|
223
|
1 /* vi:set ts=8 sts=4 sw=4:
|
|
|
2 *
|
|
|
3 * VIM - Vi IMproved by Bram Moolenaar
|
|
|
4 *
|
|
|
5 * Do ":help uganda" in Vim to read copying and usage conditions.
|
|
|
6 * Do ":help credits" in Vim to see a list of people who contributed.
|
|
|
7 * See README.txt for an overview of the Vim source code.
|
|
|
8 */
|
|
|
9
|
|
|
10 /*
|
|
|
11 * spell.c: code for spell checking
|
|
226
|
12 *
|
|
300
|
13 * The spell checking mechanism uses a tree (aka trie). Each node in the tree
|
|
|
14 * has a list of bytes that can appear (siblings). For each byte there is a
|
|
|
15 * pointer to the node with the byte that follows in the word (child).
|
|
324
|
16 *
|
|
|
17 * A NUL byte is used where the word may end. The bytes are sorted, so that
|
|
|
18 * binary searching can be used and the NUL bytes are at the start. The
|
|
|
19 * number of possible bytes is stored before the list of bytes.
|
|
|
20 *
|
|
|
21 * The tree uses two arrays: "byts" stores the characters, "idxs" stores
|
|
|
22 * either the next index or flags. The tree starts at index 0. For example,
|
|
|
23 * to lookup "vi" this sequence is followed:
|
|
|
24 * i = 0
|
|
|
25 * len = byts[i]
|
|
|
26 * n = where "v" appears in byts[i + 1] to byts[i + len]
|
|
|
27 * i = idxs[n]
|
|
|
28 * len = byts[i]
|
|
|
29 * n = where "i" appears in byts[i + 1] to byts[i + len]
|
|
|
30 * i = idxs[n]
|
|
|
31 * len = byts[i]
|
|
|
32 * find that byts[i + 1] is 0, idxs[i + 1] has flags for "vi".
|
|
300
|
33 *
|
|
339
|
34 * There are two word trees: one with case-folded words and one with words in
|
|
300
|
35 * original case. The second one is only used for keep-case words and is
|
|
|
36 * usually small.
|
|
|
37 *
|
|
481
|
38 * There is one additional tree for when not all prefixes are applied when
|
|
339
|
39 * generating the .spl file. This tree stores all the possible prefixes, as
|
|
|
40 * if they were words. At each word (prefix) end the prefix nr is stored, the
|
|
|
41 * following word must support this prefix nr. And the condition nr is
|
|
|
42 * stored, used to lookup the condition that the word must match with.
|
|
|
43 *
|
|
300
|
44 * Thanks to Olaf Seibert for providing an example implementation of this tree
|
|
|
45 * and the compression mechanism.
|
|
243
|
46 *
|
|
|
47 * Matching involves checking the caps type: Onecap ALLCAP KeepCap.
|
|
|
48 *
|
|
236
|
49 * Why doesn't Vim use aspell/ispell/myspell/etc.?
|
|
|
50 * See ":help develop-spell".
|
|
|
51 */
|
|
|
52
|
|
469
|
53 /* Use SPELL_PRINTTREE for debugging: dump the word tree after adding a word.
|
|
482
|
54 * Only use it for small word lists! */
|
|
469
|
55 #if 0
|
|
|
56 # define SPELL_PRINTTREE
|
|
482
|
57 #endif
|
|
|
58
|
|
300
|
59 /*
|
|
324
|
60 * Use this to adjust the score after finding suggestions, based on the
|
|
|
61 * suggested word sounding like the bad word. This is much faster than doing
|
|
|
62 * it for every possible suggestion.
|
|
|
63 * Disadvantage: When "the" is typed as "hte" it sounds different and goes
|
|
|
64 * down in the list.
|
|
344
|
65 * Used when 'spellsuggest' is set to "best".
|
|
|
66 */
|
|
|
67 #define RESCORE(word_score, sound_score) ((3 * word_score + sound_score) / 4)
|
|
|
68
|
|
|
69 /*
|
|
339
|
70 * Vim spell file format: <HEADER>
|
|
483
|
71 * <SECTIONS>
|
|
339
|
72 * <LWORDTREE>
|
|
|
73 * <KWORDTREE>
|
|
|
74 * <PREFIXTREE>
|
|
300
|
75 *
|
|
483
|
76 * <HEADER>: <fileID> <versionnr>
|
|
|
77 *
|
|
|
78 * <fileID> 8 bytes "VIMspell"
|
|
|
79 * <versionnr> 1 byte VIMSPELLVERSION
|
|
|
80 *
|
|
|
81 *
|
|
|
82 * Sections make it possible to add information to the .spl file without
|
|
|
83 * making it incompatible with previous versions. There are two kinds of
|
|
|
84 * sections:
|
|
|
85 * 1. Not essential for correct spell checking. E.g. for making suggestions.
|
|
|
86 * These are skipped when not supported.
|
|
|
87 * 2. Optional information, but essential for spell checking when present.
|
|
|
88 * E.g. conditions for affixes. When this section is present but not
|
|
|
89 * supported an error message is given.
|
|
300
|
90 *
|
|
483
|
91 * <SECTIONS>: <section> ... <sectionend>
|
|
|
92 *
|
|
|
93 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
94 *
|
|
|
95 * <sectionID> 1 byte number from 0 to 254 identifying the section
|
|
|
96 *
|
|
|
97 * <sectionflags> 1 byte SNF_REQUIRED: this section is required for correct
|
|
|
98 * spell checking
|
|
|
99 *
|
|
|
100 * <sectionlen> 4 bytes length of section contents, MSB first
|
|
|
101 *
|
|
|
102 * <sectionend> 1 byte SN_END
|
|
|
103 *
|
|
|
104 *
|
|
|
105 * sectionID == SN_REGION: <regionname> ...
|
|
|
106 * <regionname> 2 bytes Up to 8 region names: ca, au, etc. Lower case.
|
|
300
|
107 * First <regionname> is region 1.
|
|
|
108 *
|
|
483
|
109 * sectionID == SN_CHARFLAGS: <charflagslen> <charflags>
|
|
|
110 * <folcharslen> <folchars>
|
|
300
|
111 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
|
112 * <charflags> N bytes List of flags (first one is for character 128):
|
|
324
|
113 * 0x01 word character CF_WORD
|
|
|
114 * 0x02 upper-case character CF_UPPER
|
|
483
|
115 * <folcharslen> 2 bytes Number of bytes in <folchars>.
|
|
|
116 * <folchars> N bytes Folded characters, first one is for character 128.
|
|
300
|
117 *
|
|
483
|
118 * sectionID == SN_MIDWORD: <midword>
|
|
|
119 * <midword> N bytes Characters that are word characters only when used
|
|
366
|
120 * in the middle of a word.
|
|
|
121 *
|
|
483
|
122 * sectionID == SN_PREFCOND: <prefcondcnt> <prefcond> ...
|
|
339
|
123 * <prefcondcnt> 2 bytes Number of <prefcond> items following.
|
|
|
124 * <prefcond> : <condlen> <condstr>
|
|
|
125 * <condlen> 1 byte Length of <condstr>.
|
|
|
126 * <condstr> N bytes Condition for the prefix.
|
|
|
127 *
|
|
483
|
128 * sectionID == SN_REP: <repcount> <rep> ...
|
|
|
129 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
323
|
130 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
483
|
131 * <repfromlen> 1 byte length of <repfrom>
|
|
|
132 * <repfrom> N bytes "from" part of replacement
|
|
|
133 * <reptolen> 1 byte length of <repto>
|
|
|
134 * <repto> N bytes "to" part of replacement
|
|
323
|
135 *
|
|
483
|
136 * sectionID == SN_SAL: <salflags> <salcount> <sal> ...
|
|
|
137 * <salflags> 1 byte flags for soundsalike conversion:
|
|
323
|
138 * SAL_F0LLOWUP
|
|
|
139 * SAL_COLLAPSE
|
|
|
140 * SAL_REM_ACCENTS
|
|
483
|
141 * <salcount> 2 bytes number of <sal> items following
|
|
323
|
142 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
483
|
143 * <salfromlen> 1 byte length of <salfrom>
|
|
|
144 * <salfrom> N bytes "from" part of soundsalike
|
|
|
145 * <saltolen> 1 byte length of <salto>
|
|
|
146 * <salto> N bytes "to" part of soundsalike
|
|
323
|
147 *
|
|
483
|
148 * sectionID == SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
149 * <sofofromlen> 2 bytes length of <sofofrom>
|
|
|
150 * <sofofrom> N bytes "from" part of soundfold
|
|
|
151 * <sofotolen> 2 bytes length of <sofoto>
|
|
|
152 * <sofoto> N bytes "to" part of soundfold
|
|
323
|
153 *
|
|
483
|
154 * sectionID == SN_MAP: <mapstr>
|
|
|
155 * <mapstr> N bytes String with sequences of similar characters,
|
|
323
|
156 * separated by slashes.
|
|
300
|
157 *
|
|
483
|
158 * sectionID == SN_COMPOUND: <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
159 * <compmax> 1 byte Maximum nr of words in compound word.
|
|
|
160 * <compminlen> 1 byte Minimal word length for compounding.
|
|
|
161 * <compsylmax> 1 byte Maximum nr of syllables in compound word.
|
|
|
162 * <compflags> N bytes Flags from COMPOUNDFLAGS items, separated by
|
|
|
163 * slashes.
|
|
|
164 *
|
|
492
|
165 * sectionID == SN_NOBREAK: (empty, its presence is enough)
|
|
|
166 *
|
|
483
|
167 * sectionID == SN_SYLLABLE: <syllable>
|
|
|
168 * <syllable> N bytes String from SYLLABLE item.
|
|
300
|
169 *
|
|
|
170 * <LWORDTREE>: <wordtree>
|
|
|
171 *
|
|
339
|
172 * <KWORDTREE>: <wordtree>
|
|
|
173 *
|
|
|
174 * <PREFIXTREE>: <wordtree>
|
|
|
175 *
|
|
|
176 *
|
|
300
|
177 * <wordtree>: <nodecount> <nodedata> ...
|
|
|
178 *
|
|
|
179 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
|
180 *
|
|
|
181 * <nodedata>: <siblingcount> <sibling> ...
|
|
|
182 *
|
|
|
183 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
|
184 * follow in sorted order.
|
|
|
185 *
|
|
339
|
186 * <sibling>: <byte> [ <nodeidx> <xbyte>
|
|
481
|
187 * | <flags> [<flags2>] [<region>] [<affixID>]
|
|
|
188 * | [<pflags>] <affixID> <prefcondnr> ]
|
|
300
|
189 *
|
|
|
190 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
|
191 * BY_NOFLAGS: End of word without flags and for all
|
|
|
192 * regions.
|
|
481
|
193 * For PREFIXTREE <affixID> and
|
|
366
|
194 * <prefcondnr> follow.
|
|
390
|
195 * BY_FLAGS: End of word, <flags> follow.
|
|
481
|
196 * For PREFIXTREE <pflags>, <affixID>
|
|
455
|
197 * and <prefcondnr> follow.
|
|
390
|
198 * BY_FLAGS2: End of word, <flags> and <flags2>
|
|
|
199 * follow. Not used in PREFIXTREE.
|
|
|
200 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
300
|
201 * and <xbyte> follow.
|
|
|
202 *
|
|
|
203 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
|
204 *
|
|
|
205 * <xbyte> 1 byte byte value of the sibling.
|
|
|
206 *
|
|
|
207 * <flags> 1 byte bitmask of:
|
|
|
208 * WF_ALLCAP word must have only capitals
|
|
|
209 * WF_ONECAP first char of word must be capital
|
|
389
|
210 * WF_KEEPCAP keep-case word
|
|
|
211 * WF_FIXCAP keep-case word, all caps not allowed
|
|
300
|
212 * WF_RARE rare word
|
|
389
|
213 * WF_BANNED bad word
|
|
300
|
214 * WF_REGION <region> follows
|
|
481
|
215 * WF_AFX <affixID> follows
|
|
300
|
216 *
|
|
500
|
217 * <flags2> 1 byte Bitmask of:
|
|
390
|
218 * WF_HAS_AFF >> 8 word includes affix
|
|
500
|
219 * WF_NEEDCOMP >> 8 word only valid in compound
|
|
390
|
220 *
|
|
455
|
221 * <pflags> 1 byte bitmask of:
|
|
|
222 * WFP_RARE rare prefix
|
|
|
223 * WFP_NC non-combining prefix
|
|
|
224 * WFP_UP letter after prefix made upper case
|
|
|
225 *
|
|
300
|
226 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
|
227 * omitted it's valid in all regions.
|
|
|
228 * Lowest bit is for region 1.
|
|
|
229 *
|
|
481
|
230 * <affixID> 1 byte ID of affix that can be used with this word. In
|
|
339
|
231 * PREFIXTREE used for the required prefix ID.
|
|
|
232 *
|
|
|
233 * <prefcondnr> 2 bytes Prefix condition number, index in <prefcond> list
|
|
|
234 * from HEADER.
|
|
300
|
235 *
|
|
|
236 * All text characters are in 'encoding', but stored as single bytes.
|
|
|
237 */
|
|
|
238
|
|
223
|
239 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
|
240 # include <io.h> /* for lseek(), must be before vim.h */
|
|
|
241 #endif
|
|
|
242
|
|
|
243 #include "vim.h"
|
|
|
244
|
|
|
245 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
246
|
|
|
247 #ifdef HAVE_FCNTL_H
|
|
|
248 # include <fcntl.h>
|
|
|
249 #endif
|
|
|
250
|
|
323
|
251 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
|
252 Some places assume a word length fits in a
|
|
|
253 byte, thus it can't be above 255. */
|
|
226
|
254
|
|
490
|
255 /* Type used for indexes in the word tree need to be at least 4 bytes. If int
|
|
324
|
256 * is 8 bytes we could use something smaller, but what? */
|
|
490
|
257 #if SIZEOF_INT > 3
|
|
324
|
258 typedef int idx_T;
|
|
|
259 #else
|
|
|
260 typedef long idx_T;
|
|
|
261 #endif
|
|
|
262
|
|
|
263 /* Flags used for a word. Only the lowest byte can be used, the region byte
|
|
|
264 * comes above it. */
|
|
300
|
265 #define WF_REGION 0x01 /* region byte follows */
|
|
|
266 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
|
267 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
|
268 #define WF_RARE 0x08 /* rare word */
|
|
307
|
269 #define WF_BANNED 0x10 /* bad word */
|
|
481
|
270 #define WF_AFX 0x20 /* affix ID follows */
|
|
389
|
271 #define WF_FIXCAP 0x40 /* keep-case word, allcap not allowed */
|
|
323
|
272 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
|
273
|
|
390
|
274 /* for <flags2>, shifted up one byte to be used in wn_flags */
|
|
|
275 #define WF_HAS_AFF 0x0100 /* word includes affix */
|
|
500
|
276 #define WF_NEEDCOMP 0x0200 /* word only valid in compound */
|
|
390
|
277
|
|
389
|
278 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP | WF_FIXCAP)
|
|
300
|
279
|
|
455
|
280 /* flags for <pflags> */
|
|
|
281 #define WFP_RARE 0x01 /* rare prefix */
|
|
|
282 #define WFP_NC 0x02 /* prefix is not combining */
|
|
|
283 #define WFP_UP 0x04 /* to-upper prefix */
|
|
|
284
|
|
481
|
285 /* Flags for postponed prefixes. Must be above affixID (one byte)
|
|
390
|
286 * and prefcondnr (two bytes). */
|
|
455
|
287 #define WF_RAREPFX (WFP_RARE << 24) /* in sl_pidxs: flag for rare
|
|
|
288 * postponed prefix */
|
|
|
289 #define WF_PFX_NC (WFP_NC << 24) /* in sl_pidxs: flag for non-combining
|
|
|
290 * postponed prefix */
|
|
|
291 #define WF_PFX_UP (WFP_UP << 24) /* in sl_pidxs: flag for to-upper
|
|
|
292 * postponed prefix */
|
|
390
|
293
|
|
|
294 /* Special byte values for <byte>. Some are only used in the tree for
|
|
|
295 * postponed prefixes, some only in the other trees. This is a bit messy... */
|
|
|
296 #define BY_NOFLAGS 0 /* end of word without flags or region; for
|
|
455
|
297 * postponed prefix: no <pflags> */
|
|
|
298 #define BY_INDEX 1 /* child is shared, index follows */
|
|
|
299 #define BY_FLAGS 2 /* end of word, <flags> byte follows; for
|
|
|
300 * postponed prefix: <pflags> follows */
|
|
|
301 #define BY_FLAGS2 3 /* end of word, <flags> and <flags2> bytes
|
|
|
302 * follow; never used in prefix tree */
|
|
|
303 #define BY_SPECIAL BY_FLAGS2 /* highest special byte value */
|
|
236
|
304
|
|
323
|
305 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
344
|
306 * and si_sal. Not for sl_sal!
|
|
323
|
307 * One replacement: from "ft_from" to "ft_to". */
|
|
|
308 typedef struct fromto_S
|
|
236
|
309 {
|
|
323
|
310 char_u *ft_from;
|
|
|
311 char_u *ft_to;
|
|
|
312 } fromto_T;
|
|
236
|
313
|
|
344
|
314 /* Info from "SAL" entries in ".aff" file used in sl_sal.
|
|
|
315 * The info is split for quick processing by spell_soundfold().
|
|
|
316 * Note that "sm_oneof" and "sm_rules" point into sm_lead. */
|
|
|
317 typedef struct salitem_S
|
|
|
318 {
|
|
|
319 char_u *sm_lead; /* leading letters */
|
|
|
320 int sm_leadlen; /* length of "sm_lead" */
|
|
375
|
321 char_u *sm_oneof; /* letters from () or NULL */
|
|
344
|
322 char_u *sm_rules; /* rules like ^, $, priority */
|
|
|
323 char_u *sm_to; /* replacement. */
|
|
372
|
324 #ifdef FEAT_MBYTE
|
|
|
325 int *sm_lead_w; /* wide character copy of "sm_lead" */
|
|
375
|
326 int *sm_oneof_w; /* wide character copy of "sm_oneof" */
|
|
372
|
327 int *sm_to_w; /* wide character copy of "sm_to" */
|
|
|
328 #endif
|
|
344
|
329 } salitem_T;
|
|
|
330
|
|
375
|
331 #ifdef FEAT_MBYTE
|
|
|
332 typedef int salfirst_T;
|
|
|
333 #else
|
|
|
334 typedef short salfirst_T;
|
|
|
335 #endif
|
|
|
336
|
|
483
|
337 /* Values for SP_*ERROR are negative, positive values are used by
|
|
|
338 * read_cnt_string(). */
|
|
|
339 #define SP_TRUNCERROR -1 /* spell file truncated error */
|
|
|
340 #define SP_FORMERROR -2 /* format error in spell file */
|
|
495
|
341 #define SP_OTHERERROR -3 /* other error while reading spell file */
|
|
483
|
342
|
|
236
|
343 /*
|
|
243
|
344 * Structure used to store words and other info for one language, loaded from
|
|
|
345 * a .spl file.
|
|
300
|
346 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
|
347 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
|
348 *
|
|
|
349 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
|
350 * the number of possible bytes, sorted on byte value:
|
|
|
351 * <len> <byte1> <byte2> ...
|
|
|
352 * The "idxs" array stores the index of the child node corresponding to the
|
|
|
353 * byte in "byts".
|
|
|
354 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
481
|
355 * the flags, region mask and affixID for the word. There may be several
|
|
|
356 * zeros in sequence for alternative flag/region/affixID combinations.
|
|
236
|
357 */
|
|
|
358 typedef struct slang_S slang_T;
|
|
|
359 struct slang_S
|
|
|
360 {
|
|
|
361 slang_T *sl_next; /* next language */
|
|
|
362 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
|
310
|
363 char_u *sl_fname; /* name of .spl file */
|
|
323
|
364 int sl_add; /* TRUE if it's a .add file. */
|
|
339
|
365
|
|
300
|
366 char_u *sl_fbyts; /* case-folded word bytes */
|
|
324
|
367 idx_T *sl_fidxs; /* case-folded word indexes */
|
|
300
|
368 char_u *sl_kbyts; /* keep-case word bytes */
|
|
324
|
369 idx_T *sl_kidxs; /* keep-case word indexes */
|
|
339
|
370 char_u *sl_pbyts; /* prefix tree word bytes */
|
|
|
371 idx_T *sl_pidxs; /* prefix tree word indexes */
|
|
|
372
|
|
236
|
373 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
|
323
|
374
|
|
376
|
375 char_u *sl_midword; /* MIDWORD string or NULL */
|
|
|
376
|
|
483
|
377 int sl_compmax; /* COMPOUNDMAX (default: MAXWLEN) */
|
|
501
|
378 int sl_compminlen; /* COMPOUNDMIN (default: 0) */
|
|
483
|
379 int sl_compsylmax; /* COMPOUNDSYLMAX (default: MAXWLEN) */
|
|
|
380 regprog_T *sl_compprog; /* COMPOUNDFLAGS turned into a regexp progrm
|
|
|
381 * (NULL when no compounding) */
|
|
|
382 char_u *sl_compstartflags; /* flags for first compound word */
|
|
485
|
383 char_u *sl_compallflags; /* all flags for compound words */
|
|
492
|
384 char_u sl_nobreak; /* When TRUE: no spaces between words */
|
|
483
|
385 char_u *sl_syllable; /* SYLLABLE repeatable chars or NULL */
|
|
|
386 garray_T sl_syl_items; /* syllable items */
|
|
481
|
387
|
|
339
|
388 int sl_prefixcnt; /* number of items in "sl_prefprog" */
|
|
|
389 regprog_T **sl_prefprog; /* table with regprogs for prefixes */
|
|
|
390
|
|
323
|
391 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
|
392 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
|
393 there is none */
|
|
344
|
394 garray_T sl_sal; /* list of salitem_T entries from SAL lines */
|
|
375
|
395 salfirst_T sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
323
|
396 there is none */
|
|
375
|
397 int sl_sofo; /* SOFOFROM and SOFOTO instead of SAL items:
|
|
|
398 * "sl_sal_first" maps chars, when has_mbyte
|
|
|
399 * "sl_sal" is a list of wide char lists. */
|
|
323
|
400 int sl_followup; /* SAL followup */
|
|
|
401 int sl_collapse; /* SAL collapse_result */
|
|
|
402 int sl_rem_accents; /* SAL remove_accents */
|
|
330
|
403 int sl_has_map; /* TRUE if there is a MAP line */
|
|
|
404 #ifdef FEAT_MBYTE
|
|
|
405 hashtab_T sl_map_hash; /* MAP for multi-byte chars */
|
|
|
406 int sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
407 #else
|
|
|
408 char_u sl_map_array[256]; /* MAP for first 256 chars */
|
|
|
409 #endif
|
|
236
|
410 };
|
|
|
411
|
|
243
|
412 /* First language that is loaded, start of the linked list of loaded
|
|
|
413 * languages. */
|
|
236
|
414 static slang_T *first_lang = NULL;
|
|
|
415
|
|
323
|
416 /* Flags used in .spl file for soundsalike flags. */
|
|
|
417 #define SAL_F0LLOWUP 1
|
|
|
418 #define SAL_COLLAPSE 2
|
|
|
419 #define SAL_REM_ACCENTS 4
|
|
|
420
|
|
236
|
421 /*
|
|
|
422 * Structure used in "b_langp", filled from 'spelllang'.
|
|
|
423 */
|
|
|
424 typedef struct langp_S
|
|
|
425 {
|
|
503
|
426 slang_T *lp_slang; /* info for this language */
|
|
|
427 slang_T *lp_sallang; /* language used for sound folding or NULL */
|
|
|
428 slang_T *lp_replang; /* language used for REP items or NULL */
|
|
236
|
429 int lp_region; /* bitmask for region or REGION_ALL */
|
|
|
430 } langp_T;
|
|
|
431
|
|
|
432 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
|
433
|
|
307
|
434 #define REGION_ALL 0xff /* word valid in all regions */
|
|
|
435
|
|
483
|
436 #define VIMSPELLMAGIC "VIMspell" /* string at start of Vim spell file */
|
|
|
437 #define VIMSPELLMAGICL 8
|
|
|
438 #define VIMSPELLVERSION 50
|
|
|
439
|
|
|
440 /* Section IDs. Only renumber them when VIMSPELLVERSION changes! */
|
|
|
441 #define SN_REGION 0 /* <regionname> section */
|
|
|
442 #define SN_CHARFLAGS 1 /* charflags section */
|
|
|
443 #define SN_MIDWORD 2 /* <midword> section */
|
|
|
444 #define SN_PREFCOND 3 /* <prefcond> section */
|
|
|
445 #define SN_REP 4 /* REP items section */
|
|
|
446 #define SN_SAL 5 /* SAL items section */
|
|
|
447 #define SN_SOFO 6 /* soundfolding section */
|
|
|
448 #define SN_MAP 7 /* MAP items section */
|
|
|
449 #define SN_COMPOUND 8 /* compound words section */
|
|
|
450 #define SN_SYLLABLE 9 /* syllable section */
|
|
492
|
451 #define SN_NOBREAK 10 /* NOBREAK section */
|
|
483
|
452 #define SN_END 255 /* end of sections */
|
|
|
453
|
|
|
454 #define SNF_REQUIRED 1 /* <sectionflags>: required section */
|
|
|
455
|
|
307
|
456 /* Result values. Lower number is accepted over higher one. */
|
|
|
457 #define SP_BANNED -1
|
|
236
|
458 #define SP_OK 0
|
|
307
|
459 #define SP_RARE 1
|
|
|
460 #define SP_LOCAL 2
|
|
|
461 #define SP_BAD 3
|
|
236
|
462
|
|
381
|
463 /* file used for "zG" and "zW" */
|
|
385
|
464 static char_u *int_wordlist = NULL;
|
|
381
|
465
|
|
236
|
466 /*
|
|
323
|
467 * Information used when looking for suggestions.
|
|
|
468 */
|
|
|
469 typedef struct suginfo_S
|
|
|
470 {
|
|
|
471 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
344
|
472 int su_maxcount; /* max. number of suggestions displayed */
|
|
323
|
473 int su_maxscore; /* maximum score for adding to su_ga */
|
|
344
|
474 garray_T su_sga; /* like su_ga, sound-folded scoring */
|
|
323
|
475 char_u *su_badptr; /* start of bad word in line */
|
|
|
476 int su_badlen; /* length of detected bad word in line */
|
|
346
|
477 int su_badflags; /* caps flags for bad word */
|
|
323
|
478 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
|
479 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
534
|
480 char_u su_sal_badword[MAXWLEN]; /* su_badword soundfolded */
|
|
|
481 slang_T *su_slang_first; /* slang_T used for su_sal_badword */
|
|
323
|
482 hashtab_T su_banned; /* table with banned words */
|
|
503
|
483 slang_T *su_sallang; /* default language for sound folding */
|
|
323
|
484 } suginfo_T;
|
|
|
485
|
|
|
486 /* One word suggestion. Used in "si_ga". */
|
|
|
487 typedef struct suggest_S
|
|
|
488 {
|
|
|
489 char_u *st_word; /* suggested word, allocated string */
|
|
|
490 int st_orglen; /* length of replaced text */
|
|
|
491 int st_score; /* lower is better */
|
|
344
|
492 int st_altscore; /* used when st_score compares equal */
|
|
|
493 int st_salscore; /* st_score is for soundalike */
|
|
324
|
494 int st_had_bonus; /* bonus already included in score */
|
|
503
|
495 slang_T *st_slang; /* language used for sound folding */
|
|
323
|
496 } suggest_T;
|
|
|
497
|
|
344
|
498 #define SUG(ga, i) (((suggest_T *)(ga).ga_data)[i])
|
|
323
|
499
|
|
324
|
500 /* Number of suggestions kept when cleaning up. When rescore_suggestions() is
|
|
|
501 * called the score may change, thus we need to keep more than what is
|
|
|
502 * displayed. */
|
|
351
|
503 #define SUG_CLEAN_COUNT(su) ((su)->su_maxcount < 50 ? 50 : (su)->su_maxcount)
|
|
324
|
504
|
|
|
505 /* Threshold for sorting and cleaning up suggestions. Don't want to keep lots
|
|
|
506 * of suggestions that are not going to be displayed. */
|
|
344
|
507 #define SUG_MAX_COUNT(su) ((su)->su_maxcount + 50)
|
|
323
|
508
|
|
|
509 /* score for various changes */
|
|
344
|
510 #define SCORE_SPLIT 149 /* split bad word */
|
|
323
|
511 #define SCORE_ICASE 52 /* slightly different case */
|
|
483
|
512 #define SCORE_REGION 200 /* word is for different region */
|
|
323
|
513 #define SCORE_RARE 180 /* rare word */
|
|
|
514 #define SCORE_SWAP 90 /* swap two characters */
|
|
|
515 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
532
|
516 #define SCORE_REP 65 /* REP replacement */
|
|
323
|
517 #define SCORE_SUBST 93 /* substitute a character */
|
|
|
518 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
479
|
519 #define SCORE_SUBCOMP 33 /* substitute a composing character */
|
|
324
|
520 #define SCORE_DEL 94 /* delete a character */
|
|
532
|
521 #define SCORE_DELDUP 66 /* delete a duplicated character */
|
|
479
|
522 #define SCORE_DELCOMP 28 /* delete a composing character */
|
|
324
|
523 #define SCORE_INS 96 /* insert a character */
|
|
532
|
524 #define SCORE_INSDUP 67 /* insert a duplicate character */
|
|
475
|
525 #define SCORE_INSCOMP 30 /* insert a composing character */
|
|
366
|
526 #define SCORE_NONWORD 103 /* change non-word to word char */
|
|
323
|
527
|
|
372
|
528 #define SCORE_FILE 30 /* suggestion from a file */
|
|
323
|
529 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
|
530 * 350 allows for about three changes. */
|
|
344
|
531
|
|
|
532 #define SCORE_BIG SCORE_INS * 3 /* big difference */
|
|
323
|
533 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
|
534
|
|
|
535 /*
|
|
236
|
536 * Structure to store info for word matching.
|
|
|
537 */
|
|
|
538 typedef struct matchinf_S
|
|
|
539 {
|
|
|
540 langp_T *mi_lp; /* info for language and region */
|
|
243
|
541
|
|
|
542 /* pointers to original text to be checked */
|
|
236
|
543 char_u *mi_word; /* start of word being checked */
|
|
339
|
544 char_u *mi_end; /* end of matching word so far */
|
|
243
|
545 char_u *mi_fend; /* next char to be added to mi_fword */
|
|
300
|
546 char_u *mi_cend; /* char after what was used for
|
|
|
547 mi_capflags */
|
|
243
|
548
|
|
|
549 /* case-folded text */
|
|
|
550 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
|
300
|
551 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
|
243
|
552
|
|
339
|
553 /* for when checking word after a prefix */
|
|
|
554 int mi_prefarridx; /* index in sl_pidxs with list of
|
|
481
|
555 affixID/condition */
|
|
339
|
556 int mi_prefcnt; /* number of entries at mi_prefarridx */
|
|
|
557 int mi_prefixlen; /* byte length of prefix */
|
|
455
|
558 #ifdef FEAT_MBYTE
|
|
|
559 int mi_cprefixlen; /* byte length of prefix in original
|
|
|
560 case */
|
|
|
561 #else
|
|
|
562 # define mi_cprefixlen mi_prefixlen /* it's the same value */
|
|
|
563 #endif
|
|
339
|
564
|
|
481
|
565 /* for when checking a compound word */
|
|
|
566 int mi_compoff; /* start of following word offset */
|
|
483
|
567 char_u mi_compflags[MAXWLEN]; /* flags for compound words used */
|
|
|
568 int mi_complen; /* nr of compound words used */
|
|
481
|
569
|
|
243
|
570 /* others */
|
|
236
|
571 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
|
300
|
572 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
|
376
|
573 buf_T *mi_buf; /* buffer being checked */
|
|
492
|
574
|
|
|
575 /* for NOBREAK */
|
|
|
576 int mi_result2; /* "mi_resul" without following word */
|
|
|
577 char_u *mi_end2; /* "mi_end" without following word */
|
|
236
|
578 } matchinf_T;
|
|
|
579
|
|
307
|
580 /*
|
|
|
581 * The tables used for recognizing word characters according to spelling.
|
|
|
582 * These are only used for the first 256 characters of 'encoding'.
|
|
|
583 */
|
|
|
584 typedef struct spelltab_S
|
|
|
585 {
|
|
|
586 char_u st_isw[256]; /* flags: is word char */
|
|
|
587 char_u st_isu[256]; /* flags: is uppercase char */
|
|
|
588 char_u st_fold[256]; /* chars: folded case */
|
|
324
|
589 char_u st_upper[256]; /* chars: upper case */
|
|
307
|
590 } spelltab_T;
|
|
|
591
|
|
|
592 static spelltab_T spelltab;
|
|
|
593 static int did_set_spelltab;
|
|
|
594
|
|
324
|
595 #define CF_WORD 0x01
|
|
|
596 #define CF_UPPER 0x02
|
|
307
|
597
|
|
|
598 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
|
599 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
376
|
600 static int spell_iswordp __ARGS((char_u *p, buf_T *buf));
|
|
|
601 static int spell_iswordp_nmw __ARGS((char_u *p));
|
|
|
602 #ifdef FEAT_MBYTE
|
|
|
603 static int spell_iswordp_w __ARGS((int *p, buf_T *buf));
|
|
|
604 #endif
|
|
483
|
605 static int write_spell_prefcond __ARGS((FILE *fd, garray_T *gap));
|
|
307
|
606
|
|
|
607 /*
|
|
344
|
608 * For finding suggestions: At each node in the tree these states are tried:
|
|
330
|
609 */
|
|
|
610 typedef enum
|
|
|
611 {
|
|
344
|
612 STATE_START = 0, /* At start of node check for NUL bytes (goodword
|
|
|
613 * ends); if badword ends there is a match, otherwise
|
|
|
614 * try splitting word. */
|
|
375
|
615 STATE_NOPREFIX, /* try without prefix */
|
|
344
|
616 STATE_SPLITUNDO, /* Undo splitting. */
|
|
330
|
617 STATE_ENDNUL, /* Past NUL bytes at start of the node. */
|
|
|
618 STATE_PLAIN, /* Use each byte of the node. */
|
|
|
619 STATE_DEL, /* Delete a byte from the bad word. */
|
|
|
620 STATE_INS, /* Insert a byte in the bad word. */
|
|
|
621 STATE_SWAP, /* Swap two bytes. */
|
|
344
|
622 STATE_UNSWAP, /* Undo swap two characters. */
|
|
|
623 STATE_SWAP3, /* Swap two characters over three. */
|
|
|
624 STATE_UNSWAP3, /* Undo Swap two characters over three. */
|
|
|
625 STATE_UNROT3L, /* Undo rotate three characters left */
|
|
|
626 STATE_UNROT3R, /* Undo rotate three characters right */
|
|
330
|
627 STATE_REP_INI, /* Prepare for using REP items. */
|
|
|
628 STATE_REP, /* Use matching REP items from the .aff file. */
|
|
|
629 STATE_REP_UNDO, /* Undo a REP item replacement. */
|
|
|
630 STATE_FINAL /* End of this node. */
|
|
|
631 } state_T;
|
|
|
632
|
|
|
633 /*
|
|
346
|
634 * Struct to keep the state at each level in suggest_try_change().
|
|
323
|
635 */
|
|
|
636 typedef struct trystate_S
|
|
|
637 {
|
|
330
|
638 state_T ts_state; /* state at this level, STATE_ */
|
|
323
|
639 int ts_score; /* score */
|
|
344
|
640 idx_T ts_arridx; /* index in tree array, start of node */
|
|
330
|
641 short ts_curi; /* index in list of child nodes */
|
|
|
642 char_u ts_fidx; /* index in fword[], case-folded bad word */
|
|
|
643 char_u ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
|
644 char_u ts_twordlen; /* valid length of tword[] */
|
|
482
|
645 char_u ts_prefixdepth; /* stack depth for end of prefix or
|
|
485
|
646 * PFD_PREFIXTREE or PFD_NOPREFIX */
|
|
|
647 char_u ts_flags; /* TSF_ flags */
|
|
330
|
648 #ifdef FEAT_MBYTE
|
|
|
649 char_u ts_tcharlen; /* number of bytes in tword character */
|
|
|
650 char_u ts_tcharidx; /* current byte index in tword character */
|
|
|
651 char_u ts_isdiff; /* DIFF_ values */
|
|
|
652 char_u ts_fcharstart; /* index in fword where badword char started */
|
|
|
653 #endif
|
|
483
|
654 char_u ts_prewordlen; /* length of word in "preword[]" */
|
|
|
655 char_u ts_splitoff; /* index in "tword" after last split */
|
|
492
|
656 char_u ts_splitfidx; /* "ts_fidx" at word split */
|
|
483
|
657 char_u ts_complen; /* nr of compound words used */
|
|
485
|
658 char_u ts_compsplit; /* index for "compflags" where word was spit */
|
|
346
|
659 char_u ts_save_badflags; /* su_badflags saved here */
|
|
323
|
660 } trystate_T;
|
|
|
661
|
|
330
|
662 /* values for ts_isdiff */
|
|
|
663 #define DIFF_NONE 0 /* no different byte (yet) */
|
|
|
664 #define DIFF_YES 1 /* different byte found */
|
|
|
665 #define DIFF_INSERT 2 /* inserting character */
|
|
|
666
|
|
485
|
667 /* values for ts_flags */
|
|
|
668 #define TSF_PREFIXOK 1 /* already checked that prefix is OK */
|
|
|
669 #define TSF_DIDSPLIT 2 /* tried split at this point */
|
|
|
670
|
|
375
|
671 /* special values ts_prefixdepth */
|
|
485
|
672 #define PFD_NOPREFIX 0xff /* not using prefixes */
|
|
482
|
673 #define PFD_PREFIXTREE 0xfe /* walking through the prefix tree */
|
|
485
|
674 #define PFD_NOTSPECIAL 0xfd /* first value that's not special */
|
|
375
|
675
|
|
339
|
676 /* mode values for find_word */
|
|
481
|
677 #define FIND_FOLDWORD 0 /* find word case-folded */
|
|
|
678 #define FIND_KEEPWORD 1 /* find keep-case word */
|
|
|
679 #define FIND_PREFIX 2 /* find word after prefix */
|
|
|
680 #define FIND_COMPOUND 3 /* find case-folded compound word */
|
|
|
681 #define FIND_KEEPCOMPOUND 4 /* find keep-case compound word */
|
|
339
|
682
|
|
236
|
683 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
|
684 static void slang_free __ARGS((slang_T *lp));
|
|
310
|
685 static void slang_clear __ARGS((slang_T *lp));
|
|
339
|
686 static void find_word __ARGS((matchinf_T *mip, int mode));
|
|
483
|
687 static int can_compound __ARGS((slang_T *slang, char_u *word, char_u *flags));
|
|
455
|
688 static int valid_word_prefix __ARGS((int totprefcnt, int arridx, int flags, char_u *word, slang_T *slang, int cond_req));
|
|
485
|
689 static void find_prefix __ARGS((matchinf_T *mip, int mode));
|
|
339
|
690 static int fold_more __ARGS((matchinf_T *mip));
|
|
389
|
691 static int spell_valid_case __ARGS((int wordflags, int treeflags));
|
|
498
|
692 static int no_spell_checking __ARGS((win_T *wp));
|
|
307
|
693 static void spell_load_lang __ARGS((char_u *lang));
|
|
310
|
694 static char_u *spell_enc __ARGS((void));
|
|
385
|
695 static void int_wordlist_spl __ARGS((char_u *fname));
|
|
310
|
696 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
|
323
|
697 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
|
389
|
698 static char_u *read_cnt_string __ARGS((FILE *fd, int cnt_bytes, int *lenp));
|
|
483
|
699 static char_u *read_string __ARGS((FILE *fd, int cnt));
|
|
|
700 static int read_region_section __ARGS((FILE *fd, slang_T *slang, int len));
|
|
|
701 static int read_charflags_section __ARGS((FILE *fd));
|
|
|
702 static int read_prefcond_section __ARGS((FILE *fd, slang_T *lp));
|
|
|
703 static int read_rep_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
704 static int read_sal_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
705 static int read_sofo_section __ARGS((FILE *fd, slang_T *slang));
|
|
|
706 static int read_compound __ARGS((FILE *fd, slang_T *slang, int len));
|
|
495
|
707 static int byte_in_str __ARGS((char_u *str, int byte));
|
|
483
|
708 static int init_syl_tab __ARGS((slang_T *slang));
|
|
|
709 static int count_syllables __ARGS((slang_T *slang, char_u *word));
|
|
381
|
710 static int set_sofo __ARGS((slang_T *lp, char_u *from, char_u *to));
|
|
|
711 static void set_sal_first __ARGS((slang_T *lp));
|
|
372
|
712 #ifdef FEAT_MBYTE
|
|
|
713 static int *mb_str2wide __ARGS((char_u *s));
|
|
|
714 #endif
|
|
339
|
715 static idx_T read_tree __ARGS((FILE *fd, char_u *byts, idx_T *idxs, int maxidx, int startidx, int prefixtree, int maxprefcondnr));
|
|
376
|
716 static void clear_midword __ARGS((buf_T *buf));
|
|
|
717 static void use_midword __ARGS((slang_T *lp, buf_T *buf));
|
|
236
|
718 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
|
719 static int captype __ARGS((char_u *word, char_u *end));
|
|
474
|
720 static int badword_captype __ARGS((char_u *word, char_u *end));
|
|
323
|
721 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
|
483
|
722 static void set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
307
|
723 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
|
724 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
|
475
|
725 static int check_need_cap __ARGS((linenr_T lnum, colnr_T col));
|
|
388
|
726 static void spell_find_suggest __ARGS((char_u *badptr, suginfo_T *su, int maxcount, int banbadword, int need_cap));
|
|
372
|
727 #ifdef FEAT_EVAL
|
|
|
728 static void spell_suggest_expr __ARGS((suginfo_T *su, char_u *expr));
|
|
|
729 #endif
|
|
|
730 static void spell_suggest_file __ARGS((suginfo_T *su, char_u *fname));
|
|
|
731 static void spell_suggest_intern __ARGS((suginfo_T *su));
|
|
344
|
732 static void spell_find_cleanup __ARGS((suginfo_T *su));
|
|
324
|
733 static void onecap_copy __ARGS((char_u *word, char_u *wcopy, int upper));
|
|
344
|
734 static void allcap_copy __ARGS((char_u *word, char_u *wcopy));
|
|
346
|
735 static void suggest_try_special __ARGS((suginfo_T *su));
|
|
|
736 static void suggest_try_change __ARGS((suginfo_T *su));
|
|
323
|
737 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
455
|
738 #ifdef FEAT_MBYTE
|
|
|
739 static int nofold_len __ARGS((char_u *fword, int flen, char_u *word));
|
|
|
740 #endif
|
|
323
|
741 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
344
|
742 static void score_comp_sal __ARGS((suginfo_T *su));
|
|
|
743 static void score_combine __ARGS((suginfo_T *su));
|
|
351
|
744 static int stp_sal_score __ARGS((suggest_T *stp, suginfo_T *su, slang_T *slang, char_u *badsound));
|
|
346
|
745 static void suggest_try_soundalike __ARGS((suginfo_T *su));
|
|
323
|
746 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
330
|
747 static void set_map_str __ARGS((slang_T *lp, char_u *map));
|
|
323
|
748 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
503
|
749 static void add_suggestion __ARGS((suginfo_T *su, garray_T *gap, char_u *goodword, int badlen, int score, int altscore, int had_bonus, slang_T *slang));
|
|
323
|
750 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
751 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
|
752 static void free_banned __ARGS((suginfo_T *su));
|
|
324
|
753 static void rescore_suggestions __ARGS((suginfo_T *su));
|
|
534
|
754 static void rescore_one __ARGS((suginfo_T *su, suggest_T *stp));
|
|
344
|
755 static int cleanup_suggestions __ARGS((garray_T *gap, int maxscore, int keep));
|
|
375
|
756 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, int folded, char_u *res));
|
|
|
757 static void spell_soundfold_sofo __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
|
758 static void spell_soundfold_sal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
759 #ifdef FEAT_MBYTE
|
|
375
|
760 static void spell_soundfold_wsal __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
372
|
761 #endif
|
|
344
|
762 static int soundalike_score __ARGS((char_u *goodsound, char_u *badsound));
|
|
323
|
763 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
351
|
764 static void dump_word __ARGS((char_u *word, int round, int flags, linenr_T lnum));
|
|
500
|
765 static linenr_T dump_prefixes __ARGS((slang_T *slang, char_u *word, int round, int flags, linenr_T startlnum));
|
|
323
|
766
|
|
324
|
767 /*
|
|
|
768 * Use our own character-case definitions, because the current locale may
|
|
|
769 * differ from what the .spl file uses.
|
|
|
770 * These must not be called with negative number!
|
|
|
771 */
|
|
|
772 #ifndef FEAT_MBYTE
|
|
|
773 /* Non-multi-byte implementation. */
|
|
|
774 # define SPELL_TOFOLD(c) ((c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
775 # define SPELL_TOUPPER(c) ((c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
776 # define SPELL_ISUPPER(c) ((c) < 256 ? spelltab.st_isu[c] : FALSE)
|
|
|
777 #else
|
|
447
|
778 # if defined(HAVE_WCHAR_H)
|
|
|
779 # include <wchar.h> /* for towupper() and towlower() */
|
|
|
780 # endif
|
|
324
|
781 /* Multi-byte implementation. For Unicode we can call utf_*(), but don't do
|
|
|
782 * that for ASCII, because we don't want to use 'casemap' here. Otherwise use
|
|
|
783 * the "w" library function for characters above 255 if available. */
|
|
|
784 # ifdef HAVE_TOWLOWER
|
|
|
785 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
786 : (c) < 256 ? spelltab.st_fold[c] : towlower(c))
|
|
|
787 # else
|
|
|
788 # define SPELL_TOFOLD(c) (enc_utf8 && (c) >= 128 ? utf_fold(c) \
|
|
|
789 : (c) < 256 ? spelltab.st_fold[c] : (c))
|
|
|
790 # endif
|
|
|
791
|
|
|
792 # ifdef HAVE_TOWUPPER
|
|
|
793 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
794 : (c) < 256 ? spelltab.st_upper[c] : towupper(c))
|
|
|
795 # else
|
|
|
796 # define SPELL_TOUPPER(c) (enc_utf8 && (c) >= 128 ? utf_toupper(c) \
|
|
|
797 : (c) < 256 ? spelltab.st_upper[c] : (c))
|
|
|
798 # endif
|
|
|
799
|
|
|
800 # ifdef HAVE_ISWUPPER
|
|
|
801 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
|
802 : (c) < 256 ? spelltab.st_isu[c] : iswupper(c))
|
|
|
803 # else
|
|
|
804 # define SPELL_ISUPPER(c) (enc_utf8 && (c) >= 128 ? utf_isupper(c) \
|
|
474
|
805 : (c) < 256 ? spelltab.st_isu[c] : (FALSE))
|
|
324
|
806 # endif
|
|
|
807 #endif
|
|
|
808
|
|
307
|
809
|
|
|
810 static char *e_format = N_("E759: Format error in spell file");
|
|
381
|
811 static char *e_spell_trunc = N_("E758: Truncated spell file");
|
|
482
|
812 static char *e_afftrailing = N_("Trailing text in %s line %d: %s");
|
|
495
|
813 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
|
|
814 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
|
815 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
469
|
816 static char *msg_compressing = N_("Compressing word tree...");
|
|
236
|
817
|
|
|
818 /*
|
|
|
819 * Main spell-checking function.
|
|
300
|
820 * "ptr" points to a character that could be the start of a word.
|
|
534
|
821 * "*attrp" is set to the highlight index for a badly spelled word. For a
|
|
|
822 * non-word or when it's OK it remains unchanged.
|
|
236
|
823 * This must only be called when 'spelllang' is not empty.
|
|
323
|
824 *
|
|
385
|
825 * "capcol" is used to check for a Capitalised word after the end of a
|
|
|
826 * sentence. If it's zero then perform the check. Return the column where to
|
|
|
827 * check next, or -1 when no sentence end was found. If it's NULL then don't
|
|
|
828 * worry.
|
|
323
|
829 *
|
|
236
|
830 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
|
831 * caller can skip over the word.
|
|
|
832 */
|
|
|
833 int
|
|
385
|
834 spell_check(wp, ptr, attrp, capcol)
|
|
236
|
835 win_T *wp; /* current window */
|
|
|
836 char_u *ptr;
|
|
534
|
837 hlf_T *attrp;
|
|
385
|
838 int *capcol; /* column to check for Capital */
|
|
236
|
839 {
|
|
|
840 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
|
841 be passed to functions quickly. */
|
|
344
|
842 int nrlen = 0; /* found a number first */
|
|
385
|
843 int c;
|
|
483
|
844 int wrongcaplen = 0;
|
|
500
|
845 int lpi;
|
|
236
|
846
|
|
307
|
847 /* A word never starts at a space or a control character. Return quickly
|
|
|
848 * then, skipping over the character. */
|
|
|
849 if (*ptr <= ' ')
|
|
|
850 return 1;
|
|
483
|
851 vim_memset(&mi, 0, sizeof(matchinf_T));
|
|
236
|
852
|
|
344
|
853 /* A number is always OK. Also skip hexadecimal numbers 0xFF99 and
|
|
584
|
854 * 0X99FF. But always do check spelling to find "3GPP" and "11
|
|
|
855 * julifeest". */
|
|
300
|
856 if (*ptr >= '0' && *ptr <= '9')
|
|
|
857 {
|
|
316
|
858 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
|
859 mi.mi_end = skiphex(ptr + 2);
|
|
300
|
860 else
|
|
344
|
861 mi.mi_end = skipdigits(ptr);
|
|
584
|
862 nrlen = mi.mi_end - ptr;
|
|
|
863 }
|
|
346
|
864
|
|
|
865 /* Find the normal end of the word (until the next non-word character). */
|
|
344
|
866 mi.mi_word = ptr;
|
|
584
|
867 mi.mi_fend = ptr;
|
|
376
|
868 if (spell_iswordp(mi.mi_fend, wp->w_buffer))
|
|
344
|
869 {
|
|
|
870 do
|
|
|
871 {
|
|
307
|
872 mb_ptr_adv(mi.mi_fend);
|
|
376
|
873 } while (*mi.mi_fend != NUL && spell_iswordp(mi.mi_fend, wp->w_buffer));
|
|
385
|
874
|
|
|
875 if (capcol != NULL && *capcol == 0 && wp->w_buffer->b_cap_prog != NULL)
|
|
|
876 {
|
|
|
877 /* Check word starting with capital letter. */
|
|
455
|
878 c = PTR2CHAR(ptr);
|
|
385
|
879 if (!SPELL_ISUPPER(c))
|
|
483
|
880 wrongcaplen = (int)(mi.mi_fend - ptr);
|
|
385
|
881 }
|
|
|
882 }
|
|
|
883 if (capcol != NULL)
|
|
|
884 *capcol = -1;
|
|
344
|
885
|
|
|
886 /* We always use the characters up to the next non-word character,
|
|
|
887 * also for bad words. */
|
|
|
888 mi.mi_end = mi.mi_fend;
|
|
|
889
|
|
|
890 /* Check caps type later. */
|
|
376
|
891 mi.mi_buf = wp->w_buffer;
|
|
344
|
892
|
|
483
|
893 /* case-fold the word with one non-word character, so that we can check
|
|
|
894 * for the word end. */
|
|
344
|
895 if (*mi.mi_fend != NUL)
|
|
|
896 mb_ptr_adv(mi.mi_fend);
|
|
|
897
|
|
|
898 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
|
899 MAXWLEN + 1);
|
|
|
900 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
|
901
|
|
|
902 /* The word is bad unless we recognize it. */
|
|
|
903 mi.mi_result = SP_BAD;
|
|
492
|
904 mi.mi_result2 = SP_BAD;
|
|
344
|
905
|
|
|
906 /*
|
|
|
907 * Loop over the languages specified in 'spelllang'.
|
|
|
908 * We check them all, because a matching word may be longer than an
|
|
|
909 * already found matching word.
|
|
|
910 */
|
|
500
|
911 for (lpi = 0; lpi < wp->w_buffer->b_langp.ga_len; ++lpi)
|
|
|
912 {
|
|
|
913 mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, lpi);
|
|
|
914
|
|
|
915 /* If reloading fails the language is still in the list but everything
|
|
|
916 * has been cleared. */
|
|
|
917 if (mi.mi_lp->lp_slang->sl_fidxs == NULL)
|
|
|
918 continue;
|
|
|
919
|
|
344
|
920 /* Check for a matching word in case-folded words. */
|
|
|
921 find_word(&mi, FIND_FOLDWORD);
|
|
|
922
|
|
|
923 /* Check for a matching word in keep-case words. */
|
|
|
924 find_word(&mi, FIND_KEEPWORD);
|
|
|
925
|
|
|
926 /* Check for matching prefixes. */
|
|
485
|
927 find_prefix(&mi, FIND_FOLDWORD);
|
|
492
|
928
|
|
|
929 /* For a NOBREAK language, may want to use a word without a following
|
|
|
930 * word as a backup. */
|
|
|
931 if (mi.mi_lp->lp_slang->sl_nobreak && mi.mi_result == SP_BAD
|
|
|
932 && mi.mi_result2 != SP_BAD)
|
|
|
933 {
|
|
|
934 mi.mi_result = mi.mi_result2;
|
|
|
935 mi.mi_end = mi.mi_end2;
|
|
|
936 }
|
|
344
|
937 }
|
|
|
938
|
|
|
939 if (mi.mi_result != SP_OK)
|
|
|
940 {
|
|
346
|
941 /* If we found a number skip over it. Allows for "42nd". Do flag
|
|
|
942 * rare and local words, e.g., "3GPP". */
|
|
344
|
943 if (nrlen > 0)
|
|
346
|
944 {
|
|
|
945 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
|
946 return nrlen;
|
|
|
947 }
|
|
344
|
948
|
|
|
949 /* When we are at a non-word character there is no error, just
|
|
|
950 * skip over the character (try looking for a word after it). */
|
|
390
|
951 else if (!spell_iswordp_nmw(ptr))
|
|
243
|
952 {
|
|
385
|
953 if (capcol != NULL && wp->w_buffer->b_cap_prog != NULL)
|
|
|
954 {
|
|
|
955 regmatch_T regmatch;
|
|
|
956
|
|
|
957 /* Check for end of sentence. */
|
|
|
958 regmatch.regprog = wp->w_buffer->b_cap_prog;
|
|
|
959 regmatch.rm_ic = FALSE;
|
|
|
960 if (vim_regexec(®match, ptr, 0))
|
|
|
961 *capcol = (int)(regmatch.endp[0] - ptr);
|
|
|
962 }
|
|
|
963
|
|
344
|
964 #ifdef FEAT_MBYTE
|
|
|
965 if (has_mbyte)
|
|
474
|
966 return (*mb_ptr2len)(ptr);
|
|
344
|
967 #endif
|
|
|
968 return 1;
|
|
300
|
969 }
|
|
483
|
970 else if (mi.mi_end == ptr)
|
|
|
971 /* Always include at least one character. Required for when there
|
|
|
972 * is a mixup in "midword". */
|
|
|
973 mb_ptr_adv(mi.mi_end);
|
|
492
|
974 else if (mi.mi_result == SP_BAD
|
|
|
975 && LANGP_ENTRY(wp->w_buffer->b_langp, 0)->lp_slang->sl_nobreak)
|
|
|
976 {
|
|
|
977 char_u *p, *fp;
|
|
|
978 int save_result = mi.mi_result;
|
|
|
979
|
|
|
980 /* First language in 'spelllang' is NOBREAK. Find first position
|
|
|
981 * at which any word would be valid. */
|
|
|
982 mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
500
|
983 if (mi.mi_lp->lp_slang->sl_fidxs != NULL)
|
|
|
984 {
|
|
|
985 p = mi.mi_word;
|
|
|
986 fp = mi.mi_fword;
|
|
|
987 for (;;)
|
|
|
988 {
|
|
|
989 mb_ptr_adv(p);
|
|
|
990 mb_ptr_adv(fp);
|
|
|
991 if (p >= mi.mi_end)
|
|
|
992 break;
|
|
|
993 mi.mi_compoff = fp - mi.mi_fword;
|
|
|
994 find_word(&mi, FIND_COMPOUND);
|
|
|
995 if (mi.mi_result != SP_BAD)
|
|
|
996 {
|
|
|
997 mi.mi_end = p;
|
|
|
998 break;
|
|
|
999 }
|
|
|
1000 }
|
|
|
1001 mi.mi_result = save_result;
|
|
|
1002 }
|
|
492
|
1003 }
|
|
243
|
1004
|
|
344
|
1005 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
|
534
|
1006 *attrp = HLF_SPB;
|
|
344
|
1007 else if (mi.mi_result == SP_RARE)
|
|
534
|
1008 *attrp = HLF_SPR;
|
|
344
|
1009 else
|
|
534
|
1010 *attrp = HLF_SPL;
|
|
243
|
1011 }
|
|
|
1012
|
|
483
|
1013 if (wrongcaplen > 0 && (mi.mi_result == SP_OK || mi.mi_result == SP_RARE))
|
|
|
1014 {
|
|
|
1015 /* Report SpellCap only when the word isn't badly spelled. */
|
|
534
|
1016 *attrp = HLF_SPC;
|
|
483
|
1017 return wrongcaplen;
|
|
|
1018 }
|
|
|
1019
|
|
300
|
1020 return (int)(mi.mi_end - ptr);
|
|
236
|
1021 }
|
|
|
1022
|
|
|
1023 /*
|
|
300
|
1024 * Check if the word at "mip->mi_word" is in the tree.
|
|
339
|
1025 * When "mode" is FIND_FOLDWORD check in fold-case word tree.
|
|
|
1026 * When "mode" is FIND_KEEPWORD check in keep-case word tree.
|
|
|
1027 * When "mode" is FIND_PREFIX check for word after prefix in fold-case word
|
|
|
1028 * tree.
|
|
300
|
1029 *
|
|
|
1030 * For a match mip->mi_result is updated.
|
|
243
|
1031 */
|
|
|
1032 static void
|
|
339
|
1033 find_word(mip, mode)
|
|
243
|
1034 matchinf_T *mip;
|
|
339
|
1035 int mode;
|
|
243
|
1036 {
|
|
324
|
1037 idx_T arridx = 0;
|
|
300
|
1038 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
324
|
1039 idx_T endidx[MAXWLEN]; /* possible word endings */
|
|
300
|
1040 int endidxcnt = 0;
|
|
|
1041 int len;
|
|
|
1042 int wlen = 0;
|
|
|
1043 int flen;
|
|
|
1044 int c;
|
|
|
1045 char_u *ptr;
|
|
324
|
1046 idx_T lo, hi, m;
|
|
243
|
1047 #ifdef FEAT_MBYTE
|
|
300
|
1048 char_u *s;
|
|
490
|
1049 #endif
|
|
339
|
1050 char_u *p;
|
|
307
|
1051 int res = SP_BAD;
|
|
300
|
1052 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1053 unsigned flags;
|
|
|
1054 char_u *byts;
|
|
324
|
1055 idx_T *idxs;
|
|
481
|
1056 int word_ends;
|
|
485
|
1057 int prefix_found;
|
|
492
|
1058 int nobreak_result;
|
|
481
|
1059
|
|
|
1060 if (mode == FIND_KEEPWORD || mode == FIND_KEEPCOMPOUND)
|
|
236
|
1061 {
|
|
300
|
1062 /* Check for word with matching case in keep-case tree. */
|
|
|
1063 ptr = mip->mi_word;
|
|
|
1064 flen = 9999; /* no case folding, always enough bytes */
|
|
|
1065 byts = slang->sl_kbyts;
|
|
|
1066 idxs = slang->sl_kidxs;
|
|
481
|
1067
|
|
|
1068 if (mode == FIND_KEEPCOMPOUND)
|
|
|
1069 /* Skip over the previously found word(s). */
|
|
|
1070 wlen += mip->mi_compoff;
|
|
236
|
1071 }
|
|
|
1072 else
|
|
|
1073 {
|
|
300
|
1074 /* Check for case-folded in case-folded tree. */
|
|
|
1075 ptr = mip->mi_fword;
|
|
|
1076 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
|
1077 byts = slang->sl_fbyts;
|
|
|
1078 idxs = slang->sl_fidxs;
|
|
339
|
1079
|
|
|
1080 if (mode == FIND_PREFIX)
|
|
|
1081 {
|
|
|
1082 /* Skip over the prefix. */
|
|
|
1083 wlen = mip->mi_prefixlen;
|
|
|
1084 flen -= mip->mi_prefixlen;
|
|
|
1085 }
|
|
481
|
1086 else if (mode == FIND_COMPOUND)
|
|
|
1087 {
|
|
|
1088 /* Skip over the previously found word(s). */
|
|
|
1089 wlen = mip->mi_compoff;
|
|
|
1090 flen -= mip->mi_compoff;
|
|
|
1091 }
|
|
|
1092
|
|
243
|
1093 }
|
|
|
1094
|
|
300
|
1095 if (byts == NULL)
|
|
|
1096 return; /* array is empty */
|
|
236
|
1097
|
|
|
1098 /*
|
|
307
|
1099 * Repeat advancing in the tree until:
|
|
|
1100 * - there is a byte that doesn't match,
|
|
|
1101 * - we reach the end of the tree,
|
|
|
1102 * - or we reach the end of the line.
|
|
236
|
1103 */
|
|
300
|
1104 for (;;)
|
|
236
|
1105 {
|
|
346
|
1106 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
339
|
1107 flen = fold_more(mip);
|
|
300
|
1108
|
|
|
1109 len = byts[arridx++];
|
|
|
1110
|
|
|
1111 /* If the first possible byte is a zero the word could end here.
|
|
|
1112 * Remember this index, we first check for the longest word. */
|
|
|
1113 if (byts[arridx] == 0)
|
|
|
1114 {
|
|
307
|
1115 if (endidxcnt == MAXWLEN)
|
|
|
1116 {
|
|
|
1117 /* Must be a corrupted spell file. */
|
|
|
1118 EMSG(_(e_format));
|
|
|
1119 return;
|
|
|
1120 }
|
|
300
|
1121 endlen[endidxcnt] = wlen;
|
|
|
1122 endidx[endidxcnt++] = arridx++;
|
|
|
1123 --len;
|
|
|
1124
|
|
|
1125 /* Skip over the zeros, there can be several flag/region
|
|
|
1126 * combinations. */
|
|
|
1127 while (len > 0 && byts[arridx] == 0)
|
|
|
1128 {
|
|
|
1129 ++arridx;
|
|
|
1130 --len;
|
|
|
1131 }
|
|
|
1132 if (len == 0)
|
|
|
1133 break; /* no children, word must end here */
|
|
|
1134 }
|
|
|
1135
|
|
|
1136 /* Stop looking at end of the line. */
|
|
|
1137 if (ptr[wlen] == NUL)
|
|
|
1138 break;
|
|
|
1139
|
|
|
1140 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1141 c = ptr[wlen];
|
|
346
|
1142 if (c == TAB) /* <Tab> is handled like <Space> */
|
|
|
1143 c = ' ';
|
|
300
|
1144 lo = arridx;
|
|
|
1145 hi = arridx + len - 1;
|
|
|
1146 while (lo < hi)
|
|
|
1147 {
|
|
|
1148 m = (lo + hi) / 2;
|
|
|
1149 if (byts[m] > c)
|
|
|
1150 hi = m - 1;
|
|
|
1151 else if (byts[m] < c)
|
|
|
1152 lo = m + 1;
|
|
|
1153 else
|
|
|
1154 {
|
|
|
1155 lo = hi = m;
|
|
|
1156 break;
|
|
236
|
1157 }
|
|
|
1158 }
|
|
300
|
1159
|
|
|
1160 /* Stop if there is no matching byte. */
|
|
|
1161 if (hi < lo || byts[lo] != c)
|
|
|
1162 break;
|
|
|
1163
|
|
|
1164 /* Continue at the child (if there is one). */
|
|
|
1165 arridx = idxs[lo];
|
|
|
1166 ++wlen;
|
|
|
1167 --flen;
|
|
346
|
1168
|
|
|
1169 /* One space in the good word may stand for several spaces in the
|
|
|
1170 * checked word. */
|
|
|
1171 if (c == ' ')
|
|
|
1172 {
|
|
|
1173 for (;;)
|
|
|
1174 {
|
|
|
1175 if (flen <= 0 && *mip->mi_fend != NUL)
|
|
|
1176 flen = fold_more(mip);
|
|
|
1177 if (ptr[wlen] != ' ' && ptr[wlen] != TAB)
|
|
|
1178 break;
|
|
|
1179 ++wlen;
|
|
|
1180 --flen;
|
|
|
1181 }
|
|
|
1182 }
|
|
236
|
1183 }
|
|
|
1184
|
|
300
|
1185 /*
|
|
|
1186 * Verify that one of the possible endings is valid. Try the longest
|
|
|
1187 * first.
|
|
|
1188 */
|
|
|
1189 while (endidxcnt > 0)
|
|
|
1190 {
|
|
|
1191 --endidxcnt;
|
|
|
1192 arridx = endidx[endidxcnt];
|
|
|
1193 wlen = endlen[endidxcnt];
|
|
236
|
1194
|
|
300
|
1195 #ifdef FEAT_MBYTE
|
|
|
1196 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
|
1197 continue; /* not at first byte of character */
|
|
|
1198 #endif
|
|
376
|
1199 if (spell_iswordp(ptr + wlen, mip->mi_buf))
|
|
481
|
1200 {
|
|
492
|
1201 if (slang->sl_compprog == NULL && !slang->sl_nobreak)
|
|
481
|
1202 continue; /* next char is a word character */
|
|
|
1203 word_ends = FALSE;
|
|
|
1204 }
|
|
|
1205 else
|
|
|
1206 word_ends = TRUE;
|
|
485
|
1207 /* The prefix flag is before compound flags. Once a valid prefix flag
|
|
|
1208 * has been found we try compound flags. */
|
|
|
1209 prefix_found = FALSE;
|
|
300
|
1210
|
|
|
1211 #ifdef FEAT_MBYTE
|
|
339
|
1212 if (mode != FIND_KEEPWORD && has_mbyte)
|
|
300
|
1213 {
|
|
|
1214 /* Compute byte length in original word, length may change
|
|
339
|
1215 * when folding case. This can be slow, take a shortcut when the
|
|
|
1216 * case-folded word is equal to the keep-case word. */
|
|
300
|
1217 p = mip->mi_word;
|
|
339
|
1218 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1219 {
|
|
|
1220 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1221 mb_ptr_adv(p);
|
|
|
1222 wlen = p - mip->mi_word;
|
|
|
1223 }
|
|
300
|
1224 }
|
|
|
1225 #endif
|
|
236
|
1226
|
|
339
|
1227 /* Check flags and region. For FIND_PREFIX check the condition and
|
|
|
1228 * prefix ID.
|
|
|
1229 * Repeat this if there are more flags/region alternatives until there
|
|
|
1230 * is a match. */
|
|
|
1231 res = SP_BAD;
|
|
|
1232 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0;
|
|
|
1233 --len, ++arridx)
|
|
300
|
1234 {
|
|
|
1235 flags = idxs[arridx];
|
|
324
|
1236
|
|
339
|
1237 /* For the fold-case tree check that the case of the checked word
|
|
|
1238 * matches with what the word in the tree requires.
|
|
|
1239 * For keep-case tree the case is always right. For prefixes we
|
|
|
1240 * don't bother to check. */
|
|
|
1241 if (mode == FIND_FOLDWORD)
|
|
300
|
1242 {
|
|
|
1243 if (mip->mi_cend != mip->mi_word + wlen)
|
|
|
1244 {
|
|
323
|
1245 /* mi_capflags was set for a different word length, need
|
|
|
1246 * to do it again. */
|
|
300
|
1247 mip->mi_cend = mip->mi_word + wlen;
|
|
323
|
1248 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
|
300
|
1249 }
|
|
|
1250
|
|
346
|
1251 if (mip->mi_capflags == WF_KEEPCAP
|
|
|
1252 || !spell_valid_case(mip->mi_capflags, flags))
|
|
339
|
1253 continue;
|
|
300
|
1254 }
|
|
236
|
1255
|
|
339
|
1256 /* When mode is FIND_PREFIX the word must support the prefix:
|
|
|
1257 * check the prefix ID and the condition. Do that for the list at
|
|
366
|
1258 * mip->mi_prefarridx that find_prefix() filled. */
|
|
485
|
1259 else if (mode == FIND_PREFIX && !prefix_found)
|
|
481
|
1260 {
|
|
366
|
1261 c = valid_word_prefix(mip->mi_prefcnt, mip->mi_prefarridx,
|
|
390
|
1262 flags,
|
|
455
|
1263 mip->mi_word + mip->mi_cprefixlen, slang,
|
|
|
1264 FALSE);
|
|
366
|
1265 if (c == 0)
|
|
339
|
1266 continue;
|
|
366
|
1267
|
|
|
1268 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
1269 if (c & WF_RAREPFX)
|
|
|
1270 flags |= WF_RARE;
|
|
485
|
1271 prefix_found = TRUE;
|
|
339
|
1272 }
|
|
|
1273
|
|
492
|
1274 if (slang->sl_nobreak)
|
|
|
1275 {
|
|
|
1276 if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND)
|
|
|
1277 && (flags & WF_BANNED) == 0)
|
|
|
1278 {
|
|
|
1279 /* NOBREAK: found a valid following word. That's all we
|
|
|
1280 * need to know, so return. */
|
|
|
1281 mip->mi_result = SP_OK;
|
|
|
1282 break;
|
|
|
1283 }
|
|
|
1284 }
|
|
|
1285
|
|
|
1286 else if ((mode == FIND_COMPOUND || mode == FIND_KEEPCOMPOUND
|
|
|
1287 || !word_ends))
|
|
481
|
1288 {
|
|
483
|
1289 /* If there is no flag or the word is shorter than
|
|
|
1290 * COMPOUNDMIN reject it quickly.
|
|
|
1291 * Makes you wonder why someone puts a compound flag on a word
|
|
481
|
1292 * that's too short... Myspell compatibility requires this
|
|
|
1293 * anyway. */
|
|
490
|
1294 if (((unsigned)flags >> 24) == 0
|
|
|
1295 || wlen - mip->mi_compoff < slang->sl_compminlen)
|
|
483
|
1296 continue;
|
|
500
|
1297 #ifdef FEAT_MBYTE
|
|
|
1298 /* For multi-byte chars check character length against
|
|
|
1299 * COMPOUNDMIN. */
|
|
|
1300 if (has_mbyte
|
|
501
|
1301 && slang->sl_compminlen > 0
|
|
500
|
1302 && mb_charlen_len(mip->mi_word + mip->mi_compoff,
|
|
|
1303 wlen - mip->mi_compoff) < slang->sl_compminlen)
|
|
|
1304 continue;
|
|
|
1305 #endif
|
|
483
|
1306
|
|
490
|
1307 /* Limit the number of compound words to COMPOUNDMAX if no
|
|
|
1308 * maximum for syllables is specified. */
|
|
|
1309 if (!word_ends && mip->mi_complen + 2 > slang->sl_compmax
|
|
|
1310 && slang->sl_compsylmax == MAXWLEN)
|
|
483
|
1311 continue;
|
|
|
1312
|
|
485
|
1313 /* Quickly check if compounding is possible with this flag. */
|
|
495
|
1314 if (!byte_in_str(mip->mi_complen == 0
|
|
485
|
1315 ? slang->sl_compstartflags
|
|
|
1316 : slang->sl_compallflags,
|
|
495
|
1317 ((unsigned)flags >> 24)))
|
|
481
|
1318 continue;
|
|
|
1319
|
|
490
|
1320 if (mode == FIND_COMPOUND)
|
|
|
1321 {
|
|
|
1322 int capflags;
|
|
|
1323
|
|
|
1324 /* Need to check the caps type of the appended compound
|
|
|
1325 * word. */
|
|
|
1326 #ifdef FEAT_MBYTE
|
|
|
1327 if (has_mbyte && STRNCMP(ptr, mip->mi_word,
|
|
|
1328 mip->mi_compoff) != 0)
|
|
|
1329 {
|
|
|
1330 /* case folding may have changed the length */
|
|
|
1331 p = mip->mi_word;
|
|
|
1332 for (s = ptr; s < ptr + mip->mi_compoff; mb_ptr_adv(s))
|
|
|
1333 mb_ptr_adv(p);
|
|
|
1334 }
|
|
|
1335 else
|
|
|
1336 #endif
|
|
|
1337 p = mip->mi_word + mip->mi_compoff;
|
|
|
1338 capflags = captype(p, mip->mi_word + wlen);
|
|
|
1339 if (capflags == WF_KEEPCAP || (capflags == WF_ALLCAP
|
|
|
1340 && (flags & WF_FIXCAP) != 0))
|
|
|
1341 continue;
|
|
|
1342
|
|
|
1343 if (capflags != WF_ALLCAP)
|
|
|
1344 {
|
|
|
1345 /* When the character before the word is a word
|
|
|
1346 * character we do not accept a Onecap word. We do
|
|
|
1347 * accept a no-caps word, even when the dictionary
|
|
|
1348 * word specifies ONECAP. */
|
|
|
1349 mb_ptr_back(mip->mi_word, p);
|
|
|
1350 if (spell_iswordp_nmw(p)
|
|
|
1351 ? capflags == WF_ONECAP
|
|
|
1352 : (flags & WF_ONECAP) != 0
|
|
|
1353 && capflags != WF_ONECAP)
|
|
|
1354 continue;
|
|
|
1355 }
|
|
|
1356 }
|
|
|
1357
|
|
483
|
1358 /* If the word ends the sequence of compound flags of the
|
|
|
1359 * words must match with one of the COMPOUNDFLAGS items and
|
|
|
1360 * the number of syllables must not be too large. */
|
|
|
1361 mip->mi_compflags[mip->mi_complen] = ((unsigned)flags >> 24);
|
|
|
1362 mip->mi_compflags[mip->mi_complen + 1] = NUL;
|
|
|
1363 if (word_ends)
|
|
|
1364 {
|
|
|
1365 char_u fword[MAXWLEN];
|
|
|
1366
|
|
|
1367 if (slang->sl_compsylmax < MAXWLEN)
|
|
|
1368 {
|
|
|
1369 /* "fword" is only needed for checking syllables. */
|
|
|
1370 if (ptr == mip->mi_word)
|
|
|
1371 (void)spell_casefold(ptr, wlen, fword, MAXWLEN);
|
|
|
1372 else
|
|
|
1373 vim_strncpy(fword, ptr, endlen[endidxcnt]);
|
|
|
1374 }
|
|
|
1375 if (!can_compound(slang, fword, mip->mi_compflags))
|
|
|
1376 continue;
|
|
|
1377 }
|
|
481
|
1378 }
|
|
|
1379
|
|
500
|
1380 /* Check NEEDCOMPOUND: can't use word without compounding. */
|
|
|
1381 else if (flags & WF_NEEDCOMP)
|
|
|
1382 continue;
|
|
|
1383
|
|
492
|
1384 nobreak_result = SP_OK;
|
|
|
1385
|
|
481
|
1386 if (!word_ends)
|
|
|
1387 {
|
|
492
|
1388 int save_result = mip->mi_result;
|
|
|
1389 char_u *save_end = mip->mi_end;
|
|
501
|
1390 langp_T *save_lp = mip->mi_lp;
|
|
|
1391 int lpi;
|
|
492
|
1392
|
|
|
1393 /* Check that a valid word follows. If there is one and we
|
|
|
1394 * are compounding, it will set "mi_result", thus we are
|
|
|
1395 * always finished here. For NOBREAK we only check that a
|
|
|
1396 * valid word follows.
|
|
481
|
1397 * Recursive! */
|
|
492
|
1398 if (slang->sl_nobreak)
|
|
|
1399 mip->mi_result = SP_BAD;
|
|
481
|
1400
|
|
|
1401 /* Find following word in case-folded tree. */
|
|
|
1402 mip->mi_compoff = endlen[endidxcnt];
|
|
|
1403 #ifdef FEAT_MBYTE
|
|
|
1404 if (has_mbyte && mode == FIND_KEEPWORD)
|
|
|
1405 {
|
|
|
1406 /* Compute byte length in case-folded word from "wlen":
|
|
|
1407 * byte length in keep-case word. Length may change when
|
|
|
1408 * folding case. This can be slow, take a shortcut when
|
|
|
1409 * the case-folded word is equal to the keep-case word. */
|
|
|
1410 p = mip->mi_fword;
|
|
|
1411 if (STRNCMP(ptr, p, wlen) != 0)
|
|
|
1412 {
|
|
|
1413 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
|
1414 mb_ptr_adv(p);
|
|
|
1415 mip->mi_compoff = p - mip->mi_fword;
|
|
|
1416 }
|
|
|
1417 }
|
|
|
1418 #endif
|
|
485
|
1419 c = mip->mi_compoff;
|
|
483
|
1420 ++mip->mi_complen;
|
|
501
|
1421
|
|
|
1422 /* For NOBREAK we need to try all NOBREAK languages, at least
|
|
|
1423 * to find the ".add" file(s). */
|
|
|
1424 for (lpi = 0; lpi < mip->mi_buf->b_langp.ga_len; ++lpi)
|
|
|
1425 {
|
|
|
1426 if (slang->sl_nobreak)
|
|
|
1427 {
|
|
|
1428 mip->mi_lp = LANGP_ENTRY(mip->mi_buf->b_langp, lpi);
|
|
|
1429 if (mip->mi_lp->lp_slang->sl_fidxs == NULL
|
|
|
1430 || !mip->mi_lp->lp_slang->sl_nobreak)
|
|
|
1431 continue;
|
|
|
1432 }
|
|
|
1433
|
|
|
1434 find_word(mip, FIND_COMPOUND);
|
|
|
1435
|
|
|
1436 /* When NOBREAK any word that matches is OK. Otherwise we
|
|
|
1437 * need to find the longest match, thus try with keep-case
|
|
|
1438 * and prefix too. */
|
|
492
|
1439 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1440 {
|
|
501
|
1441 /* Find following word in keep-case tree. */
|
|
|
1442 mip->mi_compoff = wlen;
|
|
|
1443 find_word(mip, FIND_KEEPCOMPOUND);
|
|
|
1444
|
|
|
1445 if (!slang->sl_nobreak || mip->mi_result == SP_BAD)
|
|
|
1446 {
|
|
|
1447 /* Check for following word with prefix. */
|
|
|
1448 mip->mi_compoff = c;
|
|
|
1449 find_prefix(mip, FIND_COMPOUND);
|
|
|
1450 }
|
|
492
|
1451 }
|
|
501
|
1452
|
|
|
1453 if (!slang->sl_nobreak)
|
|
|
1454 break;
|
|
492
|
1455 }
|
|
483
|
1456 --mip->mi_complen;
|
|
501
|
1457 mip->mi_lp = save_lp;
|
|
485
|
1458
|
|
492
|
1459 if (slang->sl_nobreak)
|
|
|
1460 {
|
|
|
1461 nobreak_result = mip->mi_result;
|
|
|
1462 mip->mi_result = save_result;
|
|
|
1463 mip->mi_end = save_end;
|
|
|
1464 }
|
|
|
1465 else
|
|
|
1466 {
|
|
|
1467 if (mip->mi_result == SP_OK)
|
|
|
1468 break;
|
|
|
1469 continue;
|
|
|
1470 }
|
|
481
|
1471 }
|
|
|
1472
|
|
339
|
1473 if (flags & WF_BANNED)
|
|
|
1474 res = SP_BANNED;
|
|
|
1475 else if (flags & WF_REGION)
|
|
|
1476 {
|
|
|
1477 /* Check region. */
|
|
390
|
1478 if ((mip->mi_lp->lp_region & (flags >> 16)) != 0)
|
|
300
|
1479 res = SP_OK;
|
|
339
|
1480 else
|
|
|
1481 res = SP_LOCAL;
|
|
300
|
1482 }
|
|
339
|
1483 else if (flags & WF_RARE)
|
|
|
1484 res = SP_RARE;
|
|
307
|
1485 else
|
|
339
|
1486 res = SP_OK;
|
|
|
1487
|
|
492
|
1488 /* Always use the longest match and the best result. For NOBREAK
|
|
|
1489 * we separately keep the longest match without a following good
|
|
|
1490 * word as a fall-back. */
|
|
|
1491 if (nobreak_result == SP_BAD)
|
|
|
1492 {
|
|
|
1493 if (mip->mi_result2 > res)
|
|
|
1494 {
|
|
|
1495 mip->mi_result2 = res;
|
|
|
1496 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1497 }
|
|
|
1498 else if (mip->mi_result2 == res
|
|
|
1499 && mip->mi_end2 < mip->mi_word + wlen)
|
|
|
1500 mip->mi_end2 = mip->mi_word + wlen;
|
|
|
1501 }
|
|
|
1502 else if (mip->mi_result > res)
|
|
339
|
1503 {
|
|
|
1504 mip->mi_result = res;
|
|
|
1505 mip->mi_end = mip->mi_word + wlen;
|
|
|
1506 }
|
|
351
|
1507 else if (mip->mi_result == res && mip->mi_end < mip->mi_word + wlen)
|
|
339
|
1508 mip->mi_end = mip->mi_word + wlen;
|
|
|
1509
|
|
492
|
1510 if (mip->mi_result == SP_OK)
|
|
339
|
1511 break;
|
|
300
|
1512 }
|
|
|
1513
|
|
492
|
1514 if (mip->mi_result == SP_OK)
|
|
300
|
1515 break;
|
|
|
1516 }
|
|
236
|
1517 }
|
|
|
1518
|
|
323
|
1519 /*
|
|
483
|
1520 * Return TRUE if "flags" is a valid sequence of compound flags and
|
|
|
1521 * "word[len]" does not have too many syllables.
|
|
482
|
1522 */
|
|
|
1523 static int
|
|
483
|
1524 can_compound(slang, word, flags)
|
|
482
|
1525 slang_T *slang;
|
|
483
|
1526 char_u *word;
|
|
|
1527 char_u *flags;
|
|
|
1528 {
|
|
|
1529 regmatch_T regmatch;
|
|
495
|
1530 #ifdef FEAT_MBYTE
|
|
|
1531 char_u uflags[MAXWLEN * 2];
|
|
|
1532 int i;
|
|
|
1533 #endif
|
|
|
1534 char_u *p;
|
|
483
|
1535
|
|
|
1536 if (slang->sl_compprog == NULL)
|
|
|
1537 return FALSE;
|
|
495
|
1538 #ifdef FEAT_MBYTE
|
|
|
1539 if (enc_utf8)
|
|
|
1540 {
|
|
|
1541 /* Need to convert the single byte flags to utf8 characters. */
|
|
|
1542 p = uflags;
|
|
|
1543 for (i = 0; flags[i] != NUL; ++i)
|
|
|
1544 p += mb_char2bytes(flags[i], p);
|
|
|
1545 *p = NUL;
|
|
|
1546 p = uflags;
|
|
|
1547 }
|
|
|
1548 else
|
|
|
1549 #endif
|
|
|
1550 p = flags;
|
|
483
|
1551 regmatch.regprog = slang->sl_compprog;
|
|
|
1552 regmatch.rm_ic = FALSE;
|
|
495
|
1553 if (!vim_regexec(®match, p, 0))
|
|
483
|
1554 return FALSE;
|
|
|
1555
|
|
490
|
1556 /* Count the number of syllables. This may be slow, do it last. If there
|
|
|
1557 * are too many syllables AND the number of compound words is above
|
|
|
1558 * COMPOUNDMAX then compounding is not allowed. */
|
|
483
|
1559 if (slang->sl_compsylmax < MAXWLEN
|
|
|
1560 && count_syllables(slang, word) > slang->sl_compsylmax)
|
|
495
|
1561 return (int)STRLEN(flags) < slang->sl_compmax;
|
|
483
|
1562 return TRUE;
|
|
482
|
1563 }
|
|
|
1564
|
|
|
1565 /*
|
|
390
|
1566 * Return non-zero if the prefix indicated by "arridx" matches with the prefix
|
|
|
1567 * ID in "flags" for the word "word".
|
|
366
|
1568 * The WF_RAREPFX flag is included in the return value for a rare prefix.
|
|
351
|
1569 */
|
|
|
1570 static int
|
|
455
|
1571 valid_word_prefix(totprefcnt, arridx, flags, word, slang, cond_req)
|
|
351
|
1572 int totprefcnt; /* nr of prefix IDs */
|
|
|
1573 int arridx; /* idx in sl_pidxs[] */
|
|
390
|
1574 int flags;
|
|
351
|
1575 char_u *word;
|
|
|
1576 slang_T *slang;
|
|
455
|
1577 int cond_req; /* only use prefixes with a condition */
|
|
351
|
1578 {
|
|
|
1579 int prefcnt;
|
|
|
1580 int pidx;
|
|
|
1581 regprog_T *rp;
|
|
|
1582 regmatch_T regmatch;
|
|
390
|
1583 int prefid;
|
|
|
1584
|
|
|
1585 prefid = (unsigned)flags >> 24;
|
|
351
|
1586 for (prefcnt = totprefcnt - 1; prefcnt >= 0; --prefcnt)
|
|
|
1587 {
|
|
|
1588 pidx = slang->sl_pidxs[arridx + prefcnt];
|
|
|
1589
|
|
|
1590 /* Check the prefix ID. */
|
|
|
1591 if (prefid != (pidx & 0xff))
|
|
|
1592 continue;
|
|
|
1593
|
|
390
|
1594 /* Check if the prefix doesn't combine and the word already has a
|
|
|
1595 * suffix. */
|
|
|
1596 if ((flags & WF_HAS_AFF) && (pidx & WF_PFX_NC))
|
|
|
1597 continue;
|
|
|
1598
|
|
351
|
1599 /* Check the condition, if there is one. The condition index is
|
|
366
|
1600 * stored in the two bytes above the prefix ID byte. */
|
|
|
1601 rp = slang->sl_prefprog[((unsigned)pidx >> 8) & 0xffff];
|
|
351
|
1602 if (rp != NULL)
|
|
|
1603 {
|
|
|
1604 regmatch.regprog = rp;
|
|
|
1605 regmatch.rm_ic = FALSE;
|
|
|
1606 if (!vim_regexec(®match, word, 0))
|
|
|
1607 continue;
|
|
|
1608 }
|
|
455
|
1609 else if (cond_req)
|
|
|
1610 continue;
|
|
|
1611
|
|
|
1612 /* It's a match! Return the WF_ flags. */
|
|
366
|
1613 return pidx;
|
|
|
1614 }
|
|
|
1615 return 0;
|
|
351
|
1616 }
|
|
|
1617
|
|
|
1618 /*
|
|
339
|
1619 * Check if the word at "mip->mi_word" has a matching prefix.
|
|
|
1620 * If it does, then check the following word.
|
|
|
1621 *
|
|
485
|
1622 * If "mode" is "FIND_COMPOUND" then do the same after another word, find a
|
|
|
1623 * prefix in a compound word.
|
|
|
1624 *
|
|
339
|
1625 * For a match mip->mi_result is updated.
|
|
|
1626 */
|
|
|
1627 static void
|
|
485
|
1628 find_prefix(mip, mode)
|
|
339
|
1629 matchinf_T *mip;
|
|
485
|
1630 int mode;
|
|
339
|
1631 {
|
|
|
1632 idx_T arridx = 0;
|
|
|
1633 int len;
|
|
|
1634 int wlen = 0;
|
|
|
1635 int flen;
|
|
|
1636 int c;
|
|
|
1637 char_u *ptr;
|
|
|
1638 idx_T lo, hi, m;
|
|
|
1639 slang_T *slang = mip->mi_lp->lp_slang;
|
|
|
1640 char_u *byts;
|
|
|
1641 idx_T *idxs;
|
|
|
1642
|
|
375
|
1643 byts = slang->sl_pbyts;
|
|
|
1644 if (byts == NULL)
|
|
|
1645 return; /* array is empty */
|
|
|
1646
|
|
339
|
1647 /* We use the case-folded word here, since prefixes are always
|
|
|
1648 * case-folded. */
|
|
|
1649 ptr = mip->mi_fword;
|
|
|
1650 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
485
|
1651 if (mode == FIND_COMPOUND)
|
|
|
1652 {
|
|
|
1653 /* Skip over the previously found word(s). */
|
|
|
1654 ptr += mip->mi_compoff;
|
|
|
1655 flen -= mip->mi_compoff;
|
|
|
1656 }
|
|
339
|
1657 idxs = slang->sl_pidxs;
|
|
|
1658
|
|
|
1659 /*
|
|
|
1660 * Repeat advancing in the tree until:
|
|
|
1661 * - there is a byte that doesn't match,
|
|
|
1662 * - we reach the end of the tree,
|
|
|
1663 * - or we reach the end of the line.
|
|
|
1664 */
|
|
|
1665 for (;;)
|
|
|
1666 {
|
|
|
1667 if (flen == 0 && *mip->mi_fend != NUL)
|
|
|
1668 flen = fold_more(mip);
|
|
|
1669
|
|
|
1670 len = byts[arridx++];
|
|
|
1671
|
|
|
1672 /* If the first possible byte is a zero the prefix could end here.
|
|
|
1673 * Check if the following word matches and supports the prefix. */
|
|
|
1674 if (byts[arridx] == 0)
|
|
|
1675 {
|
|
|
1676 /* There can be several prefixes with different conditions. We
|
|
|
1677 * try them all, since we don't know which one will give the
|
|
|
1678 * longest match. The word is the same each time, pass the list
|
|
|
1679 * of possible prefixes to find_word(). */
|
|
|
1680 mip->mi_prefarridx = arridx;
|
|
|
1681 mip->mi_prefcnt = len;
|
|
|
1682 while (len > 0 && byts[arridx] == 0)
|
|
|
1683 {
|
|
|
1684 ++arridx;
|
|
|
1685 --len;
|
|
|
1686 }
|
|
|
1687 mip->mi_prefcnt -= len;
|
|
|
1688
|
|
|
1689 /* Find the word that comes after the prefix. */
|
|
|
1690 mip->mi_prefixlen = wlen;
|
|
485
|
1691 if (mode == FIND_COMPOUND)
|
|
|
1692 /* Skip over the previously found word(s). */
|
|
|
1693 mip->mi_prefixlen += mip->mi_compoff;
|
|
|
1694
|
|
455
|
1695 #ifdef FEAT_MBYTE
|
|
|
1696 if (has_mbyte)
|
|
|
1697 {
|
|
|
1698 /* Case-folded length may differ from original length. */
|
|
485
|
1699 mip->mi_cprefixlen = nofold_len(mip->mi_fword,
|
|
|
1700 mip->mi_prefixlen, mip->mi_word);
|
|
455
|
1701 }
|
|
|
1702 else
|
|
485
|
1703 mip->mi_cprefixlen = mip->mi_prefixlen;
|
|
455
|
1704 #endif
|
|
339
|
1705 find_word(mip, FIND_PREFIX);
|
|
|
1706
|
|
|
1707
|
|
|
1708 if (len == 0)
|
|
|
1709 break; /* no children, word must end here */
|
|
|
1710 }
|
|
|
1711
|
|
|
1712 /* Stop looking at end of the line. */
|
|
|
1713 if (ptr[wlen] == NUL)
|
|
|
1714 break;
|
|
|
1715
|
|
|
1716 /* Perform a binary search in the list of accepted bytes. */
|
|
|
1717 c = ptr[wlen];
|
|
|
1718 lo = arridx;
|
|
|
1719 hi = arridx + len - 1;
|
|
|
1720 while (lo < hi)
|
|
|
1721 {
|
|
|
1722 m = (lo + hi) / 2;
|
|
|
1723 if (byts[m] > c)
|
|
|
1724 hi = m - 1;
|
|
|
1725 else if (byts[m] < c)
|
|
|
1726 lo = m + 1;
|
|
|
1727 else
|
|
|
1728 {
|
|
|
1729 lo = hi = m;
|
|
|
1730 break;
|
|
|
1731 }
|
|
|
1732 }
|
|
|
1733
|
|
|
1734 /* Stop if there is no matching byte. */
|
|
|
1735 if (hi < lo || byts[lo] != c)
|
|
|
1736 break;
|
|
|
1737
|
|
|
1738 /* Continue at the child (if there is one). */
|
|
|
1739 arridx = idxs[lo];
|
|
|
1740 ++wlen;
|
|
|
1741 --flen;
|
|
|
1742 }
|
|
|
1743 }
|
|
|
1744
|
|
|
1745 /*
|
|
|
1746 * Need to fold at least one more character. Do until next non-word character
|
|
|
1747 * for efficiency.
|
|
|
1748 * Return the length of the folded chars in bytes.
|
|
|
1749 */
|
|
|
1750 static int
|
|
|
1751 fold_more(mip)
|
|
|
1752 matchinf_T *mip;
|
|
|
1753 {
|
|
|
1754 int flen;
|
|
|
1755 char_u *p;
|
|
|
1756
|
|
|
1757 p = mip->mi_fend;
|
|
|
1758 do
|
|
|
1759 {
|
|
|
1760 mb_ptr_adv(mip->mi_fend);
|
|
376
|
1761 } while (*mip->mi_fend != NUL && spell_iswordp(mip->mi_fend, mip->mi_buf));
|
|
339
|
1762
|
|
|
1763 /* Include the non-word character so that we can check for the
|
|
|
1764 * word end. */
|
|
|
1765 if (*mip->mi_fend != NUL)
|
|
|
1766 mb_ptr_adv(mip->mi_fend);
|
|
|
1767
|
|
|
1768 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
|
|
1769 mip->mi_fword + mip->mi_fwordlen,
|
|
|
1770 MAXWLEN - mip->mi_fwordlen);
|
|
|
1771 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
|
1772 mip->mi_fwordlen += flen;
|
|
|
1773 return flen;
|
|
|
1774 }
|
|
|
1775
|
|
|
1776 /*
|
|
323
|
1777 * Check case flags for a word. Return TRUE if the word has the requested
|
|
|
1778 * case.
|
|
|
1779 */
|
|
|
1780 static int
|
|
389
|
1781 spell_valid_case(wordflags, treeflags)
|
|
|
1782 int wordflags; /* flags for the checked word. */
|
|
323
|
1783 int treeflags; /* flags for the word in the spell tree */
|
|
|
1784 {
|
|
389
|
1785 return ((wordflags == WF_ALLCAP && (treeflags & WF_FIXCAP) == 0)
|
|
323
|
1786 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
474
|
1787 && ((treeflags & WF_ONECAP) == 0
|
|
|
1788 || (wordflags & WF_ONECAP) != 0)));
|
|
323
|
1789 }
|
|
|
1790
|
|
351
|
1791 /*
|
|
|
1792 * Return TRUE if spell checking is not enabled.
|
|
|
1793 */
|
|
|
1794 static int
|
|
498
|
1795 no_spell_checking(wp)
|
|
|
1796 win_T *wp;
|
|
|
1797 {
|
|
|
1798 if (!wp->w_p_spell || *wp->w_buffer->b_p_spl == NUL)
|
|
351
|
1799 {
|
|
|
1800 EMSG(_("E756: Spell checking is not enabled"));
|
|
|
1801 return TRUE;
|
|
|
1802 }
|
|
|
1803 return FALSE;
|
|
|
1804 }
|
|
300
|
1805
|
|
236
|
1806 /*
|
|
|
1807 * Move to next spell error.
|
|
500
|
1808 * "curline" is FALSE for "[s", "]s", "[S" and "]S".
|
|
|
1809 * "curline" is TRUE to find word under/after cursor in the same line.
|
|
483
|
1810 * For Insert mode completion "dir" is BACKWARD and "curline" is TRUE: move
|
|
|
1811 * to after badly spelled word before the cursor.
|
|
495
|
1812 * Return 0 if not found, length of the badly spelled word otherwise.
|
|
236
|
1813 */
|
|
|
1814 int
|
|
498
|
1815 spell_move_to(wp, dir, allwords, curline, attrp)
|
|
|
1816 win_T *wp;
|
|
236
|
1817 int dir; /* FORWARD or BACKWARD */
|
|
500
|
1818 int allwords; /* TRUE for "[s"/"]s", FALSE for "[S"/"]S" */
|
|
323
|
1819 int curline;
|
|
534
|
1820 hlf_T *attrp; /* return: attributes of bad word or NULL
|
|
|
1821 (only when "dir" is FORWARD) */
|
|
236
|
1822 {
|
|
249
|
1823 linenr_T lnum;
|
|
|
1824 pos_T found_pos;
|
|
495
|
1825 int found_len = 0;
|
|
236
|
1826 char_u *line;
|
|
|
1827 char_u *p;
|
|
346
|
1828 char_u *endp;
|
|
534
|
1829 hlf_T attr;
|
|
236
|
1830 int len;
|
|
498
|
1831 int has_syntax = syntax_present(wp->w_buffer);
|
|
249
|
1832 int col;
|
|
|
1833 int can_spell;
|
|
346
|
1834 char_u *buf = NULL;
|
|
|
1835 int buflen = 0;
|
|
|
1836 int skip = 0;
|
|
385
|
1837 int capcol = -1;
|
|
500
|
1838 int found_one = FALSE;
|
|
|
1839 int wrapped = FALSE;
|
|
236
|
1840
|
|
498
|
1841 if (no_spell_checking(wp))
|
|
495
|
1842 return 0;
|
|
236
|
1843
|
|
249
|
1844 /*
|
|
|
1845 * Start looking for bad word at the start of the line, because we can't
|
|
346
|
1846 * start halfway a word, we don't know where the it starts or ends.
|
|
249
|
1847 *
|
|
|
1848 * When searching backwards, we continue in the line to find the last
|
|
|
1849 * bad word (in the cursor line: before the cursor).
|
|
346
|
1850 *
|
|
|
1851 * We concatenate the start of the next line, so that wrapped words work
|
|
|
1852 * (e.g. "et<line-break>cetera"). Doesn't work when searching backwards
|
|
|
1853 * though...
|
|
249
|
1854 */
|
|
498
|
1855 lnum = wp->w_cursor.lnum;
|
|
249
|
1856 found_pos.lnum = 0;
|
|
236
|
1857
|
|
|
1858 while (!got_int)
|
|
|
1859 {
|
|
498
|
1860 line = ml_get_buf(wp->w_buffer, lnum, FALSE);
|
|
346
|
1861
|
|
|
1862 len = STRLEN(line);
|
|
|
1863 if (buflen < len + MAXWLEN + 2)
|
|
|
1864 {
|
|
|
1865 vim_free(buf);
|
|
|
1866 buflen = len + MAXWLEN + 2;
|
|
|
1867 buf = alloc(buflen);
|
|
|
1868 if (buf == NULL)
|
|
|
1869 break;
|
|
|
1870 }
|
|
|
1871
|
|
385
|
1872 /* In first line check first word for Capital. */
|
|
|
1873 if (lnum == 1)
|
|
|
1874 capcol = 0;
|
|
|
1875
|
|
|
1876 /* For checking first word with a capital skip white space. */
|
|
|
1877 if (capcol == 0)
|
|
|
1878 capcol = skipwhite(line) - line;
|
|
|
1879
|
|
346
|
1880 /* Copy the line into "buf" and append the start of the next line if
|
|
|
1881 * possible. */
|
|
|
1882 STRCPY(buf, line);
|
|
498
|
1883 if (lnum < wp->w_buffer->b_ml.ml_line_count)
|
|
346
|
1884 spell_cat_line(buf + STRLEN(buf), ml_get(lnum + 1), MAXWLEN);
|
|
|
1885
|
|
|
1886 p = buf + skip;
|
|
|
1887 endp = buf + len;
|
|
|
1888 while (p < endp)
|
|
236
|
1889 {
|
|
500
|
1890 /* When searching backward don't search after the cursor. Unless
|
|
|
1891 * we wrapped around the end of the buffer. */
|
|
300
|
1892 if (dir == BACKWARD
|
|
498
|
1893 && lnum == wp->w_cursor.lnum
|
|
500
|
1894 && !wrapped
|
|
498
|
1895 && (colnr_T)(p - buf) >= wp->w_cursor.col)
|
|
300
|
1896 break;
|
|
249
|
1897
|
|
300
|
1898 /* start of word */
|
|
534
|
1899 attr = HLF_COUNT;
|
|
498
|
1900 len = spell_check(wp, p, &attr, &capcol);
|
|
249
|
1901
|
|
534
|
1902 if (attr != HLF_COUNT)
|
|
300
|
1903 {
|
|
|
1904 /* We found a bad word. Check the attribute. */
|
|
534
|
1905 if (allwords || attr == HLF_SPB)
|
|
236
|
1906 {
|
|
500
|
1907 found_one = TRUE;
|
|
|
1908
|
|
300
|
1909 /* When searching forward only accept a bad word after
|
|
|
1910 * the cursor. */
|
|
|
1911 if (dir == BACKWARD
|
|
500
|
1912 || lnum != wp->w_cursor.lnum
|
|
498
|
1913 || (lnum == wp->w_cursor.lnum
|
|
500
|
1914 && (wrapped
|
|
|
1915 || (colnr_T)(curline ? p - buf + len
|
|
346
|
1916 : p - buf)
|
|
500
|
1917 > wp->w_cursor.col)))
|
|
236
|
1918 {
|
|
300
|
1919 if (has_syntax)
|
|
249
|
1920 {
|
|
346
|
1921 col = p - buf;
|
|
498
|
1922 (void)syn_get_id(wp, lnum, (colnr_T)col,
|
|
300
|
1923 FALSE, &can_spell);
|
|
|
1924 }
|
|
|
1925 else
|
|
|
1926 can_spell = TRUE;
|
|
249
|
1927
|
|
300
|
1928 if (can_spell)
|
|
|
1929 {
|
|
|
1930 found_pos.lnum = lnum;
|
|
346
|
1931 found_pos.col = p - buf;
|
|
249
|
1932 #ifdef FEAT_VIRTUALEDIT
|
|
300
|
1933 found_pos.coladd = 0;
|
|
249
|
1934 #endif
|
|
300
|
1935 if (dir == FORWARD)
|
|
|
1936 {
|
|
|
1937 /* No need to search further. */
|
|
498
|
1938 wp->w_cursor = found_pos;
|
|
346
|
1939 vim_free(buf);
|
|
498
|
1940 if (attrp != NULL)
|
|
|
1941 *attrp = attr;
|
|
495
|
1942 return len;
|
|
249
|
1943 }
|
|
483
|
1944 else if (curline)
|
|
|
1945 /* Insert mode completion: put cursor after
|
|
|
1946 * the bad word. */
|
|
|
1947 found_pos.col += len;
|
|
495
|
1948 found_len = len;
|
|
249
|
1949 }
|
|
236
|
1950 }
|
|
|
1951 }
|
|
|
1952 }
|
|
|
1953
|
|
300
|
1954 /* advance to character after the word */
|
|
|
1955 p += len;
|
|
385
|
1956 capcol -= len;
|
|
236
|
1957 }
|
|
|
1958
|
|
483
|
1959 if (dir == BACKWARD && found_pos.lnum != 0)
|
|
|
1960 {
|
|
500
|
1961 /* Use the last match in the line (before the cursor). */
|
|
498
|
1962 wp->w_cursor = found_pos;
|
|
483
|
1963 vim_free(buf);
|
|
495
|
1964 return found_len;
|
|
483
|
1965 }
|
|
|
1966
|
|
323
|
1967 if (curline)
|
|
346
|
1968 break; /* only check cursor line */
|
|
323
|
1969
|
|
236
|
1970 /* Advance to next line. */
|
|
249
|
1971 if (dir == BACKWARD)
|
|
|
1972 {
|
|
500
|
1973 /* If we are back at the starting line and searched it again there
|
|
|
1974 * is no match, give up. */
|
|
|
1975 if (lnum == wp->w_cursor.lnum && wrapped)
|
|
346
|
1976 break;
|
|
500
|
1977
|
|
|
1978 if (lnum > 1)
|
|
|
1979 --lnum;
|
|
|
1980 else if (!p_ws)
|
|
|
1981 break; /* at first line and 'nowrapscan' */
|
|
|
1982 else
|
|
|
1983 {
|
|
|
1984 /* Wrap around to the end of the buffer. May search the
|
|
|
1985 * starting line again and accept the last match. */
|
|
|
1986 lnum = wp->w_buffer->b_ml.ml_line_count;
|
|
|
1987 wrapped = TRUE;
|
|
503
|
1988 if (!shortmess(SHM_SEARCH))
|
|
|
1989 give_warning((char_u *)_(top_bot_msg), TRUE);
|
|
500
|
1990 }
|
|
385
|
1991 capcol = -1;
|
|
249
|
1992 }
|
|
|
1993 else
|
|
|
1994 {
|
|
500
|
1995 if (lnum < wp->w_buffer->b_ml.ml_line_count)
|
|
|
1996 ++lnum;
|
|
|
1997 else if (!p_ws)
|
|
|
1998 break; /* at first line and 'nowrapscan' */
|
|
|
1999 else
|
|
|
2000 {
|
|
|
2001 /* Wrap around to the start of the buffer. May search the
|
|
|
2002 * starting line again and accept the first match. */
|
|
|
2003 lnum = 1;
|
|
|
2004 wrapped = TRUE;
|
|
503
|
2005 if (!shortmess(SHM_SEARCH))
|
|
|
2006 give_warning((char_u *)_(bot_top_msg), TRUE);
|
|
500
|
2007 }
|
|
|
2008
|
|
|
2009 /* If we are back at the starting line and there is no match then
|
|
|
2010 * give up. */
|
|
|
2011 if (lnum == wp->w_cursor.lnum && !found_one)
|
|
346
|
2012 break;
|
|
|
2013
|
|
|
2014 /* Skip the characters at the start of the next line that were
|
|
|
2015 * included in a match crossing line boundaries. */
|
|
534
|
2016 if (attr == HLF_COUNT)
|
|
346
|
2017 skip = p - endp;
|
|
|
2018 else
|
|
|
2019 skip = 0;
|
|
385
|
2020
|
|
|
2021 /* Capscol skips over the inserted space. */
|
|
|
2022 --capcol;
|
|
|
2023
|
|
|
2024 /* But after empty line check first word in next line */
|
|
|
2025 if (*skipwhite(line) == NUL)
|
|
|
2026 capcol = 0;
|
|
249
|
2027 }
|
|
236
|
2028
|
|
|
2029 line_breakcheck();
|
|
|
2030 }
|
|
|
2031
|
|
346
|
2032 vim_free(buf);
|
|
495
|
2033 return 0;
|
|
346
|
2034 }
|
|
|
2035
|
|
|
2036 /*
|
|
|
2037 * For spell checking: concatenate the start of the following line "line" into
|
|
|
2038 * "buf", blanking-out special characters. Copy less then "maxlen" bytes.
|
|
|
2039 */
|
|
|
2040 void
|
|
|
2041 spell_cat_line(buf, line, maxlen)
|
|
|
2042 char_u *buf;
|
|
|
2043 char_u *line;
|
|
|
2044 int maxlen;
|
|
|
2045 {
|
|
|
2046 char_u *p;
|
|
|
2047 int n;
|
|
|
2048
|
|
|
2049 p = skipwhite(line);
|
|
|
2050 while (vim_strchr((char_u *)"*#/\"\t", *p) != NULL)
|
|
|
2051 p = skipwhite(p + 1);
|
|
|
2052
|
|
|
2053 if (*p != NUL)
|
|
|
2054 {
|
|
|
2055 *buf = ' ';
|
|
376
|
2056 vim_strncpy(buf + 1, line, maxlen - 2);
|
|
346
|
2057 n = p - line;
|
|
|
2058 if (n >= maxlen)
|
|
|
2059 n = maxlen - 1;
|
|
|
2060 vim_memset(buf + 1, ' ', n);
|
|
|
2061 }
|
|
236
|
2062 }
|
|
|
2063
|
|
501
|
2064 typedef struct spelload_S
|
|
|
2065 {
|
|
|
2066 char_u sl_lang[MAXWLEN + 1]; /* language name */
|
|
|
2067 slang_T *sl_slang; /* resulting slang_T struct */
|
|
|
2068 int sl_nobreak; /* NOBREAK language found */
|
|
|
2069 } spelload_T;
|
|
|
2070
|
|
236
|
2071 /*
|
|
307
|
2072 * Load word list(s) for "lang" from Vim spell file(s).
|
|
310
|
2073 * "lang" must be the language without the region: e.g., "en".
|
|
236
|
2074 */
|
|
307
|
2075 static void
|
|
236
|
2076 spell_load_lang(lang)
|
|
|
2077 char_u *lang;
|
|
|
2078 {
|
|
310
|
2079 char_u fname_enc[85];
|
|
236
|
2080 int r;
|
|
501
|
2081 spelload_T sl;
|
|
307
|
2082
|
|
310
|
2083 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
|
307
|
2084 * It's truncated when an error is detected. */
|
|
501
|
2085 STRCPY(sl.sl_lang, lang);
|
|
|
2086 sl.sl_slang = NULL;
|
|
|
2087 sl.sl_nobreak = FALSE;
|
|
307
|
2088
|
|
310
|
2089 /*
|
|
|
2090 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
|
2091 */
|
|
|
2092 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
|
2093 "spell/%s.%s.spl", lang, spell_enc());
|
|
501
|
2094 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl);
|
|
|
2095
|
|
|
2096 if (r == FAIL && *sl.sl_lang != NUL)
|
|
307
|
2097 {
|
|
|
2098 /* Try loading the ASCII version. */
|
|
310
|
2099 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
272
|
2100 "spell/%s.ascii.spl", lang);
|
|
501
|
2101 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &sl);
|
|
307
|
2102 }
|
|
|
2103
|
|
|
2104 if (r == FAIL)
|
|
483
|
2105 smsg((char_u *)_("Warning: Cannot find word list \"%s.%s.spl\" or \"%s.ascii.spl\""),
|
|
|
2106 lang, spell_enc(), lang);
|
|
501
|
2107 else if (sl.sl_slang != NULL)
|
|
|
2108 {
|
|
|
2109 /* At least one file was loaded, now load all the additions. */
|
|
310
|
2110 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
501
|
2111 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &sl);
|
|
310
|
2112 }
|
|
|
2113 }
|
|
|
2114
|
|
|
2115 /*
|
|
|
2116 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
|
2117 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
|
2118 */
|
|
|
2119 static char_u *
|
|
|
2120 spell_enc()
|
|
|
2121 {
|
|
|
2122
|
|
|
2123 #ifdef FEAT_MBYTE
|
|
|
2124 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
|
2125 return p_enc;
|
|
|
2126 #endif
|
|
|
2127 return (char_u *)"latin1";
|
|
236
|
2128 }
|
|
|
2129
|
|
|
2130 /*
|
|
385
|
2131 * Get the name of the .spl file for the internal wordlist into
|
|
|
2132 * "fname[MAXPATHL]".
|
|
|
2133 */
|
|
|
2134 static void
|
|
|
2135 int_wordlist_spl(fname)
|
|
|
2136 char_u *fname;
|
|
|
2137 {
|
|
|
2138 vim_snprintf((char *)fname, MAXPATHL, "%s.%s.spl",
|
|
|
2139 int_wordlist, spell_enc());
|
|
|
2140 }
|
|
|
2141
|
|
|
2142 /*
|
|
236
|
2143 * Allocate a new slang_T.
|
|
|
2144 * Caller must fill "sl_next".
|
|
|
2145 */
|
|
|
2146 static slang_T *
|
|
|
2147 slang_alloc(lang)
|
|
|
2148 char_u *lang;
|
|
|
2149 {
|
|
|
2150 slang_T *lp;
|
|
|
2151
|
|
300
|
2152 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
|
236
|
2153 if (lp != NULL)
|
|
|
2154 {
|
|
|
2155 lp->sl_name = vim_strsave(lang);
|
|
323
|
2156 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
483
|
2157 lp->sl_compmax = MAXWLEN;
|
|
|
2158 lp->sl_compsylmax = MAXWLEN;
|
|
236
|
2159 }
|
|
|
2160 return lp;
|
|
|
2161 }
|
|
|
2162
|
|
|
2163 /*
|
|
|
2164 * Free the contents of an slang_T and the structure itself.
|
|
|
2165 */
|
|
|
2166 static void
|
|
|
2167 slang_free(lp)
|
|
|
2168 slang_T *lp;
|
|
|
2169 {
|
|
|
2170 vim_free(lp->sl_name);
|
|
310
|
2171 vim_free(lp->sl_fname);
|
|
|
2172 slang_clear(lp);
|
|
|
2173 vim_free(lp);
|
|
|
2174 }
|
|
|
2175
|
|
|
2176 /*
|
|
|
2177 * Clear an slang_T so that the file can be reloaded.
|
|
|
2178 */
|
|
|
2179 static void
|
|
|
2180 slang_clear(lp)
|
|
|
2181 slang_T *lp;
|
|
|
2182 {
|
|
339
|
2183 garray_T *gap;
|
|
|
2184 fromto_T *ftp;
|
|
344
|
2185 salitem_T *smp;
|
|
339
|
2186 int i;
|
|
323
|
2187
|
|
300
|
2188 vim_free(lp->sl_fbyts);
|
|
310
|
2189 lp->sl_fbyts = NULL;
|
|
300
|
2190 vim_free(lp->sl_kbyts);
|
|
310
|
2191 lp->sl_kbyts = NULL;
|
|
339
|
2192 vim_free(lp->sl_pbyts);
|
|
|
2193 lp->sl_pbyts = NULL;
|
|
|
2194
|
|
300
|
2195 vim_free(lp->sl_fidxs);
|
|
310
|
2196 lp->sl_fidxs = NULL;
|
|
300
|
2197 vim_free(lp->sl_kidxs);
|
|
310
|
2198 lp->sl_kidxs = NULL;
|
|
339
|
2199 vim_free(lp->sl_pidxs);
|
|
|
2200 lp->sl_pidxs = NULL;
|
|
323
|
2201
|
|
344
|
2202 gap = &lp->sl_rep;
|
|
|
2203 while (gap->ga_len > 0)
|
|
323
|
2204 {
|
|
344
|
2205 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2206 vim_free(ftp->ft_from);
|
|
|
2207 vim_free(ftp->ft_to);
|
|
323
|
2208 }
|
|
344
|
2209 ga_clear(gap);
|
|
|
2210
|
|
|
2211 gap = &lp->sl_sal;
|
|
375
|
2212 if (lp->sl_sofo)
|
|
376
|
2213 {
|
|
|
2214 /* "ga_len" is set to 1 without adding an item for latin1 */
|
|
|
2215 if (gap->ga_data != NULL)
|
|
|
2216 /* SOFOFROM and SOFOTO items: free lists of wide characters. */
|
|
|
2217 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2218 vim_free(((int **)gap->ga_data)[i]);
|
|
|
2219 }
|
|
375
|
2220 else
|
|
|
2221 /* SAL items: free salitem_T items */
|
|
|
2222 while (gap->ga_len > 0)
|
|
|
2223 {
|
|
|
2224 smp = &((salitem_T *)gap->ga_data)[--gap->ga_len];
|
|
|
2225 vim_free(smp->sm_lead);
|
|
|
2226 /* Don't free sm_oneof and sm_rules, they point into sm_lead. */
|
|
|
2227 vim_free(smp->sm_to);
|
|
|
2228 #ifdef FEAT_MBYTE
|
|
|
2229 vim_free(smp->sm_lead_w);
|
|
|
2230 vim_free(smp->sm_oneof_w);
|
|
|
2231 vim_free(smp->sm_to_w);
|
|
|
2232 #endif
|
|
|
2233 }
|
|
344
|
2234 ga_clear(gap);
|
|
323
|
2235
|
|
339
|
2236 for (i = 0; i < lp->sl_prefixcnt; ++i)
|
|
|
2237 vim_free(lp->sl_prefprog[i]);
|
|
376
|
2238 lp->sl_prefixcnt = 0;
|
|
339
|
2239 vim_free(lp->sl_prefprog);
|
|
376
|
2240 lp->sl_prefprog = NULL;
|
|
|
2241
|
|
|
2242 vim_free(lp->sl_midword);
|
|
|
2243 lp->sl_midword = NULL;
|
|
339
|
2244
|
|
483
|
2245 vim_free(lp->sl_compprog);
|
|
|
2246 vim_free(lp->sl_compstartflags);
|
|
485
|
2247 vim_free(lp->sl_compallflags);
|
|
483
|
2248 lp->sl_compprog = NULL;
|
|
|
2249 lp->sl_compstartflags = NULL;
|
|
485
|
2250 lp->sl_compallflags = NULL;
|
|
483
|
2251
|
|
|
2252 vim_free(lp->sl_syllable);
|
|
|
2253 lp->sl_syllable = NULL;
|
|
|
2254 ga_clear(&lp->sl_syl_items);
|
|
481
|
2255
|
|
330
|
2256 #ifdef FEAT_MBYTE
|
|
|
2257 {
|
|
|
2258 int todo = lp->sl_map_hash.ht_used;
|
|
|
2259 hashitem_T *hi;
|
|
|
2260
|
|
|
2261 for (hi = lp->sl_map_hash.ht_array; todo > 0; ++hi)
|
|
|
2262 if (!HASHITEM_EMPTY(hi))
|
|
|
2263 {
|
|
|
2264 --todo;
|
|
|
2265 vim_free(hi->hi_key);
|
|
|
2266 }
|
|
|
2267 }
|
|
|
2268 hash_clear(&lp->sl_map_hash);
|
|
|
2269 #endif
|
|
483
|
2270
|
|
|
2271 lp->sl_compmax = MAXWLEN;
|
|
501
|
2272 lp->sl_compminlen = 0;
|
|
483
|
2273 lp->sl_compsylmax = MAXWLEN;
|
|
|
2274 lp->sl_regions[0] = NUL;
|
|
236
|
2275 }
|
|
|
2276
|
|
|
2277 /*
|
|
307
|
2278 * Load one spell file and store the info into a slang_T.
|
|
236
|
2279 * Invoked through do_in_runtimepath().
|
|
|
2280 */
|
|
|
2281 static void
|
|
310
|
2282 spell_load_cb(fname, cookie)
|
|
236
|
2283 char_u *fname;
|
|
501
|
2284 void *cookie;
|
|
|
2285 {
|
|
|
2286 spelload_T *slp = (spelload_T *)cookie;
|
|
|
2287 slang_T *slang;
|
|
|
2288
|
|
|
2289 slang = spell_load_file(fname, slp->sl_lang, NULL, FALSE);
|
|
|
2290 if (slang != NULL)
|
|
|
2291 {
|
|
|
2292 /* When a previously loaded file has NOBREAK also use it for the
|
|
|
2293 * ".add" files. */
|
|
|
2294 if (slp->sl_nobreak && slang->sl_add)
|
|
|
2295 slang->sl_nobreak = TRUE;
|
|
|
2296 else if (slang->sl_nobreak)
|
|
|
2297 slp->sl_nobreak = TRUE;
|
|
|
2298
|
|
|
2299 slp->sl_slang = slang;
|
|
|
2300 }
|
|
310
|
2301 }
|
|
|
2302
|
|
|
2303 /*
|
|
|
2304 * Load one spell file and store the info into a slang_T.
|
|
|
2305 *
|
|
|
2306 * This is invoked in two ways:
|
|
|
2307 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
|
2308 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
|
2309 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
|
2310 * points to the existing slang_T.
|
|
323
|
2311 * Returns the slang_T the spell file was loaded into. NULL for error.
|
|
310
|
2312 */
|
|
323
|
2313 static slang_T *
|
|
|
2314 spell_load_file(fname, lang, old_lp, silent)
|
|
310
|
2315 char_u *fname;
|
|
|
2316 char_u *lang;
|
|
|
2317 slang_T *old_lp;
|
|
323
|
2318 int silent; /* no error if file doesn't exist */
|
|
310
|
2319 {
|
|
236
|
2320 FILE *fd;
|
|
483
|
2321 char_u buf[VIMSPELLMAGICL];
|
|
236
|
2322 char_u *p;
|
|
339
|
2323 char_u *bp;
|
|
|
2324 idx_T *ip;
|
|
236
|
2325 int i;
|
|
339
|
2326 int n;
|
|
300
|
2327 int len;
|
|
236
|
2328 int round;
|
|
|
2329 char_u *save_sourcing_name = sourcing_name;
|
|
|
2330 linenr_T save_sourcing_lnum = sourcing_lnum;
|
|
307
|
2331 slang_T *lp = NULL;
|
|
324
|
2332 idx_T idx;
|
|
344
|
2333 int c = 0;
|
|
483
|
2334 int res;
|
|
236
|
2335
|
|
310
|
2336 fd = mch_fopen((char *)fname, "r");
|
|
236
|
2337 if (fd == NULL)
|
|
|
2338 {
|
|
323
|
2339 if (!silent)
|
|
|
2340 EMSG2(_(e_notopen), fname);
|
|
|
2341 else if (p_verbose > 2)
|
|
|
2342 {
|
|
|
2343 verbose_enter();
|
|
|
2344 smsg((char_u *)e_notopen, fname);
|
|
|
2345 verbose_leave();
|
|
|
2346 }
|
|
255
|
2347 goto endFAIL;
|
|
236
|
2348 }
|
|
310
|
2349 if (p_verbose > 2)
|
|
|
2350 {
|
|
|
2351 verbose_enter();
|
|
|
2352 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
|
2353 verbose_leave();
|
|
|
2354 }
|
|
|
2355
|
|
|
2356 if (old_lp == NULL)
|
|
|
2357 {
|
|
|
2358 lp = slang_alloc(lang);
|
|
|
2359 if (lp == NULL)
|
|
|
2360 goto endFAIL;
|
|
|
2361
|
|
|
2362 /* Remember the file name, used to reload the file when it's updated. */
|
|
|
2363 lp->sl_fname = vim_strsave(fname);
|
|
|
2364 if (lp->sl_fname == NULL)
|
|
|
2365 goto endFAIL;
|
|
|
2366
|
|
|
2367 /* Check for .add.spl. */
|
|
|
2368 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
|
2369 }
|
|
|
2370 else
|
|
|
2371 lp = old_lp;
|
|
307
|
2372
|
|
236
|
2373 /* Set sourcing_name, so that error messages mention the file name. */
|
|
|
2374 sourcing_name = fname;
|
|
|
2375 sourcing_lnum = 0;
|
|
|
2376
|
|
339
|
2377 /* <HEADER>: <fileID>
|
|
|
2378 */
|
|
236
|
2379 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
|
2380 buf[i] = getc(fd); /* <fileID> */
|
|
|
2381 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
|
2382 {
|
|
483
|
2383 EMSG(_("E757: This does not look like a spell file"));
|
|
381
|
2384 goto endFAIL;
|
|
483
|
2385 }
|
|
|
2386 c = getc(fd); /* <versionnr> */
|
|
|
2387 if (c < VIMSPELLVERSION)
|
|
|
2388 {
|
|
|
2389 EMSG(_("E771: Old spell file, needs to be updated"));
|
|
381
|
2390 goto endFAIL;
|
|
|
2391 }
|
|
483
|
2392 else if (c > VIMSPELLVERSION)
|
|
|
2393 {
|
|
|
2394 EMSG(_("E772: Spell file is for newer version of Vim"));
|
|
481
|
2395 goto endFAIL;
|
|
483
|
2396 }
|
|
|
2397
|
|
|
2398
|
|
|
2399 /*
|
|
|
2400 * <SECTIONS>: <section> ... <sectionend>
|
|
|
2401 * <section>: <sectionID> <sectionflags> <sectionlen> (section contents)
|
|
|
2402 */
|
|
|
2403 for (;;)
|
|
|
2404 {
|
|
|
2405 n = getc(fd); /* <sectionID> or <sectionend> */
|
|
|
2406 if (n == SN_END)
|
|
|
2407 break;
|
|
|
2408 c = getc(fd); /* <sectionflags> */
|
|
|
2409 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2410 /* <sectionlen> */
|
|
|
2411 if (len < 0)
|
|
|
2412 goto truncerr;
|
|
|
2413
|
|
|
2414 res = 0;
|
|
|
2415 switch (n)
|
|
|
2416 {
|
|
|
2417 case SN_REGION:
|
|
|
2418 res = read_region_section(fd, lp, len);
|
|
|
2419 break;
|
|
|
2420
|
|
|
2421 case SN_CHARFLAGS:
|
|
|
2422 res = read_charflags_section(fd);
|
|
|
2423 break;
|
|
|
2424
|
|
|
2425 case SN_MIDWORD:
|
|
|
2426 lp->sl_midword = read_string(fd, len); /* <midword> */
|
|
|
2427 if (lp->sl_midword == NULL)
|
|
|
2428 goto endFAIL;
|
|
|
2429 break;
|
|
|
2430
|
|
|
2431 case SN_PREFCOND:
|
|
|
2432 res = read_prefcond_section(fd, lp);
|
|
|
2433 break;
|
|
|
2434
|
|
|
2435 case SN_REP:
|
|
|
2436 res = read_rep_section(fd, lp);
|
|
|
2437 break;
|
|
|
2438
|
|
|
2439 case SN_SAL:
|
|
|
2440 res = read_sal_section(fd, lp);
|
|
|
2441 break;
|
|
|
2442
|
|
|
2443 case SN_SOFO:
|
|
|
2444 res = read_sofo_section(fd, lp);
|
|
|
2445 break;
|
|
|
2446
|
|
|
2447 case SN_MAP:
|
|
|
2448 p = read_string(fd, len); /* <mapstr> */
|
|
|
2449 if (p == NULL)
|
|
|
2450 goto endFAIL;
|
|
|
2451 set_map_str(lp, p);
|
|
|
2452 vim_free(p);
|
|
|
2453 break;
|
|
|
2454
|
|
|
2455 case SN_COMPOUND:
|
|
|
2456 res = read_compound(fd, lp, len);
|
|
|
2457 break;
|
|
|
2458
|
|
492
|
2459 case SN_NOBREAK:
|
|
|
2460 lp->sl_nobreak = TRUE;
|
|
|
2461 break;
|
|
|
2462
|
|
483
|
2463 case SN_SYLLABLE:
|
|
|
2464 lp->sl_syllable = read_string(fd, len); /* <syllable> */
|
|
|
2465 if (lp->sl_syllable == NULL)
|
|
|
2466 goto endFAIL;
|
|
|
2467 if (init_syl_tab(lp) == FAIL)
|
|
|
2468 goto endFAIL;
|
|
|
2469 break;
|
|
|
2470
|
|
|
2471 default:
|
|
|
2472 /* Unsupported section. When it's required give an error
|
|
|
2473 * message. When it's not required skip the contents. */
|
|
|
2474 if (c & SNF_REQUIRED)
|
|
|
2475 {
|
|
|
2476 EMSG(_("E770: Unsupported section in spell file"));
|
|
|
2477 goto endFAIL;
|
|
|
2478 }
|
|
|
2479 while (--len >= 0)
|
|
|
2480 if (getc(fd) < 0)
|
|
|
2481 goto truncerr;
|
|
|
2482 break;
|
|
|
2483 }
|
|
|
2484 if (res == SP_FORMERROR)
|
|
|
2485 {
|
|
|
2486 formerr:
|
|
|
2487 EMSG(_(e_format));
|
|
381
|
2488 goto endFAIL;
|
|
344
|
2489 }
|
|
483
|
2490 if (res == SP_TRUNCERROR)
|
|
|
2491 {
|
|
|
2492 truncerr:
|
|
|
2493 EMSG(_(e_spell_trunc));
|
|
375
|
2494 goto endFAIL;
|
|
483
|
2495 }
|
|
495
|
2496 if (res == SP_OTHERERROR)
|
|
483
|
2497 goto endFAIL;
|
|
389
|
2498 }
|
|
236
|
2499
|
|
300
|
2500 /* round 1: <LWORDTREE>
|
|
339
|
2501 * round 2: <KWORDTREE>
|
|
|
2502 * round 3: <PREFIXTREE> */
|
|
|
2503 for (round = 1; round <= 3; ++round)
|
|
236
|
2504 {
|
|
300
|
2505 /* The tree size was computed when writing the file, so that we can
|
|
|
2506 * allocate it as one long block. <nodecount> */
|
|
|
2507 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
2508 if (len < 0)
|
|
|
2509 goto truncerr;
|
|
|
2510 if (len > 0)
|
|
236
|
2511 {
|
|
300
|
2512 /* Allocate the byte array. */
|
|
339
|
2513 bp = lalloc((long_u)len, TRUE);
|
|
|
2514 if (bp == NULL)
|
|
300
|
2515 goto endFAIL;
|
|
|
2516 if (round == 1)
|
|
339
|
2517 lp->sl_fbyts = bp;
|
|
|
2518 else if (round == 2)
|
|
|
2519 lp->sl_kbyts = bp;
|
|
300
|
2520 else
|
|
339
|
2521 lp->sl_pbyts = bp;
|
|
236
|
2522
|
|
300
|
2523 /* Allocate the index array. */
|
|
339
|
2524 ip = (idx_T *)lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
|
2525 if (ip == NULL)
|
|
300
|
2526 goto endFAIL;
|
|
|
2527 if (round == 1)
|
|
339
|
2528 lp->sl_fidxs = ip;
|
|
|
2529 else if (round == 2)
|
|
|
2530 lp->sl_kidxs = ip;
|
|
300
|
2531 else
|
|
339
|
2532 lp->sl_pidxs = ip;
|
|
300
|
2533
|
|
|
2534 /* Read the tree and store it in the array. */
|
|
339
|
2535 idx = read_tree(fd, bp, ip, len, 0, round == 3, lp->sl_prefixcnt);
|
|
324
|
2536 if (idx == -1)
|
|
300
|
2537 goto truncerr;
|
|
324
|
2538 if (idx < 0)
|
|
236
|
2539 goto formerr;
|
|
|
2540 }
|
|
300
|
2541 }
|
|
243
|
2542
|
|
310
|
2543 /* For a new file link it in the list of spell files. */
|
|
|
2544 if (old_lp == NULL)
|
|
|
2545 {
|
|
|
2546 lp->sl_next = first_lang;
|
|
|
2547 first_lang = lp;
|
|
|
2548 }
|
|
307
|
2549
|
|
255
|
2550 goto endOK;
|
|
|
2551
|
|
|
2552 endFAIL:
|
|
310
|
2553 if (lang != NULL)
|
|
|
2554 /* truncating the name signals the error to spell_load_lang() */
|
|
|
2555 *lang = NUL;
|
|
|
2556 if (lp != NULL && old_lp == NULL)
|
|
307
|
2557 slang_free(lp);
|
|
500
|
2558 lp = NULL;
|
|
255
|
2559
|
|
|
2560 endOK:
|
|
236
|
2561 if (fd != NULL)
|
|
|
2562 fclose(fd);
|
|
|
2563 sourcing_name = save_sourcing_name;
|
|
|
2564 sourcing_lnum = save_sourcing_lnum;
|
|
323
|
2565
|
|
|
2566 return lp;
|
|
236
|
2567 }
|
|
|
2568
|
|
376
|
2569 /*
|
|
|
2570 * Read a length field from "fd" in "cnt_bytes" bytes.
|
|
381
|
2571 * Allocate memory, read the string into it and add a NUL at the end.
|
|
376
|
2572 * Returns NULL when the count is zero.
|
|
483
|
2573 * Sets "*cntp" to SP_*ERROR when there is an error, length of the result
|
|
|
2574 * otherwise.
|
|
376
|
2575 */
|
|
|
2576 static char_u *
|
|
389
|
2577 read_cnt_string(fd, cnt_bytes, cntp)
|
|
376
|
2578 FILE *fd;
|
|
|
2579 int cnt_bytes;
|
|
389
|
2580 int *cntp;
|
|
376
|
2581 {
|
|
|
2582 int cnt = 0;
|
|
|
2583 int i;
|
|
|
2584 char_u *str;
|
|
|
2585
|
|
|
2586 /* read the length bytes, MSB first */
|
|
|
2587 for (i = 0; i < cnt_bytes; ++i)
|
|
|
2588 cnt = (cnt << 8) + getc(fd);
|
|
|
2589 if (cnt < 0)
|
|
|
2590 {
|
|
483
|
2591 *cntp = SP_TRUNCERROR;
|
|
376
|
2592 return NULL;
|
|
|
2593 }
|
|
389
|
2594 *cntp = cnt;
|
|
|
2595 if (cnt == 0)
|
|
|
2596 return NULL; /* nothing to read, return NULL */
|
|
376
|
2597
|
|
483
|
2598 str = read_string(fd, cnt);
|
|
|
2599 if (str == NULL)
|
|
495
|
2600 *cntp = SP_OTHERERROR;
|
|
483
|
2601 return str;
|
|
|
2602 }
|
|
|
2603
|
|
|
2604 /*
|
|
|
2605 * Read a string of length "cnt" from "fd" into allocated memory.
|
|
|
2606 * Returns NULL when out of memory.
|
|
|
2607 */
|
|
|
2608 static char_u *
|
|
|
2609 read_string(fd, cnt)
|
|
|
2610 FILE *fd;
|
|
|
2611 int cnt;
|
|
|
2612 {
|
|
|
2613 char_u *str;
|
|
|
2614 int i;
|
|
|
2615
|
|
376
|
2616 /* allocate memory */
|
|
|
2617 str = alloc((unsigned)cnt + 1);
|
|
483
|
2618 if (str != NULL)
|
|
|
2619 {
|
|
|
2620 /* Read the string. Doesn't check for truncated file. */
|
|
|
2621 for (i = 0; i < cnt; ++i)
|
|
|
2622 str[i] = getc(fd);
|
|
|
2623 str[i] = NUL;
|
|
|
2624 }
|
|
|
2625 return str;
|
|
|
2626 }
|
|
|
2627
|
|
|
2628 /*
|
|
|
2629 * Read SN_REGION: <regionname> ...
|
|
|
2630 * Return SP_*ERROR flags.
|
|
|
2631 */
|
|
|
2632 static int
|
|
|
2633 read_region_section(fd, lp, len)
|
|
|
2634 FILE *fd;
|
|
|
2635 slang_T *lp;
|
|
|
2636 int len;
|
|
|
2637 {
|
|
|
2638 int i;
|
|
|
2639
|
|
|
2640 if (len > 16)
|
|
|
2641 return SP_FORMERROR;
|
|
|
2642 for (i = 0; i < len; ++i)
|
|
|
2643 lp->sl_regions[i] = getc(fd); /* <regionname> */
|
|
|
2644 lp->sl_regions[len] = NUL;
|
|
|
2645 return 0;
|
|
|
2646 }
|
|
|
2647
|
|
|
2648 /*
|
|
|
2649 * Read SN_CHARFLAGS section: <charflagslen> <charflags>
|
|
|
2650 * <folcharslen> <folchars>
|
|
|
2651 * Return SP_*ERROR flags.
|
|
|
2652 */
|
|
|
2653 static int
|
|
|
2654 read_charflags_section(fd)
|
|
|
2655 FILE *fd;
|
|
|
2656 {
|
|
|
2657 char_u *flags;
|
|
|
2658 char_u *fol;
|
|
|
2659 int flagslen, follen;
|
|
|
2660
|
|
|
2661 /* <charflagslen> <charflags> */
|
|
|
2662 flags = read_cnt_string(fd, 1, &flagslen);
|
|
|
2663 if (flagslen < 0)
|
|
|
2664 return flagslen;
|
|
|
2665
|
|
|
2666 /* <folcharslen> <folchars> */
|
|
|
2667 fol = read_cnt_string(fd, 2, &follen);
|
|
|
2668 if (follen < 0)
|
|
|
2669 {
|
|
|
2670 vim_free(flags);
|
|
|
2671 return follen;
|
|
|
2672 }
|
|
|
2673
|
|
|
2674 /* Set the word-char flags and fill SPELL_ISUPPER() table. */
|
|
|
2675 if (flags != NULL && fol != NULL)
|
|
|
2676 set_spell_charflags(flags, flagslen, fol);
|
|
|
2677
|
|
|
2678 vim_free(flags);
|
|
|
2679 vim_free(fol);
|
|
|
2680
|
|
|
2681 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
|
2682 if ((flags == NULL) != (fol == NULL))
|
|
|
2683 return SP_FORMERROR;
|
|
|
2684 return 0;
|
|
|
2685 }
|
|
|
2686
|
|
|
2687 /*
|
|
|
2688 * Read SN_PREFCOND section.
|
|
|
2689 * Return SP_*ERROR flags.
|
|
|
2690 */
|
|
|
2691 static int
|
|
|
2692 read_prefcond_section(fd, lp)
|
|
|
2693 FILE *fd;
|
|
|
2694 slang_T *lp;
|
|
|
2695 {
|
|
|
2696 int cnt;
|
|
|
2697 int i;
|
|
|
2698 int n;
|
|
|
2699 char_u *p;
|
|
|
2700 char_u buf[MAXWLEN + 1];
|
|
|
2701
|
|
|
2702 /* <prefcondcnt> <prefcond> ... */
|
|
|
2703 cnt = (getc(fd) << 8) + getc(fd); /* <prefcondcnt> */
|
|
|
2704 if (cnt <= 0)
|
|
|
2705 return SP_FORMERROR;
|
|
|
2706
|
|
|
2707 lp->sl_prefprog = (regprog_T **)alloc_clear(
|
|
|
2708 (unsigned)sizeof(regprog_T *) * cnt);
|
|
|
2709 if (lp->sl_prefprog == NULL)
|
|
495
|
2710 return SP_OTHERERROR;
|
|
483
|
2711 lp->sl_prefixcnt = cnt;
|
|
|
2712
|
|
376
|
2713 for (i = 0; i < cnt; ++i)
|
|
483
|
2714 {
|
|
|
2715 /* <prefcond> : <condlen> <condstr> */
|
|
|
2716 n = getc(fd); /* <condlen> */
|
|
|
2717 if (n < 0 || n >= MAXWLEN)
|
|
|
2718 return SP_FORMERROR;
|
|
|
2719
|
|
|
2720 /* When <condlen> is zero we have an empty condition. Otherwise
|
|
|
2721 * compile the regexp program used to check for the condition. */
|
|
|
2722 if (n > 0)
|
|
|
2723 {
|
|
|
2724 buf[0] = '^'; /* always match at one position only */
|
|
|
2725 p = buf + 1;
|
|
|
2726 while (n-- > 0)
|
|
|
2727 *p++ = getc(fd); /* <condstr> */
|
|
|
2728 *p = NUL;
|
|
|
2729 lp->sl_prefprog[i] = vim_regcomp(buf, RE_MAGIC + RE_STRING);
|
|
|
2730 }
|
|
|
2731 }
|
|
|
2732 return 0;
|
|
|
2733 }
|
|
|
2734
|
|
|
2735 /*
|
|
|
2736 * Read REP items section from "fd": <repcount> <rep> ...
|
|
|
2737 * Return SP_*ERROR flags.
|
|
|
2738 */
|
|
|
2739 static int
|
|
|
2740 read_rep_section(fd, slang)
|
|
|
2741 FILE *fd;
|
|
|
2742 slang_T *slang;
|
|
|
2743 {
|
|
|
2744 int cnt;
|
|
|
2745 garray_T *gap;
|
|
|
2746 fromto_T *ftp;
|
|
|
2747 short *first;
|
|
|
2748 int i;
|
|
|
2749
|
|
|
2750 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> */
|
|
|
2751 if (cnt < 0)
|
|
|
2752 return SP_TRUNCERROR;
|
|
|
2753
|
|
|
2754 gap = &slang->sl_rep;
|
|
|
2755 if (ga_grow(gap, cnt) == FAIL)
|
|
495
|
2756 return SP_OTHERERROR;
|
|
483
|
2757
|
|
|
2758 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
2759 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2760 {
|
|
|
2761 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
|
2762 ftp->ft_from = read_cnt_string(fd, 1, &i);
|
|
|
2763 if (i < 0)
|
|
|
2764 return i;
|
|
|
2765 if (i == 0)
|
|
|
2766 return SP_FORMERROR;
|
|
|
2767 ftp->ft_to = read_cnt_string(fd, 1, &i);
|
|
|
2768 if (i <= 0)
|
|
|
2769 {
|
|
|
2770 vim_free(ftp->ft_from);
|
|
|
2771 if (i < 0)
|
|
|
2772 return i;
|
|
|
2773 return SP_FORMERROR;
|
|
|
2774 }
|
|
|
2775 }
|
|
|
2776
|
|
|
2777 /* Fill the first-index table. */
|
|
|
2778 first = slang->sl_rep_first;
|
|
|
2779 for (i = 0; i < 256; ++i)
|
|
|
2780 first[i] = -1;
|
|
|
2781 for (i = 0; i < gap->ga_len; ++i)
|
|
|
2782 {
|
|
|
2783 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
2784 if (first[*ftp->ft_from] == -1)
|
|
|
2785 first[*ftp->ft_from] = i;
|
|
|
2786 }
|
|
|
2787 return 0;
|
|
|
2788 }
|
|
|
2789
|
|
|
2790 /*
|
|
|
2791 * Read SN_SAL section: <salflags> <salcount> <sal> ...
|
|
|
2792 * Return SP_*ERROR flags.
|
|
|
2793 */
|
|
|
2794 static int
|
|
|
2795 read_sal_section(fd, slang)
|
|
|
2796 FILE *fd;
|
|
|
2797 slang_T *slang;
|
|
|
2798 {
|
|
|
2799 int i;
|
|
|
2800 int cnt;
|
|
|
2801 garray_T *gap;
|
|
|
2802 salitem_T *smp;
|
|
|
2803 int ccnt;
|
|
|
2804 char_u *p;
|
|
485
|
2805 int c = NUL;
|
|
483
|
2806
|
|
|
2807 slang->sl_sofo = FALSE;
|
|
|
2808
|
|
|
2809 i = getc(fd); /* <salflags> */
|
|
|
2810 if (i & SAL_F0LLOWUP)
|
|
|
2811 slang->sl_followup = TRUE;
|
|
|
2812 if (i & SAL_COLLAPSE)
|
|
|
2813 slang->sl_collapse = TRUE;
|
|
|
2814 if (i & SAL_REM_ACCENTS)
|
|
|
2815 slang->sl_rem_accents = TRUE;
|
|
|
2816
|
|
|
2817 cnt = (getc(fd) << 8) + getc(fd); /* <salcount> */
|
|
|
2818 if (cnt < 0)
|
|
|
2819 return SP_TRUNCERROR;
|
|
|
2820
|
|
|
2821 gap = &slang->sl_sal;
|
|
|
2822 ga_init2(gap, sizeof(salitem_T), 10);
|
|
548
|
2823 if (ga_grow(gap, cnt + 1) == FAIL)
|
|
495
|
2824 return SP_OTHERERROR;
|
|
483
|
2825
|
|
|
2826 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
2827 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
|
2828 {
|
|
|
2829 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2830 ccnt = getc(fd); /* <salfromlen> */
|
|
|
2831 if (ccnt < 0)
|
|
|
2832 return SP_TRUNCERROR;
|
|
|
2833 if ((p = alloc(ccnt + 2)) == NULL)
|
|
495
|
2834 return SP_OTHERERROR;
|
|
483
|
2835 smp->sm_lead = p;
|
|
|
2836
|
|
|
2837 /* Read up to the first special char into sm_lead. */
|
|
|
2838 for (i = 0; i < ccnt; ++i)
|
|
|
2839 {
|
|
|
2840 c = getc(fd); /* <salfrom> */
|
|
|
2841 if (vim_strchr((char_u *)"0123456789(-<^$", c) != NULL)
|
|
|
2842 break;
|
|
|
2843 *p++ = c;
|
|
|
2844 }
|
|
|
2845 smp->sm_leadlen = p - smp->sm_lead;
|
|
|
2846 *p++ = NUL;
|
|
|
2847
|
|
|
2848 /* Put (abc) chars in sm_oneof, if any. */
|
|
|
2849 if (c == '(')
|
|
|
2850 {
|
|
|
2851 smp->sm_oneof = p;
|
|
|
2852 for (++i; i < ccnt; ++i)
|
|
|
2853 {
|
|
|
2854 c = getc(fd); /* <salfrom> */
|
|
|
2855 if (c == ')')
|
|
|
2856 break;
|
|
|
2857 *p++ = c;
|
|
|
2858 }
|
|
|
2859 *p++ = NUL;
|
|
|
2860 if (++i < ccnt)
|
|
|
2861 c = getc(fd);
|
|
|
2862 }
|
|
|
2863 else
|
|
|
2864 smp->sm_oneof = NULL;
|
|
|
2865
|
|
|
2866 /* Any following chars go in sm_rules. */
|
|
|
2867 smp->sm_rules = p;
|
|
|
2868 if (i < ccnt)
|
|
|
2869 /* store the char we got while checking for end of sm_lead */
|
|
|
2870 *p++ = c;
|
|
|
2871 for (++i; i < ccnt; ++i)
|
|
|
2872 *p++ = getc(fd); /* <salfrom> */
|
|
|
2873 *p++ = NUL;
|
|
|
2874
|
|
|
2875 /* <saltolen> <salto> */
|
|
|
2876 smp->sm_to = read_cnt_string(fd, 1, &ccnt);
|
|
|
2877 if (ccnt < 0)
|
|
|
2878 {
|
|
|
2879 vim_free(smp->sm_lead);
|
|
|
2880 return ccnt;
|
|
|
2881 }
|
|
|
2882
|
|
|
2883 #ifdef FEAT_MBYTE
|
|
|
2884 if (has_mbyte)
|
|
|
2885 {
|
|
|
2886 /* convert the multi-byte strings to wide char strings */
|
|
|
2887 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2888 smp->sm_leadlen = mb_charlen(smp->sm_lead);
|
|
|
2889 if (smp->sm_oneof == NULL)
|
|
|
2890 smp->sm_oneof_w = NULL;
|
|
|
2891 else
|
|
|
2892 smp->sm_oneof_w = mb_str2wide(smp->sm_oneof);
|
|
|
2893 if (smp->sm_to == NULL)
|
|
|
2894 smp->sm_to_w = NULL;
|
|
|
2895 else
|
|
|
2896 smp->sm_to_w = mb_str2wide(smp->sm_to);
|
|
|
2897 if (smp->sm_lead_w == NULL
|
|
|
2898 || (smp->sm_oneof_w == NULL && smp->sm_oneof != NULL)
|
|
|
2899 || (smp->sm_to_w == NULL && smp->sm_to != NULL))
|
|
|
2900 {
|
|
|
2901 vim_free(smp->sm_lead);
|
|
|
2902 vim_free(smp->sm_to);
|
|
|
2903 vim_free(smp->sm_lead_w);
|
|
|
2904 vim_free(smp->sm_oneof_w);
|
|
|
2905 vim_free(smp->sm_to_w);
|
|
495
|
2906 return SP_OTHERERROR;
|
|
483
|
2907 }
|
|
|
2908 }
|
|
|
2909 #endif
|
|
|
2910 }
|
|
|
2911
|
|
548
|
2912 if (gap->ga_len > 0)
|
|
|
2913 {
|
|
|
2914 /* Add one extra entry to mark the end with an empty sm_lead. Avoids
|
|
|
2915 * that we need to check the index every time. */
|
|
|
2916 smp = &((salitem_T *)gap->ga_data)[gap->ga_len];
|
|
|
2917 if ((p = alloc(1)) == NULL)
|
|
|
2918 return SP_OTHERERROR;
|
|
|
2919 p[0] = NUL;
|
|
|
2920 smp->sm_lead = p;
|
|
|
2921 smp->sm_leadlen = 0;
|
|
|
2922 smp->sm_oneof = NULL;
|
|
|
2923 smp->sm_rules = p;
|
|
|
2924 smp->sm_to = NULL;
|
|
|
2925 #ifdef FEAT_MBYTE
|
|
|
2926 if (has_mbyte)
|
|
|
2927 {
|
|
|
2928 smp->sm_lead_w = mb_str2wide(smp->sm_lead);
|
|
|
2929 smp->sm_leadlen = 0;
|
|
|
2930 smp->sm_oneof_w = NULL;
|
|
|
2931 smp->sm_to_w = NULL;
|
|
|
2932 }
|
|
|
2933 #endif
|
|
|
2934 ++gap->ga_len;
|
|
|
2935 }
|
|
|
2936
|
|
483
|
2937 /* Fill the first-index table. */
|
|
|
2938 set_sal_first(slang);
|
|
|
2939
|
|
|
2940 return 0;
|
|
|
2941 }
|
|
|
2942
|
|
|
2943 /*
|
|
|
2944 * SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
2945 * Return SP_*ERROR flags.
|
|
|
2946 */
|
|
|
2947 static int
|
|
|
2948 read_sofo_section(fd, slang)
|
|
|
2949 FILE *fd;
|
|
|
2950 slang_T *slang;
|
|
|
2951 {
|
|
|
2952 int cnt;
|
|
|
2953 char_u *from, *to;
|
|
|
2954 int res;
|
|
|
2955
|
|
|
2956 slang->sl_sofo = TRUE;
|
|
|
2957
|
|
|
2958 /* <sofofromlen> <sofofrom> */
|
|
|
2959 from = read_cnt_string(fd, 2, &cnt);
|
|
|
2960 if (cnt < 0)
|
|
|
2961 return cnt;
|
|
|
2962
|
|
|
2963 /* <sofotolen> <sofoto> */
|
|
|
2964 to = read_cnt_string(fd, 2, &cnt);
|
|
|
2965 if (cnt < 0)
|
|
|
2966 {
|
|
|
2967 vim_free(from);
|
|
|
2968 return cnt;
|
|
|
2969 }
|
|
|
2970
|
|
|
2971 /* Store the info in slang->sl_sal and/or slang->sl_sal_first. */
|
|
|
2972 if (from != NULL && to != NULL)
|
|
|
2973 res = set_sofo(slang, from, to);
|
|
|
2974 else if (from != NULL || to != NULL)
|
|
|
2975 res = SP_FORMERROR; /* only one of two strings is an error */
|
|
|
2976 else
|
|
|
2977 res = 0;
|
|
|
2978
|
|
|
2979 vim_free(from);
|
|
|
2980 vim_free(to);
|
|
|
2981 return res;
|
|
|
2982 }
|
|
|
2983
|
|
|
2984 /*
|
|
|
2985 * Read the compound section from the .spl file:
|
|
|
2986 * <compmax> <compminlen> <compsylmax> <compflags>
|
|
|
2987 * Returns SP_*ERROR flags.
|
|
|
2988 */
|
|
|
2989 static int
|
|
|
2990 read_compound(fd, slang, len)
|
|
|
2991 FILE *fd;
|
|
|
2992 slang_T *slang;
|
|
|
2993 int len;
|
|
|
2994 {
|
|
|
2995 int todo = len;
|
|
|
2996 int c;
|
|
|
2997 int atstart;
|
|
|
2998 char_u *pat;
|
|
|
2999 char_u *pp;
|
|
|
3000 char_u *cp;
|
|
485
|
3001 char_u *ap;
|
|
483
|
3002
|
|
|
3003 if (todo < 2)
|
|
|
3004 return SP_FORMERROR; /* need at least two bytes */
|
|
|
3005
|
|
|
3006 --todo;
|
|
|
3007 c = getc(fd); /* <compmax> */
|
|
|
3008 if (c < 2)
|
|
|
3009 c = MAXWLEN;
|
|
|
3010 slang->sl_compmax = c;
|
|
|
3011
|
|
|
3012 --todo;
|
|
|
3013 c = getc(fd); /* <compminlen> */
|
|
|
3014 if (c < 1)
|
|
501
|
3015 c = 0;
|
|
483
|
3016 slang->sl_compminlen = c;
|
|
|
3017
|
|
|
3018 --todo;
|
|
|
3019 c = getc(fd); /* <compsylmax> */
|
|
|
3020 if (c < 1)
|
|
|
3021 c = MAXWLEN;
|
|
|
3022 slang->sl_compsylmax = c;
|
|
|
3023
|
|
|
3024 /* Turn the COMPOUNDFLAGS items into a regexp pattern:
|
|
|
3025 * "a[bc]/a*b+" -> "^\(a[bc]\|a*b\+\)$".
|
|
495
|
3026 * Inserting backslashes may double the length, "^\(\)$<Nul>" is 7 bytes.
|
|
|
3027 * Conversion to utf-8 may double the size. */
|
|
|
3028 c = todo * 2 + 7;
|
|
|
3029 #ifdef FEAT_MBYTE
|
|
|
3030 if (enc_utf8)
|
|
|
3031 c += todo * 2;
|
|
|
3032 #endif
|
|
|
3033 pat = alloc((unsigned)c);
|
|
483
|
3034 if (pat == NULL)
|
|
495
|
3035 return SP_OTHERERROR;
|
|
483
|
3036
|
|
485
|
3037 /* We also need a list of all flags that can appear at the start and one
|
|
|
3038 * for all flags. */
|
|
483
|
3039 cp = alloc(todo + 1);
|
|
|
3040 if (cp == NULL)
|
|
|
3041 {
|
|
|
3042 vim_free(pat);
|
|
495
|
3043 return SP_OTHERERROR;
|
|
483
|
3044 }
|
|
|
3045 slang->sl_compstartflags = cp;
|
|
|
3046 *cp = NUL;
|
|
|
3047
|
|
485
|
3048 ap = alloc(todo + 1);
|
|
|
3049 if (ap == NULL)
|
|
|
3050 {
|
|
|
3051 vim_free(pat);
|
|
495
|
3052 return SP_OTHERERROR;
|
|
485
|
3053 }
|
|
|
3054 slang->sl_compallflags = ap;
|
|
|
3055 *ap = NUL;
|
|
|
3056
|
|
483
|
3057 pp = pat;
|
|
|
3058 *pp++ = '^';
|
|
|
3059 *pp++ = '\\';
|
|
|
3060 *pp++ = '(';
|
|
|
3061
|
|
|
3062 atstart = 1;
|
|
|
3063 while (todo-- > 0)
|
|
|
3064 {
|
|
|
3065 c = getc(fd); /* <compflags> */
|
|
485
|
3066
|
|
|
3067 /* Add all flags to "sl_compallflags". */
|
|
|
3068 if (vim_strchr((char_u *)"+*[]/", c) == NULL
|
|
495
|
3069 && !byte_in_str(slang->sl_compallflags, c))
|
|
485
|
3070 {
|
|
|
3071 *ap++ = c;
|
|
|
3072 *ap = NUL;
|
|
|
3073 }
|
|
|
3074
|
|
483
|
3075 if (atstart != 0)
|
|
|
3076 {
|
|
|
3077 /* At start of item: copy flags to "sl_compstartflags". For a
|
|
|
3078 * [abc] item set "atstart" to 2 and copy up to the ']'. */
|
|
|
3079 if (c == '[')
|
|
|
3080 atstart = 2;
|
|
|
3081 else if (c == ']')
|
|
|
3082 atstart = 0;
|
|
|
3083 else
|
|
|
3084 {
|
|
495
|
3085 if (!byte_in_str(slang->sl_compstartflags, c))
|
|
483
|
3086 {
|
|
|
3087 *cp++ = c;
|
|
|
3088 *cp = NUL;
|
|
|
3089 }
|
|
|
3090 if (atstart == 1)
|
|
|
3091 atstart = 0;
|
|
|
3092 }
|
|
|
3093 }
|
|
|
3094 if (c == '/') /* slash separates two items */
|
|
|
3095 {
|
|
|
3096 *pp++ = '\\';
|
|
|
3097 *pp++ = '|';
|
|
|
3098 atstart = 1;
|
|
|
3099 }
|
|
|
3100 else /* normal char, "[abc]" and '*' are copied as-is */
|
|
|
3101 {
|
|
500
|
3102 if (c == '+' || c == '~')
|
|
483
|
3103 *pp++ = '\\'; /* "a+" becomes "a\+" */
|
|
495
|
3104 #ifdef FEAT_MBYTE
|
|
|
3105 if (enc_utf8)
|
|
|
3106 pp += mb_char2bytes(c, pp);
|
|
|
3107 else
|
|
|
3108 #endif
|
|
|
3109 *pp++ = c;
|
|
483
|
3110 }
|
|
|
3111 }
|
|
|
3112
|
|
|
3113 *pp++ = '\\';
|
|
|
3114 *pp++ = ')';
|
|
|
3115 *pp++ = '$';
|
|
|
3116 *pp = NUL;
|
|
|
3117
|
|
|
3118 slang->sl_compprog = vim_regcomp(pat, RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
3119 vim_free(pat);
|
|
|
3120 if (slang->sl_compprog == NULL)
|
|
|
3121 return SP_FORMERROR;
|
|
|
3122
|
|
|
3123 return 0;
|
|
|
3124 }
|
|
|
3125
|
|
495
|
3126 /*
|
|
498
|
3127 * Return TRUE if byte "n" appears in "str".
|
|
495
|
3128 * Like strchr() but independent of locale.
|
|
|
3129 */
|
|
|
3130 static int
|
|
498
|
3131 byte_in_str(str, n)
|
|
495
|
3132 char_u *str;
|
|
498
|
3133 int n;
|
|
495
|
3134 {
|
|
|
3135 char_u *p;
|
|
|
3136
|
|
|
3137 for (p = str; *p != NUL; ++p)
|
|
498
|
3138 if (*p == n)
|
|
495
|
3139 return TRUE;
|
|
|
3140 return FALSE;
|
|
|
3141 }
|
|
|
3142
|
|
483
|
3143 #define SY_MAXLEN 30
|
|
|
3144 typedef struct syl_item_S
|
|
|
3145 {
|
|
|
3146 char_u sy_chars[SY_MAXLEN]; /* the sequence of chars */
|
|
|
3147 int sy_len;
|
|
|
3148 } syl_item_T;
|
|
|
3149
|
|
|
3150 /*
|
|
|
3151 * Truncate "slang->sl_syllable" at the first slash and put the following items
|
|
|
3152 * in "slang->sl_syl_items".
|
|
|
3153 */
|
|
|
3154 static int
|
|
|
3155 init_syl_tab(slang)
|
|
|
3156 slang_T *slang;
|
|
|
3157 {
|
|
|
3158 char_u *p;
|
|
|
3159 char_u *s;
|
|
|
3160 int l;
|
|
|
3161 syl_item_T *syl;
|
|
|
3162
|
|
|
3163 ga_init2(&slang->sl_syl_items, sizeof(syl_item_T), 4);
|
|
|
3164 p = vim_strchr(slang->sl_syllable, '/');
|
|
|
3165 while (p != NULL)
|
|
|
3166 {
|
|
|
3167 *p++ = NUL;
|
|
495
|
3168 if (*p == NUL) /* trailing slash */
|
|
483
|
3169 break;
|
|
|
3170 s = p;
|
|
|
3171 p = vim_strchr(p, '/');
|
|
|
3172 if (p == NULL)
|
|
|
3173 l = STRLEN(s);
|
|
|
3174 else
|
|
|
3175 l = p - s;
|
|
|
3176 if (l >= SY_MAXLEN)
|
|
|
3177 return SP_FORMERROR;
|
|
|
3178 if (ga_grow(&slang->sl_syl_items, 1) == FAIL)
|
|
495
|
3179 return SP_OTHERERROR;
|
|
483
|
3180 syl = ((syl_item_T *)slang->sl_syl_items.ga_data)
|
|
|
3181 + slang->sl_syl_items.ga_len++;
|
|
|
3182 vim_strncpy(syl->sy_chars, s, l);
|
|
|
3183 syl->sy_len = l;
|
|
|
3184 }
|
|
|
3185 return OK;
|
|
|
3186 }
|
|
|
3187
|
|
|
3188 /*
|
|
|
3189 * Count the number of syllables in "word".
|
|
|
3190 * When "word" contains spaces the syllables after the last space are counted.
|
|
|
3191 * Returns zero if syllables are not defines.
|
|
|
3192 */
|
|
|
3193 static int
|
|
|
3194 count_syllables(slang, word)
|
|
|
3195 slang_T *slang;
|
|
|
3196 char_u *word;
|
|
|
3197 {
|
|
|
3198 int cnt = 0;
|
|
|
3199 int skip = FALSE;
|
|
|
3200 char_u *p;
|
|
|
3201 int len;
|
|
|
3202 int i;
|
|
|
3203 syl_item_T *syl;
|
|
|
3204 int c;
|
|
|
3205
|
|
|
3206 if (slang->sl_syllable == NULL)
|
|
|
3207 return 0;
|
|
|
3208
|
|
|
3209 for (p = word; *p != NUL; p += len)
|
|
|
3210 {
|
|
|
3211 /* When running into a space reset counter. */
|
|
|
3212 if (*p == ' ')
|
|
|
3213 {
|
|
|
3214 len = 1;
|
|
|
3215 cnt = 0;
|
|
|
3216 continue;
|
|
|
3217 }
|
|
|
3218
|
|
|
3219 /* Find longest match of syllable items. */
|
|
|
3220 len = 0;
|
|
|
3221 for (i = 0; i < slang->sl_syl_items.ga_len; ++i)
|
|
|
3222 {
|
|
|
3223 syl = ((syl_item_T *)slang->sl_syl_items.ga_data) + i;
|
|
|
3224 if (syl->sy_len > len
|
|
|
3225 && STRNCMP(p, syl->sy_chars, syl->sy_len) == 0)
|
|
|
3226 len = syl->sy_len;
|
|
|
3227 }
|
|
|
3228 if (len != 0) /* found a match, count syllable */
|
|
|
3229 {
|
|
|
3230 ++cnt;
|
|
|
3231 skip = FALSE;
|
|
|
3232 }
|
|
|
3233 else
|
|
|
3234 {
|
|
|
3235 /* No recognized syllable item, at least a syllable char then? */
|
|
|
3236 #ifdef FEAT_MBYTE
|
|
|
3237 c = mb_ptr2char(p);
|
|
|
3238 len = (*mb_ptr2len)(p);
|
|
|
3239 #else
|
|
|
3240 c = *p;
|
|
|
3241 len = 1;
|
|
|
3242 #endif
|
|
|
3243 if (vim_strchr(slang->sl_syllable, c) == NULL)
|
|
|
3244 skip = FALSE; /* No, search for next syllable */
|
|
|
3245 else if (!skip)
|
|
|
3246 {
|
|
|
3247 ++cnt; /* Yes, count it */
|
|
|
3248 skip = TRUE; /* don't count following syllable chars */
|
|
|
3249 }
|
|
|
3250 }
|
|
|
3251 }
|
|
|
3252 return cnt;
|
|
376
|
3253 }
|
|
|
3254
|
|
381
|
3255 /*
|
|
|
3256 * Set the SOFOFROM and SOFOTO items in language "lp".
|
|
483
|
3257 * Returns SP_*ERROR flags when there is something wrong.
|
|
381
|
3258 */
|
|
|
3259 static int
|
|
|
3260 set_sofo(lp, from, to)
|
|
|
3261 slang_T *lp;
|
|
|
3262 char_u *from;
|
|
|
3263 char_u *to;
|
|
|
3264 {
|
|
|
3265 int i;
|
|
|
3266
|
|
|
3267 #ifdef FEAT_MBYTE
|
|
|
3268 garray_T *gap;
|
|
|
3269 char_u *s;
|
|
|
3270 char_u *p;
|
|
|
3271 int c;
|
|
|
3272 int *inp;
|
|
|
3273
|
|
|
3274 if (has_mbyte)
|
|
|
3275 {
|
|
|
3276 /* Use "sl_sal" as an array with 256 pointers to a list of wide
|
|
|
3277 * characters. The index is the low byte of the character.
|
|
|
3278 * The list contains from-to pairs with a terminating NUL.
|
|
|
3279 * sl_sal_first[] is used for latin1 "from" characters. */
|
|
|
3280 gap = &lp->sl_sal;
|
|
|
3281 ga_init2(gap, sizeof(int *), 1);
|
|
|
3282 if (ga_grow(gap, 256) == FAIL)
|
|
495
|
3283 return SP_OTHERERROR;
|
|
381
|
3284 vim_memset(gap->ga_data, 0, sizeof(int *) * 256);
|
|
|
3285 gap->ga_len = 256;
|
|
|
3286
|
|
|
3287 /* First count the number of items for each list. Temporarily use
|
|
|
3288 * sl_sal_first[] for this. */
|
|
|
3289 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3290 {
|
|
474
|
3291 c = mb_cptr2char_adv(&p);
|
|
|
3292 mb_cptr_adv(s);
|
|
381
|
3293 if (c >= 256)
|
|
|
3294 ++lp->sl_sal_first[c & 0xff];
|
|
|
3295 }
|
|
|
3296 if (*p != NUL || *s != NUL) /* lengths differ */
|
|
483
|
3297 return SP_FORMERROR;
|
|
381
|
3298
|
|
|
3299 /* Allocate the lists. */
|
|
|
3300 for (i = 0; i < 256; ++i)
|
|
|
3301 if (lp->sl_sal_first[i] > 0)
|
|
|
3302 {
|
|
|
3303 p = alloc(sizeof(int) * (lp->sl_sal_first[i] * 2 + 1));
|
|
|
3304 if (p == NULL)
|
|
495
|
3305 return SP_OTHERERROR;
|
|
381
|
3306 ((int **)gap->ga_data)[i] = (int *)p;
|
|
|
3307 *(int *)p = 0;
|
|
|
3308 }
|
|
|
3309
|
|
|
3310 /* Put the characters up to 255 in sl_sal_first[] the rest in a sl_sal
|
|
|
3311 * list. */
|
|
|
3312 vim_memset(lp->sl_sal_first, 0, sizeof(salfirst_T) * 256);
|
|
|
3313 for (p = from, s = to; *p != NUL && *s != NUL; )
|
|
|
3314 {
|
|
474
|
3315 c = mb_cptr2char_adv(&p);
|
|
|
3316 i = mb_cptr2char_adv(&s);
|
|
381
|
3317 if (c >= 256)
|
|
|
3318 {
|
|
|
3319 /* Append the from-to chars at the end of the list with
|
|
|
3320 * the low byte. */
|
|
|
3321 inp = ((int **)gap->ga_data)[c & 0xff];
|
|
|
3322 while (*inp != 0)
|
|
|
3323 ++inp;
|
|
|
3324 *inp++ = c; /* from char */
|
|
|
3325 *inp++ = i; /* to char */
|
|
|
3326 *inp++ = NUL; /* NUL at the end */
|
|
|
3327 }
|
|
|
3328 else
|
|
|
3329 /* mapping byte to char is done in sl_sal_first[] */
|
|
|
3330 lp->sl_sal_first[c] = i;
|
|
|
3331 }
|
|
|
3332 }
|
|
|
3333 else
|
|
|
3334 #endif
|
|
|
3335 {
|
|
|
3336 /* mapping bytes to bytes is done in sl_sal_first[] */
|
|
|
3337 if (STRLEN(from) != STRLEN(to))
|
|
483
|
3338 return SP_FORMERROR;
|
|
381
|
3339
|
|
|
3340 for (i = 0; to[i] != NUL; ++i)
|
|
|
3341 lp->sl_sal_first[from[i]] = to[i];
|
|
|
3342 lp->sl_sal.ga_len = 1; /* indicates we have soundfolding */
|
|
|
3343 }
|
|
|
3344
|
|
483
|
3345 return 0;
|
|
381
|
3346 }
|
|
|
3347
|
|
|
3348 /*
|
|
|
3349 * Fill the first-index table for "lp".
|
|
|
3350 */
|
|
|
3351 static void
|
|
|
3352 set_sal_first(lp)
|
|
|
3353 slang_T *lp;
|
|
|
3354 {
|
|
|
3355 salfirst_T *sfirst;
|
|
|
3356 int i;
|
|
|
3357 salitem_T *smp;
|
|
|
3358 int c;
|
|
|
3359 garray_T *gap = &lp->sl_sal;
|
|
|
3360
|
|
|
3361 sfirst = lp->sl_sal_first;
|
|
|
3362 for (i = 0; i < 256; ++i)
|
|
|
3363 sfirst[i] = -1;
|
|
|
3364 smp = (salitem_T *)gap->ga_data;
|
|
|
3365 for (i = 0; i < gap->ga_len; ++i)
|
|
|
3366 {
|
|
|
3367 #ifdef FEAT_MBYTE
|
|
|
3368 if (has_mbyte)
|
|
|
3369 /* Use the lowest byte of the first character. For latin1 it's
|
|
|
3370 * the character, for other encodings it should differ for most
|
|
|
3371 * characters. */
|
|
|
3372 c = *smp[i].sm_lead_w & 0xff;
|
|
|
3373 else
|
|
|
3374 #endif
|
|
|
3375 c = *smp[i].sm_lead;
|
|
|
3376 if (sfirst[c] == -1)
|
|
|
3377 {
|
|
|
3378 sfirst[c] = i;
|
|
|
3379 #ifdef FEAT_MBYTE
|
|
|
3380 if (has_mbyte)
|
|
|
3381 {
|
|
|
3382 int n;
|
|
|
3383
|
|
|
3384 /* Make sure all entries with this byte are following each
|
|
|
3385 * other. Move the ones that are in the wrong position. Do
|
|
|
3386 * keep the same ordering! */
|
|
|
3387 while (i + 1 < gap->ga_len
|
|
|
3388 && (*smp[i + 1].sm_lead_w & 0xff) == c)
|
|
|
3389 /* Skip over entry with same index byte. */
|
|
|
3390 ++i;
|
|
|
3391
|
|
|
3392 for (n = 1; i + n < gap->ga_len; ++n)
|
|
|
3393 if ((*smp[i + n].sm_lead_w & 0xff) == c)
|
|
|
3394 {
|
|
|
3395 salitem_T tsal;
|
|
|
3396
|
|
|
3397 /* Move entry with same index byte after the entries
|
|
|
3398 * we already found. */
|
|
|
3399 ++i;
|
|
|
3400 --n;
|
|
|
3401 tsal = smp[i + n];
|
|
|
3402 mch_memmove(smp + i + 1, smp + i,
|
|
|
3403 sizeof(salitem_T) * n);
|
|
|
3404 smp[i] = tsal;
|
|
|
3405 }
|
|
|
3406 }
|
|
|
3407 #endif
|
|
|
3408 }
|
|
|
3409 }
|
|
|
3410 }
|
|
376
|
3411
|
|
372
|
3412 #ifdef FEAT_MBYTE
|
|
|
3413 /*
|
|
|
3414 * Turn a multi-byte string into a wide character string.
|
|
|
3415 * Return it in allocated memory (NULL for out-of-memory)
|
|
|
3416 */
|
|
|
3417 static int *
|
|
|
3418 mb_str2wide(s)
|
|
|
3419 char_u *s;
|
|
|
3420 {
|
|
|
3421 int *res;
|
|
|
3422 char_u *p;
|
|
|
3423 int i = 0;
|
|
|
3424
|
|
|
3425 res = (int *)alloc(sizeof(int) * (mb_charlen(s) + 1));
|
|
|
3426 if (res != NULL)
|
|
|
3427 {
|
|
|
3428 for (p = s; *p != NUL; )
|
|
|
3429 res[i++] = mb_ptr2char_adv(&p);
|
|
|
3430 res[i] = NUL;
|
|
|
3431 }
|
|
|
3432 return res;
|
|
|
3433 }
|
|
|
3434 #endif
|
|
|
3435
|
|
236
|
3436 /*
|
|
300
|
3437 * Read one row of siblings from the spell file and store it in the byte array
|
|
|
3438 * "byts" and index array "idxs". Recursively read the children.
|
|
|
3439 *
|
|
346
|
3440 * NOTE: The code here must match put_node().
|
|
300
|
3441 *
|
|
|
3442 * Returns the index follosing the siblings.
|
|
|
3443 * Returns -1 if the file is shorter than expected.
|
|
|
3444 * Returns -2 if there is a format error.
|
|
236
|
3445 */
|
|
324
|
3446 static idx_T
|
|
339
|
3447 read_tree(fd, byts, idxs, maxidx, startidx, prefixtree, maxprefcondnr)
|
|
300
|
3448 FILE *fd;
|
|
|
3449 char_u *byts;
|
|
324
|
3450 idx_T *idxs;
|
|
300
|
3451 int maxidx; /* size of arrays */
|
|
324
|
3452 idx_T startidx; /* current index in "byts" and "idxs" */
|
|
339
|
3453 int prefixtree; /* TRUE for reading PREFIXTREE */
|
|
|
3454 int maxprefcondnr; /* maximum for <prefcondnr> */
|
|
236
|
3455 {
|
|
300
|
3456 int len;
|
|
|
3457 int i;
|
|
|
3458 int n;
|
|
324
|
3459 idx_T idx = startidx;
|
|
300
|
3460 int c;
|
|
366
|
3461 int c2;
|
|
300
|
3462 #define SHARED_MASK 0x8000000
|
|
236
|
3463
|
|
300
|
3464 len = getc(fd); /* <siblingcount> */
|
|
|
3465 if (len <= 0)
|
|
|
3466 return -1;
|
|
|
3467
|
|
|
3468 if (startidx + len >= maxidx)
|
|
|
3469 return -2;
|
|
|
3470 byts[idx++] = len;
|
|
|
3471
|
|
|
3472 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
|
3473 for (i = 1; i <= len; ++i)
|
|
236
|
3474 {
|
|
300
|
3475 c = getc(fd); /* <byte> */
|
|
|
3476 if (c < 0)
|
|
|
3477 return -1;
|
|
|
3478 if (c <= BY_SPECIAL)
|
|
|
3479 {
|
|
366
|
3480 if (c == BY_NOFLAGS && !prefixtree)
|
|
300
|
3481 {
|
|
|
3482 /* No flags, all regions. */
|
|
|
3483 idxs[idx] = 0;
|
|
|
3484 c = 0;
|
|
|
3485 }
|
|
390
|
3486 else if (c != BY_INDEX)
|
|
300
|
3487 {
|
|
339
|
3488 if (prefixtree)
|
|
|
3489 {
|
|
455
|
3490 /* Read the optional pflags byte, the prefix ID and the
|
|
|
3491 * condition nr. In idxs[] store the prefix ID in the low
|
|
|
3492 * byte, the condition index shifted up 8 bits, the flags
|
|
|
3493 * shifted up 24 bits. */
|
|
|
3494 if (c == BY_FLAGS)
|
|
|
3495 c = getc(fd) << 24; /* <pflags> */
|
|
|
3496 else
|
|
|
3497 c = 0;
|
|
|
3498
|
|
481
|
3499 c |= getc(fd); /* <affixID> */
|
|
455
|
3500
|
|
339
|
3501 n = (getc(fd) << 8) + getc(fd); /* <prefcondnr> */
|
|
|
3502 if (n >= maxprefcondnr)
|
|
|
3503 return -2;
|
|
455
|
3504 c |= (n << 8);
|
|
339
|
3505 }
|
|
390
|
3506 else /* c must be BY_FLAGS or BY_FLAGS2 */
|
|
339
|
3507 {
|
|
|
3508 /* Read flags and optional region and prefix ID. In
|
|
390
|
3509 * idxs[] the flags go in the low two bytes, region above
|
|
|
3510 * that and prefix ID above the region. */
|
|
|
3511 c2 = c;
|
|
339
|
3512 c = getc(fd); /* <flags> */
|
|
390
|
3513 if (c2 == BY_FLAGS2)
|
|
|
3514 c = (getc(fd) << 8) + c; /* <flags2> */
|
|
339
|
3515 if (c & WF_REGION)
|
|
390
|
3516 c = (getc(fd) << 16) + c; /* <region> */
|
|
481
|
3517 if (c & WF_AFX)
|
|
|
3518 c = (getc(fd) << 24) + c; /* <affixID> */
|
|
339
|
3519 }
|
|
|
3520
|
|
300
|
3521 idxs[idx] = c;
|
|
|
3522 c = 0;
|
|
|
3523 }
|
|
|
3524 else /* c == BY_INDEX */
|
|
|
3525 {
|
|
|
3526 /* <nodeidx> */
|
|
|
3527 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
|
3528 if (n < 0 || n >= maxidx)
|
|
|
3529 return -2;
|
|
|
3530 idxs[idx] = n + SHARED_MASK;
|
|
|
3531 c = getc(fd); /* <xbyte> */
|
|
|
3532 }
|
|
|
3533 }
|
|
|
3534 byts[idx++] = c;
|
|
236
|
3535 }
|
|
|
3536
|
|
300
|
3537 /* Recursively read the children for non-shared siblings.
|
|
|
3538 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
|
3539 * remove SHARED_MASK) */
|
|
|
3540 for (i = 1; i <= len; ++i)
|
|
|
3541 if (byts[startidx + i] != 0)
|
|
|
3542 {
|
|
|
3543 if (idxs[startidx + i] & SHARED_MASK)
|
|
|
3544 idxs[startidx + i] &= ~SHARED_MASK;
|
|
|
3545 else
|
|
|
3546 {
|
|
|
3547 idxs[startidx + i] = idx;
|
|
339
|
3548 idx = read_tree(fd, byts, idxs, maxidx, idx,
|
|
|
3549 prefixtree, maxprefcondnr);
|
|
300
|
3550 if (idx < 0)
|
|
|
3551 break;
|
|
|
3552 }
|
|
|
3553 }
|
|
236
|
3554
|
|
300
|
3555 return idx;
|
|
236
|
3556 }
|
|
|
3557
|
|
|
3558 /*
|
|
|
3559 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
351
|
3560 * Returns NULL if it's OK, an error message otherwise.
|
|
236
|
3561 */
|
|
|
3562 char_u *
|
|
|
3563 did_set_spelllang(buf)
|
|
|
3564 buf_T *buf;
|
|
|
3565 {
|
|
|
3566 garray_T ga;
|
|
351
|
3567 char_u *splp;
|
|
236
|
3568 char_u *region;
|
|
416
|
3569 char_u region_cp[3];
|
|
355
|
3570 int filename;
|
|
236
|
3571 int region_mask;
|
|
503
|
3572 slang_T *slang;
|
|
236
|
3573 int c;
|
|
351
|
3574 char_u lang[MAXWLEN + 1];
|
|
323
|
3575 char_u spf_name[MAXPATHL];
|
|
351
|
3576 int len;
|
|
|
3577 char_u *p;
|
|
381
|
3578 int round;
|
|
385
|
3579 char_u *spf;
|
|
389
|
3580 char_u *use_region = NULL;
|
|
|
3581 int dont_use_region = FALSE;
|
|
501
|
3582 int nobreak = FALSE;
|
|
503
|
3583 int i, j;
|
|
|
3584 langp_T *lp, *lp2;
|
|
236
|
3585
|
|
|
3586 ga_init2(&ga, sizeof(langp_T), 2);
|
|
376
|
3587 clear_midword(buf);
|
|
236
|
3588
|
|
351
|
3589 /* loop over comma separated language names. */
|
|
|
3590 for (splp = buf->b_p_spl; *splp != NUL; )
|
|
|
3591 {
|
|
|
3592 /* Get one language name. */
|
|
|
3593 copy_option_part(&splp, lang, MAXWLEN, ",");
|
|
|
3594
|
|
240
|
3595 region = NULL;
|
|
351
|
3596 len = STRLEN(lang);
|
|
355
|
3597
|
|
|
3598 /* If the name ends in ".spl" use it as the name of the spell file.
|
|
|
3599 * If there is a region name let "region" point to it and remove it
|
|
|
3600 * from the name. */
|
|
|
3601 if (len > 4 && fnamecmp(lang + len - 4, ".spl") == 0)
|
|
236
|
3602 {
|
|
355
|
3603 filename = TRUE;
|
|
|
3604
|
|
416
|
3605 /* Locate a region and remove it from the file name. */
|
|
|
3606 p = vim_strchr(gettail(lang), '_');
|
|
|
3607 if (p != NULL && ASCII_ISALPHA(p[1]) && ASCII_ISALPHA(p[2])
|
|
|
3608 && !ASCII_ISALPHA(p[3]))
|
|
|
3609 {
|
|
|
3610 vim_strncpy(region_cp, p + 1, 2);
|
|
|
3611 mch_memmove(p, p + 3, len - (p - lang) - 2);
|
|
|
3612 len -= 3;
|
|
|
3613 region = region_cp;
|
|
|
3614 }
|
|
|
3615 else
|
|
|
3616 dont_use_region = TRUE;
|
|
|
3617
|
|
355
|
3618 /* Check if we loaded this language before. */
|
|
503
|
3619 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
3620 if (fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME)
|
|
355
|
3621 break;
|
|
236
|
3622 }
|
|
355
|
3623 else
|
|
|
3624 {
|
|
|
3625 filename = FALSE;
|
|
|
3626 if (len > 3 && lang[len - 3] == '_')
|
|
|
3627 {
|
|
|
3628 region = lang + len - 2;
|
|
|
3629 len -= 3;
|
|
|
3630 lang[len] = NUL;
|
|
389
|
3631 }
|
|
|
3632 else
|
|
|
3633 dont_use_region = TRUE;
|
|
355
|
3634
|
|
|
3635 /* Check if we loaded this language before. */
|
|
503
|
3636 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
3637 if (STRICMP(lang, slang->sl_name) == 0)
|
|
355
|
3638 break;
|
|
|
3639 }
|
|
236
|
3640
|
|
416
|
3641 if (region != NULL)
|
|
|
3642 {
|
|
|
3643 /* If the region differs from what was used before then don't
|
|
|
3644 * use it for 'spellfile'. */
|
|
|
3645 if (use_region != NULL && STRCMP(region, use_region) != 0)
|
|
|
3646 dont_use_region = TRUE;
|
|
|
3647 use_region = region;
|
|
|
3648 }
|
|
|
3649
|
|
351
|
3650 /* If not found try loading the language now. */
|
|
503
|
3651 if (slang == NULL)
|
|
355
|
3652 {
|
|
|
3653 if (filename)
|
|
|
3654 (void)spell_load_file(lang, lang, NULL, FALSE);
|
|
|
3655 else
|
|
|
3656 spell_load_lang(lang);
|
|
|
3657 }
|
|
236
|
3658
|
|
307
|
3659 /*
|
|
351
|
3660 * Loop over the languages, there can be several files for "lang".
|
|
307
|
3661 */
|
|
503
|
3662 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
3663 if (filename ? fullpathcmp(lang, slang->sl_fname, FALSE) == FPC_SAME
|
|
|
3664 : STRICMP(lang, slang->sl_name) == 0)
|
|
236
|
3665 {
|
|
316
|
3666 region_mask = REGION_ALL;
|
|
355
|
3667 if (!filename && region != NULL)
|
|
236
|
3668 {
|
|
307
|
3669 /* find region in sl_regions */
|
|
503
|
3670 c = find_region(slang->sl_regions, region);
|
|
307
|
3671 if (c == REGION_ALL)
|
|
|
3672 {
|
|
503
|
3673 if (slang->sl_add)
|
|
389
|
3674 {
|
|
503
|
3675 if (*slang->sl_regions != NUL)
|
|
389
|
3676 /* This addition file is for other regions. */
|
|
|
3677 region_mask = 0;
|
|
|
3678 }
|
|
|
3679 else
|
|
|
3680 /* This is probably an error. Give a warning and
|
|
|
3681 * accept the words anyway. */
|
|
351
|
3682 smsg((char_u *)
|
|
|
3683 _("Warning: region %s not supported"),
|
|
|
3684 region);
|
|
307
|
3685 }
|
|
|
3686 else
|
|
|
3687 region_mask = 1 << c;
|
|
236
|
3688 }
|
|
307
|
3689
|
|
389
|
3690 if (region_mask != 0)
|
|
307
|
3691 {
|
|
389
|
3692 if (ga_grow(&ga, 1) == FAIL)
|
|
|
3693 {
|
|
|
3694 ga_clear(&ga);
|
|
|
3695 return e_outofmem;
|
|
|
3696 }
|
|
503
|
3697 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang;
|
|
389
|
3698 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3699 ++ga.ga_len;
|
|
503
|
3700 use_midword(slang, buf);
|
|
|
3701 if (slang->sl_nobreak)
|
|
501
|
3702 nobreak = TRUE;
|
|
307
|
3703 }
|
|
385
|
3704 }
|
|
|
3705 }
|
|
|
3706
|
|
|
3707 /* round 0: load int_wordlist, if possible.
|
|
|
3708 * round 1: load first name in 'spellfile'.
|
|
|
3709 * round 2: load second name in 'spellfile.
|
|
|
3710 * etc. */
|
|
|
3711 spf = curbuf->b_p_spf;
|
|
|
3712 for (round = 0; round == 0 || *spf != NUL; ++round)
|
|
|
3713 {
|
|
|
3714 if (round == 0)
|
|
|
3715 {
|
|
|
3716 /* Internal wordlist, if there is one. */
|
|
|
3717 if (int_wordlist == NULL)
|
|
381
|
3718 continue;
|
|
385
|
3719 int_wordlist_spl(spf_name);
|
|
381
|
3720 }
|
|
|
3721 else
|
|
|
3722 {
|
|
385
|
3723 /* One entry in 'spellfile'. */
|
|
|
3724 copy_option_part(&spf, spf_name, MAXPATHL - 5, ",");
|
|
|
3725 STRCAT(spf_name, ".spl");
|
|
|
3726
|
|
|
3727 /* If it was already found above then skip it. */
|
|
|
3728 for (c = 0; c < ga.ga_len; ++c)
|
|
500
|
3729 {
|
|
|
3730 p = LANGP_ENTRY(ga, c)->lp_slang->sl_fname;
|
|
|
3731 if (p != NULL && fullpathcmp(spf_name, p, FALSE) == FPC_SAME)
|
|
385
|
3732 break;
|
|
500
|
3733 }
|
|
385
|
3734 if (c < ga.ga_len)
|
|
381
|
3735 continue;
|
|
|
3736 }
|
|
|
3737
|
|
351
|
3738 /* Check if it was loaded already. */
|
|
503
|
3739 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
3740 if (fullpathcmp(spf_name, slang->sl_fname, FALSE) == FPC_SAME)
|
|
323
|
3741 break;
|
|
503
|
3742 if (slang == NULL)
|
|
323
|
3743 {
|
|
351
|
3744 /* Not loaded, try loading it now. The language name includes the
|
|
385
|
3745 * region name, the region is ignored otherwise. for int_wordlist
|
|
|
3746 * use an arbitrary name. */
|
|
|
3747 if (round == 0)
|
|
|
3748 STRCPY(lang, "internal wordlist");
|
|
|
3749 else
|
|
|
3750 {
|
|
|
3751 vim_strncpy(lang, gettail(spf_name), MAXWLEN);
|
|
381
|
3752 p = vim_strchr(lang, '.');
|
|
|
3753 if (p != NULL)
|
|
|
3754 *p = NUL; /* truncate at ".encoding.add" */
|
|
|
3755 }
|
|
503
|
3756 slang = spell_load_file(spf_name, lang, NULL, TRUE);
|
|
501
|
3757
|
|
|
3758 /* If one of the languages has NOBREAK we assume the addition
|
|
|
3759 * files also have this. */
|
|
503
|
3760 if (slang != NULL && nobreak)
|
|
|
3761 slang->sl_nobreak = TRUE;
|
|
|
3762 }
|
|
|
3763 if (slang != NULL && ga_grow(&ga, 1) == OK)
|
|
323
|
3764 {
|
|
389
|
3765 region_mask = REGION_ALL;
|
|
|
3766 if (use_region != NULL && !dont_use_region)
|
|
|
3767 {
|
|
|
3768 /* find region in sl_regions */
|
|
503
|
3769 c = find_region(slang->sl_regions, use_region);
|
|
389
|
3770 if (c != REGION_ALL)
|
|
|
3771 region_mask = 1 << c;
|
|
503
|
3772 else if (*slang->sl_regions != NUL)
|
|
389
|
3773 /* This spell file is for other regions. */
|
|
|
3774 region_mask = 0;
|
|
|
3775 }
|
|
|
3776
|
|
|
3777 if (region_mask != 0)
|
|
|
3778 {
|
|
503
|
3779 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = slang;
|
|
|
3780 LANGP_ENTRY(ga, ga.ga_len)->lp_sallang = NULL;
|
|
|
3781 LANGP_ENTRY(ga, ga.ga_len)->lp_replang = NULL;
|
|
389
|
3782 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
|
3783 ++ga.ga_len;
|
|
503
|
3784 use_midword(slang, buf);
|
|
389
|
3785 }
|
|
323
|
3786 }
|
|
|
3787 }
|
|
|
3788
|
|
236
|
3789 /* Everything is fine, store the new b_langp value. */
|
|
|
3790 ga_clear(&buf->b_langp);
|
|
|
3791 buf->b_langp = ga;
|
|
|
3792
|
|
503
|
3793 /* For each language figure out what language to use for sound folding and
|
|
|
3794 * REP items. If the language doesn't support it itself use another one
|
|
|
3795 * with the same name. E.g. for "en-math" use "en". */
|
|
|
3796 for (i = 0; i < ga.ga_len; ++i)
|
|
|
3797 {
|
|
|
3798 lp = LANGP_ENTRY(ga, i);
|
|
|
3799
|
|
|
3800 /* sound folding */
|
|
|
3801 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
3802 /* language does sound folding itself */
|
|
|
3803 lp->lp_sallang = lp->lp_slang;
|
|
|
3804 else
|
|
|
3805 /* find first similar language that does sound folding */
|
|
|
3806 for (j = 0; j < ga.ga_len; ++j)
|
|
|
3807 {
|
|
|
3808 lp2 = LANGP_ENTRY(ga, j);
|
|
|
3809 if (lp2->lp_slang->sl_sal.ga_len > 0
|
|
|
3810 && STRNCMP(lp->lp_slang->sl_name,
|
|
|
3811 lp2->lp_slang->sl_name, 2) == 0)
|
|
|
3812 {
|
|
|
3813 lp->lp_sallang = lp2->lp_slang;
|
|
|
3814 break;
|
|
|
3815 }
|
|
|
3816 }
|
|
|
3817
|
|
|
3818 /* REP items */
|
|
|
3819 if (lp->lp_slang->sl_rep.ga_len > 0)
|
|
|
3820 /* language has REP items itself */
|
|
|
3821 lp->lp_replang = lp->lp_slang;
|
|
|
3822 else
|
|
|
3823 /* find first similar language that does sound folding */
|
|
|
3824 for (j = 0; j < ga.ga_len; ++j)
|
|
|
3825 {
|
|
|
3826 lp2 = LANGP_ENTRY(ga, j);
|
|
|
3827 if (lp2->lp_slang->sl_rep.ga_len > 0
|
|
|
3828 && STRNCMP(lp->lp_slang->sl_name,
|
|
|
3829 lp2->lp_slang->sl_name, 2) == 0)
|
|
|
3830 {
|
|
|
3831 lp->lp_replang = lp2->lp_slang;
|
|
|
3832 break;
|
|
|
3833 }
|
|
|
3834 }
|
|
|
3835 }
|
|
|
3836
|
|
236
|
3837 return NULL;
|
|
|
3838 }
|
|
|
3839
|
|
|
3840 /*
|
|
376
|
3841 * Clear the midword characters for buffer "buf".
|
|
|
3842 */
|
|
|
3843 static void
|
|
|
3844 clear_midword(buf)
|
|
|
3845 buf_T *buf;
|
|
|
3846 {
|
|
|
3847 vim_memset(buf->b_spell_ismw, 0, 256);
|
|
|
3848 #ifdef FEAT_MBYTE
|
|
|
3849 vim_free(buf->b_spell_ismw_mb);
|
|
|
3850 buf->b_spell_ismw_mb = NULL;
|
|
|
3851 #endif
|
|
|
3852 }
|
|
|
3853
|
|
|
3854 /*
|
|
|
3855 * Use the "sl_midword" field of language "lp" for buffer "buf".
|
|
|
3856 * They add up to any currently used midword characters.
|
|
|
3857 */
|
|
|
3858 static void
|
|
|
3859 use_midword(lp, buf)
|
|
|
3860 slang_T *lp;
|
|
|
3861 buf_T *buf;
|
|
|
3862 {
|
|
|
3863 char_u *p;
|
|
|
3864
|
|
389
|
3865 if (lp->sl_midword == NULL) /* there aren't any */
|
|
|
3866 return;
|
|
|
3867
|
|
376
|
3868 for (p = lp->sl_midword; *p != NUL; )
|
|
|
3869 #ifdef FEAT_MBYTE
|
|
|
3870 if (has_mbyte)
|
|
|
3871 {
|
|
|
3872 int c, l, n;
|
|
|
3873 char_u *bp;
|
|
|
3874
|
|
|
3875 c = mb_ptr2char(p);
|
|
474
|
3876 l = (*mb_ptr2len)(p);
|
|
|
3877 if (c < 256 && l <= 2)
|
|
376
|
3878 buf->b_spell_ismw[c] = TRUE;
|
|
|
3879 else if (buf->b_spell_ismw_mb == NULL)
|
|
|
3880 /* First multi-byte char in "b_spell_ismw_mb". */
|
|
|
3881 buf->b_spell_ismw_mb = vim_strnsave(p, l);
|
|
|
3882 else
|
|
|
3883 {
|
|
|
3884 /* Append multi-byte chars to "b_spell_ismw_mb". */
|
|
|
3885 n = STRLEN(buf->b_spell_ismw_mb);
|
|
|
3886 bp = vim_strnsave(buf->b_spell_ismw_mb, n + l);
|
|
|
3887 if (bp != NULL)
|
|
|
3888 {
|
|
|
3889 vim_free(buf->b_spell_ismw_mb);
|
|
|
3890 buf->b_spell_ismw_mb = bp;
|
|
|
3891 vim_strncpy(bp + n, p, l);
|
|
|
3892 }
|
|
|
3893 }
|
|
|
3894 p += l;
|
|
|
3895 }
|
|
|
3896 else
|
|
|
3897 #endif
|
|
|
3898 buf->b_spell_ismw[*p++] = TRUE;
|
|
|
3899 }
|
|
|
3900
|
|
|
3901 /*
|
|
236
|
3902 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
|
3903 * Each region is simply stored as the two characters of it's name.
|
|
381
|
3904 * Returns the index if found (first is 0), REGION_ALL if not found.
|
|
236
|
3905 */
|
|
|
3906 static int
|
|
|
3907 find_region(rp, region)
|
|
|
3908 char_u *rp;
|
|
|
3909 char_u *region;
|
|
|
3910 {
|
|
|
3911 int i;
|
|
|
3912
|
|
|
3913 for (i = 0; ; i += 2)
|
|
|
3914 {
|
|
|
3915 if (rp[i] == NUL)
|
|
|
3916 return REGION_ALL;
|
|
|
3917 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
|
3918 break;
|
|
|
3919 }
|
|
|
3920 return i / 2;
|
|
|
3921 }
|
|
|
3922
|
|
|
3923 /*
|
|
323
|
3924 * Return case type of word:
|
|
236
|
3925 * w word 0
|
|
300
|
3926 * Word WF_ONECAP
|
|
|
3927 * W WORD WF_ALLCAP
|
|
|
3928 * WoRd wOrd WF_KEEPCAP
|
|
236
|
3929 */
|
|
|
3930 static int
|
|
|
3931 captype(word, end)
|
|
|
3932 char_u *word;
|
|
323
|
3933 char_u *end; /* When NULL use up to NUL byte. */
|
|
236
|
3934 {
|
|
|
3935 char_u *p;
|
|
|
3936 int c;
|
|
|
3937 int firstcap;
|
|
|
3938 int allcap;
|
|
|
3939 int past_second = FALSE; /* past second word char */
|
|
|
3940
|
|
|
3941 /* find first letter */
|
|
376
|
3942 for (p = word; !spell_iswordp_nmw(p); mb_ptr_adv(p))
|
|
323
|
3943 if (end == NULL ? *p == NUL : p >= end)
|
|
236
|
3944 return 0; /* only non-word characters, illegal word */
|
|
|
3945 #ifdef FEAT_MBYTE
|
|
310
|
3946 if (has_mbyte)
|
|
|
3947 c = mb_ptr2char_adv(&p);
|
|
|
3948 else
|
|
236
|
3949 #endif
|
|
310
|
3950 c = *p++;
|
|
324
|
3951 firstcap = allcap = SPELL_ISUPPER(c);
|
|
236
|
3952
|
|
|
3953 /*
|
|
|
3954 * Need to check all letters to find a word with mixed upper/lower.
|
|
|
3955 * But a word with an upper char only at start is a ONECAP.
|
|
|
3956 */
|
|
323
|
3957 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
|
376
|
3958 if (spell_iswordp_nmw(p))
|
|
236
|
3959 {
|
|
455
|
3960 c = PTR2CHAR(p);
|
|
324
|
3961 if (!SPELL_ISUPPER(c))
|
|
236
|
3962 {
|
|
|
3963 /* UUl -> KEEPCAP */
|
|
|
3964 if (past_second && allcap)
|
|
300
|
3965 return WF_KEEPCAP;
|
|
236
|
3966 allcap = FALSE;
|
|
|
3967 }
|
|
|
3968 else if (!allcap)
|
|
|
3969 /* UlU -> KEEPCAP */
|
|
300
|
3970 return WF_KEEPCAP;
|
|
236
|
3971 past_second = TRUE;
|
|
|
3972 }
|
|
|
3973
|
|
|
3974 if (allcap)
|
|
300
|
3975 return WF_ALLCAP;
|
|
236
|
3976 if (firstcap)
|
|
300
|
3977 return WF_ONECAP;
|
|
236
|
3978 return 0;
|
|
|
3979 }
|
|
|
3980
|
|
474
|
3981 /*
|
|
|
3982 * Like captype() but for a KEEPCAP word add ONECAP if the word starts with a
|
|
|
3983 * capital. So that make_case_word() can turn WOrd into Word.
|
|
|
3984 * Add ALLCAP for "WOrD".
|
|
|
3985 */
|
|
|
3986 static int
|
|
|
3987 badword_captype(word, end)
|
|
|
3988 char_u *word;
|
|
|
3989 char_u *end;
|
|
|
3990 {
|
|
|
3991 int flags = captype(word, end);
|
|
475
|
3992 int c;
|
|
474
|
3993 int l, u;
|
|
|
3994 int first;
|
|
|
3995 char_u *p;
|
|
|
3996
|
|
|
3997 if (flags & WF_KEEPCAP)
|
|
|
3998 {
|
|
|
3999 /* Count the number of UPPER and lower case letters. */
|
|
|
4000 l = u = 0;
|
|
|
4001 first = FALSE;
|
|
|
4002 for (p = word; p < end; mb_ptr_adv(p))
|
|
|
4003 {
|
|
475
|
4004 c = PTR2CHAR(p);
|
|
|
4005 if (SPELL_ISUPPER(c))
|
|
474
|
4006 {
|
|
|
4007 ++u;
|
|
|
4008 if (p == word)
|
|
|
4009 first = TRUE;
|
|
|
4010 }
|
|
|
4011 else
|
|
|
4012 ++l;
|
|
|
4013 }
|
|
|
4014
|
|
|
4015 /* If there are more UPPER than lower case letters suggest an
|
|
|
4016 * ALLCAP word. Otherwise, if the first letter is UPPER then
|
|
|
4017 * suggest ONECAP. Exception: "ALl" most likely should be "All",
|
|
|
4018 * require three upper case letters. */
|
|
|
4019 if (u > l && u > 2)
|
|
|
4020 flags |= WF_ALLCAP;
|
|
|
4021 else if (first)
|
|
|
4022 flags |= WF_ONECAP;
|
|
|
4023 }
|
|
|
4024 return flags;
|
|
|
4025 }
|
|
|
4026
|
|
355
|
4027 # if defined(FEAT_MBYTE) || defined(EXITFREE) || defined(PROTO)
|
|
|
4028 /*
|
|
|
4029 * Free all languages.
|
|
|
4030 */
|
|
|
4031 void
|
|
|
4032 spell_free_all()
|
|
|
4033 {
|
|
503
|
4034 slang_T *slang;
|
|
355
|
4035 buf_T *buf;
|
|
385
|
4036 char_u fname[MAXPATHL];
|
|
355
|
4037
|
|
|
4038 /* Go through all buffers and handle 'spelllang'. */
|
|
|
4039 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
4040 ga_clear(&buf->b_langp);
|
|
|
4041
|
|
|
4042 while (first_lang != NULL)
|
|
|
4043 {
|
|
503
|
4044 slang = first_lang;
|
|
|
4045 first_lang = slang->sl_next;
|
|
|
4046 slang_free(slang);
|
|
355
|
4047 }
|
|
366
|
4048
|
|
385
|
4049 if (int_wordlist != NULL)
|
|
|
4050 {
|
|
|
4051 /* Delete the internal wordlist and its .spl file */
|
|
|
4052 mch_remove(int_wordlist);
|
|
|
4053 int_wordlist_spl(fname);
|
|
|
4054 mch_remove(fname);
|
|
|
4055 vim_free(int_wordlist);
|
|
|
4056 int_wordlist = NULL;
|
|
381
|
4057 }
|
|
|
4058
|
|
366
|
4059 init_spell_chartab();
|
|
355
|
4060 }
|
|
|
4061 # endif
|
|
|
4062
|
|
236
|
4063 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
|
4064 /*
|
|
|
4065 * Clear all spelling tables and reload them.
|
|
307
|
4066 * Used after 'encoding' is set and when ":mkspell" was used.
|
|
236
|
4067 */
|
|
|
4068 void
|
|
|
4069 spell_reload()
|
|
|
4070 {
|
|
|
4071 buf_T *buf;
|
|
316
|
4072 win_T *wp;
|
|
236
|
4073
|
|
358
|
4074 /* Initialize the table for spell_iswordp(). */
|
|
236
|
4075 init_spell_chartab();
|
|
|
4076
|
|
|
4077 /* Unload all allocated memory. */
|
|
355
|
4078 spell_free_all();
|
|
236
|
4079
|
|
|
4080 /* Go through all buffers and handle 'spelllang'. */
|
|
|
4081 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
|
4082 {
|
|
316
|
4083 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
|
4084 * window for this buffer in which 'spell' is set. */
|
|
236
|
4085 if (*buf->b_p_spl != NUL)
|
|
316
|
4086 {
|
|
|
4087 FOR_ALL_WINDOWS(wp)
|
|
|
4088 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
|
4089 {
|
|
|
4090 (void)did_set_spelllang(buf);
|
|
|
4091 # ifdef FEAT_WINDOWS
|
|
|
4092 break;
|
|
|
4093 # endif
|
|
|
4094 }
|
|
|
4095 }
|
|
236
|
4096 }
|
|
|
4097 }
|
|
|
4098 # endif
|
|
|
4099
|
|
310
|
4100 /*
|
|
|
4101 * Reload the spell file "fname" if it's loaded.
|
|
|
4102 */
|
|
|
4103 static void
|
|
323
|
4104 spell_reload_one(fname, added_word)
|
|
310
|
4105 char_u *fname;
|
|
323
|
4106 int added_word; /* invoked through "zg" */
|
|
310
|
4107 {
|
|
503
|
4108 slang_T *slang;
|
|
323
|
4109 int didit = FALSE;
|
|
310
|
4110
|
|
503
|
4111 for (slang = first_lang; slang != NULL; slang = slang->sl_next)
|
|
|
4112 {
|
|
|
4113 if (fullpathcmp(fname, slang->sl_fname, FALSE) == FPC_SAME)
|
|
|
4114 {
|
|
|
4115 slang_clear(slang);
|
|
|
4116 if (spell_load_file(fname, NULL, slang, FALSE) == NULL)
|
|
500
|
4117 /* reloading failed, clear the language */
|
|
503
|
4118 slang_clear(slang);
|
|
310
|
4119 redraw_all_later(NOT_VALID);
|
|
323
|
4120 didit = TRUE;
|
|
310
|
4121 }
|
|
500
|
4122 }
|
|
323
|
4123
|
|
|
4124 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
|
4125 * 'spelllang' to get it loaded. */
|
|
|
4126 if (added_word && !didit)
|
|
|
4127 did_set_spelllang(curbuf);
|
|
310
|
4128 }
|
|
|
4129
|
|
|
4130
|
|
236
|
4131 /*
|
|
|
4132 * Functions for ":mkspell".
|
|
|
4133 */
|
|
|
4134
|
|
300
|
4135 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
|
236
|
4136 and .dic file. */
|
|
|
4137 /*
|
|
|
4138 * Main structure to store the contents of a ".aff" file.
|
|
|
4139 */
|
|
|
4140 typedef struct afffile_S
|
|
|
4141 {
|
|
|
4142 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
|
498
|
4143 int af_flagtype; /* AFT_CHAR, AFT_LONG, AFT_NUM or AFT_CAPLONG */
|
|
481
|
4144 int af_slash; /* character used in word for slash */
|
|
621
|
4145 unsigned af_rare; /* RARE ID for rare word */
|
|
|
4146 unsigned af_keepcase; /* KEEPCASE ID for keep-case word */
|
|
495
|
4147 unsigned af_bad; /* BAD ID for banned word */
|
|
|
4148 unsigned af_needaffix; /* NEEDAFFIX ID */
|
|
500
|
4149 unsigned af_needcomp; /* NEEDCOMPOUND ID */
|
|
339
|
4150 int af_pfxpostpone; /* postpone prefixes without chop string */
|
|
236
|
4151 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
|
4152 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
495
|
4153 hashtab_T af_comp; /* hashtable for compound flags, compitem_T */
|
|
236
|
4154 } afffile_T;
|
|
|
4155
|
|
495
|
4156 #define AFT_CHAR 0 /* flags are one character */
|
|
498
|
4157 #define AFT_LONG 1 /* flags are two characters */
|
|
|
4158 #define AFT_CAPLONG 2 /* flags are one or two characters */
|
|
|
4159 #define AFT_NUM 3 /* flags are numbers, comma separated */
|
|
495
|
4160
|
|
236
|
4161 typedef struct affentry_S affentry_T;
|
|
|
4162 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
|
4163 struct affentry_S
|
|
|
4164 {
|
|
|
4165 affentry_T *ae_next; /* next affix with same name/number */
|
|
|
4166 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
|
4167 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
|
4168 char_u *ae_cond; /* condition (NULL for ".") */
|
|
|
4169 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
|
483
|
4170 char_u ae_rare; /* rare affix */
|
|
|
4171 char_u ae_nocomp; /* word with affix not compoundable */
|
|
300
|
4172 };
|
|
|
4173
|
|
495
|
4174 #ifdef FEAT_MBYTE
|
|
|
4175 # define AH_KEY_LEN 17 /* 2 x 8 bytes + NUL */
|
|
|
4176 #else
|
|
498
|
4177 # define AH_KEY_LEN 7 /* 6 digits + NUL */
|
|
495
|
4178 #endif
|
|
455
|
4179
|
|
300
|
4180 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
|
4181 typedef struct affheader_S
|
|
|
4182 {
|
|
495
|
4183 char_u ah_key[AH_KEY_LEN]; /* key for hashtab == name of affix */
|
|
|
4184 unsigned ah_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4185 int ah_newID; /* prefix ID after renumbering; 0 if not used */
|
|
300
|
4186 int ah_combine; /* suffix may combine with prefix */
|
|
498
|
4187 int ah_follows; /* another affix block should be following */
|
|
300
|
4188 affentry_T *ah_first; /* first affix entry */
|
|
|
4189 } affheader_T;
|
|
|
4190
|
|
|
4191 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
|
4192
|
|
495
|
4193 /* Flag used in compound items. */
|
|
|
4194 typedef struct compitem_S
|
|
|
4195 {
|
|
|
4196 char_u ci_key[AH_KEY_LEN]; /* key for hashtab == name of compound */
|
|
|
4197 unsigned ci_flag; /* affix name as number, uses "af_flagtype" */
|
|
|
4198 int ci_newID; /* affix ID after renumbering. */
|
|
|
4199 } compitem_T;
|
|
|
4200
|
|
|
4201 #define HI2CI(hi) ((compitem_T *)(hi)->hi_key)
|
|
|
4202
|
|
300
|
4203 /*
|
|
|
4204 * Structure that is used to store the items in the word tree. This avoids
|
|
474
|
4205 * the need to keep track of each allocated thing, everything is freed all at
|
|
|
4206 * once after ":mkspell" is done.
|
|
300
|
4207 */
|
|
|
4208 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
|
4209 typedef struct sblock_S sblock_T;
|
|
|
4210 struct sblock_S
|
|
|
4211 {
|
|
|
4212 sblock_T *sb_next; /* next block in list */
|
|
|
4213 int sb_used; /* nr of bytes already in use */
|
|
|
4214 char_u sb_data[1]; /* data, actually longer */
|
|
236
|
4215 };
|
|
|
4216
|
|
|
4217 /*
|
|
300
|
4218 * A node in the tree.
|
|
236
|
4219 */
|
|
300
|
4220 typedef struct wordnode_S wordnode_T;
|
|
|
4221 struct wordnode_S
|
|
236
|
4222 {
|
|
346
|
4223 union /* shared to save space */
|
|
|
4224 {
|
|
469
|
4225 char_u hashkey[6]; /* the hash key, only used while compressing */
|
|
346
|
4226 int index; /* index in written nodes (valid after first
|
|
|
4227 round) */
|
|
|
4228 } wn_u1;
|
|
|
4229 union /* shared to save space */
|
|
|
4230 {
|
|
|
4231 wordnode_T *next; /* next node with same hash key */
|
|
|
4232 wordnode_T *wnode; /* parent node that will write this node */
|
|
|
4233 } wn_u2;
|
|
300
|
4234 wordnode_T *wn_child; /* child (next byte in word) */
|
|
|
4235 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
|
4236 always sorted) */
|
|
474
|
4237 int wn_refs; /* Nr. of references to this node. Only
|
|
|
4238 relevant for first node in a list of
|
|
|
4239 siblings, in following siblings it is
|
|
|
4240 always one. */
|
|
300
|
4241 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
481
|
4242 char_u wn_affixID; /* when "wn_byte" is NUL: supported/required
|
|
469
|
4243 prefix ID or 0 */
|
|
|
4244 short_u wn_flags; /* when "wn_byte" is NUL: WF_ flags */
|
|
|
4245 short wn_region; /* when "wn_byte" is NUL: region mask; for
|
|
339
|
4246 PREFIXTREE it's the prefcondnr */
|
|
469
|
4247 #ifdef SPELL_PRINTTREE
|
|
|
4248 int wn_nr; /* sequence nr for printing */
|
|
|
4249 #endif
|
|
236
|
4250 };
|
|
|
4251
|
|
390
|
4252 #define WN_MASK 0xffff /* mask relevant bits of "wn_flags" */
|
|
|
4253
|
|
300
|
4254 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
|
236
|
4255
|
|
300
|
4256 /*
|
|
|
4257 * Info used while reading the spell files.
|
|
|
4258 */
|
|
|
4259 typedef struct spellinfo_S
|
|
249
|
4260 {
|
|
300
|
4261 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
334
|
4262 long si_foldwcount; /* nr of words in si_foldroot */
|
|
474
|
4263
|
|
300
|
4264 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
334
|
4265 long si_keepwcount; /* nr of words in si_keeproot */
|
|
474
|
4266
|
|
339
|
4267 wordnode_T *si_prefroot; /* tree with postponed prefixes */
|
|
474
|
4268
|
|
300
|
4269 sblock_T *si_blocks; /* memory blocks used */
|
|
482
|
4270 long si_blocks_cnt; /* memory blocks allocated */
|
|
|
4271 long si_compress_cnt; /* words to add before lowering
|
|
|
4272 compression limit */
|
|
474
|
4273 wordnode_T *si_first_free; /* List of nodes that have been freed during
|
|
|
4274 compression, linked by "wn_child" field. */
|
|
482
|
4275 long si_free_count; /* number of nodes in si_first_free */
|
|
474
|
4276 #ifdef SPELL_PRINTTREE
|
|
|
4277 int si_wordnode_nr; /* sequence nr for nodes */
|
|
|
4278 #endif
|
|
|
4279
|
|
|
4280
|
|
300
|
4281 int si_ascii; /* handling only ASCII words */
|
|
310
|
4282 int si_add; /* addition file */
|
|
351
|
4283 int si_clear_chartab; /* when TRUE clear char tables */
|
|
300
|
4284 int si_region; /* region mask */
|
|
|
4285 vimconv_T si_conv; /* for conversion to 'encoding' */
|
|
302
|
4286 int si_memtot; /* runtime memory used */
|
|
310
|
4287 int si_verbose; /* verbose messages */
|
|
474
|
4288 int si_msg_count; /* number of words added since last message */
|
|
316
|
4289 int si_region_count; /* number of regions supported (1 when there
|
|
|
4290 are no regions) */
|
|
483
|
4291 char_u si_region_name[16]; /* region names; used only if
|
|
|
4292 * si_region_count > 1) */
|
|
323
|
4293
|
|
|
4294 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
|
4295 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
375
|
4296 char_u *si_sofofr; /* SOFOFROM text */
|
|
|
4297 char_u *si_sofoto; /* SOFOTO text */
|
|
323
|
4298 int si_followup; /* soundsalike: ? */
|
|
|
4299 int si_collapse; /* soundsalike: ? */
|
|
|
4300 int si_rem_accents; /* soundsalike: remove accents */
|
|
|
4301 garray_T si_map; /* MAP info concatenated */
|
|
495
|
4302 char_u *si_midword; /* MIDWORD chars or NULL */
|
|
483
|
4303 int si_compmax; /* max nr of words for compounding */
|
|
481
|
4304 int si_compminlen; /* minimal length for compounding */
|
|
483
|
4305 int si_compsylmax; /* max nr of syllables for compounding */
|
|
481
|
4306 char_u *si_compflags; /* flags used for compounding */
|
|
492
|
4307 char_u si_nobreak; /* NOBREAK */
|
|
483
|
4308 char_u *si_syllable; /* syllable string */
|
|
339
|
4309 garray_T si_prefcond; /* table with conditions for postponed
|
|
|
4310 * prefixes, each stored as a string */
|
|
495
|
4311 int si_newprefID; /* current value for ah_newID */
|
|
500
|
4312 int si_newcompID; /* current value for compound ID */
|
|
300
|
4313 } spellinfo_T;
|
|
249
|
4314
|
|
474
|
4315 static afffile_T *spell_read_aff __ARGS((spellinfo_T *spin, char_u *fname));
|
|
495
|
4316 static unsigned affitem2flag __ARGS((int flagtype, char_u *item, char_u *fname, int lnum));
|
|
|
4317 static unsigned get_affitem __ARGS((int flagtype, char_u **pp));
|
|
|
4318 static void process_compflags __ARGS((spellinfo_T *spin, afffile_T *aff, char_u *compflags));
|
|
500
|
4319 static void check_renumber __ARGS((spellinfo_T *spin));
|
|
495
|
4320 static int flag_in_afflist __ARGS((int flagtype, char_u *afflist, unsigned flag));
|
|
|
4321 static void aff_check_number __ARGS((int spinval, int affval, char *name));
|
|
|
4322 static void aff_check_string __ARGS((char_u *spinval, char_u *affval, char *name));
|
|
339
|
4323 static int str_equal __ARGS((char_u *s1, char_u *s2));
|
|
323
|
4324 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
|
4325 static int sal_to_bool __ARGS((char_u *s));
|
|
240
|
4326 static int has_non_ascii __ARGS((char_u *s));
|
|
300
|
4327 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
474
|
4328 static int spell_read_dic __ARGS((spellinfo_T *spin, char_u *fname, afffile_T *affile));
|
|
483
|
4329 static int get_pfxlist __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
495
|
4330 static void get_compflags __ARGS((afffile_T *affile, char_u *afflist, char_u *store_afflist));
|
|
483
|
4331 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
|
4332 static int spell_read_wordfile __ARGS((spellinfo_T *spin, char_u *fname));
|
|
|
4333 static void *getroom __ARGS((spellinfo_T *spin, size_t len, int align));
|
|
|
4334 static char_u *getroom_save __ARGS((spellinfo_T *spin, char_u *s));
|
|
300
|
4335 static void free_blocks __ARGS((sblock_T *bl));
|
|
474
|
4336 static wordnode_T *wordtree_alloc __ARGS((spellinfo_T *spin));
|
|
483
|
4337 static int store_word __ARGS((spellinfo_T *spin, char_u *word, int flags, int region, char_u *pfxlist, int need_affix));
|
|
481
|
4338 static int tree_add_word __ARGS((spellinfo_T *spin, char_u *word, wordnode_T *tree, int flags, int region, int affixID));
|
|
474
|
4339 static wordnode_T *get_wordnode __ARGS((spellinfo_T *spin));
|
|
|
4340 static void deref_wordnode __ARGS((spellinfo_T *spin, wordnode_T *node));
|
|
|
4341 static void free_wordnode __ARGS((spellinfo_T *spin, wordnode_T *n));
|
|
|
4342 static void wordtree_compress __ARGS((spellinfo_T *spin, wordnode_T *root));
|
|
|
4343 static int node_compress __ARGS((spellinfo_T *spin, wordnode_T *node, hashtab_T *ht, int *tot));
|
|
300
|
4344 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
|
500
|
4345 static int write_vim_spell __ARGS((spellinfo_T *spin, char_u *fname));
|
|
346
|
4346 static void clear_node __ARGS((wordnode_T *node));
|
|
|
4347 static int put_node __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask, int prefixtree));
|
|
323
|
4348 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
|
310
|
4349 static void init_spellfile __ARGS((void));
|
|
236
|
4350
|
|
455
|
4351 /* In the postponed prefixes tree wn_flags is used to store the WFP_ flags,
|
|
|
4352 * but it must be negative to indicate the prefix tree to tree_add_word().
|
|
|
4353 * Use a negative number with the lower 8 bits zero. */
|
|
|
4354 #define PFX_FLAGS -256
|
|
390
|
4355
|
|
483
|
4356 /*
|
|
|
4357 * Tunable parameters for when the tree is compressed. See 'mkspellmem'.
|
|
|
4358 */
|
|
|
4359 static long compress_start = 30000; /* memory / SBLOCKSIZE */
|
|
|
4360 static long compress_inc = 100; /* memory / SBLOCKSIZE */
|
|
|
4361 static long compress_added = 500000; /* word count */
|
|
|
4362
|
|
469
|
4363 #ifdef SPELL_PRINTTREE
|
|
|
4364 /*
|
|
|
4365 * For debugging the tree code: print the current tree in a (more or less)
|
|
|
4366 * readable format, so that we can see what happens when adding a word and/or
|
|
|
4367 * compressing the tree.
|
|
|
4368 * Based on code from Olaf Seibert.
|
|
|
4369 */
|
|
|
4370 #define PRINTLINESIZE 1000
|
|
|
4371 #define PRINTWIDTH 6
|
|
|
4372
|
|
|
4373 #define PRINTSOME(l, depth, fmt, a1, a2) vim_snprintf(l + depth * PRINTWIDTH, \
|
|
|
4374 PRINTLINESIZE - PRINTWIDTH * depth, fmt, a1, a2)
|
|
|
4375
|
|
|
4376 static char line1[PRINTLINESIZE];
|
|
|
4377 static char line2[PRINTLINESIZE];
|
|
|
4378 static char line3[PRINTLINESIZE];
|
|
|
4379
|
|
|
4380 static void
|
|
|
4381 spell_clear_flags(wordnode_T *node)
|
|
|
4382 {
|
|
|
4383 wordnode_T *np;
|
|
|
4384
|
|
|
4385 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
4386 {
|
|
|
4387 np->wn_u1.index = FALSE;
|
|
|
4388 spell_clear_flags(np->wn_child);
|
|
|
4389 }
|
|
|
4390 }
|
|
|
4391
|
|
|
4392 static void
|
|
|
4393 spell_print_node(wordnode_T *node, int depth)
|
|
|
4394 {
|
|
|
4395 if (node->wn_u1.index)
|
|
|
4396 {
|
|
|
4397 /* Done this node before, print the reference. */
|
|
|
4398 PRINTSOME(line1, depth, "(%d)", node->wn_nr, 0);
|
|
|
4399 PRINTSOME(line2, depth, " ", 0, 0);
|
|
|
4400 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4401 msg(line1);
|
|
|
4402 msg(line2);
|
|
|
4403 msg(line3);
|
|
|
4404 }
|
|
|
4405 else
|
|
|
4406 {
|
|
|
4407 node->wn_u1.index = TRUE;
|
|
|
4408
|
|
|
4409 if (node->wn_byte != NUL)
|
|
|
4410 {
|
|
|
4411 if (node->wn_child != NULL)
|
|
|
4412 PRINTSOME(line1, depth, " %c -> ", node->wn_byte, 0);
|
|
|
4413 else
|
|
|
4414 /* Cannot happen? */
|
|
|
4415 PRINTSOME(line1, depth, " %c ???", node->wn_byte, 0);
|
|
|
4416 }
|
|
|
4417 else
|
|
|
4418 PRINTSOME(line1, depth, " $ ", 0, 0);
|
|
|
4419
|
|
|
4420 PRINTSOME(line2, depth, "%d/%d ", node->wn_nr, node->wn_refs);
|
|
|
4421
|
|
|
4422 if (node->wn_sibling != NULL)
|
|
|
4423 PRINTSOME(line3, depth, " | ", 0, 0);
|
|
|
4424 else
|
|
|
4425 PRINTSOME(line3, depth, " ", 0, 0);
|
|
|
4426
|
|
|
4427 if (node->wn_byte == NUL)
|
|
|
4428 {
|
|
|
4429 msg(line1);
|
|
|
4430 msg(line2);
|
|
|
4431 msg(line3);
|
|
|
4432 }
|
|
|
4433
|
|
|
4434 /* do the children */
|
|
|
4435 if (node->wn_byte != NUL && node->wn_child != NULL)
|
|
|
4436 spell_print_node(node->wn_child, depth + 1);
|
|
|
4437
|
|
|
4438 /* do the siblings */
|
|
|
4439 if (node->wn_sibling != NULL)
|
|
|
4440 {
|
|
|
4441 /* get rid of all parent details except | */
|
|
|
4442 STRCPY(line1, line3);
|
|
|
4443 STRCPY(line2, line3);
|
|
|
4444 spell_print_node(node->wn_sibling, depth);
|
|
|
4445 }
|
|
|
4446 }
|
|
|
4447 }
|
|
|
4448
|
|
|
4449 static void
|
|
|
4450 spell_print_tree(wordnode_T *root)
|
|
|
4451 {
|
|
|
4452 if (root != NULL)
|
|
|
4453 {
|
|
|
4454 /* Clear the "wn_u1.index" fields, used to remember what has been
|
|
|
4455 * done. */
|
|
|
4456 spell_clear_flags(root);
|
|
|
4457
|
|
|
4458 /* Recursively print the tree. */
|
|
|
4459 spell_print_node(root, 0);
|
|
|
4460 }
|
|
|
4461 }
|
|
|
4462 #endif /* SPELL_PRINTTREE */
|
|
|
4463
|
|
236
|
4464 /*
|
|
323
|
4465 * Read the affix file "fname".
|
|
316
|
4466 * Returns an afffile_T, NULL for complete failure.
|
|
236
|
4467 */
|
|
|
4468 static afffile_T *
|
|
474
|
4469 spell_read_aff(spin, fname)
|
|
|
4470 spellinfo_T *spin;
|
|
236
|
4471 char_u *fname;
|
|
|
4472 {
|
|
|
4473 FILE *fd;
|
|
|
4474 afffile_T *aff;
|
|
|
4475 char_u rline[MAXLINELEN];
|
|
|
4476 char_u *line;
|
|
|
4477 char_u *pc = NULL;
|
|
334
|
4478 #define MAXITEMCNT 7
|
|
|
4479 char_u *(items[MAXITEMCNT]);
|
|
236
|
4480 int itemcnt;
|
|
|
4481 char_u *p;
|
|
|
4482 int lnum = 0;
|
|
|
4483 affheader_T *cur_aff = NULL;
|
|
495
|
4484 int did_postpone_prefix = FALSE;
|
|
236
|
4485 int aff_todo = 0;
|
|
|
4486 hashtab_T *tp;
|
|
255
|
4487 char_u *low = NULL;
|
|
|
4488 char_u *fol = NULL;
|
|
|
4489 char_u *upp = NULL;
|
|
323
|
4490 int do_rep;
|
|
|
4491 int do_sal;
|
|
|
4492 int do_map;
|
|
|
4493 int found_map = FALSE;
|
|
324
|
4494 hashitem_T *hi;
|
|
455
|
4495 int l;
|
|
495
|
4496 int compminlen = 0; /* COMPOUNDMIN value */
|
|
|
4497 int compsylmax = 0; /* COMPOUNDSYLMAX value */
|
|
|
4498 int compmax = 0; /* COMPOUNDMAX value */
|
|
|
4499 char_u *compflags = NULL; /* COMPOUNDFLAG and COMPOUNDFLAGS
|
|
|
4500 concatenated */
|
|
|
4501 char_u *midword = NULL; /* MIDWORD value */
|
|
|
4502 char_u *syllable = NULL; /* SYLLABLE value */
|
|
|
4503 char_u *sofofrom = NULL; /* SOFOFROM value */
|
|
|
4504 char_u *sofoto = NULL; /* SOFOTO value */
|
|
236
|
4505
|
|
300
|
4506 /*
|
|
|
4507 * Open the file.
|
|
|
4508 */
|
|
310
|
4509 fd = mch_fopen((char *)fname, "r");
|
|
236
|
4510 if (fd == NULL)
|
|
|
4511 {
|
|
|
4512 EMSG2(_(e_notopen), fname);
|
|
|
4513 return NULL;
|
|
|
4514 }
|
|
|
4515
|
|
310
|
4516 if (spin->si_verbose || p_verbose > 2)
|
|
|
4517 {
|
|
|
4518 if (!spin->si_verbose)
|
|
|
4519 verbose_enter();
|
|
366
|
4520 smsg((char_u *)_("Reading affix file %s ..."), fname);
|
|
310
|
4521 out_flush();
|
|
|
4522 if (!spin->si_verbose)
|
|
|
4523 verbose_leave();
|
|
|
4524 }
|
|
236
|
4525
|
|
323
|
4526 /* Only do REP lines when not done in another .aff file already. */
|
|
|
4527 do_rep = spin->si_rep.ga_len == 0;
|
|
|
4528
|
|
|
4529 /* Only do SAL lines when not done in another .aff file already. */
|
|
|
4530 do_sal = spin->si_sal.ga_len == 0;
|
|
|
4531
|
|
|
4532 /* Only do MAP lines when not done in another .aff file already. */
|
|
|
4533 do_map = spin->si_map.ga_len == 0;
|
|
|
4534
|
|
300
|
4535 /*
|
|
|
4536 * Allocate and init the afffile_T structure.
|
|
|
4537 */
|
|
474
|
4538 aff = (afffile_T *)getroom(spin, sizeof(afffile_T), TRUE);
|
|
236
|
4539 if (aff == NULL)
|
|
|
4540 return NULL;
|
|
|
4541 hash_init(&aff->af_pref);
|
|
|
4542 hash_init(&aff->af_suff);
|
|
495
|
4543 hash_init(&aff->af_comp);
|
|
236
|
4544
|
|
|
4545 /*
|
|
|
4546 * Read all the lines in the file one by one.
|
|
|
4547 */
|
|
255
|
4548 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
236
|
4549 {
|
|
255
|
4550 line_breakcheck();
|
|
236
|
4551 ++lnum;
|
|
|
4552
|
|
|
4553 /* Skip comment lines. */
|
|
|
4554 if (*rline == '#')
|
|
|
4555 continue;
|
|
|
4556
|
|
|
4557 /* Convert from "SET" to 'encoding' when needed. */
|
|
|
4558 vim_free(pc);
|
|
310
|
4559 #ifdef FEAT_MBYTE
|
|
300
|
4560 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
4561 {
|
|
300
|
4562 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
255
|
4563 if (pc == NULL)
|
|
|
4564 {
|
|
|
4565 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
4566 fname, lnum, rline);
|
|
|
4567 continue;
|
|
|
4568 }
|
|
236
|
4569 line = pc;
|
|
|
4570 }
|
|
|
4571 else
|
|
310
|
4572 #endif
|
|
236
|
4573 {
|
|
|
4574 pc = NULL;
|
|
|
4575 line = rline;
|
|
|
4576 }
|
|
|
4577
|
|
|
4578 /* Split the line up in white separated items. Put a NUL after each
|
|
|
4579 * item. */
|
|
|
4580 itemcnt = 0;
|
|
|
4581 for (p = line; ; )
|
|
|
4582 {
|
|
|
4583 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
|
4584 ++p;
|
|
|
4585 if (*p == NUL)
|
|
|
4586 break;
|
|
334
|
4587 if (itemcnt == MAXITEMCNT) /* too many items */
|
|
300
|
4588 break;
|
|
236
|
4589 items[itemcnt++] = p;
|
|
300
|
4590 while (*p > ' ') /* skip until white space or CR/NL */
|
|
236
|
4591 ++p;
|
|
|
4592 if (*p == NUL)
|
|
|
4593 break;
|
|
|
4594 *p++ = NUL;
|
|
|
4595 }
|
|
|
4596
|
|
|
4597 /* Handle non-empty lines. */
|
|
|
4598 if (itemcnt > 0)
|
|
|
4599 {
|
|
|
4600 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
|
4601 && aff->af_enc == NULL)
|
|
|
4602 {
|
|
310
|
4603 #ifdef FEAT_MBYTE
|
|
300
|
4604 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
|
4605 aff->af_enc = enc_canonize(items[1]);
|
|
|
4606 if (aff->af_enc != NULL && !spin->si_ascii
|
|
|
4607 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
|
4608 p_enc) == FAIL)
|
|
|
4609 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
|
4610 fname, aff->af_enc, p_enc);
|
|
375
|
4611 spin->si_conv.vc_fail = TRUE;
|
|
310
|
4612 #else
|
|
|
4613 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
4614 #endif
|
|
236
|
4615 }
|
|
495
|
4616 else if (STRCMP(items[0], "FLAG") == 0 && itemcnt == 2
|
|
|
4617 && aff->af_flagtype == AFT_CHAR)
|
|
|
4618 {
|
|
|
4619 if (STRCMP(items[1], "long") == 0)
|
|
498
|
4620 aff->af_flagtype = AFT_LONG;
|
|
495
|
4621 else if (STRCMP(items[1], "num") == 0)
|
|
498
|
4622 aff->af_flagtype = AFT_NUM;
|
|
|
4623 else if (STRCMP(items[1], "caplong") == 0)
|
|
|
4624 aff->af_flagtype = AFT_CAPLONG;
|
|
495
|
4625 else
|
|
|
4626 smsg((char_u *)_("Invalid value for FLAG in %s line %d: %s"),
|
|
|
4627 fname, lnum, items[1]);
|
|
621
|
4628 if (aff->af_rare != 0
|
|
|
4629 || aff->af_keepcase != 0
|
|
|
4630 || aff->af_bad != 0
|
|
500
|
4631 || aff->af_needaffix != 0
|
|
|
4632 || aff->af_needcomp != 0
|
|
|
4633 || compflags != NULL
|
|
495
|
4634 || aff->af_suff.ht_used > 0
|
|
|
4635 || aff->af_pref.ht_used > 0)
|
|
|
4636 smsg((char_u *)_("FLAG after using flags in %s line %d: %s"),
|
|
|
4637 fname, lnum, items[1]);
|
|
|
4638 }
|
|
|
4639 else if (STRCMP(items[0], "MIDWORD") == 0 && itemcnt == 2
|
|
|
4640 && midword == NULL)
|
|
|
4641 {
|
|
|
4642 midword = getroom_save(spin, items[1]);
|
|
366
|
4643 }
|
|
302
|
4644 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
|
4645 {
|
|
323
|
4646 /* ignored, we always split */
|
|
302
|
4647 }
|
|
323
|
4648 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
|
300
|
4649 {
|
|
323
|
4650 /* ignored, we look in the tree for what chars may appear */
|
|
300
|
4651 }
|
|
481
|
4652 else if (STRCMP(items[0], "SLASH") == 0 && itemcnt == 2
|
|
|
4653 && aff->af_slash == 0)
|
|
|
4654 {
|
|
|
4655 aff->af_slash = items[1][0];
|
|
|
4656 if (items[1][1] != NUL)
|
|
|
4657 smsg((char_u *)_("Character used for SLASH must be ASCII; in %s line %d: %s"),
|
|
|
4658 fname, lnum, items[1]);
|
|
|
4659 }
|
|
621
|
4660 /* TODO: remove "RAR" later */
|
|
|
4661 else if ((STRCMP(items[0], "RAR") == 0
|
|
|
4662 || STRCMP(items[0], "RARE") == 0) && itemcnt == 2
|
|
|
4663 && aff->af_rare == 0)
|
|
|
4664 {
|
|
|
4665 aff->af_rare = affitem2flag(aff->af_flagtype, items[1],
|
|
495
|
4666 fname, lnum);
|
|
307
|
4667 }
|
|
621
|
4668 /* TODO: remove "KEP" later */
|
|
|
4669 else if ((STRCMP(items[0], "KEP") == 0
|
|
|
4670 || STRCMP(items[0], "KEEPCASE") == 0) && itemcnt == 2
|
|
|
4671 && aff->af_keepcase == 0)
|
|
|
4672 {
|
|
|
4673 aff->af_keepcase = affitem2flag(aff->af_flagtype, items[1],
|
|
495
|
4674 fname, lnum);
|
|
307
|
4675 }
|
|
346
|
4676 else if (STRCMP(items[0], "BAD") == 0 && itemcnt == 2
|
|
|
4677 && aff->af_bad == 0)
|
|
|
4678 {
|
|
495
|
4679 aff->af_bad = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4680 fname, lnum);
|
|
346
|
4681 }
|
|
483
|
4682 else if (STRCMP(items[0], "NEEDAFFIX") == 0 && itemcnt == 2
|
|
|
4683 && aff->af_needaffix == 0)
|
|
|
4684 {
|
|
495
|
4685 aff->af_needaffix = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4686 fname, lnum);
|
|
483
|
4687 }
|
|
500
|
4688 else if (STRCMP(items[0], "NEEDCOMPOUND") == 0 && itemcnt == 2
|
|
|
4689 && aff->af_needcomp == 0)
|
|
|
4690 {
|
|
|
4691 aff->af_needcomp = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4692 fname, lnum);
|
|
|
4693 }
|
|
481
|
4694 else if (STRCMP(items[0], "COMPOUNDFLAG") == 0 && itemcnt == 2
|
|
495
|
4695 && compflags == NULL)
|
|
|
4696 {
|
|
|
4697 /* Turn flag "c" into COMPOUNDFLAGS compatible string "c+",
|
|
|
4698 * "Na" into "Na+", "1234" into "1234+". */
|
|
|
4699 p = getroom(spin, STRLEN(items[1]) + 2, FALSE);
|
|
483
|
4700 if (p != NULL)
|
|
|
4701 {
|
|
495
|
4702 STRCPY(p, items[1]);
|
|
|
4703 STRCAT(p, "+");
|
|
|
4704 compflags = p;
|
|
|
4705 }
|
|
481
|
4706 }
|
|
483
|
4707 else if (STRCMP(items[0], "COMPOUNDFLAGS") == 0 && itemcnt == 2)
|
|
|
4708 {
|
|
|
4709 /* Concatenate this string to previously defined ones, using a
|
|
|
4710 * slash to separate them. */
|
|
|
4711 l = STRLEN(items[1]) + 1;
|
|
495
|
4712 if (compflags != NULL)
|
|
|
4713 l += STRLEN(compflags) + 1;
|
|
483
|
4714 p = getroom(spin, l, FALSE);
|
|
|
4715 if (p != NULL)
|
|
|
4716 {
|
|
495
|
4717 if (compflags != NULL)
|
|
483
|
4718 {
|
|
495
|
4719 STRCPY(p, compflags);
|
|
483
|
4720 STRCAT(p, "/");
|
|
|
4721 }
|
|
|
4722 STRCAT(p, items[1]);
|
|
495
|
4723 compflags = p;
|
|
483
|
4724 }
|
|
|
4725 }
|
|
|
4726 else if (STRCMP(items[0], "COMPOUNDMAX") == 0 && itemcnt == 2
|
|
495
|
4727 && compmax == 0)
|
|
|
4728 {
|
|
|
4729 compmax = atoi((char *)items[1]);
|
|
|
4730 if (compmax == 0)
|
|
483
|
4731 smsg((char_u *)_("Wrong COMPOUNDMAX value in %s line %d: %s"),
|
|
|
4732 fname, lnum, items[1]);
|
|
481
|
4733 }
|
|
|
4734 else if (STRCMP(items[0], "COMPOUNDMIN") == 0 && itemcnt == 2
|
|
495
|
4735 && compminlen == 0)
|
|
|
4736 {
|
|
|
4737 compminlen = atoi((char *)items[1]);
|
|
|
4738 if (compminlen == 0)
|
|
481
|
4739 smsg((char_u *)_("Wrong COMPOUNDMIN value in %s line %d: %s"),
|
|
|
4740 fname, lnum, items[1]);
|
|
|
4741 }
|
|
483
|
4742 else if (STRCMP(items[0], "COMPOUNDSYLMAX") == 0 && itemcnt == 2
|
|
495
|
4743 && compsylmax == 0)
|
|
|
4744 {
|
|
|
4745 compsylmax = atoi((char *)items[1]);
|
|
|
4746 if (compsylmax == 0)
|
|
483
|
4747 smsg((char_u *)_("Wrong COMPOUNDSYLMAX value in %s line %d: %s"),
|
|
|
4748 fname, lnum, items[1]);
|
|
|
4749 }
|
|
|
4750 else if (STRCMP(items[0], "SYLLABLE") == 0 && itemcnt == 2
|
|
495
|
4751 && syllable == NULL)
|
|
|
4752 {
|
|
|
4753 syllable = getroom_save(spin, items[1]);
|
|
483
|
4754 }
|
|
492
|
4755 else if (STRCMP(items[0], "NOBREAK") == 0 && itemcnt == 1)
|
|
|
4756 {
|
|
|
4757 spin->si_nobreak = TRUE;
|
|
|
4758 }
|
|
339
|
4759 else if (STRCMP(items[0], "PFXPOSTPONE") == 0 && itemcnt == 1)
|
|
|
4760 {
|
|
|
4761 aff->af_pfxpostpone = TRUE;
|
|
|
4762 }
|
|
236
|
4763 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4764 || STRCMP(items[0], "SFX") == 0)
|
|
|
4765 && aff_todo == 0
|
|
334
|
4766 && itemcnt >= 4)
|
|
236
|
4767 {
|
|
498
|
4768 int lasti = 4;
|
|
|
4769 char_u key[AH_KEY_LEN];
|
|
|
4770
|
|
|
4771 if (*items[0] == 'P')
|
|
|
4772 tp = &aff->af_pref;
|
|
|
4773 else
|
|
|
4774 tp = &aff->af_suff;
|
|
|
4775
|
|
|
4776 /* Myspell allows the same affix name to be used multiple
|
|
|
4777 * times. The affix files that do this have an undocumented
|
|
|
4778 * "S" flag on all but the last block, thus we check for that
|
|
|
4779 * and store it in ah_follows. */
|
|
|
4780 vim_strncpy(key, items[1], AH_KEY_LEN - 1);
|
|
|
4781 hi = hash_find(tp, key);
|
|
|
4782 if (!HASHITEM_EMPTY(hi))
|
|
|
4783 {
|
|
|
4784 cur_aff = HI2AH(hi);
|
|
|
4785 if (cur_aff->ah_combine != (*items[2] == 'Y'))
|
|
|
4786 smsg((char_u *)_("Different combining flag in continued affix block in %s line %d: %s"),
|
|
|
4787 fname, lnum, items[1]);
|
|
|
4788 if (!cur_aff->ah_follows)
|
|
|
4789 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
|
4790 fname, lnum, items[1]);
|
|
|
4791 }
|
|
|
4792 else
|
|
|
4793 {
|
|
|
4794 /* New affix letter. */
|
|
|
4795 cur_aff = (affheader_T *)getroom(spin,
|
|
|
4796 sizeof(affheader_T), TRUE);
|
|
|
4797 if (cur_aff == NULL)
|
|
|
4798 break;
|
|
|
4799 cur_aff->ah_flag = affitem2flag(aff->af_flagtype, items[1],
|
|
|
4800 fname, lnum);
|
|
|
4801 if (cur_aff->ah_flag == 0 || STRLEN(items[1]) >= AH_KEY_LEN)
|
|
|
4802 break;
|
|
|
4803 if (cur_aff->ah_flag == aff->af_bad
|
|
621
|
4804 || cur_aff->ah_flag == aff->af_rare
|
|
|
4805 || cur_aff->ah_flag == aff->af_keepcase
|
|
500
|
4806 || cur_aff->ah_flag == aff->af_needaffix
|
|
|
4807 || cur_aff->ah_flag == aff->af_needcomp)
|
|
621
|
4808 smsg((char_u *)_("Affix also used for BAD/RARE/KEEPCASE/NEEDAFFIX/NEEDCOMPOUND in %s line %d: %s"),
|
|
498
|
4809 fname, lnum, items[1]);
|
|
|
4810 STRCPY(cur_aff->ah_key, items[1]);
|
|
|
4811 hash_add(tp, cur_aff->ah_key);
|
|
|
4812
|
|
|
4813 cur_aff->ah_combine = (*items[2] == 'Y');
|
|
|
4814 }
|
|
|
4815
|
|
|
4816 /* Check for the "S" flag, which apparently means that another
|
|
|
4817 * block with the same affix name is following. */
|
|
|
4818 if (itemcnt > lasti && STRCMP(items[lasti], "S") == 0)
|
|
|
4819 {
|
|
|
4820 ++lasti;
|
|
|
4821 cur_aff->ah_follows = TRUE;
|
|
|
4822 }
|
|
|
4823 else
|
|
|
4824 cur_aff->ah_follows = FALSE;
|
|
|
4825
|
|
334
|
4826 /* Myspell allows extra text after the item, but that might
|
|
|
4827 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
498
|
4828 if (itemcnt > lasti && *items[lasti] != '#')
|
|
334
|
4829 smsg((char_u *)_("Trailing text in %s line %d: %s"),
|
|
|
4830 fname, lnum, items[4]);
|
|
|
4831
|
|
498
|
4832 if (STRCMP(items[2], "Y") != 0 && STRCMP(items[2], "N") != 0)
|
|
236
|
4833 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
|
4834 fname, lnum, items[2]);
|
|
339
|
4835
|
|
498
|
4836 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
|
4837 {
|
|
|
4838 if (cur_aff->ah_newID == 0)
|
|
495
|
4839 {
|
|
|
4840 /* Use a new number in the .spl file later, to be able
|
|
|
4841 * to handle multiple .aff files. */
|
|
500
|
4842 check_renumber(spin);
|
|
495
|
4843 cur_aff->ah_newID = ++spin->si_newprefID;
|
|
|
4844
|
|
|
4845 /* We only really use ah_newID if the prefix is
|
|
|
4846 * postponed. We know that only after handling all
|
|
|
4847 * the items. */
|
|
|
4848 did_postpone_prefix = FALSE;
|
|
|
4849 }
|
|
498
|
4850 else
|
|
|
4851 /* Did use the ID in a previous block. */
|
|
|
4852 did_postpone_prefix = TRUE;
|
|
|
4853 }
|
|
|
4854
|
|
300
|
4855 aff_todo = atoi((char *)items[3]);
|
|
236
|
4856 }
|
|
|
4857 else if ((STRCMP(items[0], "PFX") == 0
|
|
|
4858 || STRCMP(items[0], "SFX") == 0)
|
|
|
4859 && aff_todo > 0
|
|
|
4860 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
334
|
4861 && itemcnt >= 5)
|
|
236
|
4862 {
|
|
|
4863 affentry_T *aff_entry;
|
|
366
|
4864 int rare = FALSE;
|
|
483
|
4865 int nocomp = FALSE;
|
|
455
|
4866 int upper = FALSE;
|
|
366
|
4867 int lasti = 5;
|
|
|
4868
|
|
483
|
4869 /* Check for "rare" and "nocomp" after the other info. */
|
|
|
4870 while (itemcnt > lasti)
|
|
|
4871 {
|
|
|
4872 if (!rare && STRICMP(items[lasti], "rare") == 0)
|
|
|
4873 {
|
|
|
4874 rare = TRUE;
|
|
|
4875 ++lasti;
|
|
|
4876 }
|
|
|
4877 else if (!nocomp && STRICMP(items[lasti], "nocomp") == 0)
|
|
|
4878 {
|
|
|
4879 nocomp = TRUE;
|
|
|
4880 ++lasti;
|
|
|
4881 }
|
|
|
4882 else
|
|
|
4883 break;
|
|
366
|
4884 }
|
|
236
|
4885
|
|
334
|
4886 /* Myspell allows extra text after the item, but that might
|
|
|
4887 * mean mistakes go unnoticed. Require a comment-starter. */
|
|
366
|
4888 if (itemcnt > lasti && *items[lasti] != '#')
|
|
482
|
4889 smsg((char_u *)_(e_afftrailing), fname, lnum, items[lasti]);
|
|
334
|
4890
|
|
236
|
4891 /* New item for an affix letter. */
|
|
|
4892 --aff_todo;
|
|
474
|
4893 aff_entry = (affentry_T *)getroom(spin,
|
|
447
|
4894 sizeof(affentry_T), TRUE);
|
|
236
|
4895 if (aff_entry == NULL)
|
|
|
4896 break;
|
|
366
|
4897 aff_entry->ae_rare = rare;
|
|
483
|
4898 aff_entry->ae_nocomp = nocomp;
|
|
240
|
4899
|
|
236
|
4900 if (STRCMP(items[2], "0") != 0)
|
|
474
|
4901 aff_entry->ae_chop = getroom_save(spin, items[2]);
|
|
236
|
4902 if (STRCMP(items[3], "0") != 0)
|
|
474
|
4903 aff_entry->ae_add = getroom_save(spin, items[3]);
|
|
236
|
4904
|
|
300
|
4905 /* Don't use an affix entry with non-ASCII characters when
|
|
|
4906 * "spin->si_ascii" is TRUE. */
|
|
|
4907 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
|
240
|
4908 || has_non_ascii(aff_entry->ae_add)))
|
|
|
4909 {
|
|
|
4910 aff_entry->ae_next = cur_aff->ah_first;
|
|
|
4911 cur_aff->ah_first = aff_entry;
|
|
300
|
4912
|
|
|
4913 if (STRCMP(items[4], ".") != 0)
|
|
|
4914 {
|
|
|
4915 char_u buf[MAXLINELEN];
|
|
|
4916
|
|
474
|
4917 aff_entry->ae_cond = getroom_save(spin, items[4]);
|
|
300
|
4918 if (*items[0] == 'P')
|
|
|
4919 sprintf((char *)buf, "^%s", items[4]);
|
|
|
4920 else
|
|
|
4921 sprintf((char *)buf, "%s$", items[4]);
|
|
|
4922 aff_entry->ae_prog = vim_regcomp(buf,
|
|
481
|
4923 RE_MAGIC + RE_STRING + RE_STRICT);
|
|
|
4924 if (aff_entry->ae_prog == NULL)
|
|
|
4925 smsg((char_u *)_("Broken condition in %s line %d: %s"),
|
|
|
4926 fname, lnum, items[4]);
|
|
300
|
4927 }
|
|
339
|
4928
|
|
|
4929 /* For postponed prefixes we need an entry in si_prefcond
|
|
|
4930 * for the condition. Use an existing one if possible. */
|
|
455
|
4931 if (*items[0] == 'P' && aff->af_pfxpostpone)
|
|
339
|
4932 {
|
|
455
|
4933 /* When the chop string is one lower-case letter and
|
|
|
4934 * the add string ends in the upper-case letter we set
|
|
|
4935 * the "upper" flag, clear "ae_chop" and remove the
|
|
|
4936 * letters from "ae_add". The condition must either
|
|
|
4937 * be empty or start with the same letter. */
|
|
|
4938 if (aff_entry->ae_chop != NULL
|
|
|
4939 && aff_entry->ae_add != NULL
|
|
|
4940 #ifdef FEAT_MBYTE
|
|
474
|
4941 && aff_entry->ae_chop[(*mb_ptr2len)(
|
|
455
|
4942 aff_entry->ae_chop)] == NUL
|
|
|
4943 #else
|
|
|
4944 && aff_entry->ae_chop[1] == NUL
|
|
|
4945 #endif
|
|
|
4946 )
|
|
339
|
4947 {
|
|
455
|
4948 int c, c_up;
|
|
|
4949
|
|
|
4950 c = PTR2CHAR(aff_entry->ae_chop);
|
|
|
4951 c_up = SPELL_TOUPPER(c);
|
|
|
4952 if (c_up != c
|
|
|
4953 && (aff_entry->ae_cond == NULL
|
|
|
4954 || PTR2CHAR(aff_entry->ae_cond) == c))
|
|
|
4955 {
|
|
|
4956 p = aff_entry->ae_add
|
|
|
4957 + STRLEN(aff_entry->ae_add);
|
|
|
4958 mb_ptr_back(aff_entry->ae_add, p);
|
|
|
4959 if (PTR2CHAR(p) == c_up)
|
|
|
4960 {
|
|
|
4961 upper = TRUE;
|
|
|
4962 aff_entry->ae_chop = NULL;
|
|
|
4963 *p = NUL;
|
|
|
4964
|
|
|
4965 /* The condition is matched with the
|
|
|
4966 * actual word, thus must check for the
|
|
|
4967 * upper-case letter. */
|
|
|
4968 if (aff_entry->ae_cond != NULL)
|
|
|
4969 {
|
|
|
4970 char_u buf[MAXLINELEN];
|
|
|
4971 #ifdef FEAT_MBYTE
|
|
|
4972 if (has_mbyte)
|
|
|
4973 {
|
|
|
4974 onecap_copy(items[4], buf, TRUE);
|
|
|
4975 aff_entry->ae_cond = getroom_save(
|
|
474
|
4976 spin, buf);
|
|
455
|
4977 }
|
|
|
4978 else
|
|
|
4979 #endif
|
|
|
4980 *aff_entry->ae_cond = c_up;
|
|
|
4981 if (aff_entry->ae_cond != NULL)
|
|
|
4982 {
|
|
|
4983 sprintf((char *)buf, "^%s",
|
|
339
|
4984 aff_entry->ae_cond);
|
|
455
|
4985 vim_free(aff_entry->ae_prog);
|
|
|
4986 aff_entry->ae_prog = vim_regcomp(
|
|
|
4987 buf, RE_MAGIC + RE_STRING);
|
|
|
4988 }
|
|
|
4989 }
|
|
|
4990 }
|
|
|
4991 }
|
|
339
|
4992 }
|
|
|
4993
|
|
455
|
4994 if (aff_entry->ae_chop == NULL)
|
|
390
|
4995 {
|
|
455
|
4996 int idx;
|
|
|
4997 char_u **pp;
|
|
|
4998 int n;
|
|
|
4999
|
|
495
|
5000 /* Find a previously used condition. */
|
|
455
|
5001 for (idx = spin->si_prefcond.ga_len - 1; idx >= 0;
|
|
|
5002 --idx)
|
|
|
5003 {
|
|
|
5004 p = ((char_u **)spin->si_prefcond.ga_data)[idx];
|
|
|
5005 if (str_equal(p, aff_entry->ae_cond))
|
|
|
5006 break;
|
|
|
5007 }
|
|
|
5008 if (idx < 0 && ga_grow(&spin->si_prefcond, 1) == OK)
|
|
|
5009 {
|
|
|
5010 /* Not found, add a new condition. */
|
|
|
5011 idx = spin->si_prefcond.ga_len++;
|
|
|
5012 pp = ((char_u **)spin->si_prefcond.ga_data)
|
|
|
5013 + idx;
|
|
|
5014 if (aff_entry->ae_cond == NULL)
|
|
|
5015 *pp = NULL;
|
|
|
5016 else
|
|
474
|
5017 *pp = getroom_save(spin,
|
|
455
|
5018 aff_entry->ae_cond);
|
|
|
5019 }
|
|
|
5020
|
|
|
5021 /* Add the prefix to the prefix tree. */
|
|
|
5022 if (aff_entry->ae_add == NULL)
|
|
|
5023 p = (char_u *)"";
|
|
390
|
5024 else
|
|
455
|
5025 p = aff_entry->ae_add;
|
|
|
5026 /* PFX_FLAGS is a negative number, so that
|
|
|
5027 * tree_add_word() knows this is the prefix tree. */
|
|
|
5028 n = PFX_FLAGS;
|
|
|
5029 if (rare)
|
|
|
5030 n |= WFP_RARE;
|
|
|
5031 if (!cur_aff->ah_combine)
|
|
|
5032 n |= WFP_NC;
|
|
|
5033 if (upper)
|
|
|
5034 n |= WFP_UP;
|
|
474
|
5035 tree_add_word(spin, p, spin->si_prefroot, n,
|
|
|
5036 idx, cur_aff->ah_newID);
|
|
495
|
5037 did_postpone_prefix = TRUE;
|
|
|
5038 }
|
|
|
5039
|
|
|
5040 /* Didn't actually use ah_newID, backup si_newprefID. */
|
|
|
5041 if (aff_todo == 0 && !did_postpone_prefix)
|
|
|
5042 {
|
|
|
5043 --spin->si_newprefID;
|
|
|
5044 cur_aff->ah_newID = 0;
|
|
390
|
5045 }
|
|
339
|
5046 }
|
|
240
|
5047 }
|
|
236
|
5048 }
|
|
495
|
5049 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2
|
|
|
5050 && fol == NULL)
|
|
|
5051 {
|
|
|
5052 fol = vim_strsave(items[1]);
|
|
|
5053 }
|
|
|
5054 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2
|
|
|
5055 && low == NULL)
|
|
|
5056 {
|
|
|
5057 low = vim_strsave(items[1]);
|
|
|
5058 }
|
|
|
5059 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2
|
|
|
5060 && upp == NULL)
|
|
|
5061 {
|
|
|
5062 upp = vim_strsave(items[1]);
|
|
255
|
5063 }
|
|
236
|
5064 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
|
323
|
5065 {
|
|
236
|
5066 /* Ignore REP count */;
|
|
323
|
5067 if (!isdigit(*items[1]))
|
|
|
5068 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
|
5069 fname, lnum);
|
|
|
5070 }
|
|
482
|
5071 else if (STRCMP(items[0], "REP") == 0 && itemcnt >= 3)
|
|
236
|
5072 {
|
|
|
5073 /* REP item */
|
|
482
|
5074 /* Myspell ignores extra arguments, we require it starts with
|
|
|
5075 * # to detect mistakes. */
|
|
|
5076 if (itemcnt > 3 && items[3][0] != '#')
|
|
|
5077 smsg((char_u *)_(e_afftrailing), fname, lnum, items[3]);
|
|
323
|
5078 if (do_rep)
|
|
532
|
5079 {
|
|
|
5080 /* Replace underscore with space (can't include a space
|
|
|
5081 * directly). */
|
|
|
5082 for (p = items[1]; *p != NUL; mb_ptr_adv(p))
|
|
|
5083 if (*p == '_')
|
|
|
5084 *p = ' ';
|
|
|
5085 for (p = items[2]; *p != NUL; mb_ptr_adv(p))
|
|
|
5086 if (*p == '_')
|
|
|
5087 *p = ' ';
|
|
323
|
5088 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
532
|
5089 }
|
|
323
|
5090 }
|
|
|
5091 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
|
5092 {
|
|
|
5093 /* MAP item or count */
|
|
|
5094 if (!found_map)
|
|
|
5095 {
|
|
|
5096 /* First line contains the count. */
|
|
|
5097 found_map = TRUE;
|
|
|
5098 if (!isdigit(*items[1]))
|
|
|
5099 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
|
5100 fname, lnum);
|
|
|
5101 }
|
|
|
5102 else if (do_map)
|
|
|
5103 {
|
|
346
|
5104 int c;
|
|
|
5105
|
|
|
5106 /* Check that every character appears only once. */
|
|
|
5107 for (p = items[1]; *p != NUL; )
|
|
|
5108 {
|
|
|
5109 #ifdef FEAT_MBYTE
|
|
|
5110 c = mb_ptr2char_adv(&p);
|
|
|
5111 #else
|
|
|
5112 c = *p++;
|
|
|
5113 #endif
|
|
|
5114 if ((spin->si_map.ga_len > 0
|
|
|
5115 && vim_strchr(spin->si_map.ga_data, c)
|
|
|
5116 != NULL)
|
|
|
5117 || vim_strchr(p, c) != NULL)
|
|
|
5118 smsg((char_u *)_("Duplicate character in MAP in %s line %d"),
|
|
|
5119 fname, lnum);
|
|
|
5120 }
|
|
|
5121
|
|
323
|
5122 /* We simply concatenate all the MAP strings, separated by
|
|
|
5123 * slashes. */
|
|
|
5124 ga_concat(&spin->si_map, items[1]);
|
|
|
5125 ga_append(&spin->si_map, '/');
|
|
|
5126 }
|
|
|
5127 }
|
|
534
|
5128 /* Accept "SAL from to" and "SAL from to # comment". */
|
|
|
5129 else if (STRCMP(items[0], "SAL") == 0
|
|
|
5130 && (itemcnt == 3 || (itemcnt > 3 && items[3][0] == '#')))
|
|
323
|
5131 {
|
|
|
5132 if (do_sal)
|
|
|
5133 {
|
|
|
5134 /* SAL item (sounds-a-like)
|
|
|
5135 * Either one of the known keys or a from-to pair. */
|
|
|
5136 if (STRCMP(items[1], "followup") == 0)
|
|
|
5137 spin->si_followup = sal_to_bool(items[2]);
|
|
|
5138 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
|
5139 spin->si_collapse = sal_to_bool(items[2]);
|
|
|
5140 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
|
5141 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
|
5142 else
|
|
|
5143 /* when "to" is "_" it means empty */
|
|
|
5144 add_fromto(spin, &spin->si_sal, items[1],
|
|
|
5145 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
|
5146 : items[2]);
|
|
|
5147 }
|
|
236
|
5148 }
|
|
375
|
5149 else if (STRCMP(items[0], "SOFOFROM") == 0 && itemcnt == 2
|
|
495
|
5150 && sofofrom == NULL)
|
|
|
5151 {
|
|
|
5152 sofofrom = getroom_save(spin, items[1]);
|
|
375
|
5153 }
|
|
|
5154 else if (STRCMP(items[0], "SOFOTO") == 0 && itemcnt == 2
|
|
495
|
5155 && sofoto == NULL)
|
|
|
5156 {
|
|
|
5157 sofoto = getroom_save(spin, items[1]);
|
|
375
|
5158 }
|
|
300
|
5159 else
|
|
481
|
5160 smsg((char_u *)_("Unrecognized or duplicate item in %s line %d: %s"),
|
|
236
|
5161 fname, lnum, items[0]);
|
|
|
5162 }
|
|
|
5163 }
|
|
|
5164
|
|
255
|
5165 if (fol != NULL || low != NULL || upp != NULL)
|
|
|
5166 {
|
|
351
|
5167 if (spin->si_clear_chartab)
|
|
|
5168 {
|
|
|
5169 /* Clear the char type tables, don't want to use any of the
|
|
|
5170 * currently used spell properties. */
|
|
|
5171 init_spell_chartab();
|
|
|
5172 spin->si_clear_chartab = FALSE;
|
|
|
5173 }
|
|
|
5174
|
|
316
|
5175 /*
|
|
|
5176 * Don't write a word table for an ASCII file, so that we don't check
|
|
|
5177 * for conflicts with a word table that matches 'encoding'.
|
|
324
|
5178 * Don't write one for utf-8 either, we use utf_*() and
|
|
316
|
5179 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
|
5180 */
|
|
|
5181 if (!spin->si_ascii
|
|
|
5182 #ifdef FEAT_MBYTE
|
|
|
5183 && !enc_utf8
|
|
|
5184 #endif
|
|
|
5185 )
|
|
260
|
5186 {
|
|
|
5187 if (fol == NULL || low == NULL || upp == NULL)
|
|
|
5188 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
|
5189 else
|
|
316
|
5190 (void)set_spell_chartab(fol, low, upp);
|
|
260
|
5191 }
|
|
255
|
5192
|
|
|
5193 vim_free(fol);
|
|
|
5194 vim_free(low);
|
|
|
5195 vim_free(upp);
|
|
|
5196 }
|
|
|
5197
|
|
481
|
5198 /* Use compound specifications of the .aff file for the spell info. */
|
|
495
|
5199 if (compmax != 0)
|
|
|
5200 {
|
|
|
5201 aff_check_number(spin->si_compmax, compmax, "COMPOUNDMAX");
|
|
|
5202 spin->si_compmax = compmax;
|
|
|
5203 }
|
|
|
5204
|
|
|
5205 if (compminlen != 0)
|
|
|
5206 {
|
|
|
5207 aff_check_number(spin->si_compminlen, compminlen, "COMPOUNDMIN");
|
|
|
5208 spin->si_compminlen = compminlen;
|
|
|
5209 }
|
|
|
5210
|
|
|
5211 if (compsylmax != 0)
|
|
|
5212 {
|
|
|
5213 if (syllable == NULL)
|
|
|
5214 smsg((char_u *)_("COMPOUNDSYLMAX used without SYLLABLE"));
|
|
|
5215 aff_check_number(spin->si_compsylmax, compsylmax, "COMPOUNDSYLMAX");
|
|
|
5216 spin->si_compsylmax = compsylmax;
|
|
|
5217 }
|
|
|
5218
|
|
|
5219 if (compflags != NULL)
|
|
|
5220 process_compflags(spin, aff, compflags);
|
|
|
5221
|
|
|
5222 /* Check that we didn't use too many renumbered flags. */
|
|
500
|
5223 if (spin->si_newcompID < spin->si_newprefID)
|
|
|
5224 {
|
|
|
5225 if (spin->si_newcompID == 127 || spin->si_newcompID == 255)
|
|
495
|
5226 MSG(_("Too many postponed prefixes"));
|
|
500
|
5227 else if (spin->si_newprefID == 0 || spin->si_newprefID == 127)
|
|
495
|
5228 MSG(_("Too many compound flags"));
|
|
481
|
5229 else
|
|
495
|
5230 MSG(_("Too many posponed prefixes and/or compound flags"));
|
|
|
5231 }
|
|
|
5232
|
|
|
5233 if (syllable != NULL)
|
|
|
5234 {
|
|
|
5235 aff_check_string(spin->si_syllable, syllable, "SYLLABLE");
|
|
|
5236 spin->si_syllable = syllable;
|
|
|
5237 }
|
|
|
5238
|
|
|
5239 if (sofofrom != NULL || sofoto != NULL)
|
|
|
5240 {
|
|
|
5241 if (sofofrom == NULL || sofoto == NULL)
|
|
|
5242 smsg((char_u *)_("Missing SOFO%s line in %s"),
|
|
|
5243 sofofrom == NULL ? "FROM" : "TO", fname);
|
|
|
5244 else if (spin->si_sal.ga_len > 0)
|
|
|
5245 smsg((char_u *)_("Both SAL and SOFO lines in %s"), fname);
|
|
483
|
5246 else
|
|
495
|
5247 {
|
|
|
5248 aff_check_string(spin->si_sofofr, sofofrom, "SOFOFROM");
|
|
|
5249 aff_check_string(spin->si_sofoto, sofoto, "SOFOTO");
|
|
|
5250 spin->si_sofofr = sofofrom;
|
|
|
5251 spin->si_sofoto = sofoto;
|
|
|
5252 }
|
|
|
5253 }
|
|
|
5254
|
|
|
5255 if (midword != NULL)
|
|
|
5256 {
|
|
|
5257 aff_check_string(spin->si_midword, midword, "MIDWORD");
|
|
|
5258 spin->si_midword = midword;
|
|
483
|
5259 }
|
|
|
5260
|
|
236
|
5261 vim_free(pc);
|
|
|
5262 fclose(fd);
|
|
|
5263 return aff;
|
|
|
5264 }
|
|
|
5265
|
|
|
5266 /*
|
|
495
|
5267 * Turn an affix flag name into a number, according to the FLAG type.
|
|
|
5268 * returns zero for failure.
|
|
|
5269 */
|
|
|
5270 static unsigned
|
|
|
5271 affitem2flag(flagtype, item, fname, lnum)
|
|
|
5272 int flagtype;
|
|
|
5273 char_u *item;
|
|
|
5274 char_u *fname;
|
|
|
5275 int lnum;
|
|
|
5276 {
|
|
|
5277 unsigned res;
|
|
|
5278 char_u *p = item;
|
|
|
5279
|
|
|
5280 res = get_affitem(flagtype, &p);
|
|
|
5281 if (res == 0)
|
|
|
5282 {
|
|
498
|
5283 if (flagtype == AFT_NUM)
|
|
495
|
5284 smsg((char_u *)_("Flag is not a number in %s line %d: %s"),
|
|
|
5285 fname, lnum, item);
|
|
|
5286 else
|
|
|
5287 smsg((char_u *)_("Illegal flag in %s line %d: %s"),
|
|
|
5288 fname, lnum, item);
|
|
|
5289 }
|
|
|
5290 if (*p != NUL)
|
|
|
5291 {
|
|
|
5292 smsg((char_u *)_(e_affname), fname, lnum, item);
|
|
|
5293 return 0;
|
|
|
5294 }
|
|
|
5295
|
|
|
5296 return res;
|
|
|
5297 }
|
|
|
5298
|
|
|
5299 /*
|
|
|
5300 * Get one affix name from "*pp" and advance the pointer.
|
|
|
5301 * Returns zero for an error, still advances the pointer then.
|
|
|
5302 */
|
|
|
5303 static unsigned
|
|
|
5304 get_affitem(flagtype, pp)
|
|
|
5305 int flagtype;
|
|
|
5306 char_u **pp;
|
|
|
5307 {
|
|
|
5308 int res;
|
|
|
5309
|
|
498
|
5310 if (flagtype == AFT_NUM)
|
|
495
|
5311 {
|
|
|
5312 if (!VIM_ISDIGIT(**pp))
|
|
|
5313 {
|
|
498
|
5314 ++*pp; /* always advance, avoid getting stuck */
|
|
495
|
5315 return 0;
|
|
|
5316 }
|
|
|
5317 res = getdigits(pp);
|
|
|
5318 }
|
|
|
5319 else
|
|
|
5320 {
|
|
|
5321 #ifdef FEAT_MBYTE
|
|
|
5322 res = mb_ptr2char_adv(pp);
|
|
|
5323 #else
|
|
|
5324 res = *(*pp)++;
|
|
|
5325 #endif
|
|
498
|
5326 if (flagtype == AFT_LONG || (flagtype == AFT_CAPLONG
|
|
495
|
5327 && res >= 'A' && res <= 'Z'))
|
|
|
5328 {
|
|
|
5329 if (**pp == NUL)
|
|
|
5330 return 0;
|
|
|
5331 #ifdef FEAT_MBYTE
|
|
|
5332 res = mb_ptr2char_adv(pp) + (res << 16);
|
|
|
5333 #else
|
|
|
5334 res = *(*pp)++ + (res << 16);
|
|
|
5335 #endif
|
|
|
5336 }
|
|
|
5337 }
|
|
|
5338 return res;
|
|
|
5339 }
|
|
|
5340
|
|
|
5341 /*
|
|
|
5342 * Process the "compflags" string used in an affix file and append it to
|
|
|
5343 * spin->si_compflags.
|
|
|
5344 * The processing involves changing the affix names to ID numbers, so that
|
|
|
5345 * they fit in one byte.
|
|
|
5346 */
|
|
|
5347 static void
|
|
|
5348 process_compflags(spin, aff, compflags)
|
|
|
5349 spellinfo_T *spin;
|
|
|
5350 afffile_T *aff;
|
|
|
5351 char_u *compflags;
|
|
|
5352 {
|
|
|
5353 char_u *p;
|
|
|
5354 char_u *prevp;
|
|
|
5355 unsigned flag;
|
|
|
5356 compitem_T *ci;
|
|
|
5357 int id;
|
|
|
5358 int len;
|
|
|
5359 char_u *tp;
|
|
|
5360 char_u key[AH_KEY_LEN];
|
|
|
5361 hashitem_T *hi;
|
|
|
5362
|
|
|
5363 /* Make room for the old and the new compflags, concatenated with a / in
|
|
|
5364 * between. Processing it makes it shorter, but we don't know by how
|
|
|
5365 * much, thus allocate the maximum. */
|
|
|
5366 len = STRLEN(compflags) + 1;
|
|
|
5367 if (spin->si_compflags != NULL)
|
|
|
5368 len += STRLEN(spin->si_compflags) + 1;
|
|
|
5369 p = getroom(spin, len, FALSE);
|
|
|
5370 if (p == NULL)
|
|
|
5371 return;
|
|
|
5372 if (spin->si_compflags != NULL)
|
|
|
5373 {
|
|
|
5374 STRCPY(p, spin->si_compflags);
|
|
|
5375 STRCAT(p, "/");
|
|
|
5376 }
|
|
|
5377 spin->si_compflags = p;
|
|
|
5378 tp = p + STRLEN(p);
|
|
|
5379
|
|
|
5380 for (p = compflags; *p != NUL; )
|
|
|
5381 {
|
|
|
5382 if (vim_strchr((char_u *)"/*+[]", *p) != NULL)
|
|
|
5383 /* Copy non-flag characters directly. */
|
|
|
5384 *tp++ = *p++;
|
|
|
5385 else
|
|
|
5386 {
|
|
|
5387 /* First get the flag number, also checks validity. */
|
|
|
5388 prevp = p;
|
|
|
5389 flag = get_affitem(aff->af_flagtype, &p);
|
|
|
5390 if (flag != 0)
|
|
|
5391 {
|
|
|
5392 /* Find the flag in the hashtable. If it was used before, use
|
|
|
5393 * the existing ID. Otherwise add a new entry. */
|
|
|
5394 vim_strncpy(key, prevp, p - prevp);
|
|
|
5395 hi = hash_find(&aff->af_comp, key);
|
|
|
5396 if (!HASHITEM_EMPTY(hi))
|
|
|
5397 id = HI2CI(hi)->ci_newID;
|
|
|
5398 else
|
|
|
5399 {
|
|
|
5400 ci = (compitem_T *)getroom(spin, sizeof(compitem_T), TRUE);
|
|
|
5401 if (ci == NULL)
|
|
|
5402 break;
|
|
|
5403 STRCPY(ci->ci_key, key);
|
|
|
5404 ci->ci_flag = flag;
|
|
|
5405 /* Avoid using a flag ID that has a special meaning in a
|
|
|
5406 * regexp (also inside []). */
|
|
|
5407 do
|
|
|
5408 {
|
|
500
|
5409 check_renumber(spin);
|
|
|
5410 id = spin->si_newcompID--;
|
|
|
5411 } while (vim_strchr((char_u *)"/+*[]\\-^", id) != NULL);
|
|
495
|
5412 ci->ci_newID = id;
|
|
|
5413 hash_add(&aff->af_comp, ci->ci_key);
|
|
|
5414 }
|
|
|
5415 *tp++ = id;
|
|
|
5416 }
|
|
498
|
5417 if (aff->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
5418 ++p;
|
|
|
5419 }
|
|
|
5420 }
|
|
|
5421
|
|
|
5422 *tp = NUL;
|
|
|
5423 }
|
|
|
5424
|
|
|
5425 /*
|
|
500
|
5426 * Check that the new IDs for postponed affixes and compounding don't overrun
|
|
|
5427 * each other. We have almost 255 available, but start at 0-127 to avoid
|
|
|
5428 * using two bytes for utf-8. When the 0-127 range is used up go to 128-255.
|
|
|
5429 * When that is used up an error message is given.
|
|
|
5430 */
|
|
|
5431 static void
|
|
|
5432 check_renumber(spin)
|
|
|
5433 spellinfo_T *spin;
|
|
|
5434 {
|
|
|
5435 if (spin->si_newprefID == spin->si_newcompID && spin->si_newcompID < 128)
|
|
|
5436 {
|
|
|
5437 spin->si_newprefID = 127;
|
|
|
5438 spin->si_newcompID = 255;
|
|
|
5439 }
|
|
|
5440 }
|
|
|
5441
|
|
|
5442 /*
|
|
495
|
5443 * Return TRUE if flag "flag" appears in affix list "afflist".
|
|
|
5444 */
|
|
|
5445 static int
|
|
|
5446 flag_in_afflist(flagtype, afflist, flag)
|
|
|
5447 int flagtype;
|
|
|
5448 char_u *afflist;
|
|
|
5449 unsigned flag;
|
|
|
5450 {
|
|
|
5451 char_u *p;
|
|
|
5452 unsigned n;
|
|
|
5453
|
|
|
5454 switch (flagtype)
|
|
|
5455 {
|
|
|
5456 case AFT_CHAR:
|
|
|
5457 return vim_strchr(afflist, flag) != NULL;
|
|
|
5458
|
|
498
|
5459 case AFT_CAPLONG:
|
|
|
5460 case AFT_LONG:
|
|
495
|
5461 for (p = afflist; *p != NUL; )
|
|
|
5462 {
|
|
|
5463 #ifdef FEAT_MBYTE
|
|
|
5464 n = mb_ptr2char_adv(&p);
|
|
|
5465 #else
|
|
|
5466 n = *p++;
|
|
|
5467 #endif
|
|
498
|
5468 if ((flagtype == AFT_LONG || (n >= 'A' && n <= 'Z'))
|
|
495
|
5469 && *p != NUL)
|
|
|
5470 #ifdef FEAT_MBYTE
|
|
|
5471 n = mb_ptr2char_adv(&p) + (n << 16);
|
|
|
5472 #else
|
|
|
5473 n = *p++ + (n << 16);
|
|
|
5474 #endif
|
|
|
5475 if (n == flag)
|
|
|
5476 return TRUE;
|
|
|
5477 }
|
|
|
5478 break;
|
|
|
5479
|
|
498
|
5480 case AFT_NUM:
|
|
495
|
5481 for (p = afflist; *p != NUL; )
|
|
|
5482 {
|
|
|
5483 n = getdigits(&p);
|
|
|
5484 if (n == flag)
|
|
|
5485 return TRUE;
|
|
|
5486 if (*p != NUL) /* skip over comma */
|
|
|
5487 ++p;
|
|
|
5488 }
|
|
|
5489 break;
|
|
|
5490 }
|
|
|
5491 return FALSE;
|
|
|
5492 }
|
|
|
5493
|
|
|
5494 /*
|
|
|
5495 * Give a warning when "spinval" and "affval" numbers are set and not the same.
|
|
|
5496 */
|
|
|
5497 static void
|
|
|
5498 aff_check_number(spinval, affval, name)
|
|
|
5499 int spinval;
|
|
|
5500 int affval;
|
|
|
5501 char *name;
|
|
|
5502 {
|
|
|
5503 if (spinval != 0 && spinval != affval)
|
|
|
5504 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
5505 }
|
|
|
5506
|
|
|
5507 /*
|
|
|
5508 * Give a warning when "spinval" and "affval" strings are set and not the same.
|
|
|
5509 */
|
|
|
5510 static void
|
|
|
5511 aff_check_string(spinval, affval, name)
|
|
|
5512 char_u *spinval;
|
|
|
5513 char_u *affval;
|
|
|
5514 char *name;
|
|
|
5515 {
|
|
|
5516 if (spinval != NULL && STRCMP(spinval, affval) != 0)
|
|
|
5517 smsg((char_u *)_("%s value differs from what is used in another .aff file"), name);
|
|
|
5518 }
|
|
|
5519
|
|
|
5520 /*
|
|
339
|
5521 * Return TRUE if strings "s1" and "s2" are equal. Also consider both being
|
|
|
5522 * NULL as equal.
|
|
|
5523 */
|
|
|
5524 static int
|
|
|
5525 str_equal(s1, s2)
|
|
|
5526 char_u *s1;
|
|
|
5527 char_u *s2;
|
|
|
5528 {
|
|
|
5529 if (s1 == NULL || s2 == NULL)
|
|
|
5530 return s1 == s2;
|
|
|
5531 return STRCMP(s1, s2) == 0;
|
|
|
5532 }
|
|
|
5533
|
|
|
5534 /*
|
|
323
|
5535 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
|
5536 * They are stored case-folded.
|
|
|
5537 */
|
|
|
5538 static void
|
|
|
5539 add_fromto(spin, gap, from, to)
|
|
|
5540 spellinfo_T *spin;
|
|
|
5541 garray_T *gap;
|
|
|
5542 char_u *from;
|
|
|
5543 char_u *to;
|
|
|
5544 {
|
|
|
5545 fromto_T *ftp;
|
|
|
5546 char_u word[MAXWLEN];
|
|
|
5547
|
|
|
5548 if (ga_grow(gap, 1) == OK)
|
|
|
5549 {
|
|
|
5550 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
|
5551 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
474
|
5552 ftp->ft_from = getroom_save(spin, word);
|
|
323
|
5553 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
474
|
5554 ftp->ft_to = getroom_save(spin, word);
|
|
323
|
5555 ++gap->ga_len;
|
|
|
5556 }
|
|
|
5557 }
|
|
|
5558
|
|
|
5559 /*
|
|
|
5560 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
|
5561 */
|
|
|
5562 static int
|
|
|
5563 sal_to_bool(s)
|
|
|
5564 char_u *s;
|
|
|
5565 {
|
|
|
5566 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
|
5567 }
|
|
|
5568
|
|
|
5569 /*
|
|
240
|
5570 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
|
5571 * When "s" is NULL FALSE is returned.
|
|
|
5572 */
|
|
|
5573 static int
|
|
|
5574 has_non_ascii(s)
|
|
|
5575 char_u *s;
|
|
|
5576 {
|
|
|
5577 char_u *p;
|
|
|
5578
|
|
|
5579 if (s != NULL)
|
|
|
5580 for (p = s; *p != NUL; ++p)
|
|
|
5581 if (*p >= 128)
|
|
|
5582 return TRUE;
|
|
|
5583 return FALSE;
|
|
|
5584 }
|
|
|
5585
|
|
|
5586 /*
|
|
236
|
5587 * Free the structure filled by spell_read_aff().
|
|
|
5588 */
|
|
|
5589 static void
|
|
|
5590 spell_free_aff(aff)
|
|
|
5591 afffile_T *aff;
|
|
|
5592 {
|
|
|
5593 hashtab_T *ht;
|
|
|
5594 hashitem_T *hi;
|
|
|
5595 int todo;
|
|
|
5596 affheader_T *ah;
|
|
300
|
5597 affentry_T *ae;
|
|
236
|
5598
|
|
|
5599 vim_free(aff->af_enc);
|
|
|
5600
|
|
339
|
5601 /* All this trouble to free the "ae_prog" items... */
|
|
236
|
5602 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
|
5603 {
|
|
|
5604 todo = ht->ht_used;
|
|
|
5605 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
|
5606 {
|
|
|
5607 if (!HASHITEM_EMPTY(hi))
|
|
|
5608 {
|
|
|
5609 --todo;
|
|
|
5610 ah = HI2AH(hi);
|
|
300
|
5611 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
|
5612 vim_free(ae->ae_prog);
|
|
236
|
5613 }
|
|
|
5614 }
|
|
|
5615 if (ht == &aff->af_suff)
|
|
|
5616 break;
|
|
|
5617 }
|
|
300
|
5618
|
|
236
|
5619 hash_clear(&aff->af_pref);
|
|
|
5620 hash_clear(&aff->af_suff);
|
|
495
|
5621 hash_clear(&aff->af_comp);
|
|
236
|
5622 }
|
|
|
5623
|
|
|
5624 /*
|
|
300
|
5625 * Read dictionary file "fname".
|
|
236
|
5626 * Returns OK or FAIL;
|
|
|
5627 */
|
|
|
5628 static int
|
|
474
|
5629 spell_read_dic(spin, fname, affile)
|
|
|
5630 spellinfo_T *spin;
|
|
236
|
5631 char_u *fname;
|
|
300
|
5632 afffile_T *affile;
|
|
236
|
5633 {
|
|
300
|
5634 hashtab_T ht;
|
|
236
|
5635 char_u line[MAXLINELEN];
|
|
481
|
5636 char_u *p;
|
|
300
|
5637 char_u *afflist;
|
|
483
|
5638 char_u store_afflist[MAXWLEN];
|
|
|
5639 int pfxlen;
|
|
|
5640 int need_affix;
|
|
300
|
5641 char_u *dw;
|
|
236
|
5642 char_u *pc;
|
|
|
5643 char_u *w;
|
|
|
5644 int l;
|
|
|
5645 hash_T hash;
|
|
|
5646 hashitem_T *hi;
|
|
|
5647 FILE *fd;
|
|
|
5648 int lnum = 1;
|
|
300
|
5649 int non_ascii = 0;
|
|
|
5650 int retval = OK;
|
|
|
5651 char_u message[MAXLINELEN + MAXWLEN];
|
|
307
|
5652 int flags;
|
|
481
|
5653 int duplicate = 0;
|
|
236
|
5654
|
|
300
|
5655 /*
|
|
|
5656 * Open the file.
|
|
|
5657 */
|
|
310
|
5658 fd = mch_fopen((char *)fname, "r");
|
|
236
|
5659 if (fd == NULL)
|
|
|
5660 {
|
|
|
5661 EMSG2(_(e_notopen), fname);
|
|
|
5662 return FAIL;
|
|
|
5663 }
|
|
|
5664
|
|
300
|
5665 /* The hashtable is only used to detect duplicated words. */
|
|
|
5666 hash_init(&ht);
|
|
|
5667
|
|
310
|
5668 if (spin->si_verbose || p_verbose > 2)
|
|
|
5669 {
|
|
|
5670 if (!spin->si_verbose)
|
|
|
5671 verbose_enter();
|
|
366
|
5672 smsg((char_u *)_("Reading dictionary file %s ..."), fname);
|
|
310
|
5673 out_flush();
|
|
|
5674 if (!spin->si_verbose)
|
|
|
5675 verbose_leave();
|
|
|
5676 }
|
|
236
|
5677
|
|
474
|
5678 /* start with a message for the first line */
|
|
|
5679 spin->si_msg_count = 999999;
|
|
|
5680
|
|
236
|
5681 /* Read and ignore the first line: word count. */
|
|
|
5682 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
324
|
5683 if (!vim_isdigit(*skipwhite(line)))
|
|
236
|
5684 EMSG2(_("E760: No word count in %s"), fname);
|
|
|
5685
|
|
|
5686 /*
|
|
|
5687 * Read all the lines in the file one by one.
|
|
|
5688 * The words are converted to 'encoding' here, before being added to
|
|
|
5689 * the hashtable.
|
|
|
5690 */
|
|
255
|
5691 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
|
236
|
5692 {
|
|
255
|
5693 line_breakcheck();
|
|
236
|
5694 ++lnum;
|
|
455
|
5695 if (line[0] == '#' || line[0] == '/')
|
|
351
|
5696 continue; /* comment line */
|
|
236
|
5697
|
|
300
|
5698 /* Remove CR, LF and white space from the end. White space halfway
|
|
|
5699 * the word is kept to allow e.g., "et al.". */
|
|
236
|
5700 l = STRLEN(line);
|
|
|
5701 while (l > 0 && line[l - 1] <= ' ')
|
|
|
5702 --l;
|
|
|
5703 if (l == 0)
|
|
|
5704 continue; /* empty line */
|
|
|
5705 line[l] = NUL;
|
|
|
5706
|
|
481
|
5707 /* Find the optional affix names. Replace the SLASH character by a
|
|
|
5708 * slash. */
|
|
|
5709 afflist = NULL;
|
|
|
5710 for (p = line; *p != NUL; mb_ptr_adv(p))
|
|
|
5711 {
|
|
|
5712 if (*p == affile->af_slash)
|
|
|
5713 *p = '/';
|
|
|
5714 else if (*p == '/')
|
|
|
5715 {
|
|
|
5716 *p = NUL;
|
|
|
5717 afflist = p + 1;
|
|
|
5718 break;
|
|
|
5719 }
|
|
|
5720 }
|
|
236
|
5721
|
|
300
|
5722 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
5723 if (spin->si_ascii && has_non_ascii(line))
|
|
|
5724 {
|
|
|
5725 ++non_ascii;
|
|
240
|
5726 continue;
|
|
300
|
5727 }
|
|
240
|
5728
|
|
310
|
5729 #ifdef FEAT_MBYTE
|
|
236
|
5730 /* Convert from "SET" to 'encoding' when needed. */
|
|
300
|
5731 if (spin->si_conv.vc_type != CONV_NONE)
|
|
236
|
5732 {
|
|
300
|
5733 pc = string_convert(&spin->si_conv, line, NULL);
|
|
255
|
5734 if (pc == NULL)
|
|
|
5735 {
|
|
|
5736 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
5737 fname, lnum, line);
|
|
|
5738 continue;
|
|
|
5739 }
|
|
236
|
5740 w = pc;
|
|
|
5741 }
|
|
|
5742 else
|
|
310
|
5743 #endif
|
|
236
|
5744 {
|
|
|
5745 pc = NULL;
|
|
|
5746 w = line;
|
|
|
5747 }
|
|
|
5748
|
|
474
|
5749 /* This takes time, print a message every 10000 words. */
|
|
|
5750 if (spin->si_verbose && spin->si_msg_count > 10000)
|
|
|
5751 {
|
|
|
5752 spin->si_msg_count = 0;
|
|
339
|
5753 vim_snprintf((char *)message, sizeof(message),
|
|
|
5754 _("line %6d, word %6d - %s"),
|
|
|
5755 lnum, spin->si_foldwcount + spin->si_keepwcount, w);
|
|
|
5756 msg_start();
|
|
|
5757 msg_puts_long_attr(message, 0);
|
|
|
5758 msg_clr_eos();
|
|
|
5759 msg_didout = FALSE;
|
|
|
5760 msg_col = 0;
|
|
|
5761 out_flush();
|
|
|
5762 }
|
|
|
5763
|
|
300
|
5764 /* Store the word in the hashtable to be able to find duplicates. */
|
|
474
|
5765 dw = (char_u *)getroom_save(spin, w);
|
|
236
|
5766 if (dw == NULL)
|
|
300
|
5767 retval = FAIL;
|
|
|
5768 vim_free(pc);
|
|
|
5769 if (retval == FAIL)
|
|
236
|
5770 break;
|
|
|
5771
|
|
300
|
5772 hash = hash_hash(dw);
|
|
|
5773 hi = hash_lookup(&ht, dw, hash);
|
|
236
|
5774 if (!HASHITEM_EMPTY(hi))
|
|
481
|
5775 {
|
|
|
5776 if (p_verbose > 0)
|
|
|
5777 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
|
375
|
5778 fname, lnum, dw);
|
|
481
|
5779 else if (duplicate == 0)
|
|
|
5780 smsg((char_u *)_("First duplicate word in %s line %d: %s"),
|
|
|
5781 fname, lnum, dw);
|
|
|
5782 ++duplicate;
|
|
|
5783 }
|
|
236
|
5784 else
|
|
300
|
5785 hash_add_item(&ht, hi, dw, hash);
|
|
|
5786
|
|
307
|
5787 flags = 0;
|
|
483
|
5788 store_afflist[0] = NUL;
|
|
|
5789 pfxlen = 0;
|
|
|
5790 need_affix = FALSE;
|
|
307
|
5791 if (afflist != NULL)
|
|
|
5792 {
|
|
|
5793 /* Check for affix name that stands for keep-case word and stands
|
|
|
5794 * for rare word (if defined). */
|
|
621
|
5795 if (affile->af_keepcase != 0 && flag_in_afflist(
|
|
|
5796 affile->af_flagtype, afflist, affile->af_keepcase))
|
|
389
|
5797 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
621
|
5798 if (affile->af_rare != 0 && flag_in_afflist(
|
|
|
5799 affile->af_flagtype, afflist, affile->af_rare))
|
|
307
|
5800 flags |= WF_RARE;
|
|
495
|
5801 if (affile->af_bad != 0 && flag_in_afflist(
|
|
|
5802 affile->af_flagtype, afflist, affile->af_bad))
|
|
346
|
5803 flags |= WF_BANNED;
|
|
495
|
5804 if (affile->af_needaffix != 0 && flag_in_afflist(
|
|
|
5805 affile->af_flagtype, afflist, affile->af_needaffix))
|
|
483
|
5806 need_affix = TRUE;
|
|
500
|
5807 if (affile->af_needcomp != 0 && flag_in_afflist(
|
|
|
5808 affile->af_flagtype, afflist, affile->af_needcomp))
|
|
|
5809 flags |= WF_NEEDCOMP;
|
|
339
|
5810
|
|
|
5811 if (affile->af_pfxpostpone)
|
|
|
5812 /* Need to store the list of prefix IDs with the word. */
|
|
483
|
5813 pfxlen = get_pfxlist(affile, afflist, store_afflist);
|
|
|
5814
|
|
|
5815 if (spin->si_compflags != NULL)
|
|
|
5816 /* Need to store the list of compound flags with the word.
|
|
|
5817 * Concatenate them to the list of prefix IDs. */
|
|
495
|
5818 get_compflags(affile, afflist, store_afflist + pfxlen);
|
|
307
|
5819 }
|
|
|
5820
|
|
300
|
5821 /* Add the word to the word tree(s). */
|
|
483
|
5822 if (store_word(spin, dw, flags, spin->si_region,
|
|
|
5823 store_afflist, need_affix) == FAIL)
|
|
300
|
5824 retval = FAIL;
|
|
236
|
5825
|
|
300
|
5826 if (afflist != NULL)
|
|
|
5827 {
|
|
|
5828 /* Find all matching suffixes and add the resulting words.
|
|
|
5829 * Additionally do matching prefixes that combine. */
|
|
474
|
5830 if (store_aff_word(spin, dw, afflist, affile,
|
|
307
|
5831 &affile->af_suff, &affile->af_pref,
|
|
483
|
5832 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5833 retval = FAIL;
|
|
|
5834
|
|
|
5835 /* Find all matching prefixes and add the resulting words. */
|
|
474
|
5836 if (store_aff_word(spin, dw, afflist, affile,
|
|
339
|
5837 &affile->af_pref, NULL,
|
|
483
|
5838 FALSE, flags, store_afflist, pfxlen) == FAIL)
|
|
300
|
5839 retval = FAIL;
|
|
|
5840 }
|
|
236
|
5841 }
|
|
|
5842
|
|
481
|
5843 if (duplicate > 0)
|
|
|
5844 smsg((char_u *)_("%d duplicate word(s) in %s"), duplicate, fname);
|
|
300
|
5845 if (spin->si_ascii && non_ascii > 0)
|
|
481
|
5846 smsg((char_u *)_("Ignored %d word(s) with non-ASCII characters in %s"),
|
|
|
5847 non_ascii, fname);
|
|
300
|
5848 hash_clear(&ht);
|
|
|
5849
|
|
236
|
5850 fclose(fd);
|
|
300
|
5851 return retval;
|
|
236
|
5852 }
|
|
|
5853
|
|
|
5854 /*
|
|
339
|
5855 * Get the list of prefix IDs from the affix list "afflist".
|
|
|
5856 * Used for PFXPOSTPONE.
|
|
483
|
5857 * Put the resulting flags in "store_afflist[MAXWLEN]" with a terminating NUL
|
|
|
5858 * and return the number of affixes.
|
|
|
5859 */
|
|
|
5860 static int
|
|
|
5861 get_pfxlist(affile, afflist, store_afflist)
|
|
339
|
5862 afffile_T *affile;
|
|
|
5863 char_u *afflist;
|
|
483
|
5864 char_u *store_afflist;
|
|
339
|
5865 {
|
|
|
5866 char_u *p;
|
|
495
|
5867 char_u *prevp;
|
|
483
|
5868 int cnt = 0;
|
|
495
|
5869 int id;
|
|
|
5870 char_u key[AH_KEY_LEN];
|
|
339
|
5871 hashitem_T *hi;
|
|
|
5872
|
|
495
|
5873 for (p = afflist; *p != NUL; )
|
|
|
5874 {
|
|
|
5875 prevp = p;
|
|
|
5876 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
5877 {
|
|
|
5878 /* A flag is a postponed prefix flag if it appears in "af_pref"
|
|
|
5879 * and it's ID is not zero. */
|
|
|
5880 vim_strncpy(key, prevp, p - prevp);
|
|
|
5881 hi = hash_find(&affile->af_pref, key);
|
|
|
5882 if (!HASHITEM_EMPTY(hi))
|
|
|
5883 {
|
|
|
5884 id = HI2AH(hi)->ah_newID;
|
|
|
5885 if (id != 0)
|
|
|
5886 store_afflist[cnt++] = id;
|
|
|
5887 }
|
|
|
5888 }
|
|
498
|
5889 if (affile->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
5890 ++p;
|
|
483
|
5891 }
|
|
|
5892
|
|
|
5893 store_afflist[cnt] = NUL;
|
|
|
5894 return cnt;
|
|
339
|
5895 }
|
|
|
5896
|
|
|
5897 /*
|
|
495
|
5898 * Get the list of compound IDs from the affix list "afflist" that are used
|
|
|
5899 * for compound words.
|
|
483
|
5900 * Puts the flags in "store_afflist[]".
|
|
|
5901 */
|
|
|
5902 static void
|
|
495
|
5903 get_compflags(affile, afflist, store_afflist)
|
|
|
5904 afffile_T *affile;
|
|
481
|
5905 char_u *afflist;
|
|
483
|
5906 char_u *store_afflist;
|
|
481
|
5907 {
|
|
|
5908 char_u *p;
|
|
495
|
5909 char_u *prevp;
|
|
483
|
5910 int cnt = 0;
|
|
495
|
5911 char_u key[AH_KEY_LEN];
|
|
|
5912 hashitem_T *hi;
|
|
|
5913
|
|
|
5914 for (p = afflist; *p != NUL; )
|
|
|
5915 {
|
|
|
5916 prevp = p;
|
|
|
5917 if (get_affitem(affile->af_flagtype, &p) != 0)
|
|
|
5918 {
|
|
|
5919 /* A flag is a compound flag if it appears in "af_comp". */
|
|
|
5920 vim_strncpy(key, prevp, p - prevp);
|
|
|
5921 hi = hash_find(&affile->af_comp, key);
|
|
|
5922 if (!HASHITEM_EMPTY(hi))
|
|
|
5923 store_afflist[cnt++] = HI2CI(hi)->ci_newID;
|
|
|
5924 }
|
|
498
|
5925 if (affile->af_flagtype == AFT_NUM && *p == ',')
|
|
495
|
5926 ++p;
|
|
|
5927 }
|
|
483
|
5928
|
|
|
5929 store_afflist[cnt] = NUL;
|
|
481
|
5930 }
|
|
|
5931
|
|
|
5932 /*
|
|
300
|
5933 * Apply affixes to a word and store the resulting words.
|
|
|
5934 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
|
5935 * prefixes or suffixes.
|
|
|
5936 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
|
5937 * the resulting words for combining affixes.
|
|
|
5938 *
|
|
|
5939 * Returns FAIL when out of memory.
|
|
236
|
5940 */
|
|
300
|
5941 static int
|
|
483
|
5942 store_aff_word(spin, word, afflist, affile, ht, xht, comb, flags,
|
|
|
5943 pfxlist, pfxlen)
|
|
474
|
5944 spellinfo_T *spin; /* spell info */
|
|
300
|
5945 char_u *word; /* basic word start */
|
|
|
5946 char_u *afflist; /* list of names of supported affixes */
|
|
339
|
5947 afffile_T *affile;
|
|
300
|
5948 hashtab_T *ht;
|
|
|
5949 hashtab_T *xht;
|
|
|
5950 int comb; /* only use affixes that combine */
|
|
307
|
5951 int flags; /* flags for the word */
|
|
339
|
5952 char_u *pfxlist; /* list of prefix IDs */
|
|
483
|
5953 int pfxlen; /* nr of flags in "pfxlist" for prefixes, rest
|
|
|
5954 * is compound flags */
|
|
236
|
5955 {
|
|
|
5956 int todo;
|
|
|
5957 hashitem_T *hi;
|
|
300
|
5958 affheader_T *ah;
|
|
|
5959 affentry_T *ae;
|
|
|
5960 regmatch_T regmatch;
|
|
|
5961 char_u newword[MAXWLEN];
|
|
|
5962 int retval = OK;
|
|
|
5963 int i;
|
|
|
5964 char_u *p;
|
|
366
|
5965 int use_flags;
|
|
390
|
5966 char_u *use_pfxlist;
|
|
483
|
5967 char_u pfx_pfxlist[MAXWLEN];
|
|
|
5968 size_t wordlen = STRLEN(word);
|
|
236
|
5969
|
|
300
|
5970 todo = ht->ht_used;
|
|
|
5971 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
|
236
|
5972 {
|
|
|
5973 if (!HASHITEM_EMPTY(hi))
|
|
|
5974 {
|
|
|
5975 --todo;
|
|
300
|
5976 ah = HI2AH(hi);
|
|
236
|
5977
|
|
300
|
5978 /* Check that the affix combines, if required, and that the word
|
|
|
5979 * supports this affix. */
|
|
495
|
5980 if ((!comb || ah->ah_combine) && flag_in_afflist(
|
|
|
5981 affile->af_flagtype, afflist, ah->ah_flag))
|
|
236
|
5982 {
|
|
300
|
5983 /* Loop over all affix entries with this name. */
|
|
|
5984 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
236
|
5985 {
|
|
300
|
5986 /* Check the condition. It's not logical to match case
|
|
|
5987 * here, but it is required for compatibility with
|
|
339
|
5988 * Myspell.
|
|
481
|
5989 * Another requirement from Myspell is that the chop
|
|
|
5990 * string is shorter than the word itself.
|
|
339
|
5991 * For prefixes, when "PFXPOSTPONE" was used, only do
|
|
|
5992 * prefixes with a chop string. */
|
|
300
|
5993 regmatch.regprog = ae->ae_prog;
|
|
|
5994 regmatch.rm_ic = FALSE;
|
|
339
|
5995 if ((xht != NULL || !affile->af_pfxpostpone
|
|
|
5996 || ae->ae_chop != NULL)
|
|
481
|
5997 && (ae->ae_chop == NULL
|
|
|
5998 || STRLEN(ae->ae_chop) < wordlen)
|
|
339
|
5999 && (ae->ae_prog == NULL
|
|
|
6000 || vim_regexec(®match, word, (colnr_T)0)))
|
|
300
|
6001 {
|
|
|
6002 /* Match. Remove the chop and add the affix. */
|
|
|
6003 if (xht == NULL)
|
|
240
|
6004 {
|
|
300
|
6005 /* prefix: chop/add at the start of the word */
|
|
|
6006 if (ae->ae_add == NULL)
|
|
|
6007 *newword = NUL;
|
|
|
6008 else
|
|
|
6009 STRCPY(newword, ae->ae_add);
|
|
|
6010 p = word;
|
|
|
6011 if (ae->ae_chop != NULL)
|
|
310
|
6012 {
|
|
300
|
6013 /* Skip chop string. */
|
|
310
|
6014 #ifdef FEAT_MBYTE
|
|
|
6015 if (has_mbyte)
|
|
324
|
6016 {
|
|
310
|
6017 i = mb_charlen(ae->ae_chop);
|
|
324
|
6018 for ( ; i > 0; --i)
|
|
|
6019 mb_ptr_adv(p);
|
|
|
6020 }
|
|
310
|
6021 else
|
|
|
6022 #endif
|
|
324
|
6023 p += STRLEN(ae->ae_chop);
|
|
310
|
6024 }
|
|
300
|
6025 STRCAT(newword, p);
|
|
|
6026 }
|
|
|
6027 else
|
|
|
6028 {
|
|
|
6029 /* suffix: chop/add at the end of the word */
|
|
|
6030 STRCPY(newword, word);
|
|
|
6031 if (ae->ae_chop != NULL)
|
|
|
6032 {
|
|
|
6033 /* Remove chop string. */
|
|
|
6034 p = newword + STRLEN(newword);
|
|
376
|
6035 i = MB_CHARLEN(ae->ae_chop);
|
|
310
|
6036 for ( ; i > 0; --i)
|
|
300
|
6037 mb_ptr_back(newword, p);
|
|
|
6038 *p = NUL;
|
|
|
6039 }
|
|
|
6040 if (ae->ae_add != NULL)
|
|
|
6041 STRCAT(newword, ae->ae_add);
|
|
240
|
6042 }
|
|
|
6043
|
|
366
|
6044 /* Obey the "rare" flag of the affix. */
|
|
|
6045 if (ae->ae_rare)
|
|
|
6046 use_flags = flags | WF_RARE;
|
|
|
6047 else
|
|
|
6048 use_flags = flags;
|
|
483
|
6049
|
|
|
6050 /* Obey the "nocomp" flag of the affix: don't use the
|
|
|
6051 * compound flags. */
|
|
390
|
6052 use_pfxlist = pfxlist;
|
|
483
|
6053 if (ae->ae_nocomp && pfxlist != NULL)
|
|
|
6054 {
|
|
|
6055 vim_strncpy(pfx_pfxlist, pfxlist, pfxlen);
|
|
|
6056 use_pfxlist = pfx_pfxlist;
|
|
|
6057 }
|
|
390
|
6058
|
|
|
6059 /* When there are postponed prefixes... */
|
|
393
|
6060 if (spin->si_prefroot != NULL
|
|
|
6061 && spin->si_prefroot->wn_sibling != NULL)
|
|
390
|
6062 {
|
|
|
6063 /* ... add a flag to indicate an affix was used. */
|
|
|
6064 use_flags |= WF_HAS_AFF;
|
|
|
6065
|
|
|
6066 /* ... don't use a prefix list if combining
|
|
483
|
6067 * affixes is not allowed. But do use the
|
|
|
6068 * compound flags after them. */
|
|
|
6069 if ((!ah->ah_combine || comb) && pfxlist != NULL)
|
|
|
6070 use_pfxlist += pfxlen;
|
|
390
|
6071 }
|
|
366
|
6072
|
|
300
|
6073 /* Store the modified word. */
|
|
474
|
6074 if (store_word(spin, newword, use_flags,
|
|
483
|
6075 spin->si_region, use_pfxlist, FALSE) == FAIL)
|
|
300
|
6076 retval = FAIL;
|
|
236
|
6077
|
|
300
|
6078 /* When added a suffix and combining is allowed also
|
|
|
6079 * try adding prefixes additionally. */
|
|
|
6080 if (xht != NULL && ah->ah_combine)
|
|
474
|
6081 if (store_aff_word(spin, newword, afflist, affile,
|
|
390
|
6082 xht, NULL, TRUE,
|
|
483
|
6083 use_flags, use_pfxlist, pfxlen) == FAIL)
|
|
300
|
6084 retval = FAIL;
|
|
236
|
6085 }
|
|
|
6086 }
|
|
|
6087 }
|
|
|
6088 }
|
|
|
6089 }
|
|
|
6090
|
|
|
6091 return retval;
|
|
|
6092 }
|
|
|
6093
|
|
|
6094 /*
|
|
300
|
6095 * Read a file with a list of words.
|
|
236
|
6096 */
|
|
|
6097 static int
|
|
474
|
6098 spell_read_wordfile(spin, fname)
|
|
|
6099 spellinfo_T *spin;
|
|
300
|
6100 char_u *fname;
|
|
236
|
6101 {
|
|
300
|
6102 FILE *fd;
|
|
|
6103 long lnum = 0;
|
|
|
6104 char_u rline[MAXLINELEN];
|
|
|
6105 char_u *line;
|
|
|
6106 char_u *pc = NULL;
|
|
381
|
6107 char_u *p;
|
|
300
|
6108 int l;
|
|
|
6109 int retval = OK;
|
|
|
6110 int did_word = FALSE;
|
|
|
6111 int non_ascii = 0;
|
|
307
|
6112 int flags;
|
|
316
|
6113 int regionmask;
|
|
236
|
6114
|
|
300
|
6115 /*
|
|
|
6116 * Open the file.
|
|
|
6117 */
|
|
310
|
6118 fd = mch_fopen((char *)fname, "r");
|
|
300
|
6119 if (fd == NULL)
|
|
236
|
6120 {
|
|
300
|
6121 EMSG2(_(e_notopen), fname);
|
|
|
6122 return FAIL;
|
|
236
|
6123 }
|
|
|
6124
|
|
310
|
6125 if (spin->si_verbose || p_verbose > 2)
|
|
|
6126 {
|
|
|
6127 if (!spin->si_verbose)
|
|
|
6128 verbose_enter();
|
|
366
|
6129 smsg((char_u *)_("Reading word file %s ..."), fname);
|
|
310
|
6130 out_flush();
|
|
|
6131 if (!spin->si_verbose)
|
|
|
6132 verbose_leave();
|
|
|
6133 }
|
|
300
|
6134
|
|
|
6135 /*
|
|
|
6136 * Read all the lines in the file one by one.
|
|
|
6137 */
|
|
|
6138 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
|
6139 {
|
|
|
6140 line_breakcheck();
|
|
|
6141 ++lnum;
|
|
|
6142
|
|
|
6143 /* Skip comment lines. */
|
|
|
6144 if (*rline == '#')
|
|
|
6145 continue;
|
|
|
6146
|
|
|
6147 /* Remove CR, LF and white space from the end. */
|
|
|
6148 l = STRLEN(rline);
|
|
|
6149 while (l > 0 && rline[l - 1] <= ' ')
|
|
|
6150 --l;
|
|
|
6151 if (l == 0)
|
|
|
6152 continue; /* empty or blank line */
|
|
|
6153 rline[l] = NUL;
|
|
|
6154
|
|
|
6155 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
|
6156 vim_free(pc);
|
|
310
|
6157 #ifdef FEAT_MBYTE
|
|
300
|
6158 if (spin->si_conv.vc_type != CONV_NONE)
|
|
|
6159 {
|
|
|
6160 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
|
6161 if (pc == NULL)
|
|
|
6162 {
|
|
|
6163 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
|
6164 fname, lnum, rline);
|
|
|
6165 continue;
|
|
|
6166 }
|
|
|
6167 line = pc;
|
|
|
6168 }
|
|
|
6169 else
|
|
310
|
6170 #endif
|
|
300
|
6171 {
|
|
|
6172 pc = NULL;
|
|
|
6173 line = rline;
|
|
|
6174 }
|
|
|
6175
|
|
307
|
6176 if (*line == '/')
|
|
300
|
6177 {
|
|
307
|
6178 ++line;
|
|
|
6179 if (STRNCMP(line, "encoding=", 9) == 0)
|
|
300
|
6180 {
|
|
|
6181 if (spin->si_conv.vc_type != CONV_NONE)
|
|
316
|
6182 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
|
6183 fname, lnum, line - 1);
|
|
300
|
6184 else if (did_word)
|
|
316
|
6185 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
|
6186 fname, lnum, line - 1);
|
|
300
|
6187 else
|
|
|
6188 {
|
|
310
|
6189 #ifdef FEAT_MBYTE
|
|
|
6190 char_u *enc;
|
|
|
6191
|
|
300
|
6192 /* Setup for conversion to 'encoding'. */
|
|
316
|
6193 line += 10;
|
|
|
6194 enc = enc_canonize(line);
|
|
300
|
6195 if (enc != NULL && !spin->si_ascii
|
|
|
6196 && convert_setup(&spin->si_conv, enc,
|
|
|
6197 p_enc) == FAIL)
|
|
|
6198 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
316
|
6199 fname, line, p_enc);
|
|
300
|
6200 vim_free(enc);
|
|
375
|
6201 spin->si_conv.vc_fail = TRUE;
|
|
310
|
6202 #else
|
|
|
6203 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
|
6204 #endif
|
|
300
|
6205 }
|
|
307
|
6206 continue;
|
|
300
|
6207 }
|
|
307
|
6208
|
|
316
|
6209 if (STRNCMP(line, "regions=", 8) == 0)
|
|
|
6210 {
|
|
|
6211 if (spin->si_region_count > 1)
|
|
|
6212 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
|
6213 fname, lnum, line);
|
|
|
6214 else
|
|
|
6215 {
|
|
|
6216 line += 8;
|
|
|
6217 if (STRLEN(line) > 16)
|
|
|
6218 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
|
6219 fname, lnum, line);
|
|
|
6220 else
|
|
|
6221 {
|
|
|
6222 spin->si_region_count = STRLEN(line) / 2;
|
|
|
6223 STRCPY(spin->si_region_name, line);
|
|
389
|
6224
|
|
|
6225 /* Adjust the mask for a word valid in all regions. */
|
|
|
6226 spin->si_region = (1 << spin->si_region_count) - 1;
|
|
316
|
6227 }
|
|
|
6228 }
|
|
|
6229 continue;
|
|
|
6230 }
|
|
|
6231
|
|
381
|
6232 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
|
|
6233 fname, lnum, line - 1);
|
|
|
6234 continue;
|
|
|
6235 }
|
|
|
6236
|
|
|
6237 flags = 0;
|
|
|
6238 regionmask = spin->si_region;
|
|
|
6239
|
|
|
6240 /* Check for flags and region after a slash. */
|
|
|
6241 p = vim_strchr(line, '/');
|
|
|
6242 if (p != NULL)
|
|
|
6243 {
|
|
|
6244 *p++ = NUL;
|
|
|
6245 while (*p != NUL)
|
|
|
6246 {
|
|
|
6247 if (*p == '=') /* keep-case word */
|
|
389
|
6248 flags |= WF_KEEPCAP | WF_FIXCAP;
|
|
381
|
6249 else if (*p == '!') /* Bad, bad, wicked word. */
|
|
|
6250 flags |= WF_BANNED;
|
|
|
6251 else if (*p == '?') /* Rare word. */
|
|
|
6252 flags |= WF_RARE;
|
|
|
6253 else if (VIM_ISDIGIT(*p)) /* region number(s) */
|
|
316
|
6254 {
|
|
381
|
6255 if ((flags & WF_REGION) == 0) /* first one */
|
|
|
6256 regionmask = 0;
|
|
|
6257 flags |= WF_REGION;
|
|
|
6258
|
|
|
6259 l = *p - '0';
|
|
316
|
6260 if (l > spin->si_region_count)
|
|
|
6261 {
|
|
|
6262 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
381
|
6263 fname, lnum, p);
|
|
316
|
6264 break;
|
|
|
6265 }
|
|
|
6266 regionmask |= 1 << (l - 1);
|
|
|
6267 }
|
|
381
|
6268 else
|
|
|
6269 {
|
|
|
6270 smsg((char_u *)_("Unrecognized flags in %s line %d: %s"),
|
|
|
6271 fname, lnum, p);
|
|
|
6272 break;
|
|
|
6273 }
|
|
|
6274 ++p;
|
|
307
|
6275 }
|
|
300
|
6276 }
|
|
|
6277
|
|
|
6278 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
|
6279 if (spin->si_ascii && has_non_ascii(line))
|
|
|
6280 {
|
|
|
6281 ++non_ascii;
|
|
|
6282 continue;
|
|
|
6283 }
|
|
|
6284
|
|
|
6285 /* Normal word: store it. */
|
|
483
|
6286 if (store_word(spin, line, flags, regionmask, NULL, FALSE) == FAIL)
|
|
300
|
6287 {
|
|
|
6288 retval = FAIL;
|
|
|
6289 break;
|
|
|
6290 }
|
|
|
6291 did_word = TRUE;
|
|
|
6292 }
|
|
|
6293
|
|
|
6294 vim_free(pc);
|
|
|
6295 fclose(fd);
|
|
|
6296
|
|
310
|
6297 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
|
6298 {
|
|
|
6299 if (p_verbose > 2)
|
|
|
6300 verbose_enter();
|
|
300
|
6301 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
|
6302 non_ascii);
|
|
310
|
6303 if (p_verbose > 2)
|
|
|
6304 verbose_leave();
|
|
|
6305 }
|
|
300
|
6306 return retval;
|
|
236
|
6307 }
|
|
|
6308
|
|
|
6309 /*
|
|
300
|
6310 * Get part of an sblock_T, "len" bytes long.
|
|
474
|
6311 * This avoids calling free() for every little struct we use (and keeping
|
|
|
6312 * track of them).
|
|
300
|
6313 * The memory is cleared to all zeros.
|
|
|
6314 * Returns NULL when out of memory.
|
|
|
6315 */
|
|
|
6316 static void *
|
|
474
|
6317 getroom(spin, len, align)
|
|
|
6318 spellinfo_T *spin;
|
|
447
|
6319 size_t len; /* length needed */
|
|
|
6320 int align; /* align for pointer */
|
|
300
|
6321 {
|
|
|
6322 char_u *p;
|
|
474
|
6323 sblock_T *bl = spin->si_blocks;
|
|
300
|
6324
|
|
447
|
6325 if (align && bl != NULL)
|
|
|
6326 /* Round size up for alignment. On some systems structures need to be
|
|
|
6327 * aligned to the size of a pointer (e.g., SPARC). */
|
|
|
6328 bl->sb_used = (bl->sb_used + sizeof(char *) - 1)
|
|
|
6329 & ~(sizeof(char *) - 1);
|
|
|
6330
|
|
300
|
6331 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
|
6332 {
|
|
|
6333 /* Allocate a block of memory. This is not freed until much later. */
|
|
|
6334 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
|
6335 if (bl == NULL)
|
|
|
6336 return NULL;
|
|
474
|
6337 bl->sb_next = spin->si_blocks;
|
|
|
6338 spin->si_blocks = bl;
|
|
300
|
6339 bl->sb_used = 0;
|
|
482
|
6340 ++spin->si_blocks_cnt;
|
|
300
|
6341 }
|
|
|
6342
|
|
|
6343 p = bl->sb_data + bl->sb_used;
|
|
|
6344 bl->sb_used += len;
|
|
|
6345
|
|
|
6346 return p;
|
|
|
6347 }
|
|
|
6348
|
|
|
6349 /*
|
|
|
6350 * Make a copy of a string into memory allocated with getroom().
|
|
|
6351 */
|
|
|
6352 static char_u *
|
|
474
|
6353 getroom_save(spin, s)
|
|
|
6354 spellinfo_T *spin;
|
|
300
|
6355 char_u *s;
|
|
|
6356 {
|
|
|
6357 char_u *sc;
|
|
|
6358
|
|
474
|
6359 sc = (char_u *)getroom(spin, STRLEN(s) + 1, FALSE);
|
|
300
|
6360 if (sc != NULL)
|
|
|
6361 STRCPY(sc, s);
|
|
|
6362 return sc;
|
|
|
6363 }
|
|
|
6364
|
|
|
6365
|
|
|
6366 /*
|
|
|
6367 * Free the list of allocated sblock_T.
|
|
236
|
6368 */
|
|
|
6369 static void
|
|
300
|
6370 free_blocks(bl)
|
|
|
6371 sblock_T *bl;
|
|
236
|
6372 {
|
|
300
|
6373 sblock_T *next;
|
|
236
|
6374
|
|
300
|
6375 while (bl != NULL)
|
|
236
|
6376 {
|
|
300
|
6377 next = bl->sb_next;
|
|
|
6378 vim_free(bl);
|
|
|
6379 bl = next;
|
|
236
|
6380 }
|
|
|
6381 }
|
|
|
6382
|
|
|
6383 /*
|
|
300
|
6384 * Allocate the root of a word tree.
|
|
236
|
6385 */
|
|
300
|
6386 static wordnode_T *
|
|
474
|
6387 wordtree_alloc(spin)
|
|
|
6388 spellinfo_T *spin;
|
|
|
6389 {
|
|
|
6390 return (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
236
|
6391 }
|
|
|
6392
|
|
|
6393 /*
|
|
300
|
6394 * Store a word in the tree(s).
|
|
390
|
6395 * Always store it in the case-folded tree. For a keep-case word this is
|
|
|
6396 * useful when the word can also be used with all caps (no WF_FIXCAP flag) and
|
|
|
6397 * used to find suggestions.
|
|
300
|
6398 * For a keep-case word also store it in the keep-case tree.
|
|
482
|
6399 * When "pfxlist" is not NULL store the word for each postponed prefix ID and
|
|
|
6400 * compound flag.
|
|
236
|
6401 */
|
|
|
6402 static int
|
|
483
|
6403 store_word(spin, word, flags, region, pfxlist, need_affix)
|
|
474
|
6404 spellinfo_T *spin;
|
|
300
|
6405 char_u *word;
|
|
307
|
6406 int flags; /* extra flags, WF_BANNED */
|
|
316
|
6407 int region; /* supported region(s) */
|
|
339
|
6408 char_u *pfxlist; /* list of prefix IDs or NULL */
|
|
483
|
6409 int need_affix; /* only store word with affix ID */
|
|
236
|
6410 {
|
|
300
|
6411 int len = STRLEN(word);
|
|
|
6412 int ct = captype(word, word + len);
|
|
|
6413 char_u foldword[MAXWLEN];
|
|
339
|
6414 int res = OK;
|
|
|
6415 char_u *p;
|
|
236
|
6416
|
|
323
|
6417 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
339
|
6418 for (p = pfxlist; res == OK; ++p)
|
|
|
6419 {
|
|
483
|
6420 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6421 res = tree_add_word(spin, foldword, spin->si_foldroot, ct | flags,
|
|
474
|
6422 region, p == NULL ? 0 : *p);
|
|
339
|
6423 if (p == NULL || *p == NUL)
|
|
|
6424 break;
|
|
|
6425 }
|
|
334
|
6426 ++spin->si_foldwcount;
|
|
307
|
6427
|
|
474
|
6428 if (res == OK && (ct == WF_KEEPCAP || (flags & WF_KEEPCAP)))
|
|
334
|
6429 {
|
|
339
|
6430 for (p = pfxlist; res == OK; ++p)
|
|
|
6431 {
|
|
483
|
6432 if (!need_affix || (p != NULL && *p != NUL))
|
|
|
6433 res = tree_add_word(spin, word, spin->si_keeproot, flags,
|
|
474
|
6434 region, p == NULL ? 0 : *p);
|
|
339
|
6435 if (p == NULL || *p == NUL)
|
|
|
6436 break;
|
|
|
6437 }
|
|
334
|
6438 ++spin->si_keepwcount;
|
|
|
6439 }
|
|
300
|
6440 return res;
|
|
236
|
6441 }
|
|
|
6442
|
|
|
6443 /*
|
|
300
|
6444 * Add word "word" to a word tree at "root".
|
|
366
|
6445 * When "flags" < 0 we are adding to the prefix tree where flags is used for
|
|
|
6446 * "rare" and "region" is the condition nr.
|
|
255
|
6447 * Returns FAIL when out of memory.
|
|
236
|
6448 */
|
|
255
|
6449 static int
|
|
481
|
6450 tree_add_word(spin, word, root, flags, region, affixID)
|
|
474
|
6451 spellinfo_T *spin;
|
|
300
|
6452 char_u *word;
|
|
|
6453 wordnode_T *root;
|
|
|
6454 int flags;
|
|
|
6455 int region;
|
|
481
|
6456 int affixID;
|
|
236
|
6457 {
|
|
300
|
6458 wordnode_T *node = root;
|
|
|
6459 wordnode_T *np;
|
|
469
|
6460 wordnode_T *copyp, **copyprev;
|
|
300
|
6461 wordnode_T **prev = NULL;
|
|
|
6462 int i;
|
|
255
|
6463
|
|
300
|
6464 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
|
6465 for (i = 0; ; ++i)
|
|
255
|
6466 {
|
|
469
|
6467 /* When there is more than one reference to this node we need to make
|
|
|
6468 * a copy, so that we can modify it. Copy the whole list of siblings
|
|
|
6469 * (we don't optimize for a partly shared list of siblings). */
|
|
|
6470 if (node != NULL && node->wn_refs > 1)
|
|
|
6471 {
|
|
|
6472 --node->wn_refs;
|
|
|
6473 copyprev = prev;
|
|
|
6474 for (copyp = node; copyp != NULL; copyp = copyp->wn_sibling)
|
|
|
6475 {
|
|
|
6476 /* Allocate a new node and copy the info. */
|
|
474
|
6477 np = get_wordnode(spin);
|
|
469
|
6478 if (np == NULL)
|
|
|
6479 return FAIL;
|
|
|
6480 np->wn_child = copyp->wn_child;
|
|
|
6481 if (np->wn_child != NULL)
|
|
|
6482 ++np->wn_child->wn_refs; /* child gets extra ref */
|
|
|
6483 np->wn_byte = copyp->wn_byte;
|
|
|
6484 if (np->wn_byte == NUL)
|
|
|
6485 {
|
|
|
6486 np->wn_flags = copyp->wn_flags;
|
|
|
6487 np->wn_region = copyp->wn_region;
|
|
481
|
6488 np->wn_affixID = copyp->wn_affixID;
|
|
469
|
6489 }
|
|
|
6490
|
|
|
6491 /* Link the new node in the list, there will be one ref. */
|
|
|
6492 np->wn_refs = 1;
|
|
|
6493 *copyprev = np;
|
|
|
6494 copyprev = &np->wn_sibling;
|
|
|
6495
|
|
|
6496 /* Let "node" point to the head of the copied list. */
|
|
|
6497 if (copyp == node)
|
|
|
6498 node = np;
|
|
|
6499 }
|
|
|
6500 }
|
|
|
6501
|
|
300
|
6502 /* Look for the sibling that has the same character. They are sorted
|
|
|
6503 * on byte value, thus stop searching when a sibling is found with a
|
|
339
|
6504 * higher byte value. For zero bytes (end of word) the sorting is
|
|
481
|
6505 * done on flags and then on affixID. */
|
|
339
|
6506 while (node != NULL
|
|
|
6507 && (node->wn_byte < word[i]
|
|
|
6508 || (node->wn_byte == NUL
|
|
|
6509 && (flags < 0
|
|
481
|
6510 ? node->wn_affixID < affixID
|
|
390
|
6511 : node->wn_flags < (flags & WN_MASK)
|
|
|
6512 || (node->wn_flags == (flags & WN_MASK)
|
|
481
|
6513 && node->wn_affixID < affixID)))))
|
|
236
|
6514 {
|
|
300
|
6515 prev = &node->wn_sibling;
|
|
|
6516 node = *prev;
|
|
236
|
6517 }
|
|
339
|
6518 if (node == NULL
|
|
|
6519 || node->wn_byte != word[i]
|
|
|
6520 || (word[i] == NUL
|
|
|
6521 && (flags < 0
|
|
390
|
6522 || node->wn_flags != (flags & WN_MASK)
|
|
481
|
6523 || node->wn_affixID != affixID)))
|
|
255
|
6524 {
|
|
300
|
6525 /* Allocate a new node. */
|
|
474
|
6526 np = get_wordnode(spin);
|
|
300
|
6527 if (np == NULL)
|
|
|
6528 return FAIL;
|
|
|
6529 np->wn_byte = word[i];
|
|
469
|
6530
|
|
|
6531 /* If "node" is NULL this is a new child or the end of the sibling
|
|
|
6532 * list: ref count is one. Otherwise use ref count of sibling and
|
|
|
6533 * make ref count of sibling one (matters when inserting in front
|
|
|
6534 * of the list of siblings). */
|
|
|
6535 if (node == NULL)
|
|
|
6536 np->wn_refs = 1;
|
|
|
6537 else
|
|
|
6538 {
|
|
|
6539 np->wn_refs = node->wn_refs;
|
|
|
6540 node->wn_refs = 1;
|
|
|
6541 }
|
|
300
|
6542 *prev = np;
|
|
|
6543 np->wn_sibling = node;
|
|
|
6544 node = np;
|
|
255
|
6545 }
|
|
300
|
6546
|
|
|
6547 if (word[i] == NUL)
|
|
|
6548 {
|
|
|
6549 node->wn_flags = flags;
|
|
|
6550 node->wn_region |= region;
|
|
481
|
6551 node->wn_affixID = affixID;
|
|
300
|
6552 break;
|
|
|
6553 }
|
|
|
6554 prev = &node->wn_child;
|
|
|
6555 node = *prev;
|
|
255
|
6556 }
|
|
469
|
6557 #ifdef SPELL_PRINTTREE
|
|
|
6558 smsg("Added \"%s\"", word);
|
|
|
6559 spell_print_tree(root->wn_sibling);
|
|
|
6560 #endif
|
|
|
6561
|
|
474
|
6562 /* count nr of words added since last message */
|
|
|
6563 ++spin->si_msg_count;
|
|
|
6564
|
|
482
|
6565 if (spin->si_compress_cnt > 1)
|
|
|
6566 {
|
|
|
6567 if (--spin->si_compress_cnt == 1)
|
|
|
6568 /* Did enough words to lower the block count limit. */
|
|
483
|
6569 spin->si_blocks_cnt += compress_inc;
|
|
482
|
6570 }
|
|
|
6571
|
|
469
|
6572 /*
|
|
482
|
6573 * When we have allocated lots of memory we need to compress the word tree
|
|
|
6574 * to free up some room. But compression is slow, and we might actually
|
|
|
6575 * need that room, thus only compress in the following situations:
|
|
|
6576 * 1. When not compressed before (si_compress_cnt == 0): when using
|
|
483
|
6577 * "compress_start" blocks.
|
|
|
6578 * 2. When compressed before and used "compress_inc" blocks before
|
|
|
6579 * adding "compress_added" words (si_compress_cnt > 1).
|
|
|
6580 * 3. When compressed before, added "compress_added" words
|
|
482
|
6581 * (si_compress_cnt == 1) and the number of free nodes drops below the
|
|
|
6582 * maximum word length.
|
|
469
|
6583 */
|
|
482
|
6584 #ifndef SPELL_PRINTTREE
|
|
|
6585 if (spin->si_compress_cnt == 1
|
|
|
6586 ? spin->si_free_count < MAXWLEN
|
|
483
|
6587 : spin->si_blocks_cnt >= compress_start)
|
|
482
|
6588 #endif
|
|
|
6589 {
|
|
|
6590 /* Decrement the block counter. The effect is that we compress again
|
|
483
|
6591 * when the freed up room has been used and another "compress_inc"
|
|
|
6592 * blocks have been allocated. Unless "compress_added" words have
|
|
482
|
6593 * been added, then the limit is put back again. */
|
|
483
|
6594 spin->si_blocks_cnt -= compress_inc;
|
|
|
6595 spin->si_compress_cnt = compress_added;
|
|
482
|
6596
|
|
474
|
6597 if (spin->si_verbose)
|
|
|
6598 {
|
|
|
6599 msg_start();
|
|
|
6600 msg_puts((char_u *)_(msg_compressing));
|
|
|
6601 msg_clr_eos();
|
|
|
6602 msg_didout = FALSE;
|
|
|
6603 msg_col = 0;
|
|
|
6604 out_flush();
|
|
|
6605 }
|
|
482
|
6606
|
|
|
6607 /* Compress both trees. Either they both have many nodes, which makes
|
|
|
6608 * compression useful, or one of them is small, which means
|
|
|
6609 * compression goes fast. */
|
|
|
6610 wordtree_compress(spin, spin->si_foldroot);
|
|
|
6611 wordtree_compress(spin, spin->si_keeproot);
|
|
469
|
6612 }
|
|
255
|
6613
|
|
|
6614 return OK;
|
|
236
|
6615 }
|
|
|
6616
|
|
469
|
6617 /*
|
|
483
|
6618 * Check the 'mkspellmem' option. Return FAIL if it's wrong.
|
|
|
6619 * Sets "sps_flags".
|
|
|
6620 */
|
|
|
6621 int
|
|
|
6622 spell_check_msm()
|
|
|
6623 {
|
|
|
6624 char_u *p = p_msm;
|
|
|
6625 long start = 0;
|
|
|
6626 long inc = 0;
|
|
|
6627 long added = 0;
|
|
|
6628
|
|
|
6629 if (!VIM_ISDIGIT(*p))
|
|
|
6630 return FAIL;
|
|
|
6631 /* block count = (value * 1024) / SBLOCKSIZE (but avoid overflow)*/
|
|
|
6632 start = (getdigits(&p) * 10) / (SBLOCKSIZE / 102);
|
|
|
6633 if (*p != ',')
|
|
|
6634 return FAIL;
|
|
|
6635 ++p;
|
|
|
6636 if (!VIM_ISDIGIT(*p))
|
|
|
6637 return FAIL;
|
|
|
6638 inc = (getdigits(&p) * 102) / (SBLOCKSIZE / 10);
|
|
|
6639 if (*p != ',')
|
|
|
6640 return FAIL;
|
|
|
6641 ++p;
|
|
|
6642 if (!VIM_ISDIGIT(*p))
|
|
|
6643 return FAIL;
|
|
|
6644 added = getdigits(&p) * 1024;
|
|
|
6645 if (*p != NUL)
|
|
|
6646 return FAIL;
|
|
|
6647
|
|
|
6648 if (start == 0 || inc == 0 || added == 0 || inc > start)
|
|
|
6649 return FAIL;
|
|
|
6650
|
|
|
6651 compress_start = start;
|
|
|
6652 compress_inc = inc;
|
|
|
6653 compress_added = added;
|
|
|
6654 return OK;
|
|
|
6655 }
|
|
|
6656
|
|
|
6657
|
|
|
6658 /*
|
|
469
|
6659 * Get a wordnode_T, either from the list of previously freed nodes or
|
|
|
6660 * allocate a new one.
|
|
|
6661 */
|
|
|
6662 static wordnode_T *
|
|
474
|
6663 get_wordnode(spin)
|
|
|
6664 spellinfo_T *spin;
|
|
469
|
6665 {
|
|
|
6666 wordnode_T *n;
|
|
|
6667
|
|
474
|
6668 if (spin->si_first_free == NULL)
|
|
|
6669 n = (wordnode_T *)getroom(spin, sizeof(wordnode_T), TRUE);
|
|
469
|
6670 else
|
|
|
6671 {
|
|
474
|
6672 n = spin->si_first_free;
|
|
|
6673 spin->si_first_free = n->wn_child;
|
|
469
|
6674 vim_memset(n, 0, sizeof(wordnode_T));
|
|
482
|
6675 --spin->si_free_count;
|
|
469
|
6676 }
|
|
|
6677 #ifdef SPELL_PRINTTREE
|
|
474
|
6678 n->wn_nr = ++spin->si_wordnode_nr;
|
|
469
|
6679 #endif
|
|
|
6680 return n;
|
|
|
6681 }
|
|
|
6682
|
|
|
6683 /*
|
|
474
|
6684 * Decrement the reference count on a node (which is the head of a list of
|
|
|
6685 * siblings). If the reference count becomes zero free the node and its
|
|
|
6686 * siblings.
|
|
|
6687 */
|
|
|
6688 static void
|
|
|
6689 deref_wordnode(spin, node)
|
|
|
6690 spellinfo_T *spin;
|
|
|
6691 wordnode_T *node;
|
|
|
6692 {
|
|
|
6693 wordnode_T *np;
|
|
|
6694
|
|
|
6695 if (--node->wn_refs == 0)
|
|
|
6696 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
6697 {
|
|
|
6698 if (np->wn_child != NULL)
|
|
|
6699 deref_wordnode(spin, np->wn_child);
|
|
|
6700 free_wordnode(spin, np);
|
|
|
6701 }
|
|
|
6702 }
|
|
|
6703
|
|
|
6704 /*
|
|
469
|
6705 * Free a wordnode_T for re-use later.
|
|
474
|
6706 * Only the "wn_child" field becomes invalid.
|
|
469
|
6707 */
|
|
|
6708 static void
|
|
474
|
6709 free_wordnode(spin, n)
|
|
|
6710 spellinfo_T *spin;
|
|
|
6711 wordnode_T *n;
|
|
|
6712 {
|
|
|
6713 n->wn_child = spin->si_first_free;
|
|
|
6714 spin->si_first_free = n;
|
|
482
|
6715 ++spin->si_free_count;
|
|
469
|
6716 }
|
|
|
6717
|
|
236
|
6718 /*
|
|
300
|
6719 * Compress a tree: find tails that are identical and can be shared.
|
|
|
6720 */
|
|
|
6721 static void
|
|
474
|
6722 wordtree_compress(spin, root)
|
|
|
6723 spellinfo_T *spin;
|
|
300
|
6724 wordnode_T *root;
|
|
|
6725 {
|
|
|
6726 hashtab_T ht;
|
|
|
6727 int n;
|
|
|
6728 int tot = 0;
|
|
469
|
6729 int perc;
|
|
300
|
6730
|
|
474
|
6731 /* Skip the root itself, it's not actually used. The first sibling is the
|
|
|
6732 * start of the tree. */
|
|
|
6733 if (root->wn_sibling != NULL)
|
|
300
|
6734 {
|
|
|
6735 hash_init(&ht);
|
|
474
|
6736 n = node_compress(spin, root->wn_sibling, &ht, &tot);
|
|
469
|
6737
|
|
|
6738 #ifndef SPELL_PRINTTREE
|
|
310
|
6739 if (spin->si_verbose || p_verbose > 2)
|
|
469
|
6740 #endif
|
|
310
|
6741 {
|
|
|
6742 if (!spin->si_verbose)
|
|
|
6743 verbose_enter();
|
|
469
|
6744 if (tot > 1000000)
|
|
|
6745 perc = (tot - n) / (tot / 100);
|
|
482
|
6746 else if (tot == 0)
|
|
|
6747 perc = 0;
|
|
469
|
6748 else
|
|
|
6749 perc = (tot - n) * 100 / tot;
|
|
310
|
6750 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
|
469
|
6751 n, tot, perc);
|
|
310
|
6752 if (p_verbose > 2)
|
|
|
6753 verbose_leave();
|
|
|
6754 }
|
|
469
|
6755 #ifdef SPELL_PRINTTREE
|
|
474
|
6756 spell_print_tree(root->wn_sibling);
|
|
469
|
6757 #endif
|
|
300
|
6758 hash_clear(&ht);
|
|
|
6759 }
|
|
|
6760 }
|
|
|
6761
|
|
|
6762 /*
|
|
|
6763 * Compress a node, its siblings and its children, depth first.
|
|
|
6764 * Returns the number of compressed nodes.
|
|
236
|
6765 */
|
|
255
|
6766 static int
|
|
474
|
6767 node_compress(spin, node, ht, tot)
|
|
|
6768 spellinfo_T *spin;
|
|
300
|
6769 wordnode_T *node;
|
|
|
6770 hashtab_T *ht;
|
|
|
6771 int *tot; /* total count of nodes before compressing,
|
|
|
6772 incremented while going through the tree */
|
|
236
|
6773 {
|
|
300
|
6774 wordnode_T *np;
|
|
|
6775 wordnode_T *tp;
|
|
|
6776 wordnode_T *child;
|
|
|
6777 hash_T hash;
|
|
236
|
6778 hashitem_T *hi;
|
|
300
|
6779 int len = 0;
|
|
|
6780 unsigned nr, n;
|
|
|
6781 int compressed = 0;
|
|
236
|
6782
|
|
300
|
6783 /*
|
|
|
6784 * Go through the list of siblings. Compress each child and then try
|
|
|
6785 * finding an identical child to replace it.
|
|
|
6786 * Note that with "child" we mean not just the node that is pointed to,
|
|
|
6787 * but the whole list of siblings, of which the node is the first.
|
|
|
6788 */
|
|
482
|
6789 for (np = node; np != NULL && !got_int; np = np->wn_sibling)
|
|
236
|
6790 {
|
|
300
|
6791 ++len;
|
|
|
6792 if ((child = np->wn_child) != NULL)
|
|
|
6793 {
|
|
346
|
6794 /* Compress the child. This fills hashkey. */
|
|
474
|
6795 compressed += node_compress(spin, child, ht, tot);
|
|
300
|
6796
|
|
|
6797 /* Try to find an identical child. */
|
|
346
|
6798 hash = hash_hash(child->wn_u1.hashkey);
|
|
|
6799 hi = hash_lookup(ht, child->wn_u1.hashkey, hash);
|
|
300
|
6800 tp = NULL;
|
|
|
6801 if (!HASHITEM_EMPTY(hi))
|
|
|
6802 {
|
|
|
6803 /* There are children with an identical hash value. Now check
|
|
|
6804 * if there is one that is really identical. */
|
|
346
|
6805 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_u2.next)
|
|
300
|
6806 if (node_equal(child, tp))
|
|
|
6807 {
|
|
|
6808 /* Found one! Now use that child in place of the
|
|
469
|
6809 * current one. This means the current child and all
|
|
|
6810 * its siblings is unlinked from the tree. */
|
|
|
6811 ++tp->wn_refs;
|
|
474
|
6812 deref_wordnode(spin, child);
|
|
300
|
6813 np->wn_child = tp;
|
|
|
6814 ++compressed;
|
|
|
6815 break;
|
|
|
6816 }
|
|
|
6817 if (tp == NULL)
|
|
|
6818 {
|
|
|
6819 /* No other child with this hash value equals the child of
|
|
|
6820 * the node, add it to the linked list after the first
|
|
|
6821 * item. */
|
|
|
6822 tp = HI2WN(hi);
|
|
346
|
6823 child->wn_u2.next = tp->wn_u2.next;
|
|
|
6824 tp->wn_u2.next = child;
|
|
300
|
6825 }
|
|
|
6826 }
|
|
|
6827 else
|
|
|
6828 /* No other child has this hash value, add it to the
|
|
|
6829 * hashtable. */
|
|
346
|
6830 hash_add_item(ht, hi, child->wn_u1.hashkey, hash);
|
|
300
|
6831 }
|
|
236
|
6832 }
|
|
300
|
6833 *tot += len;
|
|
|
6834
|
|
|
6835 /*
|
|
|
6836 * Make a hash key for the node and its siblings, so that we can quickly
|
|
|
6837 * find a lookalike node. This must be done after compressing the sibling
|
|
|
6838 * list, otherwise the hash key would become invalid by the compression.
|
|
|
6839 */
|
|
346
|
6840 node->wn_u1.hashkey[0] = len;
|
|
300
|
6841 nr = 0;
|
|
|
6842 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
6843 {
|
|
300
|
6844 if (np->wn_byte == NUL)
|
|
481
|
6845 /* end node: use wn_flags, wn_region and wn_affixID */
|
|
|
6846 n = np->wn_flags + (np->wn_region << 8) + (np->wn_affixID << 16);
|
|
300
|
6847 else
|
|
|
6848 /* byte node: use the byte value and the child pointer */
|
|
|
6849 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
|
6850 nr = nr * 101 + n;
|
|
236
|
6851 }
|
|
300
|
6852
|
|
|
6853 /* Avoid NUL bytes, it terminates the hash key. */
|
|
|
6854 n = nr & 0xff;
|
|
346
|
6855 node->wn_u1.hashkey[1] = n == 0 ? 1 : n;
|
|
300
|
6856 n = (nr >> 8) & 0xff;
|
|
346
|
6857 node->wn_u1.hashkey[2] = n == 0 ? 1 : n;
|
|
300
|
6858 n = (nr >> 16) & 0xff;
|
|
346
|
6859 node->wn_u1.hashkey[3] = n == 0 ? 1 : n;
|
|
300
|
6860 n = (nr >> 24) & 0xff;
|
|
346
|
6861 node->wn_u1.hashkey[4] = n == 0 ? 1 : n;
|
|
|
6862 node->wn_u1.hashkey[5] = NUL;
|
|
300
|
6863
|
|
482
|
6864 /* Check for CTRL-C pressed now and then. */
|
|
|
6865 fast_breakcheck();
|
|
|
6866
|
|
300
|
6867 return compressed;
|
|
|
6868 }
|
|
|
6869
|
|
|
6870 /*
|
|
|
6871 * Return TRUE when two nodes have identical siblings and children.
|
|
|
6872 */
|
|
|
6873 static int
|
|
|
6874 node_equal(n1, n2)
|
|
|
6875 wordnode_T *n1;
|
|
|
6876 wordnode_T *n2;
|
|
|
6877 {
|
|
|
6878 wordnode_T *p1;
|
|
|
6879 wordnode_T *p2;
|
|
|
6880
|
|
|
6881 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
|
6882 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
|
6883 if (p1->wn_byte != p2->wn_byte
|
|
|
6884 || (p1->wn_byte == NUL
|
|
|
6885 ? (p1->wn_flags != p2->wn_flags
|
|
339
|
6886 || p1->wn_region != p2->wn_region
|
|
481
|
6887 || p1->wn_affixID != p2->wn_affixID)
|
|
300
|
6888 : (p1->wn_child != p2->wn_child)))
|
|
|
6889 break;
|
|
|
6890
|
|
|
6891 return p1 == NULL && p2 == NULL;
|
|
236
|
6892 }
|
|
|
6893
|
|
|
6894 /*
|
|
|
6895 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
|
6896 */
|
|
255
|
6897 void
|
|
236
|
6898 put_bytes(fd, nr, len)
|
|
|
6899 FILE *fd;
|
|
|
6900 long_u nr;
|
|
|
6901 int len;
|
|
|
6902 {
|
|
|
6903 int i;
|
|
|
6904
|
|
|
6905 for (i = len - 1; i >= 0; --i)
|
|
|
6906 putc((int)(nr >> (i * 8)), fd);
|
|
|
6907 }
|
|
|
6908
|
|
323
|
6909 static int
|
|
|
6910 #ifdef __BORLANDC__
|
|
|
6911 _RTLENTRYF
|
|
|
6912 #endif
|
|
|
6913 rep_compare __ARGS((const void *s1, const void *s2));
|
|
|
6914
|
|
|
6915 /*
|
|
|
6916 * Function given to qsort() to sort the REP items on "from" string.
|
|
|
6917 */
|
|
|
6918 static int
|
|
|
6919 #ifdef __BORLANDC__
|
|
|
6920 _RTLENTRYF
|
|
|
6921 #endif
|
|
|
6922 rep_compare(s1, s2)
|
|
|
6923 const void *s1;
|
|
|
6924 const void *s2;
|
|
|
6925 {
|
|
|
6926 fromto_T *p1 = (fromto_T *)s1;
|
|
|
6927 fromto_T *p2 = (fromto_T *)s2;
|
|
|
6928
|
|
|
6929 return STRCMP(p1->ft_from, p2->ft_from);
|
|
|
6930 }
|
|
|
6931
|
|
236
|
6932 /*
|
|
483
|
6933 * Write the Vim .spl file "fname".
|
|
500
|
6934 * Return FAIL or OK;
|
|
|
6935 */
|
|
|
6936 static int
|
|
474
|
6937 write_vim_spell(spin, fname)
|
|
|
6938 spellinfo_T *spin;
|
|
236
|
6939 char_u *fname;
|
|
|
6940 {
|
|
300
|
6941 FILE *fd;
|
|
|
6942 int regionmask;
|
|
236
|
6943 int round;
|
|
300
|
6944 wordnode_T *tree;
|
|
|
6945 int nodecount;
|
|
323
|
6946 int i;
|
|
|
6947 int l;
|
|
|
6948 garray_T *gap;
|
|
|
6949 fromto_T *ftp;
|
|
|
6950 char_u *p;
|
|
|
6951 int rr;
|
|
500
|
6952 int retval = OK;
|
|
236
|
6953
|
|
310
|
6954 fd = mch_fopen((char *)fname, "w");
|
|
300
|
6955 if (fd == NULL)
|
|
236
|
6956 {
|
|
|
6957 EMSG2(_(e_notopen), fname);
|
|
500
|
6958 return FAIL;
|
|
236
|
6959 }
|
|
|
6960
|
|
483
|
6961 /* <HEADER>: <fileID> <versionnr> */
|
|
300
|
6962 /* <fileID> */
|
|
|
6963 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
500
|
6964 {
|
|
300
|
6965 EMSG(_(e_write));
|
|
500
|
6966 retval = FAIL;
|
|
|
6967 }
|
|
483
|
6968 putc(VIMSPELLVERSION, fd); /* <versionnr> */
|
|
|
6969
|
|
|
6970 /*
|
|
|
6971 * <SECTIONS>: <section> ... <sectionend>
|
|
|
6972 */
|
|
|
6973
|
|
|
6974 /* SN_REGION: <regionname> ...
|
|
|
6975 * Write the region names only if there is more than one. */
|
|
316
|
6976 if (spin->si_region_count > 1)
|
|
236
|
6977 {
|
|
483
|
6978 putc(SN_REGION, fd); /* <sectionID> */
|
|
|
6979 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
6980 l = spin->si_region_count * 2;
|
|
|
6981 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
6982 fwrite(spin->si_region_name, (size_t)l, (size_t)1, fd);
|
|
|
6983 /* <regionname> ... */
|
|
316
|
6984 regionmask = (1 << spin->si_region_count) - 1;
|
|
236
|
6985 }
|
|
|
6986 else
|
|
300
|
6987 regionmask = 0;
|
|
483
|
6988
|
|
|
6989 /* SN_CHARFLAGS: <charflagslen> <charflags> <folcharslen> <folchars>
|
|
|
6990 *
|
|
|
6991 * The table with character flags and the table for case folding.
|
|
|
6992 * This makes sure the same characters are recognized as word characters
|
|
|
6993 * when generating an when using a spell file.
|
|
260
|
6994 * Skip this for ASCII, the table may conflict with the one used for
|
|
323
|
6995 * 'encoding'.
|
|
|
6996 * Also skip this for an .add.spl file, the main spell file must contain
|
|
|
6997 * the table (avoids that it conflicts). File is shorter too.
|
|
|
6998 */
|
|
483
|
6999 if (!spin->si_ascii && !spin->si_add)
|
|
|
7000 {
|
|
|
7001 char_u folchars[128 * 8];
|
|
|
7002 int flags;
|
|
|
7003
|
|
485
|
7004 putc(SN_CHARFLAGS, fd); /* <sectionID> */
|
|
483
|
7005 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7006
|
|
|
7007 /* Form the <folchars> string first, we need to know its length. */
|
|
|
7008 l = 0;
|
|
|
7009 for (i = 128; i < 256; ++i)
|
|
|
7010 {
|
|
|
7011 #ifdef FEAT_MBYTE
|
|
|
7012 if (has_mbyte)
|
|
|
7013 l += mb_char2bytes(spelltab.st_fold[i], folchars + l);
|
|
|
7014 else
|
|
|
7015 #endif
|
|
|
7016 folchars[l++] = spelltab.st_fold[i];
|
|
|
7017 }
|
|
|
7018 put_bytes(fd, (long_u)(1 + 128 + 2 + l), 4); /* <sectionlen> */
|
|
|
7019
|
|
|
7020 fputc(128, fd); /* <charflagslen> */
|
|
|
7021 for (i = 128; i < 256; ++i)
|
|
|
7022 {
|
|
|
7023 flags = 0;
|
|
|
7024 if (spelltab.st_isw[i])
|
|
|
7025 flags |= CF_WORD;
|
|
|
7026 if (spelltab.st_isu[i])
|
|
|
7027 flags |= CF_UPPER;
|
|
|
7028 fputc(flags, fd); /* <charflags> */
|
|
|
7029 }
|
|
|
7030
|
|
|
7031 put_bytes(fd, (long_u)l, 2); /* <folcharslen> */
|
|
|
7032 fwrite(folchars, (size_t)l, (size_t)1, fd); /* <folchars> */
|
|
|
7033 }
|
|
|
7034
|
|
|
7035 /* SN_MIDWORD: <midword> */
|
|
|
7036 if (spin->si_midword != NULL)
|
|
|
7037 {
|
|
|
7038 putc(SN_MIDWORD, fd); /* <sectionID> */
|
|
|
7039 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7040
|
|
366
|
7041 i = STRLEN(spin->si_midword);
|
|
483
|
7042 put_bytes(fd, (long_u)i, 4); /* <sectionlen> */
|
|
366
|
7043 fwrite(spin->si_midword, (size_t)i, (size_t)1, fd); /* <midword> */
|
|
|
7044 }
|
|
|
7045
|
|
483
|
7046 /* SN_PREFCOND: <prefcondcnt> <prefcond> ... */
|
|
|
7047 if (spin->si_prefcond.ga_len > 0)
|
|
|
7048 {
|
|
|
7049 putc(SN_PREFCOND, fd); /* <sectionID> */
|
|
|
7050 putc(SNF_REQUIRED, fd); /* <sectionflags> */
|
|
|
7051
|
|
|
7052 l = write_spell_prefcond(NULL, &spin->si_prefcond);
|
|
|
7053 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7054
|
|
|
7055 write_spell_prefcond(fd, &spin->si_prefcond);
|
|
|
7056 }
|
|
|
7057
|
|
|
7058 /* SN_REP: <repcount> <rep> ...
|
|
|
7059 * SN_SAL: <salflags> <salcount> <sal> ... */
|
|
366
|
7060
|
|
323
|
7061 /* Sort the REP items. */
|
|
|
7062 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
|
7063 sizeof(fromto_T), rep_compare);
|
|
|
7064
|
|
483
|
7065 /* round 1: SN_REP section
|
|
|
7066 * round 2: SN_SAL section (unless SN_SOFO is used) */
|
|
323
|
7067 for (round = 1; round <= 2; ++round)
|
|
|
7068 {
|
|
|
7069 if (round == 1)
|
|
483
|
7070 {
|
|
323
|
7071 gap = &spin->si_rep;
|
|
483
|
7072 putc(SN_REP, fd); /* <sectionID> */
|
|
|
7073 }
|
|
323
|
7074 else
|
|
|
7075 {
|
|
483
|
7076 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
7077 /* using SN_SOFO section instead of SN_SAL */
|
|
|
7078 break;
|
|
323
|
7079 gap = &spin->si_sal;
|
|
483
|
7080 putc(SN_SAL, fd); /* <sectionID> */
|
|
|
7081 }
|
|
|
7082
|
|
|
7083 /* This is for making suggestions, section is not required. */
|
|
|
7084 putc(0, fd); /* <sectionflags> */
|
|
|
7085
|
|
|
7086 /* Compute the length of what follows. */
|
|
|
7087 l = 2; /* count <repcount> or <salcount> */
|
|
|
7088 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7089 {
|
|
|
7090 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
7091 l += 1 + STRLEN(ftp->ft_from); /* count <*fromlen> and <*from> */
|
|
|
7092 l += 1 + STRLEN(ftp->ft_to); /* count <*tolen> and <*to> */
|
|
|
7093 }
|
|
|
7094 if (round == 2)
|
|
|
7095 ++l; /* count <salflags> */
|
|
|
7096 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7097
|
|
|
7098 if (round == 2)
|
|
|
7099 {
|
|
323
|
7100 i = 0;
|
|
|
7101 if (spin->si_followup)
|
|
|
7102 i |= SAL_F0LLOWUP;
|
|
|
7103 if (spin->si_collapse)
|
|
|
7104 i |= SAL_COLLAPSE;
|
|
|
7105 if (spin->si_rem_accents)
|
|
|
7106 i |= SAL_REM_ACCENTS;
|
|
|
7107 putc(i, fd); /* <salflags> */
|
|
|
7108 }
|
|
|
7109
|
|
|
7110 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
|
7111 for (i = 0; i < gap->ga_len; ++i)
|
|
|
7112 {
|
|
|
7113 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
|
7114 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
|
7115 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
|
7116 for (rr = 1; rr <= 2; ++rr)
|
|
|
7117 {
|
|
|
7118 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
|
7119 l = STRLEN(p);
|
|
|
7120 putc(l, fd);
|
|
|
7121 fwrite(p, l, (size_t)1, fd);
|
|
|
7122 }
|
|
|
7123 }
|
|
483
|
7124
|
|
|
7125 }
|
|
|
7126
|
|
|
7127 /* SN_SOFO: <sofofromlen> <sofofrom> <sofotolen> <sofoto>
|
|
|
7128 * This is for making suggestions, section is not required. */
|
|
375
|
7129 if (spin->si_sofofr != NULL && spin->si_sofoto != NULL)
|
|
|
7130 {
|
|
483
|
7131 putc(SN_SOFO, fd); /* <sectionID> */
|
|
|
7132 putc(0, fd); /* <sectionflags> */
|
|
375
|
7133
|
|
|
7134 l = STRLEN(spin->si_sofofr);
|
|
483
|
7135 put_bytes(fd, (long_u)(l + STRLEN(spin->si_sofoto) + 4), 4);
|
|
|
7136 /* <sectionlen> */
|
|
|
7137
|
|
|
7138 put_bytes(fd, (long_u)l, 2); /* <sofofromlen> */
|
|
|
7139 fwrite(spin->si_sofofr, l, (size_t)1, fd); /* <sofofrom> */
|
|
375
|
7140
|
|
|
7141 l = STRLEN(spin->si_sofoto);
|
|
483
|
7142 put_bytes(fd, (long_u)l, 2); /* <sofotolen> */
|
|
|
7143 fwrite(spin->si_sofoto, l, (size_t)1, fd); /* <sofoto> */
|
|
|
7144 }
|
|
|
7145
|
|
|
7146 /* SN_MAP: <mapstr>
|
|
|
7147 * This is for making suggestions, section is not required. */
|
|
|
7148 if (spin->si_map.ga_len > 0)
|
|
|
7149 {
|
|
|
7150 putc(SN_MAP, fd); /* <sectionID> */
|
|
|
7151 putc(0, fd); /* <sectionflags> */
|
|
|
7152 l = spin->si_map.ga_len;
|
|
|
7153 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7154 fwrite(spin->si_map.ga_data, (size_t)l, (size_t)1, fd);
|
|
|
7155 /* <mapstr> */
|
|
|
7156 }
|
|
|
7157
|
|
|
7158 /* SN_COMPOUND: compound info.
|
|
|
7159 * We don't mark it required, when not supported all compound words will
|
|
|
7160 * be bad words. */
|
|
|
7161 if (spin->si_compflags != NULL)
|
|
|
7162 {
|
|
|
7163 putc(SN_COMPOUND, fd); /* <sectionID> */
|
|
|
7164 putc(0, fd); /* <sectionflags> */
|
|
|
7165
|
|
|
7166 l = STRLEN(spin->si_compflags);
|
|
|
7167 put_bytes(fd, (long_u)(l + 3), 4); /* <sectionlen> */
|
|
|
7168 putc(spin->si_compmax, fd); /* <compmax> */
|
|
|
7169 putc(spin->si_compminlen, fd); /* <compminlen> */
|
|
|
7170 putc(spin->si_compsylmax, fd); /* <compsylmax> */
|
|
|
7171 /* <compflags> */
|
|
|
7172 fwrite(spin->si_compflags, (size_t)l, (size_t)1, fd);
|
|
|
7173 }
|
|
|
7174
|
|
492
|
7175 /* SN_NOBREAK: NOBREAK flag */
|
|
|
7176 if (spin->si_nobreak)
|
|
|
7177 {
|
|
|
7178 putc(SN_NOBREAK, fd); /* <sectionID> */
|
|
|
7179 putc(0, fd); /* <sectionflags> */
|
|
|
7180
|
|
|
7181 /* It's empty, the precense of the section flags the feature. */
|
|
|
7182 put_bytes(fd, (long_u)0, 4); /* <sectionlen> */
|
|
|
7183 }
|
|
|
7184
|
|
483
|
7185 /* SN_SYLLABLE: syllable info.
|
|
|
7186 * We don't mark it required, when not supported syllables will not be
|
|
|
7187 * counted. */
|
|
|
7188 if (spin->si_syllable != NULL)
|
|
|
7189 {
|
|
|
7190 putc(SN_SYLLABLE, fd); /* <sectionID> */
|
|
|
7191 putc(0, fd); /* <sectionflags> */
|
|
|
7192
|
|
|
7193 l = STRLEN(spin->si_syllable);
|
|
|
7194 put_bytes(fd, (long_u)l, 4); /* <sectionlen> */
|
|
|
7195 fwrite(spin->si_syllable, (size_t)l, (size_t)1, fd); /* <syllable> */
|
|
|
7196 }
|
|
|
7197
|
|
|
7198 /* end of <SECTIONS> */
|
|
|
7199 putc(SN_END, fd); /* <sectionend> */
|
|
|
7200
|
|
302
|
7201
|
|
236
|
7202 /*
|
|
339
|
7203 * <LWORDTREE> <KWORDTREE> <PREFIXTREE>
|
|
236
|
7204 */
|
|
323
|
7205 spin->si_memtot = 0;
|
|
339
|
7206 for (round = 1; round <= 3; ++round)
|
|
236
|
7207 {
|
|
339
|
7208 if (round == 1)
|
|
469
|
7209 tree = spin->si_foldroot->wn_sibling;
|
|
339
|
7210 else if (round == 2)
|
|
469
|
7211 tree = spin->si_keeproot->wn_sibling;
|
|
339
|
7212 else
|
|
469
|
7213 tree = spin->si_prefroot->wn_sibling;
|
|
236
|
7214
|
|
346
|
7215 /* Clear the index and wnode fields in the tree. */
|
|
|
7216 clear_node(tree);
|
|
|
7217
|
|
300
|
7218 /* Count the number of nodes. Needed to be able to allocate the
|
|
346
|
7219 * memory when reading the nodes. Also fills in index for shared
|
|
300
|
7220 * nodes. */
|
|
346
|
7221 nodecount = put_node(NULL, tree, 0, regionmask, round == 3);
|
|
236
|
7222
|
|
300
|
7223 /* number of nodes in 4 bytes */
|
|
|
7224 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
|
302
|
7225 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
|
236
|
7226
|
|
300
|
7227 /* Write the nodes. */
|
|
346
|
7228 (void)put_node(fd, tree, 0, regionmask, round == 3);
|
|
236
|
7229 }
|
|
|
7230
|
|
500
|
7231 /* Write another byte to check for errors. */
|
|
|
7232 if (putc(0, fd) == EOF)
|
|
|
7233 retval = FAIL;
|
|
|
7234
|
|
|
7235 if (fclose(fd) == EOF)
|
|
|
7236 retval = FAIL;
|
|
|
7237
|
|
|
7238 return retval;
|
|
236
|
7239 }
|
|
|
7240
|
|
|
7241 /*
|
|
346
|
7242 * Clear the index and wnode fields of "node", it siblings and its
|
|
|
7243 * children. This is needed because they are a union with other items to save
|
|
|
7244 * space.
|
|
|
7245 */
|
|
|
7246 static void
|
|
|
7247 clear_node(node)
|
|
|
7248 wordnode_T *node;
|
|
|
7249 {
|
|
|
7250 wordnode_T *np;
|
|
|
7251
|
|
|
7252 if (node != NULL)
|
|
|
7253 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7254 {
|
|
|
7255 np->wn_u1.index = 0;
|
|
|
7256 np->wn_u2.wnode = NULL;
|
|
|
7257
|
|
|
7258 if (np->wn_byte != NUL)
|
|
|
7259 clear_node(np->wn_child);
|
|
|
7260 }
|
|
|
7261 }
|
|
|
7262
|
|
|
7263
|
|
|
7264 /*
|
|
300
|
7265 * Dump a word tree at node "node".
|
|
|
7266 *
|
|
|
7267 * This first writes the list of possible bytes (siblings). Then for each
|
|
|
7268 * byte recursively write the children.
|
|
|
7269 *
|
|
|
7270 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
|
7271 * are made about the indexes (so that we don't have to write them in the
|
|
|
7272 * file).
|
|
236
|
7273 *
|
|
300
|
7274 * Returns the number of nodes used.
|
|
236
|
7275 */
|
|
300
|
7276 static int
|
|
346
|
7277 put_node(fd, node, index, regionmask, prefixtree)
|
|
339
|
7278 FILE *fd; /* NULL when only counting */
|
|
300
|
7279 wordnode_T *node;
|
|
|
7280 int index;
|
|
|
7281 int regionmask;
|
|
339
|
7282 int prefixtree; /* TRUE for PREFIXTREE */
|
|
236
|
7283 {
|
|
300
|
7284 int newindex = index;
|
|
|
7285 int siblingcount = 0;
|
|
|
7286 wordnode_T *np;
|
|
236
|
7287 int flags;
|
|
300
|
7288
|
|
|
7289 /* If "node" is zero the tree is empty. */
|
|
|
7290 if (node == NULL)
|
|
|
7291 return 0;
|
|
|
7292
|
|
|
7293 /* Store the index where this node is written. */
|
|
346
|
7294 node->wn_u1.index = index;
|
|
236
|
7295
|
|
300
|
7296 /* Count the number of siblings. */
|
|
|
7297 for (np = node; np != NULL; np = np->wn_sibling)
|
|
|
7298 ++siblingcount;
|
|
236
|
7299
|
|
300
|
7300 /* Write the sibling count. */
|
|
|
7301 if (fd != NULL)
|
|
|
7302 putc(siblingcount, fd); /* <siblingcount> */
|
|
236
|
7303
|
|
300
|
7304 /* Write each sibling byte and optionally extra info. */
|
|
|
7305 for (np = node; np != NULL; np = np->wn_sibling)
|
|
236
|
7306 {
|
|
300
|
7307 if (np->wn_byte == 0)
|
|
|
7308 {
|
|
|
7309 if (fd != NULL)
|
|
|
7310 {
|
|
339
|
7311 /* For a NUL byte (end of word) write the flags etc. */
|
|
|
7312 if (prefixtree)
|
|
300
|
7313 {
|
|
481
|
7314 /* In PREFIXTREE write the required affixID and the
|
|
390
|
7315 * associated condition nr (stored in wn_region). The
|
|
|
7316 * byte value is misused to store the "rare" and "not
|
|
|
7317 * combining" flags */
|
|
455
|
7318 if (np->wn_flags == (short_u)PFX_FLAGS)
|
|
|
7319 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7320 else
|
|
|
7321 {
|
|
390
|
7322 putc(BY_FLAGS, fd); /* <byte> */
|
|
455
|
7323 putc(np->wn_flags, fd); /* <pflags> */
|
|
|
7324 }
|
|
481
|
7325 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7326 put_bytes(fd, (long_u)np->wn_region, 2); /* <prefcondnr> */
|
|
300
|
7327 }
|
|
|
7328 else
|
|
|
7329 {
|
|
339
|
7330 /* For word trees we write the flag/region items. */
|
|
|
7331 flags = np->wn_flags;
|
|
|
7332 if (regionmask != 0 && np->wn_region != regionmask)
|
|
|
7333 flags |= WF_REGION;
|
|
481
|
7334 if (np->wn_affixID != 0)
|
|
|
7335 flags |= WF_AFX;
|
|
339
|
7336 if (flags == 0)
|
|
|
7337 {
|
|
|
7338 /* word without flags or region */
|
|
|
7339 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
|
7340 }
|
|
|
7341 else
|
|
|
7342 {
|
|
390
|
7343 if (np->wn_flags >= 0x100)
|
|
|
7344 {
|
|
|
7345 putc(BY_FLAGS2, fd); /* <byte> */
|
|
|
7346 putc(flags, fd); /* <flags> */
|
|
|
7347 putc((unsigned)flags >> 8, fd); /* <flags2> */
|
|
|
7348 }
|
|
|
7349 else
|
|
|
7350 {
|
|
|
7351 putc(BY_FLAGS, fd); /* <byte> */
|
|
|
7352 putc(flags, fd); /* <flags> */
|
|
|
7353 }
|
|
339
|
7354 if (flags & WF_REGION)
|
|
|
7355 putc(np->wn_region, fd); /* <region> */
|
|
481
|
7356 if (flags & WF_AFX)
|
|
|
7357 putc(np->wn_affixID, fd); /* <affixID> */
|
|
339
|
7358 }
|
|
300
|
7359 }
|
|
|
7360 }
|
|
|
7361 }
|
|
|
7362 else
|
|
|
7363 {
|
|
346
|
7364 if (np->wn_child->wn_u1.index != 0
|
|
|
7365 && np->wn_child->wn_u2.wnode != node)
|
|
300
|
7366 {
|
|
|
7367 /* The child is written elsewhere, write the reference. */
|
|
|
7368 if (fd != NULL)
|
|
|
7369 {
|
|
|
7370 putc(BY_INDEX, fd); /* <byte> */
|
|
|
7371 /* <nodeidx> */
|
|
346
|
7372 put_bytes(fd, (long_u)np->wn_child->wn_u1.index, 3);
|
|
300
|
7373 }
|
|
|
7374 }
|
|
346
|
7375 else if (np->wn_child->wn_u2.wnode == NULL)
|
|
300
|
7376 /* We will write the child below and give it an index. */
|
|
346
|
7377 np->wn_child->wn_u2.wnode = node;
|
|
236
|
7378
|
|
300
|
7379 if (fd != NULL)
|
|
|
7380 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
|
7381 {
|
|
|
7382 EMSG(_(e_write));
|
|
|
7383 return 0;
|
|
|
7384 }
|
|
|
7385 }
|
|
236
|
7386 }
|
|
|
7387
|
|
300
|
7388 /* Space used in the array when reading: one for each sibling and one for
|
|
|
7389 * the count. */
|
|
|
7390 newindex += siblingcount + 1;
|
|
249
|
7391
|
|
300
|
7392 /* Recursively dump the children of each sibling. */
|
|
|
7393 for (np = node; np != NULL; np = np->wn_sibling)
|
|
346
|
7394 if (np->wn_byte != 0 && np->wn_child->wn_u2.wnode == node)
|
|
|
7395 newindex = put_node(fd, np->wn_child, newindex, regionmask,
|
|
339
|
7396 prefixtree);
|
|
249
|
7397
|
|
300
|
7398 return newindex;
|
|
236
|
7399 }
|
|
|
7400
|
|
|
7401
|
|
|
7402 /*
|
|
310
|
7403 * ":mkspell [-ascii] outfile infile ..."
|
|
|
7404 * ":mkspell [-ascii] addfile"
|
|
236
|
7405 */
|
|
|
7406 void
|
|
|
7407 ex_mkspell(eap)
|
|
|
7408 exarg_T *eap;
|
|
|
7409 {
|
|
|
7410 int fcount;
|
|
|
7411 char_u **fnames;
|
|
310
|
7412 char_u *arg = eap->arg;
|
|
|
7413 int ascii = FALSE;
|
|
|
7414
|
|
|
7415 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
|
7416 {
|
|
|
7417 ascii = TRUE;
|
|
|
7418 arg = skipwhite(arg + 6);
|
|
|
7419 }
|
|
|
7420
|
|
|
7421 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
|
7422 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
|
7423 {
|
|
323
|
7424 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
|
310
|
7425 FreeWild(fcount, fnames);
|
|
|
7426 }
|
|
|
7427 }
|
|
|
7428
|
|
|
7429 /*
|
|
|
7430 * Create a Vim spell file from one or more word lists.
|
|
|
7431 * "fnames[0]" is the output file name.
|
|
|
7432 * "fnames[fcount - 1]" is the last input file name.
|
|
|
7433 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
|
7434 * and ".spl" is appended to make the output file name.
|
|
|
7435 */
|
|
|
7436 static void
|
|
323
|
7437 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
|
310
|
7438 int fcount;
|
|
|
7439 char_u **fnames;
|
|
|
7440 int ascii; /* -ascii argument given */
|
|
|
7441 int overwrite; /* overwrite existing output file */
|
|
323
|
7442 int added_word; /* invoked through "zg" */
|
|
310
|
7443 {
|
|
236
|
7444 char_u fname[MAXPATHL];
|
|
|
7445 char_u wfname[MAXPATHL];
|
|
310
|
7446 char_u **innames;
|
|
|
7447 int incount;
|
|
236
|
7448 afffile_T *(afile[8]);
|
|
|
7449 int i;
|
|
|
7450 int len;
|
|
|
7451 struct stat st;
|
|
255
|
7452 int error = FALSE;
|
|
300
|
7453 spellinfo_T spin;
|
|
|
7454
|
|
|
7455 vim_memset(&spin, 0, sizeof(spin));
|
|
323
|
7456 spin.si_verbose = !added_word;
|
|
310
|
7457 spin.si_ascii = ascii;
|
|
323
|
7458 spin.si_followup = TRUE;
|
|
|
7459 spin.si_rem_accents = TRUE;
|
|
|
7460 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
|
7461 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
|
7462 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
|
339
|
7463 ga_init2(&spin.si_prefcond, (int)sizeof(char_u *), 50);
|
|
500
|
7464 spin.si_newcompID = 127; /* start compound ID at first maximum */
|
|
310
|
7465
|
|
|
7466 /* default: fnames[0] is output file, following are input files */
|
|
|
7467 innames = &fnames[1];
|
|
|
7468 incount = fcount - 1;
|
|
|
7469
|
|
|
7470 if (fcount >= 1)
|
|
240
|
7471 {
|
|
310
|
7472 len = STRLEN(fnames[0]);
|
|
|
7473 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
|
7474 {
|
|
|
7475 /* For ":mkspell path/en.latin1.add" output file is
|
|
|
7476 * "path/en.latin1.add.spl". */
|
|
|
7477 innames = &fnames[0];
|
|
|
7478 incount = 1;
|
|
|
7479 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
|
7480 }
|
|
366
|
7481 else if (fcount == 1)
|
|
|
7482 {
|
|
|
7483 /* For ":mkspell path/vim" output file is "path/vim.latin1.spl". */
|
|
|
7484 innames = &fnames[0];
|
|
|
7485 incount = 1;
|
|
|
7486 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
7487 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
7488 }
|
|
310
|
7489 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
|
7490 {
|
|
|
7491 /* Name ends in ".spl", use as the file name. */
|
|
323
|
7492 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
|
310
|
7493 }
|
|
|
7494 else
|
|
|
7495 /* Name should be language, make the file name from it. */
|
|
|
7496 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
|
7497 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
|
7498
|
|
|
7499 /* Check for .ascii.spl. */
|
|
|
7500 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
|
7501 spin.si_ascii = TRUE;
|
|
|
7502
|
|
|
7503 /* Check for .add.spl. */
|
|
|
7504 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
|
7505 spin.si_add = TRUE;
|
|
240
|
7506 }
|
|
|
7507
|
|
310
|
7508 if (incount <= 0)
|
|
236
|
7509 EMSG(_(e_invarg)); /* need at least output and input names */
|
|
351
|
7510 else if (vim_strchr(gettail(wfname), '_') != NULL)
|
|
|
7511 EMSG(_("E751: Output file name must not have region name"));
|
|
310
|
7512 else if (incount > 8)
|
|
236
|
7513 EMSG(_("E754: Only up to 8 regions supported"));
|
|
|
7514 else
|
|
|
7515 {
|
|
|
7516 /* Check for overwriting before doing things that may take a lot of
|
|
|
7517 * time. */
|
|
310
|
7518 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
|
236
|
7519 {
|
|
|
7520 EMSG(_(e_exists));
|
|
310
|
7521 return;
|
|
236
|
7522 }
|
|
310
|
7523 if (mch_isdir(wfname))
|
|
236
|
7524 {
|
|
310
|
7525 EMSG2(_(e_isadir2), wfname);
|
|
|
7526 return;
|
|
236
|
7527 }
|
|
|
7528
|
|
|
7529 /*
|
|
|
7530 * Init the aff and dic pointers.
|
|
|
7531 * Get the region names if there are more than 2 arguments.
|
|
|
7532 */
|
|
310
|
7533 for (i = 0; i < incount; ++i)
|
|
236
|
7534 {
|
|
310
|
7535 afile[i] = NULL;
|
|
300
|
7536
|
|
316
|
7537 if (incount > 1)
|
|
236
|
7538 {
|
|
310
|
7539 len = STRLEN(innames[i]);
|
|
|
7540 if (STRLEN(gettail(innames[i])) < 5
|
|
|
7541 || innames[i][len - 3] != '_')
|
|
236
|
7542 {
|
|
310
|
7543 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
|
7544 return;
|
|
236
|
7545 }
|
|
316
|
7546 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
|
7547 spin.si_region_name[i * 2 + 1] =
|
|
|
7548 TOLOWER_ASC(innames[i][len - 1]);
|
|
236
|
7549 }
|
|
|
7550 }
|
|
316
|
7551 spin.si_region_count = incount;
|
|
236
|
7552
|
|
474
|
7553 spin.si_foldroot = wordtree_alloc(&spin);
|
|
|
7554 spin.si_keeproot = wordtree_alloc(&spin);
|
|
|
7555 spin.si_prefroot = wordtree_alloc(&spin);
|
|
339
|
7556 if (spin.si_foldroot == NULL
|
|
|
7557 || spin.si_keeproot == NULL
|
|
|
7558 || spin.si_prefroot == NULL)
|
|
300
|
7559 {
|
|
469
|
7560 free_blocks(spin.si_blocks);
|
|
310
|
7561 return;
|
|
300
|
7562 }
|
|
|
7563
|
|
351
|
7564 /* When not producing a .add.spl file clear the character table when
|
|
|
7565 * we encounter one in the .aff file. This means we dump the current
|
|
|
7566 * one in the .spl file if the .aff file doesn't define one. That's
|
|
|
7567 * better than guessing the contents, the table will match a
|
|
|
7568 * previously loaded spell file. */
|
|
|
7569 if (!spin.si_add)
|
|
|
7570 spin.si_clear_chartab = TRUE;
|
|
|
7571
|
|
236
|
7572 /*
|
|
|
7573 * Read all the .aff and .dic files.
|
|
|
7574 * Text is converted to 'encoding'.
|
|
300
|
7575 * Words are stored in the case-folded and keep-case trees.
|
|
236
|
7576 */
|
|
310
|
7577 for (i = 0; i < incount && !error; ++i)
|
|
236
|
7578 {
|
|
300
|
7579 spin.si_conv.vc_type = CONV_NONE;
|
|
310
|
7580 spin.si_region = 1 << i;
|
|
|
7581
|
|
|
7582 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
|
300
|
7583 if (mch_stat((char *)fname, &st) >= 0)
|
|
|
7584 {
|
|
|
7585 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
|
7586 * "SET" line. */
|
|
474
|
7587 afile[i] = spell_read_aff(&spin, fname);
|
|
310
|
7588 if (afile[i] == NULL)
|
|
300
|
7589 error = TRUE;
|
|
|
7590 else
|
|
|
7591 {
|
|
|
7592 /* Read the .dic file and store the words in the trees. */
|
|
|
7593 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
|
310
|
7594 innames[i]);
|
|
474
|
7595 if (spell_read_dic(&spin, fname, afile[i]) == FAIL)
|
|
300
|
7596 error = TRUE;
|
|
|
7597 }
|
|
|
7598 }
|
|
|
7599 else
|
|
|
7600 {
|
|
|
7601 /* No .aff file, try reading the file as a word list. Store
|
|
|
7602 * the words in the trees. */
|
|
474
|
7603 if (spell_read_wordfile(&spin, innames[i]) == FAIL)
|
|
300
|
7604 error = TRUE;
|
|
|
7605 }
|
|
236
|
7606
|
|
310
|
7607 #ifdef FEAT_MBYTE
|
|
236
|
7608 /* Free any conversion stuff. */
|
|
300
|
7609 convert_setup(&spin.si_conv, NULL, NULL);
|
|
310
|
7610 #endif
|
|
236
|
7611 }
|
|
|
7612
|
|
492
|
7613 if (spin.si_compflags != NULL && spin.si_nobreak)
|
|
|
7614 MSG(_("Warning: both compounding and NOBREAK specified"));
|
|
|
7615
|
|
300
|
7616 if (!error)
|
|
236
|
7617 {
|
|
|
7618 /*
|
|
300
|
7619 * Combine tails in the tree.
|
|
236
|
7620 */
|
|
474
|
7621 if (spin.si_verbose || p_verbose > 2)
|
|
|
7622 {
|
|
|
7623 if (!spin.si_verbose)
|
|
310
|
7624 verbose_enter();
|
|
469
|
7625 MSG(_(msg_compressing));
|
|
310
|
7626 out_flush();
|
|
474
|
7627 if (!spin.si_verbose)
|
|
310
|
7628 verbose_leave();
|
|
|
7629 }
|
|
474
|
7630 wordtree_compress(&spin, spin.si_foldroot);
|
|
|
7631 wordtree_compress(&spin, spin.si_keeproot);
|
|
|
7632 wordtree_compress(&spin, spin.si_prefroot);
|
|
236
|
7633 }
|
|
|
7634
|
|
300
|
7635 if (!error)
|
|
|
7636 {
|
|
|
7637 /*
|
|
|
7638 * Write the info in the spell file.
|
|
|
7639 */
|
|
474
|
7640 if (spin.si_verbose || p_verbose > 2)
|
|
|
7641 {
|
|
|
7642 if (!spin.si_verbose)
|
|
310
|
7643 verbose_enter();
|
|
366
|
7644 smsg((char_u *)_("Writing spell file %s ..."), wfname);
|
|
310
|
7645 out_flush();
|
|
474
|
7646 if (!spin.si_verbose)
|
|
310
|
7647 verbose_leave();
|
|
|
7648 }
|
|
|
7649
|
|
500
|
7650 error = write_vim_spell(&spin, wfname) == FAIL;
|
|
474
|
7651
|
|
|
7652 if (spin.si_verbose || p_verbose > 2)
|
|
|
7653 {
|
|
|
7654 if (!spin.si_verbose)
|
|
310
|
7655 verbose_enter();
|
|
|
7656 MSG(_("Done!"));
|
|
|
7657 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
|
302
|
7658 spin.si_memtot);
|
|
310
|
7659 out_flush();
|
|
474
|
7660 if (!spin.si_verbose)
|
|
310
|
7661 verbose_leave();
|
|
|
7662 }
|
|
|
7663
|
|
|
7664 /* If the file is loaded need to reload it. */
|
|
500
|
7665 if (!error)
|
|
|
7666 spell_reload_one(wfname, added_word);
|
|
300
|
7667 }
|
|
|
7668
|
|
|
7669 /* Free the allocated memory. */
|
|
323
|
7670 ga_clear(&spin.si_rep);
|
|
|
7671 ga_clear(&spin.si_sal);
|
|
|
7672 ga_clear(&spin.si_map);
|
|
339
|
7673 ga_clear(&spin.si_prefcond);
|
|
300
|
7674
|
|
|
7675 /* Free the .aff file structures. */
|
|
310
|
7676 for (i = 0; i < incount; ++i)
|
|
|
7677 if (afile[i] != NULL)
|
|
|
7678 spell_free_aff(afile[i]);
|
|
339
|
7679
|
|
|
7680 /* Free all the bits and pieces at once. */
|
|
|
7681 free_blocks(spin.si_blocks);
|
|
236
|
7682 }
|
|
310
|
7683 }
|
|
|
7684
|
|
|
7685
|
|
|
7686 /*
|
|
385
|
7687 * ":[count]spellgood {word}"
|
|
|
7688 * ":[count]spellwrong {word}"
|
|
310
|
7689 */
|
|
|
7690 void
|
|
|
7691 ex_spell(eap)
|
|
|
7692 exarg_T *eap;
|
|
|
7693 {
|
|
381
|
7694 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong,
|
|
385
|
7695 eap->forceit ? 0 : (int)eap->line2);
|
|
236
|
7696 }
|
|
|
7697
|
|
310
|
7698 /*
|
|
|
7699 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
|
7700 */
|
|
|
7701 void
|
|
385
|
7702 spell_add_word(word, len, bad, index)
|
|
310
|
7703 char_u *word;
|
|
|
7704 int len;
|
|
|
7705 int bad;
|
|
385
|
7706 int index; /* "zG" and "zW": zero, otherwise index in
|
|
|
7707 'spellfile' */
|
|
310
|
7708 {
|
|
|
7709 FILE *fd;
|
|
385
|
7710 buf_T *buf = NULL;
|
|
351
|
7711 int new_spf = FALSE;
|
|
381
|
7712 char_u *fname;
|
|
385
|
7713 char_u fnamebuf[MAXPATHL];
|
|
|
7714 char_u line[MAXWLEN * 2];
|
|
|
7715 long fpos, fpos_next = 0;
|
|
|
7716 int i;
|
|
|
7717 char_u *spf;
|
|
|
7718
|
|
|
7719 if (index == 0) /* use internal wordlist */
|
|
|
7720 {
|
|
|
7721 if (int_wordlist == NULL)
|
|
|
7722 {
|
|
|
7723 int_wordlist = vim_tempname('s');
|
|
|
7724 if (int_wordlist == NULL)
|
|
381
|
7725 return;
|
|
|
7726 }
|
|
385
|
7727 fname = int_wordlist;
|
|
381
|
7728 }
|
|
310
|
7729 else
|
|
|
7730 {
|
|
381
|
7731 /* If 'spellfile' isn't set figure out a good default value. */
|
|
|
7732 if (*curbuf->b_p_spf == NUL)
|
|
|
7733 {
|
|
|
7734 init_spellfile();
|
|
|
7735 new_spf = TRUE;
|
|
|
7736 }
|
|
|
7737
|
|
|
7738 if (*curbuf->b_p_spf == NUL)
|
|
|
7739 {
|
|
523
|
7740 EMSG2(_(e_notset), "spellfile");
|
|
381
|
7741 return;
|
|
|
7742 }
|
|
|
7743
|
|
385
|
7744 for (spf = curbuf->b_p_spf, i = 1; *spf != NUL; ++i)
|
|
|
7745 {
|
|
|
7746 copy_option_part(&spf, fnamebuf, MAXPATHL, ",");
|
|
|
7747 if (i == index)
|
|
|
7748 break;
|
|
|
7749 if (*spf == NUL)
|
|
|
7750 {
|
|
502
|
7751 EMSGN(_("E765: 'spellfile' does not have %ld entries"), index);
|
|
385
|
7752 return;
|
|
|
7753 }
|
|
|
7754 }
|
|
|
7755
|
|
310
|
7756 /* Check that the user isn't editing the .add file somewhere. */
|
|
385
|
7757 buf = buflist_findname_exp(fnamebuf);
|
|
310
|
7758 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
|
7759 buf = NULL;
|
|
|
7760 if (buf != NULL && bufIsChanged(buf))
|
|
381
|
7761 {
|
|
310
|
7762 EMSG(_(e_bufloaded));
|
|
381
|
7763 return;
|
|
|
7764 }
|
|
|
7765
|
|
385
|
7766 fname = fnamebuf;
|
|
|
7767 }
|
|
|
7768
|
|
|
7769 if (bad)
|
|
|
7770 {
|
|
|
7771 /* When the word also appears as good word we need to remove that one,
|
|
|
7772 * since its flags sort before the one with WF_BANNED. */
|
|
|
7773 fd = mch_fopen((char *)fname, "r");
|
|
|
7774 if (fd != NULL)
|
|
|
7775 {
|
|
|
7776 while (!vim_fgets(line, MAXWLEN * 2, fd))
|
|
|
7777 {
|
|
|
7778 fpos = fpos_next;
|
|
|
7779 fpos_next = ftell(fd);
|
|
|
7780 if (STRNCMP(word, line, len) == 0
|
|
|
7781 && (line[len] == '/' || line[len] < ' '))
|
|
|
7782 {
|
|
|
7783 /* Found duplicate word. Remove it by writing a '#' at
|
|
|
7784 * the start of the line. Mixing reading and writing
|
|
|
7785 * doesn't work for all systems, close the file first. */
|
|
|
7786 fclose(fd);
|
|
|
7787 fd = mch_fopen((char *)fname, "r+");
|
|
|
7788 if (fd == NULL)
|
|
|
7789 break;
|
|
|
7790 if (fseek(fd, fpos, SEEK_SET) == 0)
|
|
|
7791 fputc('#', fd);
|
|
|
7792 fseek(fd, fpos_next, SEEK_SET);
|
|
|
7793 }
|
|
|
7794 }
|
|
|
7795 fclose(fd);
|
|
|
7796 }
|
|
381
|
7797 }
|
|
|
7798
|
|
|
7799 fd = mch_fopen((char *)fname, "a");
|
|
|
7800 if (fd == NULL && new_spf)
|
|
|
7801 {
|
|
|
7802 /* We just initialized the 'spellfile' option and can't open the file.
|
|
|
7803 * We may need to create the "spell" directory first. We already
|
|
|
7804 * checked the runtime directory is writable in init_spellfile(). */
|
|
592
|
7805 if (!dir_of_file_exists(fname))
|
|
381
|
7806 {
|
|
|
7807 /* The directory doesn't exist. Try creating it and opening the
|
|
|
7808 * file again. */
|
|
|
7809 vim_mkdir(NameBuff, 0755);
|
|
|
7810 fd = mch_fopen((char *)fname, "a");
|
|
|
7811 }
|
|
|
7812 }
|
|
|
7813
|
|
|
7814 if (fd == NULL)
|
|
|
7815 EMSG2(_(e_notopen), fname);
|
|
|
7816 else
|
|
|
7817 {
|
|
|
7818 if (bad)
|
|
|
7819 fprintf(fd, "%.*s/!\n", len, word);
|
|
310
|
7820 else
|
|
381
|
7821 fprintf(fd, "%.*s\n", len, word);
|
|
|
7822 fclose(fd);
|
|
|
7823
|
|
|
7824 /* Update the .add.spl file. */
|
|
|
7825 mkspell(1, &fname, FALSE, TRUE, TRUE);
|
|
|
7826
|
|
|
7827 /* If the .add file is edited somewhere, reload it. */
|
|
|
7828 if (buf != NULL)
|
|
|
7829 buf_reload(buf);
|
|
|
7830
|
|
|
7831 redraw_all_later(NOT_VALID);
|
|
310
|
7832 }
|
|
|
7833 }
|
|
|
7834
|
|
|
7835 /*
|
|
|
7836 * Initialize 'spellfile' for the current buffer.
|
|
|
7837 */
|
|
|
7838 static void
|
|
|
7839 init_spellfile()
|
|
|
7840 {
|
|
|
7841 char_u buf[MAXPATHL];
|
|
|
7842 int l;
|
|
500
|
7843 char_u *fname;
|
|
310
|
7844 char_u *rtp;
|
|
351
|
7845 char_u *lend;
|
|
501
|
7846 int aspath = FALSE;
|
|
|
7847 char_u *lstart = curbuf->b_p_spl;
|
|
310
|
7848
|
|
|
7849 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
|
7850 {
|
|
501
|
7851 /* Find the end of the language name. Exclude the region. If there
|
|
|
7852 * is a path separator remember the start of the tail. */
|
|
351
|
7853 for (lend = curbuf->b_p_spl; *lend != NUL
|
|
|
7854 && vim_strchr((char_u *)",._", *lend) == NULL; ++lend)
|
|
501
|
7855 if (vim_ispathsep(*lend))
|
|
|
7856 {
|
|
|
7857 aspath = TRUE;
|
|
|
7858 lstart = lend + 1;
|
|
|
7859 }
|
|
351
|
7860
|
|
|
7861 /* Loop over all entries in 'runtimepath'. Use the first one where we
|
|
|
7862 * are allowed to write. */
|
|
310
|
7863 rtp = p_rtp;
|
|
|
7864 while (*rtp != NUL)
|
|
|
7865 {
|
|
501
|
7866 if (aspath)
|
|
|
7867 /* Use directory of an entry with path, e.g., for
|
|
|
7868 * "/dir/lg.utf-8.spl" use "/dir". */
|
|
|
7869 vim_strncpy(buf, curbuf->b_p_spl, lstart - curbuf->b_p_spl - 1);
|
|
|
7870 else
|
|
|
7871 /* Copy the path from 'runtimepath' to buf[]. */
|
|
|
7872 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
310
|
7873 if (filewritable(buf) == 2)
|
|
|
7874 {
|
|
316
|
7875 /* Use the first language name from 'spelllang' and the
|
|
|
7876 * encoding used in the first loaded .spl file. */
|
|
501
|
7877 if (aspath)
|
|
|
7878 vim_strncpy(buf, curbuf->b_p_spl, lend - curbuf->b_p_spl);
|
|
|
7879 else
|
|
|
7880 {
|
|
|
7881 l = STRLEN(buf);
|
|
|
7882 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
|
|
7883 "/spell/%.*s", (int)(lend - lstart), lstart);
|
|
|
7884 }
|
|
310
|
7885 l = STRLEN(buf);
|
|
501
|
7886 fname = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang->sl_fname;
|
|
|
7887 vim_snprintf((char *)buf + l, MAXPATHL - l, ".%s.add",
|
|
|
7888 fname != NULL
|
|
|
7889 && strstr((char *)gettail(fname), ".ascii.") != NULL
|
|
|
7890 ? (char_u *)"ascii" : spell_enc());
|
|
310
|
7891 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
|
7892 break;
|
|
|
7893 }
|
|
501
|
7894 aspath = FALSE;
|
|
310
|
7895 }
|
|
|
7896 }
|
|
|
7897 }
|
|
236
|
7898
|
|
300
|
7899
|
|
307
|
7900 /*
|
|
|
7901 * Init the chartab used for spelling for ASCII.
|
|
|
7902 * EBCDIC is not supported!
|
|
|
7903 */
|
|
|
7904 static void
|
|
|
7905 clear_spell_chartab(sp)
|
|
|
7906 spelltab_T *sp;
|
|
|
7907 {
|
|
324
|
7908 int i;
|
|
307
|
7909
|
|
|
7910 /* Init everything to FALSE. */
|
|
|
7911 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
|
7912 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
|
7913 for (i = 0; i < 256; ++i)
|
|
324
|
7914 {
|
|
307
|
7915 sp->st_fold[i] = i;
|
|
324
|
7916 sp->st_upper[i] = i;
|
|
|
7917 }
|
|
307
|
7918
|
|
|
7919 /* We include digits. A word shouldn't start with a digit, but handling
|
|
|
7920 * that is done separately. */
|
|
|
7921 for (i = '0'; i <= '9'; ++i)
|
|
|
7922 sp->st_isw[i] = TRUE;
|
|
|
7923 for (i = 'A'; i <= 'Z'; ++i)
|
|
|
7924 {
|
|
|
7925 sp->st_isw[i] = TRUE;
|
|
|
7926 sp->st_isu[i] = TRUE;
|
|
|
7927 sp->st_fold[i] = i + 0x20;
|
|
|
7928 }
|
|
|
7929 for (i = 'a'; i <= 'z'; ++i)
|
|
324
|
7930 {
|
|
307
|
7931 sp->st_isw[i] = TRUE;
|
|
324
|
7932 sp->st_upper[i] = i - 0x20;
|
|
|
7933 }
|
|
307
|
7934 }
|
|
|
7935
|
|
|
7936 /*
|
|
|
7937 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
|
7938 * Called once while starting up and when 'encoding' changes.
|
|
|
7939 * The default is to use isalpha(), but the spell file should define the word
|
|
|
7940 * characters to make it possible that 'encoding' differs from the current
|
|
390
|
7941 * locale. For utf-8 we don't use isalpha() but our own functions.
|
|
307
|
7942 */
|
|
|
7943 void
|
|
|
7944 init_spell_chartab()
|
|
|
7945 {
|
|
|
7946 int i;
|
|
|
7947
|
|
|
7948 did_set_spelltab = FALSE;
|
|
|
7949 clear_spell_chartab(&spelltab);
|
|
|
7950 #ifdef FEAT_MBYTE
|
|
|
7951 if (enc_dbcs)
|
|
|
7952 {
|
|
|
7953 /* DBCS: assume double-wide characters are word characters. */
|
|
|
7954 for (i = 128; i <= 255; ++i)
|
|
|
7955 if (MB_BYTE2LEN(i) == 2)
|
|
|
7956 spelltab.st_isw[i] = TRUE;
|
|
|
7957 }
|
|
324
|
7958 else if (enc_utf8)
|
|
|
7959 {
|
|
|
7960 for (i = 128; i < 256; ++i)
|
|
|
7961 {
|
|
|
7962 spelltab.st_isu[i] = utf_isupper(i);
|
|
|
7963 spelltab.st_isw[i] = spelltab.st_isu[i] || utf_islower(i);
|
|
|
7964 spelltab.st_fold[i] = utf_fold(i);
|
|
|
7965 spelltab.st_upper[i] = utf_toupper(i);
|
|
|
7966 }
|
|
|
7967 }
|
|
307
|
7968 else
|
|
|
7969 #endif
|
|
|
7970 {
|
|
324
|
7971 /* Rough guess: use locale-dependent library functions. */
|
|
307
|
7972 for (i = 128; i < 256; ++i)
|
|
|
7973 {
|
|
|
7974 if (MB_ISUPPER(i))
|
|
|
7975 {
|
|
324
|
7976 spelltab.st_isw[i] = TRUE;
|
|
307
|
7977 spelltab.st_isu[i] = TRUE;
|
|
|
7978 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
|
7979 }
|
|
324
|
7980 else if (MB_ISLOWER(i))
|
|
|
7981 {
|
|
|
7982 spelltab.st_isw[i] = TRUE;
|
|
|
7983 spelltab.st_upper[i] = MB_TOUPPER(i);
|
|
|
7984 }
|
|
307
|
7985 }
|
|
|
7986 }
|
|
|
7987 }
|
|
|
7988
|
|
|
7989 /*
|
|
|
7990 * Set the spell character tables from strings in the affix file.
|
|
|
7991 */
|
|
|
7992 static int
|
|
|
7993 set_spell_chartab(fol, low, upp)
|
|
|
7994 char_u *fol;
|
|
|
7995 char_u *low;
|
|
|
7996 char_u *upp;
|
|
|
7997 {
|
|
|
7998 /* We build the new tables here first, so that we can compare with the
|
|
|
7999 * previous one. */
|
|
|
8000 spelltab_T new_st;
|
|
|
8001 char_u *pf = fol, *pl = low, *pu = upp;
|
|
|
8002 int f, l, u;
|
|
|
8003
|
|
|
8004 clear_spell_chartab(&new_st);
|
|
|
8005
|
|
|
8006 while (*pf != NUL)
|
|
|
8007 {
|
|
|
8008 if (*pl == NUL || *pu == NUL)
|
|
|
8009 {
|
|
|
8010 EMSG(_(e_affform));
|
|
|
8011 return FAIL;
|
|
|
8012 }
|
|
|
8013 #ifdef FEAT_MBYTE
|
|
|
8014 f = mb_ptr2char_adv(&pf);
|
|
|
8015 l = mb_ptr2char_adv(&pl);
|
|
|
8016 u = mb_ptr2char_adv(&pu);
|
|
|
8017 #else
|
|
|
8018 f = *pf++;
|
|
|
8019 l = *pl++;
|
|
|
8020 u = *pu++;
|
|
|
8021 #endif
|
|
|
8022 /* Every character that appears is a word character. */
|
|
|
8023 if (f < 256)
|
|
|
8024 new_st.st_isw[f] = TRUE;
|
|
|
8025 if (l < 256)
|
|
|
8026 new_st.st_isw[l] = TRUE;
|
|
|
8027 if (u < 256)
|
|
|
8028 new_st.st_isw[u] = TRUE;
|
|
|
8029
|
|
|
8030 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
|
8031 * case-folding */
|
|
|
8032 if (l < 256 && l != f)
|
|
|
8033 {
|
|
|
8034 if (f >= 256)
|
|
|
8035 {
|
|
|
8036 EMSG(_(e_affrange));
|
|
|
8037 return FAIL;
|
|
|
8038 }
|
|
|
8039 new_st.st_fold[l] = f;
|
|
|
8040 }
|
|
|
8041
|
|
|
8042 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
324
|
8043 * case-folding, it's upper case and the "UPP" is the upper case of
|
|
|
8044 * "FOL" . */
|
|
307
|
8045 if (u < 256 && u != f)
|
|
|
8046 {
|
|
|
8047 if (f >= 256)
|
|
|
8048 {
|
|
|
8049 EMSG(_(e_affrange));
|
|
|
8050 return FAIL;
|
|
|
8051 }
|
|
|
8052 new_st.st_fold[u] = f;
|
|
|
8053 new_st.st_isu[u] = TRUE;
|
|
324
|
8054 new_st.st_upper[f] = u;
|
|
307
|
8055 }
|
|
|
8056 }
|
|
|
8057
|
|
|
8058 if (*pl != NUL || *pu != NUL)
|
|
|
8059 {
|
|
|
8060 EMSG(_(e_affform));
|
|
|
8061 return FAIL;
|
|
|
8062 }
|
|
|
8063
|
|
|
8064 return set_spell_finish(&new_st);
|
|
|
8065 }
|
|
|
8066
|
|
|
8067 /*
|
|
|
8068 * Set the spell character tables from strings in the .spl file.
|
|
|
8069 */
|
|
483
|
8070 static void
|
|
389
|
8071 set_spell_charflags(flags, cnt, fol)
|
|
307
|
8072 char_u *flags;
|
|
389
|
8073 int cnt; /* length of "flags" */
|
|
|
8074 char_u *fol;
|
|
307
|
8075 {
|
|
|
8076 /* We build the new tables here first, so that we can compare with the
|
|
|
8077 * previous one. */
|
|
|
8078 spelltab_T new_st;
|
|
|
8079 int i;
|
|
389
|
8080 char_u *p = fol;
|
|
324
|
8081 int c;
|
|
307
|
8082
|
|
|
8083 clear_spell_chartab(&new_st);
|
|
|
8084
|
|
389
|
8085 for (i = 0; i < 128; ++i)
|
|
|
8086 {
|
|
|
8087 if (i < cnt)
|
|
|
8088 {
|
|
|
8089 new_st.st_isw[i + 128] = (flags[i] & CF_WORD) != 0;
|
|
|
8090 new_st.st_isu[i + 128] = (flags[i] & CF_UPPER) != 0;
|
|
|
8091 }
|
|
|
8092
|
|
|
8093 if (*p != NUL)
|
|
|
8094 {
|
|
307
|
8095 #ifdef FEAT_MBYTE
|
|
389
|
8096 c = mb_ptr2char_adv(&p);
|
|
307
|
8097 #else
|
|
389
|
8098 c = *p++;
|
|
307
|
8099 #endif
|
|
389
|
8100 new_st.st_fold[i + 128] = c;
|
|
|
8101 if (i + 128 != c && new_st.st_isu[i + 128] && c < 256)
|
|
|
8102 new_st.st_upper[c] = i + 128;
|
|
|
8103 }
|
|
307
|
8104 }
|
|
|
8105
|
|
483
|
8106 (void)set_spell_finish(&new_st);
|
|
307
|
8107 }
|
|
|
8108
|
|
|
8109 static int
|
|
|
8110 set_spell_finish(new_st)
|
|
|
8111 spelltab_T *new_st;
|
|
|
8112 {
|
|
|
8113 int i;
|
|
|
8114
|
|
|
8115 if (did_set_spelltab)
|
|
|
8116 {
|
|
|
8117 /* check that it's the same table */
|
|
|
8118 for (i = 0; i < 256; ++i)
|
|
|
8119 {
|
|
|
8120 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
|
8121 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
324
|
8122 || spelltab.st_fold[i] != new_st->st_fold[i]
|
|
|
8123 || spelltab.st_upper[i] != new_st->st_upper[i])
|
|
307
|
8124 {
|
|
|
8125 EMSG(_("E763: Word characters differ between spell files"));
|
|
|
8126 return FAIL;
|
|
|
8127 }
|
|
|
8128 }
|
|
|
8129 }
|
|
|
8130 else
|
|
|
8131 {
|
|
|
8132 /* copy the new spelltab into the one being used */
|
|
|
8133 spelltab = *new_st;
|
|
|
8134 did_set_spelltab = TRUE;
|
|
|
8135 }
|
|
|
8136
|
|
|
8137 return OK;
|
|
|
8138 }
|
|
|
8139
|
|
|
8140 /*
|
|
358
|
8141 * Return TRUE if "p" points to a word character.
|
|
366
|
8142 * As a special case we see "midword" characters as word character when it is
|
|
358
|
8143 * followed by a word character. This finds they'there but not 'they there'.
|
|
366
|
8144 * Thus this only works properly when past the first character of the word.
|
|
358
|
8145 */
|
|
|
8146 static int
|
|
376
|
8147 spell_iswordp(p, buf)
|
|
358
|
8148 char_u *p;
|
|
376
|
8149 buf_T *buf; /* buffer used */
|
|
358
|
8150 {
|
|
366
|
8151 #ifdef FEAT_MBYTE
|
|
358
|
8152 char_u *s;
|
|
366
|
8153 int l;
|
|
|
8154 int c;
|
|
|
8155
|
|
|
8156 if (has_mbyte)
|
|
|
8157 {
|
|
|
8158 l = MB_BYTE2LEN(*p);
|
|
358
|
8159 s = p;
|
|
366
|
8160 if (l == 1)
|
|
|
8161 {
|
|
|
8162 /* be quick for ASCII */
|
|
376
|
8163 if (buf->b_spell_ismw[*p])
|
|
366
|
8164 {
|
|
|
8165 s = p + 1; /* skip a mid-word character */
|
|
|
8166 l = MB_BYTE2LEN(*s);
|
|
|
8167 }
|
|
|
8168 }
|
|
|
8169 else
|
|
|
8170 {
|
|
|
8171 c = mb_ptr2char(p);
|
|
376
|
8172 if (c < 256 ? buf->b_spell_ismw[c]
|
|
|
8173 : (buf->b_spell_ismw_mb != NULL
|
|
|
8174 && vim_strchr(buf->b_spell_ismw_mb, c) != NULL))
|
|
366
|
8175 {
|
|
|
8176 s = p + l;
|
|
|
8177 l = MB_BYTE2LEN(*s);
|
|
|
8178 }
|
|
|
8179 }
|
|
|
8180
|
|
390
|
8181 c = mb_ptr2char(s);
|
|
|
8182 if (c > 255)
|
|
366
|
8183 return mb_get_class(s) >= 2;
|
|
390
|
8184 return spelltab.st_isw[c];
|
|
366
|
8185 }
|
|
358
|
8186 #endif
|
|
366
|
8187
|
|
376
|
8188 return spelltab.st_isw[buf->b_spell_ismw[*p] ? p[1] : p[0]];
|
|
|
8189 }
|
|
|
8190
|
|
|
8191 /*
|
|
|
8192 * Return TRUE if "p" points to a word character.
|
|
|
8193 * Unlike spell_iswordp() this doesn't check for "midword" characters.
|
|
|
8194 */
|
|
|
8195 static int
|
|
|
8196 spell_iswordp_nmw(p)
|
|
|
8197 char_u *p;
|
|
|
8198 {
|
|
|
8199 #ifdef FEAT_MBYTE
|
|
390
|
8200 int c;
|
|
|
8201
|
|
|
8202 if (has_mbyte)
|
|
|
8203 {
|
|
|
8204 c = mb_ptr2char(p);
|
|
|
8205 if (c > 255)
|
|
|
8206 return mb_get_class(p) >= 2;
|
|
|
8207 return spelltab.st_isw[c];
|
|
|
8208 }
|
|
376
|
8209 #endif
|
|
|
8210 return spelltab.st_isw[*p];
|
|
358
|
8211 }
|
|
|
8212
|
|
372
|
8213 #ifdef FEAT_MBYTE
|
|
|
8214 /*
|
|
|
8215 * Return TRUE if "p" points to a word character.
|
|
|
8216 * Wide version of spell_iswordp().
|
|
|
8217 */
|
|
|
8218 static int
|
|
376
|
8219 spell_iswordp_w(p, buf)
|
|
372
|
8220 int *p;
|
|
376
|
8221 buf_T *buf;
|
|
372
|
8222 {
|
|
|
8223 int *s;
|
|
|
8224
|
|
376
|
8225 if (*p < 256 ? buf->b_spell_ismw[*p]
|
|
|
8226 : (buf->b_spell_ismw_mb != NULL
|
|
|
8227 && vim_strchr(buf->b_spell_ismw_mb, *p) != NULL))
|
|
372
|
8228 s = p + 1;
|
|
|
8229 else
|
|
|
8230 s = p;
|
|
|
8231
|
|
390
|
8232 if (*s > 255)
|
|
372
|
8233 {
|
|
|
8234 if (enc_utf8)
|
|
|
8235 return utf_class(*s) >= 2;
|
|
|
8236 if (enc_dbcs)
|
|
|
8237 return dbcs_class((unsigned)*s >> 8, *s & 0xff) >= 2;
|
|
|
8238 return 0;
|
|
|
8239 }
|
|
|
8240 return spelltab.st_isw[*s];
|
|
|
8241 }
|
|
|
8242 #endif
|
|
|
8243
|
|
358
|
8244 /*
|
|
339
|
8245 * Write the table with prefix conditions to the .spl file.
|
|
483
|
8246 * When "fd" is NULL only count the length of what is written.
|
|
|
8247 */
|
|
|
8248 static int
|
|
339
|
8249 write_spell_prefcond(fd, gap)
|
|
|
8250 FILE *fd;
|
|
|
8251 garray_T *gap;
|
|
|
8252 {
|
|
|
8253 int i;
|
|
|
8254 char_u *p;
|
|
|
8255 int len;
|
|
483
|
8256 int totlen;
|
|
|
8257
|
|
|
8258 if (fd != NULL)
|
|
|
8259 put_bytes(fd, (long_u)gap->ga_len, 2); /* <prefcondcnt> */
|
|
|
8260
|
|
|
8261 totlen = 2 + gap->ga_len; /* length of <prefcondcnt> and <condlen> bytes */
|
|
339
|
8262
|
|
|
8263 for (i = 0; i < gap->ga_len; ++i)
|
|
|
8264 {
|
|
|
8265 /* <prefcond> : <condlen> <condstr> */
|
|
|
8266 p = ((char_u **)gap->ga_data)[i];
|
|
483
|
8267 if (p != NULL)
|
|
339
|
8268 {
|
|
|
8269 len = STRLEN(p);
|
|
483
|
8270 if (fd != NULL)
|
|
|
8271 {
|
|
|
8272 fputc(len, fd);
|
|
|
8273 fwrite(p, (size_t)len, (size_t)1, fd);
|
|
|
8274 }
|
|
|
8275 totlen += len;
|
|
|
8276 }
|
|
|
8277 else if (fd != NULL)
|
|
|
8278 fputc(0, fd);
|
|
|
8279 }
|
|
|
8280
|
|
|
8281 return totlen;
|
|
307
|
8282 }
|
|
|
8283
|
|
|
8284 /*
|
|
324
|
8285 * Case-fold "str[len]" into "buf[buflen]". The result is NUL terminated.
|
|
|
8286 * Uses the character definitions from the .spl file.
|
|
307
|
8287 * When using a multi-byte 'encoding' the length may change!
|
|
|
8288 * Returns FAIL when something wrong.
|
|
|
8289 */
|
|
|
8290 static int
|
|
324
|
8291 spell_casefold(str, len, buf, buflen)
|
|
|
8292 char_u *str;
|
|
307
|
8293 int len;
|
|
|
8294 char_u *buf;
|
|
|
8295 int buflen;
|
|
|
8296 {
|
|
|
8297 int i;
|
|
|
8298
|
|
|
8299 if (len >= buflen)
|
|
|
8300 {
|
|
|
8301 buf[0] = NUL;
|
|
|
8302 return FAIL; /* result will not fit */
|
|
|
8303 }
|
|
|
8304
|
|
|
8305 #ifdef FEAT_MBYTE
|
|
|
8306 if (has_mbyte)
|
|
|
8307 {
|
|
324
|
8308 int outi = 0;
|
|
|
8309 char_u *p;
|
|
307
|
8310 int c;
|
|
|
8311
|
|
|
8312 /* Fold one character at a time. */
|
|
324
|
8313 for (p = str; p < str + len; )
|
|
307
|
8314 {
|
|
|
8315 if (outi + MB_MAXBYTES > buflen)
|
|
|
8316 {
|
|
|
8317 buf[outi] = NUL;
|
|
|
8318 return FAIL;
|
|
|
8319 }
|
|
474
|
8320 c = mb_cptr2char_adv(&p);
|
|
324
|
8321 outi += mb_char2bytes(SPELL_TOFOLD(c), buf + outi);
|
|
307
|
8322 }
|
|
|
8323 buf[outi] = NUL;
|
|
|
8324 }
|
|
|
8325 else
|
|
|
8326 #endif
|
|
|
8327 {
|
|
|
8328 /* Be quick for non-multibyte encodings. */
|
|
|
8329 for (i = 0; i < len; ++i)
|
|
324
|
8330 buf[i] = spelltab.st_fold[str[i]];
|
|
307
|
8331 buf[i] = NUL;
|
|
|
8332 }
|
|
|
8333
|
|
|
8334 return OK;
|
|
|
8335 }
|
|
|
8336
|
|
372
|
8337 #define SPS_BEST 1
|
|
|
8338 #define SPS_FAST 2
|
|
|
8339 #define SPS_DOUBLE 4
|
|
|
8340
|
|
|
8341 static int sps_flags = SPS_BEST;
|
|
483
|
8342 static int sps_limit = 9999;
|
|
372
|
8343
|
|
|
8344 /*
|
|
|
8345 * Check the 'spellsuggest' option. Return FAIL if it's wrong.
|
|
483
|
8346 * Sets "sps_flags" and "sps_limit".
|
|
372
|
8347 */
|
|
|
8348 int
|
|
|
8349 spell_check_sps()
|
|
|
8350 {
|
|
|
8351 char_u *p;
|
|
483
|
8352 char_u *s;
|
|
372
|
8353 char_u buf[MAXPATHL];
|
|
|
8354 int f;
|
|
|
8355
|
|
|
8356 sps_flags = 0;
|
|
483
|
8357 sps_limit = 9999;
|
|
372
|
8358
|
|
|
8359 for (p = p_sps; *p != NUL; )
|
|
|
8360 {
|
|
|
8361 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
8362
|
|
|
8363 f = 0;
|
|
483
|
8364 if (VIM_ISDIGIT(*buf))
|
|
|
8365 {
|
|
|
8366 s = buf;
|
|
|
8367 sps_limit = getdigits(&s);
|
|
|
8368 if (*s != NUL && !VIM_ISDIGIT(*s))
|
|
|
8369 f = -1;
|
|
|
8370 }
|
|
|
8371 else if (STRCMP(buf, "best") == 0)
|
|
372
|
8372 f = SPS_BEST;
|
|
|
8373 else if (STRCMP(buf, "fast") == 0)
|
|
|
8374 f = SPS_FAST;
|
|
|
8375 else if (STRCMP(buf, "double") == 0)
|
|
|
8376 f = SPS_DOUBLE;
|
|
|
8377 else if (STRNCMP(buf, "expr:", 5) != 0
|
|
|
8378 && STRNCMP(buf, "file:", 5) != 0)
|
|
|
8379 f = -1;
|
|
|
8380
|
|
|
8381 if (f == -1 || (sps_flags != 0 && f != 0))
|
|
|
8382 {
|
|
|
8383 sps_flags = SPS_BEST;
|
|
483
|
8384 sps_limit = 9999;
|
|
372
|
8385 return FAIL;
|
|
|
8386 }
|
|
|
8387 if (f != 0)
|
|
|
8388 sps_flags = f;
|
|
|
8389 }
|
|
|
8390
|
|
|
8391 if (sps_flags == 0)
|
|
|
8392 sps_flags = SPS_BEST;
|
|
|
8393
|
|
|
8394 return OK;
|
|
|
8395 }
|
|
|
8396
|
|
|
8397 /* Remember what "z?" replaced. */
|
|
|
8398 static char_u *repl_from = NULL;
|
|
|
8399 static char_u *repl_to = NULL;
|
|
|
8400
|
|
323
|
8401 /*
|
|
|
8402 * "z?": Find badly spelled word under or after the cursor.
|
|
|
8403 * Give suggestions for the properly spelled word.
|
|
485
|
8404 * When "count" is non-zero use that suggestion.
|
|
323
|
8405 */
|
|
|
8406 void
|
|
485
|
8407 spell_suggest(count)
|
|
|
8408 int count;
|
|
323
|
8409 {
|
|
|
8410 char_u *line;
|
|
|
8411 pos_T prev_cursor = curwin->w_cursor;
|
|
|
8412 char_u wcopy[MAXWLEN + 2];
|
|
|
8413 char_u *p;
|
|
|
8414 int i;
|
|
|
8415 int c;
|
|
|
8416 suginfo_T sug;
|
|
|
8417 suggest_T *stp;
|
|
372
|
8418 int mouse_used;
|
|
388
|
8419 int need_cap;
|
|
483
|
8420 int limit;
|
|
485
|
8421 int selected = count;
|
|
323
|
8422
|
|
344
|
8423 /* Find the start of the badly spelled word. */
|
|
498
|
8424 if (spell_move_to(curwin, FORWARD, TRUE, TRUE, NULL) == 0
|
|
346
|
8425 || curwin->w_cursor.col > prev_cursor.col)
|
|
|
8426 {
|
|
|
8427 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
|
|
8428 return;
|
|
|
8429
|
|
|
8430 /* No bad word or it starts after the cursor: use the word under the
|
|
|
8431 * cursor. */
|
|
|
8432 curwin->w_cursor = prev_cursor;
|
|
|
8433 line = ml_get_curline();
|
|
|
8434 p = line + curwin->w_cursor.col;
|
|
|
8435 /* Backup to before start of word. */
|
|
390
|
8436 while (p > line && spell_iswordp_nmw(p))
|
|
346
|
8437 mb_ptr_back(line, p);
|
|
|
8438 /* Forward to start of word. */
|
|
390
|
8439 while (*p != NUL && !spell_iswordp_nmw(p))
|
|
346
|
8440 mb_ptr_adv(p);
|
|
|
8441
|
|
390
|
8442 if (!spell_iswordp_nmw(p)) /* No word found. */
|
|
346
|
8443 {
|
|
|
8444 beep_flush();
|
|
|
8445 return;
|
|
|
8446 }
|
|
|
8447 curwin->w_cursor.col = p - line;
|
|
323
|
8448 }
|
|
|
8449
|
|
344
|
8450 /* Get the word and its length. */
|
|
|
8451
|
|
388
|
8452 /* Figure out if the word should be capitalised. */
|
|
475
|
8453 need_cap = check_need_cap(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
|
8454
|
|
|
8455 line = ml_get_curline();
|
|
388
|
8456
|
|
483
|
8457 /* Get the list of suggestions. Limit to 'lines' - 2 or the number in
|
|
|
8458 * 'spellsuggest', whatever is smaller. */
|
|
|
8459 if (sps_limit > (int)Rows - 2)
|
|
|
8460 limit = (int)Rows - 2;
|
|
|
8461 else
|
|
|
8462 limit = sps_limit;
|
|
|
8463 spell_find_suggest(line + curwin->w_cursor.col, &sug, limit,
|
|
388
|
8464 TRUE, need_cap);
|
|
323
|
8465
|
|
|
8466 if (sug.su_ga.ga_len == 0)
|
|
|
8467 MSG(_("Sorry, no suggestions"));
|
|
485
|
8468 else if (count > 0)
|
|
|
8469 {
|
|
|
8470 if (count > sug.su_ga.ga_len)
|
|
|
8471 smsg((char_u *)_("Sorry, only %ld suggestions"),
|
|
|
8472 (long)sug.su_ga.ga_len);
|
|
|
8473 }
|
|
323
|
8474 else
|
|
|
8475 {
|
|
372
|
8476 vim_free(repl_from);
|
|
|
8477 repl_from = NULL;
|
|
|
8478 vim_free(repl_to);
|
|
|
8479 repl_to = NULL;
|
|
|
8480
|
|
474
|
8481 #ifdef FEAT_RIGHTLEFT
|
|
|
8482 /* When 'rightleft' is set the list is drawn right-left. */
|
|
|
8483 cmdmsg_rl = curwin->w_p_rl;
|
|
|
8484 if (cmdmsg_rl)
|
|
|
8485 msg_col = Columns - 1;
|
|
|
8486 #endif
|
|
|
8487
|
|
323
|
8488 /* List the suggestions. */
|
|
|
8489 msg_start();
|
|
372
|
8490 lines_left = Rows; /* avoid more prompt */
|
|
323
|
8491 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
|
8492 sug.su_badlen, sug.su_badptr);
|
|
474
|
8493 #ifdef FEAT_RIGHTLEFT
|
|
|
8494 if (cmdmsg_rl && STRNCMP(IObuff, "Change", 6) == 0)
|
|
|
8495 {
|
|
|
8496 /* And now the rabbit from the high hat: Avoid showing the
|
|
|
8497 * untranslated message rightleft. */
|
|
|
8498 vim_snprintf((char *)IObuff, IOSIZE, ":ot \"%.*s\" egnahC",
|
|
|
8499 sug.su_badlen, sug.su_badptr);
|
|
|
8500 }
|
|
|
8501 #endif
|
|
323
|
8502 msg_puts(IObuff);
|
|
|
8503 msg_clr_eos();
|
|
|
8504 msg_putchar('\n');
|
|
346
|
8505
|
|
323
|
8506 msg_scroll = TRUE;
|
|
|
8507 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
|
8508 {
|
|
344
|
8509 stp = &SUG(sug.su_ga, i);
|
|
323
|
8510
|
|
|
8511 /* The suggested word may replace only part of the bad word, add
|
|
|
8512 * the not replaced part. */
|
|
|
8513 STRCPY(wcopy, stp->st_word);
|
|
|
8514 if (sug.su_badlen > stp->st_orglen)
|
|
|
8515 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
|
8516 sug.su_badptr + stp->st_orglen,
|
|
|
8517 sug.su_badlen - stp->st_orglen);
|
|
474
|
8518 vim_snprintf((char *)IObuff, IOSIZE, "%2d", i + 1);
|
|
|
8519 #ifdef FEAT_RIGHTLEFT
|
|
|
8520 if (cmdmsg_rl)
|
|
|
8521 rl_mirror(IObuff);
|
|
|
8522 #endif
|
|
|
8523 msg_puts(IObuff);
|
|
|
8524
|
|
|
8525 vim_snprintf((char *)IObuff, IOSIZE, " \"%s\"", wcopy);
|
|
346
|
8526 msg_puts(IObuff);
|
|
|
8527
|
|
|
8528 /* The word may replace more than "su_badlen". */
|
|
|
8529 if (sug.su_badlen < stp->st_orglen)
|
|
|
8530 {
|
|
|
8531 vim_snprintf((char *)IObuff, IOSIZE, _(" < \"%.*s\""),
|
|
|
8532 stp->st_orglen, sug.su_badptr);
|
|
|
8533 msg_puts(IObuff);
|
|
|
8534 }
|
|
|
8535
|
|
324
|
8536 if (p_verbose > 0)
|
|
344
|
8537 {
|
|
346
|
8538 /* Add the score. */
|
|
351
|
8539 if (sps_flags & (SPS_DOUBLE | SPS_BEST))
|
|
474
|
8540 vim_snprintf((char *)IObuff, IOSIZE, " (%s%d - %d)",
|
|
344
|
8541 stp->st_salscore ? "s " : "",
|
|
|
8542 stp->st_score, stp->st_altscore);
|
|
|
8543 else
|
|
474
|
8544 vim_snprintf((char *)IObuff, IOSIZE, " (%d)",
|
|
346
|
8545 stp->st_score);
|
|
474
|
8546 #ifdef FEAT_RIGHTLEFT
|
|
|
8547 if (cmdmsg_rl)
|
|
|
8548 /* Mirror the numbers, but keep the leading space. */
|
|
|
8549 rl_mirror(IObuff + 1);
|
|
|
8550 #endif
|
|
346
|
8551 msg_advance(30);
|
|
|
8552 msg_puts(IObuff);
|
|
344
|
8553 }
|
|
323
|
8554 msg_putchar('\n');
|
|
|
8555 }
|
|
|
8556
|
|
474
|
8557 #ifdef FEAT_RIGHTLEFT
|
|
|
8558 cmdmsg_rl = FALSE;
|
|
|
8559 msg_col = 0;
|
|
|
8560 #endif
|
|
323
|
8561 /* Ask for choice. */
|
|
485
|
8562 selected = prompt_for_number(&mouse_used);
|
|
372
|
8563 if (mouse_used)
|
|
485
|
8564 selected -= lines_left;
|
|
|
8565 }
|
|
|
8566
|
|
|
8567 if (selected > 0 && selected <= sug.su_ga.ga_len && u_save_cursor() == OK)
|
|
|
8568 {
|
|
|
8569 /* Save the from and to text for :spellrepall. */
|
|
|
8570 stp = &SUG(sug.su_ga, selected - 1);
|
|
548
|
8571 if (sug.su_badlen > stp->st_orglen)
|
|
|
8572 {
|
|
|
8573 /* Replacing less than "su_badlen", append the remainder to
|
|
|
8574 * repl_to. */
|
|
|
8575 repl_from = vim_strnsave(sug.su_badptr, sug.su_badlen);
|
|
|
8576 vim_snprintf((char *)IObuff, IOSIZE, "%s%.*s", stp->st_word,
|
|
|
8577 sug.su_badlen - stp->st_orglen,
|
|
|
8578 sug.su_badptr + stp->st_orglen);
|
|
|
8579 repl_to = vim_strsave(IObuff);
|
|
|
8580 }
|
|
|
8581 else
|
|
|
8582 {
|
|
|
8583 /* Replacing su_badlen or more, use the whole word. */
|
|
|
8584 repl_from = vim_strnsave(sug.su_badptr, stp->st_orglen);
|
|
|
8585 repl_to = vim_strsave(stp->st_word);
|
|
|
8586 }
|
|
485
|
8587
|
|
|
8588 /* Replace the word. */
|
|
|
8589 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
|
8590 if (p != NULL)
|
|
|
8591 {
|
|
|
8592 c = sug.su_badptr - line;
|
|
|
8593 mch_memmove(p, line, c);
|
|
|
8594 STRCPY(p + c, stp->st_word);
|
|
|
8595 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
|
8596 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
8597 curwin->w_cursor.col = c;
|
|
|
8598 changed_bytes(curwin->w_cursor.lnum, c);
|
|
|
8599
|
|
|
8600 /* For redo we use a change-word command. */
|
|
|
8601 ResetRedobuff();
|
|
|
8602 AppendToRedobuff((char_u *)"ciw");
|
|
620
|
8603 AppendToRedobuffLit(p + c,
|
|
|
8604 STRLEN(stp->st_word) + sug.su_badlen - stp->st_orglen);
|
|
485
|
8605 AppendCharToRedobuff(ESC);
|
|
|
8606 }
|
|
|
8607 }
|
|
|
8608 else
|
|
|
8609 curwin->w_cursor = prev_cursor;
|
|
323
|
8610
|
|
344
|
8611 spell_find_cleanup(&sug);
|
|
|
8612 }
|
|
|
8613
|
|
|
8614 /*
|
|
475
|
8615 * Check if the word at line "lnum" column "col" is required to start with a
|
|
|
8616 * capital. This uses 'spellcapcheck' of the current buffer.
|
|
|
8617 */
|
|
|
8618 static int
|
|
|
8619 check_need_cap(lnum, col)
|
|
|
8620 linenr_T lnum;
|
|
|
8621 colnr_T col;
|
|
|
8622 {
|
|
|
8623 int need_cap = FALSE;
|
|
|
8624 char_u *line;
|
|
|
8625 char_u *line_copy = NULL;
|
|
|
8626 char_u *p;
|
|
|
8627 colnr_T endcol;
|
|
|
8628 regmatch_T regmatch;
|
|
|
8629
|
|
|
8630 if (curbuf->b_cap_prog == NULL)
|
|
|
8631 return FALSE;
|
|
|
8632
|
|
|
8633 line = ml_get_curline();
|
|
|
8634 endcol = 0;
|
|
|
8635 if ((int)(skipwhite(line) - line) >= (int)col)
|
|
|
8636 {
|
|
|
8637 /* At start of line, check if previous line is empty or sentence
|
|
|
8638 * ends there. */
|
|
|
8639 if (lnum == 1)
|
|
|
8640 need_cap = TRUE;
|
|
|
8641 else
|
|
|
8642 {
|
|
|
8643 line = ml_get(lnum - 1);
|
|
|
8644 if (*skipwhite(line) == NUL)
|
|
|
8645 need_cap = TRUE;
|
|
|
8646 else
|
|
|
8647 {
|
|
|
8648 /* Append a space in place of the line break. */
|
|
|
8649 line_copy = concat_str(line, (char_u *)" ");
|
|
|
8650 line = line_copy;
|
|
|
8651 endcol = STRLEN(line);
|
|
|
8652 }
|
|
|
8653 }
|
|
|
8654 }
|
|
|
8655 else
|
|
|
8656 endcol = col;
|
|
|
8657
|
|
|
8658 if (endcol > 0)
|
|
|
8659 {
|
|
|
8660 /* Check if sentence ends before the bad word. */
|
|
|
8661 regmatch.regprog = curbuf->b_cap_prog;
|
|
|
8662 regmatch.rm_ic = FALSE;
|
|
|
8663 p = line + endcol;
|
|
|
8664 for (;;)
|
|
|
8665 {
|
|
|
8666 mb_ptr_back(line, p);
|
|
|
8667 if (p == line || spell_iswordp_nmw(p))
|
|
|
8668 break;
|
|
|
8669 if (vim_regexec(®match, p, 0)
|
|
|
8670 && regmatch.endp[0] == line + endcol)
|
|
|
8671 {
|
|
|
8672 need_cap = TRUE;
|
|
|
8673 break;
|
|
|
8674 }
|
|
|
8675 }
|
|
|
8676 }
|
|
|
8677
|
|
|
8678 vim_free(line_copy);
|
|
|
8679
|
|
|
8680 return need_cap;
|
|
|
8681 }
|
|
|
8682
|
|
|
8683
|
|
|
8684 /*
|
|
372
|
8685 * ":spellrepall"
|
|
|
8686 */
|
|
|
8687 /*ARGSUSED*/
|
|
|
8688 void
|
|
|
8689 ex_spellrepall(eap)
|
|
|
8690 exarg_T *eap;
|
|
|
8691 {
|
|
|
8692 pos_T pos = curwin->w_cursor;
|
|
|
8693 char_u *frompat;
|
|
|
8694 int addlen;
|
|
|
8695 char_u *line;
|
|
|
8696 char_u *p;
|
|
|
8697 int save_ws = p_ws;
|
|
483
|
8698 linenr_T prev_lnum = 0;
|
|
372
|
8699
|
|
|
8700 if (repl_from == NULL || repl_to == NULL)
|
|
|
8701 {
|
|
|
8702 EMSG(_("E752: No previous spell replacement"));
|
|
|
8703 return;
|
|
|
8704 }
|
|
|
8705 addlen = STRLEN(repl_to) - STRLEN(repl_from);
|
|
|
8706
|
|
|
8707 frompat = alloc(STRLEN(repl_from) + 7);
|
|
|
8708 if (frompat == NULL)
|
|
|
8709 return;
|
|
|
8710 sprintf((char *)frompat, "\\V\\<%s\\>", repl_from);
|
|
|
8711 p_ws = FALSE;
|
|
|
8712
|
|
483
|
8713 sub_nsubs = 0;
|
|
|
8714 sub_nlines = 0;
|
|
372
|
8715 curwin->w_cursor.lnum = 0;
|
|
|
8716 while (!got_int)
|
|
|
8717 {
|
|
|
8718 if (do_search(NULL, '/', frompat, 1L, SEARCH_KEEP) == 0
|
|
|
8719 || u_save_cursor() == FAIL)
|
|
|
8720 break;
|
|
|
8721
|
|
|
8722 /* Only replace when the right word isn't there yet. This happens
|
|
|
8723 * when changing "etc" to "etc.". */
|
|
|
8724 line = ml_get_curline();
|
|
|
8725 if (addlen <= 0 || STRNCMP(line + curwin->w_cursor.col,
|
|
|
8726 repl_to, STRLEN(repl_to)) != 0)
|
|
|
8727 {
|
|
|
8728 p = alloc(STRLEN(line) + addlen + 1);
|
|
|
8729 if (p == NULL)
|
|
|
8730 break;
|
|
|
8731 mch_memmove(p, line, curwin->w_cursor.col);
|
|
|
8732 STRCPY(p + curwin->w_cursor.col, repl_to);
|
|
|
8733 STRCAT(p, line + curwin->w_cursor.col + STRLEN(repl_from));
|
|
|
8734 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
|
8735 changed_bytes(curwin->w_cursor.lnum, curwin->w_cursor.col);
|
|
483
|
8736
|
|
|
8737 if (curwin->w_cursor.lnum != prev_lnum)
|
|
|
8738 {
|
|
|
8739 ++sub_nlines;
|
|
|
8740 prev_lnum = curwin->w_cursor.lnum;
|
|
|
8741 }
|
|
|
8742 ++sub_nsubs;
|
|
372
|
8743 }
|
|
|
8744 curwin->w_cursor.col += STRLEN(repl_to);
|
|
|
8745 }
|
|
|
8746
|
|
|
8747 p_ws = save_ws;
|
|
|
8748 curwin->w_cursor = pos;
|
|
|
8749 vim_free(frompat);
|
|
|
8750
|
|
483
|
8751 if (sub_nsubs == 0)
|
|
372
|
8752 EMSG2(_("E753: Not found: %s"), repl_from);
|
|
483
|
8753 else
|
|
|
8754 do_sub_msg(FALSE);
|
|
372
|
8755 }
|
|
|
8756
|
|
|
8757 /*
|
|
344
|
8758 * Find spell suggestions for "word". Return them in the growarray "*gap" as
|
|
|
8759 * a list of allocated strings.
|
|
|
8760 */
|
|
|
8761 void
|
|
475
|
8762 spell_suggest_list(gap, word, maxcount, need_cap)
|
|
344
|
8763 garray_T *gap;
|
|
|
8764 char_u *word;
|
|
|
8765 int maxcount; /* maximum nr of suggestions */
|
|
475
|
8766 int need_cap; /* 'spellcapcheck' matched */
|
|
344
|
8767 {
|
|
|
8768 suginfo_T sug;
|
|
|
8769 int i;
|
|
|
8770 suggest_T *stp;
|
|
|
8771 char_u *wcopy;
|
|
|
8772
|
|
475
|
8773 spell_find_suggest(word, &sug, maxcount, FALSE, need_cap);
|
|
344
|
8774
|
|
|
8775 /* Make room in "gap". */
|
|
|
8776 ga_init2(gap, sizeof(char_u *), sug.su_ga.ga_len + 1);
|
|
|
8777 if (ga_grow(gap, sug.su_ga.ga_len) == FAIL)
|
|
|
8778 return;
|
|
|
8779
|
|
323
|
8780 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
344
|
8781 {
|
|
|
8782 stp = &SUG(sug.su_ga, i);
|
|
|
8783
|
|
|
8784 /* The suggested word may replace only part of "word", add the not
|
|
|
8785 * replaced part. */
|
|
|
8786 wcopy = alloc(STRLEN(stp->st_word)
|
|
|
8787 + STRLEN(sug.su_badptr + stp->st_orglen) + 1);
|
|
|
8788 if (wcopy == NULL)
|
|
|
8789 break;
|
|
|
8790 STRCPY(wcopy, stp->st_word);
|
|
|
8791 STRCAT(wcopy, sug.su_badptr + stp->st_orglen);
|
|
|
8792 ((char_u **)gap->ga_data)[gap->ga_len++] = wcopy;
|
|
|
8793 }
|
|
|
8794
|
|
|
8795 spell_find_cleanup(&sug);
|
|
|
8796 }
|
|
|
8797
|
|
|
8798 /*
|
|
|
8799 * Find spell suggestions for the word at the start of "badptr".
|
|
|
8800 * Return the suggestions in "su->su_ga".
|
|
|
8801 * The maximum number of suggestions is "maxcount".
|
|
|
8802 * Note: does use info for the current window.
|
|
|
8803 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
|
8804 */
|
|
|
8805 static void
|
|
388
|
8806 spell_find_suggest(badptr, su, maxcount, banbadword, need_cap)
|
|
344
|
8807 char_u *badptr;
|
|
|
8808 suginfo_T *su;
|
|
|
8809 int maxcount;
|
|
358
|
8810 int banbadword; /* don't include badword in suggestions */
|
|
388
|
8811 int need_cap; /* word should start with capital */
|
|
344
|
8812 {
|
|
534
|
8813 hlf_T attr = HLF_COUNT;
|
|
372
|
8814 char_u buf[MAXPATHL];
|
|
|
8815 char_u *p;
|
|
|
8816 int do_combine = FALSE;
|
|
|
8817 char_u *sps_copy;
|
|
|
8818 #ifdef FEAT_EVAL
|
|
|
8819 static int expr_busy = FALSE;
|
|
|
8820 #endif
|
|
385
|
8821 int c;
|
|
503
|
8822 int i;
|
|
|
8823 langp_T *lp;
|
|
344
|
8824
|
|
|
8825 /*
|
|
|
8826 * Set the info in "*su".
|
|
|
8827 */
|
|
|
8828 vim_memset(su, 0, sizeof(suginfo_T));
|
|
|
8829 ga_init2(&su->su_ga, (int)sizeof(suggest_T), 10);
|
|
|
8830 ga_init2(&su->su_sga, (int)sizeof(suggest_T), 10);
|
|
355
|
8831 if (*badptr == NUL)
|
|
|
8832 return;
|
|
344
|
8833 hash_init(&su->su_banned);
|
|
|
8834
|
|
|
8835 su->su_badptr = badptr;
|
|
385
|
8836 su->su_badlen = spell_check(curwin, su->su_badptr, &attr, NULL);
|
|
344
|
8837 su->su_maxcount = maxcount;
|
|
372
|
8838 su->su_maxscore = SCORE_MAXINIT;
|
|
344
|
8839
|
|
|
8840 if (su->su_badlen >= MAXWLEN)
|
|
|
8841 su->su_badlen = MAXWLEN - 1; /* just in case */
|
|
|
8842 vim_strncpy(su->su_badword, su->su_badptr, su->su_badlen);
|
|
|
8843 (void)spell_casefold(su->su_badptr, su->su_badlen,
|
|
|
8844 su->su_fbadword, MAXWLEN);
|
|
346
|
8845 /* get caps flags for bad word */
|
|
474
|
8846 su->su_badflags = badword_captype(su->su_badptr,
|
|
|
8847 su->su_badptr + su->su_badlen);
|
|
388
|
8848 if (need_cap)
|
|
|
8849 su->su_badflags |= WF_ONECAP;
|
|
344
|
8850
|
|
503
|
8851 /* Find the default language for sound folding. We simply use the first
|
|
|
8852 * one in 'spelllang' that supports sound folding. That's good for when
|
|
|
8853 * using multiple files for one language, it's not that bad when mixing
|
|
|
8854 * languages (e.g., "pl,en"). */
|
|
|
8855 for (i = 0; i < curbuf->b_langp.ga_len; ++i)
|
|
|
8856 {
|
|
|
8857 lp = LANGP_ENTRY(curbuf->b_langp, i);
|
|
|
8858 if (lp->lp_sallang != NULL)
|
|
|
8859 {
|
|
|
8860 su->su_sallang = lp->lp_sallang;
|
|
|
8861 break;
|
|
|
8862 }
|
|
|
8863 }
|
|
|
8864
|
|
534
|
8865 /* Soundfold the bad word with the default sound folding, so that we don't
|
|
|
8866 * have to do this many times. */
|
|
|
8867 if (su->su_sallang != NULL)
|
|
|
8868 spell_soundfold(su->su_sallang, su->su_fbadword, TRUE,
|
|
|
8869 su->su_sal_badword);
|
|
|
8870
|
|
385
|
8871 /* If the word is not capitalised and spell_check() doesn't consider the
|
|
|
8872 * word to be bad then it might need to be capitalised. Add a suggestion
|
|
|
8873 * for that. */
|
|
455
|
8874 c = PTR2CHAR(su->su_badptr);
|
|
534
|
8875 if (!SPELL_ISUPPER(c) && attr == HLF_COUNT)
|
|
385
|
8876 {
|
|
|
8877 make_case_word(su->su_badword, buf, WF_ONECAP);
|
|
|
8878 add_suggestion(su, &su->su_ga, buf, su->su_badlen, SCORE_ICASE,
|
|
503
|
8879 0, TRUE, su->su_sallang);
|
|
385
|
8880 }
|
|
|
8881
|
|
344
|
8882 /* Ban the bad word itself. It may appear in another region. */
|
|
358
|
8883 if (banbadword)
|
|
|
8884 add_banned(su, su->su_badword);
|
|
344
|
8885
|
|
372
|
8886 /* Make a copy of 'spellsuggest', because the expression may change it. */
|
|
|
8887 sps_copy = vim_strsave(p_sps);
|
|
|
8888 if (sps_copy == NULL)
|
|
|
8889 return;
|
|
|
8890
|
|
|
8891 /* Loop over the items in 'spellsuggest'. */
|
|
|
8892 for (p = sps_copy; *p != NUL; )
|
|
|
8893 {
|
|
|
8894 copy_option_part(&p, buf, MAXPATHL, ",");
|
|
|
8895
|
|
|
8896 if (STRNCMP(buf, "expr:", 5) == 0)
|
|
|
8897 {
|
|
|
8898 #ifdef FEAT_EVAL
|
|
375
|
8899 /* Evaluate an expression. Skip this when called recursively,
|
|
|
8900 * when using spellsuggest() in the expression. */
|
|
372
|
8901 if (!expr_busy)
|
|
|
8902 {
|
|
|
8903 expr_busy = TRUE;
|
|
|
8904 spell_suggest_expr(su, buf + 5);
|
|
|
8905 expr_busy = FALSE;
|
|
|
8906 }
|
|
|
8907 #endif
|
|
|
8908 }
|
|
|
8909 else if (STRNCMP(buf, "file:", 5) == 0)
|
|
|
8910 /* Use list of suggestions in a file. */
|
|
|
8911 spell_suggest_file(su, buf + 5);
|
|
|
8912 else
|
|
|
8913 {
|
|
|
8914 /* Use internal method. */
|
|
|
8915 spell_suggest_intern(su);
|
|
|
8916 if (sps_flags & SPS_DOUBLE)
|
|
|
8917 do_combine = TRUE;
|
|
|
8918 }
|
|
|
8919 }
|
|
|
8920
|
|
|
8921 vim_free(sps_copy);
|
|
|
8922
|
|
|
8923 if (do_combine)
|
|
|
8924 /* Combine the two list of suggestions. This must be done last,
|
|
|
8925 * because sorting changes the order again. */
|
|
|
8926 score_combine(su);
|
|
|
8927 }
|
|
|
8928
|
|
|
8929 #ifdef FEAT_EVAL
|
|
|
8930 /*
|
|
|
8931 * Find suggestions by evaluating expression "expr".
|
|
|
8932 */
|
|
|
8933 static void
|
|
|
8934 spell_suggest_expr(su, expr)
|
|
|
8935 suginfo_T *su;
|
|
|
8936 char_u *expr;
|
|
|
8937 {
|
|
|
8938 list_T *list;
|
|
|
8939 listitem_T *li;
|
|
|
8940 int score;
|
|
|
8941 char_u *p;
|
|
|
8942
|
|
|
8943 /* The work is split up in a few parts to avoid having to export
|
|
|
8944 * suginfo_T.
|
|
|
8945 * First evaluate the expression and get the resulting list. */
|
|
|
8946 list = eval_spell_expr(su->su_badword, expr);
|
|
|
8947 if (list != NULL)
|
|
|
8948 {
|
|
|
8949 /* Loop over the items in the list. */
|
|
|
8950 for (li = list->lv_first; li != NULL; li = li->li_next)
|
|
|
8951 if (li->li_tv.v_type == VAR_LIST)
|
|
|
8952 {
|
|
|
8953 /* Get the word and the score from the items. */
|
|
|
8954 score = get_spellword(li->li_tv.vval.v_list, &p);
|
|
|
8955 if (score >= 0)
|
|
|
8956 add_suggestion(su, &su->su_ga, p,
|
|
503
|
8957 su->su_badlen, score, 0, TRUE, su->su_sallang);
|
|
372
|
8958 }
|
|
|
8959 list_unref(list);
|
|
|
8960 }
|
|
|
8961
|
|
|
8962 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
8963 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
8964 }
|
|
|
8965 #endif
|
|
|
8966
|
|
|
8967 /*
|
|
474
|
8968 * Find suggestions in file "fname". Used for "file:" in 'spellsuggest'.
|
|
372
|
8969 */
|
|
|
8970 static void
|
|
|
8971 spell_suggest_file(su, fname)
|
|
|
8972 suginfo_T *su;
|
|
|
8973 char_u *fname;
|
|
|
8974 {
|
|
|
8975 FILE *fd;
|
|
|
8976 char_u line[MAXWLEN * 2];
|
|
|
8977 char_u *p;
|
|
|
8978 int len;
|
|
|
8979 char_u cword[MAXWLEN];
|
|
|
8980
|
|
|
8981 /* Open the file. */
|
|
|
8982 fd = mch_fopen((char *)fname, "r");
|
|
|
8983 if (fd == NULL)
|
|
|
8984 {
|
|
|
8985 EMSG2(_(e_notopen), fname);
|
|
|
8986 return;
|
|
|
8987 }
|
|
|
8988
|
|
|
8989 /* Read it line by line. */
|
|
|
8990 while (!vim_fgets(line, MAXWLEN * 2, fd) && !got_int)
|
|
|
8991 {
|
|
|
8992 line_breakcheck();
|
|
|
8993
|
|
|
8994 p = vim_strchr(line, '/');
|
|
|
8995 if (p == NULL)
|
|
|
8996 continue; /* No Tab found, just skip the line. */
|
|
|
8997 *p++ = NUL;
|
|
|
8998 if (STRICMP(su->su_badword, line) == 0)
|
|
|
8999 {
|
|
|
9000 /* Match! Isolate the good word, until CR or NL. */
|
|
|
9001 for (len = 0; p[len] >= ' '; ++len)
|
|
|
9002 ;
|
|
|
9003 p[len] = NUL;
|
|
|
9004
|
|
|
9005 /* If the suggestion doesn't have specific case duplicate the case
|
|
|
9006 * of the bad word. */
|
|
|
9007 if (captype(p, NULL) == 0)
|
|
|
9008 {
|
|
|
9009 make_case_word(p, cword, su->su_badflags);
|
|
|
9010 p = cword;
|
|
|
9011 }
|
|
|
9012
|
|
|
9013 add_suggestion(su, &su->su_ga, p, su->su_badlen,
|
|
503
|
9014 SCORE_FILE, 0, TRUE, su->su_sallang);
|
|
372
|
9015 }
|
|
|
9016 }
|
|
|
9017
|
|
|
9018 fclose(fd);
|
|
|
9019
|
|
|
9020 /* Sort the suggestions and truncate at "maxcount". */
|
|
|
9021 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
9022 }
|
|
|
9023
|
|
|
9024 /*
|
|
|
9025 * Find suggestions for the internal method indicated by "sps_flags".
|
|
|
9026 */
|
|
|
9027 static void
|
|
|
9028 spell_suggest_intern(su)
|
|
|
9029 suginfo_T *su;
|
|
|
9030 {
|
|
344
|
9031 /*
|
|
346
|
9032 * 1. Try special cases, such as repeating a word: "the the" -> "the".
|
|
344
|
9033 *
|
|
|
9034 * Set a maximum score to limit the combination of operations that is
|
|
|
9035 * tried.
|
|
|
9036 */
|
|
346
|
9037 suggest_try_special(su);
|
|
|
9038
|
|
|
9039 /*
|
|
|
9040 * 2. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
|
9041 * from the .aff file and inserting a space (split the word).
|
|
|
9042 */
|
|
|
9043 suggest_try_change(su);
|
|
344
|
9044
|
|
|
9045 /* For the resulting top-scorers compute the sound-a-like score. */
|
|
|
9046 if (sps_flags & SPS_DOUBLE)
|
|
|
9047 score_comp_sal(su);
|
|
|
9048
|
|
|
9049 /*
|
|
346
|
9050 * 3. Try finding sound-a-like words.
|
|
344
|
9051 *
|
|
|
9052 * Only do this when we don't have a lot of suggestions yet, because it's
|
|
|
9053 * very slow and often doesn't find new suggestions.
|
|
|
9054 */
|
|
|
9055 if ((sps_flags & SPS_DOUBLE)
|
|
|
9056 || (!(sps_flags & SPS_FAST)
|
|
|
9057 && su->su_ga.ga_len < SUG_CLEAN_COUNT(su)))
|
|
|
9058 {
|
|
|
9059 /* Allow a higher score now. */
|
|
|
9060 su->su_maxscore = SCORE_MAXMAX;
|
|
346
|
9061 suggest_try_soundalike(su);
|
|
344
|
9062 }
|
|
|
9063
|
|
|
9064 /* When CTRL-C was hit while searching do show the results. */
|
|
|
9065 ui_breakcheck();
|
|
|
9066 if (got_int)
|
|
|
9067 {
|
|
|
9068 (void)vgetc();
|
|
|
9069 got_int = FALSE;
|
|
|
9070 }
|
|
|
9071
|
|
372
|
9072 if ((sps_flags & SPS_DOUBLE) == 0 && su->su_ga.ga_len != 0)
|
|
344
|
9073 {
|
|
|
9074 if (sps_flags & SPS_BEST)
|
|
|
9075 /* Adjust the word score for how it sounds like. */
|
|
|
9076 rescore_suggestions(su);
|
|
|
9077
|
|
|
9078 /* Sort the suggestions and truncate at "maxcount". */
|
|
372
|
9079 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
344
|
9080 }
|
|
|
9081 }
|
|
|
9082
|
|
|
9083 /*
|
|
|
9084 * Free the info put in "*su" by spell_find_suggest().
|
|
|
9085 */
|
|
|
9086 static void
|
|
|
9087 spell_find_cleanup(su)
|
|
|
9088 suginfo_T *su;
|
|
|
9089 {
|
|
|
9090 int i;
|
|
|
9091
|
|
|
9092 /* Free the suggestions. */
|
|
|
9093 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
9094 vim_free(SUG(su->su_ga, i).st_word);
|
|
|
9095 ga_clear(&su->su_ga);
|
|
|
9096 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
9097 vim_free(SUG(su->su_sga, i).st_word);
|
|
|
9098 ga_clear(&su->su_sga);
|
|
323
|
9099
|
|
|
9100 /* Free the banned words. */
|
|
344
|
9101 free_banned(su);
|
|
323
|
9102 }
|
|
|
9103
|
|
|
9104 /*
|
|
324
|
9105 * Make a copy of "word", with the first letter upper or lower cased, to
|
|
|
9106 * "wcopy[MAXWLEN]". "word" must not be empty.
|
|
|
9107 * The result is NUL terminated.
|
|
323
|
9108 */
|
|
|
9109 static void
|
|
324
|
9110 onecap_copy(word, wcopy, upper)
|
|
323
|
9111 char_u *word;
|
|
|
9112 char_u *wcopy;
|
|
|
9113 int upper; /* TRUE: first letter made upper case */
|
|
|
9114 {
|
|
|
9115 char_u *p;
|
|
|
9116 int c;
|
|
|
9117 int l;
|
|
|
9118
|
|
|
9119 p = word;
|
|
|
9120 #ifdef FEAT_MBYTE
|
|
|
9121 if (has_mbyte)
|
|
474
|
9122 c = mb_cptr2char_adv(&p);
|
|
323
|
9123 else
|
|
|
9124 #endif
|
|
|
9125 c = *p++;
|
|
|
9126 if (upper)
|
|
324
|
9127 c = SPELL_TOUPPER(c);
|
|
323
|
9128 else
|
|
324
|
9129 c = SPELL_TOFOLD(c);
|
|
323
|
9130 #ifdef FEAT_MBYTE
|
|
|
9131 if (has_mbyte)
|
|
|
9132 l = mb_char2bytes(c, wcopy);
|
|
|
9133 else
|
|
|
9134 #endif
|
|
|
9135 {
|
|
|
9136 l = 1;
|
|
|
9137 wcopy[0] = c;
|
|
|
9138 }
|
|
376
|
9139 vim_strncpy(wcopy + l, p, MAXWLEN - l - 1);
|
|
323
|
9140 }
|
|
|
9141
|
|
|
9142 /*
|
|
324
|
9143 * Make a copy of "word" with all the letters upper cased into
|
|
|
9144 * "wcopy[MAXWLEN]". The result is NUL terminated.
|
|
323
|
9145 */
|
|
|
9146 static void
|
|
|
9147 allcap_copy(word, wcopy)
|
|
|
9148 char_u *word;
|
|
|
9149 char_u *wcopy;
|
|
|
9150 {
|
|
|
9151 char_u *s;
|
|
|
9152 char_u *d;
|
|
|
9153 int c;
|
|
|
9154
|
|
|
9155 d = wcopy;
|
|
|
9156 for (s = word; *s != NUL; )
|
|
|
9157 {
|
|
|
9158 #ifdef FEAT_MBYTE
|
|
|
9159 if (has_mbyte)
|
|
474
|
9160 c = mb_cptr2char_adv(&s);
|
|
323
|
9161 else
|
|
|
9162 #endif
|
|
|
9163 c = *s++;
|
|
492
|
9164
|
|
|
9165 #ifdef FEAT_MBYTE
|
|
|
9166 /* We only change ß to SS when we are certain latin1 is used. It
|
|
|
9167 * would cause weird errors in other 8-bit encodings. */
|
|
|
9168 if (enc_latin1like && c == 0xdf)
|
|
|
9169 {
|
|
|
9170 c = 'S';
|
|
|
9171 if (d - wcopy >= MAXWLEN - 1)
|
|
|
9172 break;
|
|
|
9173 *d++ = c;
|
|
|
9174 }
|
|
|
9175 else
|
|
|
9176 #endif
|
|
|
9177 c = SPELL_TOUPPER(c);
|
|
323
|
9178
|
|
|
9179 #ifdef FEAT_MBYTE
|
|
|
9180 if (has_mbyte)
|
|
|
9181 {
|
|
|
9182 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
|
9183 break;
|
|
|
9184 d += mb_char2bytes(c, d);
|
|
|
9185 }
|
|
|
9186 else
|
|
|
9187 #endif
|
|
|
9188 {
|
|
|
9189 if (d - wcopy >= MAXWLEN - 1)
|
|
|
9190 break;
|
|
|
9191 *d++ = c;
|
|
|
9192 }
|
|
|
9193 }
|
|
|
9194 *d = NUL;
|
|
|
9195 }
|
|
|
9196
|
|
|
9197 /*
|
|
346
|
9198 * Try finding suggestions by recognizing specific situations.
|
|
|
9199 */
|
|
|
9200 static void
|
|
|
9201 suggest_try_special(su)
|
|
|
9202 suginfo_T *su;
|
|
|
9203 {
|
|
|
9204 char_u *p;
|
|
376
|
9205 size_t len;
|
|
346
|
9206 int c;
|
|
|
9207 char_u word[MAXWLEN];
|
|
|
9208
|
|
|
9209 /*
|
|
|
9210 * Recognize a word that is repeated: "the the".
|
|
|
9211 */
|
|
|
9212 p = skiptowhite(su->su_fbadword);
|
|
|
9213 len = p - su->su_fbadword;
|
|
|
9214 p = skipwhite(p);
|
|
|
9215 if (STRLEN(p) == len && STRNCMP(su->su_fbadword, p, len) == 0)
|
|
|
9216 {
|
|
|
9217 /* Include badflags: if the badword is onecap or allcap
|
|
|
9218 * use that for the goodword too: "The the" -> "The". */
|
|
|
9219 c = su->su_fbadword[len];
|
|
|
9220 su->su_fbadword[len] = NUL;
|
|
|
9221 make_case_word(su->su_fbadword, word, su->su_badflags);
|
|
|
9222 su->su_fbadword[len] = c;
|
|
534
|
9223
|
|
|
9224 /* Give a soundalike score of 0, compute the score as if deleting one
|
|
|
9225 * character. */
|
|
|
9226 add_suggestion(su, &su->su_ga, word, su->su_badlen,
|
|
|
9227 RESCORE(SCORE_REP, 0), 0, TRUE, su->su_sallang);
|
|
346
|
9228 }
|
|
|
9229 }
|
|
|
9230
|
|
|
9231 /*
|
|
323
|
9232 * Try finding suggestions by adding/removing/swapping letters.
|
|
330
|
9233 *
|
|
|
9234 * This uses a state machine. At each node in the tree we try various
|
|
|
9235 * operations. When trying if an operation work "depth" is increased and the
|
|
|
9236 * stack[] is used to store info. This allows combinations, thus insert one
|
|
|
9237 * character, replace one and delete another. The number of changes is
|
|
|
9238 * limited by su->su_maxscore, checked in try_deeper().
|
|
485
|
9239 *
|
|
|
9240 * After implementing this I noticed an article by Kemal Oflazer that
|
|
|
9241 * describes something similar: "Error-tolerant Finite State Recognition with
|
|
|
9242 * Applications to Morphological Analysis and Spelling Correction" (1996).
|
|
|
9243 * The implementation in the article is simplified and requires a stack of
|
|
|
9244 * unknown depth. The implementation here only needs a stack depth of the
|
|
|
9245 * length of the word.
|
|
323
|
9246 */
|
|
|
9247 static void
|
|
346
|
9248 suggest_try_change(su)
|
|
323
|
9249 suginfo_T *su;
|
|
|
9250 {
|
|
|
9251 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
|
9252 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
|
9253 trystate_T stack[MAXWLEN];
|
|
483
|
9254 char_u preword[MAXWLEN * 3]; /* word found with proper case;
|
|
|
9255 * concatanation of prefix compound
|
|
|
9256 * words and split word. NUL terminated
|
|
|
9257 * when going deeper but not when coming
|
|
|
9258 * back. */
|
|
|
9259 char_u compflags[MAXWLEN]; /* compound flags, one for each word */
|
|
323
|
9260 trystate_T *sp;
|
|
|
9261 int newscore;
|
|
|
9262 langp_T *lp;
|
|
375
|
9263 char_u *byts, *fbyts, *pbyts;
|
|
|
9264 idx_T *idxs, *fidxs, *pidxs;
|
|
323
|
9265 int depth;
|
|
330
|
9266 int c, c2, c3;
|
|
390
|
9267 int n;
|
|
323
|
9268 int flags;
|
|
|
9269 garray_T *gap;
|
|
324
|
9270 idx_T arridx;
|
|
323
|
9271 int len;
|
|
|
9272 char_u *p;
|
|
|
9273 fromto_T *ftp;
|
|
330
|
9274 int fl = 0, tl;
|
|
346
|
9275 int repextra = 0; /* extra bytes in fword[] from REP item */
|
|
482
|
9276 slang_T *slang;
|
|
|
9277 int fword_ends;
|
|
500
|
9278 int lpi;
|
|
534
|
9279 int maysplit;
|
|
|
9280 int goodword_ends;
|
|
323
|
9281
|
|
|
9282 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
346
|
9283 * to find matches (esp. REP items). Append some more text, changing
|
|
|
9284 * chars after the bad word may help. */
|
|
323
|
9285 STRCPY(fword, su->su_fbadword);
|
|
346
|
9286 n = STRLEN(fword);
|
|
|
9287 p = su->su_badptr + su->su_badlen;
|
|
|
9288 (void)spell_casefold(p, STRLEN(p), fword + n, MAXWLEN - n);
|
|
323
|
9289
|
|
503
|
9290 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
9291 {
|
|
|
9292 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
482
|
9293 slang = lp->lp_slang;
|
|
|
9294
|
|
500
|
9295 /* If reloading a spell file fails it's still in the list but
|
|
|
9296 * everything has been cleared. */
|
|
|
9297 if (slang->sl_fbyts == NULL)
|
|
|
9298 continue;
|
|
|
9299
|
|
323
|
9300 /*
|
|
|
9301 * Go through the whole case-fold tree, try changes at each node.
|
|
|
9302 * "tword[]" contains the word collected from nodes in the tree.
|
|
|
9303 * "fword[]" the word we are trying to match with (initially the bad
|
|
|
9304 * word).
|
|
|
9305 */
|
|
|
9306 depth = 0;
|
|
375
|
9307 sp = &stack[0];
|
|
483
|
9308 vim_memset(sp, 0, sizeof(trystate_T));
|
|
375
|
9309 sp->ts_curi = 1;
|
|
330
|
9310
|
|
|
9311 /*
|
|
375
|
9312 * When there are postponed prefixes we need to use these first. At
|
|
|
9313 * the end of the prefix we continue in the case-fold tree.
|
|
|
9314 */
|
|
482
|
9315 fbyts = slang->sl_fbyts;
|
|
|
9316 fidxs = slang->sl_fidxs;
|
|
|
9317 pbyts = slang->sl_pbyts;
|
|
|
9318 pidxs = slang->sl_pidxs;
|
|
375
|
9319 if (pbyts != NULL)
|
|
|
9320 {
|
|
|
9321 byts = pbyts;
|
|
|
9322 idxs = pidxs;
|
|
482
|
9323 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
375
|
9324 sp->ts_state = STATE_NOPREFIX; /* try without prefix first */
|
|
|
9325 }
|
|
|
9326 else
|
|
|
9327 {
|
|
|
9328 byts = fbyts;
|
|
|
9329 idxs = fidxs;
|
|
482
|
9330 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
485
|
9331 sp->ts_state = STATE_START;
|
|
375
|
9332 }
|
|
|
9333
|
|
|
9334 /*
|
|
330
|
9335 * Loop to find all suggestions. At each round we either:
|
|
|
9336 * - For the current state try one operation, advance "ts_curi",
|
|
|
9337 * increase "depth".
|
|
|
9338 * - When a state is done go to the next, set "ts_state".
|
|
|
9339 * - When all states are tried decrease "depth".
|
|
|
9340 */
|
|
323
|
9341 while (depth >= 0 && !got_int)
|
|
|
9342 {
|
|
|
9343 sp = &stack[depth];
|
|
|
9344 switch (sp->ts_state)
|
|
|
9345 {
|
|
|
9346 case STATE_START:
|
|
375
|
9347 case STATE_NOPREFIX:
|
|
323
|
9348 /*
|
|
|
9349 * Start of node: Deal with NUL bytes, which means
|
|
|
9350 * tword[] may end here.
|
|
|
9351 */
|
|
|
9352 arridx = sp->ts_arridx; /* current node in the tree */
|
|
|
9353 len = byts[arridx]; /* bytes in this node */
|
|
|
9354 arridx += sp->ts_curi; /* index of current byte */
|
|
|
9355
|
|
482
|
9356 if (sp->ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
9357 {
|
|
|
9358 /* Skip over the NUL bytes, we use them later. */
|
|
|
9359 for (n = 0; n < len && byts[arridx + n] == 0; ++n)
|
|
|
9360 ;
|
|
|
9361 sp->ts_curi += n;
|
|
|
9362
|
|
390
|
9363 /* Always past NUL bytes now. */
|
|
|
9364 n = (int)sp->ts_state;
|
|
|
9365 sp->ts_state = STATE_ENDNUL;
|
|
455
|
9366 sp->ts_save_badflags = su->su_badflags;
|
|
390
|
9367
|
|
375
|
9368 /* At end of a prefix or at start of prefixtree: check for
|
|
|
9369 * following word. */
|
|
390
|
9370 if (byts[arridx] == 0 || n == (int)STATE_NOPREFIX)
|
|
375
|
9371 {
|
|
455
|
9372 /* Set su->su_badflags to the caps type at this
|
|
|
9373 * position. Use the caps type until here for the
|
|
|
9374 * prefix itself. */
|
|
|
9375 #ifdef FEAT_MBYTE
|
|
|
9376 if (has_mbyte)
|
|
|
9377 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
|
9378 else
|
|
|
9379 #endif
|
|
|
9380 n = sp->ts_fidx;
|
|
474
|
9381 flags = badword_captype(su->su_badptr,
|
|
|
9382 su->su_badptr + n);
|
|
|
9383 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
9384 su->su_badptr + su->su_badlen);
|
|
375
|
9385 ++depth;
|
|
|
9386 stack[depth] = stack[depth - 1];
|
|
|
9387 sp = &stack[depth];
|
|
|
9388 sp->ts_prefixdepth = depth - 1;
|
|
|
9389 byts = fbyts;
|
|
|
9390 idxs = fidxs;
|
|
|
9391 sp->ts_state = STATE_START;
|
|
|
9392 sp->ts_curi = 1; /* start just after length byte */
|
|
|
9393 sp->ts_arridx = 0;
|
|
|
9394
|
|
455
|
9395 /* Move the prefix to preword[] with the right case
|
|
|
9396 * and make find_keepcap_word() works. */
|
|
485
|
9397 tword[sp->ts_twordlen] = NUL;
|
|
|
9398 make_case_word(tword + sp->ts_splitoff,
|
|
|
9399 preword + sp->ts_prewordlen,
|
|
|
9400 flags);
|
|
|
9401 sp->ts_prewordlen = STRLEN(preword);
|
|
483
|
9402 sp->ts_splitoff = sp->ts_twordlen;
|
|
375
|
9403 }
|
|
|
9404 break;
|
|
|
9405 }
|
|
|
9406
|
|
346
|
9407 if (sp->ts_curi > len || byts[arridx] != 0)
|
|
323
|
9408 {
|
|
|
9409 /* Past bytes in node and/or past NUL bytes. */
|
|
|
9410 sp->ts_state = STATE_ENDNUL;
|
|
455
|
9411 sp->ts_save_badflags = su->su_badflags;
|
|
323
|
9412 break;
|
|
|
9413 }
|
|
|
9414
|
|
|
9415 /*
|
|
|
9416 * End of word in tree.
|
|
|
9417 */
|
|
|
9418 ++sp->ts_curi; /* eat one NUL byte */
|
|
|
9419
|
|
324
|
9420 flags = (int)idxs[arridx];
|
|
483
|
9421 fword_ends = (fword[sp->ts_fidx] == NUL
|
|
|
9422 || !spell_iswordp(fword + sp->ts_fidx, curbuf));
|
|
|
9423 tword[sp->ts_twordlen] = NUL;
|
|
323
|
9424
|
|
485
|
9425 if (sp->ts_prefixdepth <= PFD_NOTSPECIAL
|
|
|
9426 && (sp->ts_flags & TSF_PREFIXOK) == 0)
|
|
375
|
9427 {
|
|
|
9428 /* There was a prefix before the word. Check that the
|
|
|
9429 * prefix can be used with this word. */
|
|
|
9430 /* Count the length of the NULs in the prefix. If there
|
|
|
9431 * are none this must be the first try without a prefix.
|
|
|
9432 */
|
|
|
9433 n = stack[sp->ts_prefixdepth].ts_arridx;
|
|
|
9434 len = pbyts[n++];
|
|
|
9435 for (c = 0; c < len && pbyts[n + c] == 0; ++c)
|
|
|
9436 ;
|
|
|
9437 if (c > 0)
|
|
|
9438 {
|
|
390
|
9439 c = valid_word_prefix(c, n, flags,
|
|
483
|
9440 tword + sp->ts_splitoff, slang, FALSE);
|
|
375
|
9441 if (c == 0)
|
|
|
9442 break;
|
|
|
9443
|
|
|
9444 /* Use the WF_RARE flag for a rare prefix. */
|
|
|
9445 if (c & WF_RAREPFX)
|
|
|
9446 flags |= WF_RARE;
|
|
485
|
9447
|
|
|
9448 /* Tricky: when checking for both prefix and
|
|
|
9449 * compounding we run into the prefix flag first.
|
|
|
9450 * Remember that it's OK, so that we accept the prefix
|
|
|
9451 * when arriving at a compound flag. */
|
|
|
9452 sp->ts_flags |= TSF_PREFIXOK;
|
|
375
|
9453 }
|
|
|
9454 }
|
|
|
9455
|
|
534
|
9456 /* Check NEEDCOMPOUND: can't use word without compounding. Do
|
|
|
9457 * try appending another compound word below. */
|
|
500
|
9458 if (sp->ts_complen == sp->ts_compsplit && fword_ends
|
|
|
9459 && (flags & WF_NEEDCOMP))
|
|
534
|
9460 goodword_ends = FALSE;
|
|
|
9461 else
|
|
|
9462 goodword_ends = TRUE;
|
|
500
|
9463
|
|
485
|
9464 if (sp->ts_complen > sp->ts_compsplit)
|
|
|
9465 {
|
|
492
|
9466 if (slang->sl_nobreak)
|
|
|
9467 {
|
|
|
9468 /* There was a word before this word. When there was
|
|
|
9469 * no change in this word (it was correct) add the
|
|
|
9470 * first word as a suggestion. If this word was
|
|
|
9471 * corrected too, we need to check if a correct word
|
|
|
9472 * follows. */
|
|
|
9473 if (sp->ts_fidx - sp->ts_splitfidx
|
|
|
9474 == sp->ts_twordlen - sp->ts_splitoff
|
|
|
9475 && STRNCMP(fword + sp->ts_splitfidx,
|
|
|
9476 tword + sp->ts_splitoff,
|
|
|
9477 sp->ts_fidx - sp->ts_splitfidx) == 0)
|
|
|
9478 {
|
|
|
9479 preword[sp->ts_prewordlen] = NUL;
|
|
|
9480 add_suggestion(su, &su->su_ga, preword,
|
|
503
|
9481 sp->ts_splitfidx - repextra,
|
|
|
9482 sp->ts_score, 0, FALSE,
|
|
|
9483 lp->lp_sallang);
|
|
492
|
9484 break;
|
|
|
9485 }
|
|
|
9486 }
|
|
|
9487 else
|
|
|
9488 {
|
|
|
9489 /* There was a compound word before this word. If
|
|
|
9490 * this word does not support compounding then give up
|
|
|
9491 * (splitting is tried for the word without compound
|
|
|
9492 * flag). */
|
|
|
9493 if (((unsigned)flags >> 24) == 0
|
|
|
9494 || sp->ts_twordlen - sp->ts_splitoff
|
|
485
|
9495 < slang->sl_compminlen)
|
|
492
|
9496 break;
|
|
500
|
9497 #ifdef FEAT_MBYTE
|
|
|
9498 /* For multi-byte chars check character length against
|
|
|
9499 * COMPOUNDMIN. */
|
|
|
9500 if (has_mbyte
|
|
501
|
9501 && slang->sl_compminlen > 0
|
|
500
|
9502 && mb_charlen(tword + sp->ts_splitoff)
|
|
|
9503 < slang->sl_compminlen)
|
|
|
9504 break;
|
|
|
9505 #endif
|
|
|
9506
|
|
492
|
9507 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
9508 compflags[sp->ts_complen + 1] = NUL;
|
|
|
9509 vim_strncpy(preword + sp->ts_prewordlen,
|
|
|
9510 tword + sp->ts_splitoff,
|
|
|
9511 sp->ts_twordlen - sp->ts_splitoff);
|
|
|
9512 p = preword;
|
|
|
9513 while (*skiptowhite(p) != NUL)
|
|
|
9514 p = skipwhite(skiptowhite(p));
|
|
|
9515 if (fword_ends && !can_compound(slang, p,
|
|
485
|
9516 compflags + sp->ts_compsplit))
|
|
492
|
9517 break;
|
|
|
9518
|
|
|
9519 /* Get pointer to last char of previous word. */
|
|
|
9520 p = preword + sp->ts_prewordlen;
|
|
|
9521 mb_ptr_back(preword, p);
|
|
|
9522 }
|
|
490
|
9523 }
|
|
|
9524 else
|
|
|
9525 p = NULL;
|
|
485
|
9526
|
|
323
|
9527 /*
|
|
|
9528 * Form the word with proper case in preword.
|
|
|
9529 * If there is a word from a previous split, append.
|
|
|
9530 */
|
|
|
9531 if (flags & WF_KEEPCAP)
|
|
|
9532 /* Must find the word in the keep-case tree. */
|
|
483
|
9533 find_keepcap_word(slang, tword + sp->ts_splitoff,
|
|
|
9534 preword + sp->ts_prewordlen);
|
|
323
|
9535 else
|
|
346
|
9536 {
|
|
323
|
9537 /* Include badflags: if the badword is onecap or allcap
|
|
346
|
9538 * use that for the goodword too. But if the badword is
|
|
|
9539 * allcap and it's only one char long use onecap. */
|
|
|
9540 c = su->su_badflags;
|
|
|
9541 if ((c & WF_ALLCAP)
|
|
|
9542 #ifdef FEAT_MBYTE
|
|
474
|
9543 && su->su_badlen == (*mb_ptr2len)(su->su_badptr)
|
|
346
|
9544 #else
|
|
|
9545 && su->su_badlen == 1
|
|
|
9546 #endif
|
|
|
9547 )
|
|
|
9548 c = WF_ONECAP;
|
|
490
|
9549 c |= flags;
|
|
|
9550
|
|
|
9551 /* When appending a compound word after a word character
|
|
|
9552 * don't use Onecap. */
|
|
|
9553 if (p != NULL && spell_iswordp_nmw(p))
|
|
|
9554 c &= ~WF_ONECAP;
|
|
483
|
9555 make_case_word(tword + sp->ts_splitoff,
|
|
490
|
9556 preword + sp->ts_prewordlen, c);
|
|
346
|
9557 }
|
|
323
|
9558
|
|
|
9559 /* Don't use a banned word. It may appear again as a good
|
|
|
9560 * word, thus remember it. */
|
|
|
9561 if (flags & WF_BANNED)
|
|
|
9562 {
|
|
483
|
9563 add_banned(su, preword + sp->ts_prewordlen);
|
|
323
|
9564 break;
|
|
|
9565 }
|
|
534
|
9566 if ((sp->ts_complen == sp->ts_compsplit
|
|
|
9567 && was_banned(su, preword + sp->ts_prewordlen))
|
|
372
|
9568 || was_banned(su, preword))
|
|
534
|
9569 {
|
|
|
9570 if (slang->sl_compprog == NULL)
|
|
|
9571 break;
|
|
|
9572 /* the word so far was banned but we may try compounding */
|
|
|
9573 goodword_ends = FALSE;
|
|
|
9574 }
|
|
323
|
9575
|
|
|
9576 newscore = 0;
|
|
|
9577 if ((flags & WF_REGION)
|
|
390
|
9578 && (((unsigned)flags >> 16) & lp->lp_region) == 0)
|
|
323
|
9579 newscore += SCORE_REGION;
|
|
|
9580 if (flags & WF_RARE)
|
|
|
9581 newscore += SCORE_RARE;
|
|
|
9582
|
|
346
|
9583 if (!spell_valid_case(su->su_badflags,
|
|
483
|
9584 captype(preword + sp->ts_prewordlen, NULL)))
|
|
323
|
9585 newscore += SCORE_ICASE;
|
|
|
9586
|
|
534
|
9587 maysplit = TRUE;
|
|
|
9588 if (fword_ends && goodword_ends
|
|
|
9589 && sp->ts_fidx >= sp->ts_fidxtry)
|
|
323
|
9590 {
|
|
366
|
9591 /* The badword also ends: add suggestions. Give a penalty
|
|
|
9592 * when changing non-word char to word char, e.g., "thes,"
|
|
|
9593 * -> "these". */
|
|
|
9594 p = fword + sp->ts_fidx;
|
|
|
9595 #ifdef FEAT_MBYTE
|
|
|
9596 if (has_mbyte)
|
|
|
9597 mb_ptr_back(fword, p);
|
|
|
9598 else
|
|
|
9599 #endif
|
|
|
9600 --p;
|
|
376
|
9601 if (!spell_iswordp(p, curbuf))
|
|
366
|
9602 {
|
|
|
9603 p = preword + STRLEN(preword);
|
|
|
9604 #ifdef FEAT_MBYTE
|
|
|
9605 if (has_mbyte)
|
|
|
9606 mb_ptr_back(preword, p);
|
|
|
9607 else
|
|
|
9608 #endif
|
|
|
9609 --p;
|
|
376
|
9610 if (spell_iswordp(p, curbuf))
|
|
366
|
9611 newscore += SCORE_NONWORD;
|
|
|
9612 }
|
|
|
9613
|
|
344
|
9614 add_suggestion(su, &su->su_ga, preword,
|
|
534
|
9615 sp->ts_fidx - repextra,
|
|
|
9616 sp->ts_score + newscore, 0, FALSE,
|
|
|
9617 lp->lp_sallang);
|
|
|
9618
|
|
|
9619 /* When the bad word doesn't end yet, try changing the
|
|
|
9620 * next word. E.g., find suggestions for "the the" where
|
|
|
9621 * the second "the" is different. It's done like a split.
|
|
|
9622 */
|
|
|
9623 if (sp->ts_fidx - repextra >= su->su_badlen)
|
|
|
9624 maysplit = FALSE;
|
|
|
9625 }
|
|
|
9626
|
|
|
9627 if (maysplit
|
|
|
9628 && (sp->ts_fidx >= sp->ts_fidxtry || fword_ends)
|
|
330
|
9629 #ifdef FEAT_MBYTE
|
|
|
9630 /* Don't split halfway a character. */
|
|
|
9631 && (!has_mbyte || sp->ts_tcharlen == 0)
|
|
|
9632 #endif
|
|
|
9633 )
|
|
323
|
9634 {
|
|
482
|
9635 int try_compound;
|
|
|
9636
|
|
|
9637 /* Get here in two situations:
|
|
|
9638 * 1. The word in the tree ends but the badword continues:
|
|
|
9639 * If the word allows compounding try that. Otherwise
|
|
|
9640 * try a split by inserting a space. For both check
|
|
|
9641 * that a valid words starts at fword[sp->ts_fidx].
|
|
492
|
9642 * For NOBREAK do like compounding to be able to check
|
|
|
9643 * if the next word is valid.
|
|
482
|
9644 * 2. The badword does end, but it was due to a change
|
|
|
9645 * (e.g., a swap). No need to split, but do check that
|
|
|
9646 * the following word is valid.
|
|
|
9647 */
|
|
485
|
9648 try_compound = FALSE;
|
|
534
|
9649 if ((!fword_ends || !goodword_ends)
|
|
495
|
9650 && slang->sl_compprog != NULL
|
|
483
|
9651 && ((unsigned)flags >> 24) != 0
|
|
|
9652 && sp->ts_twordlen - sp->ts_splitoff
|
|
482
|
9653 >= slang->sl_compminlen
|
|
500
|
9654 #ifdef FEAT_MBYTE
|
|
|
9655 && (!has_mbyte
|
|
501
|
9656 || slang->sl_compminlen == 0
|
|
500
|
9657 || mb_charlen(tword + sp->ts_splitoff)
|
|
|
9658 >= slang->sl_compminlen)
|
|
|
9659 #endif
|
|
490
|
9660 && (slang->sl_compsylmax < MAXWLEN
|
|
|
9661 || sp->ts_complen + 1 - sp->ts_compsplit
|
|
|
9662 < slang->sl_compmax)
|
|
495
|
9663 && (byte_in_str(sp->ts_complen == sp->ts_compsplit
|
|
485
|
9664 ? slang->sl_compstartflags
|
|
|
9665 : slang->sl_compallflags,
|
|
495
|
9666 ((unsigned)flags >> 24))))
|
|
483
|
9667 {
|
|
482
|
9668 try_compound = TRUE;
|
|
483
|
9669 compflags[sp->ts_complen] = ((unsigned)flags >> 24);
|
|
|
9670 compflags[sp->ts_complen + 1] = NUL;
|
|
|
9671 }
|
|
485
|
9672
|
|
492
|
9673 /* For NOBREAK we never try splitting, it won't make any
|
|
|
9674 * word valid. */
|
|
|
9675 if (slang->sl_nobreak)
|
|
|
9676 try_compound = TRUE;
|
|
|
9677
|
|
485
|
9678 /* If we could add a compound word, and it's also possible
|
|
|
9679 * to split at this point, do the split first and set
|
|
|
9680 * TSF_DIDSPLIT to avoid doing it again. */
|
|
492
|
9681 else if (!fword_ends
|
|
485
|
9682 && try_compound
|
|
|
9683 && (sp->ts_flags & TSF_DIDSPLIT) == 0)
|
|
482
|
9684 {
|
|
|
9685 try_compound = FALSE;
|
|
485
|
9686 sp->ts_flags |= TSF_DIDSPLIT;
|
|
|
9687 --sp->ts_curi; /* do the same NUL again */
|
|
|
9688 compflags[sp->ts_complen] = NUL;
|
|
|
9689 }
|
|
|
9690 else
|
|
|
9691 sp->ts_flags &= ~TSF_DIDSPLIT;
|
|
|
9692
|
|
534
|
9693 if (!try_compound && (!fword_ends || !goodword_ends))
|
|
485
|
9694 {
|
|
|
9695 /* If we're going to split need to check that the
|
|
501
|
9696 * words so far are valid for compounding. If there
|
|
|
9697 * is only one word it must not have the NEEDCOMPOUND
|
|
|
9698 * flag. */
|
|
|
9699 if (sp->ts_complen == sp->ts_compsplit
|
|
|
9700 && (flags & WF_NEEDCOMP))
|
|
|
9701 break;
|
|
490
|
9702 p = preword;
|
|
|
9703 while (*skiptowhite(p) != NUL)
|
|
|
9704 p = skipwhite(skiptowhite(p));
|
|
485
|
9705 if (sp->ts_complen > sp->ts_compsplit
|
|
490
|
9706 && !can_compound(slang, p,
|
|
485
|
9707 compflags + sp->ts_compsplit))
|
|
|
9708 break;
|
|
|
9709 newscore += SCORE_SPLIT;
|
|
482
|
9710 }
|
|
|
9711
|
|
|
9712 if (try_deeper(su, stack, depth, newscore))
|
|
323
|
9713 {
|
|
|
9714 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
346
|
9715 sp->ts_save_badflags = su->su_badflags;
|
|
376
|
9716 sp->ts_state = STATE_SPLITUNDO;
|
|
|
9717
|
|
|
9718 ++depth;
|
|
|
9719 sp = &stack[depth];
|
|
323
|
9720
|
|
482
|
9721 /* Append a space to preword when splitting. */
|
|
|
9722 if (!try_compound && !fword_ends)
|
|
|
9723 STRCAT(preword, " ");
|
|
483
|
9724 sp->ts_prewordlen = STRLEN(preword);
|
|
|
9725 sp->ts_splitoff = sp->ts_twordlen;
|
|
492
|
9726 sp->ts_splitfidx = sp->ts_fidx;
|
|
376
|
9727
|
|
|
9728 /* If the badword has a non-word character at this
|
|
|
9729 * position skip it. That means replacing the
|
|
482
|
9730 * non-word character with a space. Always skip a
|
|
534
|
9731 * character when the word ends. But only when the
|
|
|
9732 * good word can end. */
|
|
|
9733 if (((!try_compound
|
|
|
9734 && !spell_iswordp_nmw(fword + sp->ts_fidx))
|
|
482
|
9735 || fword_ends)
|
|
534
|
9736 && goodword_ends)
|
|
376
|
9737 {
|
|
482
|
9738 int l;
|
|
|
9739
|
|
376
|
9740 #ifdef FEAT_MBYTE
|
|
|
9741 if (has_mbyte)
|
|
482
|
9742 l = MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
376
|
9743 else
|
|
|
9744 #endif
|
|
482
|
9745 l = 1;
|
|
|
9746 if (fword_ends)
|
|
|
9747 {
|
|
|
9748 /* Copy the skipped character to preword. */
|
|
483
|
9749 mch_memmove(preword + sp->ts_prewordlen,
|
|
482
|
9750 fword + sp->ts_fidx, l);
|
|
483
|
9751 sp->ts_prewordlen += l;
|
|
|
9752 preword[sp->ts_prewordlen] = NUL;
|
|
482
|
9753 }
|
|
|
9754 else
|
|
|
9755 sp->ts_score -= SCORE_SPLIT - SCORE_SUBST;
|
|
|
9756 sp->ts_fidx += l;
|
|
376
|
9757 }
|
|
455
|
9758
|
|
485
|
9759 /* When compounding include compound flag in
|
|
|
9760 * compflags[] (already set above). When splitting we
|
|
|
9761 * may start compounding over again. */
|
|
482
|
9762 if (try_compound)
|
|
483
|
9763 ++sp->ts_complen;
|
|
482
|
9764 else
|
|
485
|
9765 sp->ts_compsplit = sp->ts_complen;
|
|
|
9766 sp->ts_prefixdepth = PFD_NOPREFIX;
|
|
482
|
9767
|
|
455
|
9768 /* set su->su_badflags to the caps type at this
|
|
|
9769 * position */
|
|
324
|
9770 #ifdef FEAT_MBYTE
|
|
|
9771 if (has_mbyte)
|
|
455
|
9772 n = nofold_len(fword, sp->ts_fidx, su->su_badptr);
|
|
324
|
9773 else
|
|
|
9774 #endif
|
|
455
|
9775 n = sp->ts_fidx;
|
|
474
|
9776 su->su_badflags = badword_captype(su->su_badptr + n,
|
|
455
|
9777 su->su_badptr + su->su_badlen);
|
|
323
|
9778
|
|
|
9779 /* Restart at top of the tree. */
|
|
376
|
9780 sp->ts_arridx = 0;
|
|
485
|
9781
|
|
|
9782 /* If there are postponed prefixes, try these too. */
|
|
|
9783 if (pbyts != NULL)
|
|
|
9784 {
|
|
|
9785 byts = pbyts;
|
|
|
9786 idxs = pidxs;
|
|
|
9787 sp->ts_prefixdepth = PFD_PREFIXTREE;
|
|
|
9788 sp->ts_state = STATE_NOPREFIX;
|
|
|
9789 }
|
|
323
|
9790 }
|
|
|
9791 }
|
|
|
9792 break;
|
|
|
9793
|
|
|
9794 case STATE_SPLITUNDO:
|
|
482
|
9795 /* Undo the changes done for word split or compound word. */
|
|
346
|
9796 su->su_badflags = sp->ts_save_badflags;
|
|
323
|
9797
|
|
|
9798 /* Continue looking for NUL bytes. */
|
|
|
9799 sp->ts_state = STATE_START;
|
|
485
|
9800
|
|
|
9801 /* In case we went into the prefix tree. */
|
|
|
9802 byts = fbyts;
|
|
|
9803 idxs = fidxs;
|
|
323
|
9804 break;
|
|
|
9805
|
|
|
9806 case STATE_ENDNUL:
|
|
|
9807 /* Past the NUL bytes in the node. */
|
|
455
|
9808 su->su_badflags = sp->ts_save_badflags;
|
|
501
|
9809 if (fword[sp->ts_fidx] == NUL
|
|
|
9810 #ifdef FEAT_MBYTE
|
|
|
9811 && sp->ts_tcharlen == 0
|
|
|
9812 #endif
|
|
|
9813 )
|
|
323
|
9814 {
|
|
|
9815 /* The badword ends, can't use the bytes in this node. */
|
|
|
9816 sp->ts_state = STATE_DEL;
|
|
|
9817 break;
|
|
|
9818 }
|
|
|
9819 sp->ts_state = STATE_PLAIN;
|
|
|
9820 /*FALLTHROUGH*/
|
|
|
9821
|
|
|
9822 case STATE_PLAIN:
|
|
|
9823 /*
|
|
|
9824 * Go over all possible bytes at this node, add each to
|
|
|
9825 * tword[] and use child node. "ts_curi" is the index.
|
|
|
9826 */
|
|
|
9827 arridx = sp->ts_arridx;
|
|
|
9828 if (sp->ts_curi > byts[arridx])
|
|
|
9829 {
|
|
|
9830 /* Done all bytes at this node, do next state. When still
|
|
|
9831 * at already changed bytes skip the other tricks. */
|
|
|
9832 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
9833 sp->ts_state = STATE_DEL;
|
|
|
9834 else
|
|
|
9835 sp->ts_state = STATE_FINAL;
|
|
|
9836 }
|
|
|
9837 else
|
|
|
9838 {
|
|
|
9839 arridx += sp->ts_curi++;
|
|
|
9840 c = byts[arridx];
|
|
|
9841
|
|
|
9842 /* Normal byte, go one level deeper. If it's not equal to
|
|
|
9843 * the byte in the bad word adjust the score. But don't
|
|
|
9844 * even try when the byte was already changed. */
|
|
330
|
9845 if (c == fword[sp->ts_fidx]
|
|
|
9846 #ifdef FEAT_MBYTE
|
|
|
9847 || (sp->ts_tcharlen > 0
|
|
|
9848 && sp->ts_isdiff != DIFF_NONE)
|
|
|
9849 #endif
|
|
|
9850 )
|
|
323
|
9851 newscore = 0;
|
|
|
9852 else
|
|
|
9853 newscore = SCORE_SUBST;
|
|
|
9854 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
|
9855 && try_deeper(su, stack, depth, newscore))
|
|
|
9856 {
|
|
|
9857 ++depth;
|
|
330
|
9858 sp = &stack[depth];
|
|
|
9859 ++sp->ts_fidx;
|
|
|
9860 tword[sp->ts_twordlen++] = c;
|
|
|
9861 sp->ts_arridx = idxs[arridx];
|
|
|
9862 #ifdef FEAT_MBYTE
|
|
|
9863 if (newscore == SCORE_SUBST)
|
|
|
9864 sp->ts_isdiff = DIFF_YES;
|
|
|
9865 if (has_mbyte)
|
|
|
9866 {
|
|
|
9867 /* Multi-byte characters are a bit complicated to
|
|
|
9868 * handle: They differ when any of the bytes
|
|
|
9869 * differ and then their length may also differ. */
|
|
|
9870 if (sp->ts_tcharlen == 0)
|
|
|
9871 {
|
|
|
9872 /* First byte. */
|
|
|
9873 sp->ts_tcharidx = 0;
|
|
|
9874 sp->ts_tcharlen = MB_BYTE2LEN(c);
|
|
|
9875 sp->ts_fcharstart = sp->ts_fidx - 1;
|
|
|
9876 sp->ts_isdiff = (newscore != 0)
|
|
|
9877 ? DIFF_YES : DIFF_NONE;
|
|
|
9878 }
|
|
|
9879 else if (sp->ts_isdiff == DIFF_INSERT)
|
|
|
9880 /* When inserting trail bytes don't advance in
|
|
|
9881 * the bad word. */
|
|
|
9882 --sp->ts_fidx;
|
|
|
9883 if (++sp->ts_tcharidx == sp->ts_tcharlen)
|
|
|
9884 {
|
|
|
9885 /* Last byte of character. */
|
|
|
9886 if (sp->ts_isdiff == DIFF_YES)
|
|
|
9887 {
|
|
|
9888 /* Correct ts_fidx for the byte length of
|
|
|
9889 * the character (we didn't check that
|
|
|
9890 * before). */
|
|
|
9891 sp->ts_fidx = sp->ts_fcharstart
|
|
|
9892 + MB_BYTE2LEN(
|
|
|
9893 fword[sp->ts_fcharstart]);
|
|
|
9894
|
|
479
|
9895 /* For changing a composing character
|
|
|
9896 * adjust the score from SCORE_SUBST to
|
|
|
9897 * SCORE_SUBCOMP. */
|
|
|
9898 if (enc_utf8
|
|
|
9899 && utf_iscomposing(
|
|
|
9900 mb_ptr2char(tword
|
|
|
9901 + sp->ts_twordlen
|
|
|
9902 - sp->ts_tcharlen))
|
|
|
9903 && utf_iscomposing(
|
|
|
9904 mb_ptr2char(fword
|
|
|
9905 + sp->ts_fcharstart)))
|
|
|
9906 sp->ts_score -=
|
|
|
9907 SCORE_SUBST - SCORE_SUBCOMP;
|
|
|
9908
|
|
330
|
9909 /* For a similar character adjust score
|
|
|
9910 * from SCORE_SUBST to SCORE_SIMILAR. */
|
|
482
|
9911 else if (slang->sl_has_map
|
|
|
9912 && similar_chars(slang,
|
|
330
|
9913 mb_ptr2char(tword
|
|
|
9914 + sp->ts_twordlen
|
|
|
9915 - sp->ts_tcharlen),
|
|
|
9916 mb_ptr2char(fword
|
|
|
9917 + sp->ts_fcharstart)))
|
|
|
9918 sp->ts_score -=
|
|
|
9919 SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9920 }
|
|
358
|
9921 else if (sp->ts_isdiff == DIFF_INSERT
|
|
|
9922 && sp->ts_twordlen > sp->ts_tcharlen)
|
|
|
9923 {
|
|
|
9924 p = tword + sp->ts_twordlen
|
|
|
9925 - sp->ts_tcharlen;
|
|
|
9926 c = mb_ptr2char(p);
|
|
475
|
9927 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9928 {
|
|
|
9929 /* Inserting a composing char doesn't
|
|
|
9930 * count that much. */
|
|
358
|
9931 sp->ts_score -= SCORE_INS
|
|
475
|
9932 - SCORE_INSCOMP;
|
|
|
9933 }
|
|
|
9934 else
|
|
|
9935 {
|
|
|
9936 /* If the previous character was the
|
|
|
9937 * same, thus doubling a character,
|
|
|
9938 * give a bonus to the score. */
|
|
|
9939 mb_ptr_back(tword, p);
|
|
|
9940 if (c == mb_ptr2char(p))
|
|
|
9941 sp->ts_score -= SCORE_INS
|
|
358
|
9942 - SCORE_INSDUP;
|
|
475
|
9943 }
|
|
358
|
9944 }
|
|
330
|
9945
|
|
|
9946 /* Starting a new char, reset the length. */
|
|
|
9947 sp->ts_tcharlen = 0;
|
|
|
9948 }
|
|
|
9949 }
|
|
|
9950 else
|
|
|
9951 #endif
|
|
|
9952 {
|
|
|
9953 /* If we found a similar char adjust the score.
|
|
|
9954 * We do this after calling try_deeper() because
|
|
|
9955 * it's slow. */
|
|
|
9956 if (newscore != 0
|
|
482
|
9957 && slang->sl_has_map
|
|
|
9958 && similar_chars(slang,
|
|
330
|
9959 c, fword[sp->ts_fidx - 1]))
|
|
|
9960 sp->ts_score -= SCORE_SUBST - SCORE_SIMILAR;
|
|
|
9961 }
|
|
323
|
9962 }
|
|
|
9963 }
|
|
|
9964 break;
|
|
|
9965
|
|
|
9966 case STATE_DEL:
|
|
330
|
9967 #ifdef FEAT_MBYTE
|
|
|
9968 /* When past the first byte of a multi-byte char don't try
|
|
|
9969 * delete/insert/swap a character. */
|
|
|
9970 if (has_mbyte && sp->ts_tcharlen > 0)
|
|
|
9971 {
|
|
|
9972 sp->ts_state = STATE_FINAL;
|
|
|
9973 break;
|
|
|
9974 }
|
|
|
9975 #endif
|
|
|
9976 /*
|
|
|
9977 * Try skipping one character in the bad word (delete it).
|
|
|
9978 */
|
|
323
|
9979 sp->ts_state = STATE_INS;
|
|
|
9980 sp->ts_curi = 1;
|
|
|
9981 if (fword[sp->ts_fidx] != NUL
|
|
|
9982 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
|
9983 {
|
|
|
9984 ++depth;
|
|
358
|
9985
|
|
|
9986 /* Advance over the character in fword[]. Give a bonus to
|
|
|
9987 * the score if the same character is following "nn" ->
|
|
|
9988 * "n". */
|
|
330
|
9989 #ifdef FEAT_MBYTE
|
|
|
9990 if (has_mbyte)
|
|
358
|
9991 {
|
|
|
9992 c = mb_ptr2char(fword + sp->ts_fidx);
|
|
330
|
9993 stack[depth].ts_fidx += MB_BYTE2LEN(fword[sp->ts_fidx]);
|
|
479
|
9994 if (enc_utf8 && utf_iscomposing(c))
|
|
|
9995 stack[depth].ts_score -= SCORE_DEL - SCORE_DELCOMP;
|
|
|
9996 else if (c == mb_ptr2char(fword + stack[depth].ts_fidx))
|
|
358
|
9997 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
9998 }
|
|
330
|
9999 else
|
|
|
10000 #endif
|
|
358
|
10001 {
|
|
330
|
10002 ++stack[depth].ts_fidx;
|
|
358
|
10003 if (fword[sp->ts_fidx] == fword[sp->ts_fidx + 1])
|
|
|
10004 stack[depth].ts_score -= SCORE_DEL - SCORE_DELDUP;
|
|
|
10005 }
|
|
323
|
10006 break;
|
|
|
10007 }
|
|
|
10008 /*FALLTHROUGH*/
|
|
|
10009
|
|
|
10010 case STATE_INS:
|
|
330
|
10011 /* Insert one byte. Do this for each possible byte at this
|
|
323
|
10012 * node. */
|
|
|
10013 n = sp->ts_arridx;
|
|
|
10014 if (sp->ts_curi > byts[n])
|
|
|
10015 {
|
|
|
10016 /* Done all bytes at this node, do next state. */
|
|
|
10017 sp->ts_state = STATE_SWAP;
|
|
|
10018 }
|
|
|
10019 else
|
|
|
10020 {
|
|
330
|
10021 /* Do one more byte at this node. Skip NUL bytes. */
|
|
323
|
10022 n += sp->ts_curi++;
|
|
|
10023 c = byts[n];
|
|
|
10024 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
|
10025 {
|
|
|
10026 ++depth;
|
|
330
|
10027 sp = &stack[depth];
|
|
|
10028 tword[sp->ts_twordlen++] = c;
|
|
|
10029 sp->ts_arridx = idxs[n];
|
|
|
10030 #ifdef FEAT_MBYTE
|
|
|
10031 if (has_mbyte)
|
|
|
10032 {
|
|
|
10033 fl = MB_BYTE2LEN(c);
|
|
|
10034 if (fl > 1)
|
|
|
10035 {
|
|
|
10036 /* There are following bytes for the same
|
|
|
10037 * character. We must find all bytes before
|
|
|
10038 * trying delete/insert/swap/etc. */
|
|
|
10039 sp->ts_tcharlen = fl;
|
|
|
10040 sp->ts_tcharidx = 1;
|
|
|
10041 sp->ts_isdiff = DIFF_INSERT;
|
|
|
10042 }
|
|
|
10043 }
|
|
358
|
10044 else
|
|
|
10045 fl = 1;
|
|
|
10046 if (fl == 1)
|
|
330
|
10047 #endif
|
|
358
|
10048 {
|
|
|
10049 /* If the previous character was the same, thus
|
|
|
10050 * doubling a character, give a bonus to the
|
|
|
10051 * score. */
|
|
|
10052 if (sp->ts_twordlen >= 2
|
|
|
10053 && tword[sp->ts_twordlen - 2] == c)
|
|
|
10054 sp->ts_score -= SCORE_INS - SCORE_INSDUP;
|
|
|
10055 }
|
|
323
|
10056 }
|
|
|
10057 }
|
|
|
10058 break;
|
|
|
10059
|
|
|
10060 case STATE_SWAP:
|
|
330
|
10061 /*
|
|
|
10062 * Swap two bytes in the bad word: "12" -> "21".
|
|
|
10063 * We change "fword" here, it's changed back afterwards.
|
|
|
10064 */
|
|
|
10065 p = fword + sp->ts_fidx;
|
|
|
10066 c = *p;
|
|
|
10067 if (c == NUL)
|
|
|
10068 {
|
|
|
10069 /* End of word, can't swap or replace. */
|
|
|
10070 sp->ts_state = STATE_FINAL;
|
|
|
10071 break;
|
|
|
10072 }
|
|
539
|
10073
|
|
|
10074 /* Don't swap if the first character is not a word character.
|
|
|
10075 * SWAP3 etc. also don't make sense then. */
|
|
|
10076 if (!spell_iswordp(p, curbuf))
|
|
|
10077 {
|
|
|
10078 sp->ts_state = STATE_REP_INI;
|
|
|
10079 break;
|
|
|
10080 }
|
|
|
10081
|
|
330
|
10082 #ifdef FEAT_MBYTE
|
|
|
10083 if (has_mbyte)
|
|
|
10084 {
|
|
474
|
10085 n = mb_cptr2len(p);
|
|
330
|
10086 c = mb_ptr2char(p);
|
|
539
|
10087 if (!spell_iswordp(p + n, curbuf))
|
|
|
10088 c2 = c; /* don't swap non-word char */
|
|
|
10089 else
|
|
|
10090 c2 = mb_ptr2char(p + n);
|
|
330
|
10091 }
|
|
|
10092 else
|
|
|
10093 #endif
|
|
539
|
10094 {
|
|
|
10095 if (!spell_iswordp(p + 1, curbuf))
|
|
|
10096 c2 = c; /* don't swap non-word char */
|
|
|
10097 else
|
|
|
10098 c2 = p[1];
|
|
|
10099 }
|
|
|
10100
|
|
|
10101 /* When characters are identical, swap won't do anything.
|
|
|
10102 * Also get here if the second char is not a word character. */
|
|
330
|
10103 if (c == c2)
|
|
323
|
10104 {
|
|
330
|
10105 sp->ts_state = STATE_SWAP3;
|
|
|
10106 break;
|
|
|
10107 }
|
|
|
10108 if (c2 != NUL && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
|
10109 {
|
|
|
10110 sp->ts_state = STATE_UNSWAP;
|
|
323
|
10111 ++depth;
|
|
330
|
10112 #ifdef FEAT_MBYTE
|
|
|
10113 if (has_mbyte)
|
|
|
10114 {
|
|
|
10115 fl = mb_char2len(c2);
|
|
|
10116 mch_memmove(p, p + n, fl);
|
|
|
10117 mb_char2bytes(c, p + fl);
|
|
|
10118 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
10119 }
|
|
|
10120 else
|
|
|
10121 #endif
|
|
|
10122 {
|
|
|
10123 p[0] = c2;
|
|
|
10124 p[1] = c;
|
|
|
10125 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
|
10126 }
|
|
323
|
10127 }
|
|
|
10128 else
|
|
|
10129 /* If this swap doesn't work then SWAP3 won't either. */
|
|
|
10130 sp->ts_state = STATE_REP_INI;
|
|
|
10131 break;
|
|
|
10132
|
|
330
|
10133 case STATE_UNSWAP:
|
|
|
10134 /* Undo the STATE_SWAP swap: "21" -> "12". */
|
|
|
10135 p = fword + sp->ts_fidx;
|
|
|
10136 #ifdef FEAT_MBYTE
|
|
|
10137 if (has_mbyte)
|
|
|
10138 {
|
|
|
10139 n = MB_BYTE2LEN(*p);
|
|
|
10140 c = mb_ptr2char(p + n);
|
|
|
10141 mch_memmove(p + MB_BYTE2LEN(p[n]), p, n);
|
|
|
10142 mb_char2bytes(c, p);
|
|
|
10143 }
|
|
|
10144 else
|
|
|
10145 #endif
|
|
|
10146 {
|
|
|
10147 c = *p;
|
|
|
10148 *p = p[1];
|
|
|
10149 p[1] = c;
|
|
|
10150 }
|
|
|
10151 /*FALLTHROUGH*/
|
|
|
10152
|
|
|
10153 case STATE_SWAP3:
|
|
323
|
10154 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
330
|
10155 * "fword" here, it's changed back afterwards. */
|
|
|
10156 p = fword + sp->ts_fidx;
|
|
|
10157 #ifdef FEAT_MBYTE
|
|
|
10158 if (has_mbyte)
|
|
|
10159 {
|
|
474
|
10160 n = mb_cptr2len(p);
|
|
330
|
10161 c = mb_ptr2char(p);
|
|
474
|
10162 fl = mb_cptr2len(p + n);
|
|
330
|
10163 c2 = mb_ptr2char(p + n);
|
|
539
|
10164 if (!spell_iswordp(p + n + fl, curbuf))
|
|
|
10165 c3 = c; /* don't swap non-word char */
|
|
|
10166 else
|
|
|
10167 c3 = mb_ptr2char(p + n + fl);
|
|
330
|
10168 }
|
|
|
10169 else
|
|
|
10170 #endif
|
|
323
|
10171 {
|
|
330
|
10172 c = *p;
|
|
|
10173 c2 = p[1];
|
|
539
|
10174 if (!spell_iswordp(p + 2, curbuf))
|
|
|
10175 c3 = c; /* don't swap non-word char */
|
|
|
10176 else
|
|
|
10177 c3 = p[2];
|
|
330
|
10178 }
|
|
|
10179
|
|
|
10180 /* When characters are identical: "121" then SWAP3 result is
|
|
|
10181 * identical, ROT3L result is same as SWAP: "211", ROT3L
|
|
|
10182 * result is same as SWAP on next char: "112". Thus skip all
|
|
539
|
10183 * swapping. Also skip when c3 is NUL.
|
|
|
10184 * Also get here when the third character is not a word
|
|
|
10185 * character. Second character may any char: "a.b" -> "b.a" */
|
|
330
|
10186 if (c == c3 || c3 == NUL)
|
|
|
10187 {
|
|
|
10188 sp->ts_state = STATE_REP_INI;
|
|
|
10189 break;
|
|
|
10190 }
|
|
|
10191 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
10192 {
|
|
|
10193 sp->ts_state = STATE_UNSWAP3;
|
|
323
|
10194 ++depth;
|
|
330
|
10195 #ifdef FEAT_MBYTE
|
|
|
10196 if (has_mbyte)
|
|
|
10197 {
|
|
|
10198 tl = mb_char2len(c3);
|
|
|
10199 mch_memmove(p, p + n + fl, tl);
|
|
|
10200 mb_char2bytes(c2, p + tl);
|
|
|
10201 mb_char2bytes(c, p + fl + tl);
|
|
|
10202 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl + tl;
|
|
|
10203 }
|
|
|
10204 else
|
|
|
10205 #endif
|
|
|
10206 {
|
|
|
10207 p[0] = p[2];
|
|
|
10208 p[2] = c;
|
|
|
10209 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
10210 }
|
|
323
|
10211 }
|
|
|
10212 else
|
|
|
10213 sp->ts_state = STATE_REP_INI;
|
|
|
10214 break;
|
|
|
10215
|
|
330
|
10216 case STATE_UNSWAP3:
|
|
|
10217 /* Undo STATE_SWAP3: "321" -> "123" */
|
|
|
10218 p = fword + sp->ts_fidx;
|
|
|
10219 #ifdef FEAT_MBYTE
|
|
|
10220 if (has_mbyte)
|
|
|
10221 {
|
|
|
10222 n = MB_BYTE2LEN(*p);
|
|
|
10223 c2 = mb_ptr2char(p + n);
|
|
|
10224 fl = MB_BYTE2LEN(p[n]);
|
|
|
10225 c = mb_ptr2char(p + n + fl);
|
|
|
10226 tl = MB_BYTE2LEN(p[n + fl]);
|
|
|
10227 mch_memmove(p + fl + tl, p, n);
|
|
|
10228 mb_char2bytes(c, p);
|
|
|
10229 mb_char2bytes(c2, p + tl);
|
|
539
|
10230 p = p + tl;
|
|
330
|
10231 }
|
|
|
10232 else
|
|
|
10233 #endif
|
|
|
10234 {
|
|
|
10235 c = *p;
|
|
|
10236 *p = p[2];
|
|
|
10237 p[2] = c;
|
|
539
|
10238 ++p;
|
|
|
10239 }
|
|
|
10240
|
|
|
10241 if (!spell_iswordp(p, curbuf))
|
|
|
10242 {
|
|
|
10243 /* Middle char is not a word char, skip the rotate.
|
|
|
10244 * First and third char were already checked at swap
|
|
|
10245 * and swap3. */
|
|
|
10246 sp->ts_state = STATE_REP_INI;
|
|
|
10247 break;
|
|
330
|
10248 }
|
|
346
|
10249
|
|
330
|
10250 /* Rotate three characters left: "123" -> "231". We change
|
|
|
10251 * "fword" here, it's changed back afterwards. */
|
|
323
|
10252 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
10253 {
|
|
330
|
10254 sp->ts_state = STATE_UNROT3L;
|
|
323
|
10255 ++depth;
|
|
330
|
10256 p = fword + sp->ts_fidx;
|
|
|
10257 #ifdef FEAT_MBYTE
|
|
|
10258 if (has_mbyte)
|
|
|
10259 {
|
|
474
|
10260 n = mb_cptr2len(p);
|
|
330
|
10261 c = mb_ptr2char(p);
|
|
474
|
10262 fl = mb_cptr2len(p + n);
|
|
|
10263 fl += mb_cptr2len(p + n + fl);
|
|
330
|
10264 mch_memmove(p, p + n, fl);
|
|
|
10265 mb_char2bytes(c, p + fl);
|
|
|
10266 stack[depth].ts_fidxtry = sp->ts_fidx + n + fl;
|
|
|
10267 }
|
|
|
10268 else
|
|
|
10269 #endif
|
|
|
10270 {
|
|
|
10271 c = *p;
|
|
|
10272 *p = p[1];
|
|
|
10273 p[1] = p[2];
|
|
|
10274 p[2] = c;
|
|
|
10275 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
10276 }
|
|
323
|
10277 }
|
|
|
10278 else
|
|
|
10279 sp->ts_state = STATE_REP_INI;
|
|
|
10280 break;
|
|
|
10281
|
|
330
|
10282 case STATE_UNROT3L:
|
|
346
|
10283 /* Undo ROT3L: "231" -> "123" */
|
|
330
|
10284 p = fword + sp->ts_fidx;
|
|
|
10285 #ifdef FEAT_MBYTE
|
|
|
10286 if (has_mbyte)
|
|
|
10287 {
|
|
|
10288 n = MB_BYTE2LEN(*p);
|
|
|
10289 n += MB_BYTE2LEN(p[n]);
|
|
|
10290 c = mb_ptr2char(p + n);
|
|
|
10291 tl = MB_BYTE2LEN(p[n]);
|
|
|
10292 mch_memmove(p + tl, p, n);
|
|
|
10293 mb_char2bytes(c, p);
|
|
|
10294 }
|
|
|
10295 else
|
|
|
10296 #endif
|
|
|
10297 {
|
|
|
10298 c = p[2];
|
|
|
10299 p[2] = p[1];
|
|
|
10300 p[1] = *p;
|
|
|
10301 *p = c;
|
|
|
10302 }
|
|
346
|
10303
|
|
323
|
10304 /* Rotate three bytes right: "123" -> "312". We change
|
|
330
|
10305 * "fword" here, it's changed back afterwards. */
|
|
323
|
10306 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
|
10307 {
|
|
330
|
10308 sp->ts_state = STATE_UNROT3R;
|
|
323
|
10309 ++depth;
|
|
330
|
10310 p = fword + sp->ts_fidx;
|
|
|
10311 #ifdef FEAT_MBYTE
|
|
|
10312 if (has_mbyte)
|
|
|
10313 {
|
|
474
|
10314 n = mb_cptr2len(p);
|
|
|
10315 n += mb_cptr2len(p + n);
|
|
330
|
10316 c = mb_ptr2char(p + n);
|
|
474
|
10317 tl = mb_cptr2len(p + n);
|
|
330
|
10318 mch_memmove(p + tl, p, n);
|
|
|
10319 mb_char2bytes(c, p);
|
|
|
10320 stack[depth].ts_fidxtry = sp->ts_fidx + n + tl;
|
|
|
10321 }
|
|
|
10322 else
|
|
|
10323 #endif
|
|
|
10324 {
|
|
|
10325 c = p[2];
|
|
|
10326 p[2] = p[1];
|
|
|
10327 p[1] = *p;
|
|
|
10328 *p = c;
|
|
|
10329 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
|
10330 }
|
|
323
|
10331 }
|
|
|
10332 else
|
|
|
10333 sp->ts_state = STATE_REP_INI;
|
|
|
10334 break;
|
|
|
10335
|
|
330
|
10336 case STATE_UNROT3R:
|
|
346
|
10337 /* Undo ROT3R: "312" -> "123" */
|
|
330
|
10338 p = fword + sp->ts_fidx;
|
|
|
10339 #ifdef FEAT_MBYTE
|
|
|
10340 if (has_mbyte)
|
|
|
10341 {
|
|
|
10342 c = mb_ptr2char(p);
|
|
|
10343 tl = MB_BYTE2LEN(*p);
|
|
|
10344 n = MB_BYTE2LEN(p[tl]);
|
|
|
10345 n += MB_BYTE2LEN(p[tl + n]);
|
|
|
10346 mch_memmove(p, p + tl, n);
|
|
|
10347 mb_char2bytes(c, p + n);
|
|
|
10348 }
|
|
|
10349 else
|
|
|
10350 #endif
|
|
|
10351 {
|
|
|
10352 c = *p;
|
|
|
10353 *p = p[1];
|
|
|
10354 p[1] = p[2];
|
|
|
10355 p[2] = c;
|
|
|
10356 }
|
|
323
|
10357 /*FALLTHROUGH*/
|
|
|
10358
|
|
|
10359 case STATE_REP_INI:
|
|
|
10360 /* Check if matching with REP items from the .aff file would
|
|
503
|
10361 * work. Quickly skip if:
|
|
|
10362 * - there are no REP items
|
|
|
10363 * - the score is going to be too high anyway
|
|
|
10364 * - already applied a REP item or swapped here */
|
|
|
10365 if (lp->lp_replang == NULL
|
|
|
10366 || sp->ts_score + SCORE_REP >= su->su_maxscore
|
|
|
10367 || sp->ts_fidx < sp->ts_fidxtry)
|
|
323
|
10368 {
|
|
|
10369 sp->ts_state = STATE_FINAL;
|
|
|
10370 break;
|
|
|
10371 }
|
|
503
|
10372 gap = &lp->lp_replang->sl_rep;
|
|
323
|
10373
|
|
|
10374 /* Use the first byte to quickly find the first entry that
|
|
330
|
10375 * may match. If the index is -1 there is none. */
|
|
503
|
10376 sp->ts_curi = lp->lp_replang->sl_rep_first[fword[sp->ts_fidx]];
|
|
323
|
10377 if (sp->ts_curi < 0)
|
|
|
10378 {
|
|
|
10379 sp->ts_state = STATE_FINAL;
|
|
|
10380 break;
|
|
|
10381 }
|
|
|
10382
|
|
|
10383 sp->ts_state = STATE_REP;
|
|
|
10384 /*FALLTHROUGH*/
|
|
|
10385
|
|
|
10386 case STATE_REP:
|
|
|
10387 /* Try matching with REP items from the .aff file. For each
|
|
330
|
10388 * match replace the characters and check if the resulting
|
|
|
10389 * word is valid. */
|
|
323
|
10390 p = fword + sp->ts_fidx;
|
|
|
10391
|
|
503
|
10392 gap = &lp->lp_replang->sl_rep;
|
|
323
|
10393 while (sp->ts_curi < gap->ga_len)
|
|
|
10394 {
|
|
|
10395 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
|
10396 if (*ftp->ft_from != *p)
|
|
|
10397 {
|
|
|
10398 /* past possible matching entries */
|
|
|
10399 sp->ts_curi = gap->ga_len;
|
|
|
10400 break;
|
|
|
10401 }
|
|
|
10402 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
|
10403 && try_deeper(su, stack, depth, SCORE_REP))
|
|
|
10404 {
|
|
|
10405 /* Need to undo this afterwards. */
|
|
|
10406 sp->ts_state = STATE_REP_UNDO;
|
|
|
10407
|
|
|
10408 /* Change the "from" to the "to" string. */
|
|
|
10409 ++depth;
|
|
|
10410 fl = STRLEN(ftp->ft_from);
|
|
|
10411 tl = STRLEN(ftp->ft_to);
|
|
|
10412 if (fl != tl)
|
|
346
|
10413 {
|
|
323
|
10414 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
346
|
10415 repextra += tl - fl;
|
|
|
10416 }
|
|
323
|
10417 mch_memmove(p, ftp->ft_to, tl);
|
|
|
10418 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
330
|
10419 #ifdef FEAT_MBYTE
|
|
|
10420 stack[depth].ts_tcharlen = 0;
|
|
|
10421 #endif
|
|
323
|
10422 break;
|
|
|
10423 }
|
|
|
10424 }
|
|
|
10425
|
|
503
|
10426 if (sp->ts_curi >= gap->ga_len && sp->ts_state == STATE_REP)
|
|
323
|
10427 /* No (more) matches. */
|
|
|
10428 sp->ts_state = STATE_FINAL;
|
|
|
10429
|
|
|
10430 break;
|
|
|
10431
|
|
|
10432 case STATE_REP_UNDO:
|
|
|
10433 /* Undo a REP replacement and continue with the next one. */
|
|
503
|
10434 ftp = (fromto_T *)lp->lp_replang->sl_rep.ga_data
|
|
|
10435 + sp->ts_curi - 1;
|
|
323
|
10436 fl = STRLEN(ftp->ft_from);
|
|
|
10437 tl = STRLEN(ftp->ft_to);
|
|
|
10438 p = fword + sp->ts_fidx;
|
|
|
10439 if (fl != tl)
|
|
346
|
10440 {
|
|
323
|
10441 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
346
|
10442 repextra -= tl - fl;
|
|
|
10443 }
|
|
323
|
10444 mch_memmove(p, ftp->ft_from, fl);
|
|
|
10445 sp->ts_state = STATE_REP;
|
|
|
10446 break;
|
|
|
10447
|
|
|
10448 default:
|
|
|
10449 /* Did all possible states at this level, go up one level. */
|
|
|
10450 --depth;
|
|
344
|
10451
|
|
482
|
10452 if (depth >= 0 && stack[depth].ts_prefixdepth == PFD_PREFIXTREE)
|
|
375
|
10453 {
|
|
|
10454 /* Continue in or go back to the prefix tree. */
|
|
|
10455 byts = pbyts;
|
|
|
10456 idxs = pidxs;
|
|
|
10457 }
|
|
|
10458
|
|
344
|
10459 /* Don't check for CTRL-C too often, it takes time. */
|
|
|
10460 line_breakcheck();
|
|
323
|
10461 }
|
|
|
10462 }
|
|
|
10463 }
|
|
|
10464 }
|
|
|
10465
|
|
|
10466 /*
|
|
|
10467 * Try going one level deeper in the tree.
|
|
|
10468 */
|
|
|
10469 static int
|
|
|
10470 try_deeper(su, stack, depth, score_add)
|
|
|
10471 suginfo_T *su;
|
|
|
10472 trystate_T *stack;
|
|
|
10473 int depth;
|
|
|
10474 int score_add;
|
|
|
10475 {
|
|
|
10476 int newscore;
|
|
|
10477
|
|
|
10478 /* Refuse to go deeper if the scrore is getting too big. */
|
|
|
10479 newscore = stack[depth].ts_score + score_add;
|
|
|
10480 if (newscore >= su->su_maxscore)
|
|
|
10481 return FALSE;
|
|
|
10482
|
|
330
|
10483 stack[depth + 1] = stack[depth];
|
|
323
|
10484 stack[depth + 1].ts_state = STATE_START;
|
|
|
10485 stack[depth + 1].ts_score = newscore;
|
|
|
10486 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
485
|
10487 stack[depth + 1].ts_flags = 0;
|
|
323
|
10488 return TRUE;
|
|
|
10489 }
|
|
|
10490
|
|
455
|
10491 #ifdef FEAT_MBYTE
|
|
|
10492 /*
|
|
|
10493 * Case-folding may change the number of bytes: Count nr of chars in
|
|
|
10494 * fword[flen] and return the byte length of that many chars in "word".
|
|
|
10495 */
|
|
|
10496 static int
|
|
|
10497 nofold_len(fword, flen, word)
|
|
|
10498 char_u *fword;
|
|
|
10499 int flen;
|
|
|
10500 char_u *word;
|
|
|
10501 {
|
|
|
10502 char_u *p;
|
|
|
10503 int i = 0;
|
|
|
10504
|
|
|
10505 for (p = fword; p < fword + flen; mb_ptr_adv(p))
|
|
|
10506 ++i;
|
|
|
10507 for (p = word; i > 0; mb_ptr_adv(p))
|
|
|
10508 --i;
|
|
|
10509 return (int)(p - word);
|
|
|
10510 }
|
|
|
10511 #endif
|
|
|
10512
|
|
323
|
10513 /*
|
|
|
10514 * "fword" is a good word with case folded. Find the matching keep-case
|
|
|
10515 * words and put it in "kword".
|
|
|
10516 * Theoretically there could be several keep-case words that result in the
|
|
|
10517 * same case-folded word, but we only find one...
|
|
|
10518 */
|
|
|
10519 static void
|
|
|
10520 find_keepcap_word(slang, fword, kword)
|
|
|
10521 slang_T *slang;
|
|
|
10522 char_u *fword;
|
|
|
10523 char_u *kword;
|
|
|
10524 {
|
|
|
10525 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
|
10526 int depth;
|
|
324
|
10527 idx_T tryidx;
|
|
323
|
10528
|
|
|
10529 /* The following arrays are used at each depth in the tree. */
|
|
324
|
10530 idx_T arridx[MAXWLEN];
|
|
323
|
10531 int round[MAXWLEN];
|
|
|
10532 int fwordidx[MAXWLEN];
|
|
|
10533 int uwordidx[MAXWLEN];
|
|
|
10534 int kwordlen[MAXWLEN];
|
|
|
10535
|
|
|
10536 int flen, ulen;
|
|
|
10537 int l;
|
|
|
10538 int len;
|
|
|
10539 int c;
|
|
324
|
10540 idx_T lo, hi, m;
|
|
323
|
10541 char_u *p;
|
|
|
10542 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
324
|
10543 idx_T *idxs = slang->sl_kidxs; /* array with indexes */
|
|
323
|
10544
|
|
|
10545 if (byts == NULL)
|
|
|
10546 {
|
|
|
10547 /* array is empty: "cannot happen" */
|
|
|
10548 *kword = NUL;
|
|
|
10549 return;
|
|
|
10550 }
|
|
|
10551
|
|
|
10552 /* Make an all-cap version of "fword". */
|
|
|
10553 allcap_copy(fword, uword);
|
|
|
10554
|
|
|
10555 /*
|
|
|
10556 * Each character needs to be tried both case-folded and upper-case.
|
|
|
10557 * All this gets very complicated if we keep in mind that changing case
|
|
|
10558 * may change the byte length of a multi-byte character...
|
|
|
10559 */
|
|
|
10560 depth = 0;
|
|
|
10561 arridx[0] = 0;
|
|
|
10562 round[0] = 0;
|
|
|
10563 fwordidx[0] = 0;
|
|
|
10564 uwordidx[0] = 0;
|
|
|
10565 kwordlen[0] = 0;
|
|
|
10566 while (depth >= 0)
|
|
|
10567 {
|
|
|
10568 if (fword[fwordidx[depth]] == NUL)
|
|
|
10569 {
|
|
|
10570 /* We are at the end of "fword". If the tree allows a word to end
|
|
|
10571 * here we have found a match. */
|
|
|
10572 if (byts[arridx[depth] + 1] == 0)
|
|
|
10573 {
|
|
|
10574 kword[kwordlen[depth]] = NUL;
|
|
|
10575 return;
|
|
|
10576 }
|
|
|
10577
|
|
|
10578 /* kword is getting too long, continue one level up */
|
|
|
10579 --depth;
|
|
|
10580 }
|
|
|
10581 else if (++round[depth] > 2)
|
|
|
10582 {
|
|
|
10583 /* tried both fold-case and upper-case character, continue one
|
|
|
10584 * level up */
|
|
|
10585 --depth;
|
|
|
10586 }
|
|
|
10587 else
|
|
|
10588 {
|
|
|
10589 /*
|
|
|
10590 * round[depth] == 1: Try using the folded-case character.
|
|
|
10591 * round[depth] == 2: Try using the upper-case character.
|
|
|
10592 */
|
|
|
10593 #ifdef FEAT_MBYTE
|
|
|
10594 if (has_mbyte)
|
|
|
10595 {
|
|
474
|
10596 flen = mb_cptr2len(fword + fwordidx[depth]);
|
|
|
10597 ulen = mb_cptr2len(uword + uwordidx[depth]);
|
|
323
|
10598 }
|
|
|
10599 else
|
|
|
10600 #endif
|
|
|
10601 ulen = flen = 1;
|
|
|
10602 if (round[depth] == 1)
|
|
|
10603 {
|
|
|
10604 p = fword + fwordidx[depth];
|
|
|
10605 l = flen;
|
|
|
10606 }
|
|
|
10607 else
|
|
|
10608 {
|
|
|
10609 p = uword + uwordidx[depth];
|
|
|
10610 l = ulen;
|
|
|
10611 }
|
|
|
10612
|
|
|
10613 for (tryidx = arridx[depth]; l > 0; --l)
|
|
|
10614 {
|
|
|
10615 /* Perform a binary search in the list of accepted bytes. */
|
|
|
10616 len = byts[tryidx++];
|
|
|
10617 c = *p++;
|
|
|
10618 lo = tryidx;
|
|
|
10619 hi = tryidx + len - 1;
|
|
|
10620 while (lo < hi)
|
|
|
10621 {
|
|
|
10622 m = (lo + hi) / 2;
|
|
|
10623 if (byts[m] > c)
|
|
|
10624 hi = m - 1;
|
|
|
10625 else if (byts[m] < c)
|
|
|
10626 lo = m + 1;
|
|
|
10627 else
|
|
|
10628 {
|
|
|
10629 lo = hi = m;
|
|
|
10630 break;
|
|
|
10631 }
|
|
|
10632 }
|
|
|
10633
|
|
|
10634 /* Stop if there is no matching byte. */
|
|
|
10635 if (hi < lo || byts[lo] != c)
|
|
|
10636 break;
|
|
|
10637
|
|
|
10638 /* Continue at the child (if there is one). */
|
|
|
10639 tryidx = idxs[lo];
|
|
|
10640 }
|
|
|
10641
|
|
|
10642 if (l == 0)
|
|
|
10643 {
|
|
|
10644 /*
|
|
|
10645 * Found the matching char. Copy it to "kword" and go a
|
|
|
10646 * level deeper.
|
|
|
10647 */
|
|
|
10648 if (round[depth] == 1)
|
|
|
10649 {
|
|
|
10650 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
|
10651 flen);
|
|
|
10652 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
|
10653 }
|
|
|
10654 else
|
|
|
10655 {
|
|
|
10656 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
|
10657 ulen);
|
|
|
10658 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
|
10659 }
|
|
|
10660 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
|
10661 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
|
10662
|
|
|
10663 ++depth;
|
|
|
10664 arridx[depth] = tryidx;
|
|
|
10665 round[depth] = 0;
|
|
|
10666 }
|
|
|
10667 }
|
|
|
10668 }
|
|
|
10669
|
|
|
10670 /* Didn't find it: "cannot happen". */
|
|
|
10671 *kword = NUL;
|
|
|
10672 }
|
|
|
10673
|
|
|
10674 /*
|
|
344
|
10675 * Compute the sound-a-like score for suggestions in su->su_ga and add them to
|
|
|
10676 * su->su_sga.
|
|
|
10677 */
|
|
|
10678 static void
|
|
|
10679 score_comp_sal(su)
|
|
|
10680 suginfo_T *su;
|
|
|
10681 {
|
|
|
10682 langp_T *lp;
|
|
|
10683 char_u badsound[MAXWLEN];
|
|
|
10684 int i;
|
|
|
10685 suggest_T *stp;
|
|
|
10686 suggest_T *sstp;
|
|
|
10687 int score;
|
|
500
|
10688 int lpi;
|
|
344
|
10689
|
|
|
10690 if (ga_grow(&su->su_sga, su->su_ga.ga_len) == FAIL)
|
|
|
10691 return;
|
|
|
10692
|
|
|
10693 /* Use the sound-folding of the first language that supports it. */
|
|
503
|
10694 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
10695 {
|
|
|
10696 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
344
|
10697 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10698 {
|
|
|
10699 /* soundfold the bad word */
|
|
375
|
10700 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
10701
|
|
|
10702 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10703 {
|
|
|
10704 stp = &SUG(su->su_ga, i);
|
|
|
10705
|
|
351
|
10706 /* Case-fold the suggested word, sound-fold it and compute the
|
|
|
10707 * sound-a-like score. */
|
|
|
10708 score = stp_sal_score(stp, su, lp->lp_slang, badsound);
|
|
344
|
10709 if (score < SCORE_MAXMAX)
|
|
|
10710 {
|
|
|
10711 /* Add the suggestion. */
|
|
|
10712 sstp = &SUG(su->su_sga, su->su_sga.ga_len);
|
|
|
10713 sstp->st_word = vim_strsave(stp->st_word);
|
|
|
10714 if (sstp->st_word != NULL)
|
|
|
10715 {
|
|
|
10716 sstp->st_score = score;
|
|
|
10717 sstp->st_altscore = 0;
|
|
|
10718 sstp->st_orglen = stp->st_orglen;
|
|
|
10719 ++su->su_sga.ga_len;
|
|
|
10720 }
|
|
|
10721 }
|
|
|
10722 }
|
|
|
10723 break;
|
|
|
10724 }
|
|
500
|
10725 }
|
|
344
|
10726 }
|
|
|
10727
|
|
|
10728 /*
|
|
|
10729 * Combine the list of suggestions in su->su_ga and su->su_sga.
|
|
|
10730 * They are intwined.
|
|
|
10731 */
|
|
|
10732 static void
|
|
|
10733 score_combine(su)
|
|
|
10734 suginfo_T *su;
|
|
|
10735 {
|
|
|
10736 int i;
|
|
|
10737 int j;
|
|
|
10738 garray_T ga;
|
|
|
10739 garray_T *gap;
|
|
|
10740 langp_T *lp;
|
|
|
10741 suggest_T *stp;
|
|
|
10742 char_u *p;
|
|
|
10743 char_u badsound[MAXWLEN];
|
|
|
10744 int round;
|
|
500
|
10745 int lpi;
|
|
344
|
10746
|
|
|
10747 /* Add the alternate score to su_ga. */
|
|
503
|
10748 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
10749 {
|
|
|
10750 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
344
|
10751 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
10752 {
|
|
|
10753 /* soundfold the bad word */
|
|
375
|
10754 spell_soundfold(lp->lp_slang, su->su_fbadword, TRUE, badsound);
|
|
344
|
10755
|
|
|
10756 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
|
10757 {
|
|
|
10758 stp = &SUG(su->su_ga, i);
|
|
351
|
10759 stp->st_altscore = stp_sal_score(stp, su, lp->lp_slang,
|
|
|
10760 badsound);
|
|
344
|
10761 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
10762 stp->st_score = (stp->st_score * 3 + SCORE_BIG) / 4;
|
|
|
10763 else
|
|
|
10764 stp->st_score = (stp->st_score * 3
|
|
|
10765 + stp->st_altscore) / 4;
|
|
|
10766 stp->st_salscore = FALSE;
|
|
|
10767 }
|
|
|
10768 break;
|
|
|
10769 }
|
|
|
10770 }
|
|
|
10771
|
|
|
10772 /* Add the alternate score to su_sga. */
|
|
|
10773 for (i = 0; i < su->su_sga.ga_len; ++i)
|
|
|
10774 {
|
|
|
10775 stp = &SUG(su->su_sga, i);
|
|
|
10776 stp->st_altscore = spell_edit_score(su->su_badword, stp->st_word);
|
|
|
10777 if (stp->st_score == SCORE_MAXMAX)
|
|
|
10778 stp->st_score = (SCORE_BIG * 7 + stp->st_altscore) / 8;
|
|
|
10779 else
|
|
|
10780 stp->st_score = (stp->st_score * 7 + stp->st_altscore) / 8;
|
|
|
10781 stp->st_salscore = TRUE;
|
|
|
10782 }
|
|
|
10783
|
|
|
10784 /* Sort the suggestions and truncate at "maxcount" for both lists. */
|
|
|
10785 (void)cleanup_suggestions(&su->su_ga, su->su_maxscore, su->su_maxcount);
|
|
|
10786 (void)cleanup_suggestions(&su->su_sga, su->su_maxscore, su->su_maxcount);
|
|
|
10787
|
|
|
10788 ga_init2(&ga, (int)sizeof(suginfo_T), 1);
|
|
|
10789 if (ga_grow(&ga, su->su_ga.ga_len + su->su_sga.ga_len) == FAIL)
|
|
|
10790 return;
|
|
|
10791
|
|
|
10792 stp = &SUG(ga, 0);
|
|
|
10793 for (i = 0; i < su->su_ga.ga_len || i < su->su_sga.ga_len; ++i)
|
|
|
10794 {
|
|
|
10795 /* round 1: get a suggestion from su_ga
|
|
|
10796 * round 2: get a suggestion from su_sga */
|
|
|
10797 for (round = 1; round <= 2; ++round)
|
|
|
10798 {
|
|
|
10799 gap = round == 1 ? &su->su_ga : &su->su_sga;
|
|
|
10800 if (i < gap->ga_len)
|
|
|
10801 {
|
|
|
10802 /* Don't add a word if it's already there. */
|
|
|
10803 p = SUG(*gap, i).st_word;
|
|
|
10804 for (j = 0; j < ga.ga_len; ++j)
|
|
|
10805 if (STRCMP(stp[j].st_word, p) == 0)
|
|
|
10806 break;
|
|
|
10807 if (j == ga.ga_len)
|
|
|
10808 stp[ga.ga_len++] = SUG(*gap, i);
|
|
|
10809 else
|
|
|
10810 vim_free(p);
|
|
|
10811 }
|
|
|
10812 }
|
|
|
10813 }
|
|
|
10814
|
|
|
10815 ga_clear(&su->su_ga);
|
|
|
10816 ga_clear(&su->su_sga);
|
|
|
10817
|
|
|
10818 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
|
10819 if (ga.ga_len > su->su_maxcount)
|
|
|
10820 {
|
|
|
10821 for (i = su->su_maxcount; i < ga.ga_len; ++i)
|
|
|
10822 vim_free(stp[i].st_word);
|
|
|
10823 ga.ga_len = su->su_maxcount;
|
|
|
10824 }
|
|
|
10825
|
|
|
10826 su->su_ga = ga;
|
|
|
10827 }
|
|
|
10828
|
|
|
10829 /*
|
|
351
|
10830 * For the goodword in "stp" compute the soundalike score compared to the
|
|
|
10831 * badword.
|
|
|
10832 */
|
|
|
10833 static int
|
|
|
10834 stp_sal_score(stp, su, slang, badsound)
|
|
|
10835 suggest_T *stp;
|
|
|
10836 suginfo_T *su;
|
|
|
10837 slang_T *slang;
|
|
|
10838 char_u *badsound; /* sound-folded badword */
|
|
|
10839 {
|
|
|
10840 char_u *p;
|
|
534
|
10841 char_u *pbad;
|
|
|
10842 char_u *pgood;
|
|
351
|
10843 char_u badsound2[MAXWLEN];
|
|
|
10844 char_u fword[MAXWLEN];
|
|
|
10845 char_u goodsound[MAXWLEN];
|
|
534
|
10846 char_u goodword[MAXWLEN];
|
|
|
10847 int lendiff;
|
|
|
10848
|
|
|
10849 lendiff = (int)(su->su_badlen - stp->st_orglen);
|
|
|
10850 if (lendiff >= 0)
|
|
|
10851 pbad = badsound;
|
|
351
|
10852 else
|
|
|
10853 {
|
|
|
10854 /* soundfold the bad word with more characters following */
|
|
|
10855 (void)spell_casefold(su->su_badptr, stp->st_orglen, fword, MAXWLEN);
|
|
|
10856
|
|
|
10857 /* When joining two words the sound often changes a lot. E.g., "t he"
|
|
|
10858 * sounds like "t h" while "the" sounds like "@". Avoid that by
|
|
|
10859 * removing the space. Don't do it when the good word also contains a
|
|
|
10860 * space. */
|
|
|
10861 if (vim_iswhite(su->su_badptr[su->su_badlen])
|
|
|
10862 && *skiptowhite(stp->st_word) == NUL)
|
|
|
10863 for (p = fword; *(p = skiptowhite(p)) != NUL; )
|
|
|
10864 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
10865
|
|
375
|
10866 spell_soundfold(slang, fword, TRUE, badsound2);
|
|
534
|
10867 pbad = badsound2;
|
|
|
10868 }
|
|
|
10869
|
|
|
10870 if (lendiff > 0)
|
|
|
10871 {
|
|
|
10872 /* Add part of the bad word to the good word, so that we soundfold
|
|
|
10873 * what replaces the bad word. */
|
|
|
10874 STRCPY(goodword, stp->st_word);
|
|
|
10875 STRNCAT(goodword, su->su_badptr + su->su_badlen - lendiff, lendiff);
|
|
|
10876 pgood = goodword;
|
|
|
10877 }
|
|
|
10878 else
|
|
|
10879 pgood = stp->st_word;
|
|
351
|
10880
|
|
375
|
10881 /* Sound-fold the word and compute the score for the difference. */
|
|
534
|
10882 spell_soundfold(slang, pgood, FALSE, goodsound);
|
|
|
10883
|
|
|
10884 return soundalike_score(goodsound, pbad);
|
|
351
|
10885 }
|
|
|
10886
|
|
|
10887 /*
|
|
323
|
10888 * Find suggestions by comparing the word in a sound-a-like form.
|
|
503
|
10889 * Note: This doesn't support postponed prefixes.
|
|
323
|
10890 */
|
|
|
10891 static void
|
|
346
|
10892 suggest_try_soundalike(su)
|
|
323
|
10893 suginfo_T *su;
|
|
|
10894 {
|
|
|
10895 char_u salword[MAXWLEN];
|
|
|
10896 char_u tword[MAXWLEN];
|
|
|
10897 char_u tsalword[MAXWLEN];
|
|
324
|
10898 idx_T arridx[MAXWLEN];
|
|
323
|
10899 int curi[MAXWLEN];
|
|
|
10900 langp_T *lp;
|
|
|
10901 char_u *byts;
|
|
324
|
10902 idx_T *idxs;
|
|
323
|
10903 int depth;
|
|
|
10904 int c;
|
|
324
|
10905 idx_T n;
|
|
323
|
10906 int round;
|
|
|
10907 int flags;
|
|
344
|
10908 int sound_score;
|
|
483
|
10909 int local_score;
|
|
500
|
10910 int lpi;
|
|
503
|
10911 slang_T *slang;
|
|
344
|
10912
|
|
|
10913 /* Do this for all languages that support sound folding. */
|
|
503
|
10914 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
10915 {
|
|
|
10916 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
|
10917 slang = lp->lp_slang;
|
|
|
10918 if (slang->sl_sal.ga_len > 0)
|
|
323
|
10919 {
|
|
|
10920 /* soundfold the bad word */
|
|
503
|
10921 spell_soundfold(slang, su->su_fbadword, TRUE, salword);
|
|
323
|
10922
|
|
|
10923 /*
|
|
|
10924 * Go through the whole tree, soundfold each word and compare.
|
|
|
10925 * round 1: use the case-folded tree.
|
|
|
10926 * round 2: use the keep-case tree.
|
|
|
10927 */
|
|
|
10928 for (round = 1; round <= 2; ++round)
|
|
|
10929 {
|
|
|
10930 if (round == 1)
|
|
|
10931 {
|
|
503
|
10932 byts = slang->sl_fbyts;
|
|
|
10933 idxs = slang->sl_fidxs;
|
|
323
|
10934 }
|
|
|
10935 else
|
|
|
10936 {
|
|
503
|
10937 byts = slang->sl_kbyts;
|
|
|
10938 idxs = slang->sl_kidxs;
|
|
389
|
10939 if (byts == NULL) /* no keep-case words */
|
|
|
10940 continue;
|
|
323
|
10941 }
|
|
|
10942
|
|
|
10943 depth = 0;
|
|
|
10944 arridx[0] = 0;
|
|
|
10945 curi[0] = 1;
|
|
|
10946 while (depth >= 0 && !got_int)
|
|
|
10947 {
|
|
|
10948 if (curi[depth] > byts[arridx[depth]])
|
|
351
|
10949 {
|
|
323
|
10950 /* Done all bytes at this node, go up one level. */
|
|
|
10951 --depth;
|
|
351
|
10952 line_breakcheck();
|
|
|
10953 }
|
|
323
|
10954 else
|
|
|
10955 {
|
|
|
10956 /* Do one more byte at this node. */
|
|
|
10957 n = arridx[depth] + curi[depth];
|
|
|
10958 ++curi[depth];
|
|
|
10959 c = byts[n];
|
|
|
10960 if (c == 0)
|
|
|
10961 {
|
|
|
10962 /* End of word, deal with the word. */
|
|
324
|
10963 flags = (int)idxs[n];
|
|
323
|
10964 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
|
10965 {
|
|
|
10966 tword[depth] = NUL;
|
|
375
|
10967 /* Sound-fold. Only in keep-case tree need to
|
|
|
10968 * case-fold the word. */
|
|
503
|
10969 spell_soundfold(slang, tword,
|
|
375
|
10970 round == 1, tsalword);
|
|
323
|
10971
|
|
344
|
10972 /* Compute the edit distance between the
|
|
|
10973 * sound-a-like words. */
|
|
|
10974 sound_score = soundalike_score(salword,
|
|
|
10975 tsalword);
|
|
483
|
10976
|
|
|
10977 /* Add a penalty for words in another region. */
|
|
|
10978 if ((flags & WF_REGION) && (((unsigned)flags
|
|
|
10979 >> 16) & lp->lp_region) == 0)
|
|
|
10980 local_score = SCORE_REGION;
|
|
|
10981 else
|
|
|
10982 local_score = 0;
|
|
|
10983 sound_score += local_score;
|
|
|
10984
|
|
324
|
10985 if (sound_score < SCORE_MAXMAX)
|
|
323
|
10986 {
|
|
324
|
10987 char_u cword[MAXWLEN];
|
|
|
10988 char_u *p;
|
|
344
|
10989 int score;
|
|
324
|
10990
|
|
388
|
10991 flags |= su->su_badflags;
|
|
351
|
10992 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
323
|
10993 {
|
|
324
|
10994 /* Need to fix case according to
|
|
|
10995 * "flags". */
|
|
323
|
10996 make_case_word(tword, cword, flags);
|
|
324
|
10997 p = cword;
|
|
323
|
10998 }
|
|
|
10999 else
|
|
324
|
11000 p = tword;
|
|
|
11001
|
|
344
|
11002 if (sps_flags & SPS_DOUBLE)
|
|
|
11003 add_suggestion(su, &su->su_sga, p,
|
|
346
|
11004 su->su_badlen,
|
|
503
|
11005 sound_score, 0, FALSE,
|
|
|
11006 lp->lp_sallang);
|
|
344
|
11007 else
|
|
|
11008 {
|
|
|
11009 /* Compute the score. */
|
|
|
11010 score = spell_edit_score(
|
|
483
|
11011 su->su_badword, p)
|
|
|
11012 + local_score;
|
|
344
|
11013 if (sps_flags & SPS_BEST)
|
|
|
11014 /* give a bonus for the good word
|
|
|
11015 * sounding the same as the bad
|
|
|
11016 * word */
|
|
|
11017 add_suggestion(su, &su->su_ga, p,
|
|
346
|
11018 su->su_badlen,
|
|
344
|
11019 RESCORE(score, sound_score),
|
|
503
|
11020 sound_score, TRUE,
|
|
|
11021 lp->lp_sallang);
|
|
344
|
11022 else
|
|
|
11023 add_suggestion(su, &su->su_ga, p,
|
|
346
|
11024 su->su_badlen,
|
|
503
|
11025 score + sound_score,
|
|
|
11026 0, FALSE,
|
|
|
11027 lp->lp_sallang);
|
|
344
|
11028 }
|
|
323
|
11029 }
|
|
|
11030 }
|
|
|
11031
|
|
|
11032 /* Skip over other NUL bytes. */
|
|
|
11033 while (byts[n + 1] == 0)
|
|
|
11034 {
|
|
|
11035 ++n;
|
|
|
11036 ++curi[depth];
|
|
|
11037 }
|
|
|
11038 }
|
|
|
11039 else
|
|
|
11040 {
|
|
|
11041 /* Normal char, go one level deeper. */
|
|
|
11042 tword[depth++] = c;
|
|
|
11043 arridx[depth] = idxs[n];
|
|
|
11044 curi[depth] = 1;
|
|
|
11045 }
|
|
|
11046 }
|
|
|
11047 }
|
|
|
11048 }
|
|
|
11049 }
|
|
|
11050 }
|
|
|
11051 }
|
|
|
11052
|
|
|
11053 /*
|
|
324
|
11054 * Copy "fword" to "cword", fixing case according to "flags".
|
|
323
|
11055 */
|
|
|
11056 static void
|
|
|
11057 make_case_word(fword, cword, flags)
|
|
|
11058 char_u *fword;
|
|
|
11059 char_u *cword;
|
|
|
11060 int flags;
|
|
|
11061 {
|
|
|
11062 if (flags & WF_ALLCAP)
|
|
|
11063 /* Make it all upper-case */
|
|
|
11064 allcap_copy(fword, cword);
|
|
|
11065 else if (flags & WF_ONECAP)
|
|
|
11066 /* Make the first letter upper-case */
|
|
324
|
11067 onecap_copy(fword, cword, TRUE);
|
|
323
|
11068 else
|
|
|
11069 /* Use goodword as-is. */
|
|
|
11070 STRCPY(cword, fword);
|
|
|
11071 }
|
|
|
11072
|
|
330
|
11073 /*
|
|
|
11074 * Use map string "map" for languages "lp".
|
|
|
11075 */
|
|
|
11076 static void
|
|
|
11077 set_map_str(lp, map)
|
|
|
11078 slang_T *lp;
|
|
|
11079 char_u *map;
|
|
|
11080 {
|
|
|
11081 char_u *p;
|
|
|
11082 int headc = 0;
|
|
|
11083 int c;
|
|
|
11084 int i;
|
|
|
11085
|
|
|
11086 if (*map == NUL)
|
|
|
11087 {
|
|
|
11088 lp->sl_has_map = FALSE;
|
|
|
11089 return;
|
|
|
11090 }
|
|
|
11091 lp->sl_has_map = TRUE;
|
|
|
11092
|
|
|
11093 /* Init the array and hash table empty. */
|
|
|
11094 for (i = 0; i < 256; ++i)
|
|
|
11095 lp->sl_map_array[i] = 0;
|
|
|
11096 #ifdef FEAT_MBYTE
|
|
|
11097 hash_init(&lp->sl_map_hash);
|
|
|
11098 #endif
|
|
|
11099
|
|
|
11100 /*
|
|
|
11101 * The similar characters are stored separated with slashes:
|
|
|
11102 * "aaa/bbb/ccc/". Fill sl_map_array[c] with the character before c and
|
|
|
11103 * before the same slash. For characters above 255 sl_map_hash is used.
|
|
|
11104 */
|
|
|
11105 for (p = map; *p != NUL; )
|
|
|
11106 {
|
|
|
11107 #ifdef FEAT_MBYTE
|
|
474
|
11108 c = mb_cptr2char_adv(&p);
|
|
330
|
11109 #else
|
|
|
11110 c = *p++;
|
|
|
11111 #endif
|
|
|
11112 if (c == '/')
|
|
|
11113 headc = 0;
|
|
|
11114 else
|
|
|
11115 {
|
|
|
11116 if (headc == 0)
|
|
|
11117 headc = c;
|
|
|
11118
|
|
|
11119 #ifdef FEAT_MBYTE
|
|
|
11120 /* Characters above 255 don't fit in sl_map_array[], put them in
|
|
|
11121 * the hash table. Each entry is the char, a NUL the headchar and
|
|
|
11122 * a NUL. */
|
|
|
11123 if (c >= 256)
|
|
|
11124 {
|
|
|
11125 int cl = mb_char2len(c);
|
|
|
11126 int headcl = mb_char2len(headc);
|
|
|
11127 char_u *b;
|
|
|
11128 hash_T hash;
|
|
|
11129 hashitem_T *hi;
|
|
|
11130
|
|
|
11131 b = alloc((unsigned)(cl + headcl + 2));
|
|
|
11132 if (b == NULL)
|
|
|
11133 return;
|
|
|
11134 mb_char2bytes(c, b);
|
|
|
11135 b[cl] = NUL;
|
|
|
11136 mb_char2bytes(headc, b + cl + 1);
|
|
|
11137 b[cl + 1 + headcl] = NUL;
|
|
|
11138 hash = hash_hash(b);
|
|
|
11139 hi = hash_lookup(&lp->sl_map_hash, b, hash);
|
|
|
11140 if (HASHITEM_EMPTY(hi))
|
|
|
11141 hash_add_item(&lp->sl_map_hash, hi, b, hash);
|
|
|
11142 else
|
|
|
11143 {
|
|
|
11144 /* This should have been checked when generating the .spl
|
|
|
11145 * file. */
|
|
|
11146 EMSG(_("E999: duplicate char in MAP entry"));
|
|
|
11147 vim_free(b);
|
|
|
11148 }
|
|
|
11149 }
|
|
|
11150 else
|
|
|
11151 #endif
|
|
|
11152 lp->sl_map_array[c] = headc;
|
|
|
11153 }
|
|
|
11154 }
|
|
|
11155 }
|
|
|
11156
|
|
323
|
11157 /*
|
|
|
11158 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
|
11159 * lines in the .aff file.
|
|
|
11160 */
|
|
|
11161 static int
|
|
|
11162 similar_chars(slang, c1, c2)
|
|
|
11163 slang_T *slang;
|
|
|
11164 int c1;
|
|
|
11165 int c2;
|
|
|
11166 {
|
|
330
|
11167 int m1, m2;
|
|
|
11168 #ifdef FEAT_MBYTE
|
|
|
11169 char_u buf[MB_MAXBYTES];
|
|
|
11170 hashitem_T *hi;
|
|
|
11171
|
|
|
11172 if (c1 >= 256)
|
|
|
11173 {
|
|
|
11174 buf[mb_char2bytes(c1, buf)] = 0;
|
|
|
11175 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
11176 if (HASHITEM_EMPTY(hi))
|
|
|
11177 m1 = 0;
|
|
|
11178 else
|
|
|
11179 m1 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
11180 }
|
|
|
11181 else
|
|
|
11182 #endif
|
|
|
11183 m1 = slang->sl_map_array[c1];
|
|
|
11184 if (m1 == 0)
|
|
323
|
11185 return FALSE;
|
|
330
|
11186
|
|
|
11187
|
|
|
11188 #ifdef FEAT_MBYTE
|
|
|
11189 if (c2 >= 256)
|
|
|
11190 {
|
|
|
11191 buf[mb_char2bytes(c2, buf)] = 0;
|
|
|
11192 hi = hash_find(&slang->sl_map_hash, buf);
|
|
|
11193 if (HASHITEM_EMPTY(hi))
|
|
|
11194 m2 = 0;
|
|
|
11195 else
|
|
|
11196 m2 = mb_ptr2char(hi->hi_key + STRLEN(hi->hi_key) + 1);
|
|
|
11197 }
|
|
|
11198 else
|
|
|
11199 #endif
|
|
|
11200 m2 = slang->sl_map_array[c2];
|
|
|
11201
|
|
|
11202 return m1 == m2;
|
|
323
|
11203 }
|
|
|
11204
|
|
|
11205 /*
|
|
|
11206 * Add a suggestion to the list of suggestions.
|
|
|
11207 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
|
11208 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
|
11209 * with spell_edit_score().
|
|
|
11210 */
|
|
|
11211 static void
|
|
534
|
11212 add_suggestion(su, gap, goodword, badlenarg, score, altscore, had_bonus, slang)
|
|
323
|
11213 suginfo_T *su;
|
|
344
|
11214 garray_T *gap;
|
|
323
|
11215 char_u *goodword;
|
|
534
|
11216 int badlenarg; /* len of bad word replaced with "goodword" */
|
|
324
|
11217 int score;
|
|
351
|
11218 int altscore;
|
|
344
|
11219 int had_bonus; /* value for st_had_bonus */
|
|
503
|
11220 slang_T *slang; /* language for sound folding */
|
|
323
|
11221 {
|
|
534
|
11222 int goodlen = STRLEN(goodword); /* len of goodword changed */
|
|
|
11223 int badlen = badlenarg; /* len of bad word changed */
|
|
323
|
11224 suggest_T *stp;
|
|
534
|
11225 suggest_T new_sug;
|
|
323
|
11226 int i;
|
|
534
|
11227 hlf_T attr = HLF_COUNT;
|
|
532
|
11228 char_u longword[MAXWLEN + 1];
|
|
534
|
11229 char_u *pgood, *pbad;
|
|
532
|
11230
|
|
|
11231 /* Check that the word really is valid. Esp. for banned words and for
|
|
|
11232 * split words, such as "the the". Need to append what follows to check
|
|
|
11233 * for that. */
|
|
|
11234 STRCPY(longword, goodword);
|
|
|
11235 vim_strncpy(longword + goodlen, su->su_badptr + badlen, MAXWLEN - goodlen);
|
|
|
11236 (void)spell_check(curwin, longword, &attr, NULL);
|
|
534
|
11237 if (attr != HLF_COUNT)
|
|
323
|
11238 return;
|
|
|
11239
|
|
534
|
11240 /* Minimize "badlen" for consistency. Avoids that changing "the the" to
|
|
|
11241 * "thee the" is added next to changing the first "the" the "thee". */
|
|
|
11242 pgood = goodword + STRLEN(goodword);
|
|
|
11243 pbad = su->su_badptr + badlen;
|
|
|
11244 while (pgood > goodword && pbad > su->su_badptr)
|
|
|
11245 {
|
|
|
11246 mb_ptr_back(goodword, pgood);
|
|
|
11247 mb_ptr_back(su->su_badptr, pbad);
|
|
|
11248 #ifdef FEAT_MBYTE
|
|
|
11249 if (has_mbyte)
|
|
|
11250 {
|
|
|
11251 if (mb_ptr2char(pgood) != mb_ptr2char(pbad))
|
|
|
11252 break;
|
|
346
|
11253 }
|
|
|
11254 else
|
|
534
|
11255 #endif
|
|
|
11256 if (*pgood != *pbad)
|
|
|
11257 break;
|
|
|
11258 badlen = pbad - su->su_badptr;
|
|
|
11259 goodlen = pgood - goodword;
|
|
|
11260 }
|
|
|
11261 if (badlen == 0 && goodlen == 0)
|
|
|
11262 /* goodword doesn't change anything; may happen for "the the" changing
|
|
|
11263 * the first "the" to itself. */
|
|
|
11264 return;
|
|
346
|
11265
|
|
323
|
11266 if (score <= su->su_maxscore)
|
|
|
11267 {
|
|
366
|
11268 /* Check if the word is already there. Also check the length that is
|
|
|
11269 * being replaced "thes," -> "these" is a different suggestion from
|
|
|
11270 * "thes" -> "these". */
|
|
344
|
11271 stp = &SUG(*gap, 0);
|
|
|
11272 for (i = gap->ga_len - 1; i >= 0; --i)
|
|
535
|
11273 if ((int)STRLEN(stp[i].st_word) == goodlen
|
|
534
|
11274 && STRNCMP(stp[i].st_word, goodword, goodlen) == 0
|
|
366
|
11275 && stp[i].st_orglen == badlen)
|
|
323
|
11276 {
|
|
534
|
11277 /*
|
|
535
|
11278 * Found it. Remember the word with the lowest score.
|
|
534
|
11279 */
|
|
503
|
11280 if (stp[i].st_slang == NULL)
|
|
|
11281 stp[i].st_slang = slang;
|
|
534
|
11282
|
|
|
11283 new_sug.st_score = score;
|
|
|
11284 new_sug.st_altscore = altscore;
|
|
|
11285 new_sug.st_had_bonus = had_bonus;
|
|
|
11286
|
|
|
11287 if (stp[i].st_had_bonus != had_bonus)
|
|
|
11288 {
|
|
|
11289 /* Only one of the two had the soundalike score computed.
|
|
|
11290 * Need to do that for the other one now, otherwise the
|
|
|
11291 * scores can't be compared. This happens because
|
|
|
11292 * suggest_try_change() doesn't compute the soundalike
|
|
535
|
11293 * word to keep it fast, while some special methods set
|
|
|
11294 * the soundalike score to zero. */
|
|
534
|
11295 if (had_bonus)
|
|
|
11296 rescore_one(su, &stp[i]);
|
|
|
11297 else
|
|
|
11298 {
|
|
|
11299 new_sug.st_word = goodword;
|
|
|
11300 new_sug.st_slang = stp[i].st_slang;
|
|
|
11301 new_sug.st_orglen = badlen;
|
|
|
11302 rescore_one(su, &new_sug);
|
|
|
11303 }
|
|
|
11304 }
|
|
|
11305
|
|
|
11306 if (stp[i].st_score > new_sug.st_score)
|
|
|
11307 {
|
|
|
11308 stp[i].st_score = new_sug.st_score;
|
|
|
11309 stp[i].st_altscore = new_sug.st_altscore;
|
|
|
11310 stp[i].st_had_bonus = new_sug.st_had_bonus;
|
|
|
11311 }
|
|
323
|
11312 break;
|
|
|
11313 }
|
|
|
11314
|
|
344
|
11315 if (i < 0 && ga_grow(gap, 1) == OK)
|
|
323
|
11316 {
|
|
|
11317 /* Add a suggestion. */
|
|
344
|
11318 stp = &SUG(*gap, gap->ga_len);
|
|
534
|
11319 stp->st_word = vim_strnsave(goodword, goodlen);
|
|
323
|
11320 if (stp->st_word != NULL)
|
|
|
11321 {
|
|
|
11322 stp->st_score = score;
|
|
351
|
11323 stp->st_altscore = altscore;
|
|
324
|
11324 stp->st_had_bonus = had_bonus;
|
|
346
|
11325 stp->st_orglen = badlen;
|
|
503
|
11326 stp->st_slang = slang;
|
|
344
|
11327 ++gap->ga_len;
|
|
323
|
11328
|
|
|
11329 /* If we have too many suggestions now, sort the list and keep
|
|
|
11330 * the best suggestions. */
|
|
344
|
11331 if (gap->ga_len > SUG_MAX_COUNT(su))
|
|
|
11332 su->su_maxscore = cleanup_suggestions(gap, su->su_maxscore,
|
|
|
11333 SUG_CLEAN_COUNT(su));
|
|
323
|
11334 }
|
|
|
11335 }
|
|
|
11336 }
|
|
|
11337 }
|
|
|
11338
|
|
|
11339 /*
|
|
|
11340 * Add a word to be banned.
|
|
|
11341 */
|
|
|
11342 static void
|
|
|
11343 add_banned(su, word)
|
|
|
11344 suginfo_T *su;
|
|
|
11345 char_u *word;
|
|
|
11346 {
|
|
|
11347 char_u *s = vim_strsave(word);
|
|
|
11348 hash_T hash;
|
|
|
11349 hashitem_T *hi;
|
|
|
11350
|
|
|
11351 if (s != NULL)
|
|
|
11352 {
|
|
|
11353 hash = hash_hash(s);
|
|
|
11354 hi = hash_lookup(&su->su_banned, s, hash);
|
|
|
11355 if (HASHITEM_EMPTY(hi))
|
|
|
11356 hash_add_item(&su->su_banned, hi, s, hash);
|
|
355
|
11357 else
|
|
|
11358 vim_free(s);
|
|
323
|
11359 }
|
|
|
11360 }
|
|
|
11361
|
|
|
11362 /*
|
|
|
11363 * Return TRUE if a word appears in the list of banned words.
|
|
|
11364 */
|
|
|
11365 static int
|
|
|
11366 was_banned(su, word)
|
|
|
11367 suginfo_T *su;
|
|
|
11368 char_u *word;
|
|
|
11369 {
|
|
324
|
11370 hashitem_T *hi = hash_find(&su->su_banned, word);
|
|
|
11371
|
|
|
11372 return !HASHITEM_EMPTY(hi);
|
|
323
|
11373 }
|
|
|
11374
|
|
|
11375 /*
|
|
|
11376 * Free the banned words in "su".
|
|
|
11377 */
|
|
|
11378 static void
|
|
|
11379 free_banned(su)
|
|
|
11380 suginfo_T *su;
|
|
|
11381 {
|
|
|
11382 int todo;
|
|
|
11383 hashitem_T *hi;
|
|
|
11384
|
|
|
11385 todo = su->su_banned.ht_used;
|
|
|
11386 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
|
11387 {
|
|
|
11388 if (!HASHITEM_EMPTY(hi))
|
|
|
11389 {
|
|
|
11390 vim_free(hi->hi_key);
|
|
|
11391 --todo;
|
|
|
11392 }
|
|
|
11393 }
|
|
|
11394 hash_clear(&su->su_banned);
|
|
|
11395 }
|
|
|
11396
|
|
324
|
11397 /*
|
|
534
|
11398 * Recompute the score for all suggestions if sound-folding is possible. This
|
|
|
11399 * is slow, thus only done for the final results.
|
|
324
|
11400 */
|
|
|
11401 static void
|
|
|
11402 rescore_suggestions(su)
|
|
|
11403 suginfo_T *su;
|
|
|
11404 {
|
|
|
11405 int i;
|
|
534
|
11406
|
|
|
11407 if (su->su_sallang != NULL)
|
|
503
|
11408 for (i = 0; i < su->su_ga.ga_len; ++i)
|
|
534
|
11409 rescore_one(su, &SUG(su->su_ga, i));
|
|
|
11410 }
|
|
|
11411
|
|
|
11412 /*
|
|
|
11413 * Recompute the score for one suggestion if sound-folding is possible.
|
|
|
11414 */
|
|
|
11415 static void
|
|
|
11416 rescore_one(su, stp)
|
|
535
|
11417 suginfo_T *su;
|
|
|
11418 suggest_T *stp;
|
|
534
|
11419 {
|
|
|
11420 slang_T *slang = stp->st_slang;
|
|
|
11421 char_u sal_badword[MAXWLEN];
|
|
535
|
11422 char_u *p;
|
|
534
|
11423
|
|
|
11424 /* Only rescore suggestions that have no sal score yet and do have a
|
|
|
11425 * language. */
|
|
|
11426 if (slang != NULL && slang->sl_sal.ga_len > 0 && !stp->st_had_bonus)
|
|
|
11427 {
|
|
|
11428 if (slang == su->su_sallang)
|
|
535
|
11429 p = su->su_sal_badword;
|
|
534
|
11430 else
|
|
|
11431 {
|
|
|
11432 spell_soundfold(slang, su->su_fbadword, TRUE, sal_badword);
|
|
535
|
11433 p = sal_badword;
|
|
|
11434 }
|
|
|
11435
|
|
|
11436 stp->st_altscore = stp_sal_score(stp, su, slang, p);
|
|
534
|
11437 if (stp->st_altscore == SCORE_MAXMAX)
|
|
|
11438 stp->st_altscore = SCORE_BIG;
|
|
|
11439 stp->st_score = RESCORE(stp->st_score, stp->st_altscore);
|
|
|
11440 stp->st_had_bonus = TRUE;
|
|
324
|
11441 }
|
|
|
11442 }
|
|
|
11443
|
|
323
|
11444 static int
|
|
|
11445 #ifdef __BORLANDC__
|
|
|
11446 _RTLENTRYF
|
|
|
11447 #endif
|
|
|
11448 sug_compare __ARGS((const void *s1, const void *s2));
|
|
|
11449
|
|
|
11450 /*
|
|
|
11451 * Function given to qsort() to sort the suggestions on st_score.
|
|
529
|
11452 * First on "st_score", then "st_altscore" then alphabetically.
|
|
323
|
11453 */
|
|
|
11454 static int
|
|
|
11455 #ifdef __BORLANDC__
|
|
|
11456 _RTLENTRYF
|
|
|
11457 #endif
|
|
|
11458 sug_compare(s1, s2)
|
|
|
11459 const void *s1;
|
|
|
11460 const void *s2;
|
|
|
11461 {
|
|
|
11462 suggest_T *p1 = (suggest_T *)s1;
|
|
|
11463 suggest_T *p2 = (suggest_T *)s2;
|
|
344
|
11464 int n = p1->st_score - p2->st_score;
|
|
|
11465
|
|
|
11466 if (n == 0)
|
|
529
|
11467 {
|
|
|
11468 n = p1->st_altscore - p2->st_altscore;
|
|
|
11469 if (n == 0)
|
|
|
11470 n = STRICMP(p1->st_word, p2->st_word);
|
|
|
11471 }
|
|
344
|
11472 return n;
|
|
323
|
11473 }
|
|
|
11474
|
|
|
11475 /*
|
|
|
11476 * Cleanup the suggestions:
|
|
|
11477 * - Sort on score.
|
|
|
11478 * - Remove words that won't be displayed.
|
|
344
|
11479 * Returns the maximum score in the list or "maxscore" unmodified.
|
|
323
|
11480 */
|
|
344
|
11481 static int
|
|
|
11482 cleanup_suggestions(gap, maxscore, keep)
|
|
|
11483 garray_T *gap;
|
|
|
11484 int maxscore;
|
|
324
|
11485 int keep; /* nr of suggestions to keep */
|
|
323
|
11486 {
|
|
344
|
11487 suggest_T *stp = &SUG(*gap, 0);
|
|
323
|
11488 int i;
|
|
|
11489
|
|
|
11490 /* Sort the list. */
|
|
344
|
11491 qsort(gap->ga_data, (size_t)gap->ga_len, sizeof(suggest_T), sug_compare);
|
|
323
|
11492
|
|
|
11493 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
344
|
11494 if (gap->ga_len > keep)
|
|
323
|
11495 {
|
|
344
|
11496 for (i = keep; i < gap->ga_len; ++i)
|
|
323
|
11497 vim_free(stp[i].st_word);
|
|
344
|
11498 gap->ga_len = keep;
|
|
|
11499 return stp[keep - 1].st_score;
|
|
323
|
11500 }
|
|
344
|
11501 return maxscore;
|
|
323
|
11502 }
|
|
|
11503
|
|
372
|
11504 #if defined(FEAT_EVAL) || defined(PROTO)
|
|
|
11505 /*
|
|
|
11506 * Soundfold a string, for soundfold().
|
|
|
11507 * Result is in allocated memory, NULL for an error.
|
|
|
11508 */
|
|
|
11509 char_u *
|
|
|
11510 eval_soundfold(word)
|
|
|
11511 char_u *word;
|
|
|
11512 {
|
|
|
11513 langp_T *lp;
|
|
|
11514 char_u sound[MAXWLEN];
|
|
500
|
11515 int lpi;
|
|
372
|
11516
|
|
|
11517 if (curwin->w_p_spell && *curbuf->b_p_spl != NUL)
|
|
|
11518 /* Use the sound-folding of the first language that supports it. */
|
|
503
|
11519 for (lpi = 0; lpi < curbuf->b_langp.ga_len; ++lpi)
|
|
|
11520 {
|
|
|
11521 lp = LANGP_ENTRY(curbuf->b_langp, lpi);
|
|
372
|
11522 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
|
11523 {
|
|
|
11524 /* soundfold the word */
|
|
375
|
11525 spell_soundfold(lp->lp_slang, word, FALSE, sound);
|
|
372
|
11526 return vim_strsave(sound);
|
|
|
11527 }
|
|
500
|
11528 }
|
|
372
|
11529
|
|
|
11530 /* No language with sound folding, return word as-is. */
|
|
|
11531 return vim_strsave(word);
|
|
|
11532 }
|
|
|
11533 #endif
|
|
|
11534
|
|
323
|
11535 /*
|
|
|
11536 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
485
|
11537 *
|
|
|
11538 * There are many ways to turn a word into a sound-a-like representation. The
|
|
|
11539 * oldest is Soundex (1918!). A nice overview can be found in "Approximate
|
|
|
11540 * swedish name matching - survey and test of different algorithms" by Klas
|
|
|
11541 * Erikson.
|
|
|
11542 *
|
|
|
11543 * We support two methods:
|
|
|
11544 * 1. SOFOFROM/SOFOTO do a simple character mapping.
|
|
|
11545 * 2. SAL items define a more advanced sound-folding (and much slower).
|
|
323
|
11546 */
|
|
|
11547 static void
|
|
375
|
11548 spell_soundfold(slang, inword, folded, res)
|
|
|
11549 slang_T *slang;
|
|
|
11550 char_u *inword;
|
|
|
11551 int folded; /* "inword" is already case-folded */
|
|
|
11552 char_u *res;
|
|
|
11553 {
|
|
|
11554 char_u fword[MAXWLEN];
|
|
|
11555 char_u *word;
|
|
|
11556
|
|
|
11557 if (slang->sl_sofo)
|
|
|
11558 /* SOFOFROM and SOFOTO used */
|
|
|
11559 spell_soundfold_sofo(slang, inword, res);
|
|
|
11560 else
|
|
|
11561 {
|
|
|
11562 /* SAL items used. Requires the word to be case-folded. */
|
|
|
11563 if (folded)
|
|
|
11564 word = inword;
|
|
|
11565 else
|
|
|
11566 {
|
|
|
11567 (void)spell_casefold(inword, STRLEN(inword), fword, MAXWLEN);
|
|
|
11568 word = fword;
|
|
|
11569 }
|
|
|
11570
|
|
|
11571 #ifdef FEAT_MBYTE
|
|
|
11572 if (has_mbyte)
|
|
|
11573 spell_soundfold_wsal(slang, word, res);
|
|
|
11574 else
|
|
|
11575 #endif
|
|
|
11576 spell_soundfold_sal(slang, word, res);
|
|
|
11577 }
|
|
|
11578 }
|
|
|
11579
|
|
|
11580 /*
|
|
|
11581 * Perform sound folding of "inword" into "res" according to SOFOFROM and
|
|
|
11582 * SOFOTO lines.
|
|
|
11583 */
|
|
|
11584 static void
|
|
|
11585 spell_soundfold_sofo(slang, inword, res)
|
|
|
11586 slang_T *slang;
|
|
|
11587 char_u *inword;
|
|
|
11588 char_u *res;
|
|
|
11589 {
|
|
|
11590 char_u *s;
|
|
|
11591 int ri = 0;
|
|
|
11592 int c;
|
|
|
11593
|
|
|
11594 #ifdef FEAT_MBYTE
|
|
|
11595 if (has_mbyte)
|
|
|
11596 {
|
|
|
11597 int prevc = 0;
|
|
|
11598 int *ip;
|
|
|
11599
|
|
|
11600 /* The sl_sal_first[] table contains the translation for chars up to
|
|
|
11601 * 255, sl_sal the rest. */
|
|
|
11602 for (s = inword; *s != NUL; )
|
|
|
11603 {
|
|
474
|
11604 c = mb_cptr2char_adv(&s);
|
|
375
|
11605 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
11606 c = ' ';
|
|
|
11607 else if (c < 256)
|
|
|
11608 c = slang->sl_sal_first[c];
|
|
|
11609 else
|
|
|
11610 {
|
|
|
11611 ip = ((int **)slang->sl_sal.ga_data)[c & 0xff];
|
|
|
11612 if (ip == NULL) /* empty list, can't match */
|
|
|
11613 c = NUL;
|
|
|
11614 else
|
|
|
11615 for (;;) /* find "c" in the list */
|
|
|
11616 {
|
|
|
11617 if (*ip == 0) /* not found */
|
|
|
11618 {
|
|
|
11619 c = NUL;
|
|
|
11620 break;
|
|
|
11621 }
|
|
|
11622 if (*ip == c) /* match! */
|
|
|
11623 {
|
|
|
11624 c = ip[1];
|
|
|
11625 break;
|
|
|
11626 }
|
|
|
11627 ip += 2;
|
|
|
11628 }
|
|
|
11629 }
|
|
|
11630
|
|
|
11631 if (c != NUL && c != prevc)
|
|
|
11632 {
|
|
|
11633 ri += mb_char2bytes(c, res + ri);
|
|
|
11634 if (ri + MB_MAXBYTES > MAXWLEN)
|
|
|
11635 break;
|
|
|
11636 prevc = c;
|
|
|
11637 }
|
|
|
11638 }
|
|
|
11639 }
|
|
|
11640 else
|
|
|
11641 #endif
|
|
|
11642 {
|
|
|
11643 /* The sl_sal_first[] table contains the translation. */
|
|
|
11644 for (s = inword; (c = *s) != NUL; ++s)
|
|
|
11645 {
|
|
|
11646 if (vim_iswhite(c))
|
|
|
11647 c = ' ';
|
|
|
11648 else
|
|
|
11649 c = slang->sl_sal_first[c];
|
|
|
11650 if (c != NUL && (ri == 0 || res[ri - 1] != c))
|
|
|
11651 res[ri++] = c;
|
|
|
11652 }
|
|
|
11653 }
|
|
|
11654
|
|
|
11655 res[ri] = NUL;
|
|
|
11656 }
|
|
|
11657
|
|
|
11658 static void
|
|
|
11659 spell_soundfold_sal(slang, inword, res)
|
|
323
|
11660 slang_T *slang;
|
|
|
11661 char_u *inword;
|
|
|
11662 char_u *res;
|
|
|
11663 {
|
|
344
|
11664 salitem_T *smp;
|
|
323
|
11665 char_u word[MAXWLEN];
|
|
375
|
11666 char_u *s = inword;
|
|
323
|
11667 char_u *t;
|
|
344
|
11668 char_u *pf;
|
|
323
|
11669 int i, j, z;
|
|
344
|
11670 int reslen;
|
|
323
|
11671 int n, k = 0;
|
|
|
11672 int z0;
|
|
|
11673 int k0;
|
|
|
11674 int n0;
|
|
|
11675 int c;
|
|
|
11676 int pri;
|
|
|
11677 int p0 = -333;
|
|
|
11678 int c0;
|
|
|
11679
|
|
324
|
11680 /* Remove accents, if wanted. We actually remove all non-word characters.
|
|
375
|
11681 * But keep white space. We need a copy, the word may be changed here. */
|
|
323
|
11682 if (slang->sl_rem_accents)
|
|
|
11683 {
|
|
|
11684 t = word;
|
|
375
|
11685 while (*s != NUL)
|
|
323
|
11686 {
|
|
324
|
11687 if (vim_iswhite(*s))
|
|
344
|
11688 {
|
|
|
11689 *t++ = ' ';
|
|
|
11690 s = skipwhite(s);
|
|
|
11691 }
|
|
323
|
11692 else
|
|
|
11693 {
|
|
376
|
11694 if (spell_iswordp_nmw(s))
|
|
323
|
11695 *t++ = *s;
|
|
|
11696 ++s;
|
|
|
11697 }
|
|
|
11698 }
|
|
|
11699 *t = NUL;
|
|
|
11700 }
|
|
|
11701 else
|
|
375
|
11702 STRCPY(word, s);
|
|
323
|
11703
|
|
344
|
11704 smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
323
|
11705
|
|
|
11706 /*
|
|
|
11707 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
324
|
11708 * Changed to keep spaces.
|
|
323
|
11709 */
|
|
344
|
11710 i = reslen = z = 0;
|
|
323
|
11711 while ((c = word[i]) != NUL)
|
|
|
11712 {
|
|
344
|
11713 /* Start with the first rule that has the character in the word. */
|
|
323
|
11714 n = slang->sl_sal_first[c];
|
|
|
11715 z0 = 0;
|
|
|
11716
|
|
|
11717 if (n >= 0)
|
|
|
11718 {
|
|
|
11719 /* check all rules for the same letter */
|
|
344
|
11720 for (; (s = smp[n].sm_lead)[0] == c; ++n)
|
|
323
|
11721 {
|
|
344
|
11722 /* Quickly skip entries that don't match the word. Most
|
|
|
11723 * entries are less then three chars, optimize for that. */
|
|
|
11724 k = smp[n].sm_leadlen;
|
|
|
11725 if (k > 1)
|
|
323
|
11726 {
|
|
344
|
11727 if (word[i + 1] != s[1])
|
|
|
11728 continue;
|
|
|
11729 if (k > 2)
|
|
|
11730 {
|
|
|
11731 for (j = 2; j < k; ++j)
|
|
|
11732 if (word[i + j] != s[j])
|
|
|
11733 break;
|
|
|
11734 if (j < k)
|
|
|
11735 continue;
|
|
|
11736 }
|
|
323
|
11737 }
|
|
|
11738
|
|
375
|
11739 if ((pf = smp[n].sm_oneof) != NULL)
|
|
323
|
11740 {
|
|
375
|
11741 /* Check for match with one of the chars in "sm_oneof". */
|
|
344
|
11742 while (*pf != NUL && *pf != word[i + k])
|
|
|
11743 ++pf;
|
|
|
11744 if (*pf == NUL)
|
|
|
11745 continue;
|
|
|
11746 ++k;
|
|
323
|
11747 }
|
|
344
|
11748 s = smp[n].sm_rules;
|
|
|
11749 pri = 5; /* default priority */
|
|
323
|
11750
|
|
|
11751 p0 = *s;
|
|
|
11752 k0 = k;
|
|
|
11753 while (*s == '-' && k > 1)
|
|
|
11754 {
|
|
|
11755 k--;
|
|
|
11756 s++;
|
|
|
11757 }
|
|
|
11758 if (*s == '<')
|
|
|
11759 s++;
|
|
344
|
11760 if (VIM_ISDIGIT(*s))
|
|
323
|
11761 {
|
|
|
11762 /* determine priority */
|
|
|
11763 pri = *s - '0';
|
|
|
11764 s++;
|
|
|
11765 }
|
|
|
11766 if (*s == '^' && *(s + 1) == '^')
|
|
|
11767 s++;
|
|
|
11768
|
|
|
11769 if (*s == NUL
|
|
|
11770 || (*s == '^'
|
|
324
|
11771 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
11772 || spell_iswordp(word + i - 1, curbuf)))
|
|
323
|
11773 && (*(s + 1) != '$'
|
|
376
|
11774 || (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
11775 || (*s == '$' && i > 0
|
|
376
|
11776 && spell_iswordp(word + i - 1, curbuf)
|
|
|
11777 && (!spell_iswordp(word + i + k0, curbuf))))
|
|
323
|
11778 {
|
|
|
11779 /* search for followup rules, if: */
|
|
|
11780 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
11781 c0 = word[i + k - 1];
|
|
|
11782 n0 = slang->sl_sal_first[c0];
|
|
|
11783
|
|
|
11784 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
344
|
11785 && p0 != '-' && word[i + k] != NUL)
|
|
323
|
11786 {
|
|
|
11787 /* test follow-up rule for "word[i + k]" */
|
|
344
|
11788 for ( ; (s = smp[n0].sm_lead)[0] == c0; ++n0)
|
|
323
|
11789 {
|
|
344
|
11790 /* Quickly skip entries that don't match the word.
|
|
|
11791 * */
|
|
|
11792 k0 = smp[n0].sm_leadlen;
|
|
|
11793 if (k0 > 1)
|
|
323
|
11794 {
|
|
344
|
11795 if (word[i + k] != s[1])
|
|
|
11796 continue;
|
|
|
11797 if (k0 > 2)
|
|
|
11798 {
|
|
|
11799 pf = word + i + k + 1;
|
|
|
11800 for (j = 2; j < k0; ++j)
|
|
|
11801 if (*pf++ != s[j])
|
|
|
11802 break;
|
|
|
11803 if (j < k0)
|
|
|
11804 continue;
|
|
|
11805 }
|
|
323
|
11806 }
|
|
344
|
11807 k0 += k - 1;
|
|
|
11808
|
|
375
|
11809 if ((pf = smp[n0].sm_oneof) != NULL)
|
|
323
|
11810 {
|
|
344
|
11811 /* Check for match with one of the chars in
|
|
375
|
11812 * "sm_oneof". */
|
|
344
|
11813 while (*pf != NUL && *pf != word[i + k0])
|
|
|
11814 ++pf;
|
|
|
11815 if (*pf == NUL)
|
|
|
11816 continue;
|
|
|
11817 ++k0;
|
|
323
|
11818 }
|
|
344
|
11819
|
|
|
11820 p0 = 5;
|
|
|
11821 s = smp[n0].sm_rules;
|
|
323
|
11822 while (*s == '-')
|
|
|
11823 {
|
|
344
|
11824 /* "k0" gets NOT reduced because
|
|
|
11825 * "if (k0 == k)" */
|
|
323
|
11826 s++;
|
|
|
11827 }
|
|
|
11828 if (*s == '<')
|
|
|
11829 s++;
|
|
344
|
11830 if (VIM_ISDIGIT(*s))
|
|
323
|
11831 {
|
|
|
11832 p0 = *s - '0';
|
|
|
11833 s++;
|
|
|
11834 }
|
|
|
11835
|
|
|
11836 if (*s == NUL
|
|
|
11837 /* *s == '^' cuts */
|
|
|
11838 || (*s == '$'
|
|
376
|
11839 && !spell_iswordp(word + i + k0,
|
|
|
11840 curbuf)))
|
|
323
|
11841 {
|
|
|
11842 if (k0 == k)
|
|
|
11843 /* this is just a piece of the string */
|
|
|
11844 continue;
|
|
|
11845
|
|
|
11846 if (p0 < pri)
|
|
|
11847 /* priority too low */
|
|
|
11848 continue;
|
|
|
11849 /* rule fits; stop search */
|
|
|
11850 break;
|
|
|
11851 }
|
|
|
11852 }
|
|
|
11853
|
|
344
|
11854 if (p0 >= pri && smp[n0].sm_lead[0] == c0)
|
|
323
|
11855 continue;
|
|
|
11856 }
|
|
|
11857
|
|
|
11858 /* replace string */
|
|
344
|
11859 s = smp[n].sm_to;
|
|
389
|
11860 if (s == NULL)
|
|
|
11861 s = (char_u *)"";
|
|
344
|
11862 pf = smp[n].sm_rules;
|
|
|
11863 p0 = (vim_strchr(pf, '<') != NULL) ? 1 : 0;
|
|
323
|
11864 if (p0 == 1 && z == 0)
|
|
|
11865 {
|
|
|
11866 /* rule with '<' is used */
|
|
344
|
11867 if (reslen > 0 && *s != NUL && (res[reslen - 1] == c
|
|
|
11868 || res[reslen - 1] == *s))
|
|
|
11869 reslen--;
|
|
323
|
11870 z0 = 1;
|
|
|
11871 z = 1;
|
|
|
11872 k0 = 0;
|
|
372
|
11873 while (*s != NUL && word[i + k0] != NUL)
|
|
323
|
11874 {
|
|
|
11875 word[i + k0] = *s;
|
|
|
11876 k0++;
|
|
|
11877 s++;
|
|
|
11878 }
|
|
|
11879 if (k > k0)
|
|
|
11880 mch_memmove(word + i + k0, word + i + k,
|
|
|
11881 STRLEN(word + i + k) + 1);
|
|
|
11882
|
|
|
11883 /* new "actual letter" */
|
|
|
11884 c = word[i];
|
|
|
11885 }
|
|
|
11886 else
|
|
|
11887 {
|
|
|
11888 /* no '<' rule used */
|
|
|
11889 i += k - 1;
|
|
|
11890 z = 0;
|
|
344
|
11891 while (*s != NUL && s[1] != NUL && reslen < MAXWLEN)
|
|
323
|
11892 {
|
|
344
|
11893 if (reslen == 0 || res[reslen - 1] != *s)
|
|
372
|
11894 res[reslen++] = *s;
|
|
323
|
11895 s++;
|
|
|
11896 }
|
|
|
11897 /* new "actual letter" */
|
|
|
11898 c = *s;
|
|
344
|
11899 if (strstr((char *)pf, "^^") != NULL)
|
|
323
|
11900 {
|
|
|
11901 if (c != NUL)
|
|
372
|
11902 res[reslen++] = c;
|
|
323
|
11903 mch_memmove(word, word + i + 1,
|
|
|
11904 STRLEN(word + i + 1) + 1);
|
|
|
11905 i = 0;
|
|
|
11906 z0 = 1;
|
|
|
11907 }
|
|
|
11908 }
|
|
|
11909 break;
|
|
|
11910 }
|
|
|
11911 }
|
|
|
11912 }
|
|
324
|
11913 else if (vim_iswhite(c))
|
|
|
11914 {
|
|
|
11915 c = ' ';
|
|
|
11916 k = 1;
|
|
|
11917 }
|
|
323
|
11918
|
|
|
11919 if (z0 == 0)
|
|
|
11920 {
|
|
344
|
11921 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
11922 && (!slang->sl_collapse || reslen == 0
|
|
|
11923 || res[reslen - 1] != c))
|
|
323
|
11924 /* condense only double letters */
|
|
372
|
11925 res[reslen++] = c;
|
|
323
|
11926
|
|
|
11927 i++;
|
|
|
11928 z = 0;
|
|
|
11929 k = 0;
|
|
|
11930 }
|
|
|
11931 }
|
|
|
11932
|
|
344
|
11933 res[reslen] = NUL;
|
|
323
|
11934 }
|
|
|
11935
|
|
372
|
11936 #ifdef FEAT_MBYTE
|
|
|
11937 /*
|
|
|
11938 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
|
11939 * Multi-byte version of spell_soundfold().
|
|
|
11940 */
|
|
|
11941 static void
|
|
375
|
11942 spell_soundfold_wsal(slang, inword, res)
|
|
372
|
11943 slang_T *slang;
|
|
|
11944 char_u *inword;
|
|
|
11945 char_u *res;
|
|
|
11946 {
|
|
375
|
11947 salitem_T *smp = (salitem_T *)slang->sl_sal.ga_data;
|
|
372
|
11948 int word[MAXWLEN];
|
|
|
11949 int wres[MAXWLEN];
|
|
|
11950 int l;
|
|
|
11951 char_u *s;
|
|
|
11952 int *ws;
|
|
|
11953 char_u *t;
|
|
|
11954 int *pf;
|
|
|
11955 int i, j, z;
|
|
|
11956 int reslen;
|
|
|
11957 int n, k = 0;
|
|
|
11958 int z0;
|
|
|
11959 int k0;
|
|
|
11960 int n0;
|
|
|
11961 int c;
|
|
|
11962 int pri;
|
|
|
11963 int p0 = -333;
|
|
|
11964 int c0;
|
|
|
11965 int did_white = FALSE;
|
|
|
11966
|
|
|
11967 /*
|
|
|
11968 * Convert the multi-byte string to a wide-character string.
|
|
|
11969 * Remove accents, if wanted. We actually remove all non-word characters.
|
|
|
11970 * But keep white space.
|
|
|
11971 */
|
|
|
11972 n = 0;
|
|
|
11973 for (s = inword; *s != NUL; )
|
|
|
11974 {
|
|
|
11975 t = s;
|
|
474
|
11976 c = mb_cptr2char_adv(&s);
|
|
372
|
11977 if (slang->sl_rem_accents)
|
|
|
11978 {
|
|
|
11979 if (enc_utf8 ? utf_class(c) == 0 : vim_iswhite(c))
|
|
|
11980 {
|
|
|
11981 if (did_white)
|
|
|
11982 continue;
|
|
|
11983 c = ' ';
|
|
|
11984 did_white = TRUE;
|
|
|
11985 }
|
|
|
11986 else
|
|
|
11987 {
|
|
|
11988 did_white = FALSE;
|
|
376
|
11989 if (!spell_iswordp_nmw(t))
|
|
372
|
11990 continue;
|
|
|
11991 }
|
|
|
11992 }
|
|
|
11993 word[n++] = c;
|
|
|
11994 }
|
|
|
11995 word[n] = NUL;
|
|
|
11996
|
|
|
11997 /*
|
|
|
11998 * This comes from Aspell phonet.cpp.
|
|
|
11999 * Converted from C++ to C. Added support for multi-byte chars.
|
|
|
12000 * Changed to keep spaces.
|
|
|
12001 */
|
|
|
12002 i = reslen = z = 0;
|
|
|
12003 while ((c = word[i]) != NUL)
|
|
|
12004 {
|
|
|
12005 /* Start with the first rule that has the character in the word. */
|
|
|
12006 n = slang->sl_sal_first[c & 0xff];
|
|
|
12007 z0 = 0;
|
|
|
12008
|
|
|
12009 if (n >= 0)
|
|
|
12010 {
|
|
375
|
12011 /* check all rules for the same index byte */
|
|
372
|
12012 for (; ((ws = smp[n].sm_lead_w)[0] & 0xff) == (c & 0xff); ++n)
|
|
|
12013 {
|
|
|
12014 /* Quickly skip entries that don't match the word. Most
|
|
|
12015 * entries are less then three chars, optimize for that. */
|
|
375
|
12016 if (c != ws[0])
|
|
|
12017 continue;
|
|
372
|
12018 k = smp[n].sm_leadlen;
|
|
|
12019 if (k > 1)
|
|
|
12020 {
|
|
|
12021 if (word[i + 1] != ws[1])
|
|
|
12022 continue;
|
|
|
12023 if (k > 2)
|
|
|
12024 {
|
|
|
12025 for (j = 2; j < k; ++j)
|
|
|
12026 if (word[i + j] != ws[j])
|
|
|
12027 break;
|
|
|
12028 if (j < k)
|
|
|
12029 continue;
|
|
|
12030 }
|
|
|
12031 }
|
|
|
12032
|
|
375
|
12033 if ((pf = smp[n].sm_oneof_w) != NULL)
|
|
372
|
12034 {
|
|
375
|
12035 /* Check for match with one of the chars in "sm_oneof". */
|
|
372
|
12036 while (*pf != NUL && *pf != word[i + k])
|
|
|
12037 ++pf;
|
|
|
12038 if (*pf == NUL)
|
|
|
12039 continue;
|
|
|
12040 ++k;
|
|
|
12041 }
|
|
|
12042 s = smp[n].sm_rules;
|
|
|
12043 pri = 5; /* default priority */
|
|
|
12044
|
|
|
12045 p0 = *s;
|
|
|
12046 k0 = k;
|
|
|
12047 while (*s == '-' && k > 1)
|
|
|
12048 {
|
|
|
12049 k--;
|
|
|
12050 s++;
|
|
|
12051 }
|
|
|
12052 if (*s == '<')
|
|
|
12053 s++;
|
|
|
12054 if (VIM_ISDIGIT(*s))
|
|
|
12055 {
|
|
|
12056 /* determine priority */
|
|
|
12057 pri = *s - '0';
|
|
|
12058 s++;
|
|
|
12059 }
|
|
|
12060 if (*s == '^' && *(s + 1) == '^')
|
|
|
12061 s++;
|
|
|
12062
|
|
|
12063 if (*s == NUL
|
|
|
12064 || (*s == '^'
|
|
|
12065 && (i == 0 || !(word[i - 1] == ' '
|
|
376
|
12066 || spell_iswordp_w(word + i - 1, curbuf)))
|
|
372
|
12067 && (*(s + 1) != '$'
|
|
376
|
12068 || (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
12069 || (*s == '$' && i > 0
|
|
376
|
12070 && spell_iswordp_w(word + i - 1, curbuf)
|
|
|
12071 && (!spell_iswordp_w(word + i + k0, curbuf))))
|
|
372
|
12072 {
|
|
|
12073 /* search for followup rules, if: */
|
|
|
12074 /* followup and k > 1 and NO '-' in searchstring */
|
|
|
12075 c0 = word[i + k - 1];
|
|
|
12076 n0 = slang->sl_sal_first[c0 & 0xff];
|
|
|
12077
|
|
|
12078 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
|
12079 && p0 != '-' && word[i + k] != NUL)
|
|
|
12080 {
|
|
375
|
12081 /* Test follow-up rule for "word[i + k]"; loop over
|
|
|
12082 * all entries with the same index byte. */
|
|
372
|
12083 for ( ; ((ws = smp[n0].sm_lead_w)[0] & 0xff)
|
|
|
12084 == (c0 & 0xff); ++n0)
|
|
|
12085 {
|
|
|
12086 /* Quickly skip entries that don't match the word.
|
|
375
|
12087 */
|
|
|
12088 if (c0 != ws[0])
|
|
|
12089 continue;
|
|
372
|
12090 k0 = smp[n0].sm_leadlen;
|
|
|
12091 if (k0 > 1)
|
|
|
12092 {
|
|
|
12093 if (word[i + k] != ws[1])
|
|
|
12094 continue;
|
|
|
12095 if (k0 > 2)
|
|
|
12096 {
|
|
|
12097 pf = word + i + k + 1;
|
|
|
12098 for (j = 2; j < k0; ++j)
|
|
|
12099 if (*pf++ != ws[j])
|
|
|
12100 break;
|
|
|
12101 if (j < k0)
|
|
|
12102 continue;
|
|
|
12103 }
|
|
|
12104 }
|
|
|
12105 k0 += k - 1;
|
|
|
12106
|
|
375
|
12107 if ((pf = smp[n0].sm_oneof_w) != NULL)
|
|
372
|
12108 {
|
|
|
12109 /* Check for match with one of the chars in
|
|
375
|
12110 * "sm_oneof". */
|
|
372
|
12111 while (*pf != NUL && *pf != word[i + k0])
|
|
|
12112 ++pf;
|
|
|
12113 if (*pf == NUL)
|
|
|
12114 continue;
|
|
|
12115 ++k0;
|
|
|
12116 }
|
|
|
12117
|
|
|
12118 p0 = 5;
|
|
|
12119 s = smp[n0].sm_rules;
|
|
|
12120 while (*s == '-')
|
|
|
12121 {
|
|
|
12122 /* "k0" gets NOT reduced because
|
|
|
12123 * "if (k0 == k)" */
|
|
|
12124 s++;
|
|
|
12125 }
|
|
|
12126 if (*s == '<')
|
|
|
12127 s++;
|
|
|
12128 if (VIM_ISDIGIT(*s))
|
|
|
12129 {
|
|
|
12130 p0 = *s - '0';
|
|
|
12131 s++;
|
|
|
12132 }
|
|
|
12133
|
|
|
12134 if (*s == NUL
|
|
|
12135 /* *s == '^' cuts */
|
|
|
12136 || (*s == '$'
|
|
376
|
12137 && !spell_iswordp_w(word + i + k0,
|
|
|
12138 curbuf)))
|
|
372
|
12139 {
|
|
|
12140 if (k0 == k)
|
|
|
12141 /* this is just a piece of the string */
|
|
|
12142 continue;
|
|
|
12143
|
|
|
12144 if (p0 < pri)
|
|
|
12145 /* priority too low */
|
|
|
12146 continue;
|
|
|
12147 /* rule fits; stop search */
|
|
|
12148 break;
|
|
|
12149 }
|
|
|
12150 }
|
|
|
12151
|
|
|
12152 if (p0 >= pri && (smp[n0].sm_lead_w[0] & 0xff)
|
|
|
12153 == (c0 & 0xff))
|
|
|
12154 continue;
|
|
|
12155 }
|
|
|
12156
|
|
|
12157 /* replace string */
|
|
|
12158 ws = smp[n].sm_to_w;
|
|
|
12159 s = smp[n].sm_rules;
|
|
|
12160 p0 = (vim_strchr(s, '<') != NULL) ? 1 : 0;
|
|
|
12161 if (p0 == 1 && z == 0)
|
|
|
12162 {
|
|
|
12163 /* rule with '<' is used */
|
|
389
|
12164 if (reslen > 0 && ws != NULL && *ws != NUL
|
|
|
12165 && (wres[reslen - 1] == c
|
|
372
|
12166 || wres[reslen - 1] == *ws))
|
|
|
12167 reslen--;
|
|
|
12168 z0 = 1;
|
|
|
12169 z = 1;
|
|
|
12170 k0 = 0;
|
|
389
|
12171 if (ws != NULL)
|
|
|
12172 while (*ws != NUL && word[i + k0] != NUL)
|
|
|
12173 {
|
|
|
12174 word[i + k0] = *ws;
|
|
|
12175 k0++;
|
|
|
12176 ws++;
|
|
|
12177 }
|
|
372
|
12178 if (k > k0)
|
|
|
12179 mch_memmove(word + i + k0, word + i + k,
|
|
|
12180 sizeof(int) * (STRLEN(word + i + k) + 1));
|
|
|
12181
|
|
|
12182 /* new "actual letter" */
|
|
|
12183 c = word[i];
|
|
|
12184 }
|
|
|
12185 else
|
|
|
12186 {
|
|
|
12187 /* no '<' rule used */
|
|
|
12188 i += k - 1;
|
|
|
12189 z = 0;
|
|
389
|
12190 if (ws != NULL)
|
|
|
12191 while (*ws != NUL && ws[1] != NUL
|
|
|
12192 && reslen < MAXWLEN)
|
|
|
12193 {
|
|
|
12194 if (reslen == 0 || wres[reslen - 1] != *ws)
|
|
|
12195 wres[reslen++] = *ws;
|
|
|
12196 ws++;
|
|
|
12197 }
|
|
372
|
12198 /* new "actual letter" */
|
|
389
|
12199 if (ws == NULL)
|
|
|
12200 c = NUL;
|
|
|
12201 else
|
|
|
12202 c = *ws;
|
|
372
|
12203 if (strstr((char *)s, "^^") != NULL)
|
|
|
12204 {
|
|
|
12205 if (c != NUL)
|
|
|
12206 wres[reslen++] = c;
|
|
|
12207 mch_memmove(word, word + i + 1,
|
|
|
12208 sizeof(int) * (STRLEN(word + i + 1) + 1));
|
|
|
12209 i = 0;
|
|
|
12210 z0 = 1;
|
|
|
12211 }
|
|
|
12212 }
|
|
|
12213 break;
|
|
|
12214 }
|
|
|
12215 }
|
|
|
12216 }
|
|
|
12217 else if (vim_iswhite(c))
|
|
|
12218 {
|
|
|
12219 c = ' ';
|
|
|
12220 k = 1;
|
|
|
12221 }
|
|
|
12222
|
|
|
12223 if (z0 == 0)
|
|
|
12224 {
|
|
|
12225 if (k && !p0 && reslen < MAXWLEN && c != NUL
|
|
|
12226 && (!slang->sl_collapse || reslen == 0
|
|
|
12227 || wres[reslen - 1] != c))
|
|
|
12228 /* condense only double letters */
|
|
|
12229 wres[reslen++] = c;
|
|
|
12230
|
|
|
12231 i++;
|
|
|
12232 z = 0;
|
|
|
12233 k = 0;
|
|
|
12234 }
|
|
|
12235 }
|
|
|
12236
|
|
|
12237 /* Convert wide characters in "wres" to a multi-byte string in "res". */
|
|
|
12238 l = 0;
|
|
|
12239 for (n = 0; n < reslen; ++n)
|
|
|
12240 {
|
|
|
12241 l += mb_char2bytes(wres[n], res + l);
|
|
|
12242 if (l + MB_MAXBYTES > MAXWLEN)
|
|
|
12243 break;
|
|
|
12244 }
|
|
|
12245 res[l] = NUL;
|
|
|
12246 }
|
|
|
12247 #endif
|
|
|
12248
|
|
324
|
12249 /*
|
|
344
|
12250 * Compute a score for two sound-a-like words.
|
|
|
12251 * This permits up to two inserts/deletes/swaps/etc. to keep things fast.
|
|
|
12252 * Instead of a generic loop we write out the code. That keeps it fast by
|
|
|
12253 * avoiding checks that will not be possible.
|
|
|
12254 */
|
|
|
12255 static int
|
|
351
|
12256 soundalike_score(goodstart, badstart)
|
|
|
12257 char_u *goodstart; /* sound-folded good word */
|
|
|
12258 char_u *badstart; /* sound-folded bad word */
|
|
344
|
12259 {
|
|
351
|
12260 char_u *goodsound = goodstart;
|
|
|
12261 char_u *badsound = badstart;
|
|
|
12262 int goodlen;
|
|
|
12263 int badlen;
|
|
344
|
12264 int n;
|
|
|
12265 char_u *pl, *ps;
|
|
|
12266 char_u *pl2, *ps2;
|
|
351
|
12267 int score = 0;
|
|
|
12268
|
|
|
12269 /* adding/inserting "*" at the start (word starts with vowel) shouldn't be
|
|
|
12270 * counted so much, vowels halfway the word aren't counted at all. */
|
|
|
12271 if ((*badsound == '*' || *goodsound == '*') && *badsound != *goodsound)
|
|
|
12272 {
|
|
|
12273 score = SCORE_DEL / 2;
|
|
|
12274 if (*badsound == '*')
|
|
|
12275 ++badsound;
|
|
|
12276 else
|
|
|
12277 ++goodsound;
|
|
|
12278 }
|
|
|
12279
|
|
|
12280 goodlen = STRLEN(goodsound);
|
|
|
12281 badlen = STRLEN(badsound);
|
|
344
|
12282
|
|
|
12283 /* Return quickly if the lenghts are too different to be fixed by two
|
|
|
12284 * changes. */
|
|
|
12285 n = goodlen - badlen;
|
|
|
12286 if (n < -2 || n > 2)
|
|
|
12287 return SCORE_MAXMAX;
|
|
|
12288
|
|
|
12289 if (n > 0)
|
|
|
12290 {
|
|
351
|
12291 pl = goodsound; /* goodsound is longest */
|
|
344
|
12292 ps = badsound;
|
|
|
12293 }
|
|
|
12294 else
|
|
|
12295 {
|
|
351
|
12296 pl = badsound; /* badsound is longest */
|
|
344
|
12297 ps = goodsound;
|
|
|
12298 }
|
|
|
12299
|
|
|
12300 /* Skip over the identical part. */
|
|
|
12301 while (*pl == *ps && *pl != NUL)
|
|
|
12302 {
|
|
|
12303 ++pl;
|
|
|
12304 ++ps;
|
|
|
12305 }
|
|
|
12306
|
|
|
12307 switch (n)
|
|
|
12308 {
|
|
|
12309 case -2:
|
|
|
12310 case 2:
|
|
|
12311 /*
|
|
|
12312 * Must delete two characters from "pl".
|
|
|
12313 */
|
|
|
12314 ++pl; /* first delete */
|
|
|
12315 while (*pl == *ps)
|
|
|
12316 {
|
|
|
12317 ++pl;
|
|
|
12318 ++ps;
|
|
|
12319 }
|
|
|
12320 /* strings must be equal after second delete */
|
|
|
12321 if (STRCMP(pl + 1, ps) == 0)
|
|
351
|
12322 return score + SCORE_DEL * 2;
|
|
344
|
12323
|
|
|
12324 /* Failed to compare. */
|
|
|
12325 break;
|
|
|
12326
|
|
|
12327 case -1:
|
|
|
12328 case 1:
|
|
|
12329 /*
|
|
|
12330 * Minimal one delete from "pl" required.
|
|
|
12331 */
|
|
|
12332
|
|
|
12333 /* 1: delete */
|
|
|
12334 pl2 = pl + 1;
|
|
|
12335 ps2 = ps;
|
|
|
12336 while (*pl2 == *ps2)
|
|
|
12337 {
|
|
|
12338 if (*pl2 == NUL) /* reached the end */
|
|
351
|
12339 return score + SCORE_DEL;
|
|
344
|
12340 ++pl2;
|
|
|
12341 ++ps2;
|
|
|
12342 }
|
|
|
12343
|
|
|
12344 /* 2: delete then swap, then rest must be equal */
|
|
|
12345 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
12346 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
12347 return score + SCORE_DEL + SCORE_SWAP;
|
|
344
|
12348
|
|
|
12349 /* 3: delete then substitute, then the rest must be equal */
|
|
|
12350 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
12351 return score + SCORE_DEL + SCORE_SUBST;
|
|
344
|
12352
|
|
|
12353 /* 4: first swap then delete */
|
|
|
12354 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
12355 {
|
|
|
12356 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
12357 ps2 = ps + 2;
|
|
|
12358 while (*pl2 == *ps2)
|
|
|
12359 {
|
|
|
12360 ++pl2;
|
|
|
12361 ++ps2;
|
|
|
12362 }
|
|
|
12363 /* delete a char and then strings must be equal */
|
|
|
12364 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
12365 return score + SCORE_SWAP + SCORE_DEL;
|
|
344
|
12366 }
|
|
|
12367
|
|
|
12368 /* 5: first substitute then delete */
|
|
|
12369 pl2 = pl + 1; /* substitute, skip one char */
|
|
|
12370 ps2 = ps + 1;
|
|
|
12371 while (*pl2 == *ps2)
|
|
|
12372 {
|
|
|
12373 ++pl2;
|
|
|
12374 ++ps2;
|
|
|
12375 }
|
|
|
12376 /* delete a char and then strings must be equal */
|
|
|
12377 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
12378 return score + SCORE_SUBST + SCORE_DEL;
|
|
344
|
12379
|
|
|
12380 /* Failed to compare. */
|
|
|
12381 break;
|
|
|
12382
|
|
|
12383 case 0:
|
|
|
12384 /*
|
|
|
12385 * Lenghts are equal, thus changes must result in same length: An
|
|
|
12386 * insert is only possible in combination with a delete.
|
|
|
12387 * 1: check if for identical strings
|
|
|
12388 */
|
|
|
12389 if (*pl == NUL)
|
|
351
|
12390 return score;
|
|
344
|
12391
|
|
|
12392 /* 2: swap */
|
|
|
12393 if (pl[0] == ps[1] && pl[1] == ps[0])
|
|
|
12394 {
|
|
|
12395 pl2 = pl + 2; /* swap, skip two chars */
|
|
|
12396 ps2 = ps + 2;
|
|
|
12397 while (*pl2 == *ps2)
|
|
|
12398 {
|
|
|
12399 if (*pl2 == NUL) /* reached the end */
|
|
351
|
12400 return score + SCORE_SWAP;
|
|
344
|
12401 ++pl2;
|
|
|
12402 ++ps2;
|
|
|
12403 }
|
|
|
12404 /* 3: swap and swap again */
|
|
|
12405 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
12406 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
12407 return score + SCORE_SWAP + SCORE_SWAP;
|
|
344
|
12408
|
|
|
12409 /* 4: swap and substitute */
|
|
|
12410 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
12411 return score + SCORE_SWAP + SCORE_SUBST;
|
|
344
|
12412 }
|
|
|
12413
|
|
|
12414 /* 5: substitute */
|
|
|
12415 pl2 = pl + 1;
|
|
|
12416 ps2 = ps + 1;
|
|
|
12417 while (*pl2 == *ps2)
|
|
|
12418 {
|
|
|
12419 if (*pl2 == NUL) /* reached the end */
|
|
351
|
12420 return score + SCORE_SUBST;
|
|
344
|
12421 ++pl2;
|
|
|
12422 ++ps2;
|
|
|
12423 }
|
|
|
12424
|
|
|
12425 /* 6: substitute and swap */
|
|
|
12426 if (pl2[0] == ps2[1] && pl2[1] == ps2[0]
|
|
|
12427 && STRCMP(pl2 + 2, ps2 + 2) == 0)
|
|
351
|
12428 return score + SCORE_SUBST + SCORE_SWAP;
|
|
344
|
12429
|
|
|
12430 /* 7: substitute and substitute */
|
|
|
12431 if (STRCMP(pl2 + 1, ps2 + 1) == 0)
|
|
351
|
12432 return score + SCORE_SUBST + SCORE_SUBST;
|
|
344
|
12433
|
|
|
12434 /* 8: insert then delete */
|
|
|
12435 pl2 = pl;
|
|
|
12436 ps2 = ps + 1;
|
|
|
12437 while (*pl2 == *ps2)
|
|
|
12438 {
|
|
|
12439 ++pl2;
|
|
|
12440 ++ps2;
|
|
|
12441 }
|
|
|
12442 if (STRCMP(pl2 + 1, ps2) == 0)
|
|
351
|
12443 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
12444
|
|
|
12445 /* 9: delete then insert */
|
|
|
12446 pl2 = pl + 1;
|
|
|
12447 ps2 = ps;
|
|
|
12448 while (*pl2 == *ps2)
|
|
|
12449 {
|
|
|
12450 ++pl2;
|
|
|
12451 ++ps2;
|
|
|
12452 }
|
|
|
12453 if (STRCMP(pl2, ps2 + 1) == 0)
|
|
351
|
12454 return score + SCORE_INS + SCORE_DEL;
|
|
344
|
12455
|
|
|
12456 /* Failed to compare. */
|
|
|
12457 break;
|
|
|
12458 }
|
|
|
12459
|
|
|
12460 return SCORE_MAXMAX;
|
|
|
12461 }
|
|
324
|
12462
|
|
323
|
12463 /*
|
|
|
12464 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
344
|
12465 * deletes/inserts/substitutes/swaps are required the lower the score.
|
|
324
|
12466 *
|
|
485
|
12467 * The algorithm is described by Du and Chang, 1992.
|
|
|
12468 * The implementation of the algorithm comes from Aspell editdist.cpp,
|
|
|
12469 * edit_distance(). It has been converted from C++ to C and modified to
|
|
|
12470 * support multi-byte characters.
|
|
323
|
12471 */
|
|
|
12472 static int
|
|
|
12473 spell_edit_score(badword, goodword)
|
|
|
12474 char_u *badword;
|
|
|
12475 char_u *goodword;
|
|
|
12476 {
|
|
|
12477 int *cnt;
|
|
372
|
12478 int badlen, goodlen; /* lenghts including NUL */
|
|
323
|
12479 int j, i;
|
|
|
12480 int t;
|
|
|
12481 int bc, gc;
|
|
324
|
12482 int pbc, pgc;
|
|
|
12483 #ifdef FEAT_MBYTE
|
|
|
12484 char_u *p;
|
|
|
12485 int wbadword[MAXWLEN];
|
|
|
12486 int wgoodword[MAXWLEN];
|
|
|
12487
|
|
|
12488 if (has_mbyte)
|
|
|
12489 {
|
|
|
12490 /* Get the characters from the multi-byte strings and put them in an
|
|
|
12491 * int array for easy access. */
|
|
|
12492 for (p = badword, badlen = 0; *p != NUL; )
|
|
474
|
12493 wbadword[badlen++] = mb_cptr2char_adv(&p);
|
|
398
|
12494 wbadword[badlen++] = 0;
|
|
324
|
12495 for (p = goodword, goodlen = 0; *p != NUL; )
|
|
474
|
12496 wgoodword[goodlen++] = mb_cptr2char_adv(&p);
|
|
398
|
12497 wgoodword[goodlen++] = 0;
|
|
324
|
12498 }
|
|
|
12499 else
|
|
|
12500 #endif
|
|
|
12501 {
|
|
|
12502 badlen = STRLEN(badword) + 1;
|
|
|
12503 goodlen = STRLEN(goodword) + 1;
|
|
|
12504 }
|
|
323
|
12505
|
|
|
12506 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
|
12507 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
|
12508 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
|
12509 TRUE);
|
|
324
|
12510 if (cnt == NULL)
|
|
|
12511 return 0; /* out of memory */
|
|
323
|
12512
|
|
|
12513 CNT(0, 0) = 0;
|
|
|
12514 for (j = 1; j <= goodlen; ++j)
|
|
|
12515 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
|
12516
|
|
|
12517 for (i = 1; i <= badlen; ++i)
|
|
|
12518 {
|
|
|
12519 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
|
12520 for (j = 1; j <= goodlen; ++j)
|
|
|
12521 {
|
|
324
|
12522 #ifdef FEAT_MBYTE
|
|
|
12523 if (has_mbyte)
|
|
|
12524 {
|
|
|
12525 bc = wbadword[i - 1];
|
|
|
12526 gc = wgoodword[j - 1];
|
|
|
12527 }
|
|
|
12528 else
|
|
|
12529 #endif
|
|
|
12530 {
|
|
|
12531 bc = badword[i - 1];
|
|
|
12532 gc = goodword[j - 1];
|
|
|
12533 }
|
|
323
|
12534 if (bc == gc)
|
|
|
12535 CNT(i, j) = CNT(i - 1, j - 1);
|
|
|
12536 else
|
|
|
12537 {
|
|
|
12538 /* Use a better score when there is only a case difference. */
|
|
324
|
12539 if (SPELL_TOFOLD(bc) == SPELL_TOFOLD(gc))
|
|
323
|
12540 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
|
12541 else
|
|
|
12542 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
|
12543
|
|
324
|
12544 if (i > 1 && j > 1)
|
|
323
|
12545 {
|
|
324
|
12546 #ifdef FEAT_MBYTE
|
|
|
12547 if (has_mbyte)
|
|
|
12548 {
|
|
|
12549 pbc = wbadword[i - 2];
|
|
|
12550 pgc = wgoodword[j - 2];
|
|
|
12551 }
|
|
|
12552 else
|
|
|
12553 #endif
|
|
|
12554 {
|
|
|
12555 pbc = badword[i - 2];
|
|
|
12556 pgc = goodword[j - 2];
|
|
|
12557 }
|
|
|
12558 if (bc == pgc && pbc == gc)
|
|
|
12559 {
|
|
|
12560 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
|
12561 if (t < CNT(i, j))
|
|
|
12562 CNT(i, j) = t;
|
|
|
12563 }
|
|
323
|
12564 }
|
|
|
12565 t = SCORE_DEL + CNT(i - 1, j);
|
|
|
12566 if (t < CNT(i, j))
|
|
|
12567 CNT(i, j) = t;
|
|
|
12568 t = SCORE_INS + CNT(i, j - 1);
|
|
|
12569 if (t < CNT(i, j))
|
|
|
12570 CNT(i, j) = t;
|
|
|
12571 }
|
|
|
12572 }
|
|
|
12573 }
|
|
344
|
12574
|
|
|
12575 i = CNT(badlen - 1, goodlen - 1);
|
|
|
12576 vim_free(cnt);
|
|
|
12577 return i;
|
|
323
|
12578 }
|
|
307
|
12579
|
|
351
|
12580 /*
|
|
|
12581 * ":spelldump"
|
|
|
12582 */
|
|
|
12583 /*ARGSUSED*/
|
|
|
12584 void
|
|
|
12585 ex_spelldump(eap)
|
|
|
12586 exarg_T *eap;
|
|
|
12587 {
|
|
|
12588 buf_T *buf = curbuf;
|
|
|
12589 langp_T *lp;
|
|
|
12590 slang_T *slang;
|
|
|
12591 idx_T arridx[MAXWLEN];
|
|
|
12592 int curi[MAXWLEN];
|
|
|
12593 char_u word[MAXWLEN];
|
|
|
12594 int c;
|
|
|
12595 char_u *byts;
|
|
|
12596 idx_T *idxs;
|
|
|
12597 linenr_T lnum = 0;
|
|
|
12598 int round;
|
|
|
12599 int depth;
|
|
|
12600 int n;
|
|
|
12601 int flags;
|
|
381
|
12602 char_u *region_names = NULL; /* region names being used */
|
|
|
12603 int do_region = TRUE; /* dump region names and numbers */
|
|
|
12604 char_u *p;
|
|
500
|
12605 int lpi;
|
|
351
|
12606
|
|
498
|
12607 if (no_spell_checking(curwin))
|
|
351
|
12608 return;
|
|
|
12609
|
|
|
12610 /* Create a new empty buffer by splitting the window. */
|
|
|
12611 do_cmdline_cmd((char_u *)"new");
|
|
|
12612 if (!bufempty() || !buf_valid(buf))
|
|
|
12613 return;
|
|
|
12614
|
|
381
|
12615 /* Find out if we can support regions: All languages must support the same
|
|
|
12616 * regions or none at all. */
|
|
500
|
12617 for (lpi = 0; lpi < buf->b_langp.ga_len; ++lpi)
|
|
|
12618 {
|
|
|
12619 lp = LANGP_ENTRY(buf->b_langp, lpi);
|
|
381
|
12620 p = lp->lp_slang->sl_regions;
|
|
|
12621 if (p[0] != 0)
|
|
|
12622 {
|
|
|
12623 if (region_names == NULL) /* first language with regions */
|
|
|
12624 region_names = p;
|
|
|
12625 else if (STRCMP(region_names, p) != 0)
|
|
|
12626 {
|
|
|
12627 do_region = FALSE; /* region names are different */
|
|
|
12628 break;
|
|
|
12629 }
|
|
|
12630 }
|
|
|
12631 }
|
|
|
12632
|
|
|
12633 if (do_region && region_names != NULL)
|
|
|
12634 {
|
|
|
12635 vim_snprintf((char *)IObuff, IOSIZE, "/regions=%s", region_names);
|
|
|
12636 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
12637 }
|
|
|
12638 else
|
|
|
12639 do_region = FALSE;
|
|
|
12640
|
|
|
12641 /*
|
|
|
12642 * Loop over all files loaded for the entries in 'spelllang'.
|
|
|
12643 */
|
|
500
|
12644 for (lpi = 0; lpi < buf->b_langp.ga_len; ++lpi)
|
|
|
12645 {
|
|
|
12646 lp = LANGP_ENTRY(buf->b_langp, lpi);
|
|
351
|
12647 slang = lp->lp_slang;
|
|
500
|
12648 if (slang->sl_fbyts == NULL) /* reloading failed */
|
|
|
12649 continue;
|
|
351
|
12650
|
|
|
12651 vim_snprintf((char *)IObuff, IOSIZE, "# file: %s", slang->sl_fname);
|
|
|
12652 ml_append(lnum++, IObuff, (colnr_T)0, FALSE);
|
|
|
12653
|
|
|
12654 /* round 1: case-folded tree
|
|
|
12655 * round 2: keep-case tree */
|
|
|
12656 for (round = 1; round <= 2; ++round)
|
|
|
12657 {
|
|
|
12658 if (round == 1)
|
|
|
12659 {
|
|
|
12660 byts = slang->sl_fbyts;
|
|
|
12661 idxs = slang->sl_fidxs;
|
|
|
12662 }
|
|
|
12663 else
|
|
|
12664 {
|
|
|
12665 byts = slang->sl_kbyts;
|
|
|
12666 idxs = slang->sl_kidxs;
|
|
|
12667 }
|
|
|
12668 if (byts == NULL)
|
|
|
12669 continue; /* array is empty */
|
|
|
12670
|
|
|
12671 depth = 0;
|
|
|
12672 arridx[0] = 0;
|
|
|
12673 curi[0] = 1;
|
|
|
12674 while (depth >= 0 && !got_int)
|
|
|
12675 {
|
|
|
12676 if (curi[depth] > byts[arridx[depth]])
|
|
|
12677 {
|
|
|
12678 /* Done all bytes at this node, go up one level. */
|
|
|
12679 --depth;
|
|
|
12680 line_breakcheck();
|
|
|
12681 }
|
|
|
12682 else
|
|
|
12683 {
|
|
|
12684 /* Do one more byte at this node. */
|
|
|
12685 n = arridx[depth] + curi[depth];
|
|
|
12686 ++curi[depth];
|
|
|
12687 c = byts[n];
|
|
|
12688 if (c == 0)
|
|
|
12689 {
|
|
|
12690 /* End of word, deal with the word.
|
|
|
12691 * Don't use keep-case words in the fold-case tree,
|
|
|
12692 * they will appear in the keep-case tree.
|
|
|
12693 * Only use the word when the region matches. */
|
|
|
12694 flags = (int)idxs[n];
|
|
|
12695 if ((round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
500
|
12696 && (flags & WF_NEEDCOMP) == 0
|
|
381
|
12697 && (do_region
|
|
|
12698 || (flags & WF_REGION) == 0
|
|
390
|
12699 || (((unsigned)flags >> 16)
|
|
351
|
12700 & lp->lp_region) != 0))
|
|
|
12701 {
|
|
|
12702 word[depth] = NUL;
|
|
381
|
12703 if (!do_region)
|
|
|
12704 flags &= ~WF_REGION;
|
|
355
|
12705
|
|
|
12706 /* Dump the basic word if there is no prefix or
|
|
|
12707 * when it's the first one. */
|
|
390
|
12708 c = (unsigned)flags >> 24;
|
|
355
|
12709 if (c == 0 || curi[depth] == 2)
|
|
|
12710 dump_word(word, round, flags, lnum++);
|
|
351
|
12711
|
|
|
12712 /* Apply the prefix, if there is one. */
|
|
355
|
12713 if (c != 0)
|
|
500
|
12714 lnum = dump_prefixes(slang, word, round,
|
|
351
|
12715 flags, lnum);
|
|
|
12716 }
|
|
|
12717 }
|
|
|
12718 else
|
|
|
12719 {
|
|
|
12720 /* Normal char, go one level deeper. */
|
|
|
12721 word[depth++] = c;
|
|
|
12722 arridx[depth] = idxs[n];
|
|
|
12723 curi[depth] = 1;
|
|
|
12724 }
|
|
|
12725 }
|
|
|
12726 }
|
|
|
12727 }
|
|
|
12728 }
|
|
|
12729
|
|
|
12730 /* Delete the empty line that we started with. */
|
|
|
12731 if (curbuf->b_ml.ml_line_count > 1)
|
|
|
12732 ml_delete(curbuf->b_ml.ml_line_count, FALSE);
|
|
|
12733
|
|
|
12734 redraw_later(NOT_VALID);
|
|
|
12735 }
|
|
|
12736
|
|
|
12737 /*
|
|
|
12738 * Dump one word: apply case modifications and append a line to the buffer.
|
|
|
12739 */
|
|
|
12740 static void
|
|
|
12741 dump_word(word, round, flags, lnum)
|
|
|
12742 char_u *word;
|
|
|
12743 int round;
|
|
|
12744 int flags;
|
|
|
12745 linenr_T lnum;
|
|
|
12746 {
|
|
|
12747 int keepcap = FALSE;
|
|
|
12748 char_u *p;
|
|
|
12749 char_u cword[MAXWLEN];
|
|
381
|
12750 char_u badword[MAXWLEN + 10];
|
|
|
12751 int i;
|
|
351
|
12752
|
|
|
12753 if (round == 1 && (flags & WF_CAPMASK) != 0)
|
|
|
12754 {
|
|
|
12755 /* Need to fix case according to "flags". */
|
|
|
12756 make_case_word(word, cword, flags);
|
|
|
12757 p = cword;
|
|
|
12758 }
|
|
|
12759 else
|
|
|
12760 {
|
|
|
12761 p = word;
|
|
389
|
12762 if (round == 2 && ((captype(word, NULL) & WF_KEEPCAP) == 0
|
|
|
12763 || (flags & WF_FIXCAP) != 0))
|
|
351
|
12764 keepcap = TRUE;
|
|
|
12765 }
|
|
|
12766
|
|
381
|
12767 /* Add flags and regions after a slash. */
|
|
|
12768 if ((flags & (WF_BANNED | WF_RARE | WF_REGION)) || keepcap)
|
|
|
12769 {
|
|
|
12770 STRCPY(badword, p);
|
|
|
12771 STRCAT(badword, "/");
|
|
351
|
12772 if (keepcap)
|
|
|
12773 STRCAT(badword, "=");
|
|
|
12774 if (flags & WF_BANNED)
|
|
|
12775 STRCAT(badword, "!");
|
|
|
12776 else if (flags & WF_RARE)
|
|
|
12777 STRCAT(badword, "?");
|
|
381
|
12778 if (flags & WF_REGION)
|
|
|
12779 for (i = 0; i < 7; ++i)
|
|
390
|
12780 if (flags & (0x10000 << i))
|
|
381
|
12781 sprintf((char *)badword + STRLEN(badword), "%d", i + 1);
|
|
351
|
12782 p = badword;
|
|
|
12783 }
|
|
|
12784
|
|
|
12785 ml_append(lnum, p, (colnr_T)0, FALSE);
|
|
|
12786 }
|
|
|
12787
|
|
|
12788 /*
|
|
372
|
12789 * For ":spelldump": Find matching prefixes for "word". Prepend each to
|
|
|
12790 * "word" and append a line to the buffer.
|
|
351
|
12791 * Return the updated line number.
|
|
|
12792 */
|
|
|
12793 static linenr_T
|
|
500
|
12794 dump_prefixes(slang, word, round, flags, startlnum)
|
|
351
|
12795 slang_T *slang;
|
|
|
12796 char_u *word; /* case-folded word */
|
|
|
12797 int round;
|
|
|
12798 int flags; /* flags with prefix ID */
|
|
|
12799 linenr_T startlnum;
|
|
|
12800 {
|
|
|
12801 idx_T arridx[MAXWLEN];
|
|
|
12802 int curi[MAXWLEN];
|
|
|
12803 char_u prefix[MAXWLEN];
|
|
455
|
12804 char_u word_up[MAXWLEN];
|
|
|
12805 int has_word_up = FALSE;
|
|
351
|
12806 int c;
|
|
|
12807 char_u *byts;
|
|
|
12808 idx_T *idxs;
|
|
|
12809 linenr_T lnum = startlnum;
|
|
|
12810 int depth;
|
|
|
12811 int n;
|
|
|
12812 int len;
|
|
|
12813 int i;
|
|
|
12814
|
|
455
|
12815 /* if the word starts with a lower-case letter make the word with an
|
|
|
12816 * upper-case letter in word_up[]. */
|
|
|
12817 c = PTR2CHAR(word);
|
|
|
12818 if (SPELL_TOUPPER(c) != c)
|
|
|
12819 {
|
|
|
12820 onecap_copy(word, word_up, TRUE);
|
|
|
12821 has_word_up = TRUE;
|
|
|
12822 }
|
|
|
12823
|
|
351
|
12824 byts = slang->sl_pbyts;
|
|
|
12825 idxs = slang->sl_pidxs;
|
|
|
12826 if (byts != NULL) /* array not is empty */
|
|
|
12827 {
|
|
|
12828 /*
|
|
|
12829 * Loop over all prefixes, building them byte-by-byte in prefix[].
|
|
390
|
12830 * When at the end of a prefix check that it supports "flags".
|
|
351
|
12831 */
|
|
|
12832 depth = 0;
|
|
|
12833 arridx[0] = 0;
|
|
|
12834 curi[0] = 1;
|
|
|
12835 while (depth >= 0 && !got_int)
|
|
|
12836 {
|
|
390
|
12837 n = arridx[depth];
|
|
|
12838 len = byts[n];
|
|
|
12839 if (curi[depth] > len)
|
|
351
|
12840 {
|
|
|
12841 /* Done all bytes at this node, go up one level. */
|
|
|
12842 --depth;
|
|
|
12843 line_breakcheck();
|
|
|
12844 }
|
|
|
12845 else
|
|
|
12846 {
|
|
|
12847 /* Do one more byte at this node. */
|
|
390
|
12848 n += curi[depth];
|
|
351
|
12849 ++curi[depth];
|
|
|
12850 c = byts[n];
|
|
|
12851 if (c == 0)
|
|
|
12852 {
|
|
|
12853 /* End of prefix, find out how many IDs there are. */
|
|
|
12854 for (i = 1; i < len; ++i)
|
|
|
12855 if (byts[n + i] != 0)
|
|
|
12856 break;
|
|
|
12857 curi[depth] += i - 1;
|
|
|
12858
|
|
455
|
12859 c = valid_word_prefix(i, n, flags, word, slang, FALSE);
|
|
|
12860 if (c != 0)
|
|
351
|
12861 {
|
|
376
|
12862 vim_strncpy(prefix + depth, word, MAXWLEN - depth - 1);
|
|
366
|
12863 dump_word(prefix, round,
|
|
455
|
12864 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
366
|
12865 : flags, lnum++);
|
|
351
|
12866 }
|
|
455
|
12867
|
|
|
12868 /* Check for prefix that matches the word when the
|
|
|
12869 * first letter is upper-case, but only if the prefix has
|
|
|
12870 * a condition. */
|
|
|
12871 if (has_word_up)
|
|
|
12872 {
|
|
|
12873 c = valid_word_prefix(i, n, flags, word_up, slang,
|
|
|
12874 TRUE);
|
|
|
12875 if (c != 0)
|
|
|
12876 {
|
|
|
12877 vim_strncpy(prefix + depth, word_up,
|
|
|
12878 MAXWLEN - depth - 1);
|
|
|
12879 dump_word(prefix, round,
|
|
|
12880 (c & WF_RAREPFX) ? (flags | WF_RARE)
|
|
|
12881 : flags, lnum++);
|
|
|
12882 }
|
|
|
12883 }
|
|
351
|
12884 }
|
|
|
12885 else
|
|
|
12886 {
|
|
|
12887 /* Normal char, go one level deeper. */
|
|
|
12888 prefix[depth++] = c;
|
|
|
12889 arridx[depth] = idxs[n];
|
|
|
12890 curi[depth] = 1;
|
|
|
12891 }
|
|
|
12892 }
|
|
|
12893 }
|
|
|
12894 }
|
|
|
12895
|
|
|
12896 return lnum;
|
|
|
12897 }
|
|
|
12898
|
|
498
|
12899 /*
|
|
|
12900 * Move "p" to end of word.
|
|
|
12901 */
|
|
|
12902 char_u *
|
|
|
12903 spell_to_word_end(start, buf)
|
|
|
12904 char_u *start;
|
|
|
12905 buf_T *buf;
|
|
|
12906 {
|
|
|
12907 char_u *p = start;
|
|
|
12908
|
|
|
12909 while (*p != NUL && spell_iswordp(p, buf))
|
|
|
12910 mb_ptr_adv(p);
|
|
|
12911 return p;
|
|
|
12912 }
|
|
|
12913
|
|
475
|
12914 #if defined(FEAT_INS_EXPAND) || defined(PROTO)
|
|
|
12915 /*
|
|
|
12916 * Find start of the word in front of the cursor. We don't check if it is
|
|
|
12917 * badly spelled, with completion we can only change the word in front of the
|
|
|
12918 * cursor.
|
|
|
12919 * Used for Insert mode completion CTRL-X ?.
|
|
|
12920 * Returns the column number of the word.
|
|
|
12921 */
|
|
|
12922 int
|
|
|
12923 spell_word_start(startcol)
|
|
|
12924 int startcol;
|
|
|
12925 {
|
|
|
12926 char_u *line;
|
|
|
12927 char_u *p;
|
|
|
12928 int col = 0;
|
|
|
12929
|
|
498
|
12930 if (no_spell_checking(curwin))
|
|
475
|
12931 return startcol;
|
|
|
12932
|
|
|
12933 /* Find a word character before "startcol". */
|
|
|
12934 line = ml_get_curline();
|
|
|
12935 for (p = line + startcol; p > line; )
|
|
|
12936 {
|
|
|
12937 mb_ptr_back(line, p);
|
|
|
12938 if (spell_iswordp_nmw(p))
|
|
|
12939 break;
|
|
|
12940 }
|
|
|
12941
|
|
|
12942 /* Go back to start of the word. */
|
|
|
12943 while (p > line)
|
|
|
12944 {
|
|
|
12945 col = p - line;
|
|
|
12946 mb_ptr_back(line, p);
|
|
|
12947 if (!spell_iswordp(p, curbuf))
|
|
|
12948 break;
|
|
|
12949 col = 0;
|
|
|
12950 }
|
|
|
12951
|
|
535
|
12952 return col;
|
|
|
12953 }
|
|
|
12954
|
|
|
12955 /*
|
|
|
12956 * Need to check for 'spellcapcheck' now, the word is removed before
|
|
|
12957 * expand_spelling() is called. Therefore the ugly global variable.
|
|
|
12958 */
|
|
|
12959 static int spell_expand_need_cap;
|
|
|
12960
|
|
|
12961 void
|
|
|
12962 spell_expand_check_cap(col)
|
|
|
12963 colnr_T col;
|
|
|
12964 {
|
|
475
|
12965 spell_expand_need_cap = check_need_cap(curwin->w_cursor.lnum, col);
|
|
|
12966 }
|
|
|
12967
|
|
|
12968 /*
|
|
|
12969 * Get list of spelling suggestions.
|
|
|
12970 * Used for Insert mode completion CTRL-X ?.
|
|
|
12971 * Returns the number of matches. The matches are in "matchp[]", array of
|
|
|
12972 * allocated strings.
|
|
|
12973 */
|
|
|
12974 /*ARGSUSED*/
|
|
|
12975 int
|
|
|
12976 expand_spelling(lnum, col, pat, matchp)
|
|
|
12977 linenr_T lnum;
|
|
|
12978 int col;
|
|
|
12979 char_u *pat;
|
|
|
12980 char_u ***matchp;
|
|
|
12981 {
|
|
|
12982 garray_T ga;
|
|
|
12983
|
|
|
12984 spell_suggest_list(&ga, pat, 100, spell_expand_need_cap);
|
|
|
12985 *matchp = ga.ga_data;
|
|
|
12986 return ga.ga_len;
|
|
|
12987 }
|
|
|
12988 #endif
|
|
|
12989
|
|
236
|
12990 #endif /* FEAT_SYN_HL */
|
