|
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
|
|
|
77 * node points to the node after the stuff to be repeated. The operand of some
|
|
|
78 * types of node is a literal string; for others, it is a node leading into a
|
|
|
79 * sub-FSM. In particular, the operand of a BRANCH node is the first node of
|
|
|
80 * the branch. (NB this is *not* a tree structure: the tail of the branch
|
|
|
81 * connects to the thing following the set of BRANCHes.)
|
|
|
82 *
|
|
|
83 * pattern is coded like:
|
|
|
84 *
|
|
|
85 * +-----------------+
|
|
|
86 * | V
|
|
|
87 * <aa>\|<bb> BRANCH <aa> BRANCH <bb> --> END
|
|
|
88 * | ^ | ^
|
|
|
89 * +------+ +----------+
|
|
|
90 *
|
|
|
91 *
|
|
|
92 * +------------------+
|
|
|
93 * V |
|
|
|
94 * <aa>* BRANCH BRANCH <aa> --> BACK BRANCH --> NOTHING --> END
|
|
|
95 * | | ^ ^
|
|
|
96 * | +---------------+ |
|
|
|
97 * +---------------------------------------------+
|
|
|
98 *
|
|
|
99 *
|
|
|
100 * +-------------------------+
|
|
|
101 * V |
|
|
|
102 * <aa>\{} BRANCH BRACE_LIMITS --> BRACE_COMPLEX <aa> --> BACK END
|
|
|
103 * | | ^
|
|
|
104 * | +----------------+
|
|
|
105 * +-----------------------------------------------+
|
|
|
106 *
|
|
|
107 *
|
|
|
108 * <aa>\@!<bb> BRANCH NOMATCH <aa> --> END <bb> --> END
|
|
|
109 * | | ^ ^
|
|
|
110 * | +----------------+ |
|
|
|
111 * +--------------------------------+
|
|
|
112 *
|
|
|
113 * +---------+
|
|
|
114 * | V
|
|
|
115 * \z[abc] BRANCH BRANCH a BRANCH b BRANCH c BRANCH NOTHING --> END
|
|
|
116 * | | | | ^ ^
|
|
|
117 * | | | +-----+ |
|
|
|
118 * | | +----------------+ |
|
|
|
119 * | +---------------------------+ |
|
|
|
120 * +------------------------------------------------------+
|
|
|
121 *
|
|
|
122 * They all start with a BRANCH for "\|" alternaties, even when there is only
|
|
|
123 * one alternative.
|
|
|
124 */
|
|
|
125
|
|
|
126 /*
|
|
|
127 * The opcodes are:
|
|
|
128 */
|
|
|
129
|
|
|
130 /* definition number opnd? meaning */
|
|
|
131 #define END 0 /* End of program or NOMATCH operand. */
|
|
|
132 #define BOL 1 /* Match "" at beginning of line. */
|
|
|
133 #define EOL 2 /* Match "" at end of line. */
|
|
|
134 #define BRANCH 3 /* node Match this alternative, or the
|
|
|
135 * next... */
|
|
|
136 #define BACK 4 /* Match "", "next" ptr points backward. */
|
|
|
137 #define EXACTLY 5 /* str Match this string. */
|
|
|
138 #define NOTHING 6 /* Match empty string. */
|
|
|
139 #define STAR 7 /* node Match this (simple) thing 0 or more
|
|
|
140 * times. */
|
|
|
141 #define PLUS 8 /* node Match this (simple) thing 1 or more
|
|
|
142 * times. */
|
|
|
143 #define MATCH 9 /* node match the operand zero-width */
|
|
|
144 #define NOMATCH 10 /* node check for no match with operand */
|
|
|
145 #define BEHIND 11 /* node look behind for a match with operand */
|
|
|
146 #define NOBEHIND 12 /* node look behind for no match with operand */
|
|
|
147 #define SUBPAT 13 /* node match the operand here */
|
|
|
148 #define BRACE_SIMPLE 14 /* node Match this (simple) thing between m and
|
|
|
149 * n times (\{m,n\}). */
|
|
|
150 #define BOW 15 /* Match "" after [^a-zA-Z0-9_] */
|
|
|
151 #define EOW 16 /* Match "" at [^a-zA-Z0-9_] */
|
|
|
152 #define BRACE_LIMITS 17 /* nr nr define the min & max for BRACE_SIMPLE
|
|
|
153 * and BRACE_COMPLEX. */
|
|
|
154 #define NEWL 18 /* Match line-break */
|
|
|
155 #define BHPOS 19 /* End position for BEHIND or NOBEHIND */
|
|
|
156
|
|
|
157
|
|
|
158 /* character classes: 20-48 normal, 50-78 include a line-break */
|
|
|
159 #define ADD_NL 30
|
|
|
160 #define FIRST_NL ANY + ADD_NL
|
|
|
161 #define ANY 20 /* Match any one character. */
|
|
|
162 #define ANYOF 21 /* str Match any character in this string. */
|
|
|
163 #define ANYBUT 22 /* str Match any character not in this
|
|
|
164 * string. */
|
|
|
165 #define IDENT 23 /* Match identifier char */
|
|
|
166 #define SIDENT 24 /* Match identifier char but no digit */
|
|
|
167 #define KWORD 25 /* Match keyword char */
|
|
|
168 #define SKWORD 26 /* Match word char but no digit */
|
|
|
169 #define FNAME 27 /* Match file name char */
|
|
|
170 #define SFNAME 28 /* Match file name char but no digit */
|
|
|
171 #define PRINT 29 /* Match printable char */
|
|
|
172 #define SPRINT 30 /* Match printable char but no digit */
|
|
|
173 #define WHITE 31 /* Match whitespace char */
|
|
|
174 #define NWHITE 32 /* Match non-whitespace char */
|
|
|
175 #define DIGIT 33 /* Match digit char */
|
|
|
176 #define NDIGIT 34 /* Match non-digit char */
|
|
|
177 #define HEX 35 /* Match hex char */
|
|
|
178 #define NHEX 36 /* Match non-hex char */
|
|
|
179 #define OCTAL 37 /* Match octal char */
|
|
|
180 #define NOCTAL 38 /* Match non-octal char */
|
|
|
181 #define WORD 39 /* Match word char */
|
|
|
182 #define NWORD 40 /* Match non-word char */
|
|
|
183 #define HEAD 41 /* Match head char */
|
|
|
184 #define NHEAD 42 /* Match non-head char */
|
|
|
185 #define ALPHA 43 /* Match alpha char */
|
|
|
186 #define NALPHA 44 /* Match non-alpha char */
|
|
|
187 #define LOWER 45 /* Match lowercase char */
|
|
|
188 #define NLOWER 46 /* Match non-lowercase char */
|
|
|
189 #define UPPER 47 /* Match uppercase char */
|
|
|
190 #define NUPPER 48 /* Match non-uppercase char */
|
|
|
191 #define LAST_NL NUPPER + ADD_NL
|
|
|
192 #define WITH_NL(op) ((op) >= FIRST_NL && (op) <= LAST_NL)
|
|
|
193
|
|
|
194 #define MOPEN 80 /* -89 Mark this point in input as start of
|
|
|
195 * \( subexpr. MOPEN + 0 marks start of
|
|
|
196 * match. */
|
|
|
197 #define MCLOSE 90 /* -99 Analogous to MOPEN. MCLOSE + 0 marks
|
|
|
198 * end of match. */
|
|
|
199 #define BACKREF 100 /* -109 node Match same string again \1-\9 */
|
|
|
200
|
|
|
201 #ifdef FEAT_SYN_HL
|
|
|
202 # define ZOPEN 110 /* -119 Mark this point in input as start of
|
|
|
203 * \z( subexpr. */
|
|
|
204 # define ZCLOSE 120 /* -129 Analogous to ZOPEN. */
|
|
|
205 # define ZREF 130 /* -139 node Match external submatch \z1-\z9 */
|
|
|
206 #endif
|
|
|
207
|
|
|
208 #define BRACE_COMPLEX 140 /* -149 node Match nodes between m & n times */
|
|
|
209
|
|
|
210 #define NOPEN 150 /* Mark this point in input as start of
|
|
|
211 \%( subexpr. */
|
|
|
212 #define NCLOSE 151 /* Analogous to NOPEN. */
|
|
|
213
|
|
|
214 #define MULTIBYTECODE 200 /* mbc Match one multi-byte character */
|
|
|
215 #define RE_BOF 201 /* Match "" at beginning of file. */
|
|
|
216 #define RE_EOF 202 /* Match "" at end of file. */
|
|
|
217 #define CURSOR 203 /* Match location of cursor. */
|
|
|
218
|
|
|
219 #define RE_LNUM 204 /* nr cmp Match line number */
|
|
|
220 #define RE_COL 205 /* nr cmp Match column number */
|
|
|
221 #define RE_VCOL 206 /* nr cmp Match virtual column number */
|
|
|
222
|
|
|
223 /*
|
|
|
224 * Magic characters have a special meaning, they don't match literally.
|
|
|
225 * Magic characters are negative. This separates them from literal characters
|
|
|
226 * (possibly multi-byte). Only ASCII characters can be Magic.
|
|
|
227 */
|
|
|
228 #define Magic(x) ((int)(x) - 256)
|
|
|
229 #define un_Magic(x) ((x) + 256)
|
|
|
230 #define is_Magic(x) ((x) < 0)
|
|
|
231
|
|
|
232 static int no_Magic __ARGS((int x));
|
|
|
233 static int toggle_Magic __ARGS((int x));
|
|
|
234
|
|
|
235 static int
|
|
|
236 no_Magic(x)
|
|
|
237 int x;
|
|
|
238 {
|
|
|
239 if (is_Magic(x))
|
|
|
240 return un_Magic(x);
|
|
|
241 return x;
|
|
|
242 }
|
|
|
243
|
|
|
244 static int
|
|
|
245 toggle_Magic(x)
|
|
|
246 int x;
|
|
|
247 {
|
|
|
248 if (is_Magic(x))
|
|
|
249 return un_Magic(x);
|
|
|
250 return Magic(x);
|
|
|
251 }
|
|
|
252
|
|
|
253 /*
|
|
|
254 * The first byte of the regexp internal "program" is actually this magic
|
|
|
255 * number; the start node begins in the second byte. It's used to catch the
|
|
|
256 * most severe mutilation of the program by the caller.
|
|
|
257 */
|
|
|
258
|
|
|
259 #define REGMAGIC 0234
|
|
|
260
|
|
|
261 /*
|
|
|
262 * Opcode notes:
|
|
|
263 *
|
|
|
264 * BRANCH The set of branches constituting a single choice are hooked
|
|
|
265 * together with their "next" pointers, since precedence prevents
|
|
|
266 * anything being concatenated to any individual branch. The
|
|
|
267 * "next" pointer of the last BRANCH in a choice points to the
|
|
|
268 * thing following the whole choice. This is also where the
|
|
|
269 * final "next" pointer of each individual branch points; each
|
|
|
270 * branch starts with the operand node of a BRANCH node.
|
|
|
271 *
|
|
|
272 * BACK Normal "next" pointers all implicitly point forward; BACK
|
|
|
273 * exists to make loop structures possible.
|
|
|
274 *
|
|
|
275 * STAR,PLUS '=', and complex '*' and '+', are implemented as circular
|
|
|
276 * BRANCH structures using BACK. Simple cases (one character
|
|
|
277 * per match) are implemented with STAR and PLUS for speed
|
|
|
278 * and to minimize recursive plunges.
|
|
|
279 *
|
|
|
280 * BRACE_LIMITS This is always followed by a BRACE_SIMPLE or BRACE_COMPLEX
|
|
|
281 * node, and defines the min and max limits to be used for that
|
|
|
282 * node.
|
|
|
283 *
|
|
|
284 * MOPEN,MCLOSE ...are numbered at compile time.
|
|
|
285 * ZOPEN,ZCLOSE ...ditto
|
|
|
286 */
|
|
|
287
|
|
|
288 /*
|
|
|
289 * A node is one char of opcode followed by two chars of "next" pointer.
|
|
|
290 * "Next" pointers are stored as two 8-bit bytes, high order first. The
|
|
|
291 * value is a positive offset from the opcode of the node containing it.
|
|
|
292 * An operand, if any, simply follows the node. (Note that much of the
|
|
|
293 * code generation knows about this implicit relationship.)
|
|
|
294 *
|
|
|
295 * Using two bytes for the "next" pointer is vast overkill for most things,
|
|
|
296 * but allows patterns to get big without disasters.
|
|
|
297 */
|
|
|
298 #define OP(p) ((int)*(p))
|
|
|
299 #define NEXT(p) (((*((p) + 1) & 0377) << 8) + (*((p) + 2) & 0377))
|
|
|
300 #define OPERAND(p) ((p) + 3)
|
|
|
301 /* Obtain an operand that was stored as four bytes, MSB first. */
|
|
|
302 #define OPERAND_MIN(p) (((long)(p)[3] << 24) + ((long)(p)[4] << 16) \
|
|
|
303 + ((long)(p)[5] << 8) + (long)(p)[6])
|
|
|
304 /* Obtain a second operand stored as four bytes. */
|
|
|
305 #define OPERAND_MAX(p) OPERAND_MIN((p) + 4)
|
|
|
306 /* Obtain a second single-byte operand stored after a four bytes operand. */
|
|
|
307 #define OPERAND_CMP(p) (p)[7]
|
|
|
308
|
|
|
309 /*
|
|
|
310 * Utility definitions.
|
|
|
311 */
|
|
|
312 #define UCHARAT(p) ((int)*(char_u *)(p))
|
|
|
313
|
|
|
314 /* Used for an error (down from) vim_regcomp(): give the error message, set
|
|
|
315 * rc_did_emsg and return NULL */
|
|
|
316 #define EMSG_RET_NULL(m) { EMSG(m); rc_did_emsg = TRUE; return NULL; }
|
|
|
317 #define EMSG_M_RET_NULL(m, c) { EMSG2(m, c ? "" : "\\"); rc_did_emsg = TRUE; return NULL; }
|
|
|
318 #define EMSG_RET_FAIL(m) { EMSG(m); rc_did_emsg = TRUE; return FAIL; }
|
|
|
319 #define EMSG_ONE_RET_NULL EMSG_M_RET_NULL(_("E369: invalid item in %s%%[]"), reg_magic == MAGIC_ALL)
|
|
|
320
|
|
|
321 #define MAX_LIMIT (32767L << 16L)
|
|
|
322
|
|
|
323 static int re_multi_type __ARGS((int));
|
|
|
324 static int cstrncmp __ARGS((char_u *s1, char_u *s2, int *n));
|
|
|
325 static char_u *cstrchr __ARGS((char_u *, int));
|
|
|
326
|
|
|
327 #ifdef DEBUG
|
|
|
328 static void regdump __ARGS((char_u *, regprog_T *));
|
|
|
329 static char_u *regprop __ARGS((char_u *));
|
|
|
330 #endif
|
|
|
331
|
|
|
332 #define NOT_MULTI 0
|
|
|
333 #define MULTI_ONE 1
|
|
|
334 #define MULTI_MULT 2
|
|
|
335 /*
|
|
|
336 * Return NOT_MULTI if c is not a "multi" operator.
|
|
|
337 * Return MULTI_ONE if c is a single "multi" operator.
|
|
|
338 * Return MULTI_MULT if c is a multi "multi" operator.
|
|
|
339 */
|
|
|
340 static int
|
|
|
341 re_multi_type(c)
|
|
|
342 int c;
|
|
|
343 {
|
|
|
344 if (c == Magic('@') || c == Magic('=') || c == Magic('?'))
|
|
|
345 return MULTI_ONE;
|
|
|
346 if (c == Magic('*') || c == Magic('+') || c == Magic('{'))
|
|
|
347 return MULTI_MULT;
|
|
|
348 return NOT_MULTI;
|
|
|
349 }
|
|
|
350
|
|
|
351 /*
|
|
|
352 * Flags to be passed up and down.
|
|
|
353 */
|
|
|
354 #define HASWIDTH 0x1 /* Known never to match null string. */
|
|
|
355 #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */
|
|
|
356 #define SPSTART 0x4 /* Starts with * or +. */
|
|
|
357 #define HASNL 0x8 /* Contains some \n. */
|
|
|
358 #define HASLOOKBH 0x10 /* Contains "\@<=" or "\@<!". */
|
|
|
359 #define WORST 0 /* Worst case. */
|
|
|
360
|
|
|
361 /*
|
|
|
362 * When regcode is set to this value, code is not emitted and size is computed
|
|
|
363 * instead.
|
|
|
364 */
|
|
|
365 #define JUST_CALC_SIZE ((char_u *) -1)
|
|
|
366
|
|
|
367 static char_u *reg_prev_sub;
|
|
|
368
|
|
|
369 /*
|
|
|
370 * REGEXP_INRANGE contains all characters which are always special in a []
|
|
|
371 * range after '\'.
|
|
|
372 * REGEXP_ABBR contains all characters which act as abbreviations after '\'.
|
|
|
373 * These are:
|
|
|
374 * \n - New line (NL).
|
|
|
375 * \r - Carriage Return (CR).
|
|
|
376 * \t - Tab (TAB).
|
|
|
377 * \e - Escape (ESC).
|
|
|
378 * \b - Backspace (Ctrl_H).
|
|
24
|
379 * \d - Character code in decimal, eg \d123
|
|
|
380 * \o - Character code in octal, eg \o80
|
|
|
381 * \x - Character code in hex, eg \x4a
|
|
|
382 * \u - Multibyte character code, eg \u20ac
|
|
|
383 * \U - Long multibyte character code, eg \U12345678
|
|
7
|
384 */
|
|
|
385 static char_u REGEXP_INRANGE[] = "]^-n\\";
|
|
24
|
386 static char_u REGEXP_ABBR[] = "nrtebdoxuU";
|
|
7
|
387
|
|
|
388 static int backslash_trans __ARGS((int c));
|
|
|
389 static int skip_class_name __ARGS((char_u **pp));
|
|
|
390 static char_u *skip_anyof __ARGS((char_u *p));
|
|
|
391 static void init_class_tab __ARGS((void));
|
|
|
392
|
|
|
393 /*
|
|
|
394 * Translate '\x' to its control character, except "\n", which is Magic.
|
|
|
395 */
|
|
|
396 static int
|
|
|
397 backslash_trans(c)
|
|
|
398 int c;
|
|
|
399 {
|
|
|
400 switch (c)
|
|
|
401 {
|
|
|
402 case 'r': return CAR;
|
|
|
403 case 't': return TAB;
|
|
|
404 case 'e': return ESC;
|
|
|
405 case 'b': return BS;
|
|
|
406 }
|
|
|
407 return c;
|
|
|
408 }
|
|
|
409
|
|
|
410 /*
|
|
|
411 * Check for a character class name. "pp" points to the '['.
|
|
|
412 * Returns one of the CLASS_ items. CLASS_NONE means that no item was
|
|
|
413 * recognized. Otherwise "pp" is advanced to after the item.
|
|
|
414 */
|
|
|
415 static int
|
|
|
416 skip_class_name(pp)
|
|
|
417 char_u **pp;
|
|
|
418 {
|
|
|
419 static const char *(class_names[]) =
|
|
|
420 {
|
|
|
421 "alnum:]",
|
|
|
422 #define CLASS_ALNUM 0
|
|
|
423 "alpha:]",
|
|
|
424 #define CLASS_ALPHA 1
|
|
|
425 "blank:]",
|
|
|
426 #define CLASS_BLANK 2
|
|
|
427 "cntrl:]",
|
|
|
428 #define CLASS_CNTRL 3
|
|
|
429 "digit:]",
|
|
|
430 #define CLASS_DIGIT 4
|
|
|
431 "graph:]",
|
|
|
432 #define CLASS_GRAPH 5
|
|
|
433 "lower:]",
|
|
|
434 #define CLASS_LOWER 6
|
|
|
435 "print:]",
|
|
|
436 #define CLASS_PRINT 7
|
|
|
437 "punct:]",
|
|
|
438 #define CLASS_PUNCT 8
|
|
|
439 "space:]",
|
|
|
440 #define CLASS_SPACE 9
|
|
|
441 "upper:]",
|
|
|
442 #define CLASS_UPPER 10
|
|
|
443 "xdigit:]",
|
|
|
444 #define CLASS_XDIGIT 11
|
|
|
445 "tab:]",
|
|
|
446 #define CLASS_TAB 12
|
|
|
447 "return:]",
|
|
|
448 #define CLASS_RETURN 13
|
|
|
449 "backspace:]",
|
|
|
450 #define CLASS_BACKSPACE 14
|
|
|
451 "escape:]",
|
|
|
452 #define CLASS_ESCAPE 15
|
|
|
453 };
|
|
|
454 #define CLASS_NONE 99
|
|
|
455 int i;
|
|
|
456
|
|
|
457 if ((*pp)[1] == ':')
|
|
|
458 {
|
|
|
459 for (i = 0; i < sizeof(class_names) / sizeof(*class_names); ++i)
|
|
|
460 if (STRNCMP(*pp + 2, class_names[i], STRLEN(class_names[i])) == 0)
|
|
|
461 {
|
|
|
462 *pp += STRLEN(class_names[i]) + 2;
|
|
|
463 return i;
|
|
|
464 }
|
|
|
465 }
|
|
|
466 return CLASS_NONE;
|
|
|
467 }
|
|
|
468
|
|
|
469 /*
|
|
|
470 * Skip over a "[]" range.
|
|
|
471 * "p" must point to the character after the '['.
|
|
|
472 * The returned pointer is on the matching ']', or the terminating NUL.
|
|
|
473 */
|
|
|
474 static char_u *
|
|
|
475 skip_anyof(p)
|
|
|
476 char_u *p;
|
|
|
477 {
|
|
|
478 int cpo_lit; /* 'cpoptions' contains 'l' flag */
|
|
|
479 #ifdef FEAT_MBYTE
|
|
|
480 int l;
|
|
|
481 #endif
|
|
|
482
|
|
|
483 cpo_lit = (!reg_syn && vim_strchr(p_cpo, CPO_LITERAL) != NULL);
|
|
|
484
|
|
|
485 if (*p == '^') /* Complement of range. */
|
|
|
486 ++p;
|
|
|
487 if (*p == ']' || *p == '-')
|
|
|
488 ++p;
|
|
|
489 while (*p != NUL && *p != ']')
|
|
|
490 {
|
|
|
491 #ifdef FEAT_MBYTE
|
|
|
492 if (has_mbyte && (l = (*mb_ptr2len_check)(p)) > 1)
|
|
|
493 p += l;
|
|
|
494 else
|
|
|
495 #endif
|
|
|
496 if (*p == '-')
|
|
|
497 {
|
|
|
498 ++p;
|
|
|
499 if (*p != ']' && *p != NUL)
|
|
|
500 {
|
|
|
501 #ifdef FEAT_MBYTE
|
|
|
502 if (has_mbyte)
|
|
|
503 p += (*mb_ptr2len_check)(p);
|
|
|
504 else
|
|
|
505 #endif
|
|
|
506 ++p;
|
|
|
507 }
|
|
|
508 }
|
|
|
509 else if (*p == '\\'
|
|
|
510 && (vim_strchr(REGEXP_INRANGE, p[1]) != NULL
|
|
|
511 || (!cpo_lit && vim_strchr(REGEXP_ABBR, p[1]) != NULL)))
|
|
|
512 p += 2;
|
|
|
513 else if (*p == '[')
|
|
|
514 {
|
|
|
515 if (skip_class_name(&p) == CLASS_NONE)
|
|
|
516 ++p; /* It was not a class name */
|
|
|
517 }
|
|
|
518 else
|
|
|
519 ++p;
|
|
|
520 }
|
|
|
521
|
|
|
522 return p;
|
|
|
523 }
|
|
|
524
|
|
|
525 /*
|
|
|
526 * Specific version of character class functions.
|
|
|
527 * Using a table to keep this fast.
|
|
|
528 */
|
|
|
529 static short class_tab[256];
|
|
|
530
|
|
|
531 #define RI_DIGIT 0x01
|
|
|
532 #define RI_HEX 0x02
|
|
|
533 #define RI_OCTAL 0x04
|
|
|
534 #define RI_WORD 0x08
|
|
|
535 #define RI_HEAD 0x10
|
|
|
536 #define RI_ALPHA 0x20
|
|
|
537 #define RI_LOWER 0x40
|
|
|
538 #define RI_UPPER 0x80
|
|
|
539 #define RI_WHITE 0x100
|
|
|
540
|
|
|
541 static void
|
|
|
542 init_class_tab()
|
|
|
543 {
|
|
|
544 int i;
|
|
|
545 static int done = FALSE;
|
|
|
546
|
|
|
547 if (done)
|
|
|
548 return;
|
|
|
549
|
|
|
550 for (i = 0; i < 256; ++i)
|
|
|
551 {
|
|
|
552 if (i >= '0' && i <= '7')
|
|
|
553 class_tab[i] = RI_DIGIT + RI_HEX + RI_OCTAL + RI_WORD;
|
|
|
554 else if (i >= '8' && i <= '9')
|
|
|
555 class_tab[i] = RI_DIGIT + RI_HEX + RI_WORD;
|
|
|
556 else if (i >= 'a' && i <= 'f')
|
|
|
557 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER;
|
|
|
558 #ifdef EBCDIC
|
|
|
559 else if ((i >= 'g' && i <= 'i') || (i >= 'j' && i <= 'r')
|
|
|
560 || (i >= 's' && i <= 'z'))
|
|
|
561 #else
|
|
|
562 else if (i >= 'g' && i <= 'z')
|
|
|
563 #endif
|
|
|
564 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER;
|
|
|
565 else if (i >= 'A' && i <= 'F')
|
|
|
566 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER;
|
|
|
567 #ifdef EBCDIC
|
|
|
568 else if ((i >= 'G' && i <= 'I') || ( i >= 'J' && i <= 'R')
|
|
|
569 || (i >= 'S' && i <= 'Z'))
|
|
|
570 #else
|
|
|
571 else if (i >= 'G' && i <= 'Z')
|
|
|
572 #endif
|
|
|
573 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER;
|
|
|
574 else if (i == '_')
|
|
|
575 class_tab[i] = RI_WORD + RI_HEAD;
|
|
|
576 else
|
|
|
577 class_tab[i] = 0;
|
|
|
578 }
|
|
|
579 class_tab[' '] |= RI_WHITE;
|
|
|
580 class_tab['\t'] |= RI_WHITE;
|
|
|
581 done = TRUE;
|
|
|
582 }
|
|
|
583
|
|
|
584 #ifdef FEAT_MBYTE
|
|
|
585 # define ri_digit(c) (c < 0x100 && (class_tab[c] & RI_DIGIT))
|
|
|
586 # define ri_hex(c) (c < 0x100 && (class_tab[c] & RI_HEX))
|
|
|
587 # define ri_octal(c) (c < 0x100 && (class_tab[c] & RI_OCTAL))
|
|
|
588 # define ri_word(c) (c < 0x100 && (class_tab[c] & RI_WORD))
|
|
|
589 # define ri_head(c) (c < 0x100 && (class_tab[c] & RI_HEAD))
|
|
|
590 # define ri_alpha(c) (c < 0x100 && (class_tab[c] & RI_ALPHA))
|
|
|
591 # define ri_lower(c) (c < 0x100 && (class_tab[c] & RI_LOWER))
|
|
|
592 # define ri_upper(c) (c < 0x100 && (class_tab[c] & RI_UPPER))
|
|
|
593 # define ri_white(c) (c < 0x100 && (class_tab[c] & RI_WHITE))
|
|
|
594 #else
|
|
|
595 # define ri_digit(c) (class_tab[c] & RI_DIGIT)
|
|
|
596 # define ri_hex(c) (class_tab[c] & RI_HEX)
|
|
|
597 # define ri_octal(c) (class_tab[c] & RI_OCTAL)
|
|
|
598 # define ri_word(c) (class_tab[c] & RI_WORD)
|
|
|
599 # define ri_head(c) (class_tab[c] & RI_HEAD)
|
|
|
600 # define ri_alpha(c) (class_tab[c] & RI_ALPHA)
|
|
|
601 # define ri_lower(c) (class_tab[c] & RI_LOWER)
|
|
|
602 # define ri_upper(c) (class_tab[c] & RI_UPPER)
|
|
|
603 # define ri_white(c) (class_tab[c] & RI_WHITE)
|
|
|
604 #endif
|
|
|
605
|
|
|
606 /* flags for regflags */
|
|
|
607 #define RF_ICASE 1 /* ignore case */
|
|
|
608 #define RF_NOICASE 2 /* don't ignore case */
|
|
|
609 #define RF_HASNL 4 /* can match a NL */
|
|
|
610 #define RF_ICOMBINE 8 /* ignore combining characters */
|
|
|
611 #define RF_LOOKBH 16 /* uses "\@<=" or "\@<!" */
|
|
|
612
|
|
|
613 /*
|
|
|
614 * Global work variables for vim_regcomp().
|
|
|
615 */
|
|
|
616
|
|
|
617 static char_u *regparse; /* Input-scan pointer. */
|
|
|
618 static int prevchr_len; /* byte length of previous char */
|
|
|
619 static int num_complex_braces; /* Complex \{...} count */
|
|
|
620 static int regnpar; /* () count. */
|
|
|
621 #ifdef FEAT_SYN_HL
|
|
|
622 static int regnzpar; /* \z() count. */
|
|
|
623 static int re_has_z; /* \z item detected */
|
|
|
624 #endif
|
|
|
625 static char_u *regcode; /* Code-emit pointer, or JUST_CALC_SIZE */
|
|
|
626 static long regsize; /* Code size. */
|
|
|
627 static char_u had_endbrace[NSUBEXP]; /* flags, TRUE if end of () found */
|
|
|
628 static unsigned regflags; /* RF_ flags for prog */
|
|
|
629 static long brace_min[10]; /* Minimums for complex brace repeats */
|
|
|
630 static long brace_max[10]; /* Maximums for complex brace repeats */
|
|
|
631 static int brace_count[10]; /* Current counts for complex brace repeats */
|
|
|
632 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
633 static int had_eol; /* TRUE when EOL found by vim_regcomp() */
|
|
|
634 #endif
|
|
|
635 static int one_exactly = FALSE; /* only do one char for EXACTLY */
|
|
|
636
|
|
|
637 static int reg_magic; /* magicness of the pattern: */
|
|
|
638 #define MAGIC_NONE 1 /* "\V" very unmagic */
|
|
|
639 #define MAGIC_OFF 2 /* "\M" or 'magic' off */
|
|
|
640 #define MAGIC_ON 3 /* "\m" or 'magic' */
|
|
|
641 #define MAGIC_ALL 4 /* "\v" very magic */
|
|
|
642
|
|
|
643 static int reg_string; /* matching with a string instead of a buffer
|
|
|
644 line */
|
|
|
645
|
|
|
646 /*
|
|
|
647 * META contains all characters that may be magic, except '^' and '$'.
|
|
|
648 */
|
|
|
649
|
|
|
650 #ifdef EBCDIC
|
|
|
651 static char_u META[] = "%&()*+.123456789<=>?@ACDFHIKLMOPSUVWX[_acdfhiklmnopsuvwxz{|~";
|
|
|
652 #else
|
|
|
653 /* META[] is used often enough to justify turning it into a table. */
|
|
|
654 static char_u META_flags[] = {
|
|
|
655 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
|
656 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
|
|
|
657 /* % & ( ) * + . */
|
|
|
658 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0,
|
|
|
659 /* 1 2 3 4 5 6 7 8 9 < = > ? */
|
|
|
660 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1,
|
|
|
661 /* @ A C D F H I K L M O */
|
|
|
662 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1,
|
|
|
663 /* P S U V W X Z [ _ */
|
|
|
664 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1,
|
|
|
665 /* a c d f h i k l m n o */
|
|
|
666 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1,
|
|
|
667 /* p s u v w x z { | ~ */
|
|
|
668 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1
|
|
|
669 };
|
|
|
670 #endif
|
|
|
671
|
|
|
672 static int curchr;
|
|
|
673
|
|
|
674 /* arguments for reg() */
|
|
|
675 #define REG_NOPAREN 0 /* toplevel reg() */
|
|
|
676 #define REG_PAREN 1 /* \(\) */
|
|
|
677 #define REG_ZPAREN 2 /* \z(\) */
|
|
|
678 #define REG_NPAREN 3 /* \%(\) */
|
|
|
679
|
|
|
680 /*
|
|
|
681 * Forward declarations for vim_regcomp()'s friends.
|
|
|
682 */
|
|
|
683 static void initchr __ARGS((char_u *));
|
|
|
684 static int getchr __ARGS((void));
|
|
|
685 static void skipchr_keepstart __ARGS((void));
|
|
|
686 static int peekchr __ARGS((void));
|
|
|
687 static void skipchr __ARGS((void));
|
|
|
688 static void ungetchr __ARGS((void));
|
|
24
|
689 static int gethexchrs __ARGS((int maxinputlen));
|
|
|
690 static int getoctchrs __ARGS((void));
|
|
|
691 static int getdecchrs __ARGS((void));
|
|
|
692 static int coll_get_char __ARGS((void));
|
|
7
|
693 static void regcomp_start __ARGS((char_u *expr, int flags));
|
|
|
694 static char_u *reg __ARGS((int, int *));
|
|
|
695 static char_u *regbranch __ARGS((int *flagp));
|
|
|
696 static char_u *regconcat __ARGS((int *flagp));
|
|
|
697 static char_u *regpiece __ARGS((int *));
|
|
|
698 static char_u *regatom __ARGS((int *));
|
|
|
699 static char_u *regnode __ARGS((int));
|
|
|
700 static int prog_magic_wrong __ARGS((void));
|
|
|
701 static char_u *regnext __ARGS((char_u *));
|
|
|
702 static void regc __ARGS((int b));
|
|
|
703 #ifdef FEAT_MBYTE
|
|
|
704 static void regmbc __ARGS((int c));
|
|
|
705 #endif
|
|
|
706 static void reginsert __ARGS((int, char_u *));
|
|
|
707 static void reginsert_limits __ARGS((int, long, long, char_u *));
|
|
|
708 static char_u *re_put_long __ARGS((char_u *pr, long_u val));
|
|
|
709 static int read_limits __ARGS((long *, long *));
|
|
|
710 static void regtail __ARGS((char_u *, char_u *));
|
|
|
711 static void regoptail __ARGS((char_u *, char_u *));
|
|
|
712
|
|
|
713 /*
|
|
|
714 * Return TRUE if compiled regular expression "prog" can match a line break.
|
|
|
715 */
|
|
|
716 int
|
|
|
717 re_multiline(prog)
|
|
|
718 regprog_T *prog;
|
|
|
719 {
|
|
|
720 return (prog->regflags & RF_HASNL);
|
|
|
721 }
|
|
|
722
|
|
|
723 /*
|
|
|
724 * Return TRUE if compiled regular expression "prog" looks before the start
|
|
|
725 * position (pattern contains "\@<=" or "\@<!").
|
|
|
726 */
|
|
|
727 int
|
|
|
728 re_lookbehind(prog)
|
|
|
729 regprog_T *prog;
|
|
|
730 {
|
|
|
731 return (prog->regflags & RF_LOOKBH);
|
|
|
732 }
|
|
|
733
|
|
|
734 /*
|
|
|
735 * Skip past regular expression.
|
|
|
736 * Stop at end of 'p' of where 'dirc' is found ('/', '?', etc).
|
|
|
737 * Take care of characters with a backslash in front of it.
|
|
|
738 * Skip strings inside [ and ].
|
|
|
739 * When "newp" is not NULL and "dirc" is '?', make an allocated copy of the
|
|
|
740 * expression and change "\?" to "?". If "*newp" is not NULL the expression
|
|
|
741 * is changed in-place.
|
|
|
742 */
|
|
|
743 char_u *
|
|
|
744 skip_regexp(startp, dirc, magic, newp)
|
|
|
745 char_u *startp;
|
|
|
746 int dirc;
|
|
|
747 int magic;
|
|
|
748 char_u **newp;
|
|
|
749 {
|
|
|
750 int mymagic;
|
|
|
751 char_u *p = startp;
|
|
|
752
|
|
|
753 if (magic)
|
|
|
754 mymagic = MAGIC_ON;
|
|
|
755 else
|
|
|
756 mymagic = MAGIC_OFF;
|
|
|
757
|
|
|
758 for (; p[0] != NUL; ++p)
|
|
|
759 {
|
|
|
760 if (p[0] == dirc) /* found end of regexp */
|
|
|
761 break;
|
|
|
762 if ((p[0] == '[' && mymagic >= MAGIC_ON)
|
|
|
763 || (p[0] == '\\' && p[1] == '[' && mymagic <= MAGIC_OFF))
|
|
|
764 {
|
|
|
765 p = skip_anyof(p + 1);
|
|
|
766 if (p[0] == NUL)
|
|
|
767 break;
|
|
|
768 }
|
|
|
769 else if (p[0] == '\\' && p[1] != NUL)
|
|
|
770 {
|
|
|
771 if (dirc == '?' && newp != NULL && p[1] == '?')
|
|
|
772 {
|
|
|
773 /* change "\?" to "?", make a copy first. */
|
|
|
774 if (*newp == NULL)
|
|
|
775 {
|
|
|
776 *newp = vim_strsave(startp);
|
|
|
777 if (*newp != NULL)
|
|
|
778 p = *newp + (p - startp);
|
|
|
779 }
|
|
|
780 if (*newp != NULL)
|
|
|
781 mch_memmove(p, p + 1, STRLEN(p));
|
|
|
782 else
|
|
|
783 ++p;
|
|
|
784 }
|
|
|
785 else
|
|
|
786 ++p; /* skip next character */
|
|
|
787 if (*p == 'v')
|
|
|
788 mymagic = MAGIC_ALL;
|
|
|
789 else if (*p == 'V')
|
|
|
790 mymagic = MAGIC_NONE;
|
|
|
791 }
|
|
|
792 #ifdef FEAT_MBYTE
|
|
|
793 else if (has_mbyte)
|
|
|
794 p += (*mb_ptr2len_check)(p) - 1;
|
|
|
795 #endif
|
|
|
796 }
|
|
|
797 return p;
|
|
|
798 }
|
|
|
799
|
|
|
800 /*
|
|
|
801 * vim_regcomp - compile a regular expression into internal code
|
|
|
802 *
|
|
|
803 * We can't allocate space until we know how big the compiled form will be,
|
|
|
804 * but we can't compile it (and thus know how big it is) until we've got a
|
|
|
805 * place to put the code. So we cheat: we compile it twice, once with code
|
|
|
806 * generation turned off and size counting turned on, and once "for real".
|
|
|
807 * This also means that we don't allocate space until we are sure that the
|
|
|
808 * thing really will compile successfully, and we never have to move the
|
|
|
809 * code and thus invalidate pointers into it. (Note that it has to be in
|
|
|
810 * one piece because vim_free() must be able to free it all.)
|
|
|
811 *
|
|
|
812 * Whether upper/lower case is to be ignored is decided when executing the
|
|
|
813 * program, it does not matter here.
|
|
|
814 *
|
|
|
815 * Beware that the optimization-preparation code in here knows about some
|
|
|
816 * of the structure of the compiled regexp.
|
|
|
817 * "re_flags": RE_MAGIC and/or RE_STRING.
|
|
|
818 */
|
|
|
819 regprog_T *
|
|
|
820 vim_regcomp(expr, re_flags)
|
|
|
821 char_u *expr;
|
|
|
822 int re_flags;
|
|
|
823 {
|
|
|
824 regprog_T *r;
|
|
|
825 char_u *scan;
|
|
|
826 char_u *longest;
|
|
|
827 int len;
|
|
|
828 int flags;
|
|
|
829
|
|
|
830 if (expr == NULL)
|
|
|
831 EMSG_RET_NULL(_(e_null));
|
|
|
832
|
|
|
833 init_class_tab();
|
|
|
834
|
|
|
835 /*
|
|
|
836 * First pass: determine size, legality.
|
|
|
837 */
|
|
|
838 regcomp_start(expr, re_flags);
|
|
|
839 regcode = JUST_CALC_SIZE;
|
|
|
840 regc(REGMAGIC);
|
|
|
841 if (reg(REG_NOPAREN, &flags) == NULL)
|
|
|
842 return NULL;
|
|
|
843
|
|
|
844 /* Small enough for pointer-storage convention? */
|
|
|
845 #ifdef SMALL_MALLOC /* 16 bit storage allocation */
|
|
|
846 if (regsize >= 65536L - 256L)
|
|
|
847 EMSG_RET_NULL(_("E339: Pattern too long"));
|
|
|
848 #endif
|
|
|
849
|
|
|
850 /* Allocate space. */
|
|
|
851 r = (regprog_T *)lalloc(sizeof(regprog_T) + regsize, TRUE);
|
|
|
852 if (r == NULL)
|
|
|
853 return NULL;
|
|
|
854
|
|
|
855 /*
|
|
|
856 * Second pass: emit code.
|
|
|
857 */
|
|
|
858 regcomp_start(expr, re_flags);
|
|
|
859 regcode = r->program;
|
|
|
860 regc(REGMAGIC);
|
|
|
861 if (reg(REG_NOPAREN, &flags) == NULL)
|
|
|
862 {
|
|
|
863 vim_free(r);
|
|
|
864 return NULL;
|
|
|
865 }
|
|
|
866
|
|
|
867 /* Dig out information for optimizations. */
|
|
|
868 r->regstart = NUL; /* Worst-case defaults. */
|
|
|
869 r->reganch = 0;
|
|
|
870 r->regmust = NULL;
|
|
|
871 r->regmlen = 0;
|
|
|
872 r->regflags = regflags;
|
|
|
873 if (flags & HASNL)
|
|
|
874 r->regflags |= RF_HASNL;
|
|
|
875 if (flags & HASLOOKBH)
|
|
|
876 r->regflags |= RF_LOOKBH;
|
|
|
877 #ifdef FEAT_SYN_HL
|
|
|
878 /* Remember whether this pattern has any \z specials in it. */
|
|
|
879 r->reghasz = re_has_z;
|
|
|
880 #endif
|
|
|
881 scan = r->program + 1; /* First BRANCH. */
|
|
|
882 if (OP(regnext(scan)) == END) /* Only one top-level choice. */
|
|
|
883 {
|
|
|
884 scan = OPERAND(scan);
|
|
|
885
|
|
|
886 /* Starting-point info. */
|
|
|
887 if (OP(scan) == BOL || OP(scan) == RE_BOF)
|
|
|
888 {
|
|
|
889 r->reganch++;
|
|
|
890 scan = regnext(scan);
|
|
|
891 }
|
|
|
892
|
|
|
893 if (OP(scan) == EXACTLY)
|
|
|
894 {
|
|
|
895 #ifdef FEAT_MBYTE
|
|
|
896 if (has_mbyte)
|
|
|
897 r->regstart = (*mb_ptr2char)(OPERAND(scan));
|
|
|
898 else
|
|
|
899 #endif
|
|
|
900 r->regstart = *OPERAND(scan);
|
|
|
901 }
|
|
|
902 else if ((OP(scan) == BOW
|
|
|
903 || OP(scan) == EOW
|
|
|
904 || OP(scan) == NOTHING
|
|
|
905 || OP(scan) == MOPEN + 0 || OP(scan) == NOPEN
|
|
|
906 || OP(scan) == MCLOSE + 0 || OP(scan) == NCLOSE)
|
|
|
907 && OP(regnext(scan)) == EXACTLY)
|
|
|
908 {
|
|
|
909 #ifdef FEAT_MBYTE
|
|
|
910 if (has_mbyte)
|
|
|
911 r->regstart = (*mb_ptr2char)(OPERAND(regnext(scan)));
|
|
|
912 else
|
|
|
913 #endif
|
|
|
914 r->regstart = *OPERAND(regnext(scan));
|
|
|
915 }
|
|
|
916
|
|
|
917 /*
|
|
|
918 * If there's something expensive in the r.e., find the longest
|
|
|
919 * literal string that must appear and make it the regmust. Resolve
|
|
|
920 * ties in favor of later strings, since the regstart check works
|
|
|
921 * with the beginning of the r.e. and avoiding duplication
|
|
|
922 * strengthens checking. Not a strong reason, but sufficient in the
|
|
|
923 * absence of others.
|
|
|
924 */
|
|
|
925 /*
|
|
|
926 * When the r.e. starts with BOW, it is faster to look for a regmust
|
|
|
927 * first. Used a lot for "#" and "*" commands. (Added by mool).
|
|
|
928 */
|
|
|
929 if ((flags & SPSTART || OP(scan) == BOW || OP(scan) == EOW)
|
|
|
930 && !(flags & HASNL))
|
|
|
931 {
|
|
|
932 longest = NULL;
|
|
|
933 len = 0;
|
|
|
934 for (; scan != NULL; scan = regnext(scan))
|
|
|
935 if (OP(scan) == EXACTLY && STRLEN(OPERAND(scan)) >= (size_t)len)
|
|
|
936 {
|
|
|
937 longest = OPERAND(scan);
|
|
|
938 len = (int)STRLEN(OPERAND(scan));
|
|
|
939 }
|
|
|
940 r->regmust = longest;
|
|
|
941 r->regmlen = len;
|
|
|
942 }
|
|
|
943 }
|
|
|
944 #ifdef DEBUG
|
|
|
945 regdump(expr, r);
|
|
|
946 #endif
|
|
|
947 return r;
|
|
|
948 }
|
|
|
949
|
|
|
950 /*
|
|
|
951 * Setup to parse the regexp. Used once to get the length and once to do it.
|
|
|
952 */
|
|
|
953 static void
|
|
|
954 regcomp_start(expr, re_flags)
|
|
|
955 char_u *expr;
|
|
|
956 int re_flags; /* see vim_regcomp() */
|
|
|
957 {
|
|
|
958 initchr(expr);
|
|
|
959 if (re_flags & RE_MAGIC)
|
|
|
960 reg_magic = MAGIC_ON;
|
|
|
961 else
|
|
|
962 reg_magic = MAGIC_OFF;
|
|
|
963 reg_string = (re_flags & RE_STRING);
|
|
|
964
|
|
|
965 num_complex_braces = 0;
|
|
|
966 regnpar = 1;
|
|
|
967 vim_memset(had_endbrace, 0, sizeof(had_endbrace));
|
|
|
968 #ifdef FEAT_SYN_HL
|
|
|
969 regnzpar = 1;
|
|
|
970 re_has_z = 0;
|
|
|
971 #endif
|
|
|
972 regsize = 0L;
|
|
|
973 regflags = 0;
|
|
|
974 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
975 had_eol = FALSE;
|
|
|
976 #endif
|
|
|
977 }
|
|
|
978
|
|
|
979 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
980 /*
|
|
|
981 * Check if during the previous call to vim_regcomp the EOL item "$" has been
|
|
|
982 * found. This is messy, but it works fine.
|
|
|
983 */
|
|
|
984 int
|
|
|
985 vim_regcomp_had_eol()
|
|
|
986 {
|
|
|
987 return had_eol;
|
|
|
988 }
|
|
|
989 #endif
|
|
|
990
|
|
|
991 /*
|
|
|
992 * reg - regular expression, i.e. main body or parenthesized thing
|
|
|
993 *
|
|
|
994 * Caller must absorb opening parenthesis.
|
|
|
995 *
|
|
|
996 * Combining parenthesis handling with the base level of regular expression
|
|
|
997 * is a trifle forced, but the need to tie the tails of the branches to what
|
|
|
998 * follows makes it hard to avoid.
|
|
|
999 */
|
|
|
1000 static char_u *
|
|
|
1001 reg(paren, flagp)
|
|
|
1002 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */
|
|
|
1003 int *flagp;
|
|
|
1004 {
|
|
|
1005 char_u *ret;
|
|
|
1006 char_u *br;
|
|
|
1007 char_u *ender;
|
|
|
1008 int parno = 0;
|
|
|
1009 int flags;
|
|
|
1010
|
|
|
1011 *flagp = HASWIDTH; /* Tentatively. */
|
|
|
1012
|
|
|
1013 #ifdef FEAT_SYN_HL
|
|
|
1014 if (paren == REG_ZPAREN)
|
|
|
1015 {
|
|
|
1016 /* Make a ZOPEN node. */
|
|
|
1017 if (regnzpar >= NSUBEXP)
|
|
|
1018 EMSG_RET_NULL(_("E50: Too many \\z("));
|
|
|
1019 parno = regnzpar;
|
|
|
1020 regnzpar++;
|
|
|
1021 ret = regnode(ZOPEN + parno);
|
|
|
1022 }
|
|
|
1023 else
|
|
|
1024 #endif
|
|
|
1025 if (paren == REG_PAREN)
|
|
|
1026 {
|
|
|
1027 /* Make a MOPEN node. */
|
|
|
1028 if (regnpar >= NSUBEXP)
|
|
|
1029 EMSG_M_RET_NULL(_("E51: Too many %s("), reg_magic == MAGIC_ALL);
|
|
|
1030 parno = regnpar;
|
|
|
1031 ++regnpar;
|
|
|
1032 ret = regnode(MOPEN + parno);
|
|
|
1033 }
|
|
|
1034 else if (paren == REG_NPAREN)
|
|
|
1035 {
|
|
|
1036 /* Make a NOPEN node. */
|
|
|
1037 ret = regnode(NOPEN);
|
|
|
1038 }
|
|
|
1039 else
|
|
|
1040 ret = NULL;
|
|
|
1041
|
|
|
1042 /* Pick up the branches, linking them together. */
|
|
|
1043 br = regbranch(&flags);
|
|
|
1044 if (br == NULL)
|
|
|
1045 return NULL;
|
|
|
1046 if (ret != NULL)
|
|
|
1047 regtail(ret, br); /* [MZ]OPEN -> first. */
|
|
|
1048 else
|
|
|
1049 ret = br;
|
|
|
1050 /* If one of the branches can be zero-width, the whole thing can.
|
|
|
1051 * If one of the branches has * at start or matches a line-break, the
|
|
|
1052 * whole thing can. */
|
|
|
1053 if (!(flags & HASWIDTH))
|
|
|
1054 *flagp &= ~HASWIDTH;
|
|
|
1055 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH);
|
|
|
1056 while (peekchr() == Magic('|'))
|
|
|
1057 {
|
|
|
1058 skipchr();
|
|
|
1059 br = regbranch(&flags);
|
|
|
1060 if (br == NULL)
|
|
|
1061 return NULL;
|
|
|
1062 regtail(ret, br); /* BRANCH -> BRANCH. */
|
|
|
1063 if (!(flags & HASWIDTH))
|
|
|
1064 *flagp &= ~HASWIDTH;
|
|
|
1065 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH);
|
|
|
1066 }
|
|
|
1067
|
|
|
1068 /* Make a closing node, and hook it on the end. */
|
|
|
1069 ender = regnode(
|
|
|
1070 #ifdef FEAT_SYN_HL
|
|
|
1071 paren == REG_ZPAREN ? ZCLOSE + parno :
|
|
|
1072 #endif
|
|
|
1073 paren == REG_PAREN ? MCLOSE + parno :
|
|
|
1074 paren == REG_NPAREN ? NCLOSE : END);
|
|
|
1075 regtail(ret, ender);
|
|
|
1076
|
|
|
1077 /* Hook the tails of the branches to the closing node. */
|
|
|
1078 for (br = ret; br != NULL; br = regnext(br))
|
|
|
1079 regoptail(br, ender);
|
|
|
1080
|
|
|
1081 /* Check for proper termination. */
|
|
|
1082 if (paren != REG_NOPAREN && getchr() != Magic(')'))
|
|
|
1083 {
|
|
|
1084 #ifdef FEAT_SYN_HL
|
|
|
1085 if (paren == REG_ZPAREN)
|
|
|
1086 EMSG_RET_NULL(_("E52: Unmatched \\z("))
|
|
|
1087 else
|
|
|
1088 #endif
|
|
|
1089 if (paren == REG_NPAREN)
|
|
|
1090 EMSG_M_RET_NULL(_("E53: Unmatched %s%%("), reg_magic == MAGIC_ALL)
|
|
|
1091 else
|
|
|
1092 EMSG_M_RET_NULL(_("E54: Unmatched %s("), reg_magic == MAGIC_ALL)
|
|
|
1093 }
|
|
|
1094 else if (paren == REG_NOPAREN && peekchr() != NUL)
|
|
|
1095 {
|
|
|
1096 if (curchr == Magic(')'))
|
|
|
1097 EMSG_M_RET_NULL(_("E55: Unmatched %s)"), reg_magic == MAGIC_ALL)
|
|
|
1098 else
|
|
|
1099 EMSG_RET_NULL(_(e_trailing)) /* "Can't happen". */
|
|
|
1100 /* NOTREACHED */
|
|
|
1101 }
|
|
|
1102 /*
|
|
|
1103 * Here we set the flag allowing back references to this set of
|
|
|
1104 * parentheses.
|
|
|
1105 */
|
|
|
1106 if (paren == REG_PAREN)
|
|
|
1107 had_endbrace[parno] = TRUE; /* have seen the close paren */
|
|
|
1108 return ret;
|
|
|
1109 }
|
|
|
1110
|
|
|
1111 /*
|
|
|
1112 * regbranch - one alternative of an | operator
|
|
|
1113 *
|
|
|
1114 * Implements the & operator.
|
|
|
1115 */
|
|
|
1116 static char_u *
|
|
|
1117 regbranch(flagp)
|
|
|
1118 int *flagp;
|
|
|
1119 {
|
|
|
1120 char_u *ret;
|
|
|
1121 char_u *chain = NULL;
|
|
|
1122 char_u *latest;
|
|
|
1123 int flags;
|
|
|
1124
|
|
|
1125 *flagp = WORST | HASNL; /* Tentatively. */
|
|
|
1126
|
|
|
1127 ret = regnode(BRANCH);
|
|
|
1128 for (;;)
|
|
|
1129 {
|
|
|
1130 latest = regconcat(&flags);
|
|
|
1131 if (latest == NULL)
|
|
|
1132 return NULL;
|
|
|
1133 /* If one of the branches has width, the whole thing has. If one of
|
|
|
1134 * the branches anchors at start-of-line, the whole thing does.
|
|
|
1135 * If one of the branches uses look-behind, the whole thing does. */
|
|
|
1136 *flagp |= flags & (HASWIDTH | SPSTART | HASLOOKBH);
|
|
|
1137 /* If one of the branches doesn't match a line-break, the whole thing
|
|
|
1138 * doesn't. */
|
|
|
1139 *flagp &= ~HASNL | (flags & HASNL);
|
|
|
1140 if (chain != NULL)
|
|
|
1141 regtail(chain, latest);
|
|
|
1142 if (peekchr() != Magic('&'))
|
|
|
1143 break;
|
|
|
1144 skipchr();
|
|
|
1145 regtail(latest, regnode(END)); /* operand ends */
|
|
|
1146 reginsert(MATCH, latest);
|
|
|
1147 chain = latest;
|
|
|
1148 }
|
|
|
1149
|
|
|
1150 return ret;
|
|
|
1151 }
|
|
|
1152
|
|
|
1153 /*
|
|
|
1154 * regbranch - one alternative of an | or & operator
|
|
|
1155 *
|
|
|
1156 * Implements the concatenation operator.
|
|
|
1157 */
|
|
|
1158 static char_u *
|
|
|
1159 regconcat(flagp)
|
|
|
1160 int *flagp;
|
|
|
1161 {
|
|
|
1162 char_u *first = NULL;
|
|
|
1163 char_u *chain = NULL;
|
|
|
1164 char_u *latest;
|
|
|
1165 int flags;
|
|
|
1166 int cont = TRUE;
|
|
|
1167
|
|
|
1168 *flagp = WORST; /* Tentatively. */
|
|
|
1169
|
|
|
1170 while (cont)
|
|
|
1171 {
|
|
|
1172 switch (peekchr())
|
|
|
1173 {
|
|
|
1174 case NUL:
|
|
|
1175 case Magic('|'):
|
|
|
1176 case Magic('&'):
|
|
|
1177 case Magic(')'):
|
|
|
1178 cont = FALSE;
|
|
|
1179 break;
|
|
|
1180 case Magic('Z'):
|
|
|
1181 #ifdef FEAT_MBYTE
|
|
|
1182 regflags |= RF_ICOMBINE;
|
|
|
1183 #endif
|
|
|
1184 skipchr_keepstart();
|
|
|
1185 break;
|
|
|
1186 case Magic('c'):
|
|
|
1187 regflags |= RF_ICASE;
|
|
|
1188 skipchr_keepstart();
|
|
|
1189 break;
|
|
|
1190 case Magic('C'):
|
|
|
1191 regflags |= RF_NOICASE;
|
|
|
1192 skipchr_keepstart();
|
|
|
1193 break;
|
|
|
1194 case Magic('v'):
|
|
|
1195 reg_magic = MAGIC_ALL;
|
|
|
1196 skipchr_keepstart();
|
|
|
1197 curchr = -1;
|
|
|
1198 break;
|
|
|
1199 case Magic('m'):
|
|
|
1200 reg_magic = MAGIC_ON;
|
|
|
1201 skipchr_keepstart();
|
|
|
1202 curchr = -1;
|
|
|
1203 break;
|
|
|
1204 case Magic('M'):
|
|
|
1205 reg_magic = MAGIC_OFF;
|
|
|
1206 skipchr_keepstart();
|
|
|
1207 curchr = -1;
|
|
|
1208 break;
|
|
|
1209 case Magic('V'):
|
|
|
1210 reg_magic = MAGIC_NONE;
|
|
|
1211 skipchr_keepstart();
|
|
|
1212 curchr = -1;
|
|
|
1213 break;
|
|
|
1214 default:
|
|
|
1215 latest = regpiece(&flags);
|
|
|
1216 if (latest == NULL)
|
|
|
1217 return NULL;
|
|
|
1218 *flagp |= flags & (HASWIDTH | HASNL | HASLOOKBH);
|
|
|
1219 if (chain == NULL) /* First piece. */
|
|
|
1220 *flagp |= flags & SPSTART;
|
|
|
1221 else
|
|
|
1222 regtail(chain, latest);
|
|
|
1223 chain = latest;
|
|
|
1224 if (first == NULL)
|
|
|
1225 first = latest;
|
|
|
1226 break;
|
|
|
1227 }
|
|
|
1228 }
|
|
|
1229 if (first == NULL) /* Loop ran zero times. */
|
|
|
1230 first = regnode(NOTHING);
|
|
|
1231 return first;
|
|
|
1232 }
|
|
|
1233
|
|
|
1234 /*
|
|
|
1235 * regpiece - something followed by possible [*+=]
|
|
|
1236 *
|
|
|
1237 * Note that the branching code sequences used for = and the general cases
|
|
|
1238 * of * and + are somewhat optimized: they use the same NOTHING node as
|
|
|
1239 * both the endmarker for their branch list and the body of the last branch.
|
|
|
1240 * It might seem that this node could be dispensed with entirely, but the
|
|
|
1241 * endmarker role is not redundant.
|
|
|
1242 */
|
|
|
1243 static char_u *
|
|
|
1244 regpiece(flagp)
|
|
|
1245 int *flagp;
|
|
|
1246 {
|
|
|
1247 char_u *ret;
|
|
|
1248 int op;
|
|
|
1249 char_u *next;
|
|
|
1250 int flags;
|
|
|
1251 long minval;
|
|
|
1252 long maxval;
|
|
|
1253
|
|
|
1254 ret = regatom(&flags);
|
|
|
1255 if (ret == NULL)
|
|
|
1256 return NULL;
|
|
|
1257
|
|
|
1258 op = peekchr();
|
|
|
1259 if (re_multi_type(op) == NOT_MULTI)
|
|
|
1260 {
|
|
|
1261 *flagp = flags;
|
|
|
1262 return ret;
|
|
|
1263 }
|
|
|
1264 if (!(flags & HASWIDTH) && re_multi_type(op) == MULTI_MULT)
|
|
|
1265 {
|
|
|
1266 if (op == Magic('*'))
|
|
|
1267 EMSG_M_RET_NULL(_("E56: %s* operand could be empty"),
|
|
|
1268 reg_magic >= MAGIC_ON);
|
|
|
1269 if (op == Magic('+'))
|
|
|
1270 EMSG_M_RET_NULL(_("E57: %s+ operand could be empty"),
|
|
|
1271 reg_magic == MAGIC_ALL);
|
|
|
1272 /* "\{}" is checked below, it's allowed when there is an upper limit */
|
|
|
1273 }
|
|
|
1274 /* default flags */
|
|
|
1275 *flagp = (WORST | SPSTART | (flags & (HASNL | HASLOOKBH)));
|
|
|
1276
|
|
|
1277 skipchr();
|
|
|
1278 switch (op)
|
|
|
1279 {
|
|
|
1280 case Magic('*'):
|
|
|
1281 if (flags & SIMPLE)
|
|
|
1282 reginsert(STAR, ret);
|
|
|
1283 else
|
|
|
1284 {
|
|
|
1285 /* Emit x* as (x&|), where & means "self". */
|
|
|
1286 reginsert(BRANCH, ret); /* Either x */
|
|
|
1287 regoptail(ret, regnode(BACK)); /* and loop */
|
|
|
1288 regoptail(ret, ret); /* back */
|
|
|
1289 regtail(ret, regnode(BRANCH)); /* or */
|
|
|
1290 regtail(ret, regnode(NOTHING)); /* null. */
|
|
|
1291 }
|
|
|
1292 break;
|
|
|
1293
|
|
|
1294 case Magic('+'):
|
|
|
1295 if (flags & SIMPLE)
|
|
|
1296 reginsert(PLUS, ret);
|
|
|
1297 else
|
|
|
1298 {
|
|
|
1299 /* Emit x+ as x(&|), where & means "self". */
|
|
|
1300 next = regnode(BRANCH); /* Either */
|
|
|
1301 regtail(ret, next);
|
|
|
1302 regtail(regnode(BACK), ret); /* loop back */
|
|
|
1303 regtail(next, regnode(BRANCH)); /* or */
|
|
|
1304 regtail(ret, regnode(NOTHING)); /* null. */
|
|
|
1305 }
|
|
|
1306 *flagp = (WORST | HASWIDTH | (flags & (HASNL | HASLOOKBH)));
|
|
|
1307 break;
|
|
|
1308
|
|
|
1309 case Magic('@'):
|
|
|
1310 {
|
|
|
1311 int lop = END;
|
|
|
1312
|
|
|
1313 switch (no_Magic(getchr()))
|
|
|
1314 {
|
|
|
1315 case '=': lop = MATCH; break; /* \@= */
|
|
|
1316 case '!': lop = NOMATCH; break; /* \@! */
|
|
|
1317 case '>': lop = SUBPAT; break; /* \@> */
|
|
|
1318 case '<': switch (no_Magic(getchr()))
|
|
|
1319 {
|
|
|
1320 case '=': lop = BEHIND; break; /* \@<= */
|
|
|
1321 case '!': lop = NOBEHIND; break; /* \@<! */
|
|
|
1322 }
|
|
|
1323 }
|
|
|
1324 if (lop == END)
|
|
|
1325 EMSG_M_RET_NULL(_("E59: invalid character after %s@"),
|
|
|
1326 reg_magic == MAGIC_ALL);
|
|
|
1327 /* Look behind must match with behind_pos. */
|
|
|
1328 if (lop == BEHIND || lop == NOBEHIND)
|
|
|
1329 {
|
|
|
1330 regtail(ret, regnode(BHPOS));
|
|
|
1331 *flagp |= HASLOOKBH;
|
|
|
1332 }
|
|
|
1333 regtail(ret, regnode(END)); /* operand ends */
|
|
|
1334 reginsert(lop, ret);
|
|
|
1335 break;
|
|
|
1336 }
|
|
|
1337
|
|
|
1338 case Magic('?'):
|
|
|
1339 case Magic('='):
|
|
|
1340 /* Emit x= as (x|) */
|
|
|
1341 reginsert(BRANCH, ret); /* Either x */
|
|
|
1342 regtail(ret, regnode(BRANCH)); /* or */
|
|
|
1343 next = regnode(NOTHING); /* null. */
|
|
|
1344 regtail(ret, next);
|
|
|
1345 regoptail(ret, next);
|
|
|
1346 break;
|
|
|
1347
|
|
|
1348 case Magic('{'):
|
|
|
1349 if (!read_limits(&minval, &maxval))
|
|
|
1350 return NULL;
|
|
|
1351 if (!(flags & HASWIDTH) && (maxval > minval
|
|
|
1352 ? maxval >= MAX_LIMIT : minval >= MAX_LIMIT))
|
|
|
1353 EMSG_M_RET_NULL(_("E58: %s{ operand could be empty"),
|
|
|
1354 reg_magic == MAGIC_ALL);
|
|
|
1355 if (flags & SIMPLE)
|
|
|
1356 {
|
|
|
1357 reginsert(BRACE_SIMPLE, ret);
|
|
|
1358 reginsert_limits(BRACE_LIMITS, minval, maxval, ret);
|
|
|
1359 }
|
|
|
1360 else
|
|
|
1361 {
|
|
|
1362 if (num_complex_braces >= 10)
|
|
|
1363 EMSG_M_RET_NULL(_("E60: Too many complex %s{...}s"),
|
|
|
1364 reg_magic == MAGIC_ALL);
|
|
|
1365 reginsert(BRACE_COMPLEX + num_complex_braces, ret);
|
|
|
1366 regoptail(ret, regnode(BACK));
|
|
|
1367 regoptail(ret, ret);
|
|
|
1368 reginsert_limits(BRACE_LIMITS, minval, maxval, ret);
|
|
|
1369 ++num_complex_braces;
|
|
|
1370 }
|
|
|
1371 if (minval > 0 && maxval > 0)
|
|
|
1372 *flagp = (HASWIDTH | (flags & (HASNL | HASLOOKBH)));
|
|
|
1373 break;
|
|
|
1374 }
|
|
|
1375 if (re_multi_type(peekchr()) != NOT_MULTI)
|
|
|
1376 {
|
|
|
1377 /* Can't have a multi follow a multi. */
|
|
|
1378 if (peekchr() == Magic('*'))
|
|
|
1379 sprintf((char *)IObuff, _("E61: Nested %s*"),
|
|
|
1380 reg_magic >= MAGIC_ON ? "" : "\\");
|
|
|
1381 else
|
|
|
1382 sprintf((char *)IObuff, _("E62: Nested %s%c"),
|
|
|
1383 reg_magic == MAGIC_ALL ? "" : "\\", no_Magic(peekchr()));
|
|
|
1384 EMSG_RET_NULL(IObuff);
|
|
|
1385 }
|
|
|
1386
|
|
|
1387 return ret;
|
|
|
1388 }
|
|
|
1389
|
|
|
1390 /*
|
|
|
1391 * regatom - the lowest level
|
|
|
1392 *
|
|
|
1393 * Optimization: gobbles an entire sequence of ordinary characters so that
|
|
|
1394 * it can turn them into a single node, which is smaller to store and
|
|
|
1395 * faster to run. Don't do this when one_exactly is set.
|
|
|
1396 */
|
|
|
1397 static char_u *
|
|
|
1398 regatom(flagp)
|
|
|
1399 int *flagp;
|
|
|
1400 {
|
|
|
1401 char_u *ret;
|
|
|
1402 int flags;
|
|
|
1403 int cpo_lit; /* 'cpoptions' contains 'l' flag */
|
|
|
1404 int c;
|
|
|
1405 static char_u *classchars = (char_u *)".iIkKfFpPsSdDxXoOwWhHaAlLuU";
|
|
|
1406 static int classcodes[] = {ANY, IDENT, SIDENT, KWORD, SKWORD,
|
|
|
1407 FNAME, SFNAME, PRINT, SPRINT,
|
|
|
1408 WHITE, NWHITE, DIGIT, NDIGIT,
|
|
|
1409 HEX, NHEX, OCTAL, NOCTAL,
|
|
|
1410 WORD, NWORD, HEAD, NHEAD,
|
|
|
1411 ALPHA, NALPHA, LOWER, NLOWER,
|
|
|
1412 UPPER, NUPPER
|
|
|
1413 };
|
|
|
1414 char_u *p;
|
|
|
1415 int extra = 0;
|
|
|
1416
|
|
|
1417 *flagp = WORST; /* Tentatively. */
|
|
|
1418 cpo_lit = (!reg_syn && vim_strchr(p_cpo, CPO_LITERAL) != NULL);
|
|
|
1419
|
|
|
1420 c = getchr();
|
|
|
1421 switch (c)
|
|
|
1422 {
|
|
|
1423 case Magic('^'):
|
|
|
1424 ret = regnode(BOL);
|
|
|
1425 break;
|
|
|
1426
|
|
|
1427 case Magic('$'):
|
|
|
1428 ret = regnode(EOL);
|
|
|
1429 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
1430 had_eol = TRUE;
|
|
|
1431 #endif
|
|
|
1432 break;
|
|
|
1433
|
|
|
1434 case Magic('<'):
|
|
|
1435 ret = regnode(BOW);
|
|
|
1436 break;
|
|
|
1437
|
|
|
1438 case Magic('>'):
|
|
|
1439 ret = regnode(EOW);
|
|
|
1440 break;
|
|
|
1441
|
|
|
1442 case Magic('_'):
|
|
|
1443 c = no_Magic(getchr());
|
|
|
1444 if (c == '^') /* "\_^" is start-of-line */
|
|
|
1445 {
|
|
|
1446 ret = regnode(BOL);
|
|
|
1447 break;
|
|
|
1448 }
|
|
|
1449 if (c == '$') /* "\_$" is end-of-line */
|
|
|
1450 {
|
|
|
1451 ret = regnode(EOL);
|
|
|
1452 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
|
1453 had_eol = TRUE;
|
|
|
1454 #endif
|
|
|
1455 break;
|
|
|
1456 }
|
|
|
1457
|
|
|
1458 extra = ADD_NL;
|
|
|
1459 *flagp |= HASNL;
|
|
|
1460
|
|
|
1461 /* "\_[" is character range plus newline */
|
|
|
1462 if (c == '[')
|
|
|
1463 goto collection;
|
|
|
1464
|
|
|
1465 /* "\_x" is character class plus newline */
|
|
|
1466 /*FALLTHROUGH*/
|
|
|
1467
|
|
|
1468 /*
|
|
|
1469 * Character classes.
|
|
|
1470 */
|
|
|
1471 case Magic('.'):
|
|
|
1472 case Magic('i'):
|
|
|
1473 case Magic('I'):
|
|
|
1474 case Magic('k'):
|
|
|
1475 case Magic('K'):
|
|
|
1476 case Magic('f'):
|
|
|
1477 case Magic('F'):
|
|
|
1478 case Magic('p'):
|
|
|
1479 case Magic('P'):
|
|
|
1480 case Magic('s'):
|
|
|
1481 case Magic('S'):
|
|
|
1482 case Magic('d'):
|
|
|
1483 case Magic('D'):
|
|
|
1484 case Magic('x'):
|
|
|
1485 case Magic('X'):
|
|
|
1486 case Magic('o'):
|
|
|
1487 case Magic('O'):
|
|
|
1488 case Magic('w'):
|
|
|
1489 case Magic('W'):
|
|
|
1490 case Magic('h'):
|
|
|
1491 case Magic('H'):
|
|
|
1492 case Magic('a'):
|
|
|
1493 case Magic('A'):
|
|
|
1494 case Magic('l'):
|
|
|
1495 case Magic('L'):
|
|
|
1496 case Magic('u'):
|
|
|
1497 case Magic('U'):
|
|
|
1498 p = vim_strchr(classchars, no_Magic(c));
|
|
|
1499 if (p == NULL)
|
|
|
1500 EMSG_RET_NULL(_("E63: invalid use of \\_"));
|
|
|
1501 ret = regnode(classcodes[p - classchars] + extra);
|
|
|
1502 *flagp |= HASWIDTH | SIMPLE;
|
|
|
1503 break;
|
|
|
1504
|
|
|
1505 case Magic('n'):
|
|
|
1506 if (reg_string)
|
|
|
1507 {
|
|
|
1508 /* In a string "\n" matches a newline character. */
|
|
|
1509 ret = regnode(EXACTLY);
|
|
|
1510 regc(NL);
|
|
|
1511 regc(NUL);
|
|
|
1512 *flagp |= HASWIDTH | SIMPLE;
|
|
|
1513 }
|
|
|
1514 else
|
|
|
1515 {
|
|
|
1516 /* In buffer text "\n" matches the end of a line. */
|
|
|
1517 ret = regnode(NEWL);
|
|
|
1518 *flagp |= HASWIDTH | HASNL;
|
|
|
1519 }
|
|
|
1520 break;
|
|
|
1521
|
|
|
1522 case Magic('('):
|
|
|
1523 if (one_exactly)
|
|
|
1524 EMSG_ONE_RET_NULL;
|
|
|
1525 ret = reg(REG_PAREN, &flags);
|
|
|
1526 if (ret == NULL)
|
|
|
1527 return NULL;
|
|
|
1528 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH);
|
|
|
1529 break;
|
|
|
1530
|
|
|
1531 case NUL:
|
|
|
1532 case Magic('|'):
|
|
|
1533 case Magic('&'):
|
|
|
1534 case Magic(')'):
|
|
|
1535 EMSG_RET_NULL(_(e_internal)); /* Supposed to be caught earlier. */
|
|
|
1536 /* NOTREACHED */
|
|
|
1537
|
|
|
1538 case Magic('='):
|
|
|
1539 case Magic('?'):
|
|
|
1540 case Magic('+'):
|
|
|
1541 case Magic('@'):
|
|
|
1542 case Magic('{'):
|
|
|
1543 case Magic('*'):
|
|
|
1544 c = no_Magic(c);
|
|
|
1545 sprintf((char *)IObuff, _("E64: %s%c follows nothing"),
|
|
|
1546 (c == '*' ? reg_magic >= MAGIC_ON : reg_magic == MAGIC_ALL)
|
|
|
1547 ? "" : "\\", c);
|
|
|
1548 EMSG_RET_NULL(IObuff);
|
|
|
1549 /* NOTREACHED */
|
|
|
1550
|
|
|
1551 case Magic('~'): /* previous substitute pattern */
|
|
|
1552 if (reg_prev_sub)
|
|
|
1553 {
|
|
|
1554 char_u *lp;
|
|
|
1555
|
|
|
1556 ret = regnode(EXACTLY);
|
|
|
1557 lp = reg_prev_sub;
|
|
|
1558 while (*lp != NUL)
|
|
|
1559 regc(*lp++);
|
|
|
1560 regc(NUL);
|
|
|
1561 if (*reg_prev_sub != NUL)
|
|
|
1562 {
|
|
|
1563 *flagp |= HASWIDTH;
|
|
|
1564 if ((lp - reg_prev_sub) == 1)
|
|
|
1565 *flagp |= SIMPLE;
|
|
|
1566 }
|
|
|
1567 }
|
|
|
1568 else
|
|
|
1569 EMSG_RET_NULL(_(e_nopresub));
|
|
|
1570 break;
|
|
|
1571
|
|
|
1572 case Magic('1'):
|
|
|
1573 case Magic('2'):
|
|
|
1574 case Magic('3'):
|
|
|
1575 case Magic('4'):
|
|
|
1576 case Magic('5'):
|
|
|
1577 case Magic('6'):
|
|
|
1578 case Magic('7'):
|
|
|
1579 case Magic('8'):
|
|
|
1580 case Magic('9'):
|
|
|
1581 {
|
|
|
1582 int refnum;
|
|
|
1583
|
|
|
1584 refnum = c - Magic('0');
|
|
|
1585 /*
|
|
|
1586 * Check if the back reference is legal. We must have seen the
|
|
|
1587 * close brace.
|
|
|
1588 * TODO: Should also check that we don't refer to something
|
|
|
1589 * that is repeated (+*=): what instance of the repetition
|
|
|
1590 * should we match?
|
|
|
1591 */
|
|
|
1592 if (!had_endbrace[refnum])
|
|
|
1593 {
|
|
|
1594 /* Trick: check if "@<=" or "@<!" follows, in which case
|
|
|
1595 * the \1 can appear before the referenced match. */
|
|
|
1596 for (p = regparse; *p != NUL; ++p)
|
|
|
1597 if (p[0] == '@' && p[1] == '<'
|
|
|
1598 && (p[2] == '!' || p[2] == '='))
|
|
|
1599 break;
|
|
|
1600 if (*p == NUL)
|
|
|
1601 EMSG_RET_NULL(_("E65: Illegal back reference"));
|
|
|
1602 }
|
|
|
1603 ret = regnode(BACKREF + refnum);
|
|
|
1604 }
|
|
|
1605 break;
|
|
|
1606
|
|
|
1607 #ifdef FEAT_SYN_HL
|
|
|
1608 case Magic('z'):
|
|
|
1609 {
|
|
|
1610 c = no_Magic(getchr());
|
|
|
1611 switch (c)
|
|
|
1612 {
|
|
|
1613 case '(': if (reg_do_extmatch != REX_SET)
|
|
|
1614 EMSG_RET_NULL(_("E66: \\z( not allowed here"));
|
|
|
1615 if (one_exactly)
|
|
|
1616 EMSG_ONE_RET_NULL;
|
|
|
1617 ret = reg(REG_ZPAREN, &flags);
|
|
|
1618 if (ret == NULL)
|
|
|
1619 return NULL;
|
|
|
1620 *flagp |= flags & (HASWIDTH|SPSTART|HASNL|HASLOOKBH);
|
|
|
1621 re_has_z = REX_SET;
|
|
|
1622 break;
|
|
|
1623
|
|
|
1624 case '1':
|
|
|
1625 case '2':
|
|
|
1626 case '3':
|
|
|
1627 case '4':
|
|
|
1628 case '5':
|
|
|
1629 case '6':
|
|
|
1630 case '7':
|
|
|
1631 case '8':
|
|
|
1632 case '9': if (reg_do_extmatch != REX_USE)
|
|
|
1633 EMSG_RET_NULL(_("E67: \\z1 et al. not allowed here"));
|
|
|
1634 ret = regnode(ZREF + c - '0');
|
|
|
1635 re_has_z = REX_USE;
|
|
|
1636 break;
|
|
|
1637
|
|
|
1638 case 's': ret = regnode(MOPEN + 0);
|
|
|
1639 break;
|
|
|
1640
|
|
|
1641 case 'e': ret = regnode(MCLOSE + 0);
|
|
|
1642 break;
|
|
|
1643
|
|
|
1644 default: EMSG_RET_NULL(_("E68: Invalid character after \\z"));
|
|
|
1645 }
|
|
|
1646 }
|
|
|
1647 break;
|
|
|
1648 #endif
|
|
|
1649
|
|
|
1650 case Magic('%'):
|
|
|
1651 {
|
|
|
1652 c = no_Magic(getchr());
|
|
|
1653 switch (c)
|
|
|
1654 {
|
|
|
1655 /* () without a back reference */
|
|
|
1656 case '(':
|
|
|
1657 if (one_exactly)
|
|
|
1658 EMSG_ONE_RET_NULL;
|
|
|
1659 ret = reg(REG_NPAREN, &flags);
|
|
|
1660 if (ret == NULL)
|
|
|
1661 return NULL;
|
|
|
1662 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH);
|
|
|
1663 break;
|
|
|
1664
|
|
|
1665 /* Catch \%^ and \%$ regardless of where they appear in the
|
|
|
1666 * pattern -- regardless of whether or not it makes sense. */
|
|
|
1667 case '^':
|
|
|
1668 ret = regnode(RE_BOF);
|
|
|
1669 break;
|
|
|
1670
|
|
|
1671 case '$':
|
|
|
1672 ret = regnode(RE_EOF);
|
|
|
1673 break;
|
|
|
1674
|
|
|
1675 case '#':
|
|
|
1676 ret = regnode(CURSOR);
|
|
|
1677 break;
|
|
|
1678
|
|
|
1679 /* \%[abc]: Emit as a list of branches, all ending at the last
|
|
|
1680 * branch which matches nothing. */
|
|
|
1681 case '[':
|
|
|
1682 if (one_exactly) /* doesn't nest */
|
|
|
1683 EMSG_ONE_RET_NULL;
|
|
|
1684 {
|
|
|
1685 char_u *lastbranch;
|
|
|
1686 char_u *lastnode = NULL;
|
|
|
1687 char_u *br;
|
|
|
1688
|
|
|
1689 ret = NULL;
|
|
|
1690 while ((c = getchr()) != ']')
|
|
|
1691 {
|
|
|
1692 if (c == NUL)
|
|
|
1693 EMSG_M_RET_NULL(_("E69: Missing ] after %s%%["),
|
|
|
1694 reg_magic == MAGIC_ALL);
|
|
|
1695 br = regnode(BRANCH);
|
|
|
1696 if (ret == NULL)
|
|
|
1697 ret = br;
|
|
|
1698 else
|
|
|
1699 regtail(lastnode, br);
|
|
|
1700
|
|
|
1701 ungetchr();
|
|
|
1702 one_exactly = TRUE;
|
|
|
1703 lastnode = regatom(flagp);
|
|
|
1704 one_exactly = FALSE;
|
|
|
1705 if (lastnode == NULL)
|
|
|
1706 return NULL;
|
|
|
1707 }
|
|
|
1708 if (ret == NULL)
|
|
|
1709 EMSG_M_RET_NULL(_("E70: Empty %s%%[]"),
|
|
|
1710 reg_magic == MAGIC_ALL);
|
|
|
1711 lastbranch = regnode(BRANCH);
|
|
|
1712 br = regnode(NOTHING);
|
|
|
1713 if (ret != JUST_CALC_SIZE)
|
|
|
1714 {
|
|
|
1715 regtail(lastnode, br);
|
|
|
1716 regtail(lastbranch, br);
|
|
|
1717 /* connect all branches to the NOTHING
|
|
|
1718 * branch at the end */
|
|
|
1719 for (br = ret; br != lastnode; )
|
|
|
1720 {
|
|
|
1721 if (OP(br) == BRANCH)
|
|
|
1722 {
|
|
|
1723 regtail(br, lastbranch);
|
|
|
1724 br = OPERAND(br);
|
|
|
1725 }
|
|
|
1726 else
|
|
|
1727 br = regnext(br);
|
|
|
1728 }
|
|
|
1729 }
|
|
|
1730 *flagp &= ~HASWIDTH;
|
|
|
1731 break;
|
|
|
1732 }
|
|
|
1733
|
|
24
|
1734 case 'd': /* %d123 decimal */
|
|
|
1735 case 'o': /* %o123 octal */
|
|
|
1736 case 'x': /* %xab hex 2 */
|
|
|
1737 case 'u': /* %uabcd hex 4 */
|
|
|
1738 case 'U': /* %U1234abcd hex 8 */
|
|
|
1739 {
|
|
|
1740 int i;
|
|
|
1741
|
|
|
1742 switch (c)
|
|
|
1743 {
|
|
|
1744 case 'd': i = getdecchrs(); break;
|
|
|
1745 case 'o': i = getoctchrs(); break;
|
|
|
1746 case 'x': i = gethexchrs(2); break;
|
|
|
1747 case 'u': i = gethexchrs(4); break;
|
|
|
1748 case 'U': i = gethexchrs(8); break;
|
|
|
1749 default: i = -1; break;
|
|
|
1750 }
|
|
|
1751
|
|
|
1752 if (i < 0)
|
|
|
1753 EMSG_M_RET_NULL(
|
|
|
1754 _("E678: Invalid character after %s%%[dxouU]"),
|
|
|
1755 reg_magic == MAGIC_ALL);
|
|
|
1756 ret = regnode(EXACTLY);
|
|
|
1757 if (i == 0)
|
|
|
1758 regc(0x0a);
|
|
|
1759 else
|
|
|
1760 #ifdef FEAT_MBYTE
|
|
|
1761 regmbc(i);
|
|
|
1762 #else
|
|
|
1763 regc(i);
|
|
|
1764 #endif
|
|
|
1765 regc(NUL);
|
|
|
1766 *flagp |= HASWIDTH;
|
|
|
1767 break;
|
|
|
1768 }
|
|
|
1769
|
|
7
|
1770 default:
|
|
|
1771 if (VIM_ISDIGIT(c) || c == '<' || c == '>')
|
|
|
1772 {
|
|
|
1773 long_u n = 0;
|
|
|
1774 int cmp;
|
|
|
1775
|
|
|
1776 cmp = c;
|
|
|
1777 if (cmp == '<' || cmp == '>')
|
|
|
1778 c = getchr();
|
|
|
1779 while (VIM_ISDIGIT(c))
|
|
|
1780 {
|
|
|
1781 n = n * 10 + (c - '0');
|
|
|
1782 c = getchr();
|
|
|
1783 }
|
|
|
1784 if (c == 'l' || c == 'c' || c == 'v')
|
|
|
1785 {
|
|
|
1786 if (c == 'l')
|
|
|
1787 ret = regnode(RE_LNUM);
|
|
|
1788 else if (c == 'c')
|
|
|
1789 ret = regnode(RE_COL);
|
|
|
1790 else
|
|
|
1791 ret = regnode(RE_VCOL);
|
|
|
1792 if (ret == JUST_CALC_SIZE)
|
|
|
1793 regsize += 5;
|
|
|
1794 else
|
|
|
1795 {
|
|
|
1796 /* put the number and the optional
|
|
|
1797 * comparator after the opcode */
|
|
|
1798 regcode = re_put_long(regcode, n);
|
|
|
1799 *regcode++ = cmp;
|
|
|
1800 }
|
|
|
1801 break;
|
|
|
1802 }
|
|
|
1803 }
|
|
|
1804
|
|
|
1805 EMSG_M_RET_NULL(_("E71: Invalid character after %s%%"),
|
|
|
1806 reg_magic == MAGIC_ALL);
|
|
|
1807 }
|
|
|
1808 }
|
|
|
1809 break;
|
|
|
1810
|
|
|
1811 case Magic('['):
|
|
|
1812 collection:
|
|
|
1813 {
|
|
|
1814 char_u *lp;
|
|
|
1815
|
|
|
1816 /*
|
|
|
1817 * If there is no matching ']', we assume the '[' is a normal
|
|
|
1818 * character. This makes 'incsearch' and ":help [" work.
|
|
|
1819 */
|
|
|
1820 lp = skip_anyof(regparse);
|
|
|
1821 if (*lp == ']') /* there is a matching ']' */
|
|
|
1822 {
|
|
|
1823 int startc = -1; /* > 0 when next '-' is a range */
|
|
|
1824 int endc;
|
|
|
1825
|
|
|
1826 /*
|
|
|
1827 * In a character class, different parsing rules apply.
|
|
|
1828 * Not even \ is special anymore, nothing is.
|
|
|
1829 */
|
|
|
1830 if (*regparse == '^') /* Complement of range. */
|
|
|
1831 {
|
|
|
1832 ret = regnode(ANYBUT + extra);
|
|
|
1833 regparse++;
|
|
|
1834 }
|
|
|
1835 else
|
|
|
1836 ret = regnode(ANYOF + extra);
|
|
|
1837
|
|
|
1838 /* At the start ']' and '-' mean the literal character. */
|
|
|
1839 if (*regparse == ']' || *regparse == '-')
|
|
|
1840 regc(*regparse++);
|
|
|
1841
|
|
|
1842 while (*regparse != NUL && *regparse != ']')
|
|
|
1843 {
|
|
|
1844 if (*regparse == '-')
|
|
|
1845 {
|
|
|
1846 ++regparse;
|
|
|
1847 /* The '-' is not used for a range at the end and
|
|
|
1848 * after or before a '\n'. */
|
|
|
1849 if (*regparse == ']' || *regparse == NUL
|
|
|
1850 || startc == -1
|
|
|
1851 || (regparse[0] == '\\' && regparse[1] == 'n'))
|
|
|
1852 {
|
|
|
1853 regc('-');
|
|
|
1854 startc = '-'; /* [--x] is a range */
|
|
|
1855 }
|
|
|
1856 else
|
|
|
1857 {
|
|
|
1858 #ifdef FEAT_MBYTE
|
|
|
1859 if (has_mbyte)
|
|
|
1860 endc = mb_ptr2char_adv(®parse);
|
|
|
1861 else
|
|
|
1862 #endif
|
|
|
1863 endc = *regparse++;
|
|
24
|
1864
|
|
|
1865 /* Handle \o40, \x20 and \u20AC style sequences */
|
|
|
1866 if (endc == '\\' && !cpo_lit)
|
|
|
1867 endc = coll_get_char();
|
|
|
1868
|
|
7
|
1869 if (startc > endc)
|
|
|
1870 EMSG_RET_NULL(_(e_invrange));
|
|
|
1871 #ifdef FEAT_MBYTE
|
|
|
1872 if (has_mbyte && ((*mb_char2len)(startc) > 1
|
|
|
1873 || (*mb_char2len)(endc) > 1))
|
|
|
1874 {
|
|
|
1875 /* Limit to a range of 256 chars */
|
|
|
1876 if (endc > startc + 256)
|
|
|
1877 EMSG_RET_NULL(_(e_invrange));
|
|
|
1878 while (++startc <= endc)
|
|
|
1879 regmbc(startc);
|
|
|
1880 }
|
|
|
1881 else
|
|
|
1882 #endif
|
|
|
1883 {
|
|
|
1884 #ifdef EBCDIC
|
|
|
1885 int alpha_only = FALSE;
|
|
|
1886
|
|
|
1887 /* for alphabetical range skip the gaps
|
|
|
1888 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */
|
|
|
1889 if (isalpha(startc) && isalpha(endc))
|
|
|
1890 alpha_only = TRUE;
|
|
|
1891 #endif
|
|
|
1892 while (++startc <= endc)
|
|
|
1893 #ifdef EBCDIC
|
|
|
1894 if (!alpha_only || isalpha(startc))
|
|
|
1895 #endif
|
|
|
1896 regc(startc);
|
|
|
1897 }
|
|
|
1898 startc = -1;
|
|
|
1899 }
|
|
|
1900 }
|
|
|
1901 /*
|
|
|
1902 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim
|
|
|
1903 * accepts "\t", "\e", etc., but only when the 'l' flag in
|
|
|
1904 * 'cpoptions' is not included.
|
|
|
1905 */
|
|
|
1906 else if (*regparse == '\\'
|
|
|
1907 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL
|
|
|
1908 || (!cpo_lit
|
|
|
1909 && vim_strchr(REGEXP_ABBR,
|
|
|
1910 regparse[1]) != NULL)))
|
|
|
1911 {
|
|
|
1912 regparse++;
|
|
|
1913 if (*regparse == 'n')
|
|
|
1914 {
|
|
|
1915 /* '\n' in range: also match NL */
|
|
|
1916 if (ret != JUST_CALC_SIZE)
|
|
|
1917 {
|
|
|
1918 if (*ret == ANYBUT)
|
|
|
1919 *ret = ANYBUT + ADD_NL;
|
|
|
1920 else if (*ret == ANYOF)
|
|
|
1921 *ret = ANYOF + ADD_NL;
|
|
|
1922 /* else: must have had a \n already */
|
|
|
1923 }
|
|
|
1924 *flagp |= HASNL;
|
|
|
1925 regparse++;
|
|
|
1926 startc = -1;
|
|
|
1927 }
|
|
24
|
1928 else if (*regparse == 'd'
|
|
|
1929 || *regparse == 'o'
|
|
|
1930 || *regparse == 'x'
|
|
|
1931 || *regparse == 'u'
|
|
|
1932 || *regparse == 'U')
|
|
|
1933 {
|
|
|
1934 startc = coll_get_char();
|
|
|
1935 if (startc == 0)
|
|
|
1936 regc(0x0a);
|
|
|
1937 else
|
|
|
1938 #ifdef FEAT_MBYTE
|
|
|
1939 regmbc(startc);
|
|
|
1940 #else
|
|
|
1941 regc(startc);
|
|
|
1942 #endif
|
|
|
1943 }
|
|
7
|
1944 else
|
|
|
1945 {
|
|
|
1946 startc = backslash_trans(*regparse++);
|
|
|
1947 regc(startc);
|
|
|
1948 }
|
|
|
1949 }
|
|
|
1950 else if (*regparse == '[')
|
|
|
1951 {
|
|
|
1952 int c_class;
|
|
|
1953 int cu;
|
|
|
1954
|
|
|
1955 c_class = skip_class_name(®parse);
|
|
|
1956 startc = -1;
|
|
|
1957 /* Characters assumed to be 8 bits! */
|
|
|
1958 switch (c_class)
|
|
|
1959 {
|
|
|
1960 case CLASS_NONE:
|
|
|
1961 /* literal '[', allow [[-x] as a range */
|
|
|
1962 startc = *regparse++;
|
|
|
1963 regc(startc);
|
|
|
1964 break;
|
|
|
1965 case CLASS_ALNUM:
|
|
|
1966 for (cu = 1; cu <= 255; cu++)
|
|
|
1967 if (isalnum(cu))
|
|
|
1968 regc(cu);
|
|
|
1969 break;
|
|
|
1970 case CLASS_ALPHA:
|
|
|
1971 for (cu = 1; cu <= 255; cu++)
|
|
|
1972 if (isalpha(cu))
|
|
|
1973 regc(cu);
|
|
|
1974 break;
|
|
|
1975 case CLASS_BLANK:
|
|
|
1976 regc(' ');
|
|
|
1977 regc('\t');
|
|
|
1978 break;
|
|
|
1979 case CLASS_CNTRL:
|
|
|
1980 for (cu = 1; cu <= 255; cu++)
|
|
|
1981 if (iscntrl(cu))
|
|
|
1982 regc(cu);
|
|
|
1983 break;
|
|
|
1984 case CLASS_DIGIT:
|
|
|
1985 for (cu = 1; cu <= 255; cu++)
|
|
|
1986 if (VIM_ISDIGIT(cu))
|
|
|
1987 regc(cu);
|
|
|
1988 break;
|
|
|
1989 case CLASS_GRAPH:
|
|
|
1990 for (cu = 1; cu <= 255; cu++)
|
|
|
1991 if (isgraph(cu))
|
|
|
1992 regc(cu);
|
|
|
1993 break;
|
|
|
1994 case CLASS_LOWER:
|
|
|
1995 for (cu = 1; cu <= 255; cu++)
|
|
|
1996 if (islower(cu))
|
|
|
1997 regc(cu);
|
|
|
1998 break;
|
|
|
1999 case CLASS_PRINT:
|
|
|
2000 for (cu = 1; cu <= 255; cu++)
|
|
|
2001 if (vim_isprintc(cu))
|
|
|
2002 regc(cu);
|
|
|
2003 break;
|
|
|
2004 case CLASS_PUNCT:
|
|
|
2005 for (cu = 1; cu <= 255; cu++)
|
|
|
2006 if (ispunct(cu))
|
|
|
2007 regc(cu);
|
|
|
2008 break;
|
|
|
2009 case CLASS_SPACE:
|
|
|
2010 for (cu = 9; cu <= 13; cu++)
|
|
|
2011 regc(cu);
|
|
|
2012 regc(' ');
|
|
|
2013 break;
|
|
|
2014 case CLASS_UPPER:
|
|
|
2015 for (cu = 1; cu <= 255; cu++)
|
|
|
2016 if (isupper(cu))
|
|
|
2017 regc(cu);
|
|
|
2018 break;
|
|
|
2019 case CLASS_XDIGIT:
|
|
|
2020 for (cu = 1; cu <= 255; cu++)
|
|
|
2021 if (vim_isxdigit(cu))
|
|
|
2022 regc(cu);
|
|
|
2023 break;
|
|
|
2024 case CLASS_TAB:
|
|
|
2025 regc('\t');
|
|
|
2026 break;
|
|
|
2027 case CLASS_RETURN:
|
|
|
2028 regc('\r');
|
|
|
2029 break;
|
|
|
2030 case CLASS_BACKSPACE:
|
|
|
2031 regc('\b');
|
|
|
2032 break;
|
|
|
2033 case CLASS_ESCAPE:
|
|
|
2034 regc('\033');
|
|
|
2035 break;
|
|
|
2036 }
|
|
|
2037 }
|
|
|
2038 else
|
|
|
2039 {
|
|
|
2040 #ifdef FEAT_MBYTE
|
|
|
2041 if (has_mbyte)
|
|
|
2042 {
|
|
|
2043 int len;
|
|
|
2044
|
|
|
2045 /* produce a multibyte character, including any
|
|
|
2046 * following composing characters */
|
|
|
2047 startc = mb_ptr2char(regparse);
|
|
|
2048 len = (*mb_ptr2len_check)(regparse);
|
|
|
2049 if (enc_utf8 && utf_char2len(startc) != len)
|
|
|
2050 startc = -1; /* composing chars */
|
|
|
2051 while (--len >= 0)
|
|
|
2052 regc(*regparse++);
|
|
|
2053 }
|
|
|
2054 else
|
|
|
2055 #endif
|
|
|
2056 {
|
|
|
2057 startc = *regparse++;
|
|
|
2058 regc(startc);
|
|
|
2059 }
|
|
|
2060 }
|
|
|
2061 }
|
|
|
2062 regc(NUL);
|
|
|
2063 prevchr_len = 1; /* last char was the ']' */
|
|
|
2064 if (*regparse != ']')
|
|
|
2065 EMSG_RET_NULL(_(e_toomsbra)); /* Cannot happen? */
|
|
|
2066 skipchr(); /* let's be friends with the lexer again */
|
|
|
2067 *flagp |= HASWIDTH | SIMPLE;
|
|
|
2068 break;
|
|
|
2069 }
|
|
|
2070 }
|
|
|
2071 /* FALLTHROUGH */
|
|
|
2072
|
|
|
2073 default:
|
|
|
2074 {
|
|
|
2075 int len;
|
|
|
2076
|
|
|
2077 #ifdef FEAT_MBYTE
|
|
|
2078 /* A multi-byte character is handled as a separate atom if it's
|
|
|
2079 * before a multi. */
|
|
|
2080 if (has_mbyte && (*mb_char2len)(c) > 1
|
|
|
2081 && re_multi_type(peekchr()) != NOT_MULTI)
|
|
|
2082 {
|
|
|
2083 ret = regnode(MULTIBYTECODE);
|
|
|
2084 regmbc(c);
|
|
|
2085 *flagp |= HASWIDTH | SIMPLE;
|
|
|
2086 break;
|
|
|
2087 }
|
|
|
2088 #endif
|
|
|
2089
|
|
|
2090 ret = regnode(EXACTLY);
|
|
|
2091
|
|
|
2092 /*
|
|
|
2093 * Append characters as long as:
|
|
|
2094 * - there is no following multi, we then need the character in
|
|
|
2095 * front of it as a single character operand
|
|
|
2096 * - not running into a Magic character
|
|
|
2097 * - "one_exactly" is not set
|
|
|
2098 * But always emit at least one character. Might be a Multi,
|
|
|
2099 * e.g., a "[" without matching "]".
|
|
|
2100 */
|
|
|
2101 for (len = 0; c != NUL && (len == 0
|
|
|
2102 || (re_multi_type(peekchr()) == NOT_MULTI
|
|
|
2103 && !one_exactly
|
|
|
2104 && !is_Magic(c))); ++len)
|
|
|
2105 {
|
|
|
2106 c = no_Magic(c);
|
|
|
2107 #ifdef FEAT_MBYTE
|
|
|
2108 if (has_mbyte)
|
|
|
2109 {
|
|
|
2110 regmbc(c);
|
|
|
2111 if (enc_utf8)
|
|
|
2112 {
|
|
|
2113 int off;
|
|
|
2114 int l;
|
|
|
2115
|
|
|
2116 /* Need to get composing character too, directly
|
|
|
2117 * access regparse for that, because skipchr() skips
|
|
|
2118 * over composing chars. */
|
|
|
2119 ungetchr();
|
|
|
2120 if (*regparse == '\\' && regparse[1] != NUL)
|
|
|
2121 off = 1;
|
|
|
2122 else
|
|
|
2123 off = 0;
|
|
|
2124 for (;;)
|
|
|
2125 {
|
|
|
2126 l = utf_ptr2len_check(regparse + off);
|
|
|
2127 if (!UTF_COMPOSINGLIKE(regparse + off,
|
|
|
2128 regparse + off + l))
|
|
|
2129 break;
|
|
|
2130 off += l;
|
|
|
2131 regmbc(utf_ptr2char(regparse + off));
|
|
|
2132 }
|
|
|
2133 skipchr();
|
|
|
2134 }
|
|
|
2135 }
|
|
|
2136 else
|
|
|
2137 #endif
|
|
|
2138 regc(c);
|
|
|
2139 c = getchr();
|
|
|
2140 }
|
|
|
2141 ungetchr();
|
|
|
2142
|
|
|
2143 regc(NUL);
|
|
|
2144 *flagp |= HASWIDTH;
|
|
|
2145 if (len == 1)
|
|
|
2146 *flagp |= SIMPLE;
|
|
|
2147 }
|
|
|
2148 break;
|
|
|
2149 }
|
|
|
2150
|
|
|
2151 return ret;
|
|
|
2152 }
|
|
|
2153
|
|
|
2154 /*
|
|
|
2155 * emit a node
|
|
|
2156 * Return pointer to generated code.
|
|
|
2157 */
|
|
|
2158 static char_u *
|
|
|
2159 regnode(op)
|
|
|
2160 int op;
|
|
|
2161 {
|
|
|
2162 char_u *ret;
|
|
|
2163
|
|
|
2164 ret = regcode;
|
|
|
2165 if (ret == JUST_CALC_SIZE)
|
|
|
2166 regsize += 3;
|
|
|
2167 else
|
|
|
2168 {
|
|
|
2169 *regcode++ = op;
|
|
|
2170 *regcode++ = NUL; /* Null "next" pointer. */
|
|
|
2171 *regcode++ = NUL;
|
|
|
2172 }
|
|
|
2173 return ret;
|
|
|
2174 }
|
|
|
2175
|
|
|
2176 /*
|
|
|
2177 * Emit (if appropriate) a byte of code
|
|
|
2178 */
|
|
|
2179 static void
|
|
|
2180 regc(b)
|
|
|
2181 int b;
|
|
|
2182 {
|
|
|
2183 if (regcode == JUST_CALC_SIZE)
|
|
|
2184 regsize++;
|
|
|
2185 else
|
|
|
2186 *regcode++ = b;
|
|
|
2187 }
|
|
|
2188
|
|
|
2189 #ifdef FEAT_MBYTE
|
|
|
2190 /*
|
|
|
2191 * Emit (if appropriate) a multi-byte character of code
|
|
|
2192 */
|
|
|
2193 static void
|
|
|
2194 regmbc(c)
|
|
|
2195 int c;
|
|
|
2196 {
|
|
|
2197 if (regcode == JUST_CALC_SIZE)
|
|
|
2198 regsize += (*mb_char2len)(c);
|
|
|
2199 else
|
|
|
2200 regcode += (*mb_char2bytes)(c, regcode);
|
|
|
2201 }
|
|
|
2202 #endif
|
|
|
2203
|
|
|
2204 /*
|
|
|
2205 * reginsert - insert an operator in front of already-emitted operand
|
|
|
2206 *
|
|
|
2207 * Means relocating the operand.
|
|
|
2208 */
|
|
|
2209 static void
|
|
|
2210 reginsert(op, opnd)
|
|
|
2211 int op;
|
|
|
2212 char_u *opnd;
|
|
|
2213 {
|
|
|
2214 char_u *src;
|
|
|
2215 char_u *dst;
|
|
|
2216 char_u *place;
|
|
|
2217
|
|
|
2218 if (regcode == JUST_CALC_SIZE)
|
|
|
2219 {
|
|
|
2220 regsize += 3;
|
|
|
2221 return;
|
|
|
2222 }
|
|
|
2223 src = regcode;
|
|
|
2224 regcode += 3;
|
|
|
2225 dst = regcode;
|
|
|
2226 while (src > opnd)
|
|
|
2227 *--dst = *--src;
|
|
|
2228
|
|
|
2229 place = opnd; /* Op node, where operand used to be. */
|
|
|
2230 *place++ = op;
|
|
|
2231 *place++ = NUL;
|
|
|
2232 *place = NUL;
|
|
|
2233 }
|
|
|
2234
|
|
|
2235 /*
|
|
|
2236 * reginsert_limits - insert an operator in front of already-emitted operand.
|
|
|
2237 * The operator has the given limit values as operands. Also set next pointer.
|
|
|
2238 *
|
|
|
2239 * Means relocating the operand.
|
|
|
2240 */
|
|
|
2241 static void
|
|
|
2242 reginsert_limits(op, minval, maxval, opnd)
|
|
|
2243 int op;
|
|
|
2244 long minval;
|
|
|
2245 long maxval;
|
|
|
2246 char_u *opnd;
|
|
|
2247 {
|
|
|
2248 char_u *src;
|
|
|
2249 char_u *dst;
|
|
|
2250 char_u *place;
|
|
|
2251
|
|
|
2252 if (regcode == JUST_CALC_SIZE)
|
|
|
2253 {
|
|
|
2254 regsize += 11;
|
|
|
2255 return;
|
|
|
2256 }
|
|
|
2257 src = regcode;
|
|
|
2258 regcode += 11;
|
|
|
2259 dst = regcode;
|
|
|
2260 while (src > opnd)
|
|
|
2261 *--dst = *--src;
|
|
|
2262
|
|
|
2263 place = opnd; /* Op node, where operand used to be. */
|
|
|
2264 *place++ = op;
|
|
|
2265 *place++ = NUL;
|
|
|
2266 *place++ = NUL;
|
|
|
2267 place = re_put_long(place, (long_u)minval);
|
|
|
2268 place = re_put_long(place, (long_u)maxval);
|
|
|
2269 regtail(opnd, place);
|
|
|
2270 }
|
|
|
2271
|
|
|
2272 /*
|
|
|
2273 * Write a long as four bytes at "p" and return pointer to the next char.
|
|
|
2274 */
|
|
|
2275 static char_u *
|
|
|
2276 re_put_long(p, val)
|
|
|
2277 char_u *p;
|
|
|
2278 long_u val;
|
|
|
2279 {
|
|
|
2280 *p++ = (char_u) ((val >> 24) & 0377);
|
|
|
2281 *p++ = (char_u) ((val >> 16) & 0377);
|
|
|
2282 *p++ = (char_u) ((val >> 8) & 0377);
|
|
|
2283 *p++ = (char_u) (val & 0377);
|
|
|
2284 return p;
|
|
|
2285 }
|
|
|
2286
|
|
|
2287 /*
|
|
|
2288 * regtail - set the next-pointer at the end of a node chain
|
|
|
2289 */
|
|
|
2290 static void
|
|
|
2291 regtail(p, val)
|
|
|
2292 char_u *p;
|
|
|
2293 char_u *val;
|
|
|
2294 {
|
|
|
2295 char_u *scan;
|
|
|
2296 char_u *temp;
|
|
|
2297 int offset;
|
|
|
2298
|
|
|
2299 if (p == JUST_CALC_SIZE)
|
|
|
2300 return;
|
|
|
2301
|
|
|
2302 /* Find last node. */
|
|
|
2303 scan = p;
|
|
|
2304 for (;;)
|
|
|
2305 {
|
|
|
2306 temp = regnext(scan);
|
|
|
2307 if (temp == NULL)
|
|
|
2308 break;
|
|
|
2309 scan = temp;
|
|
|
2310 }
|
|
|
2311
|
|
|
2312 if (OP(scan) == BACK)
|
|
|
2313 offset = (int)(scan - val);
|
|
|
2314 else
|
|
|
2315 offset = (int)(val - scan);
|
|
|
2316 *(scan + 1) = (char_u) (((unsigned)offset >> 8) & 0377);
|
|
|
2317 *(scan + 2) = (char_u) (offset & 0377);
|
|
|
2318 }
|
|
|
2319
|
|
|
2320 /*
|
|
|
2321 * regoptail - regtail on item after a BRANCH; nop if none
|
|
|
2322 */
|
|
|
2323 static void
|
|
|
2324 regoptail(p, val)
|
|
|
2325 char_u *p;
|
|
|
2326 char_u *val;
|
|
|
2327 {
|
|
|
2328 /* When op is neither BRANCH nor BRACE_COMPLEX0-9, it is "operandless" */
|
|
|
2329 if (p == NULL || p == JUST_CALC_SIZE
|
|
|
2330 || (OP(p) != BRANCH
|
|
|
2331 && (OP(p) < BRACE_COMPLEX || OP(p) > BRACE_COMPLEX + 9)))
|
|
|
2332 return;
|
|
|
2333 regtail(OPERAND(p), val);
|
|
|
2334 }
|
|
|
2335
|
|
|
2336 /*
|
|
|
2337 * getchr() - get the next character from the pattern. We know about
|
|
|
2338 * magic and such, so therefore we need a lexical analyzer.
|
|
|
2339 */
|
|
|
2340
|
|
|
2341 /* static int curchr; */
|
|
|
2342 static int prevprevchr;
|
|
|
2343 static int prevchr;
|
|
|
2344 static int nextchr; /* used for ungetchr() */
|
|
|
2345 /*
|
|
|
2346 * Note: prevchr is sometimes -1 when we are not at the start,
|
|
|
2347 * eg in /[ ^I]^ the pattern was never found even if it existed, because ^ was
|
|
|
2348 * taken to be magic -- webb
|
|
|
2349 */
|
|
|
2350 static int at_start; /* True when on the first character */
|
|
|
2351 static int prev_at_start; /* True when on the second character */
|
|
|
2352
|
|
|
2353 static void
|
|
|
2354 initchr(str)
|
|
|
2355 char_u *str;
|
|
|
2356 {
|
|
|
2357 regparse = str;
|
|
|
2358 prevchr_len = 0;
|
|
|
2359 curchr = prevprevchr = prevchr = nextchr = -1;
|
|
|
2360 at_start = TRUE;
|
|
|
2361 prev_at_start = FALSE;
|
|
|
2362 }
|
|
|
2363
|
|
|
2364 static int
|
|
|
2365 peekchr()
|
|
|
2366 {
|
|
|
2367 if (curchr == -1)
|
|
|
2368 {
|
|
|
2369 switch (curchr = regparse[0])
|
|
|
2370 {
|
|
|
2371 case '.':
|
|
|
2372 case '[':
|
|
|
2373 case '~':
|
|
|
2374 /* magic when 'magic' is on */
|
|
|
2375 if (reg_magic >= MAGIC_ON)
|
|
|
2376 curchr = Magic(curchr);
|
|
|
2377 break;
|
|
|
2378 case '(':
|
|
|
2379 case ')':
|
|
|
2380 case '{':
|
|
|
2381 case '%':
|
|
|
2382 case '+':
|
|
|
2383 case '=':
|
|
|
2384 case '?':
|
|
|
2385 case '@':
|
|
|
2386 case '!':
|
|
|
2387 case '&':
|
|
|
2388 case '|':
|
|
|
2389 case '<':
|
|
|
2390 case '>':
|
|
|
2391 case '#': /* future ext. */
|
|
|
2392 case '"': /* future ext. */
|
|
|
2393 case '\'': /* future ext. */
|
|
|
2394 case ',': /* future ext. */
|
|
|
2395 case '-': /* future ext. */
|
|
|
2396 case ':': /* future ext. */
|
|
|
2397 case ';': /* future ext. */
|
|
|
2398 case '`': /* future ext. */
|
|
|
2399 case '/': /* Can't be used in / command */
|
|
|
2400 /* magic only after "\v" */
|
|
|
2401 if (reg_magic == MAGIC_ALL)
|
|
|
2402 curchr = Magic(curchr);
|
|
|
2403 break;
|
|
|
2404 case '*':
|
|
|
2405 /* * is not magic as the very first character, eg "?*ptr" and when
|
|
|
2406 * after '^', eg "/^*ptr" */
|
|
|
2407 if (reg_magic >= MAGIC_ON && !at_start
|
|
|
2408 && !(prev_at_start && prevchr == Magic('^')))
|
|
|
2409 curchr = Magic('*');
|
|
|
2410 break;
|
|
|
2411 case '^':
|
|
|
2412 /* '^' is only magic as the very first character and if it's after
|
|
|
2413 * "\(", "\|", "\&' or "\n" */
|
|
|
2414 if (reg_magic >= MAGIC_OFF
|
|
|
2415 && (at_start
|
|
|
2416 || reg_magic == MAGIC_ALL
|
|
|
2417 || prevchr == Magic('(')
|
|
|
2418 || prevchr == Magic('|')
|
|
|
2419 || prevchr == Magic('&')
|
|
|
2420 || prevchr == Magic('n')
|
|
|
2421 || (no_Magic(prevchr) == '('
|
|
|
2422 && prevprevchr == Magic('%'))))
|
|
|
2423 {
|
|
|
2424 curchr = Magic('^');
|
|
|
2425 at_start = TRUE;
|
|
|
2426 prev_at_start = FALSE;
|
|
|
2427 }
|
|
|
2428 break;
|
|
|
2429 case '$':
|
|
|
2430 /* '$' is only magic as the very last char and if it's in front of
|
|
|
2431 * either "\|", "\)", "\&", or "\n" */
|
|
|
2432 if (reg_magic >= MAGIC_OFF)
|
|
|
2433 {
|
|
|
2434 char_u *p = regparse + 1;
|
|
|
2435
|
|
|
2436 /* ignore \c \C \m and \M after '$' */
|
|
|
2437 while (p[0] == '\\' && (p[1] == 'c' || p[1] == 'C'
|
|
|
2438 || p[1] == 'm' || p[1] == 'M' || p[1] == 'Z'))
|
|
|
2439 p += 2;
|
|
|
2440 if (p[0] == NUL
|
|
|
2441 || (p[0] == '\\'
|
|
|
2442 && (p[1] == '|' || p[1] == '&' || p[1] == ')'
|
|
|
2443 || p[1] == 'n'))
|
|
|
2444 || reg_magic == MAGIC_ALL)
|
|
|
2445 curchr = Magic('$');
|
|
|
2446 }
|
|
|
2447 break;
|
|
|
2448 case '\\':
|
|
|
2449 {
|
|
|
2450 int c = regparse[1];
|
|
|
2451
|
|
|
2452 if (c == NUL)
|
|
|
2453 curchr = '\\'; /* trailing '\' */
|
|
|
2454 else if (
|
|
|
2455 #ifdef EBCDIC
|
|
|
2456 vim_strchr(META, c)
|
|
|
2457 #else
|
|
|
2458 c <= '~' && META_flags[c]
|
|
|
2459 #endif
|
|
|
2460 )
|
|
|
2461 {
|
|
|
2462 /*
|
|
|
2463 * META contains everything that may be magic sometimes,
|
|
|
2464 * except ^ and $ ("\^" and "\$" are only magic after
|
|
|
2465 * "\v"). We now fetch the next character and toggle its
|
|
|
2466 * magicness. Therefore, \ is so meta-magic that it is
|
|
|
2467 * not in META.
|
|
|
2468 */
|
|
|
2469 curchr = -1;
|
|
|
2470 prev_at_start = at_start;
|
|
|
2471 at_start = FALSE; /* be able to say "/\*ptr" */
|
|
|
2472 ++regparse;
|
|
|
2473 peekchr();
|
|
|
2474 --regparse;
|
|
|
2475 curchr = toggle_Magic(curchr);
|
|
|
2476 }
|
|
|
2477 else if (vim_strchr(REGEXP_ABBR, c))
|
|
|
2478 {
|
|
|
2479 /*
|
|
|
2480 * Handle abbreviations, like "\t" for TAB -- webb
|
|
|
2481 */
|
|
|
2482 curchr = backslash_trans(c);
|
|
|
2483 }
|
|
|
2484 else if (reg_magic == MAGIC_NONE && (c == '$' || c == '^'))
|
|
|
2485 curchr = toggle_Magic(c);
|
|
|
2486 else
|
|
|
2487 {
|
|
|
2488 /*
|
|
|
2489 * Next character can never be (made) magic?
|
|
|
2490 * Then backslashing it won't do anything.
|
|
|
2491 */
|
|
|
2492 #ifdef FEAT_MBYTE
|
|
|
2493 if (has_mbyte)
|
|
|
2494 curchr = (*mb_ptr2char)(regparse + 1);
|
|
|
2495 else
|
|
|
2496 #endif
|
|
|
2497 curchr = c;
|
|
|
2498 }
|
|
|
2499 break;
|
|
|
2500 }
|
|
|
2501
|
|
|
2502 #ifdef FEAT_MBYTE
|
|
|
2503 default:
|
|
|
2504 if (has_mbyte)
|
|
|
2505 curchr = (*mb_ptr2char)(regparse);
|
|
|
2506 #endif
|
|
|
2507 }
|
|
|
2508 }
|
|
|
2509
|
|
|
2510 return curchr;
|
|
|
2511 }
|
|
|
2512
|
|
|
2513 /*
|
|
|
2514 * Eat one lexed character. Do this in a way that we can undo it.
|
|
|
2515 */
|
|
|
2516 static void
|
|
|
2517 skipchr()
|
|
|
2518 {
|
|
|
2519 /* peekchr() eats a backslash, do the same here */
|
|
|
2520 if (*regparse == '\\')
|
|
|
2521 prevchr_len = 1;
|
|
|
2522 else
|
|
|
2523 prevchr_len = 0;
|
|
|
2524 if (regparse[prevchr_len] != NUL)
|
|
|
2525 {
|
|
|
2526 #ifdef FEAT_MBYTE
|
|
|
2527 if (has_mbyte)
|
|
|
2528 prevchr_len += (*mb_ptr2len_check)(regparse + prevchr_len);
|
|
|
2529 else
|
|
|
2530 #endif
|
|
|
2531 ++prevchr_len;
|
|
|
2532 }
|
|
|
2533 regparse += prevchr_len;
|
|
|
2534 prev_at_start = at_start;
|
|
|
2535 at_start = FALSE;
|
|
|
2536 prevprevchr = prevchr;
|
|
|
2537 prevchr = curchr;
|
|
|
2538 curchr = nextchr; /* use previously unget char, or -1 */
|
|
|
2539 nextchr = -1;
|
|
|
2540 }
|
|
|
2541
|
|
|
2542 /*
|
|
|
2543 * Skip a character while keeping the value of prev_at_start for at_start.
|
|
|
2544 * prevchr and prevprevchr are also kept.
|
|
|
2545 */
|
|
|
2546 static void
|
|
|
2547 skipchr_keepstart()
|
|
|
2548 {
|
|
|
2549 int as = prev_at_start;
|
|
|
2550 int pr = prevchr;
|
|
|
2551 int prpr = prevprevchr;
|
|
|
2552
|
|
|
2553 skipchr();
|
|
|
2554 at_start = as;
|
|
|
2555 prevchr = pr;
|
|
|
2556 prevprevchr = prpr;
|
|
|
2557 }
|
|
|
2558
|
|
|
2559 static int
|
|
|
2560 getchr()
|
|
|
2561 {
|
|
|
2562 int chr = peekchr();
|
|
|
2563
|
|
|
2564 skipchr();
|
|
|
2565 return chr;
|
|
|
2566 }
|
|
|
2567
|
|
|
2568 /*
|
|
|
2569 * put character back. Works only once!
|
|
|
2570 */
|
|
|
2571 static void
|
|
|
2572 ungetchr()
|
|
|
2573 {
|
|
|
2574 nextchr = curchr;
|
|
|
2575 curchr = prevchr;
|
|
|
2576 prevchr = prevprevchr;
|
|
|
2577 at_start = prev_at_start;
|
|
|
2578 prev_at_start = FALSE;
|
|
|
2579
|
|
|
2580 /* Backup regparse, so that it's at the same position as before the
|
|
|
2581 * getchr(). */
|
|
|
2582 regparse -= prevchr_len;
|
|
|
2583 }
|
|
|
2584
|
|
|
2585 /*
|
|
24
|
2586 * get and return the value of the hex string immediately after the current
|
|
|
2587 * position. Return -1 for invalid, or 0-255 for valid. Position is updated:
|
|
|
2588 * blahblah\%x20asdf
|
|
|
2589 * before-^ ^-after
|
|
|
2590 * The parameter controls the maximum number of input characters. This will be
|
|
|
2591 * 2 when reading a \%x20 sequence and 4 when reading a \%u20AC sequence.
|
|
|
2592 */
|
|
|
2593 static int
|
|
|
2594 gethexchrs(maxinputlen)
|
|
|
2595 int maxinputlen;
|
|
|
2596 {
|
|
|
2597 int nr = 0;
|
|
|
2598 int c;
|
|
|
2599 int i;
|
|
|
2600
|
|
|
2601 for (i = 0; i < maxinputlen; ++i)
|
|
|
2602 {
|
|
|
2603 c = regparse[0];
|
|
|
2604 if (!vim_isxdigit(c))
|
|
|
2605 break;
|
|
|
2606 nr <<= 4;
|
|
|
2607 nr |= hex2nr(c);
|
|
|
2608 ++regparse;
|
|
|
2609 }
|
|
|
2610
|
|
|
2611 if (i == 0)
|
|
|
2612 return -1;
|
|
|
2613 return nr;
|
|
|
2614 }
|
|
|
2615
|
|
|
2616 /*
|
|
|
2617 * get and return the value of the decimal string immediately after the
|
|
|
2618 * current position. Return -1 for invalid. Consumes all digits.
|
|
|
2619 */
|
|
|
2620 static int
|
|
|
2621 getdecchrs()
|
|
|
2622 {
|
|
|
2623 int nr = 0;
|
|
|
2624 int c;
|
|
|
2625 int i;
|
|
|
2626
|
|
|
2627 for (i = 0; ; ++i)
|
|
|
2628 {
|
|
|
2629 c = regparse[0];
|
|
|
2630 if (c < '0' || c > '9')
|
|
|
2631 break;
|
|
|
2632 nr *= 10;
|
|
|
2633 nr += c - '0';
|
|
|
2634 ++regparse;
|
|
|
2635 }
|
|
|
2636
|
|
|
2637 if (i == 0)
|
|
|
2638 return -1;
|
|
|
2639 return nr;
|
|
|
2640 }
|
|
|
2641
|
|
|
2642 /*
|
|
|
2643 * get and return the value of the octal string immediately after the current
|
|
|
2644 * position. Return -1 for invalid, or 0-255 for valid. Smart enough to handle
|
|
|
2645 * numbers > 377 correctly (for example, 400 is treated as 40) and doesn't
|
|
|
2646 * treat 8 or 9 as recognised characters. Position is updated:
|
|
|
2647 * blahblah\%o210asdf
|
|
|
2648 * before-^ ^-after
|
|
|
2649 */
|
|
|
2650 static int
|
|
|
2651 getoctchrs()
|
|
|
2652 {
|
|
|
2653 int nr = 0;
|
|
|
2654 int c;
|
|
|
2655 int i;
|
|
|
2656
|
|
|
2657 for (i = 0; i < 3 && nr < 040; ++i)
|
|
|
2658 {
|
|
|
2659 c = regparse[0];
|
|
|
2660 if (c < '0' || c > '7')
|
|
|
2661 break;
|
|
|
2662 nr <<= 3;
|
|
|
2663 nr |= hex2nr(c);
|
|
|
2664 ++regparse;
|
|
|
2665 }
|
|
|
2666
|
|
|
2667 if (i == 0)
|
|
|
2668 return -1;
|
|
|
2669 return nr;
|
|
|
2670 }
|
|
|
2671
|
|
|
2672 /*
|
|
|
2673 * Get a number after a backslash that is inside [].
|
|
|
2674 * When nothing is recognized return a backslash.
|
|
|
2675 */
|
|
|
2676 static int
|
|
|
2677 coll_get_char()
|
|
|
2678 {
|
|
|
2679 int nr = -1;
|
|
|
2680
|
|
|
2681 switch (*regparse++)
|
|
|
2682 {
|
|
|
2683 case 'd': nr = getdecchrs(); break;
|
|
|
2684 case 'o': nr = getoctchrs(); break;
|
|
|
2685 case 'x': nr = gethexchrs(2); break;
|
|
|
2686 case 'u': nr = gethexchrs(4); break;
|
|
|
2687 case 'U': nr = gethexchrs(8); break;
|
|
|
2688 }
|
|
|
2689 if (nr < 0)
|
|
|
2690 {
|
|
|
2691 /* If getting the number fails be backwards compatible: the character
|
|
|
2692 * is a backslash. */
|
|
|
2693 --regparse;
|
|
|
2694 nr = '\\';
|
|
|
2695 }
|
|
|
2696 return nr;
|
|
|
2697 }
|
|
|
2698
|
|
|
2699 /*
|
|
7
|
2700 * read_limits - Read two integers to be taken as a minimum and maximum.
|
|
|
2701 * If the first character is '-', then the range is reversed.
|
|
|
2702 * Should end with 'end'. If minval is missing, zero is default, if maxval is
|
|
|
2703 * missing, a very big number is the default.
|
|
|
2704 */
|
|
|
2705 static int
|
|
|
2706 read_limits(minval, maxval)
|
|
|
2707 long *minval;
|
|
|
2708 long *maxval;
|
|
|
2709 {
|
|
|
2710 int reverse = FALSE;
|
|
|
2711 char_u *first_char;
|
|
|
2712 long tmp;
|
|
|
2713
|
|
|
2714 if (*regparse == '-')
|
|
|
2715 {
|
|
|
2716 /* Starts with '-', so reverse the range later */
|
|
|
2717 regparse++;
|
|
|
2718 reverse = TRUE;
|
|
|
2719 }
|
|
|
2720 first_char = regparse;
|
|
|
2721 *minval = getdigits(®parse);
|
|
|
2722 if (*regparse == ',') /* There is a comma */
|
|
|
2723 {
|
|
|
2724 if (vim_isdigit(*++regparse))
|
|
|
2725 *maxval = getdigits(®parse);
|
|
|
2726 else
|
|
|
2727 *maxval = MAX_LIMIT;
|
|
|
2728 }
|
|
|
2729 else if (VIM_ISDIGIT(*first_char))
|
|
|
2730 *maxval = *minval; /* It was \{n} or \{-n} */
|
|
|
2731 else
|
|
|
2732 *maxval = MAX_LIMIT; /* It was \{} or \{-} */
|
|
|
2733 if (*regparse == '\\')
|
|
|
2734 regparse++; /* Allow either \{...} or \{...\} */
|
|
|
2735 if (*regparse != '}' || (*maxval == 0 && *minval == 0))
|
|
|
2736 {
|
|
|
2737 sprintf((char *)IObuff, _("E554: Syntax error in %s{...}"),
|
|
|
2738 reg_magic == MAGIC_ALL ? "" : "\\");
|
|
|
2739 EMSG_RET_FAIL(IObuff);
|
|
|
2740 }
|
|
|
2741
|
|
|
2742 /*
|
|
|
2743 * Reverse the range if there was a '-', or make sure it is in the right
|
|
|
2744 * order otherwise.
|
|
|
2745 */
|
|
|
2746 if ((!reverse && *minval > *maxval) || (reverse && *minval < *maxval))
|
|
|
2747 {
|
|
|
2748 tmp = *minval;
|
|
|
2749 *minval = *maxval;
|
|
|
2750 *maxval = tmp;
|
|
|
2751 }
|
|
|
2752 skipchr(); /* let's be friends with the lexer again */
|
|
|
2753 return OK;
|
|
|
2754 }
|
|
|
2755
|
|
|
2756 /*
|
|
|
2757 * vim_regexec and friends
|
|
|
2758 */
|
|
|
2759
|
|
|
2760 /*
|
|
|
2761 * Global work variables for vim_regexec().
|
|
|
2762 */
|
|
|
2763
|
|
|
2764 /* The current match-position is remembered with these variables: */
|
|
|
2765 static linenr_T reglnum; /* line number, relative to first line */
|
|
|
2766 static char_u *regline; /* start of current line */
|
|
|
2767 static char_u *reginput; /* current input, points into "regline" */
|
|
|
2768
|
|
|
2769 static int need_clear_subexpr; /* subexpressions still need to be
|
|
|
2770 * cleared */
|
|
|
2771 #ifdef FEAT_SYN_HL
|
|
|
2772 static int need_clear_zsubexpr = FALSE; /* extmatch subexpressions
|
|
|
2773 * still need to be cleared */
|
|
|
2774 #endif
|
|
|
2775
|
|
|
2776 static int out_of_stack; /* TRUE when ran out of stack space */
|
|
|
2777
|
|
|
2778 /*
|
|
|
2779 * Structure used to save the current input state, when it needs to be
|
|
|
2780 * restored after trying a match. Used by reg_save() and reg_restore().
|
|
|
2781 */
|
|
|
2782 typedef struct
|
|
|
2783 {
|
|
|
2784 union
|
|
|
2785 {
|
|
|
2786 char_u *ptr; /* reginput pointer, for single-line regexp */
|
|
|
2787 lpos_T pos; /* reginput pos, for multi-line regexp */
|
|
|
2788 } rs_u;
|
|
|
2789 } regsave_T;
|
|
|
2790
|
|
|
2791 /* struct to save start/end pointer/position in for \(\) */
|
|
|
2792 typedef struct
|
|
|
2793 {
|
|
|
2794 union
|
|
|
2795 {
|
|
|
2796 char_u *ptr;
|
|
|
2797 lpos_T pos;
|
|
|
2798 } se_u;
|
|
|
2799 } save_se_T;
|
|
|
2800
|
|
|
2801 static char_u *reg_getline __ARGS((linenr_T lnum));
|
|
|
2802 static long vim_regexec_both __ARGS((char_u *line, colnr_T col));
|
|
|
2803 static long regtry __ARGS((regprog_T *prog, colnr_T col));
|
|
|
2804 static void cleanup_subexpr __ARGS((void));
|
|
|
2805 #ifdef FEAT_SYN_HL
|
|
|
2806 static void cleanup_zsubexpr __ARGS((void));
|
|
|
2807 #endif
|
|
|
2808 static void reg_nextline __ARGS((void));
|
|
|
2809 static void reg_save __ARGS((regsave_T *save));
|
|
|
2810 static void reg_restore __ARGS((regsave_T *save));
|
|
|
2811 static int reg_save_equal __ARGS((regsave_T *save));
|
|
|
2812 static void save_se_multi __ARGS((save_se_T *savep, lpos_T *posp));
|
|
|
2813 static void save_se_one __ARGS((save_se_T *savep, char_u **pp));
|
|
|
2814
|
|
|
2815 /* Save the sub-expressions before attempting a match. */
|
|
|
2816 #define save_se(savep, posp, pp) \
|
|
|
2817 REG_MULTI ? save_se_multi((savep), (posp)) : save_se_one((savep), (pp))
|
|
|
2818
|
|
|
2819 /* After a failed match restore the sub-expressions. */
|
|
|
2820 #define restore_se(savep, posp, pp) { \
|
|
|
2821 if (REG_MULTI) \
|
|
|
2822 *(posp) = (savep)->se_u.pos; \
|
|
|
2823 else \
|
|
|
2824 *(pp) = (savep)->se_u.ptr; }
|
|
|
2825
|
|
|
2826 static int re_num_cmp __ARGS((long_u val, char_u *scan));
|
|
|
2827 static int regmatch __ARGS((char_u *prog));
|
|
|
2828 static int regrepeat __ARGS((char_u *p, long maxcount));
|
|
|
2829
|
|
|
2830 #ifdef DEBUG
|
|
|
2831 int regnarrate = 0;
|
|
|
2832 #endif
|
|
|
2833
|
|
|
2834 /*
|
|
|
2835 * Internal copy of 'ignorecase'. It is set at each call to vim_regexec().
|
|
|
2836 * Normally it gets the value of "rm_ic" or "rmm_ic", but when the pattern
|
|
|
2837 * contains '\c' or '\C' the value is overruled.
|
|
|
2838 */
|
|
|
2839 static int ireg_ic;
|
|
|
2840
|
|
|
2841 #ifdef FEAT_MBYTE
|
|
|
2842 /*
|
|
|
2843 * Similar to ireg_ic, but only for 'combining' characters. Set with \Z flag
|
|
|
2844 * in the regexp. Defaults to false, always.
|
|
|
2845 */
|
|
|
2846 static int ireg_icombine;
|
|
|
2847 #endif
|
|
|
2848
|
|
|
2849 /*
|
|
|
2850 * Sometimes need to save a copy of a line. Since alloc()/free() is very
|
|
|
2851 * slow, we keep one allocated piece of memory and only re-allocate it when
|
|
|
2852 * it's too small. It's freed in vim_regexec_both() when finished.
|
|
|
2853 */
|
|
|
2854 static char_u *reg_tofree;
|
|
|
2855 static unsigned reg_tofreelen;
|
|
|
2856
|
|
|
2857 /*
|
|
|
2858 * These variables are set when executing a regexp to speed up the execution.
|
|
|
2859 * Which ones are set depends on whethere a single-line or multi-line match is
|
|
|
2860 * done:
|
|
|
2861 * single-line multi-line
|
|
|
2862 * reg_match ®match_T NULL
|
|
|
2863 * reg_mmatch NULL ®mmatch_T
|
|
|
2864 * reg_startp reg_match->startp <invalid>
|
|
|
2865 * reg_endp reg_match->endp <invalid>
|
|
|
2866 * reg_startpos <invalid> reg_mmatch->startpos
|
|
|
2867 * reg_endpos <invalid> reg_mmatch->endpos
|
|
|
2868 * reg_win NULL window in which to search
|
|
|
2869 * reg_buf <invalid> buffer in which to search
|
|
|
2870 * reg_firstlnum <invalid> first line in which to search
|
|
|
2871 * reg_maxline 0 last line nr
|
|
|
2872 * reg_line_lbr FALSE or TRUE FALSE
|
|
|
2873 */
|
|
|
2874 static regmatch_T *reg_match;
|
|
|
2875 static regmmatch_T *reg_mmatch;
|
|
|
2876 static char_u **reg_startp = NULL;
|
|
|
2877 static char_u **reg_endp = NULL;
|
|
|
2878 static lpos_T *reg_startpos = NULL;
|
|
|
2879 static lpos_T *reg_endpos = NULL;
|
|
|
2880 static win_T *reg_win;
|
|
|
2881 static buf_T *reg_buf;
|
|
|
2882 static linenr_T reg_firstlnum;
|
|
|
2883 static linenr_T reg_maxline;
|
|
|
2884 static int reg_line_lbr; /* "\n" in string is line break */
|
|
|
2885
|
|
|
2886 /*
|
|
|
2887 * Get pointer to the line "lnum", which is relative to "reg_firstlnum".
|
|
|
2888 */
|
|
|
2889 static char_u *
|
|
|
2890 reg_getline(lnum)
|
|
|
2891 linenr_T lnum;
|
|
|
2892 {
|
|
|
2893 /* when looking behind for a match/no-match lnum is negative. But we
|
|
|
2894 * can't go before line 1 */
|
|
|
2895 if (reg_firstlnum + lnum < 1)
|
|
|
2896 return NULL;
|
|
|
2897 return ml_get_buf(reg_buf, reg_firstlnum + lnum, FALSE);
|
|
|
2898 }
|
|
|
2899
|
|
|
2900 static regsave_T behind_pos;
|
|
|
2901
|
|
|
2902 #ifdef FEAT_SYN_HL
|
|
|
2903 static char_u *reg_startzp[NSUBEXP]; /* Workspace to mark beginning */
|
|
|
2904 static char_u *reg_endzp[NSUBEXP]; /* and end of \z(...\) matches */
|
|
|
2905 static lpos_T reg_startzpos[NSUBEXP]; /* idem, beginning pos */
|
|
|
2906 static lpos_T reg_endzpos[NSUBEXP]; /* idem, end pos */
|
|
|
2907 #endif
|
|
|
2908
|
|
|
2909 /* TRUE if using multi-line regexp. */
|
|
|
2910 #define REG_MULTI (reg_match == NULL)
|
|
|
2911
|
|
|
2912 /*
|
|
|
2913 * Match a regexp against a string.
|
|
|
2914 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp().
|
|
|
2915 * Uses curbuf for line count and 'iskeyword'.
|
|
|
2916 *
|
|
|
2917 * Return TRUE if there is a match, FALSE if not.
|
|
|
2918 */
|
|
|
2919 int
|
|
|
2920 vim_regexec(rmp, line, col)
|
|
|
2921 regmatch_T *rmp;
|
|
|
2922 char_u *line; /* string to match against */
|
|
|
2923 colnr_T col; /* column to start looking for match */
|
|
|
2924 {
|
|
|
2925 reg_match = rmp;
|
|
|
2926 reg_mmatch = NULL;
|
|
|
2927 reg_maxline = 0;
|
|
|
2928 reg_line_lbr = FALSE;
|
|
|
2929 reg_win = NULL;
|
|
|
2930 ireg_ic = rmp->rm_ic;
|
|
|
2931 #ifdef FEAT_MBYTE
|
|
|
2932 ireg_icombine = FALSE;
|
|
|
2933 #endif
|
|
|
2934 return (vim_regexec_both(line, col) != 0);
|
|
|
2935 }
|
|
|
2936
|
|
|
2937 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO)
|
|
|
2938 /*
|
|
|
2939 * Like vim_regexec(), but consider a "\n" in "line" to be a line break.
|
|
|
2940 */
|
|
|
2941 int
|
|
|
2942 vim_regexec_nl(rmp, line, col)
|
|
|
2943 regmatch_T *rmp;
|
|
|
2944 char_u *line; /* string to match against */
|
|
|
2945 colnr_T col; /* column to start looking for match */
|
|
|
2946 {
|
|
|
2947 reg_match = rmp;
|
|
|
2948 reg_mmatch = NULL;
|
|
|
2949 reg_maxline = 0;
|
|
|
2950 reg_line_lbr = TRUE;
|
|
|
2951 reg_win = NULL;
|
|
|
2952 ireg_ic = rmp->rm_ic;
|
|
|
2953 #ifdef FEAT_MBYTE
|
|
|
2954 ireg_icombine = FALSE;
|
|
|
2955 #endif
|
|
|
2956 return (vim_regexec_both(line, col) != 0);
|
|
|
2957 }
|
|
|
2958 #endif
|
|
|
2959
|
|
|
2960 /*
|
|
|
2961 * Match a regexp against multiple lines.
|
|
|
2962 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp().
|
|
|
2963 * Uses curbuf for line count and 'iskeyword'.
|
|
|
2964 *
|
|
|
2965 * Return zero if there is no match. Return number of lines contained in the
|
|
|
2966 * match otherwise.
|
|
|
2967 */
|
|
|
2968 long
|
|
|
2969 vim_regexec_multi(rmp, win, buf, lnum, col)
|
|
|
2970 regmmatch_T *rmp;
|
|
|
2971 win_T *win; /* window in which to search or NULL */
|
|
|
2972 buf_T *buf; /* buffer in which to search */
|
|
|
2973 linenr_T lnum; /* nr of line to start looking for match */
|
|
|
2974 colnr_T col; /* column to start looking for match */
|
|
|
2975 {
|
|
|
2976 long r;
|
|
|
2977 buf_T *save_curbuf = curbuf;
|
|
|
2978
|
|
|
2979 reg_match = NULL;
|
|
|
2980 reg_mmatch = rmp;
|
|
|
2981 reg_buf = buf;
|
|
|
2982 reg_win = win;
|
|
|
2983 reg_firstlnum = lnum;
|
|
|
2984 reg_maxline = reg_buf->b_ml.ml_line_count - lnum;
|
|
|
2985 reg_line_lbr = FALSE;
|
|
|
2986 ireg_ic = rmp->rmm_ic;
|
|
|
2987 #ifdef FEAT_MBYTE
|
|
|
2988 ireg_icombine = FALSE;
|
|
|
2989 #endif
|
|
|
2990
|
|
|
2991 /* Need to switch to buffer "buf" to make vim_iswordc() work. */
|
|
|
2992 curbuf = buf;
|
|
|
2993 r = vim_regexec_both(NULL, col);
|
|
|
2994 curbuf = save_curbuf;
|
|
|
2995
|
|
|
2996 return r;
|
|
|
2997 }
|
|
|
2998
|
|
|
2999 /*
|
|
|
3000 * Match a regexp against a string ("line" points to the string) or multiple
|
|
|
3001 * lines ("line" is NULL, use reg_getline()).
|
|
|
3002 */
|
|
|
3003 #ifdef HAVE_SETJMP_H
|
|
|
3004 static long
|
|
|
3005 vim_regexec_both(line_arg, col_arg)
|
|
|
3006 char_u *line_arg;
|
|
|
3007 colnr_T col_arg; /* column to start looking for match */
|
|
|
3008 #else
|
|
|
3009 static long
|
|
|
3010 vim_regexec_both(line, col)
|
|
|
3011 char_u *line;
|
|
|
3012 colnr_T col; /* column to start looking for match */
|
|
|
3013 #endif
|
|
|
3014 {
|
|
|
3015 regprog_T *prog;
|
|
|
3016 char_u *s;
|
|
|
3017 long retval;
|
|
|
3018 #ifdef HAVE_SETJMP_H
|
|
|
3019 char_u *line;
|
|
|
3020 colnr_T col;
|
|
|
3021 #endif
|
|
|
3022
|
|
|
3023 reg_tofree = NULL;
|
|
|
3024
|
|
|
3025 #ifdef HAVE_TRY_EXCEPT
|
|
|
3026 __try
|
|
|
3027 {
|
|
|
3028 #endif
|
|
|
3029
|
|
|
3030 #ifdef HAVE_SETJMP_H
|
|
|
3031 /*
|
|
|
3032 * Matching with a regexp may cause a very deep recursive call of
|
|
|
3033 * regmatch(). Vim will crash when running out of stack space. Catch
|
|
|
3034 * this here if the system supports it.
|
|
|
3035 */
|
|
|
3036 mch_startjmp();
|
|
|
3037 if (SETJMP(lc_jump_env) != 0)
|
|
|
3038 {
|
|
|
3039 mch_didjmp();
|
|
|
3040 # ifdef SIGHASARG
|
|
|
3041 if (lc_signal != SIGINT)
|
|
|
3042 # endif
|
|
|
3043 EMSG(_("E361: Crash intercepted; regexp too complex?"));
|
|
|
3044 retval = 0L;
|
|
|
3045 goto theend;
|
|
|
3046 }
|
|
|
3047
|
|
|
3048 /* Trick to avoid "might be clobbered by `longjmp'" warning from gcc. */
|
|
|
3049 line = line_arg;
|
|
|
3050 col = col_arg;
|
|
|
3051 #endif
|
|
|
3052 retval = 0L;
|
|
|
3053
|
|
|
3054 if (REG_MULTI)
|
|
|
3055 {
|
|
|
3056 prog = reg_mmatch->regprog;
|
|
|
3057 line = reg_getline((linenr_T)0);
|
|
|
3058 reg_startpos = reg_mmatch->startpos;
|
|
|
3059 reg_endpos = reg_mmatch->endpos;
|
|
|
3060 }
|
|
|
3061 else
|
|
|
3062 {
|
|
|
3063 prog = reg_match->regprog;
|
|
|
3064 reg_startp = reg_match->startp;
|
|
|
3065 reg_endp = reg_match->endp;
|
|
|
3066 }
|
|
|
3067
|
|
|
3068 /* Be paranoid... */
|
|
|
3069 if (prog == NULL || line == NULL)
|
|
|
3070 {
|
|
|
3071 EMSG(_(e_null));
|
|
|
3072 goto theend;
|
|
|
3073 }
|
|
|
3074
|
|
|
3075 /* Check validity of program. */
|
|
|
3076 if (prog_magic_wrong())
|
|
|
3077 goto theend;
|
|
|
3078
|
|
|
3079 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */
|
|
|
3080 if (prog->regflags & RF_ICASE)
|
|
|
3081 ireg_ic = TRUE;
|
|
|
3082 else if (prog->regflags & RF_NOICASE)
|
|
|
3083 ireg_ic = FALSE;
|
|
|
3084
|
|
|
3085 #ifdef FEAT_MBYTE
|
|
|
3086 /* If pattern contains "\Z" overrule value of ireg_icombine */
|
|
|
3087 if (prog->regflags & RF_ICOMBINE)
|
|
|
3088 ireg_icombine = TRUE;
|
|
|
3089 #endif
|
|
|
3090
|
|
|
3091 /* If there is a "must appear" string, look for it. */
|
|
|
3092 if (prog->regmust != NULL)
|
|
|
3093 {
|
|
|
3094 int c;
|
|
|
3095
|
|
|
3096 #ifdef FEAT_MBYTE
|
|
|
3097 if (has_mbyte)
|
|
|
3098 c = (*mb_ptr2char)(prog->regmust);
|
|
|
3099 else
|
|
|
3100 #endif
|
|
|
3101 c = *prog->regmust;
|
|
|
3102 s = line + col;
|
|
|
3103 while ((s = cstrchr(s, c)) != NULL)
|
|
|
3104 {
|
|
|
3105 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0)
|
|
|
3106 break; /* Found it. */
|
|
|
3107 #ifdef FEAT_MBYTE
|
|
|
3108 if (has_mbyte)
|
|
|
3109 s += (*mb_ptr2len_check)(s);
|
|
|
3110 else
|
|
|
3111 #endif
|
|
|
3112 ++s;
|
|
|
3113 }
|
|
|
3114 if (s == NULL) /* Not present. */
|
|
|
3115 goto theend;
|
|
|
3116 }
|
|
|
3117
|
|
|
3118 regline = line;
|
|
|
3119 reglnum = 0;
|
|
|
3120 out_of_stack = FALSE;
|
|
|
3121
|
|
|
3122 /* Simplest case: Anchored match need be tried only once. */
|
|
|
3123 if (prog->reganch)
|
|
|
3124 {
|
|
|
3125 int c;
|
|
|
3126
|
|
|
3127 #ifdef FEAT_MBYTE
|
|
|
3128 if (has_mbyte)
|
|
|
3129 c = (*mb_ptr2char)(regline + col);
|
|
|
3130 else
|
|
|
3131 #endif
|
|
|
3132 c = regline[col];
|
|
|
3133 if (prog->regstart == NUL
|
|
|
3134 || prog->regstart == c
|
|
|
3135 || (ireg_ic && ((
|
|
|
3136 #ifdef FEAT_MBYTE
|
|
|
3137 (enc_utf8 && utf_fold(prog->regstart) == utf_fold(c)))
|
|
|
3138 || (c < 255 && prog->regstart < 255 &&
|
|
|
3139 #endif
|
|
|
3140 TOLOWER_LOC(prog->regstart) == TOLOWER_LOC(c)))))
|
|
|
3141 retval = regtry(prog, col);
|
|
|
3142 else
|
|
|
3143 retval = 0;
|
|
|
3144 }
|
|
|
3145 else
|
|
|
3146 {
|
|
|
3147 /* Messy cases: unanchored match. */
|
|
|
3148 while (!got_int && !out_of_stack)
|
|
|
3149 {
|
|
|
3150 if (prog->regstart != NUL)
|
|
|
3151 {
|
|
|
3152 /* Skip until the char we know it must start with. */
|
|
|
3153 s = cstrchr(regline + col, prog->regstart);
|
|
|
3154 if (s == NULL)
|
|
|
3155 {
|
|
|
3156 retval = 0;
|
|
|
3157 break;
|
|
|
3158 }
|
|
|
3159 col = (int)(s - regline);
|
|
|
3160 }
|
|
|
3161
|
|
|
3162 retval = regtry(prog, col);
|
|
|
3163 if (retval > 0)
|
|
|
3164 break;
|
|
|
3165
|
|
|
3166 /* if not currently on the first line, get it again */
|
|
|
3167 if (reglnum != 0)
|
|
|
3168 {
|
|
|
3169 regline = reg_getline((linenr_T)0);
|
|
|
3170 reglnum = 0;
|
|
|
3171 }
|
|
|
3172 if (regline[col] == NUL)
|
|
|
3173 break;
|
|
|
3174 #ifdef FEAT_MBYTE
|
|
|
3175 if (has_mbyte)
|
|
|
3176 col += (*mb_ptr2len_check)(regline + col);
|
|
|
3177 else
|
|
|
3178 #endif
|
|
|
3179 ++col;
|
|
|
3180 }
|
|
|
3181 }
|
|
|
3182
|
|
|
3183 if (out_of_stack)
|
|
|
3184 EMSG(_("E363: pattern caused out-of-stack error"));
|
|
|
3185
|
|
|
3186 #ifdef HAVE_TRY_EXCEPT
|
|
|
3187 }
|
|
|
3188 __except(EXCEPTION_EXECUTE_HANDLER)
|
|
|
3189 {
|
|
|
3190 if (GetExceptionCode() == EXCEPTION_STACK_OVERFLOW)
|
|
|
3191 {
|
|
|
3192 RESETSTKOFLW();
|
|
|
3193 EMSG(_("E363: pattern caused out-of-stack error"));
|
|
|
3194 }
|
|
|
3195 else
|
|
|
3196 EMSG(_("E361: Crash intercepted; regexp too complex?"));
|
|
|
3197 retval = 0L;
|
|
|
3198 }
|
|
|
3199 #endif
|
|
|
3200
|
|
|
3201 theend:
|
|
|
3202 /* Didn't find a match. */
|
|
|
3203 vim_free(reg_tofree);
|
|
|
3204 #ifdef HAVE_SETJMP_H
|
|
|
3205 mch_endjmp();
|
|
|
3206 #endif
|
|
|
3207 return retval;
|
|
|
3208 }
|
|
|
3209
|
|
|
3210 #ifdef FEAT_SYN_HL
|
|
|
3211 static reg_extmatch_T *make_extmatch __ARGS((void));
|
|
|
3212
|
|
|
3213 /*
|
|
|
3214 * Create a new extmatch and mark it as referenced once.
|
|
|
3215 */
|
|
|
3216 static reg_extmatch_T *
|
|
|
3217 make_extmatch()
|
|
|
3218 {
|
|
|
3219 reg_extmatch_T *em;
|
|
|
3220
|
|
|
3221 em = (reg_extmatch_T *)alloc_clear((unsigned)sizeof(reg_extmatch_T));
|
|
|
3222 if (em != NULL)
|
|
|
3223 em->refcnt = 1;
|
|
|
3224 return em;
|
|
|
3225 }
|
|
|
3226
|
|
|
3227 /*
|
|
|
3228 * Add a reference to an extmatch.
|
|
|
3229 */
|
|
|
3230 reg_extmatch_T *
|
|
|
3231 ref_extmatch(em)
|
|
|
3232 reg_extmatch_T *em;
|
|
|
3233 {
|
|
|
3234 if (em != NULL)
|
|
|
3235 em->refcnt++;
|
|
|
3236 return em;
|
|
|
3237 }
|
|
|
3238
|
|
|
3239 /*
|
|
|
3240 * Remove a reference to an extmatch. If there are no references left, free
|
|
|
3241 * the info.
|
|
|
3242 */
|
|
|
3243 void
|
|
|
3244 unref_extmatch(em)
|
|
|
3245 reg_extmatch_T *em;
|
|
|
3246 {
|
|
|
3247 int i;
|
|
|
3248
|
|
|
3249 if (em != NULL && --em->refcnt <= 0)
|
|
|
3250 {
|
|
|
3251 for (i = 0; i < NSUBEXP; ++i)
|
|
|
3252 vim_free(em->matches[i]);
|
|
|
3253 vim_free(em);
|
|
|
3254 }
|
|
|
3255 }
|
|
|
3256 #endif
|
|
|
3257
|
|
|
3258 /*
|
|
|
3259 * regtry - try match of "prog" with at regline["col"].
|
|
|
3260 * Returns 0 for failure, number of lines contained in the match otherwise.
|
|
|
3261 */
|
|
|
3262 static long
|
|
|
3263 regtry(prog, col)
|
|
|
3264 regprog_T *prog;
|
|
|
3265 colnr_T col;
|
|
|
3266 {
|
|
|
3267 reginput = regline + col;
|
|
|
3268 need_clear_subexpr = TRUE;
|
|
|
3269 #ifdef FEAT_SYN_HL
|
|
|
3270 /* Clear the external match subpointers if necessary. */
|
|
|
3271 if (prog->reghasz == REX_SET)
|
|
|
3272 need_clear_zsubexpr = TRUE;
|
|
|
3273 #endif
|
|
|
3274
|
|
|
3275 if (regmatch(prog->program + 1))
|
|
|
3276 {
|
|
|
3277 cleanup_subexpr();
|
|
|
3278 if (REG_MULTI)
|
|
|
3279 {
|
|
|
3280 if (reg_startpos[0].lnum < 0)
|
|
|
3281 {
|
|
|
3282 reg_startpos[0].lnum = 0;
|
|
|
3283 reg_startpos[0].col = col;
|
|
|
3284 }
|
|
|
3285 if (reg_endpos[0].lnum < 0)
|
|
|
3286 {
|
|
|
3287 reg_endpos[0].lnum = reglnum;
|
|
|
3288 reg_endpos[0].col = (int)(reginput - regline);
|
|
|
3289 }
|
|
|
3290 else
|
|
|
3291 /* Use line number of "\ze". */
|
|
|
3292 reglnum = reg_endpos[0].lnum;
|
|
|
3293 }
|
|
|
3294 else
|
|
|
3295 {
|
|
|
3296 if (reg_startp[0] == NULL)
|
|
|
3297 reg_startp[0] = regline + col;
|
|
|
3298 if (reg_endp[0] == NULL)
|
|
|
3299 reg_endp[0] = reginput;
|
|
|
3300 }
|
|
|
3301 #ifdef FEAT_SYN_HL
|
|
|
3302 /* Package any found \z(...\) matches for export. Default is none. */
|
|
|
3303 unref_extmatch(re_extmatch_out);
|
|
|
3304 re_extmatch_out = NULL;
|
|
|
3305
|
|
|
3306 if (prog->reghasz == REX_SET)
|
|
|
3307 {
|
|
|
3308 int i;
|
|
|
3309
|
|
|
3310 cleanup_zsubexpr();
|
|
|
3311 re_extmatch_out = make_extmatch();
|
|
|
3312 for (i = 0; i < NSUBEXP; i++)
|
|
|
3313 {
|
|
|
3314 if (REG_MULTI)
|
|
|
3315 {
|
|
|
3316 /* Only accept single line matches. */
|
|
|
3317 if (reg_startzpos[i].lnum >= 0
|
|
|
3318 && reg_endzpos[i].lnum == reg_startzpos[i].lnum)
|
|
|
3319 re_extmatch_out->matches[i] =
|
|
|
3320 vim_strnsave(reg_getline(reg_startzpos[i].lnum)
|
|
|
3321 + reg_startzpos[i].col,
|
|
|
3322 reg_endzpos[i].col - reg_startzpos[i].col);
|
|
|
3323 }
|
|
|
3324 else
|
|
|
3325 {
|
|
|
3326 if (reg_startzp[i] != NULL && reg_endzp[i] != NULL)
|
|
|
3327 re_extmatch_out->matches[i] =
|
|
|
3328 vim_strnsave(reg_startzp[i],
|
|
|
3329 (int)(reg_endzp[i] - reg_startzp[i]));
|
|
|
3330 }
|
|
|
3331 }
|
|
|
3332 }
|
|
|
3333 #endif
|
|
|
3334 return 1 + reglnum;
|
|
|
3335 }
|
|
|
3336 return 0;
|
|
|
3337 }
|
|
|
3338
|
|
|
3339 #ifdef FEAT_MBYTE
|
|
|
3340 /* multi-byte: advance reginput with a function */
|
|
|
3341 # define ADVANCE_REGINPUT() advance_reginput()
|
|
|
3342
|
|
|
3343 static void advance_reginput __ARGS((void));
|
|
|
3344 static int reg_prev_class __ARGS((void));
|
|
|
3345
|
|
|
3346 static void
|
|
|
3347 advance_reginput()
|
|
|
3348 {
|
|
|
3349 if (has_mbyte)
|
|
|
3350 reginput += (*mb_ptr2len_check)(reginput);
|
|
|
3351 else
|
|
|
3352 ++reginput;
|
|
|
3353 }
|
|
|
3354
|
|
|
3355 /*
|
|
|
3356 * Get class of previous character.
|
|
|
3357 */
|
|
|
3358 static int
|
|
|
3359 reg_prev_class()
|
|
|
3360 {
|
|
|
3361 if (reginput > regline)
|
|
|
3362 return mb_get_class(reginput - 1
|
|
|
3363 - (*mb_head_off)(regline, reginput - 1));
|
|
|
3364 return -1;
|
|
|
3365 }
|
|
|
3366
|
|
|
3367 #else
|
|
|
3368 /* No multi-byte: It's too simple to make a function for. */
|
|
|
3369 # define ADVANCE_REGINPUT() ++reginput
|
|
|
3370 #endif
|
|
|
3371
|
|
|
3372 /*
|
|
|
3373 * The arguments from BRACE_LIMITS are stored here. They are actually local
|
|
|
3374 * to regmatch(), but they are here to reduce the amount of stack space used
|
|
|
3375 * (it can be called recursively many times).
|
|
|
3376 */
|
|
|
3377 static long bl_minval;
|
|
|
3378 static long bl_maxval;
|
|
|
3379
|
|
|
3380 /*
|
|
|
3381 * regmatch - main matching routine
|
|
|
3382 *
|
|
|
3383 * Conceptually the strategy is simple: Check to see whether the current
|
|
|
3384 * node matches, call self recursively to see whether the rest matches,
|
|
|
3385 * and then act accordingly. In practice we make some effort to avoid
|
|
|
3386 * recursion, in particular by going through "ordinary" nodes (that don't
|
|
|
3387 * need to know whether the rest of the match failed) by a loop instead of
|
|
|
3388 * by recursion.
|
|
|
3389 *
|
|
|
3390 * Returns TRUE when there is a match. Leaves reginput and reglnum just after
|
|
|
3391 * the last matched character.
|
|
|
3392 * Returns FALSE when there is no match. Leaves reginput and reglnum in an
|
|
|
3393 * undefined state!
|
|
|
3394 */
|
|
|
3395 static int
|
|
|
3396 regmatch(scan)
|
|
|
3397 char_u *scan; /* Current node. */
|
|
|
3398 {
|
|
|
3399 char_u *next; /* Next node. */
|
|
|
3400 int op;
|
|
|
3401 int c;
|
|
|
3402
|
|
|
3403 #ifdef HAVE_GETRLIMIT
|
|
|
3404 /* Check if we are running out of stack space. Could be caused by
|
|
|
3405 * recursively calling ourselves. */
|
|
|
3406 if (out_of_stack || mch_stackcheck((char *)&op) == FAIL)
|
|
|
3407 {
|
|
|
3408 out_of_stack = TRUE;
|
|
|
3409 return FALSE;
|
|
|
3410 }
|
|
|
3411 #endif
|
|
|
3412
|
|
|
3413 /* Some patterns my cause a long time to match, even though they are not
|
|
|
3414 * illegal. E.g., "\([a-z]\+\)\+Q". Allow breaking them with CTRL-C. */
|
|
|
3415 fast_breakcheck();
|
|
|
3416
|
|
|
3417 #ifdef DEBUG
|
|
|
3418 if (scan != NULL && regnarrate)
|
|
|
3419 {
|
|
|
3420 mch_errmsg(regprop(scan));
|
|
|
3421 mch_errmsg("(\n");
|
|
|
3422 }
|
|
|
3423 #endif
|
|
|
3424 while (scan != NULL)
|
|
|
3425 {
|
|
|
3426 if (got_int || out_of_stack)
|
|
|
3427 return FALSE;
|
|
|
3428 #ifdef DEBUG
|
|
|
3429 if (regnarrate)
|
|
|
3430 {
|
|
|
3431 mch_errmsg(regprop(scan));
|
|
|
3432 mch_errmsg("...\n");
|
|
|
3433 # ifdef FEAT_SYN_HL
|
|
|
3434 if (re_extmatch_in != NULL)
|
|
|
3435 {
|
|
|
3436 int i;
|
|
|
3437
|
|
|
3438 mch_errmsg(_("External submatches:\n"));
|
|
|
3439 for (i = 0; i < NSUBEXP; i++)
|
|
|
3440 {
|
|
|
3441 mch_errmsg(" \"");
|
|
|
3442 if (re_extmatch_in->matches[i] != NULL)
|
|
|
3443 mch_errmsg(re_extmatch_in->matches[i]);
|
|
|
3444 mch_errmsg("\"\n");
|
|
|
3445 }
|
|
|
3446 }
|
|
|
3447 # endif
|
|
|
3448 }
|
|
|
3449 #endif
|
|
|
3450 next = regnext(scan);
|
|
|
3451
|
|
|
3452 op = OP(scan);
|
|
|
3453 /* Check for character class with NL added. */
|
|
|
3454 if (WITH_NL(op) && *reginput == NUL && reglnum < reg_maxline)
|
|
|
3455 {
|
|
|
3456 reg_nextline();
|
|
|
3457 }
|
|
|
3458 else if (reg_line_lbr && WITH_NL(op) && *reginput == '\n')
|
|
|
3459 {
|
|
|
3460 ADVANCE_REGINPUT();
|
|
|
3461 }
|
|
|
3462 else
|
|
|
3463 {
|
|
|
3464 if (WITH_NL(op))
|
|
|
3465 op -= ADD_NL;
|
|
|
3466 #ifdef FEAT_MBYTE
|
|
|
3467 if (has_mbyte)
|
|
|
3468 c = (*mb_ptr2char)(reginput);
|
|
|
3469 else
|
|
|
3470 #endif
|
|
|
3471 c = *reginput;
|
|
|
3472 switch (op)
|
|
|
3473 {
|
|
|
3474 case BOL:
|
|
|
3475 if (reginput != regline)
|
|
|
3476 return FALSE;
|
|
|
3477 break;
|
|
|
3478
|
|
|
3479 case EOL:
|
|
|
3480 if (c != NUL)
|
|
|
3481 return FALSE;
|
|
|
3482 break;
|
|
|
3483
|
|
|
3484 case RE_BOF:
|
|
|
3485 /* Passing -1 to the getline() function provided for the search
|
|
|
3486 * should always return NULL if the current line is the first
|
|
|
3487 * line of the file. */
|
|
|
3488 if (reglnum != 0 || reginput != regline
|
|
|
3489 || (REG_MULTI && reg_getline((linenr_T)-1) != NULL))
|
|
|
3490 return FALSE;
|
|
|
3491 break;
|
|
|
3492
|
|
|
3493 case RE_EOF:
|
|
|
3494 if (reglnum != reg_maxline || c != NUL)
|
|
|
3495 return FALSE;
|
|
|
3496 break;
|
|
|
3497
|
|
|
3498 case CURSOR:
|
|
|
3499 /* Check if the buffer is in a window and compare the
|
|
|
3500 * reg_win->w_cursor position to the match position. */
|
|
|
3501 if (reg_win == NULL
|
|
|
3502 || (reglnum + reg_firstlnum != reg_win->w_cursor.lnum)
|
|
|
3503 || ((colnr_T)(reginput - regline) != reg_win->w_cursor.col))
|
|
|
3504 return FALSE;
|
|
|
3505 break;
|
|
|
3506
|
|
|
3507 case RE_LNUM:
|
|
|
3508 if (!REG_MULTI || !re_num_cmp((long_u)(reglnum + reg_firstlnum),
|
|
|
3509 scan))
|
|
|
3510 return FALSE;
|
|
|
3511 break;
|
|
|
3512
|
|
|
3513 case RE_COL:
|
|
|
3514 if (!re_num_cmp((long_u)(reginput - regline) + 1, scan))
|
|
|
3515 return FALSE;
|
|
|
3516 break;
|
|
|
3517
|
|
|
3518 case RE_VCOL:
|
|
|
3519 if (!re_num_cmp((long_u)win_linetabsize(
|
|
|
3520 reg_win == NULL ? curwin : reg_win,
|
|
|
3521 regline, (colnr_T)(reginput - regline)) + 1, scan))
|
|
|
3522 return FALSE;
|
|
|
3523 break;
|
|
|
3524
|
|
|
3525 case BOW: /* \<word; reginput points to w */
|
|
|
3526 if (c == NUL) /* Can't match at end of line */
|
|
|
3527 return FALSE;
|
|
|
3528 #ifdef FEAT_MBYTE
|
|
|
3529 if (has_mbyte)
|
|
|
3530 {
|
|
|
3531 int this_class;
|
|
|
3532
|
|
|
3533 /* Get class of current and previous char (if it exists). */
|
|
|
3534 this_class = mb_get_class(reginput);
|
|
|
3535 if (this_class <= 1)
|
|
|
3536 return FALSE; /* not on a word at all */
|
|
|
3537 if (reg_prev_class() == this_class)
|
|
|
3538 return FALSE; /* previous char is in same word */
|
|
|
3539 }
|
|
|
3540 #endif
|
|
|
3541 else
|
|
|
3542 {
|
|
|
3543 if (!vim_iswordc(c)
|
|
|
3544 || (reginput > regline && vim_iswordc(reginput[-1])))
|
|
|
3545 return FALSE;
|
|
|
3546 }
|
|
|
3547 break;
|
|
|
3548
|
|
|
3549 case EOW: /* word\>; reginput points after d */
|
|
|
3550 if (reginput == regline) /* Can't match at start of line */
|
|
|
3551 return FALSE;
|
|
|
3552 #ifdef FEAT_MBYTE
|
|
|
3553 if (has_mbyte)
|
|
|
3554 {
|
|
|
3555 int this_class, prev_class;
|
|
|
3556
|
|
|
3557 /* Get class of current and previous char (if it exists). */
|
|
|
3558 this_class = mb_get_class(reginput);
|
|
|
3559 prev_class = reg_prev_class();
|
|
|
3560 if (this_class == prev_class)
|
|
|
3561 return FALSE;
|
|
|
3562 if (prev_class == 0 || prev_class == 1)
|
|
|
3563 return FALSE;
|
|
|
3564 }
|
|
|
3565 else
|
|
|
3566 #endif
|
|
|
3567 {
|
|
|
3568 if (!vim_iswordc(reginput[-1]))
|
|
|
3569 return FALSE;
|
|
|
3570 if (reginput[0] != NUL && vim_iswordc(c))
|
|
|
3571 return FALSE;
|
|
|
3572 }
|
|
|
3573 break; /* Matched with EOW */
|
|
|
3574
|
|
|
3575 case ANY:
|
|
|
3576 if (c == NUL)
|
|
|
3577 return FALSE;
|
|
|
3578 ADVANCE_REGINPUT();
|
|
|
3579 break;
|
|
|
3580
|
|
|
3581 case IDENT:
|
|
|
3582 if (!vim_isIDc(c))
|
|
|
3583 return FALSE;
|
|
|
3584 ADVANCE_REGINPUT();
|
|
|
3585 break;
|
|
|
3586
|
|
|
3587 case SIDENT:
|
|
|
3588 if (VIM_ISDIGIT(*reginput) || !vim_isIDc(c))
|
|
|
3589 return FALSE;
|
|
|
3590 ADVANCE_REGINPUT();
|
|
|
3591 break;
|
|
|
3592
|
|
|
3593 case KWORD:
|
|
|
3594 if (!vim_iswordp(reginput))
|
|
|
3595 return FALSE;
|
|
|
3596 ADVANCE_REGINPUT();
|
|
|
3597 break;
|
|
|
3598
|
|
|
3599 case SKWORD:
|
|
|
3600 if (VIM_ISDIGIT(*reginput) || !vim_iswordp(reginput))
|
|
|
3601 return FALSE;
|
|
|
3602 ADVANCE_REGINPUT();
|
|
|
3603 break;
|
|
|
3604
|
|
|
3605 case FNAME:
|
|
|
3606 if (!vim_isfilec(c))
|
|
|
3607 return FALSE;
|
|
|
3608 ADVANCE_REGINPUT();
|
|
|
3609 break;
|
|
|
3610
|
|
|
3611 case SFNAME:
|
|
|
3612 if (VIM_ISDIGIT(*reginput) || !vim_isfilec(c))
|
|
|
3613 return FALSE;
|
|
|
3614 ADVANCE_REGINPUT();
|
|
|
3615 break;
|
|
|
3616
|
|
|
3617 case PRINT:
|
|
|
3618 if (ptr2cells(reginput) != 1)
|
|
|
3619 return FALSE;
|
|
|
3620 ADVANCE_REGINPUT();
|
|
|
3621 break;
|
|
|
3622
|
|
|
3623 case SPRINT:
|
|
|
3624 if (VIM_ISDIGIT(*reginput) || ptr2cells(reginput) != 1)
|
|
|
3625 return FALSE;
|
|
|
3626 ADVANCE_REGINPUT();
|
|
|
3627 break;
|
|
|
3628
|
|
|
3629 case WHITE:
|
|
|
3630 if (!vim_iswhite(c))
|
|
|
3631 return FALSE;
|
|
|
3632 ADVANCE_REGINPUT();
|
|
|
3633 break;
|
|
|
3634
|
|
|
3635 case NWHITE:
|
|
|
3636 if (c == NUL || vim_iswhite(c))
|
|
|
3637 return FALSE;
|
|
|
3638 ADVANCE_REGINPUT();
|
|
|
3639 break;
|
|
|
3640
|
|
|
3641 case DIGIT:
|
|
|
3642 if (!ri_digit(c))
|
|
|
3643 return FALSE;
|
|
|
3644 ADVANCE_REGINPUT();
|
|
|
3645 break;
|
|
|
3646
|
|
|
3647 case NDIGIT:
|
|
|
3648 if (c == NUL || ri_digit(c))
|
|
|
3649 return FALSE;
|
|
|
3650 ADVANCE_REGINPUT();
|
|
|
3651 break;
|
|
|
3652
|
|
|
3653 case HEX:
|
|
|
3654 if (!ri_hex(c))
|
|
|
3655 return FALSE;
|
|
|
3656 ADVANCE_REGINPUT();
|
|
|
3657 break;
|
|
|
3658
|
|
|
3659 case NHEX:
|
|
|
3660 if (c == NUL || ri_hex(c))
|
|
|
3661 return FALSE;
|
|
|
3662 ADVANCE_REGINPUT();
|
|
|
3663 break;
|
|
|
3664
|
|
|
3665 case OCTAL:
|
|
|
3666 if (!ri_octal(c))
|
|
|
3667 return FALSE;
|
|
|
3668 ADVANCE_REGINPUT();
|
|
|
3669 break;
|
|
|
3670
|
|
|
3671 case NOCTAL:
|
|
|
3672 if (c == NUL || ri_octal(c))
|
|
|
3673 return FALSE;
|
|
|
3674 ADVANCE_REGINPUT();
|
|
|
3675 break;
|
|
|
3676
|
|
|
3677 case WORD:
|
|
|
3678 if (!ri_word(c))
|
|
|
3679 return FALSE;
|
|
|
3680 ADVANCE_REGINPUT();
|
|
|
3681 break;
|
|
|
3682
|
|
|
3683 case NWORD:
|
|
|
3684 if (c == NUL || ri_word(c))
|
|
|
3685 return FALSE;
|
|
|
3686 ADVANCE_REGINPUT();
|
|
|
3687 break;
|
|
|
3688
|
|
|
3689 case HEAD:
|
|
|
3690 if (!ri_head(c))
|
|
|
3691 return FALSE;
|
|
|
3692 ADVANCE_REGINPUT();
|
|
|
3693 break;
|
|
|
3694
|
|
|
3695 case NHEAD:
|
|
|
3696 if (c == NUL || ri_head(c))
|
|
|
3697 return FALSE;
|
|
|
3698 ADVANCE_REGINPUT();
|
|
|
3699 break;
|
|
|
3700
|
|
|
3701 case ALPHA:
|
|
|
3702 if (!ri_alpha(c))
|
|
|
3703 return FALSE;
|
|
|
3704 ADVANCE_REGINPUT();
|
|
|
3705 break;
|
|
|
3706
|
|
|
3707 case NALPHA:
|
|
|
3708 if (c == NUL || ri_alpha(c))
|
|
|
3709 return FALSE;
|
|
|
3710 ADVANCE_REGINPUT();
|
|
|
3711 break;
|
|
|
3712
|
|
|
3713 case LOWER:
|
|
|
3714 if (!ri_lower(c))
|
|
|
3715 return FALSE;
|
|
|
3716 ADVANCE_REGINPUT();
|
|
|
3717 break;
|
|
|
3718
|
|
|
3719 case NLOWER:
|
|
|
3720 if (c == NUL || ri_lower(c))
|
|
|
3721 return FALSE;
|
|
|
3722 ADVANCE_REGINPUT();
|
|
|
3723 break;
|
|
|
3724
|
|
|
3725 case UPPER:
|
|
|
3726 if (!ri_upper(c))
|
|
|
3727 return FALSE;
|
|
|
3728 ADVANCE_REGINPUT();
|
|
|
3729 break;
|
|
|
3730
|
|
|
3731 case NUPPER:
|
|
|
3732 if (c == NUL || ri_upper(c))
|
|
|
3733 return FALSE;
|
|
|
3734 ADVANCE_REGINPUT();
|
|
|
3735 break;
|
|
|
3736
|
|
|
3737 case EXACTLY:
|
|
|
3738 {
|
|
|
3739 int len;
|
|
|
3740 char_u *opnd;
|
|
|
3741
|
|
|
3742 opnd = OPERAND(scan);
|
|
|
3743 /* Inline the first byte, for speed. */
|
|
|
3744 if (*opnd != *reginput
|
|
|
3745 && (!ireg_ic || (
|
|
|
3746 #ifdef FEAT_MBYTE
|
|
|
3747 !enc_utf8 &&
|
|
|
3748 #endif
|
|
|
3749 TOLOWER_LOC(*opnd) != TOLOWER_LOC(*reginput))))
|
|
|
3750 return FALSE;
|
|
|
3751 if (*opnd == NUL)
|
|
|
3752 {
|
|
|
3753 /* match empty string always works; happens when "~" is
|
|
|
3754 * empty. */
|
|
|
3755 }
|
|
|
3756 else if (opnd[1] == NUL
|
|
|
3757 #ifdef FEAT_MBYTE
|
|
|
3758 && !(enc_utf8 && ireg_ic)
|
|
|
3759 #endif
|
|
|
3760 )
|
|
|
3761 ++reginput; /* matched a single char */
|
|
|
3762 else
|
|
|
3763 {
|
|
|
3764 len = (int)STRLEN(opnd);
|
|
|
3765 /* Need to match first byte again for multi-byte. */
|
|
|
3766 if (cstrncmp(opnd, reginput, &len) != 0)
|
|
|
3767 return FALSE;
|
|
|
3768 #ifdef FEAT_MBYTE
|
|
|
3769 /* Check for following composing character. */
|
|
|
3770 if (enc_utf8 && UTF_COMPOSINGLIKE(reginput, reginput + len))
|
|
|
3771 {
|
|
|
3772 /* raaron: This code makes a composing character get
|
|
|
3773 * ignored, which is the correct behavior (sometimes)
|
|
|
3774 * for voweled Hebrew texts. */
|
|
|
3775 if (!ireg_icombine)
|
|
|
3776 return FALSE;
|
|
|
3777 }
|
|
|
3778 else
|
|
|
3779 #endif
|
|
|
3780 reginput += len;
|
|
|
3781 }
|
|
|
3782 }
|
|
|
3783 break;
|
|
|
3784
|
|
|
3785 case ANYOF:
|
|
|
3786 case ANYBUT:
|
|
|
3787 if (c == NUL)
|
|
|
3788 return FALSE;
|
|
|
3789 if ((cstrchr(OPERAND(scan), c) == NULL) == (op == ANYOF))
|
|
|
3790 return FALSE;
|
|
|
3791 ADVANCE_REGINPUT();
|
|
|
3792 break;
|
|
|
3793
|
|
|
3794 #ifdef FEAT_MBYTE
|
|
|
3795 case MULTIBYTECODE:
|
|
|
3796 if (has_mbyte)
|
|
|
3797 {
|
|
|
3798 int i, len;
|
|
|
3799 char_u *opnd;
|
|
|
3800
|
|
|
3801 opnd = OPERAND(scan);
|
|
|
3802 /* Safety check (just in case 'encoding' was changed since
|
|
|
3803 * compiling the program). */
|
|
|
3804 if ((len = (*mb_ptr2len_check)(opnd)) < 2)
|
|
|
3805 return FALSE;
|
|
|
3806 for (i = 0; i < len; ++i)
|
|
|
3807 if (opnd[i] != reginput[i])
|
|
|
3808 return FALSE;
|
|
|
3809 reginput += len;
|
|
|
3810 }
|
|
|
3811 else
|
|
|
3812 return FALSE;
|
|
|
3813 break;
|
|
|
3814 #endif
|
|
|
3815
|
|
|
3816 case NOTHING:
|
|
|
3817 break;
|
|
|
3818
|
|
|
3819 case BACK:
|
|
|
3820 break;
|
|
|
3821
|
|
|
3822 case MOPEN + 0: /* Match start: \zs */
|
|
|
3823 case MOPEN + 1: /* \( */
|
|
|
3824 case MOPEN + 2:
|
|
|
3825 case MOPEN + 3:
|
|
|
3826 case MOPEN + 4:
|
|
|
3827 case MOPEN + 5:
|
|
|
3828 case MOPEN + 6:
|
|
|
3829 case MOPEN + 7:
|
|
|
3830 case MOPEN + 8:
|
|
|
3831 case MOPEN + 9:
|
|
|
3832 {
|
|
|
3833 int no;
|
|
|
3834 save_se_T save;
|
|
|
3835
|
|
|
3836 no = op - MOPEN;
|
|
|
3837 cleanup_subexpr();
|
|
|
3838 save_se(&save, ®_startpos[no], ®_startp[no]);
|
|
|
3839
|
|
|
3840 if (regmatch(next))
|
|
|
3841 return TRUE;
|
|
|
3842
|
|
|
3843 restore_se(&save, ®_startpos[no], ®_startp[no]);
|
|
|
3844 return FALSE;
|
|
|
3845 }
|
|
|
3846 /* break; Not Reached */
|
|
|
3847
|
|
|
3848 case NOPEN: /* \%( */
|
|
|
3849 case NCLOSE: /* \) after \%( */
|
|
|
3850 if (regmatch(next))
|
|
|
3851 return TRUE;
|
|
|
3852 return FALSE;
|
|
|
3853 /* break; Not Reached */
|
|
|
3854
|
|
|
3855 #ifdef FEAT_SYN_HL
|
|
|
3856 case ZOPEN + 1:
|
|
|
3857 case ZOPEN + 2:
|
|
|
3858 case ZOPEN + 3:
|
|
|
3859 case ZOPEN + 4:
|
|
|
3860 case ZOPEN + 5:
|
|
|
3861 case ZOPEN + 6:
|
|
|
3862 case ZOPEN + 7:
|
|
|
3863 case ZOPEN + 8:
|
|
|
3864 case ZOPEN + 9:
|
|
|
3865 {
|
|
|
3866 int no;
|
|
|
3867 save_se_T save;
|
|
|
3868
|
|
|
3869 no = op - ZOPEN;
|
|
|
3870 cleanup_zsubexpr();
|
|
|
3871 save_se(&save, ®_startzpos[no], ®_startzp[no]);
|
|
|
3872
|
|
|
3873 if (regmatch(next))
|
|
|
3874 return TRUE;
|
|
|
3875
|
|
|
3876 restore_se(&save, ®_startzpos[no], ®_startzp[no]);
|
|
|
3877 return FALSE;
|
|
|
3878 }
|
|
|
3879 /* break; Not Reached */
|
|
|
3880 #endif
|
|
|
3881
|
|
|
3882 case MCLOSE + 0: /* Match end: \ze */
|
|
|
3883 case MCLOSE + 1: /* \) */
|
|
|
3884 case MCLOSE + 2:
|
|
|
3885 case MCLOSE + 3:
|
|
|
3886 case MCLOSE + 4:
|
|
|
3887 case MCLOSE + 5:
|
|
|
3888 case MCLOSE + 6:
|
|
|
3889 case MCLOSE + 7:
|
|
|
3890 case MCLOSE + 8:
|
|
|
3891 case MCLOSE + 9:
|
|
|
3892 {
|
|
|
3893 int no;
|
|
|
3894 save_se_T save;
|
|
|
3895
|
|
|
3896 no = op - MCLOSE;
|
|
|
3897 cleanup_subexpr();
|
|
|
3898 save_se(&save, ®_endpos[no], ®_endp[no]);
|
|
|
3899
|
|
|
3900 if (regmatch(next))
|
|
|
3901 return TRUE;
|
|
|
3902
|
|
|
3903 restore_se(&save, ®_endpos[no], ®_endp[no]);
|
|
|
3904 return FALSE;
|
|
|
3905 }
|
|
|
3906 /* break; Not Reached */
|
|
|
3907
|
|
|
3908 #ifdef FEAT_SYN_HL
|
|
|
3909 case ZCLOSE + 1: /* \) after \z( */
|
|
|
3910 case ZCLOSE + 2:
|
|
|
3911 case ZCLOSE + 3:
|
|
|
3912 case ZCLOSE + 4:
|
|
|
3913 case ZCLOSE + 5:
|
|
|
3914 case ZCLOSE + 6:
|
|
|
3915 case ZCLOSE + 7:
|
|
|
3916 case ZCLOSE + 8:
|
|
|
3917 case ZCLOSE + 9:
|
|
|
3918 {
|
|
|
3919 int no;
|
|
|
3920 save_se_T save;
|
|
|
3921
|
|
|
3922 no = op - ZCLOSE;
|
|
|
3923 cleanup_zsubexpr();
|
|
|
3924 save_se(&save, ®_endzpos[no], ®_endzp[no]);
|
|
|
3925
|
|
|
3926 if (regmatch(next))
|
|
|
3927 return TRUE;
|
|
|
3928
|
|
|
3929 restore_se(&save, ®_endzpos[no], ®_endzp[no]);
|
|
|
3930 return FALSE;
|
|
|
3931 }
|
|
|
3932 /* break; Not Reached */
|
|
|
3933 #endif
|
|
|
3934
|
|
|
3935 case BACKREF + 1:
|
|
|
3936 case BACKREF + 2:
|
|
|
3937 case BACKREF + 3:
|
|
|
3938 case BACKREF + 4:
|
|
|
3939 case BACKREF + 5:
|
|
|
3940 case BACKREF + 6:
|
|
|
3941 case BACKREF + 7:
|
|
|
3942 case BACKREF + 8:
|
|
|
3943 case BACKREF + 9:
|
|
|
3944 {
|
|
|
3945 int no;
|
|
|
3946 int len;
|
|
|
3947 linenr_T clnum;
|
|
|
3948 colnr_T ccol;
|
|
|
3949 char_u *p;
|
|
|
3950
|
|
|
3951 no = op - BACKREF;
|
|
|
3952 cleanup_subexpr();
|
|
|
3953 if (!REG_MULTI) /* Single-line regexp */
|
|
|
3954 {
|
|
|
3955 if (reg_endp[no] == NULL)
|
|
|
3956 {
|
|
|
3957 /* Backref was not set: Match an empty string. */
|
|
|
3958 len = 0;
|
|
|
3959 }
|
|
|
3960 else
|
|
|
3961 {
|
|
|
3962 /* Compare current input with back-ref in the same
|
|
|
3963 * line. */
|
|
|
3964 len = (int)(reg_endp[no] - reg_startp[no]);
|
|
|
3965 if (cstrncmp(reg_startp[no], reginput, &len) != 0)
|
|
|
3966 return FALSE;
|
|
|
3967 }
|
|
|
3968 }
|
|
|
3969 else /* Multi-line regexp */
|
|
|
3970 {
|
|
|
3971 if (reg_endpos[no].lnum < 0)
|
|
|
3972 {
|
|
|
3973 /* Backref was not set: Match an empty string. */
|
|
|
3974 len = 0;
|
|
|
3975 }
|
|
|
3976 else
|
|
|
3977 {
|
|
|
3978 if (reg_startpos[no].lnum == reglnum
|
|
|
3979 && reg_endpos[no].lnum == reglnum)
|
|
|
3980 {
|
|
|
3981 /* Compare back-ref within the current line. */
|
|
|
3982 len = reg_endpos[no].col - reg_startpos[no].col;
|
|
|
3983 if (cstrncmp(regline + reg_startpos[no].col,
|
|
|
3984 reginput, &len) != 0)
|
|
|
3985 return FALSE;
|
|
|
3986 }
|
|
|
3987 else
|
|
|
3988 {
|
|
|
3989 /* Messy situation: Need to compare between two
|
|
|
3990 * lines. */
|
|
|
3991 ccol = reg_startpos[no].col;
|
|
|
3992 clnum = reg_startpos[no].lnum;
|
|
|
3993 for (;;)
|
|
|
3994 {
|
|
|
3995 /* Since getting one line may invalidate
|
|
|
3996 * the other, need to make copy. Slow! */
|
|
|
3997 if (regline != reg_tofree)
|
|
|
3998 {
|
|
|
3999 len = (int)STRLEN(regline);
|
|
|
4000 if (reg_tofree == NULL
|
|
|
4001 || len >= (int)reg_tofreelen)
|
|
|
4002 {
|
|
|
4003 len += 50; /* get some extra */
|
|
|
4004 vim_free(reg_tofree);
|
|
|
4005 reg_tofree = alloc(len);
|
|
|
4006 if (reg_tofree == NULL)
|
|
|
4007 return FALSE; /* out of memory! */
|
|
|
4008 reg_tofreelen = len;
|
|
|
4009 }
|
|
|
4010 STRCPY(reg_tofree, regline);
|
|
|
4011 reginput = reg_tofree
|
|
|
4012 + (reginput - regline);
|
|
|
4013 regline = reg_tofree;
|
|
|
4014 }
|
|
|
4015
|
|
|
4016 /* Get the line to compare with. */
|
|
|
4017 p = reg_getline(clnum);
|
|
|
4018 if (clnum == reg_endpos[no].lnum)
|
|
|
4019 len = reg_endpos[no].col - ccol;
|
|
|
4020 else
|
|
|
4021 len = (int)STRLEN(p + ccol);
|
|
|
4022
|
|
|
4023 if (cstrncmp(p + ccol, reginput, &len) != 0)
|
|
|
4024 return FALSE; /* doesn't match */
|
|
|
4025 if (clnum == reg_endpos[no].lnum)
|
|
|
4026 break; /* match and at end! */
|
|
|
4027 if (reglnum == reg_maxline)
|
|
|
4028 return FALSE; /* text too short */
|
|
|
4029
|
|
|
4030 /* Advance to next line. */
|
|
|
4031 reg_nextline();
|
|
|
4032 ++clnum;
|
|
|
4033 ccol = 0;
|
|
|
4034 if (got_int || out_of_stack)
|
|
|
4035 return FALSE;
|
|
|
4036 }
|
|
|
4037
|
|
|
4038 /* found a match! Note that regline may now point
|
|
|
4039 * to a copy of the line, that should not matter. */
|
|
|
4040 }
|
|
|
4041 }
|
|
|
4042 }
|
|
|
4043
|
|
|
4044 /* Matched the backref, skip over it. */
|
|
|
4045 reginput += len;
|
|
|
4046 }
|
|
|
4047 break;
|
|
|
4048
|
|
|
4049 #ifdef FEAT_SYN_HL
|
|
|
4050 case ZREF + 1:
|
|
|
4051 case ZREF + 2:
|
|
|
4052 case ZREF + 3:
|
|
|
4053 case ZREF + 4:
|
|
|
4054 case ZREF + 5:
|
|
|
4055 case ZREF + 6:
|
|
|
4056 case ZREF + 7:
|
|
|
4057 case ZREF + 8:
|
|
|
4058 case ZREF + 9:
|
|
|
4059 {
|
|
|
4060 int no;
|
|
|
4061 int len;
|
|
|
4062
|
|
|
4063 cleanup_zsubexpr();
|
|
|
4064 no = op - ZREF;
|
|
|
4065 if (re_extmatch_in != NULL
|
|
|
4066 && re_extmatch_in->matches[no] != NULL)
|
|
|
4067 {
|
|
|
4068 len = (int)STRLEN(re_extmatch_in->matches[no]);
|
|
|
4069 if (cstrncmp(re_extmatch_in->matches[no],
|
|
|
4070 reginput, &len) != 0)
|
|
|
4071 return FALSE;
|
|
|
4072 reginput += len;
|
|
|
4073 }
|
|
|
4074 else
|
|
|
4075 {
|
|
|
4076 /* Backref was not set: Match an empty string. */
|
|
|
4077 }
|
|
|
4078 }
|
|
|
4079 break;
|
|
|
4080 #endif
|
|
|
4081
|
|
|
4082 case BRANCH:
|
|
|
4083 {
|
|
|
4084 if (OP(next) != BRANCH) /* No choice. */
|
|
|
4085 next = OPERAND(scan); /* Avoid recursion. */
|
|
|
4086 else
|
|
|
4087 {
|
|
|
4088 regsave_T save;
|
|
|
4089
|
|
|
4090 do
|
|
|
4091 {
|
|
|
4092 reg_save(&save);
|
|
|
4093 if (regmatch(OPERAND(scan)))
|
|
|
4094 return TRUE;
|
|
|
4095 reg_restore(&save);
|
|
|
4096 scan = regnext(scan);
|
|
|
4097 } while (scan != NULL && OP(scan) == BRANCH);
|
|
|
4098 return FALSE;
|
|
|
4099 /* NOTREACHED */
|
|
|
4100 }
|
|
|
4101 }
|
|
|
4102 break;
|
|
|
4103
|
|
|
4104 case BRACE_LIMITS:
|
|
|
4105 {
|
|
|
4106 int no;
|
|
|
4107
|
|
|
4108 if (OP(next) == BRACE_SIMPLE)
|
|
|
4109 {
|
|
|
4110 bl_minval = OPERAND_MIN(scan);
|
|
|
4111 bl_maxval = OPERAND_MAX(scan);
|
|
|
4112 }
|
|
|
4113 else if (OP(next) >= BRACE_COMPLEX
|
|
|
4114 && OP(next) < BRACE_COMPLEX + 10)
|
|
|
4115 {
|
|
|
4116 no = OP(next) - BRACE_COMPLEX;
|
|
|
4117 brace_min[no] = OPERAND_MIN(scan);
|
|
|
4118 brace_max[no] = OPERAND_MAX(scan);
|
|
|
4119 brace_count[no] = 0;
|
|
|
4120 }
|
|
|
4121 else
|
|
|
4122 {
|
|
|
4123 EMSG(_(e_internal)); /* Shouldn't happen */
|
|
|
4124 return FALSE;
|
|
|
4125 }
|
|
|
4126 }
|
|
|
4127 break;
|
|
|
4128
|
|
|
4129 case BRACE_COMPLEX + 0:
|
|
|
4130 case BRACE_COMPLEX + 1:
|
|
|
4131 case BRACE_COMPLEX + 2:
|
|
|
4132 case BRACE_COMPLEX + 3:
|
|
|
4133 case BRACE_COMPLEX + 4:
|
|
|
4134 case BRACE_COMPLEX + 5:
|
|
|
4135 case BRACE_COMPLEX + 6:
|
|
|
4136 case BRACE_COMPLEX + 7:
|
|
|
4137 case BRACE_COMPLEX + 8:
|
|
|
4138 case BRACE_COMPLEX + 9:
|
|
|
4139 {
|
|
|
4140 int no;
|
|
|
4141 regsave_T save;
|
|
|
4142
|
|
|
4143 no = op - BRACE_COMPLEX;
|
|
|
4144 ++brace_count[no];
|
|
|
4145
|
|
|
4146 /* If not matched enough times yet, try one more */
|
|
|
4147 if (brace_count[no] <= (brace_min[no] <= brace_max[no]
|
|
|
4148 ? brace_min[no] : brace_max[no]))
|
|
|
4149 {
|
|
|
4150 reg_save(&save);
|
|
|
4151 if (regmatch(OPERAND(scan)))
|
|
|
4152 return TRUE;
|
|
|
4153 reg_restore(&save);
|
|
|
4154 --brace_count[no]; /* failed, decrement match count */
|
|
|
4155 return FALSE;
|
|
|
4156 }
|
|
|
4157
|
|
|
4158 /* If matched enough times, may try matching some more */
|
|
|
4159 if (brace_min[no] <= brace_max[no])
|
|
|
4160 {
|
|
|
4161 /* Range is the normal way around, use longest match */
|
|
|
4162 if (brace_count[no] <= brace_max[no])
|
|
|
4163 {
|
|
|
4164 reg_save(&save);
|
|
|
4165 if (regmatch(OPERAND(scan)))
|
|
|
4166 return TRUE; /* matched some more times */
|
|
|
4167 reg_restore(&save);
|
|
|
4168 --brace_count[no]; /* matched just enough times */
|
|
|
4169 /* continue with the items after \{} */
|
|
|
4170 }
|
|
|
4171 }
|
|
|
4172 else
|
|
|
4173 {
|
|
|
4174 /* Range is backwards, use shortest match first */
|
|
|
4175 if (brace_count[no] <= brace_min[no])
|
|
|
4176 {
|
|
|
4177 reg_save(&save);
|
|
|
4178 if (regmatch(next))
|
|
|
4179 return TRUE;
|
|
|
4180 reg_restore(&save);
|
|
|
4181 next = OPERAND(scan);
|
|
|
4182 /* must try to match one more item */
|
|
|
4183 }
|
|
|
4184 }
|
|
|
4185 }
|
|
|
4186 break;
|
|
|
4187
|
|
|
4188 case BRACE_SIMPLE:
|
|
|
4189 case STAR:
|
|
|
4190 case PLUS:
|
|
|
4191 {
|
|
|
4192 int nextb; /* next byte */
|
|
|
4193 int nextb_ic; /* next byte reverse case */
|
|
|
4194 long count;
|
|
|
4195 regsave_T save;
|
|
|
4196 long minval;
|
|
|
4197 long maxval;
|
|
|
4198
|
|
|
4199 /*
|
|
|
4200 * Lookahead to avoid useless match attempts when we know
|
|
|
4201 * what character comes next.
|
|
|
4202 */
|
|
|
4203 if (OP(next) == EXACTLY)
|
|
|
4204 {
|
|
|
4205 nextb = *OPERAND(next);
|
|
|
4206 if (ireg_ic)
|
|
|
4207 {
|
|
|
4208 if (isupper(nextb))
|
|
|
4209 nextb_ic = TOLOWER_LOC(nextb);
|
|
|
4210 else
|
|
|
4211 nextb_ic = TOUPPER_LOC(nextb);
|
|
|
4212 }
|
|
|
4213 else
|
|
|
4214 nextb_ic = nextb;
|
|
|
4215 }
|
|
|
4216 else
|
|
|
4217 {
|
|
|
4218 nextb = NUL;
|
|
|
4219 nextb_ic = NUL;
|
|
|
4220 }
|
|
|
4221 if (op != BRACE_SIMPLE)
|
|
|
4222 {
|
|
|
4223 minval = (op == STAR) ? 0 : 1;
|
|
|
4224 maxval = MAX_LIMIT;
|
|
|
4225 }
|
|
|
4226 else
|
|
|
4227 {
|
|
|
4228 minval = bl_minval;
|
|
|
4229 maxval = bl_maxval;
|
|
|
4230 }
|
|
|
4231
|
|
|
4232 /*
|
|
|
4233 * When maxval > minval, try matching as much as possible, up
|
|
|
4234 * to maxval. When maxval < minval, try matching at least the
|
|
|
4235 * minimal number (since the range is backwards, that's also
|
|
|
4236 * maxval!).
|
|
|
4237 */
|
|
|
4238 count = regrepeat(OPERAND(scan), maxval);
|
|
|
4239 if (got_int)
|
|
|
4240 return FALSE;
|
|
|
4241 if (minval <= maxval)
|
|
|
4242 {
|
|
|
4243 /* Range is the normal way around, use longest match */
|
|
|
4244 while (count >= minval)
|
|
|
4245 {
|
|
|
4246 /* If it could match, try it. */
|
|
|
4247 if (nextb == NUL || *reginput == nextb
|
|
|
4248 || *reginput == nextb_ic)
|
|
|
4249 {
|
|
|
4250 reg_save(&save);
|
|
|
4251 if (regmatch(next))
|
|
|
4252 return TRUE;
|
|
|
4253 reg_restore(&save);
|
|
|
4254 }
|
|
|
4255 /* Couldn't or didn't match -- back up one char. */
|
|
|
4256 if (--count < minval)
|
|
|
4257 break;
|
|
|
4258 if (reginput == regline)
|
|
|
4259 {
|
|
|
4260 /* backup to last char of previous line */
|
|
|
4261 --reglnum;
|
|
|
4262 regline = reg_getline(reglnum);
|
|
|
4263 /* Just in case regrepeat() didn't count right. */
|
|
|
4264 if (regline == NULL)
|
|
|
4265 return FALSE;
|
|
|
4266 reginput = regline + STRLEN(regline);
|
|
|
4267 fast_breakcheck();
|
|
|
4268 if (got_int || out_of_stack)
|
|
|
4269 return FALSE;
|
|
|
4270 }
|
|
|
4271 else
|
|
|
4272 {
|
|
|
4273 --reginput;
|
|
|
4274 #ifdef FEAT_MBYTE
|
|
|
4275 if (has_mbyte)
|
|
|
4276 reginput -= (*mb_head_off)(regline, reginput);
|
|
|
4277 #endif
|
|
|
4278 }
|
|
|
4279 }
|
|
|
4280 }
|
|
|
4281 else
|
|
|
4282 {
|
|
|
4283 /* Range is backwards, use shortest match first.
|
|
|
4284 * Careful: maxval and minval are exchanged! */
|
|
|
4285 if (count < maxval)
|
|
|
4286 return FALSE;
|
|
|
4287 for (;;)
|
|
|
4288 {
|
|
|
4289 /* If it could work, try it. */
|
|
|
4290 if (nextb == NUL || *reginput == nextb
|
|
|
4291 || *reginput == nextb_ic)
|
|
|
4292 {
|
|
|
4293 reg_save(&save);
|
|
|
4294 if (regmatch(next))
|
|
|
4295 return TRUE;
|
|
|
4296 reg_restore(&save);
|
|
|
4297 }
|
|
|
4298 /* Couldn't or didn't match: try advancing one char. */
|
|
|
4299 if (count == minval
|
|
|
4300 || regrepeat(OPERAND(scan), 1L) == 0)
|
|
|
4301 break;
|
|
|
4302 ++count;
|
|
|
4303 if (got_int || out_of_stack)
|
|
|
4304 return FALSE;
|
|
|
4305 }
|
|
|
4306 }
|
|
|
4307 return FALSE;
|
|
|
4308 }
|
|
|
4309 /* break; Not Reached */
|
|
|
4310
|
|
|
4311 case NOMATCH:
|
|
|
4312 {
|
|
|
4313 regsave_T save;
|
|
|
4314
|
|
|
4315 /* If the operand matches, we fail. Otherwise backup and
|
|
|
4316 * continue with the next item. */
|
|
|
4317 reg_save(&save);
|
|
|
4318 if (regmatch(OPERAND(scan)))
|
|
|
4319 return FALSE;
|
|
|
4320 reg_restore(&save);
|
|
|
4321 }
|
|
|
4322 break;
|
|
|
4323
|
|
|
4324 case MATCH:
|
|
|
4325 case SUBPAT:
|
|
|
4326 {
|
|
|
4327 regsave_T save;
|
|
|
4328
|
|
|
4329 /* If the operand doesn't match, we fail. Otherwise backup
|
|
|
4330 * and continue with the next item. */
|
|
|
4331 reg_save(&save);
|
|
|
4332 if (!regmatch(OPERAND(scan)))
|
|
|
4333 return FALSE;
|
|
|
4334 if (op == MATCH) /* zero-width */
|
|
|
4335 reg_restore(&save);
|
|
|
4336 }
|
|
|
4337 break;
|
|
|
4338
|
|
|
4339 case BEHIND:
|
|
|
4340 case NOBEHIND:
|
|
|
4341 {
|
|
|
4342 regsave_T save_after, save_start;
|
|
|
4343 regsave_T save_behind_pos;
|
|
|
4344 int needmatch = (op == BEHIND);
|
|
|
4345
|
|
|
4346 /*
|
|
|
4347 * Look back in the input of the operand matches or not. This
|
|
|
4348 * must be done at every position in the input and checking if
|
|
|
4349 * the match ends at the current position.
|
|
|
4350 * First check if the next item matches, that's probably
|
|
|
4351 * faster.
|
|
|
4352 */
|
|
|
4353 reg_save(&save_start);
|
|
|
4354 if (regmatch(next))
|
|
|
4355 {
|
|
|
4356 /* save the position after the found match for next */
|
|
|
4357 reg_save(&save_after);
|
|
|
4358
|
|
|
4359 /* start looking for a match with operand at the current
|
|
|
4360 * postion. Go back one character until we find the
|
|
|
4361 * result, hitting the start of the line or the previous
|
|
|
4362 * line (for multi-line matching).
|
|
|
4363 * Set behind_pos to where the match should end, BHPOS
|
|
|
4364 * will match it. */
|
|
|
4365 save_behind_pos = behind_pos;
|
|
|
4366 behind_pos = save_start;
|
|
|
4367 for (;;)
|
|
|
4368 {
|
|
|
4369 reg_restore(&save_start);
|
|
|
4370 if (regmatch(OPERAND(scan))
|
|
|
4371 && reg_save_equal(&behind_pos))
|
|
|
4372 {
|
|
|
4373 behind_pos = save_behind_pos;
|
|
|
4374 /* found a match that ends where "next" started */
|
|
|
4375 if (needmatch)
|
|
|
4376 {
|
|
|
4377 reg_restore(&save_after);
|
|
|
4378 return TRUE;
|
|
|
4379 }
|
|
|
4380 return FALSE;
|
|
|
4381 }
|
|
|
4382 /*
|
|
|
4383 * No match: Go back one character. May go to
|
|
|
4384 * previous line once.
|
|
|
4385 */
|
|
|
4386 if (REG_MULTI)
|
|
|
4387 {
|
|
|
4388 if (save_start.rs_u.pos.col == 0)
|
|
|
4389 {
|
|
|
4390 if (save_start.rs_u.pos.lnum
|
|
|
4391 < behind_pos.rs_u.pos.lnum
|
|
|
4392 || reg_getline(
|
|
|
4393 --save_start.rs_u.pos.lnum) == NULL)
|
|
|
4394 break;
|
|
|
4395 reg_restore(&save_start);
|
|
|
4396 save_start.rs_u.pos.col =
|
|
|
4397 (colnr_T)STRLEN(regline);
|
|
|
4398 }
|
|
|
4399 else
|
|
|
4400 --save_start.rs_u.pos.col;
|
|
|
4401 }
|
|
|
4402 else
|
|
|
4403 {
|
|
|
4404 if (save_start.rs_u.ptr == regline)
|
|
|
4405 break;
|
|
|
4406 --save_start.rs_u.ptr;
|
|
|
4407 }
|
|
|
4408 }
|
|
|
4409
|
|
|
4410 /* NOBEHIND succeeds when no match was found */
|
|
|
4411 behind_pos = save_behind_pos;
|
|
|
4412 if (!needmatch)
|
|
|
4413 {
|
|
|
4414 reg_restore(&save_after);
|
|
|
4415 return TRUE;
|
|
|
4416 }
|
|
|
4417 }
|
|
|
4418 return FALSE;
|
|
|
4419 }
|
|
|
4420
|
|
|
4421 case BHPOS:
|
|
|
4422 if (REG_MULTI)
|
|
|
4423 {
|
|
|
4424 if (behind_pos.rs_u.pos.col != (colnr_T)(reginput - regline)
|
|
|
4425 || behind_pos.rs_u.pos.lnum != reglnum)
|
|
|
4426 return FALSE;
|
|
|
4427 }
|
|
|
4428 else if (behind_pos.rs_u.ptr != reginput)
|
|
|
4429 return FALSE;
|
|
|
4430 break;
|
|
|
4431
|
|
|
4432 case NEWL:
|
|
|
4433 if ((c != NUL || reglnum == reg_maxline)
|
|
|
4434 && (c != '\n' || !reg_line_lbr))
|
|
|
4435 return FALSE;
|
|
|
4436 if (reg_line_lbr)
|
|
|
4437 ADVANCE_REGINPUT();
|
|
|
4438 else
|
|
|
4439 reg_nextline();
|
|
|
4440 break;
|
|
|
4441
|
|
|
4442 case END:
|
|
|
4443 return TRUE; /* Success! */
|
|
|
4444
|
|
|
4445 default:
|
|
|
4446 EMSG(_(e_re_corr));
|
|
|
4447 #ifdef DEBUG
|
|
|
4448 printf("Illegal op code %d\n", op);
|
|
|
4449 #endif
|
|
|
4450 return FALSE;
|
|
|
4451 }
|
|
|
4452 }
|
|
|
4453
|
|
|
4454 scan = next;
|
|
|
4455 }
|
|
|
4456
|
|
|
4457 /*
|
|
|
4458 * We get here only if there's trouble -- normally "case END" is the
|
|
|
4459 * terminating point.
|
|
|
4460 */
|
|
|
4461 EMSG(_(e_re_corr));
|
|
|
4462 #ifdef DEBUG
|
|
|
4463 printf("Premature EOL\n");
|
|
|
4464 #endif
|
|
|
4465 return FALSE;
|
|
|
4466 }
|
|
|
4467
|
|
|
4468 #ifdef FEAT_MBYTE
|
|
|
4469 # define ADVANCE_P(x) if (has_mbyte) x += (*mb_ptr2len_check)(x); else ++x
|
|
|
4470 #else
|
|
|
4471 # define ADVANCE_P(x) ++x
|
|
|
4472 #endif
|
|
|
4473
|
|
|
4474 /*
|
|
|
4475 * regrepeat - repeatedly match something simple, return how many.
|
|
|
4476 * Advances reginput (and reglnum) to just after the matched chars.
|
|
|
4477 */
|
|
|
4478 static int
|
|
|
4479 regrepeat(p, maxcount)
|
|
|
4480 char_u *p;
|
|
|
4481 long maxcount; /* maximum number of matches allowed */
|
|
|
4482 {
|
|
|
4483 long count = 0;
|
|
|
4484 char_u *scan;
|
|
|
4485 char_u *opnd;
|
|
|
4486 int mask;
|
|
|
4487 int testval = 0;
|
|
|
4488
|
|
|
4489 scan = reginput; /* Make local copy of reginput for speed. */
|
|
|
4490 opnd = OPERAND(p);
|
|
|
4491 switch (OP(p))
|
|
|
4492 {
|
|
|
4493 case ANY:
|
|
|
4494 case ANY + ADD_NL:
|
|
|
4495 while (count < maxcount)
|
|
|
4496 {
|
|
|
4497 /* Matching anything means we continue until end-of-line (or
|
|
|
4498 * end-of-file for ANY + ADD_NL), only limited by maxcount. */
|
|
|
4499 while (*scan != NUL && count < maxcount)
|
|
|
4500 {
|
|
|
4501 ++count;
|
|
|
4502 ADVANCE_P(scan);
|
|
|
4503 }
|
|
|
4504 if (!WITH_NL(OP(p)) || reglnum == reg_maxline || count == maxcount)
|
|
|
4505 break;
|
|
|
4506 ++count; /* count the line-break */
|
|
|
4507 reg_nextline();
|
|
|
4508 scan = reginput;
|
|
|
4509 if (got_int)
|
|
|
4510 break;
|
|
|
4511 }
|
|
|
4512 break;
|
|
|
4513
|
|
|
4514 case IDENT:
|
|
|
4515 case IDENT + ADD_NL:
|
|
|
4516 testval = TRUE;
|
|
|
4517 /*FALLTHROUGH*/
|
|
|
4518 case SIDENT:
|
|
|
4519 case SIDENT + ADD_NL:
|
|
|
4520 while (count < maxcount)
|
|
|
4521 {
|
|
|
4522 if (vim_isIDc(*scan) && (testval || !VIM_ISDIGIT(*scan)))
|
|
|
4523 {
|
|
|
4524 ADVANCE_P(scan);
|
|
|
4525 }
|
|
|
4526 else if (*scan == NUL)
|
|
|
4527 {
|
|
|
4528 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
|
4529 break;
|
|
|
4530 reg_nextline();
|
|
|
4531 scan = reginput;
|
|
|
4532 if (got_int)
|
|
|
4533 break;
|
|
|
4534 }
|
|
|
4535 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
|
4536 ++scan;
|
|
|
4537 else
|
|
|
4538 break;
|
|
|
4539 ++count;
|
|
|
4540 }
|
|
|
4541 break;
|
|
|
4542
|
|
|
4543 case KWORD:
|
|
|
4544 case KWORD + ADD_NL:
|
|
|
4545 testval = TRUE;
|
|
|
4546 /*FALLTHROUGH*/
|
|
|
4547 case SKWORD:
|
|
|
4548 case SKWORD + ADD_NL:
|
|
|
4549 while (count < maxcount)
|
|
|
4550 {
|
|
|
4551 if (vim_iswordp(scan) && (testval || !VIM_ISDIGIT(*scan)))
|
|
|
4552 {
|
|
|
4553 ADVANCE_P(scan);
|
|
|
4554 }
|
|
|
4555 else if (*scan == NUL)
|
|
|
4556 {
|
|
|
4557 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
|
4558 break;
|
|
|
4559 reg_nextline();
|
|
|
4560 scan = reginput;
|
|
|
4561 if (got_int)
|
|
|
4562 break;
|
|
|
4563 }
|
|
|
4564 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
|
4565 ++scan;
|
|
|
4566 else
|
|
|
4567 break;
|
|
|
4568 ++count;
|
|
|
4569 }
|
|
|
4570 break;
|
|
|
4571
|
|
|
4572 case FNAME:
|
|
|
4573 case FNAME + ADD_NL:
|
|
|
4574 testval = TRUE;
|
|
|
4575 /*FALLTHROUGH*/
|
|
|
4576 case SFNAME:
|
|
|
4577 case SFNAME + ADD_NL:
|
|
|
4578 while (count < maxcount)
|
|
|
4579 {
|
|
|
4580 if (vim_isfilec(*scan) && (testval || !VIM_ISDIGIT(*scan)))
|
|
|
4581 {
|
|
|
4582 ADVANCE_P(scan);
|
|
|
4583 }
|
|
|
4584 else if (*scan == NUL)
|
|
|
4585 {
|
|
|
4586 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
|
4587 break;
|
|
|
4588 reg_nextline();
|
|
|
4589 scan = reginput;
|
|
|
4590 if (got_int)
|
|
|
4591 break;
|
|
|
4592 }
|
|
|
4593 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
|
4594 ++scan;
|
|
|
4595 else
|
|
|
4596 break;
|
|
|
4597 ++count;
|
|
|
4598 }
|
|
|
4599 break;
|
|
|
4600
|
|
|
4601 case PRINT:
|
|
|
4602 case PRINT + ADD_NL:
|
|
|
4603 testval = TRUE;
|
|
|
4604 /*FALLTHROUGH*/
|
|
|
4605 case SPRINT:
|
|
|
4606 case SPRINT + ADD_NL:
|
|
|
4607 while (count < maxcount)
|
|
|
4608 {
|
|
|
4609 if (*scan == NUL)
|
|
|
4610 {
|
|
|
4611 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
|
4612 break;
|
|
|
4613 reg_nextline();
|
|
|
4614 scan = reginput;
|
|
|
4615 if (got_int)
|
|
|
4616 break;
|
|
|
4617 }
|
|
|
4618 else if (ptr2cells(scan) == 1 && (testval || !VIM_ISDIGIT(*scan)))
|
|
|
4619 {
|
|
|
4620 ADVANCE_P(scan);
|
|
|
4621 }
|
|
|
4622 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
|
4623 ++scan;
|
|
|
4624 else
|
|
|
4625 break;
|
|
|
4626 ++count;
|
|
|
4627 }
|
|
|
4628 break;
|
|
|
4629
|
|
|
4630 case WHITE:
|
|
|
4631 case WHITE + ADD_NL:
|
|
|
4632 testval = mask = RI_WHITE;
|
|
|
4633 do_class:
|
|
|
4634 while (count < maxcount)
|
|
|
4635 {
|
|
|
4636 #ifdef FEAT_MBYTE
|
|
|
4637 int l;
|
|
|
4638 #endif
|
|
|
4639 if (*scan == NUL)
|
|
|
4640 {
|
|
|
4641 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
|
4642 break;
|
|
|
4643 reg_nextline();
|
|
|
4644 scan = reginput;
|
|
|
4645 if (got_int)
|
|
|
4646 break;
|
|
|
4647 }
|
|
|
4648 #ifdef FEAT_MBYTE
|
|
|
4649 else if (has_mbyte && (l = (*mb_ptr2len_check)(scan)) > 1)
|
|
|
4650 {
|
|
|
4651 if (testval != 0)
|
|
|
4652 break;
|
|
|
4653 scan += l;
|
|
|
4654 }
|
|
|
4655 #endif
|
|
|
4656 else if ((class_tab[*scan] & mask) == testval)
|
|
|
4657 ++scan;
|
|
|
4658 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
|
4659 ++scan;
|
|
|
4660 else
|
|
|
4661 break;
|
|
|
4662 ++count;
|
|
|
4663 }
|
|
|
4664 break;
|
|
|
4665
|
|
|
4666 case NWHITE:
|
|
|
4667 case NWHITE + ADD_NL:
|
|
|
4668 mask = RI_WHITE;
|
|
|
4669 goto do_class;
|
|
|
4670 case DIGIT:
|
|
|
4671 case DIGIT + ADD_NL:
|
|
|
4672 testval = mask = RI_DIGIT;
|
|
|
4673 goto do_class;
|
|
|
4674 case NDIGIT:
|
|
|
4675 case NDIGIT + ADD_NL:
|
|
|
4676 mask = RI_DIGIT;
|
|
|
4677 goto do_class;
|
|
|
4678 case HEX:
|
|
|
4679 case HEX + ADD_NL:
|
|
|
4680 testval = mask = RI_HEX;
|
|
|
4681 goto do_class;
|
|
|
4682 case NHEX:
|
|
|
4683 case NHEX + ADD_NL:
|
|
|
4684 mask = RI_HEX;
|
|
|
4685 goto do_class;
|
|
|
4686 case OCTAL:
|
|
|
4687 case OCTAL + ADD_NL:
|
|
|
4688 testval = mask = RI_OCTAL;
|
|
|
4689 goto do_class;
|
|
|
4690 case NOCTAL:
|
|
|
4691 case NOCTAL + ADD_NL:
|
|
|
4692 mask = RI_OCTAL;
|
|
|
4693 goto do_class;
|
|
|
4694 case WORD:
|
|
|
4695 case WORD + ADD_NL:
|
|
|
4696 testval = mask = RI_WORD;
|
|
|
4697 goto do_class;
|
|
|
4698 case NWORD:
|
|
|
4699 case NWORD + ADD_NL:
|
|
|
4700 mask = RI_WORD;
|
|
|
4701 goto do_class;
|
|
|
4702 case HEAD:
|
|
|
4703 case HEAD + ADD_NL:
|
|
|
4704 testval = mask = RI_HEAD;
|
|
|
4705 goto do_class;
|
|
|
4706 case NHEAD:
|
|
|
4707 case NHEAD + ADD_NL:
|
|
|
4708 mask = RI_HEAD;
|
|
|
4709 goto do_class;
|
|
|
4710 case ALPHA:
|
|
|
4711 case ALPHA + ADD_NL:
|
|
|
4712 testval = mask = RI_ALPHA;
|
|
|
4713 goto do_class;
|
|
|
4714 case NALPHA:
|
|
|
4715 case NALPHA + ADD_NL:
|
|
|
4716 mask = RI_ALPHA;
|
|
|
4717 goto do_class;
|
|
|
4718 case LOWER:
|
|
|
4719 case LOWER + ADD_NL:
|
|
|
4720 testval = mask = RI_LOWER;
|
|
|
4721 goto do_class;
|
|
|
4722 case NLOWER:
|
|
|
4723 case NLOWER + ADD_NL:
|
|
|
4724 mask = RI_LOWER;
|
|
|
4725 goto do_class;
|
|
|
4726 case UPPER:
|
|
|
4727 case UPPER + ADD_NL:
|
|
|
4728 testval = mask = RI_UPPER;
|
|
|
4729 goto do_class;
|
|
|
4730 case NUPPER:
|
|
|
4731 case NUPPER + ADD_NL:
|
|
|
4732 mask = RI_UPPER;
|
|
|
4733 goto do_class;
|
|
|
4734
|
|
|
4735 case EXACTLY:
|
|
|
4736 {
|
|
|
4737 int cu, cl;
|
|
|
4738
|
|
|
4739 /* This doesn't do a multi-byte character, because a MULTIBYTECODE
|
|
|
4740 * would have been used for it. */
|
|
|
4741 if (ireg_ic)
|
|
|
4742 {
|
|
|
4743 cu = TOUPPER_LOC(*opnd);
|
|
|
4744 cl = TOLOWER_LOC(*opnd);
|
|
|
4745 while (count < maxcount && (*scan == cu || *scan == cl))
|
|
|
4746 {
|
|
|
4747 count++;
|
|
|
4748 scan++;
|
|
|
4749 }
|
|
|
4750 }
|
|
|
4751 else
|
|
|
4752 {
|
|
|
4753 cu = *opnd;
|
|
|
4754 while (count < maxcount && *scan == cu)
|
|
|
4755 {
|
|
|
4756 count++;
|
|
|
4757 scan++;
|
|
|
4758 }
|
|
|
4759 }
|
|
|
4760 break;
|
|
|
4761 }
|
|
|
4762
|
|
|
4763 #ifdef FEAT_MBYTE
|
|
|
4764 case MULTIBYTECODE:
|
|
|
4765 {
|
|
|
4766 int i, len, cf = 0;
|
|
|
4767
|
|
|
4768 /* Safety check (just in case 'encoding' was changed since
|
|
|
4769 * compiling the program). */
|
|
|
4770 if ((len = (*mb_ptr2len_check)(opnd)) > 1)
|
|
|
4771 {
|
|
|
4772 if (ireg_ic && enc_utf8)
|
|
|
4773 cf = utf_fold(utf_ptr2char(opnd));
|
|
|
4774 while (count < maxcount)
|
|
|
4775 {
|
|
|
4776 for (i = 0; i < len; ++i)
|
|
|
4777 if (opnd[i] != scan[i])
|
|
|
4778 break;
|
|
|
4779 if (i < len && (!ireg_ic || !enc_utf8
|
|
|
4780 || utf_fold(utf_ptr2char(scan)) != cf))
|
|
|
4781 break;
|
|
|
4782 scan += len;
|
|
|
4783 ++count;
|
|
|
4784 }
|
|
|
4785 }
|
|
|
4786 }
|
|
|
4787 break;
|
|
|
4788 #endif
|
|
|
4789
|
|
|
4790 case ANYOF:
|
|
|
4791 case ANYOF + ADD_NL:
|
|
|
4792 testval = TRUE;
|
|
|
4793 /*FALLTHROUGH*/
|
|
|
4794
|
|
|
4795 case ANYBUT:
|
|
|
4796 case ANYBUT + ADD_NL:
|
|
|
4797 while (count < maxcount)
|
|
|
4798 {
|
|
|
4799 #ifdef FEAT_MBYTE
|
|
|
4800 int len;
|
|
|
4801 #endif
|
|
|
4802 if (*scan == NUL)
|
|
|
4803 {
|
|
|
4804 if (!WITH_NL(OP(p)) || reglnum == reg_maxline)
|
|
|
4805 break;
|
|
|
4806 reg_nextline();
|
|
|
4807 scan = reginput;
|
|
|
4808 if (got_int)
|
|
|
4809 break;
|
|
|
4810 }
|
|
|
4811 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p)))
|
|
|
4812 ++scan;
|
|
|
4813 #ifdef FEAT_MBYTE
|
|
|
4814 else if (has_mbyte && (len = (*mb_ptr2len_check)(scan)) > 1)
|
|
|
4815 {
|
|
|
4816 if ((cstrchr(opnd, (*mb_ptr2char)(scan)) == NULL) == testval)
|
|
|
4817 break;
|
|
|
4818 scan += len;
|
|
|
4819 }
|
|
|
4820 #endif
|
|
|
4821 else
|
|
|
4822 {
|
|
|
4823 if ((cstrchr(opnd, *scan) == NULL) == testval)
|
|
|
4824 break;
|
|
|
4825 ++scan;
|
|
|
4826 }
|
|
|
4827 ++count;
|
|
|
4828 }
|
|
|
4829 break;
|
|
|
4830
|
|
|
4831 case NEWL:
|
|
|
4832 while (count < maxcount
|
|
|
4833 && ((*scan == NUL && reglnum < reg_maxline)
|
|
|
4834 || (*scan == '\n' && reg_line_lbr)))
|
|
|
4835 {
|
|
|
4836 count++;
|
|
|
4837 if (reg_line_lbr)
|
|
|
4838 ADVANCE_REGINPUT();
|
|
|
4839 else
|
|
|
4840 reg_nextline();
|
|
|
4841 scan = reginput;
|
|
|
4842 if (got_int)
|
|
|
4843 break;
|
|
|
4844 }
|
|
|
4845 break;
|
|
|
4846
|
|
|
4847 default: /* Oh dear. Called inappropriately. */
|
|
|
4848 EMSG(_(e_re_corr));
|
|
|
4849 #ifdef DEBUG
|
|
|
4850 printf("Called regrepeat with op code %d\n", OP(p));
|
|
|
4851 #endif
|
|
|
4852 break;
|
|
|
4853 }
|
|
|
4854
|
|
|
4855 reginput = scan;
|
|
|
4856
|
|
|
4857 return (int)count;
|
|
|
4858 }
|
|
|
4859
|
|
|
4860 /*
|
|
|
4861 * regnext - dig the "next" pointer out of a node
|
|
|
4862 */
|
|
|
4863 static char_u *
|
|
|
4864 regnext(p)
|
|
|
4865 char_u *p;
|
|
|
4866 {
|
|
|
4867 int offset;
|
|
|
4868
|
|
|
4869 if (p == JUST_CALC_SIZE)
|
|
|
4870 return NULL;
|
|
|
4871
|
|
|
4872 offset = NEXT(p);
|
|
|
4873 if (offset == 0)
|
|
|
4874 return NULL;
|
|
|
4875
|
|
|
4876 if (OP(p) == BACK)
|
|
|
4877 return p - offset;
|
|
|
4878 else
|
|
|
4879 return p + offset;
|
|
|
4880 }
|
|
|
4881
|
|
|
4882 /*
|
|
|
4883 * Check the regexp program for its magic number.
|
|
|
4884 * Return TRUE if it's wrong.
|
|
|
4885 */
|
|
|
4886 static int
|
|
|
4887 prog_magic_wrong()
|
|
|
4888 {
|
|
|
4889 if (UCHARAT(REG_MULTI
|
|
|
4890 ? reg_mmatch->regprog->program
|
|
|
4891 : reg_match->regprog->program) != REGMAGIC)
|
|
|
4892 {
|
|
|
4893 EMSG(_(e_re_corr));
|
|
|
4894 return TRUE;
|
|
|
4895 }
|
|
|
4896 return FALSE;
|
|
|
4897 }
|
|
|
4898
|
|
|
4899 /*
|
|
|
4900 * Cleanup the subexpressions, if this wasn't done yet.
|
|
|
4901 * This construction is used to clear the subexpressions only when they are
|
|
|
4902 * used (to increase speed).
|
|
|
4903 */
|
|
|
4904 static void
|
|
|
4905 cleanup_subexpr()
|
|
|
4906 {
|
|
|
4907 if (need_clear_subexpr)
|
|
|
4908 {
|
|
|
4909 if (REG_MULTI)
|
|
|
4910 {
|
|
|
4911 /* Use 0xff to set lnum to -1 */
|
|
|
4912 vim_memset(reg_startpos, 0xff, sizeof(lpos_T) * NSUBEXP);
|
|
|
4913 vim_memset(reg_endpos, 0xff, sizeof(lpos_T) * NSUBEXP);
|
|
|
4914 }
|
|
|
4915 else
|
|
|
4916 {
|
|
|
4917 vim_memset(reg_startp, 0, sizeof(char_u *) * NSUBEXP);
|
|
|
4918 vim_memset(reg_endp, 0, sizeof(char_u *) * NSUBEXP);
|
|
|
4919 }
|
|
|
4920 need_clear_subexpr = FALSE;
|
|
|
4921 }
|
|
|
4922 }
|
|
|
4923
|
|
|
4924 #ifdef FEAT_SYN_HL
|
|
|
4925 static void
|
|
|
4926 cleanup_zsubexpr()
|
|
|
4927 {
|
|
|
4928 if (need_clear_zsubexpr)
|
|
|
4929 {
|
|
|
4930 if (REG_MULTI)
|
|
|
4931 {
|
|
|
4932 /* Use 0xff to set lnum to -1 */
|
|
|
4933 vim_memset(reg_startzpos, 0xff, sizeof(lpos_T) * NSUBEXP);
|
|
|
4934 vim_memset(reg_endzpos, 0xff, sizeof(lpos_T) * NSUBEXP);
|
|
|
4935 }
|
|
|
4936 else
|
|
|
4937 {
|
|
|
4938 vim_memset(reg_startzp, 0, sizeof(char_u *) * NSUBEXP);
|
|
|
4939 vim_memset(reg_endzp, 0, sizeof(char_u *) * NSUBEXP);
|
|
|
4940 }
|
|
|
4941 need_clear_zsubexpr = FALSE;
|
|
|
4942 }
|
|
|
4943 }
|
|
|
4944 #endif
|
|
|
4945
|
|
|
4946 /*
|
|
|
4947 * Advance reglnum, regline and reginput to the next line.
|
|
|
4948 */
|
|
|
4949 static void
|
|
|
4950 reg_nextline()
|
|
|
4951 {
|
|
|
4952 regline = reg_getline(++reglnum);
|
|
|
4953 reginput = regline;
|
|
|
4954 fast_breakcheck();
|
|
|
4955 }
|
|
|
4956
|
|
|
4957 /*
|
|
|
4958 * Save the input line and position in a regsave_T.
|
|
|
4959 */
|
|
|
4960 static void
|
|
|
4961 reg_save(save)
|
|
|
4962 regsave_T *save;
|
|
|
4963 {
|
|
|
4964 if (REG_MULTI)
|
|
|
4965 {
|
|
|
4966 save->rs_u.pos.col = (colnr_T)(reginput - regline);
|
|
|
4967 save->rs_u.pos.lnum = reglnum;
|
|
|
4968 }
|
|
|
4969 else
|
|
|
4970 save->rs_u.ptr = reginput;
|
|
|
4971 }
|
|
|
4972
|
|
|
4973 /*
|
|
|
4974 * Restore the input line and position from a regsave_T.
|
|
|
4975 */
|
|
|
4976 static void
|
|
|
4977 reg_restore(save)
|
|
|
4978 regsave_T *save;
|
|
|
4979 {
|
|
|
4980 if (REG_MULTI)
|
|
|
4981 {
|
|
|
4982 if (reglnum != save->rs_u.pos.lnum)
|
|
|
4983 {
|
|
|
4984 /* only call reg_getline() when the line number changed to save
|
|
|
4985 * a bit of time */
|
|
|
4986 reglnum = save->rs_u.pos.lnum;
|
|
|
4987 regline = reg_getline(reglnum);
|
|
|
4988 }
|
|
|
4989 reginput = regline + save->rs_u.pos.col;
|
|
|
4990 }
|
|
|
4991 else
|
|
|
4992 reginput = save->rs_u.ptr;
|
|
|
4993 }
|
|
|
4994
|
|
|
4995 /*
|
|
|
4996 * Return TRUE if current position is equal to saved position.
|
|
|
4997 */
|
|
|
4998 static int
|
|
|
4999 reg_save_equal(save)
|
|
|
5000 regsave_T *save;
|
|
|
5001 {
|
|
|
5002 if (REG_MULTI)
|
|
|
5003 return reglnum == save->rs_u.pos.lnum
|
|
|
5004 && reginput == regline + save->rs_u.pos.col;
|
|
|
5005 return reginput == save->rs_u.ptr;
|
|
|
5006 }
|
|
|
5007
|
|
|
5008 /*
|
|
|
5009 * Tentatively set the sub-expression start to the current position (after
|
|
|
5010 * calling regmatch() they will have changed). Need to save the existing
|
|
|
5011 * values for when there is no match.
|
|
|
5012 * Use se_save() to use pointer (save_se_multi()) or position (save_se_one()),
|
|
|
5013 * depending on REG_MULTI.
|
|
|
5014 */
|
|
|
5015 static void
|
|
|
5016 save_se_multi(savep, posp)
|
|
|
5017 save_se_T *savep;
|
|
|
5018 lpos_T *posp;
|
|
|
5019 {
|
|
|
5020 savep->se_u.pos = *posp;
|
|
|
5021 posp->lnum = reglnum;
|
|
|
5022 posp->col = (colnr_T)(reginput - regline);
|
|
|
5023 }
|
|
|
5024
|
|
|
5025 static void
|
|
|
5026 save_se_one(savep, pp)
|
|
|
5027 save_se_T *savep;
|
|
|
5028 char_u **pp;
|
|
|
5029 {
|
|
|
5030 savep->se_u.ptr = *pp;
|
|
|
5031 *pp = reginput;
|
|
|
5032 }
|
|
|
5033
|
|
|
5034 /*
|
|
|
5035 * Compare a number with the operand of RE_LNUM, RE_COL or RE_VCOL.
|
|
|
5036 */
|
|
|
5037 static int
|
|
|
5038 re_num_cmp(val, scan)
|
|
|
5039 long_u val;
|
|
|
5040 char_u *scan;
|
|
|
5041 {
|
|
|
5042 long_u n = OPERAND_MIN(scan);
|
|
|
5043
|
|
|
5044 if (OPERAND_CMP(scan) == '>')
|
|
|
5045 return val > n;
|
|
|
5046 if (OPERAND_CMP(scan) == '<')
|
|
|
5047 return val < n;
|
|
|
5048 return val == n;
|
|
|
5049 }
|
|
|
5050
|
|
|
5051
|
|
|
5052 #ifdef DEBUG
|
|
|
5053
|
|
|
5054 /*
|
|
|
5055 * regdump - dump a regexp onto stdout in vaguely comprehensible form
|
|
|
5056 */
|
|
|
5057 static void
|
|
|
5058 regdump(pattern, r)
|
|
|
5059 char_u *pattern;
|
|
|
5060 regprog_T *r;
|
|
|
5061 {
|
|
|
5062 char_u *s;
|
|
|
5063 int op = EXACTLY; /* Arbitrary non-END op. */
|
|
|
5064 char_u *next;
|
|
|
5065 char_u *end = NULL;
|
|
|
5066
|
|
|
5067 printf("\r\nregcomp(%s):\r\n", pattern);
|
|
|
5068
|
|
|
5069 s = r->program + 1;
|
|
|
5070 /*
|
|
|
5071 * Loop until we find the END that isn't before a referred next (an END
|
|
|
5072 * can also appear in a NOMATCH operand).
|
|
|
5073 */
|
|
|
5074 while (op != END || s <= end)
|
|
|
5075 {
|
|
|
5076 op = OP(s);
|
|
|
5077 printf("%2d%s", (int)(s - r->program), regprop(s)); /* Where, what. */
|
|
|
5078 next = regnext(s);
|
|
|
5079 if (next == NULL) /* Next ptr. */
|
|
|
5080 printf("(0)");
|
|
|
5081 else
|
|
|
5082 printf("(%d)", (int)((s - r->program) + (next - s)));
|
|
|
5083 if (end < next)
|
|
|
5084 end = next;
|
|
|
5085 if (op == BRACE_LIMITS)
|
|
|
5086 {
|
|
|
5087 /* Two short ints */
|
|
|
5088 printf(" minval %ld, maxval %ld", OPERAND_MIN(s), OPERAND_MAX(s));
|
|
|
5089 s += 8;
|
|
|
5090 }
|
|
|
5091 s += 3;
|
|
|
5092 if (op == ANYOF || op == ANYOF + ADD_NL
|
|
|
5093 || op == ANYBUT || op == ANYBUT + ADD_NL
|
|
|
5094 || op == EXACTLY)
|
|
|
5095 {
|
|
|
5096 /* Literal string, where present. */
|
|
|
5097 while (*s != NUL)
|
|
|
5098 printf("%c", *s++);
|
|
|
5099 s++;
|
|
|
5100 }
|
|
|
5101 printf("\r\n");
|
|
|
5102 }
|
|
|
5103
|
|
|
5104 /* Header fields of interest. */
|
|
|
5105 if (r->regstart != NUL)
|
|
|
5106 printf("start `%s' 0x%x; ", r->regstart < 256
|
|
|
5107 ? (char *)transchar(r->regstart)
|
|
|
5108 : "multibyte", r->regstart);
|
|
|
5109 if (r->reganch)
|
|
|
5110 printf("anchored; ");
|
|
|
5111 if (r->regmust != NULL)
|
|
|
5112 printf("must have \"%s\"", r->regmust);
|
|
|
5113 printf("\r\n");
|
|
|
5114 }
|
|
|
5115
|
|
|
5116 /*
|
|
|
5117 * regprop - printable representation of opcode
|
|
|
5118 */
|
|
|
5119 static char_u *
|
|
|
5120 regprop(op)
|
|
|
5121 char_u *op;
|
|
|
5122 {
|
|
|
5123 char_u *p;
|
|
|
5124 static char_u buf[50];
|
|
|
5125
|
|
|
5126 (void) strcpy(buf, ":");
|
|
|
5127
|
|
|
5128 switch (OP(op))
|
|
|
5129 {
|
|
|
5130 case BOL:
|
|
|
5131 p = "BOL";
|
|
|
5132 break;
|
|
|
5133 case EOL:
|
|
|
5134 p = "EOL";
|
|
|
5135 break;
|
|
|
5136 case RE_BOF:
|
|
|
5137 p = "BOF";
|
|
|
5138 break;
|
|
|
5139 case RE_EOF:
|
|
|
5140 p = "EOF";
|
|
|
5141 break;
|
|
|
5142 case CURSOR:
|
|
|
5143 p = "CURSOR";
|
|
|
5144 break;
|
|
|
5145 case RE_LNUM:
|
|
|
5146 p = "RE_LNUM";
|
|
|
5147 break;
|
|
|
5148 case RE_COL:
|
|
|
5149 p = "RE_COL";
|
|
|
5150 break;
|
|
|
5151 case RE_VCOL:
|
|
|
5152 p = "RE_VCOL";
|
|
|
5153 break;
|
|
|
5154 case BOW:
|
|
|
5155 p = "BOW";
|
|
|
5156 break;
|
|
|
5157 case EOW:
|
|
|
5158 p = "EOW";
|
|
|
5159 break;
|
|
|
5160 case ANY:
|
|
|
5161 p = "ANY";
|
|
|
5162 break;
|
|
|
5163 case ANY + ADD_NL:
|
|
|
5164 p = "ANY+NL";
|
|
|
5165 break;
|
|
|
5166 case ANYOF:
|
|
|
5167 p = "ANYOF";
|
|
|
5168 break;
|
|
|
5169 case ANYOF + ADD_NL:
|
|
|
5170 p = "ANYOF+NL";
|
|
|
5171 break;
|
|
|
5172 case ANYBUT:
|
|
|
5173 p = "ANYBUT";
|
|
|
5174 break;
|
|
|
5175 case ANYBUT + ADD_NL:
|
|
|
5176 p = "ANYBUT+NL";
|
|
|
5177 break;
|
|
|
5178 case IDENT:
|
|
|
5179 p = "IDENT";
|
|
|
5180 break;
|
|
|
5181 case IDENT + ADD_NL:
|
|
|
5182 p = "IDENT+NL";
|
|
|
5183 break;
|
|
|
5184 case SIDENT:
|
|
|
5185 p = "SIDENT";
|
|
|
5186 break;
|
|
|
5187 case SIDENT + ADD_NL:
|
|
|
5188 p = "SIDENT+NL";
|
|
|
5189 break;
|
|
|
5190 case KWORD:
|
|
|
5191 p = "KWORD";
|
|
|
5192 break;
|
|
|
5193 case KWORD + ADD_NL:
|
|
|
5194 p = "KWORD+NL";
|
|
|
5195 break;
|
|
|
5196 case SKWORD:
|
|
|
5197 p = "SKWORD";
|
|
|
5198 break;
|
|
|
5199 case SKWORD + ADD_NL:
|
|
|
5200 p = "SKWORD+NL";
|
|
|
5201 break;
|
|
|
5202 case FNAME:
|
|
|
5203 p = "FNAME";
|
|
|
5204 break;
|
|
|
5205 case FNAME + ADD_NL:
|
|
|
5206 p = "FNAME+NL";
|
|
|
5207 break;
|
|
|
5208 case SFNAME:
|
|
|
5209 p = "SFNAME";
|
|
|
5210 break;
|
|
|
5211 case SFNAME + ADD_NL:
|
|
|
5212 p = "SFNAME+NL";
|
|
|
5213 break;
|
|
|
5214 case PRINT:
|
|
|
5215 p = "PRINT";
|
|
|
5216 break;
|
|
|
5217 case PRINT + ADD_NL:
|
|
|
5218 p = "PRINT+NL";
|
|
|
5219 break;
|
|
|
5220 case SPRINT:
|
|
|
5221 p = "SPRINT";
|
|
|
5222 break;
|
|
|
5223 case SPRINT + ADD_NL:
|
|
|
5224 p = "SPRINT+NL";
|
|
|
5225 break;
|
|
|
5226 case WHITE:
|
|
|
5227 p = "WHITE";
|
|
|
5228 break;
|
|
|
5229 case WHITE + ADD_NL:
|
|
|
5230 p = "WHITE+NL";
|
|
|
5231 break;
|
|
|
5232 case NWHITE:
|
|
|
5233 p = "NWHITE";
|
|
|
5234 break;
|
|
|
5235 case NWHITE + ADD_NL:
|
|
|
5236 p = "NWHITE+NL";
|
|
|
5237 break;
|
|
|
5238 case DIGIT:
|
|
|
5239 p = "DIGIT";
|
|
|
5240 break;
|
|
|
5241 case DIGIT + ADD_NL:
|
|
|
5242 p = "DIGIT+NL";
|
|
|
5243 break;
|
|
|
5244 case NDIGIT:
|
|
|
5245 p = "NDIGIT";
|
|
|
5246 break;
|
|
|
5247 case NDIGIT + ADD_NL:
|
|
|
5248 p = "NDIGIT+NL";
|
|
|
5249 break;
|
|
|
5250 case HEX:
|
|
|
5251 p = "HEX";
|
|
|
5252 break;
|
|
|
5253 case HEX + ADD_NL:
|
|
|
5254 p = "HEX+NL";
|
|
|
5255 break;
|
|
|
5256 case NHEX:
|
|
|
5257 p = "NHEX";
|
|
|
5258 break;
|
|
|
5259 case NHEX + ADD_NL:
|
|
|
5260 p = "NHEX+NL";
|
|
|
5261 break;
|
|
|
5262 case OCTAL:
|
|
|
5263 p = "OCTAL";
|
|
|
5264 break;
|
|
|
5265 case OCTAL + ADD_NL:
|
|
|
5266 p = "OCTAL+NL";
|
|
|
5267 break;
|
|
|
5268 case NOCTAL:
|
|
|
5269 p = "NOCTAL";
|
|
|
5270 break;
|
|
|
5271 case NOCTAL + ADD_NL:
|
|
|
5272 p = "NOCTAL+NL";
|
|
|
5273 break;
|
|
|
5274 case WORD:
|
|
|
5275 p = "WORD";
|
|
|
5276 break;
|
|
|
5277 case WORD + ADD_NL:
|
|
|
5278 p = "WORD+NL";
|
|
|
5279 break;
|
|
|
5280 case NWORD:
|
|
|
5281 p = "NWORD";
|
|
|
5282 break;
|
|
|
5283 case NWORD + ADD_NL:
|
|
|
5284 p = "NWORD+NL";
|
|
|
5285 break;
|
|
|
5286 case HEAD:
|
|
|
5287 p = "HEAD";
|
|
|
5288 break;
|
|
|
5289 case HEAD + ADD_NL:
|
|
|
5290 p = "HEAD+NL";
|
|
|
5291 break;
|
|
|
5292 case NHEAD:
|
|
|
5293 p = "NHEAD";
|
|
|
5294 break;
|
|
|
5295 case NHEAD + ADD_NL:
|
|
|
5296 p = "NHEAD+NL";
|
|
|
5297 break;
|
|
|
5298 case ALPHA:
|
|
|
5299 p = "ALPHA";
|
|
|
5300 break;
|
|
|
5301 case ALPHA + ADD_NL:
|
|
|
5302 p = "ALPHA+NL";
|
|
|
5303 break;
|
|
|
5304 case NALPHA:
|
|
|
5305 p = "NALPHA";
|
|
|
5306 break;
|
|
|
5307 case NALPHA + ADD_NL:
|
|
|
5308 p = "NALPHA+NL";
|
|
|
5309 break;
|
|
|
5310 case LOWER:
|
|
|
5311 p = "LOWER";
|
|
|
5312 break;
|
|
|
5313 case LOWER + ADD_NL:
|
|
|
5314 p = "LOWER+NL";
|
|
|
5315 break;
|
|
|
5316 case NLOWER:
|
|
|
5317 p = "NLOWER";
|
|
|
5318 break;
|
|
|
5319 case NLOWER + ADD_NL:
|
|
|
5320 p = "NLOWER+NL";
|
|
|
5321 break;
|
|
|
5322 case UPPER:
|
|
|
5323 p = "UPPER";
|
|
|
5324 break;
|
|
|
5325 case UPPER + ADD_NL:
|
|
|
5326 p = "UPPER+NL";
|
|
|
5327 break;
|
|
|
5328 case NUPPER:
|
|
|
5329 p = "NUPPER";
|
|
|
5330 break;
|
|
|
5331 case NUPPER + ADD_NL:
|
|
|
5332 p = "NUPPER+NL";
|
|
|
5333 break;
|
|
|
5334 case BRANCH:
|
|
|
5335 p = "BRANCH";
|
|
|
5336 break;
|
|
|
5337 case EXACTLY:
|
|
|
5338 p = "EXACTLY";
|
|
|
5339 break;
|
|
|
5340 case NOTHING:
|
|
|
5341 p = "NOTHING";
|
|
|
5342 break;
|
|
|
5343 case BACK:
|
|
|
5344 p = "BACK";
|
|
|
5345 break;
|
|
|
5346 case END:
|
|
|
5347 p = "END";
|
|
|
5348 break;
|
|
|
5349 case MOPEN + 0:
|
|
|
5350 p = "MATCH START";
|
|
|
5351 break;
|
|
|
5352 case MOPEN + 1:
|
|
|
5353 case MOPEN + 2:
|
|
|
5354 case MOPEN + 3:
|
|
|
5355 case MOPEN + 4:
|
|
|
5356 case MOPEN + 5:
|
|
|
5357 case MOPEN + 6:
|
|
|
5358 case MOPEN + 7:
|
|
|
5359 case MOPEN + 8:
|
|
|
5360 case MOPEN + 9:
|
|
|
5361 sprintf(buf + STRLEN(buf), "MOPEN%d", OP(op) - MOPEN);
|
|
|
5362 p = NULL;
|
|
|
5363 break;
|
|
|
5364 case MCLOSE + 0:
|
|
|
5365 p = "MATCH END";
|
|
|
5366 break;
|
|
|
5367 case MCLOSE + 1:
|
|
|
5368 case MCLOSE + 2:
|
|
|
5369 case MCLOSE + 3:
|
|
|
5370 case MCLOSE + 4:
|
|
|
5371 case MCLOSE + 5:
|
|
|
5372 case MCLOSE + 6:
|
|
|
5373 case MCLOSE + 7:
|
|
|
5374 case MCLOSE + 8:
|
|
|
5375 case MCLOSE + 9:
|
|
|
5376 sprintf(buf + STRLEN(buf), "MCLOSE%d", OP(op) - MCLOSE);
|
|
|
5377 p = NULL;
|
|
|
5378 break;
|
|
|
5379 case BACKREF + 1:
|
|
|
5380 case BACKREF + 2:
|
|
|
5381 case BACKREF + 3:
|
|
|
5382 case BACKREF + 4:
|
|
|
5383 case BACKREF + 5:
|
|
|
5384 case BACKREF + 6:
|
|
|
5385 case BACKREF + 7:
|
|
|
5386 case BACKREF + 8:
|
|
|
5387 case BACKREF + 9:
|
|
|
5388 sprintf(buf + STRLEN(buf), "BACKREF%d", OP(op) - BACKREF);
|
|
|
5389 p = NULL;
|
|
|
5390 break;
|
|
|
5391 case NOPEN:
|
|
|
5392 p = "NOPEN";
|
|
|
5393 break;
|
|
|
5394 case NCLOSE:
|
|
|
5395 p = "NCLOSE";
|
|
|
5396 break;
|
|
|
5397 #ifdef FEAT_SYN_HL
|
|
|
5398 case ZOPEN + 1:
|
|
|
5399 case ZOPEN + 2:
|
|
|
5400 case ZOPEN + 3:
|
|
|
5401 case ZOPEN + 4:
|
|
|
5402 case ZOPEN + 5:
|
|
|
5403 case ZOPEN + 6:
|
|
|
5404 case ZOPEN + 7:
|
|
|
5405 case ZOPEN + 8:
|
|
|
5406 case ZOPEN + 9:
|
|
|
5407 sprintf(buf + STRLEN(buf), "ZOPEN%d", OP(op) - ZOPEN);
|
|
|
5408 p = NULL;
|
|
|
5409 break;
|
|
|
5410 case ZCLOSE + 1:
|
|
|
5411 case ZCLOSE + 2:
|
|
|
5412 case ZCLOSE + 3:
|
|
|
5413 case ZCLOSE + 4:
|
|
|
5414 case ZCLOSE + 5:
|
|
|
5415 case ZCLOSE + 6:
|
|
|
5416 case ZCLOSE + 7:
|
|
|
5417 case ZCLOSE + 8:
|
|
|
5418 case ZCLOSE + 9:
|
|
|
5419 sprintf(buf + STRLEN(buf), "ZCLOSE%d", OP(op) - ZCLOSE);
|
|
|
5420 p = NULL;
|
|
|
5421 break;
|
|
|
5422 case ZREF + 1:
|
|
|
5423 case ZREF + 2:
|
|
|
5424 case ZREF + 3:
|
|
|
5425 case ZREF + 4:
|
|
|
5426 case ZREF + 5:
|
|
|
5427 case ZREF + 6:
|
|
|
5428 case ZREF + 7:
|
|
|
5429 case ZREF + 8:
|
|
|
5430 case ZREF + 9:
|
|
|
5431 sprintf(buf + STRLEN(buf), "ZREF%d", OP(op) - ZREF);
|
|
|
5432 p = NULL;
|
|
|
5433 break;
|
|
|
5434 #endif
|
|
|
5435 case STAR:
|
|
|
5436 p = "STAR";
|
|
|
5437 break;
|
|
|
5438 case PLUS:
|
|
|
5439 p = "PLUS";
|
|
|
5440 break;
|
|
|
5441 case NOMATCH:
|
|
|
5442 p = "NOMATCH";
|
|
|
5443 break;
|
|
|
5444 case MATCH:
|
|
|
5445 p = "MATCH";
|
|
|
5446 break;
|
|
|
5447 case BEHIND:
|
|
|
5448 p = "BEHIND";
|
|
|
5449 break;
|
|
|
5450 case NOBEHIND:
|
|
|
5451 p = "NOBEHIND";
|
|
|
5452 break;
|
|
|
5453 case SUBPAT:
|
|
|
5454 p = "SUBPAT";
|
|
|
5455 break;
|
|
|
5456 case BRACE_LIMITS:
|
|
|
5457 p = "BRACE_LIMITS";
|
|
|
5458 break;
|
|
|
5459 case BRACE_SIMPLE:
|
|
|
5460 p = "BRACE_SIMPLE";
|
|
|
5461 break;
|
|
|
5462 case BRACE_COMPLEX + 0:
|
|
|
5463 case BRACE_COMPLEX + 1:
|
|
|
5464 case BRACE_COMPLEX + 2:
|
|
|
5465 case BRACE_COMPLEX + 3:
|
|
|
5466 case BRACE_COMPLEX + 4:
|
|
|
5467 case BRACE_COMPLEX + 5:
|
|
|
5468 case BRACE_COMPLEX + 6:
|
|
|
5469 case BRACE_COMPLEX + 7:
|
|
|
5470 case BRACE_COMPLEX + 8:
|
|
|
5471 case BRACE_COMPLEX + 9:
|
|
|
5472 sprintf(buf + STRLEN(buf), "BRACE_COMPLEX%d", OP(op) - BRACE_COMPLEX);
|
|
|
5473 p = NULL;
|
|
|
5474 break;
|
|
|
5475 #ifdef FEAT_MBYTE
|
|
|
5476 case MULTIBYTECODE:
|
|
|
5477 p = "MULTIBYTECODE";
|
|
|
5478 break;
|
|
|
5479 #endif
|
|
|
5480 case NEWL:
|
|
|
5481 p = "NEWL";
|
|
|
5482 break;
|
|
|
5483 default:
|
|
|
5484 sprintf(buf + STRLEN(buf), "corrupt %d", OP(op));
|
|
|
5485 p = NULL;
|
|
|
5486 break;
|
|
|
5487 }
|
|
|
5488 if (p != NULL)
|
|
|
5489 (void) strcat(buf, p);
|
|
|
5490 return buf;
|
|
|
5491 }
|
|
|
5492 #endif
|
|
|
5493
|
|
|
5494 #ifdef FEAT_MBYTE
|
|
|
5495 static void mb_decompose __ARGS((int c, int *c1, int *c2, int *c3));
|
|
|
5496
|
|
|
5497 typedef struct
|
|
|
5498 {
|
|
|
5499 int a, b, c;
|
|
|
5500 } decomp_T;
|
|
|
5501
|
|
|
5502
|
|
|
5503 /* 0xfb20 - 0xfb4f */
|
|
|
5504 decomp_T decomp_table[0xfb4f-0xfb20+1] =
|
|
|
5505 {
|
|
|
5506 {0x5e2,0,0}, /* 0xfb20 alt ayin */
|
|
|
5507 {0x5d0,0,0}, /* 0xfb21 alt alef */
|
|
|
5508 {0x5d3,0,0}, /* 0xfb22 alt dalet */
|
|
|
5509 {0x5d4,0,0}, /* 0xfb23 alt he */
|
|
|
5510 {0x5db,0,0}, /* 0xfb24 alt kaf */
|
|
|
5511 {0x5dc,0,0}, /* 0xfb25 alt lamed */
|
|
|
5512 {0x5dd,0,0}, /* 0xfb26 alt mem-sofit */
|
|
|
5513 {0x5e8,0,0}, /* 0xfb27 alt resh */
|
|
|
5514 {0x5ea,0,0}, /* 0xfb28 alt tav */
|
|
|
5515 {'+', 0, 0}, /* 0xfb29 alt plus */
|
|
|
5516 {0x5e9, 0x5c1, 0}, /* 0xfb2a shin+shin-dot */
|
|
|
5517 {0x5e9, 0x5c2, 0}, /* 0xfb2b shin+sin-dot */
|
|
|
5518 {0x5e9, 0x5c1, 0x5bc}, /* 0xfb2c shin+shin-dot+dagesh */
|
|
|
5519 {0x5e9, 0x5c2, 0x5bc}, /* 0xfb2d shin+sin-dot+dagesh */
|
|
|
5520 {0x5d0, 0x5b7, 0}, /* 0xfb2e alef+patah */
|
|
|
5521 {0x5d0, 0x5b8, 0}, /* 0xfb2f alef+qamats */
|
|
|
5522 {0x5d0, 0x5b4, 0}, /* 0xfb30 alef+hiriq */
|
|
|
5523 {0x5d1, 0x5bc, 0}, /* 0xfb31 bet+dagesh */
|
|
|
5524 {0x5d2, 0x5bc, 0}, /* 0xfb32 gimel+dagesh */
|
|
|
5525 {0x5d3, 0x5bc, 0}, /* 0xfb33 dalet+dagesh */
|
|
|
5526 {0x5d4, 0x5bc, 0}, /* 0xfb34 he+dagesh */
|
|
|
5527 {0x5d5, 0x5bc, 0}, /* 0xfb35 vav+dagesh */
|
|
|
5528 {0x5d6, 0x5bc, 0}, /* 0xfb36 zayin+dagesh */
|
|
|
5529 {0xfb37, 0, 0}, /* 0xfb37 -- UNUSED */
|
|
|
5530 {0x5d8, 0x5bc, 0}, /* 0xfb38 tet+dagesh */
|
|
|
5531 {0x5d9, 0x5bc, 0}, /* 0xfb39 yud+dagesh */
|
|
|
5532 {0x5da, 0x5bc, 0}, /* 0xfb3a kaf sofit+dagesh */
|
|
|
5533 {0x5db, 0x5bc, 0}, /* 0xfb3b kaf+dagesh */
|
|
|
5534 {0x5dc, 0x5bc, 0}, /* 0xfb3c lamed+dagesh */
|
|
|
5535 {0xfb3d, 0, 0}, /* 0xfb3d -- UNUSED */
|
|
|
5536 {0x5de, 0x5bc, 0}, /* 0xfb3e mem+dagesh */
|
|
|
5537 {0xfb3f, 0, 0}, /* 0xfb3f -- UNUSED */
|
|
|
5538 {0x5e0, 0x5bc, 0}, /* 0xfb40 nun+dagesh */
|
|
|
5539 {0x5e1, 0x5bc, 0}, /* 0xfb41 samech+dagesh */
|
|
|
5540 {0xfb42, 0, 0}, /* 0xfb42 -- UNUSED */
|
|
|
5541 {0x5e3, 0x5bc, 0}, /* 0xfb43 pe sofit+dagesh */
|
|
|
5542 {0x5e4, 0x5bc,0}, /* 0xfb44 pe+dagesh */
|
|
|
5543 {0xfb45, 0, 0}, /* 0xfb45 -- UNUSED */
|
|
|
5544 {0x5e6, 0x5bc, 0}, /* 0xfb46 tsadi+dagesh */
|
|
|
5545 {0x5e7, 0x5bc, 0}, /* 0xfb47 qof+dagesh */
|
|
|
5546 {0x5e8, 0x5bc, 0}, /* 0xfb48 resh+dagesh */
|
|
|
5547 {0x5e9, 0x5bc, 0}, /* 0xfb49 shin+dagesh */
|
|
|
5548 {0x5ea, 0x5bc, 0}, /* 0xfb4a tav+dagesh */
|
|
|
5549 {0x5d5, 0x5b9, 0}, /* 0xfb4b vav+holam */
|
|
|
5550 {0x5d1, 0x5bf, 0}, /* 0xfb4c bet+rafe */
|
|
|
5551 {0x5db, 0x5bf, 0}, /* 0xfb4d kaf+rafe */
|
|
|
5552 {0x5e4, 0x5bf, 0}, /* 0xfb4e pe+rafe */
|
|
|
5553 {0x5d0, 0x5dc, 0} /* 0xfb4f alef-lamed */
|
|
|
5554 };
|
|
|
5555
|
|
|
5556 static void
|
|
|
5557 mb_decompose(c, c1, c2, c3)
|
|
|
5558 int c, *c1, *c2, *c3;
|
|
|
5559 {
|
|
|
5560 decomp_T d;
|
|
|
5561
|
|
|
5562 if (c >= 0x4b20 && c <= 0xfb4f)
|
|
|
5563 {
|
|
|
5564 d = decomp_table[c - 0xfb20];
|
|
|
5565 *c1 = d.a;
|
|
|
5566 *c2 = d.b;
|
|
|
5567 *c3 = d.c;
|
|
|
5568 }
|
|
|
5569 else
|
|
|
5570 {
|
|
|
5571 *c1 = c;
|
|
|
5572 *c2 = *c3 = 0;
|
|
|
5573 }
|
|
|
5574 }
|
|
|
5575 #endif
|
|
|
5576
|
|
|
5577 /*
|
|
|
5578 * Compare two strings, ignore case if ireg_ic set.
|
|
|
5579 * Return 0 if strings match, non-zero otherwise.
|
|
|
5580 * Correct the length "*n" when composing characters are ignored.
|
|
|
5581 */
|
|
|
5582 static int
|
|
|
5583 cstrncmp(s1, s2, n)
|
|
|
5584 char_u *s1, *s2;
|
|
|
5585 int *n;
|
|
|
5586 {
|
|
|
5587 int result;
|
|
|
5588
|
|
|
5589 if (!ireg_ic)
|
|
|
5590 result = STRNCMP(s1, s2, *n);
|
|
|
5591 else
|
|
|
5592 result = MB_STRNICMP(s1, s2, *n);
|
|
|
5593
|
|
|
5594 #ifdef FEAT_MBYTE
|
|
|
5595 /* if it failed and it's utf8 and we want to combineignore: */
|
|
|
5596 if (result != 0 && enc_utf8 && ireg_icombine)
|
|
|
5597 {
|
|
|
5598 char_u *str1, *str2;
|
|
|
5599 int c1, c2, c11, c12;
|
|
|
5600 int ix;
|
|
|
5601 int junk;
|
|
|
5602
|
|
|
5603 /* we have to handle the strcmp ourselves, since it is necessary to
|
|
|
5604 * deal with the composing characters by ignoring them: */
|
|
|
5605 str1 = s1;
|
|
|
5606 str2 = s2;
|
|
|
5607 c1 = c2 = 0;
|
|
|
5608 for (ix = 0; ix < *n; )
|
|
|
5609 {
|
|
|
5610 c1 = mb_ptr2char_adv(&str1);
|
|
|
5611 c2 = mb_ptr2char_adv(&str2);
|
|
|
5612 ix += utf_char2len(c1);
|
|
|
5613
|
|
|
5614 /* decompose the character if necessary, into 'base' characters
|
|
|
5615 * because I don't care about Arabic, I will hard-code the Hebrew
|
|
|
5616 * which I *do* care about! So sue me... */
|
|
|
5617 if (c1 != c2 && (!ireg_ic || utf_fold(c1) != utf_fold(c2)))
|
|
|
5618 {
|
|
|
5619 /* decomposition necessary? */
|
|
|
5620 mb_decompose(c1, &c11, &junk, &junk);
|
|
|
5621 mb_decompose(c2, &c12, &junk, &junk);
|
|
|
5622 c1 = c11;
|
|
|
5623 c2 = c12;
|
|
|
5624 if (c11 != c12 && (!ireg_ic || utf_fold(c11) != utf_fold(c12)))
|
|
|
5625 break;
|
|
|
5626 }
|
|
|
5627 }
|
|
|
5628 result = c2 - c1;
|
|
|
5629 if (result == 0)
|
|
|
5630 *n = (int)(str2 - s2);
|
|
|
5631 }
|
|
|
5632 #endif
|
|
|
5633
|
|
|
5634 return result;
|
|
|
5635 }
|
|
|
5636
|
|
|
5637 /*
|
|
|
5638 * cstrchr: This function is used a lot for simple searches, keep it fast!
|
|
|
5639 */
|
|
|
5640 static char_u *
|
|
|
5641 cstrchr(s, c)
|
|
|
5642 char_u *s;
|
|
|
5643 int c;
|
|
|
5644 {
|
|
|
5645 char_u *p;
|
|
|
5646 int cc;
|
|
|
5647
|
|
|
5648 if (!ireg_ic
|
|
|
5649 #ifdef FEAT_MBYTE
|
|
|
5650 || (!enc_utf8 && mb_char2len(c) > 1)
|
|
|
5651 #endif
|
|
|
5652 )
|
|
|
5653 return vim_strchr(s, c);
|
|
|
5654
|
|
|
5655 /* tolower() and toupper() can be slow, comparing twice should be a lot
|
|
|
5656 * faster (esp. when using MS Visual C++!).
|
|
|
5657 * For UTF-8 need to use folded case. */
|
|
|
5658 #ifdef FEAT_MBYTE
|
|
|
5659 if (enc_utf8 && c > 0x80)
|
|
|
5660 cc = utf_fold(c);
|
|
|
5661 else
|
|
|
5662 #endif
|
|
|
5663 if (isupper(c))
|
|
|
5664 cc = TOLOWER_LOC(c);
|
|
|
5665 else if (islower(c))
|
|
|
5666 cc = TOUPPER_LOC(c);
|
|
|
5667 else
|
|
|
5668 return vim_strchr(s, c);
|
|
|
5669
|
|
|
5670 #ifdef FEAT_MBYTE
|
|
|
5671 if (has_mbyte)
|
|
|
5672 {
|
|
|
5673 for (p = s; *p != NUL; p += (*mb_ptr2len_check)(p))
|
|
|
5674 {
|
|
|
5675 if (enc_utf8 && c > 0x80)
|
|
|
5676 {
|
|
|
5677 if (utf_fold(utf_ptr2char(p)) == cc)
|
|
|
5678 return p;
|
|
|
5679 }
|
|
|
5680 else if (*p == c || *p == cc)
|
|
|
5681 return p;
|
|
|
5682 }
|
|
|
5683 }
|
|
|
5684 else
|
|
|
5685 #endif
|
|
|
5686 /* Faster version for when there are no multi-byte characters. */
|
|
|
5687 for (p = s; *p != NUL; ++p)
|
|
|
5688 if (*p == c || *p == cc)
|
|
|
5689 return p;
|
|
|
5690
|
|
|
5691 return NULL;
|
|
|
5692 }
|
|
|
5693
|
|
|
5694 /***************************************************************
|
|
|
5695 * regsub stuff *
|
|
|
5696 ***************************************************************/
|
|
|
5697
|
|
|
5698 /* This stuff below really confuses cc on an SGI -- webb */
|
|
|
5699 #ifdef __sgi
|
|
|
5700 # undef __ARGS
|
|
|
5701 # define __ARGS(x) ()
|
|
|
5702 #endif
|
|
|
5703
|
|
|
5704 /*
|
|
|
5705 * We should define ftpr as a pointer to a function returning a pointer to
|
|
|
5706 * a function returning a pointer to a function ...
|
|
|
5707 * This is impossible, so we declare a pointer to a function returning a
|
|
|
5708 * pointer to a function returning void. This should work for all compilers.
|
|
|
5709 */
|
|
|
5710 typedef void (*(*fptr) __ARGS((char_u *, int)))();
|
|
|
5711
|
|
|
5712 static fptr do_upper __ARGS((char_u *, int));
|
|
|
5713 static fptr do_Upper __ARGS((char_u *, int));
|
|
|
5714 static fptr do_lower __ARGS((char_u *, int));
|
|
|
5715 static fptr do_Lower __ARGS((char_u *, int));
|
|
|
5716
|
|
|
5717 static int vim_regsub_both __ARGS((char_u *source, char_u *dest, int copy, int magic, int backslash));
|
|
|
5718
|
|
|
5719 static fptr
|
|
|
5720 do_upper(d, c)
|
|
|
5721 char_u *d;
|
|
|
5722 int c;
|
|
|
5723 {
|
|
|
5724 *d = TOUPPER_LOC(c);
|
|
|
5725
|
|
|
5726 return (fptr)NULL;
|
|
|
5727 }
|
|
|
5728
|
|
|
5729 static fptr
|
|
|
5730 do_Upper(d, c)
|
|
|
5731 char_u *d;
|
|
|
5732 int c;
|
|
|
5733 {
|
|
|
5734 *d = TOUPPER_LOC(c);
|
|
|
5735
|
|
|
5736 return (fptr)do_Upper;
|
|
|
5737 }
|
|
|
5738
|
|
|
5739 static fptr
|
|
|
5740 do_lower(d, c)
|
|
|
5741 char_u *d;
|
|
|
5742 int c;
|
|
|
5743 {
|
|
|
5744 *d = TOLOWER_LOC(c);
|
|
|
5745
|
|
|
5746 return (fptr)NULL;
|
|
|
5747 }
|
|
|
5748
|
|
|
5749 static fptr
|
|
|
5750 do_Lower(d, c)
|
|
|
5751 char_u *d;
|
|
|
5752 int c;
|
|
|
5753 {
|
|
|
5754 *d = TOLOWER_LOC(c);
|
|
|
5755
|
|
|
5756 return (fptr)do_Lower;
|
|
|
5757 }
|
|
|
5758
|
|
|
5759 /*
|
|
|
5760 * regtilde(): Replace tildes in the pattern by the old pattern.
|
|
|
5761 *
|
|
|
5762 * Short explanation of the tilde: It stands for the previous replacement
|
|
|
5763 * pattern. If that previous pattern also contains a ~ we should go back a
|
|
|
5764 * step further... But we insert the previous pattern into the current one
|
|
|
5765 * and remember that.
|
|
|
5766 * This still does not handle the case where "magic" changes. TODO?
|
|
|
5767 *
|
|
|
5768 * The tildes are parsed once before the first call to vim_regsub().
|
|
|
5769 */
|
|
|
5770 char_u *
|
|
|
5771 regtilde(source, magic)
|
|
|
5772 char_u *source;
|
|
|
5773 int magic;
|
|
|
5774 {
|
|
|
5775 char_u *newsub = source;
|
|
|
5776 char_u *tmpsub;
|
|
|
5777 char_u *p;
|
|
|
5778 int len;
|
|
|
5779 int prevlen;
|
|
|
5780
|
|
|
5781 for (p = newsub; *p; ++p)
|
|
|
5782 {
|
|
|
5783 if ((*p == '~' && magic) || (*p == '\\' && *(p + 1) == '~' && !magic))
|
|
|
5784 {
|
|
|
5785 if (reg_prev_sub != NULL)
|
|
|
5786 {
|
|
|
5787 /* length = len(newsub) - 1 + len(prev_sub) + 1 */
|
|
|
5788 prevlen = (int)STRLEN(reg_prev_sub);
|
|
|
5789 tmpsub = alloc((unsigned)(STRLEN(newsub) + prevlen));
|
|
|
5790 if (tmpsub != NULL)
|
|
|
5791 {
|
|
|
5792 /* copy prefix */
|
|
|
5793 len = (int)(p - newsub); /* not including ~ */
|
|
|
5794 mch_memmove(tmpsub, newsub, (size_t)len);
|
|
|
5795 /* interpretate tilde */
|
|
|
5796 mch_memmove(tmpsub + len, reg_prev_sub, (size_t)prevlen);
|
|
|
5797 /* copy postfix */
|
|
|
5798 if (!magic)
|
|
|
5799 ++p; /* back off \ */
|
|
|
5800 STRCPY(tmpsub + len + prevlen, p + 1);
|
|
|
5801
|
|
|
5802 if (newsub != source) /* already allocated newsub */
|
|
|
5803 vim_free(newsub);
|
|
|
5804 newsub = tmpsub;
|
|
|
5805 p = newsub + len + prevlen;
|
|
|
5806 }
|
|
|
5807 }
|
|
|
5808 else if (magic)
|
|
|
5809 STRCPY(p, p + 1); /* remove '~' */
|
|
|
5810 else
|
|
|
5811 STRCPY(p, p + 2); /* remove '\~' */
|
|
|
5812 --p;
|
|
|
5813 }
|
|
|
5814 else
|
|
|
5815 {
|
|
|
5816 if (*p == '\\' && p[1]) /* skip escaped characters */
|
|
|
5817 ++p;
|
|
|
5818 #ifdef FEAT_MBYTE
|
|
|
5819 if (has_mbyte)
|
|
|
5820 p += (*mb_ptr2len_check)(p) - 1;
|
|
|
5821 #endif
|
|
|
5822 }
|
|
|
5823 }
|
|
|
5824
|
|
|
5825 vim_free(reg_prev_sub);
|
|
|
5826 if (newsub != source) /* newsub was allocated, just keep it */
|
|
|
5827 reg_prev_sub = newsub;
|
|
|
5828 else /* no ~ found, need to save newsub */
|
|
|
5829 reg_prev_sub = vim_strsave(newsub);
|
|
|
5830 return newsub;
|
|
|
5831 }
|
|
|
5832
|
|
|
5833 #ifdef FEAT_EVAL
|
|
|
5834 static int can_f_submatch = FALSE; /* TRUE when submatch() can be used */
|
|
|
5835
|
|
|
5836 /* These pointers are used instead of reg_match and reg_mmatch for
|
|
|
5837 * reg_submatch(). Needed for when the substitution string is an expression
|
|
|
5838 * that contains a call to substitute() and submatch(). */
|
|
|
5839 static regmatch_T *submatch_match;
|
|
|
5840 static regmmatch_T *submatch_mmatch;
|
|
|
5841 #endif
|
|
|
5842
|
|
|
5843 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO)
|
|
|
5844 /*
|
|
|
5845 * vim_regsub() - perform substitutions after a vim_regexec() or
|
|
|
5846 * vim_regexec_multi() match.
|
|
|
5847 *
|
|
|
5848 * If "copy" is TRUE really copy into "dest".
|
|
|
5849 * If "copy" is FALSE nothing is copied, this is just to find out the length
|
|
|
5850 * of the result.
|
|
|
5851 *
|
|
|
5852 * If "backslash" is TRUE, a backslash will be removed later, need to double
|
|
|
5853 * them to keep them, and insert a backslash before a CR to avoid it being
|
|
|
5854 * replaced with a line break later.
|
|
|
5855 *
|
|
|
5856 * Note: The matched text must not change between the call of
|
|
|
5857 * vim_regexec()/vim_regexec_multi() and vim_regsub()! It would make the back
|
|
|
5858 * references invalid!
|
|
|
5859 *
|
|
|
5860 * Returns the size of the replacement, including terminating NUL.
|
|
|
5861 */
|
|
|
5862 int
|
|
|
5863 vim_regsub(rmp, source, dest, copy, magic, backslash)
|
|
|
5864 regmatch_T *rmp;
|
|
|
5865 char_u *source;
|
|
|
5866 char_u *dest;
|
|
|
5867 int copy;
|
|
|
5868 int magic;
|
|
|
5869 int backslash;
|
|
|
5870 {
|
|
|
5871 reg_match = rmp;
|
|
|
5872 reg_mmatch = NULL;
|
|
|
5873 reg_maxline = 0;
|
|
|
5874 return vim_regsub_both(source, dest, copy, magic, backslash);
|
|
|
5875 }
|
|
|
5876 #endif
|
|
|
5877
|
|
|
5878 int
|
|
|
5879 vim_regsub_multi(rmp, lnum, source, dest, copy, magic, backslash)
|
|
|
5880 regmmatch_T *rmp;
|
|
|
5881 linenr_T lnum;
|
|
|
5882 char_u *source;
|
|
|
5883 char_u *dest;
|
|
|
5884 int copy;
|
|
|
5885 int magic;
|
|
|
5886 int backslash;
|
|
|
5887 {
|
|
|
5888 reg_match = NULL;
|
|
|
5889 reg_mmatch = rmp;
|
|
|
5890 reg_buf = curbuf; /* always works on the current buffer! */
|
|
|
5891 reg_firstlnum = lnum;
|
|
|
5892 reg_maxline = curbuf->b_ml.ml_line_count - lnum;
|
|
|
5893 return vim_regsub_both(source, dest, copy, magic, backslash);
|
|
|
5894 }
|
|
|
5895
|
|
|
5896 static int
|
|
|
5897 vim_regsub_both(source, dest, copy, magic, backslash)
|
|
|
5898 char_u *source;
|
|
|
5899 char_u *dest;
|
|
|
5900 int copy;
|
|
|
5901 int magic;
|
|
|
5902 int backslash;
|
|
|
5903 {
|
|
|
5904 char_u *src;
|
|
|
5905 char_u *dst;
|
|
|
5906 char_u *s;
|
|
|
5907 int c;
|
|
|
5908 int no = -1;
|
|
|
5909 fptr func = (fptr)NULL;
|
|
|
5910 linenr_T clnum = 0; /* init for GCC */
|
|
|
5911 int len = 0; /* init for GCC */
|
|
|
5912 #ifdef FEAT_EVAL
|
|
|
5913 static char_u *eval_result = NULL;
|
|
|
5914 #endif
|
|
|
5915 #ifdef FEAT_MBYTE
|
|
|
5916 int l;
|
|
|
5917 #endif
|
|
|
5918
|
|
|
5919
|
|
|
5920 /* Be paranoid... */
|
|
|
5921 if (source == NULL || dest == NULL)
|
|
|
5922 {
|
|
|
5923 EMSG(_(e_null));
|
|
|
5924 return 0;
|
|
|
5925 }
|
|
|
5926 if (prog_magic_wrong())
|
|
|
5927 return 0;
|
|
|
5928 src = source;
|
|
|
5929 dst = dest;
|
|
|
5930
|
|
|
5931 /*
|
|
|
5932 * When the substitute part starts with "\=" evaluate it as an expression.
|
|
|
5933 */
|
|
|
5934 if (source[0] == '\\' && source[1] == '='
|
|
|
5935 #ifdef FEAT_EVAL
|
|
|
5936 && !can_f_submatch /* can't do this recursively */
|
|
|
5937 #endif
|
|
|
5938 )
|
|
|
5939 {
|
|
|
5940 #ifdef FEAT_EVAL
|
|
|
5941 /* To make sure that the length doesn't change between checking the
|
|
|
5942 * length and copying the string, and to speed up things, the
|
|
|
5943 * resulting string is saved from the call with "copy" == FALSE to the
|
|
|
5944 * call with "copy" == TRUE. */
|
|
|
5945 if (copy)
|
|
|
5946 {
|
|
|
5947 if (eval_result != NULL)
|
|
|
5948 {
|
|
|
5949 STRCPY(dest, eval_result);
|
|
|
5950 dst += STRLEN(eval_result);
|
|
|
5951 vim_free(eval_result);
|
|
|
5952 eval_result = NULL;
|
|
|
5953 }
|
|
|
5954 }
|
|
|
5955 else
|
|
|
5956 {
|
|
|
5957 linenr_T save_reg_maxline;
|
|
|
5958 win_T *save_reg_win;
|
|
|
5959 int save_ireg_ic;
|
|
|
5960
|
|
|
5961 vim_free(eval_result);
|
|
|
5962
|
|
|
5963 /* The expression may contain substitute(), which calls us
|
|
|
5964 * recursively. Make sure submatch() gets the text from the first
|
|
|
5965 * level. Don't need to save "reg_buf", because
|
|
|
5966 * vim_regexec_multi() can't be called recursively. */
|
|
|
5967 submatch_match = reg_match;
|
|
|
5968 submatch_mmatch = reg_mmatch;
|
|
|
5969 save_reg_maxline = reg_maxline;
|
|
|
5970 save_reg_win = reg_win;
|
|
|
5971 save_ireg_ic = ireg_ic;
|
|
|
5972 can_f_submatch = TRUE;
|
|
|
5973
|
|
|
5974 eval_result = eval_to_string(source + 2, NULL);
|
|
|
5975 if (eval_result != NULL)
|
|
|
5976 {
|
|
|
5977 for (s = eval_result; *s != NUL; ++s)
|
|
|
5978 {
|
|
|
5979 /* Change NL to CR, so that it becomes a line break.
|
|
|
5980 * Skip over a backslashed character. */
|
|
|
5981 if (*s == NL)
|
|
|
5982 *s = CAR;
|
|
|
5983 else if (*s == '\\' && s[1] != NUL)
|
|
|
5984 ++s;
|
|
|
5985 #ifdef FEAT_MBYTE
|
|
|
5986 if (has_mbyte)
|
|
|
5987 s += (*mb_ptr2len_check)(s) - 1;
|
|
|
5988 #endif
|
|
|
5989 }
|
|
|
5990
|
|
|
5991 dst += STRLEN(eval_result);
|
|
|
5992 }
|
|
|
5993
|
|
|
5994 reg_match = submatch_match;
|
|
|
5995 reg_mmatch = submatch_mmatch;
|
|
|
5996 reg_maxline = save_reg_maxline;
|
|
|
5997 reg_win = save_reg_win;
|
|
|
5998 ireg_ic = save_ireg_ic;
|
|
|
5999 can_f_submatch = FALSE;
|
|
|
6000 }
|
|
|
6001 #endif
|
|
|
6002 }
|
|
|
6003 else
|
|
|
6004 while ((c = *src++) != NUL)
|
|
|
6005 {
|
|
|
6006 if (c == '&' && magic)
|
|
|
6007 no = 0;
|
|
|
6008 else if (c == '\\' && *src != NUL)
|
|
|
6009 {
|
|
|
6010 if (*src == '&' && !magic)
|
|
|
6011 {
|
|
|
6012 ++src;
|
|
|
6013 no = 0;
|
|
|
6014 }
|
|
|
6015 else if ('0' <= *src && *src <= '9')
|
|
|
6016 {
|
|
|
6017 no = *src++ - '0';
|
|
|
6018 }
|
|
|
6019 else if (vim_strchr((char_u *)"uUlLeE", *src))
|
|
|
6020 {
|
|
|
6021 switch (*src++)
|
|
|
6022 {
|
|
|
6023 case 'u': func = (fptr)do_upper;
|
|
|
6024 continue;
|
|
|
6025 case 'U': func = (fptr)do_Upper;
|
|
|
6026 continue;
|
|
|
6027 case 'l': func = (fptr)do_lower;
|
|
|
6028 continue;
|
|
|
6029 case 'L': func = (fptr)do_Lower;
|
|
|
6030 continue;
|
|
|
6031 case 'e':
|
|
|
6032 case 'E': func = (fptr)NULL;
|
|
|
6033 continue;
|
|
|
6034 }
|
|
|
6035 }
|
|
|
6036 }
|
|
|
6037 if (no < 0) /* Ordinary character. */
|
|
|
6038 {
|
|
|
6039 if (c == '\\' && *src != NUL)
|
|
|
6040 {
|
|
|
6041 /* Check for abbreviations -- webb */
|
|
|
6042 switch (*src)
|
|
|
6043 {
|
|
|
6044 case 'r': c = CAR; ++src; break;
|
|
|
6045 case 'n': c = NL; ++src; break;
|
|
|
6046 case 't': c = TAB; ++src; break;
|
|
|
6047 /* Oh no! \e already has meaning in subst pat :-( */
|
|
|
6048 /* case 'e': c = ESC; ++src; break; */
|
|
|
6049 case 'b': c = Ctrl_H; ++src; break;
|
|
|
6050
|
|
|
6051 /* If "backslash" is TRUE the backslash will be removed
|
|
|
6052 * later. Used to insert a literal CR. */
|
|
|
6053 default: if (backslash)
|
|
|
6054 {
|
|
|
6055 if (copy)
|
|
|
6056 *dst = '\\';
|
|
|
6057 ++dst;
|
|
|
6058 }
|
|
|
6059 c = *src++;
|
|
|
6060 }
|
|
|
6061 }
|
|
|
6062
|
|
|
6063 /* Write to buffer, if copy is set. */
|
|
|
6064 #ifdef FEAT_MBYTE
|
|
|
6065 if (has_mbyte && (l = (*mb_ptr2len_check)(src - 1)) > 1)
|
|
|
6066 {
|
|
|
6067 /* TODO: should use "func" here. */
|
|
|
6068 if (copy)
|
|
|
6069 mch_memmove(dst, src - 1, l);
|
|
|
6070 dst += l - 1;
|
|
|
6071 src += l - 1;
|
|
|
6072 }
|
|
|
6073 else
|
|
|
6074 {
|
|
|
6075 #endif
|
|
|
6076 if (copy)
|
|
|
6077 {
|
|
|
6078 if (func == (fptr)NULL) /* just copy */
|
|
|
6079 *dst = c;
|
|
|
6080 else /* change case */
|
|
|
6081 func = (fptr)(func(dst, c));
|
|
|
6082 /* Turbo C complains without the typecast */
|
|
|
6083 }
|
|
|
6084 #ifdef FEAT_MBYTE
|
|
|
6085 }
|
|
|
6086 #endif
|
|
|
6087 dst++;
|
|
|
6088 }
|
|
|
6089 else
|
|
|
6090 {
|
|
|
6091 if (REG_MULTI)
|
|
|
6092 {
|
|
|
6093 clnum = reg_mmatch->startpos[no].lnum;
|
|
|
6094 if (clnum < 0 || reg_mmatch->endpos[no].lnum < 0)
|
|
|
6095 s = NULL;
|
|
|
6096 else
|
|
|
6097 {
|
|
|
6098 s = reg_getline(clnum) + reg_mmatch->startpos[no].col;
|
|
|
6099 if (reg_mmatch->endpos[no].lnum == clnum)
|
|
|
6100 len = reg_mmatch->endpos[no].col
|
|
|
6101 - reg_mmatch->startpos[no].col;
|
|
|
6102 else
|
|
|
6103 len = (int)STRLEN(s);
|
|
|
6104 }
|
|
|
6105 }
|
|
|
6106 else
|
|
|
6107 {
|
|
|
6108 s = reg_match->startp[no];
|
|
|
6109 if (reg_match->endp[no] == NULL)
|
|
|
6110 s = NULL;
|
|
|
6111 else
|
|
|
6112 len = (int)(reg_match->endp[no] - s);
|
|
|
6113 }
|
|
|
6114 if (s != NULL)
|
|
|
6115 {
|
|
|
6116 for (;;)
|
|
|
6117 {
|
|
|
6118 if (len == 0)
|
|
|
6119 {
|
|
|
6120 if (REG_MULTI)
|
|
|
6121 {
|
|
|
6122 if (reg_mmatch->endpos[no].lnum == clnum)
|
|
|
6123 break;
|
|
|
6124 if (copy)
|
|
|
6125 *dst = CAR;
|
|
|
6126 ++dst;
|
|
|
6127 s = reg_getline(++clnum);
|
|
|
6128 if (reg_mmatch->endpos[no].lnum == clnum)
|
|
|
6129 len = reg_mmatch->endpos[no].col;
|
|
|
6130 else
|
|
|
6131 len = (int)STRLEN(s);
|
|
|
6132 }
|
|
|
6133 else
|
|
|
6134 break;
|
|
|
6135 }
|
|
|
6136 else if (*s == NUL) /* we hit NUL. */
|
|
|
6137 {
|
|
|
6138 if (copy)
|
|
|
6139 EMSG(_(e_re_damg));
|
|
|
6140 goto exit;
|
|
|
6141 }
|
|
|
6142 else
|
|
|
6143 {
|
|
|
6144 if (backslash && (*s == CAR || *s == '\\'))
|
|
|
6145 {
|
|
|
6146 /*
|
|
|
6147 * Insert a backslash in front of a CR, otherwise
|
|
|
6148 * it will be replaced by a line break.
|
|
|
6149 * Number of backslashes will be halved later,
|
|
|
6150 * double them here.
|
|
|
6151 */
|
|
|
6152 if (copy)
|
|
|
6153 {
|
|
|
6154 dst[0] = '\\';
|
|
|
6155 dst[1] = *s;
|
|
|
6156 }
|
|
|
6157 dst += 2;
|
|
|
6158 }
|
|
|
6159 #ifdef FEAT_MBYTE
|
|
|
6160 else if (has_mbyte && (l = (*mb_ptr2len_check)(s)) > 1)
|
|
|
6161 {
|
|
|
6162 /* TODO: should use "func" here. */
|
|
|
6163 if (copy)
|
|
|
6164 mch_memmove(dst, s, l);
|
|
|
6165 dst += l;
|
|
|
6166 s += l - 1;
|
|
|
6167 len -= l - 1;
|
|
|
6168 }
|
|
|
6169 #endif
|
|
|
6170 else
|
|
|
6171 {
|
|
|
6172 if (copy)
|
|
|
6173 {
|
|
|
6174 if (func == (fptr)NULL) /* just copy */
|
|
|
6175 *dst = *s;
|
|
|
6176 else /* change case */
|
|
|
6177 func = (fptr)(func(dst, *s));
|
|
|
6178 /* Turbo C complains without the typecast */
|
|
|
6179 }
|
|
|
6180 ++dst;
|
|
|
6181 }
|
|
|
6182 ++s;
|
|
|
6183 --len;
|
|
|
6184 }
|
|
|
6185 }
|
|
|
6186 }
|
|
|
6187 no = -1;
|
|
|
6188 }
|
|
|
6189 }
|
|
|
6190 if (copy)
|
|
|
6191 *dst = NUL;
|
|
|
6192
|
|
|
6193 exit:
|
|
|
6194 return (int)((dst - dest) + 1);
|
|
|
6195 }
|
|
|
6196
|
|
|
6197 #ifdef FEAT_EVAL
|
|
|
6198 /*
|
|
|
6199 * Used for the submatch() function: get the string from tne n'th submatch in
|
|
|
6200 * allocated memory.
|
|
|
6201 * Returns NULL when not in a ":s" command and for a non-existing submatch.
|
|
|
6202 */
|
|
|
6203 char_u *
|
|
|
6204 reg_submatch(no)
|
|
|
6205 int no;
|
|
|
6206 {
|
|
|
6207 char_u *retval = NULL;
|
|
|
6208 char_u *s;
|
|
|
6209 int len;
|
|
|
6210 int round;
|
|
|
6211 linenr_T lnum;
|
|
|
6212
|
|
|
6213 if (!can_f_submatch)
|
|
|
6214 return NULL;
|
|
|
6215
|
|
|
6216 if (submatch_match == NULL)
|
|
|
6217 {
|
|
|
6218 /*
|
|
|
6219 * First round: compute the length and allocate memory.
|
|
|
6220 * Second round: copy the text.
|
|
|
6221 */
|
|
|
6222 for (round = 1; round <= 2; ++round)
|
|
|
6223 {
|
|
|
6224 lnum = submatch_mmatch->startpos[no].lnum;
|
|
|
6225 if (lnum < 0 || submatch_mmatch->endpos[no].lnum < 0)
|
|
|
6226 return NULL;
|
|
|
6227
|
|
|
6228 s = reg_getline(lnum) + submatch_mmatch->startpos[no].col;
|
|
|
6229 if (s == NULL) /* anti-crash check, cannot happen? */
|
|
|
6230 break;
|
|
|
6231 if (submatch_mmatch->endpos[no].lnum == lnum)
|
|
|
6232 {
|
|
|
6233 /* Within one line: take form start to end col. */
|
|
|
6234 len = submatch_mmatch->endpos[no].col
|
|
|
6235 - submatch_mmatch->startpos[no].col;
|
|
|
6236 if (round == 2)
|
|
|
6237 {
|
|
|
6238 STRNCPY(retval, s, len);
|
|
|
6239 retval[len] = NUL;
|
|
|
6240 }
|
|
|
6241 ++len;
|
|
|
6242 }
|
|
|
6243 else
|
|
|
6244 {
|
|
|
6245 /* Multiple lines: take start line from start col, middle
|
|
|
6246 * lines completely and end line up to end col. */
|
|
|
6247 len = (int)STRLEN(s);
|
|
|
6248 if (round == 2)
|
|
|
6249 {
|
|
|
6250 STRCPY(retval, s);
|
|
|
6251 retval[len] = '\n';
|
|
|
6252 }
|
|
|
6253 ++len;
|
|
|
6254 ++lnum;
|
|
|
6255 while (lnum < submatch_mmatch->endpos[no].lnum)
|
|
|
6256 {
|
|
|
6257 s = reg_getline(lnum++);
|
|
|
6258 if (round == 2)
|
|
|
6259 STRCPY(retval + len, s);
|
|
|
6260 len += (int)STRLEN(s);
|
|
|
6261 if (round == 2)
|
|
|
6262 retval[len] = '\n';
|
|
|
6263 ++len;
|
|
|
6264 }
|
|
|
6265 if (round == 2)
|
|
|
6266 STRNCPY(retval + len, reg_getline(lnum),
|
|
|
6267 submatch_mmatch->endpos[no].col);
|
|
|
6268 len += submatch_mmatch->endpos[no].col;
|
|
|
6269 if (round == 2)
|
|
|
6270 retval[len] = NUL;
|
|
|
6271 ++len;
|
|
|
6272 }
|
|
|
6273
|
|
|
6274 if (round == 1)
|
|
|
6275 {
|
|
|
6276 retval = lalloc((long_u)len, TRUE);
|
|
|
6277 if (s == NULL)
|
|
|
6278 return NULL;
|
|
|
6279 }
|
|
|
6280 }
|
|
|
6281 }
|
|
|
6282 else
|
|
|
6283 {
|
|
|
6284 if (submatch_match->endp[no] == NULL)
|
|
|
6285 retval = NULL;
|
|
|
6286 else
|
|
|
6287 {
|
|
|
6288 s = submatch_match->startp[no];
|
|
|
6289 retval = vim_strnsave(s, (int)(submatch_match->endp[no] - s));
|
|
|
6290 }
|
|
|
6291 }
|
|
|
6292
|
|
|
6293 return retval;
|
|
|
6294 }
|
|
|
6295 #endif
|
