|
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
|
|
324
|
253 /* Type used for indexes in the word tree need to be at least 3 bytes. If int
|
|
|
254 * is 8 bytes we could use something smaller, but what? */
|
|
|
255 #if SIZEOF_INT > 2
|
|
|
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 */
|
|
|
380 char_u *sl_syllable; /* SYLLABLE repeatable chars or NULL */
|
|
|
381 garray_T sl_syl_items; /* syllable items */
|
|
481
|
382
|
|
339
|
383 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
384 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
385
|
|
323
|
386 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
387 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
388 there is none */
|
|
344
|
389 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
390 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
391 there is none */
|
|
375
|
392 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
393 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
394 * "sl_sal" is a list of wide char lists. */
|
|
323
|
395 int sl_followup; /* SAL followup */
|
|
|
396 int sl_collapse; /* SAL collapse_result */
|
|
|
397 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
398 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
399 #ifdef FEAT_MBYTE
|
|
|
400 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
401 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
402 #else
|
|
|
403 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
404 #endif
|
|
236
|
405 };
|
|
|
406
|
|
243
|
407 /* First language that is loaded, start of the linked list of loaded
|
|
|
408 * languages. */
|
|
236
|
409 static slang_T *first_lang = NULL;
|
|
|
410
|
|
323
|
411 /* Flags used in .spl file for soundsalike flags. */
|
|
|
412 #define SAL_F0LLOWUP 1
|
|
|
413 #define SAL_COLLAPSE 2
|
|
|
414 #define SAL_REM_ACCENTS 4
|
|
|
415
|
|
236
|
416 /*
|
|
|
417 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
418 */
|
|
|
419 typedef struct langp_S
|
|
|
420 {
|
|
|
421 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
|
422 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
423 } langp_T;
|
|
|
424
|
|
|
425 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
426
|
|
307
|
427 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
428
|
|
483
|
429 #define VIMSPELLMAGIC "VIMspell" /* string at start of Vim spell file */
|
|
|
430 #define VIMSPELLMAGICL 8
|
|
|
431 #define VIMSPELLVERSION 50
|
|
|
432
|
|
|
433 /* Section IDs. Only renumber them when VIMSPELLVERSION changes! */
|
|
|
434 #define SN_REGION 0 /* <regionname> section */
|
|
|
435 #define SN_CHARFLAGS 1 /* charflags section */
|
|
|
436 #define SN_MIDWORD 2 /* <midword> section */
|
|
|
437 #define SN_PREFCOND 3 /* <prefcond> section */
|
|
|
438 #define SN_REP 4 /* REP items section */
|
|
|
439 #define SN_SAL 5 /* SAL items section */
|
|
|
440 #define SN_SOFO 6 /* soundfolding section */
|
|
|
441 #define SN_MAP 7 /* MAP items section */
|
|
|
442 #define SN_COMPOUND 8 /* compound words section */
|
|
|
443 #define SN_SYLLABLE 9 /* syllable section */
|
|
|
444 #define SN_END 255 /* end of sections */
|
|
|
445
|
|
|
446 #define SNF_REQUIRED 1 /* <sectionflags>: required section */
|
|
|
447
|
|
307
|
448 /* Result values. Lower number is accepted over higher one. */
|
|
|
449 #define SP_BANNED -1
|
|
236
|
450 #define SP_OK 0
|
|
307
|
451 #define SP_RARE 1
|
|
|
452 #define SP_LOCAL 2
|
|
|
453 #define SP_BAD 3
|
|
236
|
454
|
|
381
|
455 /* file used for "zG" and "zW" */
|
|
385
|
456 static char_u *int_wordlist = NULL;
|
|
381
|
457
|
|
236
|
458 /*
|
|
323
|
459 * Information used when looking for suggestions.
|
|
|
460 */
|
|
|
461 typedef struct suginfo_S
|
|
|
462 {
|
|
|
463 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
464 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
465 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
466 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
467 char_u *su_badptr; /* start of bad word in line */
|
|
|
468 int su_badlen; /* length of detected bad word in line */
|
|
346
|
469 int su_badflags; /* caps flags for bad word */
|
|
323
|
470 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
471 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
|
472 hashtab_T su_banned; /* table with banned words */
|
|
|
473 } suginfo_T;
|
|
|
474
|
|
|
475 /* One word suggestion. Used in "si_ga". */
|
|
|
476 typedef struct suggest_S
|
|
|
477 {
|
|
|
478 char_u *st_word; /* suggested word, allocated string */
|
|
|
479 int st_orglen; /* length of replaced text */
|
|
|
480 int st_score; /* lower is better */
|
|
344
|
481 int st_altscore; /* used when st_score compares equal */
|
|
|
482 int st_salscore; /* st_score is for soundalike */
|
|
324
|
483 int st_had_bonus; /* bonus already included in score */
|
|
323
|
484 } suggest_T;
|
|
|
485
|
|
344
|
486 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
487
|
|
324
|
488 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
489 * called the score may change, thus we need to keep more than what is
|
|
|
490 * displayed. */
|
|
351
|
491 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
492
|
|
|
493 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
494 * of suggestions that are not going to be displayed. */
|
|
344
|
495 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
496
|
|
|
497 /* score for various changes */
|
|
344
|
498 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
499 #define SCORE_ICASE 52 /* slightly different case */
|
|
483
|
500 #define SCORE_REGION 200 /* word is for different region */
|
|
323
|
501 #define SCORE_RARE 180 /* rare word */
|
|
|
502 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
503 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
|
504 #define SCORE_REP 87 /* REP replacement */
|
|
|
505 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
506 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
479
|
507 #define SCORE_SUBCOMP 33 /* substitute a composing character */
|
|
324
|
508 #define SCORE_DEL 94 /* delete a character */
|
|
358
|
509 #define SCORE_DELDUP 64 /* delete a duplicated character */
|
|
479
|
510 #define SCORE_DELCOMP 28 /* delete a composing character */
|
|
324
|
511 #define SCORE_INS 96 /* insert a character */
|
|
358
|
512 #define SCORE_INSDUP 66 /* insert a duplicate character */
|
|
475
|
513 #define SCORE_INSCOMP 30 /* insert a composing character */
|
|
366
|
514 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
515
|
|
372
|
516 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
517 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
518 * 350 allows for about three changes. */
|
|
344
|
519
|
|
|
520 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
521 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
522
|
|
|
523 /*
|
|
236
|
524 * Structure to store info for word matching.
|
|
|
525 */
|
|
|
526 typedef struct matchinf_S
|
|
|
527 {
|
|
|
528 langp_T *mi_lp; /* info for language and region */
|
|
243
|
529
|
|
|
530 /* pointers to original text to be checked */
|
|
236
|
531 char_u *mi_word; /* start of word being checked */
|
|
339
|
532 char_u *mi_end; /* end of matching word so far */
|
|
243
|
533 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
534 char_u *mi_cend; /* char after what was used for
|
|
|
535 mi_capflags */
|
|
243
|
536
|
|
|
537 /* case-folded text */
|
|
|
538 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
539 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
540
|
|
339
|
541 /* for when checking word after a prefix */
|
|
|
542 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
481
|
543 affixID/condition */
|
|
339
|
544 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
545 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
546 #ifdef FEAT_MBYTE
|
|
|
547 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
548 case */
|
|
|
549 #else
|
|
|
550 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
551 #endif
|
|
339
|
552
|
|
481
|
553 /* for when checking a compound word */
|
|
|
554 int mi_compoff; /* start of following word offset */
|
|
483
|
555 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */
|
|
|
556 int mi_complen; /* nr of compound words used */
|
|
481
|
557
|
|
243
|
558 /* others */
|
|
236
|
559 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
560 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
561 buf_T *mi_buf; /* buffer being checked */
|
|
236
|
562 } matchinf_T;
|
|
|
563
|
|
307
|
564 /*
|
|
|
565 * The tables used for recognizing word characters according to spelling.
|
|
|
566 * These are only used for the first 256 characters of 'encoding'.
|
|
|
567 */
|
|
|
568 typedef struct spelltab_S
|
|
|
569 {
|
|
|
570 char_u st_isw[256]; /* flags: is word char */
|
|
|
571 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
572 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
573 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
574 } spelltab_T;
|
|
|
575
|
|
|
576 static spelltab_T spelltab;
|
|
|
577 static int did_set_spelltab;
|
|
|
578
|
|
324
|
579 #define CF_WORD 0x01
|
|
|
580 #define CF_UPPER 0x02
|
|
307
|
581
|
|
|
582 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
583 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
584 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
585 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
586 #ifdef FEAT_MBYTE
|
|
|
587 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
588 #endif
|
|
483
|
589 static int write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
590
|
|
|
591 /*
|
|
344
|
592 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
593 */
|
|
|
594 typedef enum
|
|
|
595 {
|
|
344
|
596 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
597 * ends); if badword ends there is a match, otherwise
|
|
|
598 * try splitting word. */
|
|
375
|
599 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
600 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
601 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
602 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
603 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
604 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
605 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
606 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
607 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
608 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
609 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
610 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
611 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
612 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
613 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
614 STATE_FINAL /* End of this node. */
|
|
|
615 } state_T;
|
|
|
616
|
|
|
617 /*
|
|
346
|
618 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
619 */
|
|
|
620 typedef struct trystate_S
|
|
|
621 {
|
|
330
|
622 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
623 int ts_score; /* score */
|
|
344
|
624 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
625 short ts_curi; /* index in list of child nodes */
|
|
|
626 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
627 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
628 char_u ts_twordlen; /* valid length of tword[] */
|
|
482
|
629 char_u ts_prefixdepth; /* stack depth for end of prefix or
|
|
|
630 * PFD_PREFIXTREE or PFD_NOPREFIX or
|
|
|
631 * PFD_COMPOUND */
|
|
330
|
632 #ifdef FEAT_MBYTE
|
|
|
633 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
634 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
635 char_u ts_isdiff; /* DIFF_ values */
|
|
|
636 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
637 #endif
|
|
483
|
638 char_u ts_prewordlen; /* length of word in "preword[]" */
|
|
|
639 char_u ts_splitoff; /* index in "tword" after last split */
|
|
|
640 char_u ts_complen; /* nr of compound words used */
|
|
346
|
641 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
642 } trystate_T;
|
|
|
643
|
|
330
|
644 /* values for ts_isdiff */
|
|
|
645 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
646 #define DIFF_YES 1 /* different byte found */
|
|
|
647 #define DIFF_INSERT 2 /* inserting character */
|
|
|
648
|
|
375
|
649 /* special values ts_prefixdepth */
|
|
482
|
650 #define PFD_COMPOUND 0xfd /* prefixed is a compound word */
|
|
|
651 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
|
652 #define PFD_NOPREFIX 0xff /* not using prefixes */
|
|
375
|
653
|
|
339
|
654 /* mode values for find_word */
|
|
481
|
655 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
656 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
657 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
658 #define FIND_COMPOUND 3 /* find case-folded compound word */
|
|
|
659 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */
|
|
339
|
660
|
|
236
|
661 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
662 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
663 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
664 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
483
|
665 static int can_compound __ARGS((slang_T *slang, char_u *word, char_u *flags));
|
|
455
|
666 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
339
|
667 static void find_prefix __ARGS((matchinf_T *mip));
|
|
|
668 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
669 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
351
|
670 static int no_spell_checking __ARGS((void));
|
|
307
|
671 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
672 static char_u *spell_enc __ARGS((void));
|
|
385
|
673 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
674 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
675 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
676 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
483
|
677 static char_u *read_string __ARGS((FILE *fd, int cnt));
|
|
|
678 static int read_region_section __ARGS((FILE *fd, slang_T *slang, int len));
|
|
|
679 static int read_charflags_section __ARGS((FILE *fd));
|
|
|
680 static int read_prefcond_section __ARGS((FILE *fd, slang_T *lp));
|
|
|
681 static int read_rep_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
682 static int read_sal_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
683 static int read_sofo_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
684 static int read_compound __ARGS((FILE *fd, slang_T *slang, int len));
|
|
|
685 static int init_syl_tab __ARGS((slang_T *slang));
|
|
|
686 static int count_syllables __ARGS((slang_T *slang, char_u *word));
|
|
381
|
687 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
688 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
689 #ifdef FEAT_MBYTE
|
|
|
690 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
691 #endif
|
|
339
|
692 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
693 static void clear_midword __ARGS((buf_T *buf));
|
|
|
694 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
695 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
696 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
697 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
698 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
483
|
699 static void set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
700 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
701 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
475
|
702 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col));
|
|
388
|
703 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
704 #ifdef FEAT_EVAL
|
|
|
705 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
706 #endif
|
|
|
707 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
708 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
709 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
710 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
711 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
712 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
713 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
714 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
715 #ifdef FEAT_MBYTE
|
|
|
716 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
717 #endif
|
|
323
|
718 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
719 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
720 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
721 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
722 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
723 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
724 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
725 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
351
|
726 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus));
|
|
323
|
727 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
728 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
729 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
730 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
344
|
731 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
732 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
733 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
734 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
735 #ifdef FEAT_MBYTE
|
|
375
|
736 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
737 #endif
|
|
344
|
738 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
739 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
740 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
|
741 static linenr_T apply_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
742
|
|
324
|
743 /*
|
|
|
744 * Use our own character-case definitions, because the current locale may
|
|
|
745 * differ from what the .spl file uses.
|
|
|
746 * These must not be called with negative number!
|
|
|
747 */
|
|
|
748 #ifndef FEAT_MBYTE
|
|
|
749 /* Non-multi-byte implementation. */
|
|
|
750 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
751 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
752 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
753 #else
|
|
447
|
754 # if defined(HAVE_WCHAR_H)
|
|
|
755 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
756 # endif
|
|
324
|
757 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
758 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
759 * the "w" library function for characters above 255 if available. */
|
|
|
760 # ifdef HAVE_TOWLOWER
|
|
|
761 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
762 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
763 # else
|
|
|
764 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
765 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
766 # endif
|
|
|
767
|
|
|
768 # ifdef HAVE_TOWUPPER
|
|
|
769 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
770 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
771 # else
|
|
|
772 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
773 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
774 # endif
|
|
|
775
|
|
|
776 # ifdef HAVE_ISWUPPER
|
|
|
777 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
778 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
779 # else
|
|
|
780 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
781 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
782 # endif
|
|
|
783 #endif
|
|
|
784
|
|
307
|
785
|
|
|
786 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
787 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
482
|
788 static char *e_afftrailing = N_("Trailing text in %s line %d: %s");
|
|
469
|
789 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
790
|
|
|
791 /*
|
|
|
792 * Main spell-checking function.
|
|
300
|
793 * "ptr" points to a character that could be the start of a word.
|
|
236
|
794 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
|
795 * or when it's OK it remains unchanged.
|
|
|
796 * This must only be called when 'spelllang' is not empty.
|
|
323
|
797 *
|
|
385
|
798 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
799 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
800 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
801 * worry.
|
|
323
|
802 *
|
|
236
|
803 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
804 * caller can skip over the word.
|
|
|
805 */
|
|
|
806 int
|
|
385
|
807 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
808 win_T *wp; /* current window */
|
|
|
809 char_u *ptr;
|
|
|
810 int *attrp;
|
|
385
|
811 int *capcol; /* column to check for Capital */
|
|
236
|
812 {
|
|
|
813 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
814 be passed to functions quickly. */
|
|
344
|
815 int nrlen = 0; /* found a number first */
|
|
385
|
816 int c;
|
|
483
|
817 int wrongcaplen = 0;
|
|
236
|
818
|
|
307
|
819 /* A word never starts at a space or a control character. Return quickly
|
|
|
820 * then, skipping over the character. */
|
|
|
821 if (*ptr <= ' ')
|
|
|
822 return 1;
|
|
483
|
823 vim_memset(&mi, 0, sizeof(matchinf_T));
|
|
236
|
824
|
|
344
|
825 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
346
|
826 * 0X99FF. But when a word character follows do check spelling to find
|
|
|
827 * "3GPP". */
|
|
300
|
828 if (*ptr >= '0' && *ptr <= '9')
|
|
|
829 {
|
|
316
|
830 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
831 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
832 else
|
|
|
833 {
|
|
344
|
834 mi.mi_end = skipdigits(ptr);
|
|
|
835 nrlen = mi.mi_end - ptr;
|
|
300
|
836 }
|
|
376
|
837 if (!spell_iswordp(mi.mi_end, wp->w_buffer))
|
|
344
|
838 return (int)(mi.mi_end - ptr);
|
|
346
|
839
|
|
|
840 /* Try including the digits in the word. */
|
|
|
841 mi.mi_fend = ptr + nrlen;
|
|
|
842 }
|
|
|
843 else
|
|
|
844 mi.mi_fend = ptr;
|
|
|
845
|
|
|
846 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
847 mi.mi_word = ptr;
|
|
376
|
848 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
849 {
|
|
|
850 do
|
|
|
851 {
|
|
307
|
852 mb_ptr_adv(mi.mi_fend);
|
|
376
|
853 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
854
|
|
|
855 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
856 {
|
|
|
857 /* Check word starting with capital letter. */
|
|
455
|
858 c = PTR2CHAR(ptr);
|
|
385
|
859 if (!SPELL_ISUPPER(c))
|
|
483
|
860 wrongcaplen = (int)(mi.mi_fend - ptr);
|
|
385
|
861 }
|
|
|
862 }
|
|
|
863 if (capcol != NULL)
|
|
|
864 *capcol = -1;
|
|
344
|
865
|
|
|
866 /* We always use the characters up to the next non-word character,
|
|
|
867 * also for bad words. */
|
|
|
868 mi.mi_end = mi.mi_fend;
|
|
|
869
|
|
|
870 /* Check caps type later. */
|
|
376
|
871 mi.mi_buf = wp->w_buffer;
|
|
344
|
872
|
|
483
|
873 /* case-fold the word with one non-word character, so that we can check
|
|
|
874 * for the word end. */
|
|
344
|
875 if (*mi.mi_fend != NUL)
|
|
|
876 mb_ptr_adv(mi.mi_fend);
|
|
|
877
|
|
|
878 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
879 MAXWLEN + 1);
|
|
|
880 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
881
|
|
|
882 /* The word is bad unless we recognize it. */
|
|
|
883 mi.mi_result = SP_BAD;
|
|
|
884
|
|
|
885 /*
|
|
|
886 * Loop over the languages specified in 'spelllang'.
|
|
|
887 * We check them all, because a matching word may be longer than an
|
|
|
888 * already found matching word.
|
|
|
889 */
|
|
|
890 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
|
891 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
|
|
892 {
|
|
|
893 /* Check for a matching word in case-folded words. */
|
|
|
894 find_word(&mi, FIND_FOLDWORD);
|
|
|
895
|
|
|
896 /* Check for a matching word in keep-case words. */
|
|
|
897 find_word(&mi, FIND_KEEPWORD);
|
|
|
898
|
|
|
899 /* Check for matching prefixes. */
|
|
|
900 find_prefix(&mi);
|
|
|
901 }
|
|
|
902
|
|
|
903 if (mi.mi_result != SP_OK)
|
|
|
904 {
|
|
346
|
905 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
906 * rare and local words, e.g., "3GPP". */
|
|
344
|
907 if (nrlen > 0)
|
|
346
|
908 {
|
|
|
909 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
910 return nrlen;
|
|
|
911 }
|
|
344
|
912
|
|
|
913 /* When we are at a non-word character there is no error, just
|
|
|
914 * skip over the character (try looking for a word after it). */
|
|
390
|
915 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
916 {
|
|
385
|
917 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
918 {
|
|
|
919 regmatch_T regmatch;
|
|
|
920
|
|
|
921 /* Check for end of sentence. */
|
|
|
922 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
923 regmatch.rm_ic = FALSE;
|
|
|
924 if (vim_regexec(®match, ptr, 0))
|
|
|
925 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
926 }
|
|
|
927
|
|
344
|
928 #ifdef FEAT_MBYTE
|
|
|
929 if (has_mbyte)
|
|
474
|
930 return (*mb_ptr2len)(ptr);
|
|
344
|
931 #endif
|
|
|
932 return 1;
|
|
300
|
933 }
|
|
483
|
934 else if (mi.mi_end == ptr)
|
|
|
935 /* Always include at least one character. Required for when there
|
|
|
936 * is a mixup in "midword". */
|
|
|
937 mb_ptr_adv(mi.mi_end);
|
|
243
|
938
|
|
344
|
939 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
940 *attrp = highlight_attr[HLF_SPB];
|
|
|
941 else if (mi.mi_result == SP_RARE)
|
|
|
942 *attrp = highlight_attr[HLF_SPR];
|
|
|
943 else
|
|
|
944 *attrp = highlight_attr[HLF_SPL];
|
|
243
|
945 }
|
|
|
946
|
|
483
|
947 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE))
|
|
|
948 {
|
|
|
949 /* Report SpellCap only when the word isn't badly spelled. */
|
|
|
950 *attrp = highlight_attr[HLF_SPC];
|
|
|
951 return wrongcaplen;
|
|
|
952 }
|
|
|
953
|
|
300
|
954 return (int)(mi.mi_end - ptr);
|
|
236
|
955 }
|
|
|
956
|
|
|
957 /*
|
|
300
|
958 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
959 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
960 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
961 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
962 * tree.
|
|
300
|
963 *
|
|
|
964 * For a match mip->mi_result is updated.
|
|
243
|
965 */
|
|
|
966 static void
|
|
339
|
967 find_word(mip, mode)
|
|
243
|
968 matchinf_T *mip;
|
|
339
|
969 int mode;
|
|
243
|
970 {
|
|
324
|
971 idx_T arridx = 0;
|
|
300
|
972 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
973 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
974 int endidxcnt = 0;
|
|
|
975 int len;
|
|
|
976 int wlen = 0;
|
|
|
977 int flen;
|
|
|
978 int c;
|
|
|
979 char_u *ptr;
|
|
324
|
980 idx_T lo, hi, m;
|
|
243
|
981 #ifdef FEAT_MBYTE
|
|
300
|
982 char_u *s;
|
|
339
|
983 char_u *p;
|
|
307
|
984 #endif
|
|
|
985 int res = SP_BAD;
|
|
300
|
986 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
987 unsigned flags;
|
|
|
988 char_u *byts;
|
|
324
|
989 idx_T *idxs;
|
|
481
|
990 int word_ends;
|
|
|
991
|
|
|
992 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND)
|
|
236
|
993 {
|
|
300
|
994 /* Check for word with matching case in keep-case tree. */
|
|
|
995 ptr = mip->mi_word;
|
|
|
996 flen = 9999; /* no case folding, always enough bytes */
|
|
|
997 byts = slang->sl_kbyts;
|
|
|
998 idxs = slang->sl_kidxs;
|
|
481
|
999
|
|
|
1000 if (mode == FIND_KEEPCOMPOUND)
|
|
|
1001 /* Skip over the previously found word(s). */
|
|
|
1002 wlen += mip->mi_compoff;
|
|
236
|
1003 }
|
|
|
1004 else
|
|
|
1005 {
|
|
300
|
1006 /* Check for case-folded in case-folded tree. */
|
|
|
1007 ptr = mip->mi_fword;
|
|
|
1008 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1009 byts = slang->sl_fbyts;
|
|
|
1010 idxs = slang->sl_fidxs;
|
|
339
|
1011
|
|
|
1012 if (mode == FIND_PREFIX)
|
|
|
1013 {
|
|
|
1014 /* Skip over the prefix. */
|
|
|
1015 wlen = mip->mi_prefixlen;
|
|
|
1016 flen -= mip->mi_prefixlen;
|
|
|
1017 }
|
|
481
|
1018 else if (mode == FIND_COMPOUND)
|
|
|
1019 {
|
|
|
1020 /* Skip over the previously found word(s). */
|
|
|
1021 wlen = mip->mi_compoff;
|
|
|
1022 flen -= mip->mi_compoff;
|
|
|
1023 }
|
|
|
1024
|
|
243
|
1025 }
|
|
|
1026
|
|
300
|
1027 if (byts == NULL)
|
|
|
1028 return; /* array is empty */
|
|
236
|
1029
|
|
|
1030 /*
|
|
307
|
1031 * Repeat advancing in the tree until:
|
|
|
1032 * - there is a byte that doesn't match,
|
|
|
1033 * - we reach the end of the tree,
|
|
|
1034 * - or we reach the end of the line.
|
|
236
|
1035 */
|
|
300
|
1036 for (;;)
|
|
236
|
1037 {
|
|
346
|
1038 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
1039 flen = fold_more(mip);
|
|
300
|
1040
|
|
|
1041 len = byts[arridx++];
|
|
|
1042
|
|
|
1043 /* If the first possible byte is a zero the word could end here.
|
|
|
1044 * Remember this index, we first check for the longest word. */
|
|
|
1045 if (byts[arridx] == 0)
|
|
|
1046 {
|
|
307
|
1047 if (endidxcnt == MAXWLEN)
|
|
|
1048 {
|
|
|
1049 /* Must be a corrupted spell file. */
|
|
|
1050 EMSG(_(e_format));
|
|
|
1051 return;
|
|
|
1052 }
|
|
300
|
1053 endlen[endidxcnt] = wlen;
|
|
|
1054 endidx[endidxcnt++] = arridx++;
|
|
|
1055 --len;
|
|
|
1056
|
|
|
1057 /* Skip over the zeros, there can be several flag/region
|
|
|
1058 * combinations. */
|
|
|
1059 while (len > 0 && byts[arridx] == 0)
|
|
|
1060 {
|
|
|
1061 ++arridx;
|
|
|
1062 --len;
|
|
|
1063 }
|
|
|
1064 if (len == 0)
|
|
|
1065 break; /* no children, word must end here */
|
|
|
1066 }
|
|
|
1067
|
|
|
1068 /* Stop looking at end of the line. */
|
|
|
1069 if (ptr[wlen] == NUL)
|
|
|
1070 break;
|
|
|
1071
|
|
|
1072 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1073 c = ptr[wlen];
|
|
346
|
1074 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1075 c = ' ';
|
|
300
|
1076 lo = arridx;
|
|
|
1077 hi = arridx + len - 1;
|
|
|
1078 while (lo < hi)
|
|
|
1079 {
|
|
|
1080 m = (lo + hi) / 2;
|
|
|
1081 if (byts[m] > c)
|
|
|
1082 hi = m - 1;
|
|
|
1083 else if (byts[m] < c)
|
|
|
1084 lo = m + 1;
|
|
|
1085 else
|
|
|
1086 {
|
|
|
1087 lo = hi = m;
|
|
|
1088 break;
|
|
236
|
1089 }
|
|
|
1090 }
|
|
300
|
1091
|
|
|
1092 /* Stop if there is no matching byte. */
|
|
|
1093 if (hi < lo || byts[lo] != c)
|
|
|
1094 break;
|
|
|
1095
|
|
|
1096 /* Continue at the child (if there is one). */
|
|
|
1097 arridx = idxs[lo];
|
|
|
1098 ++wlen;
|
|
|
1099 --flen;
|
|
346
|
1100
|
|
|
1101 /* One space in the good word may stand for several spaces in the
|
|
|
1102 * checked word. */
|
|
|
1103 if (c == ' ')
|
|
|
1104 {
|
|
|
1105 for (;;)
|
|
|
1106 {
|
|
|
1107 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1108 flen = fold_more(mip);
|
|
|
1109 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1110 break;
|
|
|
1111 ++wlen;
|
|
|
1112 --flen;
|
|
|
1113 }
|
|
|
1114 }
|
|
236
|
1115 }
|
|
|
1116
|
|
300
|
1117 /*
|
|
|
1118 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1119 * first.
|
|
|
1120 */
|
|
|
1121 while (endidxcnt > 0)
|
|
|
1122 {
|
|
|
1123 --endidxcnt;
|
|
|
1124 arridx = endidx[endidxcnt];
|
|
|
1125 wlen = endlen[endidxcnt];
|
|
236
|
1126
|
|
300
|
1127 #ifdef FEAT_MBYTE
|
|
|
1128 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1129 continue; /* not at first byte of character */
|
|
|
1130 #endif
|
|
376
|
1131 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
481
|
1132 {
|
|
483
|
1133 if (slang->sl_compprog == NULL)
|
|
481
|
1134 continue; /* next char is a word character */
|
|
|
1135 word_ends = FALSE;
|
|
|
1136 }
|
|
|
1137 else
|
|
|
1138 word_ends = TRUE;
|
|
300
|
1139
|
|
|
1140 #ifdef FEAT_MBYTE
|
|
339
|
1141 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1142 {
|
|
|
1143 /* Compute byte length in original word, length may change
|
|
339
|
1144 * when folding case. This can be slow, take a shortcut when the
|
|
|
1145 * case-folded word is equal to the keep-case word. */
|
|
300
|
1146 p = mip->mi_word;
|
|
339
|
1147 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1148 {
|
|
|
1149 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1150 mb_ptr_adv(p);
|
|
|
1151 wlen = p - mip->mi_word;
|
|
|
1152 }
|
|
300
|
1153 }
|
|
|
1154 #endif
|
|
236
|
1155
|
|
339
|
1156 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1157 * prefix ID.
|
|
|
1158 * Repeat this if there are more flags/region alternatives until there
|
|
|
1159 * is a match. */
|
|
|
1160 res = SP_BAD;
|
|
|
1161 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1162 --len, ++arridx)
|
|
300
|
1163 {
|
|
|
1164 flags = idxs[arridx];
|
|
324
|
1165
|
|
339
|
1166 /* For the fold-case tree check that the case of the checked word
|
|
|
1167 * matches with what the word in the tree requires.
|
|
|
1168 * For keep-case tree the case is always right. For prefixes we
|
|
|
1169 * don't bother to check. */
|
|
|
1170 if (mode == FIND_FOLDWORD)
|
|
300
|
1171 {
|
|
|
1172 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1173 {
|
|
323
|
1174 /* mi_capflags was set for a different word length, need
|
|
|
1175 * to do it again. */
|
|
300
|
1176 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1177 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1178 }
|
|
|
1179
|
|
346
|
1180 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1181 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1182 continue;
|
|
300
|
1183 }
|
|
236
|
1184
|
|
339
|
1185 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1186 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1187 * mip->mi_prefarridx that find_prefix() filled. */
|
|
481
|
1188 else if (mode == FIND_PREFIX)
|
|
|
1189 {
|
|
366
|
1190 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1191 flags,
|
|
455
|
1192 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1193 FALSE);
|
|
366
|
1194 if (c == 0)
|
|
339
|
1195 continue;
|
|
366
|
1196
|
|
|
1197 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1198 if (c & WF_RAREPFX)
|
|
|
1199 flags |= WF_RARE;
|
|
339
|
1200 }
|
|
|
1201
|
|
481
|
1202 if (mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND
|
|
|
1203 || !word_ends)
|
|
|
1204 {
|
|
483
|
1205 /* If there is no flag or the word is shorter than
|
|
|
1206 * COMPOUNDMIN reject it quickly.
|
|
|
1207 * Makes you wonder why someone puts a compound flag on a word
|
|
481
|
1208 * that's too short... Myspell compatibility requires this
|
|
|
1209 * anyway. */
|
|
483
|
1210 if (((unsigned)flags >> 24) == 0 || wlen < slang->sl_compminlen)
|
|
|
1211 continue;
|
|
|
1212
|
|
|
1213 /* Limit the number of compound words to COMPOUNDMAX. */
|
|
|
1214 if (!word_ends && mip->mi_complen + 2 > slang->sl_compmax)
|
|
|
1215 continue;
|
|
|
1216
|
|
|
1217 /* At start of word quickly check if compounding is possible
|
|
|
1218 * with this flag. */
|
|
|
1219 if (mip->mi_complen == 0
|
|
|
1220 && vim_strchr(slang->sl_compstartflags,
|
|
|
1221 ((unsigned)flags >> 24)) == NULL)
|
|
481
|
1222 continue;
|
|
|
1223
|
|
483
|
1224 /* If the word ends the sequence of compound flags of the
|
|
|
1225 * words must match with one of the COMPOUNDFLAGS items and
|
|
|
1226 * the number of syllables must not be too large. */
|
|
|
1227 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24);
|
|
|
1228 mip->mi_compflags[mip->mi_complen + 1] = NUL;
|
|
|
1229 if (word_ends)
|
|
|
1230 {
|
|
|
1231 char_u fword[MAXWLEN];
|
|
|
1232
|
|
|
1233 if (slang->sl_compsylmax < MAXWLEN)
|
|
|
1234 {
|
|
|
1235 /* "fword" is only needed for checking syllables. */
|
|
|
1236 if (ptr == mip->mi_word)
|
|
|
1237 (void)spell_casefold(ptr, wlen, fword, MAXWLEN);
|
|
|
1238 else
|
|
|
1239 vim_strncpy(fword, ptr, endlen[endidxcnt]);
|
|
|
1240 }
|
|
|
1241 if (!can_compound(slang, fword, mip->mi_compflags))
|
|
|
1242 continue;
|
|
|
1243 }
|
|
481
|
1244 }
|
|
|
1245
|
|
|
1246 if (!word_ends)
|
|
|
1247 {
|
|
|
1248 /* Check that a valid word follows. If there is one, it will
|
|
|
1249 * set "mi_result", thus we are always finished here.
|
|
|
1250 * Recursive! */
|
|
|
1251
|
|
|
1252 /* Find following word in case-folded tree. */
|
|
|
1253 mip->mi_compoff = endlen[endidxcnt];
|
|
|
1254 #ifdef FEAT_MBYTE
|
|
|
1255 if (has_mbyte && mode == FIND_KEEPWORD)
|
|
|
1256 {
|
|
|
1257 /* Compute byte length in case-folded word from "wlen":
|
|
|
1258 * byte length in keep-case word. Length may change when
|
|
|
1259 * folding case. This can be slow, take a shortcut when
|
|
|
1260 * the case-folded word is equal to the keep-case word. */
|
|
|
1261 p = mip->mi_fword;
|
|
|
1262 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1263 {
|
|
|
1264 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1265 mb_ptr_adv(p);
|
|
|
1266 mip->mi_compoff = p - mip->mi_fword;
|
|
|
1267 }
|
|
|
1268 }
|
|
|
1269 #endif
|
|
483
|
1270 ++mip->mi_complen;
|
|
481
|
1271 find_word(mip, FIND_COMPOUND);
|
|
483
|
1272 --mip->mi_complen;
|
|
481
|
1273 if (mip->mi_result == SP_OK)
|
|
|
1274 break;
|
|
|
1275
|
|
|
1276 /* Find following word in keep-case tree. */
|
|
|
1277 mip->mi_compoff = wlen;
|
|
483
|
1278 ++mip->mi_complen;
|
|
481
|
1279 find_word(mip, FIND_KEEPCOMPOUND);
|
|
483
|
1280 --mip->mi_complen;
|
|
481
|
1281 if (mip->mi_result == SP_OK)
|
|
|
1282 break;
|
|
|
1283 continue;
|
|
|
1284 }
|
|
|
1285
|
|
339
|
1286 if (flags & WF_BANNED)
|
|
|
1287 res = SP_BANNED;
|
|
|
1288 else if (flags & WF_REGION)
|
|
|
1289 {
|
|
|
1290 /* Check region. */
|
|
390
|
1291 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1292 res = SP_OK;
|
|
339
|
1293 else
|
|
|
1294 res = SP_LOCAL;
|
|
300
|
1295 }
|
|
339
|
1296 else if (flags & WF_RARE)
|
|
|
1297 res = SP_RARE;
|
|
307
|
1298 else
|
|
339
|
1299 res = SP_OK;
|
|
|
1300
|
|
|
1301 /* Always use the longest match and the best result. */
|
|
|
1302 if (mip->mi_result > res)
|
|
|
1303 {
|
|
|
1304 mip->mi_result = res;
|
|
|
1305 mip->mi_end = mip->mi_word + wlen;
|
|
|
1306 }
|
|
351
|
1307 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1308 mip->mi_end = mip->mi_word + wlen;
|
|
|
1309
|
|
|
1310 if (res == SP_OK)
|
|
|
1311 break;
|
|
300
|
1312 }
|
|
|
1313
|
|
307
|
1314 if (res == SP_OK)
|
|
300
|
1315 break;
|
|
|
1316 }
|
|
236
|
1317 }
|
|
|
1318
|
|
323
|
1319 /*
|
|
483
|
1320 * Return TRUE if "flags" is a valid sequence of compound flags and
|
|
|
1321 * "word[len]" does not have too many syllables.
|
|
482
|
1322 */
|
|
|
1323 static int
|
|
483
|
1324 can_compound(slang, word, flags)
|
|
482
|
1325 slang_T *slang;
|
|
483
|
1326 char_u *word;
|
|
|
1327 char_u *flags;
|
|
|
1328 {
|
|
|
1329 regmatch_T regmatch;
|
|
|
1330
|
|
|
1331 if (slang->sl_compprog == NULL)
|
|
|
1332 return FALSE;
|
|
|
1333 regmatch.regprog = slang->sl_compprog;
|
|
|
1334 regmatch.rm_ic = FALSE;
|
|
|
1335 if (!vim_regexec(®match, flags, 0))
|
|
|
1336 return FALSE;
|
|
|
1337
|
|
|
1338 /* Count the number of syllables. This may be slow, do it last. */
|
|
|
1339 if (slang->sl_compsylmax < MAXWLEN
|
|
|
1340 && count_syllables(slang, word) > slang->sl_compsylmax)
|
|
|
1341 return FALSE;
|
|
|
1342 return TRUE;
|
|
482
|
1343 }
|
|
|
1344
|
|
|
1345 /*
|
|
390
|
1346 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1347 * ID in "flags" for the word "word".
|
|
366
|
1348 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1349 */
|
|
|
1350 static int
|
|
455
|
1351 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1352 int totprefcnt; /* nr of prefix IDs */
|
|
|
1353 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1354 int flags;
|
|
351
|
1355 char_u *word;
|
|
|
1356 slang_T *slang;
|
|
455
|
1357 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1358 {
|
|
|
1359 int prefcnt;
|
|
|
1360 int pidx;
|
|
|
1361 regprog_T *rp;
|
|
|
1362 regmatch_T regmatch;
|
|
390
|
1363 int prefid;
|
|
|
1364
|
|
|
1365 prefid = (unsigned)flags >> 24;
|
|
351
|
1366 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1367 {
|
|
|
1368 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1369
|
|
|
1370 /* Check the prefix ID. */
|
|
|
1371 if (prefid != (pidx & 0xff))
|
|
|
1372 continue;
|
|
|
1373
|
|
390
|
1374 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1375 * suffix. */
|
|
|
1376 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1377 continue;
|
|
|
1378
|
|
351
|
1379 /* Check the condition, if there is one. The condition index is
|
|
366
|
1380 * stored in the two bytes above the prefix ID byte. */
|
|
|
1381 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1382 if (rp != NULL)
|
|
|
1383 {
|
|
|
1384 regmatch.regprog = rp;
|
|
|
1385 regmatch.rm_ic = FALSE;
|
|
|
1386 if (!vim_regexec(®match, word, 0))
|
|
|
1387 continue;
|
|
|
1388 }
|
|
455
|
1389 else if (cond_req)
|
|
|
1390 continue;
|
|
|
1391
|
|
|
1392 /* It's a match! Return the WF_ flags. */
|
|
366
|
1393 return pidx;
|
|
|
1394 }
|
|
|
1395 return 0;
|
|
351
|
1396 }
|
|
|
1397
|
|
|
1398 /*
|
|
339
|
1399 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1400 * If it does, then check the following word.
|
|
|
1401 *
|
|
|
1402 * For a match mip->mi_result is updated.
|
|
|
1403 */
|
|
|
1404 static void
|
|
|
1405 find_prefix(mip)
|
|
|
1406 matchinf_T *mip;
|
|
|
1407 {
|
|
|
1408 idx_T arridx = 0;
|
|
|
1409 int len;
|
|
|
1410 int wlen = 0;
|
|
|
1411 int flen;
|
|
|
1412 int c;
|
|
|
1413 char_u *ptr;
|
|
|
1414 idx_T lo, hi, m;
|
|
|
1415 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1416 char_u *byts;
|
|
|
1417 idx_T *idxs;
|
|
|
1418
|
|
375
|
1419 byts = slang->sl_pbyts;
|
|
|
1420 if (byts == NULL)
|
|
|
1421 return; /* array is empty */
|
|
|
1422
|
|
339
|
1423 /* We use the case-folded word here, since prefixes are always
|
|
|
1424 * case-folded. */
|
|
|
1425 ptr = mip->mi_fword;
|
|
|
1426 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1427 idxs = slang->sl_pidxs;
|
|
|
1428
|
|
|
1429 /*
|
|
|
1430 * Repeat advancing in the tree until:
|
|
|
1431 * - there is a byte that doesn't match,
|
|
|
1432 * - we reach the end of the tree,
|
|
|
1433 * - or we reach the end of the line.
|
|
|
1434 */
|
|
|
1435 for (;;)
|
|
|
1436 {
|
|
|
1437 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1438 flen = fold_more(mip);
|
|
|
1439
|
|
|
1440 len = byts[arridx++];
|
|
|
1441
|
|
|
1442 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1443 * Check if the following word matches and supports the prefix. */
|
|
|
1444 if (byts[arridx] == 0)
|
|
|
1445 {
|
|
|
1446 /* There can be several prefixes with different conditions. We
|
|
|
1447 * try them all, since we don't know which one will give the
|
|
|
1448 * longest match. The word is the same each time, pass the list
|
|
|
1449 * of possible prefixes to find_word(). */
|
|
|
1450 mip->mi_prefarridx = arridx;
|
|
|
1451 mip->mi_prefcnt = len;
|
|
|
1452 while (len > 0 && byts[arridx] == 0)
|
|
|
1453 {
|
|
|
1454 ++arridx;
|
|
|
1455 --len;
|
|
|
1456 }
|
|
|
1457 mip->mi_prefcnt -= len;
|
|
|
1458
|
|
|
1459 /* Find the word that comes after the prefix. */
|
|
|
1460 mip->mi_prefixlen = wlen;
|
|
455
|
1461 #ifdef FEAT_MBYTE
|
|
|
1462 if (has_mbyte)
|
|
|
1463 {
|
|
|
1464 /* Case-folded length may differ from original length. */
|
|
|
1465 mip->mi_cprefixlen = nofold_len(mip->mi_fword, wlen,
|
|
|
1466 mip->mi_word);
|
|
|
1467 }
|
|
|
1468 else
|
|
|
1469 mip->mi_cprefixlen = wlen;
|
|
|
1470 #endif
|
|
339
|
1471 find_word(mip, FIND_PREFIX);
|
|
|
1472
|
|
|
1473
|
|
|
1474 if (len == 0)
|
|
|
1475 break; /* no children, word must end here */
|
|
|
1476 }
|
|
|
1477
|
|
|
1478 /* Stop looking at end of the line. */
|
|
|
1479 if (ptr[wlen] == NUL)
|
|
|
1480 break;
|
|
|
1481
|
|
|
1482 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1483 c = ptr[wlen];
|
|
|
1484 lo = arridx;
|
|
|
1485 hi = arridx + len - 1;
|
|
|
1486 while (lo < hi)
|
|
|
1487 {
|
|
|
1488 m = (lo + hi) / 2;
|
|
|
1489 if (byts[m] > c)
|
|
|
1490 hi = m - 1;
|
|
|
1491 else if (byts[m] < c)
|
|
|
1492 lo = m + 1;
|
|
|
1493 else
|
|
|
1494 {
|
|
|
1495 lo = hi = m;
|
|
|
1496 break;
|
|
|
1497 }
|
|
|
1498 }
|
|
|
1499
|
|
|
1500 /* Stop if there is no matching byte. */
|
|
|
1501 if (hi < lo || byts[lo] != c)
|
|
|
1502 break;
|
|
|
1503
|
|
|
1504 /* Continue at the child (if there is one). */
|
|
|
1505 arridx = idxs[lo];
|
|
|
1506 ++wlen;
|
|
|
1507 --flen;
|
|
|
1508 }
|
|
|
1509 }
|
|
|
1510
|
|
|
1511 /*
|
|
|
1512 * Need to fold at least one more character. Do until next non-word character
|
|
|
1513 * for efficiency.
|
|
|
1514 * Return the length of the folded chars in bytes.
|
|
|
1515 */
|
|
|
1516 static int
|
|
|
1517 fold_more(mip)
|
|
|
1518 matchinf_T *mip;
|
|
|
1519 {
|
|
|
1520 int flen;
|
|
|
1521 char_u *p;
|
|
|
1522
|
|
|
1523 p = mip->mi_fend;
|
|
|
1524 do
|
|
|
1525 {
|
|
|
1526 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1527 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1528
|
|
|
1529 /* Include the non-word character so that we can check for the
|
|
|
1530 * word end. */
|
|
|
1531 if (*mip->mi_fend != NUL)
|
|
|
1532 mb_ptr_adv(mip->mi_fend);
|
|
|
1533
|
|
|
1534 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1535 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1536 MAXWLEN - mip->mi_fwordlen);
|
|
|
1537 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1538 mip->mi_fwordlen += flen;
|
|
|
1539 return flen;
|
|
|
1540 }
|
|
|
1541
|
|
|
1542 /*
|
|
323
|
1543 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1544 * case.
|
|
|
1545 */
|
|
|
1546 static int
|
|
389
|
1547 spell_valid_case(wordflags, treeflags)
|
|
|
1548 int wordflags; /* flags for the checked word. */
|
|
323
|
1549 int treeflags; /* flags for the word in the spell tree */
|
|
|
1550 {
|
|
389
|
1551 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1552 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1553 && ((treeflags & WF_ONECAP) == 0
|
|
|
1554 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1555 }
|
|
|
1556
|
|
351
|
1557 /*
|
|
|
1558 * Return TRUE if spell checking is not enabled.
|
|
|
1559 */
|
|
|
1560 static int
|
|
|
1561 no_spell_checking()
|
|
|
1562 {
|
|
|
1563 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
1564 {
|
|
|
1565 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1566 return TRUE;
|
|
|
1567 }
|
|
|
1568 return FALSE;
|
|
|
1569 }
|
|
300
|
1570
|
|
236
|
1571 /*
|
|
|
1572 * Move to next spell error.
|
|
323
|
1573 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
|
483
|
1574 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move
|
|
|
1575 * to after badly spelled word before the cursor.
|
|
236
|
1576 * Return OK if found, FAIL otherwise.
|
|
|
1577 */
|
|
|
1578 int
|
|
323
|
1579 spell_move_to(dir, allwords, curline)
|
|
236
|
1580 int dir; /* FORWARD or BACKWARD */
|
|
|
1581 int allwords; /* TRUE for "[s" and "]s" */
|
|
323
|
1582 int curline;
|
|
236
|
1583 {
|
|
249
|
1584 linenr_T lnum;
|
|
|
1585 pos_T found_pos;
|
|
236
|
1586 char_u *line;
|
|
|
1587 char_u *p;
|
|
346
|
1588 char_u *endp;
|
|
|
1589 int attr;
|
|
236
|
1590 int len;
|
|
249
|
1591 int has_syntax = syntax_present(curbuf);
|
|
|
1592 int col;
|
|
|
1593 int can_spell;
|
|
346
|
1594 char_u *buf = NULL;
|
|
|
1595 int buflen = 0;
|
|
|
1596 int skip = 0;
|
|
385
|
1597 int capcol = -1;
|
|
236
|
1598
|
|
351
|
1599 if (no_spell_checking())
|
|
236
|
1600 return FAIL;
|
|
|
1601
|
|
249
|
1602 /*
|
|
|
1603 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1604 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1605 *
|
|
|
1606 * When searching backwards, we continue in the line to find the last
|
|
|
1607 * bad word (in the cursor line: before the cursor).
|
|
346
|
1608 *
|
|
|
1609 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1610 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1611 * though...
|
|
249
|
1612 */
|
|
|
1613 lnum = curwin->w_cursor.lnum;
|
|
|
1614 found_pos.lnum = 0;
|
|
236
|
1615
|
|
|
1616 while (!got_int)
|
|
|
1617 {
|
|
249
|
1618 line = ml_get(lnum);
|
|
346
|
1619
|
|
|
1620 len = STRLEN(line);
|
|
|
1621 if (buflen < len + MAXWLEN + 2)
|
|
|
1622 {
|
|
|
1623 vim_free(buf);
|
|
|
1624 buflen = len + MAXWLEN + 2;
|
|
|
1625 buf = alloc(buflen);
|
|
|
1626 if (buf == NULL)
|
|
|
1627 break;
|
|
|
1628 }
|
|
|
1629
|
|
385
|
1630 /* In first line check first word for Capital. */
|
|
|
1631 if (lnum == 1)
|
|
|
1632 capcol = 0;
|
|
|
1633
|
|
|
1634 /* For checking first word with a capital skip white space. */
|
|
|
1635 if (capcol == 0)
|
|
|
1636 capcol = skipwhite(line) - line;
|
|
|
1637
|
|
346
|
1638 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1639 * possible. */
|
|
|
1640 STRCPY(buf, line);
|
|
|
1641 if (lnum < curbuf->b_ml.ml_line_count)
|
|
|
1642 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1643
|
|
|
1644 p = buf + skip;
|
|
|
1645 endp = buf + len;
|
|
|
1646 while (p < endp)
|
|
236
|
1647 {
|
|
300
|
1648 /* When searching backward don't search after the cursor. */
|
|
|
1649 if (dir == BACKWARD
|
|
|
1650 && lnum == curwin->w_cursor.lnum
|
|
346
|
1651 && (colnr_T)(p - buf) >= curwin->w_cursor.col)
|
|
300
|
1652 break;
|
|
249
|
1653
|
|
300
|
1654 /* start of word */
|
|
346
|
1655 attr = 0;
|
|
385
|
1656 len = spell_check(curwin, p, &attr, &capcol);
|
|
249
|
1657
|
|
300
|
1658 if (attr != 0)
|
|
|
1659 {
|
|
|
1660 /* We found a bad word. Check the attribute. */
|
|
|
1661 if (allwords || attr == highlight_attr[HLF_SPB])
|
|
236
|
1662 {
|
|
300
|
1663 /* When searching forward only accept a bad word after
|
|
|
1664 * the cursor. */
|
|
|
1665 if (dir == BACKWARD
|
|
|
1666 || lnum > curwin->w_cursor.lnum
|
|
|
1667 || (lnum == curwin->w_cursor.lnum
|
|
346
|
1668 && (colnr_T)(curline ? p - buf + len
|
|
|
1669 : p - buf)
|
|
300
|
1670 > curwin->w_cursor.col))
|
|
236
|
1671 {
|
|
300
|
1672 if (has_syntax)
|
|
249
|
1673 {
|
|
346
|
1674 col = p - buf;
|
|
300
|
1675 (void)syn_get_id(lnum, (colnr_T)col,
|
|
|
1676 FALSE, &can_spell);
|
|
|
1677 }
|
|
|
1678 else
|
|
|
1679 can_spell = TRUE;
|
|
249
|
1680
|
|
300
|
1681 if (can_spell)
|
|
|
1682 {
|
|
|
1683 found_pos.lnum = lnum;
|
|
346
|
1684 found_pos.col = p - buf;
|
|
249
|
1685 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1686 found_pos.coladd = 0;
|
|
249
|
1687 #endif
|
|
300
|
1688 if (dir == FORWARD)
|
|
|
1689 {
|
|
|
1690 /* No need to search further. */
|
|
|
1691 curwin->w_cursor = found_pos;
|
|
346
|
1692 vim_free(buf);
|
|
300
|
1693 return OK;
|
|
249
|
1694 }
|
|
483
|
1695 else if (curline)
|
|
|
1696 /* Insert mode completion: put cursor after
|
|
|
1697 * the bad word. */
|
|
|
1698 found_pos.col += len;
|
|
249
|
1699 }
|
|
236
|
1700 }
|
|
|
1701 }
|
|
|
1702 }
|
|
|
1703
|
|
300
|
1704 /* advance to character after the word */
|
|
|
1705 p += len;
|
|
385
|
1706 capcol -= len;
|
|
236
|
1707 }
|
|
|
1708
|
|
483
|
1709 if (dir == BACKWARD && found_pos.lnum != 0)
|
|
|
1710 {
|
|
|
1711 /* Use the last match in the line. */
|
|
|
1712 curwin->w_cursor = found_pos;
|
|
|
1713 vim_free(buf);
|
|
|
1714 return OK;
|
|
|
1715 }
|
|
|
1716
|
|
323
|
1717 if (curline)
|
|
346
|
1718 break; /* only check cursor line */
|
|
323
|
1719
|
|
236
|
1720 /* Advance to next line. */
|
|
249
|
1721 if (dir == BACKWARD)
|
|
|
1722 {
|
|
|
1723 if (lnum == 1)
|
|
346
|
1724 break;
|
|
249
|
1725 --lnum;
|
|
385
|
1726 capcol = -1;
|
|
249
|
1727 }
|
|
|
1728 else
|
|
|
1729 {
|
|
|
1730 if (lnum == curbuf->b_ml.ml_line_count)
|
|
346
|
1731 break;
|
|
249
|
1732 ++lnum;
|
|
346
|
1733
|
|
|
1734 /* Skip the characters at the start of the next line that were
|
|
|
1735 * included in a match crossing line boundaries. */
|
|
|
1736 if (attr == 0)
|
|
|
1737 skip = p - endp;
|
|
|
1738 else
|
|
|
1739 skip = 0;
|
|
385
|
1740
|
|
|
1741 /* Capscol skips over the inserted space. */
|
|
|
1742 --capcol;
|
|
|
1743
|
|
|
1744 /* But after empty line check first word in next line */
|
|
|
1745 if (*skipwhite(line) == NUL)
|
|
|
1746 capcol = 0;
|
|
249
|
1747 }
|
|
236
|
1748
|
|
|
1749 line_breakcheck();
|
|
|
1750 }
|
|
|
1751
|
|
346
|
1752 vim_free(buf);
|
|
|
1753 return FAIL;
|
|
|
1754 }
|
|
|
1755
|
|
|
1756 /*
|
|
|
1757 * For spell checking: concatenate the start of the following line "line" into
|
|
|
1758 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
1759 */
|
|
|
1760 void
|
|
|
1761 spell_cat_line(buf, line, maxlen)
|
|
|
1762 char_u *buf;
|
|
|
1763 char_u *line;
|
|
|
1764 int maxlen;
|
|
|
1765 {
|
|
|
1766 char_u *p;
|
|
|
1767 int n;
|
|
|
1768
|
|
|
1769 p = skipwhite(line);
|
|
|
1770 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
1771 p = skipwhite(p + 1);
|
|
|
1772
|
|
|
1773 if (*p != NUL)
|
|
|
1774 {
|
|
|
1775 *buf = ' ';
|
|
376
|
1776 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
1777 n = p - line;
|
|
|
1778 if (n >= maxlen)
|
|
|
1779 n = maxlen - 1;
|
|
|
1780 vim_memset(buf + 1, ' ', n);
|
|
|
1781 }
|
|
236
|
1782 }
|
|
|
1783
|
|
|
1784 /*
|
|
307
|
1785 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
1786 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
1787 */
|
|
307
|
1788 static void
|
|
236
|
1789 spell_load_lang(lang)
|
|
|
1790 char_u *lang;
|
|
|
1791 {
|
|
310
|
1792 char_u fname_enc[85];
|
|
236
|
1793 int r;
|
|
307
|
1794 char_u langcp[MAXWLEN + 1];
|
|
|
1795
|
|
310
|
1796 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
1797 * It's truncated when an error is detected. */
|
|
|
1798 STRCPY(langcp, lang);
|
|
|
1799
|
|
310
|
1800 /*
|
|
|
1801 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
1802 */
|
|
|
1803 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
1804 "spell/%s.%s.spl", lang, spell_enc());
|
|
|
1805 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1806
|
|
|
1807 if (r == FAIL && *langcp != NUL)
|
|
|
1808 {
|
|
|
1809 /* Try loading the ASCII version. */
|
|
310
|
1810 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
1811 "spell/%s.ascii.spl", lang);
|
|
310
|
1812 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
|
307
|
1813 }
|
|
|
1814
|
|
|
1815 if (r == FAIL)
|
|
483
|
1816 smsg((char_u *)_("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""),
|
|
|
1817 lang, spell_enc(), lang);
|
|
310
|
1818 else if (*langcp != NUL)
|
|
|
1819 {
|
|
|
1820 /* Load all the additions. */
|
|
|
1821 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
|
1822 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
|
1823 }
|
|
|
1824 }
|
|
|
1825
|
|
|
1826 /*
|
|
|
1827 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
1828 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
1829 */
|
|
|
1830 static char_u *
|
|
|
1831 spell_enc()
|
|
|
1832 {
|
|
|
1833
|
|
|
1834 #ifdef FEAT_MBYTE
|
|
|
1835 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
1836 return p_enc;
|
|
|
1837 #endif
|
|
|
1838 return (char_u *)"latin1";
|
|
236
|
1839 }
|
|
|
1840
|
|
|
1841 /*
|
|
385
|
1842 * Get the name of the .spl file for the internal wordlist into
|
|
|
1843 * "fname[MAXPATHL]".
|
|
|
1844 */
|
|
|
1845 static void
|
|
|
1846 int_wordlist_spl(fname)
|
|
|
1847 char_u *fname;
|
|
|
1848 {
|
|
|
1849 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
1850 int_wordlist, spell_enc());
|
|
|
1851 }
|
|
|
1852
|
|
|
1853 /*
|
|
236
|
1854 * Allocate a new slang_T.
|
|
|
1855 * Caller must fill "sl_next".
|
|
|
1856 */
|
|
|
1857 static slang_T *
|
|
|
1858 slang_alloc(lang)
|
|
|
1859 char_u *lang;
|
|
|
1860 {
|
|
|
1861 slang_T *lp;
|
|
|
1862
|
|
300
|
1863 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
1864 if (lp != NULL)
|
|
|
1865 {
|
|
|
1866 lp->sl_name = vim_strsave(lang);
|
|
323
|
1867 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
483
|
1868 lp->sl_compmax = MAXWLEN;
|
|
|
1869 lp->sl_compminlen = MAXWLEN;
|
|
|
1870 lp->sl_compsylmax = MAXWLEN;
|
|
236
|
1871 }
|
|
|
1872 return lp;
|
|
|
1873 }
|
|
|
1874
|
|
|
1875 /*
|
|
|
1876 * Free the contents of an slang_T and the structure itself.
|
|
|
1877 */
|
|
|
1878 static void
|
|
|
1879 slang_free(lp)
|
|
|
1880 slang_T *lp;
|
|
|
1881 {
|
|
|
1882 vim_free(lp->sl_name);
|
|
310
|
1883 vim_free(lp->sl_fname);
|
|
|
1884 slang_clear(lp);
|
|
|
1885 vim_free(lp);
|
|
|
1886 }
|
|
|
1887
|
|
|
1888 /*
|
|
|
1889 * Clear an slang_T so that the file can be reloaded.
|
|
|
1890 */
|
|
|
1891 static void
|
|
|
1892 slang_clear(lp)
|
|
|
1893 slang_T *lp;
|
|
|
1894 {
|
|
339
|
1895 garray_T *gap;
|
|
|
1896 fromto_T *ftp;
|
|
344
|
1897 salitem_T *smp;
|
|
339
|
1898 int i;
|
|
323
|
1899
|
|
300
|
1900 vim_free(lp->sl_fbyts);
|
|
310
|
1901 lp->sl_fbyts = NULL;
|
|
300
|
1902 vim_free(lp->sl_kbyts);
|
|
310
|
1903 lp->sl_kbyts = NULL;
|
|
339
|
1904 vim_free(lp->sl_pbyts);
|
|
|
1905 lp->sl_pbyts = NULL;
|
|
|
1906
|
|
300
|
1907 vim_free(lp->sl_fidxs);
|
|
310
|
1908 lp->sl_fidxs = NULL;
|
|
300
|
1909 vim_free(lp->sl_kidxs);
|
|
310
|
1910 lp->sl_kidxs = NULL;
|
|
339
|
1911 vim_free(lp->sl_pidxs);
|
|
|
1912 lp->sl_pidxs = NULL;
|
|
323
|
1913
|
|
344
|
1914 gap = &lp->sl_rep;
|
|
|
1915 while (gap->ga_len > 0)
|
|
323
|
1916 {
|
|
344
|
1917 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1918 vim_free(ftp->ft_from);
|
|
|
1919 vim_free(ftp->ft_to);
|
|
323
|
1920 }
|
|
344
|
1921 ga_clear(gap);
|
|
|
1922
|
|
|
1923 gap = &lp->sl_sal;
|
|
375
|
1924 if (lp->sl_sofo)
|
|
376
|
1925 {
|
|
|
1926 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
1927 if (gap->ga_data != NULL)
|
|
|
1928 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
1929 for (i = 0; i < gap->ga_len; ++i)
|
|
|
1930 vim_free(((int **)gap->ga_data)[i]);
|
|
|
1931 }
|
|
375
|
1932 else
|
|
|
1933 /* SAL items: free salitem_T items */
|
|
|
1934 while (gap->ga_len > 0)
|
|
|
1935 {
|
|
|
1936 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
1937 vim_free(smp->sm_lead);
|
|
|
1938 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
1939 vim_free(smp->sm_to);
|
|
|
1940 #ifdef FEAT_MBYTE
|
|
|
1941 vim_free(smp->sm_lead_w);
|
|
|
1942 vim_free(smp->sm_oneof_w);
|
|
|
1943 vim_free(smp->sm_to_w);
|
|
|
1944 #endif
|
|
|
1945 }
|
|
344
|
1946 ga_clear(gap);
|
|
323
|
1947
|
|
339
|
1948 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
1949 vim_free(lp->sl_prefprog[i]);
|
|
376
|
1950 lp->sl_prefixcnt = 0;
|
|
339
|
1951 vim_free(lp->sl_prefprog);
|
|
376
|
1952 lp->sl_prefprog = NULL;
|
|
|
1953
|
|
|
1954 vim_free(lp->sl_midword);
|
|
|
1955 lp->sl_midword = NULL;
|
|
339
|
1956
|
|
483
|
1957 vim_free(lp->sl_compprog);
|
|
|
1958 vim_free(lp->sl_compstartflags);
|
|
|
1959 lp->sl_compprog = NULL;
|
|
|
1960 lp->sl_compstartflags = NULL;
|
|
|
1961
|
|
|
1962 vim_free(lp->sl_syllable);
|
|
|
1963 lp->sl_syllable = NULL;
|
|
|
1964 ga_clear(&lp->sl_syl_items);
|
|
481
|
1965
|
|
330
|
1966 #ifdef FEAT_MBYTE
|
|
|
1967 {
|
|
|
1968 int todo = lp->sl_map_hash.ht_used;
|
|
|
1969 hashitem_T *hi;
|
|
|
1970
|
|
|
1971 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
1972 if (!HASHITEM_EMPTY(hi))
|
|
|
1973 {
|
|
|
1974 --todo;
|
|
|
1975 vim_free(hi->hi_key);
|
|
|
1976 }
|
|
|
1977 }
|
|
|
1978 hash_clear(&lp->sl_map_hash);
|
|
|
1979 #endif
|
|
483
|
1980
|
|
|
1981 lp->sl_compmax = MAXWLEN;
|
|
|
1982 lp->sl_compminlen = MAXWLEN;
|
|
|
1983 lp->sl_compsylmax = MAXWLEN;
|
|
|
1984 lp->sl_regions[0] = NUL;
|
|
236
|
1985 }
|
|
|
1986
|
|
|
1987 /*
|
|
307
|
1988 * Load one spell file and store the info into a slang_T.
|
|
236
|
1989 * Invoked through do_in_runtimepath().
|
|
|
1990 */
|
|
|
1991 static void
|
|
310
|
1992 spell_load_cb(fname, cookie)
|
|
236
|
1993 char_u *fname;
|
|
307
|
1994 void *cookie; /* points to the language name */
|
|
236
|
1995 {
|
|
323
|
1996 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
|
310
|
1997 }
|
|
|
1998
|
|
|
1999 /*
|
|
|
2000 * Load one spell file and store the info into a slang_T.
|
|
|
2001 *
|
|
|
2002 * This is invoked in two ways:
|
|
|
2003 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
2004 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
2005 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
2006 * points to the existing slang_T.
|
|
323
|
2007 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
2008 */
|
|
323
|
2009 static slang_T *
|
|
|
2010 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
2011 char_u *fname;
|
|
|
2012 char_u *lang;
|
|
|
2013 slang_T *old_lp;
|
|
323
|
2014 int silent; /* no error if file doesn't exist */
|
|
310
|
2015 {
|
|
236
|
2016 FILE *fd;
|
|
483
|
2017 char_u buf[VIMSPELLMAGICL];
|
|
236
|
2018 char_u *p;
|
|
339
|
2019 char_u *bp;
|
|
|
2020 idx_T *ip;
|
|
236
|
2021 int i;
|
|
339
|
2022 int n;
|
|
300
|
2023 int len;
|
|
236
|
2024 int round;
|
|
|
2025 char_u *save_sourcing_name = sourcing_name;
|
|
|
2026 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
307
|
2027 slang_T *lp = NULL;
|
|
324
|
2028 idx_T idx;
|
|
344
|
2029 int c = 0;
|
|
483
|
2030 int res;
|
|
236
|
2031
|
|
310
|
2032 fd = mch_fopen((char *)fname, "r");
|
|
236
|
2033 if (fd == NULL)
|
|
|
2034 {
|
|
323
|
2035 if (!silent)
|
|
|
2036 EMSG2(_(e_notopen), fname);
|
|
|
2037 else if (p_verbose > 2)
|
|
|
2038 {
|
|
|
2039 verbose_enter();
|
|
|
2040 smsg((char_u *)e_notopen, fname);
|
|
|
2041 verbose_leave();
|
|
|
2042 }
|
|
255
|
2043 goto endFAIL;
|
|
236
|
2044 }
|
|
310
|
2045 if (p_verbose > 2)
|
|
|
2046 {
|
|
|
2047 verbose_enter();
|
|
|
2048 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
2049 verbose_leave();
|
|
|
2050 }
|
|
|
2051
|
|
|
2052 if (old_lp == NULL)
|
|
|
2053 {
|
|
|
2054 lp = slang_alloc(lang);
|
|
|
2055 if (lp == NULL)
|
|
|
2056 goto endFAIL;
|
|
|
2057
|
|
|
2058 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
2059 lp->sl_fname = vim_strsave(fname);
|
|
|
2060 if (lp->sl_fname == NULL)
|
|
|
2061 goto endFAIL;
|
|
|
2062
|
|
|
2063 /* Check for .add.spl. */
|
|
|
2064 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
2065 }
|
|
|
2066 else
|
|
|
2067 lp = old_lp;
|
|
307
|
2068
|
|
236
|
2069 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
2070 sourcing_name = fname;
|
|
|
2071 sourcing_lnum = 0;
|
|
|
2072
|
|
339
|
2073 /* <HEADER>: <fileID>
|
|
|
2074 */
|
|
236
|
2075 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
2076 buf[i] = getc(fd); /* <fileID> */
|
|
|
2077 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
2078 {
|
|
483
|
2079 EMSG(_("E757: This does not look like a spell file"));
|
|
381
|
2080 goto endFAIL;
|
|
483
|
2081 }
|
|
|
2082 c = getc(fd); /* <versionnr> */
|
|
|
2083 if (c < VIMSPELLVERSION)
|
|
|
2084 {
|
|
|
2085 EMSG(_("E771: Old spell file, needs to be updated"));
|
|
381
|
2086 goto endFAIL;
|
|
|
2087 }
|
|
483
|
2088 else if (c > VIMSPELLVERSION)
|
|
|
2089 {
|
|
|
2090 EMSG(_("E772: Spell file is for newer version of Vim"));
|
|
481
|
2091 goto endFAIL;
|
|
483
|
2092 }
|
|
|
2093
|
|
|
2094
|
|
|
2095 /*
|
|
|
2096 * <SECTIONS>: <section> ... <sectionend>
|
|
|
2097 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
2098 */
|
|
|
2099 for (;;)
|
|
|
2100 {
|
|
|
2101 n = getc(fd); /* <sectionID> or <sectionend> */
|
|
|
2102 if (n == SN_END)
|
|
|
2103 break;
|
|
|
2104 c = getc(fd); /* <sectionflags> */
|
|
|
2105 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2106 /* <sectionlen> */
|
|
|
2107 if (len < 0)
|
|
|
2108 goto truncerr;
|
|
|
2109
|
|
|
2110 res = 0;
|
|
|
2111 switch (n)
|
|
|
2112 {
|
|
|
2113 case SN_REGION:
|
|
|
2114 res = read_region_section(fd, lp, len);
|
|
|
2115 break;
|
|
|
2116
|
|
|
2117 case SN_CHARFLAGS:
|
|
|
2118 res = read_charflags_section(fd);
|
|
|
2119 break;
|
|
|
2120
|
|
|
2121 case SN_MIDWORD:
|
|
|
2122 lp->sl_midword = read_string(fd, len); /* <midword> */
|
|
|
2123 if (lp->sl_midword == NULL)
|
|
|
2124 goto endFAIL;
|
|
|
2125 break;
|
|
|
2126
|
|
|
2127 case SN_PREFCOND:
|
|
|
2128 res = read_prefcond_section(fd, lp);
|
|
|
2129 break;
|
|
|
2130
|
|
|
2131 case SN_REP:
|
|
|
2132 res = read_rep_section(fd, lp);
|
|
|
2133 break;
|
|
|
2134
|
|
|
2135 case SN_SAL:
|
|
|
2136 res = read_sal_section(fd, lp);
|
|
|
2137 break;
|
|
|
2138
|
|
|
2139 case SN_SOFO:
|
|
|
2140 res = read_sofo_section(fd, lp);
|
|
|
2141 break;
|
|
|
2142
|
|
|
2143 case SN_MAP:
|
|
|
2144 p = read_string(fd, len); /* <mapstr> */
|
|
|
2145 if (p == NULL)
|
|
|
2146 goto endFAIL;
|
|
|
2147 set_map_str(lp, p);
|
|
|
2148 vim_free(p);
|
|
|
2149 break;
|
|
|
2150
|
|
|
2151 case SN_COMPOUND:
|
|
|
2152 res = read_compound(fd, lp, len);
|
|
|
2153 break;
|
|
|
2154
|
|
|
2155 case SN_SYLLABLE:
|
|
|
2156 lp->sl_syllable = read_string(fd, len); /* <syllable> */
|
|
|
2157 if (lp->sl_syllable == NULL)
|
|
|
2158 goto endFAIL;
|
|
|
2159 if (init_syl_tab(lp) == FAIL)
|
|
|
2160 goto endFAIL;
|
|
|
2161 break;
|
|
|
2162
|
|
|
2163 default:
|
|
|
2164 /* Unsupported section. When it's required give an error
|
|
|
2165 * message. When it's not required skip the contents. */
|
|
|
2166 if (c & SNF_REQUIRED)
|
|
|
2167 {
|
|
|
2168 EMSG(_("E770: Unsupported section in spell file"));
|
|
|
2169 goto endFAIL;
|
|
|
2170 }
|
|
|
2171 while (--len >= 0)
|
|
|
2172 if (getc(fd) < 0)
|
|
|
2173 goto truncerr;
|
|
|
2174 break;
|
|
|
2175 }
|
|
|
2176 if (res == SP_FORMERROR)
|
|
|
2177 {
|
|
|
2178 formerr:
|
|
|
2179 EMSG(_(e_format));
|
|
381
|
2180 goto endFAIL;
|
|
344
|
2181 }
|
|
483
|
2182 if (res == SP_TRUNCERROR)
|
|
|
2183 {
|
|
|
2184 truncerr:
|
|
|
2185 EMSG(_(e_spell_trunc));
|
|
375
|
2186 goto endFAIL;
|
|
483
|
2187 }
|
|
|
2188 if (res == SP_ERROR)
|
|
|
2189 goto endFAIL;
|
|
389
|
2190 }
|
|
236
|
2191
|
|
300
|
2192 /* round 1: <LWORDTREE>
|
|
339
|
2193 * round 2: <KWORDTREE>
|
|
|
2194 * round 3: <PREFIXTREE> */
|
|
|
2195 for (round = 1; round <= 3; ++round)
|
|
236
|
2196 {
|
|
300
|
2197 /* The tree size was computed when writing the file, so that we can
|
|
|
2198 * allocate it as one long block. <nodecount> */
|
|
|
2199 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2200 if (len < 0)
|
|
|
2201 goto truncerr;
|
|
|
2202 if (len > 0)
|
|
236
|
2203 {
|
|
300
|
2204 /* Allocate the byte array. */
|
|
339
|
2205 bp = lalloc((long_u)len, TRUE);
|
|
|
2206 if (bp == NULL)
|
|
300
|
2207 goto endFAIL;
|
|
|
2208 if (round == 1)
|
|
339
|
2209 lp->sl_fbyts = bp;
|
|
|
2210 else if (round == 2)
|
|
|
2211 lp->sl_kbyts = bp;
|
|
300
|
2212 else
|
|
339
|
2213 lp->sl_pbyts = bp;
|
|
236
|
2214
|
|
300
|
2215 /* Allocate the index array. */
|
|
339
|
2216 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2217 if (ip == NULL)
|
|
300
|
2218 goto endFAIL;
|
|
|
2219 if (round == 1)
|
|
339
|
2220 lp->sl_fidxs = ip;
|
|
|
2221 else if (round == 2)
|
|
|
2222 lp->sl_kidxs = ip;
|
|
300
|
2223 else
|
|
339
|
2224 lp->sl_pidxs = ip;
|
|
300
|
2225
|
|
|
2226 /* Read the tree and store it in the array. */
|
|
339
|
2227 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2228 if (idx == -1)
|
|
300
|
2229 goto truncerr;
|
|
324
|
2230 if (idx < 0)
|
|
236
|
2231 goto formerr;
|
|
|
2232 }
|
|
300
|
2233 }
|
|
243
|
2234
|
|
310
|
2235 /* For a new file link it in the list of spell files. */
|
|
|
2236 if (old_lp == NULL)
|
|
|
2237 {
|
|
|
2238 lp->sl_next = first_lang;
|
|
|
2239 first_lang = lp;
|
|
|
2240 }
|
|
307
|
2241
|
|
255
|
2242 goto endOK;
|
|
|
2243
|
|
|
2244 endFAIL:
|
|
310
|
2245 if (lang != NULL)
|
|
|
2246 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2247 *lang = NUL;
|
|
|
2248 if (lp != NULL && old_lp == NULL)
|
|
323
|
2249 {
|
|
307
|
2250 slang_free(lp);
|
|
323
|
2251 lp = NULL;
|
|
|
2252 }
|
|
255
|
2253
|
|
|
2254 endOK:
|
|
236
|
2255 if (fd != NULL)
|
|
|
2256 fclose(fd);
|
|
|
2257 sourcing_name = save_sourcing_name;
|
|
|
2258 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2259
|
|
|
2260 return lp;
|
|
236
|
2261 }
|
|
|
2262
|
|
376
|
2263 /*
|
|
|
2264 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2265 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2266 * Returns NULL when the count is zero.
|
|
483
|
2267 * Sets "*cntp" to SP_*ERROR when there is an error, length of the result
|
|
|
2268 * otherwise.
|
|
376
|
2269 */
|
|
|
2270 static char_u *
|
|
389
|
2271 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2272 FILE *fd;
|
|
|
2273 int cnt_bytes;
|
|
389
|
2274 int *cntp;
|
|
376
|
2275 {
|
|
|
2276 int cnt = 0;
|
|
|
2277 int i;
|
|
|
2278 char_u *str;
|
|
|
2279
|
|
|
2280 /* read the length bytes, MSB first */
|
|
|
2281 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2282 cnt = (cnt << 8) + getc(fd);
|
|
|
2283 if (cnt < 0)
|
|
|
2284 {
|
|
483
|
2285 *cntp = SP_TRUNCERROR;
|
|
376
|
2286 return NULL;
|
|
|
2287 }
|
|
389
|
2288 *cntp = cnt;
|
|
|
2289 if (cnt == 0)
|
|
|
2290 return NULL; /* nothing to read, return NULL */
|
|
376
|
2291
|
|
483
|
2292 str = read_string(fd, cnt);
|
|
|
2293 if (str == NULL)
|
|
|
2294 *cntp = SP_ERROR;
|
|
|
2295 return str;
|
|
|
2296 }
|
|
|
2297
|
|
|
2298 /*
|
|
|
2299 * Read a string of length "cnt" from "fd" into allocated memory.
|
|
|
2300 * Returns NULL when out of memory.
|
|
|
2301 */
|
|
|
2302 static char_u *
|
|
|
2303 read_string(fd, cnt)
|
|
|
2304 FILE *fd;
|
|
|
2305 int cnt;
|
|
|
2306 {
|
|
|
2307 char_u *str;
|
|
|
2308 int i;
|
|
|
2309
|
|
376
|
2310 /* allocate memory */
|
|
|
2311 str = alloc((unsigned)cnt + 1);
|
|
483
|
2312 if (str != NULL)
|
|
|
2313 {
|
|
|
2314 /* Read the string. Doesn't check for truncated file. */
|
|
|
2315 for (i = 0; i < cnt; ++i)
|
|
|
2316 str[i] = getc(fd);
|
|
|
2317 str[i] = NUL;
|
|
|
2318 }
|
|
|
2319 return str;
|
|
|
2320 }
|
|
|
2321
|
|
|
2322 /*
|
|
|
2323 * Read SN_REGION: <regionname> ...
|
|
|
2324 * Return SP_*ERROR flags.
|
|
|
2325 */
|
|
|
2326 static int
|
|
|
2327 read_region_section(fd, lp, len)
|
|
|
2328 FILE *fd;
|
|
|
2329 slang_T *lp;
|
|
|
2330 int len;
|
|
|
2331 {
|
|
|
2332 int i;
|
|
|
2333
|
|
|
2334 if (len > 16)
|
|
|
2335 return SP_FORMERROR;
|
|
|
2336 for (i = 0; i < len; ++i)
|
|
|
2337 lp->sl_regions[i] = getc(fd); /* <regionname> */
|
|
|
2338 lp->sl_regions[len] = NUL;
|
|
|
2339 return 0;
|
|
|
2340 }
|
|
|
2341
|
|
|
2342 /*
|
|
|
2343 * Read SN_CHARFLAGS section: <charflagslen> <charflags>
|
|
|
2344 * <folcharslen> <folchars>
|
|
|
2345 * Return SP_*ERROR flags.
|
|
|
2346 */
|
|
|
2347 static int
|
|
|
2348 read_charflags_section(fd)
|
|
|
2349 FILE *fd;
|
|
|
2350 {
|
|
|
2351 char_u *flags;
|
|
|
2352 char_u *fol;
|
|
|
2353 int flagslen, follen;
|
|
|
2354
|
|
|
2355 /* <charflagslen> <charflags> */
|
|
|
2356 flags = read_cnt_string(fd, 1, &flagslen);
|
|
|
2357 if (flagslen < 0)
|
|
|
2358 return flagslen;
|
|
|
2359
|
|
|
2360 /* <folcharslen> <folchars> */
|
|
|
2361 fol = read_cnt_string(fd, 2, &follen);
|
|
|
2362 if (follen < 0)
|
|
|
2363 {
|
|
|
2364 vim_free(flags);
|
|
|
2365 return follen;
|
|
|
2366 }
|
|
|
2367
|
|
|
2368 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
2369 if (flags != NULL && fol != NULL)
|
|
|
2370 set_spell_charflags(flags, flagslen, fol);
|
|
|
2371
|
|
|
2372 vim_free(flags);
|
|
|
2373 vim_free(fol);
|
|
|
2374
|
|
|
2375 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
2376 if ((flags == NULL) != (fol == NULL))
|
|
|
2377 return SP_FORMERROR;
|
|
|
2378 return 0;
|
|
|
2379 }
|
|
|
2380
|
|
|
2381 /*
|
|
|
2382 * Read SN_PREFCOND section.
|
|
|
2383 * Return SP_*ERROR flags.
|
|
|
2384 */
|
|
|
2385 static int
|
|
|
2386 read_prefcond_section(fd, lp)
|
|
|
2387 FILE *fd;
|
|
|
2388 slang_T *lp;
|
|
|
2389 {
|
|
|
2390 int cnt;
|
|
|
2391 int i;
|
|
|
2392 int n;
|
|
|
2393 char_u *p;
|
|
|
2394 char_u buf[MAXWLEN + 1];
|
|
|
2395
|
|
|
2396 /* <prefcondcnt> <prefcond> ... */
|
|
|
2397 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
2398 if (cnt <= 0)
|
|
|
2399 return SP_FORMERROR;
|
|
|
2400
|
|
|
2401 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
2402 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
2403 if (lp->sl_prefprog == NULL)
|
|
|
2404 return SP_ERROR;
|
|
|
2405 lp->sl_prefixcnt = cnt;
|
|
|
2406
|
|
376
|
2407 for (i = 0; i < cnt; ++i)
|
|
483
|
2408 {
|
|
|
2409 /* <prefcond> : <condlen> <condstr> */
|
|
|
2410 n = getc(fd); /* <condlen> */
|
|
|
2411 if (n < 0 || n >= MAXWLEN)
|
|
|
2412 return SP_FORMERROR;
|
|
|
2413
|
|
|
2414 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
2415 * compile the regexp program used to check for the condition. */
|
|
|
2416 if (n > 0)
|
|
|
2417 {
|
|
|
2418 buf[0] = '^'; /* always match at one position only */
|
|
|
2419 p = buf + 1;
|
|
|
2420 while (n-- > 0)
|
|
|
2421 *p++ = getc(fd); /* <condstr> */
|
|
|
2422 *p = NUL;
|
|
|
2423 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
2424 }
|
|
|
2425 }
|
|
|
2426 return 0;
|
|
|
2427 }
|
|
|
2428
|
|
|
2429 /*
|
|
|
2430 * Read REP items section from "fd": <repcount> <rep> ...
|
|
|
2431 * Return SP_*ERROR flags.
|
|
|
2432 */
|
|
|
2433 static int
|
|
|
2434 read_rep_section(fd, slang)
|
|
|
2435 FILE *fd;
|
|
|
2436 slang_T *slang;
|
|
|
2437 {
|
|
|
2438 int cnt;
|
|
|
2439 garray_T *gap;
|
|
|
2440 fromto_T *ftp;
|
|
|
2441 short *first;
|
|
|
2442 int i;
|
|
|
2443
|
|
|
2444 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
2445 if (cnt < 0)
|
|
|
2446 return SP_TRUNCERROR;
|
|
|
2447
|
|
|
2448 gap = &slang->sl_rep;
|
|
|
2449 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2450 return SP_ERROR;
|
|
|
2451
|
|
|
2452 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
2453 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2454 {
|
|
|
2455 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
|
2456 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
|
2457 if (i < 0)
|
|
|
2458 return i;
|
|
|
2459 if (i == 0)
|
|
|
2460 return SP_FORMERROR;
|
|
|
2461 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
|
2462 if (i <= 0)
|
|
|
2463 {
|
|
|
2464 vim_free(ftp->ft_from);
|
|
|
2465 if (i < 0)
|
|
|
2466 return i;
|
|
|
2467 return SP_FORMERROR;
|
|
|
2468 }
|
|
|
2469 }
|
|
|
2470
|
|
|
2471 /* Fill the first-index table. */
|
|
|
2472 first = slang->sl_rep_first;
|
|
|
2473 for (i = 0; i < 256; ++i)
|
|
|
2474 first[i] = -1;
|
|
|
2475 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2476 {
|
|
|
2477 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
2478 if (first[*ftp->ft_from] == -1)
|
|
|
2479 first[*ftp->ft_from] = i;
|
|
|
2480 }
|
|
|
2481 return 0;
|
|
|
2482 }
|
|
|
2483
|
|
|
2484 /*
|
|
|
2485 * Read SN_SAL section: <salflags> <salcount> <sal> ...
|
|
|
2486 * Return SP_*ERROR flags.
|
|
|
2487 */
|
|
|
2488 static int
|
|
|
2489 read_sal_section(fd, slang)
|
|
|
2490 FILE *fd;
|
|
|
2491 slang_T *slang;
|
|
|
2492 {
|
|
|
2493 int i;
|
|
|
2494 int cnt;
|
|
|
2495 garray_T *gap;
|
|
|
2496 salitem_T *smp;
|
|
|
2497 int ccnt;
|
|
|
2498 char_u *p;
|
|
|
2499 int c;
|
|
|
2500
|
|
|
2501 slang->sl_sofo = FALSE;
|
|
|
2502
|
|
|
2503 i = getc(fd); /* <salflags> */
|
|
|
2504 if (i & SAL_F0LLOWUP)
|
|
|
2505 slang->sl_followup = TRUE;
|
|
|
2506 if (i & SAL_COLLAPSE)
|
|
|
2507 slang->sl_collapse = TRUE;
|
|
|
2508 if (i & SAL_REM_ACCENTS)
|
|
|
2509 slang->sl_rem_accents = TRUE;
|
|
|
2510
|
|
|
2511 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2512 if (cnt < 0)
|
|
|
2513 return SP_TRUNCERROR;
|
|
|
2514
|
|
|
2515 gap = &slang->sl_sal;
|
|
|
2516 ga_init2(gap, sizeof(salitem_T), 10);
|
|
|
2517 if (ga_grow(gap, cnt) == FAIL)
|
|
|
2518 return SP_ERROR;
|
|
|
2519
|
|
|
2520 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2521 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2522 {
|
|
|
2523 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2524 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2525 if (ccnt < 0)
|
|
|
2526 return SP_TRUNCERROR;
|
|
|
2527 if ((p = alloc(ccnt + 2)) == NULL)
|
|
|
2528 return SP_ERROR;
|
|
|
2529 smp->sm_lead = p;
|
|
|
2530
|
|
|
2531 /* Read up to the first special char into sm_lead. */
|
|
|
2532 for (i = 0; i < ccnt; ++i)
|
|
|
2533 {
|
|
|
2534 c = getc(fd); /* <salfrom> */
|
|
|
2535 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2536 break;
|
|
|
2537 *p++ = c;
|
|
|
2538 }
|
|
|
2539 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2540 *p++ = NUL;
|
|
|
2541
|
|
|
2542 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2543 if (c == '(')
|
|
|
2544 {
|
|
|
2545 smp->sm_oneof = p;
|
|
|
2546 for (++i; i < ccnt; ++i)
|
|
|
2547 {
|
|
|
2548 c = getc(fd); /* <salfrom> */
|
|
|
2549 if (c == ')')
|
|
|
2550 break;
|
|
|
2551 *p++ = c;
|
|
|
2552 }
|
|
|
2553 *p++ = NUL;
|
|
|
2554 if (++i < ccnt)
|
|
|
2555 c = getc(fd);
|
|
|
2556 }
|
|
|
2557 else
|
|
|
2558 smp->sm_oneof = NULL;
|
|
|
2559
|
|
|
2560 /* Any following chars go in sm_rules. */
|
|
|
2561 smp->sm_rules = p;
|
|
|
2562 if (i < ccnt)
|
|
|
2563 /* store the char we got while checking for end of sm_lead */
|
|
|
2564 *p++ = c;
|
|
|
2565 for (++i; i < ccnt; ++i)
|
|
|
2566 *p++ = getc(fd); /* <salfrom> */
|
|
|
2567 *p++ = NUL;
|
|
|
2568
|
|
|
2569 /* <saltolen> <salto> */
|
|
|
2570 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
|
2571 if (ccnt < 0)
|
|
|
2572 {
|
|
|
2573 vim_free(smp->sm_lead);
|
|
|
2574 return ccnt;
|
|
|
2575 }
|
|
|
2576
|
|
|
2577 #ifdef FEAT_MBYTE
|
|
|
2578 if (has_mbyte)
|
|
|
2579 {
|
|
|
2580 /* convert the multi-byte strings to wide char strings */
|
|
|
2581 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2582 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2583 if (smp->sm_oneof == NULL)
|
|
|
2584 smp->sm_oneof_w = NULL;
|
|
|
2585 else
|
|
|
2586 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
2587 if (smp->sm_to == NULL)
|
|
|
2588 smp->sm_to_w = NULL;
|
|
|
2589 else
|
|
|
2590 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
2591 if (smp->sm_lead_w == NULL
|
|
|
2592 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
2593 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
|
2594 {
|
|
|
2595 vim_free(smp->sm_lead);
|
|
|
2596 vim_free(smp->sm_to);
|
|
|
2597 vim_free(smp->sm_lead_w);
|
|
|
2598 vim_free(smp->sm_oneof_w);
|
|
|
2599 vim_free(smp->sm_to_w);
|
|
|
2600 return SP_ERROR;
|
|
|
2601 }
|
|
|
2602 }
|
|
|
2603 #endif
|
|
|
2604 }
|
|
|
2605
|
|
|
2606 /* Fill the first-index table. */
|
|
|
2607 set_sal_first(slang);
|
|
|
2608
|
|
|
2609 return 0;
|
|
|
2610 }
|
|
|
2611
|
|
|
2612 /*
|
|
|
2613 * SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
2614 * Return SP_*ERROR flags.
|
|
|
2615 */
|
|
|
2616 static int
|
|
|
2617 read_sofo_section(fd, slang)
|
|
|
2618 FILE *fd;
|
|
|
2619 slang_T *slang;
|
|
|
2620 {
|
|
|
2621 int cnt;
|
|
|
2622 char_u *from, *to;
|
|
|
2623 int res;
|
|
|
2624
|
|
|
2625 slang->sl_sofo = TRUE;
|
|
|
2626
|
|
|
2627 /* <sofofromlen> <sofofrom> */
|
|
|
2628 from = read_cnt_string(fd, 2, &cnt);
|
|
|
2629 if (cnt < 0)
|
|
|
2630 return cnt;
|
|
|
2631
|
|
|
2632 /* <sofotolen> <sofoto> */
|
|
|
2633 to = read_cnt_string(fd, 2, &cnt);
|
|
|
2634 if (cnt < 0)
|
|
|
2635 {
|
|
|
2636 vim_free(from);
|
|
|
2637 return cnt;
|
|
|
2638 }
|
|
|
2639
|
|
|
2640 /* Store the info in slang->sl_sal and/or slang->sl_sal_first. */
|
|
|
2641 if (from != NULL && to != NULL)
|
|
|
2642 res = set_sofo(slang, from, to);
|
|
|
2643 else if (from != NULL || to != NULL)
|
|
|
2644 res = SP_FORMERROR; /* only one of two strings is an error */
|
|
|
2645 else
|
|
|
2646 res = 0;
|
|
|
2647
|
|
|
2648 vim_free(from);
|
|
|
2649 vim_free(to);
|
|
|
2650 return res;
|
|
|
2651 }
|
|
|
2652
|
|
|
2653 /*
|
|
|
2654 * Read the compound section from the .spl file:
|
|
|
2655 * <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
2656 * Returns SP_*ERROR flags.
|
|
|
2657 */
|
|
|
2658 static int
|
|
|
2659 read_compound(fd, slang, len)
|
|
|
2660 FILE *fd;
|
|
|
2661 slang_T *slang;
|
|
|
2662 int len;
|
|
|
2663 {
|
|
|
2664 int todo = len;
|
|
|
2665 int c;
|
|
|
2666 int atstart;
|
|
|
2667 char_u *pat;
|
|
|
2668 char_u *pp;
|
|
|
2669 char_u *cp;
|
|
|
2670
|
|
|
2671 if (todo < 2)
|
|
|
2672 return SP_FORMERROR; /* need at least two bytes */
|
|
|
2673
|
|
|
2674 --todo;
|
|
|
2675 c = getc(fd); /* <compmax> */
|
|
|
2676 if (c < 2)
|
|
|
2677 c = MAXWLEN;
|
|
|
2678 slang->sl_compmax = c;
|
|
|
2679
|
|
|
2680 --todo;
|
|
|
2681 c = getc(fd); /* <compminlen> */
|
|
|
2682 if (c < 1)
|
|
|
2683 c = 3;
|
|
|
2684 slang->sl_compminlen = c;
|
|
|
2685
|
|
|
2686 --todo;
|
|
|
2687 c = getc(fd); /* <compsylmax> */
|
|
|
2688 if (c < 1)
|
|
|
2689 c = MAXWLEN;
|
|
|
2690 slang->sl_compsylmax = c;
|
|
|
2691
|
|
|
2692 /* Turn the COMPOUNDFLAGS items into a regexp pattern:
|
|
|
2693 * "a[bc]/a*b+" -> "^\(a[bc]\|a*b\+\)$".
|
|
|
2694 * Inserting backslashes may double the length, "^\(\)$<Nul>" is 7 bytes. */
|
|
|
2695 pat = alloc((unsigned)todo * 2 + 7);
|
|
|
2696 if (pat == NULL)
|
|
|
2697 return SP_ERROR;
|
|
|
2698
|
|
|
2699 /* We also need a list of all flags that can appear at the start. */
|
|
|
2700 cp = alloc(todo + 1);
|
|
|
2701 if (cp == NULL)
|
|
|
2702 {
|
|
|
2703 vim_free(pat);
|
|
|
2704 return SP_ERROR;
|
|
|
2705 }
|
|
|
2706 slang->sl_compstartflags = cp;
|
|
|
2707 *cp = NUL;
|
|
|
2708
|
|
|
2709 pp = pat;
|
|
|
2710 *pp++ = '^';
|
|
|
2711 *pp++ = '\\';
|
|
|
2712 *pp++ = '(';
|
|
|
2713
|
|
|
2714 atstart = 1;
|
|
|
2715 while (todo-- > 0)
|
|
|
2716 {
|
|
|
2717 c = getc(fd); /* <compflags> */
|
|
|
2718 if (atstart != 0)
|
|
|
2719 {
|
|
|
2720 /* At start of item: copy flags to "sl_compstartflags". For a
|
|
|
2721 * [abc] item set "atstart" to 2 and copy up to the ']'. */
|
|
|
2722 if (c == '[')
|
|
|
2723 atstart = 2;
|
|
|
2724 else if (c == ']')
|
|
|
2725 atstart = 0;
|
|
|
2726 else
|
|
|
2727 {
|
|
|
2728 if (vim_strchr(slang->sl_compstartflags, c) == NULL)
|
|
|
2729 {
|
|
|
2730 *cp++ = c;
|
|
|
2731 *cp = NUL;
|
|
|
2732 }
|
|
|
2733 if (atstart == 1)
|
|
|
2734 atstart = 0;
|
|
|
2735 }
|
|
|
2736 }
|
|
|
2737 if (c == '/') /* slash separates two items */
|
|
|
2738 {
|
|
|
2739 *pp++ = '\\';
|
|
|
2740 *pp++ = '|';
|
|
|
2741 atstart = 1;
|
|
|
2742 }
|
|
|
2743 else /* normal char, "[abc]" and '*' are copied as-is */
|
|
|
2744 {
|
|
|
2745 if (c == '+')
|
|
|
2746 *pp++ = '\\'; /* "a+" becomes "a\+" */
|
|
|
2747 *pp++ = c;
|
|
|
2748 }
|
|
|
2749 }
|
|
|
2750
|
|
|
2751 *pp++ = '\\';
|
|
|
2752 *pp++ = ')';
|
|
|
2753 *pp++ = '$';
|
|
|
2754 *pp = NUL;
|
|
|
2755
|
|
|
2756 slang->sl_compprog = vim_regcomp(pat, RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
2757 vim_free(pat);
|
|
|
2758 if (slang->sl_compprog == NULL)
|
|
|
2759 return SP_FORMERROR;
|
|
|
2760
|
|
|
2761 return 0;
|
|
|
2762 }
|
|
|
2763
|
|
|
2764 #define SY_MAXLEN 30
|
|
|
2765 typedef struct syl_item_S
|
|
|
2766 {
|
|
|
2767 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */
|
|
|
2768 int sy_len;
|
|
|
2769 } syl_item_T;
|
|
|
2770
|
|
|
2771 /*
|
|
|
2772 * Truncate "slang->sl_syllable" at the first slash and put the following items
|
|
|
2773 * in "slang->sl_syl_items".
|
|
|
2774 */
|
|
|
2775 static int
|
|
|
2776 init_syl_tab(slang)
|
|
|
2777 slang_T *slang;
|
|
|
2778 {
|
|
|
2779 char_u *p;
|
|
|
2780 char_u *s;
|
|
|
2781 int l;
|
|
|
2782 syl_item_T *syl;
|
|
|
2783
|
|
|
2784 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4);
|
|
|
2785 p = vim_strchr(slang->sl_syllable, '/');
|
|
|
2786 while (p != NULL)
|
|
|
2787 {
|
|
|
2788 *p++ = NUL;
|
|
|
2789 if (p == NUL)
|
|
|
2790 break;
|
|
|
2791 s = p;
|
|
|
2792 p = vim_strchr(p, '/');
|
|
|
2793 if (p == NULL)
|
|
|
2794 l = STRLEN(s);
|
|
|
2795 else
|
|
|
2796 l = p - s;
|
|
|
2797 if (l >= SY_MAXLEN)
|
|
|
2798 return SP_FORMERROR;
|
|
|
2799 if (ga_grow(&slang->sl_syl_items, 1) == FAIL)
|
|
|
2800 return SP_ERROR;
|
|
|
2801 syl = ((syl_item_T *)slang->sl_syl_items.ga_data)
|
|
|
2802 + slang->sl_syl_items.ga_len++;
|
|
|
2803 vim_strncpy(syl->sy_chars, s, l);
|
|
|
2804 syl->sy_len = l;
|
|
|
2805 }
|
|
|
2806 return OK;
|
|
|
2807 }
|
|
|
2808
|
|
|
2809 /*
|
|
|
2810 * Count the number of syllables in "word".
|
|
|
2811 * When "word" contains spaces the syllables after the last space are counted.
|
|
|
2812 * Returns zero if syllables are not defines.
|
|
|
2813 */
|
|
|
2814 static int
|
|
|
2815 count_syllables(slang, word)
|
|
|
2816 slang_T *slang;
|
|
|
2817 char_u *word;
|
|
|
2818 {
|
|
|
2819 int cnt = 0;
|
|
|
2820 int skip = FALSE;
|
|
|
2821 char_u *p;
|
|
|
2822 int len;
|
|
|
2823 int i;
|
|
|
2824 syl_item_T *syl;
|
|
|
2825 int c;
|
|
|
2826
|
|
|
2827 if (slang->sl_syllable == NULL)
|
|
|
2828 return 0;
|
|
|
2829
|
|
|
2830 for (p = word; *p != NUL; p += len)
|
|
|
2831 {
|
|
|
2832 /* When running into a space reset counter. */
|
|
|
2833 if (*p == ' ')
|
|
|
2834 {
|
|
|
2835 len = 1;
|
|
|
2836 cnt = 0;
|
|
|
2837 continue;
|
|
|
2838 }
|
|
|
2839
|
|
|
2840 /* Find longest match of syllable items. */
|
|
|
2841 len = 0;
|
|
|
2842 for (i = 0; i < slang->sl_syl_items.ga_len; ++i)
|
|
|
2843 {
|
|
|
2844 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i;
|
|
|
2845 if (syl->sy_len > len
|
|
|
2846 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0)
|
|
|
2847 len = syl->sy_len;
|
|
|
2848 }
|
|
|
2849 if (len != 0) /* found a match, count syllable */
|
|
|
2850 {
|
|
|
2851 ++cnt;
|
|
|
2852 skip = FALSE;
|
|
|
2853 }
|
|
|
2854 else
|
|
|
2855 {
|
|
|
2856 /* No recognized syllable item, at least a syllable char then? */
|
|
|
2857 #ifdef FEAT_MBYTE
|
|
|
2858 c = mb_ptr2char(p);
|
|
|
2859 len = (*mb_ptr2len)(p);
|
|
|
2860 #else
|
|
|
2861 c = *p;
|
|
|
2862 len = 1;
|
|
|
2863 #endif
|
|
|
2864 if (vim_strchr(slang->sl_syllable, c) == NULL)
|
|
|
2865 skip = FALSE; /* No, search for next syllable */
|
|
|
2866 else if (!skip)
|
|
|
2867 {
|
|
|
2868 ++cnt; /* Yes, count it */
|
|
|
2869 skip = TRUE; /* don't count following syllable chars */
|
|
|
2870 }
|
|
|
2871 }
|
|
|
2872 }
|
|
|
2873 return cnt;
|
|
376
|
2874 }
|
|
|
2875
|
|
381
|
2876 /*
|
|
|
2877 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
483
|
2878 * Returns SP_*ERROR flags when there is something wrong.
|
|
381
|
2879 */
|
|
|
2880 static int
|
|
|
2881 set_sofo(lp, from, to)
|
|
|
2882 slang_T *lp;
|
|
|
2883 char_u *from;
|
|
|
2884 char_u *to;
|
|
|
2885 {
|
|
|
2886 int i;
|
|
|
2887
|
|
|
2888 #ifdef FEAT_MBYTE
|
|
|
2889 garray_T *gap;
|
|
|
2890 char_u *s;
|
|
|
2891 char_u *p;
|
|
|
2892 int c;
|
|
|
2893 int *inp;
|
|
|
2894
|
|
|
2895 if (has_mbyte)
|
|
|
2896 {
|
|
|
2897 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
2898 * characters. The index is the low byte of the character.
|
|
|
2899 * The list contains from-to pairs with a terminating NUL.
|
|
|
2900 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
2901 gap = &lp->sl_sal;
|
|
|
2902 ga_init2(gap, sizeof(int *), 1);
|
|
|
2903 if (ga_grow(gap, 256) == FAIL)
|
|
483
|
2904 return SP_ERROR;
|
|
381
|
2905 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
2906 gap->ga_len = 256;
|
|
|
2907
|
|
|
2908 /* First count the number of items for each list. Temporarily use
|
|
|
2909 * sl_sal_first[] for this. */
|
|
|
2910 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2911 {
|
|
474
|
2912 c = mb_cptr2char_adv(&p);
|
|
|
2913 mb_cptr_adv(s);
|
|
381
|
2914 if (c >= 256)
|
|
|
2915 ++lp->sl_sal_first[c & 0xff];
|
|
|
2916 }
|
|
|
2917 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
483
|
2918 return SP_FORMERROR;
|
|
381
|
2919
|
|
|
2920 /* Allocate the lists. */
|
|
|
2921 for (i = 0; i < 256; ++i)
|
|
|
2922 if (lp->sl_sal_first[i] > 0)
|
|
|
2923 {
|
|
|
2924 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
2925 if (p == NULL)
|
|
483
|
2926 return SP_ERROR;
|
|
381
|
2927 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
2928 *(int *)p = 0;
|
|
|
2929 }
|
|
|
2930
|
|
|
2931 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
2932 * list. */
|
|
|
2933 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
2934 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
2935 {
|
|
474
|
2936 c = mb_cptr2char_adv(&p);
|
|
|
2937 i = mb_cptr2char_adv(&s);
|
|
381
|
2938 if (c >= 256)
|
|
|
2939 {
|
|
|
2940 /* Append the from-to chars at the end of the list with
|
|
|
2941 * the low byte. */
|
|
|
2942 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
2943 while (*inp != 0)
|
|
|
2944 ++inp;
|
|
|
2945 *inp++ = c; /* from char */
|
|
|
2946 *inp++ = i; /* to char */
|
|
|
2947 *inp++ = NUL; /* NUL at the end */
|
|
|
2948 }
|
|
|
2949 else
|
|
|
2950 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
2951 lp->sl_sal_first[c] = i;
|
|
|
2952 }
|
|
|
2953 }
|
|
|
2954 else
|
|
|
2955 #endif
|
|
|
2956 {
|
|
|
2957 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
2958 if (STRLEN(from) != STRLEN(to))
|
|
483
|
2959 return SP_FORMERROR;
|
|
381
|
2960
|
|
|
2961 for (i = 0; to[i] != NUL; ++i)
|
|
|
2962 lp->sl_sal_first[from[i]] = to[i];
|
|
|
2963 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
2964 }
|
|
|
2965
|
|
483
|
2966 return 0;
|
|
381
|
2967 }
|
|
|
2968
|
|
|
2969 /*
|
|
|
2970 * Fill the first-index table for "lp".
|
|
|
2971 */
|
|
|
2972 static void
|
|
|
2973 set_sal_first(lp)
|
|
|
2974 slang_T *lp;
|
|
|
2975 {
|
|
|
2976 salfirst_T *sfirst;
|
|
|
2977 int i;
|
|
|
2978 salitem_T *smp;
|
|
|
2979 int c;
|
|
|
2980 garray_T *gap = &lp->sl_sal;
|
|
|
2981
|
|
|
2982 sfirst = lp->sl_sal_first;
|
|
|
2983 for (i = 0; i < 256; ++i)
|
|
|
2984 sfirst[i] = -1;
|
|
|
2985 smp = (salitem_T *)gap->ga_data;
|
|
|
2986 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2987 {
|
|
|
2988 #ifdef FEAT_MBYTE
|
|
|
2989 if (has_mbyte)
|
|
|
2990 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
2991 * the character, for other encodings it should differ for most
|
|
|
2992 * characters. */
|
|
|
2993 c = *smp[i].sm_lead_w & 0xff;
|
|
|
2994 else
|
|
|
2995 #endif
|
|
|
2996 c = *smp[i].sm_lead;
|
|
|
2997 if (sfirst[c] == -1)
|
|
|
2998 {
|
|
|
2999 sfirst[c] = i;
|
|
|
3000 #ifdef FEAT_MBYTE
|
|
|
3001 if (has_mbyte)
|
|
|
3002 {
|
|
|
3003 int n;
|
|
|
3004
|
|
|
3005 /* Make sure all entries with this byte are following each
|
|
|
3006 * other. Move the ones that are in the wrong position. Do
|
|
|
3007 * keep the same ordering! */
|
|
|
3008 while (i + 1 < gap->ga_len
|
|
|
3009 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
3010 /* Skip over entry with same index byte. */
|
|
|
3011 ++i;
|
|
|
3012
|
|
|
3013 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
3014 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
3015 {
|
|
|
3016 salitem_T tsal;
|
|
|
3017
|
|
|
3018 /* Move entry with same index byte after the entries
|
|
|
3019 * we already found. */
|
|
|
3020 ++i;
|
|
|
3021 --n;
|
|
|
3022 tsal = smp[i + n];
|
|
|
3023 mch_memmove(smp + i + 1, smp + i,
|
|
|
3024 sizeof(salitem_T) * n);
|
|
|
3025 smp[i] = tsal;
|
|
|
3026 }
|
|
|
3027 }
|
|
|
3028 #endif
|
|
|
3029 }
|
|
|
3030 }
|
|
|
3031 }
|
|
376
|
3032
|
|
372
|
3033 #ifdef FEAT_MBYTE
|
|
|
3034 /*
|
|
|
3035 * Turn a multi-byte string into a wide character string.
|
|
|
3036 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
3037 */
|
|
|
3038 static int *
|
|
|
3039 mb_str2wide(s)
|
|
|
3040 char_u *s;
|
|
|
3041 {
|
|
|
3042 int *res;
|
|
|
3043 char_u *p;
|
|
|
3044 int i = 0;
|
|
|
3045
|
|
|
3046 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
3047 if (res != NULL)
|
|
|
3048 {
|
|
|
3049 for (p = s; *p != NUL; )
|
|
|
3050 res[i++] = mb_ptr2char_adv(&p);
|
|
|
3051 res[i] = NUL;
|
|
|
3052 }
|
|
|
3053 return res;
|
|
|
3054 }
|
|
|
3055 #endif
|
|
|
3056
|
|
236
|
3057 /*
|
|
300
|
3058 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
3059 * "byts" and index array "idxs". Recursively read the children.
|
|
|
3060 *
|
|
346
|
3061 * NOTE: The code here must match put_node().
|
|
300
|
3062 *
|
|
|
3063 * Returns the index follosing the siblings.
|
|
|
3064 * Returns -1 if the file is shorter than expected.
|
|
|
3065 * Returns -2 if there is a format error.
|
|
236
|
3066 */
|
|
324
|
3067 static idx_T
|
|
339
|
3068 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
3069 FILE *fd;
|
|
|
3070 char_u *byts;
|
|
324
|
3071 idx_T *idxs;
|
|
300
|
3072 int maxidx; /* size of arrays */
|
|
324
|
3073 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
3074 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
3075 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
3076 {
|
|
300
|
3077 int len;
|
|
|
3078 int i;
|
|
|
3079 int n;
|
|
324
|
3080 idx_T idx = startidx;
|
|
300
|
3081 int c;
|
|
366
|
3082 int c2;
|
|
300
|
3083 #define SHARED_MASK 0x8000000
|
|
236
|
3084
|
|
300
|
3085 len = getc(fd); /* <siblingcount> */
|
|
|
3086 if (len <= 0)
|
|
|
3087 return -1;
|
|
|
3088
|
|
|
3089 if (startidx + len >= maxidx)
|
|
|
3090 return -2;
|
|
|
3091 byts[idx++] = len;
|
|
|
3092
|
|
|
3093 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
3094 for (i = 1; i <= len; ++i)
|
|
236
|
3095 {
|
|
300
|
3096 c = getc(fd); /* <byte> */
|
|
|
3097 if (c < 0)
|
|
|
3098 return -1;
|
|
|
3099 if (c <= BY_SPECIAL)
|
|
|
3100 {
|
|
366
|
3101 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
3102 {
|
|
|
3103 /* No flags, all regions. */
|
|
|
3104 idxs[idx] = 0;
|
|
|
3105 c = 0;
|
|
|
3106 }
|
|
390
|
3107 else if (c != BY_INDEX)
|
|
300
|
3108 {
|
|
339
|
3109 if (prefixtree)
|
|
|
3110 {
|
|
455
|
3111 /* Read the optional pflags byte, the prefix ID and the
|
|
|
3112 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
3113 * byte, the condition index shifted up 8 bits, the flags
|
|
|
3114 * shifted up 24 bits. */
|
|
|
3115 if (c == BY_FLAGS)
|
|
|
3116 c = getc(fd) << 24; /* <pflags> */
|
|
|
3117 else
|
|
|
3118 c = 0;
|
|
|
3119
|
|
481
|
3120 c |= getc(fd); /* <affixID> */
|
|
455
|
3121
|
|
339
|
3122 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
3123 if (n >= maxprefcondnr)
|
|
|
3124 return -2;
|
|
455
|
3125 c |= (n << 8);
|
|
339
|
3126 }
|
|
390
|
3127 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
3128 {
|
|
|
3129 /* Read flags and optional region and prefix ID. In
|
|
390
|
3130 * idxs[] the flags go in the low two bytes, region above
|
|
|
3131 * that and prefix ID above the region. */
|
|
|
3132 c2 = c;
|
|
339
|
3133 c = getc(fd); /* <flags> */
|
|
390
|
3134 if (c2 == BY_FLAGS2)
|
|
|
3135 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
3136 if (c & WF_REGION)
|
|
390
|
3137 c = (getc(fd) << 16) + c; /* <region> */
|
|
481
|
3138 if (c & WF_AFX)
|
|
|
3139 c = (getc(fd) << 24) + c; /* <affixID> */
|
|
339
|
3140 }
|
|
|
3141
|
|
300
|
3142 idxs[idx] = c;
|
|
|
3143 c = 0;
|
|
|
3144 }
|
|
|
3145 else /* c == BY_INDEX */
|
|
|
3146 {
|
|
|
3147 /* <nodeidx> */
|
|
|
3148 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
3149 if (n < 0 || n >= maxidx)
|
|
|
3150 return -2;
|
|
|
3151 idxs[idx] = n + SHARED_MASK;
|
|
|
3152 c = getc(fd); /* <xbyte> */
|
|
|
3153 }
|
|
|
3154 }
|
|
|
3155 byts[idx++] = c;
|
|
236
|
3156 }
|
|
|
3157
|
|
300
|
3158 /* Recursively read the children for non-shared siblings.
|
|
|
3159 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
3160 * remove SHARED_MASK) */
|
|
|
3161 for (i = 1; i <= len; ++i)
|
|
|
3162 if (byts[startidx + i] != 0)
|
|
|
3163 {
|
|
|
3164 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
3165 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
3166 else
|
|
|
3167 {
|
|
|
3168 idxs[startidx + i] = idx;
|
|
339
|
3169 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
3170 prefixtree, maxprefcondnr);
|
|
300
|
3171 if (idx < 0)
|
|
|
3172 break;
|
|
|
3173 }
|
|
|
3174 }
|
|
236
|
3175
|
|
300
|
3176 return idx;
|
|
236
|
3177 }
|
|
|
3178
|
|
|
3179 /*
|
|
|
3180 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
3181 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
3182 */
|
|
|
3183 char_u *
|
|
|
3184 did_set_spelllang(buf)
|
|
|
3185 buf_T *buf;
|
|
|
3186 {
|
|
|
3187 garray_T ga;
|
|
351
|
3188 char_u *splp;
|
|
236
|
3189 char_u *region;
|
|
416
|
3190 char_u region_cp[3];
|
|
355
|
3191 int filename;
|
|
236
|
3192 int region_mask;
|
|
|
3193 slang_T *lp;
|
|
|
3194 int c;
|
|
351
|
3195 char_u lang[MAXWLEN + 1];
|
|
323
|
3196 char_u spf_name[MAXPATHL];
|
|
351
|
3197 int len;
|
|
|
3198 char_u *p;
|
|
381
|
3199 int round;
|
|
385
|
3200 char_u *spf;
|
|
389
|
3201 char_u *use_region = NULL;
|
|
|
3202 int dont_use_region = FALSE;
|
|
236
|
3203
|
|
|
3204 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
3205 clear_midword(buf);
|
|
236
|
3206
|
|
351
|
3207 /* loop over comma separated language names. */
|
|
|
3208 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
3209 {
|
|
|
3210 /* Get one language name. */
|
|
|
3211 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
3212
|
|
240
|
3213 region = NULL;
|
|
351
|
3214 len = STRLEN(lang);
|
|
355
|
3215
|
|
|
3216 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
3217 * If there is a region name let "region" point to it and remove it
|
|
|
3218 * from the name. */
|
|
|
3219 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
3220 {
|
|
355
|
3221 filename = TRUE;
|
|
|
3222
|
|
416
|
3223 /* Locate a region and remove it from the file name. */
|
|
|
3224 p = vim_strchr(gettail(lang), '_');
|
|
|
3225 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
3226 && !ASCII_ISALPHA(p[3]))
|
|
|
3227 {
|
|
|
3228 vim_strncpy(region_cp, p + 1, 2);
|
|
|
3229 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
3230 len -= 3;
|
|
|
3231 region = region_cp;
|
|
|
3232 }
|
|
|
3233 else
|
|
|
3234 dont_use_region = TRUE;
|
|
|
3235
|
|
355
|
3236 /* Check if we loaded this language before. */
|
|
|
3237 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3238 if (fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3239 break;
|
|
236
|
3240 }
|
|
355
|
3241 else
|
|
|
3242 {
|
|
|
3243 filename = FALSE;
|
|
|
3244 if (len > 3 && lang[len - 3] == '_')
|
|
|
3245 {
|
|
|
3246 region = lang + len - 2;
|
|
|
3247 len -= 3;
|
|
|
3248 lang[len] = NUL;
|
|
389
|
3249 }
|
|
|
3250 else
|
|
|
3251 dont_use_region = TRUE;
|
|
355
|
3252
|
|
|
3253 /* Check if we loaded this language before. */
|
|
|
3254 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3255 if (STRICMP(lang, lp->sl_name) == 0)
|
|
|
3256 break;
|
|
|
3257 }
|
|
236
|
3258
|
|
416
|
3259 if (region != NULL)
|
|
|
3260 {
|
|
|
3261 /* If the region differs from what was used before then don't
|
|
|
3262 * use it for 'spellfile'. */
|
|
|
3263 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
3264 dont_use_region = TRUE;
|
|
|
3265 use_region = region;
|
|
|
3266 }
|
|
|
3267
|
|
351
|
3268 /* If not found try loading the language now. */
|
|
236
|
3269 if (lp == NULL)
|
|
355
|
3270 {
|
|
|
3271 if (filename)
|
|
|
3272 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
3273 else
|
|
|
3274 spell_load_lang(lang);
|
|
|
3275 }
|
|
236
|
3276
|
|
307
|
3277 /*
|
|
351
|
3278 * Loop over the languages, there can be several files for "lang".
|
|
307
|
3279 */
|
|
|
3280 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
355
|
3281 if (filename ? fullpathcmp(lang, lp->sl_fname, FALSE) == FPC_SAME
|
|
|
3282 : STRICMP(lang, lp->sl_name) == 0)
|
|
236
|
3283 {
|
|
316
|
3284 region_mask = REGION_ALL;
|
|
355
|
3285 if (!filename && region != NULL)
|
|
236
|
3286 {
|
|
307
|
3287 /* find region in sl_regions */
|
|
|
3288 c = find_region(lp->sl_regions, region);
|
|
|
3289 if (c == REGION_ALL)
|
|
|
3290 {
|
|
389
|
3291 if (lp->sl_add)
|
|
|
3292 {
|
|
|
3293 if (*lp->sl_regions != NUL)
|
|
|
3294 /* This addition file is for other regions. */
|
|
|
3295 region_mask = 0;
|
|
|
3296 }
|
|
|
3297 else
|
|
|
3298 /* This is probably an error. Give a warning and
|
|
|
3299 * accept the words anyway. */
|
|
351
|
3300 smsg((char_u *)
|
|
|
3301 _("Warning: region %s not supported"),
|
|
|
3302 region);
|
|
307
|
3303 }
|
|
|
3304 else
|
|
|
3305 region_mask = 1 << c;
|
|
236
|
3306 }
|
|
307
|
3307
|
|
389
|
3308 if (region_mask != 0)
|
|
307
|
3309 {
|
|
389
|
3310 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3311 {
|
|
|
3312 ga_clear(&ga);
|
|
|
3313 return e_outofmem;
|
|
|
3314 }
|
|
|
3315 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
3316 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3317 ++ga.ga_len;
|
|
|
3318 use_midword(lp, buf);
|
|
307
|
3319 }
|
|
385
|
3320 }
|
|
|
3321 }
|
|
|
3322
|
|
|
3323 /* round 0: load int_wordlist, if possible.
|
|
|
3324 * round 1: load first name in 'spellfile'.
|
|
|
3325 * round 2: load second name in 'spellfile.
|
|
|
3326 * etc. */
|
|
|
3327 spf = curbuf->b_p_spf;
|
|
|
3328 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
3329 {
|
|
|
3330 if (round == 0)
|
|
|
3331 {
|
|
|
3332 /* Internal wordlist, if there is one. */
|
|
|
3333 if (int_wordlist == NULL)
|
|
381
|
3334 continue;
|
|
385
|
3335 int_wordlist_spl(spf_name);
|
|
381
|
3336 }
|
|
|
3337 else
|
|
|
3338 {
|
|
385
|
3339 /* One entry in 'spellfile'. */
|
|
|
3340 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
3341 STRCAT(spf_name, ".spl");
|
|
|
3342
|
|
|
3343 /* If it was already found above then skip it. */
|
|
|
3344 for (c = 0; c < ga.ga_len; ++c)
|
|
|
3345 if (fullpathcmp(spf_name,
|
|
|
3346 LANGP_ENTRY(ga, c)->lp_slang->sl_fname,
|
|
|
3347 FALSE) == FPC_SAME)
|
|
|
3348 break;
|
|
|
3349 if (c < ga.ga_len)
|
|
381
|
3350 continue;
|
|
|
3351 }
|
|
|
3352
|
|
351
|
3353 /* Check if it was loaded already. */
|
|
323
|
3354 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3355 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3356 break;
|
|
|
3357 if (lp == NULL)
|
|
|
3358 {
|
|
351
|
3359 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
3360 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
3361 * use an arbitrary name. */
|
|
|
3362 if (round == 0)
|
|
|
3363 STRCPY(lang, "internal wordlist");
|
|
|
3364 else
|
|
|
3365 {
|
|
|
3366 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
3367 p = vim_strchr(lang, '.');
|
|
|
3368 if (p != NULL)
|
|
|
3369 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
3370 }
|
|
351
|
3371 lp = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
323
|
3372 }
|
|
|
3373 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
|
3374 {
|
|
389
|
3375 region_mask = REGION_ALL;
|
|
|
3376 if (use_region != NULL && !dont_use_region)
|
|
|
3377 {
|
|
|
3378 /* find region in sl_regions */
|
|
|
3379 c = find_region(lp->sl_regions, use_region);
|
|
|
3380 if (c != REGION_ALL)
|
|
|
3381 region_mask = 1 << c;
|
|
|
3382 else if (*lp->sl_regions != NUL)
|
|
|
3383 /* This spell file is for other regions. */
|
|
|
3384 region_mask = 0;
|
|
|
3385 }
|
|
|
3386
|
|
|
3387 if (region_mask != 0)
|
|
|
3388 {
|
|
|
3389 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
|
3390 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3391 ++ga.ga_len;
|
|
|
3392 use_midword(lp, buf);
|
|
|
3393 }
|
|
323
|
3394 }
|
|
|
3395 }
|
|
|
3396
|
|
236
|
3397 /* Add a NULL entry to mark the end of the list. */
|
|
|
3398 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3399 {
|
|
|
3400 ga_clear(&ga);
|
|
|
3401 return e_outofmem;
|
|
|
3402 }
|
|
|
3403 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
|
3404 ++ga.ga_len;
|
|
|
3405
|
|
|
3406 /* Everything is fine, store the new b_langp value. */
|
|
|
3407 ga_clear(&buf->b_langp);
|
|
|
3408 buf->b_langp = ga;
|
|
|
3409
|
|
|
3410 return NULL;
|
|
|
3411 }
|
|
|
3412
|
|
|
3413 /*
|
|
376
|
3414 * Clear the midword characters for buffer "buf".
|
|
|
3415 */
|
|
|
3416 static void
|
|
|
3417 clear_midword(buf)
|
|
|
3418 buf_T *buf;
|
|
|
3419 {
|
|
|
3420 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
3421 #ifdef FEAT_MBYTE
|
|
|
3422 vim_free(buf->b_spell_ismw_mb);
|
|
|
3423 buf->b_spell_ismw_mb = NULL;
|
|
|
3424 #endif
|
|
|
3425 }
|
|
|
3426
|
|
|
3427 /*
|
|
|
3428 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
3429 * They add up to any currently used midword characters.
|
|
|
3430 */
|
|
|
3431 static void
|
|
|
3432 use_midword(lp, buf)
|
|
|
3433 slang_T *lp;
|
|
|
3434 buf_T *buf;
|
|
|
3435 {
|
|
|
3436 char_u *p;
|
|
|
3437
|
|
389
|
3438 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
3439 return;
|
|
|
3440
|
|
376
|
3441 for (p = lp->sl_midword; *p != NUL; )
|
|
|
3442 #ifdef FEAT_MBYTE
|
|
|
3443 if (has_mbyte)
|
|
|
3444 {
|
|
|
3445 int c, l, n;
|
|
|
3446 char_u *bp;
|
|
|
3447
|
|
|
3448 c = mb_ptr2char(p);
|
|
474
|
3449 l = (*mb_ptr2len)(p);
|
|
|
3450 if (c < 256 && l <= 2)
|
|
376
|
3451 buf->b_spell_ismw[c] = TRUE;
|
|
|
3452 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
3453 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
3454 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
3455 else
|
|
|
3456 {
|
|
|
3457 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
3458 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
3459 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
3460 if (bp != NULL)
|
|
|
3461 {
|
|
|
3462 vim_free(buf->b_spell_ismw_mb);
|
|
|
3463 buf->b_spell_ismw_mb = bp;
|
|
|
3464 vim_strncpy(bp + n, p, l);
|
|
|
3465 }
|
|
|
3466 }
|
|
|
3467 p += l;
|
|
|
3468 }
|
|
|
3469 else
|
|
|
3470 #endif
|
|
|
3471 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
3472 }
|
|
|
3473
|
|
|
3474 /*
|
|
236
|
3475 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
3476 * Each region is simply stored as the two characters of it's name.
|
|
381
|
3477 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
3478 */
|
|
|
3479 static int
|
|
|
3480 find_region(rp, region)
|
|
|
3481 char_u *rp;
|
|
|
3482 char_u *region;
|
|
|
3483 {
|
|
|
3484 int i;
|
|
|
3485
|
|
|
3486 for (i = 0; ; i += 2)
|
|
|
3487 {
|
|
|
3488 if (rp[i] == NUL)
|
|
|
3489 return REGION_ALL;
|
|
|
3490 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
3491 break;
|
|
|
3492 }
|
|
|
3493 return i / 2;
|
|
|
3494 }
|
|
|
3495
|
|
|
3496 /*
|
|
323
|
3497 * Return case type of word:
|
|
236
|
3498 * w word 0
|
|
300
|
3499 * Word WF_ONECAP
|
|
|
3500 * W WORD WF_ALLCAP
|
|
|
3501 * WoRd wOrd WF_KEEPCAP
|
|
236
|
3502 */
|
|
|
3503 static int
|
|
|
3504 captype(word, end)
|
|
|
3505 char_u *word;
|
|
323
|
3506 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
3507 {
|
|
|
3508 char_u *p;
|
|
|
3509 int c;
|
|
|
3510 int firstcap;
|
|
|
3511 int allcap;
|
|
|
3512 int past_second = FALSE; /* past second word char */
|
|
|
3513
|
|
|
3514 /* find first letter */
|
|
376
|
3515 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
3516 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
3517 return 0; /* only non-word characters, illegal word */
|
|
|
3518 #ifdef FEAT_MBYTE
|
|
310
|
3519 if (has_mbyte)
|
|
|
3520 c = mb_ptr2char_adv(&p);
|
|
|
3521 else
|
|
236
|
3522 #endif
|
|
310
|
3523 c = *p++;
|
|
324
|
3524 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
3525
|
|
|
3526 /*
|
|
|
3527 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
3528 * But a word with an upper char only at start is a ONECAP.
|
|
|
3529 */
|
|
323
|
3530 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
3531 if (spell_iswordp_nmw(p))
|
|
236
|
3532 {
|
|
455
|
3533 c = PTR2CHAR(p);
|
|
324
|
3534 if (!SPELL_ISUPPER(c))
|
|
236
|
3535 {
|
|
|
3536 /* UUl -> KEEPCAP */
|
|
|
3537 if (past_second && allcap)
|
|
300
|
3538 return WF_KEEPCAP;
|
|
236
|
3539 allcap = FALSE;
|
|
|
3540 }
|
|
|
3541 else if (!allcap)
|
|
|
3542 /* UlU -> KEEPCAP */
|
|
300
|
3543 return WF_KEEPCAP;
|
|
236
|
3544 past_second = TRUE;
|
|
|
3545 }
|
|
|
3546
|
|
|
3547 if (allcap)
|
|
300
|
3548 return WF_ALLCAP;
|
|
236
|
3549 if (firstcap)
|
|
300
|
3550 return WF_ONECAP;
|
|
236
|
3551 return 0;
|
|
|
3552 }
|
|
|
3553
|
|
474
|
3554 /*
|
|
|
3555 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
3556 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
3557 * Add ALLCAP for "WOrD".
|
|
|
3558 */
|
|
|
3559 static int
|
|
|
3560 badword_captype(word, end)
|
|
|
3561 char_u *word;
|
|
|
3562 char_u *end;
|
|
|
3563 {
|
|
|
3564 int flags = captype(word, end);
|
|
475
|
3565 int c;
|
|
474
|
3566 int l, u;
|
|
|
3567 int first;
|
|
|
3568 char_u *p;
|
|
|
3569
|
|
|
3570 if (flags & WF_KEEPCAP)
|
|
|
3571 {
|
|
|
3572 /* Count the number of UPPER and lower case letters. */
|
|
|
3573 l = u = 0;
|
|
|
3574 first = FALSE;
|
|
|
3575 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
3576 {
|
|
475
|
3577 c = PTR2CHAR(p);
|
|
|
3578 if (SPELL_ISUPPER(c))
|
|
474
|
3579 {
|
|
|
3580 ++u;
|
|
|
3581 if (p == word)
|
|
|
3582 first = TRUE;
|
|
|
3583 }
|
|
|
3584 else
|
|
|
3585 ++l;
|
|
|
3586 }
|
|
|
3587
|
|
|
3588 /* If there are more UPPER than lower case letters suggest an
|
|
|
3589 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
3590 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
3591 * require three upper case letters. */
|
|
|
3592 if (u > l && u > 2)
|
|
|
3593 flags |= WF_ALLCAP;
|
|
|
3594 else if (first)
|
|
|
3595 flags |= WF_ONECAP;
|
|
|
3596 }
|
|
|
3597 return flags;
|
|
|
3598 }
|
|
|
3599
|
|
355
|
3600 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
3601 /*
|
|
|
3602 * Free all languages.
|
|
|
3603 */
|
|
|
3604 void
|
|
|
3605 spell_free_all()
|
|
|
3606 {
|
|
|
3607 slang_T *lp;
|
|
|
3608 buf_T *buf;
|
|
385
|
3609 char_u fname[MAXPATHL];
|
|
355
|
3610
|
|
|
3611 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3612 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3613 ga_clear(&buf->b_langp);
|
|
|
3614
|
|
|
3615 while (first_lang != NULL)
|
|
|
3616 {
|
|
|
3617 lp = first_lang;
|
|
|
3618 first_lang = lp->sl_next;
|
|
|
3619 slang_free(lp);
|
|
|
3620 }
|
|
366
|
3621
|
|
385
|
3622 if (int_wordlist != NULL)
|
|
|
3623 {
|
|
|
3624 /* Delete the internal wordlist and its .spl file */
|
|
|
3625 mch_remove(int_wordlist);
|
|
|
3626 int_wordlist_spl(fname);
|
|
|
3627 mch_remove(fname);
|
|
|
3628 vim_free(int_wordlist);
|
|
|
3629 int_wordlist = NULL;
|
|
381
|
3630 }
|
|
|
3631
|
|
366
|
3632 init_spell_chartab();
|
|
355
|
3633 }
|
|
|
3634 # endif
|
|
|
3635
|
|
236
|
3636 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
3637 /*
|
|
|
3638 * Clear all spelling tables and reload them.
|
|
307
|
3639 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
3640 */
|
|
|
3641 void
|
|
|
3642 spell_reload()
|
|
|
3643 {
|
|
|
3644 buf_T *buf;
|
|
316
|
3645 win_T *wp;
|
|
236
|
3646
|
|
358
|
3647 /* Initialize the table for spell_iswordp(). */
|
|
236
|
3648 init_spell_chartab();
|
|
|
3649
|
|
|
3650 /* Unload all allocated memory. */
|
|
355
|
3651 spell_free_all();
|
|
236
|
3652
|
|
|
3653 /* Go through all buffers and handle 'spelllang'. */
|
|
|
3654 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
3655 {
|
|
316
|
3656 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
3657 * window for this buffer in which 'spell' is set. */
|
|
236
|
3658 if (*buf->b_p_spl != NUL)
|
|
316
|
3659 {
|
|
|
3660 FOR_ALL_WINDOWS(wp)
|
|
|
3661 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
3662 {
|
|
|
3663 (void)did_set_spelllang(buf);
|
|
|
3664 # ifdef FEAT_WINDOWS
|
|
|
3665 break;
|
|
|
3666 # endif
|
|
|
3667 }
|
|
|
3668 }
|
|
236
|
3669 }
|
|
|
3670 }
|
|
|
3671 # endif
|
|
|
3672
|
|
310
|
3673 /*
|
|
|
3674 * Reload the spell file "fname" if it's loaded.
|
|
|
3675 */
|
|
|
3676 static void
|
|
323
|
3677 spell_reload_one(fname, added_word)
|
|
310
|
3678 char_u *fname;
|
|
323
|
3679 int added_word; /* invoked through "zg" */
|
|
310
|
3680 {
|
|
|
3681 slang_T *lp;
|
|
323
|
3682 int didit = FALSE;
|
|
310
|
3683
|
|
|
3684 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
|
3685 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
|
3686 {
|
|
|
3687 slang_clear(lp);
|
|
323
|
3688 (void)spell_load_file(fname, NULL, lp, FALSE);
|
|
310
|
3689 redraw_all_later(NOT_VALID);
|
|
323
|
3690 didit = TRUE;
|
|
310
|
3691 }
|
|
323
|
3692
|
|
|
3693 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
3694 * 'spelllang' to get it loaded. */
|
|
|
3695 if (added_word && !didit)
|
|
|
3696 did_set_spelllang(curbuf);
|
|
310
|
3697 }
|
|
|
3698
|
|
|
3699
|
|
236
|
3700 /*
|
|
|
3701 * Functions for ":mkspell".
|
|
|
3702 */
|
|
|
3703
|
|
300
|
3704 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
3705 and .dic file. */
|
|
|
3706 /*
|
|
|
3707 * Main structure to store the contents of a ".aff" file.
|
|
|
3708 */
|
|
|
3709 typedef struct afffile_S
|
|
|
3710 {
|
|
|
3711 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
481
|
3712 int af_slash; /* character used in word for slash */
|
|
310
|
3713 int af_rar; /* RAR ID for rare word */
|
|
|
3714 int af_kep; /* KEP ID for keep-case word */
|
|
346
|
3715 int af_bad; /* BAD ID for banned word */
|
|
483
|
3716 int af_needaffix; /* NEEDAFFIX ID */
|
|
|
3717 char_u *af_compflags; /* COMPOUNDFLAG and COMPOUNDFLAGS concat'ed */
|
|
|
3718 int af_compmax; /* COMPOUNDMAX */
|
|
481
|
3719 int af_compminlen; /* COMPOUNDMIN */
|
|
483
|
3720 int af_compsylmax; /* COMPOUNDSYLMAX */
|
|
|
3721 char_u *af_syllable; /* SYLLABLE */
|
|
339
|
3722 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
3723 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
3724 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
|
3725 } afffile_T;
|
|
|
3726
|
|
|
3727 typedef struct affentry_S affentry_T;
|
|
|
3728 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
3729 struct affentry_S
|
|
|
3730 {
|
|
|
3731 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
3732 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
3733 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
3734 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
3735 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
483
|
3736 char_u ae_rare; /* rare affix */
|
|
|
3737 char_u ae_nocomp; /* word with affix not compoundable */
|
|
300
|
3738 };
|
|
|
3739
|
|
455
|
3740 #define AH_KEY_LEN 10
|
|
|
3741
|
|
300
|
3742 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
3743 typedef struct affheader_S
|
|
|
3744 {
|
|
455
|
3745 /* key for hashtable == name of affix entry */
|
|
|
3746 #ifdef FEAT_MBYTE
|
|
|
3747 char_u ah_key[AH_KEY_LEN]; /* multi-byte char plus NUL */
|
|
|
3748 #else
|
|
|
3749 char_u ah_key[2]; /* one byte char plus NUL */
|
|
|
3750 #endif
|
|
339
|
3751 int ah_newID; /* prefix ID after renumbering */
|
|
300
|
3752 int ah_combine; /* suffix may combine with prefix */
|
|
|
3753 affentry_T *ah_first; /* first affix entry */
|
|
|
3754 } affheader_T;
|
|
|
3755
|
|
|
3756 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
3757
|
|
|
3758 /*
|
|
|
3759 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
3760 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
3761 * once after ":mkspell" is done.
|
|
300
|
3762 */
|
|
|
3763 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
3764 typedef struct sblock_S sblock_T;
|
|
|
3765 struct sblock_S
|
|
|
3766 {
|
|
|
3767 sblock_T *sb_next; /* next block in list */
|
|
|
3768 int sb_used; /* nr of bytes already in use */
|
|
|
3769 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
3770 };
|
|
|
3771
|
|
|
3772 /*
|
|
300
|
3773 * A node in the tree.
|
|
236
|
3774 */
|
|
300
|
3775 typedef struct wordnode_S wordnode_T;
|
|
|
3776 struct wordnode_S
|
|
236
|
3777 {
|
|
346
|
3778 union /* shared to save space */
|
|
|
3779 {
|
|
469
|
3780 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
3781 int index; /* index in written nodes (valid after first
|
|
|
3782 round) */
|
|
|
3783 } wn_u1;
|
|
|
3784 union /* shared to save space */
|
|
|
3785 {
|
|
|
3786 wordnode_T *next; /* next node with same hash key */
|
|
|
3787 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
3788 } wn_u2;
|
|
300
|
3789 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
3790 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
3791 always sorted) */
|
|
474
|
3792 int wn_refs; /* Nr. of references to this node. Only
|
|
|
3793 relevant for first node in a list of
|
|
|
3794 siblings, in following siblings it is
|
|
|
3795 always one. */
|
|
300
|
3796 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
481
|
3797 char_u wn_affixID; /* when "wn_byte" is NUL: supported/required
|
|
469
|
3798 prefix ID or 0 */
|
|
|
3799 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
3800 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
3801 PREFIXTREE it's the prefcondnr */
|
|
469
|
3802 #ifdef SPELL_PRINTTREE
|
|
|
3803 int wn_nr; /* sequence nr for printing */
|
|
|
3804 #endif
|
|
236
|
3805 };
|
|
|
3806
|
|
390
|
3807 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
3808
|
|
300
|
3809 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
3810
|
|
300
|
3811 /*
|
|
|
3812 * Info used while reading the spell files.
|
|
|
3813 */
|
|
|
3814 typedef struct spellinfo_S
|
|
249
|
3815 {
|
|
300
|
3816 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
3817 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
3818
|
|
300
|
3819 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
3820 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
3821
|
|
339
|
3822 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
3823
|
|
300
|
3824 sblock_T *si_blocks; /* memory blocks used */
|
|
482
|
3825 long si_blocks_cnt; /* memory blocks allocated */
|
|
|
3826 long si_compress_cnt; /* words to add before lowering
|
|
|
3827 compression limit */
|
|
474
|
3828 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
3829 compression, linked by "wn_child" field. */
|
|
482
|
3830 long si_free_count; /* number of nodes in si_first_free */
|
|
474
|
3831 #ifdef SPELL_PRINTTREE
|
|
|
3832 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
3833 #endif
|
|
|
3834
|
|
|
3835
|
|
300
|
3836 int si_ascii; /* handling only ASCII words */
|
|
310
|
3837 int si_add; /* addition file */
|
|
351
|
3838 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
3839 int si_region; /* region mask */
|
|
|
3840 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
3841 int si_memtot; /* runtime memory used */
|
|
310
|
3842 int si_verbose; /* verbose messages */
|
|
474
|
3843 int si_msg_count; /* number of words added since last message */
|
|
316
|
3844 int si_region_count; /* number of regions supported (1 when there
|
|
|
3845 are no regions) */
|
|
483
|
3846 char_u si_region_name[16]; /* region names; used only if
|
|
|
3847 * si_region_count > 1) */
|
|
323
|
3848
|
|
|
3849 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
3850 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
3851 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
3852 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
3853 int si_followup; /* soundsalike: ? */
|
|
|
3854 int si_collapse; /* soundsalike: ? */
|
|
|
3855 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
3856 garray_T si_map; /* MAP info concatenated */
|
|
366
|
3857 char_u *si_midword; /* MIDWORD chars, alloc'ed string or NULL */
|
|
483
|
3858 int si_compmax; /* max nr of words for compounding */
|
|
481
|
3859 int si_compminlen; /* minimal length for compounding */
|
|
483
|
3860 int si_compsylmax; /* max nr of syllables for compounding */
|
|
481
|
3861 char_u *si_compflags; /* flags used for compounding */
|
|
483
|
3862 char_u *si_syllable; /* syllable string */
|
|
339
|
3863 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
3864 * prefixes, each stored as a string */
|
|
|
3865 int si_newID; /* current value for ah_newID */
|
|
300
|
3866 } spellinfo_T;
|
|
249
|
3867
|
|
474
|
3868 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
339
|
3869 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
3870 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
3871 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
3872 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
3873 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
3874 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
483
|
3875 static int get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
|
3876 static void get_compflags __ARGS((spellinfo_T *spin, char_u *afflist, char_u *store_afflist));
|
|
|
3877 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
|
3878 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
3879 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
3880 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
3881 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
3882 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
483
|
3883 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist, int need_affix));
|
|
481
|
3884 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID));
|
|
474
|
3885 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
|
3886 static void deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
|
3887 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
3888 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
3889 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
3890 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
474
|
3891 static void write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
3892 static void clear_node __ARGS((wordnode_T *node));
|
|
|
3893 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
3894 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
3895 static void init_spellfile __ARGS((void));
|
|
236
|
3896
|
|
455
|
3897 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
3898 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
3899 * Use a negative number with the lower 8 bits zero. */
|
|
|
3900 #define PFX_FLAGS -256
|
|
390
|
3901
|
|
483
|
3902 /*
|
|
|
3903 * Tunable parameters for when the tree is compressed. See 'mkspellmem'.
|
|
|
3904 */
|
|
|
3905 static long compress_start = 30000; /* memory / SBLOCKSIZE */
|
|
|
3906 static long compress_inc = 100; /* memory / SBLOCKSIZE */
|
|
|
3907 static long compress_added = 500000; /* word count */
|
|
|
3908
|
|
469
|
3909 #ifdef SPELL_PRINTTREE
|
|
|
3910 /*
|
|
|
3911 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
3912 * readable format, so that we can see what happens when adding a word and/or
|
|
|
3913 * compressing the tree.
|
|
|
3914 * Based on code from Olaf Seibert.
|
|
|
3915 */
|
|
|
3916 #define PRINTLINESIZE 1000
|
|
|
3917 #define PRINTWIDTH 6
|
|
|
3918
|
|
|
3919 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
3920 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
3921
|
|
|
3922 static char line1[PRINTLINESIZE];
|
|
|
3923 static char line2[PRINTLINESIZE];
|
|
|
3924 static char line3[PRINTLINESIZE];
|
|
|
3925
|
|
|
3926 static void
|
|
|
3927 spell_clear_flags(wordnode_T *node)
|
|
|
3928 {
|
|
|
3929 wordnode_T *np;
|
|
|
3930
|
|
|
3931 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
3932 {
|
|
|
3933 np->wn_u1.index = FALSE;
|
|
|
3934 spell_clear_flags(np->wn_child);
|
|
|
3935 }
|
|
|
3936 }
|
|
|
3937
|
|
|
3938 static void
|
|
|
3939 spell_print_node(wordnode_T *node, int depth)
|
|
|
3940 {
|
|
|
3941 if (node->wn_u1.index)
|
|
|
3942 {
|
|
|
3943 /* Done this node before, print the reference. */
|
|
|
3944 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
3945 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
3946 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3947 msg(line1);
|
|
|
3948 msg(line2);
|
|
|
3949 msg(line3);
|
|
|
3950 }
|
|
|
3951 else
|
|
|
3952 {
|
|
|
3953 node->wn_u1.index = TRUE;
|
|
|
3954
|
|
|
3955 if (node->wn_byte != NUL)
|
|
|
3956 {
|
|
|
3957 if (node->wn_child != NULL)
|
|
|
3958 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
3959 else
|
|
|
3960 /* Cannot happen? */
|
|
|
3961 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
3962 }
|
|
|
3963 else
|
|
|
3964 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
3965
|
|
|
3966 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
3967
|
|
|
3968 if (node->wn_sibling != NULL)
|
|
|
3969 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
3970 else
|
|
|
3971 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
3972
|
|
|
3973 if (node->wn_byte == NUL)
|
|
|
3974 {
|
|
|
3975 msg(line1);
|
|
|
3976 msg(line2);
|
|
|
3977 msg(line3);
|
|
|
3978 }
|
|
|
3979
|
|
|
3980 /* do the children */
|
|
|
3981 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
3982 spell_print_node(node->wn_child, depth + 1);
|
|
|
3983
|
|
|
3984 /* do the siblings */
|
|
|
3985 if (node->wn_sibling != NULL)
|
|
|
3986 {
|
|
|
3987 /* get rid of all parent details except | */
|
|
|
3988 STRCPY(line1, line3);
|
|
|
3989 STRCPY(line2, line3);
|
|
|
3990 spell_print_node(node->wn_sibling, depth);
|
|
|
3991 }
|
|
|
3992 }
|
|
|
3993 }
|
|
|
3994
|
|
|
3995 static void
|
|
|
3996 spell_print_tree(wordnode_T *root)
|
|
|
3997 {
|
|
|
3998 if (root != NULL)
|
|
|
3999 {
|
|
|
4000 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
4001 * done. */
|
|
|
4002 spell_clear_flags(root);
|
|
|
4003
|
|
|
4004 /* Recursively print the tree. */
|
|
|
4005 spell_print_node(root, 0);
|
|
|
4006 }
|
|
|
4007 }
|
|
|
4008 #endif /* SPELL_PRINTTREE */
|
|
|
4009
|
|
236
|
4010 /*
|
|
323
|
4011 * Read the affix file "fname".
|
|
316
|
4012 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
4013 */
|
|
|
4014 static afffile_T *
|
|
474
|
4015 spell_read_aff(spin, fname)
|
|
|
4016 spellinfo_T *spin;
|
|
236
|
4017 char_u *fname;
|
|
|
4018 {
|
|
|
4019 FILE *fd;
|
|
|
4020 afffile_T *aff;
|
|
|
4021 char_u rline[MAXLINELEN];
|
|
|
4022 char_u *line;
|
|
|
4023 char_u *pc = NULL;
|
|
334
|
4024 #define MAXITEMCNT 7
|
|
|
4025 char_u *(items[MAXITEMCNT]);
|
|
236
|
4026 int itemcnt;
|
|
|
4027 char_u *p;
|
|
|
4028 int lnum = 0;
|
|
|
4029 affheader_T *cur_aff = NULL;
|
|
|
4030 int aff_todo = 0;
|
|
|
4031 hashtab_T *tp;
|
|
255
|
4032 char_u *low = NULL;
|
|
|
4033 char_u *fol = NULL;
|
|
|
4034 char_u *upp = NULL;
|
|
307
|
4035 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
323
|
4036 int do_rep;
|
|
|
4037 int do_sal;
|
|
|
4038 int do_map;
|
|
366
|
4039 int do_midword;
|
|
375
|
4040 int do_sofo;
|
|
323
|
4041 int found_map = FALSE;
|
|
324
|
4042 hashitem_T *hi;
|
|
455
|
4043 int l;
|
|
236
|
4044
|
|
300
|
4045 /*
|
|
|
4046 * Open the file.
|
|
|
4047 */
|
|
310
|
4048 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4049 if (fd == NULL)
|
|
|
4050 {
|
|
|
4051 EMSG2(_(e_notopen), fname);
|
|
|
4052 return NULL;
|
|
|
4053 }
|
|
|
4054
|
|
310
|
4055 if (spin->si_verbose || p_verbose > 2)
|
|
|
4056 {
|
|
|
4057 if (!spin->si_verbose)
|
|
|
4058 verbose_enter();
|
|
366
|
4059 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
4060 out_flush();
|
|
|
4061 if (!spin->si_verbose)
|
|
|
4062 verbose_leave();
|
|
|
4063 }
|
|
236
|
4064
|
|
323
|
4065 /* Only do REP lines when not done in another .aff file already. */
|
|
|
4066 do_rep = spin->si_rep.ga_len == 0;
|
|
|
4067
|
|
|
4068 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
4069 do_sal = spin->si_sal.ga_len == 0;
|
|
|
4070
|
|
|
4071 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
4072 do_map = spin->si_map.ga_len == 0;
|
|
|
4073
|
|
366
|
4074 /* Only do MIDWORD line when not done in another .aff file already */
|
|
|
4075 do_midword = spin->si_midword == NULL;
|
|
|
4076
|
|
375
|
4077 /* Only do SOFOFROM and SOFOTO when not done in another .aff file already */
|
|
|
4078 do_sofo = spin->si_sofofr == NULL;
|
|
|
4079
|
|
300
|
4080 /*
|
|
|
4081 * Allocate and init the afffile_T structure.
|
|
|
4082 */
|
|
474
|
4083 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
4084 if (aff == NULL)
|
|
|
4085 return NULL;
|
|
|
4086 hash_init(&aff->af_pref);
|
|
|
4087 hash_init(&aff->af_suff);
|
|
|
4088
|
|
|
4089 /*
|
|
|
4090 * Read all the lines in the file one by one.
|
|
|
4091 */
|
|
255
|
4092 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
4093 {
|
|
255
|
4094 line_breakcheck();
|
|
236
|
4095 ++lnum;
|
|
|
4096
|
|
|
4097 /* Skip comment lines. */
|
|
|
4098 if (*rline == '#')
|
|
|
4099 continue;
|
|
|
4100
|
|
|
4101 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
4102 vim_free(pc);
|
|
310
|
4103 #ifdef FEAT_MBYTE
|
|
300
|
4104 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4105 {
|
|
300
|
4106 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
4107 if (pc == NULL)
|
|
|
4108 {
|
|
|
4109 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4110 fname, lnum, rline);
|
|
|
4111 continue;
|
|
|
4112 }
|
|
236
|
4113 line = pc;
|
|
|
4114 }
|
|
|
4115 else
|
|
310
|
4116 #endif
|
|
236
|
4117 {
|
|
|
4118 pc = NULL;
|
|
|
4119 line = rline;
|
|
|
4120 }
|
|
|
4121
|
|
|
4122 /* Split the line up in white separated items. Put a NUL after each
|
|
|
4123 * item. */
|
|
|
4124 itemcnt = 0;
|
|
|
4125 for (p = line; ; )
|
|
|
4126 {
|
|
|
4127 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
4128 ++p;
|
|
|
4129 if (*p == NUL)
|
|
|
4130 break;
|
|
334
|
4131 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
4132 break;
|
|
236
|
4133 items[itemcnt++] = p;
|
|
300
|
4134 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
4135 ++p;
|
|
|
4136 if (*p == NUL)
|
|
|
4137 break;
|
|
|
4138 *p++ = NUL;
|
|
|
4139 }
|
|
|
4140
|
|
|
4141 /* Handle non-empty lines. */
|
|
|
4142 if (itemcnt > 0)
|
|
|
4143 {
|
|
|
4144 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
4145 && aff->af_enc == NULL)
|
|
|
4146 {
|
|
310
|
4147 #ifdef FEAT_MBYTE
|
|
300
|
4148 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
4149 aff->af_enc = enc_canonize(items[1]);
|
|
|
4150 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
4151 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
4152 p_enc) == FAIL)
|
|
|
4153 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
4154 fname, aff->af_enc, p_enc);
|
|
375
|
4155 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4156 #else
|
|
|
4157 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4158 #endif
|
|
236
|
4159 }
|
|
366
|
4160 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2)
|
|
|
4161 {
|
|
|
4162 if (do_midword)
|
|
|
4163 spin->si_midword = vim_strsave(items[1]);
|
|
|
4164 }
|
|
302
|
4165 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
4166 {
|
|
323
|
4167 /* ignored, we always split */
|
|
302
|
4168 }
|
|
323
|
4169 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
4170 {
|
|
323
|
4171 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
4172 }
|
|
481
|
4173 else if (STRCMP(items[0], "SLASH") == 0 && itemcnt == 2
|
|
|
4174 && aff->af_slash == 0)
|
|
|
4175 {
|
|
|
4176 aff->af_slash = items[1][0];
|
|
|
4177 if (items[1][1] != NUL)
|
|
|
4178 smsg((char_u *)_("Character used for SLASH must be ASCII; in %s line %d: %s"),
|
|
|
4179 fname, lnum, items[1]);
|
|
|
4180 }
|
|
307
|
4181 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
|
4182 && aff->af_rar == 0)
|
|
|
4183 {
|
|
|
4184 aff->af_rar = items[1][0];
|
|
|
4185 if (items[1][1] != NUL)
|
|
|
4186 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4187 }
|
|
310
|
4188 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
|
4189 && aff->af_kep == 0)
|
|
307
|
4190 {
|
|
310
|
4191 aff->af_kep = items[1][0];
|
|
307
|
4192 if (items[1][1] != NUL)
|
|
|
4193 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4194 }
|
|
346
|
4195 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
4196 && aff->af_bad == 0)
|
|
|
4197 {
|
|
|
4198 aff->af_bad = items[1][0];
|
|
|
4199 if (items[1][1] != NUL)
|
|
|
4200 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4201 }
|
|
483
|
4202 else if (STRCMP(items[0], "NEEDAFFIX") == 0 && itemcnt == 2
|
|
|
4203 && aff->af_needaffix == 0)
|
|
|
4204 {
|
|
|
4205 aff->af_needaffix = items[1][0];
|
|
|
4206 if (items[1][1] != NUL)
|
|
|
4207 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4208 }
|
|
481
|
4209 else if (STRCMP(items[0], "COMPOUNDFLAG") == 0 && itemcnt == 2
|
|
483
|
4210 && aff->af_compflags == NULL)
|
|
|
4211 {
|
|
|
4212 p = getroom(spin, 3, FALSE);
|
|
|
4213 if (p != NULL)
|
|
|
4214 {
|
|
|
4215 /* Turn single flag "c" into COMPOUNDFLAGS compatible
|
|
|
4216 * string "c+". */
|
|
|
4217 p[0] = items[1][0];
|
|
|
4218 p[1] = '+';
|
|
|
4219 p[2] = NUL;
|
|
|
4220 aff->af_compflags = p;
|
|
|
4221 }
|
|
481
|
4222 if (items[1][1] != NUL)
|
|
|
4223 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
|
4224 }
|
|
483
|
4225 else if (STRCMP(items[0], "COMPOUNDFLAGS") == 0 && itemcnt == 2)
|
|
|
4226 {
|
|
|
4227 /* Concatenate this string to previously defined ones, using a
|
|
|
4228 * slash to separate them. */
|
|
|
4229 l = STRLEN(items[1]) + 1;
|
|
|
4230 if (aff->af_compflags != NULL)
|
|
|
4231 l += STRLEN(aff->af_compflags) + 1;
|
|
|
4232 p = getroom(spin, l, FALSE);
|
|
|
4233 if (p != NULL)
|
|
|
4234 {
|
|
|
4235 if (aff->af_compflags != NULL)
|
|
|
4236 {
|
|
|
4237 STRCPY(p, aff->af_compflags);
|
|
|
4238 STRCAT(p, "/");
|
|
|
4239 }
|
|
|
4240 STRCAT(p, items[1]);
|
|
|
4241 aff->af_compflags = p;
|
|
|
4242 }
|
|
|
4243 }
|
|
|
4244 else if (STRCMP(items[0], "COMPOUNDMAX") == 0 && itemcnt == 2
|
|
|
4245 && aff->af_compmax == 0)
|
|
|
4246 {
|
|
|
4247 aff->af_compmax = atoi((char *)items[1]);
|
|
|
4248 if (aff->af_compmax == 0)
|
|
|
4249 smsg((char_u *)_("Wrong COMPOUNDMAX value in %s line %d: %s"),
|
|
|
4250 fname, lnum, items[1]);
|
|
481
|
4251 }
|
|
|
4252 else if (STRCMP(items[0], "COMPOUNDMIN") == 0 && itemcnt == 2
|
|
|
4253 && aff->af_compminlen == 0)
|
|
|
4254 {
|
|
|
4255 aff->af_compminlen = atoi((char *)items[1]);
|
|
|
4256 if (aff->af_compminlen == 0)
|
|
|
4257 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"),
|
|
|
4258 fname, lnum, items[1]);
|
|
|
4259 }
|
|
483
|
4260 else if (STRCMP(items[0], "COMPOUNDSYLMAX") == 0 && itemcnt == 2
|
|
|
4261 && aff->af_compsylmax == 0)
|
|
|
4262 {
|
|
|
4263 aff->af_compsylmax = atoi((char *)items[1]);
|
|
|
4264 if (aff->af_compsylmax == 0)
|
|
|
4265 smsg((char_u *)_("Wrong COMPOUNDSYLMAX value in %s line %d: %s"),
|
|
|
4266 fname, lnum, items[1]);
|
|
|
4267 }
|
|
|
4268 else if (STRCMP(items[0], "SYLLABLE") == 0 && itemcnt == 2
|
|
|
4269 && aff->af_syllable == NULL)
|
|
|
4270 {
|
|
|
4271 aff->af_syllable = getroom_save(spin, items[1]);
|
|
|
4272 }
|
|
339
|
4273 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
4274 {
|
|
|
4275 aff->af_pfxpostpone = TRUE;
|
|
|
4276 }
|
|
236
|
4277 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4278 || STRCMP(items[0], "SFX") == 0)
|
|
|
4279 && aff_todo == 0
|
|
334
|
4280 && itemcnt >= 4)
|
|
236
|
4281 {
|
|
334
|
4282 /* Myspell allows extra text after the item, but that might
|
|
|
4283 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
|
4284 if (itemcnt > 4 && *items[4] != '#')
|
|
|
4285 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
4286 fname, lnum, items[4]);
|
|
|
4287
|
|
236
|
4288 /* New affix letter. */
|
|
474
|
4289 cur_aff = (affheader_T *)getroom(spin,
|
|
447
|
4290 sizeof(affheader_T), TRUE);
|
|
236
|
4291 if (cur_aff == NULL)
|
|
|
4292 break;
|
|
455
|
4293 #ifdef FEAT_MBYTE
|
|
|
4294 if (has_mbyte)
|
|
|
4295 {
|
|
474
|
4296 l = (*mb_ptr2len)(items[1]);
|
|
455
|
4297 if (l >= AH_KEY_LEN)
|
|
|
4298 l = 1; /* too long, must be an overlong sequence */
|
|
|
4299 else
|
|
|
4300 mch_memmove(cur_aff->ah_key, items[1], l);
|
|
|
4301 }
|
|
|
4302 else
|
|
|
4303 #endif
|
|
|
4304 {
|
|
|
4305 *cur_aff->ah_key = *items[1];
|
|
|
4306 l = 1;
|
|
|
4307 }
|
|
|
4308 cur_aff->ah_key[l] = NUL;
|
|
|
4309 if (items[1][l] != NUL)
|
|
307
|
4310 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
236
|
4311 if (*items[2] == 'Y')
|
|
|
4312 cur_aff->ah_combine = TRUE;
|
|
300
|
4313 else if (*items[2] != 'N')
|
|
236
|
4314 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
4315 fname, lnum, items[2]);
|
|
339
|
4316
|
|
236
|
4317 if (*items[0] == 'P')
|
|
339
|
4318 {
|
|
236
|
4319 tp = &aff->af_pref;
|
|
339
|
4320 /* Use a new number in the .spl file later, to be able to
|
|
|
4321 * handle multiple .aff files. */
|
|
|
4322 if (aff->af_pfxpostpone)
|
|
344
|
4323 cur_aff->ah_newID = ++spin->si_newID;
|
|
339
|
4324 }
|
|
236
|
4325 else
|
|
|
4326 tp = &aff->af_suff;
|
|
300
|
4327 aff_todo = atoi((char *)items[3]);
|
|
324
|
4328 hi = hash_find(tp, cur_aff->ah_key);
|
|
|
4329 if (!HASHITEM_EMPTY(hi))
|
|
300
|
4330 {
|
|
236
|
4331 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
4332 fname, lnum, items[1]);
|
|
300
|
4333 aff_todo = 0;
|
|
|
4334 }
|
|
236
|
4335 else
|
|
|
4336 hash_add(tp, cur_aff->ah_key);
|
|
|
4337 }
|
|
|
4338 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4339 || STRCMP(items[0], "SFX") == 0)
|
|
|
4340 && aff_todo > 0
|
|
|
4341 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
4342 && itemcnt >= 5)
|
|
236
|
4343 {
|
|
|
4344 affentry_T *aff_entry;
|
|
366
|
4345 int rare = FALSE;
|
|
483
|
4346 int nocomp = FALSE;
|
|
455
|
4347 int upper = FALSE;
|
|
366
|
4348 int lasti = 5;
|
|
|
4349
|
|
483
|
4350 /* Check for "rare" and "nocomp" after the other info. */
|
|
|
4351 while (itemcnt > lasti)
|
|
|
4352 {
|
|
|
4353 if (!rare && STRICMP(items[lasti], "rare") == 0)
|
|
|
4354 {
|
|
|
4355 rare = TRUE;
|
|
|
4356 ++lasti;
|
|
|
4357 }
|
|
|
4358 else if (!nocomp && STRICMP(items[lasti], "nocomp") == 0)
|
|
|
4359 {
|
|
|
4360 nocomp = TRUE;
|
|
|
4361 ++lasti;
|
|
|
4362 }
|
|
|
4363 else
|
|
|
4364 break;
|
|
366
|
4365 }
|
|
236
|
4366
|
|
334
|
4367 /* Myspell allows extra text after the item, but that might
|
|
|
4368 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
4369 if (itemcnt > lasti && *items[lasti] != '#')
|
|
482
|
4370 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]);
|
|
334
|
4371
|
|
236
|
4372 /* New item for an affix letter. */
|
|
|
4373 --aff_todo;
|
|
474
|
4374 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
4375 sizeof(affentry_T), TRUE);
|
|
236
|
4376 if (aff_entry == NULL)
|
|
|
4377 break;
|
|
366
|
4378 aff_entry->ae_rare = rare;
|
|
483
|
4379 aff_entry->ae_nocomp = nocomp;
|
|
240
|
4380
|
|
236
|
4381 if (STRCMP(items[2], "0") != 0)
|
|
474
|
4382 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
4383 if (STRCMP(items[3], "0") != 0)
|
|
474
|
4384 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
4385
|
|
300
|
4386 /* Don't use an affix entry with non-ASCII characters when
|
|
|
4387 * "spin->si_ascii" is TRUE. */
|
|
|
4388 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
4389 || has_non_ascii(aff_entry->ae_add)))
|
|
|
4390 {
|
|
|
4391 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
4392 cur_aff->ah_first = aff_entry;
|
|
300
|
4393
|
|
|
4394 if (STRCMP(items[4], ".") != 0)
|
|
|
4395 {
|
|
|
4396 char_u buf[MAXLINELEN];
|
|
|
4397
|
|
474
|
4398 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
4399 if (*items[0] == 'P')
|
|
|
4400 sprintf((char *)buf, "^%s", items[4]);
|
|
|
4401 else
|
|
|
4402 sprintf((char *)buf, "%s$", items[4]);
|
|
|
4403 aff_entry->ae_prog = vim_regcomp(buf,
|
|
481
|
4404 RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
4405 if (aff_entry->ae_prog == NULL)
|
|
|
4406 smsg((char_u *)_("Broken condition in %s line %d: %s"),
|
|
|
4407 fname, lnum, items[4]);
|
|
300
|
4408 }
|
|
339
|
4409
|
|
|
4410 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
4411 * for the condition. Use an existing one if possible. */
|
|
455
|
4412 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
4413 {
|
|
455
|
4414 /* When the chop string is one lower-case letter and
|
|
|
4415 * the add string ends in the upper-case letter we set
|
|
|
4416 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
4417 * letters from "ae_add". The condition must either
|
|
|
4418 * be empty or start with the same letter. */
|
|
|
4419 if (aff_entry->ae_chop != NULL
|
|
|
4420 && aff_entry->ae_add != NULL
|
|
|
4421 #ifdef FEAT_MBYTE
|
|
474
|
4422 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
4423 aff_entry->ae_chop)] == NUL
|
|
|
4424 #else
|
|
|
4425 && aff_entry->ae_chop[1] == NUL
|
|
|
4426 #endif
|
|
|
4427 )
|
|
339
|
4428 {
|
|
455
|
4429 int c, c_up;
|
|
|
4430
|
|
|
4431 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
4432 c_up = SPELL_TOUPPER(c);
|
|
|
4433 if (c_up != c
|
|
|
4434 && (aff_entry->ae_cond == NULL
|
|
|
4435 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
4436 {
|
|
|
4437 p = aff_entry->ae_add
|
|
|
4438 + STRLEN(aff_entry->ae_add);
|
|
|
4439 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
4440 if (PTR2CHAR(p) == c_up)
|
|
|
4441 {
|
|
|
4442 upper = TRUE;
|
|
|
4443 aff_entry->ae_chop = NULL;
|
|
|
4444 *p = NUL;
|
|
|
4445
|
|
|
4446 /* The condition is matched with the
|
|
|
4447 * actual word, thus must check for the
|
|
|
4448 * upper-case letter. */
|
|
|
4449 if (aff_entry->ae_cond != NULL)
|
|
|
4450 {
|
|
|
4451 char_u buf[MAXLINELEN];
|
|
|
4452 #ifdef FEAT_MBYTE
|
|
|
4453 if (has_mbyte)
|
|
|
4454 {
|
|
|
4455 onecap_copy(items[4], buf, TRUE);
|
|
|
4456 aff_entry->ae_cond = getroom_save(
|
|
474
|
4457 spin, buf);
|
|
455
|
4458 }
|
|
|
4459 else
|
|
|
4460 #endif
|
|
|
4461 *aff_entry->ae_cond = c_up;
|
|
|
4462 if (aff_entry->ae_cond != NULL)
|
|
|
4463 {
|
|
|
4464 sprintf((char *)buf, "^%s",
|
|
339
|
4465 aff_entry->ae_cond);
|
|
455
|
4466 vim_free(aff_entry->ae_prog);
|
|
|
4467 aff_entry->ae_prog = vim_regcomp(
|
|
|
4468 buf, RE_MAGIC + RE_STRING);
|
|
|
4469 }
|
|
|
4470 }
|
|
|
4471 }
|
|
|
4472 }
|
|
339
|
4473 }
|
|
|
4474
|
|
455
|
4475 if (aff_entry->ae_chop == NULL)
|
|
390
|
4476 {
|
|
455
|
4477 int idx;
|
|
|
4478 char_u **pp;
|
|
|
4479 int n;
|
|
|
4480
|
|
|
4481 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
4482 --idx)
|
|
|
4483 {
|
|
|
4484 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
4485 if (str_equal(p, aff_entry->ae_cond))
|
|
|
4486 break;
|
|
|
4487 }
|
|
|
4488 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
4489 {
|
|
|
4490 /* Not found, add a new condition. */
|
|
|
4491 idx = spin->si_prefcond.ga_len++;
|
|
|
4492 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
4493 + idx;
|
|
|
4494 if (aff_entry->ae_cond == NULL)
|
|
|
4495 *pp = NULL;
|
|
|
4496 else
|
|
474
|
4497 *pp = getroom_save(spin,
|
|
455
|
4498 aff_entry->ae_cond);
|
|
|
4499 }
|
|
|
4500
|
|
|
4501 /* Add the prefix to the prefix tree. */
|
|
|
4502 if (aff_entry->ae_add == NULL)
|
|
|
4503 p = (char_u *)"";
|
|
390
|
4504 else
|
|
455
|
4505 p = aff_entry->ae_add;
|
|
|
4506 /* PFX_FLAGS is a negative number, so that
|
|
|
4507 * tree_add_word() knows this is the prefix tree. */
|
|
|
4508 n = PFX_FLAGS;
|
|
|
4509 if (rare)
|
|
|
4510 n |= WFP_RARE;
|
|
|
4511 if (!cur_aff->ah_combine)
|
|
|
4512 n |= WFP_NC;
|
|
|
4513 if (upper)
|
|
|
4514 n |= WFP_UP;
|
|
474
|
4515 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
4516 idx, cur_aff->ah_newID);
|
|
390
|
4517 }
|
|
339
|
4518 }
|
|
240
|
4519 }
|
|
236
|
4520 }
|
|
255
|
4521 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
|
4522 {
|
|
|
4523 if (fol != NULL)
|
|
|
4524 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
|
4525 fname, lnum);
|
|
|
4526 else
|
|
|
4527 fol = vim_strsave(items[1]);
|
|
|
4528 }
|
|
|
4529 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
|
4530 {
|
|
|
4531 if (low != NULL)
|
|
|
4532 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
|
4533 fname, lnum);
|
|
|
4534 else
|
|
|
4535 low = vim_strsave(items[1]);
|
|
|
4536 }
|
|
|
4537 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
|
4538 {
|
|
|
4539 if (upp != NULL)
|
|
|
4540 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
|
4541 fname, lnum);
|
|
|
4542 else
|
|
|
4543 upp = vim_strsave(items[1]);
|
|
|
4544 }
|
|
236
|
4545 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
4546 {
|
|
236
|
4547 /* Ignore REP count */;
|
|
323
|
4548 if (!isdigit(*items[1]))
|
|
|
4549 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
4550 fname, lnum);
|
|
|
4551 }
|
|
482
|
4552 else if (STRCMP(items[0], "REP") == 0 && itemcnt >= 3)
|
|
236
|
4553 {
|
|
|
4554 /* REP item */
|
|
482
|
4555 /* Myspell ignores extra arguments, we require it starts with
|
|
|
4556 * # to detect mistakes. */
|
|
|
4557 if (itemcnt > 3 && items[3][0] != '#')
|
|
|
4558 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]);
|
|
323
|
4559 if (do_rep)
|
|
|
4560 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
|
4561 }
|
|
|
4562 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
4563 {
|
|
|
4564 /* MAP item or count */
|
|
|
4565 if (!found_map)
|
|
|
4566 {
|
|
|
4567 /* First line contains the count. */
|
|
|
4568 found_map = TRUE;
|
|
|
4569 if (!isdigit(*items[1]))
|
|
|
4570 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
4571 fname, lnum);
|
|
|
4572 }
|
|
|
4573 else if (do_map)
|
|
|
4574 {
|
|
346
|
4575 int c;
|
|
|
4576
|
|
|
4577 /* Check that every character appears only once. */
|
|
|
4578 for (p = items[1]; *p != NUL; )
|
|
|
4579 {
|
|
|
4580 #ifdef FEAT_MBYTE
|
|
|
4581 c = mb_ptr2char_adv(&p);
|
|
|
4582 #else
|
|
|
4583 c = *p++;
|
|
|
4584 #endif
|
|
|
4585 if ((spin->si_map.ga_len > 0
|
|
|
4586 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
4587 != NULL)
|
|
|
4588 || vim_strchr(p, c) != NULL)
|
|
|
4589 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
4590 fname, lnum);
|
|
|
4591 }
|
|
|
4592
|
|
323
|
4593 /* We simply concatenate all the MAP strings, separated by
|
|
|
4594 * slashes. */
|
|
|
4595 ga_concat(&spin->si_map, items[1]);
|
|
|
4596 ga_append(&spin->si_map, '/');
|
|
|
4597 }
|
|
|
4598 }
|
|
|
4599 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
|
4600 {
|
|
|
4601 if (do_sal)
|
|
|
4602 {
|
|
|
4603 /* SAL item (sounds-a-like)
|
|
|
4604 * Either one of the known keys or a from-to pair. */
|
|
|
4605 if (STRCMP(items[1], "followup") == 0)
|
|
|
4606 spin->si_followup = sal_to_bool(items[2]);
|
|
|
4607 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
4608 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
4609 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
4610 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
4611 else
|
|
|
4612 /* when "to" is "_" it means empty */
|
|
|
4613 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
4614 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
4615 : items[2]);
|
|
|
4616 }
|
|
236
|
4617 }
|
|
375
|
4618 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
|
4619 && (!do_sofo || spin->si_sofofr == NULL))
|
|
|
4620 {
|
|
|
4621 if (do_sofo)
|
|
|
4622 spin->si_sofofr = vim_strsave(items[1]);
|
|
|
4623 }
|
|
|
4624 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
|
4625 && (!do_sofo || spin->si_sofoto == NULL))
|
|
|
4626 {
|
|
|
4627 if (do_sofo)
|
|
|
4628 spin->si_sofoto = vim_strsave(items[1]);
|
|
|
4629 }
|
|
300
|
4630 else
|
|
481
|
4631 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"),
|
|
236
|
4632 fname, lnum, items[0]);
|
|
|
4633 }
|
|
|
4634 }
|
|
|
4635
|
|
375
|
4636 if (do_sofo && (spin->si_sofofr == NULL) != (spin->si_sofoto == NULL))
|
|
|
4637 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
4638 spin->si_sofofr == NULL ? "FROM" : "TO", fname);
|
|
|
4639 if (spin->si_sofofr != NULL && spin->si_sal.ga_len > 0)
|
|
|
4640 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
|
4641
|
|
255
|
4642 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
4643 {
|
|
351
|
4644 if (spin->si_clear_chartab)
|
|
|
4645 {
|
|
|
4646 /* Clear the char type tables, don't want to use any of the
|
|
|
4647 * currently used spell properties. */
|
|
|
4648 init_spell_chartab();
|
|
|
4649 spin->si_clear_chartab = FALSE;
|
|
|
4650 }
|
|
|
4651
|
|
316
|
4652 /*
|
|
|
4653 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
4654 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
4655 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
4656 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
4657 */
|
|
|
4658 if (!spin->si_ascii
|
|
|
4659 #ifdef FEAT_MBYTE
|
|
|
4660 && !enc_utf8
|
|
|
4661 #endif
|
|
|
4662 )
|
|
260
|
4663 {
|
|
|
4664 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
4665 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
4666 else
|
|
316
|
4667 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
4668 }
|
|
255
|
4669
|
|
|
4670 vim_free(fol);
|
|
|
4671 vim_free(low);
|
|
|
4672 vim_free(upp);
|
|
|
4673 }
|
|
|
4674
|
|
481
|
4675 /* Use compound specifications of the .aff file for the spell info. */
|
|
483
|
4676 if (aff->af_compmax != 0)
|
|
|
4677 {
|
|
|
4678 if (spin->si_compmax != 0 && spin->si_compmax != aff->af_compmax)
|
|
|
4679 smsg((char_u *)_("COMPOUNDMAX value differs from what is used in another .aff file"));
|
|
|
4680 else
|
|
|
4681 spin->si_compmax = aff->af_compmax;
|
|
|
4682 }
|
|
|
4683
|
|
481
|
4684 if (aff->af_compminlen != 0)
|
|
|
4685 {
|
|
|
4686 if (spin->si_compminlen != 0
|
|
|
4687 && spin->si_compminlen != aff->af_compminlen)
|
|
|
4688 smsg((char_u *)_("COMPOUNDMIN value differs from what is used in another .aff file"));
|
|
|
4689 else
|
|
|
4690 spin->si_compminlen = aff->af_compminlen;
|
|
|
4691 }
|
|
|
4692
|
|
483
|
4693 if (aff->af_compsylmax != 0)
|
|
|
4694 {
|
|
|
4695 if (aff->af_syllable == NULL)
|
|
|
4696 smsg((char_u *)_("COMPOUNDSYLMAX without SYLLABLE"));
|
|
|
4697
|
|
|
4698 if (spin->si_compsylmax != 0
|
|
|
4699 && spin->si_compsylmax != aff->af_compsylmax)
|
|
|
4700 smsg((char_u *)_("COMPOUNDSYLMAX value differs from what is used in another .aff file"));
|
|
|
4701 else
|
|
|
4702 spin->si_compsylmax = aff->af_compsylmax;
|
|
|
4703 }
|
|
|
4704
|
|
481
|
4705 if (aff->af_compflags != NULL)
|
|
|
4706 {
|
|
|
4707 if (spin->si_compflags != NULL
|
|
|
4708 && STRCMP(spin->si_compflags, aff->af_compflags) != 0)
|
|
|
4709 smsg((char_u *)_("COMPOUNDFLAG(S) value differs from what is used in another .aff file"));
|
|
|
4710 else
|
|
|
4711 spin->si_compflags = aff->af_compflags;
|
|
|
4712 }
|
|
|
4713
|
|
483
|
4714 if (aff->af_syllable != NULL)
|
|
|
4715 {
|
|
|
4716 if (spin->si_syllable != NULL
|
|
|
4717 && STRCMP(spin->si_syllable, aff->af_syllable) != 0)
|
|
|
4718 smsg((char_u *)_("SYLLABLE value differs from what is used in another .aff file"));
|
|
|
4719 else
|
|
|
4720 spin->si_syllable = aff->af_syllable;
|
|
|
4721 }
|
|
|
4722
|
|
236
|
4723 vim_free(pc);
|
|
|
4724 fclose(fd);
|
|
|
4725 return aff;
|
|
|
4726 }
|
|
|
4727
|
|
|
4728 /*
|
|
339
|
4729 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
4730 * NULL as equal.
|
|
|
4731 */
|
|
|
4732 static int
|
|
|
4733 str_equal(s1, s2)
|
|
|
4734 char_u *s1;
|
|
|
4735 char_u *s2;
|
|
|
4736 {
|
|
|
4737 if (s1 == NULL || s2 == NULL)
|
|
|
4738 return s1 == s2;
|
|
|
4739 return STRCMP(s1, s2) == 0;
|
|
|
4740 }
|
|
|
4741
|
|
|
4742 /*
|
|
323
|
4743 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
4744 * They are stored case-folded.
|
|
|
4745 */
|
|
|
4746 static void
|
|
|
4747 add_fromto(spin, gap, from, to)
|
|
|
4748 spellinfo_T *spin;
|
|
|
4749 garray_T *gap;
|
|
|
4750 char_u *from;
|
|
|
4751 char_u *to;
|
|
|
4752 {
|
|
|
4753 fromto_T *ftp;
|
|
|
4754 char_u word[MAXWLEN];
|
|
|
4755
|
|
|
4756 if (ga_grow(gap, 1) == OK)
|
|
|
4757 {
|
|
|
4758 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
4759 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
4760 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
4761 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
4762 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
4763 ++gap->ga_len;
|
|
|
4764 }
|
|
|
4765 }
|
|
|
4766
|
|
|
4767 /*
|
|
|
4768 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
4769 */
|
|
|
4770 static int
|
|
|
4771 sal_to_bool(s)
|
|
|
4772 char_u *s;
|
|
|
4773 {
|
|
|
4774 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
4775 }
|
|
|
4776
|
|
|
4777 /*
|
|
240
|
4778 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
4779 * When "s" is NULL FALSE is returned.
|
|
|
4780 */
|
|
|
4781 static int
|
|
|
4782 has_non_ascii(s)
|
|
|
4783 char_u *s;
|
|
|
4784 {
|
|
|
4785 char_u *p;
|
|
|
4786
|
|
|
4787 if (s != NULL)
|
|
|
4788 for (p = s; *p != NUL; ++p)
|
|
|
4789 if (*p >= 128)
|
|
|
4790 return TRUE;
|
|
|
4791 return FALSE;
|
|
|
4792 }
|
|
|
4793
|
|
|
4794 /*
|
|
236
|
4795 * Free the structure filled by spell_read_aff().
|
|
|
4796 */
|
|
|
4797 static void
|
|
|
4798 spell_free_aff(aff)
|
|
|
4799 afffile_T *aff;
|
|
|
4800 {
|
|
|
4801 hashtab_T *ht;
|
|
|
4802 hashitem_T *hi;
|
|
|
4803 int todo;
|
|
|
4804 affheader_T *ah;
|
|
300
|
4805 affentry_T *ae;
|
|
236
|
4806
|
|
|
4807 vim_free(aff->af_enc);
|
|
|
4808
|
|
339
|
4809 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
4810 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
4811 {
|
|
|
4812 todo = ht->ht_used;
|
|
|
4813 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
4814 {
|
|
|
4815 if (!HASHITEM_EMPTY(hi))
|
|
|
4816 {
|
|
|
4817 --todo;
|
|
|
4818 ah = HI2AH(hi);
|
|
300
|
4819 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
4820 vim_free(ae->ae_prog);
|
|
236
|
4821 }
|
|
|
4822 }
|
|
|
4823 if (ht == &aff->af_suff)
|
|
|
4824 break;
|
|
|
4825 }
|
|
300
|
4826
|
|
236
|
4827 hash_clear(&aff->af_pref);
|
|
|
4828 hash_clear(&aff->af_suff);
|
|
|
4829 }
|
|
|
4830
|
|
|
4831 /*
|
|
300
|
4832 * Read dictionary file "fname".
|
|
236
|
4833 * Returns OK or FAIL;
|
|
|
4834 */
|
|
|
4835 static int
|
|
474
|
4836 spell_read_dic(spin, fname, affile)
|
|
|
4837 spellinfo_T *spin;
|
|
236
|
4838 char_u *fname;
|
|
300
|
4839 afffile_T *affile;
|
|
236
|
4840 {
|
|
300
|
4841 hashtab_T ht;
|
|
236
|
4842 char_u line[MAXLINELEN];
|
|
481
|
4843 char_u *p;
|
|
300
|
4844 char_u *afflist;
|
|
483
|
4845 char_u store_afflist[MAXWLEN];
|
|
|
4846 int pfxlen;
|
|
|
4847 int need_affix;
|
|
300
|
4848 char_u *dw;
|
|
236
|
4849 char_u *pc;
|
|
|
4850 char_u *w;
|
|
|
4851 int l;
|
|
|
4852 hash_T hash;
|
|
|
4853 hashitem_T *hi;
|
|
|
4854 FILE *fd;
|
|
|
4855 int lnum = 1;
|
|
300
|
4856 int non_ascii = 0;
|
|
|
4857 int retval = OK;
|
|
|
4858 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
4859 int flags;
|
|
481
|
4860 int duplicate = 0;
|
|
236
|
4861
|
|
300
|
4862 /*
|
|
|
4863 * Open the file.
|
|
|
4864 */
|
|
310
|
4865 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4866 if (fd == NULL)
|
|
|
4867 {
|
|
|
4868 EMSG2(_(e_notopen), fname);
|
|
|
4869 return FAIL;
|
|
|
4870 }
|
|
|
4871
|
|
300
|
4872 /* The hashtable is only used to detect duplicated words. */
|
|
|
4873 hash_init(&ht);
|
|
|
4874
|
|
310
|
4875 if (spin->si_verbose || p_verbose > 2)
|
|
|
4876 {
|
|
|
4877 if (!spin->si_verbose)
|
|
|
4878 verbose_enter();
|
|
366
|
4879 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
4880 out_flush();
|
|
|
4881 if (!spin->si_verbose)
|
|
|
4882 verbose_leave();
|
|
|
4883 }
|
|
236
|
4884
|
|
474
|
4885 /* start with a message for the first line */
|
|
|
4886 spin->si_msg_count = 999999;
|
|
|
4887
|
|
236
|
4888 /* Read and ignore the first line: word count. */
|
|
|
4889 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
4890 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
4891 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
4892
|
|
|
4893 /*
|
|
|
4894 * Read all the lines in the file one by one.
|
|
|
4895 * The words are converted to 'encoding' here, before being added to
|
|
|
4896 * the hashtable.
|
|
|
4897 */
|
|
255
|
4898 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
4899 {
|
|
255
|
4900 line_breakcheck();
|
|
236
|
4901 ++lnum;
|
|
455
|
4902 if (line[0] == '#' || line[0] == '/')
|
|
351
|
4903 continue; /* comment line */
|
|
236
|
4904
|
|
300
|
4905 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
4906 * the word is kept to allow e.g., "et al.". */
|
|
236
|
4907 l = STRLEN(line);
|
|
|
4908 while (l > 0 && line[l - 1] <= ' ')
|
|
|
4909 --l;
|
|
|
4910 if (l == 0)
|
|
|
4911 continue; /* empty line */
|
|
|
4912 line[l] = NUL;
|
|
|
4913
|
|
481
|
4914 /* Find the optional affix names. Replace the SLASH character by a
|
|
|
4915 * slash. */
|
|
|
4916 afflist = NULL;
|
|
|
4917 for (p = line; *p != NUL; mb_ptr_adv(p))
|
|
|
4918 {
|
|
|
4919 if (*p == affile->af_slash)
|
|
|
4920 *p = '/';
|
|
|
4921 else if (*p == '/')
|
|
|
4922 {
|
|
|
4923 *p = NUL;
|
|
|
4924 afflist = p + 1;
|
|
|
4925 break;
|
|
|
4926 }
|
|
|
4927 }
|
|
236
|
4928
|
|
300
|
4929 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
4930 if (spin->si_ascii && has_non_ascii(line))
|
|
|
4931 {
|
|
|
4932 ++non_ascii;
|
|
240
|
4933 continue;
|
|
300
|
4934 }
|
|
240
|
4935
|
|
310
|
4936 #ifdef FEAT_MBYTE
|
|
236
|
4937 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
4938 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4939 {
|
|
300
|
4940 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
4941 if (pc == NULL)
|
|
|
4942 {
|
|
|
4943 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4944 fname, lnum, line);
|
|
|
4945 continue;
|
|
|
4946 }
|
|
236
|
4947 w = pc;
|
|
|
4948 }
|
|
|
4949 else
|
|
310
|
4950 #endif
|
|
236
|
4951 {
|
|
|
4952 pc = NULL;
|
|
|
4953 w = line;
|
|
|
4954 }
|
|
|
4955
|
|
474
|
4956 /* This takes time, print a message every 10000 words. */
|
|
|
4957 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
4958 {
|
|
|
4959 spin->si_msg_count = 0;
|
|
339
|
4960 vim_snprintf((char *)message, sizeof(message),
|
|
|
4961 _("line %6d, word %6d - %s"),
|
|
|
4962 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
4963 msg_start();
|
|
|
4964 msg_puts_long_attr(message, 0);
|
|
|
4965 msg_clr_eos();
|
|
|
4966 msg_didout = FALSE;
|
|
|
4967 msg_col = 0;
|
|
|
4968 out_flush();
|
|
|
4969 }
|
|
|
4970
|
|
300
|
4971 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
4972 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
4973 if (dw == NULL)
|
|
300
|
4974 retval = FAIL;
|
|
|
4975 vim_free(pc);
|
|
|
4976 if (retval == FAIL)
|
|
236
|
4977 break;
|
|
|
4978
|
|
300
|
4979 hash = hash_hash(dw);
|
|
|
4980 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
4981 if (!HASHITEM_EMPTY(hi))
|
|
481
|
4982 {
|
|
|
4983 if (p_verbose > 0)
|
|
|
4984 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
4985 fname, lnum, dw);
|
|
481
|
4986 else if (duplicate == 0)
|
|
|
4987 smsg((char_u *)_("First duplicate word in %s line %d: %s"),
|
|
|
4988 fname, lnum, dw);
|
|
|
4989 ++duplicate;
|
|
|
4990 }
|
|
236
|
4991 else
|
|
300
|
4992 hash_add_item(&ht, hi, dw, hash);
|
|
|
4993
|
|
307
|
4994 flags = 0;
|
|
483
|
4995 store_afflist[0] = NUL;
|
|
|
4996 pfxlen = 0;
|
|
|
4997 need_affix = FALSE;
|
|
307
|
4998 if (afflist != NULL)
|
|
|
4999 {
|
|
|
5000 /* Check for affix name that stands for keep-case word and stands
|
|
|
5001 * for rare word (if defined). */
|
|
310
|
5002 if (affile->af_kep != NUL
|
|
|
5003 && vim_strchr(afflist, affile->af_kep) != NULL)
|
|
389
|
5004 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
307
|
5005 if (affile->af_rar != NUL
|
|
|
5006 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
|
5007 flags |= WF_RARE;
|
|
346
|
5008 if (affile->af_bad != NUL
|
|
|
5009 && vim_strchr(afflist, affile->af_bad) != NULL)
|
|
|
5010 flags |= WF_BANNED;
|
|
483
|
5011 if (affile->af_needaffix != NUL
|
|
|
5012 && vim_strchr(afflist, affile->af_needaffix) != NULL)
|
|
|
5013 need_affix = TRUE;
|
|
339
|
5014
|
|
|
5015 if (affile->af_pfxpostpone)
|
|
|
5016 /* Need to store the list of prefix IDs with the word. */
|
|
483
|
5017 pfxlen = get_pfxlist(affile, afflist, store_afflist);
|
|
|
5018
|
|
|
5019 if (spin->si_compflags != NULL)
|
|
|
5020 /* Need to store the list of compound flags with the word.
|
|
|
5021 * Concatenate them to the list of prefix IDs. */
|
|
|
5022 get_compflags(spin, afflist, store_afflist + pfxlen);
|
|
307
|
5023 }
|
|
|
5024
|
|
300
|
5025 /* Add the word to the word tree(s). */
|
|
483
|
5026 if (store_word(spin, dw, flags, spin->si_region,
|
|
|
5027 store_afflist, need_affix) == FAIL)
|
|
300
|
5028 retval = FAIL;
|
|
236
|
5029
|
|
300
|
5030 if (afflist != NULL)
|
|
|
5031 {
|
|
|
5032 /* Find all matching suffixes and add the resulting words.
|
|
|
5033 * Additionally do matching prefixes that combine. */
|
|
474
|
5034 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
5035 &affile->af_suff, &affile->af_pref,
|
|
483
|
5036 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5037 retval = FAIL;
|
|
|
5038
|
|
|
5039 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
5040 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
5041 &affile->af_pref, NULL,
|
|
483
|
5042 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5043 retval = FAIL;
|
|
|
5044 }
|
|
236
|
5045 }
|
|
|
5046
|
|
481
|
5047 if (duplicate > 0)
|
|
|
5048 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname);
|
|
300
|
5049 if (spin->si_ascii && non_ascii > 0)
|
|
481
|
5050 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"),
|
|
|
5051 non_ascii, fname);
|
|
300
|
5052 hash_clear(&ht);
|
|
|
5053
|
|
236
|
5054 fclose(fd);
|
|
300
|
5055 return retval;
|
|
236
|
5056 }
|
|
|
5057
|
|
|
5058 /*
|
|
339
|
5059 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
5060 * Used for PFXPOSTPONE.
|
|
483
|
5061 * Put the resulting flags in "store_afflist[MAXWLEN]" with a terminating NUL
|
|
|
5062 * and return the number of affixes.
|
|
|
5063 */
|
|
|
5064 static int
|
|
|
5065 get_pfxlist(affile, afflist, store_afflist)
|
|
339
|
5066 afffile_T *affile;
|
|
|
5067 char_u *afflist;
|
|
483
|
5068 char_u *store_afflist;
|
|
339
|
5069 {
|
|
|
5070 char_u *p;
|
|
483
|
5071 int cnt = 0;
|
|
339
|
5072 char_u key[2];
|
|
|
5073 hashitem_T *hi;
|
|
|
5074
|
|
|
5075 key[1] = NUL;
|
|
483
|
5076 for (p = afflist; *p != NUL; ++p)
|
|
|
5077 {
|
|
|
5078 key[0] = *p;
|
|
|
5079 hi = hash_find(&affile->af_pref, key);
|
|
|
5080 if (!HASHITEM_EMPTY(hi))
|
|
|
5081 /* This is a prefix ID, use the new number. */
|
|
|
5082 store_afflist[cnt++] = HI2AH(hi)->ah_newID;
|
|
|
5083 }
|
|
|
5084
|
|
|
5085 store_afflist[cnt] = NUL;
|
|
|
5086 return cnt;
|
|
339
|
5087 }
|
|
|
5088
|
|
|
5089 /*
|
|
481
|
5090 * Get the list of affix IDs from the affix list "afflist" that are used for
|
|
|
5091 * compound words.
|
|
483
|
5092 * Puts the flags in "store_afflist[]".
|
|
|
5093 */
|
|
|
5094 static void
|
|
|
5095 get_compflags(spin, afflist, store_afflist)
|
|
481
|
5096 spellinfo_T *spin;
|
|
|
5097 char_u *afflist;
|
|
483
|
5098 char_u *store_afflist;
|
|
481
|
5099 {
|
|
|
5100 char_u *p;
|
|
483
|
5101 int cnt = 0;
|
|
|
5102
|
|
|
5103 for (p = afflist; *p != NUL; ++p)
|
|
|
5104 /* A flag is a compound flag if it appears in "si_compflags" and
|
|
|
5105 * it's not a special character. */
|
|
|
5106 if (vim_strchr(spin->si_compflags, *p) != NULL
|
|
|
5107 && vim_strchr((char_u *)"+*[]/", *p) == NULL)
|
|
|
5108 store_afflist[cnt++] = *p;
|
|
|
5109
|
|
|
5110 store_afflist[cnt] = NUL;
|
|
481
|
5111 }
|
|
|
5112
|
|
|
5113 /*
|
|
300
|
5114 * Apply affixes to a word and store the resulting words.
|
|
|
5115 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
5116 * prefixes or suffixes.
|
|
|
5117 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
5118 * the resulting words for combining affixes.
|
|
|
5119 *
|
|
|
5120 * Returns FAIL when out of memory.
|
|
236
|
5121 */
|
|
300
|
5122 static int
|
|
483
|
5123 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags,
|
|
|
5124 pfxlist, pfxlen)
|
|
474
|
5125 spellinfo_T *spin; /* spell info */
|
|
300
|
5126 char_u *word; /* basic word start */
|
|
|
5127 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
5128 afffile_T *affile;
|
|
300
|
5129 hashtab_T *ht;
|
|
|
5130 hashtab_T *xht;
|
|
|
5131 int comb; /* only use affixes that combine */
|
|
307
|
5132 int flags; /* flags for the word */
|
|
339
|
5133 char_u *pfxlist; /* list of prefix IDs */
|
|
483
|
5134 int pfxlen; /* nr of flags in "pfxlist" for prefixes, rest
|
|
|
5135 * is compound flags */
|
|
236
|
5136 {
|
|
|
5137 int todo;
|
|
|
5138 hashitem_T *hi;
|
|
300
|
5139 affheader_T *ah;
|
|
|
5140 affentry_T *ae;
|
|
|
5141 regmatch_T regmatch;
|
|
|
5142 char_u newword[MAXWLEN];
|
|
|
5143 int retval = OK;
|
|
|
5144 int i;
|
|
|
5145 char_u *p;
|
|
366
|
5146 int use_flags;
|
|
390
|
5147 char_u *use_pfxlist;
|
|
483
|
5148 char_u pfx_pfxlist[MAXWLEN];
|
|
455
|
5149 int c;
|
|
483
|
5150 size_t wordlen = STRLEN(word);
|
|
236
|
5151
|
|
300
|
5152 todo = ht->ht_used;
|
|
|
5153 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
5154 {
|
|
|
5155 if (!HASHITEM_EMPTY(hi))
|
|
|
5156 {
|
|
|
5157 --todo;
|
|
300
|
5158 ah = HI2AH(hi);
|
|
236
|
5159
|
|
300
|
5160 /* Check that the affix combines, if required, and that the word
|
|
|
5161 * supports this affix. */
|
|
455
|
5162 c = PTR2CHAR(ah->ah_key);
|
|
|
5163 if ((!comb || ah->ah_combine) && vim_strchr(afflist, c) != NULL)
|
|
236
|
5164 {
|
|
300
|
5165 /* Loop over all affix entries with this name. */
|
|
|
5166 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
5167 {
|
|
300
|
5168 /* Check the condition. It's not logical to match case
|
|
|
5169 * here, but it is required for compatibility with
|
|
339
|
5170 * Myspell.
|
|
481
|
5171 * Another requirement from Myspell is that the chop
|
|
|
5172 * string is shorter than the word itself.
|
|
339
|
5173 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
5174 * prefixes with a chop string. */
|
|
300
|
5175 regmatch.regprog = ae->ae_prog;
|
|
|
5176 regmatch.rm_ic = FALSE;
|
|
339
|
5177 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
5178 || ae->ae_chop != NULL)
|
|
481
|
5179 && (ae->ae_chop == NULL
|
|
|
5180 || STRLEN(ae->ae_chop) < wordlen)
|
|
339
|
5181 && (ae->ae_prog == NULL
|
|
|
5182 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
5183 {
|
|
|
5184 /* Match. Remove the chop and add the affix. */
|
|
|
5185 if (xht == NULL)
|
|
240
|
5186 {
|
|
300
|
5187 /* prefix: chop/add at the start of the word */
|
|
|
5188 if (ae->ae_add == NULL)
|
|
|
5189 *newword = NUL;
|
|
|
5190 else
|
|
|
5191 STRCPY(newword, ae->ae_add);
|
|
|
5192 p = word;
|
|
|
5193 if (ae->ae_chop != NULL)
|
|
310
|
5194 {
|
|
300
|
5195 /* Skip chop string. */
|
|
310
|
5196 #ifdef FEAT_MBYTE
|
|
|
5197 if (has_mbyte)
|
|
324
|
5198 {
|
|
310
|
5199 i = mb_charlen(ae->ae_chop);
|
|
324
|
5200 for ( ; i > 0; --i)
|
|
|
5201 mb_ptr_adv(p);
|
|
|
5202 }
|
|
310
|
5203 else
|
|
|
5204 #endif
|
|
324
|
5205 p += STRLEN(ae->ae_chop);
|
|
310
|
5206 }
|
|
300
|
5207 STRCAT(newword, p);
|
|
|
5208 }
|
|
|
5209 else
|
|
|
5210 {
|
|
|
5211 /* suffix: chop/add at the end of the word */
|
|
|
5212 STRCPY(newword, word);
|
|
|
5213 if (ae->ae_chop != NULL)
|
|
|
5214 {
|
|
|
5215 /* Remove chop string. */
|
|
|
5216 p = newword + STRLEN(newword);
|
|
376
|
5217 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
5218 for ( ; i > 0; --i)
|
|
300
|
5219 mb_ptr_back(newword, p);
|
|
|
5220 *p = NUL;
|
|
|
5221 }
|
|
|
5222 if (ae->ae_add != NULL)
|
|
|
5223 STRCAT(newword, ae->ae_add);
|
|
240
|
5224 }
|
|
|
5225
|
|
366
|
5226 /* Obey the "rare" flag of the affix. */
|
|
|
5227 if (ae->ae_rare)
|
|
|
5228 use_flags = flags | WF_RARE;
|
|
|
5229 else
|
|
|
5230 use_flags = flags;
|
|
483
|
5231
|
|
|
5232 /* Obey the "nocomp" flag of the affix: don't use the
|
|
|
5233 * compound flags. */
|
|
390
|
5234 use_pfxlist = pfxlist;
|
|
483
|
5235 if (ae->ae_nocomp && pfxlist != NULL)
|
|
|
5236 {
|
|
|
5237 vim_strncpy(pfx_pfxlist, pfxlist, pfxlen);
|
|
|
5238 use_pfxlist = pfx_pfxlist;
|
|
|
5239 }
|
|
390
|
5240
|
|
|
5241 /* When there are postponed prefixes... */
|
|
393
|
5242 if (spin->si_prefroot != NULL
|
|
|
5243 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
5244 {
|
|
|
5245 /* ... add a flag to indicate an affix was used. */
|
|
|
5246 use_flags |= WF_HAS_AFF;
|
|
|
5247
|
|
|
5248 /* ... don't use a prefix list if combining
|
|
483
|
5249 * affixes is not allowed. But do use the
|
|
|
5250 * compound flags after them. */
|
|
|
5251 if ((!ah->ah_combine || comb) && pfxlist != NULL)
|
|
|
5252 use_pfxlist += pfxlen;
|
|
390
|
5253 }
|
|
366
|
5254
|
|
300
|
5255 /* Store the modified word. */
|
|
474
|
5256 if (store_word(spin, newword, use_flags,
|
|
483
|
5257 spin->si_region, use_pfxlist, FALSE) == FAIL)
|
|
300
|
5258 retval = FAIL;
|
|
236
|
5259
|
|
300
|
5260 /* When added a suffix and combining is allowed also
|
|
|
5261 * try adding prefixes additionally. */
|
|
|
5262 if (xht != NULL && ah->ah_combine)
|
|
474
|
5263 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
5264 xht, NULL, TRUE,
|
|
483
|
5265 use_flags, use_pfxlist, pfxlen) == FAIL)
|
|
300
|
5266 retval = FAIL;
|
|
236
|
5267 }
|
|
|
5268 }
|
|
|
5269 }
|
|
|
5270 }
|
|
|
5271 }
|
|
|
5272
|
|
|
5273 return retval;
|
|
|
5274 }
|
|
|
5275
|
|
|
5276 /*
|
|
300
|
5277 * Read a file with a list of words.
|
|
236
|
5278 */
|
|
|
5279 static int
|
|
474
|
5280 spell_read_wordfile(spin, fname)
|
|
|
5281 spellinfo_T *spin;
|
|
300
|
5282 char_u *fname;
|
|
236
|
5283 {
|
|
300
|
5284 FILE *fd;
|
|
|
5285 long lnum = 0;
|
|
|
5286 char_u rline[MAXLINELEN];
|
|
|
5287 char_u *line;
|
|
|
5288 char_u *pc = NULL;
|
|
381
|
5289 char_u *p;
|
|
300
|
5290 int l;
|
|
|
5291 int retval = OK;
|
|
|
5292 int did_word = FALSE;
|
|
|
5293 int non_ascii = 0;
|
|
307
|
5294 int flags;
|
|
316
|
5295 int regionmask;
|
|
236
|
5296
|
|
300
|
5297 /*
|
|
|
5298 * Open the file.
|
|
|
5299 */
|
|
310
|
5300 fd = mch_fopen((char *)fname, "r");
|
|
300
|
5301 if (fd == NULL)
|
|
236
|
5302 {
|
|
300
|
5303 EMSG2(_(e_notopen), fname);
|
|
|
5304 return FAIL;
|
|
236
|
5305 }
|
|
|
5306
|
|
310
|
5307 if (spin->si_verbose || p_verbose > 2)
|
|
|
5308 {
|
|
|
5309 if (!spin->si_verbose)
|
|
|
5310 verbose_enter();
|
|
366
|
5311 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
5312 out_flush();
|
|
|
5313 if (!spin->si_verbose)
|
|
|
5314 verbose_leave();
|
|
|
5315 }
|
|
300
|
5316
|
|
|
5317 /*
|
|
|
5318 * Read all the lines in the file one by one.
|
|
|
5319 */
|
|
|
5320 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
5321 {
|
|
|
5322 line_breakcheck();
|
|
|
5323 ++lnum;
|
|
|
5324
|
|
|
5325 /* Skip comment lines. */
|
|
|
5326 if (*rline == '#')
|
|
|
5327 continue;
|
|
|
5328
|
|
|
5329 /* Remove CR, LF and white space from the end. */
|
|
|
5330 l = STRLEN(rline);
|
|
|
5331 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
5332 --l;
|
|
|
5333 if (l == 0)
|
|
|
5334 continue; /* empty or blank line */
|
|
|
5335 rline[l] = NUL;
|
|
|
5336
|
|
|
5337 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
5338 vim_free(pc);
|
|
310
|
5339 #ifdef FEAT_MBYTE
|
|
300
|
5340 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
5341 {
|
|
|
5342 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
5343 if (pc == NULL)
|
|
|
5344 {
|
|
|
5345 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5346 fname, lnum, rline);
|
|
|
5347 continue;
|
|
|
5348 }
|
|
|
5349 line = pc;
|
|
|
5350 }
|
|
|
5351 else
|
|
310
|
5352 #endif
|
|
300
|
5353 {
|
|
|
5354 pc = NULL;
|
|
|
5355 line = rline;
|
|
|
5356 }
|
|
|
5357
|
|
307
|
5358 if (*line == '/')
|
|
300
|
5359 {
|
|
307
|
5360 ++line;
|
|
|
5361 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
5362 {
|
|
|
5363 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
5364 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
5365 fname, lnum, line - 1);
|
|
300
|
5366 else if (did_word)
|
|
316
|
5367 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
5368 fname, lnum, line - 1);
|
|
300
|
5369 else
|
|
|
5370 {
|
|
310
|
5371 #ifdef FEAT_MBYTE
|
|
|
5372 char_u *enc;
|
|
|
5373
|
|
300
|
5374 /* Setup for conversion to 'encoding'. */
|
|
316
|
5375 line += 10;
|
|
|
5376 enc = enc_canonize(line);
|
|
300
|
5377 if (enc != NULL && !spin->si_ascii
|
|
|
5378 && convert_setup(&spin->si_conv, enc,
|
|
|
5379 p_enc) == FAIL)
|
|
|
5380 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
5381 fname, line, p_enc);
|
|
300
|
5382 vim_free(enc);
|
|
375
|
5383 spin->si_conv.vc_fail = TRUE;
|
|
310
|
5384 #else
|
|
|
5385 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
5386 #endif
|
|
300
|
5387 }
|
|
307
|
5388 continue;
|
|
300
|
5389 }
|
|
307
|
5390
|
|
316
|
5391 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
5392 {
|
|
|
5393 if (spin->si_region_count > 1)
|
|
|
5394 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
5395 fname, lnum, line);
|
|
|
5396 else
|
|
|
5397 {
|
|
|
5398 line += 8;
|
|
|
5399 if (STRLEN(line) > 16)
|
|
|
5400 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
5401 fname, lnum, line);
|
|
|
5402 else
|
|
|
5403 {
|
|
|
5404 spin->si_region_count = STRLEN(line) / 2;
|
|
|
5405 STRCPY(spin->si_region_name, line);
|
|
389
|
5406
|
|
|
5407 /* Adjust the mask for a word valid in all regions. */
|
|
|
5408 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
5409 }
|
|
|
5410 }
|
|
|
5411 continue;
|
|
|
5412 }
|
|
|
5413
|
|
381
|
5414 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
5415 fname, lnum, line - 1);
|
|
|
5416 continue;
|
|
|
5417 }
|
|
|
5418
|
|
|
5419 flags = 0;
|
|
|
5420 regionmask = spin->si_region;
|
|
|
5421
|
|
|
5422 /* Check for flags and region after a slash. */
|
|
|
5423 p = vim_strchr(line, '/');
|
|
|
5424 if (p != NULL)
|
|
|
5425 {
|
|
|
5426 *p++ = NUL;
|
|
|
5427 while (*p != NUL)
|
|
|
5428 {
|
|
|
5429 if (*p == '=') /* keep-case word */
|
|
389
|
5430 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
5431 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
5432 flags |= WF_BANNED;
|
|
|
5433 else if (*p == '?') /* Rare word. */
|
|
|
5434 flags |= WF_RARE;
|
|
|
5435 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
5436 {
|
|
381
|
5437 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
5438 regionmask = 0;
|
|
|
5439 flags |= WF_REGION;
|
|
|
5440
|
|
|
5441 l = *p - '0';
|
|
316
|
5442 if (l > spin->si_region_count)
|
|
|
5443 {
|
|
|
5444 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
5445 fname, lnum, p);
|
|
316
|
5446 break;
|
|
|
5447 }
|
|
|
5448 regionmask |= 1 << (l - 1);
|
|
|
5449 }
|
|
381
|
5450 else
|
|
|
5451 {
|
|
|
5452 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
5453 fname, lnum, p);
|
|
|
5454 break;
|
|
|
5455 }
|
|
|
5456 ++p;
|
|
307
|
5457 }
|
|
300
|
5458 }
|
|
|
5459
|
|
|
5460 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
5461 if (spin->si_ascii && has_non_ascii(line))
|
|
|
5462 {
|
|
|
5463 ++non_ascii;
|
|
|
5464 continue;
|
|
|
5465 }
|
|
|
5466
|
|
|
5467 /* Normal word: store it. */
|
|
483
|
5468 if (store_word(spin, line, flags, regionmask, NULL, FALSE) == FAIL)
|
|
300
|
5469 {
|
|
|
5470 retval = FAIL;
|
|
|
5471 break;
|
|
|
5472 }
|
|
|
5473 did_word = TRUE;
|
|
|
5474 }
|
|
|
5475
|
|
|
5476 vim_free(pc);
|
|
|
5477 fclose(fd);
|
|
|
5478
|
|
310
|
5479 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
5480 {
|
|
|
5481 if (p_verbose > 2)
|
|
|
5482 verbose_enter();
|
|
300
|
5483 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
5484 non_ascii);
|
|
310
|
5485 if (p_verbose > 2)
|
|
|
5486 verbose_leave();
|
|
|
5487 }
|
|
300
|
5488 return retval;
|
|
236
|
5489 }
|
|
|
5490
|
|
|
5491 /*
|
|
300
|
5492 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
5493 * This avoids calling free() for every little struct we use (and keeping
|
|
|
5494 * track of them).
|
|
300
|
5495 * The memory is cleared to all zeros.
|
|
|
5496 * Returns NULL when out of memory.
|
|
|
5497 */
|
|
|
5498 static void *
|
|
474
|
5499 getroom(spin, len, align)
|
|
|
5500 spellinfo_T *spin;
|
|
447
|
5501 size_t len; /* length needed */
|
|
|
5502 int align; /* align for pointer */
|
|
300
|
5503 {
|
|
|
5504 char_u *p;
|
|
474
|
5505 sblock_T *bl = spin->si_blocks;
|
|
300
|
5506
|
|
447
|
5507 if (align && bl != NULL)
|
|
|
5508 /* Round size up for alignment. On some systems structures need to be
|
|
|
5509 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
5510 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
5511 & ~(sizeof(char *) - 1);
|
|
|
5512
|
|
300
|
5513 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
5514 {
|
|
|
5515 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
5516 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
5517 if (bl == NULL)
|
|
|
5518 return NULL;
|
|
474
|
5519 bl->sb_next = spin->si_blocks;
|
|
|
5520 spin->si_blocks = bl;
|
|
300
|
5521 bl->sb_used = 0;
|
|
482
|
5522 ++spin->si_blocks_cnt;
|
|
300
|
5523 }
|
|
|
5524
|
|
|
5525 p = bl->sb_data + bl->sb_used;
|
|
|
5526 bl->sb_used += len;
|
|
|
5527
|
|
|
5528 return p;
|
|
|
5529 }
|
|
|
5530
|
|
|
5531 /*
|
|
|
5532 * Make a copy of a string into memory allocated with getroom().
|
|
|
5533 */
|
|
|
5534 static char_u *
|
|
474
|
5535 getroom_save(spin, s)
|
|
|
5536 spellinfo_T *spin;
|
|
300
|
5537 char_u *s;
|
|
|
5538 {
|
|
|
5539 char_u *sc;
|
|
|
5540
|
|
474
|
5541 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
5542 if (sc != NULL)
|
|
|
5543 STRCPY(sc, s);
|
|
|
5544 return sc;
|
|
|
5545 }
|
|
|
5546
|
|
|
5547
|
|
|
5548 /*
|
|
|
5549 * Free the list of allocated sblock_T.
|
|
236
|
5550 */
|
|
|
5551 static void
|
|
300
|
5552 free_blocks(bl)
|
|
|
5553 sblock_T *bl;
|
|
236
|
5554 {
|
|
300
|
5555 sblock_T *next;
|
|
236
|
5556
|
|
300
|
5557 while (bl != NULL)
|
|
236
|
5558 {
|
|
300
|
5559 next = bl->sb_next;
|
|
|
5560 vim_free(bl);
|
|
|
5561 bl = next;
|
|
236
|
5562 }
|
|
|
5563 }
|
|
|
5564
|
|
|
5565 /*
|
|
300
|
5566 * Allocate the root of a word tree.
|
|
236
|
5567 */
|
|
300
|
5568 static wordnode_T *
|
|
474
|
5569 wordtree_alloc(spin)
|
|
|
5570 spellinfo_T *spin;
|
|
|
5571 {
|
|
|
5572 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
5573 }
|
|
|
5574
|
|
|
5575 /*
|
|
300
|
5576 * Store a word in the tree(s).
|
|
390
|
5577 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
5578 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
5579 * used to find suggestions.
|
|
300
|
5580 * For a keep-case word also store it in the keep-case tree.
|
|
482
|
5581 * When "pfxlist" is not NULL store the word for each postponed prefix ID and
|
|
|
5582 * compound flag.
|
|
236
|
5583 */
|
|
|
5584 static int
|
|
483
|
5585 store_word(spin, word, flags, region, pfxlist, need_affix)
|
|
474
|
5586 spellinfo_T *spin;
|
|
300
|
5587 char_u *word;
|
|
307
|
5588 int flags; /* extra flags, WF_BANNED */
|
|
316
|
5589 int region; /* supported region(s) */
|
|
339
|
5590 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
483
|
5591 int need_affix; /* only store word with affix ID */
|
|
236
|
5592 {
|
|
300
|
5593 int len = STRLEN(word);
|
|
|
5594 int ct = captype(word, word + len);
|
|
|
5595 char_u foldword[MAXWLEN];
|
|
339
|
5596 int res = OK;
|
|
|
5597 char_u *p;
|
|
236
|
5598
|
|
323
|
5599 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
5600 for (p = pfxlist; res == OK; ++p)
|
|
|
5601 {
|
|
483
|
5602 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
5603 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
474
|
5604 region, p == NULL ? 0 : *p);
|
|
339
|
5605 if (p == NULL || *p == NUL)
|
|
|
5606 break;
|
|
|
5607 }
|
|
334
|
5608 ++spin->si_foldwcount;
|
|
307
|
5609
|
|
474
|
5610 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
5611 {
|
|
339
|
5612 for (p = pfxlist; res == OK; ++p)
|
|
|
5613 {
|
|
483
|
5614 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
5615 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
474
|
5616 region, p == NULL ? 0 : *p);
|
|
339
|
5617 if (p == NULL || *p == NUL)
|
|
|
5618 break;
|
|
|
5619 }
|
|
334
|
5620 ++spin->si_keepwcount;
|
|
|
5621 }
|
|
300
|
5622 return res;
|
|
236
|
5623 }
|
|
|
5624
|
|
|
5625 /*
|
|
300
|
5626 * Add word "word" to a word tree at "root".
|
|
366
|
5627 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
5628 * "rare" and "region" is the condition nr.
|
|
255
|
5629 * Returns FAIL when out of memory.
|
|
236
|
5630 */
|
|
255
|
5631 static int
|
|
481
|
5632 tree_add_word(spin, word, root, flags, region, affixID)
|
|
474
|
5633 spellinfo_T *spin;
|
|
300
|
5634 char_u *word;
|
|
|
5635 wordnode_T *root;
|
|
|
5636 int flags;
|
|
|
5637 int region;
|
|
481
|
5638 int affixID;
|
|
236
|
5639 {
|
|
300
|
5640 wordnode_T *node = root;
|
|
|
5641 wordnode_T *np;
|
|
469
|
5642 wordnode_T *copyp, **copyprev;
|
|
300
|
5643 wordnode_T **prev = NULL;
|
|
|
5644 int i;
|
|
255
|
5645
|
|
300
|
5646 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
5647 for (i = 0; ; ++i)
|
|
255
|
5648 {
|
|
469
|
5649 /* When there is more than one reference to this node we need to make
|
|
|
5650 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
5651 * (we don't optimize for a partly shared list of siblings). */
|
|
|
5652 if (node != NULL && node->wn_refs > 1)
|
|
|
5653 {
|
|
|
5654 --node->wn_refs;
|
|
|
5655 copyprev = prev;
|
|
|
5656 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
5657 {
|
|
|
5658 /* Allocate a new node and copy the info. */
|
|
474
|
5659 np = get_wordnode(spin);
|
|
469
|
5660 if (np == NULL)
|
|
|
5661 return FAIL;
|
|
|
5662 np->wn_child = copyp->wn_child;
|
|
|
5663 if (np->wn_child != NULL)
|
|
|
5664 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
5665 np->wn_byte = copyp->wn_byte;
|
|
|
5666 if (np->wn_byte == NUL)
|
|
|
5667 {
|
|
|
5668 np->wn_flags = copyp->wn_flags;
|
|
|
5669 np->wn_region = copyp->wn_region;
|
|
481
|
5670 np->wn_affixID = copyp->wn_affixID;
|
|
469
|
5671 }
|
|
|
5672
|
|
|
5673 /* Link the new node in the list, there will be one ref. */
|
|
|
5674 np->wn_refs = 1;
|
|
|
5675 *copyprev = np;
|
|
|
5676 copyprev = &np->wn_sibling;
|
|
|
5677
|
|
|
5678 /* Let "node" point to the head of the copied list. */
|
|
|
5679 if (copyp == node)
|
|
|
5680 node = np;
|
|
|
5681 }
|
|
|
5682 }
|
|
|
5683
|
|
300
|
5684 /* Look for the sibling that has the same character. They are sorted
|
|
|
5685 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
5686 * higher byte value. For zero bytes (end of word) the sorting is
|
|
481
|
5687 * done on flags and then on affixID. */
|
|
339
|
5688 while (node != NULL
|
|
|
5689 && (node->wn_byte < word[i]
|
|
|
5690 || (node->wn_byte == NUL
|
|
|
5691 && (flags < 0
|
|
481
|
5692 ? node->wn_affixID < affixID
|
|
390
|
5693 : node->wn_flags < (flags & WN_MASK)
|
|
|
5694 || (node->wn_flags == (flags & WN_MASK)
|
|
481
|
5695 && node->wn_affixID < affixID)))))
|
|
236
|
5696 {
|
|
300
|
5697 prev = &node->wn_sibling;
|
|
|
5698 node = *prev;
|
|
236
|
5699 }
|
|
339
|
5700 if (node == NULL
|
|
|
5701 || node->wn_byte != word[i]
|
|
|
5702 || (word[i] == NUL
|
|
|
5703 && (flags < 0
|
|
390
|
5704 || node->wn_flags != (flags & WN_MASK)
|
|
481
|
5705 || node->wn_affixID != affixID)))
|
|
255
|
5706 {
|
|
300
|
5707 /* Allocate a new node. */
|
|
474
|
5708 np = get_wordnode(spin);
|
|
300
|
5709 if (np == NULL)
|
|
|
5710 return FAIL;
|
|
|
5711 np->wn_byte = word[i];
|
|
469
|
5712
|
|
|
5713 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
5714 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
5715 * make ref count of sibling one (matters when inserting in front
|
|
|
5716 * of the list of siblings). */
|
|
|
5717 if (node == NULL)
|
|
|
5718 np->wn_refs = 1;
|
|
|
5719 else
|
|
|
5720 {
|
|
|
5721 np->wn_refs = node->wn_refs;
|
|
|
5722 node->wn_refs = 1;
|
|
|
5723 }
|
|
300
|
5724 *prev = np;
|
|
|
5725 np->wn_sibling = node;
|
|
|
5726 node = np;
|
|
255
|
5727 }
|
|
300
|
5728
|
|
|
5729 if (word[i] == NUL)
|
|
|
5730 {
|
|
|
5731 node->wn_flags = flags;
|
|
|
5732 node->wn_region |= region;
|
|
481
|
5733 node->wn_affixID = affixID;
|
|
300
|
5734 break;
|
|
|
5735 }
|
|
|
5736 prev = &node->wn_child;
|
|
|
5737 node = *prev;
|
|
255
|
5738 }
|
|
469
|
5739 #ifdef SPELL_PRINTTREE
|
|
|
5740 smsg("Added \"%s\"", word);
|
|
|
5741 spell_print_tree(root->wn_sibling);
|
|
|
5742 #endif
|
|
|
5743
|
|
474
|
5744 /* count nr of words added since last message */
|
|
|
5745 ++spin->si_msg_count;
|
|
|
5746
|
|
482
|
5747 if (spin->si_compress_cnt > 1)
|
|
|
5748 {
|
|
|
5749 if (--spin->si_compress_cnt == 1)
|
|
|
5750 /* Did enough words to lower the block count limit. */
|
|
483
|
5751 spin->si_blocks_cnt += compress_inc;
|
|
482
|
5752 }
|
|
|
5753
|
|
469
|
5754 /*
|
|
482
|
5755 * When we have allocated lots of memory we need to compress the word tree
|
|
|
5756 * to free up some room. But compression is slow, and we might actually
|
|
|
5757 * need that room, thus only compress in the following situations:
|
|
|
5758 * 1. When not compressed before (si_compress_cnt == 0): when using
|
|
483
|
5759 * "compress_start" blocks.
|
|
|
5760 * 2. When compressed before and used "compress_inc" blocks before
|
|
|
5761 * adding "compress_added" words (si_compress_cnt > 1).
|
|
|
5762 * 3. When compressed before, added "compress_added" words
|
|
482
|
5763 * (si_compress_cnt == 1) and the number of free nodes drops below the
|
|
|
5764 * maximum word length.
|
|
469
|
5765 */
|
|
482
|
5766 #ifndef SPELL_PRINTTREE
|
|
|
5767 if (spin->si_compress_cnt == 1
|
|
|
5768 ? spin->si_free_count < MAXWLEN
|
|
483
|
5769 : spin->si_blocks_cnt >= compress_start)
|
|
482
|
5770 #endif
|
|
|
5771 {
|
|
|
5772 /* Decrement the block counter. The effect is that we compress again
|
|
483
|
5773 * when the freed up room has been used and another "compress_inc"
|
|
|
5774 * blocks have been allocated. Unless "compress_added" words have
|
|
482
|
5775 * been added, then the limit is put back again. */
|
|
483
|
5776 spin->si_blocks_cnt -= compress_inc;
|
|
|
5777 spin->si_compress_cnt = compress_added;
|
|
482
|
5778
|
|
474
|
5779 if (spin->si_verbose)
|
|
|
5780 {
|
|
|
5781 msg_start();
|
|
|
5782 msg_puts((char_u *)_(msg_compressing));
|
|
|
5783 msg_clr_eos();
|
|
|
5784 msg_didout = FALSE;
|
|
|
5785 msg_col = 0;
|
|
|
5786 out_flush();
|
|
|
5787 }
|
|
482
|
5788
|
|
|
5789 /* Compress both trees. Either they both have many nodes, which makes
|
|
|
5790 * compression useful, or one of them is small, which means
|
|
|
5791 * compression goes fast. */
|
|
|
5792 wordtree_compress(spin, spin->si_foldroot);
|
|
|
5793 wordtree_compress(spin, spin->si_keeproot);
|
|
469
|
5794 }
|
|
255
|
5795
|
|
|
5796 return OK;
|
|
236
|
5797 }
|
|
|
5798
|
|
469
|
5799 /*
|
|
483
|
5800 * Check the 'mkspellmem' option. Return FAIL if it's wrong.
|
|
|
5801 * Sets "sps_flags".
|
|
|
5802 */
|
|
|
5803 int
|
|
|
5804 spell_check_msm()
|
|
|
5805 {
|
|
|
5806 char_u *p = p_msm;
|
|
|
5807 long start = 0;
|
|
|
5808 long inc = 0;
|
|
|
5809 long added = 0;
|
|
|
5810
|
|
|
5811 if (!VIM_ISDIGIT(*p))
|
|
|
5812 return FAIL;
|
|
|
5813 /* block count = (value * 1024) / SBLOCKSIZE (but avoid overflow)*/
|
|
|
5814 start = (getdigits(&p) * 10) / (SBLOCKSIZE / 102);
|
|
|
5815 if (*p != ',')
|
|
|
5816 return FAIL;
|
|
|
5817 ++p;
|
|
|
5818 if (!VIM_ISDIGIT(*p))
|
|
|
5819 return FAIL;
|
|
|
5820 inc = (getdigits(&p) * 102) / (SBLOCKSIZE / 10);
|
|
|
5821 if (*p != ',')
|
|
|
5822 return FAIL;
|
|
|
5823 ++p;
|
|
|
5824 if (!VIM_ISDIGIT(*p))
|
|
|
5825 return FAIL;
|
|
|
5826 added = getdigits(&p) * 1024;
|
|
|
5827 if (*p != NUL)
|
|
|
5828 return FAIL;
|
|
|
5829
|
|
|
5830 if (start == 0 || inc == 0 || added == 0 || inc > start)
|
|
|
5831 return FAIL;
|
|
|
5832
|
|
|
5833 compress_start = start;
|
|
|
5834 compress_inc = inc;
|
|
|
5835 compress_added = added;
|
|
|
5836 return OK;
|
|
|
5837 }
|
|
|
5838
|
|
|
5839
|
|
|
5840 /*
|
|
469
|
5841 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
5842 * allocate a new one.
|
|
|
5843 */
|
|
|
5844 static wordnode_T *
|
|
474
|
5845 get_wordnode(spin)
|
|
|
5846 spellinfo_T *spin;
|
|
469
|
5847 {
|
|
|
5848 wordnode_T *n;
|
|
|
5849
|
|
474
|
5850 if (spin->si_first_free == NULL)
|
|
|
5851 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
5852 else
|
|
|
5853 {
|
|
474
|
5854 n = spin->si_first_free;
|
|
|
5855 spin->si_first_free = n->wn_child;
|
|
469
|
5856 vim_memset(n, 0, sizeof(wordnode_T));
|
|
482
|
5857 --spin->si_free_count;
|
|
469
|
5858 }
|
|
|
5859 #ifdef SPELL_PRINTTREE
|
|
474
|
5860 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
5861 #endif
|
|
|
5862 return n;
|
|
|
5863 }
|
|
|
5864
|
|
|
5865 /*
|
|
474
|
5866 * Decrement the reference count on a node (which is the head of a list of
|
|
|
5867 * siblings). If the reference count becomes zero free the node and its
|
|
|
5868 * siblings.
|
|
|
5869 */
|
|
|
5870 static void
|
|
|
5871 deref_wordnode(spin, node)
|
|
|
5872 spellinfo_T *spin;
|
|
|
5873 wordnode_T *node;
|
|
|
5874 {
|
|
|
5875 wordnode_T *np;
|
|
|
5876
|
|
|
5877 if (--node->wn_refs == 0)
|
|
|
5878 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
5879 {
|
|
|
5880 if (np->wn_child != NULL)
|
|
|
5881 deref_wordnode(spin, np->wn_child);
|
|
|
5882 free_wordnode(spin, np);
|
|
|
5883 }
|
|
|
5884 }
|
|
|
5885
|
|
|
5886 /*
|
|
469
|
5887 * Free a wordnode_T for re-use later.
|
|
474
|
5888 * Only the "wn_child" field becomes invalid.
|
|
469
|
5889 */
|
|
|
5890 static void
|
|
474
|
5891 free_wordnode(spin, n)
|
|
|
5892 spellinfo_T *spin;
|
|
|
5893 wordnode_T *n;
|
|
|
5894 {
|
|
|
5895 n->wn_child = spin->si_first_free;
|
|
|
5896 spin->si_first_free = n;
|
|
482
|
5897 ++spin->si_free_count;
|
|
469
|
5898 }
|
|
|
5899
|
|
236
|
5900 /*
|
|
300
|
5901 * Compress a tree: find tails that are identical and can be shared.
|
|
|
5902 */
|
|
|
5903 static void
|
|
474
|
5904 wordtree_compress(spin, root)
|
|
|
5905 spellinfo_T *spin;
|
|
300
|
5906 wordnode_T *root;
|
|
|
5907 {
|
|
|
5908 hashtab_T ht;
|
|
|
5909 int n;
|
|
|
5910 int tot = 0;
|
|
469
|
5911 int perc;
|
|
300
|
5912
|
|
474
|
5913 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
5914 * start of the tree. */
|
|
|
5915 if (root->wn_sibling != NULL)
|
|
300
|
5916 {
|
|
|
5917 hash_init(&ht);
|
|
474
|
5918 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
5919
|
|
|
5920 #ifndef SPELL_PRINTTREE
|
|
310
|
5921 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
5922 #endif
|
|
310
|
5923 {
|
|
|
5924 if (!spin->si_verbose)
|
|
|
5925 verbose_enter();
|
|
469
|
5926 if (tot > 1000000)
|
|
|
5927 perc = (tot - n) / (tot / 100);
|
|
482
|
5928 else if (tot == 0)
|
|
|
5929 perc = 0;
|
|
469
|
5930 else
|
|
|
5931 perc = (tot - n) * 100 / tot;
|
|
310
|
5932 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
5933 n, tot, perc);
|
|
310
|
5934 if (p_verbose > 2)
|
|
|
5935 verbose_leave();
|
|
|
5936 }
|
|
469
|
5937 #ifdef SPELL_PRINTTREE
|
|
474
|
5938 spell_print_tree(root->wn_sibling);
|
|
469
|
5939 #endif
|
|
300
|
5940 hash_clear(&ht);
|
|
|
5941 }
|
|
|
5942 }
|
|
|
5943
|
|
|
5944 /*
|
|
|
5945 * Compress a node, its siblings and its children, depth first.
|
|
|
5946 * Returns the number of compressed nodes.
|
|
236
|
5947 */
|
|
255
|
5948 static int
|
|
474
|
5949 node_compress(spin, node, ht, tot)
|
|
|
5950 spellinfo_T *spin;
|
|
300
|
5951 wordnode_T *node;
|
|
|
5952 hashtab_T *ht;
|
|
|
5953 int *tot; /* total count of nodes before compressing,
|
|
|
5954 incremented while going through the tree */
|
|
236
|
5955 {
|
|
300
|
5956 wordnode_T *np;
|
|
|
5957 wordnode_T *tp;
|
|
|
5958 wordnode_T *child;
|
|
|
5959 hash_T hash;
|
|
236
|
5960 hashitem_T *hi;
|
|
300
|
5961 int len = 0;
|
|
|
5962 unsigned nr, n;
|
|
|
5963 int compressed = 0;
|
|
236
|
5964
|
|
300
|
5965 /*
|
|
|
5966 * Go through the list of siblings. Compress each child and then try
|
|
|
5967 * finding an identical child to replace it.
|
|
|
5968 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
5969 * but the whole list of siblings, of which the node is the first.
|
|
|
5970 */
|
|
482
|
5971 for (np = node; np != NULL && !got_int; np = np->wn_sibling)
|
|
236
|
5972 {
|
|
300
|
5973 ++len;
|
|
|
5974 if ((child = np->wn_child) != NULL)
|
|
|
5975 {
|
|
346
|
5976 /* Compress the child. This fills hashkey. */
|
|
474
|
5977 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
5978
|
|
|
5979 /* Try to find an identical child. */
|
|
346
|
5980 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
5981 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
5982 tp = NULL;
|
|
|
5983 if (!HASHITEM_EMPTY(hi))
|
|
|
5984 {
|
|
|
5985 /* There are children with an identical hash value. Now check
|
|
|
5986 * if there is one that is really identical. */
|
|
346
|
5987 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
5988 if (node_equal(child, tp))
|
|
|
5989 {
|
|
|
5990 /* Found one! Now use that child in place of the
|
|
469
|
5991 * current one. This means the current child and all
|
|
|
5992 * its siblings is unlinked from the tree. */
|
|
|
5993 ++tp->wn_refs;
|
|
474
|
5994 deref_wordnode(spin, child);
|
|
300
|
5995 np->wn_child = tp;
|
|
|
5996 ++compressed;
|
|
|
5997 break;
|
|
|
5998 }
|
|
|
5999 if (tp == NULL)
|
|
|
6000 {
|
|
|
6001 /* No other child with this hash value equals the child of
|
|
|
6002 * the node, add it to the linked list after the first
|
|
|
6003 * item. */
|
|
|
6004 tp = HI2WN(hi);
|
|
346
|
6005 child->wn_u2.next = tp->wn_u2.next;
|
|
|
6006 tp->wn_u2.next = child;
|
|
300
|
6007 }
|
|
|
6008 }
|
|
|
6009 else
|
|
|
6010 /* No other child has this hash value, add it to the
|
|
|
6011 * hashtable. */
|
|
346
|
6012 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
6013 }
|
|
236
|
6014 }
|
|
300
|
6015 *tot += len;
|
|
|
6016
|
|
|
6017 /*
|
|
|
6018 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
6019 * find a lookalike node. This must be done after compressing the sibling
|
|
|
6020 * list, otherwise the hash key would become invalid by the compression.
|
|
|
6021 */
|
|
346
|
6022 node->wn_u1.hashkey[0] = len;
|
|
300
|
6023 nr = 0;
|
|
|
6024 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
6025 {
|
|
300
|
6026 if (np->wn_byte == NUL)
|
|
481
|
6027 /* end node: use wn_flags, wn_region and wn_affixID */
|
|
|
6028 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16);
|
|
300
|
6029 else
|
|
|
6030 /* byte node: use the byte value and the child pointer */
|
|
|
6031 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
6032 nr = nr * 101 + n;
|
|
236
|
6033 }
|
|
300
|
6034
|
|
|
6035 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
6036 n = nr & 0xff;
|
|
346
|
6037 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
6038 n = (nr >> 8) & 0xff;
|
|
346
|
6039 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
6040 n = (nr >> 16) & 0xff;
|
|
346
|
6041 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
6042 n = (nr >> 24) & 0xff;
|
|
346
|
6043 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
6044 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
6045
|
|
482
|
6046 /* Check for CTRL-C pressed now and then. */
|
|
|
6047 fast_breakcheck();
|
|
|
6048
|
|
300
|
6049 return compressed;
|
|
|
6050 }
|
|
|
6051
|
|
|
6052 /*
|
|
|
6053 * Return TRUE when two nodes have identical siblings and children.
|
|
|
6054 */
|
|
|
6055 static int
|
|
|
6056 node_equal(n1, n2)
|
|
|
6057 wordnode_T *n1;
|
|
|
6058 wordnode_T *n2;
|
|
|
6059 {
|
|
|
6060 wordnode_T *p1;
|
|
|
6061 wordnode_T *p2;
|
|
|
6062
|
|
|
6063 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
6064 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
6065 if (p1->wn_byte != p2->wn_byte
|
|
|
6066 || (p1->wn_byte == NUL
|
|
|
6067 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
6068 || p1->wn_region != p2->wn_region
|
|
481
|
6069 || p1->wn_affixID != p2->wn_affixID)
|
|
300
|
6070 : (p1->wn_child != p2->wn_child)))
|
|
|
6071 break;
|
|
|
6072
|
|
|
6073 return p1 == NULL && p2 == NULL;
|
|
236
|
6074 }
|
|
|
6075
|
|
|
6076 /*
|
|
|
6077 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
6078 */
|
|
255
|
6079 void
|
|
236
|
6080 put_bytes(fd, nr, len)
|
|
|
6081 FILE *fd;
|
|
|
6082 long_u nr;
|
|
|
6083 int len;
|
|
|
6084 {
|
|
|
6085 int i;
|
|
|
6086
|
|
|
6087 for (i = len - 1; i >= 0; --i)
|
|
|
6088 putc((int)(nr >> (i * 8)), fd);
|
|
|
6089 }
|
|
|
6090
|
|
323
|
6091 static int
|
|
|
6092 #ifdef __BORLANDC__
|
|
|
6093 _RTLENTRYF
|
|
|
6094 #endif
|
|
|
6095 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
6096
|
|
|
6097 /*
|
|
|
6098 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
6099 */
|
|
|
6100 static int
|
|
|
6101 #ifdef __BORLANDC__
|
|
|
6102 _RTLENTRYF
|
|
|
6103 #endif
|
|
|
6104 rep_compare(s1, s2)
|
|
|
6105 const void *s1;
|
|
|
6106 const void *s2;
|
|
|
6107 {
|
|
|
6108 fromto_T *p1 = (fromto_T *)s1;
|
|
|
6109 fromto_T *p2 = (fromto_T *)s2;
|
|
|
6110
|
|
|
6111 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
6112 }
|
|
|
6113
|
|
236
|
6114 /*
|
|
483
|
6115 * Write the Vim .spl file "fname".
|
|
236
|
6116 */
|
|
|
6117 static void
|
|
474
|
6118 write_vim_spell(spin, fname)
|
|
|
6119 spellinfo_T *spin;
|
|
236
|
6120 char_u *fname;
|
|
|
6121 {
|
|
300
|
6122 FILE *fd;
|
|
|
6123 int regionmask;
|
|
236
|
6124 int round;
|
|
300
|
6125 wordnode_T *tree;
|
|
|
6126 int nodecount;
|
|
323
|
6127 int i;
|
|
|
6128 int l;
|
|
|
6129 garray_T *gap;
|
|
|
6130 fromto_T *ftp;
|
|
|
6131 char_u *p;
|
|
|
6132 int rr;
|
|
236
|
6133
|
|
310
|
6134 fd = mch_fopen((char *)fname, "w");
|
|
300
|
6135 if (fd == NULL)
|
|
236
|
6136 {
|
|
|
6137 EMSG2(_(e_notopen), fname);
|
|
|
6138 return;
|
|
|
6139 }
|
|
|
6140
|
|
483
|
6141 /* <HEADER>: <fileID> <versionnr> */
|
|
300
|
6142 /* <fileID> */
|
|
|
6143 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
|
6144 EMSG(_(e_write));
|
|
483
|
6145 putc(VIMSPELLVERSION, fd); /* <versionnr> */
|
|
|
6146
|
|
|
6147 /*
|
|
|
6148 * <SECTIONS>: <section> ... <sectionend>
|
|
|
6149 */
|
|
|
6150
|
|
|
6151 /* SN_REGION: <regionname> ...
|
|
|
6152 * Write the region names only if there is more than one. */
|
|
316
|
6153 if (spin->si_region_count > 1)
|
|
236
|
6154 {
|
|
483
|
6155 putc(SN_REGION, fd); /* <sectionID> */
|
|
|
6156 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6157 l = spin->si_region_count * 2;
|
|
|
6158 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6159 fwrite(spin->si_region_name, (size_t)l, (size_t)1, fd);
|
|
|
6160 /* <regionname> ... */
|
|
316
|
6161 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
6162 }
|
|
|
6163 else
|
|
300
|
6164 regionmask = 0;
|
|
483
|
6165
|
|
|
6166 /* SN_CHARFLAGS: <charflagslen> <charflags> <folcharslen> <folchars>
|
|
|
6167 *
|
|
|
6168 * The table with character flags and the table for case folding.
|
|
|
6169 * This makes sure the same characters are recognized as word characters
|
|
|
6170 * when generating an when using a spell file.
|
|
260
|
6171 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
6172 * 'encoding'.
|
|
|
6173 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
6174 * the table (avoids that it conflicts). File is shorter too.
|
|
|
6175 */
|
|
483
|
6176 if (!spin->si_ascii && !spin->si_add)
|
|
|
6177 {
|
|
|
6178 char_u folchars[128 * 8];
|
|
|
6179 int flags;
|
|
|
6180
|
|
|
6181 putc(SN_MIDWORD, fd); /* <sectionID> */
|
|
|
6182 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6183
|
|
|
6184 /* Form the <folchars> string first, we need to know its length. */
|
|
|
6185 l = 0;
|
|
|
6186 for (i = 128; i < 256; ++i)
|
|
|
6187 {
|
|
|
6188 #ifdef FEAT_MBYTE
|
|
|
6189 if (has_mbyte)
|
|
|
6190 l += mb_char2bytes(spelltab.st_fold[i], folchars + l);
|
|
|
6191 else
|
|
|
6192 #endif
|
|
|
6193 folchars[l++] = spelltab.st_fold[i];
|
|
|
6194 }
|
|
|
6195 put_bytes(fd, (long_u)(1 + 128 + 2 + l), 4); /* <sectionlen> */
|
|
|
6196
|
|
|
6197 fputc(128, fd); /* <charflagslen> */
|
|
|
6198 for (i = 128; i < 256; ++i)
|
|
|
6199 {
|
|
|
6200 flags = 0;
|
|
|
6201 if (spelltab.st_isw[i])
|
|
|
6202 flags |= CF_WORD;
|
|
|
6203 if (spelltab.st_isu[i])
|
|
|
6204 flags |= CF_UPPER;
|
|
|
6205 fputc(flags, fd); /* <charflags> */
|
|
|
6206 }
|
|
|
6207
|
|
|
6208 put_bytes(fd, (long_u)l, 2); /* <folcharslen> */
|
|
|
6209 fwrite(folchars, (size_t)l, (size_t)1, fd); /* <folchars> */
|
|
|
6210 }
|
|
|
6211
|
|
|
6212 /* SN_MIDWORD: <midword> */
|
|
|
6213 if (spin->si_midword != NULL)
|
|
|
6214 {
|
|
|
6215 putc(SN_MIDWORD, fd); /* <sectionID> */
|
|
|
6216 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6217
|
|
366
|
6218 i = STRLEN(spin->si_midword);
|
|
483
|
6219 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */
|
|
366
|
6220 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
6221 }
|
|
|
6222
|
|
483
|
6223 /* SN_PREFCOND: <prefcondcnt> <prefcond> ... */
|
|
|
6224 if (spin->si_prefcond.ga_len > 0)
|
|
|
6225 {
|
|
|
6226 putc(SN_PREFCOND, fd); /* <sectionID> */
|
|
|
6227 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6228
|
|
|
6229 l = write_spell_prefcond(NULL, &spin->si_prefcond);
|
|
|
6230 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6231
|
|
|
6232 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
6233 }
|
|
|
6234
|
|
|
6235 /* SN_REP: <repcount> <rep> ...
|
|
|
6236 * SN_SAL: <salflags> <salcount> <sal> ... */
|
|
366
|
6237
|
|
323
|
6238 /* Sort the REP items. */
|
|
|
6239 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
6240 sizeof(fromto_T), rep_compare);
|
|
|
6241
|
|
483
|
6242 /* round 1: SN_REP section
|
|
|
6243 * round 2: SN_SAL section (unless SN_SOFO is used) */
|
|
323
|
6244 for (round = 1; round <= 2; ++round)
|
|
|
6245 {
|
|
|
6246 if (round == 1)
|
|
483
|
6247 {
|
|
323
|
6248 gap = &spin->si_rep;
|
|
483
|
6249 putc(SN_REP, fd); /* <sectionID> */
|
|
|
6250 }
|
|
323
|
6251 else
|
|
|
6252 {
|
|
483
|
6253 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
6254 /* using SN_SOFO section instead of SN_SAL */
|
|
|
6255 break;
|
|
323
|
6256 gap = &spin->si_sal;
|
|
483
|
6257 putc(SN_SAL, fd); /* <sectionID> */
|
|
|
6258 }
|
|
|
6259
|
|
|
6260 /* This is for making suggestions, section is not required. */
|
|
|
6261 putc(0, fd); /* <sectionflags> */
|
|
|
6262
|
|
|
6263 /* Compute the length of what follows. */
|
|
|
6264 l = 2; /* count <repcount> or <salcount> */
|
|
|
6265 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6266 {
|
|
|
6267 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
6268 l += 1 + STRLEN(ftp->ft_from); /* count <*fromlen> and <*from> */
|
|
|
6269 l += 1 + STRLEN(ftp->ft_to); /* count <*tolen> and <*to> */
|
|
|
6270 }
|
|
|
6271 if (round == 2)
|
|
|
6272 ++l; /* count <salflags> */
|
|
|
6273 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6274
|
|
|
6275 if (round == 2)
|
|
|
6276 {
|
|
323
|
6277 i = 0;
|
|
|
6278 if (spin->si_followup)
|
|
|
6279 i |= SAL_F0LLOWUP;
|
|
|
6280 if (spin->si_collapse)
|
|
|
6281 i |= SAL_COLLAPSE;
|
|
|
6282 if (spin->si_rem_accents)
|
|
|
6283 i |= SAL_REM_ACCENTS;
|
|
|
6284 putc(i, fd); /* <salflags> */
|
|
|
6285 }
|
|
|
6286
|
|
|
6287 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
6288 for (i = 0; i < gap->ga_len; ++i)
|
|
|
6289 {
|
|
|
6290 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
6291 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
6292 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
6293 for (rr = 1; rr <= 2; ++rr)
|
|
|
6294 {
|
|
|
6295 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
6296 l = STRLEN(p);
|
|
|
6297 putc(l, fd);
|
|
|
6298 fwrite(p, l, (size_t)1, fd);
|
|
|
6299 }
|
|
|
6300 }
|
|
483
|
6301
|
|
|
6302 }
|
|
|
6303
|
|
|
6304 /* SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
6305 * This is for making suggestions, section is not required. */
|
|
375
|
6306 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
6307 {
|
|
483
|
6308 putc(SN_SOFO, fd); /* <sectionID> */
|
|
|
6309 putc(0, fd); /* <sectionflags> */
|
|
375
|
6310
|
|
|
6311 l = STRLEN(spin->si_sofofr);
|
|
483
|
6312 put_bytes(fd, (long_u)(l + STRLEN(spin->si_sofoto) + 4), 4);
|
|
|
6313 /* <sectionlen> */
|
|
|
6314
|
|
|
6315 put_bytes(fd, (long_u)l, 2); /* <sofofromlen> */
|
|
|
6316 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <sofofrom> */
|
|
375
|
6317
|
|
|
6318 l = STRLEN(spin->si_sofoto);
|
|
483
|
6319 put_bytes(fd, (long_u)l, 2); /* <sofotolen> */
|
|
|
6320 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <sofoto> */
|
|
|
6321 }
|
|
|
6322
|
|
|
6323 /* SN_MAP: <mapstr>
|
|
|
6324 * This is for making suggestions, section is not required. */
|
|
|
6325 if (spin->si_map.ga_len > 0)
|
|
|
6326 {
|
|
|
6327 putc(SN_MAP, fd); /* <sectionID> */
|
|
|
6328 putc(0, fd); /* <sectionflags> */
|
|
|
6329 l = spin->si_map.ga_len;
|
|
|
6330 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6331 fwrite(spin->si_map.ga_data, (size_t)l, (size_t)1, fd);
|
|
|
6332 /* <mapstr> */
|
|
|
6333 }
|
|
|
6334
|
|
|
6335 /* SN_COMPOUND: compound info.
|
|
|
6336 * We don't mark it required, when not supported all compound words will
|
|
|
6337 * be bad words. */
|
|
|
6338 if (spin->si_compflags != NULL)
|
|
|
6339 {
|
|
|
6340 putc(SN_COMPOUND, fd); /* <sectionID> */
|
|
|
6341 putc(0, fd); /* <sectionflags> */
|
|
|
6342
|
|
|
6343 l = STRLEN(spin->si_compflags);
|
|
|
6344 put_bytes(fd, (long_u)(l + 3), 4); /* <sectionlen> */
|
|
|
6345 putc(spin->si_compmax, fd); /* <compmax> */
|
|
|
6346 putc(spin->si_compminlen, fd); /* <compminlen> */
|
|
|
6347 putc(spin->si_compsylmax, fd); /* <compsylmax> */
|
|
|
6348 /* <compflags> */
|
|
|
6349 fwrite(spin->si_compflags, (size_t)l, (size_t)1, fd);
|
|
|
6350 }
|
|
|
6351
|
|
|
6352 /* SN_SYLLABLE: syllable info.
|
|
|
6353 * We don't mark it required, when not supported syllables will not be
|
|
|
6354 * counted. */
|
|
|
6355 if (spin->si_syllable != NULL)
|
|
|
6356 {
|
|
|
6357 putc(SN_SYLLABLE, fd); /* <sectionID> */
|
|
|
6358 putc(0, fd); /* <sectionflags> */
|
|
|
6359
|
|
|
6360 l = STRLEN(spin->si_syllable);
|
|
|
6361 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6362 fwrite(spin->si_syllable, (size_t)l, (size_t)1, fd); /* <syllable> */
|
|
|
6363 }
|
|
|
6364
|
|
|
6365 /* end of <SECTIONS> */
|
|
|
6366 putc(SN_END, fd); /* <sectionend> */
|
|
|
6367
|
|
302
|
6368
|
|
236
|
6369 /*
|
|
339
|
6370 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
6371 */
|
|
323
|
6372 spin->si_memtot = 0;
|
|
339
|
6373 for (round = 1; round <= 3; ++round)
|
|
236
|
6374 {
|
|
339
|
6375 if (round == 1)
|
|
469
|
6376 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
6377 else if (round == 2)
|
|
469
|
6378 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
6379 else
|
|
469
|
6380 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
6381
|
|
346
|
6382 /* Clear the index and wnode fields in the tree. */
|
|
|
6383 clear_node(tree);
|
|
|
6384
|
|
300
|
6385 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
6386 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
6387 * nodes. */
|
|
346
|
6388 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
6389
|
|
300
|
6390 /* number of nodes in 4 bytes */
|
|
|
6391 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
6392 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
6393
|
|
300
|
6394 /* Write the nodes. */
|
|
346
|
6395 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
6396 }
|
|
|
6397
|
|
300
|
6398 fclose(fd);
|
|
236
|
6399 }
|
|
|
6400
|
|
|
6401 /*
|
|
346
|
6402 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
6403 * children. This is needed because they are a union with other items to save
|
|
|
6404 * space.
|
|
|
6405 */
|
|
|
6406 static void
|
|
|
6407 clear_node(node)
|
|
|
6408 wordnode_T *node;
|
|
|
6409 {
|
|
|
6410 wordnode_T *np;
|
|
|
6411
|
|
|
6412 if (node != NULL)
|
|
|
6413 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6414 {
|
|
|
6415 np->wn_u1.index = 0;
|
|
|
6416 np->wn_u2.wnode = NULL;
|
|
|
6417
|
|
|
6418 if (np->wn_byte != NUL)
|
|
|
6419 clear_node(np->wn_child);
|
|
|
6420 }
|
|
|
6421 }
|
|
|
6422
|
|
|
6423
|
|
|
6424 /*
|
|
300
|
6425 * Dump a word tree at node "node".
|
|
|
6426 *
|
|
|
6427 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
6428 * byte recursively write the children.
|
|
|
6429 *
|
|
|
6430 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
6431 * are made about the indexes (so that we don't have to write them in the
|
|
|
6432 * file).
|
|
236
|
6433 *
|
|
300
|
6434 * Returns the number of nodes used.
|
|
236
|
6435 */
|
|
300
|
6436 static int
|
|
346
|
6437 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
6438 FILE *fd; /* NULL when only counting */
|
|
300
|
6439 wordnode_T *node;
|
|
|
6440 int index;
|
|
|
6441 int regionmask;
|
|
339
|
6442 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
6443 {
|
|
300
|
6444 int newindex = index;
|
|
|
6445 int siblingcount = 0;
|
|
|
6446 wordnode_T *np;
|
|
236
|
6447 int flags;
|
|
300
|
6448
|
|
|
6449 /* If "node" is zero the tree is empty. */
|
|
|
6450 if (node == NULL)
|
|
|
6451 return 0;
|
|
|
6452
|
|
|
6453 /* Store the index where this node is written. */
|
|
346
|
6454 node->wn_u1.index = index;
|
|
236
|
6455
|
|
300
|
6456 /* Count the number of siblings. */
|
|
|
6457 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6458 ++siblingcount;
|
|
236
|
6459
|
|
300
|
6460 /* Write the sibling count. */
|
|
|
6461 if (fd != NULL)
|
|
|
6462 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
6463
|
|
300
|
6464 /* Write each sibling byte and optionally extra info. */
|
|
|
6465 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
6466 {
|
|
300
|
6467 if (np->wn_byte == 0)
|
|
|
6468 {
|
|
|
6469 if (fd != NULL)
|
|
|
6470 {
|
|
339
|
6471 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
6472 if (prefixtree)
|
|
300
|
6473 {
|
|
481
|
6474 /* In PREFIXTREE write the required affixID and the
|
|
390
|
6475 * associated condition nr (stored in wn_region). The
|
|
|
6476 * byte value is misused to store the "rare" and "not
|
|
|
6477 * combining" flags */
|
|
455
|
6478 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
6479 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
6480 else
|
|
|
6481 {
|
|
390
|
6482 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
6483 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
6484 }
|
|
481
|
6485 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
6486 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
6487 }
|
|
|
6488 else
|
|
|
6489 {
|
|
339
|
6490 /* For word trees we write the flag/region items. */
|
|
|
6491 flags = np->wn_flags;
|
|
|
6492 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
6493 flags |= WF_REGION;
|
|
481
|
6494 if (np->wn_affixID != 0)
|
|
|
6495 flags |= WF_AFX;
|
|
339
|
6496 if (flags == 0)
|
|
|
6497 {
|
|
|
6498 /* word without flags or region */
|
|
|
6499 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
6500 }
|
|
|
6501 else
|
|
|
6502 {
|
|
390
|
6503 if (np->wn_flags >= 0x100)
|
|
|
6504 {
|
|
|
6505 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
6506 putc(flags, fd); /* <flags> */
|
|
|
6507 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
6508 }
|
|
|
6509 else
|
|
|
6510 {
|
|
|
6511 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
6512 putc(flags, fd); /* <flags> */
|
|
|
6513 }
|
|
339
|
6514 if (flags & WF_REGION)
|
|
|
6515 putc(np->wn_region, fd); /* <region> */
|
|
481
|
6516 if (flags & WF_AFX)
|
|
|
6517 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
6518 }
|
|
300
|
6519 }
|
|
|
6520 }
|
|
|
6521 }
|
|
|
6522 else
|
|
|
6523 {
|
|
346
|
6524 if (np->wn_child->wn_u1.index != 0
|
|
|
6525 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
6526 {
|
|
|
6527 /* The child is written elsewhere, write the reference. */
|
|
|
6528 if (fd != NULL)
|
|
|
6529 {
|
|
|
6530 putc(BY_INDEX, fd); /* <byte> */
|
|
|
6531 /* <nodeidx> */
|
|
346
|
6532 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
6533 }
|
|
|
6534 }
|
|
346
|
6535 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
6536 /* We will write the child below and give it an index. */
|
|
346
|
6537 np->wn_child->wn_u2.wnode = node;
|
|
236
|
6538
|
|
300
|
6539 if (fd != NULL)
|
|
|
6540 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
6541 {
|
|
|
6542 EMSG(_(e_write));
|
|
|
6543 return 0;
|
|
|
6544 }
|
|
|
6545 }
|
|
236
|
6546 }
|
|
|
6547
|
|
300
|
6548 /* Space used in the array when reading: one for each sibling and one for
|
|
|
6549 * the count. */
|
|
|
6550 newindex += siblingcount + 1;
|
|
249
|
6551
|
|
300
|
6552 /* Recursively dump the children of each sibling. */
|
|
|
6553 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
6554 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
6555 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
6556 prefixtree);
|
|
249
|
6557
|
|
300
|
6558 return newindex;
|
|
236
|
6559 }
|
|
|
6560
|
|
|
6561
|
|
|
6562 /*
|
|
310
|
6563 * ":mkspell [-ascii] outfile infile ..."
|
|
|
6564 * ":mkspell [-ascii] addfile"
|
|
236
|
6565 */
|
|
|
6566 void
|
|
|
6567 ex_mkspell(eap)
|
|
|
6568 exarg_T *eap;
|
|
|
6569 {
|
|
|
6570 int fcount;
|
|
|
6571 char_u **fnames;
|
|
310
|
6572 char_u *arg = eap->arg;
|
|
|
6573 int ascii = FALSE;
|
|
|
6574
|
|
|
6575 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
6576 {
|
|
|
6577 ascii = TRUE;
|
|
|
6578 arg = skipwhite(arg + 6);
|
|
|
6579 }
|
|
|
6580
|
|
|
6581 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
6582 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
6583 {
|
|
323
|
6584 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
6585 FreeWild(fcount, fnames);
|
|
|
6586 }
|
|
|
6587 }
|
|
|
6588
|
|
|
6589 /*
|
|
|
6590 * Create a Vim spell file from one or more word lists.
|
|
|
6591 * "fnames[0]" is the output file name.
|
|
|
6592 * "fnames[fcount - 1]" is the last input file name.
|
|
|
6593 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
6594 * and ".spl" is appended to make the output file name.
|
|
|
6595 */
|
|
|
6596 static void
|
|
323
|
6597 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
6598 int fcount;
|
|
|
6599 char_u **fnames;
|
|
|
6600 int ascii; /* -ascii argument given */
|
|
|
6601 int overwrite; /* overwrite existing output file */
|
|
323
|
6602 int added_word; /* invoked through "zg" */
|
|
310
|
6603 {
|
|
236
|
6604 char_u fname[MAXPATHL];
|
|
|
6605 char_u wfname[MAXPATHL];
|
|
310
|
6606 char_u **innames;
|
|
|
6607 int incount;
|
|
236
|
6608 afffile_T *(afile[8]);
|
|
|
6609 int i;
|
|
|
6610 int len;
|
|
|
6611 struct stat st;
|
|
255
|
6612 int error = FALSE;
|
|
300
|
6613 spellinfo_T spin;
|
|
|
6614
|
|
|
6615 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
6616 spin.si_verbose = !added_word;
|
|
310
|
6617 spin.si_ascii = ascii;
|
|
323
|
6618 spin.si_followup = TRUE;
|
|
|
6619 spin.si_rem_accents = TRUE;
|
|
|
6620 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
6621 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
6622 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
6623 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
310
|
6624
|
|
|
6625 /* default: fnames[0] is output file, following are input files */
|
|
|
6626 innames = &fnames[1];
|
|
|
6627 incount = fcount - 1;
|
|
|
6628
|
|
|
6629 if (fcount >= 1)
|
|
240
|
6630 {
|
|
310
|
6631 len = STRLEN(fnames[0]);
|
|
|
6632 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
6633 {
|
|
|
6634 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
6635 * "path/en.latin1.add.spl". */
|
|
|
6636 innames = &fnames[0];
|
|
|
6637 incount = 1;
|
|
|
6638 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
6639 }
|
|
366
|
6640 else if (fcount == 1)
|
|
|
6641 {
|
|
|
6642 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
6643 innames = &fnames[0];
|
|
|
6644 incount = 1;
|
|
|
6645 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
6646 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
6647 }
|
|
310
|
6648 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
6649 {
|
|
|
6650 /* Name ends in ".spl", use as the file name. */
|
|
323
|
6651 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
6652 }
|
|
|
6653 else
|
|
|
6654 /* Name should be language, make the file name from it. */
|
|
|
6655 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
6656 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
6657
|
|
|
6658 /* Check for .ascii.spl. */
|
|
|
6659 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
6660 spin.si_ascii = TRUE;
|
|
|
6661
|
|
|
6662 /* Check for .add.spl. */
|
|
|
6663 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
6664 spin.si_add = TRUE;
|
|
240
|
6665 }
|
|
|
6666
|
|
310
|
6667 if (incount <= 0)
|
|
236
|
6668 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
6669 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
6670 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
6671 else if (incount > 8)
|
|
236
|
6672 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
6673 else
|
|
|
6674 {
|
|
|
6675 /* Check for overwriting before doing things that may take a lot of
|
|
|
6676 * time. */
|
|
310
|
6677 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
6678 {
|
|
|
6679 EMSG(_(e_exists));
|
|
310
|
6680 return;
|
|
236
|
6681 }
|
|
310
|
6682 if (mch_isdir(wfname))
|
|
236
|
6683 {
|
|
310
|
6684 EMSG2(_(e_isadir2), wfname);
|
|
|
6685 return;
|
|
236
|
6686 }
|
|
|
6687
|
|
|
6688 /*
|
|
|
6689 * Init the aff and dic pointers.
|
|
|
6690 * Get the region names if there are more than 2 arguments.
|
|
|
6691 */
|
|
310
|
6692 for (i = 0; i < incount; ++i)
|
|
236
|
6693 {
|
|
310
|
6694 afile[i] = NULL;
|
|
300
|
6695
|
|
316
|
6696 if (incount > 1)
|
|
236
|
6697 {
|
|
310
|
6698 len = STRLEN(innames[i]);
|
|
|
6699 if (STRLEN(gettail(innames[i])) < 5
|
|
|
6700 || innames[i][len - 3] != '_')
|
|
236
|
6701 {
|
|
310
|
6702 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
6703 return;
|
|
236
|
6704 }
|
|
316
|
6705 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
6706 spin.si_region_name[i * 2 + 1] =
|
|
|
6707 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
6708 }
|
|
|
6709 }
|
|
316
|
6710 spin.si_region_count = incount;
|
|
236
|
6711
|
|
474
|
6712 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
6713 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
6714 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
6715 if (spin.si_foldroot == NULL
|
|
|
6716 || spin.si_keeproot == NULL
|
|
|
6717 || spin.si_prefroot == NULL)
|
|
300
|
6718 {
|
|
469
|
6719 free_blocks(spin.si_blocks);
|
|
310
|
6720 return;
|
|
300
|
6721 }
|
|
|
6722
|
|
351
|
6723 /* When not producing a .add.spl file clear the character table when
|
|
|
6724 * we encounter one in the .aff file. This means we dump the current
|
|
|
6725 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
6726 * better than guessing the contents, the table will match a
|
|
|
6727 * previously loaded spell file. */
|
|
|
6728 if (!spin.si_add)
|
|
|
6729 spin.si_clear_chartab = TRUE;
|
|
|
6730
|
|
236
|
6731 /*
|
|
|
6732 * Read all the .aff and .dic files.
|
|
|
6733 * Text is converted to 'encoding'.
|
|
300
|
6734 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
6735 */
|
|
310
|
6736 for (i = 0; i < incount && !error; ++i)
|
|
236
|
6737 {
|
|
300
|
6738 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
6739 spin.si_region = 1 << i;
|
|
|
6740
|
|
|
6741 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
6742 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
6743 {
|
|
|
6744 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
6745 * "SET" line. */
|
|
474
|
6746 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
6747 if (afile[i] == NULL)
|
|
300
|
6748 error = TRUE;
|
|
|
6749 else
|
|
|
6750 {
|
|
|
6751 /* Read the .dic file and store the words in the trees. */
|
|
|
6752 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
6753 innames[i]);
|
|
474
|
6754 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
6755 error = TRUE;
|
|
|
6756 }
|
|
|
6757 }
|
|
|
6758 else
|
|
|
6759 {
|
|
|
6760 /* No .aff file, try reading the file as a word list. Store
|
|
|
6761 * the words in the trees. */
|
|
474
|
6762 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
6763 error = TRUE;
|
|
|
6764 }
|
|
236
|
6765
|
|
310
|
6766 #ifdef FEAT_MBYTE
|
|
236
|
6767 /* Free any conversion stuff. */
|
|
300
|
6768 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
6769 #endif
|
|
236
|
6770 }
|
|
|
6771
|
|
300
|
6772 if (!error)
|
|
236
|
6773 {
|
|
|
6774 /*
|
|
300
|
6775 * Combine tails in the tree.
|
|
236
|
6776 */
|
|
474
|
6777 if (spin.si_verbose || p_verbose > 2)
|
|
|
6778 {
|
|
|
6779 if (!spin.si_verbose)
|
|
310
|
6780 verbose_enter();
|
|
469
|
6781 MSG(_(msg_compressing));
|
|
310
|
6782 out_flush();
|
|
474
|
6783 if (!spin.si_verbose)
|
|
310
|
6784 verbose_leave();
|
|
|
6785 }
|
|
474
|
6786 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
6787 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
6788 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
6789 }
|
|
|
6790
|
|
300
|
6791 if (!error)
|
|
|
6792 {
|
|
|
6793 /*
|
|
|
6794 * Write the info in the spell file.
|
|
|
6795 */
|
|
474
|
6796 if (spin.si_verbose || p_verbose > 2)
|
|
|
6797 {
|
|
|
6798 if (!spin.si_verbose)
|
|
310
|
6799 verbose_enter();
|
|
366
|
6800 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
6801 out_flush();
|
|
474
|
6802 if (!spin.si_verbose)
|
|
310
|
6803 verbose_leave();
|
|
|
6804 }
|
|
|
6805
|
|
474
|
6806 write_vim_spell(&spin, wfname);
|
|
|
6807
|
|
|
6808 if (spin.si_verbose || p_verbose > 2)
|
|
|
6809 {
|
|
|
6810 if (!spin.si_verbose)
|
|
310
|
6811 verbose_enter();
|
|
|
6812 MSG(_("Done!"));
|
|
|
6813 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
6814 spin.si_memtot);
|
|
310
|
6815 out_flush();
|
|
474
|
6816 if (!spin.si_verbose)
|
|
310
|
6817 verbose_leave();
|
|
|
6818 }
|
|
|
6819
|
|
|
6820 /* If the file is loaded need to reload it. */
|
|
323
|
6821 spell_reload_one(wfname, added_word);
|
|
300
|
6822 }
|
|
|
6823
|
|
|
6824 /* Free the allocated memory. */
|
|
323
|
6825 ga_clear(&spin.si_rep);
|
|
|
6826 ga_clear(&spin.si_sal);
|
|
|
6827 ga_clear(&spin.si_map);
|
|
339
|
6828 ga_clear(&spin.si_prefcond);
|
|
366
|
6829 vim_free(spin.si_midword);
|
|
375
|
6830 vim_free(spin.si_sofofr);
|
|
|
6831 vim_free(spin.si_sofoto);
|
|
300
|
6832
|
|
|
6833 /* Free the .aff file structures. */
|
|
310
|
6834 for (i = 0; i < incount; ++i)
|
|
|
6835 if (afile[i] != NULL)
|
|
|
6836 spell_free_aff(afile[i]);
|
|
339
|
6837
|
|
|
6838 /* Free all the bits and pieces at once. */
|
|
|
6839 free_blocks(spin.si_blocks);
|
|
236
|
6840 }
|
|
310
|
6841 }
|
|
|
6842
|
|
|
6843
|
|
|
6844 /*
|
|
385
|
6845 * ":[count]spellgood {word}"
|
|
|
6846 * ":[count]spellwrong {word}"
|
|
310
|
6847 */
|
|
|
6848 void
|
|
|
6849 ex_spell(eap)
|
|
|
6850 exarg_T *eap;
|
|
|
6851 {
|
|
381
|
6852 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
6853 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
6854 }
|
|
|
6855
|
|
310
|
6856 /*
|
|
|
6857 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
6858 */
|
|
|
6859 void
|
|
385
|
6860 spell_add_word(word, len, bad, index)
|
|
310
|
6861 char_u *word;
|
|
|
6862 int len;
|
|
|
6863 int bad;
|
|
385
|
6864 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
6865 'spellfile' */
|
|
310
|
6866 {
|
|
|
6867 FILE *fd;
|
|
385
|
6868 buf_T *buf = NULL;
|
|
351
|
6869 int new_spf = FALSE;
|
|
|
6870 struct stat st;
|
|
381
|
6871 char_u *fname;
|
|
385
|
6872 char_u fnamebuf[MAXPATHL];
|
|
|
6873 char_u line[MAXWLEN * 2];
|
|
|
6874 long fpos, fpos_next = 0;
|
|
|
6875 int i;
|
|
|
6876 char_u *spf;
|
|
|
6877
|
|
|
6878 if (index == 0) /* use internal wordlist */
|
|
|
6879 {
|
|
|
6880 if (int_wordlist == NULL)
|
|
|
6881 {
|
|
|
6882 int_wordlist = vim_tempname('s');
|
|
|
6883 if (int_wordlist == NULL)
|
|
381
|
6884 return;
|
|
|
6885 }
|
|
385
|
6886 fname = int_wordlist;
|
|
381
|
6887 }
|
|
310
|
6888 else
|
|
|
6889 {
|
|
381
|
6890 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
6891 if (*curbuf->b_p_spf == NUL)
|
|
|
6892 {
|
|
|
6893 init_spellfile();
|
|
|
6894 new_spf = TRUE;
|
|
|
6895 }
|
|
|
6896
|
|
|
6897 if (*curbuf->b_p_spf == NUL)
|
|
|
6898 {
|
|
|
6899 EMSG(_("E764: 'spellfile' is not set"));
|
|
|
6900 return;
|
|
|
6901 }
|
|
|
6902
|
|
385
|
6903 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
6904 {
|
|
|
6905 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
6906 if (i == index)
|
|
|
6907 break;
|
|
|
6908 if (*spf == NUL)
|
|
|
6909 {
|
|
|
6910 EMSGN(_("E765: 'spellfile' does not have %ld enties"), index);
|
|
|
6911 return;
|
|
|
6912 }
|
|
|
6913 }
|
|
|
6914
|
|
310
|
6915 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
6916 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
6917 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
6918 buf = NULL;
|
|
|
6919 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
6920 {
|
|
310
|
6921 EMSG(_(e_bufloaded));
|
|
381
|
6922 return;
|
|
|
6923 }
|
|
|
6924
|
|
385
|
6925 fname = fnamebuf;
|
|
|
6926 }
|
|
|
6927
|
|
|
6928 if (bad)
|
|
|
6929 {
|
|
|
6930 /* When the word also appears as good word we need to remove that one,
|
|
|
6931 * since its flags sort before the one with WF_BANNED. */
|
|
|
6932 fd = mch_fopen((char *)fname, "r");
|
|
|
6933 if (fd != NULL)
|
|
|
6934 {
|
|
|
6935 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
6936 {
|
|
|
6937 fpos = fpos_next;
|
|
|
6938 fpos_next = ftell(fd);
|
|
|
6939 if (STRNCMP(word, line, len) == 0
|
|
|
6940 && (line[len] == '/' || line[len] < ' '))
|
|
|
6941 {
|
|
|
6942 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
6943 * the start of the line. Mixing reading and writing
|
|
|
6944 * doesn't work for all systems, close the file first. */
|
|
|
6945 fclose(fd);
|
|
|
6946 fd = mch_fopen((char *)fname, "r+");
|
|
|
6947 if (fd == NULL)
|
|
|
6948 break;
|
|
|
6949 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
6950 fputc('#', fd);
|
|
|
6951 fseek(fd, fpos_next, SEEK_SET);
|
|
|
6952 }
|
|
|
6953 }
|
|
|
6954 fclose(fd);
|
|
|
6955 }
|
|
381
|
6956 }
|
|
|
6957
|
|
|
6958 fd = mch_fopen((char *)fname, "a");
|
|
|
6959 if (fd == NULL && new_spf)
|
|
|
6960 {
|
|
|
6961 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
6962 * We may need to create the "spell" directory first. We already
|
|
|
6963 * checked the runtime directory is writable in init_spellfile(). */
|
|
|
6964 STRCPY(NameBuff, fname);
|
|
|
6965 *gettail_sep(NameBuff) = NUL;
|
|
|
6966 if (mch_stat((char *)NameBuff, &st) < 0)
|
|
|
6967 {
|
|
|
6968 /* The directory doesn't exist. Try creating it and opening the
|
|
|
6969 * file again. */
|
|
|
6970 vim_mkdir(NameBuff, 0755);
|
|
|
6971 fd = mch_fopen((char *)fname, "a");
|
|
|
6972 }
|
|
|
6973 }
|
|
|
6974
|
|
|
6975 if (fd == NULL)
|
|
|
6976 EMSG2(_(e_notopen), fname);
|
|
|
6977 else
|
|
|
6978 {
|
|
|
6979 if (bad)
|
|
|
6980 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
6981 else
|
|
381
|
6982 fprintf(fd, "%.*s\n", len, word);
|
|
|
6983 fclose(fd);
|
|
|
6984
|
|
|
6985 /* Update the .add.spl file. */
|
|
|
6986 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
6987
|
|
|
6988 /* If the .add file is edited somewhere, reload it. */
|
|
|
6989 if (buf != NULL)
|
|
|
6990 buf_reload(buf);
|
|
|
6991
|
|
|
6992 redraw_all_later(NOT_VALID);
|
|
310
|
6993 }
|
|
|
6994 }
|
|
|
6995
|
|
|
6996 /*
|
|
|
6997 * Initialize 'spellfile' for the current buffer.
|
|
|
6998 */
|
|
|
6999 static void
|
|
|
7000 init_spellfile()
|
|
|
7001 {
|
|
|
7002 char_u buf[MAXPATHL];
|
|
|
7003 int l;
|
|
|
7004 slang_T *sl;
|
|
|
7005 char_u *rtp;
|
|
351
|
7006 char_u *lend;
|
|
310
|
7007
|
|
|
7008 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
7009 {
|
|
351
|
7010 /* Find the end of the language name. Exclude the region. */
|
|
|
7011 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
7012 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
|
7013 ;
|
|
|
7014
|
|
|
7015 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
7016 * are allowed to write. */
|
|
310
|
7017 rtp = p_rtp;
|
|
|
7018 while (*rtp != NUL)
|
|
|
7019 {
|
|
|
7020 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
7021 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
|
7022 if (filewritable(buf) == 2)
|
|
|
7023 {
|
|
316
|
7024 /* Use the first language name from 'spelllang' and the
|
|
|
7025 * encoding used in the first loaded .spl file. */
|
|
310
|
7026 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
|
7027 l = STRLEN(buf);
|
|
|
7028 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
316
|
7029 "/spell/%.*s.%s.add",
|
|
351
|
7030 (int)(lend - curbuf->b_p_spl), curbuf->b_p_spl,
|
|
310
|
7031 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
|
7032 ? (char_u *)"ascii" : spell_enc());
|
|
|
7033 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
7034 break;
|
|
|
7035 }
|
|
|
7036 }
|
|
|
7037 }
|
|
|
7038 }
|
|
236
|
7039
|
|
300
|
7040
|
|
307
|
7041 /*
|
|
|
7042 * Init the chartab used for spelling for ASCII.
|
|
|
7043 * EBCDIC is not supported!
|
|
|
7044 */
|
|
|
7045 static void
|
|
|
7046 clear_spell_chartab(sp)
|
|
|
7047 spelltab_T *sp;
|
|
|
7048 {
|
|
324
|
7049 int i;
|
|
307
|
7050
|
|
|
7051 /* Init everything to FALSE. */
|
|
|
7052 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
7053 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
7054 for (i = 0; i < 256; ++i)
|
|
324
|
7055 {
|
|
307
|
7056 sp->st_fold[i] = i;
|
|
324
|
7057 sp->st_upper[i] = i;
|
|
|
7058 }
|
|
307
|
7059
|
|
|
7060 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
7061 * that is done separately. */
|
|
|
7062 for (i = '0'; i <= '9'; ++i)
|
|
|
7063 sp->st_isw[i] = TRUE;
|
|
|
7064 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
7065 {
|
|
|
7066 sp->st_isw[i] = TRUE;
|
|
|
7067 sp->st_isu[i] = TRUE;
|
|
|
7068 sp->st_fold[i] = i + 0x20;
|
|
|
7069 }
|
|
|
7070 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
7071 {
|
|
307
|
7072 sp->st_isw[i] = TRUE;
|
|
324
|
7073 sp->st_upper[i] = i - 0x20;
|
|
|
7074 }
|
|
307
|
7075 }
|
|
|
7076
|
|
|
7077 /*
|
|
|
7078 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
7079 * Called once while starting up and when 'encoding' changes.
|
|
|
7080 * The default is to use isalpha(), but the spell file should define the word
|
|
|
7081 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
7082 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
7083 */
|
|
|
7084 void
|
|
|
7085 init_spell_chartab()
|
|
|
7086 {
|
|
|
7087 int i;
|
|
|
7088
|
|
|
7089 did_set_spelltab = FALSE;
|
|
|
7090 clear_spell_chartab(&spelltab);
|
|
|
7091 #ifdef FEAT_MBYTE
|
|
|
7092 if (enc_dbcs)
|
|
|
7093 {
|
|
|
7094 /* DBCS: assume double-wide characters are word characters. */
|
|
|
7095 for (i = 128; i <= 255; ++i)
|
|
|
7096 if (MB_BYTE2LEN(i) == 2)
|
|
|
7097 spelltab.st_isw[i] = TRUE;
|
|
|
7098 }
|
|
324
|
7099 else if (enc_utf8)
|
|
|
7100 {
|
|
|
7101 for (i = 128; i < 256; ++i)
|
|
|
7102 {
|
|
|
7103 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
7104 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
7105 spelltab.st_fold[i] = utf_fold(i);
|
|
|
7106 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
7107 }
|
|
|
7108 }
|
|
307
|
7109 else
|
|
|
7110 #endif
|
|
|
7111 {
|
|
324
|
7112 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
7113 for (i = 128; i < 256; ++i)
|
|
|
7114 {
|
|
|
7115 if (MB_ISUPPER(i))
|
|
|
7116 {
|
|
324
|
7117 spelltab.st_isw[i] = TRUE;
|
|
307
|
7118 spelltab.st_isu[i] = TRUE;
|
|
|
7119 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
7120 }
|
|
324
|
7121 else if (MB_ISLOWER(i))
|
|
|
7122 {
|
|
|
7123 spelltab.st_isw[i] = TRUE;
|
|
|
7124 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
7125 }
|
|
307
|
7126 }
|
|
|
7127 }
|
|
|
7128 }
|
|
|
7129
|
|
|
7130 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
7131 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
|
7132
|
|
|
7133 /*
|
|
|
7134 * Set the spell character tables from strings in the affix file.
|
|
|
7135 */
|
|
|
7136 static int
|
|
|
7137 set_spell_chartab(fol, low, upp)
|
|
|
7138 char_u *fol;
|
|
|
7139 char_u *low;
|
|
|
7140 char_u *upp;
|
|
|
7141 {
|
|
|
7142 /* We build the new tables here first, so that we can compare with the
|
|
|
7143 * previous one. */
|
|
|
7144 spelltab_T new_st;
|
|
|
7145 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
7146 int f, l, u;
|
|
|
7147
|
|
|
7148 clear_spell_chartab(&new_st);
|
|
|
7149
|
|
|
7150 while (*pf != NUL)
|
|
|
7151 {
|
|
|
7152 if (*pl == NUL || *pu == NUL)
|
|
|
7153 {
|
|
|
7154 EMSG(_(e_affform));
|
|
|
7155 return FAIL;
|
|
|
7156 }
|
|
|
7157 #ifdef FEAT_MBYTE
|
|
|
7158 f = mb_ptr2char_adv(&pf);
|
|
|
7159 l = mb_ptr2char_adv(&pl);
|
|
|
7160 u = mb_ptr2char_adv(&pu);
|
|
|
7161 #else
|
|
|
7162 f = *pf++;
|
|
|
7163 l = *pl++;
|
|
|
7164 u = *pu++;
|
|
|
7165 #endif
|
|
|
7166 /* Every character that appears is a word character. */
|
|
|
7167 if (f < 256)
|
|
|
7168 new_st.st_isw[f] = TRUE;
|
|
|
7169 if (l < 256)
|
|
|
7170 new_st.st_isw[l] = TRUE;
|
|
|
7171 if (u < 256)
|
|
|
7172 new_st.st_isw[u] = TRUE;
|
|
|
7173
|
|
|
7174 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
7175 * case-folding */
|
|
|
7176 if (l < 256 && l != f)
|
|
|
7177 {
|
|
|
7178 if (f >= 256)
|
|
|
7179 {
|
|
|
7180 EMSG(_(e_affrange));
|
|
|
7181 return FAIL;
|
|
|
7182 }
|
|
|
7183 new_st.st_fold[l] = f;
|
|
|
7184 }
|
|
|
7185
|
|
|
7186 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
7187 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
7188 * "FOL" . */
|
|
307
|
7189 if (u < 256 && u != f)
|
|
|
7190 {
|
|
|
7191 if (f >= 256)
|
|
|
7192 {
|
|
|
7193 EMSG(_(e_affrange));
|
|
|
7194 return FAIL;
|
|
|
7195 }
|
|
|
7196 new_st.st_fold[u] = f;
|
|
|
7197 new_st.st_isu[u] = TRUE;
|
|
324
|
7198 new_st.st_upper[f] = u;
|
|
307
|
7199 }
|
|
|
7200 }
|
|
|
7201
|
|
|
7202 if (*pl != NUL || *pu != NUL)
|
|
|
7203 {
|
|
|
7204 EMSG(_(e_affform));
|
|
|
7205 return FAIL;
|
|
|
7206 }
|
|
|
7207
|
|
|
7208 return set_spell_finish(&new_st);
|
|
|
7209 }
|
|
|
7210
|
|
|
7211 /*
|
|
|
7212 * Set the spell character tables from strings in the .spl file.
|
|
|
7213 */
|
|
483
|
7214 static void
|
|
389
|
7215 set_spell_charflags(flags, cnt, fol)
|
|
307
|
7216 char_u *flags;
|
|
389
|
7217 int cnt; /* length of "flags" */
|
|
|
7218 char_u *fol;
|
|
307
|
7219 {
|
|
|
7220 /* We build the new tables here first, so that we can compare with the
|
|
|
7221 * previous one. */
|
|
|
7222 spelltab_T new_st;
|
|
|
7223 int i;
|
|
389
|
7224 char_u *p = fol;
|
|
324
|
7225 int c;
|
|
307
|
7226
|
|
|
7227 clear_spell_chartab(&new_st);
|
|
|
7228
|
|
389
|
7229 for (i = 0; i < 128; ++i)
|
|
|
7230 {
|
|
|
7231 if (i < cnt)
|
|
|
7232 {
|
|
|
7233 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
7234 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
7235 }
|
|
|
7236
|
|
|
7237 if (*p != NUL)
|
|
|
7238 {
|
|
307
|
7239 #ifdef FEAT_MBYTE
|
|
389
|
7240 c = mb_ptr2char_adv(&p);
|
|
307
|
7241 #else
|
|
389
|
7242 c = *p++;
|
|
307
|
7243 #endif
|
|
389
|
7244 new_st.st_fold[i + 128] = c;
|
|
|
7245 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
7246 new_st.st_upper[c] = i + 128;
|
|
|
7247 }
|
|
307
|
7248 }
|
|
|
7249
|
|
483
|
7250 (void)set_spell_finish(&new_st);
|
|
307
|
7251 }
|
|
|
7252
|
|
|
7253 static int
|
|
|
7254 set_spell_finish(new_st)
|
|
|
7255 spelltab_T *new_st;
|
|
|
7256 {
|
|
|
7257 int i;
|
|
|
7258
|
|
|
7259 if (did_set_spelltab)
|
|
|
7260 {
|
|
|
7261 /* check that it's the same table */
|
|
|
7262 for (i = 0; i < 256; ++i)
|
|
|
7263 {
|
|
|
7264 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
7265 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
7266 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
7267 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
7268 {
|
|
|
7269 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
7270 return FAIL;
|
|
|
7271 }
|
|
|
7272 }
|
|
|
7273 }
|
|
|
7274 else
|
|
|
7275 {
|
|
|
7276 /* copy the new spelltab into the one being used */
|
|
|
7277 spelltab = *new_st;
|
|
|
7278 did_set_spelltab = TRUE;
|
|
|
7279 }
|
|
|
7280
|
|
|
7281 return OK;
|
|
|
7282 }
|
|
|
7283
|
|
|
7284 /*
|
|
358
|
7285 * Return TRUE if "p" points to a word character.
|
|
366
|
7286 * As a special case we see "midword" characters as word character when it is
|
|
358
|
7287 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
7288 * Thus this only works properly when past the first character of the word.
|
|
358
|
7289 */
|
|
|
7290 static int
|
|
376
|
7291 spell_iswordp(p, buf)
|
|
358
|
7292 char_u *p;
|
|
376
|
7293 buf_T *buf; /* buffer used */
|
|
358
|
7294 {
|
|
366
|
7295 #ifdef FEAT_MBYTE
|
|
358
|
7296 char_u *s;
|
|
366
|
7297 int l;
|
|
|
7298 int c;
|
|
|
7299
|
|
|
7300 if (has_mbyte)
|
|
|
7301 {
|
|
|
7302 l = MB_BYTE2LEN(*p);
|
|
358
|
7303 s = p;
|
|
366
|
7304 if (l == 1)
|
|
|
7305 {
|
|
|
7306 /* be quick for ASCII */
|
|
376
|
7307 if (buf->b_spell_ismw[*p])
|
|
366
|
7308 {
|
|
|
7309 s = p + 1; /* skip a mid-word character */
|
|
|
7310 l = MB_BYTE2LEN(*s);
|
|
|
7311 }
|
|
|
7312 }
|
|
|
7313 else
|
|
|
7314 {
|
|
|
7315 c = mb_ptr2char(p);
|
|
376
|
7316 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
7317 : (buf->b_spell_ismw_mb != NULL
|
|
|
7318 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
7319 {
|
|
|
7320 s = p + l;
|
|
|
7321 l = MB_BYTE2LEN(*s);
|
|
|
7322 }
|
|
|
7323 }
|
|
|
7324
|
|
390
|
7325 c = mb_ptr2char(s);
|
|
|
7326 if (c > 255)
|
|
366
|
7327 return mb_get_class(s) >= 2;
|
|
390
|
7328 return spelltab.st_isw[c];
|
|
366
|
7329 }
|
|
358
|
7330 #endif
|
|
366
|
7331
|
|
376
|
7332 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
7333 }
|
|
|
7334
|
|
|
7335 /*
|
|
|
7336 * Return TRUE if "p" points to a word character.
|
|
|
7337 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
7338 */
|
|
|
7339 static int
|
|
|
7340 spell_iswordp_nmw(p)
|
|
|
7341 char_u *p;
|
|
|
7342 {
|
|
|
7343 #ifdef FEAT_MBYTE
|
|
390
|
7344 int c;
|
|
|
7345
|
|
|
7346 if (has_mbyte)
|
|
|
7347 {
|
|
|
7348 c = mb_ptr2char(p);
|
|
|
7349 if (c > 255)
|
|
|
7350 return mb_get_class(p) >= 2;
|
|
|
7351 return spelltab.st_isw[c];
|
|
|
7352 }
|
|
376
|
7353 #endif
|
|
|
7354 return spelltab.st_isw[*p];
|
|
358
|
7355 }
|
|
|
7356
|
|
372
|
7357 #ifdef FEAT_MBYTE
|
|
|
7358 /*
|
|
|
7359 * Return TRUE if "p" points to a word character.
|
|
|
7360 * Wide version of spell_iswordp().
|
|
|
7361 */
|
|
|
7362 static int
|
|
376
|
7363 spell_iswordp_w(p, buf)
|
|
372
|
7364 int *p;
|
|
376
|
7365 buf_T *buf;
|
|
372
|
7366 {
|
|
|
7367 int *s;
|
|
|
7368
|
|
376
|
7369 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
7370 : (buf->b_spell_ismw_mb != NULL
|
|
|
7371 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
7372 s = p + 1;
|
|
|
7373 else
|
|
|
7374 s = p;
|
|
|
7375
|
|
390
|
7376 if (*s > 255)
|
|
372
|
7377 {
|
|
|
7378 if (enc_utf8)
|
|
|
7379 return utf_class(*s) >= 2;
|
|
|
7380 if (enc_dbcs)
|
|
|
7381 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
7382 return 0;
|
|
|
7383 }
|
|
|
7384 return spelltab.st_isw[*s];
|
|
|
7385 }
|
|
|
7386 #endif
|
|
|
7387
|
|
358
|
7388 /*
|
|
339
|
7389 * Write the table with prefix conditions to the .spl file.
|
|
483
|
7390 * When "fd" is NULL only count the length of what is written.
|
|
|
7391 */
|
|
|
7392 static int
|
|
339
|
7393 write_spell_prefcond(fd, gap)
|
|
|
7394 FILE *fd;
|
|
|
7395 garray_T *gap;
|
|
|
7396 {
|
|
|
7397 int i;
|
|
|
7398 char_u *p;
|
|
|
7399 int len;
|
|
483
|
7400 int totlen;
|
|
|
7401
|
|
|
7402 if (fd != NULL)
|
|
|
7403 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
7404
|
|
|
7405 totlen = 2 + gap->ga_len; /* length of <prefcondcnt> and <condlen> bytes */
|
|
339
|
7406
|
|
|
7407 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7408 {
|
|
|
7409 /* <prefcond> : <condlen> <condstr> */
|
|
|
7410 p = ((char_u **)gap->ga_data)[i];
|
|
483
|
7411 if (p != NULL)
|
|
339
|
7412 {
|
|
|
7413 len = STRLEN(p);
|
|
483
|
7414 if (fd != NULL)
|
|
|
7415 {
|
|
|
7416 fputc(len, fd);
|
|
|
7417 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
7418 }
|
|
|
7419 totlen += len;
|
|
|
7420 }
|
|
|
7421 else if (fd != NULL)
|
|
|
7422 fputc(0, fd);
|
|
|
7423 }
|
|
|
7424
|
|
|
7425 return totlen;
|
|
307
|
7426 }
|
|
|
7427
|
|
|
7428 /*
|
|
324
|
7429 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
7430 * Uses the character definitions from the .spl file.
|
|
307
|
7431 * When using a multi-byte 'encoding' the length may change!
|
|
|
7432 * Returns FAIL when something wrong.
|
|
|
7433 */
|
|
|
7434 static int
|
|
324
|
7435 spell_casefold(str, len, buf, buflen)
|
|
|
7436 char_u *str;
|
|
307
|
7437 int len;
|
|
|
7438 char_u *buf;
|
|
|
7439 int buflen;
|
|
|
7440 {
|
|
|
7441 int i;
|
|
|
7442
|
|
|
7443 if (len >= buflen)
|
|
|
7444 {
|
|
|
7445 buf[0] = NUL;
|
|
|
7446 return FAIL; /* result will not fit */
|
|
|
7447 }
|
|
|
7448
|
|
|
7449 #ifdef FEAT_MBYTE
|
|
|
7450 if (has_mbyte)
|
|
|
7451 {
|
|
324
|
7452 int outi = 0;
|
|
|
7453 char_u *p;
|
|
307
|
7454 int c;
|
|
|
7455
|
|
|
7456 /* Fold one character at a time. */
|
|
324
|
7457 for (p = str; p < str + len; )
|
|
307
|
7458 {
|
|
|
7459 if (outi + MB_MAXBYTES > buflen)
|
|
|
7460 {
|
|
|
7461 buf[outi] = NUL;
|
|
|
7462 return FAIL;
|
|
|
7463 }
|
|
474
|
7464 c = mb_cptr2char_adv(&p);
|
|
324
|
7465 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
7466 }
|
|
|
7467 buf[outi] = NUL;
|
|
|
7468 }
|
|
|
7469 else
|
|
|
7470 #endif
|
|
|
7471 {
|
|
|
7472 /* Be quick for non-multibyte encodings. */
|
|
|
7473 for (i = 0; i < len; ++i)
|
|
324
|
7474 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
7475 buf[i] = NUL;
|
|
|
7476 }
|
|
|
7477
|
|
|
7478 return OK;
|
|
|
7479 }
|
|
|
7480
|
|
372
|
7481 #define SPS_BEST 1
|
|
|
7482 #define SPS_FAST 2
|
|
|
7483 #define SPS_DOUBLE 4
|
|
|
7484
|
|
|
7485 static int sps_flags = SPS_BEST;
|
|
483
|
7486 static int sps_limit = 9999;
|
|
372
|
7487
|
|
|
7488 /*
|
|
|
7489 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
483
|
7490 * Sets "sps_flags" and "sps_limit".
|
|
372
|
7491 */
|
|
|
7492 int
|
|
|
7493 spell_check_sps()
|
|
|
7494 {
|
|
|
7495 char_u *p;
|
|
483
|
7496 char_u *s;
|
|
372
|
7497 char_u buf[MAXPATHL];
|
|
|
7498 int f;
|
|
|
7499
|
|
|
7500 sps_flags = 0;
|
|
483
|
7501 sps_limit = 9999;
|
|
372
|
7502
|
|
|
7503 for (p = p_sps; *p != NUL; )
|
|
|
7504 {
|
|
|
7505 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
7506
|
|
|
7507 f = 0;
|
|
483
|
7508 if (VIM_ISDIGIT(*buf))
|
|
|
7509 {
|
|
|
7510 s = buf;
|
|
|
7511 sps_limit = getdigits(&s);
|
|
|
7512 if (*s != NUL && !VIM_ISDIGIT(*s))
|
|
|
7513 f = -1;
|
|
|
7514 }
|
|
|
7515 else if (STRCMP(buf, "best") == 0)
|
|
372
|
7516 f = SPS_BEST;
|
|
|
7517 else if (STRCMP(buf, "fast") == 0)
|
|
|
7518 f = SPS_FAST;
|
|
|
7519 else if (STRCMP(buf, "double") == 0)
|
|
|
7520 f = SPS_DOUBLE;
|
|
|
7521 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
7522 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
7523 f = -1;
|
|
|
7524
|
|
|
7525 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
7526 {
|
|
|
7527 sps_flags = SPS_BEST;
|
|
483
|
7528 sps_limit = 9999;
|
|
372
|
7529 return FAIL;
|
|
|
7530 }
|
|
|
7531 if (f != 0)
|
|
|
7532 sps_flags = f;
|
|
|
7533 }
|
|
|
7534
|
|
|
7535 if (sps_flags == 0)
|
|
|
7536 sps_flags = SPS_BEST;
|
|
|
7537
|
|
|
7538 return OK;
|
|
|
7539 }
|
|
|
7540
|
|
|
7541 /* Remember what "z?" replaced. */
|
|
|
7542 static char_u *repl_from = NULL;
|
|
|
7543 static char_u *repl_to = NULL;
|
|
|
7544
|
|
323
|
7545 /*
|
|
|
7546 * "z?": Find badly spelled word under or after the cursor.
|
|
|
7547 * Give suggestions for the properly spelled word.
|
|
|
7548 */
|
|
|
7549 void
|
|
|
7550 spell_suggest()
|
|
|
7551 {
|
|
|
7552 char_u *line;
|
|
|
7553 pos_T prev_cursor = curwin->w_cursor;
|
|
|
7554 char_u wcopy[MAXWLEN + 2];
|
|
|
7555 char_u *p;
|
|
|
7556 int i;
|
|
|
7557 int c;
|
|
|
7558 suginfo_T sug;
|
|
|
7559 suggest_T *stp;
|
|
372
|
7560 int mouse_used;
|
|
388
|
7561 int need_cap;
|
|
483
|
7562 int limit;
|
|
323
|
7563
|
|
344
|
7564 /* Find the start of the badly spelled word. */
|
|
346
|
7565 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL
|
|
|
7566 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
7567 {
|
|
|
7568 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
7569 return;
|
|
|
7570
|
|
|
7571 /* No bad word or it starts after the cursor: use the word under the
|
|
|
7572 * cursor. */
|
|
|
7573 curwin->w_cursor = prev_cursor;
|
|
|
7574 line = ml_get_curline();
|
|
|
7575 p = line + curwin->w_cursor.col;
|
|
|
7576 /* Backup to before start of word. */
|
|
390
|
7577 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
7578 mb_ptr_back(line, p);
|
|
|
7579 /* Forward to start of word. */
|
|
390
|
7580 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
7581 mb_ptr_adv(p);
|
|
|
7582
|
|
390
|
7583 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
7584 {
|
|
|
7585 beep_flush();
|
|
|
7586 return;
|
|
|
7587 }
|
|
|
7588 curwin->w_cursor.col = p - line;
|
|
323
|
7589 }
|
|
|
7590
|
|
344
|
7591 /* Get the word and its length. */
|
|
|
7592
|
|
388
|
7593 /* Figure out if the word should be capitalised. */
|
|
475
|
7594 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
7595
|
|
|
7596 line = ml_get_curline();
|
|
388
|
7597
|
|
483
|
7598 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in
|
|
|
7599 * 'spellsuggest', whatever is smaller. */
|
|
|
7600 if (sps_limit > (int)Rows - 2)
|
|
|
7601 limit = (int)Rows - 2;
|
|
|
7602 else
|
|
|
7603 limit = sps_limit;
|
|
|
7604 spell_find_suggest(line + curwin->w_cursor.col, &sug, limit,
|
|
388
|
7605 TRUE, need_cap);
|
|
323
|
7606
|
|
|
7607 if (sug.su_ga.ga_len == 0)
|
|
|
7608 MSG(_("Sorry, no suggestions"));
|
|
|
7609 else
|
|
|
7610 {
|
|
372
|
7611 vim_free(repl_from);
|
|
|
7612 repl_from = NULL;
|
|
|
7613 vim_free(repl_to);
|
|
|
7614 repl_to = NULL;
|
|
|
7615
|
|
474
|
7616 #ifdef FEAT_RIGHTLEFT
|
|
|
7617 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
7618 cmdmsg_rl = curwin->w_p_rl;
|
|
|
7619 if (cmdmsg_rl)
|
|
|
7620 msg_col = Columns - 1;
|
|
|
7621 #endif
|
|
|
7622
|
|
323
|
7623 /* List the suggestions. */
|
|
|
7624 msg_start();
|
|
372
|
7625 lines_left = Rows; /* avoid more prompt */
|
|
323
|
7626 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
7627 sug.su_badlen, sug.su_badptr);
|
|
474
|
7628 #ifdef FEAT_RIGHTLEFT
|
|
|
7629 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
7630 {
|
|
|
7631 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
7632 * untranslated message rightleft. */
|
|
|
7633 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
7634 sug.su_badlen, sug.su_badptr);
|
|
|
7635 }
|
|
|
7636 #endif
|
|
323
|
7637 msg_puts(IObuff);
|
|
|
7638 msg_clr_eos();
|
|
|
7639 msg_putchar('\n');
|
|
346
|
7640
|
|
323
|
7641 msg_scroll = TRUE;
|
|
|
7642 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
7643 {
|
|
344
|
7644 stp = &SUG(sug.su_ga, i);
|
|
323
|
7645
|
|
|
7646 /* The suggested word may replace only part of the bad word, add
|
|
|
7647 * the not replaced part. */
|
|
|
7648 STRCPY(wcopy, stp->st_word);
|
|
|
7649 if (sug.su_badlen > stp->st_orglen)
|
|
|
7650 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
7651 sug.su_badptr + stp->st_orglen,
|
|
|
7652 sug.su_badlen - stp->st_orglen);
|
|
474
|
7653 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
7654 #ifdef FEAT_RIGHTLEFT
|
|
|
7655 if (cmdmsg_rl)
|
|
|
7656 rl_mirror(IObuff);
|
|
|
7657 #endif
|
|
|
7658 msg_puts(IObuff);
|
|
|
7659
|
|
|
7660 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
7661 msg_puts(IObuff);
|
|
|
7662
|
|
|
7663 /* The word may replace more than "su_badlen". */
|
|
|
7664 if (sug.su_badlen < stp->st_orglen)
|
|
|
7665 {
|
|
|
7666 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
7667 stp->st_orglen, sug.su_badptr);
|
|
|
7668 msg_puts(IObuff);
|
|
|
7669 }
|
|
|
7670
|
|
324
|
7671 if (p_verbose > 0)
|
|
344
|
7672 {
|
|
346
|
7673 /* Add the score. */
|
|
351
|
7674 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
7675 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
7676 stp->st_salscore ? "s " : "",
|
|
|
7677 stp->st_score, stp->st_altscore);
|
|
|
7678 else
|
|
474
|
7679 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
7680 stp->st_score);
|
|
474
|
7681 #ifdef FEAT_RIGHTLEFT
|
|
|
7682 if (cmdmsg_rl)
|
|
|
7683 /* Mirror the numbers, but keep the leading space. */
|
|
|
7684 rl_mirror(IObuff + 1);
|
|
|
7685 #endif
|
|
346
|
7686 msg_advance(30);
|
|
|
7687 msg_puts(IObuff);
|
|
344
|
7688 }
|
|
323
|
7689 msg_putchar('\n');
|
|
|
7690 }
|
|
|
7691
|
|
474
|
7692 #ifdef FEAT_RIGHTLEFT
|
|
|
7693 cmdmsg_rl = FALSE;
|
|
|
7694 msg_col = 0;
|
|
|
7695 #endif
|
|
323
|
7696 /* Ask for choice. */
|
|
372
|
7697 i = prompt_for_number(&mouse_used);
|
|
|
7698 if (mouse_used)
|
|
|
7699 i -= lines_left;
|
|
|
7700
|
|
344
|
7701 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
323
|
7702 {
|
|
372
|
7703 /* Save the from and to text for :spellrepall. */
|
|
|
7704 stp = &SUG(sug.su_ga, i - 1);
|
|
|
7705 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
7706 repl_to = vim_strsave(stp->st_word);
|
|
|
7707
|
|
323
|
7708 /* Replace the word. */
|
|
|
7709 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
7710 if (p != NULL)
|
|
|
7711 {
|
|
|
7712 c = sug.su_badptr - line;
|
|
|
7713 mch_memmove(p, line, c);
|
|
|
7714 STRCPY(p + c, stp->st_word);
|
|
|
7715 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
7716 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
7717 curwin->w_cursor.col = c;
|
|
|
7718 changed_bytes(curwin->w_cursor.lnum, c);
|
|
344
|
7719
|
|
|
7720 /* For redo we use a change-word command. */
|
|
|
7721 ResetRedobuff();
|
|
|
7722 AppendToRedobuff((char_u *)"ciw");
|
|
|
7723 AppendToRedobuff(stp->st_word);
|
|
|
7724 AppendCharToRedobuff(ESC);
|
|
323
|
7725 }
|
|
|
7726 }
|
|
|
7727 else
|
|
|
7728 curwin->w_cursor = prev_cursor;
|
|
|
7729 }
|
|
|
7730
|
|
344
|
7731 spell_find_cleanup(&sug);
|
|
|
7732 }
|
|
|
7733
|
|
|
7734 /*
|
|
475
|
7735 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
7736 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
7737 */
|
|
|
7738 static int
|
|
|
7739 check_need_cap(lnum, col)
|
|
|
7740 linenr_T lnum;
|
|
|
7741 colnr_T col;
|
|
|
7742 {
|
|
|
7743 int need_cap = FALSE;
|
|
|
7744 char_u *line;
|
|
|
7745 char_u *line_copy = NULL;
|
|
|
7746 char_u *p;
|
|
|
7747 colnr_T endcol;
|
|
|
7748 regmatch_T regmatch;
|
|
|
7749
|
|
|
7750 if (curbuf->b_cap_prog == NULL)
|
|
|
7751 return FALSE;
|
|
|
7752
|
|
|
7753 line = ml_get_curline();
|
|
|
7754 endcol = 0;
|
|
|
7755 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
7756 {
|
|
|
7757 /* At start of line, check if previous line is empty or sentence
|
|
|
7758 * ends there. */
|
|
|
7759 if (lnum == 1)
|
|
|
7760 need_cap = TRUE;
|
|
|
7761 else
|
|
|
7762 {
|
|
|
7763 line = ml_get(lnum - 1);
|
|
|
7764 if (*skipwhite(line) == NUL)
|
|
|
7765 need_cap = TRUE;
|
|
|
7766 else
|
|
|
7767 {
|
|
|
7768 /* Append a space in place of the line break. */
|
|
|
7769 line_copy = concat_str(line, (char_u *)" ");
|
|
|
7770 line = line_copy;
|
|
|
7771 endcol = STRLEN(line);
|
|
|
7772 }
|
|
|
7773 }
|
|
|
7774 }
|
|
|
7775 else
|
|
|
7776 endcol = col;
|
|
|
7777
|
|
|
7778 if (endcol > 0)
|
|
|
7779 {
|
|
|
7780 /* Check if sentence ends before the bad word. */
|
|
|
7781 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
7782 regmatch.rm_ic = FALSE;
|
|
|
7783 p = line + endcol;
|
|
|
7784 for (;;)
|
|
|
7785 {
|
|
|
7786 mb_ptr_back(line, p);
|
|
|
7787 if (p == line || spell_iswordp_nmw(p))
|
|
|
7788 break;
|
|
|
7789 if (vim_regexec(®match, p, 0)
|
|
|
7790 && regmatch.endp[0] == line + endcol)
|
|
|
7791 {
|
|
|
7792 need_cap = TRUE;
|
|
|
7793 break;
|
|
|
7794 }
|
|
|
7795 }
|
|
|
7796 }
|
|
|
7797
|
|
|
7798 vim_free(line_copy);
|
|
|
7799
|
|
|
7800 return need_cap;
|
|
|
7801 }
|
|
|
7802
|
|
|
7803
|
|
|
7804 /*
|
|
372
|
7805 * ":spellrepall"
|
|
|
7806 */
|
|
|
7807 /*ARGSUSED*/
|
|
|
7808 void
|
|
|
7809 ex_spellrepall(eap)
|
|
|
7810 exarg_T *eap;
|
|
|
7811 {
|
|
|
7812 pos_T pos = curwin->w_cursor;
|
|
|
7813 char_u *frompat;
|
|
|
7814 int addlen;
|
|
|
7815 char_u *line;
|
|
|
7816 char_u *p;
|
|
|
7817 int save_ws = p_ws;
|
|
483
|
7818 linenr_T prev_lnum = 0;
|
|
372
|
7819
|
|
|
7820 if (repl_from == NULL || repl_to == NULL)
|
|
|
7821 {
|
|
|
7822 EMSG(_("E752: No previous spell replacement"));
|
|
|
7823 return;
|
|
|
7824 }
|
|
|
7825 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
7826
|
|
|
7827 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
7828 if (frompat == NULL)
|
|
|
7829 return;
|
|
|
7830 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
7831 p_ws = FALSE;
|
|
|
7832
|
|
483
|
7833 sub_nsubs = 0;
|
|
|
7834 sub_nlines = 0;
|
|
372
|
7835 curwin->w_cursor.lnum = 0;
|
|
|
7836 while (!got_int)
|
|
|
7837 {
|
|
|
7838 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
7839 || u_save_cursor() == FAIL)
|
|
|
7840 break;
|
|
|
7841
|
|
|
7842 /* Only replace when the right word isn't there yet. This happens
|
|
|
7843 * when changing "etc" to "etc.". */
|
|
|
7844 line = ml_get_curline();
|
|
|
7845 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
7846 repl_to, STRLEN(repl_to)) != 0)
|
|
|
7847 {
|
|
|
7848 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
7849 if (p == NULL)
|
|
|
7850 break;
|
|
|
7851 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
7852 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
7853 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
7854 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
7855 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
483
|
7856
|
|
|
7857 if (curwin->w_cursor.lnum != prev_lnum)
|
|
|
7858 {
|
|
|
7859 ++sub_nlines;
|
|
|
7860 prev_lnum = curwin->w_cursor.lnum;
|
|
|
7861 }
|
|
|
7862 ++sub_nsubs;
|
|
372
|
7863 }
|
|
|
7864 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
7865 }
|
|
|
7866
|
|
|
7867 p_ws = save_ws;
|
|
|
7868 curwin->w_cursor = pos;
|
|
|
7869 vim_free(frompat);
|
|
|
7870
|
|
483
|
7871 if (sub_nsubs == 0)
|
|
372
|
7872 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
483
|
7873 else
|
|
|
7874 do_sub_msg(FALSE);
|
|
372
|
7875 }
|
|
|
7876
|
|
|
7877 /*
|
|
344
|
7878 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
7879 * a list of allocated strings.
|
|
|
7880 */
|
|
|
7881 void
|
|
475
|
7882 spell_suggest_list(gap, word, maxcount, need_cap)
|
|
344
|
7883 garray_T *gap;
|
|
|
7884 char_u *word;
|
|
|
7885 int maxcount; /* maximum nr of suggestions */
|
|
475
|
7886 int need_cap; /* 'spellcapcheck' matched */
|
|
344
|
7887 {
|
|
|
7888 suginfo_T sug;
|
|
|
7889 int i;
|
|
|
7890 suggest_T *stp;
|
|
|
7891 char_u *wcopy;
|
|
|
7892
|
|
475
|
7893 spell_find_suggest(word, &sug, maxcount, FALSE, need_cap);
|
|
344
|
7894
|
|
|
7895 /* Make room in "gap". */
|
|
|
7896 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
7897 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
7898 return;
|
|
|
7899
|
|
323
|
7900 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
7901 {
|
|
|
7902 stp = &SUG(sug.su_ga, i);
|
|
|
7903
|
|
|
7904 /* The suggested word may replace only part of "word", add the not
|
|
|
7905 * replaced part. */
|
|
|
7906 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
7907 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
7908 if (wcopy == NULL)
|
|
|
7909 break;
|
|
|
7910 STRCPY(wcopy, stp->st_word);
|
|
|
7911 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
7912 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
7913 }
|
|
|
7914
|
|
|
7915 spell_find_cleanup(&sug);
|
|
|
7916 }
|
|
|
7917
|
|
|
7918 /*
|
|
|
7919 * Find spell suggestions for the word at the start of "badptr".
|
|
|
7920 * Return the suggestions in "su->su_ga".
|
|
|
7921 * The maximum number of suggestions is "maxcount".
|
|
|
7922 * Note: does use info for the current window.
|
|
|
7923 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
7924 */
|
|
|
7925 static void
|
|
388
|
7926 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
7927 char_u *badptr;
|
|
|
7928 suginfo_T *su;
|
|
|
7929 int maxcount;
|
|
358
|
7930 int banbadword; /* don't include badword in suggestions */
|
|
388
|
7931 int need_cap; /* word should start with capital */
|
|
344
|
7932 {
|
|
385
|
7933 int attr = 0;
|
|
372
|
7934 char_u buf[MAXPATHL];
|
|
|
7935 char_u *p;
|
|
|
7936 int do_combine = FALSE;
|
|
|
7937 char_u *sps_copy;
|
|
|
7938 #ifdef FEAT_EVAL
|
|
|
7939 static int expr_busy = FALSE;
|
|
|
7940 #endif
|
|
385
|
7941 int c;
|
|
344
|
7942
|
|
|
7943 /*
|
|
|
7944 * Set the info in "*su".
|
|
|
7945 */
|
|
|
7946 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
7947 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
7948 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
7949 if (*badptr == NUL)
|
|
|
7950 return;
|
|
344
|
7951 hash_init(&su->su_banned);
|
|
|
7952
|
|
|
7953 su->su_badptr = badptr;
|
|
385
|
7954 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
7955 su->su_maxcount = maxcount;
|
|
372
|
7956 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
7957
|
|
|
7958 if (su->su_badlen >= MAXWLEN)
|
|
|
7959 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
7960 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
7961 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
7962 su->su_fbadword, MAXWLEN);
|
|
346
|
7963 /* get caps flags for bad word */
|
|
474
|
7964 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
7965 su->su_badptr + su->su_badlen);
|
|
388
|
7966 if (need_cap)
|
|
|
7967 su->su_badflags |= WF_ONECAP;
|
|
344
|
7968
|
|
385
|
7969 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
7970 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
7971 * for that. */
|
|
455
|
7972 c = PTR2CHAR(su->su_badptr);
|
|
385
|
7973 if (!SPELL_ISUPPER(c) && attr == 0)
|
|
|
7974 {
|
|
|
7975 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
7976 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
|
7977 0, TRUE);
|
|
|
7978 }
|
|
|
7979
|
|
344
|
7980 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
7981 if (banbadword)
|
|
|
7982 add_banned(su, su->su_badword);
|
|
344
|
7983
|
|
372
|
7984 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
7985 sps_copy = vim_strsave(p_sps);
|
|
|
7986 if (sps_copy == NULL)
|
|
|
7987 return;
|
|
|
7988
|
|
|
7989 /* Loop over the items in 'spellsuggest'. */
|
|
|
7990 for (p = sps_copy; *p != NUL; )
|
|
|
7991 {
|
|
|
7992 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
7993
|
|
|
7994 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
7995 {
|
|
|
7996 #ifdef FEAT_EVAL
|
|
375
|
7997 /* Evaluate an expression. Skip this when called recursively,
|
|
|
7998 * when using spellsuggest() in the expression. */
|
|
372
|
7999 if (!expr_busy)
|
|
|
8000 {
|
|
|
8001 expr_busy = TRUE;
|
|
|
8002 spell_suggest_expr(su, buf + 5);
|
|
|
8003 expr_busy = FALSE;
|
|
|
8004 }
|
|
|
8005 #endif
|
|
|
8006 }
|
|
|
8007 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
8008 /* Use list of suggestions in a file. */
|
|
|
8009 spell_suggest_file(su, buf + 5);
|
|
|
8010 else
|
|
|
8011 {
|
|
|
8012 /* Use internal method. */
|
|
|
8013 spell_suggest_intern(su);
|
|
|
8014 if (sps_flags & SPS_DOUBLE)
|
|
|
8015 do_combine = TRUE;
|
|
|
8016 }
|
|
|
8017 }
|
|
|
8018
|
|
|
8019 vim_free(sps_copy);
|
|
|
8020
|
|
|
8021 if (do_combine)
|
|
|
8022 /* Combine the two list of suggestions. This must be done last,
|
|
|
8023 * because sorting changes the order again. */
|
|
|
8024 score_combine(su);
|
|
|
8025 }
|
|
|
8026
|
|
|
8027 #ifdef FEAT_EVAL
|
|
|
8028 /*
|
|
|
8029 * Find suggestions by evaluating expression "expr".
|
|
|
8030 */
|
|
|
8031 static void
|
|
|
8032 spell_suggest_expr(su, expr)
|
|
|
8033 suginfo_T *su;
|
|
|
8034 char_u *expr;
|
|
|
8035 {
|
|
|
8036 list_T *list;
|
|
|
8037 listitem_T *li;
|
|
|
8038 int score;
|
|
|
8039 char_u *p;
|
|
|
8040
|
|
|
8041 /* The work is split up in a few parts to avoid having to export
|
|
|
8042 * suginfo_T.
|
|
|
8043 * First evaluate the expression and get the resulting list. */
|
|
|
8044 list = eval_spell_expr(su->su_badword, expr);
|
|
|
8045 if (list != NULL)
|
|
|
8046 {
|
|
|
8047 /* Loop over the items in the list. */
|
|
|
8048 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
8049 if (li->li_tv.v_type == VAR_LIST)
|
|
|
8050 {
|
|
|
8051 /* Get the word and the score from the items. */
|
|
|
8052 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
8053 if (score >= 0)
|
|
|
8054 add_suggestion(su, &su->su_ga, p,
|
|
|
8055 su->su_badlen, score, 0, TRUE);
|
|
|
8056 }
|
|
|
8057 list_unref(list);
|
|
|
8058 }
|
|
|
8059
|
|
|
8060 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8061 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8062 }
|
|
|
8063 #endif
|
|
|
8064
|
|
|
8065 /*
|
|
474
|
8066 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
8067 */
|
|
|
8068 static void
|
|
|
8069 spell_suggest_file(su, fname)
|
|
|
8070 suginfo_T *su;
|
|
|
8071 char_u *fname;
|
|
|
8072 {
|
|
|
8073 FILE *fd;
|
|
|
8074 char_u line[MAXWLEN * 2];
|
|
|
8075 char_u *p;
|
|
|
8076 int len;
|
|
|
8077 char_u cword[MAXWLEN];
|
|
|
8078
|
|
|
8079 /* Open the file. */
|
|
|
8080 fd = mch_fopen((char *)fname, "r");
|
|
|
8081 if (fd == NULL)
|
|
|
8082 {
|
|
|
8083 EMSG2(_(e_notopen), fname);
|
|
|
8084 return;
|
|
|
8085 }
|
|
|
8086
|
|
|
8087 /* Read it line by line. */
|
|
|
8088 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
8089 {
|
|
|
8090 line_breakcheck();
|
|
|
8091
|
|
|
8092 p = vim_strchr(line, '/');
|
|
|
8093 if (p == NULL)
|
|
|
8094 continue; /* No Tab found, just skip the line. */
|
|
|
8095 *p++ = NUL;
|
|
|
8096 if (STRICMP(su->su_badword, line) == 0)
|
|
|
8097 {
|
|
|
8098 /* Match! Isolate the good word, until CR or NL. */
|
|
|
8099 for (len = 0; p[len] >= ' '; ++len)
|
|
|
8100 ;
|
|
|
8101 p[len] = NUL;
|
|
|
8102
|
|
|
8103 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
8104 * of the bad word. */
|
|
|
8105 if (captype(p, NULL) == 0)
|
|
|
8106 {
|
|
|
8107 make_case_word(p, cword, su->su_badflags);
|
|
|
8108 p = cword;
|
|
|
8109 }
|
|
|
8110
|
|
|
8111 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
|
8112 SCORE_FILE, 0, TRUE);
|
|
|
8113 }
|
|
|
8114 }
|
|
|
8115
|
|
|
8116 fclose(fd);
|
|
|
8117
|
|
|
8118 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8119 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8120 }
|
|
|
8121
|
|
|
8122 /*
|
|
|
8123 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
8124 */
|
|
|
8125 static void
|
|
|
8126 spell_suggest_intern(su)
|
|
|
8127 suginfo_T *su;
|
|
|
8128 {
|
|
344
|
8129 /*
|
|
346
|
8130 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
8131 *
|
|
|
8132 * Set a maximum score to limit the combination of operations that is
|
|
|
8133 * tried.
|
|
|
8134 */
|
|
346
|
8135 suggest_try_special(su);
|
|
|
8136
|
|
|
8137 /*
|
|
|
8138 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
8139 * from the .aff file and inserting a space (split the word).
|
|
|
8140 */
|
|
|
8141 suggest_try_change(su);
|
|
344
|
8142
|
|
|
8143 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
8144 if (sps_flags & SPS_DOUBLE)
|
|
|
8145 score_comp_sal(su);
|
|
|
8146
|
|
|
8147 /*
|
|
346
|
8148 * 3. Try finding sound-a-like words.
|
|
344
|
8149 *
|
|
|
8150 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
8151 * very slow and often doesn't find new suggestions.
|
|
|
8152 */
|
|
|
8153 if ((sps_flags & SPS_DOUBLE)
|
|
|
8154 || (!(sps_flags & SPS_FAST)
|
|
|
8155 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
8156 {
|
|
|
8157 /* Allow a higher score now. */
|
|
|
8158 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
8159 suggest_try_soundalike(su);
|
|
344
|
8160 }
|
|
|
8161
|
|
|
8162 /* When CTRL-C was hit while searching do show the results. */
|
|
|
8163 ui_breakcheck();
|
|
|
8164 if (got_int)
|
|
|
8165 {
|
|
|
8166 (void)vgetc();
|
|
|
8167 got_int = FALSE;
|
|
|
8168 }
|
|
|
8169
|
|
372
|
8170 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
8171 {
|
|
|
8172 if (sps_flags & SPS_BEST)
|
|
|
8173 /* Adjust the word score for how it sounds like. */
|
|
|
8174 rescore_suggestions(su);
|
|
|
8175
|
|
|
8176 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
8177 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
8178 }
|
|
|
8179 }
|
|
|
8180
|
|
|
8181 /*
|
|
|
8182 * Free the info put in "*su" by spell_find_suggest().
|
|
|
8183 */
|
|
|
8184 static void
|
|
|
8185 spell_find_cleanup(su)
|
|
|
8186 suginfo_T *su;
|
|
|
8187 {
|
|
|
8188 int i;
|
|
|
8189
|
|
|
8190 /* Free the suggestions. */
|
|
|
8191 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
8192 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
8193 ga_clear(&su->su_ga);
|
|
|
8194 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
8195 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
8196 ga_clear(&su->su_sga);
|
|
323
|
8197
|
|
|
8198 /* Free the banned words. */
|
|
344
|
8199 free_banned(su);
|
|
323
|
8200 }
|
|
|
8201
|
|
|
8202 /*
|
|
324
|
8203 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
8204 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
8205 * The result is NUL terminated.
|
|
323
|
8206 */
|
|
|
8207 static void
|
|
324
|
8208 onecap_copy(word, wcopy, upper)
|
|
323
|
8209 char_u *word;
|
|
|
8210 char_u *wcopy;
|
|
|
8211 int upper; /* TRUE: first letter made upper case */
|
|
|
8212 {
|
|
|
8213 char_u *p;
|
|
|
8214 int c;
|
|
|
8215 int l;
|
|
|
8216
|
|
|
8217 p = word;
|
|
|
8218 #ifdef FEAT_MBYTE
|
|
|
8219 if (has_mbyte)
|
|
474
|
8220 c = mb_cptr2char_adv(&p);
|
|
323
|
8221 else
|
|
|
8222 #endif
|
|
|
8223 c = *p++;
|
|
|
8224 if (upper)
|
|
324
|
8225 c = SPELL_TOUPPER(c);
|
|
323
|
8226 else
|
|
324
|
8227 c = SPELL_TOFOLD(c);
|
|
323
|
8228 #ifdef FEAT_MBYTE
|
|
|
8229 if (has_mbyte)
|
|
|
8230 l = mb_char2bytes(c, wcopy);
|
|
|
8231 else
|
|
|
8232 #endif
|
|
|
8233 {
|
|
|
8234 l = 1;
|
|
|
8235 wcopy[0] = c;
|
|
|
8236 }
|
|
376
|
8237 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
8238 }
|
|
|
8239
|
|
|
8240 /*
|
|
324
|
8241 * Make a copy of "word" with all the letters upper cased into
|
|
|
8242 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
8243 */
|
|
|
8244 static void
|
|
|
8245 allcap_copy(word, wcopy)
|
|
|
8246 char_u *word;
|
|
|
8247 char_u *wcopy;
|
|
|
8248 {
|
|
|
8249 char_u *s;
|
|
|
8250 char_u *d;
|
|
|
8251 int c;
|
|
|
8252
|
|
|
8253 d = wcopy;
|
|
|
8254 for (s = word; *s != NUL; )
|
|
|
8255 {
|
|
|
8256 #ifdef FEAT_MBYTE
|
|
|
8257 if (has_mbyte)
|
|
474
|
8258 c = mb_cptr2char_adv(&s);
|
|
323
|
8259 else
|
|
|
8260 #endif
|
|
|
8261 c = *s++;
|
|
324
|
8262 c = SPELL_TOUPPER(c);
|
|
323
|
8263
|
|
|
8264 #ifdef FEAT_MBYTE
|
|
|
8265 if (has_mbyte)
|
|
|
8266 {
|
|
|
8267 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
8268 break;
|
|
|
8269 d += mb_char2bytes(c, d);
|
|
|
8270 }
|
|
|
8271 else
|
|
|
8272 #endif
|
|
|
8273 {
|
|
|
8274 if (d - wcopy >= MAXWLEN - 1)
|
|
|
8275 break;
|
|
|
8276 *d++ = c;
|
|
|
8277 }
|
|
|
8278 }
|
|
|
8279 *d = NUL;
|
|
|
8280 }
|
|
|
8281
|
|
|
8282 /*
|
|
346
|
8283 * Try finding suggestions by recognizing specific situations.
|
|
|
8284 */
|
|
|
8285 static void
|
|
|
8286 suggest_try_special(su)
|
|
|
8287 suginfo_T *su;
|
|
|
8288 {
|
|
|
8289 char_u *p;
|
|
376
|
8290 size_t len;
|
|
346
|
8291 int c;
|
|
|
8292 char_u word[MAXWLEN];
|
|
|
8293
|
|
|
8294 /*
|
|
|
8295 * Recognize a word that is repeated: "the the".
|
|
|
8296 */
|
|
|
8297 p = skiptowhite(su->su_fbadword);
|
|
|
8298 len = p - su->su_fbadword;
|
|
|
8299 p = skipwhite(p);
|
|
|
8300 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
8301 {
|
|
|
8302 /* Include badflags: if the badword is onecap or allcap
|
|
|
8303 * use that for the goodword too: "The the" -> "The". */
|
|
|
8304 c = su->su_fbadword[len];
|
|
|
8305 su->su_fbadword[len] = NUL;
|
|
|
8306 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
8307 su->su_fbadword[len] = c;
|
|
351
|
8308 add_suggestion(su, &su->su_ga, word, su->su_badlen, SCORE_DEL, 0, TRUE);
|
|
346
|
8309 }
|
|
|
8310 }
|
|
|
8311
|
|
|
8312 /*
|
|
323
|
8313 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
8314 *
|
|
|
8315 * This uses a state machine. At each node in the tree we try various
|
|
|
8316 * operations. When trying if an operation work "depth" is increased and the
|
|
|
8317 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
8318 * character, replace one and delete another. The number of changes is
|
|
|
8319 * limited by su->su_maxscore, checked in try_deeper().
|
|
323
|
8320 */
|
|
|
8321 static void
|
|
346
|
8322 suggest_try_change(su)
|
|
323
|
8323 suginfo_T *su;
|
|
|
8324 {
|
|
|
8325 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
8326 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
8327 trystate_T stack[MAXWLEN];
|
|
483
|
8328 char_u preword[MAXWLEN * 3]; /* word found with proper case;
|
|
|
8329 * concatanation of prefix compound
|
|
|
8330 * words and split word. NUL terminated
|
|
|
8331 * when going deeper but not when coming
|
|
|
8332 * back. */
|
|
|
8333 char_u compflags[MAXWLEN]; /* compound flags, one for each word */
|
|
323
|
8334 trystate_T *sp;
|
|
|
8335 int newscore;
|
|
|
8336 langp_T *lp;
|
|
375
|
8337 char_u *byts, *fbyts, *pbyts;
|
|
|
8338 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
8339 int depth;
|
|
330
|
8340 int c, c2, c3;
|
|
390
|
8341 int n;
|
|
323
|
8342 int flags;
|
|
|
8343 garray_T *gap;
|
|
324
|
8344 idx_T arridx;
|
|
323
|
8345 int len;
|
|
|
8346 char_u *p;
|
|
|
8347 fromto_T *ftp;
|
|
330
|
8348 int fl = 0, tl;
|
|
346
|
8349 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
482
|
8350 slang_T *slang;
|
|
|
8351 int fword_ends;
|
|
323
|
8352
|
|
|
8353 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
8354 * to find matches (esp. REP items). Append some more text, changing
|
|
|
8355 * chars after the bad word may help. */
|
|
323
|
8356 STRCPY(fword, su->su_fbadword);
|
|
346
|
8357 n = STRLEN(fword);
|
|
|
8358 p = su->su_badptr + su->su_badlen;
|
|
|
8359 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
8360
|
|
|
8361 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
8362 lp->lp_slang != NULL; ++lp)
|
|
|
8363 {
|
|
482
|
8364 slang = lp->lp_slang;
|
|
|
8365
|
|
323
|
8366 /*
|
|
|
8367 * Go through the whole case-fold tree, try changes at each node.
|
|
|
8368 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
8369 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
8370 * word).
|
|
|
8371 */
|
|
|
8372 depth = 0;
|
|
375
|
8373 sp = &stack[0];
|
|
483
|
8374 vim_memset(sp, 0, sizeof(trystate_T));
|
|
375
|
8375 sp->ts_state = STATE_START;
|
|
|
8376 sp->ts_curi = 1;
|
|
330
|
8377
|
|
|
8378 /*
|
|
375
|
8379 * When there are postponed prefixes we need to use these first. At
|
|
|
8380 * the end of the prefix we continue in the case-fold tree.
|
|
|
8381 */
|
|
482
|
8382 fbyts = slang->sl_fbyts;
|
|
|
8383 fidxs = slang->sl_fidxs;
|
|
|
8384 pbyts = slang->sl_pbyts;
|
|
|
8385 pidxs = slang->sl_pidxs;
|
|
375
|
8386 if (pbyts != NULL)
|
|
|
8387 {
|
|
|
8388 byts = pbyts;
|
|
|
8389 idxs = pidxs;
|
|
482
|
8390 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
375
|
8391 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
8392 }
|
|
|
8393 else
|
|
|
8394 {
|
|
|
8395 byts = fbyts;
|
|
|
8396 idxs = fidxs;
|
|
482
|
8397 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
375
|
8398 }
|
|
|
8399
|
|
|
8400 /*
|
|
330
|
8401 * Loop to find all suggestions. At each round we either:
|
|
|
8402 * - For the current state try one operation, advance "ts_curi",
|
|
|
8403 * increase "depth".
|
|
|
8404 * - When a state is done go to the next, set "ts_state".
|
|
|
8405 * - When all states are tried decrease "depth".
|
|
|
8406 */
|
|
323
|
8407 while (depth >= 0 && !got_int)
|
|
|
8408 {
|
|
|
8409 sp = &stack[depth];
|
|
|
8410 switch (sp->ts_state)
|
|
|
8411 {
|
|
|
8412 case STATE_START:
|
|
375
|
8413 case STATE_NOPREFIX:
|
|
323
|
8414 /*
|
|
|
8415 * Start of node: Deal with NUL bytes, which means
|
|
|
8416 * tword[] may end here.
|
|
|
8417 */
|
|
|
8418 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
8419 len = byts[arridx]; /* bytes in this node */
|
|
|
8420 arridx += sp->ts_curi; /* index of current byte */
|
|
|
8421
|
|
482
|
8422 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
8423 {
|
|
|
8424 /* Skip over the NUL bytes, we use them later. */
|
|
|
8425 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
8426 ;
|
|
|
8427 sp->ts_curi += n;
|
|
|
8428
|
|
390
|
8429 /* Always past NUL bytes now. */
|
|
|
8430 n = (int)sp->ts_state;
|
|
|
8431 sp->ts_state = STATE_ENDNUL;
|
|
455
|
8432 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
8433
|
|
375
|
8434 /* At end of a prefix or at start of prefixtree: check for
|
|
|
8435 * following word. */
|
|
390
|
8436 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
8437 {
|
|
455
|
8438 /* Set su->su_badflags to the caps type at this
|
|
|
8439 * position. Use the caps type until here for the
|
|
|
8440 * prefix itself. */
|
|
|
8441 #ifdef FEAT_MBYTE
|
|
|
8442 if (has_mbyte)
|
|
|
8443 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
8444 else
|
|
|
8445 #endif
|
|
|
8446 n = sp->ts_fidx;
|
|
474
|
8447 flags = badword_captype(su->su_badptr,
|
|
|
8448 su->su_badptr + n);
|
|
|
8449 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
8450 su->su_badptr + su->su_badlen);
|
|
375
|
8451 ++depth;
|
|
|
8452 stack[depth] = stack[depth - 1];
|
|
|
8453 sp = &stack[depth];
|
|
|
8454 sp->ts_prefixdepth = depth - 1;
|
|
|
8455 byts = fbyts;
|
|
|
8456 idxs = fidxs;
|
|
|
8457 sp->ts_state = STATE_START;
|
|
|
8458 sp->ts_curi = 1; /* start just after length byte */
|
|
|
8459 sp->ts_arridx = 0;
|
|
|
8460
|
|
455
|
8461 /* Move the prefix to preword[] with the right case
|
|
|
8462 * and make find_keepcap_word() works. */
|
|
483
|
8463 sp->ts_splitoff = sp->ts_twordlen;
|
|
|
8464 tword[sp->ts_splitoff] = NUL;
|
|
455
|
8465 make_case_word(tword, preword, flags);
|
|
483
|
8466 sp->ts_prewordlen = STRLEN(preword);
|
|
375
|
8467 }
|
|
|
8468 break;
|
|
|
8469 }
|
|
|
8470
|
|
346
|
8471 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
8472 {
|
|
|
8473 /* Past bytes in node and/or past NUL bytes. */
|
|
|
8474 sp->ts_state = STATE_ENDNUL;
|
|
455
|
8475 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
8476 break;
|
|
|
8477 }
|
|
|
8478
|
|
|
8479 /*
|
|
|
8480 * End of word in tree.
|
|
|
8481 */
|
|
|
8482 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
8483
|
|
324
|
8484 flags = (int)idxs[arridx];
|
|
483
|
8485 fword_ends = (fword[sp->ts_fidx] == NUL
|
|
|
8486 || !spell_iswordp(fword + sp->ts_fidx, curbuf));
|
|
|
8487 tword[sp->ts_twordlen] = NUL;
|
|
323
|
8488
|
|
482
|
8489 if (sp->ts_prefixdepth == PFD_COMPOUND)
|
|
|
8490 {
|
|
|
8491 /* There was a compound word before this word. If this
|
|
|
8492 * word does not support compounding then give up
|
|
|
8493 * (splitting is tried for the word without compound
|
|
|
8494 * flag). */
|
|
483
|
8495 if (((unsigned)flags >> 24) == 0
|
|
|
8496 || sp->ts_twordlen - sp->ts_splitoff
|
|
|
8497 < slang->sl_compminlen)
|
|
|
8498 break;
|
|
|
8499 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
8500 compflags[sp->ts_complen + 1] = NUL;
|
|
|
8501 if (fword_ends && !can_compound(slang, tword, compflags))
|
|
482
|
8502 break;
|
|
|
8503 }
|
|
|
8504 else if (sp->ts_prefixdepth < MAXWLEN)
|
|
375
|
8505 {
|
|
|
8506 /* There was a prefix before the word. Check that the
|
|
|
8507 * prefix can be used with this word. */
|
|
|
8508 /* Count the length of the NULs in the prefix. If there
|
|
|
8509 * are none this must be the first try without a prefix.
|
|
|
8510 */
|
|
|
8511 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
8512 len = pbyts[n++];
|
|
|
8513 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
8514 ;
|
|
|
8515 if (c > 0)
|
|
|
8516 {
|
|
390
|
8517 c = valid_word_prefix(c, n, flags,
|
|
483
|
8518 tword + sp->ts_splitoff, slang, FALSE);
|
|
375
|
8519 if (c == 0)
|
|
|
8520 break;
|
|
|
8521
|
|
|
8522 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
8523 if (c & WF_RAREPFX)
|
|
|
8524 flags |= WF_RARE;
|
|
|
8525 }
|
|
|
8526 }
|
|
|
8527
|
|
323
|
8528 /*
|
|
|
8529 * Form the word with proper case in preword.
|
|
|
8530 * If there is a word from a previous split, append.
|
|
|
8531 */
|
|
|
8532 if (flags & WF_KEEPCAP)
|
|
|
8533 /* Must find the word in the keep-case tree. */
|
|
483
|
8534 find_keepcap_word(slang, tword + sp->ts_splitoff,
|
|
|
8535 preword + sp->ts_prewordlen);
|
|
323
|
8536 else
|
|
346
|
8537 {
|
|
323
|
8538 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
8539 * use that for the goodword too. But if the badword is
|
|
|
8540 * allcap and it's only one char long use onecap. */
|
|
|
8541 c = su->su_badflags;
|
|
|
8542 if ((c & WF_ALLCAP)
|
|
|
8543 #ifdef FEAT_MBYTE
|
|
474
|
8544 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
8545 #else
|
|
|
8546 && su->su_badlen == 1
|
|
|
8547 #endif
|
|
|
8548 )
|
|
|
8549 c = WF_ONECAP;
|
|
483
|
8550 make_case_word(tword + sp->ts_splitoff,
|
|
|
8551 preword + sp->ts_prewordlen, flags | c);
|
|
346
|
8552 }
|
|
323
|
8553
|
|
|
8554 /* Don't use a banned word. It may appear again as a good
|
|
|
8555 * word, thus remember it. */
|
|
|
8556 if (flags & WF_BANNED)
|
|
|
8557 {
|
|
483
|
8558 add_banned(su, preword + sp->ts_prewordlen);
|
|
323
|
8559 break;
|
|
|
8560 }
|
|
483
|
8561 if (was_banned(su, preword + sp->ts_prewordlen)
|
|
372
|
8562 || was_banned(su, preword))
|
|
323
|
8563 break;
|
|
|
8564
|
|
|
8565 newscore = 0;
|
|
|
8566 if ((flags & WF_REGION)
|
|
390
|
8567 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
8568 newscore += SCORE_REGION;
|
|
|
8569 if (flags & WF_RARE)
|
|
|
8570 newscore += SCORE_RARE;
|
|
|
8571
|
|
346
|
8572 if (!spell_valid_case(su->su_badflags,
|
|
483
|
8573 captype(preword + sp->ts_prewordlen, NULL)))
|
|
323
|
8574 newscore += SCORE_ICASE;
|
|
|
8575
|
|
482
|
8576 if (fword_ends && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
8577 {
|
|
366
|
8578 /* The badword also ends: add suggestions. Give a penalty
|
|
|
8579 * when changing non-word char to word char, e.g., "thes,"
|
|
|
8580 * -> "these". */
|
|
|
8581 p = fword + sp->ts_fidx;
|
|
|
8582 #ifdef FEAT_MBYTE
|
|
|
8583 if (has_mbyte)
|
|
|
8584 mb_ptr_back(fword, p);
|
|
|
8585 else
|
|
|
8586 #endif
|
|
|
8587 --p;
|
|
376
|
8588 if (!spell_iswordp(p, curbuf))
|
|
366
|
8589 {
|
|
|
8590 p = preword + STRLEN(preword);
|
|
|
8591 #ifdef FEAT_MBYTE
|
|
|
8592 if (has_mbyte)
|
|
|
8593 mb_ptr_back(preword, p);
|
|
|
8594 else
|
|
|
8595 #endif
|
|
|
8596 --p;
|
|
376
|
8597 if (spell_iswordp(p, curbuf))
|
|
366
|
8598 newscore += SCORE_NONWORD;
|
|
|
8599 }
|
|
|
8600
|
|
344
|
8601 add_suggestion(su, &su->su_ga, preword,
|
|
346
|
8602 sp->ts_fidx - repextra,
|
|
351
|
8603 sp->ts_score + newscore, 0, FALSE);
|
|
323
|
8604 }
|
|
482
|
8605 else if ((sp->ts_fidx >= sp->ts_fidxtry || fword_ends)
|
|
330
|
8606 #ifdef FEAT_MBYTE
|
|
|
8607 /* Don't split halfway a character. */
|
|
|
8608 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
8609 #endif
|
|
|
8610 )
|
|
323
|
8611 {
|
|
482
|
8612 int try_compound;
|
|
|
8613
|
|
|
8614 /* Get here in two situations:
|
|
|
8615 * 1. The word in the tree ends but the badword continues:
|
|
|
8616 * If the word allows compounding try that. Otherwise
|
|
|
8617 * try a split by inserting a space. For both check
|
|
|
8618 * that a valid words starts at fword[sp->ts_fidx].
|
|
|
8619 * 2. The badword does end, but it was due to a change
|
|
|
8620 * (e.g., a swap). No need to split, but do check that
|
|
|
8621 * the following word is valid.
|
|
|
8622 */
|
|
|
8623 if (!fword_ends
|
|
483
|
8624 && ((unsigned)flags >> 24) != 0
|
|
|
8625 && sp->ts_twordlen - sp->ts_splitoff
|
|
482
|
8626 >= slang->sl_compminlen
|
|
483
|
8627 && sp->ts_complen + 1 <= slang->sl_compmax
|
|
|
8628 && (sp->ts_complen > 0
|
|
|
8629 || vim_strchr(slang->sl_compstartflags,
|
|
|
8630 ((unsigned)flags >> 24)) != NULL))
|
|
|
8631 {
|
|
482
|
8632 try_compound = TRUE;
|
|
483
|
8633 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
8634 compflags[sp->ts_complen + 1] = NUL;
|
|
|
8635 }
|
|
482
|
8636 else
|
|
|
8637 {
|
|
|
8638 try_compound = FALSE;
|
|
|
8639 if (!fword_ends)
|
|
|
8640 newscore += SCORE_SPLIT;
|
|
|
8641 }
|
|
|
8642
|
|
|
8643 if (try_deeper(su, stack, depth, newscore))
|
|
323
|
8644 {
|
|
|
8645 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
346
|
8646 sp->ts_save_badflags = su->su_badflags;
|
|
376
|
8647 sp->ts_state = STATE_SPLITUNDO;
|
|
|
8648
|
|
|
8649 ++depth;
|
|
|
8650 sp = &stack[depth];
|
|
323
|
8651
|
|
482
|
8652 /* Append a space to preword when splitting. */
|
|
|
8653 if (!try_compound && !fword_ends)
|
|
|
8654 STRCAT(preword, " ");
|
|
483
|
8655 sp->ts_prewordlen = STRLEN(preword);
|
|
|
8656 sp->ts_splitoff = sp->ts_twordlen;
|
|
376
|
8657
|
|
|
8658 /* If the badword has a non-word character at this
|
|
|
8659 * position skip it. That means replacing the
|
|
482
|
8660 * non-word character with a space. Always skip a
|
|
|
8661 * character when the word ends. */
|
|
|
8662 if ((!try_compound
|
|
|
8663 && !spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
|
8664 || fword_ends)
|
|
376
|
8665 {
|
|
482
|
8666 int l;
|
|
|
8667
|
|
376
|
8668 #ifdef FEAT_MBYTE
|
|
|
8669 if (has_mbyte)
|
|
482
|
8670 l = MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
376
|
8671 else
|
|
|
8672 #endif
|
|
482
|
8673 l = 1;
|
|
|
8674 if (fword_ends)
|
|
|
8675 {
|
|
|
8676 /* Copy the skipped character to preword. */
|
|
483
|
8677 mch_memmove(preword + sp->ts_prewordlen,
|
|
482
|
8678 fword + sp->ts_fidx, l);
|
|
483
|
8679 sp->ts_prewordlen += l;
|
|
|
8680 preword[sp->ts_prewordlen] = NUL;
|
|
482
|
8681 }
|
|
|
8682 else
|
|
|
8683 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
8684 sp->ts_fidx += l;
|
|
376
|
8685 }
|
|
455
|
8686
|
|
482
|
8687 /* set flag to check compound flag on following word */
|
|
|
8688 if (try_compound)
|
|
483
|
8689 {
|
|
482
|
8690 sp->ts_prefixdepth = PFD_COMPOUND;
|
|
483
|
8691 ++sp->ts_complen;
|
|
|
8692 }
|
|
482
|
8693 else
|
|
|
8694 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
|
8695
|
|
455
|
8696 /* set su->su_badflags to the caps type at this
|
|
|
8697 * position */
|
|
324
|
8698 #ifdef FEAT_MBYTE
|
|
|
8699 if (has_mbyte)
|
|
455
|
8700 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
8701 else
|
|
|
8702 #endif
|
|
455
|
8703 n = sp->ts_fidx;
|
|
474
|
8704 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
8705 su->su_badptr + su->su_badlen);
|
|
323
|
8706
|
|
|
8707 /* Restart at top of the tree. */
|
|
376
|
8708 sp->ts_arridx = 0;
|
|
323
|
8709 }
|
|
|
8710 }
|
|
|
8711 break;
|
|
|
8712
|
|
|
8713 case STATE_SPLITUNDO:
|
|
482
|
8714 /* Undo the changes done for word split or compound word. */
|
|
346
|
8715 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
8716
|
|
|
8717 /* Continue looking for NUL bytes. */
|
|
|
8718 sp->ts_state = STATE_START;
|
|
|
8719 break;
|
|
|
8720
|
|
|
8721 case STATE_ENDNUL:
|
|
|
8722 /* Past the NUL bytes in the node. */
|
|
455
|
8723 su->su_badflags = sp->ts_save_badflags;
|
|
346
|
8724 if (fword[sp->ts_fidx] == NUL)
|
|
323
|
8725 {
|
|
|
8726 /* The badword ends, can't use the bytes in this node. */
|
|
|
8727 sp->ts_state = STATE_DEL;
|
|
|
8728 break;
|
|
|
8729 }
|
|
|
8730 sp->ts_state = STATE_PLAIN;
|
|
|
8731 /*FALLTHROUGH*/
|
|
|
8732
|
|
|
8733 case STATE_PLAIN:
|
|
|
8734 /*
|
|
|
8735 * Go over all possible bytes at this node, add each to
|
|
|
8736 * tword[] and use child node. "ts_curi" is the index.
|
|
|
8737 */
|
|
|
8738 arridx = sp->ts_arridx;
|
|
|
8739 if (sp->ts_curi > byts[arridx])
|
|
|
8740 {
|
|
|
8741 /* Done all bytes at this node, do next state. When still
|
|
|
8742 * at already changed bytes skip the other tricks. */
|
|
|
8743 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
8744 sp->ts_state = STATE_DEL;
|
|
|
8745 else
|
|
|
8746 sp->ts_state = STATE_FINAL;
|
|
|
8747 }
|
|
|
8748 else
|
|
|
8749 {
|
|
|
8750 arridx += sp->ts_curi++;
|
|
|
8751 c = byts[arridx];
|
|
|
8752
|
|
|
8753 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
8754 * the byte in the bad word adjust the score. But don't
|
|
|
8755 * even try when the byte was already changed. */
|
|
330
|
8756 if (c == fword[sp->ts_fidx]
|
|
|
8757 #ifdef FEAT_MBYTE
|
|
|
8758 || (sp->ts_tcharlen > 0
|
|
|
8759 && sp->ts_isdiff != DIFF_NONE)
|
|
|
8760 #endif
|
|
|
8761 )
|
|
323
|
8762 newscore = 0;
|
|
|
8763 else
|
|
|
8764 newscore = SCORE_SUBST;
|
|
|
8765 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
8766 && try_deeper(su, stack, depth, newscore))
|
|
|
8767 {
|
|
|
8768 ++depth;
|
|
330
|
8769 sp = &stack[depth];
|
|
|
8770 ++sp->ts_fidx;
|
|
|
8771 tword[sp->ts_twordlen++] = c;
|
|
|
8772 sp->ts_arridx = idxs[arridx];
|
|
|
8773 #ifdef FEAT_MBYTE
|
|
|
8774 if (newscore == SCORE_SUBST)
|
|
|
8775 sp->ts_isdiff = DIFF_YES;
|
|
|
8776 if (has_mbyte)
|
|
|
8777 {
|
|
|
8778 /* Multi-byte characters are a bit complicated to
|
|
|
8779 * handle: They differ when any of the bytes
|
|
|
8780 * differ and then their length may also differ. */
|
|
|
8781 if (sp->ts_tcharlen == 0)
|
|
|
8782 {
|
|
|
8783 /* First byte. */
|
|
|
8784 sp->ts_tcharidx = 0;
|
|
|
8785 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
8786 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
8787 sp->ts_isdiff = (newscore != 0)
|
|
|
8788 ? DIFF_YES : DIFF_NONE;
|
|
|
8789 }
|
|
|
8790 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
8791 /* When inserting trail bytes don't advance in
|
|
|
8792 * the bad word. */
|
|
|
8793 --sp->ts_fidx;
|
|
|
8794 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
8795 {
|
|
|
8796 /* Last byte of character. */
|
|
|
8797 if (sp->ts_isdiff == DIFF_YES)
|
|
|
8798 {
|
|
|
8799 /* Correct ts_fidx for the byte length of
|
|
|
8800 * the character (we didn't check that
|
|
|
8801 * before). */
|
|
|
8802 sp->ts_fidx = sp->ts_fcharstart
|
|
|
8803 + MB_BYTE2LEN(
|
|
|
8804 fword[sp->ts_fcharstart]);
|
|
|
8805
|
|
479
|
8806 /* For changing a composing character
|
|
|
8807 * adjust the score from SCORE_SUBST to
|
|
|
8808 * SCORE_SUBCOMP. */
|
|
|
8809 if (enc_utf8
|
|
|
8810 && utf_iscomposing(
|
|
|
8811 mb_ptr2char(tword
|
|
|
8812 + sp->ts_twordlen
|
|
|
8813 - sp->ts_tcharlen))
|
|
|
8814 && utf_iscomposing(
|
|
|
8815 mb_ptr2char(fword
|
|
|
8816 + sp->ts_fcharstart)))
|
|
|
8817 sp->ts_score -=
|
|
|
8818 SCORE_SUBST - SCORE_SUBCOMP;
|
|
|
8819
|
|
330
|
8820 /* For a similar character adjust score
|
|
|
8821 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
482
|
8822 else if (slang->sl_has_map
|
|
|
8823 && similar_chars(slang,
|
|
330
|
8824 mb_ptr2char(tword
|
|
|
8825 + sp->ts_twordlen
|
|
|
8826 - sp->ts_tcharlen),
|
|
|
8827 mb_ptr2char(fword
|
|
|
8828 + sp->ts_fcharstart)))
|
|
|
8829 sp->ts_score -=
|
|
|
8830 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
8831 }
|
|
358
|
8832 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
8833 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
8834 {
|
|
|
8835 p = tword + sp->ts_twordlen
|
|
|
8836 - sp->ts_tcharlen;
|
|
|
8837 c = mb_ptr2char(p);
|
|
475
|
8838 if (enc_utf8 && utf_iscomposing(c))
|
|
|
8839 {
|
|
|
8840 /* Inserting a composing char doesn't
|
|
|
8841 * count that much. */
|
|
358
|
8842 sp->ts_score -= SCORE_INS
|
|
475
|
8843 - SCORE_INSCOMP;
|
|
|
8844 }
|
|
|
8845 else
|
|
|
8846 {
|
|
|
8847 /* If the previous character was the
|
|
|
8848 * same, thus doubling a character,
|
|
|
8849 * give a bonus to the score. */
|
|
|
8850 mb_ptr_back(tword, p);
|
|
|
8851 if (c == mb_ptr2char(p))
|
|
|
8852 sp->ts_score -= SCORE_INS
|
|
358
|
8853 - SCORE_INSDUP;
|
|
475
|
8854 }
|
|
358
|
8855 }
|
|
330
|
8856
|
|
|
8857 /* Starting a new char, reset the length. */
|
|
|
8858 sp->ts_tcharlen = 0;
|
|
|
8859 }
|
|
|
8860 }
|
|
|
8861 else
|
|
|
8862 #endif
|
|
|
8863 {
|
|
|
8864 /* If we found a similar char adjust the score.
|
|
|
8865 * We do this after calling try_deeper() because
|
|
|
8866 * it's slow. */
|
|
|
8867 if (newscore != 0
|
|
482
|
8868 && slang->sl_has_map
|
|
|
8869 && similar_chars(slang,
|
|
330
|
8870 c, fword[sp->ts_fidx - 1]))
|
|
|
8871 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
8872 }
|
|
323
|
8873 }
|
|
|
8874 }
|
|
|
8875 break;
|
|
|
8876
|
|
|
8877 case STATE_DEL:
|
|
330
|
8878 #ifdef FEAT_MBYTE
|
|
|
8879 /* When past the first byte of a multi-byte char don't try
|
|
|
8880 * delete/insert/swap a character. */
|
|
|
8881 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
8882 {
|
|
|
8883 sp->ts_state = STATE_FINAL;
|
|
|
8884 break;
|
|
|
8885 }
|
|
|
8886 #endif
|
|
|
8887 /*
|
|
|
8888 * Try skipping one character in the bad word (delete it).
|
|
|
8889 */
|
|
323
|
8890 sp->ts_state = STATE_INS;
|
|
|
8891 sp->ts_curi = 1;
|
|
|
8892 if (fword[sp->ts_fidx] != NUL
|
|
|
8893 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
8894 {
|
|
|
8895 ++depth;
|
|
358
|
8896
|
|
|
8897 /* Advance over the character in fword[]. Give a bonus to
|
|
|
8898 * the score if the same character is following "nn" ->
|
|
|
8899 * "n". */
|
|
330
|
8900 #ifdef FEAT_MBYTE
|
|
|
8901 if (has_mbyte)
|
|
358
|
8902 {
|
|
|
8903 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
8904 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
479
|
8905 if (enc_utf8 && utf_iscomposing(c))
|
|
|
8906 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
|
|
|
8907 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
358
|
8908 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
8909 }
|
|
330
|
8910 else
|
|
|
8911 #endif
|
|
358
|
8912 {
|
|
330
|
8913 ++stack[depth].ts_fidx;
|
|
358
|
8914 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
8915 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
8916 }
|
|
323
|
8917 break;
|
|
|
8918 }
|
|
|
8919 /*FALLTHROUGH*/
|
|
|
8920
|
|
|
8921 case STATE_INS:
|
|
330
|
8922 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
8923 * node. */
|
|
|
8924 n = sp->ts_arridx;
|
|
|
8925 if (sp->ts_curi > byts[n])
|
|
|
8926 {
|
|
|
8927 /* Done all bytes at this node, do next state. */
|
|
|
8928 sp->ts_state = STATE_SWAP;
|
|
|
8929 }
|
|
|
8930 else
|
|
|
8931 {
|
|
330
|
8932 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
8933 n += sp->ts_curi++;
|
|
|
8934 c = byts[n];
|
|
|
8935 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
8936 {
|
|
|
8937 ++depth;
|
|
330
|
8938 sp = &stack[depth];
|
|
|
8939 tword[sp->ts_twordlen++] = c;
|
|
|
8940 sp->ts_arridx = idxs[n];
|
|
|
8941 #ifdef FEAT_MBYTE
|
|
|
8942 if (has_mbyte)
|
|
|
8943 {
|
|
|
8944 fl = MB_BYTE2LEN(c);
|
|
|
8945 if (fl > 1)
|
|
|
8946 {
|
|
|
8947 /* There are following bytes for the same
|
|
|
8948 * character. We must find all bytes before
|
|
|
8949 * trying delete/insert/swap/etc. */
|
|
|
8950 sp->ts_tcharlen = fl;
|
|
|
8951 sp->ts_tcharidx = 1;
|
|
|
8952 sp->ts_isdiff = DIFF_INSERT;
|
|
|
8953 }
|
|
|
8954 }
|
|
358
|
8955 else
|
|
|
8956 fl = 1;
|
|
|
8957 if (fl == 1)
|
|
330
|
8958 #endif
|
|
358
|
8959 {
|
|
|
8960 /* If the previous character was the same, thus
|
|
|
8961 * doubling a character, give a bonus to the
|
|
|
8962 * score. */
|
|
|
8963 if (sp->ts_twordlen >= 2
|
|
|
8964 && tword[sp->ts_twordlen - 2] == c)
|
|
|
8965 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
8966 }
|
|
323
|
8967 }
|
|
|
8968 }
|
|
|
8969 break;
|
|
|
8970
|
|
|
8971 case STATE_SWAP:
|
|
330
|
8972 /*
|
|
|
8973 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
8974 * We change "fword" here, it's changed back afterwards.
|
|
|
8975 */
|
|
|
8976 p = fword + sp->ts_fidx;
|
|
|
8977 c = *p;
|
|
|
8978 if (c == NUL)
|
|
|
8979 {
|
|
|
8980 /* End of word, can't swap or replace. */
|
|
|
8981 sp->ts_state = STATE_FINAL;
|
|
|
8982 break;
|
|
|
8983 }
|
|
|
8984 #ifdef FEAT_MBYTE
|
|
|
8985 if (has_mbyte)
|
|
|
8986 {
|
|
474
|
8987 n = mb_cptr2len(p);
|
|
330
|
8988 c = mb_ptr2char(p);
|
|
|
8989 c2 = mb_ptr2char(p + n);
|
|
|
8990 }
|
|
|
8991 else
|
|
|
8992 #endif
|
|
|
8993 c2 = p[1];
|
|
|
8994 if (c == c2)
|
|
323
|
8995 {
|
|
330
|
8996 /* Characters are identical, swap won't do anything. */
|
|
|
8997 sp->ts_state = STATE_SWAP3;
|
|
|
8998 break;
|
|
|
8999 }
|
|
|
9000 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
9001 {
|
|
|
9002 sp->ts_state = STATE_UNSWAP;
|
|
323
|
9003 ++depth;
|
|
330
|
9004 #ifdef FEAT_MBYTE
|
|
|
9005 if (has_mbyte)
|
|
|
9006 {
|
|
|
9007 fl = mb_char2len(c2);
|
|
|
9008 mch_memmove(p, p + n, fl);
|
|
|
9009 mb_char2bytes(c, p + fl);
|
|
|
9010 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
9011 }
|
|
|
9012 else
|
|
|
9013 #endif
|
|
|
9014 {
|
|
|
9015 p[0] = c2;
|
|
|
9016 p[1] = c;
|
|
|
9017 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
9018 }
|
|
323
|
9019 }
|
|
|
9020 else
|
|
|
9021 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
9022 sp->ts_state = STATE_REP_INI;
|
|
|
9023 break;
|
|
|
9024
|
|
330
|
9025 case STATE_UNSWAP:
|
|
|
9026 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
9027 p = fword + sp->ts_fidx;
|
|
|
9028 #ifdef FEAT_MBYTE
|
|
|
9029 if (has_mbyte)
|
|
|
9030 {
|
|
|
9031 n = MB_BYTE2LEN(*p);
|
|
|
9032 c = mb_ptr2char(p + n);
|
|
|
9033 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
9034 mb_char2bytes(c, p);
|
|
|
9035 }
|
|
|
9036 else
|
|
|
9037 #endif
|
|
|
9038 {
|
|
|
9039 c = *p;
|
|
|
9040 *p = p[1];
|
|
|
9041 p[1] = c;
|
|
|
9042 }
|
|
|
9043 /*FALLTHROUGH*/
|
|
|
9044
|
|
|
9045 case STATE_SWAP3:
|
|
323
|
9046 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
9047 * "fword" here, it's changed back afterwards. */
|
|
|
9048 p = fword + sp->ts_fidx;
|
|
|
9049 #ifdef FEAT_MBYTE
|
|
|
9050 if (has_mbyte)
|
|
|
9051 {
|
|
474
|
9052 n = mb_cptr2len(p);
|
|
330
|
9053 c = mb_ptr2char(p);
|
|
474
|
9054 fl = mb_cptr2len(p + n);
|
|
330
|
9055 c2 = mb_ptr2char(p + n);
|
|
|
9056 c3 = mb_ptr2char(p + n + fl);
|
|
|
9057 }
|
|
|
9058 else
|
|
|
9059 #endif
|
|
323
|
9060 {
|
|
330
|
9061 c = *p;
|
|
|
9062 c2 = p[1];
|
|
|
9063 c3 = p[2];
|
|
|
9064 }
|
|
|
9065
|
|
|
9066 /* When characters are identical: "121" then SWAP3 result is
|
|
|
9067 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
9068 * result is same as SWAP on next char: "112". Thus skip all
|
|
|
9069 * swapping. Also skip when c3 is NUL. */
|
|
|
9070 if (c == c3 || c3 == NUL)
|
|
|
9071 {
|
|
|
9072 sp->ts_state = STATE_REP_INI;
|
|
|
9073 break;
|
|
|
9074 }
|
|
|
9075 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9076 {
|
|
|
9077 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
9078 ++depth;
|
|
330
|
9079 #ifdef FEAT_MBYTE
|
|
|
9080 if (has_mbyte)
|
|
|
9081 {
|
|
|
9082 tl = mb_char2len(c3);
|
|
|
9083 mch_memmove(p, p + n + fl, tl);
|
|
|
9084 mb_char2bytes(c2, p + tl);
|
|
|
9085 mb_char2bytes(c, p + fl + tl);
|
|
|
9086 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
9087 }
|
|
|
9088 else
|
|
|
9089 #endif
|
|
|
9090 {
|
|
|
9091 p[0] = p[2];
|
|
|
9092 p[2] = c;
|
|
|
9093 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9094 }
|
|
323
|
9095 }
|
|
|
9096 else
|
|
|
9097 sp->ts_state = STATE_REP_INI;
|
|
|
9098 break;
|
|
|
9099
|
|
330
|
9100 case STATE_UNSWAP3:
|
|
|
9101 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
9102 p = fword + sp->ts_fidx;
|
|
|
9103 #ifdef FEAT_MBYTE
|
|
|
9104 if (has_mbyte)
|
|
|
9105 {
|
|
|
9106 n = MB_BYTE2LEN(*p);
|
|
|
9107 c2 = mb_ptr2char(p + n);
|
|
|
9108 fl = MB_BYTE2LEN(p[n]);
|
|
|
9109 c = mb_ptr2char(p + n + fl);
|
|
|
9110 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
9111 mch_memmove(p + fl + tl, p, n);
|
|
|
9112 mb_char2bytes(c, p);
|
|
|
9113 mb_char2bytes(c2, p + tl);
|
|
|
9114 }
|
|
|
9115 else
|
|
|
9116 #endif
|
|
|
9117 {
|
|
|
9118 c = *p;
|
|
|
9119 *p = p[2];
|
|
|
9120 p[2] = c;
|
|
|
9121 }
|
|
346
|
9122
|
|
330
|
9123 /* Rotate three characters left: "123" -> "231". We change
|
|
|
9124 * "fword" here, it's changed back afterwards. */
|
|
323
|
9125 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9126 {
|
|
330
|
9127 sp->ts_state = STATE_UNROT3L;
|
|
323
|
9128 ++depth;
|
|
330
|
9129 p = fword + sp->ts_fidx;
|
|
|
9130 #ifdef FEAT_MBYTE
|
|
|
9131 if (has_mbyte)
|
|
|
9132 {
|
|
474
|
9133 n = mb_cptr2len(p);
|
|
330
|
9134 c = mb_ptr2char(p);
|
|
474
|
9135 fl = mb_cptr2len(p + n);
|
|
|
9136 fl += mb_cptr2len(p + n + fl);
|
|
330
|
9137 mch_memmove(p, p + n, fl);
|
|
|
9138 mb_char2bytes(c, p + fl);
|
|
|
9139 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
9140 }
|
|
|
9141 else
|
|
|
9142 #endif
|
|
|
9143 {
|
|
|
9144 c = *p;
|
|
|
9145 *p = p[1];
|
|
|
9146 p[1] = p[2];
|
|
|
9147 p[2] = c;
|
|
|
9148 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9149 }
|
|
323
|
9150 }
|
|
|
9151 else
|
|
|
9152 sp->ts_state = STATE_REP_INI;
|
|
|
9153 break;
|
|
|
9154
|
|
330
|
9155 case STATE_UNROT3L:
|
|
346
|
9156 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
9157 p = fword + sp->ts_fidx;
|
|
|
9158 #ifdef FEAT_MBYTE
|
|
|
9159 if (has_mbyte)
|
|
|
9160 {
|
|
|
9161 n = MB_BYTE2LEN(*p);
|
|
|
9162 n += MB_BYTE2LEN(p[n]);
|
|
|
9163 c = mb_ptr2char(p + n);
|
|
|
9164 tl = MB_BYTE2LEN(p[n]);
|
|
|
9165 mch_memmove(p + tl, p, n);
|
|
|
9166 mb_char2bytes(c, p);
|
|
|
9167 }
|
|
|
9168 else
|
|
|
9169 #endif
|
|
|
9170 {
|
|
|
9171 c = p[2];
|
|
|
9172 p[2] = p[1];
|
|
|
9173 p[1] = *p;
|
|
|
9174 *p = c;
|
|
|
9175 }
|
|
346
|
9176
|
|
323
|
9177 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
9178 * "fword" here, it's changed back afterwards. */
|
|
323
|
9179 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
9180 {
|
|
330
|
9181 sp->ts_state = STATE_UNROT3R;
|
|
323
|
9182 ++depth;
|
|
330
|
9183 p = fword + sp->ts_fidx;
|
|
|
9184 #ifdef FEAT_MBYTE
|
|
|
9185 if (has_mbyte)
|
|
|
9186 {
|
|
474
|
9187 n = mb_cptr2len(p);
|
|
|
9188 n += mb_cptr2len(p + n);
|
|
330
|
9189 c = mb_ptr2char(p + n);
|
|
474
|
9190 tl = mb_cptr2len(p + n);
|
|
330
|
9191 mch_memmove(p + tl, p, n);
|
|
|
9192 mb_char2bytes(c, p);
|
|
|
9193 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
9194 }
|
|
|
9195 else
|
|
|
9196 #endif
|
|
|
9197 {
|
|
|
9198 c = p[2];
|
|
|
9199 p[2] = p[1];
|
|
|
9200 p[1] = *p;
|
|
|
9201 *p = c;
|
|
|
9202 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
9203 }
|
|
323
|
9204 }
|
|
|
9205 else
|
|
|
9206 sp->ts_state = STATE_REP_INI;
|
|
|
9207 break;
|
|
|
9208
|
|
330
|
9209 case STATE_UNROT3R:
|
|
346
|
9210 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
9211 p = fword + sp->ts_fidx;
|
|
|
9212 #ifdef FEAT_MBYTE
|
|
|
9213 if (has_mbyte)
|
|
|
9214 {
|
|
|
9215 c = mb_ptr2char(p);
|
|
|
9216 tl = MB_BYTE2LEN(*p);
|
|
|
9217 n = MB_BYTE2LEN(p[tl]);
|
|
|
9218 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
9219 mch_memmove(p, p + tl, n);
|
|
|
9220 mb_char2bytes(c, p + n);
|
|
|
9221 }
|
|
|
9222 else
|
|
|
9223 #endif
|
|
|
9224 {
|
|
|
9225 c = *p;
|
|
|
9226 *p = p[1];
|
|
|
9227 p[1] = p[2];
|
|
|
9228 p[2] = c;
|
|
|
9229 }
|
|
323
|
9230 /*FALLTHROUGH*/
|
|
|
9231
|
|
|
9232 case STATE_REP_INI:
|
|
|
9233 /* Check if matching with REP items from the .aff file would
|
|
|
9234 * work. Quickly skip if there are no REP items or the score
|
|
|
9235 * is going to be too high anyway. */
|
|
482
|
9236 gap = &slang->sl_rep;
|
|
323
|
9237 if (gap->ga_len == 0
|
|
|
9238 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
|
9239 {
|
|
|
9240 sp->ts_state = STATE_FINAL;
|
|
|
9241 break;
|
|
|
9242 }
|
|
|
9243
|
|
|
9244 /* Use the first byte to quickly find the first entry that
|
|
330
|
9245 * may match. If the index is -1 there is none. */
|
|
482
|
9246 sp->ts_curi = slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
323
|
9247 if (sp->ts_curi < 0)
|
|
|
9248 {
|
|
|
9249 sp->ts_state = STATE_FINAL;
|
|
|
9250 break;
|
|
|
9251 }
|
|
|
9252
|
|
|
9253 sp->ts_state = STATE_REP;
|
|
|
9254 /*FALLTHROUGH*/
|
|
|
9255
|
|
|
9256 case STATE_REP:
|
|
|
9257 /* Try matching with REP items from the .aff file. For each
|
|
330
|
9258 * match replace the characters and check if the resulting
|
|
|
9259 * word is valid. */
|
|
323
|
9260 p = fword + sp->ts_fidx;
|
|
|
9261
|
|
482
|
9262 gap = &slang->sl_rep;
|
|
323
|
9263 while (sp->ts_curi < gap->ga_len)
|
|
|
9264 {
|
|
|
9265 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
9266 if (*ftp->ft_from != *p)
|
|
|
9267 {
|
|
|
9268 /* past possible matching entries */
|
|
|
9269 sp->ts_curi = gap->ga_len;
|
|
|
9270 break;
|
|
|
9271 }
|
|
|
9272 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
9273 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
9274 {
|
|
|
9275 /* Need to undo this afterwards. */
|
|
|
9276 sp->ts_state = STATE_REP_UNDO;
|
|
|
9277
|
|
|
9278 /* Change the "from" to the "to" string. */
|
|
|
9279 ++depth;
|
|
|
9280 fl = STRLEN(ftp->ft_from);
|
|
|
9281 tl = STRLEN(ftp->ft_to);
|
|
|
9282 if (fl != tl)
|
|
346
|
9283 {
|
|
323
|
9284 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
9285 repextra += tl - fl;
|
|
|
9286 }
|
|
323
|
9287 mch_memmove(p, ftp->ft_to, tl);
|
|
|
9288 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
9289 #ifdef FEAT_MBYTE
|
|
|
9290 stack[depth].ts_tcharlen = 0;
|
|
|
9291 #endif
|
|
323
|
9292 break;
|
|
|
9293 }
|
|
|
9294 }
|
|
|
9295
|
|
|
9296 if (sp->ts_curi >= gap->ga_len)
|
|
|
9297 /* No (more) matches. */
|
|
|
9298 sp->ts_state = STATE_FINAL;
|
|
|
9299
|
|
|
9300 break;
|
|
|
9301
|
|
|
9302 case STATE_REP_UNDO:
|
|
|
9303 /* Undo a REP replacement and continue with the next one. */
|
|
482
|
9304 ftp = (fromto_T *)slang->sl_rep.ga_data + sp->ts_curi - 1;
|
|
323
|
9305 fl = STRLEN(ftp->ft_from);
|
|
|
9306 tl = STRLEN(ftp->ft_to);
|
|
|
9307 p = fword + sp->ts_fidx;
|
|
|
9308 if (fl != tl)
|
|
346
|
9309 {
|
|
323
|
9310 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
9311 repextra -= tl - fl;
|
|
|
9312 }
|
|
323
|
9313 mch_memmove(p, ftp->ft_from, fl);
|
|
|
9314 sp->ts_state = STATE_REP;
|
|
|
9315 break;
|
|
|
9316
|
|
|
9317 default:
|
|
|
9318 /* Did all possible states at this level, go up one level. */
|
|
|
9319 --depth;
|
|
344
|
9320
|
|
482
|
9321 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
9322 {
|
|
|
9323 /* Continue in or go back to the prefix tree. */
|
|
|
9324 byts = pbyts;
|
|
|
9325 idxs = pidxs;
|
|
|
9326 }
|
|
|
9327
|
|
344
|
9328 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
9329 line_breakcheck();
|
|
323
|
9330 }
|
|
|
9331 }
|
|
|
9332 }
|
|
|
9333 }
|
|
|
9334
|
|
|
9335 /*
|
|
|
9336 * Try going one level deeper in the tree.
|
|
|
9337 */
|
|
|
9338 static int
|
|
|
9339 try_deeper(su, stack, depth, score_add)
|
|
|
9340 suginfo_T *su;
|
|
|
9341 trystate_T *stack;
|
|
|
9342 int depth;
|
|
|
9343 int score_add;
|
|
|
9344 {
|
|
|
9345 int newscore;
|
|
|
9346
|
|
|
9347 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
9348 newscore = stack[depth].ts_score + score_add;
|
|
|
9349 if (newscore >= su->su_maxscore)
|
|
|
9350 return FALSE;
|
|
|
9351
|
|
330
|
9352 stack[depth + 1] = stack[depth];
|
|
323
|
9353 stack[depth + 1].ts_state = STATE_START;
|
|
|
9354 stack[depth + 1].ts_score = newscore;
|
|
|
9355 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
|
9356 return TRUE;
|
|
|
9357 }
|
|
|
9358
|
|
455
|
9359 #ifdef FEAT_MBYTE
|
|
|
9360 /*
|
|
|
9361 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
9362 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
9363 */
|
|
|
9364 static int
|
|
|
9365 nofold_len(fword, flen, word)
|
|
|
9366 char_u *fword;
|
|
|
9367 int flen;
|
|
|
9368 char_u *word;
|
|
|
9369 {
|
|
|
9370 char_u *p;
|
|
|
9371 int i = 0;
|
|
|
9372
|
|
|
9373 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
9374 ++i;
|
|
|
9375 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
9376 --i;
|
|
|
9377 return (int)(p - word);
|
|
|
9378 }
|
|
|
9379 #endif
|
|
|
9380
|
|
323
|
9381 /*
|
|
|
9382 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
9383 * words and put it in "kword".
|
|
|
9384 * Theoretically there could be several keep-case words that result in the
|
|
|
9385 * same case-folded word, but we only find one...
|
|
|
9386 */
|
|
|
9387 static void
|
|
|
9388 find_keepcap_word(slang, fword, kword)
|
|
|
9389 slang_T *slang;
|
|
|
9390 char_u *fword;
|
|
|
9391 char_u *kword;
|
|
|
9392 {
|
|
|
9393 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
9394 int depth;
|
|
324
|
9395 idx_T tryidx;
|
|
323
|
9396
|
|
|
9397 /* The following arrays are used at each depth in the tree. */
|
|
324
|
9398 idx_T arridx[MAXWLEN];
|
|
323
|
9399 int round[MAXWLEN];
|
|
|
9400 int fwordidx[MAXWLEN];
|
|
|
9401 int uwordidx[MAXWLEN];
|
|
|
9402 int kwordlen[MAXWLEN];
|
|
|
9403
|
|
|
9404 int flen, ulen;
|
|
|
9405 int l;
|
|
|
9406 int len;
|
|
|
9407 int c;
|
|
324
|
9408 idx_T lo, hi, m;
|
|
323
|
9409 char_u *p;
|
|
|
9410 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
9411 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
9412
|
|
|
9413 if (byts == NULL)
|
|
|
9414 {
|
|
|
9415 /* array is empty: "cannot happen" */
|
|
|
9416 *kword = NUL;
|
|
|
9417 return;
|
|
|
9418 }
|
|
|
9419
|
|
|
9420 /* Make an all-cap version of "fword". */
|
|
|
9421 allcap_copy(fword, uword);
|
|
|
9422
|
|
|
9423 /*
|
|
|
9424 * Each character needs to be tried both case-folded and upper-case.
|
|
|
9425 * All this gets very complicated if we keep in mind that changing case
|
|
|
9426 * may change the byte length of a multi-byte character...
|
|
|
9427 */
|
|
|
9428 depth = 0;
|
|
|
9429 arridx[0] = 0;
|
|
|
9430 round[0] = 0;
|
|
|
9431 fwordidx[0] = 0;
|
|
|
9432 uwordidx[0] = 0;
|
|
|
9433 kwordlen[0] = 0;
|
|
|
9434 while (depth >= 0)
|
|
|
9435 {
|
|
|
9436 if (fword[fwordidx[depth]] == NUL)
|
|
|
9437 {
|
|
|
9438 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
9439 * here we have found a match. */
|
|
|
9440 if (byts[arridx[depth] + 1] == 0)
|
|
|
9441 {
|
|
|
9442 kword[kwordlen[depth]] = NUL;
|
|
|
9443 return;
|
|
|
9444 }
|
|
|
9445
|
|
|
9446 /* kword is getting too long, continue one level up */
|
|
|
9447 --depth;
|
|
|
9448 }
|
|
|
9449 else if (++round[depth] > 2)
|
|
|
9450 {
|
|
|
9451 /* tried both fold-case and upper-case character, continue one
|
|
|
9452 * level up */
|
|
|
9453 --depth;
|
|
|
9454 }
|
|
|
9455 else
|
|
|
9456 {
|
|
|
9457 /*
|
|
|
9458 * round[depth] == 1: Try using the folded-case character.
|
|
|
9459 * round[depth] == 2: Try using the upper-case character.
|
|
|
9460 */
|
|
|
9461 #ifdef FEAT_MBYTE
|
|
|
9462 if (has_mbyte)
|
|
|
9463 {
|
|
474
|
9464 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
9465 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
9466 }
|
|
|
9467 else
|
|
|
9468 #endif
|
|
|
9469 ulen = flen = 1;
|
|
|
9470 if (round[depth] == 1)
|
|
|
9471 {
|
|
|
9472 p = fword + fwordidx[depth];
|
|
|
9473 l = flen;
|
|
|
9474 }
|
|
|
9475 else
|
|
|
9476 {
|
|
|
9477 p = uword + uwordidx[depth];
|
|
|
9478 l = ulen;
|
|
|
9479 }
|
|
|
9480
|
|
|
9481 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
9482 {
|
|
|
9483 /* Perform a binary search in the list of accepted bytes. */
|
|
|
9484 len = byts[tryidx++];
|
|
|
9485 c = *p++;
|
|
|
9486 lo = tryidx;
|
|
|
9487 hi = tryidx + len - 1;
|
|
|
9488 while (lo < hi)
|
|
|
9489 {
|
|
|
9490 m = (lo + hi) / 2;
|
|
|
9491 if (byts[m] > c)
|
|
|
9492 hi = m - 1;
|
|
|
9493 else if (byts[m] < c)
|
|
|
9494 lo = m + 1;
|
|
|
9495 else
|
|
|
9496 {
|
|
|
9497 lo = hi = m;
|
|
|
9498 break;
|
|
|
9499 }
|
|
|
9500 }
|
|
|
9501
|
|
|
9502 /* Stop if there is no matching byte. */
|
|
|
9503 if (hi < lo || byts[lo] != c)
|
|
|
9504 break;
|
|
|
9505
|
|
|
9506 /* Continue at the child (if there is one). */
|
|
|
9507 tryidx = idxs[lo];
|
|
|
9508 }
|
|
|
9509
|
|
|
9510 if (l == 0)
|
|
|
9511 {
|
|
|
9512 /*
|
|
|
9513 * Found the matching char. Copy it to "kword" and go a
|
|
|
9514 * level deeper.
|
|
|
9515 */
|
|
|
9516 if (round[depth] == 1)
|
|
|
9517 {
|
|
|
9518 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
9519 flen);
|
|
|
9520 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
9521 }
|
|
|
9522 else
|
|
|
9523 {
|
|
|
9524 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
9525 ulen);
|
|
|
9526 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
9527 }
|
|
|
9528 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
9529 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
9530
|
|
|
9531 ++depth;
|
|
|
9532 arridx[depth] = tryidx;
|
|
|
9533 round[depth] = 0;
|
|
|
9534 }
|
|
|
9535 }
|
|
|
9536 }
|
|
|
9537
|
|
|
9538 /* Didn't find it: "cannot happen". */
|
|
|
9539 *kword = NUL;
|
|
|
9540 }
|
|
|
9541
|
|
|
9542 /*
|
|
344
|
9543 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
9544 * su->su_sga.
|
|
|
9545 */
|
|
|
9546 static void
|
|
|
9547 score_comp_sal(su)
|
|
|
9548 suginfo_T *su;
|
|
|
9549 {
|
|
|
9550 langp_T *lp;
|
|
|
9551 char_u badsound[MAXWLEN];
|
|
|
9552 int i;
|
|
|
9553 suggest_T *stp;
|
|
|
9554 suggest_T *sstp;
|
|
|
9555 int score;
|
|
|
9556
|
|
|
9557 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
9558 return;
|
|
|
9559
|
|
|
9560 /* Use the sound-folding of the first language that supports it. */
|
|
|
9561 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9562 lp->lp_slang != NULL; ++lp)
|
|
|
9563 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9564 {
|
|
|
9565 /* soundfold the bad word */
|
|
375
|
9566 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
9567
|
|
|
9568 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9569 {
|
|
|
9570 stp = &SUG(su->su_ga, i);
|
|
|
9571
|
|
351
|
9572 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
9573 * sound-a-like score. */
|
|
|
9574 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
9575 if (score < SCORE_MAXMAX)
|
|
|
9576 {
|
|
|
9577 /* Add the suggestion. */
|
|
|
9578 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
9579 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
9580 if (sstp->st_word != NULL)
|
|
|
9581 {
|
|
|
9582 sstp->st_score = score;
|
|
|
9583 sstp->st_altscore = 0;
|
|
|
9584 sstp->st_orglen = stp->st_orglen;
|
|
|
9585 ++su->su_sga.ga_len;
|
|
|
9586 }
|
|
|
9587 }
|
|
|
9588 }
|
|
|
9589 break;
|
|
|
9590 }
|
|
|
9591 }
|
|
|
9592
|
|
|
9593 /*
|
|
|
9594 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
9595 * They are intwined.
|
|
|
9596 */
|
|
|
9597 static void
|
|
|
9598 score_combine(su)
|
|
|
9599 suginfo_T *su;
|
|
|
9600 {
|
|
|
9601 int i;
|
|
|
9602 int j;
|
|
|
9603 garray_T ga;
|
|
|
9604 garray_T *gap;
|
|
|
9605 langp_T *lp;
|
|
|
9606 suggest_T *stp;
|
|
|
9607 char_u *p;
|
|
|
9608 char_u badsound[MAXWLEN];
|
|
|
9609 int round;
|
|
|
9610
|
|
|
9611 /* Add the alternate score to su_ga. */
|
|
|
9612 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9613 lp->lp_slang != NULL; ++lp)
|
|
|
9614 {
|
|
|
9615 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9616 {
|
|
|
9617 /* soundfold the bad word */
|
|
375
|
9618 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
9619
|
|
|
9620 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9621 {
|
|
|
9622 stp = &SUG(su->su_ga, i);
|
|
351
|
9623 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
9624 badsound);
|
|
344
|
9625 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
9626 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
9627 else
|
|
|
9628 stp->st_score = (stp->st_score * 3
|
|
|
9629 + stp->st_altscore) / 4;
|
|
|
9630 stp->st_salscore = FALSE;
|
|
|
9631 }
|
|
|
9632 break;
|
|
|
9633 }
|
|
|
9634 }
|
|
|
9635
|
|
|
9636 /* Add the alternate score to su_sga. */
|
|
|
9637 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
9638 {
|
|
|
9639 stp = &SUG(su->su_sga, i);
|
|
|
9640 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
9641 if (stp->st_score == SCORE_MAXMAX)
|
|
|
9642 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
9643 else
|
|
|
9644 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
9645 stp->st_salscore = TRUE;
|
|
|
9646 }
|
|
|
9647
|
|
|
9648 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
9649 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
9650 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
9651
|
|
|
9652 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
9653 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
9654 return;
|
|
|
9655
|
|
|
9656 stp = &SUG(ga, 0);
|
|
|
9657 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
9658 {
|
|
|
9659 /* round 1: get a suggestion from su_ga
|
|
|
9660 * round 2: get a suggestion from su_sga */
|
|
|
9661 for (round = 1; round <= 2; ++round)
|
|
|
9662 {
|
|
|
9663 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
9664 if (i < gap->ga_len)
|
|
|
9665 {
|
|
|
9666 /* Don't add a word if it's already there. */
|
|
|
9667 p = SUG(*gap, i).st_word;
|
|
|
9668 for (j = 0; j < ga.ga_len; ++j)
|
|
|
9669 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
9670 break;
|
|
|
9671 if (j == ga.ga_len)
|
|
|
9672 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
9673 else
|
|
|
9674 vim_free(p);
|
|
|
9675 }
|
|
|
9676 }
|
|
|
9677 }
|
|
|
9678
|
|
|
9679 ga_clear(&su->su_ga);
|
|
|
9680 ga_clear(&su->su_sga);
|
|
|
9681
|
|
|
9682 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
9683 if (ga.ga_len > su->su_maxcount)
|
|
|
9684 {
|
|
|
9685 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
9686 vim_free(stp[i].st_word);
|
|
|
9687 ga.ga_len = su->su_maxcount;
|
|
|
9688 }
|
|
|
9689
|
|
|
9690 su->su_ga = ga;
|
|
|
9691 }
|
|
|
9692
|
|
|
9693 /*
|
|
351
|
9694 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
9695 * badword.
|
|
|
9696 */
|
|
|
9697 static int
|
|
|
9698 stp_sal_score(stp, su, slang, badsound)
|
|
|
9699 suggest_T *stp;
|
|
|
9700 suginfo_T *su;
|
|
|
9701 slang_T *slang;
|
|
|
9702 char_u *badsound; /* sound-folded badword */
|
|
|
9703 {
|
|
|
9704 char_u *p;
|
|
|
9705 char_u badsound2[MAXWLEN];
|
|
|
9706 char_u fword[MAXWLEN];
|
|
|
9707 char_u goodsound[MAXWLEN];
|
|
|
9708
|
|
|
9709 if (stp->st_orglen <= su->su_badlen)
|
|
|
9710 p = badsound;
|
|
|
9711 else
|
|
|
9712 {
|
|
|
9713 /* soundfold the bad word with more characters following */
|
|
|
9714 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
9715
|
|
|
9716 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
9717 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
9718 * removing the space. Don't do it when the good word also contains a
|
|
|
9719 * space. */
|
|
|
9720 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
9721 && *skiptowhite(stp->st_word) == NUL)
|
|
|
9722 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
9723 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
9724
|
|
375
|
9725 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
351
|
9726 p = badsound2;
|
|
|
9727 }
|
|
|
9728
|
|
375
|
9729 /* Sound-fold the word and compute the score for the difference. */
|
|
|
9730 spell_soundfold(slang, stp->st_word, FALSE, goodsound);
|
|
351
|
9731
|
|
|
9732 return soundalike_score(goodsound, p);
|
|
|
9733 }
|
|
|
9734
|
|
|
9735 /*
|
|
323
|
9736 * Find suggestions by comparing the word in a sound-a-like form.
|
|
|
9737 */
|
|
|
9738 static void
|
|
346
|
9739 suggest_try_soundalike(su)
|
|
323
|
9740 suginfo_T *su;
|
|
|
9741 {
|
|
|
9742 char_u salword[MAXWLEN];
|
|
|
9743 char_u tword[MAXWLEN];
|
|
|
9744 char_u tsalword[MAXWLEN];
|
|
324
|
9745 idx_T arridx[MAXWLEN];
|
|
323
|
9746 int curi[MAXWLEN];
|
|
|
9747 langp_T *lp;
|
|
|
9748 char_u *byts;
|
|
324
|
9749 idx_T *idxs;
|
|
323
|
9750 int depth;
|
|
|
9751 int c;
|
|
324
|
9752 idx_T n;
|
|
323
|
9753 int round;
|
|
|
9754 int flags;
|
|
344
|
9755 int sound_score;
|
|
483
|
9756 int local_score;
|
|
344
|
9757
|
|
|
9758 /* Do this for all languages that support sound folding. */
|
|
323
|
9759 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
9760 lp->lp_slang != NULL; ++lp)
|
|
|
9761 {
|
|
|
9762 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
9763 {
|
|
|
9764 /* soundfold the bad word */
|
|
375
|
9765 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, salword);
|
|
323
|
9766
|
|
|
9767 /*
|
|
|
9768 * Go through the whole tree, soundfold each word and compare.
|
|
|
9769 * round 1: use the case-folded tree.
|
|
|
9770 * round 2: use the keep-case tree.
|
|
|
9771 */
|
|
|
9772 for (round = 1; round <= 2; ++round)
|
|
|
9773 {
|
|
|
9774 if (round == 1)
|
|
|
9775 {
|
|
|
9776 byts = lp->lp_slang->sl_fbyts;
|
|
|
9777 idxs = lp->lp_slang->sl_fidxs;
|
|
|
9778 }
|
|
|
9779 else
|
|
|
9780 {
|
|
|
9781 byts = lp->lp_slang->sl_kbyts;
|
|
|
9782 idxs = lp->lp_slang->sl_kidxs;
|
|
389
|
9783 if (byts == NULL) /* no keep-case words */
|
|
|
9784 continue;
|
|
323
|
9785 }
|
|
|
9786
|
|
|
9787 depth = 0;
|
|
|
9788 arridx[0] = 0;
|
|
|
9789 curi[0] = 1;
|
|
|
9790 while (depth >= 0 && !got_int)
|
|
|
9791 {
|
|
|
9792 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
9793 {
|
|
323
|
9794 /* Done all bytes at this node, go up one level. */
|
|
|
9795 --depth;
|
|
351
|
9796 line_breakcheck();
|
|
|
9797 }
|
|
323
|
9798 else
|
|
|
9799 {
|
|
|
9800 /* Do one more byte at this node. */
|
|
|
9801 n = arridx[depth] + curi[depth];
|
|
|
9802 ++curi[depth];
|
|
|
9803 c = byts[n];
|
|
|
9804 if (c == 0)
|
|
|
9805 {
|
|
|
9806 /* End of word, deal with the word. */
|
|
324
|
9807 flags = (int)idxs[n];
|
|
323
|
9808 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
9809 {
|
|
|
9810 tword[depth] = NUL;
|
|
375
|
9811 /* Sound-fold. Only in keep-case tree need to
|
|
|
9812 * case-fold the word. */
|
|
|
9813 spell_soundfold(lp->lp_slang, tword,
|
|
|
9814 round == 1, tsalword);
|
|
323
|
9815
|
|
344
|
9816 /* Compute the edit distance between the
|
|
|
9817 * sound-a-like words. */
|
|
|
9818 sound_score = soundalike_score(salword,
|
|
|
9819 tsalword);
|
|
483
|
9820
|
|
|
9821 /* Add a penalty for words in another region. */
|
|
|
9822 if ((flags & WF_REGION) && (((unsigned)flags
|
|
|
9823 >> 16) & lp->lp_region) == 0)
|
|
|
9824 local_score = SCORE_REGION;
|
|
|
9825 else
|
|
|
9826 local_score = 0;
|
|
|
9827 sound_score += local_score;
|
|
|
9828
|
|
324
|
9829 if (sound_score < SCORE_MAXMAX)
|
|
323
|
9830 {
|
|
324
|
9831 char_u cword[MAXWLEN];
|
|
|
9832 char_u *p;
|
|
344
|
9833 int score;
|
|
324
|
9834
|
|
388
|
9835 flags |= su->su_badflags;
|
|
351
|
9836 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
9837 {
|
|
324
|
9838 /* Need to fix case according to
|
|
|
9839 * "flags". */
|
|
323
|
9840 make_case_word(tword, cword, flags);
|
|
324
|
9841 p = cword;
|
|
323
|
9842 }
|
|
|
9843 else
|
|
324
|
9844 p = tword;
|
|
|
9845
|
|
344
|
9846 if (sps_flags & SPS_DOUBLE)
|
|
|
9847 add_suggestion(su, &su->su_sga, p,
|
|
346
|
9848 su->su_badlen,
|
|
351
|
9849 sound_score, 0, FALSE);
|
|
344
|
9850 else
|
|
|
9851 {
|
|
|
9852 /* Compute the score. */
|
|
|
9853 score = spell_edit_score(
|
|
483
|
9854 su->su_badword, p)
|
|
|
9855 + local_score;
|
|
344
|
9856 if (sps_flags & SPS_BEST)
|
|
|
9857 /* give a bonus for the good word
|
|
|
9858 * sounding the same as the bad
|
|
|
9859 * word */
|
|
|
9860 add_suggestion(su, &su->su_ga, p,
|
|
346
|
9861 su->su_badlen,
|
|
344
|
9862 RESCORE(score, sound_score),
|
|
351
|
9863 sound_score, TRUE);
|
|
344
|
9864 else
|
|
|
9865 add_suggestion(su, &su->su_ga, p,
|
|
346
|
9866 su->su_badlen,
|
|
351
|
9867 score + sound_score, 0, FALSE);
|
|
344
|
9868 }
|
|
323
|
9869 }
|
|
|
9870 }
|
|
|
9871
|
|
|
9872 /* Skip over other NUL bytes. */
|
|
|
9873 while (byts[n + 1] == 0)
|
|
|
9874 {
|
|
|
9875 ++n;
|
|
|
9876 ++curi[depth];
|
|
|
9877 }
|
|
|
9878 }
|
|
|
9879 else
|
|
|
9880 {
|
|
|
9881 /* Normal char, go one level deeper. */
|
|
|
9882 tword[depth++] = c;
|
|
|
9883 arridx[depth] = idxs[n];
|
|
|
9884 curi[depth] = 1;
|
|
|
9885 }
|
|
|
9886 }
|
|
|
9887 }
|
|
|
9888 }
|
|
|
9889 }
|
|
|
9890 }
|
|
|
9891 }
|
|
|
9892
|
|
|
9893 /*
|
|
324
|
9894 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
9895 */
|
|
|
9896 static void
|
|
|
9897 make_case_word(fword, cword, flags)
|
|
|
9898 char_u *fword;
|
|
|
9899 char_u *cword;
|
|
|
9900 int flags;
|
|
|
9901 {
|
|
|
9902 if (flags & WF_ALLCAP)
|
|
|
9903 /* Make it all upper-case */
|
|
|
9904 allcap_copy(fword, cword);
|
|
|
9905 else if (flags & WF_ONECAP)
|
|
|
9906 /* Make the first letter upper-case */
|
|
324
|
9907 onecap_copy(fword, cword, TRUE);
|
|
323
|
9908 else
|
|
|
9909 /* Use goodword as-is. */
|
|
|
9910 STRCPY(cword, fword);
|
|
|
9911 }
|
|
|
9912
|
|
330
|
9913 /*
|
|
|
9914 * Use map string "map" for languages "lp".
|
|
|
9915 */
|
|
|
9916 static void
|
|
|
9917 set_map_str(lp, map)
|
|
|
9918 slang_T *lp;
|
|
|
9919 char_u *map;
|
|
|
9920 {
|
|
|
9921 char_u *p;
|
|
|
9922 int headc = 0;
|
|
|
9923 int c;
|
|
|
9924 int i;
|
|
|
9925
|
|
|
9926 if (*map == NUL)
|
|
|
9927 {
|
|
|
9928 lp->sl_has_map = FALSE;
|
|
|
9929 return;
|
|
|
9930 }
|
|
|
9931 lp->sl_has_map = TRUE;
|
|
|
9932
|
|
|
9933 /* Init the array and hash table empty. */
|
|
|
9934 for (i = 0; i < 256; ++i)
|
|
|
9935 lp->sl_map_array[i] = 0;
|
|
|
9936 #ifdef FEAT_MBYTE
|
|
|
9937 hash_init(&lp->sl_map_hash);
|
|
|
9938 #endif
|
|
|
9939
|
|
|
9940 /*
|
|
|
9941 * The similar characters are stored separated with slashes:
|
|
|
9942 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
9943 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
9944 */
|
|
|
9945 for (p = map; *p != NUL; )
|
|
|
9946 {
|
|
|
9947 #ifdef FEAT_MBYTE
|
|
474
|
9948 c = mb_cptr2char_adv(&p);
|
|
330
|
9949 #else
|
|
|
9950 c = *p++;
|
|
|
9951 #endif
|
|
|
9952 if (c == '/')
|
|
|
9953 headc = 0;
|
|
|
9954 else
|
|
|
9955 {
|
|
|
9956 if (headc == 0)
|
|
|
9957 headc = c;
|
|
|
9958
|
|
|
9959 #ifdef FEAT_MBYTE
|
|
|
9960 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
9961 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
9962 * a NUL. */
|
|
|
9963 if (c >= 256)
|
|
|
9964 {
|
|
|
9965 int cl = mb_char2len(c);
|
|
|
9966 int headcl = mb_char2len(headc);
|
|
|
9967 char_u *b;
|
|
|
9968 hash_T hash;
|
|
|
9969 hashitem_T *hi;
|
|
|
9970
|
|
|
9971 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
9972 if (b == NULL)
|
|
|
9973 return;
|
|
|
9974 mb_char2bytes(c, b);
|
|
|
9975 b[cl] = NUL;
|
|
|
9976 mb_char2bytes(headc, b + cl + 1);
|
|
|
9977 b[cl + 1 + headcl] = NUL;
|
|
|
9978 hash = hash_hash(b);
|
|
|
9979 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
9980 if (HASHITEM_EMPTY(hi))
|
|
|
9981 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
9982 else
|
|
|
9983 {
|
|
|
9984 /* This should have been checked when generating the .spl
|
|
|
9985 * file. */
|
|
|
9986 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
9987 vim_free(b);
|
|
|
9988 }
|
|
|
9989 }
|
|
|
9990 else
|
|
|
9991 #endif
|
|
|
9992 lp->sl_map_array[c] = headc;
|
|
|
9993 }
|
|
|
9994 }
|
|
|
9995 }
|
|
|
9996
|
|
323
|
9997 /*
|
|
|
9998 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
9999 * lines in the .aff file.
|
|
|
10000 */
|
|
|
10001 static int
|
|
|
10002 similar_chars(slang, c1, c2)
|
|
|
10003 slang_T *slang;
|
|
|
10004 int c1;
|
|
|
10005 int c2;
|
|
|
10006 {
|
|
330
|
10007 int m1, m2;
|
|
|
10008 #ifdef FEAT_MBYTE
|
|
|
10009 char_u buf[MB_MAXBYTES];
|
|
|
10010 hashitem_T *hi;
|
|
|
10011
|
|
|
10012 if (c1 >= 256)
|
|
|
10013 {
|
|
|
10014 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
10015 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
10016 if (HASHITEM_EMPTY(hi))
|
|
|
10017 m1 = 0;
|
|
|
10018 else
|
|
|
10019 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
10020 }
|
|
|
10021 else
|
|
|
10022 #endif
|
|
|
10023 m1 = slang->sl_map_array[c1];
|
|
|
10024 if (m1 == 0)
|
|
323
|
10025 return FALSE;
|
|
330
|
10026
|
|
|
10027
|
|
|
10028 #ifdef FEAT_MBYTE
|
|
|
10029 if (c2 >= 256)
|
|
|
10030 {
|
|
|
10031 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
10032 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
10033 if (HASHITEM_EMPTY(hi))
|
|
|
10034 m2 = 0;
|
|
|
10035 else
|
|
|
10036 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
10037 }
|
|
|
10038 else
|
|
|
10039 #endif
|
|
|
10040 m2 = slang->sl_map_array[c2];
|
|
|
10041
|
|
|
10042 return m1 == m2;
|
|
323
|
10043 }
|
|
|
10044
|
|
|
10045 /*
|
|
|
10046 * Add a suggestion to the list of suggestions.
|
|
|
10047 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
10048 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
10049 * with spell_edit_score().
|
|
|
10050 */
|
|
|
10051 static void
|
|
351
|
10052 add_suggestion(su, gap, goodword, badlen, score, altscore, had_bonus)
|
|
323
|
10053 suginfo_T *su;
|
|
344
|
10054 garray_T *gap;
|
|
323
|
10055 char_u *goodword;
|
|
346
|
10056 int badlen; /* length of bad word used */
|
|
324
|
10057 int score;
|
|
351
|
10058 int altscore;
|
|
344
|
10059 int had_bonus; /* value for st_had_bonus */
|
|
323
|
10060 {
|
|
|
10061 suggest_T *stp;
|
|
|
10062 int i;
|
|
346
|
10063 char_u *p = NULL;
|
|
|
10064 int c = 0;
|
|
323
|
10065
|
|
|
10066 /* Check that the word wasn't banned. */
|
|
|
10067 if (was_banned(su, goodword))
|
|
|
10068 return;
|
|
|
10069
|
|
346
|
10070 /* If past "su_badlen" and the rest is identical stop at "su_badlen".
|
|
|
10071 * Remove the common part from "goodword". */
|
|
|
10072 i = badlen - su->su_badlen;
|
|
|
10073 if (i > 0)
|
|
|
10074 {
|
|
|
10075 /* This assumes there was no case folding or it didn't change the
|
|
|
10076 * length... */
|
|
|
10077 p = goodword + STRLEN(goodword) - i;
|
|
|
10078 if (p > goodword && STRNICMP(su->su_badptr + su->su_badlen, p, i) == 0)
|
|
|
10079 {
|
|
|
10080 badlen = su->su_badlen;
|
|
|
10081 c = *p;
|
|
|
10082 *p = NUL;
|
|
|
10083 }
|
|
|
10084 else
|
|
|
10085 p = NULL;
|
|
|
10086 }
|
|
372
|
10087 else if (i < 0)
|
|
|
10088 {
|
|
|
10089 /* When replacing part of the word check that we actually change
|
|
|
10090 * something. For "the the" a suggestion can be replacing the first
|
|
|
10091 * "the" with itself, since "the" wasn't banned. */
|
|
376
|
10092 if (badlen == (int)STRLEN(goodword)
|
|
372
|
10093 && STRNCMP(su->su_badword, goodword, badlen) == 0)
|
|
|
10094 return;
|
|
|
10095 }
|
|
|
10096
|
|
346
|
10097
|
|
323
|
10098 if (score <= su->su_maxscore)
|
|
|
10099 {
|
|
366
|
10100 /* Check if the word is already there. Also check the length that is
|
|
|
10101 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
10102 * "thes" -> "these". */
|
|
344
|
10103 stp = &SUG(*gap, 0);
|
|
|
10104 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
366
|
10105 if (STRCMP(stp[i].st_word, goodword) == 0
|
|
|
10106 && stp[i].st_orglen == badlen)
|
|
323
|
10107 {
|
|
|
10108 /* Found it. Remember the lowest score. */
|
|
|
10109 if (stp[i].st_score > score)
|
|
324
|
10110 {
|
|
323
|
10111 stp[i].st_score = score;
|
|
376
|
10112 stp[i].st_altscore = altscore;
|
|
324
|
10113 stp[i].st_had_bonus = had_bonus;
|
|
|
10114 }
|
|
323
|
10115 break;
|
|
|
10116 }
|
|
|
10117
|
|
344
|
10118 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
10119 {
|
|
|
10120 /* Add a suggestion. */
|
|
344
|
10121 stp = &SUG(*gap, gap->ga_len);
|
|
323
|
10122 stp->st_word = vim_strsave(goodword);
|
|
|
10123 if (stp->st_word != NULL)
|
|
|
10124 {
|
|
|
10125 stp->st_score = score;
|
|
351
|
10126 stp->st_altscore = altscore;
|
|
324
|
10127 stp->st_had_bonus = had_bonus;
|
|
346
|
10128 stp->st_orglen = badlen;
|
|
344
|
10129 ++gap->ga_len;
|
|
323
|
10130
|
|
|
10131 /* If we have too many suggestions now, sort the list and keep
|
|
|
10132 * the best suggestions. */
|
|
344
|
10133 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
10134 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
10135 SUG_CLEAN_COUNT(su));
|
|
323
|
10136 }
|
|
|
10137 }
|
|
|
10138 }
|
|
346
|
10139
|
|
|
10140 if (p != NULL)
|
|
|
10141 *p = c; /* restore "goodword" */
|
|
323
|
10142 }
|
|
|
10143
|
|
|
10144 /*
|
|
|
10145 * Add a word to be banned.
|
|
|
10146 */
|
|
|
10147 static void
|
|
|
10148 add_banned(su, word)
|
|
|
10149 suginfo_T *su;
|
|
|
10150 char_u *word;
|
|
|
10151 {
|
|
|
10152 char_u *s = vim_strsave(word);
|
|
|
10153 hash_T hash;
|
|
|
10154 hashitem_T *hi;
|
|
|
10155
|
|
|
10156 if (s != NULL)
|
|
|
10157 {
|
|
|
10158 hash = hash_hash(s);
|
|
|
10159 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
10160 if (HASHITEM_EMPTY(hi))
|
|
|
10161 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
10162 else
|
|
|
10163 vim_free(s);
|
|
323
|
10164 }
|
|
|
10165 }
|
|
|
10166
|
|
|
10167 /*
|
|
|
10168 * Return TRUE if a word appears in the list of banned words.
|
|
|
10169 */
|
|
|
10170 static int
|
|
|
10171 was_banned(su, word)
|
|
|
10172 suginfo_T *su;
|
|
|
10173 char_u *word;
|
|
|
10174 {
|
|
324
|
10175 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
10176
|
|
|
10177 return !HASHITEM_EMPTY(hi);
|
|
323
|
10178 }
|
|
|
10179
|
|
|
10180 /*
|
|
|
10181 * Free the banned words in "su".
|
|
|
10182 */
|
|
|
10183 static void
|
|
|
10184 free_banned(su)
|
|
|
10185 suginfo_T *su;
|
|
|
10186 {
|
|
|
10187 int todo;
|
|
|
10188 hashitem_T *hi;
|
|
|
10189
|
|
|
10190 todo = su->su_banned.ht_used;
|
|
|
10191 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
10192 {
|
|
|
10193 if (!HASHITEM_EMPTY(hi))
|
|
|
10194 {
|
|
|
10195 vim_free(hi->hi_key);
|
|
|
10196 --todo;
|
|
|
10197 }
|
|
|
10198 }
|
|
|
10199 hash_clear(&su->su_banned);
|
|
|
10200 }
|
|
|
10201
|
|
324
|
10202 /*
|
|
|
10203 * Recompute the score if sound-folding is possible. This is slow,
|
|
|
10204 * thus only done for the final results.
|
|
|
10205 */
|
|
|
10206 static void
|
|
|
10207 rescore_suggestions(su)
|
|
|
10208 suginfo_T *su;
|
|
|
10209 {
|
|
|
10210 langp_T *lp;
|
|
|
10211 suggest_T *stp;
|
|
|
10212 char_u sal_badword[MAXWLEN];
|
|
|
10213 int i;
|
|
|
10214
|
|
|
10215 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10216 lp->lp_slang != NULL; ++lp)
|
|
|
10217 {
|
|
|
10218 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10219 {
|
|
|
10220 /* soundfold the bad word */
|
|
375
|
10221 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, sal_badword);
|
|
324
|
10222
|
|
|
10223 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10224 {
|
|
344
|
10225 stp = &SUG(su->su_ga, i);
|
|
324
|
10226 if (!stp->st_had_bonus)
|
|
|
10227 {
|
|
351
|
10228 stp->st_altscore = stp_sal_score(stp, su,
|
|
|
10229 lp->lp_slang, sal_badword);
|
|
|
10230 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
10231 stp->st_altscore = SCORE_BIG;
|
|
|
10232 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
324
|
10233 }
|
|
|
10234 }
|
|
|
10235 break;
|
|
|
10236 }
|
|
|
10237 }
|
|
|
10238 }
|
|
|
10239
|
|
323
|
10240 static int
|
|
|
10241 #ifdef __BORLANDC__
|
|
|
10242 _RTLENTRYF
|
|
|
10243 #endif
|
|
|
10244 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
10245
|
|
|
10246 /*
|
|
|
10247 * Function given to qsort() to sort the suggestions on st_score.
|
|
|
10248 */
|
|
|
10249 static int
|
|
|
10250 #ifdef __BORLANDC__
|
|
|
10251 _RTLENTRYF
|
|
|
10252 #endif
|
|
|
10253 sug_compare(s1, s2)
|
|
|
10254 const void *s1;
|
|
|
10255 const void *s2;
|
|
|
10256 {
|
|
|
10257 suggest_T *p1 = (suggest_T *)s1;
|
|
|
10258 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
10259 int n = p1->st_score - p2->st_score;
|
|
|
10260
|
|
|
10261 if (n == 0)
|
|
|
10262 return p1->st_altscore - p2->st_altscore;
|
|
|
10263 return n;
|
|
323
|
10264 }
|
|
|
10265
|
|
|
10266 /*
|
|
|
10267 * Cleanup the suggestions:
|
|
|
10268 * - Sort on score.
|
|
|
10269 * - Remove words that won't be displayed.
|
|
344
|
10270 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
10271 */
|
|
344
|
10272 static int
|
|
|
10273 cleanup_suggestions(gap, maxscore, keep)
|
|
|
10274 garray_T *gap;
|
|
|
10275 int maxscore;
|
|
324
|
10276 int keep; /* nr of suggestions to keep */
|
|
323
|
10277 {
|
|
344
|
10278 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
10279 int i;
|
|
|
10280
|
|
|
10281 /* Sort the list. */
|
|
344
|
10282 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
10283
|
|
|
10284 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
10285 if (gap->ga_len > keep)
|
|
323
|
10286 {
|
|
344
|
10287 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
10288 vim_free(stp[i].st_word);
|
|
344
|
10289 gap->ga_len = keep;
|
|
|
10290 return stp[keep - 1].st_score;
|
|
323
|
10291 }
|
|
344
|
10292 return maxscore;
|
|
323
|
10293 }
|
|
|
10294
|
|
372
|
10295 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
10296 /*
|
|
|
10297 * Soundfold a string, for soundfold().
|
|
|
10298 * Result is in allocated memory, NULL for an error.
|
|
|
10299 */
|
|
|
10300 char_u *
|
|
|
10301 eval_soundfold(word)
|
|
|
10302 char_u *word;
|
|
|
10303 {
|
|
|
10304 langp_T *lp;
|
|
|
10305 char_u sound[MAXWLEN];
|
|
|
10306
|
|
|
10307 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
10308 /* Use the sound-folding of the first language that supports it. */
|
|
|
10309 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
|
10310 lp->lp_slang != NULL; ++lp)
|
|
|
10311 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10312 {
|
|
|
10313 /* soundfold the word */
|
|
375
|
10314 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
10315 return vim_strsave(sound);
|
|
|
10316 }
|
|
|
10317
|
|
|
10318 /* No language with sound folding, return word as-is. */
|
|
|
10319 return vim_strsave(word);
|
|
|
10320 }
|
|
|
10321 #endif
|
|
|
10322
|
|
323
|
10323 /*
|
|
|
10324 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
10325 */
|
|
|
10326 static void
|
|
375
|
10327 spell_soundfold(slang, inword, folded, res)
|
|
|
10328 slang_T *slang;
|
|
|
10329 char_u *inword;
|
|
|
10330 int folded; /* "inword" is already case-folded */
|
|
|
10331 char_u *res;
|
|
|
10332 {
|
|
|
10333 char_u fword[MAXWLEN];
|
|
|
10334 char_u *word;
|
|
|
10335
|
|
|
10336 if (slang->sl_sofo)
|
|
|
10337 /* SOFOFROM and SOFOTO used */
|
|
|
10338 spell_soundfold_sofo(slang, inword, res);
|
|
|
10339 else
|
|
|
10340 {
|
|
|
10341 /* SAL items used. Requires the word to be case-folded. */
|
|
|
10342 if (folded)
|
|
|
10343 word = inword;
|
|
|
10344 else
|
|
|
10345 {
|
|
|
10346 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
10347 word = fword;
|
|
|
10348 }
|
|
|
10349
|
|
|
10350 #ifdef FEAT_MBYTE
|
|
|
10351 if (has_mbyte)
|
|
|
10352 spell_soundfold_wsal(slang, word, res);
|
|
|
10353 else
|
|
|
10354 #endif
|
|
|
10355 spell_soundfold_sal(slang, word, res);
|
|
|
10356 }
|
|
|
10357 }
|
|
|
10358
|
|
|
10359 /*
|
|
|
10360 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
10361 * SOFOTO lines.
|
|
|
10362 */
|
|
|
10363 static void
|
|
|
10364 spell_soundfold_sofo(slang, inword, res)
|
|
|
10365 slang_T *slang;
|
|
|
10366 char_u *inword;
|
|
|
10367 char_u *res;
|
|
|
10368 {
|
|
|
10369 char_u *s;
|
|
|
10370 int ri = 0;
|
|
|
10371 int c;
|
|
|
10372
|
|
|
10373 #ifdef FEAT_MBYTE
|
|
|
10374 if (has_mbyte)
|
|
|
10375 {
|
|
|
10376 int prevc = 0;
|
|
|
10377 int *ip;
|
|
|
10378
|
|
|
10379 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
10380 * 255, sl_sal the rest. */
|
|
|
10381 for (s = inword; *s != NUL; )
|
|
|
10382 {
|
|
474
|
10383 c = mb_cptr2char_adv(&s);
|
|
375
|
10384 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
10385 c = ' ';
|
|
|
10386 else if (c < 256)
|
|
|
10387 c = slang->sl_sal_first[c];
|
|
|
10388 else
|
|
|
10389 {
|
|
|
10390 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
10391 if (ip == NULL) /* empty list, can't match */
|
|
|
10392 c = NUL;
|
|
|
10393 else
|
|
|
10394 for (;;) /* find "c" in the list */
|
|
|
10395 {
|
|
|
10396 if (*ip == 0) /* not found */
|
|
|
10397 {
|
|
|
10398 c = NUL;
|
|
|
10399 break;
|
|
|
10400 }
|
|
|
10401 if (*ip == c) /* match! */
|
|
|
10402 {
|
|
|
10403 c = ip[1];
|
|
|
10404 break;
|
|
|
10405 }
|
|
|
10406 ip += 2;
|
|
|
10407 }
|
|
|
10408 }
|
|
|
10409
|
|
|
10410 if (c != NUL && c != prevc)
|
|
|
10411 {
|
|
|
10412 ri += mb_char2bytes(c, res + ri);
|
|
|
10413 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
10414 break;
|
|
|
10415 prevc = c;
|
|
|
10416 }
|
|
|
10417 }
|
|
|
10418 }
|
|
|
10419 else
|
|
|
10420 #endif
|
|
|
10421 {
|
|
|
10422 /* The sl_sal_first[] table contains the translation. */
|
|
|
10423 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
10424 {
|
|
|
10425 if (vim_iswhite(c))
|
|
|
10426 c = ' ';
|
|
|
10427 else
|
|
|
10428 c = slang->sl_sal_first[c];
|
|
|
10429 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
10430 res[ri++] = c;
|
|
|
10431 }
|
|
|
10432 }
|
|
|
10433
|
|
|
10434 res[ri] = NUL;
|
|
|
10435 }
|
|
|
10436
|
|
|
10437 static void
|
|
|
10438 spell_soundfold_sal(slang, inword, res)
|
|
323
|
10439 slang_T *slang;
|
|
|
10440 char_u *inword;
|
|
|
10441 char_u *res;
|
|
|
10442 {
|
|
344
|
10443 salitem_T *smp;
|
|
323
|
10444 char_u word[MAXWLEN];
|
|
375
|
10445 char_u *s = inword;
|
|
323
|
10446 char_u *t;
|
|
344
|
10447 char_u *pf;
|
|
323
|
10448 int i, j, z;
|
|
344
|
10449 int reslen;
|
|
323
|
10450 int n, k = 0;
|
|
|
10451 int z0;
|
|
|
10452 int k0;
|
|
|
10453 int n0;
|
|
|
10454 int c;
|
|
|
10455 int pri;
|
|
|
10456 int p0 = -333;
|
|
|
10457 int c0;
|
|
|
10458
|
|
324
|
10459 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
10460 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
10461 if (slang->sl_rem_accents)
|
|
|
10462 {
|
|
|
10463 t = word;
|
|
375
|
10464 while (*s != NUL)
|
|
323
|
10465 {
|
|
324
|
10466 if (vim_iswhite(*s))
|
|
344
|
10467 {
|
|
|
10468 *t++ = ' ';
|
|
|
10469 s = skipwhite(s);
|
|
|
10470 }
|
|
323
|
10471 else
|
|
|
10472 {
|
|
376
|
10473 if (spell_iswordp_nmw(s))
|
|
323
|
10474 *t++ = *s;
|
|
|
10475 ++s;
|
|
|
10476 }
|
|
|
10477 }
|
|
|
10478 *t = NUL;
|
|
|
10479 }
|
|
|
10480 else
|
|
375
|
10481 STRCPY(word, s);
|
|
323
|
10482
|
|
344
|
10483 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
10484
|
|
|
10485 /*
|
|
|
10486 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
10487 * Changed to keep spaces.
|
|
323
|
10488 */
|
|
344
|
10489 i = reslen = z = 0;
|
|
323
|
10490 while ((c = word[i]) != NUL)
|
|
|
10491 {
|
|
344
|
10492 /* Start with the first rule that has the character in the word. */
|
|
323
|
10493 n = slang->sl_sal_first[c];
|
|
|
10494 z0 = 0;
|
|
|
10495
|
|
|
10496 if (n >= 0)
|
|
|
10497 {
|
|
|
10498 /* check all rules for the same letter */
|
|
344
|
10499 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
10500 {
|
|
344
|
10501 /* Quickly skip entries that don't match the word. Most
|
|
|
10502 * entries are less then three chars, optimize for that. */
|
|
|
10503 k = smp[n].sm_leadlen;
|
|
|
10504 if (k > 1)
|
|
323
|
10505 {
|
|
344
|
10506 if (word[i + 1] != s[1])
|
|
|
10507 continue;
|
|
|
10508 if (k > 2)
|
|
|
10509 {
|
|
|
10510 for (j = 2; j < k; ++j)
|
|
|
10511 if (word[i + j] != s[j])
|
|
|
10512 break;
|
|
|
10513 if (j < k)
|
|
|
10514 continue;
|
|
|
10515 }
|
|
323
|
10516 }
|
|
|
10517
|
|
375
|
10518 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
10519 {
|
|
375
|
10520 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
10521 while (*pf != NUL && *pf != word[i + k])
|
|
|
10522 ++pf;
|
|
|
10523 if (*pf == NUL)
|
|
|
10524 continue;
|
|
|
10525 ++k;
|
|
323
|
10526 }
|
|
344
|
10527 s = smp[n].sm_rules;
|
|
|
10528 pri = 5; /* default priority */
|
|
323
|
10529
|
|
|
10530 p0 = *s;
|
|
|
10531 k0 = k;
|
|
|
10532 while (*s == '-' && k > 1)
|
|
|
10533 {
|
|
|
10534 k--;
|
|
|
10535 s++;
|
|
|
10536 }
|
|
|
10537 if (*s == '<')
|
|
|
10538 s++;
|
|
344
|
10539 if (VIM_ISDIGIT(*s))
|
|
323
|
10540 {
|
|
|
10541 /* determine priority */
|
|
|
10542 pri = *s - '0';
|
|
|
10543 s++;
|
|
|
10544 }
|
|
|
10545 if (*s == '^' && *(s + 1) == '^')
|
|
|
10546 s++;
|
|
|
10547
|
|
|
10548 if (*s == NUL
|
|
|
10549 || (*s == '^'
|
|
324
|
10550 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
10551 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
10552 && (*(s + 1) != '$'
|
|
376
|
10553 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
10554 || (*s == '$' && i > 0
|
|
376
|
10555 && spell_iswordp(word + i - 1, curbuf)
|
|
|
10556 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
10557 {
|
|
|
10558 /* search for followup rules, if: */
|
|
|
10559 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
10560 c0 = word[i + k - 1];
|
|
|
10561 n0 = slang->sl_sal_first[c0];
|
|
|
10562
|
|
|
10563 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
10564 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
10565 {
|
|
|
10566 /* test follow-up rule for "word[i + k]" */
|
|
344
|
10567 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
10568 {
|
|
344
|
10569 /* Quickly skip entries that don't match the word.
|
|
|
10570 * */
|
|
|
10571 k0 = smp[n0].sm_leadlen;
|
|
|
10572 if (k0 > 1)
|
|
323
|
10573 {
|
|
344
|
10574 if (word[i + k] != s[1])
|
|
|
10575 continue;
|
|
|
10576 if (k0 > 2)
|
|
|
10577 {
|
|
|
10578 pf = word + i + k + 1;
|
|
|
10579 for (j = 2; j < k0; ++j)
|
|
|
10580 if (*pf++ != s[j])
|
|
|
10581 break;
|
|
|
10582 if (j < k0)
|
|
|
10583 continue;
|
|
|
10584 }
|
|
323
|
10585 }
|
|
344
|
10586 k0 += k - 1;
|
|
|
10587
|
|
375
|
10588 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
10589 {
|
|
344
|
10590 /* Check for match with one of the chars in
|
|
375
|
10591 * "sm_oneof". */
|
|
344
|
10592 while (*pf != NUL && *pf != word[i + k0])
|
|
|
10593 ++pf;
|
|
|
10594 if (*pf == NUL)
|
|
|
10595 continue;
|
|
|
10596 ++k0;
|
|
323
|
10597 }
|
|
344
|
10598
|
|
|
10599 p0 = 5;
|
|
|
10600 s = smp[n0].sm_rules;
|
|
323
|
10601 while (*s == '-')
|
|
|
10602 {
|
|
344
|
10603 /* "k0" gets NOT reduced because
|
|
|
10604 * "if (k0 == k)" */
|
|
323
|
10605 s++;
|
|
|
10606 }
|
|
|
10607 if (*s == '<')
|
|
|
10608 s++;
|
|
344
|
10609 if (VIM_ISDIGIT(*s))
|
|
323
|
10610 {
|
|
|
10611 p0 = *s - '0';
|
|
|
10612 s++;
|
|
|
10613 }
|
|
|
10614
|
|
|
10615 if (*s == NUL
|
|
|
10616 /* *s == '^' cuts */
|
|
|
10617 || (*s == '$'
|
|
376
|
10618 && !spell_iswordp(word + i + k0,
|
|
|
10619 curbuf)))
|
|
323
|
10620 {
|
|
|
10621 if (k0 == k)
|
|
|
10622 /* this is just a piece of the string */
|
|
|
10623 continue;
|
|
|
10624
|
|
|
10625 if (p0 < pri)
|
|
|
10626 /* priority too low */
|
|
|
10627 continue;
|
|
|
10628 /* rule fits; stop search */
|
|
|
10629 break;
|
|
|
10630 }
|
|
|
10631 }
|
|
|
10632
|
|
344
|
10633 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
10634 continue;
|
|
|
10635 }
|
|
|
10636
|
|
|
10637 /* replace string */
|
|
344
|
10638 s = smp[n].sm_to;
|
|
389
|
10639 if (s == NULL)
|
|
|
10640 s = (char_u *)"";
|
|
344
|
10641 pf = smp[n].sm_rules;
|
|
|
10642 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
10643 if (p0 == 1 && z == 0)
|
|
|
10644 {
|
|
|
10645 /* rule with '<' is used */
|
|
344
|
10646 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
10647 || res[reslen - 1] == *s))
|
|
|
10648 reslen--;
|
|
323
|
10649 z0 = 1;
|
|
|
10650 z = 1;
|
|
|
10651 k0 = 0;
|
|
372
|
10652 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
10653 {
|
|
|
10654 word[i + k0] = *s;
|
|
|
10655 k0++;
|
|
|
10656 s++;
|
|
|
10657 }
|
|
|
10658 if (k > k0)
|
|
|
10659 mch_memmove(word + i + k0, word + i + k,
|
|
|
10660 STRLEN(word + i + k) + 1);
|
|
|
10661
|
|
|
10662 /* new "actual letter" */
|
|
|
10663 c = word[i];
|
|
|
10664 }
|
|
|
10665 else
|
|
|
10666 {
|
|
|
10667 /* no '<' rule used */
|
|
|
10668 i += k - 1;
|
|
|
10669 z = 0;
|
|
344
|
10670 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
10671 {
|
|
344
|
10672 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
10673 res[reslen++] = *s;
|
|
323
|
10674 s++;
|
|
|
10675 }
|
|
|
10676 /* new "actual letter" */
|
|
|
10677 c = *s;
|
|
344
|
10678 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
10679 {
|
|
|
10680 if (c != NUL)
|
|
372
|
10681 res[reslen++] = c;
|
|
323
|
10682 mch_memmove(word, word + i + 1,
|
|
|
10683 STRLEN(word + i + 1) + 1);
|
|
|
10684 i = 0;
|
|
|
10685 z0 = 1;
|
|
|
10686 }
|
|
|
10687 }
|
|
|
10688 break;
|
|
|
10689 }
|
|
|
10690 }
|
|
|
10691 }
|
|
324
|
10692 else if (vim_iswhite(c))
|
|
|
10693 {
|
|
|
10694 c = ' ';
|
|
|
10695 k = 1;
|
|
|
10696 }
|
|
323
|
10697
|
|
|
10698 if (z0 == 0)
|
|
|
10699 {
|
|
344
|
10700 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
10701 && (!slang->sl_collapse || reslen == 0
|
|
|
10702 || res[reslen - 1] != c))
|
|
323
|
10703 /* condense only double letters */
|
|
372
|
10704 res[reslen++] = c;
|
|
323
|
10705
|
|
|
10706 i++;
|
|
|
10707 z = 0;
|
|
|
10708 k = 0;
|
|
|
10709 }
|
|
|
10710 }
|
|
|
10711
|
|
344
|
10712 res[reslen] = NUL;
|
|
323
|
10713 }
|
|
|
10714
|
|
372
|
10715 #ifdef FEAT_MBYTE
|
|
|
10716 /*
|
|
|
10717 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
10718 * Multi-byte version of spell_soundfold().
|
|
|
10719 */
|
|
|
10720 static void
|
|
375
|
10721 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
10722 slang_T *slang;
|
|
|
10723 char_u *inword;
|
|
|
10724 char_u *res;
|
|
|
10725 {
|
|
375
|
10726 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
10727 int word[MAXWLEN];
|
|
|
10728 int wres[MAXWLEN];
|
|
|
10729 int l;
|
|
|
10730 char_u *s;
|
|
|
10731 int *ws;
|
|
|
10732 char_u *t;
|
|
|
10733 int *pf;
|
|
|
10734 int i, j, z;
|
|
|
10735 int reslen;
|
|
|
10736 int n, k = 0;
|
|
|
10737 int z0;
|
|
|
10738 int k0;
|
|
|
10739 int n0;
|
|
|
10740 int c;
|
|
|
10741 int pri;
|
|
|
10742 int p0 = -333;
|
|
|
10743 int c0;
|
|
|
10744 int did_white = FALSE;
|
|
|
10745
|
|
|
10746 /*
|
|
|
10747 * Convert the multi-byte string to a wide-character string.
|
|
|
10748 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
10749 * But keep white space.
|
|
|
10750 */
|
|
|
10751 n = 0;
|
|
|
10752 for (s = inword; *s != NUL; )
|
|
|
10753 {
|
|
|
10754 t = s;
|
|
474
|
10755 c = mb_cptr2char_adv(&s);
|
|
372
|
10756 if (slang->sl_rem_accents)
|
|
|
10757 {
|
|
|
10758 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
10759 {
|
|
|
10760 if (did_white)
|
|
|
10761 continue;
|
|
|
10762 c = ' ';
|
|
|
10763 did_white = TRUE;
|
|
|
10764 }
|
|
|
10765 else
|
|
|
10766 {
|
|
|
10767 did_white = FALSE;
|
|
376
|
10768 if (!spell_iswordp_nmw(t))
|
|
372
|
10769 continue;
|
|
|
10770 }
|
|
|
10771 }
|
|
|
10772 word[n++] = c;
|
|
|
10773 }
|
|
|
10774 word[n] = NUL;
|
|
|
10775
|
|
|
10776 /*
|
|
|
10777 * This comes from Aspell phonet.cpp.
|
|
|
10778 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
10779 * Changed to keep spaces.
|
|
|
10780 */
|
|
|
10781 i = reslen = z = 0;
|
|
|
10782 while ((c = word[i]) != NUL)
|
|
|
10783 {
|
|
|
10784 /* Start with the first rule that has the character in the word. */
|
|
|
10785 n = slang->sl_sal_first[c & 0xff];
|
|
|
10786 z0 = 0;
|
|
|
10787
|
|
|
10788 if (n >= 0)
|
|
|
10789 {
|
|
375
|
10790 /* check all rules for the same index byte */
|
|
372
|
10791 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
10792 {
|
|
|
10793 /* Quickly skip entries that don't match the word. Most
|
|
|
10794 * entries are less then three chars, optimize for that. */
|
|
375
|
10795 if (c != ws[0])
|
|
|
10796 continue;
|
|
372
|
10797 k = smp[n].sm_leadlen;
|
|
|
10798 if (k > 1)
|
|
|
10799 {
|
|
|
10800 if (word[i + 1] != ws[1])
|
|
|
10801 continue;
|
|
|
10802 if (k > 2)
|
|
|
10803 {
|
|
|
10804 for (j = 2; j < k; ++j)
|
|
|
10805 if (word[i + j] != ws[j])
|
|
|
10806 break;
|
|
|
10807 if (j < k)
|
|
|
10808 continue;
|
|
|
10809 }
|
|
|
10810 }
|
|
|
10811
|
|
375
|
10812 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
10813 {
|
|
375
|
10814 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
10815 while (*pf != NUL && *pf != word[i + k])
|
|
|
10816 ++pf;
|
|
|
10817 if (*pf == NUL)
|
|
|
10818 continue;
|
|
|
10819 ++k;
|
|
|
10820 }
|
|
|
10821 s = smp[n].sm_rules;
|
|
|
10822 pri = 5; /* default priority */
|
|
|
10823
|
|
|
10824 p0 = *s;
|
|
|
10825 k0 = k;
|
|
|
10826 while (*s == '-' && k > 1)
|
|
|
10827 {
|
|
|
10828 k--;
|
|
|
10829 s++;
|
|
|
10830 }
|
|
|
10831 if (*s == '<')
|
|
|
10832 s++;
|
|
|
10833 if (VIM_ISDIGIT(*s))
|
|
|
10834 {
|
|
|
10835 /* determine priority */
|
|
|
10836 pri = *s - '0';
|
|
|
10837 s++;
|
|
|
10838 }
|
|
|
10839 if (*s == '^' && *(s + 1) == '^')
|
|
|
10840 s++;
|
|
|
10841
|
|
|
10842 if (*s == NUL
|
|
|
10843 || (*s == '^'
|
|
|
10844 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
10845 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
10846 && (*(s + 1) != '$'
|
|
376
|
10847 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
10848 || (*s == '$' && i > 0
|
|
376
|
10849 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
10850 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
10851 {
|
|
|
10852 /* search for followup rules, if: */
|
|
|
10853 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
10854 c0 = word[i + k - 1];
|
|
|
10855 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
10856
|
|
|
10857 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
10858 && p0 != '-' && word[i + k] != NUL)
|
|
|
10859 {
|
|
375
|
10860 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
10861 * all entries with the same index byte. */
|
|
372
|
10862 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
10863 == (c0 & 0xff); ++n0)
|
|
|
10864 {
|
|
|
10865 /* Quickly skip entries that don't match the word.
|
|
375
|
10866 */
|
|
|
10867 if (c0 != ws[0])
|
|
|
10868 continue;
|
|
372
|
10869 k0 = smp[n0].sm_leadlen;
|
|
|
10870 if (k0 > 1)
|
|
|
10871 {
|
|
|
10872 if (word[i + k] != ws[1])
|
|
|
10873 continue;
|
|
|
10874 if (k0 > 2)
|
|
|
10875 {
|
|
|
10876 pf = word + i + k + 1;
|
|
|
10877 for (j = 2; j < k0; ++j)
|
|
|
10878 if (*pf++ != ws[j])
|
|
|
10879 break;
|
|
|
10880 if (j < k0)
|
|
|
10881 continue;
|
|
|
10882 }
|
|
|
10883 }
|
|
|
10884 k0 += k - 1;
|
|
|
10885
|
|
375
|
10886 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
10887 {
|
|
|
10888 /* Check for match with one of the chars in
|
|
375
|
10889 * "sm_oneof". */
|
|
372
|
10890 while (*pf != NUL && *pf != word[i + k0])
|
|
|
10891 ++pf;
|
|
|
10892 if (*pf == NUL)
|
|
|
10893 continue;
|
|
|
10894 ++k0;
|
|
|
10895 }
|
|
|
10896
|
|
|
10897 p0 = 5;
|
|
|
10898 s = smp[n0].sm_rules;
|
|
|
10899 while (*s == '-')
|
|
|
10900 {
|
|
|
10901 /* "k0" gets NOT reduced because
|
|
|
10902 * "if (k0 == k)" */
|
|
|
10903 s++;
|
|
|
10904 }
|
|
|
10905 if (*s == '<')
|
|
|
10906 s++;
|
|
|
10907 if (VIM_ISDIGIT(*s))
|
|
|
10908 {
|
|
|
10909 p0 = *s - '0';
|
|
|
10910 s++;
|
|
|
10911 }
|
|
|
10912
|
|
|
10913 if (*s == NUL
|
|
|
10914 /* *s == '^' cuts */
|
|
|
10915 || (*s == '$'
|
|
376
|
10916 && !spell_iswordp_w(word + i + k0,
|
|
|
10917 curbuf)))
|
|
372
|
10918 {
|
|
|
10919 if (k0 == k)
|
|
|
10920 /* this is just a piece of the string */
|
|
|
10921 continue;
|
|
|
10922
|
|
|
10923 if (p0 < pri)
|
|
|
10924 /* priority too low */
|
|
|
10925 continue;
|
|
|
10926 /* rule fits; stop search */
|
|
|
10927 break;
|
|
|
10928 }
|
|
|
10929 }
|
|
|
10930
|
|
|
10931 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
10932 == (c0 & 0xff))
|
|
|
10933 continue;
|
|
|
10934 }
|
|
|
10935
|
|
|
10936 /* replace string */
|
|
|
10937 ws = smp[n].sm_to_w;
|
|
|
10938 s = smp[n].sm_rules;
|
|
|
10939 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
10940 if (p0 == 1 && z == 0)
|
|
|
10941 {
|
|
|
10942 /* rule with '<' is used */
|
|
389
|
10943 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
10944 && (wres[reslen - 1] == c
|
|
372
|
10945 || wres[reslen - 1] == *ws))
|
|
|
10946 reslen--;
|
|
|
10947 z0 = 1;
|
|
|
10948 z = 1;
|
|
|
10949 k0 = 0;
|
|
389
|
10950 if (ws != NULL)
|
|
|
10951 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
10952 {
|
|
|
10953 word[i + k0] = *ws;
|
|
|
10954 k0++;
|
|
|
10955 ws++;
|
|
|
10956 }
|
|
372
|
10957 if (k > k0)
|
|
|
10958 mch_memmove(word + i + k0, word + i + k,
|
|
|
10959 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
10960
|
|
|
10961 /* new "actual letter" */
|
|
|
10962 c = word[i];
|
|
|
10963 }
|
|
|
10964 else
|
|
|
10965 {
|
|
|
10966 /* no '<' rule used */
|
|
|
10967 i += k - 1;
|
|
|
10968 z = 0;
|
|
389
|
10969 if (ws != NULL)
|
|
|
10970 while (*ws != NUL && ws[1] != NUL
|
|
|
10971 && reslen < MAXWLEN)
|
|
|
10972 {
|
|
|
10973 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
10974 wres[reslen++] = *ws;
|
|
|
10975 ws++;
|
|
|
10976 }
|
|
372
|
10977 /* new "actual letter" */
|
|
389
|
10978 if (ws == NULL)
|
|
|
10979 c = NUL;
|
|
|
10980 else
|
|
|
10981 c = *ws;
|
|
372
|
10982 if (strstr((char *)s, "^^") != NULL)
|
|
|
10983 {
|
|
|
10984 if (c != NUL)
|
|
|
10985 wres[reslen++] = c;
|
|
|
10986 mch_memmove(word, word + i + 1,
|
|
|
10987 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
10988 i = 0;
|
|
|
10989 z0 = 1;
|
|
|
10990 }
|
|
|
10991 }
|
|
|
10992 break;
|
|
|
10993 }
|
|
|
10994 }
|
|
|
10995 }
|
|
|
10996 else if (vim_iswhite(c))
|
|
|
10997 {
|
|
|
10998 c = ' ';
|
|
|
10999 k = 1;
|
|
|
11000 }
|
|
|
11001
|
|
|
11002 if (z0 == 0)
|
|
|
11003 {
|
|
|
11004 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
11005 && (!slang->sl_collapse || reslen == 0
|
|
|
11006 || wres[reslen - 1] != c))
|
|
|
11007 /* condense only double letters */
|
|
|
11008 wres[reslen++] = c;
|
|
|
11009
|
|
|
11010 i++;
|
|
|
11011 z = 0;
|
|
|
11012 k = 0;
|
|
|
11013 }
|
|
|
11014 }
|
|
|
11015
|
|
|
11016 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
11017 l = 0;
|
|
|
11018 for (n = 0; n < reslen; ++n)
|
|
|
11019 {
|
|
|
11020 l += mb_char2bytes(wres[n], res + l);
|
|
|
11021 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
11022 break;
|
|
|
11023 }
|
|
|
11024 res[l] = NUL;
|
|
|
11025 }
|
|
|
11026 #endif
|
|
|
11027
|
|
324
|
11028 /*
|
|
344
|
11029 * Compute a score for two sound-a-like words.
|
|
|
11030 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
11031 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
11032 * avoiding checks that will not be possible.
|
|
|
11033 */
|
|
|
11034 static int
|
|
351
|
11035 soundalike_score(goodstart, badstart)
|
|
|
11036 char_u *goodstart; /* sound-folded good word */
|
|
|
11037 char_u *badstart; /* sound-folded bad word */
|
|
344
|
11038 {
|
|
351
|
11039 char_u *goodsound = goodstart;
|
|
|
11040 char_u *badsound = badstart;
|
|
|
11041 int goodlen;
|
|
|
11042 int badlen;
|
|
344
|
11043 int n;
|
|
|
11044 char_u *pl, *ps;
|
|
|
11045 char_u *pl2, *ps2;
|
|
351
|
11046 int score = 0;
|
|
|
11047
|
|
|
11048 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
11049 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
11050 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
11051 {
|
|
|
11052 score = SCORE_DEL / 2;
|
|
|
11053 if (*badsound == '*')
|
|
|
11054 ++badsound;
|
|
|
11055 else
|
|
|
11056 ++goodsound;
|
|
|
11057 }
|
|
|
11058
|
|
|
11059 goodlen = STRLEN(goodsound);
|
|
|
11060 badlen = STRLEN(badsound);
|
|
344
|
11061
|
|
|
11062 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
11063 * changes. */
|
|
|
11064 n = goodlen - badlen;
|
|
|
11065 if (n < -2 || n > 2)
|
|
|
11066 return SCORE_MAXMAX;
|
|
|
11067
|
|
|
11068 if (n > 0)
|
|
|
11069 {
|
|
351
|
11070 pl = goodsound; /* goodsound is longest */
|
|
344
|
11071 ps = badsound;
|
|
|
11072 }
|
|
|
11073 else
|
|
|
11074 {
|
|
351
|
11075 pl = badsound; /* badsound is longest */
|
|
344
|
11076 ps = goodsound;
|
|
|
11077 }
|
|
|
11078
|
|
|
11079 /* Skip over the identical part. */
|
|
|
11080 while (*pl == *ps && *pl != NUL)
|
|
|
11081 {
|
|
|
11082 ++pl;
|
|
|
11083 ++ps;
|
|
|
11084 }
|
|
|
11085
|
|
|
11086 switch (n)
|
|
|
11087 {
|
|
|
11088 case -2:
|
|
|
11089 case 2:
|
|
|
11090 /*
|
|
|
11091 * Must delete two characters from "pl".
|
|
|
11092 */
|
|
|
11093 ++pl; /* first delete */
|
|
|
11094 while (*pl == *ps)
|
|
|
11095 {
|
|
|
11096 ++pl;
|
|
|
11097 ++ps;
|
|
|
11098 }
|
|
|
11099 /* strings must be equal after second delete */
|
|
|
11100 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
11101 return score + SCORE_DEL * 2;
|
|
344
|
11102
|
|
|
11103 /* Failed to compare. */
|
|
|
11104 break;
|
|
|
11105
|
|
|
11106 case -1:
|
|
|
11107 case 1:
|
|
|
11108 /*
|
|
|
11109 * Minimal one delete from "pl" required.
|
|
|
11110 */
|
|
|
11111
|
|
|
11112 /* 1: delete */
|
|
|
11113 pl2 = pl + 1;
|
|
|
11114 ps2 = ps;
|
|
|
11115 while (*pl2 == *ps2)
|
|
|
11116 {
|
|
|
11117 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11118 return score + SCORE_DEL;
|
|
344
|
11119 ++pl2;
|
|
|
11120 ++ps2;
|
|
|
11121 }
|
|
|
11122
|
|
|
11123 /* 2: delete then swap, then rest must be equal */
|
|
|
11124 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11125 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11126 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
11127
|
|
|
11128 /* 3: delete then substitute, then the rest must be equal */
|
|
|
11129 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11130 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
11131
|
|
|
11132 /* 4: first swap then delete */
|
|
|
11133 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
11134 {
|
|
|
11135 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
11136 ps2 = ps + 2;
|
|
|
11137 while (*pl2 == *ps2)
|
|
|
11138 {
|
|
|
11139 ++pl2;
|
|
|
11140 ++ps2;
|
|
|
11141 }
|
|
|
11142 /* delete a char and then strings must be equal */
|
|
|
11143 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11144 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
11145 }
|
|
|
11146
|
|
|
11147 /* 5: first substitute then delete */
|
|
|
11148 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
11149 ps2 = ps + 1;
|
|
|
11150 while (*pl2 == *ps2)
|
|
|
11151 {
|
|
|
11152 ++pl2;
|
|
|
11153 ++ps2;
|
|
|
11154 }
|
|
|
11155 /* delete a char and then strings must be equal */
|
|
|
11156 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11157 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
11158
|
|
|
11159 /* Failed to compare. */
|
|
|
11160 break;
|
|
|
11161
|
|
|
11162 case 0:
|
|
|
11163 /*
|
|
|
11164 * Lenghts are equal, thus changes must result in same length: An
|
|
|
11165 * insert is only possible in combination with a delete.
|
|
|
11166 * 1: check if for identical strings
|
|
|
11167 */
|
|
|
11168 if (*pl == NUL)
|
|
351
|
11169 return score;
|
|
344
|
11170
|
|
|
11171 /* 2: swap */
|
|
|
11172 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
11173 {
|
|
|
11174 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
11175 ps2 = ps + 2;
|
|
|
11176 while (*pl2 == *ps2)
|
|
|
11177 {
|
|
|
11178 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11179 return score + SCORE_SWAP;
|
|
344
|
11180 ++pl2;
|
|
|
11181 ++ps2;
|
|
|
11182 }
|
|
|
11183 /* 3: swap and swap again */
|
|
|
11184 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11185 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11186 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
11187
|
|
|
11188 /* 4: swap and substitute */
|
|
|
11189 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11190 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
11191 }
|
|
|
11192
|
|
|
11193 /* 5: substitute */
|
|
|
11194 pl2 = pl + 1;
|
|
|
11195 ps2 = ps + 1;
|
|
|
11196 while (*pl2 == *ps2)
|
|
|
11197 {
|
|
|
11198 if (*pl2 == NUL) /* reached the end */
|
|
351
|
11199 return score + SCORE_SUBST;
|
|
344
|
11200 ++pl2;
|
|
|
11201 ++ps2;
|
|
|
11202 }
|
|
|
11203
|
|
|
11204 /* 6: substitute and swap */
|
|
|
11205 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
11206 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
11207 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
11208
|
|
|
11209 /* 7: substitute and substitute */
|
|
|
11210 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
11211 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
11212
|
|
|
11213 /* 8: insert then delete */
|
|
|
11214 pl2 = pl;
|
|
|
11215 ps2 = ps + 1;
|
|
|
11216 while (*pl2 == *ps2)
|
|
|
11217 {
|
|
|
11218 ++pl2;
|
|
|
11219 ++ps2;
|
|
|
11220 }
|
|
|
11221 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
11222 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
11223
|
|
|
11224 /* 9: delete then insert */
|
|
|
11225 pl2 = pl + 1;
|
|
|
11226 ps2 = ps;
|
|
|
11227 while (*pl2 == *ps2)
|
|
|
11228 {
|
|
|
11229 ++pl2;
|
|
|
11230 ++ps2;
|
|
|
11231 }
|
|
|
11232 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
11233 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
11234
|
|
|
11235 /* Failed to compare. */
|
|
|
11236 break;
|
|
|
11237 }
|
|
|
11238
|
|
|
11239 return SCORE_MAXMAX;
|
|
|
11240 }
|
|
324
|
11241
|
|
323
|
11242 /*
|
|
|
11243 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
11244 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
11245 *
|
|
323
|
11246 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
324
|
11247 * It has been converted from C++ to C and modified to support multi-byte
|
|
|
11248 * characters.
|
|
323
|
11249 */
|
|
|
11250 static int
|
|
|
11251 spell_edit_score(badword, goodword)
|
|
|
11252 char_u *badword;
|
|
|
11253 char_u *goodword;
|
|
|
11254 {
|
|
|
11255 int *cnt;
|
|
372
|
11256 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
11257 int j, i;
|
|
|
11258 int t;
|
|
|
11259 int bc, gc;
|
|
324
|
11260 int pbc, pgc;
|
|
|
11261 #ifdef FEAT_MBYTE
|
|
|
11262 char_u *p;
|
|
|
11263 int wbadword[MAXWLEN];
|
|
|
11264 int wgoodword[MAXWLEN];
|
|
|
11265
|
|
|
11266 if (has_mbyte)
|
|
|
11267 {
|
|
|
11268 /* Get the characters from the multi-byte strings and put them in an
|
|
|
11269 * int array for easy access. */
|
|
|
11270 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
11271 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
11272 wbadword[badlen++] = 0;
|
|
324
|
11273 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
11274 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
11275 wgoodword[goodlen++] = 0;
|
|
324
|
11276 }
|
|
|
11277 else
|
|
|
11278 #endif
|
|
|
11279 {
|
|
|
11280 badlen = STRLEN(badword) + 1;
|
|
|
11281 goodlen = STRLEN(goodword) + 1;
|
|
|
11282 }
|
|
323
|
11283
|
|
|
11284 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
11285 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
11286 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
11287 TRUE);
|
|
324
|
11288 if (cnt == NULL)
|
|
|
11289 return 0; /* out of memory */
|
|
323
|
11290
|
|
|
11291 CNT(0, 0) = 0;
|
|
|
11292 for (j = 1; j <= goodlen; ++j)
|
|
|
11293 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
11294
|
|
|
11295 for (i = 1; i <= badlen; ++i)
|
|
|
11296 {
|
|
|
11297 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
11298 for (j = 1; j <= goodlen; ++j)
|
|
|
11299 {
|
|
324
|
11300 #ifdef FEAT_MBYTE
|
|
|
11301 if (has_mbyte)
|
|
|
11302 {
|
|
|
11303 bc = wbadword[i - 1];
|
|
|
11304 gc = wgoodword[j - 1];
|
|
|
11305 }
|
|
|
11306 else
|
|
|
11307 #endif
|
|
|
11308 {
|
|
|
11309 bc = badword[i - 1];
|
|
|
11310 gc = goodword[j - 1];
|
|
|
11311 }
|
|
323
|
11312 if (bc == gc)
|
|
|
11313 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
11314 else
|
|
|
11315 {
|
|
|
11316 /* Use a better score when there is only a case difference. */
|
|
324
|
11317 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
11318 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
11319 else
|
|
|
11320 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
11321
|
|
324
|
11322 if (i > 1 && j > 1)
|
|
323
|
11323 {
|
|
324
|
11324 #ifdef FEAT_MBYTE
|
|
|
11325 if (has_mbyte)
|
|
|
11326 {
|
|
|
11327 pbc = wbadword[i - 2];
|
|
|
11328 pgc = wgoodword[j - 2];
|
|
|
11329 }
|
|
|
11330 else
|
|
|
11331 #endif
|
|
|
11332 {
|
|
|
11333 pbc = badword[i - 2];
|
|
|
11334 pgc = goodword[j - 2];
|
|
|
11335 }
|
|
|
11336 if (bc == pgc && pbc == gc)
|
|
|
11337 {
|
|
|
11338 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
11339 if (t < CNT(i, j))
|
|
|
11340 CNT(i, j) = t;
|
|
|
11341 }
|
|
323
|
11342 }
|
|
|
11343 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
11344 if (t < CNT(i, j))
|
|
|
11345 CNT(i, j) = t;
|
|
|
11346 t = SCORE_INS + CNT(i, j - 1);
|
|
|
11347 if (t < CNT(i, j))
|
|
|
11348 CNT(i, j) = t;
|
|
|
11349 }
|
|
|
11350 }
|
|
|
11351 }
|
|
344
|
11352
|
|
|
11353 i = CNT(badlen - 1, goodlen - 1);
|
|
|
11354 vim_free(cnt);
|
|
|
11355 return i;
|
|
323
|
11356 }
|
|
307
|
11357
|
|
351
|
11358 /*
|
|
|
11359 * ":spelldump"
|
|
|
11360 */
|
|
|
11361 /*ARGSUSED*/
|
|
|
11362 void
|
|
|
11363 ex_spelldump(eap)
|
|
|
11364 exarg_T *eap;
|
|
|
11365 {
|
|
|
11366 buf_T *buf = curbuf;
|
|
|
11367 langp_T *lp;
|
|
|
11368 slang_T *slang;
|
|
|
11369 idx_T arridx[MAXWLEN];
|
|
|
11370 int curi[MAXWLEN];
|
|
|
11371 char_u word[MAXWLEN];
|
|
|
11372 int c;
|
|
|
11373 char_u *byts;
|
|
|
11374 idx_T *idxs;
|
|
|
11375 linenr_T lnum = 0;
|
|
|
11376 int round;
|
|
|
11377 int depth;
|
|
|
11378 int n;
|
|
|
11379 int flags;
|
|
381
|
11380 char_u *region_names = NULL; /* region names being used */
|
|
|
11381 int do_region = TRUE; /* dump region names and numbers */
|
|
|
11382 char_u *p;
|
|
351
|
11383
|
|
|
11384 if (no_spell_checking())
|
|
|
11385 return;
|
|
|
11386
|
|
|
11387 /* Create a new empty buffer by splitting the window. */
|
|
|
11388 do_cmdline_cmd((char_u *)"new");
|
|
|
11389 if (!bufempty() || !buf_valid(buf))
|
|
|
11390 return;
|
|
|
11391
|
|
381
|
11392 /* Find out if we can support regions: All languages must support the same
|
|
|
11393 * regions or none at all. */
|
|
|
11394 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
11395 {
|
|
|
11396 p = lp->lp_slang->sl_regions;
|
|
|
11397 if (p[0] != 0)
|
|
|
11398 {
|
|
|
11399 if (region_names == NULL) /* first language with regions */
|
|
|
11400 region_names = p;
|
|
|
11401 else if (STRCMP(region_names, p) != 0)
|
|
|
11402 {
|
|
|
11403 do_region = FALSE; /* region names are different */
|
|
|
11404 break;
|
|
|
11405 }
|
|
|
11406 }
|
|
|
11407 }
|
|
|
11408
|
|
|
11409 if (do_region && region_names != NULL)
|
|
|
11410 {
|
|
|
11411 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
11412 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
11413 }
|
|
|
11414 else
|
|
|
11415 do_region = FALSE;
|
|
|
11416
|
|
|
11417 /*
|
|
|
11418 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
11419 */
|
|
351
|
11420 for (lp = LANGP_ENTRY(buf->b_langp, 0); lp->lp_slang != NULL; ++lp)
|
|
|
11421 {
|
|
|
11422 slang = lp->lp_slang;
|
|
|
11423
|
|
|
11424 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
11425 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
11426
|
|
|
11427 /* round 1: case-folded tree
|
|
|
11428 * round 2: keep-case tree */
|
|
|
11429 for (round = 1; round <= 2; ++round)
|
|
|
11430 {
|
|
|
11431 if (round == 1)
|
|
|
11432 {
|
|
|
11433 byts = slang->sl_fbyts;
|
|
|
11434 idxs = slang->sl_fidxs;
|
|
|
11435 }
|
|
|
11436 else
|
|
|
11437 {
|
|
|
11438 byts = slang->sl_kbyts;
|
|
|
11439 idxs = slang->sl_kidxs;
|
|
|
11440 }
|
|
|
11441 if (byts == NULL)
|
|
|
11442 continue; /* array is empty */
|
|
|
11443
|
|
|
11444 depth = 0;
|
|
|
11445 arridx[0] = 0;
|
|
|
11446 curi[0] = 1;
|
|
|
11447 while (depth >= 0 && !got_int)
|
|
|
11448 {
|
|
|
11449 if (curi[depth] > byts[arridx[depth]])
|
|
|
11450 {
|
|
|
11451 /* Done all bytes at this node, go up one level. */
|
|
|
11452 --depth;
|
|
|
11453 line_breakcheck();
|
|
|
11454 }
|
|
|
11455 else
|
|
|
11456 {
|
|
|
11457 /* Do one more byte at this node. */
|
|
|
11458 n = arridx[depth] + curi[depth];
|
|
|
11459 ++curi[depth];
|
|
|
11460 c = byts[n];
|
|
|
11461 if (c == 0)
|
|
|
11462 {
|
|
|
11463 /* End of word, deal with the word.
|
|
|
11464 * Don't use keep-case words in the fold-case tree,
|
|
|
11465 * they will appear in the keep-case tree.
|
|
|
11466 * Only use the word when the region matches. */
|
|
|
11467 flags = (int)idxs[n];
|
|
|
11468 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
381
|
11469 && (do_region
|
|
|
11470 || (flags & WF_REGION) == 0
|
|
390
|
11471 || (((unsigned)flags >> 16)
|
|
351
|
11472 & lp->lp_region) != 0))
|
|
|
11473 {
|
|
|
11474 word[depth] = NUL;
|
|
381
|
11475 if (!do_region)
|
|
|
11476 flags &= ~WF_REGION;
|
|
355
|
11477
|
|
|
11478 /* Dump the basic word if there is no prefix or
|
|
|
11479 * when it's the first one. */
|
|
390
|
11480 c = (unsigned)flags >> 24;
|
|
355
|
11481 if (c == 0 || curi[depth] == 2)
|
|
|
11482 dump_word(word, round, flags, lnum++);
|
|
351
|
11483
|
|
|
11484 /* Apply the prefix, if there is one. */
|
|
355
|
11485 if (c != 0)
|
|
351
|
11486 lnum = apply_prefixes(slang, word, round,
|
|
|
11487 flags, lnum);
|
|
|
11488 }
|
|
|
11489 }
|
|
|
11490 else
|
|
|
11491 {
|
|
|
11492 /* Normal char, go one level deeper. */
|
|
|
11493 word[depth++] = c;
|
|
|
11494 arridx[depth] = idxs[n];
|
|
|
11495 curi[depth] = 1;
|
|
|
11496 }
|
|
|
11497 }
|
|
|
11498 }
|
|
|
11499 }
|
|
|
11500 }
|
|
|
11501
|
|
|
11502 /* Delete the empty line that we started with. */
|
|
|
11503 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
11504 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
11505
|
|
|
11506 redraw_later(NOT_VALID);
|
|
|
11507 }
|
|
|
11508
|
|
|
11509 /*
|
|
|
11510 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
11511 */
|
|
|
11512 static void
|
|
|
11513 dump_word(word, round, flags, lnum)
|
|
|
11514 char_u *word;
|
|
|
11515 int round;
|
|
|
11516 int flags;
|
|
|
11517 linenr_T lnum;
|
|
|
11518 {
|
|
|
11519 int keepcap = FALSE;
|
|
|
11520 char_u *p;
|
|
|
11521 char_u cword[MAXWLEN];
|
|
381
|
11522 char_u badword[MAXWLEN + 10];
|
|
|
11523 int i;
|
|
351
|
11524
|
|
|
11525 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
11526 {
|
|
|
11527 /* Need to fix case according to "flags". */
|
|
|
11528 make_case_word(word, cword, flags);
|
|
|
11529 p = cword;
|
|
|
11530 }
|
|
|
11531 else
|
|
|
11532 {
|
|
|
11533 p = word;
|
|
389
|
11534 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
11535 || (flags & WF_FIXCAP) != 0))
|
|
351
|
11536 keepcap = TRUE;
|
|
|
11537 }
|
|
|
11538
|
|
381
|
11539 /* Add flags and regions after a slash. */
|
|
|
11540 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
11541 {
|
|
|
11542 STRCPY(badword, p);
|
|
|
11543 STRCAT(badword, "/");
|
|
351
|
11544 if (keepcap)
|
|
|
11545 STRCAT(badword, "=");
|
|
|
11546 if (flags & WF_BANNED)
|
|
|
11547 STRCAT(badword, "!");
|
|
|
11548 else if (flags & WF_RARE)
|
|
|
11549 STRCAT(badword, "?");
|
|
381
|
11550 if (flags & WF_REGION)
|
|
|
11551 for (i = 0; i < 7; ++i)
|
|
390
|
11552 if (flags & (0x10000 << i))
|
|
381
|
11553 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
11554 p = badword;
|
|
|
11555 }
|
|
|
11556
|
|
|
11557 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
11558 }
|
|
|
11559
|
|
|
11560 /*
|
|
372
|
11561 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
11562 * "word" and append a line to the buffer.
|
|
351
|
11563 * Return the updated line number.
|
|
|
11564 */
|
|
|
11565 static linenr_T
|
|
|
11566 apply_prefixes(slang, word, round, flags, startlnum)
|
|
|
11567 slang_T *slang;
|
|
|
11568 char_u *word; /* case-folded word */
|
|
|
11569 int round;
|
|
|
11570 int flags; /* flags with prefix ID */
|
|
|
11571 linenr_T startlnum;
|
|
|
11572 {
|
|
|
11573 idx_T arridx[MAXWLEN];
|
|
|
11574 int curi[MAXWLEN];
|
|
|
11575 char_u prefix[MAXWLEN];
|
|
455
|
11576 char_u word_up[MAXWLEN];
|
|
|
11577 int has_word_up = FALSE;
|
|
351
|
11578 int c;
|
|
|
11579 char_u *byts;
|
|
|
11580 idx_T *idxs;
|
|
|
11581 linenr_T lnum = startlnum;
|
|
|
11582 int depth;
|
|
|
11583 int n;
|
|
|
11584 int len;
|
|
|
11585 int i;
|
|
|
11586
|
|
455
|
11587 /* if the word starts with a lower-case letter make the word with an
|
|
|
11588 * upper-case letter in word_up[]. */
|
|
|
11589 c = PTR2CHAR(word);
|
|
|
11590 if (SPELL_TOUPPER(c) != c)
|
|
|
11591 {
|
|
|
11592 onecap_copy(word, word_up, TRUE);
|
|
|
11593 has_word_up = TRUE;
|
|
|
11594 }
|
|
|
11595
|
|
351
|
11596 byts = slang->sl_pbyts;
|
|
|
11597 idxs = slang->sl_pidxs;
|
|
|
11598 if (byts != NULL) /* array not is empty */
|
|
|
11599 {
|
|
|
11600 /*
|
|
|
11601 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
11602 * When at the end of a prefix check that it supports "flags".
|
|
351
|
11603 */
|
|
|
11604 depth = 0;
|
|
|
11605 arridx[0] = 0;
|
|
|
11606 curi[0] = 1;
|
|
|
11607 while (depth >= 0 && !got_int)
|
|
|
11608 {
|
|
390
|
11609 n = arridx[depth];
|
|
|
11610 len = byts[n];
|
|
|
11611 if (curi[depth] > len)
|
|
351
|
11612 {
|
|
|
11613 /* Done all bytes at this node, go up one level. */
|
|
|
11614 --depth;
|
|
|
11615 line_breakcheck();
|
|
|
11616 }
|
|
|
11617 else
|
|
|
11618 {
|
|
|
11619 /* Do one more byte at this node. */
|
|
390
|
11620 n += curi[depth];
|
|
351
|
11621 ++curi[depth];
|
|
|
11622 c = byts[n];
|
|
|
11623 if (c == 0)
|
|
|
11624 {
|
|
|
11625 /* End of prefix, find out how many IDs there are. */
|
|
|
11626 for (i = 1; i < len; ++i)
|
|
|
11627 if (byts[n + i] != 0)
|
|
|
11628 break;
|
|
|
11629 curi[depth] += i - 1;
|
|
|
11630
|
|
455
|
11631 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
11632 if (c != 0)
|
|
351
|
11633 {
|
|
376
|
11634 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
11635 dump_word(prefix, round,
|
|
455
|
11636 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
11637 : flags, lnum++);
|
|
351
|
11638 }
|
|
455
|
11639
|
|
|
11640 /* Check for prefix that matches the word when the
|
|
|
11641 * first letter is upper-case, but only if the prefix has
|
|
|
11642 * a condition. */
|
|
|
11643 if (has_word_up)
|
|
|
11644 {
|
|
|
11645 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
11646 TRUE);
|
|
|
11647 if (c != 0)
|
|
|
11648 {
|
|
|
11649 vim_strncpy(prefix + depth, word_up,
|
|
|
11650 MAXWLEN - depth - 1);
|
|
|
11651 dump_word(prefix, round,
|
|
|
11652 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
11653 : flags, lnum++);
|
|
|
11654 }
|
|
|
11655 }
|
|
351
|
11656 }
|
|
|
11657 else
|
|
|
11658 {
|
|
|
11659 /* Normal char, go one level deeper. */
|
|
|
11660 prefix[depth++] = c;
|
|
|
11661 arridx[depth] = idxs[n];
|
|
|
11662 curi[depth] = 1;
|
|
|
11663 }
|
|
|
11664 }
|
|
|
11665 }
|
|
|
11666 }
|
|
|
11667
|
|
|
11668 return lnum;
|
|
|
11669 }
|
|
|
11670
|
|
475
|
11671 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
11672 static int spell_expand_need_cap;
|
|
|
11673
|
|
|
11674 /*
|
|
|
11675 * Find start of the word in front of the cursor. We don't check if it is
|
|
|
11676 * badly spelled, with completion we can only change the word in front of the
|
|
|
11677 * cursor.
|
|
|
11678 * Used for Insert mode completion CTRL-X ?.
|
|
|
11679 * Returns the column number of the word.
|
|
|
11680 */
|
|
|
11681 int
|
|
|
11682 spell_word_start(startcol)
|
|
|
11683 int startcol;
|
|
|
11684 {
|
|
|
11685 char_u *line;
|
|
|
11686 char_u *p;
|
|
|
11687 int col = 0;
|
|
|
11688
|
|
|
11689 if (no_spell_checking())
|
|
|
11690 return startcol;
|
|
|
11691
|
|
|
11692 /* Find a word character before "startcol". */
|
|
|
11693 line = ml_get_curline();
|
|
|
11694 for (p = line + startcol; p > line; )
|
|
|
11695 {
|
|
|
11696 mb_ptr_back(line, p);
|
|
|
11697 if (spell_iswordp_nmw(p))
|
|
|
11698 break;
|
|
|
11699 }
|
|
|
11700
|
|
|
11701 /* Go back to start of the word. */
|
|
|
11702 while (p > line)
|
|
|
11703 {
|
|
|
11704 col = p - line;
|
|
|
11705 mb_ptr_back(line, p);
|
|
|
11706 if (!spell_iswordp(p, curbuf))
|
|
|
11707 break;
|
|
|
11708 col = 0;
|
|
|
11709 }
|
|
|
11710
|
|
|
11711 /* Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
11712 * expand_spelling() is called. Therefore the ugly global variable. */
|
|
|
11713 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
11714
|
|
|
11715 return col;
|
|
|
11716 }
|
|
|
11717
|
|
|
11718 /*
|
|
|
11719 * Get list of spelling suggestions.
|
|
|
11720 * Used for Insert mode completion CTRL-X ?.
|
|
|
11721 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
11722 * allocated strings.
|
|
|
11723 */
|
|
|
11724 /*ARGSUSED*/
|
|
|
11725 int
|
|
|
11726 expand_spelling(lnum, col, pat, matchp)
|
|
|
11727 linenr_T lnum;
|
|
|
11728 int col;
|
|
|
11729 char_u *pat;
|
|
|
11730 char_u ***matchp;
|
|
|
11731 {
|
|
|
11732 garray_T ga;
|
|
|
11733
|
|
|
11734 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap);
|
|
|
11735 *matchp = ga.ga_data;
|
|
|
11736 return ga.ga_len;
|
|
|
11737 }
|
|
|
11738 #endif
|
|
|
11739
|
|
236
|
11740 #endif /* FEAT_SYN_HL */
|
