|
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 *
|
|
|
165 * sectionID == SN_SYLLABLE: <syllable>
|
|
|
166 * <syllable> N bytes String from SYLLABLE item.
|
|
300
|
167 *
|
|
|
168 * <LWORDTREE>: <wordtree>
|
|
|
169 *
|
|
339
|
170 * <KWORDTREE>: <wordtree>
|
|
|
171 *
|
|
|
172 * <PREFIXTREE>: <wordtree>
|
|
|
173 *
|
|
|
174 *
|
|
300
|
175 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
176 *
|
|
|
177 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
178 *
|
|
|
179 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
180 *
|
|
|
181 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
182 * follow in sorted order.
|
|
|
183 *
|
|
339
|
184 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
481
|
185 * | <flags> [<flags2>] [<region>] [<affixID>]
|
|
|
186 * | [<pflags>] <affixID> <prefcondnr> ]
|
|
300
|
187 *
|
|
|
188 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
189 * BY_NOFLAGS: End of word without flags and for all
|
|
|
190 * regions.
|
|
481
|
191 * For PREFIXTREE <affixID> and
|
|
366
|
192 * <prefcondnr> follow.
|
|
390
|
193 * BY_FLAGS: End of word, <flags> follow.
|
|
481
|
194 * For PREFIXTREE <pflags>, <affixID>
|
|
455
|
195 * and <prefcondnr> follow.
|
|
390
|
196 * BY_FLAGS2: End of word, <flags> and <flags2>
|
|
|
197 * follow. Not used in PREFIXTREE.
|
|
|
198 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
300
|
199 * and <xbyte> follow.
|
|
|
200 *
|
|
|
201 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
202 *
|
|
|
203 * <xbyte> 1 byte byte value of the sibling.
|
|
|
204 *
|
|
|
205 * <flags> 1 byte bitmask of:
|
|
|
206 * WF_ALLCAP word must have only capitals
|
|
|
207 * WF_ONECAP first char of word must be capital
|
|
389
|
208 * WF_KEEPCAP keep-case word
|
|
|
209 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
210 * WF_RARE rare word
|
|
389
|
211 * WF_BANNED bad word
|
|
300
|
212 * WF_REGION <region> follows
|
|
481
|
213 * WF_AFX <affixID> follows
|
|
300
|
214 *
|
|
390
|
215 * <flags2> 1 byte Only used when there are postponed prefixes.
|
|
|
216 * Bitmask of:
|
|
|
217 * WF_HAS_AFF >> 8 word includes affix
|
|
|
218 *
|
|
455
|
219 * <pflags> 1 byte bitmask of:
|
|
|
220 * WFP_RARE rare prefix
|
|
|
221 * WFP_NC non-combining prefix
|
|
|
222 * WFP_UP letter after prefix made upper case
|
|
|
223 *
|
|
300
|
224 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
225 * omitted it's valid in all regions.
|
|
|
226 * Lowest bit is for region 1.
|
|
|
227 *
|
|
481
|
228 * <affixID> 1 byte ID of affix that can be used with this word. In
|
|
339
|
229 * PREFIXTREE used for the required prefix ID.
|
|
|
230 *
|
|
|
231 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
232 * from HEADER.
|
|
300
|
233 *
|
|
|
234 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
235 */
|
|
|
236
|
|
223
|
237 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
238 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
239 #endif
|
|
|
240
|
|
|
241 #include "vim.h"
|
|
|
242
|
|
|
243 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
244
|
|
|
245 #ifdef HAVE_FCNTL_H
|
|
|
246 # include <fcntl.h>
|
|
|
247 #endif
|
|
|
248
|
|
323
|
249 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
250 Some places assume a word length fits in a
|
|
|
251 byte, thus it can't be above 255. */
|
|
226
|
252
|
|
490
|
253 /* Type used for indexes in the word tree need to be at least 4 bytes. If int
|
|
324
|
254 * is 8 bytes we could use something smaller, but what? */
|
|
490
|
255 #if SIZEOF_INT > 3
|
|
324
|
256 typedef int idx_T;
|
|
|
257 #else
|
|
|
258 typedef long idx_T;
|
|
|
259 #endif
|
|
|
260
|
|
|
261 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
262 * comes above it. */
|
|
300
|
263 #define WF_REGION 0x01 /* region byte follows */
|
|
|
264 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
265 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
266 #define WF_RARE 0x08 /* rare word */
|
|
307
|
267 #define WF_BANNED 0x10 /* bad word */
|
|
481
|
268 #define WF_AFX 0x20 /* affix ID follows */
|
|
389
|
269 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
270 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
271
|
|
390
|
272 /* for <flags2>, shifted up one byte to be used in wn_flags */
|
|
|
273 #define WF_HAS_AFF 0x0100 /* word includes affix */
|
|
|
274
|
|
389
|
275 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
276
|
|
455
|
277 /* flags for <pflags> */
|
|
|
278 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
279 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
280 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
281
|
|
481
|
282 /* Flags for postponed prefixes. Must be above affixID (one byte)
|
|
390
|
283 * and prefcondnr (two bytes). */
|
|
455
|
284 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
285 * postponed prefix */
|
|
|
286 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
287 * postponed prefix */
|
|
|
288 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
289 * postponed prefix */
|
|
390
|
290
|
|
|
291 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
292 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
293 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
294 * postponed prefix: no <pflags> */
|
|
|
295 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
296 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
297 * postponed prefix: <pflags> follows */
|
|
|
298 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
299 * follow; never used in prefix tree */
|
|
|
300 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
301
|
|
323
|
302 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
303 * and si_sal. Not for sl_sal!
|
|
323
|
304 * One replacement: from "ft_from" to "ft_to". */
|
|
|
305 typedef struct fromto_S
|
|
236
|
306 {
|
|
323
|
307 char_u *ft_from;
|
|
|
308 char_u *ft_to;
|
|
|
309 } fromto_T;
|
|
236
|
310
|
|
344
|
311 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
312 * The info is split for quick processing by spell_soundfold().
|
|
|
313 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
314 typedef struct salitem_S
|
|
|
315 {
|
|
|
316 char_u *sm_lead; /* leading letters */
|
|
|
317 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
318 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
319 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
320 char_u *sm_to; /* replacement. */
|
|
372
|
321 #ifdef FEAT_MBYTE
|
|
|
322 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
323 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
324 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
325 #endif
|
|
344
|
326 } salitem_T;
|
|
|
327
|
|
375
|
328 #ifdef FEAT_MBYTE
|
|
|
329 typedef int salfirst_T;
|
|
|
330 #else
|
|
|
331 typedef short salfirst_T;
|
|
|
332 #endif
|
|
|
333
|
|
483
|
334 /* Values for SP_*ERROR are negative, positive values are used by
|
|
|
335 * read_cnt_string(). */
|
|
|
336 #define SP_TRUNCERROR -1 /* spell file truncated error */
|
|
|
337 #define SP_FORMERROR -2 /* format error in spell file */
|
|
|
338 #define SP_ERROR -3 /* other error while reading spell file */
|
|
|
339
|
|
236
|
340 /*
|
|
243
|
341 * Structure used to store words and other info for one language, loaded from
|
|
|
342 * a .spl file.
|
|
300
|
343 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
344 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
345 *
|
|
|
346 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
347 * the number of possible bytes, sorted on byte value:
|
|
|
348 * <len> <byte1> <byte2> ...
|
|
|
349 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
350 * byte in "byts".
|
|
|
351 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
481
|
352 * the flags, region mask and affixID for the word. There may be several
|
|
|
353 * zeros in sequence for alternative flag/region/affixID combinations.
|
|
236
|
354 */
|
|
|
355 typedef struct slang_S slang_T;
|
|
|
356 struct slang_S
|
|
|
357 {
|
|
|
358 slang_T *sl_next; /* next language */
|
|
|
359 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
360 char_u *sl_fname; /* name of .spl file */
|
|
323
|
361 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
362
|
|
300
|
363 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
364 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
365 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
366 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
367 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
368 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
369
|
|
236
|
370 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
371
|
|
376
|
372 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
373
|
|
483
|
374 int sl_compmax; /* COMPOUNDMAX (default: MAXWLEN) */
|
|
|
375 int sl_compminlen; /* COMPOUNDMIN (default: MAXWLEN) */
|
|
|
376 int sl_compsylmax; /* COMPOUNDSYLMAX (default: MAXWLEN) */
|
|
|
377 regprog_T *sl_compprog; /* COMPOUNDFLAGS turned into a regexp progrm
|
|
|
378 * (NULL when no compounding) */
|
|
|
379 char_u *sl_compstartflags; /* flags for first compound word */
|
|
485
|
380 char_u *sl_compallflags; /* all flags for compound words */
|
|
483
|
381 char_u *sl_syllable; /* SYLLABLE repeatable chars or NULL */
|
|
|
382 garray_T sl_syl_items; /* syllable items */
|
|
481
|
383
|
|
339
|
384 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
385 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
386
|
|
323
|
387 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
388 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
389 there is none */
|
|
344
|
390 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
391 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
392 there is none */
|
|
375
|
393 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
394 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
395 * "sl_sal" is a list of wide char lists. */
|
|
323
|
396 int sl_followup; /* SAL followup */
|
|
|
397 int sl_collapse; /* SAL collapse_result */
|
|
|
398 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
399 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
400 #ifdef FEAT_MBYTE
|
|
|
401 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
402 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
403 #else
|
|
|
404 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
405 #endif
|
|
236
|
406 };
|
|
|
407
|
|
243
|
408 /* First language that is loaded, start of the linked list of loaded
|
|
|
409 * languages. */
|
|
236
|
410 static slang_T *first_lang = NULL;
|
|
|
411
|
|
323
|
412 /* Flags used in .spl file for soundsalike flags. */
|
|
|
413 #define SAL_F0LLOWUP 1
|
|
|
414 #define SAL_COLLAPSE 2
|
|
|
415 #define SAL_REM_ACCENTS 4
|
|
|
416
|
|
236
|
417 /*
|
|
|
418 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
419 */
|
|
|
420 typedef struct langp_S
|
|
|
421 {
|
|
|
422 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
423 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
424 } langp_T;
|
|
|
425
|
|
|
426 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
427
|
|
307
|
428 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
429
|
|
483
|
430 #define VIMSPELLMAGIC "VIMspell" /* string at start of Vim spell file */
|
|
|
431 #define VIMSPELLMAGICL 8
|
|
|
432 #define VIMSPELLVERSION 50
|
|
|
433
|
|
|
434 /* Section IDs. Only renumber them when VIMSPELLVERSION changes! */
|
|
|
435 #define SN_REGION 0 /* <regionname> section */
|
|
|
436 #define SN_CHARFLAGS 1 /* charflags section */
|
|
|
437 #define SN_MIDWORD 2 /* <midword> section */
|
|
|
438 #define SN_PREFCOND 3 /* <prefcond> section */
|
|
|
439 #define SN_REP 4 /* REP items section */
|
|
|
440 #define SN_SAL 5 /* SAL items section */
|
|
|
441 #define SN_SOFO 6 /* soundfolding section */
|
|
|
442 #define SN_MAP 7 /* MAP items section */
|
|
|
443 #define SN_COMPOUND 8 /* compound words section */
|
|
|
444 #define SN_SYLLABLE 9 /* syllable section */
|
|
|
445 #define SN_END 255 /* end of sections */
|
|
|
446
|
|
|
447 #define SNF_REQUIRED 1 /* <sectionflags>: required section */
|
|
|
448
|
|
307
|
449 /* Result values. Lower number is accepted over higher one. */
|
|
|
450 #define SP_BANNED -1
|
|
236
|
451 #define SP_OK 0
|
|
307
|
452 #define SP_RARE 1
|
|
|
453 #define SP_LOCAL 2
|
|
|
454 #define SP_BAD 3
|
|
236
|
455
|
|
381
|
456 /* file used for "zG" and "zW" */
|
|
385
|
457 static char_u *int_wordlist = NULL;
|
|
381
|
458
|
|
236
|
459 /*
|
|
323
|
460 * Information used when looking for suggestions.
|
|
|
461 */
|
|
|
462 typedef struct suginfo_S
|
|
|
463 {
|
|
|
464 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
465 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
466 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
467 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
468 char_u *su_badptr; /* start of bad word in line */
|
|
|
469 int su_badlen; /* length of detected bad word in line */
|
|
346
|
470 int su_badflags; /* caps flags for bad word */
|
|
323
|
471 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
472 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
473 hashtab_T su_banned; /* table with banned words */
|
|
|
474 } suginfo_T;
|
|
|
475
|
|
|
476 /* One word suggestion. Used in "si_ga". */
|
|
|
477 typedef struct suggest_S
|
|
|
478 {
|
|
|
479 char_u *st_word; /* suggested word, allocated string */
|
|
|
480 int st_orglen; /* length of replaced text */
|
|
|
481 int st_score; /* lower is better */
|
|
344
|
482 int st_altscore; /* used when st_score compares equal */
|
|
|
483 int st_salscore; /* st_score is for soundalike */
|
|
324
|
484 int st_had_bonus; /* bonus already included in score */
|
|
323
|
485 } suggest_T;
|
|
|
486
|
|
344
|
487 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
488
|
|
324
|
489 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
490 * called the score may change, thus we need to keep more than what is
|
|
|
491 * displayed. */
|
|
351
|
492 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
493
|
|
|
494 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
495 * of suggestions that are not going to be displayed. */
|
|
344
|
496 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
497
|
|
|
498 /* score for various changes */
|
|
344
|
499 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
500 #define SCORE_ICASE 52 /* slightly different case */
|
|
483
|
501 #define SCORE_REGION 200 /* word is for different region */
|
|
323
|
502 #define SCORE_RARE 180 /* rare word */
|
|
|
503 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
504 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
505 #define SCORE_REP 87 /* REP replacement */
|
|
|
506 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
507 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
479
|
508 #define SCORE_SUBCOMP 33 /* substitute a composing character */
|
|
324
|
509 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
510 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
479
|
511 #define SCORE_DELCOMP 28 /* delete a composing character */
|
|
324
|
512 #define SCORE_INS 96 /* insert a character */
|
|
358
|
513 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
475
|
514 #define SCORE_INSCOMP 30 /* insert a composing character */
|
|
366
|
515 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
516
|
|
372
|
517 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
518 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
519 * 350 allows for about three changes. */
|
|
344
|
520
|
|
|
521 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
522 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
523
|
|
|
524 /*
|
|
236
|
525 * Structure to store info for word matching.
|
|
|
526 */
|
|
|
527 typedef struct matchinf_S
|
|
|
528 {
|
|
|
529 langp_T *mi_lp; /* info for language and region */
|
|
243
|
530
|
|
|
531 /* pointers to original text to be checked */
|
|
236
|
532 char_u *mi_word; /* start of word being checked */
|
|
339
|
533 char_u *mi_end; /* end of matching word so far */
|
|
243
|
534 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
535 char_u *mi_cend; /* char after what was used for
|
|
|
536 mi_capflags */
|
|
243
|
537
|
|
|
538 /* case-folded text */
|
|
|
539 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
540 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
541
|
|
339
|
542 /* for when checking word after a prefix */
|
|
|
543 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
481
|
544 affixID/condition */
|
|
339
|
545 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
546 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
547 #ifdef FEAT_MBYTE
|
|
|
548 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
549 case */
|
|
|
550 #else
|
|
|
551 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
552 #endif
|
|
339
|
553
|
|
481
|
554 /* for when checking a compound word */
|
|
|
555 int mi_compoff; /* start of following word offset */
|
|
483
|
556 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */
|
|
|
557 int mi_complen; /* nr of compound words used */
|
|
481
|
558
|
|
243
|
559 /* others */
|
|
236
|
560 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
561 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
562 buf_T *mi_buf; /* buffer being checked */
|
|
236
|
563 } matchinf_T;
|
|
|
564
|
|
307
|
565 /*
|
|
|
566 * The tables used for recognizing word characters according to spelling.
|
|
|
567 * These are only used for the first 256 characters of 'encoding'.
|
|
|
568 */
|
|
|
569 typedef struct spelltab_S
|
|
|
570 {
|
|
|
571 char_u st_isw[256]; /* flags: is word char */
|
|
|
572 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
573 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
574 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
575 } spelltab_T;
|
|
|
576
|
|
|
577 static spelltab_T spelltab;
|
|
|
578 static int did_set_spelltab;
|
|
|
579
|
|
324
|
580 #define CF_WORD 0x01
|
|
|
581 #define CF_UPPER 0x02
|
|
307
|
582
|
|
|
583 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
584 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
585 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
586 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
587 #ifdef FEAT_MBYTE
|
|
|
588 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
589 #endif
|
|
483
|
590 static int write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
591
|
|
|
592 /*
|
|
344
|
593 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
594 */
|
|
|
595 typedef enum
|
|
|
596 {
|
|
344
|
597 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
598 * ends); if badword ends there is a match, otherwise
|
|
|
599 * try splitting word. */
|
|
375
|
600 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
601 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
602 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
603 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
604 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
605 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
606 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
607 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
608 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
609 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
610 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
611 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
612 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
613 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
614 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
615 STATE_FINAL /* End of this node. */
|
|
|
616 } state_T;
|
|
|
617
|
|
|
618 /*
|
|
346
|
619 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
620 */
|
|
|
621 typedef struct trystate_S
|
|
|
622 {
|
|
330
|
623 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
624 int ts_score; /* score */
|
|
344
|
625 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
626 short ts_curi; /* index in list of child nodes */
|
|
|
627 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
628 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
629 char_u ts_twordlen; /* valid length of tword[] */
|
|
482
|
630 char_u ts_prefixdepth; /* stack depth for end of prefix or
|
|
485
|
631 * PFD_PREFIXTREE or PFD_NOPREFIX */
|
|
|
632 char_u ts_flags; /* TSF_ flags */
|
|
330
|
633 #ifdef FEAT_MBYTE
|
|
|
634 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
635 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
636 char_u ts_isdiff; /* DIFF_ values */
|
|
|
637 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
638 #endif
|
|
483
|
639 char_u ts_prewordlen; /* length of word in "preword[]" */
|
|
|
640 char_u ts_splitoff; /* index in "tword" after last split */
|
|
|
641 char_u ts_complen; /* nr of compound words used */
|
|
485
|
642 char_u ts_compsplit; /* index for "compflags" where word was spit */
|
|
346
|
643 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
644 } trystate_T;
|
|
|
645
|
|
330
|
646 /* values for ts_isdiff */
|
|
|
647 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
648 #define DIFF_YES 1 /* different byte found */
|
|
|
649 #define DIFF_INSERT 2 /* inserting character */
|
|
|
650
|
|
485
|
651 /* values for ts_flags */
|
|
|
652 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */
|
|
|
653 #define TSF_DIDSPLIT 2 /* tried split at this point */
|
|
|
654
|
|
375
|
655 /* special values ts_prefixdepth */
|
|
485
|
656 #define PFD_NOPREFIX 0xff /* not using prefixes */
|
|
482
|
657 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
485
|
658 #define PFD_NOTSPECIAL 0xfd /* first value that's not special */
|
|
375
|
659
|
|
339
|
660 /* mode values for find_word */
|
|
481
|
661 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
662 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
663 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
664 #define FIND_COMPOUND 3 /* find case-folded compound word */
|
|
|
665 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */
|
|
339
|
666
|
|
236
|
667 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
668 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
669 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
670 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
483
|
671 static int can_compound __ARGS((slang_T *slang, char_u *word, char_u *flags));
|
|
455
|
672 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
485
|
673 static void find_prefix __ARGS((matchinf_T *mip, int mode));
|
|
339
|
674 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
675 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
351
|
676 static int no_spell_checking __ARGS((void));
|
|
307
|
677 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
678 static char_u *spell_enc __ARGS((void));
|
|
385
|
679 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
680 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
681 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
682 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
483
|
683 static char_u *read_string __ARGS((FILE *fd, int cnt));
|
|
|
684 static int read_region_section __ARGS((FILE *fd, slang_T *slang, int len));
|
|
|
685 static int read_charflags_section __ARGS((FILE *fd));
|
|
|
686 static int read_prefcond_section __ARGS((FILE *fd, slang_T *lp));
|
|
|
687 static int read_rep_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
688 static int read_sal_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
689 static int read_sofo_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
690 static int read_compound __ARGS((FILE *fd, slang_T *slang, int len));
|
|
|
691 static int init_syl_tab __ARGS((slang_T *slang));
|
|
|
692 static int count_syllables __ARGS((slang_T *slang, char_u *word));
|
|
381
|
693 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
694 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
695 #ifdef FEAT_MBYTE
|
|
|
696 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
697 #endif
|
|
339
|
698 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
699 static void clear_midword __ARGS((buf_T *buf));
|
|
|
700 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
701 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
702 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
703 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
704 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
483
|
705 static void set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
706 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
707 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
475
|
708 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col));
|
|
388
|
709 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
710 #ifdef FEAT_EVAL
|
|
|
711 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
712 #endif
|
|
|
713 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
714 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
715 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
716 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
717 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
718 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
719 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
720 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
721 #ifdef FEAT_MBYTE
|
|
|
722 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
723 #endif
|
|
323
|
724 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
725 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
726 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
727 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
728 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
729 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
730 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
731 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
732 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
733 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
734 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
735 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
736 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
737 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
738 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
739 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
740 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
741 #ifdef FEAT_MBYTE
|
|
375
|
742 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
743 #endif
|
|
344
|
744 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
745 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
746 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
747 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
748
|
|
324
|
749 /*
|
|
|
750 * Use our own character-case definitions, because the current locale may
|
|
|
751 * differ from what the .spl file uses.
|
|
|
752 * These must not be called with negative number!
|
|
|
753 */
|
|
|
754 #ifndef FEAT_MBYTE
|
|
|
755 /* Non-multi-byte implementation. */
|
|
|
756 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
757 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
758 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
759 #else
|
|
447
|
760 # if defined(HAVE_WCHAR_H)
|
|
|
761 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
762 # endif
|
|
324
|
763 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
764 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
765 * the "w" library function for characters above 255 if available. */
|
|
|
766 # ifdef HAVE_TOWLOWER
|
|
|
767 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
768 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
769 # else
|
|
|
770 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
771 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
772 # endif
|
|
|
773
|
|
|
774 # ifdef HAVE_TOWUPPER
|
|
|
775 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
776 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
777 # else
|
|
|
778 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
779 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
780 # endif
|
|
|
781
|
|
|
782 # ifdef HAVE_ISWUPPER
|
|
|
783 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
784 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
785 # else
|
|
|
786 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
787 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
788 # endif
|
|
|
789 #endif
|
|
|
790
|
|
307
|
791
|
|
|
792 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
793 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
482
|
794 static char *e_afftrailing = N_("Trailing text in %s line %d: %s");
|
|
469
|
795 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
796
|
|
|
797 /*
|
|
|
798 * Main spell-checking function.
|
|
300
|
799 * "ptr" points to a character that could be the start of a word.
|
|
236
|
800 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
801 * or when it's OK it remains unchanged.
|
|
|
802 * This must only be called when 'spelllang' is not empty.
|
|
323
|
803 *
|
|
385
|
804 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
805 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
806 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
807 * worry.
|
|
323
|
808 *
|
|
236
|
809 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
810 * caller can skip over the word.
|
|
|
811 */
|
|
|
812 int
|
|
385
|
813 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
814 win_T *wp; /* current window */
|
|
|
815 char_u *ptr;
|
|
|
816 int *attrp;
|
|
385
|
817 int *capcol; /* column to check for Capital */
|
|
236
|
818 {
|
|
|
819 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
820 be passed to functions quickly. */
|
|
344
|
821 int nrlen = 0; /* found a number first */
|
|
385
|
822 int c;
|
|
483
|
823 int wrongcaplen = 0;
|
|
236
|
824
|
|
307
|
825 /* A word never starts at a space or a control character. Return quickly
|
|
|
826 * then, skipping over the character. */
|
|
|
827 if (*ptr <= ' ')
|
|
|
828 return 1;
|
|
483
|
829 vim_memset(&mi, 0, sizeof(matchinf_T));
|
|
236
|
830
|
|
344
|
831 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
832 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
833 * "3GPP". */
|
|
300
|
834 if (*ptr >= '0' && *ptr <= '9')
|
|
|
835 {
|
|
316
|
836 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
837 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
838 else
|
|
|
839 {
|
|
344
|
840 mi.mi_end = skipdigits(ptr);
|
|
|
841 nrlen = mi.mi_end - ptr;
|
|
300
|
842 }
|
|
376
|
843 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
844 return (int)(mi.mi_end - ptr);
|
|
346
|
845
|
|
|
846 /* Try including the digits in the word. */
|
|
|
847 mi.mi_fend = ptr + nrlen;
|
|
|
848 }
|
|
|
849 else
|
|
|
850 mi.mi_fend = ptr;
|
|
|
851
|
|
|
852 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
853 mi.mi_word = ptr;
|
|
376
|
854 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
855 {
|
|
|
856 do
|
|
|
857 {
|
|
307
|
858 mb_ptr_adv(mi.mi_fend);
|
|
376
|
859 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
860
|
|
|
861 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
862 {
|
|
|
863 /* Check word starting with capital letter. */
|
|
455
|
864 c = PTR2CHAR(ptr);
|
|
385
|
865 if (!SPELL_ISUPPER(c))
|
|
483
|
866 wrongcaplen = (int)(mi.mi_fend - ptr);
|
|
385
|
867 }
|
|
|
868 }
|
|
|
869 if (capcol != NULL)
|
|
|
870 *capcol = -1;
|
|
344
|
871
|
|
|
872 /* We always use the characters up to the next non-word character,
|
|
|
873 * also for bad words. */
|
|
|
874 mi.mi_end = mi.mi_fend;
|
|
|
875
|
|
|
876 /* Check caps type later. */
|
|
376
|
877 mi.mi_buf = wp->w_buffer;
|
|
344
|
878
|
|
483
|
879 /* case-fold the word with one non-word character, so that we can check
|
|
|
880 * for the word end. */
|
|
344
|
881 if (*mi.mi_fend != NUL)
|
|
|
882 mb_ptr_adv(mi.mi_fend);
|
|
|
883
|
|
|
884 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
885 MAXWLEN + 1);
|
|
|
886 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
887
|
|
|
888 /* The word is bad unless we recognize it. */
|
|
|
889 mi.mi_result = SP_BAD;
|
|
|
890
|
|
|
891 /*
|
|
|
892 * Loop over the languages specified in 'spelllang'.
|
|
|
893 * We check them all, because a matching word may be longer than an
|
|
|
894 * already found matching word.
|
|
|
895 */
|
|
|
896 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
897 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
898 {
|
|
|
899 /* Check for a matching word in case-folded words. */
|
|
|
900 find_word(&mi, FIND_FOLDWORD);
|
|
|
901
|
|
|
902 /* Check for a matching word in keep-case words. */
|
|
|
903 find_word(&mi, FIND_KEEPWORD);
|
|
|
904
|
|
|
905 /* Check for matching prefixes. */
|
|
485
|
906 find_prefix(&mi, FIND_FOLDWORD);
|
|
344
|
907 }
|
|
|
908
|
|
|
909 if (mi.mi_result != SP_OK)
|
|
|
910 {
|
|
346
|
911 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
912 * rare and local words, e.g., "3GPP". */
|
|
344
|
913 if (nrlen > 0)
|
|
346
|
914 {
|
|
|
915 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
916 return nrlen;
|
|
|
917 }
|
|
344
|
918
|
|
|
919 /* When we are at a non-word character there is no error, just
|
|
|
920 * skip over the character (try looking for a word after it). */
|
|
390
|
921 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
922 {
|
|
385
|
923 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
924 {
|
|
|
925 regmatch_T regmatch;
|
|
|
926
|
|
|
927 /* Check for end of sentence. */
|
|
|
928 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
929 regmatch.rm_ic = FALSE;
|
|
|
930 if (vim_regexec(®match, ptr, 0))
|
|
|
931 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
932 }
|
|
|
933
|
|
344
|
934 #ifdef FEAT_MBYTE
|
|
|
935 if (has_mbyte)
|
|
474
|
936 return (*mb_ptr2len)(ptr);
|
|
344
|
937 #endif
|
|
|
938 return 1;
|
|
300
|
939 }
|
|
483
|
940 else if (mi.mi_end == ptr)
|
|
|
941 /* Always include at least one character. Required for when there
|
|
|
942 * is a mixup in "midword". */
|
|
|
943 mb_ptr_adv(mi.mi_end);
|
|
243
|
944
|
|
344
|
945 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
946 *attrp = highlight_attr[HLF_SPB];
|
|
|
947 else if (mi.mi_result == SP_RARE)
|
|
|
948 *attrp = highlight_attr[HLF_SPR];
|
|
|
949 else
|
|
|
950 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
951 }
|
|
|
952
|
|
483
|
953 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE))
|
|
|
954 {
|
|
|
955 /* Report SpellCap only when the word isn't badly spelled. */
|
|
|
956 *attrp = highlight_attr[HLF_SPC];
|
|
|
957 return wrongcaplen;
|
|
|
958 }
|
|
|
959
|
|
300
|
960 return (int)(mi.mi_end - ptr);
|
|
236
|
961 }
|
|
|
962
|
|
|
963 /*
|
|
300
|
964 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
965 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
966 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
967 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
968 * tree.
|
|
300
|
969 *
|
|
|
970 * For a match mip->mi_result is updated.
|
|
243
|
971 */
|
|
|
972 static void
|
|
339
|
973 find_word(mip, mode)
|
|
243
|
974 matchinf_T *mip;
|
|
339
|
975 int mode;
|
|
243
|
976 {
|
|
324
|
977 idx_T arridx = 0;
|
|
300
|
978 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
979 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
980 int endidxcnt = 0;
|
|
|
981 int len;
|
|
|
982 int wlen = 0;
|
|
|
983 int flen;
|
|
|
984 int c;
|
|
|
985 char_u *ptr;
|
|
324
|
986 idx_T lo, hi, m;
|
|
243
|
987 #ifdef FEAT_MBYTE
|
|
300
|
988 char_u *s;
|
|
490
|
989 #endif
|
|
339
|
990 char_u *p;
|
|
307
|
991 int res = SP_BAD;
|
|
300
|
992 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
993 unsigned flags;
|
|
|
994 char_u *byts;
|
|
324
|
995 idx_T *idxs;
|
|
481
|
996 int word_ends;
|
|
485
|
997 int prefix_found;
|
|
481
|
998
|
|
|
999 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND)
|
|
236
|
1000 {
|
|
300
|
1001 /* Check for word with matching case in keep-case tree. */
|
|
|
1002 ptr = mip->mi_word;
|
|
|
1003 flen = 9999; /* no case folding, always enough bytes */
|
|
|
1004 byts = slang->sl_kbyts;
|
|
|
1005 idxs = slang->sl_kidxs;
|
|
481
|
1006
|
|
|
1007 if (mode == FIND_KEEPCOMPOUND)
|
|
|
1008 /* Skip over the previously found word(s). */
|
|
|
1009 wlen += mip->mi_compoff;
|
|
236
|
1010 }
|
|
|
1011 else
|
|
|
1012 {
|
|
300
|
1013 /* Check for case-folded in case-folded tree. */
|
|
|
1014 ptr = mip->mi_fword;
|
|
|
1015 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1016 byts = slang->sl_fbyts;
|
|
|
1017 idxs = slang->sl_fidxs;
|
|
339
|
1018
|
|
|
1019 if (mode == FIND_PREFIX)
|
|
|
1020 {
|
|
|
1021 /* Skip over the prefix. */
|
|
|
1022 wlen = mip->mi_prefixlen;
|
|
|
1023 flen -= mip->mi_prefixlen;
|
|
|
1024 }
|
|
481
|
1025 else if (mode == FIND_COMPOUND)
|
|
|
1026 {
|
|
|
1027 /* Skip over the previously found word(s). */
|
|
|
1028 wlen = mip->mi_compoff;
|
|
|
1029 flen -= mip->mi_compoff;
|
|
|
1030 }
|
|
|
1031
|
|
243
|
1032 }
|
|
|
1033
|
|
300
|
1034 if (byts == NULL)
|
|
|
1035 return; /* array is empty */
|
|
236
|
1036
|
|
|
1037 /*
|
|
307
|
1038 * Repeat advancing in the tree until:
|
|
|
1039 * - there is a byte that doesn't match,
|
|
|
1040 * - we reach the end of the tree,
|
|
|
1041 * - or we reach the end of the line.
|
|
236
|
1042 */
|
|
300
|
1043 for (;;)
|
|
236
|
1044 {
|
|
346
|
1045 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
1046 flen = fold_more(mip);
|
|
300
|
1047
|
|
|
1048 len = byts[arridx++];
|
|
|
1049
|
|
|
1050 /* If the first possible byte is a zero the word could end here.
|
|
|
1051 * Remember this index, we first check for the longest word. */
|
|
|
1052 if (byts[arridx] == 0)
|
|
|
1053 {
|
|
307
|
1054 if (endidxcnt == MAXWLEN)
|
|
|
1055 {
|
|
|
1056 /* Must be a corrupted spell file. */
|
|
|
1057 EMSG(_(e_format));
|
|
|
1058 return;
|
|
|
1059 }
|
|
300
|
1060 endlen[endidxcnt] = wlen;
|
|
|
1061 endidx[endidxcnt++] = arridx++;
|
|
|
1062 --len;
|
|
|
1063
|
|
|
1064 /* Skip over the zeros, there can be several flag/region
|
|
|
1065 * combinations. */
|
|
|
1066 while (len > 0 && byts[arridx] == 0)
|
|
|
1067 {
|
|
|
1068 ++arridx;
|
|
|
1069 --len;
|
|
|
1070 }
|
|
|
1071 if (len == 0)
|
|
|
1072 break; /* no children, word must end here */
|
|
|
1073 }
|
|
|
1074
|
|
|
1075 /* Stop looking at end of the line. */
|
|
|
1076 if (ptr[wlen] == NUL)
|
|
|
1077 break;
|
|
|
1078
|
|
|
1079 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1080 c = ptr[wlen];
|
|
346
|
1081 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1082 c = ' ';
|
|
300
|
1083 lo = arridx;
|
|
|
1084 hi = arridx + len - 1;
|
|
|
1085 while (lo < hi)
|
|
|
1086 {
|
|
|
1087 m = (lo + hi) / 2;
|
|
|
1088 if (byts[m] > c)
|
|
|
1089 hi = m - 1;
|
|
|
1090 else if (byts[m] < c)
|
|
|
1091 lo = m + 1;
|
|
|
1092 else
|
|
|
1093 {
|
|
|
1094 lo = hi = m;
|
|
|
1095 break;
|
|
236
|
1096 }
|
|
|
1097 }
|
|
300
|
1098
|
|
|
1099 /* Stop if there is no matching byte. */
|
|
|
1100 if (hi < lo || byts[lo] != c)
|
|
|
1101 break;
|
|
|
1102
|
|
|
1103 /* Continue at the child (if there is one). */
|
|
|
1104 arridx = idxs[lo];
|
|
|
1105 ++wlen;
|
|
|
1106 --flen;
|
|
346
|
1107
|
|
|
1108 /* One space in the good word may stand for several spaces in the
|
|
|
1109 * checked word. */
|
|
|
1110 if (c == ' ')
|
|
|
1111 {
|
|
|
1112 for (;;)
|
|
|
1113 {
|
|
|
1114 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1115 flen = fold_more(mip);
|
|
|
1116 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1117 break;
|
|
|
1118 ++wlen;
|
|
|
1119 --flen;
|
|
|
1120 }
|
|
|
1121 }
|
|
236
|
1122 }
|
|
|
1123
|
|
300
|
1124 /*
|
|
|
1125 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1126 * first.
|
|
|
1127 */
|
|
|
1128 while (endidxcnt > 0)
|
|
|
1129 {
|
|
|
1130 --endidxcnt;
|
|
|
1131 arridx = endidx[endidxcnt];
|
|
|
1132 wlen = endlen[endidxcnt];
|
|
236
|
1133
|
|
300
|
1134 #ifdef FEAT_MBYTE
|
|
|
1135 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1136 continue; /* not at first byte of character */
|
|
|
1137 #endif
|
|
376
|
1138 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
481
|
1139 {
|
|
483
|
1140 if (slang->sl_compprog == NULL)
|
|
481
|
1141 continue; /* next char is a word character */
|
|
|
1142 word_ends = FALSE;
|
|
|
1143 }
|
|
|
1144 else
|
|
|
1145 word_ends = TRUE;
|
|
485
|
1146 /* The prefix flag is before compound flags. Once a valid prefix flag
|
|
|
1147 * has been found we try compound flags. */
|
|
|
1148 prefix_found = FALSE;
|
|
300
|
1149
|
|
|
1150 #ifdef FEAT_MBYTE
|
|
339
|
1151 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1152 {
|
|
|
1153 /* Compute byte length in original word, length may change
|
|
339
|
1154 * when folding case. This can be slow, take a shortcut when the
|
|
|
1155 * case-folded word is equal to the keep-case word. */
|
|
300
|
1156 p = mip->mi_word;
|
|
339
|
1157 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1158 {
|
|
|
1159 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1160 mb_ptr_adv(p);
|
|
|
1161 wlen = p - mip->mi_word;
|
|
|
1162 }
|
|
300
|
1163 }
|
|
|
1164 #endif
|
|
236
|
1165
|
|
339
|
1166 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1167 * prefix ID.
|
|
|
1168 * Repeat this if there are more flags/region alternatives until there
|
|
|
1169 * is a match. */
|
|
|
1170 res = SP_BAD;
|
|
|
1171 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1172 --len, ++arridx)
|
|
300
|
1173 {
|
|
|
1174 flags = idxs[arridx];
|
|
324
|
1175
|
|
339
|
1176 /* For the fold-case tree check that the case of the checked word
|
|
|
1177 * matches with what the word in the tree requires.
|
|
|
1178 * For keep-case tree the case is always right. For prefixes we
|
|
|
1179 * don't bother to check. */
|
|
|
1180 if (mode == FIND_FOLDWORD)
|
|
300
|
1181 {
|
|
|
1182 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1183 {
|
|
323
|
1184 /* mi_capflags was set for a different word length, need
|
|
|
1185 * to do it again. */
|
|
300
|
1186 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1187 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1188 }
|
|
|
1189
|
|
346
|
1190 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1191 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1192 continue;
|
|
300
|
1193 }
|
|
236
|
1194
|
|
339
|
1195 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1196 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1197 * mip->mi_prefarridx that find_prefix() filled. */
|
|
485
|
1198 else if (mode == FIND_PREFIX && !prefix_found)
|
|
481
|
1199 {
|
|
366
|
1200 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1201 flags,
|
|
455
|
1202 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1203 FALSE);
|
|
366
|
1204 if (c == 0)
|
|
339
|
1205 continue;
|
|
366
|
1206
|
|
|
1207 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1208 if (c & WF_RAREPFX)
|
|
|
1209 flags |= WF_RARE;
|
|
485
|
1210 prefix_found = TRUE;
|
|
339
|
1211 }
|
|
|
1212
|
|
481
|
1213 if (mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND
|
|
|
1214 || !word_ends)
|
|
|
1215 {
|
|
483
|
1216 /* If there is no flag or the word is shorter than
|
|
|
1217 * COMPOUNDMIN reject it quickly.
|
|
|
1218 * Makes you wonder why someone puts a compound flag on a word
|
|
481
|
1219 * that's too short... Myspell compatibility requires this
|
|
|
1220 * anyway. */
|
|
490
|
1221 if (((unsigned)flags >> 24) == 0
|
|
|
1222 || wlen - mip->mi_compoff < slang->sl_compminlen)
|
|
483
|
1223 continue;
|
|
|
1224
|
|
490
|
1225 /* Limit the number of compound words to COMPOUNDMAX if no
|
|
|
1226 * maximum for syllables is specified. */
|
|
|
1227 if (!word_ends && mip->mi_complen + 2 > slang->sl_compmax
|
|
|
1228 && slang->sl_compsylmax == MAXWLEN)
|
|
483
|
1229 continue;
|
|
|
1230
|
|
485
|
1231 /* Quickly check if compounding is possible with this flag. */
|
|
|
1232 if (vim_strchr(mip->mi_complen == 0
|
|
|
1233 ? slang->sl_compstartflags
|
|
|
1234 : slang->sl_compallflags,
|
|
483
|
1235 ((unsigned)flags >> 24)) == NULL)
|
|
481
|
1236 continue;
|
|
|
1237
|
|
490
|
1238 if (mode == FIND_COMPOUND)
|
|
|
1239 {
|
|
|
1240 int capflags;
|
|
|
1241
|
|
|
1242 /* Need to check the caps type of the appended compound
|
|
|
1243 * word. */
|
|
|
1244 #ifdef FEAT_MBYTE
|
|
|
1245 if (has_mbyte && STRNCMP(ptr, mip->mi_word,
|
|
|
1246 mip->mi_compoff) != 0)
|
|
|
1247 {
|
|
|
1248 /* case folding may have changed the length */
|
|
|
1249 p = mip->mi_word;
|
|
|
1250 for (s = ptr; s < ptr + mip->mi_compoff; mb_ptr_adv(s))
|
|
|
1251 mb_ptr_adv(p);
|
|
|
1252 }
|
|
|
1253 else
|
|
|
1254 #endif
|
|
|
1255 p = mip->mi_word + mip->mi_compoff;
|
|
|
1256 capflags = captype(p, mip->mi_word + wlen);
|
|
|
1257 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP
|
|
|
1258 && (flags & WF_FIXCAP) != 0))
|
|
|
1259 continue;
|
|
|
1260
|
|
|
1261 if (capflags != WF_ALLCAP)
|
|
|
1262 {
|
|
|
1263 /* When the character before the word is a word
|
|
|
1264 * character we do not accept a Onecap word. We do
|
|
|
1265 * accept a no-caps word, even when the dictionary
|
|
|
1266 * word specifies ONECAP. */
|
|
|
1267 mb_ptr_back(mip->mi_word, p);
|
|
|
1268 if (spell_iswordp_nmw(p)
|
|
|
1269 ? capflags == WF_ONECAP
|
|
|
1270 : (flags & WF_ONECAP) != 0
|
|
|
1271 && capflags != WF_ONECAP)
|
|
|
1272 continue;
|
|
|
1273 }
|
|
|
1274 }
|
|
|
1275
|
|
483
|
1276 /* If the word ends the sequence of compound flags of the
|
|
|
1277 * words must match with one of the COMPOUNDFLAGS items and
|
|
|
1278 * the number of syllables must not be too large. */
|
|
|
1279 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24);
|
|
|
1280 mip->mi_compflags[mip->mi_complen + 1] = NUL;
|
|
|
1281 if (word_ends)
|
|
|
1282 {
|
|
|
1283 char_u fword[MAXWLEN];
|
|
|
1284
|
|
|
1285 if (slang->sl_compsylmax < MAXWLEN)
|
|
|
1286 {
|
|
|
1287 /* "fword" is only needed for checking syllables. */
|
|
|
1288 if (ptr == mip->mi_word)
|
|
|
1289 (void)spell_casefold(ptr, wlen, fword, MAXWLEN);
|
|
|
1290 else
|
|
|
1291 vim_strncpy(fword, ptr, endlen[endidxcnt]);
|
|
|
1292 }
|
|
|
1293 if (!can_compound(slang, fword, mip->mi_compflags))
|
|
|
1294 continue;
|
|
|
1295 }
|
|
481
|
1296 }
|
|
|
1297
|
|
|
1298 if (!word_ends)
|
|
|
1299 {
|
|
|
1300 /* Check that a valid word follows. If there is one, it will
|
|
|
1301 * set "mi_result", thus we are always finished here.
|
|
|
1302 * Recursive! */
|
|
|
1303
|
|
|
1304 /* Find following word in case-folded tree. */
|
|
|
1305 mip->mi_compoff = endlen[endidxcnt];
|
|
|
1306 #ifdef FEAT_MBYTE
|
|
|
1307 if (has_mbyte && mode == FIND_KEEPWORD)
|
|
|
1308 {
|
|
|
1309 /* Compute byte length in case-folded word from "wlen":
|
|
|
1310 * byte length in keep-case word. Length may change when
|
|
|
1311 * folding case. This can be slow, take a shortcut when
|
|
|
1312 * the case-folded word is equal to the keep-case word. */
|
|
|
1313 p = mip->mi_fword;
|
|
|
1314 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1315 {
|
|
|
1316 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1317 mb_ptr_adv(p);
|
|
|
1318 mip->mi_compoff = p - mip->mi_fword;
|
|
|
1319 }
|
|
|
1320 }
|
|
|
1321 #endif
|
|
485
|
1322 c = mip->mi_compoff;
|
|
483
|
1323 ++mip->mi_complen;
|
|
481
|
1324 find_word(mip, FIND_COMPOUND);
|
|
|
1325
|
|
|
1326 /* Find following word in keep-case tree. */
|
|
|
1327 mip->mi_compoff = wlen;
|
|
|
1328 find_word(mip, FIND_KEEPCOMPOUND);
|
|
485
|
1329
|
|
|
1330 /* Check for following word with prefix. */
|
|
|
1331 mip->mi_compoff = c;
|
|
|
1332 find_prefix(mip, FIND_COMPOUND);
|
|
483
|
1333 --mip->mi_complen;
|
|
485
|
1334
|
|
481
|
1335 if (mip->mi_result == SP_OK)
|
|
|
1336 break;
|
|
|
1337 continue;
|
|
|
1338 }
|
|
|
1339
|
|
339
|
1340 if (flags & WF_BANNED)
|
|
|
1341 res = SP_BANNED;
|
|
|
1342 else if (flags & WF_REGION)
|
|
|
1343 {
|
|
|
1344 /* Check region. */
|
|
390
|
1345 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1346 res = SP_OK;
|
|
339
|
1347 else
|
|
|
1348 res = SP_LOCAL;
|
|
300
|
1349 }
|
|
339
|
1350 else if (flags & WF_RARE)
|
|
|
1351 res = SP_RARE;
|
|
307
|
1352 else
|
|
339
|
1353 res = SP_OK;
|
|
|
1354
|
|
|
1355 /* Always use the longest match and the best result. */
|
|
|
1356 if (mip->mi_result > res)
|
|
|
1357 {
|
|
|
1358 mip->mi_result = res;
|
|
|
1359 mip->mi_end = mip->mi_word + wlen;
|
|
|
1360 }
|
|
351
|
1361 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1362 mip->mi_end = mip->mi_word + wlen;
|
|
|
1363
|
|
|
1364 if (res == SP_OK)
|
|
|
1365 break;
|
|
300
|
1366 }
|
|
|
1367
|
|
307
|
1368 if (res == SP_OK)
|
|
300
|
1369 break;
|
|
|
1370 }
|
|
236
|
1371 }
|
|
|
1372
|
|
323
|
1373 /*
|
|
483
|
1374 * Return TRUE if "flags" is a valid sequence of compound flags and
|
|
|
1375 * "word[len]" does not have too many syllables.
|
|
482
|
1376 */
|
|
|
1377 static int
|
|
483
|
1378 can_compound(slang, word, flags)
|
|
482
|
1379 slang_T *slang;
|
|
483
|
1380 char_u *word;
|
|
|
1381 char_u *flags;
|
|
|
1382 {
|
|
|
1383 regmatch_T regmatch;
|
|
|
1384
|
|
|
1385 if (slang->sl_compprog == NULL)
|
|
|
1386 return FALSE;
|
|
|
1387 regmatch.regprog = slang->sl_compprog;
|
|
|
1388 regmatch.rm_ic = FALSE;
|
|
|
1389 if (!vim_regexec(®match, flags, 0))
|
|
|
1390 return FALSE;
|
|
|
1391
|
|
490
|
1392 /* Count the number of syllables. This may be slow, do it last. If there
|
|
|
1393 * are too many syllables AND the number of compound words is above
|
|
|
1394 * COMPOUNDMAX then compounding is not allowed. */
|
|
483
|
1395 if (slang->sl_compsylmax < MAXWLEN
|
|
|
1396 && count_syllables(slang, word) > slang->sl_compsylmax)
|
|
490
|
1397 return STRLEN(flags) < slang->sl_compmax;
|
|
483
|
1398 return TRUE;
|
|
482
|
1399 }
|
|
|
1400
|
|
|
1401 /*
|
|
390
|
1402 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1403 * ID in "flags" for the word "word".
|
|
366
|
1404 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1405 */
|
|
|
1406 static int
|
|
455
|
1407 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1408 int totprefcnt; /* nr of prefix IDs */
|
|
|
1409 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1410 int flags;
|
|
351
|
1411 char_u *word;
|
|
|
1412 slang_T *slang;
|
|
455
|
1413 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1414 {
|
|
|
1415 int prefcnt;
|
|
|
1416 int pidx;
|
|
|
1417 regprog_T *rp;
|
|
|
1418 regmatch_T regmatch;
|
|
390
|
1419 int prefid;
|
|
|
1420
|
|
|
1421 prefid = (unsigned)flags >> 24;
|
|
351
|
1422 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1423 {
|
|
|
1424 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1425
|
|
|
1426 /* Check the prefix ID. */
|
|
|
1427 if (prefid != (pidx & 0xff))
|
|
|
1428 continue;
|
|
|
1429
|
|
390
|
1430 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1431 * suffix. */
|
|
|
1432 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1433 continue;
|
|
|
1434
|
|
351
|
1435 /* Check the condition, if there is one. The condition index is
|
|
366
|
1436 * stored in the two bytes above the prefix ID byte. */
|
|
|
1437 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1438 if (rp != NULL)
|
|
|
1439 {
|
|
|
1440 regmatch.regprog = rp;
|
|
|
1441 regmatch.rm_ic = FALSE;
|
|
|
1442 if (!vim_regexec(®match, word, 0))
|
|
|
1443 continue;
|
|
|
1444 }
|
|
455
|
1445 else if (cond_req)
|
|
|
1446 continue;
|
|
|
1447
|
|
|
1448 /* It's a match! Return the WF_ flags. */
|
|
366
|
1449 return pidx;
|
|
|
1450 }
|
|
|
1451 return 0;
|
|
351
|
1452 }
|
|
|
1453
|
|
|
1454 /*
|
|
339
|
1455 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1456 * If it does, then check the following word.
|
|
|
1457 *
|
|
485
|
1458 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a
|
|
|
1459 * prefix in a compound word.
|
|
|
1460 *
|
|
339
|
1461 * For a match mip->mi_result is updated.
|
|
|
1462 */
|
|
|
1463 static void
|
|
485
|
1464 find_prefix(mip, mode)
|
|
339
|
1465 matchinf_T *mip;
|
|
485
|
1466 int mode;
|
|
339
|
1467 {
|
|
|
1468 idx_T arridx = 0;
|
|
|
1469 int len;
|
|
|
1470 int wlen = 0;
|
|
|
1471 int flen;
|
|
|
1472 int c;
|
|
|
1473 char_u *ptr;
|
|
|
1474 idx_T lo, hi, m;
|
|
|
1475 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1476 char_u *byts;
|
|
|
1477 idx_T *idxs;
|
|
|
1478
|
|
375
|
1479 byts = slang->sl_pbyts;
|
|
|
1480 if (byts == NULL)
|
|
|
1481 return; /* array is empty */
|
|
|
1482
|
|
339
|
1483 /* We use the case-folded word here, since prefixes are always
|
|
|
1484 * case-folded. */
|
|
|
1485 ptr = mip->mi_fword;
|
|
|
1486 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
485
|
1487 if (mode == FIND_COMPOUND)
|
|
|
1488 {
|
|
|
1489 /* Skip over the previously found word(s). */
|
|
|
1490 ptr += mip->mi_compoff;
|
|
|
1491 flen -= mip->mi_compoff;
|
|
|
1492 }
|
|
339
|
1493 idxs = slang->sl_pidxs;
|
|
|
1494
|
|
|
1495 /*
|
|
|
1496 * Repeat advancing in the tree until:
|
|
|
1497 * - there is a byte that doesn't match,
|
|
|
1498 * - we reach the end of the tree,
|
|
|
1499 * - or we reach the end of the line.
|
|
|
1500 */
|
|
|
1501 for (;;)
|
|
|
1502 {
|
|
|
1503 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1504 flen = fold_more(mip);
|
|
|
1505
|
|
|
1506 len = byts[arridx++];
|
|
|
1507
|
|
|
1508 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1509 * Check if the following word matches and supports the prefix. */
|
|
|
1510 if (byts[arridx] == 0)
|
|
|
1511 {
|
|
|
1512 /* There can be several prefixes with different conditions. We
|
|
|
1513 * try them all, since we don't know which one will give the
|
|
|
1514 * longest match. The word is the same each time, pass the list
|
|
|
1515 * of possible prefixes to find_word(). */
|
|
|
1516 mip->mi_prefarridx = arridx;
|
|
|
1517 mip->mi_prefcnt = len;
|
|
|
1518 while (len > 0 && byts[arridx] == 0)
|
|
|
1519 {
|
|
|
1520 ++arridx;
|
|
|
1521 --len;
|
|
|
1522 }
|
|
|
1523 mip->mi_prefcnt -= len;
|
|
|
1524
|
|
|
1525 /* Find the word that comes after the prefix. */
|
|
|
1526 mip->mi_prefixlen = wlen;
|
|
485
|
1527 if (mode == FIND_COMPOUND)
|
|
|
1528 /* Skip over the previously found word(s). */
|
|
|
1529 mip->mi_prefixlen += mip->mi_compoff;
|
|
|
1530
|
|
455
|
1531 #ifdef FEAT_MBYTE
|
|
|
1532 if (has_mbyte)
|
|
|
1533 {
|
|
|
1534 /* Case-folded length may differ from original length. */
|
|
485
|
1535 mip->mi_cprefixlen = nofold_len(mip->mi_fword,
|
|
|
1536 mip->mi_prefixlen, mip->mi_word);
|
|
455
|
1537 }
|
|
|
1538 else
|
|
485
|
1539 mip->mi_cprefixlen = mip->mi_prefixlen;
|
|
455
|
1540 #endif
|
|
339
|
1541 find_word(mip, FIND_PREFIX);
|
|
|
1542
|
|
|
1543
|
|
|
1544 if (len == 0)
|
|
|
1545 break; /* no children, word must end here */
|
|
|
1546 }
|
|
|
1547
|
|
|
1548 /* Stop looking at end of the line. */
|
|
|
1549 if (ptr[wlen] == NUL)
|
|
|
1550 break;
|
|
|
1551
|
|
|
1552 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1553 c = ptr[wlen];
|
|
|
1554 lo = arridx;
|
|
|
1555 hi = arridx + len - 1;
|
|
|
1556 while (lo < hi)
|
|
|
1557 {
|
|
|
1558 m = (lo + hi) / 2;
|
|
|
1559 if (byts[m] > c)
|
|
|
1560 hi = m - 1;
|
|
|
1561 else if (byts[m] < c)
|
|
|
1562 lo = m + 1;
|
|
|
1563 else
|
|
|
1564 {
|
|
|
1565 lo = hi = m;
|
|
|
1566 break;
|
|
|
1567 }
|
|
|
1568 }
|
|
|
1569
|
|
|
1570 /* Stop if there is no matching byte. */
|
|
|
1571 if (hi < lo || byts[lo] != c)
|
|
|
1572 break;
|
|
|
1573
|
|
|
1574 /* Continue at the child (if there is one). */
|
|
|
1575 arridx = idxs[lo];
|
|
|
1576 ++wlen;
|
|
|
1577 --flen;
|
|
|
1578 }
|
|
|
1579 }
|
|
|
1580
|
|
|
1581 /*
|
|
|
1582 * Need to fold at least one more character. Do until next non-word character
|
|
|
1583 * for efficiency.
|
|
|
1584 * Return the length of the folded chars in bytes.
|
|
|
1585 */
|
|
|
1586 static int
|
|
|
1587 fold_more(mip)
|
|
|
1588 matchinf_T *mip;
|
|
|
1589 {
|
|
|
1590 int flen;
|
|
|
1591 char_u *p;
|
|
|
1592
|
|
|
1593 p = mip->mi_fend;
|
|
|
1594 do
|
|
|
1595 {
|
|
|
1596 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1597 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1598
|
|
|
1599 /* Include the non-word character so that we can check for the
|
|
|
1600 * word end. */
|
|
|
1601 if (*mip->mi_fend != NUL)
|
|
|
1602 mb_ptr_adv(mip->mi_fend);
|
|
|
1603
|
|
|
1604 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1605 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1606 MAXWLEN - mip->mi_fwordlen);
|
|
|
1607 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1608 mip->mi_fwordlen += flen;
|
|
|
1609 return flen;
|
|
|
1610 }
|
|
|
1611
|
|
|
1612 /*
|
|
323
|
1613 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1614 * case.
|
|
|
1615 */
|
|
|
1616 static int
|
|
389
|
1617 spell_valid_case(wordflags, treeflags)
|
|
|
1618 int wordflags; /* flags for the checked word. */
|
|
323
|
1619 int treeflags; /* flags for the word in the spell tree */
|
|
|
1620 {
|
|
389
|
1621 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1622 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1623 && ((treeflags & WF_ONECAP) == 0
|
|
|
1624 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1625 }
|
|
|
1626
|
|
351
|
1627 /*
|
|
|
1628 * Return TRUE if spell checking is not enabled.
|
|
|
1629 */
|
|
|
1630 static int
|
|
|
1631 no_spell_checking()
|
|
|
1632 {
|
|
|
1633 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1634 {
|
|
|
1635 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1636 return TRUE;
|
|
|
1637 }
|
|
|
1638 return FALSE;
|
|
|
1639 }
|
|
300
|
1640
|
|
236
|
1641 /*
|
|
|
1642 * Move to next spell error.
|
|
323
|
1643 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
483
|
1644 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move
|
|
|
1645 * to after badly spelled word before the cursor.
|
|
236
|
1646 * Return OK if found, FAIL otherwise.
|
|
|
1647 */
|
|
|
1648 int
|
|
323
|
1649 spell_move_to(dir, allwords, curline)
|
|
236
|
1650 int dir; /* FORWARD or BACKWARD */
|
|
|
1651 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1652 int curline;
|
|
236
|
1653 {
|
|
249
|
1654 linenr_T lnum;
|
|
|
1655 pos_T found_pos;
|
|
236
|
1656 char_u *line;
|
|
|
1657 char_u *p;
|
|
346
|
1658 char_u *endp;
|
|
|
1659 int attr;
|
|
236
|
1660 int len;
|
|
249
|
1661 int has_syntax = syntax_present(curbuf);
|
|
|
1662 int col;
|
|
|
1663 int can_spell;
|
|
346
|
1664 char_u *buf = NULL;
|
|
|
1665 int buflen = 0;
|
|
|
1666 int skip = 0;
|
|
385
|
1667 int capcol = -1;
|
|
236
|
1668
|
|
351
|
1669 if (no_spell_checking())
|
|
236
|
1670 return FAIL;
|
|
|
1671
|
|
249
|
1672 /*
|
|
|
1673 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1674 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1675 *
|
|
|
1676 * When searching backwards, we continue in the line to find the last
|
|
|
1677 * bad word (in the cursor line: before the cursor).
|
|
346
|
1678 *
|
|
|
1679 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1680 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1681 * though...
|
|
249
|
1682 */
|
|
|
1683 lnum = curwin->w_cursor.lnum;
|
|
|
1684 found_pos.lnum = 0;
|
|
236
|
1685
|
|
|
1686 while (!got_int)
|
|
|
1687 {
|
|
249
|
1688 line = ml_get(lnum);
|
|
346
|
1689
|
|
|
1690 len = STRLEN(line);
|
|
|
1691 if (buflen < len + MAXWLEN + 2)
|
|
|
1692 {
|
|
|
1693 vim_free(buf);
|
|
|
1694 buflen = len + MAXWLEN + 2;
|
|
|
1695 buf = alloc(buflen);
|
|
|
1696 if (buf == NULL)
|
|
|
1697 break;
|
|
|
1698 }
|
|
|
1699
|
|
385
|
1700 /* In first line check first word for Capital. */
|
|
|
1701 if (lnum == 1)
|
|
|
1702 capcol = 0;
|
|
|
1703
|
|
|
1704 /* For checking first word with a capital skip white space. */
|
|
|
1705 if (capcol == 0)
|
|
|
1706 capcol = skipwhite(line) - line;
|
|
|
1707
|
|
346
|
1708 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1709 * possible. */
|
|
|
1710 STRCPY(buf, line);
|
|
|
1711 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1712 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1713
|
|
|
1714 p = buf + skip;
|
|
|
1715 endp = buf + len;
|
|
|
1716 while (p < endp)
|
|
236
|
1717 {
|
|
300
|
1718 /* When searching backward don't search after the cursor. */
|
|
|
1719 if (dir == BACKWARD
|
|
|
1720 && lnum == curwin->w_cursor.lnum
|
|
346
|
1721 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1722 break;
|
|
249
|
1723
|
|
300
|
1724 /* start of word */
|
|
346
|
1725 attr = 0;
|
|
385
|
1726 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1727
|
|
300
|
1728 if (attr != 0)
|
|
|
1729 {
|
|
|
1730 /* We found a bad word. Check the attribute. */
|
|
|
1731 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1732 {
|
|
300
|
1733 /* When searching forward only accept a bad word after
|
|
|
1734 * the cursor. */
|
|
|
1735 if (dir == BACKWARD
|
|
|
1736 || lnum > curwin->w_cursor.lnum
|
|
|
1737 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1738 && (colnr_T)(curline ? p - buf + len
|
|
|
1739 : p - buf)
|
|
300
|
1740 > curwin->w_cursor.col))
|
|
236
|
1741 {
|
|
300
|
1742 if (has_syntax)
|
|
249
|
1743 {
|
|
346
|
1744 col = p - buf;
|
|
300
|
1745 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1746 FALSE, &can_spell);
|
|
|
1747 }
|
|
|
1748 else
|
|
|
1749 can_spell = TRUE;
|
|
249
|
1750
|
|
300
|
1751 if (can_spell)
|
|
|
1752 {
|
|
|
1753 found_pos.lnum = lnum;
|
|
346
|
1754 found_pos.col = p - buf;
|
|
249
|
1755 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1756 found_pos.coladd = 0;
|
|
249
|
1757 #endif
|
|
300
|
1758 if (dir == FORWARD)
|
|
|
1759 {
|
|
|
1760 /* No need to search further. */
|
|
|
1761 curwin->w_cursor = found_pos;
|
|
346
|
1762 vim_free(buf);
|
|
300
|
1763 return OK;
|
|
249
|
1764 }
|
|
483
|
1765 else if (curline)
|
|
|
1766 /* Insert mode completion: put cursor after
|
|
|
1767 * the bad word. */
|
|
|
1768 found_pos.col += len;
|
|
249
|
1769 }
|
|
236
|
1770 }
|
|
|
1771 }
|
|
|
1772 }
|
|
|
1773
|
|
300
|
1774 /* advance to character after the word */
|
|
|
1775 p += len;
|
|
385
|
1776 capcol -= len;
|
|
236
|
1777 }
|
|
|
1778
|
|
483
|
1779 if (dir == BACKWARD && found_pos.lnum != 0)
|
|
|
1780 {
|
|
|
1781 /* Use the last match in the line. */
|
|
|
1782 curwin->w_cursor = found_pos;
|
|
|
1783 vim_free(buf);
|
|
|
1784 return OK;
|
|
|
1785 }
|
|
|
1786
|
|
323
|
1787 if (curline)
|
|
346
|
1788 break; /* only check cursor line */
|
|
323
|
1789
|
|
236
|
1790 /* Advance to next line. */
|
|
249
|
1791 if (dir == BACKWARD)
|
|
|
1792 {
|
|
|
1793 if (lnum == 1)
|
|
346
|
1794 break;
|
|
249
|
1795 --lnum;
|
|
385
|
1796 capcol = -1;
|
|
249
|
1797 }
|
|
|
1798 else
|
|
|
1799 {
|
|
|
1800 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1801 break;
|
|
249
|
1802 ++lnum;
|
|
346
|
1803
|
|
|
1804 /* Skip the characters at the start of the next line that were
|
|
|
1805 * included in a match crossing line boundaries. */
|
|
|
1806 if (attr == 0)
|
|
|
1807 skip = p - endp;
|
|
|
1808 else
|
|
|
1809 skip = 0;
|
|
385
|
1810
|
|
|
1811 /* Capscol skips over the inserted space. */
|
|
|
1812 --capcol;
|
|
|
1813
|
|
|
1814 /* But after empty line check first word in next line */
|
|
|
1815 if (*skipwhite(line) == NUL)
|
|
|
1816 capcol = 0;
|
|
249
|
1817 }
|
|
236
|
1818
|
|
|
1819 line_breakcheck();
|
|
|
1820 }
|
|
|
1821
|
|
346
|
1822 vim_free(buf);
|
|
|
1823 return FAIL;
|
|
|
1824 }
|
|
|
1825
|
|
|
1826 /*
|
|
|
1827 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1828 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1829 */
|
|
|
1830 void
|
|
|
1831 spell_cat_line(buf, line, maxlen)
|
|
|
1832 char_u *buf;
|
|
|
1833 char_u *line;
|
|
|
1834 int maxlen;
|
|
|
1835 {
|
|
|
1836 char_u *p;
|
|
|
1837 int n;
|
|
|
1838
|
|
|
1839 p = skipwhite(line);
|
|
|
1840 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1841 p = skipwhite(p + 1);
|
|
|
1842
|
|
|
1843 if (*p != NUL)
|
|
|
1844 {
|
|
|
1845 *buf = ' ';
|
|
376
|
1846 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1847 n = p - line;
|
|
|
1848 if (n >= maxlen)
|
|
|
1849 n = maxlen - 1;
|
|
|
1850 vim_memset(buf + 1, ' ', n);
|
|
|
1851 }
|
|
236
|
1852 }
|
|
|
1853
|
|
|
1854 /*
|
|
307
|
1855 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1856 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1857 */
|
|
307
|
1858 static void
|
|
236
|
1859 spell_load_lang(lang)
|
|
|
1860 char_u *lang;
|
|
|
1861 {
|
|
310
|
1862 char_u fname_enc[85];
|
|
236
|
1863 int r;
|
|
307
|
1864 char_u langcp[MAXWLEN + 1];
|
|
|
1865
|
|
310
|
1866 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1867 * It's truncated when an error is detected. */
|
|
|
1868 STRCPY(langcp, lang);
|
|
|
1869
|
|
310
|
1870 /*
|
|
|
1871 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1872 */
|
|
|
1873 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1874 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1875 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1876
|
|
|
1877 if (r == FAIL && *langcp != NUL)
|
|
|
1878 {
|
|
|
1879 /* Try loading the ASCII version. */
|
|
310
|
1880 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1881 "spell/%s.ascii.spl", lang);
|
|
310
|
1882 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1883 }
|
|
|
1884
|
|
|
1885 if (r == FAIL)
|
|
483
|
1886 smsg((char_u *)_("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""),
|
|
|
1887 lang, spell_enc(), lang);
|
|
310
|
1888 else if (*langcp != NUL)
|
|
|
1889 {
|
|
|
1890 /* Load all the additions. */
|
|
|
1891 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1892 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1893 }
|
|
|
1894 }
|
|
|
1895
|
|
|
1896 /*
|
|
|
1897 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1898 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1899 */
|
|
|
1900 static char_u *
|
|
|
1901 spell_enc()
|
|
|
1902 {
|
|
|
1903
|
|
|
1904 #ifdef FEAT_MBYTE
|
|
|
1905 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1906 return p_enc;
|
|
|
1907 #endif
|
|
|
1908 return (char_u *)"latin1";
|
|
236
|
1909 }
|
|
|
1910
|
|
|
1911 /*
|
|
385
|
1912 * Get the name of the .spl file for the internal wordlist into
|
|
|
1913 * "fname[MAXPATHL]".
|
|
|
1914 */
|
|
|
1915 static void
|
|
|
1916 int_wordlist_spl(fname)
|
|
|
1917 char_u *fname;
|
|
|
1918 {
|
|
|
1919 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
1920 int_wordlist, spell_enc());
|
|
|
1921 }
|
|
|
1922
|
|
|
1923 /*
|
|
236
|
1924 * Allocate a new slang_T.
|
|
|
1925 * Caller must fill "sl_next".
|
|
|
1926 */
|
|
|
1927 static slang_T *
|
|
|
1928 slang_alloc(lang)
|
|
|
1929 char_u *lang;
|
|
|
1930 {
|
|
|
1931 slang_T *lp;
|
|
|
1932
|
|
300
|
1933 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1934 if (lp != NULL)
|
|
|
1935 {
|
|
|
1936 lp->sl_name = vim_strsave(lang);
|
|
323
|
1937 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
483
|
1938 lp->sl_compmax = MAXWLEN;
|
|
|
1939 lp->sl_compminlen = MAXWLEN;
|
|
|
1940 lp->sl_compsylmax = MAXWLEN;
|
|
236
|
1941 }
|
|
|
1942 return lp;
|
|
|
1943 }
|
|
|
1944
|
|
|
1945 /*
|
|
|
1946 * Free the contents of an slang_T and the structure itself.
|
|
|
1947 */
|
|
|
1948 static void
|
|
|
1949 slang_free(lp)
|
|
|
1950 slang_T *lp;
|
|
|
1951 {
|
|
|
1952 vim_free(lp->sl_name);
|
|
310
|
1953 vim_free(lp->sl_fname);
|
|
|
1954 slang_clear(lp);
|
|
|
1955 vim_free(lp);
|
|
|
1956 }
|
|
|
1957
|
|
|
1958 /*
|
|
|
1959 * Clear an slang_T so that the file can be reloaded.
|
|
|
1960 */
|
|
|
1961 static void
|
|
|
1962 slang_clear(lp)
|
|
|
1963 slang_T *lp;
|
|
|
1964 {
|
|
339
|
1965 garray_T *gap;
|
|
|
1966 fromto_T *ftp;
|
|
344
|
1967 salitem_T *smp;
|
|
339
|
1968 int i;
|
|
323
|
1969
|
|
300
|
1970 vim_free(lp->sl_fbyts);
|
|
310
|
1971 lp->sl_fbyts = NULL;
|
|
300
|
1972 vim_free(lp->sl_kbyts);
|
|
310
|
1973 lp->sl_kbyts = NULL;
|
|
339
|
1974 vim_free(lp->sl_pbyts);
|
|
|
1975 lp->sl_pbyts = NULL;
|
|
|
1976
|
|
300
|
1977 vim_free(lp->sl_fidxs);
|
|
310
|
1978 lp->sl_fidxs = NULL;
|
|
300
|
1979 vim_free(lp->sl_kidxs);
|
|
310
|
1980 lp->sl_kidxs = NULL;
|
|
339
|
1981 vim_free(lp->sl_pidxs);
|
|
|
1982 lp->sl_pidxs = NULL;
|
|
323
|
1983
|
|
344
|
1984 gap = &lp->sl_rep;
|
|
|
1985 while (gap->ga_len > 0)
|
|
323
|
1986 {
|
|
344
|
1987 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1988 vim_free(ftp->ft_from);
|
|
|
1989 vim_free(ftp->ft_to);
|
|
323
|
1990 }
|
|
344
|
1991 ga_clear(gap);
|
|
|
1992
|
|
|
1993 gap = &lp->sl_sal;
|
|
375
|
1994 if (lp->sl_sofo)
|
|
376
|
1995 {
|
|
|
1996 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
1997 if (gap->ga_data != NULL)
|
|
|
1998 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1999 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2000 vim_free(((int **)gap->ga_data)[i]);
|
|
|
2001 }
|
|
375
|
2002 else
|
|
|
2003 /* SAL items: free salitem_T items */
|
|
|
2004 while (gap->ga_len > 0)
|
|
|
2005 {
|
|
|
2006 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2007 vim_free(smp->sm_lead);
|
|
|
2008 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
2009 vim_free(smp->sm_to);
|
|
|
2010 #ifdef FEAT_MBYTE
|
|
|
2011 vim_free(smp->sm_lead_w);
|
|
|
2012 vim_free(smp->sm_oneof_w);
|
|
|
2013 vim_free(smp->sm_to_w);
|
|
|
2014 #endif
|
|
|
2015 }
|
|
344
|
2016 ga_clear(gap);
|
|
323
|
2017
|
|
339
|
2018 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
2019 vim_free(lp->sl_prefprog[i]);
|
|
376
|
2020 lp->sl_prefixcnt = 0;
|
|
339
|
2021 vim_free(lp->sl_prefprog);
|
|
376
|
2022 lp->sl_prefprog = NULL;
|
|
|
2023
|
|
|
2024 vim_free(lp->sl_midword);
|
|
|
2025 lp->sl_midword = NULL;
|
|
339
|
2026
|
|
483
|
2027 vim_free(lp->sl_compprog);
|
|
|
2028 vim_free(lp->sl_compstartflags);
|
|
485
|
2029 vim_free(lp->sl_compallflags);
|
|
483
|
2030 lp->sl_compprog = NULL;
|
|
|
2031 lp->sl_compstartflags = NULL;
|
|
485
|
2032 lp->sl_compallflags = NULL;
|
|
483
|
2033
|
|
|
2034 vim_free(lp->sl_syllable);
|
|
|
2035 lp->sl_syllable = NULL;
|
|
|
2036 ga_clear(&lp->sl_syl_items);
|
|
481
|
2037
|
|
330
|
2038 #ifdef FEAT_MBYTE
|
|
|
2039 {
|
|
|
2040 int todo = lp->sl_map_hash.ht_used;
|
|
|
2041 hashitem_T *hi;
|
|
|
2042
|
|
|
2043 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
2044 if (!HASHITEM_EMPTY(hi))
|
|
|
2045 {
|
|
|
2046 --todo;
|
|
|
2047 vim_free(hi->hi_key);
|
|
|
2048 }
|
|
|
2049 }
|
|
|
2050 hash_clear(&lp->sl_map_hash);
|
|
|
2051 #endif
|
|
483
|
2052
|
|
|
2053 lp->sl_compmax = MAXWLEN;
|
|
|
2054 lp->sl_compminlen = MAXWLEN;
|
|
|
2055 lp->sl_compsylmax = MAXWLEN;
|
|
|
2056 lp->sl_regions[0] = NUL;
|
|
236
|
2057 }
|
|
|
2058
|
|
|
2059 /*
|
|
307
|
2060 * Load one spell file and store the info into a slang_T.
|
|
236
|
2061 * Invoked through do_in_runtimepath().
|
|
|
2062 */
|
|
|
2063 static void
|
|
310
|
2064 spell_load_cb(fname, cookie)
|
|
236
|
2065 char_u *fname;
|
|
307
|
2066 void *cookie; /* points to the language name */
|
|
236
|
2067 {
|
|
323
|
2068 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
2069 }
|
|
|
2070
|
|
|
2071 /*
|
|
|
2072 * Load one spell file and store the info into a slang_T.
|
|
|
2073 *
|
|
|
2074 * This is invoked in two ways:
|
|
|
2075 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
2076 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
2077 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
2078 * points to the existing slang_T.
|
|
323
|
2079 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
2080 */
|
|
323
|
2081 static slang_T *
|
|
|
2082 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
2083 char_u *fname;
|
|
|
2084 char_u *lang;
|
|
|
2085 slang_T *old_lp;
|
|
323
|
2086 int silent; /* no error if file doesn't exist */
|
|
310
|
2087 {
|
|
236
|
2088 FILE *fd;
|
|
483
|
2089 char_u buf[VIMSPELLMAGICL];
|
|
236
|
2090 char_u *p;
|
|
339
|
2091 char_u *bp;
|
|
|
2092 idx_T *ip;
|
|
236
|
2093 int i;
|
|
339
|
2094 int n;
|
|
300
|
2095 int len;
|
|
236
|
2096 int round;
|
|
|
2097 char_u *save_sourcing_name = sourcing_name;
|
|
|
2098 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
307
|
2099 slang_T *lp = NULL;
|
|
324
|
2100 idx_T idx;
|
|
344
|
2101 int c = 0;
|
|
483
|
2102 int res;
|
|
236
|
2103
|
|
310
|
2104 fd = mch_fopen((char *)fname, "r");
|
|
236
|
2105 if (fd == NULL)
|
|
|
2106 {
|
|
323
|
2107 if (!silent)
|
|
|
2108 EMSG2(_(e_notopen), fname);
|
|
|
2109 else if (p_verbose > 2)
|
|
|
2110 {
|
|
|
2111 verbose_enter();
|
|
|
2112 smsg((char_u *)e_notopen, fname);
|
|
|
2113 verbose_leave();
|
|
|
2114 }
|
|
255
|
2115 goto endFAIL;
|
|
236
|
2116 }
|
|
310
|
2117 if (p_verbose > 2)
|
|
|
2118 {
|
|
|
2119 verbose_enter();
|
|
|
2120 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
2121 verbose_leave();
|
|
|
2122 }
|
|
|
2123
|
|
|
2124 if (old_lp == NULL)
|
|
|
2125 {
|
|
|
2126 lp = slang_alloc(lang);
|
|
|
2127 if (lp == NULL)
|
|
|
2128 goto endFAIL;
|
|
|
2129
|
|
|
2130 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
2131 lp->sl_fname = vim_strsave(fname);
|
|
|
2132 if (lp->sl_fname == NULL)
|
|
|
2133 goto endFAIL;
|
|
|
2134
|
|
|
2135 /* Check for .add.spl. */
|
|
|
2136 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
2137 }
|
|
|
2138 else
|
|
|
2139 lp = old_lp;
|
|
307
|
2140
|
|
236
|
2141 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
2142 sourcing_name = fname;
|
|
|
2143 sourcing_lnum = 0;
|
|
|
2144
|
|
339
|
2145 /* <HEADER>: <fileID>
|
|
|
2146 */
|
|
236
|
2147 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
2148 buf[i] = getc(fd); /* <fileID> */
|
|
|
2149 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
2150 {
|
|
483
|
2151 EMSG(_("E757: This does not look like a spell file"));
|
|
381
|
2152 goto endFAIL;
|
|
483
|
2153 }
|
|
|
2154 c = getc(fd); /* <versionnr> */
|
|
|
2155 if (c < VIMSPELLVERSION)
|
|
|
2156 {
|
|
|
2157 EMSG(_("E771: Old spell file, needs to be updated"));
|
|
381
|
2158 goto endFAIL;
|
|
|
2159 }
|
|
483
|
2160 else if (c > VIMSPELLVERSION)
|
|
|
2161 {
|
|
|
2162 EMSG(_("E772: Spell file is for newer version of Vim"));
|
|
481
|
2163 goto endFAIL;
|
|
483
|
2164 }
|
|
|
2165
|
|
|
2166
|
|
|
2167 /*
|
|
|
2168 * <SECTIONS>: <section> ... <sectionend>
|
|
|
2169 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
2170 */
|
|
|
2171 for (;;)
|
|
|
2172 {
|
|
|
2173 n = getc(fd); /* <sectionID> or <sectionend> */
|
|
|
2174 if (n == SN_END)
|
|
|
2175 break;
|
|
|
2176 c = getc(fd); /* <sectionflags> */
|
|
|
2177 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2178 /* <sectionlen> */
|
|
|
2179 if (len < 0)
|
|
|
2180 goto truncerr;
|
|
|
2181
|
|
|
2182 res = 0;
|
|
|
2183 switch (n)
|
|
|
2184 {
|
|
|
2185 case SN_REGION:
|
|
|
2186 res = read_region_section(fd, lp, len);
|
|
|
2187 break;
|
|
|
2188
|
|
|
2189 case SN_CHARFLAGS:
|
|
|
2190 res = read_charflags_section(fd);
|
|
|
2191 break;
|
|
|
2192
|
|
|
2193 case SN_MIDWORD:
|
|
|
2194 lp->sl_midword = read_string(fd, len); /* <midword> */
|
|
|
2195 if (lp->sl_midword == NULL)
|
|
|
2196 goto endFAIL;
|
|
|
2197 break;
|
|
|
2198
|
|
|
2199 case SN_PREFCOND:
|
|
|
2200 res = read_prefcond_section(fd, lp);
|
|
|
2201 break;
|
|
|
2202
|
|
|
2203 case SN_REP:
|
|
|
2204 res = read_rep_section(fd, lp);
|
|
|
2205 break;
|
|
|
2206
|
|
|
2207 case SN_SAL:
|
|
|
2208 res = read_sal_section(fd, lp);
|
|
|
2209 break;
|
|
|
2210
|
|
|
2211 case SN_SOFO:
|
|
|
2212 res = read_sofo_section(fd, lp);
|
|
|
2213 break;
|
|
|
2214
|
|
|
2215 case SN_MAP:
|
|
|
2216 p = read_string(fd, len); /* <mapstr> */
|
|
|
2217 if (p == NULL)
|
|
|
2218 goto endFAIL;
|
|
|
2219 set_map_str(lp, p);
|
|
|
2220 vim_free(p);
|
|
|
2221 break;
|
|
|
2222
|
|
|
2223 case SN_COMPOUND:
|
|
|
2224 res = read_compound(fd, lp, len);
|
|
|
2225 break;
|
|
|
2226
|
|
|
2227 case SN_SYLLABLE:
|
|
|
2228 lp->sl_syllable = read_string(fd, len); /* <syllable> */
|
|
|
2229 if (lp->sl_syllable == NULL)
|
|
|
2230 goto endFAIL;
|
|
|
2231 if (init_syl_tab(lp) == FAIL)
|
|
|
2232 goto endFAIL;
|
|
|
2233 break;
|
|
|
2234
|
|
|
2235 default:
|
|
|
2236 /* Unsupported section. When it's required give an error
|
|
|
2237 * message. When it's not required skip the contents. */
|
|
|
2238 if (c & SNF_REQUIRED)
|
|
|
2239 {
|
|
|
2240 EMSG(_("E770: Unsupported section in spell file"));
|
|
|
2241 goto endFAIL;
|
|
|
2242 }
|
|
|
2243 while (--len >= 0)
|
|
|
2244 if (getc(fd) < 0)
|
|
|
2245 goto truncerr;
|
|
|
2246 break;
|
|
|
2247 }
|
|
|
2248 if (res == SP_FORMERROR)
|
|
|
2249 {
|
|
|
2250 formerr:
|
|
|
2251 EMSG(_(e_format));
|
|
381
|
2252 goto endFAIL;
|
|
344
|
2253 }
|
|
483
|
2254 if (res == SP_TRUNCERROR)
|
|
|
2255 {
|
|
|
2256 truncerr:
|
|
|
2257 EMSG(_(e_spell_trunc));
|
|
375
|
2258 goto endFAIL;
|
|
483
|
2259 }
|
|
|
2260 if (res == SP_ERROR)
|
|
|
2261 goto endFAIL;
|
|
389
|
2262 }
|
|
236
|
2263
|
|
300
|
2264 /* round 1: <LWORDTREE>
|
|
339
|
2265 * round 2: <KWORDTREE>
|
|
|
2266 * round 3: <PREFIXTREE> */
|
|
|
2267 for (round = 1; round <= 3; ++round)
|
|
236
|
2268 {
|
|
300
|
2269 /* The tree size was computed when writing the file, so that we can
|
|
|
2270 * allocate it as one long block. <nodecount> */
|
|
|
2271 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2272 if (len < 0)
|
|
|
2273 goto truncerr;
|
|
|
2274 if (len > 0)
|
|
236
|
2275 {
|
|
300
|
2276 /* Allocate the byte array. */
|
|
339
|
2277 bp = lalloc((long_u)len, TRUE);
|
|
|
2278 if (bp == NULL)
|
|
300
|
2279 goto endFAIL;
|
|
|
2280 if (round == 1)
|
|
339
|
2281 lp->sl_fbyts = bp;
|
|
|
2282 else if (round == 2)
|
|
|
2283 lp->sl_kbyts = bp;
|
|
300
|
2284 else
|
|
339
|
2285 lp->sl_pbyts = bp;
|
|
236
|
2286
|
|
300
|
2287 /* Allocate the index array. */
|
|
339
|
2288 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2289 if (ip == NULL)
|
|
300
|
2290 goto endFAIL;
|
|
|
2291 if (round == 1)
|
|
339
|
2292 lp->sl_fidxs = ip;
|
|
|
2293 else if (round == 2)
|
|
|
2294 lp->sl_kidxs = ip;
|
|
300
|
2295 else
|
|
339
|
2296 lp->sl_pidxs = ip;
|
|
300
|
2297
|
|
|
2298 /* Read the tree and store it in the array. */
|
|
339
|
2299 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2300 if (idx == -1)
|
|
300
|
2301 goto truncerr;
|
|
324
|
2302 if (idx < 0)
|
|
236
|
2303 goto formerr;
|
|
|
2304 }
|
|
300
|
2305 }
|
|
243
|
2306
|
|
310
|
2307 /* For a new file link it in the list of spell files. */
|
|
|
2308 if (old_lp == NULL)
|
|
|
2309 {
|
|
|
2310 lp->sl_next = first_lang;
|
|
|
2311 first_lang = lp;
|
|
|
2312 }
|
|
307
|
2313
|
|
255
|
2314 goto endOK;
|
|
|
2315
|
|
|
2316 endFAIL:
|
|
310
|
2317 if (lang != NULL)
|
|
|
2318 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2319 *lang = NUL;
|
|
|
2320 if (lp != NULL && old_lp == NULL)
|
|
323
|
2321 {
|
|
307
|
2322 slang_free(lp);
|
|
323
|
2323 lp = NULL;
|
|
|
2324 }
|
|
255
|
2325
|
|
|
2326 endOK:
|
|
236
|
2327 if (fd != NULL)
|
|
|
2328 fclose(fd);
|
|
|
2329 sourcing_name = save_sourcing_name;
|
|
|
2330 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2331
|
|
|
2332 return lp;
|
|
236
|
2333 }
|
|
|
2334
|
|
376
|
2335 /*
|
|
|
2336 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2337 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2338 * Returns NULL when the count is zero.
|
|
483
|
2339 * Sets "*cntp" to SP_*ERROR when there is an error, length of the result
|
|
|
2340 * otherwise.
|
|
376
|
2341 */
|
|
|
2342 static char_u *
|
|
389
|
2343 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2344 FILE *fd;
|
|
|
2345 int cnt_bytes;
|
|
389
|
2346 int *cntp;
|
|
376
|
2347 {
|
|
|
2348 int cnt = 0;
|
|
|
2349 int i;
|
|
|
2350 char_u *str;
|
|
|
2351
|
|
|
2352 /* read the length bytes, MSB first */
|
|
|
2353 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2354 cnt = (cnt << 8) + getc(fd);
|
|
|
2355 if (cnt < 0)
|
|
|
2356 {
|
|
483
|
2357 *cntp = SP_TRUNCERROR;
|
|
376
|
2358 return NULL;
|
|
|
2359 }
|
|
389
|
2360 *cntp = cnt;
|
|
|
2361 if (cnt == 0)
|
|
|
2362 return NULL; /* nothing to read, return NULL */
|
|
376
|
2363
|
|
483
|
2364 str = read_string(fd, cnt);
|
|
|
2365 if (str == NULL)
|
|
|
2366 *cntp = SP_ERROR;
|
|
|
2367 return str;
|
|
|
2368 }
|
|
|
2369
|
|
|
2370 /*
|
|
|
2371 * Read a string of length "cnt" from "fd" into allocated memory.
|
|
|
2372 * Returns NULL when out of memory.
|
|
|
2373 */
|
|
|
2374 static char_u *
|
|
|
2375 read_string(fd, cnt)
|
|
|
2376 FILE *fd;
|
|
|
2377 int cnt;
|
|
|
2378 {
|
|
|
2379 char_u *str;
|
|
|
2380 int i;
|
|
|
2381
|
|
376
|
2382 /* allocate memory */
|
|
|
2383 str = alloc((unsigned)cnt + 1);
|
|
483
|
2384 if (str != NULL)
|
|
|
2385 {
|
|
|
2386 /* Read the string. Doesn't check for truncated file. */
|
|
|
2387 for (i = 0; i < cnt; ++i)
|
|
|
2388 str[i] = getc(fd);
|
|
|
2389 str[i] = NUL;
|
|
|
2390 }
|
|
|
2391 return str;
|
|
|
2392 }
|
|
|
2393
|
|
|
2394 /*
|
|
|
2395 * Read SN_REGION: <regionname> ...
|
|
|
2396 * Return SP_*ERROR flags.
|
|
|
2397 */
|
|
|
2398 static int
|
|
|
2399 read_region_section(fd, lp, len)
|
|
|
2400 FILE *fd;
|
|
|
2401 slang_T *lp;
|
|
|
2402 int len;
|
|
|
2403 {
|
|
|
2404 int i;
|
|
|
2405
|
|
|
2406 if (len > 16)
|
|
|
2407 return SP_FORMERROR;
|
|
|
2408 for (i = 0; i < len; ++i)
|
|
|
2409 lp->sl_regions[i] = getc(fd); /* <regionname> */
|
|
|
2410 lp->sl_regions[len] = NUL;
|
|
|
2411 return 0;
|
|
|
2412 }
|
|
|
2413
|
|
|
2414 /*
|
|
|
2415 * Read SN_CHARFLAGS section: <charflagslen> <charflags>
|
|
|
2416 * <folcharslen> <folchars>
|
|
|
2417 * Return SP_*ERROR flags.
|
|
|
2418 */
|
|
|
2419 static int
|
|
|
2420 read_charflags_section(fd)
|
|
|
2421 FILE *fd;
|
|
|
2422 {
|
|
|
2423 char_u *flags;
|
|
|
2424 char_u *fol;
|
|
|
2425 int flagslen, follen;
|
|
|
2426
|
|
|
2427 /* <charflagslen> <charflags> */
|
|
|
2428 flags = read_cnt_string(fd, 1, &flagslen);
|
|
|
2429 if (flagslen < 0)
|
|
|
2430 return flagslen;
|
|
|
2431
|
|
|
2432 /* <folcharslen> <folchars> */
|
|
|
2433 fol = read_cnt_string(fd, 2, &follen);
|
|
|
2434 if (follen < 0)
|
|
|
2435 {
|
|
|
2436 vim_free(flags);
|
|
|
2437 return follen;
|
|
|
2438 }
|
|
|
2439
|
|
|
2440 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
2441 if (flags != NULL && fol != NULL)
|
|
|
2442 set_spell_charflags(flags, flagslen, fol);
|
|
|
2443
|
|
|
2444 vim_free(flags);
|
|
|
2445 vim_free(fol);
|
|
|
2446
|
|
|
2447 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
2448 if ((flags == NULL) != (fol == NULL))
|
|
|
2449 return SP_FORMERROR;
|
|
|
2450 return 0;
|
|
|
2451 }
|
|
|
2452
|
|
|
2453 /*
|
|
|
2454 * Read SN_PREFCOND section.
|
|
|
2455 * Return SP_*ERROR flags.
|
|
|
2456 */
|
|
|
2457 static int
|
|
|
2458 read_prefcond_section(fd, lp)
|
|
|
2459 FILE *fd;
|
|
|
2460 slang_T *lp;
|
|
|
2461 {
|
|
|
2462 int cnt;
|
|
|
2463 int i;
|
|
|
2464 int n;
|
|
|
2465 char_u *p;
|
|
|
2466 char_u buf[MAXWLEN + 1];
|
|
|
2467
|
|
|
2468 /* <prefcondcnt> <prefcond> ... */
|
|
|
2469 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
2470 if (cnt <= 0)
|
|
|
2471 return SP_FORMERROR;
|
|
|
2472
|
|
|
2473 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
2474 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
2475 if (lp->sl_prefprog == NULL)
|
|
|
2476 return SP_ERROR;
|
|
|
2477 lp->sl_prefixcnt = cnt;
|
|
|
2478
|
|
376
|
2479 for (i = 0; i < cnt; ++i)
|
|
483
|
2480 {
|
|
|
2481 /* <prefcond> : <condlen> <condstr> */
|
|
|
2482 n = getc(fd); /* <condlen> */
|
|
|
2483 if (n < 0 || n >= MAXWLEN)
|
|
|
2484 return SP_FORMERROR;
|
|
|
2485
|
|
|
2486 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
2487 * compile the regexp program used to check for the condition. */
|
|
|
2488 if (n > 0)
|
|
|
2489 {
|
|
|
2490 buf[0] = '^'; /* always match at one position only */
|
|
|
2491 p = buf + 1;
|
|
|
2492 while (n-- > 0)
|
|
|
2493 *p++ = getc(fd); /* <condstr> */
|
|
|
2494 *p = NUL;
|
|
|
2495 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
2496 }
|
|
|
2497 }
|
|
|
2498 return 0;
|
|
|
2499 }
|
|
|
2500
|
|
|
2501 /*
|
|
|
2502 * Read REP items section from "fd": <repcount> <rep> ...
|
|
|
2503 * Return SP_*ERROR flags.
|
|
|
2504 */
|
|
|
2505 static int
|
|
|
2506 read_rep_section(fd, slang)
|
|
|
2507 FILE *fd;
|
|
|
2508 slang_T *slang;
|
|
|
2509 {
|
|
|
2510 int cnt;
|
|
|
2511 garray_T *gap;
|
|
|
2512 fromto_T *ftp;
|
|
|
2513 short *first;
|
|
|
2514 int i;
|
|
|
2515
|
|
|
2516 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
2517 if (cnt < 0)
|
|
|
2518 return SP_TRUNCERROR;
|
|
|
2519
|
|
|
2520 gap = &slang->sl_rep;
|
|
|
2521 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2522 return SP_ERROR;
|
|
|
2523
|
|
|
2524 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
2525 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2526 {
|
|
|
2527 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
|
2528 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
|
2529 if (i < 0)
|
|
|
2530 return i;
|
|
|
2531 if (i == 0)
|
|
|
2532 return SP_FORMERROR;
|
|
|
2533 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
|
2534 if (i <= 0)
|
|
|
2535 {
|
|
|
2536 vim_free(ftp->ft_from);
|
|
|
2537 if (i < 0)
|
|
|
2538 return i;
|
|
|
2539 return SP_FORMERROR;
|
|
|
2540 }
|
|
|
2541 }
|
|
|
2542
|
|
|
2543 /* Fill the first-index table. */
|
|
|
2544 first = slang->sl_rep_first;
|
|
|
2545 for (i = 0; i < 256; ++i)
|
|
|
2546 first[i] = -1;
|
|
|
2547 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2548 {
|
|
|
2549 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
2550 if (first[*ftp->ft_from] == -1)
|
|
|
2551 first[*ftp->ft_from] = i;
|
|
|
2552 }
|
|
|
2553 return 0;
|
|
|
2554 }
|
|
|
2555
|
|
|
2556 /*
|
|
|
2557 * Read SN_SAL section: <salflags> <salcount> <sal> ...
|
|
|
2558 * Return SP_*ERROR flags.
|
|
|
2559 */
|
|
|
2560 static int
|
|
|
2561 read_sal_section(fd, slang)
|
|
|
2562 FILE *fd;
|
|
|
2563 slang_T *slang;
|
|
|
2564 {
|
|
|
2565 int i;
|
|
|
2566 int cnt;
|
|
|
2567 garray_T *gap;
|
|
|
2568 salitem_T *smp;
|
|
|
2569 int ccnt;
|
|
|
2570 char_u *p;
|
|
485
|
2571 int c = NUL;
|
|
483
|
2572
|
|
|
2573 slang->sl_sofo = FALSE;
|
|
|
2574
|
|
|
2575 i = getc(fd); /* <salflags> */
|
|
|
2576 if (i & SAL_F0LLOWUP)
|
|
|
2577 slang->sl_followup = TRUE;
|
|
|
2578 if (i & SAL_COLLAPSE)
|
|
|
2579 slang->sl_collapse = TRUE;
|
|
|
2580 if (i & SAL_REM_ACCENTS)
|
|
|
2581 slang->sl_rem_accents = TRUE;
|
|
|
2582
|
|
|
2583 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2584 if (cnt < 0)
|
|
|
2585 return SP_TRUNCERROR;
|
|
|
2586
|
|
|
2587 gap = &slang->sl_sal;
|
|
|
2588 ga_init2(gap, sizeof(salitem_T), 10);
|
|
|
2589 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2590 return SP_ERROR;
|
|
|
2591
|
|
|
2592 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2593 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2594 {
|
|
|
2595 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2596 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2597 if (ccnt < 0)
|
|
|
2598 return SP_TRUNCERROR;
|
|
|
2599 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
2600 return SP_ERROR;
|
|
|
2601 smp->sm_lead = p;
|
|
|
2602
|
|
|
2603 /* Read up to the first special char into sm_lead. */
|
|
|
2604 for (i = 0; i < ccnt; ++i)
|
|
|
2605 {
|
|
|
2606 c = getc(fd); /* <salfrom> */
|
|
|
2607 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2608 break;
|
|
|
2609 *p++ = c;
|
|
|
2610 }
|
|
|
2611 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2612 *p++ = NUL;
|
|
|
2613
|
|
|
2614 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2615 if (c == '(')
|
|
|
2616 {
|
|
|
2617 smp->sm_oneof = p;
|
|
|
2618 for (++i; i < ccnt; ++i)
|
|
|
2619 {
|
|
|
2620 c = getc(fd); /* <salfrom> */
|
|
|
2621 if (c == ')')
|
|
|
2622 break;
|
|
|
2623 *p++ = c;
|
|
|
2624 }
|
|
|
2625 *p++ = NUL;
|
|
|
2626 if (++i < ccnt)
|
|
|
2627 c = getc(fd);
|
|
|
2628 }
|
|
|
2629 else
|
|
|
2630 smp->sm_oneof = NULL;
|
|
|
2631
|
|
|
2632 /* Any following chars go in sm_rules. */
|
|
|
2633 smp->sm_rules = p;
|
|
|
2634 if (i < ccnt)
|
|
|
2635 /* store the char we got while checking for end of sm_lead */
|
|
|
2636 *p++ = c;
|
|
|
2637 for (++i; i < ccnt; ++i)
|
|
|
2638 *p++ = getc(fd); /* <salfrom> */
|
|
|
2639 *p++ = NUL;
|
|
|
2640
|
|
|
2641 /* <saltolen> <salto> */
|
|
|
2642 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
|
2643 if (ccnt < 0)
|
|
|
2644 {
|
|
|
2645 vim_free(smp->sm_lead);
|
|
|
2646 return ccnt;
|
|
|
2647 }
|
|
|
2648
|
|
|
2649 #ifdef FEAT_MBYTE
|
|
|
2650 if (has_mbyte)
|
|
|
2651 {
|
|
|
2652 /* convert the multi-byte strings to wide char strings */
|
|
|
2653 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2654 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2655 if (smp->sm_oneof == NULL)
|
|
|
2656 smp->sm_oneof_w = NULL;
|
|
|
2657 else
|
|
|
2658 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
2659 if (smp->sm_to == NULL)
|
|
|
2660 smp->sm_to_w = NULL;
|
|
|
2661 else
|
|
|
2662 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
2663 if (smp->sm_lead_w == NULL
|
|
|
2664 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
2665 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
|
2666 {
|
|
|
2667 vim_free(smp->sm_lead);
|
|
|
2668 vim_free(smp->sm_to);
|
|
|
2669 vim_free(smp->sm_lead_w);
|
|
|
2670 vim_free(smp->sm_oneof_w);
|
|
|
2671 vim_free(smp->sm_to_w);
|
|
|
2672 return SP_ERROR;
|
|
|
2673 }
|
|
|
2674 }
|
|
|
2675 #endif
|
|
|
2676 }
|
|
|
2677
|
|
|
2678 /* Fill the first-index table. */
|
|
|
2679 set_sal_first(slang);
|
|
|
2680
|
|
|
2681 return 0;
|
|
|
2682 }
|
|
|
2683
|
|
|
2684 /*
|
|
|
2685 * SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
2686 * Return SP_*ERROR flags.
|
|
|
2687 */
|
|
|
2688 static int
|
|
|
2689 read_sofo_section(fd, slang)
|
|
|
2690 FILE *fd;
|
|
|
2691 slang_T *slang;
|
|
|
2692 {
|
|
|
2693 int cnt;
|
|
|
2694 char_u *from, *to;
|
|
|
2695 int res;
|
|
|
2696
|
|
|
2697 slang->sl_sofo = TRUE;
|
|
|
2698
|
|
|
2699 /* <sofofromlen> <sofofrom> */
|
|
|
2700 from = read_cnt_string(fd, 2, &cnt);
|
|
|
2701 if (cnt < 0)
|
|
|
2702 return cnt;
|
|
|
2703
|
|
|
2704 /* <sofotolen> <sofoto> */
|
|
|
2705 to = read_cnt_string(fd, 2, &cnt);
|
|
|
2706 if (cnt < 0)
|
|
|
2707 {
|
|
|
2708 vim_free(from);
|
|
|
2709 return cnt;
|
|
|
2710 }
|
|
|
2711
|
|
|
2712 /* Store the info in slang->sl_sal and/or slang->sl_sal_first. */
|
|
|
2713 if (from != NULL && to != NULL)
|
|
|
2714 res = set_sofo(slang, from, to);
|
|
|
2715 else if (from != NULL || to != NULL)
|
|
|
2716 res = SP_FORMERROR; /* only one of two strings is an error */
|
|
|
2717 else
|
|
|
2718 res = 0;
|
|
|
2719
|
|
|
2720 vim_free(from);
|
|
|
2721 vim_free(to);
|
|
|
2722 return res;
|
|
|
2723 }
|
|
|
2724
|
|
|
2725 /*
|
|
|
2726 * Read the compound section from the .spl file:
|
|
|
2727 * <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
2728 * Returns SP_*ERROR flags.
|
|
|
2729 */
|
|
|
2730 static int
|
|
|
2731 read_compound(fd, slang, len)
|
|
|
2732 FILE *fd;
|
|
|
2733 slang_T *slang;
|
|
|
2734 int len;
|
|
|
2735 {
|
|
|
2736 int todo = len;
|
|
|
2737 int c;
|
|
|
2738 int atstart;
|
|
|
2739 char_u *pat;
|
|
|
2740 char_u *pp;
|
|
|
2741 char_u *cp;
|
|
485
|
2742 char_u *ap;
|
|
483
|
2743
|
|
|
2744 if (todo < 2)
|
|
|
2745 return SP_FORMERROR; /* need at least two bytes */
|
|
|
2746
|
|
|
2747 --todo;
|
|
|
2748 c = getc(fd); /* <compmax> */
|
|
|
2749 if (c < 2)
|
|
|
2750 c = MAXWLEN;
|
|
|
2751 slang->sl_compmax = c;
|
|
|
2752
|
|
|
2753 --todo;
|
|
|
2754 c = getc(fd); /* <compminlen> */
|
|
|
2755 if (c < 1)
|
|
|
2756 c = 3;
|
|
|
2757 slang->sl_compminlen = c;
|
|
|
2758
|
|
|
2759 --todo;
|
|
|
2760 c = getc(fd); /* <compsylmax> */
|
|
|
2761 if (c < 1)
|
|
|
2762 c = MAXWLEN;
|
|
|
2763 slang->sl_compsylmax = c;
|
|
|
2764
|
|
|
2765 /* Turn the COMPOUNDFLAGS items into a regexp pattern:
|
|
|
2766 * "a[bc]/a*b+" -> "^\(a[bc]\|a*b\+\)$".
|
|
|
2767 * Inserting backslashes may double the length, "^\(\)$<Nul>" is 7 bytes. */
|
|
|
2768 pat = alloc((unsigned)todo * 2 + 7);
|
|
|
2769 if (pat == NULL)
|
|
|
2770 return SP_ERROR;
|
|
|
2771
|
|
485
|
2772 /* We also need a list of all flags that can appear at the start and one
|
|
|
2773 * for all flags. */
|
|
483
|
2774 cp = alloc(todo + 1);
|
|
|
2775 if (cp == NULL)
|
|
|
2776 {
|
|
|
2777 vim_free(pat);
|
|
|
2778 return SP_ERROR;
|
|
|
2779 }
|
|
|
2780 slang->sl_compstartflags = cp;
|
|
|
2781 *cp = NUL;
|
|
|
2782
|
|
485
|
2783 ap = alloc(todo + 1);
|
|
|
2784 if (ap == NULL)
|
|
|
2785 {
|
|
|
2786 vim_free(pat);
|
|
|
2787 return SP_ERROR;
|
|
|
2788 }
|
|
|
2789 slang->sl_compallflags = ap;
|
|
|
2790 *ap = NUL;
|
|
|
2791
|
|
483
|
2792 pp = pat;
|
|
|
2793 *pp++ = '^';
|
|
|
2794 *pp++ = '\\';
|
|
|
2795 *pp++ = '(';
|
|
|
2796
|
|
|
2797 atstart = 1;
|
|
|
2798 while (todo-- > 0)
|
|
|
2799 {
|
|
|
2800 c = getc(fd); /* <compflags> */
|
|
485
|
2801
|
|
|
2802 /* Add all flags to "sl_compallflags". */
|
|
|
2803 if (vim_strchr((char_u *)"+*[]/", c) == NULL
|
|
|
2804 && vim_strchr(slang->sl_compallflags, c) == NULL)
|
|
|
2805 {
|
|
|
2806 *ap++ = c;
|
|
|
2807 *ap = NUL;
|
|
|
2808 }
|
|
|
2809
|
|
483
|
2810 if (atstart != 0)
|
|
|
2811 {
|
|
|
2812 /* At start of item: copy flags to "sl_compstartflags". For a
|
|
|
2813 * [abc] item set "atstart" to 2 and copy up to the ']'. */
|
|
|
2814 if (c == '[')
|
|
|
2815 atstart = 2;
|
|
|
2816 else if (c == ']')
|
|
|
2817 atstart = 0;
|
|
|
2818 else
|
|
|
2819 {
|
|
|
2820 if (vim_strchr(slang->sl_compstartflags, c) == NULL)
|
|
|
2821 {
|
|
|
2822 *cp++ = c;
|
|
|
2823 *cp = NUL;
|
|
|
2824 }
|
|
|
2825 if (atstart == 1)
|
|
|
2826 atstart = 0;
|
|
|
2827 }
|
|
|
2828 }
|
|
|
2829 if (c == '/') /* slash separates two items */
|
|
|
2830 {
|
|
|
2831 *pp++ = '\\';
|
|
|
2832 *pp++ = '|';
|
|
|
2833 atstart = 1;
|
|
|
2834 }
|
|
|
2835 else /* normal char, "[abc]" and '*' are copied as-is */
|
|
|
2836 {
|
|
|
2837 if (c == '+')
|
|
|
2838 *pp++ = '\\'; /* "a+" becomes "a\+" */
|
|
|
2839 *pp++ = c;
|
|
|
2840 }
|
|
|
2841 }
|
|
|
2842
|
|
|
2843 *pp++ = '\\';
|
|
|
2844 *pp++ = ')';
|
|
|
2845 *pp++ = '$';
|
|
|
2846 *pp = NUL;
|
|
|
2847
|
|
|
2848 slang->sl_compprog = vim_regcomp(pat, RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
2849 vim_free(pat);
|
|
|
2850 if (slang->sl_compprog == NULL)
|
|
|
2851 return SP_FORMERROR;
|
|
|
2852
|
|
|
2853 return 0;
|
|
|
2854 }
|
|
|
2855
|
|
|
2856 #define SY_MAXLEN 30
|
|
|
2857 typedef struct syl_item_S
|
|
|
2858 {
|
|
|
2859 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */
|
|
|
2860 int sy_len;
|
|
|
2861 } syl_item_T;
|
|
|
2862
|
|
|
2863 /*
|
|
|
2864 * Truncate "slang->sl_syllable" at the first slash and put the following items
|
|
|
2865 * in "slang->sl_syl_items".
|
|
|
2866 */
|
|
|
2867 static int
|
|
|
2868 init_syl_tab(slang)
|
|
|
2869 slang_T *slang;
|
|
|
2870 {
|
|
|
2871 char_u *p;
|
|
|
2872 char_u *s;
|
|
|
2873 int l;
|
|
|
2874 syl_item_T *syl;
|
|
|
2875
|
|
|
2876 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4);
|
|
|
2877 p = vim_strchr(slang->sl_syllable, '/');
|
|
|
2878 while (p != NULL)
|
|
|
2879 {
|
|
|
2880 *p++ = NUL;
|
|
|
2881 if (p == NUL)
|
|
|
2882 break;
|
|
|
2883 s = p;
|
|
|
2884 p = vim_strchr(p, '/');
|
|
|
2885 if (p == NULL)
|
|
|
2886 l = STRLEN(s);
|
|
|
2887 else
|
|
|
2888 l = p - s;
|
|
|
2889 if (l >= SY_MAXLEN)
|
|
|
2890 return SP_FORMERROR;
|
|
|
2891 if (ga_grow(&slang->sl_syl_items, 1) == FAIL)
|
|
|
2892 return SP_ERROR;
|
|
|
2893 syl = ((syl_item_T *)slang->sl_syl_items.ga_data)
|
|
|
2894 + slang->sl_syl_items.ga_len++;
|
|
|
2895 vim_strncpy(syl->sy_chars, s, l);
|
|
|
2896 syl->sy_len = l;
|
|
|
2897 }
|
|
|
2898 return OK;
|
|
|
2899 }
|
|
|
2900
|
|
|
2901 /*
|
|
|
2902 * Count the number of syllables in "word".
|
|
|
2903 * When "word" contains spaces the syllables after the last space are counted.
|
|
|
2904 * Returns zero if syllables are not defines.
|
|
|
2905 */
|
|
|
2906 static int
|
|
|
2907 count_syllables(slang, word)
|
|
|
2908 slang_T *slang;
|
|
|
2909 char_u *word;
|
|
|
2910 {
|
|
|
2911 int cnt = 0;
|
|
|
2912 int skip = FALSE;
|
|
|
2913 char_u *p;
|
|
|
2914 int len;
|
|
|
2915 int i;
|
|
|
2916 syl_item_T *syl;
|
|
|
2917 int c;
|
|
|
2918
|
|
|
2919 if (slang->sl_syllable == NULL)
|
|
|
2920 return 0;
|
|
|
2921
|
|
|
2922 for (p = word; *p != NUL; p += len)
|
|
|
2923 {
|
|
|
2924 /* When running into a space reset counter. */
|
|
|
2925 if (*p == ' ')
|
|
|
2926 {
|
|
|
2927 len = 1;
|
|
|
2928 cnt = 0;
|
|
|
2929 continue;
|
|
|
2930 }
|
|
|
2931
|
|
|
2932 /* Find longest match of syllable items. */
|
|
|
2933 len = 0;
|
|
|
2934 for (i = 0; i < slang->sl_syl_items.ga_len; ++i)
|
|
|
2935 {
|
|
|
2936 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i;
|
|
|
2937 if (syl->sy_len > len
|
|
|
2938 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0)
|
|
|
2939 len = syl->sy_len;
|
|
|
2940 }
|
|
|
2941 if (len != 0) /* found a match, count syllable */
|
|
|
2942 {
|
|
|
2943 ++cnt;
|
|
|
2944 skip = FALSE;
|
|
|
2945 }
|
|
|
2946 else
|
|
|
2947 {
|
|
|
2948 /* No recognized syllable item, at least a syllable char then? */
|
|
|
2949 #ifdef FEAT_MBYTE
|
|
|
2950 c = mb_ptr2char(p);
|
|
|
2951 len = (*mb_ptr2len)(p);
|
|
|
2952 #else
|
|
|
2953 c = *p;
|
|
|
2954 len = 1;
|
|
|
2955 #endif
|
|
|
2956 if (vim_strchr(slang->sl_syllable, c) == NULL)
|
|
|
2957 skip = FALSE; /* No, search for next syllable */
|
|
|
2958 else if (!skip)
|
|
|
2959 {
|
|
|
2960 ++cnt; /* Yes, count it */
|
|
|
2961 skip = TRUE; /* don't count following syllable chars */
|
|
|
2962 }
|
|
|
2963 }
|
|
|
2964 }
|
|
|
2965 return cnt;
|
|
376
|
2966 }
|
|
|
2967
|
|
381
|
2968 /*
|
|
|
2969 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
483
|
2970 * Returns SP_*ERROR flags when there is something wrong.
|
|
381
|
2971 */
|
|
|
2972 static int
|
|
|
2973 set_sofo(lp, from, to)
|
|
|
2974 slang_T *lp;
|
|
|
2975 char_u *from;
|
|
|
2976 char_u *to;
|
|
|
2977 {
|
|
|
2978 int i;
|
|
|
2979
|
|
|
2980 #ifdef FEAT_MBYTE
|
|
|
2981 garray_T *gap;
|
|
|
2982 char_u *s;
|
|
|
2983 char_u *p;
|
|
|
2984 int c;
|
|
|
2985 int *inp;
|
|
|
2986
|
|
|
2987 if (has_mbyte)
|
|
|
2988 {
|
|
|
2989 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
2990 * characters. The index is the low byte of the character.
|
|
|
2991 * The list contains from-to pairs with a terminating NUL.
|
|
|
2992 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
2993 gap = &lp->sl_sal;
|
|
|
2994 ga_init2(gap, sizeof(int *), 1);
|
|
|
2995 if (ga_grow(gap, 256) == FAIL)
|
|
483
|
2996 return SP_ERROR;
|
|
381
|
2997 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
2998 gap->ga_len = 256;
|
|
|
2999
|
|
|
3000 /* First count the number of items for each list. Temporarily use
|
|
|
3001 * sl_sal_first[] for this. */
|
|
|
3002 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3003 {
|
|
474
|
3004 c = mb_cptr2char_adv(&p);
|
|
|
3005 mb_cptr_adv(s);
|
|
381
|
3006 if (c >= 256)
|
|
|
3007 ++lp->sl_sal_first[c & 0xff];
|
|
|
3008 }
|
|
|
3009 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
483
|
3010 return SP_FORMERROR;
|
|
381
|
3011
|
|
|
3012 /* Allocate the lists. */
|
|
|
3013 for (i = 0; i < 256; ++i)
|
|
|
3014 if (lp->sl_sal_first[i] > 0)
|
|
|
3015 {
|
|
|
3016 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
3017 if (p == NULL)
|
|
483
|
3018 return SP_ERROR;
|
|
381
|
3019 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
3020 *(int *)p = 0;
|
|
|
3021 }
|
|
|
3022
|
|
|
3023 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
3024 * list. */
|
|
|
3025 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
3026 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3027 {
|
|
474
|
3028 c = mb_cptr2char_adv(&p);
|
|
|
3029 i = mb_cptr2char_adv(&s);
|
|
381
|
3030 if (c >= 256)
|
|
|
3031 {
|
|
|
3032 /* Append the from-to chars at the end of the list with
|
|
|
3033 * the low byte. */
|
|
|
3034 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
3035 while (*inp != 0)
|
|
|
3036 ++inp;
|
|
|
3037 *inp++ = c; /* from char */
|
|
|
3038 *inp++ = i; /* to char */
|
|
|
3039 *inp++ = NUL; /* NUL at the end */
|
|
|
3040 }
|
|
|
3041 else
|
|
|
3042 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
3043 lp->sl_sal_first[c] = i;
|
|
|
3044 }
|
|
|
3045 }
|
|
|
3046 else
|
|
|
3047 #endif
|
|
|
3048 {
|
|
|
3049 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
3050 if (STRLEN(from) != STRLEN(to))
|
|
483
|
3051 return SP_FORMERROR;
|
|
381
|
3052
|
|
|
3053 for (i = 0; to[i] != NUL; ++i)
|
|
|
3054 lp->sl_sal_first[from[i]] = to[i];
|
|
|
3055 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
3056 }
|
|
|
3057
|
|
483
|
3058 return 0;
|
|
381
|
3059 }
|
|
|
3060
|
|
|
3061 /*
|
|
|
3062 * Fill the first-index table for "lp".
|
|
|
3063 */
|
|
|
3064 static void
|
|
|
3065 set_sal_first(lp)
|
|
|
3066 slang_T *lp;
|
|
|
3067 {
|
|
|
3068 salfirst_T *sfirst;
|
|
|
3069 int i;
|
|
|
3070 salitem_T *smp;
|
|
|
3071 int c;
|
|
|
3072 garray_T *gap = &lp->sl_sal;
|
|
|
3073
|
|
|
3074 sfirst = lp->sl_sal_first;
|
|
|
3075 for (i = 0; i < 256; ++i)
|
|
|
3076 sfirst[i] = -1;
|
|
|
3077 smp = (salitem_T *)gap->ga_data;
|
|
|
3078 for (i = 0; i < gap->ga_len; ++i)
|
|
|
3079 {
|
|
|
3080 #ifdef FEAT_MBYTE
|
|
|
3081 if (has_mbyte)
|
|
|
3082 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
3083 * the character, for other encodings it should differ for most
|
|
|
3084 * characters. */
|
|
|
3085 c = *smp[i].sm_lead_w & 0xff;
|
|
|
3086 else
|
|
|
3087 #endif
|
|
|
3088 c = *smp[i].sm_lead;
|
|
|
3089 if (sfirst[c] == -1)
|
|
|
3090 {
|
|
|
3091 sfirst[c] = i;
|
|
|
3092 #ifdef FEAT_MBYTE
|
|
|
3093 if (has_mbyte)
|
|
|
3094 {
|
|
|
3095 int n;
|
|
|
3096
|
|
|
3097 /* Make sure all entries with this byte are following each
|
|
|
3098 * other. Move the ones that are in the wrong position. Do
|
|
|
3099 * keep the same ordering! */
|
|
|
3100 while (i + 1 < gap->ga_len
|
|
|
3101 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
3102 /* Skip over entry with same index byte. */
|
|
|
3103 ++i;
|
|
|
3104
|
|
|
3105 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
3106 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
3107 {
|
|
|
3108 salitem_T tsal;
|
|
|
3109
|
|
|
3110 /* Move entry with same index byte after the entries
|
|
|
3111 * we already found. */
|
|
|
3112 ++i;
|
|
|
3113 --n;
|
|
|
3114 tsal = smp[i + n];
|
|
|
3115 mch_memmove(smp + i + 1, smp + i,
|
|
|
3116 sizeof(salitem_T) * n);
|
|
|
3117 smp[i] = tsal;
|
|
|
3118 }
|
|
|
3119 }
|
|
|
3120 #endif
|
|
|
3121 }
|
|
|
3122 }
|
|
|
3123 }
|
|
376
|
3124
|
|
372
|
3125 #ifdef FEAT_MBYTE
|
|
|
3126 /*
|
|
|
3127 * Turn a multi-byte string into a wide character string.
|
|
|
3128 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
3129 */
|
|
|
3130 static int *
|
|
|
3131 mb_str2wide(s)
|
|
|
3132 char_u *s;
|
|
|
3133 {
|
|
|
3134 int *res;
|
|
|
3135 char_u *p;
|
|
|
3136 int i = 0;
|
|
|
3137
|
|
|
3138 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
3139 if (res != NULL)
|
|
|
3140 {
|
|
|
3141 for (p = s; *p != NUL; )
|
|
|
3142 res[i++] = mb_ptr2char_adv(&p);
|
|
|
3143 res[i] = NUL;
|
|
|
3144 }
|
|
|
3145 return res;
|
|
|
3146 }
|
|
|
3147 #endif
|
|
|
3148
|
|
236
|
3149 /*
|
|
300
|
3150 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
3151 * "byts" and index array "idxs". Recursively read the children.
|
|
|
3152 *
|
|
346
|
3153 * NOTE: The code here must match put_node().
|
|
300
|
3154 *
|
|
|
3155 * Returns the index follosing the siblings.
|
|
|
3156 * Returns -1 if the file is shorter than expected.
|
|
|
3157 * Returns -2 if there is a format error.
|
|
236
|
3158 */
|
|
324
|
3159 static idx_T
|
|
339
|
3160 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
3161 FILE *fd;
|
|
|
3162 char_u *byts;
|
|
324
|
3163 idx_T *idxs;
|
|
300
|
3164 int maxidx; /* size of arrays */
|
|
324
|
3165 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
3166 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
3167 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
3168 {
|
|
300
|
3169 int len;
|
|
|
3170 int i;
|
|
|
3171 int n;
|
|
324
|
3172 idx_T idx = startidx;
|
|
300
|
3173 int c;
|
|
366
|
3174 int c2;
|
|
300
|
3175 #define SHARED_MASK 0x8000000
|
|
236
|
3176
|
|
300
|
3177 len = getc(fd); /* <siblingcount> */
|
|
|
3178 if (len <= 0)
|
|
|
3179 return -1;
|
|
|
3180
|
|
|
3181 if (startidx + len >= maxidx)
|
|
|
3182 return -2;
|
|
|
3183 byts[idx++] = len;
|
|
|
3184
|
|
|
3185 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
3186 for (i = 1; i <= len; ++i)
|
|
236
|
3187 {
|
|
300
|
3188 c = getc(fd); /* <byte> */
|
|
|
3189 if (c < 0)
|
|
|
3190 return -1;
|
|
|
3191 if (c <= BY_SPECIAL)
|
|
|
3192 {
|
|
366
|
3193 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
3194 {
|
|
|
3195 /* No flags, all regions. */
|
|
|
3196 idxs[idx] = 0;
|
|
|
3197 c = 0;
|
|
|
3198 }
|
|
390
|
3199 else if (c != BY_INDEX)
|
|
300
|
3200 {
|
|
339
|
3201 if (prefixtree)
|
|
|
3202 {
|
|
455
|
3203 /* Read the optional pflags byte, the prefix ID and the
|
|
|
3204 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
3205 * byte, the condition index shifted up 8 bits, the flags
|
|
|
3206 * shifted up 24 bits. */
|
|
|
3207 if (c == BY_FLAGS)
|
|
|
3208 c = getc(fd) << 24; /* <pflags> */
|
|
|
3209 else
|
|
|
3210 c = 0;
|
|
|
3211
|
|
481
|
3212 c |= getc(fd); /* <affixID> */
|
|
455
|
3213
|
|
339
|
3214 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
3215 if (n >= maxprefcondnr)
|
|
|
3216 return -2;
|
|
455
|
3217 c |= (n << 8);
|
|
339
|
3218 }
|
|
390
|
3219 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
3220 {
|
|
|
3221 /* Read flags and optional region and prefix ID. In
|
|
390
|
3222 * idxs[] the flags go in the low two bytes, region above
|
|
|
3223 * that and prefix ID above the region. */
|
|
|
3224 c2 = c;
|
|
339
|
3225 c = getc(fd); /* <flags> */
|
|
390
|
3226 if (c2 == BY_FLAGS2)
|
|
|
3227 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
3228 if (c & WF_REGION)
|
|
390
|
3229 c = (getc(fd) << 16) + c; /* <region> */
|
|
481
|
3230 if (c & WF_AFX)
|
|
|
3231 c = (getc(fd) << 24) + c; /* <affixID> */
|
|
339
|
3232 }
|
|
|
3233
|
|
300
|
3234 idxs[idx] = c;
|
|
|
3235 c = 0;
|
|
|
3236 }
|
|
|
3237 else /* c == BY_INDEX */
|
|
|
3238 {
|
|
|
3239 /* <nodeidx> */
|
|
|
3240 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
3241 if (n < 0 || n >= maxidx)
|
|
|
3242 return -2;
|
|
|
3243 idxs[idx] = n + SHARED_MASK;
|
|
|
3244 c = getc(fd); /* <xbyte> */
|
|
|
3245 }
|
|
|
3246 }
|
|
|
3247 byts[idx++] = c;
|
|
236
|
3248 }
|
|
|
3249
|
|
300
|
3250 /* Recursively read the children for non-shared siblings.
|
|
|
3251 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
3252 * remove SHARED_MASK) */
|
|
|
3253 for (i = 1; i <= len; ++i)
|
|
|
3254 if (byts[startidx + i] != 0)
|
|
|
3255 {
|
|
|
3256 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
3257 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
3258 else
|
|
|
3259 {
|
|
|
3260 idxs[startidx + i] = idx;
|
|
339
|
3261 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
3262 prefixtree, maxprefcondnr);
|
|
300
|
3263 if (idx < 0)
|
|
|
3264 break;
|
|
|
3265 }
|
|
|
3266 }
|
|
236
|
3267
|
|
300
|
3268 return idx;
|
|
236
|
3269 }
|
|
|
3270
|
|
|
3271 /*
|
|
|
3272 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
3273 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
3274 */
|
|
|
3275 char_u *
|
|
|
3276 did_set_spelllang(buf)
|
|
|
3277 buf_T *buf;
|
|
|
3278 {
|
|
|
3279 garray_T ga;
|
|
351
|
3280 char_u *splp;
|
|
236
|
3281 char_u *region;
|
|
416
|
3282 char_u region_cp[3];
|
|
355
|
3283 int filename;
|
|
236
|
3284 int region_mask;
|
|
|
3285 slang_T *lp;
|
|
|
3286 int c;
|
|
351
|
3287 char_u lang[MAXWLEN + 1];
|
|
323
|
3288 char_u spf_name[MAXPATHL];
|
|
351
|
3289 int len;
|
|
|
3290 char_u *p;
|
|
381
|
3291 int round;
|
|
385
|
3292 char_u *spf;
|
|
389
|
3293 char_u *use_region = NULL;
|
|
|
3294 int dont_use_region = FALSE;
|
|
236
|
3295
|
|
|
3296 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
3297 clear_midword(buf);
|
|
236
|
3298
|
|
351
|
3299 /* loop over comma separated language names. */
|
|
|
3300 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
3301 {
|
|
|
3302 /* Get one language name. */
|
|
|
3303 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
3304
|
|
240
|
3305 region = NULL;
|
|
351
|
3306 len = STRLEN(lang);
|
|
355
|
3307
|
|
|
3308 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
3309 * If there is a region name let "region" point to it and remove it
|
|
|
3310 * from the name. */
|
|
|
3311 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
3312 {
|
|
355
|
3313 filename = TRUE;
|
|
|
3314
|
|
416
|
3315 /* Locate a region and remove it from the file name. */
|
|
|
3316 p = vim_strchr(gettail(lang), '_');
|
|
|
3317 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
3318 && !ASCII_ISALPHA(p[3]))
|
|
|
3319 {
|
|
|
3320 vim_strncpy(region_cp, p + 1, 2);
|
|
|
3321 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
3322 len -= 3;
|
|
|
3323 region = region_cp;
|
|
|
3324 }
|
|
|
3325 else
|
|
|
3326 dont_use_region = TRUE;
|
|
|
3327
|
|
355
|
3328 /* Check if we loaded this language before. */
|
|
|
3329 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3330 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3331 break;
|
|
236
|
3332 }
|
|
355
|
3333 else
|
|
|
3334 {
|
|
|
3335 filename = FALSE;
|
|
|
3336 if (len > 3 && lang[len - 3] == '_')
|
|
|
3337 {
|
|
|
3338 region = lang + len - 2;
|
|
|
3339 len -= 3;
|
|
|
3340 lang[len] = NUL;
|
|
389
|
3341 }
|
|
|
3342 else
|
|
|
3343 dont_use_region = TRUE;
|
|
355
|
3344
|
|
|
3345 /* Check if we loaded this language before. */
|
|
|
3346 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3347 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
3348 break;
|
|
|
3349 }
|
|
236
|
3350
|
|
416
|
3351 if (region != NULL)
|
|
|
3352 {
|
|
|
3353 /* If the region differs from what was used before then don't
|
|
|
3354 * use it for 'spellfile'. */
|
|
|
3355 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
3356 dont_use_region = TRUE;
|
|
|
3357 use_region = region;
|
|
|
3358 }
|
|
|
3359
|
|
351
|
3360 /* If not found try loading the language now. */
|
|
236
|
3361 if (lp == NULL)
|
|
355
|
3362 {
|
|
|
3363 if (filename)
|
|
|
3364 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
3365 else
|
|
|
3366 spell_load_lang(lang);
|
|
|
3367 }
|
|
236
|
3368
|
|
307
|
3369 /*
|
|
351
|
3370 * Loop over the languages, there can be several files for "lang".
|
|
307
|
3371 */
|
|
|
3372 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
3373 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
3374 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
3375 {
|
|
316
|
3376 region_mask = REGION_ALL;
|
|
355
|
3377 if (!filename && region != NULL)
|
|
236
|
3378 {
|
|
307
|
3379 /* find region in sl_regions */
|
|
|
3380 c = find_region(lp->sl_regions, region);
|
|
|
3381 if (c == REGION_ALL)
|
|
|
3382 {
|
|
389
|
3383 if (lp->sl_add)
|
|
|
3384 {
|
|
|
3385 if (*lp->sl_regions != NUL)
|
|
|
3386 /* This addition file is for other regions. */
|
|
|
3387 region_mask = 0;
|
|
|
3388 }
|
|
|
3389 else
|
|
|
3390 /* This is probably an error. Give a warning and
|
|
|
3391 * accept the words anyway. */
|
|
351
|
3392 smsg((char_u *)
|
|
|
3393 _("Warning: region %s not supported"),
|
|
|
3394 region);
|
|
307
|
3395 }
|
|
|
3396 else
|
|
|
3397 region_mask = 1 << c;
|
|
236
|
3398 }
|
|
307
|
3399
|
|
389
|
3400 if (region_mask != 0)
|
|
307
|
3401 {
|
|
389
|
3402 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3403 {
|
|
|
3404 ga_clear(&ga);
|
|
|
3405 return e_outofmem;
|
|
|
3406 }
|
|
|
3407 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
3408 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3409 ++ga.ga_len;
|
|
|
3410 use_midword(lp, buf);
|
|
307
|
3411 }
|
|
385
|
3412 }
|
|
|
3413 }
|
|
|
3414
|
|
|
3415 /* round 0: load int_wordlist, if possible.
|
|
|
3416 * round 1: load first name in 'spellfile'.
|
|
|
3417 * round 2: load second name in 'spellfile.
|
|
|
3418 * etc. */
|
|
|
3419 spf = curbuf->b_p_spf;
|
|
|
3420 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
3421 {
|
|
|
3422 if (round == 0)
|
|
|
3423 {
|
|
|
3424 /* Internal wordlist, if there is one. */
|
|
|
3425 if (int_wordlist == NULL)
|
|
381
|
3426 continue;
|
|
385
|
3427 int_wordlist_spl(spf_name);
|
|
381
|
3428 }
|
|
|
3429 else
|
|
|
3430 {
|
|
385
|
3431 /* One entry in 'spellfile'. */
|
|
|
3432 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
3433 STRCAT(spf_name, ".spl");
|
|
|
3434
|
|
|
3435 /* If it was already found above then skip it. */
|
|
|
3436 for (c = 0; c < ga.ga_len; ++c)
|
|
|
3437 if (fullpathcmp(spf_name,
|
|
|
3438 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
3439 FALSE) == FPC_SAME)
|
|
|
3440 break;
|
|
|
3441 if (c < ga.ga_len)
|
|
381
|
3442 continue;
|
|
|
3443 }
|
|
|
3444
|
|
351
|
3445 /* Check if it was loaded already. */
|
|
323
|
3446 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3447 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3448 break;
|
|
|
3449 if (lp == NULL)
|
|
|
3450 {
|
|
351
|
3451 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
3452 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
3453 * use an arbitrary name. */
|
|
|
3454 if (round == 0)
|
|
|
3455 STRCPY(lang, "internal wordlist");
|
|
|
3456 else
|
|
|
3457 {
|
|
|
3458 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
3459 p = vim_strchr(lang, '.');
|
|
|
3460 if (p != NULL)
|
|
|
3461 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
3462 }
|
|
351
|
3463 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
3464 }
|
|
|
3465 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
3466 {
|
|
389
|
3467 region_mask = REGION_ALL;
|
|
|
3468 if (use_region != NULL && !dont_use_region)
|
|
|
3469 {
|
|
|
3470 /* find region in sl_regions */
|
|
|
3471 c = find_region(lp->sl_regions, use_region);
|
|
|
3472 if (c != REGION_ALL)
|
|
|
3473 region_mask = 1 << c;
|
|
|
3474 else if (*lp->sl_regions != NUL)
|
|
|
3475 /* This spell file is for other regions. */
|
|
|
3476 region_mask = 0;
|
|
|
3477 }
|
|
|
3478
|
|
|
3479 if (region_mask != 0)
|
|
|
3480 {
|
|
|
3481 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
3482 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3483 ++ga.ga_len;
|
|
|
3484 use_midword(lp, buf);
|
|
|
3485 }
|
|
323
|
3486 }
|
|
|
3487 }
|
|
|
3488
|
|
236
|
3489 /* Add a NULL entry to mark the end of the list. */
|
|
|
3490 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3491 {
|
|
|
3492 ga_clear(&ga);
|
|
|
3493 return e_outofmem;
|
|
|
3494 }
|
|
|
3495 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
3496 ++ga.ga_len;
|
|
|
3497
|
|
|
3498 /* Everything is fine, store the new b_langp value. */
|
|
|
3499 ga_clear(&buf->b_langp);
|
|
|
3500 buf->b_langp = ga;
|
|
|
3501
|
|
|
3502 return NULL;
|
|
|
3503 }
|
|
|
3504
|
|
|
3505 /*
|
|
376
|
3506 * Clear the midword characters for buffer "buf".
|
|
|
3507 */
|
|
|
3508 static void
|
|
|
3509 clear_midword(buf)
|
|
|
3510 buf_T *buf;
|
|
|
3511 {
|
|
|
3512 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
3513 #ifdef FEAT_MBYTE
|
|
|
3514 vim_free(buf->b_spell_ismw_mb);
|
|
|
3515 buf->b_spell_ismw_mb = NULL;
|
|
|
3516 #endif
|
|
|
3517 }
|
|
|
3518
|
|
|
3519 /*
|
|
|
3520 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
3521 * They add up to any currently used midword characters.
|
|
|
3522 */
|
|
|
3523 static void
|
|
|
3524 use_midword(lp, buf)
|
|
|
3525 slang_T *lp;
|
|
|
3526 buf_T *buf;
|
|
|
3527 {
|
|
|
3528 char_u *p;
|
|
|
3529
|
|
389
|
3530 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
3531 return;
|
|
|
3532
|
|
376
|
3533 for (p = lp->sl_midword; *p != NUL; )
|
|
|
3534 #ifdef FEAT_MBYTE
|
|
|
3535 if (has_mbyte)
|
|
|
3536 {
|
|
|
3537 int c, l, n;
|
|
|
3538 char_u *bp;
|
|
|
3539
|
|
|
3540 c = mb_ptr2char(p);
|
|
474
|
3541 l = (*mb_ptr2len)(p);
|
|
|
3542 if (c < 256 && l <= 2)
|
|
376
|
3543 buf->b_spell_ismw[c] = TRUE;
|
|
|
3544 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
3545 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
3546 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
3547 else
|
|
|
3548 {
|
|
|
3549 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
3550 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
3551 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
3552 if (bp != NULL)
|
|
|
3553 {
|
|
|
3554 vim_free(buf->b_spell_ismw_mb);
|
|
|
3555 buf->b_spell_ismw_mb = bp;
|
|
|
3556 vim_strncpy(bp + n, p, l);
|
|
|
3557 }
|
|
|
3558 }
|
|
|
3559 p += l;
|
|
|
3560 }
|
|
|
3561 else
|
|
|
3562 #endif
|
|
|
3563 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
3564 }
|
|
|
3565
|
|
|
3566 /*
|
|
236
|
3567 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
3568 * Each region is simply stored as the two characters of it's name.
|
|
381
|
3569 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
3570 */
|
|
|
3571 static int
|
|
|
3572 find_region(rp, region)
|
|
|
3573 char_u *rp;
|
|
|
3574 char_u *region;
|
|
|
3575 {
|
|
|
3576 int i;
|
|
|
3577
|
|
|
3578 for (i = 0; ; i += 2)
|
|
|
3579 {
|
|
|
3580 if (rp[i] == NUL)
|
|
|
3581 return REGION_ALL;
|
|
|
3582 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
3583 break;
|
|
|
3584 }
|
|
|
3585 return i / 2;
|
|
|
3586 }
|
|
|
3587
|
|
|
3588 /*
|
|
323
|
3589 * Return case type of word:
|
|
236
|
3590 * w word 0
|
|
300
|
3591 * Word WF_ONECAP
|
|
|
3592 * W WORD WF_ALLCAP
|
|
|
3593 * WoRd wOrd WF_KEEPCAP
|
|
236
|
3594 */
|
|
|
3595 static int
|
|
|
3596 captype(word, end)
|
|
|
3597 char_u *word;
|
|
323
|
3598 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
3599 {
|
|
|
3600 char_u *p;
|
|
|
3601 int c;
|
|
|
3602 int firstcap;
|
|
|
3603 int allcap;
|
|
|
3604 int past_second = FALSE; /* past second word char */
|
|
|
3605
|
|
|
3606 /* find first letter */
|
|
376
|
3607 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
3608 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
3609 return 0; /* only non-word characters, illegal word */
|
|
|
3610 #ifdef FEAT_MBYTE
|
|
310
|
3611 if (has_mbyte)
|
|
|
3612 c = mb_ptr2char_adv(&p);
|
|
|
3613 else
|
|
236
|
3614 #endif
|
|
310
|
3615 c = *p++;
|
|
324
|
3616 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
3617
|
|
|
3618 /*
|
|
|
3619 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
3620 * But a word with an upper char only at start is a ONECAP.
|
|
|
3621 */
|
|
323
|
3622 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
3623 if (spell_iswordp_nmw(p))
|
|
236
|
3624 {
|
|
455
|
3625 c = PTR2CHAR(p);
|
|
324
|
3626 if (!SPELL_ISUPPER(c))
|
|
236
|
3627 {
|
|
|
3628 /* UUl -> KEEPCAP */
|
|
|
3629 if (past_second && allcap)
|
|
300
|
3630 return WF_KEEPCAP;
|
|
236
|
3631 allcap = FALSE;
|
|
|
3632 }
|
|
|
3633 else if (!allcap)
|
|
|
3634 /* UlU -> KEEPCAP */
|
|
300
|
3635 return WF_KEEPCAP;
|
|
236
|
3636 past_second = TRUE;
|
|
|
3637 }
|
|
|
3638
|
|
|
3639 if (allcap)
|
|
300
|
3640 return WF_ALLCAP;
|
|
236
|
3641 if (firstcap)
|
|
300
|
3642 return WF_ONECAP;
|
|
236
|
3643 return 0;
|
|
|
3644 }
|
|
|
3645
|
|
474
|
3646 /*
|
|
|
3647 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
3648 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
3649 * Add ALLCAP for "WOrD".
|
|
|
3650 */
|
|
|
3651 static int
|
|
|
3652 badword_captype(word, end)
|
|
|
3653 char_u *word;
|
|
|
3654 char_u *end;
|
|
|
3655 {
|
|
|
3656 int flags = captype(word, end);
|
|
475
|
3657 int c;
|
|
474
|
3658 int l, u;
|
|
|
3659 int first;
|
|
|
3660 char_u *p;
|
|
|
3661
|
|
|
3662 if (flags & WF_KEEPCAP)
|
|
|
3663 {
|
|
|
3664 /* Count the number of UPPER and lower case letters. */
|
|
|
3665 l = u = 0;
|
|
|
3666 first = FALSE;
|
|
|
3667 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
3668 {
|
|
475
|
3669 c = PTR2CHAR(p);
|
|
|
3670 if (SPELL_ISUPPER(c))
|
|
474
|
3671 {
|
|
|
3672 ++u;
|
|
|
3673 if (p == word)
|
|
|
3674 first = TRUE;
|
|
|
3675 }
|
|
|
3676 else
|
|
|
3677 ++l;
|
|
|
3678 }
|
|
|
3679
|
|
|
3680 /* If there are more UPPER than lower case letters suggest an
|
|
|
3681 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
3682 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
3683 * require three upper case letters. */
|
|
|
3684 if (u > l && u > 2)
|
|
|
3685 flags |= WF_ALLCAP;
|
|
|
3686 else if (first)
|
|
|
3687 flags |= WF_ONECAP;
|
|
|
3688 }
|
|
|
3689 return flags;
|
|
|
3690 }
|
|
|
3691
|
|
355
|
3692 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
3693 /*
|
|
|
3694 * Free all languages.
|
|
|
3695 */
|
|
|
3696 void
|
|
|
3697 spell_free_all()
|
|
|
3698 {
|
|
|
3699 slang_T *lp;
|
|
|
3700 buf_T *buf;
|
|
385
|
3701 char_u fname[MAXPATHL];
|
|
355
|
3702
|
|
|
3703 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3704 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3705 ga_clear(&buf->b_langp);
|
|
|
3706
|
|
|
3707 while (first_lang != NULL)
|
|
|
3708 {
|
|
|
3709 lp = first_lang;
|
|
|
3710 first_lang = lp->sl_next;
|
|
|
3711 slang_free(lp);
|
|
|
3712 }
|
|
366
|
3713
|
|
385
|
3714 if (int_wordlist != NULL)
|
|
|
3715 {
|
|
|
3716 /* Delete the internal wordlist and its .spl file */
|
|
|
3717 mch_remove(int_wordlist);
|
|
|
3718 int_wordlist_spl(fname);
|
|
|
3719 mch_remove(fname);
|
|
|
3720 vim_free(int_wordlist);
|
|
|
3721 int_wordlist = NULL;
|
|
381
|
3722 }
|
|
|
3723
|
|
366
|
3724 init_spell_chartab();
|
|
355
|
3725 }
|
|
|
3726 # endif
|
|
|
3727
|
|
236
|
3728 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
3729 /*
|
|
|
3730 * Clear all spelling tables and reload them.
|
|
307
|
3731 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
3732 */
|
|
|
3733 void
|
|
|
3734 spell_reload()
|
|
|
3735 {
|
|
|
3736 buf_T *buf;
|
|
316
|
3737 win_T *wp;
|
|
236
|
3738
|
|
358
|
3739 /* Initialize the table for spell_iswordp(). */
|
|
236
|
3740 init_spell_chartab();
|
|
|
3741
|
|
|
3742 /* Unload all allocated memory. */
|
|
355
|
3743 spell_free_all();
|
|
236
|
3744
|
|
|
3745 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3746 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3747 {
|
|
316
|
3748 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
3749 * window for this buffer in which 'spell' is set. */
|
|
236
|
3750 if (*buf->b_p_spl != NUL)
|
|
316
|
3751 {
|
|
|
3752 FOR_ALL_WINDOWS(wp)
|
|
|
3753 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
3754 {
|
|
|
3755 (void)did_set_spelllang(buf);
|
|
|
3756 # ifdef FEAT_WINDOWS
|
|
|
3757 break;
|
|
|
3758 # endif
|
|
|
3759 }
|
|
|
3760 }
|
|
236
|
3761 }
|
|
|
3762 }
|
|
|
3763 # endif
|
|
|
3764
|
|
310
|
3765 /*
|
|
|
3766 * Reload the spell file "fname" if it's loaded.
|
|
|
3767 */
|
|
|
3768 static void
|
|
323
|
3769 spell_reload_one(fname, added_word)
|
|
310
|
3770 char_u *fname;
|
|
323
|
3771 int added_word; /* invoked through "zg" */
|
|
310
|
3772 {
|
|
|
3773 slang_T *lp;
|
|
323
|
3774 int didit = FALSE;
|
|
310
|
3775
|
|
|
3776 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3777 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3778 {
|
|
|
3779 slang_clear(lp);
|
|
323
|
3780 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
3781 redraw_all_later(NOT_VALID);
|
|
323
|
3782 didit = TRUE;
|
|
310
|
3783 }
|
|
323
|
3784
|
|
|
3785 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
3786 * 'spelllang' to get it loaded. */
|
|
|
3787 if (added_word && !didit)
|
|
|
3788 did_set_spelllang(curbuf);
|
|
310
|
3789 }
|
|
|
3790
|
|
|
3791
|
|
236
|
3792 /*
|
|
|
3793 * Functions for ":mkspell".
|
|
|
3794 */
|
|
|
3795
|
|
300
|
3796 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
3797 and .dic file. */
|
|
|
3798 /*
|
|
|
3799 * Main structure to store the contents of a ".aff" file.
|
|
|
3800 */
|
|
|
3801 typedef struct afffile_S
|
|
|
3802 {
|
|
|
3803 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
481
|
3804 int af_slash; /* character used in word for slash */
|
|
310
|
3805 int af_rar; /* RAR ID for rare word */
|
|
|
3806 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
3807 int af_bad; /* BAD ID for banned word */
|
|
483
|
3808 int af_needaffix; /* NEEDAFFIX ID */
|
|
|
3809 char_u *af_compflags; /* COMPOUNDFLAG and COMPOUNDFLAGS concat'ed */
|
|
|
3810 int af_compmax; /* COMPOUNDMAX */
|
|
481
|
3811 int af_compminlen; /* COMPOUNDMIN */
|
|
483
|
3812 int af_compsylmax; /* COMPOUNDSYLMAX */
|
|
|
3813 char_u *af_syllable; /* SYLLABLE */
|
|
339
|
3814 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
3815 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
3816 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
3817 } afffile_T;
|
|
|
3818
|
|
|
3819 typedef struct affentry_S affentry_T;
|
|
|
3820 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
3821 struct affentry_S
|
|
|
3822 {
|
|
|
3823 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
3824 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3825 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3826 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3827 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
483
|
3828 char_u ae_rare; /* rare affix */
|
|
|
3829 char_u ae_nocomp; /* word with affix not compoundable */
|
|
300
|
3830 };
|
|
|
3831
|
|
455
|
3832 #define AH_KEY_LEN 10
|
|
|
3833
|
|
300
|
3834 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
3835 typedef struct affheader_S
|
|
|
3836 {
|
|
455
|
3837 /* key for hashtable == name of affix entry */
|
|
|
3838 #ifdef FEAT_MBYTE
|
|
|
3839 char_u ah_key[AH_KEY_LEN]; /* multi-byte char plus NUL */
|
|
|
3840 #else
|
|
|
3841 char_u ah_key[2]; /* one byte char plus NUL */
|
|
|
3842 #endif
|
|
339
|
3843 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
3844 int ah_combine; /* suffix may combine with prefix */
|
|
|
3845 affentry_T *ah_first; /* first affix entry */
|
|
|
3846 } affheader_T;
|
|
|
3847
|
|
|
3848 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
3849
|
|
|
3850 /*
|
|
|
3851 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
3852 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
3853 * once after ":mkspell" is done.
|
|
300
|
3854 */
|
|
|
3855 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
3856 typedef struct sblock_S sblock_T;
|
|
|
3857 struct sblock_S
|
|
|
3858 {
|
|
|
3859 sblock_T *sb_next; /* next block in list */
|
|
|
3860 int sb_used; /* nr of bytes already in use */
|
|
|
3861 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
3862 };
|
|
|
3863
|
|
|
3864 /*
|
|
300
|
3865 * A node in the tree.
|
|
236
|
3866 */
|
|
300
|
3867 typedef struct wordnode_S wordnode_T;
|
|
|
3868 struct wordnode_S
|
|
236
|
3869 {
|
|
346
|
3870 union /* shared to save space */
|
|
|
3871 {
|
|
469
|
3872 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
3873 int index; /* index in written nodes (valid after first
|
|
|
3874 round) */
|
|
|
3875 } wn_u1;
|
|
|
3876 union /* shared to save space */
|
|
|
3877 {
|
|
|
3878 wordnode_T *next; /* next node with same hash key */
|
|
|
3879 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
3880 } wn_u2;
|
|
300
|
3881 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
3882 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
3883 always sorted) */
|
|
474
|
3884 int wn_refs; /* Nr. of references to this node. Only
|
|
|
3885 relevant for first node in a list of
|
|
|
3886 siblings, in following siblings it is
|
|
|
3887 always one. */
|
|
300
|
3888 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
481
|
3889 char_u wn_affixID; /* when "wn_byte" is NUL: supported/required
|
|
469
|
3890 prefix ID or 0 */
|
|
|
3891 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
3892 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
3893 PREFIXTREE it's the prefcondnr */
|
|
469
|
3894 #ifdef SPELL_PRINTTREE
|
|
|
3895 int wn_nr; /* sequence nr for printing */
|
|
|
3896 #endif
|
|
236
|
3897 };
|
|
|
3898
|
|
390
|
3899 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
3900
|
|
300
|
3901 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
3902
|
|
300
|
3903 /*
|
|
|
3904 * Info used while reading the spell files.
|
|
|
3905 */
|
|
|
3906 typedef struct spellinfo_S
|
|
249
|
3907 {
|
|
300
|
3908 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
3909 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
3910
|
|
300
|
3911 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
3912 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
3913
|
|
339
|
3914 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
3915
|
|
300
|
3916 sblock_T *si_blocks; /* memory blocks used */
|
|
482
|
3917 long si_blocks_cnt; /* memory blocks allocated */
|
|
|
3918 long si_compress_cnt; /* words to add before lowering
|
|
|
3919 compression limit */
|
|
474
|
3920 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
3921 compression, linked by "wn_child" field. */
|
|
482
|
3922 long si_free_count; /* number of nodes in si_first_free */
|
|
474
|
3923 #ifdef SPELL_PRINTTREE
|
|
|
3924 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
3925 #endif
|
|
|
3926
|
|
|
3927
|
|
300
|
3928 int si_ascii; /* handling only ASCII words */
|
|
310
|
3929 int si_add; /* addition file */
|
|
351
|
3930 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
3931 int si_region; /* region mask */
|
|
|
3932 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
3933 int si_memtot; /* runtime memory used */
|
|
310
|
3934 int si_verbose; /* verbose messages */
|
|
474
|
3935 int si_msg_count; /* number of words added since last message */
|
|
316
|
3936 int si_region_count; /* number of regions supported (1 when there
|
|
|
3937 are no regions) */
|
|
483
|
3938 char_u si_region_name[16]; /* region names; used only if
|
|
|
3939 * si_region_count > 1) */
|
|
323
|
3940
|
|
|
3941 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
3942 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
3943 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
3944 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
3945 int si_followup; /* soundsalike: ? */
|
|
|
3946 int si_collapse; /* soundsalike: ? */
|
|
|
3947 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
3948 garray_T si_map; /* MAP info concatenated */
|
|
366
|
3949 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
483
|
3950 int si_compmax; /* max nr of words for compounding */
|
|
481
|
3951 int si_compminlen; /* minimal length for compounding */
|
|
483
|
3952 int si_compsylmax; /* max nr of syllables for compounding */
|
|
481
|
3953 char_u *si_compflags; /* flags used for compounding */
|
|
483
|
3954 char_u *si_syllable; /* syllable string */
|
|
339
|
3955 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
3956 * prefixes, each stored as a string */
|
|
|
3957 int si_newID; /* current value for ah_newID */
|
|
300
|
3958 } spellinfo_T;
|
|
249
|
3959
|
|
474
|
3960 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
339
|
3961 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
3962 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
3963 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
3964 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
3965 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
3966 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
483
|
3967 static int get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
|
3968 static void get_compflags __ARGS((spellinfo_T *spin, char_u *afflist, char_u *store_afflist));
|
|
|
3969 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
|
3970 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
3971 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
3972 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
3973 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
3974 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
483
|
3975 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist, int need_affix));
|
|
481
|
3976 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID));
|
|
474
|
3977 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
|
3978 static void deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
|
3979 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
3980 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
3981 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
3982 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
474
|
3983 static void write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
3984 static void clear_node __ARGS((wordnode_T *node));
|
|
|
3985 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
3986 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
3987 static void init_spellfile __ARGS((void));
|
|
236
|
3988
|
|
455
|
3989 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
3990 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
3991 * Use a negative number with the lower 8 bits zero. */
|
|
|
3992 #define PFX_FLAGS -256
|
|
390
|
3993
|
|
483
|
3994 /*
|
|
|
3995 * Tunable parameters for when the tree is compressed. See 'mkspellmem'.
|
|
|
3996 */
|
|
|
3997 static long compress_start = 30000; /* memory / SBLOCKSIZE */
|
|
|
3998 static long compress_inc = 100; /* memory / SBLOCKSIZE */
|
|
|
3999 static long compress_added = 500000; /* word count */
|
|
|
4000
|
|
469
|
4001 #ifdef SPELL_PRINTTREE
|
|
|
4002 /*
|
|
|
4003 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
4004 * readable format, so that we can see what happens when adding a word and/or
|
|
|
4005 * compressing the tree.
|
|
|
4006 * Based on code from Olaf Seibert.
|
|
|
4007 */
|
|
|
4008 #define PRINTLINESIZE 1000
|
|
|
4009 #define PRINTWIDTH 6
|
|
|
4010
|
|
|
4011 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
4012 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
4013
|
|
|
4014 static char line1[PRINTLINESIZE];
|
|
|
4015 static char line2[PRINTLINESIZE];
|
|
|
4016 static char line3[PRINTLINESIZE];
|
|
|
4017
|
|
|
4018 static void
|
|
|
4019 spell_clear_flags(wordnode_T *node)
|
|
|
4020 {
|
|
|
4021 wordnode_T *np;
|
|
|
4022
|
|
|
4023 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4024 {
|
|
|
4025 np->wn_u1.index = FALSE;
|
|
|
4026 spell_clear_flags(np->wn_child);
|
|
|
4027 }
|
|
|
4028 }
|
|
|
4029
|
|
|
4030 static void
|
|
|
4031 spell_print_node(wordnode_T *node, int depth)
|
|
|
4032 {
|
|
|
4033 if (node->wn_u1.index)
|
|
|
4034 {
|
|
|
4035 /* Done this node before, print the reference. */
|
|
|
4036 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
4037 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
4038 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4039 msg(line1);
|
|
|
4040 msg(line2);
|
|
|
4041 msg(line3);
|
|
|
4042 }
|
|
|
4043 else
|
|
|
4044 {
|
|
|
4045 node->wn_u1.index = TRUE;
|
|
|
4046
|
|
|
4047 if (node->wn_byte != NUL)
|
|
|
4048 {
|
|
|
4049 if (node->wn_child != NULL)
|
|
|
4050 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
4051 else
|
|
|
4052 /* Cannot happen? */
|
|
|
4053 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
4054 }
|
|
|
4055 else
|
|
|
4056 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
4057
|
|
|
4058 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
4059
|
|
|
4060 if (node->wn_sibling != NULL)
|
|
|
4061 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
4062 else
|
|
|
4063 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4064
|
|
|
4065 if (node->wn_byte == NUL)
|
|
|
4066 {
|
|
|
4067 msg(line1);
|
|
|
4068 msg(line2);
|
|
|
4069 msg(line3);
|
|
|
4070 }
|
|
|
4071
|
|
|
4072 /* do the children */
|
|
|
4073 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
4074 spell_print_node(node->wn_child, depth + 1);
|
|
|
4075
|
|
|
4076 /* do the siblings */
|
|
|
4077 if (node->wn_sibling != NULL)
|
|
|
4078 {
|
|
|
4079 /* get rid of all parent details except | */
|
|
|
4080 STRCPY(line1, line3);
|
|
|
4081 STRCPY(line2, line3);
|
|
|
4082 spell_print_node(node->wn_sibling, depth);
|
|
|
4083 }
|
|
|
4084 }
|
|
|
4085 }
|
|
|
4086
|
|
|
4087 static void
|
|
|
4088 spell_print_tree(wordnode_T *root)
|
|
|
4089 {
|
|
|
4090 if (root != NULL)
|
|
|
4091 {
|
|
|
4092 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
4093 * done. */
|
|
|
4094 spell_clear_flags(root);
|
|
|
4095
|
|
|
4096 /* Recursively print the tree. */
|
|
|
4097 spell_print_node(root, 0);
|
|
|
4098 }
|
|
|
4099 }
|
|
|
4100 #endif /* SPELL_PRINTTREE */
|
|
|
4101
|
|
236
|
4102 /*
|
|
323
|
4103 * Read the affix file "fname".
|
|
316
|
4104 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
4105 */
|
|
|
4106 static afffile_T *
|
|
474
|
4107 spell_read_aff(spin, fname)
|
|
|
4108 spellinfo_T *spin;
|
|
236
|
4109 char_u *fname;
|
|
|
4110 {
|
|
|
4111 FILE *fd;
|
|
|
4112 afffile_T *aff;
|
|
|
4113 char_u rline[MAXLINELEN];
|
|
|
4114 char_u *line;
|
|
|
4115 char_u *pc = NULL;
|
|
334
|
4116 #define MAXITEMCNT 7
|
|
|
4117 char_u *(items[MAXITEMCNT]);
|
|
236
|
4118 int itemcnt;
|
|
|
4119 char_u *p;
|
|
|
4120 int lnum = 0;
|
|
|
4121 affheader_T *cur_aff = NULL;
|
|
|
4122 int aff_todo = 0;
|
|
|
4123 hashtab_T *tp;
|
|
255
|
4124 char_u *low = NULL;
|
|
|
4125 char_u *fol = NULL;
|
|
|
4126 char_u *upp = NULL;
|
|
307
|
4127 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
4128 int do_rep;
|
|
|
4129 int do_sal;
|
|
|
4130 int do_map;
|
|
366
|
4131 int do_midword;
|
|
375
|
4132 int do_sofo;
|
|
323
|
4133 int found_map = FALSE;
|
|
324
|
4134 hashitem_T *hi;
|
|
455
|
4135 int l;
|
|
236
|
4136
|
|
300
|
4137 /*
|
|
|
4138 * Open the file.
|
|
|
4139 */
|
|
310
|
4140 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4141 if (fd == NULL)
|
|
|
4142 {
|
|
|
4143 EMSG2(_(e_notopen), fname);
|
|
|
4144 return NULL;
|
|
|
4145 }
|
|
|
4146
|
|
310
|
4147 if (spin->si_verbose || p_verbose > 2)
|
|
|
4148 {
|
|
|
4149 if (!spin->si_verbose)
|
|
|
4150 verbose_enter();
|
|
366
|
4151 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
4152 out_flush();
|
|
|
4153 if (!spin->si_verbose)
|
|
|
4154 verbose_leave();
|
|
|
4155 }
|
|
236
|
4156
|
|
323
|
4157 /* Only do REP lines when not done in another .aff file already. */
|
|
|
4158 do_rep = spin->si_rep.ga_len == 0;
|
|
|
4159
|
|
|
4160 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
4161 do_sal = spin->si_sal.ga_len == 0;
|
|
|
4162
|
|
|
4163 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
4164 do_map = spin->si_map.ga_len == 0;
|
|
|
4165
|
|
366
|
4166 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
4167 do_midword = spin->si_midword == NULL;
|
|
|
4168
|
|
375
|
4169 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
4170 do_sofo = spin->si_sofofr == NULL;
|
|
|
4171
|
|
300
|
4172 /*
|
|
|
4173 * Allocate and init the afffile_T structure.
|
|
|
4174 */
|
|
474
|
4175 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
4176 if (aff == NULL)
|
|
|
4177 return NULL;
|
|
|
4178 hash_init(&aff->af_pref);
|
|
|
4179 hash_init(&aff->af_suff);
|
|
|
4180
|
|
|
4181 /*
|
|
|
4182 * Read all the lines in the file one by one.
|
|
|
4183 */
|
|
255
|
4184 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
4185 {
|
|
255
|
4186 line_breakcheck();
|
|
236
|
4187 ++lnum;
|
|
|
4188
|
|
|
4189 /* Skip comment lines. */
|
|
|
4190 if (*rline == '#')
|
|
|
4191 continue;
|
|
|
4192
|
|
|
4193 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
4194 vim_free(pc);
|
|
310
|
4195 #ifdef FEAT_MBYTE
|
|
300
|
4196 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4197 {
|
|
300
|
4198 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
4199 if (pc == NULL)
|
|
|
4200 {
|
|
|
4201 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4202 fname, lnum, rline);
|
|
|
4203 continue;
|
|
|
4204 }
|
|
236
|
4205 line = pc;
|
|
|
4206 }
|
|
|
4207 else
|
|
310
|
4208 #endif
|
|
236
|
4209 {
|
|
|
4210 pc = NULL;
|
|
|
4211 line = rline;
|
|
|
4212 }
|
|
|
4213
|
|
|
4214 /* Split the line up in white separated items. Put a NUL after each
|
|
|
4215 * item. */
|
|
|
4216 itemcnt = 0;
|
|
|
4217 for (p = line; ; )
|
|
|
4218 {
|
|
|
4219 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
4220 ++p;
|
|
|
4221 if (*p == NUL)
|
|
|
4222 break;
|
|
334
|
4223 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
4224 break;
|
|
236
|
4225 items[itemcnt++] = p;
|
|
300
|
4226 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
4227 ++p;
|
|
|
4228 if (*p == NUL)
|
|
|
4229 break;
|
|
|
4230 *p++ = NUL;
|
|
|
4231 }
|
|
|
4232
|
|
|
4233 /* Handle non-empty lines. */
|
|
|
4234 if (itemcnt > 0)
|
|
|
4235 {
|
|
|
4236 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
4237 && aff->af_enc == NULL)
|
|
|
4238 {
|
|
310
|
4239 #ifdef FEAT_MBYTE
|
|
300
|
4240 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
4241 aff->af_enc = enc_canonize(items[1]);
|
|
|
4242 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
4243 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
4244 p_enc) == FAIL)
|
|
|
4245 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
4246 fname, aff->af_enc, p_enc);
|
|
375
|
4247 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4248 #else
|
|
|
4249 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4250 #endif
|
|
236
|
4251 }
|
|
366
|
4252 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
4253 {
|
|
|
4254 if (do_midword)
|
|
|
4255 spin->si_midword = vim_strsave(items[1]);
|
|
|
4256 }
|
|
302
|
4257 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
4258 {
|
|
323
|
4259 /* ignored, we always split */
|
|
302
|
4260 }
|
|
323
|
4261 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
4262 {
|
|
323
|
4263 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
4264 }
|
|
481
|
4265 else if (STRCMP(items[0], "SLASH") == 0 && itemcnt == 2
|
|
|
4266 && aff->af_slash == 0)
|
|
|
4267 {
|
|
|
4268 aff->af_slash = items[1][0];
|
|
|
4269 if (items[1][1] != NUL)
|
|
|
4270 smsg((char_u *)_("Character used for SLASH must be ASCII; in %s line %d: %s"),
|
|
|
4271 fname, lnum, items[1]);
|
|
|
4272 }
|
|
307
|
4273 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
4274 && aff->af_rar == 0)
|
|
|
4275 {
|
|
|
4276 aff->af_rar = items[1][0];
|
|
|
4277 if (items[1][1] != NUL)
|
|
|
4278 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4279 }
|
|
310
|
4280 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
4281 && aff->af_kep == 0)
|
|
307
|
4282 {
|
|
310
|
4283 aff->af_kep = items[1][0];
|
|
307
|
4284 if (items[1][1] != NUL)
|
|
|
4285 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4286 }
|
|
346
|
4287 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
4288 && aff->af_bad == 0)
|
|
|
4289 {
|
|
|
4290 aff->af_bad = items[1][0];
|
|
|
4291 if (items[1][1] != NUL)
|
|
|
4292 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4293 }
|
|
483
|
4294 else if (STRCMP(items[0], "NEEDAFFIX") == 0 && itemcnt == 2
|
|
|
4295 && aff->af_needaffix == 0)
|
|
|
4296 {
|
|
|
4297 aff->af_needaffix = items[1][0];
|
|
|
4298 if (items[1][1] != NUL)
|
|
|
4299 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4300 }
|
|
481
|
4301 else if (STRCMP(items[0], "COMPOUNDFLAG") == 0 && itemcnt == 2
|
|
483
|
4302 && aff->af_compflags == NULL)
|
|
|
4303 {
|
|
|
4304 p = getroom(spin, 3, FALSE);
|
|
|
4305 if (p != NULL)
|
|
|
4306 {
|
|
|
4307 /* Turn single flag "c" into COMPOUNDFLAGS compatible
|
|
|
4308 * string "c+". */
|
|
|
4309 p[0] = items[1][0];
|
|
|
4310 p[1] = '+';
|
|
|
4311 p[2] = NUL;
|
|
|
4312 aff->af_compflags = p;
|
|
|
4313 }
|
|
481
|
4314 if (items[1][1] != NUL)
|
|
|
4315 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4316 }
|
|
483
|
4317 else if (STRCMP(items[0], "COMPOUNDFLAGS") == 0 && itemcnt == 2)
|
|
|
4318 {
|
|
|
4319 /* Concatenate this string to previously defined ones, using a
|
|
|
4320 * slash to separate them. */
|
|
|
4321 l = STRLEN(items[1]) + 1;
|
|
|
4322 if (aff->af_compflags != NULL)
|
|
|
4323 l += STRLEN(aff->af_compflags) + 1;
|
|
|
4324 p = getroom(spin, l, FALSE);
|
|
|
4325 if (p != NULL)
|
|
|
4326 {
|
|
|
4327 if (aff->af_compflags != NULL)
|
|
|
4328 {
|
|
|
4329 STRCPY(p, aff->af_compflags);
|
|
|
4330 STRCAT(p, "/");
|
|
|
4331 }
|
|
|
4332 STRCAT(p, items[1]);
|
|
|
4333 aff->af_compflags = p;
|
|
|
4334 }
|
|
|
4335 }
|
|
|
4336 else if (STRCMP(items[0], "COMPOUNDMAX") == 0 && itemcnt == 2
|
|
|
4337 && aff->af_compmax == 0)
|
|
|
4338 {
|
|
|
4339 aff->af_compmax = atoi((char *)items[1]);
|
|
|
4340 if (aff->af_compmax == 0)
|
|
|
4341 smsg((char_u *)_("Wrong COMPOUNDMAX value in %s line %d: %s"),
|
|
|
4342 fname, lnum, items[1]);
|
|
481
|
4343 }
|
|
|
4344 else if (STRCMP(items[0], "COMPOUNDMIN") == 0 && itemcnt == 2
|
|
|
4345 && aff->af_compminlen == 0)
|
|
|
4346 {
|
|
|
4347 aff->af_compminlen = atoi((char *)items[1]);
|
|
|
4348 if (aff->af_compminlen == 0)
|
|
|
4349 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"),
|
|
|
4350 fname, lnum, items[1]);
|
|
|
4351 }
|
|
483
|
4352 else if (STRCMP(items[0], "COMPOUNDSYLMAX") == 0 && itemcnt == 2
|
|
|
4353 && aff->af_compsylmax == 0)
|
|
|
4354 {
|
|
|
4355 aff->af_compsylmax = atoi((char *)items[1]);
|
|
|
4356 if (aff->af_compsylmax == 0)
|
|
|
4357 smsg((char_u *)_("Wrong COMPOUNDSYLMAX value in %s line %d: %s"),
|
|
|
4358 fname, lnum, items[1]);
|
|
|
4359 }
|
|
|
4360 else if (STRCMP(items[0], "SYLLABLE") == 0 && itemcnt == 2
|
|
|
4361 && aff->af_syllable == NULL)
|
|
|
4362 {
|
|
|
4363 aff->af_syllable = getroom_save(spin, items[1]);
|
|
|
4364 }
|
|
339
|
4365 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
4366 {
|
|
|
4367 aff->af_pfxpostpone = TRUE;
|
|
|
4368 }
|
|
236
|
4369 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4370 || STRCMP(items[0], "SFX") == 0)
|
|
|
4371 && aff_todo == 0
|
|
334
|
4372 && itemcnt >= 4)
|
|
236
|
4373 {
|
|
334
|
4374 /* Myspell allows extra text after the item, but that might
|
|
|
4375 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
4376 if (itemcnt > 4 && *items[4] != '#')
|
|
|
4377 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
4378 fname, lnum, items[4]);
|
|
|
4379
|
|
236
|
4380 /* New affix letter. */
|
|
474
|
4381 cur_aff = (affheader_T *)getroom(spin,
|
|
447
|
4382 sizeof(affheader_T), TRUE);
|
|
236
|
4383 if (cur_aff == NULL)
|
|
|
4384 break;
|
|
455
|
4385 #ifdef FEAT_MBYTE
|
|
|
4386 if (has_mbyte)
|
|
|
4387 {
|
|
474
|
4388 l = (*mb_ptr2len)(items[1]);
|
|
455
|
4389 if (l >= AH_KEY_LEN)
|
|
|
4390 l = 1; /* too long, must be an overlong sequence */
|
|
|
4391 else
|
|
|
4392 mch_memmove(cur_aff->ah_key, items[1], l);
|
|
|
4393 }
|
|
|
4394 else
|
|
|
4395 #endif
|
|
|
4396 {
|
|
|
4397 *cur_aff->ah_key = *items[1];
|
|
|
4398 l = 1;
|
|
|
4399 }
|
|
|
4400 cur_aff->ah_key[l] = NUL;
|
|
|
4401 if (items[1][l] != NUL)
|
|
307
|
4402 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
4403 if (*items[2] == 'Y')
|
|
|
4404 cur_aff->ah_combine = TRUE;
|
|
300
|
4405 else if (*items[2] != 'N')
|
|
236
|
4406 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
4407 fname, lnum, items[2]);
|
|
339
|
4408
|
|
236
|
4409 if (*items[0] == 'P')
|
|
339
|
4410 {
|
|
236
|
4411 tp = &aff->af_pref;
|
|
339
|
4412 /* Use a new number in the .spl file later, to be able to
|
|
|
4413 * handle multiple .aff files. */
|
|
|
4414 if (aff->af_pfxpostpone)
|
|
344
|
4415 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
4416 }
|
|
236
|
4417 else
|
|
|
4418 tp = &aff->af_suff;
|
|
300
|
4419 aff_todo = atoi((char *)items[3]);
|
|
324
|
4420 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
4421 if (!HASHITEM_EMPTY(hi))
|
|
300
|
4422 {
|
|
236
|
4423 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
4424 fname, lnum, items[1]);
|
|
300
|
4425 aff_todo = 0;
|
|
|
4426 }
|
|
236
|
4427 else
|
|
|
4428 hash_add(tp, cur_aff->ah_key);
|
|
|
4429 }
|
|
|
4430 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4431 || STRCMP(items[0], "SFX") == 0)
|
|
|
4432 && aff_todo > 0
|
|
|
4433 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
4434 && itemcnt >= 5)
|
|
236
|
4435 {
|
|
|
4436 affentry_T *aff_entry;
|
|
366
|
4437 int rare = FALSE;
|
|
483
|
4438 int nocomp = FALSE;
|
|
455
|
4439 int upper = FALSE;
|
|
366
|
4440 int lasti = 5;
|
|
|
4441
|
|
483
|
4442 /* Check for "rare" and "nocomp" after the other info. */
|
|
|
4443 while (itemcnt > lasti)
|
|
|
4444 {
|
|
|
4445 if (!rare && STRICMP(items[lasti], "rare") == 0)
|
|
|
4446 {
|
|
|
4447 rare = TRUE;
|
|
|
4448 ++lasti;
|
|
|
4449 }
|
|
|
4450 else if (!nocomp && STRICMP(items[lasti], "nocomp") == 0)
|
|
|
4451 {
|
|
|
4452 nocomp = TRUE;
|
|
|
4453 ++lasti;
|
|
|
4454 }
|
|
|
4455 else
|
|
|
4456 break;
|
|
366
|
4457 }
|
|
236
|
4458
|
|
334
|
4459 /* Myspell allows extra text after the item, but that might
|
|
|
4460 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
4461 if (itemcnt > lasti && *items[lasti] != '#')
|
|
482
|
4462 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]);
|
|
334
|
4463
|
|
236
|
4464 /* New item for an affix letter. */
|
|
|
4465 --aff_todo;
|
|
474
|
4466 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
4467 sizeof(affentry_T), TRUE);
|
|
236
|
4468 if (aff_entry == NULL)
|
|
|
4469 break;
|
|
366
|
4470 aff_entry->ae_rare = rare;
|
|
483
|
4471 aff_entry->ae_nocomp = nocomp;
|
|
240
|
4472
|
|
236
|
4473 if (STRCMP(items[2], "0") != 0)
|
|
474
|
4474 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
4475 if (STRCMP(items[3], "0") != 0)
|
|
474
|
4476 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
4477
|
|
300
|
4478 /* Don't use an affix entry with non-ASCII characters when
|
|
|
4479 * "spin->si_ascii" is TRUE. */
|
|
|
4480 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
4481 || has_non_ascii(aff_entry->ae_add)))
|
|
|
4482 {
|
|
|
4483 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
4484 cur_aff->ah_first = aff_entry;
|
|
300
|
4485
|
|
|
4486 if (STRCMP(items[4], ".") != 0)
|
|
|
4487 {
|
|
|
4488 char_u buf[MAXLINELEN];
|
|
|
4489
|
|
474
|
4490 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
4491 if (*items[0] == 'P')
|
|
|
4492 sprintf((char *)buf, "^%s", items[4]);
|
|
|
4493 else
|
|
|
4494 sprintf((char *)buf, "%s$", items[4]);
|
|
|
4495 aff_entry->ae_prog = vim_regcomp(buf,
|
|
481
|
4496 RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
4497 if (aff_entry->ae_prog == NULL)
|
|
|
4498 smsg((char_u *)_("Broken condition in %s line %d: %s"),
|
|
|
4499 fname, lnum, items[4]);
|
|
300
|
4500 }
|
|
339
|
4501
|
|
|
4502 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
4503 * for the condition. Use an existing one if possible. */
|
|
455
|
4504 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
4505 {
|
|
455
|
4506 /* When the chop string is one lower-case letter and
|
|
|
4507 * the add string ends in the upper-case letter we set
|
|
|
4508 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
4509 * letters from "ae_add". The condition must either
|
|
|
4510 * be empty or start with the same letter. */
|
|
|
4511 if (aff_entry->ae_chop != NULL
|
|
|
4512 && aff_entry->ae_add != NULL
|
|
|
4513 #ifdef FEAT_MBYTE
|
|
474
|
4514 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
4515 aff_entry->ae_chop)] == NUL
|
|
|
4516 #else
|
|
|
4517 && aff_entry->ae_chop[1] == NUL
|
|
|
4518 #endif
|
|
|
4519 )
|
|
339
|
4520 {
|
|
455
|
4521 int c, c_up;
|
|
|
4522
|
|
|
4523 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
4524 c_up = SPELL_TOUPPER(c);
|
|
|
4525 if (c_up != c
|
|
|
4526 && (aff_entry->ae_cond == NULL
|
|
|
4527 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
4528 {
|
|
|
4529 p = aff_entry->ae_add
|
|
|
4530 + STRLEN(aff_entry->ae_add);
|
|
|
4531 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
4532 if (PTR2CHAR(p) == c_up)
|
|
|
4533 {
|
|
|
4534 upper = TRUE;
|
|
|
4535 aff_entry->ae_chop = NULL;
|
|
|
4536 *p = NUL;
|
|
|
4537
|
|
|
4538 /* The condition is matched with the
|
|
|
4539 * actual word, thus must check for the
|
|
|
4540 * upper-case letter. */
|
|
|
4541 if (aff_entry->ae_cond != NULL)
|
|
|
4542 {
|
|
|
4543 char_u buf[MAXLINELEN];
|
|
|
4544 #ifdef FEAT_MBYTE
|
|
|
4545 if (has_mbyte)
|
|
|
4546 {
|
|
|
4547 onecap_copy(items[4], buf, TRUE);
|
|
|
4548 aff_entry->ae_cond = getroom_save(
|
|
474
|
4549 spin, buf);
|
|
455
|
4550 }
|
|
|
4551 else
|
|
|
4552 #endif
|
|
|
4553 *aff_entry->ae_cond = c_up;
|
|
|
4554 if (aff_entry->ae_cond != NULL)
|
|
|
4555 {
|
|
|
4556 sprintf((char *)buf, "^%s",
|
|
339
|
4557 aff_entry->ae_cond);
|
|
455
|
4558 vim_free(aff_entry->ae_prog);
|
|
|
4559 aff_entry->ae_prog = vim_regcomp(
|
|
|
4560 buf, RE_MAGIC + RE_STRING);
|
|
|
4561 }
|
|
|
4562 }
|
|
|
4563 }
|
|
|
4564 }
|
|
339
|
4565 }
|
|
|
4566
|
|
455
|
4567 if (aff_entry->ae_chop == NULL)
|
|
390
|
4568 {
|
|
455
|
4569 int idx;
|
|
|
4570 char_u **pp;
|
|
|
4571 int n;
|
|
|
4572
|
|
|
4573 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
4574 --idx)
|
|
|
4575 {
|
|
|
4576 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
4577 if (str_equal(p, aff_entry->ae_cond))
|
|
|
4578 break;
|
|
|
4579 }
|
|
|
4580 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
4581 {
|
|
|
4582 /* Not found, add a new condition. */
|
|
|
4583 idx = spin->si_prefcond.ga_len++;
|
|
|
4584 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
4585 + idx;
|
|
|
4586 if (aff_entry->ae_cond == NULL)
|
|
|
4587 *pp = NULL;
|
|
|
4588 else
|
|
474
|
4589 *pp = getroom_save(spin,
|
|
455
|
4590 aff_entry->ae_cond);
|
|
|
4591 }
|
|
|
4592
|
|
|
4593 /* Add the prefix to the prefix tree. */
|
|
|
4594 if (aff_entry->ae_add == NULL)
|
|
|
4595 p = (char_u *)"";
|
|
390
|
4596 else
|
|
455
|
4597 p = aff_entry->ae_add;
|
|
|
4598 /* PFX_FLAGS is a negative number, so that
|
|
|
4599 * tree_add_word() knows this is the prefix tree. */
|
|
|
4600 n = PFX_FLAGS;
|
|
|
4601 if (rare)
|
|
|
4602 n |= WFP_RARE;
|
|
|
4603 if (!cur_aff->ah_combine)
|
|
|
4604 n |= WFP_NC;
|
|
|
4605 if (upper)
|
|
|
4606 n |= WFP_UP;
|
|
474
|
4607 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
4608 idx, cur_aff->ah_newID);
|
|
390
|
4609 }
|
|
339
|
4610 }
|
|
240
|
4611 }
|
|
236
|
4612 }
|
|
255
|
4613 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
4614 {
|
|
|
4615 if (fol != NULL)
|
|
|
4616 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
4617 fname, lnum);
|
|
|
4618 else
|
|
|
4619 fol = vim_strsave(items[1]);
|
|
|
4620 }
|
|
|
4621 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
4622 {
|
|
|
4623 if (low != NULL)
|
|
|
4624 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
4625 fname, lnum);
|
|
|
4626 else
|
|
|
4627 low = vim_strsave(items[1]);
|
|
|
4628 }
|
|
|
4629 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
4630 {
|
|
|
4631 if (upp != NULL)
|
|
|
4632 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
4633 fname, lnum);
|
|
|
4634 else
|
|
|
4635 upp = vim_strsave(items[1]);
|
|
|
4636 }
|
|
236
|
4637 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
4638 {
|
|
236
|
4639 /* Ignore REP count */;
|
|
323
|
4640 if (!isdigit(*items[1]))
|
|
|
4641 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
4642 fname, lnum);
|
|
|
4643 }
|
|
482
|
4644 else if (STRCMP(items[0], "REP") == 0 && itemcnt >= 3)
|
|
236
|
4645 {
|
|
|
4646 /* REP item */
|
|
482
|
4647 /* Myspell ignores extra arguments, we require it starts with
|
|
|
4648 * # to detect mistakes. */
|
|
|
4649 if (itemcnt > 3 && items[3][0] != '#')
|
|
|
4650 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]);
|
|
323
|
4651 if (do_rep)
|
|
|
4652 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
4653 }
|
|
|
4654 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
4655 {
|
|
|
4656 /* MAP item or count */
|
|
|
4657 if (!found_map)
|
|
|
4658 {
|
|
|
4659 /* First line contains the count. */
|
|
|
4660 found_map = TRUE;
|
|
|
4661 if (!isdigit(*items[1]))
|
|
|
4662 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
4663 fname, lnum);
|
|
|
4664 }
|
|
|
4665 else if (do_map)
|
|
|
4666 {
|
|
346
|
4667 int c;
|
|
|
4668
|
|
|
4669 /* Check that every character appears only once. */
|
|
|
4670 for (p = items[1]; *p != NUL; )
|
|
|
4671 {
|
|
|
4672 #ifdef FEAT_MBYTE
|
|
|
4673 c = mb_ptr2char_adv(&p);
|
|
|
4674 #else
|
|
|
4675 c = *p++;
|
|
|
4676 #endif
|
|
|
4677 if ((spin->si_map.ga_len > 0
|
|
|
4678 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
4679 != NULL)
|
|
|
4680 || vim_strchr(p, c) != NULL)
|
|
|
4681 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
4682 fname, lnum);
|
|
|
4683 }
|
|
|
4684
|
|
323
|
4685 /* We simply concatenate all the MAP strings, separated by
|
|
|
4686 * slashes. */
|
|
|
4687 ga_concat(&spin->si_map, items[1]);
|
|
|
4688 ga_append(&spin->si_map, '/');
|
|
|
4689 }
|
|
|
4690 }
|
|
|
4691 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
4692 {
|
|
|
4693 if (do_sal)
|
|
|
4694 {
|
|
|
4695 /* SAL item (sounds-a-like)
|
|
|
4696 * Either one of the known keys or a from-to pair. */
|
|
|
4697 if (STRCMP(items[1], "followup") == 0)
|
|
|
4698 spin->si_followup = sal_to_bool(items[2]);
|
|
|
4699 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
4700 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
4701 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
4702 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
4703 else
|
|
|
4704 /* when "to" is "_" it means empty */
|
|
|
4705 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
4706 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
4707 : items[2]);
|
|
|
4708 }
|
|
236
|
4709 }
|
|
375
|
4710 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
4711 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
4712 {
|
|
|
4713 if (do_sofo)
|
|
|
4714 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
4715 }
|
|
|
4716 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
4717 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
4718 {
|
|
|
4719 if (do_sofo)
|
|
|
4720 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
4721 }
|
|
300
|
4722 else
|
|
481
|
4723 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"),
|
|
236
|
4724 fname, lnum, items[0]);
|
|
|
4725 }
|
|
|
4726 }
|
|
|
4727
|
|
375
|
4728 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
4729 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
4730 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
4731 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
4732 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
4733
|
|
255
|
4734 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
4735 {
|
|
351
|
4736 if (spin->si_clear_chartab)
|
|
|
4737 {
|
|
|
4738 /* Clear the char type tables, don't want to use any of the
|
|
|
4739 * currently used spell properties. */
|
|
|
4740 init_spell_chartab();
|
|
|
4741 spin->si_clear_chartab = FALSE;
|
|
|
4742 }
|
|
|
4743
|
|
316
|
4744 /*
|
|
|
4745 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
4746 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
4747 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
4748 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
4749 */
|
|
|
4750 if (!spin->si_ascii
|
|
|
4751 #ifdef FEAT_MBYTE
|
|
|
4752 && !enc_utf8
|
|
|
4753 #endif
|
|
|
4754 )
|
|
260
|
4755 {
|
|
|
4756 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
4757 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
4758 else
|
|
316
|
4759 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
4760 }
|
|
255
|
4761
|
|
|
4762 vim_free(fol);
|
|
|
4763 vim_free(low);
|
|
|
4764 vim_free(upp);
|
|
|
4765 }
|
|
|
4766
|
|
481
|
4767 /* Use compound specifications of the .aff file for the spell info. */
|
|
483
|
4768 if (aff->af_compmax != 0)
|
|
|
4769 {
|
|
|
4770 if (spin->si_compmax != 0 && spin->si_compmax != aff->af_compmax)
|
|
|
4771 smsg((char_u *)_("COMPOUNDMAX value differs from what is used in another .aff file"));
|
|
|
4772 else
|
|
|
4773 spin->si_compmax = aff->af_compmax;
|
|
|
4774 }
|
|
|
4775
|
|
481
|
4776 if (aff->af_compminlen != 0)
|
|
|
4777 {
|
|
|
4778 if (spin->si_compminlen != 0
|
|
|
4779 && spin->si_compminlen != aff->af_compminlen)
|
|
|
4780 smsg((char_u *)_("COMPOUNDMIN value differs from what is used in another .aff file"));
|
|
|
4781 else
|
|
|
4782 spin->si_compminlen = aff->af_compminlen;
|
|
|
4783 }
|
|
|
4784
|
|
483
|
4785 if (aff->af_compsylmax != 0)
|
|
|
4786 {
|
|
|
4787 if (aff->af_syllable == NULL)
|
|
|
4788 smsg((char_u *)_("COMPOUNDSYLMAX without SYLLABLE"));
|
|
|
4789
|
|
|
4790 if (spin->si_compsylmax != 0
|
|
|
4791 && spin->si_compsylmax != aff->af_compsylmax)
|
|
|
4792 smsg((char_u *)_("COMPOUNDSYLMAX value differs from what is used in another .aff file"));
|
|
|
4793 else
|
|
|
4794 spin->si_compsylmax = aff->af_compsylmax;
|
|
|
4795 }
|
|
|
4796
|
|
481
|
4797 if (aff->af_compflags != NULL)
|
|
|
4798 {
|
|
|
4799 if (spin->si_compflags != NULL
|
|
|
4800 && STRCMP(spin->si_compflags, aff->af_compflags) != 0)
|
|
|
4801 smsg((char_u *)_("COMPOUNDFLAG(S) value differs from what is used in another .aff file"));
|
|
|
4802 else
|
|
|
4803 spin->si_compflags = aff->af_compflags;
|
|
|
4804 }
|
|
|
4805
|
|
483
|
4806 if (aff->af_syllable != NULL)
|
|
|
4807 {
|
|
|
4808 if (spin->si_syllable != NULL
|
|
|
4809 && STRCMP(spin->si_syllable, aff->af_syllable) != 0)
|
|
|
4810 smsg((char_u *)_("SYLLABLE value differs from what is used in another .aff file"));
|
|
|
4811 else
|
|
|
4812 spin->si_syllable = aff->af_syllable;
|
|
|
4813 }
|
|
|
4814
|
|
236
|
4815 vim_free(pc);
|
|
|
4816 fclose(fd);
|
|
|
4817 return aff;
|
|
|
4818 }
|
|
|
4819
|
|
|
4820 /*
|
|
339
|
4821 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
4822 * NULL as equal.
|
|
|
4823 */
|
|
|
4824 static int
|
|
|
4825 str_equal(s1, s2)
|
|
|
4826 char_u *s1;
|
|
|
4827 char_u *s2;
|
|
|
4828 {
|
|
|
4829 if (s1 == NULL || s2 == NULL)
|
|
|
4830 return s1 == s2;
|
|
|
4831 return STRCMP(s1, s2) == 0;
|
|
|
4832 }
|
|
|
4833
|
|
|
4834 /*
|
|
323
|
4835 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
4836 * They are stored case-folded.
|
|
|
4837 */
|
|
|
4838 static void
|
|
|
4839 add_fromto(spin, gap, from, to)
|
|
|
4840 spellinfo_T *spin;
|
|
|
4841 garray_T *gap;
|
|
|
4842 char_u *from;
|
|
|
4843 char_u *to;
|
|
|
4844 {
|
|
|
4845 fromto_T *ftp;
|
|
|
4846 char_u word[MAXWLEN];
|
|
|
4847
|
|
|
4848 if (ga_grow(gap, 1) == OK)
|
|
|
4849 {
|
|
|
4850 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
4851 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
4852 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
4853 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
4854 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
4855 ++gap->ga_len;
|
|
|
4856 }
|
|
|
4857 }
|
|
|
4858
|
|
|
4859 /*
|
|
|
4860 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
4861 */
|
|
|
4862 static int
|
|
|
4863 sal_to_bool(s)
|
|
|
4864 char_u *s;
|
|
|
4865 {
|
|
|
4866 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
4867 }
|
|
|
4868
|
|
|
4869 /*
|
|
240
|
4870 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
4871 * When "s" is NULL FALSE is returned.
|
|
|
4872 */
|
|
|
4873 static int
|
|
|
4874 has_non_ascii(s)
|
|
|
4875 char_u *s;
|
|
|
4876 {
|
|
|
4877 char_u *p;
|
|
|
4878
|
|
|
4879 if (s != NULL)
|
|
|
4880 for (p = s; *p != NUL; ++p)
|
|
|
4881 if (*p >= 128)
|
|
|
4882 return TRUE;
|
|
|
4883 return FALSE;
|
|
|
4884 }
|
|
|
4885
|
|
|
4886 /*
|
|
236
|
4887 * Free the structure filled by spell_read_aff().
|
|
|
4888 */
|
|
|
4889 static void
|
|
|
4890 spell_free_aff(aff)
|
|
|
4891 afffile_T *aff;
|
|
|
4892 {
|
|
|
4893 hashtab_T *ht;
|
|
|
4894 hashitem_T *hi;
|
|
|
4895 int todo;
|
|
|
4896 affheader_T *ah;
|
|
300
|
4897 affentry_T *ae;
|
|
236
|
4898
|
|
|
4899 vim_free(aff->af_enc);
|
|
|
4900
|
|
339
|
4901 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
4902 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
4903 {
|
|
|
4904 todo = ht->ht_used;
|
|
|
4905 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
4906 {
|
|
|
4907 if (!HASHITEM_EMPTY(hi))
|
|
|
4908 {
|
|
|
4909 --todo;
|
|
|
4910 ah = HI2AH(hi);
|
|
300
|
4911 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
4912 vim_free(ae->ae_prog);
|
|
236
|
4913 }
|
|
|
4914 }
|
|
|
4915 if (ht == &aff->af_suff)
|
|
|
4916 break;
|
|
|
4917 }
|
|
300
|
4918
|
|
236
|
4919 hash_clear(&aff->af_pref);
|
|
|
4920 hash_clear(&aff->af_suff);
|
|
|
4921 }
|
|
|
4922
|
|
|
4923 /*
|
|
300
|
4924 * Read dictionary file "fname".
|
|
236
|
4925 * Returns OK or FAIL;
|
|
|
4926 */
|
|
|
4927 static int
|
|
474
|
4928 spell_read_dic(spin, fname, affile)
|
|
|
4929 spellinfo_T *spin;
|
|
236
|
4930 char_u *fname;
|
|
300
|
4931 afffile_T *affile;
|
|
236
|
4932 {
|
|
300
|
4933 hashtab_T ht;
|
|
236
|
4934 char_u line[MAXLINELEN];
|
|
481
|
4935 char_u *p;
|
|
300
|
4936 char_u *afflist;
|
|
483
|
4937 char_u store_afflist[MAXWLEN];
|
|
|
4938 int pfxlen;
|
|
|
4939 int need_affix;
|
|
300
|
4940 char_u *dw;
|
|
236
|
4941 char_u *pc;
|
|
|
4942 char_u *w;
|
|
|
4943 int l;
|
|
|
4944 hash_T hash;
|
|
|
4945 hashitem_T *hi;
|
|
|
4946 FILE *fd;
|
|
|
4947 int lnum = 1;
|
|
300
|
4948 int non_ascii = 0;
|
|
|
4949 int retval = OK;
|
|
|
4950 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
4951 int flags;
|
|
481
|
4952 int duplicate = 0;
|
|
236
|
4953
|
|
300
|
4954 /*
|
|
|
4955 * Open the file.
|
|
|
4956 */
|
|
310
|
4957 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4958 if (fd == NULL)
|
|
|
4959 {
|
|
|
4960 EMSG2(_(e_notopen), fname);
|
|
|
4961 return FAIL;
|
|
|
4962 }
|
|
|
4963
|
|
300
|
4964 /* The hashtable is only used to detect duplicated words. */
|
|
|
4965 hash_init(&ht);
|
|
|
4966
|
|
310
|
4967 if (spin->si_verbose || p_verbose > 2)
|
|
|
4968 {
|
|
|
4969 if (!spin->si_verbose)
|
|
|
4970 verbose_enter();
|
|
366
|
4971 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
4972 out_flush();
|
|
|
4973 if (!spin->si_verbose)
|
|
|
4974 verbose_leave();
|
|
|
4975 }
|
|
236
|
4976
|
|
474
|
4977 /* start with a message for the first line */
|
|
|
4978 spin->si_msg_count = 999999;
|
|
|
4979
|
|
236
|
4980 /* Read and ignore the first line: word count. */
|
|
|
4981 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
4982 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
4983 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
4984
|
|
|
4985 /*
|
|
|
4986 * Read all the lines in the file one by one.
|
|
|
4987 * The words are converted to 'encoding' here, before being added to
|
|
|
4988 * the hashtable.
|
|
|
4989 */
|
|
255
|
4990 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
4991 {
|
|
255
|
4992 line_breakcheck();
|
|
236
|
4993 ++lnum;
|
|
455
|
4994 if (line[0] == '#' || line[0] == '/')
|
|
351
|
4995 continue; /* comment line */
|
|
236
|
4996
|
|
300
|
4997 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
4998 * the word is kept to allow e.g., "et al.". */
|
|
236
|
4999 l = STRLEN(line);
|
|
|
5000 while (l > 0 && line[l - 1] <= ' ')
|
|
|
5001 --l;
|
|
|
5002 if (l == 0)
|
|
|
5003 continue; /* empty line */
|
|
|
5004 line[l] = NUL;
|
|
|
5005
|
|
481
|
5006 /* Find the optional affix names. Replace the SLASH character by a
|
|
|
5007 * slash. */
|
|
|
5008 afflist = NULL;
|
|
|
5009 for (p = line; *p != NUL; mb_ptr_adv(p))
|
|
|
5010 {
|
|
|
5011 if (*p == affile->af_slash)
|
|
|
5012 *p = '/';
|
|
|
5013 else if (*p == '/')
|
|
|
5014 {
|
|
|
5015 *p = NUL;
|
|
|
5016 afflist = p + 1;
|
|
|
5017 break;
|
|
|
5018 }
|
|
|
5019 }
|
|
236
|
5020
|
|
300
|
5021 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
5022 if (spin->si_ascii && has_non_ascii(line))
|
|
|
5023 {
|
|
|
5024 ++non_ascii;
|
|
240
|
5025 continue;
|
|
300
|
5026 }
|
|
240
|
5027
|
|
310
|
5028 #ifdef FEAT_MBYTE
|
|
236
|
5029 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
5030 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
5031 {
|
|
300
|
5032 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
5033 if (pc == NULL)
|
|
|
5034 {
|
|
|
5035 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5036 fname, lnum, line);
|
|
|
5037 continue;
|
|
|
5038 }
|
|
236
|
5039 w = pc;
|
|
|
5040 }
|
|
|
5041 else
|
|
310
|
5042 #endif
|
|
236
|
5043 {
|
|
|
5044 pc = NULL;
|
|
|
5045 w = line;
|
|
|
5046 }
|
|
|
5047
|
|
474
|
5048 /* This takes time, print a message every 10000 words. */
|
|
|
5049 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
5050 {
|
|
|
5051 spin->si_msg_count = 0;
|
|
339
|
5052 vim_snprintf((char *)message, sizeof(message),
|
|
|
5053 _("line %6d, word %6d - %s"),
|
|
|
5054 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
5055 msg_start();
|
|
|
5056 msg_puts_long_attr(message, 0);
|
|
|
5057 msg_clr_eos();
|
|
|
5058 msg_didout = FALSE;
|
|
|
5059 msg_col = 0;
|
|
|
5060 out_flush();
|
|
|
5061 }
|
|
|
5062
|
|
300
|
5063 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
5064 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
5065 if (dw == NULL)
|
|
300
|
5066 retval = FAIL;
|
|
|
5067 vim_free(pc);
|
|
|
5068 if (retval == FAIL)
|
|
236
|
5069 break;
|
|
|
5070
|
|
300
|
5071 hash = hash_hash(dw);
|
|
|
5072 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
5073 if (!HASHITEM_EMPTY(hi))
|
|
481
|
5074 {
|
|
|
5075 if (p_verbose > 0)
|
|
|
5076 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
5077 fname, lnum, dw);
|
|
481
|
5078 else if (duplicate == 0)
|
|
|
5079 smsg((char_u *)_("First duplicate word in %s line %d: %s"),
|
|
|
5080 fname, lnum, dw);
|
|
|
5081 ++duplicate;
|
|
|
5082 }
|
|
236
|
5083 else
|
|
300
|
5084 hash_add_item(&ht, hi, dw, hash);
|
|
|
5085
|
|
307
|
5086 flags = 0;
|
|
483
|
5087 store_afflist[0] = NUL;
|
|
|
5088 pfxlen = 0;
|
|
|
5089 need_affix = FALSE;
|
|
307
|
5090 if (afflist != NULL)
|
|
|
5091 {
|
|
|
5092 /* Check for affix name that stands for keep-case word and stands
|
|
|
5093 * for rare word (if defined). */
|
|
310
|
5094 if (affile->af_kep != NUL
|
|
|
5095 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
389
|
5096 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
307
|
5097 if (affile->af_rar != NUL
|
|
|
5098 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
5099 flags |= WF_RARE;
|
|
346
|
5100 if (affile->af_bad != NUL
|
|
|
5101 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
5102 flags |= WF_BANNED;
|
|
483
|
5103 if (affile->af_needaffix != NUL
|
|
|
5104 && vim_strchr(afflist, affile->af_needaffix) != NULL)
|
|
|
5105 need_affix = TRUE;
|
|
339
|
5106
|
|
|
5107 if (affile->af_pfxpostpone)
|
|
|
5108 /* Need to store the list of prefix IDs with the word. */
|
|
483
|
5109 pfxlen = get_pfxlist(affile, afflist, store_afflist);
|
|
|
5110
|
|
|
5111 if (spin->si_compflags != NULL)
|
|
|
5112 /* Need to store the list of compound flags with the word.
|
|
|
5113 * Concatenate them to the list of prefix IDs. */
|
|
|
5114 get_compflags(spin, afflist, store_afflist + pfxlen);
|
|
307
|
5115 }
|
|
|
5116
|
|
300
|
5117 /* Add the word to the word tree(s). */
|
|
483
|
5118 if (store_word(spin, dw, flags, spin->si_region,
|
|
|
5119 store_afflist, need_affix) == FAIL)
|
|
300
|
5120 retval = FAIL;
|
|
236
|
5121
|
|
300
|
5122 if (afflist != NULL)
|
|
|
5123 {
|
|
|
5124 /* Find all matching suffixes and add the resulting words.
|
|
|
5125 * Additionally do matching prefixes that combine. */
|
|
474
|
5126 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
5127 &affile->af_suff, &affile->af_pref,
|
|
483
|
5128 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5129 retval = FAIL;
|
|
|
5130
|
|
|
5131 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
5132 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
5133 &affile->af_pref, NULL,
|
|
483
|
5134 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5135 retval = FAIL;
|
|
|
5136 }
|
|
236
|
5137 }
|
|
|
5138
|
|
481
|
5139 if (duplicate > 0)
|
|
|
5140 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname);
|
|
300
|
5141 if (spin->si_ascii && non_ascii > 0)
|
|
481
|
5142 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"),
|
|
|
5143 non_ascii, fname);
|
|
300
|
5144 hash_clear(&ht);
|
|
|
5145
|
|
236
|
5146 fclose(fd);
|
|
300
|
5147 return retval;
|
|
236
|
5148 }
|
|
|
5149
|
|
|
5150 /*
|
|
339
|
5151 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
5152 * Used for PFXPOSTPONE.
|
|
483
|
5153 * Put the resulting flags in "store_afflist[MAXWLEN]" with a terminating NUL
|
|
|
5154 * and return the number of affixes.
|
|
|
5155 */
|
|
|
5156 static int
|
|
|
5157 get_pfxlist(affile, afflist, store_afflist)
|
|
339
|
5158 afffile_T *affile;
|
|
|
5159 char_u *afflist;
|
|
483
|
5160 char_u *store_afflist;
|
|
339
|
5161 {
|
|
|
5162 char_u *p;
|
|
483
|
5163 int cnt = 0;
|
|
339
|
5164 char_u key[2];
|
|
|
5165 hashitem_T *hi;
|
|
|
5166
|
|
|
5167 key[1] = NUL;
|
|
483
|
5168 for (p = afflist; *p != NUL; ++p)
|
|
|
5169 {
|
|
|
5170 key[0] = *p;
|
|
|
5171 hi = hash_find(&affile->af_pref, key);
|
|
|
5172 if (!HASHITEM_EMPTY(hi))
|
|
|
5173 /* This is a prefix ID, use the new number. */
|
|
|
5174 store_afflist[cnt++] = HI2AH(hi)->ah_newID;
|
|
|
5175 }
|
|
|
5176
|
|
|
5177 store_afflist[cnt] = NUL;
|
|
|
5178 return cnt;
|
|
339
|
5179 }
|
|
|
5180
|
|
|
5181 /*
|
|
481
|
5182 * Get the list of affix IDs from the affix list "afflist" that are used for
|
|
|
5183 * compound words.
|
|
483
|
5184 * Puts the flags in "store_afflist[]".
|
|
|
5185 */
|
|
|
5186 static void
|
|
|
5187 get_compflags(spin, afflist, store_afflist)
|
|
481
|
5188 spellinfo_T *spin;
|
|
|
5189 char_u *afflist;
|
|
483
|
5190 char_u *store_afflist;
|
|
481
|
5191 {
|
|
|
5192 char_u *p;
|
|
483
|
5193 int cnt = 0;
|
|
|
5194
|
|
|
5195 for (p = afflist; *p != NUL; ++p)
|
|
|
5196 /* A flag is a compound flag if it appears in "si_compflags" and
|
|
|
5197 * it's not a special character. */
|
|
|
5198 if (vim_strchr(spin->si_compflags, *p) != NULL
|
|
|
5199 && vim_strchr((char_u *)"+*[]/", *p) == NULL)
|
|
|
5200 store_afflist[cnt++] = *p;
|
|
|
5201
|
|
|
5202 store_afflist[cnt] = NUL;
|
|
481
|
5203 }
|
|
|
5204
|
|
|
5205 /*
|
|
300
|
5206 * Apply affixes to a word and store the resulting words.
|
|
|
5207 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
5208 * prefixes or suffixes.
|
|
|
5209 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
5210 * the resulting words for combining affixes.
|
|
|
5211 *
|
|
|
5212 * Returns FAIL when out of memory.
|
|
236
|
5213 */
|
|
300
|
5214 static int
|
|
483
|
5215 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags,
|
|
|
5216 pfxlist, pfxlen)
|
|
474
|
5217 spellinfo_T *spin; /* spell info */
|
|
300
|
5218 char_u *word; /* basic word start */
|
|
|
5219 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
5220 afffile_T *affile;
|
|
300
|
5221 hashtab_T *ht;
|
|
|
5222 hashtab_T *xht;
|
|
|
5223 int comb; /* only use affixes that combine */
|
|
307
|
5224 int flags; /* flags for the word */
|
|
339
|
5225 char_u *pfxlist; /* list of prefix IDs */
|
|
483
|
5226 int pfxlen; /* nr of flags in "pfxlist" for prefixes, rest
|
|
|
5227 * is compound flags */
|
|
236
|
5228 {
|
|
|
5229 int todo;
|
|
|
5230 hashitem_T *hi;
|
|
300
|
5231 affheader_T *ah;
|
|
|
5232 affentry_T *ae;
|
|
|
5233 regmatch_T regmatch;
|
|
|
5234 char_u newword[MAXWLEN];
|
|
|
5235 int retval = OK;
|
|
|
5236 int i;
|
|
|
5237 char_u *p;
|
|
366
|
5238 int use_flags;
|
|
390
|
5239 char_u *use_pfxlist;
|
|
483
|
5240 char_u pfx_pfxlist[MAXWLEN];
|
|
455
|
5241 int c;
|
|
483
|
5242 size_t wordlen = STRLEN(word);
|
|
236
|
5243
|
|
300
|
5244 todo = ht->ht_used;
|
|
|
5245 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
5246 {
|
|
|
5247 if (!HASHITEM_EMPTY(hi))
|
|
|
5248 {
|
|
|
5249 --todo;
|
|
300
|
5250 ah = HI2AH(hi);
|
|
236
|
5251
|
|
300
|
5252 /* Check that the affix combines, if required, and that the word
|
|
|
5253 * supports this affix. */
|
|
455
|
5254 c = PTR2CHAR(ah->ah_key);
|
|
|
5255 if ((!comb || ah->ah_combine) && vim_strchr(afflist, c) != NULL)
|
|
236
|
5256 {
|
|
300
|
5257 /* Loop over all affix entries with this name. */
|
|
|
5258 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
5259 {
|
|
300
|
5260 /* Check the condition. It's not logical to match case
|
|
|
5261 * here, but it is required for compatibility with
|
|
339
|
5262 * Myspell.
|
|
481
|
5263 * Another requirement from Myspell is that the chop
|
|
|
5264 * string is shorter than the word itself.
|
|
339
|
5265 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
5266 * prefixes with a chop string. */
|
|
300
|
5267 regmatch.regprog = ae->ae_prog;
|
|
|
5268 regmatch.rm_ic = FALSE;
|
|
339
|
5269 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
5270 || ae->ae_chop != NULL)
|
|
481
|
5271 && (ae->ae_chop == NULL
|
|
|
5272 || STRLEN(ae->ae_chop) < wordlen)
|
|
339
|
5273 && (ae->ae_prog == NULL
|
|
|
5274 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
5275 {
|
|
|
5276 /* Match. Remove the chop and add the affix. */
|
|
|
5277 if (xht == NULL)
|
|
240
|
5278 {
|
|
300
|
5279 /* prefix: chop/add at the start of the word */
|
|
|
5280 if (ae->ae_add == NULL)
|
|
|
5281 *newword = NUL;
|
|
|
5282 else
|
|
|
5283 STRCPY(newword, ae->ae_add);
|
|
|
5284 p = word;
|
|
|
5285 if (ae->ae_chop != NULL)
|
|
310
|
5286 {
|
|
300
|
5287 /* Skip chop string. */
|
|
310
|
5288 #ifdef FEAT_MBYTE
|
|
|
5289 if (has_mbyte)
|
|
324
|
5290 {
|
|
310
|
5291 i = mb_charlen(ae->ae_chop);
|
|
324
|
5292 for ( ; i > 0; --i)
|
|
|
5293 mb_ptr_adv(p);
|
|
|
5294 }
|
|
310
|
5295 else
|
|
|
5296 #endif
|
|
324
|
5297 p += STRLEN(ae->ae_chop);
|
|
310
|
5298 }
|
|
300
|
5299 STRCAT(newword, p);
|
|
|
5300 }
|
|
|
5301 else
|
|
|
5302 {
|
|
|
5303 /* suffix: chop/add at the end of the word */
|
|
|
5304 STRCPY(newword, word);
|
|
|
5305 if (ae->ae_chop != NULL)
|
|
|
5306 {
|
|
|
5307 /* Remove chop string. */
|
|
|
5308 p = newword + STRLEN(newword);
|
|
376
|
5309 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
5310 for ( ; i > 0; --i)
|
|
300
|
5311 mb_ptr_back(newword, p);
|
|
|
5312 *p = NUL;
|
|
|
5313 }
|
|
|
5314 if (ae->ae_add != NULL)
|
|
|
5315 STRCAT(newword, ae->ae_add);
|
|
240
|
5316 }
|
|
|
5317
|
|
366
|
5318 /* Obey the "rare" flag of the affix. */
|
|
|
5319 if (ae->ae_rare)
|
|
|
5320 use_flags = flags | WF_RARE;
|
|
|
5321 else
|
|
|
5322 use_flags = flags;
|
|
483
|
5323
|
|
|
5324 /* Obey the "nocomp" flag of the affix: don't use the
|
|
|
5325 * compound flags. */
|
|
390
|
5326 use_pfxlist = pfxlist;
|
|
483
|
5327 if (ae->ae_nocomp && pfxlist != NULL)
|
|
|
5328 {
|
|
|
5329 vim_strncpy(pfx_pfxlist, pfxlist, pfxlen);
|
|
|
5330 use_pfxlist = pfx_pfxlist;
|
|
|
5331 }
|
|
390
|
5332
|
|
|
5333 /* When there are postponed prefixes... */
|
|
393
|
5334 if (spin->si_prefroot != NULL
|
|
|
5335 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
5336 {
|
|
|
5337 /* ... add a flag to indicate an affix was used. */
|
|
|
5338 use_flags |= WF_HAS_AFF;
|
|
|
5339
|
|
|
5340 /* ... don't use a prefix list if combining
|
|
483
|
5341 * affixes is not allowed. But do use the
|
|
|
5342 * compound flags after them. */
|
|
|
5343 if ((!ah->ah_combine || comb) && pfxlist != NULL)
|
|
|
5344 use_pfxlist += pfxlen;
|
|
390
|
5345 }
|
|
366
|
5346
|
|
300
|
5347 /* Store the modified word. */
|
|
474
|
5348 if (store_word(spin, newword, use_flags,
|
|
483
|
5349 spin->si_region, use_pfxlist, FALSE) == FAIL)
|
|
300
|
5350 retval = FAIL;
|
|
236
|
5351
|
|
300
|
5352 /* When added a suffix and combining is allowed also
|
|
|
5353 * try adding prefixes additionally. */
|
|
|
5354 if (xht != NULL && ah->ah_combine)
|
|
474
|
5355 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
5356 xht, NULL, TRUE,
|
|
483
|
5357 use_flags, use_pfxlist, pfxlen) == FAIL)
|
|
300
|
5358 retval = FAIL;
|
|
236
|
5359 }
|
|
|
5360 }
|
|
|
5361 }
|
|
|
5362 }
|
|
|
5363 }
|
|
|
5364
|
|
|
5365 return retval;
|
|
|
5366 }
|
|
|
5367
|
|
|
5368 /*
|
|
300
|
5369 * Read a file with a list of words.
|
|
236
|
5370 */
|
|
|
5371 static int
|
|
474
|
5372 spell_read_wordfile(spin, fname)
|
|
|
5373 spellinfo_T *spin;
|
|
300
|
5374 char_u *fname;
|
|
236
|
5375 {
|
|
300
|
5376 FILE *fd;
|
|
|
5377 long lnum = 0;
|
|
|
5378 char_u rline[MAXLINELEN];
|
|
|
5379 char_u *line;
|
|
|
5380 char_u *pc = NULL;
|
|
381
|
5381 char_u *p;
|
|
300
|
5382 int l;
|
|
|
5383 int retval = OK;
|
|
|
5384 int did_word = FALSE;
|
|
|
5385 int non_ascii = 0;
|
|
307
|
5386 int flags;
|
|
316
|
5387 int regionmask;
|
|
236
|
5388
|
|
300
|
5389 /*
|
|
|
5390 * Open the file.
|
|
|
5391 */
|
|
310
|
5392 fd = mch_fopen((char *)fname, "r");
|
|
300
|
5393 if (fd == NULL)
|
|
236
|
5394 {
|
|
300
|
5395 EMSG2(_(e_notopen), fname);
|
|
|
5396 return FAIL;
|
|
236
|
5397 }
|
|
|
5398
|
|
310
|
5399 if (spin->si_verbose || p_verbose > 2)
|
|
|
5400 {
|
|
|
5401 if (!spin->si_verbose)
|
|
|
5402 verbose_enter();
|
|
366
|
5403 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
5404 out_flush();
|
|
|
5405 if (!spin->si_verbose)
|
|
|
5406 verbose_leave();
|
|
|
5407 }
|
|
300
|
5408
|
|
|
5409 /*
|
|
|
5410 * Read all the lines in the file one by one.
|
|
|
5411 */
|
|
|
5412 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
5413 {
|
|
|
5414 line_breakcheck();
|
|
|
5415 ++lnum;
|
|
|
5416
|
|
|
5417 /* Skip comment lines. */
|
|
|
5418 if (*rline == '#')
|
|
|
5419 continue;
|
|
|
5420
|
|
|
5421 /* Remove CR, LF and white space from the end. */
|
|
|
5422 l = STRLEN(rline);
|
|
|
5423 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
5424 --l;
|
|
|
5425 if (l == 0)
|
|
|
5426 continue; /* empty or blank line */
|
|
|
5427 rline[l] = NUL;
|
|
|
5428
|
|
|
5429 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
5430 vim_free(pc);
|
|
310
|
5431 #ifdef FEAT_MBYTE
|
|
300
|
5432 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
5433 {
|
|
|
5434 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
5435 if (pc == NULL)
|
|
|
5436 {
|
|
|
5437 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5438 fname, lnum, rline);
|
|
|
5439 continue;
|
|
|
5440 }
|
|
|
5441 line = pc;
|
|
|
5442 }
|
|
|
5443 else
|
|
310
|
5444 #endif
|
|
300
|
5445 {
|
|
|
5446 pc = NULL;
|
|
|
5447 line = rline;
|
|
|
5448 }
|
|
|
5449
|
|
307
|
5450 if (*line == '/')
|
|
300
|
5451 {
|
|
307
|
5452 ++line;
|
|
|
5453 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
5454 {
|
|
|
5455 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
5456 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
5457 fname, lnum, line - 1);
|
|
300
|
5458 else if (did_word)
|
|
316
|
5459 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
5460 fname, lnum, line - 1);
|
|
300
|
5461 else
|
|
|
5462 {
|
|
310
|
5463 #ifdef FEAT_MBYTE
|
|
|
5464 char_u *enc;
|
|
|
5465
|
|
300
|
5466 /* Setup for conversion to 'encoding'. */
|
|
316
|
5467 line += 10;
|
|
|
5468 enc = enc_canonize(line);
|
|
300
|
5469 if (enc != NULL && !spin->si_ascii
|
|
|
5470 && convert_setup(&spin->si_conv, enc,
|
|
|
5471 p_enc) == FAIL)
|
|
|
5472 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
5473 fname, line, p_enc);
|
|
300
|
5474 vim_free(enc);
|
|
375
|
5475 spin->si_conv.vc_fail = TRUE;
|
|
310
|
5476 #else
|
|
|
5477 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
5478 #endif
|
|
300
|
5479 }
|
|
307
|
5480 continue;
|
|
300
|
5481 }
|
|
307
|
5482
|
|
316
|
5483 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
5484 {
|
|
|
5485 if (spin->si_region_count > 1)
|
|
|
5486 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
5487 fname, lnum, line);
|
|
|
5488 else
|
|
|
5489 {
|
|
|
5490 line += 8;
|
|
|
5491 if (STRLEN(line) > 16)
|
|
|
5492 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
5493 fname, lnum, line);
|
|
|
5494 else
|
|
|
5495 {
|
|
|
5496 spin->si_region_count = STRLEN(line) / 2;
|
|
|
5497 STRCPY(spin->si_region_name, line);
|
|
389
|
5498
|
|
|
5499 /* Adjust the mask for a word valid in all regions. */
|
|
|
5500 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
5501 }
|
|
|
5502 }
|
|
|
5503 continue;
|
|
|
5504 }
|
|
|
5505
|
|
381
|
5506 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
5507 fname, lnum, line - 1);
|
|
|
5508 continue;
|
|
|
5509 }
|
|
|
5510
|
|
|
5511 flags = 0;
|
|
|
5512 regionmask = spin->si_region;
|
|
|
5513
|
|
|
5514 /* Check for flags and region after a slash. */
|
|
|
5515 p = vim_strchr(line, '/');
|
|
|
5516 if (p != NULL)
|
|
|
5517 {
|
|
|
5518 *p++ = NUL;
|
|
|
5519 while (*p != NUL)
|
|
|
5520 {
|
|
|
5521 if (*p == '=') /* keep-case word */
|
|
389
|
5522 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
5523 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
5524 flags |= WF_BANNED;
|
|
|
5525 else if (*p == '?') /* Rare word. */
|
|
|
5526 flags |= WF_RARE;
|
|
|
5527 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
5528 {
|
|
381
|
5529 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
5530 regionmask = 0;
|
|
|
5531 flags |= WF_REGION;
|
|
|
5532
|
|
|
5533 l = *p - '0';
|
|
316
|
5534 if (l > spin->si_region_count)
|
|
|
5535 {
|
|
|
5536 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
5537 fname, lnum, p);
|
|
316
|
5538 break;
|
|
|
5539 }
|
|
|
5540 regionmask |= 1 << (l - 1);
|
|
|
5541 }
|
|
381
|
5542 else
|
|
|
5543 {
|
|
|
5544 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
5545 fname, lnum, p);
|
|
|
5546 break;
|
|
|
5547 }
|
|
|
5548 ++p;
|
|
307
|
5549 }
|
|
300
|
5550 }
|
|
|
5551
|
|
|
5552 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
5553 if (spin->si_ascii && has_non_ascii(line))
|
|
|
5554 {
|
|
|
5555 ++non_ascii;
|
|
|
5556 continue;
|
|
|
5557 }
|
|
|
5558
|
|
|
5559 /* Normal word: store it. */
|
|
483
|
5560 if (store_word(spin, line, flags, regionmask, NULL, FALSE) == FAIL)
|
|
300
|
5561 {
|
|
|
5562 retval = FAIL;
|
|
|
5563 break;
|
|
|
5564 }
|
|
|
5565 did_word = TRUE;
|
|
|
5566 }
|
|
|
5567
|
|
|
5568 vim_free(pc);
|
|
|
5569 fclose(fd);
|
|
|
5570
|
|
310
|
5571 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
5572 {
|
|
|
5573 if (p_verbose > 2)
|
|
|
5574 verbose_enter();
|
|
300
|
5575 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
5576 non_ascii);
|
|
310
|
5577 if (p_verbose > 2)
|
|
|
5578 verbose_leave();
|
|
|
5579 }
|
|
300
|
5580 return retval;
|
|
236
|
5581 }
|
|
|
5582
|
|
|
5583 /*
|
|
300
|
5584 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
5585 * This avoids calling free() for every little struct we use (and keeping
|
|
|
5586 * track of them).
|
|
300
|
5587 * The memory is cleared to all zeros.
|
|
|
5588 * Returns NULL when out of memory.
|
|
|
5589 */
|
|
|
5590 static void *
|
|
474
|
5591 getroom(spin, len, align)
|
|
|
5592 spellinfo_T *spin;
|
|
447
|
5593 size_t len; /* length needed */
|
|
|
5594 int align; /* align for pointer */
|
|
300
|
5595 {
|
|
|
5596 char_u *p;
|
|
474
|
5597 sblock_T *bl = spin->si_blocks;
|
|
300
|
5598
|
|
447
|
5599 if (align && bl != NULL)
|
|
|
5600 /* Round size up for alignment. On some systems structures need to be
|
|
|
5601 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
5602 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
5603 & ~(sizeof(char *) - 1);
|
|
|
5604
|
|
300
|
5605 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
5606 {
|
|
|
5607 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
5608 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
5609 if (bl == NULL)
|
|
|
5610 return NULL;
|
|
474
|
5611 bl->sb_next = spin->si_blocks;
|
|
|
5612 spin->si_blocks = bl;
|
|
300
|
5613 bl->sb_used = 0;
|
|
482
|
5614 ++spin->si_blocks_cnt;
|
|
300
|
5615 }
|
|
|
5616
|
|
|
5617 p = bl->sb_data + bl->sb_used;
|
|
|
5618 bl->sb_used += len;
|
|
|
5619
|
|
|
5620 return p;
|
|
|
5621 }
|
|
|
5622
|
|
|
5623 /*
|
|
|
5624 * Make a copy of a string into memory allocated with getroom().
|
|
|
5625 */
|
|
|
5626 static char_u *
|
|
474
|
5627 getroom_save(spin, s)
|
|
|
5628 spellinfo_T *spin;
|
|
300
|
5629 char_u *s;
|
|
|
5630 {
|
|
|
5631 char_u *sc;
|
|
|
5632
|
|
474
|
5633 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
5634 if (sc != NULL)
|
|
|
5635 STRCPY(sc, s);
|
|
|
5636 return sc;
|
|
|
5637 }
|
|
|
5638
|
|
|
5639
|
|
|
5640 /*
|
|
|
5641 * Free the list of allocated sblock_T.
|
|
236
|
5642 */
|
|
|
5643 static void
|
|
300
|
5644 free_blocks(bl)
|
|
|
5645 sblock_T *bl;
|
|
236
|
5646 {
|
|
300
|
5647 sblock_T *next;
|
|
236
|
5648
|
|
300
|
5649 while (bl != NULL)
|
|
236
|
5650 {
|
|
300
|
5651 next = bl->sb_next;
|
|
|
5652 vim_free(bl);
|
|
|
5653 bl = next;
|
|
236
|
5654 }
|
|
|
5655 }
|
|
|
5656
|
|
|
5657 /*
|
|
300
|
5658 * Allocate the root of a word tree.
|
|
236
|
5659 */
|
|
300
|
5660 static wordnode_T *
|
|
474
|
5661 wordtree_alloc(spin)
|
|
|
5662 spellinfo_T *spin;
|
|
|
5663 {
|
|
|
5664 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
5665 }
|
|
|
5666
|
|
|
5667 /*
|
|
300
|
5668 * Store a word in the tree(s).
|
|
390
|
5669 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
5670 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
5671 * used to find suggestions.
|
|
300
|
5672 * For a keep-case word also store it in the keep-case tree.
|
|
482
|
5673 * When "pfxlist" is not NULL store the word for each postponed prefix ID and
|
|
|
5674 * compound flag.
|
|
236
|
5675 */
|
|
|
5676 static int
|
|
483
|
5677 store_word(spin, word, flags, region, pfxlist, need_affix)
|
|
474
|
5678 spellinfo_T *spin;
|
|
300
|
5679 char_u *word;
|
|
307
|
5680 int flags; /* extra flags, WF_BANNED */
|
|
316
|
5681 int region; /* supported region(s) */
|
|
339
|
5682 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
483
|
5683 int need_affix; /* only store word with affix ID */
|
|
236
|
5684 {
|
|
300
|
5685 int len = STRLEN(word);
|
|
|
5686 int ct = captype(word, word + len);
|
|
|
5687 char_u foldword[MAXWLEN];
|
|
339
|
5688 int res = OK;
|
|
|
5689 char_u *p;
|
|
236
|
5690
|
|
323
|
5691 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
5692 for (p = pfxlist; res == OK; ++p)
|
|
|
5693 {
|
|
483
|
5694 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
5695 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
474
|
5696 region, p == NULL ? 0 : *p);
|
|
339
|
5697 if (p == NULL || *p == NUL)
|
|
|
5698 break;
|
|
|
5699 }
|
|
334
|
5700 ++spin->si_foldwcount;
|
|
307
|
5701
|
|
474
|
5702 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
5703 {
|
|
339
|
5704 for (p = pfxlist; res == OK; ++p)
|
|
|
5705 {
|
|
483
|
5706 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
5707 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
474
|
5708 region, p == NULL ? 0 : *p);
|
|
339
|
5709 if (p == NULL || *p == NUL)
|
|
|
5710 break;
|
|
|
5711 }
|
|
334
|
5712 ++spin->si_keepwcount;
|
|
|
5713 }
|
|
300
|
5714 return res;
|
|
236
|
5715 }
|
|
|
5716
|
|
|
5717 /*
|
|
300
|
5718 * Add word "word" to a word tree at "root".
|
|
366
|
5719 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
5720 * "rare" and "region" is the condition nr.
|
|
255
|
5721 * Returns FAIL when out of memory.
|
|
236
|
5722 */
|
|
255
|
5723 static int
|
|
481
|
5724 tree_add_word(spin, word, root, flags, region, affixID)
|
|
474
|
5725 spellinfo_T *spin;
|
|
300
|
5726 char_u *word;
|
|
|
5727 wordnode_T *root;
|
|
|
5728 int flags;
|
|
|
5729 int region;
|
|
481
|
5730 int affixID;
|
|
236
|
5731 {
|
|
300
|
5732 wordnode_T *node = root;
|
|
|
5733 wordnode_T *np;
|
|
469
|
5734 wordnode_T *copyp, **copyprev;
|
|
300
|
5735 wordnode_T **prev = NULL;
|
|
|
5736 int i;
|
|
255
|
5737
|
|
300
|
5738 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
5739 for (i = 0; ; ++i)
|
|
255
|
5740 {
|
|
469
|
5741 /* When there is more than one reference to this node we need to make
|
|
|
5742 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
5743 * (we don't optimize for a partly shared list of siblings). */
|
|
|
5744 if (node != NULL && node->wn_refs > 1)
|
|
|
5745 {
|
|
|
5746 --node->wn_refs;
|
|
|
5747 copyprev = prev;
|
|
|
5748 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
5749 {
|
|
|
5750 /* Allocate a new node and copy the info. */
|
|
474
|
5751 np = get_wordnode(spin);
|
|
469
|
5752 if (np == NULL)
|
|
|
5753 return FAIL;
|
|
|
5754 np->wn_child = copyp->wn_child;
|
|
|
5755 if (np->wn_child != NULL)
|
|
|
5756 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
5757 np->wn_byte = copyp->wn_byte;
|
|
|
5758 if (np->wn_byte == NUL)
|
|
|
5759 {
|
|
|
5760 np->wn_flags = copyp->wn_flags;
|
|
|
5761 np->wn_region = copyp->wn_region;
|
|
481
|
5762 np->wn_affixID = copyp->wn_affixID;
|
|
469
|
5763 }
|
|
|
5764
|
|
|
5765 /* Link the new node in the list, there will be one ref. */
|
|
|
5766 np->wn_refs = 1;
|
|
|
5767 *copyprev = np;
|
|
|
5768 copyprev = &np->wn_sibling;
|
|
|
5769
|
|
|
5770 /* Let "node" point to the head of the copied list. */
|
|
|
5771 if (copyp == node)
|
|
|
5772 node = np;
|
|
|
5773 }
|
|
|
5774 }
|
|
|
5775
|
|
300
|
5776 /* Look for the sibling that has the same character. They are sorted
|
|
|
5777 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
5778 * higher byte value. For zero bytes (end of word) the sorting is
|
|
481
|
5779 * done on flags and then on affixID. */
|
|
339
|
5780 while (node != NULL
|
|
|
5781 && (node->wn_byte < word[i]
|
|
|
5782 || (node->wn_byte == NUL
|
|
|
5783 && (flags < 0
|
|
481
|
5784 ? node->wn_affixID < affixID
|
|
390
|
5785 : node->wn_flags < (flags & WN_MASK)
|
|
|
5786 || (node->wn_flags == (flags & WN_MASK)
|
|
481
|
5787 && node->wn_affixID < affixID)))))
|
|
236
|
5788 {
|
|
300
|
5789 prev = &node->wn_sibling;
|
|
|
5790 node = *prev;
|
|
236
|
5791 }
|
|
339
|
5792 if (node == NULL
|
|
|
5793 || node->wn_byte != word[i]
|
|
|
5794 || (word[i] == NUL
|
|
|
5795 && (flags < 0
|
|
390
|
5796 || node->wn_flags != (flags & WN_MASK)
|
|
481
|
5797 || node->wn_affixID != affixID)))
|
|
255
|
5798 {
|
|
300
|
5799 /* Allocate a new node. */
|
|
474
|
5800 np = get_wordnode(spin);
|
|
300
|
5801 if (np == NULL)
|
|
|
5802 return FAIL;
|
|
|
5803 np->wn_byte = word[i];
|
|
469
|
5804
|
|
|
5805 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
5806 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
5807 * make ref count of sibling one (matters when inserting in front
|
|
|
5808 * of the list of siblings). */
|
|
|
5809 if (node == NULL)
|
|
|
5810 np->wn_refs = 1;
|
|
|
5811 else
|
|
|
5812 {
|
|
|
5813 np->wn_refs = node->wn_refs;
|
|
|
5814 node->wn_refs = 1;
|
|
|
5815 }
|
|
300
|
5816 *prev = np;
|
|
|
5817 np->wn_sibling = node;
|
|
|
5818 node = np;
|
|
255
|
5819 }
|
|
300
|
5820
|
|
|
5821 if (word[i] == NUL)
|
|
|
5822 {
|
|
|
5823 node->wn_flags = flags;
|
|
|
5824 node->wn_region |= region;
|
|
481
|
5825 node->wn_affixID = affixID;
|
|
300
|
5826 break;
|
|
|
5827 }
|
|
|
5828 prev = &node->wn_child;
|
|
|
5829 node = *prev;
|
|
255
|
5830 }
|
|
469
|
5831 #ifdef SPELL_PRINTTREE
|
|
|
5832 smsg("Added \"%s\"", word);
|
|
|
5833 spell_print_tree(root->wn_sibling);
|
|
|
5834 #endif
|
|
|
5835
|
|
474
|
5836 /* count nr of words added since last message */
|
|
|
5837 ++spin->si_msg_count;
|
|
|
5838
|
|
482
|
5839 if (spin->si_compress_cnt > 1)
|
|
|
5840 {
|
|
|
5841 if (--spin->si_compress_cnt == 1)
|
|
|
5842 /* Did enough words to lower the block count limit. */
|
|
483
|
5843 spin->si_blocks_cnt += compress_inc;
|
|
482
|
5844 }
|
|
|
5845
|
|
469
|
5846 /*
|
|
482
|
5847 * When we have allocated lots of memory we need to compress the word tree
|
|
|
5848 * to free up some room. But compression is slow, and we might actually
|
|
|
5849 * need that room, thus only compress in the following situations:
|
|
|
5850 * 1. When not compressed before (si_compress_cnt == 0): when using
|
|
483
|
5851 * "compress_start" blocks.
|
|
|
5852 * 2. When compressed before and used "compress_inc" blocks before
|
|
|
5853 * adding "compress_added" words (si_compress_cnt > 1).
|
|
|
5854 * 3. When compressed before, added "compress_added" words
|
|
482
|
5855 * (si_compress_cnt == 1) and the number of free nodes drops below the
|
|
|
5856 * maximum word length.
|
|
469
|
5857 */
|
|
482
|
5858 #ifndef SPELL_PRINTTREE
|
|
|
5859 if (spin->si_compress_cnt == 1
|
|
|
5860 ? spin->si_free_count < MAXWLEN
|
|
483
|
5861 : spin->si_blocks_cnt >= compress_start)
|
|
482
|
5862 #endif
|
|
|
5863 {
|
|
|
5864 /* Decrement the block counter. The effect is that we compress again
|
|
483
|
5865 * when the freed up room has been used and another "compress_inc"
|
|
|
5866 * blocks have been allocated. Unless "compress_added" words have
|
|
482
|
5867 * been added, then the limit is put back again. */
|
|
483
|
5868 spin->si_blocks_cnt -= compress_inc;
|
|
|
5869 spin->si_compress_cnt = compress_added;
|
|
482
|
5870
|
|
474
|
5871 if (spin->si_verbose)
|
|
|
5872 {
|
|
|
5873 msg_start();
|
|
|
5874 msg_puts((char_u *)_(msg_compressing));
|
|
|
5875 msg_clr_eos();
|
|
|
5876 msg_didout = FALSE;
|
|
|
5877 msg_col = 0;
|
|
|
5878 out_flush();
|
|
|
5879 }
|
|
482
|
5880
|
|
|
5881 /* Compress both trees. Either they both have many nodes, which makes
|
|
|
5882 * compression useful, or one of them is small, which means
|
|
|
5883 * compression goes fast. */
|
|
|
5884 wordtree_compress(spin, spin->si_foldroot);
|
|
|
5885 wordtree_compress(spin, spin->si_keeproot);
|
|
469
|
5886 }
|
|
255
|
5887
|
|
|
5888 return OK;
|
|
236
|
5889 }
|
|
|
5890
|
|
469
|
5891 /*
|
|
483
|
5892 * Check the 'mkspellmem' option. Return FAIL if it's wrong.
|
|
|
5893 * Sets "sps_flags".
|
|
|
5894 */
|
|
|
5895 int
|
|
|
5896 spell_check_msm()
|
|
|
5897 {
|
|
|
5898 char_u *p = p_msm;
|
|
|
5899 long start = 0;
|
|
|
5900 long inc = 0;
|
|
|
5901 long added = 0;
|
|
|
5902
|
|
|
5903 if (!VIM_ISDIGIT(*p))
|
|
|
5904 return FAIL;
|
|
|
5905 /* block count = (value * 1024) / SBLOCKSIZE (but avoid overflow)*/
|
|
|
5906 start = (getdigits(&p) * 10) / (SBLOCKSIZE / 102);
|
|
|
5907 if (*p != ',')
|
|
|
5908 return FAIL;
|
|
|
5909 ++p;
|
|
|
5910 if (!VIM_ISDIGIT(*p))
|
|
|
5911 return FAIL;
|
|
|
5912 inc = (getdigits(&p) * 102) / (SBLOCKSIZE / 10);
|
|
|
5913 if (*p != ',')
|
|
|
5914 return FAIL;
|
|
|
5915 ++p;
|
|
|
5916 if (!VIM_ISDIGIT(*p))
|
|
|
5917 return FAIL;
|
|
|
5918 added = getdigits(&p) * 1024;
|
|
|
5919 if (*p != NUL)
|
|
|
5920 return FAIL;
|
|
|
5921
|
|
|
5922 if (start == 0 || inc == 0 || added == 0 || inc > start)
|
|
|
5923 return FAIL;
|
|
|
5924
|
|
|
5925 compress_start = start;
|
|
|
5926 compress_inc = inc;
|
|
|
5927 compress_added = added;
|
|
|
5928 return OK;
|
|
|
5929 }
|
|
|
5930
|
|
|
5931
|
|
|
5932 /*
|
|
469
|
5933 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
5934 * allocate a new one.
|
|
|
5935 */
|
|
|
5936 static wordnode_T *
|
|
474
|
5937 get_wordnode(spin)
|
|
|
5938 spellinfo_T *spin;
|
|
469
|
5939 {
|
|
|
5940 wordnode_T *n;
|
|
|
5941
|
|
474
|
5942 if (spin->si_first_free == NULL)
|
|
|
5943 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
5944 else
|
|
|
5945 {
|
|
474
|
5946 n = spin->si_first_free;
|
|
|
5947 spin->si_first_free = n->wn_child;
|
|
469
|
5948 vim_memset(n, 0, sizeof(wordnode_T));
|
|
482
|
5949 --spin->si_free_count;
|
|
469
|
5950 }
|
|
|
5951 #ifdef SPELL_PRINTTREE
|
|
474
|
5952 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
5953 #endif
|
|
|
5954 return n;
|
|
|
5955 }
|
|
|
5956
|
|
|
5957 /*
|
|
474
|
5958 * Decrement the reference count on a node (which is the head of a list of
|
|
|
5959 * siblings). If the reference count becomes zero free the node and its
|
|
|
5960 * siblings.
|
|
|
5961 */
|
|
|
5962 static void
|
|
|
5963 deref_wordnode(spin, node)
|
|
|
5964 spellinfo_T *spin;
|
|
|
5965 wordnode_T *node;
|
|
|
5966 {
|
|
|
5967 wordnode_T *np;
|
|
|
5968
|
|
|
5969 if (--node->wn_refs == 0)
|
|
|
5970 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5971 {
|
|
|
5972 if (np->wn_child != NULL)
|
|
|
5973 deref_wordnode(spin, np->wn_child);
|
|
|
5974 free_wordnode(spin, np);
|
|
|
5975 }
|
|
|
5976 }
|
|
|
5977
|
|
|
5978 /*
|
|
469
|
5979 * Free a wordnode_T for re-use later.
|
|
474
|
5980 * Only the "wn_child" field becomes invalid.
|
|
469
|
5981 */
|
|
|
5982 static void
|
|
474
|
5983 free_wordnode(spin, n)
|
|
|
5984 spellinfo_T *spin;
|
|
|
5985 wordnode_T *n;
|
|
|
5986 {
|
|
|
5987 n->wn_child = spin->si_first_free;
|
|
|
5988 spin->si_first_free = n;
|
|
482
|
5989 ++spin->si_free_count;
|
|
469
|
5990 }
|
|
|
5991
|
|
236
|
5992 /*
|
|
300
|
5993 * Compress a tree: find tails that are identical and can be shared.
|
|
|
5994 */
|
|
|
5995 static void
|
|
474
|
5996 wordtree_compress(spin, root)
|
|
|
5997 spellinfo_T *spin;
|
|
300
|
5998 wordnode_T *root;
|
|
|
5999 {
|
|
|
6000 hashtab_T ht;
|
|
|
6001 int n;
|
|
|
6002 int tot = 0;
|
|
469
|
6003 int perc;
|
|
300
|
6004
|
|
474
|
6005 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
6006 * start of the tree. */
|
|
|
6007 if (root->wn_sibling != NULL)
|
|
300
|
6008 {
|
|
|
6009 hash_init(&ht);
|
|
474
|
6010 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
6011
|
|
|
6012 #ifndef SPELL_PRINTTREE
|
|
310
|
6013 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
6014 #endif
|
|
310
|
6015 {
|
|
|
6016 if (!spin->si_verbose)
|
|
|
6017 verbose_enter();
|
|
469
|
6018 if (tot > 1000000)
|
|
|
6019 perc = (tot - n) / (tot / 100);
|
|
482
|
6020 else if (tot == 0)
|
|
|
6021 perc = 0;
|
|
469
|
6022 else
|
|
|
6023 perc = (tot - n) * 100 / tot;
|
|
310
|
6024 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
6025 n, tot, perc);
|
|
310
|
6026 if (p_verbose > 2)
|
|
|
6027 verbose_leave();
|
|
|
6028 }
|
|
469
|
6029 #ifdef SPELL_PRINTTREE
|
|
474
|
6030 spell_print_tree(root->wn_sibling);
|
|
469
|
6031 #endif
|
|
300
|
6032 hash_clear(&ht);
|
|
|
6033 }
|
|
|
6034 }
|
|
|
6035
|
|
|
6036 /*
|
|
|
6037 * Compress a node, its siblings and its children, depth first.
|
|
|
6038 * Returns the number of compressed nodes.
|
|
236
|
6039 */
|
|
255
|
6040 static int
|
|
474
|
6041 node_compress(spin, node, ht, tot)
|
|
|
6042 spellinfo_T *spin;
|
|
300
|
6043 wordnode_T *node;
|
|
|
6044 hashtab_T *ht;
|
|
|
6045 int *tot; /* total count of nodes before compressing,
|
|
|
6046 incremented while going through the tree */
|
|
236
|
6047 {
|
|
300
|
6048 wordnode_T *np;
|
|
|
6049 wordnode_T *tp;
|
|
|
6050 wordnode_T *child;
|
|
|
6051 hash_T hash;
|
|
236
|
6052 hashitem_T *hi;
|
|
300
|
6053 int len = 0;
|
|
|
6054 unsigned nr, n;
|
|
|
6055 int compressed = 0;
|
|
236
|
6056
|
|
300
|
6057 /*
|
|
|
6058 * Go through the list of siblings. Compress each child and then try
|
|
|
6059 * finding an identical child to replace it.
|
|
|
6060 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
6061 * but the whole list of siblings, of which the node is the first.
|
|
|
6062 */
|
|
482
|
6063 for (np = node; np != NULL && !got_int; np = np->wn_sibling)
|
|
236
|
6064 {
|
|
300
|
6065 ++len;
|
|
|
6066 if ((child = np->wn_child) != NULL)
|
|
|
6067 {
|
|
346
|
6068 /* Compress the child. This fills hashkey. */
|
|
474
|
6069 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
6070
|
|
|
6071 /* Try to find an identical child. */
|
|
346
|
6072 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
6073 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
6074 tp = NULL;
|
|
|
6075 if (!HASHITEM_EMPTY(hi))
|
|
|
6076 {
|
|
|
6077 /* There are children with an identical hash value. Now check
|
|
|
6078 * if there is one that is really identical. */
|
|
346
|
6079 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
6080 if (node_equal(child, tp))
|
|
|
6081 {
|
|
|
6082 /* Found one! Now use that child in place of the
|
|
469
|
6083 * current one. This means the current child and all
|
|
|
6084 * its siblings is unlinked from the tree. */
|
|
|
6085 ++tp->wn_refs;
|
|
474
|
6086 deref_wordnode(spin, child);
|
|
300
|
6087 np->wn_child = tp;
|
|
|
6088 ++compressed;
|
|
|
6089 break;
|
|
|
6090 }
|
|
|
6091 if (tp == NULL)
|
|
|
6092 {
|
|
|
6093 /* No other child with this hash value equals the child of
|
|
|
6094 * the node, add it to the linked list after the first
|
|
|
6095 * item. */
|
|
|
6096 tp = HI2WN(hi);
|
|
346
|
6097 child->wn_u2.next = tp->wn_u2.next;
|
|
|
6098 tp->wn_u2.next = child;
|
|
300
|
6099 }
|
|
|
6100 }
|
|
|
6101 else
|
|
|
6102 /* No other child has this hash value, add it to the
|
|
|
6103 * hashtable. */
|
|
346
|
6104 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
6105 }
|
|
236
|
6106 }
|
|
300
|
6107 *tot += len;
|
|
|
6108
|
|
|
6109 /*
|
|
|
6110 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
6111 * find a lookalike node. This must be done after compressing the sibling
|
|
|
6112 * list, otherwise the hash key would become invalid by the compression.
|
|
|
6113 */
|
|
346
|
6114 node->wn_u1.hashkey[0] = len;
|
|
300
|
6115 nr = 0;
|
|
|
6116 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
6117 {
|
|
300
|
6118 if (np->wn_byte == NUL)
|
|
481
|
6119 /* end node: use wn_flags, wn_region and wn_affixID */
|
|
|
6120 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16);
|
|
300
|
6121 else
|
|
|
6122 /* byte node: use the byte value and the child pointer */
|
|
|
6123 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
6124 nr = nr * 101 + n;
|
|
236
|
6125 }
|
|
300
|
6126
|
|
|
6127 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
6128 n = nr & 0xff;
|
|
346
|
6129 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
6130 n = (nr >> 8) & 0xff;
|
|
346
|
6131 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
6132 n = (nr >> 16) & 0xff;
|
|
346
|
6133 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
6134 n = (nr >> 24) & 0xff;
|
|
346
|
6135 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
6136 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
6137
|
|
482
|
6138 /* Check for CTRL-C pressed now and then. */
|
|
|
6139 fast_breakcheck();
|
|
|
6140
|
|
300
|
6141 return compressed;
|
|
|
6142 }
|
|
|
6143
|
|
|
6144 /*
|
|
|
6145 * Return TRUE when two nodes have identical siblings and children.
|
|
|
6146 */
|
|
|
6147 static int
|
|
|
6148 node_equal(n1, n2)
|
|
|
6149 wordnode_T *n1;
|
|
|
6150 wordnode_T *n2;
|
|
|
6151 {
|
|
|
6152 wordnode_T *p1;
|
|
|
6153 wordnode_T *p2;
|
|
|
6154
|
|
|
6155 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
6156 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
6157 if (p1->wn_byte != p2->wn_byte
|
|
|
6158 || (p1->wn_byte == NUL
|
|
|
6159 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
6160 || p1->wn_region != p2->wn_region
|
|
481
|
6161 || p1->wn_affixID != p2->wn_affixID)
|
|
300
|
6162 : (p1->wn_child != p2->wn_child)))
|
|
|
6163 break;
|
|
|
6164
|
|
|
6165 return p1 == NULL && p2 == NULL;
|
|
236
|
6166 }
|
|
|
6167
|
|
|
6168 /*
|
|
|
6169 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
6170 */
|
|
255
|
6171 void
|
|
236
|
6172 put_bytes(fd, nr, len)
|
|
|
6173 FILE *fd;
|
|
|
6174 long_u nr;
|
|
|
6175 int len;
|
|
|
6176 {
|
|
|
6177 int i;
|
|
|
6178
|
|
|
6179 for (i = len - 1; i >= 0; --i)
|
|
|
6180 putc((int)(nr >> (i * 8)), fd);
|
|
|
6181 }
|
|
|
6182
|
|
323
|
6183 static int
|
|
|
6184 #ifdef __BORLANDC__
|
|
|
6185 _RTLENTRYF
|
|
|
6186 #endif
|
|
|
6187 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
6188
|
|
|
6189 /*
|
|
|
6190 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
6191 */
|
|
|
6192 static int
|
|
|
6193 #ifdef __BORLANDC__
|
|
|
6194 _RTLENTRYF
|
|
|
6195 #endif
|
|
|
6196 rep_compare(s1, s2)
|
|
|
6197 const void *s1;
|
|
|
6198 const void *s2;
|
|
|
6199 {
|
|
|
6200 fromto_T *p1 = (fromto_T *)s1;
|
|
|
6201 fromto_T *p2 = (fromto_T *)s2;
|
|
|
6202
|
|
|
6203 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
6204 }
|
|
|
6205
|
|
236
|
6206 /*
|
|
483
|
6207 * Write the Vim .spl file "fname".
|
|
236
|
6208 */
|
|
|
6209 static void
|
|
474
|
6210 write_vim_spell(spin, fname)
|
|
|
6211 spellinfo_T *spin;
|
|
236
|
6212 char_u *fname;
|
|
|
6213 {
|
|
300
|
6214 FILE *fd;
|
|
|
6215 int regionmask;
|
|
236
|
6216 int round;
|
|
300
|
6217 wordnode_T *tree;
|
|
|
6218 int nodecount;
|
|
323
|
6219 int i;
|
|
|
6220 int l;
|
|
|
6221 garray_T *gap;
|
|
|
6222 fromto_T *ftp;
|
|
|
6223 char_u *p;
|
|
|
6224 int rr;
|
|
236
|
6225
|
|
310
|
6226 fd = mch_fopen((char *)fname, "w");
|
|
300
|
6227 if (fd == NULL)
|
|
236
|
6228 {
|
|
|
6229 EMSG2(_(e_notopen), fname);
|
|
|
6230 return;
|
|
|
6231 }
|
|
|
6232
|
|
483
|
6233 /* <HEADER>: <fileID> <versionnr> */
|
|
300
|
6234 /* <fileID> */
|
|
|
6235 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
6236 EMSG(_(e_write));
|
|
483
|
6237 putc(VIMSPELLVERSION, fd); /* <versionnr> */
|
|
|
6238
|
|
|
6239 /*
|
|
|
6240 * <SECTIONS>: <section> ... <sectionend>
|
|
|
6241 */
|
|
|
6242
|
|
|
6243 /* SN_REGION: <regionname> ...
|
|
|
6244 * Write the region names only if there is more than one. */
|
|
316
|
6245 if (spin->si_region_count > 1)
|
|
236
|
6246 {
|
|
483
|
6247 putc(SN_REGION, fd); /* <sectionID> */
|
|
|
6248 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6249 l = spin->si_region_count * 2;
|
|
|
6250 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6251 fwrite(spin->si_region_name, (size_t)l, (size_t)1, fd);
|
|
|
6252 /* <regionname> ... */
|
|
316
|
6253 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
6254 }
|
|
|
6255 else
|
|
300
|
6256 regionmask = 0;
|
|
483
|
6257
|
|
|
6258 /* SN_CHARFLAGS: <charflagslen> <charflags> <folcharslen> <folchars>
|
|
|
6259 *
|
|
|
6260 * The table with character flags and the table for case folding.
|
|
|
6261 * This makes sure the same characters are recognized as word characters
|
|
|
6262 * when generating an when using a spell file.
|
|
260
|
6263 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
6264 * 'encoding'.
|
|
|
6265 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
6266 * the table (avoids that it conflicts). File is shorter too.
|
|
|
6267 */
|
|
483
|
6268 if (!spin->si_ascii && !spin->si_add)
|
|
|
6269 {
|
|
|
6270 char_u folchars[128 * 8];
|
|
|
6271 int flags;
|
|
|
6272
|
|
485
|
6273 putc(SN_CHARFLAGS, fd); /* <sectionID> */
|
|
483
|
6274 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6275
|
|
|
6276 /* Form the <folchars> string first, we need to know its length. */
|
|
|
6277 l = 0;
|
|
|
6278 for (i = 128; i < 256; ++i)
|
|
|
6279 {
|
|
|
6280 #ifdef FEAT_MBYTE
|
|
|
6281 if (has_mbyte)
|
|
|
6282 l += mb_char2bytes(spelltab.st_fold[i], folchars + l);
|
|
|
6283 else
|
|
|
6284 #endif
|
|
|
6285 folchars[l++] = spelltab.st_fold[i];
|
|
|
6286 }
|
|
|
6287 put_bytes(fd, (long_u)(1 + 128 + 2 + l), 4); /* <sectionlen> */
|
|
|
6288
|
|
|
6289 fputc(128, fd); /* <charflagslen> */
|
|
|
6290 for (i = 128; i < 256; ++i)
|
|
|
6291 {
|
|
|
6292 flags = 0;
|
|
|
6293 if (spelltab.st_isw[i])
|
|
|
6294 flags |= CF_WORD;
|
|
|
6295 if (spelltab.st_isu[i])
|
|
|
6296 flags |= CF_UPPER;
|
|
|
6297 fputc(flags, fd); /* <charflags> */
|
|
|
6298 }
|
|
|
6299
|
|
|
6300 put_bytes(fd, (long_u)l, 2); /* <folcharslen> */
|
|
|
6301 fwrite(folchars, (size_t)l, (size_t)1, fd); /* <folchars> */
|
|
|
6302 }
|
|
|
6303
|
|
|
6304 /* SN_MIDWORD: <midword> */
|
|
|
6305 if (spin->si_midword != NULL)
|
|
|
6306 {
|
|
|
6307 putc(SN_MIDWORD, fd); /* <sectionID> */
|
|
|
6308 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6309
|
|
366
|
6310 i = STRLEN(spin->si_midword);
|
|
483
|
6311 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */
|
|
366
|
6312 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
6313 }
|
|
|
6314
|
|
483
|
6315 /* SN_PREFCOND: <prefcondcnt> <prefcond> ... */
|
|
|
6316 if (spin->si_prefcond.ga_len > 0)
|
|
|
6317 {
|
|
|
6318 putc(SN_PREFCOND, fd); /* <sectionID> */
|
|
|
6319 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6320
|
|
|
6321 l = write_spell_prefcond(NULL, &spin->si_prefcond);
|
|
|
6322 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6323
|
|
|
6324 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
6325 }
|
|
|
6326
|
|
|
6327 /* SN_REP: <repcount> <rep> ...
|
|
|
6328 * SN_SAL: <salflags> <salcount> <sal> ... */
|
|
366
|
6329
|
|
323
|
6330 /* Sort the REP items. */
|
|
|
6331 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
6332 sizeof(fromto_T), rep_compare);
|
|
|
6333
|
|
483
|
6334 /* round 1: SN_REP section
|
|
|
6335 * round 2: SN_SAL section (unless SN_SOFO is used) */
|
|
323
|
6336 for (round = 1; round <= 2; ++round)
|
|
|
6337 {
|
|
|
6338 if (round == 1)
|
|
483
|
6339 {
|
|
323
|
6340 gap = &spin->si_rep;
|
|
483
|
6341 putc(SN_REP, fd); /* <sectionID> */
|
|
|
6342 }
|
|
323
|
6343 else
|
|
|
6344 {
|
|
483
|
6345 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
6346 /* using SN_SOFO section instead of SN_SAL */
|
|
|
6347 break;
|
|
323
|
6348 gap = &spin->si_sal;
|
|
483
|
6349 putc(SN_SAL, fd); /* <sectionID> */
|
|
|
6350 }
|
|
|
6351
|
|
|
6352 /* This is for making suggestions, section is not required. */
|
|
|
6353 putc(0, fd); /* <sectionflags> */
|
|
|
6354
|
|
|
6355 /* Compute the length of what follows. */
|
|
|
6356 l = 2; /* count <repcount> or <salcount> */
|
|
|
6357 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6358 {
|
|
|
6359 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
6360 l += 1 + STRLEN(ftp->ft_from); /* count <*fromlen> and <*from> */
|
|
|
6361 l += 1 + STRLEN(ftp->ft_to); /* count <*tolen> and <*to> */
|
|
|
6362 }
|
|
|
6363 if (round == 2)
|
|
|
6364 ++l; /* count <salflags> */
|
|
|
6365 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6366
|
|
|
6367 if (round == 2)
|
|
|
6368 {
|
|
323
|
6369 i = 0;
|
|
|
6370 if (spin->si_followup)
|
|
|
6371 i |= SAL_F0LLOWUP;
|
|
|
6372 if (spin->si_collapse)
|
|
|
6373 i |= SAL_COLLAPSE;
|
|
|
6374 if (spin->si_rem_accents)
|
|
|
6375 i |= SAL_REM_ACCENTS;
|
|
|
6376 putc(i, fd); /* <salflags> */
|
|
|
6377 }
|
|
|
6378
|
|
|
6379 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
6380 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6381 {
|
|
|
6382 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
6383 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
6384 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
6385 for (rr = 1; rr <= 2; ++rr)
|
|
|
6386 {
|
|
|
6387 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
6388 l = STRLEN(p);
|
|
|
6389 putc(l, fd);
|
|
|
6390 fwrite(p, l, (size_t)1, fd);
|
|
|
6391 }
|
|
|
6392 }
|
|
483
|
6393
|
|
|
6394 }
|
|
|
6395
|
|
|
6396 /* SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
6397 * This is for making suggestions, section is not required. */
|
|
375
|
6398 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
6399 {
|
|
483
|
6400 putc(SN_SOFO, fd); /* <sectionID> */
|
|
|
6401 putc(0, fd); /* <sectionflags> */
|
|
375
|
6402
|
|
|
6403 l = STRLEN(spin->si_sofofr);
|
|
483
|
6404 put_bytes(fd, (long_u)(l + STRLEN(spin->si_sofoto) + 4), 4);
|
|
|
6405 /* <sectionlen> */
|
|
|
6406
|
|
|
6407 put_bytes(fd, (long_u)l, 2); /* <sofofromlen> */
|
|
|
6408 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <sofofrom> */
|
|
375
|
6409
|
|
|
6410 l = STRLEN(spin->si_sofoto);
|
|
483
|
6411 put_bytes(fd, (long_u)l, 2); /* <sofotolen> */
|
|
|
6412 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <sofoto> */
|
|
|
6413 }
|
|
|
6414
|
|
|
6415 /* SN_MAP: <mapstr>
|
|
|
6416 * This is for making suggestions, section is not required. */
|
|
|
6417 if (spin->si_map.ga_len > 0)
|
|
|
6418 {
|
|
|
6419 putc(SN_MAP, fd); /* <sectionID> */
|
|
|
6420 putc(0, fd); /* <sectionflags> */
|
|
|
6421 l = spin->si_map.ga_len;
|
|
|
6422 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6423 fwrite(spin->si_map.ga_data, (size_t)l, (size_t)1, fd);
|
|
|
6424 /* <mapstr> */
|
|
|
6425 }
|
|
|
6426
|
|
|
6427 /* SN_COMPOUND: compound info.
|
|
|
6428 * We don't mark it required, when not supported all compound words will
|
|
|
6429 * be bad words. */
|
|
|
6430 if (spin->si_compflags != NULL)
|
|
|
6431 {
|
|
|
6432 putc(SN_COMPOUND, fd); /* <sectionID> */
|
|
|
6433 putc(0, fd); /* <sectionflags> */
|
|
|
6434
|
|
|
6435 l = STRLEN(spin->si_compflags);
|
|
|
6436 put_bytes(fd, (long_u)(l + 3), 4); /* <sectionlen> */
|
|
|
6437 putc(spin->si_compmax, fd); /* <compmax> */
|
|
|
6438 putc(spin->si_compminlen, fd); /* <compminlen> */
|
|
|
6439 putc(spin->si_compsylmax, fd); /* <compsylmax> */
|
|
|
6440 /* <compflags> */
|
|
|
6441 fwrite(spin->si_compflags, (size_t)l, (size_t)1, fd);
|
|
|
6442 }
|
|
|
6443
|
|
|
6444 /* SN_SYLLABLE: syllable info.
|
|
|
6445 * We don't mark it required, when not supported syllables will not be
|
|
|
6446 * counted. */
|
|
|
6447 if (spin->si_syllable != NULL)
|
|
|
6448 {
|
|
|
6449 putc(SN_SYLLABLE, fd); /* <sectionID> */
|
|
|
6450 putc(0, fd); /* <sectionflags> */
|
|
|
6451
|
|
|
6452 l = STRLEN(spin->si_syllable);
|
|
|
6453 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6454 fwrite(spin->si_syllable, (size_t)l, (size_t)1, fd); /* <syllable> */
|
|
|
6455 }
|
|
|
6456
|
|
|
6457 /* end of <SECTIONS> */
|
|
|
6458 putc(SN_END, fd); /* <sectionend> */
|
|
|
6459
|
|
302
|
6460
|
|
236
|
6461 /*
|
|
339
|
6462 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
6463 */
|
|
323
|
6464 spin->si_memtot = 0;
|
|
339
|
6465 for (round = 1; round <= 3; ++round)
|
|
236
|
6466 {
|
|
339
|
6467 if (round == 1)
|
|
469
|
6468 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
6469 else if (round == 2)
|
|
469
|
6470 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
6471 else
|
|
469
|
6472 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
6473
|
|
346
|
6474 /* Clear the index and wnode fields in the tree. */
|
|
|
6475 clear_node(tree);
|
|
|
6476
|
|
300
|
6477 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
6478 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
6479 * nodes. */
|
|
346
|
6480 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
6481
|
|
300
|
6482 /* number of nodes in 4 bytes */
|
|
|
6483 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
6484 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
6485
|
|
300
|
6486 /* Write the nodes. */
|
|
346
|
6487 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
6488 }
|
|
|
6489
|
|
300
|
6490 fclose(fd);
|
|
236
|
6491 }
|
|
|
6492
|
|
|
6493 /*
|
|
346
|
6494 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
6495 * children. This is needed because they are a union with other items to save
|
|
|
6496 * space.
|
|
|
6497 */
|
|
|
6498 static void
|
|
|
6499 clear_node(node)
|
|
|
6500 wordnode_T *node;
|
|
|
6501 {
|
|
|
6502 wordnode_T *np;
|
|
|
6503
|
|
|
6504 if (node != NULL)
|
|
|
6505 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6506 {
|
|
|
6507 np->wn_u1.index = 0;
|
|
|
6508 np->wn_u2.wnode = NULL;
|
|
|
6509
|
|
|
6510 if (np->wn_byte != NUL)
|
|
|
6511 clear_node(np->wn_child);
|
|
|
6512 }
|
|
|
6513 }
|
|
|
6514
|
|
|
6515
|
|
|
6516 /*
|
|
300
|
6517 * Dump a word tree at node "node".
|
|
|
6518 *
|
|
|
6519 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
6520 * byte recursively write the children.
|
|
|
6521 *
|
|
|
6522 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
6523 * are made about the indexes (so that we don't have to write them in the
|
|
|
6524 * file).
|
|
236
|
6525 *
|
|
300
|
6526 * Returns the number of nodes used.
|
|
236
|
6527 */
|
|
300
|
6528 static int
|
|
346
|
6529 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
6530 FILE *fd; /* NULL when only counting */
|
|
300
|
6531 wordnode_T *node;
|
|
|
6532 int index;
|
|
|
6533 int regionmask;
|
|
339
|
6534 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
6535 {
|
|
300
|
6536 int newindex = index;
|
|
|
6537 int siblingcount = 0;
|
|
|
6538 wordnode_T *np;
|
|
236
|
6539 int flags;
|
|
300
|
6540
|
|
|
6541 /* If "node" is zero the tree is empty. */
|
|
|
6542 if (node == NULL)
|
|
|
6543 return 0;
|
|
|
6544
|
|
|
6545 /* Store the index where this node is written. */
|
|
346
|
6546 node->wn_u1.index = index;
|
|
236
|
6547
|
|
300
|
6548 /* Count the number of siblings. */
|
|
|
6549 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6550 ++siblingcount;
|
|
236
|
6551
|
|
300
|
6552 /* Write the sibling count. */
|
|
|
6553 if (fd != NULL)
|
|
|
6554 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
6555
|
|
300
|
6556 /* Write each sibling byte and optionally extra info. */
|
|
|
6557 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
6558 {
|
|
300
|
6559 if (np->wn_byte == 0)
|
|
|
6560 {
|
|
|
6561 if (fd != NULL)
|
|
|
6562 {
|
|
339
|
6563 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
6564 if (prefixtree)
|
|
300
|
6565 {
|
|
481
|
6566 /* In PREFIXTREE write the required affixID and the
|
|
390
|
6567 * associated condition nr (stored in wn_region). The
|
|
|
6568 * byte value is misused to store the "rare" and "not
|
|
|
6569 * combining" flags */
|
|
455
|
6570 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
6571 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
6572 else
|
|
|
6573 {
|
|
390
|
6574 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
6575 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
6576 }
|
|
481
|
6577 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
6578 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
6579 }
|
|
|
6580 else
|
|
|
6581 {
|
|
339
|
6582 /* For word trees we write the flag/region items. */
|
|
|
6583 flags = np->wn_flags;
|
|
|
6584 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
6585 flags |= WF_REGION;
|
|
481
|
6586 if (np->wn_affixID != 0)
|
|
|
6587 flags |= WF_AFX;
|
|
339
|
6588 if (flags == 0)
|
|
|
6589 {
|
|
|
6590 /* word without flags or region */
|
|
|
6591 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
6592 }
|
|
|
6593 else
|
|
|
6594 {
|
|
390
|
6595 if (np->wn_flags >= 0x100)
|
|
|
6596 {
|
|
|
6597 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
6598 putc(flags, fd); /* <flags> */
|
|
|
6599 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
6600 }
|
|
|
6601 else
|
|
|
6602 {
|
|
|
6603 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
6604 putc(flags, fd); /* <flags> */
|
|
|
6605 }
|
|
339
|
6606 if (flags & WF_REGION)
|
|
|
6607 putc(np->wn_region, fd); /* <region> */
|
|
481
|
6608 if (flags & WF_AFX)
|
|
|
6609 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
6610 }
|
|
300
|
6611 }
|
|
|
6612 }
|
|
|
6613 }
|
|
|
6614 else
|
|
|
6615 {
|
|
346
|
6616 if (np->wn_child->wn_u1.index != 0
|
|
|
6617 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
6618 {
|
|
|
6619 /* The child is written elsewhere, write the reference. */
|
|
|
6620 if (fd != NULL)
|
|
|
6621 {
|
|
|
6622 putc(BY_INDEX, fd); /* <byte> */
|
|
|
6623 /* <nodeidx> */
|
|
346
|
6624 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
6625 }
|
|
|
6626 }
|
|
346
|
6627 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
6628 /* We will write the child below and give it an index. */
|
|
346
|
6629 np->wn_child->wn_u2.wnode = node;
|
|
236
|
6630
|
|
300
|
6631 if (fd != NULL)
|
|
|
6632 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
6633 {
|
|
|
6634 EMSG(_(e_write));
|
|
|
6635 return 0;
|
|
|
6636 }
|
|
|
6637 }
|
|
236
|
6638 }
|
|
|
6639
|
|
300
|
6640 /* Space used in the array when reading: one for each sibling and one for
|
|
|
6641 * the count. */
|
|
|
6642 newindex += siblingcount + 1;
|
|
249
|
6643
|
|
300
|
6644 /* Recursively dump the children of each sibling. */
|
|
|
6645 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
6646 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
6647 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
6648 prefixtree);
|
|
249
|
6649
|
|
300
|
6650 return newindex;
|
|
236
|
6651 }
|
|
|
6652
|
|
|
6653
|
|
|
6654 /*
|
|
310
|
6655 * ":mkspell [-ascii] outfile infile ..."
|
|
|
6656 * ":mkspell [-ascii] addfile"
|
|
236
|
6657 */
|
|
|
6658 void
|
|
|
6659 ex_mkspell(eap)
|
|
|
6660 exarg_T *eap;
|
|
|
6661 {
|
|
|
6662 int fcount;
|
|
|
6663 char_u **fnames;
|
|
310
|
6664 char_u *arg = eap->arg;
|
|
|
6665 int ascii = FALSE;
|
|
|
6666
|
|
|
6667 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
6668 {
|
|
|
6669 ascii = TRUE;
|
|
|
6670 arg = skipwhite(arg + 6);
|
|
|
6671 }
|
|
|
6672
|
|
|
6673 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
6674 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
6675 {
|
|
323
|
6676 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
6677 FreeWild(fcount, fnames);
|
|
|
6678 }
|
|
|
6679 }
|
|
|
6680
|
|
|
6681 /*
|
|
|
6682 * Create a Vim spell file from one or more word lists.
|
|
|
6683 * "fnames[0]" is the output file name.
|
|
|
6684 * "fnames[fcount - 1]" is the last input file name.
|
|
|
6685 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
6686 * and ".spl" is appended to make the output file name.
|
|
|
6687 */
|
|
|
6688 static void
|
|
323
|
6689 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
6690 int fcount;
|
|
|
6691 char_u **fnames;
|
|
|
6692 int ascii; /* -ascii argument given */
|
|
|
6693 int overwrite; /* overwrite existing output file */
|
|
323
|
6694 int added_word; /* invoked through "zg" */
|
|
310
|
6695 {
|
|
236
|
6696 char_u fname[MAXPATHL];
|
|
|
6697 char_u wfname[MAXPATHL];
|
|
310
|
6698 char_u **innames;
|
|
|
6699 int incount;
|
|
236
|
6700 afffile_T *(afile[8]);
|
|
|
6701 int i;
|
|
|
6702 int len;
|
|
|
6703 struct stat st;
|
|
255
|
6704 int error = FALSE;
|
|
300
|
6705 spellinfo_T spin;
|
|
|
6706
|
|
|
6707 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
6708 spin.si_verbose = !added_word;
|
|
310
|
6709 spin.si_ascii = ascii;
|
|
323
|
6710 spin.si_followup = TRUE;
|
|
|
6711 spin.si_rem_accents = TRUE;
|
|
|
6712 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
6713 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
6714 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
6715 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
6716
|
|
|
6717 /* default: fnames[0] is output file, following are input files */
|
|
|
6718 innames = &fnames[1];
|
|
|
6719 incount = fcount - 1;
|
|
|
6720
|
|
|
6721 if (fcount >= 1)
|
|
240
|
6722 {
|
|
310
|
6723 len = STRLEN(fnames[0]);
|
|
|
6724 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
6725 {
|
|
|
6726 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
6727 * "path/en.latin1.add.spl". */
|
|
|
6728 innames = &fnames[0];
|
|
|
6729 incount = 1;
|
|
|
6730 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
6731 }
|
|
366
|
6732 else if (fcount == 1)
|
|
|
6733 {
|
|
|
6734 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
6735 innames = &fnames[0];
|
|
|
6736 incount = 1;
|
|
|
6737 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
6738 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
6739 }
|
|
310
|
6740 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
6741 {
|
|
|
6742 /* Name ends in ".spl", use as the file name. */
|
|
323
|
6743 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
6744 }
|
|
|
6745 else
|
|
|
6746 /* Name should be language, make the file name from it. */
|
|
|
6747 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
6748 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
6749
|
|
|
6750 /* Check for .ascii.spl. */
|
|
|
6751 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
6752 spin.si_ascii = TRUE;
|
|
|
6753
|
|
|
6754 /* Check for .add.spl. */
|
|
|
6755 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
6756 spin.si_add = TRUE;
|
|
240
|
6757 }
|
|
|
6758
|
|
310
|
6759 if (incount <= 0)
|
|
236
|
6760 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
6761 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
6762 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
6763 else if (incount > 8)
|
|
236
|
6764 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
6765 else
|
|
|
6766 {
|
|
|
6767 /* Check for overwriting before doing things that may take a lot of
|
|
|
6768 * time. */
|
|
310
|
6769 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
6770 {
|
|
|
6771 EMSG(_(e_exists));
|
|
310
|
6772 return;
|
|
236
|
6773 }
|
|
310
|
6774 if (mch_isdir(wfname))
|
|
236
|
6775 {
|
|
310
|
6776 EMSG2(_(e_isadir2), wfname);
|
|
|
6777 return;
|
|
236
|
6778 }
|
|
|
6779
|
|
|
6780 /*
|
|
|
6781 * Init the aff and dic pointers.
|
|
|
6782 * Get the region names if there are more than 2 arguments.
|
|
|
6783 */
|
|
310
|
6784 for (i = 0; i < incount; ++i)
|
|
236
|
6785 {
|
|
310
|
6786 afile[i] = NULL;
|
|
300
|
6787
|
|
316
|
6788 if (incount > 1)
|
|
236
|
6789 {
|
|
310
|
6790 len = STRLEN(innames[i]);
|
|
|
6791 if (STRLEN(gettail(innames[i])) < 5
|
|
|
6792 || innames[i][len - 3] != '_')
|
|
236
|
6793 {
|
|
310
|
6794 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
6795 return;
|
|
236
|
6796 }
|
|
316
|
6797 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
6798 spin.si_region_name[i * 2 + 1] =
|
|
|
6799 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
6800 }
|
|
|
6801 }
|
|
316
|
6802 spin.si_region_count = incount;
|
|
236
|
6803
|
|
474
|
6804 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
6805 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
6806 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
6807 if (spin.si_foldroot == NULL
|
|
|
6808 || spin.si_keeproot == NULL
|
|
|
6809 || spin.si_prefroot == NULL)
|
|
300
|
6810 {
|
|
469
|
6811 free_blocks(spin.si_blocks);
|
|
310
|
6812 return;
|
|
300
|
6813 }
|
|
|
6814
|
|
351
|
6815 /* When not producing a .add.spl file clear the character table when
|
|
|
6816 * we encounter one in the .aff file. This means we dump the current
|
|
|
6817 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
6818 * better than guessing the contents, the table will match a
|
|
|
6819 * previously loaded spell file. */
|
|
|
6820 if (!spin.si_add)
|
|
|
6821 spin.si_clear_chartab = TRUE;
|
|
|
6822
|
|
236
|
6823 /*
|
|
|
6824 * Read all the .aff and .dic files.
|
|
|
6825 * Text is converted to 'encoding'.
|
|
300
|
6826 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
6827 */
|
|
310
|
6828 for (i = 0; i < incount && !error; ++i)
|
|
236
|
6829 {
|
|
300
|
6830 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
6831 spin.si_region = 1 << i;
|
|
|
6832
|
|
|
6833 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
6834 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
6835 {
|
|
|
6836 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
6837 * "SET" line. */
|
|
474
|
6838 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
6839 if (afile[i] == NULL)
|
|
300
|
6840 error = TRUE;
|
|
|
6841 else
|
|
|
6842 {
|
|
|
6843 /* Read the .dic file and store the words in the trees. */
|
|
|
6844 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
6845 innames[i]);
|
|
474
|
6846 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
6847 error = TRUE;
|
|
|
6848 }
|
|
|
6849 }
|
|
|
6850 else
|
|
|
6851 {
|
|
|
6852 /* No .aff file, try reading the file as a word list. Store
|
|
|
6853 * the words in the trees. */
|
|
474
|
6854 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
6855 error = TRUE;
|
|
|
6856 }
|
|
236
|
6857
|
|
310
|
6858 #ifdef FEAT_MBYTE
|
|
236
|
6859 /* Free any conversion stuff. */
|
|
300
|
6860 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
6861 #endif
|
|
236
|
6862 }
|
|
|
6863
|
|
300
|
6864 if (!error)
|
|
236
|
6865 {
|
|
|
6866 /*
|
|
300
|
6867 * Combine tails in the tree.
|
|
236
|
6868 */
|
|
474
|
6869 if (spin.si_verbose || p_verbose > 2)
|
|
|
6870 {
|
|
|
6871 if (!spin.si_verbose)
|
|
310
|
6872 verbose_enter();
|
|
469
|
6873 MSG(_(msg_compressing));
|
|
310
|
6874 out_flush();
|
|
474
|
6875 if (!spin.si_verbose)
|
|
310
|
6876 verbose_leave();
|
|
|
6877 }
|
|
474
|
6878 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
6879 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
6880 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
6881 }
|
|
|
6882
|
|
300
|
6883 if (!error)
|
|
|
6884 {
|
|
|
6885 /*
|
|
|
6886 * Write the info in the spell file.
|
|
|
6887 */
|
|
474
|
6888 if (spin.si_verbose || p_verbose > 2)
|
|
|
6889 {
|
|
|
6890 if (!spin.si_verbose)
|
|
310
|
6891 verbose_enter();
|
|
366
|
6892 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
6893 out_flush();
|
|
474
|
6894 if (!spin.si_verbose)
|
|
310
|
6895 verbose_leave();
|
|
|
6896 }
|
|
|
6897
|
|
474
|
6898 write_vim_spell(&spin, wfname);
|
|
|
6899
|
|
|
6900 if (spin.si_verbose || p_verbose > 2)
|
|
|
6901 {
|
|
|
6902 if (!spin.si_verbose)
|
|
310
|
6903 verbose_enter();
|
|
|
6904 MSG(_("Done!"));
|
|
|
6905 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
6906 spin.si_memtot);
|
|
310
|
6907 out_flush();
|
|
474
|
6908 if (!spin.si_verbose)
|
|
310
|
6909 verbose_leave();
|
|
|
6910 }
|
|
|
6911
|
|
|
6912 /* If the file is loaded need to reload it. */
|
|
323
|
6913 spell_reload_one(wfname, added_word);
|
|
300
|
6914 }
|
|
|
6915
|
|
|
6916 /* Free the allocated memory. */
|
|
323
|
6917 ga_clear(&spin.si_rep);
|
|
|
6918 ga_clear(&spin.si_sal);
|
|
|
6919 ga_clear(&spin.si_map);
|
|
339
|
6920 ga_clear(&spin.si_prefcond);
|
|
366
|
6921 vim_free(spin.si_midword);
|
|
375
|
6922 vim_free(spin.si_sofofr);
|
|
|
6923 vim_free(spin.si_sofoto);
|
|
300
|
6924
|
|
|
6925 /* Free the .aff file structures. */
|
|
310
|
6926 for (i = 0; i < incount; ++i)
|
|
|
6927 if (afile[i] != NULL)
|
|
|
6928 spell_free_aff(afile[i]);
|
|
339
|
6929
|
|
|
6930 /* Free all the bits and pieces at once. */
|
|
|
6931 free_blocks(spin.si_blocks);
|
|
236
|
6932 }
|
|
310
|
6933 }
|
|
|
6934
|
|
|
6935
|
|
|
6936 /*
|
|
385
|
6937 * ":[count]spellgood {word}"
|
|
|
6938 * ":[count]spellwrong {word}"
|
|
310
|
6939 */
|
|
|
6940 void
|
|
|
6941 ex_spell(eap)
|
|
|
6942 exarg_T *eap;
|
|
|
6943 {
|
|
381
|
6944 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
6945 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
6946 }
|
|
|
6947
|
|
310
|
6948 /*
|
|
|
6949 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
6950 */
|
|
|
6951 void
|
|
385
|
6952 spell_add_word(word, len, bad, index)
|
|
310
|
6953 char_u *word;
|
|
|
6954 int len;
|
|
|
6955 int bad;
|
|
385
|
6956 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
6957 'spellfile' */
|
|
310
|
6958 {
|
|
|
6959 FILE *fd;
|
|
385
|
6960 buf_T *buf = NULL;
|
|
351
|
6961 int new_spf = FALSE;
|
|
|
6962 struct stat st;
|
|
381
|
6963 char_u *fname;
|
|
385
|
6964 char_u fnamebuf[MAXPATHL];
|
|
|
6965 char_u line[MAXWLEN * 2];
|
|
|
6966 long fpos, fpos_next = 0;
|
|
|
6967 int i;
|
|
|
6968 char_u *spf;
|
|
|
6969
|
|
|
6970 if (index == 0) /* use internal wordlist */
|
|
|
6971 {
|
|
|
6972 if (int_wordlist == NULL)
|
|
|
6973 {
|
|
|
6974 int_wordlist = vim_tempname('s');
|
|
|
6975 if (int_wordlist == NULL)
|
|
381
|
6976 return;
|
|
|
6977 }
|
|
385
|
6978 fname = int_wordlist;
|
|
381
|
6979 }
|
|
310
|
6980 else
|
|
|
6981 {
|
|
381
|
6982 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
6983 if (*curbuf->b_p_spf == NUL)
|
|
|
6984 {
|
|
|
6985 init_spellfile();
|
|
|
6986 new_spf = TRUE;
|
|
|
6987 }
|
|
|
6988
|
|
|
6989 if (*curbuf->b_p_spf == NUL)
|
|
|
6990 {
|
|
|
6991 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
6992 return;
|
|
|
6993 }
|
|
|
6994
|
|
385
|
6995 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
6996 {
|
|
|
6997 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
6998 if (i == index)
|
|
|
6999 break;
|
|
|
7000 if (*spf == NUL)
|
|
|
7001 {
|
|
|
7002 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
7003 return;
|
|
|
7004 }
|
|
|
7005 }
|
|
|
7006
|
|
310
|
7007 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
7008 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
7009 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
7010 buf = NULL;
|
|
|
7011 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
7012 {
|
|
310
|
7013 EMSG(_(e_bufloaded));
|
|
381
|
7014 return;
|
|
|
7015 }
|
|
|
7016
|
|
385
|
7017 fname = fnamebuf;
|
|
|
7018 }
|
|
|
7019
|
|
|
7020 if (bad)
|
|
|
7021 {
|
|
|
7022 /* When the word also appears as good word we need to remove that one,
|
|
|
7023 * since its flags sort before the one with WF_BANNED. */
|
|
|
7024 fd = mch_fopen((char *)fname, "r");
|
|
|
7025 if (fd != NULL)
|
|
|
7026 {
|
|
|
7027 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
7028 {
|
|
|
7029 fpos = fpos_next;
|
|
|
7030 fpos_next = ftell(fd);
|
|
|
7031 if (STRNCMP(word, line, len) == 0
|
|
|
7032 && (line[len] == '/' || line[len] < ' '))
|
|
|
7033 {
|
|
|
7034 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
7035 * the start of the line. Mixing reading and writing
|
|
|
7036 * doesn't work for all systems, close the file first. */
|
|
|
7037 fclose(fd);
|
|
|
7038 fd = mch_fopen((char *)fname, "r+");
|
|
|
7039 if (fd == NULL)
|
|
|
7040 break;
|
|
|
7041 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
7042 fputc('#', fd);
|
|
|
7043 fseek(fd, fpos_next, SEEK_SET);
|
|
|
7044 }
|
|
|
7045 }
|
|
|
7046 fclose(fd);
|
|
|
7047 }
|
|
381
|
7048 }
|
|
|
7049
|
|
|
7050 fd = mch_fopen((char *)fname, "a");
|
|
|
7051 if (fd == NULL && new_spf)
|
|
|
7052 {
|
|
|
7053 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
7054 * We may need to create the "spell" directory first. We already
|
|
|
7055 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
7056 STRCPY(NameBuff, fname);
|
|
|
7057 *gettail_sep(NameBuff) = NUL;
|
|
|
7058 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
7059 {
|
|
|
7060 /* The directory doesn't exist. Try creating it and opening the
|
|
|
7061 * file again. */
|
|
|
7062 vim_mkdir(NameBuff, 0755);
|
|
|
7063 fd = mch_fopen((char *)fname, "a");
|
|
|
7064 }
|
|
|
7065 }
|
|
|
7066
|
|
|
7067 if (fd == NULL)
|
|
|
7068 EMSG2(_(e_notopen), fname);
|
|
|
7069 else
|
|
|
7070 {
|
|
|
7071 if (bad)
|
|
|
7072 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
7073 else
|
|
381
|
7074 fprintf(fd, "%.*s\n", len, word);
|
|
|
7075 fclose(fd);
|
|
|
7076
|
|
|
7077 /* Update the .add.spl file. */
|
|
|
7078 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
7079
|
|
|
7080 /* If the .add file is edited somewhere, reload it. */
|
|
|
7081 if (buf != NULL)
|
|
|
7082 buf_reload(buf);
|
|
|
7083
|
|
|
7084 redraw_all_later(NOT_VALID);
|
|
310
|
7085 }
|
|
|
7086 }
|
|
|
7087
|
|
|
7088 /*
|
|
|
7089 * Initialize 'spellfile' for the current buffer.
|
|
|
7090 */
|
|
|
7091 static void
|
|
|
7092 init_spellfile()
|
|
|
7093 {
|
|
|
7094 char_u buf[MAXPATHL];
|
|
|
7095 int l;
|
|
|
7096 slang_T *sl;
|
|
|
7097 char_u *rtp;
|
|
351
|
7098 char_u *lend;
|
|
310
|
7099
|
|
|
7100 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
7101 {
|
|
351
|
7102 /* Find the end of the language name. Exclude the region. */
|
|
|
7103 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
7104 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
7105 ;
|
|
|
7106
|
|
|
7107 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
7108 * are allowed to write. */
|
|
310
|
7109 rtp = p_rtp;
|
|
|
7110 while (*rtp != NUL)
|
|
|
7111 {
|
|
|
7112 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
7113 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
7114 if (filewritable(buf) == 2)
|
|
|
7115 {
|
|
316
|
7116 /* Use the first language name from 'spelllang' and the
|
|
|
7117 * encoding used in the first loaded .spl file. */
|
|
310
|
7118 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
7119 l = STRLEN(buf);
|
|
|
7120 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
7121 "/spell/%.*s.%s.add",
|
|
351
|
7122 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
7123 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
7124 ? (char_u *)"ascii" : spell_enc());
|
|
|
7125 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
7126 break;
|
|
|
7127 }
|
|
|
7128 }
|
|
|
7129 }
|
|
|
7130 }
|
|
236
|
7131
|
|
300
|
7132
|
|
307
|
7133 /*
|
|
|
7134 * Init the chartab used for spelling for ASCII.
|
|
|
7135 * EBCDIC is not supported!
|
|
|
7136 */
|
|
|
7137 static void
|
|
|
7138 clear_spell_chartab(sp)
|
|
|
7139 spelltab_T *sp;
|
|
|
7140 {
|
|
324
|
7141 int i;
|
|
307
|
7142
|
|
|
7143 /* Init everything to FALSE. */
|
|
|
7144 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
7145 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
7146 for (i = 0; i < 256; ++i)
|
|
324
|
7147 {
|
|
307
|
7148 sp->st_fold[i] = i;
|
|
324
|
7149 sp->st_upper[i] = i;
|
|
|
7150 }
|
|
307
|
7151
|
|
|
7152 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
7153 * that is done separately. */
|
|
|
7154 for (i = '0'; i <= '9'; ++i)
|
|
|
7155 sp->st_isw[i] = TRUE;
|
|
|
7156 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
7157 {
|
|
|
7158 sp->st_isw[i] = TRUE;
|
|
|
7159 sp->st_isu[i] = TRUE;
|
|
|
7160 sp->st_fold[i] = i + 0x20;
|
|
|
7161 }
|
|
|
7162 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
7163 {
|
|
307
|
7164 sp->st_isw[i] = TRUE;
|
|
324
|
7165 sp->st_upper[i] = i - 0x20;
|
|
|
7166 }
|
|
307
|
7167 }
|
|
|
7168
|
|
|
7169 /*
|
|
|
7170 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
7171 * Called once while starting up and when 'encoding' changes.
|
|
|
7172 * The default is to use isalpha(), but the spell file should define the word
|
|
|
7173 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
7174 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
7175 */
|
|
|
7176 void
|
|
|
7177 init_spell_chartab()
|
|
|
7178 {
|
|
|
7179 int i;
|
|
|
7180
|
|
|
7181 did_set_spelltab = FALSE;
|
|
|
7182 clear_spell_chartab(&spelltab);
|
|
|
7183 #ifdef FEAT_MBYTE
|
|
|
7184 if (enc_dbcs)
|
|
|
7185 {
|
|
|
7186 /* DBCS: assume double-wide characters are word characters. */
|
|
|
7187 for (i = 128; i <= 255; ++i)
|
|
|
7188 if (MB_BYTE2LEN(i) == 2)
|
|
|
7189 spelltab.st_isw[i] = TRUE;
|
|
|
7190 }
|
|
324
|
7191 else if (enc_utf8)
|
|
|
7192 {
|
|
|
7193 for (i = 128; i < 256; ++i)
|
|
|
7194 {
|
|
|
7195 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
7196 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
7197 spelltab.st_fold[i] = utf_fold(i);
|
|
|
7198 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
7199 }
|
|
|
7200 }
|
|
307
|
7201 else
|
|
|
7202 #endif
|
|
|
7203 {
|
|
324
|
7204 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
7205 for (i = 128; i < 256; ++i)
|
|
|
7206 {
|
|
|
7207 if (MB_ISUPPER(i))
|
|
|
7208 {
|
|
324
|
7209 spelltab.st_isw[i] = TRUE;
|
|
307
|
7210 spelltab.st_isu[i] = TRUE;
|
|
|
7211 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
7212 }
|
|
324
|
7213 else if (MB_ISLOWER(i))
|
|
|
7214 {
|
|
|
7215 spelltab.st_isw[i] = TRUE;
|
|
|
7216 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
7217 }
|
|
307
|
7218 }
|
|
|
7219 }
|
|
|
7220 }
|
|
|
7221
|
|
|
7222 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
7223 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
7224
|
|
|
7225 /*
|
|
|
7226 * Set the spell character tables from strings in the affix file.
|
|
|
7227 */
|
|
|
7228 static int
|
|
|
7229 set_spell_chartab(fol, low, upp)
|
|
|
7230 char_u *fol;
|
|
|
7231 char_u *low;
|
|
|
7232 char_u *upp;
|
|
|
7233 {
|
|
|
7234 /* We build the new tables here first, so that we can compare with the
|
|
|
7235 * previous one. */
|
|
|
7236 spelltab_T new_st;
|
|
|
7237 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
7238 int f, l, u;
|
|
|
7239
|
|
|
7240 clear_spell_chartab(&new_st);
|
|
|
7241
|
|
|
7242 while (*pf != NUL)
|
|
|
7243 {
|
|
|
7244 if (*pl == NUL || *pu == NUL)
|
|
|
7245 {
|
|
|
7246 EMSG(_(e_affform));
|
|
|
7247 return FAIL;
|
|
|
7248 }
|
|
|
7249 #ifdef FEAT_MBYTE
|
|
|
7250 f = mb_ptr2char_adv(&pf);
|
|
|
7251 l = mb_ptr2char_adv(&pl);
|
|
|
7252 u = mb_ptr2char_adv(&pu);
|
|
|
7253 #else
|
|
|
7254 f = *pf++;
|
|
|
7255 l = *pl++;
|
|
|
7256 u = *pu++;
|
|
|
7257 #endif
|
|
|
7258 /* Every character that appears is a word character. */
|
|
|
7259 if (f < 256)
|
|
|
7260 new_st.st_isw[f] = TRUE;
|
|
|
7261 if (l < 256)
|
|
|
7262 new_st.st_isw[l] = TRUE;
|
|
|
7263 if (u < 256)
|
|
|
7264 new_st.st_isw[u] = TRUE;
|
|
|
7265
|
|
|
7266 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
7267 * case-folding */
|
|
|
7268 if (l < 256 && l != f)
|
|
|
7269 {
|
|
|
7270 if (f >= 256)
|
|
|
7271 {
|
|
|
7272 EMSG(_(e_affrange));
|
|
|
7273 return FAIL;
|
|
|
7274 }
|
|
|
7275 new_st.st_fold[l] = f;
|
|
|
7276 }
|
|
|
7277
|
|
|
7278 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
7279 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
7280 * "FOL" . */
|
|
307
|
7281 if (u < 256 && u != f)
|
|
|
7282 {
|
|
|
7283 if (f >= 256)
|
|
|
7284 {
|
|
|
7285 EMSG(_(e_affrange));
|
|
|
7286 return FAIL;
|
|
|
7287 }
|
|
|
7288 new_st.st_fold[u] = f;
|
|
|
7289 new_st.st_isu[u] = TRUE;
|
|
324
|
7290 new_st.st_upper[f] = u;
|
|
307
|
7291 }
|
|
|
7292 }
|
|
|
7293
|
|
|
7294 if (*pl != NUL || *pu != NUL)
|
|
|
7295 {
|
|
|
7296 EMSG(_(e_affform));
|
|
|
7297 return FAIL;
|
|
|
7298 }
|
|
|
7299
|
|
|
7300 return set_spell_finish(&new_st);
|
|
|
7301 }
|
|
|
7302
|
|
|
7303 /*
|
|
|
7304 * Set the spell character tables from strings in the .spl file.
|
|
|
7305 */
|
|
483
|
7306 static void
|
|
389
|
7307 set_spell_charflags(flags, cnt, fol)
|
|
307
|
7308 char_u *flags;
|
|
389
|
7309 int cnt; /* length of "flags" */
|
|
|
7310 char_u *fol;
|
|
307
|
7311 {
|
|
|
7312 /* We build the new tables here first, so that we can compare with the
|
|
|
7313 * previous one. */
|
|
|
7314 spelltab_T new_st;
|
|
|
7315 int i;
|
|
389
|
7316 char_u *p = fol;
|
|
324
|
7317 int c;
|
|
307
|
7318
|
|
|
7319 clear_spell_chartab(&new_st);
|
|
|
7320
|
|
389
|
7321 for (i = 0; i < 128; ++i)
|
|
|
7322 {
|
|
|
7323 if (i < cnt)
|
|
|
7324 {
|
|
|
7325 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
7326 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
7327 }
|
|
|
7328
|
|
|
7329 if (*p != NUL)
|
|
|
7330 {
|
|
307
|
7331 #ifdef FEAT_MBYTE
|
|
389
|
7332 c = mb_ptr2char_adv(&p);
|
|
307
|
7333 #else
|
|
389
|
7334 c = *p++;
|
|
307
|
7335 #endif
|
|
389
|
7336 new_st.st_fold[i + 128] = c;
|
|
|
7337 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
7338 new_st.st_upper[c] = i + 128;
|
|
|
7339 }
|
|
307
|
7340 }
|
|
|
7341
|
|
483
|
7342 (void)set_spell_finish(&new_st);
|
|
307
|
7343 }
|
|
|
7344
|
|
|
7345 static int
|
|
|
7346 set_spell_finish(new_st)
|
|
|
7347 spelltab_T *new_st;
|
|
|
7348 {
|
|
|
7349 int i;
|
|
|
7350
|
|
|
7351 if (did_set_spelltab)
|
|
|
7352 {
|
|
|
7353 /* check that it's the same table */
|
|
|
7354 for (i = 0; i < 256; ++i)
|
|
|
7355 {
|
|
|
7356 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
7357 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
7358 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
7359 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
7360 {
|
|
|
7361 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
7362 return FAIL;
|
|
|
7363 }
|
|
|
7364 }
|
|
|
7365 }
|
|
|
7366 else
|
|
|
7367 {
|
|
|
7368 /* copy the new spelltab into the one being used */
|
|
|
7369 spelltab = *new_st;
|
|
|
7370 did_set_spelltab = TRUE;
|
|
|
7371 }
|
|
|
7372
|
|
|
7373 return OK;
|
|
|
7374 }
|
|
|
7375
|
|
|
7376 /*
|
|
358
|
7377 * Return TRUE if "p" points to a word character.
|
|
366
|
7378 * As a special case we see "midword" characters as word character when it is
|
|
358
|
7379 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
7380 * Thus this only works properly when past the first character of the word.
|
|
358
|
7381 */
|
|
|
7382 static int
|
|
376
|
7383 spell_iswordp(p, buf)
|
|
358
|
7384 char_u *p;
|
|
376
|
7385 buf_T *buf; /* buffer used */
|
|
358
|
7386 {
|
|
366
|
7387 #ifdef FEAT_MBYTE
|
|
358
|
7388 char_u *s;
|
|
366
|
7389 int l;
|
|
|
7390 int c;
|
|
|
7391
|
|
|
7392 if (has_mbyte)
|
|
|
7393 {
|
|
|
7394 l = MB_BYTE2LEN(*p);
|
|
358
|
7395 s = p;
|
|
366
|
7396 if (l == 1)
|
|
|
7397 {
|
|
|
7398 /* be quick for ASCII */
|
|
376
|
7399 if (buf->b_spell_ismw[*p])
|
|
366
|
7400 {
|
|
|
7401 s = p + 1; /* skip a mid-word character */
|
|
|
7402 l = MB_BYTE2LEN(*s);
|
|
|
7403 }
|
|
|
7404 }
|
|
|
7405 else
|
|
|
7406 {
|
|
|
7407 c = mb_ptr2char(p);
|
|
376
|
7408 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
7409 : (buf->b_spell_ismw_mb != NULL
|
|
|
7410 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
7411 {
|
|
|
7412 s = p + l;
|
|
|
7413 l = MB_BYTE2LEN(*s);
|
|
|
7414 }
|
|
|
7415 }
|
|
|
7416
|
|
390
|
7417 c = mb_ptr2char(s);
|
|
|
7418 if (c > 255)
|
|
366
|
7419 return mb_get_class(s) >= 2;
|
|
390
|
7420 return spelltab.st_isw[c];
|
|
366
|
7421 }
|
|
358
|
7422 #endif
|
|
366
|
7423
|
|
376
|
7424 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
7425 }
|
|
|
7426
|
|
|
7427 /*
|
|
|
7428 * Return TRUE if "p" points to a word character.
|
|
|
7429 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
7430 */
|
|
|
7431 static int
|
|
|
7432 spell_iswordp_nmw(p)
|
|
|
7433 char_u *p;
|
|
|
7434 {
|
|
|
7435 #ifdef FEAT_MBYTE
|
|
390
|
7436 int c;
|
|
|
7437
|
|
|
7438 if (has_mbyte)
|
|
|
7439 {
|
|
|
7440 c = mb_ptr2char(p);
|
|
|
7441 if (c > 255)
|
|
|
7442 return mb_get_class(p) >= 2;
|
|
|
7443 return spelltab.st_isw[c];
|
|
|
7444 }
|
|
376
|
7445 #endif
|
|
|
7446 return spelltab.st_isw[*p];
|
|
358
|
7447 }
|
|
|
7448
|
|
372
|
7449 #ifdef FEAT_MBYTE
|
|
|
7450 /*
|
|
|
7451 * Return TRUE if "p" points to a word character.
|
|
|
7452 * Wide version of spell_iswordp().
|
|
|
7453 */
|
|
|
7454 static int
|
|
376
|
7455 spell_iswordp_w(p, buf)
|
|
372
|
7456 int *p;
|
|
376
|
7457 buf_T *buf;
|
|
372
|
7458 {
|
|
|
7459 int *s;
|
|
|
7460
|
|
376
|
7461 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
7462 : (buf->b_spell_ismw_mb != NULL
|
|
|
7463 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
7464 s = p + 1;
|
|
|
7465 else
|
|
|
7466 s = p;
|
|
|
7467
|
|
390
|
7468 if (*s > 255)
|
|
372
|
7469 {
|
|
|
7470 if (enc_utf8)
|
|
|
7471 return utf_class(*s) >= 2;
|
|
|
7472 if (enc_dbcs)
|
|
|
7473 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
7474 return 0;
|
|
|
7475 }
|
|
|
7476 return spelltab.st_isw[*s];
|
|
|
7477 }
|
|
|
7478 #endif
|
|
|
7479
|
|
358
|
7480 /*
|
|
339
|
7481 * Write the table with prefix conditions to the .spl file.
|
|
483
|
7482 * When "fd" is NULL only count the length of what is written.
|
|
|
7483 */
|
|
|
7484 static int
|
|
339
|
7485 write_spell_prefcond(fd, gap)
|
|
|
7486 FILE *fd;
|
|
|
7487 garray_T *gap;
|
|
|
7488 {
|
|
|
7489 int i;
|
|
|
7490 char_u *p;
|
|
|
7491 int len;
|
|
483
|
7492 int totlen;
|
|
|
7493
|
|
|
7494 if (fd != NULL)
|
|
|
7495 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
7496
|
|
|
7497 totlen = 2 + gap->ga_len; /* length of <prefcondcnt> and <condlen> bytes */
|
|
339
|
7498
|
|
|
7499 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7500 {
|
|
|
7501 /* <prefcond> : <condlen> <condstr> */
|
|
|
7502 p = ((char_u **)gap->ga_data)[i];
|
|
483
|
7503 if (p != NULL)
|
|
339
|
7504 {
|
|
|
7505 len = STRLEN(p);
|
|
483
|
7506 if (fd != NULL)
|
|
|
7507 {
|
|
|
7508 fputc(len, fd);
|
|
|
7509 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
7510 }
|
|
|
7511 totlen += len;
|
|
|
7512 }
|
|
|
7513 else if (fd != NULL)
|
|
|
7514 fputc(0, fd);
|
|
|
7515 }
|
|
|
7516
|
|
|
7517 return totlen;
|
|
307
|
7518 }
|
|
|
7519
|
|
|
7520 /*
|
|
324
|
7521 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
7522 * Uses the character definitions from the .spl file.
|
|
307
|
7523 * When using a multi-byte 'encoding' the length may change!
|
|
|
7524 * Returns FAIL when something wrong.
|
|
|
7525 */
|
|
|
7526 static int
|
|
324
|
7527 spell_casefold(str, len, buf, buflen)
|
|
|
7528 char_u *str;
|
|
307
|
7529 int len;
|
|
|
7530 char_u *buf;
|
|
|
7531 int buflen;
|
|
|
7532 {
|
|
|
7533 int i;
|
|
|
7534
|
|
|
7535 if (len >= buflen)
|
|
|
7536 {
|
|
|
7537 buf[0] = NUL;
|
|
|
7538 return FAIL; /* result will not fit */
|
|
|
7539 }
|
|
|
7540
|
|
|
7541 #ifdef FEAT_MBYTE
|
|
|
7542 if (has_mbyte)
|
|
|
7543 {
|
|
324
|
7544 int outi = 0;
|
|
|
7545 char_u *p;
|
|
307
|
7546 int c;
|
|
|
7547
|
|
|
7548 /* Fold one character at a time. */
|
|
324
|
7549 for (p = str; p < str + len; )
|
|
307
|
7550 {
|
|
|
7551 if (outi + MB_MAXBYTES > buflen)
|
|
|
7552 {
|
|
|
7553 buf[outi] = NUL;
|
|
|
7554 return FAIL;
|
|
|
7555 }
|
|
474
|
7556 c = mb_cptr2char_adv(&p);
|
|
324
|
7557 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
7558 }
|
|
|
7559 buf[outi] = NUL;
|
|
|
7560 }
|
|
|
7561 else
|
|
|
7562 #endif
|
|
|
7563 {
|
|
|
7564 /* Be quick for non-multibyte encodings. */
|
|
|
7565 for (i = 0; i < len; ++i)
|
|
324
|
7566 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
7567 buf[i] = NUL;
|
|
|
7568 }
|
|
|
7569
|
|
|
7570 return OK;
|
|
|
7571 }
|
|
|
7572
|
|
372
|
7573 #define SPS_BEST 1
|
|
|
7574 #define SPS_FAST 2
|
|
|
7575 #define SPS_DOUBLE 4
|
|
|
7576
|
|
|
7577 static int sps_flags = SPS_BEST;
|
|
483
|
7578 static int sps_limit = 9999;
|
|
372
|
7579
|
|
|
7580 /*
|
|
|
7581 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
483
|
7582 * Sets "sps_flags" and "sps_limit".
|
|
372
|
7583 */
|
|
|
7584 int
|
|
|
7585 spell_check_sps()
|
|
|
7586 {
|
|
|
7587 char_u *p;
|
|
483
|
7588 char_u *s;
|
|
372
|
7589 char_u buf[MAXPATHL];
|
|
|
7590 int f;
|
|
|
7591
|
|
|
7592 sps_flags = 0;
|
|
483
|
7593 sps_limit = 9999;
|
|
372
|
7594
|
|
|
7595 for (p = p_sps; *p != NUL; )
|
|
|
7596 {
|
|
|
7597 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
7598
|
|
|
7599 f = 0;
|
|
483
|
7600 if (VIM_ISDIGIT(*buf))
|
|
|
7601 {
|
|
|
7602 s = buf;
|
|
|
7603 sps_limit = getdigits(&s);
|
|
|
7604 if (*s != NUL && !VIM_ISDIGIT(*s))
|
|
|
7605 f = -1;
|
|
|
7606 }
|
|
|
7607 else if (STRCMP(buf, "best") == 0)
|
|
372
|
7608 f = SPS_BEST;
|
|
|
7609 else if (STRCMP(buf, "fast") == 0)
|
|
|
7610 f = SPS_FAST;
|
|
|
7611 else if (STRCMP(buf, "double") == 0)
|
|
|
7612 f = SPS_DOUBLE;
|
|
|
7613 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
7614 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
7615 f = -1;
|
|
|
7616
|
|
|
7617 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
7618 {
|
|
|
7619 sps_flags = SPS_BEST;
|
|
483
|
7620 sps_limit = 9999;
|
|
372
|
7621 return FAIL;
|
|
|
7622 }
|
|
|
7623 if (f != 0)
|
|
|
7624 sps_flags = f;
|
|
|
7625 }
|
|
|
7626
|
|
|
7627 if (sps_flags == 0)
|
|
|
7628 sps_flags = SPS_BEST;
|
|
|
7629
|
|
|
7630 return OK;
|
|
|
7631 }
|
|
|
7632
|
|
|
7633 /* Remember what "z?" replaced. */
|
|
|
7634 static char_u *repl_from = NULL;
|
|
|
7635 static char_u *repl_to = NULL;
|
|
|
7636
|
|
323
|
7637 /*
|
|
|
7638 * "z?": Find badly spelled word under or after the cursor.
|
|
|
7639 * Give suggestions for the properly spelled word.
|
|
485
|
7640 * When "count" is non-zero use that suggestion.
|
|
323
|
7641 */
|
|
|
7642 void
|
|
485
|
7643 spell_suggest(count)
|
|
|
7644 int count;
|
|
323
|
7645 {
|
|
|
7646 char_u *line;
|
|
|
7647 pos_T prev_cursor = curwin->w_cursor;
|
|
|
7648 char_u wcopy[MAXWLEN + 2];
|
|
|
7649 char_u *p;
|
|
|
7650 int i;
|
|
|
7651 int c;
|
|
|
7652 suginfo_T sug;
|
|
|
7653 suggest_T *stp;
|
|
372
|
7654 int mouse_used;
|
|
388
|
7655 int need_cap;
|
|
483
|
7656 int limit;
|
|
485
|
7657 int selected = count;
|
|
323
|
7658
|
|
344
|
7659 /* Find the start of the badly spelled word. */
|
|
346
|
7660 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
7661 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
7662 {
|
|
|
7663 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
7664 return;
|
|
|
7665
|
|
|
7666 /* No bad word or it starts after the cursor: use the word under the
|
|
|
7667 * cursor. */
|
|
|
7668 curwin->w_cursor = prev_cursor;
|
|
|
7669 line = ml_get_curline();
|
|
|
7670 p = line + curwin->w_cursor.col;
|
|
|
7671 /* Backup to before start of word. */
|
|
390
|
7672 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
7673 mb_ptr_back(line, p);
|
|
|
7674 /* Forward to start of word. */
|
|
390
|
7675 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
7676 mb_ptr_adv(p);
|
|
|
7677
|
|
390
|
7678 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
7679 {
|
|
|
7680 beep_flush();
|
|
|
7681 return;
|
|
|
7682 }
|
|
|
7683 curwin->w_cursor.col = p - line;
|
|
323
|
7684 }
|
|
|
7685
|
|
344
|
7686 /* Get the word and its length. */
|
|
|
7687
|
|
388
|
7688 /* Figure out if the word should be capitalised. */
|
|
475
|
7689 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
7690
|
|
|
7691 line = ml_get_curline();
|
|
388
|
7692
|
|
483
|
7693 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in
|
|
|
7694 * 'spellsuggest', whatever is smaller. */
|
|
|
7695 if (sps_limit > (int)Rows - 2)
|
|
|
7696 limit = (int)Rows - 2;
|
|
|
7697 else
|
|
|
7698 limit = sps_limit;
|
|
|
7699 spell_find_suggest(line + curwin->w_cursor.col, &sug, limit,
|
|
388
|
7700 TRUE, need_cap);
|
|
323
|
7701
|
|
|
7702 if (sug.su_ga.ga_len == 0)
|
|
|
7703 MSG(_("Sorry, no suggestions"));
|
|
485
|
7704 else if (count > 0)
|
|
|
7705 {
|
|
|
7706 if (count > sug.su_ga.ga_len)
|
|
|
7707 smsg((char_u *)_("Sorry, only %ld suggestions"),
|
|
|
7708 (long)sug.su_ga.ga_len);
|
|
|
7709 }
|
|
323
|
7710 else
|
|
|
7711 {
|
|
372
|
7712 vim_free(repl_from);
|
|
|
7713 repl_from = NULL;
|
|
|
7714 vim_free(repl_to);
|
|
|
7715 repl_to = NULL;
|
|
|
7716
|
|
474
|
7717 #ifdef FEAT_RIGHTLEFT
|
|
|
7718 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
7719 cmdmsg_rl = curwin->w_p_rl;
|
|
|
7720 if (cmdmsg_rl)
|
|
|
7721 msg_col = Columns - 1;
|
|
|
7722 #endif
|
|
|
7723
|
|
323
|
7724 /* List the suggestions. */
|
|
|
7725 msg_start();
|
|
372
|
7726 lines_left = Rows; /* avoid more prompt */
|
|
323
|
7727 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
7728 sug.su_badlen, sug.su_badptr);
|
|
474
|
7729 #ifdef FEAT_RIGHTLEFT
|
|
|
7730 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
7731 {
|
|
|
7732 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
7733 * untranslated message rightleft. */
|
|
|
7734 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
7735 sug.su_badlen, sug.su_badptr);
|
|
|
7736 }
|
|
|
7737 #endif
|
|
323
|
7738 msg_puts(IObuff);
|
|
|
7739 msg_clr_eos();
|
|
|
7740 msg_putchar('\n');
|
|
346
|
7741
|
|
323
|
7742 msg_scroll = TRUE;
|
|
|
7743 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
7744 {
|
|
344
|
7745 stp = &SUG(sug.su_ga, i);
|
|
323
|
7746
|
|
|
7747 /* The suggested word may replace only part of the bad word, add
|
|
|
7748 * the not replaced part. */
|
|
|
7749 STRCPY(wcopy, stp->st_word);
|
|
|
7750 if (sug.su_badlen > stp->st_orglen)
|
|
|
7751 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
7752 sug.su_badptr + stp->st_orglen,
|
|
|
7753 sug.su_badlen - stp->st_orglen);
|
|
474
|
7754 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
7755 #ifdef FEAT_RIGHTLEFT
|
|
|
7756 if (cmdmsg_rl)
|
|
|
7757 rl_mirror(IObuff);
|
|
|
7758 #endif
|
|
|
7759 msg_puts(IObuff);
|
|
|
7760
|
|
|
7761 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
7762 msg_puts(IObuff);
|
|
|
7763
|
|
|
7764 /* The word may replace more than "su_badlen". */
|
|
|
7765 if (sug.su_badlen < stp->st_orglen)
|
|
|
7766 {
|
|
|
7767 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
7768 stp->st_orglen, sug.su_badptr);
|
|
|
7769 msg_puts(IObuff);
|
|
|
7770 }
|
|
|
7771
|
|
324
|
7772 if (p_verbose > 0)
|
|
344
|
7773 {
|
|
346
|
7774 /* Add the score. */
|
|
351
|
7775 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
7776 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
7777 stp->st_salscore ? "s " : "",
|
|
|
7778 stp->st_score, stp->st_altscore);
|
|
|
7779 else
|
|
474
|
7780 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
7781 stp->st_score);
|
|
474
|
7782 #ifdef FEAT_RIGHTLEFT
|
|
|
7783 if (cmdmsg_rl)
|
|
|
7784 /* Mirror the numbers, but keep the leading space. */
|
|
|
7785 rl_mirror(IObuff + 1);
|
|
|
7786 #endif
|
|
346
|
7787 msg_advance(30);
|
|
|
7788 msg_puts(IObuff);
|
|
344
|
7789 }
|
|
323
|
7790 msg_putchar('\n');
|
|
|
7791 }
|
|
|
7792
|
|
474
|
7793 #ifdef FEAT_RIGHTLEFT
|
|
|
7794 cmdmsg_rl = FALSE;
|
|
|
7795 msg_col = 0;
|
|
|
7796 #endif
|
|
323
|
7797 /* Ask for choice. */
|
|
485
|
7798 selected = prompt_for_number(&mouse_used);
|
|
372
|
7799 if (mouse_used)
|
|
485
|
7800 selected -= lines_left;
|
|
|
7801 }
|
|
|
7802
|
|
|
7803 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
|
7804 {
|
|
|
7805 /* Save the from and to text for :spellrepall. */
|
|
|
7806 stp = &SUG(sug.su_ga, selected - 1);
|
|
|
7807 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
7808 repl_to = vim_strsave(stp->st_word);
|
|
|
7809
|
|
|
7810 /* Replace the word. */
|
|
|
7811 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
7812 if (p != NULL)
|
|
|
7813 {
|
|
|
7814 c = sug.su_badptr - line;
|
|
|
7815 mch_memmove(p, line, c);
|
|
|
7816 STRCPY(p + c, stp->st_word);
|
|
|
7817 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
7818 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
7819 curwin->w_cursor.col = c;
|
|
|
7820 changed_bytes(curwin->w_cursor.lnum, c);
|
|
|
7821
|
|
|
7822 /* For redo we use a change-word command. */
|
|
|
7823 ResetRedobuff();
|
|
|
7824 AppendToRedobuff((char_u *)"ciw");
|
|
|
7825 AppendToRedobuff(stp->st_word);
|
|
|
7826 AppendCharToRedobuff(ESC);
|
|
|
7827 }
|
|
|
7828 }
|
|
|
7829 else
|
|
|
7830 curwin->w_cursor = prev_cursor;
|
|
323
|
7831
|
|
344
|
7832 spell_find_cleanup(&sug);
|
|
|
7833 }
|
|
|
7834
|
|
|
7835 /*
|
|
475
|
7836 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
7837 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
7838 */
|
|
|
7839 static int
|
|
|
7840 check_need_cap(lnum, col)
|
|
|
7841 linenr_T lnum;
|
|
|
7842 colnr_T col;
|
|
|
7843 {
|
|
|
7844 int need_cap = FALSE;
|
|
|
7845 char_u *line;
|
|
|
7846 char_u *line_copy = NULL;
|
|
|
7847 char_u *p;
|
|
|
7848 colnr_T endcol;
|
|
|
7849 regmatch_T regmatch;
|
|
|
7850
|
|
|
7851 if (curbuf->b_cap_prog == NULL)
|
|
|
7852 return FALSE;
|
|
|
7853
|
|
|
7854 line = ml_get_curline();
|
|
|
7855 endcol = 0;
|
|
|
7856 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
7857 {
|
|
|
7858 /* At start of line, check if previous line is empty or sentence
|
|
|
7859 * ends there. */
|
|
|
7860 if (lnum == 1)
|
|
|
7861 need_cap = TRUE;
|
|
|
7862 else
|
|
|
7863 {
|
|
|
7864 line = ml_get(lnum - 1);
|
|
|
7865 if (*skipwhite(line) == NUL)
|
|
|
7866 need_cap = TRUE;
|
|
|
7867 else
|
|
|
7868 {
|
|
|
7869 /* Append a space in place of the line break. */
|
|
|
7870 line_copy = concat_str(line, (char_u *)" ");
|
|
|
7871 line = line_copy;
|
|
|
7872 endcol = STRLEN(line);
|
|
|
7873 }
|
|
|
7874 }
|
|
|
7875 }
|
|
|
7876 else
|
|
|
7877 endcol = col;
|
|
|
7878
|
|
|
7879 if (endcol > 0)
|
|
|
7880 {
|
|
|
7881 /* Check if sentence ends before the bad word. */
|
|
|
7882 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
7883 regmatch.rm_ic = FALSE;
|
|
|
7884 p = line + endcol;
|
|
|
7885 for (;;)
|
|
|
7886 {
|
|
|
7887 mb_ptr_back(line, p);
|
|
|
7888 if (p == line || spell_iswordp_nmw(p))
|
|
|
7889 break;
|
|
|
7890 if (vim_regexec(®match, p, 0)
|
|
|
7891 && regmatch.endp[0] == line + endcol)
|
|
|
7892 {
|
|
|
7893 need_cap = TRUE;
|
|
|
7894 break;
|
|
|
7895 }
|
|
|
7896 }
|
|
|
7897 }
|
|
|
7898
|
|
|
7899 vim_free(line_copy);
|
|
|
7900
|
|
|
7901 return need_cap;
|
|
|
7902 }
|
|
|
7903
|
|
|
7904
|
|
|
7905 /*
|
|
372
|
7906 * ":spellrepall"
|
|
|
7907 */
|
|
|
7908 /*ARGSUSED*/
|
|
|
7909 void
|
|
|
7910 ex_spellrepall(eap)
|
|
|
7911 exarg_T *eap;
|
|
|
7912 {
|
|
|
7913 pos_T pos = curwin->w_cursor;
|
|
|
7914 char_u *frompat;
|
|
|
7915 int addlen;
|
|
|
7916 char_u *line;
|
|
|
7917 char_u *p;
|
|
|
7918 int save_ws = p_ws;
|
|
483
|
7919 linenr_T prev_lnum = 0;
|
|
372
|
7920
|
|
|
7921 if (repl_from == NULL || repl_to == NULL)
|
|
|
7922 {
|
|
|
7923 EMSG(_("E752: No previous spell replacement"));
|
|
|
7924 return;
|
|
|
7925 }
|
|
|
7926 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
7927
|
|
|
7928 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
7929 if (frompat == NULL)
|
|
|
7930 return;
|
|
|
7931 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
7932 p_ws = FALSE;
|
|
|
7933
|
|
483
|
7934 sub_nsubs = 0;
|
|
|
7935 sub_nlines = 0;
|
|
372
|
7936 curwin->w_cursor.lnum = 0;
|
|
|
7937 while (!got_int)
|
|
|
7938 {
|
|
|
7939 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
7940 || u_save_cursor() == FAIL)
|
|
|
7941 break;
|
|
|
7942
|
|
|
7943 /* Only replace when the right word isn't there yet. This happens
|
|
|
7944 * when changing "etc" to "etc.". */
|
|
|
7945 line = ml_get_curline();
|
|
|
7946 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
7947 repl_to, STRLEN(repl_to)) != 0)
|
|
|
7948 {
|
|
|
7949 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
7950 if (p == NULL)
|
|
|
7951 break;
|
|
|
7952 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
7953 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
7954 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
7955 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
7956 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
483
|
7957
|
|
|
7958 if (curwin->w_cursor.lnum != prev_lnum)
|
|
|
7959 {
|
|
|
7960 ++sub_nlines;
|
|
|
7961 prev_lnum = curwin->w_cursor.lnum;
|
|
|
7962 }
|
|
|
7963 ++sub_nsubs;
|
|
372
|
7964 }
|
|
|
7965 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
7966 }
|
|
|
7967
|
|
|
7968 p_ws = save_ws;
|
|
|
7969 curwin->w_cursor = pos;
|
|
|
7970 vim_free(frompat);
|
|
|
7971
|
|
483
|
7972 if (sub_nsubs == 0)
|
|
372
|
7973 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
483
|
7974 else
|
|
|
7975 do_sub_msg(FALSE);
|
|
372
|
7976 }
|
|
|
7977
|
|
|
7978 /*
|
|
344
|
7979 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
7980 * a list of allocated strings.
|
|
|
7981 */
|
|
|
7982 void
|
|
475
|
7983 spell_suggest_list(gap, word, maxcount, need_cap)
|
|
344
|
7984 garray_T *gap;
|
|
|
7985 char_u *word;
|
|
|
7986 int maxcount; /* maximum nr of suggestions */
|
|
475
|
7987 int need_cap; /* 'spellcapcheck' matched */
|
|
344
|
7988 {
|
|
|
7989 suginfo_T sug;
|
|
|
7990 int i;
|
|
|
7991 suggest_T *stp;
|
|
|
7992 char_u *wcopy;
|
|
|
7993
|
|
475
|
7994 spell_find_suggest(word, &sug, maxcount, FALSE, need_cap);
|
|
344
|
7995
|
|
|
7996 /* Make room in "gap". */
|
|
|
7997 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
7998 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
7999 return;
|
|
|
8000
|
|
323
|
8001 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
8002 {
|
|
|
8003 stp = &SUG(sug.su_ga, i);
|
|
|
8004
|
|
|
8005 /* The suggested word may replace only part of "word", add the not
|
|
|
8006 * replaced part. */
|
|
|
8007 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
8008 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
8009 if (wcopy == NULL)
|
|
|
8010 break;
|
|
|
8011 STRCPY(wcopy, stp->st_word);
|
|
|
8012 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
8013 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
8014 }
|
|
|
8015
|
|
|
8016 spell_find_cleanup(&sug);
|
|
|
8017 }
|
|
|
8018
|
|
|
8019 /*
|
|
|
8020 * Find spell suggestions for the word at the start of "badptr".
|
|
|
8021 * Return the suggestions in "su->su_ga".
|
|
|
8022 * The maximum number of suggestions is "maxcount".
|
|
|
8023 * Note: does use info for the current window.
|
|
|
8024 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
8025 */
|
|
|
8026 static void
|
|
388
|
8027 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
8028 char_u *badptr;
|
|
|
8029 suginfo_T *su;
|
|
|
8030 int maxcount;
|
|
358
|
8031 int banbadword; /* don't include badword in suggestions */
|
|
388
|
8032 int need_cap; /* word should start with capital */
|
|
344
|
8033 {
|
|
385
|
8034 int attr = 0;
|
|
372
|
8035 char_u buf[MAXPATHL];
|
|
|
8036 char_u *p;
|
|
|
8037 int do_combine = FALSE;
|
|
|
8038 char_u *sps_copy;
|
|
|
8039 #ifdef FEAT_EVAL
|
|
|
8040 static int expr_busy = FALSE;
|
|
|
8041 #endif
|
|
385
|
8042 int c;
|
|
344
|
8043
|
|
|
8044 /*
|
|
|
8045 * Set the info in "*su".
|
|
|
8046 */
|
|
|
8047 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
8048 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
8049 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
8050 if (*badptr == NUL)
|
|
|
8051 return;
|
|
344
|
8052 hash_init(&su->su_banned);
|
|
|
8053
|
|
|
8054 su->su_badptr = badptr;
|
|
385
|
8055 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
8056 su->su_maxcount = maxcount;
|
|
372
|
8057 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
8058
|
|
|
8059 if (su->su_badlen >= MAXWLEN)
|
|
|
8060 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
8061 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
8062 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
8063 su->su_fbadword, MAXWLEN);
|
|
346
|
8064 /* get caps flags for bad word */
|
|
474
|
8065 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
8066 su->su_badptr + su->su_badlen);
|
|
388
|
8067 if (need_cap)
|
|
|
8068 su->su_badflags |= WF_ONECAP;
|
|
344
|
8069
|
|
385
|
8070 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
8071 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
8072 * for that. */
|
|
455
|
8073 c = PTR2CHAR(su->su_badptr);
|
|
385
|
8074 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
8075 {
|
|
|
8076 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
8077 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
8078 0, TRUE);
|
|
|
8079 }
|
|
|
8080
|
|
344
|
8081 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
8082 if (banbadword)
|
|
|
8083 add_banned(su, su->su_badword);
|
|
344
|
8084
|
|
372
|
8085 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
8086 sps_copy = vim_strsave(p_sps);
|
|
|
8087 if (sps_copy == NULL)
|
|
|
8088 return;
|
|
|
8089
|
|
|
8090 /* Loop over the items in 'spellsuggest'. */
|
|
|
8091 for (p = sps_copy; *p != NUL; )
|
|
|
8092 {
|
|
|
8093 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
8094
|
|
|
8095 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
8096 {
|
|
|
8097 #ifdef FEAT_EVAL
|
|
375
|
8098 /* Evaluate an expression. Skip this when called recursively,
|
|
|
8099 * when using spellsuggest() in the expression. */
|
|
372
|
8100 if (!expr_busy)
|
|
|
8101 {
|
|
|
8102 expr_busy = TRUE;
|
|
|
8103 spell_suggest_expr(su, buf + 5);
|
|
|
8104 expr_busy = FALSE;
|
|
|
8105 }
|
|
|
8106 #endif
|
|
|
8107 }
|
|
|
8108 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
8109 /* Use list of suggestions in a file. */
|
|
|
8110 spell_suggest_file(su, buf + 5);
|
|
|
8111 else
|
|
|
8112 {
|
|
|
8113 /* Use internal method. */
|
|
|
8114 spell_suggest_intern(su);
|
|
|
8115 if (sps_flags & SPS_DOUBLE)
|
|
|
8116 do_combine = TRUE;
|
|
|
8117 }
|
|
|
8118 }
|
|
|
8119
|
|
|
8120 vim_free(sps_copy);
|
|
|
8121
|
|
|
8122 if (do_combine)
|
|
|
8123 /* Combine the two list of suggestions. This must be done last,
|
|
|
8124 * because sorting changes the order again. */
|
|
|
8125 score_combine(su);
|
|
|
8126 }
|
|
|
8127
|
|
|
8128 #ifdef FEAT_EVAL
|
|
|
8129 /*
|
|
|
8130 * Find suggestions by evaluating expression "expr".
|
|
|
8131 */
|
|
|
8132 static void
|
|
|
8133 spell_suggest_expr(su, expr)
|
|
|
8134 suginfo_T *su;
|
|
|
8135 char_u *expr;
|
|
|
8136 {
|
|
|
8137 list_T *list;
|
|
|
8138 listitem_T *li;
|
|
|
8139 int score;
|
|
|
8140 char_u *p;
|
|
|
8141
|
|
|
8142 /* The work is split up in a few parts to avoid having to export
|
|
|
8143 * suginfo_T.
|
|
|
8144 * First evaluate the expression and get the resulting list. */
|
|
|
8145 list = eval_spell_expr(su->su_badword, expr);
|
|
|
8146 if (list != NULL)
|
|
|
8147 {
|
|
|
8148 /* Loop over the items in the list. */
|
|
|
8149 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
8150 if (li->li_tv.v_type == VAR_LIST)
|
|
|
8151 {
|
|
|
8152 /* Get the word and the score from the items. */
|
|
|
8153 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
8154 if (score >= 0)
|
|
|
8155 add_suggestion(su, &su->su_ga, p,
|
|
|
8156 su->su_badlen, score, 0, TRUE);
|
|
|
8157 }
|
|
|
8158 list_unref(list);
|
|
|
8159 }
|
|
|
8160
|
|
|
8161 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8162 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8163 }
|
|
|
8164 #endif
|
|
|
8165
|
|
|
8166 /*
|
|
474
|
8167 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
8168 */
|
|
|
8169 static void
|
|
|
8170 spell_suggest_file(su, fname)
|
|
|
8171 suginfo_T *su;
|
|
|
8172 char_u *fname;
|
|
|
8173 {
|
|
|
8174 FILE *fd;
|
|
|
8175 char_u line[MAXWLEN * 2];
|
|
|
8176 char_u *p;
|
|
|
8177 int len;
|
|
|
8178 char_u cword[MAXWLEN];
|
|
|
8179
|
|
|
8180 /* Open the file. */
|
|
|
8181 fd = mch_fopen((char *)fname, "r");
|
|
|
8182 if (fd == NULL)
|
|
|
8183 {
|
|
|
8184 EMSG2(_(e_notopen), fname);
|
|
|
8185 return;
|
|
|
8186 }
|
|
|
8187
|
|
|
8188 /* Read it line by line. */
|
|
|
8189 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
8190 {
|
|
|
8191 line_breakcheck();
|
|
|
8192
|
|
|
8193 p = vim_strchr(line, '/');
|
|
|
8194 if (p == NULL)
|
|
|
8195 continue; /* No Tab found, just skip the line. */
|
|
|
8196 *p++ = NUL;
|
|
|
8197 if (STRICMP(su->su_badword, line) == 0)
|
|
|
8198 {
|
|
|
8199 /* Match! Isolate the good word, until CR or NL. */
|
|
|
8200 for (len = 0; p[len] >= ' '; ++len)
|
|
|
8201 ;
|
|
|
8202 p[len] = NUL;
|
|
|
8203
|
|
|
8204 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
8205 * of the bad word. */
|
|
|
8206 if (captype(p, NULL) == 0)
|
|
|
8207 {
|
|
|
8208 make_case_word(p, cword, su->su_badflags);
|
|
|
8209 p = cword;
|
|
|
8210 }
|
|
|
8211
|
|
|
8212 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
8213 SCORE_FILE, 0, TRUE);
|
|
|
8214 }
|
|
|
8215 }
|
|
|
8216
|
|
|
8217 fclose(fd);
|
|
|
8218
|
|
|
8219 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8220 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8221 }
|
|
|
8222
|
|
|
8223 /*
|
|
|
8224 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
8225 */
|
|
|
8226 static void
|
|
|
8227 spell_suggest_intern(su)
|
|
|
8228 suginfo_T *su;
|
|
|
8229 {
|
|
344
|
8230 /*
|
|
346
|
8231 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
8232 *
|
|
|
8233 * Set a maximum score to limit the combination of operations that is
|
|
|
8234 * tried.
|
|
|
8235 */
|
|
346
|
8236 suggest_try_special(su);
|
|
|
8237
|
|
|
8238 /*
|
|
|
8239 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
8240 * from the .aff file and inserting a space (split the word).
|
|
|
8241 */
|
|
|
8242 suggest_try_change(su);
|
|
344
|
8243
|
|
|
8244 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
8245 if (sps_flags & SPS_DOUBLE)
|
|
|
8246 score_comp_sal(su);
|
|
|
8247
|
|
|
8248 /*
|
|
346
|
8249 * 3. Try finding sound-a-like words.
|
|
344
|
8250 *
|
|
|
8251 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
8252 * very slow and often doesn't find new suggestions.
|
|
|
8253 */
|
|
|
8254 if ((sps_flags & SPS_DOUBLE)
|
|
|
8255 || (!(sps_flags & SPS_FAST)
|
|
|
8256 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
8257 {
|
|
|
8258 /* Allow a higher score now. */
|
|
|
8259 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
8260 suggest_try_soundalike(su);
|
|
344
|
8261 }
|
|
|
8262
|
|
|
8263 /* When CTRL-C was hit while searching do show the results. */
|
|
|
8264 ui_breakcheck();
|
|
|
8265 if (got_int)
|
|
|
8266 {
|
|
|
8267 (void)vgetc();
|
|
|
8268 got_int = FALSE;
|
|
|
8269 }
|
|
|
8270
|
|
372
|
8271 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
8272 {
|
|
|
8273 if (sps_flags & SPS_BEST)
|
|
|
8274 /* Adjust the word score for how it sounds like. */
|
|
|
8275 rescore_suggestions(su);
|
|
|
8276
|
|
|
8277 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
8278 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
8279 }
|
|
|
8280 }
|
|
|
8281
|
|
|
8282 /*
|
|
|
8283 * Free the info put in "*su" by spell_find_suggest().
|
|
|
8284 */
|
|
|
8285 static void
|
|
|
8286 spell_find_cleanup(su)
|
|
|
8287 suginfo_T *su;
|
|
|
8288 {
|
|
|
8289 int i;
|
|
|
8290
|
|
|
8291 /* Free the suggestions. */
|
|
|
8292 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8293 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
8294 ga_clear(&su->su_ga);
|
|
|
8295 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
8296 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
8297 ga_clear(&su->su_sga);
|
|
323
|
8298
|
|
|
8299 /* Free the banned words. */
|
|
344
|
8300 free_banned(su);
|
|
323
|
8301 }
|
|
|
8302
|
|
|
8303 /*
|
|
324
|
8304 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
8305 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
8306 * The result is NUL terminated.
|
|
323
|
8307 */
|
|
|
8308 static void
|
|
324
|
8309 onecap_copy(word, wcopy, upper)
|
|
323
|
8310 char_u *word;
|
|
|
8311 char_u *wcopy;
|
|
|
8312 int upper; /* TRUE: first letter made upper case */
|
|
|
8313 {
|
|
|
8314 char_u *p;
|
|
|
8315 int c;
|
|
|
8316 int l;
|
|
|
8317
|
|
|
8318 p = word;
|
|
|
8319 #ifdef FEAT_MBYTE
|
|
|
8320 if (has_mbyte)
|
|
474
|
8321 c = mb_cptr2char_adv(&p);
|
|
323
|
8322 else
|
|
|
8323 #endif
|
|
|
8324 c = *p++;
|
|
|
8325 if (upper)
|
|
324
|
8326 c = SPELL_TOUPPER(c);
|
|
323
|
8327 else
|
|
324
|
8328 c = SPELL_TOFOLD(c);
|
|
323
|
8329 #ifdef FEAT_MBYTE
|
|
|
8330 if (has_mbyte)
|
|
|
8331 l = mb_char2bytes(c, wcopy);
|
|
|
8332 else
|
|
|
8333 #endif
|
|
|
8334 {
|
|
|
8335 l = 1;
|
|
|
8336 wcopy[0] = c;
|
|
|
8337 }
|
|
376
|
8338 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
8339 }
|
|
|
8340
|
|
|
8341 /*
|
|
324
|
8342 * Make a copy of "word" with all the letters upper cased into
|
|
|
8343 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
8344 */
|
|
|
8345 static void
|
|
|
8346 allcap_copy(word, wcopy)
|
|
|
8347 char_u *word;
|
|
|
8348 char_u *wcopy;
|
|
|
8349 {
|
|
|
8350 char_u *s;
|
|
|
8351 char_u *d;
|
|
|
8352 int c;
|
|
|
8353
|
|
|
8354 d = wcopy;
|
|
|
8355 for (s = word; *s != NUL; )
|
|
|
8356 {
|
|
|
8357 #ifdef FEAT_MBYTE
|
|
|
8358 if (has_mbyte)
|
|
474
|
8359 c = mb_cptr2char_adv(&s);
|
|
323
|
8360 else
|
|
|
8361 #endif
|
|
|
8362 c = *s++;
|
|
324
|
8363 c = SPELL_TOUPPER(c);
|
|
323
|
8364
|
|
|
8365 #ifdef FEAT_MBYTE
|
|
|
8366 if (has_mbyte)
|
|
|
8367 {
|
|
|
8368 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
8369 break;
|
|
|
8370 d += mb_char2bytes(c, d);
|
|
|
8371 }
|
|
|
8372 else
|
|
|
8373 #endif
|
|
|
8374 {
|
|
|
8375 if (d - wcopy >= MAXWLEN - 1)
|
|
|
8376 break;
|
|
|
8377 *d++ = c;
|
|
|
8378 }
|
|
|
8379 }
|
|
|
8380 *d = NUL;
|
|
|
8381 }
|
|
|
8382
|
|
|
8383 /*
|
|
346
|
8384 * Try finding suggestions by recognizing specific situations.
|
|
|
8385 */
|
|
|
8386 static void
|
|
|
8387 suggest_try_special(su)
|
|
|
8388 suginfo_T *su;
|
|
|
8389 {
|
|
|
8390 char_u *p;
|
|
376
|
8391 size_t len;
|
|
346
|
8392 int c;
|
|
|
8393 char_u word[MAXWLEN];
|
|
|
8394
|
|
|
8395 /*
|
|
|
8396 * Recognize a word that is repeated: "the the".
|
|
|
8397 */
|
|
|
8398 p = skiptowhite(su->su_fbadword);
|
|
|
8399 len = p - su->su_fbadword;
|
|
|
8400 p = skipwhite(p);
|
|
|
8401 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
8402 {
|
|
|
8403 /* Include badflags: if the badword is onecap or allcap
|
|
|
8404 * use that for the goodword too: "The the" -> "The". */
|
|
|
8405 c = su->su_fbadword[len];
|
|
|
8406 su->su_fbadword[len] = NUL;
|
|
|
8407 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
8408 su->su_fbadword[len] = c;
|
|
351
|
8409 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
8410 }
|
|
|
8411 }
|
|
|
8412
|
|
|
8413 /*
|
|
323
|
8414 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
8415 *
|
|
|
8416 * This uses a state machine. At each node in the tree we try various
|
|
|
8417 * operations. When trying if an operation work "depth" is increased and the
|
|
|
8418 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
8419 * character, replace one and delete another. The number of changes is
|
|
|
8420 * limited by su->su_maxscore, checked in try_deeper().
|
|
485
|
8421 *
|
|
|
8422 * After implementing this I noticed an article by Kemal Oflazer that
|
|
|
8423 * describes something similar: "Error-tolerant Finite State Recognition with
|
|
|
8424 * Applications to Morphological Analysis and Spelling Correction" (1996).
|
|
|
8425 * The implementation in the article is simplified and requires a stack of
|
|
|
8426 * unknown depth. The implementation here only needs a stack depth of the
|
|
|
8427 * length of the word.
|
|
323
|
8428 */
|
|
|
8429 static void
|
|
346
|
8430 suggest_try_change(su)
|
|
323
|
8431 suginfo_T *su;
|
|
|
8432 {
|
|
|
8433 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
8434 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
8435 trystate_T stack[MAXWLEN];
|
|
483
|
8436 char_u preword[MAXWLEN * 3]; /* word found with proper case;
|
|
|
8437 * concatanation of prefix compound
|
|
|
8438 * words and split word. NUL terminated
|
|
|
8439 * when going deeper but not when coming
|
|
|
8440 * back. */
|
|
|
8441 char_u compflags[MAXWLEN]; /* compound flags, one for each word */
|
|
323
|
8442 trystate_T *sp;
|
|
|
8443 int newscore;
|
|
|
8444 langp_T *lp;
|
|
375
|
8445 char_u *byts, *fbyts, *pbyts;
|
|
|
8446 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
8447 int depth;
|
|
330
|
8448 int c, c2, c3;
|
|
390
|
8449 int n;
|
|
323
|
8450 int flags;
|
|
|
8451 garray_T *gap;
|
|
324
|
8452 idx_T arridx;
|
|
323
|
8453 int len;
|
|
|
8454 char_u *p;
|
|
|
8455 fromto_T *ftp;
|
|
330
|
8456 int fl = 0, tl;
|
|
346
|
8457 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
482
|
8458 slang_T *slang;
|
|
|
8459 int fword_ends;
|
|
323
|
8460
|
|
|
8461 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
8462 * to find matches (esp. REP items). Append some more text, changing
|
|
|
8463 * chars after the bad word may help. */
|
|
323
|
8464 STRCPY(fword, su->su_fbadword);
|
|
346
|
8465 n = STRLEN(fword);
|
|
|
8466 p = su->su_badptr + su->su_badlen;
|
|
|
8467 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
8468
|
|
|
8469 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8470 lp->lp_slang != NULL; ++lp)
|
|
|
8471 {
|
|
482
|
8472 slang = lp->lp_slang;
|
|
|
8473
|
|
323
|
8474 /*
|
|
|
8475 * Go through the whole case-fold tree, try changes at each node.
|
|
|
8476 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
8477 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
8478 * word).
|
|
|
8479 */
|
|
|
8480 depth = 0;
|
|
375
|
8481 sp = &stack[0];
|
|
483
|
8482 vim_memset(sp, 0, sizeof(trystate_T));
|
|
375
|
8483 sp->ts_curi = 1;
|
|
330
|
8484
|
|
|
8485 /*
|
|
375
|
8486 * When there are postponed prefixes we need to use these first. At
|
|
|
8487 * the end of the prefix we continue in the case-fold tree.
|
|
|
8488 */
|
|
482
|
8489 fbyts = slang->sl_fbyts;
|
|
|
8490 fidxs = slang->sl_fidxs;
|
|
|
8491 pbyts = slang->sl_pbyts;
|
|
|
8492 pidxs = slang->sl_pidxs;
|
|
375
|
8493 if (pbyts != NULL)
|
|
|
8494 {
|
|
|
8495 byts = pbyts;
|
|
|
8496 idxs = pidxs;
|
|
482
|
8497 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
375
|
8498 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
8499 }
|
|
|
8500 else
|
|
|
8501 {
|
|
|
8502 byts = fbyts;
|
|
|
8503 idxs = fidxs;
|
|
482
|
8504 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
485
|
8505 sp->ts_state = STATE_START;
|
|
375
|
8506 }
|
|
|
8507
|
|
|
8508 /*
|
|
330
|
8509 * Loop to find all suggestions. At each round we either:
|
|
|
8510 * - For the current state try one operation, advance "ts_curi",
|
|
|
8511 * increase "depth".
|
|
|
8512 * - When a state is done go to the next, set "ts_state".
|
|
|
8513 * - When all states are tried decrease "depth".
|
|
|
8514 */
|
|
323
|
8515 while (depth >= 0 && !got_int)
|
|
|
8516 {
|
|
|
8517 sp = &stack[depth];
|
|
|
8518 switch (sp->ts_state)
|
|
|
8519 {
|
|
|
8520 case STATE_START:
|
|
375
|
8521 case STATE_NOPREFIX:
|
|
323
|
8522 /*
|
|
|
8523 * Start of node: Deal with NUL bytes, which means
|
|
|
8524 * tword[] may end here.
|
|
|
8525 */
|
|
|
8526 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
8527 len = byts[arridx]; /* bytes in this node */
|
|
|
8528 arridx += sp->ts_curi; /* index of current byte */
|
|
|
8529
|
|
482
|
8530 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
8531 {
|
|
|
8532 /* Skip over the NUL bytes, we use them later. */
|
|
|
8533 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
8534 ;
|
|
|
8535 sp->ts_curi += n;
|
|
|
8536
|
|
390
|
8537 /* Always past NUL bytes now. */
|
|
|
8538 n = (int)sp->ts_state;
|
|
|
8539 sp->ts_state = STATE_ENDNUL;
|
|
455
|
8540 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
8541
|
|
375
|
8542 /* At end of a prefix or at start of prefixtree: check for
|
|
|
8543 * following word. */
|
|
390
|
8544 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
8545 {
|
|
455
|
8546 /* Set su->su_badflags to the caps type at this
|
|
|
8547 * position. Use the caps type until here for the
|
|
|
8548 * prefix itself. */
|
|
|
8549 #ifdef FEAT_MBYTE
|
|
|
8550 if (has_mbyte)
|
|
|
8551 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
8552 else
|
|
|
8553 #endif
|
|
|
8554 n = sp->ts_fidx;
|
|
474
|
8555 flags = badword_captype(su->su_badptr,
|
|
|
8556 su->su_badptr + n);
|
|
|
8557 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
8558 su->su_badptr + su->su_badlen);
|
|
375
|
8559 ++depth;
|
|
|
8560 stack[depth] = stack[depth - 1];
|
|
|
8561 sp = &stack[depth];
|
|
|
8562 sp->ts_prefixdepth = depth - 1;
|
|
|
8563 byts = fbyts;
|
|
|
8564 idxs = fidxs;
|
|
|
8565 sp->ts_state = STATE_START;
|
|
|
8566 sp->ts_curi = 1; /* start just after length byte */
|
|
|
8567 sp->ts_arridx = 0;
|
|
|
8568
|
|
455
|
8569 /* Move the prefix to preword[] with the right case
|
|
|
8570 * and make find_keepcap_word() works. */
|
|
485
|
8571 tword[sp->ts_twordlen] = NUL;
|
|
|
8572 make_case_word(tword + sp->ts_splitoff,
|
|
|
8573 preword + sp->ts_prewordlen,
|
|
|
8574 flags);
|
|
|
8575 sp->ts_prewordlen = STRLEN(preword);
|
|
483
|
8576 sp->ts_splitoff = sp->ts_twordlen;
|
|
375
|
8577 }
|
|
|
8578 break;
|
|
|
8579 }
|
|
|
8580
|
|
346
|
8581 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
8582 {
|
|
|
8583 /* Past bytes in node and/or past NUL bytes. */
|
|
|
8584 sp->ts_state = STATE_ENDNUL;
|
|
455
|
8585 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
8586 break;
|
|
|
8587 }
|
|
|
8588
|
|
|
8589 /*
|
|
|
8590 * End of word in tree.
|
|
|
8591 */
|
|
|
8592 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
8593
|
|
324
|
8594 flags = (int)idxs[arridx];
|
|
483
|
8595 fword_ends = (fword[sp->ts_fidx] == NUL
|
|
|
8596 || !spell_iswordp(fword + sp->ts_fidx, curbuf));
|
|
|
8597 tword[sp->ts_twordlen] = NUL;
|
|
323
|
8598
|
|
485
|
8599 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL
|
|
|
8600 && (sp->ts_flags & TSF_PREFIXOK) == 0)
|
|
375
|
8601 {
|
|
|
8602 /* There was a prefix before the word. Check that the
|
|
|
8603 * prefix can be used with this word. */
|
|
|
8604 /* Count the length of the NULs in the prefix. If there
|
|
|
8605 * are none this must be the first try without a prefix.
|
|
|
8606 */
|
|
|
8607 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
8608 len = pbyts[n++];
|
|
|
8609 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
8610 ;
|
|
|
8611 if (c > 0)
|
|
|
8612 {
|
|
390
|
8613 c = valid_word_prefix(c, n, flags,
|
|
483
|
8614 tword + sp->ts_splitoff, slang, FALSE);
|
|
375
|
8615 if (c == 0)
|
|
|
8616 break;
|
|
|
8617
|
|
|
8618 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
8619 if (c & WF_RAREPFX)
|
|
|
8620 flags |= WF_RARE;
|
|
485
|
8621
|
|
|
8622 /* Tricky: when checking for both prefix and
|
|
|
8623 * compounding we run into the prefix flag first.
|
|
|
8624 * Remember that it's OK, so that we accept the prefix
|
|
|
8625 * when arriving at a compound flag. */
|
|
|
8626 sp->ts_flags |= TSF_PREFIXOK;
|
|
375
|
8627 }
|
|
|
8628 }
|
|
|
8629
|
|
485
|
8630 if (sp->ts_complen > sp->ts_compsplit)
|
|
|
8631 {
|
|
|
8632 /* There was a compound word before this word. If this
|
|
|
8633 * word does not support compounding then give up
|
|
|
8634 * (splitting is tried for the word without compound
|
|
|
8635 * flag). */
|
|
|
8636 if (((unsigned)flags >> 24) == 0
|
|
|
8637 || sp->ts_twordlen - sp->ts_splitoff
|
|
|
8638 < slang->sl_compminlen)
|
|
|
8639 break;
|
|
|
8640 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
8641 compflags[sp->ts_complen + 1] = NUL;
|
|
490
|
8642 vim_strncpy(preword + sp->ts_prewordlen,
|
|
|
8643 tword + sp->ts_splitoff,
|
|
|
8644 sp->ts_twordlen - sp->ts_splitoff);
|
|
|
8645 p = preword;
|
|
|
8646 while (*skiptowhite(p) != NUL)
|
|
|
8647 p = skipwhite(skiptowhite(p));
|
|
|
8648 if (fword_ends && !can_compound(slang, p,
|
|
485
|
8649 compflags + sp->ts_compsplit))
|
|
|
8650 break;
|
|
490
|
8651
|
|
|
8652 /* Get pointer to last char of previous word. */
|
|
|
8653 p = preword + sp->ts_prewordlen;
|
|
|
8654 mb_ptr_back(preword, p);
|
|
|
8655 }
|
|
|
8656 else
|
|
|
8657 p = NULL;
|
|
485
|
8658
|
|
323
|
8659 /*
|
|
|
8660 * Form the word with proper case in preword.
|
|
|
8661 * If there is a word from a previous split, append.
|
|
|
8662 */
|
|
|
8663 if (flags & WF_KEEPCAP)
|
|
|
8664 /* Must find the word in the keep-case tree. */
|
|
483
|
8665 find_keepcap_word(slang, tword + sp->ts_splitoff,
|
|
|
8666 preword + sp->ts_prewordlen);
|
|
323
|
8667 else
|
|
346
|
8668 {
|
|
323
|
8669 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
8670 * use that for the goodword too. But if the badword is
|
|
|
8671 * allcap and it's only one char long use onecap. */
|
|
|
8672 c = su->su_badflags;
|
|
|
8673 if ((c & WF_ALLCAP)
|
|
|
8674 #ifdef FEAT_MBYTE
|
|
474
|
8675 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
8676 #else
|
|
|
8677 && su->su_badlen == 1
|
|
|
8678 #endif
|
|
|
8679 )
|
|
|
8680 c = WF_ONECAP;
|
|
490
|
8681 c |= flags;
|
|
|
8682
|
|
|
8683 /* When appending a compound word after a word character
|
|
|
8684 * don't use Onecap. */
|
|
|
8685 if (p != NULL && spell_iswordp_nmw(p))
|
|
|
8686 c &= ~WF_ONECAP;
|
|
483
|
8687 make_case_word(tword + sp->ts_splitoff,
|
|
490
|
8688 preword + sp->ts_prewordlen, c);
|
|
346
|
8689 }
|
|
323
|
8690
|
|
|
8691 /* Don't use a banned word. It may appear again as a good
|
|
|
8692 * word, thus remember it. */
|
|
|
8693 if (flags & WF_BANNED)
|
|
|
8694 {
|
|
483
|
8695 add_banned(su, preword + sp->ts_prewordlen);
|
|
323
|
8696 break;
|
|
|
8697 }
|
|
483
|
8698 if (was_banned(su, preword + sp->ts_prewordlen)
|
|
372
|
8699 || was_banned(su, preword))
|
|
323
|
8700 break;
|
|
|
8701
|
|
|
8702 newscore = 0;
|
|
|
8703 if ((flags & WF_REGION)
|
|
390
|
8704 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
8705 newscore += SCORE_REGION;
|
|
|
8706 if (flags & WF_RARE)
|
|
|
8707 newscore += SCORE_RARE;
|
|
|
8708
|
|
346
|
8709 if (!spell_valid_case(su->su_badflags,
|
|
483
|
8710 captype(preword + sp->ts_prewordlen, NULL)))
|
|
323
|
8711 newscore += SCORE_ICASE;
|
|
|
8712
|
|
482
|
8713 if (fword_ends && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
8714 {
|
|
366
|
8715 /* The badword also ends: add suggestions. Give a penalty
|
|
|
8716 * when changing non-word char to word char, e.g., "thes,"
|
|
|
8717 * -> "these". */
|
|
|
8718 p = fword + sp->ts_fidx;
|
|
|
8719 #ifdef FEAT_MBYTE
|
|
|
8720 if (has_mbyte)
|
|
|
8721 mb_ptr_back(fword, p);
|
|
|
8722 else
|
|
|
8723 #endif
|
|
|
8724 --p;
|
|
376
|
8725 if (!spell_iswordp(p, curbuf))
|
|
366
|
8726 {
|
|
|
8727 p = preword + STRLEN(preword);
|
|
|
8728 #ifdef FEAT_MBYTE
|
|
|
8729 if (has_mbyte)
|
|
|
8730 mb_ptr_back(preword, p);
|
|
|
8731 else
|
|
|
8732 #endif
|
|
|
8733 --p;
|
|
376
|
8734 if (spell_iswordp(p, curbuf))
|
|
366
|
8735 newscore += SCORE_NONWORD;
|
|
|
8736 }
|
|
|
8737
|
|
344
|
8738 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
8739 sp->ts_fidx - repextra,
|
|
351
|
8740 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
8741 }
|
|
482
|
8742 else if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends)
|
|
330
|
8743 #ifdef FEAT_MBYTE
|
|
|
8744 /* Don't split halfway a character. */
|
|
|
8745 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
8746 #endif
|
|
|
8747 )
|
|
323
|
8748 {
|
|
482
|
8749 int try_compound;
|
|
|
8750
|
|
|
8751 /* Get here in two situations:
|
|
|
8752 * 1. The word in the tree ends but the badword continues:
|
|
|
8753 * If the word allows compounding try that. Otherwise
|
|
|
8754 * try a split by inserting a space. For both check
|
|
|
8755 * that a valid words starts at fword[sp->ts_fidx].
|
|
|
8756 * 2. The badword does end, but it was due to a change
|
|
|
8757 * (e.g., a swap). No need to split, but do check that
|
|
|
8758 * the following word is valid.
|
|
|
8759 */
|
|
485
|
8760 try_compound = FALSE;
|
|
482
|
8761 if (!fword_ends
|
|
483
|
8762 && ((unsigned)flags >> 24) != 0
|
|
|
8763 && sp->ts_twordlen - sp->ts_splitoff
|
|
482
|
8764 >= slang->sl_compminlen
|
|
490
|
8765 && (slang->sl_compsylmax < MAXWLEN
|
|
|
8766 || sp->ts_complen + 1 - sp->ts_compsplit
|
|
|
8767 < slang->sl_compmax)
|
|
485
|
8768 && (vim_strchr(sp->ts_complen == sp->ts_compsplit
|
|
|
8769 ? slang->sl_compstartflags
|
|
|
8770 : slang->sl_compallflags,
|
|
483
|
8771 ((unsigned)flags >> 24)) != NULL))
|
|
|
8772 {
|
|
482
|
8773 try_compound = TRUE;
|
|
483
|
8774 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
8775 compflags[sp->ts_complen + 1] = NUL;
|
|
|
8776 }
|
|
485
|
8777
|
|
|
8778 /* If we could add a compound word, and it's also possible
|
|
|
8779 * to split at this point, do the split first and set
|
|
|
8780 * TSF_DIDSPLIT to avoid doing it again. */
|
|
|
8781 if (!fword_ends
|
|
|
8782 && try_compound
|
|
|
8783 && (sp->ts_flags & TSF_DIDSPLIT) == 0)
|
|
482
|
8784 {
|
|
|
8785 try_compound = FALSE;
|
|
485
|
8786 sp->ts_flags |= TSF_DIDSPLIT;
|
|
|
8787 --sp->ts_curi; /* do the same NUL again */
|
|
|
8788 compflags[sp->ts_complen] = NUL;
|
|
|
8789 }
|
|
|
8790 else
|
|
|
8791 sp->ts_flags &= ~TSF_DIDSPLIT;
|
|
|
8792
|
|
|
8793 if (!try_compound && !fword_ends)
|
|
|
8794 {
|
|
|
8795 /* If we're going to split need to check that the
|
|
|
8796 * words so far are valid for compounding. */
|
|
490
|
8797 p = preword;
|
|
|
8798 while (*skiptowhite(p) != NUL)
|
|
|
8799 p = skipwhite(skiptowhite(p));
|
|
485
|
8800 if (sp->ts_complen > sp->ts_compsplit
|
|
490
|
8801 && !can_compound(slang, p,
|
|
485
|
8802 compflags + sp->ts_compsplit))
|
|
|
8803 break;
|
|
|
8804 newscore += SCORE_SPLIT;
|
|
482
|
8805 }
|
|
|
8806
|
|
|
8807 if (try_deeper(su, stack, depth, newscore))
|
|
323
|
8808 {
|
|
|
8809 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
346
|
8810 sp->ts_save_badflags = su->su_badflags;
|
|
376
|
8811 sp->ts_state = STATE_SPLITUNDO;
|
|
|
8812
|
|
|
8813 ++depth;
|
|
|
8814 sp = &stack[depth];
|
|
323
|
8815
|
|
482
|
8816 /* Append a space to preword when splitting. */
|
|
|
8817 if (!try_compound && !fword_ends)
|
|
|
8818 STRCAT(preword, " ");
|
|
483
|
8819 sp->ts_prewordlen = STRLEN(preword);
|
|
|
8820 sp->ts_splitoff = sp->ts_twordlen;
|
|
376
|
8821
|
|
|
8822 /* If the badword has a non-word character at this
|
|
|
8823 * position skip it. That means replacing the
|
|
482
|
8824 * non-word character with a space. Always skip a
|
|
|
8825 * character when the word ends. */
|
|
|
8826 if ((!try_compound
|
|
|
8827 && !spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
8828 || fword_ends)
|
|
376
|
8829 {
|
|
482
|
8830 int l;
|
|
|
8831
|
|
376
|
8832 #ifdef FEAT_MBYTE
|
|
|
8833 if (has_mbyte)
|
|
482
|
8834 l = MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
376
|
8835 else
|
|
|
8836 #endif
|
|
482
|
8837 l = 1;
|
|
|
8838 if (fword_ends)
|
|
|
8839 {
|
|
|
8840 /* Copy the skipped character to preword. */
|
|
483
|
8841 mch_memmove(preword + sp->ts_prewordlen,
|
|
482
|
8842 fword + sp->ts_fidx, l);
|
|
483
|
8843 sp->ts_prewordlen += l;
|
|
|
8844 preword[sp->ts_prewordlen] = NUL;
|
|
482
|
8845 }
|
|
|
8846 else
|
|
|
8847 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
8848 sp->ts_fidx += l;
|
|
376
|
8849 }
|
|
455
|
8850
|
|
485
|
8851 /* When compounding include compound flag in
|
|
|
8852 * compflags[] (already set above). When splitting we
|
|
|
8853 * may start compounding over again. */
|
|
482
|
8854 if (try_compound)
|
|
483
|
8855 ++sp->ts_complen;
|
|
482
|
8856 else
|
|
485
|
8857 sp->ts_compsplit = sp->ts_complen;
|
|
|
8858 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
482
|
8859
|
|
455
|
8860 /* set su->su_badflags to the caps type at this
|
|
|
8861 * position */
|
|
324
|
8862 #ifdef FEAT_MBYTE
|
|
|
8863 if (has_mbyte)
|
|
455
|
8864 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
8865 else
|
|
|
8866 #endif
|
|
455
|
8867 n = sp->ts_fidx;
|
|
474
|
8868 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
8869 su->su_badptr + su->su_badlen);
|
|
323
|
8870
|
|
|
8871 /* Restart at top of the tree. */
|
|
376
|
8872 sp->ts_arridx = 0;
|
|
485
|
8873
|
|
|
8874 /* If there are postponed prefixes, try these too. */
|
|
|
8875 if (pbyts != NULL)
|
|
|
8876 {
|
|
|
8877 byts = pbyts;
|
|
|
8878 idxs = pidxs;
|
|
|
8879 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
|
8880 sp->ts_state = STATE_NOPREFIX;
|
|
|
8881 }
|
|
323
|
8882 }
|
|
|
8883 }
|
|
|
8884 break;
|
|
|
8885
|
|
|
8886 case STATE_SPLITUNDO:
|
|
482
|
8887 /* Undo the changes done for word split or compound word. */
|
|
346
|
8888 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
8889
|
|
|
8890 /* Continue looking for NUL bytes. */
|
|
|
8891 sp->ts_state = STATE_START;
|
|
485
|
8892
|
|
|
8893 /* In case we went into the prefix tree. */
|
|
|
8894 byts = fbyts;
|
|
|
8895 idxs = fidxs;
|
|
323
|
8896 break;
|
|
|
8897
|
|
|
8898 case STATE_ENDNUL:
|
|
|
8899 /* Past the NUL bytes in the node. */
|
|
455
|
8900 su->su_badflags = sp->ts_save_badflags;
|
|
346
|
8901 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
8902 {
|
|
|
8903 /* The badword ends, can't use the bytes in this node. */
|
|
|
8904 sp->ts_state = STATE_DEL;
|
|
|
8905 break;
|
|
|
8906 }
|
|
|
8907 sp->ts_state = STATE_PLAIN;
|
|
|
8908 /*FALLTHROUGH*/
|
|
|
8909
|
|
|
8910 case STATE_PLAIN:
|
|
|
8911 /*
|
|
|
8912 * Go over all possible bytes at this node, add each to
|
|
|
8913 * tword[] and use child node. "ts_curi" is the index.
|
|
|
8914 */
|
|
|
8915 arridx = sp->ts_arridx;
|
|
|
8916 if (sp->ts_curi > byts[arridx])
|
|
|
8917 {
|
|
|
8918 /* Done all bytes at this node, do next state. When still
|
|
|
8919 * at already changed bytes skip the other tricks. */
|
|
|
8920 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
8921 sp->ts_state = STATE_DEL;
|
|
|
8922 else
|
|
|
8923 sp->ts_state = STATE_FINAL;
|
|
|
8924 }
|
|
|
8925 else
|
|
|
8926 {
|
|
|
8927 arridx += sp->ts_curi++;
|
|
|
8928 c = byts[arridx];
|
|
|
8929
|
|
|
8930 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
8931 * the byte in the bad word adjust the score. But don't
|
|
|
8932 * even try when the byte was already changed. */
|
|
330
|
8933 if (c == fword[sp->ts_fidx]
|
|
|
8934 #ifdef FEAT_MBYTE
|
|
|
8935 || (sp->ts_tcharlen > 0
|
|
|
8936 && sp->ts_isdiff != DIFF_NONE)
|
|
|
8937 #endif
|
|
|
8938 )
|
|
323
|
8939 newscore = 0;
|
|
|
8940 else
|
|
|
8941 newscore = SCORE_SUBST;
|
|
|
8942 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
8943 && try_deeper(su, stack, depth, newscore))
|
|
|
8944 {
|
|
|
8945 ++depth;
|
|
330
|
8946 sp = &stack[depth];
|
|
|
8947 ++sp->ts_fidx;
|
|
|
8948 tword[sp->ts_twordlen++] = c;
|
|
|
8949 sp->ts_arridx = idxs[arridx];
|
|
|
8950 #ifdef FEAT_MBYTE
|
|
|
8951 if (newscore == SCORE_SUBST)
|
|
|
8952 sp->ts_isdiff = DIFF_YES;
|
|
|
8953 if (has_mbyte)
|
|
|
8954 {
|
|
|
8955 /* Multi-byte characters are a bit complicated to
|
|
|
8956 * handle: They differ when any of the bytes
|
|
|
8957 * differ and then their length may also differ. */
|
|
|
8958 if (sp->ts_tcharlen == 0)
|
|
|
8959 {
|
|
|
8960 /* First byte. */
|
|
|
8961 sp->ts_tcharidx = 0;
|
|
|
8962 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
8963 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
8964 sp->ts_isdiff = (newscore != 0)
|
|
|
8965 ? DIFF_YES : DIFF_NONE;
|
|
|
8966 }
|
|
|
8967 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
8968 /* When inserting trail bytes don't advance in
|
|
|
8969 * the bad word. */
|
|
|
8970 --sp->ts_fidx;
|
|
|
8971 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
8972 {
|
|
|
8973 /* Last byte of character. */
|
|
|
8974 if (sp->ts_isdiff == DIFF_YES)
|
|
|
8975 {
|
|
|
8976 /* Correct ts_fidx for the byte length of
|
|
|
8977 * the character (we didn't check that
|
|
|
8978 * before). */
|
|
|
8979 sp->ts_fidx = sp->ts_fcharstart
|
|
|
8980 + MB_BYTE2LEN(
|
|
|
8981 fword[sp->ts_fcharstart]);
|
|
|
8982
|
|
479
|
8983 /* For changing a composing character
|
|
|
8984 * adjust the score from SCORE_SUBST to
|
|
|
8985 * SCORE_SUBCOMP. */
|
|
|
8986 if (enc_utf8
|
|
|
8987 && utf_iscomposing(
|
|
|
8988 mb_ptr2char(tword
|
|
|
8989 + sp->ts_twordlen
|
|
|
8990 - sp->ts_tcharlen))
|
|
|
8991 && utf_iscomposing(
|
|
|
8992 mb_ptr2char(fword
|
|
|
8993 + sp->ts_fcharstart)))
|
|
|
8994 sp->ts_score -=
|
|
|
8995 SCORE_SUBST - SCORE_SUBCOMP;
|
|
|
8996
|
|
330
|
8997 /* For a similar character adjust score
|
|
|
8998 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
482
|
8999 else if (slang->sl_has_map
|
|
|
9000 && similar_chars(slang,
|
|
330
|
9001 mb_ptr2char(tword
|
|
|
9002 + sp->ts_twordlen
|
|
|
9003 - sp->ts_tcharlen),
|
|
|
9004 mb_ptr2char(fword
|
|
|
9005 + sp->ts_fcharstart)))
|
|
|
9006 sp->ts_score -=
|
|
|
9007 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9008 }
|
|
358
|
9009 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
9010 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
9011 {
|
|
|
9012 p = tword + sp->ts_twordlen
|
|
|
9013 - sp->ts_tcharlen;
|
|
|
9014 c = mb_ptr2char(p);
|
|
475
|
9015 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9016 {
|
|
|
9017 /* Inserting a composing char doesn't
|
|
|
9018 * count that much. */
|
|
358
|
9019 sp->ts_score -= SCORE_INS
|
|
475
|
9020 - SCORE_INSCOMP;
|
|
|
9021 }
|
|
|
9022 else
|
|
|
9023 {
|
|
|
9024 /* If the previous character was the
|
|
|
9025 * same, thus doubling a character,
|
|
|
9026 * give a bonus to the score. */
|
|
|
9027 mb_ptr_back(tword, p);
|
|
|
9028 if (c == mb_ptr2char(p))
|
|
|
9029 sp->ts_score -= SCORE_INS
|
|
358
|
9030 - SCORE_INSDUP;
|
|
475
|
9031 }
|
|
358
|
9032 }
|
|
330
|
9033
|
|
|
9034 /* Starting a new char, reset the length. */
|
|
|
9035 sp->ts_tcharlen = 0;
|
|
|
9036 }
|
|
|
9037 }
|
|
|
9038 else
|
|
|
9039 #endif
|
|
|
9040 {
|
|
|
9041 /* If we found a similar char adjust the score.
|
|
|
9042 * We do this after calling try_deeper() because
|
|
|
9043 * it's slow. */
|
|
|
9044 if (newscore != 0
|
|
482
|
9045 && slang->sl_has_map
|
|
|
9046 && similar_chars(slang,
|
|
330
|
9047 c, fword[sp->ts_fidx - 1]))
|
|
|
9048 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9049 }
|
|
323
|
9050 }
|
|
|
9051 }
|
|
|
9052 break;
|
|
|
9053
|
|
|
9054 case STATE_DEL:
|
|
330
|
9055 #ifdef FEAT_MBYTE
|
|
|
9056 /* When past the first byte of a multi-byte char don't try
|
|
|
9057 * delete/insert/swap a character. */
|
|
|
9058 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
9059 {
|
|
|
9060 sp->ts_state = STATE_FINAL;
|
|
|
9061 break;
|
|
|
9062 }
|
|
|
9063 #endif
|
|
|
9064 /*
|
|
|
9065 * Try skipping one character in the bad word (delete it).
|
|
|
9066 */
|
|
323
|
9067 sp->ts_state = STATE_INS;
|
|
|
9068 sp->ts_curi = 1;
|
|
|
9069 if (fword[sp->ts_fidx] != NUL
|
|
|
9070 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
9071 {
|
|
|
9072 ++depth;
|
|
358
|
9073
|
|
|
9074 /* Advance over the character in fword[]. Give a bonus to
|
|
|
9075 * the score if the same character is following "nn" ->
|
|
|
9076 * "n". */
|
|
330
|
9077 #ifdef FEAT_MBYTE
|
|
|
9078 if (has_mbyte)
|
|
358
|
9079 {
|
|
|
9080 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
9081 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
479
|
9082 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9083 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
|
|
|
9084 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
358
|
9085 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
9086 }
|
|
330
|
9087 else
|
|
|
9088 #endif
|
|
358
|
9089 {
|
|
330
|
9090 ++stack[depth].ts_fidx;
|
|
358
|
9091 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
9092 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
9093 }
|
|
323
|
9094 break;
|
|
|
9095 }
|
|
|
9096 /*FALLTHROUGH*/
|
|
|
9097
|
|
|
9098 case STATE_INS:
|
|
330
|
9099 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
9100 * node. */
|
|
|
9101 n = sp->ts_arridx;
|
|
|
9102 if (sp->ts_curi > byts[n])
|
|
|
9103 {
|
|
|
9104 /* Done all bytes at this node, do next state. */
|
|
|
9105 sp->ts_state = STATE_SWAP;
|
|
|
9106 }
|
|
|
9107 else
|
|
|
9108 {
|
|
330
|
9109 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
9110 n += sp->ts_curi++;
|
|
|
9111 c = byts[n];
|
|
|
9112 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
9113 {
|
|
|
9114 ++depth;
|
|
330
|
9115 sp = &stack[depth];
|
|
|
9116 tword[sp->ts_twordlen++] = c;
|
|
|
9117 sp->ts_arridx = idxs[n];
|
|
|
9118 #ifdef FEAT_MBYTE
|
|
|
9119 if (has_mbyte)
|
|
|
9120 {
|
|
|
9121 fl = MB_BYTE2LEN(c);
|
|
|
9122 if (fl > 1)
|
|
|
9123 {
|
|
|
9124 /* There are following bytes for the same
|
|
|
9125 * character. We must find all bytes before
|
|
|
9126 * trying delete/insert/swap/etc. */
|
|
|
9127 sp->ts_tcharlen = fl;
|
|
|
9128 sp->ts_tcharidx = 1;
|
|
|
9129 sp->ts_isdiff = DIFF_INSERT;
|
|
|
9130 }
|
|
|
9131 }
|
|
358
|
9132 else
|
|
|
9133 fl = 1;
|
|
|
9134 if (fl == 1)
|
|
330
|
9135 #endif
|
|
358
|
9136 {
|
|
|
9137 /* If the previous character was the same, thus
|
|
|
9138 * doubling a character, give a bonus to the
|
|
|
9139 * score. */
|
|
|
9140 if (sp->ts_twordlen >= 2
|
|
|
9141 && tword[sp->ts_twordlen - 2] == c)
|
|
|
9142 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
9143 }
|
|
323
|
9144 }
|
|
|
9145 }
|
|
|
9146 break;
|
|
|
9147
|
|
|
9148 case STATE_SWAP:
|
|
330
|
9149 /*
|
|
|
9150 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
9151 * We change "fword" here, it's changed back afterwards.
|
|
|
9152 */
|
|
|
9153 p = fword + sp->ts_fidx;
|
|
|
9154 c = *p;
|
|
|
9155 if (c == NUL)
|
|
|
9156 {
|
|
|
9157 /* End of word, can't swap or replace. */
|
|
|
9158 sp->ts_state = STATE_FINAL;
|
|
|
9159 break;
|
|
|
9160 }
|
|
|
9161 #ifdef FEAT_MBYTE
|
|
|
9162 if (has_mbyte)
|
|
|
9163 {
|
|
474
|
9164 n = mb_cptr2len(p);
|
|
330
|
9165 c = mb_ptr2char(p);
|
|
|
9166 c2 = mb_ptr2char(p + n);
|
|
|
9167 }
|
|
|
9168 else
|
|
|
9169 #endif
|
|
|
9170 c2 = p[1];
|
|
|
9171 if (c == c2)
|
|
323
|
9172 {
|
|
330
|
9173 /* Characters are identical, swap won't do anything. */
|
|
|
9174 sp->ts_state = STATE_SWAP3;
|
|
|
9175 break;
|
|
|
9176 }
|
|
|
9177 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
9178 {
|
|
|
9179 sp->ts_state = STATE_UNSWAP;
|
|
323
|
9180 ++depth;
|
|
330
|
9181 #ifdef FEAT_MBYTE
|
|
|
9182 if (has_mbyte)
|
|
|
9183 {
|
|
|
9184 fl = mb_char2len(c2);
|
|
|
9185 mch_memmove(p, p + n, fl);
|
|
|
9186 mb_char2bytes(c, p + fl);
|
|
|
9187 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
9188 }
|
|
|
9189 else
|
|
|
9190 #endif
|
|
|
9191 {
|
|
|
9192 p[0] = c2;
|
|
|
9193 p[1] = c;
|
|
|
9194 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
9195 }
|
|
323
|
9196 }
|
|
|
9197 else
|
|
|
9198 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
9199 sp->ts_state = STATE_REP_INI;
|
|
|
9200 break;
|
|
|
9201
|
|
330
|
9202 case STATE_UNSWAP:
|
|
|
9203 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
9204 p = fword + sp->ts_fidx;
|
|
|
9205 #ifdef FEAT_MBYTE
|
|
|
9206 if (has_mbyte)
|
|
|
9207 {
|
|
|
9208 n = MB_BYTE2LEN(*p);
|
|
|
9209 c = mb_ptr2char(p + n);
|
|
|
9210 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
9211 mb_char2bytes(c, p);
|
|
|
9212 }
|
|
|
9213 else
|
|
|
9214 #endif
|
|
|
9215 {
|
|
|
9216 c = *p;
|
|
|
9217 *p = p[1];
|
|
|
9218 p[1] = c;
|
|
|
9219 }
|
|
|
9220 /*FALLTHROUGH*/
|
|
|
9221
|
|
|
9222 case STATE_SWAP3:
|
|
323
|
9223 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
9224 * "fword" here, it's changed back afterwards. */
|
|
|
9225 p = fword + sp->ts_fidx;
|
|
|
9226 #ifdef FEAT_MBYTE
|
|
|
9227 if (has_mbyte)
|
|
|
9228 {
|
|
474
|
9229 n = mb_cptr2len(p);
|
|
330
|
9230 c = mb_ptr2char(p);
|
|
474
|
9231 fl = mb_cptr2len(p + n);
|
|
330
|
9232 c2 = mb_ptr2char(p + n);
|
|
|
9233 c3 = mb_ptr2char(p + n + fl);
|
|
|
9234 }
|
|
|
9235 else
|
|
|
9236 #endif
|
|
323
|
9237 {
|
|
330
|
9238 c = *p;
|
|
|
9239 c2 = p[1];
|
|
|
9240 c3 = p[2];
|
|
|
9241 }
|
|
|
9242
|
|
|
9243 /* When characters are identical: "121" then SWAP3 result is
|
|
|
9244 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
9245 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
9246 * swapping. Also skip when c3 is NUL. */
|
|
|
9247 if (c == c3 || c3 == NUL)
|
|
|
9248 {
|
|
|
9249 sp->ts_state = STATE_REP_INI;
|
|
|
9250 break;
|
|
|
9251 }
|
|
|
9252 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9253 {
|
|
|
9254 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
9255 ++depth;
|
|
330
|
9256 #ifdef FEAT_MBYTE
|
|
|
9257 if (has_mbyte)
|
|
|
9258 {
|
|
|
9259 tl = mb_char2len(c3);
|
|
|
9260 mch_memmove(p, p + n + fl, tl);
|
|
|
9261 mb_char2bytes(c2, p + tl);
|
|
|
9262 mb_char2bytes(c, p + fl + tl);
|
|
|
9263 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
9264 }
|
|
|
9265 else
|
|
|
9266 #endif
|
|
|
9267 {
|
|
|
9268 p[0] = p[2];
|
|
|
9269 p[2] = c;
|
|
|
9270 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9271 }
|
|
323
|
9272 }
|
|
|
9273 else
|
|
|
9274 sp->ts_state = STATE_REP_INI;
|
|
|
9275 break;
|
|
|
9276
|
|
330
|
9277 case STATE_UNSWAP3:
|
|
|
9278 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
9279 p = fword + sp->ts_fidx;
|
|
|
9280 #ifdef FEAT_MBYTE
|
|
|
9281 if (has_mbyte)
|
|
|
9282 {
|
|
|
9283 n = MB_BYTE2LEN(*p);
|
|
|
9284 c2 = mb_ptr2char(p + n);
|
|
|
9285 fl = MB_BYTE2LEN(p[n]);
|
|
|
9286 c = mb_ptr2char(p + n + fl);
|
|
|
9287 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
9288 mch_memmove(p + fl + tl, p, n);
|
|
|
9289 mb_char2bytes(c, p);
|
|
|
9290 mb_char2bytes(c2, p + tl);
|
|
|
9291 }
|
|
|
9292 else
|
|
|
9293 #endif
|
|
|
9294 {
|
|
|
9295 c = *p;
|
|
|
9296 *p = p[2];
|
|
|
9297 p[2] = c;
|
|
|
9298 }
|
|
346
|
9299
|
|
330
|
9300 /* Rotate three characters left: "123" -> "231". We change
|
|
|
9301 * "fword" here, it's changed back afterwards. */
|
|
323
|
9302 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9303 {
|
|
330
|
9304 sp->ts_state = STATE_UNROT3L;
|
|
323
|
9305 ++depth;
|
|
330
|
9306 p = fword + sp->ts_fidx;
|
|
|
9307 #ifdef FEAT_MBYTE
|
|
|
9308 if (has_mbyte)
|
|
|
9309 {
|
|
474
|
9310 n = mb_cptr2len(p);
|
|
330
|
9311 c = mb_ptr2char(p);
|
|
474
|
9312 fl = mb_cptr2len(p + n);
|
|
|
9313 fl += mb_cptr2len(p + n + fl);
|
|
330
|
9314 mch_memmove(p, p + n, fl);
|
|
|
9315 mb_char2bytes(c, p + fl);
|
|
|
9316 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
9317 }
|
|
|
9318 else
|
|
|
9319 #endif
|
|
|
9320 {
|
|
|
9321 c = *p;
|
|
|
9322 *p = p[1];
|
|
|
9323 p[1] = p[2];
|
|
|
9324 p[2] = c;
|
|
|
9325 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9326 }
|
|
323
|
9327 }
|
|
|
9328 else
|
|
|
9329 sp->ts_state = STATE_REP_INI;
|
|
|
9330 break;
|
|
|
9331
|
|
330
|
9332 case STATE_UNROT3L:
|
|
346
|
9333 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
9334 p = fword + sp->ts_fidx;
|
|
|
9335 #ifdef FEAT_MBYTE
|
|
|
9336 if (has_mbyte)
|
|
|
9337 {
|
|
|
9338 n = MB_BYTE2LEN(*p);
|
|
|
9339 n += MB_BYTE2LEN(p[n]);
|
|
|
9340 c = mb_ptr2char(p + n);
|
|
|
9341 tl = MB_BYTE2LEN(p[n]);
|
|
|
9342 mch_memmove(p + tl, p, n);
|
|
|
9343 mb_char2bytes(c, p);
|
|
|
9344 }
|
|
|
9345 else
|
|
|
9346 #endif
|
|
|
9347 {
|
|
|
9348 c = p[2];
|
|
|
9349 p[2] = p[1];
|
|
|
9350 p[1] = *p;
|
|
|
9351 *p = c;
|
|
|
9352 }
|
|
346
|
9353
|
|
323
|
9354 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
9355 * "fword" here, it's changed back afterwards. */
|
|
323
|
9356 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9357 {
|
|
330
|
9358 sp->ts_state = STATE_UNROT3R;
|
|
323
|
9359 ++depth;
|
|
330
|
9360 p = fword + sp->ts_fidx;
|
|
|
9361 #ifdef FEAT_MBYTE
|
|
|
9362 if (has_mbyte)
|
|
|
9363 {
|
|
474
|
9364 n = mb_cptr2len(p);
|
|
|
9365 n += mb_cptr2len(p + n);
|
|
330
|
9366 c = mb_ptr2char(p + n);
|
|
474
|
9367 tl = mb_cptr2len(p + n);
|
|
330
|
9368 mch_memmove(p + tl, p, n);
|
|
|
9369 mb_char2bytes(c, p);
|
|
|
9370 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
9371 }
|
|
|
9372 else
|
|
|
9373 #endif
|
|
|
9374 {
|
|
|
9375 c = p[2];
|
|
|
9376 p[2] = p[1];
|
|
|
9377 p[1] = *p;
|
|
|
9378 *p = c;
|
|
|
9379 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9380 }
|
|
323
|
9381 }
|
|
|
9382 else
|
|
|
9383 sp->ts_state = STATE_REP_INI;
|
|
|
9384 break;
|
|
|
9385
|
|
330
|
9386 case STATE_UNROT3R:
|
|
346
|
9387 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
9388 p = fword + sp->ts_fidx;
|
|
|
9389 #ifdef FEAT_MBYTE
|
|
|
9390 if (has_mbyte)
|
|
|
9391 {
|
|
|
9392 c = mb_ptr2char(p);
|
|
|
9393 tl = MB_BYTE2LEN(*p);
|
|
|
9394 n = MB_BYTE2LEN(p[tl]);
|
|
|
9395 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
9396 mch_memmove(p, p + tl, n);
|
|
|
9397 mb_char2bytes(c, p + n);
|
|
|
9398 }
|
|
|
9399 else
|
|
|
9400 #endif
|
|
|
9401 {
|
|
|
9402 c = *p;
|
|
|
9403 *p = p[1];
|
|
|
9404 p[1] = p[2];
|
|
|
9405 p[2] = c;
|
|
|
9406 }
|
|
323
|
9407 /*FALLTHROUGH*/
|
|
|
9408
|
|
|
9409 case STATE_REP_INI:
|
|
|
9410 /* Check if matching with REP items from the .aff file would
|
|
|
9411 * work. Quickly skip if there are no REP items or the score
|
|
|
9412 * is going to be too high anyway. */
|
|
482
|
9413 gap = &slang->sl_rep;
|
|
323
|
9414 if (gap->ga_len == 0
|
|
|
9415 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
9416 {
|
|
|
9417 sp->ts_state = STATE_FINAL;
|
|
|
9418 break;
|
|
|
9419 }
|
|
|
9420
|
|
|
9421 /* Use the first byte to quickly find the first entry that
|
|
330
|
9422 * may match. If the index is -1 there is none. */
|
|
482
|
9423 sp->ts_curi = slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
323
|
9424 if (sp->ts_curi < 0)
|
|
|
9425 {
|
|
|
9426 sp->ts_state = STATE_FINAL;
|
|
|
9427 break;
|
|
|
9428 }
|
|
|
9429
|
|
|
9430 sp->ts_state = STATE_REP;
|
|
|
9431 /*FALLTHROUGH*/
|
|
|
9432
|
|
|
9433 case STATE_REP:
|
|
|
9434 /* Try matching with REP items from the .aff file. For each
|
|
330
|
9435 * match replace the characters and check if the resulting
|
|
|
9436 * word is valid. */
|
|
323
|
9437 p = fword + sp->ts_fidx;
|
|
|
9438
|
|
482
|
9439 gap = &slang->sl_rep;
|
|
323
|
9440 while (sp->ts_curi < gap->ga_len)
|
|
|
9441 {
|
|
|
9442 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
9443 if (*ftp->ft_from != *p)
|
|
|
9444 {
|
|
|
9445 /* past possible matching entries */
|
|
|
9446 sp->ts_curi = gap->ga_len;
|
|
|
9447 break;
|
|
|
9448 }
|
|
|
9449 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
9450 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
9451 {
|
|
|
9452 /* Need to undo this afterwards. */
|
|
|
9453 sp->ts_state = STATE_REP_UNDO;
|
|
|
9454
|
|
|
9455 /* Change the "from" to the "to" string. */
|
|
|
9456 ++depth;
|
|
|
9457 fl = STRLEN(ftp->ft_from);
|
|
|
9458 tl = STRLEN(ftp->ft_to);
|
|
|
9459 if (fl != tl)
|
|
346
|
9460 {
|
|
323
|
9461 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
9462 repextra += tl - fl;
|
|
|
9463 }
|
|
323
|
9464 mch_memmove(p, ftp->ft_to, tl);
|
|
|
9465 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
9466 #ifdef FEAT_MBYTE
|
|
|
9467 stack[depth].ts_tcharlen = 0;
|
|
|
9468 #endif
|
|
323
|
9469 break;
|
|
|
9470 }
|
|
|
9471 }
|
|
|
9472
|
|
|
9473 if (sp->ts_curi >= gap->ga_len)
|
|
|
9474 /* No (more) matches. */
|
|
|
9475 sp->ts_state = STATE_FINAL;
|
|
|
9476
|
|
|
9477 break;
|
|
|
9478
|
|
|
9479 case STATE_REP_UNDO:
|
|
|
9480 /* Undo a REP replacement and continue with the next one. */
|
|
482
|
9481 ftp = (fromto_T *)slang->sl_rep.ga_data + sp->ts_curi - 1;
|
|
323
|
9482 fl = STRLEN(ftp->ft_from);
|
|
|
9483 tl = STRLEN(ftp->ft_to);
|
|
|
9484 p = fword + sp->ts_fidx;
|
|
|
9485 if (fl != tl)
|
|
346
|
9486 {
|
|
323
|
9487 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
9488 repextra -= tl - fl;
|
|
|
9489 }
|
|
323
|
9490 mch_memmove(p, ftp->ft_from, fl);
|
|
|
9491 sp->ts_state = STATE_REP;
|
|
|
9492 break;
|
|
|
9493
|
|
|
9494 default:
|
|
|
9495 /* Did all possible states at this level, go up one level. */
|
|
|
9496 --depth;
|
|
344
|
9497
|
|
482
|
9498 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
9499 {
|
|
|
9500 /* Continue in or go back to the prefix tree. */
|
|
|
9501 byts = pbyts;
|
|
|
9502 idxs = pidxs;
|
|
|
9503 }
|
|
|
9504
|
|
344
|
9505 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
9506 line_breakcheck();
|
|
323
|
9507 }
|
|
|
9508 }
|
|
|
9509 }
|
|
|
9510 }
|
|
|
9511
|
|
|
9512 /*
|
|
|
9513 * Try going one level deeper in the tree.
|
|
|
9514 */
|
|
|
9515 static int
|
|
|
9516 try_deeper(su, stack, depth, score_add)
|
|
|
9517 suginfo_T *su;
|
|
|
9518 trystate_T *stack;
|
|
|
9519 int depth;
|
|
|
9520 int score_add;
|
|
|
9521 {
|
|
|
9522 int newscore;
|
|
|
9523
|
|
|
9524 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
9525 newscore = stack[depth].ts_score + score_add;
|
|
|
9526 if (newscore >= su->su_maxscore)
|
|
|
9527 return FALSE;
|
|
|
9528
|
|
330
|
9529 stack[depth + 1] = stack[depth];
|
|
323
|
9530 stack[depth + 1].ts_state = STATE_START;
|
|
|
9531 stack[depth + 1].ts_score = newscore;
|
|
|
9532 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
485
|
9533 stack[depth + 1].ts_flags = 0;
|
|
323
|
9534 return TRUE;
|
|
|
9535 }
|
|
|
9536
|
|
455
|
9537 #ifdef FEAT_MBYTE
|
|
|
9538 /*
|
|
|
9539 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
9540 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
9541 */
|
|
|
9542 static int
|
|
|
9543 nofold_len(fword, flen, word)
|
|
|
9544 char_u *fword;
|
|
|
9545 int flen;
|
|
|
9546 char_u *word;
|
|
|
9547 {
|
|
|
9548 char_u *p;
|
|
|
9549 int i = 0;
|
|
|
9550
|
|
|
9551 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
9552 ++i;
|
|
|
9553 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
9554 --i;
|
|
|
9555 return (int)(p - word);
|
|
|
9556 }
|
|
|
9557 #endif
|
|
|
9558
|
|
323
|
9559 /*
|
|
|
9560 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
9561 * words and put it in "kword".
|
|
|
9562 * Theoretically there could be several keep-case words that result in the
|
|
|
9563 * same case-folded word, but we only find one...
|
|
|
9564 */
|
|
|
9565 static void
|
|
|
9566 find_keepcap_word(slang, fword, kword)
|
|
|
9567 slang_T *slang;
|
|
|
9568 char_u *fword;
|
|
|
9569 char_u *kword;
|
|
|
9570 {
|
|
|
9571 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
9572 int depth;
|
|
324
|
9573 idx_T tryidx;
|
|
323
|
9574
|
|
|
9575 /* The following arrays are used at each depth in the tree. */
|
|
324
|
9576 idx_T arridx[MAXWLEN];
|
|
323
|
9577 int round[MAXWLEN];
|
|
|
9578 int fwordidx[MAXWLEN];
|
|
|
9579 int uwordidx[MAXWLEN];
|
|
|
9580 int kwordlen[MAXWLEN];
|
|
|
9581
|
|
|
9582 int flen, ulen;
|
|
|
9583 int l;
|
|
|
9584 int len;
|
|
|
9585 int c;
|
|
324
|
9586 idx_T lo, hi, m;
|
|
323
|
9587 char_u *p;
|
|
|
9588 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
9589 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
9590
|
|
|
9591 if (byts == NULL)
|
|
|
9592 {
|
|
|
9593 /* array is empty: "cannot happen" */
|
|
|
9594 *kword = NUL;
|
|
|
9595 return;
|
|
|
9596 }
|
|
|
9597
|
|
|
9598 /* Make an all-cap version of "fword". */
|
|
|
9599 allcap_copy(fword, uword);
|
|
|
9600
|
|
|
9601 /*
|
|
|
9602 * Each character needs to be tried both case-folded and upper-case.
|
|
|
9603 * All this gets very complicated if we keep in mind that changing case
|
|
|
9604 * may change the byte length of a multi-byte character...
|
|
|
9605 */
|
|
|
9606 depth = 0;
|
|
|
9607 arridx[0] = 0;
|
|
|
9608 round[0] = 0;
|
|
|
9609 fwordidx[0] = 0;
|
|
|
9610 uwordidx[0] = 0;
|
|
|
9611 kwordlen[0] = 0;
|
|
|
9612 while (depth >= 0)
|
|
|
9613 {
|
|
|
9614 if (fword[fwordidx[depth]] == NUL)
|
|
|
9615 {
|
|
|
9616 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
9617 * here we have found a match. */
|
|
|
9618 if (byts[arridx[depth] + 1] == 0)
|
|
|
9619 {
|
|
|
9620 kword[kwordlen[depth]] = NUL;
|
|
|
9621 return;
|
|
|
9622 }
|
|
|
9623
|
|
|
9624 /* kword is getting too long, continue one level up */
|
|
|
9625 --depth;
|
|
|
9626 }
|
|
|
9627 else if (++round[depth] > 2)
|
|
|
9628 {
|
|
|
9629 /* tried both fold-case and upper-case character, continue one
|
|
|
9630 * level up */
|
|
|
9631 --depth;
|
|
|
9632 }
|
|
|
9633 else
|
|
|
9634 {
|
|
|
9635 /*
|
|
|
9636 * round[depth] == 1: Try using the folded-case character.
|
|
|
9637 * round[depth] == 2: Try using the upper-case character.
|
|
|
9638 */
|
|
|
9639 #ifdef FEAT_MBYTE
|
|
|
9640 if (has_mbyte)
|
|
|
9641 {
|
|
474
|
9642 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
9643 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
9644 }
|
|
|
9645 else
|
|
|
9646 #endif
|
|
|
9647 ulen = flen = 1;
|
|
|
9648 if (round[depth] == 1)
|
|
|
9649 {
|
|
|
9650 p = fword + fwordidx[depth];
|
|
|
9651 l = flen;
|
|
|
9652 }
|
|
|
9653 else
|
|
|
9654 {
|
|
|
9655 p = uword + uwordidx[depth];
|
|
|
9656 l = ulen;
|
|
|
9657 }
|
|
|
9658
|
|
|
9659 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
9660 {
|
|
|
9661 /* Perform a binary search in the list of accepted bytes. */
|
|
|
9662 len = byts[tryidx++];
|
|
|
9663 c = *p++;
|
|
|
9664 lo = tryidx;
|
|
|
9665 hi = tryidx + len - 1;
|
|
|
9666 while (lo < hi)
|
|
|
9667 {
|
|
|
9668 m = (lo + hi) / 2;
|
|
|
9669 if (byts[m] > c)
|
|
|
9670 hi = m - 1;
|
|
|
9671 else if (byts[m] < c)
|
|
|
9672 lo = m + 1;
|
|
|
9673 else
|
|
|
9674 {
|
|
|
9675 lo = hi = m;
|
|
|
9676 break;
|
|
|
9677 }
|
|
|
9678 }
|
|
|
9679
|
|
|
9680 /* Stop if there is no matching byte. */
|
|
|
9681 if (hi < lo || byts[lo] != c)
|
|
|
9682 break;
|
|
|
9683
|
|
|
9684 /* Continue at the child (if there is one). */
|
|
|
9685 tryidx = idxs[lo];
|
|
|
9686 }
|
|
|
9687
|
|
|
9688 if (l == 0)
|
|
|
9689 {
|
|
|
9690 /*
|
|
|
9691 * Found the matching char. Copy it to "kword" and go a
|
|
|
9692 * level deeper.
|
|
|
9693 */
|
|
|
9694 if (round[depth] == 1)
|
|
|
9695 {
|
|
|
9696 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
9697 flen);
|
|
|
9698 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
9699 }
|
|
|
9700 else
|
|
|
9701 {
|
|
|
9702 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
9703 ulen);
|
|
|
9704 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
9705 }
|
|
|
9706 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
9707 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
9708
|
|
|
9709 ++depth;
|
|
|
9710 arridx[depth] = tryidx;
|
|
|
9711 round[depth] = 0;
|
|
|
9712 }
|
|
|
9713 }
|
|
|
9714 }
|
|
|
9715
|
|
|
9716 /* Didn't find it: "cannot happen". */
|
|
|
9717 *kword = NUL;
|
|
|
9718 }
|
|
|
9719
|
|
|
9720 /*
|
|
344
|
9721 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
9722 * su->su_sga.
|
|
|
9723 */
|
|
|
9724 static void
|
|
|
9725 score_comp_sal(su)
|
|
|
9726 suginfo_T *su;
|
|
|
9727 {
|
|
|
9728 langp_T *lp;
|
|
|
9729 char_u badsound[MAXWLEN];
|
|
|
9730 int i;
|
|
|
9731 suggest_T *stp;
|
|
|
9732 suggest_T *sstp;
|
|
|
9733 int score;
|
|
|
9734
|
|
|
9735 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
9736 return;
|
|
|
9737
|
|
|
9738 /* Use the sound-folding of the first language that supports it. */
|
|
|
9739 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9740 lp->lp_slang != NULL; ++lp)
|
|
|
9741 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9742 {
|
|
|
9743 /* soundfold the bad word */
|
|
375
|
9744 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
9745
|
|
|
9746 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9747 {
|
|
|
9748 stp = &SUG(su->su_ga, i);
|
|
|
9749
|
|
351
|
9750 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
9751 * sound-a-like score. */
|
|
|
9752 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
9753 if (score < SCORE_MAXMAX)
|
|
|
9754 {
|
|
|
9755 /* Add the suggestion. */
|
|
|
9756 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
9757 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
9758 if (sstp->st_word != NULL)
|
|
|
9759 {
|
|
|
9760 sstp->st_score = score;
|
|
|
9761 sstp->st_altscore = 0;
|
|
|
9762 sstp->st_orglen = stp->st_orglen;
|
|
|
9763 ++su->su_sga.ga_len;
|
|
|
9764 }
|
|
|
9765 }
|
|
|
9766 }
|
|
|
9767 break;
|
|
|
9768 }
|
|
|
9769 }
|
|
|
9770
|
|
|
9771 /*
|
|
|
9772 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
9773 * They are intwined.
|
|
|
9774 */
|
|
|
9775 static void
|
|
|
9776 score_combine(su)
|
|
|
9777 suginfo_T *su;
|
|
|
9778 {
|
|
|
9779 int i;
|
|
|
9780 int j;
|
|
|
9781 garray_T ga;
|
|
|
9782 garray_T *gap;
|
|
|
9783 langp_T *lp;
|
|
|
9784 suggest_T *stp;
|
|
|
9785 char_u *p;
|
|
|
9786 char_u badsound[MAXWLEN];
|
|
|
9787 int round;
|
|
|
9788
|
|
|
9789 /* Add the alternate score to su_ga. */
|
|
|
9790 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9791 lp->lp_slang != NULL; ++lp)
|
|
|
9792 {
|
|
|
9793 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9794 {
|
|
|
9795 /* soundfold the bad word */
|
|
375
|
9796 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
9797
|
|
|
9798 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9799 {
|
|
|
9800 stp = &SUG(su->su_ga, i);
|
|
351
|
9801 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
9802 badsound);
|
|
344
|
9803 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
9804 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
9805 else
|
|
|
9806 stp->st_score = (stp->st_score * 3
|
|
|
9807 + stp->st_altscore) / 4;
|
|
|
9808 stp->st_salscore = FALSE;
|
|
|
9809 }
|
|
|
9810 break;
|
|
|
9811 }
|
|
|
9812 }
|
|
|
9813
|
|
|
9814 /* Add the alternate score to su_sga. */
|
|
|
9815 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
9816 {
|
|
|
9817 stp = &SUG(su->su_sga, i);
|
|
|
9818 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
9819 if (stp->st_score == SCORE_MAXMAX)
|
|
|
9820 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
9821 else
|
|
|
9822 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
9823 stp->st_salscore = TRUE;
|
|
|
9824 }
|
|
|
9825
|
|
|
9826 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
9827 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
9828 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
9829
|
|
|
9830 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
9831 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
9832 return;
|
|
|
9833
|
|
|
9834 stp = &SUG(ga, 0);
|
|
|
9835 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
9836 {
|
|
|
9837 /* round 1: get a suggestion from su_ga
|
|
|
9838 * round 2: get a suggestion from su_sga */
|
|
|
9839 for (round = 1; round <= 2; ++round)
|
|
|
9840 {
|
|
|
9841 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
9842 if (i < gap->ga_len)
|
|
|
9843 {
|
|
|
9844 /* Don't add a word if it's already there. */
|
|
|
9845 p = SUG(*gap, i).st_word;
|
|
|
9846 for (j = 0; j < ga.ga_len; ++j)
|
|
|
9847 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
9848 break;
|
|
|
9849 if (j == ga.ga_len)
|
|
|
9850 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
9851 else
|
|
|
9852 vim_free(p);
|
|
|
9853 }
|
|
|
9854 }
|
|
|
9855 }
|
|
|
9856
|
|
|
9857 ga_clear(&su->su_ga);
|
|
|
9858 ga_clear(&su->su_sga);
|
|
|
9859
|
|
|
9860 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
9861 if (ga.ga_len > su->su_maxcount)
|
|
|
9862 {
|
|
|
9863 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
9864 vim_free(stp[i].st_word);
|
|
|
9865 ga.ga_len = su->su_maxcount;
|
|
|
9866 }
|
|
|
9867
|
|
|
9868 su->su_ga = ga;
|
|
|
9869 }
|
|
|
9870
|
|
|
9871 /*
|
|
351
|
9872 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
9873 * badword.
|
|
|
9874 */
|
|
|
9875 static int
|
|
|
9876 stp_sal_score(stp, su, slang, badsound)
|
|
|
9877 suggest_T *stp;
|
|
|
9878 suginfo_T *su;
|
|
|
9879 slang_T *slang;
|
|
|
9880 char_u *badsound; /* sound-folded badword */
|
|
|
9881 {
|
|
|
9882 char_u *p;
|
|
|
9883 char_u badsound2[MAXWLEN];
|
|
|
9884 char_u fword[MAXWLEN];
|
|
|
9885 char_u goodsound[MAXWLEN];
|
|
|
9886
|
|
|
9887 if (stp->st_orglen <= su->su_badlen)
|
|
|
9888 p = badsound;
|
|
|
9889 else
|
|
|
9890 {
|
|
|
9891 /* soundfold the bad word with more characters following */
|
|
|
9892 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
9893
|
|
|
9894 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
9895 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
9896 * removing the space. Don't do it when the good word also contains a
|
|
|
9897 * space. */
|
|
|
9898 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
9899 && *skiptowhite(stp->st_word) == NUL)
|
|
|
9900 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
9901 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
9902
|
|
375
|
9903 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
9904 p = badsound2;
|
|
|
9905 }
|
|
|
9906
|
|
375
|
9907 /* Sound-fold the word and compute the score for the difference. */
|
|
|
9908 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
9909
|
|
|
9910 return soundalike_score(goodsound, p);
|
|
|
9911 }
|
|
|
9912
|
|
|
9913 /*
|
|
323
|
9914 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
9915 */
|
|
|
9916 static void
|
|
346
|
9917 suggest_try_soundalike(su)
|
|
323
|
9918 suginfo_T *su;
|
|
|
9919 {
|
|
|
9920 char_u salword[MAXWLEN];
|
|
|
9921 char_u tword[MAXWLEN];
|
|
|
9922 char_u tsalword[MAXWLEN];
|
|
324
|
9923 idx_T arridx[MAXWLEN];
|
|
323
|
9924 int curi[MAXWLEN];
|
|
|
9925 langp_T *lp;
|
|
|
9926 char_u *byts;
|
|
324
|
9927 idx_T *idxs;
|
|
323
|
9928 int depth;
|
|
|
9929 int c;
|
|
324
|
9930 idx_T n;
|
|
323
|
9931 int round;
|
|
|
9932 int flags;
|
|
344
|
9933 int sound_score;
|
|
483
|
9934 int local_score;
|
|
344
|
9935
|
|
|
9936 /* Do this for all languages that support sound folding. */
|
|
323
|
9937 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9938 lp->lp_slang != NULL; ++lp)
|
|
|
9939 {
|
|
|
9940 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9941 {
|
|
|
9942 /* soundfold the bad word */
|
|
375
|
9943 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
9944
|
|
|
9945 /*
|
|
|
9946 * Go through the whole tree, soundfold each word and compare.
|
|
|
9947 * round 1: use the case-folded tree.
|
|
|
9948 * round 2: use the keep-case tree.
|
|
|
9949 */
|
|
|
9950 for (round = 1; round <= 2; ++round)
|
|
|
9951 {
|
|
|
9952 if (round == 1)
|
|
|
9953 {
|
|
|
9954 byts = lp->lp_slang->sl_fbyts;
|
|
|
9955 idxs = lp->lp_slang->sl_fidxs;
|
|
|
9956 }
|
|
|
9957 else
|
|
|
9958 {
|
|
|
9959 byts = lp->lp_slang->sl_kbyts;
|
|
|
9960 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
9961 if (byts == NULL) /* no keep-case words */
|
|
|
9962 continue;
|
|
323
|
9963 }
|
|
|
9964
|
|
|
9965 depth = 0;
|
|
|
9966 arridx[0] = 0;
|
|
|
9967 curi[0] = 1;
|
|
|
9968 while (depth >= 0 && !got_int)
|
|
|
9969 {
|
|
|
9970 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
9971 {
|
|
323
|
9972 /* Done all bytes at this node, go up one level. */
|
|
|
9973 --depth;
|
|
351
|
9974 line_breakcheck();
|
|
|
9975 }
|
|
323
|
9976 else
|
|
|
9977 {
|
|
|
9978 /* Do one more byte at this node. */
|
|
|
9979 n = arridx[depth] + curi[depth];
|
|
|
9980 ++curi[depth];
|
|
|
9981 c = byts[n];
|
|
|
9982 if (c == 0)
|
|
|
9983 {
|
|
|
9984 /* End of word, deal with the word. */
|
|
324
|
9985 flags = (int)idxs[n];
|
|
323
|
9986 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
9987 {
|
|
|
9988 tword[depth] = NUL;
|
|
375
|
9989 /* Sound-fold. Only in keep-case tree need to
|
|
|
9990 * case-fold the word. */
|
|
|
9991 spell_soundfold(lp->lp_slang, tword,
|
|
|
9992 round == 1, tsalword);
|
|
323
|
9993
|
|
344
|
9994 /* Compute the edit distance between the
|
|
|
9995 * sound-a-like words. */
|
|
|
9996 sound_score = soundalike_score(salword,
|
|
|
9997 tsalword);
|
|
483
|
9998
|
|
|
9999 /* Add a penalty for words in another region. */
|
|
|
10000 if ((flags & WF_REGION) && (((unsigned)flags
|
|
|
10001 >> 16) & lp->lp_region) == 0)
|
|
|
10002 local_score = SCORE_REGION;
|
|
|
10003 else
|
|
|
10004 local_score = 0;
|
|
|
10005 sound_score += local_score;
|
|
|
10006
|
|
324
|
10007 if (sound_score < SCORE_MAXMAX)
|
|
323
|
10008 {
|
|
324
|
10009 char_u cword[MAXWLEN];
|
|
|
10010 char_u *p;
|
|
344
|
10011 int score;
|
|
324
|
10012
|
|
388
|
10013 flags |= su->su_badflags;
|
|
351
|
10014 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
10015 {
|
|
324
|
10016 /* Need to fix case according to
|
|
|
10017 * "flags". */
|
|
323
|
10018 make_case_word(tword, cword, flags);
|
|
324
|
10019 p = cword;
|
|
323
|
10020 }
|
|
|
10021 else
|
|
324
|
10022 p = tword;
|
|
|
10023
|
|
344
|
10024 if (sps_flags & SPS_DOUBLE)
|
|
|
10025 add_suggestion(su, &su->su_sga, p,
|
|
346
|
10026 su->su_badlen,
|
|
351
|
10027 sound_score, 0, FALSE);
|
|
344
|
10028 else
|
|
|
10029 {
|
|
|
10030 /* Compute the score. */
|
|
|
10031 score = spell_edit_score(
|
|
483
|
10032 su->su_badword, p)
|
|
|
10033 + local_score;
|
|
344
|
10034 if (sps_flags & SPS_BEST)
|
|
|
10035 /* give a bonus for the good word
|
|
|
10036 * sounding the same as the bad
|
|
|
10037 * word */
|
|
|
10038 add_suggestion(su, &su->su_ga, p,
|
|
346
|
10039 su->su_badlen,
|
|
344
|
10040 RESCORE(score, sound_score),
|
|
351
|
10041 sound_score, TRUE);
|
|
344
|
10042 else
|
|
|
10043 add_suggestion(su, &su->su_ga, p,
|
|
346
|
10044 su->su_badlen,
|
|
351
|
10045 score + sound_score, 0, FALSE);
|
|
344
|
10046 }
|
|
323
|
10047 }
|
|
|
10048 }
|
|
|
10049
|
|
|
10050 /* Skip over other NUL bytes. */
|
|
|
10051 while (byts[n + 1] == 0)
|
|
|
10052 {
|
|
|
10053 ++n;
|
|
|
10054 ++curi[depth];
|
|
|
10055 }
|
|
|
10056 }
|
|
|
10057 else
|
|
|
10058 {
|
|
|
10059 /* Normal char, go one level deeper. */
|
|
|
10060 tword[depth++] = c;
|
|
|
10061 arridx[depth] = idxs[n];
|
|
|
10062 curi[depth] = 1;
|
|
|
10063 }
|
|
|
10064 }
|
|
|
10065 }
|
|
|
10066 }
|
|
|
10067 }
|
|
|
10068 }
|
|
|
10069 }
|
|
|
10070
|
|
|
10071 /*
|
|
324
|
10072 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
10073 */
|
|
|
10074 static void
|
|
|
10075 make_case_word(fword, cword, flags)
|
|
|
10076 char_u *fword;
|
|
|
10077 char_u *cword;
|
|
|
10078 int flags;
|
|
|
10079 {
|
|
|
10080 if (flags & WF_ALLCAP)
|
|
|
10081 /* Make it all upper-case */
|
|
|
10082 allcap_copy(fword, cword);
|
|
|
10083 else if (flags & WF_ONECAP)
|
|
|
10084 /* Make the first letter upper-case */
|
|
324
|
10085 onecap_copy(fword, cword, TRUE);
|
|
323
|
10086 else
|
|
|
10087 /* Use goodword as-is. */
|
|
|
10088 STRCPY(cword, fword);
|
|
|
10089 }
|
|
|
10090
|
|
330
|
10091 /*
|
|
|
10092 * Use map string "map" for languages "lp".
|
|
|
10093 */
|
|
|
10094 static void
|
|
|
10095 set_map_str(lp, map)
|
|
|
10096 slang_T *lp;
|
|
|
10097 char_u *map;
|
|
|
10098 {
|
|
|
10099 char_u *p;
|
|
|
10100 int headc = 0;
|
|
|
10101 int c;
|
|
|
10102 int i;
|
|
|
10103
|
|
|
10104 if (*map == NUL)
|
|
|
10105 {
|
|
|
10106 lp->sl_has_map = FALSE;
|
|
|
10107 return;
|
|
|
10108 }
|
|
|
10109 lp->sl_has_map = TRUE;
|
|
|
10110
|
|
|
10111 /* Init the array and hash table empty. */
|
|
|
10112 for (i = 0; i < 256; ++i)
|
|
|
10113 lp->sl_map_array[i] = 0;
|
|
|
10114 #ifdef FEAT_MBYTE
|
|
|
10115 hash_init(&lp->sl_map_hash);
|
|
|
10116 #endif
|
|
|
10117
|
|
|
10118 /*
|
|
|
10119 * The similar characters are stored separated with slashes:
|
|
|
10120 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
10121 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
10122 */
|
|
|
10123 for (p = map; *p != NUL; )
|
|
|
10124 {
|
|
|
10125 #ifdef FEAT_MBYTE
|
|
474
|
10126 c = mb_cptr2char_adv(&p);
|
|
330
|
10127 #else
|
|
|
10128 c = *p++;
|
|
|
10129 #endif
|
|
|
10130 if (c == '/')
|
|
|
10131 headc = 0;
|
|
|
10132 else
|
|
|
10133 {
|
|
|
10134 if (headc == 0)
|
|
|
10135 headc = c;
|
|
|
10136
|
|
|
10137 #ifdef FEAT_MBYTE
|
|
|
10138 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
10139 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
10140 * a NUL. */
|
|
|
10141 if (c >= 256)
|
|
|
10142 {
|
|
|
10143 int cl = mb_char2len(c);
|
|
|
10144 int headcl = mb_char2len(headc);
|
|
|
10145 char_u *b;
|
|
|
10146 hash_T hash;
|
|
|
10147 hashitem_T *hi;
|
|
|
10148
|
|
|
10149 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
10150 if (b == NULL)
|
|
|
10151 return;
|
|
|
10152 mb_char2bytes(c, b);
|
|
|
10153 b[cl] = NUL;
|
|
|
10154 mb_char2bytes(headc, b + cl + 1);
|
|
|
10155 b[cl + 1 + headcl] = NUL;
|
|
|
10156 hash = hash_hash(b);
|
|
|
10157 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
10158 if (HASHITEM_EMPTY(hi))
|
|
|
10159 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
10160 else
|
|
|
10161 {
|
|
|
10162 /* This should have been checked when generating the .spl
|
|
|
10163 * file. */
|
|
|
10164 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
10165 vim_free(b);
|
|
|
10166 }
|
|
|
10167 }
|
|
|
10168 else
|
|
|
10169 #endif
|
|
|
10170 lp->sl_map_array[c] = headc;
|
|
|
10171 }
|
|
|
10172 }
|
|
|
10173 }
|
|
|
10174
|
|
323
|
10175 /*
|
|
|
10176 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
10177 * lines in the .aff file.
|
|
|
10178 */
|
|
|
10179 static int
|
|
|
10180 similar_chars(slang, c1, c2)
|
|
|
10181 slang_T *slang;
|
|
|
10182 int c1;
|
|
|
10183 int c2;
|
|
|
10184 {
|
|
330
|
10185 int m1, m2;
|
|
|
10186 #ifdef FEAT_MBYTE
|
|
|
10187 char_u buf[MB_MAXBYTES];
|
|
|
10188 hashitem_T *hi;
|
|
|
10189
|
|
|
10190 if (c1 >= 256)
|
|
|
10191 {
|
|
|
10192 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
10193 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
10194 if (HASHITEM_EMPTY(hi))
|
|
|
10195 m1 = 0;
|
|
|
10196 else
|
|
|
10197 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
10198 }
|
|
|
10199 else
|
|
|
10200 #endif
|
|
|
10201 m1 = slang->sl_map_array[c1];
|
|
|
10202 if (m1 == 0)
|
|
323
|
10203 return FALSE;
|
|
330
|
10204
|
|
|
10205
|
|
|
10206 #ifdef FEAT_MBYTE
|
|
|
10207 if (c2 >= 256)
|
|
|
10208 {
|
|
|
10209 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
10210 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
10211 if (HASHITEM_EMPTY(hi))
|
|
|
10212 m2 = 0;
|
|
|
10213 else
|
|
|
10214 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
10215 }
|
|
|
10216 else
|
|
|
10217 #endif
|
|
|
10218 m2 = slang->sl_map_array[c2];
|
|
|
10219
|
|
|
10220 return m1 == m2;
|
|
323
|
10221 }
|
|
|
10222
|
|
|
10223 /*
|
|
|
10224 * Add a suggestion to the list of suggestions.
|
|
|
10225 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
10226 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
10227 * with spell_edit_score().
|
|
|
10228 */
|
|
|
10229 static void
|
|
351
|
10230 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
10231 suginfo_T *su;
|
|
344
|
10232 garray_T *gap;
|
|
323
|
10233 char_u *goodword;
|
|
346
|
10234 int badlen; /* length of bad word used */
|
|
324
|
10235 int score;
|
|
351
|
10236 int altscore;
|
|
344
|
10237 int had_bonus; /* value for st_had_bonus */
|
|
323
|
10238 {
|
|
|
10239 suggest_T *stp;
|
|
|
10240 int i;
|
|
346
|
10241 char_u *p = NULL;
|
|
|
10242 int c = 0;
|
|
323
|
10243
|
|
|
10244 /* Check that the word wasn't banned. */
|
|
|
10245 if (was_banned(su, goodword))
|
|
|
10246 return;
|
|
|
10247
|
|
346
|
10248 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
10249 * Remove the common part from "goodword". */
|
|
|
10250 i = badlen - su->su_badlen;
|
|
|
10251 if (i > 0)
|
|
|
10252 {
|
|
|
10253 /* This assumes there was no case folding or it didn't change the
|
|
|
10254 * length... */
|
|
|
10255 p = goodword + STRLEN(goodword) - i;
|
|
|
10256 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
10257 {
|
|
|
10258 badlen = su->su_badlen;
|
|
|
10259 c = *p;
|
|
|
10260 *p = NUL;
|
|
|
10261 }
|
|
|
10262 else
|
|
|
10263 p = NULL;
|
|
|
10264 }
|
|
372
|
10265 else if (i < 0)
|
|
|
10266 {
|
|
|
10267 /* When replacing part of the word check that we actually change
|
|
|
10268 * something. For "the the" a suggestion can be replacing the first
|
|
|
10269 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
10270 if (badlen == (int)STRLEN(goodword)
|
|
372
|
10271 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
10272 return;
|
|
|
10273 }
|
|
|
10274
|
|
346
|
10275
|
|
323
|
10276 if (score <= su->su_maxscore)
|
|
|
10277 {
|
|
366
|
10278 /* Check if the word is already there. Also check the length that is
|
|
|
10279 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
10280 * "thes" -> "these". */
|
|
344
|
10281 stp = &SUG(*gap, 0);
|
|
|
10282 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
10283 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
10284 && stp[i].st_orglen == badlen)
|
|
323
|
10285 {
|
|
|
10286 /* Found it. Remember the lowest score. */
|
|
|
10287 if (stp[i].st_score > score)
|
|
324
|
10288 {
|
|
323
|
10289 stp[i].st_score = score;
|
|
376
|
10290 stp[i].st_altscore = altscore;
|
|
324
|
10291 stp[i].st_had_bonus = had_bonus;
|
|
|
10292 }
|
|
323
|
10293 break;
|
|
|
10294 }
|
|
|
10295
|
|
344
|
10296 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
10297 {
|
|
|
10298 /* Add a suggestion. */
|
|
344
|
10299 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
10300 stp->st_word = vim_strsave(goodword);
|
|
|
10301 if (stp->st_word != NULL)
|
|
|
10302 {
|
|
|
10303 stp->st_score = score;
|
|
351
|
10304 stp->st_altscore = altscore;
|
|
324
|
10305 stp->st_had_bonus = had_bonus;
|
|
346
|
10306 stp->st_orglen = badlen;
|
|
344
|
10307 ++gap->ga_len;
|
|
323
|
10308
|
|
|
10309 /* If we have too many suggestions now, sort the list and keep
|
|
|
10310 * the best suggestions. */
|
|
344
|
10311 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
10312 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
10313 SUG_CLEAN_COUNT(su));
|
|
323
|
10314 }
|
|
|
10315 }
|
|
|
10316 }
|
|
346
|
10317
|
|
|
10318 if (p != NULL)
|
|
|
10319 *p = c; /* restore "goodword" */
|
|
323
|
10320 }
|
|
|
10321
|
|
|
10322 /*
|
|
|
10323 * Add a word to be banned.
|
|
|
10324 */
|
|
|
10325 static void
|
|
|
10326 add_banned(su, word)
|
|
|
10327 suginfo_T *su;
|
|
|
10328 char_u *word;
|
|
|
10329 {
|
|
|
10330 char_u *s = vim_strsave(word);
|
|
|
10331 hash_T hash;
|
|
|
10332 hashitem_T *hi;
|
|
|
10333
|
|
|
10334 if (s != NULL)
|
|
|
10335 {
|
|
|
10336 hash = hash_hash(s);
|
|
|
10337 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
10338 if (HASHITEM_EMPTY(hi))
|
|
|
10339 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
10340 else
|
|
|
10341 vim_free(s);
|
|
323
|
10342 }
|
|
|
10343 }
|
|
|
10344
|
|
|
10345 /*
|
|
|
10346 * Return TRUE if a word appears in the list of banned words.
|
|
|
10347 */
|
|
|
10348 static int
|
|
|
10349 was_banned(su, word)
|
|
|
10350 suginfo_T *su;
|
|
|
10351 char_u *word;
|
|
|
10352 {
|
|
324
|
10353 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
10354
|
|
|
10355 return !HASHITEM_EMPTY(hi);
|
|
323
|
10356 }
|
|
|
10357
|
|
|
10358 /*
|
|
|
10359 * Free the banned words in "su".
|
|
|
10360 */
|
|
|
10361 static void
|
|
|
10362 free_banned(su)
|
|
|
10363 suginfo_T *su;
|
|
|
10364 {
|
|
|
10365 int todo;
|
|
|
10366 hashitem_T *hi;
|
|
|
10367
|
|
|
10368 todo = su->su_banned.ht_used;
|
|
|
10369 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
10370 {
|
|
|
10371 if (!HASHITEM_EMPTY(hi))
|
|
|
10372 {
|
|
|
10373 vim_free(hi->hi_key);
|
|
|
10374 --todo;
|
|
|
10375 }
|
|
|
10376 }
|
|
|
10377 hash_clear(&su->su_banned);
|
|
|
10378 }
|
|
|
10379
|
|
324
|
10380 /*
|
|
|
10381 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
10382 * thus only done for the final results.
|
|
|
10383 */
|
|
|
10384 static void
|
|
|
10385 rescore_suggestions(su)
|
|
|
10386 suginfo_T *su;
|
|
|
10387 {
|
|
|
10388 langp_T *lp;
|
|
|
10389 suggest_T *stp;
|
|
|
10390 char_u sal_badword[MAXWLEN];
|
|
|
10391 int i;
|
|
|
10392
|
|
|
10393 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10394 lp->lp_slang != NULL; ++lp)
|
|
|
10395 {
|
|
|
10396 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10397 {
|
|
|
10398 /* soundfold the bad word */
|
|
375
|
10399 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
10400
|
|
|
10401 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10402 {
|
|
344
|
10403 stp = &SUG(su->su_ga, i);
|
|
324
|
10404 if (!stp->st_had_bonus)
|
|
|
10405 {
|
|
351
|
10406 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
10407 lp->lp_slang, sal_badword);
|
|
|
10408 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
10409 stp->st_altscore = SCORE_BIG;
|
|
|
10410 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
10411 }
|
|
|
10412 }
|
|
|
10413 break;
|
|
|
10414 }
|
|
|
10415 }
|
|
|
10416 }
|
|
|
10417
|
|
323
|
10418 static int
|
|
|
10419 #ifdef __BORLANDC__
|
|
|
10420 _RTLENTRYF
|
|
|
10421 #endif
|
|
|
10422 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
10423
|
|
|
10424 /*
|
|
|
10425 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
10426 */
|
|
|
10427 static int
|
|
|
10428 #ifdef __BORLANDC__
|
|
|
10429 _RTLENTRYF
|
|
|
10430 #endif
|
|
|
10431 sug_compare(s1, s2)
|
|
|
10432 const void *s1;
|
|
|
10433 const void *s2;
|
|
|
10434 {
|
|
|
10435 suggest_T *p1 = (suggest_T *)s1;
|
|
|
10436 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
10437 int n = p1->st_score - p2->st_score;
|
|
|
10438
|
|
|
10439 if (n == 0)
|
|
|
10440 return p1->st_altscore - p2->st_altscore;
|
|
|
10441 return n;
|
|
323
|
10442 }
|
|
|
10443
|
|
|
10444 /*
|
|
|
10445 * Cleanup the suggestions:
|
|
|
10446 * - Sort on score.
|
|
|
10447 * - Remove words that won't be displayed.
|
|
344
|
10448 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
10449 */
|
|
344
|
10450 static int
|
|
|
10451 cleanup_suggestions(gap, maxscore, keep)
|
|
|
10452 garray_T *gap;
|
|
|
10453 int maxscore;
|
|
324
|
10454 int keep; /* nr of suggestions to keep */
|
|
323
|
10455 {
|
|
344
|
10456 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
10457 int i;
|
|
|
10458
|
|
|
10459 /* Sort the list. */
|
|
344
|
10460 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
10461
|
|
|
10462 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
10463 if (gap->ga_len > keep)
|
|
323
|
10464 {
|
|
344
|
10465 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
10466 vim_free(stp[i].st_word);
|
|
344
|
10467 gap->ga_len = keep;
|
|
|
10468 return stp[keep - 1].st_score;
|
|
323
|
10469 }
|
|
344
|
10470 return maxscore;
|
|
323
|
10471 }
|
|
|
10472
|
|
372
|
10473 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
10474 /*
|
|
|
10475 * Soundfold a string, for soundfold().
|
|
|
10476 * Result is in allocated memory, NULL for an error.
|
|
|
10477 */
|
|
|
10478 char_u *
|
|
|
10479 eval_soundfold(word)
|
|
|
10480 char_u *word;
|
|
|
10481 {
|
|
|
10482 langp_T *lp;
|
|
|
10483 char_u sound[MAXWLEN];
|
|
|
10484
|
|
|
10485 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
10486 /* Use the sound-folding of the first language that supports it. */
|
|
|
10487 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10488 lp->lp_slang != NULL; ++lp)
|
|
|
10489 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10490 {
|
|
|
10491 /* soundfold the word */
|
|
375
|
10492 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
10493 return vim_strsave(sound);
|
|
|
10494 }
|
|
|
10495
|
|
|
10496 /* No language with sound folding, return word as-is. */
|
|
|
10497 return vim_strsave(word);
|
|
|
10498 }
|
|
|
10499 #endif
|
|
|
10500
|
|
323
|
10501 /*
|
|
|
10502 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
485
|
10503 *
|
|
|
10504 * There are many ways to turn a word into a sound-a-like representation. The
|
|
|
10505 * oldest is Soundex (1918!). A nice overview can be found in "Approximate
|
|
|
10506 * swedish name matching - survey and test of different algorithms" by Klas
|
|
|
10507 * Erikson.
|
|
|
10508 *
|
|
|
10509 * We support two methods:
|
|
|
10510 * 1. SOFOFROM/SOFOTO do a simple character mapping.
|
|
|
10511 * 2. SAL items define a more advanced sound-folding (and much slower).
|
|
323
|
10512 */
|
|
|
10513 static void
|
|
375
|
10514 spell_soundfold(slang, inword, folded, res)
|
|
|
10515 slang_T *slang;
|
|
|
10516 char_u *inword;
|
|
|
10517 int folded; /* "inword" is already case-folded */
|
|
|
10518 char_u *res;
|
|
|
10519 {
|
|
|
10520 char_u fword[MAXWLEN];
|
|
|
10521 char_u *word;
|
|
|
10522
|
|
|
10523 if (slang->sl_sofo)
|
|
|
10524 /* SOFOFROM and SOFOTO used */
|
|
|
10525 spell_soundfold_sofo(slang, inword, res);
|
|
|
10526 else
|
|
|
10527 {
|
|
|
10528 /* SAL items used. Requires the word to be case-folded. */
|
|
|
10529 if (folded)
|
|
|
10530 word = inword;
|
|
|
10531 else
|
|
|
10532 {
|
|
|
10533 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
10534 word = fword;
|
|
|
10535 }
|
|
|
10536
|
|
|
10537 #ifdef FEAT_MBYTE
|
|
|
10538 if (has_mbyte)
|
|
|
10539 spell_soundfold_wsal(slang, word, res);
|
|
|
10540 else
|
|
|
10541 #endif
|
|
|
10542 spell_soundfold_sal(slang, word, res);
|
|
|
10543 }
|
|
|
10544 }
|
|
|
10545
|
|
|
10546 /*
|
|
|
10547 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
10548 * SOFOTO lines.
|
|
|
10549 */
|
|
|
10550 static void
|
|
|
10551 spell_soundfold_sofo(slang, inword, res)
|
|
|
10552 slang_T *slang;
|
|
|
10553 char_u *inword;
|
|
|
10554 char_u *res;
|
|
|
10555 {
|
|
|
10556 char_u *s;
|
|
|
10557 int ri = 0;
|
|
|
10558 int c;
|
|
|
10559
|
|
|
10560 #ifdef FEAT_MBYTE
|
|
|
10561 if (has_mbyte)
|
|
|
10562 {
|
|
|
10563 int prevc = 0;
|
|
|
10564 int *ip;
|
|
|
10565
|
|
|
10566 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
10567 * 255, sl_sal the rest. */
|
|
|
10568 for (s = inword; *s != NUL; )
|
|
|
10569 {
|
|
474
|
10570 c = mb_cptr2char_adv(&s);
|
|
375
|
10571 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
10572 c = ' ';
|
|
|
10573 else if (c < 256)
|
|
|
10574 c = slang->sl_sal_first[c];
|
|
|
10575 else
|
|
|
10576 {
|
|
|
10577 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
10578 if (ip == NULL) /* empty list, can't match */
|
|
|
10579 c = NUL;
|
|
|
10580 else
|
|
|
10581 for (;;) /* find "c" in the list */
|
|
|
10582 {
|
|
|
10583 if (*ip == 0) /* not found */
|
|
|
10584 {
|
|
|
10585 c = NUL;
|
|
|
10586 break;
|
|
|
10587 }
|
|
|
10588 if (*ip == c) /* match! */
|
|
|
10589 {
|
|
|
10590 c = ip[1];
|
|
|
10591 break;
|
|
|
10592 }
|
|
|
10593 ip += 2;
|
|
|
10594 }
|
|
|
10595 }
|
|
|
10596
|
|
|
10597 if (c != NUL && c != prevc)
|
|
|
10598 {
|
|
|
10599 ri += mb_char2bytes(c, res + ri);
|
|
|
10600 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
10601 break;
|
|
|
10602 prevc = c;
|
|
|
10603 }
|
|
|
10604 }
|
|
|
10605 }
|
|
|
10606 else
|
|
|
10607 #endif
|
|
|
10608 {
|
|
|
10609 /* The sl_sal_first[] table contains the translation. */
|
|
|
10610 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
10611 {
|
|
|
10612 if (vim_iswhite(c))
|
|
|
10613 c = ' ';
|
|
|
10614 else
|
|
|
10615 c = slang->sl_sal_first[c];
|
|
|
10616 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
10617 res[ri++] = c;
|
|
|
10618 }
|
|
|
10619 }
|
|
|
10620
|
|
|
10621 res[ri] = NUL;
|
|
|
10622 }
|
|
|
10623
|
|
|
10624 static void
|
|
|
10625 spell_soundfold_sal(slang, inword, res)
|
|
323
|
10626 slang_T *slang;
|
|
|
10627 char_u *inword;
|
|
|
10628 char_u *res;
|
|
|
10629 {
|
|
344
|
10630 salitem_T *smp;
|
|
323
|
10631 char_u word[MAXWLEN];
|
|
375
|
10632 char_u *s = inword;
|
|
323
|
10633 char_u *t;
|
|
344
|
10634 char_u *pf;
|
|
323
|
10635 int i, j, z;
|
|
344
|
10636 int reslen;
|
|
323
|
10637 int n, k = 0;
|
|
|
10638 int z0;
|
|
|
10639 int k0;
|
|
|
10640 int n0;
|
|
|
10641 int c;
|
|
|
10642 int pri;
|
|
|
10643 int p0 = -333;
|
|
|
10644 int c0;
|
|
|
10645
|
|
324
|
10646 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
10647 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
10648 if (slang->sl_rem_accents)
|
|
|
10649 {
|
|
|
10650 t = word;
|
|
375
|
10651 while (*s != NUL)
|
|
323
|
10652 {
|
|
324
|
10653 if (vim_iswhite(*s))
|
|
344
|
10654 {
|
|
|
10655 *t++ = ' ';
|
|
|
10656 s = skipwhite(s);
|
|
|
10657 }
|
|
323
|
10658 else
|
|
|
10659 {
|
|
376
|
10660 if (spell_iswordp_nmw(s))
|
|
323
|
10661 *t++ = *s;
|
|
|
10662 ++s;
|
|
|
10663 }
|
|
|
10664 }
|
|
|
10665 *t = NUL;
|
|
|
10666 }
|
|
|
10667 else
|
|
375
|
10668 STRCPY(word, s);
|
|
323
|
10669
|
|
344
|
10670 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
10671
|
|
|
10672 /*
|
|
|
10673 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
10674 * Changed to keep spaces.
|
|
323
|
10675 */
|
|
344
|
10676 i = reslen = z = 0;
|
|
323
|
10677 while ((c = word[i]) != NUL)
|
|
|
10678 {
|
|
344
|
10679 /* Start with the first rule that has the character in the word. */
|
|
323
|
10680 n = slang->sl_sal_first[c];
|
|
|
10681 z0 = 0;
|
|
|
10682
|
|
|
10683 if (n >= 0)
|
|
|
10684 {
|
|
|
10685 /* check all rules for the same letter */
|
|
344
|
10686 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
10687 {
|
|
344
|
10688 /* Quickly skip entries that don't match the word. Most
|
|
|
10689 * entries are less then three chars, optimize for that. */
|
|
|
10690 k = smp[n].sm_leadlen;
|
|
|
10691 if (k > 1)
|
|
323
|
10692 {
|
|
344
|
10693 if (word[i + 1] != s[1])
|
|
|
10694 continue;
|
|
|
10695 if (k > 2)
|
|
|
10696 {
|
|
|
10697 for (j = 2; j < k; ++j)
|
|
|
10698 if (word[i + j] != s[j])
|
|
|
10699 break;
|
|
|
10700 if (j < k)
|
|
|
10701 continue;
|
|
|
10702 }
|
|
323
|
10703 }
|
|
|
10704
|
|
375
|
10705 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
10706 {
|
|
375
|
10707 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
10708 while (*pf != NUL && *pf != word[i + k])
|
|
|
10709 ++pf;
|
|
|
10710 if (*pf == NUL)
|
|
|
10711 continue;
|
|
|
10712 ++k;
|
|
323
|
10713 }
|
|
344
|
10714 s = smp[n].sm_rules;
|
|
|
10715 pri = 5; /* default priority */
|
|
323
|
10716
|
|
|
10717 p0 = *s;
|
|
|
10718 k0 = k;
|
|
|
10719 while (*s == '-' && k > 1)
|
|
|
10720 {
|
|
|
10721 k--;
|
|
|
10722 s++;
|
|
|
10723 }
|
|
|
10724 if (*s == '<')
|
|
|
10725 s++;
|
|
344
|
10726 if (VIM_ISDIGIT(*s))
|
|
323
|
10727 {
|
|
|
10728 /* determine priority */
|
|
|
10729 pri = *s - '0';
|
|
|
10730 s++;
|
|
|
10731 }
|
|
|
10732 if (*s == '^' && *(s + 1) == '^')
|
|
|
10733 s++;
|
|
|
10734
|
|
|
10735 if (*s == NUL
|
|
|
10736 || (*s == '^'
|
|
324
|
10737 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
10738 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
10739 && (*(s + 1) != '$'
|
|
376
|
10740 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
10741 || (*s == '$' && i > 0
|
|
376
|
10742 && spell_iswordp(word + i - 1, curbuf)
|
|
|
10743 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
10744 {
|
|
|
10745 /* search for followup rules, if: */
|
|
|
10746 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
10747 c0 = word[i + k - 1];
|
|
|
10748 n0 = slang->sl_sal_first[c0];
|
|
|
10749
|
|
|
10750 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
10751 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
10752 {
|
|
|
10753 /* test follow-up rule for "word[i + k]" */
|
|
344
|
10754 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
10755 {
|
|
344
|
10756 /* Quickly skip entries that don't match the word.
|
|
|
10757 * */
|
|
|
10758 k0 = smp[n0].sm_leadlen;
|
|
|
10759 if (k0 > 1)
|
|
323
|
10760 {
|
|
344
|
10761 if (word[i + k] != s[1])
|
|
|
10762 continue;
|
|
|
10763 if (k0 > 2)
|
|
|
10764 {
|
|
|
10765 pf = word + i + k + 1;
|
|
|
10766 for (j = 2; j < k0; ++j)
|
|
|
10767 if (*pf++ != s[j])
|
|
|
10768 break;
|
|
|
10769 if (j < k0)
|
|
|
10770 continue;
|
|
|
10771 }
|
|
323
|
10772 }
|
|
344
|
10773 k0 += k - 1;
|
|
|
10774
|
|
375
|
10775 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
10776 {
|
|
344
|
10777 /* Check for match with one of the chars in
|
|
375
|
10778 * "sm_oneof". */
|
|
344
|
10779 while (*pf != NUL && *pf != word[i + k0])
|
|
|
10780 ++pf;
|
|
|
10781 if (*pf == NUL)
|
|
|
10782 continue;
|
|
|
10783 ++k0;
|
|
323
|
10784 }
|
|
344
|
10785
|
|
|
10786 p0 = 5;
|
|
|
10787 s = smp[n0].sm_rules;
|
|
323
|
10788 while (*s == '-')
|
|
|
10789 {
|
|
344
|
10790 /* "k0" gets NOT reduced because
|
|
|
10791 * "if (k0 == k)" */
|
|
323
|
10792 s++;
|
|
|
10793 }
|
|
|
10794 if (*s == '<')
|
|
|
10795 s++;
|
|
344
|
10796 if (VIM_ISDIGIT(*s))
|
|
323
|
10797 {
|
|
|
10798 p0 = *s - '0';
|
|
|
10799 s++;
|
|
|
10800 }
|
|
|
10801
|
|
|
10802 if (*s == NUL
|
|
|
10803 /* *s == '^' cuts */
|
|
|
10804 || (*s == '$'
|
|
376
|
10805 && !spell_iswordp(word + i + k0,
|
|
|
10806 curbuf)))
|
|
323
|
10807 {
|
|
|
10808 if (k0 == k)
|
|
|
10809 /* this is just a piece of the string */
|
|
|
10810 continue;
|
|
|
10811
|
|
|
10812 if (p0 < pri)
|
|
|
10813 /* priority too low */
|
|
|
10814 continue;
|
|
|
10815 /* rule fits; stop search */
|
|
|
10816 break;
|
|
|
10817 }
|
|
|
10818 }
|
|
|
10819
|
|
344
|
10820 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
10821 continue;
|
|
|
10822 }
|
|
|
10823
|
|
|
10824 /* replace string */
|
|
344
|
10825 s = smp[n].sm_to;
|
|
389
|
10826 if (s == NULL)
|
|
|
10827 s = (char_u *)"";
|
|
344
|
10828 pf = smp[n].sm_rules;
|
|
|
10829 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
10830 if (p0 == 1 && z == 0)
|
|
|
10831 {
|
|
|
10832 /* rule with '<' is used */
|
|
344
|
10833 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
10834 || res[reslen - 1] == *s))
|
|
|
10835 reslen--;
|
|
323
|
10836 z0 = 1;
|
|
|
10837 z = 1;
|
|
|
10838 k0 = 0;
|
|
372
|
10839 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
10840 {
|
|
|
10841 word[i + k0] = *s;
|
|
|
10842 k0++;
|
|
|
10843 s++;
|
|
|
10844 }
|
|
|
10845 if (k > k0)
|
|
|
10846 mch_memmove(word + i + k0, word + i + k,
|
|
|
10847 STRLEN(word + i + k) + 1);
|
|
|
10848
|
|
|
10849 /* new "actual letter" */
|
|
|
10850 c = word[i];
|
|
|
10851 }
|
|
|
10852 else
|
|
|
10853 {
|
|
|
10854 /* no '<' rule used */
|
|
|
10855 i += k - 1;
|
|
|
10856 z = 0;
|
|
344
|
10857 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
10858 {
|
|
344
|
10859 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
10860 res[reslen++] = *s;
|
|
323
|
10861 s++;
|
|
|
10862 }
|
|
|
10863 /* new "actual letter" */
|
|
|
10864 c = *s;
|
|
344
|
10865 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
10866 {
|
|
|
10867 if (c != NUL)
|
|
372
|
10868 res[reslen++] = c;
|
|
323
|
10869 mch_memmove(word, word + i + 1,
|
|
|
10870 STRLEN(word + i + 1) + 1);
|
|
|
10871 i = 0;
|
|
|
10872 z0 = 1;
|
|
|
10873 }
|
|
|
10874 }
|
|
|
10875 break;
|
|
|
10876 }
|
|
|
10877 }
|
|
|
10878 }
|
|
324
|
10879 else if (vim_iswhite(c))
|
|
|
10880 {
|
|
|
10881 c = ' ';
|
|
|
10882 k = 1;
|
|
|
10883 }
|
|
323
|
10884
|
|
|
10885 if (z0 == 0)
|
|
|
10886 {
|
|
344
|
10887 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
10888 && (!slang->sl_collapse || reslen == 0
|
|
|
10889 || res[reslen - 1] != c))
|
|
323
|
10890 /* condense only double letters */
|
|
372
|
10891 res[reslen++] = c;
|
|
323
|
10892
|
|
|
10893 i++;
|
|
|
10894 z = 0;
|
|
|
10895 k = 0;
|
|
|
10896 }
|
|
|
10897 }
|
|
|
10898
|
|
344
|
10899 res[reslen] = NUL;
|
|
323
|
10900 }
|
|
|
10901
|
|
372
|
10902 #ifdef FEAT_MBYTE
|
|
|
10903 /*
|
|
|
10904 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
10905 * Multi-byte version of spell_soundfold().
|
|
|
10906 */
|
|
|
10907 static void
|
|
375
|
10908 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
10909 slang_T *slang;
|
|
|
10910 char_u *inword;
|
|
|
10911 char_u *res;
|
|
|
10912 {
|
|
375
|
10913 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
10914 int word[MAXWLEN];
|
|
|
10915 int wres[MAXWLEN];
|
|
|
10916 int l;
|
|
|
10917 char_u *s;
|
|
|
10918 int *ws;
|
|
|
10919 char_u *t;
|
|
|
10920 int *pf;
|
|
|
10921 int i, j, z;
|
|
|
10922 int reslen;
|
|
|
10923 int n, k = 0;
|
|
|
10924 int z0;
|
|
|
10925 int k0;
|
|
|
10926 int n0;
|
|
|
10927 int c;
|
|
|
10928 int pri;
|
|
|
10929 int p0 = -333;
|
|
|
10930 int c0;
|
|
|
10931 int did_white = FALSE;
|
|
|
10932
|
|
|
10933 /*
|
|
|
10934 * Convert the multi-byte string to a wide-character string.
|
|
|
10935 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
10936 * But keep white space.
|
|
|
10937 */
|
|
|
10938 n = 0;
|
|
|
10939 for (s = inword; *s != NUL; )
|
|
|
10940 {
|
|
|
10941 t = s;
|
|
474
|
10942 c = mb_cptr2char_adv(&s);
|
|
372
|
10943 if (slang->sl_rem_accents)
|
|
|
10944 {
|
|
|
10945 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
10946 {
|
|
|
10947 if (did_white)
|
|
|
10948 continue;
|
|
|
10949 c = ' ';
|
|
|
10950 did_white = TRUE;
|
|
|
10951 }
|
|
|
10952 else
|
|
|
10953 {
|
|
|
10954 did_white = FALSE;
|
|
376
|
10955 if (!spell_iswordp_nmw(t))
|
|
372
|
10956 continue;
|
|
|
10957 }
|
|
|
10958 }
|
|
|
10959 word[n++] = c;
|
|
|
10960 }
|
|
|
10961 word[n] = NUL;
|
|
|
10962
|
|
|
10963 /*
|
|
|
10964 * This comes from Aspell phonet.cpp.
|
|
|
10965 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
10966 * Changed to keep spaces.
|
|
|
10967 */
|
|
|
10968 i = reslen = z = 0;
|
|
|
10969 while ((c = word[i]) != NUL)
|
|
|
10970 {
|
|
|
10971 /* Start with the first rule that has the character in the word. */
|
|
|
10972 n = slang->sl_sal_first[c & 0xff];
|
|
|
10973 z0 = 0;
|
|
|
10974
|
|
|
10975 if (n >= 0)
|
|
|
10976 {
|
|
375
|
10977 /* check all rules for the same index byte */
|
|
372
|
10978 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
10979 {
|
|
|
10980 /* Quickly skip entries that don't match the word. Most
|
|
|
10981 * entries are less then three chars, optimize for that. */
|
|
375
|
10982 if (c != ws[0])
|
|
|
10983 continue;
|
|
372
|
10984 k = smp[n].sm_leadlen;
|
|
|
10985 if (k > 1)
|
|
|
10986 {
|
|
|
10987 if (word[i + 1] != ws[1])
|
|
|
10988 continue;
|
|
|
10989 if (k > 2)
|
|
|
10990 {
|
|
|
10991 for (j = 2; j < k; ++j)
|
|
|
10992 if (word[i + j] != ws[j])
|
|
|
10993 break;
|
|
|
10994 if (j < k)
|
|
|
10995 continue;
|
|
|
10996 }
|
|
|
10997 }
|
|
|
10998
|
|
375
|
10999 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
11000 {
|
|
375
|
11001 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
11002 while (*pf != NUL && *pf != word[i + k])
|
|
|
11003 ++pf;
|
|
|
11004 if (*pf == NUL)
|
|
|
11005 continue;
|
|
|
11006 ++k;
|
|
|
11007 }
|
|
|
11008 s = smp[n].sm_rules;
|
|
|
11009 pri = 5; /* default priority */
|
|
|
11010
|
|
|
11011 p0 = *s;
|
|
|
11012 k0 = k;
|
|
|
11013 while (*s == '-' && k > 1)
|
|
|
11014 {
|
|
|
11015 k--;
|
|
|
11016 s++;
|
|
|
11017 }
|
|
|
11018 if (*s == '<')
|
|
|
11019 s++;
|
|
|
11020 if (VIM_ISDIGIT(*s))
|
|
|
11021 {
|
|
|
11022 /* determine priority */
|
|
|
11023 pri = *s - '0';
|
|
|
11024 s++;
|
|
|
11025 }
|
|
|
11026 if (*s == '^' && *(s + 1) == '^')
|
|
|
11027 s++;
|
|
|
11028
|
|
|
11029 if (*s == NUL
|
|
|
11030 || (*s == '^'
|
|
|
11031 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
11032 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
11033 && (*(s + 1) != '$'
|
|
376
|
11034 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
11035 || (*s == '$' && i > 0
|
|
376
|
11036 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
11037 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
11038 {
|
|
|
11039 /* search for followup rules, if: */
|
|
|
11040 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
11041 c0 = word[i + k - 1];
|
|
|
11042 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
11043
|
|
|
11044 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
11045 && p0 != '-' && word[i + k] != NUL)
|
|
|
11046 {
|
|
375
|
11047 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
11048 * all entries with the same index byte. */
|
|
372
|
11049 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
11050 == (c0 & 0xff); ++n0)
|
|
|
11051 {
|
|
|
11052 /* Quickly skip entries that don't match the word.
|
|
375
|
11053 */
|
|
|
11054 if (c0 != ws[0])
|
|
|
11055 continue;
|
|
372
|
11056 k0 = smp[n0].sm_leadlen;
|
|
|
11057 if (k0 > 1)
|
|
|
11058 {
|
|
|
11059 if (word[i + k] != ws[1])
|
|
|
11060 continue;
|
|
|
11061 if (k0 > 2)
|
|
|
11062 {
|
|
|
11063 pf = word + i + k + 1;
|
|
|
11064 for (j = 2; j < k0; ++j)
|
|
|
11065 if (*pf++ != ws[j])
|
|
|
11066 break;
|
|
|
11067 if (j < k0)
|
|
|
11068 continue;
|
|
|
11069 }
|
|
|
11070 }
|
|
|
11071 k0 += k - 1;
|
|
|
11072
|
|
375
|
11073 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
11074 {
|
|
|
11075 /* Check for match with one of the chars in
|
|
375
|
11076 * "sm_oneof". */
|
|
372
|
11077 while (*pf != NUL && *pf != word[i + k0])
|
|
|
11078 ++pf;
|
|
|
11079 if (*pf == NUL)
|
|
|
11080 continue;
|
|
|
11081 ++k0;
|
|
|
11082 }
|
|
|
11083
|
|
|
11084 p0 = 5;
|
|
|
11085 s = smp[n0].sm_rules;
|
|
|
11086 while (*s == '-')
|
|
|
11087 {
|
|
|
11088 /* "k0" gets NOT reduced because
|
|
|
11089 * "if (k0 == k)" */
|
|
|
11090 s++;
|
|
|
11091 }
|
|
|
11092 if (*s == '<')
|
|
|
11093 s++;
|
|
|
11094 if (VIM_ISDIGIT(*s))
|
|
|
11095 {
|
|
|
11096 p0 = *s - '0';
|
|
|
11097 s++;
|
|
|
11098 }
|
|
|
11099
|
|
|
11100 if (*s == NUL
|
|
|
11101 /* *s == '^' cuts */
|
|
|
11102 || (*s == '$'
|
|
376
|
11103 && !spell_iswordp_w(word + i + k0,
|
|
|
11104 curbuf)))
|
|
372
|
11105 {
|
|
|
11106 if (k0 == k)
|
|
|
11107 /* this is just a piece of the string */
|
|
|
11108 continue;
|
|
|
11109
|
|
|
11110 if (p0 < pri)
|
|
|
11111 /* priority too low */
|
|
|
11112 continue;
|
|
|
11113 /* rule fits; stop search */
|
|
|
11114 break;
|
|
|
11115 }
|
|
|
11116 }
|
|
|
11117
|
|
|
11118 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
11119 == (c0 & 0xff))
|
|
|
11120 continue;
|
|
|
11121 }
|
|
|
11122
|
|
|
11123 /* replace string */
|
|
|
11124 ws = smp[n].sm_to_w;
|
|
|
11125 s = smp[n].sm_rules;
|
|
|
11126 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
11127 if (p0 == 1 && z == 0)
|
|
|
11128 {
|
|
|
11129 /* rule with '<' is used */
|
|
389
|
11130 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
11131 && (wres[reslen - 1] == c
|
|
372
|
11132 || wres[reslen - 1] == *ws))
|
|
|
11133 reslen--;
|
|
|
11134 z0 = 1;
|
|
|
11135 z = 1;
|
|
|
11136 k0 = 0;
|
|
389
|
11137 if (ws != NULL)
|
|
|
11138 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
11139 {
|
|
|
11140 word[i + k0] = *ws;
|
|
|
11141 k0++;
|
|
|
11142 ws++;
|
|
|
11143 }
|
|
372
|
11144 if (k > k0)
|
|
|
11145 mch_memmove(word + i + k0, word + i + k,
|
|
|
11146 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
11147
|
|
|
11148 /* new "actual letter" */
|
|
|
11149 c = word[i];
|
|
|
11150 }
|
|
|
11151 else
|
|
|
11152 {
|
|
|
11153 /* no '<' rule used */
|
|
|
11154 i += k - 1;
|
|
|
11155 z = 0;
|
|
389
|
11156 if (ws != NULL)
|
|
|
11157 while (*ws != NUL && ws[1] != NUL
|
|
|
11158 && reslen < MAXWLEN)
|
|
|
11159 {
|
|
|
11160 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
11161 wres[reslen++] = *ws;
|
|
|
11162 ws++;
|
|
|
11163 }
|
|
372
|
11164 /* new "actual letter" */
|
|
389
|
11165 if (ws == NULL)
|
|
|
11166 c = NUL;
|
|
|
11167 else
|
|
|
11168 c = *ws;
|
|
372
|
11169 if (strstr((char *)s, "^^") != NULL)
|
|
|
11170 {
|
|
|
11171 if (c != NUL)
|
|
|
11172 wres[reslen++] = c;
|
|
|
11173 mch_memmove(word, word + i + 1,
|
|
|
11174 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
11175 i = 0;
|
|
|
11176 z0 = 1;
|
|
|
11177 }
|
|
|
11178 }
|
|
|
11179 break;
|
|
|
11180 }
|
|
|
11181 }
|
|
|
11182 }
|
|
|
11183 else if (vim_iswhite(c))
|
|
|
11184 {
|
|
|
11185 c = ' ';
|
|
|
11186 k = 1;
|
|
|
11187 }
|
|
|
11188
|
|
|
11189 if (z0 == 0)
|
|
|
11190 {
|
|
|
11191 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
11192 && (!slang->sl_collapse || reslen == 0
|
|
|
11193 || wres[reslen - 1] != c))
|
|
|
11194 /* condense only double letters */
|
|
|
11195 wres[reslen++] = c;
|
|
|
11196
|
|
|
11197 i++;
|
|
|
11198 z = 0;
|
|
|
11199 k = 0;
|
|
|
11200 }
|
|
|
11201 }
|
|
|
11202
|
|
|
11203 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
11204 l = 0;
|
|
|
11205 for (n = 0; n < reslen; ++n)
|
|
|
11206 {
|
|
|
11207 l += mb_char2bytes(wres[n], res + l);
|
|
|
11208 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
11209 break;
|
|
|
11210 }
|
|
|
11211 res[l] = NUL;
|
|
|
11212 }
|
|
|
11213 #endif
|
|
|
11214
|
|
324
|
11215 /*
|
|
344
|
11216 * Compute a score for two sound-a-like words.
|
|
|
11217 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
11218 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
11219 * avoiding checks that will not be possible.
|
|
|
11220 */
|
|
|
11221 static int
|
|
351
|
11222 soundalike_score(goodstart, badstart)
|
|
|
11223 char_u *goodstart; /* sound-folded good word */
|
|
|
11224 char_u *badstart; /* sound-folded bad word */
|
|
344
|
11225 {
|
|
351
|
11226 char_u *goodsound = goodstart;
|
|
|
11227 char_u *badsound = badstart;
|
|
|
11228 int goodlen;
|
|
|
11229 int badlen;
|
|
344
|
11230 int n;
|
|
|
11231 char_u *pl, *ps;
|
|
|
11232 char_u *pl2, *ps2;
|
|
351
|
11233 int score = 0;
|
|
|
11234
|
|
|
11235 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
11236 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
11237 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
11238 {
|
|
|
11239 score = SCORE_DEL / 2;
|
|
|
11240 if (*badsound == '*')
|
|
|
11241 ++badsound;
|
|
|
11242 else
|
|
|
11243 ++goodsound;
|
|
|
11244 }
|
|
|
11245
|
|
|
11246 goodlen = STRLEN(goodsound);
|
|
|
11247 badlen = STRLEN(badsound);
|
|
344
|
11248
|
|
|
11249 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
11250 * changes. */
|
|
|
11251 n = goodlen - badlen;
|
|
|
11252 if (n < -2 || n > 2)
|
|
|
11253 return SCORE_MAXMAX;
|
|
|
11254
|
|
|
11255 if (n > 0)
|
|
|
11256 {
|
|
351
|
11257 pl = goodsound; /* goodsound is longest */
|
|
344
|
11258 ps = badsound;
|
|
|
11259 }
|
|
|
11260 else
|
|
|
11261 {
|
|
351
|
11262 pl = badsound; /* badsound is longest */
|
|
344
|
11263 ps = goodsound;
|
|
|
11264 }
|
|
|
11265
|
|
|
11266 /* Skip over the identical part. */
|
|
|
11267 while (*pl == *ps && *pl != NUL)
|
|
|
11268 {
|
|
|
11269 ++pl;
|
|
|
11270 ++ps;
|
|
|
11271 }
|
|
|
11272
|
|
|
11273 switch (n)
|
|
|
11274 {
|
|
|
11275 case -2:
|
|
|
11276 case 2:
|
|
|
11277 /*
|
|
|
11278 * Must delete two characters from "pl".
|
|
|
11279 */
|
|
|
11280 ++pl; /* first delete */
|
|
|
11281 while (*pl == *ps)
|
|
|
11282 {
|
|
|
11283 ++pl;
|
|
|
11284 ++ps;
|
|
|
11285 }
|
|
|
11286 /* strings must be equal after second delete */
|
|
|
11287 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
11288 return score + SCORE_DEL * 2;
|
|
344
|
11289
|
|
|
11290 /* Failed to compare. */
|
|
|
11291 break;
|
|
|
11292
|
|
|
11293 case -1:
|
|
|
11294 case 1:
|
|
|
11295 /*
|
|
|
11296 * Minimal one delete from "pl" required.
|
|
|
11297 */
|
|
|
11298
|
|
|
11299 /* 1: delete */
|
|
|
11300 pl2 = pl + 1;
|
|
|
11301 ps2 = ps;
|
|
|
11302 while (*pl2 == *ps2)
|
|
|
11303 {
|
|
|
11304 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11305 return score + SCORE_DEL;
|
|
344
|
11306 ++pl2;
|
|
|
11307 ++ps2;
|
|
|
11308 }
|
|
|
11309
|
|
|
11310 /* 2: delete then swap, then rest must be equal */
|
|
|
11311 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11312 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11313 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
11314
|
|
|
11315 /* 3: delete then substitute, then the rest must be equal */
|
|
|
11316 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11317 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
11318
|
|
|
11319 /* 4: first swap then delete */
|
|
|
11320 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
11321 {
|
|
|
11322 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
11323 ps2 = ps + 2;
|
|
|
11324 while (*pl2 == *ps2)
|
|
|
11325 {
|
|
|
11326 ++pl2;
|
|
|
11327 ++ps2;
|
|
|
11328 }
|
|
|
11329 /* delete a char and then strings must be equal */
|
|
|
11330 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11331 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
11332 }
|
|
|
11333
|
|
|
11334 /* 5: first substitute then delete */
|
|
|
11335 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
11336 ps2 = ps + 1;
|
|
|
11337 while (*pl2 == *ps2)
|
|
|
11338 {
|
|
|
11339 ++pl2;
|
|
|
11340 ++ps2;
|
|
|
11341 }
|
|
|
11342 /* delete a char and then strings must be equal */
|
|
|
11343 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11344 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
11345
|
|
|
11346 /* Failed to compare. */
|
|
|
11347 break;
|
|
|
11348
|
|
|
11349 case 0:
|
|
|
11350 /*
|
|
|
11351 * Lenghts are equal, thus changes must result in same length: An
|
|
|
11352 * insert is only possible in combination with a delete.
|
|
|
11353 * 1: check if for identical strings
|
|
|
11354 */
|
|
|
11355 if (*pl == NUL)
|
|
351
|
11356 return score;
|
|
344
|
11357
|
|
|
11358 /* 2: swap */
|
|
|
11359 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
11360 {
|
|
|
11361 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
11362 ps2 = ps + 2;
|
|
|
11363 while (*pl2 == *ps2)
|
|
|
11364 {
|
|
|
11365 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11366 return score + SCORE_SWAP;
|
|
344
|
11367 ++pl2;
|
|
|
11368 ++ps2;
|
|
|
11369 }
|
|
|
11370 /* 3: swap and swap again */
|
|
|
11371 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11372 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11373 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
11374
|
|
|
11375 /* 4: swap and substitute */
|
|
|
11376 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11377 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
11378 }
|
|
|
11379
|
|
|
11380 /* 5: substitute */
|
|
|
11381 pl2 = pl + 1;
|
|
|
11382 ps2 = ps + 1;
|
|
|
11383 while (*pl2 == *ps2)
|
|
|
11384 {
|
|
|
11385 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11386 return score + SCORE_SUBST;
|
|
344
|
11387 ++pl2;
|
|
|
11388 ++ps2;
|
|
|
11389 }
|
|
|
11390
|
|
|
11391 /* 6: substitute and swap */
|
|
|
11392 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11393 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11394 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
11395
|
|
|
11396 /* 7: substitute and substitute */
|
|
|
11397 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11398 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
11399
|
|
|
11400 /* 8: insert then delete */
|
|
|
11401 pl2 = pl;
|
|
|
11402 ps2 = ps + 1;
|
|
|
11403 while (*pl2 == *ps2)
|
|
|
11404 {
|
|
|
11405 ++pl2;
|
|
|
11406 ++ps2;
|
|
|
11407 }
|
|
|
11408 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11409 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
11410
|
|
|
11411 /* 9: delete then insert */
|
|
|
11412 pl2 = pl + 1;
|
|
|
11413 ps2 = ps;
|
|
|
11414 while (*pl2 == *ps2)
|
|
|
11415 {
|
|
|
11416 ++pl2;
|
|
|
11417 ++ps2;
|
|
|
11418 }
|
|
|
11419 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
11420 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
11421
|
|
|
11422 /* Failed to compare. */
|
|
|
11423 break;
|
|
|
11424 }
|
|
|
11425
|
|
|
11426 return SCORE_MAXMAX;
|
|
|
11427 }
|
|
324
|
11428
|
|
323
|
11429 /*
|
|
|
11430 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
11431 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
11432 *
|
|
485
|
11433 * The algorithm is described by Du and Chang, 1992.
|
|
|
11434 * The implementation of the algorithm comes from Aspell editdist.cpp,
|
|
|
11435 * edit_distance(). It has been converted from C++ to C and modified to
|
|
|
11436 * support multi-byte characters.
|
|
323
|
11437 */
|
|
|
11438 static int
|
|
|
11439 spell_edit_score(badword, goodword)
|
|
|
11440 char_u *badword;
|
|
|
11441 char_u *goodword;
|
|
|
11442 {
|
|
|
11443 int *cnt;
|
|
372
|
11444 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
11445 int j, i;
|
|
|
11446 int t;
|
|
|
11447 int bc, gc;
|
|
324
|
11448 int pbc, pgc;
|
|
|
11449 #ifdef FEAT_MBYTE
|
|
|
11450 char_u *p;
|
|
|
11451 int wbadword[MAXWLEN];
|
|
|
11452 int wgoodword[MAXWLEN];
|
|
|
11453
|
|
|
11454 if (has_mbyte)
|
|
|
11455 {
|
|
|
11456 /* Get the characters from the multi-byte strings and put them in an
|
|
|
11457 * int array for easy access. */
|
|
|
11458 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
11459 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
11460 wbadword[badlen++] = 0;
|
|
324
|
11461 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
11462 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
11463 wgoodword[goodlen++] = 0;
|
|
324
|
11464 }
|
|
|
11465 else
|
|
|
11466 #endif
|
|
|
11467 {
|
|
|
11468 badlen = STRLEN(badword) + 1;
|
|
|
11469 goodlen = STRLEN(goodword) + 1;
|
|
|
11470 }
|
|
323
|
11471
|
|
|
11472 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
11473 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
11474 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
11475 TRUE);
|
|
324
|
11476 if (cnt == NULL)
|
|
|
11477 return 0; /* out of memory */
|
|
323
|
11478
|
|
|
11479 CNT(0, 0) = 0;
|
|
|
11480 for (j = 1; j <= goodlen; ++j)
|
|
|
11481 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
11482
|
|
|
11483 for (i = 1; i <= badlen; ++i)
|
|
|
11484 {
|
|
|
11485 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
11486 for (j = 1; j <= goodlen; ++j)
|
|
|
11487 {
|
|
324
|
11488 #ifdef FEAT_MBYTE
|
|
|
11489 if (has_mbyte)
|
|
|
11490 {
|
|
|
11491 bc = wbadword[i - 1];
|
|
|
11492 gc = wgoodword[j - 1];
|
|
|
11493 }
|
|
|
11494 else
|
|
|
11495 #endif
|
|
|
11496 {
|
|
|
11497 bc = badword[i - 1];
|
|
|
11498 gc = goodword[j - 1];
|
|
|
11499 }
|
|
323
|
11500 if (bc == gc)
|
|
|
11501 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
11502 else
|
|
|
11503 {
|
|
|
11504 /* Use a better score when there is only a case difference. */
|
|
324
|
11505 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
11506 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
11507 else
|
|
|
11508 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
11509
|
|
324
|
11510 if (i > 1 && j > 1)
|
|
323
|
11511 {
|
|
324
|
11512 #ifdef FEAT_MBYTE
|
|
|
11513 if (has_mbyte)
|
|
|
11514 {
|
|
|
11515 pbc = wbadword[i - 2];
|
|
|
11516 pgc = wgoodword[j - 2];
|
|
|
11517 }
|
|
|
11518 else
|
|
|
11519 #endif
|
|
|
11520 {
|
|
|
11521 pbc = badword[i - 2];
|
|
|
11522 pgc = goodword[j - 2];
|
|
|
11523 }
|
|
|
11524 if (bc == pgc && pbc == gc)
|
|
|
11525 {
|
|
|
11526 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
11527 if (t < CNT(i, j))
|
|
|
11528 CNT(i, j) = t;
|
|
|
11529 }
|
|
323
|
11530 }
|
|
|
11531 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
11532 if (t < CNT(i, j))
|
|
|
11533 CNT(i, j) = t;
|
|
|
11534 t = SCORE_INS + CNT(i, j - 1);
|
|
|
11535 if (t < CNT(i, j))
|
|
|
11536 CNT(i, j) = t;
|
|
|
11537 }
|
|
|
11538 }
|
|
|
11539 }
|
|
344
|
11540
|
|
|
11541 i = CNT(badlen - 1, goodlen - 1);
|
|
|
11542 vim_free(cnt);
|
|
|
11543 return i;
|
|
323
|
11544 }
|
|
307
|
11545
|
|
351
|
11546 /*
|
|
|
11547 * ":spelldump"
|
|
|
11548 */
|
|
|
11549 /*ARGSUSED*/
|
|
|
11550 void
|
|
|
11551 ex_spelldump(eap)
|
|
|
11552 exarg_T *eap;
|
|
|
11553 {
|
|
|
11554 buf_T *buf = curbuf;
|
|
|
11555 langp_T *lp;
|
|
|
11556 slang_T *slang;
|
|
|
11557 idx_T arridx[MAXWLEN];
|
|
|
11558 int curi[MAXWLEN];
|
|
|
11559 char_u word[MAXWLEN];
|
|
|
11560 int c;
|
|
|
11561 char_u *byts;
|
|
|
11562 idx_T *idxs;
|
|
|
11563 linenr_T lnum = 0;
|
|
|
11564 int round;
|
|
|
11565 int depth;
|
|
|
11566 int n;
|
|
|
11567 int flags;
|
|
381
|
11568 char_u *region_names = NULL; /* region names being used */
|
|
|
11569 int do_region = TRUE; /* dump region names and numbers */
|
|
|
11570 char_u *p;
|
|
351
|
11571
|
|
|
11572 if (no_spell_checking())
|
|
|
11573 return;
|
|
|
11574
|
|
|
11575 /* Create a new empty buffer by splitting the window. */
|
|
|
11576 do_cmdline_cmd((char_u *)"new");
|
|
|
11577 if (!bufempty() || !buf_valid(buf))
|
|
|
11578 return;
|
|
|
11579
|
|
381
|
11580 /* Find out if we can support regions: All languages must support the same
|
|
|
11581 * regions or none at all. */
|
|
|
11582 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
11583 {
|
|
|
11584 p = lp->lp_slang->sl_regions;
|
|
|
11585 if (p[0] != 0)
|
|
|
11586 {
|
|
|
11587 if (region_names == NULL) /* first language with regions */
|
|
|
11588 region_names = p;
|
|
|
11589 else if (STRCMP(region_names, p) != 0)
|
|
|
11590 {
|
|
|
11591 do_region = FALSE; /* region names are different */
|
|
|
11592 break;
|
|
|
11593 }
|
|
|
11594 }
|
|
|
11595 }
|
|
|
11596
|
|
|
11597 if (do_region && region_names != NULL)
|
|
|
11598 {
|
|
|
11599 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
11600 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
11601 }
|
|
|
11602 else
|
|
|
11603 do_region = FALSE;
|
|
|
11604
|
|
|
11605 /*
|
|
|
11606 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
11607 */
|
|
351
|
11608 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
11609 {
|
|
|
11610 slang = lp->lp_slang;
|
|
|
11611
|
|
|
11612 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
11613 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
11614
|
|
|
11615 /* round 1: case-folded tree
|
|
|
11616 * round 2: keep-case tree */
|
|
|
11617 for (round = 1; round <= 2; ++round)
|
|
|
11618 {
|
|
|
11619 if (round == 1)
|
|
|
11620 {
|
|
|
11621 byts = slang->sl_fbyts;
|
|
|
11622 idxs = slang->sl_fidxs;
|
|
|
11623 }
|
|
|
11624 else
|
|
|
11625 {
|
|
|
11626 byts = slang->sl_kbyts;
|
|
|
11627 idxs = slang->sl_kidxs;
|
|
|
11628 }
|
|
|
11629 if (byts == NULL)
|
|
|
11630 continue; /* array is empty */
|
|
|
11631
|
|
|
11632 depth = 0;
|
|
|
11633 arridx[0] = 0;
|
|
|
11634 curi[0] = 1;
|
|
|
11635 while (depth >= 0 && !got_int)
|
|
|
11636 {
|
|
|
11637 if (curi[depth] > byts[arridx[depth]])
|
|
|
11638 {
|
|
|
11639 /* Done all bytes at this node, go up one level. */
|
|
|
11640 --depth;
|
|
|
11641 line_breakcheck();
|
|
|
11642 }
|
|
|
11643 else
|
|
|
11644 {
|
|
|
11645 /* Do one more byte at this node. */
|
|
|
11646 n = arridx[depth] + curi[depth];
|
|
|
11647 ++curi[depth];
|
|
|
11648 c = byts[n];
|
|
|
11649 if (c == 0)
|
|
|
11650 {
|
|
|
11651 /* End of word, deal with the word.
|
|
|
11652 * Don't use keep-case words in the fold-case tree,
|
|
|
11653 * they will appear in the keep-case tree.
|
|
|
11654 * Only use the word when the region matches. */
|
|
|
11655 flags = (int)idxs[n];
|
|
|
11656 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
11657 && (do_region
|
|
|
11658 || (flags & WF_REGION) == 0
|
|
390
|
11659 || (((unsigned)flags >> 16)
|
|
351
|
11660 & lp->lp_region) != 0))
|
|
|
11661 {
|
|
|
11662 word[depth] = NUL;
|
|
381
|
11663 if (!do_region)
|
|
|
11664 flags &= ~WF_REGION;
|
|
355
|
11665
|
|
|
11666 /* Dump the basic word if there is no prefix or
|
|
|
11667 * when it's the first one. */
|
|
390
|
11668 c = (unsigned)flags >> 24;
|
|
355
|
11669 if (c == 0 || curi[depth] == 2)
|
|
|
11670 dump_word(word, round, flags, lnum++);
|
|
351
|
11671
|
|
|
11672 /* Apply the prefix, if there is one. */
|
|
355
|
11673 if (c != 0)
|
|
351
|
11674 lnum = apply_prefixes(slang, word, round,
|
|
|
11675 flags, lnum);
|
|
|
11676 }
|
|
|
11677 }
|
|
|
11678 else
|
|
|
11679 {
|
|
|
11680 /* Normal char, go one level deeper. */
|
|
|
11681 word[depth++] = c;
|
|
|
11682 arridx[depth] = idxs[n];
|
|
|
11683 curi[depth] = 1;
|
|
|
11684 }
|
|
|
11685 }
|
|
|
11686 }
|
|
|
11687 }
|
|
|
11688 }
|
|
|
11689
|
|
|
11690 /* Delete the empty line that we started with. */
|
|
|
11691 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
11692 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
11693
|
|
|
11694 redraw_later(NOT_VALID);
|
|
|
11695 }
|
|
|
11696
|
|
|
11697 /*
|
|
|
11698 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
11699 */
|
|
|
11700 static void
|
|
|
11701 dump_word(word, round, flags, lnum)
|
|
|
11702 char_u *word;
|
|
|
11703 int round;
|
|
|
11704 int flags;
|
|
|
11705 linenr_T lnum;
|
|
|
11706 {
|
|
|
11707 int keepcap = FALSE;
|
|
|
11708 char_u *p;
|
|
|
11709 char_u cword[MAXWLEN];
|
|
381
|
11710 char_u badword[MAXWLEN + 10];
|
|
|
11711 int i;
|
|
351
|
11712
|
|
|
11713 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
11714 {
|
|
|
11715 /* Need to fix case according to "flags". */
|
|
|
11716 make_case_word(word, cword, flags);
|
|
|
11717 p = cword;
|
|
|
11718 }
|
|
|
11719 else
|
|
|
11720 {
|
|
|
11721 p = word;
|
|
389
|
11722 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
11723 || (flags & WF_FIXCAP) != 0))
|
|
351
|
11724 keepcap = TRUE;
|
|
|
11725 }
|
|
|
11726
|
|
381
|
11727 /* Add flags and regions after a slash. */
|
|
|
11728 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
11729 {
|
|
|
11730 STRCPY(badword, p);
|
|
|
11731 STRCAT(badword, "/");
|
|
351
|
11732 if (keepcap)
|
|
|
11733 STRCAT(badword, "=");
|
|
|
11734 if (flags & WF_BANNED)
|
|
|
11735 STRCAT(badword, "!");
|
|
|
11736 else if (flags & WF_RARE)
|
|
|
11737 STRCAT(badword, "?");
|
|
381
|
11738 if (flags & WF_REGION)
|
|
|
11739 for (i = 0; i < 7; ++i)
|
|
390
|
11740 if (flags & (0x10000 << i))
|
|
381
|
11741 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
11742 p = badword;
|
|
|
11743 }
|
|
|
11744
|
|
|
11745 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
11746 }
|
|
|
11747
|
|
|
11748 /*
|
|
372
|
11749 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
11750 * "word" and append a line to the buffer.
|
|
351
|
11751 * Return the updated line number.
|
|
|
11752 */
|
|
|
11753 static linenr_T
|
|
|
11754 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
11755 slang_T *slang;
|
|
|
11756 char_u *word; /* case-folded word */
|
|
|
11757 int round;
|
|
|
11758 int flags; /* flags with prefix ID */
|
|
|
11759 linenr_T startlnum;
|
|
|
11760 {
|
|
|
11761 idx_T arridx[MAXWLEN];
|
|
|
11762 int curi[MAXWLEN];
|
|
|
11763 char_u prefix[MAXWLEN];
|
|
455
|
11764 char_u word_up[MAXWLEN];
|
|
|
11765 int has_word_up = FALSE;
|
|
351
|
11766 int c;
|
|
|
11767 char_u *byts;
|
|
|
11768 idx_T *idxs;
|
|
|
11769 linenr_T lnum = startlnum;
|
|
|
11770 int depth;
|
|
|
11771 int n;
|
|
|
11772 int len;
|
|
|
11773 int i;
|
|
|
11774
|
|
455
|
11775 /* if the word starts with a lower-case letter make the word with an
|
|
|
11776 * upper-case letter in word_up[]. */
|
|
|
11777 c = PTR2CHAR(word);
|
|
|
11778 if (SPELL_TOUPPER(c) != c)
|
|
|
11779 {
|
|
|
11780 onecap_copy(word, word_up, TRUE);
|
|
|
11781 has_word_up = TRUE;
|
|
|
11782 }
|
|
|
11783
|
|
351
|
11784 byts = slang->sl_pbyts;
|
|
|
11785 idxs = slang->sl_pidxs;
|
|
|
11786 if (byts != NULL) /* array not is empty */
|
|
|
11787 {
|
|
|
11788 /*
|
|
|
11789 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
11790 * When at the end of a prefix check that it supports "flags".
|
|
351
|
11791 */
|
|
|
11792 depth = 0;
|
|
|
11793 arridx[0] = 0;
|
|
|
11794 curi[0] = 1;
|
|
|
11795 while (depth >= 0 && !got_int)
|
|
|
11796 {
|
|
390
|
11797 n = arridx[depth];
|
|
|
11798 len = byts[n];
|
|
|
11799 if (curi[depth] > len)
|
|
351
|
11800 {
|
|
|
11801 /* Done all bytes at this node, go up one level. */
|
|
|
11802 --depth;
|
|
|
11803 line_breakcheck();
|
|
|
11804 }
|
|
|
11805 else
|
|
|
11806 {
|
|
|
11807 /* Do one more byte at this node. */
|
|
390
|
11808 n += curi[depth];
|
|
351
|
11809 ++curi[depth];
|
|
|
11810 c = byts[n];
|
|
|
11811 if (c == 0)
|
|
|
11812 {
|
|
|
11813 /* End of prefix, find out how many IDs there are. */
|
|
|
11814 for (i = 1; i < len; ++i)
|
|
|
11815 if (byts[n + i] != 0)
|
|
|
11816 break;
|
|
|
11817 curi[depth] += i - 1;
|
|
|
11818
|
|
455
|
11819 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
11820 if (c != 0)
|
|
351
|
11821 {
|
|
376
|
11822 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
11823 dump_word(prefix, round,
|
|
455
|
11824 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
11825 : flags, lnum++);
|
|
351
|
11826 }
|
|
455
|
11827
|
|
|
11828 /* Check for prefix that matches the word when the
|
|
|
11829 * first letter is upper-case, but only if the prefix has
|
|
|
11830 * a condition. */
|
|
|
11831 if (has_word_up)
|
|
|
11832 {
|
|
|
11833 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
11834 TRUE);
|
|
|
11835 if (c != 0)
|
|
|
11836 {
|
|
|
11837 vim_strncpy(prefix + depth, word_up,
|
|
|
11838 MAXWLEN - depth - 1);
|
|
|
11839 dump_word(prefix, round,
|
|
|
11840 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
11841 : flags, lnum++);
|
|
|
11842 }
|
|
|
11843 }
|
|
351
|
11844 }
|
|
|
11845 else
|
|
|
11846 {
|
|
|
11847 /* Normal char, go one level deeper. */
|
|
|
11848 prefix[depth++] = c;
|
|
|
11849 arridx[depth] = idxs[n];
|
|
|
11850 curi[depth] = 1;
|
|
|
11851 }
|
|
|
11852 }
|
|
|
11853 }
|
|
|
11854 }
|
|
|
11855
|
|
|
11856 return lnum;
|
|
|
11857 }
|
|
|
11858
|
|
475
|
11859 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
11860 static int spell_expand_need_cap;
|
|
|
11861
|
|
|
11862 /*
|
|
|
11863 * Find start of the word in front of the cursor. We don't check if it is
|
|
|
11864 * badly spelled, with completion we can only change the word in front of the
|
|
|
11865 * cursor.
|
|
|
11866 * Used for Insert mode completion CTRL-X ?.
|
|
|
11867 * Returns the column number of the word.
|
|
|
11868 */
|
|
|
11869 int
|
|
|
11870 spell_word_start(startcol)
|
|
|
11871 int startcol;
|
|
|
11872 {
|
|
|
11873 char_u *line;
|
|
|
11874 char_u *p;
|
|
|
11875 int col = 0;
|
|
|
11876
|
|
|
11877 if (no_spell_checking())
|
|
|
11878 return startcol;
|
|
|
11879
|
|
|
11880 /* Find a word character before "startcol". */
|
|
|
11881 line = ml_get_curline();
|
|
|
11882 for (p = line + startcol; p > line; )
|
|
|
11883 {
|
|
|
11884 mb_ptr_back(line, p);
|
|
|
11885 if (spell_iswordp_nmw(p))
|
|
|
11886 break;
|
|
|
11887 }
|
|
|
11888
|
|
|
11889 /* Go back to start of the word. */
|
|
|
11890 while (p > line)
|
|
|
11891 {
|
|
|
11892 col = p - line;
|
|
|
11893 mb_ptr_back(line, p);
|
|
|
11894 if (!spell_iswordp(p, curbuf))
|
|
|
11895 break;
|
|
|
11896 col = 0;
|
|
|
11897 }
|
|
|
11898
|
|
|
11899 /* Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
11900 * expand_spelling() is called. Therefore the ugly global variable. */
|
|
|
11901 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
11902
|
|
|
11903 return col;
|
|
|
11904 }
|
|
|
11905
|
|
|
11906 /*
|
|
|
11907 * Get list of spelling suggestions.
|
|
|
11908 * Used for Insert mode completion CTRL-X ?.
|
|
|
11909 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
11910 * allocated strings.
|
|
|
11911 */
|
|
|
11912 /*ARGSUSED*/
|
|
|
11913 int
|
|
|
11914 expand_spelling(lnum, col, pat, matchp)
|
|
|
11915 linenr_T lnum;
|
|
|
11916 int col;
|
|
|
11917 char_u *pat;
|
|
|
11918 char_u ***matchp;
|
|
|
11919 {
|
|
|
11920 garray_T ga;
|
|
|
11921
|
|
|
11922 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap);
|
|
|
11923 *matchp = ga.ga_data;
|
|
|
11924 return ga.ga_len;
|
|
|
11925 }
|
|
|
11926 #endif
|
|
|
11927
|
|
236
|
11928 #endif /* FEAT_SYN_HL */
|
