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
|
|
41 #include "vim.h"
|
|
42
|
|
43 #undef DEBUG
|
|
44
|
|
45 /*
|
|
46 * The "internal use only" fields in regexp.h are present to pass info from
|
|
47 * compile to execute that permits the execute phase to run lots faster on
|
|
48 * simple cases. They are:
|
|
49 *
|
|
50 * regstart char that must begin a match; NUL if none obvious; Can be a
|
|
51 * multi-byte character.
|
|
52 * reganch is the match anchored (at beginning-of-line only)?
|
|
53 * regmust string (pointer into program) that match must include, or NULL
|
|
54 * regmlen length of regmust string
|
|
55 * regflags RF_ values or'ed together
|
|
56 *
|
|
57 * Regstart and reganch permit very fast decisions on suitable starting points
|
|
58 * for a match, cutting down the work a lot. Regmust permits fast rejection
|
|
59 * of lines that cannot possibly match. The regmust tests are costly enough
|
|
60 * that vim_regcomp() supplies a regmust only if the r.e. contains something
|
|
61 * potentially expensive (at present, the only such thing detected is * or +
|
|
62 * at the start of the r.e., which can involve a lot of backup). Regmlen is
|
|
63 * supplied because the test in vim_regexec() needs it and vim_regcomp() is
|
|
64 * computing it anyway.
|
|
65 */
|
|
66
|
|
67 /*
|
|
68 * Structure for regexp "program". This is essentially a linear encoding
|
|
69 * of a nondeterministic finite-state machine (aka syntax charts or
|
|
70 * "railroad normal form" in parsing technology). Each node is an opcode
|
|
71 * plus a "next" pointer, possibly plus an operand. "Next" pointers of
|
|
72 * all nodes except BRANCH and BRACES_COMPLEX implement concatenation; a "next"
|
|
73 * pointer with a BRANCH on both ends of it is connecting two alternatives.
|
|
74 * (Here we have one of the subtle syntax dependencies: an individual BRANCH
|
|
75 * (as opposed to a collection of them) is never concatenated with anything
|
|
76 * because of operator precedence). The "next" pointer of a BRACES_COMPLEX
|
167
|
77 * node points to the node after the stuff to be repeated.
|
|
78 * The operand of some types of node is a literal string; for others, it is a
|
|
79 * node leading into a sub-FSM. In particular, the operand of a BRANCH node
|
|
80 * is the first node of the branch.
|
|
81 * (NB this is *not* a tree structure: the tail of the branch connects to the
|
|
82 * thing following the set of BRANCHes.)
|
7
|
83 *
|
|
84 * pattern is coded like:
|
|
85 *
|
|
86 * +-----------------+
|
|
87 * | V
|
|
88 * <aa>\|<bb> BRANCH <aa> BRANCH <bb> --> END
|
|
89 * | ^ | ^
|
|
90 * +------+ +----------+
|
|
91 *
|
|
92 *
|
|
93 * +------------------+
|
|
94 * V |
|
|
95 * <aa>* BRANCH BRANCH <aa> --> BACK BRANCH --> NOTHING --> END
|
|
96 * | | ^ ^
|
|
97 * | +---------------+ |
|
|
98 * +---------------------------------------------+
|
|
99 *
|
|
100 *
|
167
|
101 * +----------------------+
|
|
102 * V |
|
|
103 * <aa>\+ BRANCH <aa> --> BRANCH --> BACK BRANCH --> NOTHING --> END
|
|
104 * | | ^ ^
|
|
105 * | +----------+ |
|
|
106 * +-------------------------------------------------+
|
|
107 *
|
|
108 *
|
7
|
109 * +-------------------------+
|
|
110 * V |
|
|
111 * <aa>\{} BRANCH BRACE_LIMITS --> BRACE_COMPLEX <aa> --> BACK END
|
|
112 * | | ^
|
|
113 * | +----------------+
|
|
114 * +-----------------------------------------------+
|
|
115 *
|
|
116 *
|
|
117 * <aa>\@!<bb> BRANCH NOMATCH <aa> --> END <bb> --> END
|
|
118 * | | ^ ^
|
|
119 * | +----------------+ |
|
|
120 * +--------------------------------+
|
|
121 *
|
|
122 * +---------+
|
|
123 * | V
|
|
124 * \z[abc] BRANCH BRANCH a BRANCH b BRANCH c BRANCH NOTHING --> END
|
|
125 * | | | | ^ ^
|
|
126 * | | | +-----+ |
|
|
127 * | | +----------------+ |
|
|
128 * | +---------------------------+ |
|
|
129 * +------------------------------------------------------+
|
|
130 *
|
|
131 * They all start with a BRANCH for "\|" alternaties, even when there is only
|
|
132 * one alternative.
|
|
133 */
|
|
134
|
|
135 /*
|
|
136 * The opcodes are:
|
|
137 */
|
|
138
|
|
139 /* definition number opnd? meaning */
|
|
140 #define END 0 /* End of program or NOMATCH operand. */
|
|
141 #define BOL 1 /* Match "" at beginning of line. */
|
|
142 #define EOL 2 /* Match "" at end of line. */
|
|
143 #define BRANCH 3 /* node Match this alternative, or the
|
|
144 * next... */
|
|
145 #define BACK 4 /* Match "", "next" ptr points backward. */
|
|
146 #define EXACTLY 5 /* str Match this string. */
|
|
147 #define NOTHING 6 /* Match empty string. */
|
|
148 #define STAR 7 /* node Match this (simple) thing 0 or more
|
|
149 * times. */
|
|
150 #define PLUS 8 /* node Match this (simple) thing 1 or more
|
|
151 * times. */
|
|
152 #define MATCH 9 /* node match the operand zero-width */
|
|
153 #define NOMATCH 10 /* node check for no match with operand */
|
|
154 #define BEHIND 11 /* node look behind for a match with operand */
|
|
155 #define NOBEHIND 12 /* node look behind for no match with operand */
|
|
156 #define SUBPAT 13 /* node match the operand here */
|
|
157 #define BRACE_SIMPLE 14 /* node Match this (simple) thing between m and
|
|
158 * n times (\{m,n\}). */
|
|
159 #define BOW 15 /* Match "" after [^a-zA-Z0-9_] */
|
|
160 #define EOW 16 /* Match "" at [^a-zA-Z0-9_] */
|
|
161 #define BRACE_LIMITS 17 /* nr nr define the min & max for BRACE_SIMPLE
|
|
162 * and BRACE_COMPLEX. */
|
|
163 #define NEWL 18 /* Match line-break */
|
|
164 #define BHPOS 19 /* End position for BEHIND or NOBEHIND */
|
|
165
|
|
166
|
|
167 /* character classes: 20-48 normal, 50-78 include a line-break */
|
|
168 #define ADD_NL 30
|
|
169 #define FIRST_NL ANY + ADD_NL
|
|
170 #define ANY 20 /* Match any one character. */
|
|
171 #define ANYOF 21 /* str Match any character in this string. */
|
|
172 #define ANYBUT 22 /* str Match any character not in this
|
|
173 * string. */
|
|
174 #define IDENT 23 /* Match identifier char */
|
|
175 #define SIDENT 24 /* Match identifier char but no digit */
|
|
176 #define KWORD 25 /* Match keyword char */
|
|
177 #define SKWORD 26 /* Match word char but no digit */
|
|
178 #define FNAME 27 /* Match file name char */
|
|
179 #define SFNAME 28 /* Match file name char but no digit */
|
|
180 #define PRINT 29 /* Match printable char */
|
|
181 #define SPRINT 30 /* Match printable char but no digit */
|
|
182 #define WHITE 31 /* Match whitespace char */
|
|
183 #define NWHITE 32 /* Match non-whitespace char */
|
|
184 #define DIGIT 33 /* Match digit char */
|
|
185 #define NDIGIT 34 /* Match non-digit char */
|
|
186 #define HEX 35 /* Match hex char */
|
|
187 #define NHEX 36 /* Match non-hex char */
|
|
188 #define OCTAL 37 /* Match octal char */
|
|
189 #define NOCTAL 38 /* Match non-octal char */
|
|
190 #define WORD 39 /* Match word char */
|
|
191 #define NWORD 40 /* Match non-word char */
|
|
192 #define HEAD 41 /* Match head char */
|
|
193 #define NHEAD 42 /* Match non-head char */
|
|
194 #define ALPHA 43 /* Match alpha char */
|
|
195 #define NALPHA 44 /* Match non-alpha char */
|
|
196 #define LOWER 45 /* Match lowercase char */
|
|
197 #define NLOWER 46 /* Match non-lowercase char */
|
|
198 #define UPPER 47 /* Match uppercase char */
|
|
199 #define NUPPER 48 /* Match non-uppercase char */
|
|
200 #define LAST_NL NUPPER + ADD_NL
|
|
201 #define WITH_NL(op) ((op) >= FIRST_NL && (op) <= LAST_NL)
|
|
202
|
|
203 #define MOPEN 80 /* -89 Mark this point in input as start of
|
|
204 * \( subexpr. MOPEN + 0 marks start of
|
|
205 * match. */
|
|
206 #define MCLOSE 90 /* -99 Analogous to MOPEN. MCLOSE + 0 marks
|
|
207 * end of match. */
|
|
208 #define BACKREF 100 /* -109 node Match same string again \1-\9 */
|
|
209
|
|
210 #ifdef FEAT_SYN_HL
|
|
211 # define ZOPEN 110 /* -119 Mark this point in input as start of
|
|
212 * \z( subexpr. */
|
|
213 # define ZCLOSE 120 /* -129 Analogous to ZOPEN. */
|
|
214 # define ZREF 130 /* -139 node Match external submatch \z1-\z9 */
|
|
215 #endif
|
|
216
|
|
217 #define BRACE_COMPLEX 140 /* -149 node Match nodes between m & n times */
|
|
218
|
|
219 #define NOPEN 150 /* Mark this point in input as start of
|
|
220 \%( subexpr. */
|
|
221 #define NCLOSE 151 /* Analogous to NOPEN. */
|
|
222
|
|
223 #define MULTIBYTECODE 200 /* mbc Match one multi-byte character */
|
|
224 #define RE_BOF 201 /* Match "" at beginning of file. */
|
|
225 #define RE_EOF 202 /* Match "" at end of file. */
|
|
226 #define CURSOR 203 /* Match location of cursor. */
|
|
227
|
|
228 #define RE_LNUM 204 /* nr cmp Match line number */
|
|
229 #define RE_COL 205 /* nr cmp Match column number */
|
|
230 #define RE_VCOL 206 /* nr cmp Match virtual column number */
|
|
231
|
|
232 /*
|
|
233 * Magic characters have a special meaning, they don't match literally.
|
|
234 * Magic characters are negative. This separates them from literal characters
|
|
235 * (possibly multi-byte). Only ASCII characters can be Magic.
|
|
236 */
|
|
237 #define Magic(x) ((int)(x) - 256)
|
|
238 #define un_Magic(x) ((x) + 256)
|
|
239 #define is_Magic(x) ((x) < 0)
|
|
240
|
|
241 static int no_Magic __ARGS((int x));
|
|
242 static int toggle_Magic __ARGS((int x));
|
|
243
|
|
244 static int
|
|
245 no_Magic(x)
|
|
246 int x;
|
|
247 {
|
|
248 if (is_Magic(x))
|
|
249 return un_Magic(x);
|
|
250 return x;
|
|
251 }
|
|
252
|
|
253 static int
|
|
254 toggle_Magic(x)
|
|
255 int x;
|
|
256 {
|
|
257 if (is_Magic(x))
|
|
258 return un_Magic(x);
|
|
259 return Magic(x);
|
|
260 }
|
|
261
|
|
262 /*
|
|
263 * The first byte of the regexp internal "program" is actually this magic
|
|
264 * number; the start node begins in the second byte. It's used to catch the
|
|
265 * most severe mutilation of the program by the caller.
|
|
266 */
|
|
267
|
|
268 #define REGMAGIC 0234
|
|
269
|
|
270 /*
|
|
271 * Opcode notes:
|
|
272 *
|
|
273 * BRANCH The set of branches constituting a single choice are hooked
|
|
274 * together with their "next" pointers, since precedence prevents
|
|
275 * anything being concatenated to any individual branch. The
|
|
276 * "next" pointer of the last BRANCH in a choice points to the
|
|
277 * thing following the whole choice. This is also where the
|
|
278 * final "next" pointer of each individual branch points; each
|
|
279 * branch starts with the operand node of a BRANCH node.
|
|
280 *
|
|
281 * BACK Normal "next" pointers all implicitly point forward; BACK
|
|
282 * exists to make loop structures possible.
|
|
283 *
|
|
284 * STAR,PLUS '=', and complex '*' and '+', are implemented as circular
|
|
285 * BRANCH structures using BACK. Simple cases (one character
|
|
286 * per match) are implemented with STAR and PLUS for speed
|
|
287 * and to minimize recursive plunges.
|
|
288 *
|
|
289 * BRACE_LIMITS This is always followed by a BRACE_SIMPLE or BRACE_COMPLEX
|
|
290 * node, and defines the min and max limits to be used for that
|
|
291 * node.
|
|
292 *
|
|
293 * MOPEN,MCLOSE ...are numbered at compile time.
|
|
294 * ZOPEN,ZCLOSE ...ditto
|
|
295 */
|
|
296
|
|
297 /*
|
|
298 * A node is one char of opcode followed by two chars of "next" pointer.
|
|
299 * "Next" pointers are stored as two 8-bit bytes, high order first. The
|
|
300 * value is a positive offset from the opcode of the node containing it.
|
|
301 * An operand, if any, simply follows the node. (Note that much of the
|
|
302 * code generation knows about this implicit relationship.)
|
|
303 *
|
|
304 * Using two bytes for the "next" pointer is vast overkill for most things,
|
|
305 * but allows patterns to get big without disasters.
|
|
306 */
|
|
307 #define OP(p) ((int)*(p))
|
|
308 #define NEXT(p) (((*((p) + 1) & 0377) << 8) + (*((p) + 2) & 0377))
|
|
309 #define OPERAND(p) ((p) + 3)
|
|
310 /* Obtain an operand that was stored as four bytes, MSB first. */
|
|
311 #define OPERAND_MIN(p) (((long)(p)[3] << 24) + ((long)(p)[4] << 16) \
|
|
312 + ((long)(p)[5] << 8) + (long)(p)[6])
|
|
313 /* Obtain a second operand stored as four bytes. */
|
|
314 #define OPERAND_MAX(p) OPERAND_MIN((p) + 4)
|
|
315 /* Obtain a second single-byte operand stored after a four bytes operand. */
|
|
316 #define OPERAND_CMP(p) (p)[7]
|
|
317
|
|
318 /*
|
|
319 * Utility definitions.
|
|
320 */
|
|
321 #define UCHARAT(p) ((int)*(char_u *)(p))
|
|
322
|
|
323 /* Used for an error (down from) vim_regcomp(): give the error message, set
|
|
324 * rc_did_emsg and return NULL */
|
|
325 #define EMSG_RET_NULL(m) { EMSG(m); rc_did_emsg = TRUE; return NULL; }
|
|
326 #define EMSG_M_RET_NULL(m, c) { EMSG2(m, c ? "" : "\\"); rc_did_emsg = TRUE; return NULL; }
|
|
327 #define EMSG_RET_FAIL(m) { EMSG(m); rc_did_emsg = TRUE; return FAIL; }
|
|
328 #define EMSG_ONE_RET_NULL EMSG_M_RET_NULL(_("E369: invalid item in %s%%[]"), reg_magic == MAGIC_ALL)
|
|
329
|
|
330 #define MAX_LIMIT (32767L << 16L)
|
|
331
|
|
332 static int re_multi_type __ARGS((int));
|
|
333 static int cstrncmp __ARGS((char_u *s1, char_u *s2, int *n));
|
|
334 static char_u *cstrchr __ARGS((char_u *, int));
|
|
335
|
|
336 #ifdef DEBUG
|
|
337 static void regdump __ARGS((char_u *, regprog_T *));
|
|
338 static char_u *regprop __ARGS((char_u *));
|
|
339 #endif
|
|
340
|
|
341 #define NOT_MULTI 0
|
|
342 #define MULTI_ONE 1
|
|
343 #define MULTI_MULT 2
|
|
344 /*
|
|
345 * Return NOT_MULTI if c is not a "multi" operator.
|
|
346 * Return MULTI_ONE if c is a single "multi" operator.
|
|
347 * Return MULTI_MULT if c is a multi "multi" operator.
|
|
348 */
|
|
349 static int
|
|
350 re_multi_type(c)
|
|
351 int c;
|
|
352 {
|
|
353 if (c == Magic('@') || c == Magic('=') || c == Magic('?'))
|
|
354 return MULTI_ONE;
|
|
355 if (c == Magic('*') || c == Magic('+') || c == Magic('{'))
|
|
356 return MULTI_MULT;
|
|
357 return NOT_MULTI;
|
|
358 }
|
|
359
|
|
360 /*
|
|
361 * Flags to be passed up and down.
|
|
362 */
|
|
363 #define HASWIDTH 0x1 /* Known never to match null string. */
|
|
364 #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */
|
|
365 #define SPSTART 0x4 /* Starts with * or +. */
|
|
366 #define HASNL 0x8 /* Contains some \n. */
|
|
367 #define HASLOOKBH 0x10 /* Contains "\@<=" or "\@<!". */
|
|
368 #define WORST 0 /* Worst case. */
|
|
369
|
|
370 /*
|
|
371 * When regcode is set to this value, code is not emitted and size is computed
|
|
372 * instead.
|
|
373 */
|
|
374 #define JUST_CALC_SIZE ((char_u *) -1)
|
|
375
|
|
376 static char_u *reg_prev_sub;
|
|
377
|
|
378 /*
|
|
379 * REGEXP_INRANGE contains all characters which are always special in a []
|
|
380 * range after '\'.
|
|
381 * REGEXP_ABBR contains all characters which act as abbreviations after '\'.
|
|
382 * These are:
|
|
383 * \n - New line (NL).
|
|
384 * \r - Carriage Return (CR).
|
|
385 * \t - Tab (TAB).
|
|
386 * \e - Escape (ESC).
|
|
387 * \b - Backspace (Ctrl_H).
|
24
|
388 * \d - Character code in decimal, eg \d123
|
|
389 * \o - Character code in octal, eg \o80
|
|
390 * \x - Character code in hex, eg \x4a
|
|
391 * \u - Multibyte character code, eg \u20ac
|
|
392 * \U - Long multibyte character code, eg \U12345678
|
7
|
393 */
|
|
394 static char_u REGEXP_INRANGE[] = "]^-n\\";
|
24
|
395 static char_u REGEXP_ABBR[] = "nrtebdoxuU";
|
7
|
396
|
|
397 static int backslash_trans __ARGS((int c));
|
167
|
398 static int get_char_class __ARGS((char_u **pp));
|
|
399 static int get_equi_class __ARGS((char_u **pp));
|
|
400 static void reg_equi_class __ARGS((int c));
|
|
401 static int get_coll_element __ARGS((char_u **pp));
|
7
|
402 static char_u *skip_anyof __ARGS((char_u *p));
|
|
403 static void init_class_tab __ARGS((void));
|
|
404
|
|
405 /*
|
|
406 * Translate '\x' to its control character, except "\n", which is Magic.
|
|
407 */
|
|
408 static int
|
|
409 backslash_trans(c)
|
|
410 int c;
|
|
411 {
|
|
412 switch (c)
|
|
413 {
|
|
414 case 'r': return CAR;
|
|
415 case 't': return TAB;
|
|
416 case 'e': return ESC;
|
|
417 case 'b': return BS;
|
|
418 }
|
|
419 return c;
|
|
420 }
|
|
421
|
|
422 /*
|
167
|
423 * Check for a character class name "[:name:]". "pp" points to the '['.
|
7
|
424 * Returns one of the CLASS_ items. CLASS_NONE means that no item was
|
|
425 * recognized. Otherwise "pp" is advanced to after the item.
|
|
426 */
|
|
427 static int
|
167
|
428 get_char_class(pp)
|
7
|
429 char_u **pp;
|
|
430 {
|
|
431 static const char *(class_names[]) =
|
|
432 {
|
|
433 "alnum:]",
|
|
434 #define CLASS_ALNUM 0
|
|
435 "alpha:]",
|
|
436 #define CLASS_ALPHA 1
|
|
437 "blank:]",
|
|
438 #define CLASS_BLANK 2
|
|
439 "cntrl:]",
|
|
440 #define CLASS_CNTRL 3
|
|
441 "digit:]",
|
|
442 #define CLASS_DIGIT 4
|
|
443 "graph:]",
|
|
444 #define CLASS_GRAPH 5
|
|
445 "lower:]",
|
|
446 #define CLASS_LOWER 6
|
|
447 "print:]",
|
|
448 #define CLASS_PRINT 7
|
|
449 "punct:]",
|
|
450 #define CLASS_PUNCT 8
|
|
451 "space:]",
|
|
452 #define CLASS_SPACE 9
|
|
453 "upper:]",
|
|
454 #define CLASS_UPPER 10
|
|
455 "xdigit:]",
|
|
456 #define CLASS_XDIGIT 11
|
|
457 "tab:]",
|
|
458 #define CLASS_TAB 12
|
|
459 "return:]",
|
|
460 #define CLASS_RETURN 13
|
|
461 "backspace:]",
|
|
462 #define CLASS_BACKSPACE 14
|
|
463 "escape:]",
|
|
464 #define CLASS_ESCAPE 15
|
|
465 };
|
|
466 #define CLASS_NONE 99
|
|
467 int i;
|
|
468
|
|
469 if ((*pp)[1] == ':')
|
|
470 {
|
|
471 for (i = 0; i < sizeof(class_names) / sizeof(*class_names); ++i)
|
|
472 if (STRNCMP(*pp + 2, class_names[i], STRLEN(class_names[i])) == 0)
|
|
473 {
|
|
474 *pp += STRLEN(class_names[i]) + 2;
|
|
475 return i;
|
|
476 }
|
|
477 }
|
|
478 return CLASS_NONE;
|
|
479 }
|
|
480
|
|
481 /*
|
|
482 * Specific version of character class functions.
|
|
483 * Using a table to keep this fast.
|
|
484 */
|
|
485 static short class_tab[256];
|
|
486
|
|
487 #define RI_DIGIT 0x01
|
|
488 #define RI_HEX 0x02
|
|
489 #define RI_OCTAL 0x04
|
|
490 #define RI_WORD 0x08
|
|
491 #define RI_HEAD 0x10
|
|
492 #define RI_ALPHA 0x20
|
|
493 #define RI_LOWER 0x40
|
|
494 #define RI_UPPER 0x80
|
|
495 #define RI_WHITE 0x100
|
|
496
|
|
497 static void
|
|
498 init_class_tab()
|
|
499 {
|
|
500 int i;
|
|
501 static int done = FALSE;
|
|
502
|
|
503 if (done)
|
|
504 return;
|
|
505
|
|
506 for (i = 0; i < 256; ++i)
|
|
507 {
|
|
508 if (i >= '0' && i <= '7')
|
|
509 class_tab[i] = RI_DIGIT + RI_HEX + RI_OCTAL + RI_WORD;
|
|
510 else if (i >= '8' && i <= '9')
|
|
511 class_tab[i] = RI_DIGIT + RI_HEX + RI_WORD;
|
|
512 else if (i >= 'a' && i <= 'f')
|
|
513 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER;
|
|
514 #ifdef EBCDIC
|
|
515 else if ((i >= 'g' && i <= 'i') || (i >= 'j' && i <= 'r')
|
|
516 || (i >= 's' && i <= 'z'))
|
|
517 #else
|
|
518 else if (i >= 'g' && i <= 'z')
|
|
519 #endif
|
|
520 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER;
|
|
521 else if (i >= 'A' && i <= 'F')
|
|
522 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER;
|
|
523 #ifdef EBCDIC
|
|
524 else if ((i >= 'G' && i <= 'I') || ( i >= 'J' && i <= 'R')
|
|
525 || (i >= 'S' && i <= 'Z'))
|
|
526 #else
|
|
527 else if (i >= 'G' && i <= 'Z')
|
|
528 #endif
|
|
529 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER;
|
|
530 else if (i == '_')
|
|
531 class_tab[i] = RI_WORD + RI_HEAD;
|
|
532 else
|
|
533 class_tab[i] = 0;
|
|
534 }
|
|
535 class_tab[' '] |= RI_WHITE;
|
|
536 class_tab['\t'] |= RI_WHITE;
|
|
537 done = TRUE;
|
|
538 }
|
|
539
|
|
540 #ifdef FEAT_MBYTE
|
|
541 # define ri_digit(c) (c < 0x100 && (class_tab[c] & RI_DIGIT))
|
|
542 # define ri_hex(c) (c < 0x100 && (class_tab[c] & RI_HEX))
|
|
543 # define ri_octal(c) (c < 0x100 && (class_tab[c] & RI_OCTAL))
|
|
544 # define ri_word(c) (c < 0x100 && (class_tab[c] & RI_WORD))
|
|
545 # define ri_head(c) (c < 0x100 && (class_tab[c] & RI_HEAD))
|
|
546 # define ri_alpha(c) (c < 0x100 && (class_tab[c] & RI_ALPHA))
|
|
547 # define ri_lower(c) (c < 0x100 && (class_tab[c] & RI_LOWER))
|
|
548 # define ri_upper(c) (c < 0x100 && (class_tab[c] & RI_UPPER))
|
|
549 # define ri_white(c) (c < 0x100 && (class_tab[c] & RI_WHITE))
|
|
550 #else
|
|
551 # define ri_digit(c) (class_tab[c] & RI_DIGIT)
|
|
552 # define ri_hex(c) (class_tab[c] & RI_HEX)
|
|
553 # define ri_octal(c) (class_tab[c] & RI_OCTAL)
|
|
554 # define ri_word(c) (class_tab[c] & RI_WORD)
|
|
555 # define ri_head(c) (class_tab[c] & RI_HEAD)
|
|
556 # define ri_alpha(c) (class_tab[c] & RI_ALPHA)
|
|
557 # define ri_lower(c) (class_tab[c] & RI_LOWER)
|
|
558 # define ri_upper(c) (class_tab[c] & RI_UPPER)
|
|
559 # define ri_white(c) (class_tab[c] & RI_WHITE)
|
|
560 #endif
|
|
561
|
|
562 /* flags for regflags */
|
|
563 #define RF_ICASE 1 /* ignore case */
|
|
564 #define RF_NOICASE 2 /* don't ignore case */
|
|
565 #define RF_HASNL 4 /* can match a NL */
|
|
566 #define RF_ICOMBINE 8 /* ignore combining characters */
|
|
567 #define RF_LOOKBH 16 /* uses "\@<=" or "\@<!" */
|
|
568
|
|
569 /*
|
|
570 * Global work variables for vim_regcomp().
|
|
571 */
|
|
572
|
|
573 static char_u *regparse; /* Input-scan pointer. */
|
|
574 static int prevchr_len; /* byte length of previous char */
|
|
575 static int num_complex_braces; /* Complex \{...} count */
|
|
576 static int regnpar; /* () count. */
|
|
577 #ifdef FEAT_SYN_HL
|
|
578 static int regnzpar; /* \z() count. */
|
|
579 static int re_has_z; /* \z item detected */
|
|
580 #endif
|
|
581 static char_u *regcode; /* Code-emit pointer, or JUST_CALC_SIZE */
|
|
582 static long regsize; /* Code size. */
|
|
583 static char_u had_endbrace[NSUBEXP]; /* flags, TRUE if end of () found */
|
|
584 static unsigned regflags; /* RF_ flags for prog */
|
|
585 static long brace_min[10]; /* Minimums for complex brace repeats */
|
|
586 static long brace_max[10]; /* Maximums for complex brace repeats */
|
|
587 static int brace_count[10]; /* Current counts for complex brace repeats */
|
|
588 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
589 static int had_eol; /* TRUE when EOL found by vim_regcomp() */
|
|
590 #endif
|
|
591 static int one_exactly = FALSE; /* only do one char for EXACTLY */
|
|
592
|
|
593 static int reg_magic; /* magicness of the pattern: */
|
|
594 #define MAGIC_NONE 1 /* "\V" very unmagic */
|
|
595 #define MAGIC_OFF 2 /* "\M" or 'magic' off */
|
|
596 #define MAGIC_ON 3 /* "\m" or 'magic' */
|
|
597 #define MAGIC_ALL 4 /* "\v" very magic */
|
|
598
|
|
599 static int reg_string; /* matching with a string instead of a buffer
|
|
600 line */
|
|
601
|
|
602 /*
|
|
603 * META contains all characters that may be magic, except '^' and '$'.
|
|
604 */
|
|
605
|
|
606 #ifdef EBCDIC
|
|
607 static char_u META[] = "%&()*+.123456789<=>?@ACDFHIKLMOPSUVWX[_acdfhiklmnopsuvwxz{|~";
|
|
608 #else
|
|
609 /* META[] is used often enough to justify turning it into a table. */
|
|
610 static char_u META_flags[] = {
|
|
611 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
612 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
613 /* % & ( ) * + . */
|
|
614 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0,
|
|
615 /* 1 2 3 4 5 6 7 8 9 < = > ? */
|
|
616 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1,
|
|
617 /* @ A C D F H I K L M O */
|
|
618 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1,
|
|
619 /* P S U V W X Z [ _ */
|
|
620 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1,
|
|
621 /* a c d f h i k l m n o */
|
|
622 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1,
|
|
623 /* p s u v w x z { | ~ */
|
|
624 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1
|
|
625 };
|
|
626 #endif
|
|
627
|
|
628 static int curchr;
|
|
629
|
|
630 /* arguments for reg() */
|
|
631 #define REG_NOPAREN 0 /* toplevel reg() */
|
|
632 #define REG_PAREN 1 /* \(\) */
|
|
633 #define REG_ZPAREN 2 /* \z(\) */
|
|
634 #define REG_NPAREN 3 /* \%(\) */
|
|
635
|
|
636 /*
|
|
637 * Forward declarations for vim_regcomp()'s friends.
|
|
638 */
|
|
639 static void initchr __ARGS((char_u *));
|
|
640 static int getchr __ARGS((void));
|
|
641 static void skipchr_keepstart __ARGS((void));
|
|
642 static int peekchr __ARGS((void));
|
|
643 static void skipchr __ARGS((void));
|
|
644 static void ungetchr __ARGS((void));
|
24
|
645 static int gethexchrs __ARGS((int maxinputlen));
|
|
646 static int getoctchrs __ARGS((void));
|
|
647 static int getdecchrs __ARGS((void));
|
|
648 static int coll_get_char __ARGS((void));
|
7
|
649 static void regcomp_start __ARGS((char_u *expr, int flags));
|
|
650 static char_u *reg __ARGS((int, int *));
|
|
651 static char_u *regbranch __ARGS((int *flagp));
|
|
652 static char_u *regconcat __ARGS((int *flagp));
|
|
653 static char_u *regpiece __ARGS((int *));
|
|
654 static char_u *regatom __ARGS((int *));
|
|
655 static char_u *regnode __ARGS((int));
|
|
656 static int prog_magic_wrong __ARGS((void));
|
|
657 static char_u *regnext __ARGS((char_u *));
|
|
658 static void regc __ARGS((int b));
|
|
659 #ifdef FEAT_MBYTE
|
|
660 static void regmbc __ARGS((int c));
|
167
|
661 #else
|
|
662 # define regmbc(c) regc(c)
|
7
|
663 #endif
|
|
664 static void reginsert __ARGS((int, char_u *));
|
|
665 static void reginsert_limits __ARGS((int, long, long, char_u *));
|
|
666 static char_u *re_put_long __ARGS((char_u *pr, long_u val));
|
|
667 static int read_limits __ARGS((long *, long *));
|
|
668 static void regtail __ARGS((char_u *, char_u *));
|
|
669 static void regoptail __ARGS((char_u *, char_u *));
|
|
670
|
|
671 /*
|
|
672 * Return TRUE if compiled regular expression "prog" can match a line break.
|
|
673 */
|
|
674 int
|
|
675 re_multiline(prog)
|
|
676 regprog_T *prog;
|
|
677 {
|
|
678 return (prog->regflags & RF_HASNL);
|
|
679 }
|
|
680
|
|
681 /*
|
|
682 * Return TRUE if compiled regular expression "prog" looks before the start
|
|
683 * position (pattern contains "\@<=" or "\@<!").
|
|
684 */
|
|
685 int
|
|
686 re_lookbehind(prog)
|
|
687 regprog_T *prog;
|
|
688 {
|
|
689 return (prog->regflags & RF_LOOKBH);
|
|
690 }
|
|
691
|
|
692 /*
|
167
|
693 * Check for an equivalence class name "[=a=]". "pp" points to the '['.
|
|
694 * Returns a character representing the class. Zero means that no item was
|
|
695 * recognized. Otherwise "pp" is advanced to after the item.
|
|
696 */
|
|
697 static int
|
|
698 get_equi_class(pp)
|
|
699 char_u **pp;
|
|
700 {
|
|
701 int c;
|
|
702 int l = 1;
|
|
703 char_u *p = *pp;
|
|
704
|
|
705 if (p[1] == '=')
|
|
706 {
|
|
707 #ifdef FEAT_MBYTE
|
|
708 if (has_mbyte)
|
|
709 l = mb_ptr2len_check(p + 2);
|
|
710 #endif
|
|
711 if (p[l + 2] == '=' && p[l + 3] == ']')
|
|
712 {
|
|
713 #ifdef FEAT_MBYTE
|
|
714 if (has_mbyte)
|
|
715 c = mb_ptr2char(p + 2);
|
|
716 else
|
|
717 #endif
|
|
718 c = p[2];
|
|
719 *pp += l + 4;
|
|
720 return c;
|
|
721 }
|
|
722 }
|
|
723 return 0;
|
|
724 }
|
|
725
|
|
726 /*
|
|
727 * Produce the bytes for equivalence class "c".
|
|
728 * Currently only handles latin1, latin9 and utf-8.
|
|
729 */
|
|
730 static void
|
|
731 reg_equi_class(c)
|
|
732 int c;
|
|
733 {
|
|
734 #ifdef FEAT_MBYTE
|
|
735 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0
|
|
736 || STRCMP(p_enc, "latin9") == 0)
|
|
737 #endif
|
|
738 {
|
|
739 switch (c)
|
|
740 {
|
|
741 case 'A': case 'À': case 'Á': case 'Â':
|
|
742 case 'Ã': case 'Ä': case 'Å':
|
|
743 regmbc('A'); regmbc('À'); regmbc('Á'); regmbc('Â');
|
|
744 regmbc('Ã'); regmbc('Ä'); regmbc('Å');
|
|
745 return;
|
|
746 case 'C': case 'Ç':
|
|
747 regmbc('C'); regmbc('Ç');
|
|
748 return;
|
|
749 case 'E': case 'È': case 'É': case 'Ê': case 'Ë':
|
|
750 regmbc('E'); regmbc('È'); regmbc('É'); regmbc('Ê');
|
|
751 regmbc('Ë');
|
|
752 return;
|
|
753 case 'I': case 'Ì': case 'Í': case 'Î': case 'Ï':
|
|
754 regmbc('I'); regmbc('Ì'); regmbc('Í'); regmbc('Î');
|
|
755 regmbc('Ï');
|
|
756 return;
|
|
757 case 'N': case 'Ñ':
|
|
758 regmbc('N'); regmbc('Ñ');
|
|
759 return;
|
|
760 case 'O': case 'Ò': case 'Ó': case 'Ô': case 'Õ': case 'Ö':
|
|
761 regmbc('O'); regmbc('Ò'); regmbc('Ó'); regmbc('Ô');
|
|
762 regmbc('Õ'); regmbc('Ö');
|
|
763 return;
|
|
764 case 'U': case 'Ù': case 'Ú': case 'Û': case 'Ü':
|
|
765 regmbc('U'); regmbc('Ù'); regmbc('Ú'); regmbc('Û');
|
|
766 regmbc('Ü');
|
|
767 return;
|
|
768 case 'Y': case 'Ý':
|
|
769 regmbc('Y'); regmbc('Ý');
|
|
770 return;
|
|
771 case 'a': case 'à': case 'á': case 'â':
|
|
772 case 'ã': case 'ä': case 'å':
|
|
773 regmbc('a'); regmbc('à'); regmbc('á'); regmbc('â');
|
|
774 regmbc('ã'); regmbc('ä'); regmbc('å');
|
|
775 return;
|
|
776 case 'c': case 'ç':
|
|
777 regmbc('c'); regmbc('ç');
|
|
778 return;
|
|
779 case 'e': case 'è': case 'é': case 'ê': case 'ë':
|
|
780 regmbc('e'); regmbc('è'); regmbc('é'); regmbc('ê');
|
|
781 regmbc('ë');
|
|
782 return;
|
|
783 case 'i': case 'ì': case 'í': case 'î': case 'ï':
|
|
784 regmbc('i'); regmbc('ì'); regmbc('í'); regmbc('î');
|
|
785 regmbc('ï');
|
|
786 return;
|
|
787 case 'n': case 'ñ':
|
|
788 regmbc('n'); regmbc('ñ');
|
|
789 return;
|
|
790 case 'o': case 'ò': case 'ó': case 'ô': case 'õ': case 'ö':
|
|
791 regmbc('o'); regmbc('ò'); regmbc('ó'); regmbc('ô');
|
|
792 regmbc('õ'); regmbc('ö');
|
|
793 return;
|
|
794 case 'u': case 'ù': case 'ú': case 'û': case 'ü':
|
|
795 regmbc('u'); regmbc('ù'); regmbc('ú'); regmbc('û');
|
|
796 regmbc('ü');
|
|
797 return;
|
|
798 case 'y': case 'ý': case 'ÿ':
|
|
799 regmbc('y'); regmbc('ý'); regmbc('ÿ');
|
|
800 return;
|
|
801 }
|
|
802 }
|
|
803 regmbc(c);
|
|
804 }
|
|
805
|
|
806 /*
|
|
807 * Check for a collating element "[.a.]". "pp" points to the '['.
|
|
808 * Returns a character. Zero means that no item was recognized. Otherwise
|
|
809 * "pp" is advanced to after the item.
|
|
810 * Currently only single characters are recognized!
|
|
811 */
|
|
812 static int
|
|
813 get_coll_element(pp)
|
|
814 char_u **pp;
|
|
815 {
|
|
816 int c;
|
|
817 int l = 1;
|
|
818 char_u *p = *pp;
|
|
819
|
|
820 if (p[1] == '.')
|
|
821 {
|
|
822 #ifdef FEAT_MBYTE
|
|
823 if (has_mbyte)
|
|
824 l = mb_ptr2len_check(p + 2);
|
|
825 #endif
|
|
826 if (p[l + 2] == '.' && p[l + 3] == ']')
|
|
827 {
|
|
828 #ifdef FEAT_MBYTE
|
|
829 if (has_mbyte)
|
|
830 c = mb_ptr2char(p + 2);
|
|
831 else
|
|
832 #endif
|
|
833 c = p[2];
|
|
834 *pp += l + 4;
|
|
835 return c;
|
|
836 }
|
|
837 }
|
|
838 return 0;
|
|
839 }
|
|
840
|
|
841
|
|
842 /*
|
|
843 * Skip over a "[]" range.
|
|
844 * "p" must point to the character after the '['.
|
|
845 * The returned pointer is on the matching ']', or the terminating NUL.
|
|
846 */
|
|
847 static char_u *
|
|
848 skip_anyof(p)
|
|
849 char_u *p;
|
|
850 {
|
|
851 int cpo_lit; /* 'cpoptions' contains 'l' flag */
|
|
852 int cpo_bsl; /* 'cpoptions' contains '\' flag */
|
|
853 #ifdef FEAT_MBYTE
|
|
854 int l;
|
|
855 #endif
|
|
856
|
|
857 cpo_lit = (!reg_syn && vim_strchr(p_cpo, CPO_LITERAL) != NULL);
|
|
858 cpo_bsl = (!reg_syn && vim_strchr(p_cpo, CPO_BACKSL) != NULL);
|
|
859
|
|
860 if (*p == '^') /* Complement of range. */
|
|
861 ++p;
|
|
862 if (*p == ']' || *p == '-')
|
|
863 ++p;
|
|
864 while (*p != NUL && *p != ']')
|
|
865 {
|
|
866 #ifdef FEAT_MBYTE
|
|
867 if (has_mbyte && (l = (*mb_ptr2len_check)(p)) > 1)
|
|
868 p += l;
|
|
869 else
|
|
870 #endif
|
|
871 if (*p == '-')
|
|
872 {
|
|
873 ++p;
|
|
874 if (*p != ']' && *p != NUL)
|
|
875 mb_ptr_adv(p);
|
|
876 }
|
|
877 else if (*p == '\\'
|
|
878 && !cpo_bsl
|
|
879 && (vim_strchr(REGEXP_INRANGE, p[1]) != NULL
|
|
880 || (!cpo_lit && vim_strchr(REGEXP_ABBR, p[1]) != NULL)))
|
|
881 p += 2;
|
|
882 else if (*p == '[')
|
|
883 {
|
|
884 if (get_char_class(&p) == CLASS_NONE
|
|
885 && get_equi_class(&p) == 0
|
|
886 && get_coll_element(&p) == 0)
|
|
887 ++p; /* It was not a class name */
|
|
888 }
|
|
889 else
|
|
890 ++p;
|
|
891 }
|
|
892
|
|
893 return p;
|
|
894 }
|
|
895
|
|
896 /*
|
7
|
897 * Skip past regular expression.
|
153
|
898 * Stop at end of "startp" or where "dirc" is found ('/', '?', etc).
|
7
|
899 * Take care of characters with a backslash in front of it.
|
|
900 * Skip strings inside [ and ].
|
|
901 * When "newp" is not NULL and "dirc" is '?', make an allocated copy of the
|
|
902 * expression and change "\?" to "?". If "*newp" is not NULL the expression
|
|
903 * is changed in-place.
|
|
904 */
|
|
905 char_u *
|
|
906 skip_regexp(startp, dirc, magic, newp)
|
|
907 char_u *startp;
|
|
908 int dirc;
|
|
909 int magic;
|
|
910 char_u **newp;
|
|
911 {
|
|
912 int mymagic;
|
|
913 char_u *p = startp;
|
|
914
|
|
915 if (magic)
|
|
916 mymagic = MAGIC_ON;
|
|
917 else
|
|
918 mymagic = MAGIC_OFF;
|
|
919
|
39
|
920 for (; p[0] != NUL; mb_ptr_adv(p))
|
7
|
921 {
|
|
922 if (p[0] == dirc) /* found end of regexp */
|
|
923 break;
|
|
924 if ((p[0] == '[' && mymagic >= MAGIC_ON)
|
|
925 || (p[0] == '\\' && p[1] == '[' && mymagic <= MAGIC_OFF))
|
|
926 {
|
|
927 p = skip_anyof(p + 1);
|
|
928 if (p[0] == NUL)
|
|
929 break;
|
|
930 }
|
|
931 else if (p[0] == '\\' && p[1] != NUL)
|
|
932 {
|
|
933 if (dirc == '?' && newp != NULL && p[1] == '?')
|
|
934 {
|
|
935 /* change "\?" to "?", make a copy first. */
|
|
936 if (*newp == NULL)
|
|
937 {
|
|
938 *newp = vim_strsave(startp);
|
|
939 if (*newp != NULL)
|
|
940 p = *newp + (p - startp);
|
|
941 }
|
|
942 if (*newp != NULL)
|
|
943 mch_memmove(p, p + 1, STRLEN(p));
|
|
944 else
|
|
945 ++p;
|
|
946 }
|
|
947 else
|
|
948 ++p; /* skip next character */
|
|
949 if (*p == 'v')
|
|
950 mymagic = MAGIC_ALL;
|
|
951 else if (*p == 'V')
|
|
952 mymagic = MAGIC_NONE;
|
|
953 }
|
|
954 }
|
|
955 return p;
|
|
956 }
|
|
957
|
|
958 /*
|
41
|
959 * vim_regcomp() - compile a regular expression into internal code
|
|
960 * Returns the program in allocated space. Returns NULL for an error.
|
7
|
961 *
|
|
962 * We can't allocate space until we know how big the compiled form will be,
|
|
963 * but we can't compile it (and thus know how big it is) until we've got a
|
|
964 * place to put the code. So we cheat: we compile it twice, once with code
|
|
965 * generation turned off and size counting turned on, and once "for real".
|
|
966 * This also means that we don't allocate space until we are sure that the
|
|
967 * thing really will compile successfully, and we never have to move the
|
|
968 * code and thus invalidate pointers into it. (Note that it has to be in
|
|
969 * one piece because vim_free() must be able to free it all.)
|
|
970 *
|
|
971 * Whether upper/lower case is to be ignored is decided when executing the
|
|
972 * program, it does not matter here.
|
|
973 *
|
|
974 * Beware that the optimization-preparation code in here knows about some
|
|
975 * of the structure of the compiled regexp.
|
|
976 * "re_flags": RE_MAGIC and/or RE_STRING.
|
|
977 */
|
|
978 regprog_T *
|
|
979 vim_regcomp(expr, re_flags)
|
|
980 char_u *expr;
|
|
981 int re_flags;
|
|
982 {
|
|
983 regprog_T *r;
|
|
984 char_u *scan;
|
|
985 char_u *longest;
|
|
986 int len;
|
|
987 int flags;
|
|
988
|
|
989 if (expr == NULL)
|
|
990 EMSG_RET_NULL(_(e_null));
|
|
991
|
|
992 init_class_tab();
|
|
993
|
|
994 /*
|
|
995 * First pass: determine size, legality.
|
|
996 */
|
|
997 regcomp_start(expr, re_flags);
|
|
998 regcode = JUST_CALC_SIZE;
|
|
999 regc(REGMAGIC);
|
|
1000 if (reg(REG_NOPAREN, &flags) == NULL)
|
|
1001 return NULL;
|
|
1002
|
|
1003 /* Small enough for pointer-storage convention? */
|
|
1004 #ifdef SMALL_MALLOC /* 16 bit storage allocation */
|
|
1005 if (regsize >= 65536L - 256L)
|
|
1006 EMSG_RET_NULL(_("E339: Pattern too long"));
|
|
1007 #endif
|
|
1008
|
|
1009 /* Allocate space. */
|
|
1010 r = (regprog_T *)lalloc(sizeof(regprog_T) + regsize, TRUE);
|
|
1011 if (r == NULL)
|
|
1012 return NULL;
|
|
1013
|
|
1014 /*
|
|
1015 * Second pass: emit code.
|
|
1016 */
|
|
1017 regcomp_start(expr, re_flags);
|
|
1018 regcode = r->program;
|
|
1019 regc(REGMAGIC);
|
|
1020 if (reg(REG_NOPAREN, &flags) == NULL)
|
|
1021 {
|
|
1022 vim_free(r);
|
|
1023 return NULL;
|
|
1024 }
|
|
1025
|
|
1026 /* Dig out information for optimizations. */
|
|
1027 r->regstart = NUL; /* Worst-case defaults. */
|
|
1028 r->reganch = 0;
|
|
1029 r->regmust = NULL;
|
|
1030 r->regmlen = 0;
|
|
1031 r->regflags = regflags;
|
|
1032 if (flags & HASNL)
|
|
1033 r->regflags |= RF_HASNL;
|
|
1034 if (flags & HASLOOKBH)
|
|
1035 r->regflags |= RF_LOOKBH;
|
|
1036 #ifdef FEAT_SYN_HL
|
|
1037 /* Remember whether this pattern has any \z specials in it. */
|
|
1038 r->reghasz = re_has_z;
|
|
1039 #endif
|
|
1040 scan = r->program + 1; /* First BRANCH. */
|
|
1041 if (OP(regnext(scan)) == END) /* Only one top-level choice. */
|
|
1042 {
|
|
1043 scan = OPERAND(scan);
|
|
1044
|
|
1045 /* Starting-point info. */
|
|
1046 if (OP(scan) == BOL || OP(scan) == RE_BOF)
|
|
1047 {
|
|
1048 r->reganch++;
|
|
1049 scan = regnext(scan);
|
|
1050 }
|
|
1051
|
|
1052 if (OP(scan) == EXACTLY)
|
|
1053 {
|
|
1054 #ifdef FEAT_MBYTE
|
|
1055 if (has_mbyte)
|
|
1056 r->regstart = (*mb_ptr2char)(OPERAND(scan));
|
|
1057 else
|
|
1058 #endif
|
|
1059 r->regstart = *OPERAND(scan);
|
|
1060 }
|
|
1061 else if ((OP(scan) == BOW
|
|
1062 || OP(scan) == EOW
|
|
1063 || OP(scan) == NOTHING
|
|
1064 || OP(scan) == MOPEN + 0 || OP(scan) == NOPEN
|
|
1065 || OP(scan) == MCLOSE + 0 || OP(scan) == NCLOSE)
|
|
1066 && OP(regnext(scan)) == EXACTLY)
|
|
1067 {
|
|
1068 #ifdef FEAT_MBYTE
|
|
1069 if (has_mbyte)
|
|
1070 r->regstart = (*mb_ptr2char)(OPERAND(regnext(scan)));
|
|
1071 else
|
|
1072 #endif
|
|
1073 r->regstart = *OPERAND(regnext(scan));
|
|
1074 }
|
|
1075
|
|
1076 /*
|
|
1077 * If there's something expensive in the r.e., find the longest
|
|
1078 * literal string that must appear and make it the regmust. Resolve
|
|
1079 * ties in favor of later strings, since the regstart check works
|
|
1080 * with the beginning of the r.e. and avoiding duplication
|
|
1081 * strengthens checking. Not a strong reason, but sufficient in the
|
|
1082 * absence of others.
|
|
1083 */
|
|
1084 /*
|
|
1085 * When the r.e. starts with BOW, it is faster to look for a regmust
|
|
1086 * first. Used a lot for "#" and "*" commands. (Added by mool).
|
|
1087 */
|
|
1088 if ((flags & SPSTART || OP(scan) == BOW || OP(scan) == EOW)
|
|
1089 && !(flags & HASNL))
|
|
1090 {
|
|
1091 longest = NULL;
|
|
1092 len = 0;
|
|
1093 for (; scan != NULL; scan = regnext(scan))
|
|
1094 if (OP(scan) == EXACTLY && STRLEN(OPERAND(scan)) >= (size_t)len)
|
|
1095 {
|
|
1096 longest = OPERAND(scan);
|
|
1097 len = (int)STRLEN(OPERAND(scan));
|
|
1098 }
|
|
1099 r->regmust = longest;
|
|
1100 r->regmlen = len;
|
|
1101 }
|
|
1102 }
|
|
1103 #ifdef DEBUG
|
|
1104 regdump(expr, r);
|
|
1105 #endif
|
|
1106 return r;
|
|
1107 }
|
|
1108
|
|
1109 /*
|
|
1110 * Setup to parse the regexp. Used once to get the length and once to do it.
|
|
1111 */
|
|
1112 static void
|
|
1113 regcomp_start(expr, re_flags)
|
|
1114 char_u *expr;
|
|
1115 int re_flags; /* see vim_regcomp() */
|
|
1116 {
|
|
1117 initchr(expr);
|
|
1118 if (re_flags & RE_MAGIC)
|
|
1119 reg_magic = MAGIC_ON;
|
|
1120 else
|
|
1121 reg_magic = MAGIC_OFF;
|
|
1122 reg_string = (re_flags & RE_STRING);
|
|
1123
|
|
1124 num_complex_braces = 0;
|
|
1125 regnpar = 1;
|
|
1126 vim_memset(had_endbrace, 0, sizeof(had_endbrace));
|
|
1127 #ifdef FEAT_SYN_HL
|
|
1128 regnzpar = 1;
|
|
1129 re_has_z = 0;
|
|
1130 #endif
|
|
1131 regsize = 0L;
|
|
1132 regflags = 0;
|
|
1133 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
1134 had_eol = FALSE;
|
|
1135 #endif
|
|
1136 }
|
|
1137
|
|
1138 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
1139 /*
|
|
1140 * Check if during the previous call to vim_regcomp the EOL item "$" has been
|
|
1141 * found. This is messy, but it works fine.
|
|
1142 */
|
|
1143 int
|
|
1144 vim_regcomp_had_eol()
|
|
1145 {
|
|
1146 return had_eol;
|
|
1147 }
|
|
1148 #endif
|
|
1149
|
|
1150 /*
|
|
1151 * reg - regular expression, i.e. main body or parenthesized thing
|
|
1152 *
|
|
1153 * Caller must absorb opening parenthesis.
|
|
1154 *
|
|
1155 * Combining parenthesis handling with the base level of regular expression
|
|
1156 * is a trifle forced, but the need to tie the tails of the branches to what
|
|
1157 * follows makes it hard to avoid.
|
|
1158 */
|
|
1159 static char_u *
|
|
1160 reg(paren, flagp)
|
|
1161 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */
|
|
1162 int *flagp;
|
|
1163 {
|
|
1164 char_u *ret;
|
|
1165 char_u *br;
|
|
1166 char_u *ender;
|
|
1167 int parno = 0;
|
|
1168 int flags;
|
|
1169
|
|
1170 *flagp = HASWIDTH; /* Tentatively. */
|
|
1171
|
|
1172 #ifdef FEAT_SYN_HL
|
|
1173 if (paren == REG_ZPAREN)
|
|
1174 {
|
|
1175 /* Make a ZOPEN node. */
|
|
1176 if (regnzpar >= NSUBEXP)
|
|
1177 EMSG_RET_NULL(_("E50: Too many \\z("));
|
|
1178 parno = regnzpar;
|
|
1179 regnzpar++;
|
|
1180 ret = regnode(ZOPEN + parno);
|
|
1181 }
|
|
1182 else
|
|
1183 #endif
|
|
1184 if (paren == REG_PAREN)
|
|
1185 {
|
|
1186 /* Make a MOPEN node. */
|
|
1187 if (regnpar >= NSUBEXP)
|
|
1188 EMSG_M_RET_NULL(_("E51: Too many %s("), reg_magic == MAGIC_ALL);
|
|
1189 parno = regnpar;
|
|
1190 ++regnpar;
|
|
1191 ret = regnode(MOPEN + parno);
|
|
1192 }
|
|
1193 else if (paren == REG_NPAREN)
|
|
1194 {
|
|
1195 /* Make a NOPEN node. */
|
|
1196 ret = regnode(NOPEN);
|
|
1197 }
|
|
1198 else
|
|
1199 ret = NULL;
|
|
1200
|
|
1201 /* Pick up the branches, linking them together. */
|
|
1202 br = regbranch(&flags);
|
|
1203 if (br == NULL)
|
|
1204 return NULL;
|
|
1205 if (ret != NULL)
|
|
1206 regtail(ret, br); /* [MZ]OPEN -> first. */
|
|
1207 else
|
|
1208 ret = br;
|
|
1209 /* If one of the branches can be zero-width, the whole thing can.
|
|
1210 * If one of the branches has * at start or matches a line-break, the
|
|
1211 * whole thing can. */
|
|
1212 if (!(flags & HASWIDTH))
|
|
1213 *flagp &= ~HASWIDTH;
|
|
1214 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH);
|
|
1215 while (peekchr() == Magic('|'))
|
|
1216 {
|
|
1217 skipchr();
|
|
1218 br = regbranch(&flags);
|
|
1219 if (br == NULL)
|
|
1220 return NULL;
|
|
1221 regtail(ret, br); /* BRANCH -> BRANCH. */
|
|
1222 if (!(flags & HASWIDTH))
|
|
1223 *flagp &= ~HASWIDTH;
|
|
1224 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH);
|
|
1225 }
|
|
1226
|
|
1227 /* Make a closing node, and hook it on the end. */
|
|
1228 ender = regnode(
|
|
1229 #ifdef FEAT_SYN_HL
|
|
1230 paren == REG_ZPAREN ? ZCLOSE + parno :
|
|
1231 #endif
|
|
1232 paren == REG_PAREN ? MCLOSE + parno :
|
|
1233 paren == REG_NPAREN ? NCLOSE : END);
|
|
1234 regtail(ret, ender);
|
|
1235
|
|
1236 /* Hook the tails of the branches to the closing node. */
|
|
1237 for (br = ret; br != NULL; br = regnext(br))
|
|
1238 regoptail(br, ender);
|
|
1239
|
|
1240 /* Check for proper termination. */
|
|
1241 if (paren != REG_NOPAREN && getchr() != Magic(')'))
|
|
1242 {
|
|
1243 #ifdef FEAT_SYN_HL
|
|
1244 if (paren == REG_ZPAREN)
|
|
1245 EMSG_RET_NULL(_("E52: Unmatched \\z("))
|
|
1246 else
|
|
1247 #endif
|
|
1248 if (paren == REG_NPAREN)
|
|
1249 EMSG_M_RET_NULL(_("E53: Unmatched %s%%("), reg_magic == MAGIC_ALL)
|
|
1250 else
|
|
1251 EMSG_M_RET_NULL(_("E54: Unmatched %s("), reg_magic == MAGIC_ALL)
|
|
1252 }
|
|
1253 else if (paren == REG_NOPAREN && peekchr() != NUL)
|
|
1254 {
|
|
1255 if (curchr == Magic(')'))
|
|
1256 EMSG_M_RET_NULL(_("E55: Unmatched %s)"), reg_magic == MAGIC_ALL)
|
|
1257 else
|
|
1258 EMSG_RET_NULL(_(e_trailing)) /* "Can't happen". */
|
|
1259 /* NOTREACHED */
|
|
1260 }
|
|
1261 /*
|
|
1262 * Here we set the flag allowing back references to this set of
|
|
1263 * parentheses.
|
|
1264 */
|
|
1265 if (paren == REG_PAREN)
|
|
1266 had_endbrace[parno] = TRUE; /* have seen the close paren */
|
|
1267 return ret;
|
|
1268 }
|
|
1269
|
|
1270 /*
|
|
1271 * regbranch - one alternative of an | operator
|
|
1272 *
|
|
1273 * Implements the & operator.
|
|
1274 */
|
|
1275 static char_u *
|
|
1276 regbranch(flagp)
|
|
1277 int *flagp;
|
|
1278 {
|
|
1279 char_u *ret;
|
|
1280 char_u *chain = NULL;
|
|
1281 char_u *latest;
|
|
1282 int flags;
|
|
1283
|
|
1284 *flagp = WORST | HASNL; /* Tentatively. */
|
|
1285
|
|
1286 ret = regnode(BRANCH);
|
|
1287 for (;;)
|
|
1288 {
|
|
1289 latest = regconcat(&flags);
|
|
1290 if (latest == NULL)
|
|
1291 return NULL;
|
|
1292 /* If one of the branches has width, the whole thing has. If one of
|
|
1293 * the branches anchors at start-of-line, the whole thing does.
|
|
1294 * If one of the branches uses look-behind, the whole thing does. */
|
|
1295 *flagp |= flags & (HASWIDTH | SPSTART | HASLOOKBH);
|
|
1296 /* If one of the branches doesn't match a line-break, the whole thing
|
|
1297 * doesn't. */
|
|
1298 *flagp &= ~HASNL | (flags & HASNL);
|
|
1299 if (chain != NULL)
|
|
1300 regtail(chain, latest);
|
|
1301 if (peekchr() != Magic('&'))
|
|
1302 break;
|
|
1303 skipchr();
|
|
1304 regtail(latest, regnode(END)); /* operand ends */
|
|
1305 reginsert(MATCH, latest);
|
|
1306 chain = latest;
|
|
1307 }
|
|
1308
|
|
1309 return ret;
|
|
1310 }
|
|
1311
|
|
1312 /*
|
|
1313 * regbranch - one alternative of an | or & operator
|
|
1314 *
|
|
1315 * Implements the concatenation operator.
|
|
1316 */
|
|
1317 static char_u *
|
|
1318 regconcat(flagp)
|
|
1319 int *flagp;
|
|
1320 {
|
|
1321 char_u *first = NULL;
|
|
1322 char_u *chain = NULL;
|
|
1323 char_u *latest;
|
|
1324 int flags;
|
|
1325 int cont = TRUE;
|
|
1326
|
|
1327 *flagp = WORST; /* Tentatively. */
|
|
1328
|
|
1329 while (cont)
|
|
1330 {
|
|
1331 switch (peekchr())
|
|
1332 {
|
|
1333 case NUL:
|
|
1334 case Magic('|'):
|
|
1335 case Magic('&'):
|
|
1336 case Magic(')'):
|
|
1337 cont = FALSE;
|
|
1338 break;
|
|
1339 case Magic('Z'):
|
|
1340 #ifdef FEAT_MBYTE
|
|
1341 regflags |= RF_ICOMBINE;
|
|
1342 #endif
|
|
1343 skipchr_keepstart();
|
|
1344 break;
|
|
1345 case Magic('c'):
|
|
1346 regflags |= RF_ICASE;
|
|
1347 skipchr_keepstart();
|
|
1348 break;
|
|
1349 case Magic('C'):
|
|
1350 regflags |= RF_NOICASE;
|
|
1351 skipchr_keepstart();
|
|
1352 break;
|
|
1353 case Magic('v'):
|
|
1354 reg_magic = MAGIC_ALL;
|
|
1355 skipchr_keepstart();
|
|
1356 curchr = -1;
|
|
1357 break;
|
|
1358 case Magic('m'):
|
|
1359 reg_magic = MAGIC_ON;
|
|
1360 skipchr_keepstart();
|
|
1361 curchr = -1;
|
|
1362 break;
|
|
1363 case Magic('M'):
|
|
1364 reg_magic = MAGIC_OFF;
|
|
1365 skipchr_keepstart();
|
|
1366 curchr = -1;
|
|
1367 break;
|
|
1368 case Magic('V'):
|
|
1369 reg_magic = MAGIC_NONE;
|
|
1370 skipchr_keepstart();
|
|
1371 curchr = -1;
|
|
1372 break;
|
|
1373 default:
|
|
1374 latest = regpiece(&flags);
|
|
1375 if (latest == NULL)
|
|
1376 return NULL;
|
|
1377 *flagp |= flags & (HASWIDTH | HASNL | HASLOOKBH);
|
|
1378 if (chain == NULL) /* First piece. */
|
|
1379 *flagp |= flags & SPSTART;
|
|
1380 else
|
|
1381 regtail(chain, latest);
|
|
1382 chain = latest;
|
|
1383 if (first == NULL)
|
|
1384 first = latest;
|
|
1385 break;
|
|
1386 }
|
|
1387 }
|
|
1388 if (first == NULL) /* Loop ran zero times. */
|
|
1389 first = regnode(NOTHING);
|
|
1390 return first;
|
|
1391 }
|
|
1392
|
|
1393 /*
|
|
1394 * regpiece - something followed by possible [*+=]
|
|
1395 *
|
|
1396 * Note that the branching code sequences used for = and the general cases
|
|
1397 * of * and + are somewhat optimized: they use the same NOTHING node as
|
|
1398 * both the endmarker for their branch list and the body of the last branch.
|
|
1399 * It might seem that this node could be dispensed with entirely, but the
|
|
1400 * endmarker role is not redundant.
|
|
1401 */
|
|
1402 static char_u *
|
|
1403 regpiece(flagp)
|
|
1404 int *flagp;
|
|
1405 {
|
|
1406 char_u *ret;
|
|
1407 int op;
|
|
1408 char_u *next;
|
|
1409 int flags;
|
|
1410 long minval;
|
|
1411 long maxval;
|
|
1412
|
|
1413 ret = regatom(&flags);
|
|
1414 if (ret == NULL)
|
|
1415 return NULL;
|
|
1416
|
|
1417 op = peekchr();
|
|
1418 if (re_multi_type(op) == NOT_MULTI)
|
|
1419 {
|
|
1420 *flagp = flags;
|
|
1421 return ret;
|
|
1422 }
|
|
1423 /* default flags */
|
|
1424 *flagp = (WORST | SPSTART | (flags & (HASNL | HASLOOKBH)));
|
|
1425
|
|
1426 skipchr();
|
|
1427 switch (op)
|
|
1428 {
|
|
1429 case Magic('*'):
|
|
1430 if (flags & SIMPLE)
|
|
1431 reginsert(STAR, ret);
|
|
1432 else
|
|
1433 {
|
|
1434 /* Emit x* as (x&|), where & means "self". */
|
|
1435 reginsert(BRANCH, ret); /* Either x */
|
|
1436 regoptail(ret, regnode(BACK)); /* and loop */
|
|
1437 regoptail(ret, ret); /* back */
|
|
1438 regtail(ret, regnode(BRANCH)); /* or */
|
|
1439 regtail(ret, regnode(NOTHING)); /* null. */
|
|
1440 }
|
|
1441 break;
|
|
1442
|
|
1443 case Magic('+'):
|
|
1444 if (flags & SIMPLE)
|
|
1445 reginsert(PLUS, ret);
|
|
1446 else
|
|
1447 {
|
|
1448 /* Emit x+ as x(&|), where & means "self". */
|
|
1449 next = regnode(BRANCH); /* Either */
|
|
1450 regtail(ret, next);
|
|
1451 regtail(regnode(BACK), ret); /* loop back */
|
|
1452 regtail(next, regnode(BRANCH)); /* or */
|
|
1453 regtail(ret, regnode(NOTHING)); /* null. */
|
|
1454 }
|
|
1455 *flagp = (WORST | HASWIDTH | (flags & (HASNL | HASLOOKBH)));
|
|
1456 break;
|
|
1457
|
|
1458 case Magic('@'):
|
|
1459 {
|
|
1460 int lop = END;
|
|
1461
|
|
1462 switch (no_Magic(getchr()))
|
|
1463 {
|
|
1464 case '=': lop = MATCH; break; /* \@= */
|
|
1465 case '!': lop = NOMATCH; break; /* \@! */
|
|
1466 case '>': lop = SUBPAT; break; /* \@> */
|
|
1467 case '<': switch (no_Magic(getchr()))
|
|
1468 {
|
|
1469 case '=': lop = BEHIND; break; /* \@<= */
|
|
1470 case '!': lop = NOBEHIND; break; /* \@<! */
|
|
1471 }
|
|
1472 }
|
|
1473 if (lop == END)
|
|
1474 EMSG_M_RET_NULL(_("E59: invalid character after %s@"),
|
|
1475 reg_magic == MAGIC_ALL);
|
|
1476 /* Look behind must match with behind_pos. */
|
|
1477 if (lop == BEHIND || lop == NOBEHIND)
|
|
1478 {
|
|
1479 regtail(ret, regnode(BHPOS));
|
|
1480 *flagp |= HASLOOKBH;
|
|
1481 }
|
|
1482 regtail(ret, regnode(END)); /* operand ends */
|
|
1483 reginsert(lop, ret);
|
|
1484 break;
|
|
1485 }
|
|
1486
|
|
1487 case Magic('?'):
|
|
1488 case Magic('='):
|
|
1489 /* Emit x= as (x|) */
|
|
1490 reginsert(BRANCH, ret); /* Either x */
|
|
1491 regtail(ret, regnode(BRANCH)); /* or */
|
|
1492 next = regnode(NOTHING); /* null. */
|
|
1493 regtail(ret, next);
|
|
1494 regoptail(ret, next);
|
|
1495 break;
|
|
1496
|
|
1497 case Magic('{'):
|
|
1498 if (!read_limits(&minval, &maxval))
|
|
1499 return NULL;
|
|
1500 if (flags & SIMPLE)
|
|
1501 {
|
|
1502 reginsert(BRACE_SIMPLE, ret);
|
|
1503 reginsert_limits(BRACE_LIMITS, minval, maxval, ret);
|
|
1504 }
|
|
1505 else
|
|
1506 {
|
|
1507 if (num_complex_braces >= 10)
|
|
1508 EMSG_M_RET_NULL(_("E60: Too many complex %s{...}s"),
|
|
1509 reg_magic == MAGIC_ALL);
|
|
1510 reginsert(BRACE_COMPLEX + num_complex_braces, ret);
|
|
1511 regoptail(ret, regnode(BACK));
|
|
1512 regoptail(ret, ret);
|
|
1513 reginsert_limits(BRACE_LIMITS, minval, maxval, ret);
|
|
1514 ++num_complex_braces;
|
|
1515 }
|
|
1516 if (minval > 0 && maxval > 0)
|
|
1517 *flagp = (HASWIDTH | (flags & (HASNL | HASLOOKBH)));
|
|
1518 break;
|
|
1519 }
|
|
1520 if (re_multi_type(peekchr()) != NOT_MULTI)
|
|
1521 {
|
|
1522 /* Can't have a multi follow a multi. */
|
|
1523 if (peekchr() == Magic('*'))
|
|
1524 sprintf((char *)IObuff, _("E61: Nested %s*"),
|
|
1525 reg_magic >= MAGIC_ON ? "" : "\\");
|
|
1526 else
|
|
1527 sprintf((char *)IObuff, _("E62: Nested %s%c"),
|
|
1528 reg_magic == MAGIC_ALL ? "" : "\\", no_Magic(peekchr()));
|
|
1529 EMSG_RET_NULL(IObuff);
|
|
1530 }
|
|
1531
|
|
1532 return ret;
|
|
1533 }
|
|
1534
|
|
1535 /*
|
|
1536 * regatom - the lowest level
|
|
1537 *
|
|
1538 * Optimization: gobbles an entire sequence of ordinary characters so that
|
|
1539 * it can turn them into a single node, which is smaller to store and
|
|
1540 * faster to run. Don't do this when one_exactly is set.
|
|
1541 */
|
|
1542 static char_u *
|
|
1543 regatom(flagp)
|
|
1544 int *flagp;
|
|
1545 {
|
|
1546 char_u *ret;
|
|
1547 int flags;
|
|
1548 int cpo_lit; /* 'cpoptions' contains 'l' flag */
|
167
|
1549 int cpo_bsl; /* 'cpoptions' contains '\' flag */
|
7
|
1550 int c;
|
|
1551 static char_u *classchars = (char_u *)".iIkKfFpPsSdDxXoOwWhHaAlLuU";
|
|
1552 static int classcodes[] = {ANY, IDENT, SIDENT, KWORD, SKWORD,
|
|
1553 FNAME, SFNAME, PRINT, SPRINT,
|
|
1554 WHITE, NWHITE, DIGIT, NDIGIT,
|
|
1555 HEX, NHEX, OCTAL, NOCTAL,
|
|
1556 WORD, NWORD, HEAD, NHEAD,
|
|
1557 ALPHA, NALPHA, LOWER, NLOWER,
|
|
1558 UPPER, NUPPER
|
|
1559 };
|
|
1560 char_u *p;
|
|
1561 int extra = 0;
|
|
1562
|
|
1563 *flagp = WORST; /* Tentatively. */
|
|
1564 cpo_lit = (!reg_syn && vim_strchr(p_cpo, CPO_LITERAL) != NULL);
|
167
|
1565 cpo_bsl = (!reg_syn && vim_strchr(p_cpo, CPO_BACKSL) != NULL);
|
7
|
1566
|
|
1567 c = getchr();
|
|
1568 switch (c)
|
|
1569 {
|
|
1570 case Magic('^'):
|
|
1571 ret = regnode(BOL);
|
|
1572 break;
|
|
1573
|
|
1574 case Magic('$'):
|
|
1575 ret = regnode(EOL);
|
|
1576 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
1577 had_eol = TRUE;
|
|
1578 #endif
|
|
1579 break;
|
|
1580
|
|
1581 case Magic('<'):
|
|
1582 ret = regnode(BOW);
|
|
1583 break;
|
|
1584
|
|
1585 case Magic('>'):
|
|
1586 ret = regnode(EOW);
|
|
1587 break;
|
|
1588
|
|
1589 case Magic('_'):
|
|
1590 c = no_Magic(getchr());
|
|
1591 if (c == '^') /* "\_^" is start-of-line */
|
|
1592 {
|
|
1593 ret = regnode(BOL);
|
|
1594 break;
|
|
1595 }
|
|
1596 if (c == '$') /* "\_$" is end-of-line */
|
|
1597 {
|
|
1598 ret = regnode(EOL);
|
|
1599 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
1600 had_eol = TRUE;
|
|
1601 #endif
|
|
1602 break;
|
|
1603 }
|
|
1604
|
|
1605 extra = ADD_NL;
|
|
1606 *flagp |= HASNL;
|
|
1607
|
|
1608 /* "\_[" is character range plus newline */
|
|
1609 if (c == '[')
|
|
1610 goto collection;
|
|
1611
|
|
1612 /* "\_x" is character class plus newline */
|
|
1613 /*FALLTHROUGH*/
|
|
1614
|
|
1615 /*
|
|
1616 * Character classes.
|
|
1617 */
|
|
1618 case Magic('.'):
|
|
1619 case Magic('i'):
|
|
1620 case Magic('I'):
|
|
1621 case Magic('k'):
|
|
1622 case Magic('K'):
|
|
1623 case Magic('f'):
|
|
1624 case Magic('F'):
|
|
1625 case Magic('p'):
|
|
1626 case Magic('P'):
|
|
1627 case Magic('s'):
|
|
1628 case Magic('S'):
|
|
1629 case Magic('d'):
|
|
1630 case Magic('D'):
|
|
1631 case Magic('x'):
|
|
1632 case Magic('X'):
|
|
1633 case Magic('o'):
|
|
1634 case Magic('O'):
|
|
1635 case Magic('w'):
|
|
1636 case Magic('W'):
|
|
1637 case Magic('h'):
|
|
1638 case Magic('H'):
|
|
1639 case Magic('a'):
|
|
1640 case Magic('A'):
|
|
1641 case Magic('l'):
|
|
1642 case Magic('L'):
|
|
1643 case Magic('u'):
|
|
1644 case Magic('U'):
|
|
1645 p = vim_strchr(classchars, no_Magic(c));
|
|
1646 if (p == NULL)
|
|
1647 EMSG_RET_NULL(_("E63: invalid use of \\_"));
|
|
1648 ret = regnode(classcodes[p - classchars] + extra);
|
|
1649 *flagp |= HASWIDTH | SIMPLE;
|
|
1650 break;
|
|
1651
|
|
1652 case Magic('n'):
|
|
1653 if (reg_string)
|
|
1654 {
|
|
1655 /* In a string "\n" matches a newline character. */
|
|
1656 ret = regnode(EXACTLY);
|
|
1657 regc(NL);
|
|
1658 regc(NUL);
|
|
1659 *flagp |= HASWIDTH | SIMPLE;
|
|
1660 }
|
|
1661 else
|
|
1662 {
|
|
1663 /* In buffer text "\n" matches the end of a line. */
|
|
1664 ret = regnode(NEWL);
|
|
1665 *flagp |= HASWIDTH | HASNL;
|
|
1666 }
|
|
1667 break;
|
|
1668
|
|
1669 case Magic('('):
|
|
1670 if (one_exactly)
|
|
1671 EMSG_ONE_RET_NULL;
|
|
1672 ret = reg(REG_PAREN, &flags);
|
|
1673 if (ret == NULL)
|
|
1674 return NULL;
|
|
1675 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH);
|
|
1676 break;
|
|
1677
|
|
1678 case NUL:
|
|
1679 case Magic('|'):
|
|
1680 case Magic('&'):
|
|
1681 case Magic(')'):
|
|
1682 EMSG_RET_NULL(_(e_internal)); /* Supposed to be caught earlier. */
|
|
1683 /* NOTREACHED */
|
|
1684
|
|
1685 case Magic('='):
|
|
1686 case Magic('?'):
|
|
1687 case Magic('+'):
|
|
1688 case Magic('@'):
|
|
1689 case Magic('{'):
|
|
1690 case Magic('*'):
|
|
1691 c = no_Magic(c);
|
|
1692 sprintf((char *)IObuff, _("E64: %s%c follows nothing"),
|
|
1693 (c == '*' ? reg_magic >= MAGIC_ON : reg_magic == MAGIC_ALL)
|
|
1694 ? "" : "\\", c);
|
|
1695 EMSG_RET_NULL(IObuff);
|
|
1696 /* NOTREACHED */
|
|
1697
|
|
1698 case Magic('~'): /* previous substitute pattern */
|
|
1699 if (reg_prev_sub)
|
|
1700 {
|
|
1701 char_u *lp;
|
|
1702
|
|
1703 ret = regnode(EXACTLY);
|
|
1704 lp = reg_prev_sub;
|
|
1705 while (*lp != NUL)
|
|
1706 regc(*lp++);
|
|
1707 regc(NUL);
|
|
1708 if (*reg_prev_sub != NUL)
|
|
1709 {
|
|
1710 *flagp |= HASWIDTH;
|
|
1711 if ((lp - reg_prev_sub) == 1)
|
|
1712 *flagp |= SIMPLE;
|
|
1713 }
|
|
1714 }
|
|
1715 else
|
|
1716 EMSG_RET_NULL(_(e_nopresub));
|
|
1717 break;
|
|
1718
|
|
1719 case Magic('1'):
|
|
1720 case Magic('2'):
|
|
1721 case Magic('3'):
|
|
1722 case Magic('4'):
|
|
1723 case Magic('5'):
|
|
1724 case Magic('6'):
|
|
1725 case Magic('7'):
|
|
1726 case Magic('8'):
|
|
1727 case Magic('9'):
|
|
1728 {
|
|
1729 int refnum;
|
|
1730
|
|
1731 refnum = c - Magic('0');
|
|
1732 /*
|
|
1733 * Check if the back reference is legal. We must have seen the
|
|
1734 * close brace.
|
|
1735 * TODO: Should also check that we don't refer to something
|
|
1736 * that is repeated (+*=): what instance of the repetition
|
|
1737 * should we match?
|
|
1738 */
|
|
1739 if (!had_endbrace[refnum])
|
|
1740 {
|
|
1741 /* Trick: check if "@<=" or "@<!" follows, in which case
|
|
1742 * the \1 can appear before the referenced match. */
|
|
1743 for (p = regparse; *p != NUL; ++p)
|
|
1744 if (p[0] == '@' && p[1] == '<'
|
|
1745 && (p[2] == '!' || p[2] == '='))
|
|
1746 break;
|
|
1747 if (*p == NUL)
|
|
1748 EMSG_RET_NULL(_("E65: Illegal back reference"));
|
|
1749 }
|
|
1750 ret = regnode(BACKREF + refnum);
|
|
1751 }
|
|
1752 break;
|
|
1753
|
|
1754 #ifdef FEAT_SYN_HL
|
|
1755 case Magic('z'):
|
|
1756 {
|
|
1757 c = no_Magic(getchr());
|
|
1758 switch (c)
|
|
1759 {
|
|
1760 case '(': if (reg_do_extmatch != REX_SET)
|
|
1761 EMSG_RET_NULL(_("E66: \\z( not allowed here"));
|
|
1762 if (one_exactly)
|
|
1763 EMSG_ONE_RET_NULL;
|
|
1764 ret = reg(REG_ZPAREN, &flags);
|
|
1765 if (ret == NULL)
|
|
1766 return NULL;
|
|
1767 *flagp |= flags & (HASWIDTH|SPSTART|HASNL|HASLOOKBH);
|
|
1768 re_has_z = REX_SET;
|
|
1769 break;
|
|
1770
|
|
1771 case '1':
|
|
1772 case '2':
|
|
1773 case '3':
|
|
1774 case '4':
|
|
1775 case '5':
|
|
1776 case '6':
|
|
1777 case '7':
|
|
1778 case '8':
|
|
1779 case '9': if (reg_do_extmatch != REX_USE)
|
|
1780 EMSG_RET_NULL(_("E67: \\z1 et al. not allowed here"));
|
|
1781 ret = regnode(ZREF + c - '0');
|
|
1782 re_has_z = REX_USE;
|
|
1783 break;
|
|
1784
|
|
1785 case 's': ret = regnode(MOPEN + 0);
|
|
1786 break;
|
|
1787
|
|
1788 case 'e': ret = regnode(MCLOSE + 0);
|
|
1789 break;
|
|
1790
|
|
1791 default: EMSG_RET_NULL(_("E68: Invalid character after \\z"));
|
|
1792 }
|
|
1793 }
|
|
1794 break;
|
|
1795 #endif
|
|
1796
|
|
1797 case Magic('%'):
|
|
1798 {
|
|
1799 c = no_Magic(getchr());
|
|
1800 switch (c)
|
|
1801 {
|
|
1802 /* () without a back reference */
|
|
1803 case '(':
|
|
1804 if (one_exactly)
|
|
1805 EMSG_ONE_RET_NULL;
|
|
1806 ret = reg(REG_NPAREN, &flags);
|
|
1807 if (ret == NULL)
|
|
1808 return NULL;
|
|
1809 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH);
|
|
1810 break;
|
|
1811
|
|
1812 /* Catch \%^ and \%$ regardless of where they appear in the
|
|
1813 * pattern -- regardless of whether or not it makes sense. */
|
|
1814 case '^':
|
|
1815 ret = regnode(RE_BOF);
|
|
1816 break;
|
|
1817
|
|
1818 case '$':
|
|
1819 ret = regnode(RE_EOF);
|
|
1820 break;
|
|
1821
|
|
1822 case '#':
|
|
1823 ret = regnode(CURSOR);
|
|
1824 break;
|
|
1825
|
|
1826 /* \%[abc]: Emit as a list of branches, all ending at the last
|
|
1827 * branch which matches nothing. */
|
|
1828 case '[':
|
|
1829 if (one_exactly) /* doesn't nest */
|
|
1830 EMSG_ONE_RET_NULL;
|
|
1831 {
|
|
1832 char_u *lastbranch;
|
|
1833 char_u *lastnode = NULL;
|
|
1834 char_u *br;
|
|
1835
|
|
1836 ret = NULL;
|
|
1837 while ((c = getchr()) != ']')
|
|
1838 {
|
|
1839 if (c == NUL)
|
|
1840 EMSG_M_RET_NULL(_("E69: Missing ] after %s%%["),
|
|
1841 reg_magic == MAGIC_ALL);
|
|
1842 br = regnode(BRANCH);
|
|
1843 if (ret == NULL)
|
|
1844 ret = br;
|
|
1845 else
|
|
1846 regtail(lastnode, br);
|
|
1847
|
|
1848 ungetchr();
|
|
1849 one_exactly = TRUE;
|
|
1850 lastnode = regatom(flagp);
|
|
1851 one_exactly = FALSE;
|
|
1852 if (lastnode == NULL)
|
|
1853 return NULL;
|
|
1854 }
|
|
1855 if (ret == NULL)
|
|
1856 EMSG_M_RET_NULL(_("E70: Empty %s%%[]"),
|
|
1857 reg_magic == MAGIC_ALL);
|
|
1858 lastbranch = regnode(BRANCH);
|
|
1859 br = regnode(NOTHING);
|
|
1860 if (ret != JUST_CALC_SIZE)
|
|
1861 {
|
|
1862 regtail(lastnode, br);
|
|
1863 regtail(lastbranch, br);
|
|
1864 /* connect all branches to the NOTHING
|
|
1865 * branch at the end */
|
|
1866 for (br = ret; br != lastnode; )
|
|
1867 {
|
|
1868 if (OP(br) == BRANCH)
|
|
1869 {
|
|
1870 regtail(br, lastbranch);
|
|
1871 br = OPERAND(br);
|
|
1872 }
|
|
1873 else
|
|
1874 br = regnext(br);
|
|
1875 }
|
|
1876 }
|
|
1877 *flagp &= ~HASWIDTH;
|
|
1878 break;
|
|
1879 }
|
|
1880
|
24
|
1881 case 'd': /* %d123 decimal */
|
|
1882 case 'o': /* %o123 octal */
|
|
1883 case 'x': /* %xab hex 2 */
|
|
1884 case 'u': /* %uabcd hex 4 */
|
|
1885 case 'U': /* %U1234abcd hex 8 */
|
|
1886 {
|
|
1887 int i;
|
|
1888
|
|
1889 switch (c)
|
|
1890 {
|
|
1891 case 'd': i = getdecchrs(); break;
|
|
1892 case 'o': i = getoctchrs(); break;
|
|
1893 case 'x': i = gethexchrs(2); break;
|
|
1894 case 'u': i = gethexchrs(4); break;
|
|
1895 case 'U': i = gethexchrs(8); break;
|
|
1896 default: i = -1; break;
|
|
1897 }
|
|
1898
|
|
1899 if (i < 0)
|
|
1900 EMSG_M_RET_NULL(
|
|
1901 _("E678: Invalid character after %s%%[dxouU]"),
|
|
1902 reg_magic == MAGIC_ALL);
|
|
1903 ret = regnode(EXACTLY);
|
|
1904 if (i == 0)
|
|
1905 regc(0x0a);
|
|
1906 else
|
|
1907 #ifdef FEAT_MBYTE
|
|
1908 regmbc(i);
|
|
1909 #else
|
|
1910 regc(i);
|
|
1911 #endif
|
|
1912 regc(NUL);
|
|
1913 *flagp |= HASWIDTH;
|
|
1914 break;
|
|
1915 }
|
|
1916
|
7
|
1917 default:
|
|
1918 if (VIM_ISDIGIT(c) || c == '<' || c == '>')
|
|
1919 {
|
|
1920 long_u n = 0;
|
|
1921 int cmp;
|
|
1922
|
|
1923 cmp = c;
|
|
1924 if (cmp == '<' || cmp == '>')
|
|
1925 c = getchr();
|
|
1926 while (VIM_ISDIGIT(c))
|
|
1927 {
|
|
1928 n = n * 10 + (c - '0');
|
|
1929 c = getchr();
|
|
1930 }
|
|
1931 if (c == 'l' || c == 'c' || c == 'v')
|
|
1932 {
|
|
1933 if (c == 'l')
|
|
1934 ret = regnode(RE_LNUM);
|
|
1935 else if (c == 'c')
|
|
1936 ret = regnode(RE_COL);
|
|
1937 else
|
|
1938 ret = regnode(RE_VCOL);
|
|
1939 if (ret == JUST_CALC_SIZE)
|
|
1940 regsize += 5;
|
|
1941 else
|
|
1942 {
|
|
1943 /* put the number and the optional
|
|
1944 * comparator after the opcode */
|
|
1945 regcode = re_put_long(regcode, n);
|
|
1946 *regcode++ = cmp;
|
|
1947 }
|
|
1948 break;
|
|
1949 }
|
|
1950 }
|
|
1951
|
|
1952 EMSG_M_RET_NULL(_("E71: Invalid character after %s%%"),
|
|
1953 reg_magic == MAGIC_ALL);
|
|
1954 }
|
|
1955 }
|
|
1956 break;
|
|
1957
|
|
1958 case Magic('['):
|
|
1959 collection:
|
|
1960 {
|
|
1961 char_u *lp;
|
|
1962
|
|
1963 /*
|
|
1964 * If there is no matching ']', we assume the '[' is a normal
|
|
1965 * character. This makes 'incsearch' and ":help [" work.
|
|
1966 */
|
|
1967 lp = skip_anyof(regparse);
|
|
1968 if (*lp == ']') /* there is a matching ']' */
|
|
1969 {
|
|
1970 int startc = -1; /* > 0 when next '-' is a range */
|
|
1971 int endc;
|
|
1972
|
|
1973 /*
|
|
1974 * In a character class, different parsing rules apply.
|
|
1975 * Not even \ is special anymore, nothing is.
|
|
1976 */
|
|
1977 if (*regparse == '^') /* Complement of range. */
|
|
1978 {
|
|
1979 ret = regnode(ANYBUT + extra);
|
|
1980 regparse++;
|
|
1981 }
|
|
1982 else
|
|
1983 ret = regnode(ANYOF + extra);
|
|
1984
|
|
1985 /* At the start ']' and '-' mean the literal character. */
|
|
1986 if (*regparse == ']' || *regparse == '-')
|
167
|
1987 {
|
|
1988 startc = *regparse;
|
7
|
1989 regc(*regparse++);
|
167
|
1990 }
|
7
|
1991
|
|
1992 while (*regparse != NUL && *regparse != ']')
|
|
1993 {
|
|
1994 if (*regparse == '-')
|
|
1995 {
|
|
1996 ++regparse;
|
|
1997 /* The '-' is not used for a range at the end and
|
|
1998 * after or before a '\n'. */
|
|
1999 if (*regparse == ']' || *regparse == NUL
|
|
2000 || startc == -1
|
|
2001 || (regparse[0] == '\\' && regparse[1] == 'n'))
|
|
2002 {
|
|
2003 regc('-');
|
|
2004 startc = '-'; /* [--x] is a range */
|
|
2005 }
|
|
2006 else
|
|
2007 {
|
167
|
2008 /* Also accept "a-[.z.]" */
|
|
2009 endc = 0;
|
|
2010 if (*regparse == '[')
|
|
2011 endc = get_coll_element(®parse);
|
|
2012 if (endc == 0)
|
|
2013 {
|
7
|
2014 #ifdef FEAT_MBYTE
|
167
|
2015 if (has_mbyte)
|
|
2016 endc = mb_ptr2char_adv(®parse);
|
|
2017 else
|
7
|
2018 #endif
|
167
|
2019 endc = *regparse++;
|
|
2020 }
|
24
|
2021
|
|
2022 /* Handle \o40, \x20 and \u20AC style sequences */
|
167
|
2023 if (endc == '\\' && !cpo_lit && !cpo_bsl)
|
24
|
2024 endc = coll_get_char();
|
|
2025
|
7
|
2026 if (startc > endc)
|
|
2027 EMSG_RET_NULL(_(e_invrange));
|
|
2028 #ifdef FEAT_MBYTE
|
|
2029 if (has_mbyte && ((*mb_char2len)(startc) > 1
|
|
2030 || (*mb_char2len)(endc) > 1))
|
|
2031 {
|
|
2032 /* Limit to a range of 256 chars */
|
|
2033 if (endc > startc + 256)
|
|
2034 EMSG_RET_NULL(_(e_invrange));
|
|
2035 while (++startc <= endc)
|
|
2036 regmbc(startc);
|
|
2037 }
|
|
2038 else
|
|
2039 #endif
|
|
2040 {
|
|
2041 #ifdef EBCDIC
|
|
2042 int alpha_only = FALSE;
|
|
2043
|
|
2044 /* for alphabetical range skip the gaps
|
|
2045 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */
|
|
2046 if (isalpha(startc) && isalpha(endc))
|
|
2047 alpha_only = TRUE;
|
|
2048 #endif
|
|
2049 while (++startc <= endc)
|
|
2050 #ifdef EBCDIC
|
|
2051 if (!alpha_only || isalpha(startc))
|
|
2052 #endif
|
|
2053 regc(startc);
|
|
2054 }
|
|
2055 startc = -1;
|
|
2056 }
|
|
2057 }
|
|
2058 /*
|
|
2059 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim
|
|
2060 * accepts "\t", "\e", etc., but only when the 'l' flag in
|
|
2061 * 'cpoptions' is not included.
|
167
|
2062 * Posix doesn't recognize backslash at all.
|
7
|
2063 */
|
|
2064 else if (*regparse == '\\'
|
167
|
2065 && !cpo_bsl
|
7
|
2066 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL
|
|
2067 || (!cpo_lit
|
|
2068 && vim_strchr(REGEXP_ABBR,
|
|
2069 regparse[1]) != NULL)))
|
|
2070 {
|
|
2071 regparse++;
|
|
2072 if (*regparse == 'n')
|
|
2073 {
|
|
2074 /* '\n' in range: also match NL */
|
|
2075 if (ret != JUST_CALC_SIZE)
|
|
2076 {
|
|
2077 if (*ret == ANYBUT)
|
|
2078 *ret = ANYBUT + ADD_NL;
|
|
2079 else if (*ret == ANYOF)
|
|
2080 *ret = ANYOF + ADD_NL;
|
|
2081 /* else: must have had a \n already */
|
|
2082 }
|
|
2083 *flagp |= HASNL;
|
|
2084 regparse++;
|
|
2085 startc = -1;
|
|
2086 }
|
24
|
2087 else if (*regparse == 'd'
|
|
2088 || *regparse == 'o'
|
|
2089 || *regparse == 'x'
|
|
2090 || *regparse == 'u'
|
|
2091 || *regparse == 'U')
|
|
2092 {
|
|
2093 startc = coll_get_char();
|
|
2094 if (startc == 0)
|
|
2095 regc(0x0a);
|
|
2096 else
|
|
2097 #ifdef FEAT_MBYTE
|
|
2098 regmbc(startc);
|
|
2099 #else
|
|
2100 regc(startc);
|
|
2101 #endif
|
|
2102 }
|
7
|
2103 else
|
|
2104 {
|
|
2105 startc = backslash_trans(*regparse++);
|
|
2106 regc(startc);
|
|
2107 }
|
|
2108 }
|
|
2109 else if (*regparse == '[')
|
|
2110 {
|
|
2111 int c_class;
|
|
2112 int cu;
|
|
2113
|
167
|
2114 c_class = get_char_class(®parse);
|
7
|
2115 startc = -1;
|
|
2116 /* Characters assumed to be 8 bits! */
|
|
2117 switch (c_class)
|
|
2118 {
|
|
2119 case CLASS_NONE:
|
167
|
2120 c_class = get_equi_class(®parse);
|
|
2121 if (c_class != 0)
|
|
2122 {
|
|
2123 /* produce equivalence class */
|
|
2124 reg_equi_class(c_class);
|
|
2125 }
|
|
2126 else if ((c_class =
|
|
2127 get_coll_element(®parse)) != 0)
|
|
2128 {
|
|
2129 /* produce a collating element */
|
|
2130 regmbc(c_class);
|
|
2131 }
|
|
2132 else
|
|
2133 {
|
|
2134 /* literal '[', allow [[-x] as a range */
|
|
2135 startc = *regparse++;
|
|
2136 regc(startc);
|
|
2137 }
|
7
|
2138 break;
|
|
2139 case CLASS_ALNUM:
|
|
2140 for (cu = 1; cu <= 255; cu++)
|
|
2141 if (isalnum(cu))
|
|
2142 regc(cu);
|
|
2143 break;
|
|
2144 case CLASS_ALPHA:
|
|
2145 for (cu = 1; cu <= 255; cu++)
|
|
2146 if (isalpha(cu))
|
|
2147 regc(cu);
|
|
2148 break;
|
|
2149 case CLASS_BLANK:
|
|
2150 regc(' ');
|
|
2151 regc('\t');
|
|
2152 break;
|
|
2153 case CLASS_CNTRL:
|
|
2154 for (cu = 1; cu <= 255; cu++)
|
|
2155 if (iscntrl(cu))
|
|
2156 regc(cu);
|
|
2157 break;
|
|
2158 case CLASS_DIGIT:
|
|
2159 for (cu = 1; cu <= 255; cu++)
|
|
2160 if (VIM_ISDIGIT(cu))
|
|
2161 regc(cu);
|
|
2162 break;
|
|
2163 case CLASS_GRAPH:
|
|
2164 for (cu = 1; cu <= 255; cu++)
|
|
2165 if (isgraph(cu))
|
|
2166 regc(cu);
|
|
2167 break;
|
|
2168 case CLASS_LOWER:
|
|
2169 for (cu = 1; cu <= 255; cu++)
|
|
2170 if (islower(cu))
|
|
2171 regc(cu);
|
|
2172 break;
|
|
2173 case CLASS_PRINT:
|
|
2174 for (cu = 1; cu <= 255; cu++)
|
|
2175 if (vim_isprintc(cu))
|
|
2176 regc(cu);
|
|
2177 break;
|
|
2178 case CLASS_PUNCT:
|
|
2179 for (cu = 1; cu <= 255; cu++)
|
|
2180 if (ispunct(cu))
|
|
2181 regc(cu);
|
|
2182 break;
|
|
2183 case CLASS_SPACE:
|
|
2184 for (cu = 9; cu <= 13; cu++)
|
|
2185 regc(cu);
|
|
2186 regc(' ');
|
|
2187 break;
|
|
2188 case CLASS_UPPER:
|
|
2189 for (cu = 1; cu <= 255; cu++)
|
|
2190 if (isupper(cu))
|
|
2191 regc(cu);
|
|
2192 break;
|
|
2193 case CLASS_XDIGIT:
|
|
2194 for (cu = 1; cu <= 255; cu++)
|
|
2195 if (vim_isxdigit(cu))
|
|
2196 regc(cu);
|
|
2197 break;
|
|
2198 case CLASS_TAB:
|
|
2199 regc('\t');
|
|
2200 break;
|
|
2201 case CLASS_RETURN:
|
|
2202 regc('\r');
|
|
2203 break;
|
|
2204 case CLASS_BACKSPACE:
|
|
2205 regc('\b');
|
|
2206 break;
|
|
2207 case CLASS_ESCAPE:
|
|
2208 regc('\033');
|
|
2209 break;
|
|
2210 }
|
|
2211 }
|
|
2212 else
|
|
2213 {
|
|
2214 #ifdef FEAT_MBYTE
|
|
2215 if (has_mbyte)
|
|
2216 {
|
|
2217 int len;
|
|
2218
|
|
2219 /* produce a multibyte character, including any
|
|
2220 * following composing characters */
|
|
2221 startc = mb_ptr2char(regparse);
|
|
2222 len = (*mb_ptr2len_check)(regparse);
|
|
2223 if (enc_utf8 && utf_char2len(startc) != len)
|
|
2224 startc = -1; /* composing chars */
|
|
2225 while (--len >= 0)
|
|
2226 regc(*regparse++);
|
|
2227 }
|
|
2228 else
|
|
2229 #endif
|
|
2230 {
|
|
2231 startc = *regparse++;
|
|
2232 regc(startc);
|
|
2233 }
|
|
2234 }
|
|
2235 }
|
|
2236 regc(NUL);
|
|
2237 prevchr_len = 1; /* last char was the ']' */
|
|
2238 if (*regparse != ']')
|
|
2239 EMSG_RET_NULL(_(e_toomsbra)); /* Cannot happen? */
|
|
2240 skipchr(); /* let's be friends with the lexer again */
|
|
2241 *flagp |= HASWIDTH | SIMPLE;
|
|
2242 break;
|
|
2243 }
|
|
2244 }
|
|
2245 /* FALLTHROUGH */
|
|
2246
|
|
2247 default:
|
|
2248 {
|
|
2249 int len;
|
|
2250
|
|
2251 #ifdef FEAT_MBYTE
|
|
2252 /* A multi-byte character is handled as a separate atom if it's
|
|
2253 * before a multi. */
|
|
2254 if (has_mbyte && (*mb_char2len)(c) > 1
|
|
2255 && re_multi_type(peekchr()) != NOT_MULTI)
|
|
2256 {
|
|
2257 ret = regnode(MULTIBYTECODE);
|
|
2258 regmbc(c);
|
|
2259 *flagp |= HASWIDTH | SIMPLE;
|
|
2260 break;
|
|
2261 }
|
|
2262 #endif
|
|
2263
|
|
2264 ret = regnode(EXACTLY);
|
|
2265
|
|
2266 /*
|
|
2267 * Append characters as long as:
|
|
2268 * - there is no following multi, we then need the character in
|
|
2269 * front of it as a single character operand
|
|
2270 * - not running into a Magic character
|
|
2271 * - "one_exactly" is not set
|
|
2272 * But always emit at least one character. Might be a Multi,
|
|
2273 * e.g., a "[" without matching "]".
|
|
2274 */
|
|
2275 for (len = 0; c != NUL && (len == 0
|
|
2276 || (re_multi_type(peekchr()) == NOT_MULTI
|
|
2277 && !one_exactly
|
|
2278 && !is_Magic(c))); ++len)
|
|
2279 {
|
|
2280 c = no_Magic(c);
|
|
2281 #ifdef FEAT_MBYTE
|
|
2282 if (has_mbyte)
|
|
2283 {
|
|
2284 regmbc(c);
|
|
2285 if (enc_utf8)
|
|
2286 {
|
|
2287 int off;
|
|
2288 int l;
|
|
2289
|
|
2290 /* Need to get composing character too, directly
|
|
2291 * access regparse for that, because skipchr() skips
|
|
2292 * over composing chars. */
|
|
2293 ungetchr();
|
|
2294 if (*regparse == '\\' && regparse[1] != NUL)
|
|
2295 off = 1;
|
|
2296 else
|
|
2297 off = 0;
|
|
2298 for (;;)
|
|
2299 {
|
|
2300 l = utf_ptr2len_check(regparse + off);
|
|
2301 if (!UTF_COMPOSINGLIKE(regparse + off,
|
|
2302 regparse + off + l))
|
|
2303 break;
|
|
2304 off += l;
|
|
2305 regmbc(utf_ptr2char(regparse + off));
|
|
2306 }
|
|
2307 skipchr();
|
|
2308 }
|
|
2309 }
|
|
2310 else
|
|
2311 #endif
|
|
2312 regc(c);
|
|
2313 c = getchr();
|
|
2314 }
|
|
2315 ungetchr();
|
|
2316
|
|
2317 regc(NUL);
|
|
2318 *flagp |= HASWIDTH;
|
|
2319 if (len == 1)
|
|
2320 *flagp |= SIMPLE;
|
|
2321 }
|
|
2322 break;
|
|
2323 }
|
|
2324
|
|
2325 return ret;
|
|
2326 }
|
|
2327
|
|
2328 /*
|
|
2329 * emit a node
|
|
2330 * Return pointer to generated code.
|
|
2331 */
|
|
2332 static char_u *
|
|
2333 regnode(op)
|
|
2334 int op;
|
|
2335 {
|
|
2336 char_u *ret;
|
|
2337
|
|
2338 ret = regcode;
|
|
2339 if (ret == JUST_CALC_SIZE)
|
|
2340 regsize += 3;
|
|
2341 else
|
|
2342 {
|
|
2343 *regcode++ = op;
|
|
2344 *regcode++ = NUL; /* Null "next" pointer. */
|
|
2345 *regcode++ = NUL;
|
|
2346 }
|
|
2347 return ret;
|
|
2348 }
|
|
2349
|
|
2350 /*
|
|
2351 * Emit (if appropriate) a byte of code
|
|
2352 */
|
|
2353 static void
|
|
2354 regc(b)
|
|
2355 int b;
|
|
2356 {
|
|
2357 if (regcode == JUST_CALC_SIZE)
|
|
2358 regsize++;
|
|
2359 else
|
|
2360 *regcode++ = b;
|
|
2361 }
|
|
2362
|
|
2363 #ifdef FEAT_MBYTE
|
|
2364 /*
|
|
2365 * Emit (if appropriate) a multi-byte character of code
|
|
2366 */
|
|
2367 static void
|
|
2368 regmbc(c)
|
|
2369 int c;
|
|
2370 {
|
|
2371 if (regcode == JUST_CALC_SIZE)
|
|
2372 regsize += (*mb_char2len)(c);
|
|
2373 else
|
|
2374 regcode += (*mb_char2bytes)(c, regcode);
|
|
2375 }
|
|
2376 #endif
|
|
2377
|
|
2378 /*
|
|
2379 * reginsert - insert an operator in front of already-emitted operand
|
|
2380 *
|
|
2381 * Means relocating the operand.
|
|
2382 */
|
|
2383 static void
|
|
2384 reginsert(op, opnd)
|
|
2385 int op;
|
|
2386 char_u *opnd;
|
|
2387 {
|
|
2388 char_u *src;
|
|
2389 char_u *dst;
|
|
2390 char_u *place;
|
|
2391
|
|
2392 if (regcode == JUST_CALC_SIZE)
|
|
2393 {
|
|
2394 regsize += 3;
|
|
2395 return;
|
|
2396 }
|
|
2397 src = regcode;
|
|
2398 regcode += 3;
|
|
2399 dst = regcode;
|
|
2400 while (src > opnd)
|
|
2401 *--dst = *--src;
|
|
2402
|
|
2403 place = opnd; /* Op node, where operand used to be. */
|
|
2404 *place++ = op;
|
|
2405 *place++ = NUL;
|
|
2406 *place = NUL;
|
|
2407 }
|
|
2408
|
|
2409 /*
|
|
2410 * reginsert_limits - insert an operator in front of already-emitted operand.
|
|
2411 * The operator has the given limit values as operands. Also set next pointer.
|
|
2412 *
|
|
2413 * Means relocating the operand.
|
|
2414 */
|
|
2415 static void
|
|
2416 reginsert_limits(op, minval, maxval, opnd)
|
|
2417 int op;
|
|
2418 long minval;
|
|
2419 long maxval;
|
|
2420 char_u *opnd;
|
|
2421 {
|
|
2422 char_u *src;
|
|
2423 char_u *dst;
|
|
2424 char_u *place;
|
|
2425
|
|
2426 if (regcode == JUST_CALC_SIZE)
|
|
2427 {
|
|
2428 regsize += 11;
|
|
2429 return;
|
|
2430 }
|
|
2431 src = regcode;
|
|
2432 regcode += 11;
|
|
2433 dst = regcode;
|
|
2434 while (src > opnd)
|
|
2435 *--dst = *--src;
|
|
2436
|
|
2437 place = opnd; /* Op node, where operand used to be. */
|
|
2438 *place++ = op;
|
|
2439 *place++ = NUL;
|
|
2440 *place++ = NUL;
|
|
2441 place = re_put_long(place, (long_u)minval);
|
|
2442 place = re_put_long(place, (long_u)maxval);
|
|
2443 regtail(opnd, place);
|
|
2444 }
|
|
2445
|
|
2446 /*
|
|
2447 * Write a long as four bytes at "p" and return pointer to the next char.
|
|
2448 */
|
|
2449 static char_u *
|
|
2450 re_put_long(p, val)
|
|
2451 char_u *p;
|
|
2452 long_u val;
|
|
2453 {
|
|
2454 *p++ = (char_u) ((val >> 24) & 0377);
|
|
2455 *p++ = (char_u) ((val >> 16) & 0377);
|
|
2456 *p++ = (char_u) ((val >> 8) & 0377);
|
|
2457 *p++ = (char_u) (val & 0377);
|
|
2458 return p;
|
|
2459 }
|
|
2460
|
|
2461 /*
|
|
2462 * regtail - set the next-pointer at the end of a node chain
|
|
2463 */
|
|
2464 static void
|
|
2465 regtail(p, val)
|
|
2466 char_u *p;
|
|
2467 char_u *val;
|
|
2468 {
|
|
2469 char_u *scan;
|
|
2470 char_u *temp;
|
|
2471 int offset;
|
|
2472
|
|
2473 if (p == JUST_CALC_SIZE)
|
|
2474 return;
|
|
2475
|
|
2476 /* Find last node. */
|
|
2477 scan = p;
|
|
2478 for (;;)
|
|
2479 {
|
|
2480 temp = regnext(scan);
|
|
2481 if (temp == NULL)
|
|
2482 break;
|
|
2483 scan = temp;
|
|
2484 }
|
|
2485
|
|
2486 if (OP(scan) == BACK)
|
|
2487 offset = (int)(scan - val);
|
|
2488 else
|
|
2489 offset = (int)(val - scan);
|
|
2490 *(scan + 1) = (char_u) (((unsigned)offset >> 8) & 0377);
|
|
2491 *(scan + 2) = (char_u) (offset & 0377);
|
|
2492 }
|
|
2493
|
|
2494 /*
|
|
2495 * regoptail - regtail on item after a BRANCH; nop if none
|
|
2496 */
|
|
2497 static void
|
|
2498 regoptail(p, val)
|
|
2499 char_u *p;
|
|
2500 char_u *val;
|
|
2501 {
|
|
2502 /* When op is neither BRANCH nor BRACE_COMPLEX0-9, it is "operandless" */
|
|
2503 if (p == NULL || p == JUST_CALC_SIZE
|
|
2504 || (OP(p) != BRANCH
|
|
2505 && (OP(p) < BRACE_COMPLEX || OP(p) > BRACE_COMPLEX + 9)))
|
|
2506 return;
|
|
2507 regtail(OPERAND(p), val);
|
|
2508 }
|
|
2509
|
|
2510 /*
|
|
2511 * getchr() - get the next character from the pattern. We know about
|
|
2512 * magic and such, so therefore we need a lexical analyzer.
|
|
2513 */
|
|
2514
|
|
2515 /* static int curchr; */
|
|
2516 static int prevprevchr;
|
|
2517 static int prevchr;
|
|
2518 static int nextchr; /* used for ungetchr() */
|
|
2519 /*
|
|
2520 * Note: prevchr is sometimes -1 when we are not at the start,
|
|
2521 * eg in /[ ^I]^ the pattern was never found even if it existed, because ^ was
|
|
2522 * taken to be magic -- webb
|
|
2523 */
|
|
2524 static int at_start; /* True when on the first character */
|
|
2525 static int prev_at_start; /* True when on the second character */
|
|
2526
|
|
2527 static void
|
|
2528 initchr(str)
|
|
2529 char_u *str;
|
|
2530 {
|
|
2531 regparse = str;
|
|
2532 prevchr_len = 0;
|
|
2533 curchr = prevprevchr = prevchr = nextchr = -1;
|
|
2534 at_start = TRUE;
|
|
2535 prev_at_start = FALSE;
|
|
2536 }
|
|
2537
|
|
2538 static int
|
|
2539 peekchr()
|
|
2540 {
|
167
|
2541 static int after_slash = FALSE;
|
|
2542
|
7
|
2543 if (curchr == -1)
|
|
2544 {
|
|
2545 switch (curchr = regparse[0])
|
|
2546 {
|
|
2547 case '.':
|
|
2548 case '[':
|
|
2549 case '~':
|
|
2550 /* magic when 'magic' is on */
|
|
2551 if (reg_magic >= MAGIC_ON)
|
|
2552 curchr = Magic(curchr);
|
|
2553 break;
|
|
2554 case '(':
|
|
2555 case ')':
|
|
2556 case '{':
|
|
2557 case '%':
|
|
2558 case '+':
|
|
2559 case '=':
|
|
2560 case '?':
|
|
2561 case '@':
|
|
2562 case '!':
|
|
2563 case '&':
|
|
2564 case '|':
|
|
2565 case '<':
|
|
2566 case '>':
|
|
2567 case '#': /* future ext. */
|
|
2568 case '"': /* future ext. */
|
|
2569 case '\'': /* future ext. */
|
|
2570 case ',': /* future ext. */
|
|
2571 case '-': /* future ext. */
|
|
2572 case ':': /* future ext. */
|
|
2573 case ';': /* future ext. */
|
|
2574 case '`': /* future ext. */
|
|
2575 case '/': /* Can't be used in / command */
|
|
2576 /* magic only after "\v" */
|
|
2577 if (reg_magic == MAGIC_ALL)
|
|
2578 curchr = Magic(curchr);
|
|
2579 break;
|
|
2580 case '*':
|
167
|
2581 /* * is not magic as the very first character, eg "?*ptr", when
|
|
2582 * after '^', eg "/^*ptr" and when after "\(", "\|", "\&". But
|
|
2583 * "\(\*" is not magic, thus must be magic if "after_slash" */
|
|
2584 if (reg_magic >= MAGIC_ON
|
|
2585 && !at_start
|
|
2586 && !(prev_at_start && prevchr == Magic('^'))
|
|
2587 && (after_slash
|
|
2588 || (prevchr != Magic('(')
|
|
2589 && prevchr != Magic('&')
|
|
2590 && prevchr != Magic('|'))))
|
7
|
2591 curchr = Magic('*');
|
|
2592 break;
|
|
2593 case '^':
|
|
2594 /* '^' is only magic as the very first character and if it's after
|
|
2595 * "\(", "\|", "\&' or "\n" */
|
|
2596 if (reg_magic >= MAGIC_OFF
|
|
2597 && (at_start
|
|
2598 || reg_magic == MAGIC_ALL
|
|
2599 || prevchr == Magic('(')
|
|
2600 || prevchr == Magic('|')
|
|
2601 || prevchr == Magic('&')
|
|
2602 || prevchr == Magic('n')
|
|
2603 || (no_Magic(prevchr) == '('
|
|
2604 && prevprevchr == Magic('%'))))
|
|
2605 {
|
|
2606 curchr = Magic('^');
|
|
2607 at_start = TRUE;
|
|
2608 prev_at_start = FALSE;
|
|
2609 }
|
|
2610 break;
|
|
2611 case '$':
|
|
2612 /* '$' is only magic as the very last char and if it's in front of
|
|
2613 * either "\|", "\)", "\&", or "\n" */
|
|
2614 if (reg_magic >= MAGIC_OFF)
|
|
2615 {
|
|
2616 char_u *p = regparse + 1;
|
|
2617
|
|
2618 /* ignore \c \C \m and \M after '$' */
|
|
2619 while (p[0] == '\\' && (p[1] == 'c' || p[1] == 'C'
|
|
2620 || p[1] == 'm' || p[1] == 'M' || p[1] == 'Z'))
|
|
2621 p += 2;
|
|
2622 if (p[0] == NUL
|
|
2623 || (p[0] == '\\'
|
|
2624 && (p[1] == '|' || p[1] == '&' || p[1] == ')'
|
|
2625 || p[1] == 'n'))
|
|
2626 || reg_magic == MAGIC_ALL)
|
|
2627 curchr = Magic('$');
|
|
2628 }
|
|
2629 break;
|
|
2630 case '\\':
|
|
2631 {
|
|
2632 int c = regparse[1];
|
|
2633
|
|
2634 if (c == NUL)
|
|
2635 curchr = '\\'; /* trailing '\' */
|
|
2636 else if (
|
|
2637 #ifdef EBCDIC
|
|
2638 vim_strchr(META, c)
|
|
2639 #else
|
|
2640 c <= '~' && META_flags[c]
|
|
2641 #endif
|
|
2642 )
|
|
2643 {
|
|
2644 /*
|
|
2645 * META contains everything that may be magic sometimes,
|
|
2646 * except ^ and $ ("\^" and "\$" are only magic after
|
|
2647 * "\v"). We now fetch the next character and toggle its
|
|
2648 * magicness. Therefore, \ is so meta-magic that it is
|
|
2649 * not in META.
|
|
2650 */
|
|
2651 curchr = -1;
|
|
2652 prev_at_start = at_start;
|
|
2653 at_start = FALSE; /* be able to say "/\*ptr" */
|
|
2654 ++regparse;
|
167
|
2655 ++after_slash;
|
7
|
2656 peekchr();
|
|
2657 --regparse;
|
167
|
2658 --after_slash;
|
7
|
2659 curchr = toggle_Magic(curchr);
|
|
2660 }
|
|
2661 else if (vim_strchr(REGEXP_ABBR, c))
|
|
2662 {
|
|
2663 /*
|
|
2664 * Handle abbreviations, like "\t" for TAB -- webb
|
|
2665 */
|
|
2666 curchr = backslash_trans(c);
|
|
2667 }
|
|
2668 else if (reg_magic == MAGIC_NONE && (c == '$' || c == '^'))
|
|
2669 curchr = toggle_Magic(c);
|
|
2670 else
|
|
2671 {
|
|
2672 /*
|
|
2673 * Next character can never be (made) magic?
|
|
2674 * Then backslashing it won't do anything.
|
|
2675 */
|
|
2676 #ifdef FEAT_MBYTE
|
|
2677 if (has_mbyte)
|
|
2678 curchr = (*mb_ptr2char)(regparse + 1);
|
|
2679 else
|
|
2680 #endif
|
|
2681 curchr = c;
|
|
2682 }
|
|
2683 break;
|
|
2684 }
|
|
2685
|
|
2686 #ifdef FEAT_MBYTE
|
|
2687 default:
|
|
2688 if (has_mbyte)
|
|
2689 curchr = (*mb_ptr2char)(regparse);
|
|
2690 #endif
|
|
2691 }
|
|
2692 }
|
|
2693
|
|
2694 return curchr;
|
|
2695 }
|
|
2696
|
|
2697 /*
|
|
2698 * Eat one lexed character. Do this in a way that we can undo it.
|
|
2699 */
|
|
2700 static void
|
|
2701 skipchr()
|
|
2702 {
|
|
2703 /* peekchr() eats a backslash, do the same here */
|
|
2704 if (*regparse == '\\')
|
|
2705 prevchr_len = 1;
|
|
2706 else
|
|
2707 prevchr_len = 0;
|
|
2708 if (regparse[prevchr_len] != NUL)
|
|
2709 {
|
|
2710 #ifdef FEAT_MBYTE
|
|
2711 if (has_mbyte)
|
|
2712 prevchr_len += (*mb_ptr2len_check)(regparse + prevchr_len);
|
|
2713 else
|
|
2714 #endif
|
|
2715 ++prevchr_len;
|
|
2716 }
|
|
2717 regparse += prevchr_len;
|
|
2718 prev_at_start = at_start;
|
|
2719 at_start = FALSE;
|
|
2720 prevprevchr = prevchr;
|
|
2721 prevchr = curchr;
|
|
2722 curchr = nextchr; /* use previously unget char, or -1 */
|
|
2723 nextchr = -1;
|
|
2724 }
|
|
2725
|
|
2726 /*
|
|
2727 * Skip a character while keeping the value of prev_at_start for at_start.
|
|
2728 * prevchr and prevprevchr are also kept.
|
|
2729 */
|
|
2730 static void
|
|
2731 skipchr_keepstart()
|
|
2732 {
|
|
2733 int as = prev_at_start;
|
|
2734 int pr = prevchr;
|
|
2735 int prpr = prevprevchr;
|
|
2736
|
|
2737 skipchr();
|
|
2738 at_start = as;
|
|
2739 prevchr = pr;
|
|
2740 prevprevchr = prpr;
|
|
2741 }
|
|
2742
|
|
2743 static int
|
|
2744 getchr()
|
|
2745 {
|
|
2746 int chr = peekchr();
|
|
2747
|
|
2748 skipchr();
|
|
2749 return chr;
|
|
2750 }
|
|
2751
|
|
2752 /*
|
|
2753 * put character back. Works only once!
|
|
2754 */
|
|
2755 static void
|
|
2756 ungetchr()
|
|
2757 {
|
|
2758 nextchr = curchr;
|
|
2759 curchr = prevchr;
|
|
2760 prevchr = prevprevchr;
|
|
2761 at_start = prev_at_start;
|
|
2762 prev_at_start = FALSE;
|
|
2763
|
|
2764 /* Backup regparse, so that it's at the same position as before the
|
|
2765 * getchr(). */
|
|
2766 regparse -= prevchr_len;
|
|
2767 }
|
|
2768
|
|
2769 /*
|
29
|
2770 * Get and return the value of the hex string at the current position.
|
|
2771 * Return -1 if there is no valid hex number.
|
|
2772 * The position is updated:
|
24
|
2773 * blahblah\%x20asdf
|
|
2774 * before-^ ^-after
|
|
2775 * The parameter controls the maximum number of input characters. This will be
|
|
2776 * 2 when reading a \%x20 sequence and 4 when reading a \%u20AC sequence.
|
|
2777 */
|
|
2778 static int
|
|
2779 gethexchrs(maxinputlen)
|
|
2780 int maxinputlen;
|
|
2781 {
|
|
2782 int nr = 0;
|
|
2783 int c;
|
|
2784 int i;
|
|
2785
|
|
2786 for (i = 0; i < maxinputlen; ++i)
|
|
2787 {
|
|
2788 c = regparse[0];
|
|
2789 if (!vim_isxdigit(c))
|
|
2790 break;
|
|
2791 nr <<= 4;
|
|
2792 nr |= hex2nr(c);
|
|
2793 ++regparse;
|
|
2794 }
|
|
2795
|
|
2796 if (i == 0)
|
|
2797 return -1;
|
|
2798 return nr;
|
|
2799 }
|
|
2800
|
|
2801 /*
|
|
2802 * get and return the value of the decimal string immediately after the
|
|
2803 * current position. Return -1 for invalid. Consumes all digits.
|
|
2804 */
|
|
2805 static int
|
|
2806 getdecchrs()
|
|
2807 {
|
|
2808 int nr = 0;
|
|
2809 int c;
|
|
2810 int i;
|
|
2811
|
|
2812 for (i = 0; ; ++i)
|
|
2813 {
|
|
2814 c = regparse[0];
|
|
2815 if (c < '0' || c > '9')
|
|
2816 break;
|
|
2817 nr *= 10;
|
|
2818 nr += c - '0';
|
|
2819 ++regparse;
|
|
2820 }
|
|
2821
|
|
2822 if (i == 0)
|
|
2823 return -1;
|
|
2824 return nr;
|
|
2825 }
|
|
2826
|
|
2827 /*
|
|
2828 * get and return the value of the octal string immediately after the current
|
|
2829 * position. Return -1 for invalid, or 0-255 for valid. Smart enough to handle
|
|
2830 * numbers > 377 correctly (for example, 400 is treated as 40) and doesn't
|
|
2831 * treat 8 or 9 as recognised characters. Position is updated:
|
|
2832 * blahblah\%o210asdf
|
|
2833 * before-^ ^-after
|
|
2834 */
|
|
2835 static int
|
|
2836 getoctchrs()
|
|
2837 {
|
|
2838 int nr = 0;
|
|
2839 int c;
|
|
2840 int i;
|
|
2841
|
|
2842 for (i = 0; i < 3 && nr < 040; ++i)
|
|
2843 {
|
|
2844 c = regparse[0];
|
|
2845 if (c < '0' || c > '7')
|
|
2846 break;
|
|
2847 nr <<= 3;
|
|
2848 nr |= hex2nr(c);
|
|
2849 ++regparse;
|
|
2850 }
|
|
2851
|
|
2852 if (i == 0)
|
|
2853 return -1;
|
|
2854 return nr;
|
|
2855 }
|
|
2856
|
|
2857 /*
|
|
2858 * Get a number after a backslash that is inside [].
|
|
2859 * When nothing is recognized return a backslash.
|
|
2860 */
|
|
2861 static int
|
|
2862 coll_get_char()
|
|
2863 {
|
|
2864 int nr = -1;
|
|
2865
|
|
2866 switch (*regparse++)
|
|
2867 {
|
|
2868 case 'd': nr = getdecchrs(); break;
|
|
2869 case 'o': nr = getoctchrs(); break;
|
|
2870 case 'x': nr = gethexchrs(2); break;
|
|
2871 case 'u': nr = gethexchrs(4); break;
|
|
2872 case 'U': nr = gethexchrs(8); break;
|
|
2873 }
|
|
2874 if (nr < 0)
|
|
2875 {
|
|
2876 /* If getting the number fails be backwards compatible: the character
|
|
2877 * is a backslash. */
|
|
2878 --regparse;
|
|
2879 nr = '\\';
|
|
2880 }
|
|
2881 return nr;
|
|
2882 }
|
|
2883
|
|
2884 /*
|
7
|
2885 * read_limits - Read two integers to be taken as a minimum and maximum.
|
|
2886 * If the first character is '-', then the range is reversed.
|
|
2887 * Should end with 'end'. If minval is missing, zero is default, if maxval is
|
|
2888 * missing, a very big number is the default.
|
|
2889 */
|
|
2890 static int
|
|
2891 read_limits(minval, maxval)
|
|
2892 long *minval;
|
|
2893 long *maxval;
|
|
2894 {
|
|
2895 int reverse = FALSE;
|
|
2896 char_u *first_char;
|
|
2897 long tmp;
|
|
2898
|
|
2899 if (*regparse == '-')
|
|
2900 {
|
|
2901 /* Starts with '-', so reverse the range later */
|
|
2902 regparse++;
|
|
2903 reverse = TRUE;
|
|
2904 }
|
|
2905 first_char = regparse;
|
|
2906 *minval = getdigits(®parse);
|
|
2907 if (*regparse == ',') /* There is a comma */
|
|
2908 {
|
|
2909 if (vim_isdigit(*++regparse))
|
|
2910 *maxval = getdigits(®parse);
|
|
2911 else
|
|
2912 *maxval = MAX_LIMIT;
|
|
2913 }
|
|
2914 else if (VIM_ISDIGIT(*first_char))
|
|
2915 *maxval = *minval; /* It was \{n} or \{-n} */
|
|
2916 else
|
|
2917 *maxval = MAX_LIMIT; /* It was \{} or \{-} */
|
|
2918 if (*regparse == '\\')
|
|
2919 regparse++; /* Allow either \{...} or \{...\} */
|
167
|
2920 if (*regparse != '}')
|
7
|
2921 {
|
|
2922 sprintf((char *)IObuff, _("E554: Syntax error in %s{...}"),
|
|
2923 reg_magic == MAGIC_ALL ? "" : "\\");
|
|
2924 EMSG_RET_FAIL(IObuff);
|
|
2925 }
|
|
2926
|
|
2927 /*
|
|
2928 * Reverse the range if there was a '-', or make sure it is in the right
|
|
2929 * order otherwise.
|
|
2930 */
|
|
2931 if ((!reverse && *minval > *maxval) || (reverse && *minval < *maxval))
|
|
2932 {
|
|
2933 tmp = *minval;
|
|
2934 *minval = *maxval;
|
|
2935 *maxval = tmp;
|
|
2936 }
|
|
2937 skipchr(); /* let's be friends with the lexer again */
|
|
2938 return OK;
|
|
2939 }
|
|
2940
|
|
2941 /*
|
|
2942 * vim_regexec and friends
|
|
2943 */
|
|
2944
|
|
2945 /*
|
|
2946 * Global work variables for vim_regexec().
|
|
2947 */
|
|
2948
|
|
2949 /* The current match-position is remembered with these variables: */
|
|
2950 static linenr_T reglnum; /* line number, relative to first line */
|
|
2951 static char_u *regline; /* start of current line */
|
|
2952 static char_u *reginput; /* current input, points into "regline" */
|
|
2953
|
|
2954 static int need_clear_subexpr; /* subexpressions still need to be
|
|
2955 * cleared */
|
|
2956 #ifdef FEAT_SYN_HL
|
|
2957 static int need_clear_zsubexpr = FALSE; /* extmatch subexpressions
|
|
2958 * still need to be cleared */
|
|
2959 #endif
|
|
2960
|
|
2961 static int out_of_stack; /* TRUE when ran out of stack space */
|
|
2962
|
|
2963 /*
|
|
2964 * Structure used to save the current input state, when it needs to be
|
|
2965 * restored after trying a match. Used by reg_save() and reg_restore().
|
|
2966 */
|
|
2967 typedef struct
|
|
2968 {
|
|
2969 union
|
|
2970 {
|
|
2971 char_u *ptr; /* reginput pointer, for single-line regexp */
|
|
2972 lpos_T pos; /* reginput pos, for multi-line regexp */
|
|
2973 } rs_u;
|
|
2974 } regsave_T;
|
|
2975
|
|
2976 /* struct to save start/end pointer/position in for \(\) */
|
|
2977 typedef struct
|
|
2978 {
|
|
2979 union
|
|
2980 {
|
|
2981 char_u *ptr;
|
|
2982 lpos_T pos;
|
|
2983 } se_u;
|
|
2984 } save_se_T;
|
|
2985
|
|
2986 static char_u *reg_getline __ARGS((linenr_T lnum));
|
|
2987 static long vim_regexec_both __ARGS((char_u *line, colnr_T col));
|
|
2988 static long regtry __ARGS((regprog_T *prog, colnr_T col));
|
|
2989 static void cleanup_subexpr __ARGS((void));
|
|
2990 #ifdef FEAT_SYN_HL
|
|
2991 static void cleanup_zsubexpr __ARGS((void));
|
|
2992 #endif
|
|
2993 static void reg_nextline __ARGS((void));
|
|
2994 static void reg_save __ARGS((regsave_T *save));
|
|
2995 static void reg_restore __ARGS((regsave_T *save));
|
|
2996 static int reg_save_equal __ARGS((regsave_T *save));
|
|
2997 static void save_se_multi __ARGS((save_se_T *savep, lpos_T *posp));
|
|
2998 static void save_se_one __ARGS((save_se_T *savep, char_u **pp));
|
|
2999
|
|
3000 /* Save the sub-expressions before attempting a match. */
|
|
3001 #define save_se(savep, posp, pp) \
|
|
3002 REG_MULTI ? save_se_multi((savep), (posp)) : save_se_one((savep), (pp))
|
|
3003
|
|
3004 /* After a failed match restore the sub-expressions. */
|
|
3005 #define restore_se(savep, posp, pp) { \
|
|
3006 if (REG_MULTI) \
|
|
3007 *(posp) = (savep)->se_u.pos; \
|
|
3008 else \
|
|
3009 *(pp) = (savep)->se_u.ptr; }
|
|
3010
|
|
3011 static int re_num_cmp __ARGS((long_u val, char_u *scan));
|
167
|
3012 static int regmatch __ARGS((char_u *prog, regsave_T *startp));
|
7
|
3013 static int regrepeat __ARGS((char_u *p, long maxcount));
|
|
3014
|
|
3015 #ifdef DEBUG
|
|
3016 int regnarrate = 0;
|
|
3017 #endif
|
|
3018
|
|
3019 /*
|
|
3020 * Internal copy of 'ignorecase'. It is set at each call to vim_regexec().
|
|
3021 * Normally it gets the value of "rm_ic" or "rmm_ic", but when the pattern
|
|
3022 * contains '\c' or '\C' the value is overruled.
|
|
3023 */
|
|
3024 static int ireg_ic;
|
|
3025
|
|
3026 #ifdef FEAT_MBYTE
|
|
3027 /*
|
|
3028 * Similar to ireg_ic, but only for 'combining' characters. Set with \Z flag
|
|
3029 * in the regexp. Defaults to false, always.
|
|
3030 */
|
|
3031 static int ireg_icombine;
|
|
3032 #endif
|
|
3033
|
|
3034 /*
|
|
3035 * Sometimes need to save a copy of a line. Since alloc()/free() is very
|
|
3036 * slow, we keep one allocated piece of memory and only re-allocate it when
|
|
3037 * it's too small. It's freed in vim_regexec_both() when finished.
|
|
3038 */
|
|
3039 static char_u *reg_tofree;
|
|
3040 static unsigned reg_tofreelen;
|
|
3041
|
|
3042 /*
|
|
3043 * These variables are set when executing a regexp to speed up the execution.
|
|
3044 * Which ones are set depends on whethere a single-line or multi-line match is
|
|
3045 * done:
|
|
3046 * single-line multi-line
|
|
3047 * reg_match ®match_T NULL
|
|
3048 * reg_mmatch NULL ®mmatch_T
|
|
3049 * reg_startp reg_match->startp <invalid>
|
|
3050 * reg_endp reg_match->endp <invalid>
|
|
3051 * reg_startpos <invalid> reg_mmatch->startpos
|
|
3052 * reg_endpos <invalid> reg_mmatch->endpos
|
|
3053 * reg_win NULL window in which to search
|
|
3054 * reg_buf <invalid> buffer in which to search
|
|
3055 * reg_firstlnum <invalid> first line in which to search
|
|
3056 * reg_maxline 0 last line nr
|
|
3057 * reg_line_lbr FALSE or TRUE FALSE
|
|
3058 */
|
|
3059 static regmatch_T *reg_match;
|
|
3060 static regmmatch_T *reg_mmatch;
|
|
3061 static char_u **reg_startp = NULL;
|
|
3062 static char_u **reg_endp = NULL;
|
|
3063 static lpos_T *reg_startpos = NULL;
|
|
3064 static lpos_T *reg_endpos = NULL;
|
|
3065 static win_T *reg_win;
|
|
3066 static buf_T *reg_buf;
|
|
3067 static linenr_T reg_firstlnum;
|
|
3068 static linenr_T reg_maxline;
|
|
3069 static int reg_line_lbr; /* "\n" in string is line break */
|
|
3070
|
|
3071 /*
|
|
3072 * Get pointer to the line "lnum", which is relative to "reg_firstlnum".
|
|
3073 */
|
|
3074 static char_u *
|
|
3075 reg_getline(lnum)
|
|
3076 linenr_T lnum;
|
|
3077 {
|
|
3078 /* when looking behind for a match/no-match lnum is negative. But we
|
|
3079 * can't go before line 1 */
|
|
3080 if (reg_firstlnum + lnum < 1)
|
|
3081 return NULL;
|
|
3082 return ml_get_buf(reg_buf, reg_firstlnum + lnum, FALSE);
|
|
3083 }
|
|
3084
|
|
3085 static regsave_T behind_pos;
|
|
3086
|
|
3087 #ifdef FEAT_SYN_HL
|
|
3088 static char_u *reg_startzp[NSUBEXP]; /* Workspace to mark beginning */
|
|
3089 static char_u *reg_endzp[NSUBEXP]; /* and end of \z(...\) matches */
|
|
3090 static lpos_T reg_startzpos[NSUBEXP]; /* idem, beginning pos */
|
|
3091 static lpos_T reg_endzpos[NSUBEXP]; /* idem, end pos */
|
|
3092 #endif
|
|
3093
|
|
3094 /* TRUE if using multi-line regexp. */
|
|
3095 #define REG_MULTI (reg_match == NULL)
|
|
3096
|
|
3097 /*
|
|
3098 * Match a regexp against a string.
|
|
3099 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp().
|
|
3100 * Uses curbuf for line count and 'iskeyword'.
|
|
3101 *
|
|
3102 * Return TRUE if there is a match, FALSE if not.
|
|
3103 */
|
|
3104 int
|
|
3105 vim_regexec(rmp, line, col)
|
|
3106 regmatch_T *rmp;
|
|
3107 char_u *line; /* string to match against */
|
|
3108 colnr_T col; /* column to start looking for match */
|
|
3109 {
|
|
3110 reg_match = rmp;
|
|
3111 reg_mmatch = NULL;
|
|
3112 reg_maxline = 0;
|
|
3113 reg_line_lbr = FALSE;
|
|
3114 reg_win = NULL;
|
|
3115 ireg_ic = rmp->rm_ic;
|
|
3116 #ifdef FEAT_MBYTE
|
|
3117 ireg_icombine = FALSE;
|
|
3118 #endif
|
|
3119 return (vim_regexec_both(line, col) != 0);
|
|
3120 }
|
|
3121
|
39
|
3122 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \
|
|
3123 || defined(FIND_REPLACE_DIALOG) || defined(PROTO)
|
7
|
3124 /*
|
|
3125 * Like vim_regexec(), but consider a "\n" in "line" to be a line break.
|
|
3126 */
|
|
3127 int
|
|
3128 vim_regexec_nl(rmp, line, col)
|
|
3129 regmatch_T *rmp;
|
|
3130 char_u *line; /* string to match against */
|
|
3131 colnr_T col; /* column to start looking for match */
|
|
3132 {
|
|
3133 reg_match = rmp;
|
|
3134 reg_mmatch = NULL;
|
|
3135 reg_maxline = 0;
|
|
3136 reg_line_lbr = TRUE;
|
|
3137 reg_win = NULL;
|
|
3138 ireg_ic = rmp->rm_ic;
|
|
3139 #ifdef FEAT_MBYTE
|
|
3140 ireg_icombine = FALSE;
|
|
3141 #endif
|
|
3142 return (vim_regexec_both(line, col) != 0);
|
|
3143 }
|
|
3144 #endif
|
|
3145
|
|
3146 /*
|
|
3147 * Match a regexp against multiple lines.
|
|
3148 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp().
|
|
3149 * Uses curbuf for line count and 'iskeyword'.
|
|
3150 *
|
|
3151 * Return zero if there is no match. Return number of lines contained in the
|
|
3152 * match otherwise.
|
|
3153 */
|
|
3154 long
|
|
3155 vim_regexec_multi(rmp, win, buf, lnum, col)
|
|
3156 regmmatch_T *rmp;
|
|
3157 win_T *win; /* window in which to search or NULL */
|
|
3158 buf_T *buf; /* buffer in which to search */
|
|
3159 linenr_T lnum; /* nr of line to start looking for match */
|
|
3160 colnr_T col; /* column to start looking for match */
|
|
3161 {
|
|
3162 long r;
|
|
3163 buf_T *save_curbuf = curbuf;
|
|
3164
|
|
3165 reg_match = NULL;
|
|
3166 reg_mmatch = rmp;
|
|
3167 reg_buf = buf;
|
|
3168 reg_win = win;
|
|
3169 reg_firstlnum = lnum;
|
|
3170 reg_maxline = reg_buf->b_ml.ml_line_count - lnum;
|
|
3171 reg_line_lbr = FALSE;
|
|
3172 ireg_ic = rmp->rmm_ic;
|
|
3173 #ifdef FEAT_MBYTE
|
|
3174 ireg_icombine = FALSE;
|
|
3175 #endif
|
|
3176
|
|
3177 /* Need to switch to buffer "buf" to make vim_iswordc() work. */
|
|
3178 curbuf = buf;
|
|
3179 r = vim_regexec_both(NULL, col);
|
|
3180 curbuf = save_curbuf;
|
|
3181
|
|
3182 return r;
|
|
3183 }
|
|
3184
|
177
|
3185 #if 0 /* this does not appear to work... */
|
|
3186 # ifdef __MINGW32__
|
|
3187 # define MINGW_TRY
|
|
3188 # endif
|
|
3189 #endif
|
|
3190
|
|
3191 #ifdef MINGW_TRY
|
|
3192 /*
|
|
3193 * Special assembly code for MingW to simulate __try / __except.
|
|
3194 * Does not work with the optimizer!
|
|
3195 */
|
|
3196 # include <excpt.h>
|
|
3197
|
|
3198 static void *ESP_save; /* used as _ESP below */
|
|
3199 static void *EBP_save; /* used as _EBP below */
|
|
3200
|
|
3201 __attribute__ ((cdecl))
|
|
3202 EXCEPTION_DISPOSITION
|
|
3203 _except_regexec_handler(
|
|
3204 struct _EXCEPTION_RECORD *ExceptionRecord,
|
|
3205 void *EstablisherFrame,
|
|
3206 struct _CONTEXT *ContextRecord,
|
|
3207 void *DispatcherContext)
|
|
3208 {
|
|
3209 __asm__ __volatile__ (
|
|
3210 "jmp regexec_reentry");
|
|
3211 return 0; /* Function does not return */
|
|
3212 }
|
|
3213 #endif
|
|
3214
|
7
|
3215 /*
|
|
3216 * Match a regexp against a string ("line" points to the string) or multiple
|
|
3217 * lines ("line" is NULL, use reg_getline()).
|
|
3218 */
|
|
3219 #ifdef HAVE_SETJMP_H
|
|
3220 static long
|
|
3221 vim_regexec_both(line_arg, col_arg)
|
|
3222 char_u *line_arg;
|
|
3223 colnr_T col_arg; /* column to start looking for match */
|
|
3224 #else
|
|
3225 static long
|
|
3226 vim_regexec_both(line, col)
|
|
3227 char_u *line;
|
|
3228 colnr_T col; /* column to start looking for match */
|
|
3229 #endif
|
|
3230 {
|
|
3231 regprog_T *prog;
|
|
3232 char_u *s;
|
|
3233 long retval;
|
|
3234 #ifdef HAVE_SETJMP_H
|
|
3235 char_u *line;
|
|
3236 colnr_T col;
|
153
|
3237 int did_mch_startjmp = FALSE;
|
7
|
3238 #endif
|
|
3239
|
|
3240 reg_tofree = NULL;
|
|
3241
|
|
3242 #ifdef HAVE_SETJMP_H
|
|
3243 /* Trick to avoid "might be clobbered by `longjmp'" warning from gcc. */
|
|
3244 line = line_arg;
|
|
3245 col = col_arg;
|
|
3246 #endif
|
|
3247 retval = 0L;
|
|
3248
|
|
3249 if (REG_MULTI)
|
|
3250 {
|
|
3251 prog = reg_mmatch->regprog;
|
|
3252 line = reg_getline((linenr_T)0);
|
|
3253 reg_startpos = reg_mmatch->startpos;
|
|
3254 reg_endpos = reg_mmatch->endpos;
|
|
3255 }
|
|
3256 else
|
|
3257 {
|
|
3258 prog = reg_match->regprog;
|
|
3259 reg_startp = reg_match->startp;
|
|
3260 reg_endp = reg_match->endp;
|
|
3261 }
|
|
3262
|
|
3263 /* Be paranoid... */
|
|
3264 if (prog == NULL || line == NULL)
|
|
3265 {
|
|
3266 EMSG(_(e_null));
|
|
3267 goto theend;
|
|
3268 }
|
|
3269
|
|
3270 /* Check validity of program. */
|
|
3271 if (prog_magic_wrong())
|
|
3272 goto theend;
|
|
3273
|
|
3274 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */
|
|
3275 if (prog->regflags & RF_ICASE)
|
|
3276 ireg_ic = TRUE;
|
|
3277 else if (prog->regflags & RF_NOICASE)
|
|
3278 ireg_ic = FALSE;
|
|
3279
|
|
3280 #ifdef FEAT_MBYTE
|
|
3281 /* If pattern contains "\Z" overrule value of ireg_icombine */
|
|
3282 if (prog->regflags & RF_ICOMBINE)
|
|
3283 ireg_icombine = TRUE;
|
|
3284 #endif
|
|
3285
|
|
3286 /* If there is a "must appear" string, look for it. */
|
|
3287 if (prog->regmust != NULL)
|
|
3288 {
|
|
3289 int c;
|
|
3290
|
|
3291 #ifdef FEAT_MBYTE
|
|
3292 if (has_mbyte)
|
|
3293 c = (*mb_ptr2char)(prog->regmust);
|
|
3294 else
|
|
3295 #endif
|
|
3296 c = *prog->regmust;
|
|
3297 s = line + col;
|
170
|
3298
|
|
3299 /*
|
|
3300 * This is used very often, esp. for ":global". Use three versions of
|
|
3301 * the loop to avoid overhead of conditions.
|
|
3302 */
|
|
3303 if (!ireg_ic
|
|
3304 #ifdef FEAT_MBYTE
|
|
3305 && !has_mbyte
|
|
3306 #endif
|
|
3307 )
|
|
3308 while ((s = vim_strbyte(s, c)) != NULL)
|
|
3309 {
|
|
3310 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0)
|
|
3311 break; /* Found it. */
|
|
3312 ++s;
|
|
3313 }
|
|
3314 #ifdef FEAT_MBYTE
|
|
3315 else if (!ireg_ic || (!enc_utf8 && mb_char2len(c) > 1))
|
|
3316 while ((s = vim_strchr(s, c)) != NULL)
|
|
3317 {
|
|
3318 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0)
|
|
3319 break; /* Found it. */
|
|
3320 mb_ptr_adv(s);
|
|
3321 }
|
|
3322 #endif
|
|
3323 else
|
|
3324 while ((s = cstrchr(s, c)) != NULL)
|
|
3325 {
|
|
3326 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0)
|
|
3327 break; /* Found it. */
|
|
3328 mb_ptr_adv(s);
|
|
3329 }
|
7
|
3330 if (s == NULL) /* Not present. */
|
|
3331 goto theend;
|
|
3332 }
|
|
3333
|
177
|
3334 #ifdef MINGW_TRY
|
|
3335 /* Ugly assembly code that is necessary to simulate "__try". */
|
|
3336 __asm__ __volatile__ (
|
|
3337 "movl %esp, _ESP_save" "\n\t"
|
|
3338 "movl %ebp, _EBP_save");
|
|
3339
|
|
3340 __asm__ __volatile__ (
|
|
3341 "pushl $__except_regexec_handler" "\n\t"
|
|
3342 "pushl %fs:0" "\n\t"
|
|
3343 "mov %esp, %fs:0");
|
|
3344 #endif
|
153
|
3345 #ifdef HAVE_TRY_EXCEPT
|
|
3346 __try
|
|
3347 {
|
|
3348 #endif
|
|
3349
|
|
3350 #ifdef HAVE_SETJMP_H
|
|
3351 /*
|
|
3352 * Matching with a regexp may cause a very deep recursive call of
|
|
3353 * regmatch(). Vim will crash when running out of stack space. Catch
|
|
3354 * this here if the system supports it.
|
|
3355 * It's a bit slow, do it after the check for "regmust".
|
|
3356 * Don't do it if the caller already set it up.
|
|
3357 */
|
|
3358 if (!lc_active)
|
|
3359 {
|
|
3360 did_mch_startjmp = TRUE;
|
|
3361 mch_startjmp();
|
|
3362 if (SETJMP(lc_jump_env) != 0)
|
|
3363 {
|
|
3364 mch_didjmp();
|
|
3365 # ifdef SIGHASARG
|
|
3366 if (lc_signal != SIGINT)
|
|
3367 # endif
|
|
3368 EMSG(_(e_complex));
|
|
3369 retval = 0L;
|
|
3370 goto inner_end;
|
|
3371 }
|
|
3372 }
|
|
3373 #endif
|
|
3374
|
7
|
3375 regline = line;
|
|
3376 reglnum = 0;
|
|
3377 out_of_stack = FALSE;
|
|
3378
|
|
3379 /* Simplest case: Anchored match need be tried only once. */
|
|
3380 if (prog->reganch)
|
|
3381 {
|
|
3382 int c;
|
|
3383
|
|
3384 #ifdef FEAT_MBYTE
|
|
3385 if (has_mbyte)
|
|
3386 c = (*mb_ptr2char)(regline + col);
|
|
3387 else
|
|
3388 #endif
|
|
3389 c = regline[col];
|
|
3390 if (prog->regstart == NUL
|
|
3391 || prog->regstart == c
|
|
3392 || (ireg_ic && ((
|
|
3393 #ifdef FEAT_MBYTE
|
|
3394 (enc_utf8 && utf_fold(prog->regstart) == utf_fold(c)))
|
|
3395 || (c < 255 && prog->regstart < 255 &&
|
|
3396 #endif
|
|
3397 TOLOWER_LOC(prog->regstart) == TOLOWER_LOC(c)))))
|
|
3398 retval = regtry(prog, col);
|
|
3399 else
|
|
3400 retval = 0;
|
|
3401 }
|
|
3402 else
|
|
3403 {
|
|
3404 /* Messy cases: unanchored match. */
|
|
3405 while (!got_int && !out_of_stack)
|
|
3406 {
|
|
3407 if (prog->regstart != NUL)
|
|
3408 {
|
170
|
3409 /* Skip until the char we know it must start with.
|
|
3410 * Used often, do some work to avoid call overhead. */
|
|
3411 if (!ireg_ic
|
|
3412 #ifdef FEAT_MBYTE
|
|
3413 && !has_mbyte
|
|
3414 #endif
|
|
3415 )
|
|
3416 s = vim_strbyte(regline + col, prog->regstart);
|
|
3417 else
|
|
3418 s = cstrchr(regline + col, prog->regstart);
|
7
|
3419 if (s == NULL)
|
|
3420 {
|
|
3421 retval = 0;
|
|
3422 break;
|
|
3423 }
|
|
3424 col = (int)(s - regline);
|
|
3425 }
|
|
3426
|
|
3427 retval = regtry(prog, col);
|
|
3428 if (retval > 0)
|
|
3429 break;
|
|
3430
|
|
3431 /* if not currently on the first line, get it again */
|
|
3432 if (reglnum != 0)
|
|
3433 {
|
|
3434 regline = reg_getline((linenr_T)0);
|
|
3435 reglnum = 0;
|
|
3436 }
|
|
3437 if (regline[col] == NUL)
|
|
3438 break;
|
|
3439 #ifdef FEAT_MBYTE
|
|
3440 if (has_mbyte)
|
|
3441 col += (*mb_ptr2len_check)(regline + col);
|
|
3442 else
|
|
3443 #endif
|
|
3444 ++col;
|
|
3445 }
|
|
3446 }
|
|
3447
|
|
3448 if (out_of_stack)
|
153
|
3449 EMSG(_(e_outofstack));
|
|
3450
|
|
3451 #ifdef HAVE_SETJMP_H
|
|
3452 inner_end:
|
170
|
3453 if (did_mch_startjmp)
|
|
3454 mch_endjmp();
|
153
|
3455 #endif
|
7
|
3456 #ifdef HAVE_TRY_EXCEPT
|
|
3457 }
|
|
3458 __except(EXCEPTION_EXECUTE_HANDLER)
|
|
3459 {
|
|
3460 if (GetExceptionCode() == EXCEPTION_STACK_OVERFLOW)
|
|
3461 {
|
|
3462 RESETSTKOFLW();
|
153
|
3463 EMSG(_(e_outofstack));
|
7
|
3464 }
|
|
3465 else
|
153
|
3466 EMSG(_(e_complex));
|
7
|
3467 retval = 0L;
|
|
3468 }
|
|
3469 #endif
|
177
|
3470 #ifdef MINGW_TRY
|
|
3471 __asm__ __volatile__ (
|
|
3472 "jmp regexec_pop" "\n"
|
|
3473 "regexec_reentry:" "\n\t"
|
|
3474 "movl _ESP_save, %esp" "\n\t"
|
|
3475 "movl _EBP_save, %ebp");
|
|
3476
|
|
3477 EMSG(_(e_complex));
|
|
3478 retval = 0L;
|
|
3479
|
|
3480 __asm__ __volatile__ (
|
|
3481 "regexec_pop:" "\n\t"
|
|
3482 "mov (%esp), %eax" "\n\t"
|
|
3483 "mov %eax, %fs:0" "\n\t"
|
|
3484 "add $8, %esp");
|
|
3485 #endif
|
7
|
3486
|
|
3487 theend:
|
|
3488 /* Didn't find a match. */
|
|
3489 vim_free(reg_tofree);
|
|
3490 return retval;
|
|
3491 }
|
|
3492
|
|
3493 #ifdef FEAT_SYN_HL
|
|
3494 static reg_extmatch_T *make_extmatch __ARGS((void));
|
|
3495
|
|
3496 /*
|
|
3497 * Create a new extmatch and mark it as referenced once.
|
|
3498 */
|
|
3499 static reg_extmatch_T *
|
|
3500 make_extmatch()
|
|
3501 {
|
|
3502 reg_extmatch_T *em;
|
|
3503
|
|
3504 em = (reg_extmatch_T *)alloc_clear((unsigned)sizeof(reg_extmatch_T));
|
|
3505 if (em != NULL)
|
|
3506 em->refcnt = 1;
|
|
3507 return em;
|
|
3508 }
|
|
3509
|
|
3510 /*
|
|
3511 * Add a reference to an extmatch.
|
|
3512 */
|
|
3513 reg_extmatch_T *
|
|
3514 ref_extmatch(em)
|
|
3515 reg_extmatch_T *em;
|
|
3516 {
|
|
3517 if (em != NULL)
|
|
3518 em->refcnt++;
|
|
3519 return em;
|
|
3520 }
|
|
3521
|
|
3522 /*
|
|
3523 * Remove a reference to an extmatch. If there are no references left, free
|
|
3524 * the info.
|
|
3525 */
|
|
3526 void
|
|
3527 unref_extmatch(em)
|
|
3528 reg_extmatch_T *em;
|
|
3529 {
|
|
3530 int i;
|
|
3531
|
|
3532 if (em != NULL && --em->refcnt <= 0)
|
|
3533 {
|
|
3534 for (i = 0; i < NSUBEXP; ++i)
|
|
3535 vim_free(em->matches[i]);
|
|
3536 vim_free(em);
|
|
3537 }
|
|
3538 }
|
|
3539 #endif
|
|
3540
|
|
3541 /*
|
|
3542 * regtry - try match of "prog" with at regline["col"].
|
|
3543 * Returns 0 for failure, number of lines contained in the match otherwise.
|
|
3544 */
|
|
3545 static long
|
|
3546 regtry(prog, col)
|
|
3547 regprog_T *prog;
|
|
3548 colnr_T col;
|
|
3549 {
|
|
3550 reginput = regline + col;
|
|
3551 need_clear_subexpr = TRUE;
|
|
3552 #ifdef FEAT_SYN_HL
|
|
3553 /* Clear the external match subpointers if necessary. */
|
|
3554 if (prog->reghasz == REX_SET)
|
|
3555 need_clear_zsubexpr = TRUE;
|
|
3556 #endif
|
|
3557
|
167
|
3558 if (regmatch(prog->program + 1, NULL))
|
7
|
3559 {
|
|
3560 cleanup_subexpr();
|
|
3561 if (REG_MULTI)
|
|
3562 {
|
|
3563 if (reg_startpos[0].lnum < 0)
|
|
3564 {
|
|
3565 reg_startpos[0].lnum = 0;
|
|
3566 reg_startpos[0].col = col;
|
|
3567 }
|
|
3568 if (reg_endpos[0].lnum < 0)
|
|
3569 {
|
|
3570 reg_endpos[0].lnum = reglnum;
|
|
3571 reg_endpos[0].col = (int)(reginput - regline);
|
|
3572 }
|
|
3573 else
|
|
3574 /* Use line number of "\ze". */
|
|
3575 reglnum = reg_endpos[0].lnum;
|
|
3576 }
|
|
3577 else
|
|
3578 {
|
|
3579 if (reg_startp[0] == NULL)
|
|
3580 reg_startp[0] = regline + col;
|
|
3581 if (reg_endp[0] == NULL)
|
|
3582 reg_endp[0] = reginput;
|
|
3583 }
|
|
3584 #ifdef FEAT_SYN_HL
|
|
3585 /* Package any found \z(...\) matches for export. Default is none. */
|
|
3586 unref_extmatch(re_extmatch_out);
|
|
3587 re_extmatch_out = NULL;
|
|
3588
|
|
3589 if (prog->reghasz == REX_SET)
|
|
3590 {
|
|
3591 int i;
|
|
3592
|
|
3593 cleanup_zsubexpr();
|
|
3594 re_extmatch_out = make_extmatch();
|
|
3595 for (i = 0; i < NSUBEXP; i++)
|
|
3596 {
|
|
3597 if (REG_MULTI)
|
|
3598 {
|
|
3599 /* Only accept single line matches. */
|
|
3600 if (reg_startzpos[i].lnum >= 0
|
|
3601 && reg_endzpos[i].lnum == reg_startzpos[i].lnum)
|
|
3602 re_extmatch_out->matches[i] =
|
|
3603 vim_strnsave(reg_getline(reg_startzpos[i].lnum)
|
|
3604 + reg_startzpos[i].col,
|
|
3605 reg_endzpos[i].col - reg_startzpos[i].col);
|
|
3606 }
|
|
3607 else
|
|
3608 {
|
|
3609 if (reg_startzp[i] != NULL && reg_endzp[i] != NULL)
|
|
3610 re_extmatch_out->matches[i] =
|
|
3611 vim_strnsave(reg_startzp[i],
|
|
3612 (int)(reg_endzp[i] - reg_startzp[i]));
|
|
3613 }
|
|
3614 }
|
|
3615 }
|
|
3616 #endif
|
|
3617 return 1 + reglnum;
|
|
3618 }
|
|
3619 return 0;
|
|
3620 }
|
|
3621
|
|
3622 #ifdef FEAT_MBYTE
|
|
3623 static int reg_prev_class __ARGS((void));
|
|
3624
|
|
3625 /*
|
|
3626 * Get class of previous character.
|
|
3627 */
|
|
3628 static int
|
|
3629 reg_prev_class()
|
|
3630 {
|
|
3631 if (reginput > regline)
|
|
3632 return mb_get_class(reginput - 1
|
|
3633 - (*mb_head_off)(regline, reginput - 1));
|
|
3634 return -1;
|
|
3635 }
|
|
3636
|
|
3637 #endif
|
39
|
3638 #define ADVANCE_REGINPUT() mb_ptr_adv(reginput)
|
7
|
3639
|
|
3640 /*
|
|
3641 * The arguments from BRACE_LIMITS are stored here. They are actually local
|
|
3642 * to regmatch(), but they are here to reduce the amount of stack space used
|
|
3643 * (it can be called recursively many times).
|
|
3644 */
|
|
3645 static long bl_minval;
|
|
3646 static long bl_maxval;
|
|
3647
|
|
3648 /*
|
|
3649 * regmatch - main matching routine
|
|
3650 *
|
|
3651 * Conceptually the strategy is simple: Check to see whether the current
|
|
3652 * node matches, call self recursively to see whether the rest matches,
|
|
3653 * and then act accordingly. In practice we make some effort to avoid
|
|
3654 * recursion, in particular by going through "ordinary" nodes (that don't
|
|
3655 * need to know whether the rest of the match failed) by a loop instead of
|
|
3656 * by recursion.
|
|
3657 *
|
|
3658 * Returns TRUE when there is a match. Leaves reginput and reglnum just after
|
|
3659 * the last matched character.
|
|
3660 * Returns FALSE when there is no match. Leaves reginput and reglnum in an
|
|
3661 * undefined state!
|
|
3662 */
|
|
3663 static int
|
167
|
3664 regmatch(scan, startp)
|
7
|
3665 char_u *scan; /* Current node. */
|
167
|
3666 regsave_T *startp; /* start position for BACK */
|
7
|
3667 {
|
|
3668 char_u *next; /* Next node. */
|
|
3669 int op;
|
|
3670 int c;
|
|
3671
|
|
3672 #ifdef HAVE_GETRLIMIT
|
|
3673 /* Check if we are running out of stack space. Could be caused by
|
|
3674 * recursively calling ourselves. */
|
|
3675 if (out_of_stack || mch_stackcheck((char *)&op) == FAIL)
|
|
3676 {
|
|
3677 out_of_stack = TRUE;
|
|
3678 return FALSE;
|
|
3679 }
|
|
3680 #endif
|
|
3681
|
|
3682 /* Some patterns my cause a long time to match, even though they are not
|
|
3683 * illegal. E.g., "\([a-z]\+\)\+Q". Allow breaking them with CTRL-C. */
|
|
3684 fast_breakcheck();
|
|
3685
|
|
3686 #ifdef DEBUG
|
|
3687 if (scan != NULL && regnarrate)
|
|
3688 {
|
|
3689 mch_errmsg(regprop(scan));
|
|
3690 mch_errmsg("(\n");
|
|
3691 }
|
|
3692 #endif
|
|
3693 while (scan != NULL)
|
|
3694 {
|
|
3695 if (got_int || out_of_stack)
|
|
3696 return FALSE;
|
|
3697 #ifdef DEBUG
|
|
3698 if (regnarrate)
|
|
3699 {
|
|
3700 mch_errmsg(regprop(scan));
|
|
3701 mch_errmsg("...\n");
|
|
3702 # ifdef FEAT_SYN_HL
|
|
3703 if (re_extmatch_in != NULL)
|
|
3704 {
|
|
3705 int i;
|
|
3706
|
|
3707 mch_errmsg(_("External submatches:\n"));
|
|
3708 for (i = 0; i < NSUBEXP; i++)
|
|
3709 {
|
|
3710 mch_errmsg(" \"");
|
|
3711 if (re_extmatch_in->matches[i] != NULL)
|
|
3712 mch_errmsg(re_extmatch_in->matches[i]);
|
|
3713 mch_errmsg("\"\n");
|
|
3714 }
|
|
3715 }
|
|
3716 # endif
|
|
3717 }
|
|
3718 #endif
|
|
3719 next = regnext(scan);
|
|
3720
|
|
3721 op = OP(scan);
|
|
3722 /* Check for character class with NL added. */
|
|
3723 if (WITH_NL(op) && *reginput == NUL && reglnum < reg_maxline)
|
|
3724 {
|
|
3725 reg_nextline();
|
|
3726 }
|
|
3727 else if (reg_line_lbr && WITH_NL(op) && *reginput == '\n')
|
|
3728 {
|
|
3729 ADVANCE_REGINPUT();
|
|
3730 }
|
|
3731 else
|
|
3732 {
|
|
3733 if (WITH_NL(op))
|
|
3734 op -= ADD_NL;
|
|
3735 #ifdef FEAT_MBYTE
|
|
3736 if (has_mbyte)
|
|
3737 c = (*mb_ptr2char)(reginput);
|
|
3738 else
|
|
3739 #endif
|
|
3740 c = *reginput;
|
|
3741 switch (op)
|
|
3742 {
|
|
3743 case BOL:
|
|
3744 if (reginput != regline)
|
|
3745 return FALSE;
|
|
3746 break;
|
|
3747
|
|
3748 case EOL:
|
|
3749 if (c != NUL)
|
|
3750 return FALSE;
|
|
3751 break;
|
|
3752
|
|
3753 case RE_BOF:
|
|
3754 /* Passing -1 to the getline() function provided for the search
|
|
3755 * should always return NULL if the current line is the first
|
|
3756 * line of the file. */
|
|
3757 if (reglnum != 0 || reginput != regline
|
|
3758 || (REG_MULTI && reg_getline((linenr_T)-1) != NULL))
|
|
3759 return FALSE;
|
|
3760 break;
|
|
3761
|
|
3762 case RE_EOF:
|
|
3763 if (reglnum != reg_maxline || c != NUL)
|
|
3764 return FALSE;
|
|
3765 break;
|
|
3766
|
|
3767 case CURSOR:
|
|
3768 /* Check if the buffer is in a window and compare the
|
|
3769 * reg_win->w_cursor position to the match position. */
|
|
3770 if (reg_win == NULL
|
|
3771 || (reglnum + reg_firstlnum != reg_win->w_cursor.lnum)
|
|
3772 || ((colnr_T)(reginput - regline) != reg_win->w_cursor.col))
|
|
3773 return FALSE;
|
|
3774 break;
|
|
3775
|
|
3776 case RE_LNUM:
|
|
3777 if (!REG_MULTI || !re_num_cmp((long_u)(reglnum + reg_firstlnum),
|
|
3778 scan))
|
|
3779 return FALSE;
|
|
3780 break;
|
|
3781
|
|
3782 case RE_COL:
|
|
3783 if (!re_num_cmp((long_u)(reginput - regline) + 1, scan))
|
|
3784 return FALSE;
|
|
3785 break;
|
|
3786
|
|
3787 case RE_VCOL:
|
|
3788 if (!re_num_cmp((long_u)win_linetabsize(
|
|
3789 reg_win == NULL ? curwin : reg_win,
|
|
3790 regline, (colnr_T)(reginput - regline)) + 1, scan))
|
|
3791 return FALSE;
|
|
3792 break;
|
|
3793
|
|
3794 case BOW: /* \<word; reginput points to w */
|
|
3795 if (c == NUL) /* Can't match at end of line */
|
|
3796 return FALSE;
|
|
3797 #ifdef FEAT_MBYTE
|
|
3798 if (has_mbyte)
|
|
3799 {
|
|
3800 int this_class;
|
|
3801
|
|
3802 /* Get class of current and previous char (if it exists). */
|
|
3803 this_class = mb_get_class(reginput);
|
|
3804 if (this_class <= 1)
|
|
3805 return FALSE; /* not on a word at all */
|
|
3806 if (reg_prev_class() == this_class)
|
|
3807 return FALSE; /* previous char is in same word */
|
|
3808 }
|
|
3809 #endif
|
|
3810 else
|
|
3811 {
|
|
3812 if (!vim_iswordc(c)
|
|
3813 || (reginput > regline && vim_iswordc(reginput[-1])))
|
|
3814 return FALSE;
|
|
3815 }
|
|
3816 break;
|
|
3817
|
|
3818 case EOW: /* word\>; reginput points after d */
|
|
3819 if (reginput == regline) /* Can't match at start of line */
|
|
3820 return FALSE;
|
|
3821 #ifdef FEAT_MBYTE
|
|
3822 if (has_mbyte)
|
|
3823 {
|
|
3824 int this_class, prev_class;
|
|
3825
|
|
3826 /* Get class of current and previous char (if it exists). */
|
|
3827 this_class = mb_get_class(reginput);
|
|
3828 prev_class = reg_prev_class();
|
|
3829 if (this_class == prev_class)
|
|
3830 return FALSE;
|
|
3831 if (prev_class == 0 || prev_class == 1)
|
|
3832 return FALSE;
|
|
3833 }
|
|
3834 else
|
|
3835 #endif
|
|
3836 {
|
|
3837 if (!vim_iswordc(reginput[-1]))
|
|
3838 return FALSE;
|
|
3839 if (reginput[0] != NUL && vim_iswordc(c))
|
|
3840 return FALSE;
|
|
3841 }
|
|
3842 break; /* Matched with EOW */
|
|
3843
|
|
3844 case ANY:
|
|
3845 if (c == NUL)
|
|
3846 return FALSE;
|
|
3847 ADVANCE_REGINPUT();
|
|
3848 break;
|
|
3849
|
|
3850 case IDENT:
|
|
3851 if (!vim_isIDc(c))
|
|
3852 return FALSE;
|
|
3853 ADVANCE_REGINPUT();
|
|
3854 break;
|
|
3855
|
|
3856 case SIDENT:
|
|
3857 if (VIM_ISDIGIT(*reginput) || !vim_isIDc(c))
|
|
3858 return FALSE;
|
|
3859 ADVANCE_REGINPUT();
|
|
3860 break;
|
|
3861
|
|
3862 case KWORD:
|
|
3863 if (!vim_iswordp(reginput))
|
|
3864 return FALSE;
|
|
3865 ADVANCE_REGINPUT();
|
|
3866 break;
|
|
3867
|
|
3868 case SKWORD:
|
|
3869 if (VIM_ISDIGIT(*reginput) || !vim_iswordp(reginput))
|
|
3870 return FALSE;
|
|
3871 ADVANCE_REGINPUT();
|
|
3872 break;
|
|
3873
|
|
3874 case FNAME:
|
|
3875 if (!vim_isfilec(c))
|
|
3876 return FALSE;
|
|
3877 ADVANCE_REGINPUT();
|
|
3878 break;
|
|
3879
|
|
3880 case SFNAME:
|
|
3881 if (VIM_ISDIGIT(*reginput) || !vim_isfilec(c))
|
|
3882 return FALSE;
|
|
3883 ADVANCE_REGINPUT();
|
|
3884 break;
|
|
3885
|
|
3886 case PRINT:
|
|
3887 if (ptr2cells(reginput) != 1)
|
|
3888 return FALSE;
|
|
3889 ADVANCE_REGINPUT();
|
|
3890 break;
|
|
3891
|
|
3892 case SPRINT:
|
|
3893 if (VIM_ISDIGIT(*reginput) || ptr2cells(reginput) != 1)
|
|
3894 return FALSE;
|
|
3895 ADVANCE_REGINPUT();
|
|
3896 break;
|
|
3897
|
|
3898 case WHITE:
|
|
3899 if (!vim_iswhite(c))
|
|
3900 return FALSE;
|
|
3901 ADVANCE_REGINPUT();
|
|
3902 break;
|
|
3903
|
|
3904 case NWHITE:
|
|
3905 if (c == NUL || vim_iswhite(c))
|
|
3906 return FALSE;
|
|
3907 ADVANCE_REGINPUT();
|
|
3908 break;
|
|
3909
|
|
3910 case DIGIT:
|
|
3911 if (!ri_digit(c))
|
|
3912 return FALSE;
|
|
3913 ADVANCE_REGINPUT();
|
|
3914 break;
|
|
3915
|
|
3916 case NDIGIT:
|
|
3917 if (c == NUL || ri_digit(c))
|
|
3918 return FALSE;
|
|
3919 ADVANCE_REGINPUT();
|
|
3920 break;
|
|
3921
|
|
3922 case HEX:
|
|
3923 if (!ri_hex(c))
|
|
3924 return FALSE;
|
|
3925 ADVANCE_REGINPUT();
|
|
3926 break;
|
|
3927
|
|
3928 case NHEX:
|
|
3929 if (c == NUL || ri_hex(c))
|
|
3930 return FALSE;
|
|
3931 ADVANCE_REGINPUT();
|
|
3932 break;
|
|
3933
|
|
3934 case OCTAL:
|
|
3935 if (!ri_octal(c))
|
|
3936 return FALSE;
|
|
3937 ADVANCE_REGINPUT();
|
|
3938 break;
|
|
3939
|
|
3940 case NOCTAL:
|
|
3941 if (c == NUL || ri_octal(c))
|
|
3942 return FALSE;
|
|
3943 ADVANCE_REGINPUT();
|
|
3944 break;
|
|
3945
|
|
3946 case WORD:
|
|
3947 if (!ri_word(c))
|
|
3948 return FALSE;
|
|
3949 ADVANCE_REGINPUT();
|
|
3950 break;
|
|
3951
|
|
3952 case NWORD:
|
|
3953 if (c == NUL || ri_word(c))
|
|
3954 return FALSE;
|
|
3955 ADVANCE_REGINPUT();
|
|
3956 break;
|
|
3957
|
|
3958 case HEAD:
|
|
3959 if (!ri_head(c))
|
|
3960 return FALSE;
|
|
3961 ADVANCE_REGINPUT();
|
|
3962 break;
|
|
3963
|
|
3964 case NHEAD:
|
|
3965 if (c == NUL || ri_head(c))
|
|
3966 return FALSE;
|
|
3967 ADVANCE_REGINPUT();
|
|
3968 break;
|
|
3969
|
|
3970 case ALPHA:
|
|
3971 if (!ri_alpha(c))
|
|
3972 return FALSE;
|
|
3973 ADVANCE_REGINPUT();
|
|
3974 break;
|
|
3975
|
|
3976 case NALPHA:
|
|
3977 if (c == NUL || ri_alpha(c))
|
|
3978 return FALSE;
|
|
3979 ADVANCE_REGINPUT();
|
|
3980 break;
|
|
3981
|
|
3982 case LOWER:
|
|
3983 if (!ri_lower(c))
|
|
3984 return FALSE;
|
|
3985 ADVANCE_REGINPUT();
|
|
3986 break;
|
|
3987
|
|
3988 case NLOWER:
|
|
3989 if (c == NUL || ri_lower(c))
|
|
3990 return FALSE;
|
|
3991 ADVANCE_REGINPUT();
|
|
3992 break;
|
|
3993
|
|
3994 case UPPER:
|
|
3995 if (!ri_upper(c))
|
|
3996 return FALSE;
|
|
3997 ADVANCE_REGINPUT();
|
|
3998 break;
|
|
3999
|
|
4000 case NUPPER:
|
|
4001 if (c == NUL || ri_upper(c))
|
|
4002 return FALSE;
|
|
4003 ADVANCE_REGINPUT();
|
|
4004 break;
|
|
4005
|
|
4006 case EXACTLY:
|
|
4007 {
|
|
4008 int len;
|
|
4009 char_u *opnd;
|
|
4010
|
|
4011 opnd = OPERAND(scan);
|
|
4012 /* Inline the first byte, for speed. */
|
|
4013 if (*opnd != *reginput
|
|
4014 && (!ireg_ic || (
|
|
4015 #ifdef FEAT_MBYTE
|
|
4016 !enc_utf8 &&
|
|
4017 #endif
|
|
4018 TOLOWER_LOC(*opnd) != TOLOWER_LOC(*reginput))))
|
|
4019 return FALSE;
|
|
4020 if (*opnd == NUL)
|
|
4021 {
|
|
4022 /* match empty string always works; happens when "~" is
|
|
4023 * empty. */
|
|
4024 }
|
|
4025 else if (opnd[1] == NUL
|
|
4026 #ifdef FEAT_MBYTE
|
|
4027 && !(enc_utf8 && ireg_ic)
|
|
4028 #endif
|
|
4029 )
|
|
4030 ++reginput; /* matched a single char */
|
|
4031 else
|
|
4032 {
|
|
4033 len = (int)STRLEN(opnd);
|
|
4034 /* Need to match first byte again for multi-byte. */
|
|
4035 if (cstrncmp(opnd, reginput, &len) != 0)
|
|
4036 return FALSE;
|
|
4037 #ifdef FEAT_MBYTE
|
|
4038 /* Check for following composing character. */
|
|
4039 if (enc_utf8 && UTF_COMPOSINGLIKE(reginput, reginput + len))
|
|
4040 {
|
|
4041 /* raaron: This code makes a composing character get
|
|
4042 * ignored, which is the correct behavior (sometimes)
|
|
4043 * for voweled Hebrew texts. */
|
|
4044 if (!ireg_icombine)
|
|
4045 return FALSE;
|
|
4046 }
|
|
4047 else
|
|
4048 #endif
|
|
4049 reginput += len;
|
|
4050 }
|
|
4051 }
|
|
4052 break;
|
|
4053
|
|
4054 case ANYOF:
|
|
4055 case ANYBUT:
|
|
4056 if (c == NUL)
|
|
4057 return FALSE;
|
|
4058 if ((cstrchr(OPERAND(scan), c) == NULL) == (op == ANYOF))
|
|
4059 return FALSE;
|
|
4060 ADVANCE_REGINPUT();
|
|
4061 break;
|
|
4062
|
|
4063 #ifdef FEAT_MBYTE
|
|
4064 case MULTIBYTECODE:
|
|
4065 if (has_mbyte)
|
|
4066 {
|
|
4067 int i, len;
|
|
4068 char_u *opnd;
|
|
4069
|
|
4070 opnd = OPERAND(scan);
|
|
4071 /* Safety check (just in case 'encoding' was changed since
|
|
4072 * compiling the program). */
|
|
4073 if ((len = (*mb_ptr2len_check)(opnd)) < 2)
|
|
4074 return FALSE;
|
|
4075 for (i = 0; i < len; ++i)
|
|
4076 if (opnd[i] != reginput[i])
|
|
4077 return FALSE;
|
|
4078 reginput += len;
|
|
4079 }
|
|
4080 else
|
|
4081 return FALSE;
|
|
4082 break;
|
|
4083 #endif
|
|
4084
|
|
4085 case NOTHING:
|
|
4086 break;
|
|
4087
|
|
4088 case BACK:
|
167
|
4089 /* When we run into BACK without matching something non-empty, we
|
|
4090 * fail. */
|
|
4091 if (startp != NULL && reg_save_equal(startp))
|
|
4092 return FALSE;
|
7
|
4093 break;
|
|
4094
|
|
4095 case MOPEN + 0: /* Match start: \zs */
|
|
4096 case MOPEN + 1: /* \( */
|
|
4097 case MOPEN + 2:
|
|
4098 case MOPEN + 3:
|
|
4099 case MOPEN + 4:
|
|
4100 case MOPEN + 5:
|
|
4101 case MOPEN + 6:
|
|
4102 case MOPEN + 7:
|
|
4103 case MOPEN + 8:
|
|
4104 case MOPEN + 9:
|
|
4105 {
|
|
4106 int no;
|
|
4107 save_se_T save;
|
|
4108
|
|
4109 no = op - MOPEN;
|
|
4110 cleanup_subexpr();
|
|
4111 save_se(&save, ®_startpos[no], ®_startp[no]);
|
|
4112
|
167
|
4113 if (regmatch(next, startp))
|
7
|
4114 return TRUE;
|
|
4115
|
|
4116 restore_se(&save, ®_startpos[no], ®_startp[no]);
|
|
4117 return FALSE;
|
|
4118 }
|
|
4119 /* break; Not Reached */
|
|
4120
|
|
4121 case NOPEN: /* \%( */
|
|
4122 case NCLOSE: /* \) after \%( */
|
167
|
4123 if (regmatch(next, startp))
|
7
|
4124 return TRUE;
|
|
4125 return FALSE;
|
|
4126 /* break; Not Reached */
|
|
4127
|
|
4128 #ifdef FEAT_SYN_HL
|
|
4129 case ZOPEN + 1:
|
|
4130 case ZOPEN + 2:
|
|
4131 case ZOPEN + 3:
|
|
4132 case ZOPEN + 4:
|
|
4133 case ZOPEN + 5:
|
|
4134 case ZOPEN + 6:
|
|
4135 case ZOPEN + 7:
|
|
4136 case ZOPEN + 8:
|
|
4137 case ZOPEN + 9:
|
|
4138 {
|
|
4139 int no;
|
|
4140 save_se_T save;
|
|
4141
|
|
4142 no = op - ZOPEN;
|
|
4143 cleanup_zsubexpr();
|
|
4144 save_se(&save, ®_startzpos[no], ®_startzp[no]);
|
|
4145
|
167
|
4146 if (regmatch(next, startp))
|
7
|
4147 return TRUE;
|
|
4148
|
|
4149 restore_se(&save, ®_startzpos[no], ®_startzp[no]);
|
|
4150 return FALSE;
|
|
4151 }
|
|
4152 /* break; Not Reached */
|
|
4153 #endif
|
|
4154
|
|
4155 case MCLOSE + 0: /* Match end: \ze */
|
|
4156 case MCLOSE + 1: /* \) */
|
|
4157 case MCLOSE + 2:
|
|
4158 case MCLOSE + 3:
|
|
4159 case MCLOSE + 4:
|
|
4160 case MCLOSE + 5:
|
|
4161 case MCLOSE + 6:
|
|
4162 case MCLOSE + 7:
|
|
4163 case MCLOSE + 8:
|
|
4164 case MCLOSE + 9:
|
|
4165 {
|
|
4166 int no;
|
|
4167 save_se_T save;
|
|
4168
|
|
4169 no = op - MCLOSE;
|
|
4170 cleanup_subexpr();
|
|
4171 save_se(&save, ®_endpos[no], ®_endp[no]);
|
|
4172
|
167
|
4173 if (regmatch(next, startp))
|
7
|
4174 return TRUE;
|
|
4175
|
|
4176 restore_se(&save, ®_endpos[no], ®_endp[no]);
|
|
4177 return FALSE;
|
|
4178 }
|
|
4179 /* break; Not Reached */
|
|
4180
|
|
4181 #ifdef FEAT_SYN_HL
|
|
4182 case ZCLOSE + 1: /* \) after \z( */
|
|
4183 case ZCLOSE + 2:
|
|
4184 case ZCLOSE + 3:
|
|
4185 case ZCLOSE + 4:
|
|
4186 case ZCLOSE + 5:
|
|
4187 case ZCLOSE + 6:
|
|
4188 case ZCLOSE + 7:
|
|
4189 case ZCLOSE + 8:
|
|
4190 case ZCLOSE + 9:
|
|
4191 {
|
|
4192 int no;
|
|
4193 save_se_T save;
|
|
4194
|
|
4195 no = op - ZCLOSE;
|
|
4196 cleanup_zsubexpr();
|
|
4197 save_se(&save, ®_endzpos[no], ®_endzp[no]);
|
|
4198
|
167
|
4199 if (regmatch(next, startp))
|
7
|
4200 return TRUE;
|
|
4201
|
|
4202 restore_se(&save, ®_endzpos[no], ®_endzp[no]);
|
|
4203 return FALSE;
|
|
4204 }
|
|
4205 /* break; Not Reached */
|
|
4206 #endif
|
|
4207
|
|
4208 case BACKREF + 1:
|
|
4209 case BACKREF + 2:
|
|
4210 case BACKREF + 3:
|
|
4211 case BACKREF + 4:
|
|
4212 case BACKREF + 5:
|
|
4213 case BACKREF + 6:
|
|
4214 case BACKREF + 7:
|
|
4215 case BACKREF + 8:
|
|
4216 case BACKREF + 9:
|
|
4217 {
|
|
4218 int no;
|
|
4219 int len;
|
|
4220 linenr_T clnum;
|
|
4221 colnr_T ccol;
|
|
4222 char_u *p;
|
|
4223
|
|
4224 no = op - BACKREF;
|
|
4225 cleanup_subexpr();
|
|
4226 if (!REG_MULTI) /* Single-line regexp */
|
|
4227 {
|
|
4228 if (reg_endp[no] == NULL)
|
|
4229 {
|
|
4230 /* Backref was not set: Match an empty string. */
|
|
4231 len = 0;
|
|
4232 }
|
|
4233 else
|
|
4234 {
|
|
4235 /* Compare current input with back-ref in the same
|
|
4236 * line. */
|
|
4237 len = (int)(reg_endp[no] - reg_startp[no]);
|
|
4238 if (cstrncmp(reg_startp[no], reginput, &len) != 0)
|
|
4239 return FALSE;
|
|
4240 }
|
|
4241 }
|
|
4242 else /* Multi-line regexp */
|
|
4243 {
|
|
4244 if (reg_endpos[no].lnum < 0)
|
|
4245 {
|
|
4246 /* Backref was not set: Match an empty string. */
|
|
4247 len = 0;
|
|
4248 }
|
|
4249 else
|
|
4250 {
|
|
4251 if (reg_startpos[no].lnum == reglnum
|
|
4252 && reg_endpos[no].lnum == reglnum)
|
|
4253 {
|
|
4254 /* Compare back-ref within the current line. */
|
|
4255 len = reg_endpos[no].col - reg_startpos[no].col;
|
|
4256 if (cstrncmp(regline + reg_startpos[no].col,
|
|
4257 reginput, &len) != 0)
|
|
4258 return FALSE;
|
|
4259 }
|
|
4260 else
|
|
4261 {
|
|
4262 /* Messy situation: Need to compare between two
|
|
4263 * lines. */
|
|
4264 ccol = reg_startpos[no].col;
|
|
4265 clnum = reg_startpos[no].lnum;
|
|
4266 for (;;)
|
|
4267 {
|
|
4268 /* Since getting one line may invalidate
|
|
4269 * the other, need to make copy. Slow! */
|
|
4270 if (regline != reg_tofree)
|
|
4271 {
|
|
4272 len = (int)STRLEN(regline);
|
|
4273 if (reg_tofree == NULL
|
|
4274 || len >= (int)reg_tofreelen)
|
|
4275 {
|
|
4276 len += 50; /* get some extra */
|
|
4277 vim_free(reg_tofree);
|
|
4278 reg_tofree = alloc(len);
|
|
4279 if (reg_tofree == NULL)
|
|
4280 return FALSE; /* out of memory! */
|
|
4281 reg_tofreelen = len;
|
|
4282 }
|
|
4283 STRCPY(reg_tofree, regline);
|
|
4284 reginput = reg_tofree
|
|
4285 + (reginput - regline);
|
|
4286 regline = reg_tofree;
|
|
4287 }
|
|
4288
|
|
4289 /* Get the line to compare with. */
|
|
4290 p = reg_getline(clnum);
|
|
4291 if (clnum == reg_endpos[no].lnum)
|
|
4292 len = reg_endpos[no].col - ccol;
|
|
4293 else
|
|
4294 len = (int)STRLEN(p + ccol);
|
|
4295
|
|
4296 if (cstrncmp(p + ccol, reginput, &len) != 0)
|
|
4297 return FALSE; /* doesn't match */
|
|
4298 if (clnum == reg_endpos[no].lnum)
|
|
4299 break; /* match and at end! */
|
|
4300 if (reglnum == reg_maxline)
|
|
4301 return FALSE; /* text too short */
|
|
4302
|
|
4303 /* Advance to next line. */
|
|
4304 reg_nextline();
|
|
4305 ++clnum;
|
|
4306 ccol = 0;
|
|
4307 if (got_int || out_of_stack)
|
|
4308 return FALSE;
|
|
4309 }
|
|
4310
|
|
4311 /* found a match! Note that regline may now point
|
|
4312 * to a copy of the line, that should not matter. */
|
|
4313 }
|
|
4314 }
|
|
4315 }
|
|
4316
|
|
4317 /* Matched the backref, skip over it. */
|
|
4318 reginput += len;
|
|
4319 }
|
|
4320 break;
|
|
4321
|
|
4322 #ifdef FEAT_SYN_HL
|
|
4323 case ZREF + 1:
|
|
4324 case ZREF + 2:
|
|
4325 case ZREF + 3:
|
|
4326 case ZREF + 4:
|
|
4327 case ZREF + 5:
|
|
4328 case ZREF + 6:
|
|
4329 case ZREF + 7:
|
|
4330 case ZREF + 8:
|
|
4331 case ZREF + 9:
|
|
4332 {
|
|
4333 int no;
|
|
4334 int len;
|
|
4335
|
|
4336 cleanup_zsubexpr();
|
|
4337 no = op - ZREF;
|
|
4338 if (re_extmatch_in != NULL
|
|
4339 && re_extmatch_in->matches[no] != NULL)
|
|
4340 {
|
|
4341 len = (int)STRLEN(re_extmatch_in->matches[no]);
|
|
4342 if (cstrncmp(re_extmatch_in->matches[no],
|
|
4343 reginput, &len) != 0)
|
|
4344 return FALSE;
|
|
4345 reginput += len;
|
|
4346 }
|
|
4347 else
|
|
4348 {
|
|
4349 /* Backref was not set: Match an empty string. */
|
|
4350 }
|
|
4351 }
|
|
4352 break;
|
|
4353 #endif
|
|
4354
|
|
4355 case BRANCH:
|
|
4356 {
|
|
4357 if (OP(next) != BRANCH) /* No choice. */
|
|
4358 next = OPERAND(scan); /* Avoid recursion. */
|
|
4359 else
|
|
4360 {
|
|
4361 regsave_T save;
|
|
4362
|
|
4363 do
|
|
4364 {
|
|
4365 reg_save(&save);
|
167
|
4366 if (regmatch(OPERAND(scan), &save))
|
7
|
4367 return TRUE;
|
|
4368 reg_restore(&save);
|
|
4369 scan = regnext(scan);
|
|
4370 } while (scan != NULL && OP(scan) == BRANCH);
|
|
4371 return FALSE;
|
|
4372 /* NOTREACHED */
|
|
4373 }
|
|
4374 }
|
|
4375 break;
|
|
4376
|
|
4377 case BRACE_LIMITS:
|
|
4378 {
|
|
4379 int no;
|
|
4380
|
|
4381 if (OP(next) == BRACE_SIMPLE)
|
|
4382 {
|
|
4383 bl_minval = OPERAND_MIN(scan);
|
|
4384 bl_maxval = OPERAND_MAX(scan);
|
|
4385 }
|
|
4386 else if (OP(next) >= BRACE_COMPLEX
|
|
4387 && OP(next) < BRACE_COMPLEX + 10)
|
|
4388 {
|
|
4389 no = OP(next) - BRACE_COMPLEX;
|
|
4390 brace_min[no] = OPERAND_MIN(scan);
|
|
4391 brace_max[no] = OPERAND_MAX(scan);
|
|
4392 brace_count[no] = 0;
|
|
4393 }
|
|
4394 else
|
|
4395 {
|
|
4396 EMSG(_(e_internal)); /* Shouldn't happen */
|
|
4397 return FALSE;
|
|
4398 }
|
|
4399 }
|
|
4400 break;
|
|
4401
|
|
4402 case BRACE_COMPLEX + 0:
|
|
4403 case BRACE_COMPLEX + 1:
|
|
4404 case BRACE_COMPLEX + 2:
|
|
4405 case BRACE_COMPLEX + 3:
|
|
4406 case BRACE_COMPLEX + 4:
|
|
4407 case BRACE_COMPLEX + 5:
|
|
4408 case BRACE_COMPLEX + 6:
|
|
4409 case BRACE_COMPLEX + 7:
|
|
4410 case BRACE_COMPLEX + 8:
|
|
4411 case BRACE_COMPLEX + 9:
|
|
4412 {
|
|
4413 int no;
|
|
4414 regsave_T save;
|
|
4415
|
|
4416 no = op - BRACE_COMPLEX;
|
|
4417 ++brace_count[no];
|
|
4418
|
|
4419 /* If not matched enough times yet, try one more */
|
|
4420 if (brace_count[no] <= (brace_min[no] <= brace_max[no]
|
|
4421 ? brace_min[no] : brace_max[no]))
|
|
4422 {
|
|
4423 reg_save(&save);
|
167
|
4424 if (regmatch(OPERAND(scan), &save))
|
7
|
4425 return TRUE;
|
|
4426 reg_restore(&save);
|
|
4427 --brace_count[no]; /* failed, decrement match count */
|
|
4428 return FALSE;
|
|
4429 }
|
|
4430
|
|
4431 /* If matched enough times, may try matching some more */
|
|
4432 if (brace_min[no] <= brace_max[no])
|
|
4433 {
|
|
4434 /* Range is the normal way around, use longest match */
|
|
4435 if (brace_count[no] <= brace_max[no])
|
|
4436 {
|
|
4437 reg_save(&save);
|
167
|
4438 if (regmatch(OPERAND(scan), &save))
|
7
|
4439 return TRUE; /* matched some more times */
|
|
4440 reg_restore(&save);
|
|
4441 --brace_count[no]; /* matched just enough times */
|
167
|
4442 /* { continue with the items after \{} */
|
7
|
4443 }
|
|
4444 }
|
|
4445 else
|
|
4446 {
|
|
4447 /* Range is backwards, use shortest match first */
|
|
4448 if (brace_count[no] <= brace_min[no])
|
|
4449 {
|
|
4450 reg_save(&save);
|
167
|
4451 if (regmatch(next, &save))
|
7
|
4452 return TRUE;
|
|
4453 reg_restore(&save);
|
|
4454 next = OPERAND(scan);
|
|
4455 /* must try to match one more item */
|
|
4456 }
|
|
4457 }
|
|
4458 }
|
|
4459 break;
|
|
4460
|
|
4461 case BRACE_SIMPLE:
|
|
4462 case STAR:
|
|
4463 case PLUS:
|
|
4464 {
|
|
4465 int nextb; /* next byte */
|
|
4466 int nextb_ic; /* next byte reverse case */
|
|
4467 long count;
|
|
4468 regsave_T save;
|
|
4469 long minval;
|
|
4470 long maxval;
|
|
4471
|
|
4472 /*
|
|
4473 * Lookahead to avoid useless match attempts when we know
|
|
4474 * what character comes next.
|
|
4475 */
|
|
4476 if (OP(next) == EXACTLY)
|
|
4477 {
|
|
4478 nextb = *OPERAND(next);
|
|
4479 if (ireg_ic)
|
|
4480 {
|
|
4481 if (isupper(nextb))
|
|
4482 nextb_ic = TOLOWER_LOC(nextb);
|
|
4483 else
|
|
4484 nextb_ic = TOUPPER_LOC(nextb);
|
|
4485 }
|
|
4486 else
|
|
4487 nextb_ic = nextb;
|
|
4488 }
|
|
4489 else
|
|
4490 {
|
|
4491 nextb = NUL;
|
|
4492 nextb_ic = NUL;
|
|
4493 }
|
|
4494 if (op != BRACE_SIMPLE)
|
|
4495 {
|
|
4496 minval = (op == STAR) ? 0 : 1;
|
|
4497 maxval = MAX_LIMIT;
|
|
4498 }
|
|
4499 else
|
|
4500 {
|
|
4501 minval = bl_minval;
|
|
4502 maxval = bl_maxval;
|
|
4503 }
|
|
4504
|
|
4505 /*
|
|
4506 * When maxval > minval, try matching as much as possible, up
|
|
4507 * to maxval. When maxval < minval, try matching at least the
|
|
4508 * minimal number (since the range is backwards, that's also
|
|
4509 * maxval!).
|
|
4510 */
|
|
4511 count = regrepeat(OPERAND(scan), maxval);
|
|
4512 if (got_int)
|
|
4513 return FALSE;
|
|
4514 if (minval <= maxval)
|
|
4515 {
|
|
4516 /* Range is the normal way around, use longest match */
|
|
4517 while (count >= minval)
|
|
4518 {
|
|
4519 /* If it could match, try it. */
|
|
4520 if (nextb == NUL || *reginput == nextb
|
|
4521 || *reginput == nextb_ic)
|
|
4522 {
|
|
4523 reg_save(&save);
|
167
|
4524 if (regmatch(next, startp))
|
7
|
4525 return TRUE;
|
|
4526 reg_restore(&save);
|
|
4527 }
|
|
4528 /* Couldn't or didn't match -- back up one char. */
|
|
4529 if (--count < minval)
|
|
4530 break;
|
|
4531 if (reginput == regline)
|
|
4532 {
|
|
4533 /* backup to last char of previous line */
|
|
4534 --reglnum;
|
|
4535 regline = reg_getline(reglnum);
|
|
4536 /* Just in case regrepeat() didn't count right. */
|
|
4537 if (regline == NULL)
|
|
4538 return FALSE;
|
|
4539 reginput = regline + STRLEN(regline);
|
|
4540 fast_breakcheck();
|
|
4541 if (got_int || out_of_stack)
|
|
4542 return FALSE;
|
|
4543 }
|
|
4544 else
|
39
|
4545 mb_ptr_back(regline, reginput);
|
7
|
4546 }
|
|
4547 }
|
|
4548 else
|
|
4549 {
|
|
4550 /* Range is backwards, use shortest match first.
|
|
4551 * Careful: maxval and minval are exchanged! */
|
|
4552 if (count < maxval)
|
|
4553 return FALSE;
|
|
4554 for (;;)
|
|
4555 {
|
|
4556 /* If it could work, try it. */
|
|
4557 if (nextb == NUL || *reginput == nextb
|
|
4558 || *reginput == nextb_ic)
|
|
4559 {
|
|
4560 reg_save(&save);
|
167
|
4561 if (regmatch(next, &save))
|
7
|
4562 return TRUE;
|
|
4563 reg_restore(&save);
|
|
4564 }
|
|
4565 /* Couldn't or didn't match: try advancing one char. */
|
|
4566 if (count == minval
|
|
4567 || regrepeat(OPERAND(scan), 1L) == 0)
|
|
4568 break;
|
|
4569 ++count;
|
|
4570 if (got_int || out_of_stack)
|
|
4571 return FALSE;
|
|
4572 }
|
|
4573 }
|
|
4574 return FALSE;
|
|
4575 }
|
|
4576 /* break; Not Reached */
|
|
4577
|
|
4578 case NOMATCH:
|
|
4579 {
|
|
4580 regsave_T save;
|
|
4581
|
|
4582 /* If the operand matches, we fail. Otherwise backup and
|
|
4583 * continue with the next item. */
|
|
4584 reg_save(&save);
|
167
|
4585 if (regmatch(OPERAND(scan), startp))
|
7
|
4586 return FALSE;
|
|
4587 reg_restore(&save);
|
|
4588 }
|
|
4589 break;
|
|
4590
|
|
4591 case MATCH:
|
|
4592 case SUBPAT:
|
|
4593 {
|
|
4594 regsave_T save;
|
|
4595
|
|
4596 /* If the operand doesn't match, we fail. Otherwise backup
|
|
4597 * and continue with the next item. */
|
|
4598 reg_save(&save);
|
167
|
4599 if (!regmatch(OPERAND(scan), startp))
|
7
|
4600 return FALSE;
|
|
4601 if (op == MATCH) /* zero-width */
|
|
4602 reg_restore(&save);
|
|
4603 }
|
|
4604 break;
|
|
4605
|
|
4606 case BEHIND:
|
|
4607 case NOBEHIND:
|
|
4608 {
|
|
4609 regsave_T save_after, save_start;
|
|
4610 regsave_T save_behind_pos;
|
|
4611 int needmatch = (op == BEHIND);
|
|
4612
|
|
4613 /*
|
|
4614 * Look back in the input of the operand matches or not. This
|
|
4615 * must be done at every position in the input and checking if
|
|
4616 * the match ends at the current position.
|
|
4617 * First check if the next item matches, that's probably
|
|
4618 * faster.
|
|
4619 */
|
|
4620 reg_save(&save_start);
|
167
|
4621 if (regmatch(next, startp))
|
7
|
4622 {
|
|
4623 /* save the position after the found match for next */
|
|
4624 reg_save(&save_after);
|
|
4625
|
|
4626 /* start looking for a match with operand at the current
|
|
4627 * postion. Go back one character until we find the
|
|
4628 * result, hitting the start of the line or the previous
|
|
4629 * line (for multi-line matching).
|
|
4630 * Set behind_pos to where the match should end, BHPOS
|
|
4631 * will match it. */
|
|
4632 save_behind_pos = behind_pos;
|
|
4633 behind_pos = save_start;
|
|
4634 for (;;)
|
|
4635 {
|
|
4636 reg_restore(&save_start);
|
167
|
4637 if (regmatch(OPERAND(scan), startp)
|
7
|
4638 && reg_save_equal(&behind_pos))
|
|
4639 {
|
|
4640 behind_pos = save_behind_pos;
|
|
4641 /* found a match that ends where "next" started */
|
|
4642 if (needmatch)
|
|
4643 {
|
|
4644 reg_restore(&save_after);
|
|
4645 return TRUE;
|
|
4646 }
|
|
4647 return FALSE;
|
|
4648 }
|
|
4649 /*
|
|
4650 * No match: Go back one character. May go to
|
|
4651 * previous line once.
|
|
4652 */
|
|
4653 if (REG_MULTI)
|
|
4654 {
|
|
4655 if (save_start.rs_u.pos.col == 0)
|
|
4656 {
|
|
4657 if (save_start.rs_u.pos.lnum
|
|
4658 < behind_pos.rs_u.pos.lnum
|
|
4659 || reg_getline(
|
|
4660 --save_start.rs_u.pos.lnum) == NULL)
|
|
4661 break;
|
|
4662 reg_restore(&save_start);
|
|
4663 save_start.rs_u.pos.col =
|
|
4664 (colnr_T)STRLEN(regline);
|
|
4665 }
|
|
4666 else
|
|
4667 --save_start.rs_u.pos.col;
|
|
4668 }
|
|
4669 else
|
|
4670 {
|
|
4671 if (save_start.rs_u.ptr == regline)
|
|
4672 break;
|
|
4673 --save_start.rs_u.ptr;
|
|
4674 }
|
|
4675 }
|
|
4676
|
|
4677 /* NOBEHIND succeeds when no match was found */
|
|
4678 behind_pos = save_behind_pos;
|
|
4679 if (!needmatch)
|
|
4680 {
|
|
4681 reg_restore(&save_after);
|
|
4682 return TRUE;
|
|
4683 }
|
|
4684 }
|
|
4685 return FALSE;
|
|
4686 }
|
|
4687
|
|
4688 case BHPOS:
|
|
4689 if (REG_MULTI)
|
|
4690 {
|
|
4691 if (behind_pos.rs_u.pos.col != (colnr_T)(reginput - regline)
|
|
4692 || behind_pos.rs_u.pos.lnum != reglnum)
|
|
4693 return FALSE;
|
|
4694 }
|
|
4695 else if (behind_pos.rs_u.ptr != reginput)
|
|
4696 return FALSE;
|
|
4697 break;
|
|
4698
|
|
4699 case NEWL:
|
|
4700 if ((c != NUL || reglnum == reg_maxline)
|
|
4701 && (c != '\n' || !reg_line_lbr))
|
|
4702 return FALSE;
|
|
4703 if (reg_line_lbr)
|
|
4704 ADVANCE_REGINPUT();
|
|
4705 else
|
|
4706 reg_nextline();
|
|
4707 break;
|
|
4708
|
|
4709 case END:
|
|
4710 return TRUE; /* Success! */
|
|
4711
|
|
4712 default:
|
|
4713 EMSG(_(e_re_corr));
|
|
4714 #ifdef DEBUG
|
|
4715 printf("Illegal op code %d\n", op);
|
|
4716 #endif
|
|
4717 return FALSE;
|
|
4718 }
|
|
4719 }
|
|
4720
|
|
4721 scan = next;
|
|
4722 }
|
|
4723
|
|
4724 /*
|
|
4725 * We get here only if there's trouble -- normally "case END" is the
|
|
4726 * terminating point.
|
|
4727 */
|
|
4728 EMSG(_(e_re_corr));
|
|
4729 #ifdef DEBUG
|
|
4730 printf("Premature EOL\n");
|
|
4731 #endif
|
|
4732 return FALSE;
|
|
4733 }
|
|
4734
|
|
4735 /*
|
|
4736 * regrepeat - repeatedly match something simple, return how many.
|
|
4737 * Advances reginput (and reglnum) to just after the matched chars.
|
|
4738 */
|
|
4739 static int
|
|
4740 regrepeat(p, maxcount)
|
|
4741 char_u *p;
|
|
4742 long maxcount; /* maximum number of matches allowed */
|
|
4743 {
|
|
4744 long count = 0;
|
|
4745 char_u *scan;
|
|
4746 char_u *opnd;
|
|
4747 int mask;
|
|
4748 int testval = 0;
|
|
4749
|
|
4750 scan = reginput; /* Make local copy of reginput for speed. */
|
|
4751 opnd = OPERAND(p);
|
|
4752 switch (OP(p))
|
|
4753 {
|
|
4754 case ANY:
|
|
4755 case ANY + ADD_NL:
|
|
4756 while (count < maxcount)
|
|
4757 {
|
|
4758 /* Matching anything means we continue until end-of-line (or
|
|
4759 * end-of-file for ANY + ADD_NL), only limited by maxcount. */
|
|
4760 while (*scan != NUL && count < maxcount)
|
|
4761 {
|
|
4762 ++count;
|
39
|
4763 mb_ptr_adv(scan);
|
7
|
4764 }
|
|
4765 if (!WITH_NL(OP(p)) || reglnum == reg_maxline || count == maxcount)
|
|
4766 break;
|
|
4767 ++count; /* count the line-break */
|
|
4768 reg_nextline();
|
|
4769 scan = reginput;
|
|
4770 if (got_int)
|
|
4771 break;
|
|
4772 }
|
|
4773 break;
|
|
4774
|
|
4775 case IDENT:
|
|
4776 case IDENT + ADD_NL:
|
|
4777 testval = TRUE;
|
|
4778 /*FALLTHROUGH*/
|
|
4779 case SIDENT:
|
|
4780 case SIDENT + ADD_NL:
|
|
4781 while (count < maxcount)
|
|
4782 {
|
|
4783 if (vim_isIDc(*scan) && (testval || !VIM_ISDIGIT(*scan)))
|
|
4784 {
|
39
|
4785 mb_ptr_adv(scan);
|
7
|
4786 }
|
|
4787 else if (*scan == NUL)
|
|
4788 {
|
|
4789 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
4790 break;
|
|
4791 reg_nextline();
|
|
4792 scan = reginput;
|
|
4793 if (got_int)
|
|
4794 break;
|
|
4795 }
|
|
4796 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
4797 ++scan;
|
|
4798 else
|
|
4799 break;
|
|
4800 ++count;
|
|
4801 }
|
|
4802 break;
|
|
4803
|
|
4804 case KWORD:
|
|
4805 case KWORD + ADD_NL:
|
|
4806 testval = TRUE;
|
|
4807 /*FALLTHROUGH*/
|
|
4808 case SKWORD:
|
|
4809 case SKWORD + ADD_NL:
|
|
4810 while (count < maxcount)
|
|
4811 {
|
|
4812 if (vim_iswordp(scan) && (testval || !VIM_ISDIGIT(*scan)))
|
|
4813 {
|
39
|
4814 mb_ptr_adv(scan);
|
7
|
4815 }
|
|
4816 else if (*scan == NUL)
|
|
4817 {
|
|
4818 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
4819 break;
|
|
4820 reg_nextline();
|
|
4821 scan = reginput;
|
|
4822 if (got_int)
|
|
4823 break;
|
|
4824 }
|
|
4825 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
4826 ++scan;
|
|
4827 else
|
|
4828 break;
|
|
4829 ++count;
|
|
4830 }
|
|
4831 break;
|
|
4832
|
|
4833 case FNAME:
|
|
4834 case FNAME + ADD_NL:
|
|
4835 testval = TRUE;
|
|
4836 /*FALLTHROUGH*/
|
|
4837 case SFNAME:
|
|
4838 case SFNAME + ADD_NL:
|
|
4839 while (count < maxcount)
|
|
4840 {
|
|
4841 if (vim_isfilec(*scan) && (testval || !VIM_ISDIGIT(*scan)))
|
|
4842 {
|
39
|
4843 mb_ptr_adv(scan);
|
7
|
4844 }
|
|
4845 else if (*scan == NUL)
|
|
4846 {
|
|
4847 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
4848 break;
|
|
4849 reg_nextline();
|
|
4850 scan = reginput;
|
|
4851 if (got_int)
|
|
4852 break;
|
|
4853 }
|
|
4854 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
4855 ++scan;
|
|
4856 else
|
|
4857 break;
|
|
4858 ++count;
|
|
4859 }
|
|
4860 break;
|
|
4861
|
|
4862 case PRINT:
|
|
4863 case PRINT + ADD_NL:
|
|
4864 testval = TRUE;
|
|
4865 /*FALLTHROUGH*/
|
|
4866 case SPRINT:
|
|
4867 case SPRINT + ADD_NL:
|
|
4868 while (count < maxcount)
|
|
4869 {
|
|
4870 if (*scan == NUL)
|
|
4871 {
|
|
4872 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
4873 break;
|
|
4874 reg_nextline();
|
|
4875 scan = reginput;
|
|
4876 if (got_int)
|
|
4877 break;
|
|
4878 }
|
|
4879 else if (ptr2cells(scan) == 1 && (testval || !VIM_ISDIGIT(*scan)))
|
|
4880 {
|
39
|
4881 mb_ptr_adv(scan);
|
7
|
4882 }
|
|
4883 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
4884 ++scan;
|
|
4885 else
|
|
4886 break;
|
|
4887 ++count;
|
|
4888 }
|
|
4889 break;
|
|
4890
|
|
4891 case WHITE:
|
|
4892 case WHITE + ADD_NL:
|
|
4893 testval = mask = RI_WHITE;
|
|
4894 do_class:
|
|
4895 while (count < maxcount)
|
|
4896 {
|
|
4897 #ifdef FEAT_MBYTE
|
|
4898 int l;
|
|
4899 #endif
|
|
4900 if (*scan == NUL)
|
|
4901 {
|
|
4902 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
4903 break;
|
|
4904 reg_nextline();
|
|
4905 scan = reginput;
|
|
4906 if (got_int)
|
|
4907 break;
|
|
4908 }
|
|
4909 #ifdef FEAT_MBYTE
|
|
4910 else if (has_mbyte && (l = (*mb_ptr2len_check)(scan)) > 1)
|
|
4911 {
|
|
4912 if (testval != 0)
|
|
4913 break;
|
|
4914 scan += l;
|
|
4915 }
|
|
4916 #endif
|
|
4917 else if ((class_tab[*scan] & mask) == testval)
|
|
4918 ++scan;
|
|
4919 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
4920 ++scan;
|
|
4921 else
|
|
4922 break;
|
|
4923 ++count;
|
|
4924 }
|
|
4925 break;
|
|
4926
|
|
4927 case NWHITE:
|
|
4928 case NWHITE + ADD_NL:
|
|
4929 mask = RI_WHITE;
|
|
4930 goto do_class;
|
|
4931 case DIGIT:
|
|
4932 case DIGIT + ADD_NL:
|
|
4933 testval = mask = RI_DIGIT;
|
|
4934 goto do_class;
|
|
4935 case NDIGIT:
|
|
4936 case NDIGIT + ADD_NL:
|
|
4937 mask = RI_DIGIT;
|
|
4938 goto do_class;
|
|
4939 case HEX:
|
|
4940 case HEX + ADD_NL:
|
|
4941 testval = mask = RI_HEX;
|
|
4942 goto do_class;
|
|
4943 case NHEX:
|
|
4944 case NHEX + ADD_NL:
|
|
4945 mask = RI_HEX;
|
|
4946 goto do_class;
|
|
4947 case OCTAL:
|
|
4948 case OCTAL + ADD_NL:
|
|
4949 testval = mask = RI_OCTAL;
|
|
4950 goto do_class;
|
|
4951 case NOCTAL:
|
|
4952 case NOCTAL + ADD_NL:
|
|
4953 mask = RI_OCTAL;
|
|
4954 goto do_class;
|
|
4955 case WORD:
|
|
4956 case WORD + ADD_NL:
|
|
4957 testval = mask = RI_WORD;
|
|
4958 goto do_class;
|
|
4959 case NWORD:
|
|
4960 case NWORD + ADD_NL:
|
|
4961 mask = RI_WORD;
|
|
4962 goto do_class;
|
|
4963 case HEAD:
|
|
4964 case HEAD + ADD_NL:
|
|
4965 testval = mask = RI_HEAD;
|
|
4966 goto do_class;
|
|
4967 case NHEAD:
|
|
4968 case NHEAD + ADD_NL:
|
|
4969 mask = RI_HEAD;
|
|
4970 goto do_class;
|
|
4971 case ALPHA:
|
|
4972 case ALPHA + ADD_NL:
|
|
4973 testval = mask = RI_ALPHA;
|
|
4974 goto do_class;
|
|
4975 case NALPHA:
|
|
4976 case NALPHA + ADD_NL:
|
|
4977 mask = RI_ALPHA;
|
|
4978 goto do_class;
|
|
4979 case LOWER:
|
|
4980 case LOWER + ADD_NL:
|
|
4981 testval = mask = RI_LOWER;
|
|
4982 goto do_class;
|
|
4983 case NLOWER:
|
|
4984 case NLOWER + ADD_NL:
|
|
4985 mask = RI_LOWER;
|
|
4986 goto do_class;
|
|
4987 case UPPER:
|
|
4988 case UPPER + ADD_NL:
|
|
4989 testval = mask = RI_UPPER;
|
|
4990 goto do_class;
|
|
4991 case NUPPER:
|
|
4992 case NUPPER + ADD_NL:
|
|
4993 mask = RI_UPPER;
|
|
4994 goto do_class;
|
|
4995
|
|
4996 case EXACTLY:
|
|
4997 {
|
|
4998 int cu, cl;
|
|
4999
|
|
5000 /* This doesn't do a multi-byte character, because a MULTIBYTECODE
|
|
5001 * would have been used for it. */
|
|
5002 if (ireg_ic)
|
|
5003 {
|
|
5004 cu = TOUPPER_LOC(*opnd);
|
|
5005 cl = TOLOWER_LOC(*opnd);
|
|
5006 while (count < maxcount && (*scan == cu || *scan == cl))
|
|
5007 {
|
|
5008 count++;
|
|
5009 scan++;
|
|
5010 }
|
|
5011 }
|
|
5012 else
|
|
5013 {
|
|
5014 cu = *opnd;
|
|
5015 while (count < maxcount && *scan == cu)
|
|
5016 {
|
|
5017 count++;
|
|
5018 scan++;
|
|
5019 }
|
|
5020 }
|
|
5021 break;
|
|
5022 }
|
|
5023
|
|
5024 #ifdef FEAT_MBYTE
|
|
5025 case MULTIBYTECODE:
|
|
5026 {
|
|
5027 int i, len, cf = 0;
|
|
5028
|
|
5029 /* Safety check (just in case 'encoding' was changed since
|
|
5030 * compiling the program). */
|
|
5031 if ((len = (*mb_ptr2len_check)(opnd)) > 1)
|
|
5032 {
|
|
5033 if (ireg_ic && enc_utf8)
|
|
5034 cf = utf_fold(utf_ptr2char(opnd));
|
|
5035 while (count < maxcount)
|
|
5036 {
|
|
5037 for (i = 0; i < len; ++i)
|
|
5038 if (opnd[i] != scan[i])
|
|
5039 break;
|
|
5040 if (i < len && (!ireg_ic || !enc_utf8
|
|
5041 || utf_fold(utf_ptr2char(scan)) != cf))
|
|
5042 break;
|
|
5043 scan += len;
|
|
5044 ++count;
|
|
5045 }
|
|
5046 }
|
|
5047 }
|
|
5048 break;
|
|
5049 #endif
|
|
5050
|
|
5051 case ANYOF:
|
|
5052 case ANYOF + ADD_NL:
|
|
5053 testval = TRUE;
|
|
5054 /*FALLTHROUGH*/
|
|
5055
|
|
5056 case ANYBUT:
|
|
5057 case ANYBUT + ADD_NL:
|
|
5058 while (count < maxcount)
|
|
5059 {
|
|
5060 #ifdef FEAT_MBYTE
|
|
5061 int len;
|
|
5062 #endif
|
|
5063 if (*scan == NUL)
|
|
5064 {
|
|
5065 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
5066 break;
|
|
5067 reg_nextline();
|
|
5068 scan = reginput;
|
|
5069 if (got_int)
|
|
5070 break;
|
|
5071 }
|
|
5072 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
5073 ++scan;
|
|
5074 #ifdef FEAT_MBYTE
|
|
5075 else if (has_mbyte && (len = (*mb_ptr2len_check)(scan)) > 1)
|
|
5076 {
|
|
5077 if ((cstrchr(opnd, (*mb_ptr2char)(scan)) == NULL) == testval)
|
|
5078 break;
|
|
5079 scan += len;
|
|
5080 }
|
|
5081 #endif
|
|
5082 else
|
|
5083 {
|
|
5084 if ((cstrchr(opnd, *scan) == NULL) == testval)
|
|
5085 break;
|
|
5086 ++scan;
|
|
5087 }
|
|
5088 ++count;
|
|
5089 }
|
|
5090 break;
|
|
5091
|
|
5092 case NEWL:
|
|
5093 while (count < maxcount
|
|
5094 && ((*scan == NUL && reglnum < reg_maxline)
|
|
5095 || (*scan == '\n' && reg_line_lbr)))
|
|
5096 {
|
|
5097 count++;
|
|
5098 if (reg_line_lbr)
|
|
5099 ADVANCE_REGINPUT();
|
|
5100 else
|
|
5101 reg_nextline();
|
|
5102 scan = reginput;
|
|
5103 if (got_int)
|
|
5104 break;
|
|
5105 }
|
|
5106 break;
|
|
5107
|
|
5108 default: /* Oh dear. Called inappropriately. */
|
|
5109 EMSG(_(e_re_corr));
|
|
5110 #ifdef DEBUG
|
|
5111 printf("Called regrepeat with op code %d\n", OP(p));
|
|
5112 #endif
|
|
5113 break;
|
|
5114 }
|
|
5115
|
|
5116 reginput = scan;
|
|
5117
|
|
5118 return (int)count;
|
|
5119 }
|
|
5120
|
|
5121 /*
|
|
5122 * regnext - dig the "next" pointer out of a node
|
|
5123 */
|
|
5124 static char_u *
|
|
5125 regnext(p)
|
|
5126 char_u *p;
|
|
5127 {
|
|
5128 int offset;
|
|
5129
|
|
5130 if (p == JUST_CALC_SIZE)
|
|
5131 return NULL;
|
|
5132
|
|
5133 offset = NEXT(p);
|
|
5134 if (offset == 0)
|
|
5135 return NULL;
|
|
5136
|
|
5137 if (OP(p) == BACK)
|
|
5138 return p - offset;
|
|
5139 else
|
|
5140 return p + offset;
|
|
5141 }
|
|
5142
|
|
5143 /*
|
|
5144 * Check the regexp program for its magic number.
|
|
5145 * Return TRUE if it's wrong.
|
|
5146 */
|
|
5147 static int
|
|
5148 prog_magic_wrong()
|
|
5149 {
|
|
5150 if (UCHARAT(REG_MULTI
|
|
5151 ? reg_mmatch->regprog->program
|
|
5152 : reg_match->regprog->program) != REGMAGIC)
|
|
5153 {
|
|
5154 EMSG(_(e_re_corr));
|
|
5155 return TRUE;
|
|
5156 }
|
|
5157 return FALSE;
|
|
5158 }
|
|
5159
|
|
5160 /*
|
|
5161 * Cleanup the subexpressions, if this wasn't done yet.
|
|
5162 * This construction is used to clear the subexpressions only when they are
|
|
5163 * used (to increase speed).
|
|
5164 */
|
|
5165 static void
|
|
5166 cleanup_subexpr()
|
|
5167 {
|
|
5168 if (need_clear_subexpr)
|
|
5169 {
|
|
5170 if (REG_MULTI)
|
|
5171 {
|
|
5172 /* Use 0xff to set lnum to -1 */
|
|
5173 vim_memset(reg_startpos, 0xff, sizeof(lpos_T) * NSUBEXP);
|
|
5174 vim_memset(reg_endpos, 0xff, sizeof(lpos_T) * NSUBEXP);
|
|
5175 }
|
|
5176 else
|
|
5177 {
|
|
5178 vim_memset(reg_startp, 0, sizeof(char_u *) * NSUBEXP);
|
|
5179 vim_memset(reg_endp, 0, sizeof(char_u *) * NSUBEXP);
|
|
5180 }
|
|
5181 need_clear_subexpr = FALSE;
|
|
5182 }
|
|
5183 }
|
|
5184
|
|
5185 #ifdef FEAT_SYN_HL
|
|
5186 static void
|
|
5187 cleanup_zsubexpr()
|
|
5188 {
|
|
5189 if (need_clear_zsubexpr)
|
|
5190 {
|
|
5191 if (REG_MULTI)
|
|
5192 {
|
|
5193 /* Use 0xff to set lnum to -1 */
|
|
5194 vim_memset(reg_startzpos, 0xff, sizeof(lpos_T) * NSUBEXP);
|
|
5195 vim_memset(reg_endzpos, 0xff, sizeof(lpos_T) * NSUBEXP);
|
|
5196 }
|
|
5197 else
|
|
5198 {
|
|
5199 vim_memset(reg_startzp, 0, sizeof(char_u *) * NSUBEXP);
|
|
5200 vim_memset(reg_endzp, 0, sizeof(char_u *) * NSUBEXP);
|
|
5201 }
|
|
5202 need_clear_zsubexpr = FALSE;
|
|
5203 }
|
|
5204 }
|
|
5205 #endif
|
|
5206
|
|
5207 /*
|
|
5208 * Advance reglnum, regline and reginput to the next line.
|
|
5209 */
|
|
5210 static void
|
|
5211 reg_nextline()
|
|
5212 {
|
|
5213 regline = reg_getline(++reglnum);
|
|
5214 reginput = regline;
|
|
5215 fast_breakcheck();
|
|
5216 }
|
|
5217
|
|
5218 /*
|
|
5219 * Save the input line and position in a regsave_T.
|
|
5220 */
|
|
5221 static void
|
|
5222 reg_save(save)
|
|
5223 regsave_T *save;
|
|
5224 {
|
|
5225 if (REG_MULTI)
|
|
5226 {
|
|
5227 save->rs_u.pos.col = (colnr_T)(reginput - regline);
|
|
5228 save->rs_u.pos.lnum = reglnum;
|
|
5229 }
|
|
5230 else
|
|
5231 save->rs_u.ptr = reginput;
|
|
5232 }
|
|
5233
|
|
5234 /*
|
|
5235 * Restore the input line and position from a regsave_T.
|
|
5236 */
|
|
5237 static void
|
|
5238 reg_restore(save)
|
|
5239 regsave_T *save;
|
|
5240 {
|
|
5241 if (REG_MULTI)
|
|
5242 {
|
|
5243 if (reglnum != save->rs_u.pos.lnum)
|
|
5244 {
|
|
5245 /* only call reg_getline() when the line number changed to save
|
|
5246 * a bit of time */
|
|
5247 reglnum = save->rs_u.pos.lnum;
|
|
5248 regline = reg_getline(reglnum);
|
|
5249 }
|
|
5250 reginput = regline + save->rs_u.pos.col;
|
|
5251 }
|
|
5252 else
|
|
5253 reginput = save->rs_u.ptr;
|
|
5254 }
|
|
5255
|
|
5256 /*
|
|
5257 * Return TRUE if current position is equal to saved position.
|
|
5258 */
|
|
5259 static int
|
|
5260 reg_save_equal(save)
|
|
5261 regsave_T *save;
|
|
5262 {
|
|
5263 if (REG_MULTI)
|
|
5264 return reglnum == save->rs_u.pos.lnum
|
|
5265 && reginput == regline + save->rs_u.pos.col;
|
|
5266 return reginput == save->rs_u.ptr;
|
|
5267 }
|
|
5268
|
|
5269 /*
|
|
5270 * Tentatively set the sub-expression start to the current position (after
|
|
5271 * calling regmatch() they will have changed). Need to save the existing
|
|
5272 * values for when there is no match.
|
|
5273 * Use se_save() to use pointer (save_se_multi()) or position (save_se_one()),
|
|
5274 * depending on REG_MULTI.
|
|
5275 */
|
|
5276 static void
|
|
5277 save_se_multi(savep, posp)
|
|
5278 save_se_T *savep;
|
|
5279 lpos_T *posp;
|
|
5280 {
|
|
5281 savep->se_u.pos = *posp;
|
|
5282 posp->lnum = reglnum;
|
|
5283 posp->col = (colnr_T)(reginput - regline);
|
|
5284 }
|
|
5285
|
|
5286 static void
|
|
5287 save_se_one(savep, pp)
|
|
5288 save_se_T *savep;
|
|
5289 char_u **pp;
|
|
5290 {
|
|
5291 savep->se_u.ptr = *pp;
|
|
5292 *pp = reginput;
|
|
5293 }
|
|
5294
|
|
5295 /*
|
|
5296 * Compare a number with the operand of RE_LNUM, RE_COL or RE_VCOL.
|
|
5297 */
|
|
5298 static int
|
|
5299 re_num_cmp(val, scan)
|
|
5300 long_u val;
|
|
5301 char_u *scan;
|
|
5302 {
|
|
5303 long_u n = OPERAND_MIN(scan);
|
|
5304
|
|
5305 if (OPERAND_CMP(scan) == '>')
|
|
5306 return val > n;
|
|
5307 if (OPERAND_CMP(scan) == '<')
|
|
5308 return val < n;
|
|
5309 return val == n;
|
|
5310 }
|
|
5311
|
|
5312
|
|
5313 #ifdef DEBUG
|
|
5314
|
|
5315 /*
|
|
5316 * regdump - dump a regexp onto stdout in vaguely comprehensible form
|
|
5317 */
|
|
5318 static void
|
|
5319 regdump(pattern, r)
|
|
5320 char_u *pattern;
|
|
5321 regprog_T *r;
|
|
5322 {
|
|
5323 char_u *s;
|
|
5324 int op = EXACTLY; /* Arbitrary non-END op. */
|
|
5325 char_u *next;
|
|
5326 char_u *end = NULL;
|
|
5327
|
|
5328 printf("\r\nregcomp(%s):\r\n", pattern);
|
|
5329
|
|
5330 s = r->program + 1;
|
|
5331 /*
|
|
5332 * Loop until we find the END that isn't before a referred next (an END
|
|
5333 * can also appear in a NOMATCH operand).
|
|
5334 */
|
|
5335 while (op != END || s <= end)
|
|
5336 {
|
|
5337 op = OP(s);
|
|
5338 printf("%2d%s", (int)(s - r->program), regprop(s)); /* Where, what. */
|
|
5339 next = regnext(s);
|
|
5340 if (next == NULL) /* Next ptr. */
|
|
5341 printf("(0)");
|
|
5342 else
|
|
5343 printf("(%d)", (int)((s - r->program) + (next - s)));
|
|
5344 if (end < next)
|
|
5345 end = next;
|
|
5346 if (op == BRACE_LIMITS)
|
|
5347 {
|
|
5348 /* Two short ints */
|
|
5349 printf(" minval %ld, maxval %ld", OPERAND_MIN(s), OPERAND_MAX(s));
|
|
5350 s += 8;
|
|
5351 }
|
|
5352 s += 3;
|
|
5353 if (op == ANYOF || op == ANYOF + ADD_NL
|
|
5354 || op == ANYBUT || op == ANYBUT + ADD_NL
|
|
5355 || op == EXACTLY)
|
|
5356 {
|
|
5357 /* Literal string, where present. */
|
|
5358 while (*s != NUL)
|
|
5359 printf("%c", *s++);
|
|
5360 s++;
|
|
5361 }
|
|
5362 printf("\r\n");
|
|
5363 }
|
|
5364
|
|
5365 /* Header fields of interest. */
|
|
5366 if (r->regstart != NUL)
|
|
5367 printf("start `%s' 0x%x; ", r->regstart < 256
|
|
5368 ? (char *)transchar(r->regstart)
|
|
5369 : "multibyte", r->regstart);
|
|
5370 if (r->reganch)
|
|
5371 printf("anchored; ");
|
|
5372 if (r->regmust != NULL)
|
|
5373 printf("must have \"%s\"", r->regmust);
|
|
5374 printf("\r\n");
|
|
5375 }
|
|
5376
|
|
5377 /*
|
|
5378 * regprop - printable representation of opcode
|
|
5379 */
|
|
5380 static char_u *
|
|
5381 regprop(op)
|
|
5382 char_u *op;
|
|
5383 {
|
|
5384 char_u *p;
|
|
5385 static char_u buf[50];
|
|
5386
|
|
5387 (void) strcpy(buf, ":");
|
|
5388
|
|
5389 switch (OP(op))
|
|
5390 {
|
|
5391 case BOL:
|
|
5392 p = "BOL";
|
|
5393 break;
|
|
5394 case EOL:
|
|
5395 p = "EOL";
|
|
5396 break;
|
|
5397 case RE_BOF:
|
|
5398 p = "BOF";
|
|
5399 break;
|
|
5400 case RE_EOF:
|
|
5401 p = "EOF";
|
|
5402 break;
|
|
5403 case CURSOR:
|
|
5404 p = "CURSOR";
|
|
5405 break;
|
|
5406 case RE_LNUM:
|
|
5407 p = "RE_LNUM";
|
|
5408 break;
|
|
5409 case RE_COL:
|
|
5410 p = "RE_COL";
|
|
5411 break;
|
|
5412 case RE_VCOL:
|
|
5413 p = "RE_VCOL";
|
|
5414 break;
|
|
5415 case BOW:
|
|
5416 p = "BOW";
|
|
5417 break;
|
|
5418 case EOW:
|
|
5419 p = "EOW";
|
|
5420 break;
|
|
5421 case ANY:
|
|
5422 p = "ANY";
|
|
5423 break;
|
|
5424 case ANY + ADD_NL:
|
|
5425 p = "ANY+NL";
|
|
5426 break;
|
|
5427 case ANYOF:
|
|
5428 p = "ANYOF";
|
|
5429 break;
|
|
5430 case ANYOF + ADD_NL:
|
|
5431 p = "ANYOF+NL";
|
|
5432 break;
|
|
5433 case ANYBUT:
|
|
5434 p = "ANYBUT";
|
|
5435 break;
|
|
5436 case ANYBUT + ADD_NL:
|
|
5437 p = "ANYBUT+NL";
|
|
5438 break;
|
|
5439 case IDENT:
|
|
5440 p = "IDENT";
|
|
5441 break;
|
|
5442 case IDENT + ADD_NL:
|
|
5443 p = "IDENT+NL";
|
|
5444 break;
|
|
5445 case SIDENT:
|
|
5446 p = "SIDENT";
|
|
5447 break;
|
|
5448 case SIDENT + ADD_NL:
|
|
5449 p = "SIDENT+NL";
|
|
5450 break;
|
|
5451 case KWORD:
|
|
5452 p = "KWORD";
|
|
5453 break;
|
|
5454 case KWORD + ADD_NL:
|
|
5455 p = "KWORD+NL";
|
|
5456 break;
|
|
5457 case SKWORD:
|
|
5458 p = "SKWORD";
|
|
5459 break;
|
|
5460 case SKWORD + ADD_NL:
|
|
5461 p = "SKWORD+NL";
|
|
5462 break;
|
|
5463 case FNAME:
|
|
5464 p = "FNAME";
|
|
5465 break;
|
|
5466 case FNAME + ADD_NL:
|
|
5467 p = "FNAME+NL";
|
|
5468 break;
|
|
5469 case SFNAME:
|
|
5470 p = "SFNAME";
|
|
5471 break;
|
|
5472 case SFNAME + ADD_NL:
|
|
5473 p = "SFNAME+NL";
|
|
5474 break;
|
|
5475 case PRINT:
|
|
5476 p = "PRINT";
|
|
5477 break;
|
|
5478 case PRINT + ADD_NL:
|
|
5479 p = "PRINT+NL";
|
|
5480 break;
|
|
5481 case SPRINT:
|
|
5482 p = "SPRINT";
|
|
5483 break;
|
|
5484 case SPRINT + ADD_NL:
|
|
5485 p = "SPRINT+NL";
|
|
5486 break;
|
|
5487 case WHITE:
|
|
5488 p = "WHITE";
|
|
5489 break;
|
|
5490 case WHITE + ADD_NL:
|
|
5491 p = "WHITE+NL";
|
|
5492 break;
|
|
5493 case NWHITE:
|
|
5494 p = "NWHITE";
|
|
5495 break;
|
|
5496 case NWHITE + ADD_NL:
|
|
5497 p = "NWHITE+NL";
|
|
5498 break;
|
|
5499 case DIGIT:
|
|
5500 p = "DIGIT";
|
|
5501 break;
|
|
5502 case DIGIT + ADD_NL:
|
|
5503 p = "DIGIT+NL";
|
|
5504 break;
|
|
5505 case NDIGIT:
|
|
5506 p = "NDIGIT";
|
|
5507 break;
|
|
5508 case NDIGIT + ADD_NL:
|
|
5509 p = "NDIGIT+NL";
|
|
5510 break;
|
|
5511 case HEX:
|
|
5512 p = "HEX";
|
|
5513 break;
|
|
5514 case HEX + ADD_NL:
|
|
5515 p = "HEX+NL";
|
|
5516 break;
|
|
5517 case NHEX:
|
|
5518 p = "NHEX";
|
|
5519 break;
|
|
5520 case NHEX + ADD_NL:
|
|
5521 p = "NHEX+NL";
|
|
5522 break;
|
|
5523 case OCTAL:
|
|
5524 p = "OCTAL";
|
|
5525 break;
|
|
5526 case OCTAL + ADD_NL:
|
|
5527 p = "OCTAL+NL";
|
|
5528 break;
|
|
5529 case NOCTAL:
|
|
5530 p = "NOCTAL";
|
|
5531 break;
|
|
5532 case NOCTAL + ADD_NL:
|
|
5533 p = "NOCTAL+NL";
|
|
5534 break;
|
|
5535 case WORD:
|
|
5536 p = "WORD";
|
|
5537 break;
|
|
5538 case WORD + ADD_NL:
|
|
5539 p = "WORD+NL";
|
|
5540 break;
|
|
5541 case NWORD:
|
|
5542 p = "NWORD";
|
|
5543 break;
|
|
5544 case NWORD + ADD_NL:
|
|
5545 p = "NWORD+NL";
|
|
5546 break;
|
|
5547 case HEAD:
|
|
5548 p = "HEAD";
|
|
5549 break;
|
|
5550 case HEAD + ADD_NL:
|
|
5551 p = "HEAD+NL";
|
|
5552 break;
|
|
5553 case NHEAD:
|
|
5554 p = "NHEAD";
|
|
5555 break;
|
|
5556 case NHEAD + ADD_NL:
|
|
5557 p = "NHEAD+NL";
|
|
5558 break;
|
|
5559 case ALPHA:
|
|
5560 p = "ALPHA";
|
|
5561 break;
|
|
5562 case ALPHA + ADD_NL:
|
|
5563 p = "ALPHA+NL";
|
|
5564 break;
|
|
5565 case NALPHA:
|
|
5566 p = "NALPHA";
|
|
5567 break;
|
|
5568 case NALPHA + ADD_NL:
|
|
5569 p = "NALPHA+NL";
|
|
5570 break;
|
|
5571 case LOWER:
|
|
5572 p = "LOWER";
|
|
5573 break;
|
|
5574 case LOWER + ADD_NL:
|
|
5575 p = "LOWER+NL";
|
|
5576 break;
|
|
5577 case NLOWER:
|
|
5578 p = "NLOWER";
|
|
5579 break;
|
|
5580 case NLOWER + ADD_NL:
|
|
5581 p = "NLOWER+NL";
|
|
5582 break;
|
|
5583 case UPPER:
|
|
5584 p = "UPPER";
|
|
5585 break;
|
|
5586 case UPPER + ADD_NL:
|
|
5587 p = "UPPER+NL";
|
|
5588 break;
|
|
5589 case NUPPER:
|
|
5590 p = "NUPPER";
|
|
5591 break;
|
|
5592 case NUPPER + ADD_NL:
|
|
5593 p = "NUPPER+NL";
|
|
5594 break;
|
|
5595 case BRANCH:
|
|
5596 p = "BRANCH";
|
|
5597 break;
|
|
5598 case EXACTLY:
|
|
5599 p = "EXACTLY";
|
|
5600 break;
|
|
5601 case NOTHING:
|
|
5602 p = "NOTHING";
|
|
5603 break;
|
|
5604 case BACK:
|
|
5605 p = "BACK";
|
|
5606 break;
|
|
5607 case END:
|
|
5608 p = "END";
|
|
5609 break;
|
|
5610 case MOPEN + 0:
|
|
5611 p = "MATCH START";
|
|
5612 break;
|
|
5613 case MOPEN + 1:
|
|
5614 case MOPEN + 2:
|
|
5615 case MOPEN + 3:
|
|
5616 case MOPEN + 4:
|
|
5617 case MOPEN + 5:
|
|
5618 case MOPEN + 6:
|
|
5619 case MOPEN + 7:
|
|
5620 case MOPEN + 8:
|
|
5621 case MOPEN + 9:
|
|
5622 sprintf(buf + STRLEN(buf), "MOPEN%d", OP(op) - MOPEN);
|
|
5623 p = NULL;
|
|
5624 break;
|
|
5625 case MCLOSE + 0:
|
|
5626 p = "MATCH END";
|
|
5627 break;
|
|
5628 case MCLOSE + 1:
|
|
5629 case MCLOSE + 2:
|
|
5630 case MCLOSE + 3:
|
|
5631 case MCLOSE + 4:
|
|
5632 case MCLOSE + 5:
|
|
5633 case MCLOSE + 6:
|
|
5634 case MCLOSE + 7:
|
|
5635 case MCLOSE + 8:
|
|
5636 case MCLOSE + 9:
|
|
5637 sprintf(buf + STRLEN(buf), "MCLOSE%d", OP(op) - MCLOSE);
|
|
5638 p = NULL;
|
|
5639 break;
|
|
5640 case BACKREF + 1:
|
|
5641 case BACKREF + 2:
|
|
5642 case BACKREF + 3:
|
|
5643 case BACKREF + 4:
|
|
5644 case BACKREF + 5:
|
|
5645 case BACKREF + 6:
|
|
5646 case BACKREF + 7:
|
|
5647 case BACKREF + 8:
|
|
5648 case BACKREF + 9:
|
|
5649 sprintf(buf + STRLEN(buf), "BACKREF%d", OP(op) - BACKREF);
|
|
5650 p = NULL;
|
|
5651 break;
|
|
5652 case NOPEN:
|
|
5653 p = "NOPEN";
|
|
5654 break;
|
|
5655 case NCLOSE:
|
|
5656 p = "NCLOSE";
|
|
5657 break;
|
|
5658 #ifdef FEAT_SYN_HL
|
|
5659 case ZOPEN + 1:
|
|
5660 case ZOPEN + 2:
|
|
5661 case ZOPEN + 3:
|
|
5662 case ZOPEN + 4:
|
|
5663 case ZOPEN + 5:
|
|
5664 case ZOPEN + 6:
|
|
5665 case ZOPEN + 7:
|
|
5666 case ZOPEN + 8:
|
|
5667 case ZOPEN + 9:
|
|
5668 sprintf(buf + STRLEN(buf), "ZOPEN%d", OP(op) - ZOPEN);
|
|
5669 p = NULL;
|
|
5670 break;
|
|
5671 case ZCLOSE + 1:
|
|
5672 case ZCLOSE + 2:
|
|
5673 case ZCLOSE + 3:
|
|
5674 case ZCLOSE + 4:
|
|
5675 case ZCLOSE + 5:
|
|
5676 case ZCLOSE + 6:
|
|
5677 case ZCLOSE + 7:
|
|
5678 case ZCLOSE + 8:
|
|
5679 case ZCLOSE + 9:
|
|
5680 sprintf(buf + STRLEN(buf), "ZCLOSE%d", OP(op) - ZCLOSE);
|
|
5681 p = NULL;
|
|
5682 break;
|
|
5683 case ZREF + 1:
|
|
5684 case ZREF + 2:
|
|
5685 case ZREF + 3:
|
|
5686 case ZREF + 4:
|
|
5687 case ZREF + 5:
|
|
5688 case ZREF + 6:
|
|
5689 case ZREF + 7:
|
|
5690 case ZREF + 8:
|
|
5691 case ZREF + 9:
|
|
5692 sprintf(buf + STRLEN(buf), "ZREF%d", OP(op) - ZREF);
|
|
5693 p = NULL;
|
|
5694 break;
|
|
5695 #endif
|
|
5696 case STAR:
|
|
5697 p = "STAR";
|
|
5698 break;
|
|
5699 case PLUS:
|
|
5700 p = "PLUS";
|
|
5701 break;
|
|
5702 case NOMATCH:
|
|
5703 p = "NOMATCH";
|
|
5704 break;
|
|
5705 case MATCH:
|
|
5706 p = "MATCH";
|
|
5707 break;
|
|
5708 case BEHIND:
|
|
5709 p = "BEHIND";
|
|
5710 break;
|
|
5711 case NOBEHIND:
|
|
5712 p = "NOBEHIND";
|
|
5713 break;
|
|
5714 case SUBPAT:
|
|
5715 p = "SUBPAT";
|
|
5716 break;
|
|
5717 case BRACE_LIMITS:
|
|
5718 p = "BRACE_LIMITS";
|
|
5719 break;
|
|
5720 case BRACE_SIMPLE:
|
|
5721 p = "BRACE_SIMPLE";
|
|
5722 break;
|
|
5723 case BRACE_COMPLEX + 0:
|
|
5724 case BRACE_COMPLEX + 1:
|
|
5725 case BRACE_COMPLEX + 2:
|
|
5726 case BRACE_COMPLEX + 3:
|
|
5727 case BRACE_COMPLEX + 4:
|
|
5728 case BRACE_COMPLEX + 5:
|
|
5729 case BRACE_COMPLEX + 6:
|
|
5730 case BRACE_COMPLEX + 7:
|
|
5731 case BRACE_COMPLEX + 8:
|
|
5732 case BRACE_COMPLEX + 9:
|
|
5733 sprintf(buf + STRLEN(buf), "BRACE_COMPLEX%d", OP(op) - BRACE_COMPLEX);
|
|
5734 p = NULL;
|
|
5735 break;
|
|
5736 #ifdef FEAT_MBYTE
|
|
5737 case MULTIBYTECODE:
|
|
5738 p = "MULTIBYTECODE";
|
|
5739 break;
|
|
5740 #endif
|
|
5741 case NEWL:
|
|
5742 p = "NEWL";
|
|
5743 break;
|
|
5744 default:
|
|
5745 sprintf(buf + STRLEN(buf), "corrupt %d", OP(op));
|
|
5746 p = NULL;
|
|
5747 break;
|
|
5748 }
|
|
5749 if (p != NULL)
|
|
5750 (void) strcat(buf, p);
|
|
5751 return buf;
|
|
5752 }
|
|
5753 #endif
|
|
5754
|
|
5755 #ifdef FEAT_MBYTE
|
|
5756 static void mb_decompose __ARGS((int c, int *c1, int *c2, int *c3));
|
|
5757
|
|
5758 typedef struct
|
|
5759 {
|
|
5760 int a, b, c;
|
|
5761 } decomp_T;
|
|
5762
|
|
5763
|
|
5764 /* 0xfb20 - 0xfb4f */
|
|
5765 decomp_T decomp_table[0xfb4f-0xfb20+1] =
|
|
5766 {
|
|
5767 {0x5e2,0,0}, /* 0xfb20 alt ayin */
|
|
5768 {0x5d0,0,0}, /* 0xfb21 alt alef */
|
|
5769 {0x5d3,0,0}, /* 0xfb22 alt dalet */
|
|
5770 {0x5d4,0,0}, /* 0xfb23 alt he */
|
|
5771 {0x5db,0,0}, /* 0xfb24 alt kaf */
|
|
5772 {0x5dc,0,0}, /* 0xfb25 alt lamed */
|
|
5773 {0x5dd,0,0}, /* 0xfb26 alt mem-sofit */
|
|
5774 {0x5e8,0,0}, /* 0xfb27 alt resh */
|
|
5775 {0x5ea,0,0}, /* 0xfb28 alt tav */
|
|
5776 {'+', 0, 0}, /* 0xfb29 alt plus */
|
|
5777 {0x5e9, 0x5c1, 0}, /* 0xfb2a shin+shin-dot */
|
|
5778 {0x5e9, 0x5c2, 0}, /* 0xfb2b shin+sin-dot */
|
|
5779 {0x5e9, 0x5c1, 0x5bc}, /* 0xfb2c shin+shin-dot+dagesh */
|
|
5780 {0x5e9, 0x5c2, 0x5bc}, /* 0xfb2d shin+sin-dot+dagesh */
|
|
5781 {0x5d0, 0x5b7, 0}, /* 0xfb2e alef+patah */
|
|
5782 {0x5d0, 0x5b8, 0}, /* 0xfb2f alef+qamats */
|
|
5783 {0x5d0, 0x5b4, 0}, /* 0xfb30 alef+hiriq */
|
|
5784 {0x5d1, 0x5bc, 0}, /* 0xfb31 bet+dagesh */
|
|
5785 {0x5d2, 0x5bc, 0}, /* 0xfb32 gimel+dagesh */
|
|
5786 {0x5d3, 0x5bc, 0}, /* 0xfb33 dalet+dagesh */
|
|
5787 {0x5d4, 0x5bc, 0}, /* 0xfb34 he+dagesh */
|
|
5788 {0x5d5, 0x5bc, 0}, /* 0xfb35 vav+dagesh */
|
|
5789 {0x5d6, 0x5bc, 0}, /* 0xfb36 zayin+dagesh */
|
|
5790 {0xfb37, 0, 0}, /* 0xfb37 -- UNUSED */
|
|
5791 {0x5d8, 0x5bc, 0}, /* 0xfb38 tet+dagesh */
|
|
5792 {0x5d9, 0x5bc, 0}, /* 0xfb39 yud+dagesh */
|
|
5793 {0x5da, 0x5bc, 0}, /* 0xfb3a kaf sofit+dagesh */
|
|
5794 {0x5db, 0x5bc, 0}, /* 0xfb3b kaf+dagesh */
|
|
5795 {0x5dc, 0x5bc, 0}, /* 0xfb3c lamed+dagesh */
|
|
5796 {0xfb3d, 0, 0}, /* 0xfb3d -- UNUSED */
|
|
5797 {0x5de, 0x5bc, 0}, /* 0xfb3e mem+dagesh */
|
|
5798 {0xfb3f, 0, 0}, /* 0xfb3f -- UNUSED */
|
|
5799 {0x5e0, 0x5bc, 0}, /* 0xfb40 nun+dagesh */
|
|
5800 {0x5e1, 0x5bc, 0}, /* 0xfb41 samech+dagesh */
|
|
5801 {0xfb42, 0, 0}, /* 0xfb42 -- UNUSED */
|
|
5802 {0x5e3, 0x5bc, 0}, /* 0xfb43 pe sofit+dagesh */
|
|
5803 {0x5e4, 0x5bc,0}, /* 0xfb44 pe+dagesh */
|
|
5804 {0xfb45, 0, 0}, /* 0xfb45 -- UNUSED */
|
|
5805 {0x5e6, 0x5bc, 0}, /* 0xfb46 tsadi+dagesh */
|
|
5806 {0x5e7, 0x5bc, 0}, /* 0xfb47 qof+dagesh */
|
|
5807 {0x5e8, 0x5bc, 0}, /* 0xfb48 resh+dagesh */
|
|
5808 {0x5e9, 0x5bc, 0}, /* 0xfb49 shin+dagesh */
|
|
5809 {0x5ea, 0x5bc, 0}, /* 0xfb4a tav+dagesh */
|
|
5810 {0x5d5, 0x5b9, 0}, /* 0xfb4b vav+holam */
|
|
5811 {0x5d1, 0x5bf, 0}, /* 0xfb4c bet+rafe */
|
|
5812 {0x5db, 0x5bf, 0}, /* 0xfb4d kaf+rafe */
|
|
5813 {0x5e4, 0x5bf, 0}, /* 0xfb4e pe+rafe */
|
|
5814 {0x5d0, 0x5dc, 0} /* 0xfb4f alef-lamed */
|
|
5815 };
|
|
5816
|
|
5817 static void
|
|
5818 mb_decompose(c, c1, c2, c3)
|
|
5819 int c, *c1, *c2, *c3;
|
|
5820 {
|
|
5821 decomp_T d;
|
|
5822
|
|
5823 if (c >= 0x4b20 && c <= 0xfb4f)
|
|
5824 {
|
|
5825 d = decomp_table[c - 0xfb20];
|
|
5826 *c1 = d.a;
|
|
5827 *c2 = d.b;
|
|
5828 *c3 = d.c;
|
|
5829 }
|
|
5830 else
|
|
5831 {
|
|
5832 *c1 = c;
|
|
5833 *c2 = *c3 = 0;
|
|
5834 }
|
|
5835 }
|
|
5836 #endif
|
|
5837
|
|
5838 /*
|
|
5839 * Compare two strings, ignore case if ireg_ic set.
|
|
5840 * Return 0 if strings match, non-zero otherwise.
|
|
5841 * Correct the length "*n" when composing characters are ignored.
|
|
5842 */
|
|
5843 static int
|
|
5844 cstrncmp(s1, s2, n)
|
|
5845 char_u *s1, *s2;
|
|
5846 int *n;
|
|
5847 {
|
|
5848 int result;
|
|
5849
|
|
5850 if (!ireg_ic)
|
|
5851 result = STRNCMP(s1, s2, *n);
|
|
5852 else
|
|
5853 result = MB_STRNICMP(s1, s2, *n);
|
|
5854
|
|
5855 #ifdef FEAT_MBYTE
|
|
5856 /* if it failed and it's utf8 and we want to combineignore: */
|
|
5857 if (result != 0 && enc_utf8 && ireg_icombine)
|
|
5858 {
|
|
5859 char_u *str1, *str2;
|
|
5860 int c1, c2, c11, c12;
|
|
5861 int ix;
|
|
5862 int junk;
|
|
5863
|
|
5864 /* we have to handle the strcmp ourselves, since it is necessary to
|
|
5865 * deal with the composing characters by ignoring them: */
|
|
5866 str1 = s1;
|
|
5867 str2 = s2;
|
|
5868 c1 = c2 = 0;
|
|
5869 for (ix = 0; ix < *n; )
|
|
5870 {
|
|
5871 c1 = mb_ptr2char_adv(&str1);
|
|
5872 c2 = mb_ptr2char_adv(&str2);
|
|
5873 ix += utf_char2len(c1);
|
|
5874
|
|
5875 /* decompose the character if necessary, into 'base' characters
|
|
5876 * because I don't care about Arabic, I will hard-code the Hebrew
|
|
5877 * which I *do* care about! So sue me... */
|
|
5878 if (c1 != c2 && (!ireg_ic || utf_fold(c1) != utf_fold(c2)))
|
|
5879 {
|
|
5880 /* decomposition necessary? */
|
|
5881 mb_decompose(c1, &c11, &junk, &junk);
|
|
5882 mb_decompose(c2, &c12, &junk, &junk);
|
|
5883 c1 = c11;
|
|
5884 c2 = c12;
|
|
5885 if (c11 != c12 && (!ireg_ic || utf_fold(c11) != utf_fold(c12)))
|
|
5886 break;
|
|
5887 }
|
|
5888 }
|
|
5889 result = c2 - c1;
|
|
5890 if (result == 0)
|
|
5891 *n = (int)(str2 - s2);
|
|
5892 }
|
|
5893 #endif
|
|
5894
|
|
5895 return result;
|
|
5896 }
|
|
5897
|
|
5898 /*
|
|
5899 * cstrchr: This function is used a lot for simple searches, keep it fast!
|
|
5900 */
|
|
5901 static char_u *
|
|
5902 cstrchr(s, c)
|
|
5903 char_u *s;
|
|
5904 int c;
|
|
5905 {
|
|
5906 char_u *p;
|
|
5907 int cc;
|
|
5908
|
|
5909 if (!ireg_ic
|
|
5910 #ifdef FEAT_MBYTE
|
|
5911 || (!enc_utf8 && mb_char2len(c) > 1)
|
|
5912 #endif
|
|
5913 )
|
|
5914 return vim_strchr(s, c);
|
|
5915
|
|
5916 /* tolower() and toupper() can be slow, comparing twice should be a lot
|
|
5917 * faster (esp. when using MS Visual C++!).
|
|
5918 * For UTF-8 need to use folded case. */
|
|
5919 #ifdef FEAT_MBYTE
|
|
5920 if (enc_utf8 && c > 0x80)
|
|
5921 cc = utf_fold(c);
|
|
5922 else
|
|
5923 #endif
|
|
5924 if (isupper(c))
|
|
5925 cc = TOLOWER_LOC(c);
|
|
5926 else if (islower(c))
|
|
5927 cc = TOUPPER_LOC(c);
|
|
5928 else
|
|
5929 return vim_strchr(s, c);
|
|
5930
|
|
5931 #ifdef FEAT_MBYTE
|
|
5932 if (has_mbyte)
|
|
5933 {
|
|
5934 for (p = s; *p != NUL; p += (*mb_ptr2len_check)(p))
|
|
5935 {
|
|
5936 if (enc_utf8 && c > 0x80)
|
|
5937 {
|
|
5938 if (utf_fold(utf_ptr2char(p)) == cc)
|
|
5939 return p;
|
|
5940 }
|
|
5941 else if (*p == c || *p == cc)
|
|
5942 return p;
|
|
5943 }
|
|
5944 }
|
|
5945 else
|
|
5946 #endif
|
|
5947 /* Faster version for when there are no multi-byte characters. */
|
|
5948 for (p = s; *p != NUL; ++p)
|
|
5949 if (*p == c || *p == cc)
|
|
5950 return p;
|
|
5951
|
|
5952 return NULL;
|
|
5953 }
|
|
5954
|
|
5955 /***************************************************************
|
|
5956 * regsub stuff *
|
|
5957 ***************************************************************/
|
|
5958
|
|
5959 /* This stuff below really confuses cc on an SGI -- webb */
|
|
5960 #ifdef __sgi
|
|
5961 # undef __ARGS
|
|
5962 # define __ARGS(x) ()
|
|
5963 #endif
|
|
5964
|
|
5965 /*
|
|
5966 * We should define ftpr as a pointer to a function returning a pointer to
|
|
5967 * a function returning a pointer to a function ...
|
|
5968 * This is impossible, so we declare a pointer to a function returning a
|
|
5969 * pointer to a function returning void. This should work for all compilers.
|
|
5970 */
|
|
5971 typedef void (*(*fptr) __ARGS((char_u *, int)))();
|
|
5972
|
|
5973 static fptr do_upper __ARGS((char_u *, int));
|
|
5974 static fptr do_Upper __ARGS((char_u *, int));
|
|
5975 static fptr do_lower __ARGS((char_u *, int));
|
|
5976 static fptr do_Lower __ARGS((char_u *, int));
|
|
5977
|
|
5978 static int vim_regsub_both __ARGS((char_u *source, char_u *dest, int copy, int magic, int backslash));
|
|
5979
|
|
5980 static fptr
|
|
5981 do_upper(d, c)
|
|
5982 char_u *d;
|
|
5983 int c;
|
|
5984 {
|
|
5985 *d = TOUPPER_LOC(c);
|
|
5986
|
|
5987 return (fptr)NULL;
|
|
5988 }
|
|
5989
|
|
5990 static fptr
|
|
5991 do_Upper(d, c)
|
|
5992 char_u *d;
|
|
5993 int c;
|
|
5994 {
|
|
5995 *d = TOUPPER_LOC(c);
|
|
5996
|
|
5997 return (fptr)do_Upper;
|
|
5998 }
|
|
5999
|
|
6000 static fptr
|
|
6001 do_lower(d, c)
|
|
6002 char_u *d;
|
|
6003 int c;
|
|
6004 {
|
|
6005 *d = TOLOWER_LOC(c);
|
|
6006
|
|
6007 return (fptr)NULL;
|
|
6008 }
|
|
6009
|
|
6010 static fptr
|
|
6011 do_Lower(d, c)
|
|
6012 char_u *d;
|
|
6013 int c;
|
|
6014 {
|
|
6015 *d = TOLOWER_LOC(c);
|
|
6016
|
|
6017 return (fptr)do_Lower;
|
|
6018 }
|
|
6019
|
|
6020 /*
|
|
6021 * regtilde(): Replace tildes in the pattern by the old pattern.
|
|
6022 *
|
|
6023 * Short explanation of the tilde: It stands for the previous replacement
|
|
6024 * pattern. If that previous pattern also contains a ~ we should go back a
|
|
6025 * step further... But we insert the previous pattern into the current one
|
|
6026 * and remember that.
|
|
6027 * This still does not handle the case where "magic" changes. TODO?
|
|
6028 *
|
|
6029 * The tildes are parsed once before the first call to vim_regsub().
|
|
6030 */
|
|
6031 char_u *
|
|
6032 regtilde(source, magic)
|
|
6033 char_u *source;
|
|
6034 int magic;
|
|
6035 {
|
|
6036 char_u *newsub = source;
|
|
6037 char_u *tmpsub;
|
|
6038 char_u *p;
|
|
6039 int len;
|
|
6040 int prevlen;
|
|
6041
|
|
6042 for (p = newsub; *p; ++p)
|
|
6043 {
|
|
6044 if ((*p == '~' && magic) || (*p == '\\' && *(p + 1) == '~' && !magic))
|
|
6045 {
|
|
6046 if (reg_prev_sub != NULL)
|
|
6047 {
|
|
6048 /* length = len(newsub) - 1 + len(prev_sub) + 1 */
|
|
6049 prevlen = (int)STRLEN(reg_prev_sub);
|
|
6050 tmpsub = alloc((unsigned)(STRLEN(newsub) + prevlen));
|
|
6051 if (tmpsub != NULL)
|
|
6052 {
|
|
6053 /* copy prefix */
|
|
6054 len = (int)(p - newsub); /* not including ~ */
|
|
6055 mch_memmove(tmpsub, newsub, (size_t)len);
|
|
6056 /* interpretate tilde */
|
|
6057 mch_memmove(tmpsub + len, reg_prev_sub, (size_t)prevlen);
|
|
6058 /* copy postfix */
|
|
6059 if (!magic)
|
|
6060 ++p; /* back off \ */
|
|
6061 STRCPY(tmpsub + len + prevlen, p + 1);
|
|
6062
|
|
6063 if (newsub != source) /* already allocated newsub */
|
|
6064 vim_free(newsub);
|
|
6065 newsub = tmpsub;
|
|
6066 p = newsub + len + prevlen;
|
|
6067 }
|
|
6068 }
|
|
6069 else if (magic)
|
|
6070 STRCPY(p, p + 1); /* remove '~' */
|
|
6071 else
|
|
6072 STRCPY(p, p + 2); /* remove '\~' */
|
|
6073 --p;
|
|
6074 }
|
|
6075 else
|
|
6076 {
|
|
6077 if (*p == '\\' && p[1]) /* skip escaped characters */
|
|
6078 ++p;
|
|
6079 #ifdef FEAT_MBYTE
|
|
6080 if (has_mbyte)
|
|
6081 p += (*mb_ptr2len_check)(p) - 1;
|
|
6082 #endif
|
|
6083 }
|
|
6084 }
|
|
6085
|
|
6086 vim_free(reg_prev_sub);
|
|
6087 if (newsub != source) /* newsub was allocated, just keep it */
|
|
6088 reg_prev_sub = newsub;
|
|
6089 else /* no ~ found, need to save newsub */
|
|
6090 reg_prev_sub = vim_strsave(newsub);
|
|
6091 return newsub;
|
|
6092 }
|
|
6093
|
|
6094 #ifdef FEAT_EVAL
|
|
6095 static int can_f_submatch = FALSE; /* TRUE when submatch() can be used */
|
|
6096
|
|
6097 /* These pointers are used instead of reg_match and reg_mmatch for
|
|
6098 * reg_submatch(). Needed for when the substitution string is an expression
|
|
6099 * that contains a call to substitute() and submatch(). */
|
|
6100 static regmatch_T *submatch_match;
|
|
6101 static regmmatch_T *submatch_mmatch;
|
|
6102 #endif
|
|
6103
|
|
6104 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO)
|
|
6105 /*
|
|
6106 * vim_regsub() - perform substitutions after a vim_regexec() or
|
|
6107 * vim_regexec_multi() match.
|
|
6108 *
|
|
6109 * If "copy" is TRUE really copy into "dest".
|
|
6110 * If "copy" is FALSE nothing is copied, this is just to find out the length
|
|
6111 * of the result.
|
|
6112 *
|
|
6113 * If "backslash" is TRUE, a backslash will be removed later, need to double
|
|
6114 * them to keep them, and insert a backslash before a CR to avoid it being
|
|
6115 * replaced with a line break later.
|
|
6116 *
|
|
6117 * Note: The matched text must not change between the call of
|
|
6118 * vim_regexec()/vim_regexec_multi() and vim_regsub()! It would make the back
|
|
6119 * references invalid!
|
|
6120 *
|
|
6121 * Returns the size of the replacement, including terminating NUL.
|
|
6122 */
|
|
6123 int
|
|
6124 vim_regsub(rmp, source, dest, copy, magic, backslash)
|
|
6125 regmatch_T *rmp;
|
|
6126 char_u *source;
|
|
6127 char_u *dest;
|
|
6128 int copy;
|
|
6129 int magic;
|
|
6130 int backslash;
|
|
6131 {
|
|
6132 reg_match = rmp;
|
|
6133 reg_mmatch = NULL;
|
|
6134 reg_maxline = 0;
|
|
6135 return vim_regsub_both(source, dest, copy, magic, backslash);
|
|
6136 }
|
|
6137 #endif
|
|
6138
|
|
6139 int
|
|
6140 vim_regsub_multi(rmp, lnum, source, dest, copy, magic, backslash)
|
|
6141 regmmatch_T *rmp;
|
|
6142 linenr_T lnum;
|
|
6143 char_u *source;
|
|
6144 char_u *dest;
|
|
6145 int copy;
|
|
6146 int magic;
|
|
6147 int backslash;
|
|
6148 {
|
|
6149 reg_match = NULL;
|
|
6150 reg_mmatch = rmp;
|
|
6151 reg_buf = curbuf; /* always works on the current buffer! */
|
|
6152 reg_firstlnum = lnum;
|
|
6153 reg_maxline = curbuf->b_ml.ml_line_count - lnum;
|
|
6154 return vim_regsub_both(source, dest, copy, magic, backslash);
|
|
6155 }
|
|
6156
|
|
6157 static int
|
|
6158 vim_regsub_both(source, dest, copy, magic, backslash)
|
|
6159 char_u *source;
|
|
6160 char_u *dest;
|
|
6161 int copy;
|
|
6162 int magic;
|
|
6163 int backslash;
|
|
6164 {
|
|
6165 char_u *src;
|
|
6166 char_u *dst;
|
|
6167 char_u *s;
|
|
6168 int c;
|
|
6169 int no = -1;
|
|
6170 fptr func = (fptr)NULL;
|
|
6171 linenr_T clnum = 0; /* init for GCC */
|
|
6172 int len = 0; /* init for GCC */
|
|
6173 #ifdef FEAT_EVAL
|
|
6174 static char_u *eval_result = NULL;
|
|
6175 #endif
|
|
6176 #ifdef FEAT_MBYTE
|
|
6177 int l;
|
|
6178 #endif
|
|
6179
|
|
6180
|
|
6181 /* Be paranoid... */
|
|
6182 if (source == NULL || dest == NULL)
|
|
6183 {
|
|
6184 EMSG(_(e_null));
|
|
6185 return 0;
|
|
6186 }
|
|
6187 if (prog_magic_wrong())
|
|
6188 return 0;
|
|
6189 src = source;
|
|
6190 dst = dest;
|
|
6191
|
|
6192 /*
|
|
6193 * When the substitute part starts with "\=" evaluate it as an expression.
|
|
6194 */
|
|
6195 if (source[0] == '\\' && source[1] == '='
|
|
6196 #ifdef FEAT_EVAL
|
|
6197 && !can_f_submatch /* can't do this recursively */
|
|
6198 #endif
|
|
6199 )
|
|
6200 {
|
|
6201 #ifdef FEAT_EVAL
|
|
6202 /* To make sure that the length doesn't change between checking the
|
|
6203 * length and copying the string, and to speed up things, the
|
|
6204 * resulting string is saved from the call with "copy" == FALSE to the
|
|
6205 * call with "copy" == TRUE. */
|
|
6206 if (copy)
|
|
6207 {
|
|
6208 if (eval_result != NULL)
|
|
6209 {
|
|
6210 STRCPY(dest, eval_result);
|
|
6211 dst += STRLEN(eval_result);
|
|
6212 vim_free(eval_result);
|
|
6213 eval_result = NULL;
|
|
6214 }
|
|
6215 }
|
|
6216 else
|
|
6217 {
|
|
6218 linenr_T save_reg_maxline;
|
|
6219 win_T *save_reg_win;
|
|
6220 int save_ireg_ic;
|
|
6221
|
|
6222 vim_free(eval_result);
|
|
6223
|
|
6224 /* The expression may contain substitute(), which calls us
|
|
6225 * recursively. Make sure submatch() gets the text from the first
|
|
6226 * level. Don't need to save "reg_buf", because
|
|
6227 * vim_regexec_multi() can't be called recursively. */
|
|
6228 submatch_match = reg_match;
|
|
6229 submatch_mmatch = reg_mmatch;
|
|
6230 save_reg_maxline = reg_maxline;
|
|
6231 save_reg_win = reg_win;
|
|
6232 save_ireg_ic = ireg_ic;
|
|
6233 can_f_submatch = TRUE;
|
|
6234
|
|
6235 eval_result = eval_to_string(source + 2, NULL);
|
|
6236 if (eval_result != NULL)
|
|
6237 {
|
39
|
6238 for (s = eval_result; *s != NUL; mb_ptr_adv(s))
|
7
|
6239 {
|
|
6240 /* Change NL to CR, so that it becomes a line break.
|
|
6241 * Skip over a backslashed character. */
|
|
6242 if (*s == NL)
|
|
6243 *s = CAR;
|
|
6244 else if (*s == '\\' && s[1] != NUL)
|
|
6245 ++s;
|
|
6246 }
|
|
6247
|
|
6248 dst += STRLEN(eval_result);
|
|
6249 }
|
|
6250
|
|
6251 reg_match = submatch_match;
|
|
6252 reg_mmatch = submatch_mmatch;
|
|
6253 reg_maxline = save_reg_maxline;
|
|
6254 reg_win = save_reg_win;
|
|
6255 ireg_ic = save_ireg_ic;
|
|
6256 can_f_submatch = FALSE;
|
|
6257 }
|
|
6258 #endif
|
|
6259 }
|
|
6260 else
|
|
6261 while ((c = *src++) != NUL)
|
|
6262 {
|
|
6263 if (c == '&' && magic)
|
|
6264 no = 0;
|
|
6265 else if (c == '\\' && *src != NUL)
|
|
6266 {
|
|
6267 if (*src == '&' && !magic)
|
|
6268 {
|
|
6269 ++src;
|
|
6270 no = 0;
|
|
6271 }
|
|
6272 else if ('0' <= *src && *src <= '9')
|
|
6273 {
|
|
6274 no = *src++ - '0';
|
|
6275 }
|
|
6276 else if (vim_strchr((char_u *)"uUlLeE", *src))
|
|
6277 {
|
|
6278 switch (*src++)
|
|
6279 {
|
|
6280 case 'u': func = (fptr)do_upper;
|
|
6281 continue;
|
|
6282 case 'U': func = (fptr)do_Upper;
|
|
6283 continue;
|
|
6284 case 'l': func = (fptr)do_lower;
|
|
6285 continue;
|
|
6286 case 'L': func = (fptr)do_Lower;
|
|
6287 continue;
|
|
6288 case 'e':
|
|
6289 case 'E': func = (fptr)NULL;
|
|
6290 continue;
|
|
6291 }
|
|
6292 }
|
|
6293 }
|
|
6294 if (no < 0) /* Ordinary character. */
|
|
6295 {
|
|
6296 if (c == '\\' && *src != NUL)
|
|
6297 {
|
|
6298 /* Check for abbreviations -- webb */
|
|
6299 switch (*src)
|
|
6300 {
|
|
6301 case 'r': c = CAR; ++src; break;
|
|
6302 case 'n': c = NL; ++src; break;
|
|
6303 case 't': c = TAB; ++src; break;
|
|
6304 /* Oh no! \e already has meaning in subst pat :-( */
|
|
6305 /* case 'e': c = ESC; ++src; break; */
|
|
6306 case 'b': c = Ctrl_H; ++src; break;
|
|
6307
|
|
6308 /* If "backslash" is TRUE the backslash will be removed
|
|
6309 * later. Used to insert a literal CR. */
|
|
6310 default: if (backslash)
|
|
6311 {
|
|
6312 if (copy)
|
|
6313 *dst = '\\';
|
|
6314 ++dst;
|
|
6315 }
|
|
6316 c = *src++;
|
|
6317 }
|
|
6318 }
|
|
6319
|
|
6320 /* Write to buffer, if copy is set. */
|
|
6321 #ifdef FEAT_MBYTE
|
|
6322 if (has_mbyte && (l = (*mb_ptr2len_check)(src - 1)) > 1)
|
|
6323 {
|
|
6324 /* TODO: should use "func" here. */
|
|
6325 if (copy)
|
|
6326 mch_memmove(dst, src - 1, l);
|
|
6327 dst += l - 1;
|
|
6328 src += l - 1;
|
|
6329 }
|
|
6330 else
|
|
6331 {
|
|
6332 #endif
|
|
6333 if (copy)
|
|
6334 {
|
|
6335 if (func == (fptr)NULL) /* just copy */
|
|
6336 *dst = c;
|
|
6337 else /* change case */
|
|
6338 func = (fptr)(func(dst, c));
|
|
6339 /* Turbo C complains without the typecast */
|
|
6340 }
|
|
6341 #ifdef FEAT_MBYTE
|
|
6342 }
|
|
6343 #endif
|
|
6344 dst++;
|
|
6345 }
|
|
6346 else
|
|
6347 {
|
|
6348 if (REG_MULTI)
|
|
6349 {
|
|
6350 clnum = reg_mmatch->startpos[no].lnum;
|
|
6351 if (clnum < 0 || reg_mmatch->endpos[no].lnum < 0)
|
|
6352 s = NULL;
|
|
6353 else
|
|
6354 {
|
|
6355 s = reg_getline(clnum) + reg_mmatch->startpos[no].col;
|
|
6356 if (reg_mmatch->endpos[no].lnum == clnum)
|
|
6357 len = reg_mmatch->endpos[no].col
|
|
6358 - reg_mmatch->startpos[no].col;
|
|
6359 else
|
|
6360 len = (int)STRLEN(s);
|
|
6361 }
|
|
6362 }
|
|
6363 else
|
|
6364 {
|
|
6365 s = reg_match->startp[no];
|
|
6366 if (reg_match->endp[no] == NULL)
|
|
6367 s = NULL;
|
|
6368 else
|
|
6369 len = (int)(reg_match->endp[no] - s);
|
|
6370 }
|
|
6371 if (s != NULL)
|
|
6372 {
|
|
6373 for (;;)
|
|
6374 {
|
|
6375 if (len == 0)
|
|
6376 {
|
|
6377 if (REG_MULTI)
|
|
6378 {
|
|
6379 if (reg_mmatch->endpos[no].lnum == clnum)
|
|
6380 break;
|
|
6381 if (copy)
|
|
6382 *dst = CAR;
|
|
6383 ++dst;
|
|
6384 s = reg_getline(++clnum);
|
|
6385 if (reg_mmatch->endpos[no].lnum == clnum)
|
|
6386 len = reg_mmatch->endpos[no].col;
|
|
6387 else
|
|
6388 len = (int)STRLEN(s);
|
|
6389 }
|
|
6390 else
|
|
6391 break;
|
|
6392 }
|
|
6393 else if (*s == NUL) /* we hit NUL. */
|
|
6394 {
|
|
6395 if (copy)
|
|
6396 EMSG(_(e_re_damg));
|
|
6397 goto exit;
|
|
6398 }
|
|
6399 else
|
|
6400 {
|
|
6401 if (backslash && (*s == CAR || *s == '\\'))
|
|
6402 {
|
|
6403 /*
|
|
6404 * Insert a backslash in front of a CR, otherwise
|
|
6405 * it will be replaced by a line break.
|
|
6406 * Number of backslashes will be halved later,
|
|
6407 * double them here.
|
|
6408 */
|
|
6409 if (copy)
|
|
6410 {
|
|
6411 dst[0] = '\\';
|
|
6412 dst[1] = *s;
|
|
6413 }
|
|
6414 dst += 2;
|
|
6415 }
|
|
6416 #ifdef FEAT_MBYTE
|
|
6417 else if (has_mbyte && (l = (*mb_ptr2len_check)(s)) > 1)
|
|
6418 {
|
|
6419 /* TODO: should use "func" here. */
|
|
6420 if (copy)
|
|
6421 mch_memmove(dst, s, l);
|
|
6422 dst += l;
|
|
6423 s += l - 1;
|
|
6424 len -= l - 1;
|
|
6425 }
|
|
6426 #endif
|
|
6427 else
|
|
6428 {
|
|
6429 if (copy)
|
|
6430 {
|
|
6431 if (func == (fptr)NULL) /* just copy */
|
|
6432 *dst = *s;
|
|
6433 else /* change case */
|
|
6434 func = (fptr)(func(dst, *s));
|
|
6435 /* Turbo C complains without the typecast */
|
|
6436 }
|
|
6437 ++dst;
|
|
6438 }
|
|
6439 ++s;
|
|
6440 --len;
|
|
6441 }
|
|
6442 }
|
|
6443 }
|
|
6444 no = -1;
|
|
6445 }
|
|
6446 }
|
|
6447 if (copy)
|
|
6448 *dst = NUL;
|
|
6449
|
|
6450 exit:
|
|
6451 return (int)((dst - dest) + 1);
|
|
6452 }
|
|
6453
|
|
6454 #ifdef FEAT_EVAL
|
|
6455 /*
|
|
6456 * Used for the submatch() function: get the string from tne n'th submatch in
|
|
6457 * allocated memory.
|
|
6458 * Returns NULL when not in a ":s" command and for a non-existing submatch.
|
|
6459 */
|
|
6460 char_u *
|
|
6461 reg_submatch(no)
|
|
6462 int no;
|
|
6463 {
|
|
6464 char_u *retval = NULL;
|
|
6465 char_u *s;
|
|
6466 int len;
|
|
6467 int round;
|
|
6468 linenr_T lnum;
|
|
6469
|
|
6470 if (!can_f_submatch)
|
|
6471 return NULL;
|
|
6472
|
|
6473 if (submatch_match == NULL)
|
|
6474 {
|
|
6475 /*
|
|
6476 * First round: compute the length and allocate memory.
|
|
6477 * Second round: copy the text.
|
|
6478 */
|
|
6479 for (round = 1; round <= 2; ++round)
|
|
6480 {
|
|
6481 lnum = submatch_mmatch->startpos[no].lnum;
|
|
6482 if (lnum < 0 || submatch_mmatch->endpos[no].lnum < 0)
|
|
6483 return NULL;
|
|
6484
|
|
6485 s = reg_getline(lnum) + submatch_mmatch->startpos[no].col;
|
|
6486 if (s == NULL) /* anti-crash check, cannot happen? */
|
|
6487 break;
|
|
6488 if (submatch_mmatch->endpos[no].lnum == lnum)
|
|
6489 {
|
|
6490 /* Within one line: take form start to end col. */
|
|
6491 len = submatch_mmatch->endpos[no].col
|
|
6492 - submatch_mmatch->startpos[no].col;
|
|
6493 if (round == 2)
|
|
6494 {
|
|
6495 STRNCPY(retval, s, len);
|
|
6496 retval[len] = NUL;
|
|
6497 }
|
|
6498 ++len;
|
|
6499 }
|
|
6500 else
|
|
6501 {
|
|
6502 /* Multiple lines: take start line from start col, middle
|
|
6503 * lines completely and end line up to end col. */
|
|
6504 len = (int)STRLEN(s);
|
|
6505 if (round == 2)
|
|
6506 {
|
|
6507 STRCPY(retval, s);
|
|
6508 retval[len] = '\n';
|
|
6509 }
|
|
6510 ++len;
|
|
6511 ++lnum;
|
|
6512 while (lnum < submatch_mmatch->endpos[no].lnum)
|
|
6513 {
|
|
6514 s = reg_getline(lnum++);
|
|
6515 if (round == 2)
|
|
6516 STRCPY(retval + len, s);
|
|
6517 len += (int)STRLEN(s);
|
|
6518 if (round == 2)
|
|
6519 retval[len] = '\n';
|
|
6520 ++len;
|
|
6521 }
|
|
6522 if (round == 2)
|
|
6523 STRNCPY(retval + len, reg_getline(lnum),
|
|
6524 submatch_mmatch->endpos[no].col);
|
|
6525 len += submatch_mmatch->endpos[no].col;
|
|
6526 if (round == 2)
|
|
6527 retval[len] = NUL;
|
|
6528 ++len;
|
|
6529 }
|
|
6530
|
|
6531 if (round == 1)
|
|
6532 {
|
|
6533 retval = lalloc((long_u)len, TRUE);
|
|
6534 if (s == NULL)
|
|
6535 return NULL;
|
|
6536 }
|
|
6537 }
|
|
6538 }
|
|
6539 else
|
|
6540 {
|
|
6541 if (submatch_match->endp[no] == NULL)
|
|
6542 retval = NULL;
|
|
6543 else
|
|
6544 {
|
|
6545 s = submatch_match->startp[no];
|
|
6546 retval = vim_strnsave(s, (int)(submatch_match->endp[no] - s));
|
|
6547 }
|
|
6548 }
|
|
6549
|
|
6550 return retval;
|
|
6551 }
|
|
6552 #endif
|