Mercurial > vim
annotate src/regexp_nfa.c @ 4573:37e444aeb33b v7.3.1034
updated for version 7.3.1034
Problem: New regexp code using strange multi-byte code.
Solution: Use the normal code to advance and backup pointers.
author | Bram Moolenaar <bram@vim.org> |
---|---|
date | Tue, 28 May 2013 22:30:35 +0200 |
parents | b2a8e3a66f8c |
children | b22bff1a6af8 |
rev | line source |
---|---|
4444 | 1 /* vi:set ts=8 sts=4 sw=4: |
2 * | |
3 * NFA regular expression implementation. | |
4 * | |
5 * This file is included in "regexp.c". | |
6 */ | |
7 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
8 /* |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
9 * Logging of NFA engine. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
10 * |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
11 * The NFA engine can write four log files: |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
12 * - Error log: Contains NFA engine's fatal errors. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
13 * - Dump log: Contains compiled NFA state machine's information. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
14 * - Run log: Contains information of matching procedure. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
15 * - Debug log: Contains detailed information of matching procedure. Can be |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
16 * disabled by undefining NFA_REGEXP_DEBUG_LOG. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
17 * The first one can also be used without debug mode. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
18 * The last three are enabled when compiled as debug mode and individually |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
19 * disabled by commenting them out. |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
20 * The log files can get quite big! |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
21 * Do disable all of this when compiling Vim for debugging, undefine DEBUG in |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
22 * regexp.c |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
23 */ |
4444 | 24 #ifdef DEBUG |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
25 # define NFA_REGEXP_ERROR_LOG "nfa_regexp_error.log" |
4444 | 26 # define ENABLE_LOG |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
27 # define NFA_REGEXP_DUMP_LOG "nfa_regexp_dump.log" |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
28 # define NFA_REGEXP_RUN_LOG "nfa_regexp_run.log" |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
29 # define NFA_REGEXP_DEBUG_LOG "nfa_regexp_debug.log" |
4444 | 30 #endif |
31 | |
32 /* Upper limit allowed for {m,n} repetitions handled by NFA */ | |
33 #define NFA_BRACES_MAXLIMIT 10 | |
34 /* For allocating space for the postfix representation */ | |
35 #define NFA_POSTFIX_MULTIPLIER (NFA_BRACES_MAXLIMIT + 2)*2 | |
36 | |
37 enum | |
38 { | |
39 NFA_SPLIT = -1024, | |
40 NFA_MATCH, | |
41 NFA_SKIP_CHAR, /* matches a 0-length char */ | |
42 NFA_END_NEG_RANGE, /* Used when expanding [^ab] */ | |
43 | |
44 NFA_CONCAT, | |
45 NFA_OR, | |
46 NFA_STAR, | |
47 NFA_PLUS, | |
48 NFA_QUEST, | |
49 NFA_QUEST_NONGREEDY, /* Non-greedy version of \? */ | |
50 NFA_NOT, /* used for [^ab] negated char ranges */ | |
51 | |
52 NFA_BOL, /* ^ Begin line */ | |
53 NFA_EOL, /* $ End line */ | |
54 NFA_BOW, /* \< Begin word */ | |
55 NFA_EOW, /* \> End word */ | |
56 NFA_BOF, /* \%^ Begin file */ | |
57 NFA_EOF, /* \%$ End file */ | |
58 NFA_NEWL, | |
59 NFA_ZSTART, /* Used for \zs */ | |
60 NFA_ZEND, /* Used for \ze */ | |
61 NFA_NOPEN, /* Start of subexpression marked with \%( */ | |
62 NFA_NCLOSE, /* End of subexpr. marked with \%( ... \) */ | |
63 NFA_START_INVISIBLE, | |
64 NFA_END_INVISIBLE, | |
65 NFA_COMPOSING, /* Next nodes in NFA are part of the | |
66 composing multibyte char */ | |
67 NFA_END_COMPOSING, /* End of a composing char in the NFA */ | |
68 | |
69 /* The following are used only in the postfix form, not in the NFA */ | |
70 NFA_PREV_ATOM_NO_WIDTH, /* Used for \@= */ | |
71 NFA_PREV_ATOM_NO_WIDTH_NEG, /* Used for \@! */ | |
72 NFA_PREV_ATOM_JUST_BEFORE, /* Used for \@<= */ | |
73 NFA_PREV_ATOM_JUST_BEFORE_NEG, /* Used for \@<! */ | |
74 NFA_PREV_ATOM_LIKE_PATTERN, /* Used for \@> */ | |
75 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
76 NFA_BACKREF1, /* \1 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
77 NFA_BACKREF2, /* \2 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
78 NFA_BACKREF3, /* \3 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
79 NFA_BACKREF4, /* \4 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
80 NFA_BACKREF5, /* \5 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
81 NFA_BACKREF6, /* \6 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
82 NFA_BACKREF7, /* \7 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
83 NFA_BACKREF8, /* \8 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
84 NFA_BACKREF9, /* \9 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
85 NFA_SKIP, /* Skip characters */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
86 |
4444 | 87 NFA_MOPEN, |
88 NFA_MCLOSE = NFA_MOPEN + NSUBEXP, | |
89 | |
90 /* NFA_FIRST_NL */ | |
91 NFA_ANY = NFA_MCLOSE + NSUBEXP, /* Match any one character. */ | |
92 NFA_ANYOF, /* Match any character in this string. */ | |
93 NFA_ANYBUT, /* Match any character not in this string. */ | |
94 NFA_IDENT, /* Match identifier char */ | |
95 NFA_SIDENT, /* Match identifier char but no digit */ | |
96 NFA_KWORD, /* Match keyword char */ | |
97 NFA_SKWORD, /* Match word char but no digit */ | |
98 NFA_FNAME, /* Match file name char */ | |
99 NFA_SFNAME, /* Match file name char but no digit */ | |
100 NFA_PRINT, /* Match printable char */ | |
101 NFA_SPRINT, /* Match printable char but no digit */ | |
102 NFA_WHITE, /* Match whitespace char */ | |
103 NFA_NWHITE, /* Match non-whitespace char */ | |
104 NFA_DIGIT, /* Match digit char */ | |
105 NFA_NDIGIT, /* Match non-digit char */ | |
106 NFA_HEX, /* Match hex char */ | |
107 NFA_NHEX, /* Match non-hex char */ | |
108 NFA_OCTAL, /* Match octal char */ | |
109 NFA_NOCTAL, /* Match non-octal char */ | |
110 NFA_WORD, /* Match word char */ | |
111 NFA_NWORD, /* Match non-word char */ | |
112 NFA_HEAD, /* Match head char */ | |
113 NFA_NHEAD, /* Match non-head char */ | |
114 NFA_ALPHA, /* Match alpha char */ | |
115 NFA_NALPHA, /* Match non-alpha char */ | |
116 NFA_LOWER, /* Match lowercase char */ | |
117 NFA_NLOWER, /* Match non-lowercase char */ | |
118 NFA_UPPER, /* Match uppercase char */ | |
119 NFA_NUPPER, /* Match non-uppercase char */ | |
120 NFA_FIRST_NL = NFA_ANY + ADD_NL, | |
121 NFA_LAST_NL = NFA_NUPPER + ADD_NL, | |
122 | |
123 /* Character classes [:alnum:] etc */ | |
124 NFA_CLASS_ALNUM, | |
125 NFA_CLASS_ALPHA, | |
126 NFA_CLASS_BLANK, | |
127 NFA_CLASS_CNTRL, | |
128 NFA_CLASS_DIGIT, | |
129 NFA_CLASS_GRAPH, | |
130 NFA_CLASS_LOWER, | |
131 NFA_CLASS_PRINT, | |
132 NFA_CLASS_PUNCT, | |
133 NFA_CLASS_SPACE, | |
134 NFA_CLASS_UPPER, | |
135 NFA_CLASS_XDIGIT, | |
136 NFA_CLASS_TAB, | |
137 NFA_CLASS_RETURN, | |
138 NFA_CLASS_BACKSPACE, | |
139 NFA_CLASS_ESCAPE | |
140 }; | |
141 | |
142 /* Keep in sync with classchars. */ | |
143 static int nfa_classcodes[] = { | |
144 NFA_ANY, NFA_IDENT, NFA_SIDENT, NFA_KWORD,NFA_SKWORD, | |
145 NFA_FNAME, NFA_SFNAME, NFA_PRINT, NFA_SPRINT, | |
146 NFA_WHITE, NFA_NWHITE, NFA_DIGIT, NFA_NDIGIT, | |
147 NFA_HEX, NFA_NHEX, NFA_OCTAL, NFA_NOCTAL, | |
148 NFA_WORD, NFA_NWORD, NFA_HEAD, NFA_NHEAD, | |
149 NFA_ALPHA, NFA_NALPHA, NFA_LOWER, NFA_NLOWER, | |
150 NFA_UPPER, NFA_NUPPER | |
151 }; | |
152 | |
153 static char_u e_misplaced[] = N_("E866: (NFA regexp) Misplaced %c"); | |
154 | |
155 /* | |
156 * NFA errors can be of 3 types: | |
157 * *** NFA runtime errors, when something unknown goes wrong. The NFA fails | |
158 * silently and revert the to backtracking engine. | |
159 * syntax_error = FALSE; | |
160 * *** Regexp syntax errors, when the input regexp is not syntactically correct. | |
161 * The NFA engine displays an error message, and nothing else happens. | |
162 * syntax_error = TRUE | |
163 * *** Unsupported features, when the input regexp uses an operator that is not | |
164 * implemented in the NFA. The NFA engine fails silently, and reverts to the | |
165 * old backtracking engine. | |
166 * syntax_error = FALSE | |
167 * "The NFA fails" means that "compiling the regexp with the NFA fails": | |
168 * nfa_regcomp() returns FAIL. | |
169 */ | |
170 static int syntax_error = FALSE; | |
171 | |
172 /* NFA regexp \ze operator encountered. */ | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
173 static int nfa_has_zend; |
4444 | 174 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
175 /* Number of sub expressions actually being used during execution. 1 if only |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
176 * the whole match (subexpr 0) is used. */ |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
177 static int nfa_nsubexpr; |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
178 |
4444 | 179 static int *post_start; /* holds the postfix form of r.e. */ |
180 static int *post_end; | |
181 static int *post_ptr; | |
182 | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
183 static int nstate; /* Number of states in the NFA. Also used when |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
184 * executing. */ |
4444 | 185 static int istate; /* Index in the state vector, used in new_state() */ |
186 | |
187 | |
188 static int nfa_regcomp_start __ARGS((char_u*expr, int re_flags)); | |
189 static int nfa_recognize_char_class __ARGS((char_u *start, char_u *end, int extra_newl)); | |
190 static int nfa_emit_equi_class __ARGS((int c, int neg)); | |
191 static int nfa_regatom __ARGS((void)); | |
192 static int nfa_regpiece __ARGS((void)); | |
193 static int nfa_regconcat __ARGS((void)); | |
194 static int nfa_regbranch __ARGS((void)); | |
195 static int nfa_reg __ARGS((int paren)); | |
196 #ifdef DEBUG | |
197 static void nfa_set_code __ARGS((int c)); | |
198 static void nfa_postfix_dump __ARGS((char_u *expr, int retval)); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
199 static void nfa_print_state __ARGS((FILE *debugf, nfa_state_T *state)); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
200 static void nfa_print_state2 __ARGS((FILE *debugf, nfa_state_T *state, garray_T *indent)); |
4444 | 201 static void nfa_dump __ARGS((nfa_regprog_T *prog)); |
202 #endif | |
203 static int *re2post __ARGS((void)); | |
204 static nfa_state_T *new_state __ARGS((int c, nfa_state_T *out, nfa_state_T *out1)); | |
205 static nfa_state_T *post2nfa __ARGS((int *postfix, int *end, int nfa_calc_size)); | |
206 static int check_char_class __ARGS((int class, int c)); | |
207 static void st_error __ARGS((int *postfix, int *end, int *p)); | |
208 static void nfa_save_listids __ARGS((nfa_state_T *start, int *list)); | |
209 static void nfa_restore_listids __ARGS((nfa_state_T *start, int *list)); | |
210 static void nfa_set_null_listids __ARGS((nfa_state_T *start)); | |
211 static void nfa_set_neg_listids __ARGS((nfa_state_T *start)); | |
212 static long nfa_regtry __ARGS((nfa_state_T *start, colnr_T col)); | |
213 static long nfa_regexec_both __ARGS((char_u *line, colnr_T col)); | |
214 static regprog_T *nfa_regcomp __ARGS((char_u *expr, int re_flags)); | |
215 static int nfa_regexec __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
216 static long nfa_regexec_multi __ARGS((regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm)); | |
217 | |
218 /* helper functions used when doing re2post() ... regatom() parsing */ | |
219 #define EMIT(c) do { \ | |
220 if (post_ptr >= post_end) \ | |
221 return FAIL; \ | |
222 *post_ptr++ = c; \ | |
223 } while (0) | |
224 | |
225 /* | |
226 * Initialize internal variables before NFA compilation. | |
227 * Return OK on success, FAIL otherwise. | |
228 */ | |
229 static int | |
230 nfa_regcomp_start(expr, re_flags) | |
231 char_u *expr; | |
232 int re_flags; /* see vim_regcomp() */ | |
233 { | |
4458 | 234 size_t postfix_size; |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
235 int nstate_max; |
4444 | 236 |
237 nstate = 0; | |
238 istate = 0; | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
239 /* A reasonable estimation for maximum size */ |
4458 | 240 nstate_max = (int)(STRLEN(expr) + 1) * NFA_POSTFIX_MULTIPLIER; |
4444 | 241 |
4454 | 242 /* Some items blow up in size, such as [A-z]. Add more space for that. |
243 * TODO: some patterns may still fail. */ | |
4458 | 244 nstate_max += 1000; |
4454 | 245 |
246 /* Size for postfix representation of expr. */ | |
4444 | 247 postfix_size = sizeof(*post_start) * nstate_max; |
4454 | 248 |
4444 | 249 post_start = (int *)lalloc(postfix_size, TRUE); |
250 if (post_start == NULL) | |
251 return FAIL; | |
252 vim_memset(post_start, 0, postfix_size); | |
253 post_ptr = post_start; | |
4454 | 254 post_end = post_start + nstate_max; |
4444 | 255 nfa_has_zend = FALSE; |
256 | |
257 regcomp_start(expr, re_flags); | |
258 | |
259 return OK; | |
260 } | |
261 | |
262 /* | |
263 * Search between "start" and "end" and try to recognize a | |
264 * character class in expanded form. For example [0-9]. | |
265 * On success, return the id the character class to be emitted. | |
266 * On failure, return 0 (=FAIL) | |
267 * Start points to the first char of the range, while end should point | |
268 * to the closing brace. | |
269 */ | |
270 static int | |
271 nfa_recognize_char_class(start, end, extra_newl) | |
272 char_u *start; | |
273 char_u *end; | |
274 int extra_newl; | |
275 { | |
276 int i; | |
277 /* Each of these variables takes up a char in "config[]", | |
278 * in the order they are here. */ | |
279 int not = FALSE, af = FALSE, AF = FALSE, az = FALSE, AZ = FALSE, | |
280 o7 = FALSE, o9 = FALSE, underscore = FALSE, newl = FALSE; | |
281 char_u *p; | |
282 #define NCONFIGS 16 | |
283 int classid[NCONFIGS] = { | |
284 NFA_DIGIT, NFA_NDIGIT, NFA_HEX, NFA_NHEX, | |
285 NFA_OCTAL, NFA_NOCTAL, NFA_WORD, NFA_NWORD, | |
286 NFA_HEAD, NFA_NHEAD, NFA_ALPHA, NFA_NALPHA, | |
287 NFA_LOWER, NFA_NLOWER, NFA_UPPER, NFA_NUPPER | |
288 }; | |
4450 | 289 char_u myconfig[10]; |
4444 | 290 char_u config[NCONFIGS][9] = { |
291 "000000100", /* digit */ | |
292 "100000100", /* non digit */ | |
293 "011000100", /* hex-digit */ | |
294 "111000100", /* non hex-digit */ | |
295 "000001000", /* octal-digit */ | |
296 "100001000", /* [^0-7] */ | |
297 "000110110", /* [0-9A-Za-z_] */ | |
298 "100110110", /* [^0-9A-Za-z_] */ | |
299 "000110010", /* head of word */ | |
300 "100110010", /* not head of word */ | |
301 "000110000", /* alphabetic char a-z */ | |
302 "100110000", /* non alphabetic char */ | |
303 "000100000", /* lowercase letter */ | |
304 "100100000", /* non lowercase */ | |
305 "000010000", /* uppercase */ | |
306 "100010000" /* non uppercase */ | |
307 }; | |
308 | |
309 if (extra_newl == TRUE) | |
310 newl = TRUE; | |
311 | |
312 if (*end != ']') | |
313 return FAIL; | |
314 p = start; | |
315 if (*p == '^') | |
316 { | |
317 not = TRUE; | |
318 p ++; | |
319 } | |
320 | |
321 while (p < end) | |
322 { | |
323 if (p + 2 < end && *(p + 1) == '-') | |
324 { | |
325 switch (*p) | |
326 { | |
327 case '0': | |
328 if (*(p + 2) == '9') | |
329 { | |
330 o9 = TRUE; | |
331 break; | |
332 } | |
333 else | |
334 if (*(p + 2) == '7') | |
335 { | |
336 o7 = TRUE; | |
337 break; | |
338 } | |
339 case 'a': | |
340 if (*(p + 2) == 'z') | |
341 { | |
342 az = TRUE; | |
343 break; | |
344 } | |
345 else | |
346 if (*(p + 2) == 'f') | |
347 { | |
348 af = TRUE; | |
349 break; | |
350 } | |
351 case 'A': | |
352 if (*(p + 2) == 'Z') | |
353 { | |
354 AZ = TRUE; | |
355 break; | |
356 } | |
357 else | |
358 if (*(p + 2) == 'F') | |
359 { | |
360 AF = TRUE; | |
361 break; | |
362 } | |
363 /* FALLTHROUGH */ | |
364 default: | |
365 return FAIL; | |
366 } | |
367 p += 3; | |
368 } | |
369 else if (p + 1 < end && *p == '\\' && *(p + 1) == 'n') | |
370 { | |
371 newl = TRUE; | |
372 p += 2; | |
373 } | |
374 else if (*p == '_') | |
375 { | |
376 underscore = TRUE; | |
377 p ++; | |
378 } | |
379 else if (*p == '\n') | |
380 { | |
381 newl = TRUE; | |
382 p ++; | |
383 } | |
384 else | |
385 return FAIL; | |
386 } /* while (p < end) */ | |
387 | |
388 if (p != end) | |
389 return FAIL; | |
390 | |
391 /* build the config that represents the ranges we gathered */ | |
392 STRCPY(myconfig, "000000000"); | |
393 if (not == TRUE) | |
394 myconfig[0] = '1'; | |
395 if (af == TRUE) | |
396 myconfig[1] = '1'; | |
397 if (AF == TRUE) | |
398 myconfig[2] = '1'; | |
399 if (az == TRUE) | |
400 myconfig[3] = '1'; | |
401 if (AZ == TRUE) | |
402 myconfig[4] = '1'; | |
403 if (o7 == TRUE) | |
404 myconfig[5] = '1'; | |
405 if (o9 == TRUE) | |
406 myconfig[6] = '1'; | |
407 if (underscore == TRUE) | |
408 myconfig[7] = '1'; | |
409 if (newl == TRUE) | |
410 { | |
411 myconfig[8] = '1'; | |
412 extra_newl = ADD_NL; | |
413 } | |
414 /* try to recognize character classes */ | |
415 for (i = 0; i < NCONFIGS; i++) | |
4450 | 416 if (STRNCMP(myconfig, config[i], 8) == 0) |
4444 | 417 return classid[i] + extra_newl; |
418 | |
419 /* fallthrough => no success so far */ | |
420 return FAIL; | |
421 | |
422 #undef NCONFIGS | |
423 } | |
424 | |
425 /* | |
426 * Produce the bytes for equivalence class "c". | |
427 * Currently only handles latin1, latin9 and utf-8. | |
428 * Emits bytes in postfix notation: 'a,b,NFA_OR,c,NFA_OR' is | |
429 * equivalent to 'a OR b OR c' | |
430 * | |
431 * NOTE! When changing this function, also update reg_equi_class() | |
432 */ | |
433 static int | |
434 nfa_emit_equi_class(c, neg) | |
435 int c; | |
436 int neg; | |
437 { | |
438 int first = TRUE; | |
439 int glue = neg == TRUE ? NFA_CONCAT : NFA_OR; | |
440 #define EMIT2(c) \ | |
441 EMIT(c); \ | |
442 if (neg == TRUE) { \ | |
443 EMIT(NFA_NOT); \ | |
444 } \ | |
445 if (first == FALSE) \ | |
446 EMIT(glue); \ | |
447 else \ | |
448 first = FALSE; \ | |
449 | |
450 #ifdef FEAT_MBYTE | |
451 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
452 || STRCMP(p_enc, "iso-8859-15") == 0) | |
453 #endif | |
454 { | |
455 switch (c) | |
456 { | |
457 case 'A': case '\300': case '\301': case '\302': | |
458 case '\303': case '\304': case '\305': | |
459 EMIT2('A'); EMIT2('\300'); EMIT2('\301'); | |
460 EMIT2('\302'); EMIT2('\303'); EMIT2('\304'); | |
461 EMIT2('\305'); | |
462 return OK; | |
463 | |
464 case 'C': case '\307': | |
465 EMIT2('C'); EMIT2('\307'); | |
466 return OK; | |
467 | |
468 case 'E': case '\310': case '\311': case '\312': case '\313': | |
469 EMIT2('E'); EMIT2('\310'); EMIT2('\311'); | |
470 EMIT2('\312'); EMIT2('\313'); | |
471 return OK; | |
472 | |
473 case 'I': case '\314': case '\315': case '\316': case '\317': | |
474 EMIT2('I'); EMIT2('\314'); EMIT2('\315'); | |
475 EMIT2('\316'); EMIT2('\317'); | |
476 return OK; | |
477 | |
478 case 'N': case '\321': | |
479 EMIT2('N'); EMIT2('\321'); | |
480 return OK; | |
481 | |
482 case 'O': case '\322': case '\323': case '\324': case '\325': | |
483 case '\326': | |
484 EMIT2('O'); EMIT2('\322'); EMIT2('\323'); | |
485 EMIT2('\324'); EMIT2('\325'); EMIT2('\326'); | |
486 return OK; | |
487 | |
488 case 'U': case '\331': case '\332': case '\333': case '\334': | |
489 EMIT2('U'); EMIT2('\331'); EMIT2('\332'); | |
490 EMIT2('\333'); EMIT2('\334'); | |
491 return OK; | |
492 | |
493 case 'Y': case '\335': | |
494 EMIT2('Y'); EMIT2('\335'); | |
495 return OK; | |
496 | |
497 case 'a': case '\340': case '\341': case '\342': | |
498 case '\343': case '\344': case '\345': | |
499 EMIT2('a'); EMIT2('\340'); EMIT2('\341'); | |
500 EMIT2('\342'); EMIT2('\343'); EMIT2('\344'); | |
501 EMIT2('\345'); | |
502 return OK; | |
503 | |
504 case 'c': case '\347': | |
505 EMIT2('c'); EMIT2('\347'); | |
506 return OK; | |
507 | |
508 case 'e': case '\350': case '\351': case '\352': case '\353': | |
509 EMIT2('e'); EMIT2('\350'); EMIT2('\351'); | |
510 EMIT2('\352'); EMIT2('\353'); | |
511 return OK; | |
512 | |
513 case 'i': case '\354': case '\355': case '\356': case '\357': | |
514 EMIT2('i'); EMIT2('\354'); EMIT2('\355'); | |
515 EMIT2('\356'); EMIT2('\357'); | |
516 return OK; | |
517 | |
518 case 'n': case '\361': | |
519 EMIT2('n'); EMIT2('\361'); | |
520 return OK; | |
521 | |
522 case 'o': case '\362': case '\363': case '\364': case '\365': | |
523 case '\366': | |
524 EMIT2('o'); EMIT2('\362'); EMIT2('\363'); | |
525 EMIT2('\364'); EMIT2('\365'); EMIT2('\366'); | |
526 return OK; | |
527 | |
528 case 'u': case '\371': case '\372': case '\373': case '\374': | |
529 EMIT2('u'); EMIT2('\371'); EMIT2('\372'); | |
530 EMIT2('\373'); EMIT2('\374'); | |
531 return OK; | |
532 | |
533 case 'y': case '\375': case '\377': | |
534 EMIT2('y'); EMIT2('\375'); EMIT2('\377'); | |
535 return OK; | |
536 | |
537 default: | |
538 return FAIL; | |
539 } | |
540 } | |
541 | |
542 EMIT(c); | |
543 return OK; | |
544 #undef EMIT2 | |
545 } | |
546 | |
547 /* | |
548 * Code to parse regular expression. | |
549 * | |
550 * We try to reuse parsing functions in regexp.c to | |
551 * minimize surprise and keep the syntax consistent. | |
552 */ | |
553 | |
554 /* | |
555 * Parse the lowest level. | |
556 * | |
557 * An atom can be one of a long list of items. Many atoms match one character | |
558 * in the text. It is often an ordinary character or a character class. | |
559 * Braces can be used to make a pattern into an atom. The "\z(\)" construct | |
560 * is only for syntax highlighting. | |
561 * | |
562 * atom ::= ordinary-atom | |
563 * or \( pattern \) | |
564 * or \%( pattern \) | |
565 * or \z( pattern \) | |
566 */ | |
567 static int | |
568 nfa_regatom() | |
569 { | |
570 int c; | |
571 int charclass; | |
572 int equiclass; | |
573 int collclass; | |
574 int got_coll_char; | |
575 char_u *p; | |
576 char_u *endp; | |
577 #ifdef FEAT_MBYTE | |
578 char_u *old_regparse = regparse; | |
579 #endif | |
580 int extra = 0; | |
581 int first; | |
582 int emit_range; | |
583 int negated; | |
584 int result; | |
585 int startc = -1; | |
586 int endc = -1; | |
587 int oldstartc = -1; | |
588 int cpo_lit; /* 'cpoptions' contains 'l' flag */ | |
589 int cpo_bsl; /* 'cpoptions' contains '\' flag */ | |
590 int glue; /* ID that will "glue" nodes together */ | |
591 | |
592 cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; | |
593 cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; | |
594 | |
595 c = getchr(); | |
596 switch (c) | |
597 { | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
598 case NUL: |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
599 syntax_error = TRUE; |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
600 EMSG_RET_FAIL(_("E865: (NFA) Regexp end encountered prematurely")); |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
601 |
4444 | 602 case Magic('^'): |
603 EMIT(NFA_BOL); | |
604 break; | |
605 | |
606 case Magic('$'): | |
607 EMIT(NFA_EOL); | |
608 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
609 had_eol = TRUE; | |
610 #endif | |
611 break; | |
612 | |
613 case Magic('<'): | |
614 EMIT(NFA_BOW); | |
615 break; | |
616 | |
617 case Magic('>'): | |
618 EMIT(NFA_EOW); | |
619 break; | |
620 | |
621 case Magic('_'): | |
622 c = no_Magic(getchr()); | |
623 if (c == '^') /* "\_^" is start-of-line */ | |
624 { | |
625 EMIT(NFA_BOL); | |
626 break; | |
627 } | |
628 if (c == '$') /* "\_$" is end-of-line */ | |
629 { | |
630 EMIT(NFA_EOL); | |
631 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
632 had_eol = TRUE; | |
633 #endif | |
634 break; | |
635 } | |
636 | |
637 extra = ADD_NL; | |
638 | |
639 /* "\_[" is collection plus newline */ | |
640 if (c == '[') | |
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
641 goto collection; |
4444 | 642 |
643 /* "\_x" is character class plus newline */ | |
644 /*FALLTHROUGH*/ | |
645 | |
646 /* | |
647 * Character classes. | |
648 */ | |
649 case Magic('.'): | |
650 case Magic('i'): | |
651 case Magic('I'): | |
652 case Magic('k'): | |
653 case Magic('K'): | |
654 case Magic('f'): | |
655 case Magic('F'): | |
656 case Magic('p'): | |
657 case Magic('P'): | |
658 case Magic('s'): | |
659 case Magic('S'): | |
660 case Magic('d'): | |
661 case Magic('D'): | |
662 case Magic('x'): | |
663 case Magic('X'): | |
664 case Magic('o'): | |
665 case Magic('O'): | |
666 case Magic('w'): | |
667 case Magic('W'): | |
668 case Magic('h'): | |
669 case Magic('H'): | |
670 case Magic('a'): | |
671 case Magic('A'): | |
672 case Magic('l'): | |
673 case Magic('L'): | |
674 case Magic('u'): | |
675 case Magic('U'): | |
676 p = vim_strchr(classchars, no_Magic(c)); | |
677 if (p == NULL) | |
678 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
679 EMSGN("INTERNAL: Unknown character class char: %ld", c); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
680 return FAIL; |
4444 | 681 } |
682 #ifdef FEAT_MBYTE | |
683 /* When '.' is followed by a composing char ignore the dot, so that | |
684 * the composing char is matched here. */ | |
685 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
686 { | |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
687 old_regparse = regparse; |
4444 | 688 c = getchr(); |
689 goto nfa_do_multibyte; | |
690 } | |
691 #endif | |
692 EMIT(nfa_classcodes[p - classchars]); | |
693 if (extra == ADD_NL) | |
694 { | |
695 EMIT(NFA_NEWL); | |
696 EMIT(NFA_OR); | |
697 regflags |= RF_HASNL; | |
698 } | |
699 break; | |
700 | |
701 case Magic('n'): | |
702 if (reg_string) | |
703 /* In a string "\n" matches a newline character. */ | |
704 EMIT(NL); | |
705 else | |
706 { | |
707 /* In buffer text "\n" matches the end of a line. */ | |
708 EMIT(NFA_NEWL); | |
709 regflags |= RF_HASNL; | |
710 } | |
711 break; | |
712 | |
713 case Magic('('): | |
714 if (nfa_reg(REG_PAREN) == FAIL) | |
715 return FAIL; /* cascaded error */ | |
716 break; | |
717 | |
718 case Magic('|'): | |
719 case Magic('&'): | |
720 case Magic(')'): | |
721 syntax_error = TRUE; | |
4450 | 722 EMSGN(_(e_misplaced), no_Magic(c)); |
4444 | 723 return FAIL; |
724 | |
725 case Magic('='): | |
726 case Magic('?'): | |
727 case Magic('+'): | |
728 case Magic('@'): | |
729 case Magic('*'): | |
730 case Magic('{'): | |
731 /* these should follow an atom, not form an atom */ | |
732 syntax_error = TRUE; | |
4450 | 733 EMSGN(_(e_misplaced), no_Magic(c)); |
4444 | 734 return FAIL; |
735 | |
736 case Magic('~'): /* previous substitute pattern */ | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
737 /* TODO: Not supported yet */ |
4444 | 738 return FAIL; |
739 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
740 case Magic('1'): EMIT(NFA_BACKREF1); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
741 case Magic('2'): EMIT(NFA_BACKREF2); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
742 case Magic('3'): EMIT(NFA_BACKREF3); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
743 case Magic('4'): EMIT(NFA_BACKREF4); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
744 case Magic('5'): EMIT(NFA_BACKREF5); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
745 case Magic('6'): EMIT(NFA_BACKREF6); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
746 case Magic('7'): EMIT(NFA_BACKREF7); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
747 case Magic('8'): EMIT(NFA_BACKREF8); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
748 case Magic('9'): EMIT(NFA_BACKREF9); break; |
4444 | 749 |
750 case Magic('z'): | |
751 c = no_Magic(getchr()); | |
752 switch (c) | |
753 { | |
754 case 's': | |
755 EMIT(NFA_ZSTART); | |
756 break; | |
757 case 'e': | |
758 EMIT(NFA_ZEND); | |
759 nfa_has_zend = TRUE; | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
760 break; |
4444 | 761 case '1': |
762 case '2': | |
763 case '3': | |
764 case '4': | |
765 case '5': | |
766 case '6': | |
767 case '7': | |
768 case '8': | |
769 case '9': | |
770 case '(': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
771 /* TODO: \z1...\z9 and \z( not yet supported */ |
4444 | 772 return FAIL; |
773 default: | |
774 syntax_error = TRUE; | |
4450 | 775 EMSGN(_("E867: (NFA) Unknown operator '\\z%c'"), |
4444 | 776 no_Magic(c)); |
777 return FAIL; | |
778 } | |
779 break; | |
780 | |
781 case Magic('%'): | |
782 c = no_Magic(getchr()); | |
783 switch (c) | |
784 { | |
785 /* () without a back reference */ | |
786 case '(': | |
787 if (nfa_reg(REG_NPAREN) == FAIL) | |
788 return FAIL; | |
789 EMIT(NFA_NOPEN); | |
790 break; | |
791 | |
792 case 'd': /* %d123 decimal */ | |
793 case 'o': /* %o123 octal */ | |
794 case 'x': /* %xab hex 2 */ | |
795 case 'u': /* %uabcd hex 4 */ | |
796 case 'U': /* %U1234abcd hex 8 */ | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
797 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
798 int nr; |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
799 |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
800 switch (c) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
801 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
802 case 'd': nr = getdecchrs(); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
803 case 'o': nr = getoctchrs(); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
804 case 'x': nr = gethexchrs(2); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
805 case 'u': nr = gethexchrs(4); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
806 case 'U': nr = gethexchrs(8); break; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
807 default: nr = -1; break; |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
808 } |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
809 |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
810 if (nr < 0) |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
811 EMSG2_RET_FAIL( |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
812 _("E678: Invalid character after %s%%[dxouU]"), |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
813 reg_magic == MAGIC_ALL); |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
814 /* TODO: what if a composing character follows? */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
815 EMIT(nr); |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
816 } |
4444 | 817 break; |
818 | |
819 /* Catch \%^ and \%$ regardless of where they appear in the | |
820 * pattern -- regardless of whether or not it makes sense. */ | |
821 case '^': | |
822 EMIT(NFA_BOF); | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
823 /* TODO: Not yet supported */ |
4444 | 824 return FAIL; |
825 break; | |
826 | |
827 case '$': | |
828 EMIT(NFA_EOF); | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
829 /* TODO: Not yet supported */ |
4444 | 830 return FAIL; |
831 break; | |
832 | |
833 case '#': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
834 /* TODO: not supported yet */ |
4444 | 835 return FAIL; |
836 break; | |
837 | |
838 case 'V': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
839 /* TODO: not supported yet */ |
4444 | 840 return FAIL; |
841 break; | |
842 | |
843 case '[': | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
844 /* TODO: \%[abc] not supported yet */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
845 return FAIL; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
846 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
847 case '0': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
848 case '1': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
849 case '2': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
850 case '3': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
851 case '4': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
852 case '5': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
853 case '6': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
854 case '7': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
855 case '8': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
856 case '9': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
857 case '<': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
858 case '>': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
859 case '\'': |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
860 /* TODO: not supported yet */ |
4444 | 861 return FAIL; |
862 | |
863 default: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
864 syntax_error = TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
865 EMSGN(_("E867: (NFA) Unknown operator '\\%%%c'"), |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
866 no_Magic(c)); |
4444 | 867 return FAIL; |
868 } | |
869 break; | |
870 | |
871 case Magic('['): | |
4517
9a2183bd8295
updated for version 7.3.1006
Bram Moolenaar <bram@vim.org>
parents:
4515
diff
changeset
|
872 collection: |
4444 | 873 /* |
874 * Glue is emitted between several atoms from the []. | |
875 * It is either NFA_OR, or NFA_CONCAT. | |
876 * | |
877 * [abc] expands to 'a b NFA_OR c NFA_OR' (in postfix notation) | |
878 * [^abc] expands to 'a NFA_NOT b NFA_NOT NFA_CONCAT c NFA_NOT | |
879 * NFA_CONCAT NFA_END_NEG_RANGE NFA_CONCAT' (in postfix | |
880 * notation) | |
881 * | |
882 */ | |
883 | |
884 | |
885 /* Emit negation atoms, if needed. | |
886 * The CONCAT below merges the NOT with the previous node. */ | |
887 #define TRY_NEG() \ | |
888 if (negated == TRUE) \ | |
889 { \ | |
890 EMIT(NFA_NOT); \ | |
891 } | |
892 | |
893 /* Emit glue between important nodes : CONCAT or OR. */ | |
894 #define EMIT_GLUE() \ | |
895 if (first == FALSE) \ | |
896 EMIT(glue); \ | |
897 else \ | |
898 first = FALSE; | |
899 | |
900 p = regparse; | |
901 endp = skip_anyof(p); | |
902 if (*endp == ']') | |
903 { | |
904 /* | |
905 * Try to reverse engineer character classes. For example, | |
906 * recognize that [0-9] stands for \d and [A-Za-z_] with \h, | |
907 * and perform the necessary substitutions in the NFA. | |
908 */ | |
909 result = nfa_recognize_char_class(regparse, endp, | |
910 extra == ADD_NL); | |
911 if (result != FAIL) | |
912 { | |
913 if (result >= NFA_DIGIT && result <= NFA_NUPPER) | |
914 EMIT(result); | |
915 else /* must be char class + newline */ | |
916 { | |
917 EMIT(result - ADD_NL); | |
918 EMIT(NFA_NEWL); | |
919 EMIT(NFA_OR); | |
920 } | |
921 regparse = endp; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
922 mb_ptr_adv(regparse); |
4444 | 923 return OK; |
924 } | |
925 /* | |
926 * Failed to recognize a character class. Use the simple | |
927 * version that turns [abc] into 'a' OR 'b' OR 'c' | |
928 */ | |
929 startc = endc = oldstartc = -1; | |
930 first = TRUE; /* Emitting first atom in this sequence? */ | |
931 negated = FALSE; | |
932 glue = NFA_OR; | |
933 if (*regparse == '^') /* negated range */ | |
934 { | |
935 negated = TRUE; | |
936 glue = NFA_CONCAT; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
937 mb_ptr_adv(regparse); |
4444 | 938 } |
939 if (*regparse == '-') | |
940 { | |
941 startc = '-'; | |
942 EMIT(startc); | |
943 TRY_NEG(); | |
944 EMIT_GLUE(); | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
945 mb_ptr_adv(regparse); |
4444 | 946 } |
947 /* Emit the OR branches for each character in the [] */ | |
948 emit_range = FALSE; | |
949 while (regparse < endp) | |
950 { | |
951 oldstartc = startc; | |
952 startc = -1; | |
953 got_coll_char = FALSE; | |
954 if (*regparse == '[') | |
955 { | |
956 /* Check for [: :], [= =], [. .] */ | |
957 equiclass = collclass = 0; | |
958 charclass = get_char_class(®parse); | |
959 if (charclass == CLASS_NONE) | |
960 { | |
961 equiclass = get_equi_class(®parse); | |
962 if (equiclass == 0) | |
963 collclass = get_coll_element(®parse); | |
964 } | |
965 | |
966 /* Character class like [:alpha:] */ | |
967 if (charclass != CLASS_NONE) | |
968 { | |
969 switch (charclass) | |
970 { | |
971 case CLASS_ALNUM: | |
972 EMIT(NFA_CLASS_ALNUM); | |
973 break; | |
974 case CLASS_ALPHA: | |
975 EMIT(NFA_CLASS_ALPHA); | |
976 break; | |
977 case CLASS_BLANK: | |
978 EMIT(NFA_CLASS_BLANK); | |
979 break; | |
980 case CLASS_CNTRL: | |
981 EMIT(NFA_CLASS_CNTRL); | |
982 break; | |
983 case CLASS_DIGIT: | |
984 EMIT(NFA_CLASS_DIGIT); | |
985 break; | |
986 case CLASS_GRAPH: | |
987 EMIT(NFA_CLASS_GRAPH); | |
988 break; | |
989 case CLASS_LOWER: | |
990 EMIT(NFA_CLASS_LOWER); | |
991 break; | |
992 case CLASS_PRINT: | |
993 EMIT(NFA_CLASS_PRINT); | |
994 break; | |
995 case CLASS_PUNCT: | |
996 EMIT(NFA_CLASS_PUNCT); | |
997 break; | |
998 case CLASS_SPACE: | |
999 EMIT(NFA_CLASS_SPACE); | |
1000 break; | |
1001 case CLASS_UPPER: | |
1002 EMIT(NFA_CLASS_UPPER); | |
1003 break; | |
1004 case CLASS_XDIGIT: | |
1005 EMIT(NFA_CLASS_XDIGIT); | |
1006 break; | |
1007 case CLASS_TAB: | |
1008 EMIT(NFA_CLASS_TAB); | |
1009 break; | |
1010 case CLASS_RETURN: | |
1011 EMIT(NFA_CLASS_RETURN); | |
1012 break; | |
1013 case CLASS_BACKSPACE: | |
1014 EMIT(NFA_CLASS_BACKSPACE); | |
1015 break; | |
1016 case CLASS_ESCAPE: | |
1017 EMIT(NFA_CLASS_ESCAPE); | |
1018 break; | |
1019 } | |
1020 TRY_NEG(); | |
1021 EMIT_GLUE(); | |
1022 continue; | |
1023 } | |
1024 /* Try equivalence class [=a=] and the like */ | |
1025 if (equiclass != 0) | |
1026 { | |
1027 result = nfa_emit_equi_class(equiclass, negated); | |
1028 if (result == FAIL) | |
1029 { | |
1030 /* should never happen */ | |
1031 EMSG_RET_FAIL(_("E868: Error building NFA with equivalence class!")); | |
1032 } | |
1033 EMIT_GLUE(); | |
1034 continue; | |
1035 } | |
1036 /* Try collating class like [. .] */ | |
1037 if (collclass != 0) | |
1038 { | |
1039 startc = collclass; /* allow [.a.]-x as a range */ | |
1040 /* Will emit the proper atom at the end of the | |
1041 * while loop. */ | |
1042 } | |
1043 } | |
1044 /* Try a range like 'a-x' or '\t-z' */ | |
1045 if (*regparse == '-') | |
1046 { | |
1047 emit_range = TRUE; | |
1048 startc = oldstartc; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1049 mb_ptr_adv(regparse); |
4444 | 1050 continue; /* reading the end of the range */ |
1051 } | |
1052 | |
1053 /* Now handle simple and escaped characters. | |
1054 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
1055 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
1056 * 'cpoptions' is not included. | |
1057 * Posix doesn't recognize backslash at all. | |
1058 */ | |
1059 if (*regparse == '\\' | |
1060 && !cpo_bsl | |
1061 && regparse + 1 <= endp | |
1062 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL | |
1063 || (!cpo_lit | |
1064 && vim_strchr(REGEXP_ABBR, regparse[1]) | |
1065 != NULL) | |
1066 ) | |
1067 ) | |
1068 { | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1069 mb_ptr_adv(regparse); |
4444 | 1070 |
4507
9dbbddb2ed10
updated for version 7.3.1001
Bram Moolenaar <bram@vim.org>
parents:
4503
diff
changeset
|
1071 if (*regparse == 'n') |
4444 | 1072 startc = reg_string ? NL : NFA_NEWL; |
1073 else | |
1074 if (*regparse == 'd' | |
1075 || *regparse == 'o' | |
1076 || *regparse == 'x' | |
1077 || *regparse == 'u' | |
1078 || *regparse == 'U' | |
1079 ) | |
1080 { | |
1081 /* TODO(RE) This needs more testing */ | |
1082 startc = coll_get_char(); | |
1083 got_coll_char = TRUE; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1084 mb_ptr_back(old_regparse, regparse); |
4444 | 1085 } |
1086 else | |
1087 { | |
1088 /* \r,\t,\e,\b */ | |
1089 startc = backslash_trans(*regparse); | |
1090 } | |
1091 } | |
1092 | |
1093 /* Normal printable char */ | |
1094 if (startc == -1) | |
1095 #ifdef FEAT_MBYTE | |
1096 startc = (*mb_ptr2char)(regparse); | |
1097 #else | |
1098 startc = *regparse; | |
1099 #endif | |
1100 | |
1101 /* Previous char was '-', so this char is end of range. */ | |
1102 if (emit_range) | |
1103 { | |
1104 endc = startc; startc = oldstartc; | |
1105 if (startc > endc) | |
1106 EMSG_RET_FAIL(_(e_invrange)); | |
1107 #ifdef FEAT_MBYTE | |
1108 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
1109 || (*mb_char2len)(endc) > 1)) | |
1110 { | |
1111 if (endc > startc + 256) | |
1112 EMSG_RET_FAIL(_(e_invrange)); | |
1113 /* Emit the range. "startc" was already emitted, so | |
1114 * skip it. */ | |
1115 for (c = startc + 1; c <= endc; c++) | |
1116 { | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1117 EMIT(c); |
4444 | 1118 TRY_NEG(); |
1119 EMIT_GLUE(); | |
1120 } | |
1121 emit_range = FALSE; | |
1122 } | |
1123 else | |
1124 #endif | |
1125 { | |
1126 #ifdef EBCDIC | |
1127 int alpha_only = FALSE; | |
1128 | |
1129 /* for alphabetical range skip the gaps | |
1130 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
1131 if (isalpha(startc) && isalpha(endc)) | |
1132 alpha_only = TRUE; | |
1133 #endif | |
1134 /* Emit the range. "startc" was already emitted, so | |
1135 * skip it. */ | |
1136 for (c = startc + 1; c <= endc; c++) | |
1137 #ifdef EBCDIC | |
1138 if (!alpha_only || isalpha(startc)) | |
1139 #endif | |
1140 { | |
1141 EMIT(c); | |
1142 TRY_NEG(); | |
1143 EMIT_GLUE(); | |
1144 } | |
1145 emit_range = FALSE; | |
1146 } | |
1147 } | |
1148 else | |
1149 { | |
1150 /* | |
1151 * This char (startc) is not part of a range. Just | |
1152 * emit it. | |
1153 * | |
1154 * Normally, simply emit startc. But if we get char | |
1155 * code=0 from a collating char, then replace it with | |
1156 * 0x0a. | |
1157 * | |
1158 * This is needed to completely mimic the behaviour of | |
1159 * the backtracking engine. | |
1160 */ | |
1161 if (got_coll_char == TRUE && startc == 0) | |
1162 EMIT(0x0a); | |
1163 else | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1164 EMIT(startc); |
4444 | 1165 TRY_NEG(); |
1166 EMIT_GLUE(); | |
1167 } | |
1168 | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1169 mb_ptr_adv(regparse); |
4444 | 1170 } /* while (p < endp) */ |
1171 | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1172 mb_ptr_back(old_regparse, regparse); |
4444 | 1173 if (*regparse == '-') /* if last, '-' is just a char */ |
1174 { | |
1175 EMIT('-'); | |
1176 TRY_NEG(); | |
1177 EMIT_GLUE(); | |
1178 } | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1179 mb_ptr_adv(regparse); |
4444 | 1180 |
1181 if (extra == ADD_NL) /* \_[] also matches \n */ | |
1182 { | |
1183 EMIT(reg_string ? NL : NFA_NEWL); | |
1184 TRY_NEG(); | |
1185 EMIT_GLUE(); | |
1186 } | |
1187 | |
1188 /* skip the trailing ] */ | |
1189 regparse = endp; | |
4573
37e444aeb33b
updated for version 7.3.1034
Bram Moolenaar <bram@vim.org>
parents:
4571
diff
changeset
|
1190 mb_ptr_adv(regparse); |
4444 | 1191 if (negated == TRUE) |
1192 { | |
1193 /* Mark end of negated char range */ | |
1194 EMIT(NFA_END_NEG_RANGE); | |
1195 EMIT(NFA_CONCAT); | |
1196 } | |
1197 return OK; | |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1198 } /* if exists closing ] */ |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1199 |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1200 if (reg_strict) |
4444 | 1201 { |
1202 syntax_error = TRUE; | |
1203 EMSG_RET_FAIL(_(e_missingbracket)); | |
1204 } | |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1205 /* FALLTHROUGH */ |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1206 |
4444 | 1207 default: |
1208 { | |
1209 #ifdef FEAT_MBYTE | |
1210 int plen; | |
1211 | |
1212 nfa_do_multibyte: | |
4543
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1213 /* plen is length of current char with composing chars */ |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1214 if (enc_utf8 && ((*mb_char2len)(c) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1215 != (plen = (*mb_ptr2len)(old_regparse)) |
08ac46980953
updated for version 7.3.1019
Bram Moolenaar <bram@vim.org>
parents:
4541
diff
changeset
|
1216 || utf_iscomposing(c))) |
4444 | 1217 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1218 int i = 0; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
1219 |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
1220 /* A base character plus composing characters, or just one |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
1221 * or more composing characters. |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1222 * This requires creating a separate atom as if enclosing |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1223 * the characters in (), where NFA_COMPOSING is the ( and |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1224 * NFA_END_COMPOSING is the ). Note that right now we are |
4444 | 1225 * building the postfix form, not the NFA itself; |
1226 * a composing char could be: a, b, c, NFA_COMPOSING | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1227 * where 'b' and 'c' are chars with codes > 256. */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1228 for (;;) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1229 { |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1230 EMIT(c); |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1231 if (i > 0) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1232 EMIT(NFA_CONCAT); |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
1233 if ((i += utf_char2len(c)) >= plen) |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1234 break; |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1235 c = utf_ptr2char(old_regparse + i); |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1236 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
1237 EMIT(NFA_COMPOSING); |
4444 | 1238 regparse = old_regparse + plen; |
1239 } | |
1240 else | |
1241 #endif | |
1242 { | |
1243 c = no_Magic(c); | |
1244 EMIT(c); | |
1245 } | |
1246 return OK; | |
1247 } | |
1248 } | |
1249 | |
1250 #undef TRY_NEG | |
1251 #undef EMIT_GLUE | |
1252 | |
1253 return OK; | |
1254 } | |
1255 | |
1256 /* | |
1257 * Parse something followed by possible [*+=]. | |
1258 * | |
1259 * A piece is an atom, possibly followed by a multi, an indication of how many | |
1260 * times the atom can be matched. Example: "a*" matches any sequence of "a" | |
1261 * characters: "", "a", "aa", etc. | |
1262 * | |
1263 * piece ::= atom | |
1264 * or atom multi | |
1265 */ | |
1266 static int | |
1267 nfa_regpiece() | |
1268 { | |
1269 int i; | |
1270 int op; | |
1271 int ret; | |
1272 long minval, maxval; | |
1273 int greedy = TRUE; /* Braces are prefixed with '-' ? */ | |
1274 char_u *old_regparse, *new_regparse; | |
1275 int c2; | |
1276 int *old_post_ptr, *my_post_start; | |
1277 int old_regnpar; | |
1278 int quest; | |
1279 | |
1280 /* Save the current position in the regexp, so that we can use it if | |
1281 * <atom>{m,n} is next. */ | |
1282 old_regparse = regparse; | |
1283 /* Save current number of open parenthesis, so we can use it if | |
1284 * <atom>{m,n} is next */ | |
1285 old_regnpar = regnpar; | |
1286 /* store current pos in the postfix form, for \{m,n} involving 0s */ | |
1287 my_post_start = post_ptr; | |
1288 | |
1289 ret = nfa_regatom(); | |
1290 if (ret == FAIL) | |
1291 return FAIL; /* cascaded error */ | |
1292 | |
1293 op = peekchr(); | |
1294 if (re_multi_type(op) == NOT_MULTI) | |
1295 return OK; | |
1296 | |
1297 skipchr(); | |
1298 switch (op) | |
1299 { | |
1300 case Magic('*'): | |
1301 EMIT(NFA_STAR); | |
1302 break; | |
1303 | |
1304 case Magic('+'): | |
1305 /* | |
1306 * Trick: Normally, (a*)\+ would match the whole input "aaa". The | |
1307 * first and only submatch would be "aaa". But the backtracking | |
1308 * engine interprets the plus as "try matching one more time", and | |
1309 * a* matches a second time at the end of the input, the empty | |
1310 * string. | |
1311 * The submatch will the empty string. | |
1312 * | |
1313 * In order to be consistent with the old engine, we disable | |
1314 * NFA_PLUS, and replace <atom>+ with <atom><atom>* | |
1315 */ | |
1316 /* EMIT(NFA_PLUS); */ | |
1317 regnpar = old_regnpar; | |
1318 regparse = old_regparse; | |
1319 curchr = -1; | |
1320 if (nfa_regatom() == FAIL) | |
1321 return FAIL; | |
1322 EMIT(NFA_STAR); | |
1323 EMIT(NFA_CONCAT); | |
1324 skipchr(); /* skip the \+ */ | |
1325 break; | |
1326 | |
1327 case Magic('@'): | |
1328 op = no_Magic(getchr()); | |
1329 switch(op) | |
1330 { | |
1331 case '=': | |
1332 EMIT(NFA_PREV_ATOM_NO_WIDTH); | |
1333 break; | |
1334 case '!': | |
1335 case '<': | |
1336 case '>': | |
1337 /* Not supported yet */ | |
1338 return FAIL; | |
1339 default: | |
1340 syntax_error = TRUE; | |
4450 | 1341 EMSGN(_("E869: (NFA) Unknown operator '\\@%c'"), op); |
4444 | 1342 return FAIL; |
1343 } | |
1344 break; | |
1345 | |
1346 case Magic('?'): | |
1347 case Magic('='): | |
1348 EMIT(NFA_QUEST); | |
1349 break; | |
1350 | |
1351 case Magic('{'): | |
1352 /* a{2,5} will expand to 'aaa?a?a?' | |
1353 * a{-1,3} will expand to 'aa??a??', where ?? is the nongreedy | |
1354 * version of '?' | |
1355 * \v(ab){2,3} will expand to '(ab)(ab)(ab)?', where all the | |
1356 * parenthesis have the same id | |
1357 */ | |
1358 | |
1359 greedy = TRUE; | |
1360 c2 = peekchr(); | |
1361 if (c2 == '-' || c2 == Magic('-')) | |
1362 { | |
1363 skipchr(); | |
1364 greedy = FALSE; | |
1365 } | |
1366 if (!read_limits(&minval, &maxval)) | |
1367 { | |
1368 syntax_error = TRUE; | |
1369 EMSG_RET_FAIL(_("E870: (NFA regexp) Error reading repetition limits")); | |
1370 } | |
1371 /* <atom>{0,inf}, <atom>{0,} and <atom>{} are equivalent to | |
1372 * <atom>* */ | |
1373 if (minval == 0 && maxval == MAX_LIMIT && greedy) | |
1374 { | |
1375 EMIT(NFA_STAR); | |
1376 break; | |
1377 } | |
1378 | |
1379 if (maxval > NFA_BRACES_MAXLIMIT) | |
1380 { | |
1381 /* This would yield a huge automaton and use too much memory. | |
1382 * Revert to old engine */ | |
1383 return FAIL; | |
1384 } | |
1385 | |
1386 /* Special case: x{0} or x{-0} */ | |
1387 if (maxval == 0) | |
1388 { | |
1389 /* Ignore result of previous call to nfa_regatom() */ | |
1390 post_ptr = my_post_start; | |
1391 /* NFA_SKIP_CHAR has 0-length and works everywhere */ | |
1392 EMIT(NFA_SKIP_CHAR); | |
1393 return OK; | |
1394 } | |
1395 | |
1396 /* Ignore previous call to nfa_regatom() */ | |
1397 post_ptr = my_post_start; | |
1398 /* Save pos after the repeated atom and the \{} */ | |
1399 new_regparse = regparse; | |
1400 | |
1401 quest = (greedy == TRUE? NFA_QUEST : NFA_QUEST_NONGREEDY); | |
1402 for (i = 0; i < maxval; i++) | |
1403 { | |
1404 /* Goto beginning of the repeated atom */ | |
1405 regparse = old_regparse; | |
1406 curchr = -1; | |
1407 /* Restore count of parenthesis */ | |
1408 regnpar = old_regnpar; | |
1409 old_post_ptr = post_ptr; | |
1410 if (nfa_regatom() == FAIL) | |
1411 return FAIL; | |
1412 /* after "minval" times, atoms are optional */ | |
1413 if (i + 1 > minval) | |
1414 EMIT(quest); | |
1415 if (old_post_ptr != my_post_start) | |
1416 EMIT(NFA_CONCAT); | |
1417 } | |
1418 | |
1419 /* Go to just after the repeated atom and the \{} */ | |
1420 regparse = new_regparse; | |
1421 curchr = -1; | |
1422 | |
1423 break; | |
1424 | |
1425 | |
1426 default: | |
1427 break; | |
1428 } /* end switch */ | |
1429 | |
1430 if (re_multi_type(peekchr()) != NOT_MULTI) | |
1431 { | |
1432 /* Can't have a multi follow a multi. */ | |
1433 syntax_error = TRUE; | |
1434 EMSG_RET_FAIL(_("E871: (NFA regexp) Can't have a multi follow a multi !")); | |
1435 } | |
1436 | |
1437 return OK; | |
1438 } | |
1439 | |
1440 /* | |
1441 * Parse one or more pieces, concatenated. It matches a match for the | |
1442 * first piece, followed by a match for the second piece, etc. Example: | |
1443 * "f[0-9]b", first matches "f", then a digit and then "b". | |
1444 * | |
1445 * concat ::= piece | |
1446 * or piece piece | |
1447 * or piece piece piece | |
1448 * etc. | |
1449 */ | |
1450 static int | |
1451 nfa_regconcat() | |
1452 { | |
1453 int cont = TRUE; | |
1454 int first = TRUE; | |
1455 | |
1456 while (cont) | |
1457 { | |
1458 switch (peekchr()) | |
1459 { | |
1460 case NUL: | |
1461 case Magic('|'): | |
1462 case Magic('&'): | |
1463 case Magic(')'): | |
1464 cont = FALSE; | |
1465 break; | |
1466 | |
1467 case Magic('Z'): | |
1468 #ifdef FEAT_MBYTE | |
1469 regflags |= RF_ICOMBINE; | |
1470 #endif | |
1471 skipchr_keepstart(); | |
1472 break; | |
1473 case Magic('c'): | |
1474 regflags |= RF_ICASE; | |
1475 skipchr_keepstart(); | |
1476 break; | |
1477 case Magic('C'): | |
1478 regflags |= RF_NOICASE; | |
1479 skipchr_keepstart(); | |
1480 break; | |
1481 case Magic('v'): | |
1482 reg_magic = MAGIC_ALL; | |
1483 skipchr_keepstart(); | |
1484 curchr = -1; | |
1485 break; | |
1486 case Magic('m'): | |
1487 reg_magic = MAGIC_ON; | |
1488 skipchr_keepstart(); | |
1489 curchr = -1; | |
1490 break; | |
1491 case Magic('M'): | |
1492 reg_magic = MAGIC_OFF; | |
1493 skipchr_keepstart(); | |
1494 curchr = -1; | |
1495 break; | |
1496 case Magic('V'): | |
1497 reg_magic = MAGIC_NONE; | |
1498 skipchr_keepstart(); | |
1499 curchr = -1; | |
1500 break; | |
1501 | |
1502 default: | |
1503 if (nfa_regpiece() == FAIL) | |
1504 return FAIL; | |
1505 if (first == FALSE) | |
1506 EMIT(NFA_CONCAT); | |
1507 else | |
1508 first = FALSE; | |
1509 break; | |
1510 } | |
1511 } | |
1512 | |
1513 return OK; | |
1514 } | |
1515 | |
1516 /* | |
1517 * Parse a branch, one or more concats, separated by "\&". It matches the | |
1518 * last concat, but only if all the preceding concats also match at the same | |
1519 * position. Examples: | |
1520 * "foobeep\&..." matches "foo" in "foobeep". | |
1521 * ".*Peter\&.*Bob" matches in a line containing both "Peter" and "Bob" | |
1522 * | |
1523 * branch ::= concat | |
1524 * or concat \& concat | |
1525 * or concat \& concat \& concat | |
1526 * etc. | |
1527 */ | |
1528 static int | |
1529 nfa_regbranch() | |
1530 { | |
1531 int ch; | |
1532 int *old_post_ptr; | |
1533 | |
1534 old_post_ptr = post_ptr; | |
1535 | |
1536 /* First branch, possibly the only one */ | |
1537 if (nfa_regconcat() == FAIL) | |
1538 return FAIL; | |
1539 | |
1540 ch = peekchr(); | |
1541 /* Try next concats */ | |
1542 while (ch == Magic('&')) | |
1543 { | |
1544 skipchr(); | |
1545 EMIT(NFA_NOPEN); | |
1546 EMIT(NFA_PREV_ATOM_NO_WIDTH); | |
1547 old_post_ptr = post_ptr; | |
1548 if (nfa_regconcat() == FAIL) | |
1549 return FAIL; | |
1550 /* if concat is empty, skip a input char. But do emit a node */ | |
1551 if (old_post_ptr == post_ptr) | |
1552 EMIT(NFA_SKIP_CHAR); | |
1553 EMIT(NFA_CONCAT); | |
1554 ch = peekchr(); | |
1555 } | |
1556 | |
1557 /* Even if a branch is empty, emit one node for it */ | |
1558 if (old_post_ptr == post_ptr) | |
1559 EMIT(NFA_SKIP_CHAR); | |
1560 | |
1561 return OK; | |
1562 } | |
1563 | |
1564 /* | |
1565 * Parse a pattern, one or more branches, separated by "\|". It matches | |
1566 * anything that matches one of the branches. Example: "foo\|beep" matches | |
1567 * "foo" and matches "beep". If more than one branch matches, the first one | |
1568 * is used. | |
1569 * | |
1570 * pattern ::= branch | |
1571 * or branch \| branch | |
1572 * or branch \| branch \| branch | |
1573 * etc. | |
1574 */ | |
1575 static int | |
1576 nfa_reg(paren) | |
1577 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
1578 { | |
1579 int parno = 0; | |
1580 | |
1581 #ifdef FEAT_SYN_HL | |
1582 #endif | |
1583 if (paren == REG_PAREN) | |
1584 { | |
1585 if (regnpar >= NSUBEXP) /* Too many `(' */ | |
1586 { | |
1587 syntax_error = TRUE; | |
1588 EMSG_RET_FAIL(_("E872: (NFA regexp) Too many '('")); | |
1589 } | |
1590 parno = regnpar++; | |
1591 } | |
1592 | |
1593 if (nfa_regbranch() == FAIL) | |
1594 return FAIL; /* cascaded error */ | |
1595 | |
1596 while (peekchr() == Magic('|')) | |
1597 { | |
1598 skipchr(); | |
1599 if (nfa_regbranch() == FAIL) | |
1600 return FAIL; /* cascaded error */ | |
1601 EMIT(NFA_OR); | |
1602 } | |
1603 | |
1604 /* Check for proper termination. */ | |
1605 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
1606 { | |
1607 syntax_error = TRUE; | |
1608 if (paren == REG_NPAREN) | |
1609 EMSG2_RET_FAIL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); | |
1610 else | |
1611 EMSG2_RET_FAIL(_(e_unmatchedp), reg_magic == MAGIC_ALL); | |
1612 } | |
1613 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
1614 { | |
1615 syntax_error = TRUE; | |
1616 if (peekchr() == Magic(')')) | |
1617 EMSG2_RET_FAIL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); | |
1618 else | |
1619 EMSG_RET_FAIL(_("E873: (NFA regexp) proper termination error")); | |
1620 } | |
1621 /* | |
1622 * Here we set the flag allowing back references to this set of | |
1623 * parentheses. | |
1624 */ | |
1625 if (paren == REG_PAREN) | |
1626 { | |
1627 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
1628 EMIT(NFA_MOPEN + parno); | |
1629 } | |
1630 | |
1631 return OK; | |
1632 } | |
1633 | |
1634 #ifdef DEBUG | |
1635 static char_u code[50]; | |
1636 | |
1637 static void | |
1638 nfa_set_code(c) | |
1639 int c; | |
1640 { | |
1641 int addnl = FALSE; | |
1642 | |
1643 if (c >= NFA_FIRST_NL && c <= NFA_LAST_NL) | |
1644 { | |
1645 addnl = TRUE; | |
1646 c -= ADD_NL; | |
1647 } | |
1648 | |
1649 STRCPY(code, ""); | |
1650 switch (c) | |
1651 { | |
1652 case NFA_MATCH: STRCPY(code, "NFA_MATCH "); break; | |
1653 case NFA_SPLIT: STRCPY(code, "NFA_SPLIT "); break; | |
1654 case NFA_CONCAT: STRCPY(code, "NFA_CONCAT "); break; | |
1655 case NFA_NEWL: STRCPY(code, "NFA_NEWL "); break; | |
1656 case NFA_ZSTART: STRCPY(code, "NFA_ZSTART"); break; | |
1657 case NFA_ZEND: STRCPY(code, "NFA_ZEND"); break; | |
1658 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1659 case NFA_BACKREF1: STRCPY(code, "NFA_BACKREF1"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1660 case NFA_BACKREF2: STRCPY(code, "NFA_BACKREF2"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1661 case NFA_BACKREF3: STRCPY(code, "NFA_BACKREF3"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1662 case NFA_BACKREF4: STRCPY(code, "NFA_BACKREF4"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1663 case NFA_BACKREF5: STRCPY(code, "NFA_BACKREF5"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1664 case NFA_BACKREF6: STRCPY(code, "NFA_BACKREF6"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1665 case NFA_BACKREF7: STRCPY(code, "NFA_BACKREF7"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1666 case NFA_BACKREF8: STRCPY(code, "NFA_BACKREF8"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1667 case NFA_BACKREF9: STRCPY(code, "NFA_BACKREF9"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1668 case NFA_SKIP: STRCPY(code, "NFA_SKIP"); break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
1669 |
4444 | 1670 case NFA_PREV_ATOM_NO_WIDTH: |
1671 STRCPY(code, "NFA_PREV_ATOM_NO_WIDTH"); break; | |
1672 case NFA_NOPEN: STRCPY(code, "NFA_MOPEN_INVISIBLE"); break; | |
1673 case NFA_NCLOSE: STRCPY(code, "NFA_MCLOSE_INVISIBLE"); break; | |
1674 case NFA_START_INVISIBLE: STRCPY(code, "NFA_START_INVISIBLE"); break; | |
1675 case NFA_END_INVISIBLE: STRCPY(code, "NFA_END_INVISIBLE"); break; | |
1676 | |
1677 case NFA_COMPOSING: STRCPY(code, "NFA_COMPOSING"); break; | |
1678 case NFA_END_COMPOSING: STRCPY(code, "NFA_END_COMPOSING"); break; | |
1679 | |
1680 case NFA_MOPEN + 0: | |
1681 case NFA_MOPEN + 1: | |
1682 case NFA_MOPEN + 2: | |
1683 case NFA_MOPEN + 3: | |
1684 case NFA_MOPEN + 4: | |
1685 case NFA_MOPEN + 5: | |
1686 case NFA_MOPEN + 6: | |
1687 case NFA_MOPEN + 7: | |
1688 case NFA_MOPEN + 8: | |
1689 case NFA_MOPEN + 9: | |
1690 STRCPY(code, "NFA_MOPEN(x)"); | |
1691 code[10] = c - NFA_MOPEN + '0'; | |
1692 break; | |
1693 case NFA_MCLOSE + 0: | |
1694 case NFA_MCLOSE + 1: | |
1695 case NFA_MCLOSE + 2: | |
1696 case NFA_MCLOSE + 3: | |
1697 case NFA_MCLOSE + 4: | |
1698 case NFA_MCLOSE + 5: | |
1699 case NFA_MCLOSE + 6: | |
1700 case NFA_MCLOSE + 7: | |
1701 case NFA_MCLOSE + 8: | |
1702 case NFA_MCLOSE + 9: | |
1703 STRCPY(code, "NFA_MCLOSE(x)"); | |
1704 code[11] = c - NFA_MCLOSE + '0'; | |
1705 break; | |
1706 case NFA_EOL: STRCPY(code, "NFA_EOL "); break; | |
1707 case NFA_BOL: STRCPY(code, "NFA_BOL "); break; | |
1708 case NFA_EOW: STRCPY(code, "NFA_EOW "); break; | |
1709 case NFA_BOW: STRCPY(code, "NFA_BOW "); break; | |
1710 case NFA_STAR: STRCPY(code, "NFA_STAR "); break; | |
1711 case NFA_PLUS: STRCPY(code, "NFA_PLUS "); break; | |
1712 case NFA_NOT: STRCPY(code, "NFA_NOT "); break; | |
1713 case NFA_SKIP_CHAR: STRCPY(code, "NFA_SKIP_CHAR"); break; | |
1714 case NFA_OR: STRCPY(code, "NFA_OR"); break; | |
1715 case NFA_QUEST: STRCPY(code, "NFA_QUEST"); break; | |
1716 case NFA_QUEST_NONGREEDY: STRCPY(code, "NFA_QUEST_NON_GREEDY"); break; | |
1717 case NFA_END_NEG_RANGE: STRCPY(code, "NFA_END_NEG_RANGE"); break; | |
1718 case NFA_CLASS_ALNUM: STRCPY(code, "NFA_CLASS_ALNUM"); break; | |
1719 case NFA_CLASS_ALPHA: STRCPY(code, "NFA_CLASS_ALPHA"); break; | |
1720 case NFA_CLASS_BLANK: STRCPY(code, "NFA_CLASS_BLANK"); break; | |
1721 case NFA_CLASS_CNTRL: STRCPY(code, "NFA_CLASS_CNTRL"); break; | |
1722 case NFA_CLASS_DIGIT: STRCPY(code, "NFA_CLASS_DIGIT"); break; | |
1723 case NFA_CLASS_GRAPH: STRCPY(code, "NFA_CLASS_GRAPH"); break; | |
1724 case NFA_CLASS_LOWER: STRCPY(code, "NFA_CLASS_LOWER"); break; | |
1725 case NFA_CLASS_PRINT: STRCPY(code, "NFA_CLASS_PRINT"); break; | |
1726 case NFA_CLASS_PUNCT: STRCPY(code, "NFA_CLASS_PUNCT"); break; | |
1727 case NFA_CLASS_SPACE: STRCPY(code, "NFA_CLASS_SPACE"); break; | |
1728 case NFA_CLASS_UPPER: STRCPY(code, "NFA_CLASS_UPPER"); break; | |
1729 case NFA_CLASS_XDIGIT: STRCPY(code, "NFA_CLASS_XDIGIT"); break; | |
1730 case NFA_CLASS_TAB: STRCPY(code, "NFA_CLASS_TAB"); break; | |
1731 case NFA_CLASS_RETURN: STRCPY(code, "NFA_CLASS_RETURN"); break; | |
1732 case NFA_CLASS_BACKSPACE: STRCPY(code, "NFA_CLASS_BACKSPACE"); break; | |
1733 case NFA_CLASS_ESCAPE: STRCPY(code, "NFA_CLASS_ESCAPE"); break; | |
1734 | |
1735 case NFA_ANY: STRCPY(code, "NFA_ANY"); break; | |
1736 case NFA_IDENT: STRCPY(code, "NFA_IDENT"); break; | |
1737 case NFA_SIDENT:STRCPY(code, "NFA_SIDENT"); break; | |
1738 case NFA_KWORD: STRCPY(code, "NFA_KWORD"); break; | |
1739 case NFA_SKWORD:STRCPY(code, "NFA_SKWORD"); break; | |
1740 case NFA_FNAME: STRCPY(code, "NFA_FNAME"); break; | |
1741 case NFA_SFNAME:STRCPY(code, "NFA_SFNAME"); break; | |
1742 case NFA_PRINT: STRCPY(code, "NFA_PRINT"); break; | |
1743 case NFA_SPRINT:STRCPY(code, "NFA_SPRINT"); break; | |
1744 case NFA_WHITE: STRCPY(code, "NFA_WHITE"); break; | |
1745 case NFA_NWHITE:STRCPY(code, "NFA_NWHITE"); break; | |
1746 case NFA_DIGIT: STRCPY(code, "NFA_DIGIT"); break; | |
1747 case NFA_NDIGIT:STRCPY(code, "NFA_NDIGIT"); break; | |
1748 case NFA_HEX: STRCPY(code, "NFA_HEX"); break; | |
1749 case NFA_NHEX: STRCPY(code, "NFA_NHEX"); break; | |
1750 case NFA_OCTAL: STRCPY(code, "NFA_OCTAL"); break; | |
1751 case NFA_NOCTAL:STRCPY(code, "NFA_NOCTAL"); break; | |
1752 case NFA_WORD: STRCPY(code, "NFA_WORD"); break; | |
1753 case NFA_NWORD: STRCPY(code, "NFA_NWORD"); break; | |
1754 case NFA_HEAD: STRCPY(code, "NFA_HEAD"); break; | |
1755 case NFA_NHEAD: STRCPY(code, "NFA_NHEAD"); break; | |
1756 case NFA_ALPHA: STRCPY(code, "NFA_ALPHA"); break; | |
1757 case NFA_NALPHA:STRCPY(code, "NFA_NALPHA"); break; | |
1758 case NFA_LOWER: STRCPY(code, "NFA_LOWER"); break; | |
1759 case NFA_NLOWER:STRCPY(code, "NFA_NLOWER"); break; | |
1760 case NFA_UPPER: STRCPY(code, "NFA_UPPER"); break; | |
1761 case NFA_NUPPER:STRCPY(code, "NFA_NUPPER"); break; | |
1762 | |
1763 default: | |
1764 STRCPY(code, "CHAR(x)"); | |
1765 code[5] = c; | |
1766 } | |
1767 | |
1768 if (addnl == TRUE) | |
1769 STRCAT(code, " + NEWLINE "); | |
1770 | |
1771 } | |
1772 | |
1773 #ifdef ENABLE_LOG | |
1774 static FILE *log_fd; | |
1775 | |
1776 /* | |
1777 * Print the postfix notation of the current regexp. | |
1778 */ | |
1779 static void | |
1780 nfa_postfix_dump(expr, retval) | |
1781 char_u *expr; | |
1782 int retval; | |
1783 { | |
1784 int *p; | |
1785 FILE *f; | |
1786 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1787 f = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
4444 | 1788 if (f != NULL) |
1789 { | |
1790 fprintf(f, "\n-------------------------\n"); | |
1791 if (retval == FAIL) | |
1792 fprintf(f, ">>> NFA engine failed ... \n"); | |
1793 else if (retval == OK) | |
1794 fprintf(f, ">>> NFA engine succeeded !\n"); | |
1795 fprintf(f, "Regexp: \"%s\"\nPostfix notation (char): \"", expr); | |
4464 | 1796 for (p = post_start; *p && p < post_end; p++) |
4444 | 1797 { |
1798 nfa_set_code(*p); | |
1799 fprintf(f, "%s, ", code); | |
1800 } | |
1801 fprintf(f, "\"\nPostfix notation (int): "); | |
4464 | 1802 for (p = post_start; *p && p < post_end; p++) |
4444 | 1803 fprintf(f, "%d ", *p); |
1804 fprintf(f, "\n\n"); | |
1805 fclose(f); | |
1806 } | |
1807 } | |
1808 | |
1809 /* | |
1810 * Print the NFA starting with a root node "state". | |
1811 */ | |
1812 static void | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1813 nfa_print_state(debugf, state) |
4444 | 1814 FILE *debugf; |
1815 nfa_state_T *state; | |
1816 { | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1817 garray_T indent; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1818 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1819 ga_init2(&indent, 1, 64); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1820 ga_append(&indent, '\0'); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1821 nfa_print_state2(debugf, state, &indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1822 ga_clear(&indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1823 } |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1824 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1825 static void |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1826 nfa_print_state2(debugf, state, indent) |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1827 FILE *debugf; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1828 nfa_state_T *state; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1829 garray_T *indent; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1830 { |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1831 char_u *p; |
4444 | 1832 |
1833 if (state == NULL) | |
1834 return; | |
1835 | |
1836 fprintf(debugf, "(%2d)", abs(state->id)); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1837 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1838 /* Output indent */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1839 p = (char_u *)indent->ga_data; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1840 if (indent->ga_len >= 3) |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1841 { |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1842 int last = indent->ga_len - 3; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1843 char_u save[2]; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1844 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1845 STRNCPY(save, &p[last], 2); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1846 STRNCPY(&p[last], "+-", 2); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1847 fprintf(debugf, " %s", p); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1848 STRNCPY(&p[last], save, 2); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1849 } |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1850 else |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1851 fprintf(debugf, " %s", p); |
4444 | 1852 |
1853 nfa_set_code(state->c); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1854 fprintf(debugf, "%s%s (%d) (id=%d)\n", |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1855 state->negated ? "NOT " : "", code, state->c, abs(state->id)); |
4444 | 1856 if (state->id < 0) |
1857 return; | |
1858 | |
1859 state->id = abs(state->id) * -1; | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1860 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1861 /* grow indent for state->out */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1862 indent->ga_len -= 1; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1863 if (state->out1) |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1864 ga_concat(indent, (char_u *)"| "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1865 else |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1866 ga_concat(indent, (char_u *)" "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1867 ga_append(indent, '\0'); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1868 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1869 nfa_print_state2(debugf, state->out, indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1870 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1871 /* replace last part of indent for state->out1 */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1872 indent->ga_len -= 3; |
4537
5cc98a5898cf
updated for version 7.3.1016
Bram Moolenaar <bram@vim.org>
parents:
4535
diff
changeset
|
1873 ga_concat(indent, (char_u *)" "); |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1874 ga_append(indent, '\0'); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1875 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1876 nfa_print_state2(debugf, state->out1, indent); |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1877 |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1878 /* shrink indent */ |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1879 indent->ga_len -= 3; |
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1880 ga_append(indent, '\0'); |
4444 | 1881 } |
1882 | |
1883 /* | |
1884 * Print the NFA state machine. | |
1885 */ | |
1886 static void | |
1887 nfa_dump(prog) | |
1888 nfa_regprog_T *prog; | |
1889 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
1890 FILE *debugf = fopen(NFA_REGEXP_DUMP_LOG, "a"); |
4444 | 1891 |
1892 if (debugf != NULL) | |
1893 { | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
1894 nfa_print_state(debugf, prog->start); |
4444 | 1895 fclose(debugf); |
1896 } | |
1897 } | |
1898 #endif /* ENABLE_LOG */ | |
1899 #endif /* DEBUG */ | |
1900 | |
1901 /* | |
1902 * Parse r.e. @expr and convert it into postfix form. | |
1903 * Return the postfix string on success, NULL otherwise. | |
1904 */ | |
1905 static int * | |
1906 re2post() | |
1907 { | |
1908 if (nfa_reg(REG_NOPAREN) == FAIL) | |
1909 return NULL; | |
1910 EMIT(NFA_MOPEN); | |
1911 return post_start; | |
1912 } | |
1913 | |
1914 /* NB. Some of the code below is inspired by Russ's. */ | |
1915 | |
1916 /* | |
1917 * Represents an NFA state plus zero or one or two arrows exiting. | |
1918 * if c == MATCH, no arrows out; matching state. | |
1919 * If c == SPLIT, unlabeled arrows to out and out1 (if != NULL). | |
1920 * If c < 256, labeled arrow with character c to out. | |
1921 */ | |
1922 | |
1923 static nfa_state_T *state_ptr; /* points to nfa_prog->state */ | |
1924 | |
1925 /* | |
1926 * Allocate and initialize nfa_state_T. | |
1927 */ | |
1928 static nfa_state_T * | |
1929 new_state(c, out, out1) | |
1930 int c; | |
1931 nfa_state_T *out; | |
1932 nfa_state_T *out1; | |
1933 { | |
1934 nfa_state_T *s; | |
1935 | |
1936 if (istate >= nstate) | |
1937 return NULL; | |
1938 | |
1939 s = &state_ptr[istate++]; | |
1940 | |
1941 s->c = c; | |
1942 s->out = out; | |
1943 s->out1 = out1; | |
1944 | |
1945 s->id = istate; | |
1946 s->lastlist = 0; | |
1947 s->negated = FALSE; | |
1948 | |
1949 return s; | |
1950 } | |
1951 | |
1952 /* | |
1953 * A partially built NFA without the matching state filled in. | |
1954 * Frag_T.start points at the start state. | |
1955 * Frag_T.out is a list of places that need to be set to the | |
1956 * next state for this fragment. | |
1957 */ | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1958 |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1959 /* Since the out pointers in the list are always |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1960 * uninitialized, we use the pointers themselves |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1961 * as storage for the Ptrlists. */ |
4444 | 1962 typedef union Ptrlist Ptrlist; |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1963 union Ptrlist |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1964 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1965 Ptrlist *next; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1966 nfa_state_T *s; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1967 }; |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1968 |
4444 | 1969 struct Frag |
1970 { | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
1971 nfa_state_T *start; |
4444 | 1972 Ptrlist *out; |
1973 }; | |
1974 typedef struct Frag Frag_T; | |
1975 | |
1976 static Frag_T frag __ARGS((nfa_state_T *start, Ptrlist *out)); | |
1977 static Ptrlist *list1 __ARGS((nfa_state_T **outp)); | |
1978 static void patch __ARGS((Ptrlist *l, nfa_state_T *s)); | |
1979 static Ptrlist *append __ARGS((Ptrlist *l1, Ptrlist *l2)); | |
1980 static void st_push __ARGS((Frag_T s, Frag_T **p, Frag_T *stack_end)); | |
1981 static Frag_T st_pop __ARGS((Frag_T **p, Frag_T *stack)); | |
1982 | |
1983 /* | |
4456 | 1984 * Initialize a Frag_T struct and return it. |
4444 | 1985 */ |
1986 static Frag_T | |
1987 frag(start, out) | |
1988 nfa_state_T *start; | |
1989 Ptrlist *out; | |
1990 { | |
4456 | 1991 Frag_T n; |
1992 | |
1993 n.start = start; | |
1994 n.out = out; | |
4444 | 1995 return n; |
1996 } | |
1997 | |
1998 /* | |
1999 * Create singleton list containing just outp. | |
2000 */ | |
2001 static Ptrlist * | |
2002 list1(outp) | |
2003 nfa_state_T **outp; | |
2004 { | |
2005 Ptrlist *l; | |
2006 | |
2007 l = (Ptrlist *)outp; | |
2008 l->next = NULL; | |
2009 return l; | |
2010 } | |
2011 | |
2012 /* | |
2013 * Patch the list of states at out to point to start. | |
2014 */ | |
2015 static void | |
2016 patch(l, s) | |
2017 Ptrlist *l; | |
2018 nfa_state_T *s; | |
2019 { | |
2020 Ptrlist *next; | |
2021 | |
2022 for (; l; l = next) | |
2023 { | |
2024 next = l->next; | |
2025 l->s = s; | |
2026 } | |
2027 } | |
2028 | |
2029 | |
2030 /* | |
2031 * Join the two lists l1 and l2, returning the combination. | |
2032 */ | |
2033 static Ptrlist * | |
2034 append(l1, l2) | |
2035 Ptrlist *l1; | |
2036 Ptrlist *l2; | |
2037 { | |
2038 Ptrlist *oldl1; | |
2039 | |
2040 oldl1 = l1; | |
2041 while (l1->next) | |
2042 l1 = l1->next; | |
2043 l1->next = l2; | |
2044 return oldl1; | |
2045 } | |
2046 | |
2047 /* | |
2048 * Stack used for transforming postfix form into NFA. | |
2049 */ | |
2050 static Frag_T empty; | |
2051 | |
2052 static void | |
2053 st_error(postfix, end, p) | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2054 int *postfix UNUSED; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2055 int *end UNUSED; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2056 int *p UNUSED; |
4444 | 2057 { |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2058 #ifdef NFA_REGEXP_ERROR_LOG |
4444 | 2059 FILE *df; |
2060 int *p2; | |
2061 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2062 df = fopen(NFA_REGEXP_ERROR_LOG, "a"); |
4444 | 2063 if (df) |
2064 { | |
2065 fprintf(df, "Error popping the stack!\n"); | |
2066 #ifdef DEBUG | |
2067 fprintf(df, "Current regexp is \"%s\"\n", nfa_regengine.expr); | |
2068 #endif | |
2069 fprintf(df, "Postfix form is: "); | |
2070 #ifdef DEBUG | |
2071 for (p2 = postfix; p2 < end; p2++) | |
2072 { | |
2073 nfa_set_code(*p2); | |
2074 fprintf(df, "%s, ", code); | |
2075 } | |
2076 nfa_set_code(*p); | |
2077 fprintf(df, "\nCurrent position is: "); | |
2078 for (p2 = postfix; p2 <= p; p2 ++) | |
2079 { | |
2080 nfa_set_code(*p2); | |
2081 fprintf(df, "%s, ", code); | |
2082 } | |
2083 #else | |
2084 for (p2 = postfix; p2 < end; p2++) | |
2085 { | |
2086 fprintf(df, "%d, ", *p2); | |
2087 } | |
2088 fprintf(df, "\nCurrent position is: "); | |
2089 for (p2 = postfix; p2 <= p; p2 ++) | |
2090 { | |
2091 fprintf(df, "%d, ", *p2); | |
2092 } | |
2093 #endif | |
2094 fprintf(df, "\n--------------------------\n"); | |
2095 fclose(df); | |
2096 } | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2097 #endif |
4444 | 2098 EMSG(_("E874: (NFA) Could not pop the stack !")); |
2099 } | |
2100 | |
2101 /* | |
2102 * Push an item onto the stack. | |
2103 */ | |
2104 static void | |
2105 st_push(s, p, stack_end) | |
2106 Frag_T s; | |
2107 Frag_T **p; | |
2108 Frag_T *stack_end; | |
2109 { | |
2110 Frag_T *stackp = *p; | |
2111 | |
2112 if (stackp >= stack_end) | |
2113 return; | |
2114 *stackp = s; | |
2115 *p = *p + 1; | |
2116 } | |
2117 | |
2118 /* | |
2119 * Pop an item from the stack. | |
2120 */ | |
2121 static Frag_T | |
2122 st_pop(p, stack) | |
2123 Frag_T **p; | |
2124 Frag_T *stack; | |
2125 { | |
2126 Frag_T *stackp; | |
2127 | |
2128 *p = *p - 1; | |
2129 stackp = *p; | |
2130 if (stackp < stack) | |
2131 return empty; | |
2132 return **p; | |
2133 } | |
2134 | |
2135 /* | |
2136 * Convert a postfix form into its equivalent NFA. | |
2137 * Return the NFA start state on success, NULL otherwise. | |
2138 */ | |
2139 static nfa_state_T * | |
2140 post2nfa(postfix, end, nfa_calc_size) | |
2141 int *postfix; | |
2142 int *end; | |
2143 int nfa_calc_size; | |
2144 { | |
2145 int *p; | |
2146 int mopen; | |
2147 int mclose; | |
2148 Frag_T *stack = NULL; | |
2149 Frag_T *stackp = NULL; | |
2150 Frag_T *stack_end = NULL; | |
2151 Frag_T e1; | |
2152 Frag_T e2; | |
2153 Frag_T e; | |
2154 nfa_state_T *s; | |
2155 nfa_state_T *s1; | |
2156 nfa_state_T *matchstate; | |
4484 | 2157 nfa_state_T *ret = NULL; |
4444 | 2158 |
2159 if (postfix == NULL) | |
2160 return NULL; | |
2161 | |
4456 | 2162 #define PUSH(s) st_push((s), &stackp, stack_end) |
4444 | 2163 #define POP() st_pop(&stackp, stack); \ |
2164 if (stackp < stack) \ | |
2165 { \ | |
2166 st_error(postfix, end, p); \ | |
2167 return NULL; \ | |
2168 } | |
2169 | |
2170 if (nfa_calc_size == FALSE) | |
2171 { | |
2172 /* Allocate space for the stack. Max states on the stack : nstate */ | |
4462 | 2173 stack = (Frag_T *) lalloc((nstate + 1) * sizeof(Frag_T), TRUE); |
4444 | 2174 stackp = stack; |
4462 | 2175 stack_end = stack + (nstate + 1); |
4444 | 2176 } |
2177 | |
2178 for (p = postfix; p < end; ++p) | |
2179 { | |
2180 switch (*p) | |
2181 { | |
2182 case NFA_CONCAT: | |
2183 /* Catenation. | |
2184 * Pay attention: this operator does not exist | |
2185 * in the r.e. itself (it is implicit, really). | |
2186 * It is added when r.e. is translated to postfix | |
2187 * form in re2post(). | |
2188 * | |
2189 * No new state added here. */ | |
2190 if (nfa_calc_size == TRUE) | |
2191 { | |
4458 | 2192 /* nstate += 0; */ |
4444 | 2193 break; |
2194 } | |
2195 e2 = POP(); | |
2196 e1 = POP(); | |
2197 patch(e1.out, e2.start); | |
2198 PUSH(frag(e1.start, e2.out)); | |
2199 break; | |
2200 | |
2201 case NFA_NOT: | |
2202 /* Negation of a character */ | |
2203 if (nfa_calc_size == TRUE) | |
2204 { | |
4458 | 2205 /* nstate += 0; */ |
4444 | 2206 break; |
2207 } | |
2208 e1 = POP(); | |
2209 e1.start->negated = TRUE; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2210 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2211 if (e1.start->c == NFA_COMPOSING) |
4444 | 2212 e1.start->out1->negated = TRUE; |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2213 #endif |
4444 | 2214 PUSH(e1); |
2215 break; | |
2216 | |
2217 case NFA_OR: | |
2218 /* Alternation */ | |
2219 if (nfa_calc_size == TRUE) | |
2220 { | |
4458 | 2221 nstate++; |
4444 | 2222 break; |
2223 } | |
2224 e2 = POP(); | |
2225 e1 = POP(); | |
2226 s = new_state(NFA_SPLIT, e1.start, e2.start); | |
2227 if (s == NULL) | |
4484 | 2228 goto theend; |
4444 | 2229 PUSH(frag(s, append(e1.out, e2.out))); |
2230 break; | |
2231 | |
2232 case NFA_STAR: | |
2233 /* Zero or more */ | |
2234 if (nfa_calc_size == TRUE) | |
2235 { | |
4458 | 2236 nstate++; |
4444 | 2237 break; |
2238 } | |
2239 e = POP(); | |
2240 s = new_state(NFA_SPLIT, e.start, NULL); | |
2241 if (s == NULL) | |
4484 | 2242 goto theend; |
4444 | 2243 patch(e.out, s); |
2244 PUSH(frag(s, list1(&s->out1))); | |
2245 break; | |
2246 | |
2247 case NFA_QUEST: | |
2248 /* one or zero atoms=> greedy match */ | |
2249 if (nfa_calc_size == TRUE) | |
2250 { | |
4458 | 2251 nstate++; |
4444 | 2252 break; |
2253 } | |
2254 e = POP(); | |
2255 s = new_state(NFA_SPLIT, e.start, NULL); | |
2256 if (s == NULL) | |
4484 | 2257 goto theend; |
4444 | 2258 PUSH(frag(s, append(e.out, list1(&s->out1)))); |
2259 break; | |
2260 | |
2261 case NFA_QUEST_NONGREEDY: | |
2262 /* zero or one atoms => non-greedy match */ | |
2263 if (nfa_calc_size == TRUE) | |
2264 { | |
4458 | 2265 nstate++; |
4444 | 2266 break; |
2267 } | |
2268 e = POP(); | |
2269 s = new_state(NFA_SPLIT, NULL, e.start); | |
2270 if (s == NULL) | |
4484 | 2271 goto theend; |
4444 | 2272 PUSH(frag(s, append(e.out, list1(&s->out)))); |
2273 break; | |
2274 | |
2275 case NFA_PLUS: | |
2276 /* One or more */ | |
2277 if (nfa_calc_size == TRUE) | |
2278 { | |
4458 | 2279 nstate++; |
4444 | 2280 break; |
2281 } | |
2282 e = POP(); | |
2283 s = new_state(NFA_SPLIT, e.start, NULL); | |
2284 if (s == NULL) | |
4484 | 2285 goto theend; |
4444 | 2286 patch(e.out, s); |
2287 PUSH(frag(e.start, list1(&s->out1))); | |
2288 break; | |
2289 | |
2290 case NFA_SKIP_CHAR: | |
2291 /* Symbol of 0-length, Used in a repetition | |
2292 * with max/min count of 0 */ | |
2293 if (nfa_calc_size == TRUE) | |
2294 { | |
4458 | 2295 nstate++; |
4444 | 2296 break; |
2297 } | |
2298 s = new_state(NFA_SKIP_CHAR, NULL, NULL); | |
2299 if (s == NULL) | |
4484 | 2300 goto theend; |
4444 | 2301 PUSH(frag(s, list1(&s->out))); |
2302 break; | |
2303 | |
2304 case NFA_PREV_ATOM_NO_WIDTH: | |
2305 /* The \@= operator: match the preceding atom with 0 width. | |
2306 * Surrounds the preceding atom with START_INVISIBLE and | |
2307 * END_INVISIBLE, similarly to MOPEN. | |
2308 */ | |
2309 /* TODO: Maybe this drops the speed? */ | |
4484 | 2310 goto theend; |
4444 | 2311 |
2312 if (nfa_calc_size == TRUE) | |
2313 { | |
2314 nstate += 2; | |
2315 break; | |
2316 } | |
2317 e = POP(); | |
2318 s1 = new_state(NFA_END_INVISIBLE, NULL, NULL); | |
2319 if (s1 == NULL) | |
4484 | 2320 goto theend; |
4444 | 2321 patch(e.out, s1); |
2322 | |
2323 s = new_state(NFA_START_INVISIBLE, e.start, s1); | |
2324 if (s == NULL) | |
4484 | 2325 goto theend; |
4444 | 2326 PUSH(frag(s, list1(&s1->out))); |
2327 break; | |
2328 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2329 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2330 case NFA_COMPOSING: /* char with composing char */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2331 #if 0 |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2332 /* TODO */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2333 if (regflags & RF_ICOMBINE) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2334 { |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
2335 /* use the base character only */ |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2336 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2337 #endif |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2338 /* FALLTHROUGH */ |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2339 #endif |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2340 |
4444 | 2341 case NFA_MOPEN + 0: /* Submatch */ |
2342 case NFA_MOPEN + 1: | |
2343 case NFA_MOPEN + 2: | |
2344 case NFA_MOPEN + 3: | |
2345 case NFA_MOPEN + 4: | |
2346 case NFA_MOPEN + 5: | |
2347 case NFA_MOPEN + 6: | |
2348 case NFA_MOPEN + 7: | |
2349 case NFA_MOPEN + 8: | |
2350 case NFA_MOPEN + 9: | |
2351 case NFA_NOPEN: /* \%( "Invisible Submatch" */ | |
2352 if (nfa_calc_size == TRUE) | |
2353 { | |
2354 nstate += 2; | |
2355 break; | |
2356 } | |
2357 | |
2358 mopen = *p; | |
2359 switch (*p) | |
2360 { | |
2361 case NFA_NOPEN: | |
2362 mclose = NFA_NCLOSE; | |
2363 break; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2364 #ifdef FEAT_MBYTE |
4444 | 2365 case NFA_COMPOSING: |
2366 mclose = NFA_END_COMPOSING; | |
2367 break; | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2368 #endif |
4444 | 2369 default: |
2370 /* NFA_MOPEN(0) ... NFA_MOPEN(9) */ | |
2371 mclose = *p + NSUBEXP; | |
2372 break; | |
2373 } | |
2374 | |
2375 /* Allow "NFA_MOPEN" as a valid postfix representation for | |
2376 * the empty regexp "". In this case, the NFA will be | |
2377 * NFA_MOPEN -> NFA_MCLOSE. Note that this also allows | |
2378 * empty groups of parenthesis, and empty mbyte chars */ | |
2379 if (stackp == stack) | |
2380 { | |
2381 s = new_state(mopen, NULL, NULL); | |
2382 if (s == NULL) | |
4484 | 2383 goto theend; |
4444 | 2384 s1 = new_state(mclose, NULL, NULL); |
2385 if (s1 == NULL) | |
4484 | 2386 goto theend; |
4444 | 2387 patch(list1(&s->out), s1); |
2388 PUSH(frag(s, list1(&s1->out))); | |
2389 break; | |
2390 } | |
2391 | |
2392 /* At least one node was emitted before NFA_MOPEN, so | |
2393 * at least one node will be between NFA_MOPEN and NFA_MCLOSE */ | |
2394 e = POP(); | |
2395 s = new_state(mopen, e.start, NULL); /* `(' */ | |
2396 if (s == NULL) | |
4484 | 2397 goto theend; |
4444 | 2398 |
2399 s1 = new_state(mclose, NULL, NULL); /* `)' */ | |
2400 if (s1 == NULL) | |
4484 | 2401 goto theend; |
4444 | 2402 patch(e.out, s1); |
2403 | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2404 #ifdef FEAT_MBYTE |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2405 if (mopen == NFA_COMPOSING) |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
2406 /* COMPOSING->out1 = END_COMPOSING */ |
4444 | 2407 patch(list1(&s->out1), s1); |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
2408 #endif |
4444 | 2409 |
2410 PUSH(frag(s, list1(&s1->out))); | |
2411 break; | |
2412 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2413 case NFA_BACKREF1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2414 case NFA_BACKREF2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2415 case NFA_BACKREF3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2416 case NFA_BACKREF4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2417 case NFA_BACKREF5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2418 case NFA_BACKREF6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2419 case NFA_BACKREF7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2420 case NFA_BACKREF8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2421 case NFA_BACKREF9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2422 if (nfa_calc_size == TRUE) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2423 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2424 nstate += 2; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2425 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2426 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2427 s = new_state(*p, NULL, NULL); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2428 if (s == NULL) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2429 goto theend; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2430 s1 = new_state(NFA_SKIP, NULL, NULL); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2431 if (s1 == NULL) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2432 goto theend; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2433 patch(list1(&s->out), s1); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2434 PUSH(frag(s, list1(&s1->out))); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2435 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2436 |
4444 | 2437 case NFA_ZSTART: |
2438 case NFA_ZEND: | |
2439 default: | |
2440 /* Operands */ | |
2441 if (nfa_calc_size == TRUE) | |
2442 { | |
4458 | 2443 nstate++; |
4444 | 2444 break; |
2445 } | |
2446 s = new_state(*p, NULL, NULL); | |
2447 if (s == NULL) | |
4484 | 2448 goto theend; |
4444 | 2449 PUSH(frag(s, list1(&s->out))); |
2450 break; | |
2451 | |
2452 } /* switch(*p) */ | |
2453 | |
2454 } /* for(p = postfix; *p; ++p) */ | |
2455 | |
2456 if (nfa_calc_size == TRUE) | |
2457 { | |
4458 | 2458 nstate++; |
4484 | 2459 goto theend; /* Return value when counting size is ignored anyway */ |
4444 | 2460 } |
2461 | |
2462 e = POP(); | |
2463 if (stackp != stack) | |
2464 EMSG_RET_NULL(_("E875: (NFA regexp) (While converting from postfix to NFA), too many states left on stack")); | |
2465 | |
2466 if (istate >= nstate) | |
2467 EMSG_RET_NULL(_("E876: (NFA regexp) Not enough space to store the whole NFA ")); | |
2468 | |
2469 matchstate = &state_ptr[istate++]; /* the match state */ | |
2470 matchstate->c = NFA_MATCH; | |
2471 matchstate->out = matchstate->out1 = NULL; | |
2472 | |
2473 patch(e.out, matchstate); | |
4484 | 2474 ret = e.start; |
2475 | |
2476 theend: | |
2477 vim_free(stack); | |
2478 return ret; | |
4444 | 2479 |
2480 #undef POP1 | |
2481 #undef PUSH1 | |
2482 #undef POP2 | |
2483 #undef PUSH2 | |
2484 #undef POP | |
2485 #undef PUSH | |
2486 } | |
2487 | |
2488 /**************************************************************** | |
2489 * NFA execution code. | |
2490 ****************************************************************/ | |
2491 | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2492 typedef struct |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2493 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2494 int in_use; /* number of subexpr with useful info */ |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2495 |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2496 /* When REG_MULTI is TRUE multilist is used, otherwise linelist. */ |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2497 union |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2498 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2499 struct multipos |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2500 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2501 lpos_T start; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2502 lpos_T end; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2503 } multilist[NSUBEXP]; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2504 struct linepos |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2505 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2506 char_u *start; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2507 char_u *end; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2508 } linelist[NSUBEXP]; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2509 }; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2510 } regsub_T; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2511 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2512 /* nfa_thread_T contains execution information of a NFA state */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2513 typedef struct |
4444 | 2514 { |
2515 nfa_state_T *state; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2516 int count; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2517 regsub_T sub; /* submatch info, only party used */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2518 } nfa_thread_T; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2519 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2520 /* nfa_list_T contains the alternative NFA execution states. */ |
4444 | 2521 typedef struct |
2522 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2523 nfa_thread_T *t; /* allocated array of states */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2524 int n; /* nr of states in "t" */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2525 int id; /* ID of the list */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2526 } nfa_list_T; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2527 |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2528 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2529 static void |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2530 log_subexpr(sub) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2531 regsub_T *sub; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2532 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2533 int j; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2534 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2535 for (j = 0; j < sub->in_use; j++) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2536 if (REG_MULTI) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2537 fprintf(log_fd, "\n *** group %d, start: c=%d, l=%d, end: c=%d, l=%d", |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2538 j, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2539 sub->multilist[j].start.col, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2540 (int)sub->multilist[j].start.lnum, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2541 sub->multilist[j].end.col, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2542 (int)sub->multilist[j].end.lnum); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2543 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2544 fprintf(log_fd, "\n *** group %d, start: \"%s\", end: \"%s\"", |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2545 j, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2546 (char *)sub->linelist[j].start, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2547 (char *)sub->linelist[j].end); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2548 fprintf(log_fd, "\n"); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2549 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2550 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2551 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2552 /* Used during execution: whether a match has been found. */ |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2553 static int nfa_match; |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2554 |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2555 static void addstate __ARGS((nfa_list_T *l, nfa_state_T *state, regsub_T *sub, int off)); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2556 static void addstate_here __ARGS((nfa_list_T *l, nfa_state_T *state, regsub_T *sub, int *ip)); |
4444 | 2557 |
2558 static void | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2559 addstate(l, state, sub, off) |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2560 nfa_list_T *l; /* runtime state list */ |
4444 | 2561 nfa_state_T *state; /* state to update */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2562 regsub_T *sub; /* pointers to subexpressions */ |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2563 int off; /* byte offset, when -1 go to next line */ |
4444 | 2564 { |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2565 int subidx; |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2566 nfa_thread_T *lastthread; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2567 lpos_T save_lpos; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2568 int save_in_use; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2569 char_u *save_ptr; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2570 int i; |
4444 | 2571 |
2572 if (l == NULL || state == NULL) | |
2573 return; | |
2574 | |
2575 switch (state->c) | |
2576 { | |
2577 case NFA_SPLIT: | |
2578 case NFA_NOT: | |
2579 case NFA_NOPEN: | |
2580 case NFA_NCLOSE: | |
2581 case NFA_MCLOSE: | |
2582 case NFA_MCLOSE + 1: | |
2583 case NFA_MCLOSE + 2: | |
2584 case NFA_MCLOSE + 3: | |
2585 case NFA_MCLOSE + 4: | |
2586 case NFA_MCLOSE + 5: | |
2587 case NFA_MCLOSE + 6: | |
2588 case NFA_MCLOSE + 7: | |
2589 case NFA_MCLOSE + 8: | |
2590 case NFA_MCLOSE + 9: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2591 /* These nodes are not added themselves but their "out" and/or |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2592 * "out1" may be added below. */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2593 break; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2594 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2595 case NFA_MOPEN: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2596 case NFA_MOPEN + 1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2597 case NFA_MOPEN + 2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2598 case NFA_MOPEN + 3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2599 case NFA_MOPEN + 4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2600 case NFA_MOPEN + 5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2601 case NFA_MOPEN + 6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2602 case NFA_MOPEN + 7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2603 case NFA_MOPEN + 8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2604 case NFA_MOPEN + 9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2605 /* These nodes do not need to be added, but we need to bail out |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2606 * when it was tried to be added to this list before. */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2607 if (state->lastlist == l->id) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2608 return; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2609 state->lastlist = l->id; |
4444 | 2610 break; |
2611 | |
2612 default: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2613 if (state->lastlist == l->id) |
4444 | 2614 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2615 /* This state is already in the list, don't add it again, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2616 * unless it is an MOPEN that is used for a backreference. */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2617 return; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2618 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2619 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2620 /* add the state to the list */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2621 state->lastlist = l->id; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2622 lastthread = &l->t[l->n++]; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2623 lastthread->state = state; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2624 lastthread->sub.in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2625 if (sub->in_use > 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2626 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2627 /* Copy the match start and end positions. */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2628 if (REG_MULTI) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2629 mch_memmove(&lastthread->sub.multilist[0], |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2630 &sub->multilist[0], |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2631 sizeof(struct multipos) * sub->in_use); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2632 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2633 mch_memmove(&lastthread->sub.linelist[0], |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2634 &sub->linelist[0], |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2635 sizeof(struct linepos) * sub->in_use); |
4444 | 2636 } |
2637 } | |
2638 | |
2639 #ifdef ENABLE_LOG | |
2640 nfa_set_code(state->c); | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2641 fprintf(log_fd, "> Adding state %d to list. Character %d: %s\n", |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2642 abs(state->id), state->c, code); |
4444 | 2643 #endif |
2644 switch (state->c) | |
2645 { | |
2646 case NFA_MATCH: | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2647 nfa_match = TRUE; |
4444 | 2648 break; |
2649 | |
2650 case NFA_SPLIT: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2651 addstate(l, state->out, sub, off); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2652 addstate(l, state->out1, sub, off); |
4444 | 2653 break; |
2654 | |
2655 #if 0 | |
2656 case NFA_END_NEG_RANGE: | |
2657 /* Nothing to handle here. nfa_regmatch() will take care of it */ | |
2658 break; | |
2659 | |
2660 case NFA_NOT: | |
2661 EMSG(_("E999: (NFA regexp internal error) Should not process NOT node !")); | |
2662 #ifdef ENABLE_LOG | |
2663 fprintf(f, "\n\n>>> E999: Added state NFA_NOT to a list ... Something went wrong ! Why wasn't it processed already? \n\n"); | |
2664 #endif | |
2665 break; | |
2666 | |
2667 case NFA_COMPOSING: | |
2668 /* nfa_regmatch() will match all the bytes of this composing char. */ | |
2669 break; | |
2670 #endif | |
2671 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2672 case NFA_SKIP_CHAR: |
4444 | 2673 case NFA_NOPEN: |
2674 case NFA_NCLOSE: | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2675 addstate(l, state->out, sub, off); |
4444 | 2676 break; |
2677 | |
2678 /* If this state is reached, then a recursive call of nfa_regmatch() | |
2679 * succeeded. the next call saves the found submatches in the | |
2680 * first state after the "invisible" branch. */ | |
2681 #if 0 | |
2682 case NFA_END_INVISIBLE: | |
2683 break; | |
2684 #endif | |
2685 | |
2686 case NFA_MOPEN + 0: | |
2687 case NFA_MOPEN + 1: | |
2688 case NFA_MOPEN + 2: | |
2689 case NFA_MOPEN + 3: | |
2690 case NFA_MOPEN + 4: | |
2691 case NFA_MOPEN + 5: | |
2692 case NFA_MOPEN + 6: | |
2693 case NFA_MOPEN + 7: | |
2694 case NFA_MOPEN + 8: | |
2695 case NFA_MOPEN + 9: | |
2696 case NFA_ZSTART: | |
2697 if (state->c == NFA_ZSTART) | |
2698 subidx = 0; | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2699 else |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2700 subidx = state->c - NFA_MOPEN; |
4444 | 2701 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2702 /* Set the position (with "off") in the subexpression. Save and |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2703 * restore it when it was in use. Otherwise fill any gap. */ |
4444 | 2704 if (REG_MULTI) |
2705 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2706 if (subidx < sub->in_use) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2707 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2708 save_lpos = sub->multilist[subidx].start; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2709 save_in_use = -1; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2710 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2711 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2712 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2713 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2714 for (i = sub->in_use; i < subidx; ++i) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2715 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2716 sub->multilist[i].start.lnum = -1; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2717 sub->multilist[i].end.lnum = -1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2718 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2719 sub->in_use = subidx + 1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2720 } |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2721 if (off == -1) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2722 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2723 sub->multilist[subidx].start.lnum = reglnum + 1; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2724 sub->multilist[subidx].start.col = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2725 } |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2726 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2727 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2728 sub->multilist[subidx].start.lnum = reglnum; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2729 sub->multilist[subidx].start.col = |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2730 (colnr_T)(reginput - regline + off); |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2731 } |
4444 | 2732 } |
2733 else | |
2734 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2735 if (subidx < sub->in_use) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2736 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2737 save_ptr = sub->linelist[subidx].start; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2738 save_in_use = -1; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2739 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2740 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2741 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2742 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2743 for (i = sub->in_use; i < subidx; ++i) |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2744 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2745 sub->linelist[i].start = NULL; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2746 sub->linelist[i].end = NULL; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2747 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2748 sub->in_use = subidx + 1; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2749 } |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2750 sub->linelist[subidx].start = reginput + off; |
4444 | 2751 } |
2752 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2753 addstate(l, state->out, sub, off); |
4444 | 2754 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2755 if (save_in_use == -1) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2756 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2757 if (REG_MULTI) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2758 sub->multilist[subidx].start = save_lpos; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2759 else |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2760 sub->linelist[subidx].start = save_ptr; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2761 } |
4444 | 2762 else |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2763 sub->in_use = save_in_use; |
4444 | 2764 break; |
2765 | |
2766 case NFA_MCLOSE + 0: | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
2767 if (nfa_has_zend) |
4444 | 2768 { |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
2769 /* Do not overwrite the position set by \ze. If no \ze |
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
2770 * encountered end will be set in nfa_regtry(). */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2771 addstate(l, state->out, sub, off); |
4444 | 2772 break; |
2773 } | |
2774 case NFA_MCLOSE + 1: | |
2775 case NFA_MCLOSE + 2: | |
2776 case NFA_MCLOSE + 3: | |
2777 case NFA_MCLOSE + 4: | |
2778 case NFA_MCLOSE + 5: | |
2779 case NFA_MCLOSE + 6: | |
2780 case NFA_MCLOSE + 7: | |
2781 case NFA_MCLOSE + 8: | |
2782 case NFA_MCLOSE + 9: | |
2783 case NFA_ZEND: | |
2784 if (state->c == NFA_ZEND) | |
2785 subidx = 0; | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2786 else |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2787 subidx = state->c - NFA_MCLOSE; |
4444 | 2788 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2789 /* We don't fill in gaps here, there must have been an MOPEN that |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
2790 * has done that. */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2791 save_in_use = sub->in_use; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2792 if (sub->in_use <= subidx) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2793 sub->in_use = subidx + 1; |
4444 | 2794 if (REG_MULTI) |
2795 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2796 save_lpos = sub->multilist[subidx].end; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2797 if (off == -1) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2798 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2799 sub->multilist[subidx].end.lnum = reglnum + 1; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2800 sub->multilist[subidx].end.col = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2801 } |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2802 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2803 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2804 sub->multilist[subidx].end.lnum = reglnum; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2805 sub->multilist[subidx].end.col = |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
2806 (colnr_T)(reginput - regline + off); |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
2807 } |
4444 | 2808 } |
2809 else | |
2810 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2811 save_ptr = sub->linelist[subidx].end; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2812 sub->linelist[subidx].end = reginput + off; |
4444 | 2813 } |
2814 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2815 addstate(l, state->out, sub, off); |
4444 | 2816 |
2817 if (REG_MULTI) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2818 sub->multilist[subidx].end = save_lpos; |
4444 | 2819 else |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2820 sub->linelist[subidx].end = save_ptr; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2821 sub->in_use = save_in_use; |
4444 | 2822 break; |
2823 } | |
2824 } | |
2825 | |
2826 /* | |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2827 * Like addstate(), but the new state(s) are put at position "*ip". |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2828 * Used for zero-width matches, next state to use is the added one. |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2829 * This makes sure the order of states to be tried does not change, which |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2830 * matters for alternatives. |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2831 */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2832 static void |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2833 addstate_here(l, state, sub, ip) |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2834 nfa_list_T *l; /* runtime state list */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2835 nfa_state_T *state; /* state to update */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2836 regsub_T *sub; /* pointers to subexpressions */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2837 int *ip; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2838 { |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2839 int tlen = l->n; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2840 int count; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2841 int i = *ip; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2842 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2843 /* first add the state(s) at the end, so that we know how many there are */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2844 addstate(l, state, sub, 0); |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2845 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2846 /* when "*ip" was at the end of the list, nothing to do */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2847 if (i + 1 == tlen) |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2848 return; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2849 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2850 /* re-order to put the new state at the current position */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2851 count = l->n - tlen; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2852 if (count > 1) |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2853 { |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2854 /* make space for new states, then move them from the |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2855 * end to the current position */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2856 mch_memmove(&(l->t[i + count]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2857 &(l->t[i + 1]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2858 sizeof(nfa_thread_T) * (l->n - i - 1)); |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2859 mch_memmove(&(l->t[i]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2860 &(l->t[l->n - 1]), |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2861 sizeof(nfa_thread_T) * count); |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2862 } |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2863 else |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2864 { |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2865 /* overwrite the current state */ |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2866 l->t[i] = l->t[l->n - 1]; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2867 } |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2868 --l->n; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2869 *ip = i - 1; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2870 } |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2871 |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
2872 /* |
4444 | 2873 * Check character class "class" against current character c. |
2874 */ | |
2875 static int | |
2876 check_char_class(class, c) | |
2877 int class; | |
2878 int c; | |
2879 { | |
2880 switch (class) | |
2881 { | |
2882 case NFA_CLASS_ALNUM: | |
4464 | 2883 if (c >= 1 && c <= 255 && isalnum(c)) |
4444 | 2884 return OK; |
2885 break; | |
2886 case NFA_CLASS_ALPHA: | |
4464 | 2887 if (c >= 1 && c <= 255 && isalpha(c)) |
4444 | 2888 return OK; |
2889 break; | |
2890 case NFA_CLASS_BLANK: | |
2891 if (c == ' ' || c == '\t') | |
2892 return OK; | |
2893 break; | |
2894 case NFA_CLASS_CNTRL: | |
4464 | 2895 if (c >= 1 && c <= 255 && iscntrl(c)) |
4444 | 2896 return OK; |
2897 break; | |
2898 case NFA_CLASS_DIGIT: | |
2899 if (VIM_ISDIGIT(c)) | |
2900 return OK; | |
2901 break; | |
2902 case NFA_CLASS_GRAPH: | |
4464 | 2903 if (c >= 1 && c <= 255 && isgraph(c)) |
4444 | 2904 return OK; |
2905 break; | |
2906 case NFA_CLASS_LOWER: | |
2907 if (MB_ISLOWER(c)) | |
2908 return OK; | |
2909 break; | |
2910 case NFA_CLASS_PRINT: | |
2911 if (vim_isprintc(c)) | |
2912 return OK; | |
2913 break; | |
2914 case NFA_CLASS_PUNCT: | |
4464 | 2915 if (c >= 1 && c <= 255 && ispunct(c)) |
4444 | 2916 return OK; |
2917 break; | |
2918 case NFA_CLASS_SPACE: | |
2919 if ((c >=9 && c <= 13) || (c == ' ')) | |
2920 return OK; | |
2921 break; | |
2922 case NFA_CLASS_UPPER: | |
2923 if (MB_ISUPPER(c)) | |
2924 return OK; | |
2925 break; | |
2926 case NFA_CLASS_XDIGIT: | |
2927 if (vim_isxdigit(c)) | |
2928 return OK; | |
2929 break; | |
2930 case NFA_CLASS_TAB: | |
2931 if (c == '\t') | |
2932 return OK; | |
2933 break; | |
2934 case NFA_CLASS_RETURN: | |
2935 if (c == '\r') | |
2936 return OK; | |
2937 break; | |
2938 case NFA_CLASS_BACKSPACE: | |
2939 if (c == '\b') | |
2940 return OK; | |
2941 break; | |
2942 case NFA_CLASS_ESCAPE: | |
2943 if (c == '\033') | |
2944 return OK; | |
2945 break; | |
2946 | |
2947 default: | |
2948 /* should not be here :P */ | |
2949 EMSG_RET_FAIL(_("E877: (NFA regexp) Invalid character class ")); | |
2950 } | |
2951 return FAIL; | |
2952 } | |
2953 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2954 static int match_backref __ARGS((regsub_T *sub, int subidx, int *bytelen)); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2955 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2956 /* |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2957 * Check for a match with subexpression "subidx". |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2958 * return TRUE if it matches. |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2959 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2960 static int |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2961 match_backref(sub, subidx, bytelen) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2962 regsub_T *sub; /* pointers to subexpressions */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2963 int subidx; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2964 int *bytelen; /* out: length of match in bytes */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2965 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2966 int len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2967 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2968 if (sub->in_use <= subidx) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2969 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2970 retempty: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2971 /* backref was not set, match an empty string */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2972 *bytelen = 0; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2973 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2974 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2975 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2976 if (REG_MULTI) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2977 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2978 if (sub->multilist[subidx].start.lnum < 0 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2979 || sub->multilist[subidx].end.lnum < 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2980 goto retempty; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2981 /* TODO: line breaks */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2982 len = sub->multilist[subidx].end.col |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2983 - sub->multilist[subidx].start.col; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2984 if (cstrncmp(regline + sub->multilist[subidx].start.col, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2985 reginput, &len) == 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2986 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2987 *bytelen = len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2988 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2989 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2990 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2991 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2992 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2993 if (sub->linelist[subidx].start == NULL |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2994 || sub->linelist[subidx].end == NULL) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2995 goto retempty; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2996 len = (int)(sub->linelist[subidx].end - sub->linelist[subidx].start); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2997 if (cstrncmp(sub->linelist[subidx].start, reginput, &len) == 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2998 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
2999 *bytelen = len; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3000 return TRUE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3001 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3002 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3003 return FALSE; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3004 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3005 |
4444 | 3006 /* |
3007 * Set all NFA nodes' list ID equal to -1. | |
3008 */ | |
3009 static void | |
3010 nfa_set_neg_listids(start) | |
3011 nfa_state_T *start; | |
3012 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3013 if (start != NULL && start->lastlist >= 0) |
4444 | 3014 { |
3015 start->lastlist = -1; | |
3016 nfa_set_neg_listids(start->out); | |
3017 nfa_set_neg_listids(start->out1); | |
3018 } | |
3019 } | |
3020 | |
3021 /* | |
3022 * Set all NFA nodes' list ID equal to 0. | |
3023 */ | |
3024 static void | |
3025 nfa_set_null_listids(start) | |
3026 nfa_state_T *start; | |
3027 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3028 if (start != NULL && start->lastlist == -1) |
4444 | 3029 { |
3030 start->lastlist = 0; | |
3031 nfa_set_null_listids(start->out); | |
3032 nfa_set_null_listids(start->out1); | |
3033 } | |
3034 } | |
3035 | |
3036 /* | |
3037 * Save list IDs for all NFA states in "list". | |
3038 */ | |
3039 static void | |
3040 nfa_save_listids(start, list) | |
3041 nfa_state_T *start; | |
3042 int *list; | |
3043 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3044 if (start != NULL && start->lastlist != -1) |
4444 | 3045 { |
3046 list[abs(start->id)] = start->lastlist; | |
3047 start->lastlist = -1; | |
3048 nfa_save_listids(start->out, list); | |
3049 nfa_save_listids(start->out1, list); | |
3050 } | |
3051 } | |
3052 | |
3053 /* | |
3054 * Restore list IDs from "list" to all NFA states. | |
3055 */ | |
3056 static void | |
3057 nfa_restore_listids(start, list) | |
3058 nfa_state_T *start; | |
3059 int *list; | |
3060 { | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3061 if (start != NULL && start->lastlist == -1) |
4444 | 3062 { |
3063 start->lastlist = list[abs(start->id)]; | |
3064 nfa_restore_listids(start->out, list); | |
3065 nfa_restore_listids(start->out1, list); | |
3066 } | |
3067 } | |
3068 | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3069 static int nfa_regmatch __ARGS((nfa_state_T *start, regsub_T *submatch, regsub_T *m)); |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3070 |
4444 | 3071 /* |
3072 * Main matching routine. | |
3073 * | |
3074 * Run NFA to determine whether it matches reginput. | |
3075 * | |
3076 * Return TRUE if there is a match, FALSE otherwise. | |
3077 * Note: Caller must ensure that: start != NULL. | |
3078 */ | |
3079 static int | |
3080 nfa_regmatch(start, submatch, m) | |
3081 nfa_state_T *start; | |
3082 regsub_T *submatch; | |
3083 regsub_T *m; | |
3084 { | |
3085 int result; | |
3086 int size = 0; | |
3087 int flag = 0; | |
3088 int old_reglnum = -1; | |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3089 int go_to_nextline = FALSE; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3090 nfa_thread_T *t; |
4444 | 3091 char_u *old_reginput = NULL; |
3092 char_u *old_regline = NULL; | |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3093 nfa_list_T list[3]; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3094 nfa_list_T *listtbl[2][2]; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3095 nfa_list_T *ll; |
4444 | 3096 int listid = 1; |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3097 int listidx; |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3098 nfa_list_T *thislist; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3099 nfa_list_T *nextlist; |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3100 nfa_list_T *neglist; |
4444 | 3101 int *listids = NULL; |
4460 | 3102 #ifdef NFA_REGEXP_DEBUG_LOG |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3103 FILE *debug = fopen(NFA_REGEXP_DEBUG_LOG, "a"); |
4444 | 3104 |
3105 if (debug == NULL) | |
3106 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3107 EMSG2(_("(NFA) COULD NOT OPEN %s !"), NFA_REGEXP_DEBUG_LOG); |
4444 | 3108 return FALSE; |
3109 } | |
3110 #endif | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3111 nfa_match = FALSE; |
4444 | 3112 |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3113 /* Allocate memory for the lists of nodes. */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3114 size = (nstate + 1) * sizeof(nfa_thread_T); |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3115 list[0].t = (nfa_thread_T *)lalloc(size, TRUE); |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3116 list[1].t = (nfa_thread_T *)lalloc(size, TRUE); |
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3117 list[2].t = (nfa_thread_T *)lalloc(size, TRUE); |
4444 | 3118 if (list[0].t == NULL || list[1].t == NULL || list[2].t == NULL) |
3119 goto theend; | |
3120 vim_memset(list[0].t, 0, size); | |
3121 vim_memset(list[1].t, 0, size); | |
3122 vim_memset(list[2].t, 0, size); | |
3123 | |
3124 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3125 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 3126 if (log_fd != NULL) |
3127 { | |
3128 fprintf(log_fd, "**********************************\n"); | |
3129 nfa_set_code(start->c); | |
3130 fprintf(log_fd, " RUNNING nfa_regmatch() starting with state %d, code %s\n", | |
3131 abs(start->id), code); | |
3132 fprintf(log_fd, "**********************************\n"); | |
3133 } | |
3134 else | |
3135 { | |
3136 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
3137 log_fd = stderr; | |
3138 } | |
3139 #endif | |
3140 | |
3141 thislist = &list[0]; | |
3142 thislist->n = 0; | |
3143 nextlist = &list[1]; | |
3144 nextlist->n = 0; | |
3145 neglist = &list[2]; | |
3146 neglist->n = 0; | |
3147 #ifdef ENABLE_LOG | |
3148 fprintf(log_fd, "(---) STARTSTATE\n"); | |
3149 #endif | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3150 thislist->id = listid; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3151 addstate(thislist, start, m, 0); |
4444 | 3152 |
3153 /* There are two cases when the NFA advances: 1. input char matches the | |
3154 * NFA node and 2. input char does not match the NFA node, but the next | |
3155 * node is NFA_NOT. The following macro calls addstate() according to | |
3156 * these rules. It is used A LOT, so use the "listtbl" table for speed */ | |
3157 listtbl[0][0] = NULL; | |
3158 listtbl[0][1] = neglist; | |
3159 listtbl[1][0] = nextlist; | |
3160 listtbl[1][1] = NULL; | |
3161 #define ADD_POS_NEG_STATE(node) \ | |
3162 ll = listtbl[result ? 1 : 0][node->negated]; \ | |
3163 if (ll != NULL) \ | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3164 addstate(ll, node->out , &t->sub, clen); |
4444 | 3165 |
3166 | |
3167 /* | |
3168 * Run for each character. | |
3169 */ | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3170 for (;;) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3171 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3172 int curc; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3173 int clen; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3174 |
4444 | 3175 #ifdef FEAT_MBYTE |
3176 if (has_mbyte) | |
3177 { | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3178 curc = (*mb_ptr2char)(reginput); |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3179 clen = (*mb_ptr2len)(reginput); |
4444 | 3180 } |
3181 else | |
3182 #endif | |
3183 { | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3184 curc = *reginput; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3185 clen = 1; |
4444 | 3186 } |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3187 if (curc == NUL) |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3188 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3189 clen = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3190 go_to_nextline = FALSE; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3191 } |
4444 | 3192 |
3193 /* swap lists */ | |
3194 thislist = &list[flag]; | |
3195 nextlist = &list[flag ^= 1]; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3196 nextlist->n = 0; /* clear nextlist */ |
4444 | 3197 listtbl[1][0] = nextlist; |
3198 ++listid; | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3199 thislist->id = listid; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3200 nextlist->id = listid + 1; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3201 neglist->id = listid + 1; |
4444 | 3202 |
3203 #ifdef ENABLE_LOG | |
3204 fprintf(log_fd, "------------------------------------------\n"); | |
3205 fprintf(log_fd, ">>> Reginput is \"%s\"\n", reginput); | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3206 fprintf(log_fd, ">>> Advanced one character ... Current char is %c (code %d) \n", curc, (int)curc); |
4444 | 3207 fprintf(log_fd, ">>> Thislist has %d states available: ", thislist->n); |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3208 { |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3209 int i; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3210 |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3211 for (i = 0; i < thislist->n; i++) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3212 fprintf(log_fd, "%d ", abs(thislist->t[i].state->id)); |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3213 } |
4444 | 3214 fprintf(log_fd, "\n"); |
3215 #endif | |
3216 | |
4460 | 3217 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 3218 fprintf(debug, "\n-------------------\n"); |
3219 #endif | |
4480 | 3220 /* |
3221 * If the state lists are empty we can stop. | |
3222 */ | |
3223 if (thislist->n == 0 && neglist->n == 0) | |
3224 break; | |
4444 | 3225 |
3226 /* compute nextlist */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3227 for (listidx = 0; listidx < thislist->n || neglist->n > 0; ++listidx) |
4444 | 3228 { |
3229 if (neglist->n > 0) | |
3230 { | |
3231 t = &neglist->t[0]; | |
4468 | 3232 neglist->n--; |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3233 listidx--; |
4444 | 3234 } |
3235 else | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3236 t = &thislist->t[listidx]; |
4444 | 3237 |
4460 | 3238 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 3239 nfa_set_code(t->state->c); |
3240 fprintf(debug, "%s, ", code); | |
3241 #endif | |
3242 #ifdef ENABLE_LOG | |
3243 nfa_set_code(t->state->c); | |
3244 fprintf(log_fd, "(%d) %s, code %d ... \n", abs(t->state->id), | |
3245 code, (int)t->state->c); | |
3246 #endif | |
3247 | |
3248 /* | |
3249 * Handle the possible codes of the current state. | |
3250 * The most important is NFA_MATCH. | |
3251 */ | |
3252 switch (t->state->c) | |
3253 { | |
3254 case NFA_MATCH: | |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3255 { |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3256 int j; |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3257 |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3258 nfa_match = TRUE; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3259 submatch->in_use = t->sub.in_use; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3260 if (REG_MULTI) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3261 for (j = 0; j < submatch->in_use; j++) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3262 { |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3263 submatch->multilist[j].start = |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3264 t->sub.multilist[j].start; |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3265 submatch->multilist[j].end = t->sub.multilist[j].end; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3266 } |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3267 else |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3268 for (j = 0; j < submatch->in_use; j++) |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3269 { |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3270 submatch->linelist[j].start = t->sub.linelist[j].start; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3271 submatch->linelist[j].end = t->sub.linelist[j].end; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3272 } |
4444 | 3273 #ifdef ENABLE_LOG |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3274 log_subexpr(&t->sub); |
4444 | 3275 #endif |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3276 /* Found the left-most longest match, do not look at any other |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3277 * states at this position. When the list of states is going |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3278 * to be empty quit without advancing, so that "reginput" is |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3279 * correct. */ |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3280 if (nextlist->n == 0 && neglist->n == 0) |
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
3281 clen = 0; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3282 goto nextchar; |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3283 } |
4444 | 3284 |
3285 case NFA_END_INVISIBLE: | |
3286 /* This is only encountered after a NFA_START_INVISIBLE node. | |
3287 * They surround a zero-width group, used with "\@=" and "\&". | |
3288 * If we got here, it means that the current "invisible" group | |
3289 * finished successfully, so return control to the parent | |
3290 * nfa_regmatch(). Submatches are stored in *m, and used in | |
3291 * the parent call. */ | |
3292 if (start->c == NFA_MOPEN + 0) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3293 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3294 else |
3295 { | |
3296 *m = t->sub; | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3297 nfa_match = TRUE; |
4444 | 3298 } |
3299 break; | |
3300 | |
3301 case NFA_START_INVISIBLE: | |
3302 /* Save global variables, and call nfa_regmatch() to check if | |
3303 * the current concat matches at this position. The concat | |
3304 * ends with the node NFA_END_INVISIBLE */ | |
3305 old_reginput = reginput; | |
3306 old_regline = regline; | |
3307 old_reglnum = reglnum; | |
3308 if (listids == NULL) | |
3309 { | |
3310 listids = (int *) lalloc(sizeof(int) * nstate, TRUE); | |
3311 if (listids == NULL) | |
3312 { | |
3313 EMSG(_("E878: (NFA) Could not allocate memory for branch traversal!")); | |
3314 return 0; | |
3315 } | |
3316 } | |
3317 #ifdef ENABLE_LOG | |
3318 if (log_fd != stderr) | |
3319 fclose(log_fd); | |
3320 log_fd = NULL; | |
3321 #endif | |
3322 /* Have to clear the listid field of the NFA nodes, so that | |
3323 * nfa_regmatch() and addstate() can run properly after | |
3324 * recursion. */ | |
3325 nfa_save_listids(start, listids); | |
3326 nfa_set_null_listids(start); | |
3327 result = nfa_regmatch(t->state->out, submatch, m); | |
3328 nfa_set_neg_listids(start); | |
3329 nfa_restore_listids(start, listids); | |
3330 | |
3331 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3332 log_fd = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 3333 if (log_fd != NULL) |
3334 { | |
3335 fprintf(log_fd, "****************************\n"); | |
3336 fprintf(log_fd, "FINISHED RUNNING nfa_regmatch() recursively\n"); | |
3337 fprintf(log_fd, "MATCH = %s\n", result == TRUE ? "OK" : "FALSE"); | |
3338 fprintf(log_fd, "****************************\n"); | |
3339 } | |
3340 else | |
3341 { | |
3342 EMSG(_("Could not open temporary log file for writing, displaying on stderr ... ")); | |
3343 log_fd = stderr; | |
3344 } | |
3345 #endif | |
3346 if (result == TRUE) | |
3347 { | |
4567
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3348 int j; |
96c1a7850097
updated for version 7.3.1031
Bram Moolenaar <bram@vim.org>
parents:
4565
diff
changeset
|
3349 |
4444 | 3350 /* Restore position in input text */ |
3351 reginput = old_reginput; | |
3352 regline = old_regline; | |
3353 reglnum = old_reglnum; | |
3354 /* Copy submatch info from the recursive call */ | |
3355 if (REG_MULTI) | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3356 for (j = 1; j < m->in_use; j++) |
4444 | 3357 { |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3358 t->sub.multilist[j].start = m->multilist[j].start; |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3359 t->sub.multilist[j].end = m->multilist[j].end; |
4444 | 3360 } |
3361 else | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3362 for (j = 1; j < m->in_use; j++) |
4444 | 3363 { |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3364 t->sub.linelist[j].start = m->linelist[j].start; |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3365 t->sub.linelist[j].end = m->linelist[j].end; |
4444 | 3366 } |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3367 t->sub.in_use = m->in_use; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3368 |
4444 | 3369 /* t->state->out1 is the corresponding END_INVISIBLE node */ |
4539
532c2e850256
updated for version 7.3.1017
Bram Moolenaar <bram@vim.org>
parents:
4537
diff
changeset
|
3370 addstate_here(thislist, t->state->out1->out, &t->sub, |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3371 &listidx); |
4444 | 3372 } |
3373 else | |
3374 { | |
3375 /* continue with next input char */ | |
3376 reginput = old_reginput; | |
3377 } | |
3378 break; | |
3379 | |
3380 case NFA_BOL: | |
3381 if (reginput == regline) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3382 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3383 break; |
3384 | |
3385 case NFA_EOL: | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3386 if (curc == NUL) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3387 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3388 break; |
3389 | |
3390 case NFA_BOW: | |
3391 { | |
3392 int bow = TRUE; | |
3393 | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3394 if (curc == NUL) |
4444 | 3395 bow = FALSE; |
3396 #ifdef FEAT_MBYTE | |
3397 else if (has_mbyte) | |
3398 { | |
3399 int this_class; | |
3400 | |
3401 /* Get class of current and previous char (if it exists). */ | |
4503 | 3402 this_class = mb_get_class_buf(reginput, reg_buf); |
4444 | 3403 if (this_class <= 1) |
3404 bow = FALSE; | |
3405 else if (reg_prev_class() == this_class) | |
3406 bow = FALSE; | |
3407 } | |
3408 #endif | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3409 else if (!vim_iswordc_buf(curc, reg_buf) |
4503 | 3410 || (reginput > regline |
3411 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
4444 | 3412 bow = FALSE; |
3413 if (bow) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3414 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3415 break; |
3416 } | |
3417 | |
3418 case NFA_EOW: | |
3419 { | |
3420 int eow = TRUE; | |
3421 | |
3422 if (reginput == regline) | |
3423 eow = FALSE; | |
3424 #ifdef FEAT_MBYTE | |
3425 else if (has_mbyte) | |
3426 { | |
3427 int this_class, prev_class; | |
3428 | |
3429 /* Get class of current and previous char (if it exists). */ | |
4503 | 3430 this_class = mb_get_class_buf(reginput, reg_buf); |
4444 | 3431 prev_class = reg_prev_class(); |
3432 if (this_class == prev_class | |
3433 || prev_class == 0 || prev_class == 1) | |
3434 eow = FALSE; | |
3435 } | |
3436 #endif | |
4503 | 3437 else if (!vim_iswordc_buf(reginput[-1], reg_buf) |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3438 || (reginput[0] != NUL |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3439 && vim_iswordc_buf(curc, reg_buf))) |
4444 | 3440 eow = FALSE; |
3441 if (eow) | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3442 addstate_here(thislist, t->state->out, &t->sub, &listidx); |
4444 | 3443 break; |
3444 } | |
3445 | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3446 #ifdef FEAT_MBYTE |
4444 | 3447 case NFA_COMPOSING: |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3448 { |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3449 int mc = curc; |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3450 int len = 0; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3451 nfa_state_T *end; |
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3452 nfa_state_T *sta; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3453 int cchars[MAX_MCO]; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3454 int ccount = 0; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3455 int j; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3456 |
4444 | 3457 sta = t->state->out; |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3458 len = 0; |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3459 if (utf_iscomposing(sta->c)) |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3460 { |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3461 /* Only match composing character(s), ignore base |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3462 * character. Used for ".{composing}" and "{composing}" |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3463 * (no preceding character). */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3464 len += mb_char2len(mc); |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3465 } |
4551
f0c37f5cf4ea
updated for version 7.3.1023
Bram Moolenaar <bram@vim.org>
parents:
4549
diff
changeset
|
3466 if (ireg_icombine && len == 0) |
4444 | 3467 { |
4535
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3468 /* If \Z was present, then ignore composing characters. |
45f97c349537
updated for version 7.3.1015
Bram Moolenaar <bram@vim.org>
parents:
4533
diff
changeset
|
3469 * When ignoring the base character this always matches. */ |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3470 /* TODO: How about negated? */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3471 if (len == 0 && sta->c != curc) |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3472 result = FAIL; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3473 else |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3474 result = OK; |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3475 while (sta->c != NFA_END_COMPOSING) |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3476 sta = sta->out; |
4444 | 3477 } |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3478 |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3479 /* Check base character matches first, unless ignored. */ |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3480 else if (len > 0 || mc == sta->c) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3481 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3482 if (len == 0) |
4529
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3483 { |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3484 len += mb_char2len(mc); |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3485 sta = sta->out; |
432a6b8c7d93
updated for version 7.3.1012
Bram Moolenaar <bram@vim.org>
parents:
4527
diff
changeset
|
3486 } |
4444 | 3487 |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3488 /* We don't care about the order of composing characters. |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3489 * Get them into cchars[] first. */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3490 while (len < clen) |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3491 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3492 mc = mb_ptr2char(reginput + len); |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3493 cchars[ccount++] = mc; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3494 len += mb_char2len(mc); |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3495 if (ccount == MAX_MCO) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3496 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3497 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3498 |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3499 /* Check that each composing char in the pattern matches a |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3500 * composing char in the text. We do not check if all |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3501 * composing chars are matched. */ |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3502 result = OK; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3503 while (sta->c != NFA_END_COMPOSING) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3504 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3505 for (j = 0; j < ccount; ++j) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3506 if (cchars[j] == sta->c) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3507 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3508 if (j == ccount) |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3509 { |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3510 result = FAIL; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3511 break; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3512 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3513 sta = sta->out; |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3514 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3515 } |
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3516 else |
4525
36ddcf4cecbc
updated for version 7.3.1010
Bram Moolenaar <bram@vim.org>
parents:
4517
diff
changeset
|
3517 result = FAIL; |
4547
fc997f05cbc7
updated for version 7.3.1021
Bram Moolenaar <bram@vim.org>
parents:
4543
diff
changeset
|
3518 |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3519 end = t->state->out1; /* NFA_END_COMPOSING */ |
4444 | 3520 ADD_POS_NEG_STATE(end); |
3521 break; | |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3522 } |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3523 #endif |
4444 | 3524 |
3525 case NFA_NEWL: | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3526 if (curc == NUL && !reg_line_lbr && REG_MULTI |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3527 && reglnum <= reg_maxline) |
4444 | 3528 { |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3529 go_to_nextline = TRUE; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3530 /* Pass -1 for the offset, which means taking the position |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3531 * at the start of the next line. */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3532 addstate(nextlist, t->state->out, &t->sub, -1); |
4444 | 3533 } |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3534 else if (curc == '\n' && reg_line_lbr) |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3535 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3536 /* match \n as if it is an ordinary character */ |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3537 addstate(nextlist, t->state->out, &t->sub, 1); |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
3538 } |
4444 | 3539 break; |
3540 | |
3541 case NFA_CLASS_ALNUM: | |
3542 case NFA_CLASS_ALPHA: | |
3543 case NFA_CLASS_BLANK: | |
3544 case NFA_CLASS_CNTRL: | |
3545 case NFA_CLASS_DIGIT: | |
3546 case NFA_CLASS_GRAPH: | |
3547 case NFA_CLASS_LOWER: | |
3548 case NFA_CLASS_PRINT: | |
3549 case NFA_CLASS_PUNCT: | |
3550 case NFA_CLASS_SPACE: | |
3551 case NFA_CLASS_UPPER: | |
3552 case NFA_CLASS_XDIGIT: | |
3553 case NFA_CLASS_TAB: | |
3554 case NFA_CLASS_RETURN: | |
3555 case NFA_CLASS_BACKSPACE: | |
3556 case NFA_CLASS_ESCAPE: | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3557 result = check_char_class(t->state->c, curc); |
4444 | 3558 ADD_POS_NEG_STATE(t->state); |
3559 break; | |
3560 | |
3561 case NFA_END_NEG_RANGE: | |
3562 /* This follows a series of negated nodes, like: | |
3563 * CHAR(x), NFA_NOT, CHAR(y), NFA_NOT etc. */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3564 if (curc > 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3565 addstate(nextlist, t->state->out, &t->sub, clen); |
4444 | 3566 break; |
3567 | |
3568 case NFA_ANY: | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3569 /* Any char except '\0', (end of input) does not match. */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3570 if (curc > 0) |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3571 addstate(nextlist, t->state->out, &t->sub, clen); |
4444 | 3572 break; |
3573 | |
3574 /* | |
3575 * Character classes like \a for alpha, \d for digit etc. | |
3576 */ | |
3577 case NFA_IDENT: /* \i */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3578 result = vim_isIDc(curc); |
4444 | 3579 ADD_POS_NEG_STATE(t->state); |
3580 break; | |
3581 | |
3582 case NFA_SIDENT: /* \I */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3583 result = !VIM_ISDIGIT(curc) && vim_isIDc(curc); |
4444 | 3584 ADD_POS_NEG_STATE(t->state); |
3585 break; | |
3586 | |
3587 case NFA_KWORD: /* \k */ | |
4503 | 3588 result = vim_iswordp_buf(reginput, reg_buf); |
4444 | 3589 ADD_POS_NEG_STATE(t->state); |
3590 break; | |
3591 | |
3592 case NFA_SKWORD: /* \K */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3593 result = !VIM_ISDIGIT(curc) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3594 && vim_iswordp_buf(reginput, reg_buf); |
4444 | 3595 ADD_POS_NEG_STATE(t->state); |
3596 break; | |
3597 | |
3598 case NFA_FNAME: /* \f */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3599 result = vim_isfilec(curc); |
4444 | 3600 ADD_POS_NEG_STATE(t->state); |
3601 break; | |
3602 | |
3603 case NFA_SFNAME: /* \F */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3604 result = !VIM_ISDIGIT(curc) && vim_isfilec(curc); |
4444 | 3605 ADD_POS_NEG_STATE(t->state); |
3606 break; | |
3607 | |
3608 case NFA_PRINT: /* \p */ | |
4470 | 3609 result = ptr2cells(reginput) == 1; |
4444 | 3610 ADD_POS_NEG_STATE(t->state); |
3611 break; | |
3612 | |
3613 case NFA_SPRINT: /* \P */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3614 result = !VIM_ISDIGIT(curc) && ptr2cells(reginput) == 1; |
4444 | 3615 ADD_POS_NEG_STATE(t->state); |
3616 break; | |
3617 | |
3618 case NFA_WHITE: /* \s */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3619 result = vim_iswhite(curc); |
4444 | 3620 ADD_POS_NEG_STATE(t->state); |
3621 break; | |
3622 | |
3623 case NFA_NWHITE: /* \S */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3624 result = curc != NUL && !vim_iswhite(curc); |
4444 | 3625 ADD_POS_NEG_STATE(t->state); |
3626 break; | |
3627 | |
3628 case NFA_DIGIT: /* \d */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3629 result = ri_digit(curc); |
4444 | 3630 ADD_POS_NEG_STATE(t->state); |
3631 break; | |
3632 | |
3633 case NFA_NDIGIT: /* \D */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3634 result = curc != NUL && !ri_digit(curc); |
4444 | 3635 ADD_POS_NEG_STATE(t->state); |
3636 break; | |
3637 | |
3638 case NFA_HEX: /* \x */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3639 result = ri_hex(curc); |
4444 | 3640 ADD_POS_NEG_STATE(t->state); |
3641 break; | |
3642 | |
3643 case NFA_NHEX: /* \X */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3644 result = curc != NUL && !ri_hex(curc); |
4444 | 3645 ADD_POS_NEG_STATE(t->state); |
3646 break; | |
3647 | |
3648 case NFA_OCTAL: /* \o */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3649 result = ri_octal(curc); |
4444 | 3650 ADD_POS_NEG_STATE(t->state); |
3651 break; | |
3652 | |
3653 case NFA_NOCTAL: /* \O */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3654 result = curc != NUL && !ri_octal(curc); |
4444 | 3655 ADD_POS_NEG_STATE(t->state); |
3656 break; | |
3657 | |
3658 case NFA_WORD: /* \w */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3659 result = ri_word(curc); |
4444 | 3660 ADD_POS_NEG_STATE(t->state); |
3661 break; | |
3662 | |
3663 case NFA_NWORD: /* \W */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3664 result = curc != NUL && !ri_word(curc); |
4444 | 3665 ADD_POS_NEG_STATE(t->state); |
3666 break; | |
3667 | |
3668 case NFA_HEAD: /* \h */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3669 result = ri_head(curc); |
4444 | 3670 ADD_POS_NEG_STATE(t->state); |
3671 break; | |
3672 | |
3673 case NFA_NHEAD: /* \H */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3674 result = curc != NUL && !ri_head(curc); |
4444 | 3675 ADD_POS_NEG_STATE(t->state); |
3676 break; | |
3677 | |
3678 case NFA_ALPHA: /* \a */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3679 result = ri_alpha(curc); |
4444 | 3680 ADD_POS_NEG_STATE(t->state); |
3681 break; | |
3682 | |
3683 case NFA_NALPHA: /* \A */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3684 result = curc != NUL && !ri_alpha(curc); |
4444 | 3685 ADD_POS_NEG_STATE(t->state); |
3686 break; | |
3687 | |
3688 case NFA_LOWER: /* \l */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3689 result = ri_lower(curc); |
4444 | 3690 ADD_POS_NEG_STATE(t->state); |
3691 break; | |
3692 | |
3693 case NFA_NLOWER: /* \L */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3694 result = curc != NUL && !ri_lower(curc); |
4444 | 3695 ADD_POS_NEG_STATE(t->state); |
3696 break; | |
3697 | |
3698 case NFA_UPPER: /* \u */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3699 result = ri_upper(curc); |
4444 | 3700 ADD_POS_NEG_STATE(t->state); |
3701 break; | |
3702 | |
3703 case NFA_NUPPER: /* \U */ | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3704 result = curc != NUL && !ri_upper(curc); |
4444 | 3705 ADD_POS_NEG_STATE(t->state); |
3706 break; | |
3707 | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3708 case NFA_BACKREF1: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3709 case NFA_BACKREF2: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3710 case NFA_BACKREF3: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3711 case NFA_BACKREF4: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3712 case NFA_BACKREF5: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3713 case NFA_BACKREF6: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3714 case NFA_BACKREF7: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3715 case NFA_BACKREF8: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3716 case NFA_BACKREF9: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3717 /* \1 .. \9 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3718 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3719 int subidx = t->state->c - NFA_BACKREF1 + 1; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3720 int bytelen; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3721 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3722 result = match_backref(&t->sub, subidx, &bytelen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3723 if (result) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3724 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3725 if (bytelen == 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3726 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3727 /* empty match always works, add NFA_SKIP with zero to |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3728 * be used next */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3729 addstate_here(thislist, t->state->out, &t->sub, |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3730 &listidx); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3731 thislist->t[listidx + 1].count = 0; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3732 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3733 else if (bytelen <= clen) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3734 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3735 /* match current character, jump ahead to out of |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3736 * NFA_SKIP */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3737 addstate(nextlist, t->state->out->out, &t->sub, clen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3738 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3739 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3740 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3741 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3742 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3743 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3744 /* skip ofer the matched characters, set character |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3745 * count in NFA_SKIP */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3746 addstate(nextlist, t->state->out, &t->sub, bytelen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3747 nextlist->t[nextlist->n - 1].count = bytelen - clen; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3748 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3749 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3750 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3751 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3752 |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3753 } |
4482 | 3754 break; |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3755 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3756 case NFA_SKIP: |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3757 /* charater of previous matching \1 .. \9 */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3758 if (t->count - clen <= 0) |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3759 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3760 /* end of match, go to what follows */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3761 addstate(nextlist, t->state->out, &t->sub, clen); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3762 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3763 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3764 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3765 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3766 else |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3767 { |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3768 /* add state again with decremented count */ |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3769 addstate(nextlist, t->state, &t->sub, 0); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3770 nextlist->t[nextlist->n - 1].count = t->count - clen; |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3771 #ifdef ENABLE_LOG |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3772 log_subexpr(&nextlist->t[nextlist->n - 1].sub); |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3773 #endif |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3774 } |
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3775 break; |
4482 | 3776 |
3777 case NFA_SKIP_CHAR: | |
3778 case NFA_ZSTART: | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
3779 case NFA_ZEND: |
4482 | 3780 /* TODO: should not happen? */ |
3781 break; | |
3782 | |
4444 | 3783 default: /* regular character */ |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3784 { |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3785 int c = t->state->c; |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3786 |
4482 | 3787 /* TODO: put this in #ifdef later */ |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3788 if (c < -256) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3789 EMSGN("INTERNAL: Negative state char: %ld", c); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3790 if (is_Magic(c)) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3791 c = un_Magic(c); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3792 result = (c == curc); |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3793 |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3794 if (!result && ireg_ic) |
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3795 result = MB_TOLOWER(c) == MB_TOLOWER(curc); |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3796 #ifdef FEAT_MBYTE |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3797 /* If there is a composing character which is not being |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3798 * ignored there can be no match. Match with composing |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3799 * character uses NFA_COMPOSING above. */ |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3800 if (result && enc_utf8 && !ireg_icombine |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3801 && clen != utf_char2len(curc)) |
4527
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3802 result = FALSE; |
55bcaa1d2749
updated for version 7.3.1011
Bram Moolenaar <bram@vim.org>
parents:
4525
diff
changeset
|
3803 #endif |
4444 | 3804 ADD_POS_NEG_STATE(t->state); |
3805 break; | |
4559
04086e297563
updated for version 7.3.1027
Bram Moolenaar <bram@vim.org>
parents:
4557
diff
changeset
|
3806 } |
4444 | 3807 } |
3808 | |
3809 } /* for (thislist = thislist; thislist->state; thislist++) */ | |
3810 | |
4557
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3811 /* Look for the start of a match in the current position by adding the |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3812 * start state to the list of states. |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3813 * The first found match is the leftmost one, thus the order of states |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3814 * matters! |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3815 * Do not add the start state in recursive calls of nfa_regmatch(), |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3816 * because recursive calls should only start in the first position. |
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3817 * Also don't start a match past the first line. */ |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3818 if (nfa_match == FALSE && start->c == NFA_MOPEN + 0 |
4557
888c12c899e5
updated for version 7.3.1026
Bram Moolenaar <bram@vim.org>
parents:
4555
diff
changeset
|
3819 && reglnum == 0 && clen != 0) |
4444 | 3820 { |
3821 #ifdef ENABLE_LOG | |
3822 fprintf(log_fd, "(---) STARTSTATE\n"); | |
3823 #endif | |
4571
b2a8e3a66f8c
updated for version 7.3.1033
Bram Moolenaar <bram@vim.org>
parents:
4569
diff
changeset
|
3824 addstate(nextlist, start, m, clen); |
4444 | 3825 } |
3826 | |
3827 #ifdef ENABLE_LOG | |
3828 fprintf(log_fd, ">>> Thislist had %d states available: ", thislist->n); | |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3829 { |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3830 int i; |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3831 |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3832 for (i = 0; i < thislist->n; i++) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3833 fprintf(log_fd, "%d ", abs(thislist->t[i].state->id)); |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3834 } |
4444 | 3835 fprintf(log_fd, "\n"); |
3836 #endif | |
3837 | |
3838 nextchar: | |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3839 /* Advance to the next character, or advance to the next line, or |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3840 * finish. */ |
4549
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3841 if (clen != 0) |
849180347ac3
updated for version 7.3.1022
Bram Moolenaar <bram@vim.org>
parents:
4547
diff
changeset
|
3842 reginput += clen; |
4515
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3843 else if (go_to_nextline) |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3844 reg_nextline(); |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3845 else |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3846 break; |
90e9917d4114
updated for version 7.3.1005
Bram Moolenaar <bram@vim.org>
parents:
4507
diff
changeset
|
3847 } |
4444 | 3848 |
3849 #ifdef ENABLE_LOG | |
3850 if (log_fd != stderr) | |
3851 fclose(log_fd); | |
3852 log_fd = NULL; | |
3853 #endif | |
3854 | |
3855 theend: | |
3856 /* Free memory */ | |
3857 vim_free(list[0].t); | |
3858 vim_free(list[1].t); | |
3859 vim_free(list[2].t); | |
3860 list[0].t = list[1].t = list[2].t = NULL; | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3861 vim_free(listids); |
4444 | 3862 #undef ADD_POS_NEG_STATE |
4460 | 3863 #ifdef NFA_REGEXP_DEBUG_LOG |
4444 | 3864 fclose(debug); |
3865 #endif | |
3866 | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3867 return nfa_match; |
4444 | 3868 } |
3869 | |
3870 /* | |
3871 * Try match of "prog" with at regline["col"]. | |
3872 * Returns 0 for failure, number of lines contained in the match otherwise. | |
3873 */ | |
3874 static long | |
3875 nfa_regtry(start, col) | |
3876 nfa_state_T *start; | |
3877 colnr_T col; | |
3878 { | |
3879 int i; | |
3880 regsub_T sub, m; | |
3881 #ifdef ENABLE_LOG | |
3882 FILE *f; | |
3883 #endif | |
3884 | |
3885 reginput = regline + col; | |
3886 need_clear_subexpr = TRUE; | |
3887 | |
3888 #ifdef ENABLE_LOG | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
3889 f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 3890 if (f != NULL) |
3891 { | |
3892 fprintf(f, "\n\n\n\n\n\n\t\t=======================================================\n"); | |
3893 fprintf(f, " =======================================================\n"); | |
3894 #ifdef DEBUG | |
3895 fprintf(f, "\tRegexp is \"%s\"\n", nfa_regengine.expr); | |
3896 #endif | |
3897 fprintf(f, "\tInput text is \"%s\" \n", reginput); | |
3898 fprintf(f, " =======================================================\n\n\n\n\n\n\n"); | |
4533
6a2005efa02b
updated for version 7.3.1014
Bram Moolenaar <bram@vim.org>
parents:
4531
diff
changeset
|
3899 nfa_print_state(f, start); |
4444 | 3900 fprintf(f, "\n\n"); |
3901 fclose(f); | |
3902 } | |
3903 else | |
3904 EMSG(_("Could not open temporary log file for writing ")); | |
3905 #endif | |
3906 | |
3907 if (REG_MULTI) | |
3908 { | |
3909 /* Use 0xff to set lnum to -1 */ | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3910 vim_memset(sub.multilist, 0xff, sizeof(struct multipos) * nfa_nsubexpr); |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3911 vim_memset(m.multilist, 0xff, sizeof(struct multipos) * nfa_nsubexpr); |
4444 | 3912 } |
3913 else | |
3914 { | |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3915 vim_memset(sub.linelist, 0, sizeof(struct linepos) * nfa_nsubexpr); |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3916 vim_memset(m.linelist, 0, sizeof(struct linepos) * nfa_nsubexpr); |
4444 | 3917 } |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3918 sub.in_use = 0; |
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3919 m.in_use = 0; |
4444 | 3920 |
3921 if (nfa_regmatch(start, &sub, &m) == FALSE) | |
3922 return 0; | |
3923 | |
3924 cleanup_subexpr(); | |
3925 if (REG_MULTI) | |
3926 { | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3927 for (i = 0; i < sub.in_use; i++) |
4444 | 3928 { |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3929 reg_startpos[i] = sub.multilist[i].start; |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3930 reg_endpos[i] = sub.multilist[i].end; |
4444 | 3931 } |
3932 | |
3933 if (reg_startpos[0].lnum < 0) | |
3934 { | |
3935 reg_startpos[0].lnum = 0; | |
3936 reg_startpos[0].col = col; | |
3937 } | |
3938 if (reg_endpos[0].lnum < 0) | |
3939 { | |
4569
f262fb02889d
updated for version 7.3.1032
Bram Moolenaar <bram@vim.org>
parents:
4567
diff
changeset
|
3940 /* pattern has a \ze but it didn't match, use current end */ |
4444 | 3941 reg_endpos[0].lnum = reglnum; |
3942 reg_endpos[0].col = (int)(reginput - regline); | |
3943 } | |
3944 else | |
3945 /* Use line number of "\ze". */ | |
3946 reglnum = reg_endpos[0].lnum; | |
3947 } | |
3948 else | |
3949 { | |
4563
e7016af0cbf9
updated for version 7.3.1029
Bram Moolenaar <bram@vim.org>
parents:
4561
diff
changeset
|
3950 for (i = 0; i < sub.in_use; i++) |
4444 | 3951 { |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3952 reg_startp[i] = sub.linelist[i].start; |
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
3953 reg_endp[i] = sub.linelist[i].end; |
4444 | 3954 } |
3955 | |
3956 if (reg_startp[0] == NULL) | |
3957 reg_startp[0] = regline + col; | |
3958 if (reg_endp[0] == NULL) | |
3959 reg_endp[0] = reginput; | |
3960 } | |
3961 | |
3962 return 1 + reglnum; | |
3963 } | |
3964 | |
3965 /* | |
3966 * Match a regexp against a string ("line" points to the string) or multiple | |
3967 * lines ("line" is NULL, use reg_getline()). | |
3968 * | |
3969 * Returns 0 for failure, number of lines contained in the match otherwise. | |
3970 */ | |
3971 static long | |
3972 nfa_regexec_both(line, col) | |
3973 char_u *line; | |
3974 colnr_T col; /* column to start looking for match */ | |
3975 { | |
3976 nfa_regprog_T *prog; | |
3977 long retval = 0L; | |
3978 int i; | |
3979 | |
3980 if (REG_MULTI) | |
3981 { | |
3982 prog = (nfa_regprog_T *)reg_mmatch->regprog; | |
3983 line = reg_getline((linenr_T)0); /* relative to the cursor */ | |
3984 reg_startpos = reg_mmatch->startpos; | |
3985 reg_endpos = reg_mmatch->endpos; | |
3986 } | |
3987 else | |
3988 { | |
3989 prog = (nfa_regprog_T *)reg_match->regprog; | |
3990 reg_startp = reg_match->startp; | |
3991 reg_endp = reg_match->endp; | |
3992 } | |
3993 | |
3994 /* Be paranoid... */ | |
3995 if (prog == NULL || line == NULL) | |
3996 { | |
3997 EMSG(_(e_null)); | |
3998 goto theend; | |
3999 } | |
4000 | |
4001 /* If the start column is past the maximum column: no need to try. */ | |
4002 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
4003 goto theend; | |
4004 | |
4005 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ | |
4006 if (prog->regflags & RF_ICASE) | |
4007 ireg_ic = TRUE; | |
4008 else if (prog->regflags & RF_NOICASE) | |
4009 ireg_ic = FALSE; | |
4010 | |
4011 #ifdef FEAT_MBYTE | |
4012 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
4013 if (prog->regflags & RF_ICOMBINE) | |
4014 ireg_icombine = TRUE; | |
4015 #endif | |
4016 | |
4017 regline = line; | |
4018 reglnum = 0; /* relative to line */ | |
4019 | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4020 nfa_has_zend = prog->has_zend; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4021 nfa_nsubexpr = prog->nsubexp; |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4022 |
4444 | 4023 nstate = prog->nstate; |
4024 for (i = 0; i < nstate; ++i) | |
4025 { | |
4026 prog->state[i].id = i; | |
4027 prog->state[i].lastlist = 0; | |
4028 } | |
4029 | |
4030 retval = nfa_regtry(prog->start, col); | |
4031 | |
4032 theend: | |
4033 return retval; | |
4034 } | |
4035 | |
4036 /* | |
4037 * Compile a regular expression into internal code for the NFA matcher. | |
4038 * Returns the program in allocated space. Returns NULL for an error. | |
4039 */ | |
4040 static regprog_T * | |
4041 nfa_regcomp(expr, re_flags) | |
4042 char_u *expr; | |
4043 int re_flags; | |
4044 { | |
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4045 nfa_regprog_T *prog = NULL; |
4458 | 4046 size_t prog_size; |
4444 | 4047 int *postfix; |
4048 | |
4049 if (expr == NULL) | |
4050 return NULL; | |
4051 | |
4052 #ifdef DEBUG | |
4053 nfa_regengine.expr = expr; | |
4054 #endif | |
4055 | |
4056 init_class_tab(); | |
4057 | |
4058 if (nfa_regcomp_start(expr, re_flags) == FAIL) | |
4059 return NULL; | |
4060 | |
4061 /* Build postfix form of the regexp. Needed to build the NFA | |
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4062 * (and count its size). */ |
4444 | 4063 postfix = re2post(); |
4064 if (postfix == NULL) | |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4065 { |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4066 /* TODO: only give this error for debugging? */ |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4067 if (post_ptr >= post_end) |
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4068 EMSGN("Internal error: estimated max number of states insufficient: %ld", post_end - post_start); |
4444 | 4069 goto fail; /* Cascaded (syntax?) error */ |
4555
b2946c06d1b6
updated for version 7.3.1025
Bram Moolenaar <bram@vim.org>
parents:
4553
diff
changeset
|
4070 } |
4444 | 4071 |
4072 /* | |
4073 * In order to build the NFA, we parse the input regexp twice: | |
4074 * 1. first pass to count size (so we can allocate space) | |
4075 * 2. second to emit code | |
4076 */ | |
4077 #ifdef ENABLE_LOG | |
4078 { | |
4531
1be43c095aff
updated for version 7.3.1013
Bram Moolenaar <bram@vim.org>
parents:
4529
diff
changeset
|
4079 FILE *f = fopen(NFA_REGEXP_RUN_LOG, "a"); |
4444 | 4080 |
4081 if (f != NULL) | |
4082 { | |
4083 fprintf(f, "\n*****************************\n\n\n\n\tCompiling regexp \"%s\" ... hold on !\n", expr); | |
4084 fclose(f); | |
4085 } | |
4086 } | |
4087 #endif | |
4088 | |
4089 /* | |
4090 * PASS 1 | |
4091 * Count number of NFA states in "nstate". Do not build the NFA. | |
4092 */ | |
4093 post2nfa(postfix, post_ptr, TRUE); | |
4541
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4094 |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4095 /* Space for compiled regexp */ |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4096 prog_size = sizeof(nfa_regprog_T) + sizeof(nfa_state_T) * nstate; |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4097 prog = (nfa_regprog_T *)lalloc(prog_size, TRUE); |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4098 if (prog == NULL) |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4099 goto fail; |
80170d61a85c
updated for version 7.3.1018
Bram Moolenaar <bram@vim.org>
parents:
4539
diff
changeset
|
4100 vim_memset(prog, 0, prog_size); |
4444 | 4101 state_ptr = prog->state; |
4102 | |
4103 /* | |
4104 * PASS 2 | |
4105 * Build the NFA | |
4106 */ | |
4107 prog->start = post2nfa(postfix, post_ptr, FALSE); | |
4108 if (prog->start == NULL) | |
4109 goto fail; | |
4110 | |
4111 prog->regflags = regflags; | |
4112 prog->engine = &nfa_regengine; | |
4113 prog->nstate = nstate; | |
4553
7b835b2969af
updated for version 7.3.1024
Bram Moolenaar <bram@vim.org>
parents:
4551
diff
changeset
|
4114 prog->has_zend = nfa_has_zend; |
4561
4d81fdda8f35
updated for version 7.3.1028
Bram Moolenaar <bram@vim.org>
parents:
4559
diff
changeset
|
4115 prog->nsubexp = regnpar; |
4444 | 4116 #ifdef ENABLE_LOG |
4117 nfa_postfix_dump(expr, OK); | |
4118 nfa_dump(prog); | |
4119 #endif | |
4120 | |
4121 out: | |
4122 vim_free(post_start); | |
4123 post_start = post_ptr = post_end = NULL; | |
4124 state_ptr = NULL; | |
4125 return (regprog_T *)prog; | |
4126 | |
4127 fail: | |
4128 vim_free(prog); | |
4129 prog = NULL; | |
4130 #ifdef ENABLE_LOG | |
4131 nfa_postfix_dump(expr, FAIL); | |
4132 #endif | |
4133 #ifdef DEBUG | |
4134 nfa_regengine.expr = NULL; | |
4135 #endif | |
4136 goto out; | |
4137 } | |
4138 | |
4139 | |
4140 /* | |
4141 * Match a regexp against a string. | |
4142 * "rmp->regprog" is a compiled regexp as returned by nfa_regcomp(). | |
4143 * Uses curbuf for line count and 'iskeyword'. | |
4144 * | |
4145 * Return TRUE if there is a match, FALSE if not. | |
4146 */ | |
4147 static int | |
4148 nfa_regexec(rmp, line, col) | |
4149 regmatch_T *rmp; | |
4150 char_u *line; /* string to match against */ | |
4151 colnr_T col; /* column to start looking for match */ | |
4152 { | |
4153 reg_match = rmp; | |
4154 reg_mmatch = NULL; | |
4155 reg_maxline = 0; | |
4156 reg_line_lbr = FALSE; | |
4157 reg_buf = curbuf; | |
4158 reg_win = NULL; | |
4159 ireg_ic = rmp->rm_ic; | |
4160 #ifdef FEAT_MBYTE | |
4161 ireg_icombine = FALSE; | |
4162 #endif | |
4163 ireg_maxcol = 0; | |
4164 return (nfa_regexec_both(line, col) != 0); | |
4165 } | |
4166 | |
4167 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
4168 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
4169 | |
4170 static int nfa_regexec_nl __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
4171 | |
4172 /* | |
4173 * Like nfa_regexec(), but consider a "\n" in "line" to be a line break. | |
4174 */ | |
4175 static int | |
4176 nfa_regexec_nl(rmp, line, col) | |
4177 regmatch_T *rmp; | |
4178 char_u *line; /* string to match against */ | |
4179 colnr_T col; /* column to start looking for match */ | |
4180 { | |
4181 reg_match = rmp; | |
4182 reg_mmatch = NULL; | |
4183 reg_maxline = 0; | |
4184 reg_line_lbr = TRUE; | |
4185 reg_buf = curbuf; | |
4186 reg_win = NULL; | |
4187 ireg_ic = rmp->rm_ic; | |
4188 #ifdef FEAT_MBYTE | |
4189 ireg_icombine = FALSE; | |
4190 #endif | |
4191 ireg_maxcol = 0; | |
4192 return (nfa_regexec_both(line, col) != 0); | |
4193 } | |
4194 #endif | |
4195 | |
4196 | |
4197 /* | |
4198 * Match a regexp against multiple lines. | |
4199 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
4200 * Uses curbuf for line count and 'iskeyword'. | |
4201 * | |
4202 * Return zero if there is no match. Return number of lines contained in the | |
4203 * match otherwise. | |
4204 * | |
4205 * Note: the body is the same as bt_regexec() except for nfa_regexec_both() | |
4206 * | |
4207 * ! Also NOTE : match may actually be in another line. e.g.: | |
4208 * when r.e. is \nc, cursor is at 'a' and the text buffer looks like | |
4209 * | |
4210 * +-------------------------+ | |
4211 * |a | | |
4212 * |b | | |
4213 * |c | | |
4214 * | | | |
4215 * +-------------------------+ | |
4216 * | |
4217 * then nfa_regexec_multi() returns 3. while the original | |
4218 * vim_regexec_multi() returns 0 and a second call at line 2 will return 2. | |
4219 * | |
4220 * FIXME if this behavior is not compatible. | |
4221 */ | |
4222 static long | |
4223 nfa_regexec_multi(rmp, win, buf, lnum, col, tm) | |
4224 regmmatch_T *rmp; | |
4225 win_T *win; /* window in which to search or NULL */ | |
4226 buf_T *buf; /* buffer in which to search */ | |
4227 linenr_T lnum; /* nr of line to start looking for match */ | |
4228 colnr_T col; /* column to start looking for match */ | |
4229 proftime_T *tm UNUSED; /* timeout limit or NULL */ | |
4230 { | |
4231 reg_match = NULL; | |
4232 reg_mmatch = rmp; | |
4233 reg_buf = buf; | |
4234 reg_win = win; | |
4235 reg_firstlnum = lnum; | |
4236 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
4237 reg_line_lbr = FALSE; | |
4238 ireg_ic = rmp->rmm_ic; | |
4239 #ifdef FEAT_MBYTE | |
4240 ireg_icombine = FALSE; | |
4241 #endif | |
4242 ireg_maxcol = rmp->rmm_maxcol; | |
4243 | |
4503 | 4244 return nfa_regexec_both(NULL, col); |
4444 | 4245 } |
4246 | |
4247 #ifdef DEBUG | |
4248 # undef ENABLE_LOG | |
4249 #endif |