Mercurial > vim
annotate src/regexp.c @ 5893:99374096a76b v7.4.289
updated for version 7.4.289
Problem: Pattern with repeated backreference does not match with new regexp
engine. (Urtica Dioica)
Solution: Also check the end of a submatch when deciding to put a state in
the state list.
author | Bram Moolenaar <bram@vim.org> |
---|---|
date | Tue, 13 May 2014 15:56:51 +0200 |
parents | 0ea551fa607d |
children | b5972833add9 |
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 */ | |
247 | |
7 | 248 /* |
249 * Magic characters have a special meaning, they don't match literally. | |
250 * Magic characters are negative. This separates them from literal characters | |
251 * (possibly multi-byte). Only ASCII characters can be Magic. | |
252 */ | |
253 #define Magic(x) ((int)(x) - 256) | |
254 #define un_Magic(x) ((x) + 256) | |
255 #define is_Magic(x) ((x) < 0) | |
256 | |
257 static int no_Magic __ARGS((int x)); | |
258 static int toggle_Magic __ARGS((int x)); | |
259 | |
260 static int | |
261 no_Magic(x) | |
262 int x; | |
263 { | |
264 if (is_Magic(x)) | |
265 return un_Magic(x); | |
266 return x; | |
267 } | |
268 | |
269 static int | |
270 toggle_Magic(x) | |
271 int x; | |
272 { | |
273 if (is_Magic(x)) | |
274 return un_Magic(x); | |
275 return Magic(x); | |
276 } | |
277 | |
278 /* | |
279 * The first byte of the regexp internal "program" is actually this magic | |
280 * number; the start node begins in the second byte. It's used to catch the | |
281 * most severe mutilation of the program by the caller. | |
282 */ | |
283 | |
284 #define REGMAGIC 0234 | |
285 | |
286 /* | |
287 * Opcode notes: | |
288 * | |
289 * BRANCH The set of branches constituting a single choice are hooked | |
290 * together with their "next" pointers, since precedence prevents | |
291 * anything being concatenated to any individual branch. The | |
292 * "next" pointer of the last BRANCH in a choice points to the | |
293 * thing following the whole choice. This is also where the | |
294 * final "next" pointer of each individual branch points; each | |
295 * branch starts with the operand node of a BRANCH node. | |
296 * | |
297 * BACK Normal "next" pointers all implicitly point forward; BACK | |
298 * exists to make loop structures possible. | |
299 * | |
300 * STAR,PLUS '=', and complex '*' and '+', are implemented as circular | |
301 * BRANCH structures using BACK. Simple cases (one character | |
302 * per match) are implemented with STAR and PLUS for speed | |
303 * and to minimize recursive plunges. | |
304 * | |
305 * BRACE_LIMITS This is always followed by a BRACE_SIMPLE or BRACE_COMPLEX | |
306 * node, and defines the min and max limits to be used for that | |
307 * node. | |
308 * | |
309 * MOPEN,MCLOSE ...are numbered at compile time. | |
310 * ZOPEN,ZCLOSE ...ditto | |
311 */ | |
312 | |
313 /* | |
314 * A node is one char of opcode followed by two chars of "next" pointer. | |
315 * "Next" pointers are stored as two 8-bit bytes, high order first. The | |
316 * value is a positive offset from the opcode of the node containing it. | |
317 * An operand, if any, simply follows the node. (Note that much of the | |
318 * code generation knows about this implicit relationship.) | |
319 * | |
320 * Using two bytes for the "next" pointer is vast overkill for most things, | |
321 * but allows patterns to get big without disasters. | |
322 */ | |
323 #define OP(p) ((int)*(p)) | |
324 #define NEXT(p) (((*((p) + 1) & 0377) << 8) + (*((p) + 2) & 0377)) | |
325 #define OPERAND(p) ((p) + 3) | |
326 /* Obtain an operand that was stored as four bytes, MSB first. */ | |
327 #define OPERAND_MIN(p) (((long)(p)[3] << 24) + ((long)(p)[4] << 16) \ | |
328 + ((long)(p)[5] << 8) + (long)(p)[6]) | |
329 /* Obtain a second operand stored as four bytes. */ | |
330 #define OPERAND_MAX(p) OPERAND_MIN((p) + 4) | |
331 /* Obtain a second single-byte operand stored after a four bytes operand. */ | |
332 #define OPERAND_CMP(p) (p)[7] | |
333 | |
334 /* | |
335 * Utility definitions. | |
336 */ | |
337 #define UCHARAT(p) ((int)*(char_u *)(p)) | |
338 | |
339 /* Used for an error (down from) vim_regcomp(): give the error message, set | |
340 * rc_did_emsg and return NULL */ | |
653 | 341 #define EMSG_RET_NULL(m) return (EMSG(m), rc_did_emsg = TRUE, (void *)NULL) |
308 | 342 #define EMSG_RET_FAIL(m) return (EMSG(m), rc_did_emsg = TRUE, FAIL) |
4444 | 343 #define EMSG2_RET_NULL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, (void *)NULL) |
344 #define EMSG2_RET_FAIL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, FAIL) | |
345 #define EMSG_ONE_RET_NULL EMSG2_RET_NULL(_("E369: invalid item in %s%%[]"), reg_magic == MAGIC_ALL) | |
7 | 346 |
347 #define MAX_LIMIT (32767L << 16L) | |
348 | |
349 static int re_multi_type __ARGS((int)); | |
350 static int cstrncmp __ARGS((char_u *s1, char_u *s2, int *n)); | |
351 static char_u *cstrchr __ARGS((char_u *, int)); | |
352 | |
4444 | 353 #ifdef BT_REGEXP_DUMP |
354 static void regdump __ARGS((char_u *, bt_regprog_T *)); | |
355 #endif | |
7 | 356 #ifdef DEBUG |
357 static char_u *regprop __ARGS((char_u *)); | |
358 #endif | |
359 | |
4444 | 360 static char_u e_missingbracket[] = N_("E769: Missing ] after %s["); |
361 static char_u e_unmatchedpp[] = N_("E53: Unmatched %s%%("); | |
362 static char_u e_unmatchedp[] = N_("E54: Unmatched %s("); | |
363 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
|
364 #ifdef FEAT_SYN_HL |
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
365 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
|
366 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
|
367 #endif |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
368 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
|
369 static char_u e_empty_sb[] = N_("E70: Empty %s%%[]"); |
7 | 370 #define NOT_MULTI 0 |
371 #define MULTI_ONE 1 | |
372 #define MULTI_MULT 2 | |
373 /* | |
374 * Return NOT_MULTI if c is not a "multi" operator. | |
375 * Return MULTI_ONE if c is a single "multi" operator. | |
376 * Return MULTI_MULT if c is a multi "multi" operator. | |
377 */ | |
378 static int | |
379 re_multi_type(c) | |
380 int c; | |
381 { | |
382 if (c == Magic('@') || c == Magic('=') || c == Magic('?')) | |
383 return MULTI_ONE; | |
384 if (c == Magic('*') || c == Magic('+') || c == Magic('{')) | |
385 return MULTI_MULT; | |
386 return NOT_MULTI; | |
387 } | |
388 | |
389 /* | |
390 * Flags to be passed up and down. | |
391 */ | |
392 #define HASWIDTH 0x1 /* Known never to match null string. */ | |
393 #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ | |
394 #define SPSTART 0x4 /* Starts with * or +. */ | |
395 #define HASNL 0x8 /* Contains some \n. */ | |
396 #define HASLOOKBH 0x10 /* Contains "\@<=" or "\@<!". */ | |
397 #define WORST 0 /* Worst case. */ | |
398 | |
399 /* | |
400 * When regcode is set to this value, code is not emitted and size is computed | |
401 * instead. | |
402 */ | |
403 #define JUST_CALC_SIZE ((char_u *) -1) | |
404 | |
359 | 405 static char_u *reg_prev_sub = NULL; |
406 | |
7 | 407 /* |
408 * REGEXP_INRANGE contains all characters which are always special in a [] | |
409 * range after '\'. | |
410 * REGEXP_ABBR contains all characters which act as abbreviations after '\'. | |
411 * These are: | |
412 * \n - New line (NL). | |
413 * \r - Carriage Return (CR). | |
414 * \t - Tab (TAB). | |
415 * \e - Escape (ESC). | |
416 * \b - Backspace (Ctrl_H). | |
24 | 417 * \d - Character code in decimal, eg \d123 |
418 * \o - Character code in octal, eg \o80 | |
419 * \x - Character code in hex, eg \x4a | |
420 * \u - Multibyte character code, eg \u20ac | |
421 * \U - Long multibyte character code, eg \U12345678 | |
7 | 422 */ |
423 static char_u REGEXP_INRANGE[] = "]^-n\\"; | |
24 | 424 static char_u REGEXP_ABBR[] = "nrtebdoxuU"; |
7 | 425 |
426 static int backslash_trans __ARGS((int c)); | |
167 | 427 static int get_char_class __ARGS((char_u **pp)); |
428 static int get_equi_class __ARGS((char_u **pp)); | |
429 static void reg_equi_class __ARGS((int c)); | |
430 static int get_coll_element __ARGS((char_u **pp)); | |
7 | 431 static char_u *skip_anyof __ARGS((char_u *p)); |
432 static void init_class_tab __ARGS((void)); | |
433 | |
434 /* | |
435 * Translate '\x' to its control character, except "\n", which is Magic. | |
436 */ | |
437 static int | |
438 backslash_trans(c) | |
439 int c; | |
440 { | |
441 switch (c) | |
442 { | |
443 case 'r': return CAR; | |
444 case 't': return TAB; | |
445 case 'e': return ESC; | |
446 case 'b': return BS; | |
447 } | |
448 return c; | |
449 } | |
450 | |
451 /* | |
167 | 452 * Check for a character class name "[:name:]". "pp" points to the '['. |
7 | 453 * Returns one of the CLASS_ items. CLASS_NONE means that no item was |
454 * recognized. Otherwise "pp" is advanced to after the item. | |
455 */ | |
456 static int | |
167 | 457 get_char_class(pp) |
7 | 458 char_u **pp; |
459 { | |
460 static const char *(class_names[]) = | |
461 { | |
462 "alnum:]", | |
463 #define CLASS_ALNUM 0 | |
464 "alpha:]", | |
465 #define CLASS_ALPHA 1 | |
466 "blank:]", | |
467 #define CLASS_BLANK 2 | |
468 "cntrl:]", | |
469 #define CLASS_CNTRL 3 | |
470 "digit:]", | |
471 #define CLASS_DIGIT 4 | |
472 "graph:]", | |
473 #define CLASS_GRAPH 5 | |
474 "lower:]", | |
475 #define CLASS_LOWER 6 | |
476 "print:]", | |
477 #define CLASS_PRINT 7 | |
478 "punct:]", | |
479 #define CLASS_PUNCT 8 | |
480 "space:]", | |
481 #define CLASS_SPACE 9 | |
482 "upper:]", | |
483 #define CLASS_UPPER 10 | |
484 "xdigit:]", | |
485 #define CLASS_XDIGIT 11 | |
486 "tab:]", | |
487 #define CLASS_TAB 12 | |
488 "return:]", | |
489 #define CLASS_RETURN 13 | |
490 "backspace:]", | |
491 #define CLASS_BACKSPACE 14 | |
492 "escape:]", | |
493 #define CLASS_ESCAPE 15 | |
494 }; | |
495 #define CLASS_NONE 99 | |
496 int i; | |
497 | |
498 if ((*pp)[1] == ':') | |
499 { | |
1877 | 500 for (i = 0; i < (int)(sizeof(class_names) / sizeof(*class_names)); ++i) |
7 | 501 if (STRNCMP(*pp + 2, class_names[i], STRLEN(class_names[i])) == 0) |
502 { | |
503 *pp += STRLEN(class_names[i]) + 2; | |
504 return i; | |
505 } | |
506 } | |
507 return CLASS_NONE; | |
508 } | |
509 | |
510 /* | |
511 * Specific version of character class functions. | |
512 * Using a table to keep this fast. | |
513 */ | |
514 static short class_tab[256]; | |
515 | |
516 #define RI_DIGIT 0x01 | |
517 #define RI_HEX 0x02 | |
518 #define RI_OCTAL 0x04 | |
519 #define RI_WORD 0x08 | |
520 #define RI_HEAD 0x10 | |
521 #define RI_ALPHA 0x20 | |
522 #define RI_LOWER 0x40 | |
523 #define RI_UPPER 0x80 | |
524 #define RI_WHITE 0x100 | |
525 | |
526 static void | |
527 init_class_tab() | |
528 { | |
529 int i; | |
530 static int done = FALSE; | |
531 | |
532 if (done) | |
533 return; | |
534 | |
535 for (i = 0; i < 256; ++i) | |
536 { | |
537 if (i >= '0' && i <= '7') | |
538 class_tab[i] = RI_DIGIT + RI_HEX + RI_OCTAL + RI_WORD; | |
539 else if (i >= '8' && i <= '9') | |
540 class_tab[i] = RI_DIGIT + RI_HEX + RI_WORD; | |
541 else if (i >= 'a' && i <= 'f') | |
542 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
543 #ifdef EBCDIC | |
544 else if ((i >= 'g' && i <= 'i') || (i >= 'j' && i <= 'r') | |
545 || (i >= 's' && i <= 'z')) | |
546 #else | |
547 else if (i >= 'g' && i <= 'z') | |
548 #endif | |
549 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
550 else if (i >= 'A' && i <= 'F') | |
551 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
552 #ifdef EBCDIC | |
553 else if ((i >= 'G' && i <= 'I') || ( i >= 'J' && i <= 'R') | |
554 || (i >= 'S' && i <= 'Z')) | |
555 #else | |
556 else if (i >= 'G' && i <= 'Z') | |
557 #endif | |
558 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
559 else if (i == '_') | |
560 class_tab[i] = RI_WORD + RI_HEAD; | |
561 else | |
562 class_tab[i] = 0; | |
563 } | |
564 class_tab[' '] |= RI_WHITE; | |
565 class_tab['\t'] |= RI_WHITE; | |
566 done = TRUE; | |
567 } | |
568 | |
569 #ifdef FEAT_MBYTE | |
570 # define ri_digit(c) (c < 0x100 && (class_tab[c] & RI_DIGIT)) | |
571 # define ri_hex(c) (c < 0x100 && (class_tab[c] & RI_HEX)) | |
572 # define ri_octal(c) (c < 0x100 && (class_tab[c] & RI_OCTAL)) | |
573 # define ri_word(c) (c < 0x100 && (class_tab[c] & RI_WORD)) | |
574 # define ri_head(c) (c < 0x100 && (class_tab[c] & RI_HEAD)) | |
575 # define ri_alpha(c) (c < 0x100 && (class_tab[c] & RI_ALPHA)) | |
576 # define ri_lower(c) (c < 0x100 && (class_tab[c] & RI_LOWER)) | |
577 # define ri_upper(c) (c < 0x100 && (class_tab[c] & RI_UPPER)) | |
578 # define ri_white(c) (c < 0x100 && (class_tab[c] & RI_WHITE)) | |
579 #else | |
580 # define ri_digit(c) (class_tab[c] & RI_DIGIT) | |
581 # define ri_hex(c) (class_tab[c] & RI_HEX) | |
582 # define ri_octal(c) (class_tab[c] & RI_OCTAL) | |
583 # define ri_word(c) (class_tab[c] & RI_WORD) | |
584 # define ri_head(c) (class_tab[c] & RI_HEAD) | |
585 # define ri_alpha(c) (class_tab[c] & RI_ALPHA) | |
586 # define ri_lower(c) (class_tab[c] & RI_LOWER) | |
587 # define ri_upper(c) (class_tab[c] & RI_UPPER) | |
588 # define ri_white(c) (class_tab[c] & RI_WHITE) | |
589 #endif | |
590 | |
591 /* flags for regflags */ | |
592 #define RF_ICASE 1 /* ignore case */ | |
593 #define RF_NOICASE 2 /* don't ignore case */ | |
594 #define RF_HASNL 4 /* can match a NL */ | |
595 #define RF_ICOMBINE 8 /* ignore combining characters */ | |
596 #define RF_LOOKBH 16 /* uses "\@<=" or "\@<!" */ | |
597 | |
598 /* | |
599 * Global work variables for vim_regcomp(). | |
600 */ | |
601 | |
602 static char_u *regparse; /* Input-scan pointer. */ | |
603 static int prevchr_len; /* byte length of previous char */ | |
604 static int num_complex_braces; /* Complex \{...} count */ | |
605 static int regnpar; /* () count. */ | |
606 #ifdef FEAT_SYN_HL | |
607 static int regnzpar; /* \z() count. */ | |
608 static int re_has_z; /* \z item detected */ | |
609 #endif | |
610 static char_u *regcode; /* Code-emit pointer, or JUST_CALC_SIZE */ | |
611 static long regsize; /* Code size. */ | |
2010 | 612 static int reg_toolong; /* TRUE when offset out of range */ |
7 | 613 static char_u had_endbrace[NSUBEXP]; /* flags, TRUE if end of () found */ |
614 static unsigned regflags; /* RF_ flags for prog */ | |
615 static long brace_min[10]; /* Minimums for complex brace repeats */ | |
616 static long brace_max[10]; /* Maximums for complex brace repeats */ | |
617 static int brace_count[10]; /* Current counts for complex brace repeats */ | |
618 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
619 static int had_eol; /* TRUE when EOL found by vim_regcomp() */ | |
620 #endif | |
621 static int one_exactly = FALSE; /* only do one char for EXACTLY */ | |
622 | |
623 static int reg_magic; /* magicness of the pattern: */ | |
624 #define MAGIC_NONE 1 /* "\V" very unmagic */ | |
625 #define MAGIC_OFF 2 /* "\M" or 'magic' off */ | |
626 #define MAGIC_ON 3 /* "\m" or 'magic' */ | |
627 #define MAGIC_ALL 4 /* "\v" very magic */ | |
628 | |
629 static int reg_string; /* matching with a string instead of a buffer | |
630 line */ | |
481 | 631 static int reg_strict; /* "[abc" is illegal */ |
7 | 632 |
633 /* | |
634 * META contains all characters that may be magic, except '^' and '$'. | |
635 */ | |
636 | |
637 #ifdef EBCDIC | |
638 static char_u META[] = "%&()*+.123456789<=>?@ACDFHIKLMOPSUVWX[_acdfhiklmnopsuvwxz{|~"; | |
639 #else | |
640 /* META[] is used often enough to justify turning it into a table. */ | |
641 static char_u META_flags[] = { | |
642 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
643 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
644 /* % & ( ) * + . */ | |
645 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0, | |
646 /* 1 2 3 4 5 6 7 8 9 < = > ? */ | |
647 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, | |
648 /* @ A C D F H I K L M O */ | |
649 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1, | |
650 /* P S U V W X Z [ _ */ | |
651 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, | |
652 /* a c d f h i k l m n o */ | |
653 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, | |
654 /* p s u v w x z { | ~ */ | |
655 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1 | |
656 }; | |
657 #endif | |
658 | |
4444 | 659 static int curchr; /* currently parsed character */ |
660 /* Previous character. Note: prevchr is sometimes -1 when we are not at the | |
661 * start, eg in /[ ^I]^ the pattern was never found even if it existed, | |
662 * because ^ was taken to be magic -- webb */ | |
663 static int prevchr; | |
664 static int prevprevchr; /* previous-previous character */ | |
665 static int nextchr; /* used for ungetchr() */ | |
7 | 666 |
667 /* arguments for reg() */ | |
668 #define REG_NOPAREN 0 /* toplevel reg() */ | |
669 #define REG_PAREN 1 /* \(\) */ | |
670 #define REG_ZPAREN 2 /* \z(\) */ | |
671 #define REG_NPAREN 3 /* \%(\) */ | |
672 | |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
673 typedef struct |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
674 { |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
675 char_u *regparse; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
676 int prevchr_len; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
677 int curchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
678 int prevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
679 int prevprevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
680 int nextchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
681 int at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
682 int prev_at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
683 int regnpar; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
684 } parse_state_T; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
685 |
7 | 686 /* |
687 * Forward declarations for vim_regcomp()'s friends. | |
688 */ | |
689 static void initchr __ARGS((char_u *)); | |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
690 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
|
691 static void restore_parse_state __ARGS((parse_state_T *ps)); |
7 | 692 static int getchr __ARGS((void)); |
693 static void skipchr_keepstart __ARGS((void)); | |
694 static int peekchr __ARGS((void)); | |
695 static void skipchr __ARGS((void)); | |
696 static void ungetchr __ARGS((void)); | |
24 | 697 static int gethexchrs __ARGS((int maxinputlen)); |
698 static int getoctchrs __ARGS((void)); | |
699 static int getdecchrs __ARGS((void)); | |
700 static int coll_get_char __ARGS((void)); | |
7 | 701 static void regcomp_start __ARGS((char_u *expr, int flags)); |
702 static char_u *reg __ARGS((int, int *)); | |
703 static char_u *regbranch __ARGS((int *flagp)); | |
704 static char_u *regconcat __ARGS((int *flagp)); | |
705 static char_u *regpiece __ARGS((int *)); | |
706 static char_u *regatom __ARGS((int *)); | |
707 static char_u *regnode __ARGS((int)); | |
714 | 708 #ifdef FEAT_MBYTE |
709 static int use_multibytecode __ARGS((int c)); | |
710 #endif | |
7 | 711 static int prog_magic_wrong __ARGS((void)); |
712 static char_u *regnext __ARGS((char_u *)); | |
713 static void regc __ARGS((int b)); | |
714 #ifdef FEAT_MBYTE | |
715 static void regmbc __ARGS((int c)); | |
2974 | 716 # define REGMBC(x) regmbc(x); |
717 # define CASEMBC(x) case x: | |
167 | 718 #else |
719 # define regmbc(c) regc(c) | |
2974 | 720 # define REGMBC(x) |
721 # define CASEMBC(x) | |
7 | 722 #endif |
723 static void reginsert __ARGS((int, char_u *)); | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
724 static void reginsert_nr __ARGS((int op, long val, char_u *opnd)); |
7 | 725 static void reginsert_limits __ARGS((int, long, long, char_u *)); |
726 static char_u *re_put_long __ARGS((char_u *pr, long_u val)); | |
727 static int read_limits __ARGS((long *, long *)); | |
728 static void regtail __ARGS((char_u *, char_u *)); | |
729 static void regoptail __ARGS((char_u *, char_u *)); | |
730 | |
4444 | 731 static regengine_T bt_regengine; |
732 static regengine_T nfa_regengine; | |
733 | |
7 | 734 /* |
735 * Return TRUE if compiled regular expression "prog" can match a line break. | |
736 */ | |
737 int | |
738 re_multiline(prog) | |
739 regprog_T *prog; | |
740 { | |
741 return (prog->regflags & RF_HASNL); | |
742 } | |
743 | |
744 /* | |
745 * Return TRUE if compiled regular expression "prog" looks before the start | |
746 * position (pattern contains "\@<=" or "\@<!"). | |
747 */ | |
748 int | |
749 re_lookbehind(prog) | |
750 regprog_T *prog; | |
751 { | |
752 return (prog->regflags & RF_LOOKBH); | |
753 } | |
754 | |
755 /* | |
167 | 756 * Check for an equivalence class name "[=a=]". "pp" points to the '['. |
757 * Returns a character representing the class. Zero means that no item was | |
758 * recognized. Otherwise "pp" is advanced to after the item. | |
759 */ | |
760 static int | |
761 get_equi_class(pp) | |
762 char_u **pp; | |
763 { | |
764 int c; | |
765 int l = 1; | |
766 char_u *p = *pp; | |
767 | |
768 if (p[1] == '=') | |
769 { | |
770 #ifdef FEAT_MBYTE | |
771 if (has_mbyte) | |
474 | 772 l = (*mb_ptr2len)(p + 2); |
167 | 773 #endif |
774 if (p[l + 2] == '=' && p[l + 3] == ']') | |
775 { | |
776 #ifdef FEAT_MBYTE | |
777 if (has_mbyte) | |
778 c = mb_ptr2char(p + 2); | |
779 else | |
780 #endif | |
781 c = p[2]; | |
782 *pp += l + 4; | |
783 return c; | |
784 } | |
785 } | |
786 return 0; | |
787 } | |
788 | |
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
789 #ifdef EBCDIC |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
790 /* |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
791 * 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
|
792 */ |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
793 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
|
794 "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
|
795 "C\x68", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
796 "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
|
797 "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
|
798 "N\x69", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
799 "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
|
800 "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
|
801 "Y\xBA", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
802 "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
|
803 "c\x48", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
804 "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
|
805 "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
|
806 "n\x49", |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
807 "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
|
808 "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
|
809 "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
|
810 }; |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
811 #endif |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
812 |
167 | 813 /* |
814 * Produce the bytes for equivalence class "c". | |
815 * Currently only handles latin1, latin9 and utf-8. | |
4444 | 816 * NOTE: When changing this function, also change nfa_emit_equi_class() |
167 | 817 */ |
818 static void | |
819 reg_equi_class(c) | |
820 int c; | |
821 { | |
822 #ifdef FEAT_MBYTE | |
823 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
492 | 824 || STRCMP(p_enc, "iso-8859-15") == 0) |
167 | 825 #endif |
826 { | |
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
827 #ifdef EBCDIC |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
828 int i; |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
829 |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
830 /* 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
|
831 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
|
832 { |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
833 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
|
834 { |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
835 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
|
836 |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
837 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
|
838 regmbc(*p++); |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
839 return; |
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
840 } |
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 #else |
167 | 843 switch (c) |
844 { | |
236 | 845 case 'A': case '\300': case '\301': case '\302': |
2974 | 846 CASEMBC(0x100) CASEMBC(0x102) CASEMBC(0x104) CASEMBC(0x1cd) |
847 CASEMBC(0x1de) CASEMBC(0x1e0) CASEMBC(0x1ea2) | |
236 | 848 case '\303': case '\304': case '\305': |
849 regmbc('A'); regmbc('\300'); regmbc('\301'); | |
850 regmbc('\302'); regmbc('\303'); regmbc('\304'); | |
851 regmbc('\305'); | |
2974 | 852 REGMBC(0x100) REGMBC(0x102) REGMBC(0x104) |
853 REGMBC(0x1cd) REGMBC(0x1de) REGMBC(0x1e0) | |
854 REGMBC(0x1ea2) | |
855 return; | |
856 case 'B': CASEMBC(0x1e02) CASEMBC(0x1e06) | |
857 regmbc('B'); REGMBC(0x1e02) REGMBC(0x1e06) | |
167 | 858 return; |
236 | 859 case 'C': case '\307': |
2974 | 860 CASEMBC(0x106) CASEMBC(0x108) CASEMBC(0x10a) CASEMBC(0x10c) |
236 | 861 regmbc('C'); regmbc('\307'); |
2974 | 862 REGMBC(0x106) REGMBC(0x108) REGMBC(0x10a) |
863 REGMBC(0x10c) | |
864 return; | |
865 case 'D': CASEMBC(0x10e) CASEMBC(0x110) CASEMBC(0x1e0a) | |
866 CASEMBC(0x1e0e) CASEMBC(0x1e10) | |
867 regmbc('D'); REGMBC(0x10e) REGMBC(0x110) | |
868 REGMBC(0x1e0a) REGMBC(0x1e0e) REGMBC(0x1e10) | |
167 | 869 return; |
236 | 870 case 'E': case '\310': case '\311': case '\312': case '\313': |
2974 | 871 CASEMBC(0x112) CASEMBC(0x114) CASEMBC(0x116) CASEMBC(0x118) |
872 CASEMBC(0x11a) CASEMBC(0x1eba) CASEMBC(0x1ebc) | |
236 | 873 regmbc('E'); regmbc('\310'); regmbc('\311'); |
874 regmbc('\312'); regmbc('\313'); | |
2974 | 875 REGMBC(0x112) REGMBC(0x114) REGMBC(0x116) |
876 REGMBC(0x118) REGMBC(0x11a) REGMBC(0x1eba) | |
877 REGMBC(0x1ebc) | |
878 return; | |
879 case 'F': CASEMBC(0x1e1e) | |
880 regmbc('F'); REGMBC(0x1e1e) | |
881 return; | |
882 case 'G': CASEMBC(0x11c) CASEMBC(0x11e) CASEMBC(0x120) | |
883 CASEMBC(0x122) CASEMBC(0x1e4) CASEMBC(0x1e6) CASEMBC(0x1f4) | |
884 CASEMBC(0x1e20) | |
885 regmbc('G'); REGMBC(0x11c) REGMBC(0x11e) | |
886 REGMBC(0x120) REGMBC(0x122) REGMBC(0x1e4) | |
887 REGMBC(0x1e6) REGMBC(0x1f4) REGMBC(0x1e20) | |
888 return; | |
889 case 'H': CASEMBC(0x124) CASEMBC(0x126) CASEMBC(0x1e22) | |
890 CASEMBC(0x1e26) CASEMBC(0x1e28) | |
891 regmbc('H'); REGMBC(0x124) REGMBC(0x126) | |
892 REGMBC(0x1e22) REGMBC(0x1e26) REGMBC(0x1e28) | |
167 | 893 return; |
236 | 894 case 'I': case '\314': case '\315': case '\316': case '\317': |
2974 | 895 CASEMBC(0x128) CASEMBC(0x12a) CASEMBC(0x12c) CASEMBC(0x12e) |
896 CASEMBC(0x130) CASEMBC(0x1cf) CASEMBC(0x1ec8) | |
236 | 897 regmbc('I'); regmbc('\314'); regmbc('\315'); |
898 regmbc('\316'); regmbc('\317'); | |
2974 | 899 REGMBC(0x128) REGMBC(0x12a) REGMBC(0x12c) |
900 REGMBC(0x12e) REGMBC(0x130) REGMBC(0x1cf) | |
901 REGMBC(0x1ec8) | |
902 return; | |
903 case 'J': CASEMBC(0x134) | |
904 regmbc('J'); REGMBC(0x134) | |
905 return; | |
906 case 'K': CASEMBC(0x136) CASEMBC(0x1e8) CASEMBC(0x1e30) | |
907 CASEMBC(0x1e34) | |
908 regmbc('K'); REGMBC(0x136) REGMBC(0x1e8) | |
909 REGMBC(0x1e30) REGMBC(0x1e34) | |
910 return; | |
911 case 'L': CASEMBC(0x139) CASEMBC(0x13b) CASEMBC(0x13d) | |
912 CASEMBC(0x13f) CASEMBC(0x141) CASEMBC(0x1e3a) | |
913 regmbc('L'); REGMBC(0x139) REGMBC(0x13b) | |
914 REGMBC(0x13d) REGMBC(0x13f) REGMBC(0x141) | |
915 REGMBC(0x1e3a) | |
916 return; | |
917 case 'M': CASEMBC(0x1e3e) CASEMBC(0x1e40) | |
918 regmbc('M'); REGMBC(0x1e3e) REGMBC(0x1e40) | |
167 | 919 return; |
236 | 920 case 'N': case '\321': |
2974 | 921 CASEMBC(0x143) CASEMBC(0x145) CASEMBC(0x147) CASEMBC(0x1e44) |
922 CASEMBC(0x1e48) | |
236 | 923 regmbc('N'); regmbc('\321'); |
2974 | 924 REGMBC(0x143) REGMBC(0x145) REGMBC(0x147) |
925 REGMBC(0x1e44) REGMBC(0x1e48) | |
167 | 926 return; |
236 | 927 case 'O': case '\322': case '\323': case '\324': case '\325': |
2974 | 928 case '\326': case '\330': |
929 CASEMBC(0x14c) CASEMBC(0x14e) CASEMBC(0x150) CASEMBC(0x1a0) | |
930 CASEMBC(0x1d1) CASEMBC(0x1ea) CASEMBC(0x1ec) CASEMBC(0x1ece) | |
236 | 931 regmbc('O'); regmbc('\322'); regmbc('\323'); |
932 regmbc('\324'); regmbc('\325'); regmbc('\326'); | |
2974 | 933 regmbc('\330'); |
934 REGMBC(0x14c) REGMBC(0x14e) REGMBC(0x150) | |
935 REGMBC(0x1a0) REGMBC(0x1d1) REGMBC(0x1ea) | |
936 REGMBC(0x1ec) REGMBC(0x1ece) | |
937 return; | |
938 case 'P': case 0x1e54: case 0x1e56: | |
939 regmbc('P'); REGMBC(0x1e54) REGMBC(0x1e56) | |
940 return; | |
941 case 'R': CASEMBC(0x154) CASEMBC(0x156) CASEMBC(0x158) | |
942 CASEMBC(0x1e58) CASEMBC(0x1e5e) | |
943 regmbc('R'); REGMBC(0x154) REGMBC(0x156) REGMBC(0x158) | |
944 REGMBC(0x1e58) REGMBC(0x1e5e) | |
945 return; | |
946 case 'S': CASEMBC(0x15a) CASEMBC(0x15c) CASEMBC(0x15e) | |
947 CASEMBC(0x160) CASEMBC(0x1e60) | |
948 regmbc('S'); REGMBC(0x15a) REGMBC(0x15c) | |
949 REGMBC(0x15e) REGMBC(0x160) REGMBC(0x1e60) | |
950 return; | |
951 case 'T': CASEMBC(0x162) CASEMBC(0x164) CASEMBC(0x166) | |
952 CASEMBC(0x1e6a) CASEMBC(0x1e6e) | |
953 regmbc('T'); REGMBC(0x162) REGMBC(0x164) | |
954 REGMBC(0x166) REGMBC(0x1e6a) REGMBC(0x1e6e) | |
167 | 955 return; |
236 | 956 case 'U': case '\331': case '\332': case '\333': case '\334': |
2974 | 957 CASEMBC(0x168) CASEMBC(0x16a) CASEMBC(0x16c) CASEMBC(0x16e) |
958 CASEMBC(0x170) CASEMBC(0x172) CASEMBC(0x1af) CASEMBC(0x1d3) | |
959 CASEMBC(0x1ee6) | |
236 | 960 regmbc('U'); regmbc('\331'); regmbc('\332'); |
961 regmbc('\333'); regmbc('\334'); | |
2974 | 962 REGMBC(0x168) REGMBC(0x16a) REGMBC(0x16c) |
963 REGMBC(0x16e) REGMBC(0x170) REGMBC(0x172) | |
964 REGMBC(0x1af) REGMBC(0x1d3) REGMBC(0x1ee6) | |
965 return; | |
966 case 'V': CASEMBC(0x1e7c) | |
967 regmbc('V'); REGMBC(0x1e7c) | |
968 return; | |
969 case 'W': CASEMBC(0x174) CASEMBC(0x1e80) CASEMBC(0x1e82) | |
970 CASEMBC(0x1e84) CASEMBC(0x1e86) | |
971 regmbc('W'); REGMBC(0x174) REGMBC(0x1e80) | |
972 REGMBC(0x1e82) REGMBC(0x1e84) REGMBC(0x1e86) | |
973 return; | |
974 case 'X': CASEMBC(0x1e8a) CASEMBC(0x1e8c) | |
975 regmbc('X'); REGMBC(0x1e8a) REGMBC(0x1e8c) | |
167 | 976 return; |
236 | 977 case 'Y': case '\335': |
2974 | 978 CASEMBC(0x176) CASEMBC(0x178) CASEMBC(0x1e8e) CASEMBC(0x1ef2) |
979 CASEMBC(0x1ef6) CASEMBC(0x1ef8) | |
236 | 980 regmbc('Y'); regmbc('\335'); |
2974 | 981 REGMBC(0x176) REGMBC(0x178) REGMBC(0x1e8e) |
982 REGMBC(0x1ef2) REGMBC(0x1ef6) REGMBC(0x1ef8) | |
983 return; | |
984 case 'Z': CASEMBC(0x179) CASEMBC(0x17b) CASEMBC(0x17d) | |
985 CASEMBC(0x1b5) CASEMBC(0x1e90) CASEMBC(0x1e94) | |
986 regmbc('Z'); REGMBC(0x179) REGMBC(0x17b) | |
987 REGMBC(0x17d) REGMBC(0x1b5) REGMBC(0x1e90) | |
988 REGMBC(0x1e94) | |
167 | 989 return; |
236 | 990 case 'a': case '\340': case '\341': case '\342': |
991 case '\343': case '\344': case '\345': | |
2974 | 992 CASEMBC(0x101) CASEMBC(0x103) CASEMBC(0x105) CASEMBC(0x1ce) |
993 CASEMBC(0x1df) CASEMBC(0x1e1) CASEMBC(0x1ea3) | |
236 | 994 regmbc('a'); regmbc('\340'); regmbc('\341'); |
995 regmbc('\342'); regmbc('\343'); regmbc('\344'); | |
996 regmbc('\345'); | |
2974 | 997 REGMBC(0x101) REGMBC(0x103) REGMBC(0x105) |
998 REGMBC(0x1ce) REGMBC(0x1df) REGMBC(0x1e1) | |
999 REGMBC(0x1ea3) | |
1000 return; | |
1001 case 'b': CASEMBC(0x1e03) CASEMBC(0x1e07) | |
1002 regmbc('b'); REGMBC(0x1e03) REGMBC(0x1e07) | |
167 | 1003 return; |
236 | 1004 case 'c': case '\347': |
2974 | 1005 CASEMBC(0x107) CASEMBC(0x109) CASEMBC(0x10b) CASEMBC(0x10d) |
236 | 1006 regmbc('c'); regmbc('\347'); |
2974 | 1007 REGMBC(0x107) REGMBC(0x109) REGMBC(0x10b) |
1008 REGMBC(0x10d) | |
1009 return; | |
1010 case 'd': CASEMBC(0x10f) CASEMBC(0x111) CASEMBC(0x1d0b) | |
1011 CASEMBC(0x1e11) | |
1012 regmbc('d'); REGMBC(0x10f) REGMBC(0x111) | |
1013 REGMBC(0x1e0b) REGMBC(0x01e0f) REGMBC(0x1e11) | |
167 | 1014 return; |
236 | 1015 case 'e': case '\350': case '\351': case '\352': case '\353': |
2974 | 1016 CASEMBC(0x113) CASEMBC(0x115) CASEMBC(0x117) CASEMBC(0x119) |
1017 CASEMBC(0x11b) CASEMBC(0x1ebb) CASEMBC(0x1ebd) | |
236 | 1018 regmbc('e'); regmbc('\350'); regmbc('\351'); |
1019 regmbc('\352'); regmbc('\353'); | |
2974 | 1020 REGMBC(0x113) REGMBC(0x115) REGMBC(0x117) |
1021 REGMBC(0x119) REGMBC(0x11b) REGMBC(0x1ebb) | |
1022 REGMBC(0x1ebd) | |
1023 return; | |
1024 case 'f': CASEMBC(0x1e1f) | |
1025 regmbc('f'); REGMBC(0x1e1f) | |
1026 return; | |
1027 case 'g': CASEMBC(0x11d) CASEMBC(0x11f) CASEMBC(0x121) | |
1028 CASEMBC(0x123) CASEMBC(0x1e5) CASEMBC(0x1e7) CASEMBC(0x1f5) | |
1029 CASEMBC(0x1e21) | |
1030 regmbc('g'); REGMBC(0x11d) REGMBC(0x11f) | |
1031 REGMBC(0x121) REGMBC(0x123) REGMBC(0x1e5) | |
1032 REGMBC(0x1e7) REGMBC(0x1f5) REGMBC(0x1e21) | |
1033 return; | |
1034 case 'h': CASEMBC(0x125) CASEMBC(0x127) CASEMBC(0x1e23) | |
1035 CASEMBC(0x1e27) CASEMBC(0x1e29) CASEMBC(0x1e96) | |
1036 regmbc('h'); REGMBC(0x125) REGMBC(0x127) | |
1037 REGMBC(0x1e23) REGMBC(0x1e27) REGMBC(0x1e29) | |
1038 REGMBC(0x1e96) | |
167 | 1039 return; |
236 | 1040 case 'i': case '\354': case '\355': case '\356': case '\357': |
2974 | 1041 CASEMBC(0x129) CASEMBC(0x12b) CASEMBC(0x12d) CASEMBC(0x12f) |
1042 CASEMBC(0x1d0) CASEMBC(0x1ec9) | |
236 | 1043 regmbc('i'); regmbc('\354'); regmbc('\355'); |
1044 regmbc('\356'); regmbc('\357'); | |
2974 | 1045 REGMBC(0x129) REGMBC(0x12b) REGMBC(0x12d) |
1046 REGMBC(0x12f) REGMBC(0x1d0) REGMBC(0x1ec9) | |
1047 return; | |
1048 case 'j': CASEMBC(0x135) CASEMBC(0x1f0) | |
1049 regmbc('j'); REGMBC(0x135) REGMBC(0x1f0) | |
1050 return; | |
1051 case 'k': CASEMBC(0x137) CASEMBC(0x1e9) CASEMBC(0x1e31) | |
1052 CASEMBC(0x1e35) | |
1053 regmbc('k'); REGMBC(0x137) REGMBC(0x1e9) | |
1054 REGMBC(0x1e31) REGMBC(0x1e35) | |
1055 return; | |
1056 case 'l': CASEMBC(0x13a) CASEMBC(0x13c) CASEMBC(0x13e) | |
1057 CASEMBC(0x140) CASEMBC(0x142) CASEMBC(0x1e3b) | |
1058 regmbc('l'); REGMBC(0x13a) REGMBC(0x13c) | |
1059 REGMBC(0x13e) REGMBC(0x140) REGMBC(0x142) | |
1060 REGMBC(0x1e3b) | |
1061 return; | |
1062 case 'm': CASEMBC(0x1e3f) CASEMBC(0x1e41) | |
1063 regmbc('m'); REGMBC(0x1e3f) REGMBC(0x1e41) | |
167 | 1064 return; |
236 | 1065 case 'n': case '\361': |
2974 | 1066 CASEMBC(0x144) CASEMBC(0x146) CASEMBC(0x148) CASEMBC(0x149) |
1067 CASEMBC(0x1e45) CASEMBC(0x1e49) | |
236 | 1068 regmbc('n'); regmbc('\361'); |
2974 | 1069 REGMBC(0x144) REGMBC(0x146) REGMBC(0x148) |
1070 REGMBC(0x149) REGMBC(0x1e45) REGMBC(0x1e49) | |
167 | 1071 return; |
236 | 1072 case 'o': case '\362': case '\363': case '\364': case '\365': |
2974 | 1073 case '\366': case '\370': |
1074 CASEMBC(0x14d) CASEMBC(0x14f) CASEMBC(0x151) CASEMBC(0x1a1) | |
1075 CASEMBC(0x1d2) CASEMBC(0x1eb) CASEMBC(0x1ed) CASEMBC(0x1ecf) | |
236 | 1076 regmbc('o'); regmbc('\362'); regmbc('\363'); |
1077 regmbc('\364'); regmbc('\365'); regmbc('\366'); | |
2974 | 1078 regmbc('\370'); |
1079 REGMBC(0x14d) REGMBC(0x14f) REGMBC(0x151) | |
1080 REGMBC(0x1a1) REGMBC(0x1d2) REGMBC(0x1eb) | |
1081 REGMBC(0x1ed) REGMBC(0x1ecf) | |
1082 return; | |
1083 case 'p': CASEMBC(0x1e55) CASEMBC(0x1e57) | |
1084 regmbc('p'); REGMBC(0x1e55) REGMBC(0x1e57) | |
1085 return; | |
1086 case 'r': CASEMBC(0x155) CASEMBC(0x157) CASEMBC(0x159) | |
1087 CASEMBC(0x1e59) CASEMBC(0x1e5f) | |
1088 regmbc('r'); REGMBC(0x155) REGMBC(0x157) REGMBC(0x159) | |
1089 REGMBC(0x1e59) REGMBC(0x1e5f) | |
1090 return; | |
1091 case 's': CASEMBC(0x15b) CASEMBC(0x15d) CASEMBC(0x15f) | |
1092 CASEMBC(0x161) CASEMBC(0x1e61) | |
1093 regmbc('s'); REGMBC(0x15b) REGMBC(0x15d) | |
1094 REGMBC(0x15f) REGMBC(0x161) REGMBC(0x1e61) | |
1095 return; | |
1096 case 't': CASEMBC(0x163) CASEMBC(0x165) CASEMBC(0x167) | |
1097 CASEMBC(0x1e6b) CASEMBC(0x1e6f) CASEMBC(0x1e97) | |
1098 regmbc('t'); REGMBC(0x163) REGMBC(0x165) REGMBC(0x167) | |
1099 REGMBC(0x1e6b) REGMBC(0x1e6f) REGMBC(0x1e97) | |
167 | 1100 return; |
236 | 1101 case 'u': case '\371': case '\372': case '\373': case '\374': |
2974 | 1102 CASEMBC(0x169) CASEMBC(0x16b) CASEMBC(0x16d) CASEMBC(0x16f) |
1103 CASEMBC(0x171) CASEMBC(0x173) CASEMBC(0x1b0) CASEMBC(0x1d4) | |
1104 CASEMBC(0x1ee7) | |
236 | 1105 regmbc('u'); regmbc('\371'); regmbc('\372'); |
1106 regmbc('\373'); regmbc('\374'); | |
2974 | 1107 REGMBC(0x169) REGMBC(0x16b) REGMBC(0x16d) |
1108 REGMBC(0x16f) REGMBC(0x171) REGMBC(0x173) | |
1109 REGMBC(0x1b0) REGMBC(0x1d4) REGMBC(0x1ee7) | |
1110 return; | |
1111 case 'v': CASEMBC(0x1e7d) | |
1112 regmbc('v'); REGMBC(0x1e7d) | |
1113 return; | |
1114 case 'w': CASEMBC(0x175) CASEMBC(0x1e81) CASEMBC(0x1e83) | |
1115 CASEMBC(0x1e85) CASEMBC(0x1e87) CASEMBC(0x1e98) | |
1116 regmbc('w'); REGMBC(0x175) REGMBC(0x1e81) | |
1117 REGMBC(0x1e83) REGMBC(0x1e85) REGMBC(0x1e87) | |
1118 REGMBC(0x1e98) | |
1119 return; | |
1120 case 'x': CASEMBC(0x1e8b) CASEMBC(0x1e8d) | |
1121 regmbc('x'); REGMBC(0x1e8b) REGMBC(0x1e8d) | |
167 | 1122 return; |
236 | 1123 case 'y': case '\375': case '\377': |
2974 | 1124 CASEMBC(0x177) CASEMBC(0x1e8f) CASEMBC(0x1e99) |
1125 CASEMBC(0x1ef3) CASEMBC(0x1ef7) CASEMBC(0x1ef9) | |
236 | 1126 regmbc('y'); regmbc('\375'); regmbc('\377'); |
2974 | 1127 REGMBC(0x177) REGMBC(0x1e8f) REGMBC(0x1e99) |
1128 REGMBC(0x1ef3) REGMBC(0x1ef7) REGMBC(0x1ef9) | |
1129 return; | |
1130 case 'z': CASEMBC(0x17a) CASEMBC(0x17c) CASEMBC(0x17e) | |
1131 CASEMBC(0x1b6) CASEMBC(0x1e91) CASEMBC(0x1e95) | |
1132 regmbc('z'); REGMBC(0x17a) REGMBC(0x17c) | |
1133 REGMBC(0x17e) REGMBC(0x1b6) REGMBC(0x1e91) | |
1134 REGMBC(0x1e95) | |
167 | 1135 return; |
1136 } | |
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
1137 #endif |
167 | 1138 } |
1139 regmbc(c); | |
1140 } | |
1141 | |
1142 /* | |
1143 * Check for a collating element "[.a.]". "pp" points to the '['. | |
1144 * Returns a character. Zero means that no item was recognized. Otherwise | |
1145 * "pp" is advanced to after the item. | |
1146 * Currently only single characters are recognized! | |
1147 */ | |
1148 static int | |
1149 get_coll_element(pp) | |
1150 char_u **pp; | |
1151 { | |
1152 int c; | |
1153 int l = 1; | |
1154 char_u *p = *pp; | |
1155 | |
1156 if (p[1] == '.') | |
1157 { | |
1158 #ifdef FEAT_MBYTE | |
1159 if (has_mbyte) | |
474 | 1160 l = (*mb_ptr2len)(p + 2); |
167 | 1161 #endif |
1162 if (p[l + 2] == '.' && p[l + 3] == ']') | |
1163 { | |
1164 #ifdef FEAT_MBYTE | |
1165 if (has_mbyte) | |
1166 c = mb_ptr2char(p + 2); | |
1167 else | |
1168 #endif | |
1169 c = p[2]; | |
1170 *pp += l + 4; | |
1171 return c; | |
1172 } | |
1173 } | |
1174 return 0; | |
1175 } | |
1176 | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1177 static void get_cpo_flags __ARGS((void)); |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1178 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
|
1179 static int reg_cpo_bsl; /* 'cpoptions' contains '\' flag */ |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1180 |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1181 static void |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1182 get_cpo_flags() |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1183 { |
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1184 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
|
1185 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
|
1186 } |
167 | 1187 |
1188 /* | |
1189 * Skip over a "[]" range. | |
1190 * "p" must point to the character after the '['. | |
1191 * The returned pointer is on the matching ']', or the terminating NUL. | |
1192 */ | |
1193 static char_u * | |
1194 skip_anyof(p) | |
1195 char_u *p; | |
1196 { | |
1197 #ifdef FEAT_MBYTE | |
1198 int l; | |
1199 #endif | |
1200 | |
1201 if (*p == '^') /* Complement of range. */ | |
1202 ++p; | |
1203 if (*p == ']' || *p == '-') | |
1204 ++p; | |
1205 while (*p != NUL && *p != ']') | |
1206 { | |
1207 #ifdef FEAT_MBYTE | |
474 | 1208 if (has_mbyte && (l = (*mb_ptr2len)(p)) > 1) |
167 | 1209 p += l; |
1210 else | |
1211 #endif | |
1212 if (*p == '-') | |
1213 { | |
1214 ++p; | |
1215 if (*p != ']' && *p != NUL) | |
1216 mb_ptr_adv(p); | |
1217 } | |
1218 else if (*p == '\\' | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1219 && !reg_cpo_bsl |
167 | 1220 && (vim_strchr(REGEXP_INRANGE, p[1]) != NULL |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1221 || (!reg_cpo_lit && vim_strchr(REGEXP_ABBR, p[1]) != NULL))) |
167 | 1222 p += 2; |
1223 else if (*p == '[') | |
1224 { | |
1225 if (get_char_class(&p) == CLASS_NONE | |
1226 && get_equi_class(&p) == 0 | |
1227 && get_coll_element(&p) == 0) | |
1228 ++p; /* It was not a class name */ | |
1229 } | |
1230 else | |
1231 ++p; | |
1232 } | |
1233 | |
1234 return p; | |
1235 } | |
1236 | |
1237 /* | |
7 | 1238 * Skip past regular expression. |
153 | 1239 * Stop at end of "startp" or where "dirc" is found ('/', '?', etc). |
7 | 1240 * Take care of characters with a backslash in front of it. |
1241 * Skip strings inside [ and ]. | |
1242 * When "newp" is not NULL and "dirc" is '?', make an allocated copy of the | |
1243 * expression and change "\?" to "?". If "*newp" is not NULL the expression | |
1244 * is changed in-place. | |
1245 */ | |
1246 char_u * | |
1247 skip_regexp(startp, dirc, magic, newp) | |
1248 char_u *startp; | |
1249 int dirc; | |
1250 int magic; | |
1251 char_u **newp; | |
1252 { | |
1253 int mymagic; | |
1254 char_u *p = startp; | |
1255 | |
1256 if (magic) | |
1257 mymagic = MAGIC_ON; | |
1258 else | |
1259 mymagic = MAGIC_OFF; | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1260 get_cpo_flags(); |
7 | 1261 |
39 | 1262 for (; p[0] != NUL; mb_ptr_adv(p)) |
7 | 1263 { |
1264 if (p[0] == dirc) /* found end of regexp */ | |
1265 break; | |
1266 if ((p[0] == '[' && mymagic >= MAGIC_ON) | |
1267 || (p[0] == '\\' && p[1] == '[' && mymagic <= MAGIC_OFF)) | |
1268 { | |
1269 p = skip_anyof(p + 1); | |
1270 if (p[0] == NUL) | |
1271 break; | |
1272 } | |
1273 else if (p[0] == '\\' && p[1] != NUL) | |
1274 { | |
1275 if (dirc == '?' && newp != NULL && p[1] == '?') | |
1276 { | |
1277 /* change "\?" to "?", make a copy first. */ | |
1278 if (*newp == NULL) | |
1279 { | |
1280 *newp = vim_strsave(startp); | |
1281 if (*newp != NULL) | |
1282 p = *newp + (p - startp); | |
1283 } | |
1284 if (*newp != NULL) | |
1621 | 1285 STRMOVE(p, p + 1); |
7 | 1286 else |
1287 ++p; | |
1288 } | |
1289 else | |
1290 ++p; /* skip next character */ | |
1291 if (*p == 'v') | |
1292 mymagic = MAGIC_ALL; | |
1293 else if (*p == 'V') | |
1294 mymagic = MAGIC_NONE; | |
1295 } | |
1296 } | |
1297 return p; | |
1298 } | |
1299 | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1300 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
|
1301 static void bt_regfree __ARGS((regprog_T *prog)); |
4444 | 1302 |
7 | 1303 /* |
4444 | 1304 * bt_regcomp() - compile a regular expression into internal code for the |
1305 * traditional back track matcher. | |
41 | 1306 * Returns the program in allocated space. Returns NULL for an error. |
7 | 1307 * |
1308 * We can't allocate space until we know how big the compiled form will be, | |
1309 * but we can't compile it (and thus know how big it is) until we've got a | |
1310 * place to put the code. So we cheat: we compile it twice, once with code | |
1311 * generation turned off and size counting turned on, and once "for real". | |
1312 * This also means that we don't allocate space until we are sure that the | |
1313 * thing really will compile successfully, and we never have to move the | |
1314 * code and thus invalidate pointers into it. (Note that it has to be in | |
1315 * one piece because vim_free() must be able to free it all.) | |
1316 * | |
1317 * Whether upper/lower case is to be ignored is decided when executing the | |
1318 * program, it does not matter here. | |
1319 * | |
1320 * Beware that the optimization-preparation code in here knows about some | |
1321 * of the structure of the compiled regexp. | |
1322 * "re_flags": RE_MAGIC and/or RE_STRING. | |
1323 */ | |
4444 | 1324 static regprog_T * |
1325 bt_regcomp(expr, re_flags) | |
7 | 1326 char_u *expr; |
1327 int re_flags; | |
1328 { | |
4444 | 1329 bt_regprog_T *r; |
7 | 1330 char_u *scan; |
1331 char_u *longest; | |
1332 int len; | |
1333 int flags; | |
1334 | |
1335 if (expr == NULL) | |
1336 EMSG_RET_NULL(_(e_null)); | |
1337 | |
1338 init_class_tab(); | |
1339 | |
1340 /* | |
1341 * First pass: determine size, legality. | |
1342 */ | |
1343 regcomp_start(expr, re_flags); | |
1344 regcode = JUST_CALC_SIZE; | |
1345 regc(REGMAGIC); | |
1346 if (reg(REG_NOPAREN, &flags) == NULL) | |
1347 return NULL; | |
1348 | |
1349 /* Small enough for pointer-storage convention? */ | |
1350 #ifdef SMALL_MALLOC /* 16 bit storage allocation */ | |
1351 if (regsize >= 65536L - 256L) | |
1352 EMSG_RET_NULL(_("E339: Pattern too long")); | |
1353 #endif | |
1354 | |
1355 /* Allocate space. */ | |
4444 | 1356 r = (bt_regprog_T *)lalloc(sizeof(bt_regprog_T) + regsize, TRUE); |
7 | 1357 if (r == NULL) |
1358 return NULL; | |
1359 | |
1360 /* | |
1361 * Second pass: emit code. | |
1362 */ | |
1363 regcomp_start(expr, re_flags); | |
1364 regcode = r->program; | |
1365 regc(REGMAGIC); | |
2010 | 1366 if (reg(REG_NOPAREN, &flags) == NULL || reg_toolong) |
7 | 1367 { |
1368 vim_free(r); | |
2010 | 1369 if (reg_toolong) |
1370 EMSG_RET_NULL(_("E339: Pattern too long")); | |
7 | 1371 return NULL; |
1372 } | |
1373 | |
1374 /* Dig out information for optimizations. */ | |
1375 r->regstart = NUL; /* Worst-case defaults. */ | |
1376 r->reganch = 0; | |
1377 r->regmust = NULL; | |
1378 r->regmlen = 0; | |
1379 r->regflags = regflags; | |
1380 if (flags & HASNL) | |
1381 r->regflags |= RF_HASNL; | |
1382 if (flags & HASLOOKBH) | |
1383 r->regflags |= RF_LOOKBH; | |
1384 #ifdef FEAT_SYN_HL | |
1385 /* Remember whether this pattern has any \z specials in it. */ | |
1386 r->reghasz = re_has_z; | |
1387 #endif | |
1388 scan = r->program + 1; /* First BRANCH. */ | |
1389 if (OP(regnext(scan)) == END) /* Only one top-level choice. */ | |
1390 { | |
1391 scan = OPERAND(scan); | |
1392 | |
1393 /* Starting-point info. */ | |
1394 if (OP(scan) == BOL || OP(scan) == RE_BOF) | |
1395 { | |
1396 r->reganch++; | |
1397 scan = regnext(scan); | |
1398 } | |
1399 | |
1400 if (OP(scan) == EXACTLY) | |
1401 { | |
1402 #ifdef FEAT_MBYTE | |
1403 if (has_mbyte) | |
1404 r->regstart = (*mb_ptr2char)(OPERAND(scan)); | |
1405 else | |
1406 #endif | |
1407 r->regstart = *OPERAND(scan); | |
1408 } | |
1409 else if ((OP(scan) == BOW | |
1410 || OP(scan) == EOW | |
1411 || OP(scan) == NOTHING | |
1412 || OP(scan) == MOPEN + 0 || OP(scan) == NOPEN | |
1413 || OP(scan) == MCLOSE + 0 || OP(scan) == NCLOSE) | |
1414 && OP(regnext(scan)) == EXACTLY) | |
1415 { | |
1416 #ifdef FEAT_MBYTE | |
1417 if (has_mbyte) | |
1418 r->regstart = (*mb_ptr2char)(OPERAND(regnext(scan))); | |
1419 else | |
1420 #endif | |
1421 r->regstart = *OPERAND(regnext(scan)); | |
1422 } | |
1423 | |
1424 /* | |
1425 * If there's something expensive in the r.e., find the longest | |
1426 * literal string that must appear and make it the regmust. Resolve | |
1427 * ties in favor of later strings, since the regstart check works | |
1428 * with the beginning of the r.e. and avoiding duplication | |
1429 * strengthens checking. Not a strong reason, but sufficient in the | |
1430 * absence of others. | |
1431 */ | |
1432 /* | |
1433 * When the r.e. starts with BOW, it is faster to look for a regmust | |
1434 * first. Used a lot for "#" and "*" commands. (Added by mool). | |
1435 */ | |
1436 if ((flags & SPSTART || OP(scan) == BOW || OP(scan) == EOW) | |
1437 && !(flags & HASNL)) | |
1438 { | |
1439 longest = NULL; | |
1440 len = 0; | |
1441 for (; scan != NULL; scan = regnext(scan)) | |
1442 if (OP(scan) == EXACTLY && STRLEN(OPERAND(scan)) >= (size_t)len) | |
1443 { | |
1444 longest = OPERAND(scan); | |
1445 len = (int)STRLEN(OPERAND(scan)); | |
1446 } | |
1447 r->regmust = longest; | |
1448 r->regmlen = len; | |
1449 } | |
1450 } | |
4444 | 1451 #ifdef BT_REGEXP_DUMP |
7 | 1452 regdump(expr, r); |
1453 #endif | |
4444 | 1454 r->engine = &bt_regengine; |
1455 return (regprog_T *)r; | |
7 | 1456 } |
1457 | |
1458 /* | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1459 * 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
|
1460 */ |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1461 static void |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1462 bt_regfree(prog) |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1463 regprog_T *prog; |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1464 { |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1465 vim_free(prog); |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1466 } |
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 /* |
7 | 1469 * Setup to parse the regexp. Used once to get the length and once to do it. |
1470 */ | |
1471 static void | |
1472 regcomp_start(expr, re_flags) | |
1473 char_u *expr; | |
1474 int re_flags; /* see vim_regcomp() */ | |
1475 { | |
1476 initchr(expr); | |
1477 if (re_flags & RE_MAGIC) | |
1478 reg_magic = MAGIC_ON; | |
1479 else | |
1480 reg_magic = MAGIC_OFF; | |
1481 reg_string = (re_flags & RE_STRING); | |
481 | 1482 reg_strict = (re_flags & RE_STRICT); |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1483 get_cpo_flags(); |
7 | 1484 |
1485 num_complex_braces = 0; | |
1486 regnpar = 1; | |
1487 vim_memset(had_endbrace, 0, sizeof(had_endbrace)); | |
1488 #ifdef FEAT_SYN_HL | |
1489 regnzpar = 1; | |
1490 re_has_z = 0; | |
1491 #endif | |
1492 regsize = 0L; | |
2010 | 1493 reg_toolong = FALSE; |
7 | 1494 regflags = 0; |
1495 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
1496 had_eol = FALSE; | |
1497 #endif | |
1498 } | |
1499 | |
1500 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
1501 /* | |
1502 * Check if during the previous call to vim_regcomp the EOL item "$" has been | |
1503 * found. This is messy, but it works fine. | |
1504 */ | |
1505 int | |
1506 vim_regcomp_had_eol() | |
1507 { | |
1508 return had_eol; | |
1509 } | |
1510 #endif | |
1511 | |
1512 /* | |
4444 | 1513 * Parse regular expression, i.e. main body or parenthesized thing. |
7 | 1514 * |
1515 * Caller must absorb opening parenthesis. | |
1516 * | |
1517 * Combining parenthesis handling with the base level of regular expression | |
1518 * is a trifle forced, but the need to tie the tails of the branches to what | |
1519 * follows makes it hard to avoid. | |
1520 */ | |
1521 static char_u * | |
1522 reg(paren, flagp) | |
1523 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
1524 int *flagp; | |
1525 { | |
1526 char_u *ret; | |
1527 char_u *br; | |
1528 char_u *ender; | |
1529 int parno = 0; | |
1530 int flags; | |
1531 | |
1532 *flagp = HASWIDTH; /* Tentatively. */ | |
1533 | |
1534 #ifdef FEAT_SYN_HL | |
1535 if (paren == REG_ZPAREN) | |
1536 { | |
1537 /* Make a ZOPEN node. */ | |
1538 if (regnzpar >= NSUBEXP) | |
1539 EMSG_RET_NULL(_("E50: Too many \\z(")); | |
1540 parno = regnzpar; | |
1541 regnzpar++; | |
1542 ret = regnode(ZOPEN + parno); | |
1543 } | |
1544 else | |
1545 #endif | |
1546 if (paren == REG_PAREN) | |
1547 { | |
1548 /* Make a MOPEN node. */ | |
1549 if (regnpar >= NSUBEXP) | |
4444 | 1550 EMSG2_RET_NULL(_("E51: Too many %s("), reg_magic == MAGIC_ALL); |
7 | 1551 parno = regnpar; |
1552 ++regnpar; | |
1553 ret = regnode(MOPEN + parno); | |
1554 } | |
1555 else if (paren == REG_NPAREN) | |
1556 { | |
1557 /* Make a NOPEN node. */ | |
1558 ret = regnode(NOPEN); | |
1559 } | |
1560 else | |
1561 ret = NULL; | |
1562 | |
1563 /* Pick up the branches, linking them together. */ | |
1564 br = regbranch(&flags); | |
1565 if (br == NULL) | |
1566 return NULL; | |
1567 if (ret != NULL) | |
1568 regtail(ret, br); /* [MZ]OPEN -> first. */ | |
1569 else | |
1570 ret = br; | |
1571 /* If one of the branches can be zero-width, the whole thing can. | |
1572 * If one of the branches has * at start or matches a line-break, the | |
1573 * whole thing can. */ | |
1574 if (!(flags & HASWIDTH)) | |
1575 *flagp &= ~HASWIDTH; | |
1576 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
1577 while (peekchr() == Magic('|')) | |
1578 { | |
1579 skipchr(); | |
1580 br = regbranch(&flags); | |
2010 | 1581 if (br == NULL || reg_toolong) |
7 | 1582 return NULL; |
1583 regtail(ret, br); /* BRANCH -> BRANCH. */ | |
1584 if (!(flags & HASWIDTH)) | |
1585 *flagp &= ~HASWIDTH; | |
1586 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
1587 } | |
1588 | |
1589 /* Make a closing node, and hook it on the end. */ | |
1590 ender = regnode( | |
1591 #ifdef FEAT_SYN_HL | |
1592 paren == REG_ZPAREN ? ZCLOSE + parno : | |
1593 #endif | |
1594 paren == REG_PAREN ? MCLOSE + parno : | |
1595 paren == REG_NPAREN ? NCLOSE : END); | |
1596 regtail(ret, ender); | |
1597 | |
1598 /* Hook the tails of the branches to the closing node. */ | |
1599 for (br = ret; br != NULL; br = regnext(br)) | |
1600 regoptail(br, ender); | |
1601 | |
1602 /* Check for proper termination. */ | |
1603 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
1604 { | |
1605 #ifdef FEAT_SYN_HL | |
1606 if (paren == REG_ZPAREN) | |
308 | 1607 EMSG_RET_NULL(_("E52: Unmatched \\z(")); |
7 | 1608 else |
1609 #endif | |
1610 if (paren == REG_NPAREN) | |
4444 | 1611 EMSG2_RET_NULL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); |
7 | 1612 else |
4444 | 1613 EMSG2_RET_NULL(_(e_unmatchedp), reg_magic == MAGIC_ALL); |
7 | 1614 } |
1615 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
1616 { | |
1617 if (curchr == Magic(')')) | |
4444 | 1618 EMSG2_RET_NULL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); |
7 | 1619 else |
308 | 1620 EMSG_RET_NULL(_(e_trailing)); /* "Can't happen". */ |
7 | 1621 /* NOTREACHED */ |
1622 } | |
1623 /* | |
1624 * Here we set the flag allowing back references to this set of | |
1625 * parentheses. | |
1626 */ | |
1627 if (paren == REG_PAREN) | |
1628 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
1629 return ret; | |
1630 } | |
1631 | |
1632 /* | |
4444 | 1633 * Parse one alternative of an | operator. |
7 | 1634 * Implements the & operator. |
1635 */ | |
1636 static char_u * | |
1637 regbranch(flagp) | |
1638 int *flagp; | |
1639 { | |
1640 char_u *ret; | |
1641 char_u *chain = NULL; | |
1642 char_u *latest; | |
1643 int flags; | |
1644 | |
1645 *flagp = WORST | HASNL; /* Tentatively. */ | |
1646 | |
1647 ret = regnode(BRANCH); | |
1648 for (;;) | |
1649 { | |
1650 latest = regconcat(&flags); | |
1651 if (latest == NULL) | |
1652 return NULL; | |
1653 /* If one of the branches has width, the whole thing has. If one of | |
1654 * the branches anchors at start-of-line, the whole thing does. | |
1655 * If one of the branches uses look-behind, the whole thing does. */ | |
1656 *flagp |= flags & (HASWIDTH | SPSTART | HASLOOKBH); | |
1657 /* If one of the branches doesn't match a line-break, the whole thing | |
1658 * doesn't. */ | |
1659 *flagp &= ~HASNL | (flags & HASNL); | |
1660 if (chain != NULL) | |
1661 regtail(chain, latest); | |
1662 if (peekchr() != Magic('&')) | |
1663 break; | |
1664 skipchr(); | |
1665 regtail(latest, regnode(END)); /* operand ends */ | |
2010 | 1666 if (reg_toolong) |
1667 break; | |
7 | 1668 reginsert(MATCH, latest); |
1669 chain = latest; | |
1670 } | |
1671 | |
1672 return ret; | |
1673 } | |
1674 | |
1675 /* | |
4444 | 1676 * Parse one alternative of an | or & operator. |
7 | 1677 * Implements the concatenation operator. |
1678 */ | |
1679 static char_u * | |
1680 regconcat(flagp) | |
1681 int *flagp; | |
1682 { | |
1683 char_u *first = NULL; | |
1684 char_u *chain = NULL; | |
1685 char_u *latest; | |
1686 int flags; | |
1687 int cont = TRUE; | |
1688 | |
1689 *flagp = WORST; /* Tentatively. */ | |
1690 | |
1691 while (cont) | |
1692 { | |
1693 switch (peekchr()) | |
1694 { | |
1695 case NUL: | |
1696 case Magic('|'): | |
1697 case Magic('&'): | |
1698 case Magic(')'): | |
1699 cont = FALSE; | |
1700 break; | |
1701 case Magic('Z'): | |
1702 #ifdef FEAT_MBYTE | |
1703 regflags |= RF_ICOMBINE; | |
1704 #endif | |
1705 skipchr_keepstart(); | |
1706 break; | |
1707 case Magic('c'): | |
1708 regflags |= RF_ICASE; | |
1709 skipchr_keepstart(); | |
1710 break; | |
1711 case Magic('C'): | |
1712 regflags |= RF_NOICASE; | |
1713 skipchr_keepstart(); | |
1714 break; | |
1715 case Magic('v'): | |
1716 reg_magic = MAGIC_ALL; | |
1717 skipchr_keepstart(); | |
1718 curchr = -1; | |
1719 break; | |
1720 case Magic('m'): | |
1721 reg_magic = MAGIC_ON; | |
1722 skipchr_keepstart(); | |
1723 curchr = -1; | |
1724 break; | |
1725 case Magic('M'): | |
1726 reg_magic = MAGIC_OFF; | |
1727 skipchr_keepstart(); | |
1728 curchr = -1; | |
1729 break; | |
1730 case Magic('V'): | |
1731 reg_magic = MAGIC_NONE; | |
1732 skipchr_keepstart(); | |
1733 curchr = -1; | |
1734 break; | |
1735 default: | |
1736 latest = regpiece(&flags); | |
2010 | 1737 if (latest == NULL || reg_toolong) |
7 | 1738 return NULL; |
1739 *flagp |= flags & (HASWIDTH | HASNL | HASLOOKBH); | |
1740 if (chain == NULL) /* First piece. */ | |
1741 *flagp |= flags & SPSTART; | |
1742 else | |
1743 regtail(chain, latest); | |
1744 chain = latest; | |
1745 if (first == NULL) | |
1746 first = latest; | |
1747 break; | |
1748 } | |
1749 } | |
1750 if (first == NULL) /* Loop ran zero times. */ | |
1751 first = regnode(NOTHING); | |
1752 return first; | |
1753 } | |
1754 | |
1755 /* | |
4444 | 1756 * Parse something followed by possible [*+=]. |
7 | 1757 * |
1758 * Note that the branching code sequences used for = and the general cases | |
1759 * of * and + are somewhat optimized: they use the same NOTHING node as | |
1760 * both the endmarker for their branch list and the body of the last branch. | |
1761 * It might seem that this node could be dispensed with entirely, but the | |
1762 * endmarker role is not redundant. | |
1763 */ | |
1764 static char_u * | |
1765 regpiece(flagp) | |
1766 int *flagp; | |
1767 { | |
1768 char_u *ret; | |
1769 int op; | |
1770 char_u *next; | |
1771 int flags; | |
1772 long minval; | |
1773 long maxval; | |
1774 | |
1775 ret = regatom(&flags); | |
1776 if (ret == NULL) | |
1777 return NULL; | |
1778 | |
1779 op = peekchr(); | |
1780 if (re_multi_type(op) == NOT_MULTI) | |
1781 { | |
1782 *flagp = flags; | |
1783 return ret; | |
1784 } | |
1785 /* default flags */ | |
1786 *flagp = (WORST | SPSTART | (flags & (HASNL | HASLOOKBH))); | |
1787 | |
1788 skipchr(); | |
1789 switch (op) | |
1790 { | |
1791 case Magic('*'): | |
1792 if (flags & SIMPLE) | |
1793 reginsert(STAR, ret); | |
1794 else | |
1795 { | |
1796 /* Emit x* as (x&|), where & means "self". */ | |
1797 reginsert(BRANCH, ret); /* Either x */ | |
1798 regoptail(ret, regnode(BACK)); /* and loop */ | |
1799 regoptail(ret, ret); /* back */ | |
1800 regtail(ret, regnode(BRANCH)); /* or */ | |
1801 regtail(ret, regnode(NOTHING)); /* null. */ | |
1802 } | |
1803 break; | |
1804 | |
1805 case Magic('+'): | |
1806 if (flags & SIMPLE) | |
1807 reginsert(PLUS, ret); | |
1808 else | |
1809 { | |
1810 /* Emit x+ as x(&|), where & means "self". */ | |
1811 next = regnode(BRANCH); /* Either */ | |
1812 regtail(ret, next); | |
233 | 1813 regtail(regnode(BACK), ret); /* loop back */ |
7 | 1814 regtail(next, regnode(BRANCH)); /* or */ |
1815 regtail(ret, regnode(NOTHING)); /* null. */ | |
1816 } | |
1817 *flagp = (WORST | HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
1818 break; | |
1819 | |
1820 case Magic('@'): | |
1821 { | |
1822 int lop = END; | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1823 int nr; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1824 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1825 nr = getdecchrs(); |
7 | 1826 switch (no_Magic(getchr())) |
1827 { | |
1828 case '=': lop = MATCH; break; /* \@= */ | |
1829 case '!': lop = NOMATCH; break; /* \@! */ | |
1830 case '>': lop = SUBPAT; break; /* \@> */ | |
1831 case '<': switch (no_Magic(getchr())) | |
1832 { | |
1833 case '=': lop = BEHIND; break; /* \@<= */ | |
1834 case '!': lop = NOBEHIND; break; /* \@<! */ | |
1835 } | |
1836 } | |
1837 if (lop == END) | |
4444 | 1838 EMSG2_RET_NULL(_("E59: invalid character after %s@"), |
7 | 1839 reg_magic == MAGIC_ALL); |
1840 /* Look behind must match with behind_pos. */ | |
1841 if (lop == BEHIND || lop == NOBEHIND) | |
1842 { | |
1843 regtail(ret, regnode(BHPOS)); | |
1844 *flagp |= HASLOOKBH; | |
1845 } | |
1846 regtail(ret, regnode(END)); /* operand ends */ | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1847 if (lop == BEHIND || lop == NOBEHIND) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1848 { |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1849 if (nr < 0) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1850 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
|
1851 reginsert_nr(lop, nr, ret); |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1852 } |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1853 else |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1854 reginsert(lop, ret); |
7 | 1855 break; |
1856 } | |
1857 | |
1858 case Magic('?'): | |
1859 case Magic('='): | |
1860 /* Emit x= as (x|) */ | |
1861 reginsert(BRANCH, ret); /* Either x */ | |
1862 regtail(ret, regnode(BRANCH)); /* or */ | |
1863 next = regnode(NOTHING); /* null. */ | |
1864 regtail(ret, next); | |
1865 regoptail(ret, next); | |
1866 break; | |
1867 | |
1868 case Magic('{'): | |
1869 if (!read_limits(&minval, &maxval)) | |
1870 return NULL; | |
1871 if (flags & SIMPLE) | |
1872 { | |
1873 reginsert(BRACE_SIMPLE, ret); | |
1874 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
1875 } | |
1876 else | |
1877 { | |
1878 if (num_complex_braces >= 10) | |
4444 | 1879 EMSG2_RET_NULL(_("E60: Too many complex %s{...}s"), |
7 | 1880 reg_magic == MAGIC_ALL); |
1881 reginsert(BRACE_COMPLEX + num_complex_braces, ret); | |
1882 regoptail(ret, regnode(BACK)); | |
1883 regoptail(ret, ret); | |
1884 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
1885 ++num_complex_braces; | |
1886 } | |
1887 if (minval > 0 && maxval > 0) | |
1888 *flagp = (HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
1889 break; | |
1890 } | |
1891 if (re_multi_type(peekchr()) != NOT_MULTI) | |
1892 { | |
1893 /* Can't have a multi follow a multi. */ | |
1894 if (peekchr() == Magic('*')) | |
1895 sprintf((char *)IObuff, _("E61: Nested %s*"), | |
1896 reg_magic >= MAGIC_ON ? "" : "\\"); | |
1897 else | |
1898 sprintf((char *)IObuff, _("E62: Nested %s%c"), | |
1899 reg_magic == MAGIC_ALL ? "" : "\\", no_Magic(peekchr())); | |
1900 EMSG_RET_NULL(IObuff); | |
1901 } | |
1902 | |
1903 return ret; | |
1904 } | |
1905 | |
4444 | 1906 /* When making changes to classchars also change nfa_classcodes. */ |
1907 static char_u *classchars = (char_u *)".iIkKfFpPsSdDxXoOwWhHaAlLuU"; | |
1908 static int classcodes[] = { | |
1909 ANY, IDENT, SIDENT, KWORD, SKWORD, | |
1910 FNAME, SFNAME, PRINT, SPRINT, | |
1911 WHITE, NWHITE, DIGIT, NDIGIT, | |
1912 HEX, NHEX, OCTAL, NOCTAL, | |
1913 WORD, NWORD, HEAD, NHEAD, | |
1914 ALPHA, NALPHA, LOWER, NLOWER, | |
1915 UPPER, NUPPER | |
1916 }; | |
1917 | |
7 | 1918 /* |
4444 | 1919 * Parse the lowest level. |
7 | 1920 * |
1921 * Optimization: gobbles an entire sequence of ordinary characters so that | |
1922 * it can turn them into a single node, which is smaller to store and | |
1923 * faster to run. Don't do this when one_exactly is set. | |
1924 */ | |
1925 static char_u * | |
1926 regatom(flagp) | |
1927 int *flagp; | |
1928 { | |
1929 char_u *ret; | |
1930 int flags; | |
1931 int c; | |
1932 char_u *p; | |
1933 int extra = 0; | |
1934 | |
1935 *flagp = WORST; /* Tentatively. */ | |
1936 | |
1937 c = getchr(); | |
1938 switch (c) | |
1939 { | |
1940 case Magic('^'): | |
1941 ret = regnode(BOL); | |
1942 break; | |
1943 | |
1944 case Magic('$'): | |
1945 ret = regnode(EOL); | |
1946 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
1947 had_eol = TRUE; | |
1948 #endif | |
1949 break; | |
1950 | |
1951 case Magic('<'): | |
1952 ret = regnode(BOW); | |
1953 break; | |
1954 | |
1955 case Magic('>'): | |
1956 ret = regnode(EOW); | |
1957 break; | |
1958 | |
1959 case Magic('_'): | |
1960 c = no_Magic(getchr()); | |
1961 if (c == '^') /* "\_^" is start-of-line */ | |
1962 { | |
1963 ret = regnode(BOL); | |
1964 break; | |
1965 } | |
1966 if (c == '$') /* "\_$" is end-of-line */ | |
1967 { | |
1968 ret = regnode(EOL); | |
1969 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
1970 had_eol = TRUE; | |
1971 #endif | |
1972 break; | |
1973 } | |
1974 | |
1975 extra = ADD_NL; | |
1976 *flagp |= HASNL; | |
1977 | |
1978 /* "\_[" is character range plus newline */ | |
1979 if (c == '[') | |
1980 goto collection; | |
1981 | |
1982 /* "\_x" is character class plus newline */ | |
1983 /*FALLTHROUGH*/ | |
1984 | |
1985 /* | |
1986 * Character classes. | |
1987 */ | |
1988 case Magic('.'): | |
1989 case Magic('i'): | |
1990 case Magic('I'): | |
1991 case Magic('k'): | |
1992 case Magic('K'): | |
1993 case Magic('f'): | |
1994 case Magic('F'): | |
1995 case Magic('p'): | |
1996 case Magic('P'): | |
1997 case Magic('s'): | |
1998 case Magic('S'): | |
1999 case Magic('d'): | |
2000 case Magic('D'): | |
2001 case Magic('x'): | |
2002 case Magic('X'): | |
2003 case Magic('o'): | |
2004 case Magic('O'): | |
2005 case Magic('w'): | |
2006 case Magic('W'): | |
2007 case Magic('h'): | |
2008 case Magic('H'): | |
2009 case Magic('a'): | |
2010 case Magic('A'): | |
2011 case Magic('l'): | |
2012 case Magic('L'): | |
2013 case Magic('u'): | |
2014 case Magic('U'): | |
2015 p = vim_strchr(classchars, no_Magic(c)); | |
2016 if (p == NULL) | |
2017 EMSG_RET_NULL(_("E63: invalid use of \\_")); | |
714 | 2018 #ifdef FEAT_MBYTE |
2019 /* When '.' is followed by a composing char ignore the dot, so that | |
2020 * the composing char is matched here. */ | |
2021 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
2022 { | |
2023 c = getchr(); | |
2024 goto do_multibyte; | |
2025 } | |
2026 #endif | |
7 | 2027 ret = regnode(classcodes[p - classchars] + extra); |
2028 *flagp |= HASWIDTH | SIMPLE; | |
2029 break; | |
2030 | |
2031 case Magic('n'): | |
2032 if (reg_string) | |
2033 { | |
2034 /* In a string "\n" matches a newline character. */ | |
2035 ret = regnode(EXACTLY); | |
2036 regc(NL); | |
2037 regc(NUL); | |
2038 *flagp |= HASWIDTH | SIMPLE; | |
2039 } | |
2040 else | |
2041 { | |
2042 /* In buffer text "\n" matches the end of a line. */ | |
2043 ret = regnode(NEWL); | |
2044 *flagp |= HASWIDTH | HASNL; | |
2045 } | |
2046 break; | |
2047 | |
2048 case Magic('('): | |
2049 if (one_exactly) | |
2050 EMSG_ONE_RET_NULL; | |
2051 ret = reg(REG_PAREN, &flags); | |
2052 if (ret == NULL) | |
2053 return NULL; | |
2054 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
2055 break; | |
2056 | |
2057 case NUL: | |
2058 case Magic('|'): | |
2059 case Magic('&'): | |
2060 case Magic(')'): | |
1468 | 2061 if (one_exactly) |
2062 EMSG_ONE_RET_NULL; | |
7 | 2063 EMSG_RET_NULL(_(e_internal)); /* Supposed to be caught earlier. */ |
2064 /* NOTREACHED */ | |
2065 | |
2066 case Magic('='): | |
2067 case Magic('?'): | |
2068 case Magic('+'): | |
2069 case Magic('@'): | |
2070 case Magic('{'): | |
2071 case Magic('*'): | |
2072 c = no_Magic(c); | |
2073 sprintf((char *)IObuff, _("E64: %s%c follows nothing"), | |
2074 (c == '*' ? reg_magic >= MAGIC_ON : reg_magic == MAGIC_ALL) | |
2075 ? "" : "\\", c); | |
2076 EMSG_RET_NULL(IObuff); | |
2077 /* NOTREACHED */ | |
2078 | |
2079 case Magic('~'): /* previous substitute pattern */ | |
359 | 2080 if (reg_prev_sub != NULL) |
7 | 2081 { |
2082 char_u *lp; | |
2083 | |
2084 ret = regnode(EXACTLY); | |
2085 lp = reg_prev_sub; | |
2086 while (*lp != NUL) | |
2087 regc(*lp++); | |
2088 regc(NUL); | |
2089 if (*reg_prev_sub != NUL) | |
2090 { | |
2091 *flagp |= HASWIDTH; | |
2092 if ((lp - reg_prev_sub) == 1) | |
2093 *flagp |= SIMPLE; | |
2094 } | |
2095 } | |
2096 else | |
2097 EMSG_RET_NULL(_(e_nopresub)); | |
2098 break; | |
2099 | |
2100 case Magic('1'): | |
2101 case Magic('2'): | |
2102 case Magic('3'): | |
2103 case Magic('4'): | |
2104 case Magic('5'): | |
2105 case Magic('6'): | |
2106 case Magic('7'): | |
2107 case Magic('8'): | |
2108 case Magic('9'): | |
2109 { | |
2110 int refnum; | |
2111 | |
2112 refnum = c - Magic('0'); | |
2113 /* | |
2114 * Check if the back reference is legal. We must have seen the | |
2115 * close brace. | |
2116 * TODO: Should also check that we don't refer to something | |
2117 * that is repeated (+*=): what instance of the repetition | |
2118 * should we match? | |
2119 */ | |
2120 if (!had_endbrace[refnum]) | |
2121 { | |
2122 /* Trick: check if "@<=" or "@<!" follows, in which case | |
2123 * the \1 can appear before the referenced match. */ | |
2124 for (p = regparse; *p != NUL; ++p) | |
2125 if (p[0] == '@' && p[1] == '<' | |
2126 && (p[2] == '!' || p[2] == '=')) | |
2127 break; | |
2128 if (*p == NUL) | |
2129 EMSG_RET_NULL(_("E65: Illegal back reference")); | |
2130 } | |
2131 ret = regnode(BACKREF + refnum); | |
2132 } | |
2133 break; | |
2134 | |
2135 case Magic('z'): | |
2136 { | |
2137 c = no_Magic(getchr()); | |
2138 switch (c) | |
2139 { | |
741 | 2140 #ifdef FEAT_SYN_HL |
7 | 2141 case '(': if (reg_do_extmatch != REX_SET) |
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
2142 EMSG_RET_NULL(_(e_z_not_allowed)); |
7 | 2143 if (one_exactly) |
2144 EMSG_ONE_RET_NULL; | |
2145 ret = reg(REG_ZPAREN, &flags); | |
2146 if (ret == NULL) | |
2147 return NULL; | |
2148 *flagp |= flags & (HASWIDTH|SPSTART|HASNL|HASLOOKBH); | |
2149 re_has_z = REX_SET; | |
2150 break; | |
2151 | |
2152 case '1': | |
2153 case '2': | |
2154 case '3': | |
2155 case '4': | |
2156 case '5': | |
2157 case '6': | |
2158 case '7': | |
2159 case '8': | |
2160 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
|
2161 EMSG_RET_NULL(_(e_z1_not_allowed)); |
7 | 2162 ret = regnode(ZREF + c - '0'); |
2163 re_has_z = REX_USE; | |
2164 break; | |
741 | 2165 #endif |
7 | 2166 |
2167 case 's': ret = regnode(MOPEN + 0); | |
2168 break; | |
2169 | |
2170 case 'e': ret = regnode(MCLOSE + 0); | |
2171 break; | |
2172 | |
2173 default: EMSG_RET_NULL(_("E68: Invalid character after \\z")); | |
2174 } | |
2175 } | |
2176 break; | |
2177 | |
2178 case Magic('%'): | |
2179 { | |
2180 c = no_Magic(getchr()); | |
2181 switch (c) | |
2182 { | |
2183 /* () without a back reference */ | |
2184 case '(': | |
2185 if (one_exactly) | |
2186 EMSG_ONE_RET_NULL; | |
2187 ret = reg(REG_NPAREN, &flags); | |
2188 if (ret == NULL) | |
2189 return NULL; | |
2190 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
2191 break; | |
2192 | |
2193 /* Catch \%^ and \%$ regardless of where they appear in the | |
2194 * pattern -- regardless of whether or not it makes sense. */ | |
2195 case '^': | |
2196 ret = regnode(RE_BOF); | |
2197 break; | |
2198 | |
2199 case '$': | |
2200 ret = regnode(RE_EOF); | |
2201 break; | |
2202 | |
2203 case '#': | |
2204 ret = regnode(CURSOR); | |
2205 break; | |
2206 | |
639 | 2207 case 'V': |
2208 ret = regnode(RE_VISUAL); | |
2209 break; | |
2210 | |
7 | 2211 /* \%[abc]: Emit as a list of branches, all ending at the last |
2212 * branch which matches nothing. */ | |
2213 case '[': | |
2214 if (one_exactly) /* doesn't nest */ | |
2215 EMSG_ONE_RET_NULL; | |
2216 { | |
2217 char_u *lastbranch; | |
2218 char_u *lastnode = NULL; | |
2219 char_u *br; | |
2220 | |
2221 ret = NULL; | |
2222 while ((c = getchr()) != ']') | |
2223 { | |
2224 if (c == NUL) | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2225 EMSG2_RET_NULL(_(e_missing_sb), |
7 | 2226 reg_magic == MAGIC_ALL); |
2227 br = regnode(BRANCH); | |
2228 if (ret == NULL) | |
2229 ret = br; | |
2230 else | |
2231 regtail(lastnode, br); | |
2232 | |
2233 ungetchr(); | |
2234 one_exactly = TRUE; | |
2235 lastnode = regatom(flagp); | |
2236 one_exactly = FALSE; | |
2237 if (lastnode == NULL) | |
2238 return NULL; | |
2239 } | |
2240 if (ret == NULL) | |
4760
532a9855bd30
updated for version 7.3.1127
Bram Moolenaar <bram@vim.org>
parents:
4746
diff
changeset
|
2241 EMSG2_RET_NULL(_(e_empty_sb), |
7 | 2242 reg_magic == MAGIC_ALL); |
2243 lastbranch = regnode(BRANCH); | |
2244 br = regnode(NOTHING); | |
2245 if (ret != JUST_CALC_SIZE) | |
2246 { | |
2247 regtail(lastnode, br); | |
2248 regtail(lastbranch, br); | |
2249 /* connect all branches to the NOTHING | |
2250 * branch at the end */ | |
2251 for (br = ret; br != lastnode; ) | |
2252 { | |
2253 if (OP(br) == BRANCH) | |
2254 { | |
2255 regtail(br, lastbranch); | |
2256 br = OPERAND(br); | |
2257 } | |
2258 else | |
2259 br = regnext(br); | |
2260 } | |
2261 } | |
1701 | 2262 *flagp &= ~(HASWIDTH | SIMPLE); |
7 | 2263 break; |
2264 } | |
2265 | |
24 | 2266 case 'd': /* %d123 decimal */ |
2267 case 'o': /* %o123 octal */ | |
2268 case 'x': /* %xab hex 2 */ | |
2269 case 'u': /* %uabcd hex 4 */ | |
2270 case 'U': /* %U1234abcd hex 8 */ | |
2271 { | |
2272 int i; | |
2273 | |
2274 switch (c) | |
2275 { | |
2276 case 'd': i = getdecchrs(); break; | |
2277 case 'o': i = getoctchrs(); break; | |
2278 case 'x': i = gethexchrs(2); break; | |
2279 case 'u': i = gethexchrs(4); break; | |
2280 case 'U': i = gethexchrs(8); break; | |
2281 default: i = -1; break; | |
2282 } | |
2283 | |
2284 if (i < 0) | |
4444 | 2285 EMSG2_RET_NULL( |
24 | 2286 _("E678: Invalid character after %s%%[dxouU]"), |
2287 reg_magic == MAGIC_ALL); | |
714 | 2288 #ifdef FEAT_MBYTE |
2289 if (use_multibytecode(i)) | |
2290 ret = regnode(MULTIBYTECODE); | |
2291 else | |
2292 #endif | |
2293 ret = regnode(EXACTLY); | |
24 | 2294 if (i == 0) |
2295 regc(0x0a); | |
2296 else | |
2297 #ifdef FEAT_MBYTE | |
2298 regmbc(i); | |
2299 #else | |
2300 regc(i); | |
2301 #endif | |
2302 regc(NUL); | |
2303 *flagp |= HASWIDTH; | |
2304 break; | |
2305 } | |
2306 | |
7 | 2307 default: |
639 | 2308 if (VIM_ISDIGIT(c) || c == '<' || c == '>' |
2309 || c == '\'') | |
7 | 2310 { |
2311 long_u n = 0; | |
2312 int cmp; | |
2313 | |
2314 cmp = c; | |
2315 if (cmp == '<' || cmp == '>') | |
2316 c = getchr(); | |
2317 while (VIM_ISDIGIT(c)) | |
2318 { | |
2319 n = n * 10 + (c - '0'); | |
2320 c = getchr(); | |
2321 } | |
639 | 2322 if (c == '\'' && n == 0) |
2323 { | |
2324 /* "\%'m", "\%<'m" and "\%>'m": Mark */ | |
2325 c = getchr(); | |
2326 ret = regnode(RE_MARK); | |
2327 if (ret == JUST_CALC_SIZE) | |
2328 regsize += 2; | |
2329 else | |
2330 { | |
2331 *regcode++ = c; | |
2332 *regcode++ = cmp; | |
2333 } | |
2334 break; | |
2335 } | |
2336 else if (c == 'l' || c == 'c' || c == 'v') | |
7 | 2337 { |
2338 if (c == 'l') | |
2339 ret = regnode(RE_LNUM); | |
2340 else if (c == 'c') | |
2341 ret = regnode(RE_COL); | |
2342 else | |
2343 ret = regnode(RE_VCOL); | |
2344 if (ret == JUST_CALC_SIZE) | |
2345 regsize += 5; | |
2346 else | |
2347 { | |
2348 /* put the number and the optional | |
2349 * comparator after the opcode */ | |
2350 regcode = re_put_long(regcode, n); | |
2351 *regcode++ = cmp; | |
2352 } | |
2353 break; | |
2354 } | |
2355 } | |
2356 | |
4444 | 2357 EMSG2_RET_NULL(_("E71: Invalid character after %s%%"), |
7 | 2358 reg_magic == MAGIC_ALL); |
2359 } | |
2360 } | |
2361 break; | |
2362 | |
2363 case Magic('['): | |
2364 collection: | |
2365 { | |
2366 char_u *lp; | |
2367 | |
2368 /* | |
2369 * If there is no matching ']', we assume the '[' is a normal | |
2370 * character. This makes 'incsearch' and ":help [" work. | |
2371 */ | |
2372 lp = skip_anyof(regparse); | |
2373 if (*lp == ']') /* there is a matching ']' */ | |
2374 { | |
2375 int startc = -1; /* > 0 when next '-' is a range */ | |
2376 int endc; | |
2377 | |
2378 /* | |
2379 * In a character class, different parsing rules apply. | |
2380 * Not even \ is special anymore, nothing is. | |
2381 */ | |
2382 if (*regparse == '^') /* Complement of range. */ | |
2383 { | |
2384 ret = regnode(ANYBUT + extra); | |
2385 regparse++; | |
2386 } | |
2387 else | |
2388 ret = regnode(ANYOF + extra); | |
2389 | |
2390 /* At the start ']' and '-' mean the literal character. */ | |
2391 if (*regparse == ']' || *regparse == '-') | |
167 | 2392 { |
2393 startc = *regparse; | |
7 | 2394 regc(*regparse++); |
167 | 2395 } |
7 | 2396 |
2397 while (*regparse != NUL && *regparse != ']') | |
2398 { | |
2399 if (*regparse == '-') | |
2400 { | |
2401 ++regparse; | |
2402 /* The '-' is not used for a range at the end and | |
2403 * after or before a '\n'. */ | |
2404 if (*regparse == ']' || *regparse == NUL | |
2405 || startc == -1 | |
2406 || (regparse[0] == '\\' && regparse[1] == 'n')) | |
2407 { | |
2408 regc('-'); | |
2409 startc = '-'; /* [--x] is a range */ | |
2410 } | |
2411 else | |
2412 { | |
167 | 2413 /* Also accept "a-[.z.]" */ |
2414 endc = 0; | |
2415 if (*regparse == '[') | |
2416 endc = get_coll_element(®parse); | |
2417 if (endc == 0) | |
2418 { | |
7 | 2419 #ifdef FEAT_MBYTE |
167 | 2420 if (has_mbyte) |
2421 endc = mb_ptr2char_adv(®parse); | |
2422 else | |
7 | 2423 #endif |
167 | 2424 endc = *regparse++; |
2425 } | |
24 | 2426 |
2427 /* Handle \o40, \x20 and \u20AC style sequences */ | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2428 if (endc == '\\' && !reg_cpo_lit && !reg_cpo_bsl) |
24 | 2429 endc = coll_get_char(); |
2430 | |
7 | 2431 if (startc > endc) |
2432 EMSG_RET_NULL(_(e_invrange)); | |
2433 #ifdef FEAT_MBYTE | |
2434 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
2435 || (*mb_char2len)(endc) > 1)) | |
2436 { | |
2437 /* Limit to a range of 256 chars */ | |
2438 if (endc > startc + 256) | |
2439 EMSG_RET_NULL(_(e_invrange)); | |
2440 while (++startc <= endc) | |
2441 regmbc(startc); | |
2442 } | |
2443 else | |
2444 #endif | |
2445 { | |
2446 #ifdef EBCDIC | |
2447 int alpha_only = FALSE; | |
2448 | |
2449 /* for alphabetical range skip the gaps | |
2450 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
2451 if (isalpha(startc) && isalpha(endc)) | |
2452 alpha_only = TRUE; | |
2453 #endif | |
2454 while (++startc <= endc) | |
2455 #ifdef EBCDIC | |
2456 if (!alpha_only || isalpha(startc)) | |
2457 #endif | |
2458 regc(startc); | |
2459 } | |
2460 startc = -1; | |
2461 } | |
2462 } | |
2463 /* | |
2464 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
2465 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
2466 * 'cpoptions' is not included. | |
167 | 2467 * Posix doesn't recognize backslash at all. |
7 | 2468 */ |
2469 else if (*regparse == '\\' | |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2470 && !reg_cpo_bsl |
7 | 2471 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL |
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2472 || (!reg_cpo_lit |
7 | 2473 && vim_strchr(REGEXP_ABBR, |
2474 regparse[1]) != NULL))) | |
2475 { | |
2476 regparse++; | |
2477 if (*regparse == 'n') | |
2478 { | |
2479 /* '\n' in range: also match NL */ | |
2480 if (ret != JUST_CALC_SIZE) | |
2481 { | |
4084 | 2482 /* Using \n inside [^] does not change what |
2483 * matches. "[^\n]" is the same as ".". */ | |
2484 if (*ret == ANYOF) | |
2485 { | |
7 | 2486 *ret = ANYOF + ADD_NL; |
4084 | 2487 *flagp |= HASNL; |
2488 } | |
7 | 2489 /* else: must have had a \n already */ |
2490 } | |
2491 regparse++; | |
2492 startc = -1; | |
2493 } | |
24 | 2494 else if (*regparse == 'd' |
2495 || *regparse == 'o' | |
2496 || *regparse == 'x' | |
2497 || *regparse == 'u' | |
2498 || *regparse == 'U') | |
2499 { | |
2500 startc = coll_get_char(); | |
2501 if (startc == 0) | |
2502 regc(0x0a); | |
2503 else | |
2504 #ifdef FEAT_MBYTE | |
2505 regmbc(startc); | |
2506 #else | |
2507 regc(startc); | |
2508 #endif | |
2509 } | |
7 | 2510 else |
2511 { | |
2512 startc = backslash_trans(*regparse++); | |
2513 regc(startc); | |
2514 } | |
2515 } | |
2516 else if (*regparse == '[') | |
2517 { | |
2518 int c_class; | |
2519 int cu; | |
2520 | |
167 | 2521 c_class = get_char_class(®parse); |
7 | 2522 startc = -1; |
2523 /* Characters assumed to be 8 bits! */ | |
2524 switch (c_class) | |
2525 { | |
2526 case CLASS_NONE: | |
167 | 2527 c_class = get_equi_class(®parse); |
2528 if (c_class != 0) | |
2529 { | |
2530 /* produce equivalence class */ | |
2531 reg_equi_class(c_class); | |
2532 } | |
2533 else if ((c_class = | |
2534 get_coll_element(®parse)) != 0) | |
2535 { | |
2536 /* produce a collating element */ | |
2537 regmbc(c_class); | |
2538 } | |
2539 else | |
2540 { | |
2541 /* literal '[', allow [[-x] as a range */ | |
2542 startc = *regparse++; | |
2543 regc(startc); | |
2544 } | |
7 | 2545 break; |
2546 case CLASS_ALNUM: | |
2547 for (cu = 1; cu <= 255; cu++) | |
2548 if (isalnum(cu)) | |
2549 regc(cu); | |
2550 break; | |
2551 case CLASS_ALPHA: | |
2552 for (cu = 1; cu <= 255; cu++) | |
2553 if (isalpha(cu)) | |
2554 regc(cu); | |
2555 break; | |
2556 case CLASS_BLANK: | |
2557 regc(' '); | |
2558 regc('\t'); | |
2559 break; | |
2560 case CLASS_CNTRL: | |
2561 for (cu = 1; cu <= 255; cu++) | |
2562 if (iscntrl(cu)) | |
2563 regc(cu); | |
2564 break; | |
2565 case CLASS_DIGIT: | |
2566 for (cu = 1; cu <= 255; cu++) | |
2567 if (VIM_ISDIGIT(cu)) | |
2568 regc(cu); | |
2569 break; | |
2570 case CLASS_GRAPH: | |
2571 for (cu = 1; cu <= 255; cu++) | |
2572 if (isgraph(cu)) | |
2573 regc(cu); | |
2574 break; | |
2575 case CLASS_LOWER: | |
2576 for (cu = 1; cu <= 255; cu++) | |
1347 | 2577 if (MB_ISLOWER(cu)) |
7 | 2578 regc(cu); |
2579 break; | |
2580 case CLASS_PRINT: | |
2581 for (cu = 1; cu <= 255; cu++) | |
2582 if (vim_isprintc(cu)) | |
2583 regc(cu); | |
2584 break; | |
2585 case CLASS_PUNCT: | |
2586 for (cu = 1; cu <= 255; cu++) | |
2587 if (ispunct(cu)) | |
2588 regc(cu); | |
2589 break; | |
2590 case CLASS_SPACE: | |
2591 for (cu = 9; cu <= 13; cu++) | |
2592 regc(cu); | |
2593 regc(' '); | |
2594 break; | |
2595 case CLASS_UPPER: | |
2596 for (cu = 1; cu <= 255; cu++) | |
1347 | 2597 if (MB_ISUPPER(cu)) |
7 | 2598 regc(cu); |
2599 break; | |
2600 case CLASS_XDIGIT: | |
2601 for (cu = 1; cu <= 255; cu++) | |
2602 if (vim_isxdigit(cu)) | |
2603 regc(cu); | |
2604 break; | |
2605 case CLASS_TAB: | |
2606 regc('\t'); | |
2607 break; | |
2608 case CLASS_RETURN: | |
2609 regc('\r'); | |
2610 break; | |
2611 case CLASS_BACKSPACE: | |
2612 regc('\b'); | |
2613 break; | |
2614 case CLASS_ESCAPE: | |
2615 regc('\033'); | |
2616 break; | |
2617 } | |
2618 } | |
2619 else | |
2620 { | |
2621 #ifdef FEAT_MBYTE | |
2622 if (has_mbyte) | |
2623 { | |
2624 int len; | |
2625 | |
2626 /* produce a multibyte character, including any | |
2627 * following composing characters */ | |
2628 startc = mb_ptr2char(regparse); | |
474 | 2629 len = (*mb_ptr2len)(regparse); |
7 | 2630 if (enc_utf8 && utf_char2len(startc) != len) |
2631 startc = -1; /* composing chars */ | |
2632 while (--len >= 0) | |
2633 regc(*regparse++); | |
2634 } | |
2635 else | |
2636 #endif | |
2637 { | |
2638 startc = *regparse++; | |
2639 regc(startc); | |
2640 } | |
2641 } | |
2642 } | |
2643 regc(NUL); | |
2644 prevchr_len = 1; /* last char was the ']' */ | |
2645 if (*regparse != ']') | |
2646 EMSG_RET_NULL(_(e_toomsbra)); /* Cannot happen? */ | |
2647 skipchr(); /* let's be friends with the lexer again */ | |
2648 *flagp |= HASWIDTH | SIMPLE; | |
2649 break; | |
2650 } | |
481 | 2651 else if (reg_strict) |
4444 | 2652 EMSG2_RET_NULL(_(e_missingbracket), reg_magic > MAGIC_OFF); |
7 | 2653 } |
2654 /* FALLTHROUGH */ | |
2655 | |
2656 default: | |
2657 { | |
2658 int len; | |
2659 | |
2660 #ifdef FEAT_MBYTE | |
2661 /* A multi-byte character is handled as a separate atom if it's | |
714 | 2662 * before a multi and when it's a composing char. */ |
2663 if (use_multibytecode(c)) | |
7 | 2664 { |
714 | 2665 do_multibyte: |
7 | 2666 ret = regnode(MULTIBYTECODE); |
2667 regmbc(c); | |
2668 *flagp |= HASWIDTH | SIMPLE; | |
2669 break; | |
2670 } | |
2671 #endif | |
2672 | |
2673 ret = regnode(EXACTLY); | |
2674 | |
2675 /* | |
2676 * Append characters as long as: | |
2677 * - there is no following multi, we then need the character in | |
2678 * front of it as a single character operand | |
2679 * - not running into a Magic character | |
2680 * - "one_exactly" is not set | |
2681 * But always emit at least one character. Might be a Multi, | |
2682 * e.g., a "[" without matching "]". | |
2683 */ | |
2684 for (len = 0; c != NUL && (len == 0 | |
2685 || (re_multi_type(peekchr()) == NOT_MULTI | |
2686 && !one_exactly | |
2687 && !is_Magic(c))); ++len) | |
2688 { | |
2689 c = no_Magic(c); | |
2690 #ifdef FEAT_MBYTE | |
2691 if (has_mbyte) | |
2692 { | |
2693 regmbc(c); | |
2694 if (enc_utf8) | |
2695 { | |
2696 int l; | |
2697 | |
714 | 2698 /* Need to get composing character too. */ |
7 | 2699 for (;;) |
2700 { | |
714 | 2701 l = utf_ptr2len(regparse); |
2702 if (!UTF_COMPOSINGLIKE(regparse, regparse + l)) | |
7 | 2703 break; |
714 | 2704 regmbc(utf_ptr2char(regparse)); |
2705 skipchr(); | |
7 | 2706 } |
2707 } | |
2708 } | |
2709 else | |
2710 #endif | |
2711 regc(c); | |
2712 c = getchr(); | |
2713 } | |
2714 ungetchr(); | |
2715 | |
2716 regc(NUL); | |
2717 *flagp |= HASWIDTH; | |
2718 if (len == 1) | |
2719 *flagp |= SIMPLE; | |
2720 } | |
2721 break; | |
2722 } | |
2723 | |
2724 return ret; | |
2725 } | |
2726 | |
714 | 2727 #ifdef FEAT_MBYTE |
2728 /* | |
2729 * Return TRUE if MULTIBYTECODE should be used instead of EXACTLY for | |
2730 * character "c". | |
2731 */ | |
2732 static int | |
2733 use_multibytecode(c) | |
2734 int c; | |
2735 { | |
2736 return has_mbyte && (*mb_char2len)(c) > 1 | |
2737 && (re_multi_type(peekchr()) != NOT_MULTI | |
2738 || (enc_utf8 && utf_iscomposing(c))); | |
2739 } | |
2740 #endif | |
2741 | |
7 | 2742 /* |
4444 | 2743 * Emit a node. |
7 | 2744 * Return pointer to generated code. |
2745 */ | |
2746 static char_u * | |
2747 regnode(op) | |
2748 int op; | |
2749 { | |
2750 char_u *ret; | |
2751 | |
2752 ret = regcode; | |
2753 if (ret == JUST_CALC_SIZE) | |
2754 regsize += 3; | |
2755 else | |
2756 { | |
2757 *regcode++ = op; | |
2758 *regcode++ = NUL; /* Null "next" pointer. */ | |
2759 *regcode++ = NUL; | |
2760 } | |
2761 return ret; | |
2762 } | |
2763 | |
2764 /* | |
2765 * Emit (if appropriate) a byte of code | |
2766 */ | |
2767 static void | |
2768 regc(b) | |
2769 int b; | |
2770 { | |
2771 if (regcode == JUST_CALC_SIZE) | |
2772 regsize++; | |
2773 else | |
2774 *regcode++ = b; | |
2775 } | |
2776 | |
2777 #ifdef FEAT_MBYTE | |
2778 /* | |
2779 * Emit (if appropriate) a multi-byte character of code | |
2780 */ | |
2781 static void | |
2782 regmbc(c) | |
2783 int c; | |
2784 { | |
2974 | 2785 if (!has_mbyte && c > 0xff) |
2786 return; | |
7 | 2787 if (regcode == JUST_CALC_SIZE) |
2788 regsize += (*mb_char2len)(c); | |
2789 else | |
2790 regcode += (*mb_char2bytes)(c, regcode); | |
2791 } | |
2792 #endif | |
2793 | |
2794 /* | |
4444 | 2795 * Insert an operator in front of already-emitted operand |
7 | 2796 * |
2797 * Means relocating the operand. | |
2798 */ | |
2799 static void | |
2800 reginsert(op, opnd) | |
2801 int op; | |
2802 char_u *opnd; | |
2803 { | |
2804 char_u *src; | |
2805 char_u *dst; | |
2806 char_u *place; | |
2807 | |
2808 if (regcode == JUST_CALC_SIZE) | |
2809 { | |
2810 regsize += 3; | |
2811 return; | |
2812 } | |
2813 src = regcode; | |
2814 regcode += 3; | |
2815 dst = regcode; | |
2816 while (src > opnd) | |
2817 *--dst = *--src; | |
2818 | |
2819 place = opnd; /* Op node, where operand used to be. */ | |
2820 *place++ = op; | |
2821 *place++ = NUL; | |
2822 *place = NUL; | |
2823 } | |
2824 | |
2825 /* | |
4444 | 2826 * 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
|
2827 * Add a number to the operator. |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2828 */ |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2829 static void |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2830 reginsert_nr(op, val, opnd) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2831 int op; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2832 long val; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2833 char_u *opnd; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2834 { |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2835 char_u *src; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2836 char_u *dst; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2837 char_u *place; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2838 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2839 if (regcode == JUST_CALC_SIZE) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2840 { |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2841 regsize += 7; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2842 return; |
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 src = regcode; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2845 regcode += 7; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2846 dst = regcode; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2847 while (src > opnd) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2848 *--dst = *--src; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2849 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2850 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
|
2851 *place++ = op; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2852 *place++ = NUL; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2853 *place++ = NUL; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2854 place = re_put_long(place, (long_u)val); |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2855 } |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2856 |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2857 /* |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2858 * Insert an operator in front of already-emitted operand. |
7 | 2859 * The operator has the given limit values as operands. Also set next pointer. |
2860 * | |
2861 * Means relocating the operand. | |
2862 */ | |
2863 static void | |
2864 reginsert_limits(op, minval, maxval, opnd) | |
2865 int op; | |
2866 long minval; | |
2867 long maxval; | |
2868 char_u *opnd; | |
2869 { | |
2870 char_u *src; | |
2871 char_u *dst; | |
2872 char_u *place; | |
2873 | |
2874 if (regcode == JUST_CALC_SIZE) | |
2875 { | |
2876 regsize += 11; | |
2877 return; | |
2878 } | |
2879 src = regcode; | |
2880 regcode += 11; | |
2881 dst = regcode; | |
2882 while (src > opnd) | |
2883 *--dst = *--src; | |
2884 | |
2885 place = opnd; /* Op node, where operand used to be. */ | |
2886 *place++ = op; | |
2887 *place++ = NUL; | |
2888 *place++ = NUL; | |
2889 place = re_put_long(place, (long_u)minval); | |
2890 place = re_put_long(place, (long_u)maxval); | |
2891 regtail(opnd, place); | |
2892 } | |
2893 | |
2894 /* | |
2895 * Write a long as four bytes at "p" and return pointer to the next char. | |
2896 */ | |
2897 static char_u * | |
2898 re_put_long(p, val) | |
2899 char_u *p; | |
2900 long_u val; | |
2901 { | |
2902 *p++ = (char_u) ((val >> 24) & 0377); | |
2903 *p++ = (char_u) ((val >> 16) & 0377); | |
2904 *p++ = (char_u) ((val >> 8) & 0377); | |
2905 *p++ = (char_u) (val & 0377); | |
2906 return p; | |
2907 } | |
2908 | |
2909 /* | |
4444 | 2910 * Set the next-pointer at the end of a node chain. |
7 | 2911 */ |
2912 static void | |
2913 regtail(p, val) | |
2914 char_u *p; | |
2915 char_u *val; | |
2916 { | |
2917 char_u *scan; | |
2918 char_u *temp; | |
2919 int offset; | |
2920 | |
2921 if (p == JUST_CALC_SIZE) | |
2922 return; | |
2923 | |
2924 /* Find last node. */ | |
2925 scan = p; | |
2926 for (;;) | |
2927 { | |
2928 temp = regnext(scan); | |
2929 if (temp == NULL) | |
2930 break; | |
2931 scan = temp; | |
2932 } | |
2933 | |
233 | 2934 if (OP(scan) == BACK) |
7 | 2935 offset = (int)(scan - val); |
2936 else | |
2937 offset = (int)(val - scan); | |
2010 | 2938 /* When the offset uses more than 16 bits it can no longer fit in the two |
2974 | 2939 * bytes available. Use a global flag to avoid having to check return |
2010 | 2940 * values in too many places. */ |
2941 if (offset > 0xffff) | |
2942 reg_toolong = TRUE; | |
2943 else | |
2944 { | |
2945 *(scan + 1) = (char_u) (((unsigned)offset >> 8) & 0377); | |
2946 *(scan + 2) = (char_u) (offset & 0377); | |
2947 } | |
7 | 2948 } |
2949 | |
2950 /* | |
4444 | 2951 * Like regtail, on item after a BRANCH; nop if none. |
7 | 2952 */ |
2953 static void | |
2954 regoptail(p, val) | |
2955 char_u *p; | |
2956 char_u *val; | |
2957 { | |
2958 /* When op is neither BRANCH nor BRACE_COMPLEX0-9, it is "operandless" */ | |
2959 if (p == NULL || p == JUST_CALC_SIZE | |
2960 || (OP(p) != BRANCH | |
2961 && (OP(p) < BRACE_COMPLEX || OP(p) > BRACE_COMPLEX + 9))) | |
2962 return; | |
2963 regtail(OPERAND(p), val); | |
2964 } | |
2965 | |
2966 /* | |
4444 | 2967 * Functions for getting characters from the regexp input. |
7 | 2968 */ |
2969 | |
2970 static int at_start; /* True when on the first character */ | |
2971 static int prev_at_start; /* True when on the second character */ | |
2972 | |
4444 | 2973 /* |
2974 * Start parsing at "str". | |
2975 */ | |
7 | 2976 static void |
2977 initchr(str) | |
2978 char_u *str; | |
2979 { | |
2980 regparse = str; | |
2981 prevchr_len = 0; | |
2982 curchr = prevprevchr = prevchr = nextchr = -1; | |
2983 at_start = TRUE; | |
2984 prev_at_start = FALSE; | |
2985 } | |
2986 | |
4444 | 2987 /* |
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2988 * 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
|
2989 * starts in the same state again. |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2990 */ |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2991 static void |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2992 save_parse_state(ps) |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2993 parse_state_T *ps; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2994 { |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2995 ps->regparse = regparse; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2996 ps->prevchr_len = prevchr_len; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2997 ps->curchr = curchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2998 ps->prevchr = prevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2999 ps->prevprevchr = prevprevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3000 ps->nextchr = nextchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3001 ps->at_start = at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3002 ps->prev_at_start = prev_at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3003 ps->regnpar = regnpar; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3004 } |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3005 |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3006 /* |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3007 * Restore a previously saved parse state. |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3008 */ |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3009 static void |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3010 restore_parse_state(ps) |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3011 parse_state_T *ps; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3012 { |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3013 regparse = ps->regparse; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3014 prevchr_len = ps->prevchr_len; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3015 curchr = ps->curchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3016 prevchr = ps->prevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3017 prevprevchr = ps->prevprevchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3018 nextchr = ps->nextchr; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3019 at_start = ps->at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3020 prev_at_start = ps->prev_at_start; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3021 regnpar = ps->regnpar; |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3022 } |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3023 |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3024 |
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3025 /* |
4444 | 3026 * Get the next character without advancing. |
3027 */ | |
7 | 3028 static int |
3029 peekchr() | |
3030 { | |
167 | 3031 static int after_slash = FALSE; |
3032 | |
7 | 3033 if (curchr == -1) |
3034 { | |
3035 switch (curchr = regparse[0]) | |
3036 { | |
3037 case '.': | |
3038 case '[': | |
3039 case '~': | |
3040 /* magic when 'magic' is on */ | |
3041 if (reg_magic >= MAGIC_ON) | |
3042 curchr = Magic(curchr); | |
3043 break; | |
3044 case '(': | |
3045 case ')': | |
3046 case '{': | |
3047 case '%': | |
3048 case '+': | |
3049 case '=': | |
3050 case '?': | |
3051 case '@': | |
3052 case '!': | |
3053 case '&': | |
3054 case '|': | |
3055 case '<': | |
3056 case '>': | |
3057 case '#': /* future ext. */ | |
3058 case '"': /* future ext. */ | |
3059 case '\'': /* future ext. */ | |
3060 case ',': /* future ext. */ | |
3061 case '-': /* future ext. */ | |
3062 case ':': /* future ext. */ | |
3063 case ';': /* future ext. */ | |
3064 case '`': /* future ext. */ | |
3065 case '/': /* Can't be used in / command */ | |
3066 /* magic only after "\v" */ | |
3067 if (reg_magic == MAGIC_ALL) | |
3068 curchr = Magic(curchr); | |
3069 break; | |
3070 case '*': | |
167 | 3071 /* * is not magic as the very first character, eg "?*ptr", when |
3072 * after '^', eg "/^*ptr" and when after "\(", "\|", "\&". But | |
3073 * "\(\*" is not magic, thus must be magic if "after_slash" */ | |
3074 if (reg_magic >= MAGIC_ON | |
3075 && !at_start | |
3076 && !(prev_at_start && prevchr == Magic('^')) | |
3077 && (after_slash | |
3078 || (prevchr != Magic('(') | |
3079 && prevchr != Magic('&') | |
3080 && prevchr != Magic('|')))) | |
7 | 3081 curchr = Magic('*'); |
3082 break; | |
3083 case '^': | |
3084 /* '^' is only magic as the very first character and if it's after | |
3085 * "\(", "\|", "\&' or "\n" */ | |
3086 if (reg_magic >= MAGIC_OFF | |
3087 && (at_start | |
3088 || reg_magic == MAGIC_ALL | |
3089 || prevchr == Magic('(') | |
3090 || prevchr == Magic('|') | |
3091 || prevchr == Magic('&') | |
3092 || prevchr == Magic('n') | |
3093 || (no_Magic(prevchr) == '(' | |
3094 && prevprevchr == Magic('%')))) | |
3095 { | |
3096 curchr = Magic('^'); | |
3097 at_start = TRUE; | |
3098 prev_at_start = FALSE; | |
3099 } | |
3100 break; | |
3101 case '$': | |
3102 /* '$' is only magic as the very last char and if it's in front of | |
3103 * either "\|", "\)", "\&", or "\n" */ | |
3104 if (reg_magic >= MAGIC_OFF) | |
3105 { | |
3106 char_u *p = regparse + 1; | |
3107 | |
3108 /* ignore \c \C \m and \M after '$' */ | |
3109 while (p[0] == '\\' && (p[1] == 'c' || p[1] == 'C' | |
3110 || p[1] == 'm' || p[1] == 'M' || p[1] == 'Z')) | |
3111 p += 2; | |
3112 if (p[0] == NUL | |
3113 || (p[0] == '\\' | |
3114 && (p[1] == '|' || p[1] == '&' || p[1] == ')' | |
3115 || p[1] == 'n')) | |
3116 || reg_magic == MAGIC_ALL) | |
3117 curchr = Magic('$'); | |
3118 } | |
3119 break; | |
3120 case '\\': | |
3121 { | |
3122 int c = regparse[1]; | |
3123 | |
3124 if (c == NUL) | |
3125 curchr = '\\'; /* trailing '\' */ | |
3126 else if ( | |
3127 #ifdef EBCDIC | |
3128 vim_strchr(META, c) | |
3129 #else | |
3130 c <= '~' && META_flags[c] | |
3131 #endif | |
3132 ) | |
3133 { | |
3134 /* | |
3135 * META contains everything that may be magic sometimes, | |
3136 * except ^ and $ ("\^" and "\$" are only magic after | |
3137 * "\v"). We now fetch the next character and toggle its | |
3138 * magicness. Therefore, \ is so meta-magic that it is | |
3139 * not in META. | |
3140 */ | |
3141 curchr = -1; | |
3142 prev_at_start = at_start; | |
3143 at_start = FALSE; /* be able to say "/\*ptr" */ | |
3144 ++regparse; | |
167 | 3145 ++after_slash; |
7 | 3146 peekchr(); |
3147 --regparse; | |
167 | 3148 --after_slash; |
7 | 3149 curchr = toggle_Magic(curchr); |
3150 } | |
3151 else if (vim_strchr(REGEXP_ABBR, c)) | |
3152 { | |
3153 /* | |
3154 * Handle abbreviations, like "\t" for TAB -- webb | |
3155 */ | |
3156 curchr = backslash_trans(c); | |
3157 } | |
3158 else if (reg_magic == MAGIC_NONE && (c == '$' || c == '^')) | |
3159 curchr = toggle_Magic(c); | |
3160 else | |
3161 { | |
3162 /* | |
3163 * Next character can never be (made) magic? | |
3164 * Then backslashing it won't do anything. | |
3165 */ | |
3166 #ifdef FEAT_MBYTE | |
3167 if (has_mbyte) | |
3168 curchr = (*mb_ptr2char)(regparse + 1); | |
3169 else | |
3170 #endif | |
3171 curchr = c; | |
3172 } | |
3173 break; | |
3174 } | |
3175 | |
3176 #ifdef FEAT_MBYTE | |
3177 default: | |
3178 if (has_mbyte) | |
3179 curchr = (*mb_ptr2char)(regparse); | |
3180 #endif | |
3181 } | |
3182 } | |
3183 | |
3184 return curchr; | |
3185 } | |
3186 | |
3187 /* | |
3188 * Eat one lexed character. Do this in a way that we can undo it. | |
3189 */ | |
3190 static void | |
3191 skipchr() | |
3192 { | |
3193 /* peekchr() eats a backslash, do the same here */ | |
3194 if (*regparse == '\\') | |
3195 prevchr_len = 1; | |
3196 else | |
3197 prevchr_len = 0; | |
3198 if (regparse[prevchr_len] != NUL) | |
3199 { | |
3200 #ifdef FEAT_MBYTE | |
714 | 3201 if (enc_utf8) |
1449 | 3202 /* exclude composing chars that mb_ptr2len does include */ |
3203 prevchr_len += utf_ptr2len(regparse + prevchr_len); | |
714 | 3204 else if (has_mbyte) |
474 | 3205 prevchr_len += (*mb_ptr2len)(regparse + prevchr_len); |
7 | 3206 else |
3207 #endif | |
3208 ++prevchr_len; | |
3209 } | |
3210 regparse += prevchr_len; | |
3211 prev_at_start = at_start; | |
3212 at_start = FALSE; | |
3213 prevprevchr = prevchr; | |
3214 prevchr = curchr; | |
3215 curchr = nextchr; /* use previously unget char, or -1 */ | |
3216 nextchr = -1; | |
3217 } | |
3218 | |
3219 /* | |
3220 * Skip a character while keeping the value of prev_at_start for at_start. | |
3221 * prevchr and prevprevchr are also kept. | |
3222 */ | |
3223 static void | |
3224 skipchr_keepstart() | |
3225 { | |
3226 int as = prev_at_start; | |
3227 int pr = prevchr; | |
3228 int prpr = prevprevchr; | |
3229 | |
3230 skipchr(); | |
3231 at_start = as; | |
3232 prevchr = pr; | |
3233 prevprevchr = prpr; | |
3234 } | |
3235 | |
4444 | 3236 /* |
3237 * Get the next character from the pattern. We know about magic and such, so | |
3238 * therefore we need a lexical analyzer. | |
3239 */ | |
7 | 3240 static int |
3241 getchr() | |
3242 { | |
3243 int chr = peekchr(); | |
3244 | |
3245 skipchr(); | |
3246 return chr; | |
3247 } | |
3248 | |
3249 /* | |
3250 * put character back. Works only once! | |
3251 */ | |
3252 static void | |
3253 ungetchr() | |
3254 { | |
3255 nextchr = curchr; | |
3256 curchr = prevchr; | |
3257 prevchr = prevprevchr; | |
3258 at_start = prev_at_start; | |
3259 prev_at_start = FALSE; | |
3260 | |
3261 /* Backup regparse, so that it's at the same position as before the | |
3262 * getchr(). */ | |
3263 regparse -= prevchr_len; | |
3264 } | |
3265 | |
3266 /* | |
29 | 3267 * Get and return the value of the hex string at the current position. |
3268 * Return -1 if there is no valid hex number. | |
3269 * The position is updated: | |
24 | 3270 * blahblah\%x20asdf |
856 | 3271 * before-^ ^-after |
24 | 3272 * The parameter controls the maximum number of input characters. This will be |
3273 * 2 when reading a \%x20 sequence and 4 when reading a \%u20AC sequence. | |
3274 */ | |
3275 static int | |
3276 gethexchrs(maxinputlen) | |
3277 int maxinputlen; | |
3278 { | |
3279 int nr = 0; | |
3280 int c; | |
3281 int i; | |
3282 | |
3283 for (i = 0; i < maxinputlen; ++i) | |
3284 { | |
3285 c = regparse[0]; | |
3286 if (!vim_isxdigit(c)) | |
3287 break; | |
3288 nr <<= 4; | |
3289 nr |= hex2nr(c); | |
3290 ++regparse; | |
3291 } | |
3292 | |
3293 if (i == 0) | |
3294 return -1; | |
3295 return nr; | |
3296 } | |
3297 | |
3298 /* | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3299 * Get and return the value of the decimal string immediately after the |
24 | 3300 * current position. Return -1 for invalid. Consumes all digits. |
3301 */ | |
3302 static int | |
3303 getdecchrs() | |
3304 { | |
3305 int nr = 0; | |
3306 int c; | |
3307 int i; | |
3308 | |
3309 for (i = 0; ; ++i) | |
3310 { | |
3311 c = regparse[0]; | |
3312 if (c < '0' || c > '9') | |
3313 break; | |
3314 nr *= 10; | |
3315 nr += c - '0'; | |
3316 ++regparse; | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3317 curchr = -1; /* no longer valid */ |
24 | 3318 } |
3319 | |
3320 if (i == 0) | |
3321 return -1; | |
3322 return nr; | |
3323 } | |
3324 | |
3325 /* | |
3326 * get and return the value of the octal string immediately after the current | |
3327 * position. Return -1 for invalid, or 0-255 for valid. Smart enough to handle | |
3328 * numbers > 377 correctly (for example, 400 is treated as 40) and doesn't | |
3329 * treat 8 or 9 as recognised characters. Position is updated: | |
3330 * blahblah\%o210asdf | |
856 | 3331 * before-^ ^-after |
24 | 3332 */ |
3333 static int | |
3334 getoctchrs() | |
3335 { | |
3336 int nr = 0; | |
3337 int c; | |
3338 int i; | |
3339 | |
3340 for (i = 0; i < 3 && nr < 040; ++i) | |
3341 { | |
3342 c = regparse[0]; | |
3343 if (c < '0' || c > '7') | |
3344 break; | |
3345 nr <<= 3; | |
3346 nr |= hex2nr(c); | |
3347 ++regparse; | |
3348 } | |
3349 | |
3350 if (i == 0) | |
3351 return -1; | |
3352 return nr; | |
3353 } | |
3354 | |
3355 /* | |
3356 * Get a number after a backslash that is inside []. | |
3357 * When nothing is recognized return a backslash. | |
3358 */ | |
3359 static int | |
3360 coll_get_char() | |
3361 { | |
3362 int nr = -1; | |
3363 | |
3364 switch (*regparse++) | |
3365 { | |
3366 case 'd': nr = getdecchrs(); break; | |
3367 case 'o': nr = getoctchrs(); break; | |
3368 case 'x': nr = gethexchrs(2); break; | |
3369 case 'u': nr = gethexchrs(4); break; | |
3370 case 'U': nr = gethexchrs(8); break; | |
3371 } | |
3372 if (nr < 0) | |
3373 { | |
3374 /* If getting the number fails be backwards compatible: the character | |
3375 * is a backslash. */ | |
3376 --regparse; | |
3377 nr = '\\'; | |
3378 } | |
3379 return nr; | |
3380 } | |
3381 | |
3382 /* | |
7 | 3383 * read_limits - Read two integers to be taken as a minimum and maximum. |
3384 * If the first character is '-', then the range is reversed. | |
3385 * Should end with 'end'. If minval is missing, zero is default, if maxval is | |
3386 * missing, a very big number is the default. | |
3387 */ | |
3388 static int | |
3389 read_limits(minval, maxval) | |
3390 long *minval; | |
3391 long *maxval; | |
3392 { | |
3393 int reverse = FALSE; | |
3394 char_u *first_char; | |
3395 long tmp; | |
3396 | |
3397 if (*regparse == '-') | |
3398 { | |
3399 /* Starts with '-', so reverse the range later */ | |
3400 regparse++; | |
3401 reverse = TRUE; | |
3402 } | |
3403 first_char = regparse; | |
3404 *minval = getdigits(®parse); | |
3405 if (*regparse == ',') /* There is a comma */ | |
3406 { | |
3407 if (vim_isdigit(*++regparse)) | |
3408 *maxval = getdigits(®parse); | |
3409 else | |
3410 *maxval = MAX_LIMIT; | |
3411 } | |
3412 else if (VIM_ISDIGIT(*first_char)) | |
3413 *maxval = *minval; /* It was \{n} or \{-n} */ | |
3414 else | |
3415 *maxval = MAX_LIMIT; /* It was \{} or \{-} */ | |
3416 if (*regparse == '\\') | |
3417 regparse++; /* Allow either \{...} or \{...\} */ | |
167 | 3418 if (*regparse != '}') |
7 | 3419 { |
3420 sprintf((char *)IObuff, _("E554: Syntax error in %s{...}"), | |
3421 reg_magic == MAGIC_ALL ? "" : "\\"); | |
3422 EMSG_RET_FAIL(IObuff); | |
3423 } | |
3424 | |
3425 /* | |
3426 * Reverse the range if there was a '-', or make sure it is in the right | |
3427 * order otherwise. | |
3428 */ | |
3429 if ((!reverse && *minval > *maxval) || (reverse && *minval < *maxval)) | |
3430 { | |
3431 tmp = *minval; | |
3432 *minval = *maxval; | |
3433 *maxval = tmp; | |
3434 } | |
3435 skipchr(); /* let's be friends with the lexer again */ | |
3436 return OK; | |
3437 } | |
3438 | |
3439 /* | |
3440 * vim_regexec and friends | |
3441 */ | |
3442 | |
3443 /* | |
3444 * Global work variables for vim_regexec(). | |
3445 */ | |
3446 | |
3447 /* The current match-position is remembered with these variables: */ | |
3448 static linenr_T reglnum; /* line number, relative to first line */ | |
3449 static char_u *regline; /* start of current line */ | |
3450 static char_u *reginput; /* current input, points into "regline" */ | |
3451 | |
3452 static int need_clear_subexpr; /* subexpressions still need to be | |
3453 * cleared */ | |
3454 #ifdef FEAT_SYN_HL | |
3455 static int need_clear_zsubexpr = FALSE; /* extmatch subexpressions | |
3456 * still need to be cleared */ | |
3457 #endif | |
3458 | |
3459 /* | |
3460 * Structure used to save the current input state, when it needs to be | |
3461 * restored after trying a match. Used by reg_save() and reg_restore(). | |
233 | 3462 * Also stores the length of "backpos". |
7 | 3463 */ |
3464 typedef struct | |
3465 { | |
3466 union | |
3467 { | |
3468 char_u *ptr; /* reginput pointer, for single-line regexp */ | |
3469 lpos_T pos; /* reginput pos, for multi-line regexp */ | |
3470 } rs_u; | |
233 | 3471 int rs_len; |
7 | 3472 } regsave_T; |
3473 | |
3474 /* struct to save start/end pointer/position in for \(\) */ | |
3475 typedef struct | |
3476 { | |
3477 union | |
3478 { | |
3479 char_u *ptr; | |
3480 lpos_T pos; | |
3481 } se_u; | |
3482 } save_se_T; | |
3483 | |
1579 | 3484 /* used for BEHIND and NOBEHIND matching */ |
3485 typedef struct regbehind_S | |
3486 { | |
3487 regsave_T save_after; | |
3488 regsave_T save_behind; | |
1602 | 3489 int save_need_clear_subexpr; |
1579 | 3490 save_se_T save_start[NSUBEXP]; |
3491 save_se_T save_end[NSUBEXP]; | |
3492 } regbehind_T; | |
3493 | |
7 | 3494 static char_u *reg_getline __ARGS((linenr_T lnum)); |
4444 | 3495 static long bt_regexec_both __ARGS((char_u *line, colnr_T col, proftime_T *tm)); |
3496 static long regtry __ARGS((bt_regprog_T *prog, colnr_T col)); | |
7 | 3497 static void cleanup_subexpr __ARGS((void)); |
3498 #ifdef FEAT_SYN_HL | |
3499 static void cleanup_zsubexpr __ARGS((void)); | |
3500 #endif | |
1579 | 3501 static void save_subexpr __ARGS((regbehind_T *bp)); |
3502 static void restore_subexpr __ARGS((regbehind_T *bp)); | |
7 | 3503 static void reg_nextline __ARGS((void)); |
233 | 3504 static void reg_save __ARGS((regsave_T *save, garray_T *gap)); |
3505 static void reg_restore __ARGS((regsave_T *save, garray_T *gap)); | |
7 | 3506 static int reg_save_equal __ARGS((regsave_T *save)); |
3507 static void save_se_multi __ARGS((save_se_T *savep, lpos_T *posp)); | |
3508 static void save_se_one __ARGS((save_se_T *savep, char_u **pp)); | |
3509 | |
3510 /* Save the sub-expressions before attempting a match. */ | |
3511 #define save_se(savep, posp, pp) \ | |
3512 REG_MULTI ? save_se_multi((savep), (posp)) : save_se_one((savep), (pp)) | |
3513 | |
3514 /* After a failed match restore the sub-expressions. */ | |
3515 #define restore_se(savep, posp, pp) { \ | |
3516 if (REG_MULTI) \ | |
3517 *(posp) = (savep)->se_u.pos; \ | |
3518 else \ | |
3519 *(pp) = (savep)->se_u.ptr; } | |
3520 | |
3521 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
|
3522 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 | 3523 static int regmatch __ARGS((char_u *prog)); |
7 | 3524 static int regrepeat __ARGS((char_u *p, long maxcount)); |
3525 | |
3526 #ifdef DEBUG | |
3527 int regnarrate = 0; | |
3528 #endif | |
3529 | |
3530 /* | |
3531 * Internal copy of 'ignorecase'. It is set at each call to vim_regexec(). | |
3532 * Normally it gets the value of "rm_ic" or "rmm_ic", but when the pattern | |
3533 * contains '\c' or '\C' the value is overruled. | |
3534 */ | |
3535 static int ireg_ic; | |
3536 | |
3537 #ifdef FEAT_MBYTE | |
3538 /* | |
3539 * Similar to ireg_ic, but only for 'combining' characters. Set with \Z flag | |
3540 * in the regexp. Defaults to false, always. | |
3541 */ | |
3542 static int ireg_icombine; | |
3543 #endif | |
3544 | |
3545 /* | |
410 | 3546 * Copy of "rmm_maxcol": maximum column to search for a match. Zero when |
3547 * there is no maximum. | |
3548 */ | |
418 | 3549 static colnr_T ireg_maxcol; |
410 | 3550 |
3551 /* | |
7 | 3552 * Sometimes need to save a copy of a line. Since alloc()/free() is very |
3553 * slow, we keep one allocated piece of memory and only re-allocate it when | |
4444 | 3554 * it's too small. It's freed in bt_regexec_both() when finished. |
7 | 3555 */ |
1468 | 3556 static char_u *reg_tofree = NULL; |
7 | 3557 static unsigned reg_tofreelen; |
3558 | |
3559 /* | |
3560 * These variables are set when executing a regexp to speed up the execution. | |
1209 | 3561 * Which ones are set depends on whether a single-line or multi-line match is |
7 | 3562 * done: |
3563 * single-line multi-line | |
3564 * reg_match ®match_T NULL | |
3565 * reg_mmatch NULL ®mmatch_T | |
3566 * reg_startp reg_match->startp <invalid> | |
3567 * reg_endp reg_match->endp <invalid> | |
3568 * reg_startpos <invalid> reg_mmatch->startpos | |
3569 * reg_endpos <invalid> reg_mmatch->endpos | |
3570 * reg_win NULL window in which to search | |
4061 | 3571 * reg_buf curbuf buffer in which to search |
7 | 3572 * reg_firstlnum <invalid> first line in which to search |
3573 * reg_maxline 0 last line nr | |
3574 * reg_line_lbr FALSE or TRUE FALSE | |
3575 */ | |
3576 static regmatch_T *reg_match; | |
3577 static regmmatch_T *reg_mmatch; | |
3578 static char_u **reg_startp = NULL; | |
3579 static char_u **reg_endp = NULL; | |
3580 static lpos_T *reg_startpos = NULL; | |
3581 static lpos_T *reg_endpos = NULL; | |
3582 static win_T *reg_win; | |
3583 static buf_T *reg_buf; | |
3584 static linenr_T reg_firstlnum; | |
3585 static linenr_T reg_maxline; | |
3586 static int reg_line_lbr; /* "\n" in string is line break */ | |
3587 | |
270 | 3588 /* Values for rs_state in regitem_T. */ |
3589 typedef enum regstate_E | |
3590 { | |
3591 RS_NOPEN = 0 /* NOPEN and NCLOSE */ | |
3592 , RS_MOPEN /* MOPEN + [0-9] */ | |
3593 , RS_MCLOSE /* MCLOSE + [0-9] */ | |
3594 #ifdef FEAT_SYN_HL | |
3595 , RS_ZOPEN /* ZOPEN + [0-9] */ | |
3596 , RS_ZCLOSE /* ZCLOSE + [0-9] */ | |
3597 #endif | |
3598 , RS_BRANCH /* BRANCH */ | |
3599 , RS_BRCPLX_MORE /* BRACE_COMPLEX and trying one more match */ | |
3600 , RS_BRCPLX_LONG /* BRACE_COMPLEX and trying longest match */ | |
3601 , RS_BRCPLX_SHORT /* BRACE_COMPLEX and trying shortest match */ | |
3602 , RS_NOMATCH /* NOMATCH */ | |
3603 , RS_BEHIND1 /* BEHIND / NOBEHIND matching rest */ | |
3604 , RS_BEHIND2 /* BEHIND / NOBEHIND matching behind part */ | |
3605 , RS_STAR_LONG /* STAR/PLUS/BRACE_SIMPLE longest match */ | |
3606 , RS_STAR_SHORT /* STAR/PLUS/BRACE_SIMPLE shortest match */ | |
3607 } regstate_T; | |
3608 | |
3609 /* | |
3610 * When there are alternatives a regstate_T is put on the regstack to remember | |
3611 * what we are doing. | |
3612 * Before it may be another type of item, depending on rs_state, to remember | |
3613 * more things. | |
3614 */ | |
3615 typedef struct regitem_S | |
3616 { | |
3617 regstate_T rs_state; /* what we are doing, one of RS_ above */ | |
3618 char_u *rs_scan; /* current node in program */ | |
3619 union | |
3620 { | |
3621 save_se_T sesave; | |
3622 regsave_T regsave; | |
3623 } rs_un; /* room for saving reginput */ | |
1579 | 3624 short rs_no; /* submatch nr or BEHIND/NOBEHIND */ |
270 | 3625 } regitem_T; |
3626 | |
3627 static regitem_T *regstack_push __ARGS((regstate_T state, char_u *scan)); | |
3628 static void regstack_pop __ARGS((char_u **scan)); | |
3629 | |
3630 /* used for STAR, PLUS and BRACE_SIMPLE matching */ | |
3631 typedef struct regstar_S | |
3632 { | |
3633 int nextb; /* next byte */ | |
3634 int nextb_ic; /* next byte reverse case */ | |
3635 long count; | |
3636 long minval; | |
3637 long maxval; | |
3638 } regstar_T; | |
3639 | |
3640 /* used to store input position when a BACK was encountered, so that we now if | |
3641 * we made any progress since the last time. */ | |
3642 typedef struct backpos_S | |
3643 { | |
3644 char_u *bp_scan; /* "scan" where BACK was encountered */ | |
3645 regsave_T bp_pos; /* last input position */ | |
3646 } backpos_T; | |
3647 | |
3648 /* | |
1520 | 3649 * "regstack" and "backpos" are used by regmatch(). They are kept over calls |
3650 * to avoid invoking malloc() and free() often. | |
3651 * "regstack" is a stack with regitem_T items, sometimes preceded by regstar_T | |
3652 * or regbehind_T. | |
3653 * "backpos_T" is a table with backpos_T for BACK | |
270 | 3654 */ |
1520 | 3655 static garray_T regstack = {0, 0, 0, 0, NULL}; |
3656 static garray_T backpos = {0, 0, 0, 0, NULL}; | |
3657 | |
3658 /* | |
3659 * Both for regstack and backpos tables we use the following strategy of | |
3660 * allocation (to reduce malloc/free calls): | |
3661 * - Initial size is fairly small. | |
3662 * - When needed, the tables are grown bigger (8 times at first, double after | |
3663 * that). | |
3664 * - After executing the match we free the memory only if the array has grown. | |
3665 * Thus the memory is kept allocated when it's at the initial size. | |
3666 * This makes it fast while not keeping a lot of memory allocated. | |
3667 * A three times speed increase was observed when using many simple patterns. | |
3668 */ | |
3669 #define REGSTACK_INITIAL 2048 | |
3670 #define BACKPOS_INITIAL 64 | |
3671 | |
3672 #if defined(EXITFREE) || defined(PROTO) | |
3673 void | |
3674 free_regexp_stuff() | |
3675 { | |
3676 ga_clear(®stack); | |
3677 ga_clear(&backpos); | |
3678 vim_free(reg_tofree); | |
3679 vim_free(reg_prev_sub); | |
3680 } | |
3681 #endif | |
270 | 3682 |
7 | 3683 /* |
3684 * Get pointer to the line "lnum", which is relative to "reg_firstlnum". | |
3685 */ | |
3686 static char_u * | |
3687 reg_getline(lnum) | |
3688 linenr_T lnum; | |
3689 { | |
3690 /* when looking behind for a match/no-match lnum is negative. But we | |
3691 * can't go before line 1 */ | |
3692 if (reg_firstlnum + lnum < 1) | |
3693 return NULL; | |
482 | 3694 if (lnum > reg_maxline) |
481 | 3695 /* Must have matched the "\n" in the last line. */ |
3696 return (char_u *)""; | |
7 | 3697 return ml_get_buf(reg_buf, reg_firstlnum + lnum, FALSE); |
3698 } | |
3699 | |
3700 static regsave_T behind_pos; | |
3701 | |
3702 #ifdef FEAT_SYN_HL | |
3703 static char_u *reg_startzp[NSUBEXP]; /* Workspace to mark beginning */ | |
3704 static char_u *reg_endzp[NSUBEXP]; /* and end of \z(...\) matches */ | |
3705 static lpos_T reg_startzpos[NSUBEXP]; /* idem, beginning pos */ | |
3706 static lpos_T reg_endzpos[NSUBEXP]; /* idem, end pos */ | |
3707 #endif | |
3708 | |
3709 /* TRUE if using multi-line regexp. */ | |
3710 #define REG_MULTI (reg_match == NULL) | |
3711 | |
5838 | 3712 static int bt_regexec_nl __ARGS((regmatch_T *rmp, char_u *line, colnr_T col, int line_lbr)); |
3713 | |
4444 | 3714 |
7 | 3715 /* |
3716 * Match a regexp against a string. | |
3717 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
3718 * Uses curbuf for line count and 'iskeyword'. | |
5838 | 3719 * if "line_lbr" is TRUE consider a "\n" in "line" to be a line break. |
7 | 3720 * |
3721 * Return TRUE if there is a match, FALSE if not. | |
3722 */ | |
4444 | 3723 static int |
5838 | 3724 bt_regexec_nl(rmp, line, col, line_lbr) |
7 | 3725 regmatch_T *rmp; |
3726 char_u *line; /* string to match against */ | |
3727 colnr_T col; /* column to start looking for match */ | |
5838 | 3728 int line_lbr; |
7 | 3729 { |
3730 reg_match = rmp; | |
3731 reg_mmatch = NULL; | |
3732 reg_maxline = 0; | |
5838 | 3733 reg_line_lbr = line_lbr; |
4061 | 3734 reg_buf = curbuf; |
7 | 3735 reg_win = NULL; |
3736 ireg_ic = rmp->rm_ic; | |
3737 #ifdef FEAT_MBYTE | |
3738 ireg_icombine = FALSE; | |
3739 #endif | |
410 | 3740 ireg_maxcol = 0; |
4444 | 3741 return (bt_regexec_both(line, col, NULL) != 0); |
7 | 3742 } |
3743 | |
4444 | 3744 static long bt_regexec_multi __ARGS((regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm)); |
3745 | |
7 | 3746 /* |
3747 * Match a regexp against multiple lines. | |
3748 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
3749 * Uses curbuf for line count and 'iskeyword'. | |
3750 * | |
3751 * Return zero if there is no match. Return number of lines contained in the | |
3752 * match otherwise. | |
3753 */ | |
4444 | 3754 static long |
3755 bt_regexec_multi(rmp, win, buf, lnum, col, tm) | |
7 | 3756 regmmatch_T *rmp; |
3757 win_T *win; /* window in which to search or NULL */ | |
3758 buf_T *buf; /* buffer in which to search */ | |
3759 linenr_T lnum; /* nr of line to start looking for match */ | |
3760 colnr_T col; /* column to start looking for match */ | |
1521 | 3761 proftime_T *tm; /* timeout limit or NULL */ |
7 | 3762 { |
3763 long r; | |
3764 | |
3765 reg_match = NULL; | |
3766 reg_mmatch = rmp; | |
3767 reg_buf = buf; | |
3768 reg_win = win; | |
3769 reg_firstlnum = lnum; | |
3770 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
3771 reg_line_lbr = FALSE; | |
3772 ireg_ic = rmp->rmm_ic; | |
3773 #ifdef FEAT_MBYTE | |
3774 ireg_icombine = FALSE; | |
3775 #endif | |
410 | 3776 ireg_maxcol = rmp->rmm_maxcol; |
7 | 3777 |
4444 | 3778 r = bt_regexec_both(NULL, col, tm); |
7 | 3779 |
3780 return r; | |
3781 } | |
3782 | |
3783 /* | |
3784 * Match a regexp against a string ("line" points to the string) or multiple | |
3785 * lines ("line" is NULL, use reg_getline()). | |
3786 */ | |
3787 static long | |
4444 | 3788 bt_regexec_both(line, col, tm) |
7 | 3789 char_u *line; |
3790 colnr_T col; /* column to start looking for match */ | |
1877 | 3791 proftime_T *tm UNUSED; /* timeout limit or NULL */ |
7 | 3792 { |
4444 | 3793 bt_regprog_T *prog; |
7 | 3794 char_u *s; |
189 | 3795 long retval = 0L; |
7 | 3796 |
1520 | 3797 /* Create "regstack" and "backpos" if they are not allocated yet. |
3798 * We allocate *_INITIAL amount of bytes first and then set the grow size | |
3799 * to much bigger value to avoid many malloc calls in case of deep regular | |
3800 * expressions. */ | |
3801 if (regstack.ga_data == NULL) | |
3802 { | |
3803 /* Use an item size of 1 byte, since we push different things | |
3804 * onto the regstack. */ | |
3805 ga_init2(®stack, 1, REGSTACK_INITIAL); | |
3806 ga_grow(®stack, REGSTACK_INITIAL); | |
3807 regstack.ga_growsize = REGSTACK_INITIAL * 8; | |
3808 } | |
3809 | |
3810 if (backpos.ga_data == NULL) | |
3811 { | |
3812 ga_init2(&backpos, sizeof(backpos_T), BACKPOS_INITIAL); | |
3813 ga_grow(&backpos, BACKPOS_INITIAL); | |
3814 backpos.ga_growsize = BACKPOS_INITIAL * 8; | |
3815 } | |
270 | 3816 |
7 | 3817 if (REG_MULTI) |
3818 { | |
4444 | 3819 prog = (bt_regprog_T *)reg_mmatch->regprog; |
7 | 3820 line = reg_getline((linenr_T)0); |
3821 reg_startpos = reg_mmatch->startpos; | |
3822 reg_endpos = reg_mmatch->endpos; | |
3823 } | |
3824 else | |
3825 { | |
4444 | 3826 prog = (bt_regprog_T *)reg_match->regprog; |
7 | 3827 reg_startp = reg_match->startp; |
3828 reg_endp = reg_match->endp; | |
3829 } | |
3830 | |
3831 /* Be paranoid... */ | |
3832 if (prog == NULL || line == NULL) | |
3833 { | |
3834 EMSG(_(e_null)); | |
3835 goto theend; | |
3836 } | |
3837 | |
3838 /* Check validity of program. */ | |
3839 if (prog_magic_wrong()) | |
3840 goto theend; | |
3841 | |
410 | 3842 /* If the start column is past the maximum column: no need to try. */ |
3843 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
3844 goto theend; | |
3845 | |
7 | 3846 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ |
3847 if (prog->regflags & RF_ICASE) | |
3848 ireg_ic = TRUE; | |
3849 else if (prog->regflags & RF_NOICASE) | |
3850 ireg_ic = FALSE; | |
3851 | |
3852 #ifdef FEAT_MBYTE | |
3853 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
3854 if (prog->regflags & RF_ICOMBINE) | |
3855 ireg_icombine = TRUE; | |
3856 #endif | |
3857 | |
3858 /* If there is a "must appear" string, look for it. */ | |
3859 if (prog->regmust != NULL) | |
3860 { | |
3861 int c; | |
3862 | |
3863 #ifdef FEAT_MBYTE | |
3864 if (has_mbyte) | |
3865 c = (*mb_ptr2char)(prog->regmust); | |
3866 else | |
3867 #endif | |
3868 c = *prog->regmust; | |
3869 s = line + col; | |
170 | 3870 |
3871 /* | |
3872 * This is used very often, esp. for ":global". Use three versions of | |
3873 * the loop to avoid overhead of conditions. | |
3874 */ | |
3875 if (!ireg_ic | |
3876 #ifdef FEAT_MBYTE | |
3877 && !has_mbyte | |
3878 #endif | |
3879 ) | |
3880 while ((s = vim_strbyte(s, c)) != NULL) | |
3881 { | |
3882 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
3883 break; /* Found it. */ | |
3884 ++s; | |
3885 } | |
3886 #ifdef FEAT_MBYTE | |
3887 else if (!ireg_ic || (!enc_utf8 && mb_char2len(c) > 1)) | |
3888 while ((s = vim_strchr(s, c)) != NULL) | |
3889 { | |
3890 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
3891 break; /* Found it. */ | |
3892 mb_ptr_adv(s); | |
3893 } | |
3894 #endif | |
3895 else | |
3896 while ((s = cstrchr(s, c)) != NULL) | |
3897 { | |
3898 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
3899 break; /* Found it. */ | |
3900 mb_ptr_adv(s); | |
3901 } | |
7 | 3902 if (s == NULL) /* Not present. */ |
3903 goto theend; | |
3904 } | |
3905 | |
3906 regline = line; | |
3907 reglnum = 0; | |
2578 | 3908 reg_toolong = FALSE; |
7 | 3909 |
3910 /* Simplest case: Anchored match need be tried only once. */ | |
3911 if (prog->reganch) | |
3912 { | |
3913 int c; | |
3914 | |
3915 #ifdef FEAT_MBYTE | |
3916 if (has_mbyte) | |
3917 c = (*mb_ptr2char)(regline + col); | |
3918 else | |
3919 #endif | |
3920 c = regline[col]; | |
3921 if (prog->regstart == NUL | |
3922 || prog->regstart == c | |
3923 || (ireg_ic && (( | |
3924 #ifdef FEAT_MBYTE | |
3925 (enc_utf8 && utf_fold(prog->regstart) == utf_fold(c))) | |
3926 || (c < 255 && prog->regstart < 255 && | |
3927 #endif | |
1347 | 3928 MB_TOLOWER(prog->regstart) == MB_TOLOWER(c))))) |
7 | 3929 retval = regtry(prog, col); |
3930 else | |
3931 retval = 0; | |
3932 } | |
3933 else | |
3934 { | |
1521 | 3935 #ifdef FEAT_RELTIME |
3936 int tm_count = 0; | |
3937 #endif | |
7 | 3938 /* Messy cases: unanchored match. */ |
180 | 3939 while (!got_int) |
7 | 3940 { |
3941 if (prog->regstart != NUL) | |
3942 { | |
170 | 3943 /* Skip until the char we know it must start with. |
3944 * Used often, do some work to avoid call overhead. */ | |
3945 if (!ireg_ic | |
3946 #ifdef FEAT_MBYTE | |
3947 && !has_mbyte | |
3948 #endif | |
3949 ) | |
3950 s = vim_strbyte(regline + col, prog->regstart); | |
3951 else | |
3952 s = cstrchr(regline + col, prog->regstart); | |
7 | 3953 if (s == NULL) |
3954 { | |
3955 retval = 0; | |
3956 break; | |
3957 } | |
3958 col = (int)(s - regline); | |
3959 } | |
3960 | |
410 | 3961 /* Check for maximum column to try. */ |
3962 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
3963 { | |
3964 retval = 0; | |
3965 break; | |
3966 } | |
3967 | |
7 | 3968 retval = regtry(prog, col); |
3969 if (retval > 0) | |
3970 break; | |
3971 | |
3972 /* if not currently on the first line, get it again */ | |
3973 if (reglnum != 0) | |
3974 { | |
481 | 3975 reglnum = 0; |
7 | 3976 regline = reg_getline((linenr_T)0); |
3977 } | |
3978 if (regline[col] == NUL) | |
3979 break; | |
3980 #ifdef FEAT_MBYTE | |
3981 if (has_mbyte) | |
474 | 3982 col += (*mb_ptr2len)(regline + col); |
7 | 3983 else |
3984 #endif | |
3985 ++col; | |
1521 | 3986 #ifdef FEAT_RELTIME |
3987 /* Check for timeout once in a twenty times to avoid overhead. */ | |
3988 if (tm != NULL && ++tm_count == 20) | |
3989 { | |
3990 tm_count = 0; | |
3991 if (profile_passed_limit(tm)) | |
3992 break; | |
3993 } | |
3994 #endif | |
7 | 3995 } |
3996 } | |
3997 | |
3998 theend: | |
1520 | 3999 /* Free "reg_tofree" when it's a bit big. |
4000 * Free regstack and backpos if they are bigger than their initial size. */ | |
4001 if (reg_tofreelen > 400) | |
4002 { | |
4003 vim_free(reg_tofree); | |
4004 reg_tofree = NULL; | |
4005 } | |
4006 if (regstack.ga_maxlen > REGSTACK_INITIAL) | |
4007 ga_clear(®stack); | |
4008 if (backpos.ga_maxlen > BACKPOS_INITIAL) | |
4009 ga_clear(&backpos); | |
270 | 4010 |
7 | 4011 return retval; |
4012 } | |
4013 | |
4014 #ifdef FEAT_SYN_HL | |
4015 static reg_extmatch_T *make_extmatch __ARGS((void)); | |
4016 | |
4017 /* | |
4018 * Create a new extmatch and mark it as referenced once. | |
4019 */ | |
4020 static reg_extmatch_T * | |
4021 make_extmatch() | |
4022 { | |
4023 reg_extmatch_T *em; | |
4024 | |
4025 em = (reg_extmatch_T *)alloc_clear((unsigned)sizeof(reg_extmatch_T)); | |
4026 if (em != NULL) | |
4027 em->refcnt = 1; | |
4028 return em; | |
4029 } | |
4030 | |
4031 /* | |
4032 * Add a reference to an extmatch. | |
4033 */ | |
4034 reg_extmatch_T * | |
4035 ref_extmatch(em) | |
4036 reg_extmatch_T *em; | |
4037 { | |
4038 if (em != NULL) | |
4039 em->refcnt++; | |
4040 return em; | |
4041 } | |
4042 | |
4043 /* | |
4044 * Remove a reference to an extmatch. If there are no references left, free | |
4045 * the info. | |
4046 */ | |
4047 void | |
4048 unref_extmatch(em) | |
4049 reg_extmatch_T *em; | |
4050 { | |
4051 int i; | |
4052 | |
4053 if (em != NULL && --em->refcnt <= 0) | |
4054 { | |
4055 for (i = 0; i < NSUBEXP; ++i) | |
4056 vim_free(em->matches[i]); | |
4057 vim_free(em); | |
4058 } | |
4059 } | |
4060 #endif | |
4061 | |
4062 /* | |
4063 * regtry - try match of "prog" with at regline["col"]. | |
4064 * Returns 0 for failure, number of lines contained in the match otherwise. | |
4065 */ | |
4066 static long | |
4067 regtry(prog, col) | |
4444 | 4068 bt_regprog_T *prog; |
7 | 4069 colnr_T col; |
4070 { | |
4071 reginput = regline + col; | |
4072 need_clear_subexpr = TRUE; | |
4073 #ifdef FEAT_SYN_HL | |
4074 /* Clear the external match subpointers if necessary. */ | |
4075 if (prog->reghasz == REX_SET) | |
4076 need_clear_zsubexpr = TRUE; | |
4077 #endif | |
4078 | |
189 | 4079 if (regmatch(prog->program + 1) == 0) |
4080 return 0; | |
4081 | |
4082 cleanup_subexpr(); | |
4083 if (REG_MULTI) | |
7 | 4084 { |
189 | 4085 if (reg_startpos[0].lnum < 0) |
7 | 4086 { |
189 | 4087 reg_startpos[0].lnum = 0; |
4088 reg_startpos[0].col = col; | |
4089 } | |
4090 if (reg_endpos[0].lnum < 0) | |
4091 { | |
4092 reg_endpos[0].lnum = reglnum; | |
4093 reg_endpos[0].col = (int)(reginput - regline); | |
7 | 4094 } |
4095 else | |
189 | 4096 /* Use line number of "\ze". */ |
4097 reglnum = reg_endpos[0].lnum; | |
4098 } | |
4099 else | |
4100 { | |
4101 if (reg_startp[0] == NULL) | |
4102 reg_startp[0] = regline + col; | |
4103 if (reg_endp[0] == NULL) | |
4104 reg_endp[0] = reginput; | |
4105 } | |
7 | 4106 #ifdef FEAT_SYN_HL |
189 | 4107 /* Package any found \z(...\) matches for export. Default is none. */ |
4108 unref_extmatch(re_extmatch_out); | |
4109 re_extmatch_out = NULL; | |
4110 | |
4111 if (prog->reghasz == REX_SET) | |
4112 { | |
4113 int i; | |
4114 | |
4115 cleanup_zsubexpr(); | |
4116 re_extmatch_out = make_extmatch(); | |
4117 for (i = 0; i < NSUBEXP; i++) | |
7 | 4118 { |
189 | 4119 if (REG_MULTI) |
7 | 4120 { |
189 | 4121 /* Only accept single line matches. */ |
4122 if (reg_startzpos[i].lnum >= 0 | |
5820 | 4123 && reg_endzpos[i].lnum == reg_startzpos[i].lnum |
4124 && reg_endzpos[i].col >= reg_startzpos[i].col) | |
189 | 4125 re_extmatch_out->matches[i] = |
4126 vim_strnsave(reg_getline(reg_startzpos[i].lnum) | |
7 | 4127 + reg_startzpos[i].col, |
189 | 4128 reg_endzpos[i].col - reg_startzpos[i].col); |
4129 } | |
4130 else | |
4131 { | |
4132 if (reg_startzp[i] != NULL && reg_endzp[i] != NULL) | |
4133 re_extmatch_out->matches[i] = | |
7 | 4134 vim_strnsave(reg_startzp[i], |
189 | 4135 (int)(reg_endzp[i] - reg_startzp[i])); |
7 | 4136 } |
4137 } | |
189 | 4138 } |
7 | 4139 #endif |
189 | 4140 return 1 + reglnum; |
7 | 4141 } |
4142 | |
4143 #ifdef FEAT_MBYTE | |
4144 static int reg_prev_class __ARGS((void)); | |
4145 | |
4146 /* | |
4147 * Get class of previous character. | |
4148 */ | |
4149 static int | |
4150 reg_prev_class() | |
4151 { | |
4152 if (reginput > regline) | |
4069 | 4153 return mb_get_class_buf(reginput - 1 |
4154 - (*mb_head_off)(regline, reginput - 1), reg_buf); | |
7 | 4155 return -1; |
4156 } | |
5735 | 4157 #endif |
4158 | |
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4159 static int reg_match_visual __ARGS((void)); |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4160 |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4161 /* |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4162 * 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
|
4163 */ |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4164 static int |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4165 reg_match_visual() |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4166 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4167 pos_T top, bot; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4168 linenr_T lnum; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4169 colnr_T col; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4170 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
|
4171 int mode; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4172 colnr_T start, end; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4173 colnr_T start2, end2; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4174 colnr_T cols; |
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 /* Check if the buffer is the current buffer. */ |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4177 if (reg_buf != curbuf || VIsual.lnum == 0) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4178 return FALSE; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4179 |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4180 if (VIsual_active) |
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 if (lt(VIsual, wp->w_cursor)) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4183 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4184 top = VIsual; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4185 bot = wp->w_cursor; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4186 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4187 else |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4188 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4189 top = wp->w_cursor; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4190 bot = VIsual; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4191 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4192 mode = VIsual_mode; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4193 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4194 else |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4195 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4196 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
|
4197 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4198 top = curbuf->b_visual.vi_start; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4199 bot = curbuf->b_visual.vi_end; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4200 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4201 else |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4202 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4203 top = curbuf->b_visual.vi_end; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4204 bot = curbuf->b_visual.vi_start; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4205 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4206 mode = curbuf->b_visual.vi_mode; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4207 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4208 lnum = reglnum + reg_firstlnum; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4209 if (lnum < top.lnum || lnum > bot.lnum) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4210 return FALSE; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4211 |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4212 if (mode == 'v') |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4213 { |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4214 col = (colnr_T)(reginput - regline); |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4215 if ((lnum == top.lnum && col < top.col) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4216 || (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
|
4217 return FALSE; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4218 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4219 else if (mode == Ctrl_V) |
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 getvvcol(wp, &top, &start, NULL, &end); |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4222 getvvcol(wp, &bot, &start2, NULL, &end2); |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4223 if (start2 < start) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4224 start = start2; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4225 if (end2 > end) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4226 end = end2; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4227 if (top.col == MAXCOL || bot.col == MAXCOL) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4228 end = MAXCOL; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4229 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
|
4230 if (cols < start || cols > end - (*p_sel == 'e')) |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4231 return FALSE; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4232 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4233 return TRUE; |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4234 } |
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4235 |
39 | 4236 #define ADVANCE_REGINPUT() mb_ptr_adv(reginput) |
7 | 4237 |
4238 /* | |
4239 * The arguments from BRACE_LIMITS are stored here. They are actually local | |
4240 * to regmatch(), but they are here to reduce the amount of stack space used | |
4241 * (it can be called recursively many times). | |
4242 */ | |
4243 static long bl_minval; | |
4244 static long bl_maxval; | |
4245 | |
4246 /* | |
4247 * regmatch - main matching routine | |
4248 * | |
180 | 4249 * Conceptually the strategy is simple: Check to see whether the current node |
4250 * matches, push an item onto the regstack and loop to see whether the rest | |
4251 * matches, and then act accordingly. In practice we make some effort to | |
4252 * avoid using the regstack, in particular by going through "ordinary" nodes | |
4253 * (that don't need to know whether the rest of the match failed) by a nested | |
4254 * loop. | |
7 | 4255 * |
4256 * Returns TRUE when there is a match. Leaves reginput and reglnum just after | |
4257 * the last matched character. | |
4258 * Returns FALSE when there is no match. Leaves reginput and reglnum in an | |
4259 * undefined state! | |
4260 */ | |
4261 static int | |
180 | 4262 regmatch(scan) |
7 | 4263 char_u *scan; /* Current node. */ |
4264 { | |
180 | 4265 char_u *next; /* Next node. */ |
4266 int op; | |
4267 int c; | |
4268 regitem_T *rp; | |
4269 int no; | |
4270 int status; /* one of the RA_ values: */ | |
4271 #define RA_FAIL 1 /* something failed, abort */ | |
4272 #define RA_CONT 2 /* continue in inner loop */ | |
4273 #define RA_BREAK 3 /* break inner loop */ | |
4274 #define RA_MATCH 4 /* successful match */ | |
4275 #define RA_NOMATCH 5 /* didn't match */ | |
270 | 4276 |
1520 | 4277 /* Make "regstack" and "backpos" empty. They are allocated and freed in |
4444 | 4278 * bt_regexec_both() to reduce malloc()/free() calls. */ |
270 | 4279 regstack.ga_len = 0; |
4280 backpos.ga_len = 0; | |
233 | 4281 |
180 | 4282 /* |
233 | 4283 * Repeat until "regstack" is empty. |
180 | 4284 */ |
4285 for (;;) | |
4286 { | |
5310 | 4287 /* Some patterns may take a long time to match, e.g., "\([a-z]\+\)\+Q". |
4288 * Allow interrupting them with CTRL-C. */ | |
7 | 4289 fast_breakcheck(); |
4290 | |
4291 #ifdef DEBUG | |
4292 if (scan != NULL && regnarrate) | |
4293 { | |
4444 | 4294 mch_errmsg((char *)regprop(scan)); |
7 | 4295 mch_errmsg("(\n"); |
4296 } | |
4297 #endif | |
180 | 4298 |
4299 /* | |
233 | 4300 * Repeat for items that can be matched sequentially, without using the |
180 | 4301 * regstack. |
4302 */ | |
4303 for (;;) | |
7 | 4304 { |
180 | 4305 if (got_int || scan == NULL) |
4306 { | |
4307 status = RA_FAIL; | |
4308 break; | |
4309 } | |
4310 status = RA_CONT; | |
4311 | |
7 | 4312 #ifdef DEBUG |
4313 if (regnarrate) | |
4314 { | |
4444 | 4315 mch_errmsg((char *)regprop(scan)); |
7 | 4316 mch_errmsg("...\n"); |
4317 # ifdef FEAT_SYN_HL | |
4318 if (re_extmatch_in != NULL) | |
4319 { | |
4320 int i; | |
4321 | |
4322 mch_errmsg(_("External submatches:\n")); | |
4323 for (i = 0; i < NSUBEXP; i++) | |
4324 { | |
4325 mch_errmsg(" \""); | |
4326 if (re_extmatch_in->matches[i] != NULL) | |
4444 | 4327 mch_errmsg((char *)re_extmatch_in->matches[i]); |
7 | 4328 mch_errmsg("\"\n"); |
4329 } | |
4330 } | |
4331 # endif | |
4332 } | |
4333 #endif | |
4334 next = regnext(scan); | |
4335 | |
4336 op = OP(scan); | |
4337 /* Check for character class with NL added. */ | |
1018 | 4338 if (!reg_line_lbr && WITH_NL(op) && REG_MULTI |
4339 && *reginput == NUL && reglnum <= reg_maxline) | |
7 | 4340 { |
4341 reg_nextline(); | |
4342 } | |
4343 else if (reg_line_lbr && WITH_NL(op) && *reginput == '\n') | |
4344 { | |
4345 ADVANCE_REGINPUT(); | |
4346 } | |
4347 else | |
4348 { | |
4349 if (WITH_NL(op)) | |
180 | 4350 op -= ADD_NL; |
7 | 4351 #ifdef FEAT_MBYTE |
4352 if (has_mbyte) | |
4353 c = (*mb_ptr2char)(reginput); | |
4354 else | |
4355 #endif | |
4356 c = *reginput; | |
4357 switch (op) | |
4358 { | |
4359 case BOL: | |
4360 if (reginput != regline) | |
180 | 4361 status = RA_NOMATCH; |
7 | 4362 break; |
4363 | |
4364 case EOL: | |
4365 if (c != NUL) | |
180 | 4366 status = RA_NOMATCH; |
7 | 4367 break; |
4368 | |
4369 case RE_BOF: | |
1458 | 4370 /* We're not at the beginning of the file when below the first |
4371 * line where we started, not at the start of the line or we | |
4372 * didn't start at the first line of the buffer. */ | |
7 | 4373 if (reglnum != 0 || reginput != regline |
1458 | 4374 || (REG_MULTI && reg_firstlnum > 1)) |
180 | 4375 status = RA_NOMATCH; |
7 | 4376 break; |
4377 | |
4378 case RE_EOF: | |
4379 if (reglnum != reg_maxline || c != NUL) | |
180 | 4380 status = RA_NOMATCH; |
7 | 4381 break; |
4382 | |
4383 case CURSOR: | |
4384 /* Check if the buffer is in a window and compare the | |
4385 * reg_win->w_cursor position to the match position. */ | |
4386 if (reg_win == NULL | |
4387 || (reglnum + reg_firstlnum != reg_win->w_cursor.lnum) | |
4388 || ((colnr_T)(reginput - regline) != reg_win->w_cursor.col)) | |
180 | 4389 status = RA_NOMATCH; |
7 | 4390 break; |
4391 | |
639 | 4392 case RE_MARK: |
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4393 /* Compare the mark position to the match position. */ |
639 | 4394 { |
4395 int mark = OPERAND(scan)[0]; | |
4396 int cmp = OPERAND(scan)[1]; | |
4397 pos_T *pos; | |
4398 | |
4043 | 4399 pos = getmark_buf(reg_buf, mark, FALSE); |
1148 | 4400 if (pos == NULL /* mark doesn't exist */ |
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4401 || pos->lnum <= 0 /* mark isn't set in reg_buf */ |
639 | 4402 || (pos->lnum == reglnum + reg_firstlnum |
4403 ? (pos->col == (colnr_T)(reginput - regline) | |
4404 ? (cmp == '<' || cmp == '>') | |
4405 : (pos->col < (colnr_T)(reginput - regline) | |
4406 ? cmp != '>' | |
4407 : cmp != '<')) | |
4408 : (pos->lnum < reglnum + reg_firstlnum | |
4409 ? cmp != '>' | |
4410 : cmp != '<'))) | |
4411 status = RA_NOMATCH; | |
4412 } | |
4413 break; | |
4414 | |
4415 case RE_VISUAL: | |
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4416 if (!reg_match_visual()) |
639 | 4417 status = RA_NOMATCH; |
4418 break; | |
4419 | |
7 | 4420 case RE_LNUM: |
4421 if (!REG_MULTI || !re_num_cmp((long_u)(reglnum + reg_firstlnum), | |
4422 scan)) | |
180 | 4423 status = RA_NOMATCH; |
7 | 4424 break; |
4425 | |
4426 case RE_COL: | |
4427 if (!re_num_cmp((long_u)(reginput - regline) + 1, scan)) | |
180 | 4428 status = RA_NOMATCH; |
7 | 4429 break; |
4430 | |
4431 case RE_VCOL: | |
4432 if (!re_num_cmp((long_u)win_linetabsize( | |
4433 reg_win == NULL ? curwin : reg_win, | |
4434 regline, (colnr_T)(reginput - regline)) + 1, scan)) | |
180 | 4435 status = RA_NOMATCH; |
7 | 4436 break; |
4437 | |
4438 case BOW: /* \<word; reginput points to w */ | |
4439 if (c == NUL) /* Can't match at end of line */ | |
180 | 4440 status = RA_NOMATCH; |
7 | 4441 #ifdef FEAT_MBYTE |
180 | 4442 else if (has_mbyte) |
7 | 4443 { |
4444 int this_class; | |
4445 | |
4446 /* Get class of current and previous char (if it exists). */ | |
4069 | 4447 this_class = mb_get_class_buf(reginput, reg_buf); |
7 | 4448 if (this_class <= 1) |
180 | 4449 status = RA_NOMATCH; /* not on a word at all */ |
4450 else if (reg_prev_class() == this_class) | |
4451 status = RA_NOMATCH; /* previous char is in same word */ | |
7 | 4452 } |
4453 #endif | |
4454 else | |
4455 { | |
4061 | 4456 if (!vim_iswordc_buf(c, reg_buf) || (reginput > regline |
4457 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
180 | 4458 status = RA_NOMATCH; |
7 | 4459 } |
4460 break; | |
4461 | |
4462 case EOW: /* word\>; reginput points after d */ | |
4463 if (reginput == regline) /* Can't match at start of line */ | |
180 | 4464 status = RA_NOMATCH; |
7 | 4465 #ifdef FEAT_MBYTE |
180 | 4466 else if (has_mbyte) |
7 | 4467 { |
4468 int this_class, prev_class; | |
4469 | |
4470 /* Get class of current and previous char (if it exists). */ | |
4069 | 4471 this_class = mb_get_class_buf(reginput, reg_buf); |
7 | 4472 prev_class = reg_prev_class(); |
180 | 4473 if (this_class == prev_class |
4474 || prev_class == 0 || prev_class == 1) | |
4475 status = RA_NOMATCH; | |
7 | 4476 } |
180 | 4477 #endif |
7 | 4478 else |
4479 { | |
4043 | 4480 if (!vim_iswordc_buf(reginput[-1], reg_buf) |
4481 || (reginput[0] != NUL && vim_iswordc_buf(c, reg_buf))) | |
180 | 4482 status = RA_NOMATCH; |
7 | 4483 } |
4484 break; /* Matched with EOW */ | |
4485 | |
4486 case ANY: | |
4084 | 4487 /* ANY does not match new lines. */ |
7 | 4488 if (c == NUL) |
180 | 4489 status = RA_NOMATCH; |
4490 else | |
4491 ADVANCE_REGINPUT(); | |
7 | 4492 break; |
4493 | |
4494 case IDENT: | |
4495 if (!vim_isIDc(c)) | |
180 | 4496 status = RA_NOMATCH; |
4497 else | |
4498 ADVANCE_REGINPUT(); | |
7 | 4499 break; |
4500 | |
4501 case SIDENT: | |
4502 if (VIM_ISDIGIT(*reginput) || !vim_isIDc(c)) | |
180 | 4503 status = RA_NOMATCH; |
4504 else | |
4505 ADVANCE_REGINPUT(); | |
7 | 4506 break; |
4507 | |
4508 case KWORD: | |
4069 | 4509 if (!vim_iswordp_buf(reginput, reg_buf)) |
180 | 4510 status = RA_NOMATCH; |
4511 else | |
4512 ADVANCE_REGINPUT(); | |
7 | 4513 break; |
4514 | |
4515 case SKWORD: | |
4069 | 4516 if (VIM_ISDIGIT(*reginput) || !vim_iswordp_buf(reginput, reg_buf)) |
180 | 4517 status = RA_NOMATCH; |
4518 else | |
4519 ADVANCE_REGINPUT(); | |
7 | 4520 break; |
4521 | |
4522 case FNAME: | |
4523 if (!vim_isfilec(c)) | |
180 | 4524 status = RA_NOMATCH; |
4525 else | |
4526 ADVANCE_REGINPUT(); | |
7 | 4527 break; |
4528 | |
4529 case SFNAME: | |
4530 if (VIM_ISDIGIT(*reginput) || !vim_isfilec(c)) | |
180 | 4531 status = RA_NOMATCH; |
4532 else | |
4533 ADVANCE_REGINPUT(); | |
7 | 4534 break; |
4535 | |
4536 case PRINT: | |
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
4537 if (!vim_isprintc(PTR2CHAR(reginput))) |
180 | 4538 status = RA_NOMATCH; |
4539 else | |
4540 ADVANCE_REGINPUT(); | |
7 | 4541 break; |
4542 | |
4543 case SPRINT: | |
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
4544 if (VIM_ISDIGIT(*reginput) || !vim_isprintc(PTR2CHAR(reginput))) |
180 | 4545 status = RA_NOMATCH; |
4546 else | |
4547 ADVANCE_REGINPUT(); | |
7 | 4548 break; |
4549 | |
4550 case WHITE: | |
4551 if (!vim_iswhite(c)) | |
180 | 4552 status = RA_NOMATCH; |
4553 else | |
4554 ADVANCE_REGINPUT(); | |
7 | 4555 break; |
4556 | |
4557 case NWHITE: | |
4558 if (c == NUL || vim_iswhite(c)) | |
180 | 4559 status = RA_NOMATCH; |
4560 else | |
4561 ADVANCE_REGINPUT(); | |
7 | 4562 break; |
4563 | |
4564 case DIGIT: | |
4565 if (!ri_digit(c)) | |
180 | 4566 status = RA_NOMATCH; |
4567 else | |
4568 ADVANCE_REGINPUT(); | |
7 | 4569 break; |
4570 | |
4571 case NDIGIT: | |
4572 if (c == NUL || ri_digit(c)) | |
180 | 4573 status = RA_NOMATCH; |
4574 else | |
4575 ADVANCE_REGINPUT(); | |
7 | 4576 break; |
4577 | |
4578 case HEX: | |
4579 if (!ri_hex(c)) | |
180 | 4580 status = RA_NOMATCH; |
4581 else | |
4582 ADVANCE_REGINPUT(); | |
7 | 4583 break; |
4584 | |
4585 case NHEX: | |
4586 if (c == NUL || ri_hex(c)) | |
180 | 4587 status = RA_NOMATCH; |
4588 else | |
4589 ADVANCE_REGINPUT(); | |
7 | 4590 break; |
4591 | |
4592 case OCTAL: | |
4593 if (!ri_octal(c)) | |
180 | 4594 status = RA_NOMATCH; |
4595 else | |
4596 ADVANCE_REGINPUT(); | |
7 | 4597 break; |
4598 | |
4599 case NOCTAL: | |
4600 if (c == NUL || ri_octal(c)) | |
180 | 4601 status = RA_NOMATCH; |
4602 else | |
4603 ADVANCE_REGINPUT(); | |
7 | 4604 break; |
4605 | |
4606 case WORD: | |
4607 if (!ri_word(c)) | |
180 | 4608 status = RA_NOMATCH; |
4609 else | |
4610 ADVANCE_REGINPUT(); | |
7 | 4611 break; |
4612 | |
4613 case NWORD: | |
4614 if (c == NUL || ri_word(c)) | |
180 | 4615 status = RA_NOMATCH; |
4616 else | |
4617 ADVANCE_REGINPUT(); | |
7 | 4618 break; |
4619 | |
4620 case HEAD: | |
4621 if (!ri_head(c)) | |
180 | 4622 status = RA_NOMATCH; |
4623 else | |
4624 ADVANCE_REGINPUT(); | |
7 | 4625 break; |
4626 | |
4627 case NHEAD: | |
4628 if (c == NUL || ri_head(c)) | |
180 | 4629 status = RA_NOMATCH; |
4630 else | |
4631 ADVANCE_REGINPUT(); | |
7 | 4632 break; |
4633 | |
4634 case ALPHA: | |
4635 if (!ri_alpha(c)) | |
180 | 4636 status = RA_NOMATCH; |
4637 else | |
4638 ADVANCE_REGINPUT(); | |
7 | 4639 break; |
4640 | |
4641 case NALPHA: | |
4642 if (c == NUL || ri_alpha(c)) | |
180 | 4643 status = RA_NOMATCH; |
4644 else | |
4645 ADVANCE_REGINPUT(); | |
7 | 4646 break; |
4647 | |
4648 case LOWER: | |
4649 if (!ri_lower(c)) | |
180 | 4650 status = RA_NOMATCH; |
4651 else | |
4652 ADVANCE_REGINPUT(); | |
7 | 4653 break; |
4654 | |
4655 case NLOWER: | |
4656 if (c == NUL || ri_lower(c)) | |
180 | 4657 status = RA_NOMATCH; |
4658 else | |
4659 ADVANCE_REGINPUT(); | |
7 | 4660 break; |
4661 | |
4662 case UPPER: | |
4663 if (!ri_upper(c)) | |
180 | 4664 status = RA_NOMATCH; |
4665 else | |
4666 ADVANCE_REGINPUT(); | |
7 | 4667 break; |
4668 | |
4669 case NUPPER: | |
4670 if (c == NUL || ri_upper(c)) | |
180 | 4671 status = RA_NOMATCH; |
4672 else | |
4673 ADVANCE_REGINPUT(); | |
7 | 4674 break; |
4675 | |
4676 case EXACTLY: | |
4677 { | |
4678 int len; | |
4679 char_u *opnd; | |
4680 | |
4681 opnd = OPERAND(scan); | |
4682 /* Inline the first byte, for speed. */ | |
4683 if (*opnd != *reginput | |
4684 && (!ireg_ic || ( | |
4685 #ifdef FEAT_MBYTE | |
4686 !enc_utf8 && | |
4687 #endif | |
1347 | 4688 MB_TOLOWER(*opnd) != MB_TOLOWER(*reginput)))) |
180 | 4689 status = RA_NOMATCH; |
4690 else if (*opnd == NUL) | |
7 | 4691 { |
4692 /* match empty string always works; happens when "~" is | |
4693 * empty. */ | |
4694 } | |
4695 else if (opnd[1] == NUL | |
4696 #ifdef FEAT_MBYTE | |
4697 && !(enc_utf8 && ireg_ic) | |
4698 #endif | |
4699 ) | |
4700 ++reginput; /* matched a single char */ | |
4701 else | |
4702 { | |
4703 len = (int)STRLEN(opnd); | |
4704 /* Need to match first byte again for multi-byte. */ | |
4705 if (cstrncmp(opnd, reginput, &len) != 0) | |
180 | 4706 status = RA_NOMATCH; |
7 | 4707 #ifdef FEAT_MBYTE |
4708 /* Check for following composing character. */ | |
180 | 4709 else if (enc_utf8 |
4710 && UTF_COMPOSINGLIKE(reginput, reginput + len)) | |
7 | 4711 { |
4712 /* raaron: This code makes a composing character get | |
4713 * ignored, which is the correct behavior (sometimes) | |
4714 * for voweled Hebrew texts. */ | |
4715 if (!ireg_icombine) | |
180 | 4716 status = RA_NOMATCH; |
7 | 4717 } |
180 | 4718 #endif |
7 | 4719 else |
180 | 4720 reginput += len; |
7 | 4721 } |
4722 } | |
4723 break; | |
4724 | |
4725 case ANYOF: | |
4726 case ANYBUT: | |
4727 if (c == NUL) | |
180 | 4728 status = RA_NOMATCH; |
4729 else if ((cstrchr(OPERAND(scan), c) == NULL) == (op == ANYOF)) | |
4730 status = RA_NOMATCH; | |
4731 else | |
4732 ADVANCE_REGINPUT(); | |
7 | 4733 break; |
4734 | |
4735 #ifdef FEAT_MBYTE | |
4736 case MULTIBYTECODE: | |
4737 if (has_mbyte) | |
4738 { | |
4739 int i, len; | |
4740 char_u *opnd; | |
944 | 4741 int opndc = 0, inpc; |
7 | 4742 |
4743 opnd = OPERAND(scan); | |
4744 /* Safety check (just in case 'encoding' was changed since | |
4745 * compiling the program). */ | |
474 | 4746 if ((len = (*mb_ptr2len)(opnd)) < 2) |
180 | 4747 { |
4748 status = RA_NOMATCH; | |
4749 break; | |
4750 } | |
714 | 4751 if (enc_utf8) |
4752 opndc = mb_ptr2char(opnd); | |
4753 if (enc_utf8 && utf_iscomposing(opndc)) | |
4754 { | |
4755 /* When only a composing char is given match at any | |
4756 * position where that composing char appears. */ | |
4757 status = RA_NOMATCH; | |
4758 for (i = 0; reginput[i] != NUL; i += utf_char2len(inpc)) | |
180 | 4759 { |
714 | 4760 inpc = mb_ptr2char(reginput + i); |
4761 if (!utf_iscomposing(inpc)) | |
4762 { | |
4763 if (i > 0) | |
4764 break; | |
4765 } | |
4766 else if (opndc == inpc) | |
4767 { | |
4768 /* Include all following composing chars. */ | |
4769 len = i + mb_ptr2len(reginput + i); | |
4770 status = RA_MATCH; | |
4771 break; | |
4772 } | |
180 | 4773 } |
714 | 4774 } |
4775 else | |
4776 for (i = 0; i < len; ++i) | |
4777 if (opnd[i] != reginput[i]) | |
4778 { | |
4779 status = RA_NOMATCH; | |
4780 break; | |
4781 } | |
7 | 4782 reginput += len; |
4783 } | |
4784 else | |
180 | 4785 status = RA_NOMATCH; |
7 | 4786 break; |
4787 #endif | |
4788 | |
4789 case NOTHING: | |
4790 break; | |
4791 | |
4792 case BACK: | |
233 | 4793 { |
4794 int i; | |
4795 backpos_T *bp; | |
4796 | |
4797 /* | |
4798 * When we run into BACK we need to check if we don't keep | |
4799 * looping without matching any input. The second and later | |
4800 * times a BACK is encountered it fails if the input is still | |
4801 * at the same position as the previous time. | |
4802 * The positions are stored in "backpos" and found by the | |
4803 * current value of "scan", the position in the RE program. | |
4804 */ | |
4805 bp = (backpos_T *)backpos.ga_data; | |
4806 for (i = 0; i < backpos.ga_len; ++i) | |
4807 if (bp[i].bp_scan == scan) | |
4808 break; | |
4809 if (i == backpos.ga_len) | |
4810 { | |
4811 /* First time at this BACK, make room to store the pos. */ | |
4812 if (ga_grow(&backpos, 1) == FAIL) | |
4813 status = RA_FAIL; | |
4814 else | |
4815 { | |
4816 /* get "ga_data" again, it may have changed */ | |
4817 bp = (backpos_T *)backpos.ga_data; | |
4818 bp[i].bp_scan = scan; | |
4819 ++backpos.ga_len; | |
4820 } | |
4821 } | |
4822 else if (reg_save_equal(&bp[i].bp_pos)) | |
4823 /* Still at same position as last time, fail. */ | |
4824 status = RA_NOMATCH; | |
4825 | |
4826 if (status != RA_FAIL && status != RA_NOMATCH) | |
4827 reg_save(&bp[i].bp_pos, &backpos); | |
4828 } | |
179 | 4829 break; |
4830 | |
7 | 4831 case MOPEN + 0: /* Match start: \zs */ |
4832 case MOPEN + 1: /* \( */ | |
4833 case MOPEN + 2: | |
4834 case MOPEN + 3: | |
4835 case MOPEN + 4: | |
4836 case MOPEN + 5: | |
4837 case MOPEN + 6: | |
4838 case MOPEN + 7: | |
4839 case MOPEN + 8: | |
4840 case MOPEN + 9: | |
4841 { | |
4842 no = op - MOPEN; | |
4843 cleanup_subexpr(); | |
270 | 4844 rp = regstack_push(RS_MOPEN, scan); |
180 | 4845 if (rp == NULL) |
4846 status = RA_FAIL; | |
4847 else | |
4848 { | |
4849 rp->rs_no = no; | |
4850 save_se(&rp->rs_un.sesave, ®_startpos[no], | |
4851 ®_startp[no]); | |
4852 /* We simply continue and handle the result when done. */ | |
4853 } | |
7 | 4854 } |
180 | 4855 break; |
7 | 4856 |
4857 case NOPEN: /* \%( */ | |
4858 case NCLOSE: /* \) after \%( */ | |
270 | 4859 if (regstack_push(RS_NOPEN, scan) == NULL) |
180 | 4860 status = RA_FAIL; |
4861 /* We simply continue and handle the result when done. */ | |
4862 break; | |
7 | 4863 |
4864 #ifdef FEAT_SYN_HL | |
4865 case ZOPEN + 1: | |
4866 case ZOPEN + 2: | |
4867 case ZOPEN + 3: | |
4868 case ZOPEN + 4: | |
4869 case ZOPEN + 5: | |
4870 case ZOPEN + 6: | |
4871 case ZOPEN + 7: | |
4872 case ZOPEN + 8: | |
4873 case ZOPEN + 9: | |
4874 { | |
4875 no = op - ZOPEN; | |
4876 cleanup_zsubexpr(); | |
270 | 4877 rp = regstack_push(RS_ZOPEN, 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, ®_startzpos[no], | |
4884 ®_startzp[no]); | |
4885 /* We simply continue and handle the result when done. */ | |
4886 } | |
7 | 4887 } |
180 | 4888 break; |
7 | 4889 #endif |
4890 | |
4891 case MCLOSE + 0: /* Match end: \ze */ | |
4892 case MCLOSE + 1: /* \) */ | |
4893 case MCLOSE + 2: | |
4894 case MCLOSE + 3: | |
4895 case MCLOSE + 4: | |
4896 case MCLOSE + 5: | |
4897 case MCLOSE + 6: | |
4898 case MCLOSE + 7: | |
4899 case MCLOSE + 8: | |
4900 case MCLOSE + 9: | |
4901 { | |
4902 no = op - MCLOSE; | |
4903 cleanup_subexpr(); | |
270 | 4904 rp = regstack_push(RS_MCLOSE, scan); |
180 | 4905 if (rp == NULL) |
4906 status = RA_FAIL; | |
4907 else | |
4908 { | |
4909 rp->rs_no = no; | |
4910 save_se(&rp->rs_un.sesave, ®_endpos[no], ®_endp[no]); | |
4911 /* We simply continue and handle the result when done. */ | |
4912 } | |
7 | 4913 } |
180 | 4914 break; |
7 | 4915 |
4916 #ifdef FEAT_SYN_HL | |
4917 case ZCLOSE + 1: /* \) after \z( */ | |
4918 case ZCLOSE + 2: | |
4919 case ZCLOSE + 3: | |
4920 case ZCLOSE + 4: | |
4921 case ZCLOSE + 5: | |
4922 case ZCLOSE + 6: | |
4923 case ZCLOSE + 7: | |
4924 case ZCLOSE + 8: | |
4925 case ZCLOSE + 9: | |
4926 { | |
4927 no = op - ZCLOSE; | |
4928 cleanup_zsubexpr(); | |
270 | 4929 rp = regstack_push(RS_ZCLOSE, scan); |
180 | 4930 if (rp == NULL) |
4931 status = RA_FAIL; | |
4932 else | |
4933 { | |
4934 rp->rs_no = no; | |
4935 save_se(&rp->rs_un.sesave, ®_endzpos[no], | |
4936 ®_endzp[no]); | |
4937 /* We simply continue and handle the result when done. */ | |
4938 } | |
7 | 4939 } |
180 | 4940 break; |
7 | 4941 #endif |
4942 | |
4943 case BACKREF + 1: | |
4944 case BACKREF + 2: | |
4945 case BACKREF + 3: | |
4946 case BACKREF + 4: | |
4947 case BACKREF + 5: | |
4948 case BACKREF + 6: | |
4949 case BACKREF + 7: | |
4950 case BACKREF + 8: | |
4951 case BACKREF + 9: | |
4952 { | |
4953 int len; | |
4954 | |
4955 no = op - BACKREF; | |
4956 cleanup_subexpr(); | |
4957 if (!REG_MULTI) /* Single-line regexp */ | |
4958 { | |
1815 | 4959 if (reg_startp[no] == NULL || reg_endp[no] == NULL) |
7 | 4960 { |
4961 /* Backref was not set: Match an empty string. */ | |
4962 len = 0; | |
4963 } | |
4964 else | |
4965 { | |
4966 /* Compare current input with back-ref in the same | |
4967 * line. */ | |
4968 len = (int)(reg_endp[no] - reg_startp[no]); | |
4969 if (cstrncmp(reg_startp[no], reginput, &len) != 0) | |
180 | 4970 status = RA_NOMATCH; |
7 | 4971 } |
4972 } | |
4973 else /* Multi-line regexp */ | |
4974 { | |
1815 | 4975 if (reg_startpos[no].lnum < 0 || reg_endpos[no].lnum < 0) |
7 | 4976 { |
4977 /* Backref was not set: Match an empty string. */ | |
4978 len = 0; | |
4979 } | |
4980 else | |
4981 { | |
4982 if (reg_startpos[no].lnum == reglnum | |
4983 && reg_endpos[no].lnum == reglnum) | |
4984 { | |
4985 /* Compare back-ref within the current line. */ | |
4986 len = reg_endpos[no].col - reg_startpos[no].col; | |
4987 if (cstrncmp(regline + reg_startpos[no].col, | |
4988 reginput, &len) != 0) | |
180 | 4989 status = RA_NOMATCH; |
7 | 4990 } |
4991 else | |
4992 { | |
4993 /* Messy situation: Need to compare between two | |
4994 * lines. */ | |
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
4995 int r = match_with_backref( |
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4996 reg_startpos[no].lnum, |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4997 reg_startpos[no].col, |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4998 reg_endpos[no].lnum, |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4999 reg_endpos[no].col, |
4899
4837fd61be52
updated for version 7.3.1195
Bram Moolenaar <bram@vim.org>
parents:
4891
diff
changeset
|
5000 &len); |
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5001 |
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5002 if (r != RA_MATCH) |
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5003 status = r; |
7 | 5004 } |
5005 } | |
5006 } | |
5007 | |
5008 /* Matched the backref, skip over it. */ | |
5009 reginput += len; | |
5010 } | |
5011 break; | |
5012 | |
5013 #ifdef FEAT_SYN_HL | |
5014 case ZREF + 1: | |
5015 case ZREF + 2: | |
5016 case ZREF + 3: | |
5017 case ZREF + 4: | |
5018 case ZREF + 5: | |
5019 case ZREF + 6: | |
5020 case ZREF + 7: | |
5021 case ZREF + 8: | |
5022 case ZREF + 9: | |
5023 { | |
5024 int len; | |
5025 | |
5026 cleanup_zsubexpr(); | |
5027 no = op - ZREF; | |
5028 if (re_extmatch_in != NULL | |
5029 && re_extmatch_in->matches[no] != NULL) | |
5030 { | |
5031 len = (int)STRLEN(re_extmatch_in->matches[no]); | |
5032 if (cstrncmp(re_extmatch_in->matches[no], | |
5033 reginput, &len) != 0) | |
180 | 5034 status = RA_NOMATCH; |
5035 else | |
5036 reginput += len; | |
7 | 5037 } |
5038 else | |
5039 { | |
5040 /* Backref was not set: Match an empty string. */ | |
5041 } | |
5042 } | |
5043 break; | |
5044 #endif | |
5045 | |
5046 case BRANCH: | |
5047 { | |
5048 if (OP(next) != BRANCH) /* No choice. */ | |
5049 next = OPERAND(scan); /* Avoid recursion. */ | |
5050 else | |
5051 { | |
270 | 5052 rp = regstack_push(RS_BRANCH, scan); |
180 | 5053 if (rp == NULL) |
5054 status = RA_FAIL; | |
5055 else | |
5056 status = RA_BREAK; /* rest is below */ | |
7 | 5057 } |
5058 } | |
5059 break; | |
5060 | |
5061 case BRACE_LIMITS: | |
5062 { | |
5063 if (OP(next) == BRACE_SIMPLE) | |
5064 { | |
5065 bl_minval = OPERAND_MIN(scan); | |
5066 bl_maxval = OPERAND_MAX(scan); | |
5067 } | |
5068 else if (OP(next) >= BRACE_COMPLEX | |
5069 && OP(next) < BRACE_COMPLEX + 10) | |
5070 { | |
5071 no = OP(next) - BRACE_COMPLEX; | |
5072 brace_min[no] = OPERAND_MIN(scan); | |
5073 brace_max[no] = OPERAND_MAX(scan); | |
5074 brace_count[no] = 0; | |
5075 } | |
5076 else | |
5077 { | |
5078 EMSG(_(e_internal)); /* Shouldn't happen */ | |
180 | 5079 status = RA_FAIL; |
7 | 5080 } |
5081 } | |
5082 break; | |
5083 | |
5084 case BRACE_COMPLEX + 0: | |
5085 case BRACE_COMPLEX + 1: | |
5086 case BRACE_COMPLEX + 2: | |
5087 case BRACE_COMPLEX + 3: | |
5088 case BRACE_COMPLEX + 4: | |
5089 case BRACE_COMPLEX + 5: | |
5090 case BRACE_COMPLEX + 6: | |
5091 case BRACE_COMPLEX + 7: | |
5092 case BRACE_COMPLEX + 8: | |
5093 case BRACE_COMPLEX + 9: | |
5094 { | |
5095 no = op - BRACE_COMPLEX; | |
5096 ++brace_count[no]; | |
5097 | |
5098 /* If not matched enough times yet, try one more */ | |
5099 if (brace_count[no] <= (brace_min[no] <= brace_max[no] | |
180 | 5100 ? brace_min[no] : brace_max[no])) |
7 | 5101 { |
270 | 5102 rp = regstack_push(RS_BRCPLX_MORE, scan); |
180 | 5103 if (rp == NULL) |
5104 status = RA_FAIL; | |
5105 else | |
5106 { | |
5107 rp->rs_no = no; | |
233 | 5108 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5109 next = OPERAND(scan); |
5110 /* We continue and handle the result when done. */ | |
5111 } | |
5112 break; | |
7 | 5113 } |
5114 | |
5115 /* If matched enough times, may try matching some more */ | |
5116 if (brace_min[no] <= brace_max[no]) | |
5117 { | |
5118 /* Range is the normal way around, use longest match */ | |
5119 if (brace_count[no] <= brace_max[no]) | |
5120 { | |
270 | 5121 rp = regstack_push(RS_BRCPLX_LONG, scan); |
180 | 5122 if (rp == NULL) |
5123 status = RA_FAIL; | |
5124 else | |
5125 { | |
5126 rp->rs_no = no; | |
233 | 5127 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5128 next = OPERAND(scan); |
5129 /* We continue and handle the result when done. */ | |
5130 } | |
7 | 5131 } |
5132 } | |
5133 else | |
5134 { | |
5135 /* Range is backwards, use shortest match first */ | |
5136 if (brace_count[no] <= brace_min[no]) | |
5137 { | |
270 | 5138 rp = regstack_push(RS_BRCPLX_SHORT, scan); |
180 | 5139 if (rp == NULL) |
5140 status = RA_FAIL; | |
5141 else | |
5142 { | |
233 | 5143 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5144 /* We continue and handle the result when done. */ |
5145 } | |
7 | 5146 } |
5147 } | |
5148 } | |
5149 break; | |
5150 | |
5151 case BRACE_SIMPLE: | |
5152 case STAR: | |
5153 case PLUS: | |
5154 { | |
180 | 5155 regstar_T rst; |
7 | 5156 |
5157 /* | |
5158 * Lookahead to avoid useless match attempts when we know | |
5159 * what character comes next. | |
5160 */ | |
5161 if (OP(next) == EXACTLY) | |
5162 { | |
180 | 5163 rst.nextb = *OPERAND(next); |
7 | 5164 if (ireg_ic) |
5165 { | |
1347 | 5166 if (MB_ISUPPER(rst.nextb)) |
5167 rst.nextb_ic = MB_TOLOWER(rst.nextb); | |
7 | 5168 else |
1347 | 5169 rst.nextb_ic = MB_TOUPPER(rst.nextb); |
7 | 5170 } |
5171 else | |
180 | 5172 rst.nextb_ic = rst.nextb; |
7 | 5173 } |
5174 else | |
5175 { | |
180 | 5176 rst.nextb = NUL; |
5177 rst.nextb_ic = NUL; | |
7 | 5178 } |
5179 if (op != BRACE_SIMPLE) | |
5180 { | |
180 | 5181 rst.minval = (op == STAR) ? 0 : 1; |
5182 rst.maxval = MAX_LIMIT; | |
7 | 5183 } |
5184 else | |
5185 { | |
180 | 5186 rst.minval = bl_minval; |
5187 rst.maxval = bl_maxval; | |
7 | 5188 } |
5189 | |
5190 /* | |
5191 * When maxval > minval, try matching as much as possible, up | |
5192 * to maxval. When maxval < minval, try matching at least the | |
5193 * minimal number (since the range is backwards, that's also | |
5194 * maxval!). | |
5195 */ | |
180 | 5196 rst.count = regrepeat(OPERAND(scan), rst.maxval); |
7 | 5197 if (got_int) |
180 | 5198 { |
5199 status = RA_FAIL; | |
5200 break; | |
5201 } | |
5202 if (rst.minval <= rst.maxval | |
5203 ? rst.count >= rst.minval : rst.count >= rst.maxval) | |
7 | 5204 { |
180 | 5205 /* It could match. Prepare for trying to match what |
5206 * follows. The code is below. Parameters are stored in | |
5207 * a regstar_T on the regstack. */ | |
212 | 5208 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
189 | 5209 { |
5210 EMSG(_(e_maxmempat)); | |
5211 status = RA_FAIL; | |
5212 } | |
5213 else if (ga_grow(®stack, sizeof(regstar_T)) == FAIL) | |
180 | 5214 status = RA_FAIL; |
5215 else | |
7 | 5216 { |
180 | 5217 regstack.ga_len += sizeof(regstar_T); |
270 | 5218 rp = regstack_push(rst.minval <= rst.maxval |
233 | 5219 ? RS_STAR_LONG : RS_STAR_SHORT, scan); |
180 | 5220 if (rp == NULL) |
5221 status = RA_FAIL; | |
5222 else | |
7 | 5223 { |
180 | 5224 *(((regstar_T *)rp) - 1) = rst; |
5225 status = RA_BREAK; /* skip the restore bits */ | |
7 | 5226 } |
5227 } | |
5228 } | |
5229 else | |
180 | 5230 status = RA_NOMATCH; |
5231 | |
7 | 5232 } |
180 | 5233 break; |
7 | 5234 |
5235 case NOMATCH: | |
5236 case MATCH: | |
5237 case SUBPAT: | |
270 | 5238 rp = regstack_push(RS_NOMATCH, scan); |
180 | 5239 if (rp == NULL) |
5240 status = RA_FAIL; | |
5241 else | |
7 | 5242 { |
180 | 5243 rp->rs_no = op; |
233 | 5244 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5245 next = OPERAND(scan); |
5246 /* We continue and handle the result when done. */ | |
7 | 5247 } |
5248 break; | |
5249 | |
5250 case BEHIND: | |
5251 case NOBEHIND: | |
180 | 5252 /* Need a bit of room to store extra positions. */ |
212 | 5253 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
189 | 5254 { |
5255 EMSG(_(e_maxmempat)); | |
5256 status = RA_FAIL; | |
5257 } | |
5258 else if (ga_grow(®stack, sizeof(regbehind_T)) == FAIL) | |
180 | 5259 status = RA_FAIL; |
5260 else | |
7 | 5261 { |
180 | 5262 regstack.ga_len += sizeof(regbehind_T); |
270 | 5263 rp = regstack_push(RS_BEHIND1, scan); |
180 | 5264 if (rp == NULL) |
5265 status = RA_FAIL; | |
5266 else | |
7 | 5267 { |
1579 | 5268 /* Need to save the subexpr to be able to restore them |
5269 * when there is a match but we don't use it. */ | |
5270 save_subexpr(((regbehind_T *)rp) - 1); | |
5271 | |
180 | 5272 rp->rs_no = op; |
233 | 5273 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5274 /* First try if what follows matches. If it does then we |
5275 * check the behind match by looping. */ | |
7 | 5276 } |
5277 } | |
180 | 5278 break; |
7 | 5279 |
5280 case BHPOS: | |
5281 if (REG_MULTI) | |
5282 { | |
5283 if (behind_pos.rs_u.pos.col != (colnr_T)(reginput - regline) | |
5284 || behind_pos.rs_u.pos.lnum != reglnum) | |
180 | 5285 status = RA_NOMATCH; |
7 | 5286 } |
5287 else if (behind_pos.rs_u.ptr != reginput) | |
180 | 5288 status = RA_NOMATCH; |
7 | 5289 break; |
5290 | |
5291 case NEWL: | |
1018 | 5292 if ((c != NUL || !REG_MULTI || reglnum > reg_maxline |
5293 || reg_line_lbr) && (c != '\n' || !reg_line_lbr)) | |
180 | 5294 status = RA_NOMATCH; |
5295 else if (reg_line_lbr) | |
7 | 5296 ADVANCE_REGINPUT(); |
5297 else | |
5298 reg_nextline(); | |
5299 break; | |
5300 | |
5301 case END: | |
180 | 5302 status = RA_MATCH; /* Success! */ |
5303 break; | |
7 | 5304 |
5305 default: | |
5306 EMSG(_(e_re_corr)); | |
5307 #ifdef DEBUG | |
5308 printf("Illegal op code %d\n", op); | |
5309 #endif | |
180 | 5310 status = RA_FAIL; |
5311 break; | |
7 | 5312 } |
5313 } | |
5314 | |
180 | 5315 /* If we can't continue sequentially, break the inner loop. */ |
5316 if (status != RA_CONT) | |
5317 break; | |
5318 | |
5319 /* Continue in inner loop, advance to next item. */ | |
7 | 5320 scan = next; |
180 | 5321 |
5322 } /* end of inner loop */ | |
7 | 5323 |
5324 /* | |
180 | 5325 * If there is something on the regstack execute the code for the state. |
233 | 5326 * If the state is popped then loop and use the older state. |
7 | 5327 */ |
180 | 5328 while (regstack.ga_len > 0 && status != RA_FAIL) |
5329 { | |
5330 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; | |
5331 switch (rp->rs_state) | |
5332 { | |
5333 case RS_NOPEN: | |
5334 /* Result is passed on as-is, simply pop the state. */ | |
270 | 5335 regstack_pop(&scan); |
180 | 5336 break; |
5337 | |
5338 case RS_MOPEN: | |
5339 /* Pop the state. Restore pointers when there is no match. */ | |
5340 if (status == RA_NOMATCH) | |
5341 restore_se(&rp->rs_un.sesave, ®_startpos[rp->rs_no], | |
5342 ®_startp[rp->rs_no]); | |
270 | 5343 regstack_pop(&scan); |
180 | 5344 break; |
5345 | |
5346 #ifdef FEAT_SYN_HL | |
5347 case RS_ZOPEN: | |
5348 /* Pop the state. Restore pointers when there is no match. */ | |
5349 if (status == RA_NOMATCH) | |
5350 restore_se(&rp->rs_un.sesave, ®_startzpos[rp->rs_no], | |
5351 ®_startzp[rp->rs_no]); | |
270 | 5352 regstack_pop(&scan); |
180 | 5353 break; |
5354 #endif | |
5355 | |
5356 case RS_MCLOSE: | |
5357 /* Pop the state. Restore pointers when there is no match. */ | |
5358 if (status == RA_NOMATCH) | |
5359 restore_se(&rp->rs_un.sesave, ®_endpos[rp->rs_no], | |
5360 ®_endp[rp->rs_no]); | |
270 | 5361 regstack_pop(&scan); |
180 | 5362 break; |
5363 | |
5364 #ifdef FEAT_SYN_HL | |
5365 case RS_ZCLOSE: | |
5366 /* Pop the state. Restore pointers when there is no match. */ | |
5367 if (status == RA_NOMATCH) | |
5368 restore_se(&rp->rs_un.sesave, ®_endzpos[rp->rs_no], | |
5369 ®_endzp[rp->rs_no]); | |
270 | 5370 regstack_pop(&scan); |
180 | 5371 break; |
5372 #endif | |
5373 | |
5374 case RS_BRANCH: | |
5375 if (status == RA_MATCH) | |
5376 /* this branch matched, use it */ | |
270 | 5377 regstack_pop(&scan); |
180 | 5378 else |
5379 { | |
5380 if (status != RA_BREAK) | |
5381 { | |
5382 /* After a non-matching branch: try next one. */ | |
233 | 5383 reg_restore(&rp->rs_un.regsave, &backpos); |
180 | 5384 scan = rp->rs_scan; |
5385 } | |
5386 if (scan == NULL || OP(scan) != BRANCH) | |
5387 { | |
5388 /* no more branches, didn't find a match */ | |
5389 status = RA_NOMATCH; | |
270 | 5390 regstack_pop(&scan); |
180 | 5391 } |
5392 else | |
5393 { | |
5394 /* Prepare to try a branch. */ | |
5395 rp->rs_scan = regnext(scan); | |
233 | 5396 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5397 scan = OPERAND(scan); |
5398 } | |
5399 } | |
5400 break; | |
5401 | |
5402 case RS_BRCPLX_MORE: | |
5403 /* Pop the state. Restore pointers when there is no match. */ | |
5404 if (status == RA_NOMATCH) | |
5405 { | |
233 | 5406 reg_restore(&rp->rs_un.regsave, &backpos); |
180 | 5407 --brace_count[rp->rs_no]; /* decrement match count */ |
5408 } | |
270 | 5409 regstack_pop(&scan); |
180 | 5410 break; |
5411 | |
5412 case RS_BRCPLX_LONG: | |
5413 /* Pop the state. Restore pointers when there is no match. */ | |
5414 if (status == RA_NOMATCH) | |
5415 { | |
5416 /* There was no match, but we did find enough matches. */ | |
233 | 5417 reg_restore(&rp->rs_un.regsave, &backpos); |
180 | 5418 --brace_count[rp->rs_no]; |
5419 /* continue with the items after "\{}" */ | |
5420 status = RA_CONT; | |
5421 } | |
270 | 5422 regstack_pop(&scan); |
180 | 5423 if (status == RA_CONT) |
5424 scan = regnext(scan); | |
5425 break; | |
5426 | |
5427 case RS_BRCPLX_SHORT: | |
5428 /* Pop the state. Restore pointers when there is no match. */ | |
5429 if (status == RA_NOMATCH) | |
5430 /* There was no match, try to match one more item. */ | |
233 | 5431 reg_restore(&rp->rs_un.regsave, &backpos); |
270 | 5432 regstack_pop(&scan); |
180 | 5433 if (status == RA_NOMATCH) |
5434 { | |
5435 scan = OPERAND(scan); | |
5436 status = RA_CONT; | |
5437 } | |
5438 break; | |
5439 | |
5440 case RS_NOMATCH: | |
5441 /* Pop the state. If the operand matches for NOMATCH or | |
5442 * doesn't match for MATCH/SUBPAT, we fail. Otherwise backup, | |
5443 * except for SUBPAT, and continue with the next item. */ | |
5444 if (status == (rp->rs_no == NOMATCH ? RA_MATCH : RA_NOMATCH)) | |
5445 status = RA_NOMATCH; | |
5446 else | |
5447 { | |
5448 status = RA_CONT; | |
233 | 5449 if (rp->rs_no != SUBPAT) /* zero-width */ |
5450 reg_restore(&rp->rs_un.regsave, &backpos); | |
180 | 5451 } |
270 | 5452 regstack_pop(&scan); |
180 | 5453 if (status == RA_CONT) |
5454 scan = regnext(scan); | |
5455 break; | |
5456 | |
5457 case RS_BEHIND1: | |
5458 if (status == RA_NOMATCH) | |
5459 { | |
270 | 5460 regstack_pop(&scan); |
180 | 5461 regstack.ga_len -= sizeof(regbehind_T); |
5462 } | |
5463 else | |
5464 { | |
5465 /* The stuff after BEHIND/NOBEHIND matches. Now try if | |
5466 * the behind part does (not) match before the current | |
5467 * position in the input. This must be done at every | |
5468 * position in the input and checking if the match ends at | |
5469 * the current position. */ | |
5470 | |
5471 /* save the position after the found match for next */ | |
233 | 5472 reg_save(&(((regbehind_T *)rp) - 1)->save_after, &backpos); |
180 | 5473 |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5474 /* Start looking for a match with operand at the current |
1209 | 5475 * position. Go back one character until we find the |
180 | 5476 * result, hitting the start of the line or the previous |
5477 * line (for multi-line matching). | |
5478 * Set behind_pos to where the match should end, BHPOS | |
5479 * will match it. Save the current value. */ | |
5480 (((regbehind_T *)rp) - 1)->save_behind = behind_pos; | |
5481 behind_pos = rp->rs_un.regsave; | |
5482 | |
5483 rp->rs_state = RS_BEHIND2; | |
5484 | |
233 | 5485 reg_restore(&rp->rs_un.regsave, &backpos); |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5486 scan = OPERAND(rp->rs_scan) + 4; |
180 | 5487 } |
5488 break; | |
5489 | |
5490 case RS_BEHIND2: | |
5491 /* | |
5492 * Looping for BEHIND / NOBEHIND match. | |
5493 */ | |
5494 if (status == RA_MATCH && reg_save_equal(&behind_pos)) | |
5495 { | |
5496 /* found a match that ends where "next" started */ | |
5497 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
5498 if (rp->rs_no == BEHIND) | |
233 | 5499 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
5500 &backpos); | |
180 | 5501 else |
1579 | 5502 { |
5503 /* But we didn't want a match. Need to restore the | |
5504 * subexpr, because what follows matched, so they have | |
5505 * been set. */ | |
180 | 5506 status = RA_NOMATCH; |
1579 | 5507 restore_subexpr(((regbehind_T *)rp) - 1); |
5508 } | |
270 | 5509 regstack_pop(&scan); |
180 | 5510 regstack.ga_len -= sizeof(regbehind_T); |
5511 } | |
5512 else | |
5513 { | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5514 long limit; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5515 |
1579 | 5516 /* No match or a match that doesn't end where we want it: Go |
5517 * back one character. May go to previous line once. */ | |
180 | 5518 no = OK; |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5519 limit = OPERAND_MIN(rp->rs_scan); |
180 | 5520 if (REG_MULTI) |
5521 { | |
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5522 if (limit > 0 |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5523 && ((rp->rs_un.regsave.rs_u.pos.lnum |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5524 < behind_pos.rs_u.pos.lnum |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5525 ? (colnr_T)STRLEN(regline) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5526 : behind_pos.rs_u.pos.col) |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5527 - 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
|
5528 no = FAIL; |
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5529 else if (rp->rs_un.regsave.rs_u.pos.col == 0) |
180 | 5530 { |
5531 if (rp->rs_un.regsave.rs_u.pos.lnum | |
5532 < behind_pos.rs_u.pos.lnum | |
5533 || reg_getline( | |
5534 --rp->rs_un.regsave.rs_u.pos.lnum) | |
5535 == NULL) | |
5536 no = FAIL; | |
5537 else | |
5538 { | |
233 | 5539 reg_restore(&rp->rs_un.regsave, &backpos); |
180 | 5540 rp->rs_un.regsave.rs_u.pos.col = |
5541 (colnr_T)STRLEN(regline); | |
5542 } | |
5543 } | |
5544 else | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5545 { |
4176 | 5546 #ifdef FEAT_MBYTE |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5547 if (has_mbyte) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5548 rp->rs_un.regsave.rs_u.pos.col -= |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5549 (*mb_head_off)(regline, regline |
4176 | 5550 + 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
|
5551 else |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5552 #endif |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5553 --rp->rs_un.regsave.rs_u.pos.col; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5554 } |
180 | 5555 } |
5556 else | |
5557 { | |
5558 if (rp->rs_un.regsave.rs_u.ptr == regline) | |
5559 no = FAIL; | |
5560 else | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5561 { |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5562 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
|
5563 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
|
5564 - rp->rs_un.regsave.rs_u.ptr) > limit) |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5565 no = FAIL; |
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5566 } |
180 | 5567 } |
5568 if (no == OK) | |
5569 { | |
5570 /* Advanced, prepare for finding match again. */ | |
233 | 5571 reg_restore(&rp->rs_un.regsave, &backpos); |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5572 scan = OPERAND(rp->rs_scan) + 4; |
1579 | 5573 if (status == RA_MATCH) |
5574 { | |
5575 /* We did match, so subexpr may have been changed, | |
5576 * need to restore them for the next try. */ | |
5577 status = RA_NOMATCH; | |
5578 restore_subexpr(((regbehind_T *)rp) - 1); | |
5579 } | |
180 | 5580 } |
5581 else | |
5582 { | |
5583 /* Can't advance. For NOBEHIND that's a match. */ | |
5584 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
5585 if (rp->rs_no == NOBEHIND) | |
5586 { | |
233 | 5587 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
5588 &backpos); | |
180 | 5589 status = RA_MATCH; |
5590 } | |
5591 else | |
1579 | 5592 { |
5593 /* We do want a proper match. Need to restore the | |
5594 * subexpr if we had a match, because they may have | |
5595 * been set. */ | |
5596 if (status == RA_MATCH) | |
5597 { | |
5598 status = RA_NOMATCH; | |
5599 restore_subexpr(((regbehind_T *)rp) - 1); | |
5600 } | |
5601 } | |
270 | 5602 regstack_pop(&scan); |
180 | 5603 regstack.ga_len -= sizeof(regbehind_T); |
5604 } | |
5605 } | |
5606 break; | |
5607 | |
5608 case RS_STAR_LONG: | |
5609 case RS_STAR_SHORT: | |
5610 { | |
5611 regstar_T *rst = ((regstar_T *)rp) - 1; | |
5612 | |
5613 if (status == RA_MATCH) | |
5614 { | |
270 | 5615 regstack_pop(&scan); |
180 | 5616 regstack.ga_len -= sizeof(regstar_T); |
5617 break; | |
5618 } | |
5619 | |
5620 /* Tried once already, restore input pointers. */ | |
5621 if (status != RA_BREAK) | |
233 | 5622 reg_restore(&rp->rs_un.regsave, &backpos); |
180 | 5623 |
5624 /* Repeat until we found a position where it could match. */ | |
5625 for (;;) | |
5626 { | |
5627 if (status != RA_BREAK) | |
5628 { | |
5629 /* Tried first position already, advance. */ | |
5630 if (rp->rs_state == RS_STAR_LONG) | |
5631 { | |
685 | 5632 /* Trying for longest match, but couldn't or |
5633 * didn't match -- back up one char. */ | |
180 | 5634 if (--rst->count < rst->minval) |
5635 break; | |
5636 if (reginput == regline) | |
5637 { | |
5638 /* backup to last char of previous line */ | |
5639 --reglnum; | |
5640 regline = reg_getline(reglnum); | |
5641 /* Just in case regrepeat() didn't count | |
5642 * right. */ | |
5643 if (regline == NULL) | |
5644 break; | |
5645 reginput = regline + STRLEN(regline); | |
5646 fast_breakcheck(); | |
5647 } | |
5648 else | |
5649 mb_ptr_back(regline, reginput); | |
5650 } | |
5651 else | |
5652 { | |
5653 /* Range is backwards, use shortest match first. | |
5654 * Careful: maxval and minval are exchanged! | |
5655 * Couldn't or didn't match: try advancing one | |
5656 * char. */ | |
5657 if (rst->count == rst->minval | |
5658 || regrepeat(OPERAND(rp->rs_scan), 1L) == 0) | |
5659 break; | |
5660 ++rst->count; | |
5661 } | |
5662 if (got_int) | |
5663 break; | |
5664 } | |
5665 else | |
5666 status = RA_NOMATCH; | |
5667 | |
5668 /* If it could match, try it. */ | |
5669 if (rst->nextb == NUL || *reginput == rst->nextb | |
5670 || *reginput == rst->nextb_ic) | |
5671 { | |
233 | 5672 reg_save(&rp->rs_un.regsave, &backpos); |
180 | 5673 scan = regnext(rp->rs_scan); |
5674 status = RA_CONT; | |
5675 break; | |
5676 } | |
5677 } | |
5678 if (status != RA_CONT) | |
5679 { | |
5680 /* Failed. */ | |
270 | 5681 regstack_pop(&scan); |
180 | 5682 regstack.ga_len -= sizeof(regstar_T); |
5683 status = RA_NOMATCH; | |
5684 } | |
5685 } | |
5686 break; | |
5687 } | |
5688 | |
685 | 5689 /* If we want to continue the inner loop or didn't pop a state |
5690 * continue matching loop */ | |
180 | 5691 if (status == RA_CONT || rp == (regitem_T *) |
5692 ((char *)regstack.ga_data + regstack.ga_len) - 1) | |
5693 break; | |
5694 } | |
5695 | |
189 | 5696 /* May need to continue with the inner loop, starting at "scan". */ |
180 | 5697 if (status == RA_CONT) |
5698 continue; | |
5699 | |
5700 /* | |
5701 * If the regstack is empty or something failed we are done. | |
5702 */ | |
5703 if (regstack.ga_len == 0 || status == RA_FAIL) | |
5704 { | |
5705 if (scan == NULL) | |
5706 { | |
5707 /* | |
5708 * We get here only if there's trouble -- normally "case END" is | |
5709 * the terminating point. | |
5710 */ | |
5711 EMSG(_(e_re_corr)); | |
7 | 5712 #ifdef DEBUG |
180 | 5713 printf("Premature EOL\n"); |
7 | 5714 #endif |
180 | 5715 } |
189 | 5716 if (status == RA_FAIL) |
5717 got_int = TRUE; | |
180 | 5718 return (status == RA_MATCH); |
5719 } | |
5720 | |
5721 } /* End of loop until the regstack is empty. */ | |
5722 | |
5723 /* NOTREACHED */ | |
5724 } | |
5725 | |
5726 /* | |
5727 * Push an item onto the regstack. | |
5728 * Returns pointer to new item. Returns NULL when out of memory. | |
5729 */ | |
5730 static regitem_T * | |
270 | 5731 regstack_push(state, scan) |
180 | 5732 regstate_T state; |
5733 char_u *scan; | |
5734 { | |
5735 regitem_T *rp; | |
5736 | |
270 | 5737 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
189 | 5738 { |
5739 EMSG(_(e_maxmempat)); | |
5740 return NULL; | |
5741 } | |
270 | 5742 if (ga_grow(®stack, sizeof(regitem_T)) == FAIL) |
180 | 5743 return NULL; |
5744 | |
270 | 5745 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len); |
180 | 5746 rp->rs_state = state; |
5747 rp->rs_scan = scan; | |
5748 | |
270 | 5749 regstack.ga_len += sizeof(regitem_T); |
180 | 5750 return rp; |
5751 } | |
5752 | |
5753 /* | |
5754 * Pop an item from the regstack. | |
5755 */ | |
5756 static void | |
270 | 5757 regstack_pop(scan) |
180 | 5758 char_u **scan; |
5759 { | |
5760 regitem_T *rp; | |
5761 | |
270 | 5762 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; |
180 | 5763 *scan = rp->rs_scan; |
5764 | |
270 | 5765 regstack.ga_len -= sizeof(regitem_T); |
7 | 5766 } |
5767 | |
5768 /* | |
5769 * regrepeat - repeatedly match something simple, return how many. | |
5770 * Advances reginput (and reglnum) to just after the matched chars. | |
5771 */ | |
5772 static int | |
5773 regrepeat(p, maxcount) | |
5774 char_u *p; | |
5775 long maxcount; /* maximum number of matches allowed */ | |
5776 { | |
5777 long count = 0; | |
5778 char_u *scan; | |
5779 char_u *opnd; | |
5780 int mask; | |
5781 int testval = 0; | |
5782 | |
5783 scan = reginput; /* Make local copy of reginput for speed. */ | |
5784 opnd = OPERAND(p); | |
5785 switch (OP(p)) | |
5786 { | |
5787 case ANY: | |
5788 case ANY + ADD_NL: | |
5789 while (count < maxcount) | |
5790 { | |
5791 /* Matching anything means we continue until end-of-line (or | |
5792 * end-of-file for ANY + ADD_NL), only limited by maxcount. */ | |
5793 while (*scan != NUL && count < maxcount) | |
5794 { | |
5795 ++count; | |
39 | 5796 mb_ptr_adv(scan); |
7 | 5797 } |
1018 | 5798 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5799 || reg_line_lbr || count == maxcount) | |
7 | 5800 break; |
5801 ++count; /* count the line-break */ | |
5802 reg_nextline(); | |
5803 scan = reginput; | |
5804 if (got_int) | |
5805 break; | |
5806 } | |
5807 break; | |
5808 | |
5809 case IDENT: | |
5810 case IDENT + ADD_NL: | |
5811 testval = TRUE; | |
5812 /*FALLTHROUGH*/ | |
5813 case SIDENT: | |
5814 case SIDENT + ADD_NL: | |
5815 while (count < maxcount) | |
5816 { | |
4466 | 5817 if (vim_isIDc(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
7 | 5818 { |
39 | 5819 mb_ptr_adv(scan); |
7 | 5820 } |
5821 else if (*scan == NUL) | |
5822 { | |
1018 | 5823 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5824 || reg_line_lbr) | |
7 | 5825 break; |
5826 reg_nextline(); | |
5827 scan = reginput; | |
5828 if (got_int) | |
5829 break; | |
5830 } | |
5831 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
5832 ++scan; | |
5833 else | |
5834 break; | |
5835 ++count; | |
5836 } | |
5837 break; | |
5838 | |
5839 case KWORD: | |
5840 case KWORD + ADD_NL: | |
5841 testval = TRUE; | |
5842 /*FALLTHROUGH*/ | |
5843 case SKWORD: | |
5844 case SKWORD + ADD_NL: | |
5845 while (count < maxcount) | |
5846 { | |
4069 | 5847 if (vim_iswordp_buf(scan, reg_buf) |
5848 && (testval || !VIM_ISDIGIT(*scan))) | |
7 | 5849 { |
39 | 5850 mb_ptr_adv(scan); |
7 | 5851 } |
5852 else if (*scan == NUL) | |
5853 { | |
1018 | 5854 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5855 || reg_line_lbr) | |
7 | 5856 break; |
5857 reg_nextline(); | |
5858 scan = reginput; | |
5859 if (got_int) | |
5860 break; | |
5861 } | |
5862 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
5863 ++scan; | |
5864 else | |
5865 break; | |
5866 ++count; | |
5867 } | |
5868 break; | |
5869 | |
5870 case FNAME: | |
5871 case FNAME + ADD_NL: | |
5872 testval = TRUE; | |
5873 /*FALLTHROUGH*/ | |
5874 case SFNAME: | |
5875 case SFNAME + ADD_NL: | |
5876 while (count < maxcount) | |
5877 { | |
4466 | 5878 if (vim_isfilec(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
7 | 5879 { |
39 | 5880 mb_ptr_adv(scan); |
7 | 5881 } |
5882 else if (*scan == NUL) | |
5883 { | |
1018 | 5884 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5885 || reg_line_lbr) | |
7 | 5886 break; |
5887 reg_nextline(); | |
5888 scan = reginput; | |
5889 if (got_int) | |
5890 break; | |
5891 } | |
5892 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
5893 ++scan; | |
5894 else | |
5895 break; | |
5896 ++count; | |
5897 } | |
5898 break; | |
5899 | |
5900 case PRINT: | |
5901 case PRINT + ADD_NL: | |
5902 testval = TRUE; | |
5903 /*FALLTHROUGH*/ | |
5904 case SPRINT: | |
5905 case SPRINT + ADD_NL: | |
5906 while (count < maxcount) | |
5907 { | |
5908 if (*scan == NUL) | |
5909 { | |
1018 | 5910 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5911 || reg_line_lbr) | |
7 | 5912 break; |
5913 reg_nextline(); | |
5914 scan = reginput; | |
5915 if (got_int) | |
5916 break; | |
5917 } | |
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
5918 else if (vim_isprintc(PTR2CHAR(scan)) == 1 |
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
5919 && (testval || !VIM_ISDIGIT(*scan))) |
7 | 5920 { |
39 | 5921 mb_ptr_adv(scan); |
7 | 5922 } |
5923 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
5924 ++scan; | |
5925 else | |
5926 break; | |
5927 ++count; | |
5928 } | |
5929 break; | |
5930 | |
5931 case WHITE: | |
5932 case WHITE + ADD_NL: | |
5933 testval = mask = RI_WHITE; | |
5934 do_class: | |
5935 while (count < maxcount) | |
5936 { | |
5937 #ifdef FEAT_MBYTE | |
5938 int l; | |
5939 #endif | |
5940 if (*scan == NUL) | |
5941 { | |
1018 | 5942 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
5943 || reg_line_lbr) | |
7 | 5944 break; |
5945 reg_nextline(); | |
5946 scan = reginput; | |
5947 if (got_int) | |
5948 break; | |
5949 } | |
5950 #ifdef FEAT_MBYTE | |
474 | 5951 else if (has_mbyte && (l = (*mb_ptr2len)(scan)) > 1) |
7 | 5952 { |
5953 if (testval != 0) | |
5954 break; | |
5955 scan += l; | |
5956 } | |
5957 #endif | |
5958 else if ((class_tab[*scan] & mask) == testval) | |
5959 ++scan; | |
5960 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
5961 ++scan; | |
5962 else | |
5963 break; | |
5964 ++count; | |
5965 } | |
5966 break; | |
5967 | |
5968 case NWHITE: | |
5969 case NWHITE + ADD_NL: | |
5970 mask = RI_WHITE; | |
5971 goto do_class; | |
5972 case DIGIT: | |
5973 case DIGIT + ADD_NL: | |
5974 testval = mask = RI_DIGIT; | |
5975 goto do_class; | |
5976 case NDIGIT: | |
5977 case NDIGIT + ADD_NL: | |
5978 mask = RI_DIGIT; | |
5979 goto do_class; | |
5980 case HEX: | |
5981 case HEX + ADD_NL: | |
5982 testval = mask = RI_HEX; | |
5983 goto do_class; | |
5984 case NHEX: | |
5985 case NHEX + ADD_NL: | |
5986 mask = RI_HEX; | |
5987 goto do_class; | |
5988 case OCTAL: | |
5989 case OCTAL + ADD_NL: | |
5990 testval = mask = RI_OCTAL; | |
5991 goto do_class; | |
5992 case NOCTAL: | |
5993 case NOCTAL + ADD_NL: | |
5994 mask = RI_OCTAL; | |
5995 goto do_class; | |
5996 case WORD: | |
5997 case WORD + ADD_NL: | |
5998 testval = mask = RI_WORD; | |
5999 goto do_class; | |
6000 case NWORD: | |
6001 case NWORD + ADD_NL: | |
6002 mask = RI_WORD; | |
6003 goto do_class; | |
6004 case HEAD: | |
6005 case HEAD + ADD_NL: | |
6006 testval = mask = RI_HEAD; | |
6007 goto do_class; | |
6008 case NHEAD: | |
6009 case NHEAD + ADD_NL: | |
6010 mask = RI_HEAD; | |
6011 goto do_class; | |
6012 case ALPHA: | |
6013 case ALPHA + ADD_NL: | |
6014 testval = mask = RI_ALPHA; | |
6015 goto do_class; | |
6016 case NALPHA: | |
6017 case NALPHA + ADD_NL: | |
6018 mask = RI_ALPHA; | |
6019 goto do_class; | |
6020 case LOWER: | |
6021 case LOWER + ADD_NL: | |
6022 testval = mask = RI_LOWER; | |
6023 goto do_class; | |
6024 case NLOWER: | |
6025 case NLOWER + ADD_NL: | |
6026 mask = RI_LOWER; | |
6027 goto do_class; | |
6028 case UPPER: | |
6029 case UPPER + ADD_NL: | |
6030 testval = mask = RI_UPPER; | |
6031 goto do_class; | |
6032 case NUPPER: | |
6033 case NUPPER + ADD_NL: | |
6034 mask = RI_UPPER; | |
6035 goto do_class; | |
6036 | |
6037 case EXACTLY: | |
6038 { | |
6039 int cu, cl; | |
6040 | |
6041 /* This doesn't do a multi-byte character, because a MULTIBYTECODE | |
1347 | 6042 * would have been used for it. It does handle single-byte |
6043 * characters, such as latin1. */ | |
7 | 6044 if (ireg_ic) |
6045 { | |
1347 | 6046 cu = MB_TOUPPER(*opnd); |
6047 cl = MB_TOLOWER(*opnd); | |
7 | 6048 while (count < maxcount && (*scan == cu || *scan == cl)) |
6049 { | |
6050 count++; | |
6051 scan++; | |
6052 } | |
6053 } | |
6054 else | |
6055 { | |
6056 cu = *opnd; | |
6057 while (count < maxcount && *scan == cu) | |
6058 { | |
6059 count++; | |
6060 scan++; | |
6061 } | |
6062 } | |
6063 break; | |
6064 } | |
6065 | |
6066 #ifdef FEAT_MBYTE | |
6067 case MULTIBYTECODE: | |
6068 { | |
6069 int i, len, cf = 0; | |
6070 | |
6071 /* Safety check (just in case 'encoding' was changed since | |
6072 * compiling the program). */ | |
474 | 6073 if ((len = (*mb_ptr2len)(opnd)) > 1) |
7 | 6074 { |
6075 if (ireg_ic && enc_utf8) | |
6076 cf = utf_fold(utf_ptr2char(opnd)); | |
6077 while (count < maxcount) | |
6078 { | |
6079 for (i = 0; i < len; ++i) | |
6080 if (opnd[i] != scan[i]) | |
6081 break; | |
6082 if (i < len && (!ireg_ic || !enc_utf8 | |
6083 || utf_fold(utf_ptr2char(scan)) != cf)) | |
6084 break; | |
6085 scan += len; | |
6086 ++count; | |
6087 } | |
6088 } | |
6089 } | |
6090 break; | |
6091 #endif | |
6092 | |
6093 case ANYOF: | |
6094 case ANYOF + ADD_NL: | |
6095 testval = TRUE; | |
6096 /*FALLTHROUGH*/ | |
6097 | |
6098 case ANYBUT: | |
6099 case ANYBUT + ADD_NL: | |
6100 while (count < maxcount) | |
6101 { | |
6102 #ifdef FEAT_MBYTE | |
6103 int len; | |
6104 #endif | |
6105 if (*scan == NUL) | |
6106 { | |
1018 | 6107 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
6108 || reg_line_lbr) | |
7 | 6109 break; |
6110 reg_nextline(); | |
6111 scan = reginput; | |
6112 if (got_int) | |
6113 break; | |
6114 } | |
6115 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
6116 ++scan; | |
6117 #ifdef FEAT_MBYTE | |
474 | 6118 else if (has_mbyte && (len = (*mb_ptr2len)(scan)) > 1) |
7 | 6119 { |
6120 if ((cstrchr(opnd, (*mb_ptr2char)(scan)) == NULL) == testval) | |
6121 break; | |
6122 scan += len; | |
6123 } | |
6124 #endif | |
6125 else | |
6126 { | |
6127 if ((cstrchr(opnd, *scan) == NULL) == testval) | |
6128 break; | |
6129 ++scan; | |
6130 } | |
6131 ++count; | |
6132 } | |
6133 break; | |
6134 | |
6135 case NEWL: | |
6136 while (count < maxcount | |
1018 | 6137 && ((*scan == NUL && reglnum <= reg_maxline && !reg_line_lbr |
6138 && REG_MULTI) || (*scan == '\n' && reg_line_lbr))) | |
7 | 6139 { |
6140 count++; | |
6141 if (reg_line_lbr) | |
6142 ADVANCE_REGINPUT(); | |
6143 else | |
6144 reg_nextline(); | |
6145 scan = reginput; | |
6146 if (got_int) | |
6147 break; | |
6148 } | |
6149 break; | |
6150 | |
6151 default: /* Oh dear. Called inappropriately. */ | |
6152 EMSG(_(e_re_corr)); | |
6153 #ifdef DEBUG | |
6154 printf("Called regrepeat with op code %d\n", OP(p)); | |
6155 #endif | |
6156 break; | |
6157 } | |
6158 | |
6159 reginput = scan; | |
6160 | |
6161 return (int)count; | |
6162 } | |
6163 | |
6164 /* | |
6165 * regnext - dig the "next" pointer out of a node | |
2010 | 6166 * Returns NULL when calculating size, when there is no next item and when |
6167 * there is an error. | |
7 | 6168 */ |
6169 static char_u * | |
6170 regnext(p) | |
6171 char_u *p; | |
6172 { | |
6173 int offset; | |
6174 | |
2010 | 6175 if (p == JUST_CALC_SIZE || reg_toolong) |
7 | 6176 return NULL; |
6177 | |
6178 offset = NEXT(p); | |
6179 if (offset == 0) | |
6180 return NULL; | |
6181 | |
233 | 6182 if (OP(p) == BACK) |
7 | 6183 return p - offset; |
6184 else | |
6185 return p + offset; | |
6186 } | |
6187 | |
6188 /* | |
6189 * Check the regexp program for its magic number. | |
6190 * Return TRUE if it's wrong. | |
6191 */ | |
6192 static int | |
6193 prog_magic_wrong() | |
6194 { | |
4444 | 6195 regprog_T *prog; |
6196 | |
6197 prog = REG_MULTI ? reg_mmatch->regprog : reg_match->regprog; | |
6198 if (prog->engine == &nfa_regengine) | |
6199 /* For NFA matcher we don't check the magic */ | |
6200 return FALSE; | |
6201 | |
6202 if (UCHARAT(((bt_regprog_T *)prog)->program) != REGMAGIC) | |
7 | 6203 { |
6204 EMSG(_(e_re_corr)); | |
6205 return TRUE; | |
6206 } | |
6207 return FALSE; | |
6208 } | |
6209 | |
6210 /* | |
6211 * Cleanup the subexpressions, if this wasn't done yet. | |
6212 * This construction is used to clear the subexpressions only when they are | |
6213 * used (to increase speed). | |
6214 */ | |
6215 static void | |
6216 cleanup_subexpr() | |
6217 { | |
6218 if (need_clear_subexpr) | |
6219 { | |
6220 if (REG_MULTI) | |
6221 { | |
6222 /* Use 0xff to set lnum to -1 */ | |
6223 vim_memset(reg_startpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
6224 vim_memset(reg_endpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
6225 } | |
6226 else | |
6227 { | |
6228 vim_memset(reg_startp, 0, sizeof(char_u *) * NSUBEXP); | |
6229 vim_memset(reg_endp, 0, sizeof(char_u *) * NSUBEXP); | |
6230 } | |
6231 need_clear_subexpr = FALSE; | |
6232 } | |
6233 } | |
6234 | |
6235 #ifdef FEAT_SYN_HL | |
6236 static void | |
6237 cleanup_zsubexpr() | |
6238 { | |
6239 if (need_clear_zsubexpr) | |
6240 { | |
6241 if (REG_MULTI) | |
6242 { | |
6243 /* Use 0xff to set lnum to -1 */ | |
6244 vim_memset(reg_startzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
6245 vim_memset(reg_endzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
6246 } | |
6247 else | |
6248 { | |
6249 vim_memset(reg_startzp, 0, sizeof(char_u *) * NSUBEXP); | |
6250 vim_memset(reg_endzp, 0, sizeof(char_u *) * NSUBEXP); | |
6251 } | |
6252 need_clear_zsubexpr = FALSE; | |
6253 } | |
6254 } | |
6255 #endif | |
6256 | |
6257 /* | |
1579 | 6258 * Save the current subexpr to "bp", so that they can be restored |
6259 * later by restore_subexpr(). | |
6260 */ | |
6261 static void | |
6262 save_subexpr(bp) | |
6263 regbehind_T *bp; | |
6264 { | |
6265 int i; | |
6266 | |
1602 | 6267 /* When "need_clear_subexpr" is set we don't need to save the values, only |
6268 * remember that this flag needs to be set again when restoring. */ | |
6269 bp->save_need_clear_subexpr = need_clear_subexpr; | |
6270 if (!need_clear_subexpr) | |
1579 | 6271 { |
1602 | 6272 for (i = 0; i < NSUBEXP; ++i) |
1579 | 6273 { |
1602 | 6274 if (REG_MULTI) |
6275 { | |
6276 bp->save_start[i].se_u.pos = reg_startpos[i]; | |
6277 bp->save_end[i].se_u.pos = reg_endpos[i]; | |
6278 } | |
6279 else | |
6280 { | |
6281 bp->save_start[i].se_u.ptr = reg_startp[i]; | |
6282 bp->save_end[i].se_u.ptr = reg_endp[i]; | |
6283 } | |
1579 | 6284 } |
6285 } | |
6286 } | |
6287 | |
6288 /* | |
6289 * Restore the subexpr from "bp". | |
6290 */ | |
6291 static void | |
6292 restore_subexpr(bp) | |
6293 regbehind_T *bp; | |
6294 { | |
6295 int i; | |
6296 | |
1602 | 6297 /* Only need to restore saved values when they are not to be cleared. */ |
6298 need_clear_subexpr = bp->save_need_clear_subexpr; | |
6299 if (!need_clear_subexpr) | |
1579 | 6300 { |
1602 | 6301 for (i = 0; i < NSUBEXP; ++i) |
1579 | 6302 { |
1602 | 6303 if (REG_MULTI) |
6304 { | |
6305 reg_startpos[i] = bp->save_start[i].se_u.pos; | |
6306 reg_endpos[i] = bp->save_end[i].se_u.pos; | |
6307 } | |
6308 else | |
6309 { | |
6310 reg_startp[i] = bp->save_start[i].se_u.ptr; | |
6311 reg_endp[i] = bp->save_end[i].se_u.ptr; | |
6312 } | |
1579 | 6313 } |
6314 } | |
6315 } | |
6316 | |
6317 /* | |
7 | 6318 * Advance reglnum, regline and reginput to the next line. |
6319 */ | |
6320 static void | |
6321 reg_nextline() | |
6322 { | |
6323 regline = reg_getline(++reglnum); | |
6324 reginput = regline; | |
6325 fast_breakcheck(); | |
6326 } | |
6327 | |
6328 /* | |
6329 * Save the input line and position in a regsave_T. | |
6330 */ | |
6331 static void | |
233 | 6332 reg_save(save, gap) |
7 | 6333 regsave_T *save; |
233 | 6334 garray_T *gap; |
7 | 6335 { |
6336 if (REG_MULTI) | |
6337 { | |
6338 save->rs_u.pos.col = (colnr_T)(reginput - regline); | |
6339 save->rs_u.pos.lnum = reglnum; | |
6340 } | |
6341 else | |
6342 save->rs_u.ptr = reginput; | |
233 | 6343 save->rs_len = gap->ga_len; |
7 | 6344 } |
6345 | |
6346 /* | |
6347 * Restore the input line and position from a regsave_T. | |
6348 */ | |
6349 static void | |
233 | 6350 reg_restore(save, gap) |
7 | 6351 regsave_T *save; |
233 | 6352 garray_T *gap; |
7 | 6353 { |
6354 if (REG_MULTI) | |
6355 { | |
6356 if (reglnum != save->rs_u.pos.lnum) | |
6357 { | |
6358 /* only call reg_getline() when the line number changed to save | |
6359 * a bit of time */ | |
6360 reglnum = save->rs_u.pos.lnum; | |
6361 regline = reg_getline(reglnum); | |
6362 } | |
6363 reginput = regline + save->rs_u.pos.col; | |
6364 } | |
6365 else | |
6366 reginput = save->rs_u.ptr; | |
233 | 6367 gap->ga_len = save->rs_len; |
7 | 6368 } |
6369 | |
6370 /* | |
6371 * Return TRUE if current position is equal to saved position. | |
6372 */ | |
6373 static int | |
6374 reg_save_equal(save) | |
6375 regsave_T *save; | |
6376 { | |
6377 if (REG_MULTI) | |
6378 return reglnum == save->rs_u.pos.lnum | |
6379 && reginput == regline + save->rs_u.pos.col; | |
6380 return reginput == save->rs_u.ptr; | |
6381 } | |
6382 | |
6383 /* | |
6384 * Tentatively set the sub-expression start to the current position (after | |
6385 * calling regmatch() they will have changed). Need to save the existing | |
6386 * values for when there is no match. | |
6387 * Use se_save() to use pointer (save_se_multi()) or position (save_se_one()), | |
6388 * depending on REG_MULTI. | |
6389 */ | |
6390 static void | |
6391 save_se_multi(savep, posp) | |
6392 save_se_T *savep; | |
6393 lpos_T *posp; | |
6394 { | |
6395 savep->se_u.pos = *posp; | |
6396 posp->lnum = reglnum; | |
6397 posp->col = (colnr_T)(reginput - regline); | |
6398 } | |
6399 | |
6400 static void | |
6401 save_se_one(savep, pp) | |
6402 save_se_T *savep; | |
6403 char_u **pp; | |
6404 { | |
6405 savep->se_u.ptr = *pp; | |
6406 *pp = reginput; | |
6407 } | |
6408 | |
6409 /* | |
6410 * Compare a number with the operand of RE_LNUM, RE_COL or RE_VCOL. | |
6411 */ | |
6412 static int | |
6413 re_num_cmp(val, scan) | |
6414 long_u val; | |
6415 char_u *scan; | |
6416 { | |
6417 long_u n = OPERAND_MIN(scan); | |
6418 | |
6419 if (OPERAND_CMP(scan) == '>') | |
6420 return val > n; | |
6421 if (OPERAND_CMP(scan) == '<') | |
6422 return val < n; | |
6423 return val == n; | |
6424 } | |
6425 | |
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6426 /* |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6427 * Check whether a backreference matches. |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6428 * Returns RA_FAIL, RA_NOMATCH or RA_MATCH. |
5504 | 6429 * If "bytelen" is not NULL, it is set to the byte length of the match in the |
6430 * last line. | |
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6431 */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6432 static int |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6433 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
|
6434 linenr_T start_lnum; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6435 colnr_T start_col; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6436 linenr_T end_lnum; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6437 colnr_T end_col; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6438 int *bytelen; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6439 { |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6440 linenr_T clnum = start_lnum; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6441 colnr_T ccol = start_col; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6442 int len; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6443 char_u *p; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6444 |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6445 if (bytelen != NULL) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6446 *bytelen = 0; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6447 for (;;) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6448 { |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6449 /* 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
|
6450 * Slow! */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6451 if (regline != reg_tofree) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6452 { |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6453 len = (int)STRLEN(regline); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6454 if (reg_tofree == NULL || len >= (int)reg_tofreelen) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6455 { |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6456 len += 50; /* get some extra */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6457 vim_free(reg_tofree); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6458 reg_tofree = alloc(len); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6459 if (reg_tofree == NULL) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6460 return RA_FAIL; /* out of memory!*/ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6461 reg_tofreelen = len; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6462 } |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6463 STRCPY(reg_tofree, regline); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6464 reginput = reg_tofree + (reginput - regline); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6465 regline = reg_tofree; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6466 } |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6467 |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6468 /* Get the line to compare with. */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6469 p = reg_getline(clnum); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6470 if (clnum == end_lnum) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6471 len = end_col - ccol; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6472 else |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6473 len = (int)STRLEN(p + ccol); |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6474 |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6475 if (cstrncmp(p + ccol, reginput, &len) != 0) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6476 return RA_NOMATCH; /* doesn't match */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6477 if (bytelen != NULL) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6478 *bytelen += len; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6479 if (clnum == end_lnum) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6480 break; /* match and at end! */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6481 if (reglnum >= reg_maxline) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6482 return RA_NOMATCH; /* text too short */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6483 |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6484 /* Advance to next line. */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6485 reg_nextline(); |
5504 | 6486 if (bytelen != NULL) |
6487 *bytelen = 0; | |
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6488 ++clnum; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6489 ccol = 0; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6490 if (got_int) |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6491 return RA_FAIL; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6492 } |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6493 |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6494 /* 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
|
6495 * that should not matter. */ |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6496 return RA_MATCH; |
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6497 } |
7 | 6498 |
4444 | 6499 #ifdef BT_REGEXP_DUMP |
7 | 6500 |
6501 /* | |
6502 * regdump - dump a regexp onto stdout in vaguely comprehensible form | |
6503 */ | |
6504 static void | |
6505 regdump(pattern, r) | |
6506 char_u *pattern; | |
4444 | 6507 bt_regprog_T *r; |
7 | 6508 { |
6509 char_u *s; | |
6510 int op = EXACTLY; /* Arbitrary non-END op. */ | |
6511 char_u *next; | |
6512 char_u *end = NULL; | |
4444 | 6513 FILE *f; |
6514 | |
6515 #ifdef BT_REGEXP_LOG | |
6516 f = fopen("bt_regexp_log.log", "a"); | |
6517 #else | |
6518 f = stdout; | |
6519 #endif | |
6520 if (f == NULL) | |
6521 return; | |
6522 fprintf(f, "-------------------------------------\n\r\nregcomp(%s):\r\n", pattern); | |
7 | 6523 |
6524 s = r->program + 1; | |
6525 /* | |
6526 * Loop until we find the END that isn't before a referred next (an END | |
6527 * can also appear in a NOMATCH operand). | |
6528 */ | |
6529 while (op != END || s <= end) | |
6530 { | |
6531 op = OP(s); | |
4444 | 6532 fprintf(f, "%2d%s", (int)(s - r->program), regprop(s)); /* Where, what. */ |
7 | 6533 next = regnext(s); |
6534 if (next == NULL) /* Next ptr. */ | |
4444 | 6535 fprintf(f, "(0)"); |
7 | 6536 else |
4444 | 6537 fprintf(f, "(%d)", (int)((s - r->program) + (next - s))); |
7 | 6538 if (end < next) |
6539 end = next; | |
6540 if (op == BRACE_LIMITS) | |
6541 { | |
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6542 /* Two ints */ |
4444 | 6543 fprintf(f, " minval %ld, maxval %ld", OPERAND_MIN(s), OPERAND_MAX(s)); |
7 | 6544 s += 8; |
6545 } | |
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6546 else if (op == BEHIND || op == NOBEHIND) |
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6547 { |
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6548 /* one int */ |
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6549 fprintf(f, " count %ld", OPERAND_MIN(s)); |
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6550 s += 4; |
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6551 } |
4770
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6552 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
|
6553 { |
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6554 /* one int plus comperator */ |
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6555 fprintf(f, " count %ld", OPERAND_MIN(s)); |
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6556 s += 5; |
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6557 } |
7 | 6558 s += 3; |
6559 if (op == ANYOF || op == ANYOF + ADD_NL | |
6560 || op == ANYBUT || op == ANYBUT + ADD_NL | |
6561 || op == EXACTLY) | |
6562 { | |
6563 /* Literal string, where present. */ | |
4444 | 6564 fprintf(f, "\nxxxxxxxxx\n"); |
7 | 6565 while (*s != NUL) |
4444 | 6566 fprintf(f, "%c", *s++); |
6567 fprintf(f, "\nxxxxxxxxx\n"); | |
7 | 6568 s++; |
6569 } | |
4444 | 6570 fprintf(f, "\r\n"); |
7 | 6571 } |
6572 | |
6573 /* Header fields of interest. */ | |
6574 if (r->regstart != NUL) | |
4444 | 6575 fprintf(f, "start `%s' 0x%x; ", r->regstart < 256 |
7 | 6576 ? (char *)transchar(r->regstart) |
6577 : "multibyte", r->regstart); | |
6578 if (r->reganch) | |
4444 | 6579 fprintf(f, "anchored; "); |
7 | 6580 if (r->regmust != NULL) |
4444 | 6581 fprintf(f, "must have \"%s\"", r->regmust); |
6582 fprintf(f, "\r\n"); | |
6583 | |
6584 #ifdef BT_REGEXP_LOG | |
6585 fclose(f); | |
6586 #endif | |
7 | 6587 } |
4444 | 6588 #endif /* BT_REGEXP_DUMP */ |
6589 | |
6590 #ifdef DEBUG | |
7 | 6591 /* |
6592 * regprop - printable representation of opcode | |
6593 */ | |
6594 static char_u * | |
6595 regprop(op) | |
6596 char_u *op; | |
6597 { | |
4444 | 6598 char *p; |
6599 static char buf[50]; | |
6600 | |
6601 STRCPY(buf, ":"); | |
6602 | |
6603 switch ((int) OP(op)) | |
7 | 6604 { |
6605 case BOL: | |
6606 p = "BOL"; | |
6607 break; | |
6608 case EOL: | |
6609 p = "EOL"; | |
6610 break; | |
6611 case RE_BOF: | |
6612 p = "BOF"; | |
6613 break; | |
6614 case RE_EOF: | |
6615 p = "EOF"; | |
6616 break; | |
6617 case CURSOR: | |
6618 p = "CURSOR"; | |
6619 break; | |
639 | 6620 case RE_VISUAL: |
6621 p = "RE_VISUAL"; | |
6622 break; | |
7 | 6623 case RE_LNUM: |
6624 p = "RE_LNUM"; | |
6625 break; | |
639 | 6626 case RE_MARK: |
6627 p = "RE_MARK"; | |
6628 break; | |
7 | 6629 case RE_COL: |
6630 p = "RE_COL"; | |
6631 break; | |
6632 case RE_VCOL: | |
6633 p = "RE_VCOL"; | |
6634 break; | |
6635 case BOW: | |
6636 p = "BOW"; | |
6637 break; | |
6638 case EOW: | |
6639 p = "EOW"; | |
6640 break; | |
6641 case ANY: | |
6642 p = "ANY"; | |
6643 break; | |
6644 case ANY + ADD_NL: | |
6645 p = "ANY+NL"; | |
6646 break; | |
6647 case ANYOF: | |
6648 p = "ANYOF"; | |
6649 break; | |
6650 case ANYOF + ADD_NL: | |
6651 p = "ANYOF+NL"; | |
6652 break; | |
6653 case ANYBUT: | |
6654 p = "ANYBUT"; | |
6655 break; | |
6656 case ANYBUT + ADD_NL: | |
6657 p = "ANYBUT+NL"; | |
6658 break; | |
6659 case IDENT: | |
6660 p = "IDENT"; | |
6661 break; | |
6662 case IDENT + ADD_NL: | |
6663 p = "IDENT+NL"; | |
6664 break; | |
6665 case SIDENT: | |
6666 p = "SIDENT"; | |
6667 break; | |
6668 case SIDENT + ADD_NL: | |
6669 p = "SIDENT+NL"; | |
6670 break; | |
6671 case KWORD: | |
6672 p = "KWORD"; | |
6673 break; | |
6674 case KWORD + ADD_NL: | |
6675 p = "KWORD+NL"; | |
6676 break; | |
6677 case SKWORD: | |
6678 p = "SKWORD"; | |
6679 break; | |
6680 case SKWORD + ADD_NL: | |
6681 p = "SKWORD+NL"; | |
6682 break; | |
6683 case FNAME: | |
6684 p = "FNAME"; | |
6685 break; | |
6686 case FNAME + ADD_NL: | |
6687 p = "FNAME+NL"; | |
6688 break; | |
6689 case SFNAME: | |
6690 p = "SFNAME"; | |
6691 break; | |
6692 case SFNAME + ADD_NL: | |
6693 p = "SFNAME+NL"; | |
6694 break; | |
6695 case PRINT: | |
6696 p = "PRINT"; | |
6697 break; | |
6698 case PRINT + ADD_NL: | |
6699 p = "PRINT+NL"; | |
6700 break; | |
6701 case SPRINT: | |
6702 p = "SPRINT"; | |
6703 break; | |
6704 case SPRINT + ADD_NL: | |
6705 p = "SPRINT+NL"; | |
6706 break; | |
6707 case WHITE: | |
6708 p = "WHITE"; | |
6709 break; | |
6710 case WHITE + ADD_NL: | |
6711 p = "WHITE+NL"; | |
6712 break; | |
6713 case NWHITE: | |
6714 p = "NWHITE"; | |
6715 break; | |
6716 case NWHITE + ADD_NL: | |
6717 p = "NWHITE+NL"; | |
6718 break; | |
6719 case DIGIT: | |
6720 p = "DIGIT"; | |
6721 break; | |
6722 case DIGIT + ADD_NL: | |
6723 p = "DIGIT+NL"; | |
6724 break; | |
6725 case NDIGIT: | |
6726 p = "NDIGIT"; | |
6727 break; | |
6728 case NDIGIT + ADD_NL: | |
6729 p = "NDIGIT+NL"; | |
6730 break; | |
6731 case HEX: | |
6732 p = "HEX"; | |
6733 break; | |
6734 case HEX + ADD_NL: | |
6735 p = "HEX+NL"; | |
6736 break; | |
6737 case NHEX: | |
6738 p = "NHEX"; | |
6739 break; | |
6740 case NHEX + ADD_NL: | |
6741 p = "NHEX+NL"; | |
6742 break; | |
6743 case OCTAL: | |
6744 p = "OCTAL"; | |
6745 break; | |
6746 case OCTAL + ADD_NL: | |
6747 p = "OCTAL+NL"; | |
6748 break; | |
6749 case NOCTAL: | |
6750 p = "NOCTAL"; | |
6751 break; | |
6752 case NOCTAL + ADD_NL: | |
6753 p = "NOCTAL+NL"; | |
6754 break; | |
6755 case WORD: | |
6756 p = "WORD"; | |
6757 break; | |
6758 case WORD + ADD_NL: | |
6759 p = "WORD+NL"; | |
6760 break; | |
6761 case NWORD: | |
6762 p = "NWORD"; | |
6763 break; | |
6764 case NWORD + ADD_NL: | |
6765 p = "NWORD+NL"; | |
6766 break; | |
6767 case HEAD: | |
6768 p = "HEAD"; | |
6769 break; | |
6770 case HEAD + ADD_NL: | |
6771 p = "HEAD+NL"; | |
6772 break; | |
6773 case NHEAD: | |
6774 p = "NHEAD"; | |
6775 break; | |
6776 case NHEAD + ADD_NL: | |
6777 p = "NHEAD+NL"; | |
6778 break; | |
6779 case ALPHA: | |
6780 p = "ALPHA"; | |
6781 break; | |
6782 case ALPHA + ADD_NL: | |
6783 p = "ALPHA+NL"; | |
6784 break; | |
6785 case NALPHA: | |
6786 p = "NALPHA"; | |
6787 break; | |
6788 case NALPHA + ADD_NL: | |
6789 p = "NALPHA+NL"; | |
6790 break; | |
6791 case LOWER: | |
6792 p = "LOWER"; | |
6793 break; | |
6794 case LOWER + ADD_NL: | |
6795 p = "LOWER+NL"; | |
6796 break; | |
6797 case NLOWER: | |
6798 p = "NLOWER"; | |
6799 break; | |
6800 case NLOWER + ADD_NL: | |
6801 p = "NLOWER+NL"; | |
6802 break; | |
6803 case UPPER: | |
6804 p = "UPPER"; | |
6805 break; | |
6806 case UPPER + ADD_NL: | |
6807 p = "UPPER+NL"; | |
6808 break; | |
6809 case NUPPER: | |
6810 p = "NUPPER"; | |
6811 break; | |
6812 case NUPPER + ADD_NL: | |
6813 p = "NUPPER+NL"; | |
6814 break; | |
6815 case BRANCH: | |
6816 p = "BRANCH"; | |
6817 break; | |
6818 case EXACTLY: | |
6819 p = "EXACTLY"; | |
6820 break; | |
6821 case NOTHING: | |
6822 p = "NOTHING"; | |
6823 break; | |
6824 case BACK: | |
6825 p = "BACK"; | |
6826 break; | |
6827 case END: | |
6828 p = "END"; | |
6829 break; | |
6830 case MOPEN + 0: | |
6831 p = "MATCH START"; | |
6832 break; | |
6833 case MOPEN + 1: | |
6834 case MOPEN + 2: | |
6835 case MOPEN + 3: | |
6836 case MOPEN + 4: | |
6837 case MOPEN + 5: | |
6838 case MOPEN + 6: | |
6839 case MOPEN + 7: | |
6840 case MOPEN + 8: | |
6841 case MOPEN + 9: | |
6842 sprintf(buf + STRLEN(buf), "MOPEN%d", OP(op) - MOPEN); | |
6843 p = NULL; | |
6844 break; | |
6845 case MCLOSE + 0: | |
6846 p = "MATCH END"; | |
6847 break; | |
6848 case MCLOSE + 1: | |
6849 case MCLOSE + 2: | |
6850 case MCLOSE + 3: | |
6851 case MCLOSE + 4: | |
6852 case MCLOSE + 5: | |
6853 case MCLOSE + 6: | |
6854 case MCLOSE + 7: | |
6855 case MCLOSE + 8: | |
6856 case MCLOSE + 9: | |
6857 sprintf(buf + STRLEN(buf), "MCLOSE%d", OP(op) - MCLOSE); | |
6858 p = NULL; | |
6859 break; | |
6860 case BACKREF + 1: | |
6861 case BACKREF + 2: | |
6862 case BACKREF + 3: | |
6863 case BACKREF + 4: | |
6864 case BACKREF + 5: | |
6865 case BACKREF + 6: | |
6866 case BACKREF + 7: | |
6867 case BACKREF + 8: | |
6868 case BACKREF + 9: | |
6869 sprintf(buf + STRLEN(buf), "BACKREF%d", OP(op) - BACKREF); | |
6870 p = NULL; | |
6871 break; | |
6872 case NOPEN: | |
6873 p = "NOPEN"; | |
6874 break; | |
6875 case NCLOSE: | |
6876 p = "NCLOSE"; | |
6877 break; | |
6878 #ifdef FEAT_SYN_HL | |
6879 case ZOPEN + 1: | |
6880 case ZOPEN + 2: | |
6881 case ZOPEN + 3: | |
6882 case ZOPEN + 4: | |
6883 case ZOPEN + 5: | |
6884 case ZOPEN + 6: | |
6885 case ZOPEN + 7: | |
6886 case ZOPEN + 8: | |
6887 case ZOPEN + 9: | |
6888 sprintf(buf + STRLEN(buf), "ZOPEN%d", OP(op) - ZOPEN); | |
6889 p = NULL; | |
6890 break; | |
6891 case ZCLOSE + 1: | |
6892 case ZCLOSE + 2: | |
6893 case ZCLOSE + 3: | |
6894 case ZCLOSE + 4: | |
6895 case ZCLOSE + 5: | |
6896 case ZCLOSE + 6: | |
6897 case ZCLOSE + 7: | |
6898 case ZCLOSE + 8: | |
6899 case ZCLOSE + 9: | |
6900 sprintf(buf + STRLEN(buf), "ZCLOSE%d", OP(op) - ZCLOSE); | |
6901 p = NULL; | |
6902 break; | |
6903 case ZREF + 1: | |
6904 case ZREF + 2: | |
6905 case ZREF + 3: | |
6906 case ZREF + 4: | |
6907 case ZREF + 5: | |
6908 case ZREF + 6: | |
6909 case ZREF + 7: | |
6910 case ZREF + 8: | |
6911 case ZREF + 9: | |
6912 sprintf(buf + STRLEN(buf), "ZREF%d", OP(op) - ZREF); | |
6913 p = NULL; | |
6914 break; | |
6915 #endif | |
6916 case STAR: | |
6917 p = "STAR"; | |
6918 break; | |
6919 case PLUS: | |
6920 p = "PLUS"; | |
6921 break; | |
6922 case NOMATCH: | |
6923 p = "NOMATCH"; | |
6924 break; | |
6925 case MATCH: | |
6926 p = "MATCH"; | |
6927 break; | |
6928 case BEHIND: | |
6929 p = "BEHIND"; | |
6930 break; | |
6931 case NOBEHIND: | |
6932 p = "NOBEHIND"; | |
6933 break; | |
6934 case SUBPAT: | |
6935 p = "SUBPAT"; | |
6936 break; | |
6937 case BRACE_LIMITS: | |
6938 p = "BRACE_LIMITS"; | |
6939 break; | |
6940 case BRACE_SIMPLE: | |
6941 p = "BRACE_SIMPLE"; | |
6942 break; | |
6943 case BRACE_COMPLEX + 0: | |
6944 case BRACE_COMPLEX + 1: | |
6945 case BRACE_COMPLEX + 2: | |
6946 case BRACE_COMPLEX + 3: | |
6947 case BRACE_COMPLEX + 4: | |
6948 case BRACE_COMPLEX + 5: | |
6949 case BRACE_COMPLEX + 6: | |
6950 case BRACE_COMPLEX + 7: | |
6951 case BRACE_COMPLEX + 8: | |
6952 case BRACE_COMPLEX + 9: | |
6953 sprintf(buf + STRLEN(buf), "BRACE_COMPLEX%d", OP(op) - BRACE_COMPLEX); | |
6954 p = NULL; | |
6955 break; | |
6956 #ifdef FEAT_MBYTE | |
6957 case MULTIBYTECODE: | |
6958 p = "MULTIBYTECODE"; | |
6959 break; | |
6960 #endif | |
6961 case NEWL: | |
6962 p = "NEWL"; | |
6963 break; | |
6964 default: | |
6965 sprintf(buf + STRLEN(buf), "corrupt %d", OP(op)); | |
6966 p = NULL; | |
6967 break; | |
6968 } | |
6969 if (p != NULL) | |
4444 | 6970 STRCAT(buf, p); |
6971 return (char_u *)buf; | |
7 | 6972 } |
4444 | 6973 #endif /* DEBUG */ |
7 | 6974 |
6975 #ifdef FEAT_MBYTE | |
6976 static void mb_decompose __ARGS((int c, int *c1, int *c2, int *c3)); | |
6977 | |
6978 typedef struct | |
6979 { | |
6980 int a, b, c; | |
6981 } decomp_T; | |
6982 | |
6983 | |
6984 /* 0xfb20 - 0xfb4f */ | |
297 | 6985 static decomp_T decomp_table[0xfb4f-0xfb20+1] = |
7 | 6986 { |
6987 {0x5e2,0,0}, /* 0xfb20 alt ayin */ | |
6988 {0x5d0,0,0}, /* 0xfb21 alt alef */ | |
6989 {0x5d3,0,0}, /* 0xfb22 alt dalet */ | |
6990 {0x5d4,0,0}, /* 0xfb23 alt he */ | |
6991 {0x5db,0,0}, /* 0xfb24 alt kaf */ | |
6992 {0x5dc,0,0}, /* 0xfb25 alt lamed */ | |
6993 {0x5dd,0,0}, /* 0xfb26 alt mem-sofit */ | |
6994 {0x5e8,0,0}, /* 0xfb27 alt resh */ | |
6995 {0x5ea,0,0}, /* 0xfb28 alt tav */ | |
6996 {'+', 0, 0}, /* 0xfb29 alt plus */ | |
6997 {0x5e9, 0x5c1, 0}, /* 0xfb2a shin+shin-dot */ | |
6998 {0x5e9, 0x5c2, 0}, /* 0xfb2b shin+sin-dot */ | |
6999 {0x5e9, 0x5c1, 0x5bc}, /* 0xfb2c shin+shin-dot+dagesh */ | |
7000 {0x5e9, 0x5c2, 0x5bc}, /* 0xfb2d shin+sin-dot+dagesh */ | |
7001 {0x5d0, 0x5b7, 0}, /* 0xfb2e alef+patah */ | |
7002 {0x5d0, 0x5b8, 0}, /* 0xfb2f alef+qamats */ | |
7003 {0x5d0, 0x5b4, 0}, /* 0xfb30 alef+hiriq */ | |
7004 {0x5d1, 0x5bc, 0}, /* 0xfb31 bet+dagesh */ | |
7005 {0x5d2, 0x5bc, 0}, /* 0xfb32 gimel+dagesh */ | |
7006 {0x5d3, 0x5bc, 0}, /* 0xfb33 dalet+dagesh */ | |
7007 {0x5d4, 0x5bc, 0}, /* 0xfb34 he+dagesh */ | |
7008 {0x5d5, 0x5bc, 0}, /* 0xfb35 vav+dagesh */ | |
7009 {0x5d6, 0x5bc, 0}, /* 0xfb36 zayin+dagesh */ | |
7010 {0xfb37, 0, 0}, /* 0xfb37 -- UNUSED */ | |
7011 {0x5d8, 0x5bc, 0}, /* 0xfb38 tet+dagesh */ | |
7012 {0x5d9, 0x5bc, 0}, /* 0xfb39 yud+dagesh */ | |
7013 {0x5da, 0x5bc, 0}, /* 0xfb3a kaf sofit+dagesh */ | |
7014 {0x5db, 0x5bc, 0}, /* 0xfb3b kaf+dagesh */ | |
7015 {0x5dc, 0x5bc, 0}, /* 0xfb3c lamed+dagesh */ | |
7016 {0xfb3d, 0, 0}, /* 0xfb3d -- UNUSED */ | |
7017 {0x5de, 0x5bc, 0}, /* 0xfb3e mem+dagesh */ | |
7018 {0xfb3f, 0, 0}, /* 0xfb3f -- UNUSED */ | |
7019 {0x5e0, 0x5bc, 0}, /* 0xfb40 nun+dagesh */ | |
7020 {0x5e1, 0x5bc, 0}, /* 0xfb41 samech+dagesh */ | |
7021 {0xfb42, 0, 0}, /* 0xfb42 -- UNUSED */ | |
7022 {0x5e3, 0x5bc, 0}, /* 0xfb43 pe sofit+dagesh */ | |
7023 {0x5e4, 0x5bc,0}, /* 0xfb44 pe+dagesh */ | |
7024 {0xfb45, 0, 0}, /* 0xfb45 -- UNUSED */ | |
7025 {0x5e6, 0x5bc, 0}, /* 0xfb46 tsadi+dagesh */ | |
7026 {0x5e7, 0x5bc, 0}, /* 0xfb47 qof+dagesh */ | |
7027 {0x5e8, 0x5bc, 0}, /* 0xfb48 resh+dagesh */ | |
7028 {0x5e9, 0x5bc, 0}, /* 0xfb49 shin+dagesh */ | |
7029 {0x5ea, 0x5bc, 0}, /* 0xfb4a tav+dagesh */ | |
7030 {0x5d5, 0x5b9, 0}, /* 0xfb4b vav+holam */ | |
7031 {0x5d1, 0x5bf, 0}, /* 0xfb4c bet+rafe */ | |
7032 {0x5db, 0x5bf, 0}, /* 0xfb4d kaf+rafe */ | |
7033 {0x5e4, 0x5bf, 0}, /* 0xfb4e pe+rafe */ | |
7034 {0x5d0, 0x5dc, 0} /* 0xfb4f alef-lamed */ | |
7035 }; | |
7036 | |
7037 static void | |
7038 mb_decompose(c, c1, c2, c3) | |
7039 int c, *c1, *c2, *c3; | |
7040 { | |
7041 decomp_T d; | |
7042 | |
4505
d037b9cbdaaa
updated for version 7.3.1000
Bram Moolenaar <bram@vim.org>
parents:
4466
diff
changeset
|
7043 if (c >= 0xfb20 && c <= 0xfb4f) |
7 | 7044 { |
7045 d = decomp_table[c - 0xfb20]; | |
7046 *c1 = d.a; | |
7047 *c2 = d.b; | |
7048 *c3 = d.c; | |
7049 } | |
7050 else | |
7051 { | |
7052 *c1 = c; | |
7053 *c2 = *c3 = 0; | |
7054 } | |
7055 } | |
7056 #endif | |
7057 | |
7058 /* | |
7059 * Compare two strings, ignore case if ireg_ic set. | |
7060 * Return 0 if strings match, non-zero otherwise. | |
7061 * Correct the length "*n" when composing characters are ignored. | |
7062 */ | |
7063 static int | |
7064 cstrncmp(s1, s2, n) | |
7065 char_u *s1, *s2; | |
7066 int *n; | |
7067 { | |
7068 int result; | |
7069 | |
7070 if (!ireg_ic) | |
7071 result = STRNCMP(s1, s2, *n); | |
7072 else | |
7073 result = MB_STRNICMP(s1, s2, *n); | |
7074 | |
7075 #ifdef FEAT_MBYTE | |
7076 /* if it failed and it's utf8 and we want to combineignore: */ | |
7077 if (result != 0 && enc_utf8 && ireg_icombine) | |
7078 { | |
7079 char_u *str1, *str2; | |
7080 int c1, c2, c11, c12; | |
7081 int junk; | |
7082 | |
7083 /* we have to handle the strcmp ourselves, since it is necessary to | |
7084 * deal with the composing characters by ignoring them: */ | |
7085 str1 = s1; | |
7086 str2 = s2; | |
7087 c1 = c2 = 0; | |
507 | 7088 while ((int)(str1 - s1) < *n) |
7 | 7089 { |
7090 c1 = mb_ptr2char_adv(&str1); | |
7091 c2 = mb_ptr2char_adv(&str2); | |
7092 | |
7093 /* decompose the character if necessary, into 'base' characters | |
7094 * because I don't care about Arabic, I will hard-code the Hebrew | |
7095 * which I *do* care about! So sue me... */ | |
7096 if (c1 != c2 && (!ireg_ic || utf_fold(c1) != utf_fold(c2))) | |
7097 { | |
7098 /* decomposition necessary? */ | |
7099 mb_decompose(c1, &c11, &junk, &junk); | |
7100 mb_decompose(c2, &c12, &junk, &junk); | |
7101 c1 = c11; | |
7102 c2 = c12; | |
7103 if (c11 != c12 && (!ireg_ic || utf_fold(c11) != utf_fold(c12))) | |
7104 break; | |
7105 } | |
7106 } | |
7107 result = c2 - c1; | |
7108 if (result == 0) | |
7109 *n = (int)(str2 - s2); | |
7110 } | |
7111 #endif | |
7112 | |
7113 return result; | |
7114 } | |
7115 | |
7116 /* | |
7117 * cstrchr: This function is used a lot for simple searches, keep it fast! | |
7118 */ | |
7119 static char_u * | |
7120 cstrchr(s, c) | |
7121 char_u *s; | |
7122 int c; | |
7123 { | |
7124 char_u *p; | |
7125 int cc; | |
7126 | |
7127 if (!ireg_ic | |
7128 #ifdef FEAT_MBYTE | |
7129 || (!enc_utf8 && mb_char2len(c) > 1) | |
7130 #endif | |
7131 ) | |
7132 return vim_strchr(s, c); | |
7133 | |
7134 /* tolower() and toupper() can be slow, comparing twice should be a lot | |
7135 * faster (esp. when using MS Visual C++!). | |
7136 * For UTF-8 need to use folded case. */ | |
7137 #ifdef FEAT_MBYTE | |
7138 if (enc_utf8 && c > 0x80) | |
7139 cc = utf_fold(c); | |
7140 else | |
7141 #endif | |
1347 | 7142 if (MB_ISUPPER(c)) |
7143 cc = MB_TOLOWER(c); | |
7144 else if (MB_ISLOWER(c)) | |
7145 cc = MB_TOUPPER(c); | |
7 | 7146 else |
7147 return vim_strchr(s, c); | |
7148 | |
7149 #ifdef FEAT_MBYTE | |
7150 if (has_mbyte) | |
7151 { | |
474 | 7152 for (p = s; *p != NUL; p += (*mb_ptr2len)(p)) |
7 | 7153 { |
7154 if (enc_utf8 && c > 0x80) | |
7155 { | |
7156 if (utf_fold(utf_ptr2char(p)) == cc) | |
7157 return p; | |
7158 } | |
7159 else if (*p == c || *p == cc) | |
7160 return p; | |
7161 } | |
7162 } | |
7163 else | |
7164 #endif | |
7165 /* Faster version for when there are no multi-byte characters. */ | |
7166 for (p = s; *p != NUL; ++p) | |
7167 if (*p == c || *p == cc) | |
7168 return p; | |
7169 | |
7170 return NULL; | |
7171 } | |
7172 | |
7173 /*************************************************************** | |
7174 * regsub stuff * | |
7175 ***************************************************************/ | |
7176 | |
7177 /* This stuff below really confuses cc on an SGI -- webb */ | |
7178 #ifdef __sgi | |
7179 # undef __ARGS | |
7180 # define __ARGS(x) () | |
7181 #endif | |
7182 | |
7183 /* | |
7184 * We should define ftpr as a pointer to a function returning a pointer to | |
7185 * a function returning a pointer to a function ... | |
7186 * This is impossible, so we declare a pointer to a function returning a | |
7187 * pointer to a function returning void. This should work for all compilers. | |
7188 */ | |
772 | 7189 typedef void (*(*fptr_T) __ARGS((int *, int)))(); |
7190 | |
7191 static fptr_T do_upper __ARGS((int *, int)); | |
7192 static fptr_T do_Upper __ARGS((int *, int)); | |
7193 static fptr_T do_lower __ARGS((int *, int)); | |
7194 static fptr_T do_Lower __ARGS((int *, int)); | |
7 | 7195 |
7196 static int vim_regsub_both __ARGS((char_u *source, char_u *dest, int copy, int magic, int backslash)); | |
7197 | |
772 | 7198 static fptr_T |
7 | 7199 do_upper(d, c) |
772 | 7200 int *d; |
7201 int c; | |
7 | 7202 { |
772 | 7203 *d = MB_TOUPPER(c); |
7204 | |
7205 return (fptr_T)NULL; | |
7 | 7206 } |
7207 | |
772 | 7208 static fptr_T |
7209 do_Upper(d, c) | |
7210 int *d; | |
7 | 7211 int c; |
7212 { | |
772 | 7213 *d = MB_TOUPPER(c); |
7214 | |
7215 return (fptr_T)do_Upper; | |
7216 } | |
7217 | |
7218 static fptr_T | |
7219 do_lower(d, c) | |
7220 int *d; | |
7221 int c; | |
7222 { | |
7223 *d = MB_TOLOWER(c); | |
7224 | |
7225 return (fptr_T)NULL; | |
7226 } | |
7227 | |
7228 static fptr_T | |
7229 do_Lower(d, c) | |
7230 int *d; | |
7231 int c; | |
7232 { | |
7233 *d = MB_TOLOWER(c); | |
7234 | |
7235 return (fptr_T)do_Lower; | |
7 | 7236 } |
7237 | |
7238 /* | |
7239 * regtilde(): Replace tildes in the pattern by the old pattern. | |
7240 * | |
7241 * Short explanation of the tilde: It stands for the previous replacement | |
7242 * pattern. If that previous pattern also contains a ~ we should go back a | |
7243 * step further... But we insert the previous pattern into the current one | |
7244 * and remember that. | |
772 | 7245 * This still does not handle the case where "magic" changes. So require the |
7246 * user to keep his hands off of "magic". | |
7 | 7247 * |
7248 * The tildes are parsed once before the first call to vim_regsub(). | |
7249 */ | |
7250 char_u * | |
7251 regtilde(source, magic) | |
7252 char_u *source; | |
7253 int magic; | |
7254 { | |
7255 char_u *newsub = source; | |
7256 char_u *tmpsub; | |
7257 char_u *p; | |
7258 int len; | |
7259 int prevlen; | |
7260 | |
7261 for (p = newsub; *p; ++p) | |
7262 { | |
7263 if ((*p == '~' && magic) || (*p == '\\' && *(p + 1) == '~' && !magic)) | |
7264 { | |
7265 if (reg_prev_sub != NULL) | |
7266 { | |
7267 /* length = len(newsub) - 1 + len(prev_sub) + 1 */ | |
7268 prevlen = (int)STRLEN(reg_prev_sub); | |
7269 tmpsub = alloc((unsigned)(STRLEN(newsub) + prevlen)); | |
7270 if (tmpsub != NULL) | |
7271 { | |
7272 /* copy prefix */ | |
7273 len = (int)(p - newsub); /* not including ~ */ | |
7274 mch_memmove(tmpsub, newsub, (size_t)len); | |
1209 | 7275 /* interpret tilde */ |
7 | 7276 mch_memmove(tmpsub + len, reg_prev_sub, (size_t)prevlen); |
7277 /* copy postfix */ | |
7278 if (!magic) | |
7279 ++p; /* back off \ */ | |
7280 STRCPY(tmpsub + len + prevlen, p + 1); | |
7281 | |
7282 if (newsub != source) /* already allocated newsub */ | |
7283 vim_free(newsub); | |
7284 newsub = tmpsub; | |
7285 p = newsub + len + prevlen; | |
7286 } | |
7287 } | |
7288 else if (magic) | |
1621 | 7289 STRMOVE(p, p + 1); /* remove '~' */ |
7 | 7290 else |
1621 | 7291 STRMOVE(p, p + 2); /* remove '\~' */ |
7 | 7292 --p; |
7293 } | |
7294 else | |
7295 { | |
7296 if (*p == '\\' && p[1]) /* skip escaped characters */ | |
7297 ++p; | |
7298 #ifdef FEAT_MBYTE | |
7299 if (has_mbyte) | |
474 | 7300 p += (*mb_ptr2len)(p) - 1; |
7 | 7301 #endif |
7302 } | |
7303 } | |
7304 | |
7305 vim_free(reg_prev_sub); | |
7306 if (newsub != source) /* newsub was allocated, just keep it */ | |
7307 reg_prev_sub = newsub; | |
7308 else /* no ~ found, need to save newsub */ | |
7309 reg_prev_sub = vim_strsave(newsub); | |
7310 return newsub; | |
7311 } | |
7312 | |
7313 #ifdef FEAT_EVAL | |
7314 static int can_f_submatch = FALSE; /* TRUE when submatch() can be used */ | |
7315 | |
7316 /* These pointers are used instead of reg_match and reg_mmatch for | |
7317 * reg_submatch(). Needed for when the substitution string is an expression | |
7318 * that contains a call to substitute() and submatch(). */ | |
7319 static regmatch_T *submatch_match; | |
7320 static regmmatch_T *submatch_mmatch; | |
2011 | 7321 static linenr_T submatch_firstlnum; |
7322 static linenr_T submatch_maxline; | |
2904 | 7323 static int submatch_line_lbr; |
7 | 7324 #endif |
7325 | |
7326 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO) | |
7327 /* | |
7328 * vim_regsub() - perform substitutions after a vim_regexec() or | |
7329 * vim_regexec_multi() match. | |
7330 * | |
7331 * If "copy" is TRUE really copy into "dest". | |
7332 * If "copy" is FALSE nothing is copied, this is just to find out the length | |
7333 * of the result. | |
7334 * | |
7335 * If "backslash" is TRUE, a backslash will be removed later, need to double | |
7336 * them to keep them, and insert a backslash before a CR to avoid it being | |
7337 * replaced with a line break later. | |
7338 * | |
7339 * Note: The matched text must not change between the call of | |
7340 * vim_regexec()/vim_regexec_multi() and vim_regsub()! It would make the back | |
7341 * references invalid! | |
7342 * | |
7343 * Returns the size of the replacement, including terminating NUL. | |
7344 */ | |
7345 int | |
7346 vim_regsub(rmp, source, dest, copy, magic, backslash) | |
7347 regmatch_T *rmp; | |
7348 char_u *source; | |
7349 char_u *dest; | |
7350 int copy; | |
7351 int magic; | |
7352 int backslash; | |
7353 { | |
7354 reg_match = rmp; | |
7355 reg_mmatch = NULL; | |
7356 reg_maxline = 0; | |
4061 | 7357 reg_buf = curbuf; |
5836 | 7358 reg_line_lbr = TRUE; |
7 | 7359 return vim_regsub_both(source, dest, copy, magic, backslash); |
7360 } | |
7361 #endif | |
7362 | |
7363 int | |
7364 vim_regsub_multi(rmp, lnum, source, dest, copy, magic, backslash) | |
7365 regmmatch_T *rmp; | |
7366 linenr_T lnum; | |
7367 char_u *source; | |
7368 char_u *dest; | |
7369 int copy; | |
7370 int magic; | |
7371 int backslash; | |
7372 { | |
7373 reg_match = NULL; | |
7374 reg_mmatch = rmp; | |
7375 reg_buf = curbuf; /* always works on the current buffer! */ | |
7376 reg_firstlnum = lnum; | |
7377 reg_maxline = curbuf->b_ml.ml_line_count - lnum; | |
5836 | 7378 reg_line_lbr = FALSE; |
7 | 7379 return vim_regsub_both(source, dest, copy, magic, backslash); |
7380 } | |
7381 | |
7382 static int | |
7383 vim_regsub_both(source, dest, copy, magic, backslash) | |
7384 char_u *source; | |
7385 char_u *dest; | |
7386 int copy; | |
7387 int magic; | |
7388 int backslash; | |
7389 { | |
7390 char_u *src; | |
7391 char_u *dst; | |
7392 char_u *s; | |
7393 int c; | |
772 | 7394 int cc; |
7 | 7395 int no = -1; |
4244 | 7396 fptr_T func_all = (fptr_T)NULL; |
7397 fptr_T func_one = (fptr_T)NULL; | |
7 | 7398 linenr_T clnum = 0; /* init for GCC */ |
7399 int len = 0; /* init for GCC */ | |
7400 #ifdef FEAT_EVAL | |
7401 static char_u *eval_result = NULL; | |
7402 #endif | |
7403 | |
7404 /* Be paranoid... */ | |
7405 if (source == NULL || dest == NULL) | |
7406 { | |
7407 EMSG(_(e_null)); | |
7408 return 0; | |
7409 } | |
7410 if (prog_magic_wrong()) | |
7411 return 0; | |
7412 src = source; | |
7413 dst = dest; | |
7414 | |
7415 /* | |
7416 * When the substitute part starts with "\=" evaluate it as an expression. | |
7417 */ | |
7418 if (source[0] == '\\' && source[1] == '=' | |
7419 #ifdef FEAT_EVAL | |
7420 && !can_f_submatch /* can't do this recursively */ | |
7421 #endif | |
7422 ) | |
7423 { | |
7424 #ifdef FEAT_EVAL | |
7425 /* To make sure that the length doesn't change between checking the | |
7426 * length and copying the string, and to speed up things, the | |
7427 * resulting string is saved from the call with "copy" == FALSE to the | |
7428 * call with "copy" == TRUE. */ | |
7429 if (copy) | |
7430 { | |
7431 if (eval_result != NULL) | |
7432 { | |
7433 STRCPY(dest, eval_result); | |
7434 dst += STRLEN(eval_result); | |
7435 vim_free(eval_result); | |
7436 eval_result = NULL; | |
7437 } | |
7438 } | |
7439 else | |
7440 { | |
7441 win_T *save_reg_win; | |
7442 int save_ireg_ic; | |
7443 | |
7444 vim_free(eval_result); | |
7445 | |
7446 /* The expression may contain substitute(), which calls us | |
7447 * recursively. Make sure submatch() gets the text from the first | |
7448 * level. Don't need to save "reg_buf", because | |
7449 * vim_regexec_multi() can't be called recursively. */ | |
7450 submatch_match = reg_match; | |
7451 submatch_mmatch = reg_mmatch; | |
2011 | 7452 submatch_firstlnum = reg_firstlnum; |
7453 submatch_maxline = reg_maxline; | |
2904 | 7454 submatch_line_lbr = reg_line_lbr; |
7 | 7455 save_reg_win = reg_win; |
7456 save_ireg_ic = ireg_ic; | |
7457 can_f_submatch = TRUE; | |
7458 | |
714 | 7459 eval_result = eval_to_string(source + 2, NULL, TRUE); |
7 | 7460 if (eval_result != NULL) |
7461 { | |
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7462 int had_backslash = FALSE; |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7463 |
39 | 7464 for (s = eval_result; *s != NUL; mb_ptr_adv(s)) |
7 | 7465 { |
2904 | 7466 /* Change NL to CR, so that it becomes a line break, |
7467 * unless called from vim_regexec_nl(). | |
7 | 7468 * Skip over a backslashed character. */ |
2904 | 7469 if (*s == NL && !submatch_line_lbr) |
7 | 7470 *s = CAR; |
7471 else if (*s == '\\' && s[1] != NUL) | |
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7472 { |
7 | 7473 ++s; |
2173 | 7474 /* Change NL to CR here too, so that this works: |
7475 * :s/abc\\\ndef/\="aaa\\\nbbb"/ on text: | |
7476 * abc\ | |
7477 * def | |
2904 | 7478 * Not when called from vim_regexec_nl(). |
2173 | 7479 */ |
2904 | 7480 if (*s == NL && !submatch_line_lbr) |
2173 | 7481 *s = CAR; |
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7482 had_backslash = TRUE; |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7483 } |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7484 } |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7485 if (had_backslash && backslash) |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7486 { |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7487 /* Backslashes will be consumed, need to double them. */ |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7488 s = vim_strsave_escaped(eval_result, (char_u *)"\\"); |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7489 if (s != NULL) |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7490 { |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7491 vim_free(eval_result); |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7492 eval_result = s; |
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7493 } |
7 | 7494 } |
7495 | |
7496 dst += STRLEN(eval_result); | |
7497 } | |
7498 | |
7499 reg_match = submatch_match; | |
7500 reg_mmatch = submatch_mmatch; | |
2011 | 7501 reg_firstlnum = submatch_firstlnum; |
7502 reg_maxline = submatch_maxline; | |
2904 | 7503 reg_line_lbr = submatch_line_lbr; |
7 | 7504 reg_win = save_reg_win; |
7505 ireg_ic = save_ireg_ic; | |
7506 can_f_submatch = FALSE; | |
7507 } | |
7508 #endif | |
7509 } | |
7510 else | |
7511 while ((c = *src++) != NUL) | |
7512 { | |
7513 if (c == '&' && magic) | |
7514 no = 0; | |
7515 else if (c == '\\' && *src != NUL) | |
7516 { | |
7517 if (*src == '&' && !magic) | |
7518 { | |
7519 ++src; | |
7520 no = 0; | |
7521 } | |
7522 else if ('0' <= *src && *src <= '9') | |
7523 { | |
7524 no = *src++ - '0'; | |
7525 } | |
7526 else if (vim_strchr((char_u *)"uUlLeE", *src)) | |
7527 { | |
7528 switch (*src++) | |
7529 { | |
4244 | 7530 case 'u': func_one = (fptr_T)do_upper; |
7 | 7531 continue; |
4244 | 7532 case 'U': func_all = (fptr_T)do_Upper; |
7 | 7533 continue; |
4244 | 7534 case 'l': func_one = (fptr_T)do_lower; |
7 | 7535 continue; |
4244 | 7536 case 'L': func_all = (fptr_T)do_Lower; |
7 | 7537 continue; |
7538 case 'e': | |
4244 | 7539 case 'E': func_one = func_all = (fptr_T)NULL; |
7 | 7540 continue; |
7541 } | |
7542 } | |
7543 } | |
7544 if (no < 0) /* Ordinary character. */ | |
7545 { | |
798 | 7546 if (c == K_SPECIAL && src[0] != NUL && src[1] != NUL) |
7547 { | |
1209 | 7548 /* Copy a special key as-is. */ |
798 | 7549 if (copy) |
7550 { | |
7551 *dst++ = c; | |
7552 *dst++ = *src++; | |
7553 *dst++ = *src++; | |
7554 } | |
7555 else | |
7556 { | |
7557 dst += 3; | |
7558 src += 2; | |
7559 } | |
7560 continue; | |
7561 } | |
7562 | |
7 | 7563 if (c == '\\' && *src != NUL) |
7564 { | |
7565 /* Check for abbreviations -- webb */ | |
7566 switch (*src) | |
7567 { | |
7568 case 'r': c = CAR; ++src; break; | |
7569 case 'n': c = NL; ++src; break; | |
7570 case 't': c = TAB; ++src; break; | |
7571 /* Oh no! \e already has meaning in subst pat :-( */ | |
7572 /* case 'e': c = ESC; ++src; break; */ | |
7573 case 'b': c = Ctrl_H; ++src; break; | |
7574 | |
7575 /* If "backslash" is TRUE the backslash will be removed | |
7576 * later. Used to insert a literal CR. */ | |
7577 default: if (backslash) | |
7578 { | |
7579 if (copy) | |
7580 *dst = '\\'; | |
7581 ++dst; | |
7582 } | |
7583 c = *src++; | |
7584 } | |
7585 } | |
798 | 7586 #ifdef FEAT_MBYTE |
7587 else if (has_mbyte) | |
7588 c = mb_ptr2char(src - 1); | |
7589 #endif | |
7 | 7590 |
7591 /* Write to buffer, if copy is set. */ | |
4244 | 7592 if (func_one != (fptr_T)NULL) |
7593 /* Turbo C complains without the typecast */ | |
7594 func_one = (fptr_T)(func_one(&cc, c)); | |
7595 else if (func_all != (fptr_T)NULL) | |
7596 /* Turbo C complains without the typecast */ | |
7597 func_all = (fptr_T)(func_all(&cc, c)); | |
7598 else /* just copy */ | |
772 | 7599 cc = c; |
7600 | |
7601 #ifdef FEAT_MBYTE | |
7602 if (has_mbyte) | |
7 | 7603 { |
2307
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7604 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
|
7605 |
7 | 7606 if (copy) |
772 | 7607 mb_char2bytes(cc, dst); |
7608 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
|
7609 if (enc_utf8) |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7610 { |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7611 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
|
7612 |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7613 /* 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
|
7614 * 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
|
7615 if (clen < totlen) |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7616 { |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7617 if (copy) |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7618 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
|
7619 (size_t)(totlen - clen)); |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7620 dst += totlen - clen; |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7621 } |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7622 } |
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7623 src += totlen - 1; |
7 | 7624 } |
7625 else | |
7626 #endif | |
7627 if (copy) | |
772 | 7628 *dst = cc; |
7 | 7629 dst++; |
7630 } | |
7631 else | |
7632 { | |
7633 if (REG_MULTI) | |
7634 { | |
7635 clnum = reg_mmatch->startpos[no].lnum; | |
7636 if (clnum < 0 || reg_mmatch->endpos[no].lnum < 0) | |
7637 s = NULL; | |
7638 else | |
7639 { | |
7640 s = reg_getline(clnum) + reg_mmatch->startpos[no].col; | |
7641 if (reg_mmatch->endpos[no].lnum == clnum) | |
7642 len = reg_mmatch->endpos[no].col | |
7643 - reg_mmatch->startpos[no].col; | |
7644 else | |
7645 len = (int)STRLEN(s); | |
7646 } | |
7647 } | |
7648 else | |
7649 { | |
7650 s = reg_match->startp[no]; | |
7651 if (reg_match->endp[no] == NULL) | |
7652 s = NULL; | |
7653 else | |
7654 len = (int)(reg_match->endp[no] - s); | |
7655 } | |
7656 if (s != NULL) | |
7657 { | |
7658 for (;;) | |
7659 { | |
7660 if (len == 0) | |
7661 { | |
7662 if (REG_MULTI) | |
7663 { | |
7664 if (reg_mmatch->endpos[no].lnum == clnum) | |
7665 break; | |
7666 if (copy) | |
7667 *dst = CAR; | |
7668 ++dst; | |
7669 s = reg_getline(++clnum); | |
7670 if (reg_mmatch->endpos[no].lnum == clnum) | |
7671 len = reg_mmatch->endpos[no].col; | |
7672 else | |
7673 len = (int)STRLEN(s); | |
7674 } | |
7675 else | |
7676 break; | |
7677 } | |
7678 else if (*s == NUL) /* we hit NUL. */ | |
7679 { | |
7680 if (copy) | |
7681 EMSG(_(e_re_damg)); | |
7682 goto exit; | |
7683 } | |
7684 else | |
7685 { | |
7686 if (backslash && (*s == CAR || *s == '\\')) | |
7687 { | |
7688 /* | |
7689 * Insert a backslash in front of a CR, otherwise | |
7690 * it will be replaced by a line break. | |
7691 * Number of backslashes will be halved later, | |
7692 * double them here. | |
7693 */ | |
7694 if (copy) | |
7695 { | |
7696 dst[0] = '\\'; | |
7697 dst[1] = *s; | |
7698 } | |
7699 dst += 2; | |
7700 } | |
7701 else | |
7702 { | |
772 | 7703 #ifdef FEAT_MBYTE |
7704 if (has_mbyte) | |
7705 c = mb_ptr2char(s); | |
7706 else | |
7707 #endif | |
7708 c = *s; | |
7709 | |
4244 | 7710 if (func_one != (fptr_T)NULL) |
7711 /* Turbo C complains without the typecast */ | |
7712 func_one = (fptr_T)(func_one(&cc, c)); | |
7713 else if (func_all != (fptr_T)NULL) | |
7714 /* Turbo C complains without the typecast */ | |
7715 func_all = (fptr_T)(func_all(&cc, c)); | |
7716 else /* just copy */ | |
772 | 7717 cc = c; |
7718 | |
7719 #ifdef FEAT_MBYTE | |
7720 if (has_mbyte) | |
7 | 7721 { |
1332 | 7722 int l; |
7723 | |
7724 /* Copy composing characters separately, one | |
7725 * at a time. */ | |
7726 if (enc_utf8) | |
7727 l = utf_ptr2len(s) - 1; | |
7728 else | |
7729 l = mb_ptr2len(s) - 1; | |
772 | 7730 |
7731 s += l; | |
7732 len -= l; | |
7733 if (copy) | |
7734 mb_char2bytes(cc, dst); | |
7735 dst += mb_char2len(cc) - 1; | |
7 | 7736 } |
772 | 7737 else |
7738 #endif | |
7739 if (copy) | |
7740 *dst = cc; | |
7741 dst++; | |
7 | 7742 } |
772 | 7743 |
7 | 7744 ++s; |
7745 --len; | |
7746 } | |
7747 } | |
7748 } | |
7749 no = -1; | |
7750 } | |
7751 } | |
7752 if (copy) | |
7753 *dst = NUL; | |
7754 | |
7755 exit: | |
7756 return (int)((dst - dest) + 1); | |
7757 } | |
7758 | |
7759 #ifdef FEAT_EVAL | |
2012 | 7760 static char_u *reg_getline_submatch __ARGS((linenr_T lnum)); |
7761 | |
7 | 7762 /* |
2011 | 7763 * Call reg_getline() with the line numbers from the submatch. If a |
7764 * substitute() was used the reg_maxline and other values have been | |
7765 * overwritten. | |
7766 */ | |
7767 static char_u * | |
7768 reg_getline_submatch(lnum) | |
7769 linenr_T lnum; | |
7770 { | |
7771 char_u *s; | |
7772 linenr_T save_first = reg_firstlnum; | |
7773 linenr_T save_max = reg_maxline; | |
7774 | |
7775 reg_firstlnum = submatch_firstlnum; | |
7776 reg_maxline = submatch_maxline; | |
7777 | |
7778 s = reg_getline(lnum); | |
7779 | |
7780 reg_firstlnum = save_first; | |
7781 reg_maxline = save_max; | |
7782 return s; | |
7783 } | |
7784 | |
7785 /* | |
1209 | 7786 * Used for the submatch() function: get the string from the n'th submatch in |
7 | 7787 * allocated memory. |
7788 * Returns NULL when not in a ":s" command and for a non-existing submatch. | |
7789 */ | |
7790 char_u * | |
7791 reg_submatch(no) | |
7792 int no; | |
7793 { | |
7794 char_u *retval = NULL; | |
7795 char_u *s; | |
7796 int len; | |
7797 int round; | |
7798 linenr_T lnum; | |
7799 | |
840 | 7800 if (!can_f_submatch || no < 0) |
7 | 7801 return NULL; |
7802 | |
7803 if (submatch_match == NULL) | |
7804 { | |
7805 /* | |
7806 * First round: compute the length and allocate memory. | |
7807 * Second round: copy the text. | |
7808 */ | |
7809 for (round = 1; round <= 2; ++round) | |
7810 { | |
7811 lnum = submatch_mmatch->startpos[no].lnum; | |
7812 if (lnum < 0 || submatch_mmatch->endpos[no].lnum < 0) | |
7813 return NULL; | |
7814 | |
2011 | 7815 s = reg_getline_submatch(lnum) + submatch_mmatch->startpos[no].col; |
7 | 7816 if (s == NULL) /* anti-crash check, cannot happen? */ |
7817 break; | |
7818 if (submatch_mmatch->endpos[no].lnum == lnum) | |
7819 { | |
7820 /* Within one line: take form start to end col. */ | |
7821 len = submatch_mmatch->endpos[no].col | |
7822 - submatch_mmatch->startpos[no].col; | |
7823 if (round == 2) | |
418 | 7824 vim_strncpy(retval, s, len); |
7 | 7825 ++len; |
7826 } | |
7827 else | |
7828 { | |
7829 /* Multiple lines: take start line from start col, middle | |
7830 * lines completely and end line up to end col. */ | |
7831 len = (int)STRLEN(s); | |
7832 if (round == 2) | |
7833 { | |
7834 STRCPY(retval, s); | |
7835 retval[len] = '\n'; | |
7836 } | |
7837 ++len; | |
7838 ++lnum; | |
7839 while (lnum < submatch_mmatch->endpos[no].lnum) | |
7840 { | |
2011 | 7841 s = reg_getline_submatch(lnum++); |
7 | 7842 if (round == 2) |
7843 STRCPY(retval + len, s); | |
7844 len += (int)STRLEN(s); | |
7845 if (round == 2) | |
7846 retval[len] = '\n'; | |
7847 ++len; | |
7848 } | |
7849 if (round == 2) | |
2011 | 7850 STRNCPY(retval + len, reg_getline_submatch(lnum), |
7 | 7851 submatch_mmatch->endpos[no].col); |
7852 len += submatch_mmatch->endpos[no].col; | |
7853 if (round == 2) | |
7854 retval[len] = NUL; | |
7855 ++len; | |
7856 } | |
7857 | |
840 | 7858 if (retval == NULL) |
7 | 7859 { |
7860 retval = lalloc((long_u)len, TRUE); | |
840 | 7861 if (retval == NULL) |
7 | 7862 return NULL; |
7863 } | |
7864 } | |
7865 } | |
7866 else | |
7867 { | |
1815 | 7868 s = submatch_match->startp[no]; |
7869 if (s == NULL || submatch_match->endp[no] == NULL) | |
7 | 7870 retval = NULL; |
7871 else | |
7872 retval = vim_strnsave(s, (int)(submatch_match->endp[no] - s)); | |
7873 } | |
7874 | |
7875 return retval; | |
7876 } | |
5794 | 7877 |
7878 /* | |
7879 * Used for the submatch() function with the optional non-zero argument: get | |
7880 * the list of strings from the n'th submatch in allocated memory with NULs | |
7881 * represented in NLs. | |
7882 * Returns a list of allocated strings. Returns NULL when not in a ":s" | |
7883 * command, for a non-existing submatch and for any error. | |
7884 */ | |
7885 list_T * | |
7886 reg_submatch_list(no) | |
7887 int no; | |
7888 { | |
7889 char_u *s; | |
7890 linenr_T slnum; | |
7891 linenr_T elnum; | |
7892 colnr_T scol; | |
7893 colnr_T ecol; | |
7894 int i; | |
7895 list_T *list; | |
7896 int error = FALSE; | |
7897 | |
7898 if (!can_f_submatch || no < 0) | |
7899 return NULL; | |
7900 | |
7901 if (submatch_match == NULL) | |
7902 { | |
7903 slnum = submatch_mmatch->startpos[no].lnum; | |
7904 elnum = submatch_mmatch->endpos[no].lnum; | |
7905 if (slnum < 0 || elnum < 0) | |
7906 return NULL; | |
7907 | |
7908 scol = submatch_mmatch->startpos[no].col; | |
7909 ecol = submatch_mmatch->endpos[no].col; | |
7910 | |
7911 list = list_alloc(); | |
7912 if (list == NULL) | |
7913 return NULL; | |
7914 | |
7915 s = reg_getline_submatch(slnum) + scol; | |
7916 if (slnum == elnum) | |
7917 { | |
7918 if (list_append_string(list, s, ecol - scol) == FAIL) | |
7919 error = TRUE; | |
7920 } | |
7921 else | |
7922 { | |
7923 if (list_append_string(list, s, -1) == FAIL) | |
7924 error = TRUE; | |
7925 for (i = 1; i < elnum - slnum; i++) | |
7926 { | |
7927 s = reg_getline_submatch(slnum + i); | |
7928 if (list_append_string(list, s, -1) == FAIL) | |
7929 error = TRUE; | |
7930 } | |
7931 s = reg_getline_submatch(elnum); | |
7932 if (list_append_string(list, s, ecol) == FAIL) | |
7933 error = TRUE; | |
7934 } | |
7935 } | |
7936 else | |
7937 { | |
7938 s = submatch_match->startp[no]; | |
7939 if (s == NULL || submatch_match->endp[no] == NULL) | |
7940 return NULL; | |
7941 list = list_alloc(); | |
7942 if (list == NULL) | |
7943 return NULL; | |
7944 if (list_append_string(list, s, | |
7945 (int)(submatch_match->endp[no] - s)) == FAIL) | |
7946 error = TRUE; | |
7947 } | |
7948 | |
7949 if (error) | |
7950 { | |
7951 list_free(list, TRUE); | |
7952 return NULL; | |
7953 } | |
7954 return list; | |
7955 } | |
7 | 7956 #endif |
4444 | 7957 |
7958 static regengine_T bt_regengine = | |
7959 { | |
7960 bt_regcomp, | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7961 bt_regfree, |
4444 | 7962 bt_regexec_nl, |
7963 bt_regexec_multi | |
7964 #ifdef DEBUG | |
7965 ,(char_u *)"" | |
7966 #endif | |
7967 }; | |
7968 | |
7969 | |
7970 #include "regexp_nfa.c" | |
7971 | |
7972 static regengine_T nfa_regengine = | |
7973 { | |
7974 nfa_regcomp, | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7975 nfa_regfree, |
4444 | 7976 nfa_regexec_nl, |
7977 nfa_regexec_multi | |
7978 #ifdef DEBUG | |
7979 ,(char_u *)"" | |
7980 #endif | |
7981 }; | |
7982 | |
7983 /* Which regexp engine to use? Needed for vim_regcomp(). | |
7984 * Must match with 'regexpengine'. */ | |
7985 static int regexp_engine = 0; | |
7986 #define AUTOMATIC_ENGINE 0 | |
7987 #define BACKTRACKING_ENGINE 1 | |
7988 #define NFA_ENGINE 2 | |
7989 #ifdef DEBUG | |
7990 static char_u regname[][30] = { | |
7991 "AUTOMATIC Regexp Engine", | |
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
7992 "BACKTRACKING Regexp Engine", |
4444 | 7993 "NFA Regexp Engine" |
7994 }; | |
7995 #endif | |
7996 | |
7997 /* | |
7998 * Compile a regular expression into internal code. | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7999 * Returns the program in allocated memory. |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8000 * Use vim_regfree() to free the memory. |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8001 * Returns NULL for an error. |
4444 | 8002 */ |
8003 regprog_T * | |
8004 vim_regcomp(expr_arg, re_flags) | |
8005 char_u *expr_arg; | |
8006 int re_flags; | |
8007 { | |
8008 regprog_T *prog = NULL; | |
8009 char_u *expr = expr_arg; | |
8010 | |
8011 regexp_engine = p_re; | |
8012 | |
8013 /* Check for prefix "\%#=", that sets the regexp engine */ | |
8014 if (STRNCMP(expr, "\\%#=", 4) == 0) | |
8015 { | |
8016 int newengine = expr[4] - '0'; | |
8017 | |
8018 if (newengine == AUTOMATIC_ENGINE | |
8019 || newengine == BACKTRACKING_ENGINE | |
8020 || newengine == NFA_ENGINE) | |
8021 { | |
8022 regexp_engine = expr[4] - '0'; | |
8023 expr += 5; | |
8024 #ifdef DEBUG | |
8025 EMSG3("New regexp mode selected (%d): %s", regexp_engine, | |
8026 regname[newengine]); | |
8027 #endif | |
8028 } | |
8029 else | |
8030 { | |
8031 EMSG(_("E864: \\%#= can only be followed by 0, 1, or 2. The automatic engine will be used ")); | |
8032 regexp_engine = AUTOMATIC_ENGINE; | |
8033 } | |
8034 } | |
8035 #ifdef DEBUG | |
8036 bt_regengine.expr = expr; | |
8037 nfa_regengine.expr = expr; | |
8038 #endif | |
8039 | |
8040 /* | |
8041 * First try the NFA engine, unless backtracking was requested. | |
8042 */ | |
8043 if (regexp_engine != BACKTRACKING_ENGINE) | |
8044 prog = nfa_regengine.regcomp(expr, re_flags); | |
8045 else | |
8046 prog = bt_regengine.regcomp(expr, re_flags); | |
8047 | |
8048 if (prog == NULL) /* error compiling regexp with initial engine */ | |
8049 { | |
4460 | 8050 #ifdef BT_REGEXP_DEBUG_LOG |
4444 | 8051 if (regexp_engine != BACKTRACKING_ENGINE) /* debugging log for NFA */ |
8052 { | |
8053 FILE *f; | |
4460 | 8054 f = fopen(BT_REGEXP_DEBUG_LOG_NAME, "a"); |
4444 | 8055 if (f) |
8056 { | |
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8057 fprintf(f, "Syntax error in \"%s\"\n", expr); |
4444 | 8058 fclose(f); |
8059 } | |
8060 else | |
4460 | 8061 EMSG2("(NFA) Could not open \"%s\" to write !!!", |
8062 BT_REGEXP_DEBUG_LOG_NAME); | |
4444 | 8063 } |
8064 #endif | |
8065 /* | |
5353 | 8066 * If the NFA engine failed, the backtracking engine won't work either. |
8067 * | |
4444 | 8068 if (regexp_engine == AUTOMATIC_ENGINE) |
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8069 prog = bt_regengine.regcomp(expr, re_flags); |
5353 | 8070 */ |
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8071 } |
4444 | 8072 |
8073 return prog; | |
8074 } | |
8075 | |
8076 /* | |
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8077 * 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
|
8078 */ |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8079 void |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8080 vim_regfree(prog) |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8081 regprog_T *prog; |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8082 { |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8083 if (prog != NULL) |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8084 prog->engine->regfree(prog); |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8085 } |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8086 |
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8087 /* |
4444 | 8088 * Match a regexp against a string. |
8089 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
8090 * Uses curbuf for line count and 'iskeyword'. | |
8091 * | |
8092 * Return TRUE if there is a match, FALSE if not. | |
8093 */ | |
8094 int | |
8095 vim_regexec(rmp, line, col) | |
8096 regmatch_T *rmp; | |
8097 char_u *line; /* string to match against */ | |
8098 colnr_T col; /* column to start looking for match */ | |
8099 { | |
5838 | 8100 return rmp->regprog->engine->regexec_nl(rmp, line, col, FALSE); |
4444 | 8101 } |
8102 | |
8103 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
8104 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
8105 /* | |
8106 * Like vim_regexec(), but consider a "\n" in "line" to be a line break. | |
8107 */ | |
8108 int | |
8109 vim_regexec_nl(rmp, line, col) | |
8110 regmatch_T *rmp; | |
8111 char_u *line; | |
8112 colnr_T col; | |
8113 { | |
5838 | 8114 return rmp->regprog->engine->regexec_nl(rmp, line, col, TRUE); |
4444 | 8115 } |
8116 #endif | |
8117 | |
8118 /* | |
8119 * Match a regexp against multiple lines. | |
8120 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
8121 * Uses curbuf for line count and 'iskeyword'. | |
8122 * | |
8123 * Return zero if there is no match. Return number of lines contained in the | |
8124 * match otherwise. | |
8125 */ | |
8126 long | |
8127 vim_regexec_multi(rmp, win, buf, lnum, col, tm) | |
8128 regmmatch_T *rmp; | |
8129 win_T *win; /* window in which to search or NULL */ | |
8130 buf_T *buf; /* buffer in which to search */ | |
8131 linenr_T lnum; /* nr of line to start looking for match */ | |
8132 colnr_T col; /* column to start looking for match */ | |
8133 proftime_T *tm; /* timeout limit or NULL */ | |
8134 { | |
8135 return rmp->regprog->engine->regexec_multi(rmp, win, buf, lnum, col, tm); | |
8136 } |