|
223
|
1 /* vi:set ts=8 sts=4 sw=4:
|
|
|
2 *
|
|
|
3 * VIM - Vi IMproved by Bram Moolenaar
|
|
|
4 *
|
|
|
5 * Do ":help uganda" in Vim to read copying and usage conditions.
|
|
|
6 * Do ":help credits" in Vim to see a list of people who contributed.
|
|
|
7 * See README.txt for an overview of the Vim source code.
|
|
|
8 */
|
|
|
9
|
|
|
10 /*
|
|
|
11 * spell.c: code for spell checking
|
|
226
|
12 *
|
|
300
|
13 * The spell checking mechanism uses a tree (aka trie). Each node in the tree
|
|
|
14 * has a list of bytes that can appear (siblings). For each byte there is a
|
|
|
15 * pointer to the node with the byte that follows in the word (child).
|
|
324
|
16 *
|
|
|
17 * A NUL byte is used where the word may end. The bytes are sorted, so that
|
|
|
18 * binary searching can be used and the NUL bytes are at the start. The
|
|
|
19 * number of possible bytes is stored before the list of bytes.
|
|
|
20 *
|
|
|
21 * The tree uses two arrays: "byts" stores the characters, "idxs" stores
|
|
|
22 * either the next index or flags. The tree starts at index 0. For example,
|
|
|
23 * to lookup "vi" this sequence is followed:
|
|
|
24 * i = 0
|
|
|
25 * len = byts[i]
|
|
|
26 * n = where "v" appears in byts[i + 1] to byts[i + len]
|
|
|
27 * i = idxs[n]
|
|
|
28 * len = byts[i]
|
|
|
29 * n = where "i" appears in byts[i + 1] to byts[i + len]
|
|
|
30 * i = idxs[n]
|
|
|
31 * len = byts[i]
|
|
|
32 * find that byts[i + 1] is 0, idxs[i + 1] has flags for "vi".
|
|
300
|
33 *
|
|
339
|
34 * There are two word trees: one with case-folded words and one with words in
|
|
300
|
35 * original case. The second one is only used for keep-case words and is
|
|
|
36 * usually small.
|
|
|
37 *
|
|
481
|
38 * There is one additional tree for when not all prefixes are applied when
|
|
339
|
39 * generating the .spl file. This tree stores all the possible prefixes, as
|
|
|
40 * if they were words. At each word (prefix) end the prefix nr is stored, the
|
|
|
41 * following word must support this prefix nr. And the condition nr is
|
|
|
42 * stored, used to lookup the condition that the word must match with.
|
|
|
43 *
|
|
300
|
44 * Thanks to Olaf Seibert for providing an example implementation of this tree
|
|
|
45 * and the compression mechanism.
|
|
243
|
46 *
|
|
|
47 * Matching involves checking the caps type: Onecap ALLCAP KeepCap.
|
|
|
48 *
|
|
236
|
49 * Why doesn't Vim use aspell/ispell/myspell/etc.?
|
|
|
50 * See ":help develop-spell".
|
|
|
51 */
|
|
|
52
|
|
469
|
53 /* Use SPELL_PRINTTREE for debugging: dump the word tree after adding a word.
|
|
482
|
54 * Only use it for small word lists! */
|
|
469
|
55 #if 0
|
|
|
56 # define SPELL_PRINTTREE
|
|
482
|
57 #endif
|
|
|
58
|
|
300
|
59 /*
|
|
324
|
60 * Use this to adjust the score after finding suggestions, based on the
|
|
|
61 * suggested word sounding like the bad word. This is much faster than doing
|
|
|
62 * it for every possible suggestion.
|
|
|
63 * Disadvantage: When "the" is typed as "hte" it sounds different and goes
|
|
|
64 * down in the list.
|
|
344
|
65 * Used when 'spellsuggest' is set to "best".
|
|
|
66 */
|
|
|
67 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4)
|
|
|
68
|
|
|
69 /*
|
|
339
|
70 * Vim spell file format: <HEADER>
|
|
483
|
71 * <SECTIONS>
|
|
339
|
72 * <LWORDTREE>
|
|
|
73 * <KWORDTREE>
|
|
|
74 * <PREFIXTREE>
|
|
300
|
75 *
|
|
483
|
76 * <HEADER>: <fileID> <versionnr>
|
|
|
77 *
|
|
|
78 * <fileID> 8 bytes "VIMspell"
|
|
|
79 * <versionnr> 1 byte VIMSPELLVERSION
|
|
|
80 *
|
|
|
81 *
|
|
|
82 * Sections make it possible to add information to the .spl file without
|
|
|
83 * making it incompatible with previous versions. There are two kinds of
|
|
|
84 * sections:
|
|
|
85 * 1. Not essential for correct spell checking. E.g. for making suggestions.
|
|
|
86 * These are skipped when not supported.
|
|
|
87 * 2. Optional information, but essential for spell checking when present.
|
|
|
88 * E.g. conditions for affixes. When this section is present but not
|
|
|
89 * supported an error message is given.
|
|
300
|
90 *
|
|
483
|
91 * <SECTIONS>: <section> ... <sectionend>
|
|
|
92 *
|
|
|
93 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
94 *
|
|
|
95 * <sectionID> 1 byte number from 0 to 254 identifying the section
|
|
|
96 *
|
|
|
97 * <sectionflags> 1 byte SNF_REQUIRED: this section is required for correct
|
|
|
98 * spell checking
|
|
|
99 *
|
|
|
100 * <sectionlen> 4 bytes length of section contents, MSB first
|
|
|
101 *
|
|
|
102 * <sectionend> 1 byte SN_END
|
|
|
103 *
|
|
|
104 *
|
|
|
105 * sectionID == SN_REGION: <regionname> ...
|
|
|
106 * <regionname> 2 bytes Up to 8 region names: ca, au, etc. Lower case.
|
|
300
|
107 * First <regionname> is region 1.
|
|
|
108 *
|
|
483
|
109 * sectionID == SN_CHARFLAGS: <charflagslen> <charflags>
|
|
|
110 * <folcharslen> <folchars>
|
|
300
|
111 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
112 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
113 * 0x01 word character CF_WORD
|
|
|
114 * 0x02 upper-case character CF_UPPER
|
|
483
|
115 * <folcharslen> 2 bytes Number of bytes in <folchars>.
|
|
|
116 * <folchars> N bytes Folded characters, first one is for character 128.
|
|
300
|
117 *
|
|
483
|
118 * sectionID == SN_MIDWORD: <midword>
|
|
|
119 * <midword> N bytes Characters that are word characters only when used
|
|
366
|
120 * in the middle of a word.
|
|
|
121 *
|
|
483
|
122 * sectionID == SN_PREFCOND: <prefcondcnt> <prefcond> ...
|
|
339
|
123 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
124 * <prefcond> : <condlen> <condstr>
|
|
|
125 * <condlen> 1 byte Length of <condstr>.
|
|
|
126 * <condstr> N bytes Condition for the prefix.
|
|
|
127 *
|
|
483
|
128 * sectionID == SN_REP: <repcount> <rep> ...
|
|
|
129 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
323
|
130 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
483
|
131 * <repfromlen> 1 byte length of <repfrom>
|
|
|
132 * <repfrom> N bytes "from" part of replacement
|
|
|
133 * <reptolen> 1 byte length of <repto>
|
|
|
134 * <repto> N bytes "to" part of replacement
|
|
323
|
135 *
|
|
483
|
136 * sectionID == SN_SAL: <salflags> <salcount> <sal> ...
|
|
|
137 * <salflags> 1 byte flags for soundsalike conversion:
|
|
323
|
138 * SAL_F0LLOWUP
|
|
|
139 * SAL_COLLAPSE
|
|
|
140 * SAL_REM_ACCENTS
|
|
483
|
141 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
142 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
483
|
143 * <salfromlen> 1 byte length of <salfrom>
|
|
|
144 * <salfrom> N bytes "from" part of soundsalike
|
|
|
145 * <saltolen> 1 byte length of <salto>
|
|
|
146 * <salto> N bytes "to" part of soundsalike
|
|
323
|
147 *
|
|
483
|
148 * sectionID == SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
149 * <sofofromlen> 2 bytes length of <sofofrom>
|
|
|
150 * <sofofrom> N bytes "from" part of soundfold
|
|
|
151 * <sofotolen> 2 bytes length of <sofoto>
|
|
|
152 * <sofoto> N bytes "to" part of soundfold
|
|
323
|
153 *
|
|
483
|
154 * sectionID == SN_MAP: <mapstr>
|
|
|
155 * <mapstr> N bytes String with sequences of similar characters,
|
|
323
|
156 * separated by slashes.
|
|
300
|
157 *
|
|
483
|
158 * sectionID == SN_COMPOUND: <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
159 * <compmax> 1 byte Maximum nr of words in compound word.
|
|
|
160 * <compminlen> 1 byte Minimal word length for compounding.
|
|
|
161 * <compsylmax> 1 byte Maximum nr of syllables in compound word.
|
|
|
162 * <compflags> N bytes Flags from COMPOUNDFLAGS items, separated by
|
|
|
163 * slashes.
|
|
|
164 *
|
|
492
|
165 * sectionID == SN_NOBREAK: (empty, its presence is enough)
|
|
|
166 *
|
|
483
|
167 * sectionID == SN_SYLLABLE: <syllable>
|
|
|
168 * <syllable> N bytes String from SYLLABLE item.
|
|
300
|
169 *
|
|
|
170 * <LWORDTREE>: <wordtree>
|
|
|
171 *
|
|
339
|
172 * <KWORDTREE>: <wordtree>
|
|
|
173 *
|
|
|
174 * <PREFIXTREE>: <wordtree>
|
|
|
175 *
|
|
|
176 *
|
|
300
|
177 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
178 *
|
|
|
179 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
180 *
|
|
|
181 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
182 *
|
|
|
183 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
184 * follow in sorted order.
|
|
|
185 *
|
|
339
|
186 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
481
|
187 * | <flags> [<flags2>] [<region>] [<affixID>]
|
|
|
188 * | [<pflags>] <affixID> <prefcondnr> ]
|
|
300
|
189 *
|
|
|
190 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
191 * BY_NOFLAGS: End of word without flags and for all
|
|
|
192 * regions.
|
|
481
|
193 * For PREFIXTREE <affixID> and
|
|
366
|
194 * <prefcondnr> follow.
|
|
390
|
195 * BY_FLAGS: End of word, <flags> follow.
|
|
481
|
196 * For PREFIXTREE <pflags>, <affixID>
|
|
455
|
197 * and <prefcondnr> follow.
|
|
390
|
198 * BY_FLAGS2: End of word, <flags> and <flags2>
|
|
|
199 * follow. Not used in PREFIXTREE.
|
|
|
200 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
300
|
201 * and <xbyte> follow.
|
|
|
202 *
|
|
|
203 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
204 *
|
|
|
205 * <xbyte> 1 byte byte value of the sibling.
|
|
|
206 *
|
|
|
207 * <flags> 1 byte bitmask of:
|
|
|
208 * WF_ALLCAP word must have only capitals
|
|
|
209 * WF_ONECAP first char of word must be capital
|
|
389
|
210 * WF_KEEPCAP keep-case word
|
|
|
211 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
212 * WF_RARE rare word
|
|
389
|
213 * WF_BANNED bad word
|
|
300
|
214 * WF_REGION <region> follows
|
|
481
|
215 * WF_AFX <affixID> follows
|
|
300
|
216 *
|
|
390
|
217 * <flags2> 1 byte Only used when there are postponed prefixes.
|
|
|
218 * Bitmask of:
|
|
|
219 * WF_HAS_AFF >> 8 word includes affix
|
|
|
220 *
|
|
455
|
221 * <pflags> 1 byte bitmask of:
|
|
|
222 * WFP_RARE rare prefix
|
|
|
223 * WFP_NC non-combining prefix
|
|
|
224 * WFP_UP letter after prefix made upper case
|
|
|
225 *
|
|
300
|
226 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
227 * omitted it's valid in all regions.
|
|
|
228 * Lowest bit is for region 1.
|
|
|
229 *
|
|
481
|
230 * <affixID> 1 byte ID of affix that can be used with this word. In
|
|
339
|
231 * PREFIXTREE used for the required prefix ID.
|
|
|
232 *
|
|
|
233 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
234 * from HEADER.
|
|
300
|
235 *
|
|
|
236 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
237 */
|
|
|
238
|
|
223
|
239 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
240 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
241 #endif
|
|
|
242
|
|
|
243 #include "vim.h"
|
|
|
244
|
|
|
245 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
246
|
|
|
247 #ifdef HAVE_FCNTL_H
|
|
|
248 # include <fcntl.h>
|
|
|
249 #endif
|
|
|
250
|
|
323
|
251 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
252 Some places assume a word length fits in a
|
|
|
253 byte, thus it can't be above 255. */
|
|
226
|
254
|
|
490
|
255 /* Type used for indexes in the word tree need to be at least 4 bytes. If int
|
|
324
|
256 * is 8 bytes we could use something smaller, but what? */
|
|
490
|
257 #if SIZEOF_INT > 3
|
|
324
|
258 typedef int idx_T;
|
|
|
259 #else
|
|
|
260 typedef long idx_T;
|
|
|
261 #endif
|
|
|
262
|
|
|
263 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
264 * comes above it. */
|
|
300
|
265 #define WF_REGION 0x01 /* region byte follows */
|
|
|
266 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
267 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
268 #define WF_RARE 0x08 /* rare word */
|
|
307
|
269 #define WF_BANNED 0x10 /* bad word */
|
|
481
|
270 #define WF_AFX 0x20 /* affix ID follows */
|
|
389
|
271 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
272 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
273
|
|
390
|
274 /* for <flags2>, shifted up one byte to be used in wn_flags */
|
|
|
275 #define WF_HAS_AFF 0x0100 /* word includes affix */
|
|
|
276
|
|
389
|
277 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
278
|
|
455
|
279 /* flags for <pflags> */
|
|
|
280 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
281 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
282 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
283
|
|
481
|
284 /* Flags for postponed prefixes. Must be above affixID (one byte)
|
|
390
|
285 * and prefcondnr (two bytes). */
|
|
455
|
286 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
287 * postponed prefix */
|
|
|
288 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
289 * postponed prefix */
|
|
|
290 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
291 * postponed prefix */
|
|
390
|
292
|
|
|
293 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
294 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
295 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
296 * postponed prefix: no <pflags> */
|
|
|
297 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
298 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
299 * postponed prefix: <pflags> follows */
|
|
|
300 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
301 * follow; never used in prefix tree */
|
|
|
302 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
303
|
|
323
|
304 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
305 * and si_sal. Not for sl_sal!
|
|
323
|
306 * One replacement: from "ft_from" to "ft_to". */
|
|
|
307 typedef struct fromto_S
|
|
236
|
308 {
|
|
323
|
309 char_u *ft_from;
|
|
|
310 char_u *ft_to;
|
|
|
311 } fromto_T;
|
|
236
|
312
|
|
344
|
313 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
314 * The info is split for quick processing by spell_soundfold().
|
|
|
315 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
316 typedef struct salitem_S
|
|
|
317 {
|
|
|
318 char_u *sm_lead; /* leading letters */
|
|
|
319 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
320 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
321 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
322 char_u *sm_to; /* replacement. */
|
|
372
|
323 #ifdef FEAT_MBYTE
|
|
|
324 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
325 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
326 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
327 #endif
|
|
344
|
328 } salitem_T;
|
|
|
329
|
|
375
|
330 #ifdef FEAT_MBYTE
|
|
|
331 typedef int salfirst_T;
|
|
|
332 #else
|
|
|
333 typedef short salfirst_T;
|
|
|
334 #endif
|
|
|
335
|
|
483
|
336 /* Values for SP_*ERROR are negative, positive values are used by
|
|
|
337 * read_cnt_string(). */
|
|
|
338 #define SP_TRUNCERROR -1 /* spell file truncated error */
|
|
|
339 #define SP_FORMERROR -2 /* format error in spell file */
|
|
495
|
340 #define SP_OTHERERROR -3 /* other error while reading spell file */
|
|
483
|
341
|
|
236
|
342 /*
|
|
243
|
343 * Structure used to store words and other info for one language, loaded from
|
|
|
344 * a .spl file.
|
|
300
|
345 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
346 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
347 *
|
|
|
348 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
349 * the number of possible bytes, sorted on byte value:
|
|
|
350 * <len> <byte1> <byte2> ...
|
|
|
351 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
352 * byte in "byts".
|
|
|
353 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
481
|
354 * the flags, region mask and affixID for the word. There may be several
|
|
|
355 * zeros in sequence for alternative flag/region/affixID combinations.
|
|
236
|
356 */
|
|
|
357 typedef struct slang_S slang_T;
|
|
|
358 struct slang_S
|
|
|
359 {
|
|
|
360 slang_T *sl_next; /* next language */
|
|
|
361 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
362 char_u *sl_fname; /* name of .spl file */
|
|
323
|
363 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
364
|
|
300
|
365 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
366 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
367 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
368 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
369 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
370 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
371
|
|
236
|
372 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
373
|
|
376
|
374 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
375
|
|
483
|
376 int sl_compmax; /* COMPOUNDMAX (default: MAXWLEN) */
|
|
|
377 int sl_compminlen; /* COMPOUNDMIN (default: MAXWLEN) */
|
|
|
378 int sl_compsylmax; /* COMPOUNDSYLMAX (default: MAXWLEN) */
|
|
|
379 regprog_T *sl_compprog; /* COMPOUNDFLAGS turned into a regexp progrm
|
|
|
380 * (NULL when no compounding) */
|
|
|
381 char_u *sl_compstartflags; /* flags for first compound word */
|
|
485
|
382 char_u *sl_compallflags; /* all flags for compound words */
|
|
492
|
383 char_u sl_nobreak; /* When TRUE: no spaces between words */
|
|
483
|
384 char_u *sl_syllable; /* SYLLABLE repeatable chars or NULL */
|
|
|
385 garray_T sl_syl_items; /* syllable items */
|
|
481
|
386
|
|
339
|
387 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
388 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
389
|
|
323
|
390 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
391 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
392 there is none */
|
|
344
|
393 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
394 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
395 there is none */
|
|
375
|
396 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
397 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
398 * "sl_sal" is a list of wide char lists. */
|
|
323
|
399 int sl_followup; /* SAL followup */
|
|
|
400 int sl_collapse; /* SAL collapse_result */
|
|
|
401 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
402 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
403 #ifdef FEAT_MBYTE
|
|
|
404 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
405 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
406 #else
|
|
|
407 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
408 #endif
|
|
236
|
409 };
|
|
|
410
|
|
243
|
411 /* First language that is loaded, start of the linked list of loaded
|
|
|
412 * languages. */
|
|
236
|
413 static slang_T *first_lang = NULL;
|
|
|
414
|
|
323
|
415 /* Flags used in .spl file for soundsalike flags. */
|
|
|
416 #define SAL_F0LLOWUP 1
|
|
|
417 #define SAL_COLLAPSE 2
|
|
|
418 #define SAL_REM_ACCENTS 4
|
|
|
419
|
|
236
|
420 /*
|
|
|
421 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
422 */
|
|
|
423 typedef struct langp_S
|
|
|
424 {
|
|
|
425 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
426 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
427 } langp_T;
|
|
|
428
|
|
|
429 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
430
|
|
307
|
431 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
432
|
|
483
|
433 #define VIMSPELLMAGIC "VIMspell" /* string at start of Vim spell file */
|
|
|
434 #define VIMSPELLMAGICL 8
|
|
|
435 #define VIMSPELLVERSION 50
|
|
|
436
|
|
|
437 /* Section IDs. Only renumber them when VIMSPELLVERSION changes! */
|
|
|
438 #define SN_REGION 0 /* <regionname> section */
|
|
|
439 #define SN_CHARFLAGS 1 /* charflags section */
|
|
|
440 #define SN_MIDWORD 2 /* <midword> section */
|
|
|
441 #define SN_PREFCOND 3 /* <prefcond> section */
|
|
|
442 #define SN_REP 4 /* REP items section */
|
|
|
443 #define SN_SAL 5 /* SAL items section */
|
|
|
444 #define SN_SOFO 6 /* soundfolding section */
|
|
|
445 #define SN_MAP 7 /* MAP items section */
|
|
|
446 #define SN_COMPOUND 8 /* compound words section */
|
|
|
447 #define SN_SYLLABLE 9 /* syllable section */
|
|
492
|
448 #define SN_NOBREAK 10 /* NOBREAK section */
|
|
483
|
449 #define SN_END 255 /* end of sections */
|
|
|
450
|
|
|
451 #define SNF_REQUIRED 1 /* <sectionflags>: required section */
|
|
|
452
|
|
307
|
453 /* Result values. Lower number is accepted over higher one. */
|
|
|
454 #define SP_BANNED -1
|
|
236
|
455 #define SP_OK 0
|
|
307
|
456 #define SP_RARE 1
|
|
|
457 #define SP_LOCAL 2
|
|
|
458 #define SP_BAD 3
|
|
236
|
459
|
|
381
|
460 /* file used for "zG" and "zW" */
|
|
385
|
461 static char_u *int_wordlist = NULL;
|
|
381
|
462
|
|
236
|
463 /*
|
|
323
|
464 * Information used when looking for suggestions.
|
|
|
465 */
|
|
|
466 typedef struct suginfo_S
|
|
|
467 {
|
|
|
468 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
469 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
470 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
471 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
472 char_u *su_badptr; /* start of bad word in line */
|
|
|
473 int su_badlen; /* length of detected bad word in line */
|
|
346
|
474 int su_badflags; /* caps flags for bad word */
|
|
323
|
475 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
476 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
477 hashtab_T su_banned; /* table with banned words */
|
|
|
478 } suginfo_T;
|
|
|
479
|
|
|
480 /* One word suggestion. Used in "si_ga". */
|
|
|
481 typedef struct suggest_S
|
|
|
482 {
|
|
|
483 char_u *st_word; /* suggested word, allocated string */
|
|
|
484 int st_orglen; /* length of replaced text */
|
|
|
485 int st_score; /* lower is better */
|
|
344
|
486 int st_altscore; /* used when st_score compares equal */
|
|
|
487 int st_salscore; /* st_score is for soundalike */
|
|
324
|
488 int st_had_bonus; /* bonus already included in score */
|
|
323
|
489 } suggest_T;
|
|
|
490
|
|
344
|
491 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
492
|
|
324
|
493 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
494 * called the score may change, thus we need to keep more than what is
|
|
|
495 * displayed. */
|
|
351
|
496 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
497
|
|
|
498 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
499 * of suggestions that are not going to be displayed. */
|
|
344
|
500 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
501
|
|
|
502 /* score for various changes */
|
|
344
|
503 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
504 #define SCORE_ICASE 52 /* slightly different case */
|
|
483
|
505 #define SCORE_REGION 200 /* word is for different region */
|
|
323
|
506 #define SCORE_RARE 180 /* rare word */
|
|
|
507 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
508 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
509 #define SCORE_REP 87 /* REP replacement */
|
|
|
510 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
511 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
479
|
512 #define SCORE_SUBCOMP 33 /* substitute a composing character */
|
|
324
|
513 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
514 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
479
|
515 #define SCORE_DELCOMP 28 /* delete a composing character */
|
|
324
|
516 #define SCORE_INS 96 /* insert a character */
|
|
358
|
517 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
475
|
518 #define SCORE_INSCOMP 30 /* insert a composing character */
|
|
366
|
519 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
520
|
|
372
|
521 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
522 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
523 * 350 allows for about three changes. */
|
|
344
|
524
|
|
|
525 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
526 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
527
|
|
|
528 /*
|
|
236
|
529 * Structure to store info for word matching.
|
|
|
530 */
|
|
|
531 typedef struct matchinf_S
|
|
|
532 {
|
|
|
533 langp_T *mi_lp; /* info for language and region */
|
|
243
|
534
|
|
|
535 /* pointers to original text to be checked */
|
|
236
|
536 char_u *mi_word; /* start of word being checked */
|
|
339
|
537 char_u *mi_end; /* end of matching word so far */
|
|
243
|
538 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
539 char_u *mi_cend; /* char after what was used for
|
|
|
540 mi_capflags */
|
|
243
|
541
|
|
|
542 /* case-folded text */
|
|
|
543 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
544 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
545
|
|
339
|
546 /* for when checking word after a prefix */
|
|
|
547 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
481
|
548 affixID/condition */
|
|
339
|
549 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
550 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
551 #ifdef FEAT_MBYTE
|
|
|
552 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
553 case */
|
|
|
554 #else
|
|
|
555 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
556 #endif
|
|
339
|
557
|
|
481
|
558 /* for when checking a compound word */
|
|
|
559 int mi_compoff; /* start of following word offset */
|
|
483
|
560 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */
|
|
|
561 int mi_complen; /* nr of compound words used */
|
|
481
|
562
|
|
243
|
563 /* others */
|
|
236
|
564 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
565 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
566 buf_T *mi_buf; /* buffer being checked */
|
|
492
|
567
|
|
|
568 /* for NOBREAK */
|
|
|
569 int mi_result2; /* "mi_resul" without following word */
|
|
|
570 char_u *mi_end2; /* "mi_end" without following word */
|
|
236
|
571 } matchinf_T;
|
|
|
572
|
|
307
|
573 /*
|
|
|
574 * The tables used for recognizing word characters according to spelling.
|
|
|
575 * These are only used for the first 256 characters of 'encoding'.
|
|
|
576 */
|
|
|
577 typedef struct spelltab_S
|
|
|
578 {
|
|
|
579 char_u st_isw[256]; /* flags: is word char */
|
|
|
580 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
581 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
582 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
583 } spelltab_T;
|
|
|
584
|
|
|
585 static spelltab_T spelltab;
|
|
|
586 static int did_set_spelltab;
|
|
|
587
|
|
324
|
588 #define CF_WORD 0x01
|
|
|
589 #define CF_UPPER 0x02
|
|
307
|
590
|
|
|
591 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
592 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
593 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
594 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
595 #ifdef FEAT_MBYTE
|
|
|
596 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
597 #endif
|
|
483
|
598 static int write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
599
|
|
|
600 /*
|
|
344
|
601 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
602 */
|
|
|
603 typedef enum
|
|
|
604 {
|
|
344
|
605 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
606 * ends); if badword ends there is a match, otherwise
|
|
|
607 * try splitting word. */
|
|
375
|
608 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
609 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
610 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
611 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
612 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
613 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
614 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
615 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
616 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
617 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
618 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
619 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
620 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
621 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
622 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
623 STATE_FINAL /* End of this node. */
|
|
|
624 } state_T;
|
|
|
625
|
|
|
626 /*
|
|
346
|
627 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
628 */
|
|
|
629 typedef struct trystate_S
|
|
|
630 {
|
|
330
|
631 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
632 int ts_score; /* score */
|
|
344
|
633 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
634 short ts_curi; /* index in list of child nodes */
|
|
|
635 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
636 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
637 char_u ts_twordlen; /* valid length of tword[] */
|
|
482
|
638 char_u ts_prefixdepth; /* stack depth for end of prefix or
|
|
485
|
639 * PFD_PREFIXTREE or PFD_NOPREFIX */
|
|
|
640 char_u ts_flags; /* TSF_ flags */
|
|
330
|
641 #ifdef FEAT_MBYTE
|
|
|
642 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
643 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
644 char_u ts_isdiff; /* DIFF_ values */
|
|
|
645 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
646 #endif
|
|
483
|
647 char_u ts_prewordlen; /* length of word in "preword[]" */
|
|
|
648 char_u ts_splitoff; /* index in "tword" after last split */
|
|
492
|
649 char_u ts_splitfidx; /* "ts_fidx" at word split */
|
|
483
|
650 char_u ts_complen; /* nr of compound words used */
|
|
485
|
651 char_u ts_compsplit; /* index for "compflags" where word was spit */
|
|
346
|
652 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
653 } trystate_T;
|
|
|
654
|
|
330
|
655 /* values for ts_isdiff */
|
|
|
656 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
657 #define DIFF_YES 1 /* different byte found */
|
|
|
658 #define DIFF_INSERT 2 /* inserting character */
|
|
|
659
|
|
485
|
660 /* values for ts_flags */
|
|
|
661 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */
|
|
|
662 #define TSF_DIDSPLIT 2 /* tried split at this point */
|
|
|
663
|
|
375
|
664 /* special values ts_prefixdepth */
|
|
485
|
665 #define PFD_NOPREFIX 0xff /* not using prefixes */
|
|
482
|
666 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
485
|
667 #define PFD_NOTSPECIAL 0xfd /* first value that's not special */
|
|
375
|
668
|
|
339
|
669 /* mode values for find_word */
|
|
481
|
670 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
671 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
672 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
673 #define FIND_COMPOUND 3 /* find case-folded compound word */
|
|
|
674 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */
|
|
339
|
675
|
|
236
|
676 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
677 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
678 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
679 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
483
|
680 static int can_compound __ARGS((slang_T *slang, char_u *word, char_u *flags));
|
|
455
|
681 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
485
|
682 static void find_prefix __ARGS((matchinf_T *mip, int mode));
|
|
339
|
683 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
684 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
351
|
685 static int no_spell_checking __ARGS((void));
|
|
307
|
686 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
687 static char_u *spell_enc __ARGS((void));
|
|
385
|
688 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
689 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
690 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
691 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
483
|
692 static char_u *read_string __ARGS((FILE *fd, int cnt));
|
|
|
693 static int read_region_section __ARGS((FILE *fd, slang_T *slang, int len));
|
|
|
694 static int read_charflags_section __ARGS((FILE *fd));
|
|
|
695 static int read_prefcond_section __ARGS((FILE *fd, slang_T *lp));
|
|
|
696 static int read_rep_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
697 static int read_sal_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
698 static int read_sofo_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
699 static int read_compound __ARGS((FILE *fd, slang_T *slang, int len));
|
|
495
|
700 static int byte_in_str __ARGS((char_u *str, int byte));
|
|
483
|
701 static int init_syl_tab __ARGS((slang_T *slang));
|
|
|
702 static int count_syllables __ARGS((slang_T *slang, char_u *word));
|
|
381
|
703 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
704 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
705 #ifdef FEAT_MBYTE
|
|
|
706 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
707 #endif
|
|
339
|
708 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
709 static void clear_midword __ARGS((buf_T *buf));
|
|
|
710 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
711 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
712 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
713 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
714 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
483
|
715 static void set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
716 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
717 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
475
|
718 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col));
|
|
388
|
719 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
720 #ifdef FEAT_EVAL
|
|
|
721 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
722 #endif
|
|
|
723 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
724 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
725 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
726 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
727 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
728 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
729 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
730 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
731 #ifdef FEAT_MBYTE
|
|
|
732 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
733 #endif
|
|
323
|
734 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
735 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
736 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
737 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
738 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
739 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
740 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
741 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
742 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
743 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
744 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
745 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
746 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
747 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
748 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
749 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
750 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
751 #ifdef FEAT_MBYTE
|
|
375
|
752 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
753 #endif
|
|
344
|
754 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
755 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
756 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
757 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
758
|
|
324
|
759 /*
|
|
|
760 * Use our own character-case definitions, because the current locale may
|
|
|
761 * differ from what the .spl file uses.
|
|
|
762 * These must not be called with negative number!
|
|
|
763 */
|
|
|
764 #ifndef FEAT_MBYTE
|
|
|
765 /* Non-multi-byte implementation. */
|
|
|
766 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
767 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
768 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
769 #else
|
|
447
|
770 # if defined(HAVE_WCHAR_H)
|
|
|
771 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
772 # endif
|
|
324
|
773 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
774 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
775 * the "w" library function for characters above 255 if available. */
|
|
|
776 # ifdef HAVE_TOWLOWER
|
|
|
777 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
778 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
779 # else
|
|
|
780 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
781 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
782 # endif
|
|
|
783
|
|
|
784 # ifdef HAVE_TOWUPPER
|
|
|
785 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
786 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
787 # else
|
|
|
788 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
789 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
790 # endif
|
|
|
791
|
|
|
792 # ifdef HAVE_ISWUPPER
|
|
|
793 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
794 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
795 # else
|
|
|
796 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
797 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
798 # endif
|
|
|
799 #endif
|
|
|
800
|
|
307
|
801
|
|
|
802 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
803 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
482
|
804 static char *e_afftrailing = N_("Trailing text in %s line %d: %s");
|
|
495
|
805 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
|
806 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
807 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
469
|
808 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
809
|
|
|
810 /*
|
|
|
811 * Main spell-checking function.
|
|
300
|
812 * "ptr" points to a character that could be the start of a word.
|
|
236
|
813 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
814 * or when it's OK it remains unchanged.
|
|
|
815 * This must only be called when 'spelllang' is not empty.
|
|
323
|
816 *
|
|
385
|
817 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
818 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
819 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
820 * worry.
|
|
323
|
821 *
|
|
236
|
822 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
823 * caller can skip over the word.
|
|
|
824 */
|
|
|
825 int
|
|
385
|
826 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
827 win_T *wp; /* current window */
|
|
|
828 char_u *ptr;
|
|
|
829 int *attrp;
|
|
385
|
830 int *capcol; /* column to check for Capital */
|
|
236
|
831 {
|
|
|
832 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
833 be passed to functions quickly. */
|
|
344
|
834 int nrlen = 0; /* found a number first */
|
|
385
|
835 int c;
|
|
483
|
836 int wrongcaplen = 0;
|
|
236
|
837
|
|
307
|
838 /* A word never starts at a space or a control character. Return quickly
|
|
|
839 * then, skipping over the character. */
|
|
|
840 if (*ptr <= ' ')
|
|
|
841 return 1;
|
|
483
|
842 vim_memset(&mi, 0, sizeof(matchinf_T));
|
|
236
|
843
|
|
344
|
844 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
845 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
846 * "3GPP". */
|
|
300
|
847 if (*ptr >= '0' && *ptr <= '9')
|
|
|
848 {
|
|
316
|
849 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
850 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
851 else
|
|
|
852 {
|
|
344
|
853 mi.mi_end = skipdigits(ptr);
|
|
|
854 nrlen = mi.mi_end - ptr;
|
|
300
|
855 }
|
|
376
|
856 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
857 return (int)(mi.mi_end - ptr);
|
|
346
|
858
|
|
|
859 /* Try including the digits in the word. */
|
|
|
860 mi.mi_fend = ptr + nrlen;
|
|
|
861 }
|
|
|
862 else
|
|
|
863 mi.mi_fend = ptr;
|
|
|
864
|
|
|
865 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
866 mi.mi_word = ptr;
|
|
376
|
867 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
868 {
|
|
|
869 do
|
|
|
870 {
|
|
307
|
871 mb_ptr_adv(mi.mi_fend);
|
|
376
|
872 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
873
|
|
|
874 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
875 {
|
|
|
876 /* Check word starting with capital letter. */
|
|
455
|
877 c = PTR2CHAR(ptr);
|
|
385
|
878 if (!SPELL_ISUPPER(c))
|
|
483
|
879 wrongcaplen = (int)(mi.mi_fend - ptr);
|
|
385
|
880 }
|
|
|
881 }
|
|
|
882 if (capcol != NULL)
|
|
|
883 *capcol = -1;
|
|
344
|
884
|
|
|
885 /* We always use the characters up to the next non-word character,
|
|
|
886 * also for bad words. */
|
|
|
887 mi.mi_end = mi.mi_fend;
|
|
|
888
|
|
|
889 /* Check caps type later. */
|
|
376
|
890 mi.mi_buf = wp->w_buffer;
|
|
344
|
891
|
|
483
|
892 /* case-fold the word with one non-word character, so that we can check
|
|
|
893 * for the word end. */
|
|
344
|
894 if (*mi.mi_fend != NUL)
|
|
|
895 mb_ptr_adv(mi.mi_fend);
|
|
|
896
|
|
|
897 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
898 MAXWLEN + 1);
|
|
|
899 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
900
|
|
|
901 /* The word is bad unless we recognize it. */
|
|
|
902 mi.mi_result = SP_BAD;
|
|
492
|
903 mi.mi_result2 = SP_BAD;
|
|
344
|
904
|
|
|
905 /*
|
|
|
906 * Loop over the languages specified in 'spelllang'.
|
|
|
907 * We check them all, because a matching word may be longer than an
|
|
|
908 * already found matching word.
|
|
|
909 */
|
|
|
910 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
911 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
912 {
|
|
|
913 /* Check for a matching word in case-folded words. */
|
|
|
914 find_word(&mi, FIND_FOLDWORD);
|
|
|
915
|
|
|
916 /* Check for a matching word in keep-case words. */
|
|
|
917 find_word(&mi, FIND_KEEPWORD);
|
|
|
918
|
|
|
919 /* Check for matching prefixes. */
|
|
485
|
920 find_prefix(&mi, FIND_FOLDWORD);
|
|
492
|
921
|
|
|
922 /* For a NOBREAK language, may want to use a word without a following
|
|
|
923 * word as a backup. */
|
|
|
924 if (mi.mi_lp->lp_slang->sl_nobreak && mi.mi_result == SP_BAD
|
|
|
925 && mi.mi_result2 != SP_BAD)
|
|
|
926 {
|
|
|
927 mi.mi_result = mi.mi_result2;
|
|
|
928 mi.mi_end = mi.mi_end2;
|
|
|
929 }
|
|
344
|
930 }
|
|
|
931
|
|
|
932 if (mi.mi_result != SP_OK)
|
|
|
933 {
|
|
346
|
934 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
935 * rare and local words, e.g., "3GPP". */
|
|
344
|
936 if (nrlen > 0)
|
|
346
|
937 {
|
|
|
938 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
939 return nrlen;
|
|
|
940 }
|
|
344
|
941
|
|
|
942 /* When we are at a non-word character there is no error, just
|
|
|
943 * skip over the character (try looking for a word after it). */
|
|
390
|
944 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
945 {
|
|
385
|
946 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
947 {
|
|
|
948 regmatch_T regmatch;
|
|
|
949
|
|
|
950 /* Check for end of sentence. */
|
|
|
951 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
952 regmatch.rm_ic = FALSE;
|
|
|
953 if (vim_regexec(®match, ptr, 0))
|
|
|
954 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
955 }
|
|
|
956
|
|
344
|
957 #ifdef FEAT_MBYTE
|
|
|
958 if (has_mbyte)
|
|
474
|
959 return (*mb_ptr2len)(ptr);
|
|
344
|
960 #endif
|
|
|
961 return 1;
|
|
300
|
962 }
|
|
483
|
963 else if (mi.mi_end == ptr)
|
|
|
964 /* Always include at least one character. Required for when there
|
|
|
965 * is a mixup in "midword". */
|
|
|
966 mb_ptr_adv(mi.mi_end);
|
|
492
|
967 else if (mi.mi_result == SP_BAD
|
|
|
968 && LANGP_ENTRY(wp->w_buffer->b_langp, 0)->lp_slang->sl_nobreak)
|
|
|
969 {
|
|
|
970 char_u *p, *fp;
|
|
|
971 int save_result = mi.mi_result;
|
|
|
972
|
|
|
973 /* First language in 'spelllang' is NOBREAK. Find first position
|
|
|
974 * at which any word would be valid. */
|
|
|
975 mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
976 p = mi.mi_word;
|
|
|
977 fp = mi.mi_fword;
|
|
|
978 for (;;)
|
|
|
979 {
|
|
|
980 mb_ptr_adv(p);
|
|
|
981 mb_ptr_adv(fp);
|
|
|
982 if (p >= mi.mi_end)
|
|
|
983 break;
|
|
|
984 mi.mi_compoff = fp - mi.mi_fword;
|
|
|
985 find_word(&mi, FIND_COMPOUND);
|
|
|
986 if (mi.mi_result != SP_BAD)
|
|
|
987 {
|
|
|
988 mi.mi_end = p;
|
|
|
989 break;
|
|
|
990 }
|
|
|
991 }
|
|
|
992 mi.mi_result = save_result;
|
|
|
993 }
|
|
243
|
994
|
|
344
|
995 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
996 *attrp = highlight_attr[HLF_SPB];
|
|
|
997 else if (mi.mi_result == SP_RARE)
|
|
|
998 *attrp = highlight_attr[HLF_SPR];
|
|
|
999 else
|
|
|
1000 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
1001 }
|
|
|
1002
|
|
483
|
1003 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE))
|
|
|
1004 {
|
|
|
1005 /* Report SpellCap only when the word isn't badly spelled. */
|
|
|
1006 *attrp = highlight_attr[HLF_SPC];
|
|
|
1007 return wrongcaplen;
|
|
|
1008 }
|
|
|
1009
|
|
300
|
1010 return (int)(mi.mi_end - ptr);
|
|
236
|
1011 }
|
|
|
1012
|
|
|
1013 /*
|
|
300
|
1014 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
1015 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
1016 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
1017 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
1018 * tree.
|
|
300
|
1019 *
|
|
|
1020 * For a match mip->mi_result is updated.
|
|
243
|
1021 */
|
|
|
1022 static void
|
|
339
|
1023 find_word(mip, mode)
|
|
243
|
1024 matchinf_T *mip;
|
|
339
|
1025 int mode;
|
|
243
|
1026 {
|
|
324
|
1027 idx_T arridx = 0;
|
|
300
|
1028 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
1029 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
1030 int endidxcnt = 0;
|
|
|
1031 int len;
|
|
|
1032 int wlen = 0;
|
|
|
1033 int flen;
|
|
|
1034 int c;
|
|
|
1035 char_u *ptr;
|
|
324
|
1036 idx_T lo, hi, m;
|
|
243
|
1037 #ifdef FEAT_MBYTE
|
|
300
|
1038 char_u *s;
|
|
490
|
1039 #endif
|
|
339
|
1040 char_u *p;
|
|
307
|
1041 int res = SP_BAD;
|
|
300
|
1042 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1043 unsigned flags;
|
|
|
1044 char_u *byts;
|
|
324
|
1045 idx_T *idxs;
|
|
481
|
1046 int word_ends;
|
|
485
|
1047 int prefix_found;
|
|
492
|
1048 int nobreak_result;
|
|
481
|
1049
|
|
|
1050 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND)
|
|
236
|
1051 {
|
|
300
|
1052 /* Check for word with matching case in keep-case tree. */
|
|
|
1053 ptr = mip->mi_word;
|
|
|
1054 flen = 9999; /* no case folding, always enough bytes */
|
|
|
1055 byts = slang->sl_kbyts;
|
|
|
1056 idxs = slang->sl_kidxs;
|
|
481
|
1057
|
|
|
1058 if (mode == FIND_KEEPCOMPOUND)
|
|
|
1059 /* Skip over the previously found word(s). */
|
|
|
1060 wlen += mip->mi_compoff;
|
|
236
|
1061 }
|
|
|
1062 else
|
|
|
1063 {
|
|
300
|
1064 /* Check for case-folded in case-folded tree. */
|
|
|
1065 ptr = mip->mi_fword;
|
|
|
1066 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1067 byts = slang->sl_fbyts;
|
|
|
1068 idxs = slang->sl_fidxs;
|
|
339
|
1069
|
|
|
1070 if (mode == FIND_PREFIX)
|
|
|
1071 {
|
|
|
1072 /* Skip over the prefix. */
|
|
|
1073 wlen = mip->mi_prefixlen;
|
|
|
1074 flen -= mip->mi_prefixlen;
|
|
|
1075 }
|
|
481
|
1076 else if (mode == FIND_COMPOUND)
|
|
|
1077 {
|
|
|
1078 /* Skip over the previously found word(s). */
|
|
|
1079 wlen = mip->mi_compoff;
|
|
|
1080 flen -= mip->mi_compoff;
|
|
|
1081 }
|
|
|
1082
|
|
243
|
1083 }
|
|
|
1084
|
|
300
|
1085 if (byts == NULL)
|
|
|
1086 return; /* array is empty */
|
|
236
|
1087
|
|
|
1088 /*
|
|
307
|
1089 * Repeat advancing in the tree until:
|
|
|
1090 * - there is a byte that doesn't match,
|
|
|
1091 * - we reach the end of the tree,
|
|
|
1092 * - or we reach the end of the line.
|
|
236
|
1093 */
|
|
300
|
1094 for (;;)
|
|
236
|
1095 {
|
|
346
|
1096 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
1097 flen = fold_more(mip);
|
|
300
|
1098
|
|
|
1099 len = byts[arridx++];
|
|
|
1100
|
|
|
1101 /* If the first possible byte is a zero the word could end here.
|
|
|
1102 * Remember this index, we first check for the longest word. */
|
|
|
1103 if (byts[arridx] == 0)
|
|
|
1104 {
|
|
307
|
1105 if (endidxcnt == MAXWLEN)
|
|
|
1106 {
|
|
|
1107 /* Must be a corrupted spell file. */
|
|
|
1108 EMSG(_(e_format));
|
|
|
1109 return;
|
|
|
1110 }
|
|
300
|
1111 endlen[endidxcnt] = wlen;
|
|
|
1112 endidx[endidxcnt++] = arridx++;
|
|
|
1113 --len;
|
|
|
1114
|
|
|
1115 /* Skip over the zeros, there can be several flag/region
|
|
|
1116 * combinations. */
|
|
|
1117 while (len > 0 && byts[arridx] == 0)
|
|
|
1118 {
|
|
|
1119 ++arridx;
|
|
|
1120 --len;
|
|
|
1121 }
|
|
|
1122 if (len == 0)
|
|
|
1123 break; /* no children, word must end here */
|
|
|
1124 }
|
|
|
1125
|
|
|
1126 /* Stop looking at end of the line. */
|
|
|
1127 if (ptr[wlen] == NUL)
|
|
|
1128 break;
|
|
|
1129
|
|
|
1130 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1131 c = ptr[wlen];
|
|
346
|
1132 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1133 c = ' ';
|
|
300
|
1134 lo = arridx;
|
|
|
1135 hi = arridx + len - 1;
|
|
|
1136 while (lo < hi)
|
|
|
1137 {
|
|
|
1138 m = (lo + hi) / 2;
|
|
|
1139 if (byts[m] > c)
|
|
|
1140 hi = m - 1;
|
|
|
1141 else if (byts[m] < c)
|
|
|
1142 lo = m + 1;
|
|
|
1143 else
|
|
|
1144 {
|
|
|
1145 lo = hi = m;
|
|
|
1146 break;
|
|
236
|
1147 }
|
|
|
1148 }
|
|
300
|
1149
|
|
|
1150 /* Stop if there is no matching byte. */
|
|
|
1151 if (hi < lo || byts[lo] != c)
|
|
|
1152 break;
|
|
|
1153
|
|
|
1154 /* Continue at the child (if there is one). */
|
|
|
1155 arridx = idxs[lo];
|
|
|
1156 ++wlen;
|
|
|
1157 --flen;
|
|
346
|
1158
|
|
|
1159 /* One space in the good word may stand for several spaces in the
|
|
|
1160 * checked word. */
|
|
|
1161 if (c == ' ')
|
|
|
1162 {
|
|
|
1163 for (;;)
|
|
|
1164 {
|
|
|
1165 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1166 flen = fold_more(mip);
|
|
|
1167 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1168 break;
|
|
|
1169 ++wlen;
|
|
|
1170 --flen;
|
|
|
1171 }
|
|
|
1172 }
|
|
236
|
1173 }
|
|
|
1174
|
|
300
|
1175 /*
|
|
|
1176 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1177 * first.
|
|
|
1178 */
|
|
|
1179 while (endidxcnt > 0)
|
|
|
1180 {
|
|
|
1181 --endidxcnt;
|
|
|
1182 arridx = endidx[endidxcnt];
|
|
|
1183 wlen = endlen[endidxcnt];
|
|
236
|
1184
|
|
300
|
1185 #ifdef FEAT_MBYTE
|
|
|
1186 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1187 continue; /* not at first byte of character */
|
|
|
1188 #endif
|
|
376
|
1189 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
481
|
1190 {
|
|
492
|
1191 if (slang->sl_compprog == NULL && !slang->sl_nobreak)
|
|
481
|
1192 continue; /* next char is a word character */
|
|
|
1193 word_ends = FALSE;
|
|
|
1194 }
|
|
|
1195 else
|
|
|
1196 word_ends = TRUE;
|
|
485
|
1197 /* The prefix flag is before compound flags. Once a valid prefix flag
|
|
|
1198 * has been found we try compound flags. */
|
|
|
1199 prefix_found = FALSE;
|
|
300
|
1200
|
|
|
1201 #ifdef FEAT_MBYTE
|
|
339
|
1202 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1203 {
|
|
|
1204 /* Compute byte length in original word, length may change
|
|
339
|
1205 * when folding case. This can be slow, take a shortcut when the
|
|
|
1206 * case-folded word is equal to the keep-case word. */
|
|
300
|
1207 p = mip->mi_word;
|
|
339
|
1208 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1209 {
|
|
|
1210 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1211 mb_ptr_adv(p);
|
|
|
1212 wlen = p - mip->mi_word;
|
|
|
1213 }
|
|
300
|
1214 }
|
|
|
1215 #endif
|
|
236
|
1216
|
|
339
|
1217 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1218 * prefix ID.
|
|
|
1219 * Repeat this if there are more flags/region alternatives until there
|
|
|
1220 * is a match. */
|
|
|
1221 res = SP_BAD;
|
|
|
1222 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1223 --len, ++arridx)
|
|
300
|
1224 {
|
|
|
1225 flags = idxs[arridx];
|
|
324
|
1226
|
|
339
|
1227 /* For the fold-case tree check that the case of the checked word
|
|
|
1228 * matches with what the word in the tree requires.
|
|
|
1229 * For keep-case tree the case is always right. For prefixes we
|
|
|
1230 * don't bother to check. */
|
|
|
1231 if (mode == FIND_FOLDWORD)
|
|
300
|
1232 {
|
|
|
1233 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1234 {
|
|
323
|
1235 /* mi_capflags was set for a different word length, need
|
|
|
1236 * to do it again. */
|
|
300
|
1237 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1238 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1239 }
|
|
|
1240
|
|
346
|
1241 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1242 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1243 continue;
|
|
300
|
1244 }
|
|
236
|
1245
|
|
339
|
1246 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1247 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1248 * mip->mi_prefarridx that find_prefix() filled. */
|
|
485
|
1249 else if (mode == FIND_PREFIX && !prefix_found)
|
|
481
|
1250 {
|
|
366
|
1251 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1252 flags,
|
|
455
|
1253 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1254 FALSE);
|
|
366
|
1255 if (c == 0)
|
|
339
|
1256 continue;
|
|
366
|
1257
|
|
|
1258 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1259 if (c & WF_RAREPFX)
|
|
|
1260 flags |= WF_RARE;
|
|
485
|
1261 prefix_found = TRUE;
|
|
339
|
1262 }
|
|
|
1263
|
|
492
|
1264 if (slang->sl_nobreak)
|
|
|
1265 {
|
|
|
1266 if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND)
|
|
|
1267 && (flags & WF_BANNED) == 0)
|
|
|
1268 {
|
|
|
1269 /* NOBREAK: found a valid following word. That's all we
|
|
|
1270 * need to know, so return. */
|
|
|
1271 mip->mi_result = SP_OK;
|
|
|
1272 break;
|
|
|
1273 }
|
|
|
1274 }
|
|
|
1275
|
|
|
1276 else if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND
|
|
|
1277 || !word_ends))
|
|
481
|
1278 {
|
|
483
|
1279 /* If there is no flag or the word is shorter than
|
|
|
1280 * COMPOUNDMIN reject it quickly.
|
|
|
1281 * Makes you wonder why someone puts a compound flag on a word
|
|
481
|
1282 * that's too short... Myspell compatibility requires this
|
|
|
1283 * anyway. */
|
|
490
|
1284 if (((unsigned)flags >> 24) == 0
|
|
|
1285 || wlen - mip->mi_compoff < slang->sl_compminlen)
|
|
483
|
1286 continue;
|
|
|
1287
|
|
490
|
1288 /* Limit the number of compound words to COMPOUNDMAX if no
|
|
|
1289 * maximum for syllables is specified. */
|
|
|
1290 if (!word_ends && mip->mi_complen + 2 > slang->sl_compmax
|
|
|
1291 && slang->sl_compsylmax == MAXWLEN)
|
|
483
|
1292 continue;
|
|
|
1293
|
|
485
|
1294 /* Quickly check if compounding is possible with this flag. */
|
|
495
|
1295 if (!byte_in_str(mip->mi_complen == 0
|
|
485
|
1296 ? slang->sl_compstartflags
|
|
|
1297 : slang->sl_compallflags,
|
|
495
|
1298 ((unsigned)flags >> 24)))
|
|
481
|
1299 continue;
|
|
|
1300
|
|
490
|
1301 if (mode == FIND_COMPOUND)
|
|
|
1302 {
|
|
|
1303 int capflags;
|
|
|
1304
|
|
|
1305 /* Need to check the caps type of the appended compound
|
|
|
1306 * word. */
|
|
|
1307 #ifdef FEAT_MBYTE
|
|
|
1308 if (has_mbyte && STRNCMP(ptr, mip->mi_word,
|
|
|
1309 mip->mi_compoff) != 0)
|
|
|
1310 {
|
|
|
1311 /* case folding may have changed the length */
|
|
|
1312 p = mip->mi_word;
|
|
|
1313 for (s = ptr; s < ptr + mip->mi_compoff; mb_ptr_adv(s))
|
|
|
1314 mb_ptr_adv(p);
|
|
|
1315 }
|
|
|
1316 else
|
|
|
1317 #endif
|
|
|
1318 p = mip->mi_word + mip->mi_compoff;
|
|
|
1319 capflags = captype(p, mip->mi_word + wlen);
|
|
|
1320 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP
|
|
|
1321 && (flags & WF_FIXCAP) != 0))
|
|
|
1322 continue;
|
|
|
1323
|
|
|
1324 if (capflags != WF_ALLCAP)
|
|
|
1325 {
|
|
|
1326 /* When the character before the word is a word
|
|
|
1327 * character we do not accept a Onecap word. We do
|
|
|
1328 * accept a no-caps word, even when the dictionary
|
|
|
1329 * word specifies ONECAP. */
|
|
|
1330 mb_ptr_back(mip->mi_word, p);
|
|
|
1331 if (spell_iswordp_nmw(p)
|
|
|
1332 ? capflags == WF_ONECAP
|
|
|
1333 : (flags & WF_ONECAP) != 0
|
|
|
1334 && capflags != WF_ONECAP)
|
|
|
1335 continue;
|
|
|
1336 }
|
|
|
1337 }
|
|
|
1338
|
|
483
|
1339 /* If the word ends the sequence of compound flags of the
|
|
|
1340 * words must match with one of the COMPOUNDFLAGS items and
|
|
|
1341 * the number of syllables must not be too large. */
|
|
|
1342 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24);
|
|
|
1343 mip->mi_compflags[mip->mi_complen + 1] = NUL;
|
|
|
1344 if (word_ends)
|
|
|
1345 {
|
|
|
1346 char_u fword[MAXWLEN];
|
|
|
1347
|
|
|
1348 if (slang->sl_compsylmax < MAXWLEN)
|
|
|
1349 {
|
|
|
1350 /* "fword" is only needed for checking syllables. */
|
|
|
1351 if (ptr == mip->mi_word)
|
|
|
1352 (void)spell_casefold(ptr, wlen, fword, MAXWLEN);
|
|
|
1353 else
|
|
|
1354 vim_strncpy(fword, ptr, endlen[endidxcnt]);
|
|
|
1355 }
|
|
|
1356 if (!can_compound(slang, fword, mip->mi_compflags))
|
|
|
1357 continue;
|
|
|
1358 }
|
|
481
|
1359 }
|
|
|
1360
|
|
492
|
1361 nobreak_result = SP_OK;
|
|
|
1362
|
|
481
|
1363 if (!word_ends)
|
|
|
1364 {
|
|
492
|
1365 int save_result = mip->mi_result;
|
|
|
1366 char_u *save_end = mip->mi_end;
|
|
|
1367
|
|
|
1368 /* Check that a valid word follows. If there is one and we
|
|
|
1369 * are compounding, it will set "mi_result", thus we are
|
|
|
1370 * always finished here. For NOBREAK we only check that a
|
|
|
1371 * valid word follows.
|
|
481
|
1372 * Recursive! */
|
|
492
|
1373 if (slang->sl_nobreak)
|
|
|
1374 mip->mi_result = SP_BAD;
|
|
481
|
1375
|
|
|
1376 /* Find following word in case-folded tree. */
|
|
|
1377 mip->mi_compoff = endlen[endidxcnt];
|
|
|
1378 #ifdef FEAT_MBYTE
|
|
|
1379 if (has_mbyte && mode == FIND_KEEPWORD)
|
|
|
1380 {
|
|
|
1381 /* Compute byte length in case-folded word from "wlen":
|
|
|
1382 * byte length in keep-case word. Length may change when
|
|
|
1383 * folding case. This can be slow, take a shortcut when
|
|
|
1384 * the case-folded word is equal to the keep-case word. */
|
|
|
1385 p = mip->mi_fword;
|
|
|
1386 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1387 {
|
|
|
1388 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1389 mb_ptr_adv(p);
|
|
|
1390 mip->mi_compoff = p - mip->mi_fword;
|
|
|
1391 }
|
|
|
1392 }
|
|
|
1393 #endif
|
|
485
|
1394 c = mip->mi_compoff;
|
|
483
|
1395 ++mip->mi_complen;
|
|
481
|
1396 find_word(mip, FIND_COMPOUND);
|
|
|
1397
|
|
492
|
1398 /* When NOBREAK any word that matches is OK. Otherwise we
|
|
|
1399 * need to find the longest match, thus try with keep-case and
|
|
|
1400 * prefix too. */
|
|
|
1401 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1402 {
|
|
|
1403 /* Find following word in keep-case tree. */
|
|
|
1404 mip->mi_compoff = wlen;
|
|
|
1405 find_word(mip, FIND_KEEPCOMPOUND);
|
|
|
1406
|
|
|
1407 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1408 {
|
|
|
1409 /* Check for following word with prefix. */
|
|
|
1410 mip->mi_compoff = c;
|
|
|
1411 find_prefix(mip, FIND_COMPOUND);
|
|
|
1412 }
|
|
|
1413 }
|
|
483
|
1414 --mip->mi_complen;
|
|
485
|
1415
|
|
492
|
1416 if (slang->sl_nobreak)
|
|
|
1417 {
|
|
|
1418 nobreak_result = mip->mi_result;
|
|
|
1419 mip->mi_result = save_result;
|
|
|
1420 mip->mi_end = save_end;
|
|
|
1421 }
|
|
|
1422 else
|
|
|
1423 {
|
|
|
1424 if (mip->mi_result == SP_OK)
|
|
|
1425 break;
|
|
|
1426 continue;
|
|
|
1427 }
|
|
481
|
1428 }
|
|
|
1429
|
|
339
|
1430 if (flags & WF_BANNED)
|
|
|
1431 res = SP_BANNED;
|
|
|
1432 else if (flags & WF_REGION)
|
|
|
1433 {
|
|
|
1434 /* Check region. */
|
|
390
|
1435 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1436 res = SP_OK;
|
|
339
|
1437 else
|
|
|
1438 res = SP_LOCAL;
|
|
300
|
1439 }
|
|
339
|
1440 else if (flags & WF_RARE)
|
|
|
1441 res = SP_RARE;
|
|
307
|
1442 else
|
|
339
|
1443 res = SP_OK;
|
|
|
1444
|
|
492
|
1445 /* Always use the longest match and the best result. For NOBREAK
|
|
|
1446 * we separately keep the longest match without a following good
|
|
|
1447 * word as a fall-back. */
|
|
|
1448 if (nobreak_result == SP_BAD)
|
|
|
1449 {
|
|
|
1450 if (mip->mi_result2 > res)
|
|
|
1451 {
|
|
|
1452 mip->mi_result2 = res;
|
|
|
1453 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1454 }
|
|
|
1455 else if (mip->mi_result2 == res
|
|
|
1456 && mip->mi_end2 < mip->mi_word + wlen)
|
|
|
1457 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1458 }
|
|
|
1459 else if (mip->mi_result > res)
|
|
339
|
1460 {
|
|
|
1461 mip->mi_result = res;
|
|
|
1462 mip->mi_end = mip->mi_word + wlen;
|
|
|
1463 }
|
|
351
|
1464 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1465 mip->mi_end = mip->mi_word + wlen;
|
|
|
1466
|
|
492
|
1467 if (mip->mi_result == SP_OK)
|
|
339
|
1468 break;
|
|
300
|
1469 }
|
|
|
1470
|
|
492
|
1471 if (mip->mi_result == SP_OK)
|
|
300
|
1472 break;
|
|
|
1473 }
|
|
236
|
1474 }
|
|
|
1475
|
|
323
|
1476 /*
|
|
483
|
1477 * Return TRUE if "flags" is a valid sequence of compound flags and
|
|
|
1478 * "word[len]" does not have too many syllables.
|
|
482
|
1479 */
|
|
|
1480 static int
|
|
483
|
1481 can_compound(slang, word, flags)
|
|
482
|
1482 slang_T *slang;
|
|
483
|
1483 char_u *word;
|
|
|
1484 char_u *flags;
|
|
|
1485 {
|
|
|
1486 regmatch_T regmatch;
|
|
495
|
1487 #ifdef FEAT_MBYTE
|
|
|
1488 char_u uflags[MAXWLEN * 2];
|
|
|
1489 int i;
|
|
|
1490 #endif
|
|
|
1491 char_u *p;
|
|
483
|
1492
|
|
|
1493 if (slang->sl_compprog == NULL)
|
|
|
1494 return FALSE;
|
|
495
|
1495 #ifdef FEAT_MBYTE
|
|
|
1496 if (enc_utf8)
|
|
|
1497 {
|
|
|
1498 /* Need to convert the single byte flags to utf8 characters. */
|
|
|
1499 p = uflags;
|
|
|
1500 for (i = 0; flags[i] != NUL; ++i)
|
|
|
1501 p += mb_char2bytes(flags[i], p);
|
|
|
1502 *p = NUL;
|
|
|
1503 p = uflags;
|
|
|
1504 }
|
|
|
1505 else
|
|
|
1506 #endif
|
|
|
1507 p = flags;
|
|
483
|
1508 regmatch.regprog = slang->sl_compprog;
|
|
|
1509 regmatch.rm_ic = FALSE;
|
|
495
|
1510 if (!vim_regexec(®match, p, 0))
|
|
483
|
1511 return FALSE;
|
|
|
1512
|
|
490
|
1513 /* Count the number of syllables. This may be slow, do it last. If there
|
|
|
1514 * are too many syllables AND the number of compound words is above
|
|
|
1515 * COMPOUNDMAX then compounding is not allowed. */
|
|
483
|
1516 if (slang->sl_compsylmax < MAXWLEN
|
|
|
1517 && count_syllables(slang, word) > slang->sl_compsylmax)
|
|
495
|
1518 return (int)STRLEN(flags) < slang->sl_compmax;
|
|
483
|
1519 return TRUE;
|
|
482
|
1520 }
|
|
|
1521
|
|
|
1522 /*
|
|
390
|
1523 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1524 * ID in "flags" for the word "word".
|
|
366
|
1525 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1526 */
|
|
|
1527 static int
|
|
455
|
1528 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1529 int totprefcnt; /* nr of prefix IDs */
|
|
|
1530 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1531 int flags;
|
|
351
|
1532 char_u *word;
|
|
|
1533 slang_T *slang;
|
|
455
|
1534 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1535 {
|
|
|
1536 int prefcnt;
|
|
|
1537 int pidx;
|
|
|
1538 regprog_T *rp;
|
|
|
1539 regmatch_T regmatch;
|
|
390
|
1540 int prefid;
|
|
|
1541
|
|
|
1542 prefid = (unsigned)flags >> 24;
|
|
351
|
1543 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1544 {
|
|
|
1545 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1546
|
|
|
1547 /* Check the prefix ID. */
|
|
|
1548 if (prefid != (pidx & 0xff))
|
|
|
1549 continue;
|
|
|
1550
|
|
390
|
1551 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1552 * suffix. */
|
|
|
1553 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1554 continue;
|
|
|
1555
|
|
351
|
1556 /* Check the condition, if there is one. The condition index is
|
|
366
|
1557 * stored in the two bytes above the prefix ID byte. */
|
|
|
1558 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1559 if (rp != NULL)
|
|
|
1560 {
|
|
|
1561 regmatch.regprog = rp;
|
|
|
1562 regmatch.rm_ic = FALSE;
|
|
|
1563 if (!vim_regexec(®match, word, 0))
|
|
|
1564 continue;
|
|
|
1565 }
|
|
455
|
1566 else if (cond_req)
|
|
|
1567 continue;
|
|
|
1568
|
|
|
1569 /* It's a match! Return the WF_ flags. */
|
|
366
|
1570 return pidx;
|
|
|
1571 }
|
|
|
1572 return 0;
|
|
351
|
1573 }
|
|
|
1574
|
|
|
1575 /*
|
|
339
|
1576 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1577 * If it does, then check the following word.
|
|
|
1578 *
|
|
485
|
1579 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a
|
|
|
1580 * prefix in a compound word.
|
|
|
1581 *
|
|
339
|
1582 * For a match mip->mi_result is updated.
|
|
|
1583 */
|
|
|
1584 static void
|
|
485
|
1585 find_prefix(mip, mode)
|
|
339
|
1586 matchinf_T *mip;
|
|
485
|
1587 int mode;
|
|
339
|
1588 {
|
|
|
1589 idx_T arridx = 0;
|
|
|
1590 int len;
|
|
|
1591 int wlen = 0;
|
|
|
1592 int flen;
|
|
|
1593 int c;
|
|
|
1594 char_u *ptr;
|
|
|
1595 idx_T lo, hi, m;
|
|
|
1596 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1597 char_u *byts;
|
|
|
1598 idx_T *idxs;
|
|
|
1599
|
|
375
|
1600 byts = slang->sl_pbyts;
|
|
|
1601 if (byts == NULL)
|
|
|
1602 return; /* array is empty */
|
|
|
1603
|
|
339
|
1604 /* We use the case-folded word here, since prefixes are always
|
|
|
1605 * case-folded. */
|
|
|
1606 ptr = mip->mi_fword;
|
|
|
1607 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
485
|
1608 if (mode == FIND_COMPOUND)
|
|
|
1609 {
|
|
|
1610 /* Skip over the previously found word(s). */
|
|
|
1611 ptr += mip->mi_compoff;
|
|
|
1612 flen -= mip->mi_compoff;
|
|
|
1613 }
|
|
339
|
1614 idxs = slang->sl_pidxs;
|
|
|
1615
|
|
|
1616 /*
|
|
|
1617 * Repeat advancing in the tree until:
|
|
|
1618 * - there is a byte that doesn't match,
|
|
|
1619 * - we reach the end of the tree,
|
|
|
1620 * - or we reach the end of the line.
|
|
|
1621 */
|
|
|
1622 for (;;)
|
|
|
1623 {
|
|
|
1624 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1625 flen = fold_more(mip);
|
|
|
1626
|
|
|
1627 len = byts[arridx++];
|
|
|
1628
|
|
|
1629 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1630 * Check if the following word matches and supports the prefix. */
|
|
|
1631 if (byts[arridx] == 0)
|
|
|
1632 {
|
|
|
1633 /* There can be several prefixes with different conditions. We
|
|
|
1634 * try them all, since we don't know which one will give the
|
|
|
1635 * longest match. The word is the same each time, pass the list
|
|
|
1636 * of possible prefixes to find_word(). */
|
|
|
1637 mip->mi_prefarridx = arridx;
|
|
|
1638 mip->mi_prefcnt = len;
|
|
|
1639 while (len > 0 && byts[arridx] == 0)
|
|
|
1640 {
|
|
|
1641 ++arridx;
|
|
|
1642 --len;
|
|
|
1643 }
|
|
|
1644 mip->mi_prefcnt -= len;
|
|
|
1645
|
|
|
1646 /* Find the word that comes after the prefix. */
|
|
|
1647 mip->mi_prefixlen = wlen;
|
|
485
|
1648 if (mode == FIND_COMPOUND)
|
|
|
1649 /* Skip over the previously found word(s). */
|
|
|
1650 mip->mi_prefixlen += mip->mi_compoff;
|
|
|
1651
|
|
455
|
1652 #ifdef FEAT_MBYTE
|
|
|
1653 if (has_mbyte)
|
|
|
1654 {
|
|
|
1655 /* Case-folded length may differ from original length. */
|
|
485
|
1656 mip->mi_cprefixlen = nofold_len(mip->mi_fword,
|
|
|
1657 mip->mi_prefixlen, mip->mi_word);
|
|
455
|
1658 }
|
|
|
1659 else
|
|
485
|
1660 mip->mi_cprefixlen = mip->mi_prefixlen;
|
|
455
|
1661 #endif
|
|
339
|
1662 find_word(mip, FIND_PREFIX);
|
|
|
1663
|
|
|
1664
|
|
|
1665 if (len == 0)
|
|
|
1666 break; /* no children, word must end here */
|
|
|
1667 }
|
|
|
1668
|
|
|
1669 /* Stop looking at end of the line. */
|
|
|
1670 if (ptr[wlen] == NUL)
|
|
|
1671 break;
|
|
|
1672
|
|
|
1673 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1674 c = ptr[wlen];
|
|
|
1675 lo = arridx;
|
|
|
1676 hi = arridx + len - 1;
|
|
|
1677 while (lo < hi)
|
|
|
1678 {
|
|
|
1679 m = (lo + hi) / 2;
|
|
|
1680 if (byts[m] > c)
|
|
|
1681 hi = m - 1;
|
|
|
1682 else if (byts[m] < c)
|
|
|
1683 lo = m + 1;
|
|
|
1684 else
|
|
|
1685 {
|
|
|
1686 lo = hi = m;
|
|
|
1687 break;
|
|
|
1688 }
|
|
|
1689 }
|
|
|
1690
|
|
|
1691 /* Stop if there is no matching byte. */
|
|
|
1692 if (hi < lo || byts[lo] != c)
|
|
|
1693 break;
|
|
|
1694
|
|
|
1695 /* Continue at the child (if there is one). */
|
|
|
1696 arridx = idxs[lo];
|
|
|
1697 ++wlen;
|
|
|
1698 --flen;
|
|
|
1699 }
|
|
|
1700 }
|
|
|
1701
|
|
|
1702 /*
|
|
|
1703 * Need to fold at least one more character. Do until next non-word character
|
|
|
1704 * for efficiency.
|
|
|
1705 * Return the length of the folded chars in bytes.
|
|
|
1706 */
|
|
|
1707 static int
|
|
|
1708 fold_more(mip)
|
|
|
1709 matchinf_T *mip;
|
|
|
1710 {
|
|
|
1711 int flen;
|
|
|
1712 char_u *p;
|
|
|
1713
|
|
|
1714 p = mip->mi_fend;
|
|
|
1715 do
|
|
|
1716 {
|
|
|
1717 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1718 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1719
|
|
|
1720 /* Include the non-word character so that we can check for the
|
|
|
1721 * word end. */
|
|
|
1722 if (*mip->mi_fend != NUL)
|
|
|
1723 mb_ptr_adv(mip->mi_fend);
|
|
|
1724
|
|
|
1725 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1726 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1727 MAXWLEN - mip->mi_fwordlen);
|
|
|
1728 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1729 mip->mi_fwordlen += flen;
|
|
|
1730 return flen;
|
|
|
1731 }
|
|
|
1732
|
|
|
1733 /*
|
|
323
|
1734 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1735 * case.
|
|
|
1736 */
|
|
|
1737 static int
|
|
389
|
1738 spell_valid_case(wordflags, treeflags)
|
|
|
1739 int wordflags; /* flags for the checked word. */
|
|
323
|
1740 int treeflags; /* flags for the word in the spell tree */
|
|
|
1741 {
|
|
389
|
1742 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1743 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1744 && ((treeflags & WF_ONECAP) == 0
|
|
|
1745 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1746 }
|
|
|
1747
|
|
351
|
1748 /*
|
|
|
1749 * Return TRUE if spell checking is not enabled.
|
|
|
1750 */
|
|
|
1751 static int
|
|
|
1752 no_spell_checking()
|
|
|
1753 {
|
|
|
1754 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1755 {
|
|
|
1756 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1757 return TRUE;
|
|
|
1758 }
|
|
|
1759 return FALSE;
|
|
|
1760 }
|
|
300
|
1761
|
|
236
|
1762 /*
|
|
|
1763 * Move to next spell error.
|
|
323
|
1764 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
483
|
1765 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move
|
|
|
1766 * to after badly spelled word before the cursor.
|
|
495
|
1767 * Return 0 if not found, length of the badly spelled word otherwise.
|
|
236
|
1768 */
|
|
|
1769 int
|
|
323
|
1770 spell_move_to(dir, allwords, curline)
|
|
236
|
1771 int dir; /* FORWARD or BACKWARD */
|
|
|
1772 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1773 int curline;
|
|
236
|
1774 {
|
|
249
|
1775 linenr_T lnum;
|
|
|
1776 pos_T found_pos;
|
|
495
|
1777 int found_len = 0;
|
|
236
|
1778 char_u *line;
|
|
|
1779 char_u *p;
|
|
346
|
1780 char_u *endp;
|
|
|
1781 int attr;
|
|
236
|
1782 int len;
|
|
249
|
1783 int has_syntax = syntax_present(curbuf);
|
|
|
1784 int col;
|
|
|
1785 int can_spell;
|
|
346
|
1786 char_u *buf = NULL;
|
|
|
1787 int buflen = 0;
|
|
|
1788 int skip = 0;
|
|
385
|
1789 int capcol = -1;
|
|
236
|
1790
|
|
351
|
1791 if (no_spell_checking())
|
|
495
|
1792 return 0;
|
|
236
|
1793
|
|
249
|
1794 /*
|
|
|
1795 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1796 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1797 *
|
|
|
1798 * When searching backwards, we continue in the line to find the last
|
|
|
1799 * bad word (in the cursor line: before the cursor).
|
|
346
|
1800 *
|
|
|
1801 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1802 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1803 * though...
|
|
249
|
1804 */
|
|
|
1805 lnum = curwin->w_cursor.lnum;
|
|
|
1806 found_pos.lnum = 0;
|
|
236
|
1807
|
|
|
1808 while (!got_int)
|
|
|
1809 {
|
|
249
|
1810 line = ml_get(lnum);
|
|
346
|
1811
|
|
|
1812 len = STRLEN(line);
|
|
|
1813 if (buflen < len + MAXWLEN + 2)
|
|
|
1814 {
|
|
|
1815 vim_free(buf);
|
|
|
1816 buflen = len + MAXWLEN + 2;
|
|
|
1817 buf = alloc(buflen);
|
|
|
1818 if (buf == NULL)
|
|
|
1819 break;
|
|
|
1820 }
|
|
|
1821
|
|
385
|
1822 /* In first line check first word for Capital. */
|
|
|
1823 if (lnum == 1)
|
|
|
1824 capcol = 0;
|
|
|
1825
|
|
|
1826 /* For checking first word with a capital skip white space. */
|
|
|
1827 if (capcol == 0)
|
|
|
1828 capcol = skipwhite(line) - line;
|
|
|
1829
|
|
346
|
1830 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1831 * possible. */
|
|
|
1832 STRCPY(buf, line);
|
|
|
1833 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1834 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1835
|
|
|
1836 p = buf + skip;
|
|
|
1837 endp = buf + len;
|
|
|
1838 while (p < endp)
|
|
236
|
1839 {
|
|
300
|
1840 /* When searching backward don't search after the cursor. */
|
|
|
1841 if (dir == BACKWARD
|
|
|
1842 && lnum == curwin->w_cursor.lnum
|
|
346
|
1843 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1844 break;
|
|
249
|
1845
|
|
300
|
1846 /* start of word */
|
|
346
|
1847 attr = 0;
|
|
385
|
1848 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1849
|
|
300
|
1850 if (attr != 0)
|
|
|
1851 {
|
|
|
1852 /* We found a bad word. Check the attribute. */
|
|
|
1853 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1854 {
|
|
300
|
1855 /* When searching forward only accept a bad word after
|
|
|
1856 * the cursor. */
|
|
|
1857 if (dir == BACKWARD
|
|
|
1858 || lnum > curwin->w_cursor.lnum
|
|
|
1859 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1860 && (colnr_T)(curline ? p - buf + len
|
|
|
1861 : p - buf)
|
|
300
|
1862 > curwin->w_cursor.col))
|
|
236
|
1863 {
|
|
300
|
1864 if (has_syntax)
|
|
249
|
1865 {
|
|
346
|
1866 col = p - buf;
|
|
300
|
1867 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1868 FALSE, &can_spell);
|
|
|
1869 }
|
|
|
1870 else
|
|
|
1871 can_spell = TRUE;
|
|
249
|
1872
|
|
300
|
1873 if (can_spell)
|
|
|
1874 {
|
|
|
1875 found_pos.lnum = lnum;
|
|
346
|
1876 found_pos.col = p - buf;
|
|
249
|
1877 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1878 found_pos.coladd = 0;
|
|
249
|
1879 #endif
|
|
300
|
1880 if (dir == FORWARD)
|
|
|
1881 {
|
|
|
1882 /* No need to search further. */
|
|
|
1883 curwin->w_cursor = found_pos;
|
|
346
|
1884 vim_free(buf);
|
|
495
|
1885 return len;
|
|
249
|
1886 }
|
|
483
|
1887 else if (curline)
|
|
|
1888 /* Insert mode completion: put cursor after
|
|
|
1889 * the bad word. */
|
|
|
1890 found_pos.col += len;
|
|
495
|
1891 found_len = len;
|
|
249
|
1892 }
|
|
236
|
1893 }
|
|
|
1894 }
|
|
|
1895 }
|
|
|
1896
|
|
300
|
1897 /* advance to character after the word */
|
|
|
1898 p += len;
|
|
385
|
1899 capcol -= len;
|
|
236
|
1900 }
|
|
|
1901
|
|
483
|
1902 if (dir == BACKWARD && found_pos.lnum != 0)
|
|
|
1903 {
|
|
|
1904 /* Use the last match in the line. */
|
|
|
1905 curwin->w_cursor = found_pos;
|
|
|
1906 vim_free(buf);
|
|
495
|
1907 return found_len;
|
|
483
|
1908 }
|
|
|
1909
|
|
323
|
1910 if (curline)
|
|
346
|
1911 break; /* only check cursor line */
|
|
323
|
1912
|
|
236
|
1913 /* Advance to next line. */
|
|
249
|
1914 if (dir == BACKWARD)
|
|
|
1915 {
|
|
|
1916 if (lnum == 1)
|
|
346
|
1917 break;
|
|
249
|
1918 --lnum;
|
|
385
|
1919 capcol = -1;
|
|
249
|
1920 }
|
|
|
1921 else
|
|
|
1922 {
|
|
|
1923 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1924 break;
|
|
249
|
1925 ++lnum;
|
|
346
|
1926
|
|
|
1927 /* Skip the characters at the start of the next line that were
|
|
|
1928 * included in a match crossing line boundaries. */
|
|
|
1929 if (attr == 0)
|
|
|
1930 skip = p - endp;
|
|
|
1931 else
|
|
|
1932 skip = 0;
|
|
385
|
1933
|
|
|
1934 /* Capscol skips over the inserted space. */
|
|
|
1935 --capcol;
|
|
|
1936
|
|
|
1937 /* But after empty line check first word in next line */
|
|
|
1938 if (*skipwhite(line) == NUL)
|
|
|
1939 capcol = 0;
|
|
249
|
1940 }
|
|
236
|
1941
|
|
|
1942 line_breakcheck();
|
|
|
1943 }
|
|
|
1944
|
|
346
|
1945 vim_free(buf);
|
|
495
|
1946 return 0;
|
|
346
|
1947 }
|
|
|
1948
|
|
|
1949 /*
|
|
|
1950 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1951 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1952 */
|
|
|
1953 void
|
|
|
1954 spell_cat_line(buf, line, maxlen)
|
|
|
1955 char_u *buf;
|
|
|
1956 char_u *line;
|
|
|
1957 int maxlen;
|
|
|
1958 {
|
|
|
1959 char_u *p;
|
|
|
1960 int n;
|
|
|
1961
|
|
|
1962 p = skipwhite(line);
|
|
|
1963 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1964 p = skipwhite(p + 1);
|
|
|
1965
|
|
|
1966 if (*p != NUL)
|
|
|
1967 {
|
|
|
1968 *buf = ' ';
|
|
376
|
1969 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1970 n = p - line;
|
|
|
1971 if (n >= maxlen)
|
|
|
1972 n = maxlen - 1;
|
|
|
1973 vim_memset(buf + 1, ' ', n);
|
|
|
1974 }
|
|
236
|
1975 }
|
|
|
1976
|
|
|
1977 /*
|
|
307
|
1978 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1979 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1980 */
|
|
307
|
1981 static void
|
|
236
|
1982 spell_load_lang(lang)
|
|
|
1983 char_u *lang;
|
|
|
1984 {
|
|
310
|
1985 char_u fname_enc[85];
|
|
236
|
1986 int r;
|
|
307
|
1987 char_u langcp[MAXWLEN + 1];
|
|
|
1988
|
|
310
|
1989 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1990 * It's truncated when an error is detected. */
|
|
|
1991 STRCPY(langcp, lang);
|
|
|
1992
|
|
310
|
1993 /*
|
|
|
1994 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1995 */
|
|
|
1996 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1997 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1998 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1999
|
|
|
2000 if (r == FAIL && *langcp != NUL)
|
|
|
2001 {
|
|
|
2002 /* Try loading the ASCII version. */
|
|
310
|
2003 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
2004 "spell/%s.ascii.spl", lang);
|
|
310
|
2005 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
2006 }
|
|
|
2007
|
|
|
2008 if (r == FAIL)
|
|
483
|
2009 smsg((char_u *)_("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""),
|
|
|
2010 lang, spell_enc(), lang);
|
|
310
|
2011 else if (*langcp != NUL)
|
|
|
2012 {
|
|
|
2013 /* Load all the additions. */
|
|
|
2014 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
2015 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
2016 }
|
|
|
2017 }
|
|
|
2018
|
|
|
2019 /*
|
|
|
2020 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
2021 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
2022 */
|
|
|
2023 static char_u *
|
|
|
2024 spell_enc()
|
|
|
2025 {
|
|
|
2026
|
|
|
2027 #ifdef FEAT_MBYTE
|
|
|
2028 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
2029 return p_enc;
|
|
|
2030 #endif
|
|
|
2031 return (char_u *)"latin1";
|
|
236
|
2032 }
|
|
|
2033
|
|
|
2034 /*
|
|
385
|
2035 * Get the name of the .spl file for the internal wordlist into
|
|
|
2036 * "fname[MAXPATHL]".
|
|
|
2037 */
|
|
|
2038 static void
|
|
|
2039 int_wordlist_spl(fname)
|
|
|
2040 char_u *fname;
|
|
|
2041 {
|
|
|
2042 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
2043 int_wordlist, spell_enc());
|
|
|
2044 }
|
|
|
2045
|
|
|
2046 /*
|
|
236
|
2047 * Allocate a new slang_T.
|
|
|
2048 * Caller must fill "sl_next".
|
|
|
2049 */
|
|
|
2050 static slang_T *
|
|
|
2051 slang_alloc(lang)
|
|
|
2052 char_u *lang;
|
|
|
2053 {
|
|
|
2054 slang_T *lp;
|
|
|
2055
|
|
300
|
2056 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
2057 if (lp != NULL)
|
|
|
2058 {
|
|
|
2059 lp->sl_name = vim_strsave(lang);
|
|
323
|
2060 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
483
|
2061 lp->sl_compmax = MAXWLEN;
|
|
|
2062 lp->sl_compminlen = MAXWLEN;
|
|
|
2063 lp->sl_compsylmax = MAXWLEN;
|
|
236
|
2064 }
|
|
|
2065 return lp;
|
|
|
2066 }
|
|
|
2067
|
|
|
2068 /*
|
|
|
2069 * Free the contents of an slang_T and the structure itself.
|
|
|
2070 */
|
|
|
2071 static void
|
|
|
2072 slang_free(lp)
|
|
|
2073 slang_T *lp;
|
|
|
2074 {
|
|
|
2075 vim_free(lp->sl_name);
|
|
310
|
2076 vim_free(lp->sl_fname);
|
|
|
2077 slang_clear(lp);
|
|
|
2078 vim_free(lp);
|
|
|
2079 }
|
|
|
2080
|
|
|
2081 /*
|
|
|
2082 * Clear an slang_T so that the file can be reloaded.
|
|
|
2083 */
|
|
|
2084 static void
|
|
|
2085 slang_clear(lp)
|
|
|
2086 slang_T *lp;
|
|
|
2087 {
|
|
339
|
2088 garray_T *gap;
|
|
|
2089 fromto_T *ftp;
|
|
344
|
2090 salitem_T *smp;
|
|
339
|
2091 int i;
|
|
323
|
2092
|
|
300
|
2093 vim_free(lp->sl_fbyts);
|
|
310
|
2094 lp->sl_fbyts = NULL;
|
|
300
|
2095 vim_free(lp->sl_kbyts);
|
|
310
|
2096 lp->sl_kbyts = NULL;
|
|
339
|
2097 vim_free(lp->sl_pbyts);
|
|
|
2098 lp->sl_pbyts = NULL;
|
|
|
2099
|
|
300
|
2100 vim_free(lp->sl_fidxs);
|
|
310
|
2101 lp->sl_fidxs = NULL;
|
|
300
|
2102 vim_free(lp->sl_kidxs);
|
|
310
|
2103 lp->sl_kidxs = NULL;
|
|
339
|
2104 vim_free(lp->sl_pidxs);
|
|
|
2105 lp->sl_pidxs = NULL;
|
|
323
|
2106
|
|
344
|
2107 gap = &lp->sl_rep;
|
|
|
2108 while (gap->ga_len > 0)
|
|
323
|
2109 {
|
|
344
|
2110 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2111 vim_free(ftp->ft_from);
|
|
|
2112 vim_free(ftp->ft_to);
|
|
323
|
2113 }
|
|
344
|
2114 ga_clear(gap);
|
|
|
2115
|
|
|
2116 gap = &lp->sl_sal;
|
|
375
|
2117 if (lp->sl_sofo)
|
|
376
|
2118 {
|
|
|
2119 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
2120 if (gap->ga_data != NULL)
|
|
|
2121 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
2122 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2123 vim_free(((int **)gap->ga_data)[i]);
|
|
|
2124 }
|
|
375
|
2125 else
|
|
|
2126 /* SAL items: free salitem_T items */
|
|
|
2127 while (gap->ga_len > 0)
|
|
|
2128 {
|
|
|
2129 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2130 vim_free(smp->sm_lead);
|
|
|
2131 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
2132 vim_free(smp->sm_to);
|
|
|
2133 #ifdef FEAT_MBYTE
|
|
|
2134 vim_free(smp->sm_lead_w);
|
|
|
2135 vim_free(smp->sm_oneof_w);
|
|
|
2136 vim_free(smp->sm_to_w);
|
|
|
2137 #endif
|
|
|
2138 }
|
|
344
|
2139 ga_clear(gap);
|
|
323
|
2140
|
|
339
|
2141 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
2142 vim_free(lp->sl_prefprog[i]);
|
|
376
|
2143 lp->sl_prefixcnt = 0;
|
|
339
|
2144 vim_free(lp->sl_prefprog);
|
|
376
|
2145 lp->sl_prefprog = NULL;
|
|
|
2146
|
|
|
2147 vim_free(lp->sl_midword);
|
|
|
2148 lp->sl_midword = NULL;
|
|
339
|
2149
|
|
483
|
2150 vim_free(lp->sl_compprog);
|
|
|
2151 vim_free(lp->sl_compstartflags);
|
|
485
|
2152 vim_free(lp->sl_compallflags);
|
|
483
|
2153 lp->sl_compprog = NULL;
|
|
|
2154 lp->sl_compstartflags = NULL;
|
|
485
|
2155 lp->sl_compallflags = NULL;
|
|
483
|
2156
|
|
|
2157 vim_free(lp->sl_syllable);
|
|
|
2158 lp->sl_syllable = NULL;
|
|
|
2159 ga_clear(&lp->sl_syl_items);
|
|
481
|
2160
|
|
330
|
2161 #ifdef FEAT_MBYTE
|
|
|
2162 {
|
|
|
2163 int todo = lp->sl_map_hash.ht_used;
|
|
|
2164 hashitem_T *hi;
|
|
|
2165
|
|
|
2166 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
2167 if (!HASHITEM_EMPTY(hi))
|
|
|
2168 {
|
|
|
2169 --todo;
|
|
|
2170 vim_free(hi->hi_key);
|
|
|
2171 }
|
|
|
2172 }
|
|
|
2173 hash_clear(&lp->sl_map_hash);
|
|
|
2174 #endif
|
|
483
|
2175
|
|
|
2176 lp->sl_compmax = MAXWLEN;
|
|
|
2177 lp->sl_compminlen = MAXWLEN;
|
|
|
2178 lp->sl_compsylmax = MAXWLEN;
|
|
|
2179 lp->sl_regions[0] = NUL;
|
|
236
|
2180 }
|
|
|
2181
|
|
|
2182 /*
|
|
307
|
2183 * Load one spell file and store the info into a slang_T.
|
|
236
|
2184 * Invoked through do_in_runtimepath().
|
|
|
2185 */
|
|
|
2186 static void
|
|
310
|
2187 spell_load_cb(fname, cookie)
|
|
236
|
2188 char_u *fname;
|
|
307
|
2189 void *cookie; /* points to the language name */
|
|
236
|
2190 {
|
|
323
|
2191 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
2192 }
|
|
|
2193
|
|
|
2194 /*
|
|
|
2195 * Load one spell file and store the info into a slang_T.
|
|
|
2196 *
|
|
|
2197 * This is invoked in two ways:
|
|
|
2198 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
2199 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
2200 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
2201 * points to the existing slang_T.
|
|
323
|
2202 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
2203 */
|
|
323
|
2204 static slang_T *
|
|
|
2205 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
2206 char_u *fname;
|
|
|
2207 char_u *lang;
|
|
|
2208 slang_T *old_lp;
|
|
323
|
2209 int silent; /* no error if file doesn't exist */
|
|
310
|
2210 {
|
|
236
|
2211 FILE *fd;
|
|
483
|
2212 char_u buf[VIMSPELLMAGICL];
|
|
236
|
2213 char_u *p;
|
|
339
|
2214 char_u *bp;
|
|
|
2215 idx_T *ip;
|
|
236
|
2216 int i;
|
|
339
|
2217 int n;
|
|
300
|
2218 int len;
|
|
236
|
2219 int round;
|
|
|
2220 char_u *save_sourcing_name = sourcing_name;
|
|
|
2221 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
307
|
2222 slang_T *lp = NULL;
|
|
324
|
2223 idx_T idx;
|
|
344
|
2224 int c = 0;
|
|
483
|
2225 int res;
|
|
236
|
2226
|
|
310
|
2227 fd = mch_fopen((char *)fname, "r");
|
|
236
|
2228 if (fd == NULL)
|
|
|
2229 {
|
|
323
|
2230 if (!silent)
|
|
|
2231 EMSG2(_(e_notopen), fname);
|
|
|
2232 else if (p_verbose > 2)
|
|
|
2233 {
|
|
|
2234 verbose_enter();
|
|
|
2235 smsg((char_u *)e_notopen, fname);
|
|
|
2236 verbose_leave();
|
|
|
2237 }
|
|
255
|
2238 goto endFAIL;
|
|
236
|
2239 }
|
|
310
|
2240 if (p_verbose > 2)
|
|
|
2241 {
|
|
|
2242 verbose_enter();
|
|
|
2243 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
2244 verbose_leave();
|
|
|
2245 }
|
|
|
2246
|
|
|
2247 if (old_lp == NULL)
|
|
|
2248 {
|
|
|
2249 lp = slang_alloc(lang);
|
|
|
2250 if (lp == NULL)
|
|
|
2251 goto endFAIL;
|
|
|
2252
|
|
|
2253 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
2254 lp->sl_fname = vim_strsave(fname);
|
|
|
2255 if (lp->sl_fname == NULL)
|
|
|
2256 goto endFAIL;
|
|
|
2257
|
|
|
2258 /* Check for .add.spl. */
|
|
|
2259 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
2260 }
|
|
|
2261 else
|
|
|
2262 lp = old_lp;
|
|
307
|
2263
|
|
236
|
2264 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
2265 sourcing_name = fname;
|
|
|
2266 sourcing_lnum = 0;
|
|
|
2267
|
|
339
|
2268 /* <HEADER>: <fileID>
|
|
|
2269 */
|
|
236
|
2270 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
2271 buf[i] = getc(fd); /* <fileID> */
|
|
|
2272 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
2273 {
|
|
483
|
2274 EMSG(_("E757: This does not look like a spell file"));
|
|
381
|
2275 goto endFAIL;
|
|
483
|
2276 }
|
|
|
2277 c = getc(fd); /* <versionnr> */
|
|
|
2278 if (c < VIMSPELLVERSION)
|
|
|
2279 {
|
|
|
2280 EMSG(_("E771: Old spell file, needs to be updated"));
|
|
381
|
2281 goto endFAIL;
|
|
|
2282 }
|
|
483
|
2283 else if (c > VIMSPELLVERSION)
|
|
|
2284 {
|
|
|
2285 EMSG(_("E772: Spell file is for newer version of Vim"));
|
|
481
|
2286 goto endFAIL;
|
|
483
|
2287 }
|
|
|
2288
|
|
|
2289
|
|
|
2290 /*
|
|
|
2291 * <SECTIONS>: <section> ... <sectionend>
|
|
|
2292 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
2293 */
|
|
|
2294 for (;;)
|
|
|
2295 {
|
|
|
2296 n = getc(fd); /* <sectionID> or <sectionend> */
|
|
|
2297 if (n == SN_END)
|
|
|
2298 break;
|
|
|
2299 c = getc(fd); /* <sectionflags> */
|
|
|
2300 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2301 /* <sectionlen> */
|
|
|
2302 if (len < 0)
|
|
|
2303 goto truncerr;
|
|
|
2304
|
|
|
2305 res = 0;
|
|
|
2306 switch (n)
|
|
|
2307 {
|
|
|
2308 case SN_REGION:
|
|
|
2309 res = read_region_section(fd, lp, len);
|
|
|
2310 break;
|
|
|
2311
|
|
|
2312 case SN_CHARFLAGS:
|
|
|
2313 res = read_charflags_section(fd);
|
|
|
2314 break;
|
|
|
2315
|
|
|
2316 case SN_MIDWORD:
|
|
|
2317 lp->sl_midword = read_string(fd, len); /* <midword> */
|
|
|
2318 if (lp->sl_midword == NULL)
|
|
|
2319 goto endFAIL;
|
|
|
2320 break;
|
|
|
2321
|
|
|
2322 case SN_PREFCOND:
|
|
|
2323 res = read_prefcond_section(fd, lp);
|
|
|
2324 break;
|
|
|
2325
|
|
|
2326 case SN_REP:
|
|
|
2327 res = read_rep_section(fd, lp);
|
|
|
2328 break;
|
|
|
2329
|
|
|
2330 case SN_SAL:
|
|
|
2331 res = read_sal_section(fd, lp);
|
|
|
2332 break;
|
|
|
2333
|
|
|
2334 case SN_SOFO:
|
|
|
2335 res = read_sofo_section(fd, lp);
|
|
|
2336 break;
|
|
|
2337
|
|
|
2338 case SN_MAP:
|
|
|
2339 p = read_string(fd, len); /* <mapstr> */
|
|
|
2340 if (p == NULL)
|
|
|
2341 goto endFAIL;
|
|
|
2342 set_map_str(lp, p);
|
|
|
2343 vim_free(p);
|
|
|
2344 break;
|
|
|
2345
|
|
|
2346 case SN_COMPOUND:
|
|
|
2347 res = read_compound(fd, lp, len);
|
|
|
2348 break;
|
|
|
2349
|
|
492
|
2350 case SN_NOBREAK:
|
|
|
2351 lp->sl_nobreak = TRUE;
|
|
|
2352 break;
|
|
|
2353
|
|
483
|
2354 case SN_SYLLABLE:
|
|
|
2355 lp->sl_syllable = read_string(fd, len); /* <syllable> */
|
|
|
2356 if (lp->sl_syllable == NULL)
|
|
|
2357 goto endFAIL;
|
|
|
2358 if (init_syl_tab(lp) == FAIL)
|
|
|
2359 goto endFAIL;
|
|
|
2360 break;
|
|
|
2361
|
|
|
2362 default:
|
|
|
2363 /* Unsupported section. When it's required give an error
|
|
|
2364 * message. When it's not required skip the contents. */
|
|
|
2365 if (c & SNF_REQUIRED)
|
|
|
2366 {
|
|
|
2367 EMSG(_("E770: Unsupported section in spell file"));
|
|
|
2368 goto endFAIL;
|
|
|
2369 }
|
|
|
2370 while (--len >= 0)
|
|
|
2371 if (getc(fd) < 0)
|
|
|
2372 goto truncerr;
|
|
|
2373 break;
|
|
|
2374 }
|
|
|
2375 if (res == SP_FORMERROR)
|
|
|
2376 {
|
|
|
2377 formerr:
|
|
|
2378 EMSG(_(e_format));
|
|
381
|
2379 goto endFAIL;
|
|
344
|
2380 }
|
|
483
|
2381 if (res == SP_TRUNCERROR)
|
|
|
2382 {
|
|
|
2383 truncerr:
|
|
|
2384 EMSG(_(e_spell_trunc));
|
|
375
|
2385 goto endFAIL;
|
|
483
|
2386 }
|
|
495
|
2387 if (res == SP_OTHERERROR)
|
|
483
|
2388 goto endFAIL;
|
|
389
|
2389 }
|
|
236
|
2390
|
|
300
|
2391 /* round 1: <LWORDTREE>
|
|
339
|
2392 * round 2: <KWORDTREE>
|
|
|
2393 * round 3: <PREFIXTREE> */
|
|
|
2394 for (round = 1; round <= 3; ++round)
|
|
236
|
2395 {
|
|
300
|
2396 /* The tree size was computed when writing the file, so that we can
|
|
|
2397 * allocate it as one long block. <nodecount> */
|
|
|
2398 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2399 if (len < 0)
|
|
|
2400 goto truncerr;
|
|
|
2401 if (len > 0)
|
|
236
|
2402 {
|
|
300
|
2403 /* Allocate the byte array. */
|
|
339
|
2404 bp = lalloc((long_u)len, TRUE);
|
|
|
2405 if (bp == NULL)
|
|
300
|
2406 goto endFAIL;
|
|
|
2407 if (round == 1)
|
|
339
|
2408 lp->sl_fbyts = bp;
|
|
|
2409 else if (round == 2)
|
|
|
2410 lp->sl_kbyts = bp;
|
|
300
|
2411 else
|
|
339
|
2412 lp->sl_pbyts = bp;
|
|
236
|
2413
|
|
300
|
2414 /* Allocate the index array. */
|
|
339
|
2415 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2416 if (ip == NULL)
|
|
300
|
2417 goto endFAIL;
|
|
|
2418 if (round == 1)
|
|
339
|
2419 lp->sl_fidxs = ip;
|
|
|
2420 else if (round == 2)
|
|
|
2421 lp->sl_kidxs = ip;
|
|
300
|
2422 else
|
|
339
|
2423 lp->sl_pidxs = ip;
|
|
300
|
2424
|
|
|
2425 /* Read the tree and store it in the array. */
|
|
339
|
2426 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2427 if (idx == -1)
|
|
300
|
2428 goto truncerr;
|
|
324
|
2429 if (idx < 0)
|
|
236
|
2430 goto formerr;
|
|
|
2431 }
|
|
300
|
2432 }
|
|
243
|
2433
|
|
310
|
2434 /* For a new file link it in the list of spell files. */
|
|
|
2435 if (old_lp == NULL)
|
|
|
2436 {
|
|
|
2437 lp->sl_next = first_lang;
|
|
|
2438 first_lang = lp;
|
|
|
2439 }
|
|
307
|
2440
|
|
255
|
2441 goto endOK;
|
|
|
2442
|
|
|
2443 endFAIL:
|
|
310
|
2444 if (lang != NULL)
|
|
|
2445 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2446 *lang = NUL;
|
|
|
2447 if (lp != NULL && old_lp == NULL)
|
|
323
|
2448 {
|
|
307
|
2449 slang_free(lp);
|
|
323
|
2450 lp = NULL;
|
|
|
2451 }
|
|
255
|
2452
|
|
|
2453 endOK:
|
|
236
|
2454 if (fd != NULL)
|
|
|
2455 fclose(fd);
|
|
|
2456 sourcing_name = save_sourcing_name;
|
|
|
2457 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2458
|
|
|
2459 return lp;
|
|
236
|
2460 }
|
|
|
2461
|
|
376
|
2462 /*
|
|
|
2463 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2464 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2465 * Returns NULL when the count is zero.
|
|
483
|
2466 * Sets "*cntp" to SP_*ERROR when there is an error, length of the result
|
|
|
2467 * otherwise.
|
|
376
|
2468 */
|
|
|
2469 static char_u *
|
|
389
|
2470 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2471 FILE *fd;
|
|
|
2472 int cnt_bytes;
|
|
389
|
2473 int *cntp;
|
|
376
|
2474 {
|
|
|
2475 int cnt = 0;
|
|
|
2476 int i;
|
|
|
2477 char_u *str;
|
|
|
2478
|
|
|
2479 /* read the length bytes, MSB first */
|
|
|
2480 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2481 cnt = (cnt << 8) + getc(fd);
|
|
|
2482 if (cnt < 0)
|
|
|
2483 {
|
|
483
|
2484 *cntp = SP_TRUNCERROR;
|
|
376
|
2485 return NULL;
|
|
|
2486 }
|
|
389
|
2487 *cntp = cnt;
|
|
|
2488 if (cnt == 0)
|
|
|
2489 return NULL; /* nothing to read, return NULL */
|
|
376
|
2490
|
|
483
|
2491 str = read_string(fd, cnt);
|
|
|
2492 if (str == NULL)
|
|
495
|
2493 *cntp = SP_OTHERERROR;
|
|
483
|
2494 return str;
|
|
|
2495 }
|
|
|
2496
|
|
|
2497 /*
|
|
|
2498 * Read a string of length "cnt" from "fd" into allocated memory.
|
|
|
2499 * Returns NULL when out of memory.
|
|
|
2500 */
|
|
|
2501 static char_u *
|
|
|
2502 read_string(fd, cnt)
|
|
|
2503 FILE *fd;
|
|
|
2504 int cnt;
|
|
|
2505 {
|
|
|
2506 char_u *str;
|
|
|
2507 int i;
|
|
|
2508
|
|
376
|
2509 /* allocate memory */
|
|
|
2510 str = alloc((unsigned)cnt + 1);
|
|
483
|
2511 if (str != NULL)
|
|
|
2512 {
|
|
|
2513 /* Read the string. Doesn't check for truncated file. */
|
|
|
2514 for (i = 0; i < cnt; ++i)
|
|
|
2515 str[i] = getc(fd);
|
|
|
2516 str[i] = NUL;
|
|
|
2517 }
|
|
|
2518 return str;
|
|
|
2519 }
|
|
|
2520
|
|
|
2521 /*
|
|
|
2522 * Read SN_REGION: <regionname> ...
|
|
|
2523 * Return SP_*ERROR flags.
|
|
|
2524 */
|
|
|
2525 static int
|
|
|
2526 read_region_section(fd, lp, len)
|
|
|
2527 FILE *fd;
|
|
|
2528 slang_T *lp;
|
|
|
2529 int len;
|
|
|
2530 {
|
|
|
2531 int i;
|
|
|
2532
|
|
|
2533 if (len > 16)
|
|
|
2534 return SP_FORMERROR;
|
|
|
2535 for (i = 0; i < len; ++i)
|
|
|
2536 lp->sl_regions[i] = getc(fd); /* <regionname> */
|
|
|
2537 lp->sl_regions[len] = NUL;
|
|
|
2538 return 0;
|
|
|
2539 }
|
|
|
2540
|
|
|
2541 /*
|
|
|
2542 * Read SN_CHARFLAGS section: <charflagslen> <charflags>
|
|
|
2543 * <folcharslen> <folchars>
|
|
|
2544 * Return SP_*ERROR flags.
|
|
|
2545 */
|
|
|
2546 static int
|
|
|
2547 read_charflags_section(fd)
|
|
|
2548 FILE *fd;
|
|
|
2549 {
|
|
|
2550 char_u *flags;
|
|
|
2551 char_u *fol;
|
|
|
2552 int flagslen, follen;
|
|
|
2553
|
|
|
2554 /* <charflagslen> <charflags> */
|
|
|
2555 flags = read_cnt_string(fd, 1, &flagslen);
|
|
|
2556 if (flagslen < 0)
|
|
|
2557 return flagslen;
|
|
|
2558
|
|
|
2559 /* <folcharslen> <folchars> */
|
|
|
2560 fol = read_cnt_string(fd, 2, &follen);
|
|
|
2561 if (follen < 0)
|
|
|
2562 {
|
|
|
2563 vim_free(flags);
|
|
|
2564 return follen;
|
|
|
2565 }
|
|
|
2566
|
|
|
2567 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
2568 if (flags != NULL && fol != NULL)
|
|
|
2569 set_spell_charflags(flags, flagslen, fol);
|
|
|
2570
|
|
|
2571 vim_free(flags);
|
|
|
2572 vim_free(fol);
|
|
|
2573
|
|
|
2574 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
2575 if ((flags == NULL) != (fol == NULL))
|
|
|
2576 return SP_FORMERROR;
|
|
|
2577 return 0;
|
|
|
2578 }
|
|
|
2579
|
|
|
2580 /*
|
|
|
2581 * Read SN_PREFCOND section.
|
|
|
2582 * Return SP_*ERROR flags.
|
|
|
2583 */
|
|
|
2584 static int
|
|
|
2585 read_prefcond_section(fd, lp)
|
|
|
2586 FILE *fd;
|
|
|
2587 slang_T *lp;
|
|
|
2588 {
|
|
|
2589 int cnt;
|
|
|
2590 int i;
|
|
|
2591 int n;
|
|
|
2592 char_u *p;
|
|
|
2593 char_u buf[MAXWLEN + 1];
|
|
|
2594
|
|
|
2595 /* <prefcondcnt> <prefcond> ... */
|
|
|
2596 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
2597 if (cnt <= 0)
|
|
|
2598 return SP_FORMERROR;
|
|
|
2599
|
|
|
2600 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
2601 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
2602 if (lp->sl_prefprog == NULL)
|
|
495
|
2603 return SP_OTHERERROR;
|
|
483
|
2604 lp->sl_prefixcnt = cnt;
|
|
|
2605
|
|
376
|
2606 for (i = 0; i < cnt; ++i)
|
|
483
|
2607 {
|
|
|
2608 /* <prefcond> : <condlen> <condstr> */
|
|
|
2609 n = getc(fd); /* <condlen> */
|
|
|
2610 if (n < 0 || n >= MAXWLEN)
|
|
|
2611 return SP_FORMERROR;
|
|
|
2612
|
|
|
2613 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
2614 * compile the regexp program used to check for the condition. */
|
|
|
2615 if (n > 0)
|
|
|
2616 {
|
|
|
2617 buf[0] = '^'; /* always match at one position only */
|
|
|
2618 p = buf + 1;
|
|
|
2619 while (n-- > 0)
|
|
|
2620 *p++ = getc(fd); /* <condstr> */
|
|
|
2621 *p = NUL;
|
|
|
2622 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
2623 }
|
|
|
2624 }
|
|
|
2625 return 0;
|
|
|
2626 }
|
|
|
2627
|
|
|
2628 /*
|
|
|
2629 * Read REP items section from "fd": <repcount> <rep> ...
|
|
|
2630 * Return SP_*ERROR flags.
|
|
|
2631 */
|
|
|
2632 static int
|
|
|
2633 read_rep_section(fd, slang)
|
|
|
2634 FILE *fd;
|
|
|
2635 slang_T *slang;
|
|
|
2636 {
|
|
|
2637 int cnt;
|
|
|
2638 garray_T *gap;
|
|
|
2639 fromto_T *ftp;
|
|
|
2640 short *first;
|
|
|
2641 int i;
|
|
|
2642
|
|
|
2643 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
2644 if (cnt < 0)
|
|
|
2645 return SP_TRUNCERROR;
|
|
|
2646
|
|
|
2647 gap = &slang->sl_rep;
|
|
|
2648 if (ga_grow(gap, cnt) == FAIL)
|
|
495
|
2649 return SP_OTHERERROR;
|
|
483
|
2650
|
|
|
2651 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
2652 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2653 {
|
|
|
2654 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
|
2655 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
|
2656 if (i < 0)
|
|
|
2657 return i;
|
|
|
2658 if (i == 0)
|
|
|
2659 return SP_FORMERROR;
|
|
|
2660 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
|
2661 if (i <= 0)
|
|
|
2662 {
|
|
|
2663 vim_free(ftp->ft_from);
|
|
|
2664 if (i < 0)
|
|
|
2665 return i;
|
|
|
2666 return SP_FORMERROR;
|
|
|
2667 }
|
|
|
2668 }
|
|
|
2669
|
|
|
2670 /* Fill the first-index table. */
|
|
|
2671 first = slang->sl_rep_first;
|
|
|
2672 for (i = 0; i < 256; ++i)
|
|
|
2673 first[i] = -1;
|
|
|
2674 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2675 {
|
|
|
2676 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
2677 if (first[*ftp->ft_from] == -1)
|
|
|
2678 first[*ftp->ft_from] = i;
|
|
|
2679 }
|
|
|
2680 return 0;
|
|
|
2681 }
|
|
|
2682
|
|
|
2683 /*
|
|
|
2684 * Read SN_SAL section: <salflags> <salcount> <sal> ...
|
|
|
2685 * Return SP_*ERROR flags.
|
|
|
2686 */
|
|
|
2687 static int
|
|
|
2688 read_sal_section(fd, slang)
|
|
|
2689 FILE *fd;
|
|
|
2690 slang_T *slang;
|
|
|
2691 {
|
|
|
2692 int i;
|
|
|
2693 int cnt;
|
|
|
2694 garray_T *gap;
|
|
|
2695 salitem_T *smp;
|
|
|
2696 int ccnt;
|
|
|
2697 char_u *p;
|
|
485
|
2698 int c = NUL;
|
|
483
|
2699
|
|
|
2700 slang->sl_sofo = FALSE;
|
|
|
2701
|
|
|
2702 i = getc(fd); /* <salflags> */
|
|
|
2703 if (i & SAL_F0LLOWUP)
|
|
|
2704 slang->sl_followup = TRUE;
|
|
|
2705 if (i & SAL_COLLAPSE)
|
|
|
2706 slang->sl_collapse = TRUE;
|
|
|
2707 if (i & SAL_REM_ACCENTS)
|
|
|
2708 slang->sl_rem_accents = TRUE;
|
|
|
2709
|
|
|
2710 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2711 if (cnt < 0)
|
|
|
2712 return SP_TRUNCERROR;
|
|
|
2713
|
|
|
2714 gap = &slang->sl_sal;
|
|
|
2715 ga_init2(gap, sizeof(salitem_T), 10);
|
|
|
2716 if (ga_grow(gap, cnt) == FAIL)
|
|
495
|
2717 return SP_OTHERERROR;
|
|
483
|
2718
|
|
|
2719 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2720 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2721 {
|
|
|
2722 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2723 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2724 if (ccnt < 0)
|
|
|
2725 return SP_TRUNCERROR;
|
|
|
2726 if ((p = alloc(ccnt + 2)) == NULL)
|
|
495
|
2727 return SP_OTHERERROR;
|
|
483
|
2728 smp->sm_lead = p;
|
|
|
2729
|
|
|
2730 /* Read up to the first special char into sm_lead. */
|
|
|
2731 for (i = 0; i < ccnt; ++i)
|
|
|
2732 {
|
|
|
2733 c = getc(fd); /* <salfrom> */
|
|
|
2734 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2735 break;
|
|
|
2736 *p++ = c;
|
|
|
2737 }
|
|
|
2738 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2739 *p++ = NUL;
|
|
|
2740
|
|
|
2741 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2742 if (c == '(')
|
|
|
2743 {
|
|
|
2744 smp->sm_oneof = p;
|
|
|
2745 for (++i; i < ccnt; ++i)
|
|
|
2746 {
|
|
|
2747 c = getc(fd); /* <salfrom> */
|
|
|
2748 if (c == ')')
|
|
|
2749 break;
|
|
|
2750 *p++ = c;
|
|
|
2751 }
|
|
|
2752 *p++ = NUL;
|
|
|
2753 if (++i < ccnt)
|
|
|
2754 c = getc(fd);
|
|
|
2755 }
|
|
|
2756 else
|
|
|
2757 smp->sm_oneof = NULL;
|
|
|
2758
|
|
|
2759 /* Any following chars go in sm_rules. */
|
|
|
2760 smp->sm_rules = p;
|
|
|
2761 if (i < ccnt)
|
|
|
2762 /* store the char we got while checking for end of sm_lead */
|
|
|
2763 *p++ = c;
|
|
|
2764 for (++i; i < ccnt; ++i)
|
|
|
2765 *p++ = getc(fd); /* <salfrom> */
|
|
|
2766 *p++ = NUL;
|
|
|
2767
|
|
|
2768 /* <saltolen> <salto> */
|
|
|
2769 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
|
2770 if (ccnt < 0)
|
|
|
2771 {
|
|
|
2772 vim_free(smp->sm_lead);
|
|
|
2773 return ccnt;
|
|
|
2774 }
|
|
|
2775
|
|
|
2776 #ifdef FEAT_MBYTE
|
|
|
2777 if (has_mbyte)
|
|
|
2778 {
|
|
|
2779 /* convert the multi-byte strings to wide char strings */
|
|
|
2780 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2781 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2782 if (smp->sm_oneof == NULL)
|
|
|
2783 smp->sm_oneof_w = NULL;
|
|
|
2784 else
|
|
|
2785 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
2786 if (smp->sm_to == NULL)
|
|
|
2787 smp->sm_to_w = NULL;
|
|
|
2788 else
|
|
|
2789 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
2790 if (smp->sm_lead_w == NULL
|
|
|
2791 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
2792 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
|
2793 {
|
|
|
2794 vim_free(smp->sm_lead);
|
|
|
2795 vim_free(smp->sm_to);
|
|
|
2796 vim_free(smp->sm_lead_w);
|
|
|
2797 vim_free(smp->sm_oneof_w);
|
|
|
2798 vim_free(smp->sm_to_w);
|
|
495
|
2799 return SP_OTHERERROR;
|
|
483
|
2800 }
|
|
|
2801 }
|
|
|
2802 #endif
|
|
|
2803 }
|
|
|
2804
|
|
|
2805 /* Fill the first-index table. */
|
|
|
2806 set_sal_first(slang);
|
|
|
2807
|
|
|
2808 return 0;
|
|
|
2809 }
|
|
|
2810
|
|
|
2811 /*
|
|
|
2812 * SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
2813 * Return SP_*ERROR flags.
|
|
|
2814 */
|
|
|
2815 static int
|
|
|
2816 read_sofo_section(fd, slang)
|
|
|
2817 FILE *fd;
|
|
|
2818 slang_T *slang;
|
|
|
2819 {
|
|
|
2820 int cnt;
|
|
|
2821 char_u *from, *to;
|
|
|
2822 int res;
|
|
|
2823
|
|
|
2824 slang->sl_sofo = TRUE;
|
|
|
2825
|
|
|
2826 /* <sofofromlen> <sofofrom> */
|
|
|
2827 from = read_cnt_string(fd, 2, &cnt);
|
|
|
2828 if (cnt < 0)
|
|
|
2829 return cnt;
|
|
|
2830
|
|
|
2831 /* <sofotolen> <sofoto> */
|
|
|
2832 to = read_cnt_string(fd, 2, &cnt);
|
|
|
2833 if (cnt < 0)
|
|
|
2834 {
|
|
|
2835 vim_free(from);
|
|
|
2836 return cnt;
|
|
|
2837 }
|
|
|
2838
|
|
|
2839 /* Store the info in slang->sl_sal and/or slang->sl_sal_first. */
|
|
|
2840 if (from != NULL && to != NULL)
|
|
|
2841 res = set_sofo(slang, from, to);
|
|
|
2842 else if (from != NULL || to != NULL)
|
|
|
2843 res = SP_FORMERROR; /* only one of two strings is an error */
|
|
|
2844 else
|
|
|
2845 res = 0;
|
|
|
2846
|
|
|
2847 vim_free(from);
|
|
|
2848 vim_free(to);
|
|
|
2849 return res;
|
|
|
2850 }
|
|
|
2851
|
|
|
2852 /*
|
|
|
2853 * Read the compound section from the .spl file:
|
|
|
2854 * <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
2855 * Returns SP_*ERROR flags.
|
|
|
2856 */
|
|
|
2857 static int
|
|
|
2858 read_compound(fd, slang, len)
|
|
|
2859 FILE *fd;
|
|
|
2860 slang_T *slang;
|
|
|
2861 int len;
|
|
|
2862 {
|
|
|
2863 int todo = len;
|
|
|
2864 int c;
|
|
|
2865 int atstart;
|
|
|
2866 char_u *pat;
|
|
|
2867 char_u *pp;
|
|
|
2868 char_u *cp;
|
|
485
|
2869 char_u *ap;
|
|
483
|
2870
|
|
|
2871 if (todo < 2)
|
|
|
2872 return SP_FORMERROR; /* need at least two bytes */
|
|
|
2873
|
|
|
2874 --todo;
|
|
|
2875 c = getc(fd); /* <compmax> */
|
|
|
2876 if (c < 2)
|
|
|
2877 c = MAXWLEN;
|
|
|
2878 slang->sl_compmax = c;
|
|
|
2879
|
|
|
2880 --todo;
|
|
|
2881 c = getc(fd); /* <compminlen> */
|
|
|
2882 if (c < 1)
|
|
|
2883 c = 3;
|
|
|
2884 slang->sl_compminlen = c;
|
|
|
2885
|
|
|
2886 --todo;
|
|
|
2887 c = getc(fd); /* <compsylmax> */
|
|
|
2888 if (c < 1)
|
|
|
2889 c = MAXWLEN;
|
|
|
2890 slang->sl_compsylmax = c;
|
|
|
2891
|
|
|
2892 /* Turn the COMPOUNDFLAGS items into a regexp pattern:
|
|
|
2893 * "a[bc]/a*b+" -> "^\(a[bc]\|a*b\+\)$".
|
|
495
|
2894 * Inserting backslashes may double the length, "^\(\)$<Nul>" is 7 bytes.
|
|
|
2895 * Conversion to utf-8 may double the size. */
|
|
|
2896 c = todo * 2 + 7;
|
|
|
2897 #ifdef FEAT_MBYTE
|
|
|
2898 if (enc_utf8)
|
|
|
2899 c += todo * 2;
|
|
|
2900 #endif
|
|
|
2901 pat = alloc((unsigned)c);
|
|
483
|
2902 if (pat == NULL)
|
|
495
|
2903 return SP_OTHERERROR;
|
|
483
|
2904
|
|
485
|
2905 /* We also need a list of all flags that can appear at the start and one
|
|
|
2906 * for all flags. */
|
|
483
|
2907 cp = alloc(todo + 1);
|
|
|
2908 if (cp == NULL)
|
|
|
2909 {
|
|
|
2910 vim_free(pat);
|
|
495
|
2911 return SP_OTHERERROR;
|
|
483
|
2912 }
|
|
|
2913 slang->sl_compstartflags = cp;
|
|
|
2914 *cp = NUL;
|
|
|
2915
|
|
485
|
2916 ap = alloc(todo + 1);
|
|
|
2917 if (ap == NULL)
|
|
|
2918 {
|
|
|
2919 vim_free(pat);
|
|
495
|
2920 return SP_OTHERERROR;
|
|
485
|
2921 }
|
|
|
2922 slang->sl_compallflags = ap;
|
|
|
2923 *ap = NUL;
|
|
|
2924
|
|
483
|
2925 pp = pat;
|
|
|
2926 *pp++ = '^';
|
|
|
2927 *pp++ = '\\';
|
|
|
2928 *pp++ = '(';
|
|
|
2929
|
|
|
2930 atstart = 1;
|
|
|
2931 while (todo-- > 0)
|
|
|
2932 {
|
|
|
2933 c = getc(fd); /* <compflags> */
|
|
485
|
2934
|
|
|
2935 /* Add all flags to "sl_compallflags". */
|
|
|
2936 if (vim_strchr((char_u *)"+*[]/", c) == NULL
|
|
495
|
2937 && !byte_in_str(slang->sl_compallflags, c))
|
|
485
|
2938 {
|
|
|
2939 *ap++ = c;
|
|
|
2940 *ap = NUL;
|
|
|
2941 }
|
|
|
2942
|
|
483
|
2943 if (atstart != 0)
|
|
|
2944 {
|
|
|
2945 /* At start of item: copy flags to "sl_compstartflags". For a
|
|
|
2946 * [abc] item set "atstart" to 2 and copy up to the ']'. */
|
|
|
2947 if (c == '[')
|
|
|
2948 atstart = 2;
|
|
|
2949 else if (c == ']')
|
|
|
2950 atstart = 0;
|
|
|
2951 else
|
|
|
2952 {
|
|
495
|
2953 if (!byte_in_str(slang->sl_compstartflags, c))
|
|
483
|
2954 {
|
|
|
2955 *cp++ = c;
|
|
|
2956 *cp = NUL;
|
|
|
2957 }
|
|
|
2958 if (atstart == 1)
|
|
|
2959 atstart = 0;
|
|
|
2960 }
|
|
|
2961 }
|
|
|
2962 if (c == '/') /* slash separates two items */
|
|
|
2963 {
|
|
|
2964 *pp++ = '\\';
|
|
|
2965 *pp++ = '|';
|
|
|
2966 atstart = 1;
|
|
|
2967 }
|
|
|
2968 else /* normal char, "[abc]" and '*' are copied as-is */
|
|
|
2969 {
|
|
|
2970 if (c == '+')
|
|
|
2971 *pp++ = '\\'; /* "a+" becomes "a\+" */
|
|
495
|
2972 #ifdef FEAT_MBYTE
|
|
|
2973 if (enc_utf8)
|
|
|
2974 pp += mb_char2bytes(c, pp);
|
|
|
2975 else
|
|
|
2976 #endif
|
|
|
2977 *pp++ = c;
|
|
483
|
2978 }
|
|
|
2979 }
|
|
|
2980
|
|
|
2981 *pp++ = '\\';
|
|
|
2982 *pp++ = ')';
|
|
|
2983 *pp++ = '$';
|
|
|
2984 *pp = NUL;
|
|
|
2985
|
|
|
2986 slang->sl_compprog = vim_regcomp(pat, RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
2987 vim_free(pat);
|
|
|
2988 if (slang->sl_compprog == NULL)
|
|
|
2989 return SP_FORMERROR;
|
|
|
2990
|
|
|
2991 return 0;
|
|
|
2992 }
|
|
|
2993
|
|
495
|
2994 /*
|
|
|
2995 * Return TRUE if "byte" appears in "str".
|
|
|
2996 * Like strchr() but independent of locale.
|
|
|
2997 */
|
|
|
2998 static int
|
|
|
2999 byte_in_str(str, byte)
|
|
|
3000 char_u *str;
|
|
|
3001 int byte;
|
|
|
3002 {
|
|
|
3003 char_u *p;
|
|
|
3004
|
|
|
3005 for (p = str; *p != NUL; ++p)
|
|
|
3006 if (*p == byte)
|
|
|
3007 return TRUE;
|
|
|
3008 return FALSE;
|
|
|
3009 }
|
|
|
3010
|
|
483
|
3011 #define SY_MAXLEN 30
|
|
|
3012 typedef struct syl_item_S
|
|
|
3013 {
|
|
|
3014 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */
|
|
|
3015 int sy_len;
|
|
|
3016 } syl_item_T;
|
|
|
3017
|
|
|
3018 /*
|
|
|
3019 * Truncate "slang->sl_syllable" at the first slash and put the following items
|
|
|
3020 * in "slang->sl_syl_items".
|
|
|
3021 */
|
|
|
3022 static int
|
|
|
3023 init_syl_tab(slang)
|
|
|
3024 slang_T *slang;
|
|
|
3025 {
|
|
|
3026 char_u *p;
|
|
|
3027 char_u *s;
|
|
|
3028 int l;
|
|
|
3029 syl_item_T *syl;
|
|
|
3030
|
|
|
3031 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4);
|
|
|
3032 p = vim_strchr(slang->sl_syllable, '/');
|
|
|
3033 while (p != NULL)
|
|
|
3034 {
|
|
|
3035 *p++ = NUL;
|
|
495
|
3036 if (*p == NUL) /* trailing slash */
|
|
483
|
3037 break;
|
|
|
3038 s = p;
|
|
|
3039 p = vim_strchr(p, '/');
|
|
|
3040 if (p == NULL)
|
|
|
3041 l = STRLEN(s);
|
|
|
3042 else
|
|
|
3043 l = p - s;
|
|
|
3044 if (l >= SY_MAXLEN)
|
|
|
3045 return SP_FORMERROR;
|
|
|
3046 if (ga_grow(&slang->sl_syl_items, 1) == FAIL)
|
|
495
|
3047 return SP_OTHERERROR;
|
|
483
|
3048 syl = ((syl_item_T *)slang->sl_syl_items.ga_data)
|
|
|
3049 + slang->sl_syl_items.ga_len++;
|
|
|
3050 vim_strncpy(syl->sy_chars, s, l);
|
|
|
3051 syl->sy_len = l;
|
|
|
3052 }
|
|
|
3053 return OK;
|
|
|
3054 }
|
|
|
3055
|
|
|
3056 /*
|
|
|
3057 * Count the number of syllables in "word".
|
|
|
3058 * When "word" contains spaces the syllables after the last space are counted.
|
|
|
3059 * Returns zero if syllables are not defines.
|
|
|
3060 */
|
|
|
3061 static int
|
|
|
3062 count_syllables(slang, word)
|
|
|
3063 slang_T *slang;
|
|
|
3064 char_u *word;
|
|
|
3065 {
|
|
|
3066 int cnt = 0;
|
|
|
3067 int skip = FALSE;
|
|
|
3068 char_u *p;
|
|
|
3069 int len;
|
|
|
3070 int i;
|
|
|
3071 syl_item_T *syl;
|
|
|
3072 int c;
|
|
|
3073
|
|
|
3074 if (slang->sl_syllable == NULL)
|
|
|
3075 return 0;
|
|
|
3076
|
|
|
3077 for (p = word; *p != NUL; p += len)
|
|
|
3078 {
|
|
|
3079 /* When running into a space reset counter. */
|
|
|
3080 if (*p == ' ')
|
|
|
3081 {
|
|
|
3082 len = 1;
|
|
|
3083 cnt = 0;
|
|
|
3084 continue;
|
|
|
3085 }
|
|
|
3086
|
|
|
3087 /* Find longest match of syllable items. */
|
|
|
3088 len = 0;
|
|
|
3089 for (i = 0; i < slang->sl_syl_items.ga_len; ++i)
|
|
|
3090 {
|
|
|
3091 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i;
|
|
|
3092 if (syl->sy_len > len
|
|
|
3093 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0)
|
|
|
3094 len = syl->sy_len;
|
|
|
3095 }
|
|
|
3096 if (len != 0) /* found a match, count syllable */
|
|
|
3097 {
|
|
|
3098 ++cnt;
|
|
|
3099 skip = FALSE;
|
|
|
3100 }
|
|
|
3101 else
|
|
|
3102 {
|
|
|
3103 /* No recognized syllable item, at least a syllable char then? */
|
|
|
3104 #ifdef FEAT_MBYTE
|
|
|
3105 c = mb_ptr2char(p);
|
|
|
3106 len = (*mb_ptr2len)(p);
|
|
|
3107 #else
|
|
|
3108 c = *p;
|
|
|
3109 len = 1;
|
|
|
3110 #endif
|
|
|
3111 if (vim_strchr(slang->sl_syllable, c) == NULL)
|
|
|
3112 skip = FALSE; /* No, search for next syllable */
|
|
|
3113 else if (!skip)
|
|
|
3114 {
|
|
|
3115 ++cnt; /* Yes, count it */
|
|
|
3116 skip = TRUE; /* don't count following syllable chars */
|
|
|
3117 }
|
|
|
3118 }
|
|
|
3119 }
|
|
|
3120 return cnt;
|
|
376
|
3121 }
|
|
|
3122
|
|
381
|
3123 /*
|
|
|
3124 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
483
|
3125 * Returns SP_*ERROR flags when there is something wrong.
|
|
381
|
3126 */
|
|
|
3127 static int
|
|
|
3128 set_sofo(lp, from, to)
|
|
|
3129 slang_T *lp;
|
|
|
3130 char_u *from;
|
|
|
3131 char_u *to;
|
|
|
3132 {
|
|
|
3133 int i;
|
|
|
3134
|
|
|
3135 #ifdef FEAT_MBYTE
|
|
|
3136 garray_T *gap;
|
|
|
3137 char_u *s;
|
|
|
3138 char_u *p;
|
|
|
3139 int c;
|
|
|
3140 int *inp;
|
|
|
3141
|
|
|
3142 if (has_mbyte)
|
|
|
3143 {
|
|
|
3144 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
3145 * characters. The index is the low byte of the character.
|
|
|
3146 * The list contains from-to pairs with a terminating NUL.
|
|
|
3147 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
3148 gap = &lp->sl_sal;
|
|
|
3149 ga_init2(gap, sizeof(int *), 1);
|
|
|
3150 if (ga_grow(gap, 256) == FAIL)
|
|
495
|
3151 return SP_OTHERERROR;
|
|
381
|
3152 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
3153 gap->ga_len = 256;
|
|
|
3154
|
|
|
3155 /* First count the number of items for each list. Temporarily use
|
|
|
3156 * sl_sal_first[] for this. */
|
|
|
3157 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3158 {
|
|
474
|
3159 c = mb_cptr2char_adv(&p);
|
|
|
3160 mb_cptr_adv(s);
|
|
381
|
3161 if (c >= 256)
|
|
|
3162 ++lp->sl_sal_first[c & 0xff];
|
|
|
3163 }
|
|
|
3164 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
483
|
3165 return SP_FORMERROR;
|
|
381
|
3166
|
|
|
3167 /* Allocate the lists. */
|
|
|
3168 for (i = 0; i < 256; ++i)
|
|
|
3169 if (lp->sl_sal_first[i] > 0)
|
|
|
3170 {
|
|
|
3171 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
3172 if (p == NULL)
|
|
495
|
3173 return SP_OTHERERROR;
|
|
381
|
3174 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
3175 *(int *)p = 0;
|
|
|
3176 }
|
|
|
3177
|
|
|
3178 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
3179 * list. */
|
|
|
3180 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
3181 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3182 {
|
|
474
|
3183 c = mb_cptr2char_adv(&p);
|
|
|
3184 i = mb_cptr2char_adv(&s);
|
|
381
|
3185 if (c >= 256)
|
|
|
3186 {
|
|
|
3187 /* Append the from-to chars at the end of the list with
|
|
|
3188 * the low byte. */
|
|
|
3189 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
3190 while (*inp != 0)
|
|
|
3191 ++inp;
|
|
|
3192 *inp++ = c; /* from char */
|
|
|
3193 *inp++ = i; /* to char */
|
|
|
3194 *inp++ = NUL; /* NUL at the end */
|
|
|
3195 }
|
|
|
3196 else
|
|
|
3197 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
3198 lp->sl_sal_first[c] = i;
|
|
|
3199 }
|
|
|
3200 }
|
|
|
3201 else
|
|
|
3202 #endif
|
|
|
3203 {
|
|
|
3204 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
3205 if (STRLEN(from) != STRLEN(to))
|
|
483
|
3206 return SP_FORMERROR;
|
|
381
|
3207
|
|
|
3208 for (i = 0; to[i] != NUL; ++i)
|
|
|
3209 lp->sl_sal_first[from[i]] = to[i];
|
|
|
3210 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
3211 }
|
|
|
3212
|
|
483
|
3213 return 0;
|
|
381
|
3214 }
|
|
|
3215
|
|
|
3216 /*
|
|
|
3217 * Fill the first-index table for "lp".
|
|
|
3218 */
|
|
|
3219 static void
|
|
|
3220 set_sal_first(lp)
|
|
|
3221 slang_T *lp;
|
|
|
3222 {
|
|
|
3223 salfirst_T *sfirst;
|
|
|
3224 int i;
|
|
|
3225 salitem_T *smp;
|
|
|
3226 int c;
|
|
|
3227 garray_T *gap = &lp->sl_sal;
|
|
|
3228
|
|
|
3229 sfirst = lp->sl_sal_first;
|
|
|
3230 for (i = 0; i < 256; ++i)
|
|
|
3231 sfirst[i] = -1;
|
|
|
3232 smp = (salitem_T *)gap->ga_data;
|
|
|
3233 for (i = 0; i < gap->ga_len; ++i)
|
|
|
3234 {
|
|
|
3235 #ifdef FEAT_MBYTE
|
|
|
3236 if (has_mbyte)
|
|
|
3237 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
3238 * the character, for other encodings it should differ for most
|
|
|
3239 * characters. */
|
|
|
3240 c = *smp[i].sm_lead_w & 0xff;
|
|
|
3241 else
|
|
|
3242 #endif
|
|
|
3243 c = *smp[i].sm_lead;
|
|
|
3244 if (sfirst[c] == -1)
|
|
|
3245 {
|
|
|
3246 sfirst[c] = i;
|
|
|
3247 #ifdef FEAT_MBYTE
|
|
|
3248 if (has_mbyte)
|
|
|
3249 {
|
|
|
3250 int n;
|
|
|
3251
|
|
|
3252 /* Make sure all entries with this byte are following each
|
|
|
3253 * other. Move the ones that are in the wrong position. Do
|
|
|
3254 * keep the same ordering! */
|
|
|
3255 while (i + 1 < gap->ga_len
|
|
|
3256 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
3257 /* Skip over entry with same index byte. */
|
|
|
3258 ++i;
|
|
|
3259
|
|
|
3260 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
3261 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
3262 {
|
|
|
3263 salitem_T tsal;
|
|
|
3264
|
|
|
3265 /* Move entry with same index byte after the entries
|
|
|
3266 * we already found. */
|
|
|
3267 ++i;
|
|
|
3268 --n;
|
|
|
3269 tsal = smp[i + n];
|
|
|
3270 mch_memmove(smp + i + 1, smp + i,
|
|
|
3271 sizeof(salitem_T) * n);
|
|
|
3272 smp[i] = tsal;
|
|
|
3273 }
|
|
|
3274 }
|
|
|
3275 #endif
|
|
|
3276 }
|
|
|
3277 }
|
|
|
3278 }
|
|
376
|
3279
|
|
372
|
3280 #ifdef FEAT_MBYTE
|
|
|
3281 /*
|
|
|
3282 * Turn a multi-byte string into a wide character string.
|
|
|
3283 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
3284 */
|
|
|
3285 static int *
|
|
|
3286 mb_str2wide(s)
|
|
|
3287 char_u *s;
|
|
|
3288 {
|
|
|
3289 int *res;
|
|
|
3290 char_u *p;
|
|
|
3291 int i = 0;
|
|
|
3292
|
|
|
3293 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
3294 if (res != NULL)
|
|
|
3295 {
|
|
|
3296 for (p = s; *p != NUL; )
|
|
|
3297 res[i++] = mb_ptr2char_adv(&p);
|
|
|
3298 res[i] = NUL;
|
|
|
3299 }
|
|
|
3300 return res;
|
|
|
3301 }
|
|
|
3302 #endif
|
|
|
3303
|
|
236
|
3304 /*
|
|
300
|
3305 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
3306 * "byts" and index array "idxs". Recursively read the children.
|
|
|
3307 *
|
|
346
|
3308 * NOTE: The code here must match put_node().
|
|
300
|
3309 *
|
|
|
3310 * Returns the index follosing the siblings.
|
|
|
3311 * Returns -1 if the file is shorter than expected.
|
|
|
3312 * Returns -2 if there is a format error.
|
|
236
|
3313 */
|
|
324
|
3314 static idx_T
|
|
339
|
3315 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
3316 FILE *fd;
|
|
|
3317 char_u *byts;
|
|
324
|
3318 idx_T *idxs;
|
|
300
|
3319 int maxidx; /* size of arrays */
|
|
324
|
3320 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
3321 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
3322 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
3323 {
|
|
300
|
3324 int len;
|
|
|
3325 int i;
|
|
|
3326 int n;
|
|
324
|
3327 idx_T idx = startidx;
|
|
300
|
3328 int c;
|
|
366
|
3329 int c2;
|
|
300
|
3330 #define SHARED_MASK 0x8000000
|
|
236
|
3331
|
|
300
|
3332 len = getc(fd); /* <siblingcount> */
|
|
|
3333 if (len <= 0)
|
|
|
3334 return -1;
|
|
|
3335
|
|
|
3336 if (startidx + len >= maxidx)
|
|
|
3337 return -2;
|
|
|
3338 byts[idx++] = len;
|
|
|
3339
|
|
|
3340 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
3341 for (i = 1; i <= len; ++i)
|
|
236
|
3342 {
|
|
300
|
3343 c = getc(fd); /* <byte> */
|
|
|
3344 if (c < 0)
|
|
|
3345 return -1;
|
|
|
3346 if (c <= BY_SPECIAL)
|
|
|
3347 {
|
|
366
|
3348 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
3349 {
|
|
|
3350 /* No flags, all regions. */
|
|
|
3351 idxs[idx] = 0;
|
|
|
3352 c = 0;
|
|
|
3353 }
|
|
390
|
3354 else if (c != BY_INDEX)
|
|
300
|
3355 {
|
|
339
|
3356 if (prefixtree)
|
|
|
3357 {
|
|
455
|
3358 /* Read the optional pflags byte, the prefix ID and the
|
|
|
3359 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
3360 * byte, the condition index shifted up 8 bits, the flags
|
|
|
3361 * shifted up 24 bits. */
|
|
|
3362 if (c == BY_FLAGS)
|
|
|
3363 c = getc(fd) << 24; /* <pflags> */
|
|
|
3364 else
|
|
|
3365 c = 0;
|
|
|
3366
|
|
481
|
3367 c |= getc(fd); /* <affixID> */
|
|
455
|
3368
|
|
339
|
3369 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
3370 if (n >= maxprefcondnr)
|
|
|
3371 return -2;
|
|
455
|
3372 c |= (n << 8);
|
|
339
|
3373 }
|
|
390
|
3374 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
3375 {
|
|
|
3376 /* Read flags and optional region and prefix ID. In
|
|
390
|
3377 * idxs[] the flags go in the low two bytes, region above
|
|
|
3378 * that and prefix ID above the region. */
|
|
|
3379 c2 = c;
|
|
339
|
3380 c = getc(fd); /* <flags> */
|
|
390
|
3381 if (c2 == BY_FLAGS2)
|
|
|
3382 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
3383 if (c & WF_REGION)
|
|
390
|
3384 c = (getc(fd) << 16) + c; /* <region> */
|
|
481
|
3385 if (c & WF_AFX)
|
|
|
3386 c = (getc(fd) << 24) + c; /* <affixID> */
|
|
339
|
3387 }
|
|
|
3388
|
|
300
|
3389 idxs[idx] = c;
|
|
|
3390 c = 0;
|
|
|
3391 }
|
|
|
3392 else /* c == BY_INDEX */
|
|
|
3393 {
|
|
|
3394 /* <nodeidx> */
|
|
|
3395 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
3396 if (n < 0 || n >= maxidx)
|
|
|
3397 return -2;
|
|
|
3398 idxs[idx] = n + SHARED_MASK;
|
|
|
3399 c = getc(fd); /* <xbyte> */
|
|
|
3400 }
|
|
|
3401 }
|
|
|
3402 byts[idx++] = c;
|
|
236
|
3403 }
|
|
|
3404
|
|
300
|
3405 /* Recursively read the children for non-shared siblings.
|
|
|
3406 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
3407 * remove SHARED_MASK) */
|
|
|
3408 for (i = 1; i <= len; ++i)
|
|
|
3409 if (byts[startidx + i] != 0)
|
|
|
3410 {
|
|
|
3411 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
3412 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
3413 else
|
|
|
3414 {
|
|
|
3415 idxs[startidx + i] = idx;
|
|
339
|
3416 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
3417 prefixtree, maxprefcondnr);
|
|
300
|
3418 if (idx < 0)
|
|
|
3419 break;
|
|
|
3420 }
|
|
|
3421 }
|
|
236
|
3422
|
|
300
|
3423 return idx;
|
|
236
|
3424 }
|
|
|
3425
|
|
|
3426 /*
|
|
|
3427 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
3428 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
3429 */
|
|
|
3430 char_u *
|
|
|
3431 did_set_spelllang(buf)
|
|
|
3432 buf_T *buf;
|
|
|
3433 {
|
|
|
3434 garray_T ga;
|
|
351
|
3435 char_u *splp;
|
|
236
|
3436 char_u *region;
|
|
416
|
3437 char_u region_cp[3];
|
|
355
|
3438 int filename;
|
|
236
|
3439 int region_mask;
|
|
|
3440 slang_T *lp;
|
|
|
3441 int c;
|
|
351
|
3442 char_u lang[MAXWLEN + 1];
|
|
323
|
3443 char_u spf_name[MAXPATHL];
|
|
351
|
3444 int len;
|
|
|
3445 char_u *p;
|
|
381
|
3446 int round;
|
|
385
|
3447 char_u *spf;
|
|
389
|
3448 char_u *use_region = NULL;
|
|
|
3449 int dont_use_region = FALSE;
|
|
236
|
3450
|
|
|
3451 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
3452 clear_midword(buf);
|
|
236
|
3453
|
|
351
|
3454 /* loop over comma separated language names. */
|
|
|
3455 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
3456 {
|
|
|
3457 /* Get one language name. */
|
|
|
3458 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
3459
|
|
240
|
3460 region = NULL;
|
|
351
|
3461 len = STRLEN(lang);
|
|
355
|
3462
|
|
|
3463 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
3464 * If there is a region name let "region" point to it and remove it
|
|
|
3465 * from the name. */
|
|
|
3466 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
3467 {
|
|
355
|
3468 filename = TRUE;
|
|
|
3469
|
|
416
|
3470 /* Locate a region and remove it from the file name. */
|
|
|
3471 p = vim_strchr(gettail(lang), '_');
|
|
|
3472 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
3473 && !ASCII_ISALPHA(p[3]))
|
|
|
3474 {
|
|
|
3475 vim_strncpy(region_cp, p + 1, 2);
|
|
|
3476 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
3477 len -= 3;
|
|
|
3478 region = region_cp;
|
|
|
3479 }
|
|
|
3480 else
|
|
|
3481 dont_use_region = TRUE;
|
|
|
3482
|
|
355
|
3483 /* Check if we loaded this language before. */
|
|
|
3484 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3485 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3486 break;
|
|
236
|
3487 }
|
|
355
|
3488 else
|
|
|
3489 {
|
|
|
3490 filename = FALSE;
|
|
|
3491 if (len > 3 && lang[len - 3] == '_')
|
|
|
3492 {
|
|
|
3493 region = lang + len - 2;
|
|
|
3494 len -= 3;
|
|
|
3495 lang[len] = NUL;
|
|
389
|
3496 }
|
|
|
3497 else
|
|
|
3498 dont_use_region = TRUE;
|
|
355
|
3499
|
|
|
3500 /* Check if we loaded this language before. */
|
|
|
3501 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3502 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
3503 break;
|
|
|
3504 }
|
|
236
|
3505
|
|
416
|
3506 if (region != NULL)
|
|
|
3507 {
|
|
|
3508 /* If the region differs from what was used before then don't
|
|
|
3509 * use it for 'spellfile'. */
|
|
|
3510 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
3511 dont_use_region = TRUE;
|
|
|
3512 use_region = region;
|
|
|
3513 }
|
|
|
3514
|
|
351
|
3515 /* If not found try loading the language now. */
|
|
236
|
3516 if (lp == NULL)
|
|
355
|
3517 {
|
|
|
3518 if (filename)
|
|
|
3519 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
3520 else
|
|
|
3521 spell_load_lang(lang);
|
|
|
3522 }
|
|
236
|
3523
|
|
307
|
3524 /*
|
|
351
|
3525 * Loop over the languages, there can be several files for "lang".
|
|
307
|
3526 */
|
|
|
3527 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
3528 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
3529 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
3530 {
|
|
316
|
3531 region_mask = REGION_ALL;
|
|
355
|
3532 if (!filename && region != NULL)
|
|
236
|
3533 {
|
|
307
|
3534 /* find region in sl_regions */
|
|
|
3535 c = find_region(lp->sl_regions, region);
|
|
|
3536 if (c == REGION_ALL)
|
|
|
3537 {
|
|
389
|
3538 if (lp->sl_add)
|
|
|
3539 {
|
|
|
3540 if (*lp->sl_regions != NUL)
|
|
|
3541 /* This addition file is for other regions. */
|
|
|
3542 region_mask = 0;
|
|
|
3543 }
|
|
|
3544 else
|
|
|
3545 /* This is probably an error. Give a warning and
|
|
|
3546 * accept the words anyway. */
|
|
351
|
3547 smsg((char_u *)
|
|
|
3548 _("Warning: region %s not supported"),
|
|
|
3549 region);
|
|
307
|
3550 }
|
|
|
3551 else
|
|
|
3552 region_mask = 1 << c;
|
|
236
|
3553 }
|
|
307
|
3554
|
|
389
|
3555 if (region_mask != 0)
|
|
307
|
3556 {
|
|
389
|
3557 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3558 {
|
|
|
3559 ga_clear(&ga);
|
|
|
3560 return e_outofmem;
|
|
|
3561 }
|
|
|
3562 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
3563 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3564 ++ga.ga_len;
|
|
|
3565 use_midword(lp, buf);
|
|
307
|
3566 }
|
|
385
|
3567 }
|
|
|
3568 }
|
|
|
3569
|
|
|
3570 /* round 0: load int_wordlist, if possible.
|
|
|
3571 * round 1: load first name in 'spellfile'.
|
|
|
3572 * round 2: load second name in 'spellfile.
|
|
|
3573 * etc. */
|
|
|
3574 spf = curbuf->b_p_spf;
|
|
|
3575 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
3576 {
|
|
|
3577 if (round == 0)
|
|
|
3578 {
|
|
|
3579 /* Internal wordlist, if there is one. */
|
|
|
3580 if (int_wordlist == NULL)
|
|
381
|
3581 continue;
|
|
385
|
3582 int_wordlist_spl(spf_name);
|
|
381
|
3583 }
|
|
|
3584 else
|
|
|
3585 {
|
|
385
|
3586 /* One entry in 'spellfile'. */
|
|
|
3587 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
3588 STRCAT(spf_name, ".spl");
|
|
|
3589
|
|
|
3590 /* If it was already found above then skip it. */
|
|
|
3591 for (c = 0; c < ga.ga_len; ++c)
|
|
|
3592 if (fullpathcmp(spf_name,
|
|
|
3593 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
3594 FALSE) == FPC_SAME)
|
|
|
3595 break;
|
|
|
3596 if (c < ga.ga_len)
|
|
381
|
3597 continue;
|
|
|
3598 }
|
|
|
3599
|
|
351
|
3600 /* Check if it was loaded already. */
|
|
323
|
3601 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3602 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3603 break;
|
|
|
3604 if (lp == NULL)
|
|
|
3605 {
|
|
351
|
3606 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
3607 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
3608 * use an arbitrary name. */
|
|
|
3609 if (round == 0)
|
|
|
3610 STRCPY(lang, "internal wordlist");
|
|
|
3611 else
|
|
|
3612 {
|
|
|
3613 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
3614 p = vim_strchr(lang, '.');
|
|
|
3615 if (p != NULL)
|
|
|
3616 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
3617 }
|
|
351
|
3618 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
3619 }
|
|
|
3620 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
3621 {
|
|
389
|
3622 region_mask = REGION_ALL;
|
|
|
3623 if (use_region != NULL && !dont_use_region)
|
|
|
3624 {
|
|
|
3625 /* find region in sl_regions */
|
|
|
3626 c = find_region(lp->sl_regions, use_region);
|
|
|
3627 if (c != REGION_ALL)
|
|
|
3628 region_mask = 1 << c;
|
|
|
3629 else if (*lp->sl_regions != NUL)
|
|
|
3630 /* This spell file is for other regions. */
|
|
|
3631 region_mask = 0;
|
|
|
3632 }
|
|
|
3633
|
|
|
3634 if (region_mask != 0)
|
|
|
3635 {
|
|
|
3636 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
3637 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3638 ++ga.ga_len;
|
|
|
3639 use_midword(lp, buf);
|
|
|
3640 }
|
|
323
|
3641 }
|
|
|
3642 }
|
|
|
3643
|
|
236
|
3644 /* Add a NULL entry to mark the end of the list. */
|
|
|
3645 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3646 {
|
|
|
3647 ga_clear(&ga);
|
|
|
3648 return e_outofmem;
|
|
|
3649 }
|
|
|
3650 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
3651 ++ga.ga_len;
|
|
|
3652
|
|
|
3653 /* Everything is fine, store the new b_langp value. */
|
|
|
3654 ga_clear(&buf->b_langp);
|
|
|
3655 buf->b_langp = ga;
|
|
|
3656
|
|
|
3657 return NULL;
|
|
|
3658 }
|
|
|
3659
|
|
|
3660 /*
|
|
376
|
3661 * Clear the midword characters for buffer "buf".
|
|
|
3662 */
|
|
|
3663 static void
|
|
|
3664 clear_midword(buf)
|
|
|
3665 buf_T *buf;
|
|
|
3666 {
|
|
|
3667 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
3668 #ifdef FEAT_MBYTE
|
|
|
3669 vim_free(buf->b_spell_ismw_mb);
|
|
|
3670 buf->b_spell_ismw_mb = NULL;
|
|
|
3671 #endif
|
|
|
3672 }
|
|
|
3673
|
|
|
3674 /*
|
|
|
3675 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
3676 * They add up to any currently used midword characters.
|
|
|
3677 */
|
|
|
3678 static void
|
|
|
3679 use_midword(lp, buf)
|
|
|
3680 slang_T *lp;
|
|
|
3681 buf_T *buf;
|
|
|
3682 {
|
|
|
3683 char_u *p;
|
|
|
3684
|
|
389
|
3685 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
3686 return;
|
|
|
3687
|
|
376
|
3688 for (p = lp->sl_midword; *p != NUL; )
|
|
|
3689 #ifdef FEAT_MBYTE
|
|
|
3690 if (has_mbyte)
|
|
|
3691 {
|
|
|
3692 int c, l, n;
|
|
|
3693 char_u *bp;
|
|
|
3694
|
|
|
3695 c = mb_ptr2char(p);
|
|
474
|
3696 l = (*mb_ptr2len)(p);
|
|
|
3697 if (c < 256 && l <= 2)
|
|
376
|
3698 buf->b_spell_ismw[c] = TRUE;
|
|
|
3699 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
3700 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
3701 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
3702 else
|
|
|
3703 {
|
|
|
3704 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
3705 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
3706 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
3707 if (bp != NULL)
|
|
|
3708 {
|
|
|
3709 vim_free(buf->b_spell_ismw_mb);
|
|
|
3710 buf->b_spell_ismw_mb = bp;
|
|
|
3711 vim_strncpy(bp + n, p, l);
|
|
|
3712 }
|
|
|
3713 }
|
|
|
3714 p += l;
|
|
|
3715 }
|
|
|
3716 else
|
|
|
3717 #endif
|
|
|
3718 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
3719 }
|
|
|
3720
|
|
|
3721 /*
|
|
236
|
3722 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
3723 * Each region is simply stored as the two characters of it's name.
|
|
381
|
3724 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
3725 */
|
|
|
3726 static int
|
|
|
3727 find_region(rp, region)
|
|
|
3728 char_u *rp;
|
|
|
3729 char_u *region;
|
|
|
3730 {
|
|
|
3731 int i;
|
|
|
3732
|
|
|
3733 for (i = 0; ; i += 2)
|
|
|
3734 {
|
|
|
3735 if (rp[i] == NUL)
|
|
|
3736 return REGION_ALL;
|
|
|
3737 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
3738 break;
|
|
|
3739 }
|
|
|
3740 return i / 2;
|
|
|
3741 }
|
|
|
3742
|
|
|
3743 /*
|
|
323
|
3744 * Return case type of word:
|
|
236
|
3745 * w word 0
|
|
300
|
3746 * Word WF_ONECAP
|
|
|
3747 * W WORD WF_ALLCAP
|
|
|
3748 * WoRd wOrd WF_KEEPCAP
|
|
236
|
3749 */
|
|
|
3750 static int
|
|
|
3751 captype(word, end)
|
|
|
3752 char_u *word;
|
|
323
|
3753 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
3754 {
|
|
|
3755 char_u *p;
|
|
|
3756 int c;
|
|
|
3757 int firstcap;
|
|
|
3758 int allcap;
|
|
|
3759 int past_second = FALSE; /* past second word char */
|
|
|
3760
|
|
|
3761 /* find first letter */
|
|
376
|
3762 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
3763 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
3764 return 0; /* only non-word characters, illegal word */
|
|
|
3765 #ifdef FEAT_MBYTE
|
|
310
|
3766 if (has_mbyte)
|
|
|
3767 c = mb_ptr2char_adv(&p);
|
|
|
3768 else
|
|
236
|
3769 #endif
|
|
310
|
3770 c = *p++;
|
|
324
|
3771 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
3772
|
|
|
3773 /*
|
|
|
3774 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
3775 * But a word with an upper char only at start is a ONECAP.
|
|
|
3776 */
|
|
323
|
3777 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
3778 if (spell_iswordp_nmw(p))
|
|
236
|
3779 {
|
|
455
|
3780 c = PTR2CHAR(p);
|
|
324
|
3781 if (!SPELL_ISUPPER(c))
|
|
236
|
3782 {
|
|
|
3783 /* UUl -> KEEPCAP */
|
|
|
3784 if (past_second && allcap)
|
|
300
|
3785 return WF_KEEPCAP;
|
|
236
|
3786 allcap = FALSE;
|
|
|
3787 }
|
|
|
3788 else if (!allcap)
|
|
|
3789 /* UlU -> KEEPCAP */
|
|
300
|
3790 return WF_KEEPCAP;
|
|
236
|
3791 past_second = TRUE;
|
|
|
3792 }
|
|
|
3793
|
|
|
3794 if (allcap)
|
|
300
|
3795 return WF_ALLCAP;
|
|
236
|
3796 if (firstcap)
|
|
300
|
3797 return WF_ONECAP;
|
|
236
|
3798 return 0;
|
|
|
3799 }
|
|
|
3800
|
|
474
|
3801 /*
|
|
|
3802 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
3803 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
3804 * Add ALLCAP for "WOrD".
|
|
|
3805 */
|
|
|
3806 static int
|
|
|
3807 badword_captype(word, end)
|
|
|
3808 char_u *word;
|
|
|
3809 char_u *end;
|
|
|
3810 {
|
|
|
3811 int flags = captype(word, end);
|
|
475
|
3812 int c;
|
|
474
|
3813 int l, u;
|
|
|
3814 int first;
|
|
|
3815 char_u *p;
|
|
|
3816
|
|
|
3817 if (flags & WF_KEEPCAP)
|
|
|
3818 {
|
|
|
3819 /* Count the number of UPPER and lower case letters. */
|
|
|
3820 l = u = 0;
|
|
|
3821 first = FALSE;
|
|
|
3822 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
3823 {
|
|
475
|
3824 c = PTR2CHAR(p);
|
|
|
3825 if (SPELL_ISUPPER(c))
|
|
474
|
3826 {
|
|
|
3827 ++u;
|
|
|
3828 if (p == word)
|
|
|
3829 first = TRUE;
|
|
|
3830 }
|
|
|
3831 else
|
|
|
3832 ++l;
|
|
|
3833 }
|
|
|
3834
|
|
|
3835 /* If there are more UPPER than lower case letters suggest an
|
|
|
3836 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
3837 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
3838 * require three upper case letters. */
|
|
|
3839 if (u > l && u > 2)
|
|
|
3840 flags |= WF_ALLCAP;
|
|
|
3841 else if (first)
|
|
|
3842 flags |= WF_ONECAP;
|
|
|
3843 }
|
|
|
3844 return flags;
|
|
|
3845 }
|
|
|
3846
|
|
355
|
3847 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
3848 /*
|
|
|
3849 * Free all languages.
|
|
|
3850 */
|
|
|
3851 void
|
|
|
3852 spell_free_all()
|
|
|
3853 {
|
|
|
3854 slang_T *lp;
|
|
|
3855 buf_T *buf;
|
|
385
|
3856 char_u fname[MAXPATHL];
|
|
355
|
3857
|
|
|
3858 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3859 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3860 ga_clear(&buf->b_langp);
|
|
|
3861
|
|
|
3862 while (first_lang != NULL)
|
|
|
3863 {
|
|
|
3864 lp = first_lang;
|
|
|
3865 first_lang = lp->sl_next;
|
|
|
3866 slang_free(lp);
|
|
|
3867 }
|
|
366
|
3868
|
|
385
|
3869 if (int_wordlist != NULL)
|
|
|
3870 {
|
|
|
3871 /* Delete the internal wordlist and its .spl file */
|
|
|
3872 mch_remove(int_wordlist);
|
|
|
3873 int_wordlist_spl(fname);
|
|
|
3874 mch_remove(fname);
|
|
|
3875 vim_free(int_wordlist);
|
|
|
3876 int_wordlist = NULL;
|
|
381
|
3877 }
|
|
|
3878
|
|
366
|
3879 init_spell_chartab();
|
|
355
|
3880 }
|
|
|
3881 # endif
|
|
|
3882
|
|
236
|
3883 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
3884 /*
|
|
|
3885 * Clear all spelling tables and reload them.
|
|
307
|
3886 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
3887 */
|
|
|
3888 void
|
|
|
3889 spell_reload()
|
|
|
3890 {
|
|
|
3891 buf_T *buf;
|
|
316
|
3892 win_T *wp;
|
|
236
|
3893
|
|
358
|
3894 /* Initialize the table for spell_iswordp(). */
|
|
236
|
3895 init_spell_chartab();
|
|
|
3896
|
|
|
3897 /* Unload all allocated memory. */
|
|
355
|
3898 spell_free_all();
|
|
236
|
3899
|
|
|
3900 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3901 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3902 {
|
|
316
|
3903 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
3904 * window for this buffer in which 'spell' is set. */
|
|
236
|
3905 if (*buf->b_p_spl != NUL)
|
|
316
|
3906 {
|
|
|
3907 FOR_ALL_WINDOWS(wp)
|
|
|
3908 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
3909 {
|
|
|
3910 (void)did_set_spelllang(buf);
|
|
|
3911 # ifdef FEAT_WINDOWS
|
|
|
3912 break;
|
|
|
3913 # endif
|
|
|
3914 }
|
|
|
3915 }
|
|
236
|
3916 }
|
|
|
3917 }
|
|
|
3918 # endif
|
|
|
3919
|
|
310
|
3920 /*
|
|
|
3921 * Reload the spell file "fname" if it's loaded.
|
|
|
3922 */
|
|
|
3923 static void
|
|
323
|
3924 spell_reload_one(fname, added_word)
|
|
310
|
3925 char_u *fname;
|
|
323
|
3926 int added_word; /* invoked through "zg" */
|
|
310
|
3927 {
|
|
|
3928 slang_T *lp;
|
|
323
|
3929 int didit = FALSE;
|
|
310
|
3930
|
|
|
3931 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3932 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3933 {
|
|
|
3934 slang_clear(lp);
|
|
323
|
3935 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
3936 redraw_all_later(NOT_VALID);
|
|
323
|
3937 didit = TRUE;
|
|
310
|
3938 }
|
|
323
|
3939
|
|
|
3940 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
3941 * 'spelllang' to get it loaded. */
|
|
|
3942 if (added_word && !didit)
|
|
|
3943 did_set_spelllang(curbuf);
|
|
310
|
3944 }
|
|
|
3945
|
|
|
3946
|
|
236
|
3947 /*
|
|
|
3948 * Functions for ":mkspell".
|
|
|
3949 */
|
|
|
3950
|
|
300
|
3951 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
3952 and .dic file. */
|
|
|
3953 /*
|
|
|
3954 * Main structure to store the contents of a ".aff" file.
|
|
|
3955 */
|
|
|
3956 typedef struct afffile_S
|
|
|
3957 {
|
|
|
3958 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
495
|
3959 int af_flagtype; /* AFT_CHAR, AFT_2CHAR, AFT_NUMBER or AFT_HUH */
|
|
481
|
3960 int af_slash; /* character used in word for slash */
|
|
495
|
3961 unsigned af_rar; /* RAR ID for rare word */
|
|
|
3962 unsigned af_kep; /* KEP ID for keep-case word */
|
|
|
3963 unsigned af_bad; /* BAD ID for banned word */
|
|
|
3964 unsigned af_needaffix; /* NEEDAFFIX ID */
|
|
339
|
3965 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
3966 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
3967 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
495
|
3968 hashtab_T af_comp; /* hashtable for compound flags, compitem_T */
|
|
236
|
3969 } afffile_T;
|
|
|
3970
|
|
495
|
3971 #define AFT_CHAR 0 /* flags are one character */
|
|
|
3972 #define AFT_2CHAR 1 /* flags are two characters */
|
|
|
3973 #define AFT_HUH 2 /* flags are one or two characters */
|
|
|
3974 #define AFT_NUMBER 3 /* flags are numbers, comma separated */
|
|
|
3975
|
|
236
|
3976 typedef struct affentry_S affentry_T;
|
|
|
3977 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
3978 struct affentry_S
|
|
|
3979 {
|
|
|
3980 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
3981 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3982 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3983 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3984 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
483
|
3985 char_u ae_rare; /* rare affix */
|
|
|
3986 char_u ae_nocomp; /* word with affix not compoundable */
|
|
300
|
3987 };
|
|
|
3988
|
|
495
|
3989 #ifdef FEAT_MBYTE
|
|
|
3990 # define AH_KEY_LEN 17 /* 2 x 8 bytes + NUL */
|
|
|
3991 #else
|
|
|
3992 # define AH_KEY_LEN 3 /* 2 x 1 byte + NUL */
|
|
|
3993 #endif
|
|
455
|
3994
|
|
300
|
3995 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
3996 typedef struct affheader_S
|
|
|
3997 {
|
|
495
|
3998 char_u ah_key[AH_KEY_LEN]; /* key for hashtab == name of affix */
|
|
|
3999 unsigned ah_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4000 int ah_newID; /* prefix ID after renumbering; 0 if not used */
|
|
300
|
4001 int ah_combine; /* suffix may combine with prefix */
|
|
|
4002 affentry_T *ah_first; /* first affix entry */
|
|
|
4003 } affheader_T;
|
|
|
4004
|
|
|
4005 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
4006
|
|
495
|
4007 /* Flag used in compound items. */
|
|
|
4008 typedef struct compitem_S
|
|
|
4009 {
|
|
|
4010 char_u ci_key[AH_KEY_LEN]; /* key for hashtab == name of compound */
|
|
|
4011 unsigned ci_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4012 int ci_newID; /* affix ID after renumbering. */
|
|
|
4013 } compitem_T;
|
|
|
4014
|
|
|
4015 #define HI2CI(hi) ((compitem_T *)(hi)->hi_key)
|
|
|
4016
|
|
300
|
4017 /*
|
|
|
4018 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
4019 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
4020 * once after ":mkspell" is done.
|
|
300
|
4021 */
|
|
|
4022 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
4023 typedef struct sblock_S sblock_T;
|
|
|
4024 struct sblock_S
|
|
|
4025 {
|
|
|
4026 sblock_T *sb_next; /* next block in list */
|
|
|
4027 int sb_used; /* nr of bytes already in use */
|
|
|
4028 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
4029 };
|
|
|
4030
|
|
|
4031 /*
|
|
300
|
4032 * A node in the tree.
|
|
236
|
4033 */
|
|
300
|
4034 typedef struct wordnode_S wordnode_T;
|
|
|
4035 struct wordnode_S
|
|
236
|
4036 {
|
|
346
|
4037 union /* shared to save space */
|
|
|
4038 {
|
|
469
|
4039 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
4040 int index; /* index in written nodes (valid after first
|
|
|
4041 round) */
|
|
|
4042 } wn_u1;
|
|
|
4043 union /* shared to save space */
|
|
|
4044 {
|
|
|
4045 wordnode_T *next; /* next node with same hash key */
|
|
|
4046 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
4047 } wn_u2;
|
|
300
|
4048 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
4049 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
4050 always sorted) */
|
|
474
|
4051 int wn_refs; /* Nr. of references to this node. Only
|
|
|
4052 relevant for first node in a list of
|
|
|
4053 siblings, in following siblings it is
|
|
|
4054 always one. */
|
|
300
|
4055 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
481
|
4056 char_u wn_affixID; /* when "wn_byte" is NUL: supported/required
|
|
469
|
4057 prefix ID or 0 */
|
|
|
4058 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
4059 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
4060 PREFIXTREE it's the prefcondnr */
|
|
469
|
4061 #ifdef SPELL_PRINTTREE
|
|
|
4062 int wn_nr; /* sequence nr for printing */
|
|
|
4063 #endif
|
|
236
|
4064 };
|
|
|
4065
|
|
390
|
4066 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
4067
|
|
300
|
4068 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
4069
|
|
300
|
4070 /*
|
|
|
4071 * Info used while reading the spell files.
|
|
|
4072 */
|
|
|
4073 typedef struct spellinfo_S
|
|
249
|
4074 {
|
|
300
|
4075 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
4076 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
4077
|
|
300
|
4078 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
4079 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
4080
|
|
339
|
4081 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
4082
|
|
300
|
4083 sblock_T *si_blocks; /* memory blocks used */
|
|
482
|
4084 long si_blocks_cnt; /* memory blocks allocated */
|
|
|
4085 long si_compress_cnt; /* words to add before lowering
|
|
|
4086 compression limit */
|
|
474
|
4087 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
4088 compression, linked by "wn_child" field. */
|
|
482
|
4089 long si_free_count; /* number of nodes in si_first_free */
|
|
474
|
4090 #ifdef SPELL_PRINTTREE
|
|
|
4091 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
4092 #endif
|
|
|
4093
|
|
|
4094
|
|
300
|
4095 int si_ascii; /* handling only ASCII words */
|
|
310
|
4096 int si_add; /* addition file */
|
|
351
|
4097 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
4098 int si_region; /* region mask */
|
|
|
4099 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
4100 int si_memtot; /* runtime memory used */
|
|
310
|
4101 int si_verbose; /* verbose messages */
|
|
474
|
4102 int si_msg_count; /* number of words added since last message */
|
|
316
|
4103 int si_region_count; /* number of regions supported (1 when there
|
|
|
4104 are no regions) */
|
|
483
|
4105 char_u si_region_name[16]; /* region names; used only if
|
|
|
4106 * si_region_count > 1) */
|
|
323
|
4107
|
|
|
4108 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
4109 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
4110 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
4111 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
4112 int si_followup; /* soundsalike: ? */
|
|
|
4113 int si_collapse; /* soundsalike: ? */
|
|
|
4114 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
4115 garray_T si_map; /* MAP info concatenated */
|
|
495
|
4116 char_u *si_midword; /* MIDWORD chars or NULL */
|
|
483
|
4117 int si_compmax; /* max nr of words for compounding */
|
|
481
|
4118 int si_compminlen; /* minimal length for compounding */
|
|
483
|
4119 int si_compsylmax; /* max nr of syllables for compounding */
|
|
481
|
4120 char_u *si_compflags; /* flags used for compounding */
|
|
492
|
4121 char_u si_nobreak; /* NOBREAK */
|
|
483
|
4122 char_u *si_syllable; /* syllable string */
|
|
339
|
4123 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
4124 * prefixes, each stored as a string */
|
|
495
|
4125 int si_newprefID; /* current value for ah_newID */
|
|
|
4126 int si_compID; /* current value for compound ID */
|
|
300
|
4127 } spellinfo_T;
|
|
249
|
4128
|
|
474
|
4129 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
495
|
4130 static unsigned affitem2flag __ARGS((int flagtype, char_u *item, char_u *fname, int lnum));
|
|
|
4131 static unsigned get_affitem __ARGS((int flagtype, char_u **pp));
|
|
|
4132 static void process_compflags __ARGS((spellinfo_T *spin, afffile_T *aff, char_u *compflags));
|
|
|
4133 static int flag_in_afflist __ARGS((int flagtype, char_u *afflist, unsigned flag));
|
|
|
4134 static void aff_check_number __ARGS((int spinval, int affval, char *name));
|
|
|
4135 static void aff_check_string __ARGS((char_u *spinval, char_u *affval, char *name));
|
|
339
|
4136 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
4137 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
4138 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
4139 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
4140 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
4141 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
483
|
4142 static int get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
495
|
4143 static void get_compflags __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
483
|
4144 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
|
4145 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
4146 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
4147 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
4148 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
4149 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
483
|
4150 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist, int need_affix));
|
|
481
|
4151 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID));
|
|
474
|
4152 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
|
4153 static void deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
|
4154 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
4155 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
4156 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
4157 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
474
|
4158 static void write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
4159 static void clear_node __ARGS((wordnode_T *node));
|
|
|
4160 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
4161 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
4162 static void init_spellfile __ARGS((void));
|
|
236
|
4163
|
|
455
|
4164 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
4165 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
4166 * Use a negative number with the lower 8 bits zero. */
|
|
|
4167 #define PFX_FLAGS -256
|
|
390
|
4168
|
|
483
|
4169 /*
|
|
|
4170 * Tunable parameters for when the tree is compressed. See 'mkspellmem'.
|
|
|
4171 */
|
|
|
4172 static long compress_start = 30000; /* memory / SBLOCKSIZE */
|
|
|
4173 static long compress_inc = 100; /* memory / SBLOCKSIZE */
|
|
|
4174 static long compress_added = 500000; /* word count */
|
|
|
4175
|
|
469
|
4176 #ifdef SPELL_PRINTTREE
|
|
|
4177 /*
|
|
|
4178 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
4179 * readable format, so that we can see what happens when adding a word and/or
|
|
|
4180 * compressing the tree.
|
|
|
4181 * Based on code from Olaf Seibert.
|
|
|
4182 */
|
|
|
4183 #define PRINTLINESIZE 1000
|
|
|
4184 #define PRINTWIDTH 6
|
|
|
4185
|
|
|
4186 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
4187 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
4188
|
|
|
4189 static char line1[PRINTLINESIZE];
|
|
|
4190 static char line2[PRINTLINESIZE];
|
|
|
4191 static char line3[PRINTLINESIZE];
|
|
|
4192
|
|
|
4193 static void
|
|
|
4194 spell_clear_flags(wordnode_T *node)
|
|
|
4195 {
|
|
|
4196 wordnode_T *np;
|
|
|
4197
|
|
|
4198 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4199 {
|
|
|
4200 np->wn_u1.index = FALSE;
|
|
|
4201 spell_clear_flags(np->wn_child);
|
|
|
4202 }
|
|
|
4203 }
|
|
|
4204
|
|
|
4205 static void
|
|
|
4206 spell_print_node(wordnode_T *node, int depth)
|
|
|
4207 {
|
|
|
4208 if (node->wn_u1.index)
|
|
|
4209 {
|
|
|
4210 /* Done this node before, print the reference. */
|
|
|
4211 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
4212 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
4213 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4214 msg(line1);
|
|
|
4215 msg(line2);
|
|
|
4216 msg(line3);
|
|
|
4217 }
|
|
|
4218 else
|
|
|
4219 {
|
|
|
4220 node->wn_u1.index = TRUE;
|
|
|
4221
|
|
|
4222 if (node->wn_byte != NUL)
|
|
|
4223 {
|
|
|
4224 if (node->wn_child != NULL)
|
|
|
4225 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
4226 else
|
|
|
4227 /* Cannot happen? */
|
|
|
4228 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
4229 }
|
|
|
4230 else
|
|
|
4231 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
4232
|
|
|
4233 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
4234
|
|
|
4235 if (node->wn_sibling != NULL)
|
|
|
4236 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
4237 else
|
|
|
4238 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4239
|
|
|
4240 if (node->wn_byte == NUL)
|
|
|
4241 {
|
|
|
4242 msg(line1);
|
|
|
4243 msg(line2);
|
|
|
4244 msg(line3);
|
|
|
4245 }
|
|
|
4246
|
|
|
4247 /* do the children */
|
|
|
4248 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
4249 spell_print_node(node->wn_child, depth + 1);
|
|
|
4250
|
|
|
4251 /* do the siblings */
|
|
|
4252 if (node->wn_sibling != NULL)
|
|
|
4253 {
|
|
|
4254 /* get rid of all parent details except | */
|
|
|
4255 STRCPY(line1, line3);
|
|
|
4256 STRCPY(line2, line3);
|
|
|
4257 spell_print_node(node->wn_sibling, depth);
|
|
|
4258 }
|
|
|
4259 }
|
|
|
4260 }
|
|
|
4261
|
|
|
4262 static void
|
|
|
4263 spell_print_tree(wordnode_T *root)
|
|
|
4264 {
|
|
|
4265 if (root != NULL)
|
|
|
4266 {
|
|
|
4267 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
4268 * done. */
|
|
|
4269 spell_clear_flags(root);
|
|
|
4270
|
|
|
4271 /* Recursively print the tree. */
|
|
|
4272 spell_print_node(root, 0);
|
|
|
4273 }
|
|
|
4274 }
|
|
|
4275 #endif /* SPELL_PRINTTREE */
|
|
|
4276
|
|
236
|
4277 /*
|
|
323
|
4278 * Read the affix file "fname".
|
|
316
|
4279 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
4280 */
|
|
|
4281 static afffile_T *
|
|
474
|
4282 spell_read_aff(spin, fname)
|
|
|
4283 spellinfo_T *spin;
|
|
236
|
4284 char_u *fname;
|
|
|
4285 {
|
|
|
4286 FILE *fd;
|
|
|
4287 afffile_T *aff;
|
|
|
4288 char_u rline[MAXLINELEN];
|
|
|
4289 char_u *line;
|
|
|
4290 char_u *pc = NULL;
|
|
334
|
4291 #define MAXITEMCNT 7
|
|
|
4292 char_u *(items[MAXITEMCNT]);
|
|
236
|
4293 int itemcnt;
|
|
|
4294 char_u *p;
|
|
|
4295 int lnum = 0;
|
|
|
4296 affheader_T *cur_aff = NULL;
|
|
495
|
4297 int did_postpone_prefix = FALSE;
|
|
236
|
4298 int aff_todo = 0;
|
|
|
4299 hashtab_T *tp;
|
|
255
|
4300 char_u *low = NULL;
|
|
|
4301 char_u *fol = NULL;
|
|
|
4302 char_u *upp = NULL;
|
|
323
|
4303 int do_rep;
|
|
|
4304 int do_sal;
|
|
|
4305 int do_map;
|
|
|
4306 int found_map = FALSE;
|
|
324
|
4307 hashitem_T *hi;
|
|
455
|
4308 int l;
|
|
495
|
4309 int compminlen = 0; /* COMPOUNDMIN value */
|
|
|
4310 int compsylmax = 0; /* COMPOUNDSYLMAX value */
|
|
|
4311 int compmax = 0; /* COMPOUNDMAX value */
|
|
|
4312 char_u *compflags = NULL; /* COMPOUNDFLAG and COMPOUNDFLAGS
|
|
|
4313 concatenated */
|
|
|
4314 char_u *midword = NULL; /* MIDWORD value */
|
|
|
4315 char_u *syllable = NULL; /* SYLLABLE value */
|
|
|
4316 char_u *sofofrom = NULL; /* SOFOFROM value */
|
|
|
4317 char_u *sofoto = NULL; /* SOFOTO value */
|
|
236
|
4318
|
|
300
|
4319 /*
|
|
|
4320 * Open the file.
|
|
|
4321 */
|
|
310
|
4322 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4323 if (fd == NULL)
|
|
|
4324 {
|
|
|
4325 EMSG2(_(e_notopen), fname);
|
|
|
4326 return NULL;
|
|
|
4327 }
|
|
|
4328
|
|
310
|
4329 if (spin->si_verbose || p_verbose > 2)
|
|
|
4330 {
|
|
|
4331 if (!spin->si_verbose)
|
|
|
4332 verbose_enter();
|
|
366
|
4333 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
4334 out_flush();
|
|
|
4335 if (!spin->si_verbose)
|
|
|
4336 verbose_leave();
|
|
|
4337 }
|
|
236
|
4338
|
|
323
|
4339 /* Only do REP lines when not done in another .aff file already. */
|
|
|
4340 do_rep = spin->si_rep.ga_len == 0;
|
|
|
4341
|
|
|
4342 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
4343 do_sal = spin->si_sal.ga_len == 0;
|
|
|
4344
|
|
|
4345 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
4346 do_map = spin->si_map.ga_len == 0;
|
|
|
4347
|
|
300
|
4348 /*
|
|
|
4349 * Allocate and init the afffile_T structure.
|
|
|
4350 */
|
|
474
|
4351 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
4352 if (aff == NULL)
|
|
|
4353 return NULL;
|
|
|
4354 hash_init(&aff->af_pref);
|
|
|
4355 hash_init(&aff->af_suff);
|
|
495
|
4356 hash_init(&aff->af_comp);
|
|
236
|
4357
|
|
|
4358 /*
|
|
|
4359 * Read all the lines in the file one by one.
|
|
|
4360 */
|
|
255
|
4361 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
4362 {
|
|
255
|
4363 line_breakcheck();
|
|
236
|
4364 ++lnum;
|
|
|
4365
|
|
|
4366 /* Skip comment lines. */
|
|
|
4367 if (*rline == '#')
|
|
|
4368 continue;
|
|
|
4369
|
|
|
4370 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
4371 vim_free(pc);
|
|
310
|
4372 #ifdef FEAT_MBYTE
|
|
300
|
4373 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4374 {
|
|
300
|
4375 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
4376 if (pc == NULL)
|
|
|
4377 {
|
|
|
4378 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4379 fname, lnum, rline);
|
|
|
4380 continue;
|
|
|
4381 }
|
|
236
|
4382 line = pc;
|
|
|
4383 }
|
|
|
4384 else
|
|
310
|
4385 #endif
|
|
236
|
4386 {
|
|
|
4387 pc = NULL;
|
|
|
4388 line = rline;
|
|
|
4389 }
|
|
|
4390
|
|
|
4391 /* Split the line up in white separated items. Put a NUL after each
|
|
|
4392 * item. */
|
|
|
4393 itemcnt = 0;
|
|
|
4394 for (p = line; ; )
|
|
|
4395 {
|
|
|
4396 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
4397 ++p;
|
|
|
4398 if (*p == NUL)
|
|
|
4399 break;
|
|
334
|
4400 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
4401 break;
|
|
236
|
4402 items[itemcnt++] = p;
|
|
300
|
4403 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
4404 ++p;
|
|
|
4405 if (*p == NUL)
|
|
|
4406 break;
|
|
|
4407 *p++ = NUL;
|
|
|
4408 }
|
|
|
4409
|
|
|
4410 /* Handle non-empty lines. */
|
|
|
4411 if (itemcnt > 0)
|
|
|
4412 {
|
|
|
4413 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
4414 && aff->af_enc == NULL)
|
|
|
4415 {
|
|
310
|
4416 #ifdef FEAT_MBYTE
|
|
300
|
4417 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
4418 aff->af_enc = enc_canonize(items[1]);
|
|
|
4419 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
4420 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
4421 p_enc) == FAIL)
|
|
|
4422 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
4423 fname, aff->af_enc, p_enc);
|
|
375
|
4424 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4425 #else
|
|
|
4426 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4427 #endif
|
|
236
|
4428 }
|
|
495
|
4429 else if (STRCMP(items[0], "FLAG") == 0 && itemcnt == 2
|
|
|
4430 && aff->af_flagtype == AFT_CHAR)
|
|
|
4431 {
|
|
|
4432 if (STRCMP(items[1], "long") == 0)
|
|
|
4433 aff->af_flagtype = AFT_2CHAR;
|
|
|
4434 else if (STRCMP(items[1], "num") == 0)
|
|
|
4435 aff->af_flagtype = AFT_NUMBER;
|
|
|
4436 else if (STRCMP(items[1], "huh") == 0)
|
|
|
4437 aff->af_flagtype = AFT_HUH;
|
|
|
4438 else
|
|
|
4439 smsg((char_u *)_("Invalid value for FLAG in %s line %d: %s"),
|
|
|
4440 fname, lnum, items[1]);
|
|
|
4441 if (aff->af_rar != 0 || aff->af_kep != 0 || aff->af_bad != 0
|
|
|
4442 || aff->af_needaffix != 0 || compflags != NULL
|
|
|
4443 || aff->af_suff.ht_used > 0
|
|
|
4444 || aff->af_pref.ht_used > 0)
|
|
|
4445 smsg((char_u *)_("FLAG after using flags in %s line %d: %s"),
|
|
|
4446 fname, lnum, items[1]);
|
|
|
4447 }
|
|
|
4448 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2
|
|
|
4449 && midword == NULL)
|
|
|
4450 {
|
|
|
4451 midword = getroom_save(spin, items[1]);
|
|
366
|
4452 }
|
|
302
|
4453 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
4454 {
|
|
323
|
4455 /* ignored, we always split */
|
|
302
|
4456 }
|
|
323
|
4457 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
4458 {
|
|
323
|
4459 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
4460 }
|
|
481
|
4461 else if (STRCMP(items[0], "SLASH") == 0 && itemcnt == 2
|
|
|
4462 && aff->af_slash == 0)
|
|
|
4463 {
|
|
|
4464 aff->af_slash = items[1][0];
|
|
|
4465 if (items[1][1] != NUL)
|
|
|
4466 smsg((char_u *)_("Character used for SLASH must be ASCII; in %s line %d: %s"),
|
|
|
4467 fname, lnum, items[1]);
|
|
|
4468 }
|
|
307
|
4469 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
4470 && aff->af_rar == 0)
|
|
|
4471 {
|
|
495
|
4472 aff->af_rar = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4473 fname, lnum);
|
|
307
|
4474 }
|
|
310
|
4475 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
4476 && aff->af_kep == 0)
|
|
307
|
4477 {
|
|
495
|
4478 aff->af_kep = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4479 fname, lnum);
|
|
307
|
4480 }
|
|
346
|
4481 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
4482 && aff->af_bad == 0)
|
|
|
4483 {
|
|
495
|
4484 aff->af_bad = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4485 fname, lnum);
|
|
346
|
4486 }
|
|
483
|
4487 else if (STRCMP(items[0], "NEEDAFFIX") == 0 && itemcnt == 2
|
|
|
4488 && aff->af_needaffix == 0)
|
|
|
4489 {
|
|
495
|
4490 aff->af_needaffix = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4491 fname, lnum);
|
|
483
|
4492 }
|
|
481
|
4493 else if (STRCMP(items[0], "COMPOUNDFLAG") == 0 && itemcnt == 2
|
|
495
|
4494 && compflags == NULL)
|
|
|
4495 {
|
|
|
4496 /* Turn flag "c" into COMPOUNDFLAGS compatible string "c+",
|
|
|
4497 * "Na" into "Na+", "1234" into "1234+". */
|
|
|
4498 p = getroom(spin, STRLEN(items[1]) + 2, FALSE);
|
|
483
|
4499 if (p != NULL)
|
|
|
4500 {
|
|
495
|
4501 STRCPY(p, items[1]);
|
|
|
4502 STRCAT(p, "+");
|
|
|
4503 compflags = p;
|
|
|
4504 }
|
|
481
|
4505 }
|
|
483
|
4506 else if (STRCMP(items[0], "COMPOUNDFLAGS") == 0 && itemcnt == 2)
|
|
|
4507 {
|
|
|
4508 /* Concatenate this string to previously defined ones, using a
|
|
|
4509 * slash to separate them. */
|
|
|
4510 l = STRLEN(items[1]) + 1;
|
|
495
|
4511 if (compflags != NULL)
|
|
|
4512 l += STRLEN(compflags) + 1;
|
|
483
|
4513 p = getroom(spin, l, FALSE);
|
|
|
4514 if (p != NULL)
|
|
|
4515 {
|
|
495
|
4516 if (compflags != NULL)
|
|
483
|
4517 {
|
|
495
|
4518 STRCPY(p, compflags);
|
|
483
|
4519 STRCAT(p, "/");
|
|
|
4520 }
|
|
|
4521 STRCAT(p, items[1]);
|
|
495
|
4522 compflags = p;
|
|
483
|
4523 }
|
|
|
4524 }
|
|
|
4525 else if (STRCMP(items[0], "COMPOUNDMAX") == 0 && itemcnt == 2
|
|
495
|
4526 && compmax == 0)
|
|
|
4527 {
|
|
|
4528 compmax = atoi((char *)items[1]);
|
|
|
4529 if (compmax == 0)
|
|
483
|
4530 smsg((char_u *)_("Wrong COMPOUNDMAX value in %s line %d: %s"),
|
|
|
4531 fname, lnum, items[1]);
|
|
481
|
4532 }
|
|
|
4533 else if (STRCMP(items[0], "COMPOUNDMIN") == 0 && itemcnt == 2
|
|
495
|
4534 && compminlen == 0)
|
|
|
4535 {
|
|
|
4536 compminlen = atoi((char *)items[1]);
|
|
|
4537 if (compminlen == 0)
|
|
481
|
4538 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"),
|
|
|
4539 fname, lnum, items[1]);
|
|
|
4540 }
|
|
483
|
4541 else if (STRCMP(items[0], "COMPOUNDSYLMAX") == 0 && itemcnt == 2
|
|
495
|
4542 && compsylmax == 0)
|
|
|
4543 {
|
|
|
4544 compsylmax = atoi((char *)items[1]);
|
|
|
4545 if (compsylmax == 0)
|
|
483
|
4546 smsg((char_u *)_("Wrong COMPOUNDSYLMAX value in %s line %d: %s"),
|
|
|
4547 fname, lnum, items[1]);
|
|
|
4548 }
|
|
|
4549 else if (STRCMP(items[0], "SYLLABLE") == 0 && itemcnt == 2
|
|
495
|
4550 && syllable == NULL)
|
|
|
4551 {
|
|
|
4552 syllable = getroom_save(spin, items[1]);
|
|
483
|
4553 }
|
|
492
|
4554 else if (STRCMP(items[0], "NOBREAK") == 0 && itemcnt == 1)
|
|
|
4555 {
|
|
|
4556 spin->si_nobreak = TRUE;
|
|
|
4557 }
|
|
339
|
4558 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
4559 {
|
|
|
4560 aff->af_pfxpostpone = TRUE;
|
|
|
4561 }
|
|
236
|
4562 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4563 || STRCMP(items[0], "SFX") == 0)
|
|
|
4564 && aff_todo == 0
|
|
334
|
4565 && itemcnt >= 4)
|
|
236
|
4566 {
|
|
334
|
4567 /* Myspell allows extra text after the item, but that might
|
|
|
4568 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
4569 if (itemcnt > 4 && *items[4] != '#')
|
|
|
4570 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
4571 fname, lnum, items[4]);
|
|
|
4572
|
|
236
|
4573 /* New affix letter. */
|
|
474
|
4574 cur_aff = (affheader_T *)getroom(spin,
|
|
447
|
4575 sizeof(affheader_T), TRUE);
|
|
236
|
4576 if (cur_aff == NULL)
|
|
|
4577 break;
|
|
495
|
4578 cur_aff->ah_flag = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4579 fname, lnum);
|
|
|
4580 if (cur_aff->ah_flag == 0 || STRLEN(items[1]) >= AH_KEY_LEN)
|
|
|
4581 break;
|
|
|
4582 STRCPY(cur_aff->ah_key, items[1]);
|
|
236
|
4583 if (*items[2] == 'Y')
|
|
|
4584 cur_aff->ah_combine = TRUE;
|
|
300
|
4585 else if (*items[2] != 'N')
|
|
236
|
4586 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
4587 fname, lnum, items[2]);
|
|
339
|
4588
|
|
236
|
4589 if (*items[0] == 'P')
|
|
339
|
4590 {
|
|
236
|
4591 tp = &aff->af_pref;
|
|
339
|
4592 if (aff->af_pfxpostpone)
|
|
495
|
4593 {
|
|
|
4594 /* Use a new number in the .spl file later, to be able
|
|
|
4595 * to handle multiple .aff files. */
|
|
|
4596 cur_aff->ah_newID = ++spin->si_newprefID;
|
|
|
4597
|
|
|
4598 /* We only really use ah_newID if the prefix is
|
|
|
4599 * postponed. We know that only after handling all
|
|
|
4600 * the items. */
|
|
|
4601 did_postpone_prefix = FALSE;
|
|
|
4602 }
|
|
339
|
4603 }
|
|
236
|
4604 else
|
|
|
4605 tp = &aff->af_suff;
|
|
300
|
4606 aff_todo = atoi((char *)items[3]);
|
|
324
|
4607 hi = hash_find(tp, cur_aff->ah_key);
|
|
495
|
4608 if (!HASHITEM_EMPTY(hi)
|
|
|
4609 || cur_aff->ah_flag == aff->af_bad
|
|
|
4610 || cur_aff->ah_flag == aff->af_rar
|
|
|
4611 || cur_aff->ah_flag == aff->af_kep
|
|
|
4612 || cur_aff->ah_flag == aff->af_needaffix)
|
|
300
|
4613 {
|
|
236
|
4614 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
4615 fname, lnum, items[1]);
|
|
300
|
4616 aff_todo = 0;
|
|
|
4617 }
|
|
236
|
4618 else
|
|
|
4619 hash_add(tp, cur_aff->ah_key);
|
|
|
4620 }
|
|
|
4621 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4622 || STRCMP(items[0], "SFX") == 0)
|
|
|
4623 && aff_todo > 0
|
|
|
4624 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
4625 && itemcnt >= 5)
|
|
236
|
4626 {
|
|
|
4627 affentry_T *aff_entry;
|
|
366
|
4628 int rare = FALSE;
|
|
483
|
4629 int nocomp = FALSE;
|
|
455
|
4630 int upper = FALSE;
|
|
366
|
4631 int lasti = 5;
|
|
|
4632
|
|
483
|
4633 /* Check for "rare" and "nocomp" after the other info. */
|
|
|
4634 while (itemcnt > lasti)
|
|
|
4635 {
|
|
|
4636 if (!rare && STRICMP(items[lasti], "rare") == 0)
|
|
|
4637 {
|
|
|
4638 rare = TRUE;
|
|
|
4639 ++lasti;
|
|
|
4640 }
|
|
|
4641 else if (!nocomp && STRICMP(items[lasti], "nocomp") == 0)
|
|
|
4642 {
|
|
|
4643 nocomp = TRUE;
|
|
|
4644 ++lasti;
|
|
|
4645 }
|
|
|
4646 else
|
|
|
4647 break;
|
|
366
|
4648 }
|
|
236
|
4649
|
|
334
|
4650 /* Myspell allows extra text after the item, but that might
|
|
|
4651 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
4652 if (itemcnt > lasti && *items[lasti] != '#')
|
|
482
|
4653 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]);
|
|
334
|
4654
|
|
236
|
4655 /* New item for an affix letter. */
|
|
|
4656 --aff_todo;
|
|
474
|
4657 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
4658 sizeof(affentry_T), TRUE);
|
|
236
|
4659 if (aff_entry == NULL)
|
|
|
4660 break;
|
|
366
|
4661 aff_entry->ae_rare = rare;
|
|
483
|
4662 aff_entry->ae_nocomp = nocomp;
|
|
240
|
4663
|
|
236
|
4664 if (STRCMP(items[2], "0") != 0)
|
|
474
|
4665 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
4666 if (STRCMP(items[3], "0") != 0)
|
|
474
|
4667 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
4668
|
|
300
|
4669 /* Don't use an affix entry with non-ASCII characters when
|
|
|
4670 * "spin->si_ascii" is TRUE. */
|
|
|
4671 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
4672 || has_non_ascii(aff_entry->ae_add)))
|
|
|
4673 {
|
|
|
4674 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
4675 cur_aff->ah_first = aff_entry;
|
|
300
|
4676
|
|
|
4677 if (STRCMP(items[4], ".") != 0)
|
|
|
4678 {
|
|
|
4679 char_u buf[MAXLINELEN];
|
|
|
4680
|
|
474
|
4681 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
4682 if (*items[0] == 'P')
|
|
|
4683 sprintf((char *)buf, "^%s", items[4]);
|
|
|
4684 else
|
|
|
4685 sprintf((char *)buf, "%s$", items[4]);
|
|
|
4686 aff_entry->ae_prog = vim_regcomp(buf,
|
|
481
|
4687 RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
4688 if (aff_entry->ae_prog == NULL)
|
|
|
4689 smsg((char_u *)_("Broken condition in %s line %d: %s"),
|
|
|
4690 fname, lnum, items[4]);
|
|
300
|
4691 }
|
|
339
|
4692
|
|
|
4693 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
4694 * for the condition. Use an existing one if possible. */
|
|
455
|
4695 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
4696 {
|
|
455
|
4697 /* When the chop string is one lower-case letter and
|
|
|
4698 * the add string ends in the upper-case letter we set
|
|
|
4699 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
4700 * letters from "ae_add". The condition must either
|
|
|
4701 * be empty or start with the same letter. */
|
|
|
4702 if (aff_entry->ae_chop != NULL
|
|
|
4703 && aff_entry->ae_add != NULL
|
|
|
4704 #ifdef FEAT_MBYTE
|
|
474
|
4705 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
4706 aff_entry->ae_chop)] == NUL
|
|
|
4707 #else
|
|
|
4708 && aff_entry->ae_chop[1] == NUL
|
|
|
4709 #endif
|
|
|
4710 )
|
|
339
|
4711 {
|
|
455
|
4712 int c, c_up;
|
|
|
4713
|
|
|
4714 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
4715 c_up = SPELL_TOUPPER(c);
|
|
|
4716 if (c_up != c
|
|
|
4717 && (aff_entry->ae_cond == NULL
|
|
|
4718 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
4719 {
|
|
|
4720 p = aff_entry->ae_add
|
|
|
4721 + STRLEN(aff_entry->ae_add);
|
|
|
4722 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
4723 if (PTR2CHAR(p) == c_up)
|
|
|
4724 {
|
|
|
4725 upper = TRUE;
|
|
|
4726 aff_entry->ae_chop = NULL;
|
|
|
4727 *p = NUL;
|
|
|
4728
|
|
|
4729 /* The condition is matched with the
|
|
|
4730 * actual word, thus must check for the
|
|
|
4731 * upper-case letter. */
|
|
|
4732 if (aff_entry->ae_cond != NULL)
|
|
|
4733 {
|
|
|
4734 char_u buf[MAXLINELEN];
|
|
|
4735 #ifdef FEAT_MBYTE
|
|
|
4736 if (has_mbyte)
|
|
|
4737 {
|
|
|
4738 onecap_copy(items[4], buf, TRUE);
|
|
|
4739 aff_entry->ae_cond = getroom_save(
|
|
474
|
4740 spin, buf);
|
|
455
|
4741 }
|
|
|
4742 else
|
|
|
4743 #endif
|
|
|
4744 *aff_entry->ae_cond = c_up;
|
|
|
4745 if (aff_entry->ae_cond != NULL)
|
|
|
4746 {
|
|
|
4747 sprintf((char *)buf, "^%s",
|
|
339
|
4748 aff_entry->ae_cond);
|
|
455
|
4749 vim_free(aff_entry->ae_prog);
|
|
|
4750 aff_entry->ae_prog = vim_regcomp(
|
|
|
4751 buf, RE_MAGIC + RE_STRING);
|
|
|
4752 }
|
|
|
4753 }
|
|
|
4754 }
|
|
|
4755 }
|
|
339
|
4756 }
|
|
|
4757
|
|
455
|
4758 if (aff_entry->ae_chop == NULL)
|
|
390
|
4759 {
|
|
455
|
4760 int idx;
|
|
|
4761 char_u **pp;
|
|
|
4762 int n;
|
|
|
4763
|
|
495
|
4764 /* Find a previously used condition. */
|
|
455
|
4765 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
4766 --idx)
|
|
|
4767 {
|
|
|
4768 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
4769 if (str_equal(p, aff_entry->ae_cond))
|
|
|
4770 break;
|
|
|
4771 }
|
|
|
4772 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
4773 {
|
|
|
4774 /* Not found, add a new condition. */
|
|
|
4775 idx = spin->si_prefcond.ga_len++;
|
|
|
4776 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
4777 + idx;
|
|
|
4778 if (aff_entry->ae_cond == NULL)
|
|
|
4779 *pp = NULL;
|
|
|
4780 else
|
|
474
|
4781 *pp = getroom_save(spin,
|
|
455
|
4782 aff_entry->ae_cond);
|
|
|
4783 }
|
|
|
4784
|
|
|
4785 /* Add the prefix to the prefix tree. */
|
|
|
4786 if (aff_entry->ae_add == NULL)
|
|
|
4787 p = (char_u *)"";
|
|
390
|
4788 else
|
|
455
|
4789 p = aff_entry->ae_add;
|
|
|
4790 /* PFX_FLAGS is a negative number, so that
|
|
|
4791 * tree_add_word() knows this is the prefix tree. */
|
|
|
4792 n = PFX_FLAGS;
|
|
|
4793 if (rare)
|
|
|
4794 n |= WFP_RARE;
|
|
|
4795 if (!cur_aff->ah_combine)
|
|
|
4796 n |= WFP_NC;
|
|
|
4797 if (upper)
|
|
|
4798 n |= WFP_UP;
|
|
474
|
4799 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
4800 idx, cur_aff->ah_newID);
|
|
495
|
4801 did_postpone_prefix = TRUE;
|
|
|
4802 }
|
|
|
4803
|
|
|
4804 /* Didn't actually use ah_newID, backup si_newprefID. */
|
|
|
4805 if (aff_todo == 0 && !did_postpone_prefix)
|
|
|
4806 {
|
|
|
4807 --spin->si_newprefID;
|
|
|
4808 cur_aff->ah_newID = 0;
|
|
390
|
4809 }
|
|
339
|
4810 }
|
|
240
|
4811 }
|
|
236
|
4812 }
|
|
495
|
4813 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2
|
|
|
4814 && fol == NULL)
|
|
|
4815 {
|
|
|
4816 fol = vim_strsave(items[1]);
|
|
|
4817 }
|
|
|
4818 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2
|
|
|
4819 && low == NULL)
|
|
|
4820 {
|
|
|
4821 low = vim_strsave(items[1]);
|
|
|
4822 }
|
|
|
4823 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2
|
|
|
4824 && upp == NULL)
|
|
|
4825 {
|
|
|
4826 upp = vim_strsave(items[1]);
|
|
255
|
4827 }
|
|
236
|
4828 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
4829 {
|
|
236
|
4830 /* Ignore REP count */;
|
|
323
|
4831 if (!isdigit(*items[1]))
|
|
|
4832 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
4833 fname, lnum);
|
|
|
4834 }
|
|
482
|
4835 else if (STRCMP(items[0], "REP") == 0 && itemcnt >= 3)
|
|
236
|
4836 {
|
|
|
4837 /* REP item */
|
|
482
|
4838 /* Myspell ignores extra arguments, we require it starts with
|
|
|
4839 * # to detect mistakes. */
|
|
|
4840 if (itemcnt > 3 && items[3][0] != '#')
|
|
|
4841 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]);
|
|
323
|
4842 if (do_rep)
|
|
|
4843 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
4844 }
|
|
|
4845 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
4846 {
|
|
|
4847 /* MAP item or count */
|
|
|
4848 if (!found_map)
|
|
|
4849 {
|
|
|
4850 /* First line contains the count. */
|
|
|
4851 found_map = TRUE;
|
|
|
4852 if (!isdigit(*items[1]))
|
|
|
4853 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
4854 fname, lnum);
|
|
|
4855 }
|
|
|
4856 else if (do_map)
|
|
|
4857 {
|
|
346
|
4858 int c;
|
|
|
4859
|
|
|
4860 /* Check that every character appears only once. */
|
|
|
4861 for (p = items[1]; *p != NUL; )
|
|
|
4862 {
|
|
|
4863 #ifdef FEAT_MBYTE
|
|
|
4864 c = mb_ptr2char_adv(&p);
|
|
|
4865 #else
|
|
|
4866 c = *p++;
|
|
|
4867 #endif
|
|
|
4868 if ((spin->si_map.ga_len > 0
|
|
|
4869 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
4870 != NULL)
|
|
|
4871 || vim_strchr(p, c) != NULL)
|
|
|
4872 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
4873 fname, lnum);
|
|
|
4874 }
|
|
|
4875
|
|
323
|
4876 /* We simply concatenate all the MAP strings, separated by
|
|
|
4877 * slashes. */
|
|
|
4878 ga_concat(&spin->si_map, items[1]);
|
|
|
4879 ga_append(&spin->si_map, '/');
|
|
|
4880 }
|
|
|
4881 }
|
|
|
4882 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
4883 {
|
|
|
4884 if (do_sal)
|
|
|
4885 {
|
|
|
4886 /* SAL item (sounds-a-like)
|
|
|
4887 * Either one of the known keys or a from-to pair. */
|
|
|
4888 if (STRCMP(items[1], "followup") == 0)
|
|
|
4889 spin->si_followup = sal_to_bool(items[2]);
|
|
|
4890 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
4891 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
4892 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
4893 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
4894 else
|
|
|
4895 /* when "to" is "_" it means empty */
|
|
|
4896 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
4897 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
4898 : items[2]);
|
|
|
4899 }
|
|
236
|
4900 }
|
|
375
|
4901 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
495
|
4902 && sofofrom == NULL)
|
|
|
4903 {
|
|
|
4904 sofofrom = getroom_save(spin, items[1]);
|
|
375
|
4905 }
|
|
|
4906 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
495
|
4907 && sofoto == NULL)
|
|
|
4908 {
|
|
|
4909 sofoto = getroom_save(spin, items[1]);
|
|
375
|
4910 }
|
|
300
|
4911 else
|
|
481
|
4912 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"),
|
|
236
|
4913 fname, lnum, items[0]);
|
|
|
4914 }
|
|
|
4915 }
|
|
|
4916
|
|
255
|
4917 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
4918 {
|
|
351
|
4919 if (spin->si_clear_chartab)
|
|
|
4920 {
|
|
|
4921 /* Clear the char type tables, don't want to use any of the
|
|
|
4922 * currently used spell properties. */
|
|
|
4923 init_spell_chartab();
|
|
|
4924 spin->si_clear_chartab = FALSE;
|
|
|
4925 }
|
|
|
4926
|
|
316
|
4927 /*
|
|
|
4928 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
4929 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
4930 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
4931 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
4932 */
|
|
|
4933 if (!spin->si_ascii
|
|
|
4934 #ifdef FEAT_MBYTE
|
|
|
4935 && !enc_utf8
|
|
|
4936 #endif
|
|
|
4937 )
|
|
260
|
4938 {
|
|
|
4939 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
4940 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
4941 else
|
|
316
|
4942 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
4943 }
|
|
255
|
4944
|
|
|
4945 vim_free(fol);
|
|
|
4946 vim_free(low);
|
|
|
4947 vim_free(upp);
|
|
|
4948 }
|
|
|
4949
|
|
481
|
4950 /* Use compound specifications of the .aff file for the spell info. */
|
|
495
|
4951 if (compmax != 0)
|
|
|
4952 {
|
|
|
4953 aff_check_number(spin->si_compmax, compmax, "COMPOUNDMAX");
|
|
|
4954 spin->si_compmax = compmax;
|
|
|
4955 }
|
|
|
4956
|
|
|
4957 if (compminlen != 0)
|
|
|
4958 {
|
|
|
4959 aff_check_number(spin->si_compminlen, compminlen, "COMPOUNDMIN");
|
|
|
4960 spin->si_compminlen = compminlen;
|
|
|
4961 }
|
|
|
4962
|
|
|
4963 if (compsylmax != 0)
|
|
|
4964 {
|
|
|
4965 if (syllable == NULL)
|
|
|
4966 smsg((char_u *)_("COMPOUNDSYLMAX used without SYLLABLE"));
|
|
|
4967 aff_check_number(spin->si_compsylmax, compsylmax, "COMPOUNDSYLMAX");
|
|
|
4968 spin->si_compsylmax = compsylmax;
|
|
|
4969 }
|
|
|
4970
|
|
|
4971 if (compflags != NULL)
|
|
|
4972 process_compflags(spin, aff, compflags);
|
|
|
4973
|
|
|
4974 /* Check that we didn't use too many renumbered flags. */
|
|
|
4975 if (spin->si_compID < spin->si_newprefID)
|
|
|
4976 {
|
|
|
4977 if (spin->si_compID == 255)
|
|
|
4978 MSG(_("Too many postponed prefixes"));
|
|
|
4979 else if (spin->si_newprefID == 0)
|
|
|
4980 MSG(_("Too many compound flags"));
|
|
481
|
4981 else
|
|
495
|
4982 MSG(_("Too many posponed prefixes and/or compound flags"));
|
|
|
4983 }
|
|
|
4984
|
|
|
4985 if (syllable != NULL)
|
|
|
4986 {
|
|
|
4987 aff_check_string(spin->si_syllable, syllable, "SYLLABLE");
|
|
|
4988 spin->si_syllable = syllable;
|
|
|
4989 }
|
|
|
4990
|
|
|
4991 if (sofofrom != NULL || sofoto != NULL)
|
|
|
4992 {
|
|
|
4993 if (sofofrom == NULL || sofoto == NULL)
|
|
|
4994 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
4995 sofofrom == NULL ? "FROM" : "TO", fname);
|
|
|
4996 else if (spin->si_sal.ga_len > 0)
|
|
|
4997 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
483
|
4998 else
|
|
495
|
4999 {
|
|
|
5000 aff_check_string(spin->si_sofofr, sofofrom, "SOFOFROM");
|
|
|
5001 aff_check_string(spin->si_sofoto, sofoto, "SOFOTO");
|
|
|
5002 spin->si_sofofr = sofofrom;
|
|
|
5003 spin->si_sofoto = sofoto;
|
|
|
5004 }
|
|
|
5005 }
|
|
|
5006
|
|
|
5007 if (midword != NULL)
|
|
|
5008 {
|
|
|
5009 aff_check_string(spin->si_midword, midword, "MIDWORD");
|
|
|
5010 spin->si_midword = midword;
|
|
483
|
5011 }
|
|
|
5012
|
|
236
|
5013 vim_free(pc);
|
|
|
5014 fclose(fd);
|
|
|
5015 return aff;
|
|
|
5016 }
|
|
|
5017
|
|
|
5018 /*
|
|
495
|
5019 * Turn an affix flag name into a number, according to the FLAG type.
|
|
|
5020 * returns zero for failure.
|
|
|
5021 */
|
|
|
5022 static unsigned
|
|
|
5023 affitem2flag(flagtype, item, fname, lnum)
|
|
|
5024 int flagtype;
|
|
|
5025 char_u *item;
|
|
|
5026 char_u *fname;
|
|
|
5027 int lnum;
|
|
|
5028 {
|
|
|
5029 unsigned res;
|
|
|
5030 char_u *p = item;
|
|
|
5031
|
|
|
5032 res = get_affitem(flagtype, &p);
|
|
|
5033 if (res == 0)
|
|
|
5034 {
|
|
|
5035 if (flagtype == AFT_NUMBER)
|
|
|
5036 smsg((char_u *)_("Flag is not a number in %s line %d: %s"),
|
|
|
5037 fname, lnum, item);
|
|
|
5038 else
|
|
|
5039 smsg((char_u *)_("Illegal flag in %s line %d: %s"),
|
|
|
5040 fname, lnum, item);
|
|
|
5041 }
|
|
|
5042 if (*p != NUL)
|
|
|
5043 {
|
|
|
5044 smsg((char_u *)_(e_affname), fname, lnum, item);
|
|
|
5045 return 0;
|
|
|
5046 }
|
|
|
5047
|
|
|
5048 return res;
|
|
|
5049 }
|
|
|
5050
|
|
|
5051 /*
|
|
|
5052 * Get one affix name from "*pp" and advance the pointer.
|
|
|
5053 * Returns zero for an error, still advances the pointer then.
|
|
|
5054 */
|
|
|
5055 static unsigned
|
|
|
5056 get_affitem(flagtype, pp)
|
|
|
5057 int flagtype;
|
|
|
5058 char_u **pp;
|
|
|
5059 {
|
|
|
5060 int res;
|
|
|
5061
|
|
|
5062 if (flagtype == AFT_NUMBER)
|
|
|
5063 {
|
|
|
5064 if (!VIM_ISDIGIT(**pp))
|
|
|
5065 {
|
|
|
5066 ++*pp;
|
|
|
5067 return 0;
|
|
|
5068 }
|
|
|
5069 res = getdigits(pp);
|
|
|
5070 }
|
|
|
5071 else
|
|
|
5072 {
|
|
|
5073 #ifdef FEAT_MBYTE
|
|
|
5074 res = mb_ptr2char_adv(pp);
|
|
|
5075 #else
|
|
|
5076 res = *(*pp)++;
|
|
|
5077 #endif
|
|
|
5078 if (flagtype == AFT_2CHAR || (flagtype == AFT_HUH
|
|
|
5079 && res >= 'A' && res <= 'Z'))
|
|
|
5080 {
|
|
|
5081 if (**pp == NUL)
|
|
|
5082 return 0;
|
|
|
5083 #ifdef FEAT_MBYTE
|
|
|
5084 res = mb_ptr2char_adv(pp) + (res << 16);
|
|
|
5085 #else
|
|
|
5086 res = *(*pp)++ + (res << 16);
|
|
|
5087 #endif
|
|
|
5088 }
|
|
|
5089 }
|
|
|
5090 return res;
|
|
|
5091 }
|
|
|
5092
|
|
|
5093 /*
|
|
|
5094 * Process the "compflags" string used in an affix file and append it to
|
|
|
5095 * spin->si_compflags.
|
|
|
5096 * The processing involves changing the affix names to ID numbers, so that
|
|
|
5097 * they fit in one byte.
|
|
|
5098 */
|
|
|
5099 static void
|
|
|
5100 process_compflags(spin, aff, compflags)
|
|
|
5101 spellinfo_T *spin;
|
|
|
5102 afffile_T *aff;
|
|
|
5103 char_u *compflags;
|
|
|
5104 {
|
|
|
5105 char_u *p;
|
|
|
5106 char_u *prevp;
|
|
|
5107 unsigned flag;
|
|
|
5108 compitem_T *ci;
|
|
|
5109 int id;
|
|
|
5110 int len;
|
|
|
5111 char_u *tp;
|
|
|
5112 char_u key[AH_KEY_LEN];
|
|
|
5113 hashitem_T *hi;
|
|
|
5114
|
|
|
5115 /* Make room for the old and the new compflags, concatenated with a / in
|
|
|
5116 * between. Processing it makes it shorter, but we don't know by how
|
|
|
5117 * much, thus allocate the maximum. */
|
|
|
5118 len = STRLEN(compflags) + 1;
|
|
|
5119 if (spin->si_compflags != NULL)
|
|
|
5120 len += STRLEN(spin->si_compflags) + 1;
|
|
|
5121 p = getroom(spin, len, FALSE);
|
|
|
5122 if (p == NULL)
|
|
|
5123 return;
|
|
|
5124 if (spin->si_compflags != NULL)
|
|
|
5125 {
|
|
|
5126 STRCPY(p, spin->si_compflags);
|
|
|
5127 STRCAT(p, "/");
|
|
|
5128 }
|
|
|
5129 else
|
|
|
5130 *p = NUL;
|
|
|
5131 spin->si_compflags = p;
|
|
|
5132 tp = p + STRLEN(p);
|
|
|
5133
|
|
|
5134 for (p = compflags; *p != NUL; )
|
|
|
5135 {
|
|
|
5136 if (vim_strchr((char_u *)"/*+[]", *p) != NULL)
|
|
|
5137 /* Copy non-flag characters directly. */
|
|
|
5138 *tp++ = *p++;
|
|
|
5139 else
|
|
|
5140 {
|
|
|
5141 /* First get the flag number, also checks validity. */
|
|
|
5142 prevp = p;
|
|
|
5143 flag = get_affitem(aff->af_flagtype, &p);
|
|
|
5144 if (flag != 0)
|
|
|
5145 {
|
|
|
5146 /* Find the flag in the hashtable. If it was used before, use
|
|
|
5147 * the existing ID. Otherwise add a new entry. */
|
|
|
5148 vim_strncpy(key, prevp, p - prevp);
|
|
|
5149 hi = hash_find(&aff->af_comp, key);
|
|
|
5150 if (!HASHITEM_EMPTY(hi))
|
|
|
5151 id = HI2CI(hi)->ci_newID;
|
|
|
5152 else
|
|
|
5153 {
|
|
|
5154 ci = (compitem_T *)getroom(spin, sizeof(compitem_T), TRUE);
|
|
|
5155 if (ci == NULL)
|
|
|
5156 break;
|
|
|
5157 STRCPY(ci->ci_key, key);
|
|
|
5158 ci->ci_flag = flag;
|
|
|
5159 /* Avoid using a flag ID that has a special meaning in a
|
|
|
5160 * regexp (also inside []). */
|
|
|
5161 do
|
|
|
5162 {
|
|
|
5163 id = spin->si_compID--;
|
|
|
5164 } while (vim_strchr((char_u *)"/*+[]\\-^", id) != NULL);
|
|
|
5165 ci->ci_newID = id;
|
|
|
5166 hash_add(&aff->af_comp, ci->ci_key);
|
|
|
5167 }
|
|
|
5168 *tp++ = id;
|
|
|
5169 }
|
|
|
5170 if (aff->af_flagtype == AFT_NUMBER && *p == ',')
|
|
|
5171 ++p;
|
|
|
5172 }
|
|
|
5173 }
|
|
|
5174
|
|
|
5175 *tp = NUL;
|
|
|
5176 }
|
|
|
5177
|
|
|
5178 /*
|
|
|
5179 * Return TRUE if flag "flag" appears in affix list "afflist".
|
|
|
5180 */
|
|
|
5181 static int
|
|
|
5182 flag_in_afflist(flagtype, afflist, flag)
|
|
|
5183 int flagtype;
|
|
|
5184 char_u *afflist;
|
|
|
5185 unsigned flag;
|
|
|
5186 {
|
|
|
5187 char_u *p;
|
|
|
5188 unsigned n;
|
|
|
5189
|
|
|
5190 switch (flagtype)
|
|
|
5191 {
|
|
|
5192 case AFT_CHAR:
|
|
|
5193 return vim_strchr(afflist, flag) != NULL;
|
|
|
5194
|
|
|
5195 case AFT_HUH:
|
|
|
5196 case AFT_2CHAR:
|
|
|
5197 for (p = afflist; *p != NUL; )
|
|
|
5198 {
|
|
|
5199 #ifdef FEAT_MBYTE
|
|
|
5200 n = mb_ptr2char_adv(&p);
|
|
|
5201 #else
|
|
|
5202 n = *p++;
|
|
|
5203 #endif
|
|
|
5204 if ((flagtype == AFT_2CHAR || (n >= 'A' && n <= 'Z'))
|
|
|
5205 && *p != NUL)
|
|
|
5206 #ifdef FEAT_MBYTE
|
|
|
5207 n = mb_ptr2char_adv(&p) + (n << 16);
|
|
|
5208 #else
|
|
|
5209 n = *p++ + (n << 16);
|
|
|
5210 #endif
|
|
|
5211 if (n == flag)
|
|
|
5212 return TRUE;
|
|
|
5213 }
|
|
|
5214 break;
|
|
|
5215
|
|
|
5216 case AFT_NUMBER:
|
|
|
5217 for (p = afflist; *p != NUL; )
|
|
|
5218 {
|
|
|
5219 n = getdigits(&p);
|
|
|
5220 if (n == flag)
|
|
|
5221 return TRUE;
|
|
|
5222 if (*p != NUL) /* skip over comma */
|
|
|
5223 ++p;
|
|
|
5224 }
|
|
|
5225 break;
|
|
|
5226 }
|
|
|
5227 return FALSE;
|
|
|
5228 }
|
|
|
5229
|
|
|
5230 /*
|
|
|
5231 * Give a warning when "spinval" and "affval" numbers are set and not the same.
|
|
|
5232 */
|
|
|
5233 static void
|
|
|
5234 aff_check_number(spinval, affval, name)
|
|
|
5235 int spinval;
|
|
|
5236 int affval;
|
|
|
5237 char *name;
|
|
|
5238 {
|
|
|
5239 if (spinval != 0 && spinval != affval)
|
|
|
5240 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
5241 }
|
|
|
5242
|
|
|
5243 /*
|
|
|
5244 * Give a warning when "spinval" and "affval" strings are set and not the same.
|
|
|
5245 */
|
|
|
5246 static void
|
|
|
5247 aff_check_string(spinval, affval, name)
|
|
|
5248 char_u *spinval;
|
|
|
5249 char_u *affval;
|
|
|
5250 char *name;
|
|
|
5251 {
|
|
|
5252 if (spinval != NULL && STRCMP(spinval, affval) != 0)
|
|
|
5253 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
5254 }
|
|
|
5255
|
|
|
5256 /*
|
|
339
|
5257 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
5258 * NULL as equal.
|
|
|
5259 */
|
|
|
5260 static int
|
|
|
5261 str_equal(s1, s2)
|
|
|
5262 char_u *s1;
|
|
|
5263 char_u *s2;
|
|
|
5264 {
|
|
|
5265 if (s1 == NULL || s2 == NULL)
|
|
|
5266 return s1 == s2;
|
|
|
5267 return STRCMP(s1, s2) == 0;
|
|
|
5268 }
|
|
|
5269
|
|
|
5270 /*
|
|
323
|
5271 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
5272 * They are stored case-folded.
|
|
|
5273 */
|
|
|
5274 static void
|
|
|
5275 add_fromto(spin, gap, from, to)
|
|
|
5276 spellinfo_T *spin;
|
|
|
5277 garray_T *gap;
|
|
|
5278 char_u *from;
|
|
|
5279 char_u *to;
|
|
|
5280 {
|
|
|
5281 fromto_T *ftp;
|
|
|
5282 char_u word[MAXWLEN];
|
|
|
5283
|
|
|
5284 if (ga_grow(gap, 1) == OK)
|
|
|
5285 {
|
|
|
5286 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
5287 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
5288 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
5289 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
5290 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
5291 ++gap->ga_len;
|
|
|
5292 }
|
|
|
5293 }
|
|
|
5294
|
|
|
5295 /*
|
|
|
5296 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
5297 */
|
|
|
5298 static int
|
|
|
5299 sal_to_bool(s)
|
|
|
5300 char_u *s;
|
|
|
5301 {
|
|
|
5302 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
5303 }
|
|
|
5304
|
|
|
5305 /*
|
|
240
|
5306 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
5307 * When "s" is NULL FALSE is returned.
|
|
|
5308 */
|
|
|
5309 static int
|
|
|
5310 has_non_ascii(s)
|
|
|
5311 char_u *s;
|
|
|
5312 {
|
|
|
5313 char_u *p;
|
|
|
5314
|
|
|
5315 if (s != NULL)
|
|
|
5316 for (p = s; *p != NUL; ++p)
|
|
|
5317 if (*p >= 128)
|
|
|
5318 return TRUE;
|
|
|
5319 return FALSE;
|
|
|
5320 }
|
|
|
5321
|
|
|
5322 /*
|
|
236
|
5323 * Free the structure filled by spell_read_aff().
|
|
|
5324 */
|
|
|
5325 static void
|
|
|
5326 spell_free_aff(aff)
|
|
|
5327 afffile_T *aff;
|
|
|
5328 {
|
|
|
5329 hashtab_T *ht;
|
|
|
5330 hashitem_T *hi;
|
|
|
5331 int todo;
|
|
|
5332 affheader_T *ah;
|
|
300
|
5333 affentry_T *ae;
|
|
236
|
5334
|
|
|
5335 vim_free(aff->af_enc);
|
|
|
5336
|
|
339
|
5337 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
5338 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
5339 {
|
|
|
5340 todo = ht->ht_used;
|
|
|
5341 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
5342 {
|
|
|
5343 if (!HASHITEM_EMPTY(hi))
|
|
|
5344 {
|
|
|
5345 --todo;
|
|
|
5346 ah = HI2AH(hi);
|
|
300
|
5347 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
5348 vim_free(ae->ae_prog);
|
|
236
|
5349 }
|
|
|
5350 }
|
|
|
5351 if (ht == &aff->af_suff)
|
|
|
5352 break;
|
|
|
5353 }
|
|
300
|
5354
|
|
236
|
5355 hash_clear(&aff->af_pref);
|
|
|
5356 hash_clear(&aff->af_suff);
|
|
495
|
5357 hash_clear(&aff->af_comp);
|
|
236
|
5358 }
|
|
|
5359
|
|
|
5360 /*
|
|
300
|
5361 * Read dictionary file "fname".
|
|
236
|
5362 * Returns OK or FAIL;
|
|
|
5363 */
|
|
|
5364 static int
|
|
474
|
5365 spell_read_dic(spin, fname, affile)
|
|
|
5366 spellinfo_T *spin;
|
|
236
|
5367 char_u *fname;
|
|
300
|
5368 afffile_T *affile;
|
|
236
|
5369 {
|
|
300
|
5370 hashtab_T ht;
|
|
236
|
5371 char_u line[MAXLINELEN];
|
|
481
|
5372 char_u *p;
|
|
300
|
5373 char_u *afflist;
|
|
483
|
5374 char_u store_afflist[MAXWLEN];
|
|
|
5375 int pfxlen;
|
|
|
5376 int need_affix;
|
|
300
|
5377 char_u *dw;
|
|
236
|
5378 char_u *pc;
|
|
|
5379 char_u *w;
|
|
|
5380 int l;
|
|
|
5381 hash_T hash;
|
|
|
5382 hashitem_T *hi;
|
|
|
5383 FILE *fd;
|
|
|
5384 int lnum = 1;
|
|
300
|
5385 int non_ascii = 0;
|
|
|
5386 int retval = OK;
|
|
|
5387 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
5388 int flags;
|
|
481
|
5389 int duplicate = 0;
|
|
236
|
5390
|
|
300
|
5391 /*
|
|
|
5392 * Open the file.
|
|
|
5393 */
|
|
310
|
5394 fd = mch_fopen((char *)fname, "r");
|
|
236
|
5395 if (fd == NULL)
|
|
|
5396 {
|
|
|
5397 EMSG2(_(e_notopen), fname);
|
|
|
5398 return FAIL;
|
|
|
5399 }
|
|
|
5400
|
|
300
|
5401 /* The hashtable is only used to detect duplicated words. */
|
|
|
5402 hash_init(&ht);
|
|
|
5403
|
|
310
|
5404 if (spin->si_verbose || p_verbose > 2)
|
|
|
5405 {
|
|
|
5406 if (!spin->si_verbose)
|
|
|
5407 verbose_enter();
|
|
366
|
5408 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
5409 out_flush();
|
|
|
5410 if (!spin->si_verbose)
|
|
|
5411 verbose_leave();
|
|
|
5412 }
|
|
236
|
5413
|
|
474
|
5414 /* start with a message for the first line */
|
|
|
5415 spin->si_msg_count = 999999;
|
|
|
5416
|
|
236
|
5417 /* Read and ignore the first line: word count. */
|
|
|
5418 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
5419 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
5420 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
5421
|
|
|
5422 /*
|
|
|
5423 * Read all the lines in the file one by one.
|
|
|
5424 * The words are converted to 'encoding' here, before being added to
|
|
|
5425 * the hashtable.
|
|
|
5426 */
|
|
255
|
5427 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
5428 {
|
|
255
|
5429 line_breakcheck();
|
|
236
|
5430 ++lnum;
|
|
455
|
5431 if (line[0] == '#' || line[0] == '/')
|
|
351
|
5432 continue; /* comment line */
|
|
236
|
5433
|
|
300
|
5434 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
5435 * the word is kept to allow e.g., "et al.". */
|
|
236
|
5436 l = STRLEN(line);
|
|
|
5437 while (l > 0 && line[l - 1] <= ' ')
|
|
|
5438 --l;
|
|
|
5439 if (l == 0)
|
|
|
5440 continue; /* empty line */
|
|
|
5441 line[l] = NUL;
|
|
|
5442
|
|
481
|
5443 /* Find the optional affix names. Replace the SLASH character by a
|
|
|
5444 * slash. */
|
|
|
5445 afflist = NULL;
|
|
|
5446 for (p = line; *p != NUL; mb_ptr_adv(p))
|
|
|
5447 {
|
|
|
5448 if (*p == affile->af_slash)
|
|
|
5449 *p = '/';
|
|
|
5450 else if (*p == '/')
|
|
|
5451 {
|
|
|
5452 *p = NUL;
|
|
|
5453 afflist = p + 1;
|
|
|
5454 break;
|
|
|
5455 }
|
|
|
5456 }
|
|
236
|
5457
|
|
300
|
5458 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
5459 if (spin->si_ascii && has_non_ascii(line))
|
|
|
5460 {
|
|
|
5461 ++non_ascii;
|
|
240
|
5462 continue;
|
|
300
|
5463 }
|
|
240
|
5464
|
|
310
|
5465 #ifdef FEAT_MBYTE
|
|
236
|
5466 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
5467 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
5468 {
|
|
300
|
5469 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
5470 if (pc == NULL)
|
|
|
5471 {
|
|
|
5472 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5473 fname, lnum, line);
|
|
|
5474 continue;
|
|
|
5475 }
|
|
236
|
5476 w = pc;
|
|
|
5477 }
|
|
|
5478 else
|
|
310
|
5479 #endif
|
|
236
|
5480 {
|
|
|
5481 pc = NULL;
|
|
|
5482 w = line;
|
|
|
5483 }
|
|
|
5484
|
|
474
|
5485 /* This takes time, print a message every 10000 words. */
|
|
|
5486 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
5487 {
|
|
|
5488 spin->si_msg_count = 0;
|
|
339
|
5489 vim_snprintf((char *)message, sizeof(message),
|
|
|
5490 _("line %6d, word %6d - %s"),
|
|
|
5491 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
5492 msg_start();
|
|
|
5493 msg_puts_long_attr(message, 0);
|
|
|
5494 msg_clr_eos();
|
|
|
5495 msg_didout = FALSE;
|
|
|
5496 msg_col = 0;
|
|
|
5497 out_flush();
|
|
|
5498 }
|
|
|
5499
|
|
300
|
5500 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
5501 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
5502 if (dw == NULL)
|
|
300
|
5503 retval = FAIL;
|
|
|
5504 vim_free(pc);
|
|
|
5505 if (retval == FAIL)
|
|
236
|
5506 break;
|
|
|
5507
|
|
300
|
5508 hash = hash_hash(dw);
|
|
|
5509 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
5510 if (!HASHITEM_EMPTY(hi))
|
|
481
|
5511 {
|
|
|
5512 if (p_verbose > 0)
|
|
|
5513 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
5514 fname, lnum, dw);
|
|
481
|
5515 else if (duplicate == 0)
|
|
|
5516 smsg((char_u *)_("First duplicate word in %s line %d: %s"),
|
|
|
5517 fname, lnum, dw);
|
|
|
5518 ++duplicate;
|
|
|
5519 }
|
|
236
|
5520 else
|
|
300
|
5521 hash_add_item(&ht, hi, dw, hash);
|
|
|
5522
|
|
307
|
5523 flags = 0;
|
|
483
|
5524 store_afflist[0] = NUL;
|
|
|
5525 pfxlen = 0;
|
|
|
5526 need_affix = FALSE;
|
|
307
|
5527 if (afflist != NULL)
|
|
|
5528 {
|
|
|
5529 /* Check for affix name that stands for keep-case word and stands
|
|
|
5530 * for rare word (if defined). */
|
|
495
|
5531 if (affile->af_kep != 0 && flag_in_afflist(
|
|
|
5532 affile->af_flagtype, afflist, affile->af_kep))
|
|
389
|
5533 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
495
|
5534 if (affile->af_rar != 0 && flag_in_afflist(
|
|
|
5535 affile->af_flagtype, afflist, affile->af_rar))
|
|
307
|
5536 flags |= WF_RARE;
|
|
495
|
5537 if (affile->af_bad != 0 && flag_in_afflist(
|
|
|
5538 affile->af_flagtype, afflist, affile->af_bad))
|
|
346
|
5539 flags |= WF_BANNED;
|
|
495
|
5540 if (affile->af_needaffix != 0 && flag_in_afflist(
|
|
|
5541 affile->af_flagtype, afflist, affile->af_needaffix))
|
|
483
|
5542 need_affix = TRUE;
|
|
339
|
5543
|
|
|
5544 if (affile->af_pfxpostpone)
|
|
|
5545 /* Need to store the list of prefix IDs with the word. */
|
|
483
|
5546 pfxlen = get_pfxlist(affile, afflist, store_afflist);
|
|
|
5547
|
|
|
5548 if (spin->si_compflags != NULL)
|
|
|
5549 /* Need to store the list of compound flags with the word.
|
|
|
5550 * Concatenate them to the list of prefix IDs. */
|
|
495
|
5551 get_compflags(affile, afflist, store_afflist + pfxlen);
|
|
307
|
5552 }
|
|
|
5553
|
|
300
|
5554 /* Add the word to the word tree(s). */
|
|
483
|
5555 if (store_word(spin, dw, flags, spin->si_region,
|
|
|
5556 store_afflist, need_affix) == FAIL)
|
|
300
|
5557 retval = FAIL;
|
|
236
|
5558
|
|
300
|
5559 if (afflist != NULL)
|
|
|
5560 {
|
|
|
5561 /* Find all matching suffixes and add the resulting words.
|
|
|
5562 * Additionally do matching prefixes that combine. */
|
|
474
|
5563 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
5564 &affile->af_suff, &affile->af_pref,
|
|
483
|
5565 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5566 retval = FAIL;
|
|
|
5567
|
|
|
5568 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
5569 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
5570 &affile->af_pref, NULL,
|
|
483
|
5571 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5572 retval = FAIL;
|
|
|
5573 }
|
|
236
|
5574 }
|
|
|
5575
|
|
481
|
5576 if (duplicate > 0)
|
|
|
5577 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname);
|
|
300
|
5578 if (spin->si_ascii && non_ascii > 0)
|
|
481
|
5579 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"),
|
|
|
5580 non_ascii, fname);
|
|
300
|
5581 hash_clear(&ht);
|
|
|
5582
|
|
236
|
5583 fclose(fd);
|
|
300
|
5584 return retval;
|
|
236
|
5585 }
|
|
|
5586
|
|
|
5587 /*
|
|
339
|
5588 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
5589 * Used for PFXPOSTPONE.
|
|
483
|
5590 * Put the resulting flags in "store_afflist[MAXWLEN]" with a terminating NUL
|
|
|
5591 * and return the number of affixes.
|
|
|
5592 */
|
|
|
5593 static int
|
|
|
5594 get_pfxlist(affile, afflist, store_afflist)
|
|
339
|
5595 afffile_T *affile;
|
|
|
5596 char_u *afflist;
|
|
483
|
5597 char_u *store_afflist;
|
|
339
|
5598 {
|
|
|
5599 char_u *p;
|
|
495
|
5600 char_u *prevp;
|
|
483
|
5601 int cnt = 0;
|
|
495
|
5602 int id;
|
|
|
5603 char_u key[AH_KEY_LEN];
|
|
339
|
5604 hashitem_T *hi;
|
|
|
5605
|
|
495
|
5606 for (p = afflist; *p != NUL; )
|
|
|
5607 {
|
|
|
5608 prevp = p;
|
|
|
5609 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
5610 {
|
|
|
5611 /* A flag is a postponed prefix flag if it appears in "af_pref"
|
|
|
5612 * and it's ID is not zero. */
|
|
|
5613 vim_strncpy(key, prevp, p - prevp);
|
|
|
5614 hi = hash_find(&affile->af_pref, key);
|
|
|
5615 if (!HASHITEM_EMPTY(hi))
|
|
|
5616 {
|
|
|
5617 id = HI2AH(hi)->ah_newID;
|
|
|
5618 if (id != 0)
|
|
|
5619 store_afflist[cnt++] = id;
|
|
|
5620 }
|
|
|
5621 }
|
|
|
5622 if (affile->af_flagtype == AFT_NUMBER && *p == ',')
|
|
|
5623 ++p;
|
|
483
|
5624 }
|
|
|
5625
|
|
|
5626 store_afflist[cnt] = NUL;
|
|
|
5627 return cnt;
|
|
339
|
5628 }
|
|
|
5629
|
|
|
5630 /*
|
|
495
|
5631 * Get the list of compound IDs from the affix list "afflist" that are used
|
|
|
5632 * for compound words.
|
|
483
|
5633 * Puts the flags in "store_afflist[]".
|
|
|
5634 */
|
|
|
5635 static void
|
|
495
|
5636 get_compflags(affile, afflist, store_afflist)
|
|
|
5637 afffile_T *affile;
|
|
481
|
5638 char_u *afflist;
|
|
483
|
5639 char_u *store_afflist;
|
|
481
|
5640 {
|
|
|
5641 char_u *p;
|
|
495
|
5642 char_u *prevp;
|
|
483
|
5643 int cnt = 0;
|
|
495
|
5644 char_u key[AH_KEY_LEN];
|
|
|
5645 hashitem_T *hi;
|
|
|
5646
|
|
|
5647 for (p = afflist; *p != NUL; )
|
|
|
5648 {
|
|
|
5649 prevp = p;
|
|
|
5650 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
5651 {
|
|
|
5652 /* A flag is a compound flag if it appears in "af_comp". */
|
|
|
5653 vim_strncpy(key, prevp, p - prevp);
|
|
|
5654 hi = hash_find(&affile->af_comp, key);
|
|
|
5655 if (!HASHITEM_EMPTY(hi))
|
|
|
5656 store_afflist[cnt++] = HI2CI(hi)->ci_newID;
|
|
|
5657 }
|
|
|
5658 if (affile->af_flagtype == AFT_NUMBER && *p == ',')
|
|
|
5659 ++p;
|
|
|
5660 }
|
|
483
|
5661
|
|
|
5662 store_afflist[cnt] = NUL;
|
|
481
|
5663 }
|
|
|
5664
|
|
|
5665 /*
|
|
300
|
5666 * Apply affixes to a word and store the resulting words.
|
|
|
5667 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
5668 * prefixes or suffixes.
|
|
|
5669 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
5670 * the resulting words for combining affixes.
|
|
|
5671 *
|
|
|
5672 * Returns FAIL when out of memory.
|
|
236
|
5673 */
|
|
300
|
5674 static int
|
|
483
|
5675 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags,
|
|
|
5676 pfxlist, pfxlen)
|
|
474
|
5677 spellinfo_T *spin; /* spell info */
|
|
300
|
5678 char_u *word; /* basic word start */
|
|
|
5679 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
5680 afffile_T *affile;
|
|
300
|
5681 hashtab_T *ht;
|
|
|
5682 hashtab_T *xht;
|
|
|
5683 int comb; /* only use affixes that combine */
|
|
307
|
5684 int flags; /* flags for the word */
|
|
339
|
5685 char_u *pfxlist; /* list of prefix IDs */
|
|
483
|
5686 int pfxlen; /* nr of flags in "pfxlist" for prefixes, rest
|
|
|
5687 * is compound flags */
|
|
236
|
5688 {
|
|
|
5689 int todo;
|
|
|
5690 hashitem_T *hi;
|
|
300
|
5691 affheader_T *ah;
|
|
|
5692 affentry_T *ae;
|
|
|
5693 regmatch_T regmatch;
|
|
|
5694 char_u newword[MAXWLEN];
|
|
|
5695 int retval = OK;
|
|
|
5696 int i;
|
|
|
5697 char_u *p;
|
|
366
|
5698 int use_flags;
|
|
390
|
5699 char_u *use_pfxlist;
|
|
483
|
5700 char_u pfx_pfxlist[MAXWLEN];
|
|
|
5701 size_t wordlen = STRLEN(word);
|
|
236
|
5702
|
|
300
|
5703 todo = ht->ht_used;
|
|
|
5704 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
5705 {
|
|
|
5706 if (!HASHITEM_EMPTY(hi))
|
|
|
5707 {
|
|
|
5708 --todo;
|
|
300
|
5709 ah = HI2AH(hi);
|
|
236
|
5710
|
|
300
|
5711 /* Check that the affix combines, if required, and that the word
|
|
|
5712 * supports this affix. */
|
|
495
|
5713 if ((!comb || ah->ah_combine) && flag_in_afflist(
|
|
|
5714 affile->af_flagtype, afflist, ah->ah_flag))
|
|
236
|
5715 {
|
|
300
|
5716 /* Loop over all affix entries with this name. */
|
|
|
5717 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
5718 {
|
|
300
|
5719 /* Check the condition. It's not logical to match case
|
|
|
5720 * here, but it is required for compatibility with
|
|
339
|
5721 * Myspell.
|
|
481
|
5722 * Another requirement from Myspell is that the chop
|
|
|
5723 * string is shorter than the word itself.
|
|
339
|
5724 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
5725 * prefixes with a chop string. */
|
|
300
|
5726 regmatch.regprog = ae->ae_prog;
|
|
|
5727 regmatch.rm_ic = FALSE;
|
|
339
|
5728 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
5729 || ae->ae_chop != NULL)
|
|
481
|
5730 && (ae->ae_chop == NULL
|
|
|
5731 || STRLEN(ae->ae_chop) < wordlen)
|
|
339
|
5732 && (ae->ae_prog == NULL
|
|
|
5733 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
5734 {
|
|
|
5735 /* Match. Remove the chop and add the affix. */
|
|
|
5736 if (xht == NULL)
|
|
240
|
5737 {
|
|
300
|
5738 /* prefix: chop/add at the start of the word */
|
|
|
5739 if (ae->ae_add == NULL)
|
|
|
5740 *newword = NUL;
|
|
|
5741 else
|
|
|
5742 STRCPY(newword, ae->ae_add);
|
|
|
5743 p = word;
|
|
|
5744 if (ae->ae_chop != NULL)
|
|
310
|
5745 {
|
|
300
|
5746 /* Skip chop string. */
|
|
310
|
5747 #ifdef FEAT_MBYTE
|
|
|
5748 if (has_mbyte)
|
|
324
|
5749 {
|
|
310
|
5750 i = mb_charlen(ae->ae_chop);
|
|
324
|
5751 for ( ; i > 0; --i)
|
|
|
5752 mb_ptr_adv(p);
|
|
|
5753 }
|
|
310
|
5754 else
|
|
|
5755 #endif
|
|
324
|
5756 p += STRLEN(ae->ae_chop);
|
|
310
|
5757 }
|
|
300
|
5758 STRCAT(newword, p);
|
|
|
5759 }
|
|
|
5760 else
|
|
|
5761 {
|
|
|
5762 /* suffix: chop/add at the end of the word */
|
|
|
5763 STRCPY(newword, word);
|
|
|
5764 if (ae->ae_chop != NULL)
|
|
|
5765 {
|
|
|
5766 /* Remove chop string. */
|
|
|
5767 p = newword + STRLEN(newword);
|
|
376
|
5768 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
5769 for ( ; i > 0; --i)
|
|
300
|
5770 mb_ptr_back(newword, p);
|
|
|
5771 *p = NUL;
|
|
|
5772 }
|
|
|
5773 if (ae->ae_add != NULL)
|
|
|
5774 STRCAT(newword, ae->ae_add);
|
|
240
|
5775 }
|
|
|
5776
|
|
366
|
5777 /* Obey the "rare" flag of the affix. */
|
|
|
5778 if (ae->ae_rare)
|
|
|
5779 use_flags = flags | WF_RARE;
|
|
|
5780 else
|
|
|
5781 use_flags = flags;
|
|
483
|
5782
|
|
|
5783 /* Obey the "nocomp" flag of the affix: don't use the
|
|
|
5784 * compound flags. */
|
|
390
|
5785 use_pfxlist = pfxlist;
|
|
483
|
5786 if (ae->ae_nocomp && pfxlist != NULL)
|
|
|
5787 {
|
|
|
5788 vim_strncpy(pfx_pfxlist, pfxlist, pfxlen);
|
|
|
5789 use_pfxlist = pfx_pfxlist;
|
|
|
5790 }
|
|
390
|
5791
|
|
|
5792 /* When there are postponed prefixes... */
|
|
393
|
5793 if (spin->si_prefroot != NULL
|
|
|
5794 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
5795 {
|
|
|
5796 /* ... add a flag to indicate an affix was used. */
|
|
|
5797 use_flags |= WF_HAS_AFF;
|
|
|
5798
|
|
|
5799 /* ... don't use a prefix list if combining
|
|
483
|
5800 * affixes is not allowed. But do use the
|
|
|
5801 * compound flags after them. */
|
|
|
5802 if ((!ah->ah_combine || comb) && pfxlist != NULL)
|
|
|
5803 use_pfxlist += pfxlen;
|
|
390
|
5804 }
|
|
366
|
5805
|
|
300
|
5806 /* Store the modified word. */
|
|
474
|
5807 if (store_word(spin, newword, use_flags,
|
|
483
|
5808 spin->si_region, use_pfxlist, FALSE) == FAIL)
|
|
300
|
5809 retval = FAIL;
|
|
236
|
5810
|
|
300
|
5811 /* When added a suffix and combining is allowed also
|
|
|
5812 * try adding prefixes additionally. */
|
|
|
5813 if (xht != NULL && ah->ah_combine)
|
|
474
|
5814 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
5815 xht, NULL, TRUE,
|
|
483
|
5816 use_flags, use_pfxlist, pfxlen) == FAIL)
|
|
300
|
5817 retval = FAIL;
|
|
236
|
5818 }
|
|
|
5819 }
|
|
|
5820 }
|
|
|
5821 }
|
|
|
5822 }
|
|
|
5823
|
|
|
5824 return retval;
|
|
|
5825 }
|
|
|
5826
|
|
|
5827 /*
|
|
300
|
5828 * Read a file with a list of words.
|
|
236
|
5829 */
|
|
|
5830 static int
|
|
474
|
5831 spell_read_wordfile(spin, fname)
|
|
|
5832 spellinfo_T *spin;
|
|
300
|
5833 char_u *fname;
|
|
236
|
5834 {
|
|
300
|
5835 FILE *fd;
|
|
|
5836 long lnum = 0;
|
|
|
5837 char_u rline[MAXLINELEN];
|
|
|
5838 char_u *line;
|
|
|
5839 char_u *pc = NULL;
|
|
381
|
5840 char_u *p;
|
|
300
|
5841 int l;
|
|
|
5842 int retval = OK;
|
|
|
5843 int did_word = FALSE;
|
|
|
5844 int non_ascii = 0;
|
|
307
|
5845 int flags;
|
|
316
|
5846 int regionmask;
|
|
236
|
5847
|
|
300
|
5848 /*
|
|
|
5849 * Open the file.
|
|
|
5850 */
|
|
310
|
5851 fd = mch_fopen((char *)fname, "r");
|
|
300
|
5852 if (fd == NULL)
|
|
236
|
5853 {
|
|
300
|
5854 EMSG2(_(e_notopen), fname);
|
|
|
5855 return FAIL;
|
|
236
|
5856 }
|
|
|
5857
|
|
310
|
5858 if (spin->si_verbose || p_verbose > 2)
|
|
|
5859 {
|
|
|
5860 if (!spin->si_verbose)
|
|
|
5861 verbose_enter();
|
|
366
|
5862 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
5863 out_flush();
|
|
|
5864 if (!spin->si_verbose)
|
|
|
5865 verbose_leave();
|
|
|
5866 }
|
|
300
|
5867
|
|
|
5868 /*
|
|
|
5869 * Read all the lines in the file one by one.
|
|
|
5870 */
|
|
|
5871 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
5872 {
|
|
|
5873 line_breakcheck();
|
|
|
5874 ++lnum;
|
|
|
5875
|
|
|
5876 /* Skip comment lines. */
|
|
|
5877 if (*rline == '#')
|
|
|
5878 continue;
|
|
|
5879
|
|
|
5880 /* Remove CR, LF and white space from the end. */
|
|
|
5881 l = STRLEN(rline);
|
|
|
5882 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
5883 --l;
|
|
|
5884 if (l == 0)
|
|
|
5885 continue; /* empty or blank line */
|
|
|
5886 rline[l] = NUL;
|
|
|
5887
|
|
|
5888 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
5889 vim_free(pc);
|
|
310
|
5890 #ifdef FEAT_MBYTE
|
|
300
|
5891 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
5892 {
|
|
|
5893 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
5894 if (pc == NULL)
|
|
|
5895 {
|
|
|
5896 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5897 fname, lnum, rline);
|
|
|
5898 continue;
|
|
|
5899 }
|
|
|
5900 line = pc;
|
|
|
5901 }
|
|
|
5902 else
|
|
310
|
5903 #endif
|
|
300
|
5904 {
|
|
|
5905 pc = NULL;
|
|
|
5906 line = rline;
|
|
|
5907 }
|
|
|
5908
|
|
307
|
5909 if (*line == '/')
|
|
300
|
5910 {
|
|
307
|
5911 ++line;
|
|
|
5912 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
5913 {
|
|
|
5914 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
5915 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
5916 fname, lnum, line - 1);
|
|
300
|
5917 else if (did_word)
|
|
316
|
5918 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
5919 fname, lnum, line - 1);
|
|
300
|
5920 else
|
|
|
5921 {
|
|
310
|
5922 #ifdef FEAT_MBYTE
|
|
|
5923 char_u *enc;
|
|
|
5924
|
|
300
|
5925 /* Setup for conversion to 'encoding'. */
|
|
316
|
5926 line += 10;
|
|
|
5927 enc = enc_canonize(line);
|
|
300
|
5928 if (enc != NULL && !spin->si_ascii
|
|
|
5929 && convert_setup(&spin->si_conv, enc,
|
|
|
5930 p_enc) == FAIL)
|
|
|
5931 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
5932 fname, line, p_enc);
|
|
300
|
5933 vim_free(enc);
|
|
375
|
5934 spin->si_conv.vc_fail = TRUE;
|
|
310
|
5935 #else
|
|
|
5936 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
5937 #endif
|
|
300
|
5938 }
|
|
307
|
5939 continue;
|
|
300
|
5940 }
|
|
307
|
5941
|
|
316
|
5942 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
5943 {
|
|
|
5944 if (spin->si_region_count > 1)
|
|
|
5945 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
5946 fname, lnum, line);
|
|
|
5947 else
|
|
|
5948 {
|
|
|
5949 line += 8;
|
|
|
5950 if (STRLEN(line) > 16)
|
|
|
5951 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
5952 fname, lnum, line);
|
|
|
5953 else
|
|
|
5954 {
|
|
|
5955 spin->si_region_count = STRLEN(line) / 2;
|
|
|
5956 STRCPY(spin->si_region_name, line);
|
|
389
|
5957
|
|
|
5958 /* Adjust the mask for a word valid in all regions. */
|
|
|
5959 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
5960 }
|
|
|
5961 }
|
|
|
5962 continue;
|
|
|
5963 }
|
|
|
5964
|
|
381
|
5965 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
5966 fname, lnum, line - 1);
|
|
|
5967 continue;
|
|
|
5968 }
|
|
|
5969
|
|
|
5970 flags = 0;
|
|
|
5971 regionmask = spin->si_region;
|
|
|
5972
|
|
|
5973 /* Check for flags and region after a slash. */
|
|
|
5974 p = vim_strchr(line, '/');
|
|
|
5975 if (p != NULL)
|
|
|
5976 {
|
|
|
5977 *p++ = NUL;
|
|
|
5978 while (*p != NUL)
|
|
|
5979 {
|
|
|
5980 if (*p == '=') /* keep-case word */
|
|
389
|
5981 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
5982 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
5983 flags |= WF_BANNED;
|
|
|
5984 else if (*p == '?') /* Rare word. */
|
|
|
5985 flags |= WF_RARE;
|
|
|
5986 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
5987 {
|
|
381
|
5988 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
5989 regionmask = 0;
|
|
|
5990 flags |= WF_REGION;
|
|
|
5991
|
|
|
5992 l = *p - '0';
|
|
316
|
5993 if (l > spin->si_region_count)
|
|
|
5994 {
|
|
|
5995 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
5996 fname, lnum, p);
|
|
316
|
5997 break;
|
|
|
5998 }
|
|
|
5999 regionmask |= 1 << (l - 1);
|
|
|
6000 }
|
|
381
|
6001 else
|
|
|
6002 {
|
|
|
6003 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
6004 fname, lnum, p);
|
|
|
6005 break;
|
|
|
6006 }
|
|
|
6007 ++p;
|
|
307
|
6008 }
|
|
300
|
6009 }
|
|
|
6010
|
|
|
6011 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
6012 if (spin->si_ascii && has_non_ascii(line))
|
|
|
6013 {
|
|
|
6014 ++non_ascii;
|
|
|
6015 continue;
|
|
|
6016 }
|
|
|
6017
|
|
|
6018 /* Normal word: store it. */
|
|
483
|
6019 if (store_word(spin, line, flags, regionmask, NULL, FALSE) == FAIL)
|
|
300
|
6020 {
|
|
|
6021 retval = FAIL;
|
|
|
6022 break;
|
|
|
6023 }
|
|
|
6024 did_word = TRUE;
|
|
|
6025 }
|
|
|
6026
|
|
|
6027 vim_free(pc);
|
|
|
6028 fclose(fd);
|
|
|
6029
|
|
310
|
6030 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
6031 {
|
|
|
6032 if (p_verbose > 2)
|
|
|
6033 verbose_enter();
|
|
300
|
6034 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
6035 non_ascii);
|
|
310
|
6036 if (p_verbose > 2)
|
|
|
6037 verbose_leave();
|
|
|
6038 }
|
|
300
|
6039 return retval;
|
|
236
|
6040 }
|
|
|
6041
|
|
|
6042 /*
|
|
300
|
6043 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
6044 * This avoids calling free() for every little struct we use (and keeping
|
|
|
6045 * track of them).
|
|
300
|
6046 * The memory is cleared to all zeros.
|
|
|
6047 * Returns NULL when out of memory.
|
|
|
6048 */
|
|
|
6049 static void *
|
|
474
|
6050 getroom(spin, len, align)
|
|
|
6051 spellinfo_T *spin;
|
|
447
|
6052 size_t len; /* length needed */
|
|
|
6053 int align; /* align for pointer */
|
|
300
|
6054 {
|
|
|
6055 char_u *p;
|
|
474
|
6056 sblock_T *bl = spin->si_blocks;
|
|
300
|
6057
|
|
447
|
6058 if (align && bl != NULL)
|
|
|
6059 /* Round size up for alignment. On some systems structures need to be
|
|
|
6060 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
6061 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
6062 & ~(sizeof(char *) - 1);
|
|
|
6063
|
|
300
|
6064 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
6065 {
|
|
|
6066 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
6067 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
6068 if (bl == NULL)
|
|
|
6069 return NULL;
|
|
474
|
6070 bl->sb_next = spin->si_blocks;
|
|
|
6071 spin->si_blocks = bl;
|
|
300
|
6072 bl->sb_used = 0;
|
|
482
|
6073 ++spin->si_blocks_cnt;
|
|
300
|
6074 }
|
|
|
6075
|
|
|
6076 p = bl->sb_data + bl->sb_used;
|
|
|
6077 bl->sb_used += len;
|
|
|
6078
|
|
|
6079 return p;
|
|
|
6080 }
|
|
|
6081
|
|
|
6082 /*
|
|
|
6083 * Make a copy of a string into memory allocated with getroom().
|
|
|
6084 */
|
|
|
6085 static char_u *
|
|
474
|
6086 getroom_save(spin, s)
|
|
|
6087 spellinfo_T *spin;
|
|
300
|
6088 char_u *s;
|
|
|
6089 {
|
|
|
6090 char_u *sc;
|
|
|
6091
|
|
474
|
6092 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
6093 if (sc != NULL)
|
|
|
6094 STRCPY(sc, s);
|
|
|
6095 return sc;
|
|
|
6096 }
|
|
|
6097
|
|
|
6098
|
|
|
6099 /*
|
|
|
6100 * Free the list of allocated sblock_T.
|
|
236
|
6101 */
|
|
|
6102 static void
|
|
300
|
6103 free_blocks(bl)
|
|
|
6104 sblock_T *bl;
|
|
236
|
6105 {
|
|
300
|
6106 sblock_T *next;
|
|
236
|
6107
|
|
300
|
6108 while (bl != NULL)
|
|
236
|
6109 {
|
|
300
|
6110 next = bl->sb_next;
|
|
|
6111 vim_free(bl);
|
|
|
6112 bl = next;
|
|
236
|
6113 }
|
|
|
6114 }
|
|
|
6115
|
|
|
6116 /*
|
|
300
|
6117 * Allocate the root of a word tree.
|
|
236
|
6118 */
|
|
300
|
6119 static wordnode_T *
|
|
474
|
6120 wordtree_alloc(spin)
|
|
|
6121 spellinfo_T *spin;
|
|
|
6122 {
|
|
|
6123 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
6124 }
|
|
|
6125
|
|
|
6126 /*
|
|
300
|
6127 * Store a word in the tree(s).
|
|
390
|
6128 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
6129 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
6130 * used to find suggestions.
|
|
300
|
6131 * For a keep-case word also store it in the keep-case tree.
|
|
482
|
6132 * When "pfxlist" is not NULL store the word for each postponed prefix ID and
|
|
|
6133 * compound flag.
|
|
236
|
6134 */
|
|
|
6135 static int
|
|
483
|
6136 store_word(spin, word, flags, region, pfxlist, need_affix)
|
|
474
|
6137 spellinfo_T *spin;
|
|
300
|
6138 char_u *word;
|
|
307
|
6139 int flags; /* extra flags, WF_BANNED */
|
|
316
|
6140 int region; /* supported region(s) */
|
|
339
|
6141 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
483
|
6142 int need_affix; /* only store word with affix ID */
|
|
236
|
6143 {
|
|
300
|
6144 int len = STRLEN(word);
|
|
|
6145 int ct = captype(word, word + len);
|
|
|
6146 char_u foldword[MAXWLEN];
|
|
339
|
6147 int res = OK;
|
|
|
6148 char_u *p;
|
|
236
|
6149
|
|
323
|
6150 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
6151 for (p = pfxlist; res == OK; ++p)
|
|
|
6152 {
|
|
483
|
6153 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6154 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
474
|
6155 region, p == NULL ? 0 : *p);
|
|
339
|
6156 if (p == NULL || *p == NUL)
|
|
|
6157 break;
|
|
|
6158 }
|
|
334
|
6159 ++spin->si_foldwcount;
|
|
307
|
6160
|
|
474
|
6161 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
6162 {
|
|
339
|
6163 for (p = pfxlist; res == OK; ++p)
|
|
|
6164 {
|
|
483
|
6165 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6166 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
474
|
6167 region, p == NULL ? 0 : *p);
|
|
339
|
6168 if (p == NULL || *p == NUL)
|
|
|
6169 break;
|
|
|
6170 }
|
|
334
|
6171 ++spin->si_keepwcount;
|
|
|
6172 }
|
|
300
|
6173 return res;
|
|
236
|
6174 }
|
|
|
6175
|
|
|
6176 /*
|
|
300
|
6177 * Add word "word" to a word tree at "root".
|
|
366
|
6178 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
6179 * "rare" and "region" is the condition nr.
|
|
255
|
6180 * Returns FAIL when out of memory.
|
|
236
|
6181 */
|
|
255
|
6182 static int
|
|
481
|
6183 tree_add_word(spin, word, root, flags, region, affixID)
|
|
474
|
6184 spellinfo_T *spin;
|
|
300
|
6185 char_u *word;
|
|
|
6186 wordnode_T *root;
|
|
|
6187 int flags;
|
|
|
6188 int region;
|
|
481
|
6189 int affixID;
|
|
236
|
6190 {
|
|
300
|
6191 wordnode_T *node = root;
|
|
|
6192 wordnode_T *np;
|
|
469
|
6193 wordnode_T *copyp, **copyprev;
|
|
300
|
6194 wordnode_T **prev = NULL;
|
|
|
6195 int i;
|
|
255
|
6196
|
|
300
|
6197 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
6198 for (i = 0; ; ++i)
|
|
255
|
6199 {
|
|
469
|
6200 /* When there is more than one reference to this node we need to make
|
|
|
6201 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
6202 * (we don't optimize for a partly shared list of siblings). */
|
|
|
6203 if (node != NULL && node->wn_refs > 1)
|
|
|
6204 {
|
|
|
6205 --node->wn_refs;
|
|
|
6206 copyprev = prev;
|
|
|
6207 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
6208 {
|
|
|
6209 /* Allocate a new node and copy the info. */
|
|
474
|
6210 np = get_wordnode(spin);
|
|
469
|
6211 if (np == NULL)
|
|
|
6212 return FAIL;
|
|
|
6213 np->wn_child = copyp->wn_child;
|
|
|
6214 if (np->wn_child != NULL)
|
|
|
6215 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
6216 np->wn_byte = copyp->wn_byte;
|
|
|
6217 if (np->wn_byte == NUL)
|
|
|
6218 {
|
|
|
6219 np->wn_flags = copyp->wn_flags;
|
|
|
6220 np->wn_region = copyp->wn_region;
|
|
481
|
6221 np->wn_affixID = copyp->wn_affixID;
|
|
469
|
6222 }
|
|
|
6223
|
|
|
6224 /* Link the new node in the list, there will be one ref. */
|
|
|
6225 np->wn_refs = 1;
|
|
|
6226 *copyprev = np;
|
|
|
6227 copyprev = &np->wn_sibling;
|
|
|
6228
|
|
|
6229 /* Let "node" point to the head of the copied list. */
|
|
|
6230 if (copyp == node)
|
|
|
6231 node = np;
|
|
|
6232 }
|
|
|
6233 }
|
|
|
6234
|
|
300
|
6235 /* Look for the sibling that has the same character. They are sorted
|
|
|
6236 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
6237 * higher byte value. For zero bytes (end of word) the sorting is
|
|
481
|
6238 * done on flags and then on affixID. */
|
|
339
|
6239 while (node != NULL
|
|
|
6240 && (node->wn_byte < word[i]
|
|
|
6241 || (node->wn_byte == NUL
|
|
|
6242 && (flags < 0
|
|
481
|
6243 ? node->wn_affixID < affixID
|
|
390
|
6244 : node->wn_flags < (flags & WN_MASK)
|
|
|
6245 || (node->wn_flags == (flags & WN_MASK)
|
|
481
|
6246 && node->wn_affixID < affixID)))))
|
|
236
|
6247 {
|
|
300
|
6248 prev = &node->wn_sibling;
|
|
|
6249 node = *prev;
|
|
236
|
6250 }
|
|
339
|
6251 if (node == NULL
|
|
|
6252 || node->wn_byte != word[i]
|
|
|
6253 || (word[i] == NUL
|
|
|
6254 && (flags < 0
|
|
390
|
6255 || node->wn_flags != (flags & WN_MASK)
|
|
481
|
6256 || node->wn_affixID != affixID)))
|
|
255
|
6257 {
|
|
300
|
6258 /* Allocate a new node. */
|
|
474
|
6259 np = get_wordnode(spin);
|
|
300
|
6260 if (np == NULL)
|
|
|
6261 return FAIL;
|
|
|
6262 np->wn_byte = word[i];
|
|
469
|
6263
|
|
|
6264 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
6265 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
6266 * make ref count of sibling one (matters when inserting in front
|
|
|
6267 * of the list of siblings). */
|
|
|
6268 if (node == NULL)
|
|
|
6269 np->wn_refs = 1;
|
|
|
6270 else
|
|
|
6271 {
|
|
|
6272 np->wn_refs = node->wn_refs;
|
|
|
6273 node->wn_refs = 1;
|
|
|
6274 }
|
|
300
|
6275 *prev = np;
|
|
|
6276 np->wn_sibling = node;
|
|
|
6277 node = np;
|
|
255
|
6278 }
|
|
300
|
6279
|
|
|
6280 if (word[i] == NUL)
|
|
|
6281 {
|
|
|
6282 node->wn_flags = flags;
|
|
|
6283 node->wn_region |= region;
|
|
481
|
6284 node->wn_affixID = affixID;
|
|
300
|
6285 break;
|
|
|
6286 }
|
|
|
6287 prev = &node->wn_child;
|
|
|
6288 node = *prev;
|
|
255
|
6289 }
|
|
469
|
6290 #ifdef SPELL_PRINTTREE
|
|
|
6291 smsg("Added \"%s\"", word);
|
|
|
6292 spell_print_tree(root->wn_sibling);
|
|
|
6293 #endif
|
|
|
6294
|
|
474
|
6295 /* count nr of words added since last message */
|
|
|
6296 ++spin->si_msg_count;
|
|
|
6297
|
|
482
|
6298 if (spin->si_compress_cnt > 1)
|
|
|
6299 {
|
|
|
6300 if (--spin->si_compress_cnt == 1)
|
|
|
6301 /* Did enough words to lower the block count limit. */
|
|
483
|
6302 spin->si_blocks_cnt += compress_inc;
|
|
482
|
6303 }
|
|
|
6304
|
|
469
|
6305 /*
|
|
482
|
6306 * When we have allocated lots of memory we need to compress the word tree
|
|
|
6307 * to free up some room. But compression is slow, and we might actually
|
|
|
6308 * need that room, thus only compress in the following situations:
|
|
|
6309 * 1. When not compressed before (si_compress_cnt == 0): when using
|
|
483
|
6310 * "compress_start" blocks.
|
|
|
6311 * 2. When compressed before and used "compress_inc" blocks before
|
|
|
6312 * adding "compress_added" words (si_compress_cnt > 1).
|
|
|
6313 * 3. When compressed before, added "compress_added" words
|
|
482
|
6314 * (si_compress_cnt == 1) and the number of free nodes drops below the
|
|
|
6315 * maximum word length.
|
|
469
|
6316 */
|
|
482
|
6317 #ifndef SPELL_PRINTTREE
|
|
|
6318 if (spin->si_compress_cnt == 1
|
|
|
6319 ? spin->si_free_count < MAXWLEN
|
|
483
|
6320 : spin->si_blocks_cnt >= compress_start)
|
|
482
|
6321 #endif
|
|
|
6322 {
|
|
|
6323 /* Decrement the block counter. The effect is that we compress again
|
|
483
|
6324 * when the freed up room has been used and another "compress_inc"
|
|
|
6325 * blocks have been allocated. Unless "compress_added" words have
|
|
482
|
6326 * been added, then the limit is put back again. */
|
|
483
|
6327 spin->si_blocks_cnt -= compress_inc;
|
|
|
6328 spin->si_compress_cnt = compress_added;
|
|
482
|
6329
|
|
474
|
6330 if (spin->si_verbose)
|
|
|
6331 {
|
|
|
6332 msg_start();
|
|
|
6333 msg_puts((char_u *)_(msg_compressing));
|
|
|
6334 msg_clr_eos();
|
|
|
6335 msg_didout = FALSE;
|
|
|
6336 msg_col = 0;
|
|
|
6337 out_flush();
|
|
|
6338 }
|
|
482
|
6339
|
|
|
6340 /* Compress both trees. Either they both have many nodes, which makes
|
|
|
6341 * compression useful, or one of them is small, which means
|
|
|
6342 * compression goes fast. */
|
|
|
6343 wordtree_compress(spin, spin->si_foldroot);
|
|
|
6344 wordtree_compress(spin, spin->si_keeproot);
|
|
469
|
6345 }
|
|
255
|
6346
|
|
|
6347 return OK;
|
|
236
|
6348 }
|
|
|
6349
|
|
469
|
6350 /*
|
|
483
|
6351 * Check the 'mkspellmem' option. Return FAIL if it's wrong.
|
|
|
6352 * Sets "sps_flags".
|
|
|
6353 */
|
|
|
6354 int
|
|
|
6355 spell_check_msm()
|
|
|
6356 {
|
|
|
6357 char_u *p = p_msm;
|
|
|
6358 long start = 0;
|
|
|
6359 long inc = 0;
|
|
|
6360 long added = 0;
|
|
|
6361
|
|
|
6362 if (!VIM_ISDIGIT(*p))
|
|
|
6363 return FAIL;
|
|
|
6364 /* block count = (value * 1024) / SBLOCKSIZE (but avoid overflow)*/
|
|
|
6365 start = (getdigits(&p) * 10) / (SBLOCKSIZE / 102);
|
|
|
6366 if (*p != ',')
|
|
|
6367 return FAIL;
|
|
|
6368 ++p;
|
|
|
6369 if (!VIM_ISDIGIT(*p))
|
|
|
6370 return FAIL;
|
|
|
6371 inc = (getdigits(&p) * 102) / (SBLOCKSIZE / 10);
|
|
|
6372 if (*p != ',')
|
|
|
6373 return FAIL;
|
|
|
6374 ++p;
|
|
|
6375 if (!VIM_ISDIGIT(*p))
|
|
|
6376 return FAIL;
|
|
|
6377 added = getdigits(&p) * 1024;
|
|
|
6378 if (*p != NUL)
|
|
|
6379 return FAIL;
|
|
|
6380
|
|
|
6381 if (start == 0 || inc == 0 || added == 0 || inc > start)
|
|
|
6382 return FAIL;
|
|
|
6383
|
|
|
6384 compress_start = start;
|
|
|
6385 compress_inc = inc;
|
|
|
6386 compress_added = added;
|
|
|
6387 return OK;
|
|
|
6388 }
|
|
|
6389
|
|
|
6390
|
|
|
6391 /*
|
|
469
|
6392 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
6393 * allocate a new one.
|
|
|
6394 */
|
|
|
6395 static wordnode_T *
|
|
474
|
6396 get_wordnode(spin)
|
|
|
6397 spellinfo_T *spin;
|
|
469
|
6398 {
|
|
|
6399 wordnode_T *n;
|
|
|
6400
|
|
474
|
6401 if (spin->si_first_free == NULL)
|
|
|
6402 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
6403 else
|
|
|
6404 {
|
|
474
|
6405 n = spin->si_first_free;
|
|
|
6406 spin->si_first_free = n->wn_child;
|
|
469
|
6407 vim_memset(n, 0, sizeof(wordnode_T));
|
|
482
|
6408 --spin->si_free_count;
|
|
469
|
6409 }
|
|
|
6410 #ifdef SPELL_PRINTTREE
|
|
474
|
6411 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
6412 #endif
|
|
|
6413 return n;
|
|
|
6414 }
|
|
|
6415
|
|
|
6416 /*
|
|
474
|
6417 * Decrement the reference count on a node (which is the head of a list of
|
|
|
6418 * siblings). If the reference count becomes zero free the node and its
|
|
|
6419 * siblings.
|
|
|
6420 */
|
|
|
6421 static void
|
|
|
6422 deref_wordnode(spin, node)
|
|
|
6423 spellinfo_T *spin;
|
|
|
6424 wordnode_T *node;
|
|
|
6425 {
|
|
|
6426 wordnode_T *np;
|
|
|
6427
|
|
|
6428 if (--node->wn_refs == 0)
|
|
|
6429 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6430 {
|
|
|
6431 if (np->wn_child != NULL)
|
|
|
6432 deref_wordnode(spin, np->wn_child);
|
|
|
6433 free_wordnode(spin, np);
|
|
|
6434 }
|
|
|
6435 }
|
|
|
6436
|
|
|
6437 /*
|
|
469
|
6438 * Free a wordnode_T for re-use later.
|
|
474
|
6439 * Only the "wn_child" field becomes invalid.
|
|
469
|
6440 */
|
|
|
6441 static void
|
|
474
|
6442 free_wordnode(spin, n)
|
|
|
6443 spellinfo_T *spin;
|
|
|
6444 wordnode_T *n;
|
|
|
6445 {
|
|
|
6446 n->wn_child = spin->si_first_free;
|
|
|
6447 spin->si_first_free = n;
|
|
482
|
6448 ++spin->si_free_count;
|
|
469
|
6449 }
|
|
|
6450
|
|
236
|
6451 /*
|
|
300
|
6452 * Compress a tree: find tails that are identical and can be shared.
|
|
|
6453 */
|
|
|
6454 static void
|
|
474
|
6455 wordtree_compress(spin, root)
|
|
|
6456 spellinfo_T *spin;
|
|
300
|
6457 wordnode_T *root;
|
|
|
6458 {
|
|
|
6459 hashtab_T ht;
|
|
|
6460 int n;
|
|
|
6461 int tot = 0;
|
|
469
|
6462 int perc;
|
|
300
|
6463
|
|
474
|
6464 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
6465 * start of the tree. */
|
|
|
6466 if (root->wn_sibling != NULL)
|
|
300
|
6467 {
|
|
|
6468 hash_init(&ht);
|
|
474
|
6469 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
6470
|
|
|
6471 #ifndef SPELL_PRINTTREE
|
|
310
|
6472 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
6473 #endif
|
|
310
|
6474 {
|
|
|
6475 if (!spin->si_verbose)
|
|
|
6476 verbose_enter();
|
|
469
|
6477 if (tot > 1000000)
|
|
|
6478 perc = (tot - n) / (tot / 100);
|
|
482
|
6479 else if (tot == 0)
|
|
|
6480 perc = 0;
|
|
469
|
6481 else
|
|
|
6482 perc = (tot - n) * 100 / tot;
|
|
310
|
6483 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
6484 n, tot, perc);
|
|
310
|
6485 if (p_verbose > 2)
|
|
|
6486 verbose_leave();
|
|
|
6487 }
|
|
469
|
6488 #ifdef SPELL_PRINTTREE
|
|
474
|
6489 spell_print_tree(root->wn_sibling);
|
|
469
|
6490 #endif
|
|
300
|
6491 hash_clear(&ht);
|
|
|
6492 }
|
|
|
6493 }
|
|
|
6494
|
|
|
6495 /*
|
|
|
6496 * Compress a node, its siblings and its children, depth first.
|
|
|
6497 * Returns the number of compressed nodes.
|
|
236
|
6498 */
|
|
255
|
6499 static int
|
|
474
|
6500 node_compress(spin, node, ht, tot)
|
|
|
6501 spellinfo_T *spin;
|
|
300
|
6502 wordnode_T *node;
|
|
|
6503 hashtab_T *ht;
|
|
|
6504 int *tot; /* total count of nodes before compressing,
|
|
|
6505 incremented while going through the tree */
|
|
236
|
6506 {
|
|
300
|
6507 wordnode_T *np;
|
|
|
6508 wordnode_T *tp;
|
|
|
6509 wordnode_T *child;
|
|
|
6510 hash_T hash;
|
|
236
|
6511 hashitem_T *hi;
|
|
300
|
6512 int len = 0;
|
|
|
6513 unsigned nr, n;
|
|
|
6514 int compressed = 0;
|
|
236
|
6515
|
|
300
|
6516 /*
|
|
|
6517 * Go through the list of siblings. Compress each child and then try
|
|
|
6518 * finding an identical child to replace it.
|
|
|
6519 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
6520 * but the whole list of siblings, of which the node is the first.
|
|
|
6521 */
|
|
482
|
6522 for (np = node; np != NULL && !got_int; np = np->wn_sibling)
|
|
236
|
6523 {
|
|
300
|
6524 ++len;
|
|
|
6525 if ((child = np->wn_child) != NULL)
|
|
|
6526 {
|
|
346
|
6527 /* Compress the child. This fills hashkey. */
|
|
474
|
6528 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
6529
|
|
|
6530 /* Try to find an identical child. */
|
|
346
|
6531 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
6532 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
6533 tp = NULL;
|
|
|
6534 if (!HASHITEM_EMPTY(hi))
|
|
|
6535 {
|
|
|
6536 /* There are children with an identical hash value. Now check
|
|
|
6537 * if there is one that is really identical. */
|
|
346
|
6538 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
6539 if (node_equal(child, tp))
|
|
|
6540 {
|
|
|
6541 /* Found one! Now use that child in place of the
|
|
469
|
6542 * current one. This means the current child and all
|
|
|
6543 * its siblings is unlinked from the tree. */
|
|
|
6544 ++tp->wn_refs;
|
|
474
|
6545 deref_wordnode(spin, child);
|
|
300
|
6546 np->wn_child = tp;
|
|
|
6547 ++compressed;
|
|
|
6548 break;
|
|
|
6549 }
|
|
|
6550 if (tp == NULL)
|
|
|
6551 {
|
|
|
6552 /* No other child with this hash value equals the child of
|
|
|
6553 * the node, add it to the linked list after the first
|
|
|
6554 * item. */
|
|
|
6555 tp = HI2WN(hi);
|
|
346
|
6556 child->wn_u2.next = tp->wn_u2.next;
|
|
|
6557 tp->wn_u2.next = child;
|
|
300
|
6558 }
|
|
|
6559 }
|
|
|
6560 else
|
|
|
6561 /* No other child has this hash value, add it to the
|
|
|
6562 * hashtable. */
|
|
346
|
6563 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
6564 }
|
|
236
|
6565 }
|
|
300
|
6566 *tot += len;
|
|
|
6567
|
|
|
6568 /*
|
|
|
6569 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
6570 * find a lookalike node. This must be done after compressing the sibling
|
|
|
6571 * list, otherwise the hash key would become invalid by the compression.
|
|
|
6572 */
|
|
346
|
6573 node->wn_u1.hashkey[0] = len;
|
|
300
|
6574 nr = 0;
|
|
|
6575 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
6576 {
|
|
300
|
6577 if (np->wn_byte == NUL)
|
|
481
|
6578 /* end node: use wn_flags, wn_region and wn_affixID */
|
|
|
6579 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16);
|
|
300
|
6580 else
|
|
|
6581 /* byte node: use the byte value and the child pointer */
|
|
|
6582 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
6583 nr = nr * 101 + n;
|
|
236
|
6584 }
|
|
300
|
6585
|
|
|
6586 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
6587 n = nr & 0xff;
|
|
346
|
6588 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
6589 n = (nr >> 8) & 0xff;
|
|
346
|
6590 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
6591 n = (nr >> 16) & 0xff;
|
|
346
|
6592 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
6593 n = (nr >> 24) & 0xff;
|
|
346
|
6594 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
6595 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
6596
|
|
482
|
6597 /* Check for CTRL-C pressed now and then. */
|
|
|
6598 fast_breakcheck();
|
|
|
6599
|
|
300
|
6600 return compressed;
|
|
|
6601 }
|
|
|
6602
|
|
|
6603 /*
|
|
|
6604 * Return TRUE when two nodes have identical siblings and children.
|
|
|
6605 */
|
|
|
6606 static int
|
|
|
6607 node_equal(n1, n2)
|
|
|
6608 wordnode_T *n1;
|
|
|
6609 wordnode_T *n2;
|
|
|
6610 {
|
|
|
6611 wordnode_T *p1;
|
|
|
6612 wordnode_T *p2;
|
|
|
6613
|
|
|
6614 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
6615 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
6616 if (p1->wn_byte != p2->wn_byte
|
|
|
6617 || (p1->wn_byte == NUL
|
|
|
6618 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
6619 || p1->wn_region != p2->wn_region
|
|
481
|
6620 || p1->wn_affixID != p2->wn_affixID)
|
|
300
|
6621 : (p1->wn_child != p2->wn_child)))
|
|
|
6622 break;
|
|
|
6623
|
|
|
6624 return p1 == NULL && p2 == NULL;
|
|
236
|
6625 }
|
|
|
6626
|
|
|
6627 /*
|
|
|
6628 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
6629 */
|
|
255
|
6630 void
|
|
236
|
6631 put_bytes(fd, nr, len)
|
|
|
6632 FILE *fd;
|
|
|
6633 long_u nr;
|
|
|
6634 int len;
|
|
|
6635 {
|
|
|
6636 int i;
|
|
|
6637
|
|
|
6638 for (i = len - 1; i >= 0; --i)
|
|
|
6639 putc((int)(nr >> (i * 8)), fd);
|
|
|
6640 }
|
|
|
6641
|
|
323
|
6642 static int
|
|
|
6643 #ifdef __BORLANDC__
|
|
|
6644 _RTLENTRYF
|
|
|
6645 #endif
|
|
|
6646 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
6647
|
|
|
6648 /*
|
|
|
6649 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
6650 */
|
|
|
6651 static int
|
|
|
6652 #ifdef __BORLANDC__
|
|
|
6653 _RTLENTRYF
|
|
|
6654 #endif
|
|
|
6655 rep_compare(s1, s2)
|
|
|
6656 const void *s1;
|
|
|
6657 const void *s2;
|
|
|
6658 {
|
|
|
6659 fromto_T *p1 = (fromto_T *)s1;
|
|
|
6660 fromto_T *p2 = (fromto_T *)s2;
|
|
|
6661
|
|
|
6662 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
6663 }
|
|
|
6664
|
|
236
|
6665 /*
|
|
483
|
6666 * Write the Vim .spl file "fname".
|
|
236
|
6667 */
|
|
|
6668 static void
|
|
474
|
6669 write_vim_spell(spin, fname)
|
|
|
6670 spellinfo_T *spin;
|
|
236
|
6671 char_u *fname;
|
|
|
6672 {
|
|
300
|
6673 FILE *fd;
|
|
|
6674 int regionmask;
|
|
236
|
6675 int round;
|
|
300
|
6676 wordnode_T *tree;
|
|
|
6677 int nodecount;
|
|
323
|
6678 int i;
|
|
|
6679 int l;
|
|
|
6680 garray_T *gap;
|
|
|
6681 fromto_T *ftp;
|
|
|
6682 char_u *p;
|
|
|
6683 int rr;
|
|
236
|
6684
|
|
310
|
6685 fd = mch_fopen((char *)fname, "w");
|
|
300
|
6686 if (fd == NULL)
|
|
236
|
6687 {
|
|
|
6688 EMSG2(_(e_notopen), fname);
|
|
|
6689 return;
|
|
|
6690 }
|
|
|
6691
|
|
483
|
6692 /* <HEADER>: <fileID> <versionnr> */
|
|
300
|
6693 /* <fileID> */
|
|
|
6694 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
6695 EMSG(_(e_write));
|
|
483
|
6696 putc(VIMSPELLVERSION, fd); /* <versionnr> */
|
|
|
6697
|
|
|
6698 /*
|
|
|
6699 * <SECTIONS>: <section> ... <sectionend>
|
|
|
6700 */
|
|
|
6701
|
|
|
6702 /* SN_REGION: <regionname> ...
|
|
|
6703 * Write the region names only if there is more than one. */
|
|
316
|
6704 if (spin->si_region_count > 1)
|
|
236
|
6705 {
|
|
483
|
6706 putc(SN_REGION, fd); /* <sectionID> */
|
|
|
6707 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6708 l = spin->si_region_count * 2;
|
|
|
6709 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6710 fwrite(spin->si_region_name, (size_t)l, (size_t)1, fd);
|
|
|
6711 /* <regionname> ... */
|
|
316
|
6712 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
6713 }
|
|
|
6714 else
|
|
300
|
6715 regionmask = 0;
|
|
483
|
6716
|
|
|
6717 /* SN_CHARFLAGS: <charflagslen> <charflags> <folcharslen> <folchars>
|
|
|
6718 *
|
|
|
6719 * The table with character flags and the table for case folding.
|
|
|
6720 * This makes sure the same characters are recognized as word characters
|
|
|
6721 * when generating an when using a spell file.
|
|
260
|
6722 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
6723 * 'encoding'.
|
|
|
6724 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
6725 * the table (avoids that it conflicts). File is shorter too.
|
|
|
6726 */
|
|
483
|
6727 if (!spin->si_ascii && !spin->si_add)
|
|
|
6728 {
|
|
|
6729 char_u folchars[128 * 8];
|
|
|
6730 int flags;
|
|
|
6731
|
|
485
|
6732 putc(SN_CHARFLAGS, fd); /* <sectionID> */
|
|
483
|
6733 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6734
|
|
|
6735 /* Form the <folchars> string first, we need to know its length. */
|
|
|
6736 l = 0;
|
|
|
6737 for (i = 128; i < 256; ++i)
|
|
|
6738 {
|
|
|
6739 #ifdef FEAT_MBYTE
|
|
|
6740 if (has_mbyte)
|
|
|
6741 l += mb_char2bytes(spelltab.st_fold[i], folchars + l);
|
|
|
6742 else
|
|
|
6743 #endif
|
|
|
6744 folchars[l++] = spelltab.st_fold[i];
|
|
|
6745 }
|
|
|
6746 put_bytes(fd, (long_u)(1 + 128 + 2 + l), 4); /* <sectionlen> */
|
|
|
6747
|
|
|
6748 fputc(128, fd); /* <charflagslen> */
|
|
|
6749 for (i = 128; i < 256; ++i)
|
|
|
6750 {
|
|
|
6751 flags = 0;
|
|
|
6752 if (spelltab.st_isw[i])
|
|
|
6753 flags |= CF_WORD;
|
|
|
6754 if (spelltab.st_isu[i])
|
|
|
6755 flags |= CF_UPPER;
|
|
|
6756 fputc(flags, fd); /* <charflags> */
|
|
|
6757 }
|
|
|
6758
|
|
|
6759 put_bytes(fd, (long_u)l, 2); /* <folcharslen> */
|
|
|
6760 fwrite(folchars, (size_t)l, (size_t)1, fd); /* <folchars> */
|
|
|
6761 }
|
|
|
6762
|
|
|
6763 /* SN_MIDWORD: <midword> */
|
|
|
6764 if (spin->si_midword != NULL)
|
|
|
6765 {
|
|
|
6766 putc(SN_MIDWORD, fd); /* <sectionID> */
|
|
|
6767 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6768
|
|
366
|
6769 i = STRLEN(spin->si_midword);
|
|
483
|
6770 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */
|
|
366
|
6771 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
6772 }
|
|
|
6773
|
|
483
|
6774 /* SN_PREFCOND: <prefcondcnt> <prefcond> ... */
|
|
|
6775 if (spin->si_prefcond.ga_len > 0)
|
|
|
6776 {
|
|
|
6777 putc(SN_PREFCOND, fd); /* <sectionID> */
|
|
|
6778 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6779
|
|
|
6780 l = write_spell_prefcond(NULL, &spin->si_prefcond);
|
|
|
6781 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6782
|
|
|
6783 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
6784 }
|
|
|
6785
|
|
|
6786 /* SN_REP: <repcount> <rep> ...
|
|
|
6787 * SN_SAL: <salflags> <salcount> <sal> ... */
|
|
366
|
6788
|
|
323
|
6789 /* Sort the REP items. */
|
|
|
6790 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
6791 sizeof(fromto_T), rep_compare);
|
|
|
6792
|
|
483
|
6793 /* round 1: SN_REP section
|
|
|
6794 * round 2: SN_SAL section (unless SN_SOFO is used) */
|
|
323
|
6795 for (round = 1; round <= 2; ++round)
|
|
|
6796 {
|
|
|
6797 if (round == 1)
|
|
483
|
6798 {
|
|
323
|
6799 gap = &spin->si_rep;
|
|
483
|
6800 putc(SN_REP, fd); /* <sectionID> */
|
|
|
6801 }
|
|
323
|
6802 else
|
|
|
6803 {
|
|
483
|
6804 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
6805 /* using SN_SOFO section instead of SN_SAL */
|
|
|
6806 break;
|
|
323
|
6807 gap = &spin->si_sal;
|
|
483
|
6808 putc(SN_SAL, fd); /* <sectionID> */
|
|
|
6809 }
|
|
|
6810
|
|
|
6811 /* This is for making suggestions, section is not required. */
|
|
|
6812 putc(0, fd); /* <sectionflags> */
|
|
|
6813
|
|
|
6814 /* Compute the length of what follows. */
|
|
|
6815 l = 2; /* count <repcount> or <salcount> */
|
|
|
6816 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6817 {
|
|
|
6818 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
6819 l += 1 + STRLEN(ftp->ft_from); /* count <*fromlen> and <*from> */
|
|
|
6820 l += 1 + STRLEN(ftp->ft_to); /* count <*tolen> and <*to> */
|
|
|
6821 }
|
|
|
6822 if (round == 2)
|
|
|
6823 ++l; /* count <salflags> */
|
|
|
6824 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6825
|
|
|
6826 if (round == 2)
|
|
|
6827 {
|
|
323
|
6828 i = 0;
|
|
|
6829 if (spin->si_followup)
|
|
|
6830 i |= SAL_F0LLOWUP;
|
|
|
6831 if (spin->si_collapse)
|
|
|
6832 i |= SAL_COLLAPSE;
|
|
|
6833 if (spin->si_rem_accents)
|
|
|
6834 i |= SAL_REM_ACCENTS;
|
|
|
6835 putc(i, fd); /* <salflags> */
|
|
|
6836 }
|
|
|
6837
|
|
|
6838 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
6839 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6840 {
|
|
|
6841 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
6842 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
6843 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
6844 for (rr = 1; rr <= 2; ++rr)
|
|
|
6845 {
|
|
|
6846 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
6847 l = STRLEN(p);
|
|
|
6848 putc(l, fd);
|
|
|
6849 fwrite(p, l, (size_t)1, fd);
|
|
|
6850 }
|
|
|
6851 }
|
|
483
|
6852
|
|
|
6853 }
|
|
|
6854
|
|
|
6855 /* SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
6856 * This is for making suggestions, section is not required. */
|
|
375
|
6857 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
6858 {
|
|
483
|
6859 putc(SN_SOFO, fd); /* <sectionID> */
|
|
|
6860 putc(0, fd); /* <sectionflags> */
|
|
375
|
6861
|
|
|
6862 l = STRLEN(spin->si_sofofr);
|
|
483
|
6863 put_bytes(fd, (long_u)(l + STRLEN(spin->si_sofoto) + 4), 4);
|
|
|
6864 /* <sectionlen> */
|
|
|
6865
|
|
|
6866 put_bytes(fd, (long_u)l, 2); /* <sofofromlen> */
|
|
|
6867 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <sofofrom> */
|
|
375
|
6868
|
|
|
6869 l = STRLEN(spin->si_sofoto);
|
|
483
|
6870 put_bytes(fd, (long_u)l, 2); /* <sofotolen> */
|
|
|
6871 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <sofoto> */
|
|
|
6872 }
|
|
|
6873
|
|
|
6874 /* SN_MAP: <mapstr>
|
|
|
6875 * This is for making suggestions, section is not required. */
|
|
|
6876 if (spin->si_map.ga_len > 0)
|
|
|
6877 {
|
|
|
6878 putc(SN_MAP, fd); /* <sectionID> */
|
|
|
6879 putc(0, fd); /* <sectionflags> */
|
|
|
6880 l = spin->si_map.ga_len;
|
|
|
6881 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6882 fwrite(spin->si_map.ga_data, (size_t)l, (size_t)1, fd);
|
|
|
6883 /* <mapstr> */
|
|
|
6884 }
|
|
|
6885
|
|
|
6886 /* SN_COMPOUND: compound info.
|
|
|
6887 * We don't mark it required, when not supported all compound words will
|
|
|
6888 * be bad words. */
|
|
|
6889 if (spin->si_compflags != NULL)
|
|
|
6890 {
|
|
|
6891 putc(SN_COMPOUND, fd); /* <sectionID> */
|
|
|
6892 putc(0, fd); /* <sectionflags> */
|
|
|
6893
|
|
|
6894 l = STRLEN(spin->si_compflags);
|
|
|
6895 put_bytes(fd, (long_u)(l + 3), 4); /* <sectionlen> */
|
|
|
6896 putc(spin->si_compmax, fd); /* <compmax> */
|
|
|
6897 putc(spin->si_compminlen, fd); /* <compminlen> */
|
|
|
6898 putc(spin->si_compsylmax, fd); /* <compsylmax> */
|
|
|
6899 /* <compflags> */
|
|
|
6900 fwrite(spin->si_compflags, (size_t)l, (size_t)1, fd);
|
|
|
6901 }
|
|
|
6902
|
|
492
|
6903 /* SN_NOBREAK: NOBREAK flag */
|
|
|
6904 if (spin->si_nobreak)
|
|
|
6905 {
|
|
|
6906 putc(SN_NOBREAK, fd); /* <sectionID> */
|
|
|
6907 putc(0, fd); /* <sectionflags> */
|
|
|
6908
|
|
|
6909 /* It's empty, the precense of the section flags the feature. */
|
|
|
6910 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */
|
|
|
6911 }
|
|
|
6912
|
|
483
|
6913 /* SN_SYLLABLE: syllable info.
|
|
|
6914 * We don't mark it required, when not supported syllables will not be
|
|
|
6915 * counted. */
|
|
|
6916 if (spin->si_syllable != NULL)
|
|
|
6917 {
|
|
|
6918 putc(SN_SYLLABLE, fd); /* <sectionID> */
|
|
|
6919 putc(0, fd); /* <sectionflags> */
|
|
|
6920
|
|
|
6921 l = STRLEN(spin->si_syllable);
|
|
|
6922 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6923 fwrite(spin->si_syllable, (size_t)l, (size_t)1, fd); /* <syllable> */
|
|
|
6924 }
|
|
|
6925
|
|
|
6926 /* end of <SECTIONS> */
|
|
|
6927 putc(SN_END, fd); /* <sectionend> */
|
|
|
6928
|
|
302
|
6929
|
|
236
|
6930 /*
|
|
339
|
6931 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
6932 */
|
|
323
|
6933 spin->si_memtot = 0;
|
|
339
|
6934 for (round = 1; round <= 3; ++round)
|
|
236
|
6935 {
|
|
339
|
6936 if (round == 1)
|
|
469
|
6937 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
6938 else if (round == 2)
|
|
469
|
6939 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
6940 else
|
|
469
|
6941 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
6942
|
|
346
|
6943 /* Clear the index and wnode fields in the tree. */
|
|
|
6944 clear_node(tree);
|
|
|
6945
|
|
300
|
6946 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
6947 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
6948 * nodes. */
|
|
346
|
6949 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
6950
|
|
300
|
6951 /* number of nodes in 4 bytes */
|
|
|
6952 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
6953 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
6954
|
|
300
|
6955 /* Write the nodes. */
|
|
346
|
6956 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
6957 }
|
|
|
6958
|
|
300
|
6959 fclose(fd);
|
|
236
|
6960 }
|
|
|
6961
|
|
|
6962 /*
|
|
346
|
6963 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
6964 * children. This is needed because they are a union with other items to save
|
|
|
6965 * space.
|
|
|
6966 */
|
|
|
6967 static void
|
|
|
6968 clear_node(node)
|
|
|
6969 wordnode_T *node;
|
|
|
6970 {
|
|
|
6971 wordnode_T *np;
|
|
|
6972
|
|
|
6973 if (node != NULL)
|
|
|
6974 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6975 {
|
|
|
6976 np->wn_u1.index = 0;
|
|
|
6977 np->wn_u2.wnode = NULL;
|
|
|
6978
|
|
|
6979 if (np->wn_byte != NUL)
|
|
|
6980 clear_node(np->wn_child);
|
|
|
6981 }
|
|
|
6982 }
|
|
|
6983
|
|
|
6984
|
|
|
6985 /*
|
|
300
|
6986 * Dump a word tree at node "node".
|
|
|
6987 *
|
|
|
6988 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
6989 * byte recursively write the children.
|
|
|
6990 *
|
|
|
6991 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
6992 * are made about the indexes (so that we don't have to write them in the
|
|
|
6993 * file).
|
|
236
|
6994 *
|
|
300
|
6995 * Returns the number of nodes used.
|
|
236
|
6996 */
|
|
300
|
6997 static int
|
|
346
|
6998 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
6999 FILE *fd; /* NULL when only counting */
|
|
300
|
7000 wordnode_T *node;
|
|
|
7001 int index;
|
|
|
7002 int regionmask;
|
|
339
|
7003 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
7004 {
|
|
300
|
7005 int newindex = index;
|
|
|
7006 int siblingcount = 0;
|
|
|
7007 wordnode_T *np;
|
|
236
|
7008 int flags;
|
|
300
|
7009
|
|
|
7010 /* If "node" is zero the tree is empty. */
|
|
|
7011 if (node == NULL)
|
|
|
7012 return 0;
|
|
|
7013
|
|
|
7014 /* Store the index where this node is written. */
|
|
346
|
7015 node->wn_u1.index = index;
|
|
236
|
7016
|
|
300
|
7017 /* Count the number of siblings. */
|
|
|
7018 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7019 ++siblingcount;
|
|
236
|
7020
|
|
300
|
7021 /* Write the sibling count. */
|
|
|
7022 if (fd != NULL)
|
|
|
7023 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
7024
|
|
300
|
7025 /* Write each sibling byte and optionally extra info. */
|
|
|
7026 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
7027 {
|
|
300
|
7028 if (np->wn_byte == 0)
|
|
|
7029 {
|
|
|
7030 if (fd != NULL)
|
|
|
7031 {
|
|
339
|
7032 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
7033 if (prefixtree)
|
|
300
|
7034 {
|
|
481
|
7035 /* In PREFIXTREE write the required affixID and the
|
|
390
|
7036 * associated condition nr (stored in wn_region). The
|
|
|
7037 * byte value is misused to store the "rare" and "not
|
|
|
7038 * combining" flags */
|
|
455
|
7039 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
7040 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7041 else
|
|
|
7042 {
|
|
390
|
7043 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
7044 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
7045 }
|
|
481
|
7046 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7047 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
7048 }
|
|
|
7049 else
|
|
|
7050 {
|
|
339
|
7051 /* For word trees we write the flag/region items. */
|
|
|
7052 flags = np->wn_flags;
|
|
|
7053 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
7054 flags |= WF_REGION;
|
|
481
|
7055 if (np->wn_affixID != 0)
|
|
|
7056 flags |= WF_AFX;
|
|
339
|
7057 if (flags == 0)
|
|
|
7058 {
|
|
|
7059 /* word without flags or region */
|
|
|
7060 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7061 }
|
|
|
7062 else
|
|
|
7063 {
|
|
390
|
7064 if (np->wn_flags >= 0x100)
|
|
|
7065 {
|
|
|
7066 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
7067 putc(flags, fd); /* <flags> */
|
|
|
7068 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
7069 }
|
|
|
7070 else
|
|
|
7071 {
|
|
|
7072 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
7073 putc(flags, fd); /* <flags> */
|
|
|
7074 }
|
|
339
|
7075 if (flags & WF_REGION)
|
|
|
7076 putc(np->wn_region, fd); /* <region> */
|
|
481
|
7077 if (flags & WF_AFX)
|
|
|
7078 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7079 }
|
|
300
|
7080 }
|
|
|
7081 }
|
|
|
7082 }
|
|
|
7083 else
|
|
|
7084 {
|
|
346
|
7085 if (np->wn_child->wn_u1.index != 0
|
|
|
7086 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
7087 {
|
|
|
7088 /* The child is written elsewhere, write the reference. */
|
|
|
7089 if (fd != NULL)
|
|
|
7090 {
|
|
|
7091 putc(BY_INDEX, fd); /* <byte> */
|
|
|
7092 /* <nodeidx> */
|
|
346
|
7093 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
7094 }
|
|
|
7095 }
|
|
346
|
7096 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
7097 /* We will write the child below and give it an index. */
|
|
346
|
7098 np->wn_child->wn_u2.wnode = node;
|
|
236
|
7099
|
|
300
|
7100 if (fd != NULL)
|
|
|
7101 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
7102 {
|
|
|
7103 EMSG(_(e_write));
|
|
|
7104 return 0;
|
|
|
7105 }
|
|
|
7106 }
|
|
236
|
7107 }
|
|
|
7108
|
|
300
|
7109 /* Space used in the array when reading: one for each sibling and one for
|
|
|
7110 * the count. */
|
|
|
7111 newindex += siblingcount + 1;
|
|
249
|
7112
|
|
300
|
7113 /* Recursively dump the children of each sibling. */
|
|
|
7114 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
7115 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
7116 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
7117 prefixtree);
|
|
249
|
7118
|
|
300
|
7119 return newindex;
|
|
236
|
7120 }
|
|
|
7121
|
|
|
7122
|
|
|
7123 /*
|
|
310
|
7124 * ":mkspell [-ascii] outfile infile ..."
|
|
|
7125 * ":mkspell [-ascii] addfile"
|
|
236
|
7126 */
|
|
|
7127 void
|
|
|
7128 ex_mkspell(eap)
|
|
|
7129 exarg_T *eap;
|
|
|
7130 {
|
|
|
7131 int fcount;
|
|
|
7132 char_u **fnames;
|
|
310
|
7133 char_u *arg = eap->arg;
|
|
|
7134 int ascii = FALSE;
|
|
|
7135
|
|
|
7136 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
7137 {
|
|
|
7138 ascii = TRUE;
|
|
|
7139 arg = skipwhite(arg + 6);
|
|
|
7140 }
|
|
|
7141
|
|
|
7142 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
7143 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
7144 {
|
|
323
|
7145 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
7146 FreeWild(fcount, fnames);
|
|
|
7147 }
|
|
|
7148 }
|
|
|
7149
|
|
|
7150 /*
|
|
|
7151 * Create a Vim spell file from one or more word lists.
|
|
|
7152 * "fnames[0]" is the output file name.
|
|
|
7153 * "fnames[fcount - 1]" is the last input file name.
|
|
|
7154 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
7155 * and ".spl" is appended to make the output file name.
|
|
|
7156 */
|
|
|
7157 static void
|
|
323
|
7158 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
7159 int fcount;
|
|
|
7160 char_u **fnames;
|
|
|
7161 int ascii; /* -ascii argument given */
|
|
|
7162 int overwrite; /* overwrite existing output file */
|
|
323
|
7163 int added_word; /* invoked through "zg" */
|
|
310
|
7164 {
|
|
236
|
7165 char_u fname[MAXPATHL];
|
|
|
7166 char_u wfname[MAXPATHL];
|
|
310
|
7167 char_u **innames;
|
|
|
7168 int incount;
|
|
236
|
7169 afffile_T *(afile[8]);
|
|
|
7170 int i;
|
|
|
7171 int len;
|
|
|
7172 struct stat st;
|
|
255
|
7173 int error = FALSE;
|
|
300
|
7174 spellinfo_T spin;
|
|
|
7175
|
|
|
7176 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
7177 spin.si_verbose = !added_word;
|
|
310
|
7178 spin.si_ascii = ascii;
|
|
323
|
7179 spin.si_followup = TRUE;
|
|
|
7180 spin.si_rem_accents = TRUE;
|
|
|
7181 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
7182 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
7183 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
7184 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
495
|
7185 spin.si_compID = 255; /* start compound ID at maximum, going down */
|
|
310
|
7186
|
|
|
7187 /* default: fnames[0] is output file, following are input files */
|
|
|
7188 innames = &fnames[1];
|
|
|
7189 incount = fcount - 1;
|
|
|
7190
|
|
|
7191 if (fcount >= 1)
|
|
240
|
7192 {
|
|
310
|
7193 len = STRLEN(fnames[0]);
|
|
|
7194 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
7195 {
|
|
|
7196 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
7197 * "path/en.latin1.add.spl". */
|
|
|
7198 innames = &fnames[0];
|
|
|
7199 incount = 1;
|
|
|
7200 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
7201 }
|
|
366
|
7202 else if (fcount == 1)
|
|
|
7203 {
|
|
|
7204 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
7205 innames = &fnames[0];
|
|
|
7206 incount = 1;
|
|
|
7207 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
7208 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
7209 }
|
|
310
|
7210 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
7211 {
|
|
|
7212 /* Name ends in ".spl", use as the file name. */
|
|
323
|
7213 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
7214 }
|
|
|
7215 else
|
|
|
7216 /* Name should be language, make the file name from it. */
|
|
|
7217 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
7218 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
7219
|
|
|
7220 /* Check for .ascii.spl. */
|
|
|
7221 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
7222 spin.si_ascii = TRUE;
|
|
|
7223
|
|
|
7224 /* Check for .add.spl. */
|
|
|
7225 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
7226 spin.si_add = TRUE;
|
|
240
|
7227 }
|
|
|
7228
|
|
310
|
7229 if (incount <= 0)
|
|
236
|
7230 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
7231 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
7232 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
7233 else if (incount > 8)
|
|
236
|
7234 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
7235 else
|
|
|
7236 {
|
|
|
7237 /* Check for overwriting before doing things that may take a lot of
|
|
|
7238 * time. */
|
|
310
|
7239 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
7240 {
|
|
|
7241 EMSG(_(e_exists));
|
|
310
|
7242 return;
|
|
236
|
7243 }
|
|
310
|
7244 if (mch_isdir(wfname))
|
|
236
|
7245 {
|
|
310
|
7246 EMSG2(_(e_isadir2), wfname);
|
|
|
7247 return;
|
|
236
|
7248 }
|
|
|
7249
|
|
|
7250 /*
|
|
|
7251 * Init the aff and dic pointers.
|
|
|
7252 * Get the region names if there are more than 2 arguments.
|
|
|
7253 */
|
|
310
|
7254 for (i = 0; i < incount; ++i)
|
|
236
|
7255 {
|
|
310
|
7256 afile[i] = NULL;
|
|
300
|
7257
|
|
316
|
7258 if (incount > 1)
|
|
236
|
7259 {
|
|
310
|
7260 len = STRLEN(innames[i]);
|
|
|
7261 if (STRLEN(gettail(innames[i])) < 5
|
|
|
7262 || innames[i][len - 3] != '_')
|
|
236
|
7263 {
|
|
310
|
7264 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
7265 return;
|
|
236
|
7266 }
|
|
316
|
7267 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
7268 spin.si_region_name[i * 2 + 1] =
|
|
|
7269 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
7270 }
|
|
|
7271 }
|
|
316
|
7272 spin.si_region_count = incount;
|
|
236
|
7273
|
|
474
|
7274 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
7275 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
7276 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
7277 if (spin.si_foldroot == NULL
|
|
|
7278 || spin.si_keeproot == NULL
|
|
|
7279 || spin.si_prefroot == NULL)
|
|
300
|
7280 {
|
|
469
|
7281 free_blocks(spin.si_blocks);
|
|
310
|
7282 return;
|
|
300
|
7283 }
|
|
|
7284
|
|
351
|
7285 /* When not producing a .add.spl file clear the character table when
|
|
|
7286 * we encounter one in the .aff file. This means we dump the current
|
|
|
7287 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
7288 * better than guessing the contents, the table will match a
|
|
|
7289 * previously loaded spell file. */
|
|
|
7290 if (!spin.si_add)
|
|
|
7291 spin.si_clear_chartab = TRUE;
|
|
|
7292
|
|
236
|
7293 /*
|
|
|
7294 * Read all the .aff and .dic files.
|
|
|
7295 * Text is converted to 'encoding'.
|
|
300
|
7296 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
7297 */
|
|
310
|
7298 for (i = 0; i < incount && !error; ++i)
|
|
236
|
7299 {
|
|
300
|
7300 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
7301 spin.si_region = 1 << i;
|
|
|
7302
|
|
|
7303 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
7304 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
7305 {
|
|
|
7306 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
7307 * "SET" line. */
|
|
474
|
7308 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
7309 if (afile[i] == NULL)
|
|
300
|
7310 error = TRUE;
|
|
|
7311 else
|
|
|
7312 {
|
|
|
7313 /* Read the .dic file and store the words in the trees. */
|
|
|
7314 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
7315 innames[i]);
|
|
474
|
7316 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
7317 error = TRUE;
|
|
|
7318 }
|
|
|
7319 }
|
|
|
7320 else
|
|
|
7321 {
|
|
|
7322 /* No .aff file, try reading the file as a word list. Store
|
|
|
7323 * the words in the trees. */
|
|
474
|
7324 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
7325 error = TRUE;
|
|
|
7326 }
|
|
236
|
7327
|
|
310
|
7328 #ifdef FEAT_MBYTE
|
|
236
|
7329 /* Free any conversion stuff. */
|
|
300
|
7330 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
7331 #endif
|
|
236
|
7332 }
|
|
|
7333
|
|
492
|
7334 if (spin.si_compflags != NULL && spin.si_nobreak)
|
|
|
7335 MSG(_("Warning: both compounding and NOBREAK specified"));
|
|
|
7336
|
|
300
|
7337 if (!error)
|
|
236
|
7338 {
|
|
|
7339 /*
|
|
300
|
7340 * Combine tails in the tree.
|
|
236
|
7341 */
|
|
474
|
7342 if (spin.si_verbose || p_verbose > 2)
|
|
|
7343 {
|
|
|
7344 if (!spin.si_verbose)
|
|
310
|
7345 verbose_enter();
|
|
469
|
7346 MSG(_(msg_compressing));
|
|
310
|
7347 out_flush();
|
|
474
|
7348 if (!spin.si_verbose)
|
|
310
|
7349 verbose_leave();
|
|
|
7350 }
|
|
474
|
7351 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
7352 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
7353 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
7354 }
|
|
|
7355
|
|
300
|
7356 if (!error)
|
|
|
7357 {
|
|
|
7358 /*
|
|
|
7359 * Write the info in the spell file.
|
|
|
7360 */
|
|
474
|
7361 if (spin.si_verbose || p_verbose > 2)
|
|
|
7362 {
|
|
|
7363 if (!spin.si_verbose)
|
|
310
|
7364 verbose_enter();
|
|
366
|
7365 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
7366 out_flush();
|
|
474
|
7367 if (!spin.si_verbose)
|
|
310
|
7368 verbose_leave();
|
|
|
7369 }
|
|
|
7370
|
|
474
|
7371 write_vim_spell(&spin, wfname);
|
|
|
7372
|
|
|
7373 if (spin.si_verbose || p_verbose > 2)
|
|
|
7374 {
|
|
|
7375 if (!spin.si_verbose)
|
|
310
|
7376 verbose_enter();
|
|
|
7377 MSG(_("Done!"));
|
|
|
7378 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
7379 spin.si_memtot);
|
|
310
|
7380 out_flush();
|
|
474
|
7381 if (!spin.si_verbose)
|
|
310
|
7382 verbose_leave();
|
|
|
7383 }
|
|
|
7384
|
|
|
7385 /* If the file is loaded need to reload it. */
|
|
323
|
7386 spell_reload_one(wfname, added_word);
|
|
300
|
7387 }
|
|
|
7388
|
|
|
7389 /* Free the allocated memory. */
|
|
323
|
7390 ga_clear(&spin.si_rep);
|
|
|
7391 ga_clear(&spin.si_sal);
|
|
|
7392 ga_clear(&spin.si_map);
|
|
339
|
7393 ga_clear(&spin.si_prefcond);
|
|
300
|
7394
|
|
|
7395 /* Free the .aff file structures. */
|
|
310
|
7396 for (i = 0; i < incount; ++i)
|
|
|
7397 if (afile[i] != NULL)
|
|
|
7398 spell_free_aff(afile[i]);
|
|
339
|
7399
|
|
|
7400 /* Free all the bits and pieces at once. */
|
|
|
7401 free_blocks(spin.si_blocks);
|
|
236
|
7402 }
|
|
310
|
7403 }
|
|
|
7404
|
|
|
7405
|
|
|
7406 /*
|
|
385
|
7407 * ":[count]spellgood {word}"
|
|
|
7408 * ":[count]spellwrong {word}"
|
|
310
|
7409 */
|
|
|
7410 void
|
|
|
7411 ex_spell(eap)
|
|
|
7412 exarg_T *eap;
|
|
|
7413 {
|
|
381
|
7414 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
7415 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
7416 }
|
|
|
7417
|
|
310
|
7418 /*
|
|
|
7419 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
7420 */
|
|
|
7421 void
|
|
385
|
7422 spell_add_word(word, len, bad, index)
|
|
310
|
7423 char_u *word;
|
|
|
7424 int len;
|
|
|
7425 int bad;
|
|
385
|
7426 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
7427 'spellfile' */
|
|
310
|
7428 {
|
|
|
7429 FILE *fd;
|
|
385
|
7430 buf_T *buf = NULL;
|
|
351
|
7431 int new_spf = FALSE;
|
|
|
7432 struct stat st;
|
|
381
|
7433 char_u *fname;
|
|
385
|
7434 char_u fnamebuf[MAXPATHL];
|
|
|
7435 char_u line[MAXWLEN * 2];
|
|
|
7436 long fpos, fpos_next = 0;
|
|
|
7437 int i;
|
|
|
7438 char_u *spf;
|
|
|
7439
|
|
|
7440 if (index == 0) /* use internal wordlist */
|
|
|
7441 {
|
|
|
7442 if (int_wordlist == NULL)
|
|
|
7443 {
|
|
|
7444 int_wordlist = vim_tempname('s');
|
|
|
7445 if (int_wordlist == NULL)
|
|
381
|
7446 return;
|
|
|
7447 }
|
|
385
|
7448 fname = int_wordlist;
|
|
381
|
7449 }
|
|
310
|
7450 else
|
|
|
7451 {
|
|
381
|
7452 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
7453 if (*curbuf->b_p_spf == NUL)
|
|
|
7454 {
|
|
|
7455 init_spellfile();
|
|
|
7456 new_spf = TRUE;
|
|
|
7457 }
|
|
|
7458
|
|
|
7459 if (*curbuf->b_p_spf == NUL)
|
|
|
7460 {
|
|
|
7461 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
7462 return;
|
|
|
7463 }
|
|
|
7464
|
|
385
|
7465 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
7466 {
|
|
|
7467 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
7468 if (i == index)
|
|
|
7469 break;
|
|
|
7470 if (*spf == NUL)
|
|
|
7471 {
|
|
|
7472 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
7473 return;
|
|
|
7474 }
|
|
|
7475 }
|
|
|
7476
|
|
310
|
7477 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
7478 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
7479 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
7480 buf = NULL;
|
|
|
7481 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
7482 {
|
|
310
|
7483 EMSG(_(e_bufloaded));
|
|
381
|
7484 return;
|
|
|
7485 }
|
|
|
7486
|
|
385
|
7487 fname = fnamebuf;
|
|
|
7488 }
|
|
|
7489
|
|
|
7490 if (bad)
|
|
|
7491 {
|
|
|
7492 /* When the word also appears as good word we need to remove that one,
|
|
|
7493 * since its flags sort before the one with WF_BANNED. */
|
|
|
7494 fd = mch_fopen((char *)fname, "r");
|
|
|
7495 if (fd != NULL)
|
|
|
7496 {
|
|
|
7497 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
7498 {
|
|
|
7499 fpos = fpos_next;
|
|
|
7500 fpos_next = ftell(fd);
|
|
|
7501 if (STRNCMP(word, line, len) == 0
|
|
|
7502 && (line[len] == '/' || line[len] < ' '))
|
|
|
7503 {
|
|
|
7504 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
7505 * the start of the line. Mixing reading and writing
|
|
|
7506 * doesn't work for all systems, close the file first. */
|
|
|
7507 fclose(fd);
|
|
|
7508 fd = mch_fopen((char *)fname, "r+");
|
|
|
7509 if (fd == NULL)
|
|
|
7510 break;
|
|
|
7511 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
7512 fputc('#', fd);
|
|
|
7513 fseek(fd, fpos_next, SEEK_SET);
|
|
|
7514 }
|
|
|
7515 }
|
|
|
7516 fclose(fd);
|
|
|
7517 }
|
|
381
|
7518 }
|
|
|
7519
|
|
|
7520 fd = mch_fopen((char *)fname, "a");
|
|
|
7521 if (fd == NULL && new_spf)
|
|
|
7522 {
|
|
|
7523 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
7524 * We may need to create the "spell" directory first. We already
|
|
|
7525 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
7526 STRCPY(NameBuff, fname);
|
|
|
7527 *gettail_sep(NameBuff) = NUL;
|
|
|
7528 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
7529 {
|
|
|
7530 /* The directory doesn't exist. Try creating it and opening the
|
|
|
7531 * file again. */
|
|
|
7532 vim_mkdir(NameBuff, 0755);
|
|
|
7533 fd = mch_fopen((char *)fname, "a");
|
|
|
7534 }
|
|
|
7535 }
|
|
|
7536
|
|
|
7537 if (fd == NULL)
|
|
|
7538 EMSG2(_(e_notopen), fname);
|
|
|
7539 else
|
|
|
7540 {
|
|
|
7541 if (bad)
|
|
|
7542 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
7543 else
|
|
381
|
7544 fprintf(fd, "%.*s\n", len, word);
|
|
|
7545 fclose(fd);
|
|
|
7546
|
|
|
7547 /* Update the .add.spl file. */
|
|
|
7548 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
7549
|
|
|
7550 /* If the .add file is edited somewhere, reload it. */
|
|
|
7551 if (buf != NULL)
|
|
|
7552 buf_reload(buf);
|
|
|
7553
|
|
|
7554 redraw_all_later(NOT_VALID);
|
|
310
|
7555 }
|
|
|
7556 }
|
|
|
7557
|
|
|
7558 /*
|
|
|
7559 * Initialize 'spellfile' for the current buffer.
|
|
|
7560 */
|
|
|
7561 static void
|
|
|
7562 init_spellfile()
|
|
|
7563 {
|
|
|
7564 char_u buf[MAXPATHL];
|
|
|
7565 int l;
|
|
|
7566 slang_T *sl;
|
|
|
7567 char_u *rtp;
|
|
351
|
7568 char_u *lend;
|
|
310
|
7569
|
|
|
7570 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
7571 {
|
|
351
|
7572 /* Find the end of the language name. Exclude the region. */
|
|
|
7573 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
7574 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
7575 ;
|
|
|
7576
|
|
|
7577 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
7578 * are allowed to write. */
|
|
310
|
7579 rtp = p_rtp;
|
|
|
7580 while (*rtp != NUL)
|
|
|
7581 {
|
|
|
7582 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
7583 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
7584 if (filewritable(buf) == 2)
|
|
|
7585 {
|
|
316
|
7586 /* Use the first language name from 'spelllang' and the
|
|
|
7587 * encoding used in the first loaded .spl file. */
|
|
310
|
7588 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
7589 l = STRLEN(buf);
|
|
|
7590 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
7591 "/spell/%.*s.%s.add",
|
|
351
|
7592 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
7593 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
7594 ? (char_u *)"ascii" : spell_enc());
|
|
|
7595 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
7596 break;
|
|
|
7597 }
|
|
|
7598 }
|
|
|
7599 }
|
|
|
7600 }
|
|
236
|
7601
|
|
300
|
7602
|
|
307
|
7603 /*
|
|
|
7604 * Init the chartab used for spelling for ASCII.
|
|
|
7605 * EBCDIC is not supported!
|
|
|
7606 */
|
|
|
7607 static void
|
|
|
7608 clear_spell_chartab(sp)
|
|
|
7609 spelltab_T *sp;
|
|
|
7610 {
|
|
324
|
7611 int i;
|
|
307
|
7612
|
|
|
7613 /* Init everything to FALSE. */
|
|
|
7614 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
7615 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
7616 for (i = 0; i < 256; ++i)
|
|
324
|
7617 {
|
|
307
|
7618 sp->st_fold[i] = i;
|
|
324
|
7619 sp->st_upper[i] = i;
|
|
|
7620 }
|
|
307
|
7621
|
|
|
7622 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
7623 * that is done separately. */
|
|
|
7624 for (i = '0'; i <= '9'; ++i)
|
|
|
7625 sp->st_isw[i] = TRUE;
|
|
|
7626 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
7627 {
|
|
|
7628 sp->st_isw[i] = TRUE;
|
|
|
7629 sp->st_isu[i] = TRUE;
|
|
|
7630 sp->st_fold[i] = i + 0x20;
|
|
|
7631 }
|
|
|
7632 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
7633 {
|
|
307
|
7634 sp->st_isw[i] = TRUE;
|
|
324
|
7635 sp->st_upper[i] = i - 0x20;
|
|
|
7636 }
|
|
307
|
7637 }
|
|
|
7638
|
|
|
7639 /*
|
|
|
7640 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
7641 * Called once while starting up and when 'encoding' changes.
|
|
|
7642 * The default is to use isalpha(), but the spell file should define the word
|
|
|
7643 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
7644 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
7645 */
|
|
|
7646 void
|
|
|
7647 init_spell_chartab()
|
|
|
7648 {
|
|
|
7649 int i;
|
|
|
7650
|
|
|
7651 did_set_spelltab = FALSE;
|
|
|
7652 clear_spell_chartab(&spelltab);
|
|
|
7653 #ifdef FEAT_MBYTE
|
|
|
7654 if (enc_dbcs)
|
|
|
7655 {
|
|
|
7656 /* DBCS: assume double-wide characters are word characters. */
|
|
|
7657 for (i = 128; i <= 255; ++i)
|
|
|
7658 if (MB_BYTE2LEN(i) == 2)
|
|
|
7659 spelltab.st_isw[i] = TRUE;
|
|
|
7660 }
|
|
324
|
7661 else if (enc_utf8)
|
|
|
7662 {
|
|
|
7663 for (i = 128; i < 256; ++i)
|
|
|
7664 {
|
|
|
7665 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
7666 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
7667 spelltab.st_fold[i] = utf_fold(i);
|
|
|
7668 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
7669 }
|
|
|
7670 }
|
|
307
|
7671 else
|
|
|
7672 #endif
|
|
|
7673 {
|
|
324
|
7674 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
7675 for (i = 128; i < 256; ++i)
|
|
|
7676 {
|
|
|
7677 if (MB_ISUPPER(i))
|
|
|
7678 {
|
|
324
|
7679 spelltab.st_isw[i] = TRUE;
|
|
307
|
7680 spelltab.st_isu[i] = TRUE;
|
|
|
7681 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
7682 }
|
|
324
|
7683 else if (MB_ISLOWER(i))
|
|
|
7684 {
|
|
|
7685 spelltab.st_isw[i] = TRUE;
|
|
|
7686 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
7687 }
|
|
307
|
7688 }
|
|
|
7689 }
|
|
|
7690 }
|
|
|
7691
|
|
|
7692 /*
|
|
|
7693 * Set the spell character tables from strings in the affix file.
|
|
|
7694 */
|
|
|
7695 static int
|
|
|
7696 set_spell_chartab(fol, low, upp)
|
|
|
7697 char_u *fol;
|
|
|
7698 char_u *low;
|
|
|
7699 char_u *upp;
|
|
|
7700 {
|
|
|
7701 /* We build the new tables here first, so that we can compare with the
|
|
|
7702 * previous one. */
|
|
|
7703 spelltab_T new_st;
|
|
|
7704 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
7705 int f, l, u;
|
|
|
7706
|
|
|
7707 clear_spell_chartab(&new_st);
|
|
|
7708
|
|
|
7709 while (*pf != NUL)
|
|
|
7710 {
|
|
|
7711 if (*pl == NUL || *pu == NUL)
|
|
|
7712 {
|
|
|
7713 EMSG(_(e_affform));
|
|
|
7714 return FAIL;
|
|
|
7715 }
|
|
|
7716 #ifdef FEAT_MBYTE
|
|
|
7717 f = mb_ptr2char_adv(&pf);
|
|
|
7718 l = mb_ptr2char_adv(&pl);
|
|
|
7719 u = mb_ptr2char_adv(&pu);
|
|
|
7720 #else
|
|
|
7721 f = *pf++;
|
|
|
7722 l = *pl++;
|
|
|
7723 u = *pu++;
|
|
|
7724 #endif
|
|
|
7725 /* Every character that appears is a word character. */
|
|
|
7726 if (f < 256)
|
|
|
7727 new_st.st_isw[f] = TRUE;
|
|
|
7728 if (l < 256)
|
|
|
7729 new_st.st_isw[l] = TRUE;
|
|
|
7730 if (u < 256)
|
|
|
7731 new_st.st_isw[u] = TRUE;
|
|
|
7732
|
|
|
7733 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
7734 * case-folding */
|
|
|
7735 if (l < 256 && l != f)
|
|
|
7736 {
|
|
|
7737 if (f >= 256)
|
|
|
7738 {
|
|
|
7739 EMSG(_(e_affrange));
|
|
|
7740 return FAIL;
|
|
|
7741 }
|
|
|
7742 new_st.st_fold[l] = f;
|
|
|
7743 }
|
|
|
7744
|
|
|
7745 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
7746 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
7747 * "FOL" . */
|
|
307
|
7748 if (u < 256 && u != f)
|
|
|
7749 {
|
|
|
7750 if (f >= 256)
|
|
|
7751 {
|
|
|
7752 EMSG(_(e_affrange));
|
|
|
7753 return FAIL;
|
|
|
7754 }
|
|
|
7755 new_st.st_fold[u] = f;
|
|
|
7756 new_st.st_isu[u] = TRUE;
|
|
324
|
7757 new_st.st_upper[f] = u;
|
|
307
|
7758 }
|
|
|
7759 }
|
|
|
7760
|
|
|
7761 if (*pl != NUL || *pu != NUL)
|
|
|
7762 {
|
|
|
7763 EMSG(_(e_affform));
|
|
|
7764 return FAIL;
|
|
|
7765 }
|
|
|
7766
|
|
|
7767 return set_spell_finish(&new_st);
|
|
|
7768 }
|
|
|
7769
|
|
|
7770 /*
|
|
|
7771 * Set the spell character tables from strings in the .spl file.
|
|
|
7772 */
|
|
483
|
7773 static void
|
|
389
|
7774 set_spell_charflags(flags, cnt, fol)
|
|
307
|
7775 char_u *flags;
|
|
389
|
7776 int cnt; /* length of "flags" */
|
|
|
7777 char_u *fol;
|
|
307
|
7778 {
|
|
|
7779 /* We build the new tables here first, so that we can compare with the
|
|
|
7780 * previous one. */
|
|
|
7781 spelltab_T new_st;
|
|
|
7782 int i;
|
|
389
|
7783 char_u *p = fol;
|
|
324
|
7784 int c;
|
|
307
|
7785
|
|
|
7786 clear_spell_chartab(&new_st);
|
|
|
7787
|
|
389
|
7788 for (i = 0; i < 128; ++i)
|
|
|
7789 {
|
|
|
7790 if (i < cnt)
|
|
|
7791 {
|
|
|
7792 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
7793 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
7794 }
|
|
|
7795
|
|
|
7796 if (*p != NUL)
|
|
|
7797 {
|
|
307
|
7798 #ifdef FEAT_MBYTE
|
|
389
|
7799 c = mb_ptr2char_adv(&p);
|
|
307
|
7800 #else
|
|
389
|
7801 c = *p++;
|
|
307
|
7802 #endif
|
|
389
|
7803 new_st.st_fold[i + 128] = c;
|
|
|
7804 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
7805 new_st.st_upper[c] = i + 128;
|
|
|
7806 }
|
|
307
|
7807 }
|
|
|
7808
|
|
483
|
7809 (void)set_spell_finish(&new_st);
|
|
307
|
7810 }
|
|
|
7811
|
|
|
7812 static int
|
|
|
7813 set_spell_finish(new_st)
|
|
|
7814 spelltab_T *new_st;
|
|
|
7815 {
|
|
|
7816 int i;
|
|
|
7817
|
|
|
7818 if (did_set_spelltab)
|
|
|
7819 {
|
|
|
7820 /* check that it's the same table */
|
|
|
7821 for (i = 0; i < 256; ++i)
|
|
|
7822 {
|
|
|
7823 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
7824 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
7825 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
7826 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
7827 {
|
|
|
7828 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
7829 return FAIL;
|
|
|
7830 }
|
|
|
7831 }
|
|
|
7832 }
|
|
|
7833 else
|
|
|
7834 {
|
|
|
7835 /* copy the new spelltab into the one being used */
|
|
|
7836 spelltab = *new_st;
|
|
|
7837 did_set_spelltab = TRUE;
|
|
|
7838 }
|
|
|
7839
|
|
|
7840 return OK;
|
|
|
7841 }
|
|
|
7842
|
|
|
7843 /*
|
|
358
|
7844 * Return TRUE if "p" points to a word character.
|
|
366
|
7845 * As a special case we see "midword" characters as word character when it is
|
|
358
|
7846 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
7847 * Thus this only works properly when past the first character of the word.
|
|
358
|
7848 */
|
|
|
7849 static int
|
|
376
|
7850 spell_iswordp(p, buf)
|
|
358
|
7851 char_u *p;
|
|
376
|
7852 buf_T *buf; /* buffer used */
|
|
358
|
7853 {
|
|
366
|
7854 #ifdef FEAT_MBYTE
|
|
358
|
7855 char_u *s;
|
|
366
|
7856 int l;
|
|
|
7857 int c;
|
|
|
7858
|
|
|
7859 if (has_mbyte)
|
|
|
7860 {
|
|
|
7861 l = MB_BYTE2LEN(*p);
|
|
358
|
7862 s = p;
|
|
366
|
7863 if (l == 1)
|
|
|
7864 {
|
|
|
7865 /* be quick for ASCII */
|
|
376
|
7866 if (buf->b_spell_ismw[*p])
|
|
366
|
7867 {
|
|
|
7868 s = p + 1; /* skip a mid-word character */
|
|
|
7869 l = MB_BYTE2LEN(*s);
|
|
|
7870 }
|
|
|
7871 }
|
|
|
7872 else
|
|
|
7873 {
|
|
|
7874 c = mb_ptr2char(p);
|
|
376
|
7875 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
7876 : (buf->b_spell_ismw_mb != NULL
|
|
|
7877 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
7878 {
|
|
|
7879 s = p + l;
|
|
|
7880 l = MB_BYTE2LEN(*s);
|
|
|
7881 }
|
|
|
7882 }
|
|
|
7883
|
|
390
|
7884 c = mb_ptr2char(s);
|
|
|
7885 if (c > 255)
|
|
366
|
7886 return mb_get_class(s) >= 2;
|
|
390
|
7887 return spelltab.st_isw[c];
|
|
366
|
7888 }
|
|
358
|
7889 #endif
|
|
366
|
7890
|
|
376
|
7891 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
7892 }
|
|
|
7893
|
|
|
7894 /*
|
|
|
7895 * Return TRUE if "p" points to a word character.
|
|
|
7896 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
7897 */
|
|
|
7898 static int
|
|
|
7899 spell_iswordp_nmw(p)
|
|
|
7900 char_u *p;
|
|
|
7901 {
|
|
|
7902 #ifdef FEAT_MBYTE
|
|
390
|
7903 int c;
|
|
|
7904
|
|
|
7905 if (has_mbyte)
|
|
|
7906 {
|
|
|
7907 c = mb_ptr2char(p);
|
|
|
7908 if (c > 255)
|
|
|
7909 return mb_get_class(p) >= 2;
|
|
|
7910 return spelltab.st_isw[c];
|
|
|
7911 }
|
|
376
|
7912 #endif
|
|
|
7913 return spelltab.st_isw[*p];
|
|
358
|
7914 }
|
|
|
7915
|
|
372
|
7916 #ifdef FEAT_MBYTE
|
|
|
7917 /*
|
|
|
7918 * Return TRUE if "p" points to a word character.
|
|
|
7919 * Wide version of spell_iswordp().
|
|
|
7920 */
|
|
|
7921 static int
|
|
376
|
7922 spell_iswordp_w(p, buf)
|
|
372
|
7923 int *p;
|
|
376
|
7924 buf_T *buf;
|
|
372
|
7925 {
|
|
|
7926 int *s;
|
|
|
7927
|
|
376
|
7928 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
7929 : (buf->b_spell_ismw_mb != NULL
|
|
|
7930 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
7931 s = p + 1;
|
|
|
7932 else
|
|
|
7933 s = p;
|
|
|
7934
|
|
390
|
7935 if (*s > 255)
|
|
372
|
7936 {
|
|
|
7937 if (enc_utf8)
|
|
|
7938 return utf_class(*s) >= 2;
|
|
|
7939 if (enc_dbcs)
|
|
|
7940 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
7941 return 0;
|
|
|
7942 }
|
|
|
7943 return spelltab.st_isw[*s];
|
|
|
7944 }
|
|
|
7945 #endif
|
|
|
7946
|
|
358
|
7947 /*
|
|
339
|
7948 * Write the table with prefix conditions to the .spl file.
|
|
483
|
7949 * When "fd" is NULL only count the length of what is written.
|
|
|
7950 */
|
|
|
7951 static int
|
|
339
|
7952 write_spell_prefcond(fd, gap)
|
|
|
7953 FILE *fd;
|
|
|
7954 garray_T *gap;
|
|
|
7955 {
|
|
|
7956 int i;
|
|
|
7957 char_u *p;
|
|
|
7958 int len;
|
|
483
|
7959 int totlen;
|
|
|
7960
|
|
|
7961 if (fd != NULL)
|
|
|
7962 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
7963
|
|
|
7964 totlen = 2 + gap->ga_len; /* length of <prefcondcnt> and <condlen> bytes */
|
|
339
|
7965
|
|
|
7966 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7967 {
|
|
|
7968 /* <prefcond> : <condlen> <condstr> */
|
|
|
7969 p = ((char_u **)gap->ga_data)[i];
|
|
483
|
7970 if (p != NULL)
|
|
339
|
7971 {
|
|
|
7972 len = STRLEN(p);
|
|
483
|
7973 if (fd != NULL)
|
|
|
7974 {
|
|
|
7975 fputc(len, fd);
|
|
|
7976 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
7977 }
|
|
|
7978 totlen += len;
|
|
|
7979 }
|
|
|
7980 else if (fd != NULL)
|
|
|
7981 fputc(0, fd);
|
|
|
7982 }
|
|
|
7983
|
|
|
7984 return totlen;
|
|
307
|
7985 }
|
|
|
7986
|
|
|
7987 /*
|
|
324
|
7988 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
7989 * Uses the character definitions from the .spl file.
|
|
307
|
7990 * When using a multi-byte 'encoding' the length may change!
|
|
|
7991 * Returns FAIL when something wrong.
|
|
|
7992 */
|
|
|
7993 static int
|
|
324
|
7994 spell_casefold(str, len, buf, buflen)
|
|
|
7995 char_u *str;
|
|
307
|
7996 int len;
|
|
|
7997 char_u *buf;
|
|
|
7998 int buflen;
|
|
|
7999 {
|
|
|
8000 int i;
|
|
|
8001
|
|
|
8002 if (len >= buflen)
|
|
|
8003 {
|
|
|
8004 buf[0] = NUL;
|
|
|
8005 return FAIL; /* result will not fit */
|
|
|
8006 }
|
|
|
8007
|
|
|
8008 #ifdef FEAT_MBYTE
|
|
|
8009 if (has_mbyte)
|
|
|
8010 {
|
|
324
|
8011 int outi = 0;
|
|
|
8012 char_u *p;
|
|
307
|
8013 int c;
|
|
|
8014
|
|
|
8015 /* Fold one character at a time. */
|
|
324
|
8016 for (p = str; p < str + len; )
|
|
307
|
8017 {
|
|
|
8018 if (outi + MB_MAXBYTES > buflen)
|
|
|
8019 {
|
|
|
8020 buf[outi] = NUL;
|
|
|
8021 return FAIL;
|
|
|
8022 }
|
|
474
|
8023 c = mb_cptr2char_adv(&p);
|
|
324
|
8024 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
8025 }
|
|
|
8026 buf[outi] = NUL;
|
|
|
8027 }
|
|
|
8028 else
|
|
|
8029 #endif
|
|
|
8030 {
|
|
|
8031 /* Be quick for non-multibyte encodings. */
|
|
|
8032 for (i = 0; i < len; ++i)
|
|
324
|
8033 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
8034 buf[i] = NUL;
|
|
|
8035 }
|
|
|
8036
|
|
|
8037 return OK;
|
|
|
8038 }
|
|
|
8039
|
|
372
|
8040 #define SPS_BEST 1
|
|
|
8041 #define SPS_FAST 2
|
|
|
8042 #define SPS_DOUBLE 4
|
|
|
8043
|
|
|
8044 static int sps_flags = SPS_BEST;
|
|
483
|
8045 static int sps_limit = 9999;
|
|
372
|
8046
|
|
|
8047 /*
|
|
|
8048 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
483
|
8049 * Sets "sps_flags" and "sps_limit".
|
|
372
|
8050 */
|
|
|
8051 int
|
|
|
8052 spell_check_sps()
|
|
|
8053 {
|
|
|
8054 char_u *p;
|
|
483
|
8055 char_u *s;
|
|
372
|
8056 char_u buf[MAXPATHL];
|
|
|
8057 int f;
|
|
|
8058
|
|
|
8059 sps_flags = 0;
|
|
483
|
8060 sps_limit = 9999;
|
|
372
|
8061
|
|
|
8062 for (p = p_sps; *p != NUL; )
|
|
|
8063 {
|
|
|
8064 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
8065
|
|
|
8066 f = 0;
|
|
483
|
8067 if (VIM_ISDIGIT(*buf))
|
|
|
8068 {
|
|
|
8069 s = buf;
|
|
|
8070 sps_limit = getdigits(&s);
|
|
|
8071 if (*s != NUL && !VIM_ISDIGIT(*s))
|
|
|
8072 f = -1;
|
|
|
8073 }
|
|
|
8074 else if (STRCMP(buf, "best") == 0)
|
|
372
|
8075 f = SPS_BEST;
|
|
|
8076 else if (STRCMP(buf, "fast") == 0)
|
|
|
8077 f = SPS_FAST;
|
|
|
8078 else if (STRCMP(buf, "double") == 0)
|
|
|
8079 f = SPS_DOUBLE;
|
|
|
8080 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
8081 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
8082 f = -1;
|
|
|
8083
|
|
|
8084 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
8085 {
|
|
|
8086 sps_flags = SPS_BEST;
|
|
483
|
8087 sps_limit = 9999;
|
|
372
|
8088 return FAIL;
|
|
|
8089 }
|
|
|
8090 if (f != 0)
|
|
|
8091 sps_flags = f;
|
|
|
8092 }
|
|
|
8093
|
|
|
8094 if (sps_flags == 0)
|
|
|
8095 sps_flags = SPS_BEST;
|
|
|
8096
|
|
|
8097 return OK;
|
|
|
8098 }
|
|
|
8099
|
|
|
8100 /* Remember what "z?" replaced. */
|
|
|
8101 static char_u *repl_from = NULL;
|
|
|
8102 static char_u *repl_to = NULL;
|
|
|
8103
|
|
323
|
8104 /*
|
|
|
8105 * "z?": Find badly spelled word under or after the cursor.
|
|
|
8106 * Give suggestions for the properly spelled word.
|
|
485
|
8107 * When "count" is non-zero use that suggestion.
|
|
323
|
8108 */
|
|
|
8109 void
|
|
485
|
8110 spell_suggest(count)
|
|
|
8111 int count;
|
|
323
|
8112 {
|
|
|
8113 char_u *line;
|
|
|
8114 pos_T prev_cursor = curwin->w_cursor;
|
|
|
8115 char_u wcopy[MAXWLEN + 2];
|
|
|
8116 char_u *p;
|
|
|
8117 int i;
|
|
|
8118 int c;
|
|
|
8119 suginfo_T sug;
|
|
|
8120 suggest_T *stp;
|
|
372
|
8121 int mouse_used;
|
|
388
|
8122 int need_cap;
|
|
483
|
8123 int limit;
|
|
485
|
8124 int selected = count;
|
|
323
|
8125
|
|
344
|
8126 /* Find the start of the badly spelled word. */
|
|
495
|
8127 if (spell_move_to(FORWARD, TRUE, TRUE) == 0
|
|
346
|
8128 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
8129 {
|
|
|
8130 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
8131 return;
|
|
|
8132
|
|
|
8133 /* No bad word or it starts after the cursor: use the word under the
|
|
|
8134 * cursor. */
|
|
|
8135 curwin->w_cursor = prev_cursor;
|
|
|
8136 line = ml_get_curline();
|
|
|
8137 p = line + curwin->w_cursor.col;
|
|
|
8138 /* Backup to before start of word. */
|
|
390
|
8139 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
8140 mb_ptr_back(line, p);
|
|
|
8141 /* Forward to start of word. */
|
|
390
|
8142 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
8143 mb_ptr_adv(p);
|
|
|
8144
|
|
390
|
8145 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
8146 {
|
|
|
8147 beep_flush();
|
|
|
8148 return;
|
|
|
8149 }
|
|
|
8150 curwin->w_cursor.col = p - line;
|
|
323
|
8151 }
|
|
|
8152
|
|
344
|
8153 /* Get the word and its length. */
|
|
|
8154
|
|
388
|
8155 /* Figure out if the word should be capitalised. */
|
|
475
|
8156 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
8157
|
|
|
8158 line = ml_get_curline();
|
|
388
|
8159
|
|
483
|
8160 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in
|
|
|
8161 * 'spellsuggest', whatever is smaller. */
|
|
|
8162 if (sps_limit > (int)Rows - 2)
|
|
|
8163 limit = (int)Rows - 2;
|
|
|
8164 else
|
|
|
8165 limit = sps_limit;
|
|
|
8166 spell_find_suggest(line + curwin->w_cursor.col, &sug, limit,
|
|
388
|
8167 TRUE, need_cap);
|
|
323
|
8168
|
|
|
8169 if (sug.su_ga.ga_len == 0)
|
|
|
8170 MSG(_("Sorry, no suggestions"));
|
|
485
|
8171 else if (count > 0)
|
|
|
8172 {
|
|
|
8173 if (count > sug.su_ga.ga_len)
|
|
|
8174 smsg((char_u *)_("Sorry, only %ld suggestions"),
|
|
|
8175 (long)sug.su_ga.ga_len);
|
|
|
8176 }
|
|
323
|
8177 else
|
|
|
8178 {
|
|
372
|
8179 vim_free(repl_from);
|
|
|
8180 repl_from = NULL;
|
|
|
8181 vim_free(repl_to);
|
|
|
8182 repl_to = NULL;
|
|
|
8183
|
|
474
|
8184 #ifdef FEAT_RIGHTLEFT
|
|
|
8185 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
8186 cmdmsg_rl = curwin->w_p_rl;
|
|
|
8187 if (cmdmsg_rl)
|
|
|
8188 msg_col = Columns - 1;
|
|
|
8189 #endif
|
|
|
8190
|
|
323
|
8191 /* List the suggestions. */
|
|
|
8192 msg_start();
|
|
372
|
8193 lines_left = Rows; /* avoid more prompt */
|
|
323
|
8194 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
8195 sug.su_badlen, sug.su_badptr);
|
|
474
|
8196 #ifdef FEAT_RIGHTLEFT
|
|
|
8197 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
8198 {
|
|
|
8199 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
8200 * untranslated message rightleft. */
|
|
|
8201 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
8202 sug.su_badlen, sug.su_badptr);
|
|
|
8203 }
|
|
|
8204 #endif
|
|
323
|
8205 msg_puts(IObuff);
|
|
|
8206 msg_clr_eos();
|
|
|
8207 msg_putchar('\n');
|
|
346
|
8208
|
|
323
|
8209 msg_scroll = TRUE;
|
|
|
8210 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
8211 {
|
|
344
|
8212 stp = &SUG(sug.su_ga, i);
|
|
323
|
8213
|
|
|
8214 /* The suggested word may replace only part of the bad word, add
|
|
|
8215 * the not replaced part. */
|
|
|
8216 STRCPY(wcopy, stp->st_word);
|
|
|
8217 if (sug.su_badlen > stp->st_orglen)
|
|
|
8218 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
8219 sug.su_badptr + stp->st_orglen,
|
|
|
8220 sug.su_badlen - stp->st_orglen);
|
|
474
|
8221 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
8222 #ifdef FEAT_RIGHTLEFT
|
|
|
8223 if (cmdmsg_rl)
|
|
|
8224 rl_mirror(IObuff);
|
|
|
8225 #endif
|
|
|
8226 msg_puts(IObuff);
|
|
|
8227
|
|
|
8228 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
8229 msg_puts(IObuff);
|
|
|
8230
|
|
|
8231 /* The word may replace more than "su_badlen". */
|
|
|
8232 if (sug.su_badlen < stp->st_orglen)
|
|
|
8233 {
|
|
|
8234 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
8235 stp->st_orglen, sug.su_badptr);
|
|
|
8236 msg_puts(IObuff);
|
|
|
8237 }
|
|
|
8238
|
|
324
|
8239 if (p_verbose > 0)
|
|
344
|
8240 {
|
|
346
|
8241 /* Add the score. */
|
|
351
|
8242 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
8243 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
8244 stp->st_salscore ? "s " : "",
|
|
|
8245 stp->st_score, stp->st_altscore);
|
|
|
8246 else
|
|
474
|
8247 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
8248 stp->st_score);
|
|
474
|
8249 #ifdef FEAT_RIGHTLEFT
|
|
|
8250 if (cmdmsg_rl)
|
|
|
8251 /* Mirror the numbers, but keep the leading space. */
|
|
|
8252 rl_mirror(IObuff + 1);
|
|
|
8253 #endif
|
|
346
|
8254 msg_advance(30);
|
|
|
8255 msg_puts(IObuff);
|
|
344
|
8256 }
|
|
323
|
8257 msg_putchar('\n');
|
|
|
8258 }
|
|
|
8259
|
|
474
|
8260 #ifdef FEAT_RIGHTLEFT
|
|
|
8261 cmdmsg_rl = FALSE;
|
|
|
8262 msg_col = 0;
|
|
|
8263 #endif
|
|
323
|
8264 /* Ask for choice. */
|
|
485
|
8265 selected = prompt_for_number(&mouse_used);
|
|
372
|
8266 if (mouse_used)
|
|
485
|
8267 selected -= lines_left;
|
|
|
8268 }
|
|
|
8269
|
|
|
8270 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
|
8271 {
|
|
|
8272 /* Save the from and to text for :spellrepall. */
|
|
|
8273 stp = &SUG(sug.su_ga, selected - 1);
|
|
|
8274 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
8275 repl_to = vim_strsave(stp->st_word);
|
|
|
8276
|
|
|
8277 /* Replace the word. */
|
|
|
8278 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
8279 if (p != NULL)
|
|
|
8280 {
|
|
|
8281 c = sug.su_badptr - line;
|
|
|
8282 mch_memmove(p, line, c);
|
|
|
8283 STRCPY(p + c, stp->st_word);
|
|
|
8284 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
8285 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
8286 curwin->w_cursor.col = c;
|
|
|
8287 changed_bytes(curwin->w_cursor.lnum, c);
|
|
|
8288
|
|
|
8289 /* For redo we use a change-word command. */
|
|
|
8290 ResetRedobuff();
|
|
|
8291 AppendToRedobuff((char_u *)"ciw");
|
|
|
8292 AppendToRedobuff(stp->st_word);
|
|
|
8293 AppendCharToRedobuff(ESC);
|
|
|
8294 }
|
|
|
8295 }
|
|
|
8296 else
|
|
|
8297 curwin->w_cursor = prev_cursor;
|
|
323
|
8298
|
|
344
|
8299 spell_find_cleanup(&sug);
|
|
|
8300 }
|
|
|
8301
|
|
|
8302 /*
|
|
475
|
8303 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
8304 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
8305 */
|
|
|
8306 static int
|
|
|
8307 check_need_cap(lnum, col)
|
|
|
8308 linenr_T lnum;
|
|
|
8309 colnr_T col;
|
|
|
8310 {
|
|
|
8311 int need_cap = FALSE;
|
|
|
8312 char_u *line;
|
|
|
8313 char_u *line_copy = NULL;
|
|
|
8314 char_u *p;
|
|
|
8315 colnr_T endcol;
|
|
|
8316 regmatch_T regmatch;
|
|
|
8317
|
|
|
8318 if (curbuf->b_cap_prog == NULL)
|
|
|
8319 return FALSE;
|
|
|
8320
|
|
|
8321 line = ml_get_curline();
|
|
|
8322 endcol = 0;
|
|
|
8323 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
8324 {
|
|
|
8325 /* At start of line, check if previous line is empty or sentence
|
|
|
8326 * ends there. */
|
|
|
8327 if (lnum == 1)
|
|
|
8328 need_cap = TRUE;
|
|
|
8329 else
|
|
|
8330 {
|
|
|
8331 line = ml_get(lnum - 1);
|
|
|
8332 if (*skipwhite(line) == NUL)
|
|
|
8333 need_cap = TRUE;
|
|
|
8334 else
|
|
|
8335 {
|
|
|
8336 /* Append a space in place of the line break. */
|
|
|
8337 line_copy = concat_str(line, (char_u *)" ");
|
|
|
8338 line = line_copy;
|
|
|
8339 endcol = STRLEN(line);
|
|
|
8340 }
|
|
|
8341 }
|
|
|
8342 }
|
|
|
8343 else
|
|
|
8344 endcol = col;
|
|
|
8345
|
|
|
8346 if (endcol > 0)
|
|
|
8347 {
|
|
|
8348 /* Check if sentence ends before the bad word. */
|
|
|
8349 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
8350 regmatch.rm_ic = FALSE;
|
|
|
8351 p = line + endcol;
|
|
|
8352 for (;;)
|
|
|
8353 {
|
|
|
8354 mb_ptr_back(line, p);
|
|
|
8355 if (p == line || spell_iswordp_nmw(p))
|
|
|
8356 break;
|
|
|
8357 if (vim_regexec(®match, p, 0)
|
|
|
8358 && regmatch.endp[0] == line + endcol)
|
|
|
8359 {
|
|
|
8360 need_cap = TRUE;
|
|
|
8361 break;
|
|
|
8362 }
|
|
|
8363 }
|
|
|
8364 }
|
|
|
8365
|
|
|
8366 vim_free(line_copy);
|
|
|
8367
|
|
|
8368 return need_cap;
|
|
|
8369 }
|
|
|
8370
|
|
|
8371
|
|
|
8372 /*
|
|
372
|
8373 * ":spellrepall"
|
|
|
8374 */
|
|
|
8375 /*ARGSUSED*/
|
|
|
8376 void
|
|
|
8377 ex_spellrepall(eap)
|
|
|
8378 exarg_T *eap;
|
|
|
8379 {
|
|
|
8380 pos_T pos = curwin->w_cursor;
|
|
|
8381 char_u *frompat;
|
|
|
8382 int addlen;
|
|
|
8383 char_u *line;
|
|
|
8384 char_u *p;
|
|
|
8385 int save_ws = p_ws;
|
|
483
|
8386 linenr_T prev_lnum = 0;
|
|
372
|
8387
|
|
|
8388 if (repl_from == NULL || repl_to == NULL)
|
|
|
8389 {
|
|
|
8390 EMSG(_("E752: No previous spell replacement"));
|
|
|
8391 return;
|
|
|
8392 }
|
|
|
8393 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
8394
|
|
|
8395 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
8396 if (frompat == NULL)
|
|
|
8397 return;
|
|
|
8398 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
8399 p_ws = FALSE;
|
|
|
8400
|
|
483
|
8401 sub_nsubs = 0;
|
|
|
8402 sub_nlines = 0;
|
|
372
|
8403 curwin->w_cursor.lnum = 0;
|
|
|
8404 while (!got_int)
|
|
|
8405 {
|
|
|
8406 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
8407 || u_save_cursor() == FAIL)
|
|
|
8408 break;
|
|
|
8409
|
|
|
8410 /* Only replace when the right word isn't there yet. This happens
|
|
|
8411 * when changing "etc" to "etc.". */
|
|
|
8412 line = ml_get_curline();
|
|
|
8413 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
8414 repl_to, STRLEN(repl_to)) != 0)
|
|
|
8415 {
|
|
|
8416 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
8417 if (p == NULL)
|
|
|
8418 break;
|
|
|
8419 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
8420 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
8421 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
8422 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
8423 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
483
|
8424
|
|
|
8425 if (curwin->w_cursor.lnum != prev_lnum)
|
|
|
8426 {
|
|
|
8427 ++sub_nlines;
|
|
|
8428 prev_lnum = curwin->w_cursor.lnum;
|
|
|
8429 }
|
|
|
8430 ++sub_nsubs;
|
|
372
|
8431 }
|
|
|
8432 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
8433 }
|
|
|
8434
|
|
|
8435 p_ws = save_ws;
|
|
|
8436 curwin->w_cursor = pos;
|
|
|
8437 vim_free(frompat);
|
|
|
8438
|
|
483
|
8439 if (sub_nsubs == 0)
|
|
372
|
8440 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
483
|
8441 else
|
|
|
8442 do_sub_msg(FALSE);
|
|
372
|
8443 }
|
|
|
8444
|
|
|
8445 /*
|
|
344
|
8446 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
8447 * a list of allocated strings.
|
|
|
8448 */
|
|
|
8449 void
|
|
475
|
8450 spell_suggest_list(gap, word, maxcount, need_cap)
|
|
344
|
8451 garray_T *gap;
|
|
|
8452 char_u *word;
|
|
|
8453 int maxcount; /* maximum nr of suggestions */
|
|
475
|
8454 int need_cap; /* 'spellcapcheck' matched */
|
|
344
|
8455 {
|
|
|
8456 suginfo_T sug;
|
|
|
8457 int i;
|
|
|
8458 suggest_T *stp;
|
|
|
8459 char_u *wcopy;
|
|
|
8460
|
|
475
|
8461 spell_find_suggest(word, &sug, maxcount, FALSE, need_cap);
|
|
344
|
8462
|
|
|
8463 /* Make room in "gap". */
|
|
|
8464 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
8465 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
8466 return;
|
|
|
8467
|
|
323
|
8468 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
8469 {
|
|
|
8470 stp = &SUG(sug.su_ga, i);
|
|
|
8471
|
|
|
8472 /* The suggested word may replace only part of "word", add the not
|
|
|
8473 * replaced part. */
|
|
|
8474 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
8475 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
8476 if (wcopy == NULL)
|
|
|
8477 break;
|
|
|
8478 STRCPY(wcopy, stp->st_word);
|
|
|
8479 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
8480 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
8481 }
|
|
|
8482
|
|
|
8483 spell_find_cleanup(&sug);
|
|
|
8484 }
|
|
|
8485
|
|
|
8486 /*
|
|
|
8487 * Find spell suggestions for the word at the start of "badptr".
|
|
|
8488 * Return the suggestions in "su->su_ga".
|
|
|
8489 * The maximum number of suggestions is "maxcount".
|
|
|
8490 * Note: does use info for the current window.
|
|
|
8491 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
8492 */
|
|
|
8493 static void
|
|
388
|
8494 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
8495 char_u *badptr;
|
|
|
8496 suginfo_T *su;
|
|
|
8497 int maxcount;
|
|
358
|
8498 int banbadword; /* don't include badword in suggestions */
|
|
388
|
8499 int need_cap; /* word should start with capital */
|
|
344
|
8500 {
|
|
385
|
8501 int attr = 0;
|
|
372
|
8502 char_u buf[MAXPATHL];
|
|
|
8503 char_u *p;
|
|
|
8504 int do_combine = FALSE;
|
|
|
8505 char_u *sps_copy;
|
|
|
8506 #ifdef FEAT_EVAL
|
|
|
8507 static int expr_busy = FALSE;
|
|
|
8508 #endif
|
|
385
|
8509 int c;
|
|
344
|
8510
|
|
|
8511 /*
|
|
|
8512 * Set the info in "*su".
|
|
|
8513 */
|
|
|
8514 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
8515 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
8516 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
8517 if (*badptr == NUL)
|
|
|
8518 return;
|
|
344
|
8519 hash_init(&su->su_banned);
|
|
|
8520
|
|
|
8521 su->su_badptr = badptr;
|
|
385
|
8522 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
8523 su->su_maxcount = maxcount;
|
|
372
|
8524 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
8525
|
|
|
8526 if (su->su_badlen >= MAXWLEN)
|
|
|
8527 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
8528 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
8529 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
8530 su->su_fbadword, MAXWLEN);
|
|
346
|
8531 /* get caps flags for bad word */
|
|
474
|
8532 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
8533 su->su_badptr + su->su_badlen);
|
|
388
|
8534 if (need_cap)
|
|
|
8535 su->su_badflags |= WF_ONECAP;
|
|
344
|
8536
|
|
385
|
8537 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
8538 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
8539 * for that. */
|
|
455
|
8540 c = PTR2CHAR(su->su_badptr);
|
|
385
|
8541 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
8542 {
|
|
|
8543 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
8544 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
8545 0, TRUE);
|
|
|
8546 }
|
|
|
8547
|
|
344
|
8548 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
8549 if (banbadword)
|
|
|
8550 add_banned(su, su->su_badword);
|
|
344
|
8551
|
|
372
|
8552 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
8553 sps_copy = vim_strsave(p_sps);
|
|
|
8554 if (sps_copy == NULL)
|
|
|
8555 return;
|
|
|
8556
|
|
|
8557 /* Loop over the items in 'spellsuggest'. */
|
|
|
8558 for (p = sps_copy; *p != NUL; )
|
|
|
8559 {
|
|
|
8560 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
8561
|
|
|
8562 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
8563 {
|
|
|
8564 #ifdef FEAT_EVAL
|
|
375
|
8565 /* Evaluate an expression. Skip this when called recursively,
|
|
|
8566 * when using spellsuggest() in the expression. */
|
|
372
|
8567 if (!expr_busy)
|
|
|
8568 {
|
|
|
8569 expr_busy = TRUE;
|
|
|
8570 spell_suggest_expr(su, buf + 5);
|
|
|
8571 expr_busy = FALSE;
|
|
|
8572 }
|
|
|
8573 #endif
|
|
|
8574 }
|
|
|
8575 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
8576 /* Use list of suggestions in a file. */
|
|
|
8577 spell_suggest_file(su, buf + 5);
|
|
|
8578 else
|
|
|
8579 {
|
|
|
8580 /* Use internal method. */
|
|
|
8581 spell_suggest_intern(su);
|
|
|
8582 if (sps_flags & SPS_DOUBLE)
|
|
|
8583 do_combine = TRUE;
|
|
|
8584 }
|
|
|
8585 }
|
|
|
8586
|
|
|
8587 vim_free(sps_copy);
|
|
|
8588
|
|
|
8589 if (do_combine)
|
|
|
8590 /* Combine the two list of suggestions. This must be done last,
|
|
|
8591 * because sorting changes the order again. */
|
|
|
8592 score_combine(su);
|
|
|
8593 }
|
|
|
8594
|
|
|
8595 #ifdef FEAT_EVAL
|
|
|
8596 /*
|
|
|
8597 * Find suggestions by evaluating expression "expr".
|
|
|
8598 */
|
|
|
8599 static void
|
|
|
8600 spell_suggest_expr(su, expr)
|
|
|
8601 suginfo_T *su;
|
|
|
8602 char_u *expr;
|
|
|
8603 {
|
|
|
8604 list_T *list;
|
|
|
8605 listitem_T *li;
|
|
|
8606 int score;
|
|
|
8607 char_u *p;
|
|
|
8608
|
|
|
8609 /* The work is split up in a few parts to avoid having to export
|
|
|
8610 * suginfo_T.
|
|
|
8611 * First evaluate the expression and get the resulting list. */
|
|
|
8612 list = eval_spell_expr(su->su_badword, expr);
|
|
|
8613 if (list != NULL)
|
|
|
8614 {
|
|
|
8615 /* Loop over the items in the list. */
|
|
|
8616 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
8617 if (li->li_tv.v_type == VAR_LIST)
|
|
|
8618 {
|
|
|
8619 /* Get the word and the score from the items. */
|
|
|
8620 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
8621 if (score >= 0)
|
|
|
8622 add_suggestion(su, &su->su_ga, p,
|
|
|
8623 su->su_badlen, score, 0, TRUE);
|
|
|
8624 }
|
|
|
8625 list_unref(list);
|
|
|
8626 }
|
|
|
8627
|
|
|
8628 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8629 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8630 }
|
|
|
8631 #endif
|
|
|
8632
|
|
|
8633 /*
|
|
474
|
8634 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
8635 */
|
|
|
8636 static void
|
|
|
8637 spell_suggest_file(su, fname)
|
|
|
8638 suginfo_T *su;
|
|
|
8639 char_u *fname;
|
|
|
8640 {
|
|
|
8641 FILE *fd;
|
|
|
8642 char_u line[MAXWLEN * 2];
|
|
|
8643 char_u *p;
|
|
|
8644 int len;
|
|
|
8645 char_u cword[MAXWLEN];
|
|
|
8646
|
|
|
8647 /* Open the file. */
|
|
|
8648 fd = mch_fopen((char *)fname, "r");
|
|
|
8649 if (fd == NULL)
|
|
|
8650 {
|
|
|
8651 EMSG2(_(e_notopen), fname);
|
|
|
8652 return;
|
|
|
8653 }
|
|
|
8654
|
|
|
8655 /* Read it line by line. */
|
|
|
8656 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
8657 {
|
|
|
8658 line_breakcheck();
|
|
|
8659
|
|
|
8660 p = vim_strchr(line, '/');
|
|
|
8661 if (p == NULL)
|
|
|
8662 continue; /* No Tab found, just skip the line. */
|
|
|
8663 *p++ = NUL;
|
|
|
8664 if (STRICMP(su->su_badword, line) == 0)
|
|
|
8665 {
|
|
|
8666 /* Match! Isolate the good word, until CR or NL. */
|
|
|
8667 for (len = 0; p[len] >= ' '; ++len)
|
|
|
8668 ;
|
|
|
8669 p[len] = NUL;
|
|
|
8670
|
|
|
8671 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
8672 * of the bad word. */
|
|
|
8673 if (captype(p, NULL) == 0)
|
|
|
8674 {
|
|
|
8675 make_case_word(p, cword, su->su_badflags);
|
|
|
8676 p = cword;
|
|
|
8677 }
|
|
|
8678
|
|
|
8679 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
8680 SCORE_FILE, 0, TRUE);
|
|
|
8681 }
|
|
|
8682 }
|
|
|
8683
|
|
|
8684 fclose(fd);
|
|
|
8685
|
|
|
8686 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8687 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8688 }
|
|
|
8689
|
|
|
8690 /*
|
|
|
8691 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
8692 */
|
|
|
8693 static void
|
|
|
8694 spell_suggest_intern(su)
|
|
|
8695 suginfo_T *su;
|
|
|
8696 {
|
|
344
|
8697 /*
|
|
346
|
8698 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
8699 *
|
|
|
8700 * Set a maximum score to limit the combination of operations that is
|
|
|
8701 * tried.
|
|
|
8702 */
|
|
346
|
8703 suggest_try_special(su);
|
|
|
8704
|
|
|
8705 /*
|
|
|
8706 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
8707 * from the .aff file and inserting a space (split the word).
|
|
|
8708 */
|
|
|
8709 suggest_try_change(su);
|
|
344
|
8710
|
|
|
8711 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
8712 if (sps_flags & SPS_DOUBLE)
|
|
|
8713 score_comp_sal(su);
|
|
|
8714
|
|
|
8715 /*
|
|
346
|
8716 * 3. Try finding sound-a-like words.
|
|
344
|
8717 *
|
|
|
8718 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
8719 * very slow and often doesn't find new suggestions.
|
|
|
8720 */
|
|
|
8721 if ((sps_flags & SPS_DOUBLE)
|
|
|
8722 || (!(sps_flags & SPS_FAST)
|
|
|
8723 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
8724 {
|
|
|
8725 /* Allow a higher score now. */
|
|
|
8726 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
8727 suggest_try_soundalike(su);
|
|
344
|
8728 }
|
|
|
8729
|
|
|
8730 /* When CTRL-C was hit while searching do show the results. */
|
|
|
8731 ui_breakcheck();
|
|
|
8732 if (got_int)
|
|
|
8733 {
|
|
|
8734 (void)vgetc();
|
|
|
8735 got_int = FALSE;
|
|
|
8736 }
|
|
|
8737
|
|
372
|
8738 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
8739 {
|
|
|
8740 if (sps_flags & SPS_BEST)
|
|
|
8741 /* Adjust the word score for how it sounds like. */
|
|
|
8742 rescore_suggestions(su);
|
|
|
8743
|
|
|
8744 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
8745 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
8746 }
|
|
|
8747 }
|
|
|
8748
|
|
|
8749 /*
|
|
|
8750 * Free the info put in "*su" by spell_find_suggest().
|
|
|
8751 */
|
|
|
8752 static void
|
|
|
8753 spell_find_cleanup(su)
|
|
|
8754 suginfo_T *su;
|
|
|
8755 {
|
|
|
8756 int i;
|
|
|
8757
|
|
|
8758 /* Free the suggestions. */
|
|
|
8759 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8760 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
8761 ga_clear(&su->su_ga);
|
|
|
8762 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
8763 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
8764 ga_clear(&su->su_sga);
|
|
323
|
8765
|
|
|
8766 /* Free the banned words. */
|
|
344
|
8767 free_banned(su);
|
|
323
|
8768 }
|
|
|
8769
|
|
|
8770 /*
|
|
324
|
8771 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
8772 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
8773 * The result is NUL terminated.
|
|
323
|
8774 */
|
|
|
8775 static void
|
|
324
|
8776 onecap_copy(word, wcopy, upper)
|
|
323
|
8777 char_u *word;
|
|
|
8778 char_u *wcopy;
|
|
|
8779 int upper; /* TRUE: first letter made upper case */
|
|
|
8780 {
|
|
|
8781 char_u *p;
|
|
|
8782 int c;
|
|
|
8783 int l;
|
|
|
8784
|
|
|
8785 p = word;
|
|
|
8786 #ifdef FEAT_MBYTE
|
|
|
8787 if (has_mbyte)
|
|
474
|
8788 c = mb_cptr2char_adv(&p);
|
|
323
|
8789 else
|
|
|
8790 #endif
|
|
|
8791 c = *p++;
|
|
|
8792 if (upper)
|
|
324
|
8793 c = SPELL_TOUPPER(c);
|
|
323
|
8794 else
|
|
324
|
8795 c = SPELL_TOFOLD(c);
|
|
323
|
8796 #ifdef FEAT_MBYTE
|
|
|
8797 if (has_mbyte)
|
|
|
8798 l = mb_char2bytes(c, wcopy);
|
|
|
8799 else
|
|
|
8800 #endif
|
|
|
8801 {
|
|
|
8802 l = 1;
|
|
|
8803 wcopy[0] = c;
|
|
|
8804 }
|
|
376
|
8805 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
8806 }
|
|
|
8807
|
|
|
8808 /*
|
|
324
|
8809 * Make a copy of "word" with all the letters upper cased into
|
|
|
8810 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
8811 */
|
|
|
8812 static void
|
|
|
8813 allcap_copy(word, wcopy)
|
|
|
8814 char_u *word;
|
|
|
8815 char_u *wcopy;
|
|
|
8816 {
|
|
|
8817 char_u *s;
|
|
|
8818 char_u *d;
|
|
|
8819 int c;
|
|
|
8820
|
|
|
8821 d = wcopy;
|
|
|
8822 for (s = word; *s != NUL; )
|
|
|
8823 {
|
|
|
8824 #ifdef FEAT_MBYTE
|
|
|
8825 if (has_mbyte)
|
|
474
|
8826 c = mb_cptr2char_adv(&s);
|
|
323
|
8827 else
|
|
|
8828 #endif
|
|
|
8829 c = *s++;
|
|
492
|
8830
|
|
|
8831 #ifdef FEAT_MBYTE
|
|
|
8832 /* We only change ß to SS when we are certain latin1 is used. It
|
|
|
8833 * would cause weird errors in other 8-bit encodings. */
|
|
|
8834 if (enc_latin1like && c == 0xdf)
|
|
|
8835 {
|
|
|
8836 c = 'S';
|
|
|
8837 if (d - wcopy >= MAXWLEN - 1)
|
|
|
8838 break;
|
|
|
8839 *d++ = c;
|
|
|
8840 }
|
|
|
8841 else
|
|
|
8842 #endif
|
|
|
8843 c = SPELL_TOUPPER(c);
|
|
323
|
8844
|
|
|
8845 #ifdef FEAT_MBYTE
|
|
|
8846 if (has_mbyte)
|
|
|
8847 {
|
|
|
8848 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
8849 break;
|
|
|
8850 d += mb_char2bytes(c, d);
|
|
|
8851 }
|
|
|
8852 else
|
|
|
8853 #endif
|
|
|
8854 {
|
|
|
8855 if (d - wcopy >= MAXWLEN - 1)
|
|
|
8856 break;
|
|
|
8857 *d++ = c;
|
|
|
8858 }
|
|
|
8859 }
|
|
|
8860 *d = NUL;
|
|
|
8861 }
|
|
|
8862
|
|
|
8863 /*
|
|
346
|
8864 * Try finding suggestions by recognizing specific situations.
|
|
|
8865 */
|
|
|
8866 static void
|
|
|
8867 suggest_try_special(su)
|
|
|
8868 suginfo_T *su;
|
|
|
8869 {
|
|
|
8870 char_u *p;
|
|
376
|
8871 size_t len;
|
|
346
|
8872 int c;
|
|
|
8873 char_u word[MAXWLEN];
|
|
|
8874
|
|
|
8875 /*
|
|
|
8876 * Recognize a word that is repeated: "the the".
|
|
|
8877 */
|
|
|
8878 p = skiptowhite(su->su_fbadword);
|
|
|
8879 len = p - su->su_fbadword;
|
|
|
8880 p = skipwhite(p);
|
|
|
8881 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
8882 {
|
|
|
8883 /* Include badflags: if the badword is onecap or allcap
|
|
|
8884 * use that for the goodword too: "The the" -> "The". */
|
|
|
8885 c = su->su_fbadword[len];
|
|
|
8886 su->su_fbadword[len] = NUL;
|
|
|
8887 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
8888 su->su_fbadword[len] = c;
|
|
351
|
8889 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
8890 }
|
|
|
8891 }
|
|
|
8892
|
|
|
8893 /*
|
|
323
|
8894 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
8895 *
|
|
|
8896 * This uses a state machine. At each node in the tree we try various
|
|
|
8897 * operations. When trying if an operation work "depth" is increased and the
|
|
|
8898 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
8899 * character, replace one and delete another. The number of changes is
|
|
|
8900 * limited by su->su_maxscore, checked in try_deeper().
|
|
485
|
8901 *
|
|
|
8902 * After implementing this I noticed an article by Kemal Oflazer that
|
|
|
8903 * describes something similar: "Error-tolerant Finite State Recognition with
|
|
|
8904 * Applications to Morphological Analysis and Spelling Correction" (1996).
|
|
|
8905 * The implementation in the article is simplified and requires a stack of
|
|
|
8906 * unknown depth. The implementation here only needs a stack depth of the
|
|
|
8907 * length of the word.
|
|
323
|
8908 */
|
|
|
8909 static void
|
|
346
|
8910 suggest_try_change(su)
|
|
323
|
8911 suginfo_T *su;
|
|
|
8912 {
|
|
|
8913 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
8914 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
8915 trystate_T stack[MAXWLEN];
|
|
483
|
8916 char_u preword[MAXWLEN * 3]; /* word found with proper case;
|
|
|
8917 * concatanation of prefix compound
|
|
|
8918 * words and split word. NUL terminated
|
|
|
8919 * when going deeper but not when coming
|
|
|
8920 * back. */
|
|
|
8921 char_u compflags[MAXWLEN]; /* compound flags, one for each word */
|
|
323
|
8922 trystate_T *sp;
|
|
|
8923 int newscore;
|
|
|
8924 langp_T *lp;
|
|
375
|
8925 char_u *byts, *fbyts, *pbyts;
|
|
|
8926 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
8927 int depth;
|
|
330
|
8928 int c, c2, c3;
|
|
390
|
8929 int n;
|
|
323
|
8930 int flags;
|
|
|
8931 garray_T *gap;
|
|
324
|
8932 idx_T arridx;
|
|
323
|
8933 int len;
|
|
|
8934 char_u *p;
|
|
|
8935 fromto_T *ftp;
|
|
330
|
8936 int fl = 0, tl;
|
|
346
|
8937 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
482
|
8938 slang_T *slang;
|
|
|
8939 int fword_ends;
|
|
323
|
8940
|
|
|
8941 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
8942 * to find matches (esp. REP items). Append some more text, changing
|
|
|
8943 * chars after the bad word may help. */
|
|
323
|
8944 STRCPY(fword, su->su_fbadword);
|
|
346
|
8945 n = STRLEN(fword);
|
|
|
8946 p = su->su_badptr + su->su_badlen;
|
|
|
8947 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
8948
|
|
|
8949 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8950 lp->lp_slang != NULL; ++lp)
|
|
|
8951 {
|
|
482
|
8952 slang = lp->lp_slang;
|
|
|
8953
|
|
323
|
8954 /*
|
|
|
8955 * Go through the whole case-fold tree, try changes at each node.
|
|
|
8956 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
8957 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
8958 * word).
|
|
|
8959 */
|
|
|
8960 depth = 0;
|
|
375
|
8961 sp = &stack[0];
|
|
483
|
8962 vim_memset(sp, 0, sizeof(trystate_T));
|
|
375
|
8963 sp->ts_curi = 1;
|
|
330
|
8964
|
|
|
8965 /*
|
|
375
|
8966 * When there are postponed prefixes we need to use these first. At
|
|
|
8967 * the end of the prefix we continue in the case-fold tree.
|
|
|
8968 */
|
|
482
|
8969 fbyts = slang->sl_fbyts;
|
|
|
8970 fidxs = slang->sl_fidxs;
|
|
|
8971 pbyts = slang->sl_pbyts;
|
|
|
8972 pidxs = slang->sl_pidxs;
|
|
375
|
8973 if (pbyts != NULL)
|
|
|
8974 {
|
|
|
8975 byts = pbyts;
|
|
|
8976 idxs = pidxs;
|
|
482
|
8977 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
375
|
8978 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
8979 }
|
|
|
8980 else
|
|
|
8981 {
|
|
|
8982 byts = fbyts;
|
|
|
8983 idxs = fidxs;
|
|
482
|
8984 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
485
|
8985 sp->ts_state = STATE_START;
|
|
375
|
8986 }
|
|
|
8987
|
|
|
8988 /*
|
|
330
|
8989 * Loop to find all suggestions. At each round we either:
|
|
|
8990 * - For the current state try one operation, advance "ts_curi",
|
|
|
8991 * increase "depth".
|
|
|
8992 * - When a state is done go to the next, set "ts_state".
|
|
|
8993 * - When all states are tried decrease "depth".
|
|
|
8994 */
|
|
323
|
8995 while (depth >= 0 && !got_int)
|
|
|
8996 {
|
|
|
8997 sp = &stack[depth];
|
|
|
8998 switch (sp->ts_state)
|
|
|
8999 {
|
|
|
9000 case STATE_START:
|
|
375
|
9001 case STATE_NOPREFIX:
|
|
323
|
9002 /*
|
|
|
9003 * Start of node: Deal with NUL bytes, which means
|
|
|
9004 * tword[] may end here.
|
|
|
9005 */
|
|
|
9006 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
9007 len = byts[arridx]; /* bytes in this node */
|
|
|
9008 arridx += sp->ts_curi; /* index of current byte */
|
|
|
9009
|
|
482
|
9010 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
9011 {
|
|
|
9012 /* Skip over the NUL bytes, we use them later. */
|
|
|
9013 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
9014 ;
|
|
|
9015 sp->ts_curi += n;
|
|
|
9016
|
|
390
|
9017 /* Always past NUL bytes now. */
|
|
|
9018 n = (int)sp->ts_state;
|
|
|
9019 sp->ts_state = STATE_ENDNUL;
|
|
455
|
9020 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
9021
|
|
375
|
9022 /* At end of a prefix or at start of prefixtree: check for
|
|
|
9023 * following word. */
|
|
390
|
9024 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
9025 {
|
|
455
|
9026 /* Set su->su_badflags to the caps type at this
|
|
|
9027 * position. Use the caps type until here for the
|
|
|
9028 * prefix itself. */
|
|
|
9029 #ifdef FEAT_MBYTE
|
|
|
9030 if (has_mbyte)
|
|
|
9031 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
9032 else
|
|
|
9033 #endif
|
|
|
9034 n = sp->ts_fidx;
|
|
474
|
9035 flags = badword_captype(su->su_badptr,
|
|
|
9036 su->su_badptr + n);
|
|
|
9037 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
9038 su->su_badptr + su->su_badlen);
|
|
375
|
9039 ++depth;
|
|
|
9040 stack[depth] = stack[depth - 1];
|
|
|
9041 sp = &stack[depth];
|
|
|
9042 sp->ts_prefixdepth = depth - 1;
|
|
|
9043 byts = fbyts;
|
|
|
9044 idxs = fidxs;
|
|
|
9045 sp->ts_state = STATE_START;
|
|
|
9046 sp->ts_curi = 1; /* start just after length byte */
|
|
|
9047 sp->ts_arridx = 0;
|
|
|
9048
|
|
455
|
9049 /* Move the prefix to preword[] with the right case
|
|
|
9050 * and make find_keepcap_word() works. */
|
|
485
|
9051 tword[sp->ts_twordlen] = NUL;
|
|
|
9052 make_case_word(tword + sp->ts_splitoff,
|
|
|
9053 preword + sp->ts_prewordlen,
|
|
|
9054 flags);
|
|
|
9055 sp->ts_prewordlen = STRLEN(preword);
|
|
483
|
9056 sp->ts_splitoff = sp->ts_twordlen;
|
|
375
|
9057 }
|
|
|
9058 break;
|
|
|
9059 }
|
|
|
9060
|
|
346
|
9061 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
9062 {
|
|
|
9063 /* Past bytes in node and/or past NUL bytes. */
|
|
|
9064 sp->ts_state = STATE_ENDNUL;
|
|
455
|
9065 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
9066 break;
|
|
|
9067 }
|
|
|
9068
|
|
|
9069 /*
|
|
|
9070 * End of word in tree.
|
|
|
9071 */
|
|
|
9072 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
9073
|
|
324
|
9074 flags = (int)idxs[arridx];
|
|
483
|
9075 fword_ends = (fword[sp->ts_fidx] == NUL
|
|
|
9076 || !spell_iswordp(fword + sp->ts_fidx, curbuf));
|
|
|
9077 tword[sp->ts_twordlen] = NUL;
|
|
323
|
9078
|
|
485
|
9079 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL
|
|
|
9080 && (sp->ts_flags & TSF_PREFIXOK) == 0)
|
|
375
|
9081 {
|
|
|
9082 /* There was a prefix before the word. Check that the
|
|
|
9083 * prefix can be used with this word. */
|
|
|
9084 /* Count the length of the NULs in the prefix. If there
|
|
|
9085 * are none this must be the first try without a prefix.
|
|
|
9086 */
|
|
|
9087 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
9088 len = pbyts[n++];
|
|
|
9089 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
9090 ;
|
|
|
9091 if (c > 0)
|
|
|
9092 {
|
|
390
|
9093 c = valid_word_prefix(c, n, flags,
|
|
483
|
9094 tword + sp->ts_splitoff, slang, FALSE);
|
|
375
|
9095 if (c == 0)
|
|
|
9096 break;
|
|
|
9097
|
|
|
9098 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
9099 if (c & WF_RAREPFX)
|
|
|
9100 flags |= WF_RARE;
|
|
485
|
9101
|
|
|
9102 /* Tricky: when checking for both prefix and
|
|
|
9103 * compounding we run into the prefix flag first.
|
|
|
9104 * Remember that it's OK, so that we accept the prefix
|
|
|
9105 * when arriving at a compound flag. */
|
|
|
9106 sp->ts_flags |= TSF_PREFIXOK;
|
|
375
|
9107 }
|
|
|
9108 }
|
|
|
9109
|
|
485
|
9110 if (sp->ts_complen > sp->ts_compsplit)
|
|
|
9111 {
|
|
492
|
9112 if (slang->sl_nobreak)
|
|
|
9113 {
|
|
|
9114 /* There was a word before this word. When there was
|
|
|
9115 * no change in this word (it was correct) add the
|
|
|
9116 * first word as a suggestion. If this word was
|
|
|
9117 * corrected too, we need to check if a correct word
|
|
|
9118 * follows. */
|
|
|
9119 if (sp->ts_fidx - sp->ts_splitfidx
|
|
|
9120 == sp->ts_twordlen - sp->ts_splitoff
|
|
|
9121 && STRNCMP(fword + sp->ts_splitfidx,
|
|
|
9122 tword + sp->ts_splitoff,
|
|
|
9123 sp->ts_fidx - sp->ts_splitfidx) == 0)
|
|
|
9124 {
|
|
|
9125 preword[sp->ts_prewordlen] = NUL;
|
|
|
9126 add_suggestion(su, &su->su_ga, preword,
|
|
|
9127 sp->ts_splitfidx - repextra,
|
|
|
9128 sp->ts_score, 0, FALSE);
|
|
|
9129 break;
|
|
|
9130 }
|
|
|
9131 }
|
|
|
9132 else
|
|
|
9133 {
|
|
|
9134 /* There was a compound word before this word. If
|
|
|
9135 * this word does not support compounding then give up
|
|
|
9136 * (splitting is tried for the word without compound
|
|
|
9137 * flag). */
|
|
|
9138 if (((unsigned)flags >> 24) == 0
|
|
|
9139 || sp->ts_twordlen - sp->ts_splitoff
|
|
485
|
9140 < slang->sl_compminlen)
|
|
492
|
9141 break;
|
|
|
9142 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
9143 compflags[sp->ts_complen + 1] = NUL;
|
|
|
9144 vim_strncpy(preword + sp->ts_prewordlen,
|
|
|
9145 tword + sp->ts_splitoff,
|
|
|
9146 sp->ts_twordlen - sp->ts_splitoff);
|
|
|
9147 p = preword;
|
|
|
9148 while (*skiptowhite(p) != NUL)
|
|
|
9149 p = skipwhite(skiptowhite(p));
|
|
|
9150 if (fword_ends && !can_compound(slang, p,
|
|
485
|
9151 compflags + sp->ts_compsplit))
|
|
492
|
9152 break;
|
|
|
9153
|
|
|
9154 /* Get pointer to last char of previous word. */
|
|
|
9155 p = preword + sp->ts_prewordlen;
|
|
|
9156 mb_ptr_back(preword, p);
|
|
|
9157 }
|
|
490
|
9158 }
|
|
|
9159 else
|
|
|
9160 p = NULL;
|
|
485
|
9161
|
|
323
|
9162 /*
|
|
|
9163 * Form the word with proper case in preword.
|
|
|
9164 * If there is a word from a previous split, append.
|
|
|
9165 */
|
|
|
9166 if (flags & WF_KEEPCAP)
|
|
|
9167 /* Must find the word in the keep-case tree. */
|
|
483
|
9168 find_keepcap_word(slang, tword + sp->ts_splitoff,
|
|
|
9169 preword + sp->ts_prewordlen);
|
|
323
|
9170 else
|
|
346
|
9171 {
|
|
323
|
9172 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
9173 * use that for the goodword too. But if the badword is
|
|
|
9174 * allcap and it's only one char long use onecap. */
|
|
|
9175 c = su->su_badflags;
|
|
|
9176 if ((c & WF_ALLCAP)
|
|
|
9177 #ifdef FEAT_MBYTE
|
|
474
|
9178 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
9179 #else
|
|
|
9180 && su->su_badlen == 1
|
|
|
9181 #endif
|
|
|
9182 )
|
|
|
9183 c = WF_ONECAP;
|
|
490
|
9184 c |= flags;
|
|
|
9185
|
|
|
9186 /* When appending a compound word after a word character
|
|
|
9187 * don't use Onecap. */
|
|
|
9188 if (p != NULL && spell_iswordp_nmw(p))
|
|
|
9189 c &= ~WF_ONECAP;
|
|
483
|
9190 make_case_word(tword + sp->ts_splitoff,
|
|
490
|
9191 preword + sp->ts_prewordlen, c);
|
|
346
|
9192 }
|
|
323
|
9193
|
|
|
9194 /* Don't use a banned word. It may appear again as a good
|
|
|
9195 * word, thus remember it. */
|
|
|
9196 if (flags & WF_BANNED)
|
|
|
9197 {
|
|
483
|
9198 add_banned(su, preword + sp->ts_prewordlen);
|
|
323
|
9199 break;
|
|
|
9200 }
|
|
483
|
9201 if (was_banned(su, preword + sp->ts_prewordlen)
|
|
372
|
9202 || was_banned(su, preword))
|
|
323
|
9203 break;
|
|
|
9204
|
|
|
9205 newscore = 0;
|
|
|
9206 if ((flags & WF_REGION)
|
|
390
|
9207 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
9208 newscore += SCORE_REGION;
|
|
|
9209 if (flags & WF_RARE)
|
|
|
9210 newscore += SCORE_RARE;
|
|
|
9211
|
|
346
|
9212 if (!spell_valid_case(su->su_badflags,
|
|
483
|
9213 captype(preword + sp->ts_prewordlen, NULL)))
|
|
323
|
9214 newscore += SCORE_ICASE;
|
|
|
9215
|
|
482
|
9216 if (fword_ends && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
9217 {
|
|
366
|
9218 /* The badword also ends: add suggestions. Give a penalty
|
|
|
9219 * when changing non-word char to word char, e.g., "thes,"
|
|
|
9220 * -> "these". */
|
|
|
9221 p = fword + sp->ts_fidx;
|
|
|
9222 #ifdef FEAT_MBYTE
|
|
|
9223 if (has_mbyte)
|
|
|
9224 mb_ptr_back(fword, p);
|
|
|
9225 else
|
|
|
9226 #endif
|
|
|
9227 --p;
|
|
376
|
9228 if (!spell_iswordp(p, curbuf))
|
|
366
|
9229 {
|
|
|
9230 p = preword + STRLEN(preword);
|
|
|
9231 #ifdef FEAT_MBYTE
|
|
|
9232 if (has_mbyte)
|
|
|
9233 mb_ptr_back(preword, p);
|
|
|
9234 else
|
|
|
9235 #endif
|
|
|
9236 --p;
|
|
376
|
9237 if (spell_iswordp(p, curbuf))
|
|
366
|
9238 newscore += SCORE_NONWORD;
|
|
|
9239 }
|
|
|
9240
|
|
344
|
9241 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
9242 sp->ts_fidx - repextra,
|
|
351
|
9243 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
9244 }
|
|
482
|
9245 else if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends)
|
|
330
|
9246 #ifdef FEAT_MBYTE
|
|
|
9247 /* Don't split halfway a character. */
|
|
|
9248 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
9249 #endif
|
|
|
9250 )
|
|
323
|
9251 {
|
|
482
|
9252 int try_compound;
|
|
|
9253
|
|
|
9254 /* Get here in two situations:
|
|
|
9255 * 1. The word in the tree ends but the badword continues:
|
|
|
9256 * If the word allows compounding try that. Otherwise
|
|
|
9257 * try a split by inserting a space. For both check
|
|
|
9258 * that a valid words starts at fword[sp->ts_fidx].
|
|
492
|
9259 * For NOBREAK do like compounding to be able to check
|
|
|
9260 * if the next word is valid.
|
|
482
|
9261 * 2. The badword does end, but it was due to a change
|
|
|
9262 * (e.g., a swap). No need to split, but do check that
|
|
|
9263 * the following word is valid.
|
|
|
9264 */
|
|
485
|
9265 try_compound = FALSE;
|
|
482
|
9266 if (!fword_ends
|
|
495
|
9267 && slang->sl_compprog != NULL
|
|
483
|
9268 && ((unsigned)flags >> 24) != 0
|
|
|
9269 && sp->ts_twordlen - sp->ts_splitoff
|
|
482
|
9270 >= slang->sl_compminlen
|
|
490
|
9271 && (slang->sl_compsylmax < MAXWLEN
|
|
|
9272 || sp->ts_complen + 1 - sp->ts_compsplit
|
|
|
9273 < slang->sl_compmax)
|
|
495
|
9274 && (byte_in_str(sp->ts_complen == sp->ts_compsplit
|
|
485
|
9275 ? slang->sl_compstartflags
|
|
|
9276 : slang->sl_compallflags,
|
|
495
|
9277 ((unsigned)flags >> 24))))
|
|
483
|
9278 {
|
|
482
|
9279 try_compound = TRUE;
|
|
483
|
9280 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
9281 compflags[sp->ts_complen + 1] = NUL;
|
|
|
9282 }
|
|
485
|
9283
|
|
492
|
9284 /* For NOBREAK we never try splitting, it won't make any
|
|
|
9285 * word valid. */
|
|
|
9286 if (slang->sl_nobreak)
|
|
|
9287 try_compound = TRUE;
|
|
|
9288
|
|
485
|
9289 /* If we could add a compound word, and it's also possible
|
|
|
9290 * to split at this point, do the split first and set
|
|
|
9291 * TSF_DIDSPLIT to avoid doing it again. */
|
|
492
|
9292 else if (!fword_ends
|
|
485
|
9293 && try_compound
|
|
|
9294 && (sp->ts_flags & TSF_DIDSPLIT) == 0)
|
|
482
|
9295 {
|
|
|
9296 try_compound = FALSE;
|
|
485
|
9297 sp->ts_flags |= TSF_DIDSPLIT;
|
|
|
9298 --sp->ts_curi; /* do the same NUL again */
|
|
|
9299 compflags[sp->ts_complen] = NUL;
|
|
|
9300 }
|
|
|
9301 else
|
|
|
9302 sp->ts_flags &= ~TSF_DIDSPLIT;
|
|
|
9303
|
|
|
9304 if (!try_compound && !fword_ends)
|
|
|
9305 {
|
|
|
9306 /* If we're going to split need to check that the
|
|
|
9307 * words so far are valid for compounding. */
|
|
490
|
9308 p = preword;
|
|
|
9309 while (*skiptowhite(p) != NUL)
|
|
|
9310 p = skipwhite(skiptowhite(p));
|
|
485
|
9311 if (sp->ts_complen > sp->ts_compsplit
|
|
490
|
9312 && !can_compound(slang, p,
|
|
485
|
9313 compflags + sp->ts_compsplit))
|
|
|
9314 break;
|
|
|
9315 newscore += SCORE_SPLIT;
|
|
482
|
9316 }
|
|
|
9317
|
|
|
9318 if (try_deeper(su, stack, depth, newscore))
|
|
323
|
9319 {
|
|
|
9320 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
346
|
9321 sp->ts_save_badflags = su->su_badflags;
|
|
376
|
9322 sp->ts_state = STATE_SPLITUNDO;
|
|
|
9323
|
|
|
9324 ++depth;
|
|
|
9325 sp = &stack[depth];
|
|
323
|
9326
|
|
482
|
9327 /* Append a space to preword when splitting. */
|
|
|
9328 if (!try_compound && !fword_ends)
|
|
|
9329 STRCAT(preword, " ");
|
|
483
|
9330 sp->ts_prewordlen = STRLEN(preword);
|
|
|
9331 sp->ts_splitoff = sp->ts_twordlen;
|
|
492
|
9332 sp->ts_splitfidx = sp->ts_fidx;
|
|
376
|
9333
|
|
|
9334 /* If the badword has a non-word character at this
|
|
|
9335 * position skip it. That means replacing the
|
|
482
|
9336 * non-word character with a space. Always skip a
|
|
|
9337 * character when the word ends. */
|
|
|
9338 if ((!try_compound
|
|
|
9339 && !spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
9340 || fword_ends)
|
|
376
|
9341 {
|
|
482
|
9342 int l;
|
|
|
9343
|
|
376
|
9344 #ifdef FEAT_MBYTE
|
|
|
9345 if (has_mbyte)
|
|
482
|
9346 l = MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
376
|
9347 else
|
|
|
9348 #endif
|
|
482
|
9349 l = 1;
|
|
|
9350 if (fword_ends)
|
|
|
9351 {
|
|
|
9352 /* Copy the skipped character to preword. */
|
|
483
|
9353 mch_memmove(preword + sp->ts_prewordlen,
|
|
482
|
9354 fword + sp->ts_fidx, l);
|
|
483
|
9355 sp->ts_prewordlen += l;
|
|
|
9356 preword[sp->ts_prewordlen] = NUL;
|
|
482
|
9357 }
|
|
|
9358 else
|
|
|
9359 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
9360 sp->ts_fidx += l;
|
|
376
|
9361 }
|
|
455
|
9362
|
|
485
|
9363 /* When compounding include compound flag in
|
|
|
9364 * compflags[] (already set above). When splitting we
|
|
|
9365 * may start compounding over again. */
|
|
482
|
9366 if (try_compound)
|
|
483
|
9367 ++sp->ts_complen;
|
|
482
|
9368 else
|
|
485
|
9369 sp->ts_compsplit = sp->ts_complen;
|
|
|
9370 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
482
|
9371
|
|
455
|
9372 /* set su->su_badflags to the caps type at this
|
|
|
9373 * position */
|
|
324
|
9374 #ifdef FEAT_MBYTE
|
|
|
9375 if (has_mbyte)
|
|
455
|
9376 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
9377 else
|
|
|
9378 #endif
|
|
455
|
9379 n = sp->ts_fidx;
|
|
474
|
9380 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
9381 su->su_badptr + su->su_badlen);
|
|
323
|
9382
|
|
|
9383 /* Restart at top of the tree. */
|
|
376
|
9384 sp->ts_arridx = 0;
|
|
485
|
9385
|
|
|
9386 /* If there are postponed prefixes, try these too. */
|
|
|
9387 if (pbyts != NULL)
|
|
|
9388 {
|
|
|
9389 byts = pbyts;
|
|
|
9390 idxs = pidxs;
|
|
|
9391 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
|
9392 sp->ts_state = STATE_NOPREFIX;
|
|
|
9393 }
|
|
323
|
9394 }
|
|
|
9395 }
|
|
|
9396 break;
|
|
|
9397
|
|
|
9398 case STATE_SPLITUNDO:
|
|
482
|
9399 /* Undo the changes done for word split or compound word. */
|
|
346
|
9400 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
9401
|
|
|
9402 /* Continue looking for NUL bytes. */
|
|
|
9403 sp->ts_state = STATE_START;
|
|
485
|
9404
|
|
|
9405 /* In case we went into the prefix tree. */
|
|
|
9406 byts = fbyts;
|
|
|
9407 idxs = fidxs;
|
|
323
|
9408 break;
|
|
|
9409
|
|
|
9410 case STATE_ENDNUL:
|
|
|
9411 /* Past the NUL bytes in the node. */
|
|
455
|
9412 su->su_badflags = sp->ts_save_badflags;
|
|
346
|
9413 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
9414 {
|
|
|
9415 /* The badword ends, can't use the bytes in this node. */
|
|
|
9416 sp->ts_state = STATE_DEL;
|
|
|
9417 break;
|
|
|
9418 }
|
|
|
9419 sp->ts_state = STATE_PLAIN;
|
|
|
9420 /*FALLTHROUGH*/
|
|
|
9421
|
|
|
9422 case STATE_PLAIN:
|
|
|
9423 /*
|
|
|
9424 * Go over all possible bytes at this node, add each to
|
|
|
9425 * tword[] and use child node. "ts_curi" is the index.
|
|
|
9426 */
|
|
|
9427 arridx = sp->ts_arridx;
|
|
|
9428 if (sp->ts_curi > byts[arridx])
|
|
|
9429 {
|
|
|
9430 /* Done all bytes at this node, do next state. When still
|
|
|
9431 * at already changed bytes skip the other tricks. */
|
|
|
9432 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
9433 sp->ts_state = STATE_DEL;
|
|
|
9434 else
|
|
|
9435 sp->ts_state = STATE_FINAL;
|
|
|
9436 }
|
|
|
9437 else
|
|
|
9438 {
|
|
|
9439 arridx += sp->ts_curi++;
|
|
|
9440 c = byts[arridx];
|
|
|
9441
|
|
|
9442 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
9443 * the byte in the bad word adjust the score. But don't
|
|
|
9444 * even try when the byte was already changed. */
|
|
330
|
9445 if (c == fword[sp->ts_fidx]
|
|
|
9446 #ifdef FEAT_MBYTE
|
|
|
9447 || (sp->ts_tcharlen > 0
|
|
|
9448 && sp->ts_isdiff != DIFF_NONE)
|
|
|
9449 #endif
|
|
|
9450 )
|
|
323
|
9451 newscore = 0;
|
|
|
9452 else
|
|
|
9453 newscore = SCORE_SUBST;
|
|
|
9454 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
9455 && try_deeper(su, stack, depth, newscore))
|
|
|
9456 {
|
|
|
9457 ++depth;
|
|
330
|
9458 sp = &stack[depth];
|
|
|
9459 ++sp->ts_fidx;
|
|
|
9460 tword[sp->ts_twordlen++] = c;
|
|
|
9461 sp->ts_arridx = idxs[arridx];
|
|
|
9462 #ifdef FEAT_MBYTE
|
|
|
9463 if (newscore == SCORE_SUBST)
|
|
|
9464 sp->ts_isdiff = DIFF_YES;
|
|
|
9465 if (has_mbyte)
|
|
|
9466 {
|
|
|
9467 /* Multi-byte characters are a bit complicated to
|
|
|
9468 * handle: They differ when any of the bytes
|
|
|
9469 * differ and then their length may also differ. */
|
|
|
9470 if (sp->ts_tcharlen == 0)
|
|
|
9471 {
|
|
|
9472 /* First byte. */
|
|
|
9473 sp->ts_tcharidx = 0;
|
|
|
9474 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
9475 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
9476 sp->ts_isdiff = (newscore != 0)
|
|
|
9477 ? DIFF_YES : DIFF_NONE;
|
|
|
9478 }
|
|
|
9479 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
9480 /* When inserting trail bytes don't advance in
|
|
|
9481 * the bad word. */
|
|
|
9482 --sp->ts_fidx;
|
|
|
9483 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
9484 {
|
|
|
9485 /* Last byte of character. */
|
|
|
9486 if (sp->ts_isdiff == DIFF_YES)
|
|
|
9487 {
|
|
|
9488 /* Correct ts_fidx for the byte length of
|
|
|
9489 * the character (we didn't check that
|
|
|
9490 * before). */
|
|
|
9491 sp->ts_fidx = sp->ts_fcharstart
|
|
|
9492 + MB_BYTE2LEN(
|
|
|
9493 fword[sp->ts_fcharstart]);
|
|
|
9494
|
|
479
|
9495 /* For changing a composing character
|
|
|
9496 * adjust the score from SCORE_SUBST to
|
|
|
9497 * SCORE_SUBCOMP. */
|
|
|
9498 if (enc_utf8
|
|
|
9499 && utf_iscomposing(
|
|
|
9500 mb_ptr2char(tword
|
|
|
9501 + sp->ts_twordlen
|
|
|
9502 - sp->ts_tcharlen))
|
|
|
9503 && utf_iscomposing(
|
|
|
9504 mb_ptr2char(fword
|
|
|
9505 + sp->ts_fcharstart)))
|
|
|
9506 sp->ts_score -=
|
|
|
9507 SCORE_SUBST - SCORE_SUBCOMP;
|
|
|
9508
|
|
330
|
9509 /* For a similar character adjust score
|
|
|
9510 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
482
|
9511 else if (slang->sl_has_map
|
|
|
9512 && similar_chars(slang,
|
|
330
|
9513 mb_ptr2char(tword
|
|
|
9514 + sp->ts_twordlen
|
|
|
9515 - sp->ts_tcharlen),
|
|
|
9516 mb_ptr2char(fword
|
|
|
9517 + sp->ts_fcharstart)))
|
|
|
9518 sp->ts_score -=
|
|
|
9519 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9520 }
|
|
358
|
9521 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
9522 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
9523 {
|
|
|
9524 p = tword + sp->ts_twordlen
|
|
|
9525 - sp->ts_tcharlen;
|
|
|
9526 c = mb_ptr2char(p);
|
|
475
|
9527 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9528 {
|
|
|
9529 /* Inserting a composing char doesn't
|
|
|
9530 * count that much. */
|
|
358
|
9531 sp->ts_score -= SCORE_INS
|
|
475
|
9532 - SCORE_INSCOMP;
|
|
|
9533 }
|
|
|
9534 else
|
|
|
9535 {
|
|
|
9536 /* If the previous character was the
|
|
|
9537 * same, thus doubling a character,
|
|
|
9538 * give a bonus to the score. */
|
|
|
9539 mb_ptr_back(tword, p);
|
|
|
9540 if (c == mb_ptr2char(p))
|
|
|
9541 sp->ts_score -= SCORE_INS
|
|
358
|
9542 - SCORE_INSDUP;
|
|
475
|
9543 }
|
|
358
|
9544 }
|
|
330
|
9545
|
|
|
9546 /* Starting a new char, reset the length. */
|
|
|
9547 sp->ts_tcharlen = 0;
|
|
|
9548 }
|
|
|
9549 }
|
|
|
9550 else
|
|
|
9551 #endif
|
|
|
9552 {
|
|
|
9553 /* If we found a similar char adjust the score.
|
|
|
9554 * We do this after calling try_deeper() because
|
|
|
9555 * it's slow. */
|
|
|
9556 if (newscore != 0
|
|
482
|
9557 && slang->sl_has_map
|
|
|
9558 && similar_chars(slang,
|
|
330
|
9559 c, fword[sp->ts_fidx - 1]))
|
|
|
9560 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9561 }
|
|
323
|
9562 }
|
|
|
9563 }
|
|
|
9564 break;
|
|
|
9565
|
|
|
9566 case STATE_DEL:
|
|
330
|
9567 #ifdef FEAT_MBYTE
|
|
|
9568 /* When past the first byte of a multi-byte char don't try
|
|
|
9569 * delete/insert/swap a character. */
|
|
|
9570 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
9571 {
|
|
|
9572 sp->ts_state = STATE_FINAL;
|
|
|
9573 break;
|
|
|
9574 }
|
|
|
9575 #endif
|
|
|
9576 /*
|
|
|
9577 * Try skipping one character in the bad word (delete it).
|
|
|
9578 */
|
|
323
|
9579 sp->ts_state = STATE_INS;
|
|
|
9580 sp->ts_curi = 1;
|
|
|
9581 if (fword[sp->ts_fidx] != NUL
|
|
|
9582 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
9583 {
|
|
|
9584 ++depth;
|
|
358
|
9585
|
|
|
9586 /* Advance over the character in fword[]. Give a bonus to
|
|
|
9587 * the score if the same character is following "nn" ->
|
|
|
9588 * "n". */
|
|
330
|
9589 #ifdef FEAT_MBYTE
|
|
|
9590 if (has_mbyte)
|
|
358
|
9591 {
|
|
|
9592 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
9593 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
479
|
9594 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9595 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
|
|
|
9596 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
358
|
9597 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
9598 }
|
|
330
|
9599 else
|
|
|
9600 #endif
|
|
358
|
9601 {
|
|
330
|
9602 ++stack[depth].ts_fidx;
|
|
358
|
9603 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
9604 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
9605 }
|
|
323
|
9606 break;
|
|
|
9607 }
|
|
|
9608 /*FALLTHROUGH*/
|
|
|
9609
|
|
|
9610 case STATE_INS:
|
|
330
|
9611 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
9612 * node. */
|
|
|
9613 n = sp->ts_arridx;
|
|
|
9614 if (sp->ts_curi > byts[n])
|
|
|
9615 {
|
|
|
9616 /* Done all bytes at this node, do next state. */
|
|
|
9617 sp->ts_state = STATE_SWAP;
|
|
|
9618 }
|
|
|
9619 else
|
|
|
9620 {
|
|
330
|
9621 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
9622 n += sp->ts_curi++;
|
|
|
9623 c = byts[n];
|
|
|
9624 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
9625 {
|
|
|
9626 ++depth;
|
|
330
|
9627 sp = &stack[depth];
|
|
|
9628 tword[sp->ts_twordlen++] = c;
|
|
|
9629 sp->ts_arridx = idxs[n];
|
|
|
9630 #ifdef FEAT_MBYTE
|
|
|
9631 if (has_mbyte)
|
|
|
9632 {
|
|
|
9633 fl = MB_BYTE2LEN(c);
|
|
|
9634 if (fl > 1)
|
|
|
9635 {
|
|
|
9636 /* There are following bytes for the same
|
|
|
9637 * character. We must find all bytes before
|
|
|
9638 * trying delete/insert/swap/etc. */
|
|
|
9639 sp->ts_tcharlen = fl;
|
|
|
9640 sp->ts_tcharidx = 1;
|
|
|
9641 sp->ts_isdiff = DIFF_INSERT;
|
|
|
9642 }
|
|
|
9643 }
|
|
358
|
9644 else
|
|
|
9645 fl = 1;
|
|
|
9646 if (fl == 1)
|
|
330
|
9647 #endif
|
|
358
|
9648 {
|
|
|
9649 /* If the previous character was the same, thus
|
|
|
9650 * doubling a character, give a bonus to the
|
|
|
9651 * score. */
|
|
|
9652 if (sp->ts_twordlen >= 2
|
|
|
9653 && tword[sp->ts_twordlen - 2] == c)
|
|
|
9654 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
9655 }
|
|
323
|
9656 }
|
|
|
9657 }
|
|
|
9658 break;
|
|
|
9659
|
|
|
9660 case STATE_SWAP:
|
|
330
|
9661 /*
|
|
|
9662 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
9663 * We change "fword" here, it's changed back afterwards.
|
|
|
9664 */
|
|
|
9665 p = fword + sp->ts_fidx;
|
|
|
9666 c = *p;
|
|
|
9667 if (c == NUL)
|
|
|
9668 {
|
|
|
9669 /* End of word, can't swap or replace. */
|
|
|
9670 sp->ts_state = STATE_FINAL;
|
|
|
9671 break;
|
|
|
9672 }
|
|
|
9673 #ifdef FEAT_MBYTE
|
|
|
9674 if (has_mbyte)
|
|
|
9675 {
|
|
474
|
9676 n = mb_cptr2len(p);
|
|
330
|
9677 c = mb_ptr2char(p);
|
|
|
9678 c2 = mb_ptr2char(p + n);
|
|
|
9679 }
|
|
|
9680 else
|
|
|
9681 #endif
|
|
|
9682 c2 = p[1];
|
|
|
9683 if (c == c2)
|
|
323
|
9684 {
|
|
330
|
9685 /* Characters are identical, swap won't do anything. */
|
|
|
9686 sp->ts_state = STATE_SWAP3;
|
|
|
9687 break;
|
|
|
9688 }
|
|
|
9689 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
9690 {
|
|
|
9691 sp->ts_state = STATE_UNSWAP;
|
|
323
|
9692 ++depth;
|
|
330
|
9693 #ifdef FEAT_MBYTE
|
|
|
9694 if (has_mbyte)
|
|
|
9695 {
|
|
|
9696 fl = mb_char2len(c2);
|
|
|
9697 mch_memmove(p, p + n, fl);
|
|
|
9698 mb_char2bytes(c, p + fl);
|
|
|
9699 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
9700 }
|
|
|
9701 else
|
|
|
9702 #endif
|
|
|
9703 {
|
|
|
9704 p[0] = c2;
|
|
|
9705 p[1] = c;
|
|
|
9706 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
9707 }
|
|
323
|
9708 }
|
|
|
9709 else
|
|
|
9710 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
9711 sp->ts_state = STATE_REP_INI;
|
|
|
9712 break;
|
|
|
9713
|
|
330
|
9714 case STATE_UNSWAP:
|
|
|
9715 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
9716 p = fword + sp->ts_fidx;
|
|
|
9717 #ifdef FEAT_MBYTE
|
|
|
9718 if (has_mbyte)
|
|
|
9719 {
|
|
|
9720 n = MB_BYTE2LEN(*p);
|
|
|
9721 c = mb_ptr2char(p + n);
|
|
|
9722 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
9723 mb_char2bytes(c, p);
|
|
|
9724 }
|
|
|
9725 else
|
|
|
9726 #endif
|
|
|
9727 {
|
|
|
9728 c = *p;
|
|
|
9729 *p = p[1];
|
|
|
9730 p[1] = c;
|
|
|
9731 }
|
|
|
9732 /*FALLTHROUGH*/
|
|
|
9733
|
|
|
9734 case STATE_SWAP3:
|
|
323
|
9735 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
9736 * "fword" here, it's changed back afterwards. */
|
|
|
9737 p = fword + sp->ts_fidx;
|
|
|
9738 #ifdef FEAT_MBYTE
|
|
|
9739 if (has_mbyte)
|
|
|
9740 {
|
|
474
|
9741 n = mb_cptr2len(p);
|
|
330
|
9742 c = mb_ptr2char(p);
|
|
474
|
9743 fl = mb_cptr2len(p + n);
|
|
330
|
9744 c2 = mb_ptr2char(p + n);
|
|
|
9745 c3 = mb_ptr2char(p + n + fl);
|
|
|
9746 }
|
|
|
9747 else
|
|
|
9748 #endif
|
|
323
|
9749 {
|
|
330
|
9750 c = *p;
|
|
|
9751 c2 = p[1];
|
|
|
9752 c3 = p[2];
|
|
|
9753 }
|
|
|
9754
|
|
|
9755 /* When characters are identical: "121" then SWAP3 result is
|
|
|
9756 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
9757 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
9758 * swapping. Also skip when c3 is NUL. */
|
|
|
9759 if (c == c3 || c3 == NUL)
|
|
|
9760 {
|
|
|
9761 sp->ts_state = STATE_REP_INI;
|
|
|
9762 break;
|
|
|
9763 }
|
|
|
9764 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9765 {
|
|
|
9766 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
9767 ++depth;
|
|
330
|
9768 #ifdef FEAT_MBYTE
|
|
|
9769 if (has_mbyte)
|
|
|
9770 {
|
|
|
9771 tl = mb_char2len(c3);
|
|
|
9772 mch_memmove(p, p + n + fl, tl);
|
|
|
9773 mb_char2bytes(c2, p + tl);
|
|
|
9774 mb_char2bytes(c, p + fl + tl);
|
|
|
9775 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
9776 }
|
|
|
9777 else
|
|
|
9778 #endif
|
|
|
9779 {
|
|
|
9780 p[0] = p[2];
|
|
|
9781 p[2] = c;
|
|
|
9782 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9783 }
|
|
323
|
9784 }
|
|
|
9785 else
|
|
|
9786 sp->ts_state = STATE_REP_INI;
|
|
|
9787 break;
|
|
|
9788
|
|
330
|
9789 case STATE_UNSWAP3:
|
|
|
9790 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
9791 p = fword + sp->ts_fidx;
|
|
|
9792 #ifdef FEAT_MBYTE
|
|
|
9793 if (has_mbyte)
|
|
|
9794 {
|
|
|
9795 n = MB_BYTE2LEN(*p);
|
|
|
9796 c2 = mb_ptr2char(p + n);
|
|
|
9797 fl = MB_BYTE2LEN(p[n]);
|
|
|
9798 c = mb_ptr2char(p + n + fl);
|
|
|
9799 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
9800 mch_memmove(p + fl + tl, p, n);
|
|
|
9801 mb_char2bytes(c, p);
|
|
|
9802 mb_char2bytes(c2, p + tl);
|
|
|
9803 }
|
|
|
9804 else
|
|
|
9805 #endif
|
|
|
9806 {
|
|
|
9807 c = *p;
|
|
|
9808 *p = p[2];
|
|
|
9809 p[2] = c;
|
|
|
9810 }
|
|
346
|
9811
|
|
330
|
9812 /* Rotate three characters left: "123" -> "231". We change
|
|
|
9813 * "fword" here, it's changed back afterwards. */
|
|
323
|
9814 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9815 {
|
|
330
|
9816 sp->ts_state = STATE_UNROT3L;
|
|
323
|
9817 ++depth;
|
|
330
|
9818 p = fword + sp->ts_fidx;
|
|
|
9819 #ifdef FEAT_MBYTE
|
|
|
9820 if (has_mbyte)
|
|
|
9821 {
|
|
474
|
9822 n = mb_cptr2len(p);
|
|
330
|
9823 c = mb_ptr2char(p);
|
|
474
|
9824 fl = mb_cptr2len(p + n);
|
|
|
9825 fl += mb_cptr2len(p + n + fl);
|
|
330
|
9826 mch_memmove(p, p + n, fl);
|
|
|
9827 mb_char2bytes(c, p + fl);
|
|
|
9828 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
9829 }
|
|
|
9830 else
|
|
|
9831 #endif
|
|
|
9832 {
|
|
|
9833 c = *p;
|
|
|
9834 *p = p[1];
|
|
|
9835 p[1] = p[2];
|
|
|
9836 p[2] = c;
|
|
|
9837 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9838 }
|
|
323
|
9839 }
|
|
|
9840 else
|
|
|
9841 sp->ts_state = STATE_REP_INI;
|
|
|
9842 break;
|
|
|
9843
|
|
330
|
9844 case STATE_UNROT3L:
|
|
346
|
9845 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
9846 p = fword + sp->ts_fidx;
|
|
|
9847 #ifdef FEAT_MBYTE
|
|
|
9848 if (has_mbyte)
|
|
|
9849 {
|
|
|
9850 n = MB_BYTE2LEN(*p);
|
|
|
9851 n += MB_BYTE2LEN(p[n]);
|
|
|
9852 c = mb_ptr2char(p + n);
|
|
|
9853 tl = MB_BYTE2LEN(p[n]);
|
|
|
9854 mch_memmove(p + tl, p, n);
|
|
|
9855 mb_char2bytes(c, p);
|
|
|
9856 }
|
|
|
9857 else
|
|
|
9858 #endif
|
|
|
9859 {
|
|
|
9860 c = p[2];
|
|
|
9861 p[2] = p[1];
|
|
|
9862 p[1] = *p;
|
|
|
9863 *p = c;
|
|
|
9864 }
|
|
346
|
9865
|
|
323
|
9866 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
9867 * "fword" here, it's changed back afterwards. */
|
|
323
|
9868 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9869 {
|
|
330
|
9870 sp->ts_state = STATE_UNROT3R;
|
|
323
|
9871 ++depth;
|
|
330
|
9872 p = fword + sp->ts_fidx;
|
|
|
9873 #ifdef FEAT_MBYTE
|
|
|
9874 if (has_mbyte)
|
|
|
9875 {
|
|
474
|
9876 n = mb_cptr2len(p);
|
|
|
9877 n += mb_cptr2len(p + n);
|
|
330
|
9878 c = mb_ptr2char(p + n);
|
|
474
|
9879 tl = mb_cptr2len(p + n);
|
|
330
|
9880 mch_memmove(p + tl, p, n);
|
|
|
9881 mb_char2bytes(c, p);
|
|
|
9882 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
9883 }
|
|
|
9884 else
|
|
|
9885 #endif
|
|
|
9886 {
|
|
|
9887 c = p[2];
|
|
|
9888 p[2] = p[1];
|
|
|
9889 p[1] = *p;
|
|
|
9890 *p = c;
|
|
|
9891 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9892 }
|
|
323
|
9893 }
|
|
|
9894 else
|
|
|
9895 sp->ts_state = STATE_REP_INI;
|
|
|
9896 break;
|
|
|
9897
|
|
330
|
9898 case STATE_UNROT3R:
|
|
346
|
9899 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
9900 p = fword + sp->ts_fidx;
|
|
|
9901 #ifdef FEAT_MBYTE
|
|
|
9902 if (has_mbyte)
|
|
|
9903 {
|
|
|
9904 c = mb_ptr2char(p);
|
|
|
9905 tl = MB_BYTE2LEN(*p);
|
|
|
9906 n = MB_BYTE2LEN(p[tl]);
|
|
|
9907 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
9908 mch_memmove(p, p + tl, n);
|
|
|
9909 mb_char2bytes(c, p + n);
|
|
|
9910 }
|
|
|
9911 else
|
|
|
9912 #endif
|
|
|
9913 {
|
|
|
9914 c = *p;
|
|
|
9915 *p = p[1];
|
|
|
9916 p[1] = p[2];
|
|
|
9917 p[2] = c;
|
|
|
9918 }
|
|
323
|
9919 /*FALLTHROUGH*/
|
|
|
9920
|
|
|
9921 case STATE_REP_INI:
|
|
|
9922 /* Check if matching with REP items from the .aff file would
|
|
|
9923 * work. Quickly skip if there are no REP items or the score
|
|
|
9924 * is going to be too high anyway. */
|
|
482
|
9925 gap = &slang->sl_rep;
|
|
323
|
9926 if (gap->ga_len == 0
|
|
|
9927 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
9928 {
|
|
|
9929 sp->ts_state = STATE_FINAL;
|
|
|
9930 break;
|
|
|
9931 }
|
|
|
9932
|
|
|
9933 /* Use the first byte to quickly find the first entry that
|
|
330
|
9934 * may match. If the index is -1 there is none. */
|
|
482
|
9935 sp->ts_curi = slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
323
|
9936 if (sp->ts_curi < 0)
|
|
|
9937 {
|
|
|
9938 sp->ts_state = STATE_FINAL;
|
|
|
9939 break;
|
|
|
9940 }
|
|
|
9941
|
|
|
9942 sp->ts_state = STATE_REP;
|
|
|
9943 /*FALLTHROUGH*/
|
|
|
9944
|
|
|
9945 case STATE_REP:
|
|
|
9946 /* Try matching with REP items from the .aff file. For each
|
|
330
|
9947 * match replace the characters and check if the resulting
|
|
|
9948 * word is valid. */
|
|
323
|
9949 p = fword + sp->ts_fidx;
|
|
|
9950
|
|
482
|
9951 gap = &slang->sl_rep;
|
|
323
|
9952 while (sp->ts_curi < gap->ga_len)
|
|
|
9953 {
|
|
|
9954 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
9955 if (*ftp->ft_from != *p)
|
|
|
9956 {
|
|
|
9957 /* past possible matching entries */
|
|
|
9958 sp->ts_curi = gap->ga_len;
|
|
|
9959 break;
|
|
|
9960 }
|
|
|
9961 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
9962 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
9963 {
|
|
|
9964 /* Need to undo this afterwards. */
|
|
|
9965 sp->ts_state = STATE_REP_UNDO;
|
|
|
9966
|
|
|
9967 /* Change the "from" to the "to" string. */
|
|
|
9968 ++depth;
|
|
|
9969 fl = STRLEN(ftp->ft_from);
|
|
|
9970 tl = STRLEN(ftp->ft_to);
|
|
|
9971 if (fl != tl)
|
|
346
|
9972 {
|
|
323
|
9973 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
9974 repextra += tl - fl;
|
|
|
9975 }
|
|
323
|
9976 mch_memmove(p, ftp->ft_to, tl);
|
|
|
9977 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
9978 #ifdef FEAT_MBYTE
|
|
|
9979 stack[depth].ts_tcharlen = 0;
|
|
|
9980 #endif
|
|
323
|
9981 break;
|
|
|
9982 }
|
|
|
9983 }
|
|
|
9984
|
|
|
9985 if (sp->ts_curi >= gap->ga_len)
|
|
|
9986 /* No (more) matches. */
|
|
|
9987 sp->ts_state = STATE_FINAL;
|
|
|
9988
|
|
|
9989 break;
|
|
|
9990
|
|
|
9991 case STATE_REP_UNDO:
|
|
|
9992 /* Undo a REP replacement and continue with the next one. */
|
|
482
|
9993 ftp = (fromto_T *)slang->sl_rep.ga_data + sp->ts_curi - 1;
|
|
323
|
9994 fl = STRLEN(ftp->ft_from);
|
|
|
9995 tl = STRLEN(ftp->ft_to);
|
|
|
9996 p = fword + sp->ts_fidx;
|
|
|
9997 if (fl != tl)
|
|
346
|
9998 {
|
|
323
|
9999 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
10000 repextra -= tl - fl;
|
|
|
10001 }
|
|
323
|
10002 mch_memmove(p, ftp->ft_from, fl);
|
|
|
10003 sp->ts_state = STATE_REP;
|
|
|
10004 break;
|
|
|
10005
|
|
|
10006 default:
|
|
|
10007 /* Did all possible states at this level, go up one level. */
|
|
|
10008 --depth;
|
|
344
|
10009
|
|
482
|
10010 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
10011 {
|
|
|
10012 /* Continue in or go back to the prefix tree. */
|
|
|
10013 byts = pbyts;
|
|
|
10014 idxs = pidxs;
|
|
|
10015 }
|
|
|
10016
|
|
344
|
10017 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
10018 line_breakcheck();
|
|
323
|
10019 }
|
|
|
10020 }
|
|
|
10021 }
|
|
|
10022 }
|
|
|
10023
|
|
|
10024 /*
|
|
|
10025 * Try going one level deeper in the tree.
|
|
|
10026 */
|
|
|
10027 static int
|
|
|
10028 try_deeper(su, stack, depth, score_add)
|
|
|
10029 suginfo_T *su;
|
|
|
10030 trystate_T *stack;
|
|
|
10031 int depth;
|
|
|
10032 int score_add;
|
|
|
10033 {
|
|
|
10034 int newscore;
|
|
|
10035
|
|
|
10036 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
10037 newscore = stack[depth].ts_score + score_add;
|
|
|
10038 if (newscore >= su->su_maxscore)
|
|
|
10039 return FALSE;
|
|
|
10040
|
|
330
|
10041 stack[depth + 1] = stack[depth];
|
|
323
|
10042 stack[depth + 1].ts_state = STATE_START;
|
|
|
10043 stack[depth + 1].ts_score = newscore;
|
|
|
10044 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
485
|
10045 stack[depth + 1].ts_flags = 0;
|
|
323
|
10046 return TRUE;
|
|
|
10047 }
|
|
|
10048
|
|
455
|
10049 #ifdef FEAT_MBYTE
|
|
|
10050 /*
|
|
|
10051 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
10052 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
10053 */
|
|
|
10054 static int
|
|
|
10055 nofold_len(fword, flen, word)
|
|
|
10056 char_u *fword;
|
|
|
10057 int flen;
|
|
|
10058 char_u *word;
|
|
|
10059 {
|
|
|
10060 char_u *p;
|
|
|
10061 int i = 0;
|
|
|
10062
|
|
|
10063 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
10064 ++i;
|
|
|
10065 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
10066 --i;
|
|
|
10067 return (int)(p - word);
|
|
|
10068 }
|
|
|
10069 #endif
|
|
|
10070
|
|
323
|
10071 /*
|
|
|
10072 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
10073 * words and put it in "kword".
|
|
|
10074 * Theoretically there could be several keep-case words that result in the
|
|
|
10075 * same case-folded word, but we only find one...
|
|
|
10076 */
|
|
|
10077 static void
|
|
|
10078 find_keepcap_word(slang, fword, kword)
|
|
|
10079 slang_T *slang;
|
|
|
10080 char_u *fword;
|
|
|
10081 char_u *kword;
|
|
|
10082 {
|
|
|
10083 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
10084 int depth;
|
|
324
|
10085 idx_T tryidx;
|
|
323
|
10086
|
|
|
10087 /* The following arrays are used at each depth in the tree. */
|
|
324
|
10088 idx_T arridx[MAXWLEN];
|
|
323
|
10089 int round[MAXWLEN];
|
|
|
10090 int fwordidx[MAXWLEN];
|
|
|
10091 int uwordidx[MAXWLEN];
|
|
|
10092 int kwordlen[MAXWLEN];
|
|
|
10093
|
|
|
10094 int flen, ulen;
|
|
|
10095 int l;
|
|
|
10096 int len;
|
|
|
10097 int c;
|
|
324
|
10098 idx_T lo, hi, m;
|
|
323
|
10099 char_u *p;
|
|
|
10100 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
10101 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
10102
|
|
|
10103 if (byts == NULL)
|
|
|
10104 {
|
|
|
10105 /* array is empty: "cannot happen" */
|
|
|
10106 *kword = NUL;
|
|
|
10107 return;
|
|
|
10108 }
|
|
|
10109
|
|
|
10110 /* Make an all-cap version of "fword". */
|
|
|
10111 allcap_copy(fword, uword);
|
|
|
10112
|
|
|
10113 /*
|
|
|
10114 * Each character needs to be tried both case-folded and upper-case.
|
|
|
10115 * All this gets very complicated if we keep in mind that changing case
|
|
|
10116 * may change the byte length of a multi-byte character...
|
|
|
10117 */
|
|
|
10118 depth = 0;
|
|
|
10119 arridx[0] = 0;
|
|
|
10120 round[0] = 0;
|
|
|
10121 fwordidx[0] = 0;
|
|
|
10122 uwordidx[0] = 0;
|
|
|
10123 kwordlen[0] = 0;
|
|
|
10124 while (depth >= 0)
|
|
|
10125 {
|
|
|
10126 if (fword[fwordidx[depth]] == NUL)
|
|
|
10127 {
|
|
|
10128 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
10129 * here we have found a match. */
|
|
|
10130 if (byts[arridx[depth] + 1] == 0)
|
|
|
10131 {
|
|
|
10132 kword[kwordlen[depth]] = NUL;
|
|
|
10133 return;
|
|
|
10134 }
|
|
|
10135
|
|
|
10136 /* kword is getting too long, continue one level up */
|
|
|
10137 --depth;
|
|
|
10138 }
|
|
|
10139 else if (++round[depth] > 2)
|
|
|
10140 {
|
|
|
10141 /* tried both fold-case and upper-case character, continue one
|
|
|
10142 * level up */
|
|
|
10143 --depth;
|
|
|
10144 }
|
|
|
10145 else
|
|
|
10146 {
|
|
|
10147 /*
|
|
|
10148 * round[depth] == 1: Try using the folded-case character.
|
|
|
10149 * round[depth] == 2: Try using the upper-case character.
|
|
|
10150 */
|
|
|
10151 #ifdef FEAT_MBYTE
|
|
|
10152 if (has_mbyte)
|
|
|
10153 {
|
|
474
|
10154 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
10155 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
10156 }
|
|
|
10157 else
|
|
|
10158 #endif
|
|
|
10159 ulen = flen = 1;
|
|
|
10160 if (round[depth] == 1)
|
|
|
10161 {
|
|
|
10162 p = fword + fwordidx[depth];
|
|
|
10163 l = flen;
|
|
|
10164 }
|
|
|
10165 else
|
|
|
10166 {
|
|
|
10167 p = uword + uwordidx[depth];
|
|
|
10168 l = ulen;
|
|
|
10169 }
|
|
|
10170
|
|
|
10171 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
10172 {
|
|
|
10173 /* Perform a binary search in the list of accepted bytes. */
|
|
|
10174 len = byts[tryidx++];
|
|
|
10175 c = *p++;
|
|
|
10176 lo = tryidx;
|
|
|
10177 hi = tryidx + len - 1;
|
|
|
10178 while (lo < hi)
|
|
|
10179 {
|
|
|
10180 m = (lo + hi) / 2;
|
|
|
10181 if (byts[m] > c)
|
|
|
10182 hi = m - 1;
|
|
|
10183 else if (byts[m] < c)
|
|
|
10184 lo = m + 1;
|
|
|
10185 else
|
|
|
10186 {
|
|
|
10187 lo = hi = m;
|
|
|
10188 break;
|
|
|
10189 }
|
|
|
10190 }
|
|
|
10191
|
|
|
10192 /* Stop if there is no matching byte. */
|
|
|
10193 if (hi < lo || byts[lo] != c)
|
|
|
10194 break;
|
|
|
10195
|
|
|
10196 /* Continue at the child (if there is one). */
|
|
|
10197 tryidx = idxs[lo];
|
|
|
10198 }
|
|
|
10199
|
|
|
10200 if (l == 0)
|
|
|
10201 {
|
|
|
10202 /*
|
|
|
10203 * Found the matching char. Copy it to "kword" and go a
|
|
|
10204 * level deeper.
|
|
|
10205 */
|
|
|
10206 if (round[depth] == 1)
|
|
|
10207 {
|
|
|
10208 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
10209 flen);
|
|
|
10210 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
10211 }
|
|
|
10212 else
|
|
|
10213 {
|
|
|
10214 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
10215 ulen);
|
|
|
10216 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
10217 }
|
|
|
10218 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
10219 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
10220
|
|
|
10221 ++depth;
|
|
|
10222 arridx[depth] = tryidx;
|
|
|
10223 round[depth] = 0;
|
|
|
10224 }
|
|
|
10225 }
|
|
|
10226 }
|
|
|
10227
|
|
|
10228 /* Didn't find it: "cannot happen". */
|
|
|
10229 *kword = NUL;
|
|
|
10230 }
|
|
|
10231
|
|
|
10232 /*
|
|
344
|
10233 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
10234 * su->su_sga.
|
|
|
10235 */
|
|
|
10236 static void
|
|
|
10237 score_comp_sal(su)
|
|
|
10238 suginfo_T *su;
|
|
|
10239 {
|
|
|
10240 langp_T *lp;
|
|
|
10241 char_u badsound[MAXWLEN];
|
|
|
10242 int i;
|
|
|
10243 suggest_T *stp;
|
|
|
10244 suggest_T *sstp;
|
|
|
10245 int score;
|
|
|
10246
|
|
|
10247 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
10248 return;
|
|
|
10249
|
|
|
10250 /* Use the sound-folding of the first language that supports it. */
|
|
|
10251 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10252 lp->lp_slang != NULL; ++lp)
|
|
|
10253 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10254 {
|
|
|
10255 /* soundfold the bad word */
|
|
375
|
10256 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
10257
|
|
|
10258 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10259 {
|
|
|
10260 stp = &SUG(su->su_ga, i);
|
|
|
10261
|
|
351
|
10262 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
10263 * sound-a-like score. */
|
|
|
10264 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
10265 if (score < SCORE_MAXMAX)
|
|
|
10266 {
|
|
|
10267 /* Add the suggestion. */
|
|
|
10268 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
10269 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
10270 if (sstp->st_word != NULL)
|
|
|
10271 {
|
|
|
10272 sstp->st_score = score;
|
|
|
10273 sstp->st_altscore = 0;
|
|
|
10274 sstp->st_orglen = stp->st_orglen;
|
|
|
10275 ++su->su_sga.ga_len;
|
|
|
10276 }
|
|
|
10277 }
|
|
|
10278 }
|
|
|
10279 break;
|
|
|
10280 }
|
|
|
10281 }
|
|
|
10282
|
|
|
10283 /*
|
|
|
10284 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
10285 * They are intwined.
|
|
|
10286 */
|
|
|
10287 static void
|
|
|
10288 score_combine(su)
|
|
|
10289 suginfo_T *su;
|
|
|
10290 {
|
|
|
10291 int i;
|
|
|
10292 int j;
|
|
|
10293 garray_T ga;
|
|
|
10294 garray_T *gap;
|
|
|
10295 langp_T *lp;
|
|
|
10296 suggest_T *stp;
|
|
|
10297 char_u *p;
|
|
|
10298 char_u badsound[MAXWLEN];
|
|
|
10299 int round;
|
|
|
10300
|
|
|
10301 /* Add the alternate score to su_ga. */
|
|
|
10302 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10303 lp->lp_slang != NULL; ++lp)
|
|
|
10304 {
|
|
|
10305 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10306 {
|
|
|
10307 /* soundfold the bad word */
|
|
375
|
10308 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
10309
|
|
|
10310 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10311 {
|
|
|
10312 stp = &SUG(su->su_ga, i);
|
|
351
|
10313 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
10314 badsound);
|
|
344
|
10315 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
10316 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
10317 else
|
|
|
10318 stp->st_score = (stp->st_score * 3
|
|
|
10319 + stp->st_altscore) / 4;
|
|
|
10320 stp->st_salscore = FALSE;
|
|
|
10321 }
|
|
|
10322 break;
|
|
|
10323 }
|
|
|
10324 }
|
|
|
10325
|
|
|
10326 /* Add the alternate score to su_sga. */
|
|
|
10327 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
10328 {
|
|
|
10329 stp = &SUG(su->su_sga, i);
|
|
|
10330 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
10331 if (stp->st_score == SCORE_MAXMAX)
|
|
|
10332 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
10333 else
|
|
|
10334 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
10335 stp->st_salscore = TRUE;
|
|
|
10336 }
|
|
|
10337
|
|
|
10338 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
10339 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
10340 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
10341
|
|
|
10342 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
10343 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
10344 return;
|
|
|
10345
|
|
|
10346 stp = &SUG(ga, 0);
|
|
|
10347 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
10348 {
|
|
|
10349 /* round 1: get a suggestion from su_ga
|
|
|
10350 * round 2: get a suggestion from su_sga */
|
|
|
10351 for (round = 1; round <= 2; ++round)
|
|
|
10352 {
|
|
|
10353 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
10354 if (i < gap->ga_len)
|
|
|
10355 {
|
|
|
10356 /* Don't add a word if it's already there. */
|
|
|
10357 p = SUG(*gap, i).st_word;
|
|
|
10358 for (j = 0; j < ga.ga_len; ++j)
|
|
|
10359 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
10360 break;
|
|
|
10361 if (j == ga.ga_len)
|
|
|
10362 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
10363 else
|
|
|
10364 vim_free(p);
|
|
|
10365 }
|
|
|
10366 }
|
|
|
10367 }
|
|
|
10368
|
|
|
10369 ga_clear(&su->su_ga);
|
|
|
10370 ga_clear(&su->su_sga);
|
|
|
10371
|
|
|
10372 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
10373 if (ga.ga_len > su->su_maxcount)
|
|
|
10374 {
|
|
|
10375 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
10376 vim_free(stp[i].st_word);
|
|
|
10377 ga.ga_len = su->su_maxcount;
|
|
|
10378 }
|
|
|
10379
|
|
|
10380 su->su_ga = ga;
|
|
|
10381 }
|
|
|
10382
|
|
|
10383 /*
|
|
351
|
10384 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
10385 * badword.
|
|
|
10386 */
|
|
|
10387 static int
|
|
|
10388 stp_sal_score(stp, su, slang, badsound)
|
|
|
10389 suggest_T *stp;
|
|
|
10390 suginfo_T *su;
|
|
|
10391 slang_T *slang;
|
|
|
10392 char_u *badsound; /* sound-folded badword */
|
|
|
10393 {
|
|
|
10394 char_u *p;
|
|
|
10395 char_u badsound2[MAXWLEN];
|
|
|
10396 char_u fword[MAXWLEN];
|
|
|
10397 char_u goodsound[MAXWLEN];
|
|
|
10398
|
|
|
10399 if (stp->st_orglen <= su->su_badlen)
|
|
|
10400 p = badsound;
|
|
|
10401 else
|
|
|
10402 {
|
|
|
10403 /* soundfold the bad word with more characters following */
|
|
|
10404 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
10405
|
|
|
10406 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
10407 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
10408 * removing the space. Don't do it when the good word also contains a
|
|
|
10409 * space. */
|
|
|
10410 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
10411 && *skiptowhite(stp->st_word) == NUL)
|
|
|
10412 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
10413 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
10414
|
|
375
|
10415 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
10416 p = badsound2;
|
|
|
10417 }
|
|
|
10418
|
|
375
|
10419 /* Sound-fold the word and compute the score for the difference. */
|
|
|
10420 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
10421
|
|
|
10422 return soundalike_score(goodsound, p);
|
|
|
10423 }
|
|
|
10424
|
|
|
10425 /*
|
|
323
|
10426 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
10427 */
|
|
|
10428 static void
|
|
346
|
10429 suggest_try_soundalike(su)
|
|
323
|
10430 suginfo_T *su;
|
|
|
10431 {
|
|
|
10432 char_u salword[MAXWLEN];
|
|
|
10433 char_u tword[MAXWLEN];
|
|
|
10434 char_u tsalword[MAXWLEN];
|
|
324
|
10435 idx_T arridx[MAXWLEN];
|
|
323
|
10436 int curi[MAXWLEN];
|
|
|
10437 langp_T *lp;
|
|
|
10438 char_u *byts;
|
|
324
|
10439 idx_T *idxs;
|
|
323
|
10440 int depth;
|
|
|
10441 int c;
|
|
324
|
10442 idx_T n;
|
|
323
|
10443 int round;
|
|
|
10444 int flags;
|
|
344
|
10445 int sound_score;
|
|
483
|
10446 int local_score;
|
|
344
|
10447
|
|
|
10448 /* Do this for all languages that support sound folding. */
|
|
323
|
10449 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10450 lp->lp_slang != NULL; ++lp)
|
|
|
10451 {
|
|
|
10452 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10453 {
|
|
|
10454 /* soundfold the bad word */
|
|
375
|
10455 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
10456
|
|
|
10457 /*
|
|
|
10458 * Go through the whole tree, soundfold each word and compare.
|
|
|
10459 * round 1: use the case-folded tree.
|
|
|
10460 * round 2: use the keep-case tree.
|
|
|
10461 */
|
|
|
10462 for (round = 1; round <= 2; ++round)
|
|
|
10463 {
|
|
|
10464 if (round == 1)
|
|
|
10465 {
|
|
|
10466 byts = lp->lp_slang->sl_fbyts;
|
|
|
10467 idxs = lp->lp_slang->sl_fidxs;
|
|
|
10468 }
|
|
|
10469 else
|
|
|
10470 {
|
|
|
10471 byts = lp->lp_slang->sl_kbyts;
|
|
|
10472 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
10473 if (byts == NULL) /* no keep-case words */
|
|
|
10474 continue;
|
|
323
|
10475 }
|
|
|
10476
|
|
|
10477 depth = 0;
|
|
|
10478 arridx[0] = 0;
|
|
|
10479 curi[0] = 1;
|
|
|
10480 while (depth >= 0 && !got_int)
|
|
|
10481 {
|
|
|
10482 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
10483 {
|
|
323
|
10484 /* Done all bytes at this node, go up one level. */
|
|
|
10485 --depth;
|
|
351
|
10486 line_breakcheck();
|
|
|
10487 }
|
|
323
|
10488 else
|
|
|
10489 {
|
|
|
10490 /* Do one more byte at this node. */
|
|
|
10491 n = arridx[depth] + curi[depth];
|
|
|
10492 ++curi[depth];
|
|
|
10493 c = byts[n];
|
|
|
10494 if (c == 0)
|
|
|
10495 {
|
|
|
10496 /* End of word, deal with the word. */
|
|
324
|
10497 flags = (int)idxs[n];
|
|
323
|
10498 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
10499 {
|
|
|
10500 tword[depth] = NUL;
|
|
375
|
10501 /* Sound-fold. Only in keep-case tree need to
|
|
|
10502 * case-fold the word. */
|
|
|
10503 spell_soundfold(lp->lp_slang, tword,
|
|
|
10504 round == 1, tsalword);
|
|
323
|
10505
|
|
344
|
10506 /* Compute the edit distance between the
|
|
|
10507 * sound-a-like words. */
|
|
|
10508 sound_score = soundalike_score(salword,
|
|
|
10509 tsalword);
|
|
483
|
10510
|
|
|
10511 /* Add a penalty for words in another region. */
|
|
|
10512 if ((flags & WF_REGION) && (((unsigned)flags
|
|
|
10513 >> 16) & lp->lp_region) == 0)
|
|
|
10514 local_score = SCORE_REGION;
|
|
|
10515 else
|
|
|
10516 local_score = 0;
|
|
|
10517 sound_score += local_score;
|
|
|
10518
|
|
324
|
10519 if (sound_score < SCORE_MAXMAX)
|
|
323
|
10520 {
|
|
324
|
10521 char_u cword[MAXWLEN];
|
|
|
10522 char_u *p;
|
|
344
|
10523 int score;
|
|
324
|
10524
|
|
388
|
10525 flags |= su->su_badflags;
|
|
351
|
10526 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
10527 {
|
|
324
|
10528 /* Need to fix case according to
|
|
|
10529 * "flags". */
|
|
323
|
10530 make_case_word(tword, cword, flags);
|
|
324
|
10531 p = cword;
|
|
323
|
10532 }
|
|
|
10533 else
|
|
324
|
10534 p = tword;
|
|
|
10535
|
|
344
|
10536 if (sps_flags & SPS_DOUBLE)
|
|
|
10537 add_suggestion(su, &su->su_sga, p,
|
|
346
|
10538 su->su_badlen,
|
|
351
|
10539 sound_score, 0, FALSE);
|
|
344
|
10540 else
|
|
|
10541 {
|
|
|
10542 /* Compute the score. */
|
|
|
10543 score = spell_edit_score(
|
|
483
|
10544 su->su_badword, p)
|
|
|
10545 + local_score;
|
|
344
|
10546 if (sps_flags & SPS_BEST)
|
|
|
10547 /* give a bonus for the good word
|
|
|
10548 * sounding the same as the bad
|
|
|
10549 * word */
|
|
|
10550 add_suggestion(su, &su->su_ga, p,
|
|
346
|
10551 su->su_badlen,
|
|
344
|
10552 RESCORE(score, sound_score),
|
|
351
|
10553 sound_score, TRUE);
|
|
344
|
10554 else
|
|
|
10555 add_suggestion(su, &su->su_ga, p,
|
|
346
|
10556 su->su_badlen,
|
|
351
|
10557 score + sound_score, 0, FALSE);
|
|
344
|
10558 }
|
|
323
|
10559 }
|
|
|
10560 }
|
|
|
10561
|
|
|
10562 /* Skip over other NUL bytes. */
|
|
|
10563 while (byts[n + 1] == 0)
|
|
|
10564 {
|
|
|
10565 ++n;
|
|
|
10566 ++curi[depth];
|
|
|
10567 }
|
|
|
10568 }
|
|
|
10569 else
|
|
|
10570 {
|
|
|
10571 /* Normal char, go one level deeper. */
|
|
|
10572 tword[depth++] = c;
|
|
|
10573 arridx[depth] = idxs[n];
|
|
|
10574 curi[depth] = 1;
|
|
|
10575 }
|
|
|
10576 }
|
|
|
10577 }
|
|
|
10578 }
|
|
|
10579 }
|
|
|
10580 }
|
|
|
10581 }
|
|
|
10582
|
|
|
10583 /*
|
|
324
|
10584 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
10585 */
|
|
|
10586 static void
|
|
|
10587 make_case_word(fword, cword, flags)
|
|
|
10588 char_u *fword;
|
|
|
10589 char_u *cword;
|
|
|
10590 int flags;
|
|
|
10591 {
|
|
|
10592 if (flags & WF_ALLCAP)
|
|
|
10593 /* Make it all upper-case */
|
|
|
10594 allcap_copy(fword, cword);
|
|
|
10595 else if (flags & WF_ONECAP)
|
|
|
10596 /* Make the first letter upper-case */
|
|
324
|
10597 onecap_copy(fword, cword, TRUE);
|
|
323
|
10598 else
|
|
|
10599 /* Use goodword as-is. */
|
|
|
10600 STRCPY(cword, fword);
|
|
|
10601 }
|
|
|
10602
|
|
330
|
10603 /*
|
|
|
10604 * Use map string "map" for languages "lp".
|
|
|
10605 */
|
|
|
10606 static void
|
|
|
10607 set_map_str(lp, map)
|
|
|
10608 slang_T *lp;
|
|
|
10609 char_u *map;
|
|
|
10610 {
|
|
|
10611 char_u *p;
|
|
|
10612 int headc = 0;
|
|
|
10613 int c;
|
|
|
10614 int i;
|
|
|
10615
|
|
|
10616 if (*map == NUL)
|
|
|
10617 {
|
|
|
10618 lp->sl_has_map = FALSE;
|
|
|
10619 return;
|
|
|
10620 }
|
|
|
10621 lp->sl_has_map = TRUE;
|
|
|
10622
|
|
|
10623 /* Init the array and hash table empty. */
|
|
|
10624 for (i = 0; i < 256; ++i)
|
|
|
10625 lp->sl_map_array[i] = 0;
|
|
|
10626 #ifdef FEAT_MBYTE
|
|
|
10627 hash_init(&lp->sl_map_hash);
|
|
|
10628 #endif
|
|
|
10629
|
|
|
10630 /*
|
|
|
10631 * The similar characters are stored separated with slashes:
|
|
|
10632 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
10633 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
10634 */
|
|
|
10635 for (p = map; *p != NUL; )
|
|
|
10636 {
|
|
|
10637 #ifdef FEAT_MBYTE
|
|
474
|
10638 c = mb_cptr2char_adv(&p);
|
|
330
|
10639 #else
|
|
|
10640 c = *p++;
|
|
|
10641 #endif
|
|
|
10642 if (c == '/')
|
|
|
10643 headc = 0;
|
|
|
10644 else
|
|
|
10645 {
|
|
|
10646 if (headc == 0)
|
|
|
10647 headc = c;
|
|
|
10648
|
|
|
10649 #ifdef FEAT_MBYTE
|
|
|
10650 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
10651 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
10652 * a NUL. */
|
|
|
10653 if (c >= 256)
|
|
|
10654 {
|
|
|
10655 int cl = mb_char2len(c);
|
|
|
10656 int headcl = mb_char2len(headc);
|
|
|
10657 char_u *b;
|
|
|
10658 hash_T hash;
|
|
|
10659 hashitem_T *hi;
|
|
|
10660
|
|
|
10661 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
10662 if (b == NULL)
|
|
|
10663 return;
|
|
|
10664 mb_char2bytes(c, b);
|
|
|
10665 b[cl] = NUL;
|
|
|
10666 mb_char2bytes(headc, b + cl + 1);
|
|
|
10667 b[cl + 1 + headcl] = NUL;
|
|
|
10668 hash = hash_hash(b);
|
|
|
10669 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
10670 if (HASHITEM_EMPTY(hi))
|
|
|
10671 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
10672 else
|
|
|
10673 {
|
|
|
10674 /* This should have been checked when generating the .spl
|
|
|
10675 * file. */
|
|
|
10676 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
10677 vim_free(b);
|
|
|
10678 }
|
|
|
10679 }
|
|
|
10680 else
|
|
|
10681 #endif
|
|
|
10682 lp->sl_map_array[c] = headc;
|
|
|
10683 }
|
|
|
10684 }
|
|
|
10685 }
|
|
|
10686
|
|
323
|
10687 /*
|
|
|
10688 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
10689 * lines in the .aff file.
|
|
|
10690 */
|
|
|
10691 static int
|
|
|
10692 similar_chars(slang, c1, c2)
|
|
|
10693 slang_T *slang;
|
|
|
10694 int c1;
|
|
|
10695 int c2;
|
|
|
10696 {
|
|
330
|
10697 int m1, m2;
|
|
|
10698 #ifdef FEAT_MBYTE
|
|
|
10699 char_u buf[MB_MAXBYTES];
|
|
|
10700 hashitem_T *hi;
|
|
|
10701
|
|
|
10702 if (c1 >= 256)
|
|
|
10703 {
|
|
|
10704 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
10705 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
10706 if (HASHITEM_EMPTY(hi))
|
|
|
10707 m1 = 0;
|
|
|
10708 else
|
|
|
10709 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
10710 }
|
|
|
10711 else
|
|
|
10712 #endif
|
|
|
10713 m1 = slang->sl_map_array[c1];
|
|
|
10714 if (m1 == 0)
|
|
323
|
10715 return FALSE;
|
|
330
|
10716
|
|
|
10717
|
|
|
10718 #ifdef FEAT_MBYTE
|
|
|
10719 if (c2 >= 256)
|
|
|
10720 {
|
|
|
10721 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
10722 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
10723 if (HASHITEM_EMPTY(hi))
|
|
|
10724 m2 = 0;
|
|
|
10725 else
|
|
|
10726 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
10727 }
|
|
|
10728 else
|
|
|
10729 #endif
|
|
|
10730 m2 = slang->sl_map_array[c2];
|
|
|
10731
|
|
|
10732 return m1 == m2;
|
|
323
|
10733 }
|
|
|
10734
|
|
|
10735 /*
|
|
|
10736 * Add a suggestion to the list of suggestions.
|
|
|
10737 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
10738 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
10739 * with spell_edit_score().
|
|
|
10740 */
|
|
|
10741 static void
|
|
351
|
10742 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
10743 suginfo_T *su;
|
|
344
|
10744 garray_T *gap;
|
|
323
|
10745 char_u *goodword;
|
|
346
|
10746 int badlen; /* length of bad word used */
|
|
324
|
10747 int score;
|
|
351
|
10748 int altscore;
|
|
344
|
10749 int had_bonus; /* value for st_had_bonus */
|
|
323
|
10750 {
|
|
|
10751 suggest_T *stp;
|
|
|
10752 int i;
|
|
346
|
10753 char_u *p = NULL;
|
|
|
10754 int c = 0;
|
|
323
|
10755
|
|
|
10756 /* Check that the word wasn't banned. */
|
|
|
10757 if (was_banned(su, goodword))
|
|
|
10758 return;
|
|
|
10759
|
|
346
|
10760 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
10761 * Remove the common part from "goodword". */
|
|
|
10762 i = badlen - su->su_badlen;
|
|
|
10763 if (i > 0)
|
|
|
10764 {
|
|
|
10765 /* This assumes there was no case folding or it didn't change the
|
|
|
10766 * length... */
|
|
|
10767 p = goodword + STRLEN(goodword) - i;
|
|
|
10768 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
10769 {
|
|
|
10770 badlen = su->su_badlen;
|
|
|
10771 c = *p;
|
|
|
10772 *p = NUL;
|
|
|
10773 }
|
|
|
10774 else
|
|
|
10775 p = NULL;
|
|
|
10776 }
|
|
372
|
10777 else if (i < 0)
|
|
|
10778 {
|
|
|
10779 /* When replacing part of the word check that we actually change
|
|
|
10780 * something. For "the the" a suggestion can be replacing the first
|
|
|
10781 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
10782 if (badlen == (int)STRLEN(goodword)
|
|
372
|
10783 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
10784 return;
|
|
|
10785 }
|
|
|
10786
|
|
346
|
10787
|
|
323
|
10788 if (score <= su->su_maxscore)
|
|
|
10789 {
|
|
366
|
10790 /* Check if the word is already there. Also check the length that is
|
|
|
10791 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
10792 * "thes" -> "these". */
|
|
344
|
10793 stp = &SUG(*gap, 0);
|
|
|
10794 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
10795 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
10796 && stp[i].st_orglen == badlen)
|
|
323
|
10797 {
|
|
|
10798 /* Found it. Remember the lowest score. */
|
|
|
10799 if (stp[i].st_score > score)
|
|
324
|
10800 {
|
|
323
|
10801 stp[i].st_score = score;
|
|
376
|
10802 stp[i].st_altscore = altscore;
|
|
324
|
10803 stp[i].st_had_bonus = had_bonus;
|
|
|
10804 }
|
|
323
|
10805 break;
|
|
|
10806 }
|
|
|
10807
|
|
344
|
10808 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
10809 {
|
|
|
10810 /* Add a suggestion. */
|
|
344
|
10811 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
10812 stp->st_word = vim_strsave(goodword);
|
|
|
10813 if (stp->st_word != NULL)
|
|
|
10814 {
|
|
|
10815 stp->st_score = score;
|
|
351
|
10816 stp->st_altscore = altscore;
|
|
324
|
10817 stp->st_had_bonus = had_bonus;
|
|
346
|
10818 stp->st_orglen = badlen;
|
|
344
|
10819 ++gap->ga_len;
|
|
323
|
10820
|
|
|
10821 /* If we have too many suggestions now, sort the list and keep
|
|
|
10822 * the best suggestions. */
|
|
344
|
10823 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
10824 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
10825 SUG_CLEAN_COUNT(su));
|
|
323
|
10826 }
|
|
|
10827 }
|
|
|
10828 }
|
|
346
|
10829
|
|
|
10830 if (p != NULL)
|
|
|
10831 *p = c; /* restore "goodword" */
|
|
323
|
10832 }
|
|
|
10833
|
|
|
10834 /*
|
|
|
10835 * Add a word to be banned.
|
|
|
10836 */
|
|
|
10837 static void
|
|
|
10838 add_banned(su, word)
|
|
|
10839 suginfo_T *su;
|
|
|
10840 char_u *word;
|
|
|
10841 {
|
|
|
10842 char_u *s = vim_strsave(word);
|
|
|
10843 hash_T hash;
|
|
|
10844 hashitem_T *hi;
|
|
|
10845
|
|
|
10846 if (s != NULL)
|
|
|
10847 {
|
|
|
10848 hash = hash_hash(s);
|
|
|
10849 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
10850 if (HASHITEM_EMPTY(hi))
|
|
|
10851 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
10852 else
|
|
|
10853 vim_free(s);
|
|
323
|
10854 }
|
|
|
10855 }
|
|
|
10856
|
|
|
10857 /*
|
|
|
10858 * Return TRUE if a word appears in the list of banned words.
|
|
|
10859 */
|
|
|
10860 static int
|
|
|
10861 was_banned(su, word)
|
|
|
10862 suginfo_T *su;
|
|
|
10863 char_u *word;
|
|
|
10864 {
|
|
324
|
10865 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
10866
|
|
|
10867 return !HASHITEM_EMPTY(hi);
|
|
323
|
10868 }
|
|
|
10869
|
|
|
10870 /*
|
|
|
10871 * Free the banned words in "su".
|
|
|
10872 */
|
|
|
10873 static void
|
|
|
10874 free_banned(su)
|
|
|
10875 suginfo_T *su;
|
|
|
10876 {
|
|
|
10877 int todo;
|
|
|
10878 hashitem_T *hi;
|
|
|
10879
|
|
|
10880 todo = su->su_banned.ht_used;
|
|
|
10881 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
10882 {
|
|
|
10883 if (!HASHITEM_EMPTY(hi))
|
|
|
10884 {
|
|
|
10885 vim_free(hi->hi_key);
|
|
|
10886 --todo;
|
|
|
10887 }
|
|
|
10888 }
|
|
|
10889 hash_clear(&su->su_banned);
|
|
|
10890 }
|
|
|
10891
|
|
324
|
10892 /*
|
|
|
10893 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
10894 * thus only done for the final results.
|
|
|
10895 */
|
|
|
10896 static void
|
|
|
10897 rescore_suggestions(su)
|
|
|
10898 suginfo_T *su;
|
|
|
10899 {
|
|
|
10900 langp_T *lp;
|
|
|
10901 suggest_T *stp;
|
|
|
10902 char_u sal_badword[MAXWLEN];
|
|
|
10903 int i;
|
|
|
10904
|
|
|
10905 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10906 lp->lp_slang != NULL; ++lp)
|
|
|
10907 {
|
|
|
10908 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10909 {
|
|
|
10910 /* soundfold the bad word */
|
|
375
|
10911 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
10912
|
|
|
10913 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10914 {
|
|
344
|
10915 stp = &SUG(su->su_ga, i);
|
|
324
|
10916 if (!stp->st_had_bonus)
|
|
|
10917 {
|
|
351
|
10918 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
10919 lp->lp_slang, sal_badword);
|
|
|
10920 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
10921 stp->st_altscore = SCORE_BIG;
|
|
|
10922 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
10923 }
|
|
|
10924 }
|
|
|
10925 break;
|
|
|
10926 }
|
|
|
10927 }
|
|
|
10928 }
|
|
|
10929
|
|
323
|
10930 static int
|
|
|
10931 #ifdef __BORLANDC__
|
|
|
10932 _RTLENTRYF
|
|
|
10933 #endif
|
|
|
10934 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
10935
|
|
|
10936 /*
|
|
|
10937 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
10938 */
|
|
|
10939 static int
|
|
|
10940 #ifdef __BORLANDC__
|
|
|
10941 _RTLENTRYF
|
|
|
10942 #endif
|
|
|
10943 sug_compare(s1, s2)
|
|
|
10944 const void *s1;
|
|
|
10945 const void *s2;
|
|
|
10946 {
|
|
|
10947 suggest_T *p1 = (suggest_T *)s1;
|
|
|
10948 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
10949 int n = p1->st_score - p2->st_score;
|
|
|
10950
|
|
|
10951 if (n == 0)
|
|
|
10952 return p1->st_altscore - p2->st_altscore;
|
|
|
10953 return n;
|
|
323
|
10954 }
|
|
|
10955
|
|
|
10956 /*
|
|
|
10957 * Cleanup the suggestions:
|
|
|
10958 * - Sort on score.
|
|
|
10959 * - Remove words that won't be displayed.
|
|
344
|
10960 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
10961 */
|
|
344
|
10962 static int
|
|
|
10963 cleanup_suggestions(gap, maxscore, keep)
|
|
|
10964 garray_T *gap;
|
|
|
10965 int maxscore;
|
|
324
|
10966 int keep; /* nr of suggestions to keep */
|
|
323
|
10967 {
|
|
344
|
10968 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
10969 int i;
|
|
|
10970
|
|
|
10971 /* Sort the list. */
|
|
344
|
10972 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
10973
|
|
|
10974 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
10975 if (gap->ga_len > keep)
|
|
323
|
10976 {
|
|
344
|
10977 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
10978 vim_free(stp[i].st_word);
|
|
344
|
10979 gap->ga_len = keep;
|
|
|
10980 return stp[keep - 1].st_score;
|
|
323
|
10981 }
|
|
344
|
10982 return maxscore;
|
|
323
|
10983 }
|
|
|
10984
|
|
372
|
10985 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
10986 /*
|
|
|
10987 * Soundfold a string, for soundfold().
|
|
|
10988 * Result is in allocated memory, NULL for an error.
|
|
|
10989 */
|
|
|
10990 char_u *
|
|
|
10991 eval_soundfold(word)
|
|
|
10992 char_u *word;
|
|
|
10993 {
|
|
|
10994 langp_T *lp;
|
|
|
10995 char_u sound[MAXWLEN];
|
|
|
10996
|
|
|
10997 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
10998 /* Use the sound-folding of the first language that supports it. */
|
|
|
10999 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
11000 lp->lp_slang != NULL; ++lp)
|
|
|
11001 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
11002 {
|
|
|
11003 /* soundfold the word */
|
|
375
|
11004 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
11005 return vim_strsave(sound);
|
|
|
11006 }
|
|
|
11007
|
|
|
11008 /* No language with sound folding, return word as-is. */
|
|
|
11009 return vim_strsave(word);
|
|
|
11010 }
|
|
|
11011 #endif
|
|
|
11012
|
|
323
|
11013 /*
|
|
|
11014 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
485
|
11015 *
|
|
|
11016 * There are many ways to turn a word into a sound-a-like representation. The
|
|
|
11017 * oldest is Soundex (1918!). A nice overview can be found in "Approximate
|
|
|
11018 * swedish name matching - survey and test of different algorithms" by Klas
|
|
|
11019 * Erikson.
|
|
|
11020 *
|
|
|
11021 * We support two methods:
|
|
|
11022 * 1. SOFOFROM/SOFOTO do a simple character mapping.
|
|
|
11023 * 2. SAL items define a more advanced sound-folding (and much slower).
|
|
323
|
11024 */
|
|
|
11025 static void
|
|
375
|
11026 spell_soundfold(slang, inword, folded, res)
|
|
|
11027 slang_T *slang;
|
|
|
11028 char_u *inword;
|
|
|
11029 int folded; /* "inword" is already case-folded */
|
|
|
11030 char_u *res;
|
|
|
11031 {
|
|
|
11032 char_u fword[MAXWLEN];
|
|
|
11033 char_u *word;
|
|
|
11034
|
|
|
11035 if (slang->sl_sofo)
|
|
|
11036 /* SOFOFROM and SOFOTO used */
|
|
|
11037 spell_soundfold_sofo(slang, inword, res);
|
|
|
11038 else
|
|
|
11039 {
|
|
|
11040 /* SAL items used. Requires the word to be case-folded. */
|
|
|
11041 if (folded)
|
|
|
11042 word = inword;
|
|
|
11043 else
|
|
|
11044 {
|
|
|
11045 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
11046 word = fword;
|
|
|
11047 }
|
|
|
11048
|
|
|
11049 #ifdef FEAT_MBYTE
|
|
|
11050 if (has_mbyte)
|
|
|
11051 spell_soundfold_wsal(slang, word, res);
|
|
|
11052 else
|
|
|
11053 #endif
|
|
|
11054 spell_soundfold_sal(slang, word, res);
|
|
|
11055 }
|
|
|
11056 }
|
|
|
11057
|
|
|
11058 /*
|
|
|
11059 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
11060 * SOFOTO lines.
|
|
|
11061 */
|
|
|
11062 static void
|
|
|
11063 spell_soundfold_sofo(slang, inword, res)
|
|
|
11064 slang_T *slang;
|
|
|
11065 char_u *inword;
|
|
|
11066 char_u *res;
|
|
|
11067 {
|
|
|
11068 char_u *s;
|
|
|
11069 int ri = 0;
|
|
|
11070 int c;
|
|
|
11071
|
|
|
11072 #ifdef FEAT_MBYTE
|
|
|
11073 if (has_mbyte)
|
|
|
11074 {
|
|
|
11075 int prevc = 0;
|
|
|
11076 int *ip;
|
|
|
11077
|
|
|
11078 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
11079 * 255, sl_sal the rest. */
|
|
|
11080 for (s = inword; *s != NUL; )
|
|
|
11081 {
|
|
474
|
11082 c = mb_cptr2char_adv(&s);
|
|
375
|
11083 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
11084 c = ' ';
|
|
|
11085 else if (c < 256)
|
|
|
11086 c = slang->sl_sal_first[c];
|
|
|
11087 else
|
|
|
11088 {
|
|
|
11089 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
11090 if (ip == NULL) /* empty list, can't match */
|
|
|
11091 c = NUL;
|
|
|
11092 else
|
|
|
11093 for (;;) /* find "c" in the list */
|
|
|
11094 {
|
|
|
11095 if (*ip == 0) /* not found */
|
|
|
11096 {
|
|
|
11097 c = NUL;
|
|
|
11098 break;
|
|
|
11099 }
|
|
|
11100 if (*ip == c) /* match! */
|
|
|
11101 {
|
|
|
11102 c = ip[1];
|
|
|
11103 break;
|
|
|
11104 }
|
|
|
11105 ip += 2;
|
|
|
11106 }
|
|
|
11107 }
|
|
|
11108
|
|
|
11109 if (c != NUL && c != prevc)
|
|
|
11110 {
|
|
|
11111 ri += mb_char2bytes(c, res + ri);
|
|
|
11112 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
11113 break;
|
|
|
11114 prevc = c;
|
|
|
11115 }
|
|
|
11116 }
|
|
|
11117 }
|
|
|
11118 else
|
|
|
11119 #endif
|
|
|
11120 {
|
|
|
11121 /* The sl_sal_first[] table contains the translation. */
|
|
|
11122 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
11123 {
|
|
|
11124 if (vim_iswhite(c))
|
|
|
11125 c = ' ';
|
|
|
11126 else
|
|
|
11127 c = slang->sl_sal_first[c];
|
|
|
11128 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
11129 res[ri++] = c;
|
|
|
11130 }
|
|
|
11131 }
|
|
|
11132
|
|
|
11133 res[ri] = NUL;
|
|
|
11134 }
|
|
|
11135
|
|
|
11136 static void
|
|
|
11137 spell_soundfold_sal(slang, inword, res)
|
|
323
|
11138 slang_T *slang;
|
|
|
11139 char_u *inword;
|
|
|
11140 char_u *res;
|
|
|
11141 {
|
|
344
|
11142 salitem_T *smp;
|
|
323
|
11143 char_u word[MAXWLEN];
|
|
375
|
11144 char_u *s = inword;
|
|
323
|
11145 char_u *t;
|
|
344
|
11146 char_u *pf;
|
|
323
|
11147 int i, j, z;
|
|
344
|
11148 int reslen;
|
|
323
|
11149 int n, k = 0;
|
|
|
11150 int z0;
|
|
|
11151 int k0;
|
|
|
11152 int n0;
|
|
|
11153 int c;
|
|
|
11154 int pri;
|
|
|
11155 int p0 = -333;
|
|
|
11156 int c0;
|
|
|
11157
|
|
324
|
11158 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
11159 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
11160 if (slang->sl_rem_accents)
|
|
|
11161 {
|
|
|
11162 t = word;
|
|
375
|
11163 while (*s != NUL)
|
|
323
|
11164 {
|
|
324
|
11165 if (vim_iswhite(*s))
|
|
344
|
11166 {
|
|
|
11167 *t++ = ' ';
|
|
|
11168 s = skipwhite(s);
|
|
|
11169 }
|
|
323
|
11170 else
|
|
|
11171 {
|
|
376
|
11172 if (spell_iswordp_nmw(s))
|
|
323
|
11173 *t++ = *s;
|
|
|
11174 ++s;
|
|
|
11175 }
|
|
|
11176 }
|
|
|
11177 *t = NUL;
|
|
|
11178 }
|
|
|
11179 else
|
|
375
|
11180 STRCPY(word, s);
|
|
323
|
11181
|
|
344
|
11182 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
11183
|
|
|
11184 /*
|
|
|
11185 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
11186 * Changed to keep spaces.
|
|
323
|
11187 */
|
|
344
|
11188 i = reslen = z = 0;
|
|
323
|
11189 while ((c = word[i]) != NUL)
|
|
|
11190 {
|
|
344
|
11191 /* Start with the first rule that has the character in the word. */
|
|
323
|
11192 n = slang->sl_sal_first[c];
|
|
|
11193 z0 = 0;
|
|
|
11194
|
|
|
11195 if (n >= 0)
|
|
|
11196 {
|
|
|
11197 /* check all rules for the same letter */
|
|
344
|
11198 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
11199 {
|
|
344
|
11200 /* Quickly skip entries that don't match the word. Most
|
|
|
11201 * entries are less then three chars, optimize for that. */
|
|
|
11202 k = smp[n].sm_leadlen;
|
|
|
11203 if (k > 1)
|
|
323
|
11204 {
|
|
344
|
11205 if (word[i + 1] != s[1])
|
|
|
11206 continue;
|
|
|
11207 if (k > 2)
|
|
|
11208 {
|
|
|
11209 for (j = 2; j < k; ++j)
|
|
|
11210 if (word[i + j] != s[j])
|
|
|
11211 break;
|
|
|
11212 if (j < k)
|
|
|
11213 continue;
|
|
|
11214 }
|
|
323
|
11215 }
|
|
|
11216
|
|
375
|
11217 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
11218 {
|
|
375
|
11219 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
11220 while (*pf != NUL && *pf != word[i + k])
|
|
|
11221 ++pf;
|
|
|
11222 if (*pf == NUL)
|
|
|
11223 continue;
|
|
|
11224 ++k;
|
|
323
|
11225 }
|
|
344
|
11226 s = smp[n].sm_rules;
|
|
|
11227 pri = 5; /* default priority */
|
|
323
|
11228
|
|
|
11229 p0 = *s;
|
|
|
11230 k0 = k;
|
|
|
11231 while (*s == '-' && k > 1)
|
|
|
11232 {
|
|
|
11233 k--;
|
|
|
11234 s++;
|
|
|
11235 }
|
|
|
11236 if (*s == '<')
|
|
|
11237 s++;
|
|
344
|
11238 if (VIM_ISDIGIT(*s))
|
|
323
|
11239 {
|
|
|
11240 /* determine priority */
|
|
|
11241 pri = *s - '0';
|
|
|
11242 s++;
|
|
|
11243 }
|
|
|
11244 if (*s == '^' && *(s + 1) == '^')
|
|
|
11245 s++;
|
|
|
11246
|
|
|
11247 if (*s == NUL
|
|
|
11248 || (*s == '^'
|
|
324
|
11249 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
11250 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
11251 && (*(s + 1) != '$'
|
|
376
|
11252 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
11253 || (*s == '$' && i > 0
|
|
376
|
11254 && spell_iswordp(word + i - 1, curbuf)
|
|
|
11255 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
11256 {
|
|
|
11257 /* search for followup rules, if: */
|
|
|
11258 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
11259 c0 = word[i + k - 1];
|
|
|
11260 n0 = slang->sl_sal_first[c0];
|
|
|
11261
|
|
|
11262 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
11263 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
11264 {
|
|
|
11265 /* test follow-up rule for "word[i + k]" */
|
|
344
|
11266 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
11267 {
|
|
344
|
11268 /* Quickly skip entries that don't match the word.
|
|
|
11269 * */
|
|
|
11270 k0 = smp[n0].sm_leadlen;
|
|
|
11271 if (k0 > 1)
|
|
323
|
11272 {
|
|
344
|
11273 if (word[i + k] != s[1])
|
|
|
11274 continue;
|
|
|
11275 if (k0 > 2)
|
|
|
11276 {
|
|
|
11277 pf = word + i + k + 1;
|
|
|
11278 for (j = 2; j < k0; ++j)
|
|
|
11279 if (*pf++ != s[j])
|
|
|
11280 break;
|
|
|
11281 if (j < k0)
|
|
|
11282 continue;
|
|
|
11283 }
|
|
323
|
11284 }
|
|
344
|
11285 k0 += k - 1;
|
|
|
11286
|
|
375
|
11287 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
11288 {
|
|
344
|
11289 /* Check for match with one of the chars in
|
|
375
|
11290 * "sm_oneof". */
|
|
344
|
11291 while (*pf != NUL && *pf != word[i + k0])
|
|
|
11292 ++pf;
|
|
|
11293 if (*pf == NUL)
|
|
|
11294 continue;
|
|
|
11295 ++k0;
|
|
323
|
11296 }
|
|
344
|
11297
|
|
|
11298 p0 = 5;
|
|
|
11299 s = smp[n0].sm_rules;
|
|
323
|
11300 while (*s == '-')
|
|
|
11301 {
|
|
344
|
11302 /* "k0" gets NOT reduced because
|
|
|
11303 * "if (k0 == k)" */
|
|
323
|
11304 s++;
|
|
|
11305 }
|
|
|
11306 if (*s == '<')
|
|
|
11307 s++;
|
|
344
|
11308 if (VIM_ISDIGIT(*s))
|
|
323
|
11309 {
|
|
|
11310 p0 = *s - '0';
|
|
|
11311 s++;
|
|
|
11312 }
|
|
|
11313
|
|
|
11314 if (*s == NUL
|
|
|
11315 /* *s == '^' cuts */
|
|
|
11316 || (*s == '$'
|
|
376
|
11317 && !spell_iswordp(word + i + k0,
|
|
|
11318 curbuf)))
|
|
323
|
11319 {
|
|
|
11320 if (k0 == k)
|
|
|
11321 /* this is just a piece of the string */
|
|
|
11322 continue;
|
|
|
11323
|
|
|
11324 if (p0 < pri)
|
|
|
11325 /* priority too low */
|
|
|
11326 continue;
|
|
|
11327 /* rule fits; stop search */
|
|
|
11328 break;
|
|
|
11329 }
|
|
|
11330 }
|
|
|
11331
|
|
344
|
11332 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
11333 continue;
|
|
|
11334 }
|
|
|
11335
|
|
|
11336 /* replace string */
|
|
344
|
11337 s = smp[n].sm_to;
|
|
389
|
11338 if (s == NULL)
|
|
|
11339 s = (char_u *)"";
|
|
344
|
11340 pf = smp[n].sm_rules;
|
|
|
11341 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
11342 if (p0 == 1 && z == 0)
|
|
|
11343 {
|
|
|
11344 /* rule with '<' is used */
|
|
344
|
11345 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
11346 || res[reslen - 1] == *s))
|
|
|
11347 reslen--;
|
|
323
|
11348 z0 = 1;
|
|
|
11349 z = 1;
|
|
|
11350 k0 = 0;
|
|
372
|
11351 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
11352 {
|
|
|
11353 word[i + k0] = *s;
|
|
|
11354 k0++;
|
|
|
11355 s++;
|
|
|
11356 }
|
|
|
11357 if (k > k0)
|
|
|
11358 mch_memmove(word + i + k0, word + i + k,
|
|
|
11359 STRLEN(word + i + k) + 1);
|
|
|
11360
|
|
|
11361 /* new "actual letter" */
|
|
|
11362 c = word[i];
|
|
|
11363 }
|
|
|
11364 else
|
|
|
11365 {
|
|
|
11366 /* no '<' rule used */
|
|
|
11367 i += k - 1;
|
|
|
11368 z = 0;
|
|
344
|
11369 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
11370 {
|
|
344
|
11371 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
11372 res[reslen++] = *s;
|
|
323
|
11373 s++;
|
|
|
11374 }
|
|
|
11375 /* new "actual letter" */
|
|
|
11376 c = *s;
|
|
344
|
11377 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
11378 {
|
|
|
11379 if (c != NUL)
|
|
372
|
11380 res[reslen++] = c;
|
|
323
|
11381 mch_memmove(word, word + i + 1,
|
|
|
11382 STRLEN(word + i + 1) + 1);
|
|
|
11383 i = 0;
|
|
|
11384 z0 = 1;
|
|
|
11385 }
|
|
|
11386 }
|
|
|
11387 break;
|
|
|
11388 }
|
|
|
11389 }
|
|
|
11390 }
|
|
324
|
11391 else if (vim_iswhite(c))
|
|
|
11392 {
|
|
|
11393 c = ' ';
|
|
|
11394 k = 1;
|
|
|
11395 }
|
|
323
|
11396
|
|
|
11397 if (z0 == 0)
|
|
|
11398 {
|
|
344
|
11399 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
11400 && (!slang->sl_collapse || reslen == 0
|
|
|
11401 || res[reslen - 1] != c))
|
|
323
|
11402 /* condense only double letters */
|
|
372
|
11403 res[reslen++] = c;
|
|
323
|
11404
|
|
|
11405 i++;
|
|
|
11406 z = 0;
|
|
|
11407 k = 0;
|
|
|
11408 }
|
|
|
11409 }
|
|
|
11410
|
|
344
|
11411 res[reslen] = NUL;
|
|
323
|
11412 }
|
|
|
11413
|
|
372
|
11414 #ifdef FEAT_MBYTE
|
|
|
11415 /*
|
|
|
11416 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
11417 * Multi-byte version of spell_soundfold().
|
|
|
11418 */
|
|
|
11419 static void
|
|
375
|
11420 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
11421 slang_T *slang;
|
|
|
11422 char_u *inword;
|
|
|
11423 char_u *res;
|
|
|
11424 {
|
|
375
|
11425 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
11426 int word[MAXWLEN];
|
|
|
11427 int wres[MAXWLEN];
|
|
|
11428 int l;
|
|
|
11429 char_u *s;
|
|
|
11430 int *ws;
|
|
|
11431 char_u *t;
|
|
|
11432 int *pf;
|
|
|
11433 int i, j, z;
|
|
|
11434 int reslen;
|
|
|
11435 int n, k = 0;
|
|
|
11436 int z0;
|
|
|
11437 int k0;
|
|
|
11438 int n0;
|
|
|
11439 int c;
|
|
|
11440 int pri;
|
|
|
11441 int p0 = -333;
|
|
|
11442 int c0;
|
|
|
11443 int did_white = FALSE;
|
|
|
11444
|
|
|
11445 /*
|
|
|
11446 * Convert the multi-byte string to a wide-character string.
|
|
|
11447 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
11448 * But keep white space.
|
|
|
11449 */
|
|
|
11450 n = 0;
|
|
|
11451 for (s = inword; *s != NUL; )
|
|
|
11452 {
|
|
|
11453 t = s;
|
|
474
|
11454 c = mb_cptr2char_adv(&s);
|
|
372
|
11455 if (slang->sl_rem_accents)
|
|
|
11456 {
|
|
|
11457 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
11458 {
|
|
|
11459 if (did_white)
|
|
|
11460 continue;
|
|
|
11461 c = ' ';
|
|
|
11462 did_white = TRUE;
|
|
|
11463 }
|
|
|
11464 else
|
|
|
11465 {
|
|
|
11466 did_white = FALSE;
|
|
376
|
11467 if (!spell_iswordp_nmw(t))
|
|
372
|
11468 continue;
|
|
|
11469 }
|
|
|
11470 }
|
|
|
11471 word[n++] = c;
|
|
|
11472 }
|
|
|
11473 word[n] = NUL;
|
|
|
11474
|
|
|
11475 /*
|
|
|
11476 * This comes from Aspell phonet.cpp.
|
|
|
11477 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
11478 * Changed to keep spaces.
|
|
|
11479 */
|
|
|
11480 i = reslen = z = 0;
|
|
|
11481 while ((c = word[i]) != NUL)
|
|
|
11482 {
|
|
|
11483 /* Start with the first rule that has the character in the word. */
|
|
|
11484 n = slang->sl_sal_first[c & 0xff];
|
|
|
11485 z0 = 0;
|
|
|
11486
|
|
|
11487 if (n >= 0)
|
|
|
11488 {
|
|
375
|
11489 /* check all rules for the same index byte */
|
|
372
|
11490 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
11491 {
|
|
|
11492 /* Quickly skip entries that don't match the word. Most
|
|
|
11493 * entries are less then three chars, optimize for that. */
|
|
375
|
11494 if (c != ws[0])
|
|
|
11495 continue;
|
|
372
|
11496 k = smp[n].sm_leadlen;
|
|
|
11497 if (k > 1)
|
|
|
11498 {
|
|
|
11499 if (word[i + 1] != ws[1])
|
|
|
11500 continue;
|
|
|
11501 if (k > 2)
|
|
|
11502 {
|
|
|
11503 for (j = 2; j < k; ++j)
|
|
|
11504 if (word[i + j] != ws[j])
|
|
|
11505 break;
|
|
|
11506 if (j < k)
|
|
|
11507 continue;
|
|
|
11508 }
|
|
|
11509 }
|
|
|
11510
|
|
375
|
11511 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
11512 {
|
|
375
|
11513 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
11514 while (*pf != NUL && *pf != word[i + k])
|
|
|
11515 ++pf;
|
|
|
11516 if (*pf == NUL)
|
|
|
11517 continue;
|
|
|
11518 ++k;
|
|
|
11519 }
|
|
|
11520 s = smp[n].sm_rules;
|
|
|
11521 pri = 5; /* default priority */
|
|
|
11522
|
|
|
11523 p0 = *s;
|
|
|
11524 k0 = k;
|
|
|
11525 while (*s == '-' && k > 1)
|
|
|
11526 {
|
|
|
11527 k--;
|
|
|
11528 s++;
|
|
|
11529 }
|
|
|
11530 if (*s == '<')
|
|
|
11531 s++;
|
|
|
11532 if (VIM_ISDIGIT(*s))
|
|
|
11533 {
|
|
|
11534 /* determine priority */
|
|
|
11535 pri = *s - '0';
|
|
|
11536 s++;
|
|
|
11537 }
|
|
|
11538 if (*s == '^' && *(s + 1) == '^')
|
|
|
11539 s++;
|
|
|
11540
|
|
|
11541 if (*s == NUL
|
|
|
11542 || (*s == '^'
|
|
|
11543 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
11544 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
11545 && (*(s + 1) != '$'
|
|
376
|
11546 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
11547 || (*s == '$' && i > 0
|
|
376
|
11548 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
11549 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
11550 {
|
|
|
11551 /* search for followup rules, if: */
|
|
|
11552 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
11553 c0 = word[i + k - 1];
|
|
|
11554 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
11555
|
|
|
11556 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
11557 && p0 != '-' && word[i + k] != NUL)
|
|
|
11558 {
|
|
375
|
11559 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
11560 * all entries with the same index byte. */
|
|
372
|
11561 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
11562 == (c0 & 0xff); ++n0)
|
|
|
11563 {
|
|
|
11564 /* Quickly skip entries that don't match the word.
|
|
375
|
11565 */
|
|
|
11566 if (c0 != ws[0])
|
|
|
11567 continue;
|
|
372
|
11568 k0 = smp[n0].sm_leadlen;
|
|
|
11569 if (k0 > 1)
|
|
|
11570 {
|
|
|
11571 if (word[i + k] != ws[1])
|
|
|
11572 continue;
|
|
|
11573 if (k0 > 2)
|
|
|
11574 {
|
|
|
11575 pf = word + i + k + 1;
|
|
|
11576 for (j = 2; j < k0; ++j)
|
|
|
11577 if (*pf++ != ws[j])
|
|
|
11578 break;
|
|
|
11579 if (j < k0)
|
|
|
11580 continue;
|
|
|
11581 }
|
|
|
11582 }
|
|
|
11583 k0 += k - 1;
|
|
|
11584
|
|
375
|
11585 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
11586 {
|
|
|
11587 /* Check for match with one of the chars in
|
|
375
|
11588 * "sm_oneof". */
|
|
372
|
11589 while (*pf != NUL && *pf != word[i + k0])
|
|
|
11590 ++pf;
|
|
|
11591 if (*pf == NUL)
|
|
|
11592 continue;
|
|
|
11593 ++k0;
|
|
|
11594 }
|
|
|
11595
|
|
|
11596 p0 = 5;
|
|
|
11597 s = smp[n0].sm_rules;
|
|
|
11598 while (*s == '-')
|
|
|
11599 {
|
|
|
11600 /* "k0" gets NOT reduced because
|
|
|
11601 * "if (k0 == k)" */
|
|
|
11602 s++;
|
|
|
11603 }
|
|
|
11604 if (*s == '<')
|
|
|
11605 s++;
|
|
|
11606 if (VIM_ISDIGIT(*s))
|
|
|
11607 {
|
|
|
11608 p0 = *s - '0';
|
|
|
11609 s++;
|
|
|
11610 }
|
|
|
11611
|
|
|
11612 if (*s == NUL
|
|
|
11613 /* *s == '^' cuts */
|
|
|
11614 || (*s == '$'
|
|
376
|
11615 && !spell_iswordp_w(word + i + k0,
|
|
|
11616 curbuf)))
|
|
372
|
11617 {
|
|
|
11618 if (k0 == k)
|
|
|
11619 /* this is just a piece of the string */
|
|
|
11620 continue;
|
|
|
11621
|
|
|
11622 if (p0 < pri)
|
|
|
11623 /* priority too low */
|
|
|
11624 continue;
|
|
|
11625 /* rule fits; stop search */
|
|
|
11626 break;
|
|
|
11627 }
|
|
|
11628 }
|
|
|
11629
|
|
|
11630 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
11631 == (c0 & 0xff))
|
|
|
11632 continue;
|
|
|
11633 }
|
|
|
11634
|
|
|
11635 /* replace string */
|
|
|
11636 ws = smp[n].sm_to_w;
|
|
|
11637 s = smp[n].sm_rules;
|
|
|
11638 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
11639 if (p0 == 1 && z == 0)
|
|
|
11640 {
|
|
|
11641 /* rule with '<' is used */
|
|
389
|
11642 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
11643 && (wres[reslen - 1] == c
|
|
372
|
11644 || wres[reslen - 1] == *ws))
|
|
|
11645 reslen--;
|
|
|
11646 z0 = 1;
|
|
|
11647 z = 1;
|
|
|
11648 k0 = 0;
|
|
389
|
11649 if (ws != NULL)
|
|
|
11650 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
11651 {
|
|
|
11652 word[i + k0] = *ws;
|
|
|
11653 k0++;
|
|
|
11654 ws++;
|
|
|
11655 }
|
|
372
|
11656 if (k > k0)
|
|
|
11657 mch_memmove(word + i + k0, word + i + k,
|
|
|
11658 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
11659
|
|
|
11660 /* new "actual letter" */
|
|
|
11661 c = word[i];
|
|
|
11662 }
|
|
|
11663 else
|
|
|
11664 {
|
|
|
11665 /* no '<' rule used */
|
|
|
11666 i += k - 1;
|
|
|
11667 z = 0;
|
|
389
|
11668 if (ws != NULL)
|
|
|
11669 while (*ws != NUL && ws[1] != NUL
|
|
|
11670 && reslen < MAXWLEN)
|
|
|
11671 {
|
|
|
11672 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
11673 wres[reslen++] = *ws;
|
|
|
11674 ws++;
|
|
|
11675 }
|
|
372
|
11676 /* new "actual letter" */
|
|
389
|
11677 if (ws == NULL)
|
|
|
11678 c = NUL;
|
|
|
11679 else
|
|
|
11680 c = *ws;
|
|
372
|
11681 if (strstr((char *)s, "^^") != NULL)
|
|
|
11682 {
|
|
|
11683 if (c != NUL)
|
|
|
11684 wres[reslen++] = c;
|
|
|
11685 mch_memmove(word, word + i + 1,
|
|
|
11686 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
11687 i = 0;
|
|
|
11688 z0 = 1;
|
|
|
11689 }
|
|
|
11690 }
|
|
|
11691 break;
|
|
|
11692 }
|
|
|
11693 }
|
|
|
11694 }
|
|
|
11695 else if (vim_iswhite(c))
|
|
|
11696 {
|
|
|
11697 c = ' ';
|
|
|
11698 k = 1;
|
|
|
11699 }
|
|
|
11700
|
|
|
11701 if (z0 == 0)
|
|
|
11702 {
|
|
|
11703 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
11704 && (!slang->sl_collapse || reslen == 0
|
|
|
11705 || wres[reslen - 1] != c))
|
|
|
11706 /* condense only double letters */
|
|
|
11707 wres[reslen++] = c;
|
|
|
11708
|
|
|
11709 i++;
|
|
|
11710 z = 0;
|
|
|
11711 k = 0;
|
|
|
11712 }
|
|
|
11713 }
|
|
|
11714
|
|
|
11715 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
11716 l = 0;
|
|
|
11717 for (n = 0; n < reslen; ++n)
|
|
|
11718 {
|
|
|
11719 l += mb_char2bytes(wres[n], res + l);
|
|
|
11720 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
11721 break;
|
|
|
11722 }
|
|
|
11723 res[l] = NUL;
|
|
|
11724 }
|
|
|
11725 #endif
|
|
|
11726
|
|
324
|
11727 /*
|
|
344
|
11728 * Compute a score for two sound-a-like words.
|
|
|
11729 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
11730 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
11731 * avoiding checks that will not be possible.
|
|
|
11732 */
|
|
|
11733 static int
|
|
351
|
11734 soundalike_score(goodstart, badstart)
|
|
|
11735 char_u *goodstart; /* sound-folded good word */
|
|
|
11736 char_u *badstart; /* sound-folded bad word */
|
|
344
|
11737 {
|
|
351
|
11738 char_u *goodsound = goodstart;
|
|
|
11739 char_u *badsound = badstart;
|
|
|
11740 int goodlen;
|
|
|
11741 int badlen;
|
|
344
|
11742 int n;
|
|
|
11743 char_u *pl, *ps;
|
|
|
11744 char_u *pl2, *ps2;
|
|
351
|
11745 int score = 0;
|
|
|
11746
|
|
|
11747 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
11748 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
11749 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
11750 {
|
|
|
11751 score = SCORE_DEL / 2;
|
|
|
11752 if (*badsound == '*')
|
|
|
11753 ++badsound;
|
|
|
11754 else
|
|
|
11755 ++goodsound;
|
|
|
11756 }
|
|
|
11757
|
|
|
11758 goodlen = STRLEN(goodsound);
|
|
|
11759 badlen = STRLEN(badsound);
|
|
344
|
11760
|
|
|
11761 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
11762 * changes. */
|
|
|
11763 n = goodlen - badlen;
|
|
|
11764 if (n < -2 || n > 2)
|
|
|
11765 return SCORE_MAXMAX;
|
|
|
11766
|
|
|
11767 if (n > 0)
|
|
|
11768 {
|
|
351
|
11769 pl = goodsound; /* goodsound is longest */
|
|
344
|
11770 ps = badsound;
|
|
|
11771 }
|
|
|
11772 else
|
|
|
11773 {
|
|
351
|
11774 pl = badsound; /* badsound is longest */
|
|
344
|
11775 ps = goodsound;
|
|
|
11776 }
|
|
|
11777
|
|
|
11778 /* Skip over the identical part. */
|
|
|
11779 while (*pl == *ps && *pl != NUL)
|
|
|
11780 {
|
|
|
11781 ++pl;
|
|
|
11782 ++ps;
|
|
|
11783 }
|
|
|
11784
|
|
|
11785 switch (n)
|
|
|
11786 {
|
|
|
11787 case -2:
|
|
|
11788 case 2:
|
|
|
11789 /*
|
|
|
11790 * Must delete two characters from "pl".
|
|
|
11791 */
|
|
|
11792 ++pl; /* first delete */
|
|
|
11793 while (*pl == *ps)
|
|
|
11794 {
|
|
|
11795 ++pl;
|
|
|
11796 ++ps;
|
|
|
11797 }
|
|
|
11798 /* strings must be equal after second delete */
|
|
|
11799 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
11800 return score + SCORE_DEL * 2;
|
|
344
|
11801
|
|
|
11802 /* Failed to compare. */
|
|
|
11803 break;
|
|
|
11804
|
|
|
11805 case -1:
|
|
|
11806 case 1:
|
|
|
11807 /*
|
|
|
11808 * Minimal one delete from "pl" required.
|
|
|
11809 */
|
|
|
11810
|
|
|
11811 /* 1: delete */
|
|
|
11812 pl2 = pl + 1;
|
|
|
11813 ps2 = ps;
|
|
|
11814 while (*pl2 == *ps2)
|
|
|
11815 {
|
|
|
11816 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11817 return score + SCORE_DEL;
|
|
344
|
11818 ++pl2;
|
|
|
11819 ++ps2;
|
|
|
11820 }
|
|
|
11821
|
|
|
11822 /* 2: delete then swap, then rest must be equal */
|
|
|
11823 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11824 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11825 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
11826
|
|
|
11827 /* 3: delete then substitute, then the rest must be equal */
|
|
|
11828 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11829 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
11830
|
|
|
11831 /* 4: first swap then delete */
|
|
|
11832 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
11833 {
|
|
|
11834 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
11835 ps2 = ps + 2;
|
|
|
11836 while (*pl2 == *ps2)
|
|
|
11837 {
|
|
|
11838 ++pl2;
|
|
|
11839 ++ps2;
|
|
|
11840 }
|
|
|
11841 /* delete a char and then strings must be equal */
|
|
|
11842 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11843 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
11844 }
|
|
|
11845
|
|
|
11846 /* 5: first substitute then delete */
|
|
|
11847 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
11848 ps2 = ps + 1;
|
|
|
11849 while (*pl2 == *ps2)
|
|
|
11850 {
|
|
|
11851 ++pl2;
|
|
|
11852 ++ps2;
|
|
|
11853 }
|
|
|
11854 /* delete a char and then strings must be equal */
|
|
|
11855 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11856 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
11857
|
|
|
11858 /* Failed to compare. */
|
|
|
11859 break;
|
|
|
11860
|
|
|
11861 case 0:
|
|
|
11862 /*
|
|
|
11863 * Lenghts are equal, thus changes must result in same length: An
|
|
|
11864 * insert is only possible in combination with a delete.
|
|
|
11865 * 1: check if for identical strings
|
|
|
11866 */
|
|
|
11867 if (*pl == NUL)
|
|
351
|
11868 return score;
|
|
344
|
11869
|
|
|
11870 /* 2: swap */
|
|
|
11871 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
11872 {
|
|
|
11873 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
11874 ps2 = ps + 2;
|
|
|
11875 while (*pl2 == *ps2)
|
|
|
11876 {
|
|
|
11877 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11878 return score + SCORE_SWAP;
|
|
344
|
11879 ++pl2;
|
|
|
11880 ++ps2;
|
|
|
11881 }
|
|
|
11882 /* 3: swap and swap again */
|
|
|
11883 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11884 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11885 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
11886
|
|
|
11887 /* 4: swap and substitute */
|
|
|
11888 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11889 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
11890 }
|
|
|
11891
|
|
|
11892 /* 5: substitute */
|
|
|
11893 pl2 = pl + 1;
|
|
|
11894 ps2 = ps + 1;
|
|
|
11895 while (*pl2 == *ps2)
|
|
|
11896 {
|
|
|
11897 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11898 return score + SCORE_SUBST;
|
|
344
|
11899 ++pl2;
|
|
|
11900 ++ps2;
|
|
|
11901 }
|
|
|
11902
|
|
|
11903 /* 6: substitute and swap */
|
|
|
11904 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11905 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11906 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
11907
|
|
|
11908 /* 7: substitute and substitute */
|
|
|
11909 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11910 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
11911
|
|
|
11912 /* 8: insert then delete */
|
|
|
11913 pl2 = pl;
|
|
|
11914 ps2 = ps + 1;
|
|
|
11915 while (*pl2 == *ps2)
|
|
|
11916 {
|
|
|
11917 ++pl2;
|
|
|
11918 ++ps2;
|
|
|
11919 }
|
|
|
11920 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11921 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
11922
|
|
|
11923 /* 9: delete then insert */
|
|
|
11924 pl2 = pl + 1;
|
|
|
11925 ps2 = ps;
|
|
|
11926 while (*pl2 == *ps2)
|
|
|
11927 {
|
|
|
11928 ++pl2;
|
|
|
11929 ++ps2;
|
|
|
11930 }
|
|
|
11931 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
11932 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
11933
|
|
|
11934 /* Failed to compare. */
|
|
|
11935 break;
|
|
|
11936 }
|
|
|
11937
|
|
|
11938 return SCORE_MAXMAX;
|
|
|
11939 }
|
|
324
|
11940
|
|
323
|
11941 /*
|
|
|
11942 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
11943 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
11944 *
|
|
485
|
11945 * The algorithm is described by Du and Chang, 1992.
|
|
|
11946 * The implementation of the algorithm comes from Aspell editdist.cpp,
|
|
|
11947 * edit_distance(). It has been converted from C++ to C and modified to
|
|
|
11948 * support multi-byte characters.
|
|
323
|
11949 */
|
|
|
11950 static int
|
|
|
11951 spell_edit_score(badword, goodword)
|
|
|
11952 char_u *badword;
|
|
|
11953 char_u *goodword;
|
|
|
11954 {
|
|
|
11955 int *cnt;
|
|
372
|
11956 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
11957 int j, i;
|
|
|
11958 int t;
|
|
|
11959 int bc, gc;
|
|
324
|
11960 int pbc, pgc;
|
|
|
11961 #ifdef FEAT_MBYTE
|
|
|
11962 char_u *p;
|
|
|
11963 int wbadword[MAXWLEN];
|
|
|
11964 int wgoodword[MAXWLEN];
|
|
|
11965
|
|
|
11966 if (has_mbyte)
|
|
|
11967 {
|
|
|
11968 /* Get the characters from the multi-byte strings and put them in an
|
|
|
11969 * int array for easy access. */
|
|
|
11970 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
11971 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
11972 wbadword[badlen++] = 0;
|
|
324
|
11973 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
11974 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
11975 wgoodword[goodlen++] = 0;
|
|
324
|
11976 }
|
|
|
11977 else
|
|
|
11978 #endif
|
|
|
11979 {
|
|
|
11980 badlen = STRLEN(badword) + 1;
|
|
|
11981 goodlen = STRLEN(goodword) + 1;
|
|
|
11982 }
|
|
323
|
11983
|
|
|
11984 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
11985 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
11986 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
11987 TRUE);
|
|
324
|
11988 if (cnt == NULL)
|
|
|
11989 return 0; /* out of memory */
|
|
323
|
11990
|
|
|
11991 CNT(0, 0) = 0;
|
|
|
11992 for (j = 1; j <= goodlen; ++j)
|
|
|
11993 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
11994
|
|
|
11995 for (i = 1; i <= badlen; ++i)
|
|
|
11996 {
|
|
|
11997 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
11998 for (j = 1; j <= goodlen; ++j)
|
|
|
11999 {
|
|
324
|
12000 #ifdef FEAT_MBYTE
|
|
|
12001 if (has_mbyte)
|
|
|
12002 {
|
|
|
12003 bc = wbadword[i - 1];
|
|
|
12004 gc = wgoodword[j - 1];
|
|
|
12005 }
|
|
|
12006 else
|
|
|
12007 #endif
|
|
|
12008 {
|
|
|
12009 bc = badword[i - 1];
|
|
|
12010 gc = goodword[j - 1];
|
|
|
12011 }
|
|
323
|
12012 if (bc == gc)
|
|
|
12013 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
12014 else
|
|
|
12015 {
|
|
|
12016 /* Use a better score when there is only a case difference. */
|
|
324
|
12017 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
12018 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
12019 else
|
|
|
12020 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
12021
|
|
324
|
12022 if (i > 1 && j > 1)
|
|
323
|
12023 {
|
|
324
|
12024 #ifdef FEAT_MBYTE
|
|
|
12025 if (has_mbyte)
|
|
|
12026 {
|
|
|
12027 pbc = wbadword[i - 2];
|
|
|
12028 pgc = wgoodword[j - 2];
|
|
|
12029 }
|
|
|
12030 else
|
|
|
12031 #endif
|
|
|
12032 {
|
|
|
12033 pbc = badword[i - 2];
|
|
|
12034 pgc = goodword[j - 2];
|
|
|
12035 }
|
|
|
12036 if (bc == pgc && pbc == gc)
|
|
|
12037 {
|
|
|
12038 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
12039 if (t < CNT(i, j))
|
|
|
12040 CNT(i, j) = t;
|
|
|
12041 }
|
|
323
|
12042 }
|
|
|
12043 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
12044 if (t < CNT(i, j))
|
|
|
12045 CNT(i, j) = t;
|
|
|
12046 t = SCORE_INS + CNT(i, j - 1);
|
|
|
12047 if (t < CNT(i, j))
|
|
|
12048 CNT(i, j) = t;
|
|
|
12049 }
|
|
|
12050 }
|
|
|
12051 }
|
|
344
|
12052
|
|
|
12053 i = CNT(badlen - 1, goodlen - 1);
|
|
|
12054 vim_free(cnt);
|
|
|
12055 return i;
|
|
323
|
12056 }
|
|
307
|
12057
|
|
351
|
12058 /*
|
|
|
12059 * ":spelldump"
|
|
|
12060 */
|
|
|
12061 /*ARGSUSED*/
|
|
|
12062 void
|
|
|
12063 ex_spelldump(eap)
|
|
|
12064 exarg_T *eap;
|
|
|
12065 {
|
|
|
12066 buf_T *buf = curbuf;
|
|
|
12067 langp_T *lp;
|
|
|
12068 slang_T *slang;
|
|
|
12069 idx_T arridx[MAXWLEN];
|
|
|
12070 int curi[MAXWLEN];
|
|
|
12071 char_u word[MAXWLEN];
|
|
|
12072 int c;
|
|
|
12073 char_u *byts;
|
|
|
12074 idx_T *idxs;
|
|
|
12075 linenr_T lnum = 0;
|
|
|
12076 int round;
|
|
|
12077 int depth;
|
|
|
12078 int n;
|
|
|
12079 int flags;
|
|
381
|
12080 char_u *region_names = NULL; /* region names being used */
|
|
|
12081 int do_region = TRUE; /* dump region names and numbers */
|
|
|
12082 char_u *p;
|
|
351
|
12083
|
|
|
12084 if (no_spell_checking())
|
|
|
12085 return;
|
|
|
12086
|
|
|
12087 /* Create a new empty buffer by splitting the window. */
|
|
|
12088 do_cmdline_cmd((char_u *)"new");
|
|
|
12089 if (!bufempty() || !buf_valid(buf))
|
|
|
12090 return;
|
|
|
12091
|
|
381
|
12092 /* Find out if we can support regions: All languages must support the same
|
|
|
12093 * regions or none at all. */
|
|
|
12094 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
12095 {
|
|
|
12096 p = lp->lp_slang->sl_regions;
|
|
|
12097 if (p[0] != 0)
|
|
|
12098 {
|
|
|
12099 if (region_names == NULL) /* first language with regions */
|
|
|
12100 region_names = p;
|
|
|
12101 else if (STRCMP(region_names, p) != 0)
|
|
|
12102 {
|
|
|
12103 do_region = FALSE; /* region names are different */
|
|
|
12104 break;
|
|
|
12105 }
|
|
|
12106 }
|
|
|
12107 }
|
|
|
12108
|
|
|
12109 if (do_region && region_names != NULL)
|
|
|
12110 {
|
|
|
12111 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
12112 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
12113 }
|
|
|
12114 else
|
|
|
12115 do_region = FALSE;
|
|
|
12116
|
|
|
12117 /*
|
|
|
12118 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
12119 */
|
|
351
|
12120 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
12121 {
|
|
|
12122 slang = lp->lp_slang;
|
|
|
12123
|
|
|
12124 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
12125 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
12126
|
|
|
12127 /* round 1: case-folded tree
|
|
|
12128 * round 2: keep-case tree */
|
|
|
12129 for (round = 1; round <= 2; ++round)
|
|
|
12130 {
|
|
|
12131 if (round == 1)
|
|
|
12132 {
|
|
|
12133 byts = slang->sl_fbyts;
|
|
|
12134 idxs = slang->sl_fidxs;
|
|
|
12135 }
|
|
|
12136 else
|
|
|
12137 {
|
|
|
12138 byts = slang->sl_kbyts;
|
|
|
12139 idxs = slang->sl_kidxs;
|
|
|
12140 }
|
|
|
12141 if (byts == NULL)
|
|
|
12142 continue; /* array is empty */
|
|
|
12143
|
|
|
12144 depth = 0;
|
|
|
12145 arridx[0] = 0;
|
|
|
12146 curi[0] = 1;
|
|
|
12147 while (depth >= 0 && !got_int)
|
|
|
12148 {
|
|
|
12149 if (curi[depth] > byts[arridx[depth]])
|
|
|
12150 {
|
|
|
12151 /* Done all bytes at this node, go up one level. */
|
|
|
12152 --depth;
|
|
|
12153 line_breakcheck();
|
|
|
12154 }
|
|
|
12155 else
|
|
|
12156 {
|
|
|
12157 /* Do one more byte at this node. */
|
|
|
12158 n = arridx[depth] + curi[depth];
|
|
|
12159 ++curi[depth];
|
|
|
12160 c = byts[n];
|
|
|
12161 if (c == 0)
|
|
|
12162 {
|
|
|
12163 /* End of word, deal with the word.
|
|
|
12164 * Don't use keep-case words in the fold-case tree,
|
|
|
12165 * they will appear in the keep-case tree.
|
|
|
12166 * Only use the word when the region matches. */
|
|
|
12167 flags = (int)idxs[n];
|
|
|
12168 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
12169 && (do_region
|
|
|
12170 || (flags & WF_REGION) == 0
|
|
390
|
12171 || (((unsigned)flags >> 16)
|
|
351
|
12172 & lp->lp_region) != 0))
|
|
|
12173 {
|
|
|
12174 word[depth] = NUL;
|
|
381
|
12175 if (!do_region)
|
|
|
12176 flags &= ~WF_REGION;
|
|
355
|
12177
|
|
|
12178 /* Dump the basic word if there is no prefix or
|
|
|
12179 * when it's the first one. */
|
|
390
|
12180 c = (unsigned)flags >> 24;
|
|
355
|
12181 if (c == 0 || curi[depth] == 2)
|
|
|
12182 dump_word(word, round, flags, lnum++);
|
|
351
|
12183
|
|
|
12184 /* Apply the prefix, if there is one. */
|
|
355
|
12185 if (c != 0)
|
|
351
|
12186 lnum = apply_prefixes(slang, word, round,
|
|
|
12187 flags, lnum);
|
|
|
12188 }
|
|
|
12189 }
|
|
|
12190 else
|
|
|
12191 {
|
|
|
12192 /* Normal char, go one level deeper. */
|
|
|
12193 word[depth++] = c;
|
|
|
12194 arridx[depth] = idxs[n];
|
|
|
12195 curi[depth] = 1;
|
|
|
12196 }
|
|
|
12197 }
|
|
|
12198 }
|
|
|
12199 }
|
|
|
12200 }
|
|
|
12201
|
|
|
12202 /* Delete the empty line that we started with. */
|
|
|
12203 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
12204 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
12205
|
|
|
12206 redraw_later(NOT_VALID);
|
|
|
12207 }
|
|
|
12208
|
|
|
12209 /*
|
|
|
12210 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
12211 */
|
|
|
12212 static void
|
|
|
12213 dump_word(word, round, flags, lnum)
|
|
|
12214 char_u *word;
|
|
|
12215 int round;
|
|
|
12216 int flags;
|
|
|
12217 linenr_T lnum;
|
|
|
12218 {
|
|
|
12219 int keepcap = FALSE;
|
|
|
12220 char_u *p;
|
|
|
12221 char_u cword[MAXWLEN];
|
|
381
|
12222 char_u badword[MAXWLEN + 10];
|
|
|
12223 int i;
|
|
351
|
12224
|
|
|
12225 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
12226 {
|
|
|
12227 /* Need to fix case according to "flags". */
|
|
|
12228 make_case_word(word, cword, flags);
|
|
|
12229 p = cword;
|
|
|
12230 }
|
|
|
12231 else
|
|
|
12232 {
|
|
|
12233 p = word;
|
|
389
|
12234 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
12235 || (flags & WF_FIXCAP) != 0))
|
|
351
|
12236 keepcap = TRUE;
|
|
|
12237 }
|
|
|
12238
|
|
381
|
12239 /* Add flags and regions after a slash. */
|
|
|
12240 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
12241 {
|
|
|
12242 STRCPY(badword, p);
|
|
|
12243 STRCAT(badword, "/");
|
|
351
|
12244 if (keepcap)
|
|
|
12245 STRCAT(badword, "=");
|
|
|
12246 if (flags & WF_BANNED)
|
|
|
12247 STRCAT(badword, "!");
|
|
|
12248 else if (flags & WF_RARE)
|
|
|
12249 STRCAT(badword, "?");
|
|
381
|
12250 if (flags & WF_REGION)
|
|
|
12251 for (i = 0; i < 7; ++i)
|
|
390
|
12252 if (flags & (0x10000 << i))
|
|
381
|
12253 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
12254 p = badword;
|
|
|
12255 }
|
|
|
12256
|
|
|
12257 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
12258 }
|
|
|
12259
|
|
|
12260 /*
|
|
372
|
12261 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
12262 * "word" and append a line to the buffer.
|
|
351
|
12263 * Return the updated line number.
|
|
|
12264 */
|
|
|
12265 static linenr_T
|
|
|
12266 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
12267 slang_T *slang;
|
|
|
12268 char_u *word; /* case-folded word */
|
|
|
12269 int round;
|
|
|
12270 int flags; /* flags with prefix ID */
|
|
|
12271 linenr_T startlnum;
|
|
|
12272 {
|
|
|
12273 idx_T arridx[MAXWLEN];
|
|
|
12274 int curi[MAXWLEN];
|
|
|
12275 char_u prefix[MAXWLEN];
|
|
455
|
12276 char_u word_up[MAXWLEN];
|
|
|
12277 int has_word_up = FALSE;
|
|
351
|
12278 int c;
|
|
|
12279 char_u *byts;
|
|
|
12280 idx_T *idxs;
|
|
|
12281 linenr_T lnum = startlnum;
|
|
|
12282 int depth;
|
|
|
12283 int n;
|
|
|
12284 int len;
|
|
|
12285 int i;
|
|
|
12286
|
|
455
|
12287 /* if the word starts with a lower-case letter make the word with an
|
|
|
12288 * upper-case letter in word_up[]. */
|
|
|
12289 c = PTR2CHAR(word);
|
|
|
12290 if (SPELL_TOUPPER(c) != c)
|
|
|
12291 {
|
|
|
12292 onecap_copy(word, word_up, TRUE);
|
|
|
12293 has_word_up = TRUE;
|
|
|
12294 }
|
|
|
12295
|
|
351
|
12296 byts = slang->sl_pbyts;
|
|
|
12297 idxs = slang->sl_pidxs;
|
|
|
12298 if (byts != NULL) /* array not is empty */
|
|
|
12299 {
|
|
|
12300 /*
|
|
|
12301 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
12302 * When at the end of a prefix check that it supports "flags".
|
|
351
|
12303 */
|
|
|
12304 depth = 0;
|
|
|
12305 arridx[0] = 0;
|
|
|
12306 curi[0] = 1;
|
|
|
12307 while (depth >= 0 && !got_int)
|
|
|
12308 {
|
|
390
|
12309 n = arridx[depth];
|
|
|
12310 len = byts[n];
|
|
|
12311 if (curi[depth] > len)
|
|
351
|
12312 {
|
|
|
12313 /* Done all bytes at this node, go up one level. */
|
|
|
12314 --depth;
|
|
|
12315 line_breakcheck();
|
|
|
12316 }
|
|
|
12317 else
|
|
|
12318 {
|
|
|
12319 /* Do one more byte at this node. */
|
|
390
|
12320 n += curi[depth];
|
|
351
|
12321 ++curi[depth];
|
|
|
12322 c = byts[n];
|
|
|
12323 if (c == 0)
|
|
|
12324 {
|
|
|
12325 /* End of prefix, find out how many IDs there are. */
|
|
|
12326 for (i = 1; i < len; ++i)
|
|
|
12327 if (byts[n + i] != 0)
|
|
|
12328 break;
|
|
|
12329 curi[depth] += i - 1;
|
|
|
12330
|
|
455
|
12331 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
12332 if (c != 0)
|
|
351
|
12333 {
|
|
376
|
12334 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
12335 dump_word(prefix, round,
|
|
455
|
12336 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
12337 : flags, lnum++);
|
|
351
|
12338 }
|
|
455
|
12339
|
|
|
12340 /* Check for prefix that matches the word when the
|
|
|
12341 * first letter is upper-case, but only if the prefix has
|
|
|
12342 * a condition. */
|
|
|
12343 if (has_word_up)
|
|
|
12344 {
|
|
|
12345 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
12346 TRUE);
|
|
|
12347 if (c != 0)
|
|
|
12348 {
|
|
|
12349 vim_strncpy(prefix + depth, word_up,
|
|
|
12350 MAXWLEN - depth - 1);
|
|
|
12351 dump_word(prefix, round,
|
|
|
12352 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
12353 : flags, lnum++);
|
|
|
12354 }
|
|
|
12355 }
|
|
351
|
12356 }
|
|
|
12357 else
|
|
|
12358 {
|
|
|
12359 /* Normal char, go one level deeper. */
|
|
|
12360 prefix[depth++] = c;
|
|
|
12361 arridx[depth] = idxs[n];
|
|
|
12362 curi[depth] = 1;
|
|
|
12363 }
|
|
|
12364 }
|
|
|
12365 }
|
|
|
12366 }
|
|
|
12367
|
|
|
12368 return lnum;
|
|
|
12369 }
|
|
|
12370
|
|
475
|
12371 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
12372 static int spell_expand_need_cap;
|
|
|
12373
|
|
|
12374 /*
|
|
|
12375 * Find start of the word in front of the cursor. We don't check if it is
|
|
|
12376 * badly spelled, with completion we can only change the word in front of the
|
|
|
12377 * cursor.
|
|
|
12378 * Used for Insert mode completion CTRL-X ?.
|
|
|
12379 * Returns the column number of the word.
|
|
|
12380 */
|
|
|
12381 int
|
|
|
12382 spell_word_start(startcol)
|
|
|
12383 int startcol;
|
|
|
12384 {
|
|
|
12385 char_u *line;
|
|
|
12386 char_u *p;
|
|
|
12387 int col = 0;
|
|
|
12388
|
|
|
12389 if (no_spell_checking())
|
|
|
12390 return startcol;
|
|
|
12391
|
|
|
12392 /* Find a word character before "startcol". */
|
|
|
12393 line = ml_get_curline();
|
|
|
12394 for (p = line + startcol; p > line; )
|
|
|
12395 {
|
|
|
12396 mb_ptr_back(line, p);
|
|
|
12397 if (spell_iswordp_nmw(p))
|
|
|
12398 break;
|
|
|
12399 }
|
|
|
12400
|
|
|
12401 /* Go back to start of the word. */
|
|
|
12402 while (p > line)
|
|
|
12403 {
|
|
|
12404 col = p - line;
|
|
|
12405 mb_ptr_back(line, p);
|
|
|
12406 if (!spell_iswordp(p, curbuf))
|
|
|
12407 break;
|
|
|
12408 col = 0;
|
|
|
12409 }
|
|
|
12410
|
|
|
12411 /* Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
12412 * expand_spelling() is called. Therefore the ugly global variable. */
|
|
|
12413 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
12414
|
|
|
12415 return col;
|
|
|
12416 }
|
|
|
12417
|
|
|
12418 /*
|
|
|
12419 * Get list of spelling suggestions.
|
|
|
12420 * Used for Insert mode completion CTRL-X ?.
|
|
|
12421 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
12422 * allocated strings.
|
|
|
12423 */
|
|
|
12424 /*ARGSUSED*/
|
|
|
12425 int
|
|
|
12426 expand_spelling(lnum, col, pat, matchp)
|
|
|
12427 linenr_T lnum;
|
|
|
12428 int col;
|
|
|
12429 char_u *pat;
|
|
|
12430 char_u ***matchp;
|
|
|
12431 {
|
|
|
12432 garray_T ga;
|
|
|
12433
|
|
|
12434 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap);
|
|
|
12435 *matchp = ga.ga_data;
|
|
|
12436 return ga.ga_len;
|
|
|
12437 }
|
|
|
12438 #endif
|
|
|
12439
|
|
236
|
12440 #endif /* FEAT_SYN_HL */
|
