Mercurial > vim
annotate src/regexp.c @ 7929:e44240b5a6da v7.4.1260
commit https://github.com/vim/vim/commit/a8343c1808f2f268282f3030ce4adaf22e8ade54
Author: Bram Moolenaar <Bram@vim.org>
Date: Thu Feb 4 22:09:48 2016 +0100
patch 7.4.1260
Problem: The channel feature doesn't work on Win32 GUI.
Solution: Use WSAGetLastError(). (Ken Takata)
| author | Christian Brabandt <cb@256bit.org> |
|---|---|
| date | Thu, 04 Feb 2016 22:15:04 +0100 |
| parents | c079097365f3 |
| children | b6b4f354df23 |
| rev | line source |
|---|---|
| 7 | 1 /* vi:set ts=8 sts=4 sw=4: |
| 2 * | |
| 3 * Handling of regular expressions: vim_regcomp(), vim_regexec(), vim_regsub() | |
| 4 * | |
| 5 * NOTICE: | |
| 6 * | |
| 7 * This is NOT the original regular expression code as written by Henry | |
| 8 * Spencer. This code has been modified specifically for use with the VIM | |
| 9 * editor, and should not be used separately from Vim. If you want a good | |
| 10 * regular expression library, get the original code. The copyright notice | |
| 11 * that follows is from the original. | |
| 12 * | |
| 13 * END NOTICE | |
| 14 * | |
| 15 * Copyright (c) 1986 by University of Toronto. | |
| 16 * Written by Henry Spencer. Not derived from licensed software. | |
| 17 * | |
| 18 * Permission is granted to anyone to use this software for any | |
| 19 * purpose on any computer system, and to redistribute it freely, | |
| 20 * subject to the following restrictions: | |
| 21 * | |
| 22 * 1. The author is not responsible for the consequences of use of | |
| 23 * this software, no matter how awful, even if they arise | |
| 24 * from defects in it. | |
| 25 * | |
| 26 * 2. The origin of this software must not be misrepresented, either | |
| 27 * by explicit claim or by omission. | |
| 28 * | |
| 29 * 3. Altered versions must be plainly marked as such, and must not | |
| 30 * be misrepresented as being the original software. | |
| 31 * | |
| 32 * Beware that some of this code is subtly aware of the way operator | |
| 33 * precedence is structured in regular expressions. Serious changes in | |
| 34 * regular-expression syntax might require a total rethink. | |
| 35 * | |
| 24 | 36 * Changes have been made by Tony Andrews, Olaf 'Rhialto' Seibert, Robert |
| 37 * Webb, Ciaran McCreesh and Bram Moolenaar. | |
| 7 | 38 * Named character class support added by Walter Briscoe (1998 Jul 01) |
| 39 */ | |
| 40 | |
| 4444 | 41 /* Uncomment the first if you do not want to see debugging logs or files |
| 42 * related to regular expressions, even when compiling with -DDEBUG. | |
| 43 * Uncomment the second to get the regexp debugging. */ | |
| 44 /* #undef DEBUG */ | |
| 45 /* #define DEBUG */ | |
| 46 | |
| 7 | 47 #include "vim.h" |
| 48 | |
| 4444 | 49 #ifdef DEBUG |
| 50 /* show/save debugging data when BT engine is used */ | |
| 51 # define BT_REGEXP_DUMP | |
| 52 /* save the debugging data to a file instead of displaying it */ | |
| 53 # define BT_REGEXP_LOG | |
| 4460 | 54 # define BT_REGEXP_DEBUG_LOG |
| 55 # define BT_REGEXP_DEBUG_LOG_NAME "bt_regexp_debug.log" | |
| 4444 | 56 #endif |
| 7 | 57 |
| 58 /* | |
| 59 * The "internal use only" fields in regexp.h are present to pass info from | |
| 60 * compile to execute that permits the execute phase to run lots faster on | |
| 61 * simple cases. They are: | |
| 62 * | |
| 63 * regstart char that must begin a match; NUL if none obvious; Can be a | |
| 64 * multi-byte character. | |
| 65 * reganch is the match anchored (at beginning-of-line only)? | |
| 66 * regmust string (pointer into program) that match must include, or NULL | |
| 67 * regmlen length of regmust string | |
| 68 * regflags RF_ values or'ed together | |
| 69 * | |
| 70 * Regstart and reganch permit very fast decisions on suitable starting points | |
| 71 * for a match, cutting down the work a lot. Regmust permits fast rejection | |
| 72 * of lines that cannot possibly match. The regmust tests are costly enough | |
| 73 * that vim_regcomp() supplies a regmust only if the r.e. contains something | |
| 74 * potentially expensive (at present, the only such thing detected is * or + | |
| 75 * at the start of the r.e., which can involve a lot of backup). Regmlen is | |
| 76 * supplied because the test in vim_regexec() needs it and vim_regcomp() is | |
| 77 * computing it anyway. | |
| 78 */ | |
| 79 | |
| 80 /* | |
| 81 * Structure for regexp "program". This is essentially a linear encoding | |
| 82 * of a nondeterministic finite-state machine (aka syntax charts or | |
| 83 * "railroad normal form" in parsing technology). Each node is an opcode | |
| 84 * plus a "next" pointer, possibly plus an operand. "Next" pointers of | |
| 85 * all nodes except BRANCH and BRACES_COMPLEX implement concatenation; a "next" | |
| 86 * pointer with a BRANCH on both ends of it is connecting two alternatives. | |
| 87 * (Here we have one of the subtle syntax dependencies: an individual BRANCH | |
| 88 * (as opposed to a collection of them) is never concatenated with anything | |
| 89 * because of operator precedence). The "next" pointer of a BRACES_COMPLEX | |
| 167 | 90 * node points to the node after the stuff to be repeated. |
| 91 * The operand of some types of node is a literal string; for others, it is a | |
| 92 * node leading into a sub-FSM. In particular, the operand of a BRANCH node | |
| 93 * is the first node of the branch. | |
| 94 * (NB this is *not* a tree structure: the tail of the branch connects to the | |
| 95 * thing following the set of BRANCHes.) | |
| 7 | 96 * |
| 97 * pattern is coded like: | |
| 98 * | |
| 99 * +-----------------+ | |
| 100 * | V | |
| 101 * <aa>\|<bb> BRANCH <aa> BRANCH <bb> --> END | |
| 102 * | ^ | ^ | |
| 103 * +------+ +----------+ | |
| 104 * | |
| 105 * | |
| 106 * +------------------+ | |
| 107 * V | | |
| 108 * <aa>* BRANCH BRANCH <aa> --> BACK BRANCH --> NOTHING --> END | |
| 109 * | | ^ ^ | |
| 110 * | +---------------+ | | |
| 111 * +---------------------------------------------+ | |
| 112 * | |
| 113 * | |
| 167 | 114 * +----------------------+ |
| 115 * V | | |
| 233 | 116 * <aa>\+ BRANCH <aa> --> BRANCH --> BACK BRANCH --> NOTHING --> END |
| 856 | 117 * | | ^ ^ |
| 118 * | +-----------+ | | |
| 179 | 119 * +--------------------------------------------------+ |
| 167 | 120 * |
| 121 * | |
| 7 | 122 * +-------------------------+ |
| 123 * V | | |
| 124 * <aa>\{} BRANCH BRACE_LIMITS --> BRACE_COMPLEX <aa> --> BACK END | |
| 125 * | | ^ | |
| 126 * | +----------------+ | |
| 127 * +-----------------------------------------------+ | |
| 128 * | |
| 129 * | |
| 130 * <aa>\@!<bb> BRANCH NOMATCH <aa> --> END <bb> --> END | |
| 131 * | | ^ ^ | |
| 132 * | +----------------+ | | |
| 133 * +--------------------------------+ | |
| 134 * | |
| 135 * +---------+ | |
| 136 * | V | |
| 137 * \z[abc] BRANCH BRANCH a BRANCH b BRANCH c BRANCH NOTHING --> END | |
| 138 * | | | | ^ ^ | |
| 139 * | | | +-----+ | | |
| 140 * | | +----------------+ | | |
| 141 * | +---------------------------+ | | |
| 142 * +------------------------------------------------------+ | |
| 143 * | |
| 1209 | 144 * They all start with a BRANCH for "\|" alternatives, even when there is only |
| 7 | 145 * one alternative. |
| 146 */ | |
| 147 | |
| 148 /* | |
| 149 * The opcodes are: | |
| 150 */ | |
| 151 | |
| 152 /* definition number opnd? meaning */ | |
| 153 #define END 0 /* End of program or NOMATCH operand. */ | |
| 154 #define BOL 1 /* Match "" at beginning of line. */ | |
| 155 #define EOL 2 /* Match "" at end of line. */ | |
| 156 #define BRANCH 3 /* node Match this alternative, or the | |
| 157 * next... */ | |
| 158 #define BACK 4 /* Match "", "next" ptr points backward. */ | |
| 159 #define EXACTLY 5 /* str Match this string. */ | |
| 160 #define NOTHING 6 /* Match empty string. */ | |
| 161 #define STAR 7 /* node Match this (simple) thing 0 or more | |
| 162 * times. */ | |
| 163 #define PLUS 8 /* node Match this (simple) thing 1 or more | |
| 164 * times. */ | |
| 165 #define MATCH 9 /* node match the operand zero-width */ | |
| 166 #define NOMATCH 10 /* node check for no match with operand */ | |
| 167 #define BEHIND 11 /* node look behind for a match with operand */ | |
| 168 #define NOBEHIND 12 /* node look behind for no match with operand */ | |
| 169 #define SUBPAT 13 /* node match the operand here */ | |
| 170 #define BRACE_SIMPLE 14 /* node Match this (simple) thing between m and | |
| 171 * n times (\{m,n\}). */ | |
| 172 #define BOW 15 /* Match "" after [^a-zA-Z0-9_] */ | |
| 173 #define EOW 16 /* Match "" at [^a-zA-Z0-9_] */ | |
| 174 #define BRACE_LIMITS 17 /* nr nr define the min & max for BRACE_SIMPLE | |
| 175 * and BRACE_COMPLEX. */ | |
| 176 #define NEWL 18 /* Match line-break */ | |
| 177 #define BHPOS 19 /* End position for BEHIND or NOBEHIND */ | |
| 178 | |
| 179 | |
| 180 /* character classes: 20-48 normal, 50-78 include a line-break */ | |
| 181 #define ADD_NL 30 | |
| 182 #define FIRST_NL ANY + ADD_NL | |
| 183 #define ANY 20 /* Match any one character. */ | |
| 184 #define ANYOF 21 /* str Match any character in this string. */ | |
| 185 #define ANYBUT 22 /* str Match any character not in this | |
| 186 * string. */ | |
| 187 #define IDENT 23 /* Match identifier char */ | |
| 188 #define SIDENT 24 /* Match identifier char but no digit */ | |
| 189 #define KWORD 25 /* Match keyword char */ | |
| 190 #define SKWORD 26 /* Match word char but no digit */ | |
| 191 #define FNAME 27 /* Match file name char */ | |
| 192 #define SFNAME 28 /* Match file name char but no digit */ | |
| 193 #define PRINT 29 /* Match printable char */ | |
| 194 #define SPRINT 30 /* Match printable char but no digit */ | |
| 195 #define WHITE 31 /* Match whitespace char */ | |
| 196 #define NWHITE 32 /* Match non-whitespace char */ | |
| 197 #define DIGIT 33 /* Match digit char */ | |
| 198 #define NDIGIT 34 /* Match non-digit char */ | |
| 199 #define HEX 35 /* Match hex char */ | |
| 200 #define NHEX 36 /* Match non-hex char */ | |
| 201 #define OCTAL 37 /* Match octal char */ | |
| 202 #define NOCTAL 38 /* Match non-octal char */ | |
| 203 #define WORD 39 /* Match word char */ | |
| 204 #define NWORD 40 /* Match non-word char */ | |
| 205 #define HEAD 41 /* Match head char */ | |
| 206 #define NHEAD 42 /* Match non-head char */ | |
| 207 #define ALPHA 43 /* Match alpha char */ | |
| 208 #define NALPHA 44 /* Match non-alpha char */ | |
| 209 #define LOWER 45 /* Match lowercase char */ | |
| 210 #define NLOWER 46 /* Match non-lowercase char */ | |
| 211 #define UPPER 47 /* Match uppercase char */ | |
| 212 #define NUPPER 48 /* Match non-uppercase char */ | |
| 213 #define LAST_NL NUPPER + ADD_NL | |
| 214 #define WITH_NL(op) ((op) >= FIRST_NL && (op) <= LAST_NL) | |
| 215 | |
| 216 #define MOPEN 80 /* -89 Mark this point in input as start of | |
| 217 * \( subexpr. MOPEN + 0 marks start of | |
| 218 * match. */ | |
| 219 #define MCLOSE 90 /* -99 Analogous to MOPEN. MCLOSE + 0 marks | |
| 220 * end of match. */ | |
| 221 #define BACKREF 100 /* -109 node Match same string again \1-\9 */ | |
| 222 | |
| 223 #ifdef FEAT_SYN_HL | |
| 224 # define ZOPEN 110 /* -119 Mark this point in input as start of | |
| 225 * \z( subexpr. */ | |
| 226 # define ZCLOSE 120 /* -129 Analogous to ZOPEN. */ | |
| 227 # define ZREF 130 /* -139 node Match external submatch \z1-\z9 */ | |
| 228 #endif | |
| 229 | |
| 230 #define BRACE_COMPLEX 140 /* -149 node Match nodes between m & n times */ | |
| 231 | |
| 232 #define NOPEN 150 /* Mark this point in input as start of | |
| 233 \%( subexpr. */ | |
| 234 #define NCLOSE 151 /* Analogous to NOPEN. */ | |
| 235 | |
| 236 #define MULTIBYTECODE 200 /* mbc Match one multi-byte character */ | |
| 237 #define RE_BOF 201 /* Match "" at beginning of file. */ | |
| 238 #define RE_EOF 202 /* Match "" at end of file. */ | |
| 239 #define CURSOR 203 /* Match location of cursor. */ | |
| 240 | |
| 241 #define RE_LNUM 204 /* nr cmp Match line number */ | |
| 242 #define RE_COL 205 /* nr cmp Match column number */ | |
| 243 #define RE_VCOL 206 /* nr cmp Match virtual column number */ | |
| 244 | |
| 639 | 245 #define RE_MARK 207 /* mark cmp Match mark position */ |
| 246 #define RE_VISUAL 208 /* Match Visual area */ | |
| 5901 | 247 #define RE_COMPOSING 209 /* any composing characters */ |
| 639 | 248 |
| 7 | 249 /* |
| 250 * Magic characters have a special meaning, they don't match literally. | |
| 251 * Magic characters are negative. This separates them from literal characters | |
| 252 * (possibly multi-byte). Only ASCII characters can be Magic. | |
| 253 */ | |
| 254 #define Magic(x) ((int)(x) - 256) | |
| 255 #define un_Magic(x) ((x) + 256) | |
| 256 #define is_Magic(x) ((x) < 0) | |
| 257 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
258 static int no_Magic(int x); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
259 static int toggle_Magic(int x); |
| 7 | 260 |
| 261 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
262 no_Magic(int x) |
| 7 | 263 { |
| 264 if (is_Magic(x)) | |
| 265 return un_Magic(x); | |
| 266 return x; | |
| 267 } | |
| 268 | |
| 269 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
270 toggle_Magic(int x) |
| 7 | 271 { |
| 272 if (is_Magic(x)) | |
| 273 return un_Magic(x); | |
| 274 return Magic(x); | |
| 275 } | |
| 276 | |
| 277 /* | |
| 278 * The first byte of the regexp internal "program" is actually this magic | |
| 279 * number; the start node begins in the second byte. It's used to catch the | |
| 280 * most severe mutilation of the program by the caller. | |
| 281 */ | |
| 282 | |
| 283 #define REGMAGIC 0234 | |
| 284 | |
| 285 /* | |
| 286 * Opcode notes: | |
| 287 * | |
| 288 * BRANCH The set of branches constituting a single choice are hooked | |
| 289 * together with their "next" pointers, since precedence prevents | |
| 290 * anything being concatenated to any individual branch. The | |
| 291 * "next" pointer of the last BRANCH in a choice points to the | |
| 292 * thing following the whole choice. This is also where the | |
| 293 * final "next" pointer of each individual branch points; each | |
| 294 * branch starts with the operand node of a BRANCH node. | |
| 295 * | |
| 296 * BACK Normal "next" pointers all implicitly point forward; BACK | |
| 297 * exists to make loop structures possible. | |
| 298 * | |
| 299 * STAR,PLUS '=', and complex '*' and '+', are implemented as circular | |
| 300 * BRANCH structures using BACK. Simple cases (one character | |
| 301 * per match) are implemented with STAR and PLUS for speed | |
| 302 * and to minimize recursive plunges. | |
| 303 * | |
| 304 * BRACE_LIMITS This is always followed by a BRACE_SIMPLE or BRACE_COMPLEX | |
| 305 * node, and defines the min and max limits to be used for that | |
| 306 * node. | |
| 307 * | |
| 308 * MOPEN,MCLOSE ...are numbered at compile time. | |
| 309 * ZOPEN,ZCLOSE ...ditto | |
| 310 */ | |
| 311 | |
| 312 /* | |
| 313 * A node is one char of opcode followed by two chars of "next" pointer. | |
| 314 * "Next" pointers are stored as two 8-bit bytes, high order first. The | |
| 315 * value is a positive offset from the opcode of the node containing it. | |
| 316 * An operand, if any, simply follows the node. (Note that much of the | |
| 317 * code generation knows about this implicit relationship.) | |
| 318 * | |
| 319 * Using two bytes for the "next" pointer is vast overkill for most things, | |
| 320 * but allows patterns to get big without disasters. | |
| 321 */ | |
| 322 #define OP(p) ((int)*(p)) | |
| 323 #define NEXT(p) (((*((p) + 1) & 0377) << 8) + (*((p) + 2) & 0377)) | |
| 324 #define OPERAND(p) ((p) + 3) | |
| 325 /* Obtain an operand that was stored as four bytes, MSB first. */ | |
| 326 #define OPERAND_MIN(p) (((long)(p)[3] << 24) + ((long)(p)[4] << 16) \ | |
| 327 + ((long)(p)[5] << 8) + (long)(p)[6]) | |
| 328 /* Obtain a second operand stored as four bytes. */ | |
| 329 #define OPERAND_MAX(p) OPERAND_MIN((p) + 4) | |
| 330 /* Obtain a second single-byte operand stored after a four bytes operand. */ | |
| 331 #define OPERAND_CMP(p) (p)[7] | |
| 332 | |
| 333 /* | |
| 334 * Utility definitions. | |
| 335 */ | |
| 336 #define UCHARAT(p) ((int)*(char_u *)(p)) | |
| 337 | |
| 338 /* Used for an error (down from) vim_regcomp(): give the error message, set | |
| 339 * rc_did_emsg and return NULL */ | |
| 653 | 340 #define EMSG_RET_NULL(m) return (EMSG(m), rc_did_emsg = TRUE, (void *)NULL) |
| 308 | 341 #define EMSG_RET_FAIL(m) return (EMSG(m), rc_did_emsg = TRUE, FAIL) |
| 4444 | 342 #define EMSG2_RET_NULL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, (void *)NULL) |
| 343 #define EMSG2_RET_FAIL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, FAIL) | |
| 344 #define EMSG_ONE_RET_NULL EMSG2_RET_NULL(_("E369: invalid item in %s%%[]"), reg_magic == MAGIC_ALL) | |
| 7 | 345 |
| 346 #define MAX_LIMIT (32767L << 16L) | |
| 347 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
348 static int re_multi_type(int); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
349 static int cstrncmp(char_u *s1, char_u *s2, int *n); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
350 static char_u *cstrchr(char_u *, int); |
| 7 | 351 |
| 4444 | 352 #ifdef BT_REGEXP_DUMP |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
353 static void regdump(char_u *, bt_regprog_T *); |
| 4444 | 354 #endif |
| 7 | 355 #ifdef DEBUG |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
356 static char_u *regprop(char_u *); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
357 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
358 |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
359 static int re_mult_next(char *what); |
| 6203 | 360 |
| 4444 | 361 static char_u e_missingbracket[] = N_("E769: Missing ] after %s["); |
| 362 static char_u e_unmatchedpp[] = N_("E53: Unmatched %s%%("); | |
| 363 static char_u e_unmatchedp[] = N_("E54: Unmatched %s("); | |
| 364 static char_u e_unmatchedpar[] = N_("E55: Unmatched %s)"); | |
|
4720
bd6bef0bd0fb
updated for version 7.3.1107
Bram Moolenaar <bram@vim.org>
parents:
4688
diff
changeset
|
365 #ifdef FEAT_SYN_HL |
|
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
366 static char_u e_z_not_allowed[] = N_("E66: \\z( not allowed here"); |
|
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
367 static char_u e_z1_not_allowed[] = N_("E67: \\z1 et al. not allowed here"); |
|
4720
bd6bef0bd0fb
updated for version 7.3.1107
Bram Moolenaar <bram@vim.org>
parents:
4688
diff
changeset
|
368 #endif |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
369 static char_u e_missing_sb[] = N_("E69: Missing ] after %s%%["); |
|
4760
532a9855bd30
updated for version 7.3.1127
Bram Moolenaar <bram@vim.org>
parents:
4746
diff
changeset
|
370 static char_u e_empty_sb[] = N_("E70: Empty %s%%[]"); |
| 7 | 371 #define NOT_MULTI 0 |
| 372 #define MULTI_ONE 1 | |
| 373 #define MULTI_MULT 2 | |
| 374 /* | |
| 375 * Return NOT_MULTI if c is not a "multi" operator. | |
| 376 * Return MULTI_ONE if c is a single "multi" operator. | |
| 377 * Return MULTI_MULT if c is a multi "multi" operator. | |
| 378 */ | |
| 379 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
380 re_multi_type(int c) |
| 7 | 381 { |
| 382 if (c == Magic('@') || c == Magic('=') || c == Magic('?')) | |
| 383 return MULTI_ONE; | |
| 384 if (c == Magic('*') || c == Magic('+') || c == Magic('{')) | |
| 385 return MULTI_MULT; | |
| 386 return NOT_MULTI; | |
| 387 } | |
| 388 | |
| 389 /* | |
| 390 * Flags to be passed up and down. | |
| 391 */ | |
| 392 #define HASWIDTH 0x1 /* Known never to match null string. */ | |
| 393 #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ | |
| 394 #define SPSTART 0x4 /* Starts with * or +. */ | |
| 395 #define HASNL 0x8 /* Contains some \n. */ | |
| 396 #define HASLOOKBH 0x10 /* Contains "\@<=" or "\@<!". */ | |
| 397 #define WORST 0 /* Worst case. */ | |
| 398 | |
| 399 /* | |
| 400 * When regcode is set to this value, code is not emitted and size is computed | |
| 401 * instead. | |
| 402 */ | |
| 403 #define JUST_CALC_SIZE ((char_u *) -1) | |
| 404 | |
| 359 | 405 static char_u *reg_prev_sub = NULL; |
| 406 | |
| 7 | 407 /* |
| 408 * REGEXP_INRANGE contains all characters which are always special in a [] | |
| 409 * range after '\'. | |
| 410 * REGEXP_ABBR contains all characters which act as abbreviations after '\'. | |
| 411 * These are: | |
| 412 * \n - New line (NL). | |
| 413 * \r - Carriage Return (CR). | |
| 414 * \t - Tab (TAB). | |
| 415 * \e - Escape (ESC). | |
| 416 * \b - Backspace (Ctrl_H). | |
| 24 | 417 * \d - Character code in decimal, eg \d123 |
| 418 * \o - Character code in octal, eg \o80 | |
| 419 * \x - Character code in hex, eg \x4a | |
| 420 * \u - Multibyte character code, eg \u20ac | |
| 421 * \U - Long multibyte character code, eg \U12345678 | |
| 7 | 422 */ |
| 423 static char_u REGEXP_INRANGE[] = "]^-n\\"; | |
| 24 | 424 static char_u REGEXP_ABBR[] = "nrtebdoxuU"; |
| 7 | 425 |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
426 static int backslash_trans(int c); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
427 static int get_char_class(char_u **pp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
428 static int get_equi_class(char_u **pp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
429 static void reg_equi_class(int c); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
430 static int get_coll_element(char_u **pp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
431 static char_u *skip_anyof(char_u *p); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
432 static void init_class_tab(void); |
| 7 | 433 |
| 434 /* | |
| 435 * Translate '\x' to its control character, except "\n", which is Magic. | |
| 436 */ | |
| 437 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
438 backslash_trans(int c) |
| 7 | 439 { |
| 440 switch (c) | |
| 441 { | |
| 442 case 'r': return CAR; | |
| 443 case 't': return TAB; | |
| 444 case 'e': return ESC; | |
| 445 case 'b': return BS; | |
| 446 } | |
| 447 return c; | |
| 448 } | |
| 449 | |
| 450 /* | |
| 167 | 451 * Check for a character class name "[:name:]". "pp" points to the '['. |
| 7 | 452 * Returns one of the CLASS_ items. CLASS_NONE means that no item was |
| 453 * recognized. Otherwise "pp" is advanced to after the item. | |
| 454 */ | |
| 455 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
456 get_char_class(char_u **pp) |
| 7 | 457 { |
| 458 static const char *(class_names[]) = | |
| 459 { | |
| 460 "alnum:]", | |
| 461 #define CLASS_ALNUM 0 | |
| 462 "alpha:]", | |
| 463 #define CLASS_ALPHA 1 | |
| 464 "blank:]", | |
| 465 #define CLASS_BLANK 2 | |
| 466 "cntrl:]", | |
| 467 #define CLASS_CNTRL 3 | |
| 468 "digit:]", | |
| 469 #define CLASS_DIGIT 4 | |
| 470 "graph:]", | |
| 471 #define CLASS_GRAPH 5 | |
| 472 "lower:]", | |
| 473 #define CLASS_LOWER 6 | |
| 474 "print:]", | |
| 475 #define CLASS_PRINT 7 | |
| 476 "punct:]", | |
| 477 #define CLASS_PUNCT 8 | |
| 478 "space:]", | |
| 479 #define CLASS_SPACE 9 | |
| 480 "upper:]", | |
| 481 #define CLASS_UPPER 10 | |
| 482 "xdigit:]", | |
| 483 #define CLASS_XDIGIT 11 | |
| 484 "tab:]", | |
| 485 #define CLASS_TAB 12 | |
| 486 "return:]", | |
| 487 #define CLASS_RETURN 13 | |
| 488 "backspace:]", | |
| 489 #define CLASS_BACKSPACE 14 | |
| 490 "escape:]", | |
| 491 #define CLASS_ESCAPE 15 | |
| 492 }; | |
| 493 #define CLASS_NONE 99 | |
| 494 int i; | |
| 495 | |
| 496 if ((*pp)[1] == ':') | |
| 497 { | |
| 1877 | 498 for (i = 0; i < (int)(sizeof(class_names) / sizeof(*class_names)); ++i) |
| 7 | 499 if (STRNCMP(*pp + 2, class_names[i], STRLEN(class_names[i])) == 0) |
| 500 { | |
| 501 *pp += STRLEN(class_names[i]) + 2; | |
| 502 return i; | |
| 503 } | |
| 504 } | |
| 505 return CLASS_NONE; | |
| 506 } | |
| 507 | |
| 508 /* | |
| 509 * Specific version of character class functions. | |
| 510 * Using a table to keep this fast. | |
| 511 */ | |
| 512 static short class_tab[256]; | |
| 513 | |
| 514 #define RI_DIGIT 0x01 | |
| 515 #define RI_HEX 0x02 | |
| 516 #define RI_OCTAL 0x04 | |
| 517 #define RI_WORD 0x08 | |
| 518 #define RI_HEAD 0x10 | |
| 519 #define RI_ALPHA 0x20 | |
| 520 #define RI_LOWER 0x40 | |
| 521 #define RI_UPPER 0x80 | |
| 522 #define RI_WHITE 0x100 | |
| 523 | |
| 524 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
525 init_class_tab(void) |
| 7 | 526 { |
| 527 int i; | |
| 528 static int done = FALSE; | |
| 529 | |
| 530 if (done) | |
| 531 return; | |
| 532 | |
| 533 for (i = 0; i < 256; ++i) | |
| 534 { | |
| 535 if (i >= '0' && i <= '7') | |
| 536 class_tab[i] = RI_DIGIT + RI_HEX + RI_OCTAL + RI_WORD; | |
| 537 else if (i >= '8' && i <= '9') | |
| 538 class_tab[i] = RI_DIGIT + RI_HEX + RI_WORD; | |
| 539 else if (i >= 'a' && i <= 'f') | |
| 540 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
| 541 #ifdef EBCDIC | |
| 542 else if ((i >= 'g' && i <= 'i') || (i >= 'j' && i <= 'r') | |
| 543 || (i >= 's' && i <= 'z')) | |
| 544 #else | |
| 545 else if (i >= 'g' && i <= 'z') | |
| 546 #endif | |
| 547 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
| 548 else if (i >= 'A' && i <= 'F') | |
| 549 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
| 550 #ifdef EBCDIC | |
| 551 else if ((i >= 'G' && i <= 'I') || ( i >= 'J' && i <= 'R') | |
| 552 || (i >= 'S' && i <= 'Z')) | |
| 553 #else | |
| 554 else if (i >= 'G' && i <= 'Z') | |
| 555 #endif | |
| 556 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
| 557 else if (i == '_') | |
| 558 class_tab[i] = RI_WORD + RI_HEAD; | |
| 559 else | |
| 560 class_tab[i] = 0; | |
| 561 } | |
| 562 class_tab[' '] |= RI_WHITE; | |
| 563 class_tab['\t'] |= RI_WHITE; | |
| 564 done = TRUE; | |
| 565 } | |
| 566 | |
| 567 #ifdef FEAT_MBYTE | |
| 568 # define ri_digit(c) (c < 0x100 && (class_tab[c] & RI_DIGIT)) | |
| 569 # define ri_hex(c) (c < 0x100 && (class_tab[c] & RI_HEX)) | |
| 570 # define ri_octal(c) (c < 0x100 && (class_tab[c] & RI_OCTAL)) | |
| 571 # define ri_word(c) (c < 0x100 && (class_tab[c] & RI_WORD)) | |
| 572 # define ri_head(c) (c < 0x100 && (class_tab[c] & RI_HEAD)) | |
| 573 # define ri_alpha(c) (c < 0x100 && (class_tab[c] & RI_ALPHA)) | |
| 574 # define ri_lower(c) (c < 0x100 && (class_tab[c] & RI_LOWER)) | |
| 575 # define ri_upper(c) (c < 0x100 && (class_tab[c] & RI_UPPER)) | |
| 576 # define ri_white(c) (c < 0x100 && (class_tab[c] & RI_WHITE)) | |
| 577 #else | |
| 578 # define ri_digit(c) (class_tab[c] & RI_DIGIT) | |
| 579 # define ri_hex(c) (class_tab[c] & RI_HEX) | |
| 580 # define ri_octal(c) (class_tab[c] & RI_OCTAL) | |
| 581 # define ri_word(c) (class_tab[c] & RI_WORD) | |
| 582 # define ri_head(c) (class_tab[c] & RI_HEAD) | |
| 583 # define ri_alpha(c) (class_tab[c] & RI_ALPHA) | |
| 584 # define ri_lower(c) (class_tab[c] & RI_LOWER) | |
| 585 # define ri_upper(c) (class_tab[c] & RI_UPPER) | |
| 586 # define ri_white(c) (class_tab[c] & RI_WHITE) | |
| 587 #endif | |
| 588 | |
| 589 /* flags for regflags */ | |
| 590 #define RF_ICASE 1 /* ignore case */ | |
| 591 #define RF_NOICASE 2 /* don't ignore case */ | |
| 592 #define RF_HASNL 4 /* can match a NL */ | |
| 593 #define RF_ICOMBINE 8 /* ignore combining characters */ | |
| 594 #define RF_LOOKBH 16 /* uses "\@<=" or "\@<!" */ | |
| 595 | |
| 596 /* | |
| 597 * Global work variables for vim_regcomp(). | |
| 598 */ | |
| 599 | |
| 600 static char_u *regparse; /* Input-scan pointer. */ | |
| 601 static int prevchr_len; /* byte length of previous char */ | |
| 602 static int num_complex_braces; /* Complex \{...} count */ | |
| 603 static int regnpar; /* () count. */ | |
| 604 #ifdef FEAT_SYN_HL | |
| 605 static int regnzpar; /* \z() count. */ | |
| 606 static int re_has_z; /* \z item detected */ | |
| 607 #endif | |
| 608 static char_u *regcode; /* Code-emit pointer, or JUST_CALC_SIZE */ | |
| 609 static long regsize; /* Code size. */ | |
| 2010 | 610 static int reg_toolong; /* TRUE when offset out of range */ |
| 7 | 611 static char_u had_endbrace[NSUBEXP]; /* flags, TRUE if end of () found */ |
| 612 static unsigned regflags; /* RF_ flags for prog */ | |
| 613 static long brace_min[10]; /* Minimums for complex brace repeats */ | |
| 614 static long brace_max[10]; /* Maximums for complex brace repeats */ | |
| 615 static int brace_count[10]; /* Current counts for complex brace repeats */ | |
| 616 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 617 static int had_eol; /* TRUE when EOL found by vim_regcomp() */ | |
| 618 #endif | |
| 619 static int one_exactly = FALSE; /* only do one char for EXACTLY */ | |
| 620 | |
| 621 static int reg_magic; /* magicness of the pattern: */ | |
| 622 #define MAGIC_NONE 1 /* "\V" very unmagic */ | |
| 623 #define MAGIC_OFF 2 /* "\M" or 'magic' off */ | |
| 624 #define MAGIC_ON 3 /* "\m" or 'magic' */ | |
| 625 #define MAGIC_ALL 4 /* "\v" very magic */ | |
| 626 | |
| 627 static int reg_string; /* matching with a string instead of a buffer | |
| 628 line */ | |
| 481 | 629 static int reg_strict; /* "[abc" is illegal */ |
| 7 | 630 |
| 631 /* | |
| 632 * META contains all characters that may be magic, except '^' and '$'. | |
| 633 */ | |
| 634 | |
| 635 #ifdef EBCDIC | |
| 636 static char_u META[] = "%&()*+.123456789<=>?@ACDFHIKLMOPSUVWX[_acdfhiklmnopsuvwxz{|~"; | |
| 637 #else | |
| 638 /* META[] is used often enough to justify turning it into a table. */ | |
| 639 static char_u META_flags[] = { | |
| 640 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| 641 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| 642 /* % & ( ) * + . */ | |
| 643 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0, | |
| 644 /* 1 2 3 4 5 6 7 8 9 < = > ? */ | |
| 645 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, | |
| 646 /* @ A C D F H I K L M O */ | |
| 647 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1, | |
| 648 /* P S U V W X Z [ _ */ | |
| 649 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, | |
| 650 /* a c d f h i k l m n o */ | |
| 651 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, | |
| 652 /* p s u v w x z { | ~ */ | |
| 653 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1 | |
| 654 }; | |
| 655 #endif | |
| 656 | |
| 4444 | 657 static int curchr; /* currently parsed character */ |
| 658 /* Previous character. Note: prevchr is sometimes -1 when we are not at the | |
| 659 * start, eg in /[ ^I]^ the pattern was never found even if it existed, | |
| 660 * because ^ was taken to be magic -- webb */ | |
| 661 static int prevchr; | |
| 662 static int prevprevchr; /* previous-previous character */ | |
| 663 static int nextchr; /* used for ungetchr() */ | |
| 7 | 664 |
| 665 /* arguments for reg() */ | |
| 666 #define REG_NOPAREN 0 /* toplevel reg() */ | |
| 667 #define REG_PAREN 1 /* \(\) */ | |
| 668 #define REG_ZPAREN 2 /* \z(\) */ | |
| 669 #define REG_NPAREN 3 /* \%(\) */ | |
| 670 | |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
671 typedef struct |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
672 { |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
673 char_u *regparse; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
674 int prevchr_len; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
675 int curchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
676 int prevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
677 int prevprevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
678 int nextchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
679 int at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
680 int prev_at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
681 int regnpar; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
682 } parse_state_T; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
683 |
| 7 | 684 /* |
| 685 * Forward declarations for vim_regcomp()'s friends. | |
| 686 */ | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
687 static void initchr(char_u *); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
688 static void save_parse_state(parse_state_T *ps); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
689 static void restore_parse_state(parse_state_T *ps); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
690 static int getchr(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
691 static void skipchr_keepstart(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
692 static int peekchr(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
693 static void skipchr(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
694 static void ungetchr(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
695 static int gethexchrs(int maxinputlen); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
696 static int getoctchrs(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
697 static int getdecchrs(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
698 static int coll_get_char(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
699 static void regcomp_start(char_u *expr, int flags); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
700 static char_u *reg(int, int *); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
701 static char_u *regbranch(int *flagp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
702 static char_u *regconcat(int *flagp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
703 static char_u *regpiece(int *); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
704 static char_u *regatom(int *); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
705 static char_u *regnode(int); |
| 714 | 706 #ifdef FEAT_MBYTE |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
707 static int use_multibytecode(int c); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
708 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
709 static int prog_magic_wrong(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
710 static char_u *regnext(char_u *); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
711 static void regc(int b); |
| 7 | 712 #ifdef FEAT_MBYTE |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
713 static void regmbc(int c); |
| 2974 | 714 # define REGMBC(x) regmbc(x); |
| 715 # define CASEMBC(x) case x: | |
| 167 | 716 #else |
| 717 # define regmbc(c) regc(c) | |
| 2974 | 718 # define REGMBC(x) |
| 719 # define CASEMBC(x) | |
| 7 | 720 #endif |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
721 static void reginsert(int, char_u *); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
722 static void reginsert_nr(int op, long val, char_u *opnd); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
723 static void reginsert_limits(int, long, long, char_u *); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
724 static char_u *re_put_long(char_u *pr, long_u val); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
725 static int read_limits(long *, long *); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
726 static void regtail(char_u *, char_u *); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
727 static void regoptail(char_u *, char_u *); |
| 7 | 728 |
| 4444 | 729 static regengine_T bt_regengine; |
| 730 static regengine_T nfa_regengine; | |
| 731 | |
| 7 | 732 /* |
| 733 * Return TRUE if compiled regular expression "prog" can match a line break. | |
| 734 */ | |
| 735 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
736 re_multiline(regprog_T *prog) |
| 7 | 737 { |
| 738 return (prog->regflags & RF_HASNL); | |
| 739 } | |
| 740 | |
| 741 /* | |
| 742 * Return TRUE if compiled regular expression "prog" looks before the start | |
| 743 * position (pattern contains "\@<=" or "\@<!"). | |
| 744 */ | |
| 745 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
746 re_lookbehind(regprog_T *prog) |
| 7 | 747 { |
| 748 return (prog->regflags & RF_LOOKBH); | |
| 749 } | |
| 750 | |
| 751 /* | |
| 167 | 752 * Check for an equivalence class name "[=a=]". "pp" points to the '['. |
| 753 * Returns a character representing the class. Zero means that no item was | |
| 754 * recognized. Otherwise "pp" is advanced to after the item. | |
| 755 */ | |
| 756 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
757 get_equi_class(char_u **pp) |
| 167 | 758 { |
| 759 int c; | |
| 760 int l = 1; | |
| 761 char_u *p = *pp; | |
| 762 | |
| 763 if (p[1] == '=') | |
| 764 { | |
| 765 #ifdef FEAT_MBYTE | |
| 766 if (has_mbyte) | |
| 474 | 767 l = (*mb_ptr2len)(p + 2); |
| 167 | 768 #endif |
| 769 if (p[l + 2] == '=' && p[l + 3] == ']') | |
| 770 { | |
| 771 #ifdef FEAT_MBYTE | |
| 772 if (has_mbyte) | |
| 773 c = mb_ptr2char(p + 2); | |
| 774 else | |
| 775 #endif | |
| 776 c = p[2]; | |
| 777 *pp += l + 4; | |
| 778 return c; | |
| 779 } | |
| 780 } | |
| 781 return 0; | |
| 782 } | |
| 783 | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
784 #ifdef EBCDIC |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
785 /* |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
786 * Table for equivalence class "c". (IBM-1047) |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
787 */ |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
788 char *EQUIVAL_CLASS_C[16] = { |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
789 "A\x62\x63\x64\x65\x66\x67", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
790 "C\x68", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
791 "E\x71\x72\x73\x74", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
792 "I\x75\x76\x77\x78", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
793 "N\x69", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
794 "O\xEB\xEC\xED\xEE\xEF", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
795 "U\xFB\xFC\xFD\xFE", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
796 "Y\xBA", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
797 "a\x42\x43\x44\x45\x46\x47", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
798 "c\x48", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
799 "e\x51\x52\x53\x54", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
800 "i\x55\x56\x57\x58", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
801 "n\x49", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
802 "o\xCB\xCC\xCD\xCE\xCF", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
803 "u\xDB\xDC\xDD\xDE", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
804 "y\x8D\xDF", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
805 }; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
806 #endif |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
807 |
| 167 | 808 /* |
| 809 * Produce the bytes for equivalence class "c". | |
| 810 * Currently only handles latin1, latin9 and utf-8. | |
| 4444 | 811 * NOTE: When changing this function, also change nfa_emit_equi_class() |
| 167 | 812 */ |
| 813 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
814 reg_equi_class(int c) |
| 167 | 815 { |
| 816 #ifdef FEAT_MBYTE | |
| 817 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
| 492 | 818 || STRCMP(p_enc, "iso-8859-15") == 0) |
| 167 | 819 #endif |
| 820 { | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
821 #ifdef EBCDIC |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
822 int i; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
823 |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
824 /* This might be slower than switch/case below. */ |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
825 for (i = 0; i < 16; i++) |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
826 { |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
827 if (vim_strchr(EQUIVAL_CLASS_C[i], c) != NULL) |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
828 { |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
829 char *p = EQUIVAL_CLASS_C[i]; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
830 |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
831 while (*p != 0) |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
832 regmbc(*p++); |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
833 return; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
834 } |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
835 } |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
836 #else |
| 167 | 837 switch (c) |
| 838 { | |
| 6765 | 839 /* Do not use '\300' style, it results in a negative number. */ |
| 840 case 'A': case 0xc0: case 0xc1: case 0xc2: | |
| 841 case 0xc3: case 0xc4: case 0xc5: | |
| 2974 | 842 CASEMBC(0x100) CASEMBC(0x102) CASEMBC(0x104) CASEMBC(0x1cd) |
| 843 CASEMBC(0x1de) CASEMBC(0x1e0) CASEMBC(0x1ea2) | |
| 6765 | 844 regmbc('A'); regmbc(0xc0); regmbc(0xc1); |
| 845 regmbc(0xc2); regmbc(0xc3); regmbc(0xc4); | |
| 846 regmbc(0xc5); | |
| 2974 | 847 REGMBC(0x100) REGMBC(0x102) REGMBC(0x104) |
| 848 REGMBC(0x1cd) REGMBC(0x1de) REGMBC(0x1e0) | |
| 849 REGMBC(0x1ea2) | |
| 850 return; | |
| 851 case 'B': CASEMBC(0x1e02) CASEMBC(0x1e06) | |
| 852 regmbc('B'); REGMBC(0x1e02) REGMBC(0x1e06) | |
| 167 | 853 return; |
| 6765 | 854 case 'C': case 0xc7: |
| 2974 | 855 CASEMBC(0x106) CASEMBC(0x108) CASEMBC(0x10a) CASEMBC(0x10c) |
| 6765 | 856 regmbc('C'); regmbc(0xc7); |
| 2974 | 857 REGMBC(0x106) REGMBC(0x108) REGMBC(0x10a) |
| 858 REGMBC(0x10c) | |
| 859 return; | |
| 860 case 'D': CASEMBC(0x10e) CASEMBC(0x110) CASEMBC(0x1e0a) | |
| 861 CASEMBC(0x1e0e) CASEMBC(0x1e10) | |
| 862 regmbc('D'); REGMBC(0x10e) REGMBC(0x110) | |
| 863 REGMBC(0x1e0a) REGMBC(0x1e0e) REGMBC(0x1e10) | |
| 167 | 864 return; |
| 6765 | 865 case 'E': case 0xc8: case 0xc9: case 0xca: case 0xcb: |
| 2974 | 866 CASEMBC(0x112) CASEMBC(0x114) CASEMBC(0x116) CASEMBC(0x118) |
| 867 CASEMBC(0x11a) CASEMBC(0x1eba) CASEMBC(0x1ebc) | |
| 6765 | 868 regmbc('E'); regmbc(0xc8); regmbc(0xc9); |
| 869 regmbc(0xca); regmbc(0xcb); | |
| 2974 | 870 REGMBC(0x112) REGMBC(0x114) REGMBC(0x116) |
| 871 REGMBC(0x118) REGMBC(0x11a) REGMBC(0x1eba) | |
| 872 REGMBC(0x1ebc) | |
| 873 return; | |
| 874 case 'F': CASEMBC(0x1e1e) | |
| 875 regmbc('F'); REGMBC(0x1e1e) | |
| 876 return; | |
| 877 case 'G': CASEMBC(0x11c) CASEMBC(0x11e) CASEMBC(0x120) | |
| 878 CASEMBC(0x122) CASEMBC(0x1e4) CASEMBC(0x1e6) CASEMBC(0x1f4) | |
| 879 CASEMBC(0x1e20) | |
| 880 regmbc('G'); REGMBC(0x11c) REGMBC(0x11e) | |
| 881 REGMBC(0x120) REGMBC(0x122) REGMBC(0x1e4) | |
| 882 REGMBC(0x1e6) REGMBC(0x1f4) REGMBC(0x1e20) | |
| 883 return; | |
| 884 case 'H': CASEMBC(0x124) CASEMBC(0x126) CASEMBC(0x1e22) | |
| 885 CASEMBC(0x1e26) CASEMBC(0x1e28) | |
| 886 regmbc('H'); REGMBC(0x124) REGMBC(0x126) | |
| 887 REGMBC(0x1e22) REGMBC(0x1e26) REGMBC(0x1e28) | |
| 167 | 888 return; |
| 6765 | 889 case 'I': case 0xcc: case 0xcd: case 0xce: case 0xcf: |
| 2974 | 890 CASEMBC(0x128) CASEMBC(0x12a) CASEMBC(0x12c) CASEMBC(0x12e) |
| 891 CASEMBC(0x130) CASEMBC(0x1cf) CASEMBC(0x1ec8) | |
| 6765 | 892 regmbc('I'); regmbc(0xcc); regmbc(0xcd); |
| 893 regmbc(0xce); regmbc(0xcf); | |
| 2974 | 894 REGMBC(0x128) REGMBC(0x12a) REGMBC(0x12c) |
| 895 REGMBC(0x12e) REGMBC(0x130) REGMBC(0x1cf) | |
| 896 REGMBC(0x1ec8) | |
| 897 return; | |
| 898 case 'J': CASEMBC(0x134) | |
| 899 regmbc('J'); REGMBC(0x134) | |
| 900 return; | |
| 901 case 'K': CASEMBC(0x136) CASEMBC(0x1e8) CASEMBC(0x1e30) | |
| 902 CASEMBC(0x1e34) | |
| 903 regmbc('K'); REGMBC(0x136) REGMBC(0x1e8) | |
| 904 REGMBC(0x1e30) REGMBC(0x1e34) | |
| 905 return; | |
| 906 case 'L': CASEMBC(0x139) CASEMBC(0x13b) CASEMBC(0x13d) | |
| 907 CASEMBC(0x13f) CASEMBC(0x141) CASEMBC(0x1e3a) | |
| 908 regmbc('L'); REGMBC(0x139) REGMBC(0x13b) | |
| 909 REGMBC(0x13d) REGMBC(0x13f) REGMBC(0x141) | |
| 910 REGMBC(0x1e3a) | |
| 911 return; | |
| 912 case 'M': CASEMBC(0x1e3e) CASEMBC(0x1e40) | |
| 913 regmbc('M'); REGMBC(0x1e3e) REGMBC(0x1e40) | |
| 167 | 914 return; |
| 6765 | 915 case 'N': case 0xd1: |
| 2974 | 916 CASEMBC(0x143) CASEMBC(0x145) CASEMBC(0x147) CASEMBC(0x1e44) |
| 917 CASEMBC(0x1e48) | |
| 6765 | 918 regmbc('N'); regmbc(0xd1); |
| 2974 | 919 REGMBC(0x143) REGMBC(0x145) REGMBC(0x147) |
| 920 REGMBC(0x1e44) REGMBC(0x1e48) | |
| 167 | 921 return; |
| 6765 | 922 case 'O': case 0xd2: case 0xd3: case 0xd4: case 0xd5: |
| 923 case 0xd6: case 0xd8: | |
| 2974 | 924 CASEMBC(0x14c) CASEMBC(0x14e) CASEMBC(0x150) CASEMBC(0x1a0) |
| 925 CASEMBC(0x1d1) CASEMBC(0x1ea) CASEMBC(0x1ec) CASEMBC(0x1ece) | |
| 6765 | 926 regmbc('O'); regmbc(0xd2); regmbc(0xd3); |
| 927 regmbc(0xd4); regmbc(0xd5); regmbc(0xd6); | |
| 928 regmbc(0xd8); | |
| 2974 | 929 REGMBC(0x14c) REGMBC(0x14e) REGMBC(0x150) |
| 930 REGMBC(0x1a0) REGMBC(0x1d1) REGMBC(0x1ea) | |
| 931 REGMBC(0x1ec) REGMBC(0x1ece) | |
| 932 return; | |
| 933 case 'P': case 0x1e54: case 0x1e56: | |
| 934 regmbc('P'); REGMBC(0x1e54) REGMBC(0x1e56) | |
| 935 return; | |
| 936 case 'R': CASEMBC(0x154) CASEMBC(0x156) CASEMBC(0x158) | |
| 937 CASEMBC(0x1e58) CASEMBC(0x1e5e) | |
| 938 regmbc('R'); REGMBC(0x154) REGMBC(0x156) REGMBC(0x158) | |
| 939 REGMBC(0x1e58) REGMBC(0x1e5e) | |
| 940 return; | |
| 941 case 'S': CASEMBC(0x15a) CASEMBC(0x15c) CASEMBC(0x15e) | |
| 942 CASEMBC(0x160) CASEMBC(0x1e60) | |
| 943 regmbc('S'); REGMBC(0x15a) REGMBC(0x15c) | |
| 944 REGMBC(0x15e) REGMBC(0x160) REGMBC(0x1e60) | |
| 945 return; | |
| 946 case 'T': CASEMBC(0x162) CASEMBC(0x164) CASEMBC(0x166) | |
| 947 CASEMBC(0x1e6a) CASEMBC(0x1e6e) | |
| 948 regmbc('T'); REGMBC(0x162) REGMBC(0x164) | |
| 949 REGMBC(0x166) REGMBC(0x1e6a) REGMBC(0x1e6e) | |
| 167 | 950 return; |
| 6765 | 951 case 'U': case 0xd9: case 0xda: case 0xdb: case 0xdc: |
| 2974 | 952 CASEMBC(0x168) CASEMBC(0x16a) CASEMBC(0x16c) CASEMBC(0x16e) |
| 953 CASEMBC(0x170) CASEMBC(0x172) CASEMBC(0x1af) CASEMBC(0x1d3) | |
| 954 CASEMBC(0x1ee6) | |
| 6765 | 955 regmbc('U'); regmbc(0xd9); regmbc(0xda); |
| 956 regmbc(0xdb); regmbc(0xdc); | |
| 2974 | 957 REGMBC(0x168) REGMBC(0x16a) REGMBC(0x16c) |
| 958 REGMBC(0x16e) REGMBC(0x170) REGMBC(0x172) | |
| 959 REGMBC(0x1af) REGMBC(0x1d3) REGMBC(0x1ee6) | |
| 960 return; | |
| 961 case 'V': CASEMBC(0x1e7c) | |
| 962 regmbc('V'); REGMBC(0x1e7c) | |
| 963 return; | |
| 964 case 'W': CASEMBC(0x174) CASEMBC(0x1e80) CASEMBC(0x1e82) | |
| 965 CASEMBC(0x1e84) CASEMBC(0x1e86) | |
| 966 regmbc('W'); REGMBC(0x174) REGMBC(0x1e80) | |
| 967 REGMBC(0x1e82) REGMBC(0x1e84) REGMBC(0x1e86) | |
| 968 return; | |
| 969 case 'X': CASEMBC(0x1e8a) CASEMBC(0x1e8c) | |
| 970 regmbc('X'); REGMBC(0x1e8a) REGMBC(0x1e8c) | |
| 167 | 971 return; |
| 6765 | 972 case 'Y': case 0xdd: |
| 2974 | 973 CASEMBC(0x176) CASEMBC(0x178) CASEMBC(0x1e8e) CASEMBC(0x1ef2) |
| 974 CASEMBC(0x1ef6) CASEMBC(0x1ef8) | |
| 6765 | 975 regmbc('Y'); regmbc(0xdd); |
| 2974 | 976 REGMBC(0x176) REGMBC(0x178) REGMBC(0x1e8e) |
| 977 REGMBC(0x1ef2) REGMBC(0x1ef6) REGMBC(0x1ef8) | |
| 978 return; | |
| 979 case 'Z': CASEMBC(0x179) CASEMBC(0x17b) CASEMBC(0x17d) | |
| 980 CASEMBC(0x1b5) CASEMBC(0x1e90) CASEMBC(0x1e94) | |
| 981 regmbc('Z'); REGMBC(0x179) REGMBC(0x17b) | |
| 982 REGMBC(0x17d) REGMBC(0x1b5) REGMBC(0x1e90) | |
| 983 REGMBC(0x1e94) | |
| 167 | 984 return; |
| 6765 | 985 case 'a': case 0xe0: case 0xe1: case 0xe2: |
| 986 case 0xe3: case 0xe4: case 0xe5: | |
| 2974 | 987 CASEMBC(0x101) CASEMBC(0x103) CASEMBC(0x105) CASEMBC(0x1ce) |
| 988 CASEMBC(0x1df) CASEMBC(0x1e1) CASEMBC(0x1ea3) | |
| 6765 | 989 regmbc('a'); regmbc(0xe0); regmbc(0xe1); |
| 990 regmbc(0xe2); regmbc(0xe3); regmbc(0xe4); | |
| 991 regmbc(0xe5); | |
| 2974 | 992 REGMBC(0x101) REGMBC(0x103) REGMBC(0x105) |
| 993 REGMBC(0x1ce) REGMBC(0x1df) REGMBC(0x1e1) | |
| 994 REGMBC(0x1ea3) | |
| 995 return; | |
| 996 case 'b': CASEMBC(0x1e03) CASEMBC(0x1e07) | |
| 997 regmbc('b'); REGMBC(0x1e03) REGMBC(0x1e07) | |
| 167 | 998 return; |
| 6765 | 999 case 'c': case 0xe7: |
| 2974 | 1000 CASEMBC(0x107) CASEMBC(0x109) CASEMBC(0x10b) CASEMBC(0x10d) |
| 6765 | 1001 regmbc('c'); regmbc(0xe7); |
| 2974 | 1002 REGMBC(0x107) REGMBC(0x109) REGMBC(0x10b) |
| 1003 REGMBC(0x10d) | |
| 1004 return; | |
| 6914 | 1005 case 'd': CASEMBC(0x10f) CASEMBC(0x111) CASEMBC(0x1e0b) |
| 1006 CASEMBC(0x1e0f) CASEMBC(0x1e11) | |
| 2974 | 1007 regmbc('d'); REGMBC(0x10f) REGMBC(0x111) |
| 6914 | 1008 REGMBC(0x1e0b) REGMBC(0x1e0f) REGMBC(0x1e11) |
| 167 | 1009 return; |
| 6765 | 1010 case 'e': case 0xe8: case 0xe9: case 0xea: case 0xeb: |
| 2974 | 1011 CASEMBC(0x113) CASEMBC(0x115) CASEMBC(0x117) CASEMBC(0x119) |
| 1012 CASEMBC(0x11b) CASEMBC(0x1ebb) CASEMBC(0x1ebd) | |
| 6765 | 1013 regmbc('e'); regmbc(0xe8); regmbc(0xe9); |
| 1014 regmbc(0xea); regmbc(0xeb); | |
| 2974 | 1015 REGMBC(0x113) REGMBC(0x115) REGMBC(0x117) |
| 1016 REGMBC(0x119) REGMBC(0x11b) REGMBC(0x1ebb) | |
| 1017 REGMBC(0x1ebd) | |
| 1018 return; | |
| 1019 case 'f': CASEMBC(0x1e1f) | |
| 1020 regmbc('f'); REGMBC(0x1e1f) | |
| 1021 return; | |
| 1022 case 'g': CASEMBC(0x11d) CASEMBC(0x11f) CASEMBC(0x121) | |
| 1023 CASEMBC(0x123) CASEMBC(0x1e5) CASEMBC(0x1e7) CASEMBC(0x1f5) | |
| 1024 CASEMBC(0x1e21) | |
| 1025 regmbc('g'); REGMBC(0x11d) REGMBC(0x11f) | |
| 1026 REGMBC(0x121) REGMBC(0x123) REGMBC(0x1e5) | |
| 1027 REGMBC(0x1e7) REGMBC(0x1f5) REGMBC(0x1e21) | |
| 1028 return; | |
| 1029 case 'h': CASEMBC(0x125) CASEMBC(0x127) CASEMBC(0x1e23) | |
| 1030 CASEMBC(0x1e27) CASEMBC(0x1e29) CASEMBC(0x1e96) | |
| 1031 regmbc('h'); REGMBC(0x125) REGMBC(0x127) | |
| 1032 REGMBC(0x1e23) REGMBC(0x1e27) REGMBC(0x1e29) | |
| 1033 REGMBC(0x1e96) | |
| 167 | 1034 return; |
| 6765 | 1035 case 'i': case 0xec: case 0xed: case 0xee: case 0xef: |
| 2974 | 1036 CASEMBC(0x129) CASEMBC(0x12b) CASEMBC(0x12d) CASEMBC(0x12f) |
| 1037 CASEMBC(0x1d0) CASEMBC(0x1ec9) | |
| 6765 | 1038 regmbc('i'); regmbc(0xec); regmbc(0xed); |
| 1039 regmbc(0xee); regmbc(0xef); | |
| 2974 | 1040 REGMBC(0x129) REGMBC(0x12b) REGMBC(0x12d) |
| 1041 REGMBC(0x12f) REGMBC(0x1d0) REGMBC(0x1ec9) | |
| 1042 return; | |
| 1043 case 'j': CASEMBC(0x135) CASEMBC(0x1f0) | |
| 1044 regmbc('j'); REGMBC(0x135) REGMBC(0x1f0) | |
| 1045 return; | |
| 1046 case 'k': CASEMBC(0x137) CASEMBC(0x1e9) CASEMBC(0x1e31) | |
| 1047 CASEMBC(0x1e35) | |
| 1048 regmbc('k'); REGMBC(0x137) REGMBC(0x1e9) | |
| 1049 REGMBC(0x1e31) REGMBC(0x1e35) | |
| 1050 return; | |
| 1051 case 'l': CASEMBC(0x13a) CASEMBC(0x13c) CASEMBC(0x13e) | |
| 1052 CASEMBC(0x140) CASEMBC(0x142) CASEMBC(0x1e3b) | |
| 1053 regmbc('l'); REGMBC(0x13a) REGMBC(0x13c) | |
| 1054 REGMBC(0x13e) REGMBC(0x140) REGMBC(0x142) | |
| 1055 REGMBC(0x1e3b) | |
| 1056 return; | |
| 1057 case 'm': CASEMBC(0x1e3f) CASEMBC(0x1e41) | |
| 1058 regmbc('m'); REGMBC(0x1e3f) REGMBC(0x1e41) | |
| 167 | 1059 return; |
| 6765 | 1060 case 'n': case 0xf1: |
| 2974 | 1061 CASEMBC(0x144) CASEMBC(0x146) CASEMBC(0x148) CASEMBC(0x149) |
| 1062 CASEMBC(0x1e45) CASEMBC(0x1e49) | |
| 6765 | 1063 regmbc('n'); regmbc(0xf1); |
| 2974 | 1064 REGMBC(0x144) REGMBC(0x146) REGMBC(0x148) |
| 1065 REGMBC(0x149) REGMBC(0x1e45) REGMBC(0x1e49) | |
| 167 | 1066 return; |
| 6765 | 1067 case 'o': case 0xf2: case 0xf3: case 0xf4: case 0xf5: |
| 1068 case 0xf6: case 0xf8: | |
| 2974 | 1069 CASEMBC(0x14d) CASEMBC(0x14f) CASEMBC(0x151) CASEMBC(0x1a1) |
| 1070 CASEMBC(0x1d2) CASEMBC(0x1eb) CASEMBC(0x1ed) CASEMBC(0x1ecf) | |
| 6765 | 1071 regmbc('o'); regmbc(0xf2); regmbc(0xf3); |
| 1072 regmbc(0xf4); regmbc(0xf5); regmbc(0xf6); | |
| 1073 regmbc(0xf8); | |
| 2974 | 1074 REGMBC(0x14d) REGMBC(0x14f) REGMBC(0x151) |
| 1075 REGMBC(0x1a1) REGMBC(0x1d2) REGMBC(0x1eb) | |
| 1076 REGMBC(0x1ed) REGMBC(0x1ecf) | |
| 1077 return; | |
| 1078 case 'p': CASEMBC(0x1e55) CASEMBC(0x1e57) | |
| 1079 regmbc('p'); REGMBC(0x1e55) REGMBC(0x1e57) | |
| 1080 return; | |
| 1081 case 'r': CASEMBC(0x155) CASEMBC(0x157) CASEMBC(0x159) | |
| 1082 CASEMBC(0x1e59) CASEMBC(0x1e5f) | |
| 1083 regmbc('r'); REGMBC(0x155) REGMBC(0x157) REGMBC(0x159) | |
| 1084 REGMBC(0x1e59) REGMBC(0x1e5f) | |
| 1085 return; | |
| 1086 case 's': CASEMBC(0x15b) CASEMBC(0x15d) CASEMBC(0x15f) | |
| 1087 CASEMBC(0x161) CASEMBC(0x1e61) | |
| 1088 regmbc('s'); REGMBC(0x15b) REGMBC(0x15d) | |
| 1089 REGMBC(0x15f) REGMBC(0x161) REGMBC(0x1e61) | |
| 1090 return; | |
| 1091 case 't': CASEMBC(0x163) CASEMBC(0x165) CASEMBC(0x167) | |
| 1092 CASEMBC(0x1e6b) CASEMBC(0x1e6f) CASEMBC(0x1e97) | |
| 1093 regmbc('t'); REGMBC(0x163) REGMBC(0x165) REGMBC(0x167) | |
| 1094 REGMBC(0x1e6b) REGMBC(0x1e6f) REGMBC(0x1e97) | |
| 167 | 1095 return; |
| 6765 | 1096 case 'u': case 0xf9: case 0xfa: case 0xfb: case 0xfc: |
| 2974 | 1097 CASEMBC(0x169) CASEMBC(0x16b) CASEMBC(0x16d) CASEMBC(0x16f) |
| 1098 CASEMBC(0x171) CASEMBC(0x173) CASEMBC(0x1b0) CASEMBC(0x1d4) | |
| 1099 CASEMBC(0x1ee7) | |
| 6765 | 1100 regmbc('u'); regmbc(0xf9); regmbc(0xfa); |
| 1101 regmbc(0xfb); regmbc(0xfc); | |
| 2974 | 1102 REGMBC(0x169) REGMBC(0x16b) REGMBC(0x16d) |
| 1103 REGMBC(0x16f) REGMBC(0x171) REGMBC(0x173) | |
| 1104 REGMBC(0x1b0) REGMBC(0x1d4) REGMBC(0x1ee7) | |
| 1105 return; | |
| 1106 case 'v': CASEMBC(0x1e7d) | |
| 1107 regmbc('v'); REGMBC(0x1e7d) | |
| 1108 return; | |
| 1109 case 'w': CASEMBC(0x175) CASEMBC(0x1e81) CASEMBC(0x1e83) | |
| 1110 CASEMBC(0x1e85) CASEMBC(0x1e87) CASEMBC(0x1e98) | |
| 1111 regmbc('w'); REGMBC(0x175) REGMBC(0x1e81) | |
| 1112 REGMBC(0x1e83) REGMBC(0x1e85) REGMBC(0x1e87) | |
| 1113 REGMBC(0x1e98) | |
| 1114 return; | |
| 1115 case 'x': CASEMBC(0x1e8b) CASEMBC(0x1e8d) | |
| 1116 regmbc('x'); REGMBC(0x1e8b) REGMBC(0x1e8d) | |
| 167 | 1117 return; |
| 6765 | 1118 case 'y': case 0xfd: case 0xff: |
| 2974 | 1119 CASEMBC(0x177) CASEMBC(0x1e8f) CASEMBC(0x1e99) |
| 1120 CASEMBC(0x1ef3) CASEMBC(0x1ef7) CASEMBC(0x1ef9) | |
| 6765 | 1121 regmbc('y'); regmbc(0xfd); regmbc(0xff); |
| 2974 | 1122 REGMBC(0x177) REGMBC(0x1e8f) REGMBC(0x1e99) |
| 1123 REGMBC(0x1ef3) REGMBC(0x1ef7) REGMBC(0x1ef9) | |
| 1124 return; | |
| 1125 case 'z': CASEMBC(0x17a) CASEMBC(0x17c) CASEMBC(0x17e) | |
| 1126 CASEMBC(0x1b6) CASEMBC(0x1e91) CASEMBC(0x1e95) | |
| 1127 regmbc('z'); REGMBC(0x17a) REGMBC(0x17c) | |
| 1128 REGMBC(0x17e) REGMBC(0x1b6) REGMBC(0x1e91) | |
| 1129 REGMBC(0x1e95) | |
| 167 | 1130 return; |
| 1131 } | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
1132 #endif |
| 167 | 1133 } |
| 1134 regmbc(c); | |
| 1135 } | |
| 1136 | |
| 1137 /* | |
| 1138 * Check for a collating element "[.a.]". "pp" points to the '['. | |
| 1139 * Returns a character. Zero means that no item was recognized. Otherwise | |
| 1140 * "pp" is advanced to after the item. | |
| 1141 * Currently only single characters are recognized! | |
| 1142 */ | |
| 1143 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1144 get_coll_element(char_u **pp) |
| 167 | 1145 { |
| 1146 int c; | |
| 1147 int l = 1; | |
| 1148 char_u *p = *pp; | |
| 1149 | |
| 6830 | 1150 if (p[0] != NUL && p[1] == '.') |
| 167 | 1151 { |
| 1152 #ifdef FEAT_MBYTE | |
| 1153 if (has_mbyte) | |
| 474 | 1154 l = (*mb_ptr2len)(p + 2); |
| 167 | 1155 #endif |
| 1156 if (p[l + 2] == '.' && p[l + 3] == ']') | |
| 1157 { | |
| 1158 #ifdef FEAT_MBYTE | |
| 1159 if (has_mbyte) | |
| 1160 c = mb_ptr2char(p + 2); | |
| 1161 else | |
| 1162 #endif | |
| 1163 c = p[2]; | |
| 1164 *pp += l + 4; | |
| 1165 return c; | |
| 1166 } | |
| 1167 } | |
| 1168 return 0; | |
| 1169 } | |
| 1170 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
1171 static void get_cpo_flags(void); |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1172 static int reg_cpo_lit; /* 'cpoptions' contains 'l' flag */ |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1173 static int reg_cpo_bsl; /* 'cpoptions' contains '\' flag */ |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1174 |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1175 static void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1176 get_cpo_flags(void) |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1177 { |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1178 reg_cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1179 reg_cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1180 } |
| 167 | 1181 |
| 1182 /* | |
| 1183 * Skip over a "[]" range. | |
| 1184 * "p" must point to the character after the '['. | |
| 1185 * The returned pointer is on the matching ']', or the terminating NUL. | |
| 1186 */ | |
| 1187 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1188 skip_anyof(char_u *p) |
| 167 | 1189 { |
| 1190 #ifdef FEAT_MBYTE | |
| 1191 int l; | |
| 1192 #endif | |
| 1193 | |
| 1194 if (*p == '^') /* Complement of range. */ | |
| 1195 ++p; | |
| 1196 if (*p == ']' || *p == '-') | |
| 1197 ++p; | |
| 1198 while (*p != NUL && *p != ']') | |
| 1199 { | |
| 1200 #ifdef FEAT_MBYTE | |
| 474 | 1201 if (has_mbyte && (l = (*mb_ptr2len)(p)) > 1) |
| 167 | 1202 p += l; |
| 1203 else | |
| 1204 #endif | |
| 1205 if (*p == '-') | |
| 1206 { | |
| 1207 ++p; | |
| 1208 if (*p != ']' && *p != NUL) | |
| 1209 mb_ptr_adv(p); | |
| 1210 } | |
| 1211 else if (*p == '\\' | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1212 && !reg_cpo_bsl |
| 167 | 1213 && (vim_strchr(REGEXP_INRANGE, p[1]) != NULL |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1214 || (!reg_cpo_lit && vim_strchr(REGEXP_ABBR, p[1]) != NULL))) |
| 167 | 1215 p += 2; |
| 1216 else if (*p == '[') | |
| 1217 { | |
| 1218 if (get_char_class(&p) == CLASS_NONE | |
| 1219 && get_equi_class(&p) == 0 | |
| 6830 | 1220 && get_coll_element(&p) == 0 |
| 1221 && *p != NUL) | |
| 1222 ++p; /* it is not a class name and not NUL */ | |
| 167 | 1223 } |
| 1224 else | |
| 1225 ++p; | |
| 1226 } | |
| 1227 | |
| 1228 return p; | |
| 1229 } | |
| 1230 | |
| 1231 /* | |
| 7 | 1232 * Skip past regular expression. |
| 153 | 1233 * Stop at end of "startp" or where "dirc" is found ('/', '?', etc). |
| 7 | 1234 * Take care of characters with a backslash in front of it. |
| 1235 * Skip strings inside [ and ]. | |
| 1236 * When "newp" is not NULL and "dirc" is '?', make an allocated copy of the | |
| 1237 * expression and change "\?" to "?". If "*newp" is not NULL the expression | |
| 1238 * is changed in-place. | |
| 1239 */ | |
| 1240 char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1241 skip_regexp( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1242 char_u *startp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1243 int dirc, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1244 int magic, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1245 char_u **newp) |
| 7 | 1246 { |
| 1247 int mymagic; | |
| 1248 char_u *p = startp; | |
| 1249 | |
| 1250 if (magic) | |
| 1251 mymagic = MAGIC_ON; | |
| 1252 else | |
| 1253 mymagic = MAGIC_OFF; | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1254 get_cpo_flags(); |
| 7 | 1255 |
| 39 | 1256 for (; p[0] != NUL; mb_ptr_adv(p)) |
| 7 | 1257 { |
| 1258 if (p[0] == dirc) /* found end of regexp */ | |
| 1259 break; | |
| 1260 if ((p[0] == '[' && mymagic >= MAGIC_ON) | |
| 1261 || (p[0] == '\\' && p[1] == '[' && mymagic <= MAGIC_OFF)) | |
| 1262 { | |
| 1263 p = skip_anyof(p + 1); | |
| 1264 if (p[0] == NUL) | |
| 1265 break; | |
| 1266 } | |
| 1267 else if (p[0] == '\\' && p[1] != NUL) | |
| 1268 { | |
| 1269 if (dirc == '?' && newp != NULL && p[1] == '?') | |
| 1270 { | |
| 1271 /* change "\?" to "?", make a copy first. */ | |
| 1272 if (*newp == NULL) | |
| 1273 { | |
| 1274 *newp = vim_strsave(startp); | |
| 1275 if (*newp != NULL) | |
| 1276 p = *newp + (p - startp); | |
| 1277 } | |
| 1278 if (*newp != NULL) | |
| 1621 | 1279 STRMOVE(p, p + 1); |
| 7 | 1280 else |
| 1281 ++p; | |
| 1282 } | |
| 1283 else | |
| 1284 ++p; /* skip next character */ | |
| 1285 if (*p == 'v') | |
| 1286 mymagic = MAGIC_ALL; | |
| 1287 else if (*p == 'V') | |
| 1288 mymagic = MAGIC_NONE; | |
| 1289 } | |
| 1290 } | |
| 1291 return p; | |
| 1292 } | |
| 1293 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
1294 static regprog_T *bt_regcomp(char_u *expr, int re_flags); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
1295 static void bt_regfree(regprog_T *prog); |
| 4444 | 1296 |
| 7 | 1297 /* |
| 4444 | 1298 * bt_regcomp() - compile a regular expression into internal code for the |
| 1299 * traditional back track matcher. | |
| 41 | 1300 * Returns the program in allocated space. Returns NULL for an error. |
| 7 | 1301 * |
| 1302 * We can't allocate space until we know how big the compiled form will be, | |
| 1303 * but we can't compile it (and thus know how big it is) until we've got a | |
| 1304 * place to put the code. So we cheat: we compile it twice, once with code | |
| 1305 * generation turned off and size counting turned on, and once "for real". | |
| 1306 * This also means that we don't allocate space until we are sure that the | |
| 1307 * thing really will compile successfully, and we never have to move the | |
| 1308 * code and thus invalidate pointers into it. (Note that it has to be in | |
| 1309 * one piece because vim_free() must be able to free it all.) | |
| 1310 * | |
| 1311 * Whether upper/lower case is to be ignored is decided when executing the | |
| 1312 * program, it does not matter here. | |
| 1313 * | |
| 1314 * Beware that the optimization-preparation code in here knows about some | |
| 1315 * of the structure of the compiled regexp. | |
| 1316 * "re_flags": RE_MAGIC and/or RE_STRING. | |
| 1317 */ | |
| 4444 | 1318 static regprog_T * |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1319 bt_regcomp(char_u *expr, int re_flags) |
| 7 | 1320 { |
| 4444 | 1321 bt_regprog_T *r; |
| 7 | 1322 char_u *scan; |
| 1323 char_u *longest; | |
| 1324 int len; | |
| 1325 int flags; | |
| 1326 | |
| 1327 if (expr == NULL) | |
| 1328 EMSG_RET_NULL(_(e_null)); | |
| 1329 | |
| 1330 init_class_tab(); | |
| 1331 | |
| 1332 /* | |
| 1333 * First pass: determine size, legality. | |
| 1334 */ | |
| 1335 regcomp_start(expr, re_flags); | |
| 1336 regcode = JUST_CALC_SIZE; | |
| 1337 regc(REGMAGIC); | |
| 1338 if (reg(REG_NOPAREN, &flags) == NULL) | |
| 1339 return NULL; | |
| 1340 | |
| 1341 /* Small enough for pointer-storage convention? */ | |
| 1342 #ifdef SMALL_MALLOC /* 16 bit storage allocation */ | |
| 1343 if (regsize >= 65536L - 256L) | |
| 1344 EMSG_RET_NULL(_("E339: Pattern too long")); | |
| 1345 #endif | |
| 1346 | |
| 1347 /* Allocate space. */ | |
| 4444 | 1348 r = (bt_regprog_T *)lalloc(sizeof(bt_regprog_T) + regsize, TRUE); |
| 7 | 1349 if (r == NULL) |
| 1350 return NULL; | |
| 1351 | |
| 1352 /* | |
| 1353 * Second pass: emit code. | |
| 1354 */ | |
| 1355 regcomp_start(expr, re_flags); | |
| 1356 regcode = r->program; | |
| 1357 regc(REGMAGIC); | |
| 2010 | 1358 if (reg(REG_NOPAREN, &flags) == NULL || reg_toolong) |
| 7 | 1359 { |
| 1360 vim_free(r); | |
| 2010 | 1361 if (reg_toolong) |
| 1362 EMSG_RET_NULL(_("E339: Pattern too long")); | |
| 7 | 1363 return NULL; |
| 1364 } | |
| 1365 | |
| 1366 /* Dig out information for optimizations. */ | |
| 1367 r->regstart = NUL; /* Worst-case defaults. */ | |
| 1368 r->reganch = 0; | |
| 1369 r->regmust = NULL; | |
| 1370 r->regmlen = 0; | |
| 1371 r->regflags = regflags; | |
| 1372 if (flags & HASNL) | |
| 1373 r->regflags |= RF_HASNL; | |
| 1374 if (flags & HASLOOKBH) | |
| 1375 r->regflags |= RF_LOOKBH; | |
| 1376 #ifdef FEAT_SYN_HL | |
| 1377 /* Remember whether this pattern has any \z specials in it. */ | |
| 1378 r->reghasz = re_has_z; | |
| 1379 #endif | |
| 1380 scan = r->program + 1; /* First BRANCH. */ | |
| 1381 if (OP(regnext(scan)) == END) /* Only one top-level choice. */ | |
| 1382 { | |
| 1383 scan = OPERAND(scan); | |
| 1384 | |
| 1385 /* Starting-point info. */ | |
| 1386 if (OP(scan) == BOL || OP(scan) == RE_BOF) | |
| 1387 { | |
| 1388 r->reganch++; | |
| 1389 scan = regnext(scan); | |
| 1390 } | |
| 1391 | |
| 1392 if (OP(scan) == EXACTLY) | |
| 1393 { | |
| 1394 #ifdef FEAT_MBYTE | |
| 1395 if (has_mbyte) | |
| 1396 r->regstart = (*mb_ptr2char)(OPERAND(scan)); | |
| 1397 else | |
| 1398 #endif | |
| 1399 r->regstart = *OPERAND(scan); | |
| 1400 } | |
| 1401 else if ((OP(scan) == BOW | |
| 1402 || OP(scan) == EOW | |
| 1403 || OP(scan) == NOTHING | |
| 1404 || OP(scan) == MOPEN + 0 || OP(scan) == NOPEN | |
| 1405 || OP(scan) == MCLOSE + 0 || OP(scan) == NCLOSE) | |
| 1406 && OP(regnext(scan)) == EXACTLY) | |
| 1407 { | |
| 1408 #ifdef FEAT_MBYTE | |
| 1409 if (has_mbyte) | |
| 1410 r->regstart = (*mb_ptr2char)(OPERAND(regnext(scan))); | |
| 1411 else | |
| 1412 #endif | |
| 1413 r->regstart = *OPERAND(regnext(scan)); | |
| 1414 } | |
| 1415 | |
| 1416 /* | |
| 1417 * If there's something expensive in the r.e., find the longest | |
| 1418 * literal string that must appear and make it the regmust. Resolve | |
| 1419 * ties in favor of later strings, since the regstart check works | |
| 1420 * with the beginning of the r.e. and avoiding duplication | |
| 1421 * strengthens checking. Not a strong reason, but sufficient in the | |
| 1422 * absence of others. | |
| 1423 */ | |
| 1424 /* | |
| 1425 * When the r.e. starts with BOW, it is faster to look for a regmust | |
| 1426 * first. Used a lot for "#" and "*" commands. (Added by mool). | |
| 1427 */ | |
| 1428 if ((flags & SPSTART || OP(scan) == BOW || OP(scan) == EOW) | |
| 1429 && !(flags & HASNL)) | |
| 1430 { | |
| 1431 longest = NULL; | |
| 1432 len = 0; | |
| 1433 for (; scan != NULL; scan = regnext(scan)) | |
| 1434 if (OP(scan) == EXACTLY && STRLEN(OPERAND(scan)) >= (size_t)len) | |
| 1435 { | |
| 1436 longest = OPERAND(scan); | |
| 1437 len = (int)STRLEN(OPERAND(scan)); | |
| 1438 } | |
| 1439 r->regmust = longest; | |
| 1440 r->regmlen = len; | |
| 1441 } | |
| 1442 } | |
| 4444 | 1443 #ifdef BT_REGEXP_DUMP |
| 7 | 1444 regdump(expr, r); |
| 1445 #endif | |
| 4444 | 1446 r->engine = &bt_regengine; |
| 1447 return (regprog_T *)r; | |
| 7 | 1448 } |
| 1449 | |
| 1450 /* | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1451 * Free a compiled regexp program, returned by bt_regcomp(). |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1452 */ |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1453 static void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1454 bt_regfree(regprog_T *prog) |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1455 { |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1456 vim_free(prog); |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1457 } |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1458 |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1459 /* |
| 7 | 1460 * Setup to parse the regexp. Used once to get the length and once to do it. |
| 1461 */ | |
| 1462 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1463 regcomp_start( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1464 char_u *expr, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1465 int re_flags) /* see vim_regcomp() */ |
| 7 | 1466 { |
| 1467 initchr(expr); | |
| 1468 if (re_flags & RE_MAGIC) | |
| 1469 reg_magic = MAGIC_ON; | |
| 1470 else | |
| 1471 reg_magic = MAGIC_OFF; | |
| 1472 reg_string = (re_flags & RE_STRING); | |
| 481 | 1473 reg_strict = (re_flags & RE_STRICT); |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1474 get_cpo_flags(); |
| 7 | 1475 |
| 1476 num_complex_braces = 0; | |
| 1477 regnpar = 1; | |
| 1478 vim_memset(had_endbrace, 0, sizeof(had_endbrace)); | |
| 1479 #ifdef FEAT_SYN_HL | |
| 1480 regnzpar = 1; | |
| 1481 re_has_z = 0; | |
| 1482 #endif | |
| 1483 regsize = 0L; | |
| 2010 | 1484 reg_toolong = FALSE; |
| 7 | 1485 regflags = 0; |
| 1486 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1487 had_eol = FALSE; | |
| 1488 #endif | |
| 1489 } | |
| 1490 | |
| 1491 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1492 /* | |
| 1493 * Check if during the previous call to vim_regcomp the EOL item "$" has been | |
| 1494 * found. This is messy, but it works fine. | |
| 1495 */ | |
| 1496 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1497 vim_regcomp_had_eol(void) |
| 7 | 1498 { |
| 1499 return had_eol; | |
| 1500 } | |
| 1501 #endif | |
| 1502 | |
| 1503 /* | |
| 4444 | 1504 * Parse regular expression, i.e. main body or parenthesized thing. |
| 7 | 1505 * |
| 1506 * Caller must absorb opening parenthesis. | |
| 1507 * | |
| 1508 * Combining parenthesis handling with the base level of regular expression | |
| 1509 * is a trifle forced, but the need to tie the tails of the branches to what | |
| 1510 * follows makes it hard to avoid. | |
| 1511 */ | |
| 1512 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1513 reg( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1514 int paren, /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1515 int *flagp) |
| 7 | 1516 { |
| 1517 char_u *ret; | |
| 1518 char_u *br; | |
| 1519 char_u *ender; | |
| 1520 int parno = 0; | |
| 1521 int flags; | |
| 1522 | |
| 1523 *flagp = HASWIDTH; /* Tentatively. */ | |
| 1524 | |
| 1525 #ifdef FEAT_SYN_HL | |
| 1526 if (paren == REG_ZPAREN) | |
| 1527 { | |
| 1528 /* Make a ZOPEN node. */ | |
| 1529 if (regnzpar >= NSUBEXP) | |
| 1530 EMSG_RET_NULL(_("E50: Too many \\z(")); | |
| 1531 parno = regnzpar; | |
| 1532 regnzpar++; | |
| 1533 ret = regnode(ZOPEN + parno); | |
| 1534 } | |
| 1535 else | |
| 1536 #endif | |
| 1537 if (paren == REG_PAREN) | |
| 1538 { | |
| 1539 /* Make a MOPEN node. */ | |
| 1540 if (regnpar >= NSUBEXP) | |
| 4444 | 1541 EMSG2_RET_NULL(_("E51: Too many %s("), reg_magic == MAGIC_ALL); |
| 7 | 1542 parno = regnpar; |
| 1543 ++regnpar; | |
| 1544 ret = regnode(MOPEN + parno); | |
| 1545 } | |
| 1546 else if (paren == REG_NPAREN) | |
| 1547 { | |
| 1548 /* Make a NOPEN node. */ | |
| 1549 ret = regnode(NOPEN); | |
| 1550 } | |
| 1551 else | |
| 1552 ret = NULL; | |
| 1553 | |
| 1554 /* Pick up the branches, linking them together. */ | |
| 1555 br = regbranch(&flags); | |
| 1556 if (br == NULL) | |
| 1557 return NULL; | |
| 1558 if (ret != NULL) | |
| 1559 regtail(ret, br); /* [MZ]OPEN -> first. */ | |
| 1560 else | |
| 1561 ret = br; | |
| 1562 /* If one of the branches can be zero-width, the whole thing can. | |
| 1563 * If one of the branches has * at start or matches a line-break, the | |
| 1564 * whole thing can. */ | |
| 1565 if (!(flags & HASWIDTH)) | |
| 1566 *flagp &= ~HASWIDTH; | |
| 1567 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1568 while (peekchr() == Magic('|')) | |
| 1569 { | |
| 1570 skipchr(); | |
| 1571 br = regbranch(&flags); | |
| 2010 | 1572 if (br == NULL || reg_toolong) |
| 7 | 1573 return NULL; |
| 1574 regtail(ret, br); /* BRANCH -> BRANCH. */ | |
| 1575 if (!(flags & HASWIDTH)) | |
| 1576 *flagp &= ~HASWIDTH; | |
| 1577 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1578 } | |
| 1579 | |
| 1580 /* Make a closing node, and hook it on the end. */ | |
| 1581 ender = regnode( | |
| 1582 #ifdef FEAT_SYN_HL | |
| 1583 paren == REG_ZPAREN ? ZCLOSE + parno : | |
| 1584 #endif | |
| 1585 paren == REG_PAREN ? MCLOSE + parno : | |
| 1586 paren == REG_NPAREN ? NCLOSE : END); | |
| 1587 regtail(ret, ender); | |
| 1588 | |
| 1589 /* Hook the tails of the branches to the closing node. */ | |
| 1590 for (br = ret; br != NULL; br = regnext(br)) | |
| 1591 regoptail(br, ender); | |
| 1592 | |
| 1593 /* Check for proper termination. */ | |
| 1594 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
| 1595 { | |
| 1596 #ifdef FEAT_SYN_HL | |
| 1597 if (paren == REG_ZPAREN) | |
| 308 | 1598 EMSG_RET_NULL(_("E52: Unmatched \\z(")); |
| 7 | 1599 else |
| 1600 #endif | |
| 1601 if (paren == REG_NPAREN) | |
| 4444 | 1602 EMSG2_RET_NULL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); |
| 7 | 1603 else |
| 4444 | 1604 EMSG2_RET_NULL(_(e_unmatchedp), reg_magic == MAGIC_ALL); |
| 7 | 1605 } |
| 1606 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
| 1607 { | |
| 1608 if (curchr == Magic(')')) | |
| 4444 | 1609 EMSG2_RET_NULL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); |
| 7 | 1610 else |
| 308 | 1611 EMSG_RET_NULL(_(e_trailing)); /* "Can't happen". */ |
| 7 | 1612 /* NOTREACHED */ |
| 1613 } | |
| 1614 /* | |
| 1615 * Here we set the flag allowing back references to this set of | |
| 1616 * parentheses. | |
| 1617 */ | |
| 1618 if (paren == REG_PAREN) | |
| 1619 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
| 1620 return ret; | |
| 1621 } | |
| 1622 | |
| 1623 /* | |
| 4444 | 1624 * Parse one alternative of an | operator. |
| 7 | 1625 * Implements the & operator. |
| 1626 */ | |
| 1627 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1628 regbranch(int *flagp) |
| 7 | 1629 { |
| 1630 char_u *ret; | |
| 1631 char_u *chain = NULL; | |
| 1632 char_u *latest; | |
| 1633 int flags; | |
| 1634 | |
| 1635 *flagp = WORST | HASNL; /* Tentatively. */ | |
| 1636 | |
| 1637 ret = regnode(BRANCH); | |
| 1638 for (;;) | |
| 1639 { | |
| 1640 latest = regconcat(&flags); | |
| 1641 if (latest == NULL) | |
| 1642 return NULL; | |
| 1643 /* If one of the branches has width, the whole thing has. If one of | |
| 1644 * the branches anchors at start-of-line, the whole thing does. | |
| 1645 * If one of the branches uses look-behind, the whole thing does. */ | |
| 1646 *flagp |= flags & (HASWIDTH | SPSTART | HASLOOKBH); | |
| 1647 /* If one of the branches doesn't match a line-break, the whole thing | |
| 1648 * doesn't. */ | |
| 1649 *flagp &= ~HASNL | (flags & HASNL); | |
| 1650 if (chain != NULL) | |
| 1651 regtail(chain, latest); | |
| 1652 if (peekchr() != Magic('&')) | |
| 1653 break; | |
| 1654 skipchr(); | |
| 1655 regtail(latest, regnode(END)); /* operand ends */ | |
| 2010 | 1656 if (reg_toolong) |
| 1657 break; | |
| 7 | 1658 reginsert(MATCH, latest); |
| 1659 chain = latest; | |
| 1660 } | |
| 1661 | |
| 1662 return ret; | |
| 1663 } | |
| 1664 | |
| 1665 /* | |
| 4444 | 1666 * Parse one alternative of an | or & operator. |
| 7 | 1667 * Implements the concatenation operator. |
| 1668 */ | |
| 1669 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1670 regconcat(int *flagp) |
| 7 | 1671 { |
| 1672 char_u *first = NULL; | |
| 1673 char_u *chain = NULL; | |
| 1674 char_u *latest; | |
| 1675 int flags; | |
| 1676 int cont = TRUE; | |
| 1677 | |
| 1678 *flagp = WORST; /* Tentatively. */ | |
| 1679 | |
| 1680 while (cont) | |
| 1681 { | |
| 1682 switch (peekchr()) | |
| 1683 { | |
| 1684 case NUL: | |
| 1685 case Magic('|'): | |
| 1686 case Magic('&'): | |
| 1687 case Magic(')'): | |
| 1688 cont = FALSE; | |
| 1689 break; | |
| 1690 case Magic('Z'): | |
| 1691 #ifdef FEAT_MBYTE | |
| 1692 regflags |= RF_ICOMBINE; | |
| 1693 #endif | |
| 1694 skipchr_keepstart(); | |
| 1695 break; | |
| 1696 case Magic('c'): | |
| 1697 regflags |= RF_ICASE; | |
| 1698 skipchr_keepstart(); | |
| 1699 break; | |
| 1700 case Magic('C'): | |
| 1701 regflags |= RF_NOICASE; | |
| 1702 skipchr_keepstart(); | |
| 1703 break; | |
| 1704 case Magic('v'): | |
| 1705 reg_magic = MAGIC_ALL; | |
| 1706 skipchr_keepstart(); | |
| 1707 curchr = -1; | |
| 1708 break; | |
| 1709 case Magic('m'): | |
| 1710 reg_magic = MAGIC_ON; | |
| 1711 skipchr_keepstart(); | |
| 1712 curchr = -1; | |
| 1713 break; | |
| 1714 case Magic('M'): | |
| 1715 reg_magic = MAGIC_OFF; | |
| 1716 skipchr_keepstart(); | |
| 1717 curchr = -1; | |
| 1718 break; | |
| 1719 case Magic('V'): | |
| 1720 reg_magic = MAGIC_NONE; | |
| 1721 skipchr_keepstart(); | |
| 1722 curchr = -1; | |
| 1723 break; | |
| 1724 default: | |
| 1725 latest = regpiece(&flags); | |
| 2010 | 1726 if (latest == NULL || reg_toolong) |
| 7 | 1727 return NULL; |
| 1728 *flagp |= flags & (HASWIDTH | HASNL | HASLOOKBH); | |
| 1729 if (chain == NULL) /* First piece. */ | |
| 1730 *flagp |= flags & SPSTART; | |
| 1731 else | |
| 1732 regtail(chain, latest); | |
| 1733 chain = latest; | |
| 1734 if (first == NULL) | |
| 1735 first = latest; | |
| 1736 break; | |
| 1737 } | |
| 1738 } | |
| 1739 if (first == NULL) /* Loop ran zero times. */ | |
| 1740 first = regnode(NOTHING); | |
| 1741 return first; | |
| 1742 } | |
| 1743 | |
| 1744 /* | |
| 4444 | 1745 * Parse something followed by possible [*+=]. |
| 7 | 1746 * |
| 1747 * Note that the branching code sequences used for = and the general cases | |
| 1748 * of * and + are somewhat optimized: they use the same NOTHING node as | |
| 1749 * both the endmarker for their branch list and the body of the last branch. | |
| 1750 * It might seem that this node could be dispensed with entirely, but the | |
| 1751 * endmarker role is not redundant. | |
| 1752 */ | |
| 1753 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1754 regpiece(int *flagp) |
| 7 | 1755 { |
| 1756 char_u *ret; | |
| 1757 int op; | |
| 1758 char_u *next; | |
| 1759 int flags; | |
| 1760 long minval; | |
| 1761 long maxval; | |
| 1762 | |
| 1763 ret = regatom(&flags); | |
| 1764 if (ret == NULL) | |
| 1765 return NULL; | |
| 1766 | |
| 1767 op = peekchr(); | |
| 1768 if (re_multi_type(op) == NOT_MULTI) | |
| 1769 { | |
| 1770 *flagp = flags; | |
| 1771 return ret; | |
| 1772 } | |
| 1773 /* default flags */ | |
| 1774 *flagp = (WORST | SPSTART | (flags & (HASNL | HASLOOKBH))); | |
| 1775 | |
| 1776 skipchr(); | |
| 1777 switch (op) | |
| 1778 { | |
| 1779 case Magic('*'): | |
| 1780 if (flags & SIMPLE) | |
| 1781 reginsert(STAR, ret); | |
| 1782 else | |
| 1783 { | |
| 1784 /* Emit x* as (x&|), where & means "self". */ | |
| 1785 reginsert(BRANCH, ret); /* Either x */ | |
| 1786 regoptail(ret, regnode(BACK)); /* and loop */ | |
| 1787 regoptail(ret, ret); /* back */ | |
| 1788 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1789 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1790 } | |
| 1791 break; | |
| 1792 | |
| 1793 case Magic('+'): | |
| 1794 if (flags & SIMPLE) | |
| 1795 reginsert(PLUS, ret); | |
| 1796 else | |
| 1797 { | |
| 1798 /* Emit x+ as x(&|), where & means "self". */ | |
| 1799 next = regnode(BRANCH); /* Either */ | |
| 1800 regtail(ret, next); | |
| 233 | 1801 regtail(regnode(BACK), ret); /* loop back */ |
| 7 | 1802 regtail(next, regnode(BRANCH)); /* or */ |
| 1803 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1804 } | |
| 1805 *flagp = (WORST | HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1806 break; | |
| 1807 | |
| 1808 case Magic('@'): | |
| 1809 { | |
| 1810 int lop = END; | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1811 int nr; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1812 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1813 nr = getdecchrs(); |
| 7 | 1814 switch (no_Magic(getchr())) |
| 1815 { | |
| 1816 case '=': lop = MATCH; break; /* \@= */ | |
| 1817 case '!': lop = NOMATCH; break; /* \@! */ | |
| 1818 case '>': lop = SUBPAT; break; /* \@> */ | |
| 1819 case '<': switch (no_Magic(getchr())) | |
| 1820 { | |
| 1821 case '=': lop = BEHIND; break; /* \@<= */ | |
| 1822 case '!': lop = NOBEHIND; break; /* \@<! */ | |
| 1823 } | |
| 1824 } | |
| 1825 if (lop == END) | |
| 4444 | 1826 EMSG2_RET_NULL(_("E59: invalid character after %s@"), |
| 7 | 1827 reg_magic == MAGIC_ALL); |
| 1828 /* Look behind must match with behind_pos. */ | |
| 1829 if (lop == BEHIND || lop == NOBEHIND) | |
| 1830 { | |
| 1831 regtail(ret, regnode(BHPOS)); | |
| 1832 *flagp |= HASLOOKBH; | |
| 1833 } | |
| 1834 regtail(ret, regnode(END)); /* operand ends */ | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1835 if (lop == BEHIND || lop == NOBEHIND) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1836 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1837 if (nr < 0) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1838 nr = 0; /* no limit is same as zero limit */ |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1839 reginsert_nr(lop, nr, ret); |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1840 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1841 else |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1842 reginsert(lop, ret); |
| 7 | 1843 break; |
| 1844 } | |
| 1845 | |
| 1846 case Magic('?'): | |
| 1847 case Magic('='): | |
| 1848 /* Emit x= as (x|) */ | |
| 1849 reginsert(BRANCH, ret); /* Either x */ | |
| 1850 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1851 next = regnode(NOTHING); /* null. */ | |
| 1852 regtail(ret, next); | |
| 1853 regoptail(ret, next); | |
| 1854 break; | |
| 1855 | |
| 1856 case Magic('{'): | |
| 1857 if (!read_limits(&minval, &maxval)) | |
| 1858 return NULL; | |
| 1859 if (flags & SIMPLE) | |
| 1860 { | |
| 1861 reginsert(BRACE_SIMPLE, ret); | |
| 1862 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1863 } | |
| 1864 else | |
| 1865 { | |
| 1866 if (num_complex_braces >= 10) | |
| 4444 | 1867 EMSG2_RET_NULL(_("E60: Too many complex %s{...}s"), |
| 7 | 1868 reg_magic == MAGIC_ALL); |
| 1869 reginsert(BRACE_COMPLEX + num_complex_braces, ret); | |
| 1870 regoptail(ret, regnode(BACK)); | |
| 1871 regoptail(ret, ret); | |
| 1872 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1873 ++num_complex_braces; | |
| 1874 } | |
| 1875 if (minval > 0 && maxval > 0) | |
| 1876 *flagp = (HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1877 break; | |
| 1878 } | |
| 1879 if (re_multi_type(peekchr()) != NOT_MULTI) | |
| 1880 { | |
| 1881 /* Can't have a multi follow a multi. */ | |
| 1882 if (peekchr() == Magic('*')) | |
| 1883 sprintf((char *)IObuff, _("E61: Nested %s*"), | |
| 1884 reg_magic >= MAGIC_ON ? "" : "\\"); | |
| 1885 else | |
| 1886 sprintf((char *)IObuff, _("E62: Nested %s%c"), | |
| 1887 reg_magic == MAGIC_ALL ? "" : "\\", no_Magic(peekchr())); | |
| 1888 EMSG_RET_NULL(IObuff); | |
| 1889 } | |
| 1890 | |
| 1891 return ret; | |
| 1892 } | |
| 1893 | |
| 4444 | 1894 /* When making changes to classchars also change nfa_classcodes. */ |
| 1895 static char_u *classchars = (char_u *)".iIkKfFpPsSdDxXoOwWhHaAlLuU"; | |
| 1896 static int classcodes[] = { | |
| 1897 ANY, IDENT, SIDENT, KWORD, SKWORD, | |
| 1898 FNAME, SFNAME, PRINT, SPRINT, | |
| 1899 WHITE, NWHITE, DIGIT, NDIGIT, | |
| 1900 HEX, NHEX, OCTAL, NOCTAL, | |
| 1901 WORD, NWORD, HEAD, NHEAD, | |
| 1902 ALPHA, NALPHA, LOWER, NLOWER, | |
| 1903 UPPER, NUPPER | |
| 1904 }; | |
| 1905 | |
| 7 | 1906 /* |
| 4444 | 1907 * Parse the lowest level. |
| 7 | 1908 * |
| 1909 * Optimization: gobbles an entire sequence of ordinary characters so that | |
| 1910 * it can turn them into a single node, which is smaller to store and | |
| 1911 * faster to run. Don't do this when one_exactly is set. | |
| 1912 */ | |
| 1913 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1914 regatom(int *flagp) |
| 7 | 1915 { |
| 1916 char_u *ret; | |
| 1917 int flags; | |
| 1918 int c; | |
| 1919 char_u *p; | |
| 1920 int extra = 0; | |
| 1921 | |
| 1922 *flagp = WORST; /* Tentatively. */ | |
| 1923 | |
| 1924 c = getchr(); | |
| 1925 switch (c) | |
| 1926 { | |
| 1927 case Magic('^'): | |
| 1928 ret = regnode(BOL); | |
| 1929 break; | |
| 1930 | |
| 1931 case Magic('$'): | |
| 1932 ret = regnode(EOL); | |
| 1933 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1934 had_eol = TRUE; | |
| 1935 #endif | |
| 1936 break; | |
| 1937 | |
| 1938 case Magic('<'): | |
| 1939 ret = regnode(BOW); | |
| 1940 break; | |
| 1941 | |
| 1942 case Magic('>'): | |
| 1943 ret = regnode(EOW); | |
| 1944 break; | |
| 1945 | |
| 1946 case Magic('_'): | |
| 1947 c = no_Magic(getchr()); | |
| 1948 if (c == '^') /* "\_^" is start-of-line */ | |
| 1949 { | |
| 1950 ret = regnode(BOL); | |
| 1951 break; | |
| 1952 } | |
| 1953 if (c == '$') /* "\_$" is end-of-line */ | |
| 1954 { | |
| 1955 ret = regnode(EOL); | |
| 1956 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1957 had_eol = TRUE; | |
| 1958 #endif | |
| 1959 break; | |
| 1960 } | |
| 1961 | |
| 1962 extra = ADD_NL; | |
| 1963 *flagp |= HASNL; | |
| 1964 | |
| 1965 /* "\_[" is character range plus newline */ | |
| 1966 if (c == '[') | |
| 1967 goto collection; | |
| 1968 | |
| 1969 /* "\_x" is character class plus newline */ | |
| 1970 /*FALLTHROUGH*/ | |
| 1971 | |
| 1972 /* | |
| 1973 * Character classes. | |
| 1974 */ | |
| 1975 case Magic('.'): | |
| 1976 case Magic('i'): | |
| 1977 case Magic('I'): | |
| 1978 case Magic('k'): | |
| 1979 case Magic('K'): | |
| 1980 case Magic('f'): | |
| 1981 case Magic('F'): | |
| 1982 case Magic('p'): | |
| 1983 case Magic('P'): | |
| 1984 case Magic('s'): | |
| 1985 case Magic('S'): | |
| 1986 case Magic('d'): | |
| 1987 case Magic('D'): | |
| 1988 case Magic('x'): | |
| 1989 case Magic('X'): | |
| 1990 case Magic('o'): | |
| 1991 case Magic('O'): | |
| 1992 case Magic('w'): | |
| 1993 case Magic('W'): | |
| 1994 case Magic('h'): | |
| 1995 case Magic('H'): | |
| 1996 case Magic('a'): | |
| 1997 case Magic('A'): | |
| 1998 case Magic('l'): | |
| 1999 case Magic('L'): | |
| 2000 case Magic('u'): | |
| 2001 case Magic('U'): | |
| 2002 p = vim_strchr(classchars, no_Magic(c)); | |
| 2003 if (p == NULL) | |
| 2004 EMSG_RET_NULL(_("E63: invalid use of \\_")); | |
| 714 | 2005 #ifdef FEAT_MBYTE |
| 2006 /* When '.' is followed by a composing char ignore the dot, so that | |
| 2007 * the composing char is matched here. */ | |
| 2008 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
| 2009 { | |
| 2010 c = getchr(); | |
| 2011 goto do_multibyte; | |
| 2012 } | |
| 2013 #endif | |
| 7 | 2014 ret = regnode(classcodes[p - classchars] + extra); |
| 2015 *flagp |= HASWIDTH | SIMPLE; | |
| 2016 break; | |
| 2017 | |
| 2018 case Magic('n'): | |
| 2019 if (reg_string) | |
| 2020 { | |
| 2021 /* In a string "\n" matches a newline character. */ | |
| 2022 ret = regnode(EXACTLY); | |
| 2023 regc(NL); | |
| 2024 regc(NUL); | |
| 2025 *flagp |= HASWIDTH | SIMPLE; | |
| 2026 } | |
| 2027 else | |
| 2028 { | |
| 2029 /* In buffer text "\n" matches the end of a line. */ | |
| 2030 ret = regnode(NEWL); | |
| 2031 *flagp |= HASWIDTH | HASNL; | |
| 2032 } | |
| 2033 break; | |
| 2034 | |
| 2035 case Magic('('): | |
| 2036 if (one_exactly) | |
| 2037 EMSG_ONE_RET_NULL; | |
| 2038 ret = reg(REG_PAREN, &flags); | |
| 2039 if (ret == NULL) | |
| 2040 return NULL; | |
| 2041 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 2042 break; | |
| 2043 | |
| 2044 case NUL: | |
| 2045 case Magic('|'): | |
| 2046 case Magic('&'): | |
| 2047 case Magic(')'): | |
| 1468 | 2048 if (one_exactly) |
| 2049 EMSG_ONE_RET_NULL; | |
| 7 | 2050 EMSG_RET_NULL(_(e_internal)); /* Supposed to be caught earlier. */ |
| 2051 /* NOTREACHED */ | |
| 2052 | |
| 2053 case Magic('='): | |
| 2054 case Magic('?'): | |
| 2055 case Magic('+'): | |
| 2056 case Magic('@'): | |
| 2057 case Magic('{'): | |
| 2058 case Magic('*'): | |
| 2059 c = no_Magic(c); | |
| 2060 sprintf((char *)IObuff, _("E64: %s%c follows nothing"), | |
| 2061 (c == '*' ? reg_magic >= MAGIC_ON : reg_magic == MAGIC_ALL) | |
| 2062 ? "" : "\\", c); | |
| 2063 EMSG_RET_NULL(IObuff); | |
| 2064 /* NOTREACHED */ | |
| 2065 | |
| 2066 case Magic('~'): /* previous substitute pattern */ | |
| 359 | 2067 if (reg_prev_sub != NULL) |
| 7 | 2068 { |
| 2069 char_u *lp; | |
| 2070 | |
| 2071 ret = regnode(EXACTLY); | |
| 2072 lp = reg_prev_sub; | |
| 2073 while (*lp != NUL) | |
| 2074 regc(*lp++); | |
| 2075 regc(NUL); | |
| 2076 if (*reg_prev_sub != NUL) | |
| 2077 { | |
| 2078 *flagp |= HASWIDTH; | |
| 2079 if ((lp - reg_prev_sub) == 1) | |
| 2080 *flagp |= SIMPLE; | |
| 2081 } | |
| 2082 } | |
| 2083 else | |
| 2084 EMSG_RET_NULL(_(e_nopresub)); | |
| 2085 break; | |
| 2086 | |
| 2087 case Magic('1'): | |
| 2088 case Magic('2'): | |
| 2089 case Magic('3'): | |
| 2090 case Magic('4'): | |
| 2091 case Magic('5'): | |
| 2092 case Magic('6'): | |
| 2093 case Magic('7'): | |
| 2094 case Magic('8'): | |
| 2095 case Magic('9'): | |
| 2096 { | |
| 2097 int refnum; | |
| 2098 | |
| 2099 refnum = c - Magic('0'); | |
| 2100 /* | |
| 2101 * Check if the back reference is legal. We must have seen the | |
| 2102 * close brace. | |
| 2103 * TODO: Should also check that we don't refer to something | |
| 2104 * that is repeated (+*=): what instance of the repetition | |
| 2105 * should we match? | |
| 2106 */ | |
| 2107 if (!had_endbrace[refnum]) | |
| 2108 { | |
| 2109 /* Trick: check if "@<=" or "@<!" follows, in which case | |
| 2110 * the \1 can appear before the referenced match. */ | |
| 2111 for (p = regparse; *p != NUL; ++p) | |
| 2112 if (p[0] == '@' && p[1] == '<' | |
| 2113 && (p[2] == '!' || p[2] == '=')) | |
| 2114 break; | |
| 2115 if (*p == NUL) | |
| 2116 EMSG_RET_NULL(_("E65: Illegal back reference")); | |
| 2117 } | |
| 2118 ret = regnode(BACKREF + refnum); | |
| 2119 } | |
| 2120 break; | |
| 2121 | |
| 2122 case Magic('z'): | |
| 2123 { | |
| 2124 c = no_Magic(getchr()); | |
| 2125 switch (c) | |
| 2126 { | |
| 741 | 2127 #ifdef FEAT_SYN_HL |
| 7 | 2128 case '(': if (reg_do_extmatch != REX_SET) |
|
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
2129 EMSG_RET_NULL(_(e_z_not_allowed)); |
| 7 | 2130 if (one_exactly) |
| 2131 EMSG_ONE_RET_NULL; | |
| 2132 ret = reg(REG_ZPAREN, &flags); | |
| 2133 if (ret == NULL) | |
| 2134 return NULL; | |
| 2135 *flagp |= flags & (HASWIDTH|SPSTART|HASNL|HASLOOKBH); | |
| 2136 re_has_z = REX_SET; | |
| 2137 break; | |
| 2138 | |
| 2139 case '1': | |
| 2140 case '2': | |
| 2141 case '3': | |
| 2142 case '4': | |
| 2143 case '5': | |
| 2144 case '6': | |
| 2145 case '7': | |
| 2146 case '8': | |
| 2147 case '9': if (reg_do_extmatch != REX_USE) | |
|
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
2148 EMSG_RET_NULL(_(e_z1_not_allowed)); |
| 7 | 2149 ret = regnode(ZREF + c - '0'); |
| 2150 re_has_z = REX_USE; | |
| 2151 break; | |
| 741 | 2152 #endif |
| 7 | 2153 |
| 2154 case 's': ret = regnode(MOPEN + 0); | |
| 6203 | 2155 if (re_mult_next("\\zs") == FAIL) |
| 2156 return NULL; | |
| 7 | 2157 break; |
| 2158 | |
| 2159 case 'e': ret = regnode(MCLOSE + 0); | |
| 6203 | 2160 if (re_mult_next("\\ze") == FAIL) |
| 2161 return NULL; | |
| 7 | 2162 break; |
| 2163 | |
| 2164 default: EMSG_RET_NULL(_("E68: Invalid character after \\z")); | |
| 2165 } | |
| 2166 } | |
| 2167 break; | |
| 2168 | |
| 2169 case Magic('%'): | |
| 2170 { | |
| 2171 c = no_Magic(getchr()); | |
| 2172 switch (c) | |
| 2173 { | |
| 2174 /* () without a back reference */ | |
| 2175 case '(': | |
| 2176 if (one_exactly) | |
| 2177 EMSG_ONE_RET_NULL; | |
| 2178 ret = reg(REG_NPAREN, &flags); | |
| 2179 if (ret == NULL) | |
| 2180 return NULL; | |
| 2181 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 2182 break; | |
| 2183 | |
| 2184 /* Catch \%^ and \%$ regardless of where they appear in the | |
| 2185 * pattern -- regardless of whether or not it makes sense. */ | |
| 2186 case '^': | |
| 2187 ret = regnode(RE_BOF); | |
| 2188 break; | |
| 2189 | |
| 2190 case '$': | |
| 2191 ret = regnode(RE_EOF); | |
| 2192 break; | |
| 2193 | |
| 2194 case '#': | |
| 2195 ret = regnode(CURSOR); | |
| 2196 break; | |
| 2197 | |
| 639 | 2198 case 'V': |
| 2199 ret = regnode(RE_VISUAL); | |
| 2200 break; | |
| 2201 | |
| 5901 | 2202 case 'C': |
| 2203 ret = regnode(RE_COMPOSING); | |
| 2204 break; | |
| 2205 | |
| 7 | 2206 /* \%[abc]: Emit as a list of branches, all ending at the last |
| 2207 * branch which matches nothing. */ | |
| 2208 case '[': | |
| 2209 if (one_exactly) /* doesn't nest */ | |
| 2210 EMSG_ONE_RET_NULL; | |
| 2211 { | |
| 2212 char_u *lastbranch; | |
| 2213 char_u *lastnode = NULL; | |
| 2214 char_u *br; | |
| 2215 | |
| 2216 ret = NULL; | |
| 2217 while ((c = getchr()) != ']') | |
| 2218 { | |
| 2219 if (c == NUL) | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2220 EMSG2_RET_NULL(_(e_missing_sb), |
| 7 | 2221 reg_magic == MAGIC_ALL); |
| 2222 br = regnode(BRANCH); | |
| 2223 if (ret == NULL) | |
| 2224 ret = br; | |
| 2225 else | |
| 2226 regtail(lastnode, br); | |
| 2227 | |
| 2228 ungetchr(); | |
| 2229 one_exactly = TRUE; | |
| 2230 lastnode = regatom(flagp); | |
| 2231 one_exactly = FALSE; | |
| 2232 if (lastnode == NULL) | |
| 2233 return NULL; | |
| 2234 } | |
| 2235 if (ret == NULL) | |
|
4760
532a9855bd30
updated for version 7.3.1127
Bram Moolenaar <bram@vim.org>
parents:
4746
diff
changeset
|
2236 EMSG2_RET_NULL(_(e_empty_sb), |
| 7 | 2237 reg_magic == MAGIC_ALL); |
| 2238 lastbranch = regnode(BRANCH); | |
| 2239 br = regnode(NOTHING); | |
| 2240 if (ret != JUST_CALC_SIZE) | |
| 2241 { | |
| 2242 regtail(lastnode, br); | |
| 2243 regtail(lastbranch, br); | |
| 2244 /* connect all branches to the NOTHING | |
| 2245 * branch at the end */ | |
| 2246 for (br = ret; br != lastnode; ) | |
| 2247 { | |
| 2248 if (OP(br) == BRANCH) | |
| 2249 { | |
| 2250 regtail(br, lastbranch); | |
| 2251 br = OPERAND(br); | |
| 2252 } | |
| 2253 else | |
| 2254 br = regnext(br); | |
| 2255 } | |
| 2256 } | |
| 1701 | 2257 *flagp &= ~(HASWIDTH | SIMPLE); |
| 7 | 2258 break; |
| 2259 } | |
| 2260 | |
| 24 | 2261 case 'd': /* %d123 decimal */ |
| 2262 case 'o': /* %o123 octal */ | |
| 2263 case 'x': /* %xab hex 2 */ | |
| 2264 case 'u': /* %uabcd hex 4 */ | |
| 2265 case 'U': /* %U1234abcd hex 8 */ | |
| 2266 { | |
| 2267 int i; | |
| 2268 | |
| 2269 switch (c) | |
| 2270 { | |
| 2271 case 'd': i = getdecchrs(); break; | |
| 2272 case 'o': i = getoctchrs(); break; | |
| 2273 case 'x': i = gethexchrs(2); break; | |
| 2274 case 'u': i = gethexchrs(4); break; | |
| 2275 case 'U': i = gethexchrs(8); break; | |
| 2276 default: i = -1; break; | |
| 2277 } | |
| 2278 | |
| 2279 if (i < 0) | |
| 4444 | 2280 EMSG2_RET_NULL( |
| 24 | 2281 _("E678: Invalid character after %s%%[dxouU]"), |
| 2282 reg_magic == MAGIC_ALL); | |
| 714 | 2283 #ifdef FEAT_MBYTE |
| 2284 if (use_multibytecode(i)) | |
| 2285 ret = regnode(MULTIBYTECODE); | |
| 2286 else | |
| 2287 #endif | |
| 2288 ret = regnode(EXACTLY); | |
| 24 | 2289 if (i == 0) |
| 2290 regc(0x0a); | |
| 2291 else | |
| 2292 #ifdef FEAT_MBYTE | |
| 2293 regmbc(i); | |
| 2294 #else | |
| 2295 regc(i); | |
| 2296 #endif | |
| 2297 regc(NUL); | |
| 2298 *flagp |= HASWIDTH; | |
| 2299 break; | |
| 2300 } | |
| 2301 | |
| 7 | 2302 default: |
| 639 | 2303 if (VIM_ISDIGIT(c) || c == '<' || c == '>' |
| 2304 || c == '\'') | |
| 7 | 2305 { |
| 2306 long_u n = 0; | |
| 2307 int cmp; | |
| 2308 | |
| 2309 cmp = c; | |
| 2310 if (cmp == '<' || cmp == '>') | |
| 2311 c = getchr(); | |
| 2312 while (VIM_ISDIGIT(c)) | |
| 2313 { | |
| 2314 n = n * 10 + (c - '0'); | |
| 2315 c = getchr(); | |
| 2316 } | |
| 639 | 2317 if (c == '\'' && n == 0) |
| 2318 { | |
| 2319 /* "\%'m", "\%<'m" and "\%>'m": Mark */ | |
| 2320 c = getchr(); | |
| 2321 ret = regnode(RE_MARK); | |
| 2322 if (ret == JUST_CALC_SIZE) | |
| 2323 regsize += 2; | |
| 2324 else | |
| 2325 { | |
| 2326 *regcode++ = c; | |
| 2327 *regcode++ = cmp; | |
| 2328 } | |
| 2329 break; | |
| 2330 } | |
| 2331 else if (c == 'l' || c == 'c' || c == 'v') | |
| 7 | 2332 { |
| 2333 if (c == 'l') | |
| 2334 ret = regnode(RE_LNUM); | |
| 2335 else if (c == 'c') | |
| 2336 ret = regnode(RE_COL); | |
| 2337 else | |
| 2338 ret = regnode(RE_VCOL); | |
| 2339 if (ret == JUST_CALC_SIZE) | |
| 2340 regsize += 5; | |
| 2341 else | |
| 2342 { | |
| 2343 /* put the number and the optional | |
| 2344 * comparator after the opcode */ | |
| 2345 regcode = re_put_long(regcode, n); | |
| 2346 *regcode++ = cmp; | |
| 2347 } | |
| 2348 break; | |
| 2349 } | |
| 2350 } | |
| 2351 | |
| 4444 | 2352 EMSG2_RET_NULL(_("E71: Invalid character after %s%%"), |
| 7 | 2353 reg_magic == MAGIC_ALL); |
| 2354 } | |
| 2355 } | |
| 2356 break; | |
| 2357 | |
| 2358 case Magic('['): | |
| 2359 collection: | |
| 2360 { | |
| 2361 char_u *lp; | |
| 2362 | |
| 2363 /* | |
| 2364 * If there is no matching ']', we assume the '[' is a normal | |
| 2365 * character. This makes 'incsearch' and ":help [" work. | |
| 2366 */ | |
| 2367 lp = skip_anyof(regparse); | |
| 2368 if (*lp == ']') /* there is a matching ']' */ | |
| 2369 { | |
| 2370 int startc = -1; /* > 0 when next '-' is a range */ | |
| 2371 int endc; | |
| 2372 | |
| 2373 /* | |
| 2374 * In a character class, different parsing rules apply. | |
| 2375 * Not even \ is special anymore, nothing is. | |
| 2376 */ | |
| 2377 if (*regparse == '^') /* Complement of range. */ | |
| 2378 { | |
| 2379 ret = regnode(ANYBUT + extra); | |
| 2380 regparse++; | |
| 2381 } | |
| 2382 else | |
| 2383 ret = regnode(ANYOF + extra); | |
| 2384 | |
| 2385 /* At the start ']' and '-' mean the literal character. */ | |
| 2386 if (*regparse == ']' || *regparse == '-') | |
| 167 | 2387 { |
| 2388 startc = *regparse; | |
| 7 | 2389 regc(*regparse++); |
| 167 | 2390 } |
| 7 | 2391 |
| 2392 while (*regparse != NUL && *regparse != ']') | |
| 2393 { | |
| 2394 if (*regparse == '-') | |
| 2395 { | |
| 2396 ++regparse; | |
| 2397 /* The '-' is not used for a range at the end and | |
| 2398 * after or before a '\n'. */ | |
| 2399 if (*regparse == ']' || *regparse == NUL | |
| 2400 || startc == -1 | |
| 2401 || (regparse[0] == '\\' && regparse[1] == 'n')) | |
| 2402 { | |
| 2403 regc('-'); | |
| 2404 startc = '-'; /* [--x] is a range */ | |
| 2405 } | |
| 2406 else | |
| 2407 { | |
| 167 | 2408 /* Also accept "a-[.z.]" */ |
| 2409 endc = 0; | |
| 2410 if (*regparse == '[') | |
| 2411 endc = get_coll_element(®parse); | |
| 2412 if (endc == 0) | |
| 2413 { | |
| 7 | 2414 #ifdef FEAT_MBYTE |
| 167 | 2415 if (has_mbyte) |
| 2416 endc = mb_ptr2char_adv(®parse); | |
| 2417 else | |
| 7 | 2418 #endif |
| 167 | 2419 endc = *regparse++; |
| 2420 } | |
| 24 | 2421 |
| 2422 /* Handle \o40, \x20 and \u20AC style sequences */ | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2423 if (endc == '\\' && !reg_cpo_lit && !reg_cpo_bsl) |
| 24 | 2424 endc = coll_get_char(); |
| 2425 | |
| 7 | 2426 if (startc > endc) |
| 2427 EMSG_RET_NULL(_(e_invrange)); | |
| 2428 #ifdef FEAT_MBYTE | |
| 2429 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
| 2430 || (*mb_char2len)(endc) > 1)) | |
| 2431 { | |
| 2432 /* Limit to a range of 256 chars */ | |
| 2433 if (endc > startc + 256) | |
| 2434 EMSG_RET_NULL(_(e_invrange)); | |
| 2435 while (++startc <= endc) | |
| 2436 regmbc(startc); | |
| 2437 } | |
| 2438 else | |
| 2439 #endif | |
| 2440 { | |
| 2441 #ifdef EBCDIC | |
| 2442 int alpha_only = FALSE; | |
| 2443 | |
| 2444 /* for alphabetical range skip the gaps | |
| 2445 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
| 2446 if (isalpha(startc) && isalpha(endc)) | |
| 2447 alpha_only = TRUE; | |
| 2448 #endif | |
| 2449 while (++startc <= endc) | |
| 2450 #ifdef EBCDIC | |
| 2451 if (!alpha_only || isalpha(startc)) | |
| 2452 #endif | |
| 2453 regc(startc); | |
| 2454 } | |
| 2455 startc = -1; | |
| 2456 } | |
| 2457 } | |
| 2458 /* | |
| 2459 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
| 2460 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
| 2461 * 'cpoptions' is not included. | |
| 167 | 2462 * Posix doesn't recognize backslash at all. |
| 7 | 2463 */ |
| 2464 else if (*regparse == '\\' | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2465 && !reg_cpo_bsl |
| 7 | 2466 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2467 || (!reg_cpo_lit |
| 7 | 2468 && vim_strchr(REGEXP_ABBR, |
| 2469 regparse[1]) != NULL))) | |
| 2470 { | |
| 2471 regparse++; | |
| 2472 if (*regparse == 'n') | |
| 2473 { | |
| 2474 /* '\n' in range: also match NL */ | |
| 2475 if (ret != JUST_CALC_SIZE) | |
| 2476 { | |
| 4084 | 2477 /* Using \n inside [^] does not change what |
| 2478 * matches. "[^\n]" is the same as ".". */ | |
| 2479 if (*ret == ANYOF) | |
| 2480 { | |
| 7 | 2481 *ret = ANYOF + ADD_NL; |
| 4084 | 2482 *flagp |= HASNL; |
| 2483 } | |
| 7 | 2484 /* else: must have had a \n already */ |
| 2485 } | |
| 2486 regparse++; | |
| 2487 startc = -1; | |
| 2488 } | |
| 24 | 2489 else if (*regparse == 'd' |
| 2490 || *regparse == 'o' | |
| 2491 || *regparse == 'x' | |
| 2492 || *regparse == 'u' | |
| 2493 || *regparse == 'U') | |
| 2494 { | |
| 2495 startc = coll_get_char(); | |
| 2496 if (startc == 0) | |
| 2497 regc(0x0a); | |
| 2498 else | |
| 2499 #ifdef FEAT_MBYTE | |
| 2500 regmbc(startc); | |
| 2501 #else | |
| 2502 regc(startc); | |
| 2503 #endif | |
| 2504 } | |
| 7 | 2505 else |
| 2506 { | |
| 2507 startc = backslash_trans(*regparse++); | |
| 2508 regc(startc); | |
| 2509 } | |
| 2510 } | |
| 2511 else if (*regparse == '[') | |
| 2512 { | |
| 2513 int c_class; | |
| 2514 int cu; | |
| 2515 | |
| 167 | 2516 c_class = get_char_class(®parse); |
| 7 | 2517 startc = -1; |
| 2518 /* Characters assumed to be 8 bits! */ | |
| 2519 switch (c_class) | |
| 2520 { | |
| 2521 case CLASS_NONE: | |
| 167 | 2522 c_class = get_equi_class(®parse); |
| 2523 if (c_class != 0) | |
| 2524 { | |
| 2525 /* produce equivalence class */ | |
| 2526 reg_equi_class(c_class); | |
| 2527 } | |
| 2528 else if ((c_class = | |
| 2529 get_coll_element(®parse)) != 0) | |
| 2530 { | |
| 2531 /* produce a collating element */ | |
| 2532 regmbc(c_class); | |
| 2533 } | |
| 2534 else | |
| 2535 { | |
| 2536 /* literal '[', allow [[-x] as a range */ | |
| 2537 startc = *regparse++; | |
| 2538 regc(startc); | |
| 2539 } | |
| 7 | 2540 break; |
| 2541 case CLASS_ALNUM: | |
| 2542 for (cu = 1; cu <= 255; cu++) | |
| 2543 if (isalnum(cu)) | |
| 2544 regc(cu); | |
| 2545 break; | |
| 2546 case CLASS_ALPHA: | |
| 2547 for (cu = 1; cu <= 255; cu++) | |
| 2548 if (isalpha(cu)) | |
| 2549 regc(cu); | |
| 2550 break; | |
| 2551 case CLASS_BLANK: | |
| 2552 regc(' '); | |
| 2553 regc('\t'); | |
| 2554 break; | |
| 2555 case CLASS_CNTRL: | |
| 2556 for (cu = 1; cu <= 255; cu++) | |
| 2557 if (iscntrl(cu)) | |
| 2558 regc(cu); | |
| 2559 break; | |
| 2560 case CLASS_DIGIT: | |
| 2561 for (cu = 1; cu <= 255; cu++) | |
| 2562 if (VIM_ISDIGIT(cu)) | |
| 2563 regc(cu); | |
| 2564 break; | |
| 2565 case CLASS_GRAPH: | |
| 2566 for (cu = 1; cu <= 255; cu++) | |
| 2567 if (isgraph(cu)) | |
| 2568 regc(cu); | |
| 2569 break; | |
| 2570 case CLASS_LOWER: | |
| 2571 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2572 if (MB_ISLOWER(cu)) |
| 7 | 2573 regc(cu); |
| 2574 break; | |
| 2575 case CLASS_PRINT: | |
| 2576 for (cu = 1; cu <= 255; cu++) | |
| 2577 if (vim_isprintc(cu)) | |
| 2578 regc(cu); | |
| 2579 break; | |
| 2580 case CLASS_PUNCT: | |
| 2581 for (cu = 1; cu <= 255; cu++) | |
| 2582 if (ispunct(cu)) | |
| 2583 regc(cu); | |
| 2584 break; | |
| 2585 case CLASS_SPACE: | |
| 2586 for (cu = 9; cu <= 13; cu++) | |
| 2587 regc(cu); | |
| 2588 regc(' '); | |
| 2589 break; | |
| 2590 case CLASS_UPPER: | |
| 2591 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2592 if (MB_ISUPPER(cu)) |
| 7 | 2593 regc(cu); |
| 2594 break; | |
| 2595 case CLASS_XDIGIT: | |
| 2596 for (cu = 1; cu <= 255; cu++) | |
| 2597 if (vim_isxdigit(cu)) | |
| 2598 regc(cu); | |
| 2599 break; | |
| 2600 case CLASS_TAB: | |
| 2601 regc('\t'); | |
| 2602 break; | |
| 2603 case CLASS_RETURN: | |
| 2604 regc('\r'); | |
| 2605 break; | |
| 2606 case CLASS_BACKSPACE: | |
| 2607 regc('\b'); | |
| 2608 break; | |
| 2609 case CLASS_ESCAPE: | |
| 2610 regc('\033'); | |
| 2611 break; | |
| 2612 } | |
| 2613 } | |
| 2614 else | |
| 2615 { | |
| 2616 #ifdef FEAT_MBYTE | |
| 2617 if (has_mbyte) | |
| 2618 { | |
| 2619 int len; | |
| 2620 | |
| 2621 /* produce a multibyte character, including any | |
| 2622 * following composing characters */ | |
| 2623 startc = mb_ptr2char(regparse); | |
| 474 | 2624 len = (*mb_ptr2len)(regparse); |
| 7 | 2625 if (enc_utf8 && utf_char2len(startc) != len) |
| 2626 startc = -1; /* composing chars */ | |
| 2627 while (--len >= 0) | |
| 2628 regc(*regparse++); | |
| 2629 } | |
| 2630 else | |
| 2631 #endif | |
| 2632 { | |
| 2633 startc = *regparse++; | |
| 2634 regc(startc); | |
| 2635 } | |
| 2636 } | |
| 2637 } | |
| 2638 regc(NUL); | |
| 2639 prevchr_len = 1; /* last char was the ']' */ | |
| 2640 if (*regparse != ']') | |
| 2641 EMSG_RET_NULL(_(e_toomsbra)); /* Cannot happen? */ | |
| 2642 skipchr(); /* let's be friends with the lexer again */ | |
| 2643 *flagp |= HASWIDTH | SIMPLE; | |
| 2644 break; | |
| 2645 } | |
| 481 | 2646 else if (reg_strict) |
| 4444 | 2647 EMSG2_RET_NULL(_(e_missingbracket), reg_magic > MAGIC_OFF); |
| 7 | 2648 } |
| 2649 /* FALLTHROUGH */ | |
| 2650 | |
| 2651 default: | |
| 2652 { | |
| 2653 int len; | |
| 2654 | |
| 2655 #ifdef FEAT_MBYTE | |
| 2656 /* A multi-byte character is handled as a separate atom if it's | |
| 714 | 2657 * before a multi and when it's a composing char. */ |
| 2658 if (use_multibytecode(c)) | |
| 7 | 2659 { |
| 714 | 2660 do_multibyte: |
| 7 | 2661 ret = regnode(MULTIBYTECODE); |
| 2662 regmbc(c); | |
| 2663 *flagp |= HASWIDTH | SIMPLE; | |
| 2664 break; | |
| 2665 } | |
| 2666 #endif | |
| 2667 | |
| 2668 ret = regnode(EXACTLY); | |
| 2669 | |
| 2670 /* | |
| 2671 * Append characters as long as: | |
| 2672 * - there is no following multi, we then need the character in | |
| 2673 * front of it as a single character operand | |
| 2674 * - not running into a Magic character | |
| 2675 * - "one_exactly" is not set | |
| 2676 * But always emit at least one character. Might be a Multi, | |
| 2677 * e.g., a "[" without matching "]". | |
| 2678 */ | |
| 2679 for (len = 0; c != NUL && (len == 0 | |
| 2680 || (re_multi_type(peekchr()) == NOT_MULTI | |
| 2681 && !one_exactly | |
| 2682 && !is_Magic(c))); ++len) | |
| 2683 { | |
| 2684 c = no_Magic(c); | |
| 2685 #ifdef FEAT_MBYTE | |
| 2686 if (has_mbyte) | |
| 2687 { | |
| 2688 regmbc(c); | |
| 2689 if (enc_utf8) | |
| 2690 { | |
| 2691 int l; | |
| 2692 | |
| 714 | 2693 /* Need to get composing character too. */ |
| 7 | 2694 for (;;) |
| 2695 { | |
| 714 | 2696 l = utf_ptr2len(regparse); |
| 2697 if (!UTF_COMPOSINGLIKE(regparse, regparse + l)) | |
| 7 | 2698 break; |
| 714 | 2699 regmbc(utf_ptr2char(regparse)); |
| 2700 skipchr(); | |
| 7 | 2701 } |
| 2702 } | |
| 2703 } | |
| 2704 else | |
| 2705 #endif | |
| 2706 regc(c); | |
| 2707 c = getchr(); | |
| 2708 } | |
| 2709 ungetchr(); | |
| 2710 | |
| 2711 regc(NUL); | |
| 2712 *flagp |= HASWIDTH; | |
| 2713 if (len == 1) | |
| 2714 *flagp |= SIMPLE; | |
| 2715 } | |
| 2716 break; | |
| 2717 } | |
| 2718 | |
| 2719 return ret; | |
| 2720 } | |
| 2721 | |
| 714 | 2722 #ifdef FEAT_MBYTE |
| 2723 /* | |
| 2724 * Return TRUE if MULTIBYTECODE should be used instead of EXACTLY for | |
| 2725 * character "c". | |
| 2726 */ | |
| 2727 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2728 use_multibytecode(int c) |
| 714 | 2729 { |
| 2730 return has_mbyte && (*mb_char2len)(c) > 1 | |
| 2731 && (re_multi_type(peekchr()) != NOT_MULTI | |
| 2732 || (enc_utf8 && utf_iscomposing(c))); | |
| 2733 } | |
| 2734 #endif | |
| 2735 | |
| 7 | 2736 /* |
| 4444 | 2737 * Emit a node. |
| 7 | 2738 * Return pointer to generated code. |
| 2739 */ | |
| 2740 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2741 regnode(int op) |
| 7 | 2742 { |
| 2743 char_u *ret; | |
| 2744 | |
| 2745 ret = regcode; | |
| 2746 if (ret == JUST_CALC_SIZE) | |
| 2747 regsize += 3; | |
| 2748 else | |
| 2749 { | |
| 2750 *regcode++ = op; | |
| 2751 *regcode++ = NUL; /* Null "next" pointer. */ | |
| 2752 *regcode++ = NUL; | |
| 2753 } | |
| 2754 return ret; | |
| 2755 } | |
| 2756 | |
| 2757 /* | |
| 2758 * Emit (if appropriate) a byte of code | |
| 2759 */ | |
| 2760 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2761 regc(int b) |
| 7 | 2762 { |
| 2763 if (regcode == JUST_CALC_SIZE) | |
| 2764 regsize++; | |
| 2765 else | |
| 2766 *regcode++ = b; | |
| 2767 } | |
| 2768 | |
| 2769 #ifdef FEAT_MBYTE | |
| 2770 /* | |
| 2771 * Emit (if appropriate) a multi-byte character of code | |
| 2772 */ | |
| 2773 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2774 regmbc(int c) |
| 7 | 2775 { |
| 2974 | 2776 if (!has_mbyte && c > 0xff) |
| 2777 return; | |
| 7 | 2778 if (regcode == JUST_CALC_SIZE) |
| 2779 regsize += (*mb_char2len)(c); | |
| 2780 else | |
| 2781 regcode += (*mb_char2bytes)(c, regcode); | |
| 2782 } | |
| 2783 #endif | |
| 2784 | |
| 2785 /* | |
| 4444 | 2786 * Insert an operator in front of already-emitted operand |
| 7 | 2787 * |
| 2788 * Means relocating the operand. | |
| 2789 */ | |
| 2790 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2791 reginsert(int op, char_u *opnd) |
| 7 | 2792 { |
| 2793 char_u *src; | |
| 2794 char_u *dst; | |
| 2795 char_u *place; | |
| 2796 | |
| 2797 if (regcode == JUST_CALC_SIZE) | |
| 2798 { | |
| 2799 regsize += 3; | |
| 2800 return; | |
| 2801 } | |
| 2802 src = regcode; | |
| 2803 regcode += 3; | |
| 2804 dst = regcode; | |
| 2805 while (src > opnd) | |
| 2806 *--dst = *--src; | |
| 2807 | |
| 2808 place = opnd; /* Op node, where operand used to be. */ | |
| 2809 *place++ = op; | |
| 2810 *place++ = NUL; | |
| 2811 *place = NUL; | |
| 2812 } | |
| 2813 | |
| 2814 /* | |
| 4444 | 2815 * Insert an operator in front of already-emitted operand. |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2816 * Add a number to the operator. |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2817 */ |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2818 static void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2819 reginsert_nr(int op, long val, char_u *opnd) |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2820 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2821 char_u *src; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2822 char_u *dst; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2823 char_u *place; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2824 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2825 if (regcode == JUST_CALC_SIZE) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2826 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2827 regsize += 7; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2828 return; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2829 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2830 src = regcode; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2831 regcode += 7; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2832 dst = regcode; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2833 while (src > opnd) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2834 *--dst = *--src; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2835 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2836 place = opnd; /* Op node, where operand used to be. */ |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2837 *place++ = op; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2838 *place++ = NUL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2839 *place++ = NUL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2840 place = re_put_long(place, (long_u)val); |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2841 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2842 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2843 /* |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2844 * Insert an operator in front of already-emitted operand. |
| 7 | 2845 * The operator has the given limit values as operands. Also set next pointer. |
| 2846 * | |
| 2847 * Means relocating the operand. | |
| 2848 */ | |
| 2849 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2850 reginsert_limits( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2851 int op, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2852 long minval, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2853 long maxval, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2854 char_u *opnd) |
| 7 | 2855 { |
| 2856 char_u *src; | |
| 2857 char_u *dst; | |
| 2858 char_u *place; | |
| 2859 | |
| 2860 if (regcode == JUST_CALC_SIZE) | |
| 2861 { | |
| 2862 regsize += 11; | |
| 2863 return; | |
| 2864 } | |
| 2865 src = regcode; | |
| 2866 regcode += 11; | |
| 2867 dst = regcode; | |
| 2868 while (src > opnd) | |
| 2869 *--dst = *--src; | |
| 2870 | |
| 2871 place = opnd; /* Op node, where operand used to be. */ | |
| 2872 *place++ = op; | |
| 2873 *place++ = NUL; | |
| 2874 *place++ = NUL; | |
| 2875 place = re_put_long(place, (long_u)minval); | |
| 2876 place = re_put_long(place, (long_u)maxval); | |
| 2877 regtail(opnd, place); | |
| 2878 } | |
| 2879 | |
| 2880 /* | |
| 2881 * Write a long as four bytes at "p" and return pointer to the next char. | |
| 2882 */ | |
| 2883 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2884 re_put_long(char_u *p, long_u val) |
| 7 | 2885 { |
| 2886 *p++ = (char_u) ((val >> 24) & 0377); | |
| 2887 *p++ = (char_u) ((val >> 16) & 0377); | |
| 2888 *p++ = (char_u) ((val >> 8) & 0377); | |
| 2889 *p++ = (char_u) (val & 0377); | |
| 2890 return p; | |
| 2891 } | |
| 2892 | |
| 2893 /* | |
| 4444 | 2894 * Set the next-pointer at the end of a node chain. |
| 7 | 2895 */ |
| 2896 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2897 regtail(char_u *p, char_u *val) |
| 7 | 2898 { |
| 2899 char_u *scan; | |
| 2900 char_u *temp; | |
| 2901 int offset; | |
| 2902 | |
| 2903 if (p == JUST_CALC_SIZE) | |
| 2904 return; | |
| 2905 | |
| 2906 /* Find last node. */ | |
| 2907 scan = p; | |
| 2908 for (;;) | |
| 2909 { | |
| 2910 temp = regnext(scan); | |
| 2911 if (temp == NULL) | |
| 2912 break; | |
| 2913 scan = temp; | |
| 2914 } | |
| 2915 | |
| 233 | 2916 if (OP(scan) == BACK) |
| 7 | 2917 offset = (int)(scan - val); |
| 2918 else | |
| 2919 offset = (int)(val - scan); | |
| 2010 | 2920 /* When the offset uses more than 16 bits it can no longer fit in the two |
| 2974 | 2921 * bytes available. Use a global flag to avoid having to check return |
| 2010 | 2922 * values in too many places. */ |
| 2923 if (offset > 0xffff) | |
| 2924 reg_toolong = TRUE; | |
| 2925 else | |
| 2926 { | |
| 2927 *(scan + 1) = (char_u) (((unsigned)offset >> 8) & 0377); | |
| 2928 *(scan + 2) = (char_u) (offset & 0377); | |
| 2929 } | |
| 7 | 2930 } |
| 2931 | |
| 2932 /* | |
| 4444 | 2933 * Like regtail, on item after a BRANCH; nop if none. |
| 7 | 2934 */ |
| 2935 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2936 regoptail(char_u *p, char_u *val) |
| 7 | 2937 { |
| 2938 /* When op is neither BRANCH nor BRACE_COMPLEX0-9, it is "operandless" */ | |
| 2939 if (p == NULL || p == JUST_CALC_SIZE | |
| 2940 || (OP(p) != BRANCH | |
| 2941 && (OP(p) < BRACE_COMPLEX || OP(p) > BRACE_COMPLEX + 9))) | |
| 2942 return; | |
| 2943 regtail(OPERAND(p), val); | |
| 2944 } | |
| 2945 | |
| 2946 /* | |
| 4444 | 2947 * Functions for getting characters from the regexp input. |
| 7 | 2948 */ |
| 2949 | |
| 2950 static int at_start; /* True when on the first character */ | |
| 2951 static int prev_at_start; /* True when on the second character */ | |
| 2952 | |
| 4444 | 2953 /* |
| 2954 * Start parsing at "str". | |
| 2955 */ | |
| 7 | 2956 static void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2957 initchr(char_u *str) |
| 7 | 2958 { |
| 2959 regparse = str; | |
| 2960 prevchr_len = 0; | |
| 2961 curchr = prevprevchr = prevchr = nextchr = -1; | |
| 2962 at_start = TRUE; | |
| 2963 prev_at_start = FALSE; | |
| 2964 } | |
| 2965 | |
| 4444 | 2966 /* |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2967 * Save the current parse state, so that it can be restored and parsing |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2968 * starts in the same state again. |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2969 */ |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2970 static void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2971 save_parse_state(parse_state_T *ps) |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2972 { |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2973 ps->regparse = regparse; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2974 ps->prevchr_len = prevchr_len; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2975 ps->curchr = curchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2976 ps->prevchr = prevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2977 ps->prevprevchr = prevprevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2978 ps->nextchr = nextchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2979 ps->at_start = at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2980 ps->prev_at_start = prev_at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2981 ps->regnpar = regnpar; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2982 } |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2983 |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2984 /* |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2985 * Restore a previously saved parse state. |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2986 */ |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2987 static void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2988 restore_parse_state(parse_state_T *ps) |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2989 { |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2990 regparse = ps->regparse; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2991 prevchr_len = ps->prevchr_len; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2992 curchr = ps->curchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2993 prevchr = ps->prevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2994 prevprevchr = ps->prevprevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2995 nextchr = ps->nextchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2996 at_start = ps->at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2997 prev_at_start = ps->prev_at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2998 regnpar = ps->regnpar; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2999 } |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3000 |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3001 |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3002 /* |
| 4444 | 3003 * Get the next character without advancing. |
| 3004 */ | |
| 7 | 3005 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3006 peekchr(void) |
| 7 | 3007 { |
| 167 | 3008 static int after_slash = FALSE; |
| 3009 | |
| 7 | 3010 if (curchr == -1) |
| 3011 { | |
| 3012 switch (curchr = regparse[0]) | |
| 3013 { | |
| 3014 case '.': | |
| 3015 case '[': | |
| 3016 case '~': | |
| 3017 /* magic when 'magic' is on */ | |
| 3018 if (reg_magic >= MAGIC_ON) | |
| 3019 curchr = Magic(curchr); | |
| 3020 break; | |
| 3021 case '(': | |
| 3022 case ')': | |
| 3023 case '{': | |
| 3024 case '%': | |
| 3025 case '+': | |
| 3026 case '=': | |
| 3027 case '?': | |
| 3028 case '@': | |
| 3029 case '!': | |
| 3030 case '&': | |
| 3031 case '|': | |
| 3032 case '<': | |
| 3033 case '>': | |
| 3034 case '#': /* future ext. */ | |
| 3035 case '"': /* future ext. */ | |
| 3036 case '\'': /* future ext. */ | |
| 3037 case ',': /* future ext. */ | |
| 3038 case '-': /* future ext. */ | |
| 3039 case ':': /* future ext. */ | |
| 3040 case ';': /* future ext. */ | |
| 3041 case '`': /* future ext. */ | |
| 3042 case '/': /* Can't be used in / command */ | |
| 3043 /* magic only after "\v" */ | |
| 3044 if (reg_magic == MAGIC_ALL) | |
| 3045 curchr = Magic(curchr); | |
| 3046 break; | |
| 3047 case '*': | |
| 167 | 3048 /* * is not magic as the very first character, eg "?*ptr", when |
| 3049 * after '^', eg "/^*ptr" and when after "\(", "\|", "\&". But | |
| 3050 * "\(\*" is not magic, thus must be magic if "after_slash" */ | |
| 3051 if (reg_magic >= MAGIC_ON | |
| 3052 && !at_start | |
| 3053 && !(prev_at_start && prevchr == Magic('^')) | |
| 3054 && (after_slash | |
| 3055 || (prevchr != Magic('(') | |
| 3056 && prevchr != Magic('&') | |
| 3057 && prevchr != Magic('|')))) | |
| 7 | 3058 curchr = Magic('*'); |
| 3059 break; | |
| 3060 case '^': | |
| 3061 /* '^' is only magic as the very first character and if it's after | |
| 3062 * "\(", "\|", "\&' or "\n" */ | |
| 3063 if (reg_magic >= MAGIC_OFF | |
| 3064 && (at_start | |
| 3065 || reg_magic == MAGIC_ALL | |
| 3066 || prevchr == Magic('(') | |
| 3067 || prevchr == Magic('|') | |
| 3068 || prevchr == Magic('&') | |
| 3069 || prevchr == Magic('n') | |
| 3070 || (no_Magic(prevchr) == '(' | |
| 3071 && prevprevchr == Magic('%')))) | |
| 3072 { | |
| 3073 curchr = Magic('^'); | |
| 3074 at_start = TRUE; | |
| 3075 prev_at_start = FALSE; | |
| 3076 } | |
| 3077 break; | |
| 3078 case '$': | |
| 3079 /* '$' is only magic as the very last char and if it's in front of | |
| 3080 * either "\|", "\)", "\&", or "\n" */ | |
| 3081 if (reg_magic >= MAGIC_OFF) | |
| 3082 { | |
| 3083 char_u *p = regparse + 1; | |
| 6041 | 3084 int is_magic_all = (reg_magic == MAGIC_ALL); |
| 3085 | |
| 3086 /* ignore \c \C \m \M \v \V and \Z after '$' */ | |
| 7 | 3087 while (p[0] == '\\' && (p[1] == 'c' || p[1] == 'C' |
| 6041 | 3088 || p[1] == 'm' || p[1] == 'M' |
| 3089 || p[1] == 'v' || p[1] == 'V' || p[1] == 'Z')) | |
| 3090 { | |
| 3091 if (p[1] == 'v') | |
| 3092 is_magic_all = TRUE; | |
| 3093 else if (p[1] == 'm' || p[1] == 'M' || p[1] == 'V') | |
| 3094 is_magic_all = FALSE; | |
| 7 | 3095 p += 2; |
| 6041 | 3096 } |
| 7 | 3097 if (p[0] == NUL |
| 3098 || (p[0] == '\\' | |
| 3099 && (p[1] == '|' || p[1] == '&' || p[1] == ')' | |
| 3100 || p[1] == 'n')) | |
| 6041 | 3101 || (is_magic_all |
| 3102 && (p[0] == '|' || p[0] == '&' || p[0] == ')')) | |
| 7 | 3103 || reg_magic == MAGIC_ALL) |
| 3104 curchr = Magic('$'); | |
| 3105 } | |
| 3106 break; | |
| 3107 case '\\': | |
| 3108 { | |
| 3109 int c = regparse[1]; | |
| 3110 | |
| 3111 if (c == NUL) | |
| 3112 curchr = '\\'; /* trailing '\' */ | |
| 3113 else if ( | |
| 3114 #ifdef EBCDIC | |
| 3115 vim_strchr(META, c) | |
| 3116 #else | |
| 3117 c <= '~' && META_flags[c] | |
| 3118 #endif | |
| 3119 ) | |
| 3120 { | |
| 3121 /* | |
| 3122 * META contains everything that may be magic sometimes, | |
| 3123 * except ^ and $ ("\^" and "\$" are only magic after | |
| 6830 | 3124 * "\V"). We now fetch the next character and toggle its |
| 7 | 3125 * magicness. Therefore, \ is so meta-magic that it is |
| 3126 * not in META. | |
| 3127 */ | |
| 3128 curchr = -1; | |
| 3129 prev_at_start = at_start; | |
| 3130 at_start = FALSE; /* be able to say "/\*ptr" */ | |
| 3131 ++regparse; | |
| 167 | 3132 ++after_slash; |
| 7 | 3133 peekchr(); |
| 3134 --regparse; | |
| 167 | 3135 --after_slash; |
| 7 | 3136 curchr = toggle_Magic(curchr); |
| 3137 } | |
| 3138 else if (vim_strchr(REGEXP_ABBR, c)) | |
| 3139 { | |
| 3140 /* | |
| 3141 * Handle abbreviations, like "\t" for TAB -- webb | |
| 3142 */ | |
| 3143 curchr = backslash_trans(c); | |
| 3144 } | |
| 3145 else if (reg_magic == MAGIC_NONE && (c == '$' || c == '^')) | |
| 3146 curchr = toggle_Magic(c); | |
| 3147 else | |
| 3148 { | |
| 3149 /* | |
| 3150 * Next character can never be (made) magic? | |
| 3151 * Then backslashing it won't do anything. | |
| 3152 */ | |
| 3153 #ifdef FEAT_MBYTE | |
| 3154 if (has_mbyte) | |
| 3155 curchr = (*mb_ptr2char)(regparse + 1); | |
| 3156 else | |
| 3157 #endif | |
| 3158 curchr = c; | |
| 3159 } | |
| 3160 break; | |
| 3161 } | |
| 3162 | |
| 3163 #ifdef FEAT_MBYTE | |
| 3164 default: | |
| 3165 if (has_mbyte) | |
| 3166 curchr = (*mb_ptr2char)(regparse); | |
| 3167 #endif | |
| 3168 } | |
| 3169 } | |
| 3170 | |
| 3171 return curchr; | |
| 3172 } | |
| 3173 | |
| 3174 /* | |
| 3175 * Eat one lexed character. Do this in a way that we can undo it. | |
| 3176 */ | |
| 3177 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3178 skipchr(void) |
| 7 | 3179 { |
| 3180 /* peekchr() eats a backslash, do the same here */ | |
| 3181 if (*regparse == '\\') | |
| 3182 prevchr_len = 1; | |
| 3183 else | |
| 3184 prevchr_len = 0; | |
| 3185 if (regparse[prevchr_len] != NUL) | |
| 3186 { | |
| 3187 #ifdef FEAT_MBYTE | |
| 714 | 3188 if (enc_utf8) |
| 1449 | 3189 /* exclude composing chars that mb_ptr2len does include */ |
| 3190 prevchr_len += utf_ptr2len(regparse + prevchr_len); | |
| 714 | 3191 else if (has_mbyte) |
| 474 | 3192 prevchr_len += (*mb_ptr2len)(regparse + prevchr_len); |
| 7 | 3193 else |
| 3194 #endif | |
| 3195 ++prevchr_len; | |
| 3196 } | |
| 3197 regparse += prevchr_len; | |
| 3198 prev_at_start = at_start; | |
| 3199 at_start = FALSE; | |
| 3200 prevprevchr = prevchr; | |
| 3201 prevchr = curchr; | |
| 3202 curchr = nextchr; /* use previously unget char, or -1 */ | |
| 3203 nextchr = -1; | |
| 3204 } | |
| 3205 | |
| 3206 /* | |
| 3207 * Skip a character while keeping the value of prev_at_start for at_start. | |
| 3208 * prevchr and prevprevchr are also kept. | |
| 3209 */ | |
| 3210 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3211 skipchr_keepstart(void) |
| 7 | 3212 { |
| 3213 int as = prev_at_start; | |
| 3214 int pr = prevchr; | |
| 3215 int prpr = prevprevchr; | |
| 3216 | |
| 3217 skipchr(); | |
| 3218 at_start = as; | |
| 3219 prevchr = pr; | |
| 3220 prevprevchr = prpr; | |
| 3221 } | |
| 3222 | |
| 4444 | 3223 /* |
| 3224 * Get the next character from the pattern. We know about magic and such, so | |
| 3225 * therefore we need a lexical analyzer. | |
| 3226 */ | |
| 7 | 3227 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3228 getchr(void) |
| 7 | 3229 { |
| 3230 int chr = peekchr(); | |
| 3231 | |
| 3232 skipchr(); | |
| 3233 return chr; | |
| 3234 } | |
| 3235 | |
| 3236 /* | |
| 3237 * put character back. Works only once! | |
| 3238 */ | |
| 3239 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3240 ungetchr(void) |
| 7 | 3241 { |
| 3242 nextchr = curchr; | |
| 3243 curchr = prevchr; | |
| 3244 prevchr = prevprevchr; | |
| 3245 at_start = prev_at_start; | |
| 3246 prev_at_start = FALSE; | |
| 3247 | |
| 3248 /* Backup regparse, so that it's at the same position as before the | |
| 3249 * getchr(). */ | |
| 3250 regparse -= prevchr_len; | |
| 3251 } | |
| 3252 | |
| 3253 /* | |
| 29 | 3254 * Get and return the value of the hex string at the current position. |
| 3255 * Return -1 if there is no valid hex number. | |
| 3256 * The position is updated: | |
| 24 | 3257 * blahblah\%x20asdf |
| 856 | 3258 * before-^ ^-after |
| 24 | 3259 * The parameter controls the maximum number of input characters. This will be |
| 3260 * 2 when reading a \%x20 sequence and 4 when reading a \%u20AC sequence. | |
| 3261 */ | |
| 3262 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3263 gethexchrs(int maxinputlen) |
| 24 | 3264 { |
| 3265 int nr = 0; | |
| 3266 int c; | |
| 3267 int i; | |
| 3268 | |
| 3269 for (i = 0; i < maxinputlen; ++i) | |
| 3270 { | |
| 3271 c = regparse[0]; | |
| 3272 if (!vim_isxdigit(c)) | |
| 3273 break; | |
| 3274 nr <<= 4; | |
| 3275 nr |= hex2nr(c); | |
| 3276 ++regparse; | |
| 3277 } | |
| 3278 | |
| 3279 if (i == 0) | |
| 3280 return -1; | |
| 3281 return nr; | |
| 3282 } | |
| 3283 | |
| 3284 /* | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3285 * Get and return the value of the decimal string immediately after the |
| 24 | 3286 * current position. Return -1 for invalid. Consumes all digits. |
| 3287 */ | |
| 3288 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3289 getdecchrs(void) |
| 24 | 3290 { |
| 3291 int nr = 0; | |
| 3292 int c; | |
| 3293 int i; | |
| 3294 | |
| 3295 for (i = 0; ; ++i) | |
| 3296 { | |
| 3297 c = regparse[0]; | |
| 3298 if (c < '0' || c > '9') | |
| 3299 break; | |
| 3300 nr *= 10; | |
| 3301 nr += c - '0'; | |
| 3302 ++regparse; | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3303 curchr = -1; /* no longer valid */ |
| 24 | 3304 } |
| 3305 | |
| 3306 if (i == 0) | |
| 3307 return -1; | |
| 3308 return nr; | |
| 3309 } | |
| 3310 | |
| 3311 /* | |
| 3312 * get and return the value of the octal string immediately after the current | |
| 3313 * position. Return -1 for invalid, or 0-255 for valid. Smart enough to handle | |
| 3314 * numbers > 377 correctly (for example, 400 is treated as 40) and doesn't | |
| 3315 * treat 8 or 9 as recognised characters. Position is updated: | |
| 3316 * blahblah\%o210asdf | |
| 856 | 3317 * before-^ ^-after |
| 24 | 3318 */ |
| 3319 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3320 getoctchrs(void) |
| 24 | 3321 { |
| 3322 int nr = 0; | |
| 3323 int c; | |
| 3324 int i; | |
| 3325 | |
| 3326 for (i = 0; i < 3 && nr < 040; ++i) | |
| 3327 { | |
| 3328 c = regparse[0]; | |
| 3329 if (c < '0' || c > '7') | |
| 3330 break; | |
| 3331 nr <<= 3; | |
| 3332 nr |= hex2nr(c); | |
| 3333 ++regparse; | |
| 3334 } | |
| 3335 | |
| 3336 if (i == 0) | |
| 3337 return -1; | |
| 3338 return nr; | |
| 3339 } | |
| 3340 | |
| 3341 /* | |
| 3342 * Get a number after a backslash that is inside []. | |
| 3343 * When nothing is recognized return a backslash. | |
| 3344 */ | |
| 3345 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3346 coll_get_char(void) |
| 24 | 3347 { |
| 3348 int nr = -1; | |
| 3349 | |
| 3350 switch (*regparse++) | |
| 3351 { | |
| 3352 case 'd': nr = getdecchrs(); break; | |
| 3353 case 'o': nr = getoctchrs(); break; | |
| 3354 case 'x': nr = gethexchrs(2); break; | |
| 3355 case 'u': nr = gethexchrs(4); break; | |
| 3356 case 'U': nr = gethexchrs(8); break; | |
| 3357 } | |
| 3358 if (nr < 0) | |
| 3359 { | |
| 3360 /* If getting the number fails be backwards compatible: the character | |
| 3361 * is a backslash. */ | |
| 3362 --regparse; | |
| 3363 nr = '\\'; | |
| 3364 } | |
| 3365 return nr; | |
| 3366 } | |
| 3367 | |
| 3368 /* | |
| 7 | 3369 * read_limits - Read two integers to be taken as a minimum and maximum. |
| 3370 * If the first character is '-', then the range is reversed. | |
| 3371 * Should end with 'end'. If minval is missing, zero is default, if maxval is | |
| 3372 * missing, a very big number is the default. | |
| 3373 */ | |
| 3374 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3375 read_limits(long *minval, long *maxval) |
| 7 | 3376 { |
| 3377 int reverse = FALSE; | |
| 3378 char_u *first_char; | |
| 3379 long tmp; | |
| 3380 | |
| 3381 if (*regparse == '-') | |
| 3382 { | |
| 3383 /* Starts with '-', so reverse the range later */ | |
| 3384 regparse++; | |
| 3385 reverse = TRUE; | |
| 3386 } | |
| 3387 first_char = regparse; | |
| 3388 *minval = getdigits(®parse); | |
| 3389 if (*regparse == ',') /* There is a comma */ | |
| 3390 { | |
| 3391 if (vim_isdigit(*++regparse)) | |
| 3392 *maxval = getdigits(®parse); | |
| 3393 else | |
| 3394 *maxval = MAX_LIMIT; | |
| 3395 } | |
| 3396 else if (VIM_ISDIGIT(*first_char)) | |
| 3397 *maxval = *minval; /* It was \{n} or \{-n} */ | |
| 3398 else | |
| 3399 *maxval = MAX_LIMIT; /* It was \{} or \{-} */ | |
| 3400 if (*regparse == '\\') | |
| 3401 regparse++; /* Allow either \{...} or \{...\} */ | |
| 167 | 3402 if (*regparse != '}') |
| 7 | 3403 { |
| 3404 sprintf((char *)IObuff, _("E554: Syntax error in %s{...}"), | |
| 3405 reg_magic == MAGIC_ALL ? "" : "\\"); | |
| 3406 EMSG_RET_FAIL(IObuff); | |
| 3407 } | |
| 3408 | |
| 3409 /* | |
| 3410 * Reverse the range if there was a '-', or make sure it is in the right | |
| 3411 * order otherwise. | |
| 3412 */ | |
| 3413 if ((!reverse && *minval > *maxval) || (reverse && *minval < *maxval)) | |
| 3414 { | |
| 3415 tmp = *minval; | |
| 3416 *minval = *maxval; | |
| 3417 *maxval = tmp; | |
| 3418 } | |
| 3419 skipchr(); /* let's be friends with the lexer again */ | |
| 3420 return OK; | |
| 3421 } | |
| 3422 | |
| 3423 /* | |
| 3424 * vim_regexec and friends | |
| 3425 */ | |
| 3426 | |
| 3427 /* | |
| 3428 * Global work variables for vim_regexec(). | |
| 3429 */ | |
| 3430 | |
| 3431 /* The current match-position is remembered with these variables: */ | |
| 3432 static linenr_T reglnum; /* line number, relative to first line */ | |
| 3433 static char_u *regline; /* start of current line */ | |
| 3434 static char_u *reginput; /* current input, points into "regline" */ | |
| 3435 | |
| 3436 static int need_clear_subexpr; /* subexpressions still need to be | |
| 3437 * cleared */ | |
| 3438 #ifdef FEAT_SYN_HL | |
| 3439 static int need_clear_zsubexpr = FALSE; /* extmatch subexpressions | |
| 3440 * still need to be cleared */ | |
| 3441 #endif | |
| 3442 | |
| 3443 /* | |
| 3444 * Structure used to save the current input state, when it needs to be | |
| 3445 * restored after trying a match. Used by reg_save() and reg_restore(). | |
| 233 | 3446 * Also stores the length of "backpos". |
| 7 | 3447 */ |
| 3448 typedef struct | |
| 3449 { | |
| 3450 union | |
| 3451 { | |
| 3452 char_u *ptr; /* reginput pointer, for single-line regexp */ | |
| 3453 lpos_T pos; /* reginput pos, for multi-line regexp */ | |
| 3454 } rs_u; | |
| 233 | 3455 int rs_len; |
| 7 | 3456 } regsave_T; |
| 3457 | |
| 3458 /* struct to save start/end pointer/position in for \(\) */ | |
| 3459 typedef struct | |
| 3460 { | |
| 3461 union | |
| 3462 { | |
| 3463 char_u *ptr; | |
| 3464 lpos_T pos; | |
| 3465 } se_u; | |
| 3466 } save_se_T; | |
| 3467 | |
| 1579 | 3468 /* used for BEHIND and NOBEHIND matching */ |
| 3469 typedef struct regbehind_S | |
| 3470 { | |
| 3471 regsave_T save_after; | |
| 3472 regsave_T save_behind; | |
| 1602 | 3473 int save_need_clear_subexpr; |
| 1579 | 3474 save_se_T save_start[NSUBEXP]; |
| 3475 save_se_T save_end[NSUBEXP]; | |
| 3476 } regbehind_T; | |
| 3477 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3478 static char_u *reg_getline(linenr_T lnum); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3479 static long bt_regexec_both(char_u *line, colnr_T col, proftime_T *tm); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3480 static long regtry(bt_regprog_T *prog, colnr_T col); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3481 static void cleanup_subexpr(void); |
| 7 | 3482 #ifdef FEAT_SYN_HL |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3483 static void cleanup_zsubexpr(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3484 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3485 static void save_subexpr(regbehind_T *bp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3486 static void restore_subexpr(regbehind_T *bp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3487 static void reg_nextline(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3488 static void reg_save(regsave_T *save, garray_T *gap); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3489 static void reg_restore(regsave_T *save, garray_T *gap); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3490 static int reg_save_equal(regsave_T *save); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3491 static void save_se_multi(save_se_T *savep, lpos_T *posp); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3492 static void save_se_one(save_se_T *savep, char_u **pp); |
| 7 | 3493 |
| 3494 /* Save the sub-expressions before attempting a match. */ | |
| 3495 #define save_se(savep, posp, pp) \ | |
| 3496 REG_MULTI ? save_se_multi((savep), (posp)) : save_se_one((savep), (pp)) | |
| 3497 | |
| 3498 /* After a failed match restore the sub-expressions. */ | |
| 3499 #define restore_se(savep, posp, pp) { \ | |
| 3500 if (REG_MULTI) \ | |
| 3501 *(posp) = (savep)->se_u.pos; \ | |
| 3502 else \ | |
| 3503 *(pp) = (savep)->se_u.ptr; } | |
| 3504 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3505 static int re_num_cmp(long_u val, char_u *scan); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3506 static int match_with_backref(linenr_T start_lnum, colnr_T start_col, linenr_T end_lnum, colnr_T end_col, int *bytelen); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3507 static int regmatch(char_u *prog); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3508 static int regrepeat(char_u *p, long maxcount); |
| 7 | 3509 |
| 3510 #ifdef DEBUG | |
| 3511 int regnarrate = 0; | |
| 3512 #endif | |
| 3513 | |
| 3514 /* | |
| 3515 * Internal copy of 'ignorecase'. It is set at each call to vim_regexec(). | |
| 3516 * Normally it gets the value of "rm_ic" or "rmm_ic", but when the pattern | |
| 3517 * contains '\c' or '\C' the value is overruled. | |
| 3518 */ | |
| 3519 static int ireg_ic; | |
| 3520 | |
| 3521 #ifdef FEAT_MBYTE | |
| 3522 /* | |
| 3523 * Similar to ireg_ic, but only for 'combining' characters. Set with \Z flag | |
| 3524 * in the regexp. Defaults to false, always. | |
| 3525 */ | |
| 3526 static int ireg_icombine; | |
| 3527 #endif | |
| 3528 | |
| 3529 /* | |
| 410 | 3530 * Copy of "rmm_maxcol": maximum column to search for a match. Zero when |
| 3531 * there is no maximum. | |
| 3532 */ | |
| 418 | 3533 static colnr_T ireg_maxcol; |
| 410 | 3534 |
| 3535 /* | |
| 7 | 3536 * Sometimes need to save a copy of a line. Since alloc()/free() is very |
| 3537 * slow, we keep one allocated piece of memory and only re-allocate it when | |
| 4444 | 3538 * it's too small. It's freed in bt_regexec_both() when finished. |
| 7 | 3539 */ |
| 1468 | 3540 static char_u *reg_tofree = NULL; |
| 7 | 3541 static unsigned reg_tofreelen; |
| 3542 | |
| 3543 /* | |
| 3544 * These variables are set when executing a regexp to speed up the execution. | |
| 1209 | 3545 * Which ones are set depends on whether a single-line or multi-line match is |
| 7 | 3546 * done: |
| 3547 * single-line multi-line | |
| 3548 * reg_match ®match_T NULL | |
| 3549 * reg_mmatch NULL ®mmatch_T | |
| 3550 * reg_startp reg_match->startp <invalid> | |
| 3551 * reg_endp reg_match->endp <invalid> | |
| 3552 * reg_startpos <invalid> reg_mmatch->startpos | |
| 3553 * reg_endpos <invalid> reg_mmatch->endpos | |
| 3554 * reg_win NULL window in which to search | |
| 4061 | 3555 * reg_buf curbuf buffer in which to search |
| 7 | 3556 * reg_firstlnum <invalid> first line in which to search |
| 3557 * reg_maxline 0 last line nr | |
| 3558 * reg_line_lbr FALSE or TRUE FALSE | |
| 3559 */ | |
| 3560 static regmatch_T *reg_match; | |
| 3561 static regmmatch_T *reg_mmatch; | |
| 3562 static char_u **reg_startp = NULL; | |
| 3563 static char_u **reg_endp = NULL; | |
| 3564 static lpos_T *reg_startpos = NULL; | |
| 3565 static lpos_T *reg_endpos = NULL; | |
| 3566 static win_T *reg_win; | |
| 3567 static buf_T *reg_buf; | |
| 3568 static linenr_T reg_firstlnum; | |
| 3569 static linenr_T reg_maxline; | |
| 3570 static int reg_line_lbr; /* "\n" in string is line break */ | |
| 3571 | |
| 270 | 3572 /* Values for rs_state in regitem_T. */ |
| 3573 typedef enum regstate_E | |
| 3574 { | |
| 3575 RS_NOPEN = 0 /* NOPEN and NCLOSE */ | |
| 3576 , RS_MOPEN /* MOPEN + [0-9] */ | |
| 3577 , RS_MCLOSE /* MCLOSE + [0-9] */ | |
| 3578 #ifdef FEAT_SYN_HL | |
| 3579 , RS_ZOPEN /* ZOPEN + [0-9] */ | |
| 3580 , RS_ZCLOSE /* ZCLOSE + [0-9] */ | |
| 3581 #endif | |
| 3582 , RS_BRANCH /* BRANCH */ | |
| 3583 , RS_BRCPLX_MORE /* BRACE_COMPLEX and trying one more match */ | |
| 3584 , RS_BRCPLX_LONG /* BRACE_COMPLEX and trying longest match */ | |
| 3585 , RS_BRCPLX_SHORT /* BRACE_COMPLEX and trying shortest match */ | |
| 3586 , RS_NOMATCH /* NOMATCH */ | |
| 3587 , RS_BEHIND1 /* BEHIND / NOBEHIND matching rest */ | |
| 3588 , RS_BEHIND2 /* BEHIND / NOBEHIND matching behind part */ | |
| 3589 , RS_STAR_LONG /* STAR/PLUS/BRACE_SIMPLE longest match */ | |
| 3590 , RS_STAR_SHORT /* STAR/PLUS/BRACE_SIMPLE shortest match */ | |
| 3591 } regstate_T; | |
| 3592 | |
| 3593 /* | |
| 3594 * When there are alternatives a regstate_T is put on the regstack to remember | |
| 3595 * what we are doing. | |
| 3596 * Before it may be another type of item, depending on rs_state, to remember | |
| 3597 * more things. | |
| 3598 */ | |
| 3599 typedef struct regitem_S | |
| 3600 { | |
| 3601 regstate_T rs_state; /* what we are doing, one of RS_ above */ | |
| 3602 char_u *rs_scan; /* current node in program */ | |
| 3603 union | |
| 3604 { | |
| 3605 save_se_T sesave; | |
| 3606 regsave_T regsave; | |
| 3607 } rs_un; /* room for saving reginput */ | |
| 1579 | 3608 short rs_no; /* submatch nr or BEHIND/NOBEHIND */ |
| 270 | 3609 } regitem_T; |
| 3610 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3611 static regitem_T *regstack_push(regstate_T state, char_u *scan); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3612 static void regstack_pop(char_u **scan); |
| 270 | 3613 |
| 3614 /* used for STAR, PLUS and BRACE_SIMPLE matching */ | |
| 3615 typedef struct regstar_S | |
| 3616 { | |
| 3617 int nextb; /* next byte */ | |
| 3618 int nextb_ic; /* next byte reverse case */ | |
| 3619 long count; | |
| 3620 long minval; | |
| 3621 long maxval; | |
| 3622 } regstar_T; | |
| 3623 | |
| 3624 /* used to store input position when a BACK was encountered, so that we now if | |
| 3625 * we made any progress since the last time. */ | |
| 3626 typedef struct backpos_S | |
| 3627 { | |
| 3628 char_u *bp_scan; /* "scan" where BACK was encountered */ | |
| 3629 regsave_T bp_pos; /* last input position */ | |
| 3630 } backpos_T; | |
| 3631 | |
| 3632 /* | |
| 1520 | 3633 * "regstack" and "backpos" are used by regmatch(). They are kept over calls |
| 3634 * to avoid invoking malloc() and free() often. | |
| 3635 * "regstack" is a stack with regitem_T items, sometimes preceded by regstar_T | |
| 3636 * or regbehind_T. | |
| 3637 * "backpos_T" is a table with backpos_T for BACK | |
| 270 | 3638 */ |
| 1520 | 3639 static garray_T regstack = {0, 0, 0, 0, NULL}; |
| 3640 static garray_T backpos = {0, 0, 0, 0, NULL}; | |
| 3641 | |
| 3642 /* | |
| 3643 * Both for regstack and backpos tables we use the following strategy of | |
| 3644 * allocation (to reduce malloc/free calls): | |
| 3645 * - Initial size is fairly small. | |
| 3646 * - When needed, the tables are grown bigger (8 times at first, double after | |
| 3647 * that). | |
| 3648 * - After executing the match we free the memory only if the array has grown. | |
| 3649 * Thus the memory is kept allocated when it's at the initial size. | |
| 3650 * This makes it fast while not keeping a lot of memory allocated. | |
| 3651 * A three times speed increase was observed when using many simple patterns. | |
| 3652 */ | |
| 3653 #define REGSTACK_INITIAL 2048 | |
| 3654 #define BACKPOS_INITIAL 64 | |
| 3655 | |
| 3656 #if defined(EXITFREE) || defined(PROTO) | |
| 3657 void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3658 free_regexp_stuff(void) |
| 1520 | 3659 { |
| 3660 ga_clear(®stack); | |
| 3661 ga_clear(&backpos); | |
| 3662 vim_free(reg_tofree); | |
| 3663 vim_free(reg_prev_sub); | |
| 3664 } | |
| 3665 #endif | |
| 270 | 3666 |
| 7 | 3667 /* |
| 3668 * Get pointer to the line "lnum", which is relative to "reg_firstlnum". | |
| 3669 */ | |
| 3670 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3671 reg_getline(linenr_T lnum) |
| 7 | 3672 { |
| 3673 /* when looking behind for a match/no-match lnum is negative. But we | |
| 3674 * can't go before line 1 */ | |
| 3675 if (reg_firstlnum + lnum < 1) | |
| 3676 return NULL; | |
| 482 | 3677 if (lnum > reg_maxline) |
| 481 | 3678 /* Must have matched the "\n" in the last line. */ |
| 3679 return (char_u *)""; | |
| 7 | 3680 return ml_get_buf(reg_buf, reg_firstlnum + lnum, FALSE); |
| 3681 } | |
| 3682 | |
| 3683 static regsave_T behind_pos; | |
| 3684 | |
| 3685 #ifdef FEAT_SYN_HL | |
| 3686 static char_u *reg_startzp[NSUBEXP]; /* Workspace to mark beginning */ | |
| 3687 static char_u *reg_endzp[NSUBEXP]; /* and end of \z(...\) matches */ | |
| 3688 static lpos_T reg_startzpos[NSUBEXP]; /* idem, beginning pos */ | |
| 3689 static lpos_T reg_endzpos[NSUBEXP]; /* idem, end pos */ | |
| 3690 #endif | |
| 3691 | |
| 3692 /* TRUE if using multi-line regexp. */ | |
| 3693 #define REG_MULTI (reg_match == NULL) | |
| 3694 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3695 static int bt_regexec_nl(regmatch_T *rmp, char_u *line, colnr_T col, int line_lbr); |
| 5838 | 3696 |
| 4444 | 3697 |
| 7 | 3698 /* |
| 3699 * Match a regexp against a string. | |
| 3700 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3701 * Uses curbuf for line count and 'iskeyword'. | |
| 5838 | 3702 * if "line_lbr" is TRUE consider a "\n" in "line" to be a line break. |
| 7 | 3703 * |
| 6390 | 3704 * Returns 0 for failure, number of lines contained in the match otherwise. |
| 7 | 3705 */ |
| 4444 | 3706 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3707 bt_regexec_nl( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3708 regmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3709 char_u *line, /* string to match against */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3710 colnr_T col, /* column to start looking for match */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3711 int line_lbr) |
| 7 | 3712 { |
| 3713 reg_match = rmp; | |
| 3714 reg_mmatch = NULL; | |
| 3715 reg_maxline = 0; | |
| 5838 | 3716 reg_line_lbr = line_lbr; |
| 4061 | 3717 reg_buf = curbuf; |
| 7 | 3718 reg_win = NULL; |
| 3719 ireg_ic = rmp->rm_ic; | |
| 3720 #ifdef FEAT_MBYTE | |
| 3721 ireg_icombine = FALSE; | |
| 3722 #endif | |
| 410 | 3723 ireg_maxcol = 0; |
| 6390 | 3724 |
| 3725 return bt_regexec_both(line, col, NULL); | |
| 7 | 3726 } |
| 3727 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3728 static long bt_regexec_multi(regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm); |
| 4444 | 3729 |
| 7 | 3730 /* |
| 3731 * Match a regexp against multiple lines. | |
| 3732 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3733 * Uses curbuf for line count and 'iskeyword'. | |
| 3734 * | |
| 3735 * Return zero if there is no match. Return number of lines contained in the | |
| 3736 * match otherwise. | |
| 3737 */ | |
| 4444 | 3738 static long |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3739 bt_regexec_multi( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3740 regmmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3741 win_T *win, /* window in which to search or NULL */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3742 buf_T *buf, /* buffer in which to search */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3743 linenr_T lnum, /* nr of line to start looking for match */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3744 colnr_T col, /* column to start looking for match */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3745 proftime_T *tm) /* timeout limit or NULL */ |
| 7 | 3746 { |
| 3747 reg_match = NULL; | |
| 3748 reg_mmatch = rmp; | |
| 3749 reg_buf = buf; | |
| 3750 reg_win = win; | |
| 3751 reg_firstlnum = lnum; | |
| 3752 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
| 3753 reg_line_lbr = FALSE; | |
| 3754 ireg_ic = rmp->rmm_ic; | |
| 3755 #ifdef FEAT_MBYTE | |
| 3756 ireg_icombine = FALSE; | |
| 3757 #endif | |
| 410 | 3758 ireg_maxcol = rmp->rmm_maxcol; |
| 7 | 3759 |
| 6390 | 3760 return bt_regexec_both(NULL, col, tm); |
| 7 | 3761 } |
| 3762 | |
| 3763 /* | |
| 3764 * Match a regexp against a string ("line" points to the string) or multiple | |
| 3765 * lines ("line" is NULL, use reg_getline()). | |
| 6390 | 3766 * Returns 0 for failure, number of lines contained in the match otherwise. |
| 7 | 3767 */ |
| 3768 static long | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3769 bt_regexec_both( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3770 char_u *line, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3771 colnr_T col, /* column to start looking for match */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3772 proftime_T *tm UNUSED) /* timeout limit or NULL */ |
| 7 | 3773 { |
| 6390 | 3774 bt_regprog_T *prog; |
| 3775 char_u *s; | |
| 3776 long retval = 0L; | |
| 7 | 3777 |
| 1520 | 3778 /* Create "regstack" and "backpos" if they are not allocated yet. |
| 3779 * We allocate *_INITIAL amount of bytes first and then set the grow size | |
| 3780 * to much bigger value to avoid many malloc calls in case of deep regular | |
| 3781 * expressions. */ | |
| 3782 if (regstack.ga_data == NULL) | |
| 3783 { | |
| 3784 /* Use an item size of 1 byte, since we push different things | |
| 3785 * onto the regstack. */ | |
| 3786 ga_init2(®stack, 1, REGSTACK_INITIAL); | |
| 7009 | 3787 (void)ga_grow(®stack, REGSTACK_INITIAL); |
| 1520 | 3788 regstack.ga_growsize = REGSTACK_INITIAL * 8; |
| 3789 } | |
| 3790 | |
| 3791 if (backpos.ga_data == NULL) | |
| 3792 { | |
| 3793 ga_init2(&backpos, sizeof(backpos_T), BACKPOS_INITIAL); | |
| 7009 | 3794 (void)ga_grow(&backpos, BACKPOS_INITIAL); |
| 1520 | 3795 backpos.ga_growsize = BACKPOS_INITIAL * 8; |
| 3796 } | |
| 270 | 3797 |
| 7 | 3798 if (REG_MULTI) |
| 3799 { | |
| 4444 | 3800 prog = (bt_regprog_T *)reg_mmatch->regprog; |
| 7 | 3801 line = reg_getline((linenr_T)0); |
| 3802 reg_startpos = reg_mmatch->startpos; | |
| 3803 reg_endpos = reg_mmatch->endpos; | |
| 3804 } | |
| 3805 else | |
| 3806 { | |
| 4444 | 3807 prog = (bt_regprog_T *)reg_match->regprog; |
| 7 | 3808 reg_startp = reg_match->startp; |
| 3809 reg_endp = reg_match->endp; | |
| 3810 } | |
| 3811 | |
| 3812 /* Be paranoid... */ | |
| 3813 if (prog == NULL || line == NULL) | |
| 3814 { | |
| 3815 EMSG(_(e_null)); | |
| 3816 goto theend; | |
| 3817 } | |
| 3818 | |
| 3819 /* Check validity of program. */ | |
| 3820 if (prog_magic_wrong()) | |
| 3821 goto theend; | |
| 3822 | |
| 410 | 3823 /* If the start column is past the maximum column: no need to try. */ |
| 3824 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3825 goto theend; | |
| 3826 | |
| 7 | 3827 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ |
| 3828 if (prog->regflags & RF_ICASE) | |
| 3829 ireg_ic = TRUE; | |
| 3830 else if (prog->regflags & RF_NOICASE) | |
| 3831 ireg_ic = FALSE; | |
| 3832 | |
| 3833 #ifdef FEAT_MBYTE | |
| 3834 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
| 3835 if (prog->regflags & RF_ICOMBINE) | |
| 3836 ireg_icombine = TRUE; | |
| 3837 #endif | |
| 3838 | |
| 3839 /* If there is a "must appear" string, look for it. */ | |
| 3840 if (prog->regmust != NULL) | |
| 3841 { | |
| 3842 int c; | |
| 3843 | |
| 3844 #ifdef FEAT_MBYTE | |
| 3845 if (has_mbyte) | |
| 3846 c = (*mb_ptr2char)(prog->regmust); | |
| 3847 else | |
| 3848 #endif | |
| 3849 c = *prog->regmust; | |
| 3850 s = line + col; | |
| 170 | 3851 |
| 3852 /* | |
| 3853 * This is used very often, esp. for ":global". Use three versions of | |
| 3854 * the loop to avoid overhead of conditions. | |
| 3855 */ | |
| 3856 if (!ireg_ic | |
| 3857 #ifdef FEAT_MBYTE | |
| 3858 && !has_mbyte | |
| 3859 #endif | |
| 3860 ) | |
| 3861 while ((s = vim_strbyte(s, c)) != NULL) | |
| 3862 { | |
| 3863 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3864 break; /* Found it. */ | |
| 3865 ++s; | |
| 3866 } | |
| 3867 #ifdef FEAT_MBYTE | |
| 3868 else if (!ireg_ic || (!enc_utf8 && mb_char2len(c) > 1)) | |
| 3869 while ((s = vim_strchr(s, c)) != NULL) | |
| 3870 { | |
| 3871 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3872 break; /* Found it. */ | |
| 3873 mb_ptr_adv(s); | |
| 3874 } | |
| 3875 #endif | |
| 3876 else | |
| 3877 while ((s = cstrchr(s, c)) != NULL) | |
| 3878 { | |
| 3879 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3880 break; /* Found it. */ | |
| 3881 mb_ptr_adv(s); | |
| 3882 } | |
| 7 | 3883 if (s == NULL) /* Not present. */ |
| 3884 goto theend; | |
| 3885 } | |
| 3886 | |
| 3887 regline = line; | |
| 3888 reglnum = 0; | |
| 2578 | 3889 reg_toolong = FALSE; |
| 7 | 3890 |
| 3891 /* Simplest case: Anchored match need be tried only once. */ | |
| 3892 if (prog->reganch) | |
| 3893 { | |
| 3894 int c; | |
| 3895 | |
| 3896 #ifdef FEAT_MBYTE | |
| 3897 if (has_mbyte) | |
| 3898 c = (*mb_ptr2char)(regline + col); | |
| 3899 else | |
| 3900 #endif | |
| 3901 c = regline[col]; | |
| 3902 if (prog->regstart == NUL | |
| 3903 || prog->regstart == c | |
| 3904 || (ireg_ic && (( | |
| 3905 #ifdef FEAT_MBYTE | |
| 3906 (enc_utf8 && utf_fold(prog->regstart) == utf_fold(c))) | |
| 3907 || (c < 255 && prog->regstart < 255 && | |
| 3908 #endif | |
| 1347 | 3909 MB_TOLOWER(prog->regstart) == MB_TOLOWER(c))))) |
| 7 | 3910 retval = regtry(prog, col); |
| 3911 else | |
| 3912 retval = 0; | |
| 3913 } | |
| 3914 else | |
| 3915 { | |
| 1521 | 3916 #ifdef FEAT_RELTIME |
| 3917 int tm_count = 0; | |
| 3918 #endif | |
| 7 | 3919 /* Messy cases: unanchored match. */ |
| 180 | 3920 while (!got_int) |
| 7 | 3921 { |
| 3922 if (prog->regstart != NUL) | |
| 3923 { | |
| 170 | 3924 /* Skip until the char we know it must start with. |
| 3925 * Used often, do some work to avoid call overhead. */ | |
| 3926 if (!ireg_ic | |
| 3927 #ifdef FEAT_MBYTE | |
| 3928 && !has_mbyte | |
| 3929 #endif | |
| 3930 ) | |
| 3931 s = vim_strbyte(regline + col, prog->regstart); | |
| 3932 else | |
| 3933 s = cstrchr(regline + col, prog->regstart); | |
| 7 | 3934 if (s == NULL) |
| 3935 { | |
| 3936 retval = 0; | |
| 3937 break; | |
| 3938 } | |
| 3939 col = (int)(s - regline); | |
| 3940 } | |
| 3941 | |
| 410 | 3942 /* Check for maximum column to try. */ |
| 3943 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3944 { | |
| 3945 retval = 0; | |
| 3946 break; | |
| 3947 } | |
| 3948 | |
| 7 | 3949 retval = regtry(prog, col); |
| 3950 if (retval > 0) | |
| 3951 break; | |
| 3952 | |
| 3953 /* if not currently on the first line, get it again */ | |
| 3954 if (reglnum != 0) | |
| 3955 { | |
| 481 | 3956 reglnum = 0; |
| 7 | 3957 regline = reg_getline((linenr_T)0); |
| 3958 } | |
| 3959 if (regline[col] == NUL) | |
| 3960 break; | |
| 3961 #ifdef FEAT_MBYTE | |
| 3962 if (has_mbyte) | |
| 474 | 3963 col += (*mb_ptr2len)(regline + col); |
| 7 | 3964 else |
| 3965 #endif | |
| 3966 ++col; | |
| 1521 | 3967 #ifdef FEAT_RELTIME |
| 3968 /* Check for timeout once in a twenty times to avoid overhead. */ | |
| 3969 if (tm != NULL && ++tm_count == 20) | |
| 3970 { | |
| 3971 tm_count = 0; | |
| 3972 if (profile_passed_limit(tm)) | |
| 3973 break; | |
| 3974 } | |
| 3975 #endif | |
| 7 | 3976 } |
| 3977 } | |
| 3978 | |
| 3979 theend: | |
| 1520 | 3980 /* Free "reg_tofree" when it's a bit big. |
| 3981 * Free regstack and backpos if they are bigger than their initial size. */ | |
| 3982 if (reg_tofreelen > 400) | |
| 3983 { | |
| 3984 vim_free(reg_tofree); | |
| 3985 reg_tofree = NULL; | |
| 3986 } | |
| 3987 if (regstack.ga_maxlen > REGSTACK_INITIAL) | |
| 3988 ga_clear(®stack); | |
| 3989 if (backpos.ga_maxlen > BACKPOS_INITIAL) | |
| 3990 ga_clear(&backpos); | |
| 270 | 3991 |
| 7 | 3992 return retval; |
| 3993 } | |
| 3994 | |
| 3995 #ifdef FEAT_SYN_HL | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3996 static reg_extmatch_T *make_extmatch(void); |
| 7 | 3997 |
| 3998 /* | |
| 3999 * Create a new extmatch and mark it as referenced once. | |
| 4000 */ | |
| 4001 static reg_extmatch_T * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4002 make_extmatch(void) |
| 7 | 4003 { |
| 4004 reg_extmatch_T *em; | |
| 4005 | |
| 4006 em = (reg_extmatch_T *)alloc_clear((unsigned)sizeof(reg_extmatch_T)); | |
| 4007 if (em != NULL) | |
| 4008 em->refcnt = 1; | |
| 4009 return em; | |
| 4010 } | |
| 4011 | |
| 4012 /* | |
| 4013 * Add a reference to an extmatch. | |
| 4014 */ | |
| 4015 reg_extmatch_T * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4016 ref_extmatch(reg_extmatch_T *em) |
| 7 | 4017 { |
| 4018 if (em != NULL) | |
| 4019 em->refcnt++; | |
| 4020 return em; | |
| 4021 } | |
| 4022 | |
| 4023 /* | |
| 4024 * Remove a reference to an extmatch. If there are no references left, free | |
| 4025 * the info. | |
| 4026 */ | |
| 4027 void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4028 unref_extmatch(reg_extmatch_T *em) |
| 7 | 4029 { |
| 4030 int i; | |
| 4031 | |
| 4032 if (em != NULL && --em->refcnt <= 0) | |
| 4033 { | |
| 4034 for (i = 0; i < NSUBEXP; ++i) | |
| 4035 vim_free(em->matches[i]); | |
| 4036 vim_free(em); | |
| 4037 } | |
| 4038 } | |
| 4039 #endif | |
| 4040 | |
| 4041 /* | |
| 4042 * regtry - try match of "prog" with at regline["col"]. | |
| 4043 * Returns 0 for failure, number of lines contained in the match otherwise. | |
| 4044 */ | |
| 4045 static long | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4046 regtry(bt_regprog_T *prog, colnr_T col) |
| 7 | 4047 { |
| 4048 reginput = regline + col; | |
| 4049 need_clear_subexpr = TRUE; | |
| 4050 #ifdef FEAT_SYN_HL | |
| 4051 /* Clear the external match subpointers if necessary. */ | |
| 4052 if (prog->reghasz == REX_SET) | |
| 4053 need_clear_zsubexpr = TRUE; | |
| 4054 #endif | |
| 4055 | |
| 189 | 4056 if (regmatch(prog->program + 1) == 0) |
| 4057 return 0; | |
| 4058 | |
| 4059 cleanup_subexpr(); | |
| 4060 if (REG_MULTI) | |
| 7 | 4061 { |
| 189 | 4062 if (reg_startpos[0].lnum < 0) |
| 7 | 4063 { |
| 189 | 4064 reg_startpos[0].lnum = 0; |
| 4065 reg_startpos[0].col = col; | |
| 4066 } | |
| 4067 if (reg_endpos[0].lnum < 0) | |
| 4068 { | |
| 4069 reg_endpos[0].lnum = reglnum; | |
| 4070 reg_endpos[0].col = (int)(reginput - regline); | |
| 7 | 4071 } |
| 4072 else | |
| 189 | 4073 /* Use line number of "\ze". */ |
| 4074 reglnum = reg_endpos[0].lnum; | |
| 4075 } | |
| 4076 else | |
| 4077 { | |
| 4078 if (reg_startp[0] == NULL) | |
| 4079 reg_startp[0] = regline + col; | |
| 4080 if (reg_endp[0] == NULL) | |
| 4081 reg_endp[0] = reginput; | |
| 4082 } | |
| 7 | 4083 #ifdef FEAT_SYN_HL |
| 189 | 4084 /* Package any found \z(...\) matches for export. Default is none. */ |
| 4085 unref_extmatch(re_extmatch_out); | |
| 4086 re_extmatch_out = NULL; | |
| 4087 | |
| 4088 if (prog->reghasz == REX_SET) | |
| 4089 { | |
| 4090 int i; | |
| 4091 | |
| 4092 cleanup_zsubexpr(); | |
| 4093 re_extmatch_out = make_extmatch(); | |
| 4094 for (i = 0; i < NSUBEXP; i++) | |
| 7 | 4095 { |
| 189 | 4096 if (REG_MULTI) |
| 7 | 4097 { |
| 189 | 4098 /* Only accept single line matches. */ |
| 4099 if (reg_startzpos[i].lnum >= 0 | |
| 5820 | 4100 && reg_endzpos[i].lnum == reg_startzpos[i].lnum |
| 4101 && reg_endzpos[i].col >= reg_startzpos[i].col) | |
| 189 | 4102 re_extmatch_out->matches[i] = |
| 4103 vim_strnsave(reg_getline(reg_startzpos[i].lnum) | |
| 7 | 4104 + reg_startzpos[i].col, |
| 189 | 4105 reg_endzpos[i].col - reg_startzpos[i].col); |
| 4106 } | |
| 4107 else | |
| 4108 { | |
| 4109 if (reg_startzp[i] != NULL && reg_endzp[i] != NULL) | |
| 4110 re_extmatch_out->matches[i] = | |
| 7 | 4111 vim_strnsave(reg_startzp[i], |
| 189 | 4112 (int)(reg_endzp[i] - reg_startzp[i])); |
| 7 | 4113 } |
| 4114 } | |
| 189 | 4115 } |
| 7 | 4116 #endif |
| 189 | 4117 return 1 + reglnum; |
| 7 | 4118 } |
| 4119 | |
| 4120 #ifdef FEAT_MBYTE | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
4121 static int reg_prev_class(void); |
| 7 | 4122 |
| 4123 /* | |
| 4124 * Get class of previous character. | |
| 4125 */ | |
| 4126 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4127 reg_prev_class(void) |
| 7 | 4128 { |
| 4129 if (reginput > regline) | |
| 4069 | 4130 return mb_get_class_buf(reginput - 1 |
| 4131 - (*mb_head_off)(regline, reginput - 1), reg_buf); | |
| 7 | 4132 return -1; |
| 4133 } | |
| 5735 | 4134 #endif |
| 4135 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
4136 static int reg_match_visual(void); |
|
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4137 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4138 /* |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4139 * Return TRUE if the current reginput position matches the Visual area. |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4140 */ |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4141 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4142 reg_match_visual(void) |
|
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4143 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4144 pos_T top, bot; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4145 linenr_T lnum; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4146 colnr_T col; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4147 win_T *wp = reg_win == NULL ? curwin : reg_win; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4148 int mode; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4149 colnr_T start, end; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4150 colnr_T start2, end2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4151 colnr_T cols; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4152 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4153 /* Check if the buffer is the current buffer. */ |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4154 if (reg_buf != curbuf || VIsual.lnum == 0) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4155 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4156 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4157 if (VIsual_active) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4158 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4159 if (lt(VIsual, wp->w_cursor)) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4160 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4161 top = VIsual; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4162 bot = wp->w_cursor; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4163 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4164 else |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4165 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4166 top = wp->w_cursor; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4167 bot = VIsual; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4168 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4169 mode = VIsual_mode; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4170 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4171 else |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4172 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4173 if (lt(curbuf->b_visual.vi_start, curbuf->b_visual.vi_end)) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4174 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4175 top = curbuf->b_visual.vi_start; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4176 bot = curbuf->b_visual.vi_end; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4177 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4178 else |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4179 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4180 top = curbuf->b_visual.vi_end; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4181 bot = curbuf->b_visual.vi_start; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4182 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4183 mode = curbuf->b_visual.vi_mode; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4184 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4185 lnum = reglnum + reg_firstlnum; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4186 if (lnum < top.lnum || lnum > bot.lnum) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4187 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4188 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4189 if (mode == 'v') |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4190 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4191 col = (colnr_T)(reginput - regline); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4192 if ((lnum == top.lnum && col < top.col) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4193 || (lnum == bot.lnum && col >= bot.col + (*p_sel != 'e'))) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4194 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4195 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4196 else if (mode == Ctrl_V) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4197 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4198 getvvcol(wp, &top, &start, NULL, &end); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4199 getvvcol(wp, &bot, &start2, NULL, &end2); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4200 if (start2 < start) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4201 start = start2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4202 if (end2 > end) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4203 end = end2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4204 if (top.col == MAXCOL || bot.col == MAXCOL) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4205 end = MAXCOL; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4206 cols = win_linetabsize(wp, regline, (colnr_T)(reginput - regline)); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4207 if (cols < start || cols > end - (*p_sel == 'e')) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4208 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4209 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4210 return TRUE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4211 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4212 |
| 39 | 4213 #define ADVANCE_REGINPUT() mb_ptr_adv(reginput) |
| 7 | 4214 |
| 4215 /* | |
| 4216 * The arguments from BRACE_LIMITS are stored here. They are actually local | |
| 4217 * to regmatch(), but they are here to reduce the amount of stack space used | |
| 4218 * (it can be called recursively many times). | |
| 4219 */ | |
| 4220 static long bl_minval; | |
| 4221 static long bl_maxval; | |
| 4222 | |
| 4223 /* | |
| 4224 * regmatch - main matching routine | |
| 4225 * | |
| 180 | 4226 * Conceptually the strategy is simple: Check to see whether the current node |
| 4227 * matches, push an item onto the regstack and loop to see whether the rest | |
| 4228 * matches, and then act accordingly. In practice we make some effort to | |
| 4229 * avoid using the regstack, in particular by going through "ordinary" nodes | |
| 4230 * (that don't need to know whether the rest of the match failed) by a nested | |
| 4231 * loop. | |
| 7 | 4232 * |
| 4233 * Returns TRUE when there is a match. Leaves reginput and reglnum just after | |
| 4234 * the last matched character. | |
| 4235 * Returns FALSE when there is no match. Leaves reginput and reglnum in an | |
| 4236 * undefined state! | |
| 4237 */ | |
| 4238 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4239 regmatch( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4240 char_u *scan) /* Current node. */ |
| 7 | 4241 { |
| 180 | 4242 char_u *next; /* Next node. */ |
| 4243 int op; | |
| 4244 int c; | |
| 4245 regitem_T *rp; | |
| 4246 int no; | |
| 4247 int status; /* one of the RA_ values: */ | |
| 4248 #define RA_FAIL 1 /* something failed, abort */ | |
| 4249 #define RA_CONT 2 /* continue in inner loop */ | |
| 4250 #define RA_BREAK 3 /* break inner loop */ | |
| 4251 #define RA_MATCH 4 /* successful match */ | |
| 4252 #define RA_NOMATCH 5 /* didn't match */ | |
| 270 | 4253 |
| 1520 | 4254 /* Make "regstack" and "backpos" empty. They are allocated and freed in |
| 4444 | 4255 * bt_regexec_both() to reduce malloc()/free() calls. */ |
| 270 | 4256 regstack.ga_len = 0; |
| 4257 backpos.ga_len = 0; | |
| 233 | 4258 |
| 180 | 4259 /* |
| 233 | 4260 * Repeat until "regstack" is empty. |
| 180 | 4261 */ |
| 4262 for (;;) | |
| 4263 { | |
| 5310 | 4264 /* Some patterns may take a long time to match, e.g., "\([a-z]\+\)\+Q". |
| 4265 * Allow interrupting them with CTRL-C. */ | |
| 7 | 4266 fast_breakcheck(); |
| 4267 | |
| 4268 #ifdef DEBUG | |
| 4269 if (scan != NULL && regnarrate) | |
| 4270 { | |
| 4444 | 4271 mch_errmsg((char *)regprop(scan)); |
| 7 | 4272 mch_errmsg("(\n"); |
| 4273 } | |
| 4274 #endif | |
| 180 | 4275 |
| 4276 /* | |
| 233 | 4277 * Repeat for items that can be matched sequentially, without using the |
| 180 | 4278 * regstack. |
| 4279 */ | |
| 4280 for (;;) | |
| 7 | 4281 { |
| 180 | 4282 if (got_int || scan == NULL) |
| 4283 { | |
| 4284 status = RA_FAIL; | |
| 4285 break; | |
| 4286 } | |
| 4287 status = RA_CONT; | |
| 4288 | |
| 7 | 4289 #ifdef DEBUG |
| 4290 if (regnarrate) | |
| 4291 { | |
| 4444 | 4292 mch_errmsg((char *)regprop(scan)); |
| 7 | 4293 mch_errmsg("...\n"); |
| 4294 # ifdef FEAT_SYN_HL | |
| 4295 if (re_extmatch_in != NULL) | |
| 4296 { | |
| 4297 int i; | |
| 4298 | |
| 4299 mch_errmsg(_("External submatches:\n")); | |
| 4300 for (i = 0; i < NSUBEXP; i++) | |
| 4301 { | |
| 4302 mch_errmsg(" \""); | |
| 4303 if (re_extmatch_in->matches[i] != NULL) | |
| 4444 | 4304 mch_errmsg((char *)re_extmatch_in->matches[i]); |
| 7 | 4305 mch_errmsg("\"\n"); |
| 4306 } | |
| 4307 } | |
| 4308 # endif | |
| 4309 } | |
| 4310 #endif | |
| 4311 next = regnext(scan); | |
| 4312 | |
| 4313 op = OP(scan); | |
| 4314 /* Check for character class with NL added. */ | |
| 1018 | 4315 if (!reg_line_lbr && WITH_NL(op) && REG_MULTI |
| 4316 && *reginput == NUL && reglnum <= reg_maxline) | |
| 7 | 4317 { |
| 4318 reg_nextline(); | |
| 4319 } | |
| 4320 else if (reg_line_lbr && WITH_NL(op) && *reginput == '\n') | |
| 4321 { | |
| 4322 ADVANCE_REGINPUT(); | |
| 4323 } | |
| 4324 else | |
| 4325 { | |
| 4326 if (WITH_NL(op)) | |
| 180 | 4327 op -= ADD_NL; |
| 7 | 4328 #ifdef FEAT_MBYTE |
| 4329 if (has_mbyte) | |
| 4330 c = (*mb_ptr2char)(reginput); | |
| 4331 else | |
| 4332 #endif | |
| 4333 c = *reginput; | |
| 4334 switch (op) | |
| 4335 { | |
| 4336 case BOL: | |
| 4337 if (reginput != regline) | |
| 180 | 4338 status = RA_NOMATCH; |
| 7 | 4339 break; |
| 4340 | |
| 4341 case EOL: | |
| 4342 if (c != NUL) | |
| 180 | 4343 status = RA_NOMATCH; |
| 7 | 4344 break; |
| 4345 | |
| 4346 case RE_BOF: | |
| 1458 | 4347 /* We're not at the beginning of the file when below the first |
| 4348 * line where we started, not at the start of the line or we | |
| 4349 * didn't start at the first line of the buffer. */ | |
| 7 | 4350 if (reglnum != 0 || reginput != regline |
| 1458 | 4351 || (REG_MULTI && reg_firstlnum > 1)) |
| 180 | 4352 status = RA_NOMATCH; |
| 7 | 4353 break; |
| 4354 | |
| 4355 case RE_EOF: | |
| 4356 if (reglnum != reg_maxline || c != NUL) | |
| 180 | 4357 status = RA_NOMATCH; |
| 7 | 4358 break; |
| 4359 | |
| 4360 case CURSOR: | |
| 4361 /* Check if the buffer is in a window and compare the | |
| 4362 * reg_win->w_cursor position to the match position. */ | |
| 4363 if (reg_win == NULL | |
| 4364 || (reglnum + reg_firstlnum != reg_win->w_cursor.lnum) | |
| 4365 || ((colnr_T)(reginput - regline) != reg_win->w_cursor.col)) | |
| 180 | 4366 status = RA_NOMATCH; |
| 7 | 4367 break; |
| 4368 | |
| 639 | 4369 case RE_MARK: |
|
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4370 /* Compare the mark position to the match position. */ |
| 639 | 4371 { |
| 4372 int mark = OPERAND(scan)[0]; | |
| 4373 int cmp = OPERAND(scan)[1]; | |
| 4374 pos_T *pos; | |
| 4375 | |
| 4043 | 4376 pos = getmark_buf(reg_buf, mark, FALSE); |
| 1148 | 4377 if (pos == NULL /* mark doesn't exist */ |
|
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4378 || pos->lnum <= 0 /* mark isn't set in reg_buf */ |
| 639 | 4379 || (pos->lnum == reglnum + reg_firstlnum |
| 4380 ? (pos->col == (colnr_T)(reginput - regline) | |
| 4381 ? (cmp == '<' || cmp == '>') | |
| 4382 : (pos->col < (colnr_T)(reginput - regline) | |
| 4383 ? cmp != '>' | |
| 4384 : cmp != '<')) | |
| 4385 : (pos->lnum < reglnum + reg_firstlnum | |
| 4386 ? cmp != '>' | |
| 4387 : cmp != '<'))) | |
| 4388 status = RA_NOMATCH; | |
| 4389 } | |
| 4390 break; | |
| 4391 | |
| 4392 case RE_VISUAL: | |
|
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4393 if (!reg_match_visual()) |
| 639 | 4394 status = RA_NOMATCH; |
| 4395 break; | |
| 4396 | |
| 7 | 4397 case RE_LNUM: |
| 4398 if (!REG_MULTI || !re_num_cmp((long_u)(reglnum + reg_firstlnum), | |
| 4399 scan)) | |
| 180 | 4400 status = RA_NOMATCH; |
| 7 | 4401 break; |
| 4402 | |
| 4403 case RE_COL: | |
| 4404 if (!re_num_cmp((long_u)(reginput - regline) + 1, scan)) | |
| 180 | 4405 status = RA_NOMATCH; |
| 7 | 4406 break; |
| 4407 | |
| 4408 case RE_VCOL: | |
| 4409 if (!re_num_cmp((long_u)win_linetabsize( | |
| 4410 reg_win == NULL ? curwin : reg_win, | |
| 4411 regline, (colnr_T)(reginput - regline)) + 1, scan)) | |
| 180 | 4412 status = RA_NOMATCH; |
| 7 | 4413 break; |
| 4414 | |
| 4415 case BOW: /* \<word; reginput points to w */ | |
| 4416 if (c == NUL) /* Can't match at end of line */ | |
| 180 | 4417 status = RA_NOMATCH; |
| 7 | 4418 #ifdef FEAT_MBYTE |
| 180 | 4419 else if (has_mbyte) |
| 7 | 4420 { |
| 4421 int this_class; | |
| 4422 | |
| 4423 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4424 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4425 if (this_class <= 1) |
| 180 | 4426 status = RA_NOMATCH; /* not on a word at all */ |
| 4427 else if (reg_prev_class() == this_class) | |
| 4428 status = RA_NOMATCH; /* previous char is in same word */ | |
| 7 | 4429 } |
| 4430 #endif | |
| 4431 else | |
| 4432 { | |
| 4061 | 4433 if (!vim_iswordc_buf(c, reg_buf) || (reginput > regline |
| 4434 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
| 180 | 4435 status = RA_NOMATCH; |
| 7 | 4436 } |
| 4437 break; | |
| 4438 | |
| 4439 case EOW: /* word\>; reginput points after d */ | |
| 4440 if (reginput == regline) /* Can't match at start of line */ | |
| 180 | 4441 status = RA_NOMATCH; |
| 7 | 4442 #ifdef FEAT_MBYTE |
| 180 | 4443 else if (has_mbyte) |
| 7 | 4444 { |
| 4445 int this_class, prev_class; | |
| 4446 | |
| 4447 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4448 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4449 prev_class = reg_prev_class(); |
| 180 | 4450 if (this_class == prev_class |
| 4451 || prev_class == 0 || prev_class == 1) | |
| 4452 status = RA_NOMATCH; | |
| 7 | 4453 } |
| 180 | 4454 #endif |
| 7 | 4455 else |
| 4456 { | |
| 4043 | 4457 if (!vim_iswordc_buf(reginput[-1], reg_buf) |
| 4458 || (reginput[0] != NUL && vim_iswordc_buf(c, reg_buf))) | |
| 180 | 4459 status = RA_NOMATCH; |
| 7 | 4460 } |
| 4461 break; /* Matched with EOW */ | |
| 4462 | |
| 4463 case ANY: | |
| 4084 | 4464 /* ANY does not match new lines. */ |
| 7 | 4465 if (c == NUL) |
| 180 | 4466 status = RA_NOMATCH; |
| 4467 else | |
| 4468 ADVANCE_REGINPUT(); | |
| 7 | 4469 break; |
| 4470 | |
| 4471 case IDENT: | |
| 4472 if (!vim_isIDc(c)) | |
| 180 | 4473 status = RA_NOMATCH; |
| 4474 else | |
| 4475 ADVANCE_REGINPUT(); | |
| 7 | 4476 break; |
| 4477 | |
| 4478 case SIDENT: | |
| 4479 if (VIM_ISDIGIT(*reginput) || !vim_isIDc(c)) | |
| 180 | 4480 status = RA_NOMATCH; |
| 4481 else | |
| 4482 ADVANCE_REGINPUT(); | |
| 7 | 4483 break; |
| 4484 | |
| 4485 case KWORD: | |
| 4069 | 4486 if (!vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4487 status = RA_NOMATCH; |
| 4488 else | |
| 4489 ADVANCE_REGINPUT(); | |
| 7 | 4490 break; |
| 4491 | |
| 4492 case SKWORD: | |
| 4069 | 4493 if (VIM_ISDIGIT(*reginput) || !vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4494 status = RA_NOMATCH; |
| 4495 else | |
| 4496 ADVANCE_REGINPUT(); | |
| 7 | 4497 break; |
| 4498 | |
| 4499 case FNAME: | |
| 4500 if (!vim_isfilec(c)) | |
| 180 | 4501 status = RA_NOMATCH; |
| 4502 else | |
| 4503 ADVANCE_REGINPUT(); | |
| 7 | 4504 break; |
| 4505 | |
| 4506 case SFNAME: | |
| 4507 if (VIM_ISDIGIT(*reginput) || !vim_isfilec(c)) | |
| 180 | 4508 status = RA_NOMATCH; |
| 4509 else | |
| 4510 ADVANCE_REGINPUT(); | |
| 7 | 4511 break; |
| 4512 | |
| 4513 case PRINT: | |
|
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
4514 if (!vim_isprintc(PTR2CHAR(reginput))) |
| 180 | 4515 status = RA_NOMATCH; |
| 4516 else | |
| 4517 ADVANCE_REGINPUT(); | |
| 7 | 4518 break; |
| 4519 | |
| 4520 case SPRINT: | |
|
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
4521 if (VIM_ISDIGIT(*reginput) || !vim_isprintc(PTR2CHAR(reginput))) |
| 180 | 4522 status = RA_NOMATCH; |
| 4523 else | |
| 4524 ADVANCE_REGINPUT(); | |
| 7 | 4525 break; |
| 4526 | |
| 4527 case WHITE: | |
| 4528 if (!vim_iswhite(c)) | |
| 180 | 4529 status = RA_NOMATCH; |
| 4530 else | |
| 4531 ADVANCE_REGINPUT(); | |
| 7 | 4532 break; |
| 4533 | |
| 4534 case NWHITE: | |
| 4535 if (c == NUL || vim_iswhite(c)) | |
| 180 | 4536 status = RA_NOMATCH; |
| 4537 else | |
| 4538 ADVANCE_REGINPUT(); | |
| 7 | 4539 break; |
| 4540 | |
| 4541 case DIGIT: | |
| 4542 if (!ri_digit(c)) | |
| 180 | 4543 status = RA_NOMATCH; |
| 4544 else | |
| 4545 ADVANCE_REGINPUT(); | |
| 7 | 4546 break; |
| 4547 | |
| 4548 case NDIGIT: | |
| 4549 if (c == NUL || ri_digit(c)) | |
| 180 | 4550 status = RA_NOMATCH; |
| 4551 else | |
| 4552 ADVANCE_REGINPUT(); | |
| 7 | 4553 break; |
| 4554 | |
| 4555 case HEX: | |
| 4556 if (!ri_hex(c)) | |
| 180 | 4557 status = RA_NOMATCH; |
| 4558 else | |
| 4559 ADVANCE_REGINPUT(); | |
| 7 | 4560 break; |
| 4561 | |
| 4562 case NHEX: | |
| 4563 if (c == NUL || ri_hex(c)) | |
| 180 | 4564 status = RA_NOMATCH; |
| 4565 else | |
| 4566 ADVANCE_REGINPUT(); | |
| 7 | 4567 break; |
| 4568 | |
| 4569 case OCTAL: | |
| 4570 if (!ri_octal(c)) | |
| 180 | 4571 status = RA_NOMATCH; |
| 4572 else | |
| 4573 ADVANCE_REGINPUT(); | |
| 7 | 4574 break; |
| 4575 | |
| 4576 case NOCTAL: | |
| 4577 if (c == NUL || ri_octal(c)) | |
| 180 | 4578 status = RA_NOMATCH; |
| 4579 else | |
| 4580 ADVANCE_REGINPUT(); | |
| 7 | 4581 break; |
| 4582 | |
| 4583 case WORD: | |
| 4584 if (!ri_word(c)) | |
| 180 | 4585 status = RA_NOMATCH; |
| 4586 else | |
| 4587 ADVANCE_REGINPUT(); | |
| 7 | 4588 break; |
| 4589 | |
| 4590 case NWORD: | |
| 4591 if (c == NUL || ri_word(c)) | |
| 180 | 4592 status = RA_NOMATCH; |
| 4593 else | |
| 4594 ADVANCE_REGINPUT(); | |
| 7 | 4595 break; |
| 4596 | |
| 4597 case HEAD: | |
| 4598 if (!ri_head(c)) | |
| 180 | 4599 status = RA_NOMATCH; |
| 4600 else | |
| 4601 ADVANCE_REGINPUT(); | |
| 7 | 4602 break; |
| 4603 | |
| 4604 case NHEAD: | |
| 4605 if (c == NUL || ri_head(c)) | |
| 180 | 4606 status = RA_NOMATCH; |
| 4607 else | |
| 4608 ADVANCE_REGINPUT(); | |
| 7 | 4609 break; |
| 4610 | |
| 4611 case ALPHA: | |
| 4612 if (!ri_alpha(c)) | |
| 180 | 4613 status = RA_NOMATCH; |
| 4614 else | |
| 4615 ADVANCE_REGINPUT(); | |
| 7 | 4616 break; |
| 4617 | |
| 4618 case NALPHA: | |
| 4619 if (c == NUL || ri_alpha(c)) | |
| 180 | 4620 status = RA_NOMATCH; |
| 4621 else | |
| 4622 ADVANCE_REGINPUT(); | |
| 7 | 4623 break; |
| 4624 | |
| 4625 case LOWER: | |
| 4626 if (!ri_lower(c)) | |
| 180 | 4627 status = RA_NOMATCH; |
| 4628 else | |
| 4629 ADVANCE_REGINPUT(); | |
| 7 | 4630 break; |
| 4631 | |
| 4632 case NLOWER: | |
| 4633 if (c == NUL || ri_lower(c)) | |
| 180 | 4634 status = RA_NOMATCH; |
| 4635 else | |
| 4636 ADVANCE_REGINPUT(); | |
| 7 | 4637 break; |
| 4638 | |
| 4639 case UPPER: | |
| 4640 if (!ri_upper(c)) | |
| 180 | 4641 status = RA_NOMATCH; |
| 4642 else | |
| 4643 ADVANCE_REGINPUT(); | |
| 7 | 4644 break; |
| 4645 | |
| 4646 case NUPPER: | |
| 4647 if (c == NUL || ri_upper(c)) | |
| 180 | 4648 status = RA_NOMATCH; |
| 4649 else | |
| 4650 ADVANCE_REGINPUT(); | |
| 7 | 4651 break; |
| 4652 | |
| 4653 case EXACTLY: | |
| 4654 { | |
| 4655 int len; | |
| 4656 char_u *opnd; | |
| 4657 | |
| 4658 opnd = OPERAND(scan); | |
| 4659 /* Inline the first byte, for speed. */ | |
| 4660 if (*opnd != *reginput | |
| 4661 && (!ireg_ic || ( | |
| 4662 #ifdef FEAT_MBYTE | |
| 4663 !enc_utf8 && | |
| 4664 #endif | |
| 1347 | 4665 MB_TOLOWER(*opnd) != MB_TOLOWER(*reginput)))) |
| 180 | 4666 status = RA_NOMATCH; |
| 4667 else if (*opnd == NUL) | |
| 7 | 4668 { |
| 4669 /* match empty string always works; happens when "~" is | |
| 4670 * empty. */ | |
| 4671 } | |
| 5899 | 4672 else |
| 4673 { | |
| 4674 if (opnd[1] == NUL | |
| 7 | 4675 #ifdef FEAT_MBYTE |
| 4676 && !(enc_utf8 && ireg_ic) | |
| 4677 #endif | |
| 4678 ) | |
| 5899 | 4679 { |
| 4680 len = 1; /* matched a single byte above */ | |
| 4681 } | |
| 4682 else | |
| 4683 { | |
| 4684 /* Need to match first byte again for multi-byte. */ | |
| 4685 len = (int)STRLEN(opnd); | |
| 4686 if (cstrncmp(opnd, reginput, &len) != 0) | |
| 4687 status = RA_NOMATCH; | |
| 4688 } | |
| 7 | 4689 #ifdef FEAT_MBYTE |
| 5901 | 4690 /* Check for following composing character, unless %C |
| 4691 * follows (skips over all composing chars). */ | |
| 5899 | 4692 if (status != RA_NOMATCH |
| 4693 && enc_utf8 | |
| 4694 && UTF_COMPOSINGLIKE(reginput, reginput + len) | |
| 5901 | 4695 && !ireg_icombine |
| 4696 && OP(next) != RE_COMPOSING) | |
| 7 | 4697 { |
| 4698 /* raaron: This code makes a composing character get | |
| 4699 * ignored, which is the correct behavior (sometimes) | |
| 4700 * for voweled Hebrew texts. */ | |
| 5899 | 4701 status = RA_NOMATCH; |
| 7 | 4702 } |
| 180 | 4703 #endif |
| 5899 | 4704 if (status != RA_NOMATCH) |
| 180 | 4705 reginput += len; |
| 7 | 4706 } |
| 4707 } | |
| 4708 break; | |
| 4709 | |
| 4710 case ANYOF: | |
| 4711 case ANYBUT: | |
| 4712 if (c == NUL) | |
| 180 | 4713 status = RA_NOMATCH; |
| 4714 else if ((cstrchr(OPERAND(scan), c) == NULL) == (op == ANYOF)) | |
| 4715 status = RA_NOMATCH; | |
| 4716 else | |
| 4717 ADVANCE_REGINPUT(); | |
| 7 | 4718 break; |
| 4719 | |
| 4720 #ifdef FEAT_MBYTE | |
| 4721 case MULTIBYTECODE: | |
| 4722 if (has_mbyte) | |
| 4723 { | |
| 4724 int i, len; | |
| 4725 char_u *opnd; | |
| 944 | 4726 int opndc = 0, inpc; |
| 7 | 4727 |
| 4728 opnd = OPERAND(scan); | |
| 4729 /* Safety check (just in case 'encoding' was changed since | |
| 4730 * compiling the program). */ | |
| 474 | 4731 if ((len = (*mb_ptr2len)(opnd)) < 2) |
| 180 | 4732 { |
| 4733 status = RA_NOMATCH; | |
| 4734 break; | |
| 4735 } | |
| 714 | 4736 if (enc_utf8) |
| 4737 opndc = mb_ptr2char(opnd); | |
| 4738 if (enc_utf8 && utf_iscomposing(opndc)) | |
| 4739 { | |
| 4740 /* When only a composing char is given match at any | |
| 4741 * position where that composing char appears. */ | |
| 4742 status = RA_NOMATCH; | |
| 6723 | 4743 for (i = 0; reginput[i] != NUL; |
| 4744 i += utf_ptr2len(reginput + i)) | |
| 180 | 4745 { |
| 714 | 4746 inpc = mb_ptr2char(reginput + i); |
| 4747 if (!utf_iscomposing(inpc)) | |
| 4748 { | |
| 4749 if (i > 0) | |
| 4750 break; | |
| 4751 } | |
| 4752 else if (opndc == inpc) | |
| 4753 { | |
| 4754 /* Include all following composing chars. */ | |
| 4755 len = i + mb_ptr2len(reginput + i); | |
| 4756 status = RA_MATCH; | |
| 4757 break; | |
| 4758 } | |
| 180 | 4759 } |
| 714 | 4760 } |
| 4761 else | |
| 4762 for (i = 0; i < len; ++i) | |
| 4763 if (opnd[i] != reginput[i]) | |
| 4764 { | |
| 4765 status = RA_NOMATCH; | |
| 4766 break; | |
| 4767 } | |
| 7 | 4768 reginput += len; |
| 4769 } | |
| 4770 else | |
| 180 | 4771 status = RA_NOMATCH; |
| 7 | 4772 break; |
| 4773 #endif | |
| 5901 | 4774 case RE_COMPOSING: |
| 4775 #ifdef FEAT_MBYTE | |
| 4776 if (enc_utf8) | |
| 4777 { | |
| 4778 /* Skip composing characters. */ | |
| 4779 while (utf_iscomposing(utf_ptr2char(reginput))) | |
| 4780 mb_cptr_adv(reginput); | |
| 4781 } | |
| 4782 #endif | |
| 4783 break; | |
| 7 | 4784 |
| 4785 case NOTHING: | |
| 4786 break; | |
| 4787 | |
| 4788 case BACK: | |
| 233 | 4789 { |
| 4790 int i; | |
| 4791 backpos_T *bp; | |
| 4792 | |
| 4793 /* | |
| 4794 * When we run into BACK we need to check if we don't keep | |
| 4795 * looping without matching any input. The second and later | |
| 4796 * times a BACK is encountered it fails if the input is still | |
| 4797 * at the same position as the previous time. | |
| 4798 * The positions are stored in "backpos" and found by the | |
| 4799 * current value of "scan", the position in the RE program. | |
| 4800 */ | |
| 4801 bp = (backpos_T *)backpos.ga_data; | |
| 4802 for (i = 0; i < backpos.ga_len; ++i) | |
| 4803 if (bp[i].bp_scan == scan) | |
| 4804 break; | |
| 4805 if (i == backpos.ga_len) | |
| 4806 { | |
| 4807 /* First time at this BACK, make room to store the pos. */ | |
| 4808 if (ga_grow(&backpos, 1) == FAIL) | |
| 4809 status = RA_FAIL; | |
| 4810 else | |
| 4811 { | |
| 4812 /* get "ga_data" again, it may have changed */ | |
| 4813 bp = (backpos_T *)backpos.ga_data; | |
| 4814 bp[i].bp_scan = scan; | |
| 4815 ++backpos.ga_len; | |
| 4816 } | |
| 4817 } | |
| 4818 else if (reg_save_equal(&bp[i].bp_pos)) | |
| 4819 /* Still at same position as last time, fail. */ | |
| 4820 status = RA_NOMATCH; | |
| 4821 | |
| 4822 if (status != RA_FAIL && status != RA_NOMATCH) | |
| 4823 reg_save(&bp[i].bp_pos, &backpos); | |
| 4824 } | |
| 179 | 4825 break; |
| 4826 | |
| 7 | 4827 case MOPEN + 0: /* Match start: \zs */ |
| 4828 case MOPEN + 1: /* \( */ | |
| 4829 case MOPEN + 2: | |
| 4830 case MOPEN + 3: | |
| 4831 case MOPEN + 4: | |
| 4832 case MOPEN + 5: | |
| 4833 case MOPEN + 6: | |
| 4834 case MOPEN + 7: | |
| 4835 case MOPEN + 8: | |
| 4836 case MOPEN + 9: | |
| 4837 { | |
| 4838 no = op - MOPEN; | |
| 4839 cleanup_subexpr(); | |
| 270 | 4840 rp = regstack_push(RS_MOPEN, scan); |
| 180 | 4841 if (rp == NULL) |
| 4842 status = RA_FAIL; | |
| 4843 else | |
| 4844 { | |
| 4845 rp->rs_no = no; | |
| 4846 save_se(&rp->rs_un.sesave, ®_startpos[no], | |
| 4847 ®_startp[no]); | |
| 4848 /* We simply continue and handle the result when done. */ | |
| 4849 } | |
| 7 | 4850 } |
| 180 | 4851 break; |
| 7 | 4852 |
| 4853 case NOPEN: /* \%( */ | |
| 4854 case NCLOSE: /* \) after \%( */ | |
| 270 | 4855 if (regstack_push(RS_NOPEN, scan) == NULL) |
| 180 | 4856 status = RA_FAIL; |
| 4857 /* We simply continue and handle the result when done. */ | |
| 4858 break; | |
| 7 | 4859 |
| 4860 #ifdef FEAT_SYN_HL | |
| 4861 case ZOPEN + 1: | |
| 4862 case ZOPEN + 2: | |
| 4863 case ZOPEN + 3: | |
| 4864 case ZOPEN + 4: | |
| 4865 case ZOPEN + 5: | |
| 4866 case ZOPEN + 6: | |
| 4867 case ZOPEN + 7: | |
| 4868 case ZOPEN + 8: | |
| 4869 case ZOPEN + 9: | |
| 4870 { | |
| 4871 no = op - ZOPEN; | |
| 4872 cleanup_zsubexpr(); | |
| 270 | 4873 rp = regstack_push(RS_ZOPEN, scan); |
| 180 | 4874 if (rp == NULL) |
| 4875 status = RA_FAIL; | |
| 4876 else | |
| 4877 { | |
| 4878 rp->rs_no = no; | |
| 4879 save_se(&rp->rs_un.sesave, ®_startzpos[no], | |
| 4880 ®_startzp[no]); | |
| 4881 /* We simply continue and handle the result when done. */ | |
| 4882 } | |
| 7 | 4883 } |
| 180 | 4884 break; |
| 7 | 4885 #endif |
| 4886 | |
| 4887 case MCLOSE + 0: /* Match end: \ze */ | |
| 4888 case MCLOSE + 1: /* \) */ | |
| 4889 case MCLOSE + 2: | |
| 4890 case MCLOSE + 3: | |
| 4891 case MCLOSE + 4: | |
| 4892 case MCLOSE + 5: | |
| 4893 case MCLOSE + 6: | |
| 4894 case MCLOSE + 7: | |
| 4895 case MCLOSE + 8: | |
| 4896 case MCLOSE + 9: | |
| 4897 { | |
| 4898 no = op - MCLOSE; | |
| 4899 cleanup_subexpr(); | |
| 270 | 4900 rp = regstack_push(RS_MCLOSE, scan); |
| 180 | 4901 if (rp == NULL) |
| 4902 status = RA_FAIL; | |
| 4903 else | |
| 4904 { | |
| 4905 rp->rs_no = no; | |
| 4906 save_se(&rp->rs_un.sesave, ®_endpos[no], ®_endp[no]); | |
| 4907 /* We simply continue and handle the result when done. */ | |
| 4908 } | |
| 7 | 4909 } |
| 180 | 4910 break; |
| 7 | 4911 |
| 4912 #ifdef FEAT_SYN_HL | |
| 4913 case ZCLOSE + 1: /* \) after \z( */ | |
| 4914 case ZCLOSE + 2: | |
| 4915 case ZCLOSE + 3: | |
| 4916 case ZCLOSE + 4: | |
| 4917 case ZCLOSE + 5: | |
| 4918 case ZCLOSE + 6: | |
| 4919 case ZCLOSE + 7: | |
| 4920 case ZCLOSE + 8: | |
| 4921 case ZCLOSE + 9: | |
| 4922 { | |
| 4923 no = op - ZCLOSE; | |
| 4924 cleanup_zsubexpr(); | |
| 270 | 4925 rp = regstack_push(RS_ZCLOSE, scan); |
| 180 | 4926 if (rp == NULL) |
| 4927 status = RA_FAIL; | |
| 4928 else | |
| 4929 { | |
| 4930 rp->rs_no = no; | |
| 4931 save_se(&rp->rs_un.sesave, ®_endzpos[no], | |
| 4932 ®_endzp[no]); | |
| 4933 /* We simply continue and handle the result when done. */ | |
| 4934 } | |
| 7 | 4935 } |
| 180 | 4936 break; |
| 7 | 4937 #endif |
| 4938 | |
| 4939 case BACKREF + 1: | |
| 4940 case BACKREF + 2: | |
| 4941 case BACKREF + 3: | |
| 4942 case BACKREF + 4: | |
| 4943 case BACKREF + 5: | |
| 4944 case BACKREF + 6: | |
| 4945 case BACKREF + 7: | |
| 4946 case BACKREF + 8: | |
| 4947 case BACKREF + 9: | |
| 4948 { | |
| 4949 int len; | |
| 4950 | |
| 4951 no = op - BACKREF; | |
| 4952 cleanup_subexpr(); | |
| 4953 if (!REG_MULTI) /* Single-line regexp */ | |
| 4954 { | |
| 1815 | 4955 if (reg_startp[no] == NULL || reg_endp[no] == NULL) |
| 7 | 4956 { |
| 4957 /* Backref was not set: Match an empty string. */ | |
| 4958 len = 0; | |
| 4959 } | |
| 4960 else | |
| 4961 { | |
| 4962 /* Compare current input with back-ref in the same | |
| 4963 * line. */ | |
| 4964 len = (int)(reg_endp[no] - reg_startp[no]); | |
| 4965 if (cstrncmp(reg_startp[no], reginput, &len) != 0) | |
| 180 | 4966 status = RA_NOMATCH; |
| 7 | 4967 } |
| 4968 } | |
| 4969 else /* Multi-line regexp */ | |
| 4970 { | |
| 1815 | 4971 if (reg_startpos[no].lnum < 0 || reg_endpos[no].lnum < 0) |
| 7 | 4972 { |
| 4973 /* Backref was not set: Match an empty string. */ | |
| 4974 len = 0; | |
| 4975 } | |
| 4976 else | |
| 4977 { | |
| 4978 if (reg_startpos[no].lnum == reglnum | |
| 4979 && reg_endpos[no].lnum == reglnum) | |
| 4980 { | |
| 4981 /* Compare back-ref within the current line. */ | |
| 4982 len = reg_endpos[no].col - reg_startpos[no].col; | |
| 4983 if (cstrncmp(regline + reg_startpos[no].col, | |
| 4984 reginput, &len) != 0) | |
| 180 | 4985 status = RA_NOMATCH; |
| 7 | 4986 } |
| 4987 else | |
| 4988 { | |
| 4989 /* Messy situation: Need to compare between two | |
| 4990 * lines. */ | |
|
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
4991 int r = match_with_backref( |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4992 reg_startpos[no].lnum, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4993 reg_startpos[no].col, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4994 reg_endpos[no].lnum, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4995 reg_endpos[no].col, |
|
4899
4837fd61be52
updated for version 7.3.1195
Bram Moolenaar <bram@vim.org>
parents:
4891
diff
changeset
|
4996 &len); |
|
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
4997 |
|
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
4998 if (r != RA_MATCH) |
|
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
4999 status = r; |
| 7 | 5000 } |
| 5001 } | |
| 5002 } | |
| 5003 | |
| 5004 /* Matched the backref, skip over it. */ | |
| 5005 reginput += len; | |
| 5006 } | |
| 5007 break; | |
| 5008 | |
| 5009 #ifdef FEAT_SYN_HL | |
| 5010 case ZREF + 1: | |
| 5011 case ZREF + 2: | |
| 5012 case ZREF + 3: | |
| 5013 case ZREF + 4: | |
| 5014 case ZREF + 5: | |
| 5015 case ZREF + 6: | |
| 5016 case ZREF + 7: | |
| 5017 case ZREF + 8: | |
| 5018 case ZREF + 9: | |
| 5019 { | |
| 5020 int len; | |
| 5021 | |
| 5022 cleanup_zsubexpr(); | |
| 5023 no = op - ZREF; | |
| 5024 if (re_extmatch_in != NULL | |
| 5025 && re_extmatch_in->matches[no] != NULL) | |
| 5026 { | |
| 5027 len = (int)STRLEN(re_extmatch_in->matches[no]); | |
| 5028 if (cstrncmp(re_extmatch_in->matches[no], | |
| 5029 reginput, &len) != 0) | |
| 180 | 5030 status = RA_NOMATCH; |
| 5031 else | |
| 5032 reginput += len; | |
| 7 | 5033 } |
| 5034 else | |
| 5035 { | |
| 5036 /* Backref was not set: Match an empty string. */ | |
| 5037 } | |
| 5038 } | |
| 5039 break; | |
| 5040 #endif | |
| 5041 | |
| 5042 case BRANCH: | |
| 5043 { | |
| 5044 if (OP(next) != BRANCH) /* No choice. */ | |
| 5045 next = OPERAND(scan); /* Avoid recursion. */ | |
| 5046 else | |
| 5047 { | |
| 270 | 5048 rp = regstack_push(RS_BRANCH, scan); |
| 180 | 5049 if (rp == NULL) |
| 5050 status = RA_FAIL; | |
| 5051 else | |
| 5052 status = RA_BREAK; /* rest is below */ | |
| 7 | 5053 } |
| 5054 } | |
| 5055 break; | |
| 5056 | |
| 5057 case BRACE_LIMITS: | |
| 5058 { | |
| 5059 if (OP(next) == BRACE_SIMPLE) | |
| 5060 { | |
| 5061 bl_minval = OPERAND_MIN(scan); | |
| 5062 bl_maxval = OPERAND_MAX(scan); | |
| 5063 } | |
| 5064 else if (OP(next) >= BRACE_COMPLEX | |
| 5065 && OP(next) < BRACE_COMPLEX + 10) | |
| 5066 { | |
| 5067 no = OP(next) - BRACE_COMPLEX; | |
| 5068 brace_min[no] = OPERAND_MIN(scan); | |
| 5069 brace_max[no] = OPERAND_MAX(scan); | |
| 5070 brace_count[no] = 0; | |
| 5071 } | |
| 5072 else | |
| 5073 { | |
| 5074 EMSG(_(e_internal)); /* Shouldn't happen */ | |
| 180 | 5075 status = RA_FAIL; |
| 7 | 5076 } |
| 5077 } | |
| 5078 break; | |
| 5079 | |
| 5080 case BRACE_COMPLEX + 0: | |
| 5081 case BRACE_COMPLEX + 1: | |
| 5082 case BRACE_COMPLEX + 2: | |
| 5083 case BRACE_COMPLEX + 3: | |
| 5084 case BRACE_COMPLEX + 4: | |
| 5085 case BRACE_COMPLEX + 5: | |
| 5086 case BRACE_COMPLEX + 6: | |
| 5087 case BRACE_COMPLEX + 7: | |
| 5088 case BRACE_COMPLEX + 8: | |
| 5089 case BRACE_COMPLEX + 9: | |
| 5090 { | |
| 5091 no = op - BRACE_COMPLEX; | |
| 5092 ++brace_count[no]; | |
| 5093 | |
| 5094 /* If not matched enough times yet, try one more */ | |
| 5095 if (brace_count[no] <= (brace_min[no] <= brace_max[no] | |
| 180 | 5096 ? brace_min[no] : brace_max[no])) |
| 7 | 5097 { |
| 270 | 5098 rp = regstack_push(RS_BRCPLX_MORE, scan); |
| 180 | 5099 if (rp == NULL) |
| 5100 status = RA_FAIL; | |
| 5101 else | |
| 5102 { | |
| 5103 rp->rs_no = no; | |
| 233 | 5104 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5105 next = OPERAND(scan); |
| 5106 /* We continue and handle the result when done. */ | |
| 5107 } | |
| 5108 break; | |
| 7 | 5109 } |
| 5110 | |
| 5111 /* If matched enough times, may try matching some more */ | |
| 5112 if (brace_min[no] <= brace_max[no]) | |
| 5113 { | |
| 5114 /* Range is the normal way around, use longest match */ | |
| 5115 if (brace_count[no] <= brace_max[no]) | |
| 5116 { | |
| 270 | 5117 rp = regstack_push(RS_BRCPLX_LONG, scan); |
| 180 | 5118 if (rp == NULL) |
| 5119 status = RA_FAIL; | |
| 5120 else | |
| 5121 { | |
| 5122 rp->rs_no = no; | |
| 233 | 5123 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5124 next = OPERAND(scan); |
| 5125 /* We continue and handle the result when done. */ | |
| 5126 } | |
| 7 | 5127 } |
| 5128 } | |
| 5129 else | |
| 5130 { | |
| 5131 /* Range is backwards, use shortest match first */ | |
| 5132 if (brace_count[no] <= brace_min[no]) | |
| 5133 { | |
| 270 | 5134 rp = regstack_push(RS_BRCPLX_SHORT, scan); |
| 180 | 5135 if (rp == NULL) |
| 5136 status = RA_FAIL; | |
| 5137 else | |
| 5138 { | |
| 233 | 5139 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5140 /* We continue and handle the result when done. */ |
| 5141 } | |
| 7 | 5142 } |
| 5143 } | |
| 5144 } | |
| 5145 break; | |
| 5146 | |
| 5147 case BRACE_SIMPLE: | |
| 5148 case STAR: | |
| 5149 case PLUS: | |
| 5150 { | |
| 180 | 5151 regstar_T rst; |
| 7 | 5152 |
| 5153 /* | |
| 5154 * Lookahead to avoid useless match attempts when we know | |
| 5155 * what character comes next. | |
| 5156 */ | |
| 5157 if (OP(next) == EXACTLY) | |
| 5158 { | |
| 180 | 5159 rst.nextb = *OPERAND(next); |
| 7 | 5160 if (ireg_ic) |
| 5161 { | |
| 1347 | 5162 if (MB_ISUPPER(rst.nextb)) |
| 5163 rst.nextb_ic = MB_TOLOWER(rst.nextb); | |
| 7 | 5164 else |
| 1347 | 5165 rst.nextb_ic = MB_TOUPPER(rst.nextb); |
| 7 | 5166 } |
| 5167 else | |
| 180 | 5168 rst.nextb_ic = rst.nextb; |
| 7 | 5169 } |
| 5170 else | |
| 5171 { | |
| 180 | 5172 rst.nextb = NUL; |
| 5173 rst.nextb_ic = NUL; | |
| 7 | 5174 } |
| 5175 if (op != BRACE_SIMPLE) | |
| 5176 { | |
| 180 | 5177 rst.minval = (op == STAR) ? 0 : 1; |
| 5178 rst.maxval = MAX_LIMIT; | |
| 7 | 5179 } |
| 5180 else | |
| 5181 { | |
| 180 | 5182 rst.minval = bl_minval; |
| 5183 rst.maxval = bl_maxval; | |
| 7 | 5184 } |
| 5185 | |
| 5186 /* | |
| 5187 * When maxval > minval, try matching as much as possible, up | |
| 5188 * to maxval. When maxval < minval, try matching at least the | |
| 5189 * minimal number (since the range is backwards, that's also | |
| 5190 * maxval!). | |
| 5191 */ | |
| 180 | 5192 rst.count = regrepeat(OPERAND(scan), rst.maxval); |
| 7 | 5193 if (got_int) |
| 180 | 5194 { |
| 5195 status = RA_FAIL; | |
| 5196 break; | |
| 5197 } | |
| 5198 if (rst.minval <= rst.maxval | |
| 5199 ? rst.count >= rst.minval : rst.count >= rst.maxval) | |
| 7 | 5200 { |
| 180 | 5201 /* It could match. Prepare for trying to match what |
| 5202 * follows. The code is below. Parameters are stored in | |
| 5203 * a regstar_T on the regstack. */ | |
| 212 | 5204 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5205 { |
| 5206 EMSG(_(e_maxmempat)); | |
| 5207 status = RA_FAIL; | |
| 5208 } | |
| 5209 else if (ga_grow(®stack, sizeof(regstar_T)) == FAIL) | |
| 180 | 5210 status = RA_FAIL; |
| 5211 else | |
| 7 | 5212 { |
| 180 | 5213 regstack.ga_len += sizeof(regstar_T); |
| 270 | 5214 rp = regstack_push(rst.minval <= rst.maxval |
| 233 | 5215 ? RS_STAR_LONG : RS_STAR_SHORT, scan); |
| 180 | 5216 if (rp == NULL) |
| 5217 status = RA_FAIL; | |
| 5218 else | |
| 7 | 5219 { |
| 180 | 5220 *(((regstar_T *)rp) - 1) = rst; |
| 5221 status = RA_BREAK; /* skip the restore bits */ | |
| 7 | 5222 } |
| 5223 } | |
| 5224 } | |
| 5225 else | |
| 180 | 5226 status = RA_NOMATCH; |
| 5227 | |
| 7 | 5228 } |
| 180 | 5229 break; |
| 7 | 5230 |
| 5231 case NOMATCH: | |
| 5232 case MATCH: | |
| 5233 case SUBPAT: | |
| 270 | 5234 rp = regstack_push(RS_NOMATCH, scan); |
| 180 | 5235 if (rp == NULL) |
| 5236 status = RA_FAIL; | |
| 5237 else | |
| 7 | 5238 { |
| 180 | 5239 rp->rs_no = op; |
| 233 | 5240 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5241 next = OPERAND(scan); |
| 5242 /* We continue and handle the result when done. */ | |
| 7 | 5243 } |
| 5244 break; | |
| 5245 | |
| 5246 case BEHIND: | |
| 5247 case NOBEHIND: | |
| 180 | 5248 /* Need a bit of room to store extra positions. */ |
| 212 | 5249 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5250 { |
| 5251 EMSG(_(e_maxmempat)); | |
| 5252 status = RA_FAIL; | |
| 5253 } | |
| 5254 else if (ga_grow(®stack, sizeof(regbehind_T)) == FAIL) | |
| 180 | 5255 status = RA_FAIL; |
| 5256 else | |
| 7 | 5257 { |
| 180 | 5258 regstack.ga_len += sizeof(regbehind_T); |
| 270 | 5259 rp = regstack_push(RS_BEHIND1, scan); |
| 180 | 5260 if (rp == NULL) |
| 5261 status = RA_FAIL; | |
| 5262 else | |
| 7 | 5263 { |
| 1579 | 5264 /* Need to save the subexpr to be able to restore them |
| 5265 * when there is a match but we don't use it. */ | |
| 5266 save_subexpr(((regbehind_T *)rp) - 1); | |
| 5267 | |
| 180 | 5268 rp->rs_no = op; |
| 233 | 5269 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5270 /* First try if what follows matches. If it does then we |
| 5271 * check the behind match by looping. */ | |
| 7 | 5272 } |
| 5273 } | |
| 180 | 5274 break; |
| 7 | 5275 |
| 5276 case BHPOS: | |
| 5277 if (REG_MULTI) | |
| 5278 { | |
| 5279 if (behind_pos.rs_u.pos.col != (colnr_T)(reginput - regline) | |
| 5280 || behind_pos.rs_u.pos.lnum != reglnum) | |
| 180 | 5281 status = RA_NOMATCH; |
| 7 | 5282 } |
| 5283 else if (behind_pos.rs_u.ptr != reginput) | |
| 180 | 5284 status = RA_NOMATCH; |
| 7 | 5285 break; |
| 5286 | |
| 5287 case NEWL: | |
| 1018 | 5288 if ((c != NUL || !REG_MULTI || reglnum > reg_maxline |
| 5289 || reg_line_lbr) && (c != '\n' || !reg_line_lbr)) | |
| 180 | 5290 status = RA_NOMATCH; |
| 5291 else if (reg_line_lbr) | |
| 7 | 5292 ADVANCE_REGINPUT(); |
| 5293 else | |
| 5294 reg_nextline(); | |
| 5295 break; | |
| 5296 | |
| 5297 case END: | |
| 180 | 5298 status = RA_MATCH; /* Success! */ |
| 5299 break; | |
| 7 | 5300 |
| 5301 default: | |
| 5302 EMSG(_(e_re_corr)); | |
| 5303 #ifdef DEBUG | |
| 5304 printf("Illegal op code %d\n", op); | |
| 5305 #endif | |
| 180 | 5306 status = RA_FAIL; |
| 5307 break; | |
| 7 | 5308 } |
| 5309 } | |
| 5310 | |
| 180 | 5311 /* If we can't continue sequentially, break the inner loop. */ |
| 5312 if (status != RA_CONT) | |
| 5313 break; | |
| 5314 | |
| 5315 /* Continue in inner loop, advance to next item. */ | |
| 7 | 5316 scan = next; |
| 180 | 5317 |
| 5318 } /* end of inner loop */ | |
| 7 | 5319 |
| 5320 /* | |
| 180 | 5321 * If there is something on the regstack execute the code for the state. |
| 233 | 5322 * If the state is popped then loop and use the older state. |
| 7 | 5323 */ |
| 180 | 5324 while (regstack.ga_len > 0 && status != RA_FAIL) |
| 5325 { | |
| 5326 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; | |
| 5327 switch (rp->rs_state) | |
| 5328 { | |
| 5329 case RS_NOPEN: | |
| 5330 /* Result is passed on as-is, simply pop the state. */ | |
| 270 | 5331 regstack_pop(&scan); |
| 180 | 5332 break; |
| 5333 | |
| 5334 case RS_MOPEN: | |
| 5335 /* Pop the state. Restore pointers when there is no match. */ | |
| 5336 if (status == RA_NOMATCH) | |
| 5337 restore_se(&rp->rs_un.sesave, ®_startpos[rp->rs_no], | |
| 5338 ®_startp[rp->rs_no]); | |
| 270 | 5339 regstack_pop(&scan); |
| 180 | 5340 break; |
| 5341 | |
| 5342 #ifdef FEAT_SYN_HL | |
| 5343 case RS_ZOPEN: | |
| 5344 /* Pop the state. Restore pointers when there is no match. */ | |
| 5345 if (status == RA_NOMATCH) | |
| 5346 restore_se(&rp->rs_un.sesave, ®_startzpos[rp->rs_no], | |
| 5347 ®_startzp[rp->rs_no]); | |
| 270 | 5348 regstack_pop(&scan); |
| 180 | 5349 break; |
| 5350 #endif | |
| 5351 | |
| 5352 case RS_MCLOSE: | |
| 5353 /* Pop the state. Restore pointers when there is no match. */ | |
| 5354 if (status == RA_NOMATCH) | |
| 5355 restore_se(&rp->rs_un.sesave, ®_endpos[rp->rs_no], | |
| 5356 ®_endp[rp->rs_no]); | |
| 270 | 5357 regstack_pop(&scan); |
| 180 | 5358 break; |
| 5359 | |
| 5360 #ifdef FEAT_SYN_HL | |
| 5361 case RS_ZCLOSE: | |
| 5362 /* Pop the state. Restore pointers when there is no match. */ | |
| 5363 if (status == RA_NOMATCH) | |
| 5364 restore_se(&rp->rs_un.sesave, ®_endzpos[rp->rs_no], | |
| 5365 ®_endzp[rp->rs_no]); | |
| 270 | 5366 regstack_pop(&scan); |
| 180 | 5367 break; |
| 5368 #endif | |
| 5369 | |
| 5370 case RS_BRANCH: | |
| 5371 if (status == RA_MATCH) | |
| 5372 /* this branch matched, use it */ | |
| 270 | 5373 regstack_pop(&scan); |
| 180 | 5374 else |
| 5375 { | |
| 5376 if (status != RA_BREAK) | |
| 5377 { | |
| 5378 /* After a non-matching branch: try next one. */ | |
| 233 | 5379 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5380 scan = rp->rs_scan; |
| 5381 } | |
| 5382 if (scan == NULL || OP(scan) != BRANCH) | |
| 5383 { | |
| 5384 /* no more branches, didn't find a match */ | |
| 5385 status = RA_NOMATCH; | |
| 270 | 5386 regstack_pop(&scan); |
| 180 | 5387 } |
| 5388 else | |
| 5389 { | |
| 5390 /* Prepare to try a branch. */ | |
| 5391 rp->rs_scan = regnext(scan); | |
| 233 | 5392 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5393 scan = OPERAND(scan); |
| 5394 } | |
| 5395 } | |
| 5396 break; | |
| 5397 | |
| 5398 case RS_BRCPLX_MORE: | |
| 5399 /* Pop the state. Restore pointers when there is no match. */ | |
| 5400 if (status == RA_NOMATCH) | |
| 5401 { | |
| 233 | 5402 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5403 --brace_count[rp->rs_no]; /* decrement match count */ |
| 5404 } | |
| 270 | 5405 regstack_pop(&scan); |
| 180 | 5406 break; |
| 5407 | |
| 5408 case RS_BRCPLX_LONG: | |
| 5409 /* Pop the state. Restore pointers when there is no match. */ | |
| 5410 if (status == RA_NOMATCH) | |
| 5411 { | |
| 5412 /* There was no match, but we did find enough matches. */ | |
| 233 | 5413 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5414 --brace_count[rp->rs_no]; |
| 5415 /* continue with the items after "\{}" */ | |
| 5416 status = RA_CONT; | |
| 5417 } | |
| 270 | 5418 regstack_pop(&scan); |
| 180 | 5419 if (status == RA_CONT) |
| 5420 scan = regnext(scan); | |
| 5421 break; | |
| 5422 | |
| 5423 case RS_BRCPLX_SHORT: | |
| 5424 /* Pop the state. Restore pointers when there is no match. */ | |
| 5425 if (status == RA_NOMATCH) | |
| 5426 /* There was no match, try to match one more item. */ | |
| 233 | 5427 reg_restore(&rp->rs_un.regsave, &backpos); |
| 270 | 5428 regstack_pop(&scan); |
| 180 | 5429 if (status == RA_NOMATCH) |
| 5430 { | |
| 5431 scan = OPERAND(scan); | |
| 5432 status = RA_CONT; | |
| 5433 } | |
| 5434 break; | |
| 5435 | |
| 5436 case RS_NOMATCH: | |
| 5437 /* Pop the state. If the operand matches for NOMATCH or | |
| 5438 * doesn't match for MATCH/SUBPAT, we fail. Otherwise backup, | |
| 5439 * except for SUBPAT, and continue with the next item. */ | |
| 5440 if (status == (rp->rs_no == NOMATCH ? RA_MATCH : RA_NOMATCH)) | |
| 5441 status = RA_NOMATCH; | |
| 5442 else | |
| 5443 { | |
| 5444 status = RA_CONT; | |
| 233 | 5445 if (rp->rs_no != SUBPAT) /* zero-width */ |
| 5446 reg_restore(&rp->rs_un.regsave, &backpos); | |
| 180 | 5447 } |
| 270 | 5448 regstack_pop(&scan); |
| 180 | 5449 if (status == RA_CONT) |
| 5450 scan = regnext(scan); | |
| 5451 break; | |
| 5452 | |
| 5453 case RS_BEHIND1: | |
| 5454 if (status == RA_NOMATCH) | |
| 5455 { | |
| 270 | 5456 regstack_pop(&scan); |
| 180 | 5457 regstack.ga_len -= sizeof(regbehind_T); |
| 5458 } | |
| 5459 else | |
| 5460 { | |
| 5461 /* The stuff after BEHIND/NOBEHIND matches. Now try if | |
| 5462 * the behind part does (not) match before the current | |
| 5463 * position in the input. This must be done at every | |
| 5464 * position in the input and checking if the match ends at | |
| 5465 * the current position. */ | |
| 5466 | |
| 5467 /* save the position after the found match for next */ | |
| 233 | 5468 reg_save(&(((regbehind_T *)rp) - 1)->save_after, &backpos); |
| 180 | 5469 |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5470 /* Start looking for a match with operand at the current |
| 1209 | 5471 * position. Go back one character until we find the |
| 180 | 5472 * result, hitting the start of the line or the previous |
| 5473 * line (for multi-line matching). | |
| 5474 * Set behind_pos to where the match should end, BHPOS | |
| 5475 * will match it. Save the current value. */ | |
| 5476 (((regbehind_T *)rp) - 1)->save_behind = behind_pos; | |
| 5477 behind_pos = rp->rs_un.regsave; | |
| 5478 | |
| 5479 rp->rs_state = RS_BEHIND2; | |
| 5480 | |
| 233 | 5481 reg_restore(&rp->rs_un.regsave, &backpos); |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5482 scan = OPERAND(rp->rs_scan) + 4; |
| 180 | 5483 } |
| 5484 break; | |
| 5485 | |
| 5486 case RS_BEHIND2: | |
| 5487 /* | |
| 5488 * Looping for BEHIND / NOBEHIND match. | |
| 5489 */ | |
| 5490 if (status == RA_MATCH && reg_save_equal(&behind_pos)) | |
| 5491 { | |
| 5492 /* found a match that ends where "next" started */ | |
| 5493 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5494 if (rp->rs_no == BEHIND) | |
| 233 | 5495 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5496 &backpos); | |
| 180 | 5497 else |
| 1579 | 5498 { |
| 5499 /* But we didn't want a match. Need to restore the | |
| 5500 * subexpr, because what follows matched, so they have | |
| 5501 * been set. */ | |
| 180 | 5502 status = RA_NOMATCH; |
| 1579 | 5503 restore_subexpr(((regbehind_T *)rp) - 1); |
| 5504 } | |
| 270 | 5505 regstack_pop(&scan); |
| 180 | 5506 regstack.ga_len -= sizeof(regbehind_T); |
| 5507 } | |
| 5508 else | |
| 5509 { | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5510 long limit; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5511 |
| 1579 | 5512 /* No match or a match that doesn't end where we want it: Go |
| 5513 * back one character. May go to previous line once. */ | |
| 180 | 5514 no = OK; |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5515 limit = OPERAND_MIN(rp->rs_scan); |
| 180 | 5516 if (REG_MULTI) |
| 5517 { | |
|
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5518 if (limit > 0 |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5519 && ((rp->rs_un.regsave.rs_u.pos.lnum |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5520 < behind_pos.rs_u.pos.lnum |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5521 ? (colnr_T)STRLEN(regline) |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5522 : behind_pos.rs_u.pos.col) |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5523 - rp->rs_un.regsave.rs_u.pos.col >= limit)) |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5524 no = FAIL; |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5525 else if (rp->rs_un.regsave.rs_u.pos.col == 0) |
| 180 | 5526 { |
| 5527 if (rp->rs_un.regsave.rs_u.pos.lnum | |
| 5528 < behind_pos.rs_u.pos.lnum | |
| 5529 || reg_getline( | |
| 5530 --rp->rs_un.regsave.rs_u.pos.lnum) | |
| 5531 == NULL) | |
| 5532 no = FAIL; | |
| 5533 else | |
| 5534 { | |
| 233 | 5535 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5536 rp->rs_un.regsave.rs_u.pos.col = |
| 5537 (colnr_T)STRLEN(regline); | |
| 5538 } | |
| 5539 } | |
| 5540 else | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5541 { |
| 4176 | 5542 #ifdef FEAT_MBYTE |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5543 if (has_mbyte) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5544 rp->rs_un.regsave.rs_u.pos.col -= |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5545 (*mb_head_off)(regline, regline |
| 4176 | 5546 + rp->rs_un.regsave.rs_u.pos.col - 1) + 1; |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5547 else |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5548 #endif |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5549 --rp->rs_un.regsave.rs_u.pos.col; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5550 } |
| 180 | 5551 } |
| 5552 else | |
| 5553 { | |
| 5554 if (rp->rs_un.regsave.rs_u.ptr == regline) | |
| 5555 no = FAIL; | |
| 5556 else | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5557 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5558 mb_ptr_back(regline, rp->rs_un.regsave.rs_u.ptr); |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5559 if (limit > 0 && (long)(behind_pos.rs_u.ptr |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5560 - rp->rs_un.regsave.rs_u.ptr) > limit) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5561 no = FAIL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5562 } |
| 180 | 5563 } |
| 5564 if (no == OK) | |
| 5565 { | |
| 5566 /* Advanced, prepare for finding match again. */ | |
| 233 | 5567 reg_restore(&rp->rs_un.regsave, &backpos); |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5568 scan = OPERAND(rp->rs_scan) + 4; |
| 1579 | 5569 if (status == RA_MATCH) |
| 5570 { | |
| 5571 /* We did match, so subexpr may have been changed, | |
| 5572 * need to restore them for the next try. */ | |
| 5573 status = RA_NOMATCH; | |
| 5574 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5575 } | |
| 180 | 5576 } |
| 5577 else | |
| 5578 { | |
| 5579 /* Can't advance. For NOBEHIND that's a match. */ | |
| 5580 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5581 if (rp->rs_no == NOBEHIND) | |
| 5582 { | |
| 233 | 5583 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5584 &backpos); | |
| 180 | 5585 status = RA_MATCH; |
| 5586 } | |
| 5587 else | |
| 1579 | 5588 { |
| 5589 /* We do want a proper match. Need to restore the | |
| 5590 * subexpr if we had a match, because they may have | |
| 5591 * been set. */ | |
| 5592 if (status == RA_MATCH) | |
| 5593 { | |
| 5594 status = RA_NOMATCH; | |
| 5595 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5596 } | |
| 5597 } | |
| 270 | 5598 regstack_pop(&scan); |
| 180 | 5599 regstack.ga_len -= sizeof(regbehind_T); |
| 5600 } | |
| 5601 } | |
| 5602 break; | |
| 5603 | |
| 5604 case RS_STAR_LONG: | |
| 5605 case RS_STAR_SHORT: | |
| 5606 { | |
| 5607 regstar_T *rst = ((regstar_T *)rp) - 1; | |
| 5608 | |
| 5609 if (status == RA_MATCH) | |
| 5610 { | |
| 270 | 5611 regstack_pop(&scan); |
| 180 | 5612 regstack.ga_len -= sizeof(regstar_T); |
| 5613 break; | |
| 5614 } | |
| 5615 | |
| 5616 /* Tried once already, restore input pointers. */ | |
| 5617 if (status != RA_BREAK) | |
| 233 | 5618 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5619 |
| 5620 /* Repeat until we found a position where it could match. */ | |
| 5621 for (;;) | |
| 5622 { | |
| 5623 if (status != RA_BREAK) | |
| 5624 { | |
| 5625 /* Tried first position already, advance. */ | |
| 5626 if (rp->rs_state == RS_STAR_LONG) | |
| 5627 { | |
| 685 | 5628 /* Trying for longest match, but couldn't or |
| 5629 * didn't match -- back up one char. */ | |
| 180 | 5630 if (--rst->count < rst->minval) |
| 5631 break; | |
| 5632 if (reginput == regline) | |
| 5633 { | |
| 5634 /* backup to last char of previous line */ | |
| 5635 --reglnum; | |
| 5636 regline = reg_getline(reglnum); | |
| 5637 /* Just in case regrepeat() didn't count | |
| 5638 * right. */ | |
| 5639 if (regline == NULL) | |
| 5640 break; | |
| 5641 reginput = regline + STRLEN(regline); | |
| 5642 fast_breakcheck(); | |
| 5643 } | |
| 5644 else | |
| 5645 mb_ptr_back(regline, reginput); | |
| 5646 } | |
| 5647 else | |
| 5648 { | |
| 5649 /* Range is backwards, use shortest match first. | |
| 5650 * Careful: maxval and minval are exchanged! | |
| 5651 * Couldn't or didn't match: try advancing one | |
| 5652 * char. */ | |
| 5653 if (rst->count == rst->minval | |
| 5654 || regrepeat(OPERAND(rp->rs_scan), 1L) == 0) | |
| 5655 break; | |
| 5656 ++rst->count; | |
| 5657 } | |
| 5658 if (got_int) | |
| 5659 break; | |
| 5660 } | |
| 5661 else | |
| 5662 status = RA_NOMATCH; | |
| 5663 | |
| 5664 /* If it could match, try it. */ | |
| 5665 if (rst->nextb == NUL || *reginput == rst->nextb | |
| 5666 || *reginput == rst->nextb_ic) | |
| 5667 { | |
| 233 | 5668 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5669 scan = regnext(rp->rs_scan); |
| 5670 status = RA_CONT; | |
| 5671 break; | |
| 5672 } | |
| 5673 } | |
| 5674 if (status != RA_CONT) | |
| 5675 { | |
| 5676 /* Failed. */ | |
| 270 | 5677 regstack_pop(&scan); |
| 180 | 5678 regstack.ga_len -= sizeof(regstar_T); |
| 5679 status = RA_NOMATCH; | |
| 5680 } | |
| 5681 } | |
| 5682 break; | |
| 5683 } | |
| 5684 | |
| 685 | 5685 /* If we want to continue the inner loop or didn't pop a state |
| 5686 * continue matching loop */ | |
| 180 | 5687 if (status == RA_CONT || rp == (regitem_T *) |
| 5688 ((char *)regstack.ga_data + regstack.ga_len) - 1) | |
| 5689 break; | |
| 5690 } | |
| 5691 | |
| 189 | 5692 /* May need to continue with the inner loop, starting at "scan". */ |
| 180 | 5693 if (status == RA_CONT) |
| 5694 continue; | |
| 5695 | |
| 5696 /* | |
| 5697 * If the regstack is empty or something failed we are done. | |
| 5698 */ | |
| 5699 if (regstack.ga_len == 0 || status == RA_FAIL) | |
| 5700 { | |
| 5701 if (scan == NULL) | |
| 5702 { | |
| 5703 /* | |
| 5704 * We get here only if there's trouble -- normally "case END" is | |
| 5705 * the terminating point. | |
| 5706 */ | |
| 5707 EMSG(_(e_re_corr)); | |
| 7 | 5708 #ifdef DEBUG |
| 180 | 5709 printf("Premature EOL\n"); |
| 7 | 5710 #endif |
| 180 | 5711 } |
| 189 | 5712 if (status == RA_FAIL) |
| 5713 got_int = TRUE; | |
| 180 | 5714 return (status == RA_MATCH); |
| 5715 } | |
| 5716 | |
| 5717 } /* End of loop until the regstack is empty. */ | |
| 5718 | |
| 5719 /* NOTREACHED */ | |
| 5720 } | |
| 5721 | |
| 5722 /* | |
| 5723 * Push an item onto the regstack. | |
| 5724 * Returns pointer to new item. Returns NULL when out of memory. | |
| 5725 */ | |
| 5726 static regitem_T * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
5727 regstack_push(regstate_T state, char_u *scan) |
| 180 | 5728 { |
| 5729 regitem_T *rp; | |
| 5730 | |
| 270 | 5731 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5732 { |
| 5733 EMSG(_(e_maxmempat)); | |
| 5734 return NULL; | |
| 5735 } | |
| 270 | 5736 if (ga_grow(®stack, sizeof(regitem_T)) == FAIL) |
| 180 | 5737 return NULL; |
| 5738 | |
| 270 | 5739 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len); |
| 180 | 5740 rp->rs_state = state; |
| 5741 rp->rs_scan = scan; | |
| 5742 | |
| 270 | 5743 regstack.ga_len += sizeof(regitem_T); |
| 180 | 5744 return rp; |
| 5745 } | |
| 5746 | |
| 5747 /* | |
| 5748 * Pop an item from the regstack. | |
| 5749 */ | |
| 5750 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
5751 regstack_pop(char_u **scan) |
| 180 | 5752 { |
| 5753 regitem_T *rp; | |
| 5754 | |
| 270 | 5755 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; |
| 180 | 5756 *scan = rp->rs_scan; |
| 5757 | |
| 270 | 5758 regstack.ga_len -= sizeof(regitem_T); |
| 7 | 5759 } |
| 5760 | |
| 5761 /* | |
| 5762 * regrepeat - repeatedly match something simple, return how many. | |
| 5763 * Advances reginput (and reglnum) to just after the matched chars. | |
| 5764 */ | |
| 5765 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
5766 regrepeat( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
5767 char_u *p, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
5768 long maxcount) /* maximum number of matches allowed */ |
| 7 | 5769 { |
| 5770 long count = 0; | |
| 5771 char_u *scan; | |
| 5772 char_u *opnd; | |
| 5773 int mask; | |
| 5774 int testval = 0; | |
| 5775 | |
| 5776 scan = reginput; /* Make local copy of reginput for speed. */ | |
| 5777 opnd = OPERAND(p); | |
| 5778 switch (OP(p)) | |
| 5779 { | |
| 5780 case ANY: | |
| 5781 case ANY + ADD_NL: | |
| 5782 while (count < maxcount) | |
| 5783 { | |
| 5784 /* Matching anything means we continue until end-of-line (or | |
| 5785 * end-of-file for ANY + ADD_NL), only limited by maxcount. */ | |
| 5786 while (*scan != NUL && count < maxcount) | |
| 5787 { | |
| 5788 ++count; | |
| 39 | 5789 mb_ptr_adv(scan); |
| 7 | 5790 } |
| 1018 | 5791 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5792 || reg_line_lbr || count == maxcount) | |
| 7 | 5793 break; |
| 5794 ++count; /* count the line-break */ | |
| 5795 reg_nextline(); | |
| 5796 scan = reginput; | |
| 5797 if (got_int) | |
| 5798 break; | |
| 5799 } | |
| 5800 break; | |
| 5801 | |
| 5802 case IDENT: | |
| 5803 case IDENT + ADD_NL: | |
| 5804 testval = TRUE; | |
| 5805 /*FALLTHROUGH*/ | |
| 5806 case SIDENT: | |
| 5807 case SIDENT + ADD_NL: | |
| 5808 while (count < maxcount) | |
| 5809 { | |
| 4466 | 5810 if (vim_isIDc(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5811 { |
| 39 | 5812 mb_ptr_adv(scan); |
| 7 | 5813 } |
| 5814 else if (*scan == NUL) | |
| 5815 { | |
| 1018 | 5816 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5817 || reg_line_lbr) | |
| 7 | 5818 break; |
| 5819 reg_nextline(); | |
| 5820 scan = reginput; | |
| 5821 if (got_int) | |
| 5822 break; | |
| 5823 } | |
| 5824 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5825 ++scan; | |
| 5826 else | |
| 5827 break; | |
| 5828 ++count; | |
| 5829 } | |
| 5830 break; | |
| 5831 | |
| 5832 case KWORD: | |
| 5833 case KWORD + ADD_NL: | |
| 5834 testval = TRUE; | |
| 5835 /*FALLTHROUGH*/ | |
| 5836 case SKWORD: | |
| 5837 case SKWORD + ADD_NL: | |
| 5838 while (count < maxcount) | |
| 5839 { | |
| 4069 | 5840 if (vim_iswordp_buf(scan, reg_buf) |
| 5841 && (testval || !VIM_ISDIGIT(*scan))) | |
| 7 | 5842 { |
| 39 | 5843 mb_ptr_adv(scan); |
| 7 | 5844 } |
| 5845 else if (*scan == NUL) | |
| 5846 { | |
| 1018 | 5847 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5848 || reg_line_lbr) | |
| 7 | 5849 break; |
| 5850 reg_nextline(); | |
| 5851 scan = reginput; | |
| 5852 if (got_int) | |
| 5853 break; | |
| 5854 } | |
| 5855 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5856 ++scan; | |
| 5857 else | |
| 5858 break; | |
| 5859 ++count; | |
| 5860 } | |
| 5861 break; | |
| 5862 | |
| 5863 case FNAME: | |
| 5864 case FNAME + ADD_NL: | |
| 5865 testval = TRUE; | |
| 5866 /*FALLTHROUGH*/ | |
| 5867 case SFNAME: | |
| 5868 case SFNAME + ADD_NL: | |
| 5869 while (count < maxcount) | |
| 5870 { | |
| 4466 | 5871 if (vim_isfilec(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5872 { |
| 39 | 5873 mb_ptr_adv(scan); |
| 7 | 5874 } |
| 5875 else if (*scan == NUL) | |
| 5876 { | |
| 1018 | 5877 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5878 || reg_line_lbr) | |
| 7 | 5879 break; |
| 5880 reg_nextline(); | |
| 5881 scan = reginput; | |
| 5882 if (got_int) | |
| 5883 break; | |
| 5884 } | |
| 5885 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5886 ++scan; | |
| 5887 else | |
| 5888 break; | |
| 5889 ++count; | |
| 5890 } | |
| 5891 break; | |
| 5892 | |
| 5893 case PRINT: | |
| 5894 case PRINT + ADD_NL: | |
| 5895 testval = TRUE; | |
| 5896 /*FALLTHROUGH*/ | |
| 5897 case SPRINT: | |
| 5898 case SPRINT + ADD_NL: | |
| 5899 while (count < maxcount) | |
| 5900 { | |
| 5901 if (*scan == NUL) | |
| 5902 { | |
| 1018 | 5903 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5904 || reg_line_lbr) | |
| 7 | 5905 break; |
| 5906 reg_nextline(); | |
| 5907 scan = reginput; | |
| 5908 if (got_int) | |
| 5909 break; | |
| 5910 } | |
|
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
5911 else if (vim_isprintc(PTR2CHAR(scan)) == 1 |
|
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
5912 && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5913 { |
| 39 | 5914 mb_ptr_adv(scan); |
| 7 | 5915 } |
| 5916 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5917 ++scan; | |
| 5918 else | |
| 5919 break; | |
| 5920 ++count; | |
| 5921 } | |
| 5922 break; | |
| 5923 | |
| 5924 case WHITE: | |
| 5925 case WHITE + ADD_NL: | |
| 5926 testval = mask = RI_WHITE; | |
| 5927 do_class: | |
| 5928 while (count < maxcount) | |
| 5929 { | |
| 5930 #ifdef FEAT_MBYTE | |
| 5931 int l; | |
| 5932 #endif | |
| 5933 if (*scan == NUL) | |
| 5934 { | |
| 1018 | 5935 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5936 || reg_line_lbr) | |
| 7 | 5937 break; |
| 5938 reg_nextline(); | |
| 5939 scan = reginput; | |
| 5940 if (got_int) | |
| 5941 break; | |
| 5942 } | |
| 5943 #ifdef FEAT_MBYTE | |
| 474 | 5944 else if (has_mbyte && (l = (*mb_ptr2len)(scan)) > 1) |
| 7 | 5945 { |
| 5946 if (testval != 0) | |
| 5947 break; | |
| 5948 scan += l; | |
| 5949 } | |
| 5950 #endif | |
| 5951 else if ((class_tab[*scan] & mask) == testval) | |
| 5952 ++scan; | |
| 5953 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5954 ++scan; | |
| 5955 else | |
| 5956 break; | |
| 5957 ++count; | |
| 5958 } | |
| 5959 break; | |
| 5960 | |
| 5961 case NWHITE: | |
| 5962 case NWHITE + ADD_NL: | |
| 5963 mask = RI_WHITE; | |
| 5964 goto do_class; | |
| 5965 case DIGIT: | |
| 5966 case DIGIT + ADD_NL: | |
| 5967 testval = mask = RI_DIGIT; | |
| 5968 goto do_class; | |
| 5969 case NDIGIT: | |
| 5970 case NDIGIT + ADD_NL: | |
| 5971 mask = RI_DIGIT; | |
| 5972 goto do_class; | |
| 5973 case HEX: | |
| 5974 case HEX + ADD_NL: | |
| 5975 testval = mask = RI_HEX; | |
| 5976 goto do_class; | |
| 5977 case NHEX: | |
| 5978 case NHEX + ADD_NL: | |
| 5979 mask = RI_HEX; | |
| 5980 goto do_class; | |
| 5981 case OCTAL: | |
| 5982 case OCTAL + ADD_NL: | |
| 5983 testval = mask = RI_OCTAL; | |
| 5984 goto do_class; | |
| 5985 case NOCTAL: | |
| 5986 case NOCTAL + ADD_NL: | |
| 5987 mask = RI_OCTAL; | |
| 5988 goto do_class; | |
| 5989 case WORD: | |
| 5990 case WORD + ADD_NL: | |
| 5991 testval = mask = RI_WORD; | |
| 5992 goto do_class; | |
| 5993 case NWORD: | |
| 5994 case NWORD + ADD_NL: | |
| 5995 mask = RI_WORD; | |
| 5996 goto do_class; | |
| 5997 case HEAD: | |
| 5998 case HEAD + ADD_NL: | |
| 5999 testval = mask = RI_HEAD; | |
| 6000 goto do_class; | |
| 6001 case NHEAD: | |
| 6002 case NHEAD + ADD_NL: | |
| 6003 mask = RI_HEAD; | |
| 6004 goto do_class; | |
| 6005 case ALPHA: | |
| 6006 case ALPHA + ADD_NL: | |
| 6007 testval = mask = RI_ALPHA; | |
| 6008 goto do_class; | |
| 6009 case NALPHA: | |
| 6010 case NALPHA + ADD_NL: | |
| 6011 mask = RI_ALPHA; | |
| 6012 goto do_class; | |
| 6013 case LOWER: | |
| 6014 case LOWER + ADD_NL: | |
| 6015 testval = mask = RI_LOWER; | |
| 6016 goto do_class; | |
| 6017 case NLOWER: | |
| 6018 case NLOWER + ADD_NL: | |
| 6019 mask = RI_LOWER; | |
| 6020 goto do_class; | |
| 6021 case UPPER: | |
| 6022 case UPPER + ADD_NL: | |
| 6023 testval = mask = RI_UPPER; | |
| 6024 goto do_class; | |
| 6025 case NUPPER: | |
| 6026 case NUPPER + ADD_NL: | |
| 6027 mask = RI_UPPER; | |
| 6028 goto do_class; | |
| 6029 | |
| 6030 case EXACTLY: | |
| 6031 { | |
| 6032 int cu, cl; | |
| 6033 | |
| 6034 /* This doesn't do a multi-byte character, because a MULTIBYTECODE | |
| 1347 | 6035 * would have been used for it. It does handle single-byte |
| 6036 * characters, such as latin1. */ | |
| 7 | 6037 if (ireg_ic) |
| 6038 { | |
| 1347 | 6039 cu = MB_TOUPPER(*opnd); |
| 6040 cl = MB_TOLOWER(*opnd); | |
| 7 | 6041 while (count < maxcount && (*scan == cu || *scan == cl)) |
| 6042 { | |
| 6043 count++; | |
| 6044 scan++; | |
| 6045 } | |
| 6046 } | |
| 6047 else | |
| 6048 { | |
| 6049 cu = *opnd; | |
| 6050 while (count < maxcount && *scan == cu) | |
| 6051 { | |
| 6052 count++; | |
| 6053 scan++; | |
| 6054 } | |
| 6055 } | |
| 6056 break; | |
| 6057 } | |
| 6058 | |
| 6059 #ifdef FEAT_MBYTE | |
| 6060 case MULTIBYTECODE: | |
| 6061 { | |
| 6062 int i, len, cf = 0; | |
| 6063 | |
| 6064 /* Safety check (just in case 'encoding' was changed since | |
| 6065 * compiling the program). */ | |
| 474 | 6066 if ((len = (*mb_ptr2len)(opnd)) > 1) |
| 7 | 6067 { |
| 6068 if (ireg_ic && enc_utf8) | |
| 6069 cf = utf_fold(utf_ptr2char(opnd)); | |
| 6785 | 6070 while (count < maxcount && (*mb_ptr2len)(scan) >= len) |
| 7 | 6071 { |
| 6072 for (i = 0; i < len; ++i) | |
| 6073 if (opnd[i] != scan[i]) | |
| 6074 break; | |
| 6075 if (i < len && (!ireg_ic || !enc_utf8 | |
| 6076 || utf_fold(utf_ptr2char(scan)) != cf)) | |
| 6077 break; | |
| 6078 scan += len; | |
| 6079 ++count; | |
| 6080 } | |
| 6081 } | |
| 6082 } | |
| 6083 break; | |
| 6084 #endif | |
| 6085 | |
| 6086 case ANYOF: | |
| 6087 case ANYOF + ADD_NL: | |
| 6088 testval = TRUE; | |
| 6089 /*FALLTHROUGH*/ | |
| 6090 | |
| 6091 case ANYBUT: | |
| 6092 case ANYBUT + ADD_NL: | |
| 6093 while (count < maxcount) | |
| 6094 { | |
| 6095 #ifdef FEAT_MBYTE | |
| 6096 int len; | |
| 6097 #endif | |
| 6098 if (*scan == NUL) | |
| 6099 { | |
| 1018 | 6100 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 6101 || reg_line_lbr) | |
| 7 | 6102 break; |
| 6103 reg_nextline(); | |
| 6104 scan = reginput; | |
| 6105 if (got_int) | |
| 6106 break; | |
| 6107 } | |
| 6108 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 6109 ++scan; | |
| 6110 #ifdef FEAT_MBYTE | |
| 474 | 6111 else if (has_mbyte && (len = (*mb_ptr2len)(scan)) > 1) |
| 7 | 6112 { |
| 6113 if ((cstrchr(opnd, (*mb_ptr2char)(scan)) == NULL) == testval) | |
| 6114 break; | |
| 6115 scan += len; | |
| 6116 } | |
| 6117 #endif | |
| 6118 else | |
| 6119 { | |
| 6120 if ((cstrchr(opnd, *scan) == NULL) == testval) | |
| 6121 break; | |
| 6122 ++scan; | |
| 6123 } | |
| 6124 ++count; | |
| 6125 } | |
| 6126 break; | |
| 6127 | |
| 6128 case NEWL: | |
| 6129 while (count < maxcount | |
| 1018 | 6130 && ((*scan == NUL && reglnum <= reg_maxline && !reg_line_lbr |
| 6131 && REG_MULTI) || (*scan == '\n' && reg_line_lbr))) | |
| 7 | 6132 { |
| 6133 count++; | |
| 6134 if (reg_line_lbr) | |
| 6135 ADVANCE_REGINPUT(); | |
| 6136 else | |
| 6137 reg_nextline(); | |
| 6138 scan = reginput; | |
| 6139 if (got_int) | |
| 6140 break; | |
| 6141 } | |
| 6142 break; | |
| 6143 | |
| 6144 default: /* Oh dear. Called inappropriately. */ | |
| 6145 EMSG(_(e_re_corr)); | |
| 6146 #ifdef DEBUG | |
| 6147 printf("Called regrepeat with op code %d\n", OP(p)); | |
| 6148 #endif | |
| 6149 break; | |
| 6150 } | |
| 6151 | |
| 6152 reginput = scan; | |
| 6153 | |
| 6154 return (int)count; | |
| 6155 } | |
| 6156 | |
| 6157 /* | |
| 6158 * regnext - dig the "next" pointer out of a node | |
| 2010 | 6159 * Returns NULL when calculating size, when there is no next item and when |
| 6160 * there is an error. | |
| 7 | 6161 */ |
| 6162 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6163 regnext(char_u *p) |
| 7 | 6164 { |
| 6165 int offset; | |
| 6166 | |
| 2010 | 6167 if (p == JUST_CALC_SIZE || reg_toolong) |
| 7 | 6168 return NULL; |
| 6169 | |
| 6170 offset = NEXT(p); | |
| 6171 if (offset == 0) | |
| 6172 return NULL; | |
| 6173 | |
| 233 | 6174 if (OP(p) == BACK) |
| 7 | 6175 return p - offset; |
| 6176 else | |
| 6177 return p + offset; | |
| 6178 } | |
| 6179 | |
| 6180 /* | |
| 6181 * Check the regexp program for its magic number. | |
| 6182 * Return TRUE if it's wrong. | |
| 6183 */ | |
| 6184 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6185 prog_magic_wrong(void) |
| 7 | 6186 { |
| 4444 | 6187 regprog_T *prog; |
| 6188 | |
| 6189 prog = REG_MULTI ? reg_mmatch->regprog : reg_match->regprog; | |
| 6190 if (prog->engine == &nfa_regengine) | |
| 6191 /* For NFA matcher we don't check the magic */ | |
| 6192 return FALSE; | |
| 6193 | |
| 6194 if (UCHARAT(((bt_regprog_T *)prog)->program) != REGMAGIC) | |
| 7 | 6195 { |
| 6196 EMSG(_(e_re_corr)); | |
| 6197 return TRUE; | |
| 6198 } | |
| 6199 return FALSE; | |
| 6200 } | |
| 6201 | |
| 6202 /* | |
| 6203 * Cleanup the subexpressions, if this wasn't done yet. | |
| 6204 * This construction is used to clear the subexpressions only when they are | |
| 6205 * used (to increase speed). | |
| 6206 */ | |
| 6207 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6208 cleanup_subexpr(void) |
| 7 | 6209 { |
| 6210 if (need_clear_subexpr) | |
| 6211 { | |
| 6212 if (REG_MULTI) | |
| 6213 { | |
| 6214 /* Use 0xff to set lnum to -1 */ | |
| 6215 vim_memset(reg_startpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6216 vim_memset(reg_endpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6217 } | |
| 6218 else | |
| 6219 { | |
| 6220 vim_memset(reg_startp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6221 vim_memset(reg_endp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6222 } | |
| 6223 need_clear_subexpr = FALSE; | |
| 6224 } | |
| 6225 } | |
| 6226 | |
| 6227 #ifdef FEAT_SYN_HL | |
| 6228 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6229 cleanup_zsubexpr(void) |
| 7 | 6230 { |
| 6231 if (need_clear_zsubexpr) | |
| 6232 { | |
| 6233 if (REG_MULTI) | |
| 6234 { | |
| 6235 /* Use 0xff to set lnum to -1 */ | |
| 6236 vim_memset(reg_startzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6237 vim_memset(reg_endzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6238 } | |
| 6239 else | |
| 6240 { | |
| 6241 vim_memset(reg_startzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6242 vim_memset(reg_endzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6243 } | |
| 6244 need_clear_zsubexpr = FALSE; | |
| 6245 } | |
| 6246 } | |
| 6247 #endif | |
| 6248 | |
| 6249 /* | |
| 1579 | 6250 * Save the current subexpr to "bp", so that they can be restored |
| 6251 * later by restore_subexpr(). | |
| 6252 */ | |
| 6253 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6254 save_subexpr(regbehind_T *bp) |
| 1579 | 6255 { |
| 6256 int i; | |
| 6257 | |
| 1602 | 6258 /* When "need_clear_subexpr" is set we don't need to save the values, only |
| 6259 * remember that this flag needs to be set again when restoring. */ | |
| 6260 bp->save_need_clear_subexpr = need_clear_subexpr; | |
| 6261 if (!need_clear_subexpr) | |
| 1579 | 6262 { |
| 1602 | 6263 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6264 { |
| 1602 | 6265 if (REG_MULTI) |
| 6266 { | |
| 6267 bp->save_start[i].se_u.pos = reg_startpos[i]; | |
| 6268 bp->save_end[i].se_u.pos = reg_endpos[i]; | |
| 6269 } | |
| 6270 else | |
| 6271 { | |
| 6272 bp->save_start[i].se_u.ptr = reg_startp[i]; | |
| 6273 bp->save_end[i].se_u.ptr = reg_endp[i]; | |
| 6274 } | |
| 1579 | 6275 } |
| 6276 } | |
| 6277 } | |
| 6278 | |
| 6279 /* | |
| 6280 * Restore the subexpr from "bp". | |
| 6281 */ | |
| 6282 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6283 restore_subexpr(regbehind_T *bp) |
| 1579 | 6284 { |
| 6285 int i; | |
| 6286 | |
| 1602 | 6287 /* Only need to restore saved values when they are not to be cleared. */ |
| 6288 need_clear_subexpr = bp->save_need_clear_subexpr; | |
| 6289 if (!need_clear_subexpr) | |
| 1579 | 6290 { |
| 1602 | 6291 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6292 { |
| 1602 | 6293 if (REG_MULTI) |
| 6294 { | |
| 6295 reg_startpos[i] = bp->save_start[i].se_u.pos; | |
| 6296 reg_endpos[i] = bp->save_end[i].se_u.pos; | |
| 6297 } | |
| 6298 else | |
| 6299 { | |
| 6300 reg_startp[i] = bp->save_start[i].se_u.ptr; | |
| 6301 reg_endp[i] = bp->save_end[i].se_u.ptr; | |
| 6302 } | |
| 1579 | 6303 } |
| 6304 } | |
| 6305 } | |
| 6306 | |
| 6307 /* | |
| 7 | 6308 * Advance reglnum, regline and reginput to the next line. |
| 6309 */ | |
| 6310 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6311 reg_nextline(void) |
| 7 | 6312 { |
| 6313 regline = reg_getline(++reglnum); | |
| 6314 reginput = regline; | |
| 6315 fast_breakcheck(); | |
| 6316 } | |
| 6317 | |
| 6318 /* | |
| 6319 * Save the input line and position in a regsave_T. | |
| 6320 */ | |
| 6321 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6322 reg_save(regsave_T *save, garray_T *gap) |
| 7 | 6323 { |
| 6324 if (REG_MULTI) | |
| 6325 { | |
| 6326 save->rs_u.pos.col = (colnr_T)(reginput - regline); | |
| 6327 save->rs_u.pos.lnum = reglnum; | |
| 6328 } | |
| 6329 else | |
| 6330 save->rs_u.ptr = reginput; | |
| 233 | 6331 save->rs_len = gap->ga_len; |
| 7 | 6332 } |
| 6333 | |
| 6334 /* | |
| 6335 * Restore the input line and position from a regsave_T. | |
| 6336 */ | |
| 6337 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6338 reg_restore(regsave_T *save, garray_T *gap) |
| 7 | 6339 { |
| 6340 if (REG_MULTI) | |
| 6341 { | |
| 6342 if (reglnum != save->rs_u.pos.lnum) | |
| 6343 { | |
| 6344 /* only call reg_getline() when the line number changed to save | |
| 6345 * a bit of time */ | |
| 6346 reglnum = save->rs_u.pos.lnum; | |
| 6347 regline = reg_getline(reglnum); | |
| 6348 } | |
| 6349 reginput = regline + save->rs_u.pos.col; | |
| 6350 } | |
| 6351 else | |
| 6352 reginput = save->rs_u.ptr; | |
| 233 | 6353 gap->ga_len = save->rs_len; |
| 7 | 6354 } |
| 6355 | |
| 6356 /* | |
| 6357 * Return TRUE if current position is equal to saved position. | |
| 6358 */ | |
| 6359 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6360 reg_save_equal(regsave_T *save) |
| 7 | 6361 { |
| 6362 if (REG_MULTI) | |
| 6363 return reglnum == save->rs_u.pos.lnum | |
| 6364 && reginput == regline + save->rs_u.pos.col; | |
| 6365 return reginput == save->rs_u.ptr; | |
| 6366 } | |
| 6367 | |
| 6368 /* | |
| 6369 * Tentatively set the sub-expression start to the current position (after | |
| 6370 * calling regmatch() they will have changed). Need to save the existing | |
| 6371 * values for when there is no match. | |
| 6372 * Use se_save() to use pointer (save_se_multi()) or position (save_se_one()), | |
| 6373 * depending on REG_MULTI. | |
| 6374 */ | |
| 6375 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6376 save_se_multi(save_se_T *savep, lpos_T *posp) |
| 7 | 6377 { |
| 6378 savep->se_u.pos = *posp; | |
| 6379 posp->lnum = reglnum; | |
| 6380 posp->col = (colnr_T)(reginput - regline); | |
| 6381 } | |
| 6382 | |
| 6383 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6384 save_se_one(save_se_T *savep, char_u **pp) |
| 7 | 6385 { |
| 6386 savep->se_u.ptr = *pp; | |
| 6387 *pp = reginput; | |
| 6388 } | |
| 6389 | |
| 6390 /* | |
| 6391 * Compare a number with the operand of RE_LNUM, RE_COL or RE_VCOL. | |
| 6392 */ | |
| 6393 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6394 re_num_cmp(long_u val, char_u *scan) |
| 7 | 6395 { |
| 6396 long_u n = OPERAND_MIN(scan); | |
| 6397 | |
| 6398 if (OPERAND_CMP(scan) == '>') | |
| 6399 return val > n; | |
| 6400 if (OPERAND_CMP(scan) == '<') | |
| 6401 return val < n; | |
| 6402 return val == n; | |
| 6403 } | |
| 6404 | |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6405 /* |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6406 * Check whether a backreference matches. |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6407 * Returns RA_FAIL, RA_NOMATCH or RA_MATCH. |
| 5504 | 6408 * If "bytelen" is not NULL, it is set to the byte length of the match in the |
| 6409 * last line. | |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6410 */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6411 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6412 match_with_backref( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6413 linenr_T start_lnum, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6414 colnr_T start_col, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6415 linenr_T end_lnum, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6416 colnr_T end_col, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6417 int *bytelen) |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6418 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6419 linenr_T clnum = start_lnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6420 colnr_T ccol = start_col; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6421 int len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6422 char_u *p; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6423 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6424 if (bytelen != NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6425 *bytelen = 0; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6426 for (;;) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6427 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6428 /* Since getting one line may invalidate the other, need to make copy. |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6429 * Slow! */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6430 if (regline != reg_tofree) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6431 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6432 len = (int)STRLEN(regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6433 if (reg_tofree == NULL || len >= (int)reg_tofreelen) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6434 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6435 len += 50; /* get some extra */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6436 vim_free(reg_tofree); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6437 reg_tofree = alloc(len); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6438 if (reg_tofree == NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6439 return RA_FAIL; /* out of memory!*/ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6440 reg_tofreelen = len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6441 } |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6442 STRCPY(reg_tofree, regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6443 reginput = reg_tofree + (reginput - regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6444 regline = reg_tofree; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6445 } |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6446 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6447 /* Get the line to compare with. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6448 p = reg_getline(clnum); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6449 if (clnum == end_lnum) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6450 len = end_col - ccol; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6451 else |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6452 len = (int)STRLEN(p + ccol); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6453 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6454 if (cstrncmp(p + ccol, reginput, &len) != 0) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6455 return RA_NOMATCH; /* doesn't match */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6456 if (bytelen != NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6457 *bytelen += len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6458 if (clnum == end_lnum) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6459 break; /* match and at end! */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6460 if (reglnum >= reg_maxline) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6461 return RA_NOMATCH; /* text too short */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6462 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6463 /* Advance to next line. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6464 reg_nextline(); |
| 5504 | 6465 if (bytelen != NULL) |
| 6466 *bytelen = 0; | |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6467 ++clnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6468 ccol = 0; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6469 if (got_int) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6470 return RA_FAIL; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6471 } |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6472 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6473 /* found a match! Note that regline may now point to a copy of the line, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6474 * that should not matter. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6475 return RA_MATCH; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6476 } |
| 7 | 6477 |
| 4444 | 6478 #ifdef BT_REGEXP_DUMP |
| 7 | 6479 |
| 6480 /* | |
| 6481 * regdump - dump a regexp onto stdout in vaguely comprehensible form | |
| 6482 */ | |
| 6483 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6484 regdump(char_u *pattern, bt_regprog_T *r) |
| 7 | 6485 { |
| 6486 char_u *s; | |
| 6487 int op = EXACTLY; /* Arbitrary non-END op. */ | |
| 6488 char_u *next; | |
| 6489 char_u *end = NULL; | |
| 4444 | 6490 FILE *f; |
| 6491 | |
| 6492 #ifdef BT_REGEXP_LOG | |
| 6493 f = fopen("bt_regexp_log.log", "a"); | |
| 6494 #else | |
| 6495 f = stdout; | |
| 6496 #endif | |
| 6497 if (f == NULL) | |
| 6498 return; | |
| 6499 fprintf(f, "-------------------------------------\n\r\nregcomp(%s):\r\n", pattern); | |
| 7 | 6500 |
| 6501 s = r->program + 1; | |
| 6502 /* | |
| 6503 * Loop until we find the END that isn't before a referred next (an END | |
| 6504 * can also appear in a NOMATCH operand). | |
| 6505 */ | |
| 6506 while (op != END || s <= end) | |
| 6507 { | |
| 6508 op = OP(s); | |
| 4444 | 6509 fprintf(f, "%2d%s", (int)(s - r->program), regprop(s)); /* Where, what. */ |
| 7 | 6510 next = regnext(s); |
| 6511 if (next == NULL) /* Next ptr. */ | |
| 4444 | 6512 fprintf(f, "(0)"); |
| 7 | 6513 else |
| 4444 | 6514 fprintf(f, "(%d)", (int)((s - r->program) + (next - s))); |
| 7 | 6515 if (end < next) |
| 6516 end = next; | |
| 6517 if (op == BRACE_LIMITS) | |
| 6518 { | |
|
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6519 /* Two ints */ |
| 4444 | 6520 fprintf(f, " minval %ld, maxval %ld", OPERAND_MIN(s), OPERAND_MAX(s)); |
| 7 | 6521 s += 8; |
| 6522 } | |
|
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6523 else if (op == BEHIND || op == NOBEHIND) |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6524 { |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6525 /* one int */ |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6526 fprintf(f, " count %ld", OPERAND_MIN(s)); |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6527 s += 4; |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6528 } |
|
4770
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6529 else if (op == RE_LNUM || op == RE_COL || op == RE_VCOL) |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6530 { |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6531 /* one int plus comperator */ |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6532 fprintf(f, " count %ld", OPERAND_MIN(s)); |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6533 s += 5; |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6534 } |
| 7 | 6535 s += 3; |
| 6536 if (op == ANYOF || op == ANYOF + ADD_NL | |
| 6537 || op == ANYBUT || op == ANYBUT + ADD_NL | |
| 6538 || op == EXACTLY) | |
| 6539 { | |
| 6540 /* Literal string, where present. */ | |
| 4444 | 6541 fprintf(f, "\nxxxxxxxxx\n"); |
| 7 | 6542 while (*s != NUL) |
| 4444 | 6543 fprintf(f, "%c", *s++); |
| 6544 fprintf(f, "\nxxxxxxxxx\n"); | |
| 7 | 6545 s++; |
| 6546 } | |
| 4444 | 6547 fprintf(f, "\r\n"); |
| 7 | 6548 } |
| 6549 | |
| 6550 /* Header fields of interest. */ | |
| 6551 if (r->regstart != NUL) | |
| 4444 | 6552 fprintf(f, "start `%s' 0x%x; ", r->regstart < 256 |
| 7 | 6553 ? (char *)transchar(r->regstart) |
| 6554 : "multibyte", r->regstart); | |
| 6555 if (r->reganch) | |
| 4444 | 6556 fprintf(f, "anchored; "); |
| 7 | 6557 if (r->regmust != NULL) |
| 4444 | 6558 fprintf(f, "must have \"%s\"", r->regmust); |
| 6559 fprintf(f, "\r\n"); | |
| 6560 | |
| 6561 #ifdef BT_REGEXP_LOG | |
| 6562 fclose(f); | |
| 6563 #endif | |
| 7 | 6564 } |
| 4444 | 6565 #endif /* BT_REGEXP_DUMP */ |
| 6566 | |
| 6567 #ifdef DEBUG | |
| 7 | 6568 /* |
| 6569 * regprop - printable representation of opcode | |
| 6570 */ | |
| 6571 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6572 regprop(char_u *op) |
| 7 | 6573 { |
| 4444 | 6574 char *p; |
| 6575 static char buf[50]; | |
| 6576 | |
| 6577 STRCPY(buf, ":"); | |
| 6578 | |
| 6579 switch ((int) OP(op)) | |
| 7 | 6580 { |
| 6581 case BOL: | |
| 6582 p = "BOL"; | |
| 6583 break; | |
| 6584 case EOL: | |
| 6585 p = "EOL"; | |
| 6586 break; | |
| 6587 case RE_BOF: | |
| 6588 p = "BOF"; | |
| 6589 break; | |
| 6590 case RE_EOF: | |
| 6591 p = "EOF"; | |
| 6592 break; | |
| 6593 case CURSOR: | |
| 6594 p = "CURSOR"; | |
| 6595 break; | |
| 639 | 6596 case RE_VISUAL: |
| 6597 p = "RE_VISUAL"; | |
| 6598 break; | |
| 7 | 6599 case RE_LNUM: |
| 6600 p = "RE_LNUM"; | |
| 6601 break; | |
| 639 | 6602 case RE_MARK: |
| 6603 p = "RE_MARK"; | |
| 6604 break; | |
| 7 | 6605 case RE_COL: |
| 6606 p = "RE_COL"; | |
| 6607 break; | |
| 6608 case RE_VCOL: | |
| 6609 p = "RE_VCOL"; | |
| 6610 break; | |
| 6611 case BOW: | |
| 6612 p = "BOW"; | |
| 6613 break; | |
| 6614 case EOW: | |
| 6615 p = "EOW"; | |
| 6616 break; | |
| 6617 case ANY: | |
| 6618 p = "ANY"; | |
| 6619 break; | |
| 6620 case ANY + ADD_NL: | |
| 6621 p = "ANY+NL"; | |
| 6622 break; | |
| 6623 case ANYOF: | |
| 6624 p = "ANYOF"; | |
| 6625 break; | |
| 6626 case ANYOF + ADD_NL: | |
| 6627 p = "ANYOF+NL"; | |
| 6628 break; | |
| 6629 case ANYBUT: | |
| 6630 p = "ANYBUT"; | |
| 6631 break; | |
| 6632 case ANYBUT + ADD_NL: | |
| 6633 p = "ANYBUT+NL"; | |
| 6634 break; | |
| 6635 case IDENT: | |
| 6636 p = "IDENT"; | |
| 6637 break; | |
| 6638 case IDENT + ADD_NL: | |
| 6639 p = "IDENT+NL"; | |
| 6640 break; | |
| 6641 case SIDENT: | |
| 6642 p = "SIDENT"; | |
| 6643 break; | |
| 6644 case SIDENT + ADD_NL: | |
| 6645 p = "SIDENT+NL"; | |
| 6646 break; | |
| 6647 case KWORD: | |
| 6648 p = "KWORD"; | |
| 6649 break; | |
| 6650 case KWORD + ADD_NL: | |
| 6651 p = "KWORD+NL"; | |
| 6652 break; | |
| 6653 case SKWORD: | |
| 6654 p = "SKWORD"; | |
| 6655 break; | |
| 6656 case SKWORD + ADD_NL: | |
| 6657 p = "SKWORD+NL"; | |
| 6658 break; | |
| 6659 case FNAME: | |
| 6660 p = "FNAME"; | |
| 6661 break; | |
| 6662 case FNAME + ADD_NL: | |
| 6663 p = "FNAME+NL"; | |
| 6664 break; | |
| 6665 case SFNAME: | |
| 6666 p = "SFNAME"; | |
| 6667 break; | |
| 6668 case SFNAME + ADD_NL: | |
| 6669 p = "SFNAME+NL"; | |
| 6670 break; | |
| 6671 case PRINT: | |
| 6672 p = "PRINT"; | |
| 6673 break; | |
| 6674 case PRINT + ADD_NL: | |
| 6675 p = "PRINT+NL"; | |
| 6676 break; | |
| 6677 case SPRINT: | |
| 6678 p = "SPRINT"; | |
| 6679 break; | |
| 6680 case SPRINT + ADD_NL: | |
| 6681 p = "SPRINT+NL"; | |
| 6682 break; | |
| 6683 case WHITE: | |
| 6684 p = "WHITE"; | |
| 6685 break; | |
| 6686 case WHITE + ADD_NL: | |
| 6687 p = "WHITE+NL"; | |
| 6688 break; | |
| 6689 case NWHITE: | |
| 6690 p = "NWHITE"; | |
| 6691 break; | |
| 6692 case NWHITE + ADD_NL: | |
| 6693 p = "NWHITE+NL"; | |
| 6694 break; | |
| 6695 case DIGIT: | |
| 6696 p = "DIGIT"; | |
| 6697 break; | |
| 6698 case DIGIT + ADD_NL: | |
| 6699 p = "DIGIT+NL"; | |
| 6700 break; | |
| 6701 case NDIGIT: | |
| 6702 p = "NDIGIT"; | |
| 6703 break; | |
| 6704 case NDIGIT + ADD_NL: | |
| 6705 p = "NDIGIT+NL"; | |
| 6706 break; | |
| 6707 case HEX: | |
| 6708 p = "HEX"; | |
| 6709 break; | |
| 6710 case HEX + ADD_NL: | |
| 6711 p = "HEX+NL"; | |
| 6712 break; | |
| 6713 case NHEX: | |
| 6714 p = "NHEX"; | |
| 6715 break; | |
| 6716 case NHEX + ADD_NL: | |
| 6717 p = "NHEX+NL"; | |
| 6718 break; | |
| 6719 case OCTAL: | |
| 6720 p = "OCTAL"; | |
| 6721 break; | |
| 6722 case OCTAL + ADD_NL: | |
| 6723 p = "OCTAL+NL"; | |
| 6724 break; | |
| 6725 case NOCTAL: | |
| 6726 p = "NOCTAL"; | |
| 6727 break; | |
| 6728 case NOCTAL + ADD_NL: | |
| 6729 p = "NOCTAL+NL"; | |
| 6730 break; | |
| 6731 case WORD: | |
| 6732 p = "WORD"; | |
| 6733 break; | |
| 6734 case WORD + ADD_NL: | |
| 6735 p = "WORD+NL"; | |
| 6736 break; | |
| 6737 case NWORD: | |
| 6738 p = "NWORD"; | |
| 6739 break; | |
| 6740 case NWORD + ADD_NL: | |
| 6741 p = "NWORD+NL"; | |
| 6742 break; | |
| 6743 case HEAD: | |
| 6744 p = "HEAD"; | |
| 6745 break; | |
| 6746 case HEAD + ADD_NL: | |
| 6747 p = "HEAD+NL"; | |
| 6748 break; | |
| 6749 case NHEAD: | |
| 6750 p = "NHEAD"; | |
| 6751 break; | |
| 6752 case NHEAD + ADD_NL: | |
| 6753 p = "NHEAD+NL"; | |
| 6754 break; | |
| 6755 case ALPHA: | |
| 6756 p = "ALPHA"; | |
| 6757 break; | |
| 6758 case ALPHA + ADD_NL: | |
| 6759 p = "ALPHA+NL"; | |
| 6760 break; | |
| 6761 case NALPHA: | |
| 6762 p = "NALPHA"; | |
| 6763 break; | |
| 6764 case NALPHA + ADD_NL: | |
| 6765 p = "NALPHA+NL"; | |
| 6766 break; | |
| 6767 case LOWER: | |
| 6768 p = "LOWER"; | |
| 6769 break; | |
| 6770 case LOWER + ADD_NL: | |
| 6771 p = "LOWER+NL"; | |
| 6772 break; | |
| 6773 case NLOWER: | |
| 6774 p = "NLOWER"; | |
| 6775 break; | |
| 6776 case NLOWER + ADD_NL: | |
| 6777 p = "NLOWER+NL"; | |
| 6778 break; | |
| 6779 case UPPER: | |
| 6780 p = "UPPER"; | |
| 6781 break; | |
| 6782 case UPPER + ADD_NL: | |
| 6783 p = "UPPER+NL"; | |
| 6784 break; | |
| 6785 case NUPPER: | |
| 6786 p = "NUPPER"; | |
| 6787 break; | |
| 6788 case NUPPER + ADD_NL: | |
| 6789 p = "NUPPER+NL"; | |
| 6790 break; | |
| 6791 case BRANCH: | |
| 6792 p = "BRANCH"; | |
| 6793 break; | |
| 6794 case EXACTLY: | |
| 6795 p = "EXACTLY"; | |
| 6796 break; | |
| 6797 case NOTHING: | |
| 6798 p = "NOTHING"; | |
| 6799 break; | |
| 6800 case BACK: | |
| 6801 p = "BACK"; | |
| 6802 break; | |
| 6803 case END: | |
| 6804 p = "END"; | |
| 6805 break; | |
| 6806 case MOPEN + 0: | |
| 6807 p = "MATCH START"; | |
| 6808 break; | |
| 6809 case MOPEN + 1: | |
| 6810 case MOPEN + 2: | |
| 6811 case MOPEN + 3: | |
| 6812 case MOPEN + 4: | |
| 6813 case MOPEN + 5: | |
| 6814 case MOPEN + 6: | |
| 6815 case MOPEN + 7: | |
| 6816 case MOPEN + 8: | |
| 6817 case MOPEN + 9: | |
| 6818 sprintf(buf + STRLEN(buf), "MOPEN%d", OP(op) - MOPEN); | |
| 6819 p = NULL; | |
| 6820 break; | |
| 6821 case MCLOSE + 0: | |
| 6822 p = "MATCH END"; | |
| 6823 break; | |
| 6824 case MCLOSE + 1: | |
| 6825 case MCLOSE + 2: | |
| 6826 case MCLOSE + 3: | |
| 6827 case MCLOSE + 4: | |
| 6828 case MCLOSE + 5: | |
| 6829 case MCLOSE + 6: | |
| 6830 case MCLOSE + 7: | |
| 6831 case MCLOSE + 8: | |
| 6832 case MCLOSE + 9: | |
| 6833 sprintf(buf + STRLEN(buf), "MCLOSE%d", OP(op) - MCLOSE); | |
| 6834 p = NULL; | |
| 6835 break; | |
| 6836 case BACKREF + 1: | |
| 6837 case BACKREF + 2: | |
| 6838 case BACKREF + 3: | |
| 6839 case BACKREF + 4: | |
| 6840 case BACKREF + 5: | |
| 6841 case BACKREF + 6: | |
| 6842 case BACKREF + 7: | |
| 6843 case BACKREF + 8: | |
| 6844 case BACKREF + 9: | |
| 6845 sprintf(buf + STRLEN(buf), "BACKREF%d", OP(op) - BACKREF); | |
| 6846 p = NULL; | |
| 6847 break; | |
| 6848 case NOPEN: | |
| 6849 p = "NOPEN"; | |
| 6850 break; | |
| 6851 case NCLOSE: | |
| 6852 p = "NCLOSE"; | |
| 6853 break; | |
| 6854 #ifdef FEAT_SYN_HL | |
| 6855 case ZOPEN + 1: | |
| 6856 case ZOPEN + 2: | |
| 6857 case ZOPEN + 3: | |
| 6858 case ZOPEN + 4: | |
| 6859 case ZOPEN + 5: | |
| 6860 case ZOPEN + 6: | |
| 6861 case ZOPEN + 7: | |
| 6862 case ZOPEN + 8: | |
| 6863 case ZOPEN + 9: | |
| 6864 sprintf(buf + STRLEN(buf), "ZOPEN%d", OP(op) - ZOPEN); | |
| 6865 p = NULL; | |
| 6866 break; | |
| 6867 case ZCLOSE + 1: | |
| 6868 case ZCLOSE + 2: | |
| 6869 case ZCLOSE + 3: | |
| 6870 case ZCLOSE + 4: | |
| 6871 case ZCLOSE + 5: | |
| 6872 case ZCLOSE + 6: | |
| 6873 case ZCLOSE + 7: | |
| 6874 case ZCLOSE + 8: | |
| 6875 case ZCLOSE + 9: | |
| 6876 sprintf(buf + STRLEN(buf), "ZCLOSE%d", OP(op) - ZCLOSE); | |
| 6877 p = NULL; | |
| 6878 break; | |
| 6879 case ZREF + 1: | |
| 6880 case ZREF + 2: | |
| 6881 case ZREF + 3: | |
| 6882 case ZREF + 4: | |
| 6883 case ZREF + 5: | |
| 6884 case ZREF + 6: | |
| 6885 case ZREF + 7: | |
| 6886 case ZREF + 8: | |
| 6887 case ZREF + 9: | |
| 6888 sprintf(buf + STRLEN(buf), "ZREF%d", OP(op) - ZREF); | |
| 6889 p = NULL; | |
| 6890 break; | |
| 6891 #endif | |
| 6892 case STAR: | |
| 6893 p = "STAR"; | |
| 6894 break; | |
| 6895 case PLUS: | |
| 6896 p = "PLUS"; | |
| 6897 break; | |
| 6898 case NOMATCH: | |
| 6899 p = "NOMATCH"; | |
| 6900 break; | |
| 6901 case MATCH: | |
| 6902 p = "MATCH"; | |
| 6903 break; | |
| 6904 case BEHIND: | |
| 6905 p = "BEHIND"; | |
| 6906 break; | |
| 6907 case NOBEHIND: | |
| 6908 p = "NOBEHIND"; | |
| 6909 break; | |
| 6910 case SUBPAT: | |
| 6911 p = "SUBPAT"; | |
| 6912 break; | |
| 6913 case BRACE_LIMITS: | |
| 6914 p = "BRACE_LIMITS"; | |
| 6915 break; | |
| 6916 case BRACE_SIMPLE: | |
| 6917 p = "BRACE_SIMPLE"; | |
| 6918 break; | |
| 6919 case BRACE_COMPLEX + 0: | |
| 6920 case BRACE_COMPLEX + 1: | |
| 6921 case BRACE_COMPLEX + 2: | |
| 6922 case BRACE_COMPLEX + 3: | |
| 6923 case BRACE_COMPLEX + 4: | |
| 6924 case BRACE_COMPLEX + 5: | |
| 6925 case BRACE_COMPLEX + 6: | |
| 6926 case BRACE_COMPLEX + 7: | |
| 6927 case BRACE_COMPLEX + 8: | |
| 6928 case BRACE_COMPLEX + 9: | |
| 6929 sprintf(buf + STRLEN(buf), "BRACE_COMPLEX%d", OP(op) - BRACE_COMPLEX); | |
| 6930 p = NULL; | |
| 6931 break; | |
| 6932 #ifdef FEAT_MBYTE | |
| 6933 case MULTIBYTECODE: | |
| 6934 p = "MULTIBYTECODE"; | |
| 6935 break; | |
| 6936 #endif | |
| 6937 case NEWL: | |
| 6938 p = "NEWL"; | |
| 6939 break; | |
| 6940 default: | |
| 6941 sprintf(buf + STRLEN(buf), "corrupt %d", OP(op)); | |
| 6942 p = NULL; | |
| 6943 break; | |
| 6944 } | |
| 6945 if (p != NULL) | |
| 4444 | 6946 STRCAT(buf, p); |
| 6947 return (char_u *)buf; | |
| 7 | 6948 } |
| 4444 | 6949 #endif /* DEBUG */ |
| 7 | 6950 |
| 6203 | 6951 /* |
| 6952 * Used in a place where no * or \+ can follow. | |
| 6953 */ | |
| 6954 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6955 re_mult_next(char *what) |
| 6203 | 6956 { |
| 6957 if (re_multi_type(peekchr()) == MULTI_MULT) | |
| 6958 EMSG2_RET_FAIL(_("E888: (NFA regexp) cannot repeat %s"), what); | |
| 6959 return OK; | |
| 6960 } | |
| 6961 | |
| 7 | 6962 #ifdef FEAT_MBYTE |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
6963 static void mb_decompose(int c, int *c1, int *c2, int *c3); |
| 7 | 6964 |
| 6965 typedef struct | |
| 6966 { | |
| 6967 int a, b, c; | |
| 6968 } decomp_T; | |
| 6969 | |
| 6970 | |
| 6971 /* 0xfb20 - 0xfb4f */ | |
| 297 | 6972 static decomp_T decomp_table[0xfb4f-0xfb20+1] = |
| 7 | 6973 { |
| 6974 {0x5e2,0,0}, /* 0xfb20 alt ayin */ | |
| 6975 {0x5d0,0,0}, /* 0xfb21 alt alef */ | |
| 6976 {0x5d3,0,0}, /* 0xfb22 alt dalet */ | |
| 6977 {0x5d4,0,0}, /* 0xfb23 alt he */ | |
| 6978 {0x5db,0,0}, /* 0xfb24 alt kaf */ | |
| 6979 {0x5dc,0,0}, /* 0xfb25 alt lamed */ | |
| 6980 {0x5dd,0,0}, /* 0xfb26 alt mem-sofit */ | |
| 6981 {0x5e8,0,0}, /* 0xfb27 alt resh */ | |
| 6982 {0x5ea,0,0}, /* 0xfb28 alt tav */ | |
| 6983 {'+', 0, 0}, /* 0xfb29 alt plus */ | |
| 6984 {0x5e9, 0x5c1, 0}, /* 0xfb2a shin+shin-dot */ | |
| 6985 {0x5e9, 0x5c2, 0}, /* 0xfb2b shin+sin-dot */ | |
| 6986 {0x5e9, 0x5c1, 0x5bc}, /* 0xfb2c shin+shin-dot+dagesh */ | |
| 6987 {0x5e9, 0x5c2, 0x5bc}, /* 0xfb2d shin+sin-dot+dagesh */ | |
| 6988 {0x5d0, 0x5b7, 0}, /* 0xfb2e alef+patah */ | |
| 6989 {0x5d0, 0x5b8, 0}, /* 0xfb2f alef+qamats */ | |
| 6990 {0x5d0, 0x5b4, 0}, /* 0xfb30 alef+hiriq */ | |
| 6991 {0x5d1, 0x5bc, 0}, /* 0xfb31 bet+dagesh */ | |
| 6992 {0x5d2, 0x5bc, 0}, /* 0xfb32 gimel+dagesh */ | |
| 6993 {0x5d3, 0x5bc, 0}, /* 0xfb33 dalet+dagesh */ | |
| 6994 {0x5d4, 0x5bc, 0}, /* 0xfb34 he+dagesh */ | |
| 6995 {0x5d5, 0x5bc, 0}, /* 0xfb35 vav+dagesh */ | |
| 6996 {0x5d6, 0x5bc, 0}, /* 0xfb36 zayin+dagesh */ | |
| 6997 {0xfb37, 0, 0}, /* 0xfb37 -- UNUSED */ | |
| 6998 {0x5d8, 0x5bc, 0}, /* 0xfb38 tet+dagesh */ | |
| 6999 {0x5d9, 0x5bc, 0}, /* 0xfb39 yud+dagesh */ | |
| 7000 {0x5da, 0x5bc, 0}, /* 0xfb3a kaf sofit+dagesh */ | |
| 7001 {0x5db, 0x5bc, 0}, /* 0xfb3b kaf+dagesh */ | |
| 7002 {0x5dc, 0x5bc, 0}, /* 0xfb3c lamed+dagesh */ | |
| 7003 {0xfb3d, 0, 0}, /* 0xfb3d -- UNUSED */ | |
| 7004 {0x5de, 0x5bc, 0}, /* 0xfb3e mem+dagesh */ | |
| 7005 {0xfb3f, 0, 0}, /* 0xfb3f -- UNUSED */ | |
| 7006 {0x5e0, 0x5bc, 0}, /* 0xfb40 nun+dagesh */ | |
| 7007 {0x5e1, 0x5bc, 0}, /* 0xfb41 samech+dagesh */ | |
| 7008 {0xfb42, 0, 0}, /* 0xfb42 -- UNUSED */ | |
| 7009 {0x5e3, 0x5bc, 0}, /* 0xfb43 pe sofit+dagesh */ | |
| 7010 {0x5e4, 0x5bc,0}, /* 0xfb44 pe+dagesh */ | |
| 7011 {0xfb45, 0, 0}, /* 0xfb45 -- UNUSED */ | |
| 7012 {0x5e6, 0x5bc, 0}, /* 0xfb46 tsadi+dagesh */ | |
| 7013 {0x5e7, 0x5bc, 0}, /* 0xfb47 qof+dagesh */ | |
| 7014 {0x5e8, 0x5bc, 0}, /* 0xfb48 resh+dagesh */ | |
| 7015 {0x5e9, 0x5bc, 0}, /* 0xfb49 shin+dagesh */ | |
| 7016 {0x5ea, 0x5bc, 0}, /* 0xfb4a tav+dagesh */ | |
| 7017 {0x5d5, 0x5b9, 0}, /* 0xfb4b vav+holam */ | |
| 7018 {0x5d1, 0x5bf, 0}, /* 0xfb4c bet+rafe */ | |
| 7019 {0x5db, 0x5bf, 0}, /* 0xfb4d kaf+rafe */ | |
| 7020 {0x5e4, 0x5bf, 0}, /* 0xfb4e pe+rafe */ | |
| 7021 {0x5d0, 0x5dc, 0} /* 0xfb4f alef-lamed */ | |
| 7022 }; | |
| 7023 | |
| 7024 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7025 mb_decompose(int c, int *c1, int *c2, int *c3) |
| 7 | 7026 { |
| 7027 decomp_T d; | |
| 7028 | |
|
4505
d037b9cbdaaa
updated for version 7.3.1000
Bram Moolenaar <bram@vim.org>
parents:
4466
diff
changeset
|
7029 if (c >= 0xfb20 && c <= 0xfb4f) |
| 7 | 7030 { |
| 7031 d = decomp_table[c - 0xfb20]; | |
| 7032 *c1 = d.a; | |
| 7033 *c2 = d.b; | |
| 7034 *c3 = d.c; | |
| 7035 } | |
| 7036 else | |
| 7037 { | |
| 7038 *c1 = c; | |
| 7039 *c2 = *c3 = 0; | |
| 7040 } | |
| 7041 } | |
| 7042 #endif | |
| 7043 | |
| 7044 /* | |
| 7045 * Compare two strings, ignore case if ireg_ic set. | |
| 7046 * Return 0 if strings match, non-zero otherwise. | |
| 7047 * Correct the length "*n" when composing characters are ignored. | |
| 7048 */ | |
| 7049 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7050 cstrncmp(char_u *s1, char_u *s2, int *n) |
| 7 | 7051 { |
| 7052 int result; | |
| 7053 | |
| 7054 if (!ireg_ic) | |
| 7055 result = STRNCMP(s1, s2, *n); | |
| 7056 else | |
| 7057 result = MB_STRNICMP(s1, s2, *n); | |
| 7058 | |
| 7059 #ifdef FEAT_MBYTE | |
| 7060 /* if it failed and it's utf8 and we want to combineignore: */ | |
| 7061 if (result != 0 && enc_utf8 && ireg_icombine) | |
| 7062 { | |
| 7063 char_u *str1, *str2; | |
| 7064 int c1, c2, c11, c12; | |
| 7065 int junk; | |
| 7066 | |
| 7067 /* we have to handle the strcmp ourselves, since it is necessary to | |
| 7068 * deal with the composing characters by ignoring them: */ | |
| 7069 str1 = s1; | |
| 7070 str2 = s2; | |
| 7071 c1 = c2 = 0; | |
| 507 | 7072 while ((int)(str1 - s1) < *n) |
| 7 | 7073 { |
| 7074 c1 = mb_ptr2char_adv(&str1); | |
| 7075 c2 = mb_ptr2char_adv(&str2); | |
| 7076 | |
| 7077 /* decompose the character if necessary, into 'base' characters | |
| 7078 * because I don't care about Arabic, I will hard-code the Hebrew | |
| 7079 * which I *do* care about! So sue me... */ | |
| 7080 if (c1 != c2 && (!ireg_ic || utf_fold(c1) != utf_fold(c2))) | |
| 7081 { | |
| 7082 /* decomposition necessary? */ | |
| 7083 mb_decompose(c1, &c11, &junk, &junk); | |
| 7084 mb_decompose(c2, &c12, &junk, &junk); | |
| 7085 c1 = c11; | |
| 7086 c2 = c12; | |
| 7087 if (c11 != c12 && (!ireg_ic || utf_fold(c11) != utf_fold(c12))) | |
| 7088 break; | |
| 7089 } | |
| 7090 } | |
| 7091 result = c2 - c1; | |
| 7092 if (result == 0) | |
| 7093 *n = (int)(str2 - s2); | |
| 7094 } | |
| 7095 #endif | |
| 7096 | |
| 7097 return result; | |
| 7098 } | |
| 7099 | |
| 7100 /* | |
| 7101 * cstrchr: This function is used a lot for simple searches, keep it fast! | |
| 7102 */ | |
| 7103 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7104 cstrchr(char_u *s, int c) |
| 7 | 7105 { |
| 7106 char_u *p; | |
| 7107 int cc; | |
| 7108 | |
| 7109 if (!ireg_ic | |
| 7110 #ifdef FEAT_MBYTE | |
| 7111 || (!enc_utf8 && mb_char2len(c) > 1) | |
| 7112 #endif | |
| 7113 ) | |
| 7114 return vim_strchr(s, c); | |
| 7115 | |
| 7116 /* tolower() and toupper() can be slow, comparing twice should be a lot | |
| 7117 * faster (esp. when using MS Visual C++!). | |
| 7118 * For UTF-8 need to use folded case. */ | |
| 7119 #ifdef FEAT_MBYTE | |
| 7120 if (enc_utf8 && c > 0x80) | |
| 7121 cc = utf_fold(c); | |
| 7122 else | |
| 7123 #endif | |
| 1347 | 7124 if (MB_ISUPPER(c)) |
| 7125 cc = MB_TOLOWER(c); | |
| 7126 else if (MB_ISLOWER(c)) | |
| 7127 cc = MB_TOUPPER(c); | |
| 7 | 7128 else |
| 7129 return vim_strchr(s, c); | |
| 7130 | |
| 7131 #ifdef FEAT_MBYTE | |
| 7132 if (has_mbyte) | |
| 7133 { | |
| 474 | 7134 for (p = s; *p != NUL; p += (*mb_ptr2len)(p)) |
| 7 | 7135 { |
| 7136 if (enc_utf8 && c > 0x80) | |
| 7137 { | |
| 7138 if (utf_fold(utf_ptr2char(p)) == cc) | |
| 7139 return p; | |
| 7140 } | |
| 7141 else if (*p == c || *p == cc) | |
| 7142 return p; | |
| 7143 } | |
| 7144 } | |
| 7145 else | |
| 7146 #endif | |
| 7147 /* Faster version for when there are no multi-byte characters. */ | |
| 7148 for (p = s; *p != NUL; ++p) | |
| 7149 if (*p == c || *p == cc) | |
| 7150 return p; | |
| 7151 | |
| 7152 return NULL; | |
| 7153 } | |
| 7154 | |
| 7155 /*************************************************************** | |
| 7156 * regsub stuff * | |
| 7157 ***************************************************************/ | |
| 7158 | |
| 7159 /* | |
| 7160 * We should define ftpr as a pointer to a function returning a pointer to | |
| 7161 * a function returning a pointer to a function ... | |
| 7162 * This is impossible, so we declare a pointer to a function returning a | |
| 7163 * pointer to a function returning void. This should work for all compilers. | |
| 7164 */ | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7165 typedef void (*(*fptr_T)(int *, int))(); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7166 |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7167 static fptr_T do_upper(int *, int); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7168 static fptr_T do_Upper(int *, int); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7169 static fptr_T do_lower(int *, int); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7170 static fptr_T do_Lower(int *, int); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7171 |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7172 static int vim_regsub_both(char_u *source, char_u *dest, int copy, int magic, int backslash); |
| 7 | 7173 |
| 772 | 7174 static fptr_T |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7175 do_upper(int *d, int c) |
| 7 | 7176 { |
| 772 | 7177 *d = MB_TOUPPER(c); |
| 7178 | |
| 7179 return (fptr_T)NULL; | |
| 7 | 7180 } |
| 7181 | |
| 772 | 7182 static fptr_T |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7183 do_Upper(int *d, int c) |
| 7 | 7184 { |
| 772 | 7185 *d = MB_TOUPPER(c); |
| 7186 | |
| 7187 return (fptr_T)do_Upper; | |
| 7188 } | |
| 7189 | |
| 7190 static fptr_T | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7191 do_lower(int *d, int c) |
| 772 | 7192 { |
| 7193 *d = MB_TOLOWER(c); | |
| 7194 | |
| 7195 return (fptr_T)NULL; | |
| 7196 } | |
| 7197 | |
| 7198 static fptr_T | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7199 do_Lower(int *d, int c) |
| 772 | 7200 { |
| 7201 *d = MB_TOLOWER(c); | |
| 7202 | |
| 7203 return (fptr_T)do_Lower; | |
| 7 | 7204 } |
| 7205 | |
| 7206 /* | |
| 7207 * regtilde(): Replace tildes in the pattern by the old pattern. | |
| 7208 * | |
| 7209 * Short explanation of the tilde: It stands for the previous replacement | |
| 7210 * pattern. If that previous pattern also contains a ~ we should go back a | |
| 7211 * step further... But we insert the previous pattern into the current one | |
| 7212 * and remember that. | |
| 772 | 7213 * This still does not handle the case where "magic" changes. So require the |
| 7214 * user to keep his hands off of "magic". | |
| 7 | 7215 * |
| 7216 * The tildes are parsed once before the first call to vim_regsub(). | |
| 7217 */ | |
| 7218 char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7219 regtilde(char_u *source, int magic) |
| 7 | 7220 { |
| 7221 char_u *newsub = source; | |
| 7222 char_u *tmpsub; | |
| 7223 char_u *p; | |
| 7224 int len; | |
| 7225 int prevlen; | |
| 7226 | |
| 7227 for (p = newsub; *p; ++p) | |
| 7228 { | |
| 7229 if ((*p == '~' && magic) || (*p == '\\' && *(p + 1) == '~' && !magic)) | |
| 7230 { | |
| 7231 if (reg_prev_sub != NULL) | |
| 7232 { | |
| 7233 /* length = len(newsub) - 1 + len(prev_sub) + 1 */ | |
| 7234 prevlen = (int)STRLEN(reg_prev_sub); | |
| 7235 tmpsub = alloc((unsigned)(STRLEN(newsub) + prevlen)); | |
| 7236 if (tmpsub != NULL) | |
| 7237 { | |
| 7238 /* copy prefix */ | |
| 7239 len = (int)(p - newsub); /* not including ~ */ | |
| 7240 mch_memmove(tmpsub, newsub, (size_t)len); | |
| 1209 | 7241 /* interpret tilde */ |
| 7 | 7242 mch_memmove(tmpsub + len, reg_prev_sub, (size_t)prevlen); |
| 7243 /* copy postfix */ | |
| 7244 if (!magic) | |
| 7245 ++p; /* back off \ */ | |
| 7246 STRCPY(tmpsub + len + prevlen, p + 1); | |
| 7247 | |
| 7248 if (newsub != source) /* already allocated newsub */ | |
| 7249 vim_free(newsub); | |
| 7250 newsub = tmpsub; | |
| 7251 p = newsub + len + prevlen; | |
| 7252 } | |
| 7253 } | |
| 7254 else if (magic) | |
| 1621 | 7255 STRMOVE(p, p + 1); /* remove '~' */ |
| 7 | 7256 else |
| 1621 | 7257 STRMOVE(p, p + 2); /* remove '\~' */ |
| 7 | 7258 --p; |
| 7259 } | |
| 7260 else | |
| 7261 { | |
| 7262 if (*p == '\\' && p[1]) /* skip escaped characters */ | |
| 7263 ++p; | |
| 7264 #ifdef FEAT_MBYTE | |
| 7265 if (has_mbyte) | |
| 474 | 7266 p += (*mb_ptr2len)(p) - 1; |
| 7 | 7267 #endif |
| 7268 } | |
| 7269 } | |
| 7270 | |
| 7271 vim_free(reg_prev_sub); | |
| 7272 if (newsub != source) /* newsub was allocated, just keep it */ | |
| 7273 reg_prev_sub = newsub; | |
| 7274 else /* no ~ found, need to save newsub */ | |
| 7275 reg_prev_sub = vim_strsave(newsub); | |
| 7276 return newsub; | |
| 7277 } | |
| 7278 | |
| 7279 #ifdef FEAT_EVAL | |
| 7280 static int can_f_submatch = FALSE; /* TRUE when submatch() can be used */ | |
| 7281 | |
| 7282 /* These pointers are used instead of reg_match and reg_mmatch for | |
| 7283 * reg_submatch(). Needed for when the substitution string is an expression | |
| 7284 * that contains a call to substitute() and submatch(). */ | |
| 7285 static regmatch_T *submatch_match; | |
| 7286 static regmmatch_T *submatch_mmatch; | |
| 2011 | 7287 static linenr_T submatch_firstlnum; |
| 7288 static linenr_T submatch_maxline; | |
| 2904 | 7289 static int submatch_line_lbr; |
| 7 | 7290 #endif |
| 7291 | |
| 7292 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO) | |
| 7293 /* | |
| 7294 * vim_regsub() - perform substitutions after a vim_regexec() or | |
| 7295 * vim_regexec_multi() match. | |
| 7296 * | |
| 7297 * If "copy" is TRUE really copy into "dest". | |
| 7298 * If "copy" is FALSE nothing is copied, this is just to find out the length | |
| 7299 * of the result. | |
| 7300 * | |
| 7301 * If "backslash" is TRUE, a backslash will be removed later, need to double | |
| 7302 * them to keep them, and insert a backslash before a CR to avoid it being | |
| 7303 * replaced with a line break later. | |
| 7304 * | |
| 7305 * Note: The matched text must not change between the call of | |
| 7306 * vim_regexec()/vim_regexec_multi() and vim_regsub()! It would make the back | |
| 7307 * references invalid! | |
| 7308 * | |
| 7309 * Returns the size of the replacement, including terminating NUL. | |
| 7310 */ | |
| 7311 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7312 vim_regsub( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7313 regmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7314 char_u *source, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7315 char_u *dest, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7316 int copy, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7317 int magic, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7318 int backslash) |
| 7 | 7319 { |
| 7320 reg_match = rmp; | |
| 7321 reg_mmatch = NULL; | |
| 7322 reg_maxline = 0; | |
| 4061 | 7323 reg_buf = curbuf; |
| 5836 | 7324 reg_line_lbr = TRUE; |
| 7 | 7325 return vim_regsub_both(source, dest, copy, magic, backslash); |
| 7326 } | |
| 7327 #endif | |
| 7328 | |
| 7329 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7330 vim_regsub_multi( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7331 regmmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7332 linenr_T lnum, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7333 char_u *source, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7334 char_u *dest, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7335 int copy, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7336 int magic, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7337 int backslash) |
| 7 | 7338 { |
| 7339 reg_match = NULL; | |
| 7340 reg_mmatch = rmp; | |
| 7341 reg_buf = curbuf; /* always works on the current buffer! */ | |
| 7342 reg_firstlnum = lnum; | |
| 7343 reg_maxline = curbuf->b_ml.ml_line_count - lnum; | |
| 5836 | 7344 reg_line_lbr = FALSE; |
| 7 | 7345 return vim_regsub_both(source, dest, copy, magic, backslash); |
| 7346 } | |
| 7347 | |
| 7348 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7349 vim_regsub_both( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7350 char_u *source, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7351 char_u *dest, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7352 int copy, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7353 int magic, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7354 int backslash) |
| 7 | 7355 { |
| 7356 char_u *src; | |
| 7357 char_u *dst; | |
| 7358 char_u *s; | |
| 7359 int c; | |
| 772 | 7360 int cc; |
| 7 | 7361 int no = -1; |
| 4244 | 7362 fptr_T func_all = (fptr_T)NULL; |
| 7363 fptr_T func_one = (fptr_T)NULL; | |
| 7 | 7364 linenr_T clnum = 0; /* init for GCC */ |
| 7365 int len = 0; /* init for GCC */ | |
| 7366 #ifdef FEAT_EVAL | |
| 7367 static char_u *eval_result = NULL; | |
| 7368 #endif | |
| 7369 | |
| 7370 /* Be paranoid... */ | |
| 7371 if (source == NULL || dest == NULL) | |
| 7372 { | |
| 7373 EMSG(_(e_null)); | |
| 7374 return 0; | |
| 7375 } | |
| 7376 if (prog_magic_wrong()) | |
| 7377 return 0; | |
| 7378 src = source; | |
| 7379 dst = dest; | |
| 7380 | |
| 7381 /* | |
| 7382 * When the substitute part starts with "\=" evaluate it as an expression. | |
| 7383 */ | |
| 7384 if (source[0] == '\\' && source[1] == '=' | |
| 7385 #ifdef FEAT_EVAL | |
| 7386 && !can_f_submatch /* can't do this recursively */ | |
| 7387 #endif | |
| 7388 ) | |
| 7389 { | |
| 7390 #ifdef FEAT_EVAL | |
| 7391 /* To make sure that the length doesn't change between checking the | |
| 7392 * length and copying the string, and to speed up things, the | |
| 7393 * resulting string is saved from the call with "copy" == FALSE to the | |
| 7394 * call with "copy" == TRUE. */ | |
| 7395 if (copy) | |
| 7396 { | |
| 7397 if (eval_result != NULL) | |
| 7398 { | |
| 7399 STRCPY(dest, eval_result); | |
| 7400 dst += STRLEN(eval_result); | |
| 7401 vim_free(eval_result); | |
| 7402 eval_result = NULL; | |
| 7403 } | |
| 7404 } | |
| 7405 else | |
| 7406 { | |
| 7407 win_T *save_reg_win; | |
| 7408 int save_ireg_ic; | |
| 7409 | |
| 7410 vim_free(eval_result); | |
| 7411 | |
| 7412 /* The expression may contain substitute(), which calls us | |
| 7413 * recursively. Make sure submatch() gets the text from the first | |
| 7414 * level. Don't need to save "reg_buf", because | |
| 7415 * vim_regexec_multi() can't be called recursively. */ | |
| 7416 submatch_match = reg_match; | |
| 7417 submatch_mmatch = reg_mmatch; | |
| 2011 | 7418 submatch_firstlnum = reg_firstlnum; |
| 7419 submatch_maxline = reg_maxline; | |
| 2904 | 7420 submatch_line_lbr = reg_line_lbr; |
| 7 | 7421 save_reg_win = reg_win; |
| 7422 save_ireg_ic = ireg_ic; | |
| 7423 can_f_submatch = TRUE; | |
| 7424 | |
| 714 | 7425 eval_result = eval_to_string(source + 2, NULL, TRUE); |
| 7 | 7426 if (eval_result != NULL) |
| 7427 { | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7428 int had_backslash = FALSE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7429 |
| 39 | 7430 for (s = eval_result; *s != NUL; mb_ptr_adv(s)) |
| 7 | 7431 { |
| 2904 | 7432 /* Change NL to CR, so that it becomes a line break, |
| 7433 * unless called from vim_regexec_nl(). | |
| 7 | 7434 * Skip over a backslashed character. */ |
| 2904 | 7435 if (*s == NL && !submatch_line_lbr) |
| 7 | 7436 *s = CAR; |
| 7437 else if (*s == '\\' && s[1] != NUL) | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7438 { |
| 7 | 7439 ++s; |
| 2173 | 7440 /* Change NL to CR here too, so that this works: |
| 7441 * :s/abc\\\ndef/\="aaa\\\nbbb"/ on text: | |
| 7442 * abc\ | |
| 7443 * def | |
| 2904 | 7444 * Not when called from vim_regexec_nl(). |
| 2173 | 7445 */ |
| 2904 | 7446 if (*s == NL && !submatch_line_lbr) |
| 2173 | 7447 *s = CAR; |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7448 had_backslash = TRUE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7449 } |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7450 } |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7451 if (had_backslash && backslash) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7452 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7453 /* Backslashes will be consumed, need to double them. */ |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7454 s = vim_strsave_escaped(eval_result, (char_u *)"\\"); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7455 if (s != NULL) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7456 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7457 vim_free(eval_result); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7458 eval_result = s; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7459 } |
| 7 | 7460 } |
| 7461 | |
| 7462 dst += STRLEN(eval_result); | |
| 7463 } | |
| 7464 | |
| 7465 reg_match = submatch_match; | |
| 7466 reg_mmatch = submatch_mmatch; | |
| 2011 | 7467 reg_firstlnum = submatch_firstlnum; |
| 7468 reg_maxline = submatch_maxline; | |
| 2904 | 7469 reg_line_lbr = submatch_line_lbr; |
| 7 | 7470 reg_win = save_reg_win; |
| 7471 ireg_ic = save_ireg_ic; | |
| 7472 can_f_submatch = FALSE; | |
| 7473 } | |
| 7474 #endif | |
| 7475 } | |
| 7476 else | |
| 7477 while ((c = *src++) != NUL) | |
| 7478 { | |
| 7479 if (c == '&' && magic) | |
| 7480 no = 0; | |
| 7481 else if (c == '\\' && *src != NUL) | |
| 7482 { | |
| 7483 if (*src == '&' && !magic) | |
| 7484 { | |
| 7485 ++src; | |
| 7486 no = 0; | |
| 7487 } | |
| 7488 else if ('0' <= *src && *src <= '9') | |
| 7489 { | |
| 7490 no = *src++ - '0'; | |
| 7491 } | |
| 7492 else if (vim_strchr((char_u *)"uUlLeE", *src)) | |
| 7493 { | |
| 7494 switch (*src++) | |
| 7495 { | |
| 4244 | 7496 case 'u': func_one = (fptr_T)do_upper; |
| 7 | 7497 continue; |
| 4244 | 7498 case 'U': func_all = (fptr_T)do_Upper; |
| 7 | 7499 continue; |
| 4244 | 7500 case 'l': func_one = (fptr_T)do_lower; |
| 7 | 7501 continue; |
| 4244 | 7502 case 'L': func_all = (fptr_T)do_Lower; |
| 7 | 7503 continue; |
| 7504 case 'e': | |
| 4244 | 7505 case 'E': func_one = func_all = (fptr_T)NULL; |
| 7 | 7506 continue; |
| 7507 } | |
| 7508 } | |
| 7509 } | |
| 7510 if (no < 0) /* Ordinary character. */ | |
| 7511 { | |
| 798 | 7512 if (c == K_SPECIAL && src[0] != NUL && src[1] != NUL) |
| 7513 { | |
| 1209 | 7514 /* Copy a special key as-is. */ |
| 798 | 7515 if (copy) |
| 7516 { | |
| 7517 *dst++ = c; | |
| 7518 *dst++ = *src++; | |
| 7519 *dst++ = *src++; | |
| 7520 } | |
| 7521 else | |
| 7522 { | |
| 7523 dst += 3; | |
| 7524 src += 2; | |
| 7525 } | |
| 7526 continue; | |
| 7527 } | |
| 7528 | |
| 7 | 7529 if (c == '\\' && *src != NUL) |
| 7530 { | |
| 7531 /* Check for abbreviations -- webb */ | |
| 7532 switch (*src) | |
| 7533 { | |
| 7534 case 'r': c = CAR; ++src; break; | |
| 7535 case 'n': c = NL; ++src; break; | |
| 7536 case 't': c = TAB; ++src; break; | |
| 7537 /* Oh no! \e already has meaning in subst pat :-( */ | |
| 7538 /* case 'e': c = ESC; ++src; break; */ | |
| 7539 case 'b': c = Ctrl_H; ++src; break; | |
| 7540 | |
| 7541 /* If "backslash" is TRUE the backslash will be removed | |
| 7542 * later. Used to insert a literal CR. */ | |
| 7543 default: if (backslash) | |
| 7544 { | |
| 7545 if (copy) | |
| 7546 *dst = '\\'; | |
| 7547 ++dst; | |
| 7548 } | |
| 7549 c = *src++; | |
| 7550 } | |
| 7551 } | |
| 798 | 7552 #ifdef FEAT_MBYTE |
| 7553 else if (has_mbyte) | |
| 7554 c = mb_ptr2char(src - 1); | |
| 7555 #endif | |
| 7 | 7556 |
| 7557 /* Write to buffer, if copy is set. */ | |
| 4244 | 7558 if (func_one != (fptr_T)NULL) |
| 7559 /* Turbo C complains without the typecast */ | |
| 7560 func_one = (fptr_T)(func_one(&cc, c)); | |
| 7561 else if (func_all != (fptr_T)NULL) | |
| 7562 /* Turbo C complains without the typecast */ | |
| 7563 func_all = (fptr_T)(func_all(&cc, c)); | |
| 7564 else /* just copy */ | |
| 772 | 7565 cc = c; |
| 7566 | |
| 7567 #ifdef FEAT_MBYTE | |
| 7568 if (has_mbyte) | |
| 7 | 7569 { |
|
2307
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7570 int totlen = mb_ptr2len(src - 1); |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7571 |
| 7 | 7572 if (copy) |
| 772 | 7573 mb_char2bytes(cc, dst); |
| 7574 dst += mb_char2len(cc) - 1; | |
|
2307
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7575 if (enc_utf8) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7576 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7577 int clen = utf_ptr2len(src - 1); |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7578 |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7579 /* If the character length is shorter than "totlen", there |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7580 * are composing characters; copy them as-is. */ |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7581 if (clen < totlen) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7582 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7583 if (copy) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7584 mch_memmove(dst + 1, src - 1 + clen, |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7585 (size_t)(totlen - clen)); |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7586 dst += totlen - clen; |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7587 } |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7588 } |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7589 src += totlen - 1; |
| 7 | 7590 } |
| 7591 else | |
| 7592 #endif | |
| 7593 if (copy) | |
| 772 | 7594 *dst = cc; |
| 7 | 7595 dst++; |
| 7596 } | |
| 7597 else | |
| 7598 { | |
| 7599 if (REG_MULTI) | |
| 7600 { | |
| 7601 clnum = reg_mmatch->startpos[no].lnum; | |
| 7602 if (clnum < 0 || reg_mmatch->endpos[no].lnum < 0) | |
| 7603 s = NULL; | |
| 7604 else | |
| 7605 { | |
| 7606 s = reg_getline(clnum) + reg_mmatch->startpos[no].col; | |
| 7607 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7608 len = reg_mmatch->endpos[no].col | |
| 7609 - reg_mmatch->startpos[no].col; | |
| 7610 else | |
| 7611 len = (int)STRLEN(s); | |
| 7612 } | |
| 7613 } | |
| 7614 else | |
| 7615 { | |
| 7616 s = reg_match->startp[no]; | |
| 7617 if (reg_match->endp[no] == NULL) | |
| 7618 s = NULL; | |
| 7619 else | |
| 7620 len = (int)(reg_match->endp[no] - s); | |
| 7621 } | |
| 7622 if (s != NULL) | |
| 7623 { | |
| 7624 for (;;) | |
| 7625 { | |
| 7626 if (len == 0) | |
| 7627 { | |
| 7628 if (REG_MULTI) | |
| 7629 { | |
| 7630 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7631 break; | |
| 7632 if (copy) | |
| 7633 *dst = CAR; | |
| 7634 ++dst; | |
| 7635 s = reg_getline(++clnum); | |
| 7636 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7637 len = reg_mmatch->endpos[no].col; | |
| 7638 else | |
| 7639 len = (int)STRLEN(s); | |
| 7640 } | |
| 7641 else | |
| 7642 break; | |
| 7643 } | |
| 7644 else if (*s == NUL) /* we hit NUL. */ | |
| 7645 { | |
| 7646 if (copy) | |
| 7647 EMSG(_(e_re_damg)); | |
| 7648 goto exit; | |
| 7649 } | |
| 7650 else | |
| 7651 { | |
| 7652 if (backslash && (*s == CAR || *s == '\\')) | |
| 7653 { | |
| 7654 /* | |
| 7655 * Insert a backslash in front of a CR, otherwise | |
| 7656 * it will be replaced by a line break. | |
| 7657 * Number of backslashes will be halved later, | |
| 7658 * double them here. | |
| 7659 */ | |
| 7660 if (copy) | |
| 7661 { | |
| 7662 dst[0] = '\\'; | |
| 7663 dst[1] = *s; | |
| 7664 } | |
| 7665 dst += 2; | |
| 7666 } | |
| 7667 else | |
| 7668 { | |
| 772 | 7669 #ifdef FEAT_MBYTE |
| 7670 if (has_mbyte) | |
| 7671 c = mb_ptr2char(s); | |
| 7672 else | |
| 7673 #endif | |
| 7674 c = *s; | |
| 7675 | |
| 4244 | 7676 if (func_one != (fptr_T)NULL) |
| 7677 /* Turbo C complains without the typecast */ | |
| 7678 func_one = (fptr_T)(func_one(&cc, c)); | |
| 7679 else if (func_all != (fptr_T)NULL) | |
| 7680 /* Turbo C complains without the typecast */ | |
| 7681 func_all = (fptr_T)(func_all(&cc, c)); | |
| 7682 else /* just copy */ | |
| 772 | 7683 cc = c; |
| 7684 | |
| 7685 #ifdef FEAT_MBYTE | |
| 7686 if (has_mbyte) | |
| 7 | 7687 { |
| 1332 | 7688 int l; |
| 7689 | |
| 7690 /* Copy composing characters separately, one | |
| 7691 * at a time. */ | |
| 7692 if (enc_utf8) | |
| 7693 l = utf_ptr2len(s) - 1; | |
| 7694 else | |
| 7695 l = mb_ptr2len(s) - 1; | |
| 772 | 7696 |
| 7697 s += l; | |
| 7698 len -= l; | |
| 7699 if (copy) | |
| 7700 mb_char2bytes(cc, dst); | |
| 7701 dst += mb_char2len(cc) - 1; | |
| 7 | 7702 } |
| 772 | 7703 else |
| 7704 #endif | |
| 7705 if (copy) | |
| 7706 *dst = cc; | |
| 7707 dst++; | |
| 7 | 7708 } |
| 772 | 7709 |
| 7 | 7710 ++s; |
| 7711 --len; | |
| 7712 } | |
| 7713 } | |
| 7714 } | |
| 7715 no = -1; | |
| 7716 } | |
| 7717 } | |
| 7718 if (copy) | |
| 7719 *dst = NUL; | |
| 7720 | |
| 7721 exit: | |
| 7722 return (int)((dst - dest) + 1); | |
| 7723 } | |
| 7724 | |
| 7725 #ifdef FEAT_EVAL | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7726 static char_u *reg_getline_submatch(linenr_T lnum); |
| 2012 | 7727 |
| 7 | 7728 /* |
| 2011 | 7729 * Call reg_getline() with the line numbers from the submatch. If a |
| 7730 * substitute() was used the reg_maxline and other values have been | |
| 7731 * overwritten. | |
| 7732 */ | |
| 7733 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7734 reg_getline_submatch(linenr_T lnum) |
| 2011 | 7735 { |
| 7736 char_u *s; | |
| 7737 linenr_T save_first = reg_firstlnum; | |
| 7738 linenr_T save_max = reg_maxline; | |
| 7739 | |
| 7740 reg_firstlnum = submatch_firstlnum; | |
| 7741 reg_maxline = submatch_maxline; | |
| 7742 | |
| 7743 s = reg_getline(lnum); | |
| 7744 | |
| 7745 reg_firstlnum = save_first; | |
| 7746 reg_maxline = save_max; | |
| 7747 return s; | |
| 7748 } | |
| 7749 | |
| 7750 /* | |
| 1209 | 7751 * Used for the submatch() function: get the string from the n'th submatch in |
| 7 | 7752 * allocated memory. |
| 7753 * Returns NULL when not in a ":s" command and for a non-existing submatch. | |
| 7754 */ | |
| 7755 char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7756 reg_submatch(int no) |
| 7 | 7757 { |
| 7758 char_u *retval = NULL; | |
| 7759 char_u *s; | |
| 7760 int len; | |
| 7761 int round; | |
| 7762 linenr_T lnum; | |
| 7763 | |
| 840 | 7764 if (!can_f_submatch || no < 0) |
| 7 | 7765 return NULL; |
| 7766 | |
| 7767 if (submatch_match == NULL) | |
| 7768 { | |
| 7769 /* | |
| 7770 * First round: compute the length and allocate memory. | |
| 7771 * Second round: copy the text. | |
| 7772 */ | |
| 7773 for (round = 1; round <= 2; ++round) | |
| 7774 { | |
| 7775 lnum = submatch_mmatch->startpos[no].lnum; | |
| 7776 if (lnum < 0 || submatch_mmatch->endpos[no].lnum < 0) | |
| 7777 return NULL; | |
| 7778 | |
| 2011 | 7779 s = reg_getline_submatch(lnum) + submatch_mmatch->startpos[no].col; |
| 7 | 7780 if (s == NULL) /* anti-crash check, cannot happen? */ |
| 7781 break; | |
| 7782 if (submatch_mmatch->endpos[no].lnum == lnum) | |
| 7783 { | |
| 7784 /* Within one line: take form start to end col. */ | |
| 7785 len = submatch_mmatch->endpos[no].col | |
| 7786 - submatch_mmatch->startpos[no].col; | |
| 7787 if (round == 2) | |
| 418 | 7788 vim_strncpy(retval, s, len); |
| 7 | 7789 ++len; |
| 7790 } | |
| 7791 else | |
| 7792 { | |
| 7793 /* Multiple lines: take start line from start col, middle | |
| 7794 * lines completely and end line up to end col. */ | |
| 7795 len = (int)STRLEN(s); | |
| 7796 if (round == 2) | |
| 7797 { | |
| 7798 STRCPY(retval, s); | |
| 7799 retval[len] = '\n'; | |
| 7800 } | |
| 7801 ++len; | |
| 7802 ++lnum; | |
| 7803 while (lnum < submatch_mmatch->endpos[no].lnum) | |
| 7804 { | |
| 2011 | 7805 s = reg_getline_submatch(lnum++); |
| 7 | 7806 if (round == 2) |
| 7807 STRCPY(retval + len, s); | |
| 7808 len += (int)STRLEN(s); | |
| 7809 if (round == 2) | |
| 7810 retval[len] = '\n'; | |
| 7811 ++len; | |
| 7812 } | |
| 7813 if (round == 2) | |
| 2011 | 7814 STRNCPY(retval + len, reg_getline_submatch(lnum), |
| 7 | 7815 submatch_mmatch->endpos[no].col); |
| 7816 len += submatch_mmatch->endpos[no].col; | |
| 7817 if (round == 2) | |
| 7818 retval[len] = NUL; | |
| 7819 ++len; | |
| 7820 } | |
| 7821 | |
| 840 | 7822 if (retval == NULL) |
| 7 | 7823 { |
| 7824 retval = lalloc((long_u)len, TRUE); | |
| 840 | 7825 if (retval == NULL) |
| 7 | 7826 return NULL; |
| 7827 } | |
| 7828 } | |
| 7829 } | |
| 7830 else | |
| 7831 { | |
| 1815 | 7832 s = submatch_match->startp[no]; |
| 7833 if (s == NULL || submatch_match->endp[no] == NULL) | |
| 7 | 7834 retval = NULL; |
| 7835 else | |
| 7836 retval = vim_strnsave(s, (int)(submatch_match->endp[no] - s)); | |
| 7837 } | |
| 7838 | |
| 7839 return retval; | |
| 7840 } | |
| 5794 | 7841 |
| 7842 /* | |
| 7843 * Used for the submatch() function with the optional non-zero argument: get | |
| 7844 * the list of strings from the n'th submatch in allocated memory with NULs | |
| 7845 * represented in NLs. | |
| 7846 * Returns a list of allocated strings. Returns NULL when not in a ":s" | |
| 7847 * command, for a non-existing submatch and for any error. | |
| 7848 */ | |
| 7849 list_T * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7850 reg_submatch_list(int no) |
| 5794 | 7851 { |
| 7852 char_u *s; | |
| 7853 linenr_T slnum; | |
| 7854 linenr_T elnum; | |
| 7855 colnr_T scol; | |
| 7856 colnr_T ecol; | |
| 7857 int i; | |
| 7858 list_T *list; | |
| 7859 int error = FALSE; | |
| 7860 | |
| 7861 if (!can_f_submatch || no < 0) | |
| 7862 return NULL; | |
| 7863 | |
| 7864 if (submatch_match == NULL) | |
| 7865 { | |
| 7866 slnum = submatch_mmatch->startpos[no].lnum; | |
| 7867 elnum = submatch_mmatch->endpos[no].lnum; | |
| 7868 if (slnum < 0 || elnum < 0) | |
| 7869 return NULL; | |
| 7870 | |
| 7871 scol = submatch_mmatch->startpos[no].col; | |
| 7872 ecol = submatch_mmatch->endpos[no].col; | |
| 7873 | |
| 7874 list = list_alloc(); | |
| 7875 if (list == NULL) | |
| 7876 return NULL; | |
| 7877 | |
| 7878 s = reg_getline_submatch(slnum) + scol; | |
| 7879 if (slnum == elnum) | |
| 7880 { | |
| 7881 if (list_append_string(list, s, ecol - scol) == FAIL) | |
| 7882 error = TRUE; | |
| 7883 } | |
| 7884 else | |
| 7885 { | |
| 7886 if (list_append_string(list, s, -1) == FAIL) | |
| 7887 error = TRUE; | |
| 7888 for (i = 1; i < elnum - slnum; i++) | |
| 7889 { | |
| 7890 s = reg_getline_submatch(slnum + i); | |
| 7891 if (list_append_string(list, s, -1) == FAIL) | |
| 7892 error = TRUE; | |
| 7893 } | |
| 7894 s = reg_getline_submatch(elnum); | |
| 7895 if (list_append_string(list, s, ecol) == FAIL) | |
| 7896 error = TRUE; | |
| 7897 } | |
| 7898 } | |
| 7899 else | |
| 7900 { | |
| 7901 s = submatch_match->startp[no]; | |
| 7902 if (s == NULL || submatch_match->endp[no] == NULL) | |
| 7903 return NULL; | |
| 7904 list = list_alloc(); | |
| 7905 if (list == NULL) | |
| 7906 return NULL; | |
| 7907 if (list_append_string(list, s, | |
| 7908 (int)(submatch_match->endp[no] - s)) == FAIL) | |
| 7909 error = TRUE; | |
| 7910 } | |
| 7911 | |
| 7912 if (error) | |
| 7913 { | |
| 7914 list_free(list, TRUE); | |
| 7915 return NULL; | |
| 7916 } | |
| 7917 return list; | |
| 7918 } | |
| 7 | 7919 #endif |
| 4444 | 7920 |
| 7921 static regengine_T bt_regengine = | |
| 7922 { | |
| 7923 bt_regcomp, | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7924 bt_regfree, |
| 4444 | 7925 bt_regexec_nl, |
| 6328 | 7926 bt_regexec_multi, |
| 7927 (char_u *)"" | |
| 4444 | 7928 }; |
| 7929 | |
| 7930 #include "regexp_nfa.c" | |
| 7931 | |
| 7932 static regengine_T nfa_regengine = | |
| 7933 { | |
| 7934 nfa_regcomp, | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7935 nfa_regfree, |
| 4444 | 7936 nfa_regexec_nl, |
| 6328 | 7937 nfa_regexec_multi, |
| 7938 (char_u *)"" | |
| 4444 | 7939 }; |
| 7940 | |
| 7941 /* Which regexp engine to use? Needed for vim_regcomp(). | |
| 7942 * Must match with 'regexpengine'. */ | |
| 7943 static int regexp_engine = 0; | |
| 6328 | 7944 |
| 4444 | 7945 #ifdef DEBUG |
| 7946 static char_u regname[][30] = { | |
| 7947 "AUTOMATIC Regexp Engine", | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
7948 "BACKTRACKING Regexp Engine", |
| 4444 | 7949 "NFA Regexp Engine" |
| 7950 }; | |
| 7951 #endif | |
| 7952 | |
| 7953 /* | |
| 7954 * Compile a regular expression into internal code. | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7955 * Returns the program in allocated memory. |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7956 * Use vim_regfree() to free the memory. |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7957 * Returns NULL for an error. |
| 4444 | 7958 */ |
| 7959 regprog_T * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7960 vim_regcomp(char_u *expr_arg, int re_flags) |
| 4444 | 7961 { |
| 7962 regprog_T *prog = NULL; | |
| 7963 char_u *expr = expr_arg; | |
| 7964 | |
| 7965 regexp_engine = p_re; | |
| 7966 | |
| 7967 /* Check for prefix "\%#=", that sets the regexp engine */ | |
| 7968 if (STRNCMP(expr, "\\%#=", 4) == 0) | |
| 7969 { | |
| 7970 int newengine = expr[4] - '0'; | |
| 7971 | |
| 7972 if (newengine == AUTOMATIC_ENGINE | |
| 7973 || newengine == BACKTRACKING_ENGINE | |
| 7974 || newengine == NFA_ENGINE) | |
| 7975 { | |
| 7976 regexp_engine = expr[4] - '0'; | |
| 7977 expr += 5; | |
| 7978 #ifdef DEBUG | |
| 5897 | 7979 smsg((char_u *)"New regexp mode selected (%d): %s", |
| 7980 regexp_engine, regname[newengine]); | |
| 4444 | 7981 #endif |
| 7982 } | |
| 7983 else | |
| 7984 { | |
| 7985 EMSG(_("E864: \\%#= can only be followed by 0, 1, or 2. The automatic engine will be used ")); | |
| 7986 regexp_engine = AUTOMATIC_ENGINE; | |
| 7987 } | |
| 7988 } | |
| 7989 bt_regengine.expr = expr; | |
| 7990 nfa_regengine.expr = expr; | |
| 7991 | |
| 7992 /* | |
| 7993 * First try the NFA engine, unless backtracking was requested. | |
| 7994 */ | |
| 7995 if (regexp_engine != BACKTRACKING_ENGINE) | |
| 6533 | 7996 prog = nfa_regengine.regcomp(expr, |
| 7997 re_flags + (regexp_engine == AUTOMATIC_ENGINE ? RE_AUTO : 0)); | |
| 4444 | 7998 else |
| 7999 prog = bt_regengine.regcomp(expr, re_flags); | |
| 8000 | |
| 6328 | 8001 /* Check for error compiling regexp with initial engine. */ |
| 8002 if (prog == NULL) | |
| 4444 | 8003 { |
| 4460 | 8004 #ifdef BT_REGEXP_DEBUG_LOG |
| 4444 | 8005 if (regexp_engine != BACKTRACKING_ENGINE) /* debugging log for NFA */ |
| 8006 { | |
| 8007 FILE *f; | |
| 4460 | 8008 f = fopen(BT_REGEXP_DEBUG_LOG_NAME, "a"); |
| 4444 | 8009 if (f) |
| 8010 { | |
|
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8011 fprintf(f, "Syntax error in \"%s\"\n", expr); |
| 4444 | 8012 fclose(f); |
| 8013 } | |
| 8014 else | |
| 4460 | 8015 EMSG2("(NFA) Could not open \"%s\" to write !!!", |
| 8016 BT_REGEXP_DEBUG_LOG_NAME); | |
| 4444 | 8017 } |
| 8018 #endif | |
| 8019 /* | |
| 6328 | 8020 * If the NFA engine failed, try the backtracking engine. |
| 6533 | 8021 * The NFA engine also fails for patterns that it can't handle well |
| 8022 * but are still valid patterns, thus a retry should work. | |
| 8023 */ | |
| 4444 | 8024 if (regexp_engine == AUTOMATIC_ENGINE) |
| 6328 | 8025 { |
| 6533 | 8026 regexp_engine = BACKTRACKING_ENGINE; |
|
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8027 prog = bt_regengine.regcomp(expr, re_flags); |
| 6328 | 8028 } |
|
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8029 } |
| 4444 | 8030 |
| 6328 | 8031 if (prog != NULL) |
| 8032 { | |
| 8033 /* Store the info needed to call regcomp() again when the engine turns | |
| 8034 * out to be very slow when executing it. */ | |
| 8035 prog->re_engine = regexp_engine; | |
| 8036 prog->re_flags = re_flags; | |
| 8037 } | |
| 8038 | |
| 4444 | 8039 return prog; |
| 8040 } | |
| 8041 | |
| 8042 /* | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8043 * Free a compiled regexp program, returned by vim_regcomp(). |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8044 */ |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8045 void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8046 vim_regfree(regprog_T *prog) |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8047 { |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8048 if (prog != NULL) |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8049 prog->engine->regfree(prog); |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8050 } |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8051 |
| 6328 | 8052 #ifdef FEAT_EVAL |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
8053 static void report_re_switch(char_u *pat); |
| 6328 | 8054 |
| 8055 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8056 report_re_switch(char_u *pat) |
| 6328 | 8057 { |
| 8058 if (p_verbose > 0) | |
| 8059 { | |
| 8060 verbose_enter(); | |
| 8061 MSG_PUTS(_("Switching to backtracking RE engine for pattern: ")); | |
| 8062 MSG_PUTS(pat); | |
| 8063 verbose_leave(); | |
| 8064 } | |
| 8065 } | |
| 8066 #endif | |
| 8067 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
8068 static int vim_regexec_both(regmatch_T *rmp, char_u *line, colnr_T col, int nl); |
| 6328 | 8069 |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8070 /* |
| 4444 | 8071 * Match a regexp against a string. |
| 8072 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 6375 | 8073 * Note: "rmp->regprog" may be freed and changed. |
| 4444 | 8074 * Uses curbuf for line count and 'iskeyword'. |
| 6328 | 8075 * When "nl" is TRUE consider a "\n" in "line" to be a line break. |
| 4444 | 8076 * |
| 8077 * Return TRUE if there is a match, FALSE if not. | |
| 8078 */ | |
| 6328 | 8079 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8080 vim_regexec_both( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8081 regmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8082 char_u *line, /* string to match against */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8083 colnr_T col, /* column to start looking for match */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8084 int nl) |
| 6328 | 8085 { |
| 8086 int result = rmp->regprog->engine->regexec_nl(rmp, line, col, nl); | |
| 8087 | |
| 8088 /* NFA engine aborted because it's very slow. */ | |
| 8089 if (rmp->regprog->re_engine == AUTOMATIC_ENGINE | |
| 8090 && result == NFA_TOO_EXPENSIVE) | |
| 8091 { | |
| 8092 int save_p_re = p_re; | |
| 8093 int re_flags = rmp->regprog->re_flags; | |
| 8094 char_u *pat = vim_strsave(((nfa_regprog_T *)rmp->regprog)->pattern); | |
| 8095 | |
| 8096 p_re = BACKTRACKING_ENGINE; | |
| 8097 vim_regfree(rmp->regprog); | |
| 8098 if (pat != NULL) | |
| 8099 { | |
| 8100 #ifdef FEAT_EVAL | |
| 8101 report_re_switch(pat); | |
| 8102 #endif | |
| 8103 rmp->regprog = vim_regcomp(pat, re_flags); | |
| 8104 if (rmp->regprog != NULL) | |
| 8105 result = rmp->regprog->engine->regexec_nl(rmp, line, col, nl); | |
| 8106 vim_free(pat); | |
| 8107 } | |
| 8108 | |
| 8109 p_re = save_p_re; | |
| 8110 } | |
| 6390 | 8111 return result > 0; |
| 6328 | 8112 } |
| 8113 | |
| 6375 | 8114 /* |
| 8115 * Note: "*prog" may be freed and changed. | |
| 6390 | 8116 * Return TRUE if there is a match, FALSE if not. |
| 6375 | 8117 */ |
| 8118 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8119 vim_regexec_prog( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8120 regprog_T **prog, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8121 int ignore_case, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8122 char_u *line, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8123 colnr_T col) |
| 6375 | 8124 { |
| 8125 int r; | |
| 8126 regmatch_T regmatch; | |
| 8127 | |
| 8128 regmatch.regprog = *prog; | |
| 8129 regmatch.rm_ic = ignore_case; | |
| 8130 r = vim_regexec_both(®match, line, col, FALSE); | |
| 8131 *prog = regmatch.regprog; | |
| 8132 return r; | |
| 8133 } | |
| 8134 | |
| 8135 /* | |
| 8136 * Note: "rmp->regprog" may be freed and changed. | |
| 6390 | 8137 * Return TRUE if there is a match, FALSE if not. |
| 6375 | 8138 */ |
| 4444 | 8139 int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8140 vim_regexec(regmatch_T *rmp, char_u *line, colnr_T col) |
| 4444 | 8141 { |
| 6328 | 8142 return vim_regexec_both(rmp, line, col, FALSE); |
| 4444 | 8143 } |
| 8144 | |
| 8145 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
| 8146 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 8147 /* | |
| 8148 * Like vim_regexec(), but consider a "\n" in "line" to be a line break. | |
| 6375 | 8149 * Note: "rmp->regprog" may be freed and changed. |
| 6390 | 8150 * Return TRUE if there is a match, FALSE if not. |
| 4444 | 8151 */ |
| 8152 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8153 vim_regexec_nl(regmatch_T *rmp, char_u *line, colnr_T col) |
| 4444 | 8154 { |
| 6328 | 8155 return vim_regexec_both(rmp, line, col, TRUE); |
| 4444 | 8156 } |
| 8157 #endif | |
| 8158 | |
| 8159 /* | |
| 8160 * Match a regexp against multiple lines. | |
| 8161 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 6375 | 8162 * Note: "rmp->regprog" may be freed and changed. |
| 4444 | 8163 * Uses curbuf for line count and 'iskeyword'. |
| 8164 * | |
| 8165 * Return zero if there is no match. Return number of lines contained in the | |
| 8166 * match otherwise. | |
| 8167 */ | |
| 8168 long | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8169 vim_regexec_multi( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8170 regmmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8171 win_T *win, /* window in which to search or NULL */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8172 buf_T *buf, /* buffer in which to search */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8173 linenr_T lnum, /* nr of line to start looking for match */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8174 colnr_T col, /* column to start looking for match */ |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8175 proftime_T *tm) /* timeout limit or NULL */ |
| 4444 | 8176 { |
| 6328 | 8177 int result = rmp->regprog->engine->regexec_multi( |
| 8178 rmp, win, buf, lnum, col, tm); | |
| 8179 | |
| 8180 /* NFA engine aborted because it's very slow. */ | |
| 8181 if (rmp->regprog->re_engine == AUTOMATIC_ENGINE | |
| 8182 && result == NFA_TOO_EXPENSIVE) | |
| 8183 { | |
| 8184 int save_p_re = p_re; | |
| 8185 int re_flags = rmp->regprog->re_flags; | |
| 8186 char_u *pat = vim_strsave(((nfa_regprog_T *)rmp->regprog)->pattern); | |
| 8187 | |
| 8188 p_re = BACKTRACKING_ENGINE; | |
| 8189 vim_regfree(rmp->regprog); | |
| 8190 if (pat != NULL) | |
| 8191 { | |
| 8192 #ifdef FEAT_EVAL | |
| 8193 report_re_switch(pat); | |
| 8194 #endif | |
| 8195 rmp->regprog = vim_regcomp(pat, re_flags); | |
| 8196 if (rmp->regprog != NULL) | |
| 8197 result = rmp->regprog->engine->regexec_multi( | |
| 8198 rmp, win, buf, lnum, col, tm); | |
| 8199 vim_free(pat); | |
| 8200 } | |
| 8201 p_re = save_p_re; | |
| 8202 } | |
| 8203 | |
| 6390 | 8204 return result <= 0 ? 0 : result; |
| 4444 | 8205 } |