|
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));
|
|
498
|
685 static int no_spell_checking __ARGS((win_T *wp));
|
|
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
|
|
498
|
1752 no_spell_checking(wp)
|
|
|
1753 win_T *wp;
|
|
|
1754 {
|
|
|
1755 if (!wp->w_p_spell || *wp->w_buffer->b_p_spl == NUL)
|
|
351
|
1756 {
|
|
|
1757 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1758 return TRUE;
|
|
|
1759 }
|
|
|
1760 return FALSE;
|
|
|
1761 }
|
|
300
|
1762
|
|
236
|
1763 /*
|
|
|
1764 * Move to next spell error.
|
|
323
|
1765 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
483
|
1766 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move
|
|
|
1767 * to after badly spelled word before the cursor.
|
|
495
|
1768 * Return 0 if not found, length of the badly spelled word otherwise.
|
|
236
|
1769 */
|
|
|
1770 int
|
|
498
|
1771 spell_move_to(wp, dir, allwords, curline, attrp)
|
|
|
1772 win_T *wp;
|
|
236
|
1773 int dir; /* FORWARD or BACKWARD */
|
|
|
1774 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1775 int curline;
|
|
498
|
1776 int *attrp; /* return: attributes of bad word or NULL */
|
|
236
|
1777 {
|
|
249
|
1778 linenr_T lnum;
|
|
|
1779 pos_T found_pos;
|
|
495
|
1780 int found_len = 0;
|
|
236
|
1781 char_u *line;
|
|
|
1782 char_u *p;
|
|
346
|
1783 char_u *endp;
|
|
|
1784 int attr;
|
|
236
|
1785 int len;
|
|
498
|
1786 int has_syntax = syntax_present(wp->w_buffer);
|
|
249
|
1787 int col;
|
|
|
1788 int can_spell;
|
|
346
|
1789 char_u *buf = NULL;
|
|
|
1790 int buflen = 0;
|
|
|
1791 int skip = 0;
|
|
385
|
1792 int capcol = -1;
|
|
236
|
1793
|
|
498
|
1794 if (no_spell_checking(wp))
|
|
495
|
1795 return 0;
|
|
236
|
1796
|
|
249
|
1797 /*
|
|
|
1798 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1799 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1800 *
|
|
|
1801 * When searching backwards, we continue in the line to find the last
|
|
|
1802 * bad word (in the cursor line: before the cursor).
|
|
346
|
1803 *
|
|
|
1804 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1805 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1806 * though...
|
|
249
|
1807 */
|
|
498
|
1808 lnum = wp->w_cursor.lnum;
|
|
249
|
1809 found_pos.lnum = 0;
|
|
236
|
1810
|
|
|
1811 while (!got_int)
|
|
|
1812 {
|
|
498
|
1813 line = ml_get_buf(wp->w_buffer, lnum, FALSE);
|
|
346
|
1814
|
|
|
1815 len = STRLEN(line);
|
|
|
1816 if (buflen < len + MAXWLEN + 2)
|
|
|
1817 {
|
|
|
1818 vim_free(buf);
|
|
|
1819 buflen = len + MAXWLEN + 2;
|
|
|
1820 buf = alloc(buflen);
|
|
|
1821 if (buf == NULL)
|
|
|
1822 break;
|
|
|
1823 }
|
|
|
1824
|
|
385
|
1825 /* In first line check first word for Capital. */
|
|
|
1826 if (lnum == 1)
|
|
|
1827 capcol = 0;
|
|
|
1828
|
|
|
1829 /* For checking first word with a capital skip white space. */
|
|
|
1830 if (capcol == 0)
|
|
|
1831 capcol = skipwhite(line) - line;
|
|
|
1832
|
|
346
|
1833 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1834 * possible. */
|
|
|
1835 STRCPY(buf, line);
|
|
498
|
1836 if (lnum < wp->w_buffer->b_ml.ml_line_count)
|
|
346
|
1837 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1838
|
|
|
1839 p = buf + skip;
|
|
|
1840 endp = buf + len;
|
|
|
1841 while (p < endp)
|
|
236
|
1842 {
|
|
300
|
1843 /* When searching backward don't search after the cursor. */
|
|
|
1844 if (dir == BACKWARD
|
|
498
|
1845 && lnum == wp->w_cursor.lnum
|
|
|
1846 && (colnr_T)(p - buf) >= wp->w_cursor.col)
|
|
300
|
1847 break;
|
|
249
|
1848
|
|
300
|
1849 /* start of word */
|
|
346
|
1850 attr = 0;
|
|
498
|
1851 len = spell_check(wp, p, &attr, &capcol);
|
|
249
|
1852
|
|
300
|
1853 if (attr != 0)
|
|
|
1854 {
|
|
|
1855 /* We found a bad word. Check the attribute. */
|
|
|
1856 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1857 {
|
|
300
|
1858 /* When searching forward only accept a bad word after
|
|
|
1859 * the cursor. */
|
|
|
1860 if (dir == BACKWARD
|
|
498
|
1861 || lnum > wp->w_cursor.lnum
|
|
|
1862 || (lnum == wp->w_cursor.lnum
|
|
346
|
1863 && (colnr_T)(curline ? p - buf + len
|
|
|
1864 : p - buf)
|
|
498
|
1865 > wp->w_cursor.col))
|
|
236
|
1866 {
|
|
300
|
1867 if (has_syntax)
|
|
249
|
1868 {
|
|
346
|
1869 col = p - buf;
|
|
498
|
1870 (void)syn_get_id(wp, lnum, (colnr_T)col,
|
|
300
|
1871 FALSE, &can_spell);
|
|
|
1872 }
|
|
|
1873 else
|
|
|
1874 can_spell = TRUE;
|
|
249
|
1875
|
|
300
|
1876 if (can_spell)
|
|
|
1877 {
|
|
|
1878 found_pos.lnum = lnum;
|
|
346
|
1879 found_pos.col = p - buf;
|
|
249
|
1880 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1881 found_pos.coladd = 0;
|
|
249
|
1882 #endif
|
|
300
|
1883 if (dir == FORWARD)
|
|
|
1884 {
|
|
|
1885 /* No need to search further. */
|
|
498
|
1886 wp->w_cursor = found_pos;
|
|
346
|
1887 vim_free(buf);
|
|
498
|
1888 if (attrp != NULL)
|
|
|
1889 *attrp = attr;
|
|
495
|
1890 return len;
|
|
249
|
1891 }
|
|
483
|
1892 else if (curline)
|
|
|
1893 /* Insert mode completion: put cursor after
|
|
|
1894 * the bad word. */
|
|
|
1895 found_pos.col += len;
|
|
495
|
1896 found_len = len;
|
|
249
|
1897 }
|
|
236
|
1898 }
|
|
|
1899 }
|
|
|
1900 }
|
|
|
1901
|
|
300
|
1902 /* advance to character after the word */
|
|
|
1903 p += len;
|
|
385
|
1904 capcol -= len;
|
|
236
|
1905 }
|
|
|
1906
|
|
483
|
1907 if (dir == BACKWARD && found_pos.lnum != 0)
|
|
|
1908 {
|
|
|
1909 /* Use the last match in the line. */
|
|
498
|
1910 wp->w_cursor = found_pos;
|
|
483
|
1911 vim_free(buf);
|
|
495
|
1912 return found_len;
|
|
483
|
1913 }
|
|
|
1914
|
|
323
|
1915 if (curline)
|
|
346
|
1916 break; /* only check cursor line */
|
|
323
|
1917
|
|
236
|
1918 /* Advance to next line. */
|
|
249
|
1919 if (dir == BACKWARD)
|
|
|
1920 {
|
|
|
1921 if (lnum == 1)
|
|
346
|
1922 break;
|
|
249
|
1923 --lnum;
|
|
385
|
1924 capcol = -1;
|
|
249
|
1925 }
|
|
|
1926 else
|
|
|
1927 {
|
|
498
|
1928 if (lnum == wp->w_buffer->b_ml.ml_line_count)
|
|
346
|
1929 break;
|
|
249
|
1930 ++lnum;
|
|
346
|
1931
|
|
|
1932 /* Skip the characters at the start of the next line that were
|
|
|
1933 * included in a match crossing line boundaries. */
|
|
|
1934 if (attr == 0)
|
|
|
1935 skip = p - endp;
|
|
|
1936 else
|
|
|
1937 skip = 0;
|
|
385
|
1938
|
|
|
1939 /* Capscol skips over the inserted space. */
|
|
|
1940 --capcol;
|
|
|
1941
|
|
|
1942 /* But after empty line check first word in next line */
|
|
|
1943 if (*skipwhite(line) == NUL)
|
|
|
1944 capcol = 0;
|
|
249
|
1945 }
|
|
236
|
1946
|
|
|
1947 line_breakcheck();
|
|
|
1948 }
|
|
|
1949
|
|
346
|
1950 vim_free(buf);
|
|
495
|
1951 return 0;
|
|
346
|
1952 }
|
|
|
1953
|
|
|
1954 /*
|
|
|
1955 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1956 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1957 */
|
|
|
1958 void
|
|
|
1959 spell_cat_line(buf, line, maxlen)
|
|
|
1960 char_u *buf;
|
|
|
1961 char_u *line;
|
|
|
1962 int maxlen;
|
|
|
1963 {
|
|
|
1964 char_u *p;
|
|
|
1965 int n;
|
|
|
1966
|
|
|
1967 p = skipwhite(line);
|
|
|
1968 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1969 p = skipwhite(p + 1);
|
|
|
1970
|
|
|
1971 if (*p != NUL)
|
|
|
1972 {
|
|
|
1973 *buf = ' ';
|
|
376
|
1974 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1975 n = p - line;
|
|
|
1976 if (n >= maxlen)
|
|
|
1977 n = maxlen - 1;
|
|
|
1978 vim_memset(buf + 1, ' ', n);
|
|
|
1979 }
|
|
236
|
1980 }
|
|
|
1981
|
|
|
1982 /*
|
|
307
|
1983 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1984 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1985 */
|
|
307
|
1986 static void
|
|
236
|
1987 spell_load_lang(lang)
|
|
|
1988 char_u *lang;
|
|
|
1989 {
|
|
310
|
1990 char_u fname_enc[85];
|
|
236
|
1991 int r;
|
|
307
|
1992 char_u langcp[MAXWLEN + 1];
|
|
|
1993
|
|
310
|
1994 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1995 * It's truncated when an error is detected. */
|
|
|
1996 STRCPY(langcp, lang);
|
|
|
1997
|
|
310
|
1998 /*
|
|
|
1999 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
2000 */
|
|
|
2001 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
2002 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
2003 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
2004
|
|
|
2005 if (r == FAIL && *langcp != NUL)
|
|
|
2006 {
|
|
|
2007 /* Try loading the ASCII version. */
|
|
310
|
2008 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
2009 "spell/%s.ascii.spl", lang);
|
|
310
|
2010 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
2011 }
|
|
|
2012
|
|
|
2013 if (r == FAIL)
|
|
483
|
2014 smsg((char_u *)_("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""),
|
|
|
2015 lang, spell_enc(), lang);
|
|
310
|
2016 else if (*langcp != NUL)
|
|
|
2017 {
|
|
|
2018 /* Load all the additions. */
|
|
|
2019 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
2020 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
2021 }
|
|
|
2022 }
|
|
|
2023
|
|
|
2024 /*
|
|
|
2025 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
2026 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
2027 */
|
|
|
2028 static char_u *
|
|
|
2029 spell_enc()
|
|
|
2030 {
|
|
|
2031
|
|
|
2032 #ifdef FEAT_MBYTE
|
|
|
2033 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
2034 return p_enc;
|
|
|
2035 #endif
|
|
|
2036 return (char_u *)"latin1";
|
|
236
|
2037 }
|
|
|
2038
|
|
|
2039 /*
|
|
385
|
2040 * Get the name of the .spl file for the internal wordlist into
|
|
|
2041 * "fname[MAXPATHL]".
|
|
|
2042 */
|
|
|
2043 static void
|
|
|
2044 int_wordlist_spl(fname)
|
|
|
2045 char_u *fname;
|
|
|
2046 {
|
|
|
2047 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
2048 int_wordlist, spell_enc());
|
|
|
2049 }
|
|
|
2050
|
|
|
2051 /*
|
|
236
|
2052 * Allocate a new slang_T.
|
|
|
2053 * Caller must fill "sl_next".
|
|
|
2054 */
|
|
|
2055 static slang_T *
|
|
|
2056 slang_alloc(lang)
|
|
|
2057 char_u *lang;
|
|
|
2058 {
|
|
|
2059 slang_T *lp;
|
|
|
2060
|
|
300
|
2061 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
2062 if (lp != NULL)
|
|
|
2063 {
|
|
|
2064 lp->sl_name = vim_strsave(lang);
|
|
323
|
2065 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
483
|
2066 lp->sl_compmax = MAXWLEN;
|
|
|
2067 lp->sl_compminlen = MAXWLEN;
|
|
|
2068 lp->sl_compsylmax = MAXWLEN;
|
|
236
|
2069 }
|
|
|
2070 return lp;
|
|
|
2071 }
|
|
|
2072
|
|
|
2073 /*
|
|
|
2074 * Free the contents of an slang_T and the structure itself.
|
|
|
2075 */
|
|
|
2076 static void
|
|
|
2077 slang_free(lp)
|
|
|
2078 slang_T *lp;
|
|
|
2079 {
|
|
|
2080 vim_free(lp->sl_name);
|
|
310
|
2081 vim_free(lp->sl_fname);
|
|
|
2082 slang_clear(lp);
|
|
|
2083 vim_free(lp);
|
|
|
2084 }
|
|
|
2085
|
|
|
2086 /*
|
|
|
2087 * Clear an slang_T so that the file can be reloaded.
|
|
|
2088 */
|
|
|
2089 static void
|
|
|
2090 slang_clear(lp)
|
|
|
2091 slang_T *lp;
|
|
|
2092 {
|
|
339
|
2093 garray_T *gap;
|
|
|
2094 fromto_T *ftp;
|
|
344
|
2095 salitem_T *smp;
|
|
339
|
2096 int i;
|
|
323
|
2097
|
|
300
|
2098 vim_free(lp->sl_fbyts);
|
|
310
|
2099 lp->sl_fbyts = NULL;
|
|
300
|
2100 vim_free(lp->sl_kbyts);
|
|
310
|
2101 lp->sl_kbyts = NULL;
|
|
339
|
2102 vim_free(lp->sl_pbyts);
|
|
|
2103 lp->sl_pbyts = NULL;
|
|
|
2104
|
|
300
|
2105 vim_free(lp->sl_fidxs);
|
|
310
|
2106 lp->sl_fidxs = NULL;
|
|
300
|
2107 vim_free(lp->sl_kidxs);
|
|
310
|
2108 lp->sl_kidxs = NULL;
|
|
339
|
2109 vim_free(lp->sl_pidxs);
|
|
|
2110 lp->sl_pidxs = NULL;
|
|
323
|
2111
|
|
344
|
2112 gap = &lp->sl_rep;
|
|
|
2113 while (gap->ga_len > 0)
|
|
323
|
2114 {
|
|
344
|
2115 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2116 vim_free(ftp->ft_from);
|
|
|
2117 vim_free(ftp->ft_to);
|
|
323
|
2118 }
|
|
344
|
2119 ga_clear(gap);
|
|
|
2120
|
|
|
2121 gap = &lp->sl_sal;
|
|
375
|
2122 if (lp->sl_sofo)
|
|
376
|
2123 {
|
|
|
2124 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
2125 if (gap->ga_data != NULL)
|
|
|
2126 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
2127 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2128 vim_free(((int **)gap->ga_data)[i]);
|
|
|
2129 }
|
|
375
|
2130 else
|
|
|
2131 /* SAL items: free salitem_T items */
|
|
|
2132 while (gap->ga_len > 0)
|
|
|
2133 {
|
|
|
2134 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2135 vim_free(smp->sm_lead);
|
|
|
2136 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
2137 vim_free(smp->sm_to);
|
|
|
2138 #ifdef FEAT_MBYTE
|
|
|
2139 vim_free(smp->sm_lead_w);
|
|
|
2140 vim_free(smp->sm_oneof_w);
|
|
|
2141 vim_free(smp->sm_to_w);
|
|
|
2142 #endif
|
|
|
2143 }
|
|
344
|
2144 ga_clear(gap);
|
|
323
|
2145
|
|
339
|
2146 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
2147 vim_free(lp->sl_prefprog[i]);
|
|
376
|
2148 lp->sl_prefixcnt = 0;
|
|
339
|
2149 vim_free(lp->sl_prefprog);
|
|
376
|
2150 lp->sl_prefprog = NULL;
|
|
|
2151
|
|
|
2152 vim_free(lp->sl_midword);
|
|
|
2153 lp->sl_midword = NULL;
|
|
339
|
2154
|
|
483
|
2155 vim_free(lp->sl_compprog);
|
|
|
2156 vim_free(lp->sl_compstartflags);
|
|
485
|
2157 vim_free(lp->sl_compallflags);
|
|
483
|
2158 lp->sl_compprog = NULL;
|
|
|
2159 lp->sl_compstartflags = NULL;
|
|
485
|
2160 lp->sl_compallflags = NULL;
|
|
483
|
2161
|
|
|
2162 vim_free(lp->sl_syllable);
|
|
|
2163 lp->sl_syllable = NULL;
|
|
|
2164 ga_clear(&lp->sl_syl_items);
|
|
481
|
2165
|
|
330
|
2166 #ifdef FEAT_MBYTE
|
|
|
2167 {
|
|
|
2168 int todo = lp->sl_map_hash.ht_used;
|
|
|
2169 hashitem_T *hi;
|
|
|
2170
|
|
|
2171 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
2172 if (!HASHITEM_EMPTY(hi))
|
|
|
2173 {
|
|
|
2174 --todo;
|
|
|
2175 vim_free(hi->hi_key);
|
|
|
2176 }
|
|
|
2177 }
|
|
|
2178 hash_clear(&lp->sl_map_hash);
|
|
|
2179 #endif
|
|
483
|
2180
|
|
|
2181 lp->sl_compmax = MAXWLEN;
|
|
|
2182 lp->sl_compminlen = MAXWLEN;
|
|
|
2183 lp->sl_compsylmax = MAXWLEN;
|
|
|
2184 lp->sl_regions[0] = NUL;
|
|
236
|
2185 }
|
|
|
2186
|
|
|
2187 /*
|
|
307
|
2188 * Load one spell file and store the info into a slang_T.
|
|
236
|
2189 * Invoked through do_in_runtimepath().
|
|
|
2190 */
|
|
|
2191 static void
|
|
310
|
2192 spell_load_cb(fname, cookie)
|
|
236
|
2193 char_u *fname;
|
|
307
|
2194 void *cookie; /* points to the language name */
|
|
236
|
2195 {
|
|
323
|
2196 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
2197 }
|
|
|
2198
|
|
|
2199 /*
|
|
|
2200 * Load one spell file and store the info into a slang_T.
|
|
|
2201 *
|
|
|
2202 * This is invoked in two ways:
|
|
|
2203 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
2204 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
2205 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
2206 * points to the existing slang_T.
|
|
323
|
2207 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
2208 */
|
|
323
|
2209 static slang_T *
|
|
|
2210 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
2211 char_u *fname;
|
|
|
2212 char_u *lang;
|
|
|
2213 slang_T *old_lp;
|
|
323
|
2214 int silent; /* no error if file doesn't exist */
|
|
310
|
2215 {
|
|
236
|
2216 FILE *fd;
|
|
483
|
2217 char_u buf[VIMSPELLMAGICL];
|
|
236
|
2218 char_u *p;
|
|
339
|
2219 char_u *bp;
|
|
|
2220 idx_T *ip;
|
|
236
|
2221 int i;
|
|
339
|
2222 int n;
|
|
300
|
2223 int len;
|
|
236
|
2224 int round;
|
|
|
2225 char_u *save_sourcing_name = sourcing_name;
|
|
|
2226 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
307
|
2227 slang_T *lp = NULL;
|
|
324
|
2228 idx_T idx;
|
|
344
|
2229 int c = 0;
|
|
483
|
2230 int res;
|
|
236
|
2231
|
|
310
|
2232 fd = mch_fopen((char *)fname, "r");
|
|
236
|
2233 if (fd == NULL)
|
|
|
2234 {
|
|
323
|
2235 if (!silent)
|
|
|
2236 EMSG2(_(e_notopen), fname);
|
|
|
2237 else if (p_verbose > 2)
|
|
|
2238 {
|
|
|
2239 verbose_enter();
|
|
|
2240 smsg((char_u *)e_notopen, fname);
|
|
|
2241 verbose_leave();
|
|
|
2242 }
|
|
255
|
2243 goto endFAIL;
|
|
236
|
2244 }
|
|
310
|
2245 if (p_verbose > 2)
|
|
|
2246 {
|
|
|
2247 verbose_enter();
|
|
|
2248 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
2249 verbose_leave();
|
|
|
2250 }
|
|
|
2251
|
|
|
2252 if (old_lp == NULL)
|
|
|
2253 {
|
|
|
2254 lp = slang_alloc(lang);
|
|
|
2255 if (lp == NULL)
|
|
|
2256 goto endFAIL;
|
|
|
2257
|
|
|
2258 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
2259 lp->sl_fname = vim_strsave(fname);
|
|
|
2260 if (lp->sl_fname == NULL)
|
|
|
2261 goto endFAIL;
|
|
|
2262
|
|
|
2263 /* Check for .add.spl. */
|
|
|
2264 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
2265 }
|
|
|
2266 else
|
|
|
2267 lp = old_lp;
|
|
307
|
2268
|
|
236
|
2269 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
2270 sourcing_name = fname;
|
|
|
2271 sourcing_lnum = 0;
|
|
|
2272
|
|
339
|
2273 /* <HEADER>: <fileID>
|
|
|
2274 */
|
|
236
|
2275 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
2276 buf[i] = getc(fd); /* <fileID> */
|
|
|
2277 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
2278 {
|
|
483
|
2279 EMSG(_("E757: This does not look like a spell file"));
|
|
381
|
2280 goto endFAIL;
|
|
483
|
2281 }
|
|
|
2282 c = getc(fd); /* <versionnr> */
|
|
|
2283 if (c < VIMSPELLVERSION)
|
|
|
2284 {
|
|
|
2285 EMSG(_("E771: Old spell file, needs to be updated"));
|
|
381
|
2286 goto endFAIL;
|
|
|
2287 }
|
|
483
|
2288 else if (c > VIMSPELLVERSION)
|
|
|
2289 {
|
|
|
2290 EMSG(_("E772: Spell file is for newer version of Vim"));
|
|
481
|
2291 goto endFAIL;
|
|
483
|
2292 }
|
|
|
2293
|
|
|
2294
|
|
|
2295 /*
|
|
|
2296 * <SECTIONS>: <section> ... <sectionend>
|
|
|
2297 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
2298 */
|
|
|
2299 for (;;)
|
|
|
2300 {
|
|
|
2301 n = getc(fd); /* <sectionID> or <sectionend> */
|
|
|
2302 if (n == SN_END)
|
|
|
2303 break;
|
|
|
2304 c = getc(fd); /* <sectionflags> */
|
|
|
2305 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2306 /* <sectionlen> */
|
|
|
2307 if (len < 0)
|
|
|
2308 goto truncerr;
|
|
|
2309
|
|
|
2310 res = 0;
|
|
|
2311 switch (n)
|
|
|
2312 {
|
|
|
2313 case SN_REGION:
|
|
|
2314 res = read_region_section(fd, lp, len);
|
|
|
2315 break;
|
|
|
2316
|
|
|
2317 case SN_CHARFLAGS:
|
|
|
2318 res = read_charflags_section(fd);
|
|
|
2319 break;
|
|
|
2320
|
|
|
2321 case SN_MIDWORD:
|
|
|
2322 lp->sl_midword = read_string(fd, len); /* <midword> */
|
|
|
2323 if (lp->sl_midword == NULL)
|
|
|
2324 goto endFAIL;
|
|
|
2325 break;
|
|
|
2326
|
|
|
2327 case SN_PREFCOND:
|
|
|
2328 res = read_prefcond_section(fd, lp);
|
|
|
2329 break;
|
|
|
2330
|
|
|
2331 case SN_REP:
|
|
|
2332 res = read_rep_section(fd, lp);
|
|
|
2333 break;
|
|
|
2334
|
|
|
2335 case SN_SAL:
|
|
|
2336 res = read_sal_section(fd, lp);
|
|
|
2337 break;
|
|
|
2338
|
|
|
2339 case SN_SOFO:
|
|
|
2340 res = read_sofo_section(fd, lp);
|
|
|
2341 break;
|
|
|
2342
|
|
|
2343 case SN_MAP:
|
|
|
2344 p = read_string(fd, len); /* <mapstr> */
|
|
|
2345 if (p == NULL)
|
|
|
2346 goto endFAIL;
|
|
|
2347 set_map_str(lp, p);
|
|
|
2348 vim_free(p);
|
|
|
2349 break;
|
|
|
2350
|
|
|
2351 case SN_COMPOUND:
|
|
|
2352 res = read_compound(fd, lp, len);
|
|
|
2353 break;
|
|
|
2354
|
|
492
|
2355 case SN_NOBREAK:
|
|
|
2356 lp->sl_nobreak = TRUE;
|
|
|
2357 break;
|
|
|
2358
|
|
483
|
2359 case SN_SYLLABLE:
|
|
|
2360 lp->sl_syllable = read_string(fd, len); /* <syllable> */
|
|
|
2361 if (lp->sl_syllable == NULL)
|
|
|
2362 goto endFAIL;
|
|
|
2363 if (init_syl_tab(lp) == FAIL)
|
|
|
2364 goto endFAIL;
|
|
|
2365 break;
|
|
|
2366
|
|
|
2367 default:
|
|
|
2368 /* Unsupported section. When it's required give an error
|
|
|
2369 * message. When it's not required skip the contents. */
|
|
|
2370 if (c & SNF_REQUIRED)
|
|
|
2371 {
|
|
|
2372 EMSG(_("E770: Unsupported section in spell file"));
|
|
|
2373 goto endFAIL;
|
|
|
2374 }
|
|
|
2375 while (--len >= 0)
|
|
|
2376 if (getc(fd) < 0)
|
|
|
2377 goto truncerr;
|
|
|
2378 break;
|
|
|
2379 }
|
|
|
2380 if (res == SP_FORMERROR)
|
|
|
2381 {
|
|
|
2382 formerr:
|
|
|
2383 EMSG(_(e_format));
|
|
381
|
2384 goto endFAIL;
|
|
344
|
2385 }
|
|
483
|
2386 if (res == SP_TRUNCERROR)
|
|
|
2387 {
|
|
|
2388 truncerr:
|
|
|
2389 EMSG(_(e_spell_trunc));
|
|
375
|
2390 goto endFAIL;
|
|
483
|
2391 }
|
|
495
|
2392 if (res == SP_OTHERERROR)
|
|
483
|
2393 goto endFAIL;
|
|
389
|
2394 }
|
|
236
|
2395
|
|
300
|
2396 /* round 1: <LWORDTREE>
|
|
339
|
2397 * round 2: <KWORDTREE>
|
|
|
2398 * round 3: <PREFIXTREE> */
|
|
|
2399 for (round = 1; round <= 3; ++round)
|
|
236
|
2400 {
|
|
300
|
2401 /* The tree size was computed when writing the file, so that we can
|
|
|
2402 * allocate it as one long block. <nodecount> */
|
|
|
2403 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2404 if (len < 0)
|
|
|
2405 goto truncerr;
|
|
|
2406 if (len > 0)
|
|
236
|
2407 {
|
|
300
|
2408 /* Allocate the byte array. */
|
|
339
|
2409 bp = lalloc((long_u)len, TRUE);
|
|
|
2410 if (bp == NULL)
|
|
300
|
2411 goto endFAIL;
|
|
|
2412 if (round == 1)
|
|
339
|
2413 lp->sl_fbyts = bp;
|
|
|
2414 else if (round == 2)
|
|
|
2415 lp->sl_kbyts = bp;
|
|
300
|
2416 else
|
|
339
|
2417 lp->sl_pbyts = bp;
|
|
236
|
2418
|
|
300
|
2419 /* Allocate the index array. */
|
|
339
|
2420 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2421 if (ip == NULL)
|
|
300
|
2422 goto endFAIL;
|
|
|
2423 if (round == 1)
|
|
339
|
2424 lp->sl_fidxs = ip;
|
|
|
2425 else if (round == 2)
|
|
|
2426 lp->sl_kidxs = ip;
|
|
300
|
2427 else
|
|
339
|
2428 lp->sl_pidxs = ip;
|
|
300
|
2429
|
|
|
2430 /* Read the tree and store it in the array. */
|
|
339
|
2431 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2432 if (idx == -1)
|
|
300
|
2433 goto truncerr;
|
|
324
|
2434 if (idx < 0)
|
|
236
|
2435 goto formerr;
|
|
|
2436 }
|
|
300
|
2437 }
|
|
243
|
2438
|
|
310
|
2439 /* For a new file link it in the list of spell files. */
|
|
|
2440 if (old_lp == NULL)
|
|
|
2441 {
|
|
|
2442 lp->sl_next = first_lang;
|
|
|
2443 first_lang = lp;
|
|
|
2444 }
|
|
307
|
2445
|
|
255
|
2446 goto endOK;
|
|
|
2447
|
|
|
2448 endFAIL:
|
|
310
|
2449 if (lang != NULL)
|
|
|
2450 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2451 *lang = NUL;
|
|
|
2452 if (lp != NULL && old_lp == NULL)
|
|
323
|
2453 {
|
|
307
|
2454 slang_free(lp);
|
|
323
|
2455 lp = NULL;
|
|
|
2456 }
|
|
255
|
2457
|
|
|
2458 endOK:
|
|
236
|
2459 if (fd != NULL)
|
|
|
2460 fclose(fd);
|
|
|
2461 sourcing_name = save_sourcing_name;
|
|
|
2462 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2463
|
|
|
2464 return lp;
|
|
236
|
2465 }
|
|
|
2466
|
|
376
|
2467 /*
|
|
|
2468 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2469 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2470 * Returns NULL when the count is zero.
|
|
483
|
2471 * Sets "*cntp" to SP_*ERROR when there is an error, length of the result
|
|
|
2472 * otherwise.
|
|
376
|
2473 */
|
|
|
2474 static char_u *
|
|
389
|
2475 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2476 FILE *fd;
|
|
|
2477 int cnt_bytes;
|
|
389
|
2478 int *cntp;
|
|
376
|
2479 {
|
|
|
2480 int cnt = 0;
|
|
|
2481 int i;
|
|
|
2482 char_u *str;
|
|
|
2483
|
|
|
2484 /* read the length bytes, MSB first */
|
|
|
2485 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2486 cnt = (cnt << 8) + getc(fd);
|
|
|
2487 if (cnt < 0)
|
|
|
2488 {
|
|
483
|
2489 *cntp = SP_TRUNCERROR;
|
|
376
|
2490 return NULL;
|
|
|
2491 }
|
|
389
|
2492 *cntp = cnt;
|
|
|
2493 if (cnt == 0)
|
|
|
2494 return NULL; /* nothing to read, return NULL */
|
|
376
|
2495
|
|
483
|
2496 str = read_string(fd, cnt);
|
|
|
2497 if (str == NULL)
|
|
495
|
2498 *cntp = SP_OTHERERROR;
|
|
483
|
2499 return str;
|
|
|
2500 }
|
|
|
2501
|
|
|
2502 /*
|
|
|
2503 * Read a string of length "cnt" from "fd" into allocated memory.
|
|
|
2504 * Returns NULL when out of memory.
|
|
|
2505 */
|
|
|
2506 static char_u *
|
|
|
2507 read_string(fd, cnt)
|
|
|
2508 FILE *fd;
|
|
|
2509 int cnt;
|
|
|
2510 {
|
|
|
2511 char_u *str;
|
|
|
2512 int i;
|
|
|
2513
|
|
376
|
2514 /* allocate memory */
|
|
|
2515 str = alloc((unsigned)cnt + 1);
|
|
483
|
2516 if (str != NULL)
|
|
|
2517 {
|
|
|
2518 /* Read the string. Doesn't check for truncated file. */
|
|
|
2519 for (i = 0; i < cnt; ++i)
|
|
|
2520 str[i] = getc(fd);
|
|
|
2521 str[i] = NUL;
|
|
|
2522 }
|
|
|
2523 return str;
|
|
|
2524 }
|
|
|
2525
|
|
|
2526 /*
|
|
|
2527 * Read SN_REGION: <regionname> ...
|
|
|
2528 * Return SP_*ERROR flags.
|
|
|
2529 */
|
|
|
2530 static int
|
|
|
2531 read_region_section(fd, lp, len)
|
|
|
2532 FILE *fd;
|
|
|
2533 slang_T *lp;
|
|
|
2534 int len;
|
|
|
2535 {
|
|
|
2536 int i;
|
|
|
2537
|
|
|
2538 if (len > 16)
|
|
|
2539 return SP_FORMERROR;
|
|
|
2540 for (i = 0; i < len; ++i)
|
|
|
2541 lp->sl_regions[i] = getc(fd); /* <regionname> */
|
|
|
2542 lp->sl_regions[len] = NUL;
|
|
|
2543 return 0;
|
|
|
2544 }
|
|
|
2545
|
|
|
2546 /*
|
|
|
2547 * Read SN_CHARFLAGS section: <charflagslen> <charflags>
|
|
|
2548 * <folcharslen> <folchars>
|
|
|
2549 * Return SP_*ERROR flags.
|
|
|
2550 */
|
|
|
2551 static int
|
|
|
2552 read_charflags_section(fd)
|
|
|
2553 FILE *fd;
|
|
|
2554 {
|
|
|
2555 char_u *flags;
|
|
|
2556 char_u *fol;
|
|
|
2557 int flagslen, follen;
|
|
|
2558
|
|
|
2559 /* <charflagslen> <charflags> */
|
|
|
2560 flags = read_cnt_string(fd, 1, &flagslen);
|
|
|
2561 if (flagslen < 0)
|
|
|
2562 return flagslen;
|
|
|
2563
|
|
|
2564 /* <folcharslen> <folchars> */
|
|
|
2565 fol = read_cnt_string(fd, 2, &follen);
|
|
|
2566 if (follen < 0)
|
|
|
2567 {
|
|
|
2568 vim_free(flags);
|
|
|
2569 return follen;
|
|
|
2570 }
|
|
|
2571
|
|
|
2572 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
2573 if (flags != NULL && fol != NULL)
|
|
|
2574 set_spell_charflags(flags, flagslen, fol);
|
|
|
2575
|
|
|
2576 vim_free(flags);
|
|
|
2577 vim_free(fol);
|
|
|
2578
|
|
|
2579 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
2580 if ((flags == NULL) != (fol == NULL))
|
|
|
2581 return SP_FORMERROR;
|
|
|
2582 return 0;
|
|
|
2583 }
|
|
|
2584
|
|
|
2585 /*
|
|
|
2586 * Read SN_PREFCOND section.
|
|
|
2587 * Return SP_*ERROR flags.
|
|
|
2588 */
|
|
|
2589 static int
|
|
|
2590 read_prefcond_section(fd, lp)
|
|
|
2591 FILE *fd;
|
|
|
2592 slang_T *lp;
|
|
|
2593 {
|
|
|
2594 int cnt;
|
|
|
2595 int i;
|
|
|
2596 int n;
|
|
|
2597 char_u *p;
|
|
|
2598 char_u buf[MAXWLEN + 1];
|
|
|
2599
|
|
|
2600 /* <prefcondcnt> <prefcond> ... */
|
|
|
2601 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
2602 if (cnt <= 0)
|
|
|
2603 return SP_FORMERROR;
|
|
|
2604
|
|
|
2605 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
2606 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
2607 if (lp->sl_prefprog == NULL)
|
|
495
|
2608 return SP_OTHERERROR;
|
|
483
|
2609 lp->sl_prefixcnt = cnt;
|
|
|
2610
|
|
376
|
2611 for (i = 0; i < cnt; ++i)
|
|
483
|
2612 {
|
|
|
2613 /* <prefcond> : <condlen> <condstr> */
|
|
|
2614 n = getc(fd); /* <condlen> */
|
|
|
2615 if (n < 0 || n >= MAXWLEN)
|
|
|
2616 return SP_FORMERROR;
|
|
|
2617
|
|
|
2618 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
2619 * compile the regexp program used to check for the condition. */
|
|
|
2620 if (n > 0)
|
|
|
2621 {
|
|
|
2622 buf[0] = '^'; /* always match at one position only */
|
|
|
2623 p = buf + 1;
|
|
|
2624 while (n-- > 0)
|
|
|
2625 *p++ = getc(fd); /* <condstr> */
|
|
|
2626 *p = NUL;
|
|
|
2627 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
2628 }
|
|
|
2629 }
|
|
|
2630 return 0;
|
|
|
2631 }
|
|
|
2632
|
|
|
2633 /*
|
|
|
2634 * Read REP items section from "fd": <repcount> <rep> ...
|
|
|
2635 * Return SP_*ERROR flags.
|
|
|
2636 */
|
|
|
2637 static int
|
|
|
2638 read_rep_section(fd, slang)
|
|
|
2639 FILE *fd;
|
|
|
2640 slang_T *slang;
|
|
|
2641 {
|
|
|
2642 int cnt;
|
|
|
2643 garray_T *gap;
|
|
|
2644 fromto_T *ftp;
|
|
|
2645 short *first;
|
|
|
2646 int i;
|
|
|
2647
|
|
|
2648 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
2649 if (cnt < 0)
|
|
|
2650 return SP_TRUNCERROR;
|
|
|
2651
|
|
|
2652 gap = &slang->sl_rep;
|
|
|
2653 if (ga_grow(gap, cnt) == FAIL)
|
|
495
|
2654 return SP_OTHERERROR;
|
|
483
|
2655
|
|
|
2656 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
2657 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2658 {
|
|
|
2659 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
|
2660 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
|
2661 if (i < 0)
|
|
|
2662 return i;
|
|
|
2663 if (i == 0)
|
|
|
2664 return SP_FORMERROR;
|
|
|
2665 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
|
2666 if (i <= 0)
|
|
|
2667 {
|
|
|
2668 vim_free(ftp->ft_from);
|
|
|
2669 if (i < 0)
|
|
|
2670 return i;
|
|
|
2671 return SP_FORMERROR;
|
|
|
2672 }
|
|
|
2673 }
|
|
|
2674
|
|
|
2675 /* Fill the first-index table. */
|
|
|
2676 first = slang->sl_rep_first;
|
|
|
2677 for (i = 0; i < 256; ++i)
|
|
|
2678 first[i] = -1;
|
|
|
2679 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2680 {
|
|
|
2681 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
2682 if (first[*ftp->ft_from] == -1)
|
|
|
2683 first[*ftp->ft_from] = i;
|
|
|
2684 }
|
|
|
2685 return 0;
|
|
|
2686 }
|
|
|
2687
|
|
|
2688 /*
|
|
|
2689 * Read SN_SAL section: <salflags> <salcount> <sal> ...
|
|
|
2690 * Return SP_*ERROR flags.
|
|
|
2691 */
|
|
|
2692 static int
|
|
|
2693 read_sal_section(fd, slang)
|
|
|
2694 FILE *fd;
|
|
|
2695 slang_T *slang;
|
|
|
2696 {
|
|
|
2697 int i;
|
|
|
2698 int cnt;
|
|
|
2699 garray_T *gap;
|
|
|
2700 salitem_T *smp;
|
|
|
2701 int ccnt;
|
|
|
2702 char_u *p;
|
|
485
|
2703 int c = NUL;
|
|
483
|
2704
|
|
|
2705 slang->sl_sofo = FALSE;
|
|
|
2706
|
|
|
2707 i = getc(fd); /* <salflags> */
|
|
|
2708 if (i & SAL_F0LLOWUP)
|
|
|
2709 slang->sl_followup = TRUE;
|
|
|
2710 if (i & SAL_COLLAPSE)
|
|
|
2711 slang->sl_collapse = TRUE;
|
|
|
2712 if (i & SAL_REM_ACCENTS)
|
|
|
2713 slang->sl_rem_accents = TRUE;
|
|
|
2714
|
|
|
2715 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2716 if (cnt < 0)
|
|
|
2717 return SP_TRUNCERROR;
|
|
|
2718
|
|
|
2719 gap = &slang->sl_sal;
|
|
|
2720 ga_init2(gap, sizeof(salitem_T), 10);
|
|
|
2721 if (ga_grow(gap, cnt) == FAIL)
|
|
495
|
2722 return SP_OTHERERROR;
|
|
483
|
2723
|
|
|
2724 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2725 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2726 {
|
|
|
2727 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2728 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2729 if (ccnt < 0)
|
|
|
2730 return SP_TRUNCERROR;
|
|
|
2731 if ((p = alloc(ccnt + 2)) == NULL)
|
|
495
|
2732 return SP_OTHERERROR;
|
|
483
|
2733 smp->sm_lead = p;
|
|
|
2734
|
|
|
2735 /* Read up to the first special char into sm_lead. */
|
|
|
2736 for (i = 0; i < ccnt; ++i)
|
|
|
2737 {
|
|
|
2738 c = getc(fd); /* <salfrom> */
|
|
|
2739 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2740 break;
|
|
|
2741 *p++ = c;
|
|
|
2742 }
|
|
|
2743 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2744 *p++ = NUL;
|
|
|
2745
|
|
|
2746 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2747 if (c == '(')
|
|
|
2748 {
|
|
|
2749 smp->sm_oneof = p;
|
|
|
2750 for (++i; i < ccnt; ++i)
|
|
|
2751 {
|
|
|
2752 c = getc(fd); /* <salfrom> */
|
|
|
2753 if (c == ')')
|
|
|
2754 break;
|
|
|
2755 *p++ = c;
|
|
|
2756 }
|
|
|
2757 *p++ = NUL;
|
|
|
2758 if (++i < ccnt)
|
|
|
2759 c = getc(fd);
|
|
|
2760 }
|
|
|
2761 else
|
|
|
2762 smp->sm_oneof = NULL;
|
|
|
2763
|
|
|
2764 /* Any following chars go in sm_rules. */
|
|
|
2765 smp->sm_rules = p;
|
|
|
2766 if (i < ccnt)
|
|
|
2767 /* store the char we got while checking for end of sm_lead */
|
|
|
2768 *p++ = c;
|
|
|
2769 for (++i; i < ccnt; ++i)
|
|
|
2770 *p++ = getc(fd); /* <salfrom> */
|
|
|
2771 *p++ = NUL;
|
|
|
2772
|
|
|
2773 /* <saltolen> <salto> */
|
|
|
2774 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
|
2775 if (ccnt < 0)
|
|
|
2776 {
|
|
|
2777 vim_free(smp->sm_lead);
|
|
|
2778 return ccnt;
|
|
|
2779 }
|
|
|
2780
|
|
|
2781 #ifdef FEAT_MBYTE
|
|
|
2782 if (has_mbyte)
|
|
|
2783 {
|
|
|
2784 /* convert the multi-byte strings to wide char strings */
|
|
|
2785 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2786 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2787 if (smp->sm_oneof == NULL)
|
|
|
2788 smp->sm_oneof_w = NULL;
|
|
|
2789 else
|
|
|
2790 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
2791 if (smp->sm_to == NULL)
|
|
|
2792 smp->sm_to_w = NULL;
|
|
|
2793 else
|
|
|
2794 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
2795 if (smp->sm_lead_w == NULL
|
|
|
2796 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
2797 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
|
2798 {
|
|
|
2799 vim_free(smp->sm_lead);
|
|
|
2800 vim_free(smp->sm_to);
|
|
|
2801 vim_free(smp->sm_lead_w);
|
|
|
2802 vim_free(smp->sm_oneof_w);
|
|
|
2803 vim_free(smp->sm_to_w);
|
|
495
|
2804 return SP_OTHERERROR;
|
|
483
|
2805 }
|
|
|
2806 }
|
|
|
2807 #endif
|
|
|
2808 }
|
|
|
2809
|
|
|
2810 /* Fill the first-index table. */
|
|
|
2811 set_sal_first(slang);
|
|
|
2812
|
|
|
2813 return 0;
|
|
|
2814 }
|
|
|
2815
|
|
|
2816 /*
|
|
|
2817 * SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
2818 * Return SP_*ERROR flags.
|
|
|
2819 */
|
|
|
2820 static int
|
|
|
2821 read_sofo_section(fd, slang)
|
|
|
2822 FILE *fd;
|
|
|
2823 slang_T *slang;
|
|
|
2824 {
|
|
|
2825 int cnt;
|
|
|
2826 char_u *from, *to;
|
|
|
2827 int res;
|
|
|
2828
|
|
|
2829 slang->sl_sofo = TRUE;
|
|
|
2830
|
|
|
2831 /* <sofofromlen> <sofofrom> */
|
|
|
2832 from = read_cnt_string(fd, 2, &cnt);
|
|
|
2833 if (cnt < 0)
|
|
|
2834 return cnt;
|
|
|
2835
|
|
|
2836 /* <sofotolen> <sofoto> */
|
|
|
2837 to = read_cnt_string(fd, 2, &cnt);
|
|
|
2838 if (cnt < 0)
|
|
|
2839 {
|
|
|
2840 vim_free(from);
|
|
|
2841 return cnt;
|
|
|
2842 }
|
|
|
2843
|
|
|
2844 /* Store the info in slang->sl_sal and/or slang->sl_sal_first. */
|
|
|
2845 if (from != NULL && to != NULL)
|
|
|
2846 res = set_sofo(slang, from, to);
|
|
|
2847 else if (from != NULL || to != NULL)
|
|
|
2848 res = SP_FORMERROR; /* only one of two strings is an error */
|
|
|
2849 else
|
|
|
2850 res = 0;
|
|
|
2851
|
|
|
2852 vim_free(from);
|
|
|
2853 vim_free(to);
|
|
|
2854 return res;
|
|
|
2855 }
|
|
|
2856
|
|
|
2857 /*
|
|
|
2858 * Read the compound section from the .spl file:
|
|
|
2859 * <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
2860 * Returns SP_*ERROR flags.
|
|
|
2861 */
|
|
|
2862 static int
|
|
|
2863 read_compound(fd, slang, len)
|
|
|
2864 FILE *fd;
|
|
|
2865 slang_T *slang;
|
|
|
2866 int len;
|
|
|
2867 {
|
|
|
2868 int todo = len;
|
|
|
2869 int c;
|
|
|
2870 int atstart;
|
|
|
2871 char_u *pat;
|
|
|
2872 char_u *pp;
|
|
|
2873 char_u *cp;
|
|
485
|
2874 char_u *ap;
|
|
483
|
2875
|
|
|
2876 if (todo < 2)
|
|
|
2877 return SP_FORMERROR; /* need at least two bytes */
|
|
|
2878
|
|
|
2879 --todo;
|
|
|
2880 c = getc(fd); /* <compmax> */
|
|
|
2881 if (c < 2)
|
|
|
2882 c = MAXWLEN;
|
|
|
2883 slang->sl_compmax = c;
|
|
|
2884
|
|
|
2885 --todo;
|
|
|
2886 c = getc(fd); /* <compminlen> */
|
|
|
2887 if (c < 1)
|
|
|
2888 c = 3;
|
|
|
2889 slang->sl_compminlen = c;
|
|
|
2890
|
|
|
2891 --todo;
|
|
|
2892 c = getc(fd); /* <compsylmax> */
|
|
|
2893 if (c < 1)
|
|
|
2894 c = MAXWLEN;
|
|
|
2895 slang->sl_compsylmax = c;
|
|
|
2896
|
|
|
2897 /* Turn the COMPOUNDFLAGS items into a regexp pattern:
|
|
|
2898 * "a[bc]/a*b+" -> "^\(a[bc]\|a*b\+\)$".
|
|
495
|
2899 * Inserting backslashes may double the length, "^\(\)$<Nul>" is 7 bytes.
|
|
|
2900 * Conversion to utf-8 may double the size. */
|
|
|
2901 c = todo * 2 + 7;
|
|
|
2902 #ifdef FEAT_MBYTE
|
|
|
2903 if (enc_utf8)
|
|
|
2904 c += todo * 2;
|
|
|
2905 #endif
|
|
|
2906 pat = alloc((unsigned)c);
|
|
483
|
2907 if (pat == NULL)
|
|
495
|
2908 return SP_OTHERERROR;
|
|
483
|
2909
|
|
485
|
2910 /* We also need a list of all flags that can appear at the start and one
|
|
|
2911 * for all flags. */
|
|
483
|
2912 cp = alloc(todo + 1);
|
|
|
2913 if (cp == NULL)
|
|
|
2914 {
|
|
|
2915 vim_free(pat);
|
|
495
|
2916 return SP_OTHERERROR;
|
|
483
|
2917 }
|
|
|
2918 slang->sl_compstartflags = cp;
|
|
|
2919 *cp = NUL;
|
|
|
2920
|
|
485
|
2921 ap = alloc(todo + 1);
|
|
|
2922 if (ap == NULL)
|
|
|
2923 {
|
|
|
2924 vim_free(pat);
|
|
495
|
2925 return SP_OTHERERROR;
|
|
485
|
2926 }
|
|
|
2927 slang->sl_compallflags = ap;
|
|
|
2928 *ap = NUL;
|
|
|
2929
|
|
483
|
2930 pp = pat;
|
|
|
2931 *pp++ = '^';
|
|
|
2932 *pp++ = '\\';
|
|
|
2933 *pp++ = '(';
|
|
|
2934
|
|
|
2935 atstart = 1;
|
|
|
2936 while (todo-- > 0)
|
|
|
2937 {
|
|
|
2938 c = getc(fd); /* <compflags> */
|
|
485
|
2939
|
|
|
2940 /* Add all flags to "sl_compallflags". */
|
|
|
2941 if (vim_strchr((char_u *)"+*[]/", c) == NULL
|
|
495
|
2942 && !byte_in_str(slang->sl_compallflags, c))
|
|
485
|
2943 {
|
|
|
2944 *ap++ = c;
|
|
|
2945 *ap = NUL;
|
|
|
2946 }
|
|
|
2947
|
|
483
|
2948 if (atstart != 0)
|
|
|
2949 {
|
|
|
2950 /* At start of item: copy flags to "sl_compstartflags". For a
|
|
|
2951 * [abc] item set "atstart" to 2 and copy up to the ']'. */
|
|
|
2952 if (c == '[')
|
|
|
2953 atstart = 2;
|
|
|
2954 else if (c == ']')
|
|
|
2955 atstart = 0;
|
|
|
2956 else
|
|
|
2957 {
|
|
495
|
2958 if (!byte_in_str(slang->sl_compstartflags, c))
|
|
483
|
2959 {
|
|
|
2960 *cp++ = c;
|
|
|
2961 *cp = NUL;
|
|
|
2962 }
|
|
|
2963 if (atstart == 1)
|
|
|
2964 atstart = 0;
|
|
|
2965 }
|
|
|
2966 }
|
|
|
2967 if (c == '/') /* slash separates two items */
|
|
|
2968 {
|
|
|
2969 *pp++ = '\\';
|
|
|
2970 *pp++ = '|';
|
|
|
2971 atstart = 1;
|
|
|
2972 }
|
|
|
2973 else /* normal char, "[abc]" and '*' are copied as-is */
|
|
|
2974 {
|
|
|
2975 if (c == '+')
|
|
|
2976 *pp++ = '\\'; /* "a+" becomes "a\+" */
|
|
495
|
2977 #ifdef FEAT_MBYTE
|
|
|
2978 if (enc_utf8)
|
|
|
2979 pp += mb_char2bytes(c, pp);
|
|
|
2980 else
|
|
|
2981 #endif
|
|
|
2982 *pp++ = c;
|
|
483
|
2983 }
|
|
|
2984 }
|
|
|
2985
|
|
|
2986 *pp++ = '\\';
|
|
|
2987 *pp++ = ')';
|
|
|
2988 *pp++ = '$';
|
|
|
2989 *pp = NUL;
|
|
|
2990
|
|
|
2991 slang->sl_compprog = vim_regcomp(pat, RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
2992 vim_free(pat);
|
|
|
2993 if (slang->sl_compprog == NULL)
|
|
|
2994 return SP_FORMERROR;
|
|
|
2995
|
|
|
2996 return 0;
|
|
|
2997 }
|
|
|
2998
|
|
495
|
2999 /*
|
|
498
|
3000 * Return TRUE if byte "n" appears in "str".
|
|
495
|
3001 * Like strchr() but independent of locale.
|
|
|
3002 */
|
|
|
3003 static int
|
|
498
|
3004 byte_in_str(str, n)
|
|
495
|
3005 char_u *str;
|
|
498
|
3006 int n;
|
|
495
|
3007 {
|
|
|
3008 char_u *p;
|
|
|
3009
|
|
|
3010 for (p = str; *p != NUL; ++p)
|
|
498
|
3011 if (*p == n)
|
|
495
|
3012 return TRUE;
|
|
|
3013 return FALSE;
|
|
|
3014 }
|
|
|
3015
|
|
483
|
3016 #define SY_MAXLEN 30
|
|
|
3017 typedef struct syl_item_S
|
|
|
3018 {
|
|
|
3019 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */
|
|
|
3020 int sy_len;
|
|
|
3021 } syl_item_T;
|
|
|
3022
|
|
|
3023 /*
|
|
|
3024 * Truncate "slang->sl_syllable" at the first slash and put the following items
|
|
|
3025 * in "slang->sl_syl_items".
|
|
|
3026 */
|
|
|
3027 static int
|
|
|
3028 init_syl_tab(slang)
|
|
|
3029 slang_T *slang;
|
|
|
3030 {
|
|
|
3031 char_u *p;
|
|
|
3032 char_u *s;
|
|
|
3033 int l;
|
|
|
3034 syl_item_T *syl;
|
|
|
3035
|
|
|
3036 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4);
|
|
|
3037 p = vim_strchr(slang->sl_syllable, '/');
|
|
|
3038 while (p != NULL)
|
|
|
3039 {
|
|
|
3040 *p++ = NUL;
|
|
495
|
3041 if (*p == NUL) /* trailing slash */
|
|
483
|
3042 break;
|
|
|
3043 s = p;
|
|
|
3044 p = vim_strchr(p, '/');
|
|
|
3045 if (p == NULL)
|
|
|
3046 l = STRLEN(s);
|
|
|
3047 else
|
|
|
3048 l = p - s;
|
|
|
3049 if (l >= SY_MAXLEN)
|
|
|
3050 return SP_FORMERROR;
|
|
|
3051 if (ga_grow(&slang->sl_syl_items, 1) == FAIL)
|
|
495
|
3052 return SP_OTHERERROR;
|
|
483
|
3053 syl = ((syl_item_T *)slang->sl_syl_items.ga_data)
|
|
|
3054 + slang->sl_syl_items.ga_len++;
|
|
|
3055 vim_strncpy(syl->sy_chars, s, l);
|
|
|
3056 syl->sy_len = l;
|
|
|
3057 }
|
|
|
3058 return OK;
|
|
|
3059 }
|
|
|
3060
|
|
|
3061 /*
|
|
|
3062 * Count the number of syllables in "word".
|
|
|
3063 * When "word" contains spaces the syllables after the last space are counted.
|
|
|
3064 * Returns zero if syllables are not defines.
|
|
|
3065 */
|
|
|
3066 static int
|
|
|
3067 count_syllables(slang, word)
|
|
|
3068 slang_T *slang;
|
|
|
3069 char_u *word;
|
|
|
3070 {
|
|
|
3071 int cnt = 0;
|
|
|
3072 int skip = FALSE;
|
|
|
3073 char_u *p;
|
|
|
3074 int len;
|
|
|
3075 int i;
|
|
|
3076 syl_item_T *syl;
|
|
|
3077 int c;
|
|
|
3078
|
|
|
3079 if (slang->sl_syllable == NULL)
|
|
|
3080 return 0;
|
|
|
3081
|
|
|
3082 for (p = word; *p != NUL; p += len)
|
|
|
3083 {
|
|
|
3084 /* When running into a space reset counter. */
|
|
|
3085 if (*p == ' ')
|
|
|
3086 {
|
|
|
3087 len = 1;
|
|
|
3088 cnt = 0;
|
|
|
3089 continue;
|
|
|
3090 }
|
|
|
3091
|
|
|
3092 /* Find longest match of syllable items. */
|
|
|
3093 len = 0;
|
|
|
3094 for (i = 0; i < slang->sl_syl_items.ga_len; ++i)
|
|
|
3095 {
|
|
|
3096 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i;
|
|
|
3097 if (syl->sy_len > len
|
|
|
3098 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0)
|
|
|
3099 len = syl->sy_len;
|
|
|
3100 }
|
|
|
3101 if (len != 0) /* found a match, count syllable */
|
|
|
3102 {
|
|
|
3103 ++cnt;
|
|
|
3104 skip = FALSE;
|
|
|
3105 }
|
|
|
3106 else
|
|
|
3107 {
|
|
|
3108 /* No recognized syllable item, at least a syllable char then? */
|
|
|
3109 #ifdef FEAT_MBYTE
|
|
|
3110 c = mb_ptr2char(p);
|
|
|
3111 len = (*mb_ptr2len)(p);
|
|
|
3112 #else
|
|
|
3113 c = *p;
|
|
|
3114 len = 1;
|
|
|
3115 #endif
|
|
|
3116 if (vim_strchr(slang->sl_syllable, c) == NULL)
|
|
|
3117 skip = FALSE; /* No, search for next syllable */
|
|
|
3118 else if (!skip)
|
|
|
3119 {
|
|
|
3120 ++cnt; /* Yes, count it */
|
|
|
3121 skip = TRUE; /* don't count following syllable chars */
|
|
|
3122 }
|
|
|
3123 }
|
|
|
3124 }
|
|
|
3125 return cnt;
|
|
376
|
3126 }
|
|
|
3127
|
|
381
|
3128 /*
|
|
|
3129 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
483
|
3130 * Returns SP_*ERROR flags when there is something wrong.
|
|
381
|
3131 */
|
|
|
3132 static int
|
|
|
3133 set_sofo(lp, from, to)
|
|
|
3134 slang_T *lp;
|
|
|
3135 char_u *from;
|
|
|
3136 char_u *to;
|
|
|
3137 {
|
|
|
3138 int i;
|
|
|
3139
|
|
|
3140 #ifdef FEAT_MBYTE
|
|
|
3141 garray_T *gap;
|
|
|
3142 char_u *s;
|
|
|
3143 char_u *p;
|
|
|
3144 int c;
|
|
|
3145 int *inp;
|
|
|
3146
|
|
|
3147 if (has_mbyte)
|
|
|
3148 {
|
|
|
3149 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
3150 * characters. The index is the low byte of the character.
|
|
|
3151 * The list contains from-to pairs with a terminating NUL.
|
|
|
3152 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
3153 gap = &lp->sl_sal;
|
|
|
3154 ga_init2(gap, sizeof(int *), 1);
|
|
|
3155 if (ga_grow(gap, 256) == FAIL)
|
|
495
|
3156 return SP_OTHERERROR;
|
|
381
|
3157 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
3158 gap->ga_len = 256;
|
|
|
3159
|
|
|
3160 /* First count the number of items for each list. Temporarily use
|
|
|
3161 * sl_sal_first[] for this. */
|
|
|
3162 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3163 {
|
|
474
|
3164 c = mb_cptr2char_adv(&p);
|
|
|
3165 mb_cptr_adv(s);
|
|
381
|
3166 if (c >= 256)
|
|
|
3167 ++lp->sl_sal_first[c & 0xff];
|
|
|
3168 }
|
|
|
3169 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
483
|
3170 return SP_FORMERROR;
|
|
381
|
3171
|
|
|
3172 /* Allocate the lists. */
|
|
|
3173 for (i = 0; i < 256; ++i)
|
|
|
3174 if (lp->sl_sal_first[i] > 0)
|
|
|
3175 {
|
|
|
3176 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
3177 if (p == NULL)
|
|
495
|
3178 return SP_OTHERERROR;
|
|
381
|
3179 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
3180 *(int *)p = 0;
|
|
|
3181 }
|
|
|
3182
|
|
|
3183 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
3184 * list. */
|
|
|
3185 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
3186 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3187 {
|
|
474
|
3188 c = mb_cptr2char_adv(&p);
|
|
|
3189 i = mb_cptr2char_adv(&s);
|
|
381
|
3190 if (c >= 256)
|
|
|
3191 {
|
|
|
3192 /* Append the from-to chars at the end of the list with
|
|
|
3193 * the low byte. */
|
|
|
3194 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
3195 while (*inp != 0)
|
|
|
3196 ++inp;
|
|
|
3197 *inp++ = c; /* from char */
|
|
|
3198 *inp++ = i; /* to char */
|
|
|
3199 *inp++ = NUL; /* NUL at the end */
|
|
|
3200 }
|
|
|
3201 else
|
|
|
3202 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
3203 lp->sl_sal_first[c] = i;
|
|
|
3204 }
|
|
|
3205 }
|
|
|
3206 else
|
|
|
3207 #endif
|
|
|
3208 {
|
|
|
3209 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
3210 if (STRLEN(from) != STRLEN(to))
|
|
483
|
3211 return SP_FORMERROR;
|
|
381
|
3212
|
|
|
3213 for (i = 0; to[i] != NUL; ++i)
|
|
|
3214 lp->sl_sal_first[from[i]] = to[i];
|
|
|
3215 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
3216 }
|
|
|
3217
|
|
483
|
3218 return 0;
|
|
381
|
3219 }
|
|
|
3220
|
|
|
3221 /*
|
|
|
3222 * Fill the first-index table for "lp".
|
|
|
3223 */
|
|
|
3224 static void
|
|
|
3225 set_sal_first(lp)
|
|
|
3226 slang_T *lp;
|
|
|
3227 {
|
|
|
3228 salfirst_T *sfirst;
|
|
|
3229 int i;
|
|
|
3230 salitem_T *smp;
|
|
|
3231 int c;
|
|
|
3232 garray_T *gap = &lp->sl_sal;
|
|
|
3233
|
|
|
3234 sfirst = lp->sl_sal_first;
|
|
|
3235 for (i = 0; i < 256; ++i)
|
|
|
3236 sfirst[i] = -1;
|
|
|
3237 smp = (salitem_T *)gap->ga_data;
|
|
|
3238 for (i = 0; i < gap->ga_len; ++i)
|
|
|
3239 {
|
|
|
3240 #ifdef FEAT_MBYTE
|
|
|
3241 if (has_mbyte)
|
|
|
3242 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
3243 * the character, for other encodings it should differ for most
|
|
|
3244 * characters. */
|
|
|
3245 c = *smp[i].sm_lead_w & 0xff;
|
|
|
3246 else
|
|
|
3247 #endif
|
|
|
3248 c = *smp[i].sm_lead;
|
|
|
3249 if (sfirst[c] == -1)
|
|
|
3250 {
|
|
|
3251 sfirst[c] = i;
|
|
|
3252 #ifdef FEAT_MBYTE
|
|
|
3253 if (has_mbyte)
|
|
|
3254 {
|
|
|
3255 int n;
|
|
|
3256
|
|
|
3257 /* Make sure all entries with this byte are following each
|
|
|
3258 * other. Move the ones that are in the wrong position. Do
|
|
|
3259 * keep the same ordering! */
|
|
|
3260 while (i + 1 < gap->ga_len
|
|
|
3261 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
3262 /* Skip over entry with same index byte. */
|
|
|
3263 ++i;
|
|
|
3264
|
|
|
3265 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
3266 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
3267 {
|
|
|
3268 salitem_T tsal;
|
|
|
3269
|
|
|
3270 /* Move entry with same index byte after the entries
|
|
|
3271 * we already found. */
|
|
|
3272 ++i;
|
|
|
3273 --n;
|
|
|
3274 tsal = smp[i + n];
|
|
|
3275 mch_memmove(smp + i + 1, smp + i,
|
|
|
3276 sizeof(salitem_T) * n);
|
|
|
3277 smp[i] = tsal;
|
|
|
3278 }
|
|
|
3279 }
|
|
|
3280 #endif
|
|
|
3281 }
|
|
|
3282 }
|
|
|
3283 }
|
|
376
|
3284
|
|
372
|
3285 #ifdef FEAT_MBYTE
|
|
|
3286 /*
|
|
|
3287 * Turn a multi-byte string into a wide character string.
|
|
|
3288 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
3289 */
|
|
|
3290 static int *
|
|
|
3291 mb_str2wide(s)
|
|
|
3292 char_u *s;
|
|
|
3293 {
|
|
|
3294 int *res;
|
|
|
3295 char_u *p;
|
|
|
3296 int i = 0;
|
|
|
3297
|
|
|
3298 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
3299 if (res != NULL)
|
|
|
3300 {
|
|
|
3301 for (p = s; *p != NUL; )
|
|
|
3302 res[i++] = mb_ptr2char_adv(&p);
|
|
|
3303 res[i] = NUL;
|
|
|
3304 }
|
|
|
3305 return res;
|
|
|
3306 }
|
|
|
3307 #endif
|
|
|
3308
|
|
236
|
3309 /*
|
|
300
|
3310 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
3311 * "byts" and index array "idxs". Recursively read the children.
|
|
|
3312 *
|
|
346
|
3313 * NOTE: The code here must match put_node().
|
|
300
|
3314 *
|
|
|
3315 * Returns the index follosing the siblings.
|
|
|
3316 * Returns -1 if the file is shorter than expected.
|
|
|
3317 * Returns -2 if there is a format error.
|
|
236
|
3318 */
|
|
324
|
3319 static idx_T
|
|
339
|
3320 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
3321 FILE *fd;
|
|
|
3322 char_u *byts;
|
|
324
|
3323 idx_T *idxs;
|
|
300
|
3324 int maxidx; /* size of arrays */
|
|
324
|
3325 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
3326 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
3327 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
3328 {
|
|
300
|
3329 int len;
|
|
|
3330 int i;
|
|
|
3331 int n;
|
|
324
|
3332 idx_T idx = startidx;
|
|
300
|
3333 int c;
|
|
366
|
3334 int c2;
|
|
300
|
3335 #define SHARED_MASK 0x8000000
|
|
236
|
3336
|
|
300
|
3337 len = getc(fd); /* <siblingcount> */
|
|
|
3338 if (len <= 0)
|
|
|
3339 return -1;
|
|
|
3340
|
|
|
3341 if (startidx + len >= maxidx)
|
|
|
3342 return -2;
|
|
|
3343 byts[idx++] = len;
|
|
|
3344
|
|
|
3345 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
3346 for (i = 1; i <= len; ++i)
|
|
236
|
3347 {
|
|
300
|
3348 c = getc(fd); /* <byte> */
|
|
|
3349 if (c < 0)
|
|
|
3350 return -1;
|
|
|
3351 if (c <= BY_SPECIAL)
|
|
|
3352 {
|
|
366
|
3353 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
3354 {
|
|
|
3355 /* No flags, all regions. */
|
|
|
3356 idxs[idx] = 0;
|
|
|
3357 c = 0;
|
|
|
3358 }
|
|
390
|
3359 else if (c != BY_INDEX)
|
|
300
|
3360 {
|
|
339
|
3361 if (prefixtree)
|
|
|
3362 {
|
|
455
|
3363 /* Read the optional pflags byte, the prefix ID and the
|
|
|
3364 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
3365 * byte, the condition index shifted up 8 bits, the flags
|
|
|
3366 * shifted up 24 bits. */
|
|
|
3367 if (c == BY_FLAGS)
|
|
|
3368 c = getc(fd) << 24; /* <pflags> */
|
|
|
3369 else
|
|
|
3370 c = 0;
|
|
|
3371
|
|
481
|
3372 c |= getc(fd); /* <affixID> */
|
|
455
|
3373
|
|
339
|
3374 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
3375 if (n >= maxprefcondnr)
|
|
|
3376 return -2;
|
|
455
|
3377 c |= (n << 8);
|
|
339
|
3378 }
|
|
390
|
3379 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
3380 {
|
|
|
3381 /* Read flags and optional region and prefix ID. In
|
|
390
|
3382 * idxs[] the flags go in the low two bytes, region above
|
|
|
3383 * that and prefix ID above the region. */
|
|
|
3384 c2 = c;
|
|
339
|
3385 c = getc(fd); /* <flags> */
|
|
390
|
3386 if (c2 == BY_FLAGS2)
|
|
|
3387 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
3388 if (c & WF_REGION)
|
|
390
|
3389 c = (getc(fd) << 16) + c; /* <region> */
|
|
481
|
3390 if (c & WF_AFX)
|
|
|
3391 c = (getc(fd) << 24) + c; /* <affixID> */
|
|
339
|
3392 }
|
|
|
3393
|
|
300
|
3394 idxs[idx] = c;
|
|
|
3395 c = 0;
|
|
|
3396 }
|
|
|
3397 else /* c == BY_INDEX */
|
|
|
3398 {
|
|
|
3399 /* <nodeidx> */
|
|
|
3400 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
3401 if (n < 0 || n >= maxidx)
|
|
|
3402 return -2;
|
|
|
3403 idxs[idx] = n + SHARED_MASK;
|
|
|
3404 c = getc(fd); /* <xbyte> */
|
|
|
3405 }
|
|
|
3406 }
|
|
|
3407 byts[idx++] = c;
|
|
236
|
3408 }
|
|
|
3409
|
|
300
|
3410 /* Recursively read the children for non-shared siblings.
|
|
|
3411 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
3412 * remove SHARED_MASK) */
|
|
|
3413 for (i = 1; i <= len; ++i)
|
|
|
3414 if (byts[startidx + i] != 0)
|
|
|
3415 {
|
|
|
3416 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
3417 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
3418 else
|
|
|
3419 {
|
|
|
3420 idxs[startidx + i] = idx;
|
|
339
|
3421 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
3422 prefixtree, maxprefcondnr);
|
|
300
|
3423 if (idx < 0)
|
|
|
3424 break;
|
|
|
3425 }
|
|
|
3426 }
|
|
236
|
3427
|
|
300
|
3428 return idx;
|
|
236
|
3429 }
|
|
|
3430
|
|
|
3431 /*
|
|
|
3432 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
3433 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
3434 */
|
|
|
3435 char_u *
|
|
|
3436 did_set_spelllang(buf)
|
|
|
3437 buf_T *buf;
|
|
|
3438 {
|
|
|
3439 garray_T ga;
|
|
351
|
3440 char_u *splp;
|
|
236
|
3441 char_u *region;
|
|
416
|
3442 char_u region_cp[3];
|
|
355
|
3443 int filename;
|
|
236
|
3444 int region_mask;
|
|
|
3445 slang_T *lp;
|
|
|
3446 int c;
|
|
351
|
3447 char_u lang[MAXWLEN + 1];
|
|
323
|
3448 char_u spf_name[MAXPATHL];
|
|
351
|
3449 int len;
|
|
|
3450 char_u *p;
|
|
381
|
3451 int round;
|
|
385
|
3452 char_u *spf;
|
|
389
|
3453 char_u *use_region = NULL;
|
|
|
3454 int dont_use_region = FALSE;
|
|
236
|
3455
|
|
|
3456 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
3457 clear_midword(buf);
|
|
236
|
3458
|
|
351
|
3459 /* loop over comma separated language names. */
|
|
|
3460 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
3461 {
|
|
|
3462 /* Get one language name. */
|
|
|
3463 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
3464
|
|
240
|
3465 region = NULL;
|
|
351
|
3466 len = STRLEN(lang);
|
|
355
|
3467
|
|
|
3468 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
3469 * If there is a region name let "region" point to it and remove it
|
|
|
3470 * from the name. */
|
|
|
3471 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
3472 {
|
|
355
|
3473 filename = TRUE;
|
|
|
3474
|
|
416
|
3475 /* Locate a region and remove it from the file name. */
|
|
|
3476 p = vim_strchr(gettail(lang), '_');
|
|
|
3477 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
3478 && !ASCII_ISALPHA(p[3]))
|
|
|
3479 {
|
|
|
3480 vim_strncpy(region_cp, p + 1, 2);
|
|
|
3481 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
3482 len -= 3;
|
|
|
3483 region = region_cp;
|
|
|
3484 }
|
|
|
3485 else
|
|
|
3486 dont_use_region = TRUE;
|
|
|
3487
|
|
355
|
3488 /* Check if we loaded this language before. */
|
|
|
3489 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3490 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3491 break;
|
|
236
|
3492 }
|
|
355
|
3493 else
|
|
|
3494 {
|
|
|
3495 filename = FALSE;
|
|
|
3496 if (len > 3 && lang[len - 3] == '_')
|
|
|
3497 {
|
|
|
3498 region = lang + len - 2;
|
|
|
3499 len -= 3;
|
|
|
3500 lang[len] = NUL;
|
|
389
|
3501 }
|
|
|
3502 else
|
|
|
3503 dont_use_region = TRUE;
|
|
355
|
3504
|
|
|
3505 /* Check if we loaded this language before. */
|
|
|
3506 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3507 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
3508 break;
|
|
|
3509 }
|
|
236
|
3510
|
|
416
|
3511 if (region != NULL)
|
|
|
3512 {
|
|
|
3513 /* If the region differs from what was used before then don't
|
|
|
3514 * use it for 'spellfile'. */
|
|
|
3515 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
3516 dont_use_region = TRUE;
|
|
|
3517 use_region = region;
|
|
|
3518 }
|
|
|
3519
|
|
351
|
3520 /* If not found try loading the language now. */
|
|
236
|
3521 if (lp == NULL)
|
|
355
|
3522 {
|
|
|
3523 if (filename)
|
|
|
3524 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
3525 else
|
|
|
3526 spell_load_lang(lang);
|
|
|
3527 }
|
|
236
|
3528
|
|
307
|
3529 /*
|
|
351
|
3530 * Loop over the languages, there can be several files for "lang".
|
|
307
|
3531 */
|
|
|
3532 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
3533 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
3534 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
3535 {
|
|
316
|
3536 region_mask = REGION_ALL;
|
|
355
|
3537 if (!filename && region != NULL)
|
|
236
|
3538 {
|
|
307
|
3539 /* find region in sl_regions */
|
|
|
3540 c = find_region(lp->sl_regions, region);
|
|
|
3541 if (c == REGION_ALL)
|
|
|
3542 {
|
|
389
|
3543 if (lp->sl_add)
|
|
|
3544 {
|
|
|
3545 if (*lp->sl_regions != NUL)
|
|
|
3546 /* This addition file is for other regions. */
|
|
|
3547 region_mask = 0;
|
|
|
3548 }
|
|
|
3549 else
|
|
|
3550 /* This is probably an error. Give a warning and
|
|
|
3551 * accept the words anyway. */
|
|
351
|
3552 smsg((char_u *)
|
|
|
3553 _("Warning: region %s not supported"),
|
|
|
3554 region);
|
|
307
|
3555 }
|
|
|
3556 else
|
|
|
3557 region_mask = 1 << c;
|
|
236
|
3558 }
|
|
307
|
3559
|
|
389
|
3560 if (region_mask != 0)
|
|
307
|
3561 {
|
|
389
|
3562 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3563 {
|
|
|
3564 ga_clear(&ga);
|
|
|
3565 return e_outofmem;
|
|
|
3566 }
|
|
|
3567 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
3568 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3569 ++ga.ga_len;
|
|
|
3570 use_midword(lp, buf);
|
|
307
|
3571 }
|
|
385
|
3572 }
|
|
|
3573 }
|
|
|
3574
|
|
|
3575 /* round 0: load int_wordlist, if possible.
|
|
|
3576 * round 1: load first name in 'spellfile'.
|
|
|
3577 * round 2: load second name in 'spellfile.
|
|
|
3578 * etc. */
|
|
|
3579 spf = curbuf->b_p_spf;
|
|
|
3580 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
3581 {
|
|
|
3582 if (round == 0)
|
|
|
3583 {
|
|
|
3584 /* Internal wordlist, if there is one. */
|
|
|
3585 if (int_wordlist == NULL)
|
|
381
|
3586 continue;
|
|
385
|
3587 int_wordlist_spl(spf_name);
|
|
381
|
3588 }
|
|
|
3589 else
|
|
|
3590 {
|
|
385
|
3591 /* One entry in 'spellfile'. */
|
|
|
3592 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
3593 STRCAT(spf_name, ".spl");
|
|
|
3594
|
|
|
3595 /* If it was already found above then skip it. */
|
|
|
3596 for (c = 0; c < ga.ga_len; ++c)
|
|
|
3597 if (fullpathcmp(spf_name,
|
|
|
3598 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
3599 FALSE) == FPC_SAME)
|
|
|
3600 break;
|
|
|
3601 if (c < ga.ga_len)
|
|
381
|
3602 continue;
|
|
|
3603 }
|
|
|
3604
|
|
351
|
3605 /* Check if it was loaded already. */
|
|
323
|
3606 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3607 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3608 break;
|
|
|
3609 if (lp == NULL)
|
|
|
3610 {
|
|
351
|
3611 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
3612 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
3613 * use an arbitrary name. */
|
|
|
3614 if (round == 0)
|
|
|
3615 STRCPY(lang, "internal wordlist");
|
|
|
3616 else
|
|
|
3617 {
|
|
|
3618 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
3619 p = vim_strchr(lang, '.');
|
|
|
3620 if (p != NULL)
|
|
|
3621 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
3622 }
|
|
351
|
3623 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
3624 }
|
|
|
3625 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
3626 {
|
|
389
|
3627 region_mask = REGION_ALL;
|
|
|
3628 if (use_region != NULL && !dont_use_region)
|
|
|
3629 {
|
|
|
3630 /* find region in sl_regions */
|
|
|
3631 c = find_region(lp->sl_regions, use_region);
|
|
|
3632 if (c != REGION_ALL)
|
|
|
3633 region_mask = 1 << c;
|
|
|
3634 else if (*lp->sl_regions != NUL)
|
|
|
3635 /* This spell file is for other regions. */
|
|
|
3636 region_mask = 0;
|
|
|
3637 }
|
|
|
3638
|
|
|
3639 if (region_mask != 0)
|
|
|
3640 {
|
|
|
3641 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
3642 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3643 ++ga.ga_len;
|
|
|
3644 use_midword(lp, buf);
|
|
|
3645 }
|
|
323
|
3646 }
|
|
|
3647 }
|
|
|
3648
|
|
236
|
3649 /* Add a NULL entry to mark the end of the list. */
|
|
|
3650 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3651 {
|
|
|
3652 ga_clear(&ga);
|
|
|
3653 return e_outofmem;
|
|
|
3654 }
|
|
|
3655 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
3656 ++ga.ga_len;
|
|
|
3657
|
|
|
3658 /* Everything is fine, store the new b_langp value. */
|
|
|
3659 ga_clear(&buf->b_langp);
|
|
|
3660 buf->b_langp = ga;
|
|
|
3661
|
|
|
3662 return NULL;
|
|
|
3663 }
|
|
|
3664
|
|
|
3665 /*
|
|
376
|
3666 * Clear the midword characters for buffer "buf".
|
|
|
3667 */
|
|
|
3668 static void
|
|
|
3669 clear_midword(buf)
|
|
|
3670 buf_T *buf;
|
|
|
3671 {
|
|
|
3672 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
3673 #ifdef FEAT_MBYTE
|
|
|
3674 vim_free(buf->b_spell_ismw_mb);
|
|
|
3675 buf->b_spell_ismw_mb = NULL;
|
|
|
3676 #endif
|
|
|
3677 }
|
|
|
3678
|
|
|
3679 /*
|
|
|
3680 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
3681 * They add up to any currently used midword characters.
|
|
|
3682 */
|
|
|
3683 static void
|
|
|
3684 use_midword(lp, buf)
|
|
|
3685 slang_T *lp;
|
|
|
3686 buf_T *buf;
|
|
|
3687 {
|
|
|
3688 char_u *p;
|
|
|
3689
|
|
389
|
3690 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
3691 return;
|
|
|
3692
|
|
376
|
3693 for (p = lp->sl_midword; *p != NUL; )
|
|
|
3694 #ifdef FEAT_MBYTE
|
|
|
3695 if (has_mbyte)
|
|
|
3696 {
|
|
|
3697 int c, l, n;
|
|
|
3698 char_u *bp;
|
|
|
3699
|
|
|
3700 c = mb_ptr2char(p);
|
|
474
|
3701 l = (*mb_ptr2len)(p);
|
|
|
3702 if (c < 256 && l <= 2)
|
|
376
|
3703 buf->b_spell_ismw[c] = TRUE;
|
|
|
3704 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
3705 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
3706 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
3707 else
|
|
|
3708 {
|
|
|
3709 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
3710 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
3711 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
3712 if (bp != NULL)
|
|
|
3713 {
|
|
|
3714 vim_free(buf->b_spell_ismw_mb);
|
|
|
3715 buf->b_spell_ismw_mb = bp;
|
|
|
3716 vim_strncpy(bp + n, p, l);
|
|
|
3717 }
|
|
|
3718 }
|
|
|
3719 p += l;
|
|
|
3720 }
|
|
|
3721 else
|
|
|
3722 #endif
|
|
|
3723 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
3724 }
|
|
|
3725
|
|
|
3726 /*
|
|
236
|
3727 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
3728 * Each region is simply stored as the two characters of it's name.
|
|
381
|
3729 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
3730 */
|
|
|
3731 static int
|
|
|
3732 find_region(rp, region)
|
|
|
3733 char_u *rp;
|
|
|
3734 char_u *region;
|
|
|
3735 {
|
|
|
3736 int i;
|
|
|
3737
|
|
|
3738 for (i = 0; ; i += 2)
|
|
|
3739 {
|
|
|
3740 if (rp[i] == NUL)
|
|
|
3741 return REGION_ALL;
|
|
|
3742 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
3743 break;
|
|
|
3744 }
|
|
|
3745 return i / 2;
|
|
|
3746 }
|
|
|
3747
|
|
|
3748 /*
|
|
323
|
3749 * Return case type of word:
|
|
236
|
3750 * w word 0
|
|
300
|
3751 * Word WF_ONECAP
|
|
|
3752 * W WORD WF_ALLCAP
|
|
|
3753 * WoRd wOrd WF_KEEPCAP
|
|
236
|
3754 */
|
|
|
3755 static int
|
|
|
3756 captype(word, end)
|
|
|
3757 char_u *word;
|
|
323
|
3758 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
3759 {
|
|
|
3760 char_u *p;
|
|
|
3761 int c;
|
|
|
3762 int firstcap;
|
|
|
3763 int allcap;
|
|
|
3764 int past_second = FALSE; /* past second word char */
|
|
|
3765
|
|
|
3766 /* find first letter */
|
|
376
|
3767 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
3768 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
3769 return 0; /* only non-word characters, illegal word */
|
|
|
3770 #ifdef FEAT_MBYTE
|
|
310
|
3771 if (has_mbyte)
|
|
|
3772 c = mb_ptr2char_adv(&p);
|
|
|
3773 else
|
|
236
|
3774 #endif
|
|
310
|
3775 c = *p++;
|
|
324
|
3776 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
3777
|
|
|
3778 /*
|
|
|
3779 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
3780 * But a word with an upper char only at start is a ONECAP.
|
|
|
3781 */
|
|
323
|
3782 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
3783 if (spell_iswordp_nmw(p))
|
|
236
|
3784 {
|
|
455
|
3785 c = PTR2CHAR(p);
|
|
324
|
3786 if (!SPELL_ISUPPER(c))
|
|
236
|
3787 {
|
|
|
3788 /* UUl -> KEEPCAP */
|
|
|
3789 if (past_second && allcap)
|
|
300
|
3790 return WF_KEEPCAP;
|
|
236
|
3791 allcap = FALSE;
|
|
|
3792 }
|
|
|
3793 else if (!allcap)
|
|
|
3794 /* UlU -> KEEPCAP */
|
|
300
|
3795 return WF_KEEPCAP;
|
|
236
|
3796 past_second = TRUE;
|
|
|
3797 }
|
|
|
3798
|
|
|
3799 if (allcap)
|
|
300
|
3800 return WF_ALLCAP;
|
|
236
|
3801 if (firstcap)
|
|
300
|
3802 return WF_ONECAP;
|
|
236
|
3803 return 0;
|
|
|
3804 }
|
|
|
3805
|
|
474
|
3806 /*
|
|
|
3807 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
3808 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
3809 * Add ALLCAP for "WOrD".
|
|
|
3810 */
|
|
|
3811 static int
|
|
|
3812 badword_captype(word, end)
|
|
|
3813 char_u *word;
|
|
|
3814 char_u *end;
|
|
|
3815 {
|
|
|
3816 int flags = captype(word, end);
|
|
475
|
3817 int c;
|
|
474
|
3818 int l, u;
|
|
|
3819 int first;
|
|
|
3820 char_u *p;
|
|
|
3821
|
|
|
3822 if (flags & WF_KEEPCAP)
|
|
|
3823 {
|
|
|
3824 /* Count the number of UPPER and lower case letters. */
|
|
|
3825 l = u = 0;
|
|
|
3826 first = FALSE;
|
|
|
3827 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
3828 {
|
|
475
|
3829 c = PTR2CHAR(p);
|
|
|
3830 if (SPELL_ISUPPER(c))
|
|
474
|
3831 {
|
|
|
3832 ++u;
|
|
|
3833 if (p == word)
|
|
|
3834 first = TRUE;
|
|
|
3835 }
|
|
|
3836 else
|
|
|
3837 ++l;
|
|
|
3838 }
|
|
|
3839
|
|
|
3840 /* If there are more UPPER than lower case letters suggest an
|
|
|
3841 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
3842 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
3843 * require three upper case letters. */
|
|
|
3844 if (u > l && u > 2)
|
|
|
3845 flags |= WF_ALLCAP;
|
|
|
3846 else if (first)
|
|
|
3847 flags |= WF_ONECAP;
|
|
|
3848 }
|
|
|
3849 return flags;
|
|
|
3850 }
|
|
|
3851
|
|
355
|
3852 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
3853 /*
|
|
|
3854 * Free all languages.
|
|
|
3855 */
|
|
|
3856 void
|
|
|
3857 spell_free_all()
|
|
|
3858 {
|
|
|
3859 slang_T *lp;
|
|
|
3860 buf_T *buf;
|
|
385
|
3861 char_u fname[MAXPATHL];
|
|
355
|
3862
|
|
|
3863 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3864 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3865 ga_clear(&buf->b_langp);
|
|
|
3866
|
|
|
3867 while (first_lang != NULL)
|
|
|
3868 {
|
|
|
3869 lp = first_lang;
|
|
|
3870 first_lang = lp->sl_next;
|
|
|
3871 slang_free(lp);
|
|
|
3872 }
|
|
366
|
3873
|
|
385
|
3874 if (int_wordlist != NULL)
|
|
|
3875 {
|
|
|
3876 /* Delete the internal wordlist and its .spl file */
|
|
|
3877 mch_remove(int_wordlist);
|
|
|
3878 int_wordlist_spl(fname);
|
|
|
3879 mch_remove(fname);
|
|
|
3880 vim_free(int_wordlist);
|
|
|
3881 int_wordlist = NULL;
|
|
381
|
3882 }
|
|
|
3883
|
|
366
|
3884 init_spell_chartab();
|
|
355
|
3885 }
|
|
|
3886 # endif
|
|
|
3887
|
|
236
|
3888 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
3889 /*
|
|
|
3890 * Clear all spelling tables and reload them.
|
|
307
|
3891 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
3892 */
|
|
|
3893 void
|
|
|
3894 spell_reload()
|
|
|
3895 {
|
|
|
3896 buf_T *buf;
|
|
316
|
3897 win_T *wp;
|
|
236
|
3898
|
|
358
|
3899 /* Initialize the table for spell_iswordp(). */
|
|
236
|
3900 init_spell_chartab();
|
|
|
3901
|
|
|
3902 /* Unload all allocated memory. */
|
|
355
|
3903 spell_free_all();
|
|
236
|
3904
|
|
|
3905 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3906 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3907 {
|
|
316
|
3908 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
3909 * window for this buffer in which 'spell' is set. */
|
|
236
|
3910 if (*buf->b_p_spl != NUL)
|
|
316
|
3911 {
|
|
|
3912 FOR_ALL_WINDOWS(wp)
|
|
|
3913 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
3914 {
|
|
|
3915 (void)did_set_spelllang(buf);
|
|
|
3916 # ifdef FEAT_WINDOWS
|
|
|
3917 break;
|
|
|
3918 # endif
|
|
|
3919 }
|
|
|
3920 }
|
|
236
|
3921 }
|
|
|
3922 }
|
|
|
3923 # endif
|
|
|
3924
|
|
310
|
3925 /*
|
|
|
3926 * Reload the spell file "fname" if it's loaded.
|
|
|
3927 */
|
|
|
3928 static void
|
|
323
|
3929 spell_reload_one(fname, added_word)
|
|
310
|
3930 char_u *fname;
|
|
323
|
3931 int added_word; /* invoked through "zg" */
|
|
310
|
3932 {
|
|
|
3933 slang_T *lp;
|
|
323
|
3934 int didit = FALSE;
|
|
310
|
3935
|
|
|
3936 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3937 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3938 {
|
|
|
3939 slang_clear(lp);
|
|
323
|
3940 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
3941 redraw_all_later(NOT_VALID);
|
|
323
|
3942 didit = TRUE;
|
|
310
|
3943 }
|
|
323
|
3944
|
|
|
3945 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
3946 * 'spelllang' to get it loaded. */
|
|
|
3947 if (added_word && !didit)
|
|
|
3948 did_set_spelllang(curbuf);
|
|
310
|
3949 }
|
|
|
3950
|
|
|
3951
|
|
236
|
3952 /*
|
|
|
3953 * Functions for ":mkspell".
|
|
|
3954 */
|
|
|
3955
|
|
300
|
3956 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
3957 and .dic file. */
|
|
|
3958 /*
|
|
|
3959 * Main structure to store the contents of a ".aff" file.
|
|
|
3960 */
|
|
|
3961 typedef struct afffile_S
|
|
|
3962 {
|
|
|
3963 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
498
|
3964 int af_flagtype; /* AFT_CHAR, AFT_LONG, AFT_NUM or AFT_CAPLONG */
|
|
481
|
3965 int af_slash; /* character used in word for slash */
|
|
495
|
3966 unsigned af_rar; /* RAR ID for rare word */
|
|
|
3967 unsigned af_kep; /* KEP ID for keep-case word */
|
|
|
3968 unsigned af_bad; /* BAD ID for banned word */
|
|
|
3969 unsigned af_needaffix; /* NEEDAFFIX ID */
|
|
339
|
3970 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
3971 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
3972 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
495
|
3973 hashtab_T af_comp; /* hashtable for compound flags, compitem_T */
|
|
236
|
3974 } afffile_T;
|
|
|
3975
|
|
495
|
3976 #define AFT_CHAR 0 /* flags are one character */
|
|
498
|
3977 #define AFT_LONG 1 /* flags are two characters */
|
|
|
3978 #define AFT_CAPLONG 2 /* flags are one or two characters */
|
|
|
3979 #define AFT_NUM 3 /* flags are numbers, comma separated */
|
|
495
|
3980
|
|
236
|
3981 typedef struct affentry_S affentry_T;
|
|
|
3982 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
3983 struct affentry_S
|
|
|
3984 {
|
|
|
3985 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
3986 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3987 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3988 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3989 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
483
|
3990 char_u ae_rare; /* rare affix */
|
|
|
3991 char_u ae_nocomp; /* word with affix not compoundable */
|
|
300
|
3992 };
|
|
|
3993
|
|
495
|
3994 #ifdef FEAT_MBYTE
|
|
|
3995 # define AH_KEY_LEN 17 /* 2 x 8 bytes + NUL */
|
|
|
3996 #else
|
|
498
|
3997 # define AH_KEY_LEN 7 /* 6 digits + NUL */
|
|
495
|
3998 #endif
|
|
455
|
3999
|
|
300
|
4000 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
4001 typedef struct affheader_S
|
|
|
4002 {
|
|
495
|
4003 char_u ah_key[AH_KEY_LEN]; /* key for hashtab == name of affix */
|
|
|
4004 unsigned ah_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4005 int ah_newID; /* prefix ID after renumbering; 0 if not used */
|
|
300
|
4006 int ah_combine; /* suffix may combine with prefix */
|
|
498
|
4007 int ah_follows; /* another affix block should be following */
|
|
300
|
4008 affentry_T *ah_first; /* first affix entry */
|
|
|
4009 } affheader_T;
|
|
|
4010
|
|
|
4011 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
4012
|
|
495
|
4013 /* Flag used in compound items. */
|
|
|
4014 typedef struct compitem_S
|
|
|
4015 {
|
|
|
4016 char_u ci_key[AH_KEY_LEN]; /* key for hashtab == name of compound */
|
|
|
4017 unsigned ci_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4018 int ci_newID; /* affix ID after renumbering. */
|
|
|
4019 } compitem_T;
|
|
|
4020
|
|
|
4021 #define HI2CI(hi) ((compitem_T *)(hi)->hi_key)
|
|
|
4022
|
|
300
|
4023 /*
|
|
|
4024 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
4025 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
4026 * once after ":mkspell" is done.
|
|
300
|
4027 */
|
|
|
4028 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
4029 typedef struct sblock_S sblock_T;
|
|
|
4030 struct sblock_S
|
|
|
4031 {
|
|
|
4032 sblock_T *sb_next; /* next block in list */
|
|
|
4033 int sb_used; /* nr of bytes already in use */
|
|
|
4034 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
4035 };
|
|
|
4036
|
|
|
4037 /*
|
|
300
|
4038 * A node in the tree.
|
|
236
|
4039 */
|
|
300
|
4040 typedef struct wordnode_S wordnode_T;
|
|
|
4041 struct wordnode_S
|
|
236
|
4042 {
|
|
346
|
4043 union /* shared to save space */
|
|
|
4044 {
|
|
469
|
4045 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
4046 int index; /* index in written nodes (valid after first
|
|
|
4047 round) */
|
|
|
4048 } wn_u1;
|
|
|
4049 union /* shared to save space */
|
|
|
4050 {
|
|
|
4051 wordnode_T *next; /* next node with same hash key */
|
|
|
4052 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
4053 } wn_u2;
|
|
300
|
4054 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
4055 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
4056 always sorted) */
|
|
474
|
4057 int wn_refs; /* Nr. of references to this node. Only
|
|
|
4058 relevant for first node in a list of
|
|
|
4059 siblings, in following siblings it is
|
|
|
4060 always one. */
|
|
300
|
4061 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
481
|
4062 char_u wn_affixID; /* when "wn_byte" is NUL: supported/required
|
|
469
|
4063 prefix ID or 0 */
|
|
|
4064 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
4065 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
4066 PREFIXTREE it's the prefcondnr */
|
|
469
|
4067 #ifdef SPELL_PRINTTREE
|
|
|
4068 int wn_nr; /* sequence nr for printing */
|
|
|
4069 #endif
|
|
236
|
4070 };
|
|
|
4071
|
|
390
|
4072 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
4073
|
|
300
|
4074 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
4075
|
|
300
|
4076 /*
|
|
|
4077 * Info used while reading the spell files.
|
|
|
4078 */
|
|
|
4079 typedef struct spellinfo_S
|
|
249
|
4080 {
|
|
300
|
4081 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
4082 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
4083
|
|
300
|
4084 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
4085 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
4086
|
|
339
|
4087 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
4088
|
|
300
|
4089 sblock_T *si_blocks; /* memory blocks used */
|
|
482
|
4090 long si_blocks_cnt; /* memory blocks allocated */
|
|
|
4091 long si_compress_cnt; /* words to add before lowering
|
|
|
4092 compression limit */
|
|
474
|
4093 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
4094 compression, linked by "wn_child" field. */
|
|
482
|
4095 long si_free_count; /* number of nodes in si_first_free */
|
|
474
|
4096 #ifdef SPELL_PRINTTREE
|
|
|
4097 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
4098 #endif
|
|
|
4099
|
|
|
4100
|
|
300
|
4101 int si_ascii; /* handling only ASCII words */
|
|
310
|
4102 int si_add; /* addition file */
|
|
351
|
4103 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
4104 int si_region; /* region mask */
|
|
|
4105 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
4106 int si_memtot; /* runtime memory used */
|
|
310
|
4107 int si_verbose; /* verbose messages */
|
|
474
|
4108 int si_msg_count; /* number of words added since last message */
|
|
316
|
4109 int si_region_count; /* number of regions supported (1 when there
|
|
|
4110 are no regions) */
|
|
483
|
4111 char_u si_region_name[16]; /* region names; used only if
|
|
|
4112 * si_region_count > 1) */
|
|
323
|
4113
|
|
|
4114 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
4115 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
4116 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
4117 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
4118 int si_followup; /* soundsalike: ? */
|
|
|
4119 int si_collapse; /* soundsalike: ? */
|
|
|
4120 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
4121 garray_T si_map; /* MAP info concatenated */
|
|
495
|
4122 char_u *si_midword; /* MIDWORD chars or NULL */
|
|
483
|
4123 int si_compmax; /* max nr of words for compounding */
|
|
481
|
4124 int si_compminlen; /* minimal length for compounding */
|
|
483
|
4125 int si_compsylmax; /* max nr of syllables for compounding */
|
|
481
|
4126 char_u *si_compflags; /* flags used for compounding */
|
|
492
|
4127 char_u si_nobreak; /* NOBREAK */
|
|
483
|
4128 char_u *si_syllable; /* syllable string */
|
|
339
|
4129 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
4130 * prefixes, each stored as a string */
|
|
495
|
4131 int si_newprefID; /* current value for ah_newID */
|
|
|
4132 int si_compID; /* current value for compound ID */
|
|
300
|
4133 } spellinfo_T;
|
|
249
|
4134
|
|
474
|
4135 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
495
|
4136 static unsigned affitem2flag __ARGS((int flagtype, char_u *item, char_u *fname, int lnum));
|
|
|
4137 static unsigned get_affitem __ARGS((int flagtype, char_u **pp));
|
|
|
4138 static void process_compflags __ARGS((spellinfo_T *spin, afffile_T *aff, char_u *compflags));
|
|
|
4139 static int flag_in_afflist __ARGS((int flagtype, char_u *afflist, unsigned flag));
|
|
|
4140 static void aff_check_number __ARGS((int spinval, int affval, char *name));
|
|
|
4141 static void aff_check_string __ARGS((char_u *spinval, char_u *affval, char *name));
|
|
339
|
4142 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
4143 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
4144 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
4145 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
4146 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
4147 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
483
|
4148 static int get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
495
|
4149 static void get_compflags __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
483
|
4150 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
|
4151 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
4152 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
4153 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
4154 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
4155 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
483
|
4156 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist, int need_affix));
|
|
481
|
4157 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID));
|
|
474
|
4158 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
|
4159 static void deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
|
4160 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
4161 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
4162 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
4163 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
474
|
4164 static void write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
4165 static void clear_node __ARGS((wordnode_T *node));
|
|
|
4166 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
4167 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
4168 static void init_spellfile __ARGS((void));
|
|
236
|
4169
|
|
455
|
4170 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
4171 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
4172 * Use a negative number with the lower 8 bits zero. */
|
|
|
4173 #define PFX_FLAGS -256
|
|
390
|
4174
|
|
483
|
4175 /*
|
|
|
4176 * Tunable parameters for when the tree is compressed. See 'mkspellmem'.
|
|
|
4177 */
|
|
|
4178 static long compress_start = 30000; /* memory / SBLOCKSIZE */
|
|
|
4179 static long compress_inc = 100; /* memory / SBLOCKSIZE */
|
|
|
4180 static long compress_added = 500000; /* word count */
|
|
|
4181
|
|
469
|
4182 #ifdef SPELL_PRINTTREE
|
|
|
4183 /*
|
|
|
4184 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
4185 * readable format, so that we can see what happens when adding a word and/or
|
|
|
4186 * compressing the tree.
|
|
|
4187 * Based on code from Olaf Seibert.
|
|
|
4188 */
|
|
|
4189 #define PRINTLINESIZE 1000
|
|
|
4190 #define PRINTWIDTH 6
|
|
|
4191
|
|
|
4192 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
4193 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
4194
|
|
|
4195 static char line1[PRINTLINESIZE];
|
|
|
4196 static char line2[PRINTLINESIZE];
|
|
|
4197 static char line3[PRINTLINESIZE];
|
|
|
4198
|
|
|
4199 static void
|
|
|
4200 spell_clear_flags(wordnode_T *node)
|
|
|
4201 {
|
|
|
4202 wordnode_T *np;
|
|
|
4203
|
|
|
4204 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4205 {
|
|
|
4206 np->wn_u1.index = FALSE;
|
|
|
4207 spell_clear_flags(np->wn_child);
|
|
|
4208 }
|
|
|
4209 }
|
|
|
4210
|
|
|
4211 static void
|
|
|
4212 spell_print_node(wordnode_T *node, int depth)
|
|
|
4213 {
|
|
|
4214 if (node->wn_u1.index)
|
|
|
4215 {
|
|
|
4216 /* Done this node before, print the reference. */
|
|
|
4217 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
4218 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
4219 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4220 msg(line1);
|
|
|
4221 msg(line2);
|
|
|
4222 msg(line3);
|
|
|
4223 }
|
|
|
4224 else
|
|
|
4225 {
|
|
|
4226 node->wn_u1.index = TRUE;
|
|
|
4227
|
|
|
4228 if (node->wn_byte != NUL)
|
|
|
4229 {
|
|
|
4230 if (node->wn_child != NULL)
|
|
|
4231 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
4232 else
|
|
|
4233 /* Cannot happen? */
|
|
|
4234 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
4235 }
|
|
|
4236 else
|
|
|
4237 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
4238
|
|
|
4239 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
4240
|
|
|
4241 if (node->wn_sibling != NULL)
|
|
|
4242 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
4243 else
|
|
|
4244 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4245
|
|
|
4246 if (node->wn_byte == NUL)
|
|
|
4247 {
|
|
|
4248 msg(line1);
|
|
|
4249 msg(line2);
|
|
|
4250 msg(line3);
|
|
|
4251 }
|
|
|
4252
|
|
|
4253 /* do the children */
|
|
|
4254 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
4255 spell_print_node(node->wn_child, depth + 1);
|
|
|
4256
|
|
|
4257 /* do the siblings */
|
|
|
4258 if (node->wn_sibling != NULL)
|
|
|
4259 {
|
|
|
4260 /* get rid of all parent details except | */
|
|
|
4261 STRCPY(line1, line3);
|
|
|
4262 STRCPY(line2, line3);
|
|
|
4263 spell_print_node(node->wn_sibling, depth);
|
|
|
4264 }
|
|
|
4265 }
|
|
|
4266 }
|
|
|
4267
|
|
|
4268 static void
|
|
|
4269 spell_print_tree(wordnode_T *root)
|
|
|
4270 {
|
|
|
4271 if (root != NULL)
|
|
|
4272 {
|
|
|
4273 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
4274 * done. */
|
|
|
4275 spell_clear_flags(root);
|
|
|
4276
|
|
|
4277 /* Recursively print the tree. */
|
|
|
4278 spell_print_node(root, 0);
|
|
|
4279 }
|
|
|
4280 }
|
|
|
4281 #endif /* SPELL_PRINTTREE */
|
|
|
4282
|
|
236
|
4283 /*
|
|
323
|
4284 * Read the affix file "fname".
|
|
316
|
4285 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
4286 */
|
|
|
4287 static afffile_T *
|
|
474
|
4288 spell_read_aff(spin, fname)
|
|
|
4289 spellinfo_T *spin;
|
|
236
|
4290 char_u *fname;
|
|
|
4291 {
|
|
|
4292 FILE *fd;
|
|
|
4293 afffile_T *aff;
|
|
|
4294 char_u rline[MAXLINELEN];
|
|
|
4295 char_u *line;
|
|
|
4296 char_u *pc = NULL;
|
|
334
|
4297 #define MAXITEMCNT 7
|
|
|
4298 char_u *(items[MAXITEMCNT]);
|
|
236
|
4299 int itemcnt;
|
|
|
4300 char_u *p;
|
|
|
4301 int lnum = 0;
|
|
|
4302 affheader_T *cur_aff = NULL;
|
|
495
|
4303 int did_postpone_prefix = FALSE;
|
|
236
|
4304 int aff_todo = 0;
|
|
|
4305 hashtab_T *tp;
|
|
255
|
4306 char_u *low = NULL;
|
|
|
4307 char_u *fol = NULL;
|
|
|
4308 char_u *upp = NULL;
|
|
323
|
4309 int do_rep;
|
|
|
4310 int do_sal;
|
|
|
4311 int do_map;
|
|
|
4312 int found_map = FALSE;
|
|
324
|
4313 hashitem_T *hi;
|
|
455
|
4314 int l;
|
|
495
|
4315 int compminlen = 0; /* COMPOUNDMIN value */
|
|
|
4316 int compsylmax = 0; /* COMPOUNDSYLMAX value */
|
|
|
4317 int compmax = 0; /* COMPOUNDMAX value */
|
|
|
4318 char_u *compflags = NULL; /* COMPOUNDFLAG and COMPOUNDFLAGS
|
|
|
4319 concatenated */
|
|
|
4320 char_u *midword = NULL; /* MIDWORD value */
|
|
|
4321 char_u *syllable = NULL; /* SYLLABLE value */
|
|
|
4322 char_u *sofofrom = NULL; /* SOFOFROM value */
|
|
|
4323 char_u *sofoto = NULL; /* SOFOTO value */
|
|
236
|
4324
|
|
300
|
4325 /*
|
|
|
4326 * Open the file.
|
|
|
4327 */
|
|
310
|
4328 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4329 if (fd == NULL)
|
|
|
4330 {
|
|
|
4331 EMSG2(_(e_notopen), fname);
|
|
|
4332 return NULL;
|
|
|
4333 }
|
|
|
4334
|
|
310
|
4335 if (spin->si_verbose || p_verbose > 2)
|
|
|
4336 {
|
|
|
4337 if (!spin->si_verbose)
|
|
|
4338 verbose_enter();
|
|
366
|
4339 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
4340 out_flush();
|
|
|
4341 if (!spin->si_verbose)
|
|
|
4342 verbose_leave();
|
|
|
4343 }
|
|
236
|
4344
|
|
323
|
4345 /* Only do REP lines when not done in another .aff file already. */
|
|
|
4346 do_rep = spin->si_rep.ga_len == 0;
|
|
|
4347
|
|
|
4348 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
4349 do_sal = spin->si_sal.ga_len == 0;
|
|
|
4350
|
|
|
4351 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
4352 do_map = spin->si_map.ga_len == 0;
|
|
|
4353
|
|
300
|
4354 /*
|
|
|
4355 * Allocate and init the afffile_T structure.
|
|
|
4356 */
|
|
474
|
4357 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
4358 if (aff == NULL)
|
|
|
4359 return NULL;
|
|
|
4360 hash_init(&aff->af_pref);
|
|
|
4361 hash_init(&aff->af_suff);
|
|
495
|
4362 hash_init(&aff->af_comp);
|
|
236
|
4363
|
|
|
4364 /*
|
|
|
4365 * Read all the lines in the file one by one.
|
|
|
4366 */
|
|
255
|
4367 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
4368 {
|
|
255
|
4369 line_breakcheck();
|
|
236
|
4370 ++lnum;
|
|
|
4371
|
|
|
4372 /* Skip comment lines. */
|
|
|
4373 if (*rline == '#')
|
|
|
4374 continue;
|
|
|
4375
|
|
|
4376 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
4377 vim_free(pc);
|
|
310
|
4378 #ifdef FEAT_MBYTE
|
|
300
|
4379 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4380 {
|
|
300
|
4381 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
4382 if (pc == NULL)
|
|
|
4383 {
|
|
|
4384 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4385 fname, lnum, rline);
|
|
|
4386 continue;
|
|
|
4387 }
|
|
236
|
4388 line = pc;
|
|
|
4389 }
|
|
|
4390 else
|
|
310
|
4391 #endif
|
|
236
|
4392 {
|
|
|
4393 pc = NULL;
|
|
|
4394 line = rline;
|
|
|
4395 }
|
|
|
4396
|
|
|
4397 /* Split the line up in white separated items. Put a NUL after each
|
|
|
4398 * item. */
|
|
|
4399 itemcnt = 0;
|
|
|
4400 for (p = line; ; )
|
|
|
4401 {
|
|
|
4402 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
4403 ++p;
|
|
|
4404 if (*p == NUL)
|
|
|
4405 break;
|
|
334
|
4406 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
4407 break;
|
|
236
|
4408 items[itemcnt++] = p;
|
|
300
|
4409 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
4410 ++p;
|
|
|
4411 if (*p == NUL)
|
|
|
4412 break;
|
|
|
4413 *p++ = NUL;
|
|
|
4414 }
|
|
|
4415
|
|
|
4416 /* Handle non-empty lines. */
|
|
|
4417 if (itemcnt > 0)
|
|
|
4418 {
|
|
|
4419 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
4420 && aff->af_enc == NULL)
|
|
|
4421 {
|
|
310
|
4422 #ifdef FEAT_MBYTE
|
|
300
|
4423 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
4424 aff->af_enc = enc_canonize(items[1]);
|
|
|
4425 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
4426 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
4427 p_enc) == FAIL)
|
|
|
4428 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
4429 fname, aff->af_enc, p_enc);
|
|
375
|
4430 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4431 #else
|
|
|
4432 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4433 #endif
|
|
236
|
4434 }
|
|
495
|
4435 else if (STRCMP(items[0], "FLAG") == 0 && itemcnt == 2
|
|
|
4436 && aff->af_flagtype == AFT_CHAR)
|
|
|
4437 {
|
|
|
4438 if (STRCMP(items[1], "long") == 0)
|
|
498
|
4439 aff->af_flagtype = AFT_LONG;
|
|
495
|
4440 else if (STRCMP(items[1], "num") == 0)
|
|
498
|
4441 aff->af_flagtype = AFT_NUM;
|
|
|
4442 else if (STRCMP(items[1], "caplong") == 0)
|
|
|
4443 aff->af_flagtype = AFT_CAPLONG;
|
|
495
|
4444 else
|
|
|
4445 smsg((char_u *)_("Invalid value for FLAG in %s line %d: %s"),
|
|
|
4446 fname, lnum, items[1]);
|
|
|
4447 if (aff->af_rar != 0 || aff->af_kep != 0 || aff->af_bad != 0
|
|
|
4448 || aff->af_needaffix != 0 || compflags != NULL
|
|
|
4449 || aff->af_suff.ht_used > 0
|
|
|
4450 || aff->af_pref.ht_used > 0)
|
|
|
4451 smsg((char_u *)_("FLAG after using flags in %s line %d: %s"),
|
|
|
4452 fname, lnum, items[1]);
|
|
|
4453 }
|
|
|
4454 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2
|
|
|
4455 && midword == NULL)
|
|
|
4456 {
|
|
|
4457 midword = getroom_save(spin, items[1]);
|
|
366
|
4458 }
|
|
302
|
4459 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
4460 {
|
|
323
|
4461 /* ignored, we always split */
|
|
302
|
4462 }
|
|
323
|
4463 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
4464 {
|
|
323
|
4465 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
4466 }
|
|
481
|
4467 else if (STRCMP(items[0], "SLASH") == 0 && itemcnt == 2
|
|
|
4468 && aff->af_slash == 0)
|
|
|
4469 {
|
|
|
4470 aff->af_slash = items[1][0];
|
|
|
4471 if (items[1][1] != NUL)
|
|
|
4472 smsg((char_u *)_("Character used for SLASH must be ASCII; in %s line %d: %s"),
|
|
|
4473 fname, lnum, items[1]);
|
|
|
4474 }
|
|
307
|
4475 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
4476 && aff->af_rar == 0)
|
|
|
4477 {
|
|
495
|
4478 aff->af_rar = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4479 fname, lnum);
|
|
307
|
4480 }
|
|
310
|
4481 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
4482 && aff->af_kep == 0)
|
|
307
|
4483 {
|
|
495
|
4484 aff->af_kep = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4485 fname, lnum);
|
|
307
|
4486 }
|
|
346
|
4487 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
4488 && aff->af_bad == 0)
|
|
|
4489 {
|
|
495
|
4490 aff->af_bad = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4491 fname, lnum);
|
|
346
|
4492 }
|
|
483
|
4493 else if (STRCMP(items[0], "NEEDAFFIX") == 0 && itemcnt == 2
|
|
|
4494 && aff->af_needaffix == 0)
|
|
|
4495 {
|
|
495
|
4496 aff->af_needaffix = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4497 fname, lnum);
|
|
483
|
4498 }
|
|
481
|
4499 else if (STRCMP(items[0], "COMPOUNDFLAG") == 0 && itemcnt == 2
|
|
495
|
4500 && compflags == NULL)
|
|
|
4501 {
|
|
|
4502 /* Turn flag "c" into COMPOUNDFLAGS compatible string "c+",
|
|
|
4503 * "Na" into "Na+", "1234" into "1234+". */
|
|
|
4504 p = getroom(spin, STRLEN(items[1]) + 2, FALSE);
|
|
483
|
4505 if (p != NULL)
|
|
|
4506 {
|
|
495
|
4507 STRCPY(p, items[1]);
|
|
|
4508 STRCAT(p, "+");
|
|
|
4509 compflags = p;
|
|
|
4510 }
|
|
481
|
4511 }
|
|
483
|
4512 else if (STRCMP(items[0], "COMPOUNDFLAGS") == 0 && itemcnt == 2)
|
|
|
4513 {
|
|
|
4514 /* Concatenate this string to previously defined ones, using a
|
|
|
4515 * slash to separate them. */
|
|
|
4516 l = STRLEN(items[1]) + 1;
|
|
495
|
4517 if (compflags != NULL)
|
|
|
4518 l += STRLEN(compflags) + 1;
|
|
483
|
4519 p = getroom(spin, l, FALSE);
|
|
|
4520 if (p != NULL)
|
|
|
4521 {
|
|
495
|
4522 if (compflags != NULL)
|
|
483
|
4523 {
|
|
495
|
4524 STRCPY(p, compflags);
|
|
483
|
4525 STRCAT(p, "/");
|
|
|
4526 }
|
|
|
4527 STRCAT(p, items[1]);
|
|
495
|
4528 compflags = p;
|
|
483
|
4529 }
|
|
|
4530 }
|
|
|
4531 else if (STRCMP(items[0], "COMPOUNDMAX") == 0 && itemcnt == 2
|
|
495
|
4532 && compmax == 0)
|
|
|
4533 {
|
|
|
4534 compmax = atoi((char *)items[1]);
|
|
|
4535 if (compmax == 0)
|
|
483
|
4536 smsg((char_u *)_("Wrong COMPOUNDMAX value in %s line %d: %s"),
|
|
|
4537 fname, lnum, items[1]);
|
|
481
|
4538 }
|
|
|
4539 else if (STRCMP(items[0], "COMPOUNDMIN") == 0 && itemcnt == 2
|
|
495
|
4540 && compminlen == 0)
|
|
|
4541 {
|
|
|
4542 compminlen = atoi((char *)items[1]);
|
|
|
4543 if (compminlen == 0)
|
|
481
|
4544 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"),
|
|
|
4545 fname, lnum, items[1]);
|
|
|
4546 }
|
|
483
|
4547 else if (STRCMP(items[0], "COMPOUNDSYLMAX") == 0 && itemcnt == 2
|
|
495
|
4548 && compsylmax == 0)
|
|
|
4549 {
|
|
|
4550 compsylmax = atoi((char *)items[1]);
|
|
|
4551 if (compsylmax == 0)
|
|
483
|
4552 smsg((char_u *)_("Wrong COMPOUNDSYLMAX value in %s line %d: %s"),
|
|
|
4553 fname, lnum, items[1]);
|
|
|
4554 }
|
|
|
4555 else if (STRCMP(items[0], "SYLLABLE") == 0 && itemcnt == 2
|
|
495
|
4556 && syllable == NULL)
|
|
|
4557 {
|
|
|
4558 syllable = getroom_save(spin, items[1]);
|
|
483
|
4559 }
|
|
492
|
4560 else if (STRCMP(items[0], "NOBREAK") == 0 && itemcnt == 1)
|
|
|
4561 {
|
|
|
4562 spin->si_nobreak = TRUE;
|
|
|
4563 }
|
|
339
|
4564 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
4565 {
|
|
|
4566 aff->af_pfxpostpone = TRUE;
|
|
|
4567 }
|
|
236
|
4568 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4569 || STRCMP(items[0], "SFX") == 0)
|
|
|
4570 && aff_todo == 0
|
|
334
|
4571 && itemcnt >= 4)
|
|
236
|
4572 {
|
|
498
|
4573 int lasti = 4;
|
|
|
4574 char_u key[AH_KEY_LEN];
|
|
|
4575
|
|
|
4576 if (*items[0] == 'P')
|
|
|
4577 tp = &aff->af_pref;
|
|
|
4578 else
|
|
|
4579 tp = &aff->af_suff;
|
|
|
4580
|
|
|
4581 /* Myspell allows the same affix name to be used multiple
|
|
|
4582 * times. The affix files that do this have an undocumented
|
|
|
4583 * "S" flag on all but the last block, thus we check for that
|
|
|
4584 * and store it in ah_follows. */
|
|
|
4585 vim_strncpy(key, items[1], AH_KEY_LEN - 1);
|
|
|
4586 hi = hash_find(tp, key);
|
|
|
4587 if (!HASHITEM_EMPTY(hi))
|
|
|
4588 {
|
|
|
4589 cur_aff = HI2AH(hi);
|
|
|
4590 if (cur_aff->ah_combine != (*items[2] == 'Y'))
|
|
|
4591 smsg((char_u *)_("Different combining flag in continued affix block in %s line %d: %s"),
|
|
|
4592 fname, lnum, items[1]);
|
|
|
4593 if (!cur_aff->ah_follows)
|
|
|
4594 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
4595 fname, lnum, items[1]);
|
|
|
4596 }
|
|
|
4597 else
|
|
|
4598 {
|
|
|
4599 /* New affix letter. */
|
|
|
4600 cur_aff = (affheader_T *)getroom(spin,
|
|
|
4601 sizeof(affheader_T), TRUE);
|
|
|
4602 if (cur_aff == NULL)
|
|
|
4603 break;
|
|
|
4604 cur_aff->ah_flag = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4605 fname, lnum);
|
|
|
4606 if (cur_aff->ah_flag == 0 || STRLEN(items[1]) >= AH_KEY_LEN)
|
|
|
4607 break;
|
|
|
4608 if (cur_aff->ah_flag == aff->af_bad
|
|
|
4609 || cur_aff->ah_flag == aff->af_rar
|
|
|
4610 || cur_aff->ah_flag == aff->af_kep
|
|
|
4611 || cur_aff->ah_flag == aff->af_needaffix)
|
|
|
4612 smsg((char_u *)_("Affix also used for BAD/RAR/KEP/NEEDAFFIX in %s line %d: %s"),
|
|
|
4613 fname, lnum, items[1]);
|
|
|
4614 STRCPY(cur_aff->ah_key, items[1]);
|
|
|
4615 hash_add(tp, cur_aff->ah_key);
|
|
|
4616
|
|
|
4617 cur_aff->ah_combine = (*items[2] == 'Y');
|
|
|
4618 }
|
|
|
4619
|
|
|
4620 /* Check for the "S" flag, which apparently means that another
|
|
|
4621 * block with the same affix name is following. */
|
|
|
4622 if (itemcnt > lasti && STRCMP(items[lasti], "S") == 0)
|
|
|
4623 {
|
|
|
4624 ++lasti;
|
|
|
4625 cur_aff->ah_follows = TRUE;
|
|
|
4626 }
|
|
|
4627 else
|
|
|
4628 cur_aff->ah_follows = FALSE;
|
|
|
4629
|
|
334
|
4630 /* Myspell allows extra text after the item, but that might
|
|
|
4631 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
498
|
4632 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
4633 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
4634 fname, lnum, items[4]);
|
|
|
4635
|
|
498
|
4636 if (STRCMP(items[2], "Y") != 0 && STRCMP(items[2], "N") != 0)
|
|
236
|
4637 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
4638 fname, lnum, items[2]);
|
|
339
|
4639
|
|
498
|
4640 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
|
4641 {
|
|
|
4642 if (cur_aff->ah_newID == 0)
|
|
495
|
4643 {
|
|
|
4644 /* Use a new number in the .spl file later, to be able
|
|
|
4645 * to handle multiple .aff files. */
|
|
|
4646 cur_aff->ah_newID = ++spin->si_newprefID;
|
|
|
4647
|
|
|
4648 /* We only really use ah_newID if the prefix is
|
|
|
4649 * postponed. We know that only after handling all
|
|
|
4650 * the items. */
|
|
|
4651 did_postpone_prefix = FALSE;
|
|
|
4652 }
|
|
498
|
4653 else
|
|
|
4654 /* Did use the ID in a previous block. */
|
|
|
4655 did_postpone_prefix = TRUE;
|
|
|
4656 }
|
|
|
4657
|
|
300
|
4658 aff_todo = atoi((char *)items[3]);
|
|
236
|
4659 }
|
|
|
4660 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4661 || STRCMP(items[0], "SFX") == 0)
|
|
|
4662 && aff_todo > 0
|
|
|
4663 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
4664 && itemcnt >= 5)
|
|
236
|
4665 {
|
|
|
4666 affentry_T *aff_entry;
|
|
366
|
4667 int rare = FALSE;
|
|
483
|
4668 int nocomp = FALSE;
|
|
455
|
4669 int upper = FALSE;
|
|
366
|
4670 int lasti = 5;
|
|
|
4671
|
|
483
|
4672 /* Check for "rare" and "nocomp" after the other info. */
|
|
|
4673 while (itemcnt > lasti)
|
|
|
4674 {
|
|
|
4675 if (!rare && STRICMP(items[lasti], "rare") == 0)
|
|
|
4676 {
|
|
|
4677 rare = TRUE;
|
|
|
4678 ++lasti;
|
|
|
4679 }
|
|
|
4680 else if (!nocomp && STRICMP(items[lasti], "nocomp") == 0)
|
|
|
4681 {
|
|
|
4682 nocomp = TRUE;
|
|
|
4683 ++lasti;
|
|
|
4684 }
|
|
|
4685 else
|
|
|
4686 break;
|
|
366
|
4687 }
|
|
236
|
4688
|
|
334
|
4689 /* Myspell allows extra text after the item, but that might
|
|
|
4690 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
4691 if (itemcnt > lasti && *items[lasti] != '#')
|
|
482
|
4692 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]);
|
|
334
|
4693
|
|
236
|
4694 /* New item for an affix letter. */
|
|
|
4695 --aff_todo;
|
|
474
|
4696 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
4697 sizeof(affentry_T), TRUE);
|
|
236
|
4698 if (aff_entry == NULL)
|
|
|
4699 break;
|
|
366
|
4700 aff_entry->ae_rare = rare;
|
|
483
|
4701 aff_entry->ae_nocomp = nocomp;
|
|
240
|
4702
|
|
236
|
4703 if (STRCMP(items[2], "0") != 0)
|
|
474
|
4704 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
4705 if (STRCMP(items[3], "0") != 0)
|
|
474
|
4706 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
4707
|
|
300
|
4708 /* Don't use an affix entry with non-ASCII characters when
|
|
|
4709 * "spin->si_ascii" is TRUE. */
|
|
|
4710 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
4711 || has_non_ascii(aff_entry->ae_add)))
|
|
|
4712 {
|
|
|
4713 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
4714 cur_aff->ah_first = aff_entry;
|
|
300
|
4715
|
|
|
4716 if (STRCMP(items[4], ".") != 0)
|
|
|
4717 {
|
|
|
4718 char_u buf[MAXLINELEN];
|
|
|
4719
|
|
474
|
4720 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
4721 if (*items[0] == 'P')
|
|
|
4722 sprintf((char *)buf, "^%s", items[4]);
|
|
|
4723 else
|
|
|
4724 sprintf((char *)buf, "%s$", items[4]);
|
|
|
4725 aff_entry->ae_prog = vim_regcomp(buf,
|
|
481
|
4726 RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
4727 if (aff_entry->ae_prog == NULL)
|
|
|
4728 smsg((char_u *)_("Broken condition in %s line %d: %s"),
|
|
|
4729 fname, lnum, items[4]);
|
|
300
|
4730 }
|
|
339
|
4731
|
|
|
4732 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
4733 * for the condition. Use an existing one if possible. */
|
|
455
|
4734 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
4735 {
|
|
455
|
4736 /* When the chop string is one lower-case letter and
|
|
|
4737 * the add string ends in the upper-case letter we set
|
|
|
4738 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
4739 * letters from "ae_add". The condition must either
|
|
|
4740 * be empty or start with the same letter. */
|
|
|
4741 if (aff_entry->ae_chop != NULL
|
|
|
4742 && aff_entry->ae_add != NULL
|
|
|
4743 #ifdef FEAT_MBYTE
|
|
474
|
4744 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
4745 aff_entry->ae_chop)] == NUL
|
|
|
4746 #else
|
|
|
4747 && aff_entry->ae_chop[1] == NUL
|
|
|
4748 #endif
|
|
|
4749 )
|
|
339
|
4750 {
|
|
455
|
4751 int c, c_up;
|
|
|
4752
|
|
|
4753 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
4754 c_up = SPELL_TOUPPER(c);
|
|
|
4755 if (c_up != c
|
|
|
4756 && (aff_entry->ae_cond == NULL
|
|
|
4757 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
4758 {
|
|
|
4759 p = aff_entry->ae_add
|
|
|
4760 + STRLEN(aff_entry->ae_add);
|
|
|
4761 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
4762 if (PTR2CHAR(p) == c_up)
|
|
|
4763 {
|
|
|
4764 upper = TRUE;
|
|
|
4765 aff_entry->ae_chop = NULL;
|
|
|
4766 *p = NUL;
|
|
|
4767
|
|
|
4768 /* The condition is matched with the
|
|
|
4769 * actual word, thus must check for the
|
|
|
4770 * upper-case letter. */
|
|
|
4771 if (aff_entry->ae_cond != NULL)
|
|
|
4772 {
|
|
|
4773 char_u buf[MAXLINELEN];
|
|
|
4774 #ifdef FEAT_MBYTE
|
|
|
4775 if (has_mbyte)
|
|
|
4776 {
|
|
|
4777 onecap_copy(items[4], buf, TRUE);
|
|
|
4778 aff_entry->ae_cond = getroom_save(
|
|
474
|
4779 spin, buf);
|
|
455
|
4780 }
|
|
|
4781 else
|
|
|
4782 #endif
|
|
|
4783 *aff_entry->ae_cond = c_up;
|
|
|
4784 if (aff_entry->ae_cond != NULL)
|
|
|
4785 {
|
|
|
4786 sprintf((char *)buf, "^%s",
|
|
339
|
4787 aff_entry->ae_cond);
|
|
455
|
4788 vim_free(aff_entry->ae_prog);
|
|
|
4789 aff_entry->ae_prog = vim_regcomp(
|
|
|
4790 buf, RE_MAGIC + RE_STRING);
|
|
|
4791 }
|
|
|
4792 }
|
|
|
4793 }
|
|
|
4794 }
|
|
339
|
4795 }
|
|
|
4796
|
|
455
|
4797 if (aff_entry->ae_chop == NULL)
|
|
390
|
4798 {
|
|
455
|
4799 int idx;
|
|
|
4800 char_u **pp;
|
|
|
4801 int n;
|
|
|
4802
|
|
495
|
4803 /* Find a previously used condition. */
|
|
455
|
4804 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
4805 --idx)
|
|
|
4806 {
|
|
|
4807 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
4808 if (str_equal(p, aff_entry->ae_cond))
|
|
|
4809 break;
|
|
|
4810 }
|
|
|
4811 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
4812 {
|
|
|
4813 /* Not found, add a new condition. */
|
|
|
4814 idx = spin->si_prefcond.ga_len++;
|
|
|
4815 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
4816 + idx;
|
|
|
4817 if (aff_entry->ae_cond == NULL)
|
|
|
4818 *pp = NULL;
|
|
|
4819 else
|
|
474
|
4820 *pp = getroom_save(spin,
|
|
455
|
4821 aff_entry->ae_cond);
|
|
|
4822 }
|
|
|
4823
|
|
|
4824 /* Add the prefix to the prefix tree. */
|
|
|
4825 if (aff_entry->ae_add == NULL)
|
|
|
4826 p = (char_u *)"";
|
|
390
|
4827 else
|
|
455
|
4828 p = aff_entry->ae_add;
|
|
|
4829 /* PFX_FLAGS is a negative number, so that
|
|
|
4830 * tree_add_word() knows this is the prefix tree. */
|
|
|
4831 n = PFX_FLAGS;
|
|
|
4832 if (rare)
|
|
|
4833 n |= WFP_RARE;
|
|
|
4834 if (!cur_aff->ah_combine)
|
|
|
4835 n |= WFP_NC;
|
|
|
4836 if (upper)
|
|
|
4837 n |= WFP_UP;
|
|
474
|
4838 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
4839 idx, cur_aff->ah_newID);
|
|
495
|
4840 did_postpone_prefix = TRUE;
|
|
|
4841 }
|
|
|
4842
|
|
|
4843 /* Didn't actually use ah_newID, backup si_newprefID. */
|
|
|
4844 if (aff_todo == 0 && !did_postpone_prefix)
|
|
|
4845 {
|
|
|
4846 --spin->si_newprefID;
|
|
|
4847 cur_aff->ah_newID = 0;
|
|
390
|
4848 }
|
|
339
|
4849 }
|
|
240
|
4850 }
|
|
236
|
4851 }
|
|
495
|
4852 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2
|
|
|
4853 && fol == NULL)
|
|
|
4854 {
|
|
|
4855 fol = vim_strsave(items[1]);
|
|
|
4856 }
|
|
|
4857 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2
|
|
|
4858 && low == NULL)
|
|
|
4859 {
|
|
|
4860 low = vim_strsave(items[1]);
|
|
|
4861 }
|
|
|
4862 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2
|
|
|
4863 && upp == NULL)
|
|
|
4864 {
|
|
|
4865 upp = vim_strsave(items[1]);
|
|
255
|
4866 }
|
|
236
|
4867 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
4868 {
|
|
236
|
4869 /* Ignore REP count */;
|
|
323
|
4870 if (!isdigit(*items[1]))
|
|
|
4871 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
4872 fname, lnum);
|
|
|
4873 }
|
|
482
|
4874 else if (STRCMP(items[0], "REP") == 0 && itemcnt >= 3)
|
|
236
|
4875 {
|
|
|
4876 /* REP item */
|
|
482
|
4877 /* Myspell ignores extra arguments, we require it starts with
|
|
|
4878 * # to detect mistakes. */
|
|
|
4879 if (itemcnt > 3 && items[3][0] != '#')
|
|
|
4880 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]);
|
|
323
|
4881 if (do_rep)
|
|
|
4882 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
4883 }
|
|
|
4884 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
4885 {
|
|
|
4886 /* MAP item or count */
|
|
|
4887 if (!found_map)
|
|
|
4888 {
|
|
|
4889 /* First line contains the count. */
|
|
|
4890 found_map = TRUE;
|
|
|
4891 if (!isdigit(*items[1]))
|
|
|
4892 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
4893 fname, lnum);
|
|
|
4894 }
|
|
|
4895 else if (do_map)
|
|
|
4896 {
|
|
346
|
4897 int c;
|
|
|
4898
|
|
|
4899 /* Check that every character appears only once. */
|
|
|
4900 for (p = items[1]; *p != NUL; )
|
|
|
4901 {
|
|
|
4902 #ifdef FEAT_MBYTE
|
|
|
4903 c = mb_ptr2char_adv(&p);
|
|
|
4904 #else
|
|
|
4905 c = *p++;
|
|
|
4906 #endif
|
|
|
4907 if ((spin->si_map.ga_len > 0
|
|
|
4908 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
4909 != NULL)
|
|
|
4910 || vim_strchr(p, c) != NULL)
|
|
|
4911 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
4912 fname, lnum);
|
|
|
4913 }
|
|
|
4914
|
|
323
|
4915 /* We simply concatenate all the MAP strings, separated by
|
|
|
4916 * slashes. */
|
|
|
4917 ga_concat(&spin->si_map, items[1]);
|
|
|
4918 ga_append(&spin->si_map, '/');
|
|
|
4919 }
|
|
|
4920 }
|
|
|
4921 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
4922 {
|
|
|
4923 if (do_sal)
|
|
|
4924 {
|
|
|
4925 /* SAL item (sounds-a-like)
|
|
|
4926 * Either one of the known keys or a from-to pair. */
|
|
|
4927 if (STRCMP(items[1], "followup") == 0)
|
|
|
4928 spin->si_followup = sal_to_bool(items[2]);
|
|
|
4929 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
4930 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
4931 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
4932 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
4933 else
|
|
|
4934 /* when "to" is "_" it means empty */
|
|
|
4935 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
4936 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
4937 : items[2]);
|
|
|
4938 }
|
|
236
|
4939 }
|
|
375
|
4940 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
495
|
4941 && sofofrom == NULL)
|
|
|
4942 {
|
|
|
4943 sofofrom = getroom_save(spin, items[1]);
|
|
375
|
4944 }
|
|
|
4945 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
495
|
4946 && sofoto == NULL)
|
|
|
4947 {
|
|
|
4948 sofoto = getroom_save(spin, items[1]);
|
|
375
|
4949 }
|
|
300
|
4950 else
|
|
481
|
4951 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"),
|
|
236
|
4952 fname, lnum, items[0]);
|
|
|
4953 }
|
|
|
4954 }
|
|
|
4955
|
|
255
|
4956 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
4957 {
|
|
351
|
4958 if (spin->si_clear_chartab)
|
|
|
4959 {
|
|
|
4960 /* Clear the char type tables, don't want to use any of the
|
|
|
4961 * currently used spell properties. */
|
|
|
4962 init_spell_chartab();
|
|
|
4963 spin->si_clear_chartab = FALSE;
|
|
|
4964 }
|
|
|
4965
|
|
316
|
4966 /*
|
|
|
4967 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
4968 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
4969 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
4970 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
4971 */
|
|
|
4972 if (!spin->si_ascii
|
|
|
4973 #ifdef FEAT_MBYTE
|
|
|
4974 && !enc_utf8
|
|
|
4975 #endif
|
|
|
4976 )
|
|
260
|
4977 {
|
|
|
4978 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
4979 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
4980 else
|
|
316
|
4981 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
4982 }
|
|
255
|
4983
|
|
|
4984 vim_free(fol);
|
|
|
4985 vim_free(low);
|
|
|
4986 vim_free(upp);
|
|
|
4987 }
|
|
|
4988
|
|
481
|
4989 /* Use compound specifications of the .aff file for the spell info. */
|
|
495
|
4990 if (compmax != 0)
|
|
|
4991 {
|
|
|
4992 aff_check_number(spin->si_compmax, compmax, "COMPOUNDMAX");
|
|
|
4993 spin->si_compmax = compmax;
|
|
|
4994 }
|
|
|
4995
|
|
|
4996 if (compminlen != 0)
|
|
|
4997 {
|
|
|
4998 aff_check_number(spin->si_compminlen, compminlen, "COMPOUNDMIN");
|
|
|
4999 spin->si_compminlen = compminlen;
|
|
|
5000 }
|
|
|
5001
|
|
|
5002 if (compsylmax != 0)
|
|
|
5003 {
|
|
|
5004 if (syllable == NULL)
|
|
|
5005 smsg((char_u *)_("COMPOUNDSYLMAX used without SYLLABLE"));
|
|
|
5006 aff_check_number(spin->si_compsylmax, compsylmax, "COMPOUNDSYLMAX");
|
|
|
5007 spin->si_compsylmax = compsylmax;
|
|
|
5008 }
|
|
|
5009
|
|
|
5010 if (compflags != NULL)
|
|
|
5011 process_compflags(spin, aff, compflags);
|
|
|
5012
|
|
|
5013 /* Check that we didn't use too many renumbered flags. */
|
|
|
5014 if (spin->si_compID < spin->si_newprefID)
|
|
|
5015 {
|
|
|
5016 if (spin->si_compID == 255)
|
|
|
5017 MSG(_("Too many postponed prefixes"));
|
|
|
5018 else if (spin->si_newprefID == 0)
|
|
|
5019 MSG(_("Too many compound flags"));
|
|
481
|
5020 else
|
|
495
|
5021 MSG(_("Too many posponed prefixes and/or compound flags"));
|
|
|
5022 }
|
|
|
5023
|
|
|
5024 if (syllable != NULL)
|
|
|
5025 {
|
|
|
5026 aff_check_string(spin->si_syllable, syllable, "SYLLABLE");
|
|
|
5027 spin->si_syllable = syllable;
|
|
|
5028 }
|
|
|
5029
|
|
|
5030 if (sofofrom != NULL || sofoto != NULL)
|
|
|
5031 {
|
|
|
5032 if (sofofrom == NULL || sofoto == NULL)
|
|
|
5033 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
5034 sofofrom == NULL ? "FROM" : "TO", fname);
|
|
|
5035 else if (spin->si_sal.ga_len > 0)
|
|
|
5036 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
483
|
5037 else
|
|
495
|
5038 {
|
|
|
5039 aff_check_string(spin->si_sofofr, sofofrom, "SOFOFROM");
|
|
|
5040 aff_check_string(spin->si_sofoto, sofoto, "SOFOTO");
|
|
|
5041 spin->si_sofofr = sofofrom;
|
|
|
5042 spin->si_sofoto = sofoto;
|
|
|
5043 }
|
|
|
5044 }
|
|
|
5045
|
|
|
5046 if (midword != NULL)
|
|
|
5047 {
|
|
|
5048 aff_check_string(spin->si_midword, midword, "MIDWORD");
|
|
|
5049 spin->si_midword = midword;
|
|
483
|
5050 }
|
|
|
5051
|
|
236
|
5052 vim_free(pc);
|
|
|
5053 fclose(fd);
|
|
|
5054 return aff;
|
|
|
5055 }
|
|
|
5056
|
|
|
5057 /*
|
|
495
|
5058 * Turn an affix flag name into a number, according to the FLAG type.
|
|
|
5059 * returns zero for failure.
|
|
|
5060 */
|
|
|
5061 static unsigned
|
|
|
5062 affitem2flag(flagtype, item, fname, lnum)
|
|
|
5063 int flagtype;
|
|
|
5064 char_u *item;
|
|
|
5065 char_u *fname;
|
|
|
5066 int lnum;
|
|
|
5067 {
|
|
|
5068 unsigned res;
|
|
|
5069 char_u *p = item;
|
|
|
5070
|
|
|
5071 res = get_affitem(flagtype, &p);
|
|
|
5072 if (res == 0)
|
|
|
5073 {
|
|
498
|
5074 if (flagtype == AFT_NUM)
|
|
495
|
5075 smsg((char_u *)_("Flag is not a number in %s line %d: %s"),
|
|
|
5076 fname, lnum, item);
|
|
|
5077 else
|
|
|
5078 smsg((char_u *)_("Illegal flag in %s line %d: %s"),
|
|
|
5079 fname, lnum, item);
|
|
|
5080 }
|
|
|
5081 if (*p != NUL)
|
|
|
5082 {
|
|
|
5083 smsg((char_u *)_(e_affname), fname, lnum, item);
|
|
|
5084 return 0;
|
|
|
5085 }
|
|
|
5086
|
|
|
5087 return res;
|
|
|
5088 }
|
|
|
5089
|
|
|
5090 /*
|
|
|
5091 * Get one affix name from "*pp" and advance the pointer.
|
|
|
5092 * Returns zero for an error, still advances the pointer then.
|
|
|
5093 */
|
|
|
5094 static unsigned
|
|
|
5095 get_affitem(flagtype, pp)
|
|
|
5096 int flagtype;
|
|
|
5097 char_u **pp;
|
|
|
5098 {
|
|
|
5099 int res;
|
|
|
5100
|
|
498
|
5101 if (flagtype == AFT_NUM)
|
|
495
|
5102 {
|
|
|
5103 if (!VIM_ISDIGIT(**pp))
|
|
|
5104 {
|
|
498
|
5105 ++*pp; /* always advance, avoid getting stuck */
|
|
495
|
5106 return 0;
|
|
|
5107 }
|
|
|
5108 res = getdigits(pp);
|
|
|
5109 }
|
|
|
5110 else
|
|
|
5111 {
|
|
|
5112 #ifdef FEAT_MBYTE
|
|
|
5113 res = mb_ptr2char_adv(pp);
|
|
|
5114 #else
|
|
|
5115 res = *(*pp)++;
|
|
|
5116 #endif
|
|
498
|
5117 if (flagtype == AFT_LONG || (flagtype == AFT_CAPLONG
|
|
495
|
5118 && res >= 'A' && res <= 'Z'))
|
|
|
5119 {
|
|
|
5120 if (**pp == NUL)
|
|
|
5121 return 0;
|
|
|
5122 #ifdef FEAT_MBYTE
|
|
|
5123 res = mb_ptr2char_adv(pp) + (res << 16);
|
|
|
5124 #else
|
|
|
5125 res = *(*pp)++ + (res << 16);
|
|
|
5126 #endif
|
|
|
5127 }
|
|
|
5128 }
|
|
|
5129 return res;
|
|
|
5130 }
|
|
|
5131
|
|
|
5132 /*
|
|
|
5133 * Process the "compflags" string used in an affix file and append it to
|
|
|
5134 * spin->si_compflags.
|
|
|
5135 * The processing involves changing the affix names to ID numbers, so that
|
|
|
5136 * they fit in one byte.
|
|
|
5137 */
|
|
|
5138 static void
|
|
|
5139 process_compflags(spin, aff, compflags)
|
|
|
5140 spellinfo_T *spin;
|
|
|
5141 afffile_T *aff;
|
|
|
5142 char_u *compflags;
|
|
|
5143 {
|
|
|
5144 char_u *p;
|
|
|
5145 char_u *prevp;
|
|
|
5146 unsigned flag;
|
|
|
5147 compitem_T *ci;
|
|
|
5148 int id;
|
|
|
5149 int len;
|
|
|
5150 char_u *tp;
|
|
|
5151 char_u key[AH_KEY_LEN];
|
|
|
5152 hashitem_T *hi;
|
|
|
5153
|
|
|
5154 /* Make room for the old and the new compflags, concatenated with a / in
|
|
|
5155 * between. Processing it makes it shorter, but we don't know by how
|
|
|
5156 * much, thus allocate the maximum. */
|
|
|
5157 len = STRLEN(compflags) + 1;
|
|
|
5158 if (spin->si_compflags != NULL)
|
|
|
5159 len += STRLEN(spin->si_compflags) + 1;
|
|
|
5160 p = getroom(spin, len, FALSE);
|
|
|
5161 if (p == NULL)
|
|
|
5162 return;
|
|
|
5163 if (spin->si_compflags != NULL)
|
|
|
5164 {
|
|
|
5165 STRCPY(p, spin->si_compflags);
|
|
|
5166 STRCAT(p, "/");
|
|
|
5167 }
|
|
|
5168 else
|
|
|
5169 *p = NUL;
|
|
|
5170 spin->si_compflags = p;
|
|
|
5171 tp = p + STRLEN(p);
|
|
|
5172
|
|
|
5173 for (p = compflags; *p != NUL; )
|
|
|
5174 {
|
|
|
5175 if (vim_strchr((char_u *)"/*+[]", *p) != NULL)
|
|
|
5176 /* Copy non-flag characters directly. */
|
|
|
5177 *tp++ = *p++;
|
|
|
5178 else
|
|
|
5179 {
|
|
|
5180 /* First get the flag number, also checks validity. */
|
|
|
5181 prevp = p;
|
|
|
5182 flag = get_affitem(aff->af_flagtype, &p);
|
|
|
5183 if (flag != 0)
|
|
|
5184 {
|
|
|
5185 /* Find the flag in the hashtable. If it was used before, use
|
|
|
5186 * the existing ID. Otherwise add a new entry. */
|
|
|
5187 vim_strncpy(key, prevp, p - prevp);
|
|
|
5188 hi = hash_find(&aff->af_comp, key);
|
|
|
5189 if (!HASHITEM_EMPTY(hi))
|
|
|
5190 id = HI2CI(hi)->ci_newID;
|
|
|
5191 else
|
|
|
5192 {
|
|
|
5193 ci = (compitem_T *)getroom(spin, sizeof(compitem_T), TRUE);
|
|
|
5194 if (ci == NULL)
|
|
|
5195 break;
|
|
|
5196 STRCPY(ci->ci_key, key);
|
|
|
5197 ci->ci_flag = flag;
|
|
|
5198 /* Avoid using a flag ID that has a special meaning in a
|
|
|
5199 * regexp (also inside []). */
|
|
|
5200 do
|
|
|
5201 {
|
|
|
5202 id = spin->si_compID--;
|
|
|
5203 } while (vim_strchr((char_u *)"/*+[]\\-^", id) != NULL);
|
|
|
5204 ci->ci_newID = id;
|
|
|
5205 hash_add(&aff->af_comp, ci->ci_key);
|
|
|
5206 }
|
|
|
5207 *tp++ = id;
|
|
|
5208 }
|
|
498
|
5209 if (aff->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
5210 ++p;
|
|
|
5211 }
|
|
|
5212 }
|
|
|
5213
|
|
|
5214 *tp = NUL;
|
|
|
5215 }
|
|
|
5216
|
|
|
5217 /*
|
|
|
5218 * Return TRUE if flag "flag" appears in affix list "afflist".
|
|
|
5219 */
|
|
|
5220 static int
|
|
|
5221 flag_in_afflist(flagtype, afflist, flag)
|
|
|
5222 int flagtype;
|
|
|
5223 char_u *afflist;
|
|
|
5224 unsigned flag;
|
|
|
5225 {
|
|
|
5226 char_u *p;
|
|
|
5227 unsigned n;
|
|
|
5228
|
|
|
5229 switch (flagtype)
|
|
|
5230 {
|
|
|
5231 case AFT_CHAR:
|
|
|
5232 return vim_strchr(afflist, flag) != NULL;
|
|
|
5233
|
|
498
|
5234 case AFT_CAPLONG:
|
|
|
5235 case AFT_LONG:
|
|
495
|
5236 for (p = afflist; *p != NUL; )
|
|
|
5237 {
|
|
|
5238 #ifdef FEAT_MBYTE
|
|
|
5239 n = mb_ptr2char_adv(&p);
|
|
|
5240 #else
|
|
|
5241 n = *p++;
|
|
|
5242 #endif
|
|
498
|
5243 if ((flagtype == AFT_LONG || (n >= 'A' && n <= 'Z'))
|
|
495
|
5244 && *p != NUL)
|
|
|
5245 #ifdef FEAT_MBYTE
|
|
|
5246 n = mb_ptr2char_adv(&p) + (n << 16);
|
|
|
5247 #else
|
|
|
5248 n = *p++ + (n << 16);
|
|
|
5249 #endif
|
|
|
5250 if (n == flag)
|
|
|
5251 return TRUE;
|
|
|
5252 }
|
|
|
5253 break;
|
|
|
5254
|
|
498
|
5255 case AFT_NUM:
|
|
495
|
5256 for (p = afflist; *p != NUL; )
|
|
|
5257 {
|
|
|
5258 n = getdigits(&p);
|
|
|
5259 if (n == flag)
|
|
|
5260 return TRUE;
|
|
|
5261 if (*p != NUL) /* skip over comma */
|
|
|
5262 ++p;
|
|
|
5263 }
|
|
|
5264 break;
|
|
|
5265 }
|
|
|
5266 return FALSE;
|
|
|
5267 }
|
|
|
5268
|
|
|
5269 /*
|
|
|
5270 * Give a warning when "spinval" and "affval" numbers are set and not the same.
|
|
|
5271 */
|
|
|
5272 static void
|
|
|
5273 aff_check_number(spinval, affval, name)
|
|
|
5274 int spinval;
|
|
|
5275 int affval;
|
|
|
5276 char *name;
|
|
|
5277 {
|
|
|
5278 if (spinval != 0 && spinval != affval)
|
|
|
5279 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
5280 }
|
|
|
5281
|
|
|
5282 /*
|
|
|
5283 * Give a warning when "spinval" and "affval" strings are set and not the same.
|
|
|
5284 */
|
|
|
5285 static void
|
|
|
5286 aff_check_string(spinval, affval, name)
|
|
|
5287 char_u *spinval;
|
|
|
5288 char_u *affval;
|
|
|
5289 char *name;
|
|
|
5290 {
|
|
|
5291 if (spinval != NULL && STRCMP(spinval, affval) != 0)
|
|
|
5292 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
5293 }
|
|
|
5294
|
|
|
5295 /*
|
|
339
|
5296 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
5297 * NULL as equal.
|
|
|
5298 */
|
|
|
5299 static int
|
|
|
5300 str_equal(s1, s2)
|
|
|
5301 char_u *s1;
|
|
|
5302 char_u *s2;
|
|
|
5303 {
|
|
|
5304 if (s1 == NULL || s2 == NULL)
|
|
|
5305 return s1 == s2;
|
|
|
5306 return STRCMP(s1, s2) == 0;
|
|
|
5307 }
|
|
|
5308
|
|
|
5309 /*
|
|
323
|
5310 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
5311 * They are stored case-folded.
|
|
|
5312 */
|
|
|
5313 static void
|
|
|
5314 add_fromto(spin, gap, from, to)
|
|
|
5315 spellinfo_T *spin;
|
|
|
5316 garray_T *gap;
|
|
|
5317 char_u *from;
|
|
|
5318 char_u *to;
|
|
|
5319 {
|
|
|
5320 fromto_T *ftp;
|
|
|
5321 char_u word[MAXWLEN];
|
|
|
5322
|
|
|
5323 if (ga_grow(gap, 1) == OK)
|
|
|
5324 {
|
|
|
5325 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
5326 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
5327 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
5328 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
5329 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
5330 ++gap->ga_len;
|
|
|
5331 }
|
|
|
5332 }
|
|
|
5333
|
|
|
5334 /*
|
|
|
5335 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
5336 */
|
|
|
5337 static int
|
|
|
5338 sal_to_bool(s)
|
|
|
5339 char_u *s;
|
|
|
5340 {
|
|
|
5341 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
5342 }
|
|
|
5343
|
|
|
5344 /*
|
|
240
|
5345 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
5346 * When "s" is NULL FALSE is returned.
|
|
|
5347 */
|
|
|
5348 static int
|
|
|
5349 has_non_ascii(s)
|
|
|
5350 char_u *s;
|
|
|
5351 {
|
|
|
5352 char_u *p;
|
|
|
5353
|
|
|
5354 if (s != NULL)
|
|
|
5355 for (p = s; *p != NUL; ++p)
|
|
|
5356 if (*p >= 128)
|
|
|
5357 return TRUE;
|
|
|
5358 return FALSE;
|
|
|
5359 }
|
|
|
5360
|
|
|
5361 /*
|
|
236
|
5362 * Free the structure filled by spell_read_aff().
|
|
|
5363 */
|
|
|
5364 static void
|
|
|
5365 spell_free_aff(aff)
|
|
|
5366 afffile_T *aff;
|
|
|
5367 {
|
|
|
5368 hashtab_T *ht;
|
|
|
5369 hashitem_T *hi;
|
|
|
5370 int todo;
|
|
|
5371 affheader_T *ah;
|
|
300
|
5372 affentry_T *ae;
|
|
236
|
5373
|
|
|
5374 vim_free(aff->af_enc);
|
|
|
5375
|
|
339
|
5376 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
5377 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
5378 {
|
|
|
5379 todo = ht->ht_used;
|
|
|
5380 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
5381 {
|
|
|
5382 if (!HASHITEM_EMPTY(hi))
|
|
|
5383 {
|
|
|
5384 --todo;
|
|
|
5385 ah = HI2AH(hi);
|
|
300
|
5386 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
5387 vim_free(ae->ae_prog);
|
|
236
|
5388 }
|
|
|
5389 }
|
|
|
5390 if (ht == &aff->af_suff)
|
|
|
5391 break;
|
|
|
5392 }
|
|
300
|
5393
|
|
236
|
5394 hash_clear(&aff->af_pref);
|
|
|
5395 hash_clear(&aff->af_suff);
|
|
495
|
5396 hash_clear(&aff->af_comp);
|
|
236
|
5397 }
|
|
|
5398
|
|
|
5399 /*
|
|
300
|
5400 * Read dictionary file "fname".
|
|
236
|
5401 * Returns OK or FAIL;
|
|
|
5402 */
|
|
|
5403 static int
|
|
474
|
5404 spell_read_dic(spin, fname, affile)
|
|
|
5405 spellinfo_T *spin;
|
|
236
|
5406 char_u *fname;
|
|
300
|
5407 afffile_T *affile;
|
|
236
|
5408 {
|
|
300
|
5409 hashtab_T ht;
|
|
236
|
5410 char_u line[MAXLINELEN];
|
|
481
|
5411 char_u *p;
|
|
300
|
5412 char_u *afflist;
|
|
483
|
5413 char_u store_afflist[MAXWLEN];
|
|
|
5414 int pfxlen;
|
|
|
5415 int need_affix;
|
|
300
|
5416 char_u *dw;
|
|
236
|
5417 char_u *pc;
|
|
|
5418 char_u *w;
|
|
|
5419 int l;
|
|
|
5420 hash_T hash;
|
|
|
5421 hashitem_T *hi;
|
|
|
5422 FILE *fd;
|
|
|
5423 int lnum = 1;
|
|
300
|
5424 int non_ascii = 0;
|
|
|
5425 int retval = OK;
|
|
|
5426 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
5427 int flags;
|
|
481
|
5428 int duplicate = 0;
|
|
236
|
5429
|
|
300
|
5430 /*
|
|
|
5431 * Open the file.
|
|
|
5432 */
|
|
310
|
5433 fd = mch_fopen((char *)fname, "r");
|
|
236
|
5434 if (fd == NULL)
|
|
|
5435 {
|
|
|
5436 EMSG2(_(e_notopen), fname);
|
|
|
5437 return FAIL;
|
|
|
5438 }
|
|
|
5439
|
|
300
|
5440 /* The hashtable is only used to detect duplicated words. */
|
|
|
5441 hash_init(&ht);
|
|
|
5442
|
|
310
|
5443 if (spin->si_verbose || p_verbose > 2)
|
|
|
5444 {
|
|
|
5445 if (!spin->si_verbose)
|
|
|
5446 verbose_enter();
|
|
366
|
5447 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
5448 out_flush();
|
|
|
5449 if (!spin->si_verbose)
|
|
|
5450 verbose_leave();
|
|
|
5451 }
|
|
236
|
5452
|
|
474
|
5453 /* start with a message for the first line */
|
|
|
5454 spin->si_msg_count = 999999;
|
|
|
5455
|
|
236
|
5456 /* Read and ignore the first line: word count. */
|
|
|
5457 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
5458 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
5459 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
5460
|
|
|
5461 /*
|
|
|
5462 * Read all the lines in the file one by one.
|
|
|
5463 * The words are converted to 'encoding' here, before being added to
|
|
|
5464 * the hashtable.
|
|
|
5465 */
|
|
255
|
5466 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
5467 {
|
|
255
|
5468 line_breakcheck();
|
|
236
|
5469 ++lnum;
|
|
455
|
5470 if (line[0] == '#' || line[0] == '/')
|
|
351
|
5471 continue; /* comment line */
|
|
236
|
5472
|
|
300
|
5473 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
5474 * the word is kept to allow e.g., "et al.". */
|
|
236
|
5475 l = STRLEN(line);
|
|
|
5476 while (l > 0 && line[l - 1] <= ' ')
|
|
|
5477 --l;
|
|
|
5478 if (l == 0)
|
|
|
5479 continue; /* empty line */
|
|
|
5480 line[l] = NUL;
|
|
|
5481
|
|
481
|
5482 /* Find the optional affix names. Replace the SLASH character by a
|
|
|
5483 * slash. */
|
|
|
5484 afflist = NULL;
|
|
|
5485 for (p = line; *p != NUL; mb_ptr_adv(p))
|
|
|
5486 {
|
|
|
5487 if (*p == affile->af_slash)
|
|
|
5488 *p = '/';
|
|
|
5489 else if (*p == '/')
|
|
|
5490 {
|
|
|
5491 *p = NUL;
|
|
|
5492 afflist = p + 1;
|
|
|
5493 break;
|
|
|
5494 }
|
|
|
5495 }
|
|
236
|
5496
|
|
300
|
5497 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
5498 if (spin->si_ascii && has_non_ascii(line))
|
|
|
5499 {
|
|
|
5500 ++non_ascii;
|
|
240
|
5501 continue;
|
|
300
|
5502 }
|
|
240
|
5503
|
|
310
|
5504 #ifdef FEAT_MBYTE
|
|
236
|
5505 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
5506 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
5507 {
|
|
300
|
5508 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
5509 if (pc == NULL)
|
|
|
5510 {
|
|
|
5511 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5512 fname, lnum, line);
|
|
|
5513 continue;
|
|
|
5514 }
|
|
236
|
5515 w = pc;
|
|
|
5516 }
|
|
|
5517 else
|
|
310
|
5518 #endif
|
|
236
|
5519 {
|
|
|
5520 pc = NULL;
|
|
|
5521 w = line;
|
|
|
5522 }
|
|
|
5523
|
|
474
|
5524 /* This takes time, print a message every 10000 words. */
|
|
|
5525 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
5526 {
|
|
|
5527 spin->si_msg_count = 0;
|
|
339
|
5528 vim_snprintf((char *)message, sizeof(message),
|
|
|
5529 _("line %6d, word %6d - %s"),
|
|
|
5530 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
5531 msg_start();
|
|
|
5532 msg_puts_long_attr(message, 0);
|
|
|
5533 msg_clr_eos();
|
|
|
5534 msg_didout = FALSE;
|
|
|
5535 msg_col = 0;
|
|
|
5536 out_flush();
|
|
|
5537 }
|
|
|
5538
|
|
300
|
5539 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
5540 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
5541 if (dw == NULL)
|
|
300
|
5542 retval = FAIL;
|
|
|
5543 vim_free(pc);
|
|
|
5544 if (retval == FAIL)
|
|
236
|
5545 break;
|
|
|
5546
|
|
300
|
5547 hash = hash_hash(dw);
|
|
|
5548 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
5549 if (!HASHITEM_EMPTY(hi))
|
|
481
|
5550 {
|
|
|
5551 if (p_verbose > 0)
|
|
|
5552 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
5553 fname, lnum, dw);
|
|
481
|
5554 else if (duplicate == 0)
|
|
|
5555 smsg((char_u *)_("First duplicate word in %s line %d: %s"),
|
|
|
5556 fname, lnum, dw);
|
|
|
5557 ++duplicate;
|
|
|
5558 }
|
|
236
|
5559 else
|
|
300
|
5560 hash_add_item(&ht, hi, dw, hash);
|
|
|
5561
|
|
307
|
5562 flags = 0;
|
|
483
|
5563 store_afflist[0] = NUL;
|
|
|
5564 pfxlen = 0;
|
|
|
5565 need_affix = FALSE;
|
|
307
|
5566 if (afflist != NULL)
|
|
|
5567 {
|
|
|
5568 /* Check for affix name that stands for keep-case word and stands
|
|
|
5569 * for rare word (if defined). */
|
|
495
|
5570 if (affile->af_kep != 0 && flag_in_afflist(
|
|
|
5571 affile->af_flagtype, afflist, affile->af_kep))
|
|
389
|
5572 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
495
|
5573 if (affile->af_rar != 0 && flag_in_afflist(
|
|
|
5574 affile->af_flagtype, afflist, affile->af_rar))
|
|
307
|
5575 flags |= WF_RARE;
|
|
495
|
5576 if (affile->af_bad != 0 && flag_in_afflist(
|
|
|
5577 affile->af_flagtype, afflist, affile->af_bad))
|
|
346
|
5578 flags |= WF_BANNED;
|
|
495
|
5579 if (affile->af_needaffix != 0 && flag_in_afflist(
|
|
|
5580 affile->af_flagtype, afflist, affile->af_needaffix))
|
|
483
|
5581 need_affix = TRUE;
|
|
339
|
5582
|
|
|
5583 if (affile->af_pfxpostpone)
|
|
|
5584 /* Need to store the list of prefix IDs with the word. */
|
|
483
|
5585 pfxlen = get_pfxlist(affile, afflist, store_afflist);
|
|
|
5586
|
|
|
5587 if (spin->si_compflags != NULL)
|
|
|
5588 /* Need to store the list of compound flags with the word.
|
|
|
5589 * Concatenate them to the list of prefix IDs. */
|
|
495
|
5590 get_compflags(affile, afflist, store_afflist + pfxlen);
|
|
307
|
5591 }
|
|
|
5592
|
|
300
|
5593 /* Add the word to the word tree(s). */
|
|
483
|
5594 if (store_word(spin, dw, flags, spin->si_region,
|
|
|
5595 store_afflist, need_affix) == FAIL)
|
|
300
|
5596 retval = FAIL;
|
|
236
|
5597
|
|
300
|
5598 if (afflist != NULL)
|
|
|
5599 {
|
|
|
5600 /* Find all matching suffixes and add the resulting words.
|
|
|
5601 * Additionally do matching prefixes that combine. */
|
|
474
|
5602 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
5603 &affile->af_suff, &affile->af_pref,
|
|
483
|
5604 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5605 retval = FAIL;
|
|
|
5606
|
|
|
5607 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
5608 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
5609 &affile->af_pref, NULL,
|
|
483
|
5610 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5611 retval = FAIL;
|
|
|
5612 }
|
|
236
|
5613 }
|
|
|
5614
|
|
481
|
5615 if (duplicate > 0)
|
|
|
5616 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname);
|
|
300
|
5617 if (spin->si_ascii && non_ascii > 0)
|
|
481
|
5618 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"),
|
|
|
5619 non_ascii, fname);
|
|
300
|
5620 hash_clear(&ht);
|
|
|
5621
|
|
236
|
5622 fclose(fd);
|
|
300
|
5623 return retval;
|
|
236
|
5624 }
|
|
|
5625
|
|
|
5626 /*
|
|
339
|
5627 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
5628 * Used for PFXPOSTPONE.
|
|
483
|
5629 * Put the resulting flags in "store_afflist[MAXWLEN]" with a terminating NUL
|
|
|
5630 * and return the number of affixes.
|
|
|
5631 */
|
|
|
5632 static int
|
|
|
5633 get_pfxlist(affile, afflist, store_afflist)
|
|
339
|
5634 afffile_T *affile;
|
|
|
5635 char_u *afflist;
|
|
483
|
5636 char_u *store_afflist;
|
|
339
|
5637 {
|
|
|
5638 char_u *p;
|
|
495
|
5639 char_u *prevp;
|
|
483
|
5640 int cnt = 0;
|
|
495
|
5641 int id;
|
|
|
5642 char_u key[AH_KEY_LEN];
|
|
339
|
5643 hashitem_T *hi;
|
|
|
5644
|
|
495
|
5645 for (p = afflist; *p != NUL; )
|
|
|
5646 {
|
|
|
5647 prevp = p;
|
|
|
5648 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
5649 {
|
|
|
5650 /* A flag is a postponed prefix flag if it appears in "af_pref"
|
|
|
5651 * and it's ID is not zero. */
|
|
|
5652 vim_strncpy(key, prevp, p - prevp);
|
|
|
5653 hi = hash_find(&affile->af_pref, key);
|
|
|
5654 if (!HASHITEM_EMPTY(hi))
|
|
|
5655 {
|
|
|
5656 id = HI2AH(hi)->ah_newID;
|
|
|
5657 if (id != 0)
|
|
|
5658 store_afflist[cnt++] = id;
|
|
|
5659 }
|
|
|
5660 }
|
|
498
|
5661 if (affile->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
5662 ++p;
|
|
483
|
5663 }
|
|
|
5664
|
|
|
5665 store_afflist[cnt] = NUL;
|
|
|
5666 return cnt;
|
|
339
|
5667 }
|
|
|
5668
|
|
|
5669 /*
|
|
495
|
5670 * Get the list of compound IDs from the affix list "afflist" that are used
|
|
|
5671 * for compound words.
|
|
483
|
5672 * Puts the flags in "store_afflist[]".
|
|
|
5673 */
|
|
|
5674 static void
|
|
495
|
5675 get_compflags(affile, afflist, store_afflist)
|
|
|
5676 afffile_T *affile;
|
|
481
|
5677 char_u *afflist;
|
|
483
|
5678 char_u *store_afflist;
|
|
481
|
5679 {
|
|
|
5680 char_u *p;
|
|
495
|
5681 char_u *prevp;
|
|
483
|
5682 int cnt = 0;
|
|
495
|
5683 char_u key[AH_KEY_LEN];
|
|
|
5684 hashitem_T *hi;
|
|
|
5685
|
|
|
5686 for (p = afflist; *p != NUL; )
|
|
|
5687 {
|
|
|
5688 prevp = p;
|
|
|
5689 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
5690 {
|
|
|
5691 /* A flag is a compound flag if it appears in "af_comp". */
|
|
|
5692 vim_strncpy(key, prevp, p - prevp);
|
|
|
5693 hi = hash_find(&affile->af_comp, key);
|
|
|
5694 if (!HASHITEM_EMPTY(hi))
|
|
|
5695 store_afflist[cnt++] = HI2CI(hi)->ci_newID;
|
|
|
5696 }
|
|
498
|
5697 if (affile->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
5698 ++p;
|
|
|
5699 }
|
|
483
|
5700
|
|
|
5701 store_afflist[cnt] = NUL;
|
|
481
|
5702 }
|
|
|
5703
|
|
|
5704 /*
|
|
300
|
5705 * Apply affixes to a word and store the resulting words.
|
|
|
5706 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
5707 * prefixes or suffixes.
|
|
|
5708 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
5709 * the resulting words for combining affixes.
|
|
|
5710 *
|
|
|
5711 * Returns FAIL when out of memory.
|
|
236
|
5712 */
|
|
300
|
5713 static int
|
|
483
|
5714 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags,
|
|
|
5715 pfxlist, pfxlen)
|
|
474
|
5716 spellinfo_T *spin; /* spell info */
|
|
300
|
5717 char_u *word; /* basic word start */
|
|
|
5718 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
5719 afffile_T *affile;
|
|
300
|
5720 hashtab_T *ht;
|
|
|
5721 hashtab_T *xht;
|
|
|
5722 int comb; /* only use affixes that combine */
|
|
307
|
5723 int flags; /* flags for the word */
|
|
339
|
5724 char_u *pfxlist; /* list of prefix IDs */
|
|
483
|
5725 int pfxlen; /* nr of flags in "pfxlist" for prefixes, rest
|
|
|
5726 * is compound flags */
|
|
236
|
5727 {
|
|
|
5728 int todo;
|
|
|
5729 hashitem_T *hi;
|
|
300
|
5730 affheader_T *ah;
|
|
|
5731 affentry_T *ae;
|
|
|
5732 regmatch_T regmatch;
|
|
|
5733 char_u newword[MAXWLEN];
|
|
|
5734 int retval = OK;
|
|
|
5735 int i;
|
|
|
5736 char_u *p;
|
|
366
|
5737 int use_flags;
|
|
390
|
5738 char_u *use_pfxlist;
|
|
483
|
5739 char_u pfx_pfxlist[MAXWLEN];
|
|
|
5740 size_t wordlen = STRLEN(word);
|
|
236
|
5741
|
|
300
|
5742 todo = ht->ht_used;
|
|
|
5743 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
5744 {
|
|
|
5745 if (!HASHITEM_EMPTY(hi))
|
|
|
5746 {
|
|
|
5747 --todo;
|
|
300
|
5748 ah = HI2AH(hi);
|
|
236
|
5749
|
|
300
|
5750 /* Check that the affix combines, if required, and that the word
|
|
|
5751 * supports this affix. */
|
|
495
|
5752 if ((!comb || ah->ah_combine) && flag_in_afflist(
|
|
|
5753 affile->af_flagtype, afflist, ah->ah_flag))
|
|
236
|
5754 {
|
|
300
|
5755 /* Loop over all affix entries with this name. */
|
|
|
5756 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
5757 {
|
|
300
|
5758 /* Check the condition. It's not logical to match case
|
|
|
5759 * here, but it is required for compatibility with
|
|
339
|
5760 * Myspell.
|
|
481
|
5761 * Another requirement from Myspell is that the chop
|
|
|
5762 * string is shorter than the word itself.
|
|
339
|
5763 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
5764 * prefixes with a chop string. */
|
|
300
|
5765 regmatch.regprog = ae->ae_prog;
|
|
|
5766 regmatch.rm_ic = FALSE;
|
|
339
|
5767 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
5768 || ae->ae_chop != NULL)
|
|
481
|
5769 && (ae->ae_chop == NULL
|
|
|
5770 || STRLEN(ae->ae_chop) < wordlen)
|
|
339
|
5771 && (ae->ae_prog == NULL
|
|
|
5772 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
5773 {
|
|
|
5774 /* Match. Remove the chop and add the affix. */
|
|
|
5775 if (xht == NULL)
|
|
240
|
5776 {
|
|
300
|
5777 /* prefix: chop/add at the start of the word */
|
|
|
5778 if (ae->ae_add == NULL)
|
|
|
5779 *newword = NUL;
|
|
|
5780 else
|
|
|
5781 STRCPY(newword, ae->ae_add);
|
|
|
5782 p = word;
|
|
|
5783 if (ae->ae_chop != NULL)
|
|
310
|
5784 {
|
|
300
|
5785 /* Skip chop string. */
|
|
310
|
5786 #ifdef FEAT_MBYTE
|
|
|
5787 if (has_mbyte)
|
|
324
|
5788 {
|
|
310
|
5789 i = mb_charlen(ae->ae_chop);
|
|
324
|
5790 for ( ; i > 0; --i)
|
|
|
5791 mb_ptr_adv(p);
|
|
|
5792 }
|
|
310
|
5793 else
|
|
|
5794 #endif
|
|
324
|
5795 p += STRLEN(ae->ae_chop);
|
|
310
|
5796 }
|
|
300
|
5797 STRCAT(newword, p);
|
|
|
5798 }
|
|
|
5799 else
|
|
|
5800 {
|
|
|
5801 /* suffix: chop/add at the end of the word */
|
|
|
5802 STRCPY(newword, word);
|
|
|
5803 if (ae->ae_chop != NULL)
|
|
|
5804 {
|
|
|
5805 /* Remove chop string. */
|
|
|
5806 p = newword + STRLEN(newword);
|
|
376
|
5807 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
5808 for ( ; i > 0; --i)
|
|
300
|
5809 mb_ptr_back(newword, p);
|
|
|
5810 *p = NUL;
|
|
|
5811 }
|
|
|
5812 if (ae->ae_add != NULL)
|
|
|
5813 STRCAT(newword, ae->ae_add);
|
|
240
|
5814 }
|
|
|
5815
|
|
366
|
5816 /* Obey the "rare" flag of the affix. */
|
|
|
5817 if (ae->ae_rare)
|
|
|
5818 use_flags = flags | WF_RARE;
|
|
|
5819 else
|
|
|
5820 use_flags = flags;
|
|
483
|
5821
|
|
|
5822 /* Obey the "nocomp" flag of the affix: don't use the
|
|
|
5823 * compound flags. */
|
|
390
|
5824 use_pfxlist = pfxlist;
|
|
483
|
5825 if (ae->ae_nocomp && pfxlist != NULL)
|
|
|
5826 {
|
|
|
5827 vim_strncpy(pfx_pfxlist, pfxlist, pfxlen);
|
|
|
5828 use_pfxlist = pfx_pfxlist;
|
|
|
5829 }
|
|
390
|
5830
|
|
|
5831 /* When there are postponed prefixes... */
|
|
393
|
5832 if (spin->si_prefroot != NULL
|
|
|
5833 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
5834 {
|
|
|
5835 /* ... add a flag to indicate an affix was used. */
|
|
|
5836 use_flags |= WF_HAS_AFF;
|
|
|
5837
|
|
|
5838 /* ... don't use a prefix list if combining
|
|
483
|
5839 * affixes is not allowed. But do use the
|
|
|
5840 * compound flags after them. */
|
|
|
5841 if ((!ah->ah_combine || comb) && pfxlist != NULL)
|
|
|
5842 use_pfxlist += pfxlen;
|
|
390
|
5843 }
|
|
366
|
5844
|
|
300
|
5845 /* Store the modified word. */
|
|
474
|
5846 if (store_word(spin, newword, use_flags,
|
|
483
|
5847 spin->si_region, use_pfxlist, FALSE) == FAIL)
|
|
300
|
5848 retval = FAIL;
|
|
236
|
5849
|
|
300
|
5850 /* When added a suffix and combining is allowed also
|
|
|
5851 * try adding prefixes additionally. */
|
|
|
5852 if (xht != NULL && ah->ah_combine)
|
|
474
|
5853 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
5854 xht, NULL, TRUE,
|
|
483
|
5855 use_flags, use_pfxlist, pfxlen) == FAIL)
|
|
300
|
5856 retval = FAIL;
|
|
236
|
5857 }
|
|
|
5858 }
|
|
|
5859 }
|
|
|
5860 }
|
|
|
5861 }
|
|
|
5862
|
|
|
5863 return retval;
|
|
|
5864 }
|
|
|
5865
|
|
|
5866 /*
|
|
300
|
5867 * Read a file with a list of words.
|
|
236
|
5868 */
|
|
|
5869 static int
|
|
474
|
5870 spell_read_wordfile(spin, fname)
|
|
|
5871 spellinfo_T *spin;
|
|
300
|
5872 char_u *fname;
|
|
236
|
5873 {
|
|
300
|
5874 FILE *fd;
|
|
|
5875 long lnum = 0;
|
|
|
5876 char_u rline[MAXLINELEN];
|
|
|
5877 char_u *line;
|
|
|
5878 char_u *pc = NULL;
|
|
381
|
5879 char_u *p;
|
|
300
|
5880 int l;
|
|
|
5881 int retval = OK;
|
|
|
5882 int did_word = FALSE;
|
|
|
5883 int non_ascii = 0;
|
|
307
|
5884 int flags;
|
|
316
|
5885 int regionmask;
|
|
236
|
5886
|
|
300
|
5887 /*
|
|
|
5888 * Open the file.
|
|
|
5889 */
|
|
310
|
5890 fd = mch_fopen((char *)fname, "r");
|
|
300
|
5891 if (fd == NULL)
|
|
236
|
5892 {
|
|
300
|
5893 EMSG2(_(e_notopen), fname);
|
|
|
5894 return FAIL;
|
|
236
|
5895 }
|
|
|
5896
|
|
310
|
5897 if (spin->si_verbose || p_verbose > 2)
|
|
|
5898 {
|
|
|
5899 if (!spin->si_verbose)
|
|
|
5900 verbose_enter();
|
|
366
|
5901 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
5902 out_flush();
|
|
|
5903 if (!spin->si_verbose)
|
|
|
5904 verbose_leave();
|
|
|
5905 }
|
|
300
|
5906
|
|
|
5907 /*
|
|
|
5908 * Read all the lines in the file one by one.
|
|
|
5909 */
|
|
|
5910 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
5911 {
|
|
|
5912 line_breakcheck();
|
|
|
5913 ++lnum;
|
|
|
5914
|
|
|
5915 /* Skip comment lines. */
|
|
|
5916 if (*rline == '#')
|
|
|
5917 continue;
|
|
|
5918
|
|
|
5919 /* Remove CR, LF and white space from the end. */
|
|
|
5920 l = STRLEN(rline);
|
|
|
5921 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
5922 --l;
|
|
|
5923 if (l == 0)
|
|
|
5924 continue; /* empty or blank line */
|
|
|
5925 rline[l] = NUL;
|
|
|
5926
|
|
|
5927 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
5928 vim_free(pc);
|
|
310
|
5929 #ifdef FEAT_MBYTE
|
|
300
|
5930 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
5931 {
|
|
|
5932 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
5933 if (pc == NULL)
|
|
|
5934 {
|
|
|
5935 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5936 fname, lnum, rline);
|
|
|
5937 continue;
|
|
|
5938 }
|
|
|
5939 line = pc;
|
|
|
5940 }
|
|
|
5941 else
|
|
310
|
5942 #endif
|
|
300
|
5943 {
|
|
|
5944 pc = NULL;
|
|
|
5945 line = rline;
|
|
|
5946 }
|
|
|
5947
|
|
307
|
5948 if (*line == '/')
|
|
300
|
5949 {
|
|
307
|
5950 ++line;
|
|
|
5951 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
5952 {
|
|
|
5953 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
5954 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
5955 fname, lnum, line - 1);
|
|
300
|
5956 else if (did_word)
|
|
316
|
5957 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
5958 fname, lnum, line - 1);
|
|
300
|
5959 else
|
|
|
5960 {
|
|
310
|
5961 #ifdef FEAT_MBYTE
|
|
|
5962 char_u *enc;
|
|
|
5963
|
|
300
|
5964 /* Setup for conversion to 'encoding'. */
|
|
316
|
5965 line += 10;
|
|
|
5966 enc = enc_canonize(line);
|
|
300
|
5967 if (enc != NULL && !spin->si_ascii
|
|
|
5968 && convert_setup(&spin->si_conv, enc,
|
|
|
5969 p_enc) == FAIL)
|
|
|
5970 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
5971 fname, line, p_enc);
|
|
300
|
5972 vim_free(enc);
|
|
375
|
5973 spin->si_conv.vc_fail = TRUE;
|
|
310
|
5974 #else
|
|
|
5975 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
5976 #endif
|
|
300
|
5977 }
|
|
307
|
5978 continue;
|
|
300
|
5979 }
|
|
307
|
5980
|
|
316
|
5981 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
5982 {
|
|
|
5983 if (spin->si_region_count > 1)
|
|
|
5984 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
5985 fname, lnum, line);
|
|
|
5986 else
|
|
|
5987 {
|
|
|
5988 line += 8;
|
|
|
5989 if (STRLEN(line) > 16)
|
|
|
5990 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
5991 fname, lnum, line);
|
|
|
5992 else
|
|
|
5993 {
|
|
|
5994 spin->si_region_count = STRLEN(line) / 2;
|
|
|
5995 STRCPY(spin->si_region_name, line);
|
|
389
|
5996
|
|
|
5997 /* Adjust the mask for a word valid in all regions. */
|
|
|
5998 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
5999 }
|
|
|
6000 }
|
|
|
6001 continue;
|
|
|
6002 }
|
|
|
6003
|
|
381
|
6004 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
6005 fname, lnum, line - 1);
|
|
|
6006 continue;
|
|
|
6007 }
|
|
|
6008
|
|
|
6009 flags = 0;
|
|
|
6010 regionmask = spin->si_region;
|
|
|
6011
|
|
|
6012 /* Check for flags and region after a slash. */
|
|
|
6013 p = vim_strchr(line, '/');
|
|
|
6014 if (p != NULL)
|
|
|
6015 {
|
|
|
6016 *p++ = NUL;
|
|
|
6017 while (*p != NUL)
|
|
|
6018 {
|
|
|
6019 if (*p == '=') /* keep-case word */
|
|
389
|
6020 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
6021 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
6022 flags |= WF_BANNED;
|
|
|
6023 else if (*p == '?') /* Rare word. */
|
|
|
6024 flags |= WF_RARE;
|
|
|
6025 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
6026 {
|
|
381
|
6027 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
6028 regionmask = 0;
|
|
|
6029 flags |= WF_REGION;
|
|
|
6030
|
|
|
6031 l = *p - '0';
|
|
316
|
6032 if (l > spin->si_region_count)
|
|
|
6033 {
|
|
|
6034 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
6035 fname, lnum, p);
|
|
316
|
6036 break;
|
|
|
6037 }
|
|
|
6038 regionmask |= 1 << (l - 1);
|
|
|
6039 }
|
|
381
|
6040 else
|
|
|
6041 {
|
|
|
6042 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
6043 fname, lnum, p);
|
|
|
6044 break;
|
|
|
6045 }
|
|
|
6046 ++p;
|
|
307
|
6047 }
|
|
300
|
6048 }
|
|
|
6049
|
|
|
6050 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
6051 if (spin->si_ascii && has_non_ascii(line))
|
|
|
6052 {
|
|
|
6053 ++non_ascii;
|
|
|
6054 continue;
|
|
|
6055 }
|
|
|
6056
|
|
|
6057 /* Normal word: store it. */
|
|
483
|
6058 if (store_word(spin, line, flags, regionmask, NULL, FALSE) == FAIL)
|
|
300
|
6059 {
|
|
|
6060 retval = FAIL;
|
|
|
6061 break;
|
|
|
6062 }
|
|
|
6063 did_word = TRUE;
|
|
|
6064 }
|
|
|
6065
|
|
|
6066 vim_free(pc);
|
|
|
6067 fclose(fd);
|
|
|
6068
|
|
310
|
6069 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
6070 {
|
|
|
6071 if (p_verbose > 2)
|
|
|
6072 verbose_enter();
|
|
300
|
6073 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
6074 non_ascii);
|
|
310
|
6075 if (p_verbose > 2)
|
|
|
6076 verbose_leave();
|
|
|
6077 }
|
|
300
|
6078 return retval;
|
|
236
|
6079 }
|
|
|
6080
|
|
|
6081 /*
|
|
300
|
6082 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
6083 * This avoids calling free() for every little struct we use (and keeping
|
|
|
6084 * track of them).
|
|
300
|
6085 * The memory is cleared to all zeros.
|
|
|
6086 * Returns NULL when out of memory.
|
|
|
6087 */
|
|
|
6088 static void *
|
|
474
|
6089 getroom(spin, len, align)
|
|
|
6090 spellinfo_T *spin;
|
|
447
|
6091 size_t len; /* length needed */
|
|
|
6092 int align; /* align for pointer */
|
|
300
|
6093 {
|
|
|
6094 char_u *p;
|
|
474
|
6095 sblock_T *bl = spin->si_blocks;
|
|
300
|
6096
|
|
447
|
6097 if (align && bl != NULL)
|
|
|
6098 /* Round size up for alignment. On some systems structures need to be
|
|
|
6099 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
6100 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
6101 & ~(sizeof(char *) - 1);
|
|
|
6102
|
|
300
|
6103 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
6104 {
|
|
|
6105 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
6106 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
6107 if (bl == NULL)
|
|
|
6108 return NULL;
|
|
474
|
6109 bl->sb_next = spin->si_blocks;
|
|
|
6110 spin->si_blocks = bl;
|
|
300
|
6111 bl->sb_used = 0;
|
|
482
|
6112 ++spin->si_blocks_cnt;
|
|
300
|
6113 }
|
|
|
6114
|
|
|
6115 p = bl->sb_data + bl->sb_used;
|
|
|
6116 bl->sb_used += len;
|
|
|
6117
|
|
|
6118 return p;
|
|
|
6119 }
|
|
|
6120
|
|
|
6121 /*
|
|
|
6122 * Make a copy of a string into memory allocated with getroom().
|
|
|
6123 */
|
|
|
6124 static char_u *
|
|
474
|
6125 getroom_save(spin, s)
|
|
|
6126 spellinfo_T *spin;
|
|
300
|
6127 char_u *s;
|
|
|
6128 {
|
|
|
6129 char_u *sc;
|
|
|
6130
|
|
474
|
6131 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
6132 if (sc != NULL)
|
|
|
6133 STRCPY(sc, s);
|
|
|
6134 return sc;
|
|
|
6135 }
|
|
|
6136
|
|
|
6137
|
|
|
6138 /*
|
|
|
6139 * Free the list of allocated sblock_T.
|
|
236
|
6140 */
|
|
|
6141 static void
|
|
300
|
6142 free_blocks(bl)
|
|
|
6143 sblock_T *bl;
|
|
236
|
6144 {
|
|
300
|
6145 sblock_T *next;
|
|
236
|
6146
|
|
300
|
6147 while (bl != NULL)
|
|
236
|
6148 {
|
|
300
|
6149 next = bl->sb_next;
|
|
|
6150 vim_free(bl);
|
|
|
6151 bl = next;
|
|
236
|
6152 }
|
|
|
6153 }
|
|
|
6154
|
|
|
6155 /*
|
|
300
|
6156 * Allocate the root of a word tree.
|
|
236
|
6157 */
|
|
300
|
6158 static wordnode_T *
|
|
474
|
6159 wordtree_alloc(spin)
|
|
|
6160 spellinfo_T *spin;
|
|
|
6161 {
|
|
|
6162 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
6163 }
|
|
|
6164
|
|
|
6165 /*
|
|
300
|
6166 * Store a word in the tree(s).
|
|
390
|
6167 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
6168 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
6169 * used to find suggestions.
|
|
300
|
6170 * For a keep-case word also store it in the keep-case tree.
|
|
482
|
6171 * When "pfxlist" is not NULL store the word for each postponed prefix ID and
|
|
|
6172 * compound flag.
|
|
236
|
6173 */
|
|
|
6174 static int
|
|
483
|
6175 store_word(spin, word, flags, region, pfxlist, need_affix)
|
|
474
|
6176 spellinfo_T *spin;
|
|
300
|
6177 char_u *word;
|
|
307
|
6178 int flags; /* extra flags, WF_BANNED */
|
|
316
|
6179 int region; /* supported region(s) */
|
|
339
|
6180 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
483
|
6181 int need_affix; /* only store word with affix ID */
|
|
236
|
6182 {
|
|
300
|
6183 int len = STRLEN(word);
|
|
|
6184 int ct = captype(word, word + len);
|
|
|
6185 char_u foldword[MAXWLEN];
|
|
339
|
6186 int res = OK;
|
|
|
6187 char_u *p;
|
|
236
|
6188
|
|
323
|
6189 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
6190 for (p = pfxlist; res == OK; ++p)
|
|
|
6191 {
|
|
483
|
6192 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6193 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
474
|
6194 region, p == NULL ? 0 : *p);
|
|
339
|
6195 if (p == NULL || *p == NUL)
|
|
|
6196 break;
|
|
|
6197 }
|
|
334
|
6198 ++spin->si_foldwcount;
|
|
307
|
6199
|
|
474
|
6200 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
6201 {
|
|
339
|
6202 for (p = pfxlist; res == OK; ++p)
|
|
|
6203 {
|
|
483
|
6204 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6205 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
474
|
6206 region, p == NULL ? 0 : *p);
|
|
339
|
6207 if (p == NULL || *p == NUL)
|
|
|
6208 break;
|
|
|
6209 }
|
|
334
|
6210 ++spin->si_keepwcount;
|
|
|
6211 }
|
|
300
|
6212 return res;
|
|
236
|
6213 }
|
|
|
6214
|
|
|
6215 /*
|
|
300
|
6216 * Add word "word" to a word tree at "root".
|
|
366
|
6217 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
6218 * "rare" and "region" is the condition nr.
|
|
255
|
6219 * Returns FAIL when out of memory.
|
|
236
|
6220 */
|
|
255
|
6221 static int
|
|
481
|
6222 tree_add_word(spin, word, root, flags, region, affixID)
|
|
474
|
6223 spellinfo_T *spin;
|
|
300
|
6224 char_u *word;
|
|
|
6225 wordnode_T *root;
|
|
|
6226 int flags;
|
|
|
6227 int region;
|
|
481
|
6228 int affixID;
|
|
236
|
6229 {
|
|
300
|
6230 wordnode_T *node = root;
|
|
|
6231 wordnode_T *np;
|
|
469
|
6232 wordnode_T *copyp, **copyprev;
|
|
300
|
6233 wordnode_T **prev = NULL;
|
|
|
6234 int i;
|
|
255
|
6235
|
|
300
|
6236 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
6237 for (i = 0; ; ++i)
|
|
255
|
6238 {
|
|
469
|
6239 /* When there is more than one reference to this node we need to make
|
|
|
6240 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
6241 * (we don't optimize for a partly shared list of siblings). */
|
|
|
6242 if (node != NULL && node->wn_refs > 1)
|
|
|
6243 {
|
|
|
6244 --node->wn_refs;
|
|
|
6245 copyprev = prev;
|
|
|
6246 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
6247 {
|
|
|
6248 /* Allocate a new node and copy the info. */
|
|
474
|
6249 np = get_wordnode(spin);
|
|
469
|
6250 if (np == NULL)
|
|
|
6251 return FAIL;
|
|
|
6252 np->wn_child = copyp->wn_child;
|
|
|
6253 if (np->wn_child != NULL)
|
|
|
6254 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
6255 np->wn_byte = copyp->wn_byte;
|
|
|
6256 if (np->wn_byte == NUL)
|
|
|
6257 {
|
|
|
6258 np->wn_flags = copyp->wn_flags;
|
|
|
6259 np->wn_region = copyp->wn_region;
|
|
481
|
6260 np->wn_affixID = copyp->wn_affixID;
|
|
469
|
6261 }
|
|
|
6262
|
|
|
6263 /* Link the new node in the list, there will be one ref. */
|
|
|
6264 np->wn_refs = 1;
|
|
|
6265 *copyprev = np;
|
|
|
6266 copyprev = &np->wn_sibling;
|
|
|
6267
|
|
|
6268 /* Let "node" point to the head of the copied list. */
|
|
|
6269 if (copyp == node)
|
|
|
6270 node = np;
|
|
|
6271 }
|
|
|
6272 }
|
|
|
6273
|
|
300
|
6274 /* Look for the sibling that has the same character. They are sorted
|
|
|
6275 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
6276 * higher byte value. For zero bytes (end of word) the sorting is
|
|
481
|
6277 * done on flags and then on affixID. */
|
|
339
|
6278 while (node != NULL
|
|
|
6279 && (node->wn_byte < word[i]
|
|
|
6280 || (node->wn_byte == NUL
|
|
|
6281 && (flags < 0
|
|
481
|
6282 ? node->wn_affixID < affixID
|
|
390
|
6283 : node->wn_flags < (flags & WN_MASK)
|
|
|
6284 || (node->wn_flags == (flags & WN_MASK)
|
|
481
|
6285 && node->wn_affixID < affixID)))))
|
|
236
|
6286 {
|
|
300
|
6287 prev = &node->wn_sibling;
|
|
|
6288 node = *prev;
|
|
236
|
6289 }
|
|
339
|
6290 if (node == NULL
|
|
|
6291 || node->wn_byte != word[i]
|
|
|
6292 || (word[i] == NUL
|
|
|
6293 && (flags < 0
|
|
390
|
6294 || node->wn_flags != (flags & WN_MASK)
|
|
481
|
6295 || node->wn_affixID != affixID)))
|
|
255
|
6296 {
|
|
300
|
6297 /* Allocate a new node. */
|
|
474
|
6298 np = get_wordnode(spin);
|
|
300
|
6299 if (np == NULL)
|
|
|
6300 return FAIL;
|
|
|
6301 np->wn_byte = word[i];
|
|
469
|
6302
|
|
|
6303 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
6304 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
6305 * make ref count of sibling one (matters when inserting in front
|
|
|
6306 * of the list of siblings). */
|
|
|
6307 if (node == NULL)
|
|
|
6308 np->wn_refs = 1;
|
|
|
6309 else
|
|
|
6310 {
|
|
|
6311 np->wn_refs = node->wn_refs;
|
|
|
6312 node->wn_refs = 1;
|
|
|
6313 }
|
|
300
|
6314 *prev = np;
|
|
|
6315 np->wn_sibling = node;
|
|
|
6316 node = np;
|
|
255
|
6317 }
|
|
300
|
6318
|
|
|
6319 if (word[i] == NUL)
|
|
|
6320 {
|
|
|
6321 node->wn_flags = flags;
|
|
|
6322 node->wn_region |= region;
|
|
481
|
6323 node->wn_affixID = affixID;
|
|
300
|
6324 break;
|
|
|
6325 }
|
|
|
6326 prev = &node->wn_child;
|
|
|
6327 node = *prev;
|
|
255
|
6328 }
|
|
469
|
6329 #ifdef SPELL_PRINTTREE
|
|
|
6330 smsg("Added \"%s\"", word);
|
|
|
6331 spell_print_tree(root->wn_sibling);
|
|
|
6332 #endif
|
|
|
6333
|
|
474
|
6334 /* count nr of words added since last message */
|
|
|
6335 ++spin->si_msg_count;
|
|
|
6336
|
|
482
|
6337 if (spin->si_compress_cnt > 1)
|
|
|
6338 {
|
|
|
6339 if (--spin->si_compress_cnt == 1)
|
|
|
6340 /* Did enough words to lower the block count limit. */
|
|
483
|
6341 spin->si_blocks_cnt += compress_inc;
|
|
482
|
6342 }
|
|
|
6343
|
|
469
|
6344 /*
|
|
482
|
6345 * When we have allocated lots of memory we need to compress the word tree
|
|
|
6346 * to free up some room. But compression is slow, and we might actually
|
|
|
6347 * need that room, thus only compress in the following situations:
|
|
|
6348 * 1. When not compressed before (si_compress_cnt == 0): when using
|
|
483
|
6349 * "compress_start" blocks.
|
|
|
6350 * 2. When compressed before and used "compress_inc" blocks before
|
|
|
6351 * adding "compress_added" words (si_compress_cnt > 1).
|
|
|
6352 * 3. When compressed before, added "compress_added" words
|
|
482
|
6353 * (si_compress_cnt == 1) and the number of free nodes drops below the
|
|
|
6354 * maximum word length.
|
|
469
|
6355 */
|
|
482
|
6356 #ifndef SPELL_PRINTTREE
|
|
|
6357 if (spin->si_compress_cnt == 1
|
|
|
6358 ? spin->si_free_count < MAXWLEN
|
|
483
|
6359 : spin->si_blocks_cnt >= compress_start)
|
|
482
|
6360 #endif
|
|
|
6361 {
|
|
|
6362 /* Decrement the block counter. The effect is that we compress again
|
|
483
|
6363 * when the freed up room has been used and another "compress_inc"
|
|
|
6364 * blocks have been allocated. Unless "compress_added" words have
|
|
482
|
6365 * been added, then the limit is put back again. */
|
|
483
|
6366 spin->si_blocks_cnt -= compress_inc;
|
|
|
6367 spin->si_compress_cnt = compress_added;
|
|
482
|
6368
|
|
474
|
6369 if (spin->si_verbose)
|
|
|
6370 {
|
|
|
6371 msg_start();
|
|
|
6372 msg_puts((char_u *)_(msg_compressing));
|
|
|
6373 msg_clr_eos();
|
|
|
6374 msg_didout = FALSE;
|
|
|
6375 msg_col = 0;
|
|
|
6376 out_flush();
|
|
|
6377 }
|
|
482
|
6378
|
|
|
6379 /* Compress both trees. Either they both have many nodes, which makes
|
|
|
6380 * compression useful, or one of them is small, which means
|
|
|
6381 * compression goes fast. */
|
|
|
6382 wordtree_compress(spin, spin->si_foldroot);
|
|
|
6383 wordtree_compress(spin, spin->si_keeproot);
|
|
469
|
6384 }
|
|
255
|
6385
|
|
|
6386 return OK;
|
|
236
|
6387 }
|
|
|
6388
|
|
469
|
6389 /*
|
|
483
|
6390 * Check the 'mkspellmem' option. Return FAIL if it's wrong.
|
|
|
6391 * Sets "sps_flags".
|
|
|
6392 */
|
|
|
6393 int
|
|
|
6394 spell_check_msm()
|
|
|
6395 {
|
|
|
6396 char_u *p = p_msm;
|
|
|
6397 long start = 0;
|
|
|
6398 long inc = 0;
|
|
|
6399 long added = 0;
|
|
|
6400
|
|
|
6401 if (!VIM_ISDIGIT(*p))
|
|
|
6402 return FAIL;
|
|
|
6403 /* block count = (value * 1024) / SBLOCKSIZE (but avoid overflow)*/
|
|
|
6404 start = (getdigits(&p) * 10) / (SBLOCKSIZE / 102);
|
|
|
6405 if (*p != ',')
|
|
|
6406 return FAIL;
|
|
|
6407 ++p;
|
|
|
6408 if (!VIM_ISDIGIT(*p))
|
|
|
6409 return FAIL;
|
|
|
6410 inc = (getdigits(&p) * 102) / (SBLOCKSIZE / 10);
|
|
|
6411 if (*p != ',')
|
|
|
6412 return FAIL;
|
|
|
6413 ++p;
|
|
|
6414 if (!VIM_ISDIGIT(*p))
|
|
|
6415 return FAIL;
|
|
|
6416 added = getdigits(&p) * 1024;
|
|
|
6417 if (*p != NUL)
|
|
|
6418 return FAIL;
|
|
|
6419
|
|
|
6420 if (start == 0 || inc == 0 || added == 0 || inc > start)
|
|
|
6421 return FAIL;
|
|
|
6422
|
|
|
6423 compress_start = start;
|
|
|
6424 compress_inc = inc;
|
|
|
6425 compress_added = added;
|
|
|
6426 return OK;
|
|
|
6427 }
|
|
|
6428
|
|
|
6429
|
|
|
6430 /*
|
|
469
|
6431 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
6432 * allocate a new one.
|
|
|
6433 */
|
|
|
6434 static wordnode_T *
|
|
474
|
6435 get_wordnode(spin)
|
|
|
6436 spellinfo_T *spin;
|
|
469
|
6437 {
|
|
|
6438 wordnode_T *n;
|
|
|
6439
|
|
474
|
6440 if (spin->si_first_free == NULL)
|
|
|
6441 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
6442 else
|
|
|
6443 {
|
|
474
|
6444 n = spin->si_first_free;
|
|
|
6445 spin->si_first_free = n->wn_child;
|
|
469
|
6446 vim_memset(n, 0, sizeof(wordnode_T));
|
|
482
|
6447 --spin->si_free_count;
|
|
469
|
6448 }
|
|
|
6449 #ifdef SPELL_PRINTTREE
|
|
474
|
6450 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
6451 #endif
|
|
|
6452 return n;
|
|
|
6453 }
|
|
|
6454
|
|
|
6455 /*
|
|
474
|
6456 * Decrement the reference count on a node (which is the head of a list of
|
|
|
6457 * siblings). If the reference count becomes zero free the node and its
|
|
|
6458 * siblings.
|
|
|
6459 */
|
|
|
6460 static void
|
|
|
6461 deref_wordnode(spin, node)
|
|
|
6462 spellinfo_T *spin;
|
|
|
6463 wordnode_T *node;
|
|
|
6464 {
|
|
|
6465 wordnode_T *np;
|
|
|
6466
|
|
|
6467 if (--node->wn_refs == 0)
|
|
|
6468 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6469 {
|
|
|
6470 if (np->wn_child != NULL)
|
|
|
6471 deref_wordnode(spin, np->wn_child);
|
|
|
6472 free_wordnode(spin, np);
|
|
|
6473 }
|
|
|
6474 }
|
|
|
6475
|
|
|
6476 /*
|
|
469
|
6477 * Free a wordnode_T for re-use later.
|
|
474
|
6478 * Only the "wn_child" field becomes invalid.
|
|
469
|
6479 */
|
|
|
6480 static void
|
|
474
|
6481 free_wordnode(spin, n)
|
|
|
6482 spellinfo_T *spin;
|
|
|
6483 wordnode_T *n;
|
|
|
6484 {
|
|
|
6485 n->wn_child = spin->si_first_free;
|
|
|
6486 spin->si_first_free = n;
|
|
482
|
6487 ++spin->si_free_count;
|
|
469
|
6488 }
|
|
|
6489
|
|
236
|
6490 /*
|
|
300
|
6491 * Compress a tree: find tails that are identical and can be shared.
|
|
|
6492 */
|
|
|
6493 static void
|
|
474
|
6494 wordtree_compress(spin, root)
|
|
|
6495 spellinfo_T *spin;
|
|
300
|
6496 wordnode_T *root;
|
|
|
6497 {
|
|
|
6498 hashtab_T ht;
|
|
|
6499 int n;
|
|
|
6500 int tot = 0;
|
|
469
|
6501 int perc;
|
|
300
|
6502
|
|
474
|
6503 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
6504 * start of the tree. */
|
|
|
6505 if (root->wn_sibling != NULL)
|
|
300
|
6506 {
|
|
|
6507 hash_init(&ht);
|
|
474
|
6508 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
6509
|
|
|
6510 #ifndef SPELL_PRINTTREE
|
|
310
|
6511 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
6512 #endif
|
|
310
|
6513 {
|
|
|
6514 if (!spin->si_verbose)
|
|
|
6515 verbose_enter();
|
|
469
|
6516 if (tot > 1000000)
|
|
|
6517 perc = (tot - n) / (tot / 100);
|
|
482
|
6518 else if (tot == 0)
|
|
|
6519 perc = 0;
|
|
469
|
6520 else
|
|
|
6521 perc = (tot - n) * 100 / tot;
|
|
310
|
6522 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
6523 n, tot, perc);
|
|
310
|
6524 if (p_verbose > 2)
|
|
|
6525 verbose_leave();
|
|
|
6526 }
|
|
469
|
6527 #ifdef SPELL_PRINTTREE
|
|
474
|
6528 spell_print_tree(root->wn_sibling);
|
|
469
|
6529 #endif
|
|
300
|
6530 hash_clear(&ht);
|
|
|
6531 }
|
|
|
6532 }
|
|
|
6533
|
|
|
6534 /*
|
|
|
6535 * Compress a node, its siblings and its children, depth first.
|
|
|
6536 * Returns the number of compressed nodes.
|
|
236
|
6537 */
|
|
255
|
6538 static int
|
|
474
|
6539 node_compress(spin, node, ht, tot)
|
|
|
6540 spellinfo_T *spin;
|
|
300
|
6541 wordnode_T *node;
|
|
|
6542 hashtab_T *ht;
|
|
|
6543 int *tot; /* total count of nodes before compressing,
|
|
|
6544 incremented while going through the tree */
|
|
236
|
6545 {
|
|
300
|
6546 wordnode_T *np;
|
|
|
6547 wordnode_T *tp;
|
|
|
6548 wordnode_T *child;
|
|
|
6549 hash_T hash;
|
|
236
|
6550 hashitem_T *hi;
|
|
300
|
6551 int len = 0;
|
|
|
6552 unsigned nr, n;
|
|
|
6553 int compressed = 0;
|
|
236
|
6554
|
|
300
|
6555 /*
|
|
|
6556 * Go through the list of siblings. Compress each child and then try
|
|
|
6557 * finding an identical child to replace it.
|
|
|
6558 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
6559 * but the whole list of siblings, of which the node is the first.
|
|
|
6560 */
|
|
482
|
6561 for (np = node; np != NULL && !got_int; np = np->wn_sibling)
|
|
236
|
6562 {
|
|
300
|
6563 ++len;
|
|
|
6564 if ((child = np->wn_child) != NULL)
|
|
|
6565 {
|
|
346
|
6566 /* Compress the child. This fills hashkey. */
|
|
474
|
6567 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
6568
|
|
|
6569 /* Try to find an identical child. */
|
|
346
|
6570 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
6571 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
6572 tp = NULL;
|
|
|
6573 if (!HASHITEM_EMPTY(hi))
|
|
|
6574 {
|
|
|
6575 /* There are children with an identical hash value. Now check
|
|
|
6576 * if there is one that is really identical. */
|
|
346
|
6577 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
6578 if (node_equal(child, tp))
|
|
|
6579 {
|
|
|
6580 /* Found one! Now use that child in place of the
|
|
469
|
6581 * current one. This means the current child and all
|
|
|
6582 * its siblings is unlinked from the tree. */
|
|
|
6583 ++tp->wn_refs;
|
|
474
|
6584 deref_wordnode(spin, child);
|
|
300
|
6585 np->wn_child = tp;
|
|
|
6586 ++compressed;
|
|
|
6587 break;
|
|
|
6588 }
|
|
|
6589 if (tp == NULL)
|
|
|
6590 {
|
|
|
6591 /* No other child with this hash value equals the child of
|
|
|
6592 * the node, add it to the linked list after the first
|
|
|
6593 * item. */
|
|
|
6594 tp = HI2WN(hi);
|
|
346
|
6595 child->wn_u2.next = tp->wn_u2.next;
|
|
|
6596 tp->wn_u2.next = child;
|
|
300
|
6597 }
|
|
|
6598 }
|
|
|
6599 else
|
|
|
6600 /* No other child has this hash value, add it to the
|
|
|
6601 * hashtable. */
|
|
346
|
6602 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
6603 }
|
|
236
|
6604 }
|
|
300
|
6605 *tot += len;
|
|
|
6606
|
|
|
6607 /*
|
|
|
6608 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
6609 * find a lookalike node. This must be done after compressing the sibling
|
|
|
6610 * list, otherwise the hash key would become invalid by the compression.
|
|
|
6611 */
|
|
346
|
6612 node->wn_u1.hashkey[0] = len;
|
|
300
|
6613 nr = 0;
|
|
|
6614 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
6615 {
|
|
300
|
6616 if (np->wn_byte == NUL)
|
|
481
|
6617 /* end node: use wn_flags, wn_region and wn_affixID */
|
|
|
6618 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16);
|
|
300
|
6619 else
|
|
|
6620 /* byte node: use the byte value and the child pointer */
|
|
|
6621 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
6622 nr = nr * 101 + n;
|
|
236
|
6623 }
|
|
300
|
6624
|
|
|
6625 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
6626 n = nr & 0xff;
|
|
346
|
6627 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
6628 n = (nr >> 8) & 0xff;
|
|
346
|
6629 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
6630 n = (nr >> 16) & 0xff;
|
|
346
|
6631 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
6632 n = (nr >> 24) & 0xff;
|
|
346
|
6633 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
6634 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
6635
|
|
482
|
6636 /* Check for CTRL-C pressed now and then. */
|
|
|
6637 fast_breakcheck();
|
|
|
6638
|
|
300
|
6639 return compressed;
|
|
|
6640 }
|
|
|
6641
|
|
|
6642 /*
|
|
|
6643 * Return TRUE when two nodes have identical siblings and children.
|
|
|
6644 */
|
|
|
6645 static int
|
|
|
6646 node_equal(n1, n2)
|
|
|
6647 wordnode_T *n1;
|
|
|
6648 wordnode_T *n2;
|
|
|
6649 {
|
|
|
6650 wordnode_T *p1;
|
|
|
6651 wordnode_T *p2;
|
|
|
6652
|
|
|
6653 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
6654 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
6655 if (p1->wn_byte != p2->wn_byte
|
|
|
6656 || (p1->wn_byte == NUL
|
|
|
6657 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
6658 || p1->wn_region != p2->wn_region
|
|
481
|
6659 || p1->wn_affixID != p2->wn_affixID)
|
|
300
|
6660 : (p1->wn_child != p2->wn_child)))
|
|
|
6661 break;
|
|
|
6662
|
|
|
6663 return p1 == NULL && p2 == NULL;
|
|
236
|
6664 }
|
|
|
6665
|
|
|
6666 /*
|
|
|
6667 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
6668 */
|
|
255
|
6669 void
|
|
236
|
6670 put_bytes(fd, nr, len)
|
|
|
6671 FILE *fd;
|
|
|
6672 long_u nr;
|
|
|
6673 int len;
|
|
|
6674 {
|
|
|
6675 int i;
|
|
|
6676
|
|
|
6677 for (i = len - 1; i >= 0; --i)
|
|
|
6678 putc((int)(nr >> (i * 8)), fd);
|
|
|
6679 }
|
|
|
6680
|
|
323
|
6681 static int
|
|
|
6682 #ifdef __BORLANDC__
|
|
|
6683 _RTLENTRYF
|
|
|
6684 #endif
|
|
|
6685 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
6686
|
|
|
6687 /*
|
|
|
6688 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
6689 */
|
|
|
6690 static int
|
|
|
6691 #ifdef __BORLANDC__
|
|
|
6692 _RTLENTRYF
|
|
|
6693 #endif
|
|
|
6694 rep_compare(s1, s2)
|
|
|
6695 const void *s1;
|
|
|
6696 const void *s2;
|
|
|
6697 {
|
|
|
6698 fromto_T *p1 = (fromto_T *)s1;
|
|
|
6699 fromto_T *p2 = (fromto_T *)s2;
|
|
|
6700
|
|
|
6701 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
6702 }
|
|
|
6703
|
|
236
|
6704 /*
|
|
483
|
6705 * Write the Vim .spl file "fname".
|
|
236
|
6706 */
|
|
|
6707 static void
|
|
474
|
6708 write_vim_spell(spin, fname)
|
|
|
6709 spellinfo_T *spin;
|
|
236
|
6710 char_u *fname;
|
|
|
6711 {
|
|
300
|
6712 FILE *fd;
|
|
|
6713 int regionmask;
|
|
236
|
6714 int round;
|
|
300
|
6715 wordnode_T *tree;
|
|
|
6716 int nodecount;
|
|
323
|
6717 int i;
|
|
|
6718 int l;
|
|
|
6719 garray_T *gap;
|
|
|
6720 fromto_T *ftp;
|
|
|
6721 char_u *p;
|
|
|
6722 int rr;
|
|
236
|
6723
|
|
310
|
6724 fd = mch_fopen((char *)fname, "w");
|
|
300
|
6725 if (fd == NULL)
|
|
236
|
6726 {
|
|
|
6727 EMSG2(_(e_notopen), fname);
|
|
|
6728 return;
|
|
|
6729 }
|
|
|
6730
|
|
483
|
6731 /* <HEADER>: <fileID> <versionnr> */
|
|
300
|
6732 /* <fileID> */
|
|
|
6733 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
6734 EMSG(_(e_write));
|
|
483
|
6735 putc(VIMSPELLVERSION, fd); /* <versionnr> */
|
|
|
6736
|
|
|
6737 /*
|
|
|
6738 * <SECTIONS>: <section> ... <sectionend>
|
|
|
6739 */
|
|
|
6740
|
|
|
6741 /* SN_REGION: <regionname> ...
|
|
|
6742 * Write the region names only if there is more than one. */
|
|
316
|
6743 if (spin->si_region_count > 1)
|
|
236
|
6744 {
|
|
483
|
6745 putc(SN_REGION, fd); /* <sectionID> */
|
|
|
6746 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6747 l = spin->si_region_count * 2;
|
|
|
6748 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6749 fwrite(spin->si_region_name, (size_t)l, (size_t)1, fd);
|
|
|
6750 /* <regionname> ... */
|
|
316
|
6751 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
6752 }
|
|
|
6753 else
|
|
300
|
6754 regionmask = 0;
|
|
483
|
6755
|
|
|
6756 /* SN_CHARFLAGS: <charflagslen> <charflags> <folcharslen> <folchars>
|
|
|
6757 *
|
|
|
6758 * The table with character flags and the table for case folding.
|
|
|
6759 * This makes sure the same characters are recognized as word characters
|
|
|
6760 * when generating an when using a spell file.
|
|
260
|
6761 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
6762 * 'encoding'.
|
|
|
6763 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
6764 * the table (avoids that it conflicts). File is shorter too.
|
|
|
6765 */
|
|
483
|
6766 if (!spin->si_ascii && !spin->si_add)
|
|
|
6767 {
|
|
|
6768 char_u folchars[128 * 8];
|
|
|
6769 int flags;
|
|
|
6770
|
|
485
|
6771 putc(SN_CHARFLAGS, fd); /* <sectionID> */
|
|
483
|
6772 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6773
|
|
|
6774 /* Form the <folchars> string first, we need to know its length. */
|
|
|
6775 l = 0;
|
|
|
6776 for (i = 128; i < 256; ++i)
|
|
|
6777 {
|
|
|
6778 #ifdef FEAT_MBYTE
|
|
|
6779 if (has_mbyte)
|
|
|
6780 l += mb_char2bytes(spelltab.st_fold[i], folchars + l);
|
|
|
6781 else
|
|
|
6782 #endif
|
|
|
6783 folchars[l++] = spelltab.st_fold[i];
|
|
|
6784 }
|
|
|
6785 put_bytes(fd, (long_u)(1 + 128 + 2 + l), 4); /* <sectionlen> */
|
|
|
6786
|
|
|
6787 fputc(128, fd); /* <charflagslen> */
|
|
|
6788 for (i = 128; i < 256; ++i)
|
|
|
6789 {
|
|
|
6790 flags = 0;
|
|
|
6791 if (spelltab.st_isw[i])
|
|
|
6792 flags |= CF_WORD;
|
|
|
6793 if (spelltab.st_isu[i])
|
|
|
6794 flags |= CF_UPPER;
|
|
|
6795 fputc(flags, fd); /* <charflags> */
|
|
|
6796 }
|
|
|
6797
|
|
|
6798 put_bytes(fd, (long_u)l, 2); /* <folcharslen> */
|
|
|
6799 fwrite(folchars, (size_t)l, (size_t)1, fd); /* <folchars> */
|
|
|
6800 }
|
|
|
6801
|
|
|
6802 /* SN_MIDWORD: <midword> */
|
|
|
6803 if (spin->si_midword != NULL)
|
|
|
6804 {
|
|
|
6805 putc(SN_MIDWORD, fd); /* <sectionID> */
|
|
|
6806 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6807
|
|
366
|
6808 i = STRLEN(spin->si_midword);
|
|
483
|
6809 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */
|
|
366
|
6810 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
6811 }
|
|
|
6812
|
|
483
|
6813 /* SN_PREFCOND: <prefcondcnt> <prefcond> ... */
|
|
|
6814 if (spin->si_prefcond.ga_len > 0)
|
|
|
6815 {
|
|
|
6816 putc(SN_PREFCOND, fd); /* <sectionID> */
|
|
|
6817 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6818
|
|
|
6819 l = write_spell_prefcond(NULL, &spin->si_prefcond);
|
|
|
6820 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6821
|
|
|
6822 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
6823 }
|
|
|
6824
|
|
|
6825 /* SN_REP: <repcount> <rep> ...
|
|
|
6826 * SN_SAL: <salflags> <salcount> <sal> ... */
|
|
366
|
6827
|
|
323
|
6828 /* Sort the REP items. */
|
|
|
6829 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
6830 sizeof(fromto_T), rep_compare);
|
|
|
6831
|
|
483
|
6832 /* round 1: SN_REP section
|
|
|
6833 * round 2: SN_SAL section (unless SN_SOFO is used) */
|
|
323
|
6834 for (round = 1; round <= 2; ++round)
|
|
|
6835 {
|
|
|
6836 if (round == 1)
|
|
483
|
6837 {
|
|
323
|
6838 gap = &spin->si_rep;
|
|
483
|
6839 putc(SN_REP, fd); /* <sectionID> */
|
|
|
6840 }
|
|
323
|
6841 else
|
|
|
6842 {
|
|
483
|
6843 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
6844 /* using SN_SOFO section instead of SN_SAL */
|
|
|
6845 break;
|
|
323
|
6846 gap = &spin->si_sal;
|
|
483
|
6847 putc(SN_SAL, fd); /* <sectionID> */
|
|
|
6848 }
|
|
|
6849
|
|
|
6850 /* This is for making suggestions, section is not required. */
|
|
|
6851 putc(0, fd); /* <sectionflags> */
|
|
|
6852
|
|
|
6853 /* Compute the length of what follows. */
|
|
|
6854 l = 2; /* count <repcount> or <salcount> */
|
|
|
6855 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6856 {
|
|
|
6857 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
6858 l += 1 + STRLEN(ftp->ft_from); /* count <*fromlen> and <*from> */
|
|
|
6859 l += 1 + STRLEN(ftp->ft_to); /* count <*tolen> and <*to> */
|
|
|
6860 }
|
|
|
6861 if (round == 2)
|
|
|
6862 ++l; /* count <salflags> */
|
|
|
6863 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6864
|
|
|
6865 if (round == 2)
|
|
|
6866 {
|
|
323
|
6867 i = 0;
|
|
|
6868 if (spin->si_followup)
|
|
|
6869 i |= SAL_F0LLOWUP;
|
|
|
6870 if (spin->si_collapse)
|
|
|
6871 i |= SAL_COLLAPSE;
|
|
|
6872 if (spin->si_rem_accents)
|
|
|
6873 i |= SAL_REM_ACCENTS;
|
|
|
6874 putc(i, fd); /* <salflags> */
|
|
|
6875 }
|
|
|
6876
|
|
|
6877 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
6878 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6879 {
|
|
|
6880 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
6881 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
6882 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
6883 for (rr = 1; rr <= 2; ++rr)
|
|
|
6884 {
|
|
|
6885 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
6886 l = STRLEN(p);
|
|
|
6887 putc(l, fd);
|
|
|
6888 fwrite(p, l, (size_t)1, fd);
|
|
|
6889 }
|
|
|
6890 }
|
|
483
|
6891
|
|
|
6892 }
|
|
|
6893
|
|
|
6894 /* SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
6895 * This is for making suggestions, section is not required. */
|
|
375
|
6896 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
6897 {
|
|
483
|
6898 putc(SN_SOFO, fd); /* <sectionID> */
|
|
|
6899 putc(0, fd); /* <sectionflags> */
|
|
375
|
6900
|
|
|
6901 l = STRLEN(spin->si_sofofr);
|
|
483
|
6902 put_bytes(fd, (long_u)(l + STRLEN(spin->si_sofoto) + 4), 4);
|
|
|
6903 /* <sectionlen> */
|
|
|
6904
|
|
|
6905 put_bytes(fd, (long_u)l, 2); /* <sofofromlen> */
|
|
|
6906 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <sofofrom> */
|
|
375
|
6907
|
|
|
6908 l = STRLEN(spin->si_sofoto);
|
|
483
|
6909 put_bytes(fd, (long_u)l, 2); /* <sofotolen> */
|
|
|
6910 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <sofoto> */
|
|
|
6911 }
|
|
|
6912
|
|
|
6913 /* SN_MAP: <mapstr>
|
|
|
6914 * This is for making suggestions, section is not required. */
|
|
|
6915 if (spin->si_map.ga_len > 0)
|
|
|
6916 {
|
|
|
6917 putc(SN_MAP, fd); /* <sectionID> */
|
|
|
6918 putc(0, fd); /* <sectionflags> */
|
|
|
6919 l = spin->si_map.ga_len;
|
|
|
6920 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6921 fwrite(spin->si_map.ga_data, (size_t)l, (size_t)1, fd);
|
|
|
6922 /* <mapstr> */
|
|
|
6923 }
|
|
|
6924
|
|
|
6925 /* SN_COMPOUND: compound info.
|
|
|
6926 * We don't mark it required, when not supported all compound words will
|
|
|
6927 * be bad words. */
|
|
|
6928 if (spin->si_compflags != NULL)
|
|
|
6929 {
|
|
|
6930 putc(SN_COMPOUND, fd); /* <sectionID> */
|
|
|
6931 putc(0, fd); /* <sectionflags> */
|
|
|
6932
|
|
|
6933 l = STRLEN(spin->si_compflags);
|
|
|
6934 put_bytes(fd, (long_u)(l + 3), 4); /* <sectionlen> */
|
|
|
6935 putc(spin->si_compmax, fd); /* <compmax> */
|
|
|
6936 putc(spin->si_compminlen, fd); /* <compminlen> */
|
|
|
6937 putc(spin->si_compsylmax, fd); /* <compsylmax> */
|
|
|
6938 /* <compflags> */
|
|
|
6939 fwrite(spin->si_compflags, (size_t)l, (size_t)1, fd);
|
|
|
6940 }
|
|
|
6941
|
|
492
|
6942 /* SN_NOBREAK: NOBREAK flag */
|
|
|
6943 if (spin->si_nobreak)
|
|
|
6944 {
|
|
|
6945 putc(SN_NOBREAK, fd); /* <sectionID> */
|
|
|
6946 putc(0, fd); /* <sectionflags> */
|
|
|
6947
|
|
|
6948 /* It's empty, the precense of the section flags the feature. */
|
|
|
6949 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */
|
|
|
6950 }
|
|
|
6951
|
|
483
|
6952 /* SN_SYLLABLE: syllable info.
|
|
|
6953 * We don't mark it required, when not supported syllables will not be
|
|
|
6954 * counted. */
|
|
|
6955 if (spin->si_syllable != NULL)
|
|
|
6956 {
|
|
|
6957 putc(SN_SYLLABLE, fd); /* <sectionID> */
|
|
|
6958 putc(0, fd); /* <sectionflags> */
|
|
|
6959
|
|
|
6960 l = STRLEN(spin->si_syllable);
|
|
|
6961 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6962 fwrite(spin->si_syllable, (size_t)l, (size_t)1, fd); /* <syllable> */
|
|
|
6963 }
|
|
|
6964
|
|
|
6965 /* end of <SECTIONS> */
|
|
|
6966 putc(SN_END, fd); /* <sectionend> */
|
|
|
6967
|
|
302
|
6968
|
|
236
|
6969 /*
|
|
339
|
6970 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
6971 */
|
|
323
|
6972 spin->si_memtot = 0;
|
|
339
|
6973 for (round = 1; round <= 3; ++round)
|
|
236
|
6974 {
|
|
339
|
6975 if (round == 1)
|
|
469
|
6976 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
6977 else if (round == 2)
|
|
469
|
6978 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
6979 else
|
|
469
|
6980 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
6981
|
|
346
|
6982 /* Clear the index and wnode fields in the tree. */
|
|
|
6983 clear_node(tree);
|
|
|
6984
|
|
300
|
6985 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
6986 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
6987 * nodes. */
|
|
346
|
6988 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
6989
|
|
300
|
6990 /* number of nodes in 4 bytes */
|
|
|
6991 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
6992 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
6993
|
|
300
|
6994 /* Write the nodes. */
|
|
346
|
6995 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
6996 }
|
|
|
6997
|
|
300
|
6998 fclose(fd);
|
|
236
|
6999 }
|
|
|
7000
|
|
|
7001 /*
|
|
346
|
7002 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
7003 * children. This is needed because they are a union with other items to save
|
|
|
7004 * space.
|
|
|
7005 */
|
|
|
7006 static void
|
|
|
7007 clear_node(node)
|
|
|
7008 wordnode_T *node;
|
|
|
7009 {
|
|
|
7010 wordnode_T *np;
|
|
|
7011
|
|
|
7012 if (node != NULL)
|
|
|
7013 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7014 {
|
|
|
7015 np->wn_u1.index = 0;
|
|
|
7016 np->wn_u2.wnode = NULL;
|
|
|
7017
|
|
|
7018 if (np->wn_byte != NUL)
|
|
|
7019 clear_node(np->wn_child);
|
|
|
7020 }
|
|
|
7021 }
|
|
|
7022
|
|
|
7023
|
|
|
7024 /*
|
|
300
|
7025 * Dump a word tree at node "node".
|
|
|
7026 *
|
|
|
7027 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
7028 * byte recursively write the children.
|
|
|
7029 *
|
|
|
7030 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
7031 * are made about the indexes (so that we don't have to write them in the
|
|
|
7032 * file).
|
|
236
|
7033 *
|
|
300
|
7034 * Returns the number of nodes used.
|
|
236
|
7035 */
|
|
300
|
7036 static int
|
|
346
|
7037 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
7038 FILE *fd; /* NULL when only counting */
|
|
300
|
7039 wordnode_T *node;
|
|
|
7040 int index;
|
|
|
7041 int regionmask;
|
|
339
|
7042 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
7043 {
|
|
300
|
7044 int newindex = index;
|
|
|
7045 int siblingcount = 0;
|
|
|
7046 wordnode_T *np;
|
|
236
|
7047 int flags;
|
|
300
|
7048
|
|
|
7049 /* If "node" is zero the tree is empty. */
|
|
|
7050 if (node == NULL)
|
|
|
7051 return 0;
|
|
|
7052
|
|
|
7053 /* Store the index where this node is written. */
|
|
346
|
7054 node->wn_u1.index = index;
|
|
236
|
7055
|
|
300
|
7056 /* Count the number of siblings. */
|
|
|
7057 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7058 ++siblingcount;
|
|
236
|
7059
|
|
300
|
7060 /* Write the sibling count. */
|
|
|
7061 if (fd != NULL)
|
|
|
7062 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
7063
|
|
300
|
7064 /* Write each sibling byte and optionally extra info. */
|
|
|
7065 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
7066 {
|
|
300
|
7067 if (np->wn_byte == 0)
|
|
|
7068 {
|
|
|
7069 if (fd != NULL)
|
|
|
7070 {
|
|
339
|
7071 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
7072 if (prefixtree)
|
|
300
|
7073 {
|
|
481
|
7074 /* In PREFIXTREE write the required affixID and the
|
|
390
|
7075 * associated condition nr (stored in wn_region). The
|
|
|
7076 * byte value is misused to store the "rare" and "not
|
|
|
7077 * combining" flags */
|
|
455
|
7078 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
7079 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7080 else
|
|
|
7081 {
|
|
390
|
7082 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
7083 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
7084 }
|
|
481
|
7085 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7086 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
7087 }
|
|
|
7088 else
|
|
|
7089 {
|
|
339
|
7090 /* For word trees we write the flag/region items. */
|
|
|
7091 flags = np->wn_flags;
|
|
|
7092 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
7093 flags |= WF_REGION;
|
|
481
|
7094 if (np->wn_affixID != 0)
|
|
|
7095 flags |= WF_AFX;
|
|
339
|
7096 if (flags == 0)
|
|
|
7097 {
|
|
|
7098 /* word without flags or region */
|
|
|
7099 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7100 }
|
|
|
7101 else
|
|
|
7102 {
|
|
390
|
7103 if (np->wn_flags >= 0x100)
|
|
|
7104 {
|
|
|
7105 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
7106 putc(flags, fd); /* <flags> */
|
|
|
7107 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
7108 }
|
|
|
7109 else
|
|
|
7110 {
|
|
|
7111 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
7112 putc(flags, fd); /* <flags> */
|
|
|
7113 }
|
|
339
|
7114 if (flags & WF_REGION)
|
|
|
7115 putc(np->wn_region, fd); /* <region> */
|
|
481
|
7116 if (flags & WF_AFX)
|
|
|
7117 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7118 }
|
|
300
|
7119 }
|
|
|
7120 }
|
|
|
7121 }
|
|
|
7122 else
|
|
|
7123 {
|
|
346
|
7124 if (np->wn_child->wn_u1.index != 0
|
|
|
7125 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
7126 {
|
|
|
7127 /* The child is written elsewhere, write the reference. */
|
|
|
7128 if (fd != NULL)
|
|
|
7129 {
|
|
|
7130 putc(BY_INDEX, fd); /* <byte> */
|
|
|
7131 /* <nodeidx> */
|
|
346
|
7132 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
7133 }
|
|
|
7134 }
|
|
346
|
7135 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
7136 /* We will write the child below and give it an index. */
|
|
346
|
7137 np->wn_child->wn_u2.wnode = node;
|
|
236
|
7138
|
|
300
|
7139 if (fd != NULL)
|
|
|
7140 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
7141 {
|
|
|
7142 EMSG(_(e_write));
|
|
|
7143 return 0;
|
|
|
7144 }
|
|
|
7145 }
|
|
236
|
7146 }
|
|
|
7147
|
|
300
|
7148 /* Space used in the array when reading: one for each sibling and one for
|
|
|
7149 * the count. */
|
|
|
7150 newindex += siblingcount + 1;
|
|
249
|
7151
|
|
300
|
7152 /* Recursively dump the children of each sibling. */
|
|
|
7153 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
7154 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
7155 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
7156 prefixtree);
|
|
249
|
7157
|
|
300
|
7158 return newindex;
|
|
236
|
7159 }
|
|
|
7160
|
|
|
7161
|
|
|
7162 /*
|
|
310
|
7163 * ":mkspell [-ascii] outfile infile ..."
|
|
|
7164 * ":mkspell [-ascii] addfile"
|
|
236
|
7165 */
|
|
|
7166 void
|
|
|
7167 ex_mkspell(eap)
|
|
|
7168 exarg_T *eap;
|
|
|
7169 {
|
|
|
7170 int fcount;
|
|
|
7171 char_u **fnames;
|
|
310
|
7172 char_u *arg = eap->arg;
|
|
|
7173 int ascii = FALSE;
|
|
|
7174
|
|
|
7175 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
7176 {
|
|
|
7177 ascii = TRUE;
|
|
|
7178 arg = skipwhite(arg + 6);
|
|
|
7179 }
|
|
|
7180
|
|
|
7181 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
7182 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
7183 {
|
|
323
|
7184 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
7185 FreeWild(fcount, fnames);
|
|
|
7186 }
|
|
|
7187 }
|
|
|
7188
|
|
|
7189 /*
|
|
|
7190 * Create a Vim spell file from one or more word lists.
|
|
|
7191 * "fnames[0]" is the output file name.
|
|
|
7192 * "fnames[fcount - 1]" is the last input file name.
|
|
|
7193 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
7194 * and ".spl" is appended to make the output file name.
|
|
|
7195 */
|
|
|
7196 static void
|
|
323
|
7197 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
7198 int fcount;
|
|
|
7199 char_u **fnames;
|
|
|
7200 int ascii; /* -ascii argument given */
|
|
|
7201 int overwrite; /* overwrite existing output file */
|
|
323
|
7202 int added_word; /* invoked through "zg" */
|
|
310
|
7203 {
|
|
236
|
7204 char_u fname[MAXPATHL];
|
|
|
7205 char_u wfname[MAXPATHL];
|
|
310
|
7206 char_u **innames;
|
|
|
7207 int incount;
|
|
236
|
7208 afffile_T *(afile[8]);
|
|
|
7209 int i;
|
|
|
7210 int len;
|
|
|
7211 struct stat st;
|
|
255
|
7212 int error = FALSE;
|
|
300
|
7213 spellinfo_T spin;
|
|
|
7214
|
|
|
7215 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
7216 spin.si_verbose = !added_word;
|
|
310
|
7217 spin.si_ascii = ascii;
|
|
323
|
7218 spin.si_followup = TRUE;
|
|
|
7219 spin.si_rem_accents = TRUE;
|
|
|
7220 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
7221 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
7222 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
7223 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
495
|
7224 spin.si_compID = 255; /* start compound ID at maximum, going down */
|
|
310
|
7225
|
|
|
7226 /* default: fnames[0] is output file, following are input files */
|
|
|
7227 innames = &fnames[1];
|
|
|
7228 incount = fcount - 1;
|
|
|
7229
|
|
|
7230 if (fcount >= 1)
|
|
240
|
7231 {
|
|
310
|
7232 len = STRLEN(fnames[0]);
|
|
|
7233 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
7234 {
|
|
|
7235 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
7236 * "path/en.latin1.add.spl". */
|
|
|
7237 innames = &fnames[0];
|
|
|
7238 incount = 1;
|
|
|
7239 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
7240 }
|
|
366
|
7241 else if (fcount == 1)
|
|
|
7242 {
|
|
|
7243 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
7244 innames = &fnames[0];
|
|
|
7245 incount = 1;
|
|
|
7246 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
7247 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
7248 }
|
|
310
|
7249 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
7250 {
|
|
|
7251 /* Name ends in ".spl", use as the file name. */
|
|
323
|
7252 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
7253 }
|
|
|
7254 else
|
|
|
7255 /* Name should be language, make the file name from it. */
|
|
|
7256 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
7257 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
7258
|
|
|
7259 /* Check for .ascii.spl. */
|
|
|
7260 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
7261 spin.si_ascii = TRUE;
|
|
|
7262
|
|
|
7263 /* Check for .add.spl. */
|
|
|
7264 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
7265 spin.si_add = TRUE;
|
|
240
|
7266 }
|
|
|
7267
|
|
310
|
7268 if (incount <= 0)
|
|
236
|
7269 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
7270 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
7271 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
7272 else if (incount > 8)
|
|
236
|
7273 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
7274 else
|
|
|
7275 {
|
|
|
7276 /* Check for overwriting before doing things that may take a lot of
|
|
|
7277 * time. */
|
|
310
|
7278 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
7279 {
|
|
|
7280 EMSG(_(e_exists));
|
|
310
|
7281 return;
|
|
236
|
7282 }
|
|
310
|
7283 if (mch_isdir(wfname))
|
|
236
|
7284 {
|
|
310
|
7285 EMSG2(_(e_isadir2), wfname);
|
|
|
7286 return;
|
|
236
|
7287 }
|
|
|
7288
|
|
|
7289 /*
|
|
|
7290 * Init the aff and dic pointers.
|
|
|
7291 * Get the region names if there are more than 2 arguments.
|
|
|
7292 */
|
|
310
|
7293 for (i = 0; i < incount; ++i)
|
|
236
|
7294 {
|
|
310
|
7295 afile[i] = NULL;
|
|
300
|
7296
|
|
316
|
7297 if (incount > 1)
|
|
236
|
7298 {
|
|
310
|
7299 len = STRLEN(innames[i]);
|
|
|
7300 if (STRLEN(gettail(innames[i])) < 5
|
|
|
7301 || innames[i][len - 3] != '_')
|
|
236
|
7302 {
|
|
310
|
7303 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
7304 return;
|
|
236
|
7305 }
|
|
316
|
7306 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
7307 spin.si_region_name[i * 2 + 1] =
|
|
|
7308 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
7309 }
|
|
|
7310 }
|
|
316
|
7311 spin.si_region_count = incount;
|
|
236
|
7312
|
|
474
|
7313 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
7314 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
7315 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
7316 if (spin.si_foldroot == NULL
|
|
|
7317 || spin.si_keeproot == NULL
|
|
|
7318 || spin.si_prefroot == NULL)
|
|
300
|
7319 {
|
|
469
|
7320 free_blocks(spin.si_blocks);
|
|
310
|
7321 return;
|
|
300
|
7322 }
|
|
|
7323
|
|
351
|
7324 /* When not producing a .add.spl file clear the character table when
|
|
|
7325 * we encounter one in the .aff file. This means we dump the current
|
|
|
7326 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
7327 * better than guessing the contents, the table will match a
|
|
|
7328 * previously loaded spell file. */
|
|
|
7329 if (!spin.si_add)
|
|
|
7330 spin.si_clear_chartab = TRUE;
|
|
|
7331
|
|
236
|
7332 /*
|
|
|
7333 * Read all the .aff and .dic files.
|
|
|
7334 * Text is converted to 'encoding'.
|
|
300
|
7335 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
7336 */
|
|
310
|
7337 for (i = 0; i < incount && !error; ++i)
|
|
236
|
7338 {
|
|
300
|
7339 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
7340 spin.si_region = 1 << i;
|
|
|
7341
|
|
|
7342 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
7343 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
7344 {
|
|
|
7345 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
7346 * "SET" line. */
|
|
474
|
7347 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
7348 if (afile[i] == NULL)
|
|
300
|
7349 error = TRUE;
|
|
|
7350 else
|
|
|
7351 {
|
|
|
7352 /* Read the .dic file and store the words in the trees. */
|
|
|
7353 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
7354 innames[i]);
|
|
474
|
7355 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
7356 error = TRUE;
|
|
|
7357 }
|
|
|
7358 }
|
|
|
7359 else
|
|
|
7360 {
|
|
|
7361 /* No .aff file, try reading the file as a word list. Store
|
|
|
7362 * the words in the trees. */
|
|
474
|
7363 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
7364 error = TRUE;
|
|
|
7365 }
|
|
236
|
7366
|
|
310
|
7367 #ifdef FEAT_MBYTE
|
|
236
|
7368 /* Free any conversion stuff. */
|
|
300
|
7369 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
7370 #endif
|
|
236
|
7371 }
|
|
|
7372
|
|
492
|
7373 if (spin.si_compflags != NULL && spin.si_nobreak)
|
|
|
7374 MSG(_("Warning: both compounding and NOBREAK specified"));
|
|
|
7375
|
|
300
|
7376 if (!error)
|
|
236
|
7377 {
|
|
|
7378 /*
|
|
300
|
7379 * Combine tails in the tree.
|
|
236
|
7380 */
|
|
474
|
7381 if (spin.si_verbose || p_verbose > 2)
|
|
|
7382 {
|
|
|
7383 if (!spin.si_verbose)
|
|
310
|
7384 verbose_enter();
|
|
469
|
7385 MSG(_(msg_compressing));
|
|
310
|
7386 out_flush();
|
|
474
|
7387 if (!spin.si_verbose)
|
|
310
|
7388 verbose_leave();
|
|
|
7389 }
|
|
474
|
7390 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
7391 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
7392 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
7393 }
|
|
|
7394
|
|
300
|
7395 if (!error)
|
|
|
7396 {
|
|
|
7397 /*
|
|
|
7398 * Write the info in the spell file.
|
|
|
7399 */
|
|
474
|
7400 if (spin.si_verbose || p_verbose > 2)
|
|
|
7401 {
|
|
|
7402 if (!spin.si_verbose)
|
|
310
|
7403 verbose_enter();
|
|
366
|
7404 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
7405 out_flush();
|
|
474
|
7406 if (!spin.si_verbose)
|
|
310
|
7407 verbose_leave();
|
|
|
7408 }
|
|
|
7409
|
|
474
|
7410 write_vim_spell(&spin, wfname);
|
|
|
7411
|
|
|
7412 if (spin.si_verbose || p_verbose > 2)
|
|
|
7413 {
|
|
|
7414 if (!spin.si_verbose)
|
|
310
|
7415 verbose_enter();
|
|
|
7416 MSG(_("Done!"));
|
|
|
7417 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
7418 spin.si_memtot);
|
|
310
|
7419 out_flush();
|
|
474
|
7420 if (!spin.si_verbose)
|
|
310
|
7421 verbose_leave();
|
|
|
7422 }
|
|
|
7423
|
|
|
7424 /* If the file is loaded need to reload it. */
|
|
323
|
7425 spell_reload_one(wfname, added_word);
|
|
300
|
7426 }
|
|
|
7427
|
|
|
7428 /* Free the allocated memory. */
|
|
323
|
7429 ga_clear(&spin.si_rep);
|
|
|
7430 ga_clear(&spin.si_sal);
|
|
|
7431 ga_clear(&spin.si_map);
|
|
339
|
7432 ga_clear(&spin.si_prefcond);
|
|
300
|
7433
|
|
|
7434 /* Free the .aff file structures. */
|
|
310
|
7435 for (i = 0; i < incount; ++i)
|
|
|
7436 if (afile[i] != NULL)
|
|
|
7437 spell_free_aff(afile[i]);
|
|
339
|
7438
|
|
|
7439 /* Free all the bits and pieces at once. */
|
|
|
7440 free_blocks(spin.si_blocks);
|
|
236
|
7441 }
|
|
310
|
7442 }
|
|
|
7443
|
|
|
7444
|
|
|
7445 /*
|
|
385
|
7446 * ":[count]spellgood {word}"
|
|
|
7447 * ":[count]spellwrong {word}"
|
|
310
|
7448 */
|
|
|
7449 void
|
|
|
7450 ex_spell(eap)
|
|
|
7451 exarg_T *eap;
|
|
|
7452 {
|
|
381
|
7453 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
7454 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
7455 }
|
|
|
7456
|
|
310
|
7457 /*
|
|
|
7458 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
7459 */
|
|
|
7460 void
|
|
385
|
7461 spell_add_word(word, len, bad, index)
|
|
310
|
7462 char_u *word;
|
|
|
7463 int len;
|
|
|
7464 int bad;
|
|
385
|
7465 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
7466 'spellfile' */
|
|
310
|
7467 {
|
|
|
7468 FILE *fd;
|
|
385
|
7469 buf_T *buf = NULL;
|
|
351
|
7470 int new_spf = FALSE;
|
|
|
7471 struct stat st;
|
|
381
|
7472 char_u *fname;
|
|
385
|
7473 char_u fnamebuf[MAXPATHL];
|
|
|
7474 char_u line[MAXWLEN * 2];
|
|
|
7475 long fpos, fpos_next = 0;
|
|
|
7476 int i;
|
|
|
7477 char_u *spf;
|
|
|
7478
|
|
|
7479 if (index == 0) /* use internal wordlist */
|
|
|
7480 {
|
|
|
7481 if (int_wordlist == NULL)
|
|
|
7482 {
|
|
|
7483 int_wordlist = vim_tempname('s');
|
|
|
7484 if (int_wordlist == NULL)
|
|
381
|
7485 return;
|
|
|
7486 }
|
|
385
|
7487 fname = int_wordlist;
|
|
381
|
7488 }
|
|
310
|
7489 else
|
|
|
7490 {
|
|
381
|
7491 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
7492 if (*curbuf->b_p_spf == NUL)
|
|
|
7493 {
|
|
|
7494 init_spellfile();
|
|
|
7495 new_spf = TRUE;
|
|
|
7496 }
|
|
|
7497
|
|
|
7498 if (*curbuf->b_p_spf == NUL)
|
|
|
7499 {
|
|
|
7500 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
7501 return;
|
|
|
7502 }
|
|
|
7503
|
|
385
|
7504 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
7505 {
|
|
|
7506 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
7507 if (i == index)
|
|
|
7508 break;
|
|
|
7509 if (*spf == NUL)
|
|
|
7510 {
|
|
|
7511 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
7512 return;
|
|
|
7513 }
|
|
|
7514 }
|
|
|
7515
|
|
310
|
7516 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
7517 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
7518 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
7519 buf = NULL;
|
|
|
7520 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
7521 {
|
|
310
|
7522 EMSG(_(e_bufloaded));
|
|
381
|
7523 return;
|
|
|
7524 }
|
|
|
7525
|
|
385
|
7526 fname = fnamebuf;
|
|
|
7527 }
|
|
|
7528
|
|
|
7529 if (bad)
|
|
|
7530 {
|
|
|
7531 /* When the word also appears as good word we need to remove that one,
|
|
|
7532 * since its flags sort before the one with WF_BANNED. */
|
|
|
7533 fd = mch_fopen((char *)fname, "r");
|
|
|
7534 if (fd != NULL)
|
|
|
7535 {
|
|
|
7536 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
7537 {
|
|
|
7538 fpos = fpos_next;
|
|
|
7539 fpos_next = ftell(fd);
|
|
|
7540 if (STRNCMP(word, line, len) == 0
|
|
|
7541 && (line[len] == '/' || line[len] < ' '))
|
|
|
7542 {
|
|
|
7543 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
7544 * the start of the line. Mixing reading and writing
|
|
|
7545 * doesn't work for all systems, close the file first. */
|
|
|
7546 fclose(fd);
|
|
|
7547 fd = mch_fopen((char *)fname, "r+");
|
|
|
7548 if (fd == NULL)
|
|
|
7549 break;
|
|
|
7550 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
7551 fputc('#', fd);
|
|
|
7552 fseek(fd, fpos_next, SEEK_SET);
|
|
|
7553 }
|
|
|
7554 }
|
|
|
7555 fclose(fd);
|
|
|
7556 }
|
|
381
|
7557 }
|
|
|
7558
|
|
|
7559 fd = mch_fopen((char *)fname, "a");
|
|
|
7560 if (fd == NULL && new_spf)
|
|
|
7561 {
|
|
|
7562 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
7563 * We may need to create the "spell" directory first. We already
|
|
|
7564 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
7565 STRCPY(NameBuff, fname);
|
|
|
7566 *gettail_sep(NameBuff) = NUL;
|
|
|
7567 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
7568 {
|
|
|
7569 /* The directory doesn't exist. Try creating it and opening the
|
|
|
7570 * file again. */
|
|
|
7571 vim_mkdir(NameBuff, 0755);
|
|
|
7572 fd = mch_fopen((char *)fname, "a");
|
|
|
7573 }
|
|
|
7574 }
|
|
|
7575
|
|
|
7576 if (fd == NULL)
|
|
|
7577 EMSG2(_(e_notopen), fname);
|
|
|
7578 else
|
|
|
7579 {
|
|
|
7580 if (bad)
|
|
|
7581 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
7582 else
|
|
381
|
7583 fprintf(fd, "%.*s\n", len, word);
|
|
|
7584 fclose(fd);
|
|
|
7585
|
|
|
7586 /* Update the .add.spl file. */
|
|
|
7587 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
7588
|
|
|
7589 /* If the .add file is edited somewhere, reload it. */
|
|
|
7590 if (buf != NULL)
|
|
|
7591 buf_reload(buf);
|
|
|
7592
|
|
|
7593 redraw_all_later(NOT_VALID);
|
|
310
|
7594 }
|
|
|
7595 }
|
|
|
7596
|
|
|
7597 /*
|
|
|
7598 * Initialize 'spellfile' for the current buffer.
|
|
|
7599 */
|
|
|
7600 static void
|
|
|
7601 init_spellfile()
|
|
|
7602 {
|
|
|
7603 char_u buf[MAXPATHL];
|
|
|
7604 int l;
|
|
|
7605 slang_T *sl;
|
|
|
7606 char_u *rtp;
|
|
351
|
7607 char_u *lend;
|
|
310
|
7608
|
|
|
7609 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
7610 {
|
|
351
|
7611 /* Find the end of the language name. Exclude the region. */
|
|
|
7612 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
7613 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
7614 ;
|
|
|
7615
|
|
|
7616 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
7617 * are allowed to write. */
|
|
310
|
7618 rtp = p_rtp;
|
|
|
7619 while (*rtp != NUL)
|
|
|
7620 {
|
|
|
7621 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
7622 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
7623 if (filewritable(buf) == 2)
|
|
|
7624 {
|
|
316
|
7625 /* Use the first language name from 'spelllang' and the
|
|
|
7626 * encoding used in the first loaded .spl file. */
|
|
310
|
7627 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
7628 l = STRLEN(buf);
|
|
|
7629 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
7630 "/spell/%.*s.%s.add",
|
|
351
|
7631 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
7632 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
7633 ? (char_u *)"ascii" : spell_enc());
|
|
|
7634 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
7635 break;
|
|
|
7636 }
|
|
|
7637 }
|
|
|
7638 }
|
|
|
7639 }
|
|
236
|
7640
|
|
300
|
7641
|
|
307
|
7642 /*
|
|
|
7643 * Init the chartab used for spelling for ASCII.
|
|
|
7644 * EBCDIC is not supported!
|
|
|
7645 */
|
|
|
7646 static void
|
|
|
7647 clear_spell_chartab(sp)
|
|
|
7648 spelltab_T *sp;
|
|
|
7649 {
|
|
324
|
7650 int i;
|
|
307
|
7651
|
|
|
7652 /* Init everything to FALSE. */
|
|
|
7653 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
7654 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
7655 for (i = 0; i < 256; ++i)
|
|
324
|
7656 {
|
|
307
|
7657 sp->st_fold[i] = i;
|
|
324
|
7658 sp->st_upper[i] = i;
|
|
|
7659 }
|
|
307
|
7660
|
|
|
7661 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
7662 * that is done separately. */
|
|
|
7663 for (i = '0'; i <= '9'; ++i)
|
|
|
7664 sp->st_isw[i] = TRUE;
|
|
|
7665 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
7666 {
|
|
|
7667 sp->st_isw[i] = TRUE;
|
|
|
7668 sp->st_isu[i] = TRUE;
|
|
|
7669 sp->st_fold[i] = i + 0x20;
|
|
|
7670 }
|
|
|
7671 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
7672 {
|
|
307
|
7673 sp->st_isw[i] = TRUE;
|
|
324
|
7674 sp->st_upper[i] = i - 0x20;
|
|
|
7675 }
|
|
307
|
7676 }
|
|
|
7677
|
|
|
7678 /*
|
|
|
7679 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
7680 * Called once while starting up and when 'encoding' changes.
|
|
|
7681 * The default is to use isalpha(), but the spell file should define the word
|
|
|
7682 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
7683 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
7684 */
|
|
|
7685 void
|
|
|
7686 init_spell_chartab()
|
|
|
7687 {
|
|
|
7688 int i;
|
|
|
7689
|
|
|
7690 did_set_spelltab = FALSE;
|
|
|
7691 clear_spell_chartab(&spelltab);
|
|
|
7692 #ifdef FEAT_MBYTE
|
|
|
7693 if (enc_dbcs)
|
|
|
7694 {
|
|
|
7695 /* DBCS: assume double-wide characters are word characters. */
|
|
|
7696 for (i = 128; i <= 255; ++i)
|
|
|
7697 if (MB_BYTE2LEN(i) == 2)
|
|
|
7698 spelltab.st_isw[i] = TRUE;
|
|
|
7699 }
|
|
324
|
7700 else if (enc_utf8)
|
|
|
7701 {
|
|
|
7702 for (i = 128; i < 256; ++i)
|
|
|
7703 {
|
|
|
7704 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
7705 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
7706 spelltab.st_fold[i] = utf_fold(i);
|
|
|
7707 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
7708 }
|
|
|
7709 }
|
|
307
|
7710 else
|
|
|
7711 #endif
|
|
|
7712 {
|
|
324
|
7713 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
7714 for (i = 128; i < 256; ++i)
|
|
|
7715 {
|
|
|
7716 if (MB_ISUPPER(i))
|
|
|
7717 {
|
|
324
|
7718 spelltab.st_isw[i] = TRUE;
|
|
307
|
7719 spelltab.st_isu[i] = TRUE;
|
|
|
7720 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
7721 }
|
|
324
|
7722 else if (MB_ISLOWER(i))
|
|
|
7723 {
|
|
|
7724 spelltab.st_isw[i] = TRUE;
|
|
|
7725 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
7726 }
|
|
307
|
7727 }
|
|
|
7728 }
|
|
|
7729 }
|
|
|
7730
|
|
|
7731 /*
|
|
|
7732 * Set the spell character tables from strings in the affix file.
|
|
|
7733 */
|
|
|
7734 static int
|
|
|
7735 set_spell_chartab(fol, low, upp)
|
|
|
7736 char_u *fol;
|
|
|
7737 char_u *low;
|
|
|
7738 char_u *upp;
|
|
|
7739 {
|
|
|
7740 /* We build the new tables here first, so that we can compare with the
|
|
|
7741 * previous one. */
|
|
|
7742 spelltab_T new_st;
|
|
|
7743 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
7744 int f, l, u;
|
|
|
7745
|
|
|
7746 clear_spell_chartab(&new_st);
|
|
|
7747
|
|
|
7748 while (*pf != NUL)
|
|
|
7749 {
|
|
|
7750 if (*pl == NUL || *pu == NUL)
|
|
|
7751 {
|
|
|
7752 EMSG(_(e_affform));
|
|
|
7753 return FAIL;
|
|
|
7754 }
|
|
|
7755 #ifdef FEAT_MBYTE
|
|
|
7756 f = mb_ptr2char_adv(&pf);
|
|
|
7757 l = mb_ptr2char_adv(&pl);
|
|
|
7758 u = mb_ptr2char_adv(&pu);
|
|
|
7759 #else
|
|
|
7760 f = *pf++;
|
|
|
7761 l = *pl++;
|
|
|
7762 u = *pu++;
|
|
|
7763 #endif
|
|
|
7764 /* Every character that appears is a word character. */
|
|
|
7765 if (f < 256)
|
|
|
7766 new_st.st_isw[f] = TRUE;
|
|
|
7767 if (l < 256)
|
|
|
7768 new_st.st_isw[l] = TRUE;
|
|
|
7769 if (u < 256)
|
|
|
7770 new_st.st_isw[u] = TRUE;
|
|
|
7771
|
|
|
7772 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
7773 * case-folding */
|
|
|
7774 if (l < 256 && l != f)
|
|
|
7775 {
|
|
|
7776 if (f >= 256)
|
|
|
7777 {
|
|
|
7778 EMSG(_(e_affrange));
|
|
|
7779 return FAIL;
|
|
|
7780 }
|
|
|
7781 new_st.st_fold[l] = f;
|
|
|
7782 }
|
|
|
7783
|
|
|
7784 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
7785 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
7786 * "FOL" . */
|
|
307
|
7787 if (u < 256 && u != f)
|
|
|
7788 {
|
|
|
7789 if (f >= 256)
|
|
|
7790 {
|
|
|
7791 EMSG(_(e_affrange));
|
|
|
7792 return FAIL;
|
|
|
7793 }
|
|
|
7794 new_st.st_fold[u] = f;
|
|
|
7795 new_st.st_isu[u] = TRUE;
|
|
324
|
7796 new_st.st_upper[f] = u;
|
|
307
|
7797 }
|
|
|
7798 }
|
|
|
7799
|
|
|
7800 if (*pl != NUL || *pu != NUL)
|
|
|
7801 {
|
|
|
7802 EMSG(_(e_affform));
|
|
|
7803 return FAIL;
|
|
|
7804 }
|
|
|
7805
|
|
|
7806 return set_spell_finish(&new_st);
|
|
|
7807 }
|
|
|
7808
|
|
|
7809 /*
|
|
|
7810 * Set the spell character tables from strings in the .spl file.
|
|
|
7811 */
|
|
483
|
7812 static void
|
|
389
|
7813 set_spell_charflags(flags, cnt, fol)
|
|
307
|
7814 char_u *flags;
|
|
389
|
7815 int cnt; /* length of "flags" */
|
|
|
7816 char_u *fol;
|
|
307
|
7817 {
|
|
|
7818 /* We build the new tables here first, so that we can compare with the
|
|
|
7819 * previous one. */
|
|
|
7820 spelltab_T new_st;
|
|
|
7821 int i;
|
|
389
|
7822 char_u *p = fol;
|
|
324
|
7823 int c;
|
|
307
|
7824
|
|
|
7825 clear_spell_chartab(&new_st);
|
|
|
7826
|
|
389
|
7827 for (i = 0; i < 128; ++i)
|
|
|
7828 {
|
|
|
7829 if (i < cnt)
|
|
|
7830 {
|
|
|
7831 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
7832 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
7833 }
|
|
|
7834
|
|
|
7835 if (*p != NUL)
|
|
|
7836 {
|
|
307
|
7837 #ifdef FEAT_MBYTE
|
|
389
|
7838 c = mb_ptr2char_adv(&p);
|
|
307
|
7839 #else
|
|
389
|
7840 c = *p++;
|
|
307
|
7841 #endif
|
|
389
|
7842 new_st.st_fold[i + 128] = c;
|
|
|
7843 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
7844 new_st.st_upper[c] = i + 128;
|
|
|
7845 }
|
|
307
|
7846 }
|
|
|
7847
|
|
483
|
7848 (void)set_spell_finish(&new_st);
|
|
307
|
7849 }
|
|
|
7850
|
|
|
7851 static int
|
|
|
7852 set_spell_finish(new_st)
|
|
|
7853 spelltab_T *new_st;
|
|
|
7854 {
|
|
|
7855 int i;
|
|
|
7856
|
|
|
7857 if (did_set_spelltab)
|
|
|
7858 {
|
|
|
7859 /* check that it's the same table */
|
|
|
7860 for (i = 0; i < 256; ++i)
|
|
|
7861 {
|
|
|
7862 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
7863 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
7864 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
7865 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
7866 {
|
|
|
7867 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
7868 return FAIL;
|
|
|
7869 }
|
|
|
7870 }
|
|
|
7871 }
|
|
|
7872 else
|
|
|
7873 {
|
|
|
7874 /* copy the new spelltab into the one being used */
|
|
|
7875 spelltab = *new_st;
|
|
|
7876 did_set_spelltab = TRUE;
|
|
|
7877 }
|
|
|
7878
|
|
|
7879 return OK;
|
|
|
7880 }
|
|
|
7881
|
|
|
7882 /*
|
|
358
|
7883 * Return TRUE if "p" points to a word character.
|
|
366
|
7884 * As a special case we see "midword" characters as word character when it is
|
|
358
|
7885 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
7886 * Thus this only works properly when past the first character of the word.
|
|
358
|
7887 */
|
|
|
7888 static int
|
|
376
|
7889 spell_iswordp(p, buf)
|
|
358
|
7890 char_u *p;
|
|
376
|
7891 buf_T *buf; /* buffer used */
|
|
358
|
7892 {
|
|
366
|
7893 #ifdef FEAT_MBYTE
|
|
358
|
7894 char_u *s;
|
|
366
|
7895 int l;
|
|
|
7896 int c;
|
|
|
7897
|
|
|
7898 if (has_mbyte)
|
|
|
7899 {
|
|
|
7900 l = MB_BYTE2LEN(*p);
|
|
358
|
7901 s = p;
|
|
366
|
7902 if (l == 1)
|
|
|
7903 {
|
|
|
7904 /* be quick for ASCII */
|
|
376
|
7905 if (buf->b_spell_ismw[*p])
|
|
366
|
7906 {
|
|
|
7907 s = p + 1; /* skip a mid-word character */
|
|
|
7908 l = MB_BYTE2LEN(*s);
|
|
|
7909 }
|
|
|
7910 }
|
|
|
7911 else
|
|
|
7912 {
|
|
|
7913 c = mb_ptr2char(p);
|
|
376
|
7914 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
7915 : (buf->b_spell_ismw_mb != NULL
|
|
|
7916 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
7917 {
|
|
|
7918 s = p + l;
|
|
|
7919 l = MB_BYTE2LEN(*s);
|
|
|
7920 }
|
|
|
7921 }
|
|
|
7922
|
|
390
|
7923 c = mb_ptr2char(s);
|
|
|
7924 if (c > 255)
|
|
366
|
7925 return mb_get_class(s) >= 2;
|
|
390
|
7926 return spelltab.st_isw[c];
|
|
366
|
7927 }
|
|
358
|
7928 #endif
|
|
366
|
7929
|
|
376
|
7930 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
7931 }
|
|
|
7932
|
|
|
7933 /*
|
|
|
7934 * Return TRUE if "p" points to a word character.
|
|
|
7935 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
7936 */
|
|
|
7937 static int
|
|
|
7938 spell_iswordp_nmw(p)
|
|
|
7939 char_u *p;
|
|
|
7940 {
|
|
|
7941 #ifdef FEAT_MBYTE
|
|
390
|
7942 int c;
|
|
|
7943
|
|
|
7944 if (has_mbyte)
|
|
|
7945 {
|
|
|
7946 c = mb_ptr2char(p);
|
|
|
7947 if (c > 255)
|
|
|
7948 return mb_get_class(p) >= 2;
|
|
|
7949 return spelltab.st_isw[c];
|
|
|
7950 }
|
|
376
|
7951 #endif
|
|
|
7952 return spelltab.st_isw[*p];
|
|
358
|
7953 }
|
|
|
7954
|
|
372
|
7955 #ifdef FEAT_MBYTE
|
|
|
7956 /*
|
|
|
7957 * Return TRUE if "p" points to a word character.
|
|
|
7958 * Wide version of spell_iswordp().
|
|
|
7959 */
|
|
|
7960 static int
|
|
376
|
7961 spell_iswordp_w(p, buf)
|
|
372
|
7962 int *p;
|
|
376
|
7963 buf_T *buf;
|
|
372
|
7964 {
|
|
|
7965 int *s;
|
|
|
7966
|
|
376
|
7967 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
7968 : (buf->b_spell_ismw_mb != NULL
|
|
|
7969 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
7970 s = p + 1;
|
|
|
7971 else
|
|
|
7972 s = p;
|
|
|
7973
|
|
390
|
7974 if (*s > 255)
|
|
372
|
7975 {
|
|
|
7976 if (enc_utf8)
|
|
|
7977 return utf_class(*s) >= 2;
|
|
|
7978 if (enc_dbcs)
|
|
|
7979 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
7980 return 0;
|
|
|
7981 }
|
|
|
7982 return spelltab.st_isw[*s];
|
|
|
7983 }
|
|
|
7984 #endif
|
|
|
7985
|
|
358
|
7986 /*
|
|
339
|
7987 * Write the table with prefix conditions to the .spl file.
|
|
483
|
7988 * When "fd" is NULL only count the length of what is written.
|
|
|
7989 */
|
|
|
7990 static int
|
|
339
|
7991 write_spell_prefcond(fd, gap)
|
|
|
7992 FILE *fd;
|
|
|
7993 garray_T *gap;
|
|
|
7994 {
|
|
|
7995 int i;
|
|
|
7996 char_u *p;
|
|
|
7997 int len;
|
|
483
|
7998 int totlen;
|
|
|
7999
|
|
|
8000 if (fd != NULL)
|
|
|
8001 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
8002
|
|
|
8003 totlen = 2 + gap->ga_len; /* length of <prefcondcnt> and <condlen> bytes */
|
|
339
|
8004
|
|
|
8005 for (i = 0; i < gap->ga_len; ++i)
|
|
|
8006 {
|
|
|
8007 /* <prefcond> : <condlen> <condstr> */
|
|
|
8008 p = ((char_u **)gap->ga_data)[i];
|
|
483
|
8009 if (p != NULL)
|
|
339
|
8010 {
|
|
|
8011 len = STRLEN(p);
|
|
483
|
8012 if (fd != NULL)
|
|
|
8013 {
|
|
|
8014 fputc(len, fd);
|
|
|
8015 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
8016 }
|
|
|
8017 totlen += len;
|
|
|
8018 }
|
|
|
8019 else if (fd != NULL)
|
|
|
8020 fputc(0, fd);
|
|
|
8021 }
|
|
|
8022
|
|
|
8023 return totlen;
|
|
307
|
8024 }
|
|
|
8025
|
|
|
8026 /*
|
|
324
|
8027 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
8028 * Uses the character definitions from the .spl file.
|
|
307
|
8029 * When using a multi-byte 'encoding' the length may change!
|
|
|
8030 * Returns FAIL when something wrong.
|
|
|
8031 */
|
|
|
8032 static int
|
|
324
|
8033 spell_casefold(str, len, buf, buflen)
|
|
|
8034 char_u *str;
|
|
307
|
8035 int len;
|
|
|
8036 char_u *buf;
|
|
|
8037 int buflen;
|
|
|
8038 {
|
|
|
8039 int i;
|
|
|
8040
|
|
|
8041 if (len >= buflen)
|
|
|
8042 {
|
|
|
8043 buf[0] = NUL;
|
|
|
8044 return FAIL; /* result will not fit */
|
|
|
8045 }
|
|
|
8046
|
|
|
8047 #ifdef FEAT_MBYTE
|
|
|
8048 if (has_mbyte)
|
|
|
8049 {
|
|
324
|
8050 int outi = 0;
|
|
|
8051 char_u *p;
|
|
307
|
8052 int c;
|
|
|
8053
|
|
|
8054 /* Fold one character at a time. */
|
|
324
|
8055 for (p = str; p < str + len; )
|
|
307
|
8056 {
|
|
|
8057 if (outi + MB_MAXBYTES > buflen)
|
|
|
8058 {
|
|
|
8059 buf[outi] = NUL;
|
|
|
8060 return FAIL;
|
|
|
8061 }
|
|
474
|
8062 c = mb_cptr2char_adv(&p);
|
|
324
|
8063 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
8064 }
|
|
|
8065 buf[outi] = NUL;
|
|
|
8066 }
|
|
|
8067 else
|
|
|
8068 #endif
|
|
|
8069 {
|
|
|
8070 /* Be quick for non-multibyte encodings. */
|
|
|
8071 for (i = 0; i < len; ++i)
|
|
324
|
8072 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
8073 buf[i] = NUL;
|
|
|
8074 }
|
|
|
8075
|
|
|
8076 return OK;
|
|
|
8077 }
|
|
|
8078
|
|
372
|
8079 #define SPS_BEST 1
|
|
|
8080 #define SPS_FAST 2
|
|
|
8081 #define SPS_DOUBLE 4
|
|
|
8082
|
|
|
8083 static int sps_flags = SPS_BEST;
|
|
483
|
8084 static int sps_limit = 9999;
|
|
372
|
8085
|
|
|
8086 /*
|
|
|
8087 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
483
|
8088 * Sets "sps_flags" and "sps_limit".
|
|
372
|
8089 */
|
|
|
8090 int
|
|
|
8091 spell_check_sps()
|
|
|
8092 {
|
|
|
8093 char_u *p;
|
|
483
|
8094 char_u *s;
|
|
372
|
8095 char_u buf[MAXPATHL];
|
|
|
8096 int f;
|
|
|
8097
|
|
|
8098 sps_flags = 0;
|
|
483
|
8099 sps_limit = 9999;
|
|
372
|
8100
|
|
|
8101 for (p = p_sps; *p != NUL; )
|
|
|
8102 {
|
|
|
8103 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
8104
|
|
|
8105 f = 0;
|
|
483
|
8106 if (VIM_ISDIGIT(*buf))
|
|
|
8107 {
|
|
|
8108 s = buf;
|
|
|
8109 sps_limit = getdigits(&s);
|
|
|
8110 if (*s != NUL && !VIM_ISDIGIT(*s))
|
|
|
8111 f = -1;
|
|
|
8112 }
|
|
|
8113 else if (STRCMP(buf, "best") == 0)
|
|
372
|
8114 f = SPS_BEST;
|
|
|
8115 else if (STRCMP(buf, "fast") == 0)
|
|
|
8116 f = SPS_FAST;
|
|
|
8117 else if (STRCMP(buf, "double") == 0)
|
|
|
8118 f = SPS_DOUBLE;
|
|
|
8119 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
8120 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
8121 f = -1;
|
|
|
8122
|
|
|
8123 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
8124 {
|
|
|
8125 sps_flags = SPS_BEST;
|
|
483
|
8126 sps_limit = 9999;
|
|
372
|
8127 return FAIL;
|
|
|
8128 }
|
|
|
8129 if (f != 0)
|
|
|
8130 sps_flags = f;
|
|
|
8131 }
|
|
|
8132
|
|
|
8133 if (sps_flags == 0)
|
|
|
8134 sps_flags = SPS_BEST;
|
|
|
8135
|
|
|
8136 return OK;
|
|
|
8137 }
|
|
|
8138
|
|
|
8139 /* Remember what "z?" replaced. */
|
|
|
8140 static char_u *repl_from = NULL;
|
|
|
8141 static char_u *repl_to = NULL;
|
|
|
8142
|
|
323
|
8143 /*
|
|
|
8144 * "z?": Find badly spelled word under or after the cursor.
|
|
|
8145 * Give suggestions for the properly spelled word.
|
|
485
|
8146 * When "count" is non-zero use that suggestion.
|
|
323
|
8147 */
|
|
|
8148 void
|
|
485
|
8149 spell_suggest(count)
|
|
|
8150 int count;
|
|
323
|
8151 {
|
|
|
8152 char_u *line;
|
|
|
8153 pos_T prev_cursor = curwin->w_cursor;
|
|
|
8154 char_u wcopy[MAXWLEN + 2];
|
|
|
8155 char_u *p;
|
|
|
8156 int i;
|
|
|
8157 int c;
|
|
|
8158 suginfo_T sug;
|
|
|
8159 suggest_T *stp;
|
|
372
|
8160 int mouse_used;
|
|
388
|
8161 int need_cap;
|
|
483
|
8162 int limit;
|
|
485
|
8163 int selected = count;
|
|
323
|
8164
|
|
344
|
8165 /* Find the start of the badly spelled word. */
|
|
498
|
8166 if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0
|
|
346
|
8167 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
8168 {
|
|
|
8169 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
8170 return;
|
|
|
8171
|
|
|
8172 /* No bad word or it starts after the cursor: use the word under the
|
|
|
8173 * cursor. */
|
|
|
8174 curwin->w_cursor = prev_cursor;
|
|
|
8175 line = ml_get_curline();
|
|
|
8176 p = line + curwin->w_cursor.col;
|
|
|
8177 /* Backup to before start of word. */
|
|
390
|
8178 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
8179 mb_ptr_back(line, p);
|
|
|
8180 /* Forward to start of word. */
|
|
390
|
8181 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
8182 mb_ptr_adv(p);
|
|
|
8183
|
|
390
|
8184 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
8185 {
|
|
|
8186 beep_flush();
|
|
|
8187 return;
|
|
|
8188 }
|
|
|
8189 curwin->w_cursor.col = p - line;
|
|
323
|
8190 }
|
|
|
8191
|
|
344
|
8192 /* Get the word and its length. */
|
|
|
8193
|
|
388
|
8194 /* Figure out if the word should be capitalised. */
|
|
475
|
8195 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
8196
|
|
|
8197 line = ml_get_curline();
|
|
388
|
8198
|
|
483
|
8199 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in
|
|
|
8200 * 'spellsuggest', whatever is smaller. */
|
|
|
8201 if (sps_limit > (int)Rows - 2)
|
|
|
8202 limit = (int)Rows - 2;
|
|
|
8203 else
|
|
|
8204 limit = sps_limit;
|
|
|
8205 spell_find_suggest(line + curwin->w_cursor.col, &sug, limit,
|
|
388
|
8206 TRUE, need_cap);
|
|
323
|
8207
|
|
|
8208 if (sug.su_ga.ga_len == 0)
|
|
|
8209 MSG(_("Sorry, no suggestions"));
|
|
485
|
8210 else if (count > 0)
|
|
|
8211 {
|
|
|
8212 if (count > sug.su_ga.ga_len)
|
|
|
8213 smsg((char_u *)_("Sorry, only %ld suggestions"),
|
|
|
8214 (long)sug.su_ga.ga_len);
|
|
|
8215 }
|
|
323
|
8216 else
|
|
|
8217 {
|
|
372
|
8218 vim_free(repl_from);
|
|
|
8219 repl_from = NULL;
|
|
|
8220 vim_free(repl_to);
|
|
|
8221 repl_to = NULL;
|
|
|
8222
|
|
474
|
8223 #ifdef FEAT_RIGHTLEFT
|
|
|
8224 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
8225 cmdmsg_rl = curwin->w_p_rl;
|
|
|
8226 if (cmdmsg_rl)
|
|
|
8227 msg_col = Columns - 1;
|
|
|
8228 #endif
|
|
|
8229
|
|
323
|
8230 /* List the suggestions. */
|
|
|
8231 msg_start();
|
|
372
|
8232 lines_left = Rows; /* avoid more prompt */
|
|
323
|
8233 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
8234 sug.su_badlen, sug.su_badptr);
|
|
474
|
8235 #ifdef FEAT_RIGHTLEFT
|
|
|
8236 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
8237 {
|
|
|
8238 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
8239 * untranslated message rightleft. */
|
|
|
8240 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
8241 sug.su_badlen, sug.su_badptr);
|
|
|
8242 }
|
|
|
8243 #endif
|
|
323
|
8244 msg_puts(IObuff);
|
|
|
8245 msg_clr_eos();
|
|
|
8246 msg_putchar('\n');
|
|
346
|
8247
|
|
323
|
8248 msg_scroll = TRUE;
|
|
|
8249 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
8250 {
|
|
344
|
8251 stp = &SUG(sug.su_ga, i);
|
|
323
|
8252
|
|
|
8253 /* The suggested word may replace only part of the bad word, add
|
|
|
8254 * the not replaced part. */
|
|
|
8255 STRCPY(wcopy, stp->st_word);
|
|
|
8256 if (sug.su_badlen > stp->st_orglen)
|
|
|
8257 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
8258 sug.su_badptr + stp->st_orglen,
|
|
|
8259 sug.su_badlen - stp->st_orglen);
|
|
474
|
8260 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
8261 #ifdef FEAT_RIGHTLEFT
|
|
|
8262 if (cmdmsg_rl)
|
|
|
8263 rl_mirror(IObuff);
|
|
|
8264 #endif
|
|
|
8265 msg_puts(IObuff);
|
|
|
8266
|
|
|
8267 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
8268 msg_puts(IObuff);
|
|
|
8269
|
|
|
8270 /* The word may replace more than "su_badlen". */
|
|
|
8271 if (sug.su_badlen < stp->st_orglen)
|
|
|
8272 {
|
|
|
8273 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
8274 stp->st_orglen, sug.su_badptr);
|
|
|
8275 msg_puts(IObuff);
|
|
|
8276 }
|
|
|
8277
|
|
324
|
8278 if (p_verbose > 0)
|
|
344
|
8279 {
|
|
346
|
8280 /* Add the score. */
|
|
351
|
8281 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
8282 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
8283 stp->st_salscore ? "s " : "",
|
|
|
8284 stp->st_score, stp->st_altscore);
|
|
|
8285 else
|
|
474
|
8286 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
8287 stp->st_score);
|
|
474
|
8288 #ifdef FEAT_RIGHTLEFT
|
|
|
8289 if (cmdmsg_rl)
|
|
|
8290 /* Mirror the numbers, but keep the leading space. */
|
|
|
8291 rl_mirror(IObuff + 1);
|
|
|
8292 #endif
|
|
346
|
8293 msg_advance(30);
|
|
|
8294 msg_puts(IObuff);
|
|
344
|
8295 }
|
|
323
|
8296 msg_putchar('\n');
|
|
|
8297 }
|
|
|
8298
|
|
474
|
8299 #ifdef FEAT_RIGHTLEFT
|
|
|
8300 cmdmsg_rl = FALSE;
|
|
|
8301 msg_col = 0;
|
|
|
8302 #endif
|
|
323
|
8303 /* Ask for choice. */
|
|
485
|
8304 selected = prompt_for_number(&mouse_used);
|
|
372
|
8305 if (mouse_used)
|
|
485
|
8306 selected -= lines_left;
|
|
|
8307 }
|
|
|
8308
|
|
|
8309 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
|
8310 {
|
|
|
8311 /* Save the from and to text for :spellrepall. */
|
|
|
8312 stp = &SUG(sug.su_ga, selected - 1);
|
|
|
8313 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
8314 repl_to = vim_strsave(stp->st_word);
|
|
|
8315
|
|
|
8316 /* Replace the word. */
|
|
|
8317 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
8318 if (p != NULL)
|
|
|
8319 {
|
|
|
8320 c = sug.su_badptr - line;
|
|
|
8321 mch_memmove(p, line, c);
|
|
|
8322 STRCPY(p + c, stp->st_word);
|
|
|
8323 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
8324 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
8325 curwin->w_cursor.col = c;
|
|
|
8326 changed_bytes(curwin->w_cursor.lnum, c);
|
|
|
8327
|
|
|
8328 /* For redo we use a change-word command. */
|
|
|
8329 ResetRedobuff();
|
|
|
8330 AppendToRedobuff((char_u *)"ciw");
|
|
|
8331 AppendToRedobuff(stp->st_word);
|
|
|
8332 AppendCharToRedobuff(ESC);
|
|
|
8333 }
|
|
|
8334 }
|
|
|
8335 else
|
|
|
8336 curwin->w_cursor = prev_cursor;
|
|
323
|
8337
|
|
344
|
8338 spell_find_cleanup(&sug);
|
|
|
8339 }
|
|
|
8340
|
|
|
8341 /*
|
|
475
|
8342 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
8343 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
8344 */
|
|
|
8345 static int
|
|
|
8346 check_need_cap(lnum, col)
|
|
|
8347 linenr_T lnum;
|
|
|
8348 colnr_T col;
|
|
|
8349 {
|
|
|
8350 int need_cap = FALSE;
|
|
|
8351 char_u *line;
|
|
|
8352 char_u *line_copy = NULL;
|
|
|
8353 char_u *p;
|
|
|
8354 colnr_T endcol;
|
|
|
8355 regmatch_T regmatch;
|
|
|
8356
|
|
|
8357 if (curbuf->b_cap_prog == NULL)
|
|
|
8358 return FALSE;
|
|
|
8359
|
|
|
8360 line = ml_get_curline();
|
|
|
8361 endcol = 0;
|
|
|
8362 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
8363 {
|
|
|
8364 /* At start of line, check if previous line is empty or sentence
|
|
|
8365 * ends there. */
|
|
|
8366 if (lnum == 1)
|
|
|
8367 need_cap = TRUE;
|
|
|
8368 else
|
|
|
8369 {
|
|
|
8370 line = ml_get(lnum - 1);
|
|
|
8371 if (*skipwhite(line) == NUL)
|
|
|
8372 need_cap = TRUE;
|
|
|
8373 else
|
|
|
8374 {
|
|
|
8375 /* Append a space in place of the line break. */
|
|
|
8376 line_copy = concat_str(line, (char_u *)" ");
|
|
|
8377 line = line_copy;
|
|
|
8378 endcol = STRLEN(line);
|
|
|
8379 }
|
|
|
8380 }
|
|
|
8381 }
|
|
|
8382 else
|
|
|
8383 endcol = col;
|
|
|
8384
|
|
|
8385 if (endcol > 0)
|
|
|
8386 {
|
|
|
8387 /* Check if sentence ends before the bad word. */
|
|
|
8388 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
8389 regmatch.rm_ic = FALSE;
|
|
|
8390 p = line + endcol;
|
|
|
8391 for (;;)
|
|
|
8392 {
|
|
|
8393 mb_ptr_back(line, p);
|
|
|
8394 if (p == line || spell_iswordp_nmw(p))
|
|
|
8395 break;
|
|
|
8396 if (vim_regexec(®match, p, 0)
|
|
|
8397 && regmatch.endp[0] == line + endcol)
|
|
|
8398 {
|
|
|
8399 need_cap = TRUE;
|
|
|
8400 break;
|
|
|
8401 }
|
|
|
8402 }
|
|
|
8403 }
|
|
|
8404
|
|
|
8405 vim_free(line_copy);
|
|
|
8406
|
|
|
8407 return need_cap;
|
|
|
8408 }
|
|
|
8409
|
|
|
8410
|
|
|
8411 /*
|
|
372
|
8412 * ":spellrepall"
|
|
|
8413 */
|
|
|
8414 /*ARGSUSED*/
|
|
|
8415 void
|
|
|
8416 ex_spellrepall(eap)
|
|
|
8417 exarg_T *eap;
|
|
|
8418 {
|
|
|
8419 pos_T pos = curwin->w_cursor;
|
|
|
8420 char_u *frompat;
|
|
|
8421 int addlen;
|
|
|
8422 char_u *line;
|
|
|
8423 char_u *p;
|
|
|
8424 int save_ws = p_ws;
|
|
483
|
8425 linenr_T prev_lnum = 0;
|
|
372
|
8426
|
|
|
8427 if (repl_from == NULL || repl_to == NULL)
|
|
|
8428 {
|
|
|
8429 EMSG(_("E752: No previous spell replacement"));
|
|
|
8430 return;
|
|
|
8431 }
|
|
|
8432 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
8433
|
|
|
8434 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
8435 if (frompat == NULL)
|
|
|
8436 return;
|
|
|
8437 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
8438 p_ws = FALSE;
|
|
|
8439
|
|
483
|
8440 sub_nsubs = 0;
|
|
|
8441 sub_nlines = 0;
|
|
372
|
8442 curwin->w_cursor.lnum = 0;
|
|
|
8443 while (!got_int)
|
|
|
8444 {
|
|
|
8445 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
8446 || u_save_cursor() == FAIL)
|
|
|
8447 break;
|
|
|
8448
|
|
|
8449 /* Only replace when the right word isn't there yet. This happens
|
|
|
8450 * when changing "etc" to "etc.". */
|
|
|
8451 line = ml_get_curline();
|
|
|
8452 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
8453 repl_to, STRLEN(repl_to)) != 0)
|
|
|
8454 {
|
|
|
8455 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
8456 if (p == NULL)
|
|
|
8457 break;
|
|
|
8458 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
8459 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
8460 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
8461 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
8462 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
483
|
8463
|
|
|
8464 if (curwin->w_cursor.lnum != prev_lnum)
|
|
|
8465 {
|
|
|
8466 ++sub_nlines;
|
|
|
8467 prev_lnum = curwin->w_cursor.lnum;
|
|
|
8468 }
|
|
|
8469 ++sub_nsubs;
|
|
372
|
8470 }
|
|
|
8471 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
8472 }
|
|
|
8473
|
|
|
8474 p_ws = save_ws;
|
|
|
8475 curwin->w_cursor = pos;
|
|
|
8476 vim_free(frompat);
|
|
|
8477
|
|
483
|
8478 if (sub_nsubs == 0)
|
|
372
|
8479 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
483
|
8480 else
|
|
|
8481 do_sub_msg(FALSE);
|
|
372
|
8482 }
|
|
|
8483
|
|
|
8484 /*
|
|
344
|
8485 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
8486 * a list of allocated strings.
|
|
|
8487 */
|
|
|
8488 void
|
|
475
|
8489 spell_suggest_list(gap, word, maxcount, need_cap)
|
|
344
|
8490 garray_T *gap;
|
|
|
8491 char_u *word;
|
|
|
8492 int maxcount; /* maximum nr of suggestions */
|
|
475
|
8493 int need_cap; /* 'spellcapcheck' matched */
|
|
344
|
8494 {
|
|
|
8495 suginfo_T sug;
|
|
|
8496 int i;
|
|
|
8497 suggest_T *stp;
|
|
|
8498 char_u *wcopy;
|
|
|
8499
|
|
475
|
8500 spell_find_suggest(word, &sug, maxcount, FALSE, need_cap);
|
|
344
|
8501
|
|
|
8502 /* Make room in "gap". */
|
|
|
8503 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
8504 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
8505 return;
|
|
|
8506
|
|
323
|
8507 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
8508 {
|
|
|
8509 stp = &SUG(sug.su_ga, i);
|
|
|
8510
|
|
|
8511 /* The suggested word may replace only part of "word", add the not
|
|
|
8512 * replaced part. */
|
|
|
8513 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
8514 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
8515 if (wcopy == NULL)
|
|
|
8516 break;
|
|
|
8517 STRCPY(wcopy, stp->st_word);
|
|
|
8518 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
8519 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
8520 }
|
|
|
8521
|
|
|
8522 spell_find_cleanup(&sug);
|
|
|
8523 }
|
|
|
8524
|
|
|
8525 /*
|
|
|
8526 * Find spell suggestions for the word at the start of "badptr".
|
|
|
8527 * Return the suggestions in "su->su_ga".
|
|
|
8528 * The maximum number of suggestions is "maxcount".
|
|
|
8529 * Note: does use info for the current window.
|
|
|
8530 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
8531 */
|
|
|
8532 static void
|
|
388
|
8533 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
8534 char_u *badptr;
|
|
|
8535 suginfo_T *su;
|
|
|
8536 int maxcount;
|
|
358
|
8537 int banbadword; /* don't include badword in suggestions */
|
|
388
|
8538 int need_cap; /* word should start with capital */
|
|
344
|
8539 {
|
|
385
|
8540 int attr = 0;
|
|
372
|
8541 char_u buf[MAXPATHL];
|
|
|
8542 char_u *p;
|
|
|
8543 int do_combine = FALSE;
|
|
|
8544 char_u *sps_copy;
|
|
|
8545 #ifdef FEAT_EVAL
|
|
|
8546 static int expr_busy = FALSE;
|
|
|
8547 #endif
|
|
385
|
8548 int c;
|
|
344
|
8549
|
|
|
8550 /*
|
|
|
8551 * Set the info in "*su".
|
|
|
8552 */
|
|
|
8553 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
8554 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
8555 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
8556 if (*badptr == NUL)
|
|
|
8557 return;
|
|
344
|
8558 hash_init(&su->su_banned);
|
|
|
8559
|
|
|
8560 su->su_badptr = badptr;
|
|
385
|
8561 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
8562 su->su_maxcount = maxcount;
|
|
372
|
8563 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
8564
|
|
|
8565 if (su->su_badlen >= MAXWLEN)
|
|
|
8566 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
8567 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
8568 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
8569 su->su_fbadword, MAXWLEN);
|
|
346
|
8570 /* get caps flags for bad word */
|
|
474
|
8571 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
8572 su->su_badptr + su->su_badlen);
|
|
388
|
8573 if (need_cap)
|
|
|
8574 su->su_badflags |= WF_ONECAP;
|
|
344
|
8575
|
|
385
|
8576 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
8577 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
8578 * for that. */
|
|
455
|
8579 c = PTR2CHAR(su->su_badptr);
|
|
385
|
8580 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
8581 {
|
|
|
8582 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
8583 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
8584 0, TRUE);
|
|
|
8585 }
|
|
|
8586
|
|
344
|
8587 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
8588 if (banbadword)
|
|
|
8589 add_banned(su, su->su_badword);
|
|
344
|
8590
|
|
372
|
8591 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
8592 sps_copy = vim_strsave(p_sps);
|
|
|
8593 if (sps_copy == NULL)
|
|
|
8594 return;
|
|
|
8595
|
|
|
8596 /* Loop over the items in 'spellsuggest'. */
|
|
|
8597 for (p = sps_copy; *p != NUL; )
|
|
|
8598 {
|
|
|
8599 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
8600
|
|
|
8601 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
8602 {
|
|
|
8603 #ifdef FEAT_EVAL
|
|
375
|
8604 /* Evaluate an expression. Skip this when called recursively,
|
|
|
8605 * when using spellsuggest() in the expression. */
|
|
372
|
8606 if (!expr_busy)
|
|
|
8607 {
|
|
|
8608 expr_busy = TRUE;
|
|
|
8609 spell_suggest_expr(su, buf + 5);
|
|
|
8610 expr_busy = FALSE;
|
|
|
8611 }
|
|
|
8612 #endif
|
|
|
8613 }
|
|
|
8614 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
8615 /* Use list of suggestions in a file. */
|
|
|
8616 spell_suggest_file(su, buf + 5);
|
|
|
8617 else
|
|
|
8618 {
|
|
|
8619 /* Use internal method. */
|
|
|
8620 spell_suggest_intern(su);
|
|
|
8621 if (sps_flags & SPS_DOUBLE)
|
|
|
8622 do_combine = TRUE;
|
|
|
8623 }
|
|
|
8624 }
|
|
|
8625
|
|
|
8626 vim_free(sps_copy);
|
|
|
8627
|
|
|
8628 if (do_combine)
|
|
|
8629 /* Combine the two list of suggestions. This must be done last,
|
|
|
8630 * because sorting changes the order again. */
|
|
|
8631 score_combine(su);
|
|
|
8632 }
|
|
|
8633
|
|
|
8634 #ifdef FEAT_EVAL
|
|
|
8635 /*
|
|
|
8636 * Find suggestions by evaluating expression "expr".
|
|
|
8637 */
|
|
|
8638 static void
|
|
|
8639 spell_suggest_expr(su, expr)
|
|
|
8640 suginfo_T *su;
|
|
|
8641 char_u *expr;
|
|
|
8642 {
|
|
|
8643 list_T *list;
|
|
|
8644 listitem_T *li;
|
|
|
8645 int score;
|
|
|
8646 char_u *p;
|
|
|
8647
|
|
|
8648 /* The work is split up in a few parts to avoid having to export
|
|
|
8649 * suginfo_T.
|
|
|
8650 * First evaluate the expression and get the resulting list. */
|
|
|
8651 list = eval_spell_expr(su->su_badword, expr);
|
|
|
8652 if (list != NULL)
|
|
|
8653 {
|
|
|
8654 /* Loop over the items in the list. */
|
|
|
8655 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
8656 if (li->li_tv.v_type == VAR_LIST)
|
|
|
8657 {
|
|
|
8658 /* Get the word and the score from the items. */
|
|
|
8659 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
8660 if (score >= 0)
|
|
|
8661 add_suggestion(su, &su->su_ga, p,
|
|
|
8662 su->su_badlen, score, 0, TRUE);
|
|
|
8663 }
|
|
|
8664 list_unref(list);
|
|
|
8665 }
|
|
|
8666
|
|
|
8667 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8668 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8669 }
|
|
|
8670 #endif
|
|
|
8671
|
|
|
8672 /*
|
|
474
|
8673 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
8674 */
|
|
|
8675 static void
|
|
|
8676 spell_suggest_file(su, fname)
|
|
|
8677 suginfo_T *su;
|
|
|
8678 char_u *fname;
|
|
|
8679 {
|
|
|
8680 FILE *fd;
|
|
|
8681 char_u line[MAXWLEN * 2];
|
|
|
8682 char_u *p;
|
|
|
8683 int len;
|
|
|
8684 char_u cword[MAXWLEN];
|
|
|
8685
|
|
|
8686 /* Open the file. */
|
|
|
8687 fd = mch_fopen((char *)fname, "r");
|
|
|
8688 if (fd == NULL)
|
|
|
8689 {
|
|
|
8690 EMSG2(_(e_notopen), fname);
|
|
|
8691 return;
|
|
|
8692 }
|
|
|
8693
|
|
|
8694 /* Read it line by line. */
|
|
|
8695 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
8696 {
|
|
|
8697 line_breakcheck();
|
|
|
8698
|
|
|
8699 p = vim_strchr(line, '/');
|
|
|
8700 if (p == NULL)
|
|
|
8701 continue; /* No Tab found, just skip the line. */
|
|
|
8702 *p++ = NUL;
|
|
|
8703 if (STRICMP(su->su_badword, line) == 0)
|
|
|
8704 {
|
|
|
8705 /* Match! Isolate the good word, until CR or NL. */
|
|
|
8706 for (len = 0; p[len] >= ' '; ++len)
|
|
|
8707 ;
|
|
|
8708 p[len] = NUL;
|
|
|
8709
|
|
|
8710 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
8711 * of the bad word. */
|
|
|
8712 if (captype(p, NULL) == 0)
|
|
|
8713 {
|
|
|
8714 make_case_word(p, cword, su->su_badflags);
|
|
|
8715 p = cword;
|
|
|
8716 }
|
|
|
8717
|
|
|
8718 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
8719 SCORE_FILE, 0, TRUE);
|
|
|
8720 }
|
|
|
8721 }
|
|
|
8722
|
|
|
8723 fclose(fd);
|
|
|
8724
|
|
|
8725 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8726 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8727 }
|
|
|
8728
|
|
|
8729 /*
|
|
|
8730 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
8731 */
|
|
|
8732 static void
|
|
|
8733 spell_suggest_intern(su)
|
|
|
8734 suginfo_T *su;
|
|
|
8735 {
|
|
344
|
8736 /*
|
|
346
|
8737 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
8738 *
|
|
|
8739 * Set a maximum score to limit the combination of operations that is
|
|
|
8740 * tried.
|
|
|
8741 */
|
|
346
|
8742 suggest_try_special(su);
|
|
|
8743
|
|
|
8744 /*
|
|
|
8745 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
8746 * from the .aff file and inserting a space (split the word).
|
|
|
8747 */
|
|
|
8748 suggest_try_change(su);
|
|
344
|
8749
|
|
|
8750 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
8751 if (sps_flags & SPS_DOUBLE)
|
|
|
8752 score_comp_sal(su);
|
|
|
8753
|
|
|
8754 /*
|
|
346
|
8755 * 3. Try finding sound-a-like words.
|
|
344
|
8756 *
|
|
|
8757 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
8758 * very slow and often doesn't find new suggestions.
|
|
|
8759 */
|
|
|
8760 if ((sps_flags & SPS_DOUBLE)
|
|
|
8761 || (!(sps_flags & SPS_FAST)
|
|
|
8762 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
8763 {
|
|
|
8764 /* Allow a higher score now. */
|
|
|
8765 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
8766 suggest_try_soundalike(su);
|
|
344
|
8767 }
|
|
|
8768
|
|
|
8769 /* When CTRL-C was hit while searching do show the results. */
|
|
|
8770 ui_breakcheck();
|
|
|
8771 if (got_int)
|
|
|
8772 {
|
|
|
8773 (void)vgetc();
|
|
|
8774 got_int = FALSE;
|
|
|
8775 }
|
|
|
8776
|
|
372
|
8777 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
8778 {
|
|
|
8779 if (sps_flags & SPS_BEST)
|
|
|
8780 /* Adjust the word score for how it sounds like. */
|
|
|
8781 rescore_suggestions(su);
|
|
|
8782
|
|
|
8783 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
8784 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
8785 }
|
|
|
8786 }
|
|
|
8787
|
|
|
8788 /*
|
|
|
8789 * Free the info put in "*su" by spell_find_suggest().
|
|
|
8790 */
|
|
|
8791 static void
|
|
|
8792 spell_find_cleanup(su)
|
|
|
8793 suginfo_T *su;
|
|
|
8794 {
|
|
|
8795 int i;
|
|
|
8796
|
|
|
8797 /* Free the suggestions. */
|
|
|
8798 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8799 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
8800 ga_clear(&su->su_ga);
|
|
|
8801 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
8802 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
8803 ga_clear(&su->su_sga);
|
|
323
|
8804
|
|
|
8805 /* Free the banned words. */
|
|
344
|
8806 free_banned(su);
|
|
323
|
8807 }
|
|
|
8808
|
|
|
8809 /*
|
|
324
|
8810 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
8811 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
8812 * The result is NUL terminated.
|
|
323
|
8813 */
|
|
|
8814 static void
|
|
324
|
8815 onecap_copy(word, wcopy, upper)
|
|
323
|
8816 char_u *word;
|
|
|
8817 char_u *wcopy;
|
|
|
8818 int upper; /* TRUE: first letter made upper case */
|
|
|
8819 {
|
|
|
8820 char_u *p;
|
|
|
8821 int c;
|
|
|
8822 int l;
|
|
|
8823
|
|
|
8824 p = word;
|
|
|
8825 #ifdef FEAT_MBYTE
|
|
|
8826 if (has_mbyte)
|
|
474
|
8827 c = mb_cptr2char_adv(&p);
|
|
323
|
8828 else
|
|
|
8829 #endif
|
|
|
8830 c = *p++;
|
|
|
8831 if (upper)
|
|
324
|
8832 c = SPELL_TOUPPER(c);
|
|
323
|
8833 else
|
|
324
|
8834 c = SPELL_TOFOLD(c);
|
|
323
|
8835 #ifdef FEAT_MBYTE
|
|
|
8836 if (has_mbyte)
|
|
|
8837 l = mb_char2bytes(c, wcopy);
|
|
|
8838 else
|
|
|
8839 #endif
|
|
|
8840 {
|
|
|
8841 l = 1;
|
|
|
8842 wcopy[0] = c;
|
|
|
8843 }
|
|
376
|
8844 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
8845 }
|
|
|
8846
|
|
|
8847 /*
|
|
324
|
8848 * Make a copy of "word" with all the letters upper cased into
|
|
|
8849 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
8850 */
|
|
|
8851 static void
|
|
|
8852 allcap_copy(word, wcopy)
|
|
|
8853 char_u *word;
|
|
|
8854 char_u *wcopy;
|
|
|
8855 {
|
|
|
8856 char_u *s;
|
|
|
8857 char_u *d;
|
|
|
8858 int c;
|
|
|
8859
|
|
|
8860 d = wcopy;
|
|
|
8861 for (s = word; *s != NUL; )
|
|
|
8862 {
|
|
|
8863 #ifdef FEAT_MBYTE
|
|
|
8864 if (has_mbyte)
|
|
474
|
8865 c = mb_cptr2char_adv(&s);
|
|
323
|
8866 else
|
|
|
8867 #endif
|
|
|
8868 c = *s++;
|
|
492
|
8869
|
|
|
8870 #ifdef FEAT_MBYTE
|
|
|
8871 /* We only change ß to SS when we are certain latin1 is used. It
|
|
|
8872 * would cause weird errors in other 8-bit encodings. */
|
|
|
8873 if (enc_latin1like && c == 0xdf)
|
|
|
8874 {
|
|
|
8875 c = 'S';
|
|
|
8876 if (d - wcopy >= MAXWLEN - 1)
|
|
|
8877 break;
|
|
|
8878 *d++ = c;
|
|
|
8879 }
|
|
|
8880 else
|
|
|
8881 #endif
|
|
|
8882 c = SPELL_TOUPPER(c);
|
|
323
|
8883
|
|
|
8884 #ifdef FEAT_MBYTE
|
|
|
8885 if (has_mbyte)
|
|
|
8886 {
|
|
|
8887 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
8888 break;
|
|
|
8889 d += mb_char2bytes(c, d);
|
|
|
8890 }
|
|
|
8891 else
|
|
|
8892 #endif
|
|
|
8893 {
|
|
|
8894 if (d - wcopy >= MAXWLEN - 1)
|
|
|
8895 break;
|
|
|
8896 *d++ = c;
|
|
|
8897 }
|
|
|
8898 }
|
|
|
8899 *d = NUL;
|
|
|
8900 }
|
|
|
8901
|
|
|
8902 /*
|
|
346
|
8903 * Try finding suggestions by recognizing specific situations.
|
|
|
8904 */
|
|
|
8905 static void
|
|
|
8906 suggest_try_special(su)
|
|
|
8907 suginfo_T *su;
|
|
|
8908 {
|
|
|
8909 char_u *p;
|
|
376
|
8910 size_t len;
|
|
346
|
8911 int c;
|
|
|
8912 char_u word[MAXWLEN];
|
|
|
8913
|
|
|
8914 /*
|
|
|
8915 * Recognize a word that is repeated: "the the".
|
|
|
8916 */
|
|
|
8917 p = skiptowhite(su->su_fbadword);
|
|
|
8918 len = p - su->su_fbadword;
|
|
|
8919 p = skipwhite(p);
|
|
|
8920 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
8921 {
|
|
|
8922 /* Include badflags: if the badword is onecap or allcap
|
|
|
8923 * use that for the goodword too: "The the" -> "The". */
|
|
|
8924 c = su->su_fbadword[len];
|
|
|
8925 su->su_fbadword[len] = NUL;
|
|
|
8926 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
8927 su->su_fbadword[len] = c;
|
|
351
|
8928 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
8929 }
|
|
|
8930 }
|
|
|
8931
|
|
|
8932 /*
|
|
323
|
8933 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
8934 *
|
|
|
8935 * This uses a state machine. At each node in the tree we try various
|
|
|
8936 * operations. When trying if an operation work "depth" is increased and the
|
|
|
8937 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
8938 * character, replace one and delete another. The number of changes is
|
|
|
8939 * limited by su->su_maxscore, checked in try_deeper().
|
|
485
|
8940 *
|
|
|
8941 * After implementing this I noticed an article by Kemal Oflazer that
|
|
|
8942 * describes something similar: "Error-tolerant Finite State Recognition with
|
|
|
8943 * Applications to Morphological Analysis and Spelling Correction" (1996).
|
|
|
8944 * The implementation in the article is simplified and requires a stack of
|
|
|
8945 * unknown depth. The implementation here only needs a stack depth of the
|
|
|
8946 * length of the word.
|
|
323
|
8947 */
|
|
|
8948 static void
|
|
346
|
8949 suggest_try_change(su)
|
|
323
|
8950 suginfo_T *su;
|
|
|
8951 {
|
|
|
8952 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
8953 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
8954 trystate_T stack[MAXWLEN];
|
|
483
|
8955 char_u preword[MAXWLEN * 3]; /* word found with proper case;
|
|
|
8956 * concatanation of prefix compound
|
|
|
8957 * words and split word. NUL terminated
|
|
|
8958 * when going deeper but not when coming
|
|
|
8959 * back. */
|
|
|
8960 char_u compflags[MAXWLEN]; /* compound flags, one for each word */
|
|
323
|
8961 trystate_T *sp;
|
|
|
8962 int newscore;
|
|
|
8963 langp_T *lp;
|
|
375
|
8964 char_u *byts, *fbyts, *pbyts;
|
|
|
8965 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
8966 int depth;
|
|
330
|
8967 int c, c2, c3;
|
|
390
|
8968 int n;
|
|
323
|
8969 int flags;
|
|
|
8970 garray_T *gap;
|
|
324
|
8971 idx_T arridx;
|
|
323
|
8972 int len;
|
|
|
8973 char_u *p;
|
|
|
8974 fromto_T *ftp;
|
|
330
|
8975 int fl = 0, tl;
|
|
346
|
8976 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
482
|
8977 slang_T *slang;
|
|
|
8978 int fword_ends;
|
|
323
|
8979
|
|
|
8980 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
8981 * to find matches (esp. REP items). Append some more text, changing
|
|
|
8982 * chars after the bad word may help. */
|
|
323
|
8983 STRCPY(fword, su->su_fbadword);
|
|
346
|
8984 n = STRLEN(fword);
|
|
|
8985 p = su->su_badptr + su->su_badlen;
|
|
|
8986 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
8987
|
|
|
8988 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8989 lp->lp_slang != NULL; ++lp)
|
|
|
8990 {
|
|
482
|
8991 slang = lp->lp_slang;
|
|
|
8992
|
|
323
|
8993 /*
|
|
|
8994 * Go through the whole case-fold tree, try changes at each node.
|
|
|
8995 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
8996 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
8997 * word).
|
|
|
8998 */
|
|
|
8999 depth = 0;
|
|
375
|
9000 sp = &stack[0];
|
|
483
|
9001 vim_memset(sp, 0, sizeof(trystate_T));
|
|
375
|
9002 sp->ts_curi = 1;
|
|
330
|
9003
|
|
|
9004 /*
|
|
375
|
9005 * When there are postponed prefixes we need to use these first. At
|
|
|
9006 * the end of the prefix we continue in the case-fold tree.
|
|
|
9007 */
|
|
482
|
9008 fbyts = slang->sl_fbyts;
|
|
|
9009 fidxs = slang->sl_fidxs;
|
|
|
9010 pbyts = slang->sl_pbyts;
|
|
|
9011 pidxs = slang->sl_pidxs;
|
|
375
|
9012 if (pbyts != NULL)
|
|
|
9013 {
|
|
|
9014 byts = pbyts;
|
|
|
9015 idxs = pidxs;
|
|
482
|
9016 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
375
|
9017 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
9018 }
|
|
|
9019 else
|
|
|
9020 {
|
|
|
9021 byts = fbyts;
|
|
|
9022 idxs = fidxs;
|
|
482
|
9023 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
485
|
9024 sp->ts_state = STATE_START;
|
|
375
|
9025 }
|
|
|
9026
|
|
|
9027 /*
|
|
330
|
9028 * Loop to find all suggestions. At each round we either:
|
|
|
9029 * - For the current state try one operation, advance "ts_curi",
|
|
|
9030 * increase "depth".
|
|
|
9031 * - When a state is done go to the next, set "ts_state".
|
|
|
9032 * - When all states are tried decrease "depth".
|
|
|
9033 */
|
|
323
|
9034 while (depth >= 0 && !got_int)
|
|
|
9035 {
|
|
|
9036 sp = &stack[depth];
|
|
|
9037 switch (sp->ts_state)
|
|
|
9038 {
|
|
|
9039 case STATE_START:
|
|
375
|
9040 case STATE_NOPREFIX:
|
|
323
|
9041 /*
|
|
|
9042 * Start of node: Deal with NUL bytes, which means
|
|
|
9043 * tword[] may end here.
|
|
|
9044 */
|
|
|
9045 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
9046 len = byts[arridx]; /* bytes in this node */
|
|
|
9047 arridx += sp->ts_curi; /* index of current byte */
|
|
|
9048
|
|
482
|
9049 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
9050 {
|
|
|
9051 /* Skip over the NUL bytes, we use them later. */
|
|
|
9052 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
9053 ;
|
|
|
9054 sp->ts_curi += n;
|
|
|
9055
|
|
390
|
9056 /* Always past NUL bytes now. */
|
|
|
9057 n = (int)sp->ts_state;
|
|
|
9058 sp->ts_state = STATE_ENDNUL;
|
|
455
|
9059 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
9060
|
|
375
|
9061 /* At end of a prefix or at start of prefixtree: check for
|
|
|
9062 * following word. */
|
|
390
|
9063 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
9064 {
|
|
455
|
9065 /* Set su->su_badflags to the caps type at this
|
|
|
9066 * position. Use the caps type until here for the
|
|
|
9067 * prefix itself. */
|
|
|
9068 #ifdef FEAT_MBYTE
|
|
|
9069 if (has_mbyte)
|
|
|
9070 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
9071 else
|
|
|
9072 #endif
|
|
|
9073 n = sp->ts_fidx;
|
|
474
|
9074 flags = badword_captype(su->su_badptr,
|
|
|
9075 su->su_badptr + n);
|
|
|
9076 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
9077 su->su_badptr + su->su_badlen);
|
|
375
|
9078 ++depth;
|
|
|
9079 stack[depth] = stack[depth - 1];
|
|
|
9080 sp = &stack[depth];
|
|
|
9081 sp->ts_prefixdepth = depth - 1;
|
|
|
9082 byts = fbyts;
|
|
|
9083 idxs = fidxs;
|
|
|
9084 sp->ts_state = STATE_START;
|
|
|
9085 sp->ts_curi = 1; /* start just after length byte */
|
|
|
9086 sp->ts_arridx = 0;
|
|
|
9087
|
|
455
|
9088 /* Move the prefix to preword[] with the right case
|
|
|
9089 * and make find_keepcap_word() works. */
|
|
485
|
9090 tword[sp->ts_twordlen] = NUL;
|
|
|
9091 make_case_word(tword + sp->ts_splitoff,
|
|
|
9092 preword + sp->ts_prewordlen,
|
|
|
9093 flags);
|
|
|
9094 sp->ts_prewordlen = STRLEN(preword);
|
|
483
|
9095 sp->ts_splitoff = sp->ts_twordlen;
|
|
375
|
9096 }
|
|
|
9097 break;
|
|
|
9098 }
|
|
|
9099
|
|
346
|
9100 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
9101 {
|
|
|
9102 /* Past bytes in node and/or past NUL bytes. */
|
|
|
9103 sp->ts_state = STATE_ENDNUL;
|
|
455
|
9104 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
9105 break;
|
|
|
9106 }
|
|
|
9107
|
|
|
9108 /*
|
|
|
9109 * End of word in tree.
|
|
|
9110 */
|
|
|
9111 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
9112
|
|
324
|
9113 flags = (int)idxs[arridx];
|
|
483
|
9114 fword_ends = (fword[sp->ts_fidx] == NUL
|
|
|
9115 || !spell_iswordp(fword + sp->ts_fidx, curbuf));
|
|
|
9116 tword[sp->ts_twordlen] = NUL;
|
|
323
|
9117
|
|
485
|
9118 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL
|
|
|
9119 && (sp->ts_flags & TSF_PREFIXOK) == 0)
|
|
375
|
9120 {
|
|
|
9121 /* There was a prefix before the word. Check that the
|
|
|
9122 * prefix can be used with this word. */
|
|
|
9123 /* Count the length of the NULs in the prefix. If there
|
|
|
9124 * are none this must be the first try without a prefix.
|
|
|
9125 */
|
|
|
9126 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
9127 len = pbyts[n++];
|
|
|
9128 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
9129 ;
|
|
|
9130 if (c > 0)
|
|
|
9131 {
|
|
390
|
9132 c = valid_word_prefix(c, n, flags,
|
|
483
|
9133 tword + sp->ts_splitoff, slang, FALSE);
|
|
375
|
9134 if (c == 0)
|
|
|
9135 break;
|
|
|
9136
|
|
|
9137 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
9138 if (c & WF_RAREPFX)
|
|
|
9139 flags |= WF_RARE;
|
|
485
|
9140
|
|
|
9141 /* Tricky: when checking for both prefix and
|
|
|
9142 * compounding we run into the prefix flag first.
|
|
|
9143 * Remember that it's OK, so that we accept the prefix
|
|
|
9144 * when arriving at a compound flag. */
|
|
|
9145 sp->ts_flags |= TSF_PREFIXOK;
|
|
375
|
9146 }
|
|
|
9147 }
|
|
|
9148
|
|
485
|
9149 if (sp->ts_complen > sp->ts_compsplit)
|
|
|
9150 {
|
|
492
|
9151 if (slang->sl_nobreak)
|
|
|
9152 {
|
|
|
9153 /* There was a word before this word. When there was
|
|
|
9154 * no change in this word (it was correct) add the
|
|
|
9155 * first word as a suggestion. If this word was
|
|
|
9156 * corrected too, we need to check if a correct word
|
|
|
9157 * follows. */
|
|
|
9158 if (sp->ts_fidx - sp->ts_splitfidx
|
|
|
9159 == sp->ts_twordlen - sp->ts_splitoff
|
|
|
9160 && STRNCMP(fword + sp->ts_splitfidx,
|
|
|
9161 tword + sp->ts_splitoff,
|
|
|
9162 sp->ts_fidx - sp->ts_splitfidx) == 0)
|
|
|
9163 {
|
|
|
9164 preword[sp->ts_prewordlen] = NUL;
|
|
|
9165 add_suggestion(su, &su->su_ga, preword,
|
|
|
9166 sp->ts_splitfidx - repextra,
|
|
|
9167 sp->ts_score, 0, FALSE);
|
|
|
9168 break;
|
|
|
9169 }
|
|
|
9170 }
|
|
|
9171 else
|
|
|
9172 {
|
|
|
9173 /* There was a compound word before this word. If
|
|
|
9174 * this word does not support compounding then give up
|
|
|
9175 * (splitting is tried for the word without compound
|
|
|
9176 * flag). */
|
|
|
9177 if (((unsigned)flags >> 24) == 0
|
|
|
9178 || sp->ts_twordlen - sp->ts_splitoff
|
|
485
|
9179 < slang->sl_compminlen)
|
|
492
|
9180 break;
|
|
|
9181 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
9182 compflags[sp->ts_complen + 1] = NUL;
|
|
|
9183 vim_strncpy(preword + sp->ts_prewordlen,
|
|
|
9184 tword + sp->ts_splitoff,
|
|
|
9185 sp->ts_twordlen - sp->ts_splitoff);
|
|
|
9186 p = preword;
|
|
|
9187 while (*skiptowhite(p) != NUL)
|
|
|
9188 p = skipwhite(skiptowhite(p));
|
|
|
9189 if (fword_ends && !can_compound(slang, p,
|
|
485
|
9190 compflags + sp->ts_compsplit))
|
|
492
|
9191 break;
|
|
|
9192
|
|
|
9193 /* Get pointer to last char of previous word. */
|
|
|
9194 p = preword + sp->ts_prewordlen;
|
|
|
9195 mb_ptr_back(preword, p);
|
|
|
9196 }
|
|
490
|
9197 }
|
|
|
9198 else
|
|
|
9199 p = NULL;
|
|
485
|
9200
|
|
323
|
9201 /*
|
|
|
9202 * Form the word with proper case in preword.
|
|
|
9203 * If there is a word from a previous split, append.
|
|
|
9204 */
|
|
|
9205 if (flags & WF_KEEPCAP)
|
|
|
9206 /* Must find the word in the keep-case tree. */
|
|
483
|
9207 find_keepcap_word(slang, tword + sp->ts_splitoff,
|
|
|
9208 preword + sp->ts_prewordlen);
|
|
323
|
9209 else
|
|
346
|
9210 {
|
|
323
|
9211 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
9212 * use that for the goodword too. But if the badword is
|
|
|
9213 * allcap and it's only one char long use onecap. */
|
|
|
9214 c = su->su_badflags;
|
|
|
9215 if ((c & WF_ALLCAP)
|
|
|
9216 #ifdef FEAT_MBYTE
|
|
474
|
9217 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
9218 #else
|
|
|
9219 && su->su_badlen == 1
|
|
|
9220 #endif
|
|
|
9221 )
|
|
|
9222 c = WF_ONECAP;
|
|
490
|
9223 c |= flags;
|
|
|
9224
|
|
|
9225 /* When appending a compound word after a word character
|
|
|
9226 * don't use Onecap. */
|
|
|
9227 if (p != NULL && spell_iswordp_nmw(p))
|
|
|
9228 c &= ~WF_ONECAP;
|
|
483
|
9229 make_case_word(tword + sp->ts_splitoff,
|
|
490
|
9230 preword + sp->ts_prewordlen, c);
|
|
346
|
9231 }
|
|
323
|
9232
|
|
|
9233 /* Don't use a banned word. It may appear again as a good
|
|
|
9234 * word, thus remember it. */
|
|
|
9235 if (flags & WF_BANNED)
|
|
|
9236 {
|
|
483
|
9237 add_banned(su, preword + sp->ts_prewordlen);
|
|
323
|
9238 break;
|
|
|
9239 }
|
|
483
|
9240 if (was_banned(su, preword + sp->ts_prewordlen)
|
|
372
|
9241 || was_banned(su, preword))
|
|
323
|
9242 break;
|
|
|
9243
|
|
|
9244 newscore = 0;
|
|
|
9245 if ((flags & WF_REGION)
|
|
390
|
9246 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
9247 newscore += SCORE_REGION;
|
|
|
9248 if (flags & WF_RARE)
|
|
|
9249 newscore += SCORE_RARE;
|
|
|
9250
|
|
346
|
9251 if (!spell_valid_case(su->su_badflags,
|
|
483
|
9252 captype(preword + sp->ts_prewordlen, NULL)))
|
|
323
|
9253 newscore += SCORE_ICASE;
|
|
|
9254
|
|
482
|
9255 if (fword_ends && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
9256 {
|
|
366
|
9257 /* The badword also ends: add suggestions. Give a penalty
|
|
|
9258 * when changing non-word char to word char, e.g., "thes,"
|
|
|
9259 * -> "these". */
|
|
|
9260 p = fword + sp->ts_fidx;
|
|
|
9261 #ifdef FEAT_MBYTE
|
|
|
9262 if (has_mbyte)
|
|
|
9263 mb_ptr_back(fword, p);
|
|
|
9264 else
|
|
|
9265 #endif
|
|
|
9266 --p;
|
|
376
|
9267 if (!spell_iswordp(p, curbuf))
|
|
366
|
9268 {
|
|
|
9269 p = preword + STRLEN(preword);
|
|
|
9270 #ifdef FEAT_MBYTE
|
|
|
9271 if (has_mbyte)
|
|
|
9272 mb_ptr_back(preword, p);
|
|
|
9273 else
|
|
|
9274 #endif
|
|
|
9275 --p;
|
|
376
|
9276 if (spell_iswordp(p, curbuf))
|
|
366
|
9277 newscore += SCORE_NONWORD;
|
|
|
9278 }
|
|
|
9279
|
|
344
|
9280 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
9281 sp->ts_fidx - repextra,
|
|
351
|
9282 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
9283 }
|
|
482
|
9284 else if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends)
|
|
330
|
9285 #ifdef FEAT_MBYTE
|
|
|
9286 /* Don't split halfway a character. */
|
|
|
9287 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
9288 #endif
|
|
|
9289 )
|
|
323
|
9290 {
|
|
482
|
9291 int try_compound;
|
|
|
9292
|
|
|
9293 /* Get here in two situations:
|
|
|
9294 * 1. The word in the tree ends but the badword continues:
|
|
|
9295 * If the word allows compounding try that. Otherwise
|
|
|
9296 * try a split by inserting a space. For both check
|
|
|
9297 * that a valid words starts at fword[sp->ts_fidx].
|
|
492
|
9298 * For NOBREAK do like compounding to be able to check
|
|
|
9299 * if the next word is valid.
|
|
482
|
9300 * 2. The badword does end, but it was due to a change
|
|
|
9301 * (e.g., a swap). No need to split, but do check that
|
|
|
9302 * the following word is valid.
|
|
|
9303 */
|
|
485
|
9304 try_compound = FALSE;
|
|
482
|
9305 if (!fword_ends
|
|
495
|
9306 && slang->sl_compprog != NULL
|
|
483
|
9307 && ((unsigned)flags >> 24) != 0
|
|
|
9308 && sp->ts_twordlen - sp->ts_splitoff
|
|
482
|
9309 >= slang->sl_compminlen
|
|
490
|
9310 && (slang->sl_compsylmax < MAXWLEN
|
|
|
9311 || sp->ts_complen + 1 - sp->ts_compsplit
|
|
|
9312 < slang->sl_compmax)
|
|
495
|
9313 && (byte_in_str(sp->ts_complen == sp->ts_compsplit
|
|
485
|
9314 ? slang->sl_compstartflags
|
|
|
9315 : slang->sl_compallflags,
|
|
495
|
9316 ((unsigned)flags >> 24))))
|
|
483
|
9317 {
|
|
482
|
9318 try_compound = TRUE;
|
|
483
|
9319 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
9320 compflags[sp->ts_complen + 1] = NUL;
|
|
|
9321 }
|
|
485
|
9322
|
|
492
|
9323 /* For NOBREAK we never try splitting, it won't make any
|
|
|
9324 * word valid. */
|
|
|
9325 if (slang->sl_nobreak)
|
|
|
9326 try_compound = TRUE;
|
|
|
9327
|
|
485
|
9328 /* If we could add a compound word, and it's also possible
|
|
|
9329 * to split at this point, do the split first and set
|
|
|
9330 * TSF_DIDSPLIT to avoid doing it again. */
|
|
492
|
9331 else if (!fword_ends
|
|
485
|
9332 && try_compound
|
|
|
9333 && (sp->ts_flags & TSF_DIDSPLIT) == 0)
|
|
482
|
9334 {
|
|
|
9335 try_compound = FALSE;
|
|
485
|
9336 sp->ts_flags |= TSF_DIDSPLIT;
|
|
|
9337 --sp->ts_curi; /* do the same NUL again */
|
|
|
9338 compflags[sp->ts_complen] = NUL;
|
|
|
9339 }
|
|
|
9340 else
|
|
|
9341 sp->ts_flags &= ~TSF_DIDSPLIT;
|
|
|
9342
|
|
|
9343 if (!try_compound && !fword_ends)
|
|
|
9344 {
|
|
|
9345 /* If we're going to split need to check that the
|
|
|
9346 * words so far are valid for compounding. */
|
|
490
|
9347 p = preword;
|
|
|
9348 while (*skiptowhite(p) != NUL)
|
|
|
9349 p = skipwhite(skiptowhite(p));
|
|
485
|
9350 if (sp->ts_complen > sp->ts_compsplit
|
|
490
|
9351 && !can_compound(slang, p,
|
|
485
|
9352 compflags + sp->ts_compsplit))
|
|
|
9353 break;
|
|
|
9354 newscore += SCORE_SPLIT;
|
|
482
|
9355 }
|
|
|
9356
|
|
|
9357 if (try_deeper(su, stack, depth, newscore))
|
|
323
|
9358 {
|
|
|
9359 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
346
|
9360 sp->ts_save_badflags = su->su_badflags;
|
|
376
|
9361 sp->ts_state = STATE_SPLITUNDO;
|
|
|
9362
|
|
|
9363 ++depth;
|
|
|
9364 sp = &stack[depth];
|
|
323
|
9365
|
|
482
|
9366 /* Append a space to preword when splitting. */
|
|
|
9367 if (!try_compound && !fword_ends)
|
|
|
9368 STRCAT(preword, " ");
|
|
483
|
9369 sp->ts_prewordlen = STRLEN(preword);
|
|
|
9370 sp->ts_splitoff = sp->ts_twordlen;
|
|
492
|
9371 sp->ts_splitfidx = sp->ts_fidx;
|
|
376
|
9372
|
|
|
9373 /* If the badword has a non-word character at this
|
|
|
9374 * position skip it. That means replacing the
|
|
482
|
9375 * non-word character with a space. Always skip a
|
|
|
9376 * character when the word ends. */
|
|
|
9377 if ((!try_compound
|
|
|
9378 && !spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
9379 || fword_ends)
|
|
376
|
9380 {
|
|
482
|
9381 int l;
|
|
|
9382
|
|
376
|
9383 #ifdef FEAT_MBYTE
|
|
|
9384 if (has_mbyte)
|
|
482
|
9385 l = MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
376
|
9386 else
|
|
|
9387 #endif
|
|
482
|
9388 l = 1;
|
|
|
9389 if (fword_ends)
|
|
|
9390 {
|
|
|
9391 /* Copy the skipped character to preword. */
|
|
483
|
9392 mch_memmove(preword + sp->ts_prewordlen,
|
|
482
|
9393 fword + sp->ts_fidx, l);
|
|
483
|
9394 sp->ts_prewordlen += l;
|
|
|
9395 preword[sp->ts_prewordlen] = NUL;
|
|
482
|
9396 }
|
|
|
9397 else
|
|
|
9398 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
9399 sp->ts_fidx += l;
|
|
376
|
9400 }
|
|
455
|
9401
|
|
485
|
9402 /* When compounding include compound flag in
|
|
|
9403 * compflags[] (already set above). When splitting we
|
|
|
9404 * may start compounding over again. */
|
|
482
|
9405 if (try_compound)
|
|
483
|
9406 ++sp->ts_complen;
|
|
482
|
9407 else
|
|
485
|
9408 sp->ts_compsplit = sp->ts_complen;
|
|
|
9409 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
482
|
9410
|
|
455
|
9411 /* set su->su_badflags to the caps type at this
|
|
|
9412 * position */
|
|
324
|
9413 #ifdef FEAT_MBYTE
|
|
|
9414 if (has_mbyte)
|
|
455
|
9415 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
9416 else
|
|
|
9417 #endif
|
|
455
|
9418 n = sp->ts_fidx;
|
|
474
|
9419 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
9420 su->su_badptr + su->su_badlen);
|
|
323
|
9421
|
|
|
9422 /* Restart at top of the tree. */
|
|
376
|
9423 sp->ts_arridx = 0;
|
|
485
|
9424
|
|
|
9425 /* If there are postponed prefixes, try these too. */
|
|
|
9426 if (pbyts != NULL)
|
|
|
9427 {
|
|
|
9428 byts = pbyts;
|
|
|
9429 idxs = pidxs;
|
|
|
9430 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
|
9431 sp->ts_state = STATE_NOPREFIX;
|
|
|
9432 }
|
|
323
|
9433 }
|
|
|
9434 }
|
|
|
9435 break;
|
|
|
9436
|
|
|
9437 case STATE_SPLITUNDO:
|
|
482
|
9438 /* Undo the changes done for word split or compound word. */
|
|
346
|
9439 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
9440
|
|
|
9441 /* Continue looking for NUL bytes. */
|
|
|
9442 sp->ts_state = STATE_START;
|
|
485
|
9443
|
|
|
9444 /* In case we went into the prefix tree. */
|
|
|
9445 byts = fbyts;
|
|
|
9446 idxs = fidxs;
|
|
323
|
9447 break;
|
|
|
9448
|
|
|
9449 case STATE_ENDNUL:
|
|
|
9450 /* Past the NUL bytes in the node. */
|
|
455
|
9451 su->su_badflags = sp->ts_save_badflags;
|
|
346
|
9452 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
9453 {
|
|
|
9454 /* The badword ends, can't use the bytes in this node. */
|
|
|
9455 sp->ts_state = STATE_DEL;
|
|
|
9456 break;
|
|
|
9457 }
|
|
|
9458 sp->ts_state = STATE_PLAIN;
|
|
|
9459 /*FALLTHROUGH*/
|
|
|
9460
|
|
|
9461 case STATE_PLAIN:
|
|
|
9462 /*
|
|
|
9463 * Go over all possible bytes at this node, add each to
|
|
|
9464 * tword[] and use child node. "ts_curi" is the index.
|
|
|
9465 */
|
|
|
9466 arridx = sp->ts_arridx;
|
|
|
9467 if (sp->ts_curi > byts[arridx])
|
|
|
9468 {
|
|
|
9469 /* Done all bytes at this node, do next state. When still
|
|
|
9470 * at already changed bytes skip the other tricks. */
|
|
|
9471 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
9472 sp->ts_state = STATE_DEL;
|
|
|
9473 else
|
|
|
9474 sp->ts_state = STATE_FINAL;
|
|
|
9475 }
|
|
|
9476 else
|
|
|
9477 {
|
|
|
9478 arridx += sp->ts_curi++;
|
|
|
9479 c = byts[arridx];
|
|
|
9480
|
|
|
9481 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
9482 * the byte in the bad word adjust the score. But don't
|
|
|
9483 * even try when the byte was already changed. */
|
|
330
|
9484 if (c == fword[sp->ts_fidx]
|
|
|
9485 #ifdef FEAT_MBYTE
|
|
|
9486 || (sp->ts_tcharlen > 0
|
|
|
9487 && sp->ts_isdiff != DIFF_NONE)
|
|
|
9488 #endif
|
|
|
9489 )
|
|
323
|
9490 newscore = 0;
|
|
|
9491 else
|
|
|
9492 newscore = SCORE_SUBST;
|
|
|
9493 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
9494 && try_deeper(su, stack, depth, newscore))
|
|
|
9495 {
|
|
|
9496 ++depth;
|
|
330
|
9497 sp = &stack[depth];
|
|
|
9498 ++sp->ts_fidx;
|
|
|
9499 tword[sp->ts_twordlen++] = c;
|
|
|
9500 sp->ts_arridx = idxs[arridx];
|
|
|
9501 #ifdef FEAT_MBYTE
|
|
|
9502 if (newscore == SCORE_SUBST)
|
|
|
9503 sp->ts_isdiff = DIFF_YES;
|
|
|
9504 if (has_mbyte)
|
|
|
9505 {
|
|
|
9506 /* Multi-byte characters are a bit complicated to
|
|
|
9507 * handle: They differ when any of the bytes
|
|
|
9508 * differ and then their length may also differ. */
|
|
|
9509 if (sp->ts_tcharlen == 0)
|
|
|
9510 {
|
|
|
9511 /* First byte. */
|
|
|
9512 sp->ts_tcharidx = 0;
|
|
|
9513 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
9514 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
9515 sp->ts_isdiff = (newscore != 0)
|
|
|
9516 ? DIFF_YES : DIFF_NONE;
|
|
|
9517 }
|
|
|
9518 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
9519 /* When inserting trail bytes don't advance in
|
|
|
9520 * the bad word. */
|
|
|
9521 --sp->ts_fidx;
|
|
|
9522 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
9523 {
|
|
|
9524 /* Last byte of character. */
|
|
|
9525 if (sp->ts_isdiff == DIFF_YES)
|
|
|
9526 {
|
|
|
9527 /* Correct ts_fidx for the byte length of
|
|
|
9528 * the character (we didn't check that
|
|
|
9529 * before). */
|
|
|
9530 sp->ts_fidx = sp->ts_fcharstart
|
|
|
9531 + MB_BYTE2LEN(
|
|
|
9532 fword[sp->ts_fcharstart]);
|
|
|
9533
|
|
479
|
9534 /* For changing a composing character
|
|
|
9535 * adjust the score from SCORE_SUBST to
|
|
|
9536 * SCORE_SUBCOMP. */
|
|
|
9537 if (enc_utf8
|
|
|
9538 && utf_iscomposing(
|
|
|
9539 mb_ptr2char(tword
|
|
|
9540 + sp->ts_twordlen
|
|
|
9541 - sp->ts_tcharlen))
|
|
|
9542 && utf_iscomposing(
|
|
|
9543 mb_ptr2char(fword
|
|
|
9544 + sp->ts_fcharstart)))
|
|
|
9545 sp->ts_score -=
|
|
|
9546 SCORE_SUBST - SCORE_SUBCOMP;
|
|
|
9547
|
|
330
|
9548 /* For a similar character adjust score
|
|
|
9549 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
482
|
9550 else if (slang->sl_has_map
|
|
|
9551 && similar_chars(slang,
|
|
330
|
9552 mb_ptr2char(tword
|
|
|
9553 + sp->ts_twordlen
|
|
|
9554 - sp->ts_tcharlen),
|
|
|
9555 mb_ptr2char(fword
|
|
|
9556 + sp->ts_fcharstart)))
|
|
|
9557 sp->ts_score -=
|
|
|
9558 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9559 }
|
|
358
|
9560 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
9561 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
9562 {
|
|
|
9563 p = tword + sp->ts_twordlen
|
|
|
9564 - sp->ts_tcharlen;
|
|
|
9565 c = mb_ptr2char(p);
|
|
475
|
9566 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9567 {
|
|
|
9568 /* Inserting a composing char doesn't
|
|
|
9569 * count that much. */
|
|
358
|
9570 sp->ts_score -= SCORE_INS
|
|
475
|
9571 - SCORE_INSCOMP;
|
|
|
9572 }
|
|
|
9573 else
|
|
|
9574 {
|
|
|
9575 /* If the previous character was the
|
|
|
9576 * same, thus doubling a character,
|
|
|
9577 * give a bonus to the score. */
|
|
|
9578 mb_ptr_back(tword, p);
|
|
|
9579 if (c == mb_ptr2char(p))
|
|
|
9580 sp->ts_score -= SCORE_INS
|
|
358
|
9581 - SCORE_INSDUP;
|
|
475
|
9582 }
|
|
358
|
9583 }
|
|
330
|
9584
|
|
|
9585 /* Starting a new char, reset the length. */
|
|
|
9586 sp->ts_tcharlen = 0;
|
|
|
9587 }
|
|
|
9588 }
|
|
|
9589 else
|
|
|
9590 #endif
|
|
|
9591 {
|
|
|
9592 /* If we found a similar char adjust the score.
|
|
|
9593 * We do this after calling try_deeper() because
|
|
|
9594 * it's slow. */
|
|
|
9595 if (newscore != 0
|
|
482
|
9596 && slang->sl_has_map
|
|
|
9597 && similar_chars(slang,
|
|
330
|
9598 c, fword[sp->ts_fidx - 1]))
|
|
|
9599 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9600 }
|
|
323
|
9601 }
|
|
|
9602 }
|
|
|
9603 break;
|
|
|
9604
|
|
|
9605 case STATE_DEL:
|
|
330
|
9606 #ifdef FEAT_MBYTE
|
|
|
9607 /* When past the first byte of a multi-byte char don't try
|
|
|
9608 * delete/insert/swap a character. */
|
|
|
9609 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
9610 {
|
|
|
9611 sp->ts_state = STATE_FINAL;
|
|
|
9612 break;
|
|
|
9613 }
|
|
|
9614 #endif
|
|
|
9615 /*
|
|
|
9616 * Try skipping one character in the bad word (delete it).
|
|
|
9617 */
|
|
323
|
9618 sp->ts_state = STATE_INS;
|
|
|
9619 sp->ts_curi = 1;
|
|
|
9620 if (fword[sp->ts_fidx] != NUL
|
|
|
9621 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
9622 {
|
|
|
9623 ++depth;
|
|
358
|
9624
|
|
|
9625 /* Advance over the character in fword[]. Give a bonus to
|
|
|
9626 * the score if the same character is following "nn" ->
|
|
|
9627 * "n". */
|
|
330
|
9628 #ifdef FEAT_MBYTE
|
|
|
9629 if (has_mbyte)
|
|
358
|
9630 {
|
|
|
9631 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
9632 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
479
|
9633 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9634 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
|
|
|
9635 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
358
|
9636 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
9637 }
|
|
330
|
9638 else
|
|
|
9639 #endif
|
|
358
|
9640 {
|
|
330
|
9641 ++stack[depth].ts_fidx;
|
|
358
|
9642 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
9643 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
9644 }
|
|
323
|
9645 break;
|
|
|
9646 }
|
|
|
9647 /*FALLTHROUGH*/
|
|
|
9648
|
|
|
9649 case STATE_INS:
|
|
330
|
9650 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
9651 * node. */
|
|
|
9652 n = sp->ts_arridx;
|
|
|
9653 if (sp->ts_curi > byts[n])
|
|
|
9654 {
|
|
|
9655 /* Done all bytes at this node, do next state. */
|
|
|
9656 sp->ts_state = STATE_SWAP;
|
|
|
9657 }
|
|
|
9658 else
|
|
|
9659 {
|
|
330
|
9660 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
9661 n += sp->ts_curi++;
|
|
|
9662 c = byts[n];
|
|
|
9663 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
9664 {
|
|
|
9665 ++depth;
|
|
330
|
9666 sp = &stack[depth];
|
|
|
9667 tword[sp->ts_twordlen++] = c;
|
|
|
9668 sp->ts_arridx = idxs[n];
|
|
|
9669 #ifdef FEAT_MBYTE
|
|
|
9670 if (has_mbyte)
|
|
|
9671 {
|
|
|
9672 fl = MB_BYTE2LEN(c);
|
|
|
9673 if (fl > 1)
|
|
|
9674 {
|
|
|
9675 /* There are following bytes for the same
|
|
|
9676 * character. We must find all bytes before
|
|
|
9677 * trying delete/insert/swap/etc. */
|
|
|
9678 sp->ts_tcharlen = fl;
|
|
|
9679 sp->ts_tcharidx = 1;
|
|
|
9680 sp->ts_isdiff = DIFF_INSERT;
|
|
|
9681 }
|
|
|
9682 }
|
|
358
|
9683 else
|
|
|
9684 fl = 1;
|
|
|
9685 if (fl == 1)
|
|
330
|
9686 #endif
|
|
358
|
9687 {
|
|
|
9688 /* If the previous character was the same, thus
|
|
|
9689 * doubling a character, give a bonus to the
|
|
|
9690 * score. */
|
|
|
9691 if (sp->ts_twordlen >= 2
|
|
|
9692 && tword[sp->ts_twordlen - 2] == c)
|
|
|
9693 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
9694 }
|
|
323
|
9695 }
|
|
|
9696 }
|
|
|
9697 break;
|
|
|
9698
|
|
|
9699 case STATE_SWAP:
|
|
330
|
9700 /*
|
|
|
9701 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
9702 * We change "fword" here, it's changed back afterwards.
|
|
|
9703 */
|
|
|
9704 p = fword + sp->ts_fidx;
|
|
|
9705 c = *p;
|
|
|
9706 if (c == NUL)
|
|
|
9707 {
|
|
|
9708 /* End of word, can't swap or replace. */
|
|
|
9709 sp->ts_state = STATE_FINAL;
|
|
|
9710 break;
|
|
|
9711 }
|
|
|
9712 #ifdef FEAT_MBYTE
|
|
|
9713 if (has_mbyte)
|
|
|
9714 {
|
|
474
|
9715 n = mb_cptr2len(p);
|
|
330
|
9716 c = mb_ptr2char(p);
|
|
|
9717 c2 = mb_ptr2char(p + n);
|
|
|
9718 }
|
|
|
9719 else
|
|
|
9720 #endif
|
|
|
9721 c2 = p[1];
|
|
|
9722 if (c == c2)
|
|
323
|
9723 {
|
|
330
|
9724 /* Characters are identical, swap won't do anything. */
|
|
|
9725 sp->ts_state = STATE_SWAP3;
|
|
|
9726 break;
|
|
|
9727 }
|
|
|
9728 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
9729 {
|
|
|
9730 sp->ts_state = STATE_UNSWAP;
|
|
323
|
9731 ++depth;
|
|
330
|
9732 #ifdef FEAT_MBYTE
|
|
|
9733 if (has_mbyte)
|
|
|
9734 {
|
|
|
9735 fl = mb_char2len(c2);
|
|
|
9736 mch_memmove(p, p + n, fl);
|
|
|
9737 mb_char2bytes(c, p + fl);
|
|
|
9738 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
9739 }
|
|
|
9740 else
|
|
|
9741 #endif
|
|
|
9742 {
|
|
|
9743 p[0] = c2;
|
|
|
9744 p[1] = c;
|
|
|
9745 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
9746 }
|
|
323
|
9747 }
|
|
|
9748 else
|
|
|
9749 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
9750 sp->ts_state = STATE_REP_INI;
|
|
|
9751 break;
|
|
|
9752
|
|
330
|
9753 case STATE_UNSWAP:
|
|
|
9754 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
9755 p = fword + sp->ts_fidx;
|
|
|
9756 #ifdef FEAT_MBYTE
|
|
|
9757 if (has_mbyte)
|
|
|
9758 {
|
|
|
9759 n = MB_BYTE2LEN(*p);
|
|
|
9760 c = mb_ptr2char(p + n);
|
|
|
9761 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
9762 mb_char2bytes(c, p);
|
|
|
9763 }
|
|
|
9764 else
|
|
|
9765 #endif
|
|
|
9766 {
|
|
|
9767 c = *p;
|
|
|
9768 *p = p[1];
|
|
|
9769 p[1] = c;
|
|
|
9770 }
|
|
|
9771 /*FALLTHROUGH*/
|
|
|
9772
|
|
|
9773 case STATE_SWAP3:
|
|
323
|
9774 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
9775 * "fword" here, it's changed back afterwards. */
|
|
|
9776 p = fword + sp->ts_fidx;
|
|
|
9777 #ifdef FEAT_MBYTE
|
|
|
9778 if (has_mbyte)
|
|
|
9779 {
|
|
474
|
9780 n = mb_cptr2len(p);
|
|
330
|
9781 c = mb_ptr2char(p);
|
|
474
|
9782 fl = mb_cptr2len(p + n);
|
|
330
|
9783 c2 = mb_ptr2char(p + n);
|
|
|
9784 c3 = mb_ptr2char(p + n + fl);
|
|
|
9785 }
|
|
|
9786 else
|
|
|
9787 #endif
|
|
323
|
9788 {
|
|
330
|
9789 c = *p;
|
|
|
9790 c2 = p[1];
|
|
|
9791 c3 = p[2];
|
|
|
9792 }
|
|
|
9793
|
|
|
9794 /* When characters are identical: "121" then SWAP3 result is
|
|
|
9795 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
9796 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
9797 * swapping. Also skip when c3 is NUL. */
|
|
|
9798 if (c == c3 || c3 == NUL)
|
|
|
9799 {
|
|
|
9800 sp->ts_state = STATE_REP_INI;
|
|
|
9801 break;
|
|
|
9802 }
|
|
|
9803 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9804 {
|
|
|
9805 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
9806 ++depth;
|
|
330
|
9807 #ifdef FEAT_MBYTE
|
|
|
9808 if (has_mbyte)
|
|
|
9809 {
|
|
|
9810 tl = mb_char2len(c3);
|
|
|
9811 mch_memmove(p, p + n + fl, tl);
|
|
|
9812 mb_char2bytes(c2, p + tl);
|
|
|
9813 mb_char2bytes(c, p + fl + tl);
|
|
|
9814 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
9815 }
|
|
|
9816 else
|
|
|
9817 #endif
|
|
|
9818 {
|
|
|
9819 p[0] = p[2];
|
|
|
9820 p[2] = c;
|
|
|
9821 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9822 }
|
|
323
|
9823 }
|
|
|
9824 else
|
|
|
9825 sp->ts_state = STATE_REP_INI;
|
|
|
9826 break;
|
|
|
9827
|
|
330
|
9828 case STATE_UNSWAP3:
|
|
|
9829 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
9830 p = fword + sp->ts_fidx;
|
|
|
9831 #ifdef FEAT_MBYTE
|
|
|
9832 if (has_mbyte)
|
|
|
9833 {
|
|
|
9834 n = MB_BYTE2LEN(*p);
|
|
|
9835 c2 = mb_ptr2char(p + n);
|
|
|
9836 fl = MB_BYTE2LEN(p[n]);
|
|
|
9837 c = mb_ptr2char(p + n + fl);
|
|
|
9838 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
9839 mch_memmove(p + fl + tl, p, n);
|
|
|
9840 mb_char2bytes(c, p);
|
|
|
9841 mb_char2bytes(c2, p + tl);
|
|
|
9842 }
|
|
|
9843 else
|
|
|
9844 #endif
|
|
|
9845 {
|
|
|
9846 c = *p;
|
|
|
9847 *p = p[2];
|
|
|
9848 p[2] = c;
|
|
|
9849 }
|
|
346
|
9850
|
|
330
|
9851 /* Rotate three characters left: "123" -> "231". We change
|
|
|
9852 * "fword" here, it's changed back afterwards. */
|
|
323
|
9853 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9854 {
|
|
330
|
9855 sp->ts_state = STATE_UNROT3L;
|
|
323
|
9856 ++depth;
|
|
330
|
9857 p = fword + sp->ts_fidx;
|
|
|
9858 #ifdef FEAT_MBYTE
|
|
|
9859 if (has_mbyte)
|
|
|
9860 {
|
|
474
|
9861 n = mb_cptr2len(p);
|
|
330
|
9862 c = mb_ptr2char(p);
|
|
474
|
9863 fl = mb_cptr2len(p + n);
|
|
|
9864 fl += mb_cptr2len(p + n + fl);
|
|
330
|
9865 mch_memmove(p, p + n, fl);
|
|
|
9866 mb_char2bytes(c, p + fl);
|
|
|
9867 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
9868 }
|
|
|
9869 else
|
|
|
9870 #endif
|
|
|
9871 {
|
|
|
9872 c = *p;
|
|
|
9873 *p = p[1];
|
|
|
9874 p[1] = p[2];
|
|
|
9875 p[2] = c;
|
|
|
9876 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9877 }
|
|
323
|
9878 }
|
|
|
9879 else
|
|
|
9880 sp->ts_state = STATE_REP_INI;
|
|
|
9881 break;
|
|
|
9882
|
|
330
|
9883 case STATE_UNROT3L:
|
|
346
|
9884 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
9885 p = fword + sp->ts_fidx;
|
|
|
9886 #ifdef FEAT_MBYTE
|
|
|
9887 if (has_mbyte)
|
|
|
9888 {
|
|
|
9889 n = MB_BYTE2LEN(*p);
|
|
|
9890 n += MB_BYTE2LEN(p[n]);
|
|
|
9891 c = mb_ptr2char(p + n);
|
|
|
9892 tl = MB_BYTE2LEN(p[n]);
|
|
|
9893 mch_memmove(p + tl, p, n);
|
|
|
9894 mb_char2bytes(c, p);
|
|
|
9895 }
|
|
|
9896 else
|
|
|
9897 #endif
|
|
|
9898 {
|
|
|
9899 c = p[2];
|
|
|
9900 p[2] = p[1];
|
|
|
9901 p[1] = *p;
|
|
|
9902 *p = c;
|
|
|
9903 }
|
|
346
|
9904
|
|
323
|
9905 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
9906 * "fword" here, it's changed back afterwards. */
|
|
323
|
9907 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9908 {
|
|
330
|
9909 sp->ts_state = STATE_UNROT3R;
|
|
323
|
9910 ++depth;
|
|
330
|
9911 p = fword + sp->ts_fidx;
|
|
|
9912 #ifdef FEAT_MBYTE
|
|
|
9913 if (has_mbyte)
|
|
|
9914 {
|
|
474
|
9915 n = mb_cptr2len(p);
|
|
|
9916 n += mb_cptr2len(p + n);
|
|
330
|
9917 c = mb_ptr2char(p + n);
|
|
474
|
9918 tl = mb_cptr2len(p + n);
|
|
330
|
9919 mch_memmove(p + tl, p, n);
|
|
|
9920 mb_char2bytes(c, p);
|
|
|
9921 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
9922 }
|
|
|
9923 else
|
|
|
9924 #endif
|
|
|
9925 {
|
|
|
9926 c = p[2];
|
|
|
9927 p[2] = p[1];
|
|
|
9928 p[1] = *p;
|
|
|
9929 *p = c;
|
|
|
9930 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9931 }
|
|
323
|
9932 }
|
|
|
9933 else
|
|
|
9934 sp->ts_state = STATE_REP_INI;
|
|
|
9935 break;
|
|
|
9936
|
|
330
|
9937 case STATE_UNROT3R:
|
|
346
|
9938 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
9939 p = fword + sp->ts_fidx;
|
|
|
9940 #ifdef FEAT_MBYTE
|
|
|
9941 if (has_mbyte)
|
|
|
9942 {
|
|
|
9943 c = mb_ptr2char(p);
|
|
|
9944 tl = MB_BYTE2LEN(*p);
|
|
|
9945 n = MB_BYTE2LEN(p[tl]);
|
|
|
9946 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
9947 mch_memmove(p, p + tl, n);
|
|
|
9948 mb_char2bytes(c, p + n);
|
|
|
9949 }
|
|
|
9950 else
|
|
|
9951 #endif
|
|
|
9952 {
|
|
|
9953 c = *p;
|
|
|
9954 *p = p[1];
|
|
|
9955 p[1] = p[2];
|
|
|
9956 p[2] = c;
|
|
|
9957 }
|
|
323
|
9958 /*FALLTHROUGH*/
|
|
|
9959
|
|
|
9960 case STATE_REP_INI:
|
|
|
9961 /* Check if matching with REP items from the .aff file would
|
|
|
9962 * work. Quickly skip if there are no REP items or the score
|
|
|
9963 * is going to be too high anyway. */
|
|
482
|
9964 gap = &slang->sl_rep;
|
|
323
|
9965 if (gap->ga_len == 0
|
|
|
9966 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
9967 {
|
|
|
9968 sp->ts_state = STATE_FINAL;
|
|
|
9969 break;
|
|
|
9970 }
|
|
|
9971
|
|
|
9972 /* Use the first byte to quickly find the first entry that
|
|
330
|
9973 * may match. If the index is -1 there is none. */
|
|
482
|
9974 sp->ts_curi = slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
323
|
9975 if (sp->ts_curi < 0)
|
|
|
9976 {
|
|
|
9977 sp->ts_state = STATE_FINAL;
|
|
|
9978 break;
|
|
|
9979 }
|
|
|
9980
|
|
|
9981 sp->ts_state = STATE_REP;
|
|
|
9982 /*FALLTHROUGH*/
|
|
|
9983
|
|
|
9984 case STATE_REP:
|
|
|
9985 /* Try matching with REP items from the .aff file. For each
|
|
330
|
9986 * match replace the characters and check if the resulting
|
|
|
9987 * word is valid. */
|
|
323
|
9988 p = fword + sp->ts_fidx;
|
|
|
9989
|
|
482
|
9990 gap = &slang->sl_rep;
|
|
323
|
9991 while (sp->ts_curi < gap->ga_len)
|
|
|
9992 {
|
|
|
9993 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
9994 if (*ftp->ft_from != *p)
|
|
|
9995 {
|
|
|
9996 /* past possible matching entries */
|
|
|
9997 sp->ts_curi = gap->ga_len;
|
|
|
9998 break;
|
|
|
9999 }
|
|
|
10000 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
10001 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
10002 {
|
|
|
10003 /* Need to undo this afterwards. */
|
|
|
10004 sp->ts_state = STATE_REP_UNDO;
|
|
|
10005
|
|
|
10006 /* Change the "from" to the "to" string. */
|
|
|
10007 ++depth;
|
|
|
10008 fl = STRLEN(ftp->ft_from);
|
|
|
10009 tl = STRLEN(ftp->ft_to);
|
|
|
10010 if (fl != tl)
|
|
346
|
10011 {
|
|
323
|
10012 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
10013 repextra += tl - fl;
|
|
|
10014 }
|
|
323
|
10015 mch_memmove(p, ftp->ft_to, tl);
|
|
|
10016 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
10017 #ifdef FEAT_MBYTE
|
|
|
10018 stack[depth].ts_tcharlen = 0;
|
|
|
10019 #endif
|
|
323
|
10020 break;
|
|
|
10021 }
|
|
|
10022 }
|
|
|
10023
|
|
|
10024 if (sp->ts_curi >= gap->ga_len)
|
|
|
10025 /* No (more) matches. */
|
|
|
10026 sp->ts_state = STATE_FINAL;
|
|
|
10027
|
|
|
10028 break;
|
|
|
10029
|
|
|
10030 case STATE_REP_UNDO:
|
|
|
10031 /* Undo a REP replacement and continue with the next one. */
|
|
482
|
10032 ftp = (fromto_T *)slang->sl_rep.ga_data + sp->ts_curi - 1;
|
|
323
|
10033 fl = STRLEN(ftp->ft_from);
|
|
|
10034 tl = STRLEN(ftp->ft_to);
|
|
|
10035 p = fword + sp->ts_fidx;
|
|
|
10036 if (fl != tl)
|
|
346
|
10037 {
|
|
323
|
10038 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
10039 repextra -= tl - fl;
|
|
|
10040 }
|
|
323
|
10041 mch_memmove(p, ftp->ft_from, fl);
|
|
|
10042 sp->ts_state = STATE_REP;
|
|
|
10043 break;
|
|
|
10044
|
|
|
10045 default:
|
|
|
10046 /* Did all possible states at this level, go up one level. */
|
|
|
10047 --depth;
|
|
344
|
10048
|
|
482
|
10049 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
10050 {
|
|
|
10051 /* Continue in or go back to the prefix tree. */
|
|
|
10052 byts = pbyts;
|
|
|
10053 idxs = pidxs;
|
|
|
10054 }
|
|
|
10055
|
|
344
|
10056 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
10057 line_breakcheck();
|
|
323
|
10058 }
|
|
|
10059 }
|
|
|
10060 }
|
|
|
10061 }
|
|
|
10062
|
|
|
10063 /*
|
|
|
10064 * Try going one level deeper in the tree.
|
|
|
10065 */
|
|
|
10066 static int
|
|
|
10067 try_deeper(su, stack, depth, score_add)
|
|
|
10068 suginfo_T *su;
|
|
|
10069 trystate_T *stack;
|
|
|
10070 int depth;
|
|
|
10071 int score_add;
|
|
|
10072 {
|
|
|
10073 int newscore;
|
|
|
10074
|
|
|
10075 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
10076 newscore = stack[depth].ts_score + score_add;
|
|
|
10077 if (newscore >= su->su_maxscore)
|
|
|
10078 return FALSE;
|
|
|
10079
|
|
330
|
10080 stack[depth + 1] = stack[depth];
|
|
323
|
10081 stack[depth + 1].ts_state = STATE_START;
|
|
|
10082 stack[depth + 1].ts_score = newscore;
|
|
|
10083 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
485
|
10084 stack[depth + 1].ts_flags = 0;
|
|
323
|
10085 return TRUE;
|
|
|
10086 }
|
|
|
10087
|
|
455
|
10088 #ifdef FEAT_MBYTE
|
|
|
10089 /*
|
|
|
10090 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
10091 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
10092 */
|
|
|
10093 static int
|
|
|
10094 nofold_len(fword, flen, word)
|
|
|
10095 char_u *fword;
|
|
|
10096 int flen;
|
|
|
10097 char_u *word;
|
|
|
10098 {
|
|
|
10099 char_u *p;
|
|
|
10100 int i = 0;
|
|
|
10101
|
|
|
10102 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
10103 ++i;
|
|
|
10104 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
10105 --i;
|
|
|
10106 return (int)(p - word);
|
|
|
10107 }
|
|
|
10108 #endif
|
|
|
10109
|
|
323
|
10110 /*
|
|
|
10111 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
10112 * words and put it in "kword".
|
|
|
10113 * Theoretically there could be several keep-case words that result in the
|
|
|
10114 * same case-folded word, but we only find one...
|
|
|
10115 */
|
|
|
10116 static void
|
|
|
10117 find_keepcap_word(slang, fword, kword)
|
|
|
10118 slang_T *slang;
|
|
|
10119 char_u *fword;
|
|
|
10120 char_u *kword;
|
|
|
10121 {
|
|
|
10122 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
10123 int depth;
|
|
324
|
10124 idx_T tryidx;
|
|
323
|
10125
|
|
|
10126 /* The following arrays are used at each depth in the tree. */
|
|
324
|
10127 idx_T arridx[MAXWLEN];
|
|
323
|
10128 int round[MAXWLEN];
|
|
|
10129 int fwordidx[MAXWLEN];
|
|
|
10130 int uwordidx[MAXWLEN];
|
|
|
10131 int kwordlen[MAXWLEN];
|
|
|
10132
|
|
|
10133 int flen, ulen;
|
|
|
10134 int l;
|
|
|
10135 int len;
|
|
|
10136 int c;
|
|
324
|
10137 idx_T lo, hi, m;
|
|
323
|
10138 char_u *p;
|
|
|
10139 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
10140 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
10141
|
|
|
10142 if (byts == NULL)
|
|
|
10143 {
|
|
|
10144 /* array is empty: "cannot happen" */
|
|
|
10145 *kword = NUL;
|
|
|
10146 return;
|
|
|
10147 }
|
|
|
10148
|
|
|
10149 /* Make an all-cap version of "fword". */
|
|
|
10150 allcap_copy(fword, uword);
|
|
|
10151
|
|
|
10152 /*
|
|
|
10153 * Each character needs to be tried both case-folded and upper-case.
|
|
|
10154 * All this gets very complicated if we keep in mind that changing case
|
|
|
10155 * may change the byte length of a multi-byte character...
|
|
|
10156 */
|
|
|
10157 depth = 0;
|
|
|
10158 arridx[0] = 0;
|
|
|
10159 round[0] = 0;
|
|
|
10160 fwordidx[0] = 0;
|
|
|
10161 uwordidx[0] = 0;
|
|
|
10162 kwordlen[0] = 0;
|
|
|
10163 while (depth >= 0)
|
|
|
10164 {
|
|
|
10165 if (fword[fwordidx[depth]] == NUL)
|
|
|
10166 {
|
|
|
10167 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
10168 * here we have found a match. */
|
|
|
10169 if (byts[arridx[depth] + 1] == 0)
|
|
|
10170 {
|
|
|
10171 kword[kwordlen[depth]] = NUL;
|
|
|
10172 return;
|
|
|
10173 }
|
|
|
10174
|
|
|
10175 /* kword is getting too long, continue one level up */
|
|
|
10176 --depth;
|
|
|
10177 }
|
|
|
10178 else if (++round[depth] > 2)
|
|
|
10179 {
|
|
|
10180 /* tried both fold-case and upper-case character, continue one
|
|
|
10181 * level up */
|
|
|
10182 --depth;
|
|
|
10183 }
|
|
|
10184 else
|
|
|
10185 {
|
|
|
10186 /*
|
|
|
10187 * round[depth] == 1: Try using the folded-case character.
|
|
|
10188 * round[depth] == 2: Try using the upper-case character.
|
|
|
10189 */
|
|
|
10190 #ifdef FEAT_MBYTE
|
|
|
10191 if (has_mbyte)
|
|
|
10192 {
|
|
474
|
10193 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
10194 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
10195 }
|
|
|
10196 else
|
|
|
10197 #endif
|
|
|
10198 ulen = flen = 1;
|
|
|
10199 if (round[depth] == 1)
|
|
|
10200 {
|
|
|
10201 p = fword + fwordidx[depth];
|
|
|
10202 l = flen;
|
|
|
10203 }
|
|
|
10204 else
|
|
|
10205 {
|
|
|
10206 p = uword + uwordidx[depth];
|
|
|
10207 l = ulen;
|
|
|
10208 }
|
|
|
10209
|
|
|
10210 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
10211 {
|
|
|
10212 /* Perform a binary search in the list of accepted bytes. */
|
|
|
10213 len = byts[tryidx++];
|
|
|
10214 c = *p++;
|
|
|
10215 lo = tryidx;
|
|
|
10216 hi = tryidx + len - 1;
|
|
|
10217 while (lo < hi)
|
|
|
10218 {
|
|
|
10219 m = (lo + hi) / 2;
|
|
|
10220 if (byts[m] > c)
|
|
|
10221 hi = m - 1;
|
|
|
10222 else if (byts[m] < c)
|
|
|
10223 lo = m + 1;
|
|
|
10224 else
|
|
|
10225 {
|
|
|
10226 lo = hi = m;
|
|
|
10227 break;
|
|
|
10228 }
|
|
|
10229 }
|
|
|
10230
|
|
|
10231 /* Stop if there is no matching byte. */
|
|
|
10232 if (hi < lo || byts[lo] != c)
|
|
|
10233 break;
|
|
|
10234
|
|
|
10235 /* Continue at the child (if there is one). */
|
|
|
10236 tryidx = idxs[lo];
|
|
|
10237 }
|
|
|
10238
|
|
|
10239 if (l == 0)
|
|
|
10240 {
|
|
|
10241 /*
|
|
|
10242 * Found the matching char. Copy it to "kword" and go a
|
|
|
10243 * level deeper.
|
|
|
10244 */
|
|
|
10245 if (round[depth] == 1)
|
|
|
10246 {
|
|
|
10247 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
10248 flen);
|
|
|
10249 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
10250 }
|
|
|
10251 else
|
|
|
10252 {
|
|
|
10253 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
10254 ulen);
|
|
|
10255 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
10256 }
|
|
|
10257 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
10258 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
10259
|
|
|
10260 ++depth;
|
|
|
10261 arridx[depth] = tryidx;
|
|
|
10262 round[depth] = 0;
|
|
|
10263 }
|
|
|
10264 }
|
|
|
10265 }
|
|
|
10266
|
|
|
10267 /* Didn't find it: "cannot happen". */
|
|
|
10268 *kword = NUL;
|
|
|
10269 }
|
|
|
10270
|
|
|
10271 /*
|
|
344
|
10272 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
10273 * su->su_sga.
|
|
|
10274 */
|
|
|
10275 static void
|
|
|
10276 score_comp_sal(su)
|
|
|
10277 suginfo_T *su;
|
|
|
10278 {
|
|
|
10279 langp_T *lp;
|
|
|
10280 char_u badsound[MAXWLEN];
|
|
|
10281 int i;
|
|
|
10282 suggest_T *stp;
|
|
|
10283 suggest_T *sstp;
|
|
|
10284 int score;
|
|
|
10285
|
|
|
10286 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
10287 return;
|
|
|
10288
|
|
|
10289 /* Use the sound-folding of the first language that supports it. */
|
|
|
10290 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10291 lp->lp_slang != NULL; ++lp)
|
|
|
10292 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10293 {
|
|
|
10294 /* soundfold the bad word */
|
|
375
|
10295 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
10296
|
|
|
10297 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10298 {
|
|
|
10299 stp = &SUG(su->su_ga, i);
|
|
|
10300
|
|
351
|
10301 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
10302 * sound-a-like score. */
|
|
|
10303 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
10304 if (score < SCORE_MAXMAX)
|
|
|
10305 {
|
|
|
10306 /* Add the suggestion. */
|
|
|
10307 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
10308 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
10309 if (sstp->st_word != NULL)
|
|
|
10310 {
|
|
|
10311 sstp->st_score = score;
|
|
|
10312 sstp->st_altscore = 0;
|
|
|
10313 sstp->st_orglen = stp->st_orglen;
|
|
|
10314 ++su->su_sga.ga_len;
|
|
|
10315 }
|
|
|
10316 }
|
|
|
10317 }
|
|
|
10318 break;
|
|
|
10319 }
|
|
|
10320 }
|
|
|
10321
|
|
|
10322 /*
|
|
|
10323 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
10324 * They are intwined.
|
|
|
10325 */
|
|
|
10326 static void
|
|
|
10327 score_combine(su)
|
|
|
10328 suginfo_T *su;
|
|
|
10329 {
|
|
|
10330 int i;
|
|
|
10331 int j;
|
|
|
10332 garray_T ga;
|
|
|
10333 garray_T *gap;
|
|
|
10334 langp_T *lp;
|
|
|
10335 suggest_T *stp;
|
|
|
10336 char_u *p;
|
|
|
10337 char_u badsound[MAXWLEN];
|
|
|
10338 int round;
|
|
|
10339
|
|
|
10340 /* Add the alternate score to su_ga. */
|
|
|
10341 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10342 lp->lp_slang != NULL; ++lp)
|
|
|
10343 {
|
|
|
10344 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10345 {
|
|
|
10346 /* soundfold the bad word */
|
|
375
|
10347 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
10348
|
|
|
10349 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10350 {
|
|
|
10351 stp = &SUG(su->su_ga, i);
|
|
351
|
10352 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
10353 badsound);
|
|
344
|
10354 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
10355 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
10356 else
|
|
|
10357 stp->st_score = (stp->st_score * 3
|
|
|
10358 + stp->st_altscore) / 4;
|
|
|
10359 stp->st_salscore = FALSE;
|
|
|
10360 }
|
|
|
10361 break;
|
|
|
10362 }
|
|
|
10363 }
|
|
|
10364
|
|
|
10365 /* Add the alternate score to su_sga. */
|
|
|
10366 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
10367 {
|
|
|
10368 stp = &SUG(su->su_sga, i);
|
|
|
10369 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
10370 if (stp->st_score == SCORE_MAXMAX)
|
|
|
10371 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
10372 else
|
|
|
10373 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
10374 stp->st_salscore = TRUE;
|
|
|
10375 }
|
|
|
10376
|
|
|
10377 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
10378 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
10379 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
10380
|
|
|
10381 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
10382 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
10383 return;
|
|
|
10384
|
|
|
10385 stp = &SUG(ga, 0);
|
|
|
10386 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
10387 {
|
|
|
10388 /* round 1: get a suggestion from su_ga
|
|
|
10389 * round 2: get a suggestion from su_sga */
|
|
|
10390 for (round = 1; round <= 2; ++round)
|
|
|
10391 {
|
|
|
10392 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
10393 if (i < gap->ga_len)
|
|
|
10394 {
|
|
|
10395 /* Don't add a word if it's already there. */
|
|
|
10396 p = SUG(*gap, i).st_word;
|
|
|
10397 for (j = 0; j < ga.ga_len; ++j)
|
|
|
10398 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
10399 break;
|
|
|
10400 if (j == ga.ga_len)
|
|
|
10401 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
10402 else
|
|
|
10403 vim_free(p);
|
|
|
10404 }
|
|
|
10405 }
|
|
|
10406 }
|
|
|
10407
|
|
|
10408 ga_clear(&su->su_ga);
|
|
|
10409 ga_clear(&su->su_sga);
|
|
|
10410
|
|
|
10411 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
10412 if (ga.ga_len > su->su_maxcount)
|
|
|
10413 {
|
|
|
10414 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
10415 vim_free(stp[i].st_word);
|
|
|
10416 ga.ga_len = su->su_maxcount;
|
|
|
10417 }
|
|
|
10418
|
|
|
10419 su->su_ga = ga;
|
|
|
10420 }
|
|
|
10421
|
|
|
10422 /*
|
|
351
|
10423 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
10424 * badword.
|
|
|
10425 */
|
|
|
10426 static int
|
|
|
10427 stp_sal_score(stp, su, slang, badsound)
|
|
|
10428 suggest_T *stp;
|
|
|
10429 suginfo_T *su;
|
|
|
10430 slang_T *slang;
|
|
|
10431 char_u *badsound; /* sound-folded badword */
|
|
|
10432 {
|
|
|
10433 char_u *p;
|
|
|
10434 char_u badsound2[MAXWLEN];
|
|
|
10435 char_u fword[MAXWLEN];
|
|
|
10436 char_u goodsound[MAXWLEN];
|
|
|
10437
|
|
|
10438 if (stp->st_orglen <= su->su_badlen)
|
|
|
10439 p = badsound;
|
|
|
10440 else
|
|
|
10441 {
|
|
|
10442 /* soundfold the bad word with more characters following */
|
|
|
10443 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
10444
|
|
|
10445 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
10446 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
10447 * removing the space. Don't do it when the good word also contains a
|
|
|
10448 * space. */
|
|
|
10449 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
10450 && *skiptowhite(stp->st_word) == NUL)
|
|
|
10451 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
10452 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
10453
|
|
375
|
10454 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
10455 p = badsound2;
|
|
|
10456 }
|
|
|
10457
|
|
375
|
10458 /* Sound-fold the word and compute the score for the difference. */
|
|
|
10459 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
10460
|
|
|
10461 return soundalike_score(goodsound, p);
|
|
|
10462 }
|
|
|
10463
|
|
|
10464 /*
|
|
323
|
10465 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
10466 */
|
|
|
10467 static void
|
|
346
|
10468 suggest_try_soundalike(su)
|
|
323
|
10469 suginfo_T *su;
|
|
|
10470 {
|
|
|
10471 char_u salword[MAXWLEN];
|
|
|
10472 char_u tword[MAXWLEN];
|
|
|
10473 char_u tsalword[MAXWLEN];
|
|
324
|
10474 idx_T arridx[MAXWLEN];
|
|
323
|
10475 int curi[MAXWLEN];
|
|
|
10476 langp_T *lp;
|
|
|
10477 char_u *byts;
|
|
324
|
10478 idx_T *idxs;
|
|
323
|
10479 int depth;
|
|
|
10480 int c;
|
|
324
|
10481 idx_T n;
|
|
323
|
10482 int round;
|
|
|
10483 int flags;
|
|
344
|
10484 int sound_score;
|
|
483
|
10485 int local_score;
|
|
344
|
10486
|
|
|
10487 /* Do this for all languages that support sound folding. */
|
|
323
|
10488 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10489 lp->lp_slang != NULL; ++lp)
|
|
|
10490 {
|
|
|
10491 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10492 {
|
|
|
10493 /* soundfold the bad word */
|
|
375
|
10494 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
10495
|
|
|
10496 /*
|
|
|
10497 * Go through the whole tree, soundfold each word and compare.
|
|
|
10498 * round 1: use the case-folded tree.
|
|
|
10499 * round 2: use the keep-case tree.
|
|
|
10500 */
|
|
|
10501 for (round = 1; round <= 2; ++round)
|
|
|
10502 {
|
|
|
10503 if (round == 1)
|
|
|
10504 {
|
|
|
10505 byts = lp->lp_slang->sl_fbyts;
|
|
|
10506 idxs = lp->lp_slang->sl_fidxs;
|
|
|
10507 }
|
|
|
10508 else
|
|
|
10509 {
|
|
|
10510 byts = lp->lp_slang->sl_kbyts;
|
|
|
10511 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
10512 if (byts == NULL) /* no keep-case words */
|
|
|
10513 continue;
|
|
323
|
10514 }
|
|
|
10515
|
|
|
10516 depth = 0;
|
|
|
10517 arridx[0] = 0;
|
|
|
10518 curi[0] = 1;
|
|
|
10519 while (depth >= 0 && !got_int)
|
|
|
10520 {
|
|
|
10521 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
10522 {
|
|
323
|
10523 /* Done all bytes at this node, go up one level. */
|
|
|
10524 --depth;
|
|
351
|
10525 line_breakcheck();
|
|
|
10526 }
|
|
323
|
10527 else
|
|
|
10528 {
|
|
|
10529 /* Do one more byte at this node. */
|
|
|
10530 n = arridx[depth] + curi[depth];
|
|
|
10531 ++curi[depth];
|
|
|
10532 c = byts[n];
|
|
|
10533 if (c == 0)
|
|
|
10534 {
|
|
|
10535 /* End of word, deal with the word. */
|
|
324
|
10536 flags = (int)idxs[n];
|
|
323
|
10537 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
10538 {
|
|
|
10539 tword[depth] = NUL;
|
|
375
|
10540 /* Sound-fold. Only in keep-case tree need to
|
|
|
10541 * case-fold the word. */
|
|
|
10542 spell_soundfold(lp->lp_slang, tword,
|
|
|
10543 round == 1, tsalword);
|
|
323
|
10544
|
|
344
|
10545 /* Compute the edit distance between the
|
|
|
10546 * sound-a-like words. */
|
|
|
10547 sound_score = soundalike_score(salword,
|
|
|
10548 tsalword);
|
|
483
|
10549
|
|
|
10550 /* Add a penalty for words in another region. */
|
|
|
10551 if ((flags & WF_REGION) && (((unsigned)flags
|
|
|
10552 >> 16) & lp->lp_region) == 0)
|
|
|
10553 local_score = SCORE_REGION;
|
|
|
10554 else
|
|
|
10555 local_score = 0;
|
|
|
10556 sound_score += local_score;
|
|
|
10557
|
|
324
|
10558 if (sound_score < SCORE_MAXMAX)
|
|
323
|
10559 {
|
|
324
|
10560 char_u cword[MAXWLEN];
|
|
|
10561 char_u *p;
|
|
344
|
10562 int score;
|
|
324
|
10563
|
|
388
|
10564 flags |= su->su_badflags;
|
|
351
|
10565 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
10566 {
|
|
324
|
10567 /* Need to fix case according to
|
|
|
10568 * "flags". */
|
|
323
|
10569 make_case_word(tword, cword, flags);
|
|
324
|
10570 p = cword;
|
|
323
|
10571 }
|
|
|
10572 else
|
|
324
|
10573 p = tword;
|
|
|
10574
|
|
344
|
10575 if (sps_flags & SPS_DOUBLE)
|
|
|
10576 add_suggestion(su, &su->su_sga, p,
|
|
346
|
10577 su->su_badlen,
|
|
351
|
10578 sound_score, 0, FALSE);
|
|
344
|
10579 else
|
|
|
10580 {
|
|
|
10581 /* Compute the score. */
|
|
|
10582 score = spell_edit_score(
|
|
483
|
10583 su->su_badword, p)
|
|
|
10584 + local_score;
|
|
344
|
10585 if (sps_flags & SPS_BEST)
|
|
|
10586 /* give a bonus for the good word
|
|
|
10587 * sounding the same as the bad
|
|
|
10588 * word */
|
|
|
10589 add_suggestion(su, &su->su_ga, p,
|
|
346
|
10590 su->su_badlen,
|
|
344
|
10591 RESCORE(score, sound_score),
|
|
351
|
10592 sound_score, TRUE);
|
|
344
|
10593 else
|
|
|
10594 add_suggestion(su, &su->su_ga, p,
|
|
346
|
10595 su->su_badlen,
|
|
351
|
10596 score + sound_score, 0, FALSE);
|
|
344
|
10597 }
|
|
323
|
10598 }
|
|
|
10599 }
|
|
|
10600
|
|
|
10601 /* Skip over other NUL bytes. */
|
|
|
10602 while (byts[n + 1] == 0)
|
|
|
10603 {
|
|
|
10604 ++n;
|
|
|
10605 ++curi[depth];
|
|
|
10606 }
|
|
|
10607 }
|
|
|
10608 else
|
|
|
10609 {
|
|
|
10610 /* Normal char, go one level deeper. */
|
|
|
10611 tword[depth++] = c;
|
|
|
10612 arridx[depth] = idxs[n];
|
|
|
10613 curi[depth] = 1;
|
|
|
10614 }
|
|
|
10615 }
|
|
|
10616 }
|
|
|
10617 }
|
|
|
10618 }
|
|
|
10619 }
|
|
|
10620 }
|
|
|
10621
|
|
|
10622 /*
|
|
324
|
10623 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
10624 */
|
|
|
10625 static void
|
|
|
10626 make_case_word(fword, cword, flags)
|
|
|
10627 char_u *fword;
|
|
|
10628 char_u *cword;
|
|
|
10629 int flags;
|
|
|
10630 {
|
|
|
10631 if (flags & WF_ALLCAP)
|
|
|
10632 /* Make it all upper-case */
|
|
|
10633 allcap_copy(fword, cword);
|
|
|
10634 else if (flags & WF_ONECAP)
|
|
|
10635 /* Make the first letter upper-case */
|
|
324
|
10636 onecap_copy(fword, cword, TRUE);
|
|
323
|
10637 else
|
|
|
10638 /* Use goodword as-is. */
|
|
|
10639 STRCPY(cword, fword);
|
|
|
10640 }
|
|
|
10641
|
|
330
|
10642 /*
|
|
|
10643 * Use map string "map" for languages "lp".
|
|
|
10644 */
|
|
|
10645 static void
|
|
|
10646 set_map_str(lp, map)
|
|
|
10647 slang_T *lp;
|
|
|
10648 char_u *map;
|
|
|
10649 {
|
|
|
10650 char_u *p;
|
|
|
10651 int headc = 0;
|
|
|
10652 int c;
|
|
|
10653 int i;
|
|
|
10654
|
|
|
10655 if (*map == NUL)
|
|
|
10656 {
|
|
|
10657 lp->sl_has_map = FALSE;
|
|
|
10658 return;
|
|
|
10659 }
|
|
|
10660 lp->sl_has_map = TRUE;
|
|
|
10661
|
|
|
10662 /* Init the array and hash table empty. */
|
|
|
10663 for (i = 0; i < 256; ++i)
|
|
|
10664 lp->sl_map_array[i] = 0;
|
|
|
10665 #ifdef FEAT_MBYTE
|
|
|
10666 hash_init(&lp->sl_map_hash);
|
|
|
10667 #endif
|
|
|
10668
|
|
|
10669 /*
|
|
|
10670 * The similar characters are stored separated with slashes:
|
|
|
10671 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
10672 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
10673 */
|
|
|
10674 for (p = map; *p != NUL; )
|
|
|
10675 {
|
|
|
10676 #ifdef FEAT_MBYTE
|
|
474
|
10677 c = mb_cptr2char_adv(&p);
|
|
330
|
10678 #else
|
|
|
10679 c = *p++;
|
|
|
10680 #endif
|
|
|
10681 if (c == '/')
|
|
|
10682 headc = 0;
|
|
|
10683 else
|
|
|
10684 {
|
|
|
10685 if (headc == 0)
|
|
|
10686 headc = c;
|
|
|
10687
|
|
|
10688 #ifdef FEAT_MBYTE
|
|
|
10689 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
10690 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
10691 * a NUL. */
|
|
|
10692 if (c >= 256)
|
|
|
10693 {
|
|
|
10694 int cl = mb_char2len(c);
|
|
|
10695 int headcl = mb_char2len(headc);
|
|
|
10696 char_u *b;
|
|
|
10697 hash_T hash;
|
|
|
10698 hashitem_T *hi;
|
|
|
10699
|
|
|
10700 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
10701 if (b == NULL)
|
|
|
10702 return;
|
|
|
10703 mb_char2bytes(c, b);
|
|
|
10704 b[cl] = NUL;
|
|
|
10705 mb_char2bytes(headc, b + cl + 1);
|
|
|
10706 b[cl + 1 + headcl] = NUL;
|
|
|
10707 hash = hash_hash(b);
|
|
|
10708 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
10709 if (HASHITEM_EMPTY(hi))
|
|
|
10710 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
10711 else
|
|
|
10712 {
|
|
|
10713 /* This should have been checked when generating the .spl
|
|
|
10714 * file. */
|
|
|
10715 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
10716 vim_free(b);
|
|
|
10717 }
|
|
|
10718 }
|
|
|
10719 else
|
|
|
10720 #endif
|
|
|
10721 lp->sl_map_array[c] = headc;
|
|
|
10722 }
|
|
|
10723 }
|
|
|
10724 }
|
|
|
10725
|
|
323
|
10726 /*
|
|
|
10727 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
10728 * lines in the .aff file.
|
|
|
10729 */
|
|
|
10730 static int
|
|
|
10731 similar_chars(slang, c1, c2)
|
|
|
10732 slang_T *slang;
|
|
|
10733 int c1;
|
|
|
10734 int c2;
|
|
|
10735 {
|
|
330
|
10736 int m1, m2;
|
|
|
10737 #ifdef FEAT_MBYTE
|
|
|
10738 char_u buf[MB_MAXBYTES];
|
|
|
10739 hashitem_T *hi;
|
|
|
10740
|
|
|
10741 if (c1 >= 256)
|
|
|
10742 {
|
|
|
10743 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
10744 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
10745 if (HASHITEM_EMPTY(hi))
|
|
|
10746 m1 = 0;
|
|
|
10747 else
|
|
|
10748 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
10749 }
|
|
|
10750 else
|
|
|
10751 #endif
|
|
|
10752 m1 = slang->sl_map_array[c1];
|
|
|
10753 if (m1 == 0)
|
|
323
|
10754 return FALSE;
|
|
330
|
10755
|
|
|
10756
|
|
|
10757 #ifdef FEAT_MBYTE
|
|
|
10758 if (c2 >= 256)
|
|
|
10759 {
|
|
|
10760 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
10761 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
10762 if (HASHITEM_EMPTY(hi))
|
|
|
10763 m2 = 0;
|
|
|
10764 else
|
|
|
10765 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
10766 }
|
|
|
10767 else
|
|
|
10768 #endif
|
|
|
10769 m2 = slang->sl_map_array[c2];
|
|
|
10770
|
|
|
10771 return m1 == m2;
|
|
323
|
10772 }
|
|
|
10773
|
|
|
10774 /*
|
|
|
10775 * Add a suggestion to the list of suggestions.
|
|
|
10776 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
10777 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
10778 * with spell_edit_score().
|
|
|
10779 */
|
|
|
10780 static void
|
|
351
|
10781 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
10782 suginfo_T *su;
|
|
344
|
10783 garray_T *gap;
|
|
323
|
10784 char_u *goodword;
|
|
346
|
10785 int badlen; /* length of bad word used */
|
|
324
|
10786 int score;
|
|
351
|
10787 int altscore;
|
|
344
|
10788 int had_bonus; /* value for st_had_bonus */
|
|
323
|
10789 {
|
|
|
10790 suggest_T *stp;
|
|
|
10791 int i;
|
|
346
|
10792 char_u *p = NULL;
|
|
|
10793 int c = 0;
|
|
323
|
10794
|
|
|
10795 /* Check that the word wasn't banned. */
|
|
|
10796 if (was_banned(su, goodword))
|
|
|
10797 return;
|
|
|
10798
|
|
346
|
10799 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
10800 * Remove the common part from "goodword". */
|
|
|
10801 i = badlen - su->su_badlen;
|
|
|
10802 if (i > 0)
|
|
|
10803 {
|
|
|
10804 /* This assumes there was no case folding or it didn't change the
|
|
|
10805 * length... */
|
|
|
10806 p = goodword + STRLEN(goodword) - i;
|
|
|
10807 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
10808 {
|
|
|
10809 badlen = su->su_badlen;
|
|
|
10810 c = *p;
|
|
|
10811 *p = NUL;
|
|
|
10812 }
|
|
|
10813 else
|
|
|
10814 p = NULL;
|
|
|
10815 }
|
|
372
|
10816 else if (i < 0)
|
|
|
10817 {
|
|
|
10818 /* When replacing part of the word check that we actually change
|
|
|
10819 * something. For "the the" a suggestion can be replacing the first
|
|
|
10820 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
10821 if (badlen == (int)STRLEN(goodword)
|
|
372
|
10822 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
10823 return;
|
|
|
10824 }
|
|
|
10825
|
|
346
|
10826
|
|
323
|
10827 if (score <= su->su_maxscore)
|
|
|
10828 {
|
|
366
|
10829 /* Check if the word is already there. Also check the length that is
|
|
|
10830 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
10831 * "thes" -> "these". */
|
|
344
|
10832 stp = &SUG(*gap, 0);
|
|
|
10833 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
10834 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
10835 && stp[i].st_orglen == badlen)
|
|
323
|
10836 {
|
|
|
10837 /* Found it. Remember the lowest score. */
|
|
|
10838 if (stp[i].st_score > score)
|
|
324
|
10839 {
|
|
323
|
10840 stp[i].st_score = score;
|
|
376
|
10841 stp[i].st_altscore = altscore;
|
|
324
|
10842 stp[i].st_had_bonus = had_bonus;
|
|
|
10843 }
|
|
323
|
10844 break;
|
|
|
10845 }
|
|
|
10846
|
|
344
|
10847 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
10848 {
|
|
|
10849 /* Add a suggestion. */
|
|
344
|
10850 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
10851 stp->st_word = vim_strsave(goodword);
|
|
|
10852 if (stp->st_word != NULL)
|
|
|
10853 {
|
|
|
10854 stp->st_score = score;
|
|
351
|
10855 stp->st_altscore = altscore;
|
|
324
|
10856 stp->st_had_bonus = had_bonus;
|
|
346
|
10857 stp->st_orglen = badlen;
|
|
344
|
10858 ++gap->ga_len;
|
|
323
|
10859
|
|
|
10860 /* If we have too many suggestions now, sort the list and keep
|
|
|
10861 * the best suggestions. */
|
|
344
|
10862 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
10863 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
10864 SUG_CLEAN_COUNT(su));
|
|
323
|
10865 }
|
|
|
10866 }
|
|
|
10867 }
|
|
346
|
10868
|
|
|
10869 if (p != NULL)
|
|
|
10870 *p = c; /* restore "goodword" */
|
|
323
|
10871 }
|
|
|
10872
|
|
|
10873 /*
|
|
|
10874 * Add a word to be banned.
|
|
|
10875 */
|
|
|
10876 static void
|
|
|
10877 add_banned(su, word)
|
|
|
10878 suginfo_T *su;
|
|
|
10879 char_u *word;
|
|
|
10880 {
|
|
|
10881 char_u *s = vim_strsave(word);
|
|
|
10882 hash_T hash;
|
|
|
10883 hashitem_T *hi;
|
|
|
10884
|
|
|
10885 if (s != NULL)
|
|
|
10886 {
|
|
|
10887 hash = hash_hash(s);
|
|
|
10888 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
10889 if (HASHITEM_EMPTY(hi))
|
|
|
10890 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
10891 else
|
|
|
10892 vim_free(s);
|
|
323
|
10893 }
|
|
|
10894 }
|
|
|
10895
|
|
|
10896 /*
|
|
|
10897 * Return TRUE if a word appears in the list of banned words.
|
|
|
10898 */
|
|
|
10899 static int
|
|
|
10900 was_banned(su, word)
|
|
|
10901 suginfo_T *su;
|
|
|
10902 char_u *word;
|
|
|
10903 {
|
|
324
|
10904 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
10905
|
|
|
10906 return !HASHITEM_EMPTY(hi);
|
|
323
|
10907 }
|
|
|
10908
|
|
|
10909 /*
|
|
|
10910 * Free the banned words in "su".
|
|
|
10911 */
|
|
|
10912 static void
|
|
|
10913 free_banned(su)
|
|
|
10914 suginfo_T *su;
|
|
|
10915 {
|
|
|
10916 int todo;
|
|
|
10917 hashitem_T *hi;
|
|
|
10918
|
|
|
10919 todo = su->su_banned.ht_used;
|
|
|
10920 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
10921 {
|
|
|
10922 if (!HASHITEM_EMPTY(hi))
|
|
|
10923 {
|
|
|
10924 vim_free(hi->hi_key);
|
|
|
10925 --todo;
|
|
|
10926 }
|
|
|
10927 }
|
|
|
10928 hash_clear(&su->su_banned);
|
|
|
10929 }
|
|
|
10930
|
|
324
|
10931 /*
|
|
|
10932 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
10933 * thus only done for the final results.
|
|
|
10934 */
|
|
|
10935 static void
|
|
|
10936 rescore_suggestions(su)
|
|
|
10937 suginfo_T *su;
|
|
|
10938 {
|
|
|
10939 langp_T *lp;
|
|
|
10940 suggest_T *stp;
|
|
|
10941 char_u sal_badword[MAXWLEN];
|
|
|
10942 int i;
|
|
|
10943
|
|
|
10944 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10945 lp->lp_slang != NULL; ++lp)
|
|
|
10946 {
|
|
|
10947 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10948 {
|
|
|
10949 /* soundfold the bad word */
|
|
375
|
10950 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
10951
|
|
|
10952 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10953 {
|
|
344
|
10954 stp = &SUG(su->su_ga, i);
|
|
324
|
10955 if (!stp->st_had_bonus)
|
|
|
10956 {
|
|
351
|
10957 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
10958 lp->lp_slang, sal_badword);
|
|
|
10959 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
10960 stp->st_altscore = SCORE_BIG;
|
|
|
10961 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
10962 }
|
|
|
10963 }
|
|
|
10964 break;
|
|
|
10965 }
|
|
|
10966 }
|
|
|
10967 }
|
|
|
10968
|
|
323
|
10969 static int
|
|
|
10970 #ifdef __BORLANDC__
|
|
|
10971 _RTLENTRYF
|
|
|
10972 #endif
|
|
|
10973 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
10974
|
|
|
10975 /*
|
|
|
10976 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
10977 */
|
|
|
10978 static int
|
|
|
10979 #ifdef __BORLANDC__
|
|
|
10980 _RTLENTRYF
|
|
|
10981 #endif
|
|
|
10982 sug_compare(s1, s2)
|
|
|
10983 const void *s1;
|
|
|
10984 const void *s2;
|
|
|
10985 {
|
|
|
10986 suggest_T *p1 = (suggest_T *)s1;
|
|
|
10987 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
10988 int n = p1->st_score - p2->st_score;
|
|
|
10989
|
|
|
10990 if (n == 0)
|
|
|
10991 return p1->st_altscore - p2->st_altscore;
|
|
|
10992 return n;
|
|
323
|
10993 }
|
|
|
10994
|
|
|
10995 /*
|
|
|
10996 * Cleanup the suggestions:
|
|
|
10997 * - Sort on score.
|
|
|
10998 * - Remove words that won't be displayed.
|
|
344
|
10999 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
11000 */
|
|
344
|
11001 static int
|
|
|
11002 cleanup_suggestions(gap, maxscore, keep)
|
|
|
11003 garray_T *gap;
|
|
|
11004 int maxscore;
|
|
324
|
11005 int keep; /* nr of suggestions to keep */
|
|
323
|
11006 {
|
|
344
|
11007 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
11008 int i;
|
|
|
11009
|
|
|
11010 /* Sort the list. */
|
|
344
|
11011 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
11012
|
|
|
11013 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
11014 if (gap->ga_len > keep)
|
|
323
|
11015 {
|
|
344
|
11016 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
11017 vim_free(stp[i].st_word);
|
|
344
|
11018 gap->ga_len = keep;
|
|
|
11019 return stp[keep - 1].st_score;
|
|
323
|
11020 }
|
|
344
|
11021 return maxscore;
|
|
323
|
11022 }
|
|
|
11023
|
|
372
|
11024 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
11025 /*
|
|
|
11026 * Soundfold a string, for soundfold().
|
|
|
11027 * Result is in allocated memory, NULL for an error.
|
|
|
11028 */
|
|
|
11029 char_u *
|
|
|
11030 eval_soundfold(word)
|
|
|
11031 char_u *word;
|
|
|
11032 {
|
|
|
11033 langp_T *lp;
|
|
|
11034 char_u sound[MAXWLEN];
|
|
|
11035
|
|
|
11036 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
11037 /* Use the sound-folding of the first language that supports it. */
|
|
|
11038 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
11039 lp->lp_slang != NULL; ++lp)
|
|
|
11040 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
11041 {
|
|
|
11042 /* soundfold the word */
|
|
375
|
11043 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
11044 return vim_strsave(sound);
|
|
|
11045 }
|
|
|
11046
|
|
|
11047 /* No language with sound folding, return word as-is. */
|
|
|
11048 return vim_strsave(word);
|
|
|
11049 }
|
|
|
11050 #endif
|
|
|
11051
|
|
323
|
11052 /*
|
|
|
11053 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
485
|
11054 *
|
|
|
11055 * There are many ways to turn a word into a sound-a-like representation. The
|
|
|
11056 * oldest is Soundex (1918!). A nice overview can be found in "Approximate
|
|
|
11057 * swedish name matching - survey and test of different algorithms" by Klas
|
|
|
11058 * Erikson.
|
|
|
11059 *
|
|
|
11060 * We support two methods:
|
|
|
11061 * 1. SOFOFROM/SOFOTO do a simple character mapping.
|
|
|
11062 * 2. SAL items define a more advanced sound-folding (and much slower).
|
|
323
|
11063 */
|
|
|
11064 static void
|
|
375
|
11065 spell_soundfold(slang, inword, folded, res)
|
|
|
11066 slang_T *slang;
|
|
|
11067 char_u *inword;
|
|
|
11068 int folded; /* "inword" is already case-folded */
|
|
|
11069 char_u *res;
|
|
|
11070 {
|
|
|
11071 char_u fword[MAXWLEN];
|
|
|
11072 char_u *word;
|
|
|
11073
|
|
|
11074 if (slang->sl_sofo)
|
|
|
11075 /* SOFOFROM and SOFOTO used */
|
|
|
11076 spell_soundfold_sofo(slang, inword, res);
|
|
|
11077 else
|
|
|
11078 {
|
|
|
11079 /* SAL items used. Requires the word to be case-folded. */
|
|
|
11080 if (folded)
|
|
|
11081 word = inword;
|
|
|
11082 else
|
|
|
11083 {
|
|
|
11084 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
11085 word = fword;
|
|
|
11086 }
|
|
|
11087
|
|
|
11088 #ifdef FEAT_MBYTE
|
|
|
11089 if (has_mbyte)
|
|
|
11090 spell_soundfold_wsal(slang, word, res);
|
|
|
11091 else
|
|
|
11092 #endif
|
|
|
11093 spell_soundfold_sal(slang, word, res);
|
|
|
11094 }
|
|
|
11095 }
|
|
|
11096
|
|
|
11097 /*
|
|
|
11098 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
11099 * SOFOTO lines.
|
|
|
11100 */
|
|
|
11101 static void
|
|
|
11102 spell_soundfold_sofo(slang, inword, res)
|
|
|
11103 slang_T *slang;
|
|
|
11104 char_u *inword;
|
|
|
11105 char_u *res;
|
|
|
11106 {
|
|
|
11107 char_u *s;
|
|
|
11108 int ri = 0;
|
|
|
11109 int c;
|
|
|
11110
|
|
|
11111 #ifdef FEAT_MBYTE
|
|
|
11112 if (has_mbyte)
|
|
|
11113 {
|
|
|
11114 int prevc = 0;
|
|
|
11115 int *ip;
|
|
|
11116
|
|
|
11117 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
11118 * 255, sl_sal the rest. */
|
|
|
11119 for (s = inword; *s != NUL; )
|
|
|
11120 {
|
|
474
|
11121 c = mb_cptr2char_adv(&s);
|
|
375
|
11122 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
11123 c = ' ';
|
|
|
11124 else if (c < 256)
|
|
|
11125 c = slang->sl_sal_first[c];
|
|
|
11126 else
|
|
|
11127 {
|
|
|
11128 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
11129 if (ip == NULL) /* empty list, can't match */
|
|
|
11130 c = NUL;
|
|
|
11131 else
|
|
|
11132 for (;;) /* find "c" in the list */
|
|
|
11133 {
|
|
|
11134 if (*ip == 0) /* not found */
|
|
|
11135 {
|
|
|
11136 c = NUL;
|
|
|
11137 break;
|
|
|
11138 }
|
|
|
11139 if (*ip == c) /* match! */
|
|
|
11140 {
|
|
|
11141 c = ip[1];
|
|
|
11142 break;
|
|
|
11143 }
|
|
|
11144 ip += 2;
|
|
|
11145 }
|
|
|
11146 }
|
|
|
11147
|
|
|
11148 if (c != NUL && c != prevc)
|
|
|
11149 {
|
|
|
11150 ri += mb_char2bytes(c, res + ri);
|
|
|
11151 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
11152 break;
|
|
|
11153 prevc = c;
|
|
|
11154 }
|
|
|
11155 }
|
|
|
11156 }
|
|
|
11157 else
|
|
|
11158 #endif
|
|
|
11159 {
|
|
|
11160 /* The sl_sal_first[] table contains the translation. */
|
|
|
11161 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
11162 {
|
|
|
11163 if (vim_iswhite(c))
|
|
|
11164 c = ' ';
|
|
|
11165 else
|
|
|
11166 c = slang->sl_sal_first[c];
|
|
|
11167 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
11168 res[ri++] = c;
|
|
|
11169 }
|
|
|
11170 }
|
|
|
11171
|
|
|
11172 res[ri] = NUL;
|
|
|
11173 }
|
|
|
11174
|
|
|
11175 static void
|
|
|
11176 spell_soundfold_sal(slang, inword, res)
|
|
323
|
11177 slang_T *slang;
|
|
|
11178 char_u *inword;
|
|
|
11179 char_u *res;
|
|
|
11180 {
|
|
344
|
11181 salitem_T *smp;
|
|
323
|
11182 char_u word[MAXWLEN];
|
|
375
|
11183 char_u *s = inword;
|
|
323
|
11184 char_u *t;
|
|
344
|
11185 char_u *pf;
|
|
323
|
11186 int i, j, z;
|
|
344
|
11187 int reslen;
|
|
323
|
11188 int n, k = 0;
|
|
|
11189 int z0;
|
|
|
11190 int k0;
|
|
|
11191 int n0;
|
|
|
11192 int c;
|
|
|
11193 int pri;
|
|
|
11194 int p0 = -333;
|
|
|
11195 int c0;
|
|
|
11196
|
|
324
|
11197 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
11198 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
11199 if (slang->sl_rem_accents)
|
|
|
11200 {
|
|
|
11201 t = word;
|
|
375
|
11202 while (*s != NUL)
|
|
323
|
11203 {
|
|
324
|
11204 if (vim_iswhite(*s))
|
|
344
|
11205 {
|
|
|
11206 *t++ = ' ';
|
|
|
11207 s = skipwhite(s);
|
|
|
11208 }
|
|
323
|
11209 else
|
|
|
11210 {
|
|
376
|
11211 if (spell_iswordp_nmw(s))
|
|
323
|
11212 *t++ = *s;
|
|
|
11213 ++s;
|
|
|
11214 }
|
|
|
11215 }
|
|
|
11216 *t = NUL;
|
|
|
11217 }
|
|
|
11218 else
|
|
375
|
11219 STRCPY(word, s);
|
|
323
|
11220
|
|
344
|
11221 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
11222
|
|
|
11223 /*
|
|
|
11224 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
11225 * Changed to keep spaces.
|
|
323
|
11226 */
|
|
344
|
11227 i = reslen = z = 0;
|
|
323
|
11228 while ((c = word[i]) != NUL)
|
|
|
11229 {
|
|
344
|
11230 /* Start with the first rule that has the character in the word. */
|
|
323
|
11231 n = slang->sl_sal_first[c];
|
|
|
11232 z0 = 0;
|
|
|
11233
|
|
|
11234 if (n >= 0)
|
|
|
11235 {
|
|
|
11236 /* check all rules for the same letter */
|
|
344
|
11237 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
11238 {
|
|
344
|
11239 /* Quickly skip entries that don't match the word. Most
|
|
|
11240 * entries are less then three chars, optimize for that. */
|
|
|
11241 k = smp[n].sm_leadlen;
|
|
|
11242 if (k > 1)
|
|
323
|
11243 {
|
|
344
|
11244 if (word[i + 1] != s[1])
|
|
|
11245 continue;
|
|
|
11246 if (k > 2)
|
|
|
11247 {
|
|
|
11248 for (j = 2; j < k; ++j)
|
|
|
11249 if (word[i + j] != s[j])
|
|
|
11250 break;
|
|
|
11251 if (j < k)
|
|
|
11252 continue;
|
|
|
11253 }
|
|
323
|
11254 }
|
|
|
11255
|
|
375
|
11256 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
11257 {
|
|
375
|
11258 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
11259 while (*pf != NUL && *pf != word[i + k])
|
|
|
11260 ++pf;
|
|
|
11261 if (*pf == NUL)
|
|
|
11262 continue;
|
|
|
11263 ++k;
|
|
323
|
11264 }
|
|
344
|
11265 s = smp[n].sm_rules;
|
|
|
11266 pri = 5; /* default priority */
|
|
323
|
11267
|
|
|
11268 p0 = *s;
|
|
|
11269 k0 = k;
|
|
|
11270 while (*s == '-' && k > 1)
|
|
|
11271 {
|
|
|
11272 k--;
|
|
|
11273 s++;
|
|
|
11274 }
|
|
|
11275 if (*s == '<')
|
|
|
11276 s++;
|
|
344
|
11277 if (VIM_ISDIGIT(*s))
|
|
323
|
11278 {
|
|
|
11279 /* determine priority */
|
|
|
11280 pri = *s - '0';
|
|
|
11281 s++;
|
|
|
11282 }
|
|
|
11283 if (*s == '^' && *(s + 1) == '^')
|
|
|
11284 s++;
|
|
|
11285
|
|
|
11286 if (*s == NUL
|
|
|
11287 || (*s == '^'
|
|
324
|
11288 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
11289 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
11290 && (*(s + 1) != '$'
|
|
376
|
11291 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
11292 || (*s == '$' && i > 0
|
|
376
|
11293 && spell_iswordp(word + i - 1, curbuf)
|
|
|
11294 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
11295 {
|
|
|
11296 /* search for followup rules, if: */
|
|
|
11297 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
11298 c0 = word[i + k - 1];
|
|
|
11299 n0 = slang->sl_sal_first[c0];
|
|
|
11300
|
|
|
11301 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
11302 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
11303 {
|
|
|
11304 /* test follow-up rule for "word[i + k]" */
|
|
344
|
11305 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
11306 {
|
|
344
|
11307 /* Quickly skip entries that don't match the word.
|
|
|
11308 * */
|
|
|
11309 k0 = smp[n0].sm_leadlen;
|
|
|
11310 if (k0 > 1)
|
|
323
|
11311 {
|
|
344
|
11312 if (word[i + k] != s[1])
|
|
|
11313 continue;
|
|
|
11314 if (k0 > 2)
|
|
|
11315 {
|
|
|
11316 pf = word + i + k + 1;
|
|
|
11317 for (j = 2; j < k0; ++j)
|
|
|
11318 if (*pf++ != s[j])
|
|
|
11319 break;
|
|
|
11320 if (j < k0)
|
|
|
11321 continue;
|
|
|
11322 }
|
|
323
|
11323 }
|
|
344
|
11324 k0 += k - 1;
|
|
|
11325
|
|
375
|
11326 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
11327 {
|
|
344
|
11328 /* Check for match with one of the chars in
|
|
375
|
11329 * "sm_oneof". */
|
|
344
|
11330 while (*pf != NUL && *pf != word[i + k0])
|
|
|
11331 ++pf;
|
|
|
11332 if (*pf == NUL)
|
|
|
11333 continue;
|
|
|
11334 ++k0;
|
|
323
|
11335 }
|
|
344
|
11336
|
|
|
11337 p0 = 5;
|
|
|
11338 s = smp[n0].sm_rules;
|
|
323
|
11339 while (*s == '-')
|
|
|
11340 {
|
|
344
|
11341 /* "k0" gets NOT reduced because
|
|
|
11342 * "if (k0 == k)" */
|
|
323
|
11343 s++;
|
|
|
11344 }
|
|
|
11345 if (*s == '<')
|
|
|
11346 s++;
|
|
344
|
11347 if (VIM_ISDIGIT(*s))
|
|
323
|
11348 {
|
|
|
11349 p0 = *s - '0';
|
|
|
11350 s++;
|
|
|
11351 }
|
|
|
11352
|
|
|
11353 if (*s == NUL
|
|
|
11354 /* *s == '^' cuts */
|
|
|
11355 || (*s == '$'
|
|
376
|
11356 && !spell_iswordp(word + i + k0,
|
|
|
11357 curbuf)))
|
|
323
|
11358 {
|
|
|
11359 if (k0 == k)
|
|
|
11360 /* this is just a piece of the string */
|
|
|
11361 continue;
|
|
|
11362
|
|
|
11363 if (p0 < pri)
|
|
|
11364 /* priority too low */
|
|
|
11365 continue;
|
|
|
11366 /* rule fits; stop search */
|
|
|
11367 break;
|
|
|
11368 }
|
|
|
11369 }
|
|
|
11370
|
|
344
|
11371 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
11372 continue;
|
|
|
11373 }
|
|
|
11374
|
|
|
11375 /* replace string */
|
|
344
|
11376 s = smp[n].sm_to;
|
|
389
|
11377 if (s == NULL)
|
|
|
11378 s = (char_u *)"";
|
|
344
|
11379 pf = smp[n].sm_rules;
|
|
|
11380 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
11381 if (p0 == 1 && z == 0)
|
|
|
11382 {
|
|
|
11383 /* rule with '<' is used */
|
|
344
|
11384 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
11385 || res[reslen - 1] == *s))
|
|
|
11386 reslen--;
|
|
323
|
11387 z0 = 1;
|
|
|
11388 z = 1;
|
|
|
11389 k0 = 0;
|
|
372
|
11390 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
11391 {
|
|
|
11392 word[i + k0] = *s;
|
|
|
11393 k0++;
|
|
|
11394 s++;
|
|
|
11395 }
|
|
|
11396 if (k > k0)
|
|
|
11397 mch_memmove(word + i + k0, word + i + k,
|
|
|
11398 STRLEN(word + i + k) + 1);
|
|
|
11399
|
|
|
11400 /* new "actual letter" */
|
|
|
11401 c = word[i];
|
|
|
11402 }
|
|
|
11403 else
|
|
|
11404 {
|
|
|
11405 /* no '<' rule used */
|
|
|
11406 i += k - 1;
|
|
|
11407 z = 0;
|
|
344
|
11408 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
11409 {
|
|
344
|
11410 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
11411 res[reslen++] = *s;
|
|
323
|
11412 s++;
|
|
|
11413 }
|
|
|
11414 /* new "actual letter" */
|
|
|
11415 c = *s;
|
|
344
|
11416 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
11417 {
|
|
|
11418 if (c != NUL)
|
|
372
|
11419 res[reslen++] = c;
|
|
323
|
11420 mch_memmove(word, word + i + 1,
|
|
|
11421 STRLEN(word + i + 1) + 1);
|
|
|
11422 i = 0;
|
|
|
11423 z0 = 1;
|
|
|
11424 }
|
|
|
11425 }
|
|
|
11426 break;
|
|
|
11427 }
|
|
|
11428 }
|
|
|
11429 }
|
|
324
|
11430 else if (vim_iswhite(c))
|
|
|
11431 {
|
|
|
11432 c = ' ';
|
|
|
11433 k = 1;
|
|
|
11434 }
|
|
323
|
11435
|
|
|
11436 if (z0 == 0)
|
|
|
11437 {
|
|
344
|
11438 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
11439 && (!slang->sl_collapse || reslen == 0
|
|
|
11440 || res[reslen - 1] != c))
|
|
323
|
11441 /* condense only double letters */
|
|
372
|
11442 res[reslen++] = c;
|
|
323
|
11443
|
|
|
11444 i++;
|
|
|
11445 z = 0;
|
|
|
11446 k = 0;
|
|
|
11447 }
|
|
|
11448 }
|
|
|
11449
|
|
344
|
11450 res[reslen] = NUL;
|
|
323
|
11451 }
|
|
|
11452
|
|
372
|
11453 #ifdef FEAT_MBYTE
|
|
|
11454 /*
|
|
|
11455 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
11456 * Multi-byte version of spell_soundfold().
|
|
|
11457 */
|
|
|
11458 static void
|
|
375
|
11459 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
11460 slang_T *slang;
|
|
|
11461 char_u *inword;
|
|
|
11462 char_u *res;
|
|
|
11463 {
|
|
375
|
11464 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
11465 int word[MAXWLEN];
|
|
|
11466 int wres[MAXWLEN];
|
|
|
11467 int l;
|
|
|
11468 char_u *s;
|
|
|
11469 int *ws;
|
|
|
11470 char_u *t;
|
|
|
11471 int *pf;
|
|
|
11472 int i, j, z;
|
|
|
11473 int reslen;
|
|
|
11474 int n, k = 0;
|
|
|
11475 int z0;
|
|
|
11476 int k0;
|
|
|
11477 int n0;
|
|
|
11478 int c;
|
|
|
11479 int pri;
|
|
|
11480 int p0 = -333;
|
|
|
11481 int c0;
|
|
|
11482 int did_white = FALSE;
|
|
|
11483
|
|
|
11484 /*
|
|
|
11485 * Convert the multi-byte string to a wide-character string.
|
|
|
11486 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
11487 * But keep white space.
|
|
|
11488 */
|
|
|
11489 n = 0;
|
|
|
11490 for (s = inword; *s != NUL; )
|
|
|
11491 {
|
|
|
11492 t = s;
|
|
474
|
11493 c = mb_cptr2char_adv(&s);
|
|
372
|
11494 if (slang->sl_rem_accents)
|
|
|
11495 {
|
|
|
11496 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
11497 {
|
|
|
11498 if (did_white)
|
|
|
11499 continue;
|
|
|
11500 c = ' ';
|
|
|
11501 did_white = TRUE;
|
|
|
11502 }
|
|
|
11503 else
|
|
|
11504 {
|
|
|
11505 did_white = FALSE;
|
|
376
|
11506 if (!spell_iswordp_nmw(t))
|
|
372
|
11507 continue;
|
|
|
11508 }
|
|
|
11509 }
|
|
|
11510 word[n++] = c;
|
|
|
11511 }
|
|
|
11512 word[n] = NUL;
|
|
|
11513
|
|
|
11514 /*
|
|
|
11515 * This comes from Aspell phonet.cpp.
|
|
|
11516 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
11517 * Changed to keep spaces.
|
|
|
11518 */
|
|
|
11519 i = reslen = z = 0;
|
|
|
11520 while ((c = word[i]) != NUL)
|
|
|
11521 {
|
|
|
11522 /* Start with the first rule that has the character in the word. */
|
|
|
11523 n = slang->sl_sal_first[c & 0xff];
|
|
|
11524 z0 = 0;
|
|
|
11525
|
|
|
11526 if (n >= 0)
|
|
|
11527 {
|
|
375
|
11528 /* check all rules for the same index byte */
|
|
372
|
11529 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
11530 {
|
|
|
11531 /* Quickly skip entries that don't match the word. Most
|
|
|
11532 * entries are less then three chars, optimize for that. */
|
|
375
|
11533 if (c != ws[0])
|
|
|
11534 continue;
|
|
372
|
11535 k = smp[n].sm_leadlen;
|
|
|
11536 if (k > 1)
|
|
|
11537 {
|
|
|
11538 if (word[i + 1] != ws[1])
|
|
|
11539 continue;
|
|
|
11540 if (k > 2)
|
|
|
11541 {
|
|
|
11542 for (j = 2; j < k; ++j)
|
|
|
11543 if (word[i + j] != ws[j])
|
|
|
11544 break;
|
|
|
11545 if (j < k)
|
|
|
11546 continue;
|
|
|
11547 }
|
|
|
11548 }
|
|
|
11549
|
|
375
|
11550 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
11551 {
|
|
375
|
11552 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
11553 while (*pf != NUL && *pf != word[i + k])
|
|
|
11554 ++pf;
|
|
|
11555 if (*pf == NUL)
|
|
|
11556 continue;
|
|
|
11557 ++k;
|
|
|
11558 }
|
|
|
11559 s = smp[n].sm_rules;
|
|
|
11560 pri = 5; /* default priority */
|
|
|
11561
|
|
|
11562 p0 = *s;
|
|
|
11563 k0 = k;
|
|
|
11564 while (*s == '-' && k > 1)
|
|
|
11565 {
|
|
|
11566 k--;
|
|
|
11567 s++;
|
|
|
11568 }
|
|
|
11569 if (*s == '<')
|
|
|
11570 s++;
|
|
|
11571 if (VIM_ISDIGIT(*s))
|
|
|
11572 {
|
|
|
11573 /* determine priority */
|
|
|
11574 pri = *s - '0';
|
|
|
11575 s++;
|
|
|
11576 }
|
|
|
11577 if (*s == '^' && *(s + 1) == '^')
|
|
|
11578 s++;
|
|
|
11579
|
|
|
11580 if (*s == NUL
|
|
|
11581 || (*s == '^'
|
|
|
11582 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
11583 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
11584 && (*(s + 1) != '$'
|
|
376
|
11585 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
11586 || (*s == '$' && i > 0
|
|
376
|
11587 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
11588 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
11589 {
|
|
|
11590 /* search for followup rules, if: */
|
|
|
11591 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
11592 c0 = word[i + k - 1];
|
|
|
11593 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
11594
|
|
|
11595 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
11596 && p0 != '-' && word[i + k] != NUL)
|
|
|
11597 {
|
|
375
|
11598 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
11599 * all entries with the same index byte. */
|
|
372
|
11600 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
11601 == (c0 & 0xff); ++n0)
|
|
|
11602 {
|
|
|
11603 /* Quickly skip entries that don't match the word.
|
|
375
|
11604 */
|
|
|
11605 if (c0 != ws[0])
|
|
|
11606 continue;
|
|
372
|
11607 k0 = smp[n0].sm_leadlen;
|
|
|
11608 if (k0 > 1)
|
|
|
11609 {
|
|
|
11610 if (word[i + k] != ws[1])
|
|
|
11611 continue;
|
|
|
11612 if (k0 > 2)
|
|
|
11613 {
|
|
|
11614 pf = word + i + k + 1;
|
|
|
11615 for (j = 2; j < k0; ++j)
|
|
|
11616 if (*pf++ != ws[j])
|
|
|
11617 break;
|
|
|
11618 if (j < k0)
|
|
|
11619 continue;
|
|
|
11620 }
|
|
|
11621 }
|
|
|
11622 k0 += k - 1;
|
|
|
11623
|
|
375
|
11624 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
11625 {
|
|
|
11626 /* Check for match with one of the chars in
|
|
375
|
11627 * "sm_oneof". */
|
|
372
|
11628 while (*pf != NUL && *pf != word[i + k0])
|
|
|
11629 ++pf;
|
|
|
11630 if (*pf == NUL)
|
|
|
11631 continue;
|
|
|
11632 ++k0;
|
|
|
11633 }
|
|
|
11634
|
|
|
11635 p0 = 5;
|
|
|
11636 s = smp[n0].sm_rules;
|
|
|
11637 while (*s == '-')
|
|
|
11638 {
|
|
|
11639 /* "k0" gets NOT reduced because
|
|
|
11640 * "if (k0 == k)" */
|
|
|
11641 s++;
|
|
|
11642 }
|
|
|
11643 if (*s == '<')
|
|
|
11644 s++;
|
|
|
11645 if (VIM_ISDIGIT(*s))
|
|
|
11646 {
|
|
|
11647 p0 = *s - '0';
|
|
|
11648 s++;
|
|
|
11649 }
|
|
|
11650
|
|
|
11651 if (*s == NUL
|
|
|
11652 /* *s == '^' cuts */
|
|
|
11653 || (*s == '$'
|
|
376
|
11654 && !spell_iswordp_w(word + i + k0,
|
|
|
11655 curbuf)))
|
|
372
|
11656 {
|
|
|
11657 if (k0 == k)
|
|
|
11658 /* this is just a piece of the string */
|
|
|
11659 continue;
|
|
|
11660
|
|
|
11661 if (p0 < pri)
|
|
|
11662 /* priority too low */
|
|
|
11663 continue;
|
|
|
11664 /* rule fits; stop search */
|
|
|
11665 break;
|
|
|
11666 }
|
|
|
11667 }
|
|
|
11668
|
|
|
11669 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
11670 == (c0 & 0xff))
|
|
|
11671 continue;
|
|
|
11672 }
|
|
|
11673
|
|
|
11674 /* replace string */
|
|
|
11675 ws = smp[n].sm_to_w;
|
|
|
11676 s = smp[n].sm_rules;
|
|
|
11677 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
11678 if (p0 == 1 && z == 0)
|
|
|
11679 {
|
|
|
11680 /* rule with '<' is used */
|
|
389
|
11681 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
11682 && (wres[reslen - 1] == c
|
|
372
|
11683 || wres[reslen - 1] == *ws))
|
|
|
11684 reslen--;
|
|
|
11685 z0 = 1;
|
|
|
11686 z = 1;
|
|
|
11687 k0 = 0;
|
|
389
|
11688 if (ws != NULL)
|
|
|
11689 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
11690 {
|
|
|
11691 word[i + k0] = *ws;
|
|
|
11692 k0++;
|
|
|
11693 ws++;
|
|
|
11694 }
|
|
372
|
11695 if (k > k0)
|
|
|
11696 mch_memmove(word + i + k0, word + i + k,
|
|
|
11697 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
11698
|
|
|
11699 /* new "actual letter" */
|
|
|
11700 c = word[i];
|
|
|
11701 }
|
|
|
11702 else
|
|
|
11703 {
|
|
|
11704 /* no '<' rule used */
|
|
|
11705 i += k - 1;
|
|
|
11706 z = 0;
|
|
389
|
11707 if (ws != NULL)
|
|
|
11708 while (*ws != NUL && ws[1] != NUL
|
|
|
11709 && reslen < MAXWLEN)
|
|
|
11710 {
|
|
|
11711 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
11712 wres[reslen++] = *ws;
|
|
|
11713 ws++;
|
|
|
11714 }
|
|
372
|
11715 /* new "actual letter" */
|
|
389
|
11716 if (ws == NULL)
|
|
|
11717 c = NUL;
|
|
|
11718 else
|
|
|
11719 c = *ws;
|
|
372
|
11720 if (strstr((char *)s, "^^") != NULL)
|
|
|
11721 {
|
|
|
11722 if (c != NUL)
|
|
|
11723 wres[reslen++] = c;
|
|
|
11724 mch_memmove(word, word + i + 1,
|
|
|
11725 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
11726 i = 0;
|
|
|
11727 z0 = 1;
|
|
|
11728 }
|
|
|
11729 }
|
|
|
11730 break;
|
|
|
11731 }
|
|
|
11732 }
|
|
|
11733 }
|
|
|
11734 else if (vim_iswhite(c))
|
|
|
11735 {
|
|
|
11736 c = ' ';
|
|
|
11737 k = 1;
|
|
|
11738 }
|
|
|
11739
|
|
|
11740 if (z0 == 0)
|
|
|
11741 {
|
|
|
11742 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
11743 && (!slang->sl_collapse || reslen == 0
|
|
|
11744 || wres[reslen - 1] != c))
|
|
|
11745 /* condense only double letters */
|
|
|
11746 wres[reslen++] = c;
|
|
|
11747
|
|
|
11748 i++;
|
|
|
11749 z = 0;
|
|
|
11750 k = 0;
|
|
|
11751 }
|
|
|
11752 }
|
|
|
11753
|
|
|
11754 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
11755 l = 0;
|
|
|
11756 for (n = 0; n < reslen; ++n)
|
|
|
11757 {
|
|
|
11758 l += mb_char2bytes(wres[n], res + l);
|
|
|
11759 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
11760 break;
|
|
|
11761 }
|
|
|
11762 res[l] = NUL;
|
|
|
11763 }
|
|
|
11764 #endif
|
|
|
11765
|
|
324
|
11766 /*
|
|
344
|
11767 * Compute a score for two sound-a-like words.
|
|
|
11768 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
11769 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
11770 * avoiding checks that will not be possible.
|
|
|
11771 */
|
|
|
11772 static int
|
|
351
|
11773 soundalike_score(goodstart, badstart)
|
|
|
11774 char_u *goodstart; /* sound-folded good word */
|
|
|
11775 char_u *badstart; /* sound-folded bad word */
|
|
344
|
11776 {
|
|
351
|
11777 char_u *goodsound = goodstart;
|
|
|
11778 char_u *badsound = badstart;
|
|
|
11779 int goodlen;
|
|
|
11780 int badlen;
|
|
344
|
11781 int n;
|
|
|
11782 char_u *pl, *ps;
|
|
|
11783 char_u *pl2, *ps2;
|
|
351
|
11784 int score = 0;
|
|
|
11785
|
|
|
11786 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
11787 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
11788 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
11789 {
|
|
|
11790 score = SCORE_DEL / 2;
|
|
|
11791 if (*badsound == '*')
|
|
|
11792 ++badsound;
|
|
|
11793 else
|
|
|
11794 ++goodsound;
|
|
|
11795 }
|
|
|
11796
|
|
|
11797 goodlen = STRLEN(goodsound);
|
|
|
11798 badlen = STRLEN(badsound);
|
|
344
|
11799
|
|
|
11800 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
11801 * changes. */
|
|
|
11802 n = goodlen - badlen;
|
|
|
11803 if (n < -2 || n > 2)
|
|
|
11804 return SCORE_MAXMAX;
|
|
|
11805
|
|
|
11806 if (n > 0)
|
|
|
11807 {
|
|
351
|
11808 pl = goodsound; /* goodsound is longest */
|
|
344
|
11809 ps = badsound;
|
|
|
11810 }
|
|
|
11811 else
|
|
|
11812 {
|
|
351
|
11813 pl = badsound; /* badsound is longest */
|
|
344
|
11814 ps = goodsound;
|
|
|
11815 }
|
|
|
11816
|
|
|
11817 /* Skip over the identical part. */
|
|
|
11818 while (*pl == *ps && *pl != NUL)
|
|
|
11819 {
|
|
|
11820 ++pl;
|
|
|
11821 ++ps;
|
|
|
11822 }
|
|
|
11823
|
|
|
11824 switch (n)
|
|
|
11825 {
|
|
|
11826 case -2:
|
|
|
11827 case 2:
|
|
|
11828 /*
|
|
|
11829 * Must delete two characters from "pl".
|
|
|
11830 */
|
|
|
11831 ++pl; /* first delete */
|
|
|
11832 while (*pl == *ps)
|
|
|
11833 {
|
|
|
11834 ++pl;
|
|
|
11835 ++ps;
|
|
|
11836 }
|
|
|
11837 /* strings must be equal after second delete */
|
|
|
11838 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
11839 return score + SCORE_DEL * 2;
|
|
344
|
11840
|
|
|
11841 /* Failed to compare. */
|
|
|
11842 break;
|
|
|
11843
|
|
|
11844 case -1:
|
|
|
11845 case 1:
|
|
|
11846 /*
|
|
|
11847 * Minimal one delete from "pl" required.
|
|
|
11848 */
|
|
|
11849
|
|
|
11850 /* 1: delete */
|
|
|
11851 pl2 = pl + 1;
|
|
|
11852 ps2 = ps;
|
|
|
11853 while (*pl2 == *ps2)
|
|
|
11854 {
|
|
|
11855 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11856 return score + SCORE_DEL;
|
|
344
|
11857 ++pl2;
|
|
|
11858 ++ps2;
|
|
|
11859 }
|
|
|
11860
|
|
|
11861 /* 2: delete then swap, then rest must be equal */
|
|
|
11862 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11863 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11864 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
11865
|
|
|
11866 /* 3: delete then substitute, then the rest must be equal */
|
|
|
11867 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11868 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
11869
|
|
|
11870 /* 4: first swap then delete */
|
|
|
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 ++pl2;
|
|
|
11878 ++ps2;
|
|
|
11879 }
|
|
|
11880 /* delete a char and then strings must be equal */
|
|
|
11881 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11882 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
11883 }
|
|
|
11884
|
|
|
11885 /* 5: first substitute then delete */
|
|
|
11886 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
11887 ps2 = ps + 1;
|
|
|
11888 while (*pl2 == *ps2)
|
|
|
11889 {
|
|
|
11890 ++pl2;
|
|
|
11891 ++ps2;
|
|
|
11892 }
|
|
|
11893 /* delete a char and then strings must be equal */
|
|
|
11894 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11895 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
11896
|
|
|
11897 /* Failed to compare. */
|
|
|
11898 break;
|
|
|
11899
|
|
|
11900 case 0:
|
|
|
11901 /*
|
|
|
11902 * Lenghts are equal, thus changes must result in same length: An
|
|
|
11903 * insert is only possible in combination with a delete.
|
|
|
11904 * 1: check if for identical strings
|
|
|
11905 */
|
|
|
11906 if (*pl == NUL)
|
|
351
|
11907 return score;
|
|
344
|
11908
|
|
|
11909 /* 2: swap */
|
|
|
11910 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
11911 {
|
|
|
11912 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
11913 ps2 = ps + 2;
|
|
|
11914 while (*pl2 == *ps2)
|
|
|
11915 {
|
|
|
11916 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11917 return score + SCORE_SWAP;
|
|
344
|
11918 ++pl2;
|
|
|
11919 ++ps2;
|
|
|
11920 }
|
|
|
11921 /* 3: swap and swap again */
|
|
|
11922 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11923 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11924 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
11925
|
|
|
11926 /* 4: swap and substitute */
|
|
|
11927 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11928 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
11929 }
|
|
|
11930
|
|
|
11931 /* 5: substitute */
|
|
|
11932 pl2 = pl + 1;
|
|
|
11933 ps2 = ps + 1;
|
|
|
11934 while (*pl2 == *ps2)
|
|
|
11935 {
|
|
|
11936 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11937 return score + SCORE_SUBST;
|
|
344
|
11938 ++pl2;
|
|
|
11939 ++ps2;
|
|
|
11940 }
|
|
|
11941
|
|
|
11942 /* 6: substitute and swap */
|
|
|
11943 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11944 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11945 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
11946
|
|
|
11947 /* 7: substitute and substitute */
|
|
|
11948 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11949 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
11950
|
|
|
11951 /* 8: insert then delete */
|
|
|
11952 pl2 = pl;
|
|
|
11953 ps2 = ps + 1;
|
|
|
11954 while (*pl2 == *ps2)
|
|
|
11955 {
|
|
|
11956 ++pl2;
|
|
|
11957 ++ps2;
|
|
|
11958 }
|
|
|
11959 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11960 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
11961
|
|
|
11962 /* 9: delete then insert */
|
|
|
11963 pl2 = pl + 1;
|
|
|
11964 ps2 = ps;
|
|
|
11965 while (*pl2 == *ps2)
|
|
|
11966 {
|
|
|
11967 ++pl2;
|
|
|
11968 ++ps2;
|
|
|
11969 }
|
|
|
11970 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
11971 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
11972
|
|
|
11973 /* Failed to compare. */
|
|
|
11974 break;
|
|
|
11975 }
|
|
|
11976
|
|
|
11977 return SCORE_MAXMAX;
|
|
|
11978 }
|
|
324
|
11979
|
|
323
|
11980 /*
|
|
|
11981 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
11982 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
11983 *
|
|
485
|
11984 * The algorithm is described by Du and Chang, 1992.
|
|
|
11985 * The implementation of the algorithm comes from Aspell editdist.cpp,
|
|
|
11986 * edit_distance(). It has been converted from C++ to C and modified to
|
|
|
11987 * support multi-byte characters.
|
|
323
|
11988 */
|
|
|
11989 static int
|
|
|
11990 spell_edit_score(badword, goodword)
|
|
|
11991 char_u *badword;
|
|
|
11992 char_u *goodword;
|
|
|
11993 {
|
|
|
11994 int *cnt;
|
|
372
|
11995 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
11996 int j, i;
|
|
|
11997 int t;
|
|
|
11998 int bc, gc;
|
|
324
|
11999 int pbc, pgc;
|
|
|
12000 #ifdef FEAT_MBYTE
|
|
|
12001 char_u *p;
|
|
|
12002 int wbadword[MAXWLEN];
|
|
|
12003 int wgoodword[MAXWLEN];
|
|
|
12004
|
|
|
12005 if (has_mbyte)
|
|
|
12006 {
|
|
|
12007 /* Get the characters from the multi-byte strings and put them in an
|
|
|
12008 * int array for easy access. */
|
|
|
12009 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
12010 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
12011 wbadword[badlen++] = 0;
|
|
324
|
12012 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
12013 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
12014 wgoodword[goodlen++] = 0;
|
|
324
|
12015 }
|
|
|
12016 else
|
|
|
12017 #endif
|
|
|
12018 {
|
|
|
12019 badlen = STRLEN(badword) + 1;
|
|
|
12020 goodlen = STRLEN(goodword) + 1;
|
|
|
12021 }
|
|
323
|
12022
|
|
|
12023 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
12024 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
12025 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
12026 TRUE);
|
|
324
|
12027 if (cnt == NULL)
|
|
|
12028 return 0; /* out of memory */
|
|
323
|
12029
|
|
|
12030 CNT(0, 0) = 0;
|
|
|
12031 for (j = 1; j <= goodlen; ++j)
|
|
|
12032 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
12033
|
|
|
12034 for (i = 1; i <= badlen; ++i)
|
|
|
12035 {
|
|
|
12036 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
12037 for (j = 1; j <= goodlen; ++j)
|
|
|
12038 {
|
|
324
|
12039 #ifdef FEAT_MBYTE
|
|
|
12040 if (has_mbyte)
|
|
|
12041 {
|
|
|
12042 bc = wbadword[i - 1];
|
|
|
12043 gc = wgoodword[j - 1];
|
|
|
12044 }
|
|
|
12045 else
|
|
|
12046 #endif
|
|
|
12047 {
|
|
|
12048 bc = badword[i - 1];
|
|
|
12049 gc = goodword[j - 1];
|
|
|
12050 }
|
|
323
|
12051 if (bc == gc)
|
|
|
12052 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
12053 else
|
|
|
12054 {
|
|
|
12055 /* Use a better score when there is only a case difference. */
|
|
324
|
12056 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
12057 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
12058 else
|
|
|
12059 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
12060
|
|
324
|
12061 if (i > 1 && j > 1)
|
|
323
|
12062 {
|
|
324
|
12063 #ifdef FEAT_MBYTE
|
|
|
12064 if (has_mbyte)
|
|
|
12065 {
|
|
|
12066 pbc = wbadword[i - 2];
|
|
|
12067 pgc = wgoodword[j - 2];
|
|
|
12068 }
|
|
|
12069 else
|
|
|
12070 #endif
|
|
|
12071 {
|
|
|
12072 pbc = badword[i - 2];
|
|
|
12073 pgc = goodword[j - 2];
|
|
|
12074 }
|
|
|
12075 if (bc == pgc && pbc == gc)
|
|
|
12076 {
|
|
|
12077 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
12078 if (t < CNT(i, j))
|
|
|
12079 CNT(i, j) = t;
|
|
|
12080 }
|
|
323
|
12081 }
|
|
|
12082 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
12083 if (t < CNT(i, j))
|
|
|
12084 CNT(i, j) = t;
|
|
|
12085 t = SCORE_INS + CNT(i, j - 1);
|
|
|
12086 if (t < CNT(i, j))
|
|
|
12087 CNT(i, j) = t;
|
|
|
12088 }
|
|
|
12089 }
|
|
|
12090 }
|
|
344
|
12091
|
|
|
12092 i = CNT(badlen - 1, goodlen - 1);
|
|
|
12093 vim_free(cnt);
|
|
|
12094 return i;
|
|
323
|
12095 }
|
|
307
|
12096
|
|
351
|
12097 /*
|
|
|
12098 * ":spelldump"
|
|
|
12099 */
|
|
|
12100 /*ARGSUSED*/
|
|
|
12101 void
|
|
|
12102 ex_spelldump(eap)
|
|
|
12103 exarg_T *eap;
|
|
|
12104 {
|
|
|
12105 buf_T *buf = curbuf;
|
|
|
12106 langp_T *lp;
|
|
|
12107 slang_T *slang;
|
|
|
12108 idx_T arridx[MAXWLEN];
|
|
|
12109 int curi[MAXWLEN];
|
|
|
12110 char_u word[MAXWLEN];
|
|
|
12111 int c;
|
|
|
12112 char_u *byts;
|
|
|
12113 idx_T *idxs;
|
|
|
12114 linenr_T lnum = 0;
|
|
|
12115 int round;
|
|
|
12116 int depth;
|
|
|
12117 int n;
|
|
|
12118 int flags;
|
|
381
|
12119 char_u *region_names = NULL; /* region names being used */
|
|
|
12120 int do_region = TRUE; /* dump region names and numbers */
|
|
|
12121 char_u *p;
|
|
351
|
12122
|
|
498
|
12123 if (no_spell_checking(curwin))
|
|
351
|
12124 return;
|
|
|
12125
|
|
|
12126 /* Create a new empty buffer by splitting the window. */
|
|
|
12127 do_cmdline_cmd((char_u *)"new");
|
|
|
12128 if (!bufempty() || !buf_valid(buf))
|
|
|
12129 return;
|
|
|
12130
|
|
381
|
12131 /* Find out if we can support regions: All languages must support the same
|
|
|
12132 * regions or none at all. */
|
|
|
12133 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
12134 {
|
|
|
12135 p = lp->lp_slang->sl_regions;
|
|
|
12136 if (p[0] != 0)
|
|
|
12137 {
|
|
|
12138 if (region_names == NULL) /* first language with regions */
|
|
|
12139 region_names = p;
|
|
|
12140 else if (STRCMP(region_names, p) != 0)
|
|
|
12141 {
|
|
|
12142 do_region = FALSE; /* region names are different */
|
|
|
12143 break;
|
|
|
12144 }
|
|
|
12145 }
|
|
|
12146 }
|
|
|
12147
|
|
|
12148 if (do_region && region_names != NULL)
|
|
|
12149 {
|
|
|
12150 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
12151 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
12152 }
|
|
|
12153 else
|
|
|
12154 do_region = FALSE;
|
|
|
12155
|
|
|
12156 /*
|
|
|
12157 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
12158 */
|
|
351
|
12159 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
12160 {
|
|
|
12161 slang = lp->lp_slang;
|
|
|
12162
|
|
|
12163 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
12164 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
12165
|
|
|
12166 /* round 1: case-folded tree
|
|
|
12167 * round 2: keep-case tree */
|
|
|
12168 for (round = 1; round <= 2; ++round)
|
|
|
12169 {
|
|
|
12170 if (round == 1)
|
|
|
12171 {
|
|
|
12172 byts = slang->sl_fbyts;
|
|
|
12173 idxs = slang->sl_fidxs;
|
|
|
12174 }
|
|
|
12175 else
|
|
|
12176 {
|
|
|
12177 byts = slang->sl_kbyts;
|
|
|
12178 idxs = slang->sl_kidxs;
|
|
|
12179 }
|
|
|
12180 if (byts == NULL)
|
|
|
12181 continue; /* array is empty */
|
|
|
12182
|
|
|
12183 depth = 0;
|
|
|
12184 arridx[0] = 0;
|
|
|
12185 curi[0] = 1;
|
|
|
12186 while (depth >= 0 && !got_int)
|
|
|
12187 {
|
|
|
12188 if (curi[depth] > byts[arridx[depth]])
|
|
|
12189 {
|
|
|
12190 /* Done all bytes at this node, go up one level. */
|
|
|
12191 --depth;
|
|
|
12192 line_breakcheck();
|
|
|
12193 }
|
|
|
12194 else
|
|
|
12195 {
|
|
|
12196 /* Do one more byte at this node. */
|
|
|
12197 n = arridx[depth] + curi[depth];
|
|
|
12198 ++curi[depth];
|
|
|
12199 c = byts[n];
|
|
|
12200 if (c == 0)
|
|
|
12201 {
|
|
|
12202 /* End of word, deal with the word.
|
|
|
12203 * Don't use keep-case words in the fold-case tree,
|
|
|
12204 * they will appear in the keep-case tree.
|
|
|
12205 * Only use the word when the region matches. */
|
|
|
12206 flags = (int)idxs[n];
|
|
|
12207 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
12208 && (do_region
|
|
|
12209 || (flags & WF_REGION) == 0
|
|
390
|
12210 || (((unsigned)flags >> 16)
|
|
351
|
12211 & lp->lp_region) != 0))
|
|
|
12212 {
|
|
|
12213 word[depth] = NUL;
|
|
381
|
12214 if (!do_region)
|
|
|
12215 flags &= ~WF_REGION;
|
|
355
|
12216
|
|
|
12217 /* Dump the basic word if there is no prefix or
|
|
|
12218 * when it's the first one. */
|
|
390
|
12219 c = (unsigned)flags >> 24;
|
|
355
|
12220 if (c == 0 || curi[depth] == 2)
|
|
|
12221 dump_word(word, round, flags, lnum++);
|
|
351
|
12222
|
|
|
12223 /* Apply the prefix, if there is one. */
|
|
355
|
12224 if (c != 0)
|
|
351
|
12225 lnum = apply_prefixes(slang, word, round,
|
|
|
12226 flags, lnum);
|
|
|
12227 }
|
|
|
12228 }
|
|
|
12229 else
|
|
|
12230 {
|
|
|
12231 /* Normal char, go one level deeper. */
|
|
|
12232 word[depth++] = c;
|
|
|
12233 arridx[depth] = idxs[n];
|
|
|
12234 curi[depth] = 1;
|
|
|
12235 }
|
|
|
12236 }
|
|
|
12237 }
|
|
|
12238 }
|
|
|
12239 }
|
|
|
12240
|
|
|
12241 /* Delete the empty line that we started with. */
|
|
|
12242 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
12243 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
12244
|
|
|
12245 redraw_later(NOT_VALID);
|
|
|
12246 }
|
|
|
12247
|
|
|
12248 /*
|
|
|
12249 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
12250 */
|
|
|
12251 static void
|
|
|
12252 dump_word(word, round, flags, lnum)
|
|
|
12253 char_u *word;
|
|
|
12254 int round;
|
|
|
12255 int flags;
|
|
|
12256 linenr_T lnum;
|
|
|
12257 {
|
|
|
12258 int keepcap = FALSE;
|
|
|
12259 char_u *p;
|
|
|
12260 char_u cword[MAXWLEN];
|
|
381
|
12261 char_u badword[MAXWLEN + 10];
|
|
|
12262 int i;
|
|
351
|
12263
|
|
|
12264 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
12265 {
|
|
|
12266 /* Need to fix case according to "flags". */
|
|
|
12267 make_case_word(word, cword, flags);
|
|
|
12268 p = cword;
|
|
|
12269 }
|
|
|
12270 else
|
|
|
12271 {
|
|
|
12272 p = word;
|
|
389
|
12273 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
12274 || (flags & WF_FIXCAP) != 0))
|
|
351
|
12275 keepcap = TRUE;
|
|
|
12276 }
|
|
|
12277
|
|
381
|
12278 /* Add flags and regions after a slash. */
|
|
|
12279 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
12280 {
|
|
|
12281 STRCPY(badword, p);
|
|
|
12282 STRCAT(badword, "/");
|
|
351
|
12283 if (keepcap)
|
|
|
12284 STRCAT(badword, "=");
|
|
|
12285 if (flags & WF_BANNED)
|
|
|
12286 STRCAT(badword, "!");
|
|
|
12287 else if (flags & WF_RARE)
|
|
|
12288 STRCAT(badword, "?");
|
|
381
|
12289 if (flags & WF_REGION)
|
|
|
12290 for (i = 0; i < 7; ++i)
|
|
390
|
12291 if (flags & (0x10000 << i))
|
|
381
|
12292 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
12293 p = badword;
|
|
|
12294 }
|
|
|
12295
|
|
|
12296 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
12297 }
|
|
|
12298
|
|
|
12299 /*
|
|
372
|
12300 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
12301 * "word" and append a line to the buffer.
|
|
351
|
12302 * Return the updated line number.
|
|
|
12303 */
|
|
|
12304 static linenr_T
|
|
|
12305 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
12306 slang_T *slang;
|
|
|
12307 char_u *word; /* case-folded word */
|
|
|
12308 int round;
|
|
|
12309 int flags; /* flags with prefix ID */
|
|
|
12310 linenr_T startlnum;
|
|
|
12311 {
|
|
|
12312 idx_T arridx[MAXWLEN];
|
|
|
12313 int curi[MAXWLEN];
|
|
|
12314 char_u prefix[MAXWLEN];
|
|
455
|
12315 char_u word_up[MAXWLEN];
|
|
|
12316 int has_word_up = FALSE;
|
|
351
|
12317 int c;
|
|
|
12318 char_u *byts;
|
|
|
12319 idx_T *idxs;
|
|
|
12320 linenr_T lnum = startlnum;
|
|
|
12321 int depth;
|
|
|
12322 int n;
|
|
|
12323 int len;
|
|
|
12324 int i;
|
|
|
12325
|
|
455
|
12326 /* if the word starts with a lower-case letter make the word with an
|
|
|
12327 * upper-case letter in word_up[]. */
|
|
|
12328 c = PTR2CHAR(word);
|
|
|
12329 if (SPELL_TOUPPER(c) != c)
|
|
|
12330 {
|
|
|
12331 onecap_copy(word, word_up, TRUE);
|
|
|
12332 has_word_up = TRUE;
|
|
|
12333 }
|
|
|
12334
|
|
351
|
12335 byts = slang->sl_pbyts;
|
|
|
12336 idxs = slang->sl_pidxs;
|
|
|
12337 if (byts != NULL) /* array not is empty */
|
|
|
12338 {
|
|
|
12339 /*
|
|
|
12340 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
12341 * When at the end of a prefix check that it supports "flags".
|
|
351
|
12342 */
|
|
|
12343 depth = 0;
|
|
|
12344 arridx[0] = 0;
|
|
|
12345 curi[0] = 1;
|
|
|
12346 while (depth >= 0 && !got_int)
|
|
|
12347 {
|
|
390
|
12348 n = arridx[depth];
|
|
|
12349 len = byts[n];
|
|
|
12350 if (curi[depth] > len)
|
|
351
|
12351 {
|
|
|
12352 /* Done all bytes at this node, go up one level. */
|
|
|
12353 --depth;
|
|
|
12354 line_breakcheck();
|
|
|
12355 }
|
|
|
12356 else
|
|
|
12357 {
|
|
|
12358 /* Do one more byte at this node. */
|
|
390
|
12359 n += curi[depth];
|
|
351
|
12360 ++curi[depth];
|
|
|
12361 c = byts[n];
|
|
|
12362 if (c == 0)
|
|
|
12363 {
|
|
|
12364 /* End of prefix, find out how many IDs there are. */
|
|
|
12365 for (i = 1; i < len; ++i)
|
|
|
12366 if (byts[n + i] != 0)
|
|
|
12367 break;
|
|
|
12368 curi[depth] += i - 1;
|
|
|
12369
|
|
455
|
12370 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
12371 if (c != 0)
|
|
351
|
12372 {
|
|
376
|
12373 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
12374 dump_word(prefix, round,
|
|
455
|
12375 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
12376 : flags, lnum++);
|
|
351
|
12377 }
|
|
455
|
12378
|
|
|
12379 /* Check for prefix that matches the word when the
|
|
|
12380 * first letter is upper-case, but only if the prefix has
|
|
|
12381 * a condition. */
|
|
|
12382 if (has_word_up)
|
|
|
12383 {
|
|
|
12384 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
12385 TRUE);
|
|
|
12386 if (c != 0)
|
|
|
12387 {
|
|
|
12388 vim_strncpy(prefix + depth, word_up,
|
|
|
12389 MAXWLEN - depth - 1);
|
|
|
12390 dump_word(prefix, round,
|
|
|
12391 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
12392 : flags, lnum++);
|
|
|
12393 }
|
|
|
12394 }
|
|
351
|
12395 }
|
|
|
12396 else
|
|
|
12397 {
|
|
|
12398 /* Normal char, go one level deeper. */
|
|
|
12399 prefix[depth++] = c;
|
|
|
12400 arridx[depth] = idxs[n];
|
|
|
12401 curi[depth] = 1;
|
|
|
12402 }
|
|
|
12403 }
|
|
|
12404 }
|
|
|
12405 }
|
|
|
12406
|
|
|
12407 return lnum;
|
|
|
12408 }
|
|
|
12409
|
|
498
|
12410 /*
|
|
|
12411 * Move "p" to end of word.
|
|
|
12412 */
|
|
|
12413 char_u *
|
|
|
12414 spell_to_word_end(start, buf)
|
|
|
12415 char_u *start;
|
|
|
12416 buf_T *buf;
|
|
|
12417 {
|
|
|
12418 char_u *p = start;
|
|
|
12419
|
|
|
12420 while (*p != NUL && spell_iswordp(p, buf))
|
|
|
12421 mb_ptr_adv(p);
|
|
|
12422 return p;
|
|
|
12423 }
|
|
|
12424
|
|
475
|
12425 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
12426 static int spell_expand_need_cap;
|
|
|
12427
|
|
|
12428 /*
|
|
|
12429 * Find start of the word in front of the cursor. We don't check if it is
|
|
|
12430 * badly spelled, with completion we can only change the word in front of the
|
|
|
12431 * cursor.
|
|
|
12432 * Used for Insert mode completion CTRL-X ?.
|
|
|
12433 * Returns the column number of the word.
|
|
|
12434 */
|
|
|
12435 int
|
|
|
12436 spell_word_start(startcol)
|
|
|
12437 int startcol;
|
|
|
12438 {
|
|
|
12439 char_u *line;
|
|
|
12440 char_u *p;
|
|
|
12441 int col = 0;
|
|
|
12442
|
|
498
|
12443 if (no_spell_checking(curwin))
|
|
475
|
12444 return startcol;
|
|
|
12445
|
|
|
12446 /* Find a word character before "startcol". */
|
|
|
12447 line = ml_get_curline();
|
|
|
12448 for (p = line + startcol; p > line; )
|
|
|
12449 {
|
|
|
12450 mb_ptr_back(line, p);
|
|
|
12451 if (spell_iswordp_nmw(p))
|
|
|
12452 break;
|
|
|
12453 }
|
|
|
12454
|
|
|
12455 /* Go back to start of the word. */
|
|
|
12456 while (p > line)
|
|
|
12457 {
|
|
|
12458 col = p - line;
|
|
|
12459 mb_ptr_back(line, p);
|
|
|
12460 if (!spell_iswordp(p, curbuf))
|
|
|
12461 break;
|
|
|
12462 col = 0;
|
|
|
12463 }
|
|
|
12464
|
|
|
12465 /* Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
12466 * expand_spelling() is called. Therefore the ugly global variable. */
|
|
|
12467 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
12468
|
|
|
12469 return col;
|
|
|
12470 }
|
|
|
12471
|
|
|
12472 /*
|
|
|
12473 * Get list of spelling suggestions.
|
|
|
12474 * Used for Insert mode completion CTRL-X ?.
|
|
|
12475 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
12476 * allocated strings.
|
|
|
12477 */
|
|
|
12478 /*ARGSUSED*/
|
|
|
12479 int
|
|
|
12480 expand_spelling(lnum, col, pat, matchp)
|
|
|
12481 linenr_T lnum;
|
|
|
12482 int col;
|
|
|
12483 char_u *pat;
|
|
|
12484 char_u ***matchp;
|
|
|
12485 {
|
|
|
12486 garray_T ga;
|
|
|
12487
|
|
|
12488 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap);
|
|
|
12489 *matchp = ga.ga_data;
|
|
|
12490 return ga.ga_len;
|
|
|
12491 }
|
|
|
12492 #endif
|
|
|
12493
|
|
236
|
12494 #endif /* FEAT_SYN_HL */
|
