|
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 *
|
|
500
|
217 * <flags2> 1 byte Bitmask of:
|
|
390
|
218 * WF_HAS_AFF >> 8 word includes affix
|
|
500
|
219 * WF_NEEDCOMP >> 8 word only valid in compound
|
|
390
|
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 */
|
|
500
|
276 #define WF_NEEDCOMP 0x0200 /* word only valid in compound */
|
|
390
|
277
|
|
389
|
278 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
279
|
|
455
|
280 /* flags for <pflags> */
|
|
|
281 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
282 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
283 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
284
|
|
481
|
285 /* Flags for postponed prefixes. Must be above affixID (one byte)
|
|
390
|
286 * and prefcondnr (two bytes). */
|
|
455
|
287 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
288 * postponed prefix */
|
|
|
289 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
290 * postponed prefix */
|
|
|
291 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
292 * postponed prefix */
|
|
390
|
293
|
|
|
294 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
295 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
296 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
297 * postponed prefix: no <pflags> */
|
|
|
298 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
299 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
300 * postponed prefix: <pflags> follows */
|
|
|
301 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
302 * follow; never used in prefix tree */
|
|
|
303 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
304
|
|
323
|
305 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
306 * and si_sal. Not for sl_sal!
|
|
323
|
307 * One replacement: from "ft_from" to "ft_to". */
|
|
|
308 typedef struct fromto_S
|
|
236
|
309 {
|
|
323
|
310 char_u *ft_from;
|
|
|
311 char_u *ft_to;
|
|
|
312 } fromto_T;
|
|
236
|
313
|
|
344
|
314 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
315 * The info is split for quick processing by spell_soundfold().
|
|
|
316 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
317 typedef struct salitem_S
|
|
|
318 {
|
|
|
319 char_u *sm_lead; /* leading letters */
|
|
|
320 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
321 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
322 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
323 char_u *sm_to; /* replacement. */
|
|
372
|
324 #ifdef FEAT_MBYTE
|
|
|
325 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
326 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
327 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
328 #endif
|
|
344
|
329 } salitem_T;
|
|
|
330
|
|
375
|
331 #ifdef FEAT_MBYTE
|
|
|
332 typedef int salfirst_T;
|
|
|
333 #else
|
|
|
334 typedef short salfirst_T;
|
|
|
335 #endif
|
|
|
336
|
|
483
|
337 /* Values for SP_*ERROR are negative, positive values are used by
|
|
|
338 * read_cnt_string(). */
|
|
|
339 #define SP_TRUNCERROR -1 /* spell file truncated error */
|
|
|
340 #define SP_FORMERROR -2 /* format error in spell file */
|
|
495
|
341 #define SP_OTHERERROR -3 /* other error while reading spell file */
|
|
483
|
342
|
|
236
|
343 /*
|
|
243
|
344 * Structure used to store words and other info for one language, loaded from
|
|
|
345 * a .spl file.
|
|
300
|
346 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
347 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
348 *
|
|
|
349 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
350 * the number of possible bytes, sorted on byte value:
|
|
|
351 * <len> <byte1> <byte2> ...
|
|
|
352 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
353 * byte in "byts".
|
|
|
354 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
481
|
355 * the flags, region mask and affixID for the word. There may be several
|
|
|
356 * zeros in sequence for alternative flag/region/affixID combinations.
|
|
236
|
357 */
|
|
|
358 typedef struct slang_S slang_T;
|
|
|
359 struct slang_S
|
|
|
360 {
|
|
|
361 slang_T *sl_next; /* next language */
|
|
|
362 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
363 char_u *sl_fname; /* name of .spl file */
|
|
323
|
364 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
365
|
|
300
|
366 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
367 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
368 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
369 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
370 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
371 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
372
|
|
236
|
373 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
374
|
|
376
|
375 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
376
|
|
483
|
377 int sl_compmax; /* COMPOUNDMAX (default: MAXWLEN) */
|
|
501
|
378 int sl_compminlen; /* COMPOUNDMIN (default: 0) */
|
|
483
|
379 int sl_compsylmax; /* COMPOUNDSYLMAX (default: MAXWLEN) */
|
|
|
380 regprog_T *sl_compprog; /* COMPOUNDFLAGS turned into a regexp progrm
|
|
|
381 * (NULL when no compounding) */
|
|
|
382 char_u *sl_compstartflags; /* flags for first compound word */
|
|
485
|
383 char_u *sl_compallflags; /* all flags for compound words */
|
|
492
|
384 char_u sl_nobreak; /* When TRUE: no spaces between words */
|
|
483
|
385 char_u *sl_syllable; /* SYLLABLE repeatable chars or NULL */
|
|
|
386 garray_T sl_syl_items; /* syllable items */
|
|
481
|
387
|
|
339
|
388 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
389 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
390
|
|
323
|
391 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
392 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
393 there is none */
|
|
344
|
394 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
395 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
396 there is none */
|
|
375
|
397 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
398 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
399 * "sl_sal" is a list of wide char lists. */
|
|
323
|
400 int sl_followup; /* SAL followup */
|
|
|
401 int sl_collapse; /* SAL collapse_result */
|
|
|
402 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
403 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
404 #ifdef FEAT_MBYTE
|
|
|
405 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
406 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
407 #else
|
|
|
408 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
409 #endif
|
|
236
|
410 };
|
|
|
411
|
|
243
|
412 /* First language that is loaded, start of the linked list of loaded
|
|
|
413 * languages. */
|
|
236
|
414 static slang_T *first_lang = NULL;
|
|
|
415
|
|
323
|
416 /* Flags used in .spl file for soundsalike flags. */
|
|
|
417 #define SAL_F0LLOWUP 1
|
|
|
418 #define SAL_COLLAPSE 2
|
|
|
419 #define SAL_REM_ACCENTS 4
|
|
|
420
|
|
236
|
421 /*
|
|
|
422 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
423 */
|
|
|
424 typedef struct langp_S
|
|
|
425 {
|
|
503
|
426 slang_T *lp_slang; /* info for this language */
|
|
|
427 slang_T *lp_sallang; /* language used for sound folding or NULL */
|
|
|
428 slang_T *lp_replang; /* language used for REP items or NULL */
|
|
236
|
429 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
430 } langp_T;
|
|
|
431
|
|
|
432 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
433
|
|
307
|
434 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
435
|
|
483
|
436 #define VIMSPELLMAGIC "VIMspell" /* string at start of Vim spell file */
|
|
|
437 #define VIMSPELLMAGICL 8
|
|
|
438 #define VIMSPELLVERSION 50
|
|
|
439
|
|
|
440 /* Section IDs. Only renumber them when VIMSPELLVERSION changes! */
|
|
|
441 #define SN_REGION 0 /* <regionname> section */
|
|
|
442 #define SN_CHARFLAGS 1 /* charflags section */
|
|
|
443 #define SN_MIDWORD 2 /* <midword> section */
|
|
|
444 #define SN_PREFCOND 3 /* <prefcond> section */
|
|
|
445 #define SN_REP 4 /* REP items section */
|
|
|
446 #define SN_SAL 5 /* SAL items section */
|
|
|
447 #define SN_SOFO 6 /* soundfolding section */
|
|
|
448 #define SN_MAP 7 /* MAP items section */
|
|
|
449 #define SN_COMPOUND 8 /* compound words section */
|
|
|
450 #define SN_SYLLABLE 9 /* syllable section */
|
|
492
|
451 #define SN_NOBREAK 10 /* NOBREAK section */
|
|
483
|
452 #define SN_END 255 /* end of sections */
|
|
|
453
|
|
|
454 #define SNF_REQUIRED 1 /* <sectionflags>: required section */
|
|
|
455
|
|
307
|
456 /* Result values. Lower number is accepted over higher one. */
|
|
|
457 #define SP_BANNED -1
|
|
236
|
458 #define SP_OK 0
|
|
307
|
459 #define SP_RARE 1
|
|
|
460 #define SP_LOCAL 2
|
|
|
461 #define SP_BAD 3
|
|
236
|
462
|
|
381
|
463 /* file used for "zG" and "zW" */
|
|
385
|
464 static char_u *int_wordlist = NULL;
|
|
381
|
465
|
|
236
|
466 /*
|
|
323
|
467 * Information used when looking for suggestions.
|
|
|
468 */
|
|
|
469 typedef struct suginfo_S
|
|
|
470 {
|
|
|
471 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
472 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
473 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
474 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
475 char_u *su_badptr; /* start of bad word in line */
|
|
|
476 int su_badlen; /* length of detected bad word in line */
|
|
346
|
477 int su_badflags; /* caps flags for bad word */
|
|
323
|
478 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
479 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
534
|
480 char_u su_sal_badword[MAXWLEN]; /* su_badword soundfolded */
|
|
|
481 slang_T *su_slang_first; /* slang_T used for su_sal_badword */
|
|
323
|
482 hashtab_T su_banned; /* table with banned words */
|
|
503
|
483 slang_T *su_sallang; /* default language for sound folding */
|
|
323
|
484 } suginfo_T;
|
|
|
485
|
|
|
486 /* One word suggestion. Used in "si_ga". */
|
|
|
487 typedef struct suggest_S
|
|
|
488 {
|
|
|
489 char_u *st_word; /* suggested word, allocated string */
|
|
|
490 int st_orglen; /* length of replaced text */
|
|
|
491 int st_score; /* lower is better */
|
|
344
|
492 int st_altscore; /* used when st_score compares equal */
|
|
|
493 int st_salscore; /* st_score is for soundalike */
|
|
324
|
494 int st_had_bonus; /* bonus already included in score */
|
|
503
|
495 slang_T *st_slang; /* language used for sound folding */
|
|
323
|
496 } suggest_T;
|
|
|
497
|
|
344
|
498 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
499
|
|
324
|
500 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
501 * called the score may change, thus we need to keep more than what is
|
|
|
502 * displayed. */
|
|
351
|
503 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
504
|
|
|
505 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
506 * of suggestions that are not going to be displayed. */
|
|
344
|
507 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
508
|
|
|
509 /* score for various changes */
|
|
344
|
510 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
511 #define SCORE_ICASE 52 /* slightly different case */
|
|
483
|
512 #define SCORE_REGION 200 /* word is for different region */
|
|
323
|
513 #define SCORE_RARE 180 /* rare word */
|
|
|
514 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
515 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
532
|
516 #define SCORE_REP 65 /* REP replacement */
|
|
323
|
517 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
518 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
479
|
519 #define SCORE_SUBCOMP 33 /* substitute a composing character */
|
|
324
|
520 #define SCORE_DEL 94 /* delete a character */
|
|
532
|
521 #define SCORE_DELDUP 66 /* delete a duplicated character */
|
|
479
|
522 #define SCORE_DELCOMP 28 /* delete a composing character */
|
|
324
|
523 #define SCORE_INS 96 /* insert a character */
|
|
532
|
524 #define SCORE_INSDUP 67 /* insert a duplicate character */
|
|
475
|
525 #define SCORE_INSCOMP 30 /* insert a composing character */
|
|
366
|
526 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
527
|
|
372
|
528 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
529 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
530 * 350 allows for about three changes. */
|
|
344
|
531
|
|
|
532 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
533 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
534
|
|
|
535 /*
|
|
236
|
536 * Structure to store info for word matching.
|
|
|
537 */
|
|
|
538 typedef struct matchinf_S
|
|
|
539 {
|
|
|
540 langp_T *mi_lp; /* info for language and region */
|
|
243
|
541
|
|
|
542 /* pointers to original text to be checked */
|
|
236
|
543 char_u *mi_word; /* start of word being checked */
|
|
339
|
544 char_u *mi_end; /* end of matching word so far */
|
|
243
|
545 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
546 char_u *mi_cend; /* char after what was used for
|
|
|
547 mi_capflags */
|
|
243
|
548
|
|
|
549 /* case-folded text */
|
|
|
550 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
551 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
552
|
|
339
|
553 /* for when checking word after a prefix */
|
|
|
554 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
481
|
555 affixID/condition */
|
|
339
|
556 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
557 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
558 #ifdef FEAT_MBYTE
|
|
|
559 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
560 case */
|
|
|
561 #else
|
|
|
562 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
563 #endif
|
|
339
|
564
|
|
481
|
565 /* for when checking a compound word */
|
|
|
566 int mi_compoff; /* start of following word offset */
|
|
483
|
567 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */
|
|
|
568 int mi_complen; /* nr of compound words used */
|
|
481
|
569
|
|
243
|
570 /* others */
|
|
236
|
571 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
572 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
573 buf_T *mi_buf; /* buffer being checked */
|
|
492
|
574
|
|
|
575 /* for NOBREAK */
|
|
|
576 int mi_result2; /* "mi_resul" without following word */
|
|
|
577 char_u *mi_end2; /* "mi_end" without following word */
|
|
236
|
578 } matchinf_T;
|
|
|
579
|
|
307
|
580 /*
|
|
|
581 * The tables used for recognizing word characters according to spelling.
|
|
|
582 * These are only used for the first 256 characters of 'encoding'.
|
|
|
583 */
|
|
|
584 typedef struct spelltab_S
|
|
|
585 {
|
|
|
586 char_u st_isw[256]; /* flags: is word char */
|
|
|
587 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
588 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
589 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
590 } spelltab_T;
|
|
|
591
|
|
|
592 static spelltab_T spelltab;
|
|
|
593 static int did_set_spelltab;
|
|
|
594
|
|
324
|
595 #define CF_WORD 0x01
|
|
|
596 #define CF_UPPER 0x02
|
|
307
|
597
|
|
|
598 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
599 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
600 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
601 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
602 #ifdef FEAT_MBYTE
|
|
|
603 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
604 #endif
|
|
483
|
605 static int write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
606
|
|
|
607 /*
|
|
344
|
608 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
609 */
|
|
|
610 typedef enum
|
|
|
611 {
|
|
344
|
612 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
613 * ends); if badword ends there is a match, otherwise
|
|
|
614 * try splitting word. */
|
|
375
|
615 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
616 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
617 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
618 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
619 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
620 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
621 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
622 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
623 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
624 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
625 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
626 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
627 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
628 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
629 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
630 STATE_FINAL /* End of this node. */
|
|
|
631 } state_T;
|
|
|
632
|
|
|
633 /*
|
|
346
|
634 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
635 */
|
|
|
636 typedef struct trystate_S
|
|
|
637 {
|
|
330
|
638 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
639 int ts_score; /* score */
|
|
344
|
640 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
641 short ts_curi; /* index in list of child nodes */
|
|
|
642 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
643 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
644 char_u ts_twordlen; /* valid length of tword[] */
|
|
482
|
645 char_u ts_prefixdepth; /* stack depth for end of prefix or
|
|
485
|
646 * PFD_PREFIXTREE or PFD_NOPREFIX */
|
|
|
647 char_u ts_flags; /* TSF_ flags */
|
|
330
|
648 #ifdef FEAT_MBYTE
|
|
|
649 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
650 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
651 char_u ts_isdiff; /* DIFF_ values */
|
|
|
652 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
653 #endif
|
|
483
|
654 char_u ts_prewordlen; /* length of word in "preword[]" */
|
|
|
655 char_u ts_splitoff; /* index in "tword" after last split */
|
|
492
|
656 char_u ts_splitfidx; /* "ts_fidx" at word split */
|
|
483
|
657 char_u ts_complen; /* nr of compound words used */
|
|
485
|
658 char_u ts_compsplit; /* index for "compflags" where word was spit */
|
|
346
|
659 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
660 } trystate_T;
|
|
|
661
|
|
330
|
662 /* values for ts_isdiff */
|
|
|
663 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
664 #define DIFF_YES 1 /* different byte found */
|
|
|
665 #define DIFF_INSERT 2 /* inserting character */
|
|
|
666
|
|
485
|
667 /* values for ts_flags */
|
|
|
668 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */
|
|
|
669 #define TSF_DIDSPLIT 2 /* tried split at this point */
|
|
|
670
|
|
375
|
671 /* special values ts_prefixdepth */
|
|
485
|
672 #define PFD_NOPREFIX 0xff /* not using prefixes */
|
|
482
|
673 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
485
|
674 #define PFD_NOTSPECIAL 0xfd /* first value that's not special */
|
|
375
|
675
|
|
339
|
676 /* mode values for find_word */
|
|
481
|
677 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
678 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
679 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
680 #define FIND_COMPOUND 3 /* find case-folded compound word */
|
|
|
681 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */
|
|
339
|
682
|
|
236
|
683 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
684 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
685 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
686 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
483
|
687 static int can_compound __ARGS((slang_T *slang, char_u *word, char_u *flags));
|
|
455
|
688 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
485
|
689 static void find_prefix __ARGS((matchinf_T *mip, int mode));
|
|
339
|
690 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
691 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
498
|
692 static int no_spell_checking __ARGS((win_T *wp));
|
|
307
|
693 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
694 static char_u *spell_enc __ARGS((void));
|
|
385
|
695 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
696 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
697 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
698 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
483
|
699 static char_u *read_string __ARGS((FILE *fd, int cnt));
|
|
|
700 static int read_region_section __ARGS((FILE *fd, slang_T *slang, int len));
|
|
|
701 static int read_charflags_section __ARGS((FILE *fd));
|
|
|
702 static int read_prefcond_section __ARGS((FILE *fd, slang_T *lp));
|
|
|
703 static int read_rep_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
704 static int read_sal_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
705 static int read_sofo_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
706 static int read_compound __ARGS((FILE *fd, slang_T *slang, int len));
|
|
495
|
707 static int byte_in_str __ARGS((char_u *str, int byte));
|
|
483
|
708 static int init_syl_tab __ARGS((slang_T *slang));
|
|
|
709 static int count_syllables __ARGS((slang_T *slang, char_u *word));
|
|
381
|
710 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
711 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
712 #ifdef FEAT_MBYTE
|
|
|
713 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
714 #endif
|
|
339
|
715 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
716 static void clear_midword __ARGS((buf_T *buf));
|
|
|
717 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
718 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
719 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
720 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
721 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
483
|
722 static void set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
723 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
724 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
475
|
725 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col));
|
|
388
|
726 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
727 #ifdef FEAT_EVAL
|
|
|
728 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
729 #endif
|
|
|
730 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
731 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
732 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
733 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
734 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
735 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
736 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
737 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
738 #ifdef FEAT_MBYTE
|
|
|
739 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
740 #endif
|
|
323
|
741 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
742 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
743 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
744 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
745 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
746 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
747 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
748 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
503
|
749 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus, slang_T *slang));
|
|
323
|
750 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
751 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
752 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
753 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
534
|
754 static void rescore_one __ARGS((suginfo_T *su, suggest_T *stp));
|
|
344
|
755 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
756 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
757 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
758 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
759 #ifdef FEAT_MBYTE
|
|
375
|
760 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
761 #endif
|
|
344
|
762 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
763 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
764 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
500
|
765 static linenr_T dump_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
766
|
|
324
|
767 /*
|
|
|
768 * Use our own character-case definitions, because the current locale may
|
|
|
769 * differ from what the .spl file uses.
|
|
|
770 * These must not be called with negative number!
|
|
|
771 */
|
|
|
772 #ifndef FEAT_MBYTE
|
|
|
773 /* Non-multi-byte implementation. */
|
|
|
774 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
775 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
776 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
777 #else
|
|
447
|
778 # if defined(HAVE_WCHAR_H)
|
|
|
779 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
780 # endif
|
|
324
|
781 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
782 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
783 * the "w" library function for characters above 255 if available. */
|
|
|
784 # ifdef HAVE_TOWLOWER
|
|
|
785 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
786 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
787 # else
|
|
|
788 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
789 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
790 # endif
|
|
|
791
|
|
|
792 # ifdef HAVE_TOWUPPER
|
|
|
793 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
794 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
795 # else
|
|
|
796 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
797 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
798 # endif
|
|
|
799
|
|
|
800 # ifdef HAVE_ISWUPPER
|
|
|
801 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
802 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
803 # else
|
|
|
804 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
805 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
806 # endif
|
|
|
807 #endif
|
|
|
808
|
|
307
|
809
|
|
|
810 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
811 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
482
|
812 static char *e_afftrailing = N_("Trailing text in %s line %d: %s");
|
|
495
|
813 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
|
814 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
815 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
469
|
816 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
817
|
|
|
818 /*
|
|
|
819 * Main spell-checking function.
|
|
300
|
820 * "ptr" points to a character that could be the start of a word.
|
|
534
|
821 * "*attrp" is set to the highlight index for a badly spelled word. For a
|
|
|
822 * non-word or when it's OK it remains unchanged.
|
|
236
|
823 * This must only be called when 'spelllang' is not empty.
|
|
323
|
824 *
|
|
385
|
825 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
826 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
827 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
828 * worry.
|
|
323
|
829 *
|
|
236
|
830 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
831 * caller can skip over the word.
|
|
|
832 */
|
|
|
833 int
|
|
385
|
834 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
835 win_T *wp; /* current window */
|
|
|
836 char_u *ptr;
|
|
534
|
837 hlf_T *attrp;
|
|
385
|
838 int *capcol; /* column to check for Capital */
|
|
236
|
839 {
|
|
|
840 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
841 be passed to functions quickly. */
|
|
344
|
842 int nrlen = 0; /* found a number first */
|
|
385
|
843 int c;
|
|
483
|
844 int wrongcaplen = 0;
|
|
500
|
845 int lpi;
|
|
236
|
846
|
|
307
|
847 /* A word never starts at a space or a control character. Return quickly
|
|
|
848 * then, skipping over the character. */
|
|
|
849 if (*ptr <= ' ')
|
|
|
850 return 1;
|
|
483
|
851 vim_memset(&mi, 0, sizeof(matchinf_T));
|
|
236
|
852
|
|
344
|
853 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
854 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
855 * "3GPP". */
|
|
300
|
856 if (*ptr >= '0' && *ptr <= '9')
|
|
|
857 {
|
|
316
|
858 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
859 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
860 else
|
|
|
861 {
|
|
344
|
862 mi.mi_end = skipdigits(ptr);
|
|
|
863 nrlen = mi.mi_end - ptr;
|
|
300
|
864 }
|
|
376
|
865 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
866 return (int)(mi.mi_end - ptr);
|
|
346
|
867
|
|
|
868 /* Try including the digits in the word. */
|
|
|
869 mi.mi_fend = ptr + nrlen;
|
|
|
870 }
|
|
|
871 else
|
|
|
872 mi.mi_fend = ptr;
|
|
|
873
|
|
|
874 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
875 mi.mi_word = ptr;
|
|
376
|
876 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
877 {
|
|
|
878 do
|
|
|
879 {
|
|
307
|
880 mb_ptr_adv(mi.mi_fend);
|
|
376
|
881 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
882
|
|
|
883 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
884 {
|
|
|
885 /* Check word starting with capital letter. */
|
|
455
|
886 c = PTR2CHAR(ptr);
|
|
385
|
887 if (!SPELL_ISUPPER(c))
|
|
483
|
888 wrongcaplen = (int)(mi.mi_fend - ptr);
|
|
385
|
889 }
|
|
|
890 }
|
|
|
891 if (capcol != NULL)
|
|
|
892 *capcol = -1;
|
|
344
|
893
|
|
|
894 /* We always use the characters up to the next non-word character,
|
|
|
895 * also for bad words. */
|
|
|
896 mi.mi_end = mi.mi_fend;
|
|
|
897
|
|
|
898 /* Check caps type later. */
|
|
376
|
899 mi.mi_buf = wp->w_buffer;
|
|
344
|
900
|
|
483
|
901 /* case-fold the word with one non-word character, so that we can check
|
|
|
902 * for the word end. */
|
|
344
|
903 if (*mi.mi_fend != NUL)
|
|
|
904 mb_ptr_adv(mi.mi_fend);
|
|
|
905
|
|
|
906 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
907 MAXWLEN + 1);
|
|
|
908 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
909
|
|
|
910 /* The word is bad unless we recognize it. */
|
|
|
911 mi.mi_result = SP_BAD;
|
|
492
|
912 mi.mi_result2 = SP_BAD;
|
|
344
|
913
|
|
|
914 /*
|
|
|
915 * Loop over the languages specified in 'spelllang'.
|
|
|
916 * We check them all, because a matching word may be longer than an
|
|
|
917 * already found matching word.
|
|
|
918 */
|
|
500
|
919 for (lpi = 0; lpi < wp->w_buffer->b_langp.ga_len; ++lpi)
|
|
|
920 {
|
|
|
921 mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, lpi);
|
|
|
922
|
|
|
923 /* If reloading fails the language is still in the list but everything
|
|
|
924 * has been cleared. */
|
|
|
925 if (mi.mi_lp->lp_slang->sl_fidxs == NULL)
|
|
|
926 continue;
|
|
|
927
|
|
344
|
928 /* Check for a matching word in case-folded words. */
|
|
|
929 find_word(&mi, FIND_FOLDWORD);
|
|
|
930
|
|
|
931 /* Check for a matching word in keep-case words. */
|
|
|
932 find_word(&mi, FIND_KEEPWORD);
|
|
|
933
|
|
|
934 /* Check for matching prefixes. */
|
|
485
|
935 find_prefix(&mi, FIND_FOLDWORD);
|
|
492
|
936
|
|
|
937 /* For a NOBREAK language, may want to use a word without a following
|
|
|
938 * word as a backup. */
|
|
|
939 if (mi.mi_lp->lp_slang->sl_nobreak && mi.mi_result == SP_BAD
|
|
|
940 && mi.mi_result2 != SP_BAD)
|
|
|
941 {
|
|
|
942 mi.mi_result = mi.mi_result2;
|
|
|
943 mi.mi_end = mi.mi_end2;
|
|
|
944 }
|
|
344
|
945 }
|
|
|
946
|
|
|
947 if (mi.mi_result != SP_OK)
|
|
|
948 {
|
|
346
|
949 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
950 * rare and local words, e.g., "3GPP". */
|
|
344
|
951 if (nrlen > 0)
|
|
346
|
952 {
|
|
|
953 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
954 return nrlen;
|
|
|
955 }
|
|
344
|
956
|
|
|
957 /* When we are at a non-word character there is no error, just
|
|
|
958 * skip over the character (try looking for a word after it). */
|
|
390
|
959 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
960 {
|
|
385
|
961 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
962 {
|
|
|
963 regmatch_T regmatch;
|
|
|
964
|
|
|
965 /* Check for end of sentence. */
|
|
|
966 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
967 regmatch.rm_ic = FALSE;
|
|
|
968 if (vim_regexec(®match, ptr, 0))
|
|
|
969 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
970 }
|
|
|
971
|
|
344
|
972 #ifdef FEAT_MBYTE
|
|
|
973 if (has_mbyte)
|
|
474
|
974 return (*mb_ptr2len)(ptr);
|
|
344
|
975 #endif
|
|
|
976 return 1;
|
|
300
|
977 }
|
|
483
|
978 else if (mi.mi_end == ptr)
|
|
|
979 /* Always include at least one character. Required for when there
|
|
|
980 * is a mixup in "midword". */
|
|
|
981 mb_ptr_adv(mi.mi_end);
|
|
492
|
982 else if (mi.mi_result == SP_BAD
|
|
|
983 && LANGP_ENTRY(wp->w_buffer->b_langp, 0)->lp_slang->sl_nobreak)
|
|
|
984 {
|
|
|
985 char_u *p, *fp;
|
|
|
986 int save_result = mi.mi_result;
|
|
|
987
|
|
|
988 /* First language in 'spelllang' is NOBREAK. Find first position
|
|
|
989 * at which any word would be valid. */
|
|
|
990 mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
500
|
991 if (mi.mi_lp->lp_slang->sl_fidxs != NULL)
|
|
|
992 {
|
|
|
993 p = mi.mi_word;
|
|
|
994 fp = mi.mi_fword;
|
|
|
995 for (;;)
|
|
|
996 {
|
|
|
997 mb_ptr_adv(p);
|
|
|
998 mb_ptr_adv(fp);
|
|
|
999 if (p >= mi.mi_end)
|
|
|
1000 break;
|
|
|
1001 mi.mi_compoff = fp - mi.mi_fword;
|
|
|
1002 find_word(&mi, FIND_COMPOUND);
|
|
|
1003 if (mi.mi_result != SP_BAD)
|
|
|
1004 {
|
|
|
1005 mi.mi_end = p;
|
|
|
1006 break;
|
|
|
1007 }
|
|
|
1008 }
|
|
|
1009 mi.mi_result = save_result;
|
|
|
1010 }
|
|
492
|
1011 }
|
|
243
|
1012
|
|
344
|
1013 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
534
|
1014 *attrp = HLF_SPB;
|
|
344
|
1015 else if (mi.mi_result == SP_RARE)
|
|
534
|
1016 *attrp = HLF_SPR;
|
|
344
|
1017 else
|
|
534
|
1018 *attrp = HLF_SPL;
|
|
243
|
1019 }
|
|
|
1020
|
|
483
|
1021 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE))
|
|
|
1022 {
|
|
|
1023 /* Report SpellCap only when the word isn't badly spelled. */
|
|
534
|
1024 *attrp = HLF_SPC;
|
|
483
|
1025 return wrongcaplen;
|
|
|
1026 }
|
|
|
1027
|
|
300
|
1028 return (int)(mi.mi_end - ptr);
|
|
236
|
1029 }
|
|
|
1030
|
|
|
1031 /*
|
|
300
|
1032 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
1033 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
1034 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
1035 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
1036 * tree.
|
|
300
|
1037 *
|
|
|
1038 * For a match mip->mi_result is updated.
|
|
243
|
1039 */
|
|
|
1040 static void
|
|
339
|
1041 find_word(mip, mode)
|
|
243
|
1042 matchinf_T *mip;
|
|
339
|
1043 int mode;
|
|
243
|
1044 {
|
|
324
|
1045 idx_T arridx = 0;
|
|
300
|
1046 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
1047 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
1048 int endidxcnt = 0;
|
|
|
1049 int len;
|
|
|
1050 int wlen = 0;
|
|
|
1051 int flen;
|
|
|
1052 int c;
|
|
|
1053 char_u *ptr;
|
|
324
|
1054 idx_T lo, hi, m;
|
|
243
|
1055 #ifdef FEAT_MBYTE
|
|
300
|
1056 char_u *s;
|
|
490
|
1057 #endif
|
|
339
|
1058 char_u *p;
|
|
307
|
1059 int res = SP_BAD;
|
|
300
|
1060 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1061 unsigned flags;
|
|
|
1062 char_u *byts;
|
|
324
|
1063 idx_T *idxs;
|
|
481
|
1064 int word_ends;
|
|
485
|
1065 int prefix_found;
|
|
492
|
1066 int nobreak_result;
|
|
481
|
1067
|
|
|
1068 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND)
|
|
236
|
1069 {
|
|
300
|
1070 /* Check for word with matching case in keep-case tree. */
|
|
|
1071 ptr = mip->mi_word;
|
|
|
1072 flen = 9999; /* no case folding, always enough bytes */
|
|
|
1073 byts = slang->sl_kbyts;
|
|
|
1074 idxs = slang->sl_kidxs;
|
|
481
|
1075
|
|
|
1076 if (mode == FIND_KEEPCOMPOUND)
|
|
|
1077 /* Skip over the previously found word(s). */
|
|
|
1078 wlen += mip->mi_compoff;
|
|
236
|
1079 }
|
|
|
1080 else
|
|
|
1081 {
|
|
300
|
1082 /* Check for case-folded in case-folded tree. */
|
|
|
1083 ptr = mip->mi_fword;
|
|
|
1084 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1085 byts = slang->sl_fbyts;
|
|
|
1086 idxs = slang->sl_fidxs;
|
|
339
|
1087
|
|
|
1088 if (mode == FIND_PREFIX)
|
|
|
1089 {
|
|
|
1090 /* Skip over the prefix. */
|
|
|
1091 wlen = mip->mi_prefixlen;
|
|
|
1092 flen -= mip->mi_prefixlen;
|
|
|
1093 }
|
|
481
|
1094 else if (mode == FIND_COMPOUND)
|
|
|
1095 {
|
|
|
1096 /* Skip over the previously found word(s). */
|
|
|
1097 wlen = mip->mi_compoff;
|
|
|
1098 flen -= mip->mi_compoff;
|
|
|
1099 }
|
|
|
1100
|
|
243
|
1101 }
|
|
|
1102
|
|
300
|
1103 if (byts == NULL)
|
|
|
1104 return; /* array is empty */
|
|
236
|
1105
|
|
|
1106 /*
|
|
307
|
1107 * Repeat advancing in the tree until:
|
|
|
1108 * - there is a byte that doesn't match,
|
|
|
1109 * - we reach the end of the tree,
|
|
|
1110 * - or we reach the end of the line.
|
|
236
|
1111 */
|
|
300
|
1112 for (;;)
|
|
236
|
1113 {
|
|
346
|
1114 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
1115 flen = fold_more(mip);
|
|
300
|
1116
|
|
|
1117 len = byts[arridx++];
|
|
|
1118
|
|
|
1119 /* If the first possible byte is a zero the word could end here.
|
|
|
1120 * Remember this index, we first check for the longest word. */
|
|
|
1121 if (byts[arridx] == 0)
|
|
|
1122 {
|
|
307
|
1123 if (endidxcnt == MAXWLEN)
|
|
|
1124 {
|
|
|
1125 /* Must be a corrupted spell file. */
|
|
|
1126 EMSG(_(e_format));
|
|
|
1127 return;
|
|
|
1128 }
|
|
300
|
1129 endlen[endidxcnt] = wlen;
|
|
|
1130 endidx[endidxcnt++] = arridx++;
|
|
|
1131 --len;
|
|
|
1132
|
|
|
1133 /* Skip over the zeros, there can be several flag/region
|
|
|
1134 * combinations. */
|
|
|
1135 while (len > 0 && byts[arridx] == 0)
|
|
|
1136 {
|
|
|
1137 ++arridx;
|
|
|
1138 --len;
|
|
|
1139 }
|
|
|
1140 if (len == 0)
|
|
|
1141 break; /* no children, word must end here */
|
|
|
1142 }
|
|
|
1143
|
|
|
1144 /* Stop looking at end of the line. */
|
|
|
1145 if (ptr[wlen] == NUL)
|
|
|
1146 break;
|
|
|
1147
|
|
|
1148 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1149 c = ptr[wlen];
|
|
346
|
1150 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1151 c = ' ';
|
|
300
|
1152 lo = arridx;
|
|
|
1153 hi = arridx + len - 1;
|
|
|
1154 while (lo < hi)
|
|
|
1155 {
|
|
|
1156 m = (lo + hi) / 2;
|
|
|
1157 if (byts[m] > c)
|
|
|
1158 hi = m - 1;
|
|
|
1159 else if (byts[m] < c)
|
|
|
1160 lo = m + 1;
|
|
|
1161 else
|
|
|
1162 {
|
|
|
1163 lo = hi = m;
|
|
|
1164 break;
|
|
236
|
1165 }
|
|
|
1166 }
|
|
300
|
1167
|
|
|
1168 /* Stop if there is no matching byte. */
|
|
|
1169 if (hi < lo || byts[lo] != c)
|
|
|
1170 break;
|
|
|
1171
|
|
|
1172 /* Continue at the child (if there is one). */
|
|
|
1173 arridx = idxs[lo];
|
|
|
1174 ++wlen;
|
|
|
1175 --flen;
|
|
346
|
1176
|
|
|
1177 /* One space in the good word may stand for several spaces in the
|
|
|
1178 * checked word. */
|
|
|
1179 if (c == ' ')
|
|
|
1180 {
|
|
|
1181 for (;;)
|
|
|
1182 {
|
|
|
1183 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1184 flen = fold_more(mip);
|
|
|
1185 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1186 break;
|
|
|
1187 ++wlen;
|
|
|
1188 --flen;
|
|
|
1189 }
|
|
|
1190 }
|
|
236
|
1191 }
|
|
|
1192
|
|
300
|
1193 /*
|
|
|
1194 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1195 * first.
|
|
|
1196 */
|
|
|
1197 while (endidxcnt > 0)
|
|
|
1198 {
|
|
|
1199 --endidxcnt;
|
|
|
1200 arridx = endidx[endidxcnt];
|
|
|
1201 wlen = endlen[endidxcnt];
|
|
236
|
1202
|
|
300
|
1203 #ifdef FEAT_MBYTE
|
|
|
1204 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1205 continue; /* not at first byte of character */
|
|
|
1206 #endif
|
|
376
|
1207 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
481
|
1208 {
|
|
492
|
1209 if (slang->sl_compprog == NULL && !slang->sl_nobreak)
|
|
481
|
1210 continue; /* next char is a word character */
|
|
|
1211 word_ends = FALSE;
|
|
|
1212 }
|
|
|
1213 else
|
|
|
1214 word_ends = TRUE;
|
|
485
|
1215 /* The prefix flag is before compound flags. Once a valid prefix flag
|
|
|
1216 * has been found we try compound flags. */
|
|
|
1217 prefix_found = FALSE;
|
|
300
|
1218
|
|
|
1219 #ifdef FEAT_MBYTE
|
|
339
|
1220 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1221 {
|
|
|
1222 /* Compute byte length in original word, length may change
|
|
339
|
1223 * when folding case. This can be slow, take a shortcut when the
|
|
|
1224 * case-folded word is equal to the keep-case word. */
|
|
300
|
1225 p = mip->mi_word;
|
|
339
|
1226 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1227 {
|
|
|
1228 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1229 mb_ptr_adv(p);
|
|
|
1230 wlen = p - mip->mi_word;
|
|
|
1231 }
|
|
300
|
1232 }
|
|
|
1233 #endif
|
|
236
|
1234
|
|
339
|
1235 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1236 * prefix ID.
|
|
|
1237 * Repeat this if there are more flags/region alternatives until there
|
|
|
1238 * is a match. */
|
|
|
1239 res = SP_BAD;
|
|
|
1240 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1241 --len, ++arridx)
|
|
300
|
1242 {
|
|
|
1243 flags = idxs[arridx];
|
|
324
|
1244
|
|
339
|
1245 /* For the fold-case tree check that the case of the checked word
|
|
|
1246 * matches with what the word in the tree requires.
|
|
|
1247 * For keep-case tree the case is always right. For prefixes we
|
|
|
1248 * don't bother to check. */
|
|
|
1249 if (mode == FIND_FOLDWORD)
|
|
300
|
1250 {
|
|
|
1251 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1252 {
|
|
323
|
1253 /* mi_capflags was set for a different word length, need
|
|
|
1254 * to do it again. */
|
|
300
|
1255 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1256 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1257 }
|
|
|
1258
|
|
346
|
1259 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1260 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1261 continue;
|
|
300
|
1262 }
|
|
236
|
1263
|
|
339
|
1264 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1265 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1266 * mip->mi_prefarridx that find_prefix() filled. */
|
|
485
|
1267 else if (mode == FIND_PREFIX && !prefix_found)
|
|
481
|
1268 {
|
|
366
|
1269 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1270 flags,
|
|
455
|
1271 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1272 FALSE);
|
|
366
|
1273 if (c == 0)
|
|
339
|
1274 continue;
|
|
366
|
1275
|
|
|
1276 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1277 if (c & WF_RAREPFX)
|
|
|
1278 flags |= WF_RARE;
|
|
485
|
1279 prefix_found = TRUE;
|
|
339
|
1280 }
|
|
|
1281
|
|
492
|
1282 if (slang->sl_nobreak)
|
|
|
1283 {
|
|
|
1284 if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND)
|
|
|
1285 && (flags & WF_BANNED) == 0)
|
|
|
1286 {
|
|
|
1287 /* NOBREAK: found a valid following word. That's all we
|
|
|
1288 * need to know, so return. */
|
|
|
1289 mip->mi_result = SP_OK;
|
|
|
1290 break;
|
|
|
1291 }
|
|
|
1292 }
|
|
|
1293
|
|
|
1294 else if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND
|
|
|
1295 || !word_ends))
|
|
481
|
1296 {
|
|
483
|
1297 /* If there is no flag or the word is shorter than
|
|
|
1298 * COMPOUNDMIN reject it quickly.
|
|
|
1299 * Makes you wonder why someone puts a compound flag on a word
|
|
481
|
1300 * that's too short... Myspell compatibility requires this
|
|
|
1301 * anyway. */
|
|
490
|
1302 if (((unsigned)flags >> 24) == 0
|
|
|
1303 || wlen - mip->mi_compoff < slang->sl_compminlen)
|
|
483
|
1304 continue;
|
|
500
|
1305 #ifdef FEAT_MBYTE
|
|
|
1306 /* For multi-byte chars check character length against
|
|
|
1307 * COMPOUNDMIN. */
|
|
|
1308 if (has_mbyte
|
|
501
|
1309 && slang->sl_compminlen > 0
|
|
500
|
1310 && mb_charlen_len(mip->mi_word + mip->mi_compoff,
|
|
|
1311 wlen - mip->mi_compoff) < slang->sl_compminlen)
|
|
|
1312 continue;
|
|
|
1313 #endif
|
|
483
|
1314
|
|
490
|
1315 /* Limit the number of compound words to COMPOUNDMAX if no
|
|
|
1316 * maximum for syllables is specified. */
|
|
|
1317 if (!word_ends && mip->mi_complen + 2 > slang->sl_compmax
|
|
|
1318 && slang->sl_compsylmax == MAXWLEN)
|
|
483
|
1319 continue;
|
|
|
1320
|
|
485
|
1321 /* Quickly check if compounding is possible with this flag. */
|
|
495
|
1322 if (!byte_in_str(mip->mi_complen == 0
|
|
485
|
1323 ? slang->sl_compstartflags
|
|
|
1324 : slang->sl_compallflags,
|
|
495
|
1325 ((unsigned)flags >> 24)))
|
|
481
|
1326 continue;
|
|
|
1327
|
|
490
|
1328 if (mode == FIND_COMPOUND)
|
|
|
1329 {
|
|
|
1330 int capflags;
|
|
|
1331
|
|
|
1332 /* Need to check the caps type of the appended compound
|
|
|
1333 * word. */
|
|
|
1334 #ifdef FEAT_MBYTE
|
|
|
1335 if (has_mbyte && STRNCMP(ptr, mip->mi_word,
|
|
|
1336 mip->mi_compoff) != 0)
|
|
|
1337 {
|
|
|
1338 /* case folding may have changed the length */
|
|
|
1339 p = mip->mi_word;
|
|
|
1340 for (s = ptr; s < ptr + mip->mi_compoff; mb_ptr_adv(s))
|
|
|
1341 mb_ptr_adv(p);
|
|
|
1342 }
|
|
|
1343 else
|
|
|
1344 #endif
|
|
|
1345 p = mip->mi_word + mip->mi_compoff;
|
|
|
1346 capflags = captype(p, mip->mi_word + wlen);
|
|
|
1347 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP
|
|
|
1348 && (flags & WF_FIXCAP) != 0))
|
|
|
1349 continue;
|
|
|
1350
|
|
|
1351 if (capflags != WF_ALLCAP)
|
|
|
1352 {
|
|
|
1353 /* When the character before the word is a word
|
|
|
1354 * character we do not accept a Onecap word. We do
|
|
|
1355 * accept a no-caps word, even when the dictionary
|
|
|
1356 * word specifies ONECAP. */
|
|
|
1357 mb_ptr_back(mip->mi_word, p);
|
|
|
1358 if (spell_iswordp_nmw(p)
|
|
|
1359 ? capflags == WF_ONECAP
|
|
|
1360 : (flags & WF_ONECAP) != 0
|
|
|
1361 && capflags != WF_ONECAP)
|
|
|
1362 continue;
|
|
|
1363 }
|
|
|
1364 }
|
|
|
1365
|
|
483
|
1366 /* If the word ends the sequence of compound flags of the
|
|
|
1367 * words must match with one of the COMPOUNDFLAGS items and
|
|
|
1368 * the number of syllables must not be too large. */
|
|
|
1369 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24);
|
|
|
1370 mip->mi_compflags[mip->mi_complen + 1] = NUL;
|
|
|
1371 if (word_ends)
|
|
|
1372 {
|
|
|
1373 char_u fword[MAXWLEN];
|
|
|
1374
|
|
|
1375 if (slang->sl_compsylmax < MAXWLEN)
|
|
|
1376 {
|
|
|
1377 /* "fword" is only needed for checking syllables. */
|
|
|
1378 if (ptr == mip->mi_word)
|
|
|
1379 (void)spell_casefold(ptr, wlen, fword, MAXWLEN);
|
|
|
1380 else
|
|
|
1381 vim_strncpy(fword, ptr, endlen[endidxcnt]);
|
|
|
1382 }
|
|
|
1383 if (!can_compound(slang, fword, mip->mi_compflags))
|
|
|
1384 continue;
|
|
|
1385 }
|
|
481
|
1386 }
|
|
|
1387
|
|
500
|
1388 /* Check NEEDCOMPOUND: can't use word without compounding. */
|
|
|
1389 else if (flags & WF_NEEDCOMP)
|
|
|
1390 continue;
|
|
|
1391
|
|
492
|
1392 nobreak_result = SP_OK;
|
|
|
1393
|
|
481
|
1394 if (!word_ends)
|
|
|
1395 {
|
|
492
|
1396 int save_result = mip->mi_result;
|
|
|
1397 char_u *save_end = mip->mi_end;
|
|
501
|
1398 langp_T *save_lp = mip->mi_lp;
|
|
|
1399 int lpi;
|
|
492
|
1400
|
|
|
1401 /* Check that a valid word follows. If there is one and we
|
|
|
1402 * are compounding, it will set "mi_result", thus we are
|
|
|
1403 * always finished here. For NOBREAK we only check that a
|
|
|
1404 * valid word follows.
|
|
481
|
1405 * Recursive! */
|
|
492
|
1406 if (slang->sl_nobreak)
|
|
|
1407 mip->mi_result = SP_BAD;
|
|
481
|
1408
|
|
|
1409 /* Find following word in case-folded tree. */
|
|
|
1410 mip->mi_compoff = endlen[endidxcnt];
|
|
|
1411 #ifdef FEAT_MBYTE
|
|
|
1412 if (has_mbyte && mode == FIND_KEEPWORD)
|
|
|
1413 {
|
|
|
1414 /* Compute byte length in case-folded word from "wlen":
|
|
|
1415 * byte length in keep-case word. Length may change when
|
|
|
1416 * folding case. This can be slow, take a shortcut when
|
|
|
1417 * the case-folded word is equal to the keep-case word. */
|
|
|
1418 p = mip->mi_fword;
|
|
|
1419 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1420 {
|
|
|
1421 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1422 mb_ptr_adv(p);
|
|
|
1423 mip->mi_compoff = p - mip->mi_fword;
|
|
|
1424 }
|
|
|
1425 }
|
|
|
1426 #endif
|
|
485
|
1427 c = mip->mi_compoff;
|
|
483
|
1428 ++mip->mi_complen;
|
|
501
|
1429
|
|
|
1430 /* For NOBREAK we need to try all NOBREAK languages, at least
|
|
|
1431 * to find the ".add" file(s). */
|
|
|
1432 for (lpi = 0; lpi < mip->mi_buf->b_langp.ga_len; ++lpi)
|
|
|
1433 {
|
|
|
1434 if (slang->sl_nobreak)
|
|
|
1435 {
|
|
|
1436 mip->mi_lp = LANGP_ENTRY(mip->mi_buf->b_langp, lpi);
|
|
|
1437 if (mip->mi_lp->lp_slang->sl_fidxs == NULL
|
|
|
1438 || !mip->mi_lp->lp_slang->sl_nobreak)
|
|
|
1439 continue;
|
|
|
1440 }
|
|
|
1441
|
|
|
1442 find_word(mip, FIND_COMPOUND);
|
|
|
1443
|
|
|
1444 /* When NOBREAK any word that matches is OK. Otherwise we
|
|
|
1445 * need to find the longest match, thus try with keep-case
|
|
|
1446 * and prefix too. */
|
|
492
|
1447 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1448 {
|
|
501
|
1449 /* Find following word in keep-case tree. */
|
|
|
1450 mip->mi_compoff = wlen;
|
|
|
1451 find_word(mip, FIND_KEEPCOMPOUND);
|
|
|
1452
|
|
|
1453 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1454 {
|
|
|
1455 /* Check for following word with prefix. */
|
|
|
1456 mip->mi_compoff = c;
|
|
|
1457 find_prefix(mip, FIND_COMPOUND);
|
|
|
1458 }
|
|
492
|
1459 }
|
|
501
|
1460
|
|
|
1461 if (!slang->sl_nobreak)
|
|
|
1462 break;
|
|
492
|
1463 }
|
|
483
|
1464 --mip->mi_complen;
|
|
501
|
1465 mip->mi_lp = save_lp;
|
|
485
|
1466
|
|
492
|
1467 if (slang->sl_nobreak)
|
|
|
1468 {
|
|
|
1469 nobreak_result = mip->mi_result;
|
|
|
1470 mip->mi_result = save_result;
|
|
|
1471 mip->mi_end = save_end;
|
|
|
1472 }
|
|
|
1473 else
|
|
|
1474 {
|
|
|
1475 if (mip->mi_result == SP_OK)
|
|
|
1476 break;
|
|
|
1477 continue;
|
|
|
1478 }
|
|
481
|
1479 }
|
|
|
1480
|
|
339
|
1481 if (flags & WF_BANNED)
|
|
|
1482 res = SP_BANNED;
|
|
|
1483 else if (flags & WF_REGION)
|
|
|
1484 {
|
|
|
1485 /* Check region. */
|
|
390
|
1486 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1487 res = SP_OK;
|
|
339
|
1488 else
|
|
|
1489 res = SP_LOCAL;
|
|
300
|
1490 }
|
|
339
|
1491 else if (flags & WF_RARE)
|
|
|
1492 res = SP_RARE;
|
|
307
|
1493 else
|
|
339
|
1494 res = SP_OK;
|
|
|
1495
|
|
492
|
1496 /* Always use the longest match and the best result. For NOBREAK
|
|
|
1497 * we separately keep the longest match without a following good
|
|
|
1498 * word as a fall-back. */
|
|
|
1499 if (nobreak_result == SP_BAD)
|
|
|
1500 {
|
|
|
1501 if (mip->mi_result2 > res)
|
|
|
1502 {
|
|
|
1503 mip->mi_result2 = res;
|
|
|
1504 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1505 }
|
|
|
1506 else if (mip->mi_result2 == res
|
|
|
1507 && mip->mi_end2 < mip->mi_word + wlen)
|
|
|
1508 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1509 }
|
|
|
1510 else if (mip->mi_result > res)
|
|
339
|
1511 {
|
|
|
1512 mip->mi_result = res;
|
|
|
1513 mip->mi_end = mip->mi_word + wlen;
|
|
|
1514 }
|
|
351
|
1515 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1516 mip->mi_end = mip->mi_word + wlen;
|
|
|
1517
|
|
492
|
1518 if (mip->mi_result == SP_OK)
|
|
339
|
1519 break;
|
|
300
|
1520 }
|
|
|
1521
|
|
492
|
1522 if (mip->mi_result == SP_OK)
|
|
300
|
1523 break;
|
|
|
1524 }
|
|
236
|
1525 }
|
|
|
1526
|
|
323
|
1527 /*
|
|
483
|
1528 * Return TRUE if "flags" is a valid sequence of compound flags and
|
|
|
1529 * "word[len]" does not have too many syllables.
|
|
482
|
1530 */
|
|
|
1531 static int
|
|
483
|
1532 can_compound(slang, word, flags)
|
|
482
|
1533 slang_T *slang;
|
|
483
|
1534 char_u *word;
|
|
|
1535 char_u *flags;
|
|
|
1536 {
|
|
|
1537 regmatch_T regmatch;
|
|
495
|
1538 #ifdef FEAT_MBYTE
|
|
|
1539 char_u uflags[MAXWLEN * 2];
|
|
|
1540 int i;
|
|
|
1541 #endif
|
|
|
1542 char_u *p;
|
|
483
|
1543
|
|
|
1544 if (slang->sl_compprog == NULL)
|
|
|
1545 return FALSE;
|
|
495
|
1546 #ifdef FEAT_MBYTE
|
|
|
1547 if (enc_utf8)
|
|
|
1548 {
|
|
|
1549 /* Need to convert the single byte flags to utf8 characters. */
|
|
|
1550 p = uflags;
|
|
|
1551 for (i = 0; flags[i] != NUL; ++i)
|
|
|
1552 p += mb_char2bytes(flags[i], p);
|
|
|
1553 *p = NUL;
|
|
|
1554 p = uflags;
|
|
|
1555 }
|
|
|
1556 else
|
|
|
1557 #endif
|
|
|
1558 p = flags;
|
|
483
|
1559 regmatch.regprog = slang->sl_compprog;
|
|
|
1560 regmatch.rm_ic = FALSE;
|
|
495
|
1561 if (!vim_regexec(®match, p, 0))
|
|
483
|
1562 return FALSE;
|
|
|
1563
|
|
490
|
1564 /* Count the number of syllables. This may be slow, do it last. If there
|
|
|
1565 * are too many syllables AND the number of compound words is above
|
|
|
1566 * COMPOUNDMAX then compounding is not allowed. */
|
|
483
|
1567 if (slang->sl_compsylmax < MAXWLEN
|
|
|
1568 && count_syllables(slang, word) > slang->sl_compsylmax)
|
|
495
|
1569 return (int)STRLEN(flags) < slang->sl_compmax;
|
|
483
|
1570 return TRUE;
|
|
482
|
1571 }
|
|
|
1572
|
|
|
1573 /*
|
|
390
|
1574 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1575 * ID in "flags" for the word "word".
|
|
366
|
1576 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1577 */
|
|
|
1578 static int
|
|
455
|
1579 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1580 int totprefcnt; /* nr of prefix IDs */
|
|
|
1581 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1582 int flags;
|
|
351
|
1583 char_u *word;
|
|
|
1584 slang_T *slang;
|
|
455
|
1585 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1586 {
|
|
|
1587 int prefcnt;
|
|
|
1588 int pidx;
|
|
|
1589 regprog_T *rp;
|
|
|
1590 regmatch_T regmatch;
|
|
390
|
1591 int prefid;
|
|
|
1592
|
|
|
1593 prefid = (unsigned)flags >> 24;
|
|
351
|
1594 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1595 {
|
|
|
1596 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1597
|
|
|
1598 /* Check the prefix ID. */
|
|
|
1599 if (prefid != (pidx & 0xff))
|
|
|
1600 continue;
|
|
|
1601
|
|
390
|
1602 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1603 * suffix. */
|
|
|
1604 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1605 continue;
|
|
|
1606
|
|
351
|
1607 /* Check the condition, if there is one. The condition index is
|
|
366
|
1608 * stored in the two bytes above the prefix ID byte. */
|
|
|
1609 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1610 if (rp != NULL)
|
|
|
1611 {
|
|
|
1612 regmatch.regprog = rp;
|
|
|
1613 regmatch.rm_ic = FALSE;
|
|
|
1614 if (!vim_regexec(®match, word, 0))
|
|
|
1615 continue;
|
|
|
1616 }
|
|
455
|
1617 else if (cond_req)
|
|
|
1618 continue;
|
|
|
1619
|
|
|
1620 /* It's a match! Return the WF_ flags. */
|
|
366
|
1621 return pidx;
|
|
|
1622 }
|
|
|
1623 return 0;
|
|
351
|
1624 }
|
|
|
1625
|
|
|
1626 /*
|
|
339
|
1627 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1628 * If it does, then check the following word.
|
|
|
1629 *
|
|
485
|
1630 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a
|
|
|
1631 * prefix in a compound word.
|
|
|
1632 *
|
|
339
|
1633 * For a match mip->mi_result is updated.
|
|
|
1634 */
|
|
|
1635 static void
|
|
485
|
1636 find_prefix(mip, mode)
|
|
339
|
1637 matchinf_T *mip;
|
|
485
|
1638 int mode;
|
|
339
|
1639 {
|
|
|
1640 idx_T arridx = 0;
|
|
|
1641 int len;
|
|
|
1642 int wlen = 0;
|
|
|
1643 int flen;
|
|
|
1644 int c;
|
|
|
1645 char_u *ptr;
|
|
|
1646 idx_T lo, hi, m;
|
|
|
1647 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1648 char_u *byts;
|
|
|
1649 idx_T *idxs;
|
|
|
1650
|
|
375
|
1651 byts = slang->sl_pbyts;
|
|
|
1652 if (byts == NULL)
|
|
|
1653 return; /* array is empty */
|
|
|
1654
|
|
339
|
1655 /* We use the case-folded word here, since prefixes are always
|
|
|
1656 * case-folded. */
|
|
|
1657 ptr = mip->mi_fword;
|
|
|
1658 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
485
|
1659 if (mode == FIND_COMPOUND)
|
|
|
1660 {
|
|
|
1661 /* Skip over the previously found word(s). */
|
|
|
1662 ptr += mip->mi_compoff;
|
|
|
1663 flen -= mip->mi_compoff;
|
|
|
1664 }
|
|
339
|
1665 idxs = slang->sl_pidxs;
|
|
|
1666
|
|
|
1667 /*
|
|
|
1668 * Repeat advancing in the tree until:
|
|
|
1669 * - there is a byte that doesn't match,
|
|
|
1670 * - we reach the end of the tree,
|
|
|
1671 * - or we reach the end of the line.
|
|
|
1672 */
|
|
|
1673 for (;;)
|
|
|
1674 {
|
|
|
1675 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1676 flen = fold_more(mip);
|
|
|
1677
|
|
|
1678 len = byts[arridx++];
|
|
|
1679
|
|
|
1680 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1681 * Check if the following word matches and supports the prefix. */
|
|
|
1682 if (byts[arridx] == 0)
|
|
|
1683 {
|
|
|
1684 /* There can be several prefixes with different conditions. We
|
|
|
1685 * try them all, since we don't know which one will give the
|
|
|
1686 * longest match. The word is the same each time, pass the list
|
|
|
1687 * of possible prefixes to find_word(). */
|
|
|
1688 mip->mi_prefarridx = arridx;
|
|
|
1689 mip->mi_prefcnt = len;
|
|
|
1690 while (len > 0 && byts[arridx] == 0)
|
|
|
1691 {
|
|
|
1692 ++arridx;
|
|
|
1693 --len;
|
|
|
1694 }
|
|
|
1695 mip->mi_prefcnt -= len;
|
|
|
1696
|
|
|
1697 /* Find the word that comes after the prefix. */
|
|
|
1698 mip->mi_prefixlen = wlen;
|
|
485
|
1699 if (mode == FIND_COMPOUND)
|
|
|
1700 /* Skip over the previously found word(s). */
|
|
|
1701 mip->mi_prefixlen += mip->mi_compoff;
|
|
|
1702
|
|
455
|
1703 #ifdef FEAT_MBYTE
|
|
|
1704 if (has_mbyte)
|
|
|
1705 {
|
|
|
1706 /* Case-folded length may differ from original length. */
|
|
485
|
1707 mip->mi_cprefixlen = nofold_len(mip->mi_fword,
|
|
|
1708 mip->mi_prefixlen, mip->mi_word);
|
|
455
|
1709 }
|
|
|
1710 else
|
|
485
|
1711 mip->mi_cprefixlen = mip->mi_prefixlen;
|
|
455
|
1712 #endif
|
|
339
|
1713 find_word(mip, FIND_PREFIX);
|
|
|
1714
|
|
|
1715
|
|
|
1716 if (len == 0)
|
|
|
1717 break; /* no children, word must end here */
|
|
|
1718 }
|
|
|
1719
|
|
|
1720 /* Stop looking at end of the line. */
|
|
|
1721 if (ptr[wlen] == NUL)
|
|
|
1722 break;
|
|
|
1723
|
|
|
1724 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1725 c = ptr[wlen];
|
|
|
1726 lo = arridx;
|
|
|
1727 hi = arridx + len - 1;
|
|
|
1728 while (lo < hi)
|
|
|
1729 {
|
|
|
1730 m = (lo + hi) / 2;
|
|
|
1731 if (byts[m] > c)
|
|
|
1732 hi = m - 1;
|
|
|
1733 else if (byts[m] < c)
|
|
|
1734 lo = m + 1;
|
|
|
1735 else
|
|
|
1736 {
|
|
|
1737 lo = hi = m;
|
|
|
1738 break;
|
|
|
1739 }
|
|
|
1740 }
|
|
|
1741
|
|
|
1742 /* Stop if there is no matching byte. */
|
|
|
1743 if (hi < lo || byts[lo] != c)
|
|
|
1744 break;
|
|
|
1745
|
|
|
1746 /* Continue at the child (if there is one). */
|
|
|
1747 arridx = idxs[lo];
|
|
|
1748 ++wlen;
|
|
|
1749 --flen;
|
|
|
1750 }
|
|
|
1751 }
|
|
|
1752
|
|
|
1753 /*
|
|
|
1754 * Need to fold at least one more character. Do until next non-word character
|
|
|
1755 * for efficiency.
|
|
|
1756 * Return the length of the folded chars in bytes.
|
|
|
1757 */
|
|
|
1758 static int
|
|
|
1759 fold_more(mip)
|
|
|
1760 matchinf_T *mip;
|
|
|
1761 {
|
|
|
1762 int flen;
|
|
|
1763 char_u *p;
|
|
|
1764
|
|
|
1765 p = mip->mi_fend;
|
|
|
1766 do
|
|
|
1767 {
|
|
|
1768 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1769 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1770
|
|
|
1771 /* Include the non-word character so that we can check for the
|
|
|
1772 * word end. */
|
|
|
1773 if (*mip->mi_fend != NUL)
|
|
|
1774 mb_ptr_adv(mip->mi_fend);
|
|
|
1775
|
|
|
1776 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1777 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1778 MAXWLEN - mip->mi_fwordlen);
|
|
|
1779 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1780 mip->mi_fwordlen += flen;
|
|
|
1781 return flen;
|
|
|
1782 }
|
|
|
1783
|
|
|
1784 /*
|
|
323
|
1785 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1786 * case.
|
|
|
1787 */
|
|
|
1788 static int
|
|
389
|
1789 spell_valid_case(wordflags, treeflags)
|
|
|
1790 int wordflags; /* flags for the checked word. */
|
|
323
|
1791 int treeflags; /* flags for the word in the spell tree */
|
|
|
1792 {
|
|
389
|
1793 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1794 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1795 && ((treeflags & WF_ONECAP) == 0
|
|
|
1796 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1797 }
|
|
|
1798
|
|
351
|
1799 /*
|
|
|
1800 * Return TRUE if spell checking is not enabled.
|
|
|
1801 */
|
|
|
1802 static int
|
|
498
|
1803 no_spell_checking(wp)
|
|
|
1804 win_T *wp;
|
|
|
1805 {
|
|
|
1806 if (!wp->w_p_spell || *wp->w_buffer->b_p_spl == NUL)
|
|
351
|
1807 {
|
|
|
1808 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1809 return TRUE;
|
|
|
1810 }
|
|
|
1811 return FALSE;
|
|
|
1812 }
|
|
300
|
1813
|
|
236
|
1814 /*
|
|
|
1815 * Move to next spell error.
|
|
500
|
1816 * "curline" is FALSE for "[s", "]s", "[S" and "]S".
|
|
|
1817 * "curline" is TRUE to find word under/after cursor in the same line.
|
|
483
|
1818 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move
|
|
|
1819 * to after badly spelled word before the cursor.
|
|
495
|
1820 * Return 0 if not found, length of the badly spelled word otherwise.
|
|
236
|
1821 */
|
|
|
1822 int
|
|
498
|
1823 spell_move_to(wp, dir, allwords, curline, attrp)
|
|
|
1824 win_T *wp;
|
|
236
|
1825 int dir; /* FORWARD or BACKWARD */
|
|
500
|
1826 int allwords; /* TRUE for "[s"/"]s", FALSE for "[S"/"]S" */
|
|
323
|
1827 int curline;
|
|
534
|
1828 hlf_T *attrp; /* return: attributes of bad word or NULL
|
|
|
1829 (only when "dir" is FORWARD) */
|
|
236
|
1830 {
|
|
249
|
1831 linenr_T lnum;
|
|
|
1832 pos_T found_pos;
|
|
495
|
1833 int found_len = 0;
|
|
236
|
1834 char_u *line;
|
|
|
1835 char_u *p;
|
|
346
|
1836 char_u *endp;
|
|
534
|
1837 hlf_T attr;
|
|
236
|
1838 int len;
|
|
498
|
1839 int has_syntax = syntax_present(wp->w_buffer);
|
|
249
|
1840 int col;
|
|
|
1841 int can_spell;
|
|
346
|
1842 char_u *buf = NULL;
|
|
|
1843 int buflen = 0;
|
|
|
1844 int skip = 0;
|
|
385
|
1845 int capcol = -1;
|
|
500
|
1846 int found_one = FALSE;
|
|
|
1847 int wrapped = FALSE;
|
|
236
|
1848
|
|
498
|
1849 if (no_spell_checking(wp))
|
|
495
|
1850 return 0;
|
|
236
|
1851
|
|
249
|
1852 /*
|
|
|
1853 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1854 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1855 *
|
|
|
1856 * When searching backwards, we continue in the line to find the last
|
|
|
1857 * bad word (in the cursor line: before the cursor).
|
|
346
|
1858 *
|
|
|
1859 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1860 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1861 * though...
|
|
249
|
1862 */
|
|
498
|
1863 lnum = wp->w_cursor.lnum;
|
|
249
|
1864 found_pos.lnum = 0;
|
|
236
|
1865
|
|
|
1866 while (!got_int)
|
|
|
1867 {
|
|
498
|
1868 line = ml_get_buf(wp->w_buffer, lnum, FALSE);
|
|
346
|
1869
|
|
|
1870 len = STRLEN(line);
|
|
|
1871 if (buflen < len + MAXWLEN + 2)
|
|
|
1872 {
|
|
|
1873 vim_free(buf);
|
|
|
1874 buflen = len + MAXWLEN + 2;
|
|
|
1875 buf = alloc(buflen);
|
|
|
1876 if (buf == NULL)
|
|
|
1877 break;
|
|
|
1878 }
|
|
|
1879
|
|
385
|
1880 /* In first line check first word for Capital. */
|
|
|
1881 if (lnum == 1)
|
|
|
1882 capcol = 0;
|
|
|
1883
|
|
|
1884 /* For checking first word with a capital skip white space. */
|
|
|
1885 if (capcol == 0)
|
|
|
1886 capcol = skipwhite(line) - line;
|
|
|
1887
|
|
346
|
1888 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1889 * possible. */
|
|
|
1890 STRCPY(buf, line);
|
|
498
|
1891 if (lnum < wp->w_buffer->b_ml.ml_line_count)
|
|
346
|
1892 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1893
|
|
|
1894 p = buf + skip;
|
|
|
1895 endp = buf + len;
|
|
|
1896 while (p < endp)
|
|
236
|
1897 {
|
|
500
|
1898 /* When searching backward don't search after the cursor. Unless
|
|
|
1899 * we wrapped around the end of the buffer. */
|
|
300
|
1900 if (dir == BACKWARD
|
|
498
|
1901 && lnum == wp->w_cursor.lnum
|
|
500
|
1902 && !wrapped
|
|
498
|
1903 && (colnr_T)(p - buf) >= wp->w_cursor.col)
|
|
300
|
1904 break;
|
|
249
|
1905
|
|
300
|
1906 /* start of word */
|
|
534
|
1907 attr = HLF_COUNT;
|
|
498
|
1908 len = spell_check(wp, p, &attr, &capcol);
|
|
249
|
1909
|
|
534
|
1910 if (attr != HLF_COUNT)
|
|
300
|
1911 {
|
|
|
1912 /* We found a bad word. Check the attribute. */
|
|
534
|
1913 if (allwords || attr == HLF_SPB)
|
|
236
|
1914 {
|
|
500
|
1915 found_one = TRUE;
|
|
|
1916
|
|
300
|
1917 /* When searching forward only accept a bad word after
|
|
|
1918 * the cursor. */
|
|
|
1919 if (dir == BACKWARD
|
|
500
|
1920 || lnum != wp->w_cursor.lnum
|
|
498
|
1921 || (lnum == wp->w_cursor.lnum
|
|
500
|
1922 && (wrapped
|
|
|
1923 || (colnr_T)(curline ? p - buf + len
|
|
346
|
1924 : p - buf)
|
|
500
|
1925 > wp->w_cursor.col)))
|
|
236
|
1926 {
|
|
300
|
1927 if (has_syntax)
|
|
249
|
1928 {
|
|
346
|
1929 col = p - buf;
|
|
498
|
1930 (void)syn_get_id(wp, lnum, (colnr_T)col,
|
|
300
|
1931 FALSE, &can_spell);
|
|
|
1932 }
|
|
|
1933 else
|
|
|
1934 can_spell = TRUE;
|
|
249
|
1935
|
|
300
|
1936 if (can_spell)
|
|
|
1937 {
|
|
|
1938 found_pos.lnum = lnum;
|
|
346
|
1939 found_pos.col = p - buf;
|
|
249
|
1940 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1941 found_pos.coladd = 0;
|
|
249
|
1942 #endif
|
|
300
|
1943 if (dir == FORWARD)
|
|
|
1944 {
|
|
|
1945 /* No need to search further. */
|
|
498
|
1946 wp->w_cursor = found_pos;
|
|
346
|
1947 vim_free(buf);
|
|
498
|
1948 if (attrp != NULL)
|
|
|
1949 *attrp = attr;
|
|
495
|
1950 return len;
|
|
249
|
1951 }
|
|
483
|
1952 else if (curline)
|
|
|
1953 /* Insert mode completion: put cursor after
|
|
|
1954 * the bad word. */
|
|
|
1955 found_pos.col += len;
|
|
495
|
1956 found_len = len;
|
|
249
|
1957 }
|
|
236
|
1958 }
|
|
|
1959 }
|
|
|
1960 }
|
|
|
1961
|
|
300
|
1962 /* advance to character after the word */
|
|
|
1963 p += len;
|
|
385
|
1964 capcol -= len;
|
|
236
|
1965 }
|
|
|
1966
|
|
483
|
1967 if (dir == BACKWARD && found_pos.lnum != 0)
|
|
|
1968 {
|
|
500
|
1969 /* Use the last match in the line (before the cursor). */
|
|
498
|
1970 wp->w_cursor = found_pos;
|
|
483
|
1971 vim_free(buf);
|
|
495
|
1972 return found_len;
|
|
483
|
1973 }
|
|
|
1974
|
|
323
|
1975 if (curline)
|
|
346
|
1976 break; /* only check cursor line */
|
|
323
|
1977
|
|
236
|
1978 /* Advance to next line. */
|
|
249
|
1979 if (dir == BACKWARD)
|
|
|
1980 {
|
|
500
|
1981 /* If we are back at the starting line and searched it again there
|
|
|
1982 * is no match, give up. */
|
|
|
1983 if (lnum == wp->w_cursor.lnum && wrapped)
|
|
346
|
1984 break;
|
|
500
|
1985
|
|
|
1986 if (lnum > 1)
|
|
|
1987 --lnum;
|
|
|
1988 else if (!p_ws)
|
|
|
1989 break; /* at first line and 'nowrapscan' */
|
|
|
1990 else
|
|
|
1991 {
|
|
|
1992 /* Wrap around to the end of the buffer. May search the
|
|
|
1993 * starting line again and accept the last match. */
|
|
|
1994 lnum = wp->w_buffer->b_ml.ml_line_count;
|
|
|
1995 wrapped = TRUE;
|
|
503
|
1996 if (!shortmess(SHM_SEARCH))
|
|
|
1997 give_warning((char_u *)_(top_bot_msg), TRUE);
|
|
500
|
1998 }
|
|
385
|
1999 capcol = -1;
|
|
249
|
2000 }
|
|
|
2001 else
|
|
|
2002 {
|
|
500
|
2003 if (lnum < wp->w_buffer->b_ml.ml_line_count)
|
|
|
2004 ++lnum;
|
|
|
2005 else if (!p_ws)
|
|
|
2006 break; /* at first line and 'nowrapscan' */
|
|
|
2007 else
|
|
|
2008 {
|
|
|
2009 /* Wrap around to the start of the buffer. May search the
|
|
|
2010 * starting line again and accept the first match. */
|
|
|
2011 lnum = 1;
|
|
|
2012 wrapped = TRUE;
|
|
503
|
2013 if (!shortmess(SHM_SEARCH))
|
|
|
2014 give_warning((char_u *)_(bot_top_msg), TRUE);
|
|
500
|
2015 }
|
|
|
2016
|
|
|
2017 /* If we are back at the starting line and there is no match then
|
|
|
2018 * give up. */
|
|
|
2019 if (lnum == wp->w_cursor.lnum && !found_one)
|
|
346
|
2020 break;
|
|
|
2021
|
|
|
2022 /* Skip the characters at the start of the next line that were
|
|
|
2023 * included in a match crossing line boundaries. */
|
|
534
|
2024 if (attr == HLF_COUNT)
|
|
346
|
2025 skip = p - endp;
|
|
|
2026 else
|
|
|
2027 skip = 0;
|
|
385
|
2028
|
|
|
2029 /* Capscol skips over the inserted space. */
|
|
|
2030 --capcol;
|
|
|
2031
|
|
|
2032 /* But after empty line check first word in next line */
|
|
|
2033 if (*skipwhite(line) == NUL)
|
|
|
2034 capcol = 0;
|
|
249
|
2035 }
|
|
236
|
2036
|
|
|
2037 line_breakcheck();
|
|
|
2038 }
|
|
|
2039
|
|
346
|
2040 vim_free(buf);
|
|
495
|
2041 return 0;
|
|
346
|
2042 }
|
|
|
2043
|
|
|
2044 /*
|
|
|
2045 * For spell checking: concatenate the start of the following line "line" into
|
|
|
2046 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
2047 */
|
|
|
2048 void
|
|
|
2049 spell_cat_line(buf, line, maxlen)
|
|
|
2050 char_u *buf;
|
|
|
2051 char_u *line;
|
|
|
2052 int maxlen;
|
|
|
2053 {
|
|
|
2054 char_u *p;
|
|
|
2055 int n;
|
|
|
2056
|
|
|
2057 p = skipwhite(line);
|
|
|
2058 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
2059 p = skipwhite(p + 1);
|
|
|
2060
|
|
|
2061 if (*p != NUL)
|
|
|
2062 {
|
|
|
2063 *buf = ' ';
|
|
376
|
2064 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
2065 n = p - line;
|
|
|
2066 if (n >= maxlen)
|
|
|
2067 n = maxlen - 1;
|
|
|
2068 vim_memset(buf + 1, ' ', n);
|
|
|
2069 }
|
|
236
|
2070 }
|
|
|
2071
|
|
501
|
2072 typedef struct spelload_S
|
|
|
2073 {
|
|
|
2074 char_u sl_lang[MAXWLEN + 1]; /* language name */
|
|
|
2075 slang_T *sl_slang; /* resulting slang_T struct */
|
|
|
2076 int sl_nobreak; /* NOBREAK language found */
|
|
|
2077 } spelload_T;
|
|
|
2078
|
|
236
|
2079 /*
|
|
307
|
2080 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
2081 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
2082 */
|
|
307
|
2083 static void
|
|
236
|
2084 spell_load_lang(lang)
|
|
|
2085 char_u *lang;
|
|
|
2086 {
|
|
310
|
2087 char_u fname_enc[85];
|
|
236
|
2088 int r;
|
|
501
|
2089 spelload_T sl;
|
|
307
|
2090
|
|
310
|
2091 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
2092 * It's truncated when an error is detected. */
|
|
501
|
2093 STRCPY(sl.sl_lang, lang);
|
|
|
2094 sl.sl_slang = NULL;
|
|
|
2095 sl.sl_nobreak = FALSE;
|
|
307
|
2096
|
|
310
|
2097 /*
|
|
|
2098 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
2099 */
|
|
|
2100 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
2101 "spell/%s.%s.spl", lang, spell_enc());
|
|
501
|
2102 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl);
|
|
|
2103
|
|
|
2104 if (r == FAIL && *sl.sl_lang != NUL)
|
|
307
|
2105 {
|
|
|
2106 /* Try loading the ASCII version. */
|
|
310
|
2107 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
2108 "spell/%s.ascii.spl", lang);
|
|
501
|
2109 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl);
|
|
307
|
2110 }
|
|
|
2111
|
|
|
2112 if (r == FAIL)
|
|
483
|
2113 smsg((char_u *)_("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""),
|
|
|
2114 lang, spell_enc(), lang);
|
|
501
|
2115 else if (sl.sl_slang != NULL)
|
|
|
2116 {
|
|
|
2117 /* At least one file was loaded, now load all the additions. */
|
|
310
|
2118 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
501
|
2119 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &sl);
|
|
310
|
2120 }
|
|
|
2121 }
|
|
|
2122
|
|
|
2123 /*
|
|
|
2124 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
2125 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
2126 */
|
|
|
2127 static char_u *
|
|
|
2128 spell_enc()
|
|
|
2129 {
|
|
|
2130
|
|
|
2131 #ifdef FEAT_MBYTE
|
|
|
2132 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
2133 return p_enc;
|
|
|
2134 #endif
|
|
|
2135 return (char_u *)"latin1";
|
|
236
|
2136 }
|
|
|
2137
|
|
|
2138 /*
|
|
385
|
2139 * Get the name of the .spl file for the internal wordlist into
|
|
|
2140 * "fname[MAXPATHL]".
|
|
|
2141 */
|
|
|
2142 static void
|
|
|
2143 int_wordlist_spl(fname)
|
|
|
2144 char_u *fname;
|
|
|
2145 {
|
|
|
2146 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
2147 int_wordlist, spell_enc());
|
|
|
2148 }
|
|
|
2149
|
|
|
2150 /*
|
|
236
|
2151 * Allocate a new slang_T.
|
|
|
2152 * Caller must fill "sl_next".
|
|
|
2153 */
|
|
|
2154 static slang_T *
|
|
|
2155 slang_alloc(lang)
|
|
|
2156 char_u *lang;
|
|
|
2157 {
|
|
|
2158 slang_T *lp;
|
|
|
2159
|
|
300
|
2160 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
2161 if (lp != NULL)
|
|
|
2162 {
|
|
|
2163 lp->sl_name = vim_strsave(lang);
|
|
323
|
2164 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
483
|
2165 lp->sl_compmax = MAXWLEN;
|
|
|
2166 lp->sl_compsylmax = MAXWLEN;
|
|
236
|
2167 }
|
|
|
2168 return lp;
|
|
|
2169 }
|
|
|
2170
|
|
|
2171 /*
|
|
|
2172 * Free the contents of an slang_T and the structure itself.
|
|
|
2173 */
|
|
|
2174 static void
|
|
|
2175 slang_free(lp)
|
|
|
2176 slang_T *lp;
|
|
|
2177 {
|
|
|
2178 vim_free(lp->sl_name);
|
|
310
|
2179 vim_free(lp->sl_fname);
|
|
|
2180 slang_clear(lp);
|
|
|
2181 vim_free(lp);
|
|
|
2182 }
|
|
|
2183
|
|
|
2184 /*
|
|
|
2185 * Clear an slang_T so that the file can be reloaded.
|
|
|
2186 */
|
|
|
2187 static void
|
|
|
2188 slang_clear(lp)
|
|
|
2189 slang_T *lp;
|
|
|
2190 {
|
|
339
|
2191 garray_T *gap;
|
|
|
2192 fromto_T *ftp;
|
|
344
|
2193 salitem_T *smp;
|
|
339
|
2194 int i;
|
|
323
|
2195
|
|
300
|
2196 vim_free(lp->sl_fbyts);
|
|
310
|
2197 lp->sl_fbyts = NULL;
|
|
300
|
2198 vim_free(lp->sl_kbyts);
|
|
310
|
2199 lp->sl_kbyts = NULL;
|
|
339
|
2200 vim_free(lp->sl_pbyts);
|
|
|
2201 lp->sl_pbyts = NULL;
|
|
|
2202
|
|
300
|
2203 vim_free(lp->sl_fidxs);
|
|
310
|
2204 lp->sl_fidxs = NULL;
|
|
300
|
2205 vim_free(lp->sl_kidxs);
|
|
310
|
2206 lp->sl_kidxs = NULL;
|
|
339
|
2207 vim_free(lp->sl_pidxs);
|
|
|
2208 lp->sl_pidxs = NULL;
|
|
323
|
2209
|
|
344
|
2210 gap = &lp->sl_rep;
|
|
|
2211 while (gap->ga_len > 0)
|
|
323
|
2212 {
|
|
344
|
2213 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2214 vim_free(ftp->ft_from);
|
|
|
2215 vim_free(ftp->ft_to);
|
|
323
|
2216 }
|
|
344
|
2217 ga_clear(gap);
|
|
|
2218
|
|
|
2219 gap = &lp->sl_sal;
|
|
375
|
2220 if (lp->sl_sofo)
|
|
376
|
2221 {
|
|
|
2222 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
2223 if (gap->ga_data != NULL)
|
|
|
2224 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
2225 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2226 vim_free(((int **)gap->ga_data)[i]);
|
|
|
2227 }
|
|
375
|
2228 else
|
|
|
2229 /* SAL items: free salitem_T items */
|
|
|
2230 while (gap->ga_len > 0)
|
|
|
2231 {
|
|
|
2232 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2233 vim_free(smp->sm_lead);
|
|
|
2234 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
2235 vim_free(smp->sm_to);
|
|
|
2236 #ifdef FEAT_MBYTE
|
|
|
2237 vim_free(smp->sm_lead_w);
|
|
|
2238 vim_free(smp->sm_oneof_w);
|
|
|
2239 vim_free(smp->sm_to_w);
|
|
|
2240 #endif
|
|
|
2241 }
|
|
344
|
2242 ga_clear(gap);
|
|
323
|
2243
|
|
339
|
2244 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
2245 vim_free(lp->sl_prefprog[i]);
|
|
376
|
2246 lp->sl_prefixcnt = 0;
|
|
339
|
2247 vim_free(lp->sl_prefprog);
|
|
376
|
2248 lp->sl_prefprog = NULL;
|
|
|
2249
|
|
|
2250 vim_free(lp->sl_midword);
|
|
|
2251 lp->sl_midword = NULL;
|
|
339
|
2252
|
|
483
|
2253 vim_free(lp->sl_compprog);
|
|
|
2254 vim_free(lp->sl_compstartflags);
|
|
485
|
2255 vim_free(lp->sl_compallflags);
|
|
483
|
2256 lp->sl_compprog = NULL;
|
|
|
2257 lp->sl_compstartflags = NULL;
|
|
485
|
2258 lp->sl_compallflags = NULL;
|
|
483
|
2259
|
|
|
2260 vim_free(lp->sl_syllable);
|
|
|
2261 lp->sl_syllable = NULL;
|
|
|
2262 ga_clear(&lp->sl_syl_items);
|
|
481
|
2263
|
|
330
|
2264 #ifdef FEAT_MBYTE
|
|
|
2265 {
|
|
|
2266 int todo = lp->sl_map_hash.ht_used;
|
|
|
2267 hashitem_T *hi;
|
|
|
2268
|
|
|
2269 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
2270 if (!HASHITEM_EMPTY(hi))
|
|
|
2271 {
|
|
|
2272 --todo;
|
|
|
2273 vim_free(hi->hi_key);
|
|
|
2274 }
|
|
|
2275 }
|
|
|
2276 hash_clear(&lp->sl_map_hash);
|
|
|
2277 #endif
|
|
483
|
2278
|
|
|
2279 lp->sl_compmax = MAXWLEN;
|
|
501
|
2280 lp->sl_compminlen = 0;
|
|
483
|
2281 lp->sl_compsylmax = MAXWLEN;
|
|
|
2282 lp->sl_regions[0] = NUL;
|
|
236
|
2283 }
|
|
|
2284
|
|
|
2285 /*
|
|
307
|
2286 * Load one spell file and store the info into a slang_T.
|
|
236
|
2287 * Invoked through do_in_runtimepath().
|
|
|
2288 */
|
|
|
2289 static void
|
|
310
|
2290 spell_load_cb(fname, cookie)
|
|
236
|
2291 char_u *fname;
|
|
501
|
2292 void *cookie;
|
|
|
2293 {
|
|
|
2294 spelload_T *slp = (spelload_T *)cookie;
|
|
|
2295 slang_T *slang;
|
|
|
2296
|
|
|
2297 slang = spell_load_file(fname, slp->sl_lang, NULL, FALSE);
|
|
|
2298 if (slang != NULL)
|
|
|
2299 {
|
|
|
2300 /* When a previously loaded file has NOBREAK also use it for the
|
|
|
2301 * ".add" files. */
|
|
|
2302 if (slp->sl_nobreak && slang->sl_add)
|
|
|
2303 slang->sl_nobreak = TRUE;
|
|
|
2304 else if (slang->sl_nobreak)
|
|
|
2305 slp->sl_nobreak = TRUE;
|
|
|
2306
|
|
|
2307 slp->sl_slang = slang;
|
|
|
2308 }
|
|
310
|
2309 }
|
|
|
2310
|
|
|
2311 /*
|
|
|
2312 * Load one spell file and store the info into a slang_T.
|
|
|
2313 *
|
|
|
2314 * This is invoked in two ways:
|
|
|
2315 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
2316 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
2317 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
2318 * points to the existing slang_T.
|
|
323
|
2319 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
2320 */
|
|
323
|
2321 static slang_T *
|
|
|
2322 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
2323 char_u *fname;
|
|
|
2324 char_u *lang;
|
|
|
2325 slang_T *old_lp;
|
|
323
|
2326 int silent; /* no error if file doesn't exist */
|
|
310
|
2327 {
|
|
236
|
2328 FILE *fd;
|
|
483
|
2329 char_u buf[VIMSPELLMAGICL];
|
|
236
|
2330 char_u *p;
|
|
339
|
2331 char_u *bp;
|
|
|
2332 idx_T *ip;
|
|
236
|
2333 int i;
|
|
339
|
2334 int n;
|
|
300
|
2335 int len;
|
|
236
|
2336 int round;
|
|
|
2337 char_u *save_sourcing_name = sourcing_name;
|
|
|
2338 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
307
|
2339 slang_T *lp = NULL;
|
|
324
|
2340 idx_T idx;
|
|
344
|
2341 int c = 0;
|
|
483
|
2342 int res;
|
|
236
|
2343
|
|
310
|
2344 fd = mch_fopen((char *)fname, "r");
|
|
236
|
2345 if (fd == NULL)
|
|
|
2346 {
|
|
323
|
2347 if (!silent)
|
|
|
2348 EMSG2(_(e_notopen), fname);
|
|
|
2349 else if (p_verbose > 2)
|
|
|
2350 {
|
|
|
2351 verbose_enter();
|
|
|
2352 smsg((char_u *)e_notopen, fname);
|
|
|
2353 verbose_leave();
|
|
|
2354 }
|
|
255
|
2355 goto endFAIL;
|
|
236
|
2356 }
|
|
310
|
2357 if (p_verbose > 2)
|
|
|
2358 {
|
|
|
2359 verbose_enter();
|
|
|
2360 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
2361 verbose_leave();
|
|
|
2362 }
|
|
|
2363
|
|
|
2364 if (old_lp == NULL)
|
|
|
2365 {
|
|
|
2366 lp = slang_alloc(lang);
|
|
|
2367 if (lp == NULL)
|
|
|
2368 goto endFAIL;
|
|
|
2369
|
|
|
2370 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
2371 lp->sl_fname = vim_strsave(fname);
|
|
|
2372 if (lp->sl_fname == NULL)
|
|
|
2373 goto endFAIL;
|
|
|
2374
|
|
|
2375 /* Check for .add.spl. */
|
|
|
2376 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
2377 }
|
|
|
2378 else
|
|
|
2379 lp = old_lp;
|
|
307
|
2380
|
|
236
|
2381 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
2382 sourcing_name = fname;
|
|
|
2383 sourcing_lnum = 0;
|
|
|
2384
|
|
339
|
2385 /* <HEADER>: <fileID>
|
|
|
2386 */
|
|
236
|
2387 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
2388 buf[i] = getc(fd); /* <fileID> */
|
|
|
2389 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
2390 {
|
|
483
|
2391 EMSG(_("E757: This does not look like a spell file"));
|
|
381
|
2392 goto endFAIL;
|
|
483
|
2393 }
|
|
|
2394 c = getc(fd); /* <versionnr> */
|
|
|
2395 if (c < VIMSPELLVERSION)
|
|
|
2396 {
|
|
|
2397 EMSG(_("E771: Old spell file, needs to be updated"));
|
|
381
|
2398 goto endFAIL;
|
|
|
2399 }
|
|
483
|
2400 else if (c > VIMSPELLVERSION)
|
|
|
2401 {
|
|
|
2402 EMSG(_("E772: Spell file is for newer version of Vim"));
|
|
481
|
2403 goto endFAIL;
|
|
483
|
2404 }
|
|
|
2405
|
|
|
2406
|
|
|
2407 /*
|
|
|
2408 * <SECTIONS>: <section> ... <sectionend>
|
|
|
2409 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
2410 */
|
|
|
2411 for (;;)
|
|
|
2412 {
|
|
|
2413 n = getc(fd); /* <sectionID> or <sectionend> */
|
|
|
2414 if (n == SN_END)
|
|
|
2415 break;
|
|
|
2416 c = getc(fd); /* <sectionflags> */
|
|
|
2417 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2418 /* <sectionlen> */
|
|
|
2419 if (len < 0)
|
|
|
2420 goto truncerr;
|
|
|
2421
|
|
|
2422 res = 0;
|
|
|
2423 switch (n)
|
|
|
2424 {
|
|
|
2425 case SN_REGION:
|
|
|
2426 res = read_region_section(fd, lp, len);
|
|
|
2427 break;
|
|
|
2428
|
|
|
2429 case SN_CHARFLAGS:
|
|
|
2430 res = read_charflags_section(fd);
|
|
|
2431 break;
|
|
|
2432
|
|
|
2433 case SN_MIDWORD:
|
|
|
2434 lp->sl_midword = read_string(fd, len); /* <midword> */
|
|
|
2435 if (lp->sl_midword == NULL)
|
|
|
2436 goto endFAIL;
|
|
|
2437 break;
|
|
|
2438
|
|
|
2439 case SN_PREFCOND:
|
|
|
2440 res = read_prefcond_section(fd, lp);
|
|
|
2441 break;
|
|
|
2442
|
|
|
2443 case SN_REP:
|
|
|
2444 res = read_rep_section(fd, lp);
|
|
|
2445 break;
|
|
|
2446
|
|
|
2447 case SN_SAL:
|
|
|
2448 res = read_sal_section(fd, lp);
|
|
|
2449 break;
|
|
|
2450
|
|
|
2451 case SN_SOFO:
|
|
|
2452 res = read_sofo_section(fd, lp);
|
|
|
2453 break;
|
|
|
2454
|
|
|
2455 case SN_MAP:
|
|
|
2456 p = read_string(fd, len); /* <mapstr> */
|
|
|
2457 if (p == NULL)
|
|
|
2458 goto endFAIL;
|
|
|
2459 set_map_str(lp, p);
|
|
|
2460 vim_free(p);
|
|
|
2461 break;
|
|
|
2462
|
|
|
2463 case SN_COMPOUND:
|
|
|
2464 res = read_compound(fd, lp, len);
|
|
|
2465 break;
|
|
|
2466
|
|
492
|
2467 case SN_NOBREAK:
|
|
|
2468 lp->sl_nobreak = TRUE;
|
|
|
2469 break;
|
|
|
2470
|
|
483
|
2471 case SN_SYLLABLE:
|
|
|
2472 lp->sl_syllable = read_string(fd, len); /* <syllable> */
|
|
|
2473 if (lp->sl_syllable == NULL)
|
|
|
2474 goto endFAIL;
|
|
|
2475 if (init_syl_tab(lp) == FAIL)
|
|
|
2476 goto endFAIL;
|
|
|
2477 break;
|
|
|
2478
|
|
|
2479 default:
|
|
|
2480 /* Unsupported section. When it's required give an error
|
|
|
2481 * message. When it's not required skip the contents. */
|
|
|
2482 if (c & SNF_REQUIRED)
|
|
|
2483 {
|
|
|
2484 EMSG(_("E770: Unsupported section in spell file"));
|
|
|
2485 goto endFAIL;
|
|
|
2486 }
|
|
|
2487 while (--len >= 0)
|
|
|
2488 if (getc(fd) < 0)
|
|
|
2489 goto truncerr;
|
|
|
2490 break;
|
|
|
2491 }
|
|
|
2492 if (res == SP_FORMERROR)
|
|
|
2493 {
|
|
|
2494 formerr:
|
|
|
2495 EMSG(_(e_format));
|
|
381
|
2496 goto endFAIL;
|
|
344
|
2497 }
|
|
483
|
2498 if (res == SP_TRUNCERROR)
|
|
|
2499 {
|
|
|
2500 truncerr:
|
|
|
2501 EMSG(_(e_spell_trunc));
|
|
375
|
2502 goto endFAIL;
|
|
483
|
2503 }
|
|
495
|
2504 if (res == SP_OTHERERROR)
|
|
483
|
2505 goto endFAIL;
|
|
389
|
2506 }
|
|
236
|
2507
|
|
300
|
2508 /* round 1: <LWORDTREE>
|
|
339
|
2509 * round 2: <KWORDTREE>
|
|
|
2510 * round 3: <PREFIXTREE> */
|
|
|
2511 for (round = 1; round <= 3; ++round)
|
|
236
|
2512 {
|
|
300
|
2513 /* The tree size was computed when writing the file, so that we can
|
|
|
2514 * allocate it as one long block. <nodecount> */
|
|
|
2515 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2516 if (len < 0)
|
|
|
2517 goto truncerr;
|
|
|
2518 if (len > 0)
|
|
236
|
2519 {
|
|
300
|
2520 /* Allocate the byte array. */
|
|
339
|
2521 bp = lalloc((long_u)len, TRUE);
|
|
|
2522 if (bp == NULL)
|
|
300
|
2523 goto endFAIL;
|
|
|
2524 if (round == 1)
|
|
339
|
2525 lp->sl_fbyts = bp;
|
|
|
2526 else if (round == 2)
|
|
|
2527 lp->sl_kbyts = bp;
|
|
300
|
2528 else
|
|
339
|
2529 lp->sl_pbyts = bp;
|
|
236
|
2530
|
|
300
|
2531 /* Allocate the index array. */
|
|
339
|
2532 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2533 if (ip == NULL)
|
|
300
|
2534 goto endFAIL;
|
|
|
2535 if (round == 1)
|
|
339
|
2536 lp->sl_fidxs = ip;
|
|
|
2537 else if (round == 2)
|
|
|
2538 lp->sl_kidxs = ip;
|
|
300
|
2539 else
|
|
339
|
2540 lp->sl_pidxs = ip;
|
|
300
|
2541
|
|
|
2542 /* Read the tree and store it in the array. */
|
|
339
|
2543 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2544 if (idx == -1)
|
|
300
|
2545 goto truncerr;
|
|
324
|
2546 if (idx < 0)
|
|
236
|
2547 goto formerr;
|
|
|
2548 }
|
|
300
|
2549 }
|
|
243
|
2550
|
|
310
|
2551 /* For a new file link it in the list of spell files. */
|
|
|
2552 if (old_lp == NULL)
|
|
|
2553 {
|
|
|
2554 lp->sl_next = first_lang;
|
|
|
2555 first_lang = lp;
|
|
|
2556 }
|
|
307
|
2557
|
|
255
|
2558 goto endOK;
|
|
|
2559
|
|
|
2560 endFAIL:
|
|
310
|
2561 if (lang != NULL)
|
|
|
2562 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2563 *lang = NUL;
|
|
|
2564 if (lp != NULL && old_lp == NULL)
|
|
307
|
2565 slang_free(lp);
|
|
500
|
2566 lp = NULL;
|
|
255
|
2567
|
|
|
2568 endOK:
|
|
236
|
2569 if (fd != NULL)
|
|
|
2570 fclose(fd);
|
|
|
2571 sourcing_name = save_sourcing_name;
|
|
|
2572 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2573
|
|
|
2574 return lp;
|
|
236
|
2575 }
|
|
|
2576
|
|
376
|
2577 /*
|
|
|
2578 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2579 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2580 * Returns NULL when the count is zero.
|
|
483
|
2581 * Sets "*cntp" to SP_*ERROR when there is an error, length of the result
|
|
|
2582 * otherwise.
|
|
376
|
2583 */
|
|
|
2584 static char_u *
|
|
389
|
2585 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2586 FILE *fd;
|
|
|
2587 int cnt_bytes;
|
|
389
|
2588 int *cntp;
|
|
376
|
2589 {
|
|
|
2590 int cnt = 0;
|
|
|
2591 int i;
|
|
|
2592 char_u *str;
|
|
|
2593
|
|
|
2594 /* read the length bytes, MSB first */
|
|
|
2595 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2596 cnt = (cnt << 8) + getc(fd);
|
|
|
2597 if (cnt < 0)
|
|
|
2598 {
|
|
483
|
2599 *cntp = SP_TRUNCERROR;
|
|
376
|
2600 return NULL;
|
|
|
2601 }
|
|
389
|
2602 *cntp = cnt;
|
|
|
2603 if (cnt == 0)
|
|
|
2604 return NULL; /* nothing to read, return NULL */
|
|
376
|
2605
|
|
483
|
2606 str = read_string(fd, cnt);
|
|
|
2607 if (str == NULL)
|
|
495
|
2608 *cntp = SP_OTHERERROR;
|
|
483
|
2609 return str;
|
|
|
2610 }
|
|
|
2611
|
|
|
2612 /*
|
|
|
2613 * Read a string of length "cnt" from "fd" into allocated memory.
|
|
|
2614 * Returns NULL when out of memory.
|
|
|
2615 */
|
|
|
2616 static char_u *
|
|
|
2617 read_string(fd, cnt)
|
|
|
2618 FILE *fd;
|
|
|
2619 int cnt;
|
|
|
2620 {
|
|
|
2621 char_u *str;
|
|
|
2622 int i;
|
|
|
2623
|
|
376
|
2624 /* allocate memory */
|
|
|
2625 str = alloc((unsigned)cnt + 1);
|
|
483
|
2626 if (str != NULL)
|
|
|
2627 {
|
|
|
2628 /* Read the string. Doesn't check for truncated file. */
|
|
|
2629 for (i = 0; i < cnt; ++i)
|
|
|
2630 str[i] = getc(fd);
|
|
|
2631 str[i] = NUL;
|
|
|
2632 }
|
|
|
2633 return str;
|
|
|
2634 }
|
|
|
2635
|
|
|
2636 /*
|
|
|
2637 * Read SN_REGION: <regionname> ...
|
|
|
2638 * Return SP_*ERROR flags.
|
|
|
2639 */
|
|
|
2640 static int
|
|
|
2641 read_region_section(fd, lp, len)
|
|
|
2642 FILE *fd;
|
|
|
2643 slang_T *lp;
|
|
|
2644 int len;
|
|
|
2645 {
|
|
|
2646 int i;
|
|
|
2647
|
|
|
2648 if (len > 16)
|
|
|
2649 return SP_FORMERROR;
|
|
|
2650 for (i = 0; i < len; ++i)
|
|
|
2651 lp->sl_regions[i] = getc(fd); /* <regionname> */
|
|
|
2652 lp->sl_regions[len] = NUL;
|
|
|
2653 return 0;
|
|
|
2654 }
|
|
|
2655
|
|
|
2656 /*
|
|
|
2657 * Read SN_CHARFLAGS section: <charflagslen> <charflags>
|
|
|
2658 * <folcharslen> <folchars>
|
|
|
2659 * Return SP_*ERROR flags.
|
|
|
2660 */
|
|
|
2661 static int
|
|
|
2662 read_charflags_section(fd)
|
|
|
2663 FILE *fd;
|
|
|
2664 {
|
|
|
2665 char_u *flags;
|
|
|
2666 char_u *fol;
|
|
|
2667 int flagslen, follen;
|
|
|
2668
|
|
|
2669 /* <charflagslen> <charflags> */
|
|
|
2670 flags = read_cnt_string(fd, 1, &flagslen);
|
|
|
2671 if (flagslen < 0)
|
|
|
2672 return flagslen;
|
|
|
2673
|
|
|
2674 /* <folcharslen> <folchars> */
|
|
|
2675 fol = read_cnt_string(fd, 2, &follen);
|
|
|
2676 if (follen < 0)
|
|
|
2677 {
|
|
|
2678 vim_free(flags);
|
|
|
2679 return follen;
|
|
|
2680 }
|
|
|
2681
|
|
|
2682 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
2683 if (flags != NULL && fol != NULL)
|
|
|
2684 set_spell_charflags(flags, flagslen, fol);
|
|
|
2685
|
|
|
2686 vim_free(flags);
|
|
|
2687 vim_free(fol);
|
|
|
2688
|
|
|
2689 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
2690 if ((flags == NULL) != (fol == NULL))
|
|
|
2691 return SP_FORMERROR;
|
|
|
2692 return 0;
|
|
|
2693 }
|
|
|
2694
|
|
|
2695 /*
|
|
|
2696 * Read SN_PREFCOND section.
|
|
|
2697 * Return SP_*ERROR flags.
|
|
|
2698 */
|
|
|
2699 static int
|
|
|
2700 read_prefcond_section(fd, lp)
|
|
|
2701 FILE *fd;
|
|
|
2702 slang_T *lp;
|
|
|
2703 {
|
|
|
2704 int cnt;
|
|
|
2705 int i;
|
|
|
2706 int n;
|
|
|
2707 char_u *p;
|
|
|
2708 char_u buf[MAXWLEN + 1];
|
|
|
2709
|
|
|
2710 /* <prefcondcnt> <prefcond> ... */
|
|
|
2711 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
2712 if (cnt <= 0)
|
|
|
2713 return SP_FORMERROR;
|
|
|
2714
|
|
|
2715 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
2716 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
2717 if (lp->sl_prefprog == NULL)
|
|
495
|
2718 return SP_OTHERERROR;
|
|
483
|
2719 lp->sl_prefixcnt = cnt;
|
|
|
2720
|
|
376
|
2721 for (i = 0; i < cnt; ++i)
|
|
483
|
2722 {
|
|
|
2723 /* <prefcond> : <condlen> <condstr> */
|
|
|
2724 n = getc(fd); /* <condlen> */
|
|
|
2725 if (n < 0 || n >= MAXWLEN)
|
|
|
2726 return SP_FORMERROR;
|
|
|
2727
|
|
|
2728 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
2729 * compile the regexp program used to check for the condition. */
|
|
|
2730 if (n > 0)
|
|
|
2731 {
|
|
|
2732 buf[0] = '^'; /* always match at one position only */
|
|
|
2733 p = buf + 1;
|
|
|
2734 while (n-- > 0)
|
|
|
2735 *p++ = getc(fd); /* <condstr> */
|
|
|
2736 *p = NUL;
|
|
|
2737 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
2738 }
|
|
|
2739 }
|
|
|
2740 return 0;
|
|
|
2741 }
|
|
|
2742
|
|
|
2743 /*
|
|
|
2744 * Read REP items section from "fd": <repcount> <rep> ...
|
|
|
2745 * Return SP_*ERROR flags.
|
|
|
2746 */
|
|
|
2747 static int
|
|
|
2748 read_rep_section(fd, slang)
|
|
|
2749 FILE *fd;
|
|
|
2750 slang_T *slang;
|
|
|
2751 {
|
|
|
2752 int cnt;
|
|
|
2753 garray_T *gap;
|
|
|
2754 fromto_T *ftp;
|
|
|
2755 short *first;
|
|
|
2756 int i;
|
|
|
2757
|
|
|
2758 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
2759 if (cnt < 0)
|
|
|
2760 return SP_TRUNCERROR;
|
|
|
2761
|
|
|
2762 gap = &slang->sl_rep;
|
|
|
2763 if (ga_grow(gap, cnt) == FAIL)
|
|
495
|
2764 return SP_OTHERERROR;
|
|
483
|
2765
|
|
|
2766 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
2767 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2768 {
|
|
|
2769 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
|
2770 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
|
2771 if (i < 0)
|
|
|
2772 return i;
|
|
|
2773 if (i == 0)
|
|
|
2774 return SP_FORMERROR;
|
|
|
2775 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
|
2776 if (i <= 0)
|
|
|
2777 {
|
|
|
2778 vim_free(ftp->ft_from);
|
|
|
2779 if (i < 0)
|
|
|
2780 return i;
|
|
|
2781 return SP_FORMERROR;
|
|
|
2782 }
|
|
|
2783 }
|
|
|
2784
|
|
|
2785 /* Fill the first-index table. */
|
|
|
2786 first = slang->sl_rep_first;
|
|
|
2787 for (i = 0; i < 256; ++i)
|
|
|
2788 first[i] = -1;
|
|
|
2789 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2790 {
|
|
|
2791 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
2792 if (first[*ftp->ft_from] == -1)
|
|
|
2793 first[*ftp->ft_from] = i;
|
|
|
2794 }
|
|
|
2795 return 0;
|
|
|
2796 }
|
|
|
2797
|
|
|
2798 /*
|
|
|
2799 * Read SN_SAL section: <salflags> <salcount> <sal> ...
|
|
|
2800 * Return SP_*ERROR flags.
|
|
|
2801 */
|
|
|
2802 static int
|
|
|
2803 read_sal_section(fd, slang)
|
|
|
2804 FILE *fd;
|
|
|
2805 slang_T *slang;
|
|
|
2806 {
|
|
|
2807 int i;
|
|
|
2808 int cnt;
|
|
|
2809 garray_T *gap;
|
|
|
2810 salitem_T *smp;
|
|
|
2811 int ccnt;
|
|
|
2812 char_u *p;
|
|
485
|
2813 int c = NUL;
|
|
483
|
2814
|
|
|
2815 slang->sl_sofo = FALSE;
|
|
|
2816
|
|
|
2817 i = getc(fd); /* <salflags> */
|
|
|
2818 if (i & SAL_F0LLOWUP)
|
|
|
2819 slang->sl_followup = TRUE;
|
|
|
2820 if (i & SAL_COLLAPSE)
|
|
|
2821 slang->sl_collapse = TRUE;
|
|
|
2822 if (i & SAL_REM_ACCENTS)
|
|
|
2823 slang->sl_rem_accents = TRUE;
|
|
|
2824
|
|
|
2825 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2826 if (cnt < 0)
|
|
|
2827 return SP_TRUNCERROR;
|
|
|
2828
|
|
|
2829 gap = &slang->sl_sal;
|
|
|
2830 ga_init2(gap, sizeof(salitem_T), 10);
|
|
|
2831 if (ga_grow(gap, cnt) == FAIL)
|
|
495
|
2832 return SP_OTHERERROR;
|
|
483
|
2833
|
|
|
2834 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2835 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2836 {
|
|
|
2837 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2838 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2839 if (ccnt < 0)
|
|
|
2840 return SP_TRUNCERROR;
|
|
|
2841 if ((p = alloc(ccnt + 2)) == NULL)
|
|
495
|
2842 return SP_OTHERERROR;
|
|
483
|
2843 smp->sm_lead = p;
|
|
|
2844
|
|
|
2845 /* Read up to the first special char into sm_lead. */
|
|
|
2846 for (i = 0; i < ccnt; ++i)
|
|
|
2847 {
|
|
|
2848 c = getc(fd); /* <salfrom> */
|
|
|
2849 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2850 break;
|
|
|
2851 *p++ = c;
|
|
|
2852 }
|
|
|
2853 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2854 *p++ = NUL;
|
|
|
2855
|
|
|
2856 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2857 if (c == '(')
|
|
|
2858 {
|
|
|
2859 smp->sm_oneof = p;
|
|
|
2860 for (++i; i < ccnt; ++i)
|
|
|
2861 {
|
|
|
2862 c = getc(fd); /* <salfrom> */
|
|
|
2863 if (c == ')')
|
|
|
2864 break;
|
|
|
2865 *p++ = c;
|
|
|
2866 }
|
|
|
2867 *p++ = NUL;
|
|
|
2868 if (++i < ccnt)
|
|
|
2869 c = getc(fd);
|
|
|
2870 }
|
|
|
2871 else
|
|
|
2872 smp->sm_oneof = NULL;
|
|
|
2873
|
|
|
2874 /* Any following chars go in sm_rules. */
|
|
|
2875 smp->sm_rules = p;
|
|
|
2876 if (i < ccnt)
|
|
|
2877 /* store the char we got while checking for end of sm_lead */
|
|
|
2878 *p++ = c;
|
|
|
2879 for (++i; i < ccnt; ++i)
|
|
|
2880 *p++ = getc(fd); /* <salfrom> */
|
|
|
2881 *p++ = NUL;
|
|
|
2882
|
|
|
2883 /* <saltolen> <salto> */
|
|
|
2884 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
|
2885 if (ccnt < 0)
|
|
|
2886 {
|
|
|
2887 vim_free(smp->sm_lead);
|
|
|
2888 return ccnt;
|
|
|
2889 }
|
|
|
2890
|
|
|
2891 #ifdef FEAT_MBYTE
|
|
|
2892 if (has_mbyte)
|
|
|
2893 {
|
|
|
2894 /* convert the multi-byte strings to wide char strings */
|
|
|
2895 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2896 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2897 if (smp->sm_oneof == NULL)
|
|
|
2898 smp->sm_oneof_w = NULL;
|
|
|
2899 else
|
|
|
2900 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
2901 if (smp->sm_to == NULL)
|
|
|
2902 smp->sm_to_w = NULL;
|
|
|
2903 else
|
|
|
2904 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
2905 if (smp->sm_lead_w == NULL
|
|
|
2906 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
2907 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
|
2908 {
|
|
|
2909 vim_free(smp->sm_lead);
|
|
|
2910 vim_free(smp->sm_to);
|
|
|
2911 vim_free(smp->sm_lead_w);
|
|
|
2912 vim_free(smp->sm_oneof_w);
|
|
|
2913 vim_free(smp->sm_to_w);
|
|
495
|
2914 return SP_OTHERERROR;
|
|
483
|
2915 }
|
|
|
2916 }
|
|
|
2917 #endif
|
|
|
2918 }
|
|
|
2919
|
|
|
2920 /* Fill the first-index table. */
|
|
|
2921 set_sal_first(slang);
|
|
|
2922
|
|
|
2923 return 0;
|
|
|
2924 }
|
|
|
2925
|
|
|
2926 /*
|
|
|
2927 * SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
2928 * Return SP_*ERROR flags.
|
|
|
2929 */
|
|
|
2930 static int
|
|
|
2931 read_sofo_section(fd, slang)
|
|
|
2932 FILE *fd;
|
|
|
2933 slang_T *slang;
|
|
|
2934 {
|
|
|
2935 int cnt;
|
|
|
2936 char_u *from, *to;
|
|
|
2937 int res;
|
|
|
2938
|
|
|
2939 slang->sl_sofo = TRUE;
|
|
|
2940
|
|
|
2941 /* <sofofromlen> <sofofrom> */
|
|
|
2942 from = read_cnt_string(fd, 2, &cnt);
|
|
|
2943 if (cnt < 0)
|
|
|
2944 return cnt;
|
|
|
2945
|
|
|
2946 /* <sofotolen> <sofoto> */
|
|
|
2947 to = read_cnt_string(fd, 2, &cnt);
|
|
|
2948 if (cnt < 0)
|
|
|
2949 {
|
|
|
2950 vim_free(from);
|
|
|
2951 return cnt;
|
|
|
2952 }
|
|
|
2953
|
|
|
2954 /* Store the info in slang->sl_sal and/or slang->sl_sal_first. */
|
|
|
2955 if (from != NULL && to != NULL)
|
|
|
2956 res = set_sofo(slang, from, to);
|
|
|
2957 else if (from != NULL || to != NULL)
|
|
|
2958 res = SP_FORMERROR; /* only one of two strings is an error */
|
|
|
2959 else
|
|
|
2960 res = 0;
|
|
|
2961
|
|
|
2962 vim_free(from);
|
|
|
2963 vim_free(to);
|
|
|
2964 return res;
|
|
|
2965 }
|
|
|
2966
|
|
|
2967 /*
|
|
|
2968 * Read the compound section from the .spl file:
|
|
|
2969 * <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
2970 * Returns SP_*ERROR flags.
|
|
|
2971 */
|
|
|
2972 static int
|
|
|
2973 read_compound(fd, slang, len)
|
|
|
2974 FILE *fd;
|
|
|
2975 slang_T *slang;
|
|
|
2976 int len;
|
|
|
2977 {
|
|
|
2978 int todo = len;
|
|
|
2979 int c;
|
|
|
2980 int atstart;
|
|
|
2981 char_u *pat;
|
|
|
2982 char_u *pp;
|
|
|
2983 char_u *cp;
|
|
485
|
2984 char_u *ap;
|
|
483
|
2985
|
|
|
2986 if (todo < 2)
|
|
|
2987 return SP_FORMERROR; /* need at least two bytes */
|
|
|
2988
|
|
|
2989 --todo;
|
|
|
2990 c = getc(fd); /* <compmax> */
|
|
|
2991 if (c < 2)
|
|
|
2992 c = MAXWLEN;
|
|
|
2993 slang->sl_compmax = c;
|
|
|
2994
|
|
|
2995 --todo;
|
|
|
2996 c = getc(fd); /* <compminlen> */
|
|
|
2997 if (c < 1)
|
|
501
|
2998 c = 0;
|
|
483
|
2999 slang->sl_compminlen = c;
|
|
|
3000
|
|
|
3001 --todo;
|
|
|
3002 c = getc(fd); /* <compsylmax> */
|
|
|
3003 if (c < 1)
|
|
|
3004 c = MAXWLEN;
|
|
|
3005 slang->sl_compsylmax = c;
|
|
|
3006
|
|
|
3007 /* Turn the COMPOUNDFLAGS items into a regexp pattern:
|
|
|
3008 * "a[bc]/a*b+" -> "^\(a[bc]\|a*b\+\)$".
|
|
495
|
3009 * Inserting backslashes may double the length, "^\(\)$<Nul>" is 7 bytes.
|
|
|
3010 * Conversion to utf-8 may double the size. */
|
|
|
3011 c = todo * 2 + 7;
|
|
|
3012 #ifdef FEAT_MBYTE
|
|
|
3013 if (enc_utf8)
|
|
|
3014 c += todo * 2;
|
|
|
3015 #endif
|
|
|
3016 pat = alloc((unsigned)c);
|
|
483
|
3017 if (pat == NULL)
|
|
495
|
3018 return SP_OTHERERROR;
|
|
483
|
3019
|
|
485
|
3020 /* We also need a list of all flags that can appear at the start and one
|
|
|
3021 * for all flags. */
|
|
483
|
3022 cp = alloc(todo + 1);
|
|
|
3023 if (cp == NULL)
|
|
|
3024 {
|
|
|
3025 vim_free(pat);
|
|
495
|
3026 return SP_OTHERERROR;
|
|
483
|
3027 }
|
|
|
3028 slang->sl_compstartflags = cp;
|
|
|
3029 *cp = NUL;
|
|
|
3030
|
|
485
|
3031 ap = alloc(todo + 1);
|
|
|
3032 if (ap == NULL)
|
|
|
3033 {
|
|
|
3034 vim_free(pat);
|
|
495
|
3035 return SP_OTHERERROR;
|
|
485
|
3036 }
|
|
|
3037 slang->sl_compallflags = ap;
|
|
|
3038 *ap = NUL;
|
|
|
3039
|
|
483
|
3040 pp = pat;
|
|
|
3041 *pp++ = '^';
|
|
|
3042 *pp++ = '\\';
|
|
|
3043 *pp++ = '(';
|
|
|
3044
|
|
|
3045 atstart = 1;
|
|
|
3046 while (todo-- > 0)
|
|
|
3047 {
|
|
|
3048 c = getc(fd); /* <compflags> */
|
|
485
|
3049
|
|
|
3050 /* Add all flags to "sl_compallflags". */
|
|
|
3051 if (vim_strchr((char_u *)"+*[]/", c) == NULL
|
|
495
|
3052 && !byte_in_str(slang->sl_compallflags, c))
|
|
485
|
3053 {
|
|
|
3054 *ap++ = c;
|
|
|
3055 *ap = NUL;
|
|
|
3056 }
|
|
|
3057
|
|
483
|
3058 if (atstart != 0)
|
|
|
3059 {
|
|
|
3060 /* At start of item: copy flags to "sl_compstartflags". For a
|
|
|
3061 * [abc] item set "atstart" to 2 and copy up to the ']'. */
|
|
|
3062 if (c == '[')
|
|
|
3063 atstart = 2;
|
|
|
3064 else if (c == ']')
|
|
|
3065 atstart = 0;
|
|
|
3066 else
|
|
|
3067 {
|
|
495
|
3068 if (!byte_in_str(slang->sl_compstartflags, c))
|
|
483
|
3069 {
|
|
|
3070 *cp++ = c;
|
|
|
3071 *cp = NUL;
|
|
|
3072 }
|
|
|
3073 if (atstart == 1)
|
|
|
3074 atstart = 0;
|
|
|
3075 }
|
|
|
3076 }
|
|
|
3077 if (c == '/') /* slash separates two items */
|
|
|
3078 {
|
|
|
3079 *pp++ = '\\';
|
|
|
3080 *pp++ = '|';
|
|
|
3081 atstart = 1;
|
|
|
3082 }
|
|
|
3083 else /* normal char, "[abc]" and '*' are copied as-is */
|
|
|
3084 {
|
|
500
|
3085 if (c == '+' || c == '~')
|
|
483
|
3086 *pp++ = '\\'; /* "a+" becomes "a\+" */
|
|
495
|
3087 #ifdef FEAT_MBYTE
|
|
|
3088 if (enc_utf8)
|
|
|
3089 pp += mb_char2bytes(c, pp);
|
|
|
3090 else
|
|
|
3091 #endif
|
|
|
3092 *pp++ = c;
|
|
483
|
3093 }
|
|
|
3094 }
|
|
|
3095
|
|
|
3096 *pp++ = '\\';
|
|
|
3097 *pp++ = ')';
|
|
|
3098 *pp++ = '$';
|
|
|
3099 *pp = NUL;
|
|
|
3100
|
|
|
3101 slang->sl_compprog = vim_regcomp(pat, RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
3102 vim_free(pat);
|
|
|
3103 if (slang->sl_compprog == NULL)
|
|
|
3104 return SP_FORMERROR;
|
|
|
3105
|
|
|
3106 return 0;
|
|
|
3107 }
|
|
|
3108
|
|
495
|
3109 /*
|
|
498
|
3110 * Return TRUE if byte "n" appears in "str".
|
|
495
|
3111 * Like strchr() but independent of locale.
|
|
|
3112 */
|
|
|
3113 static int
|
|
498
|
3114 byte_in_str(str, n)
|
|
495
|
3115 char_u *str;
|
|
498
|
3116 int n;
|
|
495
|
3117 {
|
|
|
3118 char_u *p;
|
|
|
3119
|
|
|
3120 for (p = str; *p != NUL; ++p)
|
|
498
|
3121 if (*p == n)
|
|
495
|
3122 return TRUE;
|
|
|
3123 return FALSE;
|
|
|
3124 }
|
|
|
3125
|
|
483
|
3126 #define SY_MAXLEN 30
|
|
|
3127 typedef struct syl_item_S
|
|
|
3128 {
|
|
|
3129 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */
|
|
|
3130 int sy_len;
|
|
|
3131 } syl_item_T;
|
|
|
3132
|
|
|
3133 /*
|
|
|
3134 * Truncate "slang->sl_syllable" at the first slash and put the following items
|
|
|
3135 * in "slang->sl_syl_items".
|
|
|
3136 */
|
|
|
3137 static int
|
|
|
3138 init_syl_tab(slang)
|
|
|
3139 slang_T *slang;
|
|
|
3140 {
|
|
|
3141 char_u *p;
|
|
|
3142 char_u *s;
|
|
|
3143 int l;
|
|
|
3144 syl_item_T *syl;
|
|
|
3145
|
|
|
3146 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4);
|
|
|
3147 p = vim_strchr(slang->sl_syllable, '/');
|
|
|
3148 while (p != NULL)
|
|
|
3149 {
|
|
|
3150 *p++ = NUL;
|
|
495
|
3151 if (*p == NUL) /* trailing slash */
|
|
483
|
3152 break;
|
|
|
3153 s = p;
|
|
|
3154 p = vim_strchr(p, '/');
|
|
|
3155 if (p == NULL)
|
|
|
3156 l = STRLEN(s);
|
|
|
3157 else
|
|
|
3158 l = p - s;
|
|
|
3159 if (l >= SY_MAXLEN)
|
|
|
3160 return SP_FORMERROR;
|
|
|
3161 if (ga_grow(&slang->sl_syl_items, 1) == FAIL)
|
|
495
|
3162 return SP_OTHERERROR;
|
|
483
|
3163 syl = ((syl_item_T *)slang->sl_syl_items.ga_data)
|
|
|
3164 + slang->sl_syl_items.ga_len++;
|
|
|
3165 vim_strncpy(syl->sy_chars, s, l);
|
|
|
3166 syl->sy_len = l;
|
|
|
3167 }
|
|
|
3168 return OK;
|
|
|
3169 }
|
|
|
3170
|
|
|
3171 /*
|
|
|
3172 * Count the number of syllables in "word".
|
|
|
3173 * When "word" contains spaces the syllables after the last space are counted.
|
|
|
3174 * Returns zero if syllables are not defines.
|
|
|
3175 */
|
|
|
3176 static int
|
|
|
3177 count_syllables(slang, word)
|
|
|
3178 slang_T *slang;
|
|
|
3179 char_u *word;
|
|
|
3180 {
|
|
|
3181 int cnt = 0;
|
|
|
3182 int skip = FALSE;
|
|
|
3183 char_u *p;
|
|
|
3184 int len;
|
|
|
3185 int i;
|
|
|
3186 syl_item_T *syl;
|
|
|
3187 int c;
|
|
|
3188
|
|
|
3189 if (slang->sl_syllable == NULL)
|
|
|
3190 return 0;
|
|
|
3191
|
|
|
3192 for (p = word; *p != NUL; p += len)
|
|
|
3193 {
|
|
|
3194 /* When running into a space reset counter. */
|
|
|
3195 if (*p == ' ')
|
|
|
3196 {
|
|
|
3197 len = 1;
|
|
|
3198 cnt = 0;
|
|
|
3199 continue;
|
|
|
3200 }
|
|
|
3201
|
|
|
3202 /* Find longest match of syllable items. */
|
|
|
3203 len = 0;
|
|
|
3204 for (i = 0; i < slang->sl_syl_items.ga_len; ++i)
|
|
|
3205 {
|
|
|
3206 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i;
|
|
|
3207 if (syl->sy_len > len
|
|
|
3208 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0)
|
|
|
3209 len = syl->sy_len;
|
|
|
3210 }
|
|
|
3211 if (len != 0) /* found a match, count syllable */
|
|
|
3212 {
|
|
|
3213 ++cnt;
|
|
|
3214 skip = FALSE;
|
|
|
3215 }
|
|
|
3216 else
|
|
|
3217 {
|
|
|
3218 /* No recognized syllable item, at least a syllable char then? */
|
|
|
3219 #ifdef FEAT_MBYTE
|
|
|
3220 c = mb_ptr2char(p);
|
|
|
3221 len = (*mb_ptr2len)(p);
|
|
|
3222 #else
|
|
|
3223 c = *p;
|
|
|
3224 len = 1;
|
|
|
3225 #endif
|
|
|
3226 if (vim_strchr(slang->sl_syllable, c) == NULL)
|
|
|
3227 skip = FALSE; /* No, search for next syllable */
|
|
|
3228 else if (!skip)
|
|
|
3229 {
|
|
|
3230 ++cnt; /* Yes, count it */
|
|
|
3231 skip = TRUE; /* don't count following syllable chars */
|
|
|
3232 }
|
|
|
3233 }
|
|
|
3234 }
|
|
|
3235 return cnt;
|
|
376
|
3236 }
|
|
|
3237
|
|
381
|
3238 /*
|
|
|
3239 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
483
|
3240 * Returns SP_*ERROR flags when there is something wrong.
|
|
381
|
3241 */
|
|
|
3242 static int
|
|
|
3243 set_sofo(lp, from, to)
|
|
|
3244 slang_T *lp;
|
|
|
3245 char_u *from;
|
|
|
3246 char_u *to;
|
|
|
3247 {
|
|
|
3248 int i;
|
|
|
3249
|
|
|
3250 #ifdef FEAT_MBYTE
|
|
|
3251 garray_T *gap;
|
|
|
3252 char_u *s;
|
|
|
3253 char_u *p;
|
|
|
3254 int c;
|
|
|
3255 int *inp;
|
|
|
3256
|
|
|
3257 if (has_mbyte)
|
|
|
3258 {
|
|
|
3259 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
3260 * characters. The index is the low byte of the character.
|
|
|
3261 * The list contains from-to pairs with a terminating NUL.
|
|
|
3262 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
3263 gap = &lp->sl_sal;
|
|
|
3264 ga_init2(gap, sizeof(int *), 1);
|
|
|
3265 if (ga_grow(gap, 256) == FAIL)
|
|
495
|
3266 return SP_OTHERERROR;
|
|
381
|
3267 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
3268 gap->ga_len = 256;
|
|
|
3269
|
|
|
3270 /* First count the number of items for each list. Temporarily use
|
|
|
3271 * sl_sal_first[] for this. */
|
|
|
3272 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3273 {
|
|
474
|
3274 c = mb_cptr2char_adv(&p);
|
|
|
3275 mb_cptr_adv(s);
|
|
381
|
3276 if (c >= 256)
|
|
|
3277 ++lp->sl_sal_first[c & 0xff];
|
|
|
3278 }
|
|
|
3279 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
483
|
3280 return SP_FORMERROR;
|
|
381
|
3281
|
|
|
3282 /* Allocate the lists. */
|
|
|
3283 for (i = 0; i < 256; ++i)
|
|
|
3284 if (lp->sl_sal_first[i] > 0)
|
|
|
3285 {
|
|
|
3286 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
3287 if (p == NULL)
|
|
495
|
3288 return SP_OTHERERROR;
|
|
381
|
3289 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
3290 *(int *)p = 0;
|
|
|
3291 }
|
|
|
3292
|
|
|
3293 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
3294 * list. */
|
|
|
3295 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
3296 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3297 {
|
|
474
|
3298 c = mb_cptr2char_adv(&p);
|
|
|
3299 i = mb_cptr2char_adv(&s);
|
|
381
|
3300 if (c >= 256)
|
|
|
3301 {
|
|
|
3302 /* Append the from-to chars at the end of the list with
|
|
|
3303 * the low byte. */
|
|
|
3304 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
3305 while (*inp != 0)
|
|
|
3306 ++inp;
|
|
|
3307 *inp++ = c; /* from char */
|
|
|
3308 *inp++ = i; /* to char */
|
|
|
3309 *inp++ = NUL; /* NUL at the end */
|
|
|
3310 }
|
|
|
3311 else
|
|
|
3312 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
3313 lp->sl_sal_first[c] = i;
|
|
|
3314 }
|
|
|
3315 }
|
|
|
3316 else
|
|
|
3317 #endif
|
|
|
3318 {
|
|
|
3319 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
3320 if (STRLEN(from) != STRLEN(to))
|
|
483
|
3321 return SP_FORMERROR;
|
|
381
|
3322
|
|
|
3323 for (i = 0; to[i] != NUL; ++i)
|
|
|
3324 lp->sl_sal_first[from[i]] = to[i];
|
|
|
3325 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
3326 }
|
|
|
3327
|
|
483
|
3328 return 0;
|
|
381
|
3329 }
|
|
|
3330
|
|
|
3331 /*
|
|
|
3332 * Fill the first-index table for "lp".
|
|
|
3333 */
|
|
|
3334 static void
|
|
|
3335 set_sal_first(lp)
|
|
|
3336 slang_T *lp;
|
|
|
3337 {
|
|
|
3338 salfirst_T *sfirst;
|
|
|
3339 int i;
|
|
|
3340 salitem_T *smp;
|
|
|
3341 int c;
|
|
|
3342 garray_T *gap = &lp->sl_sal;
|
|
|
3343
|
|
|
3344 sfirst = lp->sl_sal_first;
|
|
|
3345 for (i = 0; i < 256; ++i)
|
|
|
3346 sfirst[i] = -1;
|
|
|
3347 smp = (salitem_T *)gap->ga_data;
|
|
|
3348 for (i = 0; i < gap->ga_len; ++i)
|
|
|
3349 {
|
|
|
3350 #ifdef FEAT_MBYTE
|
|
|
3351 if (has_mbyte)
|
|
|
3352 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
3353 * the character, for other encodings it should differ for most
|
|
|
3354 * characters. */
|
|
|
3355 c = *smp[i].sm_lead_w & 0xff;
|
|
|
3356 else
|
|
|
3357 #endif
|
|
|
3358 c = *smp[i].sm_lead;
|
|
|
3359 if (sfirst[c] == -1)
|
|
|
3360 {
|
|
|
3361 sfirst[c] = i;
|
|
|
3362 #ifdef FEAT_MBYTE
|
|
|
3363 if (has_mbyte)
|
|
|
3364 {
|
|
|
3365 int n;
|
|
|
3366
|
|
|
3367 /* Make sure all entries with this byte are following each
|
|
|
3368 * other. Move the ones that are in the wrong position. Do
|
|
|
3369 * keep the same ordering! */
|
|
|
3370 while (i + 1 < gap->ga_len
|
|
|
3371 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
3372 /* Skip over entry with same index byte. */
|
|
|
3373 ++i;
|
|
|
3374
|
|
|
3375 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
3376 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
3377 {
|
|
|
3378 salitem_T tsal;
|
|
|
3379
|
|
|
3380 /* Move entry with same index byte after the entries
|
|
|
3381 * we already found. */
|
|
|
3382 ++i;
|
|
|
3383 --n;
|
|
|
3384 tsal = smp[i + n];
|
|
|
3385 mch_memmove(smp + i + 1, smp + i,
|
|
|
3386 sizeof(salitem_T) * n);
|
|
|
3387 smp[i] = tsal;
|
|
|
3388 }
|
|
|
3389 }
|
|
|
3390 #endif
|
|
|
3391 }
|
|
|
3392 }
|
|
|
3393 }
|
|
376
|
3394
|
|
372
|
3395 #ifdef FEAT_MBYTE
|
|
|
3396 /*
|
|
|
3397 * Turn a multi-byte string into a wide character string.
|
|
|
3398 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
3399 */
|
|
|
3400 static int *
|
|
|
3401 mb_str2wide(s)
|
|
|
3402 char_u *s;
|
|
|
3403 {
|
|
|
3404 int *res;
|
|
|
3405 char_u *p;
|
|
|
3406 int i = 0;
|
|
|
3407
|
|
|
3408 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
3409 if (res != NULL)
|
|
|
3410 {
|
|
|
3411 for (p = s; *p != NUL; )
|
|
|
3412 res[i++] = mb_ptr2char_adv(&p);
|
|
|
3413 res[i] = NUL;
|
|
|
3414 }
|
|
|
3415 return res;
|
|
|
3416 }
|
|
|
3417 #endif
|
|
|
3418
|
|
236
|
3419 /*
|
|
300
|
3420 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
3421 * "byts" and index array "idxs". Recursively read the children.
|
|
|
3422 *
|
|
346
|
3423 * NOTE: The code here must match put_node().
|
|
300
|
3424 *
|
|
|
3425 * Returns the index follosing the siblings.
|
|
|
3426 * Returns -1 if the file is shorter than expected.
|
|
|
3427 * Returns -2 if there is a format error.
|
|
236
|
3428 */
|
|
324
|
3429 static idx_T
|
|
339
|
3430 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
3431 FILE *fd;
|
|
|
3432 char_u *byts;
|
|
324
|
3433 idx_T *idxs;
|
|
300
|
3434 int maxidx; /* size of arrays */
|
|
324
|
3435 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
3436 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
3437 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
3438 {
|
|
300
|
3439 int len;
|
|
|
3440 int i;
|
|
|
3441 int n;
|
|
324
|
3442 idx_T idx = startidx;
|
|
300
|
3443 int c;
|
|
366
|
3444 int c2;
|
|
300
|
3445 #define SHARED_MASK 0x8000000
|
|
236
|
3446
|
|
300
|
3447 len = getc(fd); /* <siblingcount> */
|
|
|
3448 if (len <= 0)
|
|
|
3449 return -1;
|
|
|
3450
|
|
|
3451 if (startidx + len >= maxidx)
|
|
|
3452 return -2;
|
|
|
3453 byts[idx++] = len;
|
|
|
3454
|
|
|
3455 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
3456 for (i = 1; i <= len; ++i)
|
|
236
|
3457 {
|
|
300
|
3458 c = getc(fd); /* <byte> */
|
|
|
3459 if (c < 0)
|
|
|
3460 return -1;
|
|
|
3461 if (c <= BY_SPECIAL)
|
|
|
3462 {
|
|
366
|
3463 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
3464 {
|
|
|
3465 /* No flags, all regions. */
|
|
|
3466 idxs[idx] = 0;
|
|
|
3467 c = 0;
|
|
|
3468 }
|
|
390
|
3469 else if (c != BY_INDEX)
|
|
300
|
3470 {
|
|
339
|
3471 if (prefixtree)
|
|
|
3472 {
|
|
455
|
3473 /* Read the optional pflags byte, the prefix ID and the
|
|
|
3474 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
3475 * byte, the condition index shifted up 8 bits, the flags
|
|
|
3476 * shifted up 24 bits. */
|
|
|
3477 if (c == BY_FLAGS)
|
|
|
3478 c = getc(fd) << 24; /* <pflags> */
|
|
|
3479 else
|
|
|
3480 c = 0;
|
|
|
3481
|
|
481
|
3482 c |= getc(fd); /* <affixID> */
|
|
455
|
3483
|
|
339
|
3484 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
3485 if (n >= maxprefcondnr)
|
|
|
3486 return -2;
|
|
455
|
3487 c |= (n << 8);
|
|
339
|
3488 }
|
|
390
|
3489 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
3490 {
|
|
|
3491 /* Read flags and optional region and prefix ID. In
|
|
390
|
3492 * idxs[] the flags go in the low two bytes, region above
|
|
|
3493 * that and prefix ID above the region. */
|
|
|
3494 c2 = c;
|
|
339
|
3495 c = getc(fd); /* <flags> */
|
|
390
|
3496 if (c2 == BY_FLAGS2)
|
|
|
3497 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
3498 if (c & WF_REGION)
|
|
390
|
3499 c = (getc(fd) << 16) + c; /* <region> */
|
|
481
|
3500 if (c & WF_AFX)
|
|
|
3501 c = (getc(fd) << 24) + c; /* <affixID> */
|
|
339
|
3502 }
|
|
|
3503
|
|
300
|
3504 idxs[idx] = c;
|
|
|
3505 c = 0;
|
|
|
3506 }
|
|
|
3507 else /* c == BY_INDEX */
|
|
|
3508 {
|
|
|
3509 /* <nodeidx> */
|
|
|
3510 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
3511 if (n < 0 || n >= maxidx)
|
|
|
3512 return -2;
|
|
|
3513 idxs[idx] = n + SHARED_MASK;
|
|
|
3514 c = getc(fd); /* <xbyte> */
|
|
|
3515 }
|
|
|
3516 }
|
|
|
3517 byts[idx++] = c;
|
|
236
|
3518 }
|
|
|
3519
|
|
300
|
3520 /* Recursively read the children for non-shared siblings.
|
|
|
3521 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
3522 * remove SHARED_MASK) */
|
|
|
3523 for (i = 1; i <= len; ++i)
|
|
|
3524 if (byts[startidx + i] != 0)
|
|
|
3525 {
|
|
|
3526 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
3527 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
3528 else
|
|
|
3529 {
|
|
|
3530 idxs[startidx + i] = idx;
|
|
339
|
3531 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
3532 prefixtree, maxprefcondnr);
|
|
300
|
3533 if (idx < 0)
|
|
|
3534 break;
|
|
|
3535 }
|
|
|
3536 }
|
|
236
|
3537
|
|
300
|
3538 return idx;
|
|
236
|
3539 }
|
|
|
3540
|
|
|
3541 /*
|
|
|
3542 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
3543 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
3544 */
|
|
|
3545 char_u *
|
|
|
3546 did_set_spelllang(buf)
|
|
|
3547 buf_T *buf;
|
|
|
3548 {
|
|
|
3549 garray_T ga;
|
|
351
|
3550 char_u *splp;
|
|
236
|
3551 char_u *region;
|
|
416
|
3552 char_u region_cp[3];
|
|
355
|
3553 int filename;
|
|
236
|
3554 int region_mask;
|
|
503
|
3555 slang_T *slang;
|
|
236
|
3556 int c;
|
|
351
|
3557 char_u lang[MAXWLEN + 1];
|
|
323
|
3558 char_u spf_name[MAXPATHL];
|
|
351
|
3559 int len;
|
|
|
3560 char_u *p;
|
|
381
|
3561 int round;
|
|
385
|
3562 char_u *spf;
|
|
389
|
3563 char_u *use_region = NULL;
|
|
|
3564 int dont_use_region = FALSE;
|
|
501
|
3565 int nobreak = FALSE;
|
|
503
|
3566 int i, j;
|
|
|
3567 langp_T *lp, *lp2;
|
|
236
|
3568
|
|
|
3569 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
3570 clear_midword(buf);
|
|
236
|
3571
|
|
351
|
3572 /* loop over comma separated language names. */
|
|
|
3573 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
3574 {
|
|
|
3575 /* Get one language name. */
|
|
|
3576 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
3577
|
|
240
|
3578 region = NULL;
|
|
351
|
3579 len = STRLEN(lang);
|
|
355
|
3580
|
|
|
3581 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
3582 * If there is a region name let "region" point to it and remove it
|
|
|
3583 * from the name. */
|
|
|
3584 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
3585 {
|
|
355
|
3586 filename = TRUE;
|
|
|
3587
|
|
416
|
3588 /* Locate a region and remove it from the file name. */
|
|
|
3589 p = vim_strchr(gettail(lang), '_');
|
|
|
3590 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
3591 && !ASCII_ISALPHA(p[3]))
|
|
|
3592 {
|
|
|
3593 vim_strncpy(region_cp, p + 1, 2);
|
|
|
3594 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
3595 len -= 3;
|
|
|
3596 region = region_cp;
|
|
|
3597 }
|
|
|
3598 else
|
|
|
3599 dont_use_region = TRUE;
|
|
|
3600
|
|
355
|
3601 /* Check if we loaded this language before. */
|
|
503
|
3602 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
3603 if (fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME)
|
|
355
|
3604 break;
|
|
236
|
3605 }
|
|
355
|
3606 else
|
|
|
3607 {
|
|
|
3608 filename = FALSE;
|
|
|
3609 if (len > 3 && lang[len - 3] == '_')
|
|
|
3610 {
|
|
|
3611 region = lang + len - 2;
|
|
|
3612 len -= 3;
|
|
|
3613 lang[len] = NUL;
|
|
389
|
3614 }
|
|
|
3615 else
|
|
|
3616 dont_use_region = TRUE;
|
|
355
|
3617
|
|
|
3618 /* Check if we loaded this language before. */
|
|
503
|
3619 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
3620 if (STRICMP(lang, slang->sl_name) == 0)
|
|
355
|
3621 break;
|
|
|
3622 }
|
|
236
|
3623
|
|
416
|
3624 if (region != NULL)
|
|
|
3625 {
|
|
|
3626 /* If the region differs from what was used before then don't
|
|
|
3627 * use it for 'spellfile'. */
|
|
|
3628 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
3629 dont_use_region = TRUE;
|
|
|
3630 use_region = region;
|
|
|
3631 }
|
|
|
3632
|
|
351
|
3633 /* If not found try loading the language now. */
|
|
503
|
3634 if (slang == NULL)
|
|
355
|
3635 {
|
|
|
3636 if (filename)
|
|
|
3637 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
3638 else
|
|
|
3639 spell_load_lang(lang);
|
|
|
3640 }
|
|
236
|
3641
|
|
307
|
3642 /*
|
|
351
|
3643 * Loop over the languages, there can be several files for "lang".
|
|
307
|
3644 */
|
|
503
|
3645 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
3646 if (filename ? fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME
|
|
|
3647 : STRICMP(lang, slang->sl_name) == 0)
|
|
236
|
3648 {
|
|
316
|
3649 region_mask = REGION_ALL;
|
|
355
|
3650 if (!filename && region != NULL)
|
|
236
|
3651 {
|
|
307
|
3652 /* find region in sl_regions */
|
|
503
|
3653 c = find_region(slang->sl_regions, region);
|
|
307
|
3654 if (c == REGION_ALL)
|
|
|
3655 {
|
|
503
|
3656 if (slang->sl_add)
|
|
389
|
3657 {
|
|
503
|
3658 if (*slang->sl_regions != NUL)
|
|
389
|
3659 /* This addition file is for other regions. */
|
|
|
3660 region_mask = 0;
|
|
|
3661 }
|
|
|
3662 else
|
|
|
3663 /* This is probably an error. Give a warning and
|
|
|
3664 * accept the words anyway. */
|
|
351
|
3665 smsg((char_u *)
|
|
|
3666 _("Warning: region %s not supported"),
|
|
|
3667 region);
|
|
307
|
3668 }
|
|
|
3669 else
|
|
|
3670 region_mask = 1 << c;
|
|
236
|
3671 }
|
|
307
|
3672
|
|
389
|
3673 if (region_mask != 0)
|
|
307
|
3674 {
|
|
389
|
3675 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3676 {
|
|
|
3677 ga_clear(&ga);
|
|
|
3678 return e_outofmem;
|
|
|
3679 }
|
|
503
|
3680 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang;
|
|
389
|
3681 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3682 ++ga.ga_len;
|
|
503
|
3683 use_midword(slang, buf);
|
|
|
3684 if (slang->sl_nobreak)
|
|
501
|
3685 nobreak = TRUE;
|
|
307
|
3686 }
|
|
385
|
3687 }
|
|
|
3688 }
|
|
|
3689
|
|
|
3690 /* round 0: load int_wordlist, if possible.
|
|
|
3691 * round 1: load first name in 'spellfile'.
|
|
|
3692 * round 2: load second name in 'spellfile.
|
|
|
3693 * etc. */
|
|
|
3694 spf = curbuf->b_p_spf;
|
|
|
3695 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
3696 {
|
|
|
3697 if (round == 0)
|
|
|
3698 {
|
|
|
3699 /* Internal wordlist, if there is one. */
|
|
|
3700 if (int_wordlist == NULL)
|
|
381
|
3701 continue;
|
|
385
|
3702 int_wordlist_spl(spf_name);
|
|
381
|
3703 }
|
|
|
3704 else
|
|
|
3705 {
|
|
385
|
3706 /* One entry in 'spellfile'. */
|
|
|
3707 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
3708 STRCAT(spf_name, ".spl");
|
|
|
3709
|
|
|
3710 /* If it was already found above then skip it. */
|
|
|
3711 for (c = 0; c < ga.ga_len; ++c)
|
|
500
|
3712 {
|
|
|
3713 p = LANGP_ENTRY(ga, c)->lp_slang->sl_fname;
|
|
|
3714 if (p != NULL && fullpathcmp(spf_name, p, FALSE) == FPC_SAME)
|
|
385
|
3715 break;
|
|
500
|
3716 }
|
|
385
|
3717 if (c < ga.ga_len)
|
|
381
|
3718 continue;
|
|
|
3719 }
|
|
|
3720
|
|
351
|
3721 /* Check if it was loaded already. */
|
|
503
|
3722 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
3723 if (fullpathcmp(spf_name, slang->sl_fname, FALSE) == FPC_SAME)
|
|
323
|
3724 break;
|
|
503
|
3725 if (slang == NULL)
|
|
323
|
3726 {
|
|
351
|
3727 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
3728 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
3729 * use an arbitrary name. */
|
|
|
3730 if (round == 0)
|
|
|
3731 STRCPY(lang, "internal wordlist");
|
|
|
3732 else
|
|
|
3733 {
|
|
|
3734 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
3735 p = vim_strchr(lang, '.');
|
|
|
3736 if (p != NULL)
|
|
|
3737 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
3738 }
|
|
503
|
3739 slang = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
501
|
3740
|
|
|
3741 /* If one of the languages has NOBREAK we assume the addition
|
|
|
3742 * files also have this. */
|
|
503
|
3743 if (slang != NULL && nobreak)
|
|
|
3744 slang->sl_nobreak = TRUE;
|
|
|
3745 }
|
|
|
3746 if (slang != NULL && ga_grow(&ga, 1) == OK)
|
|
323
|
3747 {
|
|
389
|
3748 region_mask = REGION_ALL;
|
|
|
3749 if (use_region != NULL && !dont_use_region)
|
|
|
3750 {
|
|
|
3751 /* find region in sl_regions */
|
|
503
|
3752 c = find_region(slang->sl_regions, use_region);
|
|
389
|
3753 if (c != REGION_ALL)
|
|
|
3754 region_mask = 1 << c;
|
|
503
|
3755 else if (*slang->sl_regions != NUL)
|
|
389
|
3756 /* This spell file is for other regions. */
|
|
|
3757 region_mask = 0;
|
|
|
3758 }
|
|
|
3759
|
|
|
3760 if (region_mask != 0)
|
|
|
3761 {
|
|
503
|
3762 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang;
|
|
|
3763 LANGP_ENTRY(ga, ga.ga_len)->lp_sallang = NULL;
|
|
|
3764 LANGP_ENTRY(ga, ga.ga_len)->lp_replang = NULL;
|
|
389
|
3765 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3766 ++ga.ga_len;
|
|
503
|
3767 use_midword(slang, buf);
|
|
389
|
3768 }
|
|
323
|
3769 }
|
|
|
3770 }
|
|
|
3771
|
|
236
|
3772 /* Everything is fine, store the new b_langp value. */
|
|
|
3773 ga_clear(&buf->b_langp);
|
|
|
3774 buf->b_langp = ga;
|
|
|
3775
|
|
503
|
3776 /* For each language figure out what language to use for sound folding and
|
|
|
3777 * REP items. If the language doesn't support it itself use another one
|
|
|
3778 * with the same name. E.g. for "en-math" use "en". */
|
|
|
3779 for (i = 0; i < ga.ga_len; ++i)
|
|
|
3780 {
|
|
|
3781 lp = LANGP_ENTRY(ga, i);
|
|
|
3782
|
|
|
3783 /* sound folding */
|
|
|
3784 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
3785 /* language does sound folding itself */
|
|
|
3786 lp->lp_sallang = lp->lp_slang;
|
|
|
3787 else
|
|
|
3788 /* find first similar language that does sound folding */
|
|
|
3789 for (j = 0; j < ga.ga_len; ++j)
|
|
|
3790 {
|
|
|
3791 lp2 = LANGP_ENTRY(ga, j);
|
|
|
3792 if (lp2->lp_slang->sl_sal.ga_len > 0
|
|
|
3793 && STRNCMP(lp->lp_slang->sl_name,
|
|
|
3794 lp2->lp_slang->sl_name, 2) == 0)
|
|
|
3795 {
|
|
|
3796 lp->lp_sallang = lp2->lp_slang;
|
|
|
3797 break;
|
|
|
3798 }
|
|
|
3799 }
|
|
|
3800
|
|
|
3801 /* REP items */
|
|
|
3802 if (lp->lp_slang->sl_rep.ga_len > 0)
|
|
|
3803 /* language has REP items itself */
|
|
|
3804 lp->lp_replang = lp->lp_slang;
|
|
|
3805 else
|
|
|
3806 /* find first similar language that does sound folding */
|
|
|
3807 for (j = 0; j < ga.ga_len; ++j)
|
|
|
3808 {
|
|
|
3809 lp2 = LANGP_ENTRY(ga, j);
|
|
|
3810 if (lp2->lp_slang->sl_rep.ga_len > 0
|
|
|
3811 && STRNCMP(lp->lp_slang->sl_name,
|
|
|
3812 lp2->lp_slang->sl_name, 2) == 0)
|
|
|
3813 {
|
|
|
3814 lp->lp_replang = lp2->lp_slang;
|
|
|
3815 break;
|
|
|
3816 }
|
|
|
3817 }
|
|
|
3818 }
|
|
|
3819
|
|
236
|
3820 return NULL;
|
|
|
3821 }
|
|
|
3822
|
|
|
3823 /*
|
|
376
|
3824 * Clear the midword characters for buffer "buf".
|
|
|
3825 */
|
|
|
3826 static void
|
|
|
3827 clear_midword(buf)
|
|
|
3828 buf_T *buf;
|
|
|
3829 {
|
|
|
3830 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
3831 #ifdef FEAT_MBYTE
|
|
|
3832 vim_free(buf->b_spell_ismw_mb);
|
|
|
3833 buf->b_spell_ismw_mb = NULL;
|
|
|
3834 #endif
|
|
|
3835 }
|
|
|
3836
|
|
|
3837 /*
|
|
|
3838 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
3839 * They add up to any currently used midword characters.
|
|
|
3840 */
|
|
|
3841 static void
|
|
|
3842 use_midword(lp, buf)
|
|
|
3843 slang_T *lp;
|
|
|
3844 buf_T *buf;
|
|
|
3845 {
|
|
|
3846 char_u *p;
|
|
|
3847
|
|
389
|
3848 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
3849 return;
|
|
|
3850
|
|
376
|
3851 for (p = lp->sl_midword; *p != NUL; )
|
|
|
3852 #ifdef FEAT_MBYTE
|
|
|
3853 if (has_mbyte)
|
|
|
3854 {
|
|
|
3855 int c, l, n;
|
|
|
3856 char_u *bp;
|
|
|
3857
|
|
|
3858 c = mb_ptr2char(p);
|
|
474
|
3859 l = (*mb_ptr2len)(p);
|
|
|
3860 if (c < 256 && l <= 2)
|
|
376
|
3861 buf->b_spell_ismw[c] = TRUE;
|
|
|
3862 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
3863 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
3864 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
3865 else
|
|
|
3866 {
|
|
|
3867 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
3868 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
3869 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
3870 if (bp != NULL)
|
|
|
3871 {
|
|
|
3872 vim_free(buf->b_spell_ismw_mb);
|
|
|
3873 buf->b_spell_ismw_mb = bp;
|
|
|
3874 vim_strncpy(bp + n, p, l);
|
|
|
3875 }
|
|
|
3876 }
|
|
|
3877 p += l;
|
|
|
3878 }
|
|
|
3879 else
|
|
|
3880 #endif
|
|
|
3881 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
3882 }
|
|
|
3883
|
|
|
3884 /*
|
|
236
|
3885 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
3886 * Each region is simply stored as the two characters of it's name.
|
|
381
|
3887 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
3888 */
|
|
|
3889 static int
|
|
|
3890 find_region(rp, region)
|
|
|
3891 char_u *rp;
|
|
|
3892 char_u *region;
|
|
|
3893 {
|
|
|
3894 int i;
|
|
|
3895
|
|
|
3896 for (i = 0; ; i += 2)
|
|
|
3897 {
|
|
|
3898 if (rp[i] == NUL)
|
|
|
3899 return REGION_ALL;
|
|
|
3900 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
3901 break;
|
|
|
3902 }
|
|
|
3903 return i / 2;
|
|
|
3904 }
|
|
|
3905
|
|
|
3906 /*
|
|
323
|
3907 * Return case type of word:
|
|
236
|
3908 * w word 0
|
|
300
|
3909 * Word WF_ONECAP
|
|
|
3910 * W WORD WF_ALLCAP
|
|
|
3911 * WoRd wOrd WF_KEEPCAP
|
|
236
|
3912 */
|
|
|
3913 static int
|
|
|
3914 captype(word, end)
|
|
|
3915 char_u *word;
|
|
323
|
3916 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
3917 {
|
|
|
3918 char_u *p;
|
|
|
3919 int c;
|
|
|
3920 int firstcap;
|
|
|
3921 int allcap;
|
|
|
3922 int past_second = FALSE; /* past second word char */
|
|
|
3923
|
|
|
3924 /* find first letter */
|
|
376
|
3925 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
3926 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
3927 return 0; /* only non-word characters, illegal word */
|
|
|
3928 #ifdef FEAT_MBYTE
|
|
310
|
3929 if (has_mbyte)
|
|
|
3930 c = mb_ptr2char_adv(&p);
|
|
|
3931 else
|
|
236
|
3932 #endif
|
|
310
|
3933 c = *p++;
|
|
324
|
3934 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
3935
|
|
|
3936 /*
|
|
|
3937 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
3938 * But a word with an upper char only at start is a ONECAP.
|
|
|
3939 */
|
|
323
|
3940 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
3941 if (spell_iswordp_nmw(p))
|
|
236
|
3942 {
|
|
455
|
3943 c = PTR2CHAR(p);
|
|
324
|
3944 if (!SPELL_ISUPPER(c))
|
|
236
|
3945 {
|
|
|
3946 /* UUl -> KEEPCAP */
|
|
|
3947 if (past_second && allcap)
|
|
300
|
3948 return WF_KEEPCAP;
|
|
236
|
3949 allcap = FALSE;
|
|
|
3950 }
|
|
|
3951 else if (!allcap)
|
|
|
3952 /* UlU -> KEEPCAP */
|
|
300
|
3953 return WF_KEEPCAP;
|
|
236
|
3954 past_second = TRUE;
|
|
|
3955 }
|
|
|
3956
|
|
|
3957 if (allcap)
|
|
300
|
3958 return WF_ALLCAP;
|
|
236
|
3959 if (firstcap)
|
|
300
|
3960 return WF_ONECAP;
|
|
236
|
3961 return 0;
|
|
|
3962 }
|
|
|
3963
|
|
474
|
3964 /*
|
|
|
3965 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
3966 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
3967 * Add ALLCAP for "WOrD".
|
|
|
3968 */
|
|
|
3969 static int
|
|
|
3970 badword_captype(word, end)
|
|
|
3971 char_u *word;
|
|
|
3972 char_u *end;
|
|
|
3973 {
|
|
|
3974 int flags = captype(word, end);
|
|
475
|
3975 int c;
|
|
474
|
3976 int l, u;
|
|
|
3977 int first;
|
|
|
3978 char_u *p;
|
|
|
3979
|
|
|
3980 if (flags & WF_KEEPCAP)
|
|
|
3981 {
|
|
|
3982 /* Count the number of UPPER and lower case letters. */
|
|
|
3983 l = u = 0;
|
|
|
3984 first = FALSE;
|
|
|
3985 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
3986 {
|
|
475
|
3987 c = PTR2CHAR(p);
|
|
|
3988 if (SPELL_ISUPPER(c))
|
|
474
|
3989 {
|
|
|
3990 ++u;
|
|
|
3991 if (p == word)
|
|
|
3992 first = TRUE;
|
|
|
3993 }
|
|
|
3994 else
|
|
|
3995 ++l;
|
|
|
3996 }
|
|
|
3997
|
|
|
3998 /* If there are more UPPER than lower case letters suggest an
|
|
|
3999 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
4000 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
4001 * require three upper case letters. */
|
|
|
4002 if (u > l && u > 2)
|
|
|
4003 flags |= WF_ALLCAP;
|
|
|
4004 else if (first)
|
|
|
4005 flags |= WF_ONECAP;
|
|
|
4006 }
|
|
|
4007 return flags;
|
|
|
4008 }
|
|
|
4009
|
|
355
|
4010 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
4011 /*
|
|
|
4012 * Free all languages.
|
|
|
4013 */
|
|
|
4014 void
|
|
|
4015 spell_free_all()
|
|
|
4016 {
|
|
503
|
4017 slang_T *slang;
|
|
355
|
4018 buf_T *buf;
|
|
385
|
4019 char_u fname[MAXPATHL];
|
|
355
|
4020
|
|
|
4021 /* Go through all buffers and handle 'spelllang'. */
|
|
|
4022 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
4023 ga_clear(&buf->b_langp);
|
|
|
4024
|
|
|
4025 while (first_lang != NULL)
|
|
|
4026 {
|
|
503
|
4027 slang = first_lang;
|
|
|
4028 first_lang = slang->sl_next;
|
|
|
4029 slang_free(slang);
|
|
355
|
4030 }
|
|
366
|
4031
|
|
385
|
4032 if (int_wordlist != NULL)
|
|
|
4033 {
|
|
|
4034 /* Delete the internal wordlist and its .spl file */
|
|
|
4035 mch_remove(int_wordlist);
|
|
|
4036 int_wordlist_spl(fname);
|
|
|
4037 mch_remove(fname);
|
|
|
4038 vim_free(int_wordlist);
|
|
|
4039 int_wordlist = NULL;
|
|
381
|
4040 }
|
|
|
4041
|
|
366
|
4042 init_spell_chartab();
|
|
355
|
4043 }
|
|
|
4044 # endif
|
|
|
4045
|
|
236
|
4046 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
4047 /*
|
|
|
4048 * Clear all spelling tables and reload them.
|
|
307
|
4049 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
4050 */
|
|
|
4051 void
|
|
|
4052 spell_reload()
|
|
|
4053 {
|
|
|
4054 buf_T *buf;
|
|
316
|
4055 win_T *wp;
|
|
236
|
4056
|
|
358
|
4057 /* Initialize the table for spell_iswordp(). */
|
|
236
|
4058 init_spell_chartab();
|
|
|
4059
|
|
|
4060 /* Unload all allocated memory. */
|
|
355
|
4061 spell_free_all();
|
|
236
|
4062
|
|
|
4063 /* Go through all buffers and handle 'spelllang'. */
|
|
|
4064 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
4065 {
|
|
316
|
4066 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
4067 * window for this buffer in which 'spell' is set. */
|
|
236
|
4068 if (*buf->b_p_spl != NUL)
|
|
316
|
4069 {
|
|
|
4070 FOR_ALL_WINDOWS(wp)
|
|
|
4071 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
4072 {
|
|
|
4073 (void)did_set_spelllang(buf);
|
|
|
4074 # ifdef FEAT_WINDOWS
|
|
|
4075 break;
|
|
|
4076 # endif
|
|
|
4077 }
|
|
|
4078 }
|
|
236
|
4079 }
|
|
|
4080 }
|
|
|
4081 # endif
|
|
|
4082
|
|
310
|
4083 /*
|
|
|
4084 * Reload the spell file "fname" if it's loaded.
|
|
|
4085 */
|
|
|
4086 static void
|
|
323
|
4087 spell_reload_one(fname, added_word)
|
|
310
|
4088 char_u *fname;
|
|
323
|
4089 int added_word; /* invoked through "zg" */
|
|
310
|
4090 {
|
|
503
|
4091 slang_T *slang;
|
|
323
|
4092 int didit = FALSE;
|
|
310
|
4093
|
|
503
|
4094 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
4095 {
|
|
|
4096 if (fullpathcmp(fname, slang->sl_fname, FALSE) == FPC_SAME)
|
|
|
4097 {
|
|
|
4098 slang_clear(slang);
|
|
|
4099 if (spell_load_file(fname, NULL, slang, FALSE) == NULL)
|
|
500
|
4100 /* reloading failed, clear the language */
|
|
503
|
4101 slang_clear(slang);
|
|
310
|
4102 redraw_all_later(NOT_VALID);
|
|
323
|
4103 didit = TRUE;
|
|
310
|
4104 }
|
|
500
|
4105 }
|
|
323
|
4106
|
|
|
4107 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
4108 * 'spelllang' to get it loaded. */
|
|
|
4109 if (added_word && !didit)
|
|
|
4110 did_set_spelllang(curbuf);
|
|
310
|
4111 }
|
|
|
4112
|
|
|
4113
|
|
236
|
4114 /*
|
|
|
4115 * Functions for ":mkspell".
|
|
|
4116 */
|
|
|
4117
|
|
300
|
4118 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
4119 and .dic file. */
|
|
|
4120 /*
|
|
|
4121 * Main structure to store the contents of a ".aff" file.
|
|
|
4122 */
|
|
|
4123 typedef struct afffile_S
|
|
|
4124 {
|
|
|
4125 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
498
|
4126 int af_flagtype; /* AFT_CHAR, AFT_LONG, AFT_NUM or AFT_CAPLONG */
|
|
481
|
4127 int af_slash; /* character used in word for slash */
|
|
495
|
4128 unsigned af_rar; /* RAR ID for rare word */
|
|
|
4129 unsigned af_kep; /* KEP ID for keep-case word */
|
|
|
4130 unsigned af_bad; /* BAD ID for banned word */
|
|
|
4131 unsigned af_needaffix; /* NEEDAFFIX ID */
|
|
500
|
4132 unsigned af_needcomp; /* NEEDCOMPOUND ID */
|
|
339
|
4133 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
4134 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
4135 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
495
|
4136 hashtab_T af_comp; /* hashtable for compound flags, compitem_T */
|
|
236
|
4137 } afffile_T;
|
|
|
4138
|
|
495
|
4139 #define AFT_CHAR 0 /* flags are one character */
|
|
498
|
4140 #define AFT_LONG 1 /* flags are two characters */
|
|
|
4141 #define AFT_CAPLONG 2 /* flags are one or two characters */
|
|
|
4142 #define AFT_NUM 3 /* flags are numbers, comma separated */
|
|
495
|
4143
|
|
236
|
4144 typedef struct affentry_S affentry_T;
|
|
|
4145 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
4146 struct affentry_S
|
|
|
4147 {
|
|
|
4148 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
4149 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
4150 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
4151 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
4152 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
483
|
4153 char_u ae_rare; /* rare affix */
|
|
|
4154 char_u ae_nocomp; /* word with affix not compoundable */
|
|
300
|
4155 };
|
|
|
4156
|
|
495
|
4157 #ifdef FEAT_MBYTE
|
|
|
4158 # define AH_KEY_LEN 17 /* 2 x 8 bytes + NUL */
|
|
|
4159 #else
|
|
498
|
4160 # define AH_KEY_LEN 7 /* 6 digits + NUL */
|
|
495
|
4161 #endif
|
|
455
|
4162
|
|
300
|
4163 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
4164 typedef struct affheader_S
|
|
|
4165 {
|
|
495
|
4166 char_u ah_key[AH_KEY_LEN]; /* key for hashtab == name of affix */
|
|
|
4167 unsigned ah_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4168 int ah_newID; /* prefix ID after renumbering; 0 if not used */
|
|
300
|
4169 int ah_combine; /* suffix may combine with prefix */
|
|
498
|
4170 int ah_follows; /* another affix block should be following */
|
|
300
|
4171 affentry_T *ah_first; /* first affix entry */
|
|
|
4172 } affheader_T;
|
|
|
4173
|
|
|
4174 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
4175
|
|
495
|
4176 /* Flag used in compound items. */
|
|
|
4177 typedef struct compitem_S
|
|
|
4178 {
|
|
|
4179 char_u ci_key[AH_KEY_LEN]; /* key for hashtab == name of compound */
|
|
|
4180 unsigned ci_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4181 int ci_newID; /* affix ID after renumbering. */
|
|
|
4182 } compitem_T;
|
|
|
4183
|
|
|
4184 #define HI2CI(hi) ((compitem_T *)(hi)->hi_key)
|
|
|
4185
|
|
300
|
4186 /*
|
|
|
4187 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
4188 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
4189 * once after ":mkspell" is done.
|
|
300
|
4190 */
|
|
|
4191 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
4192 typedef struct sblock_S sblock_T;
|
|
|
4193 struct sblock_S
|
|
|
4194 {
|
|
|
4195 sblock_T *sb_next; /* next block in list */
|
|
|
4196 int sb_used; /* nr of bytes already in use */
|
|
|
4197 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
4198 };
|
|
|
4199
|
|
|
4200 /*
|
|
300
|
4201 * A node in the tree.
|
|
236
|
4202 */
|
|
300
|
4203 typedef struct wordnode_S wordnode_T;
|
|
|
4204 struct wordnode_S
|
|
236
|
4205 {
|
|
346
|
4206 union /* shared to save space */
|
|
|
4207 {
|
|
469
|
4208 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
4209 int index; /* index in written nodes (valid after first
|
|
|
4210 round) */
|
|
|
4211 } wn_u1;
|
|
|
4212 union /* shared to save space */
|
|
|
4213 {
|
|
|
4214 wordnode_T *next; /* next node with same hash key */
|
|
|
4215 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
4216 } wn_u2;
|
|
300
|
4217 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
4218 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
4219 always sorted) */
|
|
474
|
4220 int wn_refs; /* Nr. of references to this node. Only
|
|
|
4221 relevant for first node in a list of
|
|
|
4222 siblings, in following siblings it is
|
|
|
4223 always one. */
|
|
300
|
4224 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
481
|
4225 char_u wn_affixID; /* when "wn_byte" is NUL: supported/required
|
|
469
|
4226 prefix ID or 0 */
|
|
|
4227 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
4228 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
4229 PREFIXTREE it's the prefcondnr */
|
|
469
|
4230 #ifdef SPELL_PRINTTREE
|
|
|
4231 int wn_nr; /* sequence nr for printing */
|
|
|
4232 #endif
|
|
236
|
4233 };
|
|
|
4234
|
|
390
|
4235 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
4236
|
|
300
|
4237 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
4238
|
|
300
|
4239 /*
|
|
|
4240 * Info used while reading the spell files.
|
|
|
4241 */
|
|
|
4242 typedef struct spellinfo_S
|
|
249
|
4243 {
|
|
300
|
4244 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
4245 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
4246
|
|
300
|
4247 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
4248 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
4249
|
|
339
|
4250 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
4251
|
|
300
|
4252 sblock_T *si_blocks; /* memory blocks used */
|
|
482
|
4253 long si_blocks_cnt; /* memory blocks allocated */
|
|
|
4254 long si_compress_cnt; /* words to add before lowering
|
|
|
4255 compression limit */
|
|
474
|
4256 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
4257 compression, linked by "wn_child" field. */
|
|
482
|
4258 long si_free_count; /* number of nodes in si_first_free */
|
|
474
|
4259 #ifdef SPELL_PRINTTREE
|
|
|
4260 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
4261 #endif
|
|
|
4262
|
|
|
4263
|
|
300
|
4264 int si_ascii; /* handling only ASCII words */
|
|
310
|
4265 int si_add; /* addition file */
|
|
351
|
4266 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
4267 int si_region; /* region mask */
|
|
|
4268 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
4269 int si_memtot; /* runtime memory used */
|
|
310
|
4270 int si_verbose; /* verbose messages */
|
|
474
|
4271 int si_msg_count; /* number of words added since last message */
|
|
316
|
4272 int si_region_count; /* number of regions supported (1 when there
|
|
|
4273 are no regions) */
|
|
483
|
4274 char_u si_region_name[16]; /* region names; used only if
|
|
|
4275 * si_region_count > 1) */
|
|
323
|
4276
|
|
|
4277 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
4278 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
4279 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
4280 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
4281 int si_followup; /* soundsalike: ? */
|
|
|
4282 int si_collapse; /* soundsalike: ? */
|
|
|
4283 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
4284 garray_T si_map; /* MAP info concatenated */
|
|
495
|
4285 char_u *si_midword; /* MIDWORD chars or NULL */
|
|
483
|
4286 int si_compmax; /* max nr of words for compounding */
|
|
481
|
4287 int si_compminlen; /* minimal length for compounding */
|
|
483
|
4288 int si_compsylmax; /* max nr of syllables for compounding */
|
|
481
|
4289 char_u *si_compflags; /* flags used for compounding */
|
|
492
|
4290 char_u si_nobreak; /* NOBREAK */
|
|
483
|
4291 char_u *si_syllable; /* syllable string */
|
|
339
|
4292 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
4293 * prefixes, each stored as a string */
|
|
495
|
4294 int si_newprefID; /* current value for ah_newID */
|
|
500
|
4295 int si_newcompID; /* current value for compound ID */
|
|
300
|
4296 } spellinfo_T;
|
|
249
|
4297
|
|
474
|
4298 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
495
|
4299 static unsigned affitem2flag __ARGS((int flagtype, char_u *item, char_u *fname, int lnum));
|
|
|
4300 static unsigned get_affitem __ARGS((int flagtype, char_u **pp));
|
|
|
4301 static void process_compflags __ARGS((spellinfo_T *spin, afffile_T *aff, char_u *compflags));
|
|
500
|
4302 static void check_renumber __ARGS((spellinfo_T *spin));
|
|
495
|
4303 static int flag_in_afflist __ARGS((int flagtype, char_u *afflist, unsigned flag));
|
|
|
4304 static void aff_check_number __ARGS((int spinval, int affval, char *name));
|
|
|
4305 static void aff_check_string __ARGS((char_u *spinval, char_u *affval, char *name));
|
|
339
|
4306 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
4307 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
4308 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
4309 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
4310 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
4311 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
483
|
4312 static int get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
495
|
4313 static void get_compflags __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
483
|
4314 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
|
4315 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
4316 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
4317 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
4318 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
4319 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
483
|
4320 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist, int need_affix));
|
|
481
|
4321 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID));
|
|
474
|
4322 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
|
4323 static void deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
|
4324 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
4325 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
4326 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
4327 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
500
|
4328 static int write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
4329 static void clear_node __ARGS((wordnode_T *node));
|
|
|
4330 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
4331 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
4332 static void init_spellfile __ARGS((void));
|
|
236
|
4333
|
|
455
|
4334 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
4335 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
4336 * Use a negative number with the lower 8 bits zero. */
|
|
|
4337 #define PFX_FLAGS -256
|
|
390
|
4338
|
|
483
|
4339 /*
|
|
|
4340 * Tunable parameters for when the tree is compressed. See 'mkspellmem'.
|
|
|
4341 */
|
|
|
4342 static long compress_start = 30000; /* memory / SBLOCKSIZE */
|
|
|
4343 static long compress_inc = 100; /* memory / SBLOCKSIZE */
|
|
|
4344 static long compress_added = 500000; /* word count */
|
|
|
4345
|
|
469
|
4346 #ifdef SPELL_PRINTTREE
|
|
|
4347 /*
|
|
|
4348 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
4349 * readable format, so that we can see what happens when adding a word and/or
|
|
|
4350 * compressing the tree.
|
|
|
4351 * Based on code from Olaf Seibert.
|
|
|
4352 */
|
|
|
4353 #define PRINTLINESIZE 1000
|
|
|
4354 #define PRINTWIDTH 6
|
|
|
4355
|
|
|
4356 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
4357 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
4358
|
|
|
4359 static char line1[PRINTLINESIZE];
|
|
|
4360 static char line2[PRINTLINESIZE];
|
|
|
4361 static char line3[PRINTLINESIZE];
|
|
|
4362
|
|
|
4363 static void
|
|
|
4364 spell_clear_flags(wordnode_T *node)
|
|
|
4365 {
|
|
|
4366 wordnode_T *np;
|
|
|
4367
|
|
|
4368 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4369 {
|
|
|
4370 np->wn_u1.index = FALSE;
|
|
|
4371 spell_clear_flags(np->wn_child);
|
|
|
4372 }
|
|
|
4373 }
|
|
|
4374
|
|
|
4375 static void
|
|
|
4376 spell_print_node(wordnode_T *node, int depth)
|
|
|
4377 {
|
|
|
4378 if (node->wn_u1.index)
|
|
|
4379 {
|
|
|
4380 /* Done this node before, print the reference. */
|
|
|
4381 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
4382 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
4383 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4384 msg(line1);
|
|
|
4385 msg(line2);
|
|
|
4386 msg(line3);
|
|
|
4387 }
|
|
|
4388 else
|
|
|
4389 {
|
|
|
4390 node->wn_u1.index = TRUE;
|
|
|
4391
|
|
|
4392 if (node->wn_byte != NUL)
|
|
|
4393 {
|
|
|
4394 if (node->wn_child != NULL)
|
|
|
4395 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
4396 else
|
|
|
4397 /* Cannot happen? */
|
|
|
4398 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
4399 }
|
|
|
4400 else
|
|
|
4401 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
4402
|
|
|
4403 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
4404
|
|
|
4405 if (node->wn_sibling != NULL)
|
|
|
4406 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
4407 else
|
|
|
4408 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4409
|
|
|
4410 if (node->wn_byte == NUL)
|
|
|
4411 {
|
|
|
4412 msg(line1);
|
|
|
4413 msg(line2);
|
|
|
4414 msg(line3);
|
|
|
4415 }
|
|
|
4416
|
|
|
4417 /* do the children */
|
|
|
4418 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
4419 spell_print_node(node->wn_child, depth + 1);
|
|
|
4420
|
|
|
4421 /* do the siblings */
|
|
|
4422 if (node->wn_sibling != NULL)
|
|
|
4423 {
|
|
|
4424 /* get rid of all parent details except | */
|
|
|
4425 STRCPY(line1, line3);
|
|
|
4426 STRCPY(line2, line3);
|
|
|
4427 spell_print_node(node->wn_sibling, depth);
|
|
|
4428 }
|
|
|
4429 }
|
|
|
4430 }
|
|
|
4431
|
|
|
4432 static void
|
|
|
4433 spell_print_tree(wordnode_T *root)
|
|
|
4434 {
|
|
|
4435 if (root != NULL)
|
|
|
4436 {
|
|
|
4437 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
4438 * done. */
|
|
|
4439 spell_clear_flags(root);
|
|
|
4440
|
|
|
4441 /* Recursively print the tree. */
|
|
|
4442 spell_print_node(root, 0);
|
|
|
4443 }
|
|
|
4444 }
|
|
|
4445 #endif /* SPELL_PRINTTREE */
|
|
|
4446
|
|
236
|
4447 /*
|
|
323
|
4448 * Read the affix file "fname".
|
|
316
|
4449 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
4450 */
|
|
|
4451 static afffile_T *
|
|
474
|
4452 spell_read_aff(spin, fname)
|
|
|
4453 spellinfo_T *spin;
|
|
236
|
4454 char_u *fname;
|
|
|
4455 {
|
|
|
4456 FILE *fd;
|
|
|
4457 afffile_T *aff;
|
|
|
4458 char_u rline[MAXLINELEN];
|
|
|
4459 char_u *line;
|
|
|
4460 char_u *pc = NULL;
|
|
334
|
4461 #define MAXITEMCNT 7
|
|
|
4462 char_u *(items[MAXITEMCNT]);
|
|
236
|
4463 int itemcnt;
|
|
|
4464 char_u *p;
|
|
|
4465 int lnum = 0;
|
|
|
4466 affheader_T *cur_aff = NULL;
|
|
495
|
4467 int did_postpone_prefix = FALSE;
|
|
236
|
4468 int aff_todo = 0;
|
|
|
4469 hashtab_T *tp;
|
|
255
|
4470 char_u *low = NULL;
|
|
|
4471 char_u *fol = NULL;
|
|
|
4472 char_u *upp = NULL;
|
|
323
|
4473 int do_rep;
|
|
|
4474 int do_sal;
|
|
|
4475 int do_map;
|
|
|
4476 int found_map = FALSE;
|
|
324
|
4477 hashitem_T *hi;
|
|
455
|
4478 int l;
|
|
495
|
4479 int compminlen = 0; /* COMPOUNDMIN value */
|
|
|
4480 int compsylmax = 0; /* COMPOUNDSYLMAX value */
|
|
|
4481 int compmax = 0; /* COMPOUNDMAX value */
|
|
|
4482 char_u *compflags = NULL; /* COMPOUNDFLAG and COMPOUNDFLAGS
|
|
|
4483 concatenated */
|
|
|
4484 char_u *midword = NULL; /* MIDWORD value */
|
|
|
4485 char_u *syllable = NULL; /* SYLLABLE value */
|
|
|
4486 char_u *sofofrom = NULL; /* SOFOFROM value */
|
|
|
4487 char_u *sofoto = NULL; /* SOFOTO value */
|
|
236
|
4488
|
|
300
|
4489 /*
|
|
|
4490 * Open the file.
|
|
|
4491 */
|
|
310
|
4492 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4493 if (fd == NULL)
|
|
|
4494 {
|
|
|
4495 EMSG2(_(e_notopen), fname);
|
|
|
4496 return NULL;
|
|
|
4497 }
|
|
|
4498
|
|
310
|
4499 if (spin->si_verbose || p_verbose > 2)
|
|
|
4500 {
|
|
|
4501 if (!spin->si_verbose)
|
|
|
4502 verbose_enter();
|
|
366
|
4503 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
4504 out_flush();
|
|
|
4505 if (!spin->si_verbose)
|
|
|
4506 verbose_leave();
|
|
|
4507 }
|
|
236
|
4508
|
|
323
|
4509 /* Only do REP lines when not done in another .aff file already. */
|
|
|
4510 do_rep = spin->si_rep.ga_len == 0;
|
|
|
4511
|
|
|
4512 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
4513 do_sal = spin->si_sal.ga_len == 0;
|
|
|
4514
|
|
|
4515 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
4516 do_map = spin->si_map.ga_len == 0;
|
|
|
4517
|
|
300
|
4518 /*
|
|
|
4519 * Allocate and init the afffile_T structure.
|
|
|
4520 */
|
|
474
|
4521 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
4522 if (aff == NULL)
|
|
|
4523 return NULL;
|
|
|
4524 hash_init(&aff->af_pref);
|
|
|
4525 hash_init(&aff->af_suff);
|
|
495
|
4526 hash_init(&aff->af_comp);
|
|
236
|
4527
|
|
|
4528 /*
|
|
|
4529 * Read all the lines in the file one by one.
|
|
|
4530 */
|
|
255
|
4531 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
4532 {
|
|
255
|
4533 line_breakcheck();
|
|
236
|
4534 ++lnum;
|
|
|
4535
|
|
|
4536 /* Skip comment lines. */
|
|
|
4537 if (*rline == '#')
|
|
|
4538 continue;
|
|
|
4539
|
|
|
4540 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
4541 vim_free(pc);
|
|
310
|
4542 #ifdef FEAT_MBYTE
|
|
300
|
4543 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4544 {
|
|
300
|
4545 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
4546 if (pc == NULL)
|
|
|
4547 {
|
|
|
4548 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4549 fname, lnum, rline);
|
|
|
4550 continue;
|
|
|
4551 }
|
|
236
|
4552 line = pc;
|
|
|
4553 }
|
|
|
4554 else
|
|
310
|
4555 #endif
|
|
236
|
4556 {
|
|
|
4557 pc = NULL;
|
|
|
4558 line = rline;
|
|
|
4559 }
|
|
|
4560
|
|
|
4561 /* Split the line up in white separated items. Put a NUL after each
|
|
|
4562 * item. */
|
|
|
4563 itemcnt = 0;
|
|
|
4564 for (p = line; ; )
|
|
|
4565 {
|
|
|
4566 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
4567 ++p;
|
|
|
4568 if (*p == NUL)
|
|
|
4569 break;
|
|
334
|
4570 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
4571 break;
|
|
236
|
4572 items[itemcnt++] = p;
|
|
300
|
4573 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
4574 ++p;
|
|
|
4575 if (*p == NUL)
|
|
|
4576 break;
|
|
|
4577 *p++ = NUL;
|
|
|
4578 }
|
|
|
4579
|
|
|
4580 /* Handle non-empty lines. */
|
|
|
4581 if (itemcnt > 0)
|
|
|
4582 {
|
|
|
4583 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
4584 && aff->af_enc == NULL)
|
|
|
4585 {
|
|
310
|
4586 #ifdef FEAT_MBYTE
|
|
300
|
4587 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
4588 aff->af_enc = enc_canonize(items[1]);
|
|
|
4589 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
4590 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
4591 p_enc) == FAIL)
|
|
|
4592 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
4593 fname, aff->af_enc, p_enc);
|
|
375
|
4594 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4595 #else
|
|
|
4596 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4597 #endif
|
|
236
|
4598 }
|
|
495
|
4599 else if (STRCMP(items[0], "FLAG") == 0 && itemcnt == 2
|
|
|
4600 && aff->af_flagtype == AFT_CHAR)
|
|
|
4601 {
|
|
|
4602 if (STRCMP(items[1], "long") == 0)
|
|
498
|
4603 aff->af_flagtype = AFT_LONG;
|
|
495
|
4604 else if (STRCMP(items[1], "num") == 0)
|
|
498
|
4605 aff->af_flagtype = AFT_NUM;
|
|
|
4606 else if (STRCMP(items[1], "caplong") == 0)
|
|
|
4607 aff->af_flagtype = AFT_CAPLONG;
|
|
495
|
4608 else
|
|
|
4609 smsg((char_u *)_("Invalid value for FLAG in %s line %d: %s"),
|
|
|
4610 fname, lnum, items[1]);
|
|
|
4611 if (aff->af_rar != 0 || aff->af_kep != 0 || aff->af_bad != 0
|
|
500
|
4612 || aff->af_needaffix != 0
|
|
|
4613 || aff->af_needcomp != 0
|
|
|
4614 || compflags != NULL
|
|
495
|
4615 || aff->af_suff.ht_used > 0
|
|
|
4616 || aff->af_pref.ht_used > 0)
|
|
|
4617 smsg((char_u *)_("FLAG after using flags in %s line %d: %s"),
|
|
|
4618 fname, lnum, items[1]);
|
|
|
4619 }
|
|
|
4620 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2
|
|
|
4621 && midword == NULL)
|
|
|
4622 {
|
|
|
4623 midword = getroom_save(spin, items[1]);
|
|
366
|
4624 }
|
|
302
|
4625 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
4626 {
|
|
323
|
4627 /* ignored, we always split */
|
|
302
|
4628 }
|
|
323
|
4629 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
4630 {
|
|
323
|
4631 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
4632 }
|
|
481
|
4633 else if (STRCMP(items[0], "SLASH") == 0 && itemcnt == 2
|
|
|
4634 && aff->af_slash == 0)
|
|
|
4635 {
|
|
|
4636 aff->af_slash = items[1][0];
|
|
|
4637 if (items[1][1] != NUL)
|
|
|
4638 smsg((char_u *)_("Character used for SLASH must be ASCII; in %s line %d: %s"),
|
|
|
4639 fname, lnum, items[1]);
|
|
|
4640 }
|
|
307
|
4641 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
4642 && aff->af_rar == 0)
|
|
|
4643 {
|
|
495
|
4644 aff->af_rar = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4645 fname, lnum);
|
|
307
|
4646 }
|
|
310
|
4647 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
4648 && aff->af_kep == 0)
|
|
307
|
4649 {
|
|
495
|
4650 aff->af_kep = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4651 fname, lnum);
|
|
307
|
4652 }
|
|
346
|
4653 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
4654 && aff->af_bad == 0)
|
|
|
4655 {
|
|
495
|
4656 aff->af_bad = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4657 fname, lnum);
|
|
346
|
4658 }
|
|
483
|
4659 else if (STRCMP(items[0], "NEEDAFFIX") == 0 && itemcnt == 2
|
|
|
4660 && aff->af_needaffix == 0)
|
|
|
4661 {
|
|
495
|
4662 aff->af_needaffix = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4663 fname, lnum);
|
|
483
|
4664 }
|
|
500
|
4665 else if (STRCMP(items[0], "NEEDCOMPOUND") == 0 && itemcnt == 2
|
|
|
4666 && aff->af_needcomp == 0)
|
|
|
4667 {
|
|
|
4668 aff->af_needcomp = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4669 fname, lnum);
|
|
|
4670 }
|
|
481
|
4671 else if (STRCMP(items[0], "COMPOUNDFLAG") == 0 && itemcnt == 2
|
|
495
|
4672 && compflags == NULL)
|
|
|
4673 {
|
|
|
4674 /* Turn flag "c" into COMPOUNDFLAGS compatible string "c+",
|
|
|
4675 * "Na" into "Na+", "1234" into "1234+". */
|
|
|
4676 p = getroom(spin, STRLEN(items[1]) + 2, FALSE);
|
|
483
|
4677 if (p != NULL)
|
|
|
4678 {
|
|
495
|
4679 STRCPY(p, items[1]);
|
|
|
4680 STRCAT(p, "+");
|
|
|
4681 compflags = p;
|
|
|
4682 }
|
|
481
|
4683 }
|
|
483
|
4684 else if (STRCMP(items[0], "COMPOUNDFLAGS") == 0 && itemcnt == 2)
|
|
|
4685 {
|
|
|
4686 /* Concatenate this string to previously defined ones, using a
|
|
|
4687 * slash to separate them. */
|
|
|
4688 l = STRLEN(items[1]) + 1;
|
|
495
|
4689 if (compflags != NULL)
|
|
|
4690 l += STRLEN(compflags) + 1;
|
|
483
|
4691 p = getroom(spin, l, FALSE);
|
|
|
4692 if (p != NULL)
|
|
|
4693 {
|
|
495
|
4694 if (compflags != NULL)
|
|
483
|
4695 {
|
|
495
|
4696 STRCPY(p, compflags);
|
|
483
|
4697 STRCAT(p, "/");
|
|
|
4698 }
|
|
|
4699 STRCAT(p, items[1]);
|
|
495
|
4700 compflags = p;
|
|
483
|
4701 }
|
|
|
4702 }
|
|
|
4703 else if (STRCMP(items[0], "COMPOUNDMAX") == 0 && itemcnt == 2
|
|
495
|
4704 && compmax == 0)
|
|
|
4705 {
|
|
|
4706 compmax = atoi((char *)items[1]);
|
|
|
4707 if (compmax == 0)
|
|
483
|
4708 smsg((char_u *)_("Wrong COMPOUNDMAX value in %s line %d: %s"),
|
|
|
4709 fname, lnum, items[1]);
|
|
481
|
4710 }
|
|
|
4711 else if (STRCMP(items[0], "COMPOUNDMIN") == 0 && itemcnt == 2
|
|
495
|
4712 && compminlen == 0)
|
|
|
4713 {
|
|
|
4714 compminlen = atoi((char *)items[1]);
|
|
|
4715 if (compminlen == 0)
|
|
481
|
4716 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"),
|
|
|
4717 fname, lnum, items[1]);
|
|
|
4718 }
|
|
483
|
4719 else if (STRCMP(items[0], "COMPOUNDSYLMAX") == 0 && itemcnt == 2
|
|
495
|
4720 && compsylmax == 0)
|
|
|
4721 {
|
|
|
4722 compsylmax = atoi((char *)items[1]);
|
|
|
4723 if (compsylmax == 0)
|
|
483
|
4724 smsg((char_u *)_("Wrong COMPOUNDSYLMAX value in %s line %d: %s"),
|
|
|
4725 fname, lnum, items[1]);
|
|
|
4726 }
|
|
|
4727 else if (STRCMP(items[0], "SYLLABLE") == 0 && itemcnt == 2
|
|
495
|
4728 && syllable == NULL)
|
|
|
4729 {
|
|
|
4730 syllable = getroom_save(spin, items[1]);
|
|
483
|
4731 }
|
|
492
|
4732 else if (STRCMP(items[0], "NOBREAK") == 0 && itemcnt == 1)
|
|
|
4733 {
|
|
|
4734 spin->si_nobreak = TRUE;
|
|
|
4735 }
|
|
339
|
4736 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
4737 {
|
|
|
4738 aff->af_pfxpostpone = TRUE;
|
|
|
4739 }
|
|
236
|
4740 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4741 || STRCMP(items[0], "SFX") == 0)
|
|
|
4742 && aff_todo == 0
|
|
334
|
4743 && itemcnt >= 4)
|
|
236
|
4744 {
|
|
498
|
4745 int lasti = 4;
|
|
|
4746 char_u key[AH_KEY_LEN];
|
|
|
4747
|
|
|
4748 if (*items[0] == 'P')
|
|
|
4749 tp = &aff->af_pref;
|
|
|
4750 else
|
|
|
4751 tp = &aff->af_suff;
|
|
|
4752
|
|
|
4753 /* Myspell allows the same affix name to be used multiple
|
|
|
4754 * times. The affix files that do this have an undocumented
|
|
|
4755 * "S" flag on all but the last block, thus we check for that
|
|
|
4756 * and store it in ah_follows. */
|
|
|
4757 vim_strncpy(key, items[1], AH_KEY_LEN - 1);
|
|
|
4758 hi = hash_find(tp, key);
|
|
|
4759 if (!HASHITEM_EMPTY(hi))
|
|
|
4760 {
|
|
|
4761 cur_aff = HI2AH(hi);
|
|
|
4762 if (cur_aff->ah_combine != (*items[2] == 'Y'))
|
|
|
4763 smsg((char_u *)_("Different combining flag in continued affix block in %s line %d: %s"),
|
|
|
4764 fname, lnum, items[1]);
|
|
|
4765 if (!cur_aff->ah_follows)
|
|
|
4766 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
4767 fname, lnum, items[1]);
|
|
|
4768 }
|
|
|
4769 else
|
|
|
4770 {
|
|
|
4771 /* New affix letter. */
|
|
|
4772 cur_aff = (affheader_T *)getroom(spin,
|
|
|
4773 sizeof(affheader_T), TRUE);
|
|
|
4774 if (cur_aff == NULL)
|
|
|
4775 break;
|
|
|
4776 cur_aff->ah_flag = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4777 fname, lnum);
|
|
|
4778 if (cur_aff->ah_flag == 0 || STRLEN(items[1]) >= AH_KEY_LEN)
|
|
|
4779 break;
|
|
|
4780 if (cur_aff->ah_flag == aff->af_bad
|
|
|
4781 || cur_aff->ah_flag == aff->af_rar
|
|
|
4782 || cur_aff->ah_flag == aff->af_kep
|
|
500
|
4783 || cur_aff->ah_flag == aff->af_needaffix
|
|
|
4784 || cur_aff->ah_flag == aff->af_needcomp)
|
|
|
4785 smsg((char_u *)_("Affix also used for BAD/RAR/KEP/NEEDAFFIX/NEEDCOMPOUND in %s line %d: %s"),
|
|
498
|
4786 fname, lnum, items[1]);
|
|
|
4787 STRCPY(cur_aff->ah_key, items[1]);
|
|
|
4788 hash_add(tp, cur_aff->ah_key);
|
|
|
4789
|
|
|
4790 cur_aff->ah_combine = (*items[2] == 'Y');
|
|
|
4791 }
|
|
|
4792
|
|
|
4793 /* Check for the "S" flag, which apparently means that another
|
|
|
4794 * block with the same affix name is following. */
|
|
|
4795 if (itemcnt > lasti && STRCMP(items[lasti], "S") == 0)
|
|
|
4796 {
|
|
|
4797 ++lasti;
|
|
|
4798 cur_aff->ah_follows = TRUE;
|
|
|
4799 }
|
|
|
4800 else
|
|
|
4801 cur_aff->ah_follows = FALSE;
|
|
|
4802
|
|
334
|
4803 /* Myspell allows extra text after the item, but that might
|
|
|
4804 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
498
|
4805 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
4806 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
4807 fname, lnum, items[4]);
|
|
|
4808
|
|
498
|
4809 if (STRCMP(items[2], "Y") != 0 && STRCMP(items[2], "N") != 0)
|
|
236
|
4810 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
4811 fname, lnum, items[2]);
|
|
339
|
4812
|
|
498
|
4813 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
|
4814 {
|
|
|
4815 if (cur_aff->ah_newID == 0)
|
|
495
|
4816 {
|
|
|
4817 /* Use a new number in the .spl file later, to be able
|
|
|
4818 * to handle multiple .aff files. */
|
|
500
|
4819 check_renumber(spin);
|
|
495
|
4820 cur_aff->ah_newID = ++spin->si_newprefID;
|
|
|
4821
|
|
|
4822 /* We only really use ah_newID if the prefix is
|
|
|
4823 * postponed. We know that only after handling all
|
|
|
4824 * the items. */
|
|
|
4825 did_postpone_prefix = FALSE;
|
|
|
4826 }
|
|
498
|
4827 else
|
|
|
4828 /* Did use the ID in a previous block. */
|
|
|
4829 did_postpone_prefix = TRUE;
|
|
|
4830 }
|
|
|
4831
|
|
300
|
4832 aff_todo = atoi((char *)items[3]);
|
|
236
|
4833 }
|
|
|
4834 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4835 || STRCMP(items[0], "SFX") == 0)
|
|
|
4836 && aff_todo > 0
|
|
|
4837 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
4838 && itemcnt >= 5)
|
|
236
|
4839 {
|
|
|
4840 affentry_T *aff_entry;
|
|
366
|
4841 int rare = FALSE;
|
|
483
|
4842 int nocomp = FALSE;
|
|
455
|
4843 int upper = FALSE;
|
|
366
|
4844 int lasti = 5;
|
|
|
4845
|
|
483
|
4846 /* Check for "rare" and "nocomp" after the other info. */
|
|
|
4847 while (itemcnt > lasti)
|
|
|
4848 {
|
|
|
4849 if (!rare && STRICMP(items[lasti], "rare") == 0)
|
|
|
4850 {
|
|
|
4851 rare = TRUE;
|
|
|
4852 ++lasti;
|
|
|
4853 }
|
|
|
4854 else if (!nocomp && STRICMP(items[lasti], "nocomp") == 0)
|
|
|
4855 {
|
|
|
4856 nocomp = TRUE;
|
|
|
4857 ++lasti;
|
|
|
4858 }
|
|
|
4859 else
|
|
|
4860 break;
|
|
366
|
4861 }
|
|
236
|
4862
|
|
334
|
4863 /* Myspell allows extra text after the item, but that might
|
|
|
4864 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
4865 if (itemcnt > lasti && *items[lasti] != '#')
|
|
482
|
4866 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]);
|
|
334
|
4867
|
|
236
|
4868 /* New item for an affix letter. */
|
|
|
4869 --aff_todo;
|
|
474
|
4870 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
4871 sizeof(affentry_T), TRUE);
|
|
236
|
4872 if (aff_entry == NULL)
|
|
|
4873 break;
|
|
366
|
4874 aff_entry->ae_rare = rare;
|
|
483
|
4875 aff_entry->ae_nocomp = nocomp;
|
|
240
|
4876
|
|
236
|
4877 if (STRCMP(items[2], "0") != 0)
|
|
474
|
4878 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
4879 if (STRCMP(items[3], "0") != 0)
|
|
474
|
4880 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
4881
|
|
300
|
4882 /* Don't use an affix entry with non-ASCII characters when
|
|
|
4883 * "spin->si_ascii" is TRUE. */
|
|
|
4884 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
4885 || has_non_ascii(aff_entry->ae_add)))
|
|
|
4886 {
|
|
|
4887 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
4888 cur_aff->ah_first = aff_entry;
|
|
300
|
4889
|
|
|
4890 if (STRCMP(items[4], ".") != 0)
|
|
|
4891 {
|
|
|
4892 char_u buf[MAXLINELEN];
|
|
|
4893
|
|
474
|
4894 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
4895 if (*items[0] == 'P')
|
|
|
4896 sprintf((char *)buf, "^%s", items[4]);
|
|
|
4897 else
|
|
|
4898 sprintf((char *)buf, "%s$", items[4]);
|
|
|
4899 aff_entry->ae_prog = vim_regcomp(buf,
|
|
481
|
4900 RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
4901 if (aff_entry->ae_prog == NULL)
|
|
|
4902 smsg((char_u *)_("Broken condition in %s line %d: %s"),
|
|
|
4903 fname, lnum, items[4]);
|
|
300
|
4904 }
|
|
339
|
4905
|
|
|
4906 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
4907 * for the condition. Use an existing one if possible. */
|
|
455
|
4908 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
4909 {
|
|
455
|
4910 /* When the chop string is one lower-case letter and
|
|
|
4911 * the add string ends in the upper-case letter we set
|
|
|
4912 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
4913 * letters from "ae_add". The condition must either
|
|
|
4914 * be empty or start with the same letter. */
|
|
|
4915 if (aff_entry->ae_chop != NULL
|
|
|
4916 && aff_entry->ae_add != NULL
|
|
|
4917 #ifdef FEAT_MBYTE
|
|
474
|
4918 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
4919 aff_entry->ae_chop)] == NUL
|
|
|
4920 #else
|
|
|
4921 && aff_entry->ae_chop[1] == NUL
|
|
|
4922 #endif
|
|
|
4923 )
|
|
339
|
4924 {
|
|
455
|
4925 int c, c_up;
|
|
|
4926
|
|
|
4927 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
4928 c_up = SPELL_TOUPPER(c);
|
|
|
4929 if (c_up != c
|
|
|
4930 && (aff_entry->ae_cond == NULL
|
|
|
4931 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
4932 {
|
|
|
4933 p = aff_entry->ae_add
|
|
|
4934 + STRLEN(aff_entry->ae_add);
|
|
|
4935 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
4936 if (PTR2CHAR(p) == c_up)
|
|
|
4937 {
|
|
|
4938 upper = TRUE;
|
|
|
4939 aff_entry->ae_chop = NULL;
|
|
|
4940 *p = NUL;
|
|
|
4941
|
|
|
4942 /* The condition is matched with the
|
|
|
4943 * actual word, thus must check for the
|
|
|
4944 * upper-case letter. */
|
|
|
4945 if (aff_entry->ae_cond != NULL)
|
|
|
4946 {
|
|
|
4947 char_u buf[MAXLINELEN];
|
|
|
4948 #ifdef FEAT_MBYTE
|
|
|
4949 if (has_mbyte)
|
|
|
4950 {
|
|
|
4951 onecap_copy(items[4], buf, TRUE);
|
|
|
4952 aff_entry->ae_cond = getroom_save(
|
|
474
|
4953 spin, buf);
|
|
455
|
4954 }
|
|
|
4955 else
|
|
|
4956 #endif
|
|
|
4957 *aff_entry->ae_cond = c_up;
|
|
|
4958 if (aff_entry->ae_cond != NULL)
|
|
|
4959 {
|
|
|
4960 sprintf((char *)buf, "^%s",
|
|
339
|
4961 aff_entry->ae_cond);
|
|
455
|
4962 vim_free(aff_entry->ae_prog);
|
|
|
4963 aff_entry->ae_prog = vim_regcomp(
|
|
|
4964 buf, RE_MAGIC + RE_STRING);
|
|
|
4965 }
|
|
|
4966 }
|
|
|
4967 }
|
|
|
4968 }
|
|
339
|
4969 }
|
|
|
4970
|
|
455
|
4971 if (aff_entry->ae_chop == NULL)
|
|
390
|
4972 {
|
|
455
|
4973 int idx;
|
|
|
4974 char_u **pp;
|
|
|
4975 int n;
|
|
|
4976
|
|
495
|
4977 /* Find a previously used condition. */
|
|
455
|
4978 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
4979 --idx)
|
|
|
4980 {
|
|
|
4981 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
4982 if (str_equal(p, aff_entry->ae_cond))
|
|
|
4983 break;
|
|
|
4984 }
|
|
|
4985 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
4986 {
|
|
|
4987 /* Not found, add a new condition. */
|
|
|
4988 idx = spin->si_prefcond.ga_len++;
|
|
|
4989 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
4990 + idx;
|
|
|
4991 if (aff_entry->ae_cond == NULL)
|
|
|
4992 *pp = NULL;
|
|
|
4993 else
|
|
474
|
4994 *pp = getroom_save(spin,
|
|
455
|
4995 aff_entry->ae_cond);
|
|
|
4996 }
|
|
|
4997
|
|
|
4998 /* Add the prefix to the prefix tree. */
|
|
|
4999 if (aff_entry->ae_add == NULL)
|
|
|
5000 p = (char_u *)"";
|
|
390
|
5001 else
|
|
455
|
5002 p = aff_entry->ae_add;
|
|
|
5003 /* PFX_FLAGS is a negative number, so that
|
|
|
5004 * tree_add_word() knows this is the prefix tree. */
|
|
|
5005 n = PFX_FLAGS;
|
|
|
5006 if (rare)
|
|
|
5007 n |= WFP_RARE;
|
|
|
5008 if (!cur_aff->ah_combine)
|
|
|
5009 n |= WFP_NC;
|
|
|
5010 if (upper)
|
|
|
5011 n |= WFP_UP;
|
|
474
|
5012 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
5013 idx, cur_aff->ah_newID);
|
|
495
|
5014 did_postpone_prefix = TRUE;
|
|
|
5015 }
|
|
|
5016
|
|
|
5017 /* Didn't actually use ah_newID, backup si_newprefID. */
|
|
|
5018 if (aff_todo == 0 && !did_postpone_prefix)
|
|
|
5019 {
|
|
|
5020 --spin->si_newprefID;
|
|
|
5021 cur_aff->ah_newID = 0;
|
|
390
|
5022 }
|
|
339
|
5023 }
|
|
240
|
5024 }
|
|
236
|
5025 }
|
|
495
|
5026 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2
|
|
|
5027 && fol == NULL)
|
|
|
5028 {
|
|
|
5029 fol = vim_strsave(items[1]);
|
|
|
5030 }
|
|
|
5031 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2
|
|
|
5032 && low == NULL)
|
|
|
5033 {
|
|
|
5034 low = vim_strsave(items[1]);
|
|
|
5035 }
|
|
|
5036 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2
|
|
|
5037 && upp == NULL)
|
|
|
5038 {
|
|
|
5039 upp = vim_strsave(items[1]);
|
|
255
|
5040 }
|
|
236
|
5041 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
5042 {
|
|
236
|
5043 /* Ignore REP count */;
|
|
323
|
5044 if (!isdigit(*items[1]))
|
|
|
5045 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
5046 fname, lnum);
|
|
|
5047 }
|
|
482
|
5048 else if (STRCMP(items[0], "REP") == 0 && itemcnt >= 3)
|
|
236
|
5049 {
|
|
|
5050 /* REP item */
|
|
482
|
5051 /* Myspell ignores extra arguments, we require it starts with
|
|
|
5052 * # to detect mistakes. */
|
|
|
5053 if (itemcnt > 3 && items[3][0] != '#')
|
|
|
5054 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]);
|
|
323
|
5055 if (do_rep)
|
|
532
|
5056 {
|
|
|
5057 /* Replace underscore with space (can't include a space
|
|
|
5058 * directly). */
|
|
|
5059 for (p = items[1]; *p != NUL; mb_ptr_adv(p))
|
|
|
5060 if (*p == '_')
|
|
|
5061 *p = ' ';
|
|
|
5062 for (p = items[2]; *p != NUL; mb_ptr_adv(p))
|
|
|
5063 if (*p == '_')
|
|
|
5064 *p = ' ';
|
|
323
|
5065 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
532
|
5066 }
|
|
323
|
5067 }
|
|
|
5068 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
5069 {
|
|
|
5070 /* MAP item or count */
|
|
|
5071 if (!found_map)
|
|
|
5072 {
|
|
|
5073 /* First line contains the count. */
|
|
|
5074 found_map = TRUE;
|
|
|
5075 if (!isdigit(*items[1]))
|
|
|
5076 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
5077 fname, lnum);
|
|
|
5078 }
|
|
|
5079 else if (do_map)
|
|
|
5080 {
|
|
346
|
5081 int c;
|
|
|
5082
|
|
|
5083 /* Check that every character appears only once. */
|
|
|
5084 for (p = items[1]; *p != NUL; )
|
|
|
5085 {
|
|
|
5086 #ifdef FEAT_MBYTE
|
|
|
5087 c = mb_ptr2char_adv(&p);
|
|
|
5088 #else
|
|
|
5089 c = *p++;
|
|
|
5090 #endif
|
|
|
5091 if ((spin->si_map.ga_len > 0
|
|
|
5092 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
5093 != NULL)
|
|
|
5094 || vim_strchr(p, c) != NULL)
|
|
|
5095 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
5096 fname, lnum);
|
|
|
5097 }
|
|
|
5098
|
|
323
|
5099 /* We simply concatenate all the MAP strings, separated by
|
|
|
5100 * slashes. */
|
|
|
5101 ga_concat(&spin->si_map, items[1]);
|
|
|
5102 ga_append(&spin->si_map, '/');
|
|
|
5103 }
|
|
|
5104 }
|
|
534
|
5105 /* Accept "SAL from to" and "SAL from to # comment". */
|
|
|
5106 else if (STRCMP(items[0], "SAL") == 0
|
|
|
5107 && (itemcnt == 3 || (itemcnt > 3 && items[3][0] == '#')))
|
|
323
|
5108 {
|
|
|
5109 if (do_sal)
|
|
|
5110 {
|
|
|
5111 /* SAL item (sounds-a-like)
|
|
|
5112 * Either one of the known keys or a from-to pair. */
|
|
|
5113 if (STRCMP(items[1], "followup") == 0)
|
|
|
5114 spin->si_followup = sal_to_bool(items[2]);
|
|
|
5115 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
5116 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
5117 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
5118 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
5119 else
|
|
|
5120 /* when "to" is "_" it means empty */
|
|
|
5121 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
5122 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
5123 : items[2]);
|
|
|
5124 }
|
|
236
|
5125 }
|
|
375
|
5126 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
495
|
5127 && sofofrom == NULL)
|
|
|
5128 {
|
|
|
5129 sofofrom = getroom_save(spin, items[1]);
|
|
375
|
5130 }
|
|
|
5131 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
495
|
5132 && sofoto == NULL)
|
|
|
5133 {
|
|
|
5134 sofoto = getroom_save(spin, items[1]);
|
|
375
|
5135 }
|
|
300
|
5136 else
|
|
481
|
5137 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"),
|
|
236
|
5138 fname, lnum, items[0]);
|
|
|
5139 }
|
|
|
5140 }
|
|
|
5141
|
|
255
|
5142 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
5143 {
|
|
351
|
5144 if (spin->si_clear_chartab)
|
|
|
5145 {
|
|
|
5146 /* Clear the char type tables, don't want to use any of the
|
|
|
5147 * currently used spell properties. */
|
|
|
5148 init_spell_chartab();
|
|
|
5149 spin->si_clear_chartab = FALSE;
|
|
|
5150 }
|
|
|
5151
|
|
316
|
5152 /*
|
|
|
5153 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
5154 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
5155 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
5156 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
5157 */
|
|
|
5158 if (!spin->si_ascii
|
|
|
5159 #ifdef FEAT_MBYTE
|
|
|
5160 && !enc_utf8
|
|
|
5161 #endif
|
|
|
5162 )
|
|
260
|
5163 {
|
|
|
5164 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
5165 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
5166 else
|
|
316
|
5167 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
5168 }
|
|
255
|
5169
|
|
|
5170 vim_free(fol);
|
|
|
5171 vim_free(low);
|
|
|
5172 vim_free(upp);
|
|
|
5173 }
|
|
|
5174
|
|
481
|
5175 /* Use compound specifications of the .aff file for the spell info. */
|
|
495
|
5176 if (compmax != 0)
|
|
|
5177 {
|
|
|
5178 aff_check_number(spin->si_compmax, compmax, "COMPOUNDMAX");
|
|
|
5179 spin->si_compmax = compmax;
|
|
|
5180 }
|
|
|
5181
|
|
|
5182 if (compminlen != 0)
|
|
|
5183 {
|
|
|
5184 aff_check_number(spin->si_compminlen, compminlen, "COMPOUNDMIN");
|
|
|
5185 spin->si_compminlen = compminlen;
|
|
|
5186 }
|
|
|
5187
|
|
|
5188 if (compsylmax != 0)
|
|
|
5189 {
|
|
|
5190 if (syllable == NULL)
|
|
|
5191 smsg((char_u *)_("COMPOUNDSYLMAX used without SYLLABLE"));
|
|
|
5192 aff_check_number(spin->si_compsylmax, compsylmax, "COMPOUNDSYLMAX");
|
|
|
5193 spin->si_compsylmax = compsylmax;
|
|
|
5194 }
|
|
|
5195
|
|
|
5196 if (compflags != NULL)
|
|
|
5197 process_compflags(spin, aff, compflags);
|
|
|
5198
|
|
|
5199 /* Check that we didn't use too many renumbered flags. */
|
|
500
|
5200 if (spin->si_newcompID < spin->si_newprefID)
|
|
|
5201 {
|
|
|
5202 if (spin->si_newcompID == 127 || spin->si_newcompID == 255)
|
|
495
|
5203 MSG(_("Too many postponed prefixes"));
|
|
500
|
5204 else if (spin->si_newprefID == 0 || spin->si_newprefID == 127)
|
|
495
|
5205 MSG(_("Too many compound flags"));
|
|
481
|
5206 else
|
|
495
|
5207 MSG(_("Too many posponed prefixes and/or compound flags"));
|
|
|
5208 }
|
|
|
5209
|
|
|
5210 if (syllable != NULL)
|
|
|
5211 {
|
|
|
5212 aff_check_string(spin->si_syllable, syllable, "SYLLABLE");
|
|
|
5213 spin->si_syllable = syllable;
|
|
|
5214 }
|
|
|
5215
|
|
|
5216 if (sofofrom != NULL || sofoto != NULL)
|
|
|
5217 {
|
|
|
5218 if (sofofrom == NULL || sofoto == NULL)
|
|
|
5219 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
5220 sofofrom == NULL ? "FROM" : "TO", fname);
|
|
|
5221 else if (spin->si_sal.ga_len > 0)
|
|
|
5222 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
483
|
5223 else
|
|
495
|
5224 {
|
|
|
5225 aff_check_string(spin->si_sofofr, sofofrom, "SOFOFROM");
|
|
|
5226 aff_check_string(spin->si_sofoto, sofoto, "SOFOTO");
|
|
|
5227 spin->si_sofofr = sofofrom;
|
|
|
5228 spin->si_sofoto = sofoto;
|
|
|
5229 }
|
|
|
5230 }
|
|
|
5231
|
|
|
5232 if (midword != NULL)
|
|
|
5233 {
|
|
|
5234 aff_check_string(spin->si_midword, midword, "MIDWORD");
|
|
|
5235 spin->si_midword = midword;
|
|
483
|
5236 }
|
|
|
5237
|
|
236
|
5238 vim_free(pc);
|
|
|
5239 fclose(fd);
|
|
|
5240 return aff;
|
|
|
5241 }
|
|
|
5242
|
|
|
5243 /*
|
|
495
|
5244 * Turn an affix flag name into a number, according to the FLAG type.
|
|
|
5245 * returns zero for failure.
|
|
|
5246 */
|
|
|
5247 static unsigned
|
|
|
5248 affitem2flag(flagtype, item, fname, lnum)
|
|
|
5249 int flagtype;
|
|
|
5250 char_u *item;
|
|
|
5251 char_u *fname;
|
|
|
5252 int lnum;
|
|
|
5253 {
|
|
|
5254 unsigned res;
|
|
|
5255 char_u *p = item;
|
|
|
5256
|
|
|
5257 res = get_affitem(flagtype, &p);
|
|
|
5258 if (res == 0)
|
|
|
5259 {
|
|
498
|
5260 if (flagtype == AFT_NUM)
|
|
495
|
5261 smsg((char_u *)_("Flag is not a number in %s line %d: %s"),
|
|
|
5262 fname, lnum, item);
|
|
|
5263 else
|
|
|
5264 smsg((char_u *)_("Illegal flag in %s line %d: %s"),
|
|
|
5265 fname, lnum, item);
|
|
|
5266 }
|
|
|
5267 if (*p != NUL)
|
|
|
5268 {
|
|
|
5269 smsg((char_u *)_(e_affname), fname, lnum, item);
|
|
|
5270 return 0;
|
|
|
5271 }
|
|
|
5272
|
|
|
5273 return res;
|
|
|
5274 }
|
|
|
5275
|
|
|
5276 /*
|
|
|
5277 * Get one affix name from "*pp" and advance the pointer.
|
|
|
5278 * Returns zero for an error, still advances the pointer then.
|
|
|
5279 */
|
|
|
5280 static unsigned
|
|
|
5281 get_affitem(flagtype, pp)
|
|
|
5282 int flagtype;
|
|
|
5283 char_u **pp;
|
|
|
5284 {
|
|
|
5285 int res;
|
|
|
5286
|
|
498
|
5287 if (flagtype == AFT_NUM)
|
|
495
|
5288 {
|
|
|
5289 if (!VIM_ISDIGIT(**pp))
|
|
|
5290 {
|
|
498
|
5291 ++*pp; /* always advance, avoid getting stuck */
|
|
495
|
5292 return 0;
|
|
|
5293 }
|
|
|
5294 res = getdigits(pp);
|
|
|
5295 }
|
|
|
5296 else
|
|
|
5297 {
|
|
|
5298 #ifdef FEAT_MBYTE
|
|
|
5299 res = mb_ptr2char_adv(pp);
|
|
|
5300 #else
|
|
|
5301 res = *(*pp)++;
|
|
|
5302 #endif
|
|
498
|
5303 if (flagtype == AFT_LONG || (flagtype == AFT_CAPLONG
|
|
495
|
5304 && res >= 'A' && res <= 'Z'))
|
|
|
5305 {
|
|
|
5306 if (**pp == NUL)
|
|
|
5307 return 0;
|
|
|
5308 #ifdef FEAT_MBYTE
|
|
|
5309 res = mb_ptr2char_adv(pp) + (res << 16);
|
|
|
5310 #else
|
|
|
5311 res = *(*pp)++ + (res << 16);
|
|
|
5312 #endif
|
|
|
5313 }
|
|
|
5314 }
|
|
|
5315 return res;
|
|
|
5316 }
|
|
|
5317
|
|
|
5318 /*
|
|
|
5319 * Process the "compflags" string used in an affix file and append it to
|
|
|
5320 * spin->si_compflags.
|
|
|
5321 * The processing involves changing the affix names to ID numbers, so that
|
|
|
5322 * they fit in one byte.
|
|
|
5323 */
|
|
|
5324 static void
|
|
|
5325 process_compflags(spin, aff, compflags)
|
|
|
5326 spellinfo_T *spin;
|
|
|
5327 afffile_T *aff;
|
|
|
5328 char_u *compflags;
|
|
|
5329 {
|
|
|
5330 char_u *p;
|
|
|
5331 char_u *prevp;
|
|
|
5332 unsigned flag;
|
|
|
5333 compitem_T *ci;
|
|
|
5334 int id;
|
|
|
5335 int len;
|
|
|
5336 char_u *tp;
|
|
|
5337 char_u key[AH_KEY_LEN];
|
|
|
5338 hashitem_T *hi;
|
|
|
5339
|
|
|
5340 /* Make room for the old and the new compflags, concatenated with a / in
|
|
|
5341 * between. Processing it makes it shorter, but we don't know by how
|
|
|
5342 * much, thus allocate the maximum. */
|
|
|
5343 len = STRLEN(compflags) + 1;
|
|
|
5344 if (spin->si_compflags != NULL)
|
|
|
5345 len += STRLEN(spin->si_compflags) + 1;
|
|
|
5346 p = getroom(spin, len, FALSE);
|
|
|
5347 if (p == NULL)
|
|
|
5348 return;
|
|
|
5349 if (spin->si_compflags != NULL)
|
|
|
5350 {
|
|
|
5351 STRCPY(p, spin->si_compflags);
|
|
|
5352 STRCAT(p, "/");
|
|
|
5353 }
|
|
|
5354 spin->si_compflags = p;
|
|
|
5355 tp = p + STRLEN(p);
|
|
|
5356
|
|
|
5357 for (p = compflags; *p != NUL; )
|
|
|
5358 {
|
|
|
5359 if (vim_strchr((char_u *)"/*+[]", *p) != NULL)
|
|
|
5360 /* Copy non-flag characters directly. */
|
|
|
5361 *tp++ = *p++;
|
|
|
5362 else
|
|
|
5363 {
|
|
|
5364 /* First get the flag number, also checks validity. */
|
|
|
5365 prevp = p;
|
|
|
5366 flag = get_affitem(aff->af_flagtype, &p);
|
|
|
5367 if (flag != 0)
|
|
|
5368 {
|
|
|
5369 /* Find the flag in the hashtable. If it was used before, use
|
|
|
5370 * the existing ID. Otherwise add a new entry. */
|
|
|
5371 vim_strncpy(key, prevp, p - prevp);
|
|
|
5372 hi = hash_find(&aff->af_comp, key);
|
|
|
5373 if (!HASHITEM_EMPTY(hi))
|
|
|
5374 id = HI2CI(hi)->ci_newID;
|
|
|
5375 else
|
|
|
5376 {
|
|
|
5377 ci = (compitem_T *)getroom(spin, sizeof(compitem_T), TRUE);
|
|
|
5378 if (ci == NULL)
|
|
|
5379 break;
|
|
|
5380 STRCPY(ci->ci_key, key);
|
|
|
5381 ci->ci_flag = flag;
|
|
|
5382 /* Avoid using a flag ID that has a special meaning in a
|
|
|
5383 * regexp (also inside []). */
|
|
|
5384 do
|
|
|
5385 {
|
|
500
|
5386 check_renumber(spin);
|
|
|
5387 id = spin->si_newcompID--;
|
|
|
5388 } while (vim_strchr((char_u *)"/+*[]\\-^", id) != NULL);
|
|
495
|
5389 ci->ci_newID = id;
|
|
|
5390 hash_add(&aff->af_comp, ci->ci_key);
|
|
|
5391 }
|
|
|
5392 *tp++ = id;
|
|
|
5393 }
|
|
498
|
5394 if (aff->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
5395 ++p;
|
|
|
5396 }
|
|
|
5397 }
|
|
|
5398
|
|
|
5399 *tp = NUL;
|
|
|
5400 }
|
|
|
5401
|
|
|
5402 /*
|
|
500
|
5403 * Check that the new IDs for postponed affixes and compounding don't overrun
|
|
|
5404 * each other. We have almost 255 available, but start at 0-127 to avoid
|
|
|
5405 * using two bytes for utf-8. When the 0-127 range is used up go to 128-255.
|
|
|
5406 * When that is used up an error message is given.
|
|
|
5407 */
|
|
|
5408 static void
|
|
|
5409 check_renumber(spin)
|
|
|
5410 spellinfo_T *spin;
|
|
|
5411 {
|
|
|
5412 if (spin->si_newprefID == spin->si_newcompID && spin->si_newcompID < 128)
|
|
|
5413 {
|
|
|
5414 spin->si_newprefID = 127;
|
|
|
5415 spin->si_newcompID = 255;
|
|
|
5416 }
|
|
|
5417 }
|
|
|
5418
|
|
|
5419 /*
|
|
495
|
5420 * Return TRUE if flag "flag" appears in affix list "afflist".
|
|
|
5421 */
|
|
|
5422 static int
|
|
|
5423 flag_in_afflist(flagtype, afflist, flag)
|
|
|
5424 int flagtype;
|
|
|
5425 char_u *afflist;
|
|
|
5426 unsigned flag;
|
|
|
5427 {
|
|
|
5428 char_u *p;
|
|
|
5429 unsigned n;
|
|
|
5430
|
|
|
5431 switch (flagtype)
|
|
|
5432 {
|
|
|
5433 case AFT_CHAR:
|
|
|
5434 return vim_strchr(afflist, flag) != NULL;
|
|
|
5435
|
|
498
|
5436 case AFT_CAPLONG:
|
|
|
5437 case AFT_LONG:
|
|
495
|
5438 for (p = afflist; *p != NUL; )
|
|
|
5439 {
|
|
|
5440 #ifdef FEAT_MBYTE
|
|
|
5441 n = mb_ptr2char_adv(&p);
|
|
|
5442 #else
|
|
|
5443 n = *p++;
|
|
|
5444 #endif
|
|
498
|
5445 if ((flagtype == AFT_LONG || (n >= 'A' && n <= 'Z'))
|
|
495
|
5446 && *p != NUL)
|
|
|
5447 #ifdef FEAT_MBYTE
|
|
|
5448 n = mb_ptr2char_adv(&p) + (n << 16);
|
|
|
5449 #else
|
|
|
5450 n = *p++ + (n << 16);
|
|
|
5451 #endif
|
|
|
5452 if (n == flag)
|
|
|
5453 return TRUE;
|
|
|
5454 }
|
|
|
5455 break;
|
|
|
5456
|
|
498
|
5457 case AFT_NUM:
|
|
495
|
5458 for (p = afflist; *p != NUL; )
|
|
|
5459 {
|
|
|
5460 n = getdigits(&p);
|
|
|
5461 if (n == flag)
|
|
|
5462 return TRUE;
|
|
|
5463 if (*p != NUL) /* skip over comma */
|
|
|
5464 ++p;
|
|
|
5465 }
|
|
|
5466 break;
|
|
|
5467 }
|
|
|
5468 return FALSE;
|
|
|
5469 }
|
|
|
5470
|
|
|
5471 /*
|
|
|
5472 * Give a warning when "spinval" and "affval" numbers are set and not the same.
|
|
|
5473 */
|
|
|
5474 static void
|
|
|
5475 aff_check_number(spinval, affval, name)
|
|
|
5476 int spinval;
|
|
|
5477 int affval;
|
|
|
5478 char *name;
|
|
|
5479 {
|
|
|
5480 if (spinval != 0 && spinval != affval)
|
|
|
5481 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
5482 }
|
|
|
5483
|
|
|
5484 /*
|
|
|
5485 * Give a warning when "spinval" and "affval" strings are set and not the same.
|
|
|
5486 */
|
|
|
5487 static void
|
|
|
5488 aff_check_string(spinval, affval, name)
|
|
|
5489 char_u *spinval;
|
|
|
5490 char_u *affval;
|
|
|
5491 char *name;
|
|
|
5492 {
|
|
|
5493 if (spinval != NULL && STRCMP(spinval, affval) != 0)
|
|
|
5494 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
5495 }
|
|
|
5496
|
|
|
5497 /*
|
|
339
|
5498 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
5499 * NULL as equal.
|
|
|
5500 */
|
|
|
5501 static int
|
|
|
5502 str_equal(s1, s2)
|
|
|
5503 char_u *s1;
|
|
|
5504 char_u *s2;
|
|
|
5505 {
|
|
|
5506 if (s1 == NULL || s2 == NULL)
|
|
|
5507 return s1 == s2;
|
|
|
5508 return STRCMP(s1, s2) == 0;
|
|
|
5509 }
|
|
|
5510
|
|
|
5511 /*
|
|
323
|
5512 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
5513 * They are stored case-folded.
|
|
|
5514 */
|
|
|
5515 static void
|
|
|
5516 add_fromto(spin, gap, from, to)
|
|
|
5517 spellinfo_T *spin;
|
|
|
5518 garray_T *gap;
|
|
|
5519 char_u *from;
|
|
|
5520 char_u *to;
|
|
|
5521 {
|
|
|
5522 fromto_T *ftp;
|
|
|
5523 char_u word[MAXWLEN];
|
|
|
5524
|
|
|
5525 if (ga_grow(gap, 1) == OK)
|
|
|
5526 {
|
|
|
5527 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
5528 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
5529 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
5530 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
5531 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
5532 ++gap->ga_len;
|
|
|
5533 }
|
|
|
5534 }
|
|
|
5535
|
|
|
5536 /*
|
|
|
5537 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
5538 */
|
|
|
5539 static int
|
|
|
5540 sal_to_bool(s)
|
|
|
5541 char_u *s;
|
|
|
5542 {
|
|
|
5543 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
5544 }
|
|
|
5545
|
|
|
5546 /*
|
|
240
|
5547 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
5548 * When "s" is NULL FALSE is returned.
|
|
|
5549 */
|
|
|
5550 static int
|
|
|
5551 has_non_ascii(s)
|
|
|
5552 char_u *s;
|
|
|
5553 {
|
|
|
5554 char_u *p;
|
|
|
5555
|
|
|
5556 if (s != NULL)
|
|
|
5557 for (p = s; *p != NUL; ++p)
|
|
|
5558 if (*p >= 128)
|
|
|
5559 return TRUE;
|
|
|
5560 return FALSE;
|
|
|
5561 }
|
|
|
5562
|
|
|
5563 /*
|
|
236
|
5564 * Free the structure filled by spell_read_aff().
|
|
|
5565 */
|
|
|
5566 static void
|
|
|
5567 spell_free_aff(aff)
|
|
|
5568 afffile_T *aff;
|
|
|
5569 {
|
|
|
5570 hashtab_T *ht;
|
|
|
5571 hashitem_T *hi;
|
|
|
5572 int todo;
|
|
|
5573 affheader_T *ah;
|
|
300
|
5574 affentry_T *ae;
|
|
236
|
5575
|
|
|
5576 vim_free(aff->af_enc);
|
|
|
5577
|
|
339
|
5578 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
5579 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
5580 {
|
|
|
5581 todo = ht->ht_used;
|
|
|
5582 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
5583 {
|
|
|
5584 if (!HASHITEM_EMPTY(hi))
|
|
|
5585 {
|
|
|
5586 --todo;
|
|
|
5587 ah = HI2AH(hi);
|
|
300
|
5588 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
5589 vim_free(ae->ae_prog);
|
|
236
|
5590 }
|
|
|
5591 }
|
|
|
5592 if (ht == &aff->af_suff)
|
|
|
5593 break;
|
|
|
5594 }
|
|
300
|
5595
|
|
236
|
5596 hash_clear(&aff->af_pref);
|
|
|
5597 hash_clear(&aff->af_suff);
|
|
495
|
5598 hash_clear(&aff->af_comp);
|
|
236
|
5599 }
|
|
|
5600
|
|
|
5601 /*
|
|
300
|
5602 * Read dictionary file "fname".
|
|
236
|
5603 * Returns OK or FAIL;
|
|
|
5604 */
|
|
|
5605 static int
|
|
474
|
5606 spell_read_dic(spin, fname, affile)
|
|
|
5607 spellinfo_T *spin;
|
|
236
|
5608 char_u *fname;
|
|
300
|
5609 afffile_T *affile;
|
|
236
|
5610 {
|
|
300
|
5611 hashtab_T ht;
|
|
236
|
5612 char_u line[MAXLINELEN];
|
|
481
|
5613 char_u *p;
|
|
300
|
5614 char_u *afflist;
|
|
483
|
5615 char_u store_afflist[MAXWLEN];
|
|
|
5616 int pfxlen;
|
|
|
5617 int need_affix;
|
|
300
|
5618 char_u *dw;
|
|
236
|
5619 char_u *pc;
|
|
|
5620 char_u *w;
|
|
|
5621 int l;
|
|
|
5622 hash_T hash;
|
|
|
5623 hashitem_T *hi;
|
|
|
5624 FILE *fd;
|
|
|
5625 int lnum = 1;
|
|
300
|
5626 int non_ascii = 0;
|
|
|
5627 int retval = OK;
|
|
|
5628 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
5629 int flags;
|
|
481
|
5630 int duplicate = 0;
|
|
236
|
5631
|
|
300
|
5632 /*
|
|
|
5633 * Open the file.
|
|
|
5634 */
|
|
310
|
5635 fd = mch_fopen((char *)fname, "r");
|
|
236
|
5636 if (fd == NULL)
|
|
|
5637 {
|
|
|
5638 EMSG2(_(e_notopen), fname);
|
|
|
5639 return FAIL;
|
|
|
5640 }
|
|
|
5641
|
|
300
|
5642 /* The hashtable is only used to detect duplicated words. */
|
|
|
5643 hash_init(&ht);
|
|
|
5644
|
|
310
|
5645 if (spin->si_verbose || p_verbose > 2)
|
|
|
5646 {
|
|
|
5647 if (!spin->si_verbose)
|
|
|
5648 verbose_enter();
|
|
366
|
5649 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
5650 out_flush();
|
|
|
5651 if (!spin->si_verbose)
|
|
|
5652 verbose_leave();
|
|
|
5653 }
|
|
236
|
5654
|
|
474
|
5655 /* start with a message for the first line */
|
|
|
5656 spin->si_msg_count = 999999;
|
|
|
5657
|
|
236
|
5658 /* Read and ignore the first line: word count. */
|
|
|
5659 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
5660 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
5661 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
5662
|
|
|
5663 /*
|
|
|
5664 * Read all the lines in the file one by one.
|
|
|
5665 * The words are converted to 'encoding' here, before being added to
|
|
|
5666 * the hashtable.
|
|
|
5667 */
|
|
255
|
5668 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
5669 {
|
|
255
|
5670 line_breakcheck();
|
|
236
|
5671 ++lnum;
|
|
455
|
5672 if (line[0] == '#' || line[0] == '/')
|
|
351
|
5673 continue; /* comment line */
|
|
236
|
5674
|
|
300
|
5675 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
5676 * the word is kept to allow e.g., "et al.". */
|
|
236
|
5677 l = STRLEN(line);
|
|
|
5678 while (l > 0 && line[l - 1] <= ' ')
|
|
|
5679 --l;
|
|
|
5680 if (l == 0)
|
|
|
5681 continue; /* empty line */
|
|
|
5682 line[l] = NUL;
|
|
|
5683
|
|
481
|
5684 /* Find the optional affix names. Replace the SLASH character by a
|
|
|
5685 * slash. */
|
|
|
5686 afflist = NULL;
|
|
|
5687 for (p = line; *p != NUL; mb_ptr_adv(p))
|
|
|
5688 {
|
|
|
5689 if (*p == affile->af_slash)
|
|
|
5690 *p = '/';
|
|
|
5691 else if (*p == '/')
|
|
|
5692 {
|
|
|
5693 *p = NUL;
|
|
|
5694 afflist = p + 1;
|
|
|
5695 break;
|
|
|
5696 }
|
|
|
5697 }
|
|
236
|
5698
|
|
300
|
5699 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
5700 if (spin->si_ascii && has_non_ascii(line))
|
|
|
5701 {
|
|
|
5702 ++non_ascii;
|
|
240
|
5703 continue;
|
|
300
|
5704 }
|
|
240
|
5705
|
|
310
|
5706 #ifdef FEAT_MBYTE
|
|
236
|
5707 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
5708 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
5709 {
|
|
300
|
5710 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
5711 if (pc == NULL)
|
|
|
5712 {
|
|
|
5713 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5714 fname, lnum, line);
|
|
|
5715 continue;
|
|
|
5716 }
|
|
236
|
5717 w = pc;
|
|
|
5718 }
|
|
|
5719 else
|
|
310
|
5720 #endif
|
|
236
|
5721 {
|
|
|
5722 pc = NULL;
|
|
|
5723 w = line;
|
|
|
5724 }
|
|
|
5725
|
|
474
|
5726 /* This takes time, print a message every 10000 words. */
|
|
|
5727 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
5728 {
|
|
|
5729 spin->si_msg_count = 0;
|
|
339
|
5730 vim_snprintf((char *)message, sizeof(message),
|
|
|
5731 _("line %6d, word %6d - %s"),
|
|
|
5732 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
5733 msg_start();
|
|
|
5734 msg_puts_long_attr(message, 0);
|
|
|
5735 msg_clr_eos();
|
|
|
5736 msg_didout = FALSE;
|
|
|
5737 msg_col = 0;
|
|
|
5738 out_flush();
|
|
|
5739 }
|
|
|
5740
|
|
300
|
5741 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
5742 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
5743 if (dw == NULL)
|
|
300
|
5744 retval = FAIL;
|
|
|
5745 vim_free(pc);
|
|
|
5746 if (retval == FAIL)
|
|
236
|
5747 break;
|
|
|
5748
|
|
300
|
5749 hash = hash_hash(dw);
|
|
|
5750 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
5751 if (!HASHITEM_EMPTY(hi))
|
|
481
|
5752 {
|
|
|
5753 if (p_verbose > 0)
|
|
|
5754 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
5755 fname, lnum, dw);
|
|
481
|
5756 else if (duplicate == 0)
|
|
|
5757 smsg((char_u *)_("First duplicate word in %s line %d: %s"),
|
|
|
5758 fname, lnum, dw);
|
|
|
5759 ++duplicate;
|
|
|
5760 }
|
|
236
|
5761 else
|
|
300
|
5762 hash_add_item(&ht, hi, dw, hash);
|
|
|
5763
|
|
307
|
5764 flags = 0;
|
|
483
|
5765 store_afflist[0] = NUL;
|
|
|
5766 pfxlen = 0;
|
|
|
5767 need_affix = FALSE;
|
|
307
|
5768 if (afflist != NULL)
|
|
|
5769 {
|
|
|
5770 /* Check for affix name that stands for keep-case word and stands
|
|
|
5771 * for rare word (if defined). */
|
|
495
|
5772 if (affile->af_kep != 0 && flag_in_afflist(
|
|
|
5773 affile->af_flagtype, afflist, affile->af_kep))
|
|
389
|
5774 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
495
|
5775 if (affile->af_rar != 0 && flag_in_afflist(
|
|
|
5776 affile->af_flagtype, afflist, affile->af_rar))
|
|
307
|
5777 flags |= WF_RARE;
|
|
495
|
5778 if (affile->af_bad != 0 && flag_in_afflist(
|
|
|
5779 affile->af_flagtype, afflist, affile->af_bad))
|
|
346
|
5780 flags |= WF_BANNED;
|
|
495
|
5781 if (affile->af_needaffix != 0 && flag_in_afflist(
|
|
|
5782 affile->af_flagtype, afflist, affile->af_needaffix))
|
|
483
|
5783 need_affix = TRUE;
|
|
500
|
5784 if (affile->af_needcomp != 0 && flag_in_afflist(
|
|
|
5785 affile->af_flagtype, afflist, affile->af_needcomp))
|
|
|
5786 flags |= WF_NEEDCOMP;
|
|
339
|
5787
|
|
|
5788 if (affile->af_pfxpostpone)
|
|
|
5789 /* Need to store the list of prefix IDs with the word. */
|
|
483
|
5790 pfxlen = get_pfxlist(affile, afflist, store_afflist);
|
|
|
5791
|
|
|
5792 if (spin->si_compflags != NULL)
|
|
|
5793 /* Need to store the list of compound flags with the word.
|
|
|
5794 * Concatenate them to the list of prefix IDs. */
|
|
495
|
5795 get_compflags(affile, afflist, store_afflist + pfxlen);
|
|
307
|
5796 }
|
|
|
5797
|
|
300
|
5798 /* Add the word to the word tree(s). */
|
|
483
|
5799 if (store_word(spin, dw, flags, spin->si_region,
|
|
|
5800 store_afflist, need_affix) == FAIL)
|
|
300
|
5801 retval = FAIL;
|
|
236
|
5802
|
|
300
|
5803 if (afflist != NULL)
|
|
|
5804 {
|
|
|
5805 /* Find all matching suffixes and add the resulting words.
|
|
|
5806 * Additionally do matching prefixes that combine. */
|
|
474
|
5807 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
5808 &affile->af_suff, &affile->af_pref,
|
|
483
|
5809 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5810 retval = FAIL;
|
|
|
5811
|
|
|
5812 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
5813 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
5814 &affile->af_pref, NULL,
|
|
483
|
5815 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5816 retval = FAIL;
|
|
|
5817 }
|
|
236
|
5818 }
|
|
|
5819
|
|
481
|
5820 if (duplicate > 0)
|
|
|
5821 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname);
|
|
300
|
5822 if (spin->si_ascii && non_ascii > 0)
|
|
481
|
5823 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"),
|
|
|
5824 non_ascii, fname);
|
|
300
|
5825 hash_clear(&ht);
|
|
|
5826
|
|
236
|
5827 fclose(fd);
|
|
300
|
5828 return retval;
|
|
236
|
5829 }
|
|
|
5830
|
|
|
5831 /*
|
|
339
|
5832 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
5833 * Used for PFXPOSTPONE.
|
|
483
|
5834 * Put the resulting flags in "store_afflist[MAXWLEN]" with a terminating NUL
|
|
|
5835 * and return the number of affixes.
|
|
|
5836 */
|
|
|
5837 static int
|
|
|
5838 get_pfxlist(affile, afflist, store_afflist)
|
|
339
|
5839 afffile_T *affile;
|
|
|
5840 char_u *afflist;
|
|
483
|
5841 char_u *store_afflist;
|
|
339
|
5842 {
|
|
|
5843 char_u *p;
|
|
495
|
5844 char_u *prevp;
|
|
483
|
5845 int cnt = 0;
|
|
495
|
5846 int id;
|
|
|
5847 char_u key[AH_KEY_LEN];
|
|
339
|
5848 hashitem_T *hi;
|
|
|
5849
|
|
495
|
5850 for (p = afflist; *p != NUL; )
|
|
|
5851 {
|
|
|
5852 prevp = p;
|
|
|
5853 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
5854 {
|
|
|
5855 /* A flag is a postponed prefix flag if it appears in "af_pref"
|
|
|
5856 * and it's ID is not zero. */
|
|
|
5857 vim_strncpy(key, prevp, p - prevp);
|
|
|
5858 hi = hash_find(&affile->af_pref, key);
|
|
|
5859 if (!HASHITEM_EMPTY(hi))
|
|
|
5860 {
|
|
|
5861 id = HI2AH(hi)->ah_newID;
|
|
|
5862 if (id != 0)
|
|
|
5863 store_afflist[cnt++] = id;
|
|
|
5864 }
|
|
|
5865 }
|
|
498
|
5866 if (affile->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
5867 ++p;
|
|
483
|
5868 }
|
|
|
5869
|
|
|
5870 store_afflist[cnt] = NUL;
|
|
|
5871 return cnt;
|
|
339
|
5872 }
|
|
|
5873
|
|
|
5874 /*
|
|
495
|
5875 * Get the list of compound IDs from the affix list "afflist" that are used
|
|
|
5876 * for compound words.
|
|
483
|
5877 * Puts the flags in "store_afflist[]".
|
|
|
5878 */
|
|
|
5879 static void
|
|
495
|
5880 get_compflags(affile, afflist, store_afflist)
|
|
|
5881 afffile_T *affile;
|
|
481
|
5882 char_u *afflist;
|
|
483
|
5883 char_u *store_afflist;
|
|
481
|
5884 {
|
|
|
5885 char_u *p;
|
|
495
|
5886 char_u *prevp;
|
|
483
|
5887 int cnt = 0;
|
|
495
|
5888 char_u key[AH_KEY_LEN];
|
|
|
5889 hashitem_T *hi;
|
|
|
5890
|
|
|
5891 for (p = afflist; *p != NUL; )
|
|
|
5892 {
|
|
|
5893 prevp = p;
|
|
|
5894 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
5895 {
|
|
|
5896 /* A flag is a compound flag if it appears in "af_comp". */
|
|
|
5897 vim_strncpy(key, prevp, p - prevp);
|
|
|
5898 hi = hash_find(&affile->af_comp, key);
|
|
|
5899 if (!HASHITEM_EMPTY(hi))
|
|
|
5900 store_afflist[cnt++] = HI2CI(hi)->ci_newID;
|
|
|
5901 }
|
|
498
|
5902 if (affile->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
5903 ++p;
|
|
|
5904 }
|
|
483
|
5905
|
|
|
5906 store_afflist[cnt] = NUL;
|
|
481
|
5907 }
|
|
|
5908
|
|
|
5909 /*
|
|
300
|
5910 * Apply affixes to a word and store the resulting words.
|
|
|
5911 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
5912 * prefixes or suffixes.
|
|
|
5913 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
5914 * the resulting words for combining affixes.
|
|
|
5915 *
|
|
|
5916 * Returns FAIL when out of memory.
|
|
236
|
5917 */
|
|
300
|
5918 static int
|
|
483
|
5919 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags,
|
|
|
5920 pfxlist, pfxlen)
|
|
474
|
5921 spellinfo_T *spin; /* spell info */
|
|
300
|
5922 char_u *word; /* basic word start */
|
|
|
5923 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
5924 afffile_T *affile;
|
|
300
|
5925 hashtab_T *ht;
|
|
|
5926 hashtab_T *xht;
|
|
|
5927 int comb; /* only use affixes that combine */
|
|
307
|
5928 int flags; /* flags for the word */
|
|
339
|
5929 char_u *pfxlist; /* list of prefix IDs */
|
|
483
|
5930 int pfxlen; /* nr of flags in "pfxlist" for prefixes, rest
|
|
|
5931 * is compound flags */
|
|
236
|
5932 {
|
|
|
5933 int todo;
|
|
|
5934 hashitem_T *hi;
|
|
300
|
5935 affheader_T *ah;
|
|
|
5936 affentry_T *ae;
|
|
|
5937 regmatch_T regmatch;
|
|
|
5938 char_u newword[MAXWLEN];
|
|
|
5939 int retval = OK;
|
|
|
5940 int i;
|
|
|
5941 char_u *p;
|
|
366
|
5942 int use_flags;
|
|
390
|
5943 char_u *use_pfxlist;
|
|
483
|
5944 char_u pfx_pfxlist[MAXWLEN];
|
|
|
5945 size_t wordlen = STRLEN(word);
|
|
236
|
5946
|
|
300
|
5947 todo = ht->ht_used;
|
|
|
5948 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
5949 {
|
|
|
5950 if (!HASHITEM_EMPTY(hi))
|
|
|
5951 {
|
|
|
5952 --todo;
|
|
300
|
5953 ah = HI2AH(hi);
|
|
236
|
5954
|
|
300
|
5955 /* Check that the affix combines, if required, and that the word
|
|
|
5956 * supports this affix. */
|
|
495
|
5957 if ((!comb || ah->ah_combine) && flag_in_afflist(
|
|
|
5958 affile->af_flagtype, afflist, ah->ah_flag))
|
|
236
|
5959 {
|
|
300
|
5960 /* Loop over all affix entries with this name. */
|
|
|
5961 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
5962 {
|
|
300
|
5963 /* Check the condition. It's not logical to match case
|
|
|
5964 * here, but it is required for compatibility with
|
|
339
|
5965 * Myspell.
|
|
481
|
5966 * Another requirement from Myspell is that the chop
|
|
|
5967 * string is shorter than the word itself.
|
|
339
|
5968 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
5969 * prefixes with a chop string. */
|
|
300
|
5970 regmatch.regprog = ae->ae_prog;
|
|
|
5971 regmatch.rm_ic = FALSE;
|
|
339
|
5972 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
5973 || ae->ae_chop != NULL)
|
|
481
|
5974 && (ae->ae_chop == NULL
|
|
|
5975 || STRLEN(ae->ae_chop) < wordlen)
|
|
339
|
5976 && (ae->ae_prog == NULL
|
|
|
5977 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
5978 {
|
|
|
5979 /* Match. Remove the chop and add the affix. */
|
|
|
5980 if (xht == NULL)
|
|
240
|
5981 {
|
|
300
|
5982 /* prefix: chop/add at the start of the word */
|
|
|
5983 if (ae->ae_add == NULL)
|
|
|
5984 *newword = NUL;
|
|
|
5985 else
|
|
|
5986 STRCPY(newword, ae->ae_add);
|
|
|
5987 p = word;
|
|
|
5988 if (ae->ae_chop != NULL)
|
|
310
|
5989 {
|
|
300
|
5990 /* Skip chop string. */
|
|
310
|
5991 #ifdef FEAT_MBYTE
|
|
|
5992 if (has_mbyte)
|
|
324
|
5993 {
|
|
310
|
5994 i = mb_charlen(ae->ae_chop);
|
|
324
|
5995 for ( ; i > 0; --i)
|
|
|
5996 mb_ptr_adv(p);
|
|
|
5997 }
|
|
310
|
5998 else
|
|
|
5999 #endif
|
|
324
|
6000 p += STRLEN(ae->ae_chop);
|
|
310
|
6001 }
|
|
300
|
6002 STRCAT(newword, p);
|
|
|
6003 }
|
|
|
6004 else
|
|
|
6005 {
|
|
|
6006 /* suffix: chop/add at the end of the word */
|
|
|
6007 STRCPY(newword, word);
|
|
|
6008 if (ae->ae_chop != NULL)
|
|
|
6009 {
|
|
|
6010 /* Remove chop string. */
|
|
|
6011 p = newword + STRLEN(newword);
|
|
376
|
6012 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
6013 for ( ; i > 0; --i)
|
|
300
|
6014 mb_ptr_back(newword, p);
|
|
|
6015 *p = NUL;
|
|
|
6016 }
|
|
|
6017 if (ae->ae_add != NULL)
|
|
|
6018 STRCAT(newword, ae->ae_add);
|
|
240
|
6019 }
|
|
|
6020
|
|
366
|
6021 /* Obey the "rare" flag of the affix. */
|
|
|
6022 if (ae->ae_rare)
|
|
|
6023 use_flags = flags | WF_RARE;
|
|
|
6024 else
|
|
|
6025 use_flags = flags;
|
|
483
|
6026
|
|
|
6027 /* Obey the "nocomp" flag of the affix: don't use the
|
|
|
6028 * compound flags. */
|
|
390
|
6029 use_pfxlist = pfxlist;
|
|
483
|
6030 if (ae->ae_nocomp && pfxlist != NULL)
|
|
|
6031 {
|
|
|
6032 vim_strncpy(pfx_pfxlist, pfxlist, pfxlen);
|
|
|
6033 use_pfxlist = pfx_pfxlist;
|
|
|
6034 }
|
|
390
|
6035
|
|
|
6036 /* When there are postponed prefixes... */
|
|
393
|
6037 if (spin->si_prefroot != NULL
|
|
|
6038 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
6039 {
|
|
|
6040 /* ... add a flag to indicate an affix was used. */
|
|
|
6041 use_flags |= WF_HAS_AFF;
|
|
|
6042
|
|
|
6043 /* ... don't use a prefix list if combining
|
|
483
|
6044 * affixes is not allowed. But do use the
|
|
|
6045 * compound flags after them. */
|
|
|
6046 if ((!ah->ah_combine || comb) && pfxlist != NULL)
|
|
|
6047 use_pfxlist += pfxlen;
|
|
390
|
6048 }
|
|
366
|
6049
|
|
300
|
6050 /* Store the modified word. */
|
|
474
|
6051 if (store_word(spin, newword, use_flags,
|
|
483
|
6052 spin->si_region, use_pfxlist, FALSE) == FAIL)
|
|
300
|
6053 retval = FAIL;
|
|
236
|
6054
|
|
300
|
6055 /* When added a suffix and combining is allowed also
|
|
|
6056 * try adding prefixes additionally. */
|
|
|
6057 if (xht != NULL && ah->ah_combine)
|
|
474
|
6058 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
6059 xht, NULL, TRUE,
|
|
483
|
6060 use_flags, use_pfxlist, pfxlen) == FAIL)
|
|
300
|
6061 retval = FAIL;
|
|
236
|
6062 }
|
|
|
6063 }
|
|
|
6064 }
|
|
|
6065 }
|
|
|
6066 }
|
|
|
6067
|
|
|
6068 return retval;
|
|
|
6069 }
|
|
|
6070
|
|
|
6071 /*
|
|
300
|
6072 * Read a file with a list of words.
|
|
236
|
6073 */
|
|
|
6074 static int
|
|
474
|
6075 spell_read_wordfile(spin, fname)
|
|
|
6076 spellinfo_T *spin;
|
|
300
|
6077 char_u *fname;
|
|
236
|
6078 {
|
|
300
|
6079 FILE *fd;
|
|
|
6080 long lnum = 0;
|
|
|
6081 char_u rline[MAXLINELEN];
|
|
|
6082 char_u *line;
|
|
|
6083 char_u *pc = NULL;
|
|
381
|
6084 char_u *p;
|
|
300
|
6085 int l;
|
|
|
6086 int retval = OK;
|
|
|
6087 int did_word = FALSE;
|
|
|
6088 int non_ascii = 0;
|
|
307
|
6089 int flags;
|
|
316
|
6090 int regionmask;
|
|
236
|
6091
|
|
300
|
6092 /*
|
|
|
6093 * Open the file.
|
|
|
6094 */
|
|
310
|
6095 fd = mch_fopen((char *)fname, "r");
|
|
300
|
6096 if (fd == NULL)
|
|
236
|
6097 {
|
|
300
|
6098 EMSG2(_(e_notopen), fname);
|
|
|
6099 return FAIL;
|
|
236
|
6100 }
|
|
|
6101
|
|
310
|
6102 if (spin->si_verbose || p_verbose > 2)
|
|
|
6103 {
|
|
|
6104 if (!spin->si_verbose)
|
|
|
6105 verbose_enter();
|
|
366
|
6106 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
6107 out_flush();
|
|
|
6108 if (!spin->si_verbose)
|
|
|
6109 verbose_leave();
|
|
|
6110 }
|
|
300
|
6111
|
|
|
6112 /*
|
|
|
6113 * Read all the lines in the file one by one.
|
|
|
6114 */
|
|
|
6115 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
6116 {
|
|
|
6117 line_breakcheck();
|
|
|
6118 ++lnum;
|
|
|
6119
|
|
|
6120 /* Skip comment lines. */
|
|
|
6121 if (*rline == '#')
|
|
|
6122 continue;
|
|
|
6123
|
|
|
6124 /* Remove CR, LF and white space from the end. */
|
|
|
6125 l = STRLEN(rline);
|
|
|
6126 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
6127 --l;
|
|
|
6128 if (l == 0)
|
|
|
6129 continue; /* empty or blank line */
|
|
|
6130 rline[l] = NUL;
|
|
|
6131
|
|
|
6132 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
6133 vim_free(pc);
|
|
310
|
6134 #ifdef FEAT_MBYTE
|
|
300
|
6135 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
6136 {
|
|
|
6137 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
6138 if (pc == NULL)
|
|
|
6139 {
|
|
|
6140 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
6141 fname, lnum, rline);
|
|
|
6142 continue;
|
|
|
6143 }
|
|
|
6144 line = pc;
|
|
|
6145 }
|
|
|
6146 else
|
|
310
|
6147 #endif
|
|
300
|
6148 {
|
|
|
6149 pc = NULL;
|
|
|
6150 line = rline;
|
|
|
6151 }
|
|
|
6152
|
|
307
|
6153 if (*line == '/')
|
|
300
|
6154 {
|
|
307
|
6155 ++line;
|
|
|
6156 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
6157 {
|
|
|
6158 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
6159 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
6160 fname, lnum, line - 1);
|
|
300
|
6161 else if (did_word)
|
|
316
|
6162 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
6163 fname, lnum, line - 1);
|
|
300
|
6164 else
|
|
|
6165 {
|
|
310
|
6166 #ifdef FEAT_MBYTE
|
|
|
6167 char_u *enc;
|
|
|
6168
|
|
300
|
6169 /* Setup for conversion to 'encoding'. */
|
|
316
|
6170 line += 10;
|
|
|
6171 enc = enc_canonize(line);
|
|
300
|
6172 if (enc != NULL && !spin->si_ascii
|
|
|
6173 && convert_setup(&spin->si_conv, enc,
|
|
|
6174 p_enc) == FAIL)
|
|
|
6175 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
6176 fname, line, p_enc);
|
|
300
|
6177 vim_free(enc);
|
|
375
|
6178 spin->si_conv.vc_fail = TRUE;
|
|
310
|
6179 #else
|
|
|
6180 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
6181 #endif
|
|
300
|
6182 }
|
|
307
|
6183 continue;
|
|
300
|
6184 }
|
|
307
|
6185
|
|
316
|
6186 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
6187 {
|
|
|
6188 if (spin->si_region_count > 1)
|
|
|
6189 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
6190 fname, lnum, line);
|
|
|
6191 else
|
|
|
6192 {
|
|
|
6193 line += 8;
|
|
|
6194 if (STRLEN(line) > 16)
|
|
|
6195 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
6196 fname, lnum, line);
|
|
|
6197 else
|
|
|
6198 {
|
|
|
6199 spin->si_region_count = STRLEN(line) / 2;
|
|
|
6200 STRCPY(spin->si_region_name, line);
|
|
389
|
6201
|
|
|
6202 /* Adjust the mask for a word valid in all regions. */
|
|
|
6203 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
6204 }
|
|
|
6205 }
|
|
|
6206 continue;
|
|
|
6207 }
|
|
|
6208
|
|
381
|
6209 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
6210 fname, lnum, line - 1);
|
|
|
6211 continue;
|
|
|
6212 }
|
|
|
6213
|
|
|
6214 flags = 0;
|
|
|
6215 regionmask = spin->si_region;
|
|
|
6216
|
|
|
6217 /* Check for flags and region after a slash. */
|
|
|
6218 p = vim_strchr(line, '/');
|
|
|
6219 if (p != NULL)
|
|
|
6220 {
|
|
|
6221 *p++ = NUL;
|
|
|
6222 while (*p != NUL)
|
|
|
6223 {
|
|
|
6224 if (*p == '=') /* keep-case word */
|
|
389
|
6225 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
6226 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
6227 flags |= WF_BANNED;
|
|
|
6228 else if (*p == '?') /* Rare word. */
|
|
|
6229 flags |= WF_RARE;
|
|
|
6230 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
6231 {
|
|
381
|
6232 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
6233 regionmask = 0;
|
|
|
6234 flags |= WF_REGION;
|
|
|
6235
|
|
|
6236 l = *p - '0';
|
|
316
|
6237 if (l > spin->si_region_count)
|
|
|
6238 {
|
|
|
6239 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
6240 fname, lnum, p);
|
|
316
|
6241 break;
|
|
|
6242 }
|
|
|
6243 regionmask |= 1 << (l - 1);
|
|
|
6244 }
|
|
381
|
6245 else
|
|
|
6246 {
|
|
|
6247 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
6248 fname, lnum, p);
|
|
|
6249 break;
|
|
|
6250 }
|
|
|
6251 ++p;
|
|
307
|
6252 }
|
|
300
|
6253 }
|
|
|
6254
|
|
|
6255 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
6256 if (spin->si_ascii && has_non_ascii(line))
|
|
|
6257 {
|
|
|
6258 ++non_ascii;
|
|
|
6259 continue;
|
|
|
6260 }
|
|
|
6261
|
|
|
6262 /* Normal word: store it. */
|
|
483
|
6263 if (store_word(spin, line, flags, regionmask, NULL, FALSE) == FAIL)
|
|
300
|
6264 {
|
|
|
6265 retval = FAIL;
|
|
|
6266 break;
|
|
|
6267 }
|
|
|
6268 did_word = TRUE;
|
|
|
6269 }
|
|
|
6270
|
|
|
6271 vim_free(pc);
|
|
|
6272 fclose(fd);
|
|
|
6273
|
|
310
|
6274 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
6275 {
|
|
|
6276 if (p_verbose > 2)
|
|
|
6277 verbose_enter();
|
|
300
|
6278 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
6279 non_ascii);
|
|
310
|
6280 if (p_verbose > 2)
|
|
|
6281 verbose_leave();
|
|
|
6282 }
|
|
300
|
6283 return retval;
|
|
236
|
6284 }
|
|
|
6285
|
|
|
6286 /*
|
|
300
|
6287 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
6288 * This avoids calling free() for every little struct we use (and keeping
|
|
|
6289 * track of them).
|
|
300
|
6290 * The memory is cleared to all zeros.
|
|
|
6291 * Returns NULL when out of memory.
|
|
|
6292 */
|
|
|
6293 static void *
|
|
474
|
6294 getroom(spin, len, align)
|
|
|
6295 spellinfo_T *spin;
|
|
447
|
6296 size_t len; /* length needed */
|
|
|
6297 int align; /* align for pointer */
|
|
300
|
6298 {
|
|
|
6299 char_u *p;
|
|
474
|
6300 sblock_T *bl = spin->si_blocks;
|
|
300
|
6301
|
|
447
|
6302 if (align && bl != NULL)
|
|
|
6303 /* Round size up for alignment. On some systems structures need to be
|
|
|
6304 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
6305 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
6306 & ~(sizeof(char *) - 1);
|
|
|
6307
|
|
300
|
6308 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
6309 {
|
|
|
6310 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
6311 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
6312 if (bl == NULL)
|
|
|
6313 return NULL;
|
|
474
|
6314 bl->sb_next = spin->si_blocks;
|
|
|
6315 spin->si_blocks = bl;
|
|
300
|
6316 bl->sb_used = 0;
|
|
482
|
6317 ++spin->si_blocks_cnt;
|
|
300
|
6318 }
|
|
|
6319
|
|
|
6320 p = bl->sb_data + bl->sb_used;
|
|
|
6321 bl->sb_used += len;
|
|
|
6322
|
|
|
6323 return p;
|
|
|
6324 }
|
|
|
6325
|
|
|
6326 /*
|
|
|
6327 * Make a copy of a string into memory allocated with getroom().
|
|
|
6328 */
|
|
|
6329 static char_u *
|
|
474
|
6330 getroom_save(spin, s)
|
|
|
6331 spellinfo_T *spin;
|
|
300
|
6332 char_u *s;
|
|
|
6333 {
|
|
|
6334 char_u *sc;
|
|
|
6335
|
|
474
|
6336 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
6337 if (sc != NULL)
|
|
|
6338 STRCPY(sc, s);
|
|
|
6339 return sc;
|
|
|
6340 }
|
|
|
6341
|
|
|
6342
|
|
|
6343 /*
|
|
|
6344 * Free the list of allocated sblock_T.
|
|
236
|
6345 */
|
|
|
6346 static void
|
|
300
|
6347 free_blocks(bl)
|
|
|
6348 sblock_T *bl;
|
|
236
|
6349 {
|
|
300
|
6350 sblock_T *next;
|
|
236
|
6351
|
|
300
|
6352 while (bl != NULL)
|
|
236
|
6353 {
|
|
300
|
6354 next = bl->sb_next;
|
|
|
6355 vim_free(bl);
|
|
|
6356 bl = next;
|
|
236
|
6357 }
|
|
|
6358 }
|
|
|
6359
|
|
|
6360 /*
|
|
300
|
6361 * Allocate the root of a word tree.
|
|
236
|
6362 */
|
|
300
|
6363 static wordnode_T *
|
|
474
|
6364 wordtree_alloc(spin)
|
|
|
6365 spellinfo_T *spin;
|
|
|
6366 {
|
|
|
6367 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
6368 }
|
|
|
6369
|
|
|
6370 /*
|
|
300
|
6371 * Store a word in the tree(s).
|
|
390
|
6372 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
6373 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
6374 * used to find suggestions.
|
|
300
|
6375 * For a keep-case word also store it in the keep-case tree.
|
|
482
|
6376 * When "pfxlist" is not NULL store the word for each postponed prefix ID and
|
|
|
6377 * compound flag.
|
|
236
|
6378 */
|
|
|
6379 static int
|
|
483
|
6380 store_word(spin, word, flags, region, pfxlist, need_affix)
|
|
474
|
6381 spellinfo_T *spin;
|
|
300
|
6382 char_u *word;
|
|
307
|
6383 int flags; /* extra flags, WF_BANNED */
|
|
316
|
6384 int region; /* supported region(s) */
|
|
339
|
6385 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
483
|
6386 int need_affix; /* only store word with affix ID */
|
|
236
|
6387 {
|
|
300
|
6388 int len = STRLEN(word);
|
|
|
6389 int ct = captype(word, word + len);
|
|
|
6390 char_u foldword[MAXWLEN];
|
|
339
|
6391 int res = OK;
|
|
|
6392 char_u *p;
|
|
236
|
6393
|
|
323
|
6394 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
6395 for (p = pfxlist; res == OK; ++p)
|
|
|
6396 {
|
|
483
|
6397 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6398 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
474
|
6399 region, p == NULL ? 0 : *p);
|
|
339
|
6400 if (p == NULL || *p == NUL)
|
|
|
6401 break;
|
|
|
6402 }
|
|
334
|
6403 ++spin->si_foldwcount;
|
|
307
|
6404
|
|
474
|
6405 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
6406 {
|
|
339
|
6407 for (p = pfxlist; res == OK; ++p)
|
|
|
6408 {
|
|
483
|
6409 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6410 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
474
|
6411 region, p == NULL ? 0 : *p);
|
|
339
|
6412 if (p == NULL || *p == NUL)
|
|
|
6413 break;
|
|
|
6414 }
|
|
334
|
6415 ++spin->si_keepwcount;
|
|
|
6416 }
|
|
300
|
6417 return res;
|
|
236
|
6418 }
|
|
|
6419
|
|
|
6420 /*
|
|
300
|
6421 * Add word "word" to a word tree at "root".
|
|
366
|
6422 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
6423 * "rare" and "region" is the condition nr.
|
|
255
|
6424 * Returns FAIL when out of memory.
|
|
236
|
6425 */
|
|
255
|
6426 static int
|
|
481
|
6427 tree_add_word(spin, word, root, flags, region, affixID)
|
|
474
|
6428 spellinfo_T *spin;
|
|
300
|
6429 char_u *word;
|
|
|
6430 wordnode_T *root;
|
|
|
6431 int flags;
|
|
|
6432 int region;
|
|
481
|
6433 int affixID;
|
|
236
|
6434 {
|
|
300
|
6435 wordnode_T *node = root;
|
|
|
6436 wordnode_T *np;
|
|
469
|
6437 wordnode_T *copyp, **copyprev;
|
|
300
|
6438 wordnode_T **prev = NULL;
|
|
|
6439 int i;
|
|
255
|
6440
|
|
300
|
6441 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
6442 for (i = 0; ; ++i)
|
|
255
|
6443 {
|
|
469
|
6444 /* When there is more than one reference to this node we need to make
|
|
|
6445 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
6446 * (we don't optimize for a partly shared list of siblings). */
|
|
|
6447 if (node != NULL && node->wn_refs > 1)
|
|
|
6448 {
|
|
|
6449 --node->wn_refs;
|
|
|
6450 copyprev = prev;
|
|
|
6451 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
6452 {
|
|
|
6453 /* Allocate a new node and copy the info. */
|
|
474
|
6454 np = get_wordnode(spin);
|
|
469
|
6455 if (np == NULL)
|
|
|
6456 return FAIL;
|
|
|
6457 np->wn_child = copyp->wn_child;
|
|
|
6458 if (np->wn_child != NULL)
|
|
|
6459 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
6460 np->wn_byte = copyp->wn_byte;
|
|
|
6461 if (np->wn_byte == NUL)
|
|
|
6462 {
|
|
|
6463 np->wn_flags = copyp->wn_flags;
|
|
|
6464 np->wn_region = copyp->wn_region;
|
|
481
|
6465 np->wn_affixID = copyp->wn_affixID;
|
|
469
|
6466 }
|
|
|
6467
|
|
|
6468 /* Link the new node in the list, there will be one ref. */
|
|
|
6469 np->wn_refs = 1;
|
|
|
6470 *copyprev = np;
|
|
|
6471 copyprev = &np->wn_sibling;
|
|
|
6472
|
|
|
6473 /* Let "node" point to the head of the copied list. */
|
|
|
6474 if (copyp == node)
|
|
|
6475 node = np;
|
|
|
6476 }
|
|
|
6477 }
|
|
|
6478
|
|
300
|
6479 /* Look for the sibling that has the same character. They are sorted
|
|
|
6480 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
6481 * higher byte value. For zero bytes (end of word) the sorting is
|
|
481
|
6482 * done on flags and then on affixID. */
|
|
339
|
6483 while (node != NULL
|
|
|
6484 && (node->wn_byte < word[i]
|
|
|
6485 || (node->wn_byte == NUL
|
|
|
6486 && (flags < 0
|
|
481
|
6487 ? node->wn_affixID < affixID
|
|
390
|
6488 : node->wn_flags < (flags & WN_MASK)
|
|
|
6489 || (node->wn_flags == (flags & WN_MASK)
|
|
481
|
6490 && node->wn_affixID < affixID)))))
|
|
236
|
6491 {
|
|
300
|
6492 prev = &node->wn_sibling;
|
|
|
6493 node = *prev;
|
|
236
|
6494 }
|
|
339
|
6495 if (node == NULL
|
|
|
6496 || node->wn_byte != word[i]
|
|
|
6497 || (word[i] == NUL
|
|
|
6498 && (flags < 0
|
|
390
|
6499 || node->wn_flags != (flags & WN_MASK)
|
|
481
|
6500 || node->wn_affixID != affixID)))
|
|
255
|
6501 {
|
|
300
|
6502 /* Allocate a new node. */
|
|
474
|
6503 np = get_wordnode(spin);
|
|
300
|
6504 if (np == NULL)
|
|
|
6505 return FAIL;
|
|
|
6506 np->wn_byte = word[i];
|
|
469
|
6507
|
|
|
6508 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
6509 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
6510 * make ref count of sibling one (matters when inserting in front
|
|
|
6511 * of the list of siblings). */
|
|
|
6512 if (node == NULL)
|
|
|
6513 np->wn_refs = 1;
|
|
|
6514 else
|
|
|
6515 {
|
|
|
6516 np->wn_refs = node->wn_refs;
|
|
|
6517 node->wn_refs = 1;
|
|
|
6518 }
|
|
300
|
6519 *prev = np;
|
|
|
6520 np->wn_sibling = node;
|
|
|
6521 node = np;
|
|
255
|
6522 }
|
|
300
|
6523
|
|
|
6524 if (word[i] == NUL)
|
|
|
6525 {
|
|
|
6526 node->wn_flags = flags;
|
|
|
6527 node->wn_region |= region;
|
|
481
|
6528 node->wn_affixID = affixID;
|
|
300
|
6529 break;
|
|
|
6530 }
|
|
|
6531 prev = &node->wn_child;
|
|
|
6532 node = *prev;
|
|
255
|
6533 }
|
|
469
|
6534 #ifdef SPELL_PRINTTREE
|
|
|
6535 smsg("Added \"%s\"", word);
|
|
|
6536 spell_print_tree(root->wn_sibling);
|
|
|
6537 #endif
|
|
|
6538
|
|
474
|
6539 /* count nr of words added since last message */
|
|
|
6540 ++spin->si_msg_count;
|
|
|
6541
|
|
482
|
6542 if (spin->si_compress_cnt > 1)
|
|
|
6543 {
|
|
|
6544 if (--spin->si_compress_cnt == 1)
|
|
|
6545 /* Did enough words to lower the block count limit. */
|
|
483
|
6546 spin->si_blocks_cnt += compress_inc;
|
|
482
|
6547 }
|
|
|
6548
|
|
469
|
6549 /*
|
|
482
|
6550 * When we have allocated lots of memory we need to compress the word tree
|
|
|
6551 * to free up some room. But compression is slow, and we might actually
|
|
|
6552 * need that room, thus only compress in the following situations:
|
|
|
6553 * 1. When not compressed before (si_compress_cnt == 0): when using
|
|
483
|
6554 * "compress_start" blocks.
|
|
|
6555 * 2. When compressed before and used "compress_inc" blocks before
|
|
|
6556 * adding "compress_added" words (si_compress_cnt > 1).
|
|
|
6557 * 3. When compressed before, added "compress_added" words
|
|
482
|
6558 * (si_compress_cnt == 1) and the number of free nodes drops below the
|
|
|
6559 * maximum word length.
|
|
469
|
6560 */
|
|
482
|
6561 #ifndef SPELL_PRINTTREE
|
|
|
6562 if (spin->si_compress_cnt == 1
|
|
|
6563 ? spin->si_free_count < MAXWLEN
|
|
483
|
6564 : spin->si_blocks_cnt >= compress_start)
|
|
482
|
6565 #endif
|
|
|
6566 {
|
|
|
6567 /* Decrement the block counter. The effect is that we compress again
|
|
483
|
6568 * when the freed up room has been used and another "compress_inc"
|
|
|
6569 * blocks have been allocated. Unless "compress_added" words have
|
|
482
|
6570 * been added, then the limit is put back again. */
|
|
483
|
6571 spin->si_blocks_cnt -= compress_inc;
|
|
|
6572 spin->si_compress_cnt = compress_added;
|
|
482
|
6573
|
|
474
|
6574 if (spin->si_verbose)
|
|
|
6575 {
|
|
|
6576 msg_start();
|
|
|
6577 msg_puts((char_u *)_(msg_compressing));
|
|
|
6578 msg_clr_eos();
|
|
|
6579 msg_didout = FALSE;
|
|
|
6580 msg_col = 0;
|
|
|
6581 out_flush();
|
|
|
6582 }
|
|
482
|
6583
|
|
|
6584 /* Compress both trees. Either they both have many nodes, which makes
|
|
|
6585 * compression useful, or one of them is small, which means
|
|
|
6586 * compression goes fast. */
|
|
|
6587 wordtree_compress(spin, spin->si_foldroot);
|
|
|
6588 wordtree_compress(spin, spin->si_keeproot);
|
|
469
|
6589 }
|
|
255
|
6590
|
|
|
6591 return OK;
|
|
236
|
6592 }
|
|
|
6593
|
|
469
|
6594 /*
|
|
483
|
6595 * Check the 'mkspellmem' option. Return FAIL if it's wrong.
|
|
|
6596 * Sets "sps_flags".
|
|
|
6597 */
|
|
|
6598 int
|
|
|
6599 spell_check_msm()
|
|
|
6600 {
|
|
|
6601 char_u *p = p_msm;
|
|
|
6602 long start = 0;
|
|
|
6603 long inc = 0;
|
|
|
6604 long added = 0;
|
|
|
6605
|
|
|
6606 if (!VIM_ISDIGIT(*p))
|
|
|
6607 return FAIL;
|
|
|
6608 /* block count = (value * 1024) / SBLOCKSIZE (but avoid overflow)*/
|
|
|
6609 start = (getdigits(&p) * 10) / (SBLOCKSIZE / 102);
|
|
|
6610 if (*p != ',')
|
|
|
6611 return FAIL;
|
|
|
6612 ++p;
|
|
|
6613 if (!VIM_ISDIGIT(*p))
|
|
|
6614 return FAIL;
|
|
|
6615 inc = (getdigits(&p) * 102) / (SBLOCKSIZE / 10);
|
|
|
6616 if (*p != ',')
|
|
|
6617 return FAIL;
|
|
|
6618 ++p;
|
|
|
6619 if (!VIM_ISDIGIT(*p))
|
|
|
6620 return FAIL;
|
|
|
6621 added = getdigits(&p) * 1024;
|
|
|
6622 if (*p != NUL)
|
|
|
6623 return FAIL;
|
|
|
6624
|
|
|
6625 if (start == 0 || inc == 0 || added == 0 || inc > start)
|
|
|
6626 return FAIL;
|
|
|
6627
|
|
|
6628 compress_start = start;
|
|
|
6629 compress_inc = inc;
|
|
|
6630 compress_added = added;
|
|
|
6631 return OK;
|
|
|
6632 }
|
|
|
6633
|
|
|
6634
|
|
|
6635 /*
|
|
469
|
6636 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
6637 * allocate a new one.
|
|
|
6638 */
|
|
|
6639 static wordnode_T *
|
|
474
|
6640 get_wordnode(spin)
|
|
|
6641 spellinfo_T *spin;
|
|
469
|
6642 {
|
|
|
6643 wordnode_T *n;
|
|
|
6644
|
|
474
|
6645 if (spin->si_first_free == NULL)
|
|
|
6646 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
6647 else
|
|
|
6648 {
|
|
474
|
6649 n = spin->si_first_free;
|
|
|
6650 spin->si_first_free = n->wn_child;
|
|
469
|
6651 vim_memset(n, 0, sizeof(wordnode_T));
|
|
482
|
6652 --spin->si_free_count;
|
|
469
|
6653 }
|
|
|
6654 #ifdef SPELL_PRINTTREE
|
|
474
|
6655 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
6656 #endif
|
|
|
6657 return n;
|
|
|
6658 }
|
|
|
6659
|
|
|
6660 /*
|
|
474
|
6661 * Decrement the reference count on a node (which is the head of a list of
|
|
|
6662 * siblings). If the reference count becomes zero free the node and its
|
|
|
6663 * siblings.
|
|
|
6664 */
|
|
|
6665 static void
|
|
|
6666 deref_wordnode(spin, node)
|
|
|
6667 spellinfo_T *spin;
|
|
|
6668 wordnode_T *node;
|
|
|
6669 {
|
|
|
6670 wordnode_T *np;
|
|
|
6671
|
|
|
6672 if (--node->wn_refs == 0)
|
|
|
6673 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6674 {
|
|
|
6675 if (np->wn_child != NULL)
|
|
|
6676 deref_wordnode(spin, np->wn_child);
|
|
|
6677 free_wordnode(spin, np);
|
|
|
6678 }
|
|
|
6679 }
|
|
|
6680
|
|
|
6681 /*
|
|
469
|
6682 * Free a wordnode_T for re-use later.
|
|
474
|
6683 * Only the "wn_child" field becomes invalid.
|
|
469
|
6684 */
|
|
|
6685 static void
|
|
474
|
6686 free_wordnode(spin, n)
|
|
|
6687 spellinfo_T *spin;
|
|
|
6688 wordnode_T *n;
|
|
|
6689 {
|
|
|
6690 n->wn_child = spin->si_first_free;
|
|
|
6691 spin->si_first_free = n;
|
|
482
|
6692 ++spin->si_free_count;
|
|
469
|
6693 }
|
|
|
6694
|
|
236
|
6695 /*
|
|
300
|
6696 * Compress a tree: find tails that are identical and can be shared.
|
|
|
6697 */
|
|
|
6698 static void
|
|
474
|
6699 wordtree_compress(spin, root)
|
|
|
6700 spellinfo_T *spin;
|
|
300
|
6701 wordnode_T *root;
|
|
|
6702 {
|
|
|
6703 hashtab_T ht;
|
|
|
6704 int n;
|
|
|
6705 int tot = 0;
|
|
469
|
6706 int perc;
|
|
300
|
6707
|
|
474
|
6708 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
6709 * start of the tree. */
|
|
|
6710 if (root->wn_sibling != NULL)
|
|
300
|
6711 {
|
|
|
6712 hash_init(&ht);
|
|
474
|
6713 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
6714
|
|
|
6715 #ifndef SPELL_PRINTTREE
|
|
310
|
6716 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
6717 #endif
|
|
310
|
6718 {
|
|
|
6719 if (!spin->si_verbose)
|
|
|
6720 verbose_enter();
|
|
469
|
6721 if (tot > 1000000)
|
|
|
6722 perc = (tot - n) / (tot / 100);
|
|
482
|
6723 else if (tot == 0)
|
|
|
6724 perc = 0;
|
|
469
|
6725 else
|
|
|
6726 perc = (tot - n) * 100 / tot;
|
|
310
|
6727 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
6728 n, tot, perc);
|
|
310
|
6729 if (p_verbose > 2)
|
|
|
6730 verbose_leave();
|
|
|
6731 }
|
|
469
|
6732 #ifdef SPELL_PRINTTREE
|
|
474
|
6733 spell_print_tree(root->wn_sibling);
|
|
469
|
6734 #endif
|
|
300
|
6735 hash_clear(&ht);
|
|
|
6736 }
|
|
|
6737 }
|
|
|
6738
|
|
|
6739 /*
|
|
|
6740 * Compress a node, its siblings and its children, depth first.
|
|
|
6741 * Returns the number of compressed nodes.
|
|
236
|
6742 */
|
|
255
|
6743 static int
|
|
474
|
6744 node_compress(spin, node, ht, tot)
|
|
|
6745 spellinfo_T *spin;
|
|
300
|
6746 wordnode_T *node;
|
|
|
6747 hashtab_T *ht;
|
|
|
6748 int *tot; /* total count of nodes before compressing,
|
|
|
6749 incremented while going through the tree */
|
|
236
|
6750 {
|
|
300
|
6751 wordnode_T *np;
|
|
|
6752 wordnode_T *tp;
|
|
|
6753 wordnode_T *child;
|
|
|
6754 hash_T hash;
|
|
236
|
6755 hashitem_T *hi;
|
|
300
|
6756 int len = 0;
|
|
|
6757 unsigned nr, n;
|
|
|
6758 int compressed = 0;
|
|
236
|
6759
|
|
300
|
6760 /*
|
|
|
6761 * Go through the list of siblings. Compress each child and then try
|
|
|
6762 * finding an identical child to replace it.
|
|
|
6763 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
6764 * but the whole list of siblings, of which the node is the first.
|
|
|
6765 */
|
|
482
|
6766 for (np = node; np != NULL && !got_int; np = np->wn_sibling)
|
|
236
|
6767 {
|
|
300
|
6768 ++len;
|
|
|
6769 if ((child = np->wn_child) != NULL)
|
|
|
6770 {
|
|
346
|
6771 /* Compress the child. This fills hashkey. */
|
|
474
|
6772 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
6773
|
|
|
6774 /* Try to find an identical child. */
|
|
346
|
6775 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
6776 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
6777 tp = NULL;
|
|
|
6778 if (!HASHITEM_EMPTY(hi))
|
|
|
6779 {
|
|
|
6780 /* There are children with an identical hash value. Now check
|
|
|
6781 * if there is one that is really identical. */
|
|
346
|
6782 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
6783 if (node_equal(child, tp))
|
|
|
6784 {
|
|
|
6785 /* Found one! Now use that child in place of the
|
|
469
|
6786 * current one. This means the current child and all
|
|
|
6787 * its siblings is unlinked from the tree. */
|
|
|
6788 ++tp->wn_refs;
|
|
474
|
6789 deref_wordnode(spin, child);
|
|
300
|
6790 np->wn_child = tp;
|
|
|
6791 ++compressed;
|
|
|
6792 break;
|
|
|
6793 }
|
|
|
6794 if (tp == NULL)
|
|
|
6795 {
|
|
|
6796 /* No other child with this hash value equals the child of
|
|
|
6797 * the node, add it to the linked list after the first
|
|
|
6798 * item. */
|
|
|
6799 tp = HI2WN(hi);
|
|
346
|
6800 child->wn_u2.next = tp->wn_u2.next;
|
|
|
6801 tp->wn_u2.next = child;
|
|
300
|
6802 }
|
|
|
6803 }
|
|
|
6804 else
|
|
|
6805 /* No other child has this hash value, add it to the
|
|
|
6806 * hashtable. */
|
|
346
|
6807 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
6808 }
|
|
236
|
6809 }
|
|
300
|
6810 *tot += len;
|
|
|
6811
|
|
|
6812 /*
|
|
|
6813 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
6814 * find a lookalike node. This must be done after compressing the sibling
|
|
|
6815 * list, otherwise the hash key would become invalid by the compression.
|
|
|
6816 */
|
|
346
|
6817 node->wn_u1.hashkey[0] = len;
|
|
300
|
6818 nr = 0;
|
|
|
6819 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
6820 {
|
|
300
|
6821 if (np->wn_byte == NUL)
|
|
481
|
6822 /* end node: use wn_flags, wn_region and wn_affixID */
|
|
|
6823 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16);
|
|
300
|
6824 else
|
|
|
6825 /* byte node: use the byte value and the child pointer */
|
|
|
6826 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
6827 nr = nr * 101 + n;
|
|
236
|
6828 }
|
|
300
|
6829
|
|
|
6830 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
6831 n = nr & 0xff;
|
|
346
|
6832 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
6833 n = (nr >> 8) & 0xff;
|
|
346
|
6834 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
6835 n = (nr >> 16) & 0xff;
|
|
346
|
6836 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
6837 n = (nr >> 24) & 0xff;
|
|
346
|
6838 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
6839 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
6840
|
|
482
|
6841 /* Check for CTRL-C pressed now and then. */
|
|
|
6842 fast_breakcheck();
|
|
|
6843
|
|
300
|
6844 return compressed;
|
|
|
6845 }
|
|
|
6846
|
|
|
6847 /*
|
|
|
6848 * Return TRUE when two nodes have identical siblings and children.
|
|
|
6849 */
|
|
|
6850 static int
|
|
|
6851 node_equal(n1, n2)
|
|
|
6852 wordnode_T *n1;
|
|
|
6853 wordnode_T *n2;
|
|
|
6854 {
|
|
|
6855 wordnode_T *p1;
|
|
|
6856 wordnode_T *p2;
|
|
|
6857
|
|
|
6858 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
6859 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
6860 if (p1->wn_byte != p2->wn_byte
|
|
|
6861 || (p1->wn_byte == NUL
|
|
|
6862 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
6863 || p1->wn_region != p2->wn_region
|
|
481
|
6864 || p1->wn_affixID != p2->wn_affixID)
|
|
300
|
6865 : (p1->wn_child != p2->wn_child)))
|
|
|
6866 break;
|
|
|
6867
|
|
|
6868 return p1 == NULL && p2 == NULL;
|
|
236
|
6869 }
|
|
|
6870
|
|
|
6871 /*
|
|
|
6872 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
6873 */
|
|
255
|
6874 void
|
|
236
|
6875 put_bytes(fd, nr, len)
|
|
|
6876 FILE *fd;
|
|
|
6877 long_u nr;
|
|
|
6878 int len;
|
|
|
6879 {
|
|
|
6880 int i;
|
|
|
6881
|
|
|
6882 for (i = len - 1; i >= 0; --i)
|
|
|
6883 putc((int)(nr >> (i * 8)), fd);
|
|
|
6884 }
|
|
|
6885
|
|
323
|
6886 static int
|
|
|
6887 #ifdef __BORLANDC__
|
|
|
6888 _RTLENTRYF
|
|
|
6889 #endif
|
|
|
6890 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
6891
|
|
|
6892 /*
|
|
|
6893 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
6894 */
|
|
|
6895 static int
|
|
|
6896 #ifdef __BORLANDC__
|
|
|
6897 _RTLENTRYF
|
|
|
6898 #endif
|
|
|
6899 rep_compare(s1, s2)
|
|
|
6900 const void *s1;
|
|
|
6901 const void *s2;
|
|
|
6902 {
|
|
|
6903 fromto_T *p1 = (fromto_T *)s1;
|
|
|
6904 fromto_T *p2 = (fromto_T *)s2;
|
|
|
6905
|
|
|
6906 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
6907 }
|
|
|
6908
|
|
236
|
6909 /*
|
|
483
|
6910 * Write the Vim .spl file "fname".
|
|
500
|
6911 * Return FAIL or OK;
|
|
|
6912 */
|
|
|
6913 static int
|
|
474
|
6914 write_vim_spell(spin, fname)
|
|
|
6915 spellinfo_T *spin;
|
|
236
|
6916 char_u *fname;
|
|
|
6917 {
|
|
300
|
6918 FILE *fd;
|
|
|
6919 int regionmask;
|
|
236
|
6920 int round;
|
|
300
|
6921 wordnode_T *tree;
|
|
|
6922 int nodecount;
|
|
323
|
6923 int i;
|
|
|
6924 int l;
|
|
|
6925 garray_T *gap;
|
|
|
6926 fromto_T *ftp;
|
|
|
6927 char_u *p;
|
|
|
6928 int rr;
|
|
500
|
6929 int retval = OK;
|
|
236
|
6930
|
|
310
|
6931 fd = mch_fopen((char *)fname, "w");
|
|
300
|
6932 if (fd == NULL)
|
|
236
|
6933 {
|
|
|
6934 EMSG2(_(e_notopen), fname);
|
|
500
|
6935 return FAIL;
|
|
236
|
6936 }
|
|
|
6937
|
|
483
|
6938 /* <HEADER>: <fileID> <versionnr> */
|
|
300
|
6939 /* <fileID> */
|
|
|
6940 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
500
|
6941 {
|
|
300
|
6942 EMSG(_(e_write));
|
|
500
|
6943 retval = FAIL;
|
|
|
6944 }
|
|
483
|
6945 putc(VIMSPELLVERSION, fd); /* <versionnr> */
|
|
|
6946
|
|
|
6947 /*
|
|
|
6948 * <SECTIONS>: <section> ... <sectionend>
|
|
|
6949 */
|
|
|
6950
|
|
|
6951 /* SN_REGION: <regionname> ...
|
|
|
6952 * Write the region names only if there is more than one. */
|
|
316
|
6953 if (spin->si_region_count > 1)
|
|
236
|
6954 {
|
|
483
|
6955 putc(SN_REGION, fd); /* <sectionID> */
|
|
|
6956 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6957 l = spin->si_region_count * 2;
|
|
|
6958 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6959 fwrite(spin->si_region_name, (size_t)l, (size_t)1, fd);
|
|
|
6960 /* <regionname> ... */
|
|
316
|
6961 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
6962 }
|
|
|
6963 else
|
|
300
|
6964 regionmask = 0;
|
|
483
|
6965
|
|
|
6966 /* SN_CHARFLAGS: <charflagslen> <charflags> <folcharslen> <folchars>
|
|
|
6967 *
|
|
|
6968 * The table with character flags and the table for case folding.
|
|
|
6969 * This makes sure the same characters are recognized as word characters
|
|
|
6970 * when generating an when using a spell file.
|
|
260
|
6971 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
6972 * 'encoding'.
|
|
|
6973 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
6974 * the table (avoids that it conflicts). File is shorter too.
|
|
|
6975 */
|
|
483
|
6976 if (!spin->si_ascii && !spin->si_add)
|
|
|
6977 {
|
|
|
6978 char_u folchars[128 * 8];
|
|
|
6979 int flags;
|
|
|
6980
|
|
485
|
6981 putc(SN_CHARFLAGS, fd); /* <sectionID> */
|
|
483
|
6982 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6983
|
|
|
6984 /* Form the <folchars> string first, we need to know its length. */
|
|
|
6985 l = 0;
|
|
|
6986 for (i = 128; i < 256; ++i)
|
|
|
6987 {
|
|
|
6988 #ifdef FEAT_MBYTE
|
|
|
6989 if (has_mbyte)
|
|
|
6990 l += mb_char2bytes(spelltab.st_fold[i], folchars + l);
|
|
|
6991 else
|
|
|
6992 #endif
|
|
|
6993 folchars[l++] = spelltab.st_fold[i];
|
|
|
6994 }
|
|
|
6995 put_bytes(fd, (long_u)(1 + 128 + 2 + l), 4); /* <sectionlen> */
|
|
|
6996
|
|
|
6997 fputc(128, fd); /* <charflagslen> */
|
|
|
6998 for (i = 128; i < 256; ++i)
|
|
|
6999 {
|
|
|
7000 flags = 0;
|
|
|
7001 if (spelltab.st_isw[i])
|
|
|
7002 flags |= CF_WORD;
|
|
|
7003 if (spelltab.st_isu[i])
|
|
|
7004 flags |= CF_UPPER;
|
|
|
7005 fputc(flags, fd); /* <charflags> */
|
|
|
7006 }
|
|
|
7007
|
|
|
7008 put_bytes(fd, (long_u)l, 2); /* <folcharslen> */
|
|
|
7009 fwrite(folchars, (size_t)l, (size_t)1, fd); /* <folchars> */
|
|
|
7010 }
|
|
|
7011
|
|
|
7012 /* SN_MIDWORD: <midword> */
|
|
|
7013 if (spin->si_midword != NULL)
|
|
|
7014 {
|
|
|
7015 putc(SN_MIDWORD, fd); /* <sectionID> */
|
|
|
7016 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7017
|
|
366
|
7018 i = STRLEN(spin->si_midword);
|
|
483
|
7019 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */
|
|
366
|
7020 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
7021 }
|
|
|
7022
|
|
483
|
7023 /* SN_PREFCOND: <prefcondcnt> <prefcond> ... */
|
|
|
7024 if (spin->si_prefcond.ga_len > 0)
|
|
|
7025 {
|
|
|
7026 putc(SN_PREFCOND, fd); /* <sectionID> */
|
|
|
7027 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7028
|
|
|
7029 l = write_spell_prefcond(NULL, &spin->si_prefcond);
|
|
|
7030 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7031
|
|
|
7032 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
7033 }
|
|
|
7034
|
|
|
7035 /* SN_REP: <repcount> <rep> ...
|
|
|
7036 * SN_SAL: <salflags> <salcount> <sal> ... */
|
|
366
|
7037
|
|
323
|
7038 /* Sort the REP items. */
|
|
|
7039 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
7040 sizeof(fromto_T), rep_compare);
|
|
|
7041
|
|
483
|
7042 /* round 1: SN_REP section
|
|
|
7043 * round 2: SN_SAL section (unless SN_SOFO is used) */
|
|
323
|
7044 for (round = 1; round <= 2; ++round)
|
|
|
7045 {
|
|
|
7046 if (round == 1)
|
|
483
|
7047 {
|
|
323
|
7048 gap = &spin->si_rep;
|
|
483
|
7049 putc(SN_REP, fd); /* <sectionID> */
|
|
|
7050 }
|
|
323
|
7051 else
|
|
|
7052 {
|
|
483
|
7053 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
7054 /* using SN_SOFO section instead of SN_SAL */
|
|
|
7055 break;
|
|
323
|
7056 gap = &spin->si_sal;
|
|
483
|
7057 putc(SN_SAL, fd); /* <sectionID> */
|
|
|
7058 }
|
|
|
7059
|
|
|
7060 /* This is for making suggestions, section is not required. */
|
|
|
7061 putc(0, fd); /* <sectionflags> */
|
|
|
7062
|
|
|
7063 /* Compute the length of what follows. */
|
|
|
7064 l = 2; /* count <repcount> or <salcount> */
|
|
|
7065 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7066 {
|
|
|
7067 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
7068 l += 1 + STRLEN(ftp->ft_from); /* count <*fromlen> and <*from> */
|
|
|
7069 l += 1 + STRLEN(ftp->ft_to); /* count <*tolen> and <*to> */
|
|
|
7070 }
|
|
|
7071 if (round == 2)
|
|
|
7072 ++l; /* count <salflags> */
|
|
|
7073 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7074
|
|
|
7075 if (round == 2)
|
|
|
7076 {
|
|
323
|
7077 i = 0;
|
|
|
7078 if (spin->si_followup)
|
|
|
7079 i |= SAL_F0LLOWUP;
|
|
|
7080 if (spin->si_collapse)
|
|
|
7081 i |= SAL_COLLAPSE;
|
|
|
7082 if (spin->si_rem_accents)
|
|
|
7083 i |= SAL_REM_ACCENTS;
|
|
|
7084 putc(i, fd); /* <salflags> */
|
|
|
7085 }
|
|
|
7086
|
|
|
7087 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
7088 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7089 {
|
|
|
7090 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
7091 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
7092 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
7093 for (rr = 1; rr <= 2; ++rr)
|
|
|
7094 {
|
|
|
7095 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
7096 l = STRLEN(p);
|
|
|
7097 putc(l, fd);
|
|
|
7098 fwrite(p, l, (size_t)1, fd);
|
|
|
7099 }
|
|
|
7100 }
|
|
483
|
7101
|
|
|
7102 }
|
|
|
7103
|
|
|
7104 /* SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
7105 * This is for making suggestions, section is not required. */
|
|
375
|
7106 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
7107 {
|
|
483
|
7108 putc(SN_SOFO, fd); /* <sectionID> */
|
|
|
7109 putc(0, fd); /* <sectionflags> */
|
|
375
|
7110
|
|
|
7111 l = STRLEN(spin->si_sofofr);
|
|
483
|
7112 put_bytes(fd, (long_u)(l + STRLEN(spin->si_sofoto) + 4), 4);
|
|
|
7113 /* <sectionlen> */
|
|
|
7114
|
|
|
7115 put_bytes(fd, (long_u)l, 2); /* <sofofromlen> */
|
|
|
7116 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <sofofrom> */
|
|
375
|
7117
|
|
|
7118 l = STRLEN(spin->si_sofoto);
|
|
483
|
7119 put_bytes(fd, (long_u)l, 2); /* <sofotolen> */
|
|
|
7120 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <sofoto> */
|
|
|
7121 }
|
|
|
7122
|
|
|
7123 /* SN_MAP: <mapstr>
|
|
|
7124 * This is for making suggestions, section is not required. */
|
|
|
7125 if (spin->si_map.ga_len > 0)
|
|
|
7126 {
|
|
|
7127 putc(SN_MAP, fd); /* <sectionID> */
|
|
|
7128 putc(0, fd); /* <sectionflags> */
|
|
|
7129 l = spin->si_map.ga_len;
|
|
|
7130 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7131 fwrite(spin->si_map.ga_data, (size_t)l, (size_t)1, fd);
|
|
|
7132 /* <mapstr> */
|
|
|
7133 }
|
|
|
7134
|
|
|
7135 /* SN_COMPOUND: compound info.
|
|
|
7136 * We don't mark it required, when not supported all compound words will
|
|
|
7137 * be bad words. */
|
|
|
7138 if (spin->si_compflags != NULL)
|
|
|
7139 {
|
|
|
7140 putc(SN_COMPOUND, fd); /* <sectionID> */
|
|
|
7141 putc(0, fd); /* <sectionflags> */
|
|
|
7142
|
|
|
7143 l = STRLEN(spin->si_compflags);
|
|
|
7144 put_bytes(fd, (long_u)(l + 3), 4); /* <sectionlen> */
|
|
|
7145 putc(spin->si_compmax, fd); /* <compmax> */
|
|
|
7146 putc(spin->si_compminlen, fd); /* <compminlen> */
|
|
|
7147 putc(spin->si_compsylmax, fd); /* <compsylmax> */
|
|
|
7148 /* <compflags> */
|
|
|
7149 fwrite(spin->si_compflags, (size_t)l, (size_t)1, fd);
|
|
|
7150 }
|
|
|
7151
|
|
492
|
7152 /* SN_NOBREAK: NOBREAK flag */
|
|
|
7153 if (spin->si_nobreak)
|
|
|
7154 {
|
|
|
7155 putc(SN_NOBREAK, fd); /* <sectionID> */
|
|
|
7156 putc(0, fd); /* <sectionflags> */
|
|
|
7157
|
|
|
7158 /* It's empty, the precense of the section flags the feature. */
|
|
|
7159 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */
|
|
|
7160 }
|
|
|
7161
|
|
483
|
7162 /* SN_SYLLABLE: syllable info.
|
|
|
7163 * We don't mark it required, when not supported syllables will not be
|
|
|
7164 * counted. */
|
|
|
7165 if (spin->si_syllable != NULL)
|
|
|
7166 {
|
|
|
7167 putc(SN_SYLLABLE, fd); /* <sectionID> */
|
|
|
7168 putc(0, fd); /* <sectionflags> */
|
|
|
7169
|
|
|
7170 l = STRLEN(spin->si_syllable);
|
|
|
7171 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7172 fwrite(spin->si_syllable, (size_t)l, (size_t)1, fd); /* <syllable> */
|
|
|
7173 }
|
|
|
7174
|
|
|
7175 /* end of <SECTIONS> */
|
|
|
7176 putc(SN_END, fd); /* <sectionend> */
|
|
|
7177
|
|
302
|
7178
|
|
236
|
7179 /*
|
|
339
|
7180 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
7181 */
|
|
323
|
7182 spin->si_memtot = 0;
|
|
339
|
7183 for (round = 1; round <= 3; ++round)
|
|
236
|
7184 {
|
|
339
|
7185 if (round == 1)
|
|
469
|
7186 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
7187 else if (round == 2)
|
|
469
|
7188 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
7189 else
|
|
469
|
7190 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
7191
|
|
346
|
7192 /* Clear the index and wnode fields in the tree. */
|
|
|
7193 clear_node(tree);
|
|
|
7194
|
|
300
|
7195 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
7196 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
7197 * nodes. */
|
|
346
|
7198 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
7199
|
|
300
|
7200 /* number of nodes in 4 bytes */
|
|
|
7201 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
7202 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
7203
|
|
300
|
7204 /* Write the nodes. */
|
|
346
|
7205 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
7206 }
|
|
|
7207
|
|
500
|
7208 /* Write another byte to check for errors. */
|
|
|
7209 if (putc(0, fd) == EOF)
|
|
|
7210 retval = FAIL;
|
|
|
7211
|
|
|
7212 if (fclose(fd) == EOF)
|
|
|
7213 retval = FAIL;
|
|
|
7214
|
|
|
7215 return retval;
|
|
236
|
7216 }
|
|
|
7217
|
|
|
7218 /*
|
|
346
|
7219 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
7220 * children. This is needed because they are a union with other items to save
|
|
|
7221 * space.
|
|
|
7222 */
|
|
|
7223 static void
|
|
|
7224 clear_node(node)
|
|
|
7225 wordnode_T *node;
|
|
|
7226 {
|
|
|
7227 wordnode_T *np;
|
|
|
7228
|
|
|
7229 if (node != NULL)
|
|
|
7230 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7231 {
|
|
|
7232 np->wn_u1.index = 0;
|
|
|
7233 np->wn_u2.wnode = NULL;
|
|
|
7234
|
|
|
7235 if (np->wn_byte != NUL)
|
|
|
7236 clear_node(np->wn_child);
|
|
|
7237 }
|
|
|
7238 }
|
|
|
7239
|
|
|
7240
|
|
|
7241 /*
|
|
300
|
7242 * Dump a word tree at node "node".
|
|
|
7243 *
|
|
|
7244 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
7245 * byte recursively write the children.
|
|
|
7246 *
|
|
|
7247 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
7248 * are made about the indexes (so that we don't have to write them in the
|
|
|
7249 * file).
|
|
236
|
7250 *
|
|
300
|
7251 * Returns the number of nodes used.
|
|
236
|
7252 */
|
|
300
|
7253 static int
|
|
346
|
7254 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
7255 FILE *fd; /* NULL when only counting */
|
|
300
|
7256 wordnode_T *node;
|
|
|
7257 int index;
|
|
|
7258 int regionmask;
|
|
339
|
7259 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
7260 {
|
|
300
|
7261 int newindex = index;
|
|
|
7262 int siblingcount = 0;
|
|
|
7263 wordnode_T *np;
|
|
236
|
7264 int flags;
|
|
300
|
7265
|
|
|
7266 /* If "node" is zero the tree is empty. */
|
|
|
7267 if (node == NULL)
|
|
|
7268 return 0;
|
|
|
7269
|
|
|
7270 /* Store the index where this node is written. */
|
|
346
|
7271 node->wn_u1.index = index;
|
|
236
|
7272
|
|
300
|
7273 /* Count the number of siblings. */
|
|
|
7274 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7275 ++siblingcount;
|
|
236
|
7276
|
|
300
|
7277 /* Write the sibling count. */
|
|
|
7278 if (fd != NULL)
|
|
|
7279 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
7280
|
|
300
|
7281 /* Write each sibling byte and optionally extra info. */
|
|
|
7282 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
7283 {
|
|
300
|
7284 if (np->wn_byte == 0)
|
|
|
7285 {
|
|
|
7286 if (fd != NULL)
|
|
|
7287 {
|
|
339
|
7288 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
7289 if (prefixtree)
|
|
300
|
7290 {
|
|
481
|
7291 /* In PREFIXTREE write the required affixID and the
|
|
390
|
7292 * associated condition nr (stored in wn_region). The
|
|
|
7293 * byte value is misused to store the "rare" and "not
|
|
|
7294 * combining" flags */
|
|
455
|
7295 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
7296 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7297 else
|
|
|
7298 {
|
|
390
|
7299 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
7300 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
7301 }
|
|
481
|
7302 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7303 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
7304 }
|
|
|
7305 else
|
|
|
7306 {
|
|
339
|
7307 /* For word trees we write the flag/region items. */
|
|
|
7308 flags = np->wn_flags;
|
|
|
7309 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
7310 flags |= WF_REGION;
|
|
481
|
7311 if (np->wn_affixID != 0)
|
|
|
7312 flags |= WF_AFX;
|
|
339
|
7313 if (flags == 0)
|
|
|
7314 {
|
|
|
7315 /* word without flags or region */
|
|
|
7316 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7317 }
|
|
|
7318 else
|
|
|
7319 {
|
|
390
|
7320 if (np->wn_flags >= 0x100)
|
|
|
7321 {
|
|
|
7322 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
7323 putc(flags, fd); /* <flags> */
|
|
|
7324 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
7325 }
|
|
|
7326 else
|
|
|
7327 {
|
|
|
7328 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
7329 putc(flags, fd); /* <flags> */
|
|
|
7330 }
|
|
339
|
7331 if (flags & WF_REGION)
|
|
|
7332 putc(np->wn_region, fd); /* <region> */
|
|
481
|
7333 if (flags & WF_AFX)
|
|
|
7334 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7335 }
|
|
300
|
7336 }
|
|
|
7337 }
|
|
|
7338 }
|
|
|
7339 else
|
|
|
7340 {
|
|
346
|
7341 if (np->wn_child->wn_u1.index != 0
|
|
|
7342 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
7343 {
|
|
|
7344 /* The child is written elsewhere, write the reference. */
|
|
|
7345 if (fd != NULL)
|
|
|
7346 {
|
|
|
7347 putc(BY_INDEX, fd); /* <byte> */
|
|
|
7348 /* <nodeidx> */
|
|
346
|
7349 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
7350 }
|
|
|
7351 }
|
|
346
|
7352 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
7353 /* We will write the child below and give it an index. */
|
|
346
|
7354 np->wn_child->wn_u2.wnode = node;
|
|
236
|
7355
|
|
300
|
7356 if (fd != NULL)
|
|
|
7357 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
7358 {
|
|
|
7359 EMSG(_(e_write));
|
|
|
7360 return 0;
|
|
|
7361 }
|
|
|
7362 }
|
|
236
|
7363 }
|
|
|
7364
|
|
300
|
7365 /* Space used in the array when reading: one for each sibling and one for
|
|
|
7366 * the count. */
|
|
|
7367 newindex += siblingcount + 1;
|
|
249
|
7368
|
|
300
|
7369 /* Recursively dump the children of each sibling. */
|
|
|
7370 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
7371 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
7372 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
7373 prefixtree);
|
|
249
|
7374
|
|
300
|
7375 return newindex;
|
|
236
|
7376 }
|
|
|
7377
|
|
|
7378
|
|
|
7379 /*
|
|
310
|
7380 * ":mkspell [-ascii] outfile infile ..."
|
|
|
7381 * ":mkspell [-ascii] addfile"
|
|
236
|
7382 */
|
|
|
7383 void
|
|
|
7384 ex_mkspell(eap)
|
|
|
7385 exarg_T *eap;
|
|
|
7386 {
|
|
|
7387 int fcount;
|
|
|
7388 char_u **fnames;
|
|
310
|
7389 char_u *arg = eap->arg;
|
|
|
7390 int ascii = FALSE;
|
|
|
7391
|
|
|
7392 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
7393 {
|
|
|
7394 ascii = TRUE;
|
|
|
7395 arg = skipwhite(arg + 6);
|
|
|
7396 }
|
|
|
7397
|
|
|
7398 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
7399 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
7400 {
|
|
323
|
7401 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
7402 FreeWild(fcount, fnames);
|
|
|
7403 }
|
|
|
7404 }
|
|
|
7405
|
|
|
7406 /*
|
|
|
7407 * Create a Vim spell file from one or more word lists.
|
|
|
7408 * "fnames[0]" is the output file name.
|
|
|
7409 * "fnames[fcount - 1]" is the last input file name.
|
|
|
7410 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
7411 * and ".spl" is appended to make the output file name.
|
|
|
7412 */
|
|
|
7413 static void
|
|
323
|
7414 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
7415 int fcount;
|
|
|
7416 char_u **fnames;
|
|
|
7417 int ascii; /* -ascii argument given */
|
|
|
7418 int overwrite; /* overwrite existing output file */
|
|
323
|
7419 int added_word; /* invoked through "zg" */
|
|
310
|
7420 {
|
|
236
|
7421 char_u fname[MAXPATHL];
|
|
|
7422 char_u wfname[MAXPATHL];
|
|
310
|
7423 char_u **innames;
|
|
|
7424 int incount;
|
|
236
|
7425 afffile_T *(afile[8]);
|
|
|
7426 int i;
|
|
|
7427 int len;
|
|
|
7428 struct stat st;
|
|
255
|
7429 int error = FALSE;
|
|
300
|
7430 spellinfo_T spin;
|
|
|
7431
|
|
|
7432 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
7433 spin.si_verbose = !added_word;
|
|
310
|
7434 spin.si_ascii = ascii;
|
|
323
|
7435 spin.si_followup = TRUE;
|
|
|
7436 spin.si_rem_accents = TRUE;
|
|
|
7437 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
7438 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
7439 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
7440 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
500
|
7441 spin.si_newcompID = 127; /* start compound ID at first maximum */
|
|
310
|
7442
|
|
|
7443 /* default: fnames[0] is output file, following are input files */
|
|
|
7444 innames = &fnames[1];
|
|
|
7445 incount = fcount - 1;
|
|
|
7446
|
|
|
7447 if (fcount >= 1)
|
|
240
|
7448 {
|
|
310
|
7449 len = STRLEN(fnames[0]);
|
|
|
7450 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
7451 {
|
|
|
7452 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
7453 * "path/en.latin1.add.spl". */
|
|
|
7454 innames = &fnames[0];
|
|
|
7455 incount = 1;
|
|
|
7456 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
7457 }
|
|
366
|
7458 else if (fcount == 1)
|
|
|
7459 {
|
|
|
7460 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
7461 innames = &fnames[0];
|
|
|
7462 incount = 1;
|
|
|
7463 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
7464 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
7465 }
|
|
310
|
7466 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
7467 {
|
|
|
7468 /* Name ends in ".spl", use as the file name. */
|
|
323
|
7469 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
7470 }
|
|
|
7471 else
|
|
|
7472 /* Name should be language, make the file name from it. */
|
|
|
7473 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
7474 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
7475
|
|
|
7476 /* Check for .ascii.spl. */
|
|
|
7477 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
7478 spin.si_ascii = TRUE;
|
|
|
7479
|
|
|
7480 /* Check for .add.spl. */
|
|
|
7481 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
7482 spin.si_add = TRUE;
|
|
240
|
7483 }
|
|
|
7484
|
|
310
|
7485 if (incount <= 0)
|
|
236
|
7486 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
7487 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
7488 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
7489 else if (incount > 8)
|
|
236
|
7490 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
7491 else
|
|
|
7492 {
|
|
|
7493 /* Check for overwriting before doing things that may take a lot of
|
|
|
7494 * time. */
|
|
310
|
7495 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
7496 {
|
|
|
7497 EMSG(_(e_exists));
|
|
310
|
7498 return;
|
|
236
|
7499 }
|
|
310
|
7500 if (mch_isdir(wfname))
|
|
236
|
7501 {
|
|
310
|
7502 EMSG2(_(e_isadir2), wfname);
|
|
|
7503 return;
|
|
236
|
7504 }
|
|
|
7505
|
|
|
7506 /*
|
|
|
7507 * Init the aff and dic pointers.
|
|
|
7508 * Get the region names if there are more than 2 arguments.
|
|
|
7509 */
|
|
310
|
7510 for (i = 0; i < incount; ++i)
|
|
236
|
7511 {
|
|
310
|
7512 afile[i] = NULL;
|
|
300
|
7513
|
|
316
|
7514 if (incount > 1)
|
|
236
|
7515 {
|
|
310
|
7516 len = STRLEN(innames[i]);
|
|
|
7517 if (STRLEN(gettail(innames[i])) < 5
|
|
|
7518 || innames[i][len - 3] != '_')
|
|
236
|
7519 {
|
|
310
|
7520 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
7521 return;
|
|
236
|
7522 }
|
|
316
|
7523 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
7524 spin.si_region_name[i * 2 + 1] =
|
|
|
7525 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
7526 }
|
|
|
7527 }
|
|
316
|
7528 spin.si_region_count = incount;
|
|
236
|
7529
|
|
474
|
7530 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
7531 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
7532 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
7533 if (spin.si_foldroot == NULL
|
|
|
7534 || spin.si_keeproot == NULL
|
|
|
7535 || spin.si_prefroot == NULL)
|
|
300
|
7536 {
|
|
469
|
7537 free_blocks(spin.si_blocks);
|
|
310
|
7538 return;
|
|
300
|
7539 }
|
|
|
7540
|
|
351
|
7541 /* When not producing a .add.spl file clear the character table when
|
|
|
7542 * we encounter one in the .aff file. This means we dump the current
|
|
|
7543 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
7544 * better than guessing the contents, the table will match a
|
|
|
7545 * previously loaded spell file. */
|
|
|
7546 if (!spin.si_add)
|
|
|
7547 spin.si_clear_chartab = TRUE;
|
|
|
7548
|
|
236
|
7549 /*
|
|
|
7550 * Read all the .aff and .dic files.
|
|
|
7551 * Text is converted to 'encoding'.
|
|
300
|
7552 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
7553 */
|
|
310
|
7554 for (i = 0; i < incount && !error; ++i)
|
|
236
|
7555 {
|
|
300
|
7556 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
7557 spin.si_region = 1 << i;
|
|
|
7558
|
|
|
7559 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
7560 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
7561 {
|
|
|
7562 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
7563 * "SET" line. */
|
|
474
|
7564 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
7565 if (afile[i] == NULL)
|
|
300
|
7566 error = TRUE;
|
|
|
7567 else
|
|
|
7568 {
|
|
|
7569 /* Read the .dic file and store the words in the trees. */
|
|
|
7570 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
7571 innames[i]);
|
|
474
|
7572 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
7573 error = TRUE;
|
|
|
7574 }
|
|
|
7575 }
|
|
|
7576 else
|
|
|
7577 {
|
|
|
7578 /* No .aff file, try reading the file as a word list. Store
|
|
|
7579 * the words in the trees. */
|
|
474
|
7580 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
7581 error = TRUE;
|
|
|
7582 }
|
|
236
|
7583
|
|
310
|
7584 #ifdef FEAT_MBYTE
|
|
236
|
7585 /* Free any conversion stuff. */
|
|
300
|
7586 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
7587 #endif
|
|
236
|
7588 }
|
|
|
7589
|
|
492
|
7590 if (spin.si_compflags != NULL && spin.si_nobreak)
|
|
|
7591 MSG(_("Warning: both compounding and NOBREAK specified"));
|
|
|
7592
|
|
300
|
7593 if (!error)
|
|
236
|
7594 {
|
|
|
7595 /*
|
|
300
|
7596 * Combine tails in the tree.
|
|
236
|
7597 */
|
|
474
|
7598 if (spin.si_verbose || p_verbose > 2)
|
|
|
7599 {
|
|
|
7600 if (!spin.si_verbose)
|
|
310
|
7601 verbose_enter();
|
|
469
|
7602 MSG(_(msg_compressing));
|
|
310
|
7603 out_flush();
|
|
474
|
7604 if (!spin.si_verbose)
|
|
310
|
7605 verbose_leave();
|
|
|
7606 }
|
|
474
|
7607 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
7608 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
7609 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
7610 }
|
|
|
7611
|
|
300
|
7612 if (!error)
|
|
|
7613 {
|
|
|
7614 /*
|
|
|
7615 * Write the info in the spell file.
|
|
|
7616 */
|
|
474
|
7617 if (spin.si_verbose || p_verbose > 2)
|
|
|
7618 {
|
|
|
7619 if (!spin.si_verbose)
|
|
310
|
7620 verbose_enter();
|
|
366
|
7621 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
7622 out_flush();
|
|
474
|
7623 if (!spin.si_verbose)
|
|
310
|
7624 verbose_leave();
|
|
|
7625 }
|
|
|
7626
|
|
500
|
7627 error = write_vim_spell(&spin, wfname) == FAIL;
|
|
474
|
7628
|
|
|
7629 if (spin.si_verbose || p_verbose > 2)
|
|
|
7630 {
|
|
|
7631 if (!spin.si_verbose)
|
|
310
|
7632 verbose_enter();
|
|
|
7633 MSG(_("Done!"));
|
|
|
7634 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
7635 spin.si_memtot);
|
|
310
|
7636 out_flush();
|
|
474
|
7637 if (!spin.si_verbose)
|
|
310
|
7638 verbose_leave();
|
|
|
7639 }
|
|
|
7640
|
|
|
7641 /* If the file is loaded need to reload it. */
|
|
500
|
7642 if (!error)
|
|
|
7643 spell_reload_one(wfname, added_word);
|
|
300
|
7644 }
|
|
|
7645
|
|
|
7646 /* Free the allocated memory. */
|
|
323
|
7647 ga_clear(&spin.si_rep);
|
|
|
7648 ga_clear(&spin.si_sal);
|
|
|
7649 ga_clear(&spin.si_map);
|
|
339
|
7650 ga_clear(&spin.si_prefcond);
|
|
300
|
7651
|
|
|
7652 /* Free the .aff file structures. */
|
|
310
|
7653 for (i = 0; i < incount; ++i)
|
|
|
7654 if (afile[i] != NULL)
|
|
|
7655 spell_free_aff(afile[i]);
|
|
339
|
7656
|
|
|
7657 /* Free all the bits and pieces at once. */
|
|
|
7658 free_blocks(spin.si_blocks);
|
|
236
|
7659 }
|
|
310
|
7660 }
|
|
|
7661
|
|
|
7662
|
|
|
7663 /*
|
|
385
|
7664 * ":[count]spellgood {word}"
|
|
|
7665 * ":[count]spellwrong {word}"
|
|
310
|
7666 */
|
|
|
7667 void
|
|
|
7668 ex_spell(eap)
|
|
|
7669 exarg_T *eap;
|
|
|
7670 {
|
|
381
|
7671 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
7672 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
7673 }
|
|
|
7674
|
|
310
|
7675 /*
|
|
|
7676 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
7677 */
|
|
|
7678 void
|
|
385
|
7679 spell_add_word(word, len, bad, index)
|
|
310
|
7680 char_u *word;
|
|
|
7681 int len;
|
|
|
7682 int bad;
|
|
385
|
7683 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
7684 'spellfile' */
|
|
310
|
7685 {
|
|
|
7686 FILE *fd;
|
|
385
|
7687 buf_T *buf = NULL;
|
|
351
|
7688 int new_spf = FALSE;
|
|
|
7689 struct stat st;
|
|
381
|
7690 char_u *fname;
|
|
385
|
7691 char_u fnamebuf[MAXPATHL];
|
|
|
7692 char_u line[MAXWLEN * 2];
|
|
|
7693 long fpos, fpos_next = 0;
|
|
|
7694 int i;
|
|
|
7695 char_u *spf;
|
|
|
7696
|
|
|
7697 if (index == 0) /* use internal wordlist */
|
|
|
7698 {
|
|
|
7699 if (int_wordlist == NULL)
|
|
|
7700 {
|
|
|
7701 int_wordlist = vim_tempname('s');
|
|
|
7702 if (int_wordlist == NULL)
|
|
381
|
7703 return;
|
|
|
7704 }
|
|
385
|
7705 fname = int_wordlist;
|
|
381
|
7706 }
|
|
310
|
7707 else
|
|
|
7708 {
|
|
381
|
7709 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
7710 if (*curbuf->b_p_spf == NUL)
|
|
|
7711 {
|
|
|
7712 init_spellfile();
|
|
|
7713 new_spf = TRUE;
|
|
|
7714 }
|
|
|
7715
|
|
|
7716 if (*curbuf->b_p_spf == NUL)
|
|
|
7717 {
|
|
523
|
7718 EMSG2(_(e_notset), "spellfile");
|
|
381
|
7719 return;
|
|
|
7720 }
|
|
|
7721
|
|
385
|
7722 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
7723 {
|
|
|
7724 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
7725 if (i == index)
|
|
|
7726 break;
|
|
|
7727 if (*spf == NUL)
|
|
|
7728 {
|
|
502
|
7729 EMSGN(_("E765: 'spellfile' does not have %ld entries"), index);
|
|
385
|
7730 return;
|
|
|
7731 }
|
|
|
7732 }
|
|
|
7733
|
|
310
|
7734 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
7735 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
7736 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
7737 buf = NULL;
|
|
|
7738 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
7739 {
|
|
310
|
7740 EMSG(_(e_bufloaded));
|
|
381
|
7741 return;
|
|
|
7742 }
|
|
|
7743
|
|
385
|
7744 fname = fnamebuf;
|
|
|
7745 }
|
|
|
7746
|
|
|
7747 if (bad)
|
|
|
7748 {
|
|
|
7749 /* When the word also appears as good word we need to remove that one,
|
|
|
7750 * since its flags sort before the one with WF_BANNED. */
|
|
|
7751 fd = mch_fopen((char *)fname, "r");
|
|
|
7752 if (fd != NULL)
|
|
|
7753 {
|
|
|
7754 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
7755 {
|
|
|
7756 fpos = fpos_next;
|
|
|
7757 fpos_next = ftell(fd);
|
|
|
7758 if (STRNCMP(word, line, len) == 0
|
|
|
7759 && (line[len] == '/' || line[len] < ' '))
|
|
|
7760 {
|
|
|
7761 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
7762 * the start of the line. Mixing reading and writing
|
|
|
7763 * doesn't work for all systems, close the file first. */
|
|
|
7764 fclose(fd);
|
|
|
7765 fd = mch_fopen((char *)fname, "r+");
|
|
|
7766 if (fd == NULL)
|
|
|
7767 break;
|
|
|
7768 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
7769 fputc('#', fd);
|
|
|
7770 fseek(fd, fpos_next, SEEK_SET);
|
|
|
7771 }
|
|
|
7772 }
|
|
|
7773 fclose(fd);
|
|
|
7774 }
|
|
381
|
7775 }
|
|
|
7776
|
|
|
7777 fd = mch_fopen((char *)fname, "a");
|
|
|
7778 if (fd == NULL && new_spf)
|
|
|
7779 {
|
|
|
7780 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
7781 * We may need to create the "spell" directory first. We already
|
|
|
7782 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
7783 STRCPY(NameBuff, fname);
|
|
|
7784 *gettail_sep(NameBuff) = NUL;
|
|
|
7785 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
7786 {
|
|
|
7787 /* The directory doesn't exist. Try creating it and opening the
|
|
|
7788 * file again. */
|
|
|
7789 vim_mkdir(NameBuff, 0755);
|
|
|
7790 fd = mch_fopen((char *)fname, "a");
|
|
|
7791 }
|
|
|
7792 }
|
|
|
7793
|
|
|
7794 if (fd == NULL)
|
|
|
7795 EMSG2(_(e_notopen), fname);
|
|
|
7796 else
|
|
|
7797 {
|
|
|
7798 if (bad)
|
|
|
7799 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
7800 else
|
|
381
|
7801 fprintf(fd, "%.*s\n", len, word);
|
|
|
7802 fclose(fd);
|
|
|
7803
|
|
|
7804 /* Update the .add.spl file. */
|
|
|
7805 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
7806
|
|
|
7807 /* If the .add file is edited somewhere, reload it. */
|
|
|
7808 if (buf != NULL)
|
|
|
7809 buf_reload(buf);
|
|
|
7810
|
|
|
7811 redraw_all_later(NOT_VALID);
|
|
310
|
7812 }
|
|
|
7813 }
|
|
|
7814
|
|
|
7815 /*
|
|
|
7816 * Initialize 'spellfile' for the current buffer.
|
|
|
7817 */
|
|
|
7818 static void
|
|
|
7819 init_spellfile()
|
|
|
7820 {
|
|
|
7821 char_u buf[MAXPATHL];
|
|
|
7822 int l;
|
|
500
|
7823 char_u *fname;
|
|
310
|
7824 char_u *rtp;
|
|
351
|
7825 char_u *lend;
|
|
501
|
7826 int aspath = FALSE;
|
|
|
7827 char_u *lstart = curbuf->b_p_spl;
|
|
310
|
7828
|
|
|
7829 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
7830 {
|
|
501
|
7831 /* Find the end of the language name. Exclude the region. If there
|
|
|
7832 * is a path separator remember the start of the tail. */
|
|
351
|
7833 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
7834 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
501
|
7835 if (vim_ispathsep(*lend))
|
|
|
7836 {
|
|
|
7837 aspath = TRUE;
|
|
|
7838 lstart = lend + 1;
|
|
|
7839 }
|
|
351
|
7840
|
|
|
7841 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
7842 * are allowed to write. */
|
|
310
|
7843 rtp = p_rtp;
|
|
|
7844 while (*rtp != NUL)
|
|
|
7845 {
|
|
501
|
7846 if (aspath)
|
|
|
7847 /* Use directory of an entry with path, e.g., for
|
|
|
7848 * "/dir/lg.utf-8.spl" use "/dir". */
|
|
|
7849 vim_strncpy(buf, curbuf->b_p_spl, lstart - curbuf->b_p_spl - 1);
|
|
|
7850 else
|
|
|
7851 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
7852 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
310
|
7853 if (filewritable(buf) == 2)
|
|
|
7854 {
|
|
316
|
7855 /* Use the first language name from 'spelllang' and the
|
|
|
7856 * encoding used in the first loaded .spl file. */
|
|
501
|
7857 if (aspath)
|
|
|
7858 vim_strncpy(buf, curbuf->b_p_spl, lend - curbuf->b_p_spl);
|
|
|
7859 else
|
|
|
7860 {
|
|
|
7861 l = STRLEN(buf);
|
|
|
7862 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
|
7863 "/spell/%.*s", (int)(lend - lstart), lstart);
|
|
|
7864 }
|
|
310
|
7865 l = STRLEN(buf);
|
|
501
|
7866 fname = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang->sl_fname;
|
|
|
7867 vim_snprintf((char *)buf + l, MAXPATHL - l, ".%s.add",
|
|
|
7868 fname != NULL
|
|
|
7869 && strstr((char *)gettail(fname), ".ascii.") != NULL
|
|
|
7870 ? (char_u *)"ascii" : spell_enc());
|
|
310
|
7871 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
7872 break;
|
|
|
7873 }
|
|
501
|
7874 aspath = FALSE;
|
|
310
|
7875 }
|
|
|
7876 }
|
|
|
7877 }
|
|
236
|
7878
|
|
300
|
7879
|
|
307
|
7880 /*
|
|
|
7881 * Init the chartab used for spelling for ASCII.
|
|
|
7882 * EBCDIC is not supported!
|
|
|
7883 */
|
|
|
7884 static void
|
|
|
7885 clear_spell_chartab(sp)
|
|
|
7886 spelltab_T *sp;
|
|
|
7887 {
|
|
324
|
7888 int i;
|
|
307
|
7889
|
|
|
7890 /* Init everything to FALSE. */
|
|
|
7891 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
7892 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
7893 for (i = 0; i < 256; ++i)
|
|
324
|
7894 {
|
|
307
|
7895 sp->st_fold[i] = i;
|
|
324
|
7896 sp->st_upper[i] = i;
|
|
|
7897 }
|
|
307
|
7898
|
|
|
7899 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
7900 * that is done separately. */
|
|
|
7901 for (i = '0'; i <= '9'; ++i)
|
|
|
7902 sp->st_isw[i] = TRUE;
|
|
|
7903 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
7904 {
|
|
|
7905 sp->st_isw[i] = TRUE;
|
|
|
7906 sp->st_isu[i] = TRUE;
|
|
|
7907 sp->st_fold[i] = i + 0x20;
|
|
|
7908 }
|
|
|
7909 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
7910 {
|
|
307
|
7911 sp->st_isw[i] = TRUE;
|
|
324
|
7912 sp->st_upper[i] = i - 0x20;
|
|
|
7913 }
|
|
307
|
7914 }
|
|
|
7915
|
|
|
7916 /*
|
|
|
7917 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
7918 * Called once while starting up and when 'encoding' changes.
|
|
|
7919 * The default is to use isalpha(), but the spell file should define the word
|
|
|
7920 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
7921 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
7922 */
|
|
|
7923 void
|
|
|
7924 init_spell_chartab()
|
|
|
7925 {
|
|
|
7926 int i;
|
|
|
7927
|
|
|
7928 did_set_spelltab = FALSE;
|
|
|
7929 clear_spell_chartab(&spelltab);
|
|
|
7930 #ifdef FEAT_MBYTE
|
|
|
7931 if (enc_dbcs)
|
|
|
7932 {
|
|
|
7933 /* DBCS: assume double-wide characters are word characters. */
|
|
|
7934 for (i = 128; i <= 255; ++i)
|
|
|
7935 if (MB_BYTE2LEN(i) == 2)
|
|
|
7936 spelltab.st_isw[i] = TRUE;
|
|
|
7937 }
|
|
324
|
7938 else if (enc_utf8)
|
|
|
7939 {
|
|
|
7940 for (i = 128; i < 256; ++i)
|
|
|
7941 {
|
|
|
7942 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
7943 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
7944 spelltab.st_fold[i] = utf_fold(i);
|
|
|
7945 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
7946 }
|
|
|
7947 }
|
|
307
|
7948 else
|
|
|
7949 #endif
|
|
|
7950 {
|
|
324
|
7951 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
7952 for (i = 128; i < 256; ++i)
|
|
|
7953 {
|
|
|
7954 if (MB_ISUPPER(i))
|
|
|
7955 {
|
|
324
|
7956 spelltab.st_isw[i] = TRUE;
|
|
307
|
7957 spelltab.st_isu[i] = TRUE;
|
|
|
7958 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
7959 }
|
|
324
|
7960 else if (MB_ISLOWER(i))
|
|
|
7961 {
|
|
|
7962 spelltab.st_isw[i] = TRUE;
|
|
|
7963 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
7964 }
|
|
307
|
7965 }
|
|
|
7966 }
|
|
|
7967 }
|
|
|
7968
|
|
|
7969 /*
|
|
|
7970 * Set the spell character tables from strings in the affix file.
|
|
|
7971 */
|
|
|
7972 static int
|
|
|
7973 set_spell_chartab(fol, low, upp)
|
|
|
7974 char_u *fol;
|
|
|
7975 char_u *low;
|
|
|
7976 char_u *upp;
|
|
|
7977 {
|
|
|
7978 /* We build the new tables here first, so that we can compare with the
|
|
|
7979 * previous one. */
|
|
|
7980 spelltab_T new_st;
|
|
|
7981 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
7982 int f, l, u;
|
|
|
7983
|
|
|
7984 clear_spell_chartab(&new_st);
|
|
|
7985
|
|
|
7986 while (*pf != NUL)
|
|
|
7987 {
|
|
|
7988 if (*pl == NUL || *pu == NUL)
|
|
|
7989 {
|
|
|
7990 EMSG(_(e_affform));
|
|
|
7991 return FAIL;
|
|
|
7992 }
|
|
|
7993 #ifdef FEAT_MBYTE
|
|
|
7994 f = mb_ptr2char_adv(&pf);
|
|
|
7995 l = mb_ptr2char_adv(&pl);
|
|
|
7996 u = mb_ptr2char_adv(&pu);
|
|
|
7997 #else
|
|
|
7998 f = *pf++;
|
|
|
7999 l = *pl++;
|
|
|
8000 u = *pu++;
|
|
|
8001 #endif
|
|
|
8002 /* Every character that appears is a word character. */
|
|
|
8003 if (f < 256)
|
|
|
8004 new_st.st_isw[f] = TRUE;
|
|
|
8005 if (l < 256)
|
|
|
8006 new_st.st_isw[l] = TRUE;
|
|
|
8007 if (u < 256)
|
|
|
8008 new_st.st_isw[u] = TRUE;
|
|
|
8009
|
|
|
8010 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
8011 * case-folding */
|
|
|
8012 if (l < 256 && l != f)
|
|
|
8013 {
|
|
|
8014 if (f >= 256)
|
|
|
8015 {
|
|
|
8016 EMSG(_(e_affrange));
|
|
|
8017 return FAIL;
|
|
|
8018 }
|
|
|
8019 new_st.st_fold[l] = f;
|
|
|
8020 }
|
|
|
8021
|
|
|
8022 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
8023 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
8024 * "FOL" . */
|
|
307
|
8025 if (u < 256 && u != f)
|
|
|
8026 {
|
|
|
8027 if (f >= 256)
|
|
|
8028 {
|
|
|
8029 EMSG(_(e_affrange));
|
|
|
8030 return FAIL;
|
|
|
8031 }
|
|
|
8032 new_st.st_fold[u] = f;
|
|
|
8033 new_st.st_isu[u] = TRUE;
|
|
324
|
8034 new_st.st_upper[f] = u;
|
|
307
|
8035 }
|
|
|
8036 }
|
|
|
8037
|
|
|
8038 if (*pl != NUL || *pu != NUL)
|
|
|
8039 {
|
|
|
8040 EMSG(_(e_affform));
|
|
|
8041 return FAIL;
|
|
|
8042 }
|
|
|
8043
|
|
|
8044 return set_spell_finish(&new_st);
|
|
|
8045 }
|
|
|
8046
|
|
|
8047 /*
|
|
|
8048 * Set the spell character tables from strings in the .spl file.
|
|
|
8049 */
|
|
483
|
8050 static void
|
|
389
|
8051 set_spell_charflags(flags, cnt, fol)
|
|
307
|
8052 char_u *flags;
|
|
389
|
8053 int cnt; /* length of "flags" */
|
|
|
8054 char_u *fol;
|
|
307
|
8055 {
|
|
|
8056 /* We build the new tables here first, so that we can compare with the
|
|
|
8057 * previous one. */
|
|
|
8058 spelltab_T new_st;
|
|
|
8059 int i;
|
|
389
|
8060 char_u *p = fol;
|
|
324
|
8061 int c;
|
|
307
|
8062
|
|
|
8063 clear_spell_chartab(&new_st);
|
|
|
8064
|
|
389
|
8065 for (i = 0; i < 128; ++i)
|
|
|
8066 {
|
|
|
8067 if (i < cnt)
|
|
|
8068 {
|
|
|
8069 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
8070 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
8071 }
|
|
|
8072
|
|
|
8073 if (*p != NUL)
|
|
|
8074 {
|
|
307
|
8075 #ifdef FEAT_MBYTE
|
|
389
|
8076 c = mb_ptr2char_adv(&p);
|
|
307
|
8077 #else
|
|
389
|
8078 c = *p++;
|
|
307
|
8079 #endif
|
|
389
|
8080 new_st.st_fold[i + 128] = c;
|
|
|
8081 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
8082 new_st.st_upper[c] = i + 128;
|
|
|
8083 }
|
|
307
|
8084 }
|
|
|
8085
|
|
483
|
8086 (void)set_spell_finish(&new_st);
|
|
307
|
8087 }
|
|
|
8088
|
|
|
8089 static int
|
|
|
8090 set_spell_finish(new_st)
|
|
|
8091 spelltab_T *new_st;
|
|
|
8092 {
|
|
|
8093 int i;
|
|
|
8094
|
|
|
8095 if (did_set_spelltab)
|
|
|
8096 {
|
|
|
8097 /* check that it's the same table */
|
|
|
8098 for (i = 0; i < 256; ++i)
|
|
|
8099 {
|
|
|
8100 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
8101 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
8102 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
8103 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
8104 {
|
|
|
8105 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
8106 return FAIL;
|
|
|
8107 }
|
|
|
8108 }
|
|
|
8109 }
|
|
|
8110 else
|
|
|
8111 {
|
|
|
8112 /* copy the new spelltab into the one being used */
|
|
|
8113 spelltab = *new_st;
|
|
|
8114 did_set_spelltab = TRUE;
|
|
|
8115 }
|
|
|
8116
|
|
|
8117 return OK;
|
|
|
8118 }
|
|
|
8119
|
|
|
8120 /*
|
|
358
|
8121 * Return TRUE if "p" points to a word character.
|
|
366
|
8122 * As a special case we see "midword" characters as word character when it is
|
|
358
|
8123 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
8124 * Thus this only works properly when past the first character of the word.
|
|
358
|
8125 */
|
|
|
8126 static int
|
|
376
|
8127 spell_iswordp(p, buf)
|
|
358
|
8128 char_u *p;
|
|
376
|
8129 buf_T *buf; /* buffer used */
|
|
358
|
8130 {
|
|
366
|
8131 #ifdef FEAT_MBYTE
|
|
358
|
8132 char_u *s;
|
|
366
|
8133 int l;
|
|
|
8134 int c;
|
|
|
8135
|
|
|
8136 if (has_mbyte)
|
|
|
8137 {
|
|
|
8138 l = MB_BYTE2LEN(*p);
|
|
358
|
8139 s = p;
|
|
366
|
8140 if (l == 1)
|
|
|
8141 {
|
|
|
8142 /* be quick for ASCII */
|
|
376
|
8143 if (buf->b_spell_ismw[*p])
|
|
366
|
8144 {
|
|
|
8145 s = p + 1; /* skip a mid-word character */
|
|
|
8146 l = MB_BYTE2LEN(*s);
|
|
|
8147 }
|
|
|
8148 }
|
|
|
8149 else
|
|
|
8150 {
|
|
|
8151 c = mb_ptr2char(p);
|
|
376
|
8152 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
8153 : (buf->b_spell_ismw_mb != NULL
|
|
|
8154 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
8155 {
|
|
|
8156 s = p + l;
|
|
|
8157 l = MB_BYTE2LEN(*s);
|
|
|
8158 }
|
|
|
8159 }
|
|
|
8160
|
|
390
|
8161 c = mb_ptr2char(s);
|
|
|
8162 if (c > 255)
|
|
366
|
8163 return mb_get_class(s) >= 2;
|
|
390
|
8164 return spelltab.st_isw[c];
|
|
366
|
8165 }
|
|
358
|
8166 #endif
|
|
366
|
8167
|
|
376
|
8168 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
8169 }
|
|
|
8170
|
|
|
8171 /*
|
|
|
8172 * Return TRUE if "p" points to a word character.
|
|
|
8173 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
8174 */
|
|
|
8175 static int
|
|
|
8176 spell_iswordp_nmw(p)
|
|
|
8177 char_u *p;
|
|
|
8178 {
|
|
|
8179 #ifdef FEAT_MBYTE
|
|
390
|
8180 int c;
|
|
|
8181
|
|
|
8182 if (has_mbyte)
|
|
|
8183 {
|
|
|
8184 c = mb_ptr2char(p);
|
|
|
8185 if (c > 255)
|
|
|
8186 return mb_get_class(p) >= 2;
|
|
|
8187 return spelltab.st_isw[c];
|
|
|
8188 }
|
|
376
|
8189 #endif
|
|
|
8190 return spelltab.st_isw[*p];
|
|
358
|
8191 }
|
|
|
8192
|
|
372
|
8193 #ifdef FEAT_MBYTE
|
|
|
8194 /*
|
|
|
8195 * Return TRUE if "p" points to a word character.
|
|
|
8196 * Wide version of spell_iswordp().
|
|
|
8197 */
|
|
|
8198 static int
|
|
376
|
8199 spell_iswordp_w(p, buf)
|
|
372
|
8200 int *p;
|
|
376
|
8201 buf_T *buf;
|
|
372
|
8202 {
|
|
|
8203 int *s;
|
|
|
8204
|
|
376
|
8205 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
8206 : (buf->b_spell_ismw_mb != NULL
|
|
|
8207 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
8208 s = p + 1;
|
|
|
8209 else
|
|
|
8210 s = p;
|
|
|
8211
|
|
390
|
8212 if (*s > 255)
|
|
372
|
8213 {
|
|
|
8214 if (enc_utf8)
|
|
|
8215 return utf_class(*s) >= 2;
|
|
|
8216 if (enc_dbcs)
|
|
|
8217 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
8218 return 0;
|
|
|
8219 }
|
|
|
8220 return spelltab.st_isw[*s];
|
|
|
8221 }
|
|
|
8222 #endif
|
|
|
8223
|
|
358
|
8224 /*
|
|
339
|
8225 * Write the table with prefix conditions to the .spl file.
|
|
483
|
8226 * When "fd" is NULL only count the length of what is written.
|
|
|
8227 */
|
|
|
8228 static int
|
|
339
|
8229 write_spell_prefcond(fd, gap)
|
|
|
8230 FILE *fd;
|
|
|
8231 garray_T *gap;
|
|
|
8232 {
|
|
|
8233 int i;
|
|
|
8234 char_u *p;
|
|
|
8235 int len;
|
|
483
|
8236 int totlen;
|
|
|
8237
|
|
|
8238 if (fd != NULL)
|
|
|
8239 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
8240
|
|
|
8241 totlen = 2 + gap->ga_len; /* length of <prefcondcnt> and <condlen> bytes */
|
|
339
|
8242
|
|
|
8243 for (i = 0; i < gap->ga_len; ++i)
|
|
|
8244 {
|
|
|
8245 /* <prefcond> : <condlen> <condstr> */
|
|
|
8246 p = ((char_u **)gap->ga_data)[i];
|
|
483
|
8247 if (p != NULL)
|
|
339
|
8248 {
|
|
|
8249 len = STRLEN(p);
|
|
483
|
8250 if (fd != NULL)
|
|
|
8251 {
|
|
|
8252 fputc(len, fd);
|
|
|
8253 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
8254 }
|
|
|
8255 totlen += len;
|
|
|
8256 }
|
|
|
8257 else if (fd != NULL)
|
|
|
8258 fputc(0, fd);
|
|
|
8259 }
|
|
|
8260
|
|
|
8261 return totlen;
|
|
307
|
8262 }
|
|
|
8263
|
|
|
8264 /*
|
|
324
|
8265 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
8266 * Uses the character definitions from the .spl file.
|
|
307
|
8267 * When using a multi-byte 'encoding' the length may change!
|
|
|
8268 * Returns FAIL when something wrong.
|
|
|
8269 */
|
|
|
8270 static int
|
|
324
|
8271 spell_casefold(str, len, buf, buflen)
|
|
|
8272 char_u *str;
|
|
307
|
8273 int len;
|
|
|
8274 char_u *buf;
|
|
|
8275 int buflen;
|
|
|
8276 {
|
|
|
8277 int i;
|
|
|
8278
|
|
|
8279 if (len >= buflen)
|
|
|
8280 {
|
|
|
8281 buf[0] = NUL;
|
|
|
8282 return FAIL; /* result will not fit */
|
|
|
8283 }
|
|
|
8284
|
|
|
8285 #ifdef FEAT_MBYTE
|
|
|
8286 if (has_mbyte)
|
|
|
8287 {
|
|
324
|
8288 int outi = 0;
|
|
|
8289 char_u *p;
|
|
307
|
8290 int c;
|
|
|
8291
|
|
|
8292 /* Fold one character at a time. */
|
|
324
|
8293 for (p = str; p < str + len; )
|
|
307
|
8294 {
|
|
|
8295 if (outi + MB_MAXBYTES > buflen)
|
|
|
8296 {
|
|
|
8297 buf[outi] = NUL;
|
|
|
8298 return FAIL;
|
|
|
8299 }
|
|
474
|
8300 c = mb_cptr2char_adv(&p);
|
|
324
|
8301 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
8302 }
|
|
|
8303 buf[outi] = NUL;
|
|
|
8304 }
|
|
|
8305 else
|
|
|
8306 #endif
|
|
|
8307 {
|
|
|
8308 /* Be quick for non-multibyte encodings. */
|
|
|
8309 for (i = 0; i < len; ++i)
|
|
324
|
8310 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
8311 buf[i] = NUL;
|
|
|
8312 }
|
|
|
8313
|
|
|
8314 return OK;
|
|
|
8315 }
|
|
|
8316
|
|
372
|
8317 #define SPS_BEST 1
|
|
|
8318 #define SPS_FAST 2
|
|
|
8319 #define SPS_DOUBLE 4
|
|
|
8320
|
|
|
8321 static int sps_flags = SPS_BEST;
|
|
483
|
8322 static int sps_limit = 9999;
|
|
372
|
8323
|
|
|
8324 /*
|
|
|
8325 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
483
|
8326 * Sets "sps_flags" and "sps_limit".
|
|
372
|
8327 */
|
|
|
8328 int
|
|
|
8329 spell_check_sps()
|
|
|
8330 {
|
|
|
8331 char_u *p;
|
|
483
|
8332 char_u *s;
|
|
372
|
8333 char_u buf[MAXPATHL];
|
|
|
8334 int f;
|
|
|
8335
|
|
|
8336 sps_flags = 0;
|
|
483
|
8337 sps_limit = 9999;
|
|
372
|
8338
|
|
|
8339 for (p = p_sps; *p != NUL; )
|
|
|
8340 {
|
|
|
8341 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
8342
|
|
|
8343 f = 0;
|
|
483
|
8344 if (VIM_ISDIGIT(*buf))
|
|
|
8345 {
|
|
|
8346 s = buf;
|
|
|
8347 sps_limit = getdigits(&s);
|
|
|
8348 if (*s != NUL && !VIM_ISDIGIT(*s))
|
|
|
8349 f = -1;
|
|
|
8350 }
|
|
|
8351 else if (STRCMP(buf, "best") == 0)
|
|
372
|
8352 f = SPS_BEST;
|
|
|
8353 else if (STRCMP(buf, "fast") == 0)
|
|
|
8354 f = SPS_FAST;
|
|
|
8355 else if (STRCMP(buf, "double") == 0)
|
|
|
8356 f = SPS_DOUBLE;
|
|
|
8357 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
8358 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
8359 f = -1;
|
|
|
8360
|
|
|
8361 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
8362 {
|
|
|
8363 sps_flags = SPS_BEST;
|
|
483
|
8364 sps_limit = 9999;
|
|
372
|
8365 return FAIL;
|
|
|
8366 }
|
|
|
8367 if (f != 0)
|
|
|
8368 sps_flags = f;
|
|
|
8369 }
|
|
|
8370
|
|
|
8371 if (sps_flags == 0)
|
|
|
8372 sps_flags = SPS_BEST;
|
|
|
8373
|
|
|
8374 return OK;
|
|
|
8375 }
|
|
|
8376
|
|
|
8377 /* Remember what "z?" replaced. */
|
|
|
8378 static char_u *repl_from = NULL;
|
|
|
8379 static char_u *repl_to = NULL;
|
|
|
8380
|
|
323
|
8381 /*
|
|
|
8382 * "z?": Find badly spelled word under or after the cursor.
|
|
|
8383 * Give suggestions for the properly spelled word.
|
|
485
|
8384 * When "count" is non-zero use that suggestion.
|
|
323
|
8385 */
|
|
|
8386 void
|
|
485
|
8387 spell_suggest(count)
|
|
|
8388 int count;
|
|
323
|
8389 {
|
|
|
8390 char_u *line;
|
|
|
8391 pos_T prev_cursor = curwin->w_cursor;
|
|
|
8392 char_u wcopy[MAXWLEN + 2];
|
|
|
8393 char_u *p;
|
|
|
8394 int i;
|
|
|
8395 int c;
|
|
|
8396 suginfo_T sug;
|
|
|
8397 suggest_T *stp;
|
|
372
|
8398 int mouse_used;
|
|
388
|
8399 int need_cap;
|
|
483
|
8400 int limit;
|
|
485
|
8401 int selected = count;
|
|
323
|
8402
|
|
344
|
8403 /* Find the start of the badly spelled word. */
|
|
498
|
8404 if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0
|
|
346
|
8405 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
8406 {
|
|
|
8407 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
8408 return;
|
|
|
8409
|
|
|
8410 /* No bad word or it starts after the cursor: use the word under the
|
|
|
8411 * cursor. */
|
|
|
8412 curwin->w_cursor = prev_cursor;
|
|
|
8413 line = ml_get_curline();
|
|
|
8414 p = line + curwin->w_cursor.col;
|
|
|
8415 /* Backup to before start of word. */
|
|
390
|
8416 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
8417 mb_ptr_back(line, p);
|
|
|
8418 /* Forward to start of word. */
|
|
390
|
8419 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
8420 mb_ptr_adv(p);
|
|
|
8421
|
|
390
|
8422 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
8423 {
|
|
|
8424 beep_flush();
|
|
|
8425 return;
|
|
|
8426 }
|
|
|
8427 curwin->w_cursor.col = p - line;
|
|
323
|
8428 }
|
|
|
8429
|
|
344
|
8430 /* Get the word and its length. */
|
|
|
8431
|
|
388
|
8432 /* Figure out if the word should be capitalised. */
|
|
475
|
8433 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
8434
|
|
|
8435 line = ml_get_curline();
|
|
388
|
8436
|
|
483
|
8437 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in
|
|
|
8438 * 'spellsuggest', whatever is smaller. */
|
|
|
8439 if (sps_limit > (int)Rows - 2)
|
|
|
8440 limit = (int)Rows - 2;
|
|
|
8441 else
|
|
|
8442 limit = sps_limit;
|
|
|
8443 spell_find_suggest(line + curwin->w_cursor.col, &sug, limit,
|
|
388
|
8444 TRUE, need_cap);
|
|
323
|
8445
|
|
|
8446 if (sug.su_ga.ga_len == 0)
|
|
|
8447 MSG(_("Sorry, no suggestions"));
|
|
485
|
8448 else if (count > 0)
|
|
|
8449 {
|
|
|
8450 if (count > sug.su_ga.ga_len)
|
|
|
8451 smsg((char_u *)_("Sorry, only %ld suggestions"),
|
|
|
8452 (long)sug.su_ga.ga_len);
|
|
|
8453 }
|
|
323
|
8454 else
|
|
|
8455 {
|
|
372
|
8456 vim_free(repl_from);
|
|
|
8457 repl_from = NULL;
|
|
|
8458 vim_free(repl_to);
|
|
|
8459 repl_to = NULL;
|
|
|
8460
|
|
474
|
8461 #ifdef FEAT_RIGHTLEFT
|
|
|
8462 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
8463 cmdmsg_rl = curwin->w_p_rl;
|
|
|
8464 if (cmdmsg_rl)
|
|
|
8465 msg_col = Columns - 1;
|
|
|
8466 #endif
|
|
|
8467
|
|
323
|
8468 /* List the suggestions. */
|
|
|
8469 msg_start();
|
|
372
|
8470 lines_left = Rows; /* avoid more prompt */
|
|
323
|
8471 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
8472 sug.su_badlen, sug.su_badptr);
|
|
474
|
8473 #ifdef FEAT_RIGHTLEFT
|
|
|
8474 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
8475 {
|
|
|
8476 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
8477 * untranslated message rightleft. */
|
|
|
8478 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
8479 sug.su_badlen, sug.su_badptr);
|
|
|
8480 }
|
|
|
8481 #endif
|
|
323
|
8482 msg_puts(IObuff);
|
|
|
8483 msg_clr_eos();
|
|
|
8484 msg_putchar('\n');
|
|
346
|
8485
|
|
323
|
8486 msg_scroll = TRUE;
|
|
|
8487 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
8488 {
|
|
344
|
8489 stp = &SUG(sug.su_ga, i);
|
|
323
|
8490
|
|
|
8491 /* The suggested word may replace only part of the bad word, add
|
|
|
8492 * the not replaced part. */
|
|
|
8493 STRCPY(wcopy, stp->st_word);
|
|
|
8494 if (sug.su_badlen > stp->st_orglen)
|
|
|
8495 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
8496 sug.su_badptr + stp->st_orglen,
|
|
|
8497 sug.su_badlen - stp->st_orglen);
|
|
474
|
8498 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
8499 #ifdef FEAT_RIGHTLEFT
|
|
|
8500 if (cmdmsg_rl)
|
|
|
8501 rl_mirror(IObuff);
|
|
|
8502 #endif
|
|
|
8503 msg_puts(IObuff);
|
|
|
8504
|
|
|
8505 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
8506 msg_puts(IObuff);
|
|
|
8507
|
|
|
8508 /* The word may replace more than "su_badlen". */
|
|
|
8509 if (sug.su_badlen < stp->st_orglen)
|
|
|
8510 {
|
|
|
8511 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
8512 stp->st_orglen, sug.su_badptr);
|
|
|
8513 msg_puts(IObuff);
|
|
|
8514 }
|
|
|
8515
|
|
324
|
8516 if (p_verbose > 0)
|
|
344
|
8517 {
|
|
346
|
8518 /* Add the score. */
|
|
351
|
8519 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
8520 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
8521 stp->st_salscore ? "s " : "",
|
|
|
8522 stp->st_score, stp->st_altscore);
|
|
|
8523 else
|
|
474
|
8524 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
8525 stp->st_score);
|
|
474
|
8526 #ifdef FEAT_RIGHTLEFT
|
|
|
8527 if (cmdmsg_rl)
|
|
|
8528 /* Mirror the numbers, but keep the leading space. */
|
|
|
8529 rl_mirror(IObuff + 1);
|
|
|
8530 #endif
|
|
346
|
8531 msg_advance(30);
|
|
|
8532 msg_puts(IObuff);
|
|
344
|
8533 }
|
|
323
|
8534 msg_putchar('\n');
|
|
|
8535 }
|
|
|
8536
|
|
474
|
8537 #ifdef FEAT_RIGHTLEFT
|
|
|
8538 cmdmsg_rl = FALSE;
|
|
|
8539 msg_col = 0;
|
|
|
8540 #endif
|
|
323
|
8541 /* Ask for choice. */
|
|
485
|
8542 selected = prompt_for_number(&mouse_used);
|
|
372
|
8543 if (mouse_used)
|
|
485
|
8544 selected -= lines_left;
|
|
|
8545 }
|
|
|
8546
|
|
|
8547 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
|
8548 {
|
|
|
8549 /* Save the from and to text for :spellrepall. */
|
|
|
8550 stp = &SUG(sug.su_ga, selected - 1);
|
|
|
8551 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
8552 repl_to = vim_strsave(stp->st_word);
|
|
|
8553
|
|
|
8554 /* Replace the word. */
|
|
|
8555 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
8556 if (p != NULL)
|
|
|
8557 {
|
|
|
8558 c = sug.su_badptr - line;
|
|
|
8559 mch_memmove(p, line, c);
|
|
|
8560 STRCPY(p + c, stp->st_word);
|
|
|
8561 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
8562 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
8563 curwin->w_cursor.col = c;
|
|
|
8564 changed_bytes(curwin->w_cursor.lnum, c);
|
|
|
8565
|
|
|
8566 /* For redo we use a change-word command. */
|
|
|
8567 ResetRedobuff();
|
|
|
8568 AppendToRedobuff((char_u *)"ciw");
|
|
|
8569 AppendToRedobuff(stp->st_word);
|
|
|
8570 AppendCharToRedobuff(ESC);
|
|
|
8571 }
|
|
|
8572 }
|
|
|
8573 else
|
|
|
8574 curwin->w_cursor = prev_cursor;
|
|
323
|
8575
|
|
344
|
8576 spell_find_cleanup(&sug);
|
|
|
8577 }
|
|
|
8578
|
|
|
8579 /*
|
|
475
|
8580 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
8581 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
8582 */
|
|
|
8583 static int
|
|
|
8584 check_need_cap(lnum, col)
|
|
|
8585 linenr_T lnum;
|
|
|
8586 colnr_T col;
|
|
|
8587 {
|
|
|
8588 int need_cap = FALSE;
|
|
|
8589 char_u *line;
|
|
|
8590 char_u *line_copy = NULL;
|
|
|
8591 char_u *p;
|
|
|
8592 colnr_T endcol;
|
|
|
8593 regmatch_T regmatch;
|
|
|
8594
|
|
|
8595 if (curbuf->b_cap_prog == NULL)
|
|
|
8596 return FALSE;
|
|
|
8597
|
|
|
8598 line = ml_get_curline();
|
|
|
8599 endcol = 0;
|
|
|
8600 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
8601 {
|
|
|
8602 /* At start of line, check if previous line is empty or sentence
|
|
|
8603 * ends there. */
|
|
|
8604 if (lnum == 1)
|
|
|
8605 need_cap = TRUE;
|
|
|
8606 else
|
|
|
8607 {
|
|
|
8608 line = ml_get(lnum - 1);
|
|
|
8609 if (*skipwhite(line) == NUL)
|
|
|
8610 need_cap = TRUE;
|
|
|
8611 else
|
|
|
8612 {
|
|
|
8613 /* Append a space in place of the line break. */
|
|
|
8614 line_copy = concat_str(line, (char_u *)" ");
|
|
|
8615 line = line_copy;
|
|
|
8616 endcol = STRLEN(line);
|
|
|
8617 }
|
|
|
8618 }
|
|
|
8619 }
|
|
|
8620 else
|
|
|
8621 endcol = col;
|
|
|
8622
|
|
|
8623 if (endcol > 0)
|
|
|
8624 {
|
|
|
8625 /* Check if sentence ends before the bad word. */
|
|
|
8626 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
8627 regmatch.rm_ic = FALSE;
|
|
|
8628 p = line + endcol;
|
|
|
8629 for (;;)
|
|
|
8630 {
|
|
|
8631 mb_ptr_back(line, p);
|
|
|
8632 if (p == line || spell_iswordp_nmw(p))
|
|
|
8633 break;
|
|
|
8634 if (vim_regexec(®match, p, 0)
|
|
|
8635 && regmatch.endp[0] == line + endcol)
|
|
|
8636 {
|
|
|
8637 need_cap = TRUE;
|
|
|
8638 break;
|
|
|
8639 }
|
|
|
8640 }
|
|
|
8641 }
|
|
|
8642
|
|
|
8643 vim_free(line_copy);
|
|
|
8644
|
|
|
8645 return need_cap;
|
|
|
8646 }
|
|
|
8647
|
|
|
8648
|
|
|
8649 /*
|
|
372
|
8650 * ":spellrepall"
|
|
|
8651 */
|
|
|
8652 /*ARGSUSED*/
|
|
|
8653 void
|
|
|
8654 ex_spellrepall(eap)
|
|
|
8655 exarg_T *eap;
|
|
|
8656 {
|
|
|
8657 pos_T pos = curwin->w_cursor;
|
|
|
8658 char_u *frompat;
|
|
|
8659 int addlen;
|
|
|
8660 char_u *line;
|
|
|
8661 char_u *p;
|
|
|
8662 int save_ws = p_ws;
|
|
483
|
8663 linenr_T prev_lnum = 0;
|
|
372
|
8664
|
|
|
8665 if (repl_from == NULL || repl_to == NULL)
|
|
|
8666 {
|
|
|
8667 EMSG(_("E752: No previous spell replacement"));
|
|
|
8668 return;
|
|
|
8669 }
|
|
|
8670 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
8671
|
|
|
8672 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
8673 if (frompat == NULL)
|
|
|
8674 return;
|
|
|
8675 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
8676 p_ws = FALSE;
|
|
|
8677
|
|
483
|
8678 sub_nsubs = 0;
|
|
|
8679 sub_nlines = 0;
|
|
372
|
8680 curwin->w_cursor.lnum = 0;
|
|
|
8681 while (!got_int)
|
|
|
8682 {
|
|
|
8683 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
8684 || u_save_cursor() == FAIL)
|
|
|
8685 break;
|
|
|
8686
|
|
|
8687 /* Only replace when the right word isn't there yet. This happens
|
|
|
8688 * when changing "etc" to "etc.". */
|
|
|
8689 line = ml_get_curline();
|
|
|
8690 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
8691 repl_to, STRLEN(repl_to)) != 0)
|
|
|
8692 {
|
|
|
8693 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
8694 if (p == NULL)
|
|
|
8695 break;
|
|
|
8696 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
8697 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
8698 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
8699 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
8700 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
483
|
8701
|
|
|
8702 if (curwin->w_cursor.lnum != prev_lnum)
|
|
|
8703 {
|
|
|
8704 ++sub_nlines;
|
|
|
8705 prev_lnum = curwin->w_cursor.lnum;
|
|
|
8706 }
|
|
|
8707 ++sub_nsubs;
|
|
372
|
8708 }
|
|
|
8709 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
8710 }
|
|
|
8711
|
|
|
8712 p_ws = save_ws;
|
|
|
8713 curwin->w_cursor = pos;
|
|
|
8714 vim_free(frompat);
|
|
|
8715
|
|
483
|
8716 if (sub_nsubs == 0)
|
|
372
|
8717 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
483
|
8718 else
|
|
|
8719 do_sub_msg(FALSE);
|
|
372
|
8720 }
|
|
|
8721
|
|
|
8722 /*
|
|
344
|
8723 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
8724 * a list of allocated strings.
|
|
|
8725 */
|
|
|
8726 void
|
|
475
|
8727 spell_suggest_list(gap, word, maxcount, need_cap)
|
|
344
|
8728 garray_T *gap;
|
|
|
8729 char_u *word;
|
|
|
8730 int maxcount; /* maximum nr of suggestions */
|
|
475
|
8731 int need_cap; /* 'spellcapcheck' matched */
|
|
344
|
8732 {
|
|
|
8733 suginfo_T sug;
|
|
|
8734 int i;
|
|
|
8735 suggest_T *stp;
|
|
|
8736 char_u *wcopy;
|
|
|
8737
|
|
475
|
8738 spell_find_suggest(word, &sug, maxcount, FALSE, need_cap);
|
|
344
|
8739
|
|
|
8740 /* Make room in "gap". */
|
|
|
8741 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
8742 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
8743 return;
|
|
|
8744
|
|
323
|
8745 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
8746 {
|
|
|
8747 stp = &SUG(sug.su_ga, i);
|
|
|
8748
|
|
|
8749 /* The suggested word may replace only part of "word", add the not
|
|
|
8750 * replaced part. */
|
|
|
8751 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
8752 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
8753 if (wcopy == NULL)
|
|
|
8754 break;
|
|
|
8755 STRCPY(wcopy, stp->st_word);
|
|
|
8756 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
8757 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
8758 }
|
|
|
8759
|
|
|
8760 spell_find_cleanup(&sug);
|
|
|
8761 }
|
|
|
8762
|
|
|
8763 /*
|
|
|
8764 * Find spell suggestions for the word at the start of "badptr".
|
|
|
8765 * Return the suggestions in "su->su_ga".
|
|
|
8766 * The maximum number of suggestions is "maxcount".
|
|
|
8767 * Note: does use info for the current window.
|
|
|
8768 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
8769 */
|
|
|
8770 static void
|
|
388
|
8771 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
8772 char_u *badptr;
|
|
|
8773 suginfo_T *su;
|
|
|
8774 int maxcount;
|
|
358
|
8775 int banbadword; /* don't include badword in suggestions */
|
|
388
|
8776 int need_cap; /* word should start with capital */
|
|
344
|
8777 {
|
|
534
|
8778 hlf_T attr = HLF_COUNT;
|
|
372
|
8779 char_u buf[MAXPATHL];
|
|
|
8780 char_u *p;
|
|
|
8781 int do_combine = FALSE;
|
|
|
8782 char_u *sps_copy;
|
|
|
8783 #ifdef FEAT_EVAL
|
|
|
8784 static int expr_busy = FALSE;
|
|
|
8785 #endif
|
|
385
|
8786 int c;
|
|
503
|
8787 int i;
|
|
|
8788 langp_T *lp;
|
|
344
|
8789
|
|
|
8790 /*
|
|
|
8791 * Set the info in "*su".
|
|
|
8792 */
|
|
|
8793 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
8794 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
8795 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
8796 if (*badptr == NUL)
|
|
|
8797 return;
|
|
344
|
8798 hash_init(&su->su_banned);
|
|
|
8799
|
|
|
8800 su->su_badptr = badptr;
|
|
385
|
8801 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
8802 su->su_maxcount = maxcount;
|
|
372
|
8803 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
8804
|
|
|
8805 if (su->su_badlen >= MAXWLEN)
|
|
|
8806 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
8807 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
8808 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
8809 su->su_fbadword, MAXWLEN);
|
|
346
|
8810 /* get caps flags for bad word */
|
|
474
|
8811 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
8812 su->su_badptr + su->su_badlen);
|
|
388
|
8813 if (need_cap)
|
|
|
8814 su->su_badflags |= WF_ONECAP;
|
|
344
|
8815
|
|
503
|
8816 /* Find the default language for sound folding. We simply use the first
|
|
|
8817 * one in 'spelllang' that supports sound folding. That's good for when
|
|
|
8818 * using multiple files for one language, it's not that bad when mixing
|
|
|
8819 * languages (e.g., "pl,en"). */
|
|
|
8820 for (i = 0; i < curbuf->b_langp.ga_len; ++i)
|
|
|
8821 {
|
|
|
8822 lp = LANGP_ENTRY(curbuf->b_langp, i);
|
|
|
8823 if (lp->lp_sallang != NULL)
|
|
|
8824 {
|
|
|
8825 su->su_sallang = lp->lp_sallang;
|
|
|
8826 break;
|
|
|
8827 }
|
|
|
8828 }
|
|
|
8829
|
|
534
|
8830 /* Soundfold the bad word with the default sound folding, so that we don't
|
|
|
8831 * have to do this many times. */
|
|
|
8832 if (su->su_sallang != NULL)
|
|
|
8833 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE,
|
|
|
8834 su->su_sal_badword);
|
|
|
8835
|
|
385
|
8836 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
8837 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
8838 * for that. */
|
|
455
|
8839 c = PTR2CHAR(su->su_badptr);
|
|
534
|
8840 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT)
|
|
385
|
8841 {
|
|
|
8842 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
8843 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
503
|
8844 0, TRUE, su->su_sallang);
|
|
385
|
8845 }
|
|
|
8846
|
|
344
|
8847 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
8848 if (banbadword)
|
|
|
8849 add_banned(su, su->su_badword);
|
|
344
|
8850
|
|
372
|
8851 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
8852 sps_copy = vim_strsave(p_sps);
|
|
|
8853 if (sps_copy == NULL)
|
|
|
8854 return;
|
|
|
8855
|
|
|
8856 /* Loop over the items in 'spellsuggest'. */
|
|
|
8857 for (p = sps_copy; *p != NUL; )
|
|
|
8858 {
|
|
|
8859 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
8860
|
|
|
8861 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
8862 {
|
|
|
8863 #ifdef FEAT_EVAL
|
|
375
|
8864 /* Evaluate an expression. Skip this when called recursively,
|
|
|
8865 * when using spellsuggest() in the expression. */
|
|
372
|
8866 if (!expr_busy)
|
|
|
8867 {
|
|
|
8868 expr_busy = TRUE;
|
|
|
8869 spell_suggest_expr(su, buf + 5);
|
|
|
8870 expr_busy = FALSE;
|
|
|
8871 }
|
|
|
8872 #endif
|
|
|
8873 }
|
|
|
8874 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
8875 /* Use list of suggestions in a file. */
|
|
|
8876 spell_suggest_file(su, buf + 5);
|
|
|
8877 else
|
|
|
8878 {
|
|
|
8879 /* Use internal method. */
|
|
|
8880 spell_suggest_intern(su);
|
|
|
8881 if (sps_flags & SPS_DOUBLE)
|
|
|
8882 do_combine = TRUE;
|
|
|
8883 }
|
|
|
8884 }
|
|
|
8885
|
|
|
8886 vim_free(sps_copy);
|
|
|
8887
|
|
|
8888 if (do_combine)
|
|
|
8889 /* Combine the two list of suggestions. This must be done last,
|
|
|
8890 * because sorting changes the order again. */
|
|
|
8891 score_combine(su);
|
|
|
8892 }
|
|
|
8893
|
|
|
8894 #ifdef FEAT_EVAL
|
|
|
8895 /*
|
|
|
8896 * Find suggestions by evaluating expression "expr".
|
|
|
8897 */
|
|
|
8898 static void
|
|
|
8899 spell_suggest_expr(su, expr)
|
|
|
8900 suginfo_T *su;
|
|
|
8901 char_u *expr;
|
|
|
8902 {
|
|
|
8903 list_T *list;
|
|
|
8904 listitem_T *li;
|
|
|
8905 int score;
|
|
|
8906 char_u *p;
|
|
|
8907
|
|
|
8908 /* The work is split up in a few parts to avoid having to export
|
|
|
8909 * suginfo_T.
|
|
|
8910 * First evaluate the expression and get the resulting list. */
|
|
|
8911 list = eval_spell_expr(su->su_badword, expr);
|
|
|
8912 if (list != NULL)
|
|
|
8913 {
|
|
|
8914 /* Loop over the items in the list. */
|
|
|
8915 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
8916 if (li->li_tv.v_type == VAR_LIST)
|
|
|
8917 {
|
|
|
8918 /* Get the word and the score from the items. */
|
|
|
8919 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
8920 if (score >= 0)
|
|
|
8921 add_suggestion(su, &su->su_ga, p,
|
|
503
|
8922 su->su_badlen, score, 0, TRUE, su->su_sallang);
|
|
372
|
8923 }
|
|
|
8924 list_unref(list);
|
|
|
8925 }
|
|
|
8926
|
|
|
8927 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8928 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8929 }
|
|
|
8930 #endif
|
|
|
8931
|
|
|
8932 /*
|
|
474
|
8933 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
8934 */
|
|
|
8935 static void
|
|
|
8936 spell_suggest_file(su, fname)
|
|
|
8937 suginfo_T *su;
|
|
|
8938 char_u *fname;
|
|
|
8939 {
|
|
|
8940 FILE *fd;
|
|
|
8941 char_u line[MAXWLEN * 2];
|
|
|
8942 char_u *p;
|
|
|
8943 int len;
|
|
|
8944 char_u cword[MAXWLEN];
|
|
|
8945
|
|
|
8946 /* Open the file. */
|
|
|
8947 fd = mch_fopen((char *)fname, "r");
|
|
|
8948 if (fd == NULL)
|
|
|
8949 {
|
|
|
8950 EMSG2(_(e_notopen), fname);
|
|
|
8951 return;
|
|
|
8952 }
|
|
|
8953
|
|
|
8954 /* Read it line by line. */
|
|
|
8955 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
8956 {
|
|
|
8957 line_breakcheck();
|
|
|
8958
|
|
|
8959 p = vim_strchr(line, '/');
|
|
|
8960 if (p == NULL)
|
|
|
8961 continue; /* No Tab found, just skip the line. */
|
|
|
8962 *p++ = NUL;
|
|
|
8963 if (STRICMP(su->su_badword, line) == 0)
|
|
|
8964 {
|
|
|
8965 /* Match! Isolate the good word, until CR or NL. */
|
|
|
8966 for (len = 0; p[len] >= ' '; ++len)
|
|
|
8967 ;
|
|
|
8968 p[len] = NUL;
|
|
|
8969
|
|
|
8970 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
8971 * of the bad word. */
|
|
|
8972 if (captype(p, NULL) == 0)
|
|
|
8973 {
|
|
|
8974 make_case_word(p, cword, su->su_badflags);
|
|
|
8975 p = cword;
|
|
|
8976 }
|
|
|
8977
|
|
|
8978 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
503
|
8979 SCORE_FILE, 0, TRUE, su->su_sallang);
|
|
372
|
8980 }
|
|
|
8981 }
|
|
|
8982
|
|
|
8983 fclose(fd);
|
|
|
8984
|
|
|
8985 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8986 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8987 }
|
|
|
8988
|
|
|
8989 /*
|
|
|
8990 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
8991 */
|
|
|
8992 static void
|
|
|
8993 spell_suggest_intern(su)
|
|
|
8994 suginfo_T *su;
|
|
|
8995 {
|
|
344
|
8996 /*
|
|
346
|
8997 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
8998 *
|
|
|
8999 * Set a maximum score to limit the combination of operations that is
|
|
|
9000 * tried.
|
|
|
9001 */
|
|
346
|
9002 suggest_try_special(su);
|
|
|
9003
|
|
|
9004 /*
|
|
|
9005 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
9006 * from the .aff file and inserting a space (split the word).
|
|
|
9007 */
|
|
|
9008 suggest_try_change(su);
|
|
344
|
9009
|
|
|
9010 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
9011 if (sps_flags & SPS_DOUBLE)
|
|
|
9012 score_comp_sal(su);
|
|
|
9013
|
|
|
9014 /*
|
|
346
|
9015 * 3. Try finding sound-a-like words.
|
|
344
|
9016 *
|
|
|
9017 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
9018 * very slow and often doesn't find new suggestions.
|
|
|
9019 */
|
|
|
9020 if ((sps_flags & SPS_DOUBLE)
|
|
|
9021 || (!(sps_flags & SPS_FAST)
|
|
|
9022 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
9023 {
|
|
|
9024 /* Allow a higher score now. */
|
|
|
9025 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
9026 suggest_try_soundalike(su);
|
|
344
|
9027 }
|
|
|
9028
|
|
|
9029 /* When CTRL-C was hit while searching do show the results. */
|
|
|
9030 ui_breakcheck();
|
|
|
9031 if (got_int)
|
|
|
9032 {
|
|
|
9033 (void)vgetc();
|
|
|
9034 got_int = FALSE;
|
|
|
9035 }
|
|
|
9036
|
|
372
|
9037 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
9038 {
|
|
|
9039 if (sps_flags & SPS_BEST)
|
|
|
9040 /* Adjust the word score for how it sounds like. */
|
|
|
9041 rescore_suggestions(su);
|
|
|
9042
|
|
|
9043 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
9044 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
9045 }
|
|
|
9046 }
|
|
|
9047
|
|
|
9048 /*
|
|
|
9049 * Free the info put in "*su" by spell_find_suggest().
|
|
|
9050 */
|
|
|
9051 static void
|
|
|
9052 spell_find_cleanup(su)
|
|
|
9053 suginfo_T *su;
|
|
|
9054 {
|
|
|
9055 int i;
|
|
|
9056
|
|
|
9057 /* Free the suggestions. */
|
|
|
9058 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9059 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
9060 ga_clear(&su->su_ga);
|
|
|
9061 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
9062 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
9063 ga_clear(&su->su_sga);
|
|
323
|
9064
|
|
|
9065 /* Free the banned words. */
|
|
344
|
9066 free_banned(su);
|
|
323
|
9067 }
|
|
|
9068
|
|
|
9069 /*
|
|
324
|
9070 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
9071 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
9072 * The result is NUL terminated.
|
|
323
|
9073 */
|
|
|
9074 static void
|
|
324
|
9075 onecap_copy(word, wcopy, upper)
|
|
323
|
9076 char_u *word;
|
|
|
9077 char_u *wcopy;
|
|
|
9078 int upper; /* TRUE: first letter made upper case */
|
|
|
9079 {
|
|
|
9080 char_u *p;
|
|
|
9081 int c;
|
|
|
9082 int l;
|
|
|
9083
|
|
|
9084 p = word;
|
|
|
9085 #ifdef FEAT_MBYTE
|
|
|
9086 if (has_mbyte)
|
|
474
|
9087 c = mb_cptr2char_adv(&p);
|
|
323
|
9088 else
|
|
|
9089 #endif
|
|
|
9090 c = *p++;
|
|
|
9091 if (upper)
|
|
324
|
9092 c = SPELL_TOUPPER(c);
|
|
323
|
9093 else
|
|
324
|
9094 c = SPELL_TOFOLD(c);
|
|
323
|
9095 #ifdef FEAT_MBYTE
|
|
|
9096 if (has_mbyte)
|
|
|
9097 l = mb_char2bytes(c, wcopy);
|
|
|
9098 else
|
|
|
9099 #endif
|
|
|
9100 {
|
|
|
9101 l = 1;
|
|
|
9102 wcopy[0] = c;
|
|
|
9103 }
|
|
376
|
9104 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
9105 }
|
|
|
9106
|
|
|
9107 /*
|
|
324
|
9108 * Make a copy of "word" with all the letters upper cased into
|
|
|
9109 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
9110 */
|
|
|
9111 static void
|
|
|
9112 allcap_copy(word, wcopy)
|
|
|
9113 char_u *word;
|
|
|
9114 char_u *wcopy;
|
|
|
9115 {
|
|
|
9116 char_u *s;
|
|
|
9117 char_u *d;
|
|
|
9118 int c;
|
|
|
9119
|
|
|
9120 d = wcopy;
|
|
|
9121 for (s = word; *s != NUL; )
|
|
|
9122 {
|
|
|
9123 #ifdef FEAT_MBYTE
|
|
|
9124 if (has_mbyte)
|
|
474
|
9125 c = mb_cptr2char_adv(&s);
|
|
323
|
9126 else
|
|
|
9127 #endif
|
|
|
9128 c = *s++;
|
|
492
|
9129
|
|
|
9130 #ifdef FEAT_MBYTE
|
|
|
9131 /* We only change ß to SS when we are certain latin1 is used. It
|
|
|
9132 * would cause weird errors in other 8-bit encodings. */
|
|
|
9133 if (enc_latin1like && c == 0xdf)
|
|
|
9134 {
|
|
|
9135 c = 'S';
|
|
|
9136 if (d - wcopy >= MAXWLEN - 1)
|
|
|
9137 break;
|
|
|
9138 *d++ = c;
|
|
|
9139 }
|
|
|
9140 else
|
|
|
9141 #endif
|
|
|
9142 c = SPELL_TOUPPER(c);
|
|
323
|
9143
|
|
|
9144 #ifdef FEAT_MBYTE
|
|
|
9145 if (has_mbyte)
|
|
|
9146 {
|
|
|
9147 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
9148 break;
|
|
|
9149 d += mb_char2bytes(c, d);
|
|
|
9150 }
|
|
|
9151 else
|
|
|
9152 #endif
|
|
|
9153 {
|
|
|
9154 if (d - wcopy >= MAXWLEN - 1)
|
|
|
9155 break;
|
|
|
9156 *d++ = c;
|
|
|
9157 }
|
|
|
9158 }
|
|
|
9159 *d = NUL;
|
|
|
9160 }
|
|
|
9161
|
|
|
9162 /*
|
|
346
|
9163 * Try finding suggestions by recognizing specific situations.
|
|
|
9164 */
|
|
|
9165 static void
|
|
|
9166 suggest_try_special(su)
|
|
|
9167 suginfo_T *su;
|
|
|
9168 {
|
|
|
9169 char_u *p;
|
|
376
|
9170 size_t len;
|
|
346
|
9171 int c;
|
|
|
9172 char_u word[MAXWLEN];
|
|
|
9173
|
|
|
9174 /*
|
|
|
9175 * Recognize a word that is repeated: "the the".
|
|
|
9176 */
|
|
|
9177 p = skiptowhite(su->su_fbadword);
|
|
|
9178 len = p - su->su_fbadword;
|
|
|
9179 p = skipwhite(p);
|
|
|
9180 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
9181 {
|
|
|
9182 /* Include badflags: if the badword is onecap or allcap
|
|
|
9183 * use that for the goodword too: "The the" -> "The". */
|
|
|
9184 c = su->su_fbadword[len];
|
|
|
9185 su->su_fbadword[len] = NUL;
|
|
|
9186 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
9187 su->su_fbadword[len] = c;
|
|
534
|
9188
|
|
|
9189 /* Give a soundalike score of 0, compute the score as if deleting one
|
|
|
9190 * character. */
|
|
|
9191 add_suggestion(su, &su->su_ga, word, su->su_badlen,
|
|
|
9192 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang);
|
|
346
|
9193 }
|
|
|
9194 }
|
|
|
9195
|
|
|
9196 /*
|
|
323
|
9197 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
9198 *
|
|
|
9199 * This uses a state machine. At each node in the tree we try various
|
|
|
9200 * operations. When trying if an operation work "depth" is increased and the
|
|
|
9201 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
9202 * character, replace one and delete another. The number of changes is
|
|
|
9203 * limited by su->su_maxscore, checked in try_deeper().
|
|
485
|
9204 *
|
|
|
9205 * After implementing this I noticed an article by Kemal Oflazer that
|
|
|
9206 * describes something similar: "Error-tolerant Finite State Recognition with
|
|
|
9207 * Applications to Morphological Analysis and Spelling Correction" (1996).
|
|
|
9208 * The implementation in the article is simplified and requires a stack of
|
|
|
9209 * unknown depth. The implementation here only needs a stack depth of the
|
|
|
9210 * length of the word.
|
|
323
|
9211 */
|
|
|
9212 static void
|
|
346
|
9213 suggest_try_change(su)
|
|
323
|
9214 suginfo_T *su;
|
|
|
9215 {
|
|
|
9216 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
9217 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
9218 trystate_T stack[MAXWLEN];
|
|
483
|
9219 char_u preword[MAXWLEN * 3]; /* word found with proper case;
|
|
|
9220 * concatanation of prefix compound
|
|
|
9221 * words and split word. NUL terminated
|
|
|
9222 * when going deeper but not when coming
|
|
|
9223 * back. */
|
|
|
9224 char_u compflags[MAXWLEN]; /* compound flags, one for each word */
|
|
323
|
9225 trystate_T *sp;
|
|
|
9226 int newscore;
|
|
|
9227 langp_T *lp;
|
|
375
|
9228 char_u *byts, *fbyts, *pbyts;
|
|
|
9229 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
9230 int depth;
|
|
330
|
9231 int c, c2, c3;
|
|
390
|
9232 int n;
|
|
323
|
9233 int flags;
|
|
|
9234 garray_T *gap;
|
|
324
|
9235 idx_T arridx;
|
|
323
|
9236 int len;
|
|
|
9237 char_u *p;
|
|
|
9238 fromto_T *ftp;
|
|
330
|
9239 int fl = 0, tl;
|
|
346
|
9240 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
482
|
9241 slang_T *slang;
|
|
|
9242 int fword_ends;
|
|
500
|
9243 int lpi;
|
|
534
|
9244 int maysplit;
|
|
|
9245 int goodword_ends;
|
|
323
|
9246
|
|
|
9247 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
9248 * to find matches (esp. REP items). Append some more text, changing
|
|
|
9249 * chars after the bad word may help. */
|
|
323
|
9250 STRCPY(fword, su->su_fbadword);
|
|
346
|
9251 n = STRLEN(fword);
|
|
|
9252 p = su->su_badptr + su->su_badlen;
|
|
|
9253 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
9254
|
|
503
|
9255 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
9256 {
|
|
|
9257 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
482
|
9258 slang = lp->lp_slang;
|
|
|
9259
|
|
500
|
9260 /* If reloading a spell file fails it's still in the list but
|
|
|
9261 * everything has been cleared. */
|
|
|
9262 if (slang->sl_fbyts == NULL)
|
|
|
9263 continue;
|
|
|
9264
|
|
323
|
9265 /*
|
|
|
9266 * Go through the whole case-fold tree, try changes at each node.
|
|
|
9267 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
9268 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
9269 * word).
|
|
|
9270 */
|
|
|
9271 depth = 0;
|
|
375
|
9272 sp = &stack[0];
|
|
483
|
9273 vim_memset(sp, 0, sizeof(trystate_T));
|
|
375
|
9274 sp->ts_curi = 1;
|
|
330
|
9275
|
|
|
9276 /*
|
|
375
|
9277 * When there are postponed prefixes we need to use these first. At
|
|
|
9278 * the end of the prefix we continue in the case-fold tree.
|
|
|
9279 */
|
|
482
|
9280 fbyts = slang->sl_fbyts;
|
|
|
9281 fidxs = slang->sl_fidxs;
|
|
|
9282 pbyts = slang->sl_pbyts;
|
|
|
9283 pidxs = slang->sl_pidxs;
|
|
375
|
9284 if (pbyts != NULL)
|
|
|
9285 {
|
|
|
9286 byts = pbyts;
|
|
|
9287 idxs = pidxs;
|
|
482
|
9288 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
375
|
9289 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
9290 }
|
|
|
9291 else
|
|
|
9292 {
|
|
|
9293 byts = fbyts;
|
|
|
9294 idxs = fidxs;
|
|
482
|
9295 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
485
|
9296 sp->ts_state = STATE_START;
|
|
375
|
9297 }
|
|
|
9298
|
|
|
9299 /*
|
|
330
|
9300 * Loop to find all suggestions. At each round we either:
|
|
|
9301 * - For the current state try one operation, advance "ts_curi",
|
|
|
9302 * increase "depth".
|
|
|
9303 * - When a state is done go to the next, set "ts_state".
|
|
|
9304 * - When all states are tried decrease "depth".
|
|
|
9305 */
|
|
323
|
9306 while (depth >= 0 && !got_int)
|
|
|
9307 {
|
|
|
9308 sp = &stack[depth];
|
|
|
9309 switch (sp->ts_state)
|
|
|
9310 {
|
|
|
9311 case STATE_START:
|
|
375
|
9312 case STATE_NOPREFIX:
|
|
323
|
9313 /*
|
|
|
9314 * Start of node: Deal with NUL bytes, which means
|
|
|
9315 * tword[] may end here.
|
|
|
9316 */
|
|
|
9317 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
9318 len = byts[arridx]; /* bytes in this node */
|
|
|
9319 arridx += sp->ts_curi; /* index of current byte */
|
|
|
9320
|
|
482
|
9321 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
9322 {
|
|
|
9323 /* Skip over the NUL bytes, we use them later. */
|
|
|
9324 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
9325 ;
|
|
|
9326 sp->ts_curi += n;
|
|
|
9327
|
|
390
|
9328 /* Always past NUL bytes now. */
|
|
|
9329 n = (int)sp->ts_state;
|
|
|
9330 sp->ts_state = STATE_ENDNUL;
|
|
455
|
9331 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
9332
|
|
375
|
9333 /* At end of a prefix or at start of prefixtree: check for
|
|
|
9334 * following word. */
|
|
390
|
9335 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
9336 {
|
|
455
|
9337 /* Set su->su_badflags to the caps type at this
|
|
|
9338 * position. Use the caps type until here for the
|
|
|
9339 * prefix itself. */
|
|
|
9340 #ifdef FEAT_MBYTE
|
|
|
9341 if (has_mbyte)
|
|
|
9342 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
9343 else
|
|
|
9344 #endif
|
|
|
9345 n = sp->ts_fidx;
|
|
474
|
9346 flags = badword_captype(su->su_badptr,
|
|
|
9347 su->su_badptr + n);
|
|
|
9348 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
9349 su->su_badptr + su->su_badlen);
|
|
375
|
9350 ++depth;
|
|
|
9351 stack[depth] = stack[depth - 1];
|
|
|
9352 sp = &stack[depth];
|
|
|
9353 sp->ts_prefixdepth = depth - 1;
|
|
|
9354 byts = fbyts;
|
|
|
9355 idxs = fidxs;
|
|
|
9356 sp->ts_state = STATE_START;
|
|
|
9357 sp->ts_curi = 1; /* start just after length byte */
|
|
|
9358 sp->ts_arridx = 0;
|
|
|
9359
|
|
455
|
9360 /* Move the prefix to preword[] with the right case
|
|
|
9361 * and make find_keepcap_word() works. */
|
|
485
|
9362 tword[sp->ts_twordlen] = NUL;
|
|
|
9363 make_case_word(tword + sp->ts_splitoff,
|
|
|
9364 preword + sp->ts_prewordlen,
|
|
|
9365 flags);
|
|
|
9366 sp->ts_prewordlen = STRLEN(preword);
|
|
483
|
9367 sp->ts_splitoff = sp->ts_twordlen;
|
|
375
|
9368 }
|
|
|
9369 break;
|
|
|
9370 }
|
|
|
9371
|
|
346
|
9372 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
9373 {
|
|
|
9374 /* Past bytes in node and/or past NUL bytes. */
|
|
|
9375 sp->ts_state = STATE_ENDNUL;
|
|
455
|
9376 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
9377 break;
|
|
|
9378 }
|
|
|
9379
|
|
|
9380 /*
|
|
|
9381 * End of word in tree.
|
|
|
9382 */
|
|
|
9383 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
9384
|
|
324
|
9385 flags = (int)idxs[arridx];
|
|
483
|
9386 fword_ends = (fword[sp->ts_fidx] == NUL
|
|
|
9387 || !spell_iswordp(fword + sp->ts_fidx, curbuf));
|
|
|
9388 tword[sp->ts_twordlen] = NUL;
|
|
323
|
9389
|
|
485
|
9390 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL
|
|
|
9391 && (sp->ts_flags & TSF_PREFIXOK) == 0)
|
|
375
|
9392 {
|
|
|
9393 /* There was a prefix before the word. Check that the
|
|
|
9394 * prefix can be used with this word. */
|
|
|
9395 /* Count the length of the NULs in the prefix. If there
|
|
|
9396 * are none this must be the first try without a prefix.
|
|
|
9397 */
|
|
|
9398 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
9399 len = pbyts[n++];
|
|
|
9400 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
9401 ;
|
|
|
9402 if (c > 0)
|
|
|
9403 {
|
|
390
|
9404 c = valid_word_prefix(c, n, flags,
|
|
483
|
9405 tword + sp->ts_splitoff, slang, FALSE);
|
|
375
|
9406 if (c == 0)
|
|
|
9407 break;
|
|
|
9408
|
|
|
9409 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
9410 if (c & WF_RAREPFX)
|
|
|
9411 flags |= WF_RARE;
|
|
485
|
9412
|
|
|
9413 /* Tricky: when checking for both prefix and
|
|
|
9414 * compounding we run into the prefix flag first.
|
|
|
9415 * Remember that it's OK, so that we accept the prefix
|
|
|
9416 * when arriving at a compound flag. */
|
|
|
9417 sp->ts_flags |= TSF_PREFIXOK;
|
|
375
|
9418 }
|
|
|
9419 }
|
|
|
9420
|
|
534
|
9421 /* Check NEEDCOMPOUND: can't use word without compounding. Do
|
|
|
9422 * try appending another compound word below. */
|
|
500
|
9423 if (sp->ts_complen == sp->ts_compsplit && fword_ends
|
|
|
9424 && (flags & WF_NEEDCOMP))
|
|
534
|
9425 goodword_ends = FALSE;
|
|
|
9426 else
|
|
|
9427 goodword_ends = TRUE;
|
|
500
|
9428
|
|
485
|
9429 if (sp->ts_complen > sp->ts_compsplit)
|
|
|
9430 {
|
|
492
|
9431 if (slang->sl_nobreak)
|
|
|
9432 {
|
|
|
9433 /* There was a word before this word. When there was
|
|
|
9434 * no change in this word (it was correct) add the
|
|
|
9435 * first word as a suggestion. If this word was
|
|
|
9436 * corrected too, we need to check if a correct word
|
|
|
9437 * follows. */
|
|
|
9438 if (sp->ts_fidx - sp->ts_splitfidx
|
|
|
9439 == sp->ts_twordlen - sp->ts_splitoff
|
|
|
9440 && STRNCMP(fword + sp->ts_splitfidx,
|
|
|
9441 tword + sp->ts_splitoff,
|
|
|
9442 sp->ts_fidx - sp->ts_splitfidx) == 0)
|
|
|
9443 {
|
|
|
9444 preword[sp->ts_prewordlen] = NUL;
|
|
|
9445 add_suggestion(su, &su->su_ga, preword,
|
|
503
|
9446 sp->ts_splitfidx - repextra,
|
|
|
9447 sp->ts_score, 0, FALSE,
|
|
|
9448 lp->lp_sallang);
|
|
492
|
9449 break;
|
|
|
9450 }
|
|
|
9451 }
|
|
|
9452 else
|
|
|
9453 {
|
|
|
9454 /* There was a compound word before this word. If
|
|
|
9455 * this word does not support compounding then give up
|
|
|
9456 * (splitting is tried for the word without compound
|
|
|
9457 * flag). */
|
|
|
9458 if (((unsigned)flags >> 24) == 0
|
|
|
9459 || sp->ts_twordlen - sp->ts_splitoff
|
|
485
|
9460 < slang->sl_compminlen)
|
|
492
|
9461 break;
|
|
500
|
9462 #ifdef FEAT_MBYTE
|
|
|
9463 /* For multi-byte chars check character length against
|
|
|
9464 * COMPOUNDMIN. */
|
|
|
9465 if (has_mbyte
|
|
501
|
9466 && slang->sl_compminlen > 0
|
|
500
|
9467 && mb_charlen(tword + sp->ts_splitoff)
|
|
|
9468 < slang->sl_compminlen)
|
|
|
9469 break;
|
|
|
9470 #endif
|
|
|
9471
|
|
492
|
9472 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
9473 compflags[sp->ts_complen + 1] = NUL;
|
|
|
9474 vim_strncpy(preword + sp->ts_prewordlen,
|
|
|
9475 tword + sp->ts_splitoff,
|
|
|
9476 sp->ts_twordlen - sp->ts_splitoff);
|
|
|
9477 p = preword;
|
|
|
9478 while (*skiptowhite(p) != NUL)
|
|
|
9479 p = skipwhite(skiptowhite(p));
|
|
|
9480 if (fword_ends && !can_compound(slang, p,
|
|
485
|
9481 compflags + sp->ts_compsplit))
|
|
492
|
9482 break;
|
|
|
9483
|
|
|
9484 /* Get pointer to last char of previous word. */
|
|
|
9485 p = preword + sp->ts_prewordlen;
|
|
|
9486 mb_ptr_back(preword, p);
|
|
|
9487 }
|
|
490
|
9488 }
|
|
|
9489 else
|
|
|
9490 p = NULL;
|
|
485
|
9491
|
|
323
|
9492 /*
|
|
|
9493 * Form the word with proper case in preword.
|
|
|
9494 * If there is a word from a previous split, append.
|
|
|
9495 */
|
|
|
9496 if (flags & WF_KEEPCAP)
|
|
|
9497 /* Must find the word in the keep-case tree. */
|
|
483
|
9498 find_keepcap_word(slang, tword + sp->ts_splitoff,
|
|
|
9499 preword + sp->ts_prewordlen);
|
|
323
|
9500 else
|
|
346
|
9501 {
|
|
323
|
9502 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
9503 * use that for the goodword too. But if the badword is
|
|
|
9504 * allcap and it's only one char long use onecap. */
|
|
|
9505 c = su->su_badflags;
|
|
|
9506 if ((c & WF_ALLCAP)
|
|
|
9507 #ifdef FEAT_MBYTE
|
|
474
|
9508 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
9509 #else
|
|
|
9510 && su->su_badlen == 1
|
|
|
9511 #endif
|
|
|
9512 )
|
|
|
9513 c = WF_ONECAP;
|
|
490
|
9514 c |= flags;
|
|
|
9515
|
|
|
9516 /* When appending a compound word after a word character
|
|
|
9517 * don't use Onecap. */
|
|
|
9518 if (p != NULL && spell_iswordp_nmw(p))
|
|
|
9519 c &= ~WF_ONECAP;
|
|
483
|
9520 make_case_word(tword + sp->ts_splitoff,
|
|
490
|
9521 preword + sp->ts_prewordlen, c);
|
|
346
|
9522 }
|
|
323
|
9523
|
|
|
9524 /* Don't use a banned word. It may appear again as a good
|
|
|
9525 * word, thus remember it. */
|
|
|
9526 if (flags & WF_BANNED)
|
|
|
9527 {
|
|
483
|
9528 add_banned(su, preword + sp->ts_prewordlen);
|
|
323
|
9529 break;
|
|
|
9530 }
|
|
534
|
9531 if ((sp->ts_complen == sp->ts_compsplit
|
|
|
9532 && was_banned(su, preword + sp->ts_prewordlen))
|
|
372
|
9533 || was_banned(su, preword))
|
|
534
|
9534 {
|
|
|
9535 if (slang->sl_compprog == NULL)
|
|
|
9536 break;
|
|
|
9537 /* the word so far was banned but we may try compounding */
|
|
|
9538 goodword_ends = FALSE;
|
|
|
9539 }
|
|
323
|
9540
|
|
|
9541 newscore = 0;
|
|
|
9542 if ((flags & WF_REGION)
|
|
390
|
9543 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
9544 newscore += SCORE_REGION;
|
|
|
9545 if (flags & WF_RARE)
|
|
|
9546 newscore += SCORE_RARE;
|
|
|
9547
|
|
346
|
9548 if (!spell_valid_case(su->su_badflags,
|
|
483
|
9549 captype(preword + sp->ts_prewordlen, NULL)))
|
|
323
|
9550 newscore += SCORE_ICASE;
|
|
|
9551
|
|
534
|
9552 maysplit = TRUE;
|
|
|
9553 if (fword_ends && goodword_ends
|
|
|
9554 && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
9555 {
|
|
366
|
9556 /* The badword also ends: add suggestions. Give a penalty
|
|
|
9557 * when changing non-word char to word char, e.g., "thes,"
|
|
|
9558 * -> "these". */
|
|
|
9559 p = fword + sp->ts_fidx;
|
|
|
9560 #ifdef FEAT_MBYTE
|
|
|
9561 if (has_mbyte)
|
|
|
9562 mb_ptr_back(fword, p);
|
|
|
9563 else
|
|
|
9564 #endif
|
|
|
9565 --p;
|
|
376
|
9566 if (!spell_iswordp(p, curbuf))
|
|
366
|
9567 {
|
|
|
9568 p = preword + STRLEN(preword);
|
|
|
9569 #ifdef FEAT_MBYTE
|
|
|
9570 if (has_mbyte)
|
|
|
9571 mb_ptr_back(preword, p);
|
|
|
9572 else
|
|
|
9573 #endif
|
|
|
9574 --p;
|
|
376
|
9575 if (spell_iswordp(p, curbuf))
|
|
366
|
9576 newscore += SCORE_NONWORD;
|
|
|
9577 }
|
|
|
9578
|
|
344
|
9579 add_suggestion(su, &su->su_ga, preword,
|
|
534
|
9580 sp->ts_fidx - repextra,
|
|
|
9581 sp->ts_score + newscore, 0, FALSE,
|
|
|
9582 lp->lp_sallang);
|
|
|
9583
|
|
|
9584 /* When the bad word doesn't end yet, try changing the
|
|
|
9585 * next word. E.g., find suggestions for "the the" where
|
|
|
9586 * the second "the" is different. It's done like a split.
|
|
|
9587 */
|
|
|
9588 if (sp->ts_fidx - repextra >= su->su_badlen)
|
|
|
9589 maysplit = FALSE;
|
|
|
9590 }
|
|
|
9591
|
|
|
9592 if (maysplit
|
|
|
9593 && (sp->ts_fidx >= sp->ts_fidxtry || fword_ends)
|
|
330
|
9594 #ifdef FEAT_MBYTE
|
|
|
9595 /* Don't split halfway a character. */
|
|
|
9596 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
9597 #endif
|
|
|
9598 )
|
|
323
|
9599 {
|
|
482
|
9600 int try_compound;
|
|
|
9601
|
|
|
9602 /* Get here in two situations:
|
|
|
9603 * 1. The word in the tree ends but the badword continues:
|
|
|
9604 * If the word allows compounding try that. Otherwise
|
|
|
9605 * try a split by inserting a space. For both check
|
|
|
9606 * that a valid words starts at fword[sp->ts_fidx].
|
|
492
|
9607 * For NOBREAK do like compounding to be able to check
|
|
|
9608 * if the next word is valid.
|
|
482
|
9609 * 2. The badword does end, but it was due to a change
|
|
|
9610 * (e.g., a swap). No need to split, but do check that
|
|
|
9611 * the following word is valid.
|
|
|
9612 */
|
|
485
|
9613 try_compound = FALSE;
|
|
534
|
9614 if ((!fword_ends || !goodword_ends)
|
|
495
|
9615 && slang->sl_compprog != NULL
|
|
483
|
9616 && ((unsigned)flags >> 24) != 0
|
|
|
9617 && sp->ts_twordlen - sp->ts_splitoff
|
|
482
|
9618 >= slang->sl_compminlen
|
|
500
|
9619 #ifdef FEAT_MBYTE
|
|
|
9620 && (!has_mbyte
|
|
501
|
9621 || slang->sl_compminlen == 0
|
|
500
|
9622 || mb_charlen(tword + sp->ts_splitoff)
|
|
|
9623 >= slang->sl_compminlen)
|
|
|
9624 #endif
|
|
490
|
9625 && (slang->sl_compsylmax < MAXWLEN
|
|
|
9626 || sp->ts_complen + 1 - sp->ts_compsplit
|
|
|
9627 < slang->sl_compmax)
|
|
495
|
9628 && (byte_in_str(sp->ts_complen == sp->ts_compsplit
|
|
485
|
9629 ? slang->sl_compstartflags
|
|
|
9630 : slang->sl_compallflags,
|
|
495
|
9631 ((unsigned)flags >> 24))))
|
|
483
|
9632 {
|
|
482
|
9633 try_compound = TRUE;
|
|
483
|
9634 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
9635 compflags[sp->ts_complen + 1] = NUL;
|
|
|
9636 }
|
|
485
|
9637
|
|
492
|
9638 /* For NOBREAK we never try splitting, it won't make any
|
|
|
9639 * word valid. */
|
|
|
9640 if (slang->sl_nobreak)
|
|
|
9641 try_compound = TRUE;
|
|
|
9642
|
|
485
|
9643 /* If we could add a compound word, and it's also possible
|
|
|
9644 * to split at this point, do the split first and set
|
|
|
9645 * TSF_DIDSPLIT to avoid doing it again. */
|
|
492
|
9646 else if (!fword_ends
|
|
485
|
9647 && try_compound
|
|
|
9648 && (sp->ts_flags & TSF_DIDSPLIT) == 0)
|
|
482
|
9649 {
|
|
|
9650 try_compound = FALSE;
|
|
485
|
9651 sp->ts_flags |= TSF_DIDSPLIT;
|
|
|
9652 --sp->ts_curi; /* do the same NUL again */
|
|
|
9653 compflags[sp->ts_complen] = NUL;
|
|
|
9654 }
|
|
|
9655 else
|
|
|
9656 sp->ts_flags &= ~TSF_DIDSPLIT;
|
|
|
9657
|
|
534
|
9658 if (!try_compound && (!fword_ends || !goodword_ends))
|
|
485
|
9659 {
|
|
|
9660 /* If we're going to split need to check that the
|
|
501
|
9661 * words so far are valid for compounding. If there
|
|
|
9662 * is only one word it must not have the NEEDCOMPOUND
|
|
|
9663 * flag. */
|
|
|
9664 if (sp->ts_complen == sp->ts_compsplit
|
|
|
9665 && (flags & WF_NEEDCOMP))
|
|
|
9666 break;
|
|
490
|
9667 p = preword;
|
|
|
9668 while (*skiptowhite(p) != NUL)
|
|
|
9669 p = skipwhite(skiptowhite(p));
|
|
485
|
9670 if (sp->ts_complen > sp->ts_compsplit
|
|
490
|
9671 && !can_compound(slang, p,
|
|
485
|
9672 compflags + sp->ts_compsplit))
|
|
|
9673 break;
|
|
|
9674 newscore += SCORE_SPLIT;
|
|
482
|
9675 }
|
|
|
9676
|
|
|
9677 if (try_deeper(su, stack, depth, newscore))
|
|
323
|
9678 {
|
|
|
9679 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
346
|
9680 sp->ts_save_badflags = su->su_badflags;
|
|
376
|
9681 sp->ts_state = STATE_SPLITUNDO;
|
|
|
9682
|
|
|
9683 ++depth;
|
|
|
9684 sp = &stack[depth];
|
|
323
|
9685
|
|
482
|
9686 /* Append a space to preword when splitting. */
|
|
|
9687 if (!try_compound && !fword_ends)
|
|
|
9688 STRCAT(preword, " ");
|
|
483
|
9689 sp->ts_prewordlen = STRLEN(preword);
|
|
|
9690 sp->ts_splitoff = sp->ts_twordlen;
|
|
492
|
9691 sp->ts_splitfidx = sp->ts_fidx;
|
|
376
|
9692
|
|
|
9693 /* If the badword has a non-word character at this
|
|
|
9694 * position skip it. That means replacing the
|
|
482
|
9695 * non-word character with a space. Always skip a
|
|
534
|
9696 * character when the word ends. But only when the
|
|
|
9697 * good word can end. */
|
|
|
9698 if (((!try_compound
|
|
|
9699 && !spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
482
|
9700 || fword_ends)
|
|
534
|
9701 && goodword_ends)
|
|
376
|
9702 {
|
|
482
|
9703 int l;
|
|
|
9704
|
|
376
|
9705 #ifdef FEAT_MBYTE
|
|
|
9706 if (has_mbyte)
|
|
482
|
9707 l = MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
376
|
9708 else
|
|
|
9709 #endif
|
|
482
|
9710 l = 1;
|
|
|
9711 if (fword_ends)
|
|
|
9712 {
|
|
|
9713 /* Copy the skipped character to preword. */
|
|
483
|
9714 mch_memmove(preword + sp->ts_prewordlen,
|
|
482
|
9715 fword + sp->ts_fidx, l);
|
|
483
|
9716 sp->ts_prewordlen += l;
|
|
|
9717 preword[sp->ts_prewordlen] = NUL;
|
|
482
|
9718 }
|
|
|
9719 else
|
|
|
9720 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
9721 sp->ts_fidx += l;
|
|
376
|
9722 }
|
|
455
|
9723
|
|
485
|
9724 /* When compounding include compound flag in
|
|
|
9725 * compflags[] (already set above). When splitting we
|
|
|
9726 * may start compounding over again. */
|
|
482
|
9727 if (try_compound)
|
|
483
|
9728 ++sp->ts_complen;
|
|
482
|
9729 else
|
|
485
|
9730 sp->ts_compsplit = sp->ts_complen;
|
|
|
9731 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
482
|
9732
|
|
455
|
9733 /* set su->su_badflags to the caps type at this
|
|
|
9734 * position */
|
|
324
|
9735 #ifdef FEAT_MBYTE
|
|
|
9736 if (has_mbyte)
|
|
455
|
9737 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
9738 else
|
|
|
9739 #endif
|
|
455
|
9740 n = sp->ts_fidx;
|
|
474
|
9741 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
9742 su->su_badptr + su->su_badlen);
|
|
323
|
9743
|
|
|
9744 /* Restart at top of the tree. */
|
|
376
|
9745 sp->ts_arridx = 0;
|
|
485
|
9746
|
|
|
9747 /* If there are postponed prefixes, try these too. */
|
|
|
9748 if (pbyts != NULL)
|
|
|
9749 {
|
|
|
9750 byts = pbyts;
|
|
|
9751 idxs = pidxs;
|
|
|
9752 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
|
9753 sp->ts_state = STATE_NOPREFIX;
|
|
|
9754 }
|
|
323
|
9755 }
|
|
|
9756 }
|
|
|
9757 break;
|
|
|
9758
|
|
|
9759 case STATE_SPLITUNDO:
|
|
482
|
9760 /* Undo the changes done for word split or compound word. */
|
|
346
|
9761 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
9762
|
|
|
9763 /* Continue looking for NUL bytes. */
|
|
|
9764 sp->ts_state = STATE_START;
|
|
485
|
9765
|
|
|
9766 /* In case we went into the prefix tree. */
|
|
|
9767 byts = fbyts;
|
|
|
9768 idxs = fidxs;
|
|
323
|
9769 break;
|
|
|
9770
|
|
|
9771 case STATE_ENDNUL:
|
|
|
9772 /* Past the NUL bytes in the node. */
|
|
455
|
9773 su->su_badflags = sp->ts_save_badflags;
|
|
501
|
9774 if (fword[sp->ts_fidx] == NUL
|
|
|
9775 #ifdef FEAT_MBYTE
|
|
|
9776 && sp->ts_tcharlen == 0
|
|
|
9777 #endif
|
|
|
9778 )
|
|
323
|
9779 {
|
|
|
9780 /* The badword ends, can't use the bytes in this node. */
|
|
|
9781 sp->ts_state = STATE_DEL;
|
|
|
9782 break;
|
|
|
9783 }
|
|
|
9784 sp->ts_state = STATE_PLAIN;
|
|
|
9785 /*FALLTHROUGH*/
|
|
|
9786
|
|
|
9787 case STATE_PLAIN:
|
|
|
9788 /*
|
|
|
9789 * Go over all possible bytes at this node, add each to
|
|
|
9790 * tword[] and use child node. "ts_curi" is the index.
|
|
|
9791 */
|
|
|
9792 arridx = sp->ts_arridx;
|
|
|
9793 if (sp->ts_curi > byts[arridx])
|
|
|
9794 {
|
|
|
9795 /* Done all bytes at this node, do next state. When still
|
|
|
9796 * at already changed bytes skip the other tricks. */
|
|
|
9797 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
9798 sp->ts_state = STATE_DEL;
|
|
|
9799 else
|
|
|
9800 sp->ts_state = STATE_FINAL;
|
|
|
9801 }
|
|
|
9802 else
|
|
|
9803 {
|
|
|
9804 arridx += sp->ts_curi++;
|
|
|
9805 c = byts[arridx];
|
|
|
9806
|
|
|
9807 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
9808 * the byte in the bad word adjust the score. But don't
|
|
|
9809 * even try when the byte was already changed. */
|
|
330
|
9810 if (c == fword[sp->ts_fidx]
|
|
|
9811 #ifdef FEAT_MBYTE
|
|
|
9812 || (sp->ts_tcharlen > 0
|
|
|
9813 && sp->ts_isdiff != DIFF_NONE)
|
|
|
9814 #endif
|
|
|
9815 )
|
|
323
|
9816 newscore = 0;
|
|
|
9817 else
|
|
|
9818 newscore = SCORE_SUBST;
|
|
|
9819 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
9820 && try_deeper(su, stack, depth, newscore))
|
|
|
9821 {
|
|
|
9822 ++depth;
|
|
330
|
9823 sp = &stack[depth];
|
|
|
9824 ++sp->ts_fidx;
|
|
|
9825 tword[sp->ts_twordlen++] = c;
|
|
|
9826 sp->ts_arridx = idxs[arridx];
|
|
|
9827 #ifdef FEAT_MBYTE
|
|
|
9828 if (newscore == SCORE_SUBST)
|
|
|
9829 sp->ts_isdiff = DIFF_YES;
|
|
|
9830 if (has_mbyte)
|
|
|
9831 {
|
|
|
9832 /* Multi-byte characters are a bit complicated to
|
|
|
9833 * handle: They differ when any of the bytes
|
|
|
9834 * differ and then their length may also differ. */
|
|
|
9835 if (sp->ts_tcharlen == 0)
|
|
|
9836 {
|
|
|
9837 /* First byte. */
|
|
|
9838 sp->ts_tcharidx = 0;
|
|
|
9839 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
9840 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
9841 sp->ts_isdiff = (newscore != 0)
|
|
|
9842 ? DIFF_YES : DIFF_NONE;
|
|
|
9843 }
|
|
|
9844 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
9845 /* When inserting trail bytes don't advance in
|
|
|
9846 * the bad word. */
|
|
|
9847 --sp->ts_fidx;
|
|
|
9848 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
9849 {
|
|
|
9850 /* Last byte of character. */
|
|
|
9851 if (sp->ts_isdiff == DIFF_YES)
|
|
|
9852 {
|
|
|
9853 /* Correct ts_fidx for the byte length of
|
|
|
9854 * the character (we didn't check that
|
|
|
9855 * before). */
|
|
|
9856 sp->ts_fidx = sp->ts_fcharstart
|
|
|
9857 + MB_BYTE2LEN(
|
|
|
9858 fword[sp->ts_fcharstart]);
|
|
|
9859
|
|
479
|
9860 /* For changing a composing character
|
|
|
9861 * adjust the score from SCORE_SUBST to
|
|
|
9862 * SCORE_SUBCOMP. */
|
|
|
9863 if (enc_utf8
|
|
|
9864 && utf_iscomposing(
|
|
|
9865 mb_ptr2char(tword
|
|
|
9866 + sp->ts_twordlen
|
|
|
9867 - sp->ts_tcharlen))
|
|
|
9868 && utf_iscomposing(
|
|
|
9869 mb_ptr2char(fword
|
|
|
9870 + sp->ts_fcharstart)))
|
|
|
9871 sp->ts_score -=
|
|
|
9872 SCORE_SUBST - SCORE_SUBCOMP;
|
|
|
9873
|
|
330
|
9874 /* For a similar character adjust score
|
|
|
9875 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
482
|
9876 else if (slang->sl_has_map
|
|
|
9877 && similar_chars(slang,
|
|
330
|
9878 mb_ptr2char(tword
|
|
|
9879 + sp->ts_twordlen
|
|
|
9880 - sp->ts_tcharlen),
|
|
|
9881 mb_ptr2char(fword
|
|
|
9882 + sp->ts_fcharstart)))
|
|
|
9883 sp->ts_score -=
|
|
|
9884 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9885 }
|
|
358
|
9886 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
9887 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
9888 {
|
|
|
9889 p = tword + sp->ts_twordlen
|
|
|
9890 - sp->ts_tcharlen;
|
|
|
9891 c = mb_ptr2char(p);
|
|
475
|
9892 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9893 {
|
|
|
9894 /* Inserting a composing char doesn't
|
|
|
9895 * count that much. */
|
|
358
|
9896 sp->ts_score -= SCORE_INS
|
|
475
|
9897 - SCORE_INSCOMP;
|
|
|
9898 }
|
|
|
9899 else
|
|
|
9900 {
|
|
|
9901 /* If the previous character was the
|
|
|
9902 * same, thus doubling a character,
|
|
|
9903 * give a bonus to the score. */
|
|
|
9904 mb_ptr_back(tword, p);
|
|
|
9905 if (c == mb_ptr2char(p))
|
|
|
9906 sp->ts_score -= SCORE_INS
|
|
358
|
9907 - SCORE_INSDUP;
|
|
475
|
9908 }
|
|
358
|
9909 }
|
|
330
|
9910
|
|
|
9911 /* Starting a new char, reset the length. */
|
|
|
9912 sp->ts_tcharlen = 0;
|
|
|
9913 }
|
|
|
9914 }
|
|
|
9915 else
|
|
|
9916 #endif
|
|
|
9917 {
|
|
|
9918 /* If we found a similar char adjust the score.
|
|
|
9919 * We do this after calling try_deeper() because
|
|
|
9920 * it's slow. */
|
|
|
9921 if (newscore != 0
|
|
482
|
9922 && slang->sl_has_map
|
|
|
9923 && similar_chars(slang,
|
|
330
|
9924 c, fword[sp->ts_fidx - 1]))
|
|
|
9925 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9926 }
|
|
323
|
9927 }
|
|
|
9928 }
|
|
|
9929 break;
|
|
|
9930
|
|
|
9931 case STATE_DEL:
|
|
330
|
9932 #ifdef FEAT_MBYTE
|
|
|
9933 /* When past the first byte of a multi-byte char don't try
|
|
|
9934 * delete/insert/swap a character. */
|
|
|
9935 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
9936 {
|
|
|
9937 sp->ts_state = STATE_FINAL;
|
|
|
9938 break;
|
|
|
9939 }
|
|
|
9940 #endif
|
|
|
9941 /*
|
|
|
9942 * Try skipping one character in the bad word (delete it).
|
|
|
9943 */
|
|
323
|
9944 sp->ts_state = STATE_INS;
|
|
|
9945 sp->ts_curi = 1;
|
|
|
9946 if (fword[sp->ts_fidx] != NUL
|
|
|
9947 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
9948 {
|
|
|
9949 ++depth;
|
|
358
|
9950
|
|
|
9951 /* Advance over the character in fword[]. Give a bonus to
|
|
|
9952 * the score if the same character is following "nn" ->
|
|
|
9953 * "n". */
|
|
330
|
9954 #ifdef FEAT_MBYTE
|
|
|
9955 if (has_mbyte)
|
|
358
|
9956 {
|
|
|
9957 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
9958 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
479
|
9959 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9960 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
|
|
|
9961 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
358
|
9962 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
9963 }
|
|
330
|
9964 else
|
|
|
9965 #endif
|
|
358
|
9966 {
|
|
330
|
9967 ++stack[depth].ts_fidx;
|
|
358
|
9968 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
9969 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
9970 }
|
|
323
|
9971 break;
|
|
|
9972 }
|
|
|
9973 /*FALLTHROUGH*/
|
|
|
9974
|
|
|
9975 case STATE_INS:
|
|
330
|
9976 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
9977 * node. */
|
|
|
9978 n = sp->ts_arridx;
|
|
|
9979 if (sp->ts_curi > byts[n])
|
|
|
9980 {
|
|
|
9981 /* Done all bytes at this node, do next state. */
|
|
|
9982 sp->ts_state = STATE_SWAP;
|
|
|
9983 }
|
|
|
9984 else
|
|
|
9985 {
|
|
330
|
9986 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
9987 n += sp->ts_curi++;
|
|
|
9988 c = byts[n];
|
|
|
9989 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
9990 {
|
|
|
9991 ++depth;
|
|
330
|
9992 sp = &stack[depth];
|
|
|
9993 tword[sp->ts_twordlen++] = c;
|
|
|
9994 sp->ts_arridx = idxs[n];
|
|
|
9995 #ifdef FEAT_MBYTE
|
|
|
9996 if (has_mbyte)
|
|
|
9997 {
|
|
|
9998 fl = MB_BYTE2LEN(c);
|
|
|
9999 if (fl > 1)
|
|
|
10000 {
|
|
|
10001 /* There are following bytes for the same
|
|
|
10002 * character. We must find all bytes before
|
|
|
10003 * trying delete/insert/swap/etc. */
|
|
|
10004 sp->ts_tcharlen = fl;
|
|
|
10005 sp->ts_tcharidx = 1;
|
|
|
10006 sp->ts_isdiff = DIFF_INSERT;
|
|
|
10007 }
|
|
|
10008 }
|
|
358
|
10009 else
|
|
|
10010 fl = 1;
|
|
|
10011 if (fl == 1)
|
|
330
|
10012 #endif
|
|
358
|
10013 {
|
|
|
10014 /* If the previous character was the same, thus
|
|
|
10015 * doubling a character, give a bonus to the
|
|
|
10016 * score. */
|
|
|
10017 if (sp->ts_twordlen >= 2
|
|
|
10018 && tword[sp->ts_twordlen - 2] == c)
|
|
|
10019 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
10020 }
|
|
323
|
10021 }
|
|
|
10022 }
|
|
|
10023 break;
|
|
|
10024
|
|
|
10025 case STATE_SWAP:
|
|
330
|
10026 /*
|
|
|
10027 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
10028 * We change "fword" here, it's changed back afterwards.
|
|
|
10029 */
|
|
|
10030 p = fword + sp->ts_fidx;
|
|
|
10031 c = *p;
|
|
|
10032 if (c == NUL)
|
|
|
10033 {
|
|
|
10034 /* End of word, can't swap or replace. */
|
|
|
10035 sp->ts_state = STATE_FINAL;
|
|
|
10036 break;
|
|
|
10037 }
|
|
|
10038 #ifdef FEAT_MBYTE
|
|
|
10039 if (has_mbyte)
|
|
|
10040 {
|
|
474
|
10041 n = mb_cptr2len(p);
|
|
330
|
10042 c = mb_ptr2char(p);
|
|
|
10043 c2 = mb_ptr2char(p + n);
|
|
|
10044 }
|
|
|
10045 else
|
|
|
10046 #endif
|
|
|
10047 c2 = p[1];
|
|
|
10048 if (c == c2)
|
|
323
|
10049 {
|
|
330
|
10050 /* Characters are identical, swap won't do anything. */
|
|
|
10051 sp->ts_state = STATE_SWAP3;
|
|
|
10052 break;
|
|
|
10053 }
|
|
|
10054 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
10055 {
|
|
|
10056 sp->ts_state = STATE_UNSWAP;
|
|
323
|
10057 ++depth;
|
|
330
|
10058 #ifdef FEAT_MBYTE
|
|
|
10059 if (has_mbyte)
|
|
|
10060 {
|
|
|
10061 fl = mb_char2len(c2);
|
|
|
10062 mch_memmove(p, p + n, fl);
|
|
|
10063 mb_char2bytes(c, p + fl);
|
|
|
10064 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
10065 }
|
|
|
10066 else
|
|
|
10067 #endif
|
|
|
10068 {
|
|
|
10069 p[0] = c2;
|
|
|
10070 p[1] = c;
|
|
|
10071 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
10072 }
|
|
323
|
10073 }
|
|
|
10074 else
|
|
|
10075 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
10076 sp->ts_state = STATE_REP_INI;
|
|
|
10077 break;
|
|
|
10078
|
|
330
|
10079 case STATE_UNSWAP:
|
|
|
10080 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
10081 p = fword + sp->ts_fidx;
|
|
|
10082 #ifdef FEAT_MBYTE
|
|
|
10083 if (has_mbyte)
|
|
|
10084 {
|
|
|
10085 n = MB_BYTE2LEN(*p);
|
|
|
10086 c = mb_ptr2char(p + n);
|
|
|
10087 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
10088 mb_char2bytes(c, p);
|
|
|
10089 }
|
|
|
10090 else
|
|
|
10091 #endif
|
|
|
10092 {
|
|
|
10093 c = *p;
|
|
|
10094 *p = p[1];
|
|
|
10095 p[1] = c;
|
|
|
10096 }
|
|
|
10097 /*FALLTHROUGH*/
|
|
|
10098
|
|
|
10099 case STATE_SWAP3:
|
|
323
|
10100 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
10101 * "fword" here, it's changed back afterwards. */
|
|
|
10102 p = fword + sp->ts_fidx;
|
|
|
10103 #ifdef FEAT_MBYTE
|
|
|
10104 if (has_mbyte)
|
|
|
10105 {
|
|
474
|
10106 n = mb_cptr2len(p);
|
|
330
|
10107 c = mb_ptr2char(p);
|
|
474
|
10108 fl = mb_cptr2len(p + n);
|
|
330
|
10109 c2 = mb_ptr2char(p + n);
|
|
|
10110 c3 = mb_ptr2char(p + n + fl);
|
|
|
10111 }
|
|
|
10112 else
|
|
|
10113 #endif
|
|
323
|
10114 {
|
|
330
|
10115 c = *p;
|
|
|
10116 c2 = p[1];
|
|
|
10117 c3 = p[2];
|
|
|
10118 }
|
|
|
10119
|
|
|
10120 /* When characters are identical: "121" then SWAP3 result is
|
|
|
10121 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
10122 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
10123 * swapping. Also skip when c3 is NUL. */
|
|
|
10124 if (c == c3 || c3 == NUL)
|
|
|
10125 {
|
|
|
10126 sp->ts_state = STATE_REP_INI;
|
|
|
10127 break;
|
|
|
10128 }
|
|
|
10129 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
10130 {
|
|
|
10131 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
10132 ++depth;
|
|
330
|
10133 #ifdef FEAT_MBYTE
|
|
|
10134 if (has_mbyte)
|
|
|
10135 {
|
|
|
10136 tl = mb_char2len(c3);
|
|
|
10137 mch_memmove(p, p + n + fl, tl);
|
|
|
10138 mb_char2bytes(c2, p + tl);
|
|
|
10139 mb_char2bytes(c, p + fl + tl);
|
|
|
10140 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
10141 }
|
|
|
10142 else
|
|
|
10143 #endif
|
|
|
10144 {
|
|
|
10145 p[0] = p[2];
|
|
|
10146 p[2] = c;
|
|
|
10147 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
10148 }
|
|
323
|
10149 }
|
|
|
10150 else
|
|
|
10151 sp->ts_state = STATE_REP_INI;
|
|
|
10152 break;
|
|
|
10153
|
|
330
|
10154 case STATE_UNSWAP3:
|
|
|
10155 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
10156 p = fword + sp->ts_fidx;
|
|
|
10157 #ifdef FEAT_MBYTE
|
|
|
10158 if (has_mbyte)
|
|
|
10159 {
|
|
|
10160 n = MB_BYTE2LEN(*p);
|
|
|
10161 c2 = mb_ptr2char(p + n);
|
|
|
10162 fl = MB_BYTE2LEN(p[n]);
|
|
|
10163 c = mb_ptr2char(p + n + fl);
|
|
|
10164 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
10165 mch_memmove(p + fl + tl, p, n);
|
|
|
10166 mb_char2bytes(c, p);
|
|
|
10167 mb_char2bytes(c2, p + tl);
|
|
|
10168 }
|
|
|
10169 else
|
|
|
10170 #endif
|
|
|
10171 {
|
|
|
10172 c = *p;
|
|
|
10173 *p = p[2];
|
|
|
10174 p[2] = c;
|
|
|
10175 }
|
|
346
|
10176
|
|
330
|
10177 /* Rotate three characters left: "123" -> "231". We change
|
|
|
10178 * "fword" here, it's changed back afterwards. */
|
|
323
|
10179 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
10180 {
|
|
330
|
10181 sp->ts_state = STATE_UNROT3L;
|
|
323
|
10182 ++depth;
|
|
330
|
10183 p = fword + sp->ts_fidx;
|
|
|
10184 #ifdef FEAT_MBYTE
|
|
|
10185 if (has_mbyte)
|
|
|
10186 {
|
|
474
|
10187 n = mb_cptr2len(p);
|
|
330
|
10188 c = mb_ptr2char(p);
|
|
474
|
10189 fl = mb_cptr2len(p + n);
|
|
|
10190 fl += mb_cptr2len(p + n + fl);
|
|
330
|
10191 mch_memmove(p, p + n, fl);
|
|
|
10192 mb_char2bytes(c, p + fl);
|
|
|
10193 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
10194 }
|
|
|
10195 else
|
|
|
10196 #endif
|
|
|
10197 {
|
|
|
10198 c = *p;
|
|
|
10199 *p = p[1];
|
|
|
10200 p[1] = p[2];
|
|
|
10201 p[2] = c;
|
|
|
10202 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
10203 }
|
|
323
|
10204 }
|
|
|
10205 else
|
|
|
10206 sp->ts_state = STATE_REP_INI;
|
|
|
10207 break;
|
|
|
10208
|
|
330
|
10209 case STATE_UNROT3L:
|
|
346
|
10210 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
10211 p = fword + sp->ts_fidx;
|
|
|
10212 #ifdef FEAT_MBYTE
|
|
|
10213 if (has_mbyte)
|
|
|
10214 {
|
|
|
10215 n = MB_BYTE2LEN(*p);
|
|
|
10216 n += MB_BYTE2LEN(p[n]);
|
|
|
10217 c = mb_ptr2char(p + n);
|
|
|
10218 tl = MB_BYTE2LEN(p[n]);
|
|
|
10219 mch_memmove(p + tl, p, n);
|
|
|
10220 mb_char2bytes(c, p);
|
|
|
10221 }
|
|
|
10222 else
|
|
|
10223 #endif
|
|
|
10224 {
|
|
|
10225 c = p[2];
|
|
|
10226 p[2] = p[1];
|
|
|
10227 p[1] = *p;
|
|
|
10228 *p = c;
|
|
|
10229 }
|
|
346
|
10230
|
|
323
|
10231 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
10232 * "fword" here, it's changed back afterwards. */
|
|
323
|
10233 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
10234 {
|
|
330
|
10235 sp->ts_state = STATE_UNROT3R;
|
|
323
|
10236 ++depth;
|
|
330
|
10237 p = fword + sp->ts_fidx;
|
|
|
10238 #ifdef FEAT_MBYTE
|
|
|
10239 if (has_mbyte)
|
|
|
10240 {
|
|
474
|
10241 n = mb_cptr2len(p);
|
|
|
10242 n += mb_cptr2len(p + n);
|
|
330
|
10243 c = mb_ptr2char(p + n);
|
|
474
|
10244 tl = mb_cptr2len(p + n);
|
|
330
|
10245 mch_memmove(p + tl, p, n);
|
|
|
10246 mb_char2bytes(c, p);
|
|
|
10247 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
10248 }
|
|
|
10249 else
|
|
|
10250 #endif
|
|
|
10251 {
|
|
|
10252 c = p[2];
|
|
|
10253 p[2] = p[1];
|
|
|
10254 p[1] = *p;
|
|
|
10255 *p = c;
|
|
|
10256 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
10257 }
|
|
323
|
10258 }
|
|
|
10259 else
|
|
|
10260 sp->ts_state = STATE_REP_INI;
|
|
|
10261 break;
|
|
|
10262
|
|
330
|
10263 case STATE_UNROT3R:
|
|
346
|
10264 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
10265 p = fword + sp->ts_fidx;
|
|
|
10266 #ifdef FEAT_MBYTE
|
|
|
10267 if (has_mbyte)
|
|
|
10268 {
|
|
|
10269 c = mb_ptr2char(p);
|
|
|
10270 tl = MB_BYTE2LEN(*p);
|
|
|
10271 n = MB_BYTE2LEN(p[tl]);
|
|
|
10272 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
10273 mch_memmove(p, p + tl, n);
|
|
|
10274 mb_char2bytes(c, p + n);
|
|
|
10275 }
|
|
|
10276 else
|
|
|
10277 #endif
|
|
|
10278 {
|
|
|
10279 c = *p;
|
|
|
10280 *p = p[1];
|
|
|
10281 p[1] = p[2];
|
|
|
10282 p[2] = c;
|
|
|
10283 }
|
|
323
|
10284 /*FALLTHROUGH*/
|
|
|
10285
|
|
|
10286 case STATE_REP_INI:
|
|
|
10287 /* Check if matching with REP items from the .aff file would
|
|
503
|
10288 * work. Quickly skip if:
|
|
|
10289 * - there are no REP items
|
|
|
10290 * - the score is going to be too high anyway
|
|
|
10291 * - already applied a REP item or swapped here */
|
|
|
10292 if (lp->lp_replang == NULL
|
|
|
10293 || sp->ts_score + SCORE_REP >= su->su_maxscore
|
|
|
10294 || sp->ts_fidx < sp->ts_fidxtry)
|
|
323
|
10295 {
|
|
|
10296 sp->ts_state = STATE_FINAL;
|
|
|
10297 break;
|
|
|
10298 }
|
|
503
|
10299 gap = &lp->lp_replang->sl_rep;
|
|
323
|
10300
|
|
|
10301 /* Use the first byte to quickly find the first entry that
|
|
330
|
10302 * may match. If the index is -1 there is none. */
|
|
503
|
10303 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]];
|
|
323
|
10304 if (sp->ts_curi < 0)
|
|
|
10305 {
|
|
|
10306 sp->ts_state = STATE_FINAL;
|
|
|
10307 break;
|
|
|
10308 }
|
|
|
10309
|
|
|
10310 sp->ts_state = STATE_REP;
|
|
|
10311 /*FALLTHROUGH*/
|
|
|
10312
|
|
|
10313 case STATE_REP:
|
|
|
10314 /* Try matching with REP items from the .aff file. For each
|
|
330
|
10315 * match replace the characters and check if the resulting
|
|
|
10316 * word is valid. */
|
|
323
|
10317 p = fword + sp->ts_fidx;
|
|
|
10318
|
|
503
|
10319 gap = &lp->lp_replang->sl_rep;
|
|
323
|
10320 while (sp->ts_curi < gap->ga_len)
|
|
|
10321 {
|
|
|
10322 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
10323 if (*ftp->ft_from != *p)
|
|
|
10324 {
|
|
|
10325 /* past possible matching entries */
|
|
|
10326 sp->ts_curi = gap->ga_len;
|
|
|
10327 break;
|
|
|
10328 }
|
|
|
10329 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
10330 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
10331 {
|
|
|
10332 /* Need to undo this afterwards. */
|
|
|
10333 sp->ts_state = STATE_REP_UNDO;
|
|
|
10334
|
|
|
10335 /* Change the "from" to the "to" string. */
|
|
|
10336 ++depth;
|
|
|
10337 fl = STRLEN(ftp->ft_from);
|
|
|
10338 tl = STRLEN(ftp->ft_to);
|
|
|
10339 if (fl != tl)
|
|
346
|
10340 {
|
|
323
|
10341 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
10342 repextra += tl - fl;
|
|
|
10343 }
|
|
323
|
10344 mch_memmove(p, ftp->ft_to, tl);
|
|
|
10345 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
10346 #ifdef FEAT_MBYTE
|
|
|
10347 stack[depth].ts_tcharlen = 0;
|
|
|
10348 #endif
|
|
323
|
10349 break;
|
|
|
10350 }
|
|
|
10351 }
|
|
|
10352
|
|
503
|
10353 if (sp->ts_curi >= gap->ga_len && sp->ts_state == STATE_REP)
|
|
323
|
10354 /* No (more) matches. */
|
|
|
10355 sp->ts_state = STATE_FINAL;
|
|
|
10356
|
|
|
10357 break;
|
|
|
10358
|
|
|
10359 case STATE_REP_UNDO:
|
|
|
10360 /* Undo a REP replacement and continue with the next one. */
|
|
503
|
10361 ftp = (fromto_T *)lp->lp_replang->sl_rep.ga_data
|
|
|
10362 + sp->ts_curi - 1;
|
|
323
|
10363 fl = STRLEN(ftp->ft_from);
|
|
|
10364 tl = STRLEN(ftp->ft_to);
|
|
|
10365 p = fword + sp->ts_fidx;
|
|
|
10366 if (fl != tl)
|
|
346
|
10367 {
|
|
323
|
10368 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
10369 repextra -= tl - fl;
|
|
|
10370 }
|
|
323
|
10371 mch_memmove(p, ftp->ft_from, fl);
|
|
|
10372 sp->ts_state = STATE_REP;
|
|
|
10373 break;
|
|
|
10374
|
|
|
10375 default:
|
|
|
10376 /* Did all possible states at this level, go up one level. */
|
|
|
10377 --depth;
|
|
344
|
10378
|
|
482
|
10379 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
10380 {
|
|
|
10381 /* Continue in or go back to the prefix tree. */
|
|
|
10382 byts = pbyts;
|
|
|
10383 idxs = pidxs;
|
|
|
10384 }
|
|
|
10385
|
|
344
|
10386 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
10387 line_breakcheck();
|
|
323
|
10388 }
|
|
|
10389 }
|
|
|
10390 }
|
|
|
10391 }
|
|
|
10392
|
|
|
10393 /*
|
|
|
10394 * Try going one level deeper in the tree.
|
|
|
10395 */
|
|
|
10396 static int
|
|
|
10397 try_deeper(su, stack, depth, score_add)
|
|
|
10398 suginfo_T *su;
|
|
|
10399 trystate_T *stack;
|
|
|
10400 int depth;
|
|
|
10401 int score_add;
|
|
|
10402 {
|
|
|
10403 int newscore;
|
|
|
10404
|
|
|
10405 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
10406 newscore = stack[depth].ts_score + score_add;
|
|
|
10407 if (newscore >= su->su_maxscore)
|
|
|
10408 return FALSE;
|
|
|
10409
|
|
330
|
10410 stack[depth + 1] = stack[depth];
|
|
323
|
10411 stack[depth + 1].ts_state = STATE_START;
|
|
|
10412 stack[depth + 1].ts_score = newscore;
|
|
|
10413 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
485
|
10414 stack[depth + 1].ts_flags = 0;
|
|
323
|
10415 return TRUE;
|
|
|
10416 }
|
|
|
10417
|
|
455
|
10418 #ifdef FEAT_MBYTE
|
|
|
10419 /*
|
|
|
10420 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
10421 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
10422 */
|
|
|
10423 static int
|
|
|
10424 nofold_len(fword, flen, word)
|
|
|
10425 char_u *fword;
|
|
|
10426 int flen;
|
|
|
10427 char_u *word;
|
|
|
10428 {
|
|
|
10429 char_u *p;
|
|
|
10430 int i = 0;
|
|
|
10431
|
|
|
10432 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
10433 ++i;
|
|
|
10434 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
10435 --i;
|
|
|
10436 return (int)(p - word);
|
|
|
10437 }
|
|
|
10438 #endif
|
|
|
10439
|
|
323
|
10440 /*
|
|
|
10441 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
10442 * words and put it in "kword".
|
|
|
10443 * Theoretically there could be several keep-case words that result in the
|
|
|
10444 * same case-folded word, but we only find one...
|
|
|
10445 */
|
|
|
10446 static void
|
|
|
10447 find_keepcap_word(slang, fword, kword)
|
|
|
10448 slang_T *slang;
|
|
|
10449 char_u *fword;
|
|
|
10450 char_u *kword;
|
|
|
10451 {
|
|
|
10452 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
10453 int depth;
|
|
324
|
10454 idx_T tryidx;
|
|
323
|
10455
|
|
|
10456 /* The following arrays are used at each depth in the tree. */
|
|
324
|
10457 idx_T arridx[MAXWLEN];
|
|
323
|
10458 int round[MAXWLEN];
|
|
|
10459 int fwordidx[MAXWLEN];
|
|
|
10460 int uwordidx[MAXWLEN];
|
|
|
10461 int kwordlen[MAXWLEN];
|
|
|
10462
|
|
|
10463 int flen, ulen;
|
|
|
10464 int l;
|
|
|
10465 int len;
|
|
|
10466 int c;
|
|
324
|
10467 idx_T lo, hi, m;
|
|
323
|
10468 char_u *p;
|
|
|
10469 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
10470 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
10471
|
|
|
10472 if (byts == NULL)
|
|
|
10473 {
|
|
|
10474 /* array is empty: "cannot happen" */
|
|
|
10475 *kword = NUL;
|
|
|
10476 return;
|
|
|
10477 }
|
|
|
10478
|
|
|
10479 /* Make an all-cap version of "fword". */
|
|
|
10480 allcap_copy(fword, uword);
|
|
|
10481
|
|
|
10482 /*
|
|
|
10483 * Each character needs to be tried both case-folded and upper-case.
|
|
|
10484 * All this gets very complicated if we keep in mind that changing case
|
|
|
10485 * may change the byte length of a multi-byte character...
|
|
|
10486 */
|
|
|
10487 depth = 0;
|
|
|
10488 arridx[0] = 0;
|
|
|
10489 round[0] = 0;
|
|
|
10490 fwordidx[0] = 0;
|
|
|
10491 uwordidx[0] = 0;
|
|
|
10492 kwordlen[0] = 0;
|
|
|
10493 while (depth >= 0)
|
|
|
10494 {
|
|
|
10495 if (fword[fwordidx[depth]] == NUL)
|
|
|
10496 {
|
|
|
10497 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
10498 * here we have found a match. */
|
|
|
10499 if (byts[arridx[depth] + 1] == 0)
|
|
|
10500 {
|
|
|
10501 kword[kwordlen[depth]] = NUL;
|
|
|
10502 return;
|
|
|
10503 }
|
|
|
10504
|
|
|
10505 /* kword is getting too long, continue one level up */
|
|
|
10506 --depth;
|
|
|
10507 }
|
|
|
10508 else if (++round[depth] > 2)
|
|
|
10509 {
|
|
|
10510 /* tried both fold-case and upper-case character, continue one
|
|
|
10511 * level up */
|
|
|
10512 --depth;
|
|
|
10513 }
|
|
|
10514 else
|
|
|
10515 {
|
|
|
10516 /*
|
|
|
10517 * round[depth] == 1: Try using the folded-case character.
|
|
|
10518 * round[depth] == 2: Try using the upper-case character.
|
|
|
10519 */
|
|
|
10520 #ifdef FEAT_MBYTE
|
|
|
10521 if (has_mbyte)
|
|
|
10522 {
|
|
474
|
10523 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
10524 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
10525 }
|
|
|
10526 else
|
|
|
10527 #endif
|
|
|
10528 ulen = flen = 1;
|
|
|
10529 if (round[depth] == 1)
|
|
|
10530 {
|
|
|
10531 p = fword + fwordidx[depth];
|
|
|
10532 l = flen;
|
|
|
10533 }
|
|
|
10534 else
|
|
|
10535 {
|
|
|
10536 p = uword + uwordidx[depth];
|
|
|
10537 l = ulen;
|
|
|
10538 }
|
|
|
10539
|
|
|
10540 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
10541 {
|
|
|
10542 /* Perform a binary search in the list of accepted bytes. */
|
|
|
10543 len = byts[tryidx++];
|
|
|
10544 c = *p++;
|
|
|
10545 lo = tryidx;
|
|
|
10546 hi = tryidx + len - 1;
|
|
|
10547 while (lo < hi)
|
|
|
10548 {
|
|
|
10549 m = (lo + hi) / 2;
|
|
|
10550 if (byts[m] > c)
|
|
|
10551 hi = m - 1;
|
|
|
10552 else if (byts[m] < c)
|
|
|
10553 lo = m + 1;
|
|
|
10554 else
|
|
|
10555 {
|
|
|
10556 lo = hi = m;
|
|
|
10557 break;
|
|
|
10558 }
|
|
|
10559 }
|
|
|
10560
|
|
|
10561 /* Stop if there is no matching byte. */
|
|
|
10562 if (hi < lo || byts[lo] != c)
|
|
|
10563 break;
|
|
|
10564
|
|
|
10565 /* Continue at the child (if there is one). */
|
|
|
10566 tryidx = idxs[lo];
|
|
|
10567 }
|
|
|
10568
|
|
|
10569 if (l == 0)
|
|
|
10570 {
|
|
|
10571 /*
|
|
|
10572 * Found the matching char. Copy it to "kword" and go a
|
|
|
10573 * level deeper.
|
|
|
10574 */
|
|
|
10575 if (round[depth] == 1)
|
|
|
10576 {
|
|
|
10577 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
10578 flen);
|
|
|
10579 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
10580 }
|
|
|
10581 else
|
|
|
10582 {
|
|
|
10583 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
10584 ulen);
|
|
|
10585 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
10586 }
|
|
|
10587 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
10588 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
10589
|
|
|
10590 ++depth;
|
|
|
10591 arridx[depth] = tryidx;
|
|
|
10592 round[depth] = 0;
|
|
|
10593 }
|
|
|
10594 }
|
|
|
10595 }
|
|
|
10596
|
|
|
10597 /* Didn't find it: "cannot happen". */
|
|
|
10598 *kword = NUL;
|
|
|
10599 }
|
|
|
10600
|
|
|
10601 /*
|
|
344
|
10602 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
10603 * su->su_sga.
|
|
|
10604 */
|
|
|
10605 static void
|
|
|
10606 score_comp_sal(su)
|
|
|
10607 suginfo_T *su;
|
|
|
10608 {
|
|
|
10609 langp_T *lp;
|
|
|
10610 char_u badsound[MAXWLEN];
|
|
|
10611 int i;
|
|
|
10612 suggest_T *stp;
|
|
|
10613 suggest_T *sstp;
|
|
|
10614 int score;
|
|
500
|
10615 int lpi;
|
|
344
|
10616
|
|
|
10617 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
10618 return;
|
|
|
10619
|
|
|
10620 /* Use the sound-folding of the first language that supports it. */
|
|
503
|
10621 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
10622 {
|
|
|
10623 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
344
|
10624 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10625 {
|
|
|
10626 /* soundfold the bad word */
|
|
375
|
10627 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
10628
|
|
|
10629 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10630 {
|
|
|
10631 stp = &SUG(su->su_ga, i);
|
|
|
10632
|
|
351
|
10633 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
10634 * sound-a-like score. */
|
|
|
10635 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
10636 if (score < SCORE_MAXMAX)
|
|
|
10637 {
|
|
|
10638 /* Add the suggestion. */
|
|
|
10639 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
10640 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
10641 if (sstp->st_word != NULL)
|
|
|
10642 {
|
|
|
10643 sstp->st_score = score;
|
|
|
10644 sstp->st_altscore = 0;
|
|
|
10645 sstp->st_orglen = stp->st_orglen;
|
|
|
10646 ++su->su_sga.ga_len;
|
|
|
10647 }
|
|
|
10648 }
|
|
|
10649 }
|
|
|
10650 break;
|
|
|
10651 }
|
|
500
|
10652 }
|
|
344
|
10653 }
|
|
|
10654
|
|
|
10655 /*
|
|
|
10656 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
10657 * They are intwined.
|
|
|
10658 */
|
|
|
10659 static void
|
|
|
10660 score_combine(su)
|
|
|
10661 suginfo_T *su;
|
|
|
10662 {
|
|
|
10663 int i;
|
|
|
10664 int j;
|
|
|
10665 garray_T ga;
|
|
|
10666 garray_T *gap;
|
|
|
10667 langp_T *lp;
|
|
|
10668 suggest_T *stp;
|
|
|
10669 char_u *p;
|
|
|
10670 char_u badsound[MAXWLEN];
|
|
|
10671 int round;
|
|
500
|
10672 int lpi;
|
|
344
|
10673
|
|
|
10674 /* Add the alternate score to su_ga. */
|
|
503
|
10675 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
10676 {
|
|
|
10677 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
344
|
10678 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10679 {
|
|
|
10680 /* soundfold the bad word */
|
|
375
|
10681 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
10682
|
|
|
10683 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10684 {
|
|
|
10685 stp = &SUG(su->su_ga, i);
|
|
351
|
10686 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
10687 badsound);
|
|
344
|
10688 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
10689 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
10690 else
|
|
|
10691 stp->st_score = (stp->st_score * 3
|
|
|
10692 + stp->st_altscore) / 4;
|
|
|
10693 stp->st_salscore = FALSE;
|
|
|
10694 }
|
|
|
10695 break;
|
|
|
10696 }
|
|
|
10697 }
|
|
|
10698
|
|
|
10699 /* Add the alternate score to su_sga. */
|
|
|
10700 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
10701 {
|
|
|
10702 stp = &SUG(su->su_sga, i);
|
|
|
10703 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
10704 if (stp->st_score == SCORE_MAXMAX)
|
|
|
10705 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
10706 else
|
|
|
10707 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
10708 stp->st_salscore = TRUE;
|
|
|
10709 }
|
|
|
10710
|
|
|
10711 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
10712 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
10713 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
10714
|
|
|
10715 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
10716 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
10717 return;
|
|
|
10718
|
|
|
10719 stp = &SUG(ga, 0);
|
|
|
10720 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
10721 {
|
|
|
10722 /* round 1: get a suggestion from su_ga
|
|
|
10723 * round 2: get a suggestion from su_sga */
|
|
|
10724 for (round = 1; round <= 2; ++round)
|
|
|
10725 {
|
|
|
10726 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
10727 if (i < gap->ga_len)
|
|
|
10728 {
|
|
|
10729 /* Don't add a word if it's already there. */
|
|
|
10730 p = SUG(*gap, i).st_word;
|
|
|
10731 for (j = 0; j < ga.ga_len; ++j)
|
|
|
10732 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
10733 break;
|
|
|
10734 if (j == ga.ga_len)
|
|
|
10735 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
10736 else
|
|
|
10737 vim_free(p);
|
|
|
10738 }
|
|
|
10739 }
|
|
|
10740 }
|
|
|
10741
|
|
|
10742 ga_clear(&su->su_ga);
|
|
|
10743 ga_clear(&su->su_sga);
|
|
|
10744
|
|
|
10745 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
10746 if (ga.ga_len > su->su_maxcount)
|
|
|
10747 {
|
|
|
10748 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
10749 vim_free(stp[i].st_word);
|
|
|
10750 ga.ga_len = su->su_maxcount;
|
|
|
10751 }
|
|
|
10752
|
|
|
10753 su->su_ga = ga;
|
|
|
10754 }
|
|
|
10755
|
|
|
10756 /*
|
|
351
|
10757 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
10758 * badword.
|
|
|
10759 */
|
|
|
10760 static int
|
|
|
10761 stp_sal_score(stp, su, slang, badsound)
|
|
|
10762 suggest_T *stp;
|
|
|
10763 suginfo_T *su;
|
|
|
10764 slang_T *slang;
|
|
|
10765 char_u *badsound; /* sound-folded badword */
|
|
|
10766 {
|
|
|
10767 char_u *p;
|
|
534
|
10768 char_u *pbad;
|
|
|
10769 char_u *pgood;
|
|
351
|
10770 char_u badsound2[MAXWLEN];
|
|
|
10771 char_u fword[MAXWLEN];
|
|
|
10772 char_u goodsound[MAXWLEN];
|
|
534
|
10773 char_u goodword[MAXWLEN];
|
|
|
10774 int lendiff;
|
|
|
10775
|
|
|
10776 lendiff = (int)(su->su_badlen - stp->st_orglen);
|
|
|
10777 if (lendiff >= 0)
|
|
|
10778 pbad = badsound;
|
|
351
|
10779 else
|
|
|
10780 {
|
|
|
10781 /* soundfold the bad word with more characters following */
|
|
|
10782 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
10783
|
|
|
10784 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
10785 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
10786 * removing the space. Don't do it when the good word also contains a
|
|
|
10787 * space. */
|
|
|
10788 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
10789 && *skiptowhite(stp->st_word) == NUL)
|
|
|
10790 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
10791 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
10792
|
|
375
|
10793 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
534
|
10794 pbad = badsound2;
|
|
|
10795 }
|
|
|
10796
|
|
|
10797 if (lendiff > 0)
|
|
|
10798 {
|
|
|
10799 /* Add part of the bad word to the good word, so that we soundfold
|
|
|
10800 * what replaces the bad word. */
|
|
|
10801 STRCPY(goodword, stp->st_word);
|
|
|
10802 STRNCAT(goodword, su->su_badptr + su->su_badlen - lendiff, lendiff);
|
|
|
10803 pgood = goodword;
|
|
|
10804 }
|
|
|
10805 else
|
|
|
10806 pgood = stp->st_word;
|
|
351
|
10807
|
|
375
|
10808 /* Sound-fold the word and compute the score for the difference. */
|
|
534
|
10809 spell_soundfold(slang, pgood, FALSE, goodsound);
|
|
|
10810
|
|
|
10811 return soundalike_score(goodsound, pbad);
|
|
351
|
10812 }
|
|
|
10813
|
|
|
10814 /*
|
|
323
|
10815 * Find suggestions by comparing the word in a sound-a-like form.
|
|
503
|
10816 * Note: This doesn't support postponed prefixes.
|
|
323
|
10817 */
|
|
|
10818 static void
|
|
346
|
10819 suggest_try_soundalike(su)
|
|
323
|
10820 suginfo_T *su;
|
|
|
10821 {
|
|
|
10822 char_u salword[MAXWLEN];
|
|
|
10823 char_u tword[MAXWLEN];
|
|
|
10824 char_u tsalword[MAXWLEN];
|
|
324
|
10825 idx_T arridx[MAXWLEN];
|
|
323
|
10826 int curi[MAXWLEN];
|
|
|
10827 langp_T *lp;
|
|
|
10828 char_u *byts;
|
|
324
|
10829 idx_T *idxs;
|
|
323
|
10830 int depth;
|
|
|
10831 int c;
|
|
324
|
10832 idx_T n;
|
|
323
|
10833 int round;
|
|
|
10834 int flags;
|
|
344
|
10835 int sound_score;
|
|
483
|
10836 int local_score;
|
|
500
|
10837 int lpi;
|
|
503
|
10838 slang_T *slang;
|
|
344
|
10839
|
|
|
10840 /* Do this for all languages that support sound folding. */
|
|
503
|
10841 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
10842 {
|
|
|
10843 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
|
10844 slang = lp->lp_slang;
|
|
|
10845 if (slang->sl_sal.ga_len > 0)
|
|
323
|
10846 {
|
|
|
10847 /* soundfold the bad word */
|
|
503
|
10848 spell_soundfold(slang, su->su_fbadword, TRUE, salword);
|
|
323
|
10849
|
|
|
10850 /*
|
|
|
10851 * Go through the whole tree, soundfold each word and compare.
|
|
|
10852 * round 1: use the case-folded tree.
|
|
|
10853 * round 2: use the keep-case tree.
|
|
|
10854 */
|
|
|
10855 for (round = 1; round <= 2; ++round)
|
|
|
10856 {
|
|
|
10857 if (round == 1)
|
|
|
10858 {
|
|
503
|
10859 byts = slang->sl_fbyts;
|
|
|
10860 idxs = slang->sl_fidxs;
|
|
323
|
10861 }
|
|
|
10862 else
|
|
|
10863 {
|
|
503
|
10864 byts = slang->sl_kbyts;
|
|
|
10865 idxs = slang->sl_kidxs;
|
|
389
|
10866 if (byts == NULL) /* no keep-case words */
|
|
|
10867 continue;
|
|
323
|
10868 }
|
|
|
10869
|
|
|
10870 depth = 0;
|
|
|
10871 arridx[0] = 0;
|
|
|
10872 curi[0] = 1;
|
|
|
10873 while (depth >= 0 && !got_int)
|
|
|
10874 {
|
|
|
10875 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
10876 {
|
|
323
|
10877 /* Done all bytes at this node, go up one level. */
|
|
|
10878 --depth;
|
|
351
|
10879 line_breakcheck();
|
|
|
10880 }
|
|
323
|
10881 else
|
|
|
10882 {
|
|
|
10883 /* Do one more byte at this node. */
|
|
|
10884 n = arridx[depth] + curi[depth];
|
|
|
10885 ++curi[depth];
|
|
|
10886 c = byts[n];
|
|
|
10887 if (c == 0)
|
|
|
10888 {
|
|
|
10889 /* End of word, deal with the word. */
|
|
324
|
10890 flags = (int)idxs[n];
|
|
323
|
10891 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
10892 {
|
|
|
10893 tword[depth] = NUL;
|
|
375
|
10894 /* Sound-fold. Only in keep-case tree need to
|
|
|
10895 * case-fold the word. */
|
|
503
|
10896 spell_soundfold(slang, tword,
|
|
375
|
10897 round == 1, tsalword);
|
|
323
|
10898
|
|
344
|
10899 /* Compute the edit distance between the
|
|
|
10900 * sound-a-like words. */
|
|
|
10901 sound_score = soundalike_score(salword,
|
|
|
10902 tsalword);
|
|
483
|
10903
|
|
|
10904 /* Add a penalty for words in another region. */
|
|
|
10905 if ((flags & WF_REGION) && (((unsigned)flags
|
|
|
10906 >> 16) & lp->lp_region) == 0)
|
|
|
10907 local_score = SCORE_REGION;
|
|
|
10908 else
|
|
|
10909 local_score = 0;
|
|
|
10910 sound_score += local_score;
|
|
|
10911
|
|
324
|
10912 if (sound_score < SCORE_MAXMAX)
|
|
323
|
10913 {
|
|
324
|
10914 char_u cword[MAXWLEN];
|
|
|
10915 char_u *p;
|
|
344
|
10916 int score;
|
|
324
|
10917
|
|
388
|
10918 flags |= su->su_badflags;
|
|
351
|
10919 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
10920 {
|
|
324
|
10921 /* Need to fix case according to
|
|
|
10922 * "flags". */
|
|
323
|
10923 make_case_word(tword, cword, flags);
|
|
324
|
10924 p = cword;
|
|
323
|
10925 }
|
|
|
10926 else
|
|
324
|
10927 p = tword;
|
|
|
10928
|
|
344
|
10929 if (sps_flags & SPS_DOUBLE)
|
|
|
10930 add_suggestion(su, &su->su_sga, p,
|
|
346
|
10931 su->su_badlen,
|
|
503
|
10932 sound_score, 0, FALSE,
|
|
|
10933 lp->lp_sallang);
|
|
344
|
10934 else
|
|
|
10935 {
|
|
|
10936 /* Compute the score. */
|
|
|
10937 score = spell_edit_score(
|
|
483
|
10938 su->su_badword, p)
|
|
|
10939 + local_score;
|
|
344
|
10940 if (sps_flags & SPS_BEST)
|
|
|
10941 /* give a bonus for the good word
|
|
|
10942 * sounding the same as the bad
|
|
|
10943 * word */
|
|
|
10944 add_suggestion(su, &su->su_ga, p,
|
|
346
|
10945 su->su_badlen,
|
|
344
|
10946 RESCORE(score, sound_score),
|
|
503
|
10947 sound_score, TRUE,
|
|
|
10948 lp->lp_sallang);
|
|
344
|
10949 else
|
|
|
10950 add_suggestion(su, &su->su_ga, p,
|
|
346
|
10951 su->su_badlen,
|
|
503
|
10952 score + sound_score,
|
|
|
10953 0, FALSE,
|
|
|
10954 lp->lp_sallang);
|
|
344
|
10955 }
|
|
323
|
10956 }
|
|
|
10957 }
|
|
|
10958
|
|
|
10959 /* Skip over other NUL bytes. */
|
|
|
10960 while (byts[n + 1] == 0)
|
|
|
10961 {
|
|
|
10962 ++n;
|
|
|
10963 ++curi[depth];
|
|
|
10964 }
|
|
|
10965 }
|
|
|
10966 else
|
|
|
10967 {
|
|
|
10968 /* Normal char, go one level deeper. */
|
|
|
10969 tword[depth++] = c;
|
|
|
10970 arridx[depth] = idxs[n];
|
|
|
10971 curi[depth] = 1;
|
|
|
10972 }
|
|
|
10973 }
|
|
|
10974 }
|
|
|
10975 }
|
|
|
10976 }
|
|
|
10977 }
|
|
|
10978 }
|
|
|
10979
|
|
|
10980 /*
|
|
324
|
10981 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
10982 */
|
|
|
10983 static void
|
|
|
10984 make_case_word(fword, cword, flags)
|
|
|
10985 char_u *fword;
|
|
|
10986 char_u *cword;
|
|
|
10987 int flags;
|
|
|
10988 {
|
|
|
10989 if (flags & WF_ALLCAP)
|
|
|
10990 /* Make it all upper-case */
|
|
|
10991 allcap_copy(fword, cword);
|
|
|
10992 else if (flags & WF_ONECAP)
|
|
|
10993 /* Make the first letter upper-case */
|
|
324
|
10994 onecap_copy(fword, cword, TRUE);
|
|
323
|
10995 else
|
|
|
10996 /* Use goodword as-is. */
|
|
|
10997 STRCPY(cword, fword);
|
|
|
10998 }
|
|
|
10999
|
|
330
|
11000 /*
|
|
|
11001 * Use map string "map" for languages "lp".
|
|
|
11002 */
|
|
|
11003 static void
|
|
|
11004 set_map_str(lp, map)
|
|
|
11005 slang_T *lp;
|
|
|
11006 char_u *map;
|
|
|
11007 {
|
|
|
11008 char_u *p;
|
|
|
11009 int headc = 0;
|
|
|
11010 int c;
|
|
|
11011 int i;
|
|
|
11012
|
|
|
11013 if (*map == NUL)
|
|
|
11014 {
|
|
|
11015 lp->sl_has_map = FALSE;
|
|
|
11016 return;
|
|
|
11017 }
|
|
|
11018 lp->sl_has_map = TRUE;
|
|
|
11019
|
|
|
11020 /* Init the array and hash table empty. */
|
|
|
11021 for (i = 0; i < 256; ++i)
|
|
|
11022 lp->sl_map_array[i] = 0;
|
|
|
11023 #ifdef FEAT_MBYTE
|
|
|
11024 hash_init(&lp->sl_map_hash);
|
|
|
11025 #endif
|
|
|
11026
|
|
|
11027 /*
|
|
|
11028 * The similar characters are stored separated with slashes:
|
|
|
11029 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
11030 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
11031 */
|
|
|
11032 for (p = map; *p != NUL; )
|
|
|
11033 {
|
|
|
11034 #ifdef FEAT_MBYTE
|
|
474
|
11035 c = mb_cptr2char_adv(&p);
|
|
330
|
11036 #else
|
|
|
11037 c = *p++;
|
|
|
11038 #endif
|
|
|
11039 if (c == '/')
|
|
|
11040 headc = 0;
|
|
|
11041 else
|
|
|
11042 {
|
|
|
11043 if (headc == 0)
|
|
|
11044 headc = c;
|
|
|
11045
|
|
|
11046 #ifdef FEAT_MBYTE
|
|
|
11047 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
11048 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
11049 * a NUL. */
|
|
|
11050 if (c >= 256)
|
|
|
11051 {
|
|
|
11052 int cl = mb_char2len(c);
|
|
|
11053 int headcl = mb_char2len(headc);
|
|
|
11054 char_u *b;
|
|
|
11055 hash_T hash;
|
|
|
11056 hashitem_T *hi;
|
|
|
11057
|
|
|
11058 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
11059 if (b == NULL)
|
|
|
11060 return;
|
|
|
11061 mb_char2bytes(c, b);
|
|
|
11062 b[cl] = NUL;
|
|
|
11063 mb_char2bytes(headc, b + cl + 1);
|
|
|
11064 b[cl + 1 + headcl] = NUL;
|
|
|
11065 hash = hash_hash(b);
|
|
|
11066 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
11067 if (HASHITEM_EMPTY(hi))
|
|
|
11068 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
11069 else
|
|
|
11070 {
|
|
|
11071 /* This should have been checked when generating the .spl
|
|
|
11072 * file. */
|
|
|
11073 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
11074 vim_free(b);
|
|
|
11075 }
|
|
|
11076 }
|
|
|
11077 else
|
|
|
11078 #endif
|
|
|
11079 lp->sl_map_array[c] = headc;
|
|
|
11080 }
|
|
|
11081 }
|
|
|
11082 }
|
|
|
11083
|
|
323
|
11084 /*
|
|
|
11085 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
11086 * lines in the .aff file.
|
|
|
11087 */
|
|
|
11088 static int
|
|
|
11089 similar_chars(slang, c1, c2)
|
|
|
11090 slang_T *slang;
|
|
|
11091 int c1;
|
|
|
11092 int c2;
|
|
|
11093 {
|
|
330
|
11094 int m1, m2;
|
|
|
11095 #ifdef FEAT_MBYTE
|
|
|
11096 char_u buf[MB_MAXBYTES];
|
|
|
11097 hashitem_T *hi;
|
|
|
11098
|
|
|
11099 if (c1 >= 256)
|
|
|
11100 {
|
|
|
11101 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
11102 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
11103 if (HASHITEM_EMPTY(hi))
|
|
|
11104 m1 = 0;
|
|
|
11105 else
|
|
|
11106 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
11107 }
|
|
|
11108 else
|
|
|
11109 #endif
|
|
|
11110 m1 = slang->sl_map_array[c1];
|
|
|
11111 if (m1 == 0)
|
|
323
|
11112 return FALSE;
|
|
330
|
11113
|
|
|
11114
|
|
|
11115 #ifdef FEAT_MBYTE
|
|
|
11116 if (c2 >= 256)
|
|
|
11117 {
|
|
|
11118 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
11119 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
11120 if (HASHITEM_EMPTY(hi))
|
|
|
11121 m2 = 0;
|
|
|
11122 else
|
|
|
11123 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
11124 }
|
|
|
11125 else
|
|
|
11126 #endif
|
|
|
11127 m2 = slang->sl_map_array[c2];
|
|
|
11128
|
|
|
11129 return m1 == m2;
|
|
323
|
11130 }
|
|
|
11131
|
|
|
11132 /*
|
|
|
11133 * Add a suggestion to the list of suggestions.
|
|
|
11134 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
11135 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
11136 * with spell_edit_score().
|
|
|
11137 */
|
|
|
11138 static void
|
|
534
|
11139 add_suggestion(su, gap, goodword, badlenarg, score, altscore, had_bonus, slang)
|
|
323
|
11140 suginfo_T *su;
|
|
344
|
11141 garray_T *gap;
|
|
323
|
11142 char_u *goodword;
|
|
534
|
11143 int badlenarg; /* len of bad word replaced with "goodword" */
|
|
324
|
11144 int score;
|
|
351
|
11145 int altscore;
|
|
344
|
11146 int had_bonus; /* value for st_had_bonus */
|
|
503
|
11147 slang_T *slang; /* language for sound folding */
|
|
323
|
11148 {
|
|
534
|
11149 int goodlen = STRLEN(goodword); /* len of goodword changed */
|
|
|
11150 int badlen = badlenarg; /* len of bad word changed */
|
|
323
|
11151 suggest_T *stp;
|
|
534
|
11152 suggest_T new_sug;
|
|
323
|
11153 int i;
|
|
534
|
11154 hlf_T attr = HLF_COUNT;
|
|
532
|
11155 char_u longword[MAXWLEN + 1];
|
|
534
|
11156 char_u *pgood, *pbad;
|
|
532
|
11157
|
|
|
11158 /* Check that the word really is valid. Esp. for banned words and for
|
|
|
11159 * split words, such as "the the". Need to append what follows to check
|
|
|
11160 * for that. */
|
|
|
11161 STRCPY(longword, goodword);
|
|
|
11162 vim_strncpy(longword + goodlen, su->su_badptr + badlen, MAXWLEN - goodlen);
|
|
|
11163 (void)spell_check(curwin, longword, &attr, NULL);
|
|
534
|
11164 if (attr != HLF_COUNT)
|
|
323
|
11165 return;
|
|
|
11166
|
|
534
|
11167 /* Minimize "badlen" for consistency. Avoids that changing "the the" to
|
|
|
11168 * "thee the" is added next to changing the first "the" the "thee". */
|
|
|
11169 pgood = goodword + STRLEN(goodword);
|
|
|
11170 pbad = su->su_badptr + badlen;
|
|
|
11171 while (pgood > goodword && pbad > su->su_badptr)
|
|
|
11172 {
|
|
|
11173 mb_ptr_back(goodword, pgood);
|
|
|
11174 mb_ptr_back(su->su_badptr, pbad);
|
|
|
11175 #ifdef FEAT_MBYTE
|
|
|
11176 if (has_mbyte)
|
|
|
11177 {
|
|
|
11178 if (mb_ptr2char(pgood) != mb_ptr2char(pbad))
|
|
|
11179 break;
|
|
346
|
11180 }
|
|
|
11181 else
|
|
534
|
11182 #endif
|
|
|
11183 if (*pgood != *pbad)
|
|
|
11184 break;
|
|
|
11185 badlen = pbad - su->su_badptr;
|
|
|
11186 goodlen = pgood - goodword;
|
|
|
11187 }
|
|
|
11188 if (badlen == 0 && goodlen == 0)
|
|
|
11189 /* goodword doesn't change anything; may happen for "the the" changing
|
|
|
11190 * the first "the" to itself. */
|
|
|
11191 return;
|
|
346
|
11192
|
|
323
|
11193 if (score <= su->su_maxscore)
|
|
|
11194 {
|
|
366
|
11195 /* Check if the word is already there. Also check the length that is
|
|
|
11196 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
11197 * "thes" -> "these". */
|
|
344
|
11198 stp = &SUG(*gap, 0);
|
|
|
11199 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
535
|
11200 if ((int)STRLEN(stp[i].st_word) == goodlen
|
|
534
|
11201 && STRNCMP(stp[i].st_word, goodword, goodlen) == 0
|
|
366
|
11202 && stp[i].st_orglen == badlen)
|
|
323
|
11203 {
|
|
534
|
11204 /*
|
|
535
|
11205 * Found it. Remember the word with the lowest score.
|
|
534
|
11206 */
|
|
503
|
11207 if (stp[i].st_slang == NULL)
|
|
|
11208 stp[i].st_slang = slang;
|
|
534
|
11209
|
|
|
11210 new_sug.st_score = score;
|
|
|
11211 new_sug.st_altscore = altscore;
|
|
|
11212 new_sug.st_had_bonus = had_bonus;
|
|
|
11213
|
|
|
11214 if (stp[i].st_had_bonus != had_bonus)
|
|
|
11215 {
|
|
|
11216 /* Only one of the two had the soundalike score computed.
|
|
|
11217 * Need to do that for the other one now, otherwise the
|
|
|
11218 * scores can't be compared. This happens because
|
|
|
11219 * suggest_try_change() doesn't compute the soundalike
|
|
535
|
11220 * word to keep it fast, while some special methods set
|
|
|
11221 * the soundalike score to zero. */
|
|
534
|
11222 if (had_bonus)
|
|
|
11223 rescore_one(su, &stp[i]);
|
|
|
11224 else
|
|
|
11225 {
|
|
|
11226 new_sug.st_word = goodword;
|
|
|
11227 new_sug.st_slang = stp[i].st_slang;
|
|
|
11228 new_sug.st_orglen = badlen;
|
|
|
11229 rescore_one(su, &new_sug);
|
|
|
11230 }
|
|
|
11231 }
|
|
|
11232
|
|
|
11233 if (stp[i].st_score > new_sug.st_score)
|
|
|
11234 {
|
|
|
11235 stp[i].st_score = new_sug.st_score;
|
|
|
11236 stp[i].st_altscore = new_sug.st_altscore;
|
|
|
11237 stp[i].st_had_bonus = new_sug.st_had_bonus;
|
|
|
11238 }
|
|
323
|
11239 break;
|
|
|
11240 }
|
|
|
11241
|
|
344
|
11242 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
11243 {
|
|
|
11244 /* Add a suggestion. */
|
|
344
|
11245 stp = &SUG(*gap, gap->ga_len);
|
|
534
|
11246 stp->st_word = vim_strnsave(goodword, goodlen);
|
|
323
|
11247 if (stp->st_word != NULL)
|
|
|
11248 {
|
|
|
11249 stp->st_score = score;
|
|
351
|
11250 stp->st_altscore = altscore;
|
|
324
|
11251 stp->st_had_bonus = had_bonus;
|
|
346
|
11252 stp->st_orglen = badlen;
|
|
503
|
11253 stp->st_slang = slang;
|
|
344
|
11254 ++gap->ga_len;
|
|
323
|
11255
|
|
|
11256 /* If we have too many suggestions now, sort the list and keep
|
|
|
11257 * the best suggestions. */
|
|
344
|
11258 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
11259 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
11260 SUG_CLEAN_COUNT(su));
|
|
323
|
11261 }
|
|
|
11262 }
|
|
|
11263 }
|
|
|
11264 }
|
|
|
11265
|
|
|
11266 /*
|
|
|
11267 * Add a word to be banned.
|
|
|
11268 */
|
|
|
11269 static void
|
|
|
11270 add_banned(su, word)
|
|
|
11271 suginfo_T *su;
|
|
|
11272 char_u *word;
|
|
|
11273 {
|
|
|
11274 char_u *s = vim_strsave(word);
|
|
|
11275 hash_T hash;
|
|
|
11276 hashitem_T *hi;
|
|
|
11277
|
|
|
11278 if (s != NULL)
|
|
|
11279 {
|
|
|
11280 hash = hash_hash(s);
|
|
|
11281 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
11282 if (HASHITEM_EMPTY(hi))
|
|
|
11283 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
11284 else
|
|
|
11285 vim_free(s);
|
|
323
|
11286 }
|
|
|
11287 }
|
|
|
11288
|
|
|
11289 /*
|
|
|
11290 * Return TRUE if a word appears in the list of banned words.
|
|
|
11291 */
|
|
|
11292 static int
|
|
|
11293 was_banned(su, word)
|
|
|
11294 suginfo_T *su;
|
|
|
11295 char_u *word;
|
|
|
11296 {
|
|
324
|
11297 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
11298
|
|
|
11299 return !HASHITEM_EMPTY(hi);
|
|
323
|
11300 }
|
|
|
11301
|
|
|
11302 /*
|
|
|
11303 * Free the banned words in "su".
|
|
|
11304 */
|
|
|
11305 static void
|
|
|
11306 free_banned(su)
|
|
|
11307 suginfo_T *su;
|
|
|
11308 {
|
|
|
11309 int todo;
|
|
|
11310 hashitem_T *hi;
|
|
|
11311
|
|
|
11312 todo = su->su_banned.ht_used;
|
|
|
11313 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
11314 {
|
|
|
11315 if (!HASHITEM_EMPTY(hi))
|
|
|
11316 {
|
|
|
11317 vim_free(hi->hi_key);
|
|
|
11318 --todo;
|
|
|
11319 }
|
|
|
11320 }
|
|
|
11321 hash_clear(&su->su_banned);
|
|
|
11322 }
|
|
|
11323
|
|
324
|
11324 /*
|
|
534
|
11325 * Recompute the score for all suggestions if sound-folding is possible. This
|
|
|
11326 * is slow, thus only done for the final results.
|
|
324
|
11327 */
|
|
|
11328 static void
|
|
|
11329 rescore_suggestions(su)
|
|
|
11330 suginfo_T *su;
|
|
|
11331 {
|
|
|
11332 int i;
|
|
534
|
11333
|
|
|
11334 if (su->su_sallang != NULL)
|
|
503
|
11335 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
534
|
11336 rescore_one(su, &SUG(su->su_ga, i));
|
|
|
11337 }
|
|
|
11338
|
|
|
11339 /*
|
|
|
11340 * Recompute the score for one suggestion if sound-folding is possible.
|
|
|
11341 */
|
|
|
11342 static void
|
|
|
11343 rescore_one(su, stp)
|
|
535
|
11344 suginfo_T *su;
|
|
|
11345 suggest_T *stp;
|
|
534
|
11346 {
|
|
|
11347 slang_T *slang = stp->st_slang;
|
|
|
11348 char_u sal_badword[MAXWLEN];
|
|
535
|
11349 char_u *p;
|
|
534
|
11350
|
|
|
11351 /* Only rescore suggestions that have no sal score yet and do have a
|
|
|
11352 * language. */
|
|
|
11353 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus)
|
|
|
11354 {
|
|
|
11355 if (slang == su->su_sallang)
|
|
535
|
11356 p = su->su_sal_badword;
|
|
534
|
11357 else
|
|
|
11358 {
|
|
|
11359 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword);
|
|
535
|
11360 p = sal_badword;
|
|
|
11361 }
|
|
|
11362
|
|
|
11363 stp->st_altscore = stp_sal_score(stp, su, slang, p);
|
|
534
|
11364 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
11365 stp->st_altscore = SCORE_BIG;
|
|
|
11366 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
|
11367 stp->st_had_bonus = TRUE;
|
|
324
|
11368 }
|
|
|
11369 }
|
|
|
11370
|
|
323
|
11371 static int
|
|
|
11372 #ifdef __BORLANDC__
|
|
|
11373 _RTLENTRYF
|
|
|
11374 #endif
|
|
|
11375 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
11376
|
|
|
11377 /*
|
|
|
11378 * Function given to qsort() to sort the suggestions on st_score.
|
|
529
|
11379 * First on "st_score", then "st_altscore" then alphabetically.
|
|
323
|
11380 */
|
|
|
11381 static int
|
|
|
11382 #ifdef __BORLANDC__
|
|
|
11383 _RTLENTRYF
|
|
|
11384 #endif
|
|
|
11385 sug_compare(s1, s2)
|
|
|
11386 const void *s1;
|
|
|
11387 const void *s2;
|
|
|
11388 {
|
|
|
11389 suggest_T *p1 = (suggest_T *)s1;
|
|
|
11390 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
11391 int n = p1->st_score - p2->st_score;
|
|
|
11392
|
|
|
11393 if (n == 0)
|
|
529
|
11394 {
|
|
|
11395 n = p1->st_altscore - p2->st_altscore;
|
|
|
11396 if (n == 0)
|
|
|
11397 n = STRICMP(p1->st_word, p2->st_word);
|
|
|
11398 }
|
|
344
|
11399 return n;
|
|
323
|
11400 }
|
|
|
11401
|
|
|
11402 /*
|
|
|
11403 * Cleanup the suggestions:
|
|
|
11404 * - Sort on score.
|
|
|
11405 * - Remove words that won't be displayed.
|
|
344
|
11406 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
11407 */
|
|
344
|
11408 static int
|
|
|
11409 cleanup_suggestions(gap, maxscore, keep)
|
|
|
11410 garray_T *gap;
|
|
|
11411 int maxscore;
|
|
324
|
11412 int keep; /* nr of suggestions to keep */
|
|
323
|
11413 {
|
|
344
|
11414 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
11415 int i;
|
|
|
11416
|
|
|
11417 /* Sort the list. */
|
|
344
|
11418 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
11419
|
|
|
11420 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
11421 if (gap->ga_len > keep)
|
|
323
|
11422 {
|
|
344
|
11423 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
11424 vim_free(stp[i].st_word);
|
|
344
|
11425 gap->ga_len = keep;
|
|
|
11426 return stp[keep - 1].st_score;
|
|
323
|
11427 }
|
|
344
|
11428 return maxscore;
|
|
323
|
11429 }
|
|
|
11430
|
|
372
|
11431 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
11432 /*
|
|
|
11433 * Soundfold a string, for soundfold().
|
|
|
11434 * Result is in allocated memory, NULL for an error.
|
|
|
11435 */
|
|
|
11436 char_u *
|
|
|
11437 eval_soundfold(word)
|
|
|
11438 char_u *word;
|
|
|
11439 {
|
|
|
11440 langp_T *lp;
|
|
|
11441 char_u sound[MAXWLEN];
|
|
500
|
11442 int lpi;
|
|
372
|
11443
|
|
|
11444 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
11445 /* Use the sound-folding of the first language that supports it. */
|
|
503
|
11446 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
11447 {
|
|
|
11448 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
372
|
11449 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
11450 {
|
|
|
11451 /* soundfold the word */
|
|
375
|
11452 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
11453 return vim_strsave(sound);
|
|
|
11454 }
|
|
500
|
11455 }
|
|
372
|
11456
|
|
|
11457 /* No language with sound folding, return word as-is. */
|
|
|
11458 return vim_strsave(word);
|
|
|
11459 }
|
|
|
11460 #endif
|
|
|
11461
|
|
323
|
11462 /*
|
|
|
11463 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
485
|
11464 *
|
|
|
11465 * There are many ways to turn a word into a sound-a-like representation. The
|
|
|
11466 * oldest is Soundex (1918!). A nice overview can be found in "Approximate
|
|
|
11467 * swedish name matching - survey and test of different algorithms" by Klas
|
|
|
11468 * Erikson.
|
|
|
11469 *
|
|
|
11470 * We support two methods:
|
|
|
11471 * 1. SOFOFROM/SOFOTO do a simple character mapping.
|
|
|
11472 * 2. SAL items define a more advanced sound-folding (and much slower).
|
|
323
|
11473 */
|
|
|
11474 static void
|
|
375
|
11475 spell_soundfold(slang, inword, folded, res)
|
|
|
11476 slang_T *slang;
|
|
|
11477 char_u *inword;
|
|
|
11478 int folded; /* "inword" is already case-folded */
|
|
|
11479 char_u *res;
|
|
|
11480 {
|
|
|
11481 char_u fword[MAXWLEN];
|
|
|
11482 char_u *word;
|
|
|
11483
|
|
|
11484 if (slang->sl_sofo)
|
|
|
11485 /* SOFOFROM and SOFOTO used */
|
|
|
11486 spell_soundfold_sofo(slang, inword, res);
|
|
|
11487 else
|
|
|
11488 {
|
|
|
11489 /* SAL items used. Requires the word to be case-folded. */
|
|
|
11490 if (folded)
|
|
|
11491 word = inword;
|
|
|
11492 else
|
|
|
11493 {
|
|
|
11494 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
11495 word = fword;
|
|
|
11496 }
|
|
|
11497
|
|
|
11498 #ifdef FEAT_MBYTE
|
|
|
11499 if (has_mbyte)
|
|
|
11500 spell_soundfold_wsal(slang, word, res);
|
|
|
11501 else
|
|
|
11502 #endif
|
|
|
11503 spell_soundfold_sal(slang, word, res);
|
|
|
11504 }
|
|
|
11505 }
|
|
|
11506
|
|
|
11507 /*
|
|
|
11508 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
11509 * SOFOTO lines.
|
|
|
11510 */
|
|
|
11511 static void
|
|
|
11512 spell_soundfold_sofo(slang, inword, res)
|
|
|
11513 slang_T *slang;
|
|
|
11514 char_u *inword;
|
|
|
11515 char_u *res;
|
|
|
11516 {
|
|
|
11517 char_u *s;
|
|
|
11518 int ri = 0;
|
|
|
11519 int c;
|
|
|
11520
|
|
|
11521 #ifdef FEAT_MBYTE
|
|
|
11522 if (has_mbyte)
|
|
|
11523 {
|
|
|
11524 int prevc = 0;
|
|
|
11525 int *ip;
|
|
|
11526
|
|
|
11527 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
11528 * 255, sl_sal the rest. */
|
|
|
11529 for (s = inword; *s != NUL; )
|
|
|
11530 {
|
|
474
|
11531 c = mb_cptr2char_adv(&s);
|
|
375
|
11532 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
11533 c = ' ';
|
|
|
11534 else if (c < 256)
|
|
|
11535 c = slang->sl_sal_first[c];
|
|
|
11536 else
|
|
|
11537 {
|
|
|
11538 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
11539 if (ip == NULL) /* empty list, can't match */
|
|
|
11540 c = NUL;
|
|
|
11541 else
|
|
|
11542 for (;;) /* find "c" in the list */
|
|
|
11543 {
|
|
|
11544 if (*ip == 0) /* not found */
|
|
|
11545 {
|
|
|
11546 c = NUL;
|
|
|
11547 break;
|
|
|
11548 }
|
|
|
11549 if (*ip == c) /* match! */
|
|
|
11550 {
|
|
|
11551 c = ip[1];
|
|
|
11552 break;
|
|
|
11553 }
|
|
|
11554 ip += 2;
|
|
|
11555 }
|
|
|
11556 }
|
|
|
11557
|
|
|
11558 if (c != NUL && c != prevc)
|
|
|
11559 {
|
|
|
11560 ri += mb_char2bytes(c, res + ri);
|
|
|
11561 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
11562 break;
|
|
|
11563 prevc = c;
|
|
|
11564 }
|
|
|
11565 }
|
|
|
11566 }
|
|
|
11567 else
|
|
|
11568 #endif
|
|
|
11569 {
|
|
|
11570 /* The sl_sal_first[] table contains the translation. */
|
|
|
11571 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
11572 {
|
|
|
11573 if (vim_iswhite(c))
|
|
|
11574 c = ' ';
|
|
|
11575 else
|
|
|
11576 c = slang->sl_sal_first[c];
|
|
|
11577 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
11578 res[ri++] = c;
|
|
|
11579 }
|
|
|
11580 }
|
|
|
11581
|
|
|
11582 res[ri] = NUL;
|
|
|
11583 }
|
|
|
11584
|
|
|
11585 static void
|
|
|
11586 spell_soundfold_sal(slang, inword, res)
|
|
323
|
11587 slang_T *slang;
|
|
|
11588 char_u *inword;
|
|
|
11589 char_u *res;
|
|
|
11590 {
|
|
344
|
11591 salitem_T *smp;
|
|
323
|
11592 char_u word[MAXWLEN];
|
|
375
|
11593 char_u *s = inword;
|
|
323
|
11594 char_u *t;
|
|
344
|
11595 char_u *pf;
|
|
323
|
11596 int i, j, z;
|
|
344
|
11597 int reslen;
|
|
323
|
11598 int n, k = 0;
|
|
|
11599 int z0;
|
|
|
11600 int k0;
|
|
|
11601 int n0;
|
|
|
11602 int c;
|
|
|
11603 int pri;
|
|
|
11604 int p0 = -333;
|
|
|
11605 int c0;
|
|
|
11606
|
|
324
|
11607 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
11608 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
11609 if (slang->sl_rem_accents)
|
|
|
11610 {
|
|
|
11611 t = word;
|
|
375
|
11612 while (*s != NUL)
|
|
323
|
11613 {
|
|
324
|
11614 if (vim_iswhite(*s))
|
|
344
|
11615 {
|
|
|
11616 *t++ = ' ';
|
|
|
11617 s = skipwhite(s);
|
|
|
11618 }
|
|
323
|
11619 else
|
|
|
11620 {
|
|
376
|
11621 if (spell_iswordp_nmw(s))
|
|
323
|
11622 *t++ = *s;
|
|
|
11623 ++s;
|
|
|
11624 }
|
|
|
11625 }
|
|
|
11626 *t = NUL;
|
|
|
11627 }
|
|
|
11628 else
|
|
375
|
11629 STRCPY(word, s);
|
|
323
|
11630
|
|
344
|
11631 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
11632
|
|
|
11633 /*
|
|
|
11634 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
11635 * Changed to keep spaces.
|
|
323
|
11636 */
|
|
344
|
11637 i = reslen = z = 0;
|
|
323
|
11638 while ((c = word[i]) != NUL)
|
|
|
11639 {
|
|
344
|
11640 /* Start with the first rule that has the character in the word. */
|
|
323
|
11641 n = slang->sl_sal_first[c];
|
|
|
11642 z0 = 0;
|
|
|
11643
|
|
|
11644 if (n >= 0)
|
|
|
11645 {
|
|
|
11646 /* check all rules for the same letter */
|
|
344
|
11647 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
11648 {
|
|
344
|
11649 /* Quickly skip entries that don't match the word. Most
|
|
|
11650 * entries are less then three chars, optimize for that. */
|
|
|
11651 k = smp[n].sm_leadlen;
|
|
|
11652 if (k > 1)
|
|
323
|
11653 {
|
|
344
|
11654 if (word[i + 1] != s[1])
|
|
|
11655 continue;
|
|
|
11656 if (k > 2)
|
|
|
11657 {
|
|
|
11658 for (j = 2; j < k; ++j)
|
|
|
11659 if (word[i + j] != s[j])
|
|
|
11660 break;
|
|
|
11661 if (j < k)
|
|
|
11662 continue;
|
|
|
11663 }
|
|
323
|
11664 }
|
|
|
11665
|
|
375
|
11666 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
11667 {
|
|
375
|
11668 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
11669 while (*pf != NUL && *pf != word[i + k])
|
|
|
11670 ++pf;
|
|
|
11671 if (*pf == NUL)
|
|
|
11672 continue;
|
|
|
11673 ++k;
|
|
323
|
11674 }
|
|
344
|
11675 s = smp[n].sm_rules;
|
|
|
11676 pri = 5; /* default priority */
|
|
323
|
11677
|
|
|
11678 p0 = *s;
|
|
|
11679 k0 = k;
|
|
|
11680 while (*s == '-' && k > 1)
|
|
|
11681 {
|
|
|
11682 k--;
|
|
|
11683 s++;
|
|
|
11684 }
|
|
|
11685 if (*s == '<')
|
|
|
11686 s++;
|
|
344
|
11687 if (VIM_ISDIGIT(*s))
|
|
323
|
11688 {
|
|
|
11689 /* determine priority */
|
|
|
11690 pri = *s - '0';
|
|
|
11691 s++;
|
|
|
11692 }
|
|
|
11693 if (*s == '^' && *(s + 1) == '^')
|
|
|
11694 s++;
|
|
|
11695
|
|
|
11696 if (*s == NUL
|
|
|
11697 || (*s == '^'
|
|
324
|
11698 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
11699 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
11700 && (*(s + 1) != '$'
|
|
376
|
11701 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
11702 || (*s == '$' && i > 0
|
|
376
|
11703 && spell_iswordp(word + i - 1, curbuf)
|
|
|
11704 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
11705 {
|
|
|
11706 /* search for followup rules, if: */
|
|
|
11707 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
11708 c0 = word[i + k - 1];
|
|
|
11709 n0 = slang->sl_sal_first[c0];
|
|
|
11710
|
|
|
11711 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
11712 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
11713 {
|
|
|
11714 /* test follow-up rule for "word[i + k]" */
|
|
344
|
11715 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
11716 {
|
|
344
|
11717 /* Quickly skip entries that don't match the word.
|
|
|
11718 * */
|
|
|
11719 k0 = smp[n0].sm_leadlen;
|
|
|
11720 if (k0 > 1)
|
|
323
|
11721 {
|
|
344
|
11722 if (word[i + k] != s[1])
|
|
|
11723 continue;
|
|
|
11724 if (k0 > 2)
|
|
|
11725 {
|
|
|
11726 pf = word + i + k + 1;
|
|
|
11727 for (j = 2; j < k0; ++j)
|
|
|
11728 if (*pf++ != s[j])
|
|
|
11729 break;
|
|
|
11730 if (j < k0)
|
|
|
11731 continue;
|
|
|
11732 }
|
|
323
|
11733 }
|
|
344
|
11734 k0 += k - 1;
|
|
|
11735
|
|
375
|
11736 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
11737 {
|
|
344
|
11738 /* Check for match with one of the chars in
|
|
375
|
11739 * "sm_oneof". */
|
|
344
|
11740 while (*pf != NUL && *pf != word[i + k0])
|
|
|
11741 ++pf;
|
|
|
11742 if (*pf == NUL)
|
|
|
11743 continue;
|
|
|
11744 ++k0;
|
|
323
|
11745 }
|
|
344
|
11746
|
|
|
11747 p0 = 5;
|
|
|
11748 s = smp[n0].sm_rules;
|
|
323
|
11749 while (*s == '-')
|
|
|
11750 {
|
|
344
|
11751 /* "k0" gets NOT reduced because
|
|
|
11752 * "if (k0 == k)" */
|
|
323
|
11753 s++;
|
|
|
11754 }
|
|
|
11755 if (*s == '<')
|
|
|
11756 s++;
|
|
344
|
11757 if (VIM_ISDIGIT(*s))
|
|
323
|
11758 {
|
|
|
11759 p0 = *s - '0';
|
|
|
11760 s++;
|
|
|
11761 }
|
|
|
11762
|
|
|
11763 if (*s == NUL
|
|
|
11764 /* *s == '^' cuts */
|
|
|
11765 || (*s == '$'
|
|
376
|
11766 && !spell_iswordp(word + i + k0,
|
|
|
11767 curbuf)))
|
|
323
|
11768 {
|
|
|
11769 if (k0 == k)
|
|
|
11770 /* this is just a piece of the string */
|
|
|
11771 continue;
|
|
|
11772
|
|
|
11773 if (p0 < pri)
|
|
|
11774 /* priority too low */
|
|
|
11775 continue;
|
|
|
11776 /* rule fits; stop search */
|
|
|
11777 break;
|
|
|
11778 }
|
|
|
11779 }
|
|
|
11780
|
|
344
|
11781 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
11782 continue;
|
|
|
11783 }
|
|
|
11784
|
|
|
11785 /* replace string */
|
|
344
|
11786 s = smp[n].sm_to;
|
|
389
|
11787 if (s == NULL)
|
|
|
11788 s = (char_u *)"";
|
|
344
|
11789 pf = smp[n].sm_rules;
|
|
|
11790 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
11791 if (p0 == 1 && z == 0)
|
|
|
11792 {
|
|
|
11793 /* rule with '<' is used */
|
|
344
|
11794 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
11795 || res[reslen - 1] == *s))
|
|
|
11796 reslen--;
|
|
323
|
11797 z0 = 1;
|
|
|
11798 z = 1;
|
|
|
11799 k0 = 0;
|
|
372
|
11800 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
11801 {
|
|
|
11802 word[i + k0] = *s;
|
|
|
11803 k0++;
|
|
|
11804 s++;
|
|
|
11805 }
|
|
|
11806 if (k > k0)
|
|
|
11807 mch_memmove(word + i + k0, word + i + k,
|
|
|
11808 STRLEN(word + i + k) + 1);
|
|
|
11809
|
|
|
11810 /* new "actual letter" */
|
|
|
11811 c = word[i];
|
|
|
11812 }
|
|
|
11813 else
|
|
|
11814 {
|
|
|
11815 /* no '<' rule used */
|
|
|
11816 i += k - 1;
|
|
|
11817 z = 0;
|
|
344
|
11818 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
11819 {
|
|
344
|
11820 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
11821 res[reslen++] = *s;
|
|
323
|
11822 s++;
|
|
|
11823 }
|
|
|
11824 /* new "actual letter" */
|
|
|
11825 c = *s;
|
|
344
|
11826 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
11827 {
|
|
|
11828 if (c != NUL)
|
|
372
|
11829 res[reslen++] = c;
|
|
323
|
11830 mch_memmove(word, word + i + 1,
|
|
|
11831 STRLEN(word + i + 1) + 1);
|
|
|
11832 i = 0;
|
|
|
11833 z0 = 1;
|
|
|
11834 }
|
|
|
11835 }
|
|
|
11836 break;
|
|
|
11837 }
|
|
|
11838 }
|
|
|
11839 }
|
|
324
|
11840 else if (vim_iswhite(c))
|
|
|
11841 {
|
|
|
11842 c = ' ';
|
|
|
11843 k = 1;
|
|
|
11844 }
|
|
323
|
11845
|
|
|
11846 if (z0 == 0)
|
|
|
11847 {
|
|
344
|
11848 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
11849 && (!slang->sl_collapse || reslen == 0
|
|
|
11850 || res[reslen - 1] != c))
|
|
323
|
11851 /* condense only double letters */
|
|
372
|
11852 res[reslen++] = c;
|
|
323
|
11853
|
|
|
11854 i++;
|
|
|
11855 z = 0;
|
|
|
11856 k = 0;
|
|
|
11857 }
|
|
|
11858 }
|
|
|
11859
|
|
344
|
11860 res[reslen] = NUL;
|
|
323
|
11861 }
|
|
|
11862
|
|
372
|
11863 #ifdef FEAT_MBYTE
|
|
|
11864 /*
|
|
|
11865 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
11866 * Multi-byte version of spell_soundfold().
|
|
|
11867 */
|
|
|
11868 static void
|
|
375
|
11869 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
11870 slang_T *slang;
|
|
|
11871 char_u *inword;
|
|
|
11872 char_u *res;
|
|
|
11873 {
|
|
375
|
11874 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
11875 int word[MAXWLEN];
|
|
|
11876 int wres[MAXWLEN];
|
|
|
11877 int l;
|
|
|
11878 char_u *s;
|
|
|
11879 int *ws;
|
|
|
11880 char_u *t;
|
|
|
11881 int *pf;
|
|
|
11882 int i, j, z;
|
|
|
11883 int reslen;
|
|
|
11884 int n, k = 0;
|
|
|
11885 int z0;
|
|
|
11886 int k0;
|
|
|
11887 int n0;
|
|
|
11888 int c;
|
|
|
11889 int pri;
|
|
|
11890 int p0 = -333;
|
|
|
11891 int c0;
|
|
|
11892 int did_white = FALSE;
|
|
|
11893
|
|
|
11894 /*
|
|
|
11895 * Convert the multi-byte string to a wide-character string.
|
|
|
11896 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
11897 * But keep white space.
|
|
|
11898 */
|
|
|
11899 n = 0;
|
|
|
11900 for (s = inword; *s != NUL; )
|
|
|
11901 {
|
|
|
11902 t = s;
|
|
474
|
11903 c = mb_cptr2char_adv(&s);
|
|
372
|
11904 if (slang->sl_rem_accents)
|
|
|
11905 {
|
|
|
11906 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
11907 {
|
|
|
11908 if (did_white)
|
|
|
11909 continue;
|
|
|
11910 c = ' ';
|
|
|
11911 did_white = TRUE;
|
|
|
11912 }
|
|
|
11913 else
|
|
|
11914 {
|
|
|
11915 did_white = FALSE;
|
|
376
|
11916 if (!spell_iswordp_nmw(t))
|
|
372
|
11917 continue;
|
|
|
11918 }
|
|
|
11919 }
|
|
|
11920 word[n++] = c;
|
|
|
11921 }
|
|
|
11922 word[n] = NUL;
|
|
|
11923
|
|
|
11924 /*
|
|
|
11925 * This comes from Aspell phonet.cpp.
|
|
|
11926 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
11927 * Changed to keep spaces.
|
|
|
11928 */
|
|
|
11929 i = reslen = z = 0;
|
|
|
11930 while ((c = word[i]) != NUL)
|
|
|
11931 {
|
|
|
11932 /* Start with the first rule that has the character in the word. */
|
|
|
11933 n = slang->sl_sal_first[c & 0xff];
|
|
|
11934 z0 = 0;
|
|
|
11935
|
|
|
11936 if (n >= 0)
|
|
|
11937 {
|
|
375
|
11938 /* check all rules for the same index byte */
|
|
372
|
11939 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
11940 {
|
|
|
11941 /* Quickly skip entries that don't match the word. Most
|
|
|
11942 * entries are less then three chars, optimize for that. */
|
|
375
|
11943 if (c != ws[0])
|
|
|
11944 continue;
|
|
372
|
11945 k = smp[n].sm_leadlen;
|
|
|
11946 if (k > 1)
|
|
|
11947 {
|
|
|
11948 if (word[i + 1] != ws[1])
|
|
|
11949 continue;
|
|
|
11950 if (k > 2)
|
|
|
11951 {
|
|
|
11952 for (j = 2; j < k; ++j)
|
|
|
11953 if (word[i + j] != ws[j])
|
|
|
11954 break;
|
|
|
11955 if (j < k)
|
|
|
11956 continue;
|
|
|
11957 }
|
|
|
11958 }
|
|
|
11959
|
|
375
|
11960 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
11961 {
|
|
375
|
11962 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
11963 while (*pf != NUL && *pf != word[i + k])
|
|
|
11964 ++pf;
|
|
|
11965 if (*pf == NUL)
|
|
|
11966 continue;
|
|
|
11967 ++k;
|
|
|
11968 }
|
|
|
11969 s = smp[n].sm_rules;
|
|
|
11970 pri = 5; /* default priority */
|
|
|
11971
|
|
|
11972 p0 = *s;
|
|
|
11973 k0 = k;
|
|
|
11974 while (*s == '-' && k > 1)
|
|
|
11975 {
|
|
|
11976 k--;
|
|
|
11977 s++;
|
|
|
11978 }
|
|
|
11979 if (*s == '<')
|
|
|
11980 s++;
|
|
|
11981 if (VIM_ISDIGIT(*s))
|
|
|
11982 {
|
|
|
11983 /* determine priority */
|
|
|
11984 pri = *s - '0';
|
|
|
11985 s++;
|
|
|
11986 }
|
|
|
11987 if (*s == '^' && *(s + 1) == '^')
|
|
|
11988 s++;
|
|
|
11989
|
|
|
11990 if (*s == NUL
|
|
|
11991 || (*s == '^'
|
|
|
11992 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
11993 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
11994 && (*(s + 1) != '$'
|
|
376
|
11995 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
11996 || (*s == '$' && i > 0
|
|
376
|
11997 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
11998 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
11999 {
|
|
|
12000 /* search for followup rules, if: */
|
|
|
12001 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
12002 c0 = word[i + k - 1];
|
|
|
12003 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
12004
|
|
|
12005 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
12006 && p0 != '-' && word[i + k] != NUL)
|
|
|
12007 {
|
|
375
|
12008 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
12009 * all entries with the same index byte. */
|
|
372
|
12010 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
12011 == (c0 & 0xff); ++n0)
|
|
|
12012 {
|
|
|
12013 /* Quickly skip entries that don't match the word.
|
|
375
|
12014 */
|
|
|
12015 if (c0 != ws[0])
|
|
|
12016 continue;
|
|
372
|
12017 k0 = smp[n0].sm_leadlen;
|
|
|
12018 if (k0 > 1)
|
|
|
12019 {
|
|
|
12020 if (word[i + k] != ws[1])
|
|
|
12021 continue;
|
|
|
12022 if (k0 > 2)
|
|
|
12023 {
|
|
|
12024 pf = word + i + k + 1;
|
|
|
12025 for (j = 2; j < k0; ++j)
|
|
|
12026 if (*pf++ != ws[j])
|
|
|
12027 break;
|
|
|
12028 if (j < k0)
|
|
|
12029 continue;
|
|
|
12030 }
|
|
|
12031 }
|
|
|
12032 k0 += k - 1;
|
|
|
12033
|
|
375
|
12034 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
12035 {
|
|
|
12036 /* Check for match with one of the chars in
|
|
375
|
12037 * "sm_oneof". */
|
|
372
|
12038 while (*pf != NUL && *pf != word[i + k0])
|
|
|
12039 ++pf;
|
|
|
12040 if (*pf == NUL)
|
|
|
12041 continue;
|
|
|
12042 ++k0;
|
|
|
12043 }
|
|
|
12044
|
|
|
12045 p0 = 5;
|
|
|
12046 s = smp[n0].sm_rules;
|
|
|
12047 while (*s == '-')
|
|
|
12048 {
|
|
|
12049 /* "k0" gets NOT reduced because
|
|
|
12050 * "if (k0 == k)" */
|
|
|
12051 s++;
|
|
|
12052 }
|
|
|
12053 if (*s == '<')
|
|
|
12054 s++;
|
|
|
12055 if (VIM_ISDIGIT(*s))
|
|
|
12056 {
|
|
|
12057 p0 = *s - '0';
|
|
|
12058 s++;
|
|
|
12059 }
|
|
|
12060
|
|
|
12061 if (*s == NUL
|
|
|
12062 /* *s == '^' cuts */
|
|
|
12063 || (*s == '$'
|
|
376
|
12064 && !spell_iswordp_w(word + i + k0,
|
|
|
12065 curbuf)))
|
|
372
|
12066 {
|
|
|
12067 if (k0 == k)
|
|
|
12068 /* this is just a piece of the string */
|
|
|
12069 continue;
|
|
|
12070
|
|
|
12071 if (p0 < pri)
|
|
|
12072 /* priority too low */
|
|
|
12073 continue;
|
|
|
12074 /* rule fits; stop search */
|
|
|
12075 break;
|
|
|
12076 }
|
|
|
12077 }
|
|
|
12078
|
|
|
12079 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
12080 == (c0 & 0xff))
|
|
|
12081 continue;
|
|
|
12082 }
|
|
|
12083
|
|
|
12084 /* replace string */
|
|
|
12085 ws = smp[n].sm_to_w;
|
|
|
12086 s = smp[n].sm_rules;
|
|
|
12087 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
12088 if (p0 == 1 && z == 0)
|
|
|
12089 {
|
|
|
12090 /* rule with '<' is used */
|
|
389
|
12091 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
12092 && (wres[reslen - 1] == c
|
|
372
|
12093 || wres[reslen - 1] == *ws))
|
|
|
12094 reslen--;
|
|
|
12095 z0 = 1;
|
|
|
12096 z = 1;
|
|
|
12097 k0 = 0;
|
|
389
|
12098 if (ws != NULL)
|
|
|
12099 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
12100 {
|
|
|
12101 word[i + k0] = *ws;
|
|
|
12102 k0++;
|
|
|
12103 ws++;
|
|
|
12104 }
|
|
372
|
12105 if (k > k0)
|
|
|
12106 mch_memmove(word + i + k0, word + i + k,
|
|
|
12107 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
12108
|
|
|
12109 /* new "actual letter" */
|
|
|
12110 c = word[i];
|
|
|
12111 }
|
|
|
12112 else
|
|
|
12113 {
|
|
|
12114 /* no '<' rule used */
|
|
|
12115 i += k - 1;
|
|
|
12116 z = 0;
|
|
389
|
12117 if (ws != NULL)
|
|
|
12118 while (*ws != NUL && ws[1] != NUL
|
|
|
12119 && reslen < MAXWLEN)
|
|
|
12120 {
|
|
|
12121 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
12122 wres[reslen++] = *ws;
|
|
|
12123 ws++;
|
|
|
12124 }
|
|
372
|
12125 /* new "actual letter" */
|
|
389
|
12126 if (ws == NULL)
|
|
|
12127 c = NUL;
|
|
|
12128 else
|
|
|
12129 c = *ws;
|
|
372
|
12130 if (strstr((char *)s, "^^") != NULL)
|
|
|
12131 {
|
|
|
12132 if (c != NUL)
|
|
|
12133 wres[reslen++] = c;
|
|
|
12134 mch_memmove(word, word + i + 1,
|
|
|
12135 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
12136 i = 0;
|
|
|
12137 z0 = 1;
|
|
|
12138 }
|
|
|
12139 }
|
|
|
12140 break;
|
|
|
12141 }
|
|
|
12142 }
|
|
|
12143 }
|
|
|
12144 else if (vim_iswhite(c))
|
|
|
12145 {
|
|
|
12146 c = ' ';
|
|
|
12147 k = 1;
|
|
|
12148 }
|
|
|
12149
|
|
|
12150 if (z0 == 0)
|
|
|
12151 {
|
|
|
12152 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
12153 && (!slang->sl_collapse || reslen == 0
|
|
|
12154 || wres[reslen - 1] != c))
|
|
|
12155 /* condense only double letters */
|
|
|
12156 wres[reslen++] = c;
|
|
|
12157
|
|
|
12158 i++;
|
|
|
12159 z = 0;
|
|
|
12160 k = 0;
|
|
|
12161 }
|
|
|
12162 }
|
|
|
12163
|
|
|
12164 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
12165 l = 0;
|
|
|
12166 for (n = 0; n < reslen; ++n)
|
|
|
12167 {
|
|
|
12168 l += mb_char2bytes(wres[n], res + l);
|
|
|
12169 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
12170 break;
|
|
|
12171 }
|
|
|
12172 res[l] = NUL;
|
|
|
12173 }
|
|
|
12174 #endif
|
|
|
12175
|
|
324
|
12176 /*
|
|
344
|
12177 * Compute a score for two sound-a-like words.
|
|
|
12178 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
12179 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
12180 * avoiding checks that will not be possible.
|
|
|
12181 */
|
|
|
12182 static int
|
|
351
|
12183 soundalike_score(goodstart, badstart)
|
|
|
12184 char_u *goodstart; /* sound-folded good word */
|
|
|
12185 char_u *badstart; /* sound-folded bad word */
|
|
344
|
12186 {
|
|
351
|
12187 char_u *goodsound = goodstart;
|
|
|
12188 char_u *badsound = badstart;
|
|
|
12189 int goodlen;
|
|
|
12190 int badlen;
|
|
344
|
12191 int n;
|
|
|
12192 char_u *pl, *ps;
|
|
|
12193 char_u *pl2, *ps2;
|
|
351
|
12194 int score = 0;
|
|
|
12195
|
|
|
12196 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
12197 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
12198 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
12199 {
|
|
|
12200 score = SCORE_DEL / 2;
|
|
|
12201 if (*badsound == '*')
|
|
|
12202 ++badsound;
|
|
|
12203 else
|
|
|
12204 ++goodsound;
|
|
|
12205 }
|
|
|
12206
|
|
|
12207 goodlen = STRLEN(goodsound);
|
|
|
12208 badlen = STRLEN(badsound);
|
|
344
|
12209
|
|
|
12210 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
12211 * changes. */
|
|
|
12212 n = goodlen - badlen;
|
|
|
12213 if (n < -2 || n > 2)
|
|
|
12214 return SCORE_MAXMAX;
|
|
|
12215
|
|
|
12216 if (n > 0)
|
|
|
12217 {
|
|
351
|
12218 pl = goodsound; /* goodsound is longest */
|
|
344
|
12219 ps = badsound;
|
|
|
12220 }
|
|
|
12221 else
|
|
|
12222 {
|
|
351
|
12223 pl = badsound; /* badsound is longest */
|
|
344
|
12224 ps = goodsound;
|
|
|
12225 }
|
|
|
12226
|
|
|
12227 /* Skip over the identical part. */
|
|
|
12228 while (*pl == *ps && *pl != NUL)
|
|
|
12229 {
|
|
|
12230 ++pl;
|
|
|
12231 ++ps;
|
|
|
12232 }
|
|
|
12233
|
|
|
12234 switch (n)
|
|
|
12235 {
|
|
|
12236 case -2:
|
|
|
12237 case 2:
|
|
|
12238 /*
|
|
|
12239 * Must delete two characters from "pl".
|
|
|
12240 */
|
|
|
12241 ++pl; /* first delete */
|
|
|
12242 while (*pl == *ps)
|
|
|
12243 {
|
|
|
12244 ++pl;
|
|
|
12245 ++ps;
|
|
|
12246 }
|
|
|
12247 /* strings must be equal after second delete */
|
|
|
12248 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
12249 return score + SCORE_DEL * 2;
|
|
344
|
12250
|
|
|
12251 /* Failed to compare. */
|
|
|
12252 break;
|
|
|
12253
|
|
|
12254 case -1:
|
|
|
12255 case 1:
|
|
|
12256 /*
|
|
|
12257 * Minimal one delete from "pl" required.
|
|
|
12258 */
|
|
|
12259
|
|
|
12260 /* 1: delete */
|
|
|
12261 pl2 = pl + 1;
|
|
|
12262 ps2 = ps;
|
|
|
12263 while (*pl2 == *ps2)
|
|
|
12264 {
|
|
|
12265 if (*pl2 == NUL) /* reached the end */
|
|
351
|
12266 return score + SCORE_DEL;
|
|
344
|
12267 ++pl2;
|
|
|
12268 ++ps2;
|
|
|
12269 }
|
|
|
12270
|
|
|
12271 /* 2: delete then swap, then rest must be equal */
|
|
|
12272 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
12273 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
12274 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
12275
|
|
|
12276 /* 3: delete then substitute, then the rest must be equal */
|
|
|
12277 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
12278 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
12279
|
|
|
12280 /* 4: first swap then delete */
|
|
|
12281 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
12282 {
|
|
|
12283 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
12284 ps2 = ps + 2;
|
|
|
12285 while (*pl2 == *ps2)
|
|
|
12286 {
|
|
|
12287 ++pl2;
|
|
|
12288 ++ps2;
|
|
|
12289 }
|
|
|
12290 /* delete a char and then strings must be equal */
|
|
|
12291 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
12292 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
12293 }
|
|
|
12294
|
|
|
12295 /* 5: first substitute then delete */
|
|
|
12296 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
12297 ps2 = ps + 1;
|
|
|
12298 while (*pl2 == *ps2)
|
|
|
12299 {
|
|
|
12300 ++pl2;
|
|
|
12301 ++ps2;
|
|
|
12302 }
|
|
|
12303 /* delete a char and then strings must be equal */
|
|
|
12304 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
12305 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
12306
|
|
|
12307 /* Failed to compare. */
|
|
|
12308 break;
|
|
|
12309
|
|
|
12310 case 0:
|
|
|
12311 /*
|
|
|
12312 * Lenghts are equal, thus changes must result in same length: An
|
|
|
12313 * insert is only possible in combination with a delete.
|
|
|
12314 * 1: check if for identical strings
|
|
|
12315 */
|
|
|
12316 if (*pl == NUL)
|
|
351
|
12317 return score;
|
|
344
|
12318
|
|
|
12319 /* 2: swap */
|
|
|
12320 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
12321 {
|
|
|
12322 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
12323 ps2 = ps + 2;
|
|
|
12324 while (*pl2 == *ps2)
|
|
|
12325 {
|
|
|
12326 if (*pl2 == NUL) /* reached the end */
|
|
351
|
12327 return score + SCORE_SWAP;
|
|
344
|
12328 ++pl2;
|
|
|
12329 ++ps2;
|
|
|
12330 }
|
|
|
12331 /* 3: swap and swap again */
|
|
|
12332 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
12333 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
12334 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
12335
|
|
|
12336 /* 4: swap and substitute */
|
|
|
12337 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
12338 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
12339 }
|
|
|
12340
|
|
|
12341 /* 5: substitute */
|
|
|
12342 pl2 = pl + 1;
|
|
|
12343 ps2 = ps + 1;
|
|
|
12344 while (*pl2 == *ps2)
|
|
|
12345 {
|
|
|
12346 if (*pl2 == NUL) /* reached the end */
|
|
351
|
12347 return score + SCORE_SUBST;
|
|
344
|
12348 ++pl2;
|
|
|
12349 ++ps2;
|
|
|
12350 }
|
|
|
12351
|
|
|
12352 /* 6: substitute and swap */
|
|
|
12353 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
12354 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
12355 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
12356
|
|
|
12357 /* 7: substitute and substitute */
|
|
|
12358 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
12359 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
12360
|
|
|
12361 /* 8: insert then delete */
|
|
|
12362 pl2 = pl;
|
|
|
12363 ps2 = ps + 1;
|
|
|
12364 while (*pl2 == *ps2)
|
|
|
12365 {
|
|
|
12366 ++pl2;
|
|
|
12367 ++ps2;
|
|
|
12368 }
|
|
|
12369 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
12370 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
12371
|
|
|
12372 /* 9: delete then insert */
|
|
|
12373 pl2 = pl + 1;
|
|
|
12374 ps2 = ps;
|
|
|
12375 while (*pl2 == *ps2)
|
|
|
12376 {
|
|
|
12377 ++pl2;
|
|
|
12378 ++ps2;
|
|
|
12379 }
|
|
|
12380 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
12381 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
12382
|
|
|
12383 /* Failed to compare. */
|
|
|
12384 break;
|
|
|
12385 }
|
|
|
12386
|
|
|
12387 return SCORE_MAXMAX;
|
|
|
12388 }
|
|
324
|
12389
|
|
323
|
12390 /*
|
|
|
12391 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
12392 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
12393 *
|
|
485
|
12394 * The algorithm is described by Du and Chang, 1992.
|
|
|
12395 * The implementation of the algorithm comes from Aspell editdist.cpp,
|
|
|
12396 * edit_distance(). It has been converted from C++ to C and modified to
|
|
|
12397 * support multi-byte characters.
|
|
323
|
12398 */
|
|
|
12399 static int
|
|
|
12400 spell_edit_score(badword, goodword)
|
|
|
12401 char_u *badword;
|
|
|
12402 char_u *goodword;
|
|
|
12403 {
|
|
|
12404 int *cnt;
|
|
372
|
12405 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
12406 int j, i;
|
|
|
12407 int t;
|
|
|
12408 int bc, gc;
|
|
324
|
12409 int pbc, pgc;
|
|
|
12410 #ifdef FEAT_MBYTE
|
|
|
12411 char_u *p;
|
|
|
12412 int wbadword[MAXWLEN];
|
|
|
12413 int wgoodword[MAXWLEN];
|
|
|
12414
|
|
|
12415 if (has_mbyte)
|
|
|
12416 {
|
|
|
12417 /* Get the characters from the multi-byte strings and put them in an
|
|
|
12418 * int array for easy access. */
|
|
|
12419 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
12420 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
12421 wbadword[badlen++] = 0;
|
|
324
|
12422 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
12423 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
12424 wgoodword[goodlen++] = 0;
|
|
324
|
12425 }
|
|
|
12426 else
|
|
|
12427 #endif
|
|
|
12428 {
|
|
|
12429 badlen = STRLEN(badword) + 1;
|
|
|
12430 goodlen = STRLEN(goodword) + 1;
|
|
|
12431 }
|
|
323
|
12432
|
|
|
12433 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
12434 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
12435 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
12436 TRUE);
|
|
324
|
12437 if (cnt == NULL)
|
|
|
12438 return 0; /* out of memory */
|
|
323
|
12439
|
|
|
12440 CNT(0, 0) = 0;
|
|
|
12441 for (j = 1; j <= goodlen; ++j)
|
|
|
12442 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
12443
|
|
|
12444 for (i = 1; i <= badlen; ++i)
|
|
|
12445 {
|
|
|
12446 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
12447 for (j = 1; j <= goodlen; ++j)
|
|
|
12448 {
|
|
324
|
12449 #ifdef FEAT_MBYTE
|
|
|
12450 if (has_mbyte)
|
|
|
12451 {
|
|
|
12452 bc = wbadword[i - 1];
|
|
|
12453 gc = wgoodword[j - 1];
|
|
|
12454 }
|
|
|
12455 else
|
|
|
12456 #endif
|
|
|
12457 {
|
|
|
12458 bc = badword[i - 1];
|
|
|
12459 gc = goodword[j - 1];
|
|
|
12460 }
|
|
323
|
12461 if (bc == gc)
|
|
|
12462 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
12463 else
|
|
|
12464 {
|
|
|
12465 /* Use a better score when there is only a case difference. */
|
|
324
|
12466 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
12467 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
12468 else
|
|
|
12469 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
12470
|
|
324
|
12471 if (i > 1 && j > 1)
|
|
323
|
12472 {
|
|
324
|
12473 #ifdef FEAT_MBYTE
|
|
|
12474 if (has_mbyte)
|
|
|
12475 {
|
|
|
12476 pbc = wbadword[i - 2];
|
|
|
12477 pgc = wgoodword[j - 2];
|
|
|
12478 }
|
|
|
12479 else
|
|
|
12480 #endif
|
|
|
12481 {
|
|
|
12482 pbc = badword[i - 2];
|
|
|
12483 pgc = goodword[j - 2];
|
|
|
12484 }
|
|
|
12485 if (bc == pgc && pbc == gc)
|
|
|
12486 {
|
|
|
12487 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
12488 if (t < CNT(i, j))
|
|
|
12489 CNT(i, j) = t;
|
|
|
12490 }
|
|
323
|
12491 }
|
|
|
12492 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
12493 if (t < CNT(i, j))
|
|
|
12494 CNT(i, j) = t;
|
|
|
12495 t = SCORE_INS + CNT(i, j - 1);
|
|
|
12496 if (t < CNT(i, j))
|
|
|
12497 CNT(i, j) = t;
|
|
|
12498 }
|
|
|
12499 }
|
|
|
12500 }
|
|
344
|
12501
|
|
|
12502 i = CNT(badlen - 1, goodlen - 1);
|
|
|
12503 vim_free(cnt);
|
|
|
12504 return i;
|
|
323
|
12505 }
|
|
307
|
12506
|
|
351
|
12507 /*
|
|
|
12508 * ":spelldump"
|
|
|
12509 */
|
|
|
12510 /*ARGSUSED*/
|
|
|
12511 void
|
|
|
12512 ex_spelldump(eap)
|
|
|
12513 exarg_T *eap;
|
|
|
12514 {
|
|
|
12515 buf_T *buf = curbuf;
|
|
|
12516 langp_T *lp;
|
|
|
12517 slang_T *slang;
|
|
|
12518 idx_T arridx[MAXWLEN];
|
|
|
12519 int curi[MAXWLEN];
|
|
|
12520 char_u word[MAXWLEN];
|
|
|
12521 int c;
|
|
|
12522 char_u *byts;
|
|
|
12523 idx_T *idxs;
|
|
|
12524 linenr_T lnum = 0;
|
|
|
12525 int round;
|
|
|
12526 int depth;
|
|
|
12527 int n;
|
|
|
12528 int flags;
|
|
381
|
12529 char_u *region_names = NULL; /* region names being used */
|
|
|
12530 int do_region = TRUE; /* dump region names and numbers */
|
|
|
12531 char_u *p;
|
|
500
|
12532 int lpi;
|
|
351
|
12533
|
|
498
|
12534 if (no_spell_checking(curwin))
|
|
351
|
12535 return;
|
|
|
12536
|
|
|
12537 /* Create a new empty buffer by splitting the window. */
|
|
|
12538 do_cmdline_cmd((char_u *)"new");
|
|
|
12539 if (!bufempty() || !buf_valid(buf))
|
|
|
12540 return;
|
|
|
12541
|
|
381
|
12542 /* Find out if we can support regions: All languages must support the same
|
|
|
12543 * regions or none at all. */
|
|
500
|
12544 for (lpi = 0; lpi < buf->b_langp.ga_len; ++lpi)
|
|
|
12545 {
|
|
|
12546 lp = LANGP_ENTRY(buf->b_langp, lpi);
|
|
381
|
12547 p = lp->lp_slang->sl_regions;
|
|
|
12548 if (p[0] != 0)
|
|
|
12549 {
|
|
|
12550 if (region_names == NULL) /* first language with regions */
|
|
|
12551 region_names = p;
|
|
|
12552 else if (STRCMP(region_names, p) != 0)
|
|
|
12553 {
|
|
|
12554 do_region = FALSE; /* region names are different */
|
|
|
12555 break;
|
|
|
12556 }
|
|
|
12557 }
|
|
|
12558 }
|
|
|
12559
|
|
|
12560 if (do_region && region_names != NULL)
|
|
|
12561 {
|
|
|
12562 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
12563 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
12564 }
|
|
|
12565 else
|
|
|
12566 do_region = FALSE;
|
|
|
12567
|
|
|
12568 /*
|
|
|
12569 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
12570 */
|
|
500
|
12571 for (lpi = 0; lpi < buf->b_langp.ga_len; ++lpi)
|
|
|
12572 {
|
|
|
12573 lp = LANGP_ENTRY(buf->b_langp, lpi);
|
|
351
|
12574 slang = lp->lp_slang;
|
|
500
|
12575 if (slang->sl_fbyts == NULL) /* reloading failed */
|
|
|
12576 continue;
|
|
351
|
12577
|
|
|
12578 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
12579 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
12580
|
|
|
12581 /* round 1: case-folded tree
|
|
|
12582 * round 2: keep-case tree */
|
|
|
12583 for (round = 1; round <= 2; ++round)
|
|
|
12584 {
|
|
|
12585 if (round == 1)
|
|
|
12586 {
|
|
|
12587 byts = slang->sl_fbyts;
|
|
|
12588 idxs = slang->sl_fidxs;
|
|
|
12589 }
|
|
|
12590 else
|
|
|
12591 {
|
|
|
12592 byts = slang->sl_kbyts;
|
|
|
12593 idxs = slang->sl_kidxs;
|
|
|
12594 }
|
|
|
12595 if (byts == NULL)
|
|
|
12596 continue; /* array is empty */
|
|
|
12597
|
|
|
12598 depth = 0;
|
|
|
12599 arridx[0] = 0;
|
|
|
12600 curi[0] = 1;
|
|
|
12601 while (depth >= 0 && !got_int)
|
|
|
12602 {
|
|
|
12603 if (curi[depth] > byts[arridx[depth]])
|
|
|
12604 {
|
|
|
12605 /* Done all bytes at this node, go up one level. */
|
|
|
12606 --depth;
|
|
|
12607 line_breakcheck();
|
|
|
12608 }
|
|
|
12609 else
|
|
|
12610 {
|
|
|
12611 /* Do one more byte at this node. */
|
|
|
12612 n = arridx[depth] + curi[depth];
|
|
|
12613 ++curi[depth];
|
|
|
12614 c = byts[n];
|
|
|
12615 if (c == 0)
|
|
|
12616 {
|
|
|
12617 /* End of word, deal with the word.
|
|
|
12618 * Don't use keep-case words in the fold-case tree,
|
|
|
12619 * they will appear in the keep-case tree.
|
|
|
12620 * Only use the word when the region matches. */
|
|
|
12621 flags = (int)idxs[n];
|
|
|
12622 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
500
|
12623 && (flags & WF_NEEDCOMP) == 0
|
|
381
|
12624 && (do_region
|
|
|
12625 || (flags & WF_REGION) == 0
|
|
390
|
12626 || (((unsigned)flags >> 16)
|
|
351
|
12627 & lp->lp_region) != 0))
|
|
|
12628 {
|
|
|
12629 word[depth] = NUL;
|
|
381
|
12630 if (!do_region)
|
|
|
12631 flags &= ~WF_REGION;
|
|
355
|
12632
|
|
|
12633 /* Dump the basic word if there is no prefix or
|
|
|
12634 * when it's the first one. */
|
|
390
|
12635 c = (unsigned)flags >> 24;
|
|
355
|
12636 if (c == 0 || curi[depth] == 2)
|
|
|
12637 dump_word(word, round, flags, lnum++);
|
|
351
|
12638
|
|
|
12639 /* Apply the prefix, if there is one. */
|
|
355
|
12640 if (c != 0)
|
|
500
|
12641 lnum = dump_prefixes(slang, word, round,
|
|
351
|
12642 flags, lnum);
|
|
|
12643 }
|
|
|
12644 }
|
|
|
12645 else
|
|
|
12646 {
|
|
|
12647 /* Normal char, go one level deeper. */
|
|
|
12648 word[depth++] = c;
|
|
|
12649 arridx[depth] = idxs[n];
|
|
|
12650 curi[depth] = 1;
|
|
|
12651 }
|
|
|
12652 }
|
|
|
12653 }
|
|
|
12654 }
|
|
|
12655 }
|
|
|
12656
|
|
|
12657 /* Delete the empty line that we started with. */
|
|
|
12658 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
12659 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
12660
|
|
|
12661 redraw_later(NOT_VALID);
|
|
|
12662 }
|
|
|
12663
|
|
|
12664 /*
|
|
|
12665 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
12666 */
|
|
|
12667 static void
|
|
|
12668 dump_word(word, round, flags, lnum)
|
|
|
12669 char_u *word;
|
|
|
12670 int round;
|
|
|
12671 int flags;
|
|
|
12672 linenr_T lnum;
|
|
|
12673 {
|
|
|
12674 int keepcap = FALSE;
|
|
|
12675 char_u *p;
|
|
|
12676 char_u cword[MAXWLEN];
|
|
381
|
12677 char_u badword[MAXWLEN + 10];
|
|
|
12678 int i;
|
|
351
|
12679
|
|
|
12680 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
12681 {
|
|
|
12682 /* Need to fix case according to "flags". */
|
|
|
12683 make_case_word(word, cword, flags);
|
|
|
12684 p = cword;
|
|
|
12685 }
|
|
|
12686 else
|
|
|
12687 {
|
|
|
12688 p = word;
|
|
389
|
12689 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
12690 || (flags & WF_FIXCAP) != 0))
|
|
351
|
12691 keepcap = TRUE;
|
|
|
12692 }
|
|
|
12693
|
|
381
|
12694 /* Add flags and regions after a slash. */
|
|
|
12695 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
12696 {
|
|
|
12697 STRCPY(badword, p);
|
|
|
12698 STRCAT(badword, "/");
|
|
351
|
12699 if (keepcap)
|
|
|
12700 STRCAT(badword, "=");
|
|
|
12701 if (flags & WF_BANNED)
|
|
|
12702 STRCAT(badword, "!");
|
|
|
12703 else if (flags & WF_RARE)
|
|
|
12704 STRCAT(badword, "?");
|
|
381
|
12705 if (flags & WF_REGION)
|
|
|
12706 for (i = 0; i < 7; ++i)
|
|
390
|
12707 if (flags & (0x10000 << i))
|
|
381
|
12708 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
12709 p = badword;
|
|
|
12710 }
|
|
|
12711
|
|
|
12712 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
12713 }
|
|
|
12714
|
|
|
12715 /*
|
|
372
|
12716 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
12717 * "word" and append a line to the buffer.
|
|
351
|
12718 * Return the updated line number.
|
|
|
12719 */
|
|
|
12720 static linenr_T
|
|
500
|
12721 dump_prefixes(slang, word, round, flags, startlnum)
|
|
351
|
12722 slang_T *slang;
|
|
|
12723 char_u *word; /* case-folded word */
|
|
|
12724 int round;
|
|
|
12725 int flags; /* flags with prefix ID */
|
|
|
12726 linenr_T startlnum;
|
|
|
12727 {
|
|
|
12728 idx_T arridx[MAXWLEN];
|
|
|
12729 int curi[MAXWLEN];
|
|
|
12730 char_u prefix[MAXWLEN];
|
|
455
|
12731 char_u word_up[MAXWLEN];
|
|
|
12732 int has_word_up = FALSE;
|
|
351
|
12733 int c;
|
|
|
12734 char_u *byts;
|
|
|
12735 idx_T *idxs;
|
|
|
12736 linenr_T lnum = startlnum;
|
|
|
12737 int depth;
|
|
|
12738 int n;
|
|
|
12739 int len;
|
|
|
12740 int i;
|
|
|
12741
|
|
455
|
12742 /* if the word starts with a lower-case letter make the word with an
|
|
|
12743 * upper-case letter in word_up[]. */
|
|
|
12744 c = PTR2CHAR(word);
|
|
|
12745 if (SPELL_TOUPPER(c) != c)
|
|
|
12746 {
|
|
|
12747 onecap_copy(word, word_up, TRUE);
|
|
|
12748 has_word_up = TRUE;
|
|
|
12749 }
|
|
|
12750
|
|
351
|
12751 byts = slang->sl_pbyts;
|
|
|
12752 idxs = slang->sl_pidxs;
|
|
|
12753 if (byts != NULL) /* array not is empty */
|
|
|
12754 {
|
|
|
12755 /*
|
|
|
12756 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
12757 * When at the end of a prefix check that it supports "flags".
|
|
351
|
12758 */
|
|
|
12759 depth = 0;
|
|
|
12760 arridx[0] = 0;
|
|
|
12761 curi[0] = 1;
|
|
|
12762 while (depth >= 0 && !got_int)
|
|
|
12763 {
|
|
390
|
12764 n = arridx[depth];
|
|
|
12765 len = byts[n];
|
|
|
12766 if (curi[depth] > len)
|
|
351
|
12767 {
|
|
|
12768 /* Done all bytes at this node, go up one level. */
|
|
|
12769 --depth;
|
|
|
12770 line_breakcheck();
|
|
|
12771 }
|
|
|
12772 else
|
|
|
12773 {
|
|
|
12774 /* Do one more byte at this node. */
|
|
390
|
12775 n += curi[depth];
|
|
351
|
12776 ++curi[depth];
|
|
|
12777 c = byts[n];
|
|
|
12778 if (c == 0)
|
|
|
12779 {
|
|
|
12780 /* End of prefix, find out how many IDs there are. */
|
|
|
12781 for (i = 1; i < len; ++i)
|
|
|
12782 if (byts[n + i] != 0)
|
|
|
12783 break;
|
|
|
12784 curi[depth] += i - 1;
|
|
|
12785
|
|
455
|
12786 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
12787 if (c != 0)
|
|
351
|
12788 {
|
|
376
|
12789 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
12790 dump_word(prefix, round,
|
|
455
|
12791 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
12792 : flags, lnum++);
|
|
351
|
12793 }
|
|
455
|
12794
|
|
|
12795 /* Check for prefix that matches the word when the
|
|
|
12796 * first letter is upper-case, but only if the prefix has
|
|
|
12797 * a condition. */
|
|
|
12798 if (has_word_up)
|
|
|
12799 {
|
|
|
12800 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
12801 TRUE);
|
|
|
12802 if (c != 0)
|
|
|
12803 {
|
|
|
12804 vim_strncpy(prefix + depth, word_up,
|
|
|
12805 MAXWLEN - depth - 1);
|
|
|
12806 dump_word(prefix, round,
|
|
|
12807 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
12808 : flags, lnum++);
|
|
|
12809 }
|
|
|
12810 }
|
|
351
|
12811 }
|
|
|
12812 else
|
|
|
12813 {
|
|
|
12814 /* Normal char, go one level deeper. */
|
|
|
12815 prefix[depth++] = c;
|
|
|
12816 arridx[depth] = idxs[n];
|
|
|
12817 curi[depth] = 1;
|
|
|
12818 }
|
|
|
12819 }
|
|
|
12820 }
|
|
|
12821 }
|
|
|
12822
|
|
|
12823 return lnum;
|
|
|
12824 }
|
|
|
12825
|
|
498
|
12826 /*
|
|
|
12827 * Move "p" to end of word.
|
|
|
12828 */
|
|
|
12829 char_u *
|
|
|
12830 spell_to_word_end(start, buf)
|
|
|
12831 char_u *start;
|
|
|
12832 buf_T *buf;
|
|
|
12833 {
|
|
|
12834 char_u *p = start;
|
|
|
12835
|
|
|
12836 while (*p != NUL && spell_iswordp(p, buf))
|
|
|
12837 mb_ptr_adv(p);
|
|
|
12838 return p;
|
|
|
12839 }
|
|
|
12840
|
|
475
|
12841 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
12842 /*
|
|
|
12843 * Find start of the word in front of the cursor. We don't check if it is
|
|
|
12844 * badly spelled, with completion we can only change the word in front of the
|
|
|
12845 * cursor.
|
|
|
12846 * Used for Insert mode completion CTRL-X ?.
|
|
|
12847 * Returns the column number of the word.
|
|
|
12848 */
|
|
|
12849 int
|
|
|
12850 spell_word_start(startcol)
|
|
|
12851 int startcol;
|
|
|
12852 {
|
|
|
12853 char_u *line;
|
|
|
12854 char_u *p;
|
|
|
12855 int col = 0;
|
|
|
12856
|
|
498
|
12857 if (no_spell_checking(curwin))
|
|
475
|
12858 return startcol;
|
|
|
12859
|
|
|
12860 /* Find a word character before "startcol". */
|
|
|
12861 line = ml_get_curline();
|
|
|
12862 for (p = line + startcol; p > line; )
|
|
|
12863 {
|
|
|
12864 mb_ptr_back(line, p);
|
|
|
12865 if (spell_iswordp_nmw(p))
|
|
|
12866 break;
|
|
|
12867 }
|
|
|
12868
|
|
|
12869 /* Go back to start of the word. */
|
|
|
12870 while (p > line)
|
|
|
12871 {
|
|
|
12872 col = p - line;
|
|
|
12873 mb_ptr_back(line, p);
|
|
|
12874 if (!spell_iswordp(p, curbuf))
|
|
|
12875 break;
|
|
|
12876 col = 0;
|
|
|
12877 }
|
|
|
12878
|
|
535
|
12879 return col;
|
|
|
12880 }
|
|
|
12881
|
|
|
12882 /*
|
|
|
12883 * Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
12884 * expand_spelling() is called. Therefore the ugly global variable.
|
|
|
12885 */
|
|
|
12886 static int spell_expand_need_cap;
|
|
|
12887
|
|
|
12888 void
|
|
|
12889 spell_expand_check_cap(col)
|
|
|
12890 colnr_T col;
|
|
|
12891 {
|
|
475
|
12892 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
12893 }
|
|
|
12894
|
|
|
12895 /*
|
|
|
12896 * Get list of spelling suggestions.
|
|
|
12897 * Used for Insert mode completion CTRL-X ?.
|
|
|
12898 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
12899 * allocated strings.
|
|
|
12900 */
|
|
|
12901 /*ARGSUSED*/
|
|
|
12902 int
|
|
|
12903 expand_spelling(lnum, col, pat, matchp)
|
|
|
12904 linenr_T lnum;
|
|
|
12905 int col;
|
|
|
12906 char_u *pat;
|
|
|
12907 char_u ***matchp;
|
|
|
12908 {
|
|
|
12909 garray_T ga;
|
|
|
12910
|
|
|
12911 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap);
|
|
|
12912 *matchp = ga.ga_data;
|
|
|
12913 return ga.ga_len;
|
|
|
12914 }
|
|
|
12915 #endif
|
|
|
12916
|
|
236
|
12917 #endif /* FEAT_SYN_HL */
|
