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