Mercurial > vim
annotate src/regexp.c @ 6158:353442863d85 v7.4.415
Update version number to 7.4.415
author | Bram Moolenaar <bram@vim.org> |
---|---|
date | Sat, 23 Aug 2014 14:18:52 +0200 |
parents | 761687c6808f |
children | 8515b42f939c |
rev | line source |
---|---|
7 | 1 /* vi:set ts=8 sts=4 sw=4: |
2 * | |
3 * Handling of regular expressions: vim_regcomp(), vim_regexec(), vim_regsub() | |
4 * | |
5 * NOTICE: | |
6 * | |
7 * This is NOT the original regular expression code as written by Henry | |
8 * Spencer. This code has been modified specifically for use with the VIM | |
9 * editor, and should not be used separately from Vim. If you want a good | |
10 * regular expression library, get the original code. The copyright notice | |
11 * that follows is from the original. | |
12 * | |
13 * END NOTICE | |
14 * | |
15 * Copyright (c) 1986 by University of Toronto. | |
16 * Written by Henry Spencer. Not derived from licensed software. | |
17 * | |
18 * Permission is granted to anyone to use this software for any | |
19 * purpose on any computer system, and to redistribute it freely, | |
20 * subject to the following restrictions: | |
21 * | |
22 * 1. The author is not responsible for the consequences of use of | |
23 * this software, no matter how awful, even if they arise | |
24 * from defects in it. | |
25 * | |
26 * 2. The origin of this software must not be misrepresented, either | |
27 * by explicit claim or by omission. | |
28 * | |
29 * 3. Altered versions must be plainly marked as such, and must not | |
30 * be misrepresented as being the original software. | |
31 * | |
32 * Beware that some of this code is subtly aware of the way operator | |
33 * precedence is structured in regular expressions. Serious changes in | |
34 * regular-expression syntax might require a total rethink. | |
35 * | |
24 | 36 * Changes have been made by Tony Andrews, Olaf 'Rhialto' Seibert, Robert |
37 * Webb, Ciaran McCreesh and Bram Moolenaar. | |
7 | 38 * Named character class support added by Walter Briscoe (1998 Jul 01) |
39 */ | |
40 | |
4444 | 41 /* Uncomment the first if you do not want to see debugging logs or files |
42 * related to regular expressions, even when compiling with -DDEBUG. | |
43 * Uncomment the second to get the regexp debugging. */ | |
44 /* #undef DEBUG */ | |
45 /* #define DEBUG */ | |
46 | |
7 | 47 #include "vim.h" |
48 | |
4444 | 49 #ifdef DEBUG |
50 /* show/save debugging data when BT engine is used */ | |
51 # define BT_REGEXP_DUMP | |
52 /* save the debugging data to a file instead of displaying it */ | |
53 # define BT_REGEXP_LOG | |
4460 | 54 # define BT_REGEXP_DEBUG_LOG |
55 # define BT_REGEXP_DEBUG_LOG_NAME "bt_regexp_debug.log" | |
4444 | 56 #endif |
7 | 57 |
58 /* | |
59 * The "internal use only" fields in regexp.h are present to pass info from | |
60 * compile to execute that permits the execute phase to run lots faster on | |
61 * simple cases. They are: | |
62 * | |
63 * regstart char that must begin a match; NUL if none obvious; Can be a | |
64 * multi-byte character. | |
65 * reganch is the match anchored (at beginning-of-line only)? | |
66 * regmust string (pointer into program) that match must include, or NULL | |
67 * regmlen length of regmust string | |
68 * regflags RF_ values or'ed together | |
69 * | |
70 * Regstart and reganch permit very fast decisions on suitable starting points | |
71 * for a match, cutting down the work a lot. Regmust permits fast rejection | |
72 * of lines that cannot possibly match. The regmust tests are costly enough | |
73 * that vim_regcomp() supplies a regmust only if the r.e. contains something | |
74 * potentially expensive (at present, the only such thing detected is * or + | |
75 * at the start of the r.e., which can involve a lot of backup). Regmlen is | |
76 * supplied because the test in vim_regexec() needs it and vim_regcomp() is | |
77 * computing it anyway. | |
78 */ | |
79 | |
80 /* | |
81 * Structure for regexp "program". This is essentially a linear encoding | |
82 * of a nondeterministic finite-state machine (aka syntax charts or | |
83 * "railroad normal form" in parsing technology). Each node is an opcode | |
84 * plus a "next" pointer, possibly plus an operand. "Next" pointers of | |
85 * all nodes except BRANCH and BRACES_COMPLEX implement concatenation; a "next" | |
86 * pointer with a BRANCH on both ends of it is connecting two alternatives. | |
87 * (Here we have one of the subtle syntax dependencies: an individual BRANCH | |
88 * (as opposed to a collection of them) is never concatenated with anything | |
89 * because of operator precedence). The "next" pointer of a BRACES_COMPLEX | |
167 | 90 * node points to the node after the stuff to be repeated. |
91 * The operand of some types of node is a literal string; for others, it is a | |
92 * node leading into a sub-FSM. In particular, the operand of a BRANCH node | |
93 * is the first node of the branch. | |
94 * (NB this is *not* a tree structure: the tail of the branch connects to the | |
95 * thing following the set of BRANCHes.) | |
7 | 96 * |
97 * pattern is coded like: | |
98 * | |
99 * +-----------------+ | |
100 * | V | |
101 * <aa>\|<bb> BRANCH <aa> BRANCH <bb> --> END | |
102 * | ^ | ^ | |
103 * +------+ +----------+ | |
104 * | |
105 * | |
106 * +------------------+ | |
107 * V | | |
108 * <aa>* BRANCH BRANCH <aa> --> BACK BRANCH --> NOTHING --> END | |
109 * | | ^ ^ | |
110 * | +---------------+ | | |
111 * +---------------------------------------------+ | |
112 * | |
113 * | |
167 | 114 * +----------------------+ |
115 * V | | |
233 | 116 * <aa>\+ BRANCH <aa> --> BRANCH --> BACK BRANCH --> NOTHING --> END |
856 | 117 * | | ^ ^ |
118 * | +-----------+ | | |
179 | 119 * +--------------------------------------------------+ |
167 | 120 * |
121 * | |
7 | 122 * +-------------------------+ |
123 * V | | |
124 * <aa>\{} BRANCH BRACE_LIMITS --> BRACE_COMPLEX <aa> --> BACK END | |
125 * | | ^ | |
126 * | +----------------+ | |
127 * +-----------------------------------------------+ | |
128 * | |
129 * | |
130 * <aa>\@!<bb> BRANCH NOMATCH <aa> --> END <bb> --> END | |
131 * | | ^ ^ | |
132 * | +----------------+ | | |
133 * +--------------------------------+ | |
134 * | |
135 * +---------+ | |
136 * | V | |
137 * \z[abc] BRANCH BRANCH a BRANCH b BRANCH c BRANCH NOTHING --> END | |
138 * | | | | ^ ^ | |
139 * | | | +-----+ | | |
140 * | | +----------------+ | | |
141 * | +---------------------------+ | | |
142 * +------------------------------------------------------+ | |
143 * | |
1209 | 144 * They all start with a BRANCH for "\|" alternatives, even when there is only |
7 | 145 * one alternative. |
146 */ | |
147 | |
148 /* | |
149 * The opcodes are: | |
150 */ | |
151 | |
152 /* definition number opnd? meaning */ | |
153 #define END 0 /* End of program or NOMATCH operand. */ | |
154 #define BOL 1 /* Match "" at beginning of line. */ | |
155 #define EOL 2 /* Match "" at end of line. */ | |
156 #define BRANCH 3 /* node Match this alternative, or the | |
157 * next... */ | |
158 #define BACK 4 /* Match "", "next" ptr points backward. */ | |
159 #define EXACTLY 5 /* str Match this string. */ | |
160 #define NOTHING 6 /* Match empty string. */ | |
161 #define STAR 7 /* node Match this (simple) thing 0 or more | |
162 * times. */ | |
163 #define PLUS 8 /* node Match this (simple) thing 1 or more | |
164 * times. */ | |
165 #define MATCH 9 /* node match the operand zero-width */ | |
166 #define NOMATCH 10 /* node check for no match with operand */ | |
167 #define BEHIND 11 /* node look behind for a match with operand */ | |
168 #define NOBEHIND 12 /* node look behind for no match with operand */ | |
169 #define SUBPAT 13 /* node match the operand here */ | |
170 #define BRACE_SIMPLE 14 /* node Match this (simple) thing between m and | |
171 * n times (\{m,n\}). */ | |
172 #define BOW 15 /* Match "" after [^a-zA-Z0-9_] */ | |
173 #define EOW 16 /* Match "" at [^a-zA-Z0-9_] */ | |
174 #define BRACE_LIMITS 17 /* nr nr define the min & max for BRACE_SIMPLE | |
175 * and BRACE_COMPLEX. */ | |
176 #define NEWL 18 /* Match line-break */ | |
177 #define BHPOS 19 /* End position for BEHIND or NOBEHIND */ | |
178 | |
179 | |
180 /* character classes: 20-48 normal, 50-78 include a line-break */ | |
181 #define ADD_NL 30 | |
182 #define FIRST_NL ANY + ADD_NL | |
183 #define ANY 20 /* Match any one character. */ | |
184 #define ANYOF 21 /* str Match any character in this string. */ | |
185 #define ANYBUT 22 /* str Match any character not in this | |
186 * string. */ | |
187 #define IDENT 23 /* Match identifier char */ | |
188 #define SIDENT 24 /* Match identifier char but no digit */ | |
189 #define KWORD 25 /* Match keyword char */ | |
190 #define SKWORD 26 /* Match word char but no digit */ | |
191 #define FNAME 27 /* Match file name char */ | |
192 #define SFNAME 28 /* Match file name char but no digit */ | |
193 #define PRINT 29 /* Match printable char */ | |
194 #define SPRINT 30 /* Match printable char but no digit */ | |
195 #define WHITE 31 /* Match whitespace char */ | |
196 #define NWHITE 32 /* Match non-whitespace char */ | |
197 #define DIGIT 33 /* Match digit char */ | |
198 #define NDIGIT 34 /* Match non-digit char */ | |
199 #define HEX 35 /* Match hex char */ | |
200 #define NHEX 36 /* Match non-hex char */ | |
201 #define OCTAL 37 /* Match octal char */ | |
202 #define NOCTAL 38 /* Match non-octal char */ | |
203 #define WORD 39 /* Match word char */ | |
204 #define NWORD 40 /* Match non-word char */ | |
205 #define HEAD 41 /* Match head char */ | |
206 #define NHEAD 42 /* Match non-head char */ | |
207 #define ALPHA 43 /* Match alpha char */ | |
208 #define NALPHA 44 /* Match non-alpha char */ | |
209 #define LOWER 45 /* Match lowercase char */ | |
210 #define NLOWER 46 /* Match non-lowercase char */ | |
211 #define UPPER 47 /* Match uppercase char */ | |
212 #define NUPPER 48 /* Match non-uppercase char */ | |
213 #define LAST_NL NUPPER + ADD_NL | |
214 #define WITH_NL(op) ((op) >= FIRST_NL && (op) <= LAST_NL) | |
215 | |
216 #define MOPEN 80 /* -89 Mark this point in input as start of | |
217 * \( subexpr. MOPEN + 0 marks start of | |
218 * match. */ | |
219 #define MCLOSE 90 /* -99 Analogous to MOPEN. MCLOSE + 0 marks | |
220 * end of match. */ | |
221 #define BACKREF 100 /* -109 node Match same string again \1-\9 */ | |
222 | |
223 #ifdef FEAT_SYN_HL | |
224 # define ZOPEN 110 /* -119 Mark this point in input as start of | |
225 * \z( subexpr. */ | |
226 # define ZCLOSE 120 /* -129 Analogous to ZOPEN. */ | |
227 # define ZREF 130 /* -139 node Match external submatch \z1-\z9 */ | |
228 #endif | |
229 | |
230 #define BRACE_COMPLEX 140 /* -149 node Match nodes between m & n times */ | |
231 | |
232 #define NOPEN 150 /* Mark this point in input as start of | |
233 \%( subexpr. */ | |
234 #define NCLOSE 151 /* Analogous to NOPEN. */ | |
235 | |
236 #define MULTIBYTECODE 200 /* mbc Match one multi-byte character */ | |
237 #define RE_BOF 201 /* Match "" at beginning of file. */ | |
238 #define RE_EOF 202 /* Match "" at end of file. */ | |
239 #define CURSOR 203 /* Match location of cursor. */ | |
240 | |
241 #define RE_LNUM 204 /* nr cmp Match line number */ | |
242 #define RE_COL 205 /* nr cmp Match column number */ | |
243 #define RE_VCOL 206 /* nr cmp Match virtual column number */ | |
244 | |
639 | 245 #define RE_MARK 207 /* mark cmp Match mark position */ |
246 #define RE_VISUAL 208 /* Match Visual area */ | |
5901 | 247 #define RE_COMPOSING 209 /* any composing characters */ |
639 | 248 |
7 | 249 /* |
250 * Magic characters have a special meaning, they don't match literally. | |
251 * Magic characters are negative. This separates them from literal characters | |
252 * (possibly multi-byte). Only ASCII characters can be Magic. | |
253 */ | |
254 #define Magic(x) ((int)(x) - 256) | |
255 #define un_Magic(x) ((x) + 256) | |
256 #define is_Magic(x) ((x) < 0) | |
257 | |
258 static int no_Magic __ARGS((int x)); | |
259 static int toggle_Magic __ARGS((int x)); | |
260 | |
261 static int | |
262 no_Magic(x) | |
263 int x; | |
264 { | |
265 if (is_Magic(x)) | |
266 return un_Magic(x); | |
267 return x; | |
268 } | |
269 | |
270 static int | |
271 toggle_Magic(x) | |
272 int x; | |
273 { | |
274 if (is_Magic(x)) | |
275 return un_Magic(x); | |
276 return Magic(x); | |
277 } | |
278 | |
279 /* | |
280 * The first byte of the regexp internal "program" is actually this magic | |
281 * number; the start node begins in the second byte. It's used to catch the | |
282 * most severe mutilation of the program by the caller. | |
283 */ | |
284 | |
285 #define REGMAGIC 0234 | |
286 | |
287 /* | |
288 * Opcode notes: | |
289 * | |
290 * BRANCH The set of branches constituting a single choice are hooked | |
291 * together with their "next" pointers, since precedence prevents | |
292 * anything being concatenated to any individual branch. The | |
293 * "next" pointer of the last BRANCH in a choice points to the | |
294 * thing following the whole choice. This is also where the | |
295 * final "next" pointer of each individual branch points; each | |
296 * branch starts with the operand node of a BRANCH node. | |
297 * | |
298 * BACK Normal "next" pointers all implicitly point forward; BACK | |
299 * exists to make loop structures possible. | |
300 * | |
301 * STAR,PLUS '=', and complex '*' and '+', are implemented as circular | |
302 * BRANCH structures using BACK. Simple cases (one character | |
303 * per match) are implemented with STAR and PLUS for speed | |
304 * and to minimize recursive plunges. | |
305 * | |
306 * BRACE_LIMITS This is always followed by a BRACE_SIMPLE or BRACE_COMPLEX | |
307 * node, and defines the min and max limits to be used for that | |
308 * node. | |
309 * | |
310 * MOPEN,MCLOSE ...are numbered at compile time. | |
311 * ZOPEN,ZCLOSE ...ditto | |
312 */ | |
313 | |
314 /* | |
315 * A node is one char of opcode followed by two chars of "next" pointer. | |
316 * "Next" pointers are stored as two 8-bit bytes, high order first. The | |
317 * value is a positive offset from the opcode of the node containing it. | |
318 * An operand, if any, simply follows the node. (Note that much of the | |
319 * code generation knows about this implicit relationship.) | |
320 * | |
321 * Using two bytes for the "next" pointer is vast overkill for most things, | |
322 * but allows patterns to get big without disasters. | |
323 */ | |
324 #define OP(p) ((int)*(p)) | |
325 #define NEXT(p) (((*((p) + 1) & 0377) << 8) + (*((p) + 2) & 0377)) | |
326 #define OPERAND(p) ((p) + 3) | |
327 /* Obtain an operand that was stored as four bytes, MSB first. */ | |
328 #define OPERAND_MIN(p) (((long)(p)[3] << 24) + ((long)(p)[4] << 16) \ | |
329 + ((long)(p)[5] << 8) + (long)(p)[6]) | |
330 /* Obtain a second operand stored as four bytes. */ | |
331 #define OPERAND_MAX(p) OPERAND_MIN((p) + 4) | |
332 /* Obtain a second single-byte operand stored after a four bytes operand. */ | |
333 #define OPERAND_CMP(p) (p)[7] | |
334 | |
335 /* | |
336 * Utility definitions. | |
337 */ | |
338 #define UCHARAT(p) ((int)*(char_u *)(p)) | |
339 | |
340 /* Used for an error (down from) vim_regcomp(): give the error message, set | |
341 * rc_did_emsg and return NULL */ | |
653 | 342 #define EMSG_RET_NULL(m) return (EMSG(m), rc_did_emsg = TRUE, (void *)NULL) |
308 | 343 #define EMSG_RET_FAIL(m) return (EMSG(m), rc_did_emsg = TRUE, FAIL) |
4444 | 344 #define EMSG2_RET_NULL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, (void *)NULL) |
345 #define EMSG2_RET_FAIL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, FAIL) | |
346 #define EMSG_ONE_RET_NULL EMSG2_RET_NULL(_("E369: invalid item in %s%%[]"), reg_magic == MAGIC_ALL) | |
7 | 347 |
348 #define MAX_LIMIT (32767L << 16L) | |
349 | |
350 static int re_multi_type __ARGS((int)); | |
351 static int cstrncmp __ARGS((char_u *s1, char_u *s2, int *n)); | |
352 static char_u *cstrchr __ARGS((char_u *, int)); | |
353 | |
4444 | 354 #ifdef BT_REGEXP_DUMP |
355 static void regdump __ARGS((char_u *, bt_regprog_T *)); | |
356 #endif | |
7 | 357 #ifdef DEBUG |
358 static char_u *regprop __ARGS((char_u *)); | |
359 #endif | |
360 | |
4444 | 361 static char_u e_missingbracket[] = N_("E769: Missing ] after %s["); |
362 static char_u e_unmatchedpp[] = N_("E53: Unmatched %s%%("); | |
363 static char_u e_unmatchedp[] = N_("E54: Unmatched %s("); | |
364 static char_u e_unmatchedpar[] = N_("E55: Unmatched %s)"); | |
4720
bd6bef0bd0fb
updated for version 7.3.1107
Bram Moolenaar <bram@vim.org>
parents:
4688
diff
changeset
|
365 #ifdef FEAT_SYN_HL |
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
366 static char_u e_z_not_allowed[] = N_("E66: \\z( not allowed here"); |
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
367 static char_u e_z1_not_allowed[] = N_("E67: \\z1 et al. not allowed here"); |
4720
bd6bef0bd0fb
updated for version 7.3.1107
Bram Moolenaar <bram@vim.org>
parents:
4688
diff
changeset
|
368 #endif |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
369 static char_u e_missing_sb[] = N_("E69: Missing ] after %s%%["); |
4760
532a9855bd30
updated for version 7.3.1127
Bram Moolenaar <bram@vim.org>
parents:
4746
diff
changeset
|
370 static char_u e_empty_sb[] = N_("E70: Empty %s%%[]"); |
7 | 371 #define NOT_MULTI 0 |
372 #define MULTI_ONE 1 | |
373 #define MULTI_MULT 2 | |
374 /* | |
375 * Return NOT_MULTI if c is not a "multi" operator. | |
376 * Return MULTI_ONE if c is a single "multi" operator. | |
377 * Return MULTI_MULT if c is a multi "multi" operator. | |
378 */ | |
379 static int | |
380 re_multi_type(c) | |
381 int c; | |
382 { | |
383 if (c == Magic('@') || c == Magic('=') || c == Magic('?')) | |
384 return MULTI_ONE; | |
385 if (c == Magic('*') || c == Magic('+') || c == Magic('{')) | |
386 return MULTI_MULT; | |
387 return NOT_MULTI; | |
388 } | |
389 | |
390 /* | |
391 * Flags to be passed up and down. | |
392 */ | |
393 #define HASWIDTH 0x1 /* Known never to match null string. */ | |
394 #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ | |
395 #define SPSTART 0x4 /* Starts with * or +. */ | |
396 #define HASNL 0x8 /* Contains some \n. */ | |
397 #define HASLOOKBH 0x10 /* Contains "\@<=" or "\@<!". */ | |
398 #define WORST 0 /* Worst case. */ | |
399 | |
400 /* | |
401 * When regcode is set to this value, code is not emitted and size is computed | |
402 * instead. | |
403 */ | |
404 #define JUST_CALC_SIZE ((char_u *) -1) | |
405 | |
359 | 406 static char_u *reg_prev_sub = NULL; |
407 | |
7 | 408 /* |
409 * REGEXP_INRANGE contains all characters which are always special in a [] | |
410 * range after '\'. | |
411 * REGEXP_ABBR contains all characters which act as abbreviations after '\'. | |
412 * These are: | |
413 * \n - New line (NL). | |
414 * \r - Carriage Return (CR). | |
415 * \t - Tab (TAB). | |
416 * \e - Escape (ESC). | |
417 * \b - Backspace (Ctrl_H). | |
24 | 418 * \d - Character code in decimal, eg \d123 |
419 * \o - Character code in octal, eg \o80 | |
420 * \x - Character code in hex, eg \x4a | |
421 * \u - Multibyte character code, eg \u20ac | |
422 * \U - Long multibyte character code, eg \U12345678 | |
7 | 423 */ |
424 static char_u REGEXP_INRANGE[] = "]^-n\\"; | |
24 | 425 static char_u REGEXP_ABBR[] = "nrtebdoxuU"; |
7 | 426 |
427 static int backslash_trans __ARGS((int c)); | |
167 | 428 static int get_char_class __ARGS((char_u **pp)); |
429 static int get_equi_class __ARGS((char_u **pp)); | |
430 static void reg_equi_class __ARGS((int c)); | |
431 static int get_coll_element __ARGS((char_u **pp)); | |
7 | 432 static char_u *skip_anyof __ARGS((char_u *p)); |
433 static void init_class_tab __ARGS((void)); | |
434 | |
435 /* | |
436 * Translate '\x' to its control character, except "\n", which is Magic. | |
437 */ | |
438 static int | |
439 backslash_trans(c) | |
440 int c; | |
441 { | |
442 switch (c) | |
443 { | |
444 case 'r': return CAR; | |
445 case 't': return TAB; | |
446 case 'e': return ESC; | |
447 case 'b': return BS; | |
448 } | |
449 return c; | |
450 } | |
451 | |
452 /* | |
167 | 453 * Check for a character class name "[:name:]". "pp" points to the '['. |
7 | 454 * Returns one of the CLASS_ items. CLASS_NONE means that no item was |
455 * recognized. Otherwise "pp" is advanced to after the item. | |
456 */ | |
457 static int | |
167 | 458 get_char_class(pp) |
7 | 459 char_u **pp; |
460 { | |
461 static const char *(class_names[]) = | |
462 { | |
463 "alnum:]", | |
464 #define CLASS_ALNUM 0 | |
465 "alpha:]", | |
466 #define CLASS_ALPHA 1 | |
467 "blank:]", | |
468 #define CLASS_BLANK 2 | |
469 "cntrl:]", | |
470 #define CLASS_CNTRL 3 | |
471 "digit:]", | |
472 #define CLASS_DIGIT 4 | |
473 "graph:]", | |
474 #define CLASS_GRAPH 5 | |
475 "lower:]", | |
476 #define CLASS_LOWER 6 | |
477 "print:]", | |
478 #define CLASS_PRINT 7 | |
479 "punct:]", | |
480 #define CLASS_PUNCT 8 | |
481 "space:]", | |
482 #define CLASS_SPACE 9 | |
483 "upper:]", | |
484 #define CLASS_UPPER 10 | |
485 "xdigit:]", | |
486 #define CLASS_XDIGIT 11 | |
487 "tab:]", | |
488 #define CLASS_TAB 12 | |
489 "return:]", | |
490 #define CLASS_RETURN 13 | |
491 "backspace:]", | |
492 #define CLASS_BACKSPACE 14 | |
493 "escape:]", | |
494 #define CLASS_ESCAPE 15 | |
495 }; | |
496 #define CLASS_NONE 99 | |
497 int i; | |
498 | |
499 if ((*pp)[1] == ':') | |
500 { | |
1877 | 501 for (i = 0; i < (int)(sizeof(class_names) / sizeof(*class_names)); ++i) |
7 | 502 if (STRNCMP(*pp + 2, class_names[i], STRLEN(class_names[i])) == 0) |
503 { | |
504 *pp += STRLEN(class_names[i]) + 2; | |
505 return i; | |
506 } | |
507 } | |
508 return CLASS_NONE; | |
509 } | |
510 | |
511 /* | |
512 * Specific version of character class functions. | |
513 * Using a table to keep this fast. | |
514 */ | |
515 static short class_tab[256]; | |
516 | |
517 #define RI_DIGIT 0x01 | |
518 #define RI_HEX 0x02 | |
519 #define RI_OCTAL 0x04 | |
520 #define RI_WORD 0x08 | |
521 #define RI_HEAD 0x10 | |
522 #define RI_ALPHA 0x20 | |
523 #define RI_LOWER 0x40 | |
524 #define RI_UPPER 0x80 | |
525 #define RI_WHITE 0x100 | |
526 | |
527 static void | |
528 init_class_tab() | |
529 { | |
530 int i; | |
531 static int done = FALSE; | |
532 | |
533 if (done) | |
534 return; | |
535 | |
536 for (i = 0; i < 256; ++i) | |
537 { | |
538 if (i >= '0' && i <= '7') | |
539 class_tab[i] = RI_DIGIT + RI_HEX + RI_OCTAL + RI_WORD; | |
540 else if (i >= '8' && i <= '9') | |
541 class_tab[i] = RI_DIGIT + RI_HEX + RI_WORD; | |
542 else if (i >= 'a' && i <= 'f') | |
543 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
544 #ifdef EBCDIC | |
545 else if ((i >= 'g' && i <= 'i') || (i >= 'j' && i <= 'r') | |
546 || (i >= 's' && i <= 'z')) | |
547 #else | |
548 else if (i >= 'g' && i <= 'z') | |
549 #endif | |
550 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
551 else if (i >= 'A' && i <= 'F') | |
552 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
553 #ifdef EBCDIC | |
554 else if ((i >= 'G' && i <= 'I') || ( i >= 'J' && i <= 'R') | |
555 || (i >= 'S' && i <= 'Z')) | |
556 #else | |
557 else if (i >= 'G' && i <= 'Z') | |
558 #endif | |
559 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
560 else if (i == '_') | |
561 class_tab[i] = RI_WORD + RI_HEAD; | |
562 else | |
563 class_tab[i] = 0; | |
564 } | |
565 class_tab[' '] |= RI_WHITE; | |
566 class_tab['\t'] |= RI_WHITE; | |
567 done = TRUE; | |
568 } | |
569 | |
570 #ifdef FEAT_MBYTE | |
571 # define ri_digit(c) (c < 0x100 && (class_tab[c] & RI_DIGIT)) | |
572 # define ri_hex(c) (c < 0x100 && (class_tab[c] & RI_HEX)) | |
573 # define ri_octal(c) (c < 0x100 && (class_tab[c] & RI_OCTAL)) | |
574 # define ri_word(c) (c < 0x100 && (class_tab[c] & RI_WORD)) | |
575 # define ri_head(c) (c < 0x100 && (class_tab[c] & RI_HEAD)) | |
576 # define ri_alpha(c) (c < 0x100 && (class_tab[c] & RI_ALPHA)) | |
577 # define ri_lower(c) (c < 0x100 && (class_tab[c] & RI_LOWER)) | |
578 # define ri_upper(c) (c < 0x100 && (class_tab[c] & RI_UPPER)) | |
579 # define ri_white(c) (c < 0x100 && (class_tab[c] & RI_WHITE)) | |
580 #else | |
581 # define ri_digit(c) (class_tab[c] & RI_DIGIT) | |
582 # define ri_hex(c) (class_tab[c] & RI_HEX) | |
583 # define ri_octal(c) (class_tab[c] & RI_OCTAL) | |
584 # define ri_word(c) (class_tab[c] & RI_WORD) | |
585 # define ri_head(c) (class_tab[c] & RI_HEAD) | |
586 # define ri_alpha(c) (class_tab[c] & RI_ALPHA) | |
587 # define ri_lower(c) (class_tab[c] & RI_LOWER) | |
588 # define ri_upper(c) (class_tab[c] & RI_UPPER) | |
589 # define ri_white(c) (class_tab[c] & RI_WHITE) | |
590 #endif | |
591 | |
592 /* flags for regflags */ | |
593 #define RF_ICASE 1 /* ignore case */ | |
594 #define RF_NOICASE 2 /* don't ignore case */ | |
595 #define RF_HASNL 4 /* can match a NL */ | |
596 #define RF_ICOMBINE 8 /* ignore combining characters */ | |
597 #define RF_LOOKBH 16 /* uses "\@<=" or "\@<!" */ | |
598 | |
599 /* | |
600 * Global work variables for vim_regcomp(). | |
601 */ | |
602 | |
603 static char_u *regparse; /* Input-scan pointer. */ | |
604 static int prevchr_len; /* byte length of previous char */ | |
605 static int num_complex_braces; /* Complex \{...} count */ | |
606 static int regnpar; /* () count. */ | |
607 #ifdef FEAT_SYN_HL | |
608 static int regnzpar; /* \z() count. */ | |
609 static int re_has_z; /* \z item detected */ | |
610 #endif | |
611 static char_u *regcode; /* Code-emit pointer, or JUST_CALC_SIZE */ | |
612 static long regsize; /* Code size. */ | |
2010 | 613 static int reg_toolong; /* TRUE when offset out of range */ |
7 | 614 static char_u had_endbrace[NSUBEXP]; /* flags, TRUE if end of () found */ |
615 static unsigned regflags; /* RF_ flags for prog */ | |
616 static long brace_min[10]; /* Minimums for complex brace repeats */ | |
617 static long brace_max[10]; /* Maximums for complex brace repeats */ | |
618 static int brace_count[10]; /* Current counts for complex brace repeats */ | |
619 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
620 static int had_eol; /* TRUE when EOL found by vim_regcomp() */ | |
621 #endif | |
622 static int one_exactly = FALSE; /* only do one char for EXACTLY */ | |
623 | |
624 static int reg_magic; /* magicness of the pattern: */ | |
625 #define MAGIC_NONE 1 /* "\V" very unmagic */ | |
626 #define MAGIC_OFF 2 /* "\M" or 'magic' off */ | |
627 #define MAGIC_ON 3 /* "\m" or 'magic' */ | |
628 #define MAGIC_ALL 4 /* "\v" very magic */ | |
629 | |
630 static int reg_string; /* matching with a string instead of a buffer | |
631 line */ | |
481 | 632 static int reg_strict; /* "[abc" is illegal */ |
7 | 633 |
634 /* | |
635 * META contains all characters that may be magic, except '^' and '$'. | |
636 */ | |
637 | |
638 #ifdef EBCDIC | |
639 static char_u META[] = "%&()*+.123456789<=>?@ACDFHIKLMOPSUVWX[_acdfhiklmnopsuvwxz{|~"; | |
640 #else | |
641 /* META[] is used often enough to justify turning it into a table. */ | |
642 static char_u META_flags[] = { | |
643 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
644 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
645 /* % & ( ) * + . */ | |
646 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0, | |
647 /* 1 2 3 4 5 6 7 8 9 < = > ? */ | |
648 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, | |
649 /* @ A C D F H I K L M O */ | |
650 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1, | |
651 /* P S U V W X Z [ _ */ | |
652 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, | |
653 /* a c d f h i k l m n o */ | |
654 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, | |
655 /* p s u v w x z { | ~ */ | |
656 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1 | |
657 }; | |
658 #endif | |
659 | |
4444 | 660 static int curchr; /* currently parsed character */ |
661 /* Previous character. Note: prevchr is sometimes -1 when we are not at the | |
662 * start, eg in /[ ^I]^ the pattern was never found even if it existed, | |
663 * because ^ was taken to be magic -- webb */ | |
664 static int prevchr; | |
665 static int prevprevchr; /* previous-previous character */ | |
666 static int nextchr; /* used for ungetchr() */ | |
7 | 667 |
668 /* arguments for reg() */ | |
669 #define REG_NOPAREN 0 /* toplevel reg() */ | |
670 #define REG_PAREN 1 /* \(\) */ | |
671 #define REG_ZPAREN 2 /* \z(\) */ | |
672 #define REG_NPAREN 3 /* \%(\) */ | |
673 | |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
674 typedef struct |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
675 { |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
676 char_u *regparse; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
677 int prevchr_len; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
678 int curchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
679 int prevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
680 int prevprevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
681 int nextchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
682 int at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
683 int prev_at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
684 int regnpar; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
685 } parse_state_T; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
686 |
7 | 687 /* |
688 * Forward declarations for vim_regcomp()'s friends. | |
689 */ | |
690 static void initchr __ARGS((char_u *)); | |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
691 static void save_parse_state __ARGS((parse_state_T *ps)); |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
692 static void restore_parse_state __ARGS((parse_state_T *ps)); |
7 | 693 static int getchr __ARGS((void)); |
694 static void skipchr_keepstart __ARGS((void)); | |
695 static int peekchr __ARGS((void)); | |
696 static void skipchr __ARGS((void)); | |
697 static void ungetchr __ARGS((void)); | |
24 | 698 static int gethexchrs __ARGS((int maxinputlen)); |
699 static int getoctchrs __ARGS((void)); | |
700 static int getdecchrs __ARGS((void)); | |
701 static int coll_get_char __ARGS((void)); | |
7 | 702 static void regcomp_start __ARGS((char_u *expr, int flags)); |
703 static char_u *reg __ARGS((int, int *)); | |
704 static char_u *regbranch __ARGS((int *flagp)); | |
705 static char_u *regconcat __ARGS((int *flagp)); | |
706 static char_u *regpiece __ARGS((int *)); | |
707 static char_u *regatom __ARGS((int *)); | |
708 static char_u *regnode __ARGS((int)); | |
714 | 709 #ifdef FEAT_MBYTE |
710 static int use_multibytecode __ARGS((int c)); | |
711 #endif | |
7 | 712 static int prog_magic_wrong __ARGS((void)); |
713 static char_u *regnext __ARGS((char_u *)); | |
714 static void regc __ARGS((int b)); | |
715 #ifdef FEAT_MBYTE | |
716 static void regmbc __ARGS((int c)); | |
2974 | 717 # define REGMBC(x) regmbc(x); |
718 # define CASEMBC(x) case x: | |
167 | 719 #else |
720 # define regmbc(c) regc(c) | |
2974 | 721 # define REGMBC(x) |
722 # define CASEMBC(x) | |
7 | 723 #endif |
724 static void reginsert __ARGS((int, char_u *)); | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
725 static void reginsert_nr __ARGS((int op, long val, char_u *opnd)); |
7 | 726 static void reginsert_limits __ARGS((int, long, long, char_u *)); |
727 static char_u *re_put_long __ARGS((char_u *pr, long_u val)); | |
728 static int read_limits __ARGS((long *, long *)); | |
729 static void regtail __ARGS((char_u *, char_u *)); | |
730 static void regoptail __ARGS((char_u *, char_u *)); | |
731 | |
4444 | 732 static regengine_T bt_regengine; |
733 static regengine_T nfa_regengine; | |
734 | |
7 | 735 /* |
736 * Return TRUE if compiled regular expression "prog" can match a line break. | |
737 */ | |
738 int | |
739 re_multiline(prog) | |
740 regprog_T *prog; | |
741 { | |
742 return (prog->regflags & RF_HASNL); | |
743 } | |
744 | |
745 /* | |
746 * Return TRUE if compiled regular expression "prog" looks before the start | |
747 * position (pattern contains "\@<=" or "\@<!"). | |
748 */ | |
749 int | |
750 re_lookbehind(prog) | |
751 regprog_T *prog; | |
752 { | |
753 return (prog->regflags & RF_LOOKBH); | |
754 } | |
755 | |
756 /* | |
167 | 757 * Check for an equivalence class name "[=a=]". "pp" points to the '['. |
758 * Returns a character representing the class. Zero means that no item was | |
759 * recognized. Otherwise "pp" is advanced to after the item. | |
760 */ | |
761 static int | |
762 get_equi_class(pp) | |
763 char_u **pp; | |
764 { | |
765 int c; | |
766 int l = 1; | |
767 char_u *p = *pp; | |
768 | |
769 if (p[1] == '=') | |
770 { | |
771 #ifdef FEAT_MBYTE | |
772 if (has_mbyte) | |
474 | 773 l = (*mb_ptr2len)(p + 2); |
167 | 774 #endif |
775 if (p[l + 2] == '=' && p[l + 3] == ']') | |
776 { | |
777 #ifdef FEAT_MBYTE | |
778 if (has_mbyte) | |
779 c = mb_ptr2char(p + 2); | |
780 else | |
781 #endif | |
782 c = p[2]; | |
783 *pp += l + 4; | |
784 return c; | |
785 } | |
786 } | |
787 return 0; | |
788 } | |
789 | |
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
790 #ifdef EBCDIC |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
791 /* |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
792 * Table for equivalence class "c". (IBM-1047) |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
793 */ |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
794 char *EQUIVAL_CLASS_C[16] = { |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
795 "A\x62\x63\x64\x65\x66\x67", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
796 "C\x68", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
797 "E\x71\x72\x73\x74", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
798 "I\x75\x76\x77\x78", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
799 "N\x69", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
800 "O\xEB\xEC\xED\xEE\xEF", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
801 "U\xFB\xFC\xFD\xFE", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
802 "Y\xBA", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
803 "a\x42\x43\x44\x45\x46\x47", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
804 "c\x48", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
805 "e\x51\x52\x53\x54", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
806 "i\x55\x56\x57\x58", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
807 "n\x49", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
808 "o\xCB\xCC\xCD\xCE\xCF", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
809 "u\xDB\xDC\xDD\xDE", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
810 "y\x8D\xDF", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
811 }; |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
812 #endif |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
813 |
167 | 814 /* |
815 * Produce the bytes for equivalence class "c". | |
816 * Currently only handles latin1, latin9 and utf-8. | |
4444 | 817 * NOTE: When changing this function, also change nfa_emit_equi_class() |
167 | 818 */ |
819 static void | |
820 reg_equi_class(c) | |
821 int c; | |
822 { | |
823 #ifdef FEAT_MBYTE | |
824 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
492 | 825 || STRCMP(p_enc, "iso-8859-15") == 0) |
167 | 826 #endif |
827 { | |
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
828 #ifdef EBCDIC |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
829 int i; |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
830 |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
831 /* This might be slower than switch/case below. */ |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
832 for (i = 0; i < 16; i++) |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
833 { |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
834 if (vim_strchr(EQUIVAL_CLASS_C[i], c) != NULL) |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
835 { |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
836 char *p = EQUIVAL_CLASS_C[i]; |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
837 |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
838 while (*p != 0) |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
839 regmbc(*p++); |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
840 return; |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
841 } |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
842 } |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
843 #else |
167 | 844 switch (c) |
845 { | |
236 | 846 case 'A': case '\300': case '\301': case '\302': |
2974 | 847 CASEMBC(0x100) CASEMBC(0x102) CASEMBC(0x104) CASEMBC(0x1cd) |
848 CASEMBC(0x1de) CASEMBC(0x1e0) CASEMBC(0x1ea2) | |
236 | 849 case '\303': case '\304': case '\305': |
850 regmbc('A'); regmbc('\300'); regmbc('\301'); | |
851 regmbc('\302'); regmbc('\303'); regmbc('\304'); | |
852 regmbc('\305'); | |
2974 | 853 REGMBC(0x100) REGMBC(0x102) REGMBC(0x104) |
854 REGMBC(0x1cd) REGMBC(0x1de) REGMBC(0x1e0) | |
855 REGMBC(0x1ea2) | |
856 return; | |
857 case 'B': CASEMBC(0x1e02) CASEMBC(0x1e06) | |
858 regmbc('B'); REGMBC(0x1e02) REGMBC(0x1e06) | |
167 | 859 return; |
236 | 860 case 'C': case '\307': |
2974 | 861 CASEMBC(0x106) CASEMBC(0x108) CASEMBC(0x10a) CASEMBC(0x10c) |
236 | 862 regmbc('C'); regmbc('\307'); |
2974 | 863 REGMBC(0x106) REGMBC(0x108) REGMBC(0x10a) |
864 REGMBC(0x10c) | |
865 return; | |
866 case 'D': CASEMBC(0x10e) CASEMBC(0x110) CASEMBC(0x1e0a) | |
867 CASEMBC(0x1e0e) CASEMBC(0x1e10) | |
868 regmbc('D'); REGMBC(0x10e) REGMBC(0x110) | |
869 REGMBC(0x1e0a) REGMBC(0x1e0e) REGMBC(0x1e10) | |
167 | 870 return; |
236 | 871 case 'E': case '\310': case '\311': case '\312': case '\313': |
2974 | 872 CASEMBC(0x112) CASEMBC(0x114) CASEMBC(0x116) CASEMBC(0x118) |
873 CASEMBC(0x11a) CASEMBC(0x1eba) CASEMBC(0x1ebc) | |
236 | 874 regmbc('E'); regmbc('\310'); regmbc('\311'); |
875 regmbc('\312'); regmbc('\313'); | |
2974 | 876 REGMBC(0x112) REGMBC(0x114) REGMBC(0x116) |
877 REGMBC(0x118) REGMBC(0x11a) REGMBC(0x1eba) | |
878 REGMBC(0x1ebc) | |
879 return; | |
880 case 'F': CASEMBC(0x1e1e) | |
881 regmbc('F'); REGMBC(0x1e1e) | |
882 return; | |
883 case 'G': CASEMBC(0x11c) CASEMBC(0x11e) CASEMBC(0x120) | |
884 CASEMBC(0x122) CASEMBC(0x1e4) CASEMBC(0x1e6) CASEMBC(0x1f4) | |
885 CASEMBC(0x1e20) | |
886 regmbc('G'); REGMBC(0x11c) REGMBC(0x11e) | |
887 REGMBC(0x120) REGMBC(0x122) REGMBC(0x1e4) | |
888 REGMBC(0x1e6) REGMBC(0x1f4) REGMBC(0x1e20) | |
889 return; | |
890 case 'H': CASEMBC(0x124) CASEMBC(0x126) CASEMBC(0x1e22) | |
891 CASEMBC(0x1e26) CASEMBC(0x1e28) | |
892 regmbc('H'); REGMBC(0x124) REGMBC(0x126) | |
893 REGMBC(0x1e22) REGMBC(0x1e26) REGMBC(0x1e28) | |
167 | 894 return; |
236 | 895 case 'I': case '\314': case '\315': case '\316': case '\317': |
2974 | 896 CASEMBC(0x128) CASEMBC(0x12a) CASEMBC(0x12c) CASEMBC(0x12e) |
897 CASEMBC(0x130) CASEMBC(0x1cf) CASEMBC(0x1ec8) | |
236 | 898 regmbc('I'); regmbc('\314'); regmbc('\315'); |
899 regmbc('\316'); regmbc('\317'); | |
2974 | 900 REGMBC(0x128) REGMBC(0x12a) REGMBC(0x12c) |
901 REGMBC(0x12e) REGMBC(0x130) REGMBC(0x1cf) | |
902 REGMBC(0x1ec8) | |
903 return; | |
904 case 'J': CASEMBC(0x134) | |
905 regmbc('J'); REGMBC(0x134) | |
906 return; | |
907 case 'K': CASEMBC(0x136) CASEMBC(0x1e8) CASEMBC(0x1e30) | |
908 CASEMBC(0x1e34) | |
909 regmbc('K'); REGMBC(0x136) REGMBC(0x1e8) | |
910 REGMBC(0x1e30) REGMBC(0x1e34) | |
911 return; | |
912 case 'L': CASEMBC(0x139) CASEMBC(0x13b) CASEMBC(0x13d) | |
913 CASEMBC(0x13f) CASEMBC(0x141) CASEMBC(0x1e3a) | |
914 regmbc('L'); REGMBC(0x139) REGMBC(0x13b) | |
915 REGMBC(0x13d) REGMBC(0x13f) REGMBC(0x141) | |
916 REGMBC(0x1e3a) | |
917 return; | |
918 case 'M': CASEMBC(0x1e3e) CASEMBC(0x1e40) | |
919 regmbc('M'); REGMBC(0x1e3e) REGMBC(0x1e40) | |
167 | 920 return; |
236 | 921 case 'N': case '\321': |
2974 | 922 CASEMBC(0x143) CASEMBC(0x145) CASEMBC(0x147) CASEMBC(0x1e44) |
923 CASEMBC(0x1e48) | |
236 | 924 regmbc('N'); regmbc('\321'); |
2974 | 925 REGMBC(0x143) REGMBC(0x145) REGMBC(0x147) |
926 REGMBC(0x1e44) REGMBC(0x1e48) | |
167 | 927 return; |
236 | 928 case 'O': case '\322': case '\323': case '\324': case '\325': |
2974 | 929 case '\326': case '\330': |
930 CASEMBC(0x14c) CASEMBC(0x14e) CASEMBC(0x150) CASEMBC(0x1a0) | |
931 CASEMBC(0x1d1) CASEMBC(0x1ea) CASEMBC(0x1ec) CASEMBC(0x1ece) | |
236 | 932 regmbc('O'); regmbc('\322'); regmbc('\323'); |
933 regmbc('\324'); regmbc('\325'); regmbc('\326'); | |
2974 | 934 regmbc('\330'); |
935 REGMBC(0x14c) REGMBC(0x14e) REGMBC(0x150) | |
936 REGMBC(0x1a0) REGMBC(0x1d1) REGMBC(0x1ea) | |
937 REGMBC(0x1ec) REGMBC(0x1ece) | |
938 return; | |
939 case 'P': case 0x1e54: case 0x1e56: | |
940 regmbc('P'); REGMBC(0x1e54) REGMBC(0x1e56) | |
941 return; | |
942 case 'R': CASEMBC(0x154) CASEMBC(0x156) CASEMBC(0x158) | |
943 CASEMBC(0x1e58) CASEMBC(0x1e5e) | |
944 regmbc('R'); REGMBC(0x154) REGMBC(0x156) REGMBC(0x158) | |
945 REGMBC(0x1e58) REGMBC(0x1e5e) | |
946 return; | |
947 case 'S': CASEMBC(0x15a) CASEMBC(0x15c) CASEMBC(0x15e) | |
948 CASEMBC(0x160) CASEMBC(0x1e60) | |
949 regmbc('S'); REGMBC(0x15a) REGMBC(0x15c) | |
950 REGMBC(0x15e) REGMBC(0x160) REGMBC(0x1e60) | |
951 return; | |
952 case 'T': CASEMBC(0x162) CASEMBC(0x164) CASEMBC(0x166) | |
953 CASEMBC(0x1e6a) CASEMBC(0x1e6e) | |
954 regmbc('T'); REGMBC(0x162) REGMBC(0x164) | |
955 REGMBC(0x166) REGMBC(0x1e6a) REGMBC(0x1e6e) | |
167 | 956 return; |
236 | 957 case 'U': case '\331': case '\332': case '\333': case '\334': |
2974 | 958 CASEMBC(0x168) CASEMBC(0x16a) CASEMBC(0x16c) CASEMBC(0x16e) |
959 CASEMBC(0x170) CASEMBC(0x172) CASEMBC(0x1af) CASEMBC(0x1d3) | |
960 CASEMBC(0x1ee6) | |
236 | 961 regmbc('U'); regmbc('\331'); regmbc('\332'); |
962 regmbc('\333'); regmbc('\334'); | |
2974 | 963 REGMBC(0x168) REGMBC(0x16a) REGMBC(0x16c) |
964 REGMBC(0x16e) REGMBC(0x170) REGMBC(0x172) | |
965 REGMBC(0x1af) REGMBC(0x1d3) REGMBC(0x1ee6) | |
966 return; | |
967 case 'V': CASEMBC(0x1e7c) | |
968 regmbc('V'); REGMBC(0x1e7c) | |
969 return; | |
970 case 'W': CASEMBC(0x174) CASEMBC(0x1e80) CASEMBC(0x1e82) | |
971 CASEMBC(0x1e84) CASEMBC(0x1e86) | |
972 regmbc('W'); REGMBC(0x174) REGMBC(0x1e80) | |
973 REGMBC(0x1e82) REGMBC(0x1e84) REGMBC(0x1e86) | |
974 return; | |
975 case 'X': CASEMBC(0x1e8a) CASEMBC(0x1e8c) | |
976 regmbc('X'); REGMBC(0x1e8a) REGMBC(0x1e8c) | |
167 | 977 return; |
236 | 978 case 'Y': case '\335': |
2974 | 979 CASEMBC(0x176) CASEMBC(0x178) CASEMBC(0x1e8e) CASEMBC(0x1ef2) |
980 CASEMBC(0x1ef6) CASEMBC(0x1ef8) | |
236 | 981 regmbc('Y'); regmbc('\335'); |
2974 | 982 REGMBC(0x176) REGMBC(0x178) REGMBC(0x1e8e) |
983 REGMBC(0x1ef2) REGMBC(0x1ef6) REGMBC(0x1ef8) | |
984 return; | |
985 case 'Z': CASEMBC(0x179) CASEMBC(0x17b) CASEMBC(0x17d) | |
986 CASEMBC(0x1b5) CASEMBC(0x1e90) CASEMBC(0x1e94) | |
987 regmbc('Z'); REGMBC(0x179) REGMBC(0x17b) | |
988 REGMBC(0x17d) REGMBC(0x1b5) REGMBC(0x1e90) | |
989 REGMBC(0x1e94) | |
167 | 990 return; |
236 | 991 case 'a': case '\340': case '\341': case '\342': |
992 case '\343': case '\344': case '\345': | |
2974 | 993 CASEMBC(0x101) CASEMBC(0x103) CASEMBC(0x105) CASEMBC(0x1ce) |
994 CASEMBC(0x1df) CASEMBC(0x1e1) CASEMBC(0x1ea3) | |
236 | 995 regmbc('a'); regmbc('\340'); regmbc('\341'); |
996 regmbc('\342'); regmbc('\343'); regmbc('\344'); | |
997 regmbc('\345'); | |
2974 | 998 REGMBC(0x101) REGMBC(0x103) REGMBC(0x105) |
999 REGMBC(0x1ce) REGMBC(0x1df) REGMBC(0x1e1) | |
1000 REGMBC(0x1ea3) | |
1001 return; | |
1002 case 'b': CASEMBC(0x1e03) CASEMBC(0x1e07) | |
1003 regmbc('b'); REGMBC(0x1e03) REGMBC(0x1e07) | |
167 | 1004 return; |
236 | 1005 case 'c': case '\347': |
2974 | 1006 CASEMBC(0x107) CASEMBC(0x109) CASEMBC(0x10b) CASEMBC(0x10d) |
236 | 1007 regmbc('c'); regmbc('\347'); |
2974 | 1008 REGMBC(0x107) REGMBC(0x109) REGMBC(0x10b) |
1009 REGMBC(0x10d) | |
1010 return; | |
1011 case 'd': CASEMBC(0x10f) CASEMBC(0x111) CASEMBC(0x1d0b) | |
1012 CASEMBC(0x1e11) | |
1013 regmbc('d'); REGMBC(0x10f) REGMBC(0x111) | |
1014 REGMBC(0x1e0b) REGMBC(0x01e0f) REGMBC(0x1e11) | |
167 | 1015 return; |
236 | 1016 case 'e': case '\350': case '\351': case '\352': case '\353': |
2974 | 1017 CASEMBC(0x113) CASEMBC(0x115) CASEMBC(0x117) CASEMBC(0x119) |
1018 CASEMBC(0x11b) CASEMBC(0x1ebb) CASEMBC(0x1ebd) | |
236 | 1019 regmbc('e'); regmbc('\350'); regmbc('\351'); |
1020 regmbc('\352'); regmbc('\353'); | |
2974 | 1021 REGMBC(0x113) REGMBC(0x115) REGMBC(0x117) |
1022 REGMBC(0x119) REGMBC(0x11b) REGMBC(0x1ebb) | |
1023 REGMBC(0x1ebd) | |
1024 return; | |
1025 case 'f': CASEMBC(0x1e1f) | |
1026 regmbc('f'); REGMBC(0x1e1f) | |
1027 return; | |
1028 case 'g': CASEMBC(0x11d) CASEMBC(0x11f) CASEMBC(0x121) | |
1029 CASEMBC(0x123) CASEMBC(0x1e5) CASEMBC(0x1e7) CASEMBC(0x1f5) | |
1030 CASEMBC(0x1e21) | |
1031 regmbc('g'); REGMBC(0x11d) REGMBC(0x11f) | |
1032 REGMBC(0x121) REGMBC(0x123) REGMBC(0x1e5) | |
1033 REGMBC(0x1e7) REGMBC(0x1f5) REGMBC(0x1e21) | |
1034 return; | |
1035 case 'h': CASEMBC(0x125) CASEMBC(0x127) CASEMBC(0x1e23) | |
1036 CASEMBC(0x1e27) CASEMBC(0x1e29) CASEMBC(0x1e96) | |
1037 regmbc('h'); REGMBC(0x125) REGMBC(0x127) | |
1038 REGMBC(0x1e23) REGMBC(0x1e27) REGMBC(0x1e29) | |
1039 REGMBC(0x1e96) | |
167 | 1040 return; |
236 | 1041 case 'i': case '\354': case '\355': case '\356': case '\357': |
2974 | 1042 CASEMBC(0x129) CASEMBC(0x12b) CASEMBC(0x12d) CASEMBC(0x12f) |
1043 CASEMBC(0x1d0) CASEMBC(0x1ec9) | |
236 | 1044 regmbc('i'); regmbc('\354'); regmbc('\355'); |
1045 regmbc('\356'); regmbc('\357'); | |
2974 | 1046 REGMBC(0x129) REGMBC(0x12b) REGMBC(0x12d) |
1047 REGMBC(0x12f) REGMBC(0x1d0) REGMBC(0x1ec9) | |
1048 return; | |
1049 case 'j': CASEMBC(0x135) CASEMBC(0x1f0) | |
1050 regmbc('j'); REGMBC(0x135) REGMBC(0x1f0) | |
1051 return; | |
1052 case 'k': CASEMBC(0x137) CASEMBC(0x1e9) CASEMBC(0x1e31) | |
1053 CASEMBC(0x1e35) | |
1054 regmbc('k'); REGMBC(0x137) REGMBC(0x1e9) | |
1055 REGMBC(0x1e31) REGMBC(0x1e35) | |
1056 return; | |
1057 case 'l': CASEMBC(0x13a) CASEMBC(0x13c) CASEMBC(0x13e) | |
1058 CASEMBC(0x140) CASEMBC(0x142) CASEMBC(0x1e3b) | |
1059 regmbc('l'); REGMBC(0x13a) REGMBC(0x13c) | |
1060 REGMBC(0x13e) REGMBC(0x140) REGMBC(0x142) | |
1061 REGMBC(0x1e3b) | |
1062 return; | |
1063 case 'm': CASEMBC(0x1e3f) CASEMBC(0x1e41) | |
1064 regmbc('m'); REGMBC(0x1e3f) REGMBC(0x1e41) | |
167 | 1065 return; |
236 | 1066 case 'n': case '\361': |
2974 | 1067 CASEMBC(0x144) CASEMBC(0x146) CASEMBC(0x148) CASEMBC(0x149) |
1068 CASEMBC(0x1e45) CASEMBC(0x1e49) | |
236 | 1069 regmbc('n'); regmbc('\361'); |
2974 | 1070 REGMBC(0x144) REGMBC(0x146) REGMBC(0x148) |
1071 REGMBC(0x149) REGMBC(0x1e45) REGMBC(0x1e49) | |
167 | 1072 return; |
236 | 1073 case 'o': case '\362': case '\363': case '\364': case '\365': |
2974 | 1074 case '\366': case '\370': |
1075 CASEMBC(0x14d) CASEMBC(0x14f) CASEMBC(0x151) CASEMBC(0x1a1) | |
1076 CASEMBC(0x1d2) CASEMBC(0x1eb) CASEMBC(0x1ed) CASEMBC(0x1ecf) | |
236 | 1077 regmbc('o'); regmbc('\362'); regmbc('\363'); |
1078 regmbc('\364'); regmbc('\365'); regmbc('\366'); | |
2974 | 1079 regmbc('\370'); |
1080 REGMBC(0x14d) REGMBC(0x14f) REGMBC(0x151) | |
1081 REGMBC(0x1a1) REGMBC(0x1d2) REGMBC(0x1eb) | |
1082 REGMBC(0x1ed) REGMBC(0x1ecf) | |
1083 return; | |
1084 case 'p': CASEMBC(0x1e55) CASEMBC(0x1e57) | |
1085 regmbc('p'); REGMBC(0x1e55) REGMBC(0x1e57) | |
1086 return; | |
1087 case 'r': CASEMBC(0x155) CASEMBC(0x157) CASEMBC(0x159) | |
1088 CASEMBC(0x1e59) CASEMBC(0x1e5f) | |
1089 regmbc('r'); REGMBC(0x155) REGMBC(0x157) REGMBC(0x159) | |
1090 REGMBC(0x1e59) REGMBC(0x1e5f) | |
1091 return; | |
1092 case 's': CASEMBC(0x15b) CASEMBC(0x15d) CASEMBC(0x15f) | |
1093 CASEMBC(0x161) CASEMBC(0x1e61) | |
1094 regmbc('s'); REGMBC(0x15b) REGMBC(0x15d) | |
1095 REGMBC(0x15f) REGMBC(0x161) REGMBC(0x1e61) | |
1096 return; | |
1097 case 't': CASEMBC(0x163) CASEMBC(0x165) CASEMBC(0x167) | |
1098 CASEMBC(0x1e6b) CASEMBC(0x1e6f) CASEMBC(0x1e97) | |
1099 regmbc('t'); REGMBC(0x163) REGMBC(0x165) REGMBC(0x167) | |
1100 REGMBC(0x1e6b) REGMBC(0x1e6f) REGMBC(0x1e97) | |
167 | 1101 return; |
236 | 1102 case 'u': case '\371': case '\372': case '\373': case '\374': |
2974 | 1103 CASEMBC(0x169) CASEMBC(0x16b) CASEMBC(0x16d) CASEMBC(0x16f) |
1104 CASEMBC(0x171) CASEMBC(0x173) CASEMBC(0x1b0) CASEMBC(0x1d4) | |
1105 CASEMBC(0x1ee7) | |
236 | 1106 regmbc('u'); regmbc('\371'); regmbc('\372'); |
1107 regmbc('\373'); regmbc('\374'); | |
2974 | 1108 REGMBC(0x169) REGMBC(0x16b) REGMBC(0x16d) |
1109 REGMBC(0x16f) REGMBC(0x171) REGMBC(0x173) | |
1110 REGMBC(0x1b0) REGMBC(0x1d4) REGMBC(0x1ee7) | |
1111 return; | |
1112 case 'v': CASEMBC(0x1e7d) | |
1113 regmbc('v'); REGMBC(0x1e7d) | |
1114 return; | |
1115 case 'w': CASEMBC(0x175) CASEMBC(0x1e81) CASEMBC(0x1e83) | |
1116 CASEMBC(0x1e85) CASEMBC(0x1e87) CASEMBC(0x1e98) | |
1117 regmbc('w'); REGMBC(0x175) REGMBC(0x1e81) | |
1118 REGMBC(0x1e83) REGMBC(0x1e85) REGMBC(0x1e87) | |
1119 REGMBC(0x1e98) | |
1120 return; | |
1121 case 'x': CASEMBC(0x1e8b) CASEMBC(0x1e8d) | |
1122 regmbc('x'); REGMBC(0x1e8b) REGMBC(0x1e8d) | |
167 | 1123 return; |
236 | 1124 case 'y': case '\375': case '\377': |
2974 | 1125 CASEMBC(0x177) CASEMBC(0x1e8f) CASEMBC(0x1e99) |
1126 CASEMBC(0x1ef3) CASEMBC(0x1ef7) CASEMBC(0x1ef9) | |
236 | 1127 regmbc('y'); regmbc('\375'); regmbc('\377'); |
2974 | 1128 REGMBC(0x177) REGMBC(0x1e8f) REGMBC(0x1e99) |
1129 REGMBC(0x1ef3) REGMBC(0x1ef7) REGMBC(0x1ef9) | |
1130 return; | |
1131 case 'z': CASEMBC(0x17a) CASEMBC(0x17c) CASEMBC(0x17e) | |
1132 CASEMBC(0x1b6) CASEMBC(0x1e91) CASEMBC(0x1e95) | |
1133 regmbc('z'); REGMBC(0x17a) REGMBC(0x17c) | |
1134 REGMBC(0x17e) REGMBC(0x1b6) REGMBC(0x1e91) | |
1135 REGMBC(0x1e95) | |
167 | 1136 return; |
1137 } | |
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
1138 #endif |
167 | 1139 } |
1140 regmbc(c); | |
1141 } | |
1142 | |
1143 /* | |
1144 * Check for a collating element "[.a.]". "pp" points to the '['. | |
1145 * Returns a character. Zero means that no item was recognized. Otherwise | |
1146 * "pp" is advanced to after the item. | |
1147 * Currently only single characters are recognized! | |
1148 */ | |
1149 static int | |
1150 get_coll_element(pp) | |
1151 char_u **pp; | |
1152 { | |
1153 int c; | |
1154 int l = 1; | |
1155 char_u *p = *pp; | |
1156 | |
1157 if (p[1] == '.') | |
1158 { | |
1159 #ifdef FEAT_MBYTE | |
1160 if (has_mbyte) | |
474 | 1161 l = (*mb_ptr2len)(p + 2); |
167 | 1162 #endif |
1163 if (p[l + 2] == '.' && p[l + 3] == ']') | |
1164 { | |
1165 #ifdef FEAT_MBYTE | |
1166 if (has_mbyte) | |
1167 c = mb_ptr2char(p + 2); | |
1168 else | |
1169 #endif | |
1170 c = p[2]; | |
1171 *pp += l + 4; | |
1172 return c; | |
1173 } | |
1174 } | |
1175 return 0; | |
1176 } | |
1177 | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1178 static void get_cpo_flags __ARGS((void)); |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1179 static int reg_cpo_lit; /* 'cpoptions' contains 'l' flag */ |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1180 static int reg_cpo_bsl; /* 'cpoptions' contains '\' flag */ |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1181 |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1182 static void |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1183 get_cpo_flags() |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1184 { |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1185 reg_cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1186 reg_cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1187 } |
167 | 1188 |
1189 /* | |
1190 * Skip over a "[]" range. | |
1191 * "p" must point to the character after the '['. | |
1192 * The returned pointer is on the matching ']', or the terminating NUL. | |
1193 */ | |
1194 static char_u * | |
1195 skip_anyof(p) | |
1196 char_u *p; | |
1197 { | |
1198 #ifdef FEAT_MBYTE | |
1199 int l; | |
1200 #endif | |
1201 | |
1202 if (*p == '^') /* Complement of range. */ | |
1203 ++p; | |
1204 if (*p == ']' || *p == '-') | |
1205 ++p; | |
1206 while (*p != NUL && *p != ']') | |
1207 { | |
1208 #ifdef FEAT_MBYTE | |
474 | 1209 if (has_mbyte && (l = (*mb_ptr2len)(p)) > 1) |
167 | 1210 p += l; |
1211 else | |
1212 #endif | |
1213 if (*p == '-') | |
1214 { | |
1215 ++p; | |
1216 if (*p != ']' && *p != NUL) | |
1217 mb_ptr_adv(p); | |
1218 } | |
1219 else if (*p == '\\' | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1220 && !reg_cpo_bsl |
167 | 1221 && (vim_strchr(REGEXP_INRANGE, p[1]) != NULL |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1222 || (!reg_cpo_lit && vim_strchr(REGEXP_ABBR, p[1]) != NULL))) |
167 | 1223 p += 2; |
1224 else if (*p == '[') | |
1225 { | |
1226 if (get_char_class(&p) == CLASS_NONE | |
1227 && get_equi_class(&p) == 0 | |
1228 && get_coll_element(&p) == 0) | |
1229 ++p; /* It was not a class name */ | |
1230 } | |
1231 else | |
1232 ++p; | |
1233 } | |
1234 | |
1235 return p; | |
1236 } | |
1237 | |
1238 /* | |
7 | 1239 * Skip past regular expression. |
153 | 1240 * Stop at end of "startp" or where "dirc" is found ('/', '?', etc). |
7 | 1241 * Take care of characters with a backslash in front of it. |
1242 * Skip strings inside [ and ]. | |
1243 * When "newp" is not NULL and "dirc" is '?', make an allocated copy of the | |
1244 * expression and change "\?" to "?". If "*newp" is not NULL the expression | |
1245 * is changed in-place. | |
1246 */ | |
1247 char_u * | |
1248 skip_regexp(startp, dirc, magic, newp) | |
1249 char_u *startp; | |
1250 int dirc; | |
1251 int magic; | |
1252 char_u **newp; | |
1253 { | |
1254 int mymagic; | |
1255 char_u *p = startp; | |
1256 | |
1257 if (magic) | |
1258 mymagic = MAGIC_ON; | |
1259 else | |
1260 mymagic = MAGIC_OFF; | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1261 get_cpo_flags(); |
7 | 1262 |
39 | 1263 for (; p[0] != NUL; mb_ptr_adv(p)) |
7 | 1264 { |
1265 if (p[0] == dirc) /* found end of regexp */ | |
1266 break; | |
1267 if ((p[0] == '[' && mymagic >= MAGIC_ON) | |
1268 || (p[0] == '\\' && p[1] == '[' && mymagic <= MAGIC_OFF)) | |
1269 { | |
1270 p = skip_anyof(p + 1); | |
1271 if (p[0] == NUL) | |
1272 break; | |
1273 } | |
1274 else if (p[0] == '\\' && p[1] != NUL) | |
1275 { | |
1276 if (dirc == '?' && newp != NULL && p[1] == '?') | |
1277 { | |
1278 /* change "\?" to "?", make a copy first. */ | |
1279 if (*newp == NULL) | |
1280 { | |
1281 *newp = vim_strsave(startp); | |
1282 if (*newp != NULL) | |
1283 p = *newp + (p - startp); | |
1284 } | |
1285 if (*newp != NULL) | |
1621 | 1286 STRMOVE(p, p + 1); |
7 | 1287 else |
1288 ++p; | |
1289 } | |
1290 else | |
1291 ++p; /* skip next character */ | |
1292 if (*p == 'v') | |
1293 mymagic = MAGIC_ALL; | |
1294 else if (*p == 'V') | |
1295 mymagic = MAGIC_NONE; | |
1296 } | |
1297 } | |
1298 return p; | |
1299 } | |
1300 | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1301 static regprog_T *bt_regcomp __ARGS((char_u *expr, int re_flags)); |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1302 static void bt_regfree __ARGS((regprog_T *prog)); |
4444 | 1303 |
7 | 1304 /* |
4444 | 1305 * bt_regcomp() - compile a regular expression into internal code for the |
1306 * traditional back track matcher. | |
41 | 1307 * Returns the program in allocated space. Returns NULL for an error. |
7 | 1308 * |
1309 * We can't allocate space until we know how big the compiled form will be, | |
1310 * but we can't compile it (and thus know how big it is) until we've got a | |
1311 * place to put the code. So we cheat: we compile it twice, once with code | |
1312 * generation turned off and size counting turned on, and once "for real". | |
1313 * This also means that we don't allocate space until we are sure that the | |
1314 * thing really will compile successfully, and we never have to move the | |
1315 * code and thus invalidate pointers into it. (Note that it has to be in | |
1316 * one piece because vim_free() must be able to free it all.) | |
1317 * | |
1318 * Whether upper/lower case is to be ignored is decided when executing the | |
1319 * program, it does not matter here. | |
1320 * | |
1321 * Beware that the optimization-preparation code in here knows about some | |
1322 * of the structure of the compiled regexp. | |
1323 * "re_flags": RE_MAGIC and/or RE_STRING. | |
1324 */ | |
4444 | 1325 static regprog_T * |
1326 bt_regcomp(expr, re_flags) | |
7 | 1327 char_u *expr; |
1328 int re_flags; | |
1329 { | |
4444 | 1330 bt_regprog_T *r; |
7 | 1331 char_u *scan; |
1332 char_u *longest; | |
1333 int len; | |
1334 int flags; | |
1335 | |
1336 if (expr == NULL) | |
1337 EMSG_RET_NULL(_(e_null)); | |
1338 | |
1339 init_class_tab(); | |
1340 | |
1341 /* | |
1342 * First pass: determine size, legality. | |
1343 */ | |
1344 regcomp_start(expr, re_flags); | |
1345 regcode = JUST_CALC_SIZE; | |
1346 regc(REGMAGIC); | |
1347 if (reg(REG_NOPAREN, &flags) == NULL) | |
1348 return NULL; | |
1349 | |
1350 /* Small enough for pointer-storage convention? */ | |
1351 #ifdef SMALL_MALLOC /* 16 bit storage allocation */ | |
1352 if (regsize >= 65536L - 256L) | |
1353 EMSG_RET_NULL(_("E339: Pattern too long")); | |
1354 #endif | |
1355 | |
1356 /* Allocate space. */ | |
4444 | 1357 r = (bt_regprog_T *)lalloc(sizeof(bt_regprog_T) + regsize, TRUE); |
7 | 1358 if (r == NULL) |
1359 return NULL; | |
1360 | |
1361 /* | |
1362 * Second pass: emit code. | |
1363 */ | |
1364 regcomp_start(expr, re_flags); | |
1365 regcode = r->program; | |
1366 regc(REGMAGIC); | |
2010 | 1367 if (reg(REG_NOPAREN, &flags) == NULL || reg_toolong) |
7 | 1368 { |
1369 vim_free(r); | |
2010 | 1370 if (reg_toolong) |
1371 EMSG_RET_NULL(_("E339: Pattern too long")); | |
7 | 1372 return NULL; |
1373 } | |
1374 | |
1375 /* Dig out information for optimizations. */ | |
1376 r->regstart = NUL; /* Worst-case defaults. */ | |
1377 r->reganch = 0; | |
1378 r->regmust = NULL; | |
1379 r->regmlen = 0; | |
1380 r->regflags = regflags; | |
1381 if (flags & HASNL) | |
1382 r->regflags |= RF_HASNL; | |
1383 if (flags & HASLOOKBH) | |
1384 r->regflags |= RF_LOOKBH; | |
1385 #ifdef FEAT_SYN_HL | |
1386 /* Remember whether this pattern has any \z specials in it. */ | |
1387 r->reghasz = re_has_z; | |
1388 #endif | |
1389 scan = r->program + 1; /* First BRANCH. */ | |
1390 if (OP(regnext(scan)) == END) /* Only one top-level choice. */ | |
1391 { | |
1392 scan = OPERAND(scan); | |
1393 | |
1394 /* Starting-point info. */ | |
1395 if (OP(scan) == BOL || OP(scan) == RE_BOF) | |
1396 { | |
1397 r->reganch++; | |
1398 scan = regnext(scan); | |
1399 } | |
1400 | |
1401 if (OP(scan) == EXACTLY) | |
1402 { | |
1403 #ifdef FEAT_MBYTE | |
1404 if (has_mbyte) | |
1405 r->regstart = (*mb_ptr2char)(OPERAND(scan)); | |
1406 else | |
1407 #endif | |
1408 r->regstart = *OPERAND(scan); | |
1409 } | |
1410 else if ((OP(scan) == BOW | |
1411 || OP(scan) == EOW | |
1412 || OP(scan) == NOTHING | |
1413 || OP(scan) == MOPEN + 0 || OP(scan) == NOPEN | |
1414 || OP(scan) == MCLOSE + 0 || OP(scan) == NCLOSE) | |
1415 && OP(regnext(scan)) == EXACTLY) | |
1416 { | |
1417 #ifdef FEAT_MBYTE | |
1418 if (has_mbyte) | |
1419 r->regstart = (*mb_ptr2char)(OPERAND(regnext(scan))); | |
1420 else | |
1421 #endif | |
1422 r->regstart = *OPERAND(regnext(scan)); | |
1423 } | |
1424 | |
1425 /* | |
1426 * If there's something expensive in the r.e., find the longest | |
1427 * literal string that must appear and make it the regmust. Resolve | |
1428 * ties in favor of later strings, since the regstart check works | |
1429 * with the beginning of the r.e. and avoiding duplication | |
1430 * strengthens checking. Not a strong reason, but sufficient in the | |
1431 * absence of others. | |
1432 */ | |
1433 /* | |
1434 * When the r.e. starts with BOW, it is faster to look for a regmust | |
1435 * first. Used a lot for "#" and "*" commands. (Added by mool). | |
1436 */ | |
1437 if ((flags & SPSTART || OP(scan) == BOW || OP(scan) == EOW) | |
1438 && !(flags & HASNL)) | |
1439 { | |
1440 longest = NULL; | |
1441 len = 0; | |
1442 for (; scan != NULL; scan = regnext(scan)) | |
1443 if (OP(scan) == EXACTLY && STRLEN(OPERAND(scan)) >= (size_t)len) | |
1444 { | |
1445 longest = OPERAND(scan); | |
1446 len = (int)STRLEN(OPERAND(scan)); | |
1447 } | |
1448 r->regmust = longest; | |
1449 r->regmlen = len; | |
1450 } | |
1451 } | |
4444 | 1452 #ifdef BT_REGEXP_DUMP |
7 | 1453 regdump(expr, r); |
1454 #endif | |
4444 | 1455 r->engine = &bt_regengine; |
1456 return (regprog_T *)r; | |
7 | 1457 } |
1458 | |
1459 /* | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1460 * Free a compiled regexp program, returned by bt_regcomp(). |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1461 */ |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1462 static void |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1463 bt_regfree(prog) |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1464 regprog_T *prog; |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1465 { |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1466 vim_free(prog); |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1467 } |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1468 |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1469 /* |
7 | 1470 * Setup to parse the regexp. Used once to get the length and once to do it. |
1471 */ | |
1472 static void | |
1473 regcomp_start(expr, re_flags) | |
1474 char_u *expr; | |
1475 int re_flags; /* see vim_regcomp() */ | |
1476 { | |
1477 initchr(expr); | |
1478 if (re_flags & RE_MAGIC) | |
1479 reg_magic = MAGIC_ON; | |
1480 else | |
1481 reg_magic = MAGIC_OFF; | |
1482 reg_string = (re_flags & RE_STRING); | |
481 | 1483 reg_strict = (re_flags & RE_STRICT); |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1484 get_cpo_flags(); |
7 | 1485 |
1486 num_complex_braces = 0; | |
1487 regnpar = 1; | |
1488 vim_memset(had_endbrace, 0, sizeof(had_endbrace)); | |
1489 #ifdef FEAT_SYN_HL | |
1490 regnzpar = 1; | |
1491 re_has_z = 0; | |
1492 #endif | |
1493 regsize = 0L; | |
2010 | 1494 reg_toolong = FALSE; |
7 | 1495 regflags = 0; |
1496 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
1497 had_eol = FALSE; | |
1498 #endif | |
1499 } | |
1500 | |
1501 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
1502 /* | |
1503 * Check if during the previous call to vim_regcomp the EOL item "$" has been | |
1504 * found. This is messy, but it works fine. | |
1505 */ | |
1506 int | |
1507 vim_regcomp_had_eol() | |
1508 { | |
1509 return had_eol; | |
1510 } | |
1511 #endif | |
1512 | |
1513 /* | |
4444 | 1514 * Parse regular expression, i.e. main body or parenthesized thing. |
7 | 1515 * |
1516 * Caller must absorb opening parenthesis. | |
1517 * | |
1518 * Combining parenthesis handling with the base level of regular expression | |
1519 * is a trifle forced, but the need to tie the tails of the branches to what | |
1520 * follows makes it hard to avoid. | |
1521 */ | |
1522 static char_u * | |
1523 reg(paren, flagp) | |
1524 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
1525 int *flagp; | |
1526 { | |
1527 char_u *ret; | |
1528 char_u *br; | |
1529 char_u *ender; | |
1530 int parno = 0; | |
1531 int flags; | |
1532 | |
1533 *flagp = HASWIDTH; /* Tentatively. */ | |
1534 | |
1535 #ifdef FEAT_SYN_HL | |
1536 if (paren == REG_ZPAREN) | |
1537 { | |
1538 /* Make a ZOPEN node. */ | |
1539 if (regnzpar >= NSUBEXP) | |
1540 EMSG_RET_NULL(_("E50: Too many \\z(")); | |
1541 parno = regnzpar; | |
1542 regnzpar++; | |
1543 ret = regnode(ZOPEN + parno); | |
1544 } | |
1545 else | |
1546 #endif | |
1547 if (paren == REG_PAREN) | |
1548 { | |
1549 /* Make a MOPEN node. */ | |
1550 if (regnpar >= NSUBEXP) | |
4444 | 1551 EMSG2_RET_NULL(_("E51: Too many %s("), reg_magic == MAGIC_ALL); |
7 | 1552 parno = regnpar; |
1553 ++regnpar; | |
1554 ret = regnode(MOPEN + parno); | |
1555 } | |
1556 else if (paren == REG_NPAREN) | |
1557 { | |
1558 /* Make a NOPEN node. */ | |
1559 ret = regnode(NOPEN); | |
1560 } | |
1561 else | |
1562 ret = NULL; | |
1563 | |
1564 /* Pick up the branches, linking them together. */ | |
1565 br = regbranch(&flags); | |
1566 if (br == NULL) | |
1567 return NULL; | |
1568 if (ret != NULL) | |
1569 regtail(ret, br); /* [MZ]OPEN -> first. */ | |
1570 else | |
1571 ret = br; | |
1572 /* If one of the branches can be zero-width, the whole thing can. | |
1573 * If one of the branches has * at start or matches a line-break, the | |
1574 * whole thing can. */ | |
1575 if (!(flags & HASWIDTH)) | |
1576 *flagp &= ~HASWIDTH; | |
1577 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
1578 while (peekchr() == Magic('|')) | |
1579 { | |
1580 skipchr(); | |
1581 br = regbranch(&flags); | |
2010 | 1582 if (br == NULL || reg_toolong) |
7 | 1583 return NULL; |
1584 regtail(ret, br); /* BRANCH -> BRANCH. */ | |
1585 if (!(flags & HASWIDTH)) | |
1586 *flagp &= ~HASWIDTH; | |
1587 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
1588 } | |
1589 | |
1590 /* Make a closing node, and hook it on the end. */ | |
1591 ender = regnode( | |
1592 #ifdef FEAT_SYN_HL | |
1593 paren == REG_ZPAREN ? ZCLOSE + parno : | |
1594 #endif | |
1595 paren == REG_PAREN ? MCLOSE + parno : | |
1596 paren == REG_NPAREN ? NCLOSE : END); | |
1597 regtail(ret, ender); | |
1598 | |
1599 /* Hook the tails of the branches to the closing node. */ | |
1600 for (br = ret; br != NULL; br = regnext(br)) | |
1601 regoptail(br, ender); | |
1602 | |
1603 /* Check for proper termination. */ | |
1604 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
1605 { | |
1606 #ifdef FEAT_SYN_HL | |
1607 if (paren == REG_ZPAREN) | |
308 | 1608 EMSG_RET_NULL(_("E52: Unmatched \\z(")); |
7 | 1609 else |
1610 #endif | |
1611 if (paren == REG_NPAREN) | |
4444 | 1612 EMSG2_RET_NULL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); |
7 | 1613 else |
4444 | 1614 EMSG2_RET_NULL(_(e_unmatchedp), reg_magic == MAGIC_ALL); |
7 | 1615 } |
1616 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
1617 { | |
1618 if (curchr == Magic(')')) | |
4444 | 1619 EMSG2_RET_NULL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); |
7 | 1620 else |
308 | 1621 EMSG_RET_NULL(_(e_trailing)); /* "Can't happen". */ |
7 | 1622 /* NOTREACHED */ |
1623 } | |
1624 /* | |
1625 * Here we set the flag allowing back references to this set of | |
1626 * parentheses. | |
1627 */ | |
1628 if (paren == REG_PAREN) | |
1629 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
1630 return ret; | |
1631 } | |
1632 | |
1633 /* | |
4444 | 1634 * Parse one alternative of an | operator. |
7 | 1635 * Implements the & operator. |
1636 */ | |
1637 static char_u * | |
1638 regbranch(flagp) | |
1639 int *flagp; | |
1640 { | |
1641 char_u *ret; | |
1642 char_u *chain = NULL; | |
1643 char_u *latest; | |
1644 int flags; | |
1645 | |
1646 *flagp = WORST | HASNL; /* Tentatively. */ | |
1647 | |
1648 ret = regnode(BRANCH); | |
1649 for (;;) | |
1650 { | |
1651 latest = regconcat(&flags); | |
1652 if (latest == NULL) | |
1653 return NULL; | |
1654 /* If one of the branches has width, the whole thing has. If one of | |
1655 * the branches anchors at start-of-line, the whole thing does. | |
1656 * If one of the branches uses look-behind, the whole thing does. */ | |
1657 *flagp |= flags & (HASWIDTH | SPSTART | HASLOOKBH); | |
1658 /* If one of the branches doesn't match a line-break, the whole thing | |
1659 * doesn't. */ | |
1660 *flagp &= ~HASNL | (flags & HASNL); | |
1661 if (chain != NULL) | |
1662 regtail(chain, latest); | |
1663 if (peekchr() != Magic('&')) | |
1664 break; | |
1665 skipchr(); | |
1666 regtail(latest, regnode(END)); /* operand ends */ | |
2010 | 1667 if (reg_toolong) |
1668 break; | |
7 | 1669 reginsert(MATCH, latest); |
1670 chain = latest; | |
1671 } | |
1672 | |
1673 return ret; | |
1674 } | |
1675 | |
1676 /* | |
4444 | 1677 * Parse one alternative of an | or & operator. |
7 | 1678 * Implements the concatenation operator. |
1679 */ | |
1680 static char_u * | |
1681 regconcat(flagp) | |
1682 int *flagp; | |
1683 { | |
1684 char_u *first = NULL; | |
1685 char_u *chain = NULL; | |
1686 char_u *latest; | |
1687 int flags; | |
1688 int cont = TRUE; | |
1689 | |
1690 *flagp = WORST; /* Tentatively. */ | |
1691 | |
1692 while (cont) | |
1693 { | |
1694 switch (peekchr()) | |
1695 { | |
1696 case NUL: | |
1697 case Magic('|'): | |
1698 case Magic('&'): | |
1699 case Magic(')'): | |
1700 cont = FALSE; | |
1701 break; | |
1702 case Magic('Z'): | |
1703 #ifdef FEAT_MBYTE | |
1704 regflags |= RF_ICOMBINE; | |
1705 #endif | |
1706 skipchr_keepstart(); | |
1707 break; | |
1708 case Magic('c'): | |
1709 regflags |= RF_ICASE; | |
1710 skipchr_keepstart(); | |
1711 break; | |
1712 case Magic('C'): | |
1713 regflags |= RF_NOICASE; | |
1714 skipchr_keepstart(); | |
1715 break; | |
1716 case Magic('v'): | |
1717 reg_magic = MAGIC_ALL; | |
1718 skipchr_keepstart(); | |
1719 curchr = -1; | |
1720 break; | |
1721 case Magic('m'): | |
1722 reg_magic = MAGIC_ON; | |
1723 skipchr_keepstart(); | |
1724 curchr = -1; | |
1725 break; | |
1726 case Magic('M'): | |
1727 reg_magic = MAGIC_OFF; | |
1728 skipchr_keepstart(); | |
1729 curchr = -1; | |
1730 break; | |
1731 case Magic('V'): | |
1732 reg_magic = MAGIC_NONE; | |
1733 skipchr_keepstart(); | |
1734 curchr = -1; | |
1735 break; | |
1736 default: | |
1737 latest = regpiece(&flags); | |
2010 | 1738 if (latest == NULL || reg_toolong) |
7 | 1739 return NULL; |
1740 *flagp |= flags & (HASWIDTH | HASNL | HASLOOKBH); | |
1741 if (chain == NULL) /* First piece. */ | |
1742 *flagp |= flags & SPSTART; | |
1743 else | |
1744 regtail(chain, latest); | |
1745 chain = latest; | |
1746 if (first == NULL) | |
1747 first = latest; | |
1748 break; | |
1749 } | |
1750 } | |
1751 if (first == NULL) /* Loop ran zero times. */ | |
1752 first = regnode(NOTHING); | |
1753 return first; | |
1754 } | |
1755 | |
1756 /* | |
4444 | 1757 * Parse something followed by possible [*+=]. |
7 | 1758 * |
1759 * Note that the branching code sequences used for = and the general cases | |
1760 * of * and + are somewhat optimized: they use the same NOTHING node as | |
1761 * both the endmarker for their branch list and the body of the last branch. | |
1762 * It might seem that this node could be dispensed with entirely, but the | |
1763 * endmarker role is not redundant. | |
1764 */ | |
1765 static char_u * | |
1766 regpiece(flagp) | |
1767 int *flagp; | |
1768 { | |
1769 char_u *ret; | |
1770 int op; | |
1771 char_u *next; | |
1772 int flags; | |
1773 long minval; | |
1774 long maxval; | |
1775 | |
1776 ret = regatom(&flags); | |
1777 if (ret == NULL) | |
1778 return NULL; | |
1779 | |
1780 op = peekchr(); | |
1781 if (re_multi_type(op) == NOT_MULTI) | |
1782 { | |
1783 *flagp = flags; | |
1784 return ret; | |
1785 } | |
1786 /* default flags */ | |
1787 *flagp = (WORST | SPSTART | (flags & (HASNL | HASLOOKBH))); | |
1788 | |
1789 skipchr(); | |
1790 switch (op) | |
1791 { | |
1792 case Magic('*'): | |
1793 if (flags & SIMPLE) | |
1794 reginsert(STAR, ret); | |
1795 else | |
1796 { | |
1797 /* Emit x* as (x&|), where & means "self". */ | |
1798 reginsert(BRANCH, ret); /* Either x */ | |
1799 regoptail(ret, regnode(BACK)); /* and loop */ | |
1800 regoptail(ret, ret); /* back */ | |
1801 regtail(ret, regnode(BRANCH)); /* or */ | |
1802 regtail(ret, regnode(NOTHING)); /* null. */ | |
1803 } | |
1804 break; | |
1805 | |
1806 case Magic('+'): | |
1807 if (flags & SIMPLE) | |
1808 reginsert(PLUS, ret); | |
1809 else | |
1810 { | |
1811 /* Emit x+ as x(&|), where & means "self". */ | |
1812 next = regnode(BRANCH); /* Either */ | |
1813 regtail(ret, next); | |
233 | 1814 regtail(regnode(BACK), ret); /* loop back */ |
7 | 1815 regtail(next, regnode(BRANCH)); /* or */ |
1816 regtail(ret, regnode(NOTHING)); /* null. */ | |
1817 } | |
1818 *flagp = (WORST | HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
1819 break; | |
1820 | |
1821 case Magic('@'): | |
1822 { | |
1823 int lop = END; | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1824 int nr; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1825 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1826 nr = getdecchrs(); |
7 | 1827 switch (no_Magic(getchr())) |
1828 { | |
1829 case '=': lop = MATCH; break; /* \@= */ | |
1830 case '!': lop = NOMATCH; break; /* \@! */ | |
1831 case '>': lop = SUBPAT; break; /* \@> */ | |
1832 case '<': switch (no_Magic(getchr())) | |
1833 { | |
1834 case '=': lop = BEHIND; break; /* \@<= */ | |
1835 case '!': lop = NOBEHIND; break; /* \@<! */ | |
1836 } | |
1837 } | |
1838 if (lop == END) | |
4444 | 1839 EMSG2_RET_NULL(_("E59: invalid character after %s@"), |
7 | 1840 reg_magic == MAGIC_ALL); |
1841 /* Look behind must match with behind_pos. */ | |
1842 if (lop == BEHIND || lop == NOBEHIND) | |
1843 { | |
1844 regtail(ret, regnode(BHPOS)); | |
1845 *flagp |= HASLOOKBH; | |
1846 } | |
1847 regtail(ret, regnode(END)); /* operand ends */ | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1848 if (lop == BEHIND || lop == NOBEHIND) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1849 { |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1850 if (nr < 0) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1851 nr = 0; /* no limit is same as zero limit */ |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1852 reginsert_nr(lop, nr, ret); |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1853 } |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1854 else |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1855 reginsert(lop, ret); |
7 | 1856 break; |
1857 } | |
1858 | |
1859 case Magic('?'): | |
1860 case Magic('='): | |
1861 /* Emit x= as (x|) */ | |
1862 reginsert(BRANCH, ret); /* Either x */ | |
1863 regtail(ret, regnode(BRANCH)); /* or */ | |
1864 next = regnode(NOTHING); /* null. */ | |
1865 regtail(ret, next); | |
1866 regoptail(ret, next); | |
1867 break; | |
1868 | |
1869 case Magic('{'): | |
1870 if (!read_limits(&minval, &maxval)) | |
1871 return NULL; | |
1872 if (flags & SIMPLE) | |
1873 { | |
1874 reginsert(BRACE_SIMPLE, ret); | |
1875 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
1876 } | |
1877 else | |
1878 { | |
1879 if (num_complex_braces >= 10) | |
4444 | 1880 EMSG2_RET_NULL(_("E60: Too many complex %s{...}s"), |
7 | 1881 reg_magic == MAGIC_ALL); |
1882 reginsert(BRACE_COMPLEX + num_complex_braces, ret); | |
1883 regoptail(ret, regnode(BACK)); | |
1884 regoptail(ret, ret); | |
1885 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
1886 ++num_complex_braces; | |
1887 } | |
1888 if (minval > 0 && maxval > 0) | |
1889 *flagp = (HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
1890 break; | |
1891 } | |
1892 if (re_multi_type(peekchr()) != NOT_MULTI) | |
1893 { | |
1894 /* Can't have a multi follow a multi. */ | |
1895 if (peekchr() == Magic('*')) | |
1896 sprintf((char *)IObuff, _("E61: Nested %s*"), | |
1897 reg_magic >= MAGIC_ON ? "" : "\\"); | |
1898 else | |
1899 sprintf((char *)IObuff, _("E62: Nested %s%c"), | |
1900 reg_magic == MAGIC_ALL ? "" : "\\", no_Magic(peekchr())); | |
1901 EMSG_RET_NULL(IObuff); | |
1902 } | |
1903 | |
1904 return ret; | |
1905 } | |
1906 | |
4444 | 1907 /* When making changes to classchars also change nfa_classcodes. */ |
1908 static char_u *classchars = (char_u *)".iIkKfFpPsSdDxXoOwWhHaAlLuU"; | |
1909 static int classcodes[] = { | |
1910 ANY, IDENT, SIDENT, KWORD, SKWORD, | |
1911 FNAME, SFNAME, PRINT, SPRINT, | |
1912 WHITE, NWHITE, DIGIT, NDIGIT, | |
1913 HEX, NHEX, OCTAL, NOCTAL, | |
1914 WORD, NWORD, HEAD, NHEAD, | |
1915 ALPHA, NALPHA, LOWER, NLOWER, | |
1916 UPPER, NUPPER | |
1917 }; | |
1918 | |
7 | 1919 /* |
4444 | 1920 * Parse the lowest level. |
7 | 1921 * |
1922 * Optimization: gobbles an entire sequence of ordinary characters so that | |
1923 * it can turn them into a single node, which is smaller to store and | |
1924 * faster to run. Don't do this when one_exactly is set. | |
1925 */ | |
1926 static char_u * | |
1927 regatom(flagp) | |
1928 int *flagp; | |
1929 { | |
1930 char_u *ret; | |
1931 int flags; | |
1932 int c; | |
1933 char_u *p; | |
1934 int extra = 0; | |
1935 | |
1936 *flagp = WORST; /* Tentatively. */ | |
1937 | |
1938 c = getchr(); | |
1939 switch (c) | |
1940 { | |
1941 case Magic('^'): | |
1942 ret = regnode(BOL); | |
1943 break; | |
1944 | |
1945 case Magic('$'): | |
1946 ret = regnode(EOL); | |
1947 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
1948 had_eol = TRUE; | |
1949 #endif | |
1950 break; | |
1951 | |
1952 case Magic('<'): | |
1953 ret = regnode(BOW); | |
1954 break; | |
1955 | |
1956 case Magic('>'): | |
1957 ret = regnode(EOW); | |
1958 break; | |
1959 | |
1960 case Magic('_'): | |
1961 c = no_Magic(getchr()); | |
1962 if (c == '^') /* "\_^" is start-of-line */ | |
1963 { | |
1964 ret = regnode(BOL); | |
1965 break; | |
1966 } | |
1967 if (c == '$') /* "\_$" is end-of-line */ | |
1968 { | |
1969 ret = regnode(EOL); | |
1970 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
1971 had_eol = TRUE; | |
1972 #endif | |
1973 break; | |
1974 } | |
1975 | |
1976 extra = ADD_NL; | |
1977 *flagp |= HASNL; | |
1978 | |
1979 /* "\_[" is character range plus newline */ | |
1980 if (c == '[') | |
1981 goto collection; | |
1982 | |
1983 /* "\_x" is character class plus newline */ | |
1984 /*FALLTHROUGH*/ | |
1985 | |
1986 /* | |
1987 * Character classes. | |
1988 */ | |
1989 case Magic('.'): | |
1990 case Magic('i'): | |
1991 case Magic('I'): | |
1992 case Magic('k'): | |
1993 case Magic('K'): | |
1994 case Magic('f'): | |
1995 case Magic('F'): | |
1996 case Magic('p'): | |
1997 case Magic('P'): | |
1998 case Magic('s'): | |
1999 case Magic('S'): | |
2000 case Magic('d'): | |
2001 case Magic('D'): | |
2002 case Magic('x'): | |
2003 case Magic('X'): | |
2004 case Magic('o'): | |
2005 case Magic('O'): | |
2006 case Magic('w'): | |
2007 case Magic('W'): | |
2008 case Magic('h'): | |
2009 case Magic('H'): | |
2010 case Magic('a'): | |
2011 case Magic('A'): | |
2012 case Magic('l'): | |
2013 case Magic('L'): | |
2014 case Magic('u'): | |
2015 case Magic('U'): | |
2016 p = vim_strchr(classchars, no_Magic(c)); | |
2017 if (p == NULL) | |
2018 EMSG_RET_NULL(_("E63: invalid use of \\_")); | |
714 | 2019 #ifdef FEAT_MBYTE |
2020 /* When '.' is followed by a composing char ignore the dot, so that | |
2021 * the composing char is matched here. */ | |
2022 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
2023 { | |
2024 c = getchr(); | |
2025 goto do_multibyte; | |
2026 } | |
2027 #endif | |
7 | 2028 ret = regnode(classcodes[p - classchars] + extra); |
2029 *flagp |= HASWIDTH | SIMPLE; | |
2030 break; | |
2031 | |
2032 case Magic('n'): | |
2033 if (reg_string) | |
2034 { | |
2035 /* In a string "\n" matches a newline character. */ | |
2036 ret = regnode(EXACTLY); | |
2037 regc(NL); | |
2038 regc(NUL); | |
2039 *flagp |= HASWIDTH | SIMPLE; | |
2040 } | |
2041 else | |
2042 { | |
2043 /* In buffer text "\n" matches the end of a line. */ | |
2044 ret = regnode(NEWL); | |
2045 *flagp |= HASWIDTH | HASNL; | |
2046 } | |
2047 break; | |
2048 | |
2049 case Magic('('): | |
2050 if (one_exactly) | |
2051 EMSG_ONE_RET_NULL; | |
2052 ret = reg(REG_PAREN, &flags); | |
2053 if (ret == NULL) | |
2054 return NULL; | |
2055 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
2056 break; | |
2057 | |
2058 case NUL: | |
2059 case Magic('|'): | |
2060 case Magic('&'): | |
2061 case Magic(')'): | |
1468 | 2062 if (one_exactly) |
2063 EMSG_ONE_RET_NULL; | |
7 | 2064 EMSG_RET_NULL(_(e_internal)); /* Supposed to be caught earlier. */ |
2065 /* NOTREACHED */ | |
2066 | |
2067 case Magic('='): | |
2068 case Magic('?'): | |
2069 case Magic('+'): | |
2070 case Magic('@'): | |
2071 case Magic('{'): | |
2072 case Magic('*'): | |
2073 c = no_Magic(c); | |
2074 sprintf((char *)IObuff, _("E64: %s%c follows nothing"), | |
2075 (c == '*' ? reg_magic >= MAGIC_ON : reg_magic == MAGIC_ALL) | |
2076 ? "" : "\\", c); | |
2077 EMSG_RET_NULL(IObuff); | |
2078 /* NOTREACHED */ | |
2079 | |
2080 case Magic('~'): /* previous substitute pattern */ | |
359 | 2081 if (reg_prev_sub != NULL) |
7 | 2082 { |
2083 char_u *lp; | |
2084 | |
2085 ret = regnode(EXACTLY); | |
2086 lp = reg_prev_sub; | |
2087 while (*lp != NUL) | |
2088 regc(*lp++); | |
2089 regc(NUL); | |
2090 if (*reg_prev_sub != NUL) | |
2091 { | |
2092 *flagp |= HASWIDTH; | |
2093 if ((lp - reg_prev_sub) == 1) | |
2094 *flagp |= SIMPLE; | |
2095 } | |
2096 } | |
2097 else | |
2098 EMSG_RET_NULL(_(e_nopresub)); | |
2099 break; | |
2100 | |
2101 case Magic('1'): | |
2102 case Magic('2'): | |
2103 case Magic('3'): | |
2104 case Magic('4'): | |
2105 case Magic('5'): | |
2106 case Magic('6'): | |
2107 case Magic('7'): | |
2108 case Magic('8'): | |
2109 case Magic('9'): | |
2110 { | |
2111 int refnum; | |
2112 | |
2113 refnum = c - Magic('0'); | |
2114 /* | |
2115 * Check if the back reference is legal. We must have seen the | |
2116 * close brace. | |
2117 * TODO: Should also check that we don't refer to something | |
2118 * that is repeated (+*=): what instance of the repetition | |
2119 * should we match? | |
2120 */ | |
2121 if (!had_endbrace[refnum]) | |
2122 { | |
2123 /* Trick: check if "@<=" or "@<!" follows, in which case | |
2124 * the \1 can appear before the referenced match. */ | |
2125 for (p = regparse; *p != NUL; ++p) | |
2126 if (p[0] == '@' && p[1] == '<' | |
2127 && (p[2] == '!' || p[2] == '=')) | |
2128 break; | |
2129 if (*p == NUL) | |
2130 EMSG_RET_NULL(_("E65: Illegal back reference")); | |
2131 } | |
2132 ret = regnode(BACKREF + refnum); | |
2133 } | |
2134 break; | |
2135 | |
2136 case Magic('z'): | |
2137 { | |
2138 c = no_Magic(getchr()); | |
2139 switch (c) | |
2140 { | |
741 | 2141 #ifdef FEAT_SYN_HL |
7 | 2142 case '(': if (reg_do_extmatch != REX_SET) |
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
2143 EMSG_RET_NULL(_(e_z_not_allowed)); |
7 | 2144 if (one_exactly) |
2145 EMSG_ONE_RET_NULL; | |
2146 ret = reg(REG_ZPAREN, &flags); | |
2147 if (ret == NULL) | |
2148 return NULL; | |
2149 *flagp |= flags & (HASWIDTH|SPSTART|HASNL|HASLOOKBH); | |
2150 re_has_z = REX_SET; | |
2151 break; | |
2152 | |
2153 case '1': | |
2154 case '2': | |
2155 case '3': | |
2156 case '4': | |
2157 case '5': | |
2158 case '6': | |
2159 case '7': | |
2160 case '8': | |
2161 case '9': if (reg_do_extmatch != REX_USE) | |
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
2162 EMSG_RET_NULL(_(e_z1_not_allowed)); |
7 | 2163 ret = regnode(ZREF + c - '0'); |
2164 re_has_z = REX_USE; | |
2165 break; | |
741 | 2166 #endif |
7 | 2167 |
2168 case 's': ret = regnode(MOPEN + 0); | |
2169 break; | |
2170 | |
2171 case 'e': ret = regnode(MCLOSE + 0); | |
2172 break; | |
2173 | |
2174 default: EMSG_RET_NULL(_("E68: Invalid character after \\z")); | |
2175 } | |
2176 } | |
2177 break; | |
2178 | |
2179 case Magic('%'): | |
2180 { | |
2181 c = no_Magic(getchr()); | |
2182 switch (c) | |
2183 { | |
2184 /* () without a back reference */ | |
2185 case '(': | |
2186 if (one_exactly) | |
2187 EMSG_ONE_RET_NULL; | |
2188 ret = reg(REG_NPAREN, &flags); | |
2189 if (ret == NULL) | |
2190 return NULL; | |
2191 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
2192 break; | |
2193 | |
2194 /* Catch \%^ and \%$ regardless of where they appear in the | |
2195 * pattern -- regardless of whether or not it makes sense. */ | |
2196 case '^': | |
2197 ret = regnode(RE_BOF); | |
2198 break; | |
2199 | |
2200 case '$': | |
2201 ret = regnode(RE_EOF); | |
2202 break; | |
2203 | |
2204 case '#': | |
2205 ret = regnode(CURSOR); | |
2206 break; | |
2207 | |
639 | 2208 case 'V': |
2209 ret = regnode(RE_VISUAL); | |
2210 break; | |
2211 | |
5901 | 2212 case 'C': |
2213 ret = regnode(RE_COMPOSING); | |
2214 break; | |
2215 | |
7 | 2216 /* \%[abc]: Emit as a list of branches, all ending at the last |
2217 * branch which matches nothing. */ | |
2218 case '[': | |
2219 if (one_exactly) /* doesn't nest */ | |
2220 EMSG_ONE_RET_NULL; | |
2221 { | |
2222 char_u *lastbranch; | |
2223 char_u *lastnode = NULL; | |
2224 char_u *br; | |
2225 | |
2226 ret = NULL; | |
2227 while ((c = getchr()) != ']') | |
2228 { | |
2229 if (c == NUL) | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2230 EMSG2_RET_NULL(_(e_missing_sb), |
7 | 2231 reg_magic == MAGIC_ALL); |
2232 br = regnode(BRANCH); | |
2233 if (ret == NULL) | |
2234 ret = br; | |
2235 else | |
2236 regtail(lastnode, br); | |
2237 | |
2238 ungetchr(); | |
2239 one_exactly = TRUE; | |
2240 lastnode = regatom(flagp); | |
2241 one_exactly = FALSE; | |
2242 if (lastnode == NULL) | |
2243 return NULL; | |
2244 } | |
2245 if (ret == NULL) | |
4760
532a9855bd30
updated for version 7.3.1127
Bram Moolenaar <bram@vim.org>
parents:
4746
diff
changeset
|
2246 EMSG2_RET_NULL(_(e_empty_sb), |
7 | 2247 reg_magic == MAGIC_ALL); |
2248 lastbranch = regnode(BRANCH); | |
2249 br = regnode(NOTHING); | |
2250 if (ret != JUST_CALC_SIZE) | |
2251 { | |
2252 regtail(lastnode, br); | |
2253 regtail(lastbranch, br); | |
2254 /* connect all branches to the NOTHING | |
2255 * branch at the end */ | |
2256 for (br = ret; br != lastnode; ) | |
2257 { | |
2258 if (OP(br) == BRANCH) | |
2259 { | |
2260 regtail(br, lastbranch); | |
2261 br = OPERAND(br); | |
2262 } | |
2263 else | |
2264 br = regnext(br); | |
2265 } | |
2266 } | |
1701 | 2267 *flagp &= ~(HASWIDTH | SIMPLE); |
7 | 2268 break; |
2269 } | |
2270 | |
24 | 2271 case 'd': /* %d123 decimal */ |
2272 case 'o': /* %o123 octal */ | |
2273 case 'x': /* %xab hex 2 */ | |
2274 case 'u': /* %uabcd hex 4 */ | |
2275 case 'U': /* %U1234abcd hex 8 */ | |
2276 { | |
2277 int i; | |
2278 | |
2279 switch (c) | |
2280 { | |
2281 case 'd': i = getdecchrs(); break; | |
2282 case 'o': i = getoctchrs(); break; | |
2283 case 'x': i = gethexchrs(2); break; | |
2284 case 'u': i = gethexchrs(4); break; | |
2285 case 'U': i = gethexchrs(8); break; | |
2286 default: i = -1; break; | |
2287 } | |
2288 | |
2289 if (i < 0) | |
4444 | 2290 EMSG2_RET_NULL( |
24 | 2291 _("E678: Invalid character after %s%%[dxouU]"), |
2292 reg_magic == MAGIC_ALL); | |
714 | 2293 #ifdef FEAT_MBYTE |
2294 if (use_multibytecode(i)) | |
2295 ret = regnode(MULTIBYTECODE); | |
2296 else | |
2297 #endif | |
2298 ret = regnode(EXACTLY); | |
24 | 2299 if (i == 0) |
2300 regc(0x0a); | |
2301 else | |
2302 #ifdef FEAT_MBYTE | |
2303 regmbc(i); | |
2304 #else | |
2305 regc(i); | |
2306 #endif | |
2307 regc(NUL); | |
2308 *flagp |= HASWIDTH; | |
2309 break; | |
2310 } | |
2311 | |
7 | 2312 default: |
639 | 2313 if (VIM_ISDIGIT(c) || c == '<' || c == '>' |
2314 || c == '\'') | |
7 | 2315 { |
2316 long_u n = 0; | |
2317 int cmp; | |
2318 | |
2319 cmp = c; | |
2320 if (cmp == '<' || cmp == '>') | |
2321 c = getchr(); | |
2322 while (VIM_ISDIGIT(c)) | |
2323 { | |
2324 n = n * 10 + (c - '0'); | |
2325 c = getchr(); | |
2326 } | |
639 | 2327 if (c == '\'' && n == 0) |
2328 { | |
2329 /* "\%'m", "\%<'m" and "\%>'m": Mark */ | |
2330 c = getchr(); | |
2331 ret = regnode(RE_MARK); | |
2332 if (ret == JUST_CALC_SIZE) | |
2333 regsize += 2; | |
2334 else | |
2335 { | |
2336 *regcode++ = c; | |
2337 *regcode++ = cmp; | |
2338 } | |
2339 break; | |
2340 } | |
2341 else if (c == 'l' || c == 'c' || c == 'v') | |
7 | 2342 { |
2343 if (c == 'l') | |
2344 ret = regnode(RE_LNUM); | |
2345 else if (c == 'c') | |
2346 ret = regnode(RE_COL); | |
2347 else | |
2348 ret = regnode(RE_VCOL); | |
2349 if (ret == JUST_CALC_SIZE) | |
2350 regsize += 5; | |
2351 else | |
2352 { | |
2353 /* put the number and the optional | |
2354 * comparator after the opcode */ | |
2355 regcode = re_put_long(regcode, n); | |
2356 *regcode++ = cmp; | |
2357 } | |
2358 break; | |
2359 } | |
2360 } | |
2361 | |
4444 | 2362 EMSG2_RET_NULL(_("E71: Invalid character after %s%%"), |
7 | 2363 reg_magic == MAGIC_ALL); |
2364 } | |
2365 } | |
2366 break; | |
2367 | |
2368 case Magic('['): | |
2369 collection: | |
2370 { | |
2371 char_u *lp; | |
2372 | |
2373 /* | |
2374 * If there is no matching ']', we assume the '[' is a normal | |
2375 * character. This makes 'incsearch' and ":help [" work. | |
2376 */ | |
2377 lp = skip_anyof(regparse); | |
2378 if (*lp == ']') /* there is a matching ']' */ | |
2379 { | |
2380 int startc = -1; /* > 0 when next '-' is a range */ | |
2381 int endc; | |
2382 | |
2383 /* | |
2384 * In a character class, different parsing rules apply. | |
2385 * Not even \ is special anymore, nothing is. | |
2386 */ | |
2387 if (*regparse == '^') /* Complement of range. */ | |
2388 { | |
2389 ret = regnode(ANYBUT + extra); | |
2390 regparse++; | |
2391 } | |
2392 else | |
2393 ret = regnode(ANYOF + extra); | |
2394 | |
2395 /* At the start ']' and '-' mean the literal character. */ | |
2396 if (*regparse == ']' || *regparse == '-') | |
167 | 2397 { |
2398 startc = *regparse; | |
7 | 2399 regc(*regparse++); |
167 | 2400 } |
7 | 2401 |
2402 while (*regparse != NUL && *regparse != ']') | |
2403 { | |
2404 if (*regparse == '-') | |
2405 { | |
2406 ++regparse; | |
2407 /* The '-' is not used for a range at the end and | |
2408 * after or before a '\n'. */ | |
2409 if (*regparse == ']' || *regparse == NUL | |
2410 || startc == -1 | |
2411 || (regparse[0] == '\\' && regparse[1] == 'n')) | |
2412 { | |
2413 regc('-'); | |
2414 startc = '-'; /* [--x] is a range */ | |
2415 } | |
2416 else | |
2417 { | |
167 | 2418 /* Also accept "a-[.z.]" */ |
2419 endc = 0; | |
2420 if (*regparse == '[') | |
2421 endc = get_coll_element(®parse); | |
2422 if (endc == 0) | |
2423 { | |
7 | 2424 #ifdef FEAT_MBYTE |
167 | 2425 if (has_mbyte) |
2426 endc = mb_ptr2char_adv(®parse); | |
2427 else | |
7 | 2428 #endif |
167 | 2429 endc = *regparse++; |
2430 } | |
24 | 2431 |
2432 /* Handle \o40, \x20 and \u20AC style sequences */ | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2433 if (endc == '\\' && !reg_cpo_lit && !reg_cpo_bsl) |
24 | 2434 endc = coll_get_char(); |
2435 | |
7 | 2436 if (startc > endc) |
2437 EMSG_RET_NULL(_(e_invrange)); | |
2438 #ifdef FEAT_MBYTE | |
2439 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
2440 || (*mb_char2len)(endc) > 1)) | |
2441 { | |
2442 /* Limit to a range of 256 chars */ | |
2443 if (endc > startc + 256) | |
2444 EMSG_RET_NULL(_(e_invrange)); | |
2445 while (++startc <= endc) | |
2446 regmbc(startc); | |
2447 } | |
2448 else | |
2449 #endif | |
2450 { | |
2451 #ifdef EBCDIC | |
2452 int alpha_only = FALSE; | |
2453 | |
2454 /* for alphabetical range skip the gaps | |
2455 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
2456 if (isalpha(startc) && isalpha(endc)) | |
2457 alpha_only = TRUE; | |
2458 #endif | |
2459 while (++startc <= endc) | |
2460 #ifdef EBCDIC | |
2461 if (!alpha_only || isalpha(startc)) | |
2462 #endif | |
2463 regc(startc); | |
2464 } | |
2465 startc = -1; | |
2466 } | |
2467 } | |
2468 /* | |
2469 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
2470 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
2471 * 'cpoptions' is not included. | |
167 | 2472 * Posix doesn't recognize backslash at all. |
7 | 2473 */ |
2474 else if (*regparse == '\\' | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2475 && !reg_cpo_bsl |
7 | 2476 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2477 || (!reg_cpo_lit |
7 | 2478 && vim_strchr(REGEXP_ABBR, |
2479 regparse[1]) != NULL))) | |
2480 { | |
2481 regparse++; | |
2482 if (*regparse == 'n') | |
2483 { | |
2484 /* '\n' in range: also match NL */ | |
2485 if (ret != JUST_CALC_SIZE) | |
2486 { | |
4084 | 2487 /* Using \n inside [^] does not change what |
2488 * matches. "[^\n]" is the same as ".". */ | |
2489 if (*ret == ANYOF) | |
2490 { | |
7 | 2491 *ret = ANYOF + ADD_NL; |
4084 | 2492 *flagp |= HASNL; |
2493 } | |
7 | 2494 /* else: must have had a \n already */ |
2495 } | |
2496 regparse++; | |
2497 startc = -1; | |
2498 } | |
24 | 2499 else if (*regparse == 'd' |
2500 || *regparse == 'o' | |
2501 || *regparse == 'x' | |
2502 || *regparse == 'u' | |
2503 || *regparse == 'U') | |
2504 { | |
2505 startc = coll_get_char(); | |
2506 if (startc == 0) | |
2507 regc(0x0a); | |
2508 else | |
2509 #ifdef FEAT_MBYTE | |
2510 regmbc(startc); | |
2511 #else | |
2512 regc(startc); | |
2513 #endif | |
2514 } | |
7 | 2515 else |
2516 { | |
2517 startc = backslash_trans(*regparse++); | |
2518 regc(startc); | |
2519 } | |
2520 } | |
2521 else if (*regparse == '[') | |
2522 { | |
2523 int c_class; | |
2524 int cu; | |
2525 | |
167 | 2526 c_class = get_char_class(®parse); |
7 | 2527 startc = -1; |
2528 /* Characters assumed to be 8 bits! */ | |
2529 switch (c_class) | |
2530 { | |
2531 case CLASS_NONE: | |
167 | 2532 c_class = get_equi_class(®parse); |
2533 if (c_class != 0) | |
2534 { | |
2535 /* produce equivalence class */ | |
2536 reg_equi_class(c_class); | |
2537 } | |
2538 else if ((c_class = | |
2539 get_coll_element(®parse)) != 0) | |
2540 { | |
2541 /* produce a collating element */ | |
2542 regmbc(c_class); | |
2543 } | |
2544 else | |
2545 { | |
2546 /* literal '[', allow [[-x] as a range */ | |
2547 startc = *regparse++; | |
2548 regc(startc); | |
2549 } | |
7 | 2550 break; |
2551 case CLASS_ALNUM: | |
2552 for (cu = 1; cu <= 255; cu++) | |
2553 if (isalnum(cu)) | |
2554 regc(cu); | |
2555 break; | |
2556 case CLASS_ALPHA: | |
2557 for (cu = 1; cu <= 255; cu++) | |
2558 if (isalpha(cu)) | |
2559 regc(cu); | |
2560 break; | |
2561 case CLASS_BLANK: | |
2562 regc(' '); | |
2563 regc('\t'); | |
2564 break; | |
2565 case CLASS_CNTRL: | |
2566 for (cu = 1; cu <= 255; cu++) | |
2567 if (iscntrl(cu)) | |
2568 regc(cu); | |
2569 break; | |
2570 case CLASS_DIGIT: | |
2571 for (cu = 1; cu <= 255; cu++) | |
2572 if (VIM_ISDIGIT(cu)) | |
2573 regc(cu); | |
2574 break; | |
2575 case CLASS_GRAPH: | |
2576 for (cu = 1; cu <= 255; cu++) | |
2577 if (isgraph(cu)) | |
2578 regc(cu); | |
2579 break; | |
2580 case CLASS_LOWER: | |
2581 for (cu = 1; cu <= 255; cu++) | |
1347 | 2582 if (MB_ISLOWER(cu)) |
7 | 2583 regc(cu); |
2584 break; | |
2585 case CLASS_PRINT: | |
2586 for (cu = 1; cu <= 255; cu++) | |
2587 if (vim_isprintc(cu)) | |
2588 regc(cu); | |
2589 break; | |
2590 case CLASS_PUNCT: | |
2591 for (cu = 1; cu <= 255; cu++) | |
2592 if (ispunct(cu)) | |
2593 regc(cu); | |
2594 break; | |
2595 case CLASS_SPACE: | |
2596 for (cu = 9; cu <= 13; cu++) | |
2597 regc(cu); | |
2598 regc(' '); | |
2599 break; | |
2600 case CLASS_UPPER: | |
2601 for (cu = 1; cu <= 255; cu++) | |
1347 | 2602 if (MB_ISUPPER(cu)) |
7 | 2603 regc(cu); |
2604 break; | |
2605 case CLASS_XDIGIT: | |
2606 for (cu = 1; cu <= 255; cu++) | |
2607 if (vim_isxdigit(cu)) | |
2608 regc(cu); | |
2609 break; | |
2610 case CLASS_TAB: | |
2611 regc('\t'); | |
2612 break; | |
2613 case CLASS_RETURN: | |
2614 regc('\r'); | |
2615 break; | |
2616 case CLASS_BACKSPACE: | |
2617 regc('\b'); | |
2618 break; | |
2619 case CLASS_ESCAPE: | |
2620 regc('\033'); | |
2621 break; | |
2622 } | |
2623 } | |
2624 else | |
2625 { | |
2626 #ifdef FEAT_MBYTE | |
2627 if (has_mbyte) | |
2628 { | |
2629 int len; | |
2630 | |
2631 /* produce a multibyte character, including any | |
2632 * following composing characters */ | |
2633 startc = mb_ptr2char(regparse); | |
474 | 2634 len = (*mb_ptr2len)(regparse); |
7 | 2635 if (enc_utf8 && utf_char2len(startc) != len) |
2636 startc = -1; /* composing chars */ | |
2637 while (--len >= 0) | |
2638 regc(*regparse++); | |
2639 } | |
2640 else | |
2641 #endif | |
2642 { | |
2643 startc = *regparse++; | |
2644 regc(startc); | |
2645 } | |
2646 } | |
2647 } | |
2648 regc(NUL); | |
2649 prevchr_len = 1; /* last char was the ']' */ | |
2650 if (*regparse != ']') | |
2651 EMSG_RET_NULL(_(e_toomsbra)); /* Cannot happen? */ | |
2652 skipchr(); /* let's be friends with the lexer again */ | |
2653 *flagp |= HASWIDTH | SIMPLE; | |
2654 break; | |
2655 } | |
481 | 2656 else if (reg_strict) |
4444 | 2657 EMSG2_RET_NULL(_(e_missingbracket), reg_magic > MAGIC_OFF); |
7 | 2658 } |
2659 /* FALLTHROUGH */ | |
2660 | |
2661 default: | |
2662 { | |
2663 int len; | |
2664 | |
2665 #ifdef FEAT_MBYTE | |
2666 /* A multi-byte character is handled as a separate atom if it's | |
714 | 2667 * before a multi and when it's a composing char. */ |
2668 if (use_multibytecode(c)) | |
7 | 2669 { |
714 | 2670 do_multibyte: |
7 | 2671 ret = regnode(MULTIBYTECODE); |
2672 regmbc(c); | |
2673 *flagp |= HASWIDTH | SIMPLE; | |
2674 break; | |
2675 } | |
2676 #endif | |
2677 | |
2678 ret = regnode(EXACTLY); | |
2679 | |
2680 /* | |
2681 * Append characters as long as: | |
2682 * - there is no following multi, we then need the character in | |
2683 * front of it as a single character operand | |
2684 * - not running into a Magic character | |
2685 * - "one_exactly" is not set | |
2686 * But always emit at least one character. Might be a Multi, | |
2687 * e.g., a "[" without matching "]". | |
2688 */ | |
2689 for (len = 0; c != NUL && (len == 0 | |
2690 || (re_multi_type(peekchr()) == NOT_MULTI | |
2691 && !one_exactly | |
2692 && !is_Magic(c))); ++len) | |
2693 { | |
2694 c = no_Magic(c); | |
2695 #ifdef FEAT_MBYTE | |
2696 if (has_mbyte) | |
2697 { | |
2698 regmbc(c); | |
2699 if (enc_utf8) | |
2700 { | |
2701 int l; | |
2702 | |
714 | 2703 /* Need to get composing character too. */ |
7 | 2704 for (;;) |
2705 { | |
714 | 2706 l = utf_ptr2len(regparse); |
2707 if (!UTF_COMPOSINGLIKE(regparse, regparse + l)) | |
7 | 2708 break; |
714 | 2709 regmbc(utf_ptr2char(regparse)); |
2710 skipchr(); | |
7 | 2711 } |
2712 } | |
2713 } | |
2714 else | |
2715 #endif | |
2716 regc(c); | |
2717 c = getchr(); | |
2718 } | |
2719 ungetchr(); | |
2720 | |
2721 regc(NUL); | |
2722 *flagp |= HASWIDTH; | |
2723 if (len == 1) | |
2724 *flagp |= SIMPLE; | |
2725 } | |
2726 break; | |
2727 } | |
2728 | |
2729 return ret; | |
2730 } | |
2731 | |
714 | 2732 #ifdef FEAT_MBYTE |
2733 /* | |
2734 * Return TRUE if MULTIBYTECODE should be used instead of EXACTLY for | |
2735 * character "c". | |
2736 */ | |
2737 static int | |
2738 use_multibytecode(c) | |
2739 int c; | |
2740 { | |
2741 return has_mbyte && (*mb_char2len)(c) > 1 | |
2742 && (re_multi_type(peekchr()) != NOT_MULTI | |
2743 || (enc_utf8 && utf_iscomposing(c))); | |
2744 } | |
2745 #endif | |
2746 | |
7 | 2747 /* |
4444 | 2748 * Emit a node. |
7 | 2749 * Return pointer to generated code. |
2750 */ | |
2751 static char_u * | |
2752 regnode(op) | |
2753 int op; | |
2754 { | |
2755 char_u *ret; | |
2756 | |
2757 ret = regcode; | |
2758 if (ret == JUST_CALC_SIZE) | |
2759 regsize += 3; | |
2760 else | |
2761 { | |
2762 *regcode++ = op; | |
2763 *regcode++ = NUL; /* Null "next" pointer. */ | |
2764 *regcode++ = NUL; | |
2765 } | |
2766 return ret; | |
2767 } | |
2768 | |
2769 /* | |
2770 * Emit (if appropriate) a byte of code | |
2771 */ | |
2772 static void | |
2773 regc(b) | |
2774 int b; | |
2775 { | |
2776 if (regcode == JUST_CALC_SIZE) | |
2777 regsize++; | |
2778 else | |
2779 *regcode++ = b; | |
2780 } | |
2781 | |
2782 #ifdef FEAT_MBYTE | |
2783 /* | |
2784 * Emit (if appropriate) a multi-byte character of code | |
2785 */ | |
2786 static void | |
2787 regmbc(c) | |
2788 int c; | |
2789 { | |
2974 | 2790 if (!has_mbyte && c > 0xff) |
2791 return; | |
7 | 2792 if (regcode == JUST_CALC_SIZE) |
2793 regsize += (*mb_char2len)(c); | |
2794 else | |
2795 regcode += (*mb_char2bytes)(c, regcode); | |
2796 } | |
2797 #endif | |
2798 | |
2799 /* | |
4444 | 2800 * Insert an operator in front of already-emitted operand |
7 | 2801 * |
2802 * Means relocating the operand. | |
2803 */ | |
2804 static void | |
2805 reginsert(op, opnd) | |
2806 int op; | |
2807 char_u *opnd; | |
2808 { | |
2809 char_u *src; | |
2810 char_u *dst; | |
2811 char_u *place; | |
2812 | |
2813 if (regcode == JUST_CALC_SIZE) | |
2814 { | |
2815 regsize += 3; | |
2816 return; | |
2817 } | |
2818 src = regcode; | |
2819 regcode += 3; | |
2820 dst = regcode; | |
2821 while (src > opnd) | |
2822 *--dst = *--src; | |
2823 | |
2824 place = opnd; /* Op node, where operand used to be. */ | |
2825 *place++ = op; | |
2826 *place++ = NUL; | |
2827 *place = NUL; | |
2828 } | |
2829 | |
2830 /* | |
4444 | 2831 * Insert an operator in front of already-emitted operand. |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2832 * Add a number to the operator. |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2833 */ |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2834 static void |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2835 reginsert_nr(op, val, opnd) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2836 int op; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2837 long val; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2838 char_u *opnd; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2839 { |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2840 char_u *src; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2841 char_u *dst; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2842 char_u *place; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2843 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2844 if (regcode == JUST_CALC_SIZE) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2845 { |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2846 regsize += 7; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2847 return; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2848 } |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2849 src = regcode; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2850 regcode += 7; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2851 dst = regcode; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2852 while (src > opnd) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2853 *--dst = *--src; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2854 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2855 place = opnd; /* Op node, where operand used to be. */ |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2856 *place++ = op; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2857 *place++ = NUL; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2858 *place++ = NUL; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2859 place = re_put_long(place, (long_u)val); |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2860 } |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2861 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2862 /* |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2863 * Insert an operator in front of already-emitted operand. |
7 | 2864 * The operator has the given limit values as operands. Also set next pointer. |
2865 * | |
2866 * Means relocating the operand. | |
2867 */ | |
2868 static void | |
2869 reginsert_limits(op, minval, maxval, opnd) | |
2870 int op; | |
2871 long minval; | |
2872 long maxval; | |
2873 char_u *opnd; | |
2874 { | |
2875 char_u *src; | |
2876 char_u *dst; | |
2877 char_u *place; | |
2878 | |
2879 if (regcode == JUST_CALC_SIZE) | |
2880 { | |
2881 regsize += 11; | |
2882 return; | |
2883 } | |
2884 src = regcode; | |
2885 regcode += 11; | |
2886 dst = regcode; | |
2887 while (src > opnd) | |
2888 *--dst = *--src; | |
2889 | |
2890 place = opnd; /* Op node, where operand used to be. */ | |
2891 *place++ = op; | |
2892 *place++ = NUL; | |
2893 *place++ = NUL; | |
2894 place = re_put_long(place, (long_u)minval); | |
2895 place = re_put_long(place, (long_u)maxval); | |
2896 regtail(opnd, place); | |
2897 } | |
2898 | |
2899 /* | |
2900 * Write a long as four bytes at "p" and return pointer to the next char. | |
2901 */ | |
2902 static char_u * | |
2903 re_put_long(p, val) | |
2904 char_u *p; | |
2905 long_u val; | |
2906 { | |
2907 *p++ = (char_u) ((val >> 24) & 0377); | |
2908 *p++ = (char_u) ((val >> 16) & 0377); | |
2909 *p++ = (char_u) ((val >> 8) & 0377); | |
2910 *p++ = (char_u) (val & 0377); | |
2911 return p; | |
2912 } | |
2913 | |
2914 /* | |
4444 | 2915 * Set the next-pointer at the end of a node chain. |
7 | 2916 */ |
2917 static void | |
2918 regtail(p, val) | |
2919 char_u *p; | |
2920 char_u *val; | |
2921 { | |
2922 char_u *scan; | |
2923 char_u *temp; | |
2924 int offset; | |
2925 | |
2926 if (p == JUST_CALC_SIZE) | |
2927 return; | |
2928 | |
2929 /* Find last node. */ | |
2930 scan = p; | |
2931 for (;;) | |
2932 { | |
2933 temp = regnext(scan); | |
2934 if (temp == NULL) | |
2935 break; | |
2936 scan = temp; | |
2937 } | |
2938 | |
233 | 2939 if (OP(scan) == BACK) |
7 | 2940 offset = (int)(scan - val); |
2941 else | |
2942 offset = (int)(val - scan); | |
2010 | 2943 /* When the offset uses more than 16 bits it can no longer fit in the two |
2974 | 2944 * bytes available. Use a global flag to avoid having to check return |
2010 | 2945 * values in too many places. */ |
2946 if (offset > 0xffff) | |
2947 reg_toolong = TRUE; | |
2948 else | |
2949 { | |
2950 *(scan + 1) = (char_u) (((unsigned)offset >> 8) & 0377); | |
2951 *(scan + 2) = (char_u) (offset & 0377); | |
2952 } | |
7 | 2953 } |
2954 | |
2955 /* | |
4444 | 2956 * Like regtail, on item after a BRANCH; nop if none. |
7 | 2957 */ |
2958 static void | |
2959 regoptail(p, val) | |
2960 char_u *p; | |
2961 char_u *val; | |
2962 { | |
2963 /* When op is neither BRANCH nor BRACE_COMPLEX0-9, it is "operandless" */ | |
2964 if (p == NULL || p == JUST_CALC_SIZE | |
2965 || (OP(p) != BRANCH | |
2966 && (OP(p) < BRACE_COMPLEX || OP(p) > BRACE_COMPLEX + 9))) | |
2967 return; | |
2968 regtail(OPERAND(p), val); | |
2969 } | |
2970 | |
2971 /* | |
4444 | 2972 * Functions for getting characters from the regexp input. |
7 | 2973 */ |
2974 | |
2975 static int at_start; /* True when on the first character */ | |
2976 static int prev_at_start; /* True when on the second character */ | |
2977 | |
4444 | 2978 /* |
2979 * Start parsing at "str". | |
2980 */ | |
7 | 2981 static void |
2982 initchr(str) | |
2983 char_u *str; | |
2984 { | |
2985 regparse = str; | |
2986 prevchr_len = 0; | |
2987 curchr = prevprevchr = prevchr = nextchr = -1; | |
2988 at_start = TRUE; | |
2989 prev_at_start = FALSE; | |
2990 } | |
2991 | |
4444 | 2992 /* |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2993 * Save the current parse state, so that it can be restored and parsing |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2994 * starts in the same state again. |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2995 */ |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2996 static void |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2997 save_parse_state(ps) |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2998 parse_state_T *ps; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2999 { |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3000 ps->regparse = regparse; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3001 ps->prevchr_len = prevchr_len; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3002 ps->curchr = curchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3003 ps->prevchr = prevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3004 ps->prevprevchr = prevprevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3005 ps->nextchr = nextchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3006 ps->at_start = at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3007 ps->prev_at_start = prev_at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3008 ps->regnpar = regnpar; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3009 } |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3010 |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3011 /* |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3012 * Restore a previously saved parse state. |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3013 */ |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3014 static void |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3015 restore_parse_state(ps) |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3016 parse_state_T *ps; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3017 { |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3018 regparse = ps->regparse; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3019 prevchr_len = ps->prevchr_len; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3020 curchr = ps->curchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3021 prevchr = ps->prevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3022 prevprevchr = ps->prevprevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3023 nextchr = ps->nextchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3024 at_start = ps->at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3025 prev_at_start = ps->prev_at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3026 regnpar = ps->regnpar; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3027 } |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3028 |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3029 |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3030 /* |
4444 | 3031 * Get the next character without advancing. |
3032 */ | |
7 | 3033 static int |
3034 peekchr() | |
3035 { | |
167 | 3036 static int after_slash = FALSE; |
3037 | |
7 | 3038 if (curchr == -1) |
3039 { | |
3040 switch (curchr = regparse[0]) | |
3041 { | |
3042 case '.': | |
3043 case '[': | |
3044 case '~': | |
3045 /* magic when 'magic' is on */ | |
3046 if (reg_magic >= MAGIC_ON) | |
3047 curchr = Magic(curchr); | |
3048 break; | |
3049 case '(': | |
3050 case ')': | |
3051 case '{': | |
3052 case '%': | |
3053 case '+': | |
3054 case '=': | |
3055 case '?': | |
3056 case '@': | |
3057 case '!': | |
3058 case '&': | |
3059 case '|': | |
3060 case '<': | |
3061 case '>': | |
3062 case '#': /* future ext. */ | |
3063 case '"': /* future ext. */ | |
3064 case '\'': /* future ext. */ | |
3065 case ',': /* future ext. */ | |
3066 case '-': /* future ext. */ | |
3067 case ':': /* future ext. */ | |
3068 case ';': /* future ext. */ | |
3069 case '`': /* future ext. */ | |
3070 case '/': /* Can't be used in / command */ | |
3071 /* magic only after "\v" */ | |
3072 if (reg_magic == MAGIC_ALL) | |
3073 curchr = Magic(curchr); | |
3074 break; | |
3075 case '*': | |
167 | 3076 /* * is not magic as the very first character, eg "?*ptr", when |
3077 * after '^', eg "/^*ptr" and when after "\(", "\|", "\&". But | |
3078 * "\(\*" is not magic, thus must be magic if "after_slash" */ | |
3079 if (reg_magic >= MAGIC_ON | |
3080 && !at_start | |
3081 && !(prev_at_start && prevchr == Magic('^')) | |
3082 && (after_slash | |
3083 || (prevchr != Magic('(') | |
3084 && prevchr != Magic('&') | |
3085 && prevchr != Magic('|')))) | |
7 | 3086 curchr = Magic('*'); |
3087 break; | |
3088 case '^': | |
3089 /* '^' is only magic as the very first character and if it's after | |
3090 * "\(", "\|", "\&' or "\n" */ | |
3091 if (reg_magic >= MAGIC_OFF | |
3092 && (at_start | |
3093 || reg_magic == MAGIC_ALL | |
3094 || prevchr == Magic('(') | |
3095 || prevchr == Magic('|') | |
3096 || prevchr == Magic('&') | |
3097 || prevchr == Magic('n') | |
3098 || (no_Magic(prevchr) == '(' | |
3099 && prevprevchr == Magic('%')))) | |
3100 { | |
3101 curchr = Magic('^'); | |
3102 at_start = TRUE; | |
3103 prev_at_start = FALSE; | |
3104 } | |
3105 break; | |
3106 case '$': | |
3107 /* '$' is only magic as the very last char and if it's in front of | |
3108 * either "\|", "\)", "\&", or "\n" */ | |
3109 if (reg_magic >= MAGIC_OFF) | |
3110 { | |
3111 char_u *p = regparse + 1; | |
6041 | 3112 int is_magic_all = (reg_magic == MAGIC_ALL); |
3113 | |
3114 /* ignore \c \C \m \M \v \V and \Z after '$' */ | |
7 | 3115 while (p[0] == '\\' && (p[1] == 'c' || p[1] == 'C' |
6041 | 3116 || p[1] == 'm' || p[1] == 'M' |
3117 || p[1] == 'v' || p[1] == 'V' || p[1] == 'Z')) | |
3118 { | |
3119 if (p[1] == 'v') | |
3120 is_magic_all = TRUE; | |
3121 else if (p[1] == 'm' || p[1] == 'M' || p[1] == 'V') | |
3122 is_magic_all = FALSE; | |
7 | 3123 p += 2; |
6041 | 3124 } |
7 | 3125 if (p[0] == NUL |
3126 || (p[0] == '\\' | |
3127 && (p[1] == '|' || p[1] == '&' || p[1] == ')' | |
3128 || p[1] == 'n')) | |
6041 | 3129 || (is_magic_all |
3130 && (p[0] == '|' || p[0] == '&' || p[0] == ')')) | |
7 | 3131 || reg_magic == MAGIC_ALL) |
3132 curchr = Magic('$'); | |
3133 } | |
3134 break; | |
3135 case '\\': | |
3136 { | |
3137 int c = regparse[1]; | |
3138 | |
3139 if (c == NUL) | |
3140 curchr = '\\'; /* trailing '\' */ | |
3141 else if ( | |
3142 #ifdef EBCDIC | |
3143 vim_strchr(META, c) | |
3144 #else | |
3145 c <= '~' && META_flags[c] | |
3146 #endif | |
3147 ) | |
3148 { | |
3149 /* | |
3150 * META contains everything that may be magic sometimes, | |
3151 * except ^ and $ ("\^" and "\$" are only magic after | |
3152 * "\v"). We now fetch the next character and toggle its | |
3153 * magicness. Therefore, \ is so meta-magic that it is | |
3154 * not in META. | |
3155 */ | |
3156 curchr = -1; | |
3157 prev_at_start = at_start; | |
3158 at_start = FALSE; /* be able to say "/\*ptr" */ | |
3159 ++regparse; | |
167 | 3160 ++after_slash; |
7 | 3161 peekchr(); |
3162 --regparse; | |
167 | 3163 --after_slash; |
7 | 3164 curchr = toggle_Magic(curchr); |
3165 } | |
3166 else if (vim_strchr(REGEXP_ABBR, c)) | |
3167 { | |
3168 /* | |
3169 * Handle abbreviations, like "\t" for TAB -- webb | |
3170 */ | |
3171 curchr = backslash_trans(c); | |
3172 } | |
3173 else if (reg_magic == MAGIC_NONE && (c == '$' || c == '^')) | |
3174 curchr = toggle_Magic(c); | |
3175 else | |
3176 { | |
3177 /* | |
3178 * Next character can never be (made) magic? | |
3179 * Then backslashing it won't do anything. | |
3180 */ | |
3181 #ifdef FEAT_MBYTE | |
3182 if (has_mbyte) | |
3183 curchr = (*mb_ptr2char)(regparse + 1); | |
3184 else | |
3185 #endif | |
3186 curchr = c; | |
3187 } | |
3188 break; | |
3189 } | |
3190 | |
3191 #ifdef FEAT_MBYTE | |
3192 default: | |
3193 if (has_mbyte) | |
3194 curchr = (*mb_ptr2char)(regparse); | |
3195 #endif | |
3196 } | |
3197 } | |
3198 | |
3199 return curchr; | |
3200 } | |
3201 | |
3202 /* | |
3203 * Eat one lexed character. Do this in a way that we can undo it. | |
3204 */ | |
3205 static void | |
3206 skipchr() | |
3207 { | |
3208 /* peekchr() eats a backslash, do the same here */ | |
3209 if (*regparse == '\\') | |
3210 prevchr_len = 1; | |
3211 else | |
3212 prevchr_len = 0; | |
3213 if (regparse[prevchr_len] != NUL) | |
3214 { | |
3215 #ifdef FEAT_MBYTE | |
714 | 3216 if (enc_utf8) |
1449 | 3217 /* exclude composing chars that mb_ptr2len does include */ |
3218 prevchr_len += utf_ptr2len(regparse + prevchr_len); | |
714 | 3219 else if (has_mbyte) |
474 | 3220 prevchr_len += (*mb_ptr2len)(regparse + prevchr_len); |
7 | 3221 else |
3222 #endif | |
3223 ++prevchr_len; | |
3224 } | |
3225 regparse += prevchr_len; | |
3226 prev_at_start = at_start; | |
3227 at_start = FALSE; | |
3228 prevprevchr = prevchr; | |
3229 prevchr = curchr; | |
3230 curchr = nextchr; /* use previously unget char, or -1 */ | |
3231 nextchr = -1; | |
3232 } | |
3233 | |
3234 /* | |
3235 * Skip a character while keeping the value of prev_at_start for at_start. | |
3236 * prevchr and prevprevchr are also kept. | |
3237 */ | |
3238 static void | |
3239 skipchr_keepstart() | |
3240 { | |
3241 int as = prev_at_start; | |
3242 int pr = prevchr; | |
3243 int prpr = prevprevchr; | |
3244 | |
3245 skipchr(); | |
3246 at_start = as; | |
3247 prevchr = pr; | |
3248 prevprevchr = prpr; | |
3249 } | |
3250 | |
4444 | 3251 /* |
3252 * Get the next character from the pattern. We know about magic and such, so | |
3253 * therefore we need a lexical analyzer. | |
3254 */ | |
7 | 3255 static int |
3256 getchr() | |
3257 { | |
3258 int chr = peekchr(); | |
3259 | |
3260 skipchr(); | |
3261 return chr; | |
3262 } | |
3263 | |
3264 /* | |
3265 * put character back. Works only once! | |
3266 */ | |
3267 static void | |
3268 ungetchr() | |
3269 { | |
3270 nextchr = curchr; | |
3271 curchr = prevchr; | |
3272 prevchr = prevprevchr; | |
3273 at_start = prev_at_start; | |
3274 prev_at_start = FALSE; | |
3275 | |
3276 /* Backup regparse, so that it's at the same position as before the | |
3277 * getchr(). */ | |
3278 regparse -= prevchr_len; | |
3279 } | |
3280 | |
3281 /* | |
29 | 3282 * Get and return the value of the hex string at the current position. |
3283 * Return -1 if there is no valid hex number. | |
3284 * The position is updated: | |
24 | 3285 * blahblah\%x20asdf |
856 | 3286 * before-^ ^-after |
24 | 3287 * The parameter controls the maximum number of input characters. This will be |
3288 * 2 when reading a \%x20 sequence and 4 when reading a \%u20AC sequence. | |
3289 */ | |
3290 static int | |
3291 gethexchrs(maxinputlen) | |
3292 int maxinputlen; | |
3293 { | |
3294 int nr = 0; | |
3295 int c; | |
3296 int i; | |
3297 | |
3298 for (i = 0; i < maxinputlen; ++i) | |
3299 { | |
3300 c = regparse[0]; | |
3301 if (!vim_isxdigit(c)) | |
3302 break; | |
3303 nr <<= 4; | |
3304 nr |= hex2nr(c); | |
3305 ++regparse; | |
3306 } | |
3307 | |
3308 if (i == 0) | |
3309 return -1; | |
3310 return nr; | |
3311 } | |
3312 | |
3313 /* | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3314 * Get and return the value of the decimal string immediately after the |
24 | 3315 * current position. Return -1 for invalid. Consumes all digits. |
3316 */ | |
3317 static int | |
3318 getdecchrs() | |
3319 { | |
3320 int nr = 0; | |
3321 int c; | |
3322 int i; | |
3323 | |
3324 for (i = 0; ; ++i) | |
3325 { | |
3326 c = regparse[0]; | |
3327 if (c < '0' || c > '9') | |
3328 break; | |
3329 nr *= 10; | |
3330 nr += c - '0'; | |
3331 ++regparse; | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3332 curchr = -1; /* no longer valid */ |
24 | 3333 } |
3334 | |
3335 if (i == 0) | |
3336 return -1; | |
3337 return nr; | |
3338 } | |
3339 | |
3340 /* | |
3341 * get and return the value of the octal string immediately after the current | |
3342 * position. Return -1 for invalid, or 0-255 for valid. Smart enough to handle | |
3343 * numbers > 377 correctly (for example, 400 is treated as 40) and doesn't | |
3344 * treat 8 or 9 as recognised characters. Position is updated: | |
3345 * blahblah\%o210asdf | |
856 | 3346 * before-^ ^-after |
24 | 3347 */ |
3348 static int | |
3349 getoctchrs() | |
3350 { | |
3351 int nr = 0; | |
3352 int c; | |
3353 int i; | |
3354 | |
3355 for (i = 0; i < 3 && nr < 040; ++i) | |
3356 { | |
3357 c = regparse[0]; | |
3358 if (c < '0' || c > '7') | |
3359 break; | |
3360 nr <<= 3; | |
3361 nr |= hex2nr(c); | |
3362 ++regparse; | |
3363 } | |
3364 | |
3365 if (i == 0) | |
3366 return -1; | |
3367 return nr; | |
3368 } | |
3369 | |
3370 /* | |
3371 * Get a number after a backslash that is inside []. | |
3372 * When nothing is recognized return a backslash. | |
3373 */ | |
3374 static int | |
3375 coll_get_char() | |
3376 { | |
3377 int nr = -1; | |
3378 | |
3379 switch (*regparse++) | |
3380 { | |
3381 case 'd': nr = getdecchrs(); break; | |
3382 case 'o': nr = getoctchrs(); break; | |
3383 case 'x': nr = gethexchrs(2); break; | |
3384 case 'u': nr = gethexchrs(4); break; | |
3385 case 'U': nr = gethexchrs(8); break; | |
3386 } | |
3387 if (nr < 0) | |
3388 { | |
3389 /* If getting the number fails be backwards compatible: the character | |
3390 * is a backslash. */ | |
3391 --regparse; | |
3392 nr = '\\'; | |
3393 } | |
3394 return nr; | |
3395 } | |
3396 | |
3397 /* | |
7 | 3398 * read_limits - Read two integers to be taken as a minimum and maximum. |
3399 * If the first character is '-', then the range is reversed. | |
3400 * Should end with 'end'. If minval is missing, zero is default, if maxval is | |
3401 * missing, a very big number is the default. | |
3402 */ | |
3403 static int | |
3404 read_limits(minval, maxval) | |
3405 long *minval; | |
3406 long *maxval; | |
3407 { | |
3408 int reverse = FALSE; | |
3409 char_u *first_char; | |
3410 long tmp; | |
3411 | |
3412 if (*regparse == '-') | |
3413 { | |
3414 /* Starts with '-', so reverse the range later */ | |
3415 regparse++; | |
3416 reverse = TRUE; | |
3417 } | |
3418 first_char = regparse; | |
3419 *minval = getdigits(®parse); | |
3420 if (*regparse == ',') /* There is a comma */ | |
3421 { | |
3422 if (vim_isdigit(*++regparse)) | |
3423 *maxval = getdigits(®parse); | |
3424 else | |
3425 *maxval = MAX_LIMIT; | |
3426 } | |
3427 else if (VIM_ISDIGIT(*first_char)) | |
3428 *maxval = *minval; /* It was \{n} or \{-n} */ | |
3429 else | |
3430 *maxval = MAX_LIMIT; /* It was \{} or \{-} */ | |
3431 if (*regparse == '\\') | |
3432 regparse++; /* Allow either \{...} or \{...\} */ | |
167 | 3433 if (*regparse != '}') |
7 | 3434 { |
3435 sprintf((char *)IObuff, _("E554: Syntax error in %s{...}"), | |
3436 reg_magic == MAGIC_ALL ? "" : "\\"); | |
3437 EMSG_RET_FAIL(IObuff); | |
3438 } | |
3439 | |
3440 /* | |
3441 * Reverse the range if there was a '-', or make sure it is in the right | |
3442 * order otherwise. | |
3443 */ | |
3444 if ((!reverse && *minval > *maxval) || (reverse && *minval < *maxval)) | |
3445 { | |
3446 tmp = *minval; | |
3447 *minval = *maxval; | |
3448 *maxval = tmp; | |
3449 } | |
3450 skipchr(); /* let's be friends with the lexer again */ | |
3451 return OK; | |
3452 } | |
3453 | |
3454 /* | |
3455 * vim_regexec and friends | |
3456 */ | |
3457 | |
3458 /* | |
3459 * Global work variables for vim_regexec(). | |
3460 */ | |
3461 | |
3462 /* The current match-position is remembered with these variables: */ | |
3463 static linenr_T reglnum; /* line number, relative to first line */ | |
3464 static char_u *regline; /* start of current line */ | |
3465 static char_u *reginput; /* current input, points into "regline" */ | |
3466 | |
3467 static int need_clear_subexpr; /* subexpressions still need to be | |
3468 * cleared */ | |
3469 #ifdef FEAT_SYN_HL | |
3470 static int need_clear_zsubexpr = FALSE; /* extmatch subexpressions | |
3471 * still need to be cleared */ | |
3472 #endif | |
3473 | |
3474 /* | |
3475 * Structure used to save the current input state, when it needs to be | |
3476 * restored after trying a match. Used by reg_save() and reg_restore(). | |
233 | 3477 * Also stores the length of "backpos". |
7 | 3478 */ |
3479 typedef struct | |
3480 { | |
3481 union | |
3482 { | |
3483 char_u *ptr; /* reginput pointer, for single-line regexp */ | |
3484 lpos_T pos; /* reginput pos, for multi-line regexp */ | |
3485 } rs_u; | |
233 | 3486 int rs_len; |
7 | 3487 } regsave_T; |
3488 | |
3489 /* struct to save start/end pointer/position in for \(\) */ | |
3490 typedef struct | |
3491 { | |
3492 union | |
3493 { | |
3494 char_u *ptr; | |
3495 lpos_T pos; | |
3496 } se_u; | |
3497 } save_se_T; | |
3498 | |
1579 | 3499 /* used for BEHIND and NOBEHIND matching */ |
3500 typedef struct regbehind_S | |
3501 { | |
3502 regsave_T save_after; | |
3503 regsave_T save_behind; | |
1602 | 3504 int save_need_clear_subexpr; |
1579 | 3505 save_se_T save_start[NSUBEXP]; |
3506 save_se_T save_end[NSUBEXP]; | |
3507 } regbehind_T; | |
3508 | |
7 | 3509 static char_u *reg_getline __ARGS((linenr_T lnum)); |
4444 | 3510 static long bt_regexec_both __ARGS((char_u *line, colnr_T col, proftime_T *tm)); |
3511 static long regtry __ARGS((bt_regprog_T *prog, colnr_T col)); | |
7 | 3512 static void cleanup_subexpr __ARGS((void)); |
3513 #ifdef FEAT_SYN_HL | |
3514 static void cleanup_zsubexpr __ARGS((void)); | |
3515 #endif | |
1579 | 3516 static void save_subexpr __ARGS((regbehind_T *bp)); |
3517 static void restore_subexpr __ARGS((regbehind_T *bp)); | |
7 | 3518 static void reg_nextline __ARGS((void)); |
233 | 3519 static void reg_save __ARGS((regsave_T *save, garray_T *gap)); |
3520 static void reg_restore __ARGS((regsave_T *save, garray_T *gap)); | |
7 | 3521 static int reg_save_equal __ARGS((regsave_T *save)); |
3522 static void save_se_multi __ARGS((save_se_T *savep, lpos_T *posp)); | |
3523 static void save_se_one __ARGS((save_se_T *savep, char_u **pp)); | |
3524 | |
3525 /* Save the sub-expressions before attempting a match. */ | |
3526 #define save_se(savep, posp, pp) \ | |
3527 REG_MULTI ? save_se_multi((savep), (posp)) : save_se_one((savep), (pp)) | |
3528 | |
3529 /* After a failed match restore the sub-expressions. */ | |
3530 #define restore_se(savep, posp, pp) { \ | |
3531 if (REG_MULTI) \ | |
3532 *(posp) = (savep)->se_u.pos; \ | |
3533 else \ | |
3534 *(pp) = (savep)->se_u.ptr; } | |
3535 | |
3536 static int re_num_cmp __ARGS((long_u val, char_u *scan)); | |
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
3537 static int match_with_backref __ARGS((linenr_T start_lnum, colnr_T start_col, linenr_T end_lnum, colnr_T end_col, int *bytelen)); |
180 | 3538 static int regmatch __ARGS((char_u *prog)); |
7 | 3539 static int regrepeat __ARGS((char_u *p, long maxcount)); |
3540 | |
3541 #ifdef DEBUG | |
3542 int regnarrate = 0; | |
3543 #endif | |
3544 | |
3545 /* | |
3546 * Internal copy of 'ignorecase'. It is set at each call to vim_regexec(). | |
3547 * Normally it gets the value of "rm_ic" or "rmm_ic", but when the pattern | |
3548 * contains '\c' or '\C' the value is overruled. | |
3549 */ | |
3550 static int ireg_ic; | |
3551 | |
3552 #ifdef FEAT_MBYTE | |
3553 /* | |
3554 * Similar to ireg_ic, but only for 'combining' characters. Set with \Z flag | |
3555 * in the regexp. Defaults to false, always. | |
3556 */ | |
3557 static int ireg_icombine; | |
3558 #endif | |
3559 | |
3560 /* | |
410 | 3561 * Copy of "rmm_maxcol": maximum column to search for a match. Zero when |
3562 * there is no maximum. | |
3563 */ | |
418 | 3564 static colnr_T ireg_maxcol; |
410 | 3565 |
3566 /* | |
7 | 3567 * Sometimes need to save a copy of a line. Since alloc()/free() is very |
3568 * slow, we keep one allocated piece of memory and only re-allocate it when | |
4444 | 3569 * it's too small. It's freed in bt_regexec_both() when finished. |
7 | 3570 */ |
1468 | 3571 static char_u *reg_tofree = NULL; |
7 | 3572 static unsigned reg_tofreelen; |
3573 | |
3574 /* | |
3575 * These variables are set when executing a regexp to speed up the execution. | |
1209 | 3576 * Which ones are set depends on whether a single-line or multi-line match is |
7 | 3577 * done: |
3578 * single-line multi-line | |
3579 * reg_match ®match_T NULL | |
3580 * reg_mmatch NULL ®mmatch_T | |
3581 * reg_startp reg_match->startp <invalid> | |
3582 * reg_endp reg_match->endp <invalid> | |
3583 * reg_startpos <invalid> reg_mmatch->startpos | |
3584 * reg_endpos <invalid> reg_mmatch->endpos | |
3585 * reg_win NULL window in which to search | |
4061 | 3586 * reg_buf curbuf buffer in which to search |
7 | 3587 * reg_firstlnum <invalid> first line in which to search |
3588 * reg_maxline 0 last line nr | |
3589 * reg_line_lbr FALSE or TRUE FALSE | |
3590 */ | |
3591 static regmatch_T *reg_match; | |
3592 static regmmatch_T *reg_mmatch; | |
3593 static char_u **reg_startp = NULL; | |
3594 static char_u **reg_endp = NULL; | |
3595 static lpos_T *reg_startpos = NULL; | |
3596 static lpos_T *reg_endpos = NULL; | |
3597 static win_T *reg_win; | |
3598 static buf_T *reg_buf; | |
3599 static linenr_T reg_firstlnum; | |
3600 static linenr_T reg_maxline; | |
3601 static int reg_line_lbr; /* "\n" in string is line break */ | |
3602 | |
270 | 3603 /* Values for rs_state in regitem_T. */ |
3604 typedef enum regstate_E | |
3605 { | |
3606 RS_NOPEN = 0 /* NOPEN and NCLOSE */ | |
3607 , RS_MOPEN /* MOPEN + [0-9] */ | |
3608 , RS_MCLOSE /* MCLOSE + [0-9] */ | |
3609 #ifdef FEAT_SYN_HL | |
3610 , RS_ZOPEN /* ZOPEN + [0-9] */ | |
3611 , RS_ZCLOSE /* ZCLOSE + [0-9] */ | |
3612 #endif | |
3613 , RS_BRANCH /* BRANCH */ | |
3614 , RS_BRCPLX_MORE /* BRACE_COMPLEX and trying one more match */ | |
3615 , RS_BRCPLX_LONG /* BRACE_COMPLEX and trying longest match */ | |
3616 , RS_BRCPLX_SHORT /* BRACE_COMPLEX and trying shortest match */ | |
3617 , RS_NOMATCH /* NOMATCH */ | |
3618 , RS_BEHIND1 /* BEHIND / NOBEHIND matching rest */ | |
3619 , RS_BEHIND2 /* BEHIND / NOBEHIND matching behind part */ | |
3620 , RS_STAR_LONG /* STAR/PLUS/BRACE_SIMPLE longest match */ | |
3621 , RS_STAR_SHORT /* STAR/PLUS/BRACE_SIMPLE shortest match */ | |
3622 } regstate_T; | |
3623 | |
3624 /* | |
3625 * When there are alternatives a regstate_T is put on the regstack to remember | |
3626 * what we are doing. | |
3627 * Before it may be another type of item, depending on rs_state, to remember | |
3628 * more things. | |
3629 */ | |
3630 typedef struct regitem_S | |
3631 { | |
3632 regstate_T rs_state; /* what we are doing, one of RS_ above */ | |
3633 char_u *rs_scan; /* current node in program */ | |
3634 union | |
3635 { | |
3636 save_se_T sesave; | |
3637 regsave_T regsave; | |
3638 } rs_un; /* room for saving reginput */ | |
1579 | 3639 short rs_no; /* submatch nr or BEHIND/NOBEHIND */ |
270 | 3640 } regitem_T; |
3641 | |
3642 static regitem_T *regstack_push __ARGS((regstate_T state, char_u *scan)); | |
3643 static void regstack_pop __ARGS((char_u **scan)); | |
3644 | |
3645 /* used for STAR, PLUS and BRACE_SIMPLE matching */ | |
3646 typedef struct regstar_S | |
3647 { | |
3648 int nextb; /* next byte */ | |
3649 int nextb_ic; /* next byte reverse case */ | |
3650 long count; | |
3651 long minval; | |
3652 long maxval; | |
3653 } regstar_T; | |
3654 | |
3655 /* used to store input position when a BACK was encountered, so that we now if | |
3656 * we made any progress since the last time. */ | |
3657 typedef struct backpos_S | |
3658 { | |
3659 char_u *bp_scan; /* "scan" where BACK was encountered */ | |
3660 regsave_T bp_pos; /* last input position */ | |
3661 } backpos_T; | |
3662 | |
3663 /* | |
1520 | 3664 * "regstack" and "backpos" are used by regmatch(). They are kept over calls |
3665 * to avoid invoking malloc() and free() often. | |
3666 * "regstack" is a stack with regitem_T items, sometimes preceded by regstar_T | |
3667 * or regbehind_T. | |
3668 * "backpos_T" is a table with backpos_T for BACK | |
270 | 3669 */ |
1520 | 3670 static garray_T regstack = {0, 0, 0, 0, NULL}; |
3671 static garray_T backpos = {0, 0, 0, 0, NULL}; | |
3672 | |
3673 /* | |
3674 * Both for regstack and backpos tables we use the following strategy of | |
3675 * allocation (to reduce malloc/free calls): | |
3676 * - Initial size is fairly small. | |
3677 * - When needed, the tables are grown bigger (8 times at first, double after | |
3678 * that). | |
3679 * - After executing the match we free the memory only if the array has grown. | |
3680 * Thus the memory is kept allocated when it's at the initial size. | |
3681 * This makes it fast while not keeping a lot of memory allocated. | |
3682 * A three times speed increase was observed when using many simple patterns. | |
3683 */ | |
3684 #define REGSTACK_INITIAL 2048 | |
3685 #define BACKPOS_INITIAL 64 | |
3686 | |
3687 #if defined(EXITFREE) || defined(PROTO) | |
3688 void | |
3689 free_regexp_stuff() | |
3690 { | |
3691 ga_clear(®stack); | |
3692 ga_clear(&backpos); | |
3693 vim_free(reg_tofree); | |
3694 vim_free(reg_prev_sub); | |
3695 } | |
3696 #endif | |
270 | 3697 |
7 | 3698 /* |
3699 * Get pointer to the line "lnum", which is relative to "reg_firstlnum". | |
3700 */ | |
3701 static char_u * | |
3702 reg_getline(lnum) | |
3703 linenr_T lnum; | |
3704 { | |
3705 /* when looking behind for a match/no-match lnum is negative. But we | |
3706 * can't go before line 1 */ | |
3707 if (reg_firstlnum + lnum < 1) | |
3708 return NULL; | |
482 | 3709 if (lnum > reg_maxline) |
481 | 3710 /* Must have matched the "\n" in the last line. */ |
3711 return (char_u *)""; | |
7 | 3712 return ml_get_buf(reg_buf, reg_firstlnum + lnum, FALSE); |
3713 } | |
3714 | |
3715 static regsave_T behind_pos; | |
3716 | |
3717 #ifdef FEAT_SYN_HL | |
3718 static char_u *reg_startzp[NSUBEXP]; /* Workspace to mark beginning */ | |
3719 static char_u *reg_endzp[NSUBEXP]; /* and end of \z(...\) matches */ | |
3720 static lpos_T reg_startzpos[NSUBEXP]; /* idem, beginning pos */ | |
3721 static lpos_T reg_endzpos[NSUBEXP]; /* idem, end pos */ | |
3722 #endif | |
3723 | |
3724 /* TRUE if using multi-line regexp. */ | |
3725 #define REG_MULTI (reg_match == NULL) | |
3726 | |
5838 | 3727 static int bt_regexec_nl __ARGS((regmatch_T *rmp, char_u *line, colnr_T col, int line_lbr)); |
3728 | |
4444 | 3729 |
7 | 3730 /* |
3731 * Match a regexp against a string. | |
3732 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
3733 * Uses curbuf for line count and 'iskeyword'. | |
5838 | 3734 * if "line_lbr" is TRUE consider a "\n" in "line" to be a line break. |
7 | 3735 * |
3736 * Return TRUE if there is a match, FALSE if not. | |
3737 */ | |
4444 | 3738 static int |
5838 | 3739 bt_regexec_nl(rmp, line, col, line_lbr) |
7 | 3740 regmatch_T *rmp; |
3741 char_u *line; /* string to match against */ | |
3742 colnr_T col; /* column to start looking for match */ | |
5838 | 3743 int line_lbr; |
7 | 3744 { |
3745 reg_match = rmp; | |
3746 reg_mmatch = NULL; | |
3747 reg_maxline = 0; | |
5838 | 3748 reg_line_lbr = line_lbr; |
4061 | 3749 reg_buf = curbuf; |
7 | 3750 reg_win = NULL; |
3751 ireg_ic = rmp->rm_ic; | |
3752 #ifdef FEAT_MBYTE | |
3753 ireg_icombine = FALSE; | |
3754 #endif | |
410 | 3755 ireg_maxcol = 0; |
4444 | 3756 return (bt_regexec_both(line, col, NULL) != 0); |
7 | 3757 } |
3758 | |
4444 | 3759 static long bt_regexec_multi __ARGS((regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm)); |
3760 | |
7 | 3761 /* |
3762 * Match a regexp against multiple lines. | |
3763 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
3764 * Uses curbuf for line count and 'iskeyword'. | |
3765 * | |
3766 * Return zero if there is no match. Return number of lines contained in the | |
3767 * match otherwise. | |
3768 */ | |
4444 | 3769 static long |
3770 bt_regexec_multi(rmp, win, buf, lnum, col, tm) | |
7 | 3771 regmmatch_T *rmp; |
3772 win_T *win; /* window in which to search or NULL */ | |
3773 buf_T *buf; /* buffer in which to search */ | |
3774 linenr_T lnum; /* nr of line to start looking for match */ | |
3775 colnr_T col; /* column to start looking for match */ | |
1521 | 3776 proftime_T *tm; /* timeout limit or NULL */ |
7 | 3777 { |
3778 long r; | |
3779 | |
3780 reg_match = NULL; | |
3781 reg_mmatch = rmp; | |
3782 reg_buf = buf; | |
3783 reg_win = win; | |
3784 reg_firstlnum = lnum; | |
3785 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
3786 reg_line_lbr = FALSE; | |
3787 ireg_ic = rmp->rmm_ic; | |
3788 #ifdef FEAT_MBYTE | |
3789 ireg_icombine = FALSE; | |
3790 #endif | |
410 | 3791 ireg_maxcol = rmp->rmm_maxcol; |
7 | 3792 |
4444 | 3793 r = bt_regexec_both(NULL, col, tm); |
7 | 3794 |
3795 return r; | |
3796 } | |
3797 | |
3798 /* | |
3799 * Match a regexp against a string ("line" points to the string) or multiple | |
3800 * lines ("line" is NULL, use reg_getline()). | |
3801 */ | |
3802 static long | |
4444 | 3803 bt_regexec_both(line, col, tm) |
7 | 3804 char_u *line; |
3805 colnr_T col; /* column to start looking for match */ | |
1877 | 3806 proftime_T *tm UNUSED; /* timeout limit or NULL */ |
7 | 3807 { |
4444 | 3808 bt_regprog_T *prog; |
7 | 3809 char_u *s; |
189 | 3810 long retval = 0L; |
7 | 3811 |
1520 | 3812 /* Create "regstack" and "backpos" if they are not allocated yet. |
3813 * We allocate *_INITIAL amount of bytes first and then set the grow size | |
3814 * to much bigger value to avoid many malloc calls in case of deep regular | |
3815 * expressions. */ | |
3816 if (regstack.ga_data == NULL) | |
3817 { | |
3818 /* Use an item size of 1 byte, since we push different things | |
3819 * onto the regstack. */ | |
3820 ga_init2(®stack, 1, REGSTACK_INITIAL); | |
3821 ga_grow(®stack, REGSTACK_INITIAL); | |
3822 regstack.ga_growsize = REGSTACK_INITIAL * 8; | |
3823 } | |
3824 | |
3825 if (backpos.ga_data == NULL) | |
3826 { | |
3827 ga_init2(&backpos, sizeof(backpos_T), BACKPOS_INITIAL); | |
3828 ga_grow(&backpos, BACKPOS_INITIAL); | |
3829 backpos.ga_growsize = BACKPOS_INITIAL * 8; | |
3830 } | |
270 | 3831 |
7 | 3832 if (REG_MULTI) |
3833 { | |
4444 | 3834 prog = (bt_regprog_T *)reg_mmatch->regprog; |
7 | 3835 line = reg_getline((linenr_T)0); |
3836 reg_startpos = reg_mmatch->startpos; | |
3837 reg_endpos = reg_mmatch->endpos; | |
3838 } | |
3839 else | |
3840 { | |
4444 | 3841 prog = (bt_regprog_T *)reg_match->regprog; |
7 | 3842 reg_startp = reg_match->startp; |
3843 reg_endp = reg_match->endp; | |
3844 } | |
3845 | |
3846 /* Be paranoid... */ | |
3847 if (prog == NULL || line == NULL) | |
3848 { | |
3849 EMSG(_(e_null)); | |
3850 goto theend; | |
3851 } | |
3852 | |
3853 /* Check validity of program. */ | |
3854 if (prog_magic_wrong()) | |
3855 goto theend; | |
3856 | |
410 | 3857 /* If the start column is past the maximum column: no need to try. */ |
3858 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
3859 goto theend; | |
3860 | |
7 | 3861 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ |
3862 if (prog->regflags & RF_ICASE) | |
3863 ireg_ic = TRUE; | |
3864 else if (prog->regflags & RF_NOICASE) | |
3865 ireg_ic = FALSE; | |
3866 | |
3867 #ifdef FEAT_MBYTE | |
3868 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
3869 if (prog->regflags & RF_ICOMBINE) | |
3870 ireg_icombine = TRUE; | |
3871 #endif | |
3872 | |
3873 /* If there is a "must appear" string, look for it. */ | |
3874 if (prog->regmust != NULL) | |
3875 { | |
3876 int c; | |
3877 | |
3878 #ifdef FEAT_MBYTE | |
3879 if (has_mbyte) | |
3880 c = (*mb_ptr2char)(prog->regmust); | |
3881 else | |
3882 #endif | |
3883 c = *prog->regmust; | |
3884 s = line + col; | |
170 | 3885 |
3886 /* | |
3887 * This is used very often, esp. for ":global". Use three versions of | |
3888 * the loop to avoid overhead of conditions. | |
3889 */ | |
3890 if (!ireg_ic | |
3891 #ifdef FEAT_MBYTE | |
3892 && !has_mbyte | |
3893 #endif | |
3894 ) | |
3895 while ((s = vim_strbyte(s, c)) != NULL) | |
3896 { | |
3897 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
3898 break; /* Found it. */ | |
3899 ++s; | |
3900 } | |
3901 #ifdef FEAT_MBYTE | |
3902 else if (!ireg_ic || (!enc_utf8 && mb_char2len(c) > 1)) | |
3903 while ((s = vim_strchr(s, c)) != NULL) | |
3904 { | |
3905 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
3906 break; /* Found it. */ | |
3907 mb_ptr_adv(s); | |
3908 } | |
3909 #endif | |
3910 else | |
3911 while ((s = cstrchr(s, c)) != NULL) | |
3912 { | |
3913 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
3914 break; /* Found it. */ | |
3915 mb_ptr_adv(s); | |
3916 } | |
7 | 3917 if (s == NULL) /* Not present. */ |
3918 goto theend; | |
3919 } | |
3920 | |
3921 regline = line; | |
3922 reglnum = 0; | |
2578 | 3923 reg_toolong = FALSE; |
7 | 3924 |
3925 /* Simplest case: Anchored match need be tried only once. */ | |
3926 if (prog->reganch) | |
3927 { | |
3928 int c; | |
3929 | |
3930 #ifdef FEAT_MBYTE | |
3931 if (has_mbyte) | |
3932 c = (*mb_ptr2char)(regline + col); | |
3933 else | |
3934 #endif | |
3935 c = regline[col]; | |
3936 if (prog->regstart == NUL | |
3937 || prog->regstart == c | |
3938 || (ireg_ic && (( | |
3939 #ifdef FEAT_MBYTE | |
3940 (enc_utf8 && utf_fold(prog->regstart) == utf_fold(c))) | |
3941 || (c < 255 && prog->regstart < 255 && | |
3942 #endif | |
1347 | 3943 MB_TOLOWER(prog->regstart) == MB_TOLOWER(c))))) |
7 | 3944 retval = regtry(prog, col); |
3945 else | |
3946 retval = 0; | |
3947 } | |
3948 else | |
3949 { | |
1521 | 3950 #ifdef FEAT_RELTIME |
3951 int tm_count = 0; | |
3952 #endif | |
7 | 3953 /* Messy cases: unanchored match. */ |
180 | 3954 while (!got_int) |
7 | 3955 { |
3956 if (prog->regstart != NUL) | |
3957 { | |
170 | 3958 /* Skip until the char we know it must start with. |
3959 * Used often, do some work to avoid call overhead. */ | |
3960 if (!ireg_ic | |
3961 #ifdef FEAT_MBYTE | |
3962 && !has_mbyte | |
3963 #endif | |
3964 ) | |
3965 s = vim_strbyte(regline + col, prog->regstart); | |
3966 else | |
3967 s = cstrchr(regline + col, prog->regstart); | |
7 | 3968 if (s == NULL) |
3969 { | |
3970 retval = 0; | |
3971 break; | |
3972 } | |
3973 col = (int)(s - regline); | |
3974 } | |
3975 | |
410 | 3976 /* Check for maximum column to try. */ |
3977 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
3978 { | |
3979 retval = 0; | |
3980 break; | |
3981 } | |
3982 | |
7 | 3983 retval = regtry(prog, col); |
3984 if (retval > 0) | |
3985 break; | |
3986 | |
3987 /* if not currently on the first line, get it again */ | |
3988 if (reglnum != 0) | |
3989 { | |
481 | 3990 reglnum = 0; |
7 | 3991 regline = reg_getline((linenr_T)0); |
3992 } | |
3993 if (regline[col] == NUL) | |
3994 break; | |
3995 #ifdef FEAT_MBYTE | |
3996 if (has_mbyte) | |
474 | 3997 col += (*mb_ptr2len)(regline + col); |
7 | 3998 else |
3999 #endif | |
4000 ++col; | |
1521 | 4001 #ifdef FEAT_RELTIME |
4002 /* Check for timeout once in a twenty times to avoid overhead. */ | |
4003 if (tm != NULL && ++tm_count == 20) | |
4004 { | |
4005 tm_count = 0; | |
4006 if (profile_passed_limit(tm)) | |
4007 break; | |
4008 } | |
4009 #endif | |
7 | 4010 } |
4011 } | |
4012 | |
4013 theend: | |
1520 | 4014 /* Free "reg_tofree" when it's a bit big. |
4015 * Free regstack and backpos if they are bigger than their initial size. */ | |
4016 if (reg_tofreelen > 400) | |
4017 { | |
4018 vim_free(reg_tofree); | |
4019 reg_tofree = NULL; | |
4020 } | |
4021 if (regstack.ga_maxlen > REGSTACK_INITIAL) | |
4022 ga_clear(®stack); | |
4023 if (backpos.ga_maxlen > BACKPOS_INITIAL) | |
4024 ga_clear(&backpos); | |
270 | 4025 |
7 | 4026 return retval; |
4027 } | |
4028 | |
4029 #ifdef FEAT_SYN_HL | |
4030 static reg_extmatch_T *make_extmatch __ARGS((void)); | |
4031 | |
4032 /* | |
4033 * Create a new extmatch and mark it as referenced once. | |
4034 */ | |
4035 static reg_extmatch_T * | |
4036 make_extmatch() | |
4037 { | |
4038 reg_extmatch_T *em; | |
4039 | |
4040 em = (reg_extmatch_T *)alloc_clear((unsigned)sizeof(reg_extmatch_T)); | |
4041 if (em != NULL) | |
4042 em->refcnt = 1; | |
4043 return em; | |
4044 } | |
4045 | |
4046 /* | |
4047 * Add a reference to an extmatch. | |
4048 */ | |
4049 reg_extmatch_T * | |
4050 ref_extmatch(em) | |
4051 reg_extmatch_T *em; | |
4052 { | |
4053 if (em != NULL) | |
4054 em->refcnt++; | |
4055 return em; | |
4056 } | |
4057 | |
4058 /* | |
4059 * Remove a reference to an extmatch. If there are no references left, free | |
4060 * the info. | |
4061 */ | |
4062 void | |
4063 unref_extmatch(em) | |
4064 reg_extmatch_T *em; | |
4065 { | |
4066 int i; | |
4067 | |
4068 if (em != NULL && --em->refcnt <= 0) | |
4069 { | |
4070 for (i = 0; i < NSUBEXP; ++i) | |
4071 vim_free(em->matches[i]); | |
4072 vim_free(em); | |
4073 } | |
4074 } | |
4075 #endif | |
4076 | |
4077 /* | |
4078 * regtry - try match of "prog" with at regline["col"]. | |
4079 * Returns 0 for failure, number of lines contained in the match otherwise. | |
4080 */ | |
4081 static long | |
4082 regtry(prog, col) | |
4444 | 4083 bt_regprog_T *prog; |
7 | 4084 colnr_T col; |
4085 { | |
4086 reginput = regline + col; | |
4087 need_clear_subexpr = TRUE; | |
4088 #ifdef FEAT_SYN_HL | |
4089 /* Clear the external match subpointers if necessary. */ | |
4090 if (prog->reghasz == REX_SET) | |
4091 need_clear_zsubexpr = TRUE; | |
4092 #endif | |
4093 | |
189 | 4094 if (regmatch(prog->program + 1) == 0) |
4095 return 0; | |
4096 | |
4097 cleanup_subexpr(); | |
4098 if (REG_MULTI) | |
7 | 4099 { |
189 | 4100 if (reg_startpos[0].lnum < 0) |
7 | 4101 { |
189 | 4102 reg_startpos[0].lnum = 0; |
4103 reg_startpos[0].col = col; | |
4104 } | |
4105 if (reg_endpos[0].lnum < 0) | |
4106 { | |
4107 reg_endpos[0].lnum = reglnum; | |
4108 reg_endpos[0].col = (int)(reginput - regline); | |
7 | 4109 } |
4110 else | |
189 | 4111 /* Use line number of "\ze". */ |
4112 reglnum = reg_endpos[0].lnum; | |
4113 } | |
4114 else | |
4115 { | |
4116 if (reg_startp[0] == NULL) | |
4117 reg_startp[0] = regline + col; | |
4118 if (reg_endp[0] == NULL) | |
4119 reg_endp[0] = reginput; | |
4120 } | |
7 | 4121 #ifdef FEAT_SYN_HL |
189 | 4122 /* Package any found \z(...\) matches for export. Default is none. */ |
4123 unref_extmatch(re_extmatch_out); | |
4124 re_extmatch_out = NULL; | |
4125 | |
4126 if (prog->reghasz == REX_SET) | |
4127 { | |
4128 int i; | |
4129 | |
4130 cleanup_zsubexpr(); | |
4131 re_extmatch_out = make_extmatch(); | |
4132 for (i = 0; i < NSUBEXP; i++) | |
7 | 4133 { |
189 | 4134 if (REG_MULTI) |
7 | 4135 { |
189 | 4136 /* Only accept single line matches. */ |
4137 if (reg_startzpos[i].lnum >= 0 | |
5820 | 4138 && reg_endzpos[i].lnum == reg_startzpos[i].lnum |
4139 && reg_endzpos[i].col >= reg_startzpos[i].col) | |
189 | 4140 re_extmatch_out->matches[i] = |
4141 vim_strnsave(reg_getline(reg_startzpos[i].lnum) | |
7 | 4142 + reg_startzpos[i].col, |
189 | 4143 reg_endzpos[i].col - reg_startzpos[i].col); |
4144 } | |
4145 else | |
4146 { | |
4147 if (reg_startzp[i] != NULL && reg_endzp[i] != NULL) | |
4148 re_extmatch_out->matches[i] = | |
7 | 4149 vim_strnsave(reg_startzp[i], |
189 | 4150 (int)(reg_endzp[i] - reg_startzp[i])); |
7 | 4151 } |
4152 } | |
189 | 4153 } |
7 | 4154 #endif |
189 | 4155 return 1 + reglnum; |
7 | 4156 } |
4157 | |
4158 #ifdef FEAT_MBYTE | |
4159 static int reg_prev_class __ARGS((void)); | |
4160 | |
4161 /* | |
4162 * Get class of previous character. | |
4163 */ | |
4164 static int | |
4165 reg_prev_class() | |
4166 { | |
4167 if (reginput > regline) | |
4069 | 4168 return mb_get_class_buf(reginput - 1 |
4169 - (*mb_head_off)(regline, reginput - 1), reg_buf); | |
7 | 4170 return -1; |
4171 } | |
5735 | 4172 #endif |
4173 | |
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4174 static int reg_match_visual __ARGS((void)); |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4175 |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4176 /* |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4177 * Return TRUE if the current reginput position matches the Visual area. |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4178 */ |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4179 static int |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4180 reg_match_visual() |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4181 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4182 pos_T top, bot; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4183 linenr_T lnum; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4184 colnr_T col; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4185 win_T *wp = reg_win == NULL ? curwin : reg_win; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4186 int mode; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4187 colnr_T start, end; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4188 colnr_T start2, end2; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4189 colnr_T cols; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4190 |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4191 /* Check if the buffer is the current buffer. */ |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4192 if (reg_buf != curbuf || VIsual.lnum == 0) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4193 return FALSE; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4194 |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4195 if (VIsual_active) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4196 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4197 if (lt(VIsual, wp->w_cursor)) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4198 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4199 top = VIsual; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4200 bot = wp->w_cursor; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4201 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4202 else |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4203 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4204 top = wp->w_cursor; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4205 bot = VIsual; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4206 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4207 mode = VIsual_mode; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4208 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4209 else |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4210 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4211 if (lt(curbuf->b_visual.vi_start, curbuf->b_visual.vi_end)) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4212 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4213 top = curbuf->b_visual.vi_start; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4214 bot = curbuf->b_visual.vi_end; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4215 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4216 else |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4217 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4218 top = curbuf->b_visual.vi_end; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4219 bot = curbuf->b_visual.vi_start; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4220 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4221 mode = curbuf->b_visual.vi_mode; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4222 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4223 lnum = reglnum + reg_firstlnum; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4224 if (lnum < top.lnum || lnum > bot.lnum) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4225 return FALSE; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4226 |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4227 if (mode == 'v') |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4228 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4229 col = (colnr_T)(reginput - regline); |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4230 if ((lnum == top.lnum && col < top.col) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4231 || (lnum == bot.lnum && col >= bot.col + (*p_sel != 'e'))) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4232 return FALSE; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4233 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4234 else if (mode == Ctrl_V) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4235 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4236 getvvcol(wp, &top, &start, NULL, &end); |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4237 getvvcol(wp, &bot, &start2, NULL, &end2); |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4238 if (start2 < start) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4239 start = start2; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4240 if (end2 > end) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4241 end = end2; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4242 if (top.col == MAXCOL || bot.col == MAXCOL) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4243 end = MAXCOL; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4244 cols = win_linetabsize(wp, regline, (colnr_T)(reginput - regline)); |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4245 if (cols < start || cols > end - (*p_sel == 'e')) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4246 return FALSE; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4247 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4248 return TRUE; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4249 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4250 |
39 | 4251 #define ADVANCE_REGINPUT() mb_ptr_adv(reginput) |
7 | 4252 |
4253 /* | |
4254 * The arguments from BRACE_LIMITS are stored here. They are actually local | |
4255 * to regmatch(), but they are here to reduce the amount of stack space used | |
4256 * (it can be called recursively many times). | |
4257 */ | |
4258 static long bl_minval; | |
4259 static long bl_maxval; | |
4260 | |
4261 /* | |
4262 * regmatch - main matching routine | |
4263 * | |
180 | 4264 * Conceptually the strategy is simple: Check to see whether the current node |
4265 * matches, push an item onto the regstack and loop to see whether the rest | |
4266 * matches, and then act accordingly. In practice we make some effort to | |
4267 * avoid using the regstack, in particular by going through "ordinary" nodes | |
4268 * (that don't need to know whether the rest of the match failed) by a nested | |
4269 * loop. | |
7 | 4270 * |
4271 * Returns TRUE when there is a match. Leaves reginput and reglnum just after | |
4272 * the last matched character. | |
4273 * Returns FALSE when there is no match. Leaves reginput and reglnum in an | |
4274 * undefined state! | |
4275 */ | |
4276 static int | |
180 | 4277 regmatch(scan) |
7 | 4278 char_u *scan; /* Current node. */ |
4279 { | |
180 | 4280 char_u *next; /* Next node. */ |
4281 int op; | |
4282 int c; | |
4283 regitem_T *rp; | |
4284 int no; | |
4285 int status; /* one of the RA_ values: */ | |
4286 #define RA_FAIL 1 /* something failed, abort */ | |
4287 #define RA_CONT 2 /* continue in inner loop */ | |
4288 #define RA_BREAK 3 /* break inner loop */ | |
4289 #define RA_MATCH 4 /* successful match */ | |
4290 #define RA_NOMATCH 5 /* didn't match */ | |
270 | 4291 |
1520 | 4292 /* Make "regstack" and "backpos" empty. They are allocated and freed in |
4444 | 4293 * bt_regexec_both() to reduce malloc()/free() calls. */ |
270 | 4294 regstack.ga_len = 0; |
4295 backpos.ga_len = 0; | |
233 | 4296 |
180 | 4297 /* |
233 | 4298 * Repeat until "regstack" is empty. |
180 | 4299 */ |
4300 for (;;) | |
4301 { | |
5310 | 4302 /* Some patterns may take a long time to match, e.g., "\([a-z]\+\)\+Q". |
4303 * Allow interrupting them with CTRL-C. */ | |
7 | 4304 fast_breakcheck(); |
4305 | |
4306 #ifdef DEBUG | |
4307 if (scan != NULL && regnarrate) | |
4308 { | |
4444 | 4309 mch_errmsg((char *)regprop(scan)); |
7 | 4310 mch_errmsg("(\n"); |
4311 } | |
4312 #endif | |
180 | 4313 |
4314 /* | |
233 | 4315 * Repeat for items that can be matched sequentially, without using the |
180 | 4316 * regstack. |
4317 */ | |
4318 for (;;) | |
7 | 4319 { |
180 | 4320 if (got_int || scan == NULL) |
4321 { | |
4322 status = RA_FAIL; | |
4323 break; | |
4324 } | |
4325 status = RA_CONT; | |
4326 | |
7 | 4327 #ifdef DEBUG |
4328 if (regnarrate) | |
4329 { | |
4444 | 4330 mch_errmsg((char *)regprop(scan)); |
7 | 4331 mch_errmsg("...\n"); |
4332 # ifdef FEAT_SYN_HL | |
4333 if (re_extmatch_in != NULL) | |
4334 { | |
4335 int i; | |
4336 | |
4337 mch_errmsg(_("External submatches:\n")); | |
4338 for (i = 0; i < NSUBEXP; i++) | |
4339 { | |
4340 mch_errmsg(" \""); | |
4341 if (re_extmatch_in->matches[i] != NULL) | |
4444 | 4342 mch_errmsg((char *)re_extmatch_in->matches[i]); |
7 | 4343 mch_errmsg("\"\n"); |
4344 } | |
4345 } | |
4346 # endif | |
4347 } | |
4348 #endif | |
4349 next = regnext(scan); | |
4350 | |
4351 op = OP(scan); | |
4352 /* Check for character class with NL added. */ | |
1018 | 4353 if (!reg_line_lbr && WITH_NL(op) && REG_MULTI |
4354 && *reginput == NUL && reglnum <= reg_maxline) | |
7 | 4355 { |
4356 reg_nextline(); | |
4357 } | |
4358 else if (reg_line_lbr && WITH_NL(op) && *reginput == '\n') | |
4359 { | |
4360 ADVANCE_REGINPUT(); | |
4361 } | |
4362 else | |
4363 { | |
4364 if (WITH_NL(op)) | |
180 | 4365 op -= ADD_NL; |
7 | 4366 #ifdef FEAT_MBYTE |
4367 if (has_mbyte) | |
4368 c = (*mb_ptr2char)(reginput); | |
4369 else | |
4370 #endif | |
4371 c = *reginput; | |
4372 switch (op) | |
4373 { | |
4374 case BOL: | |
4375 if (reginput != regline) | |
180 | 4376 status = RA_NOMATCH; |
7 | 4377 break; |
4378 | |
4379 case EOL: | |
4380 if (c != NUL) | |
180 | 4381 status = RA_NOMATCH; |
7 | 4382 break; |
4383 | |
4384 case RE_BOF: | |
1458 | 4385 /* We're not at the beginning of the file when below the first |
4386 * line where we started, not at the start of the line or we | |
4387 * didn't start at the first line of the buffer. */ | |
7 | 4388 if (reglnum != 0 || reginput != regline |
1458 | 4389 || (REG_MULTI && reg_firstlnum > 1)) |
180 | 4390 status = RA_NOMATCH; |
7 | 4391 break; |
4392 | |
4393 case RE_EOF: | |
4394 if (reglnum != reg_maxline || c != NUL) | |
180 | 4395 status = RA_NOMATCH; |
7 | 4396 break; |
4397 | |
4398 case CURSOR: | |
4399 /* Check if the buffer is in a window and compare the | |
4400 * reg_win->w_cursor position to the match position. */ | |
4401 if (reg_win == NULL | |
4402 || (reglnum + reg_firstlnum != reg_win->w_cursor.lnum) | |
4403 || ((colnr_T)(reginput - regline) != reg_win->w_cursor.col)) | |
180 | 4404 status = RA_NOMATCH; |
7 | 4405 break; |
4406 | |
639 | 4407 case RE_MARK: |
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4408 /* Compare the mark position to the match position. */ |
639 | 4409 { |
4410 int mark = OPERAND(scan)[0]; | |
4411 int cmp = OPERAND(scan)[1]; | |
4412 pos_T *pos; | |
4413 | |
4043 | 4414 pos = getmark_buf(reg_buf, mark, FALSE); |
1148 | 4415 if (pos == NULL /* mark doesn't exist */ |
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4416 || pos->lnum <= 0 /* mark isn't set in reg_buf */ |
639 | 4417 || (pos->lnum == reglnum + reg_firstlnum |
4418 ? (pos->col == (colnr_T)(reginput - regline) | |
4419 ? (cmp == '<' || cmp == '>') | |
4420 : (pos->col < (colnr_T)(reginput - regline) | |
4421 ? cmp != '>' | |
4422 : cmp != '<')) | |
4423 : (pos->lnum < reglnum + reg_firstlnum | |
4424 ? cmp != '>' | |
4425 : cmp != '<'))) | |
4426 status = RA_NOMATCH; | |
4427 } | |
4428 break; | |
4429 | |
4430 case RE_VISUAL: | |
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4431 if (!reg_match_visual()) |
639 | 4432 status = RA_NOMATCH; |
4433 break; | |
4434 | |
7 | 4435 case RE_LNUM: |
4436 if (!REG_MULTI || !re_num_cmp((long_u)(reglnum + reg_firstlnum), | |
4437 scan)) | |
180 | 4438 status = RA_NOMATCH; |
7 | 4439 break; |
4440 | |
4441 case RE_COL: | |
4442 if (!re_num_cmp((long_u)(reginput - regline) + 1, scan)) | |
180 | 4443 status = RA_NOMATCH; |
7 | 4444 break; |
4445 | |
4446 case RE_VCOL: | |
4447 if (!re_num_cmp((long_u)win_linetabsize( | |
4448 reg_win == NULL ? curwin : reg_win, | |
4449 regline, (colnr_T)(reginput - regline)) + 1, scan)) | |
180 | 4450 status = RA_NOMATCH; |
7 | 4451 break; |
4452 | |
4453 case BOW: /* \<word; reginput points to w */ | |
4454 if (c == NUL) /* Can't match at end of line */ | |
180 | 4455 status = RA_NOMATCH; |
7 | 4456 #ifdef FEAT_MBYTE |
180 | 4457 else if (has_mbyte) |
7 | 4458 { |
4459 int this_class; | |
4460 | |
4461 /* Get class of current and previous char (if it exists). */ | |
4069 | 4462 this_class = mb_get_class_buf(reginput, reg_buf); |
7 | 4463 if (this_class <= 1) |
180 | 4464 status = RA_NOMATCH; /* not on a word at all */ |
4465 else if (reg_prev_class() == this_class) | |
4466 status = RA_NOMATCH; /* previous char is in same word */ | |
7 | 4467 } |
4468 #endif | |
4469 else | |
4470 { | |
4061 | 4471 if (!vim_iswordc_buf(c, reg_buf) || (reginput > regline |
4472 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
180 | 4473 status = RA_NOMATCH; |
7 | 4474 } |
4475 break; | |
4476 | |
4477 case EOW: /* word\>; reginput points after d */ | |
4478 if (reginput == regline) /* Can't match at start of line */ | |
180 | 4479 status = RA_NOMATCH; |
7 | 4480 #ifdef FEAT_MBYTE |
180 | 4481 else if (has_mbyte) |
7 | 4482 { |
4483 int this_class, prev_class; | |
4484 | |
4485 /* Get class of current and previous char (if it exists). */ | |
4069 | 4486 this_class = mb_get_class_buf(reginput, reg_buf); |
7 | 4487 prev_class = reg_prev_class(); |
180 | 4488 if (this_class == prev_class |
4489 || prev_class == 0 || prev_class == 1) | |
4490 status = RA_NOMATCH; | |
7 | 4491 } |
180 | 4492 #endif |
7 | 4493 else |
4494 { | |
4043 | 4495 if (!vim_iswordc_buf(reginput[-1], reg_buf) |
4496 || (reginput[0] != NUL && vim_iswordc_buf(c, reg_buf))) | |
180 | 4497 status = RA_NOMATCH; |
7 | 4498 } |
4499 break; /* Matched with EOW */ | |
4500 | |
4501 case ANY: | |
4084 | 4502 /* ANY does not match new lines. */ |
7 | 4503 if (c == NUL) |
180 | 4504 status = RA_NOMATCH; |
4505 else | |
4506 ADVANCE_REGINPUT(); | |
7 | 4507 break; |
4508 | |
4509 case IDENT: | |
4510 if (!vim_isIDc(c)) | |
180 | 4511 status = RA_NOMATCH; |
4512 else | |
4513 ADVANCE_REGINPUT(); | |
7 | 4514 break; |
4515 | |
4516 case SIDENT: | |
4517 if (VIM_ISDIGIT(*reginput) || !vim_isIDc(c)) | |
180 | 4518 status = RA_NOMATCH; |
4519 else | |
4520 ADVANCE_REGINPUT(); | |
7 | 4521 break; |
4522 | |
4523 case KWORD: | |
4069 | 4524 if (!vim_iswordp_buf(reginput, reg_buf)) |
180 | 4525 status = RA_NOMATCH; |
4526 else | |
4527 ADVANCE_REGINPUT(); | |
7 | 4528 break; |
4529 | |
4530 case SKWORD: | |
4069 | 4531 if (VIM_ISDIGIT(*reginput) || !vim_iswordp_buf(reginput, reg_buf)) |
180 | 4532 status = RA_NOMATCH; |
4533 else | |
4534 ADVANCE_REGINPUT(); | |
7 | 4535 break; |
4536 | |
4537 case FNAME: | |
4538 if (!vim_isfilec(c)) | |
180 | 4539 status = RA_NOMATCH; |
4540 else | |
4541 ADVANCE_REGINPUT(); | |
7 | 4542 break; |
4543 | |
4544 case SFNAME: | |
4545 if (VIM_ISDIGIT(*reginput) || !vim_isfilec(c)) | |
180 | 4546 status = RA_NOMATCH; |
4547 else | |
4548 ADVANCE_REGINPUT(); | |
7 | 4549 break; |
4550 | |
4551 case PRINT: | |
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
4552 if (!vim_isprintc(PTR2CHAR(reginput))) |
180 | 4553 status = RA_NOMATCH; |
4554 else | |
4555 ADVANCE_REGINPUT(); | |
7 | 4556 break; |
4557 | |
4558 case SPRINT: | |
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
4559 if (VIM_ISDIGIT(*reginput) || !vim_isprintc(PTR2CHAR(reginput))) |
180 | 4560 status = RA_NOMATCH; |
4561 else | |
4562 ADVANCE_REGINPUT(); | |
7 | 4563 break; |
4564 | |
4565 case WHITE: | |
4566 if (!vim_iswhite(c)) | |
180 | 4567 status = RA_NOMATCH; |
4568 else | |
4569 ADVANCE_REGINPUT(); | |
7 | 4570 break; |
4571 | |
4572 case NWHITE: | |
4573 if (c == NUL || vim_iswhite(c)) | |
180 | 4574 status = RA_NOMATCH; |
4575 else | |
4576 ADVANCE_REGINPUT(); | |
7 | 4577 break; |
4578 | |
4579 case DIGIT: | |
4580 if (!ri_digit(c)) | |
180 | 4581 status = RA_NOMATCH; |
4582 else | |
4583 ADVANCE_REGINPUT(); | |
7 | 4584 break; |
4585 | |
4586 case NDIGIT: | |
4587 if (c == NUL || ri_digit(c)) | |
180 | 4588 status = RA_NOMATCH; |
4589 else | |
4590 ADVANCE_REGINPUT(); | |
7 | 4591 break; |
4592 | |
4593 case HEX: | |
4594 if (!ri_hex(c)) | |
180 | 4595 status = RA_NOMATCH; |
4596 else | |
4597 ADVANCE_REGINPUT(); | |
7 | 4598 break; |
4599 | |
4600 case NHEX: | |
4601 if (c == NUL || ri_hex(c)) | |
180 | 4602 status = RA_NOMATCH; |
4603 else | |
4604 ADVANCE_REGINPUT(); | |
7 | 4605 break; |
4606 | |
4607 case OCTAL: | |
4608 if (!ri_octal(c)) | |
180 | 4609 status = RA_NOMATCH; |
4610 else | |
4611 ADVANCE_REGINPUT(); | |
7 | 4612 break; |
4613 | |
4614 case NOCTAL: | |
4615 if (c == NUL || ri_octal(c)) | |
180 | 4616 status = RA_NOMATCH; |
4617 else | |
4618 ADVANCE_REGINPUT(); | |
7 | 4619 break; |
4620 | |
4621 case WORD: | |
4622 if (!ri_word(c)) | |
180 | 4623 status = RA_NOMATCH; |
4624 else | |
4625 ADVANCE_REGINPUT(); | |
7 | 4626 break; |
4627 | |
4628 case NWORD: | |
4629 if (c == NUL || ri_word(c)) | |
180 | 4630 status = RA_NOMATCH; |
4631 else | |
4632 ADVANCE_REGINPUT(); | |
7 | 4633 break; |
4634 | |
4635 case HEAD: | |
4636 if (!ri_head(c)) | |
180 | 4637 status = RA_NOMATCH; |
4638 else | |
4639 ADVANCE_REGINPUT(); | |
7 | 4640 break; |
4641 | |
4642 case NHEAD: | |
4643 if (c == NUL || ri_head(c)) | |
180 | 4644 status = RA_NOMATCH; |
4645 else | |
4646 ADVANCE_REGINPUT(); | |
7 | 4647 break; |
4648 | |
4649 case ALPHA: | |
4650 if (!ri_alpha(c)) | |
180 | 4651 status = RA_NOMATCH; |
4652 else | |
4653 ADVANCE_REGINPUT(); | |
7 | 4654 break; |
4655 | |
4656 case NALPHA: | |
4657 if (c == NUL || ri_alpha(c)) | |
180 | 4658 status = RA_NOMATCH; |
4659 else | |
4660 ADVANCE_REGINPUT(); | |
7 | 4661 break; |
4662 | |
4663 case LOWER: | |
4664 if (!ri_lower(c)) | |
180 | 4665 status = RA_NOMATCH; |
4666 else | |
4667 ADVANCE_REGINPUT(); | |
7 | 4668 break; |
4669 | |
4670 case NLOWER: | |
4671 if (c == NUL || ri_lower(c)) | |
180 | 4672 status = RA_NOMATCH; |
4673 else | |
4674 ADVANCE_REGINPUT(); | |
7 | 4675 break; |
4676 | |
4677 case UPPER: | |
4678 if (!ri_upper(c)) | |
180 | 4679 status = RA_NOMATCH; |
4680 else | |
4681 ADVANCE_REGINPUT(); | |
7 | 4682 break; |
4683 | |
4684 case NUPPER: | |
4685 if (c == NUL || ri_upper(c)) | |
180 | 4686 status = RA_NOMATCH; |
4687 else | |
4688 ADVANCE_REGINPUT(); | |
7 | 4689 break; |
4690 | |
4691 case EXACTLY: | |
4692 { | |
4693 int len; | |
4694 char_u *opnd; | |
4695 | |
4696 opnd = OPERAND(scan); | |
4697 /* Inline the first byte, for speed. */ | |
4698 if (*opnd != *reginput | |
4699 && (!ireg_ic || ( | |
4700 #ifdef FEAT_MBYTE | |
4701 !enc_utf8 && | |
4702 #endif | |
1347 | 4703 MB_TOLOWER(*opnd) != MB_TOLOWER(*reginput)))) |
180 | 4704 status = RA_NOMATCH; |
4705 else if (*opnd == NUL) | |
7 | 4706 { |
4707 /* match empty string always works; happens when "~" is | |
4708 * empty. */ | |
4709 } | |
5899 | 4710 else |
4711 { | |
4712 if (opnd[1] == NUL | |
7 | 4713 #ifdef FEAT_MBYTE |
4714 && !(enc_utf8 && ireg_ic) | |
4715 #endif | |
4716 ) | |
5899 | 4717 { |
4718 len = 1; /* matched a single byte above */ | |
4719 } | |
4720 else | |
4721 { | |
4722 /* Need to match first byte again for multi-byte. */ | |
4723 len = (int)STRLEN(opnd); | |
4724 if (cstrncmp(opnd, reginput, &len) != 0) | |
4725 status = RA_NOMATCH; | |
4726 } | |
7 | 4727 #ifdef FEAT_MBYTE |
5901 | 4728 /* Check for following composing character, unless %C |
4729 * follows (skips over all composing chars). */ | |
5899 | 4730 if (status != RA_NOMATCH |
4731 && enc_utf8 | |
4732 && UTF_COMPOSINGLIKE(reginput, reginput + len) | |
5901 | 4733 && !ireg_icombine |
4734 && OP(next) != RE_COMPOSING) | |
7 | 4735 { |
4736 /* raaron: This code makes a composing character get | |
4737 * ignored, which is the correct behavior (sometimes) | |
4738 * for voweled Hebrew texts. */ | |
5899 | 4739 status = RA_NOMATCH; |
7 | 4740 } |
180 | 4741 #endif |
5899 | 4742 if (status != RA_NOMATCH) |
180 | 4743 reginput += len; |
7 | 4744 } |
4745 } | |
4746 break; | |
4747 | |
4748 case ANYOF: | |
4749 case ANYBUT: | |
4750 if (c == NUL) | |
180 | 4751 status = RA_NOMATCH; |
4752 else if ((cstrchr(OPERAND(scan), c) == NULL) == (op == ANYOF)) | |
4753 status = RA_NOMATCH; | |
4754 else | |
4755 ADVANCE_REGINPUT(); | |
7 | 4756 break; |
4757 | |
4758 #ifdef FEAT_MBYTE | |
4759 case MULTIBYTECODE: | |
4760 if (has_mbyte) | |
4761 { | |
4762 int i, len; | |
4763 char_u *opnd; | |
944 | 4764 int opndc = 0, inpc; |
7 | 4765 |
4766 opnd = OPERAND(scan); | |
4767 /* Safety check (just in case 'encoding' was changed since | |
4768 * compiling the program). */ | |
474 | 4769 if ((len = (*mb_ptr2len)(opnd)) < 2) |
180 | 4770 { |
4771 status = RA_NOMATCH; | |
4772 break; | |
4773 } | |
714 | 4774 if (enc_utf8) |
4775 opndc = mb_ptr2char(opnd); | |
4776 if (enc_utf8 && utf_iscomposing(opndc)) | |
4777 { | |
4778 /* When only a composing char is given match at any | |
4779 * position where that composing char appears. */ | |
4780 status = RA_NOMATCH; | |
4781 for (i = 0; reginput[i] != NUL; i += utf_char2len(inpc)) | |
180 | 4782 { |
714 | 4783 inpc = mb_ptr2char(reginput + i); |
4784 if (!utf_iscomposing(inpc)) | |
4785 { | |
4786 if (i > 0) | |
4787 break; | |
4788 } | |
4789 else if (opndc == inpc) | |
4790 { | |
4791 /* Include all following composing chars. */ | |
4792 len = i + mb_ptr2len(reginput + i); | |
4793 status = RA_MATCH; | |
4794 break; | |
4795 } | |
180 | 4796 } |
714 | 4797 } |
4798 else | |
4799 for (i = 0; i < len; ++i) | |
4800 if (opnd[i] != reginput[i]) | |
4801 { | |
4802 status = RA_NOMATCH; | |
4803 break; | |
4804 } | |
7 | 4805 reginput += len; |
4806 } | |
4807 else | |
180 | 4808 status = RA_NOMATCH; |
7 | 4809 break; |
4810 #endif | |
5901 | 4811 case RE_COMPOSING: |
4812 #ifdef FEAT_MBYTE | |
4813 if (enc_utf8) | |
4814 { | |
4815 /* Skip composing characters. */ | |
4816 while (utf_iscomposing(utf_ptr2char(reginput))) | |
4817 mb_cptr_adv(reginput); | |
4818 } | |
4819 #endif | |
4820 break; | |
7 | 4821 |
4822 case NOTHING: | |
4823 break; | |
4824 | |
4825 case BACK: | |
233 | 4826 { |
4827 int i; | |
4828 backpos_T *bp; | |
4829 | |
4830 /* | |
4831 * When we run into BACK we need to check if we don't keep | |
4832 * looping without matching any input. The second and later | |
4833 * times a BACK is encountered it fails if the input is still | |
4834 * at the same position as the previous time. | |
4835 * The positions are stored in "backpos" and found by the | |
4836 * current value of "scan", the position in the RE program. | |
4837 */ | |
4838 bp = (backpos_T *)backpos.ga_data; | |
4839 for (i = 0; i < backpos.ga_len; ++i) | |
4840 if (bp[i].bp_scan == scan) | |
4841 break; | |
4842 if (i == backpos.ga_len) | |
4843 { | |
4844 /* First time at this BACK, make room to store the pos. */ | |
4845 if (ga_grow(&backpos, 1) == FAIL) | |
4846 status = RA_FAIL; | |
4847 else | |
4848 { | |
4849 /* get "ga_data" again, it may have changed */ | |
4850 bp = (backpos_T *)backpos.ga_data; | |
4851 bp[i].bp_scan = scan; | |
4852 ++backpos.ga_len; | |
4853 } | |
4854 } | |
4855 else if (reg_save_equal(&bp[i].bp_pos)) | |
4856 /* Still at same position as last time, fail. */ | |
4857 status = RA_NOMATCH; | |
4858 | |
4859 if (status != RA_FAIL && status != RA_NOMATCH) | |
4860 reg_save(&bp[i].bp_pos, &backpos); | |
4861 } | |
179 | 4862 break; |
4863 | |
7 | 4864 case MOPEN + 0: /* Match start: \zs */ |
4865 case MOPEN + 1: /* \( */ | |
4866 case MOPEN + 2: | |
4867 case MOPEN + 3: | |
4868 case MOPEN + 4: | |
4869 case MOPEN + 5: | |
4870 case MOPEN + 6: | |
4871 case MOPEN + 7: | |
4872 case MOPEN + 8: | |
4873 case MOPEN + 9: | |
4874 { | |
4875 no = op - MOPEN; | |
4876 cleanup_subexpr(); | |
270 | 4877 rp = regstack_push(RS_MOPEN, scan); |
180 | 4878 if (rp == NULL) |
4879 status = RA_FAIL; | |
4880 else | |
4881 { | |
4882 rp->rs_no = no; | |
4883 save_se(&rp->rs_un.sesave, ®_startpos[no], | |
4884 ®_startp[no]); | |
4885 /* We simply continue and handle the result when done. */ | |
4886 } | |
7 | 4887 } |
180 | 4888 break; |
7 | 4889 |
4890 case NOPEN: /* \%( */ | |
4891 case NCLOSE: /* \) after \%( */ | |
270 | 4892 if (regstack_push(RS_NOPEN, scan) == NULL) |
180 | 4893 status = RA_FAIL; |
4894 /* We simply continue and handle the result when done. */ | |
4895 break; | |
7 | 4896 |
4897 #ifdef FEAT_SYN_HL | |
4898 case ZOPEN + 1: | |
4899 case ZOPEN + 2: | |
4900 case ZOPEN + 3: | |
4901 case ZOPEN + 4: | |
4902 case ZOPEN + 5: | |
4903 case ZOPEN + 6: | |
4904 case ZOPEN + 7: | |
4905 case ZOPEN + 8: | |
4906 case ZOPEN + 9: | |
4907 { | |
4908 no = op - ZOPEN; | |
4909 cleanup_zsubexpr(); | |
270 | 4910 rp = regstack_push(RS_ZOPEN, scan); |
180 | 4911 if (rp == NULL) |
4912 status = RA_FAIL; | |
4913 else | |
4914 { | |
4915 rp->rs_no = no; | |
4916 save_se(&rp->rs_un.sesave, ®_startzpos[no], | |
4917 ®_startzp[no]); | |
4918 /* We simply continue and handle the result when done. */ | |
4919 } | |
7 | 4920 } |
180 | 4921 break; |
7 | 4922 #endif |
4923 | |
4924 case MCLOSE + 0: /* Match end: \ze */ | |
4925 case MCLOSE + 1: /* \) */ | |
4926 case MCLOSE + 2: | |
4927 case MCLOSE + 3: | |
4928 case MCLOSE + 4: | |
4929 case MCLOSE + 5: | |
4930 case MCLOSE + 6: | |
4931 case MCLOSE + 7: | |
4932 case MCLOSE + 8: | |
4933 case MCLOSE + 9: | |
4934 { | |
4935 no = op - MCLOSE; | |
4936 cleanup_subexpr(); | |
270 | 4937 rp = regstack_push(RS_MCLOSE, scan); |
180 | 4938 if (rp == NULL) |
4939 status = RA_FAIL; | |
4940 else | |
4941 { | |
4942 rp->rs_no = no; | |
4943 save_se(&rp->rs_un.sesave, ®_endpos[no], ®_endp[no]); | |
4944 /* We simply continue and handle the result when done. */ | |
4945 } | |
7 | 4946 } |
180 | 4947 break; |
7 | 4948 |
4949 #ifdef FEAT_SYN_HL | |
4950 case ZCLOSE + 1: /* \) after \z( */ | |
4951 case ZCLOSE + 2: | |
4952 case ZCLOSE + 3: | |
4953 case ZCLOSE + 4: | |
4954 case ZCLOSE + 5: | |
4955 case ZCLOSE + 6: | |
4956 case ZCLOSE + 7: | |
4957 case ZCLOSE + 8: | |
4958 case ZCLOSE + 9: | |
4959 { | |
4960 no = op - ZCLOSE; | |
4961 cleanup_zsubexpr(); | |
270 | 4962 rp = regstack_push(RS_ZCLOSE, scan); |
180 | 4963 if (rp == NULL) |
4964 status = RA_FAIL; | |
4965 else | |
4966 { | |
4967 rp->rs_no = no; | |
4968 save_se(&rp->rs_un.sesave, ®_endzpos[no], | |
4969 ®_endzp[no]); | |
4970 /* We simply continue and handle the result when done. */ | |
4971 } | |
7 | 4972 } |
180 | 4973 break; |
7 | 4974 #endif |
4975 | |
4976 case BACKREF + 1: | |
4977 case BACKREF + 2: | |
4978 case BACKREF + 3: | |
4979 case BACKREF + 4: | |
4980 case BACKREF + 5: | |
4981 case BACKREF + 6: | |
4982 case BACKREF + 7: | |
4983 case BACKREF + 8: | |
4984 case BACKREF + 9: | |
4985 { | |
4986 int len; | |
4987 | |
4988 no = op - BACKREF; | |
4989 cleanup_subexpr(); | |
4990 if (!REG_MULTI) /* Single-line regexp */ | |
4991 { | |
1815 | 4992 if (reg_startp[no] == NULL || reg_endp[no] == NULL) |
7 | 4993 { |
4994 /* Backref was not set: Match an empty string. */ | |
4995 len = 0; | |
4996 } | |
4997 else | |
4998 { | |
4999 /* Compare current input with back-ref in the same | |
5000 * line. */ | |
5001 len = (int)(reg_endp[no] - reg_startp[no]); | |
5002 if (cstrncmp(reg_startp[no], reginput, &len) != 0) | |
180 | 5003 status = RA_NOMATCH; |
7 | 5004 } |
5005 } | |
5006 else /* Multi-line regexp */ | |
5007 { | |
1815 | 5008 if (reg_startpos[no].lnum < 0 || reg_endpos[no].lnum < 0) |
7 | 5009 { |
5010 /* Backref was not set: Match an empty string. */ | |
5011 len = 0; | |
5012 } | |
5013 else | |
5014 { | |
5015 if (reg_startpos[no].lnum == reglnum | |
5016 && reg_endpos[no].lnum == reglnum) | |
5017 { | |
5018 /* Compare back-ref within the current line. */ | |
5019 len = reg_endpos[no].col - reg_startpos[no].col; | |
5020 if (cstrncmp(regline + reg_startpos[no].col, | |
5021 reginput, &len) != 0) | |
180 | 5022 status = RA_NOMATCH; |
7 | 5023 } |
5024 else | |
5025 { | |
5026 /* Messy situation: Need to compare between two | |
5027 * lines. */ | |
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5028 int r = match_with_backref( |
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5029 reg_startpos[no].lnum, |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5030 reg_startpos[no].col, |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5031 reg_endpos[no].lnum, |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5032 reg_endpos[no].col, |
4899
4837fd61be52
updated for version 7.3.1195
Bram Moolenaar <bram@vim.org>
parents:
4891
diff
changeset
|
5033 &len); |
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5034 |
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5035 if (r != RA_MATCH) |
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5036 status = r; |
7 | 5037 } |
5038 } | |
5039 } | |
5040 | |
5041 /* Matched the backref, skip over it. */ | |
5042 reginput += len; | |
5043 } | |
5044 break; | |
5045 | |
5046 #ifdef FEAT_SYN_HL | |
5047 case ZREF + 1: | |
5048 case ZREF + 2: | |
5049 case ZREF + 3: | |
5050 case ZREF + 4: | |
5051 case ZREF + 5: | |
5052 case ZREF + 6: | |
5053 case ZREF + 7: | |
5054 case ZREF + 8: | |
5055 case ZREF + 9: | |
5056 { | |
5057 int len; | |
5058 | |
5059 cleanup_zsubexpr(); | |
5060 no = op - ZREF; | |
5061 if (re_extmatch_in != NULL | |
5062 && re_extmatch_in->matches[no] != NULL) | |
5063 { | |
5064 len = (int)STRLEN(re_extmatch_in->matches[no]); | |
5065 if (cstrncmp(re_extmatch_in->matches[no], | |
5066 reginput, &len) != 0) | |
180 | 5067 status = RA_NOMATCH; |
5068 else | |
5069 reginput += len; | |
7 | 5070 } |
5071 else | |
5072 { | |
5073 /* Backref was not set: Match an empty string. */ | |
5074 } | |
5075 } | |
5076 break; | |
5077 #endif | |
5078 | |
5079 case BRANCH: | |
5080 { | |
5081 if (OP(next) != BRANCH) /* No choice. */ | |
5082 next = OPERAND(scan); /* Avoid recursion. */ | |
5083 else | |
5084 { | |
270 | 5085 rp = regstack_push(RS_BRANCH, scan); |
180 | 5086 if (rp == NULL) |
5087 status = RA_FAIL; | |
5088 else | |
5089 status = RA_BREAK; /* rest is below */ | |
7 | 5090 } |
5091 } | |
5092 break; | |
5093 | |
5094 case BRACE_LIMITS: | |
5095 { | |
5096 if (OP(next) == BRACE_SIMPLE) | |
5097 { | |
5098 bl_minval = OPERAND_MIN(scan); | |
5099 bl_maxval = OPERAND_MAX(scan); | |
5100 } | |
5101 else if (OP(next) >= BRACE_COMPLEX | |
5102 && OP(next) < BRACE_COMPLEX + 10) | |
5103 { | |
5104 no = OP(next) - BRACE_COMPLEX; | |
5105 brace_min[no] = OPERAND_MIN(scan); | |
5106 brace_max[no] = OPERAND_MAX(scan); | |
5107 brace_count[no] = 0; | |
5108 } | |
5109 else | |
5110 { | |
5111 EMSG(_(e_internal)); /* Shouldn't happen */ | |
180 | 5112 status = RA_FAIL; |
7 | 5113 } |
5114 } | |
5115 break; | |
5116 | |
5117 case BRACE_COMPLEX + 0: | |
5118 case BRACE_COMPLEX + 1: | |
5119 case BRACE_COMPLEX + 2: | |
5120 case BRACE_COMPLEX + 3: | |
5121 case BRACE_COMPLEX + 4: | |
5122 case BRACE_COMPLEX + 5: | |
5123 case BRACE_COMPLEX + 6: | |
5124 case BRACE_COMPLEX + 7: | |
5125 case BRACE_COMPLEX + 8: | |
5126 case BRACE_COMPLEX + 9: | |
5127 { | |
5128 no = op - BRACE_COMPLEX; | |
5129 ++brace_count[no]; | |
5130 | |
5131 /* If not matched enough times yet, try one more */ | |
5132 if (brace_count[no] <= (brace_min[no] <= brace_max[no] | |
180 | 5133 ? brace_min[no] : brace_max[no])) |
7 | 5134 { |
270 | 5135 rp = regstack_push(RS_BRCPLX_MORE, scan); |
180 | 5136 if (rp == NULL) |
5137 status = RA_FAIL; | |
5138 else | |
5139 { | |
5140 rp->rs_no = no; | |
233 | 5141 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5142 next = OPERAND(scan); |
5143 /* We continue and handle the result when done. */ | |
5144 } | |
5145 break; | |
7 | 5146 } |
5147 | |
5148 /* If matched enough times, may try matching some more */ | |
5149 if (brace_min[no] <= brace_max[no]) | |
5150 { | |
5151 /* Range is the normal way around, use longest match */ | |
5152 if (brace_count[no] <= brace_max[no]) | |
5153 { | |
270 | 5154 rp = regstack_push(RS_BRCPLX_LONG, scan); |
180 | 5155 if (rp == NULL) |
5156 status = RA_FAIL; | |
5157 else | |
5158 { | |
5159 rp->rs_no = no; | |
233 | 5160 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5161 next = OPERAND(scan); |
5162 /* We continue and handle the result when done. */ | |
5163 } | |
7 | 5164 } |
5165 } | |
5166 else | |
5167 { | |
5168 /* Range is backwards, use shortest match first */ | |
5169 if (brace_count[no] <= brace_min[no]) | |
5170 { | |
270 | 5171 rp = regstack_push(RS_BRCPLX_SHORT, scan); |
180 | 5172 if (rp == NULL) |
5173 status = RA_FAIL; | |
5174 else | |
5175 { | |
233 | 5176 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5177 /* We continue and handle the result when done. */ |
5178 } | |
7 | 5179 } |
5180 } | |
5181 } | |
5182 break; | |
5183 | |
5184 case BRACE_SIMPLE: | |
5185 case STAR: | |
5186 case PLUS: | |
5187 { | |
180 | 5188 regstar_T rst; |
7 | 5189 |
5190 /* | |
5191 * Lookahead to avoid useless match attempts when we know | |
5192 * what character comes next. | |
5193 */ | |
5194 if (OP(next) == EXACTLY) | |
5195 { | |
180 | 5196 rst.nextb = *OPERAND(next); |
7 | 5197 if (ireg_ic) |
5198 { | |
1347 | 5199 if (MB_ISUPPER(rst.nextb)) |
5200 rst.nextb_ic = MB_TOLOWER(rst.nextb); | |
7 | 5201 else |
1347 | 5202 rst.nextb_ic = MB_TOUPPER(rst.nextb); |
7 | 5203 } |
5204 else | |
180 | 5205 rst.nextb_ic = rst.nextb; |
7 | 5206 } |
5207 else | |
5208 { | |
180 | 5209 rst.nextb = NUL; |
5210 rst.nextb_ic = NUL; | |
7 | 5211 } |
5212 if (op != BRACE_SIMPLE) | |
5213 { | |
180 | 5214 rst.minval = (op == STAR) ? 0 : 1; |
5215 rst.maxval = MAX_LIMIT; | |
7 | 5216 } |
5217 else | |
5218 { | |
180 | 5219 rst.minval = bl_minval; |
5220 rst.maxval = bl_maxval; | |
7 | 5221 } |
5222 | |
5223 /* | |
5224 * When maxval > minval, try matching as much as possible, up | |
5225 * to maxval. When maxval < minval, try matching at least the | |
5226 * minimal number (since the range is backwards, that's also | |
5227 * maxval!). | |
5228 */ | |
180 | 5229 rst.count = regrepeat(OPERAND(scan), rst.maxval); |
7 | 5230 if (got_int) |
180 | 5231 { |
5232 status = RA_FAIL; | |
5233 break; | |
5234 } | |
5235 if (rst.minval <= rst.maxval | |
5236 ? rst.count >= rst.minval : rst.count >= rst.maxval) | |
7 | 5237 { |
180 | 5238 /* It could match. Prepare for trying to match what |
5239 * follows. The code is below. Parameters are stored in | |
5240 * a regstar_T on the regstack. */ | |
212 | 5241 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
189 | 5242 { |
5243 EMSG(_(e_maxmempat)); | |
5244 status = RA_FAIL; | |
5245 } | |
5246 else if (ga_grow(®stack, sizeof(regstar_T)) == FAIL) | |
180 | 5247 status = RA_FAIL; |
5248 else | |
7 | 5249 { |
180 | 5250 regstack.ga_len += sizeof(regstar_T); |
270 | 5251 rp = regstack_push(rst.minval <= rst.maxval |
233 | 5252 ? RS_STAR_LONG : RS_STAR_SHORT, scan); |
180 | 5253 if (rp == NULL) |
5254 status = RA_FAIL; | |
5255 else | |
7 | 5256 { |
180 | 5257 *(((regstar_T *)rp) - 1) = rst; |
5258 status = RA_BREAK; /* skip the restore bits */ | |
7 | 5259 } |
5260 } | |
5261 } | |
5262 else | |
180 | 5263 status = RA_NOMATCH; |
5264 | |
7 | 5265 } |
180 | 5266 break; |
7 | 5267 |
5268 case NOMATCH: | |
5269 case MATCH: | |
5270 case SUBPAT: | |
270 | 5271 rp = regstack_push(RS_NOMATCH, scan); |
180 | 5272 if (rp == NULL) |
5273 status = RA_FAIL; | |
5274 else | |
7 | 5275 { |
180 | 5276 rp->rs_no = op; |
233 | 5277 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5278 next = OPERAND(scan); |
5279 /* We continue and handle the result when done. */ | |
7 | 5280 } |
5281 break; | |
5282 | |
5283 case BEHIND: | |
5284 case NOBEHIND: | |
180 | 5285 /* Need a bit of room to store extra positions. */ |
212 | 5286 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
189 | 5287 { |
5288 EMSG(_(e_maxmempat)); | |
5289 status = RA_FAIL; | |
5290 } | |
5291 else if (ga_grow(®stack, sizeof(regbehind_T)) == FAIL) | |
180 | 5292 status = RA_FAIL; |
5293 else | |
7 | 5294 { |
180 | 5295 regstack.ga_len += sizeof(regbehind_T); |
270 | 5296 rp = regstack_push(RS_BEHIND1, scan); |
180 | 5297 if (rp == NULL) |
5298 status = RA_FAIL; | |
5299 else | |
7 | 5300 { |
1579 | 5301 /* Need to save the subexpr to be able to restore them |
5302 * when there is a match but we don't use it. */ | |
5303 save_subexpr(((regbehind_T *)rp) - 1); | |
5304 | |
180 | 5305 rp->rs_no = op; |
233 | 5306 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5307 /* First try if what follows matches. If it does then we |
5308 * check the behind match by looping. */ | |
7 | 5309 } |
5310 } | |
180 | 5311 break; |
7 | 5312 |
5313 case BHPOS: | |
5314 if (REG_MULTI) | |
5315 { | |
5316 if (behind_pos.rs_u.pos.col != (colnr_T)(reginput - regline) | |
5317 || behind_pos.rs_u.pos.lnum != reglnum) | |
180 | 5318 status = RA_NOMATCH; |
7 | 5319 } |
5320 else if (behind_pos.rs_u.ptr != reginput) | |
180 | 5321 status = RA_NOMATCH; |
7 | 5322 break; |
5323 | |
5324 case NEWL: | |
1018 | 5325 if ((c != NUL || !REG_MULTI || reglnum > reg_maxline |
5326 || reg_line_lbr) && (c != '\n' || !reg_line_lbr)) | |
180 | 5327 status = RA_NOMATCH; |
5328 else if (reg_line_lbr) | |
7 | 5329 ADVANCE_REGINPUT(); |
5330 else | |
5331 reg_nextline(); | |
5332 break; | |
5333 | |
5334 case END: | |
180 | 5335 status = RA_MATCH; /* Success! */ |
5336 break; | |
7 | 5337 |
5338 default: | |
5339 EMSG(_(e_re_corr)); | |
5340 #ifdef DEBUG | |
5341 printf("Illegal op code %d\n", op); | |
5342 #endif | |
180 | 5343 status = RA_FAIL; |
5344 break; | |
7 | 5345 } |
5346 } | |
5347 | |
180 | 5348 /* If we can't continue sequentially, break the inner loop. */ |
5349 if (status != RA_CONT) | |
5350 break; | |
5351 | |
5352 /* Continue in inner loop, advance to next item. */ | |
7 | 5353 scan = next; |
180 | 5354 |
5355 } /* end of inner loop */ | |
7 | 5356 |
5357 /* | |
180 | 5358 * If there is something on the regstack execute the code for the state. |
233 | 5359 * If the state is popped then loop and use the older state. |
7 | 5360 */ |
180 | 5361 while (regstack.ga_len > 0 && status != RA_FAIL) |
5362 { | |
5363 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; | |
5364 switch (rp->rs_state) | |
5365 { | |
5366 case RS_NOPEN: | |
5367 /* Result is passed on as-is, simply pop the state. */ | |
270 | 5368 regstack_pop(&scan); |
180 | 5369 break; |
5370 | |
5371 case RS_MOPEN: | |
5372 /* Pop the state. Restore pointers when there is no match. */ | |
5373 if (status == RA_NOMATCH) | |
5374 restore_se(&rp->rs_un.sesave, ®_startpos[rp->rs_no], | |
5375 ®_startp[rp->rs_no]); | |
270 | 5376 regstack_pop(&scan); |
180 | 5377 break; |
5378 | |
5379 #ifdef FEAT_SYN_HL | |
5380 case RS_ZOPEN: | |
5381 /* Pop the state. Restore pointers when there is no match. */ | |
5382 if (status == RA_NOMATCH) | |
5383 restore_se(&rp->rs_un.sesave, ®_startzpos[rp->rs_no], | |
5384 ®_startzp[rp->rs_no]); | |
270 | 5385 regstack_pop(&scan); |
180 | 5386 break; |
5387 #endif | |
5388 | |
5389 case RS_MCLOSE: | |
5390 /* Pop the state. Restore pointers when there is no match. */ | |
5391 if (status == RA_NOMATCH) | |
5392 restore_se(&rp->rs_un.sesave, ®_endpos[rp->rs_no], | |
5393 ®_endp[rp->rs_no]); | |
270 | 5394 regstack_pop(&scan); |
180 | 5395 break; |
5396 | |
5397 #ifdef FEAT_SYN_HL | |
5398 case RS_ZCLOSE: | |
5399 /* Pop the state. Restore pointers when there is no match. */ | |
5400 if (status == RA_NOMATCH) | |
5401 restore_se(&rp->rs_un.sesave, ®_endzpos[rp->rs_no], | |
5402 ®_endzp[rp->rs_no]); | |
270 | 5403 regstack_pop(&scan); |
180 | 5404 break; |
5405 #endif | |
5406 | |
5407 case RS_BRANCH: | |
5408 if (status == RA_MATCH) | |
5409 /* this branch matched, use it */ | |
270 | 5410 regstack_pop(&scan); |
180 | 5411 else |
5412 { | |
5413 if (status != RA_BREAK) | |
5414 { | |
5415 /* After a non-matching branch: try next one. */ | |
233 | 5416 reg_restore(&rp->rs_un.regsave, &backpos); |
180 | 5417 scan = rp->rs_scan; |
5418 } | |
5419 if (scan == NULL || OP(scan) != BRANCH) | |
5420 { | |
5421 /* no more branches, didn't find a match */ | |
5422 status = RA_NOMATCH; | |
270 | 5423 regstack_pop(&scan); |
180 | 5424 } |
5425 else | |
5426 { | |
5427 /* Prepare to try a branch. */ | |
5428 rp->rs_scan = regnext(scan); | |
233 | 5429 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5430 scan = OPERAND(scan); |
5431 } | |
5432 } | |
5433 break; | |
5434 | |
5435 case RS_BRCPLX_MORE: | |
5436 /* Pop the state. Restore pointers when there is no match. */ | |
5437 if (status == RA_NOMATCH) | |
5438 { | |
233 | 5439 reg_restore(&rp->rs_un.regsave, &backpos); |
180 | 5440 --brace_count[rp->rs_no]; /* decrement match count */ |
5441 } | |
270 | 5442 regstack_pop(&scan); |
180 | 5443 break; |
5444 | |
5445 case RS_BRCPLX_LONG: | |
5446 /* Pop the state. Restore pointers when there is no match. */ | |
5447 if (status == RA_NOMATCH) | |
5448 { | |
5449 /* There was no match, but we did find enough matches. */ | |
233 | 5450 reg_restore(&rp->rs_un.regsave, &backpos); |
180 | 5451 --brace_count[rp->rs_no]; |
5452 /* continue with the items after "\{}" */ | |
5453 status = RA_CONT; | |
5454 } | |
270 | 5455 regstack_pop(&scan); |
180 | 5456 if (status == RA_CONT) |
5457 scan = regnext(scan); | |
5458 break; | |
5459 | |
5460 case RS_BRCPLX_SHORT: | |
5461 /* Pop the state. Restore pointers when there is no match. */ | |
5462 if (status == RA_NOMATCH) | |
5463 /* There was no match, try to match one more item. */ | |
233 | 5464 reg_restore(&rp->rs_un.regsave, &backpos); |
270 | 5465 regstack_pop(&scan); |
180 | 5466 if (status == RA_NOMATCH) |
5467 { | |
5468 scan = OPERAND(scan); | |
5469 status = RA_CONT; | |
5470 } | |
5471 break; | |
5472 | |
5473 case RS_NOMATCH: | |
5474 /* Pop the state. If the operand matches for NOMATCH or | |
5475 * doesn't match for MATCH/SUBPAT, we fail. Otherwise backup, | |
5476 * except for SUBPAT, and continue with the next item. */ | |
5477 if (status == (rp->rs_no == NOMATCH ? RA_MATCH : RA_NOMATCH)) | |
5478 status = RA_NOMATCH; | |
5479 else | |
5480 { | |
5481 status = RA_CONT; | |
233 | 5482 if (rp->rs_no != SUBPAT) /* zero-width */ |
5483 reg_restore(&rp->rs_un.regsave, &backpos); | |
180 | 5484 } |
270 | 5485 regstack_pop(&scan); |
180 | 5486 if (status == RA_CONT) |
5487 scan = regnext(scan); | |
5488 break; | |
5489 | |
5490 case RS_BEHIND1: | |
5491 if (status == RA_NOMATCH) | |
5492 { | |
270 | 5493 regstack_pop(&scan); |
180 | 5494 regstack.ga_len -= sizeof(regbehind_T); |
5495 } | |
5496 else | |
5497 { | |
5498 /* The stuff after BEHIND/NOBEHIND matches. Now try if | |
5499 * the behind part does (not) match before the current | |
5500 * position in the input. This must be done at every | |
5501 * position in the input and checking if the match ends at | |
5502 * the current position. */ | |
5503 | |
5504 /* save the position after the found match for next */ | |
233 | 5505 reg_save(&(((regbehind_T *)rp) - 1)->save_after, &backpos); |
180 | 5506 |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5507 /* Start looking for a match with operand at the current |
1209 | 5508 * position. Go back one character until we find the |
180 | 5509 * result, hitting the start of the line or the previous |
5510 * line (for multi-line matching). | |
5511 * Set behind_pos to where the match should end, BHPOS | |
5512 * will match it. Save the current value. */ | |
5513 (((regbehind_T *)rp) - 1)->save_behind = behind_pos; | |
5514 behind_pos = rp->rs_un.regsave; | |
5515 | |
5516 rp->rs_state = RS_BEHIND2; | |
5517 | |
233 | 5518 reg_restore(&rp->rs_un.regsave, &backpos); |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5519 scan = OPERAND(rp->rs_scan) + 4; |
180 | 5520 } |
5521 break; | |
5522 | |
5523 case RS_BEHIND2: | |
5524 /* | |
5525 * Looping for BEHIND / NOBEHIND match. | |
5526 */ | |
5527 if (status == RA_MATCH && reg_save_equal(&behind_pos)) | |
5528 { | |
5529 /* found a match that ends where "next" started */ | |
5530 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
5531 if (rp->rs_no == BEHIND) | |
233 | 5532 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
5533 &backpos); | |
180 | 5534 else |
1579 | 5535 { |
5536 /* But we didn't want a match. Need to restore the | |
5537 * subexpr, because what follows matched, so they have | |
5538 * been set. */ | |
180 | 5539 status = RA_NOMATCH; |
1579 | 5540 restore_subexpr(((regbehind_T *)rp) - 1); |
5541 } | |
270 | 5542 regstack_pop(&scan); |
180 | 5543 regstack.ga_len -= sizeof(regbehind_T); |
5544 } | |
5545 else | |
5546 { | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5547 long limit; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5548 |
1579 | 5549 /* No match or a match that doesn't end where we want it: Go |
5550 * back one character. May go to previous line once. */ | |
180 | 5551 no = OK; |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5552 limit = OPERAND_MIN(rp->rs_scan); |
180 | 5553 if (REG_MULTI) |
5554 { | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5555 if (limit > 0 |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5556 && ((rp->rs_un.regsave.rs_u.pos.lnum |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5557 < behind_pos.rs_u.pos.lnum |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5558 ? (colnr_T)STRLEN(regline) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5559 : behind_pos.rs_u.pos.col) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5560 - rp->rs_un.regsave.rs_u.pos.col >= limit)) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5561 no = FAIL; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5562 else if (rp->rs_un.regsave.rs_u.pos.col == 0) |
180 | 5563 { |
5564 if (rp->rs_un.regsave.rs_u.pos.lnum | |
5565 < behind_pos.rs_u.pos.lnum | |
5566 || reg_getline( | |
5567 --rp->rs_un.regsave.rs_u.pos.lnum) | |
5568 == NULL) | |
5569 no = FAIL; | |
5570 else | |
5571 { | |
233 | 5572 reg_restore(&rp->rs_un.regsave, &backpos); |
180 | 5573 rp->rs_un.regsave.rs_u.pos.col = |
5574 (colnr_T)STRLEN(regline); | |
5575 } | |
5576 } | |
5577 else | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5578 { |
4176 | 5579 #ifdef FEAT_MBYTE |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5580 if (has_mbyte) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5581 rp->rs_un.regsave.rs_u.pos.col -= |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5582 (*mb_head_off)(regline, regline |
4176 | 5583 + rp->rs_un.regsave.rs_u.pos.col - 1) + 1; |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5584 else |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5585 #endif |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5586 --rp->rs_un.regsave.rs_u.pos.col; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5587 } |
180 | 5588 } |
5589 else | |
5590 { | |
5591 if (rp->rs_un.regsave.rs_u.ptr == regline) | |
5592 no = FAIL; | |
5593 else | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5594 { |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5595 mb_ptr_back(regline, rp->rs_un.regsave.rs_u.ptr); |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5596 if (limit > 0 && (long)(behind_pos.rs_u.ptr |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5597 - rp->rs_un.regsave.rs_u.ptr) > limit) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5598 no = FAIL; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5599 } |
180 | 5600 } |
5601 if (no == OK) | |
5602 { | |
5603 /* Advanced, prepare for finding match again. */ | |
233 | 5604 reg_restore(&rp->rs_un.regsave, &backpos); |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5605 scan = OPERAND(rp->rs_scan) + 4; |
1579 | 5606 if (status == RA_MATCH) |
5607 { | |
5608 /* We did match, so subexpr may have been changed, | |
5609 * need to restore them for the next try. */ | |
5610 status = RA_NOMATCH; | |
5611 restore_subexpr(((regbehind_T *)rp) - 1); | |
5612 } | |
180 | 5613 } |
5614 else | |
5615 { | |
5616 /* Can't advance. For NOBEHIND that's a match. */ | |
5617 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
5618 if (rp->rs_no == NOBEHIND) | |
5619 { | |
233 | 5620 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
5621 &backpos); | |
180 | 5622 status = RA_MATCH; |
5623 } | |
5624 else | |
1579 | 5625 { |
5626 /* We do want a proper match. Need to restore the | |
5627 * subexpr if we had a match, because they may have | |
5628 * been set. */ | |
5629 if (status == RA_MATCH) | |
5630 { | |
5631 status = RA_NOMATCH; | |
5632 restore_subexpr(((regbehind_T *)rp) - 1); | |
5633 } | |
5634 } | |
270 | 5635 regstack_pop(&scan); |
180 | 5636 regstack.ga_len -= sizeof(regbehind_T); |
5637 } | |
5638 } | |
5639 break; | |
5640 | |
5641 case RS_STAR_LONG: | |
5642 case RS_STAR_SHORT: | |
5643 { | |
5644 regstar_T *rst = ((regstar_T *)rp) - 1; | |
5645 | |
5646 if (status == RA_MATCH) | |
5647 { | |
270 | 5648 regstack_pop(&scan); |
180 | 5649 regstack.ga_len -= sizeof(regstar_T); |
5650 break; | |
5651 } | |
5652 | |
5653 /* Tried once already, restore input pointers. */ | |
5654 if (status != RA_BREAK) | |
233 | 5655 reg_restore(&rp->rs_un.regsave, &backpos); |
180 | 5656 |
5657 /* Repeat until we found a position where it could match. */ | |
5658 for (;;) | |
5659 { | |
5660 if (status != RA_BREAK) | |
5661 { | |
5662 /* Tried first position already, advance. */ | |
5663 if (rp->rs_state == RS_STAR_LONG) | |
5664 { | |
685 | 5665 /* Trying for longest match, but couldn't or |
5666 * didn't match -- back up one char. */ | |
180 | 5667 if (--rst->count < rst->minval) |
5668 break; | |
5669 if (reginput == regline) | |
5670 { | |
5671 /* backup to last char of previous line */ | |
5672 --reglnum; | |
5673 regline = reg_getline(reglnum); | |
5674 /* Just in case regrepeat() didn't count | |
5675 * right. */ | |
5676 if (regline == NULL) | |
5677 break; | |
5678 reginput = regline + STRLEN(regline); | |
5679 fast_breakcheck(); | |
5680 } | |
5681 else | |
5682 mb_ptr_back(regline, reginput); | |
5683 } | |
5684 else | |
5685 { | |
5686 /* Range is backwards, use shortest match first. | |
5687 * Careful: maxval and minval are exchanged! | |
5688 * Couldn't or didn't match: try advancing one | |
5689 * char. */ | |
5690 if (rst->count == rst->minval | |
5691 || regrepeat(OPERAND(rp->rs_scan), 1L) == 0) | |
5692 break; | |
5693 ++rst->count; | |
5694 } | |
5695 if (got_int) | |
5696 break; | |
5697 } | |
5698 else | |
5699 status = RA_NOMATCH; | |
5700 | |
5701 /* If it could match, try it. */ | |
5702 if (rst->nextb == NUL || *reginput == rst->nextb | |
5703 || *reginput == rst->nextb_ic) | |
5704 { | |
233 | 5705 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5706 scan = regnext(rp->rs_scan); |
5707 status = RA_CONT; | |
5708 break; | |
5709 } | |
5710 } | |
5711 if (status != RA_CONT) | |
5712 { | |
5713 /* Failed. */ | |
270 | 5714 regstack_pop(&scan); |
180 | 5715 regstack.ga_len -= sizeof(regstar_T); |
5716 status = RA_NOMATCH; | |
5717 } | |
5718 } | |
5719 break; | |
5720 } | |
5721 | |
685 | 5722 /* If we want to continue the inner loop or didn't pop a state |
5723 * continue matching loop */ | |
180 | 5724 if (status == RA_CONT || rp == (regitem_T *) |
5725 ((char *)regstack.ga_data + regstack.ga_len) - 1) | |
5726 break; | |
5727 } | |
5728 | |
189 | 5729 /* May need to continue with the inner loop, starting at "scan". */ |
180 | 5730 if (status == RA_CONT) |
5731 continue; | |
5732 | |
5733 /* | |
5734 * If the regstack is empty or something failed we are done. | |
5735 */ | |
5736 if (regstack.ga_len == 0 || status == RA_FAIL) | |
5737 { | |
5738 if (scan == NULL) | |
5739 { | |
5740 /* | |
5741 * We get here only if there's trouble -- normally "case END" is | |
5742 * the terminating point. | |
5743 */ | |
5744 EMSG(_(e_re_corr)); | |
7 | 5745 #ifdef DEBUG |
180 | 5746 printf("Premature EOL\n"); |
7 | 5747 #endif |
180 | 5748 } |
189 | 5749 if (status == RA_FAIL) |
5750 got_int = TRUE; | |
180 | 5751 return (status == RA_MATCH); |
5752 } | |
5753 | |
5754 } /* End of loop until the regstack is empty. */ | |
5755 | |
5756 /* NOTREACHED */ | |
5757 } | |
5758 | |
5759 /* | |
5760 * Push an item onto the regstack. | |
5761 * Returns pointer to new item. Returns NULL when out of memory. | |
5762 */ | |
5763 static regitem_T * | |
270 | 5764 regstack_push(state, scan) |
180 | 5765 regstate_T state; |
5766 char_u *scan; | |
5767 { | |
5768 regitem_T *rp; | |
5769 | |
270 | 5770 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
189 | 5771 { |
5772 EMSG(_(e_maxmempat)); | |
5773 return NULL; | |
5774 } | |
270 | 5775 if (ga_grow(®stack, sizeof(regitem_T)) == FAIL) |
180 | 5776 return NULL; |
5777 | |
270 | 5778 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len); |
180 | 5779 rp->rs_state = state; |
5780 rp->rs_scan = scan; | |
5781 | |
270 | 5782 regstack.ga_len += sizeof(regitem_T); |
180 | 5783 return rp; |
5784 } | |
5785 | |
5786 /* | |
5787 * Pop an item from the regstack. | |
5788 */ | |
5789 static void | |
270 | 5790 regstack_pop(scan) |
180 | 5791 char_u **scan; |
5792 { | |
5793 regitem_T *rp; | |
5794 | |
270 | 5795 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; |
180 | 5796 *scan = rp->rs_scan; |
5797 | |
270 | 5798 regstack.ga_len -= sizeof(regitem_T); |
7 | 5799 } |
5800 | |
5801 /* | |
5802 * regrepeat - repeatedly match something simple, return how many. | |
5803 * Advances reginput (and reglnum) to just after the matched chars. | |
5804 */ | |
5805 static int | |
5806 regrepeat(p, maxcount) | |
5807 char_u *p; | |
5808 long maxcount; /* maximum number of matches allowed */ | |
5809 { | |
5810 long count = 0; | |
5811 char_u *scan; | |
5812 char_u *opnd; | |
5813 int mask; | |
5814 int testval = 0; | |
5815 | |
5816 scan = reginput; /* Make local copy of reginput for speed. */ | |
5817 opnd = OPERAND(p); | |
5818 switch (OP(p)) | |
5819 { | |
5820 case ANY: | |
5821 case ANY + ADD_NL: | |
5822 while (count < maxcount) | |
5823 { | |
5824 /* Matching anything means we continue until end-of-line (or | |
5825 * end-of-file for ANY + ADD_NL), only limited by maxcount. */ | |
5826 while (*scan != NUL && count < maxcount) | |
5827 { | |
5828 ++count; | |
39 | 5829 mb_ptr_adv(scan); |
7 | 5830 } |
1018 | 5831 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5832 || reg_line_lbr || count == maxcount) | |
7 | 5833 break; |
5834 ++count; /* count the line-break */ | |
5835 reg_nextline(); | |
5836 scan = reginput; | |
5837 if (got_int) | |
5838 break; | |
5839 } | |
5840 break; | |
5841 | |
5842 case IDENT: | |
5843 case IDENT + ADD_NL: | |
5844 testval = TRUE; | |
5845 /*FALLTHROUGH*/ | |
5846 case SIDENT: | |
5847 case SIDENT + ADD_NL: | |
5848 while (count < maxcount) | |
5849 { | |
4466 | 5850 if (vim_isIDc(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
7 | 5851 { |
39 | 5852 mb_ptr_adv(scan); |
7 | 5853 } |
5854 else if (*scan == NUL) | |
5855 { | |
1018 | 5856 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5857 || reg_line_lbr) | |
7 | 5858 break; |
5859 reg_nextline(); | |
5860 scan = reginput; | |
5861 if (got_int) | |
5862 break; | |
5863 } | |
5864 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
5865 ++scan; | |
5866 else | |
5867 break; | |
5868 ++count; | |
5869 } | |
5870 break; | |
5871 | |
5872 case KWORD: | |
5873 case KWORD + ADD_NL: | |
5874 testval = TRUE; | |
5875 /*FALLTHROUGH*/ | |
5876 case SKWORD: | |
5877 case SKWORD + ADD_NL: | |
5878 while (count < maxcount) | |
5879 { | |
4069 | 5880 if (vim_iswordp_buf(scan, reg_buf) |
5881 && (testval || !VIM_ISDIGIT(*scan))) | |
7 | 5882 { |
39 | 5883 mb_ptr_adv(scan); |
7 | 5884 } |
5885 else if (*scan == NUL) | |
5886 { | |
1018 | 5887 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5888 || reg_line_lbr) | |
7 | 5889 break; |
5890 reg_nextline(); | |
5891 scan = reginput; | |
5892 if (got_int) | |
5893 break; | |
5894 } | |
5895 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
5896 ++scan; | |
5897 else | |
5898 break; | |
5899 ++count; | |
5900 } | |
5901 break; | |
5902 | |
5903 case FNAME: | |
5904 case FNAME + ADD_NL: | |
5905 testval = TRUE; | |
5906 /*FALLTHROUGH*/ | |
5907 case SFNAME: | |
5908 case SFNAME + ADD_NL: | |
5909 while (count < maxcount) | |
5910 { | |
4466 | 5911 if (vim_isfilec(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
7 | 5912 { |
39 | 5913 mb_ptr_adv(scan); |
7 | 5914 } |
5915 else if (*scan == NUL) | |
5916 { | |
1018 | 5917 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5918 || reg_line_lbr) | |
7 | 5919 break; |
5920 reg_nextline(); | |
5921 scan = reginput; | |
5922 if (got_int) | |
5923 break; | |
5924 } | |
5925 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
5926 ++scan; | |
5927 else | |
5928 break; | |
5929 ++count; | |
5930 } | |
5931 break; | |
5932 | |
5933 case PRINT: | |
5934 case PRINT + ADD_NL: | |
5935 testval = TRUE; | |
5936 /*FALLTHROUGH*/ | |
5937 case SPRINT: | |
5938 case SPRINT + ADD_NL: | |
5939 while (count < maxcount) | |
5940 { | |
5941 if (*scan == NUL) | |
5942 { | |
1018 | 5943 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5944 || reg_line_lbr) | |
7 | 5945 break; |
5946 reg_nextline(); | |
5947 scan = reginput; | |
5948 if (got_int) | |
5949 break; | |
5950 } | |
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
5951 else if (vim_isprintc(PTR2CHAR(scan)) == 1 |
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
5952 && (testval || !VIM_ISDIGIT(*scan))) |
7 | 5953 { |
39 | 5954 mb_ptr_adv(scan); |
7 | 5955 } |
5956 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
5957 ++scan; | |
5958 else | |
5959 break; | |
5960 ++count; | |
5961 } | |
5962 break; | |
5963 | |
5964 case WHITE: | |
5965 case WHITE + ADD_NL: | |
5966 testval = mask = RI_WHITE; | |
5967 do_class: | |
5968 while (count < maxcount) | |
5969 { | |
5970 #ifdef FEAT_MBYTE | |
5971 int l; | |
5972 #endif | |
5973 if (*scan == NUL) | |
5974 { | |
1018 | 5975 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5976 || reg_line_lbr) | |
7 | 5977 break; |
5978 reg_nextline(); | |
5979 scan = reginput; | |
5980 if (got_int) | |
5981 break; | |
5982 } | |
5983 #ifdef FEAT_MBYTE | |
474 | 5984 else if (has_mbyte && (l = (*mb_ptr2len)(scan)) > 1) |
7 | 5985 { |
5986 if (testval != 0) | |
5987 break; | |
5988 scan += l; | |
5989 } | |
5990 #endif | |
5991 else if ((class_tab[*scan] & mask) == testval) | |
5992 ++scan; | |
5993 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
5994 ++scan; | |
5995 else | |
5996 break; | |
5997 ++count; | |
5998 } | |
5999 break; | |
6000 | |
6001 case NWHITE: | |
6002 case NWHITE + ADD_NL: | |
6003 mask = RI_WHITE; | |
6004 goto do_class; | |
6005 case DIGIT: | |
6006 case DIGIT + ADD_NL: | |
6007 testval = mask = RI_DIGIT; | |
6008 goto do_class; | |
6009 case NDIGIT: | |
6010 case NDIGIT + ADD_NL: | |
6011 mask = RI_DIGIT; | |
6012 goto do_class; | |
6013 case HEX: | |
6014 case HEX + ADD_NL: | |
6015 testval = mask = RI_HEX; | |
6016 goto do_class; | |
6017 case NHEX: | |
6018 case NHEX + ADD_NL: | |
6019 mask = RI_HEX; | |
6020 goto do_class; | |
6021 case OCTAL: | |
6022 case OCTAL + ADD_NL: | |
6023 testval = mask = RI_OCTAL; | |
6024 goto do_class; | |
6025 case NOCTAL: | |
6026 case NOCTAL + ADD_NL: | |
6027 mask = RI_OCTAL; | |
6028 goto do_class; | |
6029 case WORD: | |
6030 case WORD + ADD_NL: | |
6031 testval = mask = RI_WORD; | |
6032 goto do_class; | |
6033 case NWORD: | |
6034 case NWORD + ADD_NL: | |
6035 mask = RI_WORD; | |
6036 goto do_class; | |
6037 case HEAD: | |
6038 case HEAD + ADD_NL: | |
6039 testval = mask = RI_HEAD; | |
6040 goto do_class; | |
6041 case NHEAD: | |
6042 case NHEAD + ADD_NL: | |
6043 mask = RI_HEAD; | |
6044 goto do_class; | |
6045 case ALPHA: | |
6046 case ALPHA + ADD_NL: | |
6047 testval = mask = RI_ALPHA; | |
6048 goto do_class; | |
6049 case NALPHA: | |
6050 case NALPHA + ADD_NL: | |
6051 mask = RI_ALPHA; | |
6052 goto do_class; | |
6053 case LOWER: | |
6054 case LOWER + ADD_NL: | |
6055 testval = mask = RI_LOWER; | |
6056 goto do_class; | |
6057 case NLOWER: | |
6058 case NLOWER + ADD_NL: | |
6059 mask = RI_LOWER; | |
6060 goto do_class; | |
6061 case UPPER: | |
6062 case UPPER + ADD_NL: | |
6063 testval = mask = RI_UPPER; | |
6064 goto do_class; | |
6065 case NUPPER: | |
6066 case NUPPER + ADD_NL: | |
6067 mask = RI_UPPER; | |
6068 goto do_class; | |
6069 | |
6070 case EXACTLY: | |
6071 { | |
6072 int cu, cl; | |
6073 | |
6074 /* This doesn't do a multi-byte character, because a MULTIBYTECODE | |
1347 | 6075 * would have been used for it. It does handle single-byte |
6076 * characters, such as latin1. */ | |
7 | 6077 if (ireg_ic) |
6078 { | |
1347 | 6079 cu = MB_TOUPPER(*opnd); |
6080 cl = MB_TOLOWER(*opnd); | |
7 | 6081 while (count < maxcount && (*scan == cu || *scan == cl)) |
6082 { | |
6083 count++; | |
6084 scan++; | |
6085 } | |
6086 } | |
6087 else | |
6088 { | |
6089 cu = *opnd; | |
6090 while (count < maxcount && *scan == cu) | |
6091 { | |
6092 count++; | |
6093 scan++; | |
6094 } | |
6095 } | |
6096 break; | |
6097 } | |
6098 | |
6099 #ifdef FEAT_MBYTE | |
6100 case MULTIBYTECODE: | |
6101 { | |
6102 int i, len, cf = 0; | |
6103 | |
6104 /* Safety check (just in case 'encoding' was changed since | |
6105 * compiling the program). */ | |
474 | 6106 if ((len = (*mb_ptr2len)(opnd)) > 1) |
7 | 6107 { |
6108 if (ireg_ic && enc_utf8) | |
6109 cf = utf_fold(utf_ptr2char(opnd)); | |
6110 while (count < maxcount) | |
6111 { | |
6112 for (i = 0; i < len; ++i) | |
6113 if (opnd[i] != scan[i]) | |
6114 break; | |
6115 if (i < len && (!ireg_ic || !enc_utf8 | |
6116 || utf_fold(utf_ptr2char(scan)) != cf)) | |
6117 break; | |
6118 scan += len; | |
6119 ++count; | |
6120 } | |
6121 } | |
6122 } | |
6123 break; | |
6124 #endif | |
6125 | |
6126 case ANYOF: | |
6127 case ANYOF + ADD_NL: | |
6128 testval = TRUE; | |
6129 /*FALLTHROUGH*/ | |
6130 | |
6131 case ANYBUT: | |
6132 case ANYBUT + ADD_NL: | |
6133 while (count < maxcount) | |
6134 { | |
6135 #ifdef FEAT_MBYTE | |
6136 int len; | |
6137 #endif | |
6138 if (*scan == NUL) | |
6139 { | |
1018 | 6140 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
6141 || reg_line_lbr) | |
7 | 6142 break; |
6143 reg_nextline(); | |
6144 scan = reginput; | |
6145 if (got_int) | |
6146 break; | |
6147 } | |
6148 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
6149 ++scan; | |
6150 #ifdef FEAT_MBYTE | |
474 | 6151 else if (has_mbyte && (len = (*mb_ptr2len)(scan)) > 1) |
7 | 6152 { |
6153 if ((cstrchr(opnd, (*mb_ptr2char)(scan)) == NULL) == testval) | |
6154 break; | |
6155 scan += len; | |
6156 } | |
6157 #endif | |
6158 else | |
6159 { | |
6160 if ((cstrchr(opnd, *scan) == NULL) == testval) | |
6161 break; | |
6162 ++scan; | |
6163 } | |
6164 ++count; | |
6165 } | |
6166 break; | |
6167 | |
6168 case NEWL: | |
6169 while (count < maxcount | |
1018 | 6170 && ((*scan == NUL && reglnum <= reg_maxline && !reg_line_lbr |
6171 && REG_MULTI) || (*scan == '\n' && reg_line_lbr))) | |
7 | 6172 { |
6173 count++; | |
6174 if (reg_line_lbr) | |
6175 ADVANCE_REGINPUT(); | |
6176 else | |
6177 reg_nextline(); | |
6178 scan = reginput; | |
6179 if (got_int) | |
6180 break; | |
6181 } | |
6182 break; | |
6183 | |
6184 default: /* Oh dear. Called inappropriately. */ | |
6185 EMSG(_(e_re_corr)); | |
6186 #ifdef DEBUG | |
6187 printf("Called regrepeat with op code %d\n", OP(p)); | |
6188 #endif | |
6189 break; | |
6190 } | |
6191 | |
6192 reginput = scan; | |
6193 | |
6194 return (int)count; | |
6195 } | |
6196 | |
6197 /* | |
6198 * regnext - dig the "next" pointer out of a node | |
2010 | 6199 * Returns NULL when calculating size, when there is no next item and when |
6200 * there is an error. | |
7 | 6201 */ |
6202 static char_u * | |
6203 regnext(p) | |
6204 char_u *p; | |
6205 { | |
6206 int offset; | |
6207 | |
2010 | 6208 if (p == JUST_CALC_SIZE || reg_toolong) |
7 | 6209 return NULL; |
6210 | |
6211 offset = NEXT(p); | |
6212 if (offset == 0) | |
6213 return NULL; | |
6214 | |
233 | 6215 if (OP(p) == BACK) |
7 | 6216 return p - offset; |
6217 else | |
6218 return p + offset; | |
6219 } | |
6220 | |
6221 /* | |
6222 * Check the regexp program for its magic number. | |
6223 * Return TRUE if it's wrong. | |
6224 */ | |
6225 static int | |
6226 prog_magic_wrong() | |
6227 { | |
4444 | 6228 regprog_T *prog; |
6229 | |
6230 prog = REG_MULTI ? reg_mmatch->regprog : reg_match->regprog; | |
6231 if (prog->engine == &nfa_regengine) | |
6232 /* For NFA matcher we don't check the magic */ | |
6233 return FALSE; | |
6234 | |
6235 if (UCHARAT(((bt_regprog_T *)prog)->program) != REGMAGIC) | |
7 | 6236 { |
6237 EMSG(_(e_re_corr)); | |
6238 return TRUE; | |
6239 } | |
6240 return FALSE; | |
6241 } | |
6242 | |
6243 /* | |
6244 * Cleanup the subexpressions, if this wasn't done yet. | |
6245 * This construction is used to clear the subexpressions only when they are | |
6246 * used (to increase speed). | |
6247 */ | |
6248 static void | |
6249 cleanup_subexpr() | |
6250 { | |
6251 if (need_clear_subexpr) | |
6252 { | |
6253 if (REG_MULTI) | |
6254 { | |
6255 /* Use 0xff to set lnum to -1 */ | |
6256 vim_memset(reg_startpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
6257 vim_memset(reg_endpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
6258 } | |
6259 else | |
6260 { | |
6261 vim_memset(reg_startp, 0, sizeof(char_u *) * NSUBEXP); | |
6262 vim_memset(reg_endp, 0, sizeof(char_u *) * NSUBEXP); | |
6263 } | |
6264 need_clear_subexpr = FALSE; | |
6265 } | |
6266 } | |
6267 | |
6268 #ifdef FEAT_SYN_HL | |
6269 static void | |
6270 cleanup_zsubexpr() | |
6271 { | |
6272 if (need_clear_zsubexpr) | |
6273 { | |
6274 if (REG_MULTI) | |
6275 { | |
6276 /* Use 0xff to set lnum to -1 */ | |
6277 vim_memset(reg_startzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
6278 vim_memset(reg_endzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
6279 } | |
6280 else | |
6281 { | |
6282 vim_memset(reg_startzp, 0, sizeof(char_u *) * NSUBEXP); | |
6283 vim_memset(reg_endzp, 0, sizeof(char_u *) * NSUBEXP); | |
6284 } | |
6285 need_clear_zsubexpr = FALSE; | |
6286 } | |
6287 } | |
6288 #endif | |
6289 | |
6290 /* | |
1579 | 6291 * Save the current subexpr to "bp", so that they can be restored |
6292 * later by restore_subexpr(). | |
6293 */ | |
6294 static void | |
6295 save_subexpr(bp) | |
6296 regbehind_T *bp; | |
6297 { | |
6298 int i; | |
6299 | |
1602 | 6300 /* When "need_clear_subexpr" is set we don't need to save the values, only |
6301 * remember that this flag needs to be set again when restoring. */ | |
6302 bp->save_need_clear_subexpr = need_clear_subexpr; | |
6303 if (!need_clear_subexpr) | |
1579 | 6304 { |
1602 | 6305 for (i = 0; i < NSUBEXP; ++i) |
1579 | 6306 { |
1602 | 6307 if (REG_MULTI) |
6308 { | |
6309 bp->save_start[i].se_u.pos = reg_startpos[i]; | |
6310 bp->save_end[i].se_u.pos = reg_endpos[i]; | |
6311 } | |
6312 else | |
6313 { | |
6314 bp->save_start[i].se_u.ptr = reg_startp[i]; | |
6315 bp->save_end[i].se_u.ptr = reg_endp[i]; | |
6316 } | |
1579 | 6317 } |
6318 } | |
6319 } | |
6320 | |
6321 /* | |
6322 * Restore the subexpr from "bp". | |
6323 */ | |
6324 static void | |
6325 restore_subexpr(bp) | |
6326 regbehind_T *bp; | |
6327 { | |
6328 int i; | |
6329 | |
1602 | 6330 /* Only need to restore saved values when they are not to be cleared. */ |
6331 need_clear_subexpr = bp->save_need_clear_subexpr; | |
6332 if (!need_clear_subexpr) | |
1579 | 6333 { |
1602 | 6334 for (i = 0; i < NSUBEXP; ++i) |
1579 | 6335 { |
1602 | 6336 if (REG_MULTI) |
6337 { | |
6338 reg_startpos[i] = bp->save_start[i].se_u.pos; | |
6339 reg_endpos[i] = bp->save_end[i].se_u.pos; | |
6340 } | |
6341 else | |
6342 { | |
6343 reg_startp[i] = bp->save_start[i].se_u.ptr; | |
6344 reg_endp[i] = bp->save_end[i].se_u.ptr; | |
6345 } | |
1579 | 6346 } |
6347 } | |
6348 } | |
6349 | |
6350 /* | |
7 | 6351 * Advance reglnum, regline and reginput to the next line. |
6352 */ | |
6353 static void | |
6354 reg_nextline() | |
6355 { | |
6356 regline = reg_getline(++reglnum); | |
6357 reginput = regline; | |
6358 fast_breakcheck(); | |
6359 } | |
6360 | |
6361 /* | |
6362 * Save the input line and position in a regsave_T. | |
6363 */ | |
6364 static void | |
233 | 6365 reg_save(save, gap) |
7 | 6366 regsave_T *save; |
233 | 6367 garray_T *gap; |
7 | 6368 { |
6369 if (REG_MULTI) | |
6370 { | |
6371 save->rs_u.pos.col = (colnr_T)(reginput - regline); | |
6372 save->rs_u.pos.lnum = reglnum; | |
6373 } | |
6374 else | |
6375 save->rs_u.ptr = reginput; | |
233 | 6376 save->rs_len = gap->ga_len; |
7 | 6377 } |
6378 | |
6379 /* | |
6380 * Restore the input line and position from a regsave_T. | |
6381 */ | |
6382 static void | |
233 | 6383 reg_restore(save, gap) |
7 | 6384 regsave_T *save; |
233 | 6385 garray_T *gap; |
7 | 6386 { |
6387 if (REG_MULTI) | |
6388 { | |
6389 if (reglnum != save->rs_u.pos.lnum) | |
6390 { | |
6391 /* only call reg_getline() when the line number changed to save | |
6392 * a bit of time */ | |
6393 reglnum = save->rs_u.pos.lnum; | |
6394 regline = reg_getline(reglnum); | |
6395 } | |
6396 reginput = regline + save->rs_u.pos.col; | |
6397 } | |
6398 else | |
6399 reginput = save->rs_u.ptr; | |
233 | 6400 gap->ga_len = save->rs_len; |
7 | 6401 } |
6402 | |
6403 /* | |
6404 * Return TRUE if current position is equal to saved position. | |
6405 */ | |
6406 static int | |
6407 reg_save_equal(save) | |
6408 regsave_T *save; | |
6409 { | |
6410 if (REG_MULTI) | |
6411 return reglnum == save->rs_u.pos.lnum | |
6412 && reginput == regline + save->rs_u.pos.col; | |
6413 return reginput == save->rs_u.ptr; | |
6414 } | |
6415 | |
6416 /* | |
6417 * Tentatively set the sub-expression start to the current position (after | |
6418 * calling regmatch() they will have changed). Need to save the existing | |
6419 * values for when there is no match. | |
6420 * Use se_save() to use pointer (save_se_multi()) or position (save_se_one()), | |
6421 * depending on REG_MULTI. | |
6422 */ | |
6423 static void | |
6424 save_se_multi(savep, posp) | |
6425 save_se_T *savep; | |
6426 lpos_T *posp; | |
6427 { | |
6428 savep->se_u.pos = *posp; | |
6429 posp->lnum = reglnum; | |
6430 posp->col = (colnr_T)(reginput - regline); | |
6431 } | |
6432 | |
6433 static void | |
6434 save_se_one(savep, pp) | |
6435 save_se_T *savep; | |
6436 char_u **pp; | |
6437 { | |
6438 savep->se_u.ptr = *pp; | |
6439 *pp = reginput; | |
6440 } | |
6441 | |
6442 /* | |
6443 * Compare a number with the operand of RE_LNUM, RE_COL or RE_VCOL. | |
6444 */ | |
6445 static int | |
6446 re_num_cmp(val, scan) | |
6447 long_u val; | |
6448 char_u *scan; | |
6449 { | |
6450 long_u n = OPERAND_MIN(scan); | |
6451 | |
6452 if (OPERAND_CMP(scan) == '>') | |
6453 return val > n; | |
6454 if (OPERAND_CMP(scan) == '<') | |
6455 return val < n; | |
6456 return val == n; | |
6457 } | |
6458 | |
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6459 /* |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6460 * Check whether a backreference matches. |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6461 * Returns RA_FAIL, RA_NOMATCH or RA_MATCH. |
5504 | 6462 * If "bytelen" is not NULL, it is set to the byte length of the match in the |
6463 * last line. | |
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6464 */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6465 static int |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6466 match_with_backref(start_lnum, start_col, end_lnum, end_col, bytelen) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6467 linenr_T start_lnum; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6468 colnr_T start_col; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6469 linenr_T end_lnum; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6470 colnr_T end_col; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6471 int *bytelen; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6472 { |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6473 linenr_T clnum = start_lnum; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6474 colnr_T ccol = start_col; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6475 int len; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6476 char_u *p; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6477 |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6478 if (bytelen != NULL) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6479 *bytelen = 0; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6480 for (;;) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6481 { |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6482 /* Since getting one line may invalidate the other, need to make copy. |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6483 * Slow! */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6484 if (regline != reg_tofree) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6485 { |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6486 len = (int)STRLEN(regline); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6487 if (reg_tofree == NULL || len >= (int)reg_tofreelen) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6488 { |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6489 len += 50; /* get some extra */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6490 vim_free(reg_tofree); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6491 reg_tofree = alloc(len); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6492 if (reg_tofree == NULL) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6493 return RA_FAIL; /* out of memory!*/ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6494 reg_tofreelen = len; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6495 } |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6496 STRCPY(reg_tofree, regline); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6497 reginput = reg_tofree + (reginput - regline); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6498 regline = reg_tofree; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6499 } |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6500 |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6501 /* Get the line to compare with. */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6502 p = reg_getline(clnum); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6503 if (clnum == end_lnum) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6504 len = end_col - ccol; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6505 else |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6506 len = (int)STRLEN(p + ccol); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6507 |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6508 if (cstrncmp(p + ccol, reginput, &len) != 0) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6509 return RA_NOMATCH; /* doesn't match */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6510 if (bytelen != NULL) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6511 *bytelen += len; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6512 if (clnum == end_lnum) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6513 break; /* match and at end! */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6514 if (reglnum >= reg_maxline) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6515 return RA_NOMATCH; /* text too short */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6516 |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6517 /* Advance to next line. */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6518 reg_nextline(); |
5504 | 6519 if (bytelen != NULL) |
6520 *bytelen = 0; | |
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6521 ++clnum; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6522 ccol = 0; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6523 if (got_int) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6524 return RA_FAIL; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6525 } |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6526 |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6527 /* found a match! Note that regline may now point to a copy of the line, |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6528 * that should not matter. */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6529 return RA_MATCH; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6530 } |
7 | 6531 |
4444 | 6532 #ifdef BT_REGEXP_DUMP |
7 | 6533 |
6534 /* | |
6535 * regdump - dump a regexp onto stdout in vaguely comprehensible form | |
6536 */ | |
6537 static void | |
6538 regdump(pattern, r) | |
6539 char_u *pattern; | |
4444 | 6540 bt_regprog_T *r; |
7 | 6541 { |
6542 char_u *s; | |
6543 int op = EXACTLY; /* Arbitrary non-END op. */ | |
6544 char_u *next; | |
6545 char_u *end = NULL; | |
4444 | 6546 FILE *f; |
6547 | |
6548 #ifdef BT_REGEXP_LOG | |
6549 f = fopen("bt_regexp_log.log", "a"); | |
6550 #else | |
6551 f = stdout; | |
6552 #endif | |
6553 if (f == NULL) | |
6554 return; | |
6555 fprintf(f, "-------------------------------------\n\r\nregcomp(%s):\r\n", pattern); | |
7 | 6556 |
6557 s = r->program + 1; | |
6558 /* | |
6559 * Loop until we find the END that isn't before a referred next (an END | |
6560 * can also appear in a NOMATCH operand). | |
6561 */ | |
6562 while (op != END || s <= end) | |
6563 { | |
6564 op = OP(s); | |
4444 | 6565 fprintf(f, "%2d%s", (int)(s - r->program), regprop(s)); /* Where, what. */ |
7 | 6566 next = regnext(s); |
6567 if (next == NULL) /* Next ptr. */ | |
4444 | 6568 fprintf(f, "(0)"); |
7 | 6569 else |
4444 | 6570 fprintf(f, "(%d)", (int)((s - r->program) + (next - s))); |
7 | 6571 if (end < next) |
6572 end = next; | |
6573 if (op == BRACE_LIMITS) | |
6574 { | |
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6575 /* Two ints */ |
4444 | 6576 fprintf(f, " minval %ld, maxval %ld", OPERAND_MIN(s), OPERAND_MAX(s)); |
7 | 6577 s += 8; |
6578 } | |
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6579 else if (op == BEHIND || op == NOBEHIND) |
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6580 { |
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6581 /* one int */ |
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6582 fprintf(f, " count %ld", OPERAND_MIN(s)); |
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6583 s += 4; |
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6584 } |
4770
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6585 else if (op == RE_LNUM || op == RE_COL || op == RE_VCOL) |
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6586 { |
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6587 /* one int plus comperator */ |
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6588 fprintf(f, " count %ld", OPERAND_MIN(s)); |
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6589 s += 5; |
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6590 } |
7 | 6591 s += 3; |
6592 if (op == ANYOF || op == ANYOF + ADD_NL | |
6593 || op == ANYBUT || op == ANYBUT + ADD_NL | |
6594 || op == EXACTLY) | |
6595 { | |
6596 /* Literal string, where present. */ | |
4444 | 6597 fprintf(f, "\nxxxxxxxxx\n"); |
7 | 6598 while (*s != NUL) |
4444 | 6599 fprintf(f, "%c", *s++); |
6600 fprintf(f, "\nxxxxxxxxx\n"); | |
7 | 6601 s++; |
6602 } | |
4444 | 6603 fprintf(f, "\r\n"); |
7 | 6604 } |
6605 | |
6606 /* Header fields of interest. */ | |
6607 if (r->regstart != NUL) | |
4444 | 6608 fprintf(f, "start `%s' 0x%x; ", r->regstart < 256 |
7 | 6609 ? (char *)transchar(r->regstart) |
6610 : "multibyte", r->regstart); | |
6611 if (r->reganch) | |
4444 | 6612 fprintf(f, "anchored; "); |
7 | 6613 if (r->regmust != NULL) |
4444 | 6614 fprintf(f, "must have \"%s\"", r->regmust); |
6615 fprintf(f, "\r\n"); | |
6616 | |
6617 #ifdef BT_REGEXP_LOG | |
6618 fclose(f); | |
6619 #endif | |
7 | 6620 } |
4444 | 6621 #endif /* BT_REGEXP_DUMP */ |
6622 | |
6623 #ifdef DEBUG | |
7 | 6624 /* |
6625 * regprop - printable representation of opcode | |
6626 */ | |
6627 static char_u * | |
6628 regprop(op) | |
6629 char_u *op; | |
6630 { | |
4444 | 6631 char *p; |
6632 static char buf[50]; | |
6633 | |
6634 STRCPY(buf, ":"); | |
6635 | |
6636 switch ((int) OP(op)) | |
7 | 6637 { |
6638 case BOL: | |
6639 p = "BOL"; | |
6640 break; | |
6641 case EOL: | |
6642 p = "EOL"; | |
6643 break; | |
6644 case RE_BOF: | |
6645 p = "BOF"; | |
6646 break; | |
6647 case RE_EOF: | |
6648 p = "EOF"; | |
6649 break; | |
6650 case CURSOR: | |
6651 p = "CURSOR"; | |
6652 break; | |
639 | 6653 case RE_VISUAL: |
6654 p = "RE_VISUAL"; | |
6655 break; | |
7 | 6656 case RE_LNUM: |
6657 p = "RE_LNUM"; | |
6658 break; | |
639 | 6659 case RE_MARK: |
6660 p = "RE_MARK"; | |
6661 break; | |
7 | 6662 case RE_COL: |
6663 p = "RE_COL"; | |
6664 break; | |
6665 case RE_VCOL: | |
6666 p = "RE_VCOL"; | |
6667 break; | |
6668 case BOW: | |
6669 p = "BOW"; | |
6670 break; | |
6671 case EOW: | |
6672 p = "EOW"; | |
6673 break; | |
6674 case ANY: | |
6675 p = "ANY"; | |
6676 break; | |
6677 case ANY + ADD_NL: | |
6678 p = "ANY+NL"; | |
6679 break; | |
6680 case ANYOF: | |
6681 p = "ANYOF"; | |
6682 break; | |
6683 case ANYOF + ADD_NL: | |
6684 p = "ANYOF+NL"; | |
6685 break; | |
6686 case ANYBUT: | |
6687 p = "ANYBUT"; | |
6688 break; | |
6689 case ANYBUT + ADD_NL: | |
6690 p = "ANYBUT+NL"; | |
6691 break; | |
6692 case IDENT: | |
6693 p = "IDENT"; | |
6694 break; | |
6695 case IDENT + ADD_NL: | |
6696 p = "IDENT+NL"; | |
6697 break; | |
6698 case SIDENT: | |
6699 p = "SIDENT"; | |
6700 break; | |
6701 case SIDENT + ADD_NL: | |
6702 p = "SIDENT+NL"; | |
6703 break; | |
6704 case KWORD: | |
6705 p = "KWORD"; | |
6706 break; | |
6707 case KWORD + ADD_NL: | |
6708 p = "KWORD+NL"; | |
6709 break; | |
6710 case SKWORD: | |
6711 p = "SKWORD"; | |
6712 break; | |
6713 case SKWORD + ADD_NL: | |
6714 p = "SKWORD+NL"; | |
6715 break; | |
6716 case FNAME: | |
6717 p = "FNAME"; | |
6718 break; | |
6719 case FNAME + ADD_NL: | |
6720 p = "FNAME+NL"; | |
6721 break; | |
6722 case SFNAME: | |
6723 p = "SFNAME"; | |
6724 break; | |
6725 case SFNAME + ADD_NL: | |
6726 p = "SFNAME+NL"; | |
6727 break; | |
6728 case PRINT: | |
6729 p = "PRINT"; | |
6730 break; | |
6731 case PRINT + ADD_NL: | |
6732 p = "PRINT+NL"; | |
6733 break; | |
6734 case SPRINT: | |
6735 p = "SPRINT"; | |
6736 break; | |
6737 case SPRINT + ADD_NL: | |
6738 p = "SPRINT+NL"; | |
6739 break; | |
6740 case WHITE: | |
6741 p = "WHITE"; | |
6742 break; | |
6743 case WHITE + ADD_NL: | |
6744 p = "WHITE+NL"; | |
6745 break; | |
6746 case NWHITE: | |
6747 p = "NWHITE"; | |
6748 break; | |
6749 case NWHITE + ADD_NL: | |
6750 p = "NWHITE+NL"; | |
6751 break; | |
6752 case DIGIT: | |
6753 p = "DIGIT"; | |
6754 break; | |
6755 case DIGIT + ADD_NL: | |
6756 p = "DIGIT+NL"; | |
6757 break; | |
6758 case NDIGIT: | |
6759 p = "NDIGIT"; | |
6760 break; | |
6761 case NDIGIT + ADD_NL: | |
6762 p = "NDIGIT+NL"; | |
6763 break; | |
6764 case HEX: | |
6765 p = "HEX"; | |
6766 break; | |
6767 case HEX + ADD_NL: | |
6768 p = "HEX+NL"; | |
6769 break; | |
6770 case NHEX: | |
6771 p = "NHEX"; | |
6772 break; | |
6773 case NHEX + ADD_NL: | |
6774 p = "NHEX+NL"; | |
6775 break; | |
6776 case OCTAL: | |
6777 p = "OCTAL"; | |
6778 break; | |
6779 case OCTAL + ADD_NL: | |
6780 p = "OCTAL+NL"; | |
6781 break; | |
6782 case NOCTAL: | |
6783 p = "NOCTAL"; | |
6784 break; | |
6785 case NOCTAL + ADD_NL: | |
6786 p = "NOCTAL+NL"; | |
6787 break; | |
6788 case WORD: | |
6789 p = "WORD"; | |
6790 break; | |
6791 case WORD + ADD_NL: | |
6792 p = "WORD+NL"; | |
6793 break; | |
6794 case NWORD: | |
6795 p = "NWORD"; | |
6796 break; | |
6797 case NWORD + ADD_NL: | |
6798 p = "NWORD+NL"; | |
6799 break; | |
6800 case HEAD: | |
6801 p = "HEAD"; | |
6802 break; | |
6803 case HEAD + ADD_NL: | |
6804 p = "HEAD+NL"; | |
6805 break; | |
6806 case NHEAD: | |
6807 p = "NHEAD"; | |
6808 break; | |
6809 case NHEAD + ADD_NL: | |
6810 p = "NHEAD+NL"; | |
6811 break; | |
6812 case ALPHA: | |
6813 p = "ALPHA"; | |
6814 break; | |
6815 case ALPHA + ADD_NL: | |
6816 p = "ALPHA+NL"; | |
6817 break; | |
6818 case NALPHA: | |
6819 p = "NALPHA"; | |
6820 break; | |
6821 case NALPHA + ADD_NL: | |
6822 p = "NALPHA+NL"; | |
6823 break; | |
6824 case LOWER: | |
6825 p = "LOWER"; | |
6826 break; | |
6827 case LOWER + ADD_NL: | |
6828 p = "LOWER+NL"; | |
6829 break; | |
6830 case NLOWER: | |
6831 p = "NLOWER"; | |
6832 break; | |
6833 case NLOWER + ADD_NL: | |
6834 p = "NLOWER+NL"; | |
6835 break; | |
6836 case UPPER: | |
6837 p = "UPPER"; | |
6838 break; | |
6839 case UPPER + ADD_NL: | |
6840 p = "UPPER+NL"; | |
6841 break; | |
6842 case NUPPER: | |
6843 p = "NUPPER"; | |
6844 break; | |
6845 case NUPPER + ADD_NL: | |
6846 p = "NUPPER+NL"; | |
6847 break; | |
6848 case BRANCH: | |
6849 p = "BRANCH"; | |
6850 break; | |
6851 case EXACTLY: | |
6852 p = "EXACTLY"; | |
6853 break; | |
6854 case NOTHING: | |
6855 p = "NOTHING"; | |
6856 break; | |
6857 case BACK: | |
6858 p = "BACK"; | |
6859 break; | |
6860 case END: | |
6861 p = "END"; | |
6862 break; | |
6863 case MOPEN + 0: | |
6864 p = "MATCH START"; | |
6865 break; | |
6866 case MOPEN + 1: | |
6867 case MOPEN + 2: | |
6868 case MOPEN + 3: | |
6869 case MOPEN + 4: | |
6870 case MOPEN + 5: | |
6871 case MOPEN + 6: | |
6872 case MOPEN + 7: | |
6873 case MOPEN + 8: | |
6874 case MOPEN + 9: | |
6875 sprintf(buf + STRLEN(buf), "MOPEN%d", OP(op) - MOPEN); | |
6876 p = NULL; | |
6877 break; | |
6878 case MCLOSE + 0: | |
6879 p = "MATCH END"; | |
6880 break; | |
6881 case MCLOSE + 1: | |
6882 case MCLOSE + 2: | |
6883 case MCLOSE + 3: | |
6884 case MCLOSE + 4: | |
6885 case MCLOSE + 5: | |
6886 case MCLOSE + 6: | |
6887 case MCLOSE + 7: | |
6888 case MCLOSE + 8: | |
6889 case MCLOSE + 9: | |
6890 sprintf(buf + STRLEN(buf), "MCLOSE%d", OP(op) - MCLOSE); | |
6891 p = NULL; | |
6892 break; | |
6893 case BACKREF + 1: | |
6894 case BACKREF + 2: | |
6895 case BACKREF + 3: | |
6896 case BACKREF + 4: | |
6897 case BACKREF + 5: | |
6898 case BACKREF + 6: | |
6899 case BACKREF + 7: | |
6900 case BACKREF + 8: | |
6901 case BACKREF + 9: | |
6902 sprintf(buf + STRLEN(buf), "BACKREF%d", OP(op) - BACKREF); | |
6903 p = NULL; | |
6904 break; | |
6905 case NOPEN: | |
6906 p = "NOPEN"; | |
6907 break; | |
6908 case NCLOSE: | |
6909 p = "NCLOSE"; | |
6910 break; | |
6911 #ifdef FEAT_SYN_HL | |
6912 case ZOPEN + 1: | |
6913 case ZOPEN + 2: | |
6914 case ZOPEN + 3: | |
6915 case ZOPEN + 4: | |
6916 case ZOPEN + 5: | |
6917 case ZOPEN + 6: | |
6918 case ZOPEN + 7: | |
6919 case ZOPEN + 8: | |
6920 case ZOPEN + 9: | |
6921 sprintf(buf + STRLEN(buf), "ZOPEN%d", OP(op) - ZOPEN); | |
6922 p = NULL; | |
6923 break; | |
6924 case ZCLOSE + 1: | |
6925 case ZCLOSE + 2: | |
6926 case ZCLOSE + 3: | |
6927 case ZCLOSE + 4: | |
6928 case ZCLOSE + 5: | |
6929 case ZCLOSE + 6: | |
6930 case ZCLOSE + 7: | |
6931 case ZCLOSE + 8: | |
6932 case ZCLOSE + 9: | |
6933 sprintf(buf + STRLEN(buf), "ZCLOSE%d", OP(op) - ZCLOSE); | |
6934 p = NULL; | |
6935 break; | |
6936 case ZREF + 1: | |
6937 case ZREF + 2: | |
6938 case ZREF + 3: | |
6939 case ZREF + 4: | |
6940 case ZREF + 5: | |
6941 case ZREF + 6: | |
6942 case ZREF + 7: | |
6943 case ZREF + 8: | |
6944 case ZREF + 9: | |
6945 sprintf(buf + STRLEN(buf), "ZREF%d", OP(op) - ZREF); | |
6946 p = NULL; | |
6947 break; | |
6948 #endif | |
6949 case STAR: | |
6950 p = "STAR"; | |
6951 break; | |
6952 case PLUS: | |
6953 p = "PLUS"; | |
6954 break; | |
6955 case NOMATCH: | |
6956 p = "NOMATCH"; | |
6957 break; | |
6958 case MATCH: | |
6959 p = "MATCH"; | |
6960 break; | |
6961 case BEHIND: | |
6962 p = "BEHIND"; | |
6963 break; | |
6964 case NOBEHIND: | |
6965 p = "NOBEHIND"; | |
6966 break; | |
6967 case SUBPAT: | |
6968 p = "SUBPAT"; | |
6969 break; | |
6970 case BRACE_LIMITS: | |
6971 p = "BRACE_LIMITS"; | |
6972 break; | |
6973 case BRACE_SIMPLE: | |
6974 p = "BRACE_SIMPLE"; | |
6975 break; | |
6976 case BRACE_COMPLEX + 0: | |
6977 case BRACE_COMPLEX + 1: | |
6978 case BRACE_COMPLEX + 2: | |
6979 case BRACE_COMPLEX + 3: | |
6980 case BRACE_COMPLEX + 4: | |
6981 case BRACE_COMPLEX + 5: | |
6982 case BRACE_COMPLEX + 6: | |
6983 case BRACE_COMPLEX + 7: | |
6984 case BRACE_COMPLEX + 8: | |
6985 case BRACE_COMPLEX + 9: | |
6986 sprintf(buf + STRLEN(buf), "BRACE_COMPLEX%d", OP(op) - BRACE_COMPLEX); | |
6987 p = NULL; | |
6988 break; | |
6989 #ifdef FEAT_MBYTE | |
6990 case MULTIBYTECODE: | |
6991 p = "MULTIBYTECODE"; | |
6992 break; | |
6993 #endif | |
6994 case NEWL: | |
6995 p = "NEWL"; | |
6996 break; | |
6997 default: | |
6998 sprintf(buf + STRLEN(buf), "corrupt %d", OP(op)); | |
6999 p = NULL; | |
7000 break; | |
7001 } | |
7002 if (p != NULL) | |
4444 | 7003 STRCAT(buf, p); |
7004 return (char_u *)buf; | |
7 | 7005 } |
4444 | 7006 #endif /* DEBUG */ |
7 | 7007 |
7008 #ifdef FEAT_MBYTE | |
7009 static void mb_decompose __ARGS((int c, int *c1, int *c2, int *c3)); | |
7010 | |
7011 typedef struct | |
7012 { | |
7013 int a, b, c; | |
7014 } decomp_T; | |
7015 | |
7016 | |
7017 /* 0xfb20 - 0xfb4f */ | |
297 | 7018 static decomp_T decomp_table[0xfb4f-0xfb20+1] = |
7 | 7019 { |
7020 {0x5e2,0,0}, /* 0xfb20 alt ayin */ | |
7021 {0x5d0,0,0}, /* 0xfb21 alt alef */ | |
7022 {0x5d3,0,0}, /* 0xfb22 alt dalet */ | |
7023 {0x5d4,0,0}, /* 0xfb23 alt he */ | |
7024 {0x5db,0,0}, /* 0xfb24 alt kaf */ | |
7025 {0x5dc,0,0}, /* 0xfb25 alt lamed */ | |
7026 {0x5dd,0,0}, /* 0xfb26 alt mem-sofit */ | |
7027 {0x5e8,0,0}, /* 0xfb27 alt resh */ | |
7028 {0x5ea,0,0}, /* 0xfb28 alt tav */ | |
7029 {'+', 0, 0}, /* 0xfb29 alt plus */ | |
7030 {0x5e9, 0x5c1, 0}, /* 0xfb2a shin+shin-dot */ | |
7031 {0x5e9, 0x5c2, 0}, /* 0xfb2b shin+sin-dot */ | |
7032 {0x5e9, 0x5c1, 0x5bc}, /* 0xfb2c shin+shin-dot+dagesh */ | |
7033 {0x5e9, 0x5c2, 0x5bc}, /* 0xfb2d shin+sin-dot+dagesh */ | |
7034 {0x5d0, 0x5b7, 0}, /* 0xfb2e alef+patah */ | |
7035 {0x5d0, 0x5b8, 0}, /* 0xfb2f alef+qamats */ | |
7036 {0x5d0, 0x5b4, 0}, /* 0xfb30 alef+hiriq */ | |
7037 {0x5d1, 0x5bc, 0}, /* 0xfb31 bet+dagesh */ | |
7038 {0x5d2, 0x5bc, 0}, /* 0xfb32 gimel+dagesh */ | |
7039 {0x5d3, 0x5bc, 0}, /* 0xfb33 dalet+dagesh */ | |
7040 {0x5d4, 0x5bc, 0}, /* 0xfb34 he+dagesh */ | |
7041 {0x5d5, 0x5bc, 0}, /* 0xfb35 vav+dagesh */ | |
7042 {0x5d6, 0x5bc, 0}, /* 0xfb36 zayin+dagesh */ | |
7043 {0xfb37, 0, 0}, /* 0xfb37 -- UNUSED */ | |
7044 {0x5d8, 0x5bc, 0}, /* 0xfb38 tet+dagesh */ | |
7045 {0x5d9, 0x5bc, 0}, /* 0xfb39 yud+dagesh */ | |
7046 {0x5da, 0x5bc, 0}, /* 0xfb3a kaf sofit+dagesh */ | |
7047 {0x5db, 0x5bc, 0}, /* 0xfb3b kaf+dagesh */ | |
7048 {0x5dc, 0x5bc, 0}, /* 0xfb3c lamed+dagesh */ | |
7049 {0xfb3d, 0, 0}, /* 0xfb3d -- UNUSED */ | |
7050 {0x5de, 0x5bc, 0}, /* 0xfb3e mem+dagesh */ | |
7051 {0xfb3f, 0, 0}, /* 0xfb3f -- UNUSED */ | |
7052 {0x5e0, 0x5bc, 0}, /* 0xfb40 nun+dagesh */ | |
7053 {0x5e1, 0x5bc, 0}, /* 0xfb41 samech+dagesh */ | |
7054 {0xfb42, 0, 0}, /* 0xfb42 -- UNUSED */ | |
7055 {0x5e3, 0x5bc, 0}, /* 0xfb43 pe sofit+dagesh */ | |
7056 {0x5e4, 0x5bc,0}, /* 0xfb44 pe+dagesh */ | |
7057 {0xfb45, 0, 0}, /* 0xfb45 -- UNUSED */ | |
7058 {0x5e6, 0x5bc, 0}, /* 0xfb46 tsadi+dagesh */ | |
7059 {0x5e7, 0x5bc, 0}, /* 0xfb47 qof+dagesh */ | |
7060 {0x5e8, 0x5bc, 0}, /* 0xfb48 resh+dagesh */ | |
7061 {0x5e9, 0x5bc, 0}, /* 0xfb49 shin+dagesh */ | |
7062 {0x5ea, 0x5bc, 0}, /* 0xfb4a tav+dagesh */ | |
7063 {0x5d5, 0x5b9, 0}, /* 0xfb4b vav+holam */ | |
7064 {0x5d1, 0x5bf, 0}, /* 0xfb4c bet+rafe */ | |
7065 {0x5db, 0x5bf, 0}, /* 0xfb4d kaf+rafe */ | |
7066 {0x5e4, 0x5bf, 0}, /* 0xfb4e pe+rafe */ | |
7067 {0x5d0, 0x5dc, 0} /* 0xfb4f alef-lamed */ | |
7068 }; | |
7069 | |
7070 static void | |
7071 mb_decompose(c, c1, c2, c3) | |
7072 int c, *c1, *c2, *c3; | |
7073 { | |
7074 decomp_T d; | |
7075 | |
4505
d037b9cbdaaa
updated for version 7.3.1000
Bram Moolenaar <bram@vim.org>
parents:
4466
diff
changeset
|
7076 if (c >= 0xfb20 && c <= 0xfb4f) |
7 | 7077 { |
7078 d = decomp_table[c - 0xfb20]; | |
7079 *c1 = d.a; | |
7080 *c2 = d.b; | |
7081 *c3 = d.c; | |
7082 } | |
7083 else | |
7084 { | |
7085 *c1 = c; | |
7086 *c2 = *c3 = 0; | |
7087 } | |
7088 } | |
7089 #endif | |
7090 | |
7091 /* | |
7092 * Compare two strings, ignore case if ireg_ic set. | |
7093 * Return 0 if strings match, non-zero otherwise. | |
7094 * Correct the length "*n" when composing characters are ignored. | |
7095 */ | |
7096 static int | |
7097 cstrncmp(s1, s2, n) | |
7098 char_u *s1, *s2; | |
7099 int *n; | |
7100 { | |
7101 int result; | |
7102 | |
7103 if (!ireg_ic) | |
7104 result = STRNCMP(s1, s2, *n); | |
7105 else | |
7106 result = MB_STRNICMP(s1, s2, *n); | |
7107 | |
7108 #ifdef FEAT_MBYTE | |
7109 /* if it failed and it's utf8 and we want to combineignore: */ | |
7110 if (result != 0 && enc_utf8 && ireg_icombine) | |
7111 { | |
7112 char_u *str1, *str2; | |
7113 int c1, c2, c11, c12; | |
7114 int junk; | |
7115 | |
7116 /* we have to handle the strcmp ourselves, since it is necessary to | |
7117 * deal with the composing characters by ignoring them: */ | |
7118 str1 = s1; | |
7119 str2 = s2; | |
7120 c1 = c2 = 0; | |
507 | 7121 while ((int)(str1 - s1) < *n) |
7 | 7122 { |
7123 c1 = mb_ptr2char_adv(&str1); | |
7124 c2 = mb_ptr2char_adv(&str2); | |
7125 | |
7126 /* decompose the character if necessary, into 'base' characters | |
7127 * because I don't care about Arabic, I will hard-code the Hebrew | |
7128 * which I *do* care about! So sue me... */ | |
7129 if (c1 != c2 && (!ireg_ic || utf_fold(c1) != utf_fold(c2))) | |
7130 { | |
7131 /* decomposition necessary? */ | |
7132 mb_decompose(c1, &c11, &junk, &junk); | |
7133 mb_decompose(c2, &c12, &junk, &junk); | |
7134 c1 = c11; | |
7135 c2 = c12; | |
7136 if (c11 != c12 && (!ireg_ic || utf_fold(c11) != utf_fold(c12))) | |
7137 break; | |
7138 } | |
7139 } | |
7140 result = c2 - c1; | |
7141 if (result == 0) | |
7142 *n = (int)(str2 - s2); | |
7143 } | |
7144 #endif | |
7145 | |
7146 return result; | |
7147 } | |
7148 | |
7149 /* | |
7150 * cstrchr: This function is used a lot for simple searches, keep it fast! | |
7151 */ | |
7152 static char_u * | |
7153 cstrchr(s, c) | |
7154 char_u *s; | |
7155 int c; | |
7156 { | |
7157 char_u *p; | |
7158 int cc; | |
7159 | |
7160 if (!ireg_ic | |
7161 #ifdef FEAT_MBYTE | |
7162 || (!enc_utf8 && mb_char2len(c) > 1) | |
7163 #endif | |
7164 ) | |
7165 return vim_strchr(s, c); | |
7166 | |
7167 /* tolower() and toupper() can be slow, comparing twice should be a lot | |
7168 * faster (esp. when using MS Visual C++!). | |
7169 * For UTF-8 need to use folded case. */ | |
7170 #ifdef FEAT_MBYTE | |
7171 if (enc_utf8 && c > 0x80) | |
7172 cc = utf_fold(c); | |
7173 else | |
7174 #endif | |
1347 | 7175 if (MB_ISUPPER(c)) |
7176 cc = MB_TOLOWER(c); | |
7177 else if (MB_ISLOWER(c)) | |
7178 cc = MB_TOUPPER(c); | |
7 | 7179 else |
7180 return vim_strchr(s, c); | |
7181 | |
7182 #ifdef FEAT_MBYTE | |
7183 if (has_mbyte) | |
7184 { | |
474 | 7185 for (p = s; *p != NUL; p += (*mb_ptr2len)(p)) |
7 | 7186 { |
7187 if (enc_utf8 && c > 0x80) | |
7188 { | |
7189 if (utf_fold(utf_ptr2char(p)) == cc) | |
7190 return p; | |
7191 } | |
7192 else if (*p == c || *p == cc) | |
7193 return p; | |
7194 } | |
7195 } | |
7196 else | |
7197 #endif | |
7198 /* Faster version for when there are no multi-byte characters. */ | |
7199 for (p = s; *p != NUL; ++p) | |
7200 if (*p == c || *p == cc) | |
7201 return p; | |
7202 | |
7203 return NULL; | |
7204 } | |
7205 | |
7206 /*************************************************************** | |
7207 * regsub stuff * | |
7208 ***************************************************************/ | |
7209 | |
7210 /* This stuff below really confuses cc on an SGI -- webb */ | |
7211 #ifdef __sgi | |
7212 # undef __ARGS | |
7213 # define __ARGS(x) () | |
7214 #endif | |
7215 | |
7216 /* | |
7217 * We should define ftpr as a pointer to a function returning a pointer to | |
7218 * a function returning a pointer to a function ... | |
7219 * This is impossible, so we declare a pointer to a function returning a | |
7220 * pointer to a function returning void. This should work for all compilers. | |
7221 */ | |
772 | 7222 typedef void (*(*fptr_T) __ARGS((int *, int)))(); |
7223 | |
7224 static fptr_T do_upper __ARGS((int *, int)); | |
7225 static fptr_T do_Upper __ARGS((int *, int)); | |
7226 static fptr_T do_lower __ARGS((int *, int)); | |
7227 static fptr_T do_Lower __ARGS((int *, int)); | |
7 | 7228 |
7229 static int vim_regsub_both __ARGS((char_u *source, char_u *dest, int copy, int magic, int backslash)); | |
7230 | |
772 | 7231 static fptr_T |
7 | 7232 do_upper(d, c) |
772 | 7233 int *d; |
7234 int c; | |
7 | 7235 { |
772 | 7236 *d = MB_TOUPPER(c); |
7237 | |
7238 return (fptr_T)NULL; | |
7 | 7239 } |
7240 | |
772 | 7241 static fptr_T |
7242 do_Upper(d, c) | |
7243 int *d; | |
7 | 7244 int c; |
7245 { | |
772 | 7246 *d = MB_TOUPPER(c); |
7247 | |
7248 return (fptr_T)do_Upper; | |
7249 } | |
7250 | |
7251 static fptr_T | |
7252 do_lower(d, c) | |
7253 int *d; | |
7254 int c; | |
7255 { | |
7256 *d = MB_TOLOWER(c); | |
7257 | |
7258 return (fptr_T)NULL; | |
7259 } | |
7260 | |
7261 static fptr_T | |
7262 do_Lower(d, c) | |
7263 int *d; | |
7264 int c; | |
7265 { | |
7266 *d = MB_TOLOWER(c); | |
7267 | |
7268 return (fptr_T)do_Lower; | |
7 | 7269 } |
7270 | |
7271 /* | |
7272 * regtilde(): Replace tildes in the pattern by the old pattern. | |
7273 * | |
7274 * Short explanation of the tilde: It stands for the previous replacement | |
7275 * pattern. If that previous pattern also contains a ~ we should go back a | |
7276 * step further... But we insert the previous pattern into the current one | |
7277 * and remember that. | |
772 | 7278 * This still does not handle the case where "magic" changes. So require the |
7279 * user to keep his hands off of "magic". | |
7 | 7280 * |
7281 * The tildes are parsed once before the first call to vim_regsub(). | |
7282 */ | |
7283 char_u * | |
7284 regtilde(source, magic) | |
7285 char_u *source; | |
7286 int magic; | |
7287 { | |
7288 char_u *newsub = source; | |
7289 char_u *tmpsub; | |
7290 char_u *p; | |
7291 int len; | |
7292 int prevlen; | |
7293 | |
7294 for (p = newsub; *p; ++p) | |
7295 { | |
7296 if ((*p == '~' && magic) || (*p == '\\' && *(p + 1) == '~' && !magic)) | |
7297 { | |
7298 if (reg_prev_sub != NULL) | |
7299 { | |
7300 /* length = len(newsub) - 1 + len(prev_sub) + 1 */ | |
7301 prevlen = (int)STRLEN(reg_prev_sub); | |
7302 tmpsub = alloc((unsigned)(STRLEN(newsub) + prevlen)); | |
7303 if (tmpsub != NULL) | |
7304 { | |
7305 /* copy prefix */ | |
7306 len = (int)(p - newsub); /* not including ~ */ | |
7307 mch_memmove(tmpsub, newsub, (size_t)len); | |
1209 | 7308 /* interpret tilde */ |
7 | 7309 mch_memmove(tmpsub + len, reg_prev_sub, (size_t)prevlen); |
7310 /* copy postfix */ | |
7311 if (!magic) | |
7312 ++p; /* back off \ */ | |
7313 STRCPY(tmpsub + len + prevlen, p + 1); | |
7314 | |
7315 if (newsub != source) /* already allocated newsub */ | |
7316 vim_free(newsub); | |
7317 newsub = tmpsub; | |
7318 p = newsub + len + prevlen; | |
7319 } | |
7320 } | |
7321 else if (magic) | |
1621 | 7322 STRMOVE(p, p + 1); /* remove '~' */ |
7 | 7323 else |
1621 | 7324 STRMOVE(p, p + 2); /* remove '\~' */ |
7 | 7325 --p; |
7326 } | |
7327 else | |
7328 { | |
7329 if (*p == '\\' && p[1]) /* skip escaped characters */ | |
7330 ++p; | |
7331 #ifdef FEAT_MBYTE | |
7332 if (has_mbyte) | |
474 | 7333 p += (*mb_ptr2len)(p) - 1; |
7 | 7334 #endif |
7335 } | |
7336 } | |
7337 | |
7338 vim_free(reg_prev_sub); | |
7339 if (newsub != source) /* newsub was allocated, just keep it */ | |
7340 reg_prev_sub = newsub; | |
7341 else /* no ~ found, need to save newsub */ | |
7342 reg_prev_sub = vim_strsave(newsub); | |
7343 return newsub; | |
7344 } | |
7345 | |
7346 #ifdef FEAT_EVAL | |
7347 static int can_f_submatch = FALSE; /* TRUE when submatch() can be used */ | |
7348 | |
7349 /* These pointers are used instead of reg_match and reg_mmatch for | |
7350 * reg_submatch(). Needed for when the substitution string is an expression | |
7351 * that contains a call to substitute() and submatch(). */ | |
7352 static regmatch_T *submatch_match; | |
7353 static regmmatch_T *submatch_mmatch; | |
2011 | 7354 static linenr_T submatch_firstlnum; |
7355 static linenr_T submatch_maxline; | |
2904 | 7356 static int submatch_line_lbr; |
7 | 7357 #endif |
7358 | |
7359 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO) | |
7360 /* | |
7361 * vim_regsub() - perform substitutions after a vim_regexec() or | |
7362 * vim_regexec_multi() match. | |
7363 * | |
7364 * If "copy" is TRUE really copy into "dest". | |
7365 * If "copy" is FALSE nothing is copied, this is just to find out the length | |
7366 * of the result. | |
7367 * | |
7368 * If "backslash" is TRUE, a backslash will be removed later, need to double | |
7369 * them to keep them, and insert a backslash before a CR to avoid it being | |
7370 * replaced with a line break later. | |
7371 * | |
7372 * Note: The matched text must not change between the call of | |
7373 * vim_regexec()/vim_regexec_multi() and vim_regsub()! It would make the back | |
7374 * references invalid! | |
7375 * | |
7376 * Returns the size of the replacement, including terminating NUL. | |
7377 */ | |
7378 int | |
7379 vim_regsub(rmp, source, dest, copy, magic, backslash) | |
7380 regmatch_T *rmp; | |
7381 char_u *source; | |
7382 char_u *dest; | |
7383 int copy; | |
7384 int magic; | |
7385 int backslash; | |
7386 { | |
7387 reg_match = rmp; | |
7388 reg_mmatch = NULL; | |
7389 reg_maxline = 0; | |
4061 | 7390 reg_buf = curbuf; |
5836 | 7391 reg_line_lbr = TRUE; |
7 | 7392 return vim_regsub_both(source, dest, copy, magic, backslash); |
7393 } | |
7394 #endif | |
7395 | |
7396 int | |
7397 vim_regsub_multi(rmp, lnum, source, dest, copy, magic, backslash) | |
7398 regmmatch_T *rmp; | |
7399 linenr_T lnum; | |
7400 char_u *source; | |
7401 char_u *dest; | |
7402 int copy; | |
7403 int magic; | |
7404 int backslash; | |
7405 { | |
7406 reg_match = NULL; | |
7407 reg_mmatch = rmp; | |
7408 reg_buf = curbuf; /* always works on the current buffer! */ | |
7409 reg_firstlnum = lnum; | |
7410 reg_maxline = curbuf->b_ml.ml_line_count - lnum; | |
5836 | 7411 reg_line_lbr = FALSE; |
7 | 7412 return vim_regsub_both(source, dest, copy, magic, backslash); |
7413 } | |
7414 | |
7415 static int | |
7416 vim_regsub_both(source, dest, copy, magic, backslash) | |
7417 char_u *source; | |
7418 char_u *dest; | |
7419 int copy; | |
7420 int magic; | |
7421 int backslash; | |
7422 { | |
7423 char_u *src; | |
7424 char_u *dst; | |
7425 char_u *s; | |
7426 int c; | |
772 | 7427 int cc; |
7 | 7428 int no = -1; |
4244 | 7429 fptr_T func_all = (fptr_T)NULL; |
7430 fptr_T func_one = (fptr_T)NULL; | |
7 | 7431 linenr_T clnum = 0; /* init for GCC */ |
7432 int len = 0; /* init for GCC */ | |
7433 #ifdef FEAT_EVAL | |
7434 static char_u *eval_result = NULL; | |
7435 #endif | |
7436 | |
7437 /* Be paranoid... */ | |
7438 if (source == NULL || dest == NULL) | |
7439 { | |
7440 EMSG(_(e_null)); | |
7441 return 0; | |
7442 } | |
7443 if (prog_magic_wrong()) | |
7444 return 0; | |
7445 src = source; | |
7446 dst = dest; | |
7447 | |
7448 /* | |
7449 * When the substitute part starts with "\=" evaluate it as an expression. | |
7450 */ | |
7451 if (source[0] == '\\' && source[1] == '=' | |
7452 #ifdef FEAT_EVAL | |
7453 && !can_f_submatch /* can't do this recursively */ | |
7454 #endif | |
7455 ) | |
7456 { | |
7457 #ifdef FEAT_EVAL | |
7458 /* To make sure that the length doesn't change between checking the | |
7459 * length and copying the string, and to speed up things, the | |
7460 * resulting string is saved from the call with "copy" == FALSE to the | |
7461 * call with "copy" == TRUE. */ | |
7462 if (copy) | |
7463 { | |
7464 if (eval_result != NULL) | |
7465 { | |
7466 STRCPY(dest, eval_result); | |
7467 dst += STRLEN(eval_result); | |
7468 vim_free(eval_result); | |
7469 eval_result = NULL; | |
7470 } | |
7471 } | |
7472 else | |
7473 { | |
7474 win_T *save_reg_win; | |
7475 int save_ireg_ic; | |
7476 | |
7477 vim_free(eval_result); | |
7478 | |
7479 /* The expression may contain substitute(), which calls us | |
7480 * recursively. Make sure submatch() gets the text from the first | |
7481 * level. Don't need to save "reg_buf", because | |
7482 * vim_regexec_multi() can't be called recursively. */ | |
7483 submatch_match = reg_match; | |
7484 submatch_mmatch = reg_mmatch; | |
2011 | 7485 submatch_firstlnum = reg_firstlnum; |
7486 submatch_maxline = reg_maxline; | |
2904 | 7487 submatch_line_lbr = reg_line_lbr; |
7 | 7488 save_reg_win = reg_win; |
7489 save_ireg_ic = ireg_ic; | |
7490 can_f_submatch = TRUE; | |
7491 | |
714 | 7492 eval_result = eval_to_string(source + 2, NULL, TRUE); |
7 | 7493 if (eval_result != NULL) |
7494 { | |
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7495 int had_backslash = FALSE; |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7496 |
39 | 7497 for (s = eval_result; *s != NUL; mb_ptr_adv(s)) |
7 | 7498 { |
2904 | 7499 /* Change NL to CR, so that it becomes a line break, |
7500 * unless called from vim_regexec_nl(). | |
7 | 7501 * Skip over a backslashed character. */ |
2904 | 7502 if (*s == NL && !submatch_line_lbr) |
7 | 7503 *s = CAR; |
7504 else if (*s == '\\' && s[1] != NUL) | |
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7505 { |
7 | 7506 ++s; |
2173 | 7507 /* Change NL to CR here too, so that this works: |
7508 * :s/abc\\\ndef/\="aaa\\\nbbb"/ on text: | |
7509 * abc\ | |
7510 * def | |
2904 | 7511 * Not when called from vim_regexec_nl(). |
2173 | 7512 */ |
2904 | 7513 if (*s == NL && !submatch_line_lbr) |
2173 | 7514 *s = CAR; |
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7515 had_backslash = TRUE; |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7516 } |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7517 } |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7518 if (had_backslash && backslash) |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7519 { |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7520 /* Backslashes will be consumed, need to double them. */ |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7521 s = vim_strsave_escaped(eval_result, (char_u *)"\\"); |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7522 if (s != NULL) |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7523 { |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7524 vim_free(eval_result); |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7525 eval_result = s; |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7526 } |
7 | 7527 } |
7528 | |
7529 dst += STRLEN(eval_result); | |
7530 } | |
7531 | |
7532 reg_match = submatch_match; | |
7533 reg_mmatch = submatch_mmatch; | |
2011 | 7534 reg_firstlnum = submatch_firstlnum; |
7535 reg_maxline = submatch_maxline; | |
2904 | 7536 reg_line_lbr = submatch_line_lbr; |
7 | 7537 reg_win = save_reg_win; |
7538 ireg_ic = save_ireg_ic; | |
7539 can_f_submatch = FALSE; | |
7540 } | |
7541 #endif | |
7542 } | |
7543 else | |
7544 while ((c = *src++) != NUL) | |
7545 { | |
7546 if (c == '&' && magic) | |
7547 no = 0; | |
7548 else if (c == '\\' && *src != NUL) | |
7549 { | |
7550 if (*src == '&' && !magic) | |
7551 { | |
7552 ++src; | |
7553 no = 0; | |
7554 } | |
7555 else if ('0' <= *src && *src <= '9') | |
7556 { | |
7557 no = *src++ - '0'; | |
7558 } | |
7559 else if (vim_strchr((char_u *)"uUlLeE", *src)) | |
7560 { | |
7561 switch (*src++) | |
7562 { | |
4244 | 7563 case 'u': func_one = (fptr_T)do_upper; |
7 | 7564 continue; |
4244 | 7565 case 'U': func_all = (fptr_T)do_Upper; |
7 | 7566 continue; |
4244 | 7567 case 'l': func_one = (fptr_T)do_lower; |
7 | 7568 continue; |
4244 | 7569 case 'L': func_all = (fptr_T)do_Lower; |
7 | 7570 continue; |
7571 case 'e': | |
4244 | 7572 case 'E': func_one = func_all = (fptr_T)NULL; |
7 | 7573 continue; |
7574 } | |
7575 } | |
7576 } | |
7577 if (no < 0) /* Ordinary character. */ | |
7578 { | |
798 | 7579 if (c == K_SPECIAL && src[0] != NUL && src[1] != NUL) |
7580 { | |
1209 | 7581 /* Copy a special key as-is. */ |
798 | 7582 if (copy) |
7583 { | |
7584 *dst++ = c; | |
7585 *dst++ = *src++; | |
7586 *dst++ = *src++; | |
7587 } | |
7588 else | |
7589 { | |
7590 dst += 3; | |
7591 src += 2; | |
7592 } | |
7593 continue; | |
7594 } | |
7595 | |
7 | 7596 if (c == '\\' && *src != NUL) |
7597 { | |
7598 /* Check for abbreviations -- webb */ | |
7599 switch (*src) | |
7600 { | |
7601 case 'r': c = CAR; ++src; break; | |
7602 case 'n': c = NL; ++src; break; | |
7603 case 't': c = TAB; ++src; break; | |
7604 /* Oh no! \e already has meaning in subst pat :-( */ | |
7605 /* case 'e': c = ESC; ++src; break; */ | |
7606 case 'b': c = Ctrl_H; ++src; break; | |
7607 | |
7608 /* If "backslash" is TRUE the backslash will be removed | |
7609 * later. Used to insert a literal CR. */ | |
7610 default: if (backslash) | |
7611 { | |
7612 if (copy) | |
7613 *dst = '\\'; | |
7614 ++dst; | |
7615 } | |
7616 c = *src++; | |
7617 } | |
7618 } | |
798 | 7619 #ifdef FEAT_MBYTE |
7620 else if (has_mbyte) | |
7621 c = mb_ptr2char(src - 1); | |
7622 #endif | |
7 | 7623 |
7624 /* Write to buffer, if copy is set. */ | |
4244 | 7625 if (func_one != (fptr_T)NULL) |
7626 /* Turbo C complains without the typecast */ | |
7627 func_one = (fptr_T)(func_one(&cc, c)); | |
7628 else if (func_all != (fptr_T)NULL) | |
7629 /* Turbo C complains without the typecast */ | |
7630 func_all = (fptr_T)(func_all(&cc, c)); | |
7631 else /* just copy */ | |
772 | 7632 cc = c; |
7633 | |
7634 #ifdef FEAT_MBYTE | |
7635 if (has_mbyte) | |
7 | 7636 { |
2307
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7637 int totlen = mb_ptr2len(src - 1); |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7638 |
7 | 7639 if (copy) |
772 | 7640 mb_char2bytes(cc, dst); |
7641 dst += mb_char2len(cc) - 1; | |
2307
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7642 if (enc_utf8) |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7643 { |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7644 int clen = utf_ptr2len(src - 1); |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7645 |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7646 /* If the character length is shorter than "totlen", there |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7647 * are composing characters; copy them as-is. */ |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7648 if (clen < totlen) |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7649 { |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7650 if (copy) |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7651 mch_memmove(dst + 1, src - 1 + clen, |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7652 (size_t)(totlen - clen)); |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7653 dst += totlen - clen; |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7654 } |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7655 } |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7656 src += totlen - 1; |
7 | 7657 } |
7658 else | |
7659 #endif | |
7660 if (copy) | |
772 | 7661 *dst = cc; |
7 | 7662 dst++; |
7663 } | |
7664 else | |
7665 { | |
7666 if (REG_MULTI) | |
7667 { | |
7668 clnum = reg_mmatch->startpos[no].lnum; | |
7669 if (clnum < 0 || reg_mmatch->endpos[no].lnum < 0) | |
7670 s = NULL; | |
7671 else | |
7672 { | |
7673 s = reg_getline(clnum) + reg_mmatch->startpos[no].col; | |
7674 if (reg_mmatch->endpos[no].lnum == clnum) | |
7675 len = reg_mmatch->endpos[no].col | |
7676 - reg_mmatch->startpos[no].col; | |
7677 else | |
7678 len = (int)STRLEN(s); | |
7679 } | |
7680 } | |
7681 else | |
7682 { | |
7683 s = reg_match->startp[no]; | |
7684 if (reg_match->endp[no] == NULL) | |
7685 s = NULL; | |
7686 else | |
7687 len = (int)(reg_match->endp[no] - s); | |
7688 } | |
7689 if (s != NULL) | |
7690 { | |
7691 for (;;) | |
7692 { | |
7693 if (len == 0) | |
7694 { | |
7695 if (REG_MULTI) | |
7696 { | |
7697 if (reg_mmatch->endpos[no].lnum == clnum) | |
7698 break; | |
7699 if (copy) | |
7700 *dst = CAR; | |
7701 ++dst; | |
7702 s = reg_getline(++clnum); | |
7703 if (reg_mmatch->endpos[no].lnum == clnum) | |
7704 len = reg_mmatch->endpos[no].col; | |
7705 else | |
7706 len = (int)STRLEN(s); | |
7707 } | |
7708 else | |
7709 break; | |
7710 } | |
7711 else if (*s == NUL) /* we hit NUL. */ | |
7712 { | |
7713 if (copy) | |
7714 EMSG(_(e_re_damg)); | |
7715 goto exit; | |
7716 } | |
7717 else | |
7718 { | |
7719 if (backslash && (*s == CAR || *s == '\\')) | |
7720 { | |
7721 /* | |
7722 * Insert a backslash in front of a CR, otherwise | |
7723 * it will be replaced by a line break. | |
7724 * Number of backslashes will be halved later, | |
7725 * double them here. | |
7726 */ | |
7727 if (copy) | |
7728 { | |
7729 dst[0] = '\\'; | |
7730 dst[1] = *s; | |
7731 } | |
7732 dst += 2; | |
7733 } | |
7734 else | |
7735 { | |
772 | 7736 #ifdef FEAT_MBYTE |
7737 if (has_mbyte) | |
7738 c = mb_ptr2char(s); | |
7739 else | |
7740 #endif | |
7741 c = *s; | |
7742 | |
4244 | 7743 if (func_one != (fptr_T)NULL) |
7744 /* Turbo C complains without the typecast */ | |
7745 func_one = (fptr_T)(func_one(&cc, c)); | |
7746 else if (func_all != (fptr_T)NULL) | |
7747 /* Turbo C complains without the typecast */ | |
7748 func_all = (fptr_T)(func_all(&cc, c)); | |
7749 else /* just copy */ | |
772 | 7750 cc = c; |
7751 | |
7752 #ifdef FEAT_MBYTE | |
7753 if (has_mbyte) | |
7 | 7754 { |
1332 | 7755 int l; |
7756 | |
7757 /* Copy composing characters separately, one | |
7758 * at a time. */ | |
7759 if (enc_utf8) | |
7760 l = utf_ptr2len(s) - 1; | |
7761 else | |
7762 l = mb_ptr2len(s) - 1; | |
772 | 7763 |
7764 s += l; | |
7765 len -= l; | |
7766 if (copy) | |
7767 mb_char2bytes(cc, dst); | |
7768 dst += mb_char2len(cc) - 1; | |
7 | 7769 } |
772 | 7770 else |
7771 #endif | |
7772 if (copy) | |
7773 *dst = cc; | |
7774 dst++; | |
7 | 7775 } |
772 | 7776 |
7 | 7777 ++s; |
7778 --len; | |
7779 } | |
7780 } | |
7781 } | |
7782 no = -1; | |
7783 } | |
7784 } | |
7785 if (copy) | |
7786 *dst = NUL; | |
7787 | |
7788 exit: | |
7789 return (int)((dst - dest) + 1); | |
7790 } | |
7791 | |
7792 #ifdef FEAT_EVAL | |
2012 | 7793 static char_u *reg_getline_submatch __ARGS((linenr_T lnum)); |
7794 | |
7 | 7795 /* |
2011 | 7796 * Call reg_getline() with the line numbers from the submatch. If a |
7797 * substitute() was used the reg_maxline and other values have been | |
7798 * overwritten. | |
7799 */ | |
7800 static char_u * | |
7801 reg_getline_submatch(lnum) | |
7802 linenr_T lnum; | |
7803 { | |
7804 char_u *s; | |
7805 linenr_T save_first = reg_firstlnum; | |
7806 linenr_T save_max = reg_maxline; | |
7807 | |
7808 reg_firstlnum = submatch_firstlnum; | |
7809 reg_maxline = submatch_maxline; | |
7810 | |
7811 s = reg_getline(lnum); | |
7812 | |
7813 reg_firstlnum = save_first; | |
7814 reg_maxline = save_max; | |
7815 return s; | |
7816 } | |
7817 | |
7818 /* | |
1209 | 7819 * Used for the submatch() function: get the string from the n'th submatch in |
7 | 7820 * allocated memory. |
7821 * Returns NULL when not in a ":s" command and for a non-existing submatch. | |
7822 */ | |
7823 char_u * | |
7824 reg_submatch(no) | |
7825 int no; | |
7826 { | |
7827 char_u *retval = NULL; | |
7828 char_u *s; | |
7829 int len; | |
7830 int round; | |
7831 linenr_T lnum; | |
7832 | |
840 | 7833 if (!can_f_submatch || no < 0) |
7 | 7834 return NULL; |
7835 | |
7836 if (submatch_match == NULL) | |
7837 { | |
7838 /* | |
7839 * First round: compute the length and allocate memory. | |
7840 * Second round: copy the text. | |
7841 */ | |
7842 for (round = 1; round <= 2; ++round) | |
7843 { | |
7844 lnum = submatch_mmatch->startpos[no].lnum; | |
7845 if (lnum < 0 || submatch_mmatch->endpos[no].lnum < 0) | |
7846 return NULL; | |
7847 | |
2011 | 7848 s = reg_getline_submatch(lnum) + submatch_mmatch->startpos[no].col; |
7 | 7849 if (s == NULL) /* anti-crash check, cannot happen? */ |
7850 break; | |
7851 if (submatch_mmatch->endpos[no].lnum == lnum) | |
7852 { | |
7853 /* Within one line: take form start to end col. */ | |
7854 len = submatch_mmatch->endpos[no].col | |
7855 - submatch_mmatch->startpos[no].col; | |
7856 if (round == 2) | |
418 | 7857 vim_strncpy(retval, s, len); |
7 | 7858 ++len; |
7859 } | |
7860 else | |
7861 { | |
7862 /* Multiple lines: take start line from start col, middle | |
7863 * lines completely and end line up to end col. */ | |
7864 len = (int)STRLEN(s); | |
7865 if (round == 2) | |
7866 { | |
7867 STRCPY(retval, s); | |
7868 retval[len] = '\n'; | |
7869 } | |
7870 ++len; | |
7871 ++lnum; | |
7872 while (lnum < submatch_mmatch->endpos[no].lnum) | |
7873 { | |
2011 | 7874 s = reg_getline_submatch(lnum++); |
7 | 7875 if (round == 2) |
7876 STRCPY(retval + len, s); | |
7877 len += (int)STRLEN(s); | |
7878 if (round == 2) | |
7879 retval[len] = '\n'; | |
7880 ++len; | |
7881 } | |
7882 if (round == 2) | |
2011 | 7883 STRNCPY(retval + len, reg_getline_submatch(lnum), |
7 | 7884 submatch_mmatch->endpos[no].col); |
7885 len += submatch_mmatch->endpos[no].col; | |
7886 if (round == 2) | |
7887 retval[len] = NUL; | |
7888 ++len; | |
7889 } | |
7890 | |
840 | 7891 if (retval == NULL) |
7 | 7892 { |
7893 retval = lalloc((long_u)len, TRUE); | |
840 | 7894 if (retval == NULL) |
7 | 7895 return NULL; |
7896 } | |
7897 } | |
7898 } | |
7899 else | |
7900 { | |
1815 | 7901 s = submatch_match->startp[no]; |
7902 if (s == NULL || submatch_match->endp[no] == NULL) | |
7 | 7903 retval = NULL; |
7904 else | |
7905 retval = vim_strnsave(s, (int)(submatch_match->endp[no] - s)); | |
7906 } | |
7907 | |
7908 return retval; | |
7909 } | |
5794 | 7910 |
7911 /* | |
7912 * Used for the submatch() function with the optional non-zero argument: get | |
7913 * the list of strings from the n'th submatch in allocated memory with NULs | |
7914 * represented in NLs. | |
7915 * Returns a list of allocated strings. Returns NULL when not in a ":s" | |
7916 * command, for a non-existing submatch and for any error. | |
7917 */ | |
7918 list_T * | |
7919 reg_submatch_list(no) | |
7920 int no; | |
7921 { | |
7922 char_u *s; | |
7923 linenr_T slnum; | |
7924 linenr_T elnum; | |
7925 colnr_T scol; | |
7926 colnr_T ecol; | |
7927 int i; | |
7928 list_T *list; | |
7929 int error = FALSE; | |
7930 | |
7931 if (!can_f_submatch || no < 0) | |
7932 return NULL; | |
7933 | |
7934 if (submatch_match == NULL) | |
7935 { | |
7936 slnum = submatch_mmatch->startpos[no].lnum; | |
7937 elnum = submatch_mmatch->endpos[no].lnum; | |
7938 if (slnum < 0 || elnum < 0) | |
7939 return NULL; | |
7940 | |
7941 scol = submatch_mmatch->startpos[no].col; | |
7942 ecol = submatch_mmatch->endpos[no].col; | |
7943 | |
7944 list = list_alloc(); | |
7945 if (list == NULL) | |
7946 return NULL; | |
7947 | |
7948 s = reg_getline_submatch(slnum) + scol; | |
7949 if (slnum == elnum) | |
7950 { | |
7951 if (list_append_string(list, s, ecol - scol) == FAIL) | |
7952 error = TRUE; | |
7953 } | |
7954 else | |
7955 { | |
7956 if (list_append_string(list, s, -1) == FAIL) | |
7957 error = TRUE; | |
7958 for (i = 1; i < elnum - slnum; i++) | |
7959 { | |
7960 s = reg_getline_submatch(slnum + i); | |
7961 if (list_append_string(list, s, -1) == FAIL) | |
7962 error = TRUE; | |
7963 } | |
7964 s = reg_getline_submatch(elnum); | |
7965 if (list_append_string(list, s, ecol) == FAIL) | |
7966 error = TRUE; | |
7967 } | |
7968 } | |
7969 else | |
7970 { | |
7971 s = submatch_match->startp[no]; | |
7972 if (s == NULL || submatch_match->endp[no] == NULL) | |
7973 return NULL; | |
7974 list = list_alloc(); | |
7975 if (list == NULL) | |
7976 return NULL; | |
7977 if (list_append_string(list, s, | |
7978 (int)(submatch_match->endp[no] - s)) == FAIL) | |
7979 error = TRUE; | |
7980 } | |
7981 | |
7982 if (error) | |
7983 { | |
7984 list_free(list, TRUE); | |
7985 return NULL; | |
7986 } | |
7987 return list; | |
7988 } | |
7 | 7989 #endif |
4444 | 7990 |
7991 static regengine_T bt_regengine = | |
7992 { | |
7993 bt_regcomp, | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7994 bt_regfree, |
4444 | 7995 bt_regexec_nl, |
7996 bt_regexec_multi | |
7997 #ifdef DEBUG | |
7998 ,(char_u *)"" | |
7999 #endif | |
8000 }; | |
8001 | |
8002 | |
8003 #include "regexp_nfa.c" | |
8004 | |
8005 static regengine_T nfa_regengine = | |
8006 { | |
8007 nfa_regcomp, | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8008 nfa_regfree, |
4444 | 8009 nfa_regexec_nl, |
8010 nfa_regexec_multi | |
8011 #ifdef DEBUG | |
8012 ,(char_u *)"" | |
8013 #endif | |
8014 }; | |
8015 | |
8016 /* Which regexp engine to use? Needed for vim_regcomp(). | |
8017 * Must match with 'regexpengine'. */ | |
8018 static int regexp_engine = 0; | |
8019 #define AUTOMATIC_ENGINE 0 | |
8020 #define BACKTRACKING_ENGINE 1 | |
8021 #define NFA_ENGINE 2 | |
8022 #ifdef DEBUG | |
8023 static char_u regname[][30] = { | |
8024 "AUTOMATIC Regexp Engine", | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
8025 "BACKTRACKING Regexp Engine", |
4444 | 8026 "NFA Regexp Engine" |
8027 }; | |
8028 #endif | |
8029 | |
8030 /* | |
8031 * Compile a regular expression into internal code. | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8032 * Returns the program in allocated memory. |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8033 * Use vim_regfree() to free the memory. |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8034 * Returns NULL for an error. |
4444 | 8035 */ |
8036 regprog_T * | |
8037 vim_regcomp(expr_arg, re_flags) | |
8038 char_u *expr_arg; | |
8039 int re_flags; | |
8040 { | |
8041 regprog_T *prog = NULL; | |
8042 char_u *expr = expr_arg; | |
8043 | |
8044 regexp_engine = p_re; | |
8045 | |
8046 /* Check for prefix "\%#=", that sets the regexp engine */ | |
8047 if (STRNCMP(expr, "\\%#=", 4) == 0) | |
8048 { | |
8049 int newengine = expr[4] - '0'; | |
8050 | |
8051 if (newengine == AUTOMATIC_ENGINE | |
8052 || newengine == BACKTRACKING_ENGINE | |
8053 || newengine == NFA_ENGINE) | |
8054 { | |
8055 regexp_engine = expr[4] - '0'; | |
8056 expr += 5; | |
8057 #ifdef DEBUG | |
5897 | 8058 smsg((char_u *)"New regexp mode selected (%d): %s", |
8059 regexp_engine, regname[newengine]); | |
4444 | 8060 #endif |
8061 } | |
8062 else | |
8063 { | |
8064 EMSG(_("E864: \\%#= can only be followed by 0, 1, or 2. The automatic engine will be used ")); | |
8065 regexp_engine = AUTOMATIC_ENGINE; | |
8066 } | |
8067 } | |
8068 #ifdef DEBUG | |
8069 bt_regengine.expr = expr; | |
8070 nfa_regengine.expr = expr; | |
8071 #endif | |
8072 | |
8073 /* | |
8074 * First try the NFA engine, unless backtracking was requested. | |
8075 */ | |
8076 if (regexp_engine != BACKTRACKING_ENGINE) | |
8077 prog = nfa_regengine.regcomp(expr, re_flags); | |
8078 else | |
8079 prog = bt_regengine.regcomp(expr, re_flags); | |
8080 | |
8081 if (prog == NULL) /* error compiling regexp with initial engine */ | |
8082 { | |
4460 | 8083 #ifdef BT_REGEXP_DEBUG_LOG |
4444 | 8084 if (regexp_engine != BACKTRACKING_ENGINE) /* debugging log for NFA */ |
8085 { | |
8086 FILE *f; | |
4460 | 8087 f = fopen(BT_REGEXP_DEBUG_LOG_NAME, "a"); |
4444 | 8088 if (f) |
8089 { | |
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8090 fprintf(f, "Syntax error in \"%s\"\n", expr); |
4444 | 8091 fclose(f); |
8092 } | |
8093 else | |
4460 | 8094 EMSG2("(NFA) Could not open \"%s\" to write !!!", |
8095 BT_REGEXP_DEBUG_LOG_NAME); | |
4444 | 8096 } |
8097 #endif | |
8098 /* | |
5353 | 8099 * If the NFA engine failed, the backtracking engine won't work either. |
8100 * | |
4444 | 8101 if (regexp_engine == AUTOMATIC_ENGINE) |
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8102 prog = bt_regengine.regcomp(expr, re_flags); |
5353 | 8103 */ |
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8104 } |
4444 | 8105 |
8106 return prog; | |
8107 } | |
8108 | |
8109 /* | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8110 * Free a compiled regexp program, returned by vim_regcomp(). |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8111 */ |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8112 void |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8113 vim_regfree(prog) |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8114 regprog_T *prog; |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8115 { |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8116 if (prog != NULL) |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8117 prog->engine->regfree(prog); |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8118 } |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8119 |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8120 /* |
4444 | 8121 * Match a regexp against a string. |
8122 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
8123 * Uses curbuf for line count and 'iskeyword'. | |
8124 * | |
8125 * Return TRUE if there is a match, FALSE if not. | |
8126 */ | |
8127 int | |
8128 vim_regexec(rmp, line, col) | |
8129 regmatch_T *rmp; | |
8130 char_u *line; /* string to match against */ | |
8131 colnr_T col; /* column to start looking for match */ | |
8132 { | |
5838 | 8133 return rmp->regprog->engine->regexec_nl(rmp, line, col, FALSE); |
4444 | 8134 } |
8135 | |
8136 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
8137 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
8138 /* | |
8139 * Like vim_regexec(), but consider a "\n" in "line" to be a line break. | |
8140 */ | |
8141 int | |
8142 vim_regexec_nl(rmp, line, col) | |
8143 regmatch_T *rmp; | |
8144 char_u *line; | |
8145 colnr_T col; | |
8146 { | |
5838 | 8147 return rmp->regprog->engine->regexec_nl(rmp, line, col, TRUE); |
4444 | 8148 } |
8149 #endif | |
8150 | |
8151 /* | |
8152 * Match a regexp against multiple lines. | |
8153 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
8154 * Uses curbuf for line count and 'iskeyword'. | |
8155 * | |
8156 * Return zero if there is no match. Return number of lines contained in the | |
8157 * match otherwise. | |
8158 */ | |
8159 long | |
8160 vim_regexec_multi(rmp, win, buf, lnum, col, tm) | |
8161 regmmatch_T *rmp; | |
8162 win_T *win; /* window in which to search or NULL */ | |
8163 buf_T *buf; /* buffer in which to search */ | |
8164 linenr_T lnum; /* nr of line to start looking for match */ | |
8165 colnr_T col; /* column to start looking for match */ | |
8166 proftime_T *tm; /* timeout limit or NULL */ | |
8167 { | |
8168 return rmp->regprog->engine->regexec_multi(rmp, win, buf, lnum, col, tm); | |
8169 } |