Mercurial > vim
annotate src/regexp.c @ 8860:3c50de41b173
Added tag v7.4.1717 for changeset 03250bc0c63aad58f86f10e3a423ae62fa9defe9
| author | Christian Brabandt <cb@256bit.org> |
|---|---|
| date | Thu, 07 Apr 2016 21:45:06 +0200 |
| parents | 318eaa6fa973 |
| children | e1b84109506a |
| 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", |
|
8825
318eaa6fa973
commit https://github.com/vim/vim/commit/22e421549d54147d003f6444de007cb1d73f1d27
Christian Brabandt <cb@256bit.org>
parents:
8212
diff
changeset
|
794 "O\xEB\xEC\xED\xEE\xEF\x80", |
|
2247
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", |
|
8825
318eaa6fa973
commit https://github.com/vim/vim/commit/22e421549d54147d003f6444de007cb1d73f1d27
Christian Brabandt <cb@256bit.org>
parents:
8212
diff
changeset
|
802 "o\xCB\xCC\xCD\xCE\xCF\x70", |
|
2247
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 /* Allocate space. */ | |
| 4444 | 1342 r = (bt_regprog_T *)lalloc(sizeof(bt_regprog_T) + regsize, TRUE); |
| 7 | 1343 if (r == NULL) |
| 1344 return NULL; | |
| 1345 | |
| 1346 /* | |
| 1347 * Second pass: emit code. | |
| 1348 */ | |
| 1349 regcomp_start(expr, re_flags); | |
| 1350 regcode = r->program; | |
| 1351 regc(REGMAGIC); | |
| 2010 | 1352 if (reg(REG_NOPAREN, &flags) == NULL || reg_toolong) |
| 7 | 1353 { |
| 1354 vim_free(r); | |
| 2010 | 1355 if (reg_toolong) |
| 1356 EMSG_RET_NULL(_("E339: Pattern too long")); | |
| 7 | 1357 return NULL; |
| 1358 } | |
| 1359 | |
| 1360 /* Dig out information for optimizations. */ | |
| 1361 r->regstart = NUL; /* Worst-case defaults. */ | |
| 1362 r->reganch = 0; | |
| 1363 r->regmust = NULL; | |
| 1364 r->regmlen = 0; | |
| 1365 r->regflags = regflags; | |
| 1366 if (flags & HASNL) | |
| 1367 r->regflags |= RF_HASNL; | |
| 1368 if (flags & HASLOOKBH) | |
| 1369 r->regflags |= RF_LOOKBH; | |
| 1370 #ifdef FEAT_SYN_HL | |
| 1371 /* Remember whether this pattern has any \z specials in it. */ | |
| 1372 r->reghasz = re_has_z; | |
| 1373 #endif | |
| 1374 scan = r->program + 1; /* First BRANCH. */ | |
| 1375 if (OP(regnext(scan)) == END) /* Only one top-level choice. */ | |
| 1376 { | |
| 1377 scan = OPERAND(scan); | |
| 1378 | |
| 1379 /* Starting-point info. */ | |
| 1380 if (OP(scan) == BOL || OP(scan) == RE_BOF) | |
| 1381 { | |
| 1382 r->reganch++; | |
| 1383 scan = regnext(scan); | |
| 1384 } | |
| 1385 | |
| 1386 if (OP(scan) == EXACTLY) | |
| 1387 { | |
| 1388 #ifdef FEAT_MBYTE | |
| 1389 if (has_mbyte) | |
| 1390 r->regstart = (*mb_ptr2char)(OPERAND(scan)); | |
| 1391 else | |
| 1392 #endif | |
| 1393 r->regstart = *OPERAND(scan); | |
| 1394 } | |
| 1395 else if ((OP(scan) == BOW | |
| 1396 || OP(scan) == EOW | |
| 1397 || OP(scan) == NOTHING | |
| 1398 || OP(scan) == MOPEN + 0 || OP(scan) == NOPEN | |
| 1399 || OP(scan) == MCLOSE + 0 || OP(scan) == NCLOSE) | |
| 1400 && OP(regnext(scan)) == EXACTLY) | |
| 1401 { | |
| 1402 #ifdef FEAT_MBYTE | |
| 1403 if (has_mbyte) | |
| 1404 r->regstart = (*mb_ptr2char)(OPERAND(regnext(scan))); | |
| 1405 else | |
| 1406 #endif | |
| 1407 r->regstart = *OPERAND(regnext(scan)); | |
| 1408 } | |
| 1409 | |
| 1410 /* | |
| 1411 * If there's something expensive in the r.e., find the longest | |
| 1412 * literal string that must appear and make it the regmust. Resolve | |
| 1413 * ties in favor of later strings, since the regstart check works | |
| 1414 * with the beginning of the r.e. and avoiding duplication | |
| 1415 * strengthens checking. Not a strong reason, but sufficient in the | |
| 1416 * absence of others. | |
| 1417 */ | |
| 1418 /* | |
| 1419 * When the r.e. starts with BOW, it is faster to look for a regmust | |
| 1420 * first. Used a lot for "#" and "*" commands. (Added by mool). | |
| 1421 */ | |
| 1422 if ((flags & SPSTART || OP(scan) == BOW || OP(scan) == EOW) | |
| 1423 && !(flags & HASNL)) | |
| 1424 { | |
| 1425 longest = NULL; | |
| 1426 len = 0; | |
| 1427 for (; scan != NULL; scan = regnext(scan)) | |
| 1428 if (OP(scan) == EXACTLY && STRLEN(OPERAND(scan)) >= (size_t)len) | |
| 1429 { | |
| 1430 longest = OPERAND(scan); | |
| 1431 len = (int)STRLEN(OPERAND(scan)); | |
| 1432 } | |
| 1433 r->regmust = longest; | |
| 1434 r->regmlen = len; | |
| 1435 } | |
| 1436 } | |
| 4444 | 1437 #ifdef BT_REGEXP_DUMP |
| 7 | 1438 regdump(expr, r); |
| 1439 #endif | |
| 4444 | 1440 r->engine = &bt_regengine; |
| 1441 return (regprog_T *)r; | |
| 7 | 1442 } |
| 1443 | |
| 1444 /* | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1445 * 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
|
1446 */ |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1447 static void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1448 bt_regfree(regprog_T *prog) |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1449 { |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1450 vim_free(prog); |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1451 } |
|
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 /* |
| 7 | 1454 * Setup to parse the regexp. Used once to get the length and once to do it. |
| 1455 */ | |
| 1456 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1457 regcomp_start( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1458 char_u *expr, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1459 int re_flags) /* see vim_regcomp() */ |
| 7 | 1460 { |
| 1461 initchr(expr); | |
| 1462 if (re_flags & RE_MAGIC) | |
| 1463 reg_magic = MAGIC_ON; | |
| 1464 else | |
| 1465 reg_magic = MAGIC_OFF; | |
| 1466 reg_string = (re_flags & RE_STRING); | |
| 481 | 1467 reg_strict = (re_flags & RE_STRICT); |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1468 get_cpo_flags(); |
| 7 | 1469 |
| 1470 num_complex_braces = 0; | |
| 1471 regnpar = 1; | |
| 1472 vim_memset(had_endbrace, 0, sizeof(had_endbrace)); | |
| 1473 #ifdef FEAT_SYN_HL | |
| 1474 regnzpar = 1; | |
| 1475 re_has_z = 0; | |
| 1476 #endif | |
| 1477 regsize = 0L; | |
| 2010 | 1478 reg_toolong = FALSE; |
| 7 | 1479 regflags = 0; |
| 1480 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1481 had_eol = FALSE; | |
| 1482 #endif | |
| 1483 } | |
| 1484 | |
| 1485 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1486 /* | |
| 1487 * Check if during the previous call to vim_regcomp the EOL item "$" has been | |
| 1488 * found. This is messy, but it works fine. | |
| 1489 */ | |
| 1490 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1491 vim_regcomp_had_eol(void) |
| 7 | 1492 { |
| 1493 return had_eol; | |
| 1494 } | |
| 1495 #endif | |
| 1496 | |
|
8021
b6b4f354df23
commit https://github.com/vim/vim/commit/7c29f387819b5817b003d2ba73e2b5cf3cb3d0dd
Christian Brabandt <cb@256bit.org>
parents:
7833
diff
changeset
|
1497 /* variables for parsing reginput */ |
|
b6b4f354df23
commit https://github.com/vim/vim/commit/7c29f387819b5817b003d2ba73e2b5cf3cb3d0dd
Christian Brabandt <cb@256bit.org>
parents:
7833
diff
changeset
|
1498 static int at_start; /* True when on the first character */ |
|
b6b4f354df23
commit https://github.com/vim/vim/commit/7c29f387819b5817b003d2ba73e2b5cf3cb3d0dd
Christian Brabandt <cb@256bit.org>
parents:
7833
diff
changeset
|
1499 static int prev_at_start; /* True when on the second character */ |
|
b6b4f354df23
commit https://github.com/vim/vim/commit/7c29f387819b5817b003d2ba73e2b5cf3cb3d0dd
Christian Brabandt <cb@256bit.org>
parents:
7833
diff
changeset
|
1500 |
| 7 | 1501 /* |
| 4444 | 1502 * Parse regular expression, i.e. main body or parenthesized thing. |
| 7 | 1503 * |
| 1504 * Caller must absorb opening parenthesis. | |
| 1505 * | |
| 1506 * Combining parenthesis handling with the base level of regular expression | |
| 1507 * is a trifle forced, but the need to tie the tails of the branches to what | |
| 1508 * follows makes it hard to avoid. | |
| 1509 */ | |
| 1510 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1511 reg( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1512 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
|
1513 int *flagp) |
| 7 | 1514 { |
| 1515 char_u *ret; | |
| 1516 char_u *br; | |
| 1517 char_u *ender; | |
| 1518 int parno = 0; | |
| 1519 int flags; | |
| 1520 | |
| 1521 *flagp = HASWIDTH; /* Tentatively. */ | |
| 1522 | |
| 1523 #ifdef FEAT_SYN_HL | |
| 1524 if (paren == REG_ZPAREN) | |
| 1525 { | |
| 1526 /* Make a ZOPEN node. */ | |
| 1527 if (regnzpar >= NSUBEXP) | |
| 1528 EMSG_RET_NULL(_("E50: Too many \\z(")); | |
| 1529 parno = regnzpar; | |
| 1530 regnzpar++; | |
| 1531 ret = regnode(ZOPEN + parno); | |
| 1532 } | |
| 1533 else | |
| 1534 #endif | |
| 1535 if (paren == REG_PAREN) | |
| 1536 { | |
| 1537 /* Make a MOPEN node. */ | |
| 1538 if (regnpar >= NSUBEXP) | |
| 4444 | 1539 EMSG2_RET_NULL(_("E51: Too many %s("), reg_magic == MAGIC_ALL); |
| 7 | 1540 parno = regnpar; |
| 1541 ++regnpar; | |
| 1542 ret = regnode(MOPEN + parno); | |
| 1543 } | |
| 1544 else if (paren == REG_NPAREN) | |
| 1545 { | |
| 1546 /* Make a NOPEN node. */ | |
| 1547 ret = regnode(NOPEN); | |
| 1548 } | |
| 1549 else | |
| 1550 ret = NULL; | |
| 1551 | |
| 1552 /* Pick up the branches, linking them together. */ | |
| 1553 br = regbranch(&flags); | |
| 1554 if (br == NULL) | |
| 1555 return NULL; | |
| 1556 if (ret != NULL) | |
| 1557 regtail(ret, br); /* [MZ]OPEN -> first. */ | |
| 1558 else | |
| 1559 ret = br; | |
| 1560 /* If one of the branches can be zero-width, the whole thing can. | |
| 1561 * If one of the branches has * at start or matches a line-break, the | |
| 1562 * whole thing can. */ | |
| 1563 if (!(flags & HASWIDTH)) | |
| 1564 *flagp &= ~HASWIDTH; | |
| 1565 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1566 while (peekchr() == Magic('|')) | |
| 1567 { | |
| 1568 skipchr(); | |
| 1569 br = regbranch(&flags); | |
| 2010 | 1570 if (br == NULL || reg_toolong) |
| 7 | 1571 return NULL; |
| 1572 regtail(ret, br); /* BRANCH -> BRANCH. */ | |
| 1573 if (!(flags & HASWIDTH)) | |
| 1574 *flagp &= ~HASWIDTH; | |
| 1575 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1576 } | |
| 1577 | |
| 1578 /* Make a closing node, and hook it on the end. */ | |
| 1579 ender = regnode( | |
| 1580 #ifdef FEAT_SYN_HL | |
| 1581 paren == REG_ZPAREN ? ZCLOSE + parno : | |
| 1582 #endif | |
| 1583 paren == REG_PAREN ? MCLOSE + parno : | |
| 1584 paren == REG_NPAREN ? NCLOSE : END); | |
| 1585 regtail(ret, ender); | |
| 1586 | |
| 1587 /* Hook the tails of the branches to the closing node. */ | |
| 1588 for (br = ret; br != NULL; br = regnext(br)) | |
| 1589 regoptail(br, ender); | |
| 1590 | |
| 1591 /* Check for proper termination. */ | |
| 1592 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
| 1593 { | |
| 1594 #ifdef FEAT_SYN_HL | |
| 1595 if (paren == REG_ZPAREN) | |
| 308 | 1596 EMSG_RET_NULL(_("E52: Unmatched \\z(")); |
| 7 | 1597 else |
| 1598 #endif | |
| 1599 if (paren == REG_NPAREN) | |
| 4444 | 1600 EMSG2_RET_NULL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); |
| 7 | 1601 else |
| 4444 | 1602 EMSG2_RET_NULL(_(e_unmatchedp), reg_magic == MAGIC_ALL); |
| 7 | 1603 } |
| 1604 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
| 1605 { | |
| 1606 if (curchr == Magic(')')) | |
| 4444 | 1607 EMSG2_RET_NULL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); |
| 7 | 1608 else |
| 308 | 1609 EMSG_RET_NULL(_(e_trailing)); /* "Can't happen". */ |
| 7 | 1610 /* NOTREACHED */ |
| 1611 } | |
| 1612 /* | |
| 1613 * Here we set the flag allowing back references to this set of | |
| 1614 * parentheses. | |
| 1615 */ | |
| 1616 if (paren == REG_PAREN) | |
| 1617 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
| 1618 return ret; | |
| 1619 } | |
| 1620 | |
| 1621 /* | |
| 4444 | 1622 * Parse one alternative of an | operator. |
| 7 | 1623 * Implements the & operator. |
| 1624 */ | |
| 1625 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1626 regbranch(int *flagp) |
| 7 | 1627 { |
| 1628 char_u *ret; | |
| 1629 char_u *chain = NULL; | |
| 1630 char_u *latest; | |
| 1631 int flags; | |
| 1632 | |
| 1633 *flagp = WORST | HASNL; /* Tentatively. */ | |
| 1634 | |
| 1635 ret = regnode(BRANCH); | |
| 1636 for (;;) | |
| 1637 { | |
| 1638 latest = regconcat(&flags); | |
| 1639 if (latest == NULL) | |
| 1640 return NULL; | |
| 1641 /* If one of the branches has width, the whole thing has. If one of | |
| 1642 * the branches anchors at start-of-line, the whole thing does. | |
| 1643 * If one of the branches uses look-behind, the whole thing does. */ | |
| 1644 *flagp |= flags & (HASWIDTH | SPSTART | HASLOOKBH); | |
| 1645 /* If one of the branches doesn't match a line-break, the whole thing | |
| 1646 * doesn't. */ | |
| 1647 *flagp &= ~HASNL | (flags & HASNL); | |
| 1648 if (chain != NULL) | |
| 1649 regtail(chain, latest); | |
| 1650 if (peekchr() != Magic('&')) | |
| 1651 break; | |
| 1652 skipchr(); | |
| 1653 regtail(latest, regnode(END)); /* operand ends */ | |
| 2010 | 1654 if (reg_toolong) |
| 1655 break; | |
| 7 | 1656 reginsert(MATCH, latest); |
| 1657 chain = latest; | |
| 1658 } | |
| 1659 | |
| 1660 return ret; | |
| 1661 } | |
| 1662 | |
| 1663 /* | |
| 4444 | 1664 * Parse one alternative of an | or & operator. |
| 7 | 1665 * Implements the concatenation operator. |
| 1666 */ | |
| 1667 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1668 regconcat(int *flagp) |
| 7 | 1669 { |
| 1670 char_u *first = NULL; | |
| 1671 char_u *chain = NULL; | |
| 1672 char_u *latest; | |
| 1673 int flags; | |
| 1674 int cont = TRUE; | |
| 1675 | |
| 1676 *flagp = WORST; /* Tentatively. */ | |
| 1677 | |
| 1678 while (cont) | |
| 1679 { | |
| 1680 switch (peekchr()) | |
| 1681 { | |
| 1682 case NUL: | |
| 1683 case Magic('|'): | |
| 1684 case Magic('&'): | |
| 1685 case Magic(')'): | |
| 1686 cont = FALSE; | |
| 1687 break; | |
| 1688 case Magic('Z'): | |
| 1689 #ifdef FEAT_MBYTE | |
| 1690 regflags |= RF_ICOMBINE; | |
| 1691 #endif | |
| 1692 skipchr_keepstart(); | |
| 1693 break; | |
| 1694 case Magic('c'): | |
| 1695 regflags |= RF_ICASE; | |
| 1696 skipchr_keepstart(); | |
| 1697 break; | |
| 1698 case Magic('C'): | |
| 1699 regflags |= RF_NOICASE; | |
| 1700 skipchr_keepstart(); | |
| 1701 break; | |
| 1702 case Magic('v'): | |
| 1703 reg_magic = MAGIC_ALL; | |
| 1704 skipchr_keepstart(); | |
| 1705 curchr = -1; | |
| 1706 break; | |
| 1707 case Magic('m'): | |
| 1708 reg_magic = MAGIC_ON; | |
| 1709 skipchr_keepstart(); | |
| 1710 curchr = -1; | |
| 1711 break; | |
| 1712 case Magic('M'): | |
| 1713 reg_magic = MAGIC_OFF; | |
| 1714 skipchr_keepstart(); | |
| 1715 curchr = -1; | |
| 1716 break; | |
| 1717 case Magic('V'): | |
| 1718 reg_magic = MAGIC_NONE; | |
| 1719 skipchr_keepstart(); | |
| 1720 curchr = -1; | |
| 1721 break; | |
| 1722 default: | |
| 1723 latest = regpiece(&flags); | |
| 2010 | 1724 if (latest == NULL || reg_toolong) |
| 7 | 1725 return NULL; |
| 1726 *flagp |= flags & (HASWIDTH | HASNL | HASLOOKBH); | |
| 1727 if (chain == NULL) /* First piece. */ | |
| 1728 *flagp |= flags & SPSTART; | |
| 1729 else | |
| 1730 regtail(chain, latest); | |
| 1731 chain = latest; | |
| 1732 if (first == NULL) | |
| 1733 first = latest; | |
| 1734 break; | |
| 1735 } | |
| 1736 } | |
| 1737 if (first == NULL) /* Loop ran zero times. */ | |
| 1738 first = regnode(NOTHING); | |
| 1739 return first; | |
| 1740 } | |
| 1741 | |
| 1742 /* | |
| 4444 | 1743 * Parse something followed by possible [*+=]. |
| 7 | 1744 * |
| 1745 * Note that the branching code sequences used for = and the general cases | |
| 1746 * of * and + are somewhat optimized: they use the same NOTHING node as | |
| 1747 * both the endmarker for their branch list and the body of the last branch. | |
| 1748 * It might seem that this node could be dispensed with entirely, but the | |
| 1749 * endmarker role is not redundant. | |
| 1750 */ | |
| 1751 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1752 regpiece(int *flagp) |
| 7 | 1753 { |
| 1754 char_u *ret; | |
| 1755 int op; | |
| 1756 char_u *next; | |
| 1757 int flags; | |
| 1758 long minval; | |
| 1759 long maxval; | |
| 1760 | |
| 1761 ret = regatom(&flags); | |
| 1762 if (ret == NULL) | |
| 1763 return NULL; | |
| 1764 | |
| 1765 op = peekchr(); | |
| 1766 if (re_multi_type(op) == NOT_MULTI) | |
| 1767 { | |
| 1768 *flagp = flags; | |
| 1769 return ret; | |
| 1770 } | |
| 1771 /* default flags */ | |
| 1772 *flagp = (WORST | SPSTART | (flags & (HASNL | HASLOOKBH))); | |
| 1773 | |
| 1774 skipchr(); | |
| 1775 switch (op) | |
| 1776 { | |
| 1777 case Magic('*'): | |
| 1778 if (flags & SIMPLE) | |
| 1779 reginsert(STAR, ret); | |
| 1780 else | |
| 1781 { | |
| 1782 /* Emit x* as (x&|), where & means "self". */ | |
| 1783 reginsert(BRANCH, ret); /* Either x */ | |
| 1784 regoptail(ret, regnode(BACK)); /* and loop */ | |
| 1785 regoptail(ret, ret); /* back */ | |
| 1786 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1787 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1788 } | |
| 1789 break; | |
| 1790 | |
| 1791 case Magic('+'): | |
| 1792 if (flags & SIMPLE) | |
| 1793 reginsert(PLUS, ret); | |
| 1794 else | |
| 1795 { | |
| 1796 /* Emit x+ as x(&|), where & means "self". */ | |
| 1797 next = regnode(BRANCH); /* Either */ | |
| 1798 regtail(ret, next); | |
| 233 | 1799 regtail(regnode(BACK), ret); /* loop back */ |
| 7 | 1800 regtail(next, regnode(BRANCH)); /* or */ |
| 1801 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1802 } | |
| 1803 *flagp = (WORST | HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1804 break; | |
| 1805 | |
| 1806 case Magic('@'): | |
| 1807 { | |
| 1808 int lop = END; | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1809 int nr; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1810 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1811 nr = getdecchrs(); |
| 7 | 1812 switch (no_Magic(getchr())) |
| 1813 { | |
| 1814 case '=': lop = MATCH; break; /* \@= */ | |
| 1815 case '!': lop = NOMATCH; break; /* \@! */ | |
| 1816 case '>': lop = SUBPAT; break; /* \@> */ | |
| 1817 case '<': switch (no_Magic(getchr())) | |
| 1818 { | |
| 1819 case '=': lop = BEHIND; break; /* \@<= */ | |
| 1820 case '!': lop = NOBEHIND; break; /* \@<! */ | |
| 1821 } | |
| 1822 } | |
| 1823 if (lop == END) | |
| 4444 | 1824 EMSG2_RET_NULL(_("E59: invalid character after %s@"), |
| 7 | 1825 reg_magic == MAGIC_ALL); |
| 1826 /* Look behind must match with behind_pos. */ | |
| 1827 if (lop == BEHIND || lop == NOBEHIND) | |
| 1828 { | |
| 1829 regtail(ret, regnode(BHPOS)); | |
| 1830 *flagp |= HASLOOKBH; | |
| 1831 } | |
| 1832 regtail(ret, regnode(END)); /* operand ends */ | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1833 if (lop == BEHIND || lop == NOBEHIND) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1834 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1835 if (nr < 0) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1836 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
|
1837 reginsert_nr(lop, nr, ret); |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1838 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1839 else |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1840 reginsert(lop, ret); |
| 7 | 1841 break; |
| 1842 } | |
| 1843 | |
| 1844 case Magic('?'): | |
| 1845 case Magic('='): | |
| 1846 /* Emit x= as (x|) */ | |
| 1847 reginsert(BRANCH, ret); /* Either x */ | |
| 1848 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1849 next = regnode(NOTHING); /* null. */ | |
| 1850 regtail(ret, next); | |
| 1851 regoptail(ret, next); | |
| 1852 break; | |
| 1853 | |
| 1854 case Magic('{'): | |
| 1855 if (!read_limits(&minval, &maxval)) | |
| 1856 return NULL; | |
| 1857 if (flags & SIMPLE) | |
| 1858 { | |
| 1859 reginsert(BRACE_SIMPLE, ret); | |
| 1860 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1861 } | |
| 1862 else | |
| 1863 { | |
| 1864 if (num_complex_braces >= 10) | |
| 4444 | 1865 EMSG2_RET_NULL(_("E60: Too many complex %s{...}s"), |
| 7 | 1866 reg_magic == MAGIC_ALL); |
| 1867 reginsert(BRACE_COMPLEX + num_complex_braces, ret); | |
| 1868 regoptail(ret, regnode(BACK)); | |
| 1869 regoptail(ret, ret); | |
| 1870 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1871 ++num_complex_braces; | |
| 1872 } | |
| 1873 if (minval > 0 && maxval > 0) | |
| 1874 *flagp = (HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1875 break; | |
| 1876 } | |
| 1877 if (re_multi_type(peekchr()) != NOT_MULTI) | |
| 1878 { | |
| 1879 /* Can't have a multi follow a multi. */ | |
| 1880 if (peekchr() == Magic('*')) | |
| 1881 sprintf((char *)IObuff, _("E61: Nested %s*"), | |
| 1882 reg_magic >= MAGIC_ON ? "" : "\\"); | |
| 1883 else | |
| 1884 sprintf((char *)IObuff, _("E62: Nested %s%c"), | |
| 1885 reg_magic == MAGIC_ALL ? "" : "\\", no_Magic(peekchr())); | |
| 1886 EMSG_RET_NULL(IObuff); | |
| 1887 } | |
| 1888 | |
| 1889 return ret; | |
| 1890 } | |
| 1891 | |
| 4444 | 1892 /* When making changes to classchars also change nfa_classcodes. */ |
| 1893 static char_u *classchars = (char_u *)".iIkKfFpPsSdDxXoOwWhHaAlLuU"; | |
| 1894 static int classcodes[] = { | |
| 1895 ANY, IDENT, SIDENT, KWORD, SKWORD, | |
| 1896 FNAME, SFNAME, PRINT, SPRINT, | |
| 1897 WHITE, NWHITE, DIGIT, NDIGIT, | |
| 1898 HEX, NHEX, OCTAL, NOCTAL, | |
| 1899 WORD, NWORD, HEAD, NHEAD, | |
| 1900 ALPHA, NALPHA, LOWER, NLOWER, | |
| 1901 UPPER, NUPPER | |
| 1902 }; | |
| 1903 | |
| 7 | 1904 /* |
| 4444 | 1905 * Parse the lowest level. |
| 7 | 1906 * |
| 1907 * Optimization: gobbles an entire sequence of ordinary characters so that | |
| 1908 * it can turn them into a single node, which is smaller to store and | |
| 1909 * faster to run. Don't do this when one_exactly is set. | |
| 1910 */ | |
| 1911 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
1912 regatom(int *flagp) |
| 7 | 1913 { |
| 1914 char_u *ret; | |
| 1915 int flags; | |
| 1916 int c; | |
| 1917 char_u *p; | |
| 1918 int extra = 0; | |
|
8021
b6b4f354df23
commit https://github.com/vim/vim/commit/7c29f387819b5817b003d2ba73e2b5cf3cb3d0dd
Christian Brabandt <cb@256bit.org>
parents:
7833
diff
changeset
|
1919 int save_prev_at_start = prev_at_start; |
| 7 | 1920 |
| 1921 *flagp = WORST; /* Tentatively. */ | |
| 1922 | |
| 1923 c = getchr(); | |
| 1924 switch (c) | |
| 1925 { | |
| 1926 case Magic('^'): | |
| 1927 ret = regnode(BOL); | |
| 1928 break; | |
| 1929 | |
| 1930 case Magic('$'): | |
| 1931 ret = regnode(EOL); | |
| 1932 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1933 had_eol = TRUE; | |
| 1934 #endif | |
| 1935 break; | |
| 1936 | |
| 1937 case Magic('<'): | |
| 1938 ret = regnode(BOW); | |
| 1939 break; | |
| 1940 | |
| 1941 case Magic('>'): | |
| 1942 ret = regnode(EOW); | |
| 1943 break; | |
| 1944 | |
| 1945 case Magic('_'): | |
| 1946 c = no_Magic(getchr()); | |
| 1947 if (c == '^') /* "\_^" is start-of-line */ | |
| 1948 { | |
| 1949 ret = regnode(BOL); | |
| 1950 break; | |
| 1951 } | |
| 1952 if (c == '$') /* "\_$" is end-of-line */ | |
| 1953 { | |
| 1954 ret = regnode(EOL); | |
| 1955 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1956 had_eol = TRUE; | |
| 1957 #endif | |
| 1958 break; | |
| 1959 } | |
| 1960 | |
| 1961 extra = ADD_NL; | |
| 1962 *flagp |= HASNL; | |
| 1963 | |
| 1964 /* "\_[" is character range plus newline */ | |
| 1965 if (c == '[') | |
| 1966 goto collection; | |
| 1967 | |
| 1968 /* "\_x" is character class plus newline */ | |
| 1969 /*FALLTHROUGH*/ | |
| 1970 | |
| 1971 /* | |
| 1972 * Character classes. | |
| 1973 */ | |
| 1974 case Magic('.'): | |
| 1975 case Magic('i'): | |
| 1976 case Magic('I'): | |
| 1977 case Magic('k'): | |
| 1978 case Magic('K'): | |
| 1979 case Magic('f'): | |
| 1980 case Magic('F'): | |
| 1981 case Magic('p'): | |
| 1982 case Magic('P'): | |
| 1983 case Magic('s'): | |
| 1984 case Magic('S'): | |
| 1985 case Magic('d'): | |
| 1986 case Magic('D'): | |
| 1987 case Magic('x'): | |
| 1988 case Magic('X'): | |
| 1989 case Magic('o'): | |
| 1990 case Magic('O'): | |
| 1991 case Magic('w'): | |
| 1992 case Magic('W'): | |
| 1993 case Magic('h'): | |
| 1994 case Magic('H'): | |
| 1995 case Magic('a'): | |
| 1996 case Magic('A'): | |
| 1997 case Magic('l'): | |
| 1998 case Magic('L'): | |
| 1999 case Magic('u'): | |
| 2000 case Magic('U'): | |
| 2001 p = vim_strchr(classchars, no_Magic(c)); | |
| 2002 if (p == NULL) | |
| 2003 EMSG_RET_NULL(_("E63: invalid use of \\_")); | |
| 714 | 2004 #ifdef FEAT_MBYTE |
| 2005 /* When '.' is followed by a composing char ignore the dot, so that | |
| 2006 * the composing char is matched here. */ | |
| 2007 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
| 2008 { | |
| 2009 c = getchr(); | |
| 2010 goto do_multibyte; | |
| 2011 } | |
| 2012 #endif | |
| 7 | 2013 ret = regnode(classcodes[p - classchars] + extra); |
| 2014 *flagp |= HASWIDTH | SIMPLE; | |
| 2015 break; | |
| 2016 | |
| 2017 case Magic('n'): | |
| 2018 if (reg_string) | |
| 2019 { | |
| 2020 /* In a string "\n" matches a newline character. */ | |
| 2021 ret = regnode(EXACTLY); | |
| 2022 regc(NL); | |
| 2023 regc(NUL); | |
| 2024 *flagp |= HASWIDTH | SIMPLE; | |
| 2025 } | |
| 2026 else | |
| 2027 { | |
| 2028 /* In buffer text "\n" matches the end of a line. */ | |
| 2029 ret = regnode(NEWL); | |
| 2030 *flagp |= HASWIDTH | HASNL; | |
| 2031 } | |
| 2032 break; | |
| 2033 | |
| 2034 case Magic('('): | |
| 2035 if (one_exactly) | |
| 2036 EMSG_ONE_RET_NULL; | |
| 2037 ret = reg(REG_PAREN, &flags); | |
| 2038 if (ret == NULL) | |
| 2039 return NULL; | |
| 2040 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 2041 break; | |
| 2042 | |
| 2043 case NUL: | |
| 2044 case Magic('|'): | |
| 2045 case Magic('&'): | |
| 2046 case Magic(')'): | |
| 1468 | 2047 if (one_exactly) |
| 2048 EMSG_ONE_RET_NULL; | |
| 7 | 2049 EMSG_RET_NULL(_(e_internal)); /* Supposed to be caught earlier. */ |
| 2050 /* NOTREACHED */ | |
| 2051 | |
| 2052 case Magic('='): | |
| 2053 case Magic('?'): | |
| 2054 case Magic('+'): | |
| 2055 case Magic('@'): | |
| 2056 case Magic('{'): | |
| 2057 case Magic('*'): | |
| 2058 c = no_Magic(c); | |
| 2059 sprintf((char *)IObuff, _("E64: %s%c follows nothing"), | |
| 2060 (c == '*' ? reg_magic >= MAGIC_ON : reg_magic == MAGIC_ALL) | |
| 2061 ? "" : "\\", c); | |
| 2062 EMSG_RET_NULL(IObuff); | |
| 2063 /* NOTREACHED */ | |
| 2064 | |
| 2065 case Magic('~'): /* previous substitute pattern */ | |
| 359 | 2066 if (reg_prev_sub != NULL) |
| 7 | 2067 { |
| 2068 char_u *lp; | |
| 2069 | |
| 2070 ret = regnode(EXACTLY); | |
| 2071 lp = reg_prev_sub; | |
| 2072 while (*lp != NUL) | |
| 2073 regc(*lp++); | |
| 2074 regc(NUL); | |
| 2075 if (*reg_prev_sub != NUL) | |
| 2076 { | |
| 2077 *flagp |= HASWIDTH; | |
| 2078 if ((lp - reg_prev_sub) == 1) | |
| 2079 *flagp |= SIMPLE; | |
| 2080 } | |
| 2081 } | |
| 2082 else | |
| 2083 EMSG_RET_NULL(_(e_nopresub)); | |
| 2084 break; | |
| 2085 | |
| 2086 case Magic('1'): | |
| 2087 case Magic('2'): | |
| 2088 case Magic('3'): | |
| 2089 case Magic('4'): | |
| 2090 case Magic('5'): | |
| 2091 case Magic('6'): | |
| 2092 case Magic('7'): | |
| 2093 case Magic('8'): | |
| 2094 case Magic('9'): | |
| 2095 { | |
| 2096 int refnum; | |
| 2097 | |
| 2098 refnum = c - Magic('0'); | |
| 2099 /* | |
| 2100 * Check if the back reference is legal. We must have seen the | |
| 2101 * close brace. | |
| 2102 * TODO: Should also check that we don't refer to something | |
| 2103 * that is repeated (+*=): what instance of the repetition | |
| 2104 * should we match? | |
| 2105 */ | |
| 2106 if (!had_endbrace[refnum]) | |
| 2107 { | |
| 2108 /* Trick: check if "@<=" or "@<!" follows, in which case | |
| 2109 * the \1 can appear before the referenced match. */ | |
| 2110 for (p = regparse; *p != NUL; ++p) | |
| 2111 if (p[0] == '@' && p[1] == '<' | |
| 2112 && (p[2] == '!' || p[2] == '=')) | |
| 2113 break; | |
| 2114 if (*p == NUL) | |
| 2115 EMSG_RET_NULL(_("E65: Illegal back reference")); | |
| 2116 } | |
| 2117 ret = regnode(BACKREF + refnum); | |
| 2118 } | |
| 2119 break; | |
| 2120 | |
| 2121 case Magic('z'): | |
| 2122 { | |
| 2123 c = no_Magic(getchr()); | |
| 2124 switch (c) | |
| 2125 { | |
| 741 | 2126 #ifdef FEAT_SYN_HL |
| 7 | 2127 case '(': if (reg_do_extmatch != REX_SET) |
|
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
2128 EMSG_RET_NULL(_(e_z_not_allowed)); |
| 7 | 2129 if (one_exactly) |
| 2130 EMSG_ONE_RET_NULL; | |
| 2131 ret = reg(REG_ZPAREN, &flags); | |
| 2132 if (ret == NULL) | |
| 2133 return NULL; | |
| 2134 *flagp |= flags & (HASWIDTH|SPSTART|HASNL|HASLOOKBH); | |
| 2135 re_has_z = REX_SET; | |
| 2136 break; | |
| 2137 | |
| 2138 case '1': | |
| 2139 case '2': | |
| 2140 case '3': | |
| 2141 case '4': | |
| 2142 case '5': | |
| 2143 case '6': | |
| 2144 case '7': | |
| 2145 case '8': | |
| 2146 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
|
2147 EMSG_RET_NULL(_(e_z1_not_allowed)); |
| 7 | 2148 ret = regnode(ZREF + c - '0'); |
| 2149 re_has_z = REX_USE; | |
| 2150 break; | |
| 741 | 2151 #endif |
| 7 | 2152 |
| 2153 case 's': ret = regnode(MOPEN + 0); | |
| 6203 | 2154 if (re_mult_next("\\zs") == FAIL) |
| 2155 return NULL; | |
| 7 | 2156 break; |
| 2157 | |
| 2158 case 'e': ret = regnode(MCLOSE + 0); | |
| 6203 | 2159 if (re_mult_next("\\ze") == FAIL) |
| 2160 return NULL; | |
| 7 | 2161 break; |
| 2162 | |
| 2163 default: EMSG_RET_NULL(_("E68: Invalid character after \\z")); | |
| 2164 } | |
| 2165 } | |
| 2166 break; | |
| 2167 | |
| 2168 case Magic('%'): | |
| 2169 { | |
| 2170 c = no_Magic(getchr()); | |
| 2171 switch (c) | |
| 2172 { | |
| 2173 /* () without a back reference */ | |
| 2174 case '(': | |
| 2175 if (one_exactly) | |
| 2176 EMSG_ONE_RET_NULL; | |
| 2177 ret = reg(REG_NPAREN, &flags); | |
| 2178 if (ret == NULL) | |
| 2179 return NULL; | |
| 2180 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 2181 break; | |
| 2182 | |
| 2183 /* Catch \%^ and \%$ regardless of where they appear in the | |
| 2184 * pattern -- regardless of whether or not it makes sense. */ | |
| 2185 case '^': | |
| 2186 ret = regnode(RE_BOF); | |
| 2187 break; | |
| 2188 | |
| 2189 case '$': | |
| 2190 ret = regnode(RE_EOF); | |
| 2191 break; | |
| 2192 | |
| 2193 case '#': | |
| 2194 ret = regnode(CURSOR); | |
| 2195 break; | |
| 2196 | |
| 639 | 2197 case 'V': |
| 2198 ret = regnode(RE_VISUAL); | |
| 2199 break; | |
| 2200 | |
| 5901 | 2201 case 'C': |
| 2202 ret = regnode(RE_COMPOSING); | |
| 2203 break; | |
| 2204 | |
| 7 | 2205 /* \%[abc]: Emit as a list of branches, all ending at the last |
| 2206 * branch which matches nothing. */ | |
| 2207 case '[': | |
| 2208 if (one_exactly) /* doesn't nest */ | |
| 2209 EMSG_ONE_RET_NULL; | |
| 2210 { | |
| 2211 char_u *lastbranch; | |
| 2212 char_u *lastnode = NULL; | |
| 2213 char_u *br; | |
| 2214 | |
| 2215 ret = NULL; | |
| 2216 while ((c = getchr()) != ']') | |
| 2217 { | |
| 2218 if (c == NUL) | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2219 EMSG2_RET_NULL(_(e_missing_sb), |
| 7 | 2220 reg_magic == MAGIC_ALL); |
| 2221 br = regnode(BRANCH); | |
| 2222 if (ret == NULL) | |
| 2223 ret = br; | |
| 2224 else | |
| 2225 regtail(lastnode, br); | |
| 2226 | |
| 2227 ungetchr(); | |
| 2228 one_exactly = TRUE; | |
| 2229 lastnode = regatom(flagp); | |
| 2230 one_exactly = FALSE; | |
| 2231 if (lastnode == NULL) | |
| 2232 return NULL; | |
| 2233 } | |
| 2234 if (ret == NULL) | |
|
4760
532a9855bd30
updated for version 7.3.1127
Bram Moolenaar <bram@vim.org>
parents:
4746
diff
changeset
|
2235 EMSG2_RET_NULL(_(e_empty_sb), |
| 7 | 2236 reg_magic == MAGIC_ALL); |
| 2237 lastbranch = regnode(BRANCH); | |
| 2238 br = regnode(NOTHING); | |
| 2239 if (ret != JUST_CALC_SIZE) | |
| 2240 { | |
| 2241 regtail(lastnode, br); | |
| 2242 regtail(lastbranch, br); | |
| 2243 /* connect all branches to the NOTHING | |
| 2244 * branch at the end */ | |
| 2245 for (br = ret; br != lastnode; ) | |
| 2246 { | |
| 2247 if (OP(br) == BRANCH) | |
| 2248 { | |
| 2249 regtail(br, lastbranch); | |
| 2250 br = OPERAND(br); | |
| 2251 } | |
| 2252 else | |
| 2253 br = regnext(br); | |
| 2254 } | |
| 2255 } | |
| 1701 | 2256 *flagp &= ~(HASWIDTH | SIMPLE); |
| 7 | 2257 break; |
| 2258 } | |
| 2259 | |
| 24 | 2260 case 'd': /* %d123 decimal */ |
| 2261 case 'o': /* %o123 octal */ | |
| 2262 case 'x': /* %xab hex 2 */ | |
| 2263 case 'u': /* %uabcd hex 4 */ | |
| 2264 case 'U': /* %U1234abcd hex 8 */ | |
| 2265 { | |
| 2266 int i; | |
| 2267 | |
| 2268 switch (c) | |
| 2269 { | |
| 2270 case 'd': i = getdecchrs(); break; | |
| 2271 case 'o': i = getoctchrs(); break; | |
| 2272 case 'x': i = gethexchrs(2); break; | |
| 2273 case 'u': i = gethexchrs(4); break; | |
| 2274 case 'U': i = gethexchrs(8); break; | |
| 2275 default: i = -1; break; | |
| 2276 } | |
| 2277 | |
| 2278 if (i < 0) | |
| 4444 | 2279 EMSG2_RET_NULL( |
| 24 | 2280 _("E678: Invalid character after %s%%[dxouU]"), |
| 2281 reg_magic == MAGIC_ALL); | |
| 714 | 2282 #ifdef FEAT_MBYTE |
| 2283 if (use_multibytecode(i)) | |
| 2284 ret = regnode(MULTIBYTECODE); | |
| 2285 else | |
| 2286 #endif | |
| 2287 ret = regnode(EXACTLY); | |
| 24 | 2288 if (i == 0) |
| 2289 regc(0x0a); | |
| 2290 else | |
| 2291 #ifdef FEAT_MBYTE | |
| 2292 regmbc(i); | |
| 2293 #else | |
| 2294 regc(i); | |
| 2295 #endif | |
| 2296 regc(NUL); | |
| 2297 *flagp |= HASWIDTH; | |
| 2298 break; | |
| 2299 } | |
| 2300 | |
| 7 | 2301 default: |
| 639 | 2302 if (VIM_ISDIGIT(c) || c == '<' || c == '>' |
| 2303 || c == '\'') | |
| 7 | 2304 { |
| 2305 long_u n = 0; | |
| 2306 int cmp; | |
| 2307 | |
| 2308 cmp = c; | |
| 2309 if (cmp == '<' || cmp == '>') | |
| 2310 c = getchr(); | |
| 2311 while (VIM_ISDIGIT(c)) | |
| 2312 { | |
| 2313 n = n * 10 + (c - '0'); | |
| 2314 c = getchr(); | |
| 2315 } | |
| 639 | 2316 if (c == '\'' && n == 0) |
| 2317 { | |
| 2318 /* "\%'m", "\%<'m" and "\%>'m": Mark */ | |
| 2319 c = getchr(); | |
| 2320 ret = regnode(RE_MARK); | |
| 2321 if (ret == JUST_CALC_SIZE) | |
| 2322 regsize += 2; | |
| 2323 else | |
| 2324 { | |
| 2325 *regcode++ = c; | |
| 2326 *regcode++ = cmp; | |
| 2327 } | |
| 2328 break; | |
| 2329 } | |
| 2330 else if (c == 'l' || c == 'c' || c == 'v') | |
| 7 | 2331 { |
| 2332 if (c == 'l') | |
|
8021
b6b4f354df23
commit https://github.com/vim/vim/commit/7c29f387819b5817b003d2ba73e2b5cf3cb3d0dd
Christian Brabandt <cb@256bit.org>
parents:
7833
diff
changeset
|
2333 { |
| 7 | 2334 ret = regnode(RE_LNUM); |
|
8021
b6b4f354df23
commit https://github.com/vim/vim/commit/7c29f387819b5817b003d2ba73e2b5cf3cb3d0dd
Christian Brabandt <cb@256bit.org>
parents:
7833
diff
changeset
|
2335 if (save_prev_at_start) |
|
b6b4f354df23
commit https://github.com/vim/vim/commit/7c29f387819b5817b003d2ba73e2b5cf3cb3d0dd
Christian Brabandt <cb@256bit.org>
parents:
7833
diff
changeset
|
2336 at_start = TRUE; |
|
b6b4f354df23
commit https://github.com/vim/vim/commit/7c29f387819b5817b003d2ba73e2b5cf3cb3d0dd
Christian Brabandt <cb@256bit.org>
parents:
7833
diff
changeset
|
2337 } |
| 7 | 2338 else if (c == 'c') |
| 2339 ret = regnode(RE_COL); | |
| 2340 else | |
| 2341 ret = regnode(RE_VCOL); | |
| 2342 if (ret == JUST_CALC_SIZE) | |
| 2343 regsize += 5; | |
| 2344 else | |
| 2345 { | |
| 2346 /* put the number and the optional | |
| 2347 * comparator after the opcode */ | |
| 2348 regcode = re_put_long(regcode, n); | |
| 2349 *regcode++ = cmp; | |
| 2350 } | |
| 2351 break; | |
| 2352 } | |
| 2353 } | |
| 2354 | |
| 4444 | 2355 EMSG2_RET_NULL(_("E71: Invalid character after %s%%"), |
| 7 | 2356 reg_magic == MAGIC_ALL); |
| 2357 } | |
| 2358 } | |
| 2359 break; | |
| 2360 | |
| 2361 case Magic('['): | |
| 2362 collection: | |
| 2363 { | |
| 2364 char_u *lp; | |
| 2365 | |
| 2366 /* | |
| 2367 * If there is no matching ']', we assume the '[' is a normal | |
| 2368 * character. This makes 'incsearch' and ":help [" work. | |
| 2369 */ | |
| 2370 lp = skip_anyof(regparse); | |
| 2371 if (*lp == ']') /* there is a matching ']' */ | |
| 2372 { | |
| 2373 int startc = -1; /* > 0 when next '-' is a range */ | |
| 2374 int endc; | |
| 2375 | |
| 2376 /* | |
| 2377 * In a character class, different parsing rules apply. | |
| 2378 * Not even \ is special anymore, nothing is. | |
| 2379 */ | |
| 2380 if (*regparse == '^') /* Complement of range. */ | |
| 2381 { | |
| 2382 ret = regnode(ANYBUT + extra); | |
| 2383 regparse++; | |
| 2384 } | |
| 2385 else | |
| 2386 ret = regnode(ANYOF + extra); | |
| 2387 | |
| 2388 /* At the start ']' and '-' mean the literal character. */ | |
| 2389 if (*regparse == ']' || *regparse == '-') | |
| 167 | 2390 { |
| 2391 startc = *regparse; | |
| 7 | 2392 regc(*regparse++); |
| 167 | 2393 } |
| 7 | 2394 |
| 2395 while (*regparse != NUL && *regparse != ']') | |
| 2396 { | |
| 2397 if (*regparse == '-') | |
| 2398 { | |
| 2399 ++regparse; | |
| 2400 /* The '-' is not used for a range at the end and | |
| 2401 * after or before a '\n'. */ | |
| 2402 if (*regparse == ']' || *regparse == NUL | |
| 2403 || startc == -1 | |
| 2404 || (regparse[0] == '\\' && regparse[1] == 'n')) | |
| 2405 { | |
| 2406 regc('-'); | |
| 2407 startc = '-'; /* [--x] is a range */ | |
| 2408 } | |
| 2409 else | |
| 2410 { | |
| 167 | 2411 /* Also accept "a-[.z.]" */ |
| 2412 endc = 0; | |
| 2413 if (*regparse == '[') | |
| 2414 endc = get_coll_element(®parse); | |
| 2415 if (endc == 0) | |
| 2416 { | |
| 7 | 2417 #ifdef FEAT_MBYTE |
| 167 | 2418 if (has_mbyte) |
| 2419 endc = mb_ptr2char_adv(®parse); | |
| 2420 else | |
| 7 | 2421 #endif |
| 167 | 2422 endc = *regparse++; |
| 2423 } | |
| 24 | 2424 |
| 2425 /* Handle \o40, \x20 and \u20AC style sequences */ | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2426 if (endc == '\\' && !reg_cpo_lit && !reg_cpo_bsl) |
| 24 | 2427 endc = coll_get_char(); |
| 2428 | |
| 7 | 2429 if (startc > endc) |
| 2430 EMSG_RET_NULL(_(e_invrange)); | |
| 2431 #ifdef FEAT_MBYTE | |
| 2432 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
| 2433 || (*mb_char2len)(endc) > 1)) | |
| 2434 { | |
| 2435 /* Limit to a range of 256 chars */ | |
| 2436 if (endc > startc + 256) | |
| 2437 EMSG_RET_NULL(_(e_invrange)); | |
| 2438 while (++startc <= endc) | |
| 2439 regmbc(startc); | |
| 2440 } | |
| 2441 else | |
| 2442 #endif | |
| 2443 { | |
| 2444 #ifdef EBCDIC | |
| 2445 int alpha_only = FALSE; | |
| 2446 | |
| 2447 /* for alphabetical range skip the gaps | |
| 2448 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
| 2449 if (isalpha(startc) && isalpha(endc)) | |
| 2450 alpha_only = TRUE; | |
| 2451 #endif | |
| 2452 while (++startc <= endc) | |
| 2453 #ifdef EBCDIC | |
| 2454 if (!alpha_only || isalpha(startc)) | |
| 2455 #endif | |
| 2456 regc(startc); | |
| 2457 } | |
| 2458 startc = -1; | |
| 2459 } | |
| 2460 } | |
| 2461 /* | |
| 2462 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
| 2463 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
| 2464 * 'cpoptions' is not included. | |
| 167 | 2465 * Posix doesn't recognize backslash at all. |
| 7 | 2466 */ |
| 2467 else if (*regparse == '\\' | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2468 && !reg_cpo_bsl |
| 7 | 2469 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2470 || (!reg_cpo_lit |
| 7 | 2471 && vim_strchr(REGEXP_ABBR, |
| 2472 regparse[1]) != NULL))) | |
| 2473 { | |
| 2474 regparse++; | |
| 2475 if (*regparse == 'n') | |
| 2476 { | |
| 2477 /* '\n' in range: also match NL */ | |
| 2478 if (ret != JUST_CALC_SIZE) | |
| 2479 { | |
| 4084 | 2480 /* Using \n inside [^] does not change what |
| 2481 * matches. "[^\n]" is the same as ".". */ | |
| 2482 if (*ret == ANYOF) | |
| 2483 { | |
| 7 | 2484 *ret = ANYOF + ADD_NL; |
| 4084 | 2485 *flagp |= HASNL; |
| 2486 } | |
| 7 | 2487 /* else: must have had a \n already */ |
| 2488 } | |
| 2489 regparse++; | |
| 2490 startc = -1; | |
| 2491 } | |
| 24 | 2492 else if (*regparse == 'd' |
| 2493 || *regparse == 'o' | |
| 2494 || *regparse == 'x' | |
| 2495 || *regparse == 'u' | |
| 2496 || *regparse == 'U') | |
| 2497 { | |
| 2498 startc = coll_get_char(); | |
| 2499 if (startc == 0) | |
| 2500 regc(0x0a); | |
| 2501 else | |
| 2502 #ifdef FEAT_MBYTE | |
| 2503 regmbc(startc); | |
| 2504 #else | |
| 2505 regc(startc); | |
| 2506 #endif | |
| 2507 } | |
| 7 | 2508 else |
| 2509 { | |
| 2510 startc = backslash_trans(*regparse++); | |
| 2511 regc(startc); | |
| 2512 } | |
| 2513 } | |
| 2514 else if (*regparse == '[') | |
| 2515 { | |
| 2516 int c_class; | |
| 2517 int cu; | |
| 2518 | |
| 167 | 2519 c_class = get_char_class(®parse); |
| 7 | 2520 startc = -1; |
| 2521 /* Characters assumed to be 8 bits! */ | |
| 2522 switch (c_class) | |
| 2523 { | |
| 2524 case CLASS_NONE: | |
| 167 | 2525 c_class = get_equi_class(®parse); |
| 2526 if (c_class != 0) | |
| 2527 { | |
| 2528 /* produce equivalence class */ | |
| 2529 reg_equi_class(c_class); | |
| 2530 } | |
| 2531 else if ((c_class = | |
| 2532 get_coll_element(®parse)) != 0) | |
| 2533 { | |
| 2534 /* produce a collating element */ | |
| 2535 regmbc(c_class); | |
| 2536 } | |
| 2537 else | |
| 2538 { | |
| 2539 /* literal '[', allow [[-x] as a range */ | |
| 2540 startc = *regparse++; | |
| 2541 regc(startc); | |
| 2542 } | |
| 7 | 2543 break; |
| 2544 case CLASS_ALNUM: | |
| 2545 for (cu = 1; cu <= 255; cu++) | |
| 2546 if (isalnum(cu)) | |
| 2547 regc(cu); | |
| 2548 break; | |
| 2549 case CLASS_ALPHA: | |
| 2550 for (cu = 1; cu <= 255; cu++) | |
| 2551 if (isalpha(cu)) | |
| 2552 regc(cu); | |
| 2553 break; | |
| 2554 case CLASS_BLANK: | |
| 2555 regc(' '); | |
| 2556 regc('\t'); | |
| 2557 break; | |
| 2558 case CLASS_CNTRL: | |
| 2559 for (cu = 1; cu <= 255; cu++) | |
| 2560 if (iscntrl(cu)) | |
| 2561 regc(cu); | |
| 2562 break; | |
| 2563 case CLASS_DIGIT: | |
| 2564 for (cu = 1; cu <= 255; cu++) | |
| 2565 if (VIM_ISDIGIT(cu)) | |
| 2566 regc(cu); | |
| 2567 break; | |
| 2568 case CLASS_GRAPH: | |
| 2569 for (cu = 1; cu <= 255; cu++) | |
| 2570 if (isgraph(cu)) | |
| 2571 regc(cu); | |
| 2572 break; | |
| 2573 case CLASS_LOWER: | |
| 2574 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2575 if (MB_ISLOWER(cu)) |
| 7 | 2576 regc(cu); |
| 2577 break; | |
| 2578 case CLASS_PRINT: | |
| 2579 for (cu = 1; cu <= 255; cu++) | |
| 2580 if (vim_isprintc(cu)) | |
| 2581 regc(cu); | |
| 2582 break; | |
| 2583 case CLASS_PUNCT: | |
| 2584 for (cu = 1; cu <= 255; cu++) | |
| 2585 if (ispunct(cu)) | |
| 2586 regc(cu); | |
| 2587 break; | |
| 2588 case CLASS_SPACE: | |
| 2589 for (cu = 9; cu <= 13; cu++) | |
| 2590 regc(cu); | |
| 2591 regc(' '); | |
| 2592 break; | |
| 2593 case CLASS_UPPER: | |
| 2594 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2595 if (MB_ISUPPER(cu)) |
| 7 | 2596 regc(cu); |
| 2597 break; | |
| 2598 case CLASS_XDIGIT: | |
| 2599 for (cu = 1; cu <= 255; cu++) | |
| 2600 if (vim_isxdigit(cu)) | |
| 2601 regc(cu); | |
| 2602 break; | |
| 2603 case CLASS_TAB: | |
| 2604 regc('\t'); | |
| 2605 break; | |
| 2606 case CLASS_RETURN: | |
| 2607 regc('\r'); | |
| 2608 break; | |
| 2609 case CLASS_BACKSPACE: | |
| 2610 regc('\b'); | |
| 2611 break; | |
| 2612 case CLASS_ESCAPE: | |
| 2613 regc('\033'); | |
| 2614 break; | |
| 2615 } | |
| 2616 } | |
| 2617 else | |
| 2618 { | |
| 2619 #ifdef FEAT_MBYTE | |
| 2620 if (has_mbyte) | |
| 2621 { | |
| 2622 int len; | |
| 2623 | |
| 2624 /* produce a multibyte character, including any | |
| 2625 * following composing characters */ | |
| 2626 startc = mb_ptr2char(regparse); | |
| 474 | 2627 len = (*mb_ptr2len)(regparse); |
| 7 | 2628 if (enc_utf8 && utf_char2len(startc) != len) |
| 2629 startc = -1; /* composing chars */ | |
| 2630 while (--len >= 0) | |
| 2631 regc(*regparse++); | |
| 2632 } | |
| 2633 else | |
| 2634 #endif | |
| 2635 { | |
| 2636 startc = *regparse++; | |
| 2637 regc(startc); | |
| 2638 } | |
| 2639 } | |
| 2640 } | |
| 2641 regc(NUL); | |
| 2642 prevchr_len = 1; /* last char was the ']' */ | |
| 2643 if (*regparse != ']') | |
| 2644 EMSG_RET_NULL(_(e_toomsbra)); /* Cannot happen? */ | |
| 2645 skipchr(); /* let's be friends with the lexer again */ | |
| 2646 *flagp |= HASWIDTH | SIMPLE; | |
| 2647 break; | |
| 2648 } | |
| 481 | 2649 else if (reg_strict) |
| 4444 | 2650 EMSG2_RET_NULL(_(e_missingbracket), reg_magic > MAGIC_OFF); |
| 7 | 2651 } |
| 2652 /* FALLTHROUGH */ | |
| 2653 | |
| 2654 default: | |
| 2655 { | |
| 2656 int len; | |
| 2657 | |
| 2658 #ifdef FEAT_MBYTE | |
| 2659 /* A multi-byte character is handled as a separate atom if it's | |
| 714 | 2660 * before a multi and when it's a composing char. */ |
| 2661 if (use_multibytecode(c)) | |
| 7 | 2662 { |
| 714 | 2663 do_multibyte: |
| 7 | 2664 ret = regnode(MULTIBYTECODE); |
| 2665 regmbc(c); | |
| 2666 *flagp |= HASWIDTH | SIMPLE; | |
| 2667 break; | |
| 2668 } | |
| 2669 #endif | |
| 2670 | |
| 2671 ret = regnode(EXACTLY); | |
| 2672 | |
| 2673 /* | |
| 2674 * Append characters as long as: | |
| 2675 * - there is no following multi, we then need the character in | |
| 2676 * front of it as a single character operand | |
| 2677 * - not running into a Magic character | |
| 2678 * - "one_exactly" is not set | |
| 2679 * But always emit at least one character. Might be a Multi, | |
| 2680 * e.g., a "[" without matching "]". | |
| 2681 */ | |
| 2682 for (len = 0; c != NUL && (len == 0 | |
| 2683 || (re_multi_type(peekchr()) == NOT_MULTI | |
| 2684 && !one_exactly | |
| 2685 && !is_Magic(c))); ++len) | |
| 2686 { | |
| 2687 c = no_Magic(c); | |
| 2688 #ifdef FEAT_MBYTE | |
| 2689 if (has_mbyte) | |
| 2690 { | |
| 2691 regmbc(c); | |
| 2692 if (enc_utf8) | |
| 2693 { | |
| 2694 int l; | |
| 2695 | |
| 714 | 2696 /* Need to get composing character too. */ |
| 7 | 2697 for (;;) |
| 2698 { | |
| 714 | 2699 l = utf_ptr2len(regparse); |
| 2700 if (!UTF_COMPOSINGLIKE(regparse, regparse + l)) | |
| 7 | 2701 break; |
| 714 | 2702 regmbc(utf_ptr2char(regparse)); |
| 2703 skipchr(); | |
| 7 | 2704 } |
| 2705 } | |
| 2706 } | |
| 2707 else | |
| 2708 #endif | |
| 2709 regc(c); | |
| 2710 c = getchr(); | |
| 2711 } | |
| 2712 ungetchr(); | |
| 2713 | |
| 2714 regc(NUL); | |
| 2715 *flagp |= HASWIDTH; | |
| 2716 if (len == 1) | |
| 2717 *flagp |= SIMPLE; | |
| 2718 } | |
| 2719 break; | |
| 2720 } | |
| 2721 | |
| 2722 return ret; | |
| 2723 } | |
| 2724 | |
| 714 | 2725 #ifdef FEAT_MBYTE |
| 2726 /* | |
| 2727 * Return TRUE if MULTIBYTECODE should be used instead of EXACTLY for | |
| 2728 * character "c". | |
| 2729 */ | |
| 2730 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2731 use_multibytecode(int c) |
| 714 | 2732 { |
| 2733 return has_mbyte && (*mb_char2len)(c) > 1 | |
| 2734 && (re_multi_type(peekchr()) != NOT_MULTI | |
| 2735 || (enc_utf8 && utf_iscomposing(c))); | |
| 2736 } | |
| 2737 #endif | |
| 2738 | |
| 7 | 2739 /* |
| 4444 | 2740 * Emit a node. |
| 7 | 2741 * Return pointer to generated code. |
| 2742 */ | |
| 2743 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2744 regnode(int op) |
| 7 | 2745 { |
| 2746 char_u *ret; | |
| 2747 | |
| 2748 ret = regcode; | |
| 2749 if (ret == JUST_CALC_SIZE) | |
| 2750 regsize += 3; | |
| 2751 else | |
| 2752 { | |
| 2753 *regcode++ = op; | |
| 2754 *regcode++ = NUL; /* Null "next" pointer. */ | |
| 2755 *regcode++ = NUL; | |
| 2756 } | |
| 2757 return ret; | |
| 2758 } | |
| 2759 | |
| 2760 /* | |
| 2761 * Emit (if appropriate) a byte of code | |
| 2762 */ | |
| 2763 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2764 regc(int b) |
| 7 | 2765 { |
| 2766 if (regcode == JUST_CALC_SIZE) | |
| 2767 regsize++; | |
| 2768 else | |
| 2769 *regcode++ = b; | |
| 2770 } | |
| 2771 | |
| 2772 #ifdef FEAT_MBYTE | |
| 2773 /* | |
| 2774 * Emit (if appropriate) a multi-byte character of code | |
| 2775 */ | |
| 2776 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2777 regmbc(int c) |
| 7 | 2778 { |
| 2974 | 2779 if (!has_mbyte && c > 0xff) |
| 2780 return; | |
| 7 | 2781 if (regcode == JUST_CALC_SIZE) |
| 2782 regsize += (*mb_char2len)(c); | |
| 2783 else | |
| 2784 regcode += (*mb_char2bytes)(c, regcode); | |
| 2785 } | |
| 2786 #endif | |
| 2787 | |
| 2788 /* | |
| 4444 | 2789 * Insert an operator in front of already-emitted operand |
| 7 | 2790 * |
| 2791 * Means relocating the operand. | |
| 2792 */ | |
| 2793 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2794 reginsert(int op, char_u *opnd) |
| 7 | 2795 { |
| 2796 char_u *src; | |
| 2797 char_u *dst; | |
| 2798 char_u *place; | |
| 2799 | |
| 2800 if (regcode == JUST_CALC_SIZE) | |
| 2801 { | |
| 2802 regsize += 3; | |
| 2803 return; | |
| 2804 } | |
| 2805 src = regcode; | |
| 2806 regcode += 3; | |
| 2807 dst = regcode; | |
| 2808 while (src > opnd) | |
| 2809 *--dst = *--src; | |
| 2810 | |
| 2811 place = opnd; /* Op node, where operand used to be. */ | |
| 2812 *place++ = op; | |
| 2813 *place++ = NUL; | |
| 2814 *place = NUL; | |
| 2815 } | |
| 2816 | |
| 2817 /* | |
| 4444 | 2818 * 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
|
2819 * Add a number to the operator. |
|
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 static void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2822 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
|
2823 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2824 char_u *src; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2825 char_u *dst; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2826 char_u *place; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2827 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2828 if (regcode == JUST_CALC_SIZE) |
|
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 regsize += 7; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2831 return; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2832 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2833 src = regcode; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2834 regcode += 7; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2835 dst = regcode; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2836 while (src > opnd) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2837 *--dst = *--src; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2838 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2839 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
|
2840 *place++ = op; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2841 *place++ = NUL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2842 *place++ = NUL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2843 place = re_put_long(place, (long_u)val); |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2844 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2845 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2846 /* |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2847 * Insert an operator in front of already-emitted operand. |
| 7 | 2848 * The operator has the given limit values as operands. Also set next pointer. |
| 2849 * | |
| 2850 * Means relocating the operand. | |
| 2851 */ | |
| 2852 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2853 reginsert_limits( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2854 int op, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2855 long minval, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2856 long maxval, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2857 char_u *opnd) |
| 7 | 2858 { |
| 2859 char_u *src; | |
| 2860 char_u *dst; | |
| 2861 char_u *place; | |
| 2862 | |
| 2863 if (regcode == JUST_CALC_SIZE) | |
| 2864 { | |
| 2865 regsize += 11; | |
| 2866 return; | |
| 2867 } | |
| 2868 src = regcode; | |
| 2869 regcode += 11; | |
| 2870 dst = regcode; | |
| 2871 while (src > opnd) | |
| 2872 *--dst = *--src; | |
| 2873 | |
| 2874 place = opnd; /* Op node, where operand used to be. */ | |
| 2875 *place++ = op; | |
| 2876 *place++ = NUL; | |
| 2877 *place++ = NUL; | |
| 2878 place = re_put_long(place, (long_u)minval); | |
| 2879 place = re_put_long(place, (long_u)maxval); | |
| 2880 regtail(opnd, place); | |
| 2881 } | |
| 2882 | |
| 2883 /* | |
| 2884 * Write a long as four bytes at "p" and return pointer to the next char. | |
| 2885 */ | |
| 2886 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2887 re_put_long(char_u *p, long_u val) |
| 7 | 2888 { |
| 2889 *p++ = (char_u) ((val >> 24) & 0377); | |
| 2890 *p++ = (char_u) ((val >> 16) & 0377); | |
| 2891 *p++ = (char_u) ((val >> 8) & 0377); | |
| 2892 *p++ = (char_u) (val & 0377); | |
| 2893 return p; | |
| 2894 } | |
| 2895 | |
| 2896 /* | |
| 4444 | 2897 * Set the next-pointer at the end of a node chain. |
| 7 | 2898 */ |
| 2899 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2900 regtail(char_u *p, char_u *val) |
| 7 | 2901 { |
| 2902 char_u *scan; | |
| 2903 char_u *temp; | |
| 2904 int offset; | |
| 2905 | |
| 2906 if (p == JUST_CALC_SIZE) | |
| 2907 return; | |
| 2908 | |
| 2909 /* Find last node. */ | |
| 2910 scan = p; | |
| 2911 for (;;) | |
| 2912 { | |
| 2913 temp = regnext(scan); | |
| 2914 if (temp == NULL) | |
| 2915 break; | |
| 2916 scan = temp; | |
| 2917 } | |
| 2918 | |
| 233 | 2919 if (OP(scan) == BACK) |
| 7 | 2920 offset = (int)(scan - val); |
| 2921 else | |
| 2922 offset = (int)(val - scan); | |
| 2010 | 2923 /* When the offset uses more than 16 bits it can no longer fit in the two |
| 2974 | 2924 * bytes available. Use a global flag to avoid having to check return |
| 2010 | 2925 * values in too many places. */ |
| 2926 if (offset > 0xffff) | |
| 2927 reg_toolong = TRUE; | |
| 2928 else | |
| 2929 { | |
| 2930 *(scan + 1) = (char_u) (((unsigned)offset >> 8) & 0377); | |
| 2931 *(scan + 2) = (char_u) (offset & 0377); | |
| 2932 } | |
| 7 | 2933 } |
| 2934 | |
| 2935 /* | |
| 4444 | 2936 * Like regtail, on item after a BRANCH; nop if none. |
| 7 | 2937 */ |
| 2938 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2939 regoptail(char_u *p, char_u *val) |
| 7 | 2940 { |
| 2941 /* When op is neither BRANCH nor BRACE_COMPLEX0-9, it is "operandless" */ | |
| 2942 if (p == NULL || p == JUST_CALC_SIZE | |
| 2943 || (OP(p) != BRANCH | |
| 2944 && (OP(p) < BRACE_COMPLEX || OP(p) > BRACE_COMPLEX + 9))) | |
| 2945 return; | |
| 2946 regtail(OPERAND(p), val); | |
| 2947 } | |
| 2948 | |
| 2949 /* | |
| 4444 | 2950 * Functions for getting characters from the regexp input. |
| 7 | 2951 */ |
| 4444 | 2952 /* |
| 2953 * Start parsing at "str". | |
| 2954 */ | |
| 7 | 2955 static void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2956 initchr(char_u *str) |
| 7 | 2957 { |
| 2958 regparse = str; | |
| 2959 prevchr_len = 0; | |
| 2960 curchr = prevprevchr = prevchr = nextchr = -1; | |
| 2961 at_start = TRUE; | |
| 2962 prev_at_start = FALSE; | |
| 2963 } | |
| 2964 | |
| 4444 | 2965 /* |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2966 * 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
|
2967 * starts in the same state again. |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2968 */ |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2969 static void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2970 save_parse_state(parse_state_T *ps) |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2971 { |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2972 ps->regparse = regparse; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2973 ps->prevchr_len = prevchr_len; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2974 ps->curchr = curchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2975 ps->prevchr = prevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2976 ps->prevprevchr = prevprevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2977 ps->nextchr = nextchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2978 ps->at_start = at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2979 ps->prev_at_start = prev_at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2980 ps->regnpar = regnpar; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2981 } |
|
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 * Restore a previously saved parse state. |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2985 */ |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2986 static void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
2987 restore_parse_state(parse_state_T *ps) |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2988 { |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2989 regparse = ps->regparse; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2990 prevchr_len = ps->prevchr_len; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2991 curchr = ps->curchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2992 prevchr = ps->prevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2993 prevprevchr = ps->prevprevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2994 nextchr = ps->nextchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2995 at_start = ps->at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2996 prev_at_start = ps->prev_at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2997 regnpar = ps->regnpar; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2998 } |
|
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 /* |
| 4444 | 3002 * Get the next character without advancing. |
| 3003 */ | |
| 7 | 3004 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3005 peekchr(void) |
| 7 | 3006 { |
| 167 | 3007 static int after_slash = FALSE; |
| 3008 | |
| 7 | 3009 if (curchr == -1) |
| 3010 { | |
| 3011 switch (curchr = regparse[0]) | |
| 3012 { | |
| 3013 case '.': | |
| 3014 case '[': | |
| 3015 case '~': | |
| 3016 /* magic when 'magic' is on */ | |
| 3017 if (reg_magic >= MAGIC_ON) | |
| 3018 curchr = Magic(curchr); | |
| 3019 break; | |
| 3020 case '(': | |
| 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 '#': /* future ext. */ | |
| 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 '/': /* Can't be used in / command */ | |
| 3042 /* magic only after "\v" */ | |
| 3043 if (reg_magic == MAGIC_ALL) | |
| 3044 curchr = Magic(curchr); | |
| 3045 break; | |
| 3046 case '*': | |
| 167 | 3047 /* * is not magic as the very first character, eg "?*ptr", when |
| 3048 * after '^', eg "/^*ptr" and when after "\(", "\|", "\&". But | |
| 3049 * "\(\*" is not magic, thus must be magic if "after_slash" */ | |
| 3050 if (reg_magic >= MAGIC_ON | |
| 3051 && !at_start | |
| 3052 && !(prev_at_start && prevchr == Magic('^')) | |
| 3053 && (after_slash | |
| 3054 || (prevchr != Magic('(') | |
| 3055 && prevchr != Magic('&') | |
| 3056 && prevchr != Magic('|')))) | |
| 7 | 3057 curchr = Magic('*'); |
| 3058 break; | |
| 3059 case '^': | |
| 3060 /* '^' is only magic as the very first character and if it's after | |
| 3061 * "\(", "\|", "\&' or "\n" */ | |
| 3062 if (reg_magic >= MAGIC_OFF | |
| 3063 && (at_start | |
| 3064 || reg_magic == MAGIC_ALL | |
| 3065 || prevchr == Magic('(') | |
| 3066 || prevchr == Magic('|') | |
| 3067 || prevchr == Magic('&') | |
| 3068 || prevchr == Magic('n') | |
| 3069 || (no_Magic(prevchr) == '(' | |
| 3070 && prevprevchr == Magic('%')))) | |
| 3071 { | |
| 3072 curchr = Magic('^'); | |
| 3073 at_start = TRUE; | |
| 3074 prev_at_start = FALSE; | |
| 3075 } | |
| 3076 break; | |
| 3077 case '$': | |
| 3078 /* '$' is only magic as the very last char and if it's in front of | |
| 3079 * either "\|", "\)", "\&", or "\n" */ | |
| 3080 if (reg_magic >= MAGIC_OFF) | |
| 3081 { | |
| 3082 char_u *p = regparse + 1; | |
| 6041 | 3083 int is_magic_all = (reg_magic == MAGIC_ALL); |
| 3084 | |
| 3085 /* ignore \c \C \m \M \v \V and \Z after '$' */ | |
| 7 | 3086 while (p[0] == '\\' && (p[1] == 'c' || p[1] == 'C' |
| 6041 | 3087 || p[1] == 'm' || p[1] == 'M' |
| 3088 || p[1] == 'v' || p[1] == 'V' || p[1] == 'Z')) | |
| 3089 { | |
| 3090 if (p[1] == 'v') | |
| 3091 is_magic_all = TRUE; | |
| 3092 else if (p[1] == 'm' || p[1] == 'M' || p[1] == 'V') | |
| 3093 is_magic_all = FALSE; | |
| 7 | 3094 p += 2; |
| 6041 | 3095 } |
| 7 | 3096 if (p[0] == NUL |
| 3097 || (p[0] == '\\' | |
| 3098 && (p[1] == '|' || p[1] == '&' || p[1] == ')' | |
| 3099 || p[1] == 'n')) | |
| 6041 | 3100 || (is_magic_all |
| 3101 && (p[0] == '|' || p[0] == '&' || p[0] == ')')) | |
| 7 | 3102 || reg_magic == MAGIC_ALL) |
| 3103 curchr = Magic('$'); | |
| 3104 } | |
| 3105 break; | |
| 3106 case '\\': | |
| 3107 { | |
| 3108 int c = regparse[1]; | |
| 3109 | |
| 3110 if (c == NUL) | |
| 3111 curchr = '\\'; /* trailing '\' */ | |
| 3112 else if ( | |
| 3113 #ifdef EBCDIC | |
| 3114 vim_strchr(META, c) | |
| 3115 #else | |
| 3116 c <= '~' && META_flags[c] | |
| 3117 #endif | |
| 3118 ) | |
| 3119 { | |
| 3120 /* | |
| 3121 * META contains everything that may be magic sometimes, | |
| 3122 * except ^ and $ ("\^" and "\$" are only magic after | |
| 6830 | 3123 * "\V"). We now fetch the next character and toggle its |
| 7 | 3124 * magicness. Therefore, \ is so meta-magic that it is |
| 3125 * not in META. | |
| 3126 */ | |
| 3127 curchr = -1; | |
| 3128 prev_at_start = at_start; | |
| 3129 at_start = FALSE; /* be able to say "/\*ptr" */ | |
| 3130 ++regparse; | |
| 167 | 3131 ++after_slash; |
| 7 | 3132 peekchr(); |
| 3133 --regparse; | |
| 167 | 3134 --after_slash; |
| 7 | 3135 curchr = toggle_Magic(curchr); |
| 3136 } | |
| 3137 else if (vim_strchr(REGEXP_ABBR, c)) | |
| 3138 { | |
| 3139 /* | |
| 3140 * Handle abbreviations, like "\t" for TAB -- webb | |
| 3141 */ | |
| 3142 curchr = backslash_trans(c); | |
| 3143 } | |
| 3144 else if (reg_magic == MAGIC_NONE && (c == '$' || c == '^')) | |
| 3145 curchr = toggle_Magic(c); | |
| 3146 else | |
| 3147 { | |
| 3148 /* | |
| 3149 * Next character can never be (made) magic? | |
| 3150 * Then backslashing it won't do anything. | |
| 3151 */ | |
| 3152 #ifdef FEAT_MBYTE | |
| 3153 if (has_mbyte) | |
| 3154 curchr = (*mb_ptr2char)(regparse + 1); | |
| 3155 else | |
| 3156 #endif | |
| 3157 curchr = c; | |
| 3158 } | |
| 3159 break; | |
| 3160 } | |
| 3161 | |
| 3162 #ifdef FEAT_MBYTE | |
| 3163 default: | |
| 3164 if (has_mbyte) | |
| 3165 curchr = (*mb_ptr2char)(regparse); | |
| 3166 #endif | |
| 3167 } | |
| 3168 } | |
| 3169 | |
| 3170 return curchr; | |
| 3171 } | |
| 3172 | |
| 3173 /* | |
| 3174 * Eat one lexed character. Do this in a way that we can undo it. | |
| 3175 */ | |
| 3176 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3177 skipchr(void) |
| 7 | 3178 { |
| 3179 /* peekchr() eats a backslash, do the same here */ | |
| 3180 if (*regparse == '\\') | |
| 3181 prevchr_len = 1; | |
| 3182 else | |
| 3183 prevchr_len = 0; | |
| 3184 if (regparse[prevchr_len] != NUL) | |
| 3185 { | |
| 3186 #ifdef FEAT_MBYTE | |
| 714 | 3187 if (enc_utf8) |
| 1449 | 3188 /* exclude composing chars that mb_ptr2len does include */ |
| 3189 prevchr_len += utf_ptr2len(regparse + prevchr_len); | |
| 714 | 3190 else if (has_mbyte) |
| 474 | 3191 prevchr_len += (*mb_ptr2len)(regparse + prevchr_len); |
| 7 | 3192 else |
| 3193 #endif | |
| 3194 ++prevchr_len; | |
| 3195 } | |
| 3196 regparse += prevchr_len; | |
| 3197 prev_at_start = at_start; | |
| 3198 at_start = FALSE; | |
| 3199 prevprevchr = prevchr; | |
| 3200 prevchr = curchr; | |
| 3201 curchr = nextchr; /* use previously unget char, or -1 */ | |
| 3202 nextchr = -1; | |
| 3203 } | |
| 3204 | |
| 3205 /* | |
| 3206 * Skip a character while keeping the value of prev_at_start for at_start. | |
| 3207 * prevchr and prevprevchr are also kept. | |
| 3208 */ | |
| 3209 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3210 skipchr_keepstart(void) |
| 7 | 3211 { |
| 3212 int as = prev_at_start; | |
| 3213 int pr = prevchr; | |
| 3214 int prpr = prevprevchr; | |
| 3215 | |
| 3216 skipchr(); | |
| 3217 at_start = as; | |
| 3218 prevchr = pr; | |
| 3219 prevprevchr = prpr; | |
| 3220 } | |
| 3221 | |
| 4444 | 3222 /* |
| 3223 * Get the next character from the pattern. We know about magic and such, so | |
| 3224 * therefore we need a lexical analyzer. | |
| 3225 */ | |
| 7 | 3226 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3227 getchr(void) |
| 7 | 3228 { |
| 3229 int chr = peekchr(); | |
| 3230 | |
| 3231 skipchr(); | |
| 3232 return chr; | |
| 3233 } | |
| 3234 | |
| 3235 /* | |
| 3236 * put character back. Works only once! | |
| 3237 */ | |
| 3238 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3239 ungetchr(void) |
| 7 | 3240 { |
| 3241 nextchr = curchr; | |
| 3242 curchr = prevchr; | |
| 3243 prevchr = prevprevchr; | |
| 3244 at_start = prev_at_start; | |
| 3245 prev_at_start = FALSE; | |
| 3246 | |
| 3247 /* Backup regparse, so that it's at the same position as before the | |
| 3248 * getchr(). */ | |
| 3249 regparse -= prevchr_len; | |
| 3250 } | |
| 3251 | |
| 3252 /* | |
| 29 | 3253 * Get and return the value of the hex string at the current position. |
| 3254 * Return -1 if there is no valid hex number. | |
| 3255 * The position is updated: | |
| 24 | 3256 * blahblah\%x20asdf |
| 856 | 3257 * before-^ ^-after |
| 24 | 3258 * The parameter controls the maximum number of input characters. This will be |
| 3259 * 2 when reading a \%x20 sequence and 4 when reading a \%u20AC sequence. | |
| 3260 */ | |
| 3261 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3262 gethexchrs(int maxinputlen) |
| 24 | 3263 { |
| 3264 int nr = 0; | |
| 3265 int c; | |
| 3266 int i; | |
| 3267 | |
| 3268 for (i = 0; i < maxinputlen; ++i) | |
| 3269 { | |
| 3270 c = regparse[0]; | |
| 3271 if (!vim_isxdigit(c)) | |
| 3272 break; | |
| 3273 nr <<= 4; | |
| 3274 nr |= hex2nr(c); | |
| 3275 ++regparse; | |
| 3276 } | |
| 3277 | |
| 3278 if (i == 0) | |
| 3279 return -1; | |
| 3280 return nr; | |
| 3281 } | |
| 3282 | |
| 3283 /* | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3284 * Get and return the value of the decimal string immediately after the |
| 24 | 3285 * current position. Return -1 for invalid. Consumes all digits. |
| 3286 */ | |
| 3287 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3288 getdecchrs(void) |
| 24 | 3289 { |
| 3290 int nr = 0; | |
| 3291 int c; | |
| 3292 int i; | |
| 3293 | |
| 3294 for (i = 0; ; ++i) | |
| 3295 { | |
| 3296 c = regparse[0]; | |
| 3297 if (c < '0' || c > '9') | |
| 3298 break; | |
| 3299 nr *= 10; | |
| 3300 nr += c - '0'; | |
| 3301 ++regparse; | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3302 curchr = -1; /* no longer valid */ |
| 24 | 3303 } |
| 3304 | |
| 3305 if (i == 0) | |
| 3306 return -1; | |
| 3307 return nr; | |
| 3308 } | |
| 3309 | |
| 3310 /* | |
| 3311 * get and return the value of the octal string immediately after the current | |
| 3312 * position. Return -1 for invalid, or 0-255 for valid. Smart enough to handle | |
| 3313 * numbers > 377 correctly (for example, 400 is treated as 40) and doesn't | |
| 3314 * treat 8 or 9 as recognised characters. Position is updated: | |
| 3315 * blahblah\%o210asdf | |
| 856 | 3316 * before-^ ^-after |
| 24 | 3317 */ |
| 3318 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3319 getoctchrs(void) |
| 24 | 3320 { |
| 3321 int nr = 0; | |
| 3322 int c; | |
| 3323 int i; | |
| 3324 | |
| 3325 for (i = 0; i < 3 && nr < 040; ++i) | |
| 3326 { | |
| 3327 c = regparse[0]; | |
| 3328 if (c < '0' || c > '7') | |
| 3329 break; | |
| 3330 nr <<= 3; | |
| 3331 nr |= hex2nr(c); | |
| 3332 ++regparse; | |
| 3333 } | |
| 3334 | |
| 3335 if (i == 0) | |
| 3336 return -1; | |
| 3337 return nr; | |
| 3338 } | |
| 3339 | |
| 3340 /* | |
| 3341 * Get a number after a backslash that is inside []. | |
| 3342 * When nothing is recognized return a backslash. | |
| 3343 */ | |
| 3344 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3345 coll_get_char(void) |
| 24 | 3346 { |
| 3347 int nr = -1; | |
| 3348 | |
| 3349 switch (*regparse++) | |
| 3350 { | |
| 3351 case 'd': nr = getdecchrs(); break; | |
| 3352 case 'o': nr = getoctchrs(); break; | |
| 3353 case 'x': nr = gethexchrs(2); break; | |
| 3354 case 'u': nr = gethexchrs(4); break; | |
| 3355 case 'U': nr = gethexchrs(8); break; | |
| 3356 } | |
| 3357 if (nr < 0) | |
| 3358 { | |
| 3359 /* If getting the number fails be backwards compatible: the character | |
| 3360 * is a backslash. */ | |
| 3361 --regparse; | |
| 3362 nr = '\\'; | |
| 3363 } | |
| 3364 return nr; | |
| 3365 } | |
| 3366 | |
| 3367 /* | |
| 7 | 3368 * read_limits - Read two integers to be taken as a minimum and maximum. |
| 3369 * If the first character is '-', then the range is reversed. | |
| 3370 * Should end with 'end'. If minval is missing, zero is default, if maxval is | |
| 3371 * missing, a very big number is the default. | |
| 3372 */ | |
| 3373 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3374 read_limits(long *minval, long *maxval) |
| 7 | 3375 { |
| 3376 int reverse = FALSE; | |
| 3377 char_u *first_char; | |
| 3378 long tmp; | |
| 3379 | |
| 3380 if (*regparse == '-') | |
| 3381 { | |
| 3382 /* Starts with '-', so reverse the range later */ | |
| 3383 regparse++; | |
| 3384 reverse = TRUE; | |
| 3385 } | |
| 3386 first_char = regparse; | |
| 3387 *minval = getdigits(®parse); | |
| 3388 if (*regparse == ',') /* There is a comma */ | |
| 3389 { | |
| 3390 if (vim_isdigit(*++regparse)) | |
| 3391 *maxval = getdigits(®parse); | |
| 3392 else | |
| 3393 *maxval = MAX_LIMIT; | |
| 3394 } | |
| 3395 else if (VIM_ISDIGIT(*first_char)) | |
| 3396 *maxval = *minval; /* It was \{n} or \{-n} */ | |
| 3397 else | |
| 3398 *maxval = MAX_LIMIT; /* It was \{} or \{-} */ | |
| 3399 if (*regparse == '\\') | |
| 3400 regparse++; /* Allow either \{...} or \{...\} */ | |
| 167 | 3401 if (*regparse != '}') |
| 7 | 3402 { |
| 3403 sprintf((char *)IObuff, _("E554: Syntax error in %s{...}"), | |
| 3404 reg_magic == MAGIC_ALL ? "" : "\\"); | |
| 3405 EMSG_RET_FAIL(IObuff); | |
| 3406 } | |
| 3407 | |
| 3408 /* | |
| 3409 * Reverse the range if there was a '-', or make sure it is in the right | |
| 3410 * order otherwise. | |
| 3411 */ | |
| 3412 if ((!reverse && *minval > *maxval) || (reverse && *minval < *maxval)) | |
| 3413 { | |
| 3414 tmp = *minval; | |
| 3415 *minval = *maxval; | |
| 3416 *maxval = tmp; | |
| 3417 } | |
| 3418 skipchr(); /* let's be friends with the lexer again */ | |
| 3419 return OK; | |
| 3420 } | |
| 3421 | |
| 3422 /* | |
| 3423 * vim_regexec and friends | |
| 3424 */ | |
| 3425 | |
| 3426 /* | |
| 3427 * Global work variables for vim_regexec(). | |
| 3428 */ | |
| 3429 | |
| 3430 /* The current match-position is remembered with these variables: */ | |
| 3431 static linenr_T reglnum; /* line number, relative to first line */ | |
| 3432 static char_u *regline; /* start of current line */ | |
| 3433 static char_u *reginput; /* current input, points into "regline" */ | |
| 3434 | |
| 3435 static int need_clear_subexpr; /* subexpressions still need to be | |
| 3436 * cleared */ | |
| 3437 #ifdef FEAT_SYN_HL | |
| 3438 static int need_clear_zsubexpr = FALSE; /* extmatch subexpressions | |
| 3439 * still need to be cleared */ | |
| 3440 #endif | |
| 3441 | |
| 3442 /* | |
| 3443 * Structure used to save the current input state, when it needs to be | |
| 3444 * restored after trying a match. Used by reg_save() and reg_restore(). | |
| 233 | 3445 * Also stores the length of "backpos". |
| 7 | 3446 */ |
| 3447 typedef struct | |
| 3448 { | |
| 3449 union | |
| 3450 { | |
| 3451 char_u *ptr; /* reginput pointer, for single-line regexp */ | |
| 3452 lpos_T pos; /* reginput pos, for multi-line regexp */ | |
| 3453 } rs_u; | |
| 233 | 3454 int rs_len; |
| 7 | 3455 } regsave_T; |
| 3456 | |
| 3457 /* struct to save start/end pointer/position in for \(\) */ | |
| 3458 typedef struct | |
| 3459 { | |
| 3460 union | |
| 3461 { | |
| 3462 char_u *ptr; | |
| 3463 lpos_T pos; | |
| 3464 } se_u; | |
| 3465 } save_se_T; | |
| 3466 | |
| 1579 | 3467 /* used for BEHIND and NOBEHIND matching */ |
| 3468 typedef struct regbehind_S | |
| 3469 { | |
| 3470 regsave_T save_after; | |
| 3471 regsave_T save_behind; | |
| 1602 | 3472 int save_need_clear_subexpr; |
| 1579 | 3473 save_se_T save_start[NSUBEXP]; |
| 3474 save_se_T save_end[NSUBEXP]; | |
| 3475 } regbehind_T; | |
| 3476 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3477 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
|
3478 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
|
3479 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
|
3480 static void cleanup_subexpr(void); |
| 7 | 3481 #ifdef FEAT_SYN_HL |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3482 static void cleanup_zsubexpr(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3483 #endif |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3484 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
|
3485 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
|
3486 static void reg_nextline(void); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3487 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
|
3488 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
|
3489 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
|
3490 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
|
3491 static void save_se_one(save_se_T *savep, char_u **pp); |
| 7 | 3492 |
| 3493 /* Save the sub-expressions before attempting a match. */ | |
| 3494 #define save_se(savep, posp, pp) \ | |
| 3495 REG_MULTI ? save_se_multi((savep), (posp)) : save_se_one((savep), (pp)) | |
| 3496 | |
| 3497 /* After a failed match restore the sub-expressions. */ | |
| 3498 #define restore_se(savep, posp, pp) { \ | |
| 3499 if (REG_MULTI) \ | |
| 3500 *(posp) = (savep)->se_u.pos; \ | |
| 3501 else \ | |
| 3502 *(pp) = (savep)->se_u.ptr; } | |
| 3503 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3504 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
|
3505 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
|
3506 static int regmatch(char_u *prog); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3507 static int regrepeat(char_u *p, long maxcount); |
| 7 | 3508 |
| 3509 #ifdef DEBUG | |
| 3510 int regnarrate = 0; | |
| 3511 #endif | |
| 3512 | |
| 3513 /* | |
| 3514 * Internal copy of 'ignorecase'. It is set at each call to vim_regexec(). | |
| 3515 * Normally it gets the value of "rm_ic" or "rmm_ic", but when the pattern | |
| 3516 * contains '\c' or '\C' the value is overruled. | |
| 3517 */ | |
| 3518 static int ireg_ic; | |
| 3519 | |
| 3520 #ifdef FEAT_MBYTE | |
| 3521 /* | |
| 3522 * Similar to ireg_ic, but only for 'combining' characters. Set with \Z flag | |
| 3523 * in the regexp. Defaults to false, always. | |
| 3524 */ | |
| 3525 static int ireg_icombine; | |
| 3526 #endif | |
| 3527 | |
| 3528 /* | |
| 410 | 3529 * Copy of "rmm_maxcol": maximum column to search for a match. Zero when |
| 3530 * there is no maximum. | |
| 3531 */ | |
| 418 | 3532 static colnr_T ireg_maxcol; |
| 410 | 3533 |
| 3534 /* | |
| 7 | 3535 * Sometimes need to save a copy of a line. Since alloc()/free() is very |
| 3536 * slow, we keep one allocated piece of memory and only re-allocate it when | |
| 4444 | 3537 * it's too small. It's freed in bt_regexec_both() when finished. |
| 7 | 3538 */ |
| 1468 | 3539 static char_u *reg_tofree = NULL; |
| 7 | 3540 static unsigned reg_tofreelen; |
| 3541 | |
| 3542 /* | |
| 3543 * These variables are set when executing a regexp to speed up the execution. | |
| 1209 | 3544 * Which ones are set depends on whether a single-line or multi-line match is |
| 7 | 3545 * done: |
| 3546 * single-line multi-line | |
| 3547 * reg_match ®match_T NULL | |
| 3548 * reg_mmatch NULL ®mmatch_T | |
| 3549 * reg_startp reg_match->startp <invalid> | |
| 3550 * reg_endp reg_match->endp <invalid> | |
| 3551 * reg_startpos <invalid> reg_mmatch->startpos | |
| 3552 * reg_endpos <invalid> reg_mmatch->endpos | |
| 3553 * reg_win NULL window in which to search | |
| 4061 | 3554 * reg_buf curbuf buffer in which to search |
| 7 | 3555 * reg_firstlnum <invalid> first line in which to search |
| 3556 * reg_maxline 0 last line nr | |
| 3557 * reg_line_lbr FALSE or TRUE FALSE | |
| 3558 */ | |
| 3559 static regmatch_T *reg_match; | |
| 3560 static regmmatch_T *reg_mmatch; | |
| 3561 static char_u **reg_startp = NULL; | |
| 3562 static char_u **reg_endp = NULL; | |
| 3563 static lpos_T *reg_startpos = NULL; | |
| 3564 static lpos_T *reg_endpos = NULL; | |
| 3565 static win_T *reg_win; | |
| 3566 static buf_T *reg_buf; | |
| 3567 static linenr_T reg_firstlnum; | |
| 3568 static linenr_T reg_maxline; | |
| 3569 static int reg_line_lbr; /* "\n" in string is line break */ | |
| 3570 | |
| 270 | 3571 /* Values for rs_state in regitem_T. */ |
| 3572 typedef enum regstate_E | |
| 3573 { | |
| 3574 RS_NOPEN = 0 /* NOPEN and NCLOSE */ | |
| 3575 , RS_MOPEN /* MOPEN + [0-9] */ | |
| 3576 , RS_MCLOSE /* MCLOSE + [0-9] */ | |
| 3577 #ifdef FEAT_SYN_HL | |
| 3578 , RS_ZOPEN /* ZOPEN + [0-9] */ | |
| 3579 , RS_ZCLOSE /* ZCLOSE + [0-9] */ | |
| 3580 #endif | |
| 3581 , RS_BRANCH /* BRANCH */ | |
| 3582 , RS_BRCPLX_MORE /* BRACE_COMPLEX and trying one more match */ | |
| 3583 , RS_BRCPLX_LONG /* BRACE_COMPLEX and trying longest match */ | |
| 3584 , RS_BRCPLX_SHORT /* BRACE_COMPLEX and trying shortest match */ | |
| 3585 , RS_NOMATCH /* NOMATCH */ | |
| 3586 , RS_BEHIND1 /* BEHIND / NOBEHIND matching rest */ | |
| 3587 , RS_BEHIND2 /* BEHIND / NOBEHIND matching behind part */ | |
| 3588 , RS_STAR_LONG /* STAR/PLUS/BRACE_SIMPLE longest match */ | |
| 3589 , RS_STAR_SHORT /* STAR/PLUS/BRACE_SIMPLE shortest match */ | |
| 3590 } regstate_T; | |
| 3591 | |
| 3592 /* | |
| 3593 * When there are alternatives a regstate_T is put on the regstack to remember | |
| 3594 * what we are doing. | |
| 3595 * Before it may be another type of item, depending on rs_state, to remember | |
| 3596 * more things. | |
| 3597 */ | |
| 3598 typedef struct regitem_S | |
| 3599 { | |
| 3600 regstate_T rs_state; /* what we are doing, one of RS_ above */ | |
| 3601 char_u *rs_scan; /* current node in program */ | |
| 3602 union | |
| 3603 { | |
| 3604 save_se_T sesave; | |
| 3605 regsave_T regsave; | |
| 3606 } rs_un; /* room for saving reginput */ | |
| 1579 | 3607 short rs_no; /* submatch nr or BEHIND/NOBEHIND */ |
| 270 | 3608 } regitem_T; |
| 3609 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3610 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
|
3611 static void regstack_pop(char_u **scan); |
| 270 | 3612 |
| 3613 /* used for STAR, PLUS and BRACE_SIMPLE matching */ | |
| 3614 typedef struct regstar_S | |
| 3615 { | |
| 3616 int nextb; /* next byte */ | |
| 3617 int nextb_ic; /* next byte reverse case */ | |
| 3618 long count; | |
| 3619 long minval; | |
| 3620 long maxval; | |
| 3621 } regstar_T; | |
| 3622 | |
| 3623 /* used to store input position when a BACK was encountered, so that we now if | |
| 3624 * we made any progress since the last time. */ | |
| 3625 typedef struct backpos_S | |
| 3626 { | |
| 3627 char_u *bp_scan; /* "scan" where BACK was encountered */ | |
| 3628 regsave_T bp_pos; /* last input position */ | |
| 3629 } backpos_T; | |
| 3630 | |
| 3631 /* | |
| 1520 | 3632 * "regstack" and "backpos" are used by regmatch(). They are kept over calls |
| 3633 * to avoid invoking malloc() and free() often. | |
| 3634 * "regstack" is a stack with regitem_T items, sometimes preceded by regstar_T | |
| 3635 * or regbehind_T. | |
| 3636 * "backpos_T" is a table with backpos_T for BACK | |
| 270 | 3637 */ |
| 1520 | 3638 static garray_T regstack = {0, 0, 0, 0, NULL}; |
| 3639 static garray_T backpos = {0, 0, 0, 0, NULL}; | |
| 3640 | |
| 3641 /* | |
| 3642 * Both for regstack and backpos tables we use the following strategy of | |
| 3643 * allocation (to reduce malloc/free calls): | |
| 3644 * - Initial size is fairly small. | |
| 3645 * - When needed, the tables are grown bigger (8 times at first, double after | |
| 3646 * that). | |
| 3647 * - After executing the match we free the memory only if the array has grown. | |
| 3648 * Thus the memory is kept allocated when it's at the initial size. | |
| 3649 * This makes it fast while not keeping a lot of memory allocated. | |
| 3650 * A three times speed increase was observed when using many simple patterns. | |
| 3651 */ | |
| 3652 #define REGSTACK_INITIAL 2048 | |
| 3653 #define BACKPOS_INITIAL 64 | |
| 3654 | |
| 3655 #if defined(EXITFREE) || defined(PROTO) | |
| 3656 void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3657 free_regexp_stuff(void) |
| 1520 | 3658 { |
| 3659 ga_clear(®stack); | |
| 3660 ga_clear(&backpos); | |
| 3661 vim_free(reg_tofree); | |
| 3662 vim_free(reg_prev_sub); | |
| 3663 } | |
| 3664 #endif | |
| 270 | 3665 |
| 7 | 3666 /* |
| 3667 * Get pointer to the line "lnum", which is relative to "reg_firstlnum". | |
| 3668 */ | |
| 3669 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3670 reg_getline(linenr_T lnum) |
| 7 | 3671 { |
| 3672 /* when looking behind for a match/no-match lnum is negative. But we | |
| 3673 * can't go before line 1 */ | |
| 3674 if (reg_firstlnum + lnum < 1) | |
| 3675 return NULL; | |
| 482 | 3676 if (lnum > reg_maxline) |
| 481 | 3677 /* Must have matched the "\n" in the last line. */ |
| 3678 return (char_u *)""; | |
| 7 | 3679 return ml_get_buf(reg_buf, reg_firstlnum + lnum, FALSE); |
| 3680 } | |
| 3681 | |
| 3682 static regsave_T behind_pos; | |
| 3683 | |
| 3684 #ifdef FEAT_SYN_HL | |
| 3685 static char_u *reg_startzp[NSUBEXP]; /* Workspace to mark beginning */ | |
| 3686 static char_u *reg_endzp[NSUBEXP]; /* and end of \z(...\) matches */ | |
| 3687 static lpos_T reg_startzpos[NSUBEXP]; /* idem, beginning pos */ | |
| 3688 static lpos_T reg_endzpos[NSUBEXP]; /* idem, end pos */ | |
| 3689 #endif | |
| 3690 | |
| 3691 /* TRUE if using multi-line regexp. */ | |
| 3692 #define REG_MULTI (reg_match == NULL) | |
| 3693 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3694 static int bt_regexec_nl(regmatch_T *rmp, char_u *line, colnr_T col, int line_lbr); |
| 5838 | 3695 |
| 4444 | 3696 |
| 7 | 3697 /* |
| 3698 * Match a regexp against a string. | |
| 3699 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3700 * Uses curbuf for line count and 'iskeyword'. | |
| 5838 | 3701 * if "line_lbr" is TRUE consider a "\n" in "line" to be a line break. |
| 7 | 3702 * |
| 6390 | 3703 * Returns 0 for failure, number of lines contained in the match otherwise. |
| 7 | 3704 */ |
| 4444 | 3705 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3706 bt_regexec_nl( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3707 regmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3708 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
|
3709 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
|
3710 int line_lbr) |
| 7 | 3711 { |
| 3712 reg_match = rmp; | |
| 3713 reg_mmatch = NULL; | |
| 3714 reg_maxline = 0; | |
| 5838 | 3715 reg_line_lbr = line_lbr; |
| 4061 | 3716 reg_buf = curbuf; |
| 7 | 3717 reg_win = NULL; |
| 3718 ireg_ic = rmp->rm_ic; | |
| 3719 #ifdef FEAT_MBYTE | |
| 3720 ireg_icombine = FALSE; | |
| 3721 #endif | |
| 410 | 3722 ireg_maxcol = 0; |
| 6390 | 3723 |
| 3724 return bt_regexec_both(line, col, NULL); | |
| 7 | 3725 } |
| 3726 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3727 static long bt_regexec_multi(regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm); |
| 4444 | 3728 |
| 7 | 3729 /* |
| 3730 * Match a regexp against multiple lines. | |
| 3731 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3732 * Uses curbuf for line count and 'iskeyword'. | |
| 3733 * | |
| 3734 * Return zero if there is no match. Return number of lines contained in the | |
| 3735 * match otherwise. | |
| 3736 */ | |
| 4444 | 3737 static long |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3738 bt_regexec_multi( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3739 regmmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3740 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
|
3741 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
|
3742 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
|
3743 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
|
3744 proftime_T *tm) /* timeout limit or NULL */ |
| 7 | 3745 { |
| 3746 reg_match = NULL; | |
| 3747 reg_mmatch = rmp; | |
| 3748 reg_buf = buf; | |
| 3749 reg_win = win; | |
| 3750 reg_firstlnum = lnum; | |
| 3751 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
| 3752 reg_line_lbr = FALSE; | |
| 3753 ireg_ic = rmp->rmm_ic; | |
| 3754 #ifdef FEAT_MBYTE | |
| 3755 ireg_icombine = FALSE; | |
| 3756 #endif | |
| 410 | 3757 ireg_maxcol = rmp->rmm_maxcol; |
| 7 | 3758 |
| 6390 | 3759 return bt_regexec_both(NULL, col, tm); |
| 7 | 3760 } |
| 3761 | |
| 3762 /* | |
| 3763 * Match a regexp against a string ("line" points to the string) or multiple | |
| 3764 * lines ("line" is NULL, use reg_getline()). | |
| 6390 | 3765 * Returns 0 for failure, number of lines contained in the match otherwise. |
| 7 | 3766 */ |
| 3767 static long | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3768 bt_regexec_both( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3769 char_u *line, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
3770 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
|
3771 proftime_T *tm UNUSED) /* timeout limit or NULL */ |
| 7 | 3772 { |
| 6390 | 3773 bt_regprog_T *prog; |
| 3774 char_u *s; | |
| 3775 long retval = 0L; | |
| 7 | 3776 |
| 1520 | 3777 /* Create "regstack" and "backpos" if they are not allocated yet. |
| 3778 * We allocate *_INITIAL amount of bytes first and then set the grow size | |
| 3779 * to much bigger value to avoid many malloc calls in case of deep regular | |
| 3780 * expressions. */ | |
| 3781 if (regstack.ga_data == NULL) | |
| 3782 { | |
| 3783 /* Use an item size of 1 byte, since we push different things | |
| 3784 * onto the regstack. */ | |
| 3785 ga_init2(®stack, 1, REGSTACK_INITIAL); | |
| 7009 | 3786 (void)ga_grow(®stack, REGSTACK_INITIAL); |
| 1520 | 3787 regstack.ga_growsize = REGSTACK_INITIAL * 8; |
| 3788 } | |
| 3789 | |
| 3790 if (backpos.ga_data == NULL) | |
| 3791 { | |
| 3792 ga_init2(&backpos, sizeof(backpos_T), BACKPOS_INITIAL); | |
| 7009 | 3793 (void)ga_grow(&backpos, BACKPOS_INITIAL); |
| 1520 | 3794 backpos.ga_growsize = BACKPOS_INITIAL * 8; |
| 3795 } | |
| 270 | 3796 |
| 7 | 3797 if (REG_MULTI) |
| 3798 { | |
| 4444 | 3799 prog = (bt_regprog_T *)reg_mmatch->regprog; |
| 7 | 3800 line = reg_getline((linenr_T)0); |
| 3801 reg_startpos = reg_mmatch->startpos; | |
| 3802 reg_endpos = reg_mmatch->endpos; | |
| 3803 } | |
| 3804 else | |
| 3805 { | |
| 4444 | 3806 prog = (bt_regprog_T *)reg_match->regprog; |
| 7 | 3807 reg_startp = reg_match->startp; |
| 3808 reg_endp = reg_match->endp; | |
| 3809 } | |
| 3810 | |
| 3811 /* Be paranoid... */ | |
| 3812 if (prog == NULL || line == NULL) | |
| 3813 { | |
| 3814 EMSG(_(e_null)); | |
| 3815 goto theend; | |
| 3816 } | |
| 3817 | |
| 3818 /* Check validity of program. */ | |
| 3819 if (prog_magic_wrong()) | |
| 3820 goto theend; | |
| 3821 | |
| 410 | 3822 /* If the start column is past the maximum column: no need to try. */ |
| 3823 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3824 goto theend; | |
| 3825 | |
| 7 | 3826 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ |
| 3827 if (prog->regflags & RF_ICASE) | |
| 3828 ireg_ic = TRUE; | |
| 3829 else if (prog->regflags & RF_NOICASE) | |
| 3830 ireg_ic = FALSE; | |
| 3831 | |
| 3832 #ifdef FEAT_MBYTE | |
| 3833 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
| 3834 if (prog->regflags & RF_ICOMBINE) | |
| 3835 ireg_icombine = TRUE; | |
| 3836 #endif | |
| 3837 | |
| 3838 /* If there is a "must appear" string, look for it. */ | |
| 3839 if (prog->regmust != NULL) | |
| 3840 { | |
| 3841 int c; | |
| 3842 | |
| 3843 #ifdef FEAT_MBYTE | |
| 3844 if (has_mbyte) | |
| 3845 c = (*mb_ptr2char)(prog->regmust); | |
| 3846 else | |
| 3847 #endif | |
| 3848 c = *prog->regmust; | |
| 3849 s = line + col; | |
| 170 | 3850 |
| 3851 /* | |
| 3852 * This is used very often, esp. for ":global". Use three versions of | |
| 3853 * the loop to avoid overhead of conditions. | |
| 3854 */ | |
| 3855 if (!ireg_ic | |
| 3856 #ifdef FEAT_MBYTE | |
| 3857 && !has_mbyte | |
| 3858 #endif | |
| 3859 ) | |
| 3860 while ((s = vim_strbyte(s, c)) != NULL) | |
| 3861 { | |
| 3862 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3863 break; /* Found it. */ | |
| 3864 ++s; | |
| 3865 } | |
| 3866 #ifdef FEAT_MBYTE | |
| 3867 else if (!ireg_ic || (!enc_utf8 && mb_char2len(c) > 1)) | |
| 3868 while ((s = vim_strchr(s, c)) != NULL) | |
| 3869 { | |
| 3870 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3871 break; /* Found it. */ | |
| 3872 mb_ptr_adv(s); | |
| 3873 } | |
| 3874 #endif | |
| 3875 else | |
| 3876 while ((s = cstrchr(s, c)) != NULL) | |
| 3877 { | |
| 3878 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3879 break; /* Found it. */ | |
| 3880 mb_ptr_adv(s); | |
| 3881 } | |
| 7 | 3882 if (s == NULL) /* Not present. */ |
| 3883 goto theend; | |
| 3884 } | |
| 3885 | |
| 3886 regline = line; | |
| 3887 reglnum = 0; | |
| 2578 | 3888 reg_toolong = FALSE; |
| 7 | 3889 |
| 3890 /* Simplest case: Anchored match need be tried only once. */ | |
| 3891 if (prog->reganch) | |
| 3892 { | |
| 3893 int c; | |
| 3894 | |
| 3895 #ifdef FEAT_MBYTE | |
| 3896 if (has_mbyte) | |
| 3897 c = (*mb_ptr2char)(regline + col); | |
| 3898 else | |
| 3899 #endif | |
| 3900 c = regline[col]; | |
| 3901 if (prog->regstart == NUL | |
| 3902 || prog->regstart == c | |
| 3903 || (ireg_ic && (( | |
| 3904 #ifdef FEAT_MBYTE | |
| 3905 (enc_utf8 && utf_fold(prog->regstart) == utf_fold(c))) | |
| 3906 || (c < 255 && prog->regstart < 255 && | |
| 3907 #endif | |
| 1347 | 3908 MB_TOLOWER(prog->regstart) == MB_TOLOWER(c))))) |
| 7 | 3909 retval = regtry(prog, col); |
| 3910 else | |
| 3911 retval = 0; | |
| 3912 } | |
| 3913 else | |
| 3914 { | |
| 1521 | 3915 #ifdef FEAT_RELTIME |
| 3916 int tm_count = 0; | |
| 3917 #endif | |
| 7 | 3918 /* Messy cases: unanchored match. */ |
| 180 | 3919 while (!got_int) |
| 7 | 3920 { |
| 3921 if (prog->regstart != NUL) | |
| 3922 { | |
| 170 | 3923 /* Skip until the char we know it must start with. |
| 3924 * Used often, do some work to avoid call overhead. */ | |
| 3925 if (!ireg_ic | |
| 3926 #ifdef FEAT_MBYTE | |
| 3927 && !has_mbyte | |
| 3928 #endif | |
| 3929 ) | |
| 3930 s = vim_strbyte(regline + col, prog->regstart); | |
| 3931 else | |
| 3932 s = cstrchr(regline + col, prog->regstart); | |
| 7 | 3933 if (s == NULL) |
| 3934 { | |
| 3935 retval = 0; | |
| 3936 break; | |
| 3937 } | |
| 3938 col = (int)(s - regline); | |
| 3939 } | |
| 3940 | |
| 410 | 3941 /* Check for maximum column to try. */ |
| 3942 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3943 { | |
| 3944 retval = 0; | |
| 3945 break; | |
| 3946 } | |
| 3947 | |
| 7 | 3948 retval = regtry(prog, col); |
| 3949 if (retval > 0) | |
| 3950 break; | |
| 3951 | |
| 3952 /* if not currently on the first line, get it again */ | |
| 3953 if (reglnum != 0) | |
| 3954 { | |
| 481 | 3955 reglnum = 0; |
| 7 | 3956 regline = reg_getline((linenr_T)0); |
| 3957 } | |
| 3958 if (regline[col] == NUL) | |
| 3959 break; | |
| 3960 #ifdef FEAT_MBYTE | |
| 3961 if (has_mbyte) | |
| 474 | 3962 col += (*mb_ptr2len)(regline + col); |
| 7 | 3963 else |
| 3964 #endif | |
| 3965 ++col; | |
| 1521 | 3966 #ifdef FEAT_RELTIME |
| 3967 /* Check for timeout once in a twenty times to avoid overhead. */ | |
| 3968 if (tm != NULL && ++tm_count == 20) | |
| 3969 { | |
| 3970 tm_count = 0; | |
| 3971 if (profile_passed_limit(tm)) | |
| 3972 break; | |
| 3973 } | |
| 3974 #endif | |
| 7 | 3975 } |
| 3976 } | |
| 3977 | |
| 3978 theend: | |
| 1520 | 3979 /* Free "reg_tofree" when it's a bit big. |
| 3980 * Free regstack and backpos if they are bigger than their initial size. */ | |
| 3981 if (reg_tofreelen > 400) | |
| 3982 { | |
| 3983 vim_free(reg_tofree); | |
| 3984 reg_tofree = NULL; | |
| 3985 } | |
| 3986 if (regstack.ga_maxlen > REGSTACK_INITIAL) | |
| 3987 ga_clear(®stack); | |
| 3988 if (backpos.ga_maxlen > BACKPOS_INITIAL) | |
| 3989 ga_clear(&backpos); | |
| 270 | 3990 |
| 7 | 3991 return retval; |
| 3992 } | |
| 3993 | |
| 3994 #ifdef FEAT_SYN_HL | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
3995 static reg_extmatch_T *make_extmatch(void); |
| 7 | 3996 |
| 3997 /* | |
| 3998 * Create a new extmatch and mark it as referenced once. | |
| 3999 */ | |
| 4000 static reg_extmatch_T * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4001 make_extmatch(void) |
| 7 | 4002 { |
| 4003 reg_extmatch_T *em; | |
| 4004 | |
| 4005 em = (reg_extmatch_T *)alloc_clear((unsigned)sizeof(reg_extmatch_T)); | |
| 4006 if (em != NULL) | |
| 4007 em->refcnt = 1; | |
| 4008 return em; | |
| 4009 } | |
| 4010 | |
| 4011 /* | |
| 4012 * Add a reference to an extmatch. | |
| 4013 */ | |
| 4014 reg_extmatch_T * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4015 ref_extmatch(reg_extmatch_T *em) |
| 7 | 4016 { |
| 4017 if (em != NULL) | |
| 4018 em->refcnt++; | |
| 4019 return em; | |
| 4020 } | |
| 4021 | |
| 4022 /* | |
| 4023 * Remove a reference to an extmatch. If there are no references left, free | |
| 4024 * the info. | |
| 4025 */ | |
| 4026 void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4027 unref_extmatch(reg_extmatch_T *em) |
| 7 | 4028 { |
| 4029 int i; | |
| 4030 | |
| 4031 if (em != NULL && --em->refcnt <= 0) | |
| 4032 { | |
| 4033 for (i = 0; i < NSUBEXP; ++i) | |
| 4034 vim_free(em->matches[i]); | |
| 4035 vim_free(em); | |
| 4036 } | |
| 4037 } | |
| 4038 #endif | |
| 4039 | |
| 4040 /* | |
| 4041 * regtry - try match of "prog" with at regline["col"]. | |
| 4042 * Returns 0 for failure, number of lines contained in the match otherwise. | |
| 4043 */ | |
| 4044 static long | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4045 regtry(bt_regprog_T *prog, colnr_T col) |
| 7 | 4046 { |
| 4047 reginput = regline + col; | |
| 4048 need_clear_subexpr = TRUE; | |
| 4049 #ifdef FEAT_SYN_HL | |
| 4050 /* Clear the external match subpointers if necessary. */ | |
| 4051 if (prog->reghasz == REX_SET) | |
| 4052 need_clear_zsubexpr = TRUE; | |
| 4053 #endif | |
| 4054 | |
| 189 | 4055 if (regmatch(prog->program + 1) == 0) |
| 4056 return 0; | |
| 4057 | |
| 4058 cleanup_subexpr(); | |
| 4059 if (REG_MULTI) | |
| 7 | 4060 { |
| 189 | 4061 if (reg_startpos[0].lnum < 0) |
| 7 | 4062 { |
| 189 | 4063 reg_startpos[0].lnum = 0; |
| 4064 reg_startpos[0].col = col; | |
| 4065 } | |
| 4066 if (reg_endpos[0].lnum < 0) | |
| 4067 { | |
| 4068 reg_endpos[0].lnum = reglnum; | |
| 4069 reg_endpos[0].col = (int)(reginput - regline); | |
| 7 | 4070 } |
| 4071 else | |
| 189 | 4072 /* Use line number of "\ze". */ |
| 4073 reglnum = reg_endpos[0].lnum; | |
| 4074 } | |
| 4075 else | |
| 4076 { | |
| 4077 if (reg_startp[0] == NULL) | |
| 4078 reg_startp[0] = regline + col; | |
| 4079 if (reg_endp[0] == NULL) | |
| 4080 reg_endp[0] = reginput; | |
| 4081 } | |
| 7 | 4082 #ifdef FEAT_SYN_HL |
| 189 | 4083 /* Package any found \z(...\) matches for export. Default is none. */ |
| 4084 unref_extmatch(re_extmatch_out); | |
| 4085 re_extmatch_out = NULL; | |
| 4086 | |
| 4087 if (prog->reghasz == REX_SET) | |
| 4088 { | |
| 4089 int i; | |
| 4090 | |
| 4091 cleanup_zsubexpr(); | |
| 4092 re_extmatch_out = make_extmatch(); | |
| 4093 for (i = 0; i < NSUBEXP; i++) | |
| 7 | 4094 { |
| 189 | 4095 if (REG_MULTI) |
| 7 | 4096 { |
| 189 | 4097 /* Only accept single line matches. */ |
| 4098 if (reg_startzpos[i].lnum >= 0 | |
| 5820 | 4099 && reg_endzpos[i].lnum == reg_startzpos[i].lnum |
| 4100 && reg_endzpos[i].col >= reg_startzpos[i].col) | |
| 189 | 4101 re_extmatch_out->matches[i] = |
| 4102 vim_strnsave(reg_getline(reg_startzpos[i].lnum) | |
| 7 | 4103 + reg_startzpos[i].col, |
| 189 | 4104 reg_endzpos[i].col - reg_startzpos[i].col); |
| 4105 } | |
| 4106 else | |
| 4107 { | |
| 4108 if (reg_startzp[i] != NULL && reg_endzp[i] != NULL) | |
| 4109 re_extmatch_out->matches[i] = | |
| 7 | 4110 vim_strnsave(reg_startzp[i], |
| 189 | 4111 (int)(reg_endzp[i] - reg_startzp[i])); |
| 7 | 4112 } |
| 4113 } | |
| 189 | 4114 } |
| 7 | 4115 #endif |
| 189 | 4116 return 1 + reglnum; |
| 7 | 4117 } |
| 4118 | |
| 4119 #ifdef FEAT_MBYTE | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
4120 static int reg_prev_class(void); |
| 7 | 4121 |
| 4122 /* | |
| 4123 * Get class of previous character. | |
| 4124 */ | |
| 4125 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4126 reg_prev_class(void) |
| 7 | 4127 { |
| 4128 if (reginput > regline) | |
| 4069 | 4129 return mb_get_class_buf(reginput - 1 |
| 4130 - (*mb_head_off)(regline, reginput - 1), reg_buf); | |
| 7 | 4131 return -1; |
| 4132 } | |
| 5735 | 4133 #endif |
| 4134 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
4135 static int reg_match_visual(void); |
|
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4136 |
|
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 * 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
|
4139 */ |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4140 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4141 reg_match_visual(void) |
|
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4142 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4143 pos_T top, bot; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4144 linenr_T lnum; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4145 colnr_T col; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4146 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
|
4147 int mode; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4148 colnr_T start, end; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4149 colnr_T start2, end2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4150 colnr_T cols; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4151 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4152 /* Check if the buffer is the current buffer. */ |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4153 if (reg_buf != curbuf || VIsual.lnum == 0) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4154 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4155 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4156 if (VIsual_active) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4157 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4158 if (lt(VIsual, wp->w_cursor)) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4159 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4160 top = VIsual; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4161 bot = wp->w_cursor; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4162 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4163 else |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4164 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4165 top = wp->w_cursor; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4166 bot = VIsual; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4167 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4168 mode = VIsual_mode; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4169 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4170 else |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4171 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4172 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
|
4173 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4174 top = curbuf->b_visual.vi_start; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4175 bot = curbuf->b_visual.vi_end; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4176 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4177 else |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4178 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4179 top = curbuf->b_visual.vi_end; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4180 bot = curbuf->b_visual.vi_start; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4181 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4182 mode = curbuf->b_visual.vi_mode; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4183 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4184 lnum = reglnum + reg_firstlnum; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4185 if (lnum < top.lnum || lnum > bot.lnum) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4186 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4187 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4188 if (mode == 'v') |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4189 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4190 col = (colnr_T)(reginput - regline); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4191 if ((lnum == top.lnum && col < top.col) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4192 || (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
|
4193 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4194 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4195 else if (mode == Ctrl_V) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4196 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4197 getvvcol(wp, &top, &start, NULL, &end); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4198 getvvcol(wp, &bot, &start2, NULL, &end2); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4199 if (start2 < start) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4200 start = start2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4201 if (end2 > end) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4202 end = end2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4203 if (top.col == MAXCOL || bot.col == MAXCOL) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4204 end = MAXCOL; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4205 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
|
4206 if (cols < start || cols > end - (*p_sel == 'e')) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4207 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4208 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4209 return TRUE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4210 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4211 |
| 39 | 4212 #define ADVANCE_REGINPUT() mb_ptr_adv(reginput) |
| 7 | 4213 |
| 4214 /* | |
| 4215 * The arguments from BRACE_LIMITS are stored here. They are actually local | |
| 4216 * to regmatch(), but they are here to reduce the amount of stack space used | |
| 4217 * (it can be called recursively many times). | |
| 4218 */ | |
| 4219 static long bl_minval; | |
| 4220 static long bl_maxval; | |
| 4221 | |
| 4222 /* | |
| 4223 * regmatch - main matching routine | |
| 4224 * | |
| 180 | 4225 * Conceptually the strategy is simple: Check to see whether the current node |
| 4226 * matches, push an item onto the regstack and loop to see whether the rest | |
| 4227 * matches, and then act accordingly. In practice we make some effort to | |
| 4228 * avoid using the regstack, in particular by going through "ordinary" nodes | |
| 4229 * (that don't need to know whether the rest of the match failed) by a nested | |
| 4230 * loop. | |
| 7 | 4231 * |
| 4232 * Returns TRUE when there is a match. Leaves reginput and reglnum just after | |
| 4233 * the last matched character. | |
| 4234 * Returns FALSE when there is no match. Leaves reginput and reglnum in an | |
| 4235 * undefined state! | |
| 4236 */ | |
| 4237 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4238 regmatch( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
4239 char_u *scan) /* Current node. */ |
| 7 | 4240 { |
| 180 | 4241 char_u *next; /* Next node. */ |
| 4242 int op; | |
| 4243 int c; | |
| 4244 regitem_T *rp; | |
| 4245 int no; | |
| 4246 int status; /* one of the RA_ values: */ | |
| 4247 #define RA_FAIL 1 /* something failed, abort */ | |
| 4248 #define RA_CONT 2 /* continue in inner loop */ | |
| 4249 #define RA_BREAK 3 /* break inner loop */ | |
| 4250 #define RA_MATCH 4 /* successful match */ | |
| 4251 #define RA_NOMATCH 5 /* didn't match */ | |
| 270 | 4252 |
| 1520 | 4253 /* Make "regstack" and "backpos" empty. They are allocated and freed in |
| 4444 | 4254 * bt_regexec_both() to reduce malloc()/free() calls. */ |
| 270 | 4255 regstack.ga_len = 0; |
| 4256 backpos.ga_len = 0; | |
| 233 | 4257 |
| 180 | 4258 /* |
| 233 | 4259 * Repeat until "regstack" is empty. |
| 180 | 4260 */ |
| 4261 for (;;) | |
| 4262 { | |
| 5310 | 4263 /* Some patterns may take a long time to match, e.g., "\([a-z]\+\)\+Q". |
| 4264 * Allow interrupting them with CTRL-C. */ | |
| 7 | 4265 fast_breakcheck(); |
| 4266 | |
| 4267 #ifdef DEBUG | |
| 4268 if (scan != NULL && regnarrate) | |
| 4269 { | |
| 4444 | 4270 mch_errmsg((char *)regprop(scan)); |
| 7 | 4271 mch_errmsg("(\n"); |
| 4272 } | |
| 4273 #endif | |
| 180 | 4274 |
| 4275 /* | |
| 233 | 4276 * Repeat for items that can be matched sequentially, without using the |
| 180 | 4277 * regstack. |
| 4278 */ | |
| 4279 for (;;) | |
| 7 | 4280 { |
| 180 | 4281 if (got_int || scan == NULL) |
| 4282 { | |
| 4283 status = RA_FAIL; | |
| 4284 break; | |
| 4285 } | |
| 4286 status = RA_CONT; | |
| 4287 | |
| 7 | 4288 #ifdef DEBUG |
| 4289 if (regnarrate) | |
| 4290 { | |
| 4444 | 4291 mch_errmsg((char *)regprop(scan)); |
| 7 | 4292 mch_errmsg("...\n"); |
| 4293 # ifdef FEAT_SYN_HL | |
| 4294 if (re_extmatch_in != NULL) | |
| 4295 { | |
| 4296 int i; | |
| 4297 | |
| 4298 mch_errmsg(_("External submatches:\n")); | |
| 4299 for (i = 0; i < NSUBEXP; i++) | |
| 4300 { | |
| 4301 mch_errmsg(" \""); | |
| 4302 if (re_extmatch_in->matches[i] != NULL) | |
| 4444 | 4303 mch_errmsg((char *)re_extmatch_in->matches[i]); |
| 7 | 4304 mch_errmsg("\"\n"); |
| 4305 } | |
| 4306 } | |
| 4307 # endif | |
| 4308 } | |
| 4309 #endif | |
| 4310 next = regnext(scan); | |
| 4311 | |
| 4312 op = OP(scan); | |
| 4313 /* Check for character class with NL added. */ | |
| 1018 | 4314 if (!reg_line_lbr && WITH_NL(op) && REG_MULTI |
| 4315 && *reginput == NUL && reglnum <= reg_maxline) | |
| 7 | 4316 { |
| 4317 reg_nextline(); | |
| 4318 } | |
| 4319 else if (reg_line_lbr && WITH_NL(op) && *reginput == '\n') | |
| 4320 { | |
| 4321 ADVANCE_REGINPUT(); | |
| 4322 } | |
| 4323 else | |
| 4324 { | |
| 4325 if (WITH_NL(op)) | |
| 180 | 4326 op -= ADD_NL; |
| 7 | 4327 #ifdef FEAT_MBYTE |
| 4328 if (has_mbyte) | |
| 4329 c = (*mb_ptr2char)(reginput); | |
| 4330 else | |
| 4331 #endif | |
| 4332 c = *reginput; | |
| 4333 switch (op) | |
| 4334 { | |
| 4335 case BOL: | |
| 4336 if (reginput != regline) | |
| 180 | 4337 status = RA_NOMATCH; |
| 7 | 4338 break; |
| 4339 | |
| 4340 case EOL: | |
| 4341 if (c != NUL) | |
| 180 | 4342 status = RA_NOMATCH; |
| 7 | 4343 break; |
| 4344 | |
| 4345 case RE_BOF: | |
| 1458 | 4346 /* We're not at the beginning of the file when below the first |
| 4347 * line where we started, not at the start of the line or we | |
| 4348 * didn't start at the first line of the buffer. */ | |
| 7 | 4349 if (reglnum != 0 || reginput != regline |
| 1458 | 4350 || (REG_MULTI && reg_firstlnum > 1)) |
| 180 | 4351 status = RA_NOMATCH; |
| 7 | 4352 break; |
| 4353 | |
| 4354 case RE_EOF: | |
| 4355 if (reglnum != reg_maxline || c != NUL) | |
| 180 | 4356 status = RA_NOMATCH; |
| 7 | 4357 break; |
| 4358 | |
| 4359 case CURSOR: | |
| 4360 /* Check if the buffer is in a window and compare the | |
| 4361 * reg_win->w_cursor position to the match position. */ | |
| 4362 if (reg_win == NULL | |
| 4363 || (reglnum + reg_firstlnum != reg_win->w_cursor.lnum) | |
| 4364 || ((colnr_T)(reginput - regline) != reg_win->w_cursor.col)) | |
| 180 | 4365 status = RA_NOMATCH; |
| 7 | 4366 break; |
| 4367 | |
| 639 | 4368 case RE_MARK: |
|
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4369 /* Compare the mark position to the match position. */ |
| 639 | 4370 { |
| 4371 int mark = OPERAND(scan)[0]; | |
| 4372 int cmp = OPERAND(scan)[1]; | |
| 4373 pos_T *pos; | |
| 4374 | |
| 4043 | 4375 pos = getmark_buf(reg_buf, mark, FALSE); |
| 1148 | 4376 if (pos == NULL /* mark doesn't exist */ |
|
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4377 || pos->lnum <= 0 /* mark isn't set in reg_buf */ |
| 639 | 4378 || (pos->lnum == reglnum + reg_firstlnum |
| 4379 ? (pos->col == (colnr_T)(reginput - regline) | |
| 4380 ? (cmp == '<' || cmp == '>') | |
| 4381 : (pos->col < (colnr_T)(reginput - regline) | |
| 4382 ? cmp != '>' | |
| 4383 : cmp != '<')) | |
| 4384 : (pos->lnum < reglnum + reg_firstlnum | |
| 4385 ? cmp != '>' | |
| 4386 : cmp != '<'))) | |
| 4387 status = RA_NOMATCH; | |
| 4388 } | |
| 4389 break; | |
| 4390 | |
| 4391 case RE_VISUAL: | |
|
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4392 if (!reg_match_visual()) |
| 639 | 4393 status = RA_NOMATCH; |
| 4394 break; | |
| 4395 | |
| 7 | 4396 case RE_LNUM: |
| 4397 if (!REG_MULTI || !re_num_cmp((long_u)(reglnum + reg_firstlnum), | |
| 4398 scan)) | |
| 180 | 4399 status = RA_NOMATCH; |
| 7 | 4400 break; |
| 4401 | |
| 4402 case RE_COL: | |
| 4403 if (!re_num_cmp((long_u)(reginput - regline) + 1, scan)) | |
| 180 | 4404 status = RA_NOMATCH; |
| 7 | 4405 break; |
| 4406 | |
| 4407 case RE_VCOL: | |
| 4408 if (!re_num_cmp((long_u)win_linetabsize( | |
| 4409 reg_win == NULL ? curwin : reg_win, | |
| 4410 regline, (colnr_T)(reginput - regline)) + 1, scan)) | |
| 180 | 4411 status = RA_NOMATCH; |
| 7 | 4412 break; |
| 4413 | |
| 4414 case BOW: /* \<word; reginput points to w */ | |
| 4415 if (c == NUL) /* Can't match at end of line */ | |
| 180 | 4416 status = RA_NOMATCH; |
| 7 | 4417 #ifdef FEAT_MBYTE |
| 180 | 4418 else if (has_mbyte) |
| 7 | 4419 { |
| 4420 int this_class; | |
| 4421 | |
| 4422 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4423 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4424 if (this_class <= 1) |
| 180 | 4425 status = RA_NOMATCH; /* not on a word at all */ |
| 4426 else if (reg_prev_class() == this_class) | |
| 4427 status = RA_NOMATCH; /* previous char is in same word */ | |
| 7 | 4428 } |
| 4429 #endif | |
| 4430 else | |
| 4431 { | |
| 4061 | 4432 if (!vim_iswordc_buf(c, reg_buf) || (reginput > regline |
| 4433 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
| 180 | 4434 status = RA_NOMATCH; |
| 7 | 4435 } |
| 4436 break; | |
| 4437 | |
| 4438 case EOW: /* word\>; reginput points after d */ | |
| 4439 if (reginput == regline) /* Can't match at start of line */ | |
| 180 | 4440 status = RA_NOMATCH; |
| 7 | 4441 #ifdef FEAT_MBYTE |
| 180 | 4442 else if (has_mbyte) |
| 7 | 4443 { |
| 4444 int this_class, prev_class; | |
| 4445 | |
| 4446 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4447 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4448 prev_class = reg_prev_class(); |
| 180 | 4449 if (this_class == prev_class |
| 4450 || prev_class == 0 || prev_class == 1) | |
| 4451 status = RA_NOMATCH; | |
| 7 | 4452 } |
| 180 | 4453 #endif |
| 7 | 4454 else |
| 4455 { | |
| 4043 | 4456 if (!vim_iswordc_buf(reginput[-1], reg_buf) |
| 4457 || (reginput[0] != NUL && vim_iswordc_buf(c, reg_buf))) | |
| 180 | 4458 status = RA_NOMATCH; |
| 7 | 4459 } |
| 4460 break; /* Matched with EOW */ | |
| 4461 | |
| 4462 case ANY: | |
| 4084 | 4463 /* ANY does not match new lines. */ |
| 7 | 4464 if (c == NUL) |
| 180 | 4465 status = RA_NOMATCH; |
| 4466 else | |
| 4467 ADVANCE_REGINPUT(); | |
| 7 | 4468 break; |
| 4469 | |
| 4470 case IDENT: | |
| 4471 if (!vim_isIDc(c)) | |
| 180 | 4472 status = RA_NOMATCH; |
| 4473 else | |
| 4474 ADVANCE_REGINPUT(); | |
| 7 | 4475 break; |
| 4476 | |
| 4477 case SIDENT: | |
| 4478 if (VIM_ISDIGIT(*reginput) || !vim_isIDc(c)) | |
| 180 | 4479 status = RA_NOMATCH; |
| 4480 else | |
| 4481 ADVANCE_REGINPUT(); | |
| 7 | 4482 break; |
| 4483 | |
| 4484 case KWORD: | |
| 4069 | 4485 if (!vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4486 status = RA_NOMATCH; |
| 4487 else | |
| 4488 ADVANCE_REGINPUT(); | |
| 7 | 4489 break; |
| 4490 | |
| 4491 case SKWORD: | |
| 4069 | 4492 if (VIM_ISDIGIT(*reginput) || !vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4493 status = RA_NOMATCH; |
| 4494 else | |
| 4495 ADVANCE_REGINPUT(); | |
| 7 | 4496 break; |
| 4497 | |
| 4498 case FNAME: | |
| 4499 if (!vim_isfilec(c)) | |
| 180 | 4500 status = RA_NOMATCH; |
| 4501 else | |
| 4502 ADVANCE_REGINPUT(); | |
| 7 | 4503 break; |
| 4504 | |
| 4505 case SFNAME: | |
| 4506 if (VIM_ISDIGIT(*reginput) || !vim_isfilec(c)) | |
| 180 | 4507 status = RA_NOMATCH; |
| 4508 else | |
| 4509 ADVANCE_REGINPUT(); | |
| 7 | 4510 break; |
| 4511 | |
| 4512 case PRINT: | |
|
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
4513 if (!vim_isprintc(PTR2CHAR(reginput))) |
| 180 | 4514 status = RA_NOMATCH; |
| 4515 else | |
| 4516 ADVANCE_REGINPUT(); | |
| 7 | 4517 break; |
| 4518 | |
| 4519 case SPRINT: | |
|
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
4520 if (VIM_ISDIGIT(*reginput) || !vim_isprintc(PTR2CHAR(reginput))) |
| 180 | 4521 status = RA_NOMATCH; |
| 4522 else | |
| 4523 ADVANCE_REGINPUT(); | |
| 7 | 4524 break; |
| 4525 | |
| 4526 case WHITE: | |
| 4527 if (!vim_iswhite(c)) | |
| 180 | 4528 status = RA_NOMATCH; |
| 4529 else | |
| 4530 ADVANCE_REGINPUT(); | |
| 7 | 4531 break; |
| 4532 | |
| 4533 case NWHITE: | |
| 4534 if (c == NUL || vim_iswhite(c)) | |
| 180 | 4535 status = RA_NOMATCH; |
| 4536 else | |
| 4537 ADVANCE_REGINPUT(); | |
| 7 | 4538 break; |
| 4539 | |
| 4540 case DIGIT: | |
| 4541 if (!ri_digit(c)) | |
| 180 | 4542 status = RA_NOMATCH; |
| 4543 else | |
| 4544 ADVANCE_REGINPUT(); | |
| 7 | 4545 break; |
| 4546 | |
| 4547 case NDIGIT: | |
| 4548 if (c == NUL || ri_digit(c)) | |
| 180 | 4549 status = RA_NOMATCH; |
| 4550 else | |
| 4551 ADVANCE_REGINPUT(); | |
| 7 | 4552 break; |
| 4553 | |
| 4554 case HEX: | |
| 4555 if (!ri_hex(c)) | |
| 180 | 4556 status = RA_NOMATCH; |
| 4557 else | |
| 4558 ADVANCE_REGINPUT(); | |
| 7 | 4559 break; |
| 4560 | |
| 4561 case NHEX: | |
| 4562 if (c == NUL || ri_hex(c)) | |
| 180 | 4563 status = RA_NOMATCH; |
| 4564 else | |
| 4565 ADVANCE_REGINPUT(); | |
| 7 | 4566 break; |
| 4567 | |
| 4568 case OCTAL: | |
| 4569 if (!ri_octal(c)) | |
| 180 | 4570 status = RA_NOMATCH; |
| 4571 else | |
| 4572 ADVANCE_REGINPUT(); | |
| 7 | 4573 break; |
| 4574 | |
| 4575 case NOCTAL: | |
| 4576 if (c == NUL || ri_octal(c)) | |
| 180 | 4577 status = RA_NOMATCH; |
| 4578 else | |
| 4579 ADVANCE_REGINPUT(); | |
| 7 | 4580 break; |
| 4581 | |
| 4582 case WORD: | |
| 4583 if (!ri_word(c)) | |
| 180 | 4584 status = RA_NOMATCH; |
| 4585 else | |
| 4586 ADVANCE_REGINPUT(); | |
| 7 | 4587 break; |
| 4588 | |
| 4589 case NWORD: | |
| 4590 if (c == NUL || ri_word(c)) | |
| 180 | 4591 status = RA_NOMATCH; |
| 4592 else | |
| 4593 ADVANCE_REGINPUT(); | |
| 7 | 4594 break; |
| 4595 | |
| 4596 case HEAD: | |
| 4597 if (!ri_head(c)) | |
| 180 | 4598 status = RA_NOMATCH; |
| 4599 else | |
| 4600 ADVANCE_REGINPUT(); | |
| 7 | 4601 break; |
| 4602 | |
| 4603 case NHEAD: | |
| 4604 if (c == NUL || ri_head(c)) | |
| 180 | 4605 status = RA_NOMATCH; |
| 4606 else | |
| 4607 ADVANCE_REGINPUT(); | |
| 7 | 4608 break; |
| 4609 | |
| 4610 case ALPHA: | |
| 4611 if (!ri_alpha(c)) | |
| 180 | 4612 status = RA_NOMATCH; |
| 4613 else | |
| 4614 ADVANCE_REGINPUT(); | |
| 7 | 4615 break; |
| 4616 | |
| 4617 case NALPHA: | |
| 4618 if (c == NUL || ri_alpha(c)) | |
| 180 | 4619 status = RA_NOMATCH; |
| 4620 else | |
| 4621 ADVANCE_REGINPUT(); | |
| 7 | 4622 break; |
| 4623 | |
| 4624 case LOWER: | |
| 4625 if (!ri_lower(c)) | |
| 180 | 4626 status = RA_NOMATCH; |
| 4627 else | |
| 4628 ADVANCE_REGINPUT(); | |
| 7 | 4629 break; |
| 4630 | |
| 4631 case NLOWER: | |
| 4632 if (c == NUL || ri_lower(c)) | |
| 180 | 4633 status = RA_NOMATCH; |
| 4634 else | |
| 4635 ADVANCE_REGINPUT(); | |
| 7 | 4636 break; |
| 4637 | |
| 4638 case UPPER: | |
| 4639 if (!ri_upper(c)) | |
| 180 | 4640 status = RA_NOMATCH; |
| 4641 else | |
| 4642 ADVANCE_REGINPUT(); | |
| 7 | 4643 break; |
| 4644 | |
| 4645 case NUPPER: | |
| 4646 if (c == NUL || ri_upper(c)) | |
| 180 | 4647 status = RA_NOMATCH; |
| 4648 else | |
| 4649 ADVANCE_REGINPUT(); | |
| 7 | 4650 break; |
| 4651 | |
| 4652 case EXACTLY: | |
| 4653 { | |
| 4654 int len; | |
| 4655 char_u *opnd; | |
| 4656 | |
| 4657 opnd = OPERAND(scan); | |
| 4658 /* Inline the first byte, for speed. */ | |
| 4659 if (*opnd != *reginput | |
| 4660 && (!ireg_ic || ( | |
| 4661 #ifdef FEAT_MBYTE | |
| 4662 !enc_utf8 && | |
| 4663 #endif | |
| 1347 | 4664 MB_TOLOWER(*opnd) != MB_TOLOWER(*reginput)))) |
| 180 | 4665 status = RA_NOMATCH; |
| 4666 else if (*opnd == NUL) | |
| 7 | 4667 { |
| 4668 /* match empty string always works; happens when "~" is | |
| 4669 * empty. */ | |
| 4670 } | |
| 5899 | 4671 else |
| 4672 { | |
| 4673 if (opnd[1] == NUL | |
| 7 | 4674 #ifdef FEAT_MBYTE |
| 4675 && !(enc_utf8 && ireg_ic) | |
| 4676 #endif | |
| 4677 ) | |
| 5899 | 4678 { |
| 4679 len = 1; /* matched a single byte above */ | |
| 4680 } | |
| 4681 else | |
| 4682 { | |
| 4683 /* Need to match first byte again for multi-byte. */ | |
| 4684 len = (int)STRLEN(opnd); | |
| 4685 if (cstrncmp(opnd, reginput, &len) != 0) | |
| 4686 status = RA_NOMATCH; | |
| 4687 } | |
| 7 | 4688 #ifdef FEAT_MBYTE |
| 5901 | 4689 /* Check for following composing character, unless %C |
| 4690 * follows (skips over all composing chars). */ | |
| 5899 | 4691 if (status != RA_NOMATCH |
| 4692 && enc_utf8 | |
| 4693 && UTF_COMPOSINGLIKE(reginput, reginput + len) | |
| 5901 | 4694 && !ireg_icombine |
| 4695 && OP(next) != RE_COMPOSING) | |
| 7 | 4696 { |
| 4697 /* raaron: This code makes a composing character get | |
| 4698 * ignored, which is the correct behavior (sometimes) | |
| 4699 * for voweled Hebrew texts. */ | |
| 5899 | 4700 status = RA_NOMATCH; |
| 7 | 4701 } |
| 180 | 4702 #endif |
| 5899 | 4703 if (status != RA_NOMATCH) |
| 180 | 4704 reginput += len; |
| 7 | 4705 } |
| 4706 } | |
| 4707 break; | |
| 4708 | |
| 4709 case ANYOF: | |
| 4710 case ANYBUT: | |
| 4711 if (c == NUL) | |
| 180 | 4712 status = RA_NOMATCH; |
| 4713 else if ((cstrchr(OPERAND(scan), c) == NULL) == (op == ANYOF)) | |
| 4714 status = RA_NOMATCH; | |
| 4715 else | |
| 4716 ADVANCE_REGINPUT(); | |
| 7 | 4717 break; |
| 4718 | |
| 4719 #ifdef FEAT_MBYTE | |
| 4720 case MULTIBYTECODE: | |
| 4721 if (has_mbyte) | |
| 4722 { | |
| 4723 int i, len; | |
| 4724 char_u *opnd; | |
| 944 | 4725 int opndc = 0, inpc; |
| 7 | 4726 |
| 4727 opnd = OPERAND(scan); | |
| 4728 /* Safety check (just in case 'encoding' was changed since | |
| 4729 * compiling the program). */ | |
| 474 | 4730 if ((len = (*mb_ptr2len)(opnd)) < 2) |
| 180 | 4731 { |
| 4732 status = RA_NOMATCH; | |
| 4733 break; | |
| 4734 } | |
| 714 | 4735 if (enc_utf8) |
| 4736 opndc = mb_ptr2char(opnd); | |
| 4737 if (enc_utf8 && utf_iscomposing(opndc)) | |
| 4738 { | |
| 4739 /* When only a composing char is given match at any | |
| 4740 * position where that composing char appears. */ | |
| 4741 status = RA_NOMATCH; | |
| 6723 | 4742 for (i = 0; reginput[i] != NUL; |
| 4743 i += utf_ptr2len(reginput + i)) | |
| 180 | 4744 { |
| 714 | 4745 inpc = mb_ptr2char(reginput + i); |
| 4746 if (!utf_iscomposing(inpc)) | |
| 4747 { | |
| 4748 if (i > 0) | |
| 4749 break; | |
| 4750 } | |
| 4751 else if (opndc == inpc) | |
| 4752 { | |
| 4753 /* Include all following composing chars. */ | |
| 4754 len = i + mb_ptr2len(reginput + i); | |
| 4755 status = RA_MATCH; | |
| 4756 break; | |
| 4757 } | |
| 180 | 4758 } |
| 714 | 4759 } |
| 4760 else | |
| 4761 for (i = 0; i < len; ++i) | |
| 4762 if (opnd[i] != reginput[i]) | |
| 4763 { | |
| 4764 status = RA_NOMATCH; | |
| 4765 break; | |
| 4766 } | |
| 7 | 4767 reginput += len; |
| 4768 } | |
| 4769 else | |
| 180 | 4770 status = RA_NOMATCH; |
| 7 | 4771 break; |
| 4772 #endif | |
| 5901 | 4773 case RE_COMPOSING: |
| 4774 #ifdef FEAT_MBYTE | |
| 4775 if (enc_utf8) | |
| 4776 { | |
| 4777 /* Skip composing characters. */ | |
| 4778 while (utf_iscomposing(utf_ptr2char(reginput))) | |
| 4779 mb_cptr_adv(reginput); | |
| 4780 } | |
| 4781 #endif | |
| 4782 break; | |
| 7 | 4783 |
| 4784 case NOTHING: | |
| 4785 break; | |
| 4786 | |
| 4787 case BACK: | |
| 233 | 4788 { |
| 4789 int i; | |
| 4790 backpos_T *bp; | |
| 4791 | |
| 4792 /* | |
| 4793 * When we run into BACK we need to check if we don't keep | |
| 4794 * looping without matching any input. The second and later | |
| 4795 * times a BACK is encountered it fails if the input is still | |
| 4796 * at the same position as the previous time. | |
| 4797 * The positions are stored in "backpos" and found by the | |
| 4798 * current value of "scan", the position in the RE program. | |
| 4799 */ | |
| 4800 bp = (backpos_T *)backpos.ga_data; | |
| 4801 for (i = 0; i < backpos.ga_len; ++i) | |
| 4802 if (bp[i].bp_scan == scan) | |
| 4803 break; | |
| 4804 if (i == backpos.ga_len) | |
| 4805 { | |
| 4806 /* First time at this BACK, make room to store the pos. */ | |
| 4807 if (ga_grow(&backpos, 1) == FAIL) | |
| 4808 status = RA_FAIL; | |
| 4809 else | |
| 4810 { | |
| 4811 /* get "ga_data" again, it may have changed */ | |
| 4812 bp = (backpos_T *)backpos.ga_data; | |
| 4813 bp[i].bp_scan = scan; | |
| 4814 ++backpos.ga_len; | |
| 4815 } | |
| 4816 } | |
| 4817 else if (reg_save_equal(&bp[i].bp_pos)) | |
| 4818 /* Still at same position as last time, fail. */ | |
| 4819 status = RA_NOMATCH; | |
| 4820 | |
| 4821 if (status != RA_FAIL && status != RA_NOMATCH) | |
| 4822 reg_save(&bp[i].bp_pos, &backpos); | |
| 4823 } | |
| 179 | 4824 break; |
| 4825 | |
| 7 | 4826 case MOPEN + 0: /* Match start: \zs */ |
| 4827 case MOPEN + 1: /* \( */ | |
| 4828 case MOPEN + 2: | |
| 4829 case MOPEN + 3: | |
| 4830 case MOPEN + 4: | |
| 4831 case MOPEN + 5: | |
| 4832 case MOPEN + 6: | |
| 4833 case MOPEN + 7: | |
| 4834 case MOPEN + 8: | |
| 4835 case MOPEN + 9: | |
| 4836 { | |
| 4837 no = op - MOPEN; | |
| 4838 cleanup_subexpr(); | |
| 270 | 4839 rp = regstack_push(RS_MOPEN, scan); |
| 180 | 4840 if (rp == NULL) |
| 4841 status = RA_FAIL; | |
| 4842 else | |
| 4843 { | |
| 4844 rp->rs_no = no; | |
| 4845 save_se(&rp->rs_un.sesave, ®_startpos[no], | |
| 4846 ®_startp[no]); | |
| 4847 /* We simply continue and handle the result when done. */ | |
| 4848 } | |
| 7 | 4849 } |
| 180 | 4850 break; |
| 7 | 4851 |
| 4852 case NOPEN: /* \%( */ | |
| 4853 case NCLOSE: /* \) after \%( */ | |
| 270 | 4854 if (regstack_push(RS_NOPEN, scan) == NULL) |
| 180 | 4855 status = RA_FAIL; |
| 4856 /* We simply continue and handle the result when done. */ | |
| 4857 break; | |
| 7 | 4858 |
| 4859 #ifdef FEAT_SYN_HL | |
| 4860 case ZOPEN + 1: | |
| 4861 case ZOPEN + 2: | |
| 4862 case ZOPEN + 3: | |
| 4863 case ZOPEN + 4: | |
| 4864 case ZOPEN + 5: | |
| 4865 case ZOPEN + 6: | |
| 4866 case ZOPEN + 7: | |
| 4867 case ZOPEN + 8: | |
| 4868 case ZOPEN + 9: | |
| 4869 { | |
| 4870 no = op - ZOPEN; | |
| 4871 cleanup_zsubexpr(); | |
| 270 | 4872 rp = regstack_push(RS_ZOPEN, scan); |
| 180 | 4873 if (rp == NULL) |
| 4874 status = RA_FAIL; | |
| 4875 else | |
| 4876 { | |
| 4877 rp->rs_no = no; | |
| 4878 save_se(&rp->rs_un.sesave, ®_startzpos[no], | |
| 4879 ®_startzp[no]); | |
| 4880 /* We simply continue and handle the result when done. */ | |
| 4881 } | |
| 7 | 4882 } |
| 180 | 4883 break; |
| 7 | 4884 #endif |
| 4885 | |
| 4886 case MCLOSE + 0: /* Match end: \ze */ | |
| 4887 case MCLOSE + 1: /* \) */ | |
| 4888 case MCLOSE + 2: | |
| 4889 case MCLOSE + 3: | |
| 4890 case MCLOSE + 4: | |
| 4891 case MCLOSE + 5: | |
| 4892 case MCLOSE + 6: | |
| 4893 case MCLOSE + 7: | |
| 4894 case MCLOSE + 8: | |
| 4895 case MCLOSE + 9: | |
| 4896 { | |
| 4897 no = op - MCLOSE; | |
| 4898 cleanup_subexpr(); | |
| 270 | 4899 rp = regstack_push(RS_MCLOSE, scan); |
| 180 | 4900 if (rp == NULL) |
| 4901 status = RA_FAIL; | |
| 4902 else | |
| 4903 { | |
| 4904 rp->rs_no = no; | |
| 4905 save_se(&rp->rs_un.sesave, ®_endpos[no], ®_endp[no]); | |
| 4906 /* We simply continue and handle the result when done. */ | |
| 4907 } | |
| 7 | 4908 } |
| 180 | 4909 break; |
| 7 | 4910 |
| 4911 #ifdef FEAT_SYN_HL | |
| 4912 case ZCLOSE + 1: /* \) after \z( */ | |
| 4913 case ZCLOSE + 2: | |
| 4914 case ZCLOSE + 3: | |
| 4915 case ZCLOSE + 4: | |
| 4916 case ZCLOSE + 5: | |
| 4917 case ZCLOSE + 6: | |
| 4918 case ZCLOSE + 7: | |
| 4919 case ZCLOSE + 8: | |
| 4920 case ZCLOSE + 9: | |
| 4921 { | |
| 4922 no = op - ZCLOSE; | |
| 4923 cleanup_zsubexpr(); | |
| 270 | 4924 rp = regstack_push(RS_ZCLOSE, scan); |
| 180 | 4925 if (rp == NULL) |
| 4926 status = RA_FAIL; | |
| 4927 else | |
| 4928 { | |
| 4929 rp->rs_no = no; | |
| 4930 save_se(&rp->rs_un.sesave, ®_endzpos[no], | |
| 4931 ®_endzp[no]); | |
| 4932 /* We simply continue and handle the result when done. */ | |
| 4933 } | |
| 7 | 4934 } |
| 180 | 4935 break; |
| 7 | 4936 #endif |
| 4937 | |
| 4938 case BACKREF + 1: | |
| 4939 case BACKREF + 2: | |
| 4940 case BACKREF + 3: | |
| 4941 case BACKREF + 4: | |
| 4942 case BACKREF + 5: | |
| 4943 case BACKREF + 6: | |
| 4944 case BACKREF + 7: | |
| 4945 case BACKREF + 8: | |
| 4946 case BACKREF + 9: | |
| 4947 { | |
| 4948 int len; | |
| 4949 | |
| 4950 no = op - BACKREF; | |
| 4951 cleanup_subexpr(); | |
| 4952 if (!REG_MULTI) /* Single-line regexp */ | |
| 4953 { | |
| 1815 | 4954 if (reg_startp[no] == NULL || reg_endp[no] == NULL) |
| 7 | 4955 { |
| 4956 /* Backref was not set: Match an empty string. */ | |
| 4957 len = 0; | |
| 4958 } | |
| 4959 else | |
| 4960 { | |
| 4961 /* Compare current input with back-ref in the same | |
| 4962 * line. */ | |
| 4963 len = (int)(reg_endp[no] - reg_startp[no]); | |
| 4964 if (cstrncmp(reg_startp[no], reginput, &len) != 0) | |
| 180 | 4965 status = RA_NOMATCH; |
| 7 | 4966 } |
| 4967 } | |
| 4968 else /* Multi-line regexp */ | |
| 4969 { | |
| 1815 | 4970 if (reg_startpos[no].lnum < 0 || reg_endpos[no].lnum < 0) |
| 7 | 4971 { |
| 4972 /* Backref was not set: Match an empty string. */ | |
| 4973 len = 0; | |
| 4974 } | |
| 4975 else | |
| 4976 { | |
| 4977 if (reg_startpos[no].lnum == reglnum | |
| 4978 && reg_endpos[no].lnum == reglnum) | |
| 4979 { | |
| 4980 /* Compare back-ref within the current line. */ | |
| 4981 len = reg_endpos[no].col - reg_startpos[no].col; | |
| 4982 if (cstrncmp(regline + reg_startpos[no].col, | |
| 4983 reginput, &len) != 0) | |
| 180 | 4984 status = RA_NOMATCH; |
| 7 | 4985 } |
| 4986 else | |
| 4987 { | |
| 4988 /* Messy situation: Need to compare between two | |
| 4989 * lines. */ | |
|
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
4990 int r = match_with_backref( |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4991 reg_startpos[no].lnum, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4992 reg_startpos[no].col, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4993 reg_endpos[no].lnum, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
4994 reg_endpos[no].col, |
|
4899
4837fd61be52
updated for version 7.3.1195
Bram Moolenaar <bram@vim.org>
parents:
4891
diff
changeset
|
4995 &len); |
|
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
4996 |
|
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
4997 if (r != RA_MATCH) |
|
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
4998 status = r; |
| 7 | 4999 } |
| 5000 } | |
| 5001 } | |
| 5002 | |
| 5003 /* Matched the backref, skip over it. */ | |
| 5004 reginput += len; | |
| 5005 } | |
| 5006 break; | |
| 5007 | |
| 5008 #ifdef FEAT_SYN_HL | |
| 5009 case ZREF + 1: | |
| 5010 case ZREF + 2: | |
| 5011 case ZREF + 3: | |
| 5012 case ZREF + 4: | |
| 5013 case ZREF + 5: | |
| 5014 case ZREF + 6: | |
| 5015 case ZREF + 7: | |
| 5016 case ZREF + 8: | |
| 5017 case ZREF + 9: | |
| 5018 { | |
| 5019 int len; | |
| 5020 | |
| 5021 cleanup_zsubexpr(); | |
| 5022 no = op - ZREF; | |
| 5023 if (re_extmatch_in != NULL | |
| 5024 && re_extmatch_in->matches[no] != NULL) | |
| 5025 { | |
| 5026 len = (int)STRLEN(re_extmatch_in->matches[no]); | |
| 5027 if (cstrncmp(re_extmatch_in->matches[no], | |
| 5028 reginput, &len) != 0) | |
| 180 | 5029 status = RA_NOMATCH; |
| 5030 else | |
| 5031 reginput += len; | |
| 7 | 5032 } |
| 5033 else | |
| 5034 { | |
| 5035 /* Backref was not set: Match an empty string. */ | |
| 5036 } | |
| 5037 } | |
| 5038 break; | |
| 5039 #endif | |
| 5040 | |
| 5041 case BRANCH: | |
| 5042 { | |
| 5043 if (OP(next) != BRANCH) /* No choice. */ | |
| 5044 next = OPERAND(scan); /* Avoid recursion. */ | |
| 5045 else | |
| 5046 { | |
| 270 | 5047 rp = regstack_push(RS_BRANCH, scan); |
| 180 | 5048 if (rp == NULL) |
| 5049 status = RA_FAIL; | |
| 5050 else | |
| 5051 status = RA_BREAK; /* rest is below */ | |
| 7 | 5052 } |
| 5053 } | |
| 5054 break; | |
| 5055 | |
| 5056 case BRACE_LIMITS: | |
| 5057 { | |
| 5058 if (OP(next) == BRACE_SIMPLE) | |
| 5059 { | |
| 5060 bl_minval = OPERAND_MIN(scan); | |
| 5061 bl_maxval = OPERAND_MAX(scan); | |
| 5062 } | |
| 5063 else if (OP(next) >= BRACE_COMPLEX | |
| 5064 && OP(next) < BRACE_COMPLEX + 10) | |
| 5065 { | |
| 5066 no = OP(next) - BRACE_COMPLEX; | |
| 5067 brace_min[no] = OPERAND_MIN(scan); | |
| 5068 brace_max[no] = OPERAND_MAX(scan); | |
| 5069 brace_count[no] = 0; | |
| 5070 } | |
| 5071 else | |
| 5072 { | |
| 5073 EMSG(_(e_internal)); /* Shouldn't happen */ | |
| 180 | 5074 status = RA_FAIL; |
| 7 | 5075 } |
| 5076 } | |
| 5077 break; | |
| 5078 | |
| 5079 case BRACE_COMPLEX + 0: | |
| 5080 case BRACE_COMPLEX + 1: | |
| 5081 case BRACE_COMPLEX + 2: | |
| 5082 case BRACE_COMPLEX + 3: | |
| 5083 case BRACE_COMPLEX + 4: | |
| 5084 case BRACE_COMPLEX + 5: | |
| 5085 case BRACE_COMPLEX + 6: | |
| 5086 case BRACE_COMPLEX + 7: | |
| 5087 case BRACE_COMPLEX + 8: | |
| 5088 case BRACE_COMPLEX + 9: | |
| 5089 { | |
| 5090 no = op - BRACE_COMPLEX; | |
| 5091 ++brace_count[no]; | |
| 5092 | |
| 5093 /* If not matched enough times yet, try one more */ | |
| 5094 if (brace_count[no] <= (brace_min[no] <= brace_max[no] | |
| 180 | 5095 ? brace_min[no] : brace_max[no])) |
| 7 | 5096 { |
| 270 | 5097 rp = regstack_push(RS_BRCPLX_MORE, scan); |
| 180 | 5098 if (rp == NULL) |
| 5099 status = RA_FAIL; | |
| 5100 else | |
| 5101 { | |
| 5102 rp->rs_no = no; | |
| 233 | 5103 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5104 next = OPERAND(scan); |
| 5105 /* We continue and handle the result when done. */ | |
| 5106 } | |
| 5107 break; | |
| 7 | 5108 } |
| 5109 | |
| 5110 /* If matched enough times, may try matching some more */ | |
| 5111 if (brace_min[no] <= brace_max[no]) | |
| 5112 { | |
| 5113 /* Range is the normal way around, use longest match */ | |
| 5114 if (brace_count[no] <= brace_max[no]) | |
| 5115 { | |
| 270 | 5116 rp = regstack_push(RS_BRCPLX_LONG, scan); |
| 180 | 5117 if (rp == NULL) |
| 5118 status = RA_FAIL; | |
| 5119 else | |
| 5120 { | |
| 5121 rp->rs_no = no; | |
| 233 | 5122 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5123 next = OPERAND(scan); |
| 5124 /* We continue and handle the result when done. */ | |
| 5125 } | |
| 7 | 5126 } |
| 5127 } | |
| 5128 else | |
| 5129 { | |
| 5130 /* Range is backwards, use shortest match first */ | |
| 5131 if (brace_count[no] <= brace_min[no]) | |
| 5132 { | |
| 270 | 5133 rp = regstack_push(RS_BRCPLX_SHORT, scan); |
| 180 | 5134 if (rp == NULL) |
| 5135 status = RA_FAIL; | |
| 5136 else | |
| 5137 { | |
| 233 | 5138 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5139 /* We continue and handle the result when done. */ |
| 5140 } | |
| 7 | 5141 } |
| 5142 } | |
| 5143 } | |
| 5144 break; | |
| 5145 | |
| 5146 case BRACE_SIMPLE: | |
| 5147 case STAR: | |
| 5148 case PLUS: | |
| 5149 { | |
| 180 | 5150 regstar_T rst; |
| 7 | 5151 |
| 5152 /* | |
| 5153 * Lookahead to avoid useless match attempts when we know | |
| 5154 * what character comes next. | |
| 5155 */ | |
| 5156 if (OP(next) == EXACTLY) | |
| 5157 { | |
| 180 | 5158 rst.nextb = *OPERAND(next); |
| 7 | 5159 if (ireg_ic) |
| 5160 { | |
| 1347 | 5161 if (MB_ISUPPER(rst.nextb)) |
| 5162 rst.nextb_ic = MB_TOLOWER(rst.nextb); | |
| 7 | 5163 else |
| 1347 | 5164 rst.nextb_ic = MB_TOUPPER(rst.nextb); |
| 7 | 5165 } |
| 5166 else | |
| 180 | 5167 rst.nextb_ic = rst.nextb; |
| 7 | 5168 } |
| 5169 else | |
| 5170 { | |
| 180 | 5171 rst.nextb = NUL; |
| 5172 rst.nextb_ic = NUL; | |
| 7 | 5173 } |
| 5174 if (op != BRACE_SIMPLE) | |
| 5175 { | |
| 180 | 5176 rst.minval = (op == STAR) ? 0 : 1; |
| 5177 rst.maxval = MAX_LIMIT; | |
| 7 | 5178 } |
| 5179 else | |
| 5180 { | |
| 180 | 5181 rst.minval = bl_minval; |
| 5182 rst.maxval = bl_maxval; | |
| 7 | 5183 } |
| 5184 | |
| 5185 /* | |
| 5186 * When maxval > minval, try matching as much as possible, up | |
| 5187 * to maxval. When maxval < minval, try matching at least the | |
| 5188 * minimal number (since the range is backwards, that's also | |
| 5189 * maxval!). | |
| 5190 */ | |
| 180 | 5191 rst.count = regrepeat(OPERAND(scan), rst.maxval); |
| 7 | 5192 if (got_int) |
| 180 | 5193 { |
| 5194 status = RA_FAIL; | |
| 5195 break; | |
| 5196 } | |
| 5197 if (rst.minval <= rst.maxval | |
| 5198 ? rst.count >= rst.minval : rst.count >= rst.maxval) | |
| 7 | 5199 { |
| 180 | 5200 /* It could match. Prepare for trying to match what |
| 5201 * follows. The code is below. Parameters are stored in | |
| 5202 * a regstar_T on the regstack. */ | |
| 212 | 5203 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5204 { |
| 5205 EMSG(_(e_maxmempat)); | |
| 5206 status = RA_FAIL; | |
| 5207 } | |
| 5208 else if (ga_grow(®stack, sizeof(regstar_T)) == FAIL) | |
| 180 | 5209 status = RA_FAIL; |
| 5210 else | |
| 7 | 5211 { |
| 180 | 5212 regstack.ga_len += sizeof(regstar_T); |
| 270 | 5213 rp = regstack_push(rst.minval <= rst.maxval |
| 233 | 5214 ? RS_STAR_LONG : RS_STAR_SHORT, scan); |
| 180 | 5215 if (rp == NULL) |
| 5216 status = RA_FAIL; | |
| 5217 else | |
| 7 | 5218 { |
| 180 | 5219 *(((regstar_T *)rp) - 1) = rst; |
| 5220 status = RA_BREAK; /* skip the restore bits */ | |
| 7 | 5221 } |
| 5222 } | |
| 5223 } | |
| 5224 else | |
| 180 | 5225 status = RA_NOMATCH; |
| 5226 | |
| 7 | 5227 } |
| 180 | 5228 break; |
| 7 | 5229 |
| 5230 case NOMATCH: | |
| 5231 case MATCH: | |
| 5232 case SUBPAT: | |
| 270 | 5233 rp = regstack_push(RS_NOMATCH, scan); |
| 180 | 5234 if (rp == NULL) |
| 5235 status = RA_FAIL; | |
| 5236 else | |
| 7 | 5237 { |
| 180 | 5238 rp->rs_no = op; |
| 233 | 5239 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5240 next = OPERAND(scan); |
| 5241 /* We continue and handle the result when done. */ | |
| 7 | 5242 } |
| 5243 break; | |
| 5244 | |
| 5245 case BEHIND: | |
| 5246 case NOBEHIND: | |
| 180 | 5247 /* Need a bit of room to store extra positions. */ |
| 212 | 5248 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5249 { |
| 5250 EMSG(_(e_maxmempat)); | |
| 5251 status = RA_FAIL; | |
| 5252 } | |
| 5253 else if (ga_grow(®stack, sizeof(regbehind_T)) == FAIL) | |
| 180 | 5254 status = RA_FAIL; |
| 5255 else | |
| 7 | 5256 { |
| 180 | 5257 regstack.ga_len += sizeof(regbehind_T); |
| 270 | 5258 rp = regstack_push(RS_BEHIND1, scan); |
| 180 | 5259 if (rp == NULL) |
| 5260 status = RA_FAIL; | |
| 5261 else | |
| 7 | 5262 { |
| 1579 | 5263 /* Need to save the subexpr to be able to restore them |
| 5264 * when there is a match but we don't use it. */ | |
| 5265 save_subexpr(((regbehind_T *)rp) - 1); | |
| 5266 | |
| 180 | 5267 rp->rs_no = op; |
| 233 | 5268 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5269 /* First try if what follows matches. If it does then we |
| 5270 * check the behind match by looping. */ | |
| 7 | 5271 } |
| 5272 } | |
| 180 | 5273 break; |
| 7 | 5274 |
| 5275 case BHPOS: | |
| 5276 if (REG_MULTI) | |
| 5277 { | |
| 5278 if (behind_pos.rs_u.pos.col != (colnr_T)(reginput - regline) | |
| 5279 || behind_pos.rs_u.pos.lnum != reglnum) | |
| 180 | 5280 status = RA_NOMATCH; |
| 7 | 5281 } |
| 5282 else if (behind_pos.rs_u.ptr != reginput) | |
| 180 | 5283 status = RA_NOMATCH; |
| 7 | 5284 break; |
| 5285 | |
| 5286 case NEWL: | |
| 1018 | 5287 if ((c != NUL || !REG_MULTI || reglnum > reg_maxline |
| 5288 || reg_line_lbr) && (c != '\n' || !reg_line_lbr)) | |
| 180 | 5289 status = RA_NOMATCH; |
| 5290 else if (reg_line_lbr) | |
| 7 | 5291 ADVANCE_REGINPUT(); |
| 5292 else | |
| 5293 reg_nextline(); | |
| 5294 break; | |
| 5295 | |
| 5296 case END: | |
| 180 | 5297 status = RA_MATCH; /* Success! */ |
| 5298 break; | |
| 7 | 5299 |
| 5300 default: | |
| 5301 EMSG(_(e_re_corr)); | |
| 5302 #ifdef DEBUG | |
| 5303 printf("Illegal op code %d\n", op); | |
| 5304 #endif | |
| 180 | 5305 status = RA_FAIL; |
| 5306 break; | |
| 7 | 5307 } |
| 5308 } | |
| 5309 | |
| 180 | 5310 /* If we can't continue sequentially, break the inner loop. */ |
| 5311 if (status != RA_CONT) | |
| 5312 break; | |
| 5313 | |
| 5314 /* Continue in inner loop, advance to next item. */ | |
| 7 | 5315 scan = next; |
| 180 | 5316 |
| 5317 } /* end of inner loop */ | |
| 7 | 5318 |
| 5319 /* | |
| 180 | 5320 * If there is something on the regstack execute the code for the state. |
| 233 | 5321 * If the state is popped then loop and use the older state. |
| 7 | 5322 */ |
| 180 | 5323 while (regstack.ga_len > 0 && status != RA_FAIL) |
| 5324 { | |
| 5325 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; | |
| 5326 switch (rp->rs_state) | |
| 5327 { | |
| 5328 case RS_NOPEN: | |
| 5329 /* Result is passed on as-is, simply pop the state. */ | |
| 270 | 5330 regstack_pop(&scan); |
| 180 | 5331 break; |
| 5332 | |
| 5333 case RS_MOPEN: | |
| 5334 /* Pop the state. Restore pointers when there is no match. */ | |
| 5335 if (status == RA_NOMATCH) | |
| 5336 restore_se(&rp->rs_un.sesave, ®_startpos[rp->rs_no], | |
| 5337 ®_startp[rp->rs_no]); | |
| 270 | 5338 regstack_pop(&scan); |
| 180 | 5339 break; |
| 5340 | |
| 5341 #ifdef FEAT_SYN_HL | |
| 5342 case RS_ZOPEN: | |
| 5343 /* Pop the state. Restore pointers when there is no match. */ | |
| 5344 if (status == RA_NOMATCH) | |
| 5345 restore_se(&rp->rs_un.sesave, ®_startzpos[rp->rs_no], | |
| 5346 ®_startzp[rp->rs_no]); | |
| 270 | 5347 regstack_pop(&scan); |
| 180 | 5348 break; |
| 5349 #endif | |
| 5350 | |
| 5351 case RS_MCLOSE: | |
| 5352 /* Pop the state. Restore pointers when there is no match. */ | |
| 5353 if (status == RA_NOMATCH) | |
| 5354 restore_se(&rp->rs_un.sesave, ®_endpos[rp->rs_no], | |
| 5355 ®_endp[rp->rs_no]); | |
| 270 | 5356 regstack_pop(&scan); |
| 180 | 5357 break; |
| 5358 | |
| 5359 #ifdef FEAT_SYN_HL | |
| 5360 case RS_ZCLOSE: | |
| 5361 /* Pop the state. Restore pointers when there is no match. */ | |
| 5362 if (status == RA_NOMATCH) | |
| 5363 restore_se(&rp->rs_un.sesave, ®_endzpos[rp->rs_no], | |
| 5364 ®_endzp[rp->rs_no]); | |
| 270 | 5365 regstack_pop(&scan); |
| 180 | 5366 break; |
| 5367 #endif | |
| 5368 | |
| 5369 case RS_BRANCH: | |
| 5370 if (status == RA_MATCH) | |
| 5371 /* this branch matched, use it */ | |
| 270 | 5372 regstack_pop(&scan); |
| 180 | 5373 else |
| 5374 { | |
| 5375 if (status != RA_BREAK) | |
| 5376 { | |
| 5377 /* After a non-matching branch: try next one. */ | |
| 233 | 5378 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5379 scan = rp->rs_scan; |
| 5380 } | |
| 5381 if (scan == NULL || OP(scan) != BRANCH) | |
| 5382 { | |
| 5383 /* no more branches, didn't find a match */ | |
| 5384 status = RA_NOMATCH; | |
| 270 | 5385 regstack_pop(&scan); |
| 180 | 5386 } |
| 5387 else | |
| 5388 { | |
| 5389 /* Prepare to try a branch. */ | |
| 5390 rp->rs_scan = regnext(scan); | |
| 233 | 5391 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5392 scan = OPERAND(scan); |
| 5393 } | |
| 5394 } | |
| 5395 break; | |
| 5396 | |
| 5397 case RS_BRCPLX_MORE: | |
| 5398 /* Pop the state. Restore pointers when there is no match. */ | |
| 5399 if (status == RA_NOMATCH) | |
| 5400 { | |
| 233 | 5401 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5402 --brace_count[rp->rs_no]; /* decrement match count */ |
| 5403 } | |
| 270 | 5404 regstack_pop(&scan); |
| 180 | 5405 break; |
| 5406 | |
| 5407 case RS_BRCPLX_LONG: | |
| 5408 /* Pop the state. Restore pointers when there is no match. */ | |
| 5409 if (status == RA_NOMATCH) | |
| 5410 { | |
| 5411 /* There was no match, but we did find enough matches. */ | |
| 233 | 5412 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5413 --brace_count[rp->rs_no]; |
| 5414 /* continue with the items after "\{}" */ | |
| 5415 status = RA_CONT; | |
| 5416 } | |
| 270 | 5417 regstack_pop(&scan); |
| 180 | 5418 if (status == RA_CONT) |
| 5419 scan = regnext(scan); | |
| 5420 break; | |
| 5421 | |
| 5422 case RS_BRCPLX_SHORT: | |
| 5423 /* Pop the state. Restore pointers when there is no match. */ | |
| 5424 if (status == RA_NOMATCH) | |
| 5425 /* There was no match, try to match one more item. */ | |
| 233 | 5426 reg_restore(&rp->rs_un.regsave, &backpos); |
| 270 | 5427 regstack_pop(&scan); |
| 180 | 5428 if (status == RA_NOMATCH) |
| 5429 { | |
| 5430 scan = OPERAND(scan); | |
| 5431 status = RA_CONT; | |
| 5432 } | |
| 5433 break; | |
| 5434 | |
| 5435 case RS_NOMATCH: | |
| 5436 /* Pop the state. If the operand matches for NOMATCH or | |
| 5437 * doesn't match for MATCH/SUBPAT, we fail. Otherwise backup, | |
| 5438 * except for SUBPAT, and continue with the next item. */ | |
| 5439 if (status == (rp->rs_no == NOMATCH ? RA_MATCH : RA_NOMATCH)) | |
| 5440 status = RA_NOMATCH; | |
| 5441 else | |
| 5442 { | |
| 5443 status = RA_CONT; | |
| 233 | 5444 if (rp->rs_no != SUBPAT) /* zero-width */ |
| 5445 reg_restore(&rp->rs_un.regsave, &backpos); | |
| 180 | 5446 } |
| 270 | 5447 regstack_pop(&scan); |
| 180 | 5448 if (status == RA_CONT) |
| 5449 scan = regnext(scan); | |
| 5450 break; | |
| 5451 | |
| 5452 case RS_BEHIND1: | |
| 5453 if (status == RA_NOMATCH) | |
| 5454 { | |
| 270 | 5455 regstack_pop(&scan); |
| 180 | 5456 regstack.ga_len -= sizeof(regbehind_T); |
| 5457 } | |
| 5458 else | |
| 5459 { | |
| 5460 /* The stuff after BEHIND/NOBEHIND matches. Now try if | |
| 5461 * the behind part does (not) match before the current | |
| 5462 * position in the input. This must be done at every | |
| 5463 * position in the input and checking if the match ends at | |
| 5464 * the current position. */ | |
| 5465 | |
| 5466 /* save the position after the found match for next */ | |
| 233 | 5467 reg_save(&(((regbehind_T *)rp) - 1)->save_after, &backpos); |
| 180 | 5468 |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5469 /* Start looking for a match with operand at the current |
| 1209 | 5470 * position. Go back one character until we find the |
| 180 | 5471 * result, hitting the start of the line or the previous |
| 5472 * line (for multi-line matching). | |
| 5473 * Set behind_pos to where the match should end, BHPOS | |
| 5474 * will match it. Save the current value. */ | |
| 5475 (((regbehind_T *)rp) - 1)->save_behind = behind_pos; | |
| 5476 behind_pos = rp->rs_un.regsave; | |
| 5477 | |
| 5478 rp->rs_state = RS_BEHIND2; | |
| 5479 | |
| 233 | 5480 reg_restore(&rp->rs_un.regsave, &backpos); |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5481 scan = OPERAND(rp->rs_scan) + 4; |
| 180 | 5482 } |
| 5483 break; | |
| 5484 | |
| 5485 case RS_BEHIND2: | |
| 5486 /* | |
| 5487 * Looping for BEHIND / NOBEHIND match. | |
| 5488 */ | |
| 5489 if (status == RA_MATCH && reg_save_equal(&behind_pos)) | |
| 5490 { | |
| 5491 /* found a match that ends where "next" started */ | |
| 5492 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5493 if (rp->rs_no == BEHIND) | |
| 233 | 5494 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5495 &backpos); | |
| 180 | 5496 else |
| 1579 | 5497 { |
| 5498 /* But we didn't want a match. Need to restore the | |
| 5499 * subexpr, because what follows matched, so they have | |
| 5500 * been set. */ | |
| 180 | 5501 status = RA_NOMATCH; |
| 1579 | 5502 restore_subexpr(((regbehind_T *)rp) - 1); |
| 5503 } | |
| 270 | 5504 regstack_pop(&scan); |
| 180 | 5505 regstack.ga_len -= sizeof(regbehind_T); |
| 5506 } | |
| 5507 else | |
| 5508 { | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5509 long limit; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5510 |
| 1579 | 5511 /* No match or a match that doesn't end where we want it: Go |
| 5512 * back one character. May go to previous line once. */ | |
| 180 | 5513 no = OK; |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5514 limit = OPERAND_MIN(rp->rs_scan); |
| 180 | 5515 if (REG_MULTI) |
| 5516 { | |
|
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5517 if (limit > 0 |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5518 && ((rp->rs_un.regsave.rs_u.pos.lnum |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5519 < behind_pos.rs_u.pos.lnum |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5520 ? (colnr_T)STRLEN(regline) |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5521 : behind_pos.rs_u.pos.col) |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5522 - 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
|
5523 no = FAIL; |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5524 else if (rp->rs_un.regsave.rs_u.pos.col == 0) |
| 180 | 5525 { |
| 5526 if (rp->rs_un.regsave.rs_u.pos.lnum | |
| 5527 < behind_pos.rs_u.pos.lnum | |
| 5528 || reg_getline( | |
| 5529 --rp->rs_un.regsave.rs_u.pos.lnum) | |
| 5530 == NULL) | |
| 5531 no = FAIL; | |
| 5532 else | |
| 5533 { | |
| 233 | 5534 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5535 rp->rs_un.regsave.rs_u.pos.col = |
| 5536 (colnr_T)STRLEN(regline); | |
| 5537 } | |
| 5538 } | |
| 5539 else | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5540 { |
| 4176 | 5541 #ifdef FEAT_MBYTE |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5542 if (has_mbyte) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5543 rp->rs_un.regsave.rs_u.pos.col -= |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5544 (*mb_head_off)(regline, regline |
| 4176 | 5545 + 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
|
5546 else |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5547 #endif |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5548 --rp->rs_un.regsave.rs_u.pos.col; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5549 } |
| 180 | 5550 } |
| 5551 else | |
| 5552 { | |
| 5553 if (rp->rs_un.regsave.rs_u.ptr == regline) | |
| 5554 no = FAIL; | |
| 5555 else | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5556 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5557 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
|
5558 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
|
5559 - rp->rs_un.regsave.rs_u.ptr) > limit) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5560 no = FAIL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5561 } |
| 180 | 5562 } |
| 5563 if (no == OK) | |
| 5564 { | |
| 5565 /* Advanced, prepare for finding match again. */ | |
| 233 | 5566 reg_restore(&rp->rs_un.regsave, &backpos); |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5567 scan = OPERAND(rp->rs_scan) + 4; |
| 1579 | 5568 if (status == RA_MATCH) |
| 5569 { | |
| 5570 /* We did match, so subexpr may have been changed, | |
| 5571 * need to restore them for the next try. */ | |
| 5572 status = RA_NOMATCH; | |
| 5573 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5574 } | |
| 180 | 5575 } |
| 5576 else | |
| 5577 { | |
| 5578 /* Can't advance. For NOBEHIND that's a match. */ | |
| 5579 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5580 if (rp->rs_no == NOBEHIND) | |
| 5581 { | |
| 233 | 5582 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5583 &backpos); | |
| 180 | 5584 status = RA_MATCH; |
| 5585 } | |
| 5586 else | |
| 1579 | 5587 { |
| 5588 /* We do want a proper match. Need to restore the | |
| 5589 * subexpr if we had a match, because they may have | |
| 5590 * been set. */ | |
| 5591 if (status == RA_MATCH) | |
| 5592 { | |
| 5593 status = RA_NOMATCH; | |
| 5594 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5595 } | |
| 5596 } | |
| 270 | 5597 regstack_pop(&scan); |
| 180 | 5598 regstack.ga_len -= sizeof(regbehind_T); |
| 5599 } | |
| 5600 } | |
| 5601 break; | |
| 5602 | |
| 5603 case RS_STAR_LONG: | |
| 5604 case RS_STAR_SHORT: | |
| 5605 { | |
| 5606 regstar_T *rst = ((regstar_T *)rp) - 1; | |
| 5607 | |
| 5608 if (status == RA_MATCH) | |
| 5609 { | |
| 270 | 5610 regstack_pop(&scan); |
| 180 | 5611 regstack.ga_len -= sizeof(regstar_T); |
| 5612 break; | |
| 5613 } | |
| 5614 | |
| 5615 /* Tried once already, restore input pointers. */ | |
| 5616 if (status != RA_BREAK) | |
| 233 | 5617 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5618 |
| 5619 /* Repeat until we found a position where it could match. */ | |
| 5620 for (;;) | |
| 5621 { | |
| 5622 if (status != RA_BREAK) | |
| 5623 { | |
| 5624 /* Tried first position already, advance. */ | |
| 5625 if (rp->rs_state == RS_STAR_LONG) | |
| 5626 { | |
| 685 | 5627 /* Trying for longest match, but couldn't or |
| 5628 * didn't match -- back up one char. */ | |
| 180 | 5629 if (--rst->count < rst->minval) |
| 5630 break; | |
| 5631 if (reginput == regline) | |
| 5632 { | |
| 5633 /* backup to last char of previous line */ | |
| 5634 --reglnum; | |
| 5635 regline = reg_getline(reglnum); | |
| 5636 /* Just in case regrepeat() didn't count | |
| 5637 * right. */ | |
| 5638 if (regline == NULL) | |
| 5639 break; | |
| 5640 reginput = regline + STRLEN(regline); | |
| 5641 fast_breakcheck(); | |
| 5642 } | |
| 5643 else | |
| 5644 mb_ptr_back(regline, reginput); | |
| 5645 } | |
| 5646 else | |
| 5647 { | |
| 5648 /* Range is backwards, use shortest match first. | |
| 5649 * Careful: maxval and minval are exchanged! | |
| 5650 * Couldn't or didn't match: try advancing one | |
| 5651 * char. */ | |
| 5652 if (rst->count == rst->minval | |
| 5653 || regrepeat(OPERAND(rp->rs_scan), 1L) == 0) | |
| 5654 break; | |
| 5655 ++rst->count; | |
| 5656 } | |
| 5657 if (got_int) | |
| 5658 break; | |
| 5659 } | |
| 5660 else | |
| 5661 status = RA_NOMATCH; | |
| 5662 | |
| 5663 /* If it could match, try it. */ | |
| 5664 if (rst->nextb == NUL || *reginput == rst->nextb | |
| 5665 || *reginput == rst->nextb_ic) | |
| 5666 { | |
| 233 | 5667 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5668 scan = regnext(rp->rs_scan); |
| 5669 status = RA_CONT; | |
| 5670 break; | |
| 5671 } | |
| 5672 } | |
| 5673 if (status != RA_CONT) | |
| 5674 { | |
| 5675 /* Failed. */ | |
| 270 | 5676 regstack_pop(&scan); |
| 180 | 5677 regstack.ga_len -= sizeof(regstar_T); |
| 5678 status = RA_NOMATCH; | |
| 5679 } | |
| 5680 } | |
| 5681 break; | |
| 5682 } | |
| 5683 | |
| 685 | 5684 /* If we want to continue the inner loop or didn't pop a state |
| 5685 * continue matching loop */ | |
| 180 | 5686 if (status == RA_CONT || rp == (regitem_T *) |
| 5687 ((char *)regstack.ga_data + regstack.ga_len) - 1) | |
| 5688 break; | |
| 5689 } | |
| 5690 | |
| 189 | 5691 /* May need to continue with the inner loop, starting at "scan". */ |
| 180 | 5692 if (status == RA_CONT) |
| 5693 continue; | |
| 5694 | |
| 5695 /* | |
| 5696 * If the regstack is empty or something failed we are done. | |
| 5697 */ | |
| 5698 if (regstack.ga_len == 0 || status == RA_FAIL) | |
| 5699 { | |
| 5700 if (scan == NULL) | |
| 5701 { | |
| 5702 /* | |
| 5703 * We get here only if there's trouble -- normally "case END" is | |
| 5704 * the terminating point. | |
| 5705 */ | |
| 5706 EMSG(_(e_re_corr)); | |
| 7 | 5707 #ifdef DEBUG |
| 180 | 5708 printf("Premature EOL\n"); |
| 7 | 5709 #endif |
| 180 | 5710 } |
| 189 | 5711 if (status == RA_FAIL) |
| 5712 got_int = TRUE; | |
| 180 | 5713 return (status == RA_MATCH); |
| 5714 } | |
| 5715 | |
| 5716 } /* End of loop until the regstack is empty. */ | |
| 5717 | |
| 5718 /* NOTREACHED */ | |
| 5719 } | |
| 5720 | |
| 5721 /* | |
| 5722 * Push an item onto the regstack. | |
| 5723 * Returns pointer to new item. Returns NULL when out of memory. | |
| 5724 */ | |
| 5725 static regitem_T * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
5726 regstack_push(regstate_T state, char_u *scan) |
| 180 | 5727 { |
| 5728 regitem_T *rp; | |
| 5729 | |
| 270 | 5730 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5731 { |
| 5732 EMSG(_(e_maxmempat)); | |
| 5733 return NULL; | |
| 5734 } | |
| 270 | 5735 if (ga_grow(®stack, sizeof(regitem_T)) == FAIL) |
| 180 | 5736 return NULL; |
| 5737 | |
| 270 | 5738 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len); |
| 180 | 5739 rp->rs_state = state; |
| 5740 rp->rs_scan = scan; | |
| 5741 | |
| 270 | 5742 regstack.ga_len += sizeof(regitem_T); |
| 180 | 5743 return rp; |
| 5744 } | |
| 5745 | |
| 5746 /* | |
| 5747 * Pop an item from the regstack. | |
| 5748 */ | |
| 5749 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
5750 regstack_pop(char_u **scan) |
| 180 | 5751 { |
| 5752 regitem_T *rp; | |
| 5753 | |
| 270 | 5754 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; |
| 180 | 5755 *scan = rp->rs_scan; |
| 5756 | |
| 270 | 5757 regstack.ga_len -= sizeof(regitem_T); |
| 7 | 5758 } |
| 5759 | |
| 5760 /* | |
| 5761 * regrepeat - repeatedly match something simple, return how many. | |
| 5762 * Advances reginput (and reglnum) to just after the matched chars. | |
| 5763 */ | |
| 5764 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
5765 regrepeat( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
5766 char_u *p, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
5767 long maxcount) /* maximum number of matches allowed */ |
| 7 | 5768 { |
| 5769 long count = 0; | |
| 5770 char_u *scan; | |
| 5771 char_u *opnd; | |
| 5772 int mask; | |
| 5773 int testval = 0; | |
| 5774 | |
| 5775 scan = reginput; /* Make local copy of reginput for speed. */ | |
| 5776 opnd = OPERAND(p); | |
| 5777 switch (OP(p)) | |
| 5778 { | |
| 5779 case ANY: | |
| 5780 case ANY + ADD_NL: | |
| 5781 while (count < maxcount) | |
| 5782 { | |
| 5783 /* Matching anything means we continue until end-of-line (or | |
| 5784 * end-of-file for ANY + ADD_NL), only limited by maxcount. */ | |
| 5785 while (*scan != NUL && count < maxcount) | |
| 5786 { | |
| 5787 ++count; | |
| 39 | 5788 mb_ptr_adv(scan); |
| 7 | 5789 } |
| 1018 | 5790 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5791 || reg_line_lbr || count == maxcount) | |
| 7 | 5792 break; |
| 5793 ++count; /* count the line-break */ | |
| 5794 reg_nextline(); | |
| 5795 scan = reginput; | |
| 5796 if (got_int) | |
| 5797 break; | |
| 5798 } | |
| 5799 break; | |
| 5800 | |
| 5801 case IDENT: | |
| 5802 case IDENT + ADD_NL: | |
| 5803 testval = TRUE; | |
| 5804 /*FALLTHROUGH*/ | |
| 5805 case SIDENT: | |
| 5806 case SIDENT + ADD_NL: | |
| 5807 while (count < maxcount) | |
| 5808 { | |
| 4466 | 5809 if (vim_isIDc(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5810 { |
| 39 | 5811 mb_ptr_adv(scan); |
| 7 | 5812 } |
| 5813 else if (*scan == NUL) | |
| 5814 { | |
| 1018 | 5815 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5816 || reg_line_lbr) | |
| 7 | 5817 break; |
| 5818 reg_nextline(); | |
| 5819 scan = reginput; | |
| 5820 if (got_int) | |
| 5821 break; | |
| 5822 } | |
| 5823 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5824 ++scan; | |
| 5825 else | |
| 5826 break; | |
| 5827 ++count; | |
| 5828 } | |
| 5829 break; | |
| 5830 | |
| 5831 case KWORD: | |
| 5832 case KWORD + ADD_NL: | |
| 5833 testval = TRUE; | |
| 5834 /*FALLTHROUGH*/ | |
| 5835 case SKWORD: | |
| 5836 case SKWORD + ADD_NL: | |
| 5837 while (count < maxcount) | |
| 5838 { | |
| 4069 | 5839 if (vim_iswordp_buf(scan, reg_buf) |
| 5840 && (testval || !VIM_ISDIGIT(*scan))) | |
| 7 | 5841 { |
| 39 | 5842 mb_ptr_adv(scan); |
| 7 | 5843 } |
| 5844 else if (*scan == NUL) | |
| 5845 { | |
| 1018 | 5846 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5847 || reg_line_lbr) | |
| 7 | 5848 break; |
| 5849 reg_nextline(); | |
| 5850 scan = reginput; | |
| 5851 if (got_int) | |
| 5852 break; | |
| 5853 } | |
| 5854 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5855 ++scan; | |
| 5856 else | |
| 5857 break; | |
| 5858 ++count; | |
| 5859 } | |
| 5860 break; | |
| 5861 | |
| 5862 case FNAME: | |
| 5863 case FNAME + ADD_NL: | |
| 5864 testval = TRUE; | |
| 5865 /*FALLTHROUGH*/ | |
| 5866 case SFNAME: | |
| 5867 case SFNAME + ADD_NL: | |
| 5868 while (count < maxcount) | |
| 5869 { | |
| 4466 | 5870 if (vim_isfilec(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5871 { |
| 39 | 5872 mb_ptr_adv(scan); |
| 7 | 5873 } |
| 5874 else if (*scan == NUL) | |
| 5875 { | |
| 1018 | 5876 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5877 || reg_line_lbr) | |
| 7 | 5878 break; |
| 5879 reg_nextline(); | |
| 5880 scan = reginput; | |
| 5881 if (got_int) | |
| 5882 break; | |
| 5883 } | |
| 5884 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5885 ++scan; | |
| 5886 else | |
| 5887 break; | |
| 5888 ++count; | |
| 5889 } | |
| 5890 break; | |
| 5891 | |
| 5892 case PRINT: | |
| 5893 case PRINT + ADD_NL: | |
| 5894 testval = TRUE; | |
| 5895 /*FALLTHROUGH*/ | |
| 5896 case SPRINT: | |
| 5897 case SPRINT + ADD_NL: | |
| 5898 while (count < maxcount) | |
| 5899 { | |
| 5900 if (*scan == NUL) | |
| 5901 { | |
| 1018 | 5902 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5903 || reg_line_lbr) | |
| 7 | 5904 break; |
| 5905 reg_nextline(); | |
| 5906 scan = reginput; | |
| 5907 if (got_int) | |
| 5908 break; | |
| 5909 } | |
|
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
5910 else if (vim_isprintc(PTR2CHAR(scan)) == 1 |
|
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
5911 && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5912 { |
| 39 | 5913 mb_ptr_adv(scan); |
| 7 | 5914 } |
| 5915 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5916 ++scan; | |
| 5917 else | |
| 5918 break; | |
| 5919 ++count; | |
| 5920 } | |
| 5921 break; | |
| 5922 | |
| 5923 case WHITE: | |
| 5924 case WHITE + ADD_NL: | |
| 5925 testval = mask = RI_WHITE; | |
| 5926 do_class: | |
| 5927 while (count < maxcount) | |
| 5928 { | |
| 5929 #ifdef FEAT_MBYTE | |
| 5930 int l; | |
| 5931 #endif | |
| 5932 if (*scan == NUL) | |
| 5933 { | |
| 1018 | 5934 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5935 || reg_line_lbr) | |
| 7 | 5936 break; |
| 5937 reg_nextline(); | |
| 5938 scan = reginput; | |
| 5939 if (got_int) | |
| 5940 break; | |
| 5941 } | |
| 5942 #ifdef FEAT_MBYTE | |
| 474 | 5943 else if (has_mbyte && (l = (*mb_ptr2len)(scan)) > 1) |
| 7 | 5944 { |
| 5945 if (testval != 0) | |
| 5946 break; | |
| 5947 scan += l; | |
| 5948 } | |
| 5949 #endif | |
| 5950 else if ((class_tab[*scan] & mask) == testval) | |
| 5951 ++scan; | |
| 5952 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5953 ++scan; | |
| 5954 else | |
| 5955 break; | |
| 5956 ++count; | |
| 5957 } | |
| 5958 break; | |
| 5959 | |
| 5960 case NWHITE: | |
| 5961 case NWHITE + ADD_NL: | |
| 5962 mask = RI_WHITE; | |
| 5963 goto do_class; | |
| 5964 case DIGIT: | |
| 5965 case DIGIT + ADD_NL: | |
| 5966 testval = mask = RI_DIGIT; | |
| 5967 goto do_class; | |
| 5968 case NDIGIT: | |
| 5969 case NDIGIT + ADD_NL: | |
| 5970 mask = RI_DIGIT; | |
| 5971 goto do_class; | |
| 5972 case HEX: | |
| 5973 case HEX + ADD_NL: | |
| 5974 testval = mask = RI_HEX; | |
| 5975 goto do_class; | |
| 5976 case NHEX: | |
| 5977 case NHEX + ADD_NL: | |
| 5978 mask = RI_HEX; | |
| 5979 goto do_class; | |
| 5980 case OCTAL: | |
| 5981 case OCTAL + ADD_NL: | |
| 5982 testval = mask = RI_OCTAL; | |
| 5983 goto do_class; | |
| 5984 case NOCTAL: | |
| 5985 case NOCTAL + ADD_NL: | |
| 5986 mask = RI_OCTAL; | |
| 5987 goto do_class; | |
| 5988 case WORD: | |
| 5989 case WORD + ADD_NL: | |
| 5990 testval = mask = RI_WORD; | |
| 5991 goto do_class; | |
| 5992 case NWORD: | |
| 5993 case NWORD + ADD_NL: | |
| 5994 mask = RI_WORD; | |
| 5995 goto do_class; | |
| 5996 case HEAD: | |
| 5997 case HEAD + ADD_NL: | |
| 5998 testval = mask = RI_HEAD; | |
| 5999 goto do_class; | |
| 6000 case NHEAD: | |
| 6001 case NHEAD + ADD_NL: | |
| 6002 mask = RI_HEAD; | |
| 6003 goto do_class; | |
| 6004 case ALPHA: | |
| 6005 case ALPHA + ADD_NL: | |
| 6006 testval = mask = RI_ALPHA; | |
| 6007 goto do_class; | |
| 6008 case NALPHA: | |
| 6009 case NALPHA + ADD_NL: | |
| 6010 mask = RI_ALPHA; | |
| 6011 goto do_class; | |
| 6012 case LOWER: | |
| 6013 case LOWER + ADD_NL: | |
| 6014 testval = mask = RI_LOWER; | |
| 6015 goto do_class; | |
| 6016 case NLOWER: | |
| 6017 case NLOWER + ADD_NL: | |
| 6018 mask = RI_LOWER; | |
| 6019 goto do_class; | |
| 6020 case UPPER: | |
| 6021 case UPPER + ADD_NL: | |
| 6022 testval = mask = RI_UPPER; | |
| 6023 goto do_class; | |
| 6024 case NUPPER: | |
| 6025 case NUPPER + ADD_NL: | |
| 6026 mask = RI_UPPER; | |
| 6027 goto do_class; | |
| 6028 | |
| 6029 case EXACTLY: | |
| 6030 { | |
| 6031 int cu, cl; | |
| 6032 | |
| 6033 /* This doesn't do a multi-byte character, because a MULTIBYTECODE | |
| 1347 | 6034 * would have been used for it. It does handle single-byte |
| 6035 * characters, such as latin1. */ | |
| 7 | 6036 if (ireg_ic) |
| 6037 { | |
| 1347 | 6038 cu = MB_TOUPPER(*opnd); |
| 6039 cl = MB_TOLOWER(*opnd); | |
| 7 | 6040 while (count < maxcount && (*scan == cu || *scan == cl)) |
| 6041 { | |
| 6042 count++; | |
| 6043 scan++; | |
| 6044 } | |
| 6045 } | |
| 6046 else | |
| 6047 { | |
| 6048 cu = *opnd; | |
| 6049 while (count < maxcount && *scan == cu) | |
| 6050 { | |
| 6051 count++; | |
| 6052 scan++; | |
| 6053 } | |
| 6054 } | |
| 6055 break; | |
| 6056 } | |
| 6057 | |
| 6058 #ifdef FEAT_MBYTE | |
| 6059 case MULTIBYTECODE: | |
| 6060 { | |
| 6061 int i, len, cf = 0; | |
| 6062 | |
| 6063 /* Safety check (just in case 'encoding' was changed since | |
| 6064 * compiling the program). */ | |
| 474 | 6065 if ((len = (*mb_ptr2len)(opnd)) > 1) |
| 7 | 6066 { |
| 6067 if (ireg_ic && enc_utf8) | |
| 6068 cf = utf_fold(utf_ptr2char(opnd)); | |
| 6785 | 6069 while (count < maxcount && (*mb_ptr2len)(scan) >= len) |
| 7 | 6070 { |
| 6071 for (i = 0; i < len; ++i) | |
| 6072 if (opnd[i] != scan[i]) | |
| 6073 break; | |
| 6074 if (i < len && (!ireg_ic || !enc_utf8 | |
| 6075 || utf_fold(utf_ptr2char(scan)) != cf)) | |
| 6076 break; | |
| 6077 scan += len; | |
| 6078 ++count; | |
| 6079 } | |
| 6080 } | |
| 6081 } | |
| 6082 break; | |
| 6083 #endif | |
| 6084 | |
| 6085 case ANYOF: | |
| 6086 case ANYOF + ADD_NL: | |
| 6087 testval = TRUE; | |
| 6088 /*FALLTHROUGH*/ | |
| 6089 | |
| 6090 case ANYBUT: | |
| 6091 case ANYBUT + ADD_NL: | |
| 6092 while (count < maxcount) | |
| 6093 { | |
| 6094 #ifdef FEAT_MBYTE | |
| 6095 int len; | |
| 6096 #endif | |
| 6097 if (*scan == NUL) | |
| 6098 { | |
| 1018 | 6099 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 6100 || reg_line_lbr) | |
| 7 | 6101 break; |
| 6102 reg_nextline(); | |
| 6103 scan = reginput; | |
| 6104 if (got_int) | |
| 6105 break; | |
| 6106 } | |
| 6107 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 6108 ++scan; | |
| 6109 #ifdef FEAT_MBYTE | |
| 474 | 6110 else if (has_mbyte && (len = (*mb_ptr2len)(scan)) > 1) |
| 7 | 6111 { |
| 6112 if ((cstrchr(opnd, (*mb_ptr2char)(scan)) == NULL) == testval) | |
| 6113 break; | |
| 6114 scan += len; | |
| 6115 } | |
| 6116 #endif | |
| 6117 else | |
| 6118 { | |
| 6119 if ((cstrchr(opnd, *scan) == NULL) == testval) | |
| 6120 break; | |
| 6121 ++scan; | |
| 6122 } | |
| 6123 ++count; | |
| 6124 } | |
| 6125 break; | |
| 6126 | |
| 6127 case NEWL: | |
| 6128 while (count < maxcount | |
| 1018 | 6129 && ((*scan == NUL && reglnum <= reg_maxline && !reg_line_lbr |
| 6130 && REG_MULTI) || (*scan == '\n' && reg_line_lbr))) | |
| 7 | 6131 { |
| 6132 count++; | |
| 6133 if (reg_line_lbr) | |
| 6134 ADVANCE_REGINPUT(); | |
| 6135 else | |
| 6136 reg_nextline(); | |
| 6137 scan = reginput; | |
| 6138 if (got_int) | |
| 6139 break; | |
| 6140 } | |
| 6141 break; | |
| 6142 | |
| 6143 default: /* Oh dear. Called inappropriately. */ | |
| 6144 EMSG(_(e_re_corr)); | |
| 6145 #ifdef DEBUG | |
| 6146 printf("Called regrepeat with op code %d\n", OP(p)); | |
| 6147 #endif | |
| 6148 break; | |
| 6149 } | |
| 6150 | |
| 6151 reginput = scan; | |
| 6152 | |
| 6153 return (int)count; | |
| 6154 } | |
| 6155 | |
| 6156 /* | |
| 6157 * regnext - dig the "next" pointer out of a node | |
| 2010 | 6158 * Returns NULL when calculating size, when there is no next item and when |
| 6159 * there is an error. | |
| 7 | 6160 */ |
| 6161 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6162 regnext(char_u *p) |
| 7 | 6163 { |
| 6164 int offset; | |
| 6165 | |
| 2010 | 6166 if (p == JUST_CALC_SIZE || reg_toolong) |
| 7 | 6167 return NULL; |
| 6168 | |
| 6169 offset = NEXT(p); | |
| 6170 if (offset == 0) | |
| 6171 return NULL; | |
| 6172 | |
| 233 | 6173 if (OP(p) == BACK) |
| 7 | 6174 return p - offset; |
| 6175 else | |
| 6176 return p + offset; | |
| 6177 } | |
| 6178 | |
| 6179 /* | |
| 6180 * Check the regexp program for its magic number. | |
| 6181 * Return TRUE if it's wrong. | |
| 6182 */ | |
| 6183 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6184 prog_magic_wrong(void) |
| 7 | 6185 { |
| 4444 | 6186 regprog_T *prog; |
| 6187 | |
| 6188 prog = REG_MULTI ? reg_mmatch->regprog : reg_match->regprog; | |
| 6189 if (prog->engine == &nfa_regengine) | |
| 6190 /* For NFA matcher we don't check the magic */ | |
| 6191 return FALSE; | |
| 6192 | |
| 6193 if (UCHARAT(((bt_regprog_T *)prog)->program) != REGMAGIC) | |
| 7 | 6194 { |
| 6195 EMSG(_(e_re_corr)); | |
| 6196 return TRUE; | |
| 6197 } | |
| 6198 return FALSE; | |
| 6199 } | |
| 6200 | |
| 6201 /* | |
| 6202 * Cleanup the subexpressions, if this wasn't done yet. | |
| 6203 * This construction is used to clear the subexpressions only when they are | |
| 6204 * used (to increase speed). | |
| 6205 */ | |
| 6206 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6207 cleanup_subexpr(void) |
| 7 | 6208 { |
| 6209 if (need_clear_subexpr) | |
| 6210 { | |
| 6211 if (REG_MULTI) | |
| 6212 { | |
| 6213 /* Use 0xff to set lnum to -1 */ | |
| 6214 vim_memset(reg_startpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6215 vim_memset(reg_endpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6216 } | |
| 6217 else | |
| 6218 { | |
| 6219 vim_memset(reg_startp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6220 vim_memset(reg_endp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6221 } | |
| 6222 need_clear_subexpr = FALSE; | |
| 6223 } | |
| 6224 } | |
| 6225 | |
| 6226 #ifdef FEAT_SYN_HL | |
| 6227 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6228 cleanup_zsubexpr(void) |
| 7 | 6229 { |
| 6230 if (need_clear_zsubexpr) | |
| 6231 { | |
| 6232 if (REG_MULTI) | |
| 6233 { | |
| 6234 /* Use 0xff to set lnum to -1 */ | |
| 6235 vim_memset(reg_startzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6236 vim_memset(reg_endzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6237 } | |
| 6238 else | |
| 6239 { | |
| 6240 vim_memset(reg_startzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6241 vim_memset(reg_endzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6242 } | |
| 6243 need_clear_zsubexpr = FALSE; | |
| 6244 } | |
| 6245 } | |
| 6246 #endif | |
| 6247 | |
| 6248 /* | |
| 1579 | 6249 * Save the current subexpr to "bp", so that they can be restored |
| 6250 * later by restore_subexpr(). | |
| 6251 */ | |
| 6252 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6253 save_subexpr(regbehind_T *bp) |
| 1579 | 6254 { |
| 6255 int i; | |
| 6256 | |
| 1602 | 6257 /* When "need_clear_subexpr" is set we don't need to save the values, only |
| 6258 * remember that this flag needs to be set again when restoring. */ | |
| 6259 bp->save_need_clear_subexpr = need_clear_subexpr; | |
| 6260 if (!need_clear_subexpr) | |
| 1579 | 6261 { |
| 1602 | 6262 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6263 { |
| 1602 | 6264 if (REG_MULTI) |
| 6265 { | |
| 6266 bp->save_start[i].se_u.pos = reg_startpos[i]; | |
| 6267 bp->save_end[i].se_u.pos = reg_endpos[i]; | |
| 6268 } | |
| 6269 else | |
| 6270 { | |
| 6271 bp->save_start[i].se_u.ptr = reg_startp[i]; | |
| 6272 bp->save_end[i].se_u.ptr = reg_endp[i]; | |
| 6273 } | |
| 1579 | 6274 } |
| 6275 } | |
| 6276 } | |
| 6277 | |
| 6278 /* | |
| 6279 * Restore the subexpr from "bp". | |
| 6280 */ | |
| 6281 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6282 restore_subexpr(regbehind_T *bp) |
| 1579 | 6283 { |
| 6284 int i; | |
| 6285 | |
| 1602 | 6286 /* Only need to restore saved values when they are not to be cleared. */ |
| 6287 need_clear_subexpr = bp->save_need_clear_subexpr; | |
| 6288 if (!need_clear_subexpr) | |
| 1579 | 6289 { |
| 1602 | 6290 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6291 { |
| 1602 | 6292 if (REG_MULTI) |
| 6293 { | |
| 6294 reg_startpos[i] = bp->save_start[i].se_u.pos; | |
| 6295 reg_endpos[i] = bp->save_end[i].se_u.pos; | |
| 6296 } | |
| 6297 else | |
| 6298 { | |
| 6299 reg_startp[i] = bp->save_start[i].se_u.ptr; | |
| 6300 reg_endp[i] = bp->save_end[i].se_u.ptr; | |
| 6301 } | |
| 1579 | 6302 } |
| 6303 } | |
| 6304 } | |
| 6305 | |
| 6306 /* | |
| 7 | 6307 * Advance reglnum, regline and reginput to the next line. |
| 6308 */ | |
| 6309 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6310 reg_nextline(void) |
| 7 | 6311 { |
| 6312 regline = reg_getline(++reglnum); | |
| 6313 reginput = regline; | |
| 6314 fast_breakcheck(); | |
| 6315 } | |
| 6316 | |
| 6317 /* | |
| 6318 * Save the input line and position in a regsave_T. | |
| 6319 */ | |
| 6320 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6321 reg_save(regsave_T *save, garray_T *gap) |
| 7 | 6322 { |
| 6323 if (REG_MULTI) | |
| 6324 { | |
| 6325 save->rs_u.pos.col = (colnr_T)(reginput - regline); | |
| 6326 save->rs_u.pos.lnum = reglnum; | |
| 6327 } | |
| 6328 else | |
| 6329 save->rs_u.ptr = reginput; | |
| 233 | 6330 save->rs_len = gap->ga_len; |
| 7 | 6331 } |
| 6332 | |
| 6333 /* | |
| 6334 * Restore the input line and position from a regsave_T. | |
| 6335 */ | |
| 6336 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6337 reg_restore(regsave_T *save, garray_T *gap) |
| 7 | 6338 { |
| 6339 if (REG_MULTI) | |
| 6340 { | |
| 6341 if (reglnum != save->rs_u.pos.lnum) | |
| 6342 { | |
| 6343 /* only call reg_getline() when the line number changed to save | |
| 6344 * a bit of time */ | |
| 6345 reglnum = save->rs_u.pos.lnum; | |
| 6346 regline = reg_getline(reglnum); | |
| 6347 } | |
| 6348 reginput = regline + save->rs_u.pos.col; | |
| 6349 } | |
| 6350 else | |
| 6351 reginput = save->rs_u.ptr; | |
| 233 | 6352 gap->ga_len = save->rs_len; |
| 7 | 6353 } |
| 6354 | |
| 6355 /* | |
| 6356 * Return TRUE if current position is equal to saved position. | |
| 6357 */ | |
| 6358 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6359 reg_save_equal(regsave_T *save) |
| 7 | 6360 { |
| 6361 if (REG_MULTI) | |
| 6362 return reglnum == save->rs_u.pos.lnum | |
| 6363 && reginput == regline + save->rs_u.pos.col; | |
| 6364 return reginput == save->rs_u.ptr; | |
| 6365 } | |
| 6366 | |
| 6367 /* | |
| 6368 * Tentatively set the sub-expression start to the current position (after | |
| 6369 * calling regmatch() they will have changed). Need to save the existing | |
| 6370 * values for when there is no match. | |
| 6371 * Use se_save() to use pointer (save_se_multi()) or position (save_se_one()), | |
| 6372 * depending on REG_MULTI. | |
| 6373 */ | |
| 6374 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6375 save_se_multi(save_se_T *savep, lpos_T *posp) |
| 7 | 6376 { |
| 6377 savep->se_u.pos = *posp; | |
| 6378 posp->lnum = reglnum; | |
| 6379 posp->col = (colnr_T)(reginput - regline); | |
| 6380 } | |
| 6381 | |
| 6382 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6383 save_se_one(save_se_T *savep, char_u **pp) |
| 7 | 6384 { |
| 6385 savep->se_u.ptr = *pp; | |
| 6386 *pp = reginput; | |
| 6387 } | |
| 6388 | |
| 6389 /* | |
| 6390 * Compare a number with the operand of RE_LNUM, RE_COL or RE_VCOL. | |
| 6391 */ | |
| 6392 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6393 re_num_cmp(long_u val, char_u *scan) |
| 7 | 6394 { |
| 6395 long_u n = OPERAND_MIN(scan); | |
| 6396 | |
| 6397 if (OPERAND_CMP(scan) == '>') | |
| 6398 return val > n; | |
| 6399 if (OPERAND_CMP(scan) == '<') | |
| 6400 return val < n; | |
| 6401 return val == n; | |
| 6402 } | |
| 6403 | |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6404 /* |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6405 * Check whether a backreference matches. |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6406 * Returns RA_FAIL, RA_NOMATCH or RA_MATCH. |
| 5504 | 6407 * If "bytelen" is not NULL, it is set to the byte length of the match in the |
| 6408 * last line. | |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6409 */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6410 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6411 match_with_backref( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6412 linenr_T start_lnum, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6413 colnr_T start_col, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6414 linenr_T end_lnum, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6415 colnr_T end_col, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6416 int *bytelen) |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6417 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6418 linenr_T clnum = start_lnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6419 colnr_T ccol = start_col; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6420 int len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6421 char_u *p; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6422 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6423 if (bytelen != NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6424 *bytelen = 0; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6425 for (;;) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6426 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6427 /* 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
|
6428 * Slow! */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6429 if (regline != reg_tofree) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6430 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6431 len = (int)STRLEN(regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6432 if (reg_tofree == NULL || len >= (int)reg_tofreelen) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6433 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6434 len += 50; /* get some extra */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6435 vim_free(reg_tofree); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6436 reg_tofree = alloc(len); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6437 if (reg_tofree == NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6438 return RA_FAIL; /* out of memory!*/ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6439 reg_tofreelen = len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6440 } |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6441 STRCPY(reg_tofree, regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6442 reginput = reg_tofree + (reginput - regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6443 regline = reg_tofree; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6444 } |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6445 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6446 /* Get the line to compare with. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6447 p = reg_getline(clnum); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6448 if (clnum == end_lnum) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6449 len = end_col - ccol; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6450 else |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6451 len = (int)STRLEN(p + ccol); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6452 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6453 if (cstrncmp(p + ccol, reginput, &len) != 0) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6454 return RA_NOMATCH; /* doesn't match */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6455 if (bytelen != NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6456 *bytelen += len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6457 if (clnum == end_lnum) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6458 break; /* match and at end! */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6459 if (reglnum >= reg_maxline) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6460 return RA_NOMATCH; /* text too short */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6461 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6462 /* Advance to next line. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6463 reg_nextline(); |
| 5504 | 6464 if (bytelen != NULL) |
| 6465 *bytelen = 0; | |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6466 ++clnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6467 ccol = 0; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6468 if (got_int) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6469 return RA_FAIL; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6470 } |
|
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 /* 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
|
6473 * that should not matter. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6474 return RA_MATCH; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6475 } |
| 7 | 6476 |
| 4444 | 6477 #ifdef BT_REGEXP_DUMP |
| 7 | 6478 |
| 6479 /* | |
| 6480 * regdump - dump a regexp onto stdout in vaguely comprehensible form | |
| 6481 */ | |
| 6482 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6483 regdump(char_u *pattern, bt_regprog_T *r) |
| 7 | 6484 { |
| 6485 char_u *s; | |
| 6486 int op = EXACTLY; /* Arbitrary non-END op. */ | |
| 6487 char_u *next; | |
| 6488 char_u *end = NULL; | |
| 4444 | 6489 FILE *f; |
| 6490 | |
| 6491 #ifdef BT_REGEXP_LOG | |
| 6492 f = fopen("bt_regexp_log.log", "a"); | |
| 6493 #else | |
| 6494 f = stdout; | |
| 6495 #endif | |
| 6496 if (f == NULL) | |
| 6497 return; | |
| 6498 fprintf(f, "-------------------------------------\n\r\nregcomp(%s):\r\n", pattern); | |
| 7 | 6499 |
| 6500 s = r->program + 1; | |
| 6501 /* | |
| 6502 * Loop until we find the END that isn't before a referred next (an END | |
| 6503 * can also appear in a NOMATCH operand). | |
| 6504 */ | |
| 6505 while (op != END || s <= end) | |
| 6506 { | |
| 6507 op = OP(s); | |
| 4444 | 6508 fprintf(f, "%2d%s", (int)(s - r->program), regprop(s)); /* Where, what. */ |
| 7 | 6509 next = regnext(s); |
| 6510 if (next == NULL) /* Next ptr. */ | |
| 4444 | 6511 fprintf(f, "(0)"); |
| 7 | 6512 else |
| 4444 | 6513 fprintf(f, "(%d)", (int)((s - r->program) + (next - s))); |
| 7 | 6514 if (end < next) |
| 6515 end = next; | |
| 6516 if (op == BRACE_LIMITS) | |
| 6517 { | |
|
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6518 /* Two ints */ |
| 4444 | 6519 fprintf(f, " minval %ld, maxval %ld", OPERAND_MIN(s), OPERAND_MAX(s)); |
| 7 | 6520 s += 8; |
| 6521 } | |
|
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6522 else if (op == BEHIND || op == NOBEHIND) |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6523 { |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6524 /* one int */ |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6525 fprintf(f, " count %ld", OPERAND_MIN(s)); |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6526 s += 4; |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6527 } |
|
4770
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6528 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
|
6529 { |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6530 /* one int plus comperator */ |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6531 fprintf(f, " count %ld", OPERAND_MIN(s)); |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6532 s += 5; |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6533 } |
| 7 | 6534 s += 3; |
| 6535 if (op == ANYOF || op == ANYOF + ADD_NL | |
| 6536 || op == ANYBUT || op == ANYBUT + ADD_NL | |
| 6537 || op == EXACTLY) | |
| 6538 { | |
| 6539 /* Literal string, where present. */ | |
| 4444 | 6540 fprintf(f, "\nxxxxxxxxx\n"); |
| 7 | 6541 while (*s != NUL) |
| 4444 | 6542 fprintf(f, "%c", *s++); |
| 6543 fprintf(f, "\nxxxxxxxxx\n"); | |
| 7 | 6544 s++; |
| 6545 } | |
| 4444 | 6546 fprintf(f, "\r\n"); |
| 7 | 6547 } |
| 6548 | |
| 6549 /* Header fields of interest. */ | |
| 6550 if (r->regstart != NUL) | |
| 4444 | 6551 fprintf(f, "start `%s' 0x%x; ", r->regstart < 256 |
| 7 | 6552 ? (char *)transchar(r->regstart) |
| 6553 : "multibyte", r->regstart); | |
| 6554 if (r->reganch) | |
| 4444 | 6555 fprintf(f, "anchored; "); |
| 7 | 6556 if (r->regmust != NULL) |
| 4444 | 6557 fprintf(f, "must have \"%s\"", r->regmust); |
| 6558 fprintf(f, "\r\n"); | |
| 6559 | |
| 6560 #ifdef BT_REGEXP_LOG | |
| 6561 fclose(f); | |
| 6562 #endif | |
| 7 | 6563 } |
| 4444 | 6564 #endif /* BT_REGEXP_DUMP */ |
| 6565 | |
| 6566 #ifdef DEBUG | |
| 7 | 6567 /* |
| 6568 * regprop - printable representation of opcode | |
| 6569 */ | |
| 6570 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6571 regprop(char_u *op) |
| 7 | 6572 { |
| 4444 | 6573 char *p; |
| 6574 static char buf[50]; | |
| 6575 | |
| 6576 STRCPY(buf, ":"); | |
| 6577 | |
| 6578 switch ((int) OP(op)) | |
| 7 | 6579 { |
| 6580 case BOL: | |
| 6581 p = "BOL"; | |
| 6582 break; | |
| 6583 case EOL: | |
| 6584 p = "EOL"; | |
| 6585 break; | |
| 6586 case RE_BOF: | |
| 6587 p = "BOF"; | |
| 6588 break; | |
| 6589 case RE_EOF: | |
| 6590 p = "EOF"; | |
| 6591 break; | |
| 6592 case CURSOR: | |
| 6593 p = "CURSOR"; | |
| 6594 break; | |
| 639 | 6595 case RE_VISUAL: |
| 6596 p = "RE_VISUAL"; | |
| 6597 break; | |
| 7 | 6598 case RE_LNUM: |
| 6599 p = "RE_LNUM"; | |
| 6600 break; | |
| 639 | 6601 case RE_MARK: |
| 6602 p = "RE_MARK"; | |
| 6603 break; | |
| 7 | 6604 case RE_COL: |
| 6605 p = "RE_COL"; | |
| 6606 break; | |
| 6607 case RE_VCOL: | |
| 6608 p = "RE_VCOL"; | |
| 6609 break; | |
| 6610 case BOW: | |
| 6611 p = "BOW"; | |
| 6612 break; | |
| 6613 case EOW: | |
| 6614 p = "EOW"; | |
| 6615 break; | |
| 6616 case ANY: | |
| 6617 p = "ANY"; | |
| 6618 break; | |
| 6619 case ANY + ADD_NL: | |
| 6620 p = "ANY+NL"; | |
| 6621 break; | |
| 6622 case ANYOF: | |
| 6623 p = "ANYOF"; | |
| 6624 break; | |
| 6625 case ANYOF + ADD_NL: | |
| 6626 p = "ANYOF+NL"; | |
| 6627 break; | |
| 6628 case ANYBUT: | |
| 6629 p = "ANYBUT"; | |
| 6630 break; | |
| 6631 case ANYBUT + ADD_NL: | |
| 6632 p = "ANYBUT+NL"; | |
| 6633 break; | |
| 6634 case IDENT: | |
| 6635 p = "IDENT"; | |
| 6636 break; | |
| 6637 case IDENT + ADD_NL: | |
| 6638 p = "IDENT+NL"; | |
| 6639 break; | |
| 6640 case SIDENT: | |
| 6641 p = "SIDENT"; | |
| 6642 break; | |
| 6643 case SIDENT + ADD_NL: | |
| 6644 p = "SIDENT+NL"; | |
| 6645 break; | |
| 6646 case KWORD: | |
| 6647 p = "KWORD"; | |
| 6648 break; | |
| 6649 case KWORD + ADD_NL: | |
| 6650 p = "KWORD+NL"; | |
| 6651 break; | |
| 6652 case SKWORD: | |
| 6653 p = "SKWORD"; | |
| 6654 break; | |
| 6655 case SKWORD + ADD_NL: | |
| 6656 p = "SKWORD+NL"; | |
| 6657 break; | |
| 6658 case FNAME: | |
| 6659 p = "FNAME"; | |
| 6660 break; | |
| 6661 case FNAME + ADD_NL: | |
| 6662 p = "FNAME+NL"; | |
| 6663 break; | |
| 6664 case SFNAME: | |
| 6665 p = "SFNAME"; | |
| 6666 break; | |
| 6667 case SFNAME + ADD_NL: | |
| 6668 p = "SFNAME+NL"; | |
| 6669 break; | |
| 6670 case PRINT: | |
| 6671 p = "PRINT"; | |
| 6672 break; | |
| 6673 case PRINT + ADD_NL: | |
| 6674 p = "PRINT+NL"; | |
| 6675 break; | |
| 6676 case SPRINT: | |
| 6677 p = "SPRINT"; | |
| 6678 break; | |
| 6679 case SPRINT + ADD_NL: | |
| 6680 p = "SPRINT+NL"; | |
| 6681 break; | |
| 6682 case WHITE: | |
| 6683 p = "WHITE"; | |
| 6684 break; | |
| 6685 case WHITE + ADD_NL: | |
| 6686 p = "WHITE+NL"; | |
| 6687 break; | |
| 6688 case NWHITE: | |
| 6689 p = "NWHITE"; | |
| 6690 break; | |
| 6691 case NWHITE + ADD_NL: | |
| 6692 p = "NWHITE+NL"; | |
| 6693 break; | |
| 6694 case DIGIT: | |
| 6695 p = "DIGIT"; | |
| 6696 break; | |
| 6697 case DIGIT + ADD_NL: | |
| 6698 p = "DIGIT+NL"; | |
| 6699 break; | |
| 6700 case NDIGIT: | |
| 6701 p = "NDIGIT"; | |
| 6702 break; | |
| 6703 case NDIGIT + ADD_NL: | |
| 6704 p = "NDIGIT+NL"; | |
| 6705 break; | |
| 6706 case HEX: | |
| 6707 p = "HEX"; | |
| 6708 break; | |
| 6709 case HEX + ADD_NL: | |
| 6710 p = "HEX+NL"; | |
| 6711 break; | |
| 6712 case NHEX: | |
| 6713 p = "NHEX"; | |
| 6714 break; | |
| 6715 case NHEX + ADD_NL: | |
| 6716 p = "NHEX+NL"; | |
| 6717 break; | |
| 6718 case OCTAL: | |
| 6719 p = "OCTAL"; | |
| 6720 break; | |
| 6721 case OCTAL + ADD_NL: | |
| 6722 p = "OCTAL+NL"; | |
| 6723 break; | |
| 6724 case NOCTAL: | |
| 6725 p = "NOCTAL"; | |
| 6726 break; | |
| 6727 case NOCTAL + ADD_NL: | |
| 6728 p = "NOCTAL+NL"; | |
| 6729 break; | |
| 6730 case WORD: | |
| 6731 p = "WORD"; | |
| 6732 break; | |
| 6733 case WORD + ADD_NL: | |
| 6734 p = "WORD+NL"; | |
| 6735 break; | |
| 6736 case NWORD: | |
| 6737 p = "NWORD"; | |
| 6738 break; | |
| 6739 case NWORD + ADD_NL: | |
| 6740 p = "NWORD+NL"; | |
| 6741 break; | |
| 6742 case HEAD: | |
| 6743 p = "HEAD"; | |
| 6744 break; | |
| 6745 case HEAD + ADD_NL: | |
| 6746 p = "HEAD+NL"; | |
| 6747 break; | |
| 6748 case NHEAD: | |
| 6749 p = "NHEAD"; | |
| 6750 break; | |
| 6751 case NHEAD + ADD_NL: | |
| 6752 p = "NHEAD+NL"; | |
| 6753 break; | |
| 6754 case ALPHA: | |
| 6755 p = "ALPHA"; | |
| 6756 break; | |
| 6757 case ALPHA + ADD_NL: | |
| 6758 p = "ALPHA+NL"; | |
| 6759 break; | |
| 6760 case NALPHA: | |
| 6761 p = "NALPHA"; | |
| 6762 break; | |
| 6763 case NALPHA + ADD_NL: | |
| 6764 p = "NALPHA+NL"; | |
| 6765 break; | |
| 6766 case LOWER: | |
| 6767 p = "LOWER"; | |
| 6768 break; | |
| 6769 case LOWER + ADD_NL: | |
| 6770 p = "LOWER+NL"; | |
| 6771 break; | |
| 6772 case NLOWER: | |
| 6773 p = "NLOWER"; | |
| 6774 break; | |
| 6775 case NLOWER + ADD_NL: | |
| 6776 p = "NLOWER+NL"; | |
| 6777 break; | |
| 6778 case UPPER: | |
| 6779 p = "UPPER"; | |
| 6780 break; | |
| 6781 case UPPER + ADD_NL: | |
| 6782 p = "UPPER+NL"; | |
| 6783 break; | |
| 6784 case NUPPER: | |
| 6785 p = "NUPPER"; | |
| 6786 break; | |
| 6787 case NUPPER + ADD_NL: | |
| 6788 p = "NUPPER+NL"; | |
| 6789 break; | |
| 6790 case BRANCH: | |
| 6791 p = "BRANCH"; | |
| 6792 break; | |
| 6793 case EXACTLY: | |
| 6794 p = "EXACTLY"; | |
| 6795 break; | |
| 6796 case NOTHING: | |
| 6797 p = "NOTHING"; | |
| 6798 break; | |
| 6799 case BACK: | |
| 6800 p = "BACK"; | |
| 6801 break; | |
| 6802 case END: | |
| 6803 p = "END"; | |
| 6804 break; | |
| 6805 case MOPEN + 0: | |
| 6806 p = "MATCH START"; | |
| 6807 break; | |
| 6808 case MOPEN + 1: | |
| 6809 case MOPEN + 2: | |
| 6810 case MOPEN + 3: | |
| 6811 case MOPEN + 4: | |
| 6812 case MOPEN + 5: | |
| 6813 case MOPEN + 6: | |
| 6814 case MOPEN + 7: | |
| 6815 case MOPEN + 8: | |
| 6816 case MOPEN + 9: | |
| 6817 sprintf(buf + STRLEN(buf), "MOPEN%d", OP(op) - MOPEN); | |
| 6818 p = NULL; | |
| 6819 break; | |
| 6820 case MCLOSE + 0: | |
| 6821 p = "MATCH END"; | |
| 6822 break; | |
| 6823 case MCLOSE + 1: | |
| 6824 case MCLOSE + 2: | |
| 6825 case MCLOSE + 3: | |
| 6826 case MCLOSE + 4: | |
| 6827 case MCLOSE + 5: | |
| 6828 case MCLOSE + 6: | |
| 6829 case MCLOSE + 7: | |
| 6830 case MCLOSE + 8: | |
| 6831 case MCLOSE + 9: | |
| 6832 sprintf(buf + STRLEN(buf), "MCLOSE%d", OP(op) - MCLOSE); | |
| 6833 p = NULL; | |
| 6834 break; | |
| 6835 case BACKREF + 1: | |
| 6836 case BACKREF + 2: | |
| 6837 case BACKREF + 3: | |
| 6838 case BACKREF + 4: | |
| 6839 case BACKREF + 5: | |
| 6840 case BACKREF + 6: | |
| 6841 case BACKREF + 7: | |
| 6842 case BACKREF + 8: | |
| 6843 case BACKREF + 9: | |
| 6844 sprintf(buf + STRLEN(buf), "BACKREF%d", OP(op) - BACKREF); | |
| 6845 p = NULL; | |
| 6846 break; | |
| 6847 case NOPEN: | |
| 6848 p = "NOPEN"; | |
| 6849 break; | |
| 6850 case NCLOSE: | |
| 6851 p = "NCLOSE"; | |
| 6852 break; | |
| 6853 #ifdef FEAT_SYN_HL | |
| 6854 case ZOPEN + 1: | |
| 6855 case ZOPEN + 2: | |
| 6856 case ZOPEN + 3: | |
| 6857 case ZOPEN + 4: | |
| 6858 case ZOPEN + 5: | |
| 6859 case ZOPEN + 6: | |
| 6860 case ZOPEN + 7: | |
| 6861 case ZOPEN + 8: | |
| 6862 case ZOPEN + 9: | |
| 6863 sprintf(buf + STRLEN(buf), "ZOPEN%d", OP(op) - ZOPEN); | |
| 6864 p = NULL; | |
| 6865 break; | |
| 6866 case ZCLOSE + 1: | |
| 6867 case ZCLOSE + 2: | |
| 6868 case ZCLOSE + 3: | |
| 6869 case ZCLOSE + 4: | |
| 6870 case ZCLOSE + 5: | |
| 6871 case ZCLOSE + 6: | |
| 6872 case ZCLOSE + 7: | |
| 6873 case ZCLOSE + 8: | |
| 6874 case ZCLOSE + 9: | |
| 6875 sprintf(buf + STRLEN(buf), "ZCLOSE%d", OP(op) - ZCLOSE); | |
| 6876 p = NULL; | |
| 6877 break; | |
| 6878 case ZREF + 1: | |
| 6879 case ZREF + 2: | |
| 6880 case ZREF + 3: | |
| 6881 case ZREF + 4: | |
| 6882 case ZREF + 5: | |
| 6883 case ZREF + 6: | |
| 6884 case ZREF + 7: | |
| 6885 case ZREF + 8: | |
| 6886 case ZREF + 9: | |
| 6887 sprintf(buf + STRLEN(buf), "ZREF%d", OP(op) - ZREF); | |
| 6888 p = NULL; | |
| 6889 break; | |
| 6890 #endif | |
| 6891 case STAR: | |
| 6892 p = "STAR"; | |
| 6893 break; | |
| 6894 case PLUS: | |
| 6895 p = "PLUS"; | |
| 6896 break; | |
| 6897 case NOMATCH: | |
| 6898 p = "NOMATCH"; | |
| 6899 break; | |
| 6900 case MATCH: | |
| 6901 p = "MATCH"; | |
| 6902 break; | |
| 6903 case BEHIND: | |
| 6904 p = "BEHIND"; | |
| 6905 break; | |
| 6906 case NOBEHIND: | |
| 6907 p = "NOBEHIND"; | |
| 6908 break; | |
| 6909 case SUBPAT: | |
| 6910 p = "SUBPAT"; | |
| 6911 break; | |
| 6912 case BRACE_LIMITS: | |
| 6913 p = "BRACE_LIMITS"; | |
| 6914 break; | |
| 6915 case BRACE_SIMPLE: | |
| 6916 p = "BRACE_SIMPLE"; | |
| 6917 break; | |
| 6918 case BRACE_COMPLEX + 0: | |
| 6919 case BRACE_COMPLEX + 1: | |
| 6920 case BRACE_COMPLEX + 2: | |
| 6921 case BRACE_COMPLEX + 3: | |
| 6922 case BRACE_COMPLEX + 4: | |
| 6923 case BRACE_COMPLEX + 5: | |
| 6924 case BRACE_COMPLEX + 6: | |
| 6925 case BRACE_COMPLEX + 7: | |
| 6926 case BRACE_COMPLEX + 8: | |
| 6927 case BRACE_COMPLEX + 9: | |
| 6928 sprintf(buf + STRLEN(buf), "BRACE_COMPLEX%d", OP(op) - BRACE_COMPLEX); | |
| 6929 p = NULL; | |
| 6930 break; | |
| 6931 #ifdef FEAT_MBYTE | |
| 6932 case MULTIBYTECODE: | |
| 6933 p = "MULTIBYTECODE"; | |
| 6934 break; | |
| 6935 #endif | |
| 6936 case NEWL: | |
| 6937 p = "NEWL"; | |
| 6938 break; | |
| 6939 default: | |
| 6940 sprintf(buf + STRLEN(buf), "corrupt %d", OP(op)); | |
| 6941 p = NULL; | |
| 6942 break; | |
| 6943 } | |
| 6944 if (p != NULL) | |
| 4444 | 6945 STRCAT(buf, p); |
| 6946 return (char_u *)buf; | |
| 7 | 6947 } |
| 4444 | 6948 #endif /* DEBUG */ |
| 7 | 6949 |
| 6203 | 6950 /* |
| 6951 * Used in a place where no * or \+ can follow. | |
| 6952 */ | |
| 6953 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
6954 re_mult_next(char *what) |
| 6203 | 6955 { |
| 6956 if (re_multi_type(peekchr()) == MULTI_MULT) | |
| 6957 EMSG2_RET_FAIL(_("E888: (NFA regexp) cannot repeat %s"), what); | |
| 6958 return OK; | |
| 6959 } | |
| 6960 | |
| 7 | 6961 #ifdef FEAT_MBYTE |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
6962 static void mb_decompose(int c, int *c1, int *c2, int *c3); |
| 7 | 6963 |
| 6964 typedef struct | |
| 6965 { | |
| 6966 int a, b, c; | |
| 6967 } decomp_T; | |
| 6968 | |
| 6969 | |
| 6970 /* 0xfb20 - 0xfb4f */ | |
| 297 | 6971 static decomp_T decomp_table[0xfb4f-0xfb20+1] = |
| 7 | 6972 { |
| 6973 {0x5e2,0,0}, /* 0xfb20 alt ayin */ | |
| 6974 {0x5d0,0,0}, /* 0xfb21 alt alef */ | |
| 6975 {0x5d3,0,0}, /* 0xfb22 alt dalet */ | |
| 6976 {0x5d4,0,0}, /* 0xfb23 alt he */ | |
| 6977 {0x5db,0,0}, /* 0xfb24 alt kaf */ | |
| 6978 {0x5dc,0,0}, /* 0xfb25 alt lamed */ | |
| 6979 {0x5dd,0,0}, /* 0xfb26 alt mem-sofit */ | |
| 6980 {0x5e8,0,0}, /* 0xfb27 alt resh */ | |
| 6981 {0x5ea,0,0}, /* 0xfb28 alt tav */ | |
| 6982 {'+', 0, 0}, /* 0xfb29 alt plus */ | |
| 6983 {0x5e9, 0x5c1, 0}, /* 0xfb2a shin+shin-dot */ | |
| 6984 {0x5e9, 0x5c2, 0}, /* 0xfb2b shin+sin-dot */ | |
| 6985 {0x5e9, 0x5c1, 0x5bc}, /* 0xfb2c shin+shin-dot+dagesh */ | |
| 6986 {0x5e9, 0x5c2, 0x5bc}, /* 0xfb2d shin+sin-dot+dagesh */ | |
| 6987 {0x5d0, 0x5b7, 0}, /* 0xfb2e alef+patah */ | |
| 6988 {0x5d0, 0x5b8, 0}, /* 0xfb2f alef+qamats */ | |
| 6989 {0x5d0, 0x5b4, 0}, /* 0xfb30 alef+hiriq */ | |
| 6990 {0x5d1, 0x5bc, 0}, /* 0xfb31 bet+dagesh */ | |
| 6991 {0x5d2, 0x5bc, 0}, /* 0xfb32 gimel+dagesh */ | |
| 6992 {0x5d3, 0x5bc, 0}, /* 0xfb33 dalet+dagesh */ | |
| 6993 {0x5d4, 0x5bc, 0}, /* 0xfb34 he+dagesh */ | |
| 6994 {0x5d5, 0x5bc, 0}, /* 0xfb35 vav+dagesh */ | |
| 6995 {0x5d6, 0x5bc, 0}, /* 0xfb36 zayin+dagesh */ | |
| 6996 {0xfb37, 0, 0}, /* 0xfb37 -- UNUSED */ | |
| 6997 {0x5d8, 0x5bc, 0}, /* 0xfb38 tet+dagesh */ | |
| 6998 {0x5d9, 0x5bc, 0}, /* 0xfb39 yud+dagesh */ | |
| 6999 {0x5da, 0x5bc, 0}, /* 0xfb3a kaf sofit+dagesh */ | |
| 7000 {0x5db, 0x5bc, 0}, /* 0xfb3b kaf+dagesh */ | |
| 7001 {0x5dc, 0x5bc, 0}, /* 0xfb3c lamed+dagesh */ | |
| 7002 {0xfb3d, 0, 0}, /* 0xfb3d -- UNUSED */ | |
| 7003 {0x5de, 0x5bc, 0}, /* 0xfb3e mem+dagesh */ | |
| 7004 {0xfb3f, 0, 0}, /* 0xfb3f -- UNUSED */ | |
| 7005 {0x5e0, 0x5bc, 0}, /* 0xfb40 nun+dagesh */ | |
| 7006 {0x5e1, 0x5bc, 0}, /* 0xfb41 samech+dagesh */ | |
| 7007 {0xfb42, 0, 0}, /* 0xfb42 -- UNUSED */ | |
| 7008 {0x5e3, 0x5bc, 0}, /* 0xfb43 pe sofit+dagesh */ | |
| 7009 {0x5e4, 0x5bc,0}, /* 0xfb44 pe+dagesh */ | |
| 7010 {0xfb45, 0, 0}, /* 0xfb45 -- UNUSED */ | |
| 7011 {0x5e6, 0x5bc, 0}, /* 0xfb46 tsadi+dagesh */ | |
| 7012 {0x5e7, 0x5bc, 0}, /* 0xfb47 qof+dagesh */ | |
| 7013 {0x5e8, 0x5bc, 0}, /* 0xfb48 resh+dagesh */ | |
| 7014 {0x5e9, 0x5bc, 0}, /* 0xfb49 shin+dagesh */ | |
| 7015 {0x5ea, 0x5bc, 0}, /* 0xfb4a tav+dagesh */ | |
| 7016 {0x5d5, 0x5b9, 0}, /* 0xfb4b vav+holam */ | |
| 7017 {0x5d1, 0x5bf, 0}, /* 0xfb4c bet+rafe */ | |
| 7018 {0x5db, 0x5bf, 0}, /* 0xfb4d kaf+rafe */ | |
| 7019 {0x5e4, 0x5bf, 0}, /* 0xfb4e pe+rafe */ | |
| 7020 {0x5d0, 0x5dc, 0} /* 0xfb4f alef-lamed */ | |
| 7021 }; | |
| 7022 | |
| 7023 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7024 mb_decompose(int c, int *c1, int *c2, int *c3) |
| 7 | 7025 { |
| 7026 decomp_T d; | |
| 7027 | |
|
4505
d037b9cbdaaa
updated for version 7.3.1000
Bram Moolenaar <bram@vim.org>
parents:
4466
diff
changeset
|
7028 if (c >= 0xfb20 && c <= 0xfb4f) |
| 7 | 7029 { |
| 7030 d = decomp_table[c - 0xfb20]; | |
| 7031 *c1 = d.a; | |
| 7032 *c2 = d.b; | |
| 7033 *c3 = d.c; | |
| 7034 } | |
| 7035 else | |
| 7036 { | |
| 7037 *c1 = c; | |
| 7038 *c2 = *c3 = 0; | |
| 7039 } | |
| 7040 } | |
| 7041 #endif | |
| 7042 | |
| 7043 /* | |
| 7044 * Compare two strings, ignore case if ireg_ic set. | |
| 7045 * Return 0 if strings match, non-zero otherwise. | |
| 7046 * Correct the length "*n" when composing characters are ignored. | |
| 7047 */ | |
| 7048 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7049 cstrncmp(char_u *s1, char_u *s2, int *n) |
| 7 | 7050 { |
| 7051 int result; | |
| 7052 | |
| 7053 if (!ireg_ic) | |
| 7054 result = STRNCMP(s1, s2, *n); | |
| 7055 else | |
| 7056 result = MB_STRNICMP(s1, s2, *n); | |
| 7057 | |
| 7058 #ifdef FEAT_MBYTE | |
| 7059 /* if it failed and it's utf8 and we want to combineignore: */ | |
| 7060 if (result != 0 && enc_utf8 && ireg_icombine) | |
| 7061 { | |
| 7062 char_u *str1, *str2; | |
| 7063 int c1, c2, c11, c12; | |
| 7064 int junk; | |
| 7065 | |
| 7066 /* we have to handle the strcmp ourselves, since it is necessary to | |
| 7067 * deal with the composing characters by ignoring them: */ | |
| 7068 str1 = s1; | |
| 7069 str2 = s2; | |
| 7070 c1 = c2 = 0; | |
| 507 | 7071 while ((int)(str1 - s1) < *n) |
| 7 | 7072 { |
| 7073 c1 = mb_ptr2char_adv(&str1); | |
| 7074 c2 = mb_ptr2char_adv(&str2); | |
| 7075 | |
| 7076 /* decompose the character if necessary, into 'base' characters | |
| 7077 * because I don't care about Arabic, I will hard-code the Hebrew | |
| 7078 * which I *do* care about! So sue me... */ | |
| 7079 if (c1 != c2 && (!ireg_ic || utf_fold(c1) != utf_fold(c2))) | |
| 7080 { | |
| 7081 /* decomposition necessary? */ | |
| 7082 mb_decompose(c1, &c11, &junk, &junk); | |
| 7083 mb_decompose(c2, &c12, &junk, &junk); | |
| 7084 c1 = c11; | |
| 7085 c2 = c12; | |
| 7086 if (c11 != c12 && (!ireg_ic || utf_fold(c11) != utf_fold(c12))) | |
| 7087 break; | |
| 7088 } | |
| 7089 } | |
| 7090 result = c2 - c1; | |
| 7091 if (result == 0) | |
| 7092 *n = (int)(str2 - s2); | |
| 7093 } | |
| 7094 #endif | |
| 7095 | |
| 7096 return result; | |
| 7097 } | |
| 7098 | |
| 7099 /* | |
| 7100 * cstrchr: This function is used a lot for simple searches, keep it fast! | |
| 7101 */ | |
| 7102 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7103 cstrchr(char_u *s, int c) |
| 7 | 7104 { |
| 7105 char_u *p; | |
| 7106 int cc; | |
| 7107 | |
| 7108 if (!ireg_ic | |
| 7109 #ifdef FEAT_MBYTE | |
| 7110 || (!enc_utf8 && mb_char2len(c) > 1) | |
| 7111 #endif | |
| 7112 ) | |
| 7113 return vim_strchr(s, c); | |
| 7114 | |
| 7115 /* tolower() and toupper() can be slow, comparing twice should be a lot | |
| 7116 * faster (esp. when using MS Visual C++!). | |
| 7117 * For UTF-8 need to use folded case. */ | |
| 7118 #ifdef FEAT_MBYTE | |
| 7119 if (enc_utf8 && c > 0x80) | |
| 7120 cc = utf_fold(c); | |
| 7121 else | |
| 7122 #endif | |
| 1347 | 7123 if (MB_ISUPPER(c)) |
| 7124 cc = MB_TOLOWER(c); | |
| 7125 else if (MB_ISLOWER(c)) | |
| 7126 cc = MB_TOUPPER(c); | |
| 7 | 7127 else |
| 7128 return vim_strchr(s, c); | |
| 7129 | |
| 7130 #ifdef FEAT_MBYTE | |
| 7131 if (has_mbyte) | |
| 7132 { | |
| 474 | 7133 for (p = s; *p != NUL; p += (*mb_ptr2len)(p)) |
| 7 | 7134 { |
| 7135 if (enc_utf8 && c > 0x80) | |
| 7136 { | |
| 7137 if (utf_fold(utf_ptr2char(p)) == cc) | |
| 7138 return p; | |
| 7139 } | |
| 7140 else if (*p == c || *p == cc) | |
| 7141 return p; | |
| 7142 } | |
| 7143 } | |
| 7144 else | |
| 7145 #endif | |
| 7146 /* Faster version for when there are no multi-byte characters. */ | |
| 7147 for (p = s; *p != NUL; ++p) | |
| 7148 if (*p == c || *p == cc) | |
| 7149 return p; | |
| 7150 | |
| 7151 return NULL; | |
| 7152 } | |
| 7153 | |
| 7154 /*************************************************************** | |
| 7155 * regsub stuff * | |
| 7156 ***************************************************************/ | |
| 7157 | |
| 7158 /* | |
| 7159 * We should define ftpr as a pointer to a function returning a pointer to | |
| 7160 * a function returning a pointer to a function ... | |
| 7161 * This is impossible, so we declare a pointer to a function returning a | |
| 7162 * pointer to a function returning void. This should work for all compilers. | |
| 7163 */ | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7164 typedef void (*(*fptr_T)(int *, int))(); |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7165 |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7166 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
|
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_lower(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 |
|
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7171 static int vim_regsub_both(char_u *source, char_u *dest, int copy, int magic, int backslash); |
| 7 | 7172 |
| 772 | 7173 static fptr_T |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7174 do_upper(int *d, int c) |
| 7 | 7175 { |
| 772 | 7176 *d = MB_TOUPPER(c); |
| 7177 | |
| 7178 return (fptr_T)NULL; | |
| 7 | 7179 } |
| 7180 | |
| 772 | 7181 static fptr_T |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7182 do_Upper(int *d, int c) |
| 7 | 7183 { |
| 772 | 7184 *d = MB_TOUPPER(c); |
| 7185 | |
| 7186 return (fptr_T)do_Upper; | |
| 7187 } | |
| 7188 | |
| 7189 static fptr_T | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7190 do_lower(int *d, int c) |
| 772 | 7191 { |
| 7192 *d = MB_TOLOWER(c); | |
| 7193 | |
| 7194 return (fptr_T)NULL; | |
| 7195 } | |
| 7196 | |
| 7197 static fptr_T | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7198 do_Lower(int *d, int c) |
| 772 | 7199 { |
| 7200 *d = MB_TOLOWER(c); | |
| 7201 | |
| 7202 return (fptr_T)do_Lower; | |
| 7 | 7203 } |
| 7204 | |
| 7205 /* | |
| 7206 * regtilde(): Replace tildes in the pattern by the old pattern. | |
| 7207 * | |
| 7208 * Short explanation of the tilde: It stands for the previous replacement | |
| 7209 * pattern. If that previous pattern also contains a ~ we should go back a | |
| 7210 * step further... But we insert the previous pattern into the current one | |
| 7211 * and remember that. | |
| 772 | 7212 * This still does not handle the case where "magic" changes. So require the |
| 7213 * user to keep his hands off of "magic". | |
| 7 | 7214 * |
| 7215 * The tildes are parsed once before the first call to vim_regsub(). | |
| 7216 */ | |
| 7217 char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7218 regtilde(char_u *source, int magic) |
| 7 | 7219 { |
| 7220 char_u *newsub = source; | |
| 7221 char_u *tmpsub; | |
| 7222 char_u *p; | |
| 7223 int len; | |
| 7224 int prevlen; | |
| 7225 | |
| 7226 for (p = newsub; *p; ++p) | |
| 7227 { | |
| 7228 if ((*p == '~' && magic) || (*p == '\\' && *(p + 1) == '~' && !magic)) | |
| 7229 { | |
| 7230 if (reg_prev_sub != NULL) | |
| 7231 { | |
| 7232 /* length = len(newsub) - 1 + len(prev_sub) + 1 */ | |
| 7233 prevlen = (int)STRLEN(reg_prev_sub); | |
| 7234 tmpsub = alloc((unsigned)(STRLEN(newsub) + prevlen)); | |
| 7235 if (tmpsub != NULL) | |
| 7236 { | |
| 7237 /* copy prefix */ | |
| 7238 len = (int)(p - newsub); /* not including ~ */ | |
| 7239 mch_memmove(tmpsub, newsub, (size_t)len); | |
| 1209 | 7240 /* interpret tilde */ |
| 7 | 7241 mch_memmove(tmpsub + len, reg_prev_sub, (size_t)prevlen); |
| 7242 /* copy postfix */ | |
| 7243 if (!magic) | |
| 7244 ++p; /* back off \ */ | |
| 7245 STRCPY(tmpsub + len + prevlen, p + 1); | |
| 7246 | |
| 7247 if (newsub != source) /* already allocated newsub */ | |
| 7248 vim_free(newsub); | |
| 7249 newsub = tmpsub; | |
| 7250 p = newsub + len + prevlen; | |
| 7251 } | |
| 7252 } | |
| 7253 else if (magic) | |
| 1621 | 7254 STRMOVE(p, p + 1); /* remove '~' */ |
| 7 | 7255 else |
| 1621 | 7256 STRMOVE(p, p + 2); /* remove '\~' */ |
| 7 | 7257 --p; |
| 7258 } | |
| 7259 else | |
| 7260 { | |
| 7261 if (*p == '\\' && p[1]) /* skip escaped characters */ | |
| 7262 ++p; | |
| 7263 #ifdef FEAT_MBYTE | |
| 7264 if (has_mbyte) | |
| 474 | 7265 p += (*mb_ptr2len)(p) - 1; |
| 7 | 7266 #endif |
| 7267 } | |
| 7268 } | |
| 7269 | |
| 7270 vim_free(reg_prev_sub); | |
| 7271 if (newsub != source) /* newsub was allocated, just keep it */ | |
| 7272 reg_prev_sub = newsub; | |
| 7273 else /* no ~ found, need to save newsub */ | |
| 7274 reg_prev_sub = vim_strsave(newsub); | |
| 7275 return newsub; | |
| 7276 } | |
| 7277 | |
| 7278 #ifdef FEAT_EVAL | |
| 7279 static int can_f_submatch = FALSE; /* TRUE when submatch() can be used */ | |
| 7280 | |
| 7281 /* These pointers are used instead of reg_match and reg_mmatch for | |
| 7282 * reg_submatch(). Needed for when the substitution string is an expression | |
| 7283 * that contains a call to substitute() and submatch(). */ | |
| 7284 static regmatch_T *submatch_match; | |
| 7285 static regmmatch_T *submatch_mmatch; | |
| 2011 | 7286 static linenr_T submatch_firstlnum; |
| 7287 static linenr_T submatch_maxline; | |
| 2904 | 7288 static int submatch_line_lbr; |
| 7 | 7289 #endif |
| 7290 | |
| 7291 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO) | |
| 7292 /* | |
| 7293 * vim_regsub() - perform substitutions after a vim_regexec() or | |
| 7294 * vim_regexec_multi() match. | |
| 7295 * | |
| 7296 * If "copy" is TRUE really copy into "dest". | |
| 7297 * If "copy" is FALSE nothing is copied, this is just to find out the length | |
| 7298 * of the result. | |
| 7299 * | |
| 7300 * If "backslash" is TRUE, a backslash will be removed later, need to double | |
| 7301 * them to keep them, and insert a backslash before a CR to avoid it being | |
| 7302 * replaced with a line break later. | |
| 7303 * | |
| 7304 * Note: The matched text must not change between the call of | |
| 7305 * vim_regexec()/vim_regexec_multi() and vim_regsub()! It would make the back | |
| 7306 * references invalid! | |
| 7307 * | |
| 7308 * Returns the size of the replacement, including terminating NUL. | |
| 7309 */ | |
| 7310 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7311 vim_regsub( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7312 regmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7313 char_u *source, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7314 char_u *dest, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7315 int copy, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7316 int magic, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7317 int backslash) |
| 7 | 7318 { |
| 7319 reg_match = rmp; | |
| 7320 reg_mmatch = NULL; | |
| 7321 reg_maxline = 0; | |
| 4061 | 7322 reg_buf = curbuf; |
| 5836 | 7323 reg_line_lbr = TRUE; |
| 7 | 7324 return vim_regsub_both(source, dest, copy, magic, backslash); |
| 7325 } | |
| 7326 #endif | |
| 7327 | |
| 7328 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7329 vim_regsub_multi( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7330 regmmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7331 linenr_T lnum, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7332 char_u *source, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7333 char_u *dest, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7334 int copy, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7335 int magic, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7336 int backslash) |
| 7 | 7337 { |
| 7338 reg_match = NULL; | |
| 7339 reg_mmatch = rmp; | |
| 7340 reg_buf = curbuf; /* always works on the current buffer! */ | |
| 7341 reg_firstlnum = lnum; | |
| 7342 reg_maxline = curbuf->b_ml.ml_line_count - lnum; | |
| 5836 | 7343 reg_line_lbr = FALSE; |
| 7 | 7344 return vim_regsub_both(source, dest, copy, magic, backslash); |
| 7345 } | |
| 7346 | |
| 7347 static int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7348 vim_regsub_both( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7349 char_u *source, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7350 char_u *dest, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7351 int copy, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7352 int magic, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7353 int backslash) |
| 7 | 7354 { |
| 7355 char_u *src; | |
| 7356 char_u *dst; | |
| 7357 char_u *s; | |
| 7358 int c; | |
| 772 | 7359 int cc; |
| 7 | 7360 int no = -1; |
| 4244 | 7361 fptr_T func_all = (fptr_T)NULL; |
| 7362 fptr_T func_one = (fptr_T)NULL; | |
| 7 | 7363 linenr_T clnum = 0; /* init for GCC */ |
| 7364 int len = 0; /* init for GCC */ | |
| 7365 #ifdef FEAT_EVAL | |
| 7366 static char_u *eval_result = NULL; | |
| 7367 #endif | |
| 7368 | |
| 7369 /* Be paranoid... */ | |
| 7370 if (source == NULL || dest == NULL) | |
| 7371 { | |
| 7372 EMSG(_(e_null)); | |
| 7373 return 0; | |
| 7374 } | |
| 7375 if (prog_magic_wrong()) | |
| 7376 return 0; | |
| 7377 src = source; | |
| 7378 dst = dest; | |
| 7379 | |
| 7380 /* | |
| 7381 * When the substitute part starts with "\=" evaluate it as an expression. | |
| 7382 */ | |
| 7383 if (source[0] == '\\' && source[1] == '=' | |
| 7384 #ifdef FEAT_EVAL | |
| 7385 && !can_f_submatch /* can't do this recursively */ | |
| 7386 #endif | |
| 7387 ) | |
| 7388 { | |
| 7389 #ifdef FEAT_EVAL | |
| 7390 /* To make sure that the length doesn't change between checking the | |
| 7391 * length and copying the string, and to speed up things, the | |
| 7392 * resulting string is saved from the call with "copy" == FALSE to the | |
| 7393 * call with "copy" == TRUE. */ | |
| 7394 if (copy) | |
| 7395 { | |
| 7396 if (eval_result != NULL) | |
| 7397 { | |
| 7398 STRCPY(dest, eval_result); | |
| 7399 dst += STRLEN(eval_result); | |
| 7400 vim_free(eval_result); | |
| 7401 eval_result = NULL; | |
| 7402 } | |
| 7403 } | |
| 7404 else | |
| 7405 { | |
| 7406 win_T *save_reg_win; | |
| 7407 int save_ireg_ic; | |
| 7408 | |
| 7409 vim_free(eval_result); | |
| 7410 | |
| 7411 /* The expression may contain substitute(), which calls us | |
| 7412 * recursively. Make sure submatch() gets the text from the first | |
| 7413 * level. Don't need to save "reg_buf", because | |
| 7414 * vim_regexec_multi() can't be called recursively. */ | |
| 7415 submatch_match = reg_match; | |
| 7416 submatch_mmatch = reg_mmatch; | |
| 2011 | 7417 submatch_firstlnum = reg_firstlnum; |
| 7418 submatch_maxline = reg_maxline; | |
| 2904 | 7419 submatch_line_lbr = reg_line_lbr; |
| 7 | 7420 save_reg_win = reg_win; |
| 7421 save_ireg_ic = ireg_ic; | |
| 7422 can_f_submatch = TRUE; | |
| 7423 | |
| 714 | 7424 eval_result = eval_to_string(source + 2, NULL, TRUE); |
| 7 | 7425 if (eval_result != NULL) |
| 7426 { | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7427 int had_backslash = FALSE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7428 |
| 39 | 7429 for (s = eval_result; *s != NUL; mb_ptr_adv(s)) |
| 7 | 7430 { |
| 2904 | 7431 /* Change NL to CR, so that it becomes a line break, |
| 7432 * unless called from vim_regexec_nl(). | |
| 7 | 7433 * Skip over a backslashed character. */ |
| 2904 | 7434 if (*s == NL && !submatch_line_lbr) |
| 7 | 7435 *s = CAR; |
| 7436 else if (*s == '\\' && s[1] != NUL) | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7437 { |
| 7 | 7438 ++s; |
| 2173 | 7439 /* Change NL to CR here too, so that this works: |
| 7440 * :s/abc\\\ndef/\="aaa\\\nbbb"/ on text: | |
| 7441 * abc\ | |
| 7442 * def | |
| 2904 | 7443 * Not when called from vim_regexec_nl(). |
| 2173 | 7444 */ |
| 2904 | 7445 if (*s == NL && !submatch_line_lbr) |
| 2173 | 7446 *s = CAR; |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7447 had_backslash = TRUE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7448 } |
|
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 if (had_backslash && backslash) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7451 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7452 /* Backslashes will be consumed, need to double them. */ |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7453 s = vim_strsave_escaped(eval_result, (char_u *)"\\"); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7454 if (s != NULL) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7455 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7456 vim_free(eval_result); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7457 eval_result = s; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7458 } |
| 7 | 7459 } |
| 7460 | |
| 7461 dst += STRLEN(eval_result); | |
| 7462 } | |
| 7463 | |
| 7464 reg_match = submatch_match; | |
| 7465 reg_mmatch = submatch_mmatch; | |
| 2011 | 7466 reg_firstlnum = submatch_firstlnum; |
| 7467 reg_maxline = submatch_maxline; | |
| 2904 | 7468 reg_line_lbr = submatch_line_lbr; |
| 7 | 7469 reg_win = save_reg_win; |
| 7470 ireg_ic = save_ireg_ic; | |
| 7471 can_f_submatch = FALSE; | |
| 7472 } | |
| 7473 #endif | |
| 7474 } | |
| 7475 else | |
| 7476 while ((c = *src++) != NUL) | |
| 7477 { | |
| 7478 if (c == '&' && magic) | |
| 7479 no = 0; | |
| 7480 else if (c == '\\' && *src != NUL) | |
| 7481 { | |
| 7482 if (*src == '&' && !magic) | |
| 7483 { | |
| 7484 ++src; | |
| 7485 no = 0; | |
| 7486 } | |
| 7487 else if ('0' <= *src && *src <= '9') | |
| 7488 { | |
| 7489 no = *src++ - '0'; | |
| 7490 } | |
| 7491 else if (vim_strchr((char_u *)"uUlLeE", *src)) | |
| 7492 { | |
| 7493 switch (*src++) | |
| 7494 { | |
| 4244 | 7495 case 'u': func_one = (fptr_T)do_upper; |
| 7 | 7496 continue; |
| 4244 | 7497 case 'U': func_all = (fptr_T)do_Upper; |
| 7 | 7498 continue; |
| 4244 | 7499 case 'l': func_one = (fptr_T)do_lower; |
| 7 | 7500 continue; |
| 4244 | 7501 case 'L': func_all = (fptr_T)do_Lower; |
| 7 | 7502 continue; |
| 7503 case 'e': | |
| 4244 | 7504 case 'E': func_one = func_all = (fptr_T)NULL; |
| 7 | 7505 continue; |
| 7506 } | |
| 7507 } | |
| 7508 } | |
| 7509 if (no < 0) /* Ordinary character. */ | |
| 7510 { | |
| 798 | 7511 if (c == K_SPECIAL && src[0] != NUL && src[1] != NUL) |
| 7512 { | |
| 1209 | 7513 /* Copy a special key as-is. */ |
| 798 | 7514 if (copy) |
| 7515 { | |
| 7516 *dst++ = c; | |
| 7517 *dst++ = *src++; | |
| 7518 *dst++ = *src++; | |
| 7519 } | |
| 7520 else | |
| 7521 { | |
| 7522 dst += 3; | |
| 7523 src += 2; | |
| 7524 } | |
| 7525 continue; | |
| 7526 } | |
| 7527 | |
| 7 | 7528 if (c == '\\' && *src != NUL) |
| 7529 { | |
| 7530 /* Check for abbreviations -- webb */ | |
| 7531 switch (*src) | |
| 7532 { | |
| 7533 case 'r': c = CAR; ++src; break; | |
| 7534 case 'n': c = NL; ++src; break; | |
| 7535 case 't': c = TAB; ++src; break; | |
| 7536 /* Oh no! \e already has meaning in subst pat :-( */ | |
| 7537 /* case 'e': c = ESC; ++src; break; */ | |
| 7538 case 'b': c = Ctrl_H; ++src; break; | |
| 7539 | |
| 7540 /* If "backslash" is TRUE the backslash will be removed | |
| 7541 * later. Used to insert a literal CR. */ | |
| 7542 default: if (backslash) | |
| 7543 { | |
| 7544 if (copy) | |
| 7545 *dst = '\\'; | |
| 7546 ++dst; | |
| 7547 } | |
| 7548 c = *src++; | |
| 7549 } | |
| 7550 } | |
| 798 | 7551 #ifdef FEAT_MBYTE |
| 7552 else if (has_mbyte) | |
| 7553 c = mb_ptr2char(src - 1); | |
| 7554 #endif | |
| 7 | 7555 |
| 7556 /* Write to buffer, if copy is set. */ | |
| 4244 | 7557 if (func_one != (fptr_T)NULL) |
| 7558 /* Turbo C complains without the typecast */ | |
| 7559 func_one = (fptr_T)(func_one(&cc, c)); | |
| 7560 else if (func_all != (fptr_T)NULL) | |
| 7561 /* Turbo C complains without the typecast */ | |
| 7562 func_all = (fptr_T)(func_all(&cc, c)); | |
| 7563 else /* just copy */ | |
| 772 | 7564 cc = c; |
| 7565 | |
| 7566 #ifdef FEAT_MBYTE | |
| 7567 if (has_mbyte) | |
| 7 | 7568 { |
|
2307
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7569 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
|
7570 |
| 7 | 7571 if (copy) |
| 772 | 7572 mb_char2bytes(cc, dst); |
| 7573 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
|
7574 if (enc_utf8) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7575 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7576 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
|
7577 |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7578 /* 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
|
7579 * 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
|
7580 if (clen < totlen) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7581 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7582 if (copy) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7583 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
|
7584 (size_t)(totlen - clen)); |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7585 dst += totlen - clen; |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7586 } |
|
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 src += totlen - 1; |
| 7 | 7589 } |
| 7590 else | |
| 7591 #endif | |
| 7592 if (copy) | |
| 772 | 7593 *dst = cc; |
| 7 | 7594 dst++; |
| 7595 } | |
| 7596 else | |
| 7597 { | |
| 7598 if (REG_MULTI) | |
| 7599 { | |
| 7600 clnum = reg_mmatch->startpos[no].lnum; | |
| 7601 if (clnum < 0 || reg_mmatch->endpos[no].lnum < 0) | |
| 7602 s = NULL; | |
| 7603 else | |
| 7604 { | |
| 7605 s = reg_getline(clnum) + reg_mmatch->startpos[no].col; | |
| 7606 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7607 len = reg_mmatch->endpos[no].col | |
| 7608 - reg_mmatch->startpos[no].col; | |
| 7609 else | |
| 7610 len = (int)STRLEN(s); | |
| 7611 } | |
| 7612 } | |
| 7613 else | |
| 7614 { | |
| 7615 s = reg_match->startp[no]; | |
| 7616 if (reg_match->endp[no] == NULL) | |
| 7617 s = NULL; | |
| 7618 else | |
| 7619 len = (int)(reg_match->endp[no] - s); | |
| 7620 } | |
| 7621 if (s != NULL) | |
| 7622 { | |
| 7623 for (;;) | |
| 7624 { | |
| 7625 if (len == 0) | |
| 7626 { | |
| 7627 if (REG_MULTI) | |
| 7628 { | |
| 7629 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7630 break; | |
| 7631 if (copy) | |
| 7632 *dst = CAR; | |
| 7633 ++dst; | |
| 7634 s = reg_getline(++clnum); | |
| 7635 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7636 len = reg_mmatch->endpos[no].col; | |
| 7637 else | |
| 7638 len = (int)STRLEN(s); | |
| 7639 } | |
| 7640 else | |
| 7641 break; | |
| 7642 } | |
| 7643 else if (*s == NUL) /* we hit NUL. */ | |
| 7644 { | |
| 7645 if (copy) | |
| 7646 EMSG(_(e_re_damg)); | |
| 7647 goto exit; | |
| 7648 } | |
| 7649 else | |
| 7650 { | |
| 7651 if (backslash && (*s == CAR || *s == '\\')) | |
| 7652 { | |
| 7653 /* | |
| 7654 * Insert a backslash in front of a CR, otherwise | |
| 7655 * it will be replaced by a line break. | |
| 7656 * Number of backslashes will be halved later, | |
| 7657 * double them here. | |
| 7658 */ | |
| 7659 if (copy) | |
| 7660 { | |
| 7661 dst[0] = '\\'; | |
| 7662 dst[1] = *s; | |
| 7663 } | |
| 7664 dst += 2; | |
| 7665 } | |
| 7666 else | |
| 7667 { | |
| 772 | 7668 #ifdef FEAT_MBYTE |
| 7669 if (has_mbyte) | |
| 7670 c = mb_ptr2char(s); | |
| 7671 else | |
| 7672 #endif | |
| 7673 c = *s; | |
| 7674 | |
| 4244 | 7675 if (func_one != (fptr_T)NULL) |
| 7676 /* Turbo C complains without the typecast */ | |
| 7677 func_one = (fptr_T)(func_one(&cc, c)); | |
| 7678 else if (func_all != (fptr_T)NULL) | |
| 7679 /* Turbo C complains without the typecast */ | |
| 7680 func_all = (fptr_T)(func_all(&cc, c)); | |
| 7681 else /* just copy */ | |
| 772 | 7682 cc = c; |
| 7683 | |
| 7684 #ifdef FEAT_MBYTE | |
| 7685 if (has_mbyte) | |
| 7 | 7686 { |
| 1332 | 7687 int l; |
| 7688 | |
| 7689 /* Copy composing characters separately, one | |
| 7690 * at a time. */ | |
| 7691 if (enc_utf8) | |
| 7692 l = utf_ptr2len(s) - 1; | |
| 7693 else | |
| 7694 l = mb_ptr2len(s) - 1; | |
| 772 | 7695 |
| 7696 s += l; | |
| 7697 len -= l; | |
| 7698 if (copy) | |
| 7699 mb_char2bytes(cc, dst); | |
| 7700 dst += mb_char2len(cc) - 1; | |
| 7 | 7701 } |
| 772 | 7702 else |
| 7703 #endif | |
| 7704 if (copy) | |
| 7705 *dst = cc; | |
| 7706 dst++; | |
| 7 | 7707 } |
| 772 | 7708 |
| 7 | 7709 ++s; |
| 7710 --len; | |
| 7711 } | |
| 7712 } | |
| 7713 } | |
| 7714 no = -1; | |
| 7715 } | |
| 7716 } | |
| 7717 if (copy) | |
| 7718 *dst = NUL; | |
| 7719 | |
| 7720 exit: | |
| 7721 return (int)((dst - dest) + 1); | |
| 7722 } | |
| 7723 | |
| 7724 #ifdef FEAT_EVAL | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
7725 static char_u *reg_getline_submatch(linenr_T lnum); |
| 2012 | 7726 |
| 7 | 7727 /* |
| 2011 | 7728 * Call reg_getline() with the line numbers from the submatch. If a |
| 7729 * substitute() was used the reg_maxline and other values have been | |
| 7730 * overwritten. | |
| 7731 */ | |
| 7732 static char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7733 reg_getline_submatch(linenr_T lnum) |
| 2011 | 7734 { |
| 7735 char_u *s; | |
| 7736 linenr_T save_first = reg_firstlnum; | |
| 7737 linenr_T save_max = reg_maxline; | |
| 7738 | |
| 7739 reg_firstlnum = submatch_firstlnum; | |
| 7740 reg_maxline = submatch_maxline; | |
| 7741 | |
| 7742 s = reg_getline(lnum); | |
| 7743 | |
| 7744 reg_firstlnum = save_first; | |
| 7745 reg_maxline = save_max; | |
| 7746 return s; | |
| 7747 } | |
| 7748 | |
| 7749 /* | |
| 1209 | 7750 * Used for the submatch() function: get the string from the n'th submatch in |
| 7 | 7751 * allocated memory. |
| 7752 * Returns NULL when not in a ":s" command and for a non-existing submatch. | |
| 7753 */ | |
| 7754 char_u * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7755 reg_submatch(int no) |
| 7 | 7756 { |
| 7757 char_u *retval = NULL; | |
| 7758 char_u *s; | |
| 7759 int len; | |
| 7760 int round; | |
| 7761 linenr_T lnum; | |
| 7762 | |
| 840 | 7763 if (!can_f_submatch || no < 0) |
| 7 | 7764 return NULL; |
| 7765 | |
| 7766 if (submatch_match == NULL) | |
| 7767 { | |
| 7768 /* | |
| 7769 * First round: compute the length and allocate memory. | |
| 7770 * Second round: copy the text. | |
| 7771 */ | |
| 7772 for (round = 1; round <= 2; ++round) | |
| 7773 { | |
| 7774 lnum = submatch_mmatch->startpos[no].lnum; | |
| 7775 if (lnum < 0 || submatch_mmatch->endpos[no].lnum < 0) | |
| 7776 return NULL; | |
| 7777 | |
| 2011 | 7778 s = reg_getline_submatch(lnum) + submatch_mmatch->startpos[no].col; |
| 7 | 7779 if (s == NULL) /* anti-crash check, cannot happen? */ |
| 7780 break; | |
| 7781 if (submatch_mmatch->endpos[no].lnum == lnum) | |
| 7782 { | |
| 7783 /* Within one line: take form start to end col. */ | |
| 7784 len = submatch_mmatch->endpos[no].col | |
| 7785 - submatch_mmatch->startpos[no].col; | |
| 7786 if (round == 2) | |
| 418 | 7787 vim_strncpy(retval, s, len); |
| 7 | 7788 ++len; |
| 7789 } | |
| 7790 else | |
| 7791 { | |
| 7792 /* Multiple lines: take start line from start col, middle | |
| 7793 * lines completely and end line up to end col. */ | |
| 7794 len = (int)STRLEN(s); | |
| 7795 if (round == 2) | |
| 7796 { | |
| 7797 STRCPY(retval, s); | |
| 7798 retval[len] = '\n'; | |
| 7799 } | |
| 7800 ++len; | |
| 7801 ++lnum; | |
| 7802 while (lnum < submatch_mmatch->endpos[no].lnum) | |
| 7803 { | |
| 2011 | 7804 s = reg_getline_submatch(lnum++); |
| 7 | 7805 if (round == 2) |
| 7806 STRCPY(retval + len, s); | |
| 7807 len += (int)STRLEN(s); | |
| 7808 if (round == 2) | |
| 7809 retval[len] = '\n'; | |
| 7810 ++len; | |
| 7811 } | |
| 7812 if (round == 2) | |
| 2011 | 7813 STRNCPY(retval + len, reg_getline_submatch(lnum), |
| 7 | 7814 submatch_mmatch->endpos[no].col); |
| 7815 len += submatch_mmatch->endpos[no].col; | |
| 7816 if (round == 2) | |
| 7817 retval[len] = NUL; | |
| 7818 ++len; | |
| 7819 } | |
| 7820 | |
| 840 | 7821 if (retval == NULL) |
| 7 | 7822 { |
| 7823 retval = lalloc((long_u)len, TRUE); | |
| 840 | 7824 if (retval == NULL) |
| 7 | 7825 return NULL; |
| 7826 } | |
| 7827 } | |
| 7828 } | |
| 7829 else | |
| 7830 { | |
| 1815 | 7831 s = submatch_match->startp[no]; |
| 7832 if (s == NULL || submatch_match->endp[no] == NULL) | |
| 7 | 7833 retval = NULL; |
| 7834 else | |
| 7835 retval = vim_strnsave(s, (int)(submatch_match->endp[no] - s)); | |
| 7836 } | |
| 7837 | |
| 7838 return retval; | |
| 7839 } | |
| 5794 | 7840 |
| 7841 /* | |
| 7842 * Used for the submatch() function with the optional non-zero argument: get | |
| 7843 * the list of strings from the n'th submatch in allocated memory with NULs | |
| 7844 * represented in NLs. | |
| 7845 * Returns a list of allocated strings. Returns NULL when not in a ":s" | |
| 7846 * command, for a non-existing submatch and for any error. | |
| 7847 */ | |
| 7848 list_T * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7849 reg_submatch_list(int no) |
| 5794 | 7850 { |
| 7851 char_u *s; | |
| 7852 linenr_T slnum; | |
| 7853 linenr_T elnum; | |
| 7854 colnr_T scol; | |
| 7855 colnr_T ecol; | |
| 7856 int i; | |
| 7857 list_T *list; | |
| 7858 int error = FALSE; | |
| 7859 | |
| 7860 if (!can_f_submatch || no < 0) | |
| 7861 return NULL; | |
| 7862 | |
| 7863 if (submatch_match == NULL) | |
| 7864 { | |
| 7865 slnum = submatch_mmatch->startpos[no].lnum; | |
| 7866 elnum = submatch_mmatch->endpos[no].lnum; | |
| 7867 if (slnum < 0 || elnum < 0) | |
| 7868 return NULL; | |
| 7869 | |
| 7870 scol = submatch_mmatch->startpos[no].col; | |
| 7871 ecol = submatch_mmatch->endpos[no].col; | |
| 7872 | |
| 7873 list = list_alloc(); | |
| 7874 if (list == NULL) | |
| 7875 return NULL; | |
| 7876 | |
| 7877 s = reg_getline_submatch(slnum) + scol; | |
| 7878 if (slnum == elnum) | |
| 7879 { | |
| 7880 if (list_append_string(list, s, ecol - scol) == FAIL) | |
| 7881 error = TRUE; | |
| 7882 } | |
| 7883 else | |
| 7884 { | |
| 7885 if (list_append_string(list, s, -1) == FAIL) | |
| 7886 error = TRUE; | |
| 7887 for (i = 1; i < elnum - slnum; i++) | |
| 7888 { | |
| 7889 s = reg_getline_submatch(slnum + i); | |
| 7890 if (list_append_string(list, s, -1) == FAIL) | |
| 7891 error = TRUE; | |
| 7892 } | |
| 7893 s = reg_getline_submatch(elnum); | |
| 7894 if (list_append_string(list, s, ecol) == FAIL) | |
| 7895 error = TRUE; | |
| 7896 } | |
| 7897 } | |
| 7898 else | |
| 7899 { | |
| 7900 s = submatch_match->startp[no]; | |
| 7901 if (s == NULL || submatch_match->endp[no] == NULL) | |
| 7902 return NULL; | |
| 7903 list = list_alloc(); | |
| 7904 if (list == NULL) | |
| 7905 return NULL; | |
| 7906 if (list_append_string(list, s, | |
| 7907 (int)(submatch_match->endp[no] - s)) == FAIL) | |
| 7908 error = TRUE; | |
| 7909 } | |
| 7910 | |
| 7911 if (error) | |
| 7912 { | |
| 7913 list_free(list, TRUE); | |
| 7914 return NULL; | |
| 7915 } | |
| 7916 return list; | |
| 7917 } | |
| 7 | 7918 #endif |
| 4444 | 7919 |
| 7920 static regengine_T bt_regengine = | |
| 7921 { | |
| 7922 bt_regcomp, | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7923 bt_regfree, |
| 4444 | 7924 bt_regexec_nl, |
| 6328 | 7925 bt_regexec_multi, |
| 7926 (char_u *)"" | |
| 4444 | 7927 }; |
| 7928 | |
| 7929 #include "regexp_nfa.c" | |
| 7930 | |
| 7931 static regengine_T nfa_regengine = | |
| 7932 { | |
| 7933 nfa_regcomp, | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7934 nfa_regfree, |
| 4444 | 7935 nfa_regexec_nl, |
| 6328 | 7936 nfa_regexec_multi, |
| 7937 (char_u *)"" | |
| 4444 | 7938 }; |
| 7939 | |
| 7940 /* Which regexp engine to use? Needed for vim_regcomp(). | |
| 7941 * Must match with 'regexpengine'. */ | |
| 7942 static int regexp_engine = 0; | |
| 6328 | 7943 |
| 4444 | 7944 #ifdef DEBUG |
| 7945 static char_u regname[][30] = { | |
| 7946 "AUTOMATIC Regexp Engine", | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
7947 "BACKTRACKING Regexp Engine", |
| 4444 | 7948 "NFA Regexp Engine" |
| 7949 }; | |
| 7950 #endif | |
| 7951 | |
| 7952 /* | |
| 7953 * Compile a regular expression into internal code. | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7954 * Returns the program in allocated memory. |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7955 * Use vim_regfree() to free the memory. |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7956 * Returns NULL for an error. |
| 4444 | 7957 */ |
| 7958 regprog_T * | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
7959 vim_regcomp(char_u *expr_arg, int re_flags) |
| 4444 | 7960 { |
| 7961 regprog_T *prog = NULL; | |
| 7962 char_u *expr = expr_arg; | |
| 7963 | |
| 7964 regexp_engine = p_re; | |
| 7965 | |
| 7966 /* Check for prefix "\%#=", that sets the regexp engine */ | |
| 7967 if (STRNCMP(expr, "\\%#=", 4) == 0) | |
| 7968 { | |
| 7969 int newengine = expr[4] - '0'; | |
| 7970 | |
| 7971 if (newengine == AUTOMATIC_ENGINE | |
| 7972 || newengine == BACKTRACKING_ENGINE | |
| 7973 || newengine == NFA_ENGINE) | |
| 7974 { | |
| 7975 regexp_engine = expr[4] - '0'; | |
| 7976 expr += 5; | |
| 7977 #ifdef DEBUG | |
| 5897 | 7978 smsg((char_u *)"New regexp mode selected (%d): %s", |
| 7979 regexp_engine, regname[newengine]); | |
| 4444 | 7980 #endif |
| 7981 } | |
| 7982 else | |
| 7983 { | |
| 7984 EMSG(_("E864: \\%#= can only be followed by 0, 1, or 2. The automatic engine will be used ")); | |
| 7985 regexp_engine = AUTOMATIC_ENGINE; | |
| 7986 } | |
| 7987 } | |
| 7988 bt_regengine.expr = expr; | |
| 7989 nfa_regengine.expr = expr; | |
| 7990 | |
| 7991 /* | |
| 7992 * First try the NFA engine, unless backtracking was requested. | |
| 7993 */ | |
| 7994 if (regexp_engine != BACKTRACKING_ENGINE) | |
| 6533 | 7995 prog = nfa_regengine.regcomp(expr, |
| 7996 re_flags + (regexp_engine == AUTOMATIC_ENGINE ? RE_AUTO : 0)); | |
| 4444 | 7997 else |
| 7998 prog = bt_regengine.regcomp(expr, re_flags); | |
| 7999 | |
| 6328 | 8000 /* Check for error compiling regexp with initial engine. */ |
| 8001 if (prog == NULL) | |
| 4444 | 8002 { |
| 4460 | 8003 #ifdef BT_REGEXP_DEBUG_LOG |
| 4444 | 8004 if (regexp_engine != BACKTRACKING_ENGINE) /* debugging log for NFA */ |
| 8005 { | |
| 8006 FILE *f; | |
| 4460 | 8007 f = fopen(BT_REGEXP_DEBUG_LOG_NAME, "a"); |
| 4444 | 8008 if (f) |
| 8009 { | |
|
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8010 fprintf(f, "Syntax error in \"%s\"\n", expr); |
| 4444 | 8011 fclose(f); |
| 8012 } | |
| 8013 else | |
| 4460 | 8014 EMSG2("(NFA) Could not open \"%s\" to write !!!", |
| 8015 BT_REGEXP_DEBUG_LOG_NAME); | |
| 4444 | 8016 } |
| 8017 #endif | |
| 8018 /* | |
| 6328 | 8019 * If the NFA engine failed, try the backtracking engine. |
| 6533 | 8020 * The NFA engine also fails for patterns that it can't handle well |
| 8021 * but are still valid patterns, thus a retry should work. | |
| 8022 */ | |
| 4444 | 8023 if (regexp_engine == AUTOMATIC_ENGINE) |
| 6328 | 8024 { |
| 6533 | 8025 regexp_engine = BACKTRACKING_ENGINE; |
|
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8026 prog = bt_regengine.regcomp(expr, re_flags); |
| 6328 | 8027 } |
|
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8028 } |
| 4444 | 8029 |
| 6328 | 8030 if (prog != NULL) |
| 8031 { | |
| 8032 /* Store the info needed to call regcomp() again when the engine turns | |
| 8033 * out to be very slow when executing it. */ | |
| 8034 prog->re_engine = regexp_engine; | |
| 8035 prog->re_flags = re_flags; | |
| 8036 } | |
| 8037 | |
| 4444 | 8038 return prog; |
| 8039 } | |
| 8040 | |
| 8041 /* | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8042 * 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
|
8043 */ |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8044 void |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8045 vim_regfree(regprog_T *prog) |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8046 { |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8047 if (prog != NULL) |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8048 prog->engine->regfree(prog); |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8049 } |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8050 |
| 6328 | 8051 #ifdef FEAT_EVAL |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
8052 static void report_re_switch(char_u *pat); |
| 6328 | 8053 |
| 8054 static void | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8055 report_re_switch(char_u *pat) |
| 6328 | 8056 { |
| 8057 if (p_verbose > 0) | |
| 8058 { | |
| 8059 verbose_enter(); | |
| 8060 MSG_PUTS(_("Switching to backtracking RE engine for pattern: ")); | |
| 8061 MSG_PUTS(pat); | |
| 8062 verbose_leave(); | |
| 8063 } | |
| 8064 } | |
| 8065 #endif | |
| 8066 | |
|
7805
0b6c37dd858d
commit https://github.com/vim/vim/commit/baaa7e9ec7398a813e21285c272fa99792642077
Christian Brabandt <cb@256bit.org>
parents:
7009
diff
changeset
|
8067 static int vim_regexec_both(regmatch_T *rmp, char_u *line, colnr_T col, int nl); |
| 6328 | 8068 |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8069 /* |
| 4444 | 8070 * Match a regexp against a string. |
| 8071 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 6375 | 8072 * Note: "rmp->regprog" may be freed and changed. |
| 4444 | 8073 * Uses curbuf for line count and 'iskeyword'. |
| 6328 | 8074 * When "nl" is TRUE consider a "\n" in "line" to be a line break. |
| 4444 | 8075 * |
| 8076 * Return TRUE if there is a match, FALSE if not. | |
| 8077 */ | |
| 6328 | 8078 static int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8079 vim_regexec_both( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8080 regmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8081 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
|
8082 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
|
8083 int nl) |
| 6328 | 8084 { |
| 8085 int result = rmp->regprog->engine->regexec_nl(rmp, line, col, nl); | |
| 8086 | |
| 8087 /* NFA engine aborted because it's very slow. */ | |
| 8088 if (rmp->regprog->re_engine == AUTOMATIC_ENGINE | |
| 8089 && result == NFA_TOO_EXPENSIVE) | |
| 8090 { | |
| 8091 int save_p_re = p_re; | |
| 8092 int re_flags = rmp->regprog->re_flags; | |
| 8093 char_u *pat = vim_strsave(((nfa_regprog_T *)rmp->regprog)->pattern); | |
| 8094 | |
| 8095 p_re = BACKTRACKING_ENGINE; | |
| 8096 vim_regfree(rmp->regprog); | |
| 8097 if (pat != NULL) | |
| 8098 { | |
| 8099 #ifdef FEAT_EVAL | |
| 8100 report_re_switch(pat); | |
| 8101 #endif | |
| 8102 rmp->regprog = vim_regcomp(pat, re_flags); | |
| 8103 if (rmp->regprog != NULL) | |
| 8104 result = rmp->regprog->engine->regexec_nl(rmp, line, col, nl); | |
| 8105 vim_free(pat); | |
| 8106 } | |
| 8107 | |
| 8108 p_re = save_p_re; | |
| 8109 } | |
| 6390 | 8110 return result > 0; |
| 6328 | 8111 } |
| 8112 | |
| 6375 | 8113 /* |
| 8114 * Note: "*prog" may be freed and changed. | |
| 6390 | 8115 * Return TRUE if there is a match, FALSE if not. |
| 6375 | 8116 */ |
| 8117 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8118 vim_regexec_prog( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8119 regprog_T **prog, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8120 int ignore_case, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8121 char_u *line, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8122 colnr_T col) |
| 6375 | 8123 { |
| 8124 int r; | |
| 8125 regmatch_T regmatch; | |
| 8126 | |
| 8127 regmatch.regprog = *prog; | |
| 8128 regmatch.rm_ic = ignore_case; | |
| 8129 r = vim_regexec_both(®match, line, col, FALSE); | |
| 8130 *prog = regmatch.regprog; | |
| 8131 return r; | |
| 8132 } | |
| 8133 | |
| 8134 /* | |
| 8135 * Note: "rmp->regprog" may be freed and changed. | |
| 6390 | 8136 * Return TRUE if there is a match, FALSE if not. |
| 6375 | 8137 */ |
| 4444 | 8138 int |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8139 vim_regexec(regmatch_T *rmp, char_u *line, colnr_T col) |
| 4444 | 8140 { |
| 6328 | 8141 return vim_regexec_both(rmp, line, col, FALSE); |
| 4444 | 8142 } |
| 8143 | |
| 8144 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
| 8145 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 8146 /* | |
| 8147 * Like vim_regexec(), but consider a "\n" in "line" to be a line break. | |
| 6375 | 8148 * Note: "rmp->regprog" may be freed and changed. |
| 6390 | 8149 * Return TRUE if there is a match, FALSE if not. |
| 4444 | 8150 */ |
| 8151 int | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8152 vim_regexec_nl(regmatch_T *rmp, char_u *line, colnr_T col) |
| 4444 | 8153 { |
| 6328 | 8154 return vim_regexec_both(rmp, line, col, TRUE); |
| 4444 | 8155 } |
| 8156 #endif | |
| 8157 | |
| 8158 /* | |
| 8159 * Match a regexp against multiple lines. | |
| 8160 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 6375 | 8161 * Note: "rmp->regprog" may be freed and changed. |
| 4444 | 8162 * Uses curbuf for line count and 'iskeyword'. |
| 8163 * | |
| 8164 * Return zero if there is no match. Return number of lines contained in the | |
| 8165 * match otherwise. | |
| 8166 */ | |
| 8167 long | |
|
7833
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8168 vim_regexec_multi( |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8169 regmmatch_T *rmp, |
|
c079097365f3
commit https://github.com/vim/vim/commit/055409764ca5f7978d4c399d2c440af0ce971c4f
Christian Brabandt <cb@256bit.org>
parents:
7807
diff
changeset
|
8170 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
|
8171 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
|
8172 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
|
8173 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
|
8174 proftime_T *tm) /* timeout limit or NULL */ |
| 4444 | 8175 { |
| 6328 | 8176 int result = rmp->regprog->engine->regexec_multi( |
| 8177 rmp, win, buf, lnum, col, tm); | |
| 8178 | |
| 8179 /* NFA engine aborted because it's very slow. */ | |
| 8180 if (rmp->regprog->re_engine == AUTOMATIC_ENGINE | |
| 8181 && result == NFA_TOO_EXPENSIVE) | |
| 8182 { | |
| 8183 int save_p_re = p_re; | |
| 8184 int re_flags = rmp->regprog->re_flags; | |
| 8185 char_u *pat = vim_strsave(((nfa_regprog_T *)rmp->regprog)->pattern); | |
| 8186 | |
| 8187 p_re = BACKTRACKING_ENGINE; | |
| 8188 vim_regfree(rmp->regprog); | |
| 8189 if (pat != NULL) | |
| 8190 { | |
| 8191 #ifdef FEAT_EVAL | |
| 8192 report_re_switch(pat); | |
| 8193 #endif | |
| 8194 rmp->regprog = vim_regcomp(pat, re_flags); | |
| 8195 if (rmp->regprog != NULL) | |
| 8196 result = rmp->regprog->engine->regexec_multi( | |
| 8197 rmp, win, buf, lnum, col, tm); | |
| 8198 vim_free(pat); | |
| 8199 } | |
| 8200 p_re = save_p_re; | |
| 8201 } | |
| 8202 | |
| 6390 | 8203 return result <= 0 ? 0 : result; |
| 4444 | 8204 } |