Mercurial > vim
annotate src/regexp.c @ 5153:b5abda4e7c4b v7.4a.003
updated for version 7.4a.003
Problem: Copyright year is oudated.
Solution: Only use the first year.
| author | Bram Moolenaar <bram@vim.org> |
|---|---|
| date | Tue, 09 Jul 2013 12:54:21 +0200 |
| parents | 56fbf60e9476 |
| children | 9982ec574beb |
| 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 */ | |
| 247 | |
| 7 | 248 /* |
| 249 * Magic characters have a special meaning, they don't match literally. | |
| 250 * Magic characters are negative. This separates them from literal characters | |
| 251 * (possibly multi-byte). Only ASCII characters can be Magic. | |
| 252 */ | |
| 253 #define Magic(x) ((int)(x) - 256) | |
| 254 #define un_Magic(x) ((x) + 256) | |
| 255 #define is_Magic(x) ((x) < 0) | |
| 256 | |
| 257 static int no_Magic __ARGS((int x)); | |
| 258 static int toggle_Magic __ARGS((int x)); | |
| 259 | |
| 260 static int | |
| 261 no_Magic(x) | |
| 262 int x; | |
| 263 { | |
| 264 if (is_Magic(x)) | |
| 265 return un_Magic(x); | |
| 266 return x; | |
| 267 } | |
| 268 | |
| 269 static int | |
| 270 toggle_Magic(x) | |
| 271 int x; | |
| 272 { | |
| 273 if (is_Magic(x)) | |
| 274 return un_Magic(x); | |
| 275 return Magic(x); | |
| 276 } | |
| 277 | |
| 278 /* | |
| 279 * The first byte of the regexp internal "program" is actually this magic | |
| 280 * number; the start node begins in the second byte. It's used to catch the | |
| 281 * most severe mutilation of the program by the caller. | |
| 282 */ | |
| 283 | |
| 284 #define REGMAGIC 0234 | |
| 285 | |
| 286 /* | |
| 287 * Opcode notes: | |
| 288 * | |
| 289 * BRANCH The set of branches constituting a single choice are hooked | |
| 290 * together with their "next" pointers, since precedence prevents | |
| 291 * anything being concatenated to any individual branch. The | |
| 292 * "next" pointer of the last BRANCH in a choice points to the | |
| 293 * thing following the whole choice. This is also where the | |
| 294 * final "next" pointer of each individual branch points; each | |
| 295 * branch starts with the operand node of a BRANCH node. | |
| 296 * | |
| 297 * BACK Normal "next" pointers all implicitly point forward; BACK | |
| 298 * exists to make loop structures possible. | |
| 299 * | |
| 300 * STAR,PLUS '=', and complex '*' and '+', are implemented as circular | |
| 301 * BRANCH structures using BACK. Simple cases (one character | |
| 302 * per match) are implemented with STAR and PLUS for speed | |
| 303 * and to minimize recursive plunges. | |
| 304 * | |
| 305 * BRACE_LIMITS This is always followed by a BRACE_SIMPLE or BRACE_COMPLEX | |
| 306 * node, and defines the min and max limits to be used for that | |
| 307 * node. | |
| 308 * | |
| 309 * MOPEN,MCLOSE ...are numbered at compile time. | |
| 310 * ZOPEN,ZCLOSE ...ditto | |
| 311 */ | |
| 312 | |
| 313 /* | |
| 314 * A node is one char of opcode followed by two chars of "next" pointer. | |
| 315 * "Next" pointers are stored as two 8-bit bytes, high order first. The | |
| 316 * value is a positive offset from the opcode of the node containing it. | |
| 317 * An operand, if any, simply follows the node. (Note that much of the | |
| 318 * code generation knows about this implicit relationship.) | |
| 319 * | |
| 320 * Using two bytes for the "next" pointer is vast overkill for most things, | |
| 321 * but allows patterns to get big without disasters. | |
| 322 */ | |
| 323 #define OP(p) ((int)*(p)) | |
| 324 #define NEXT(p) (((*((p) + 1) & 0377) << 8) + (*((p) + 2) & 0377)) | |
| 325 #define OPERAND(p) ((p) + 3) | |
| 326 /* Obtain an operand that was stored as four bytes, MSB first. */ | |
| 327 #define OPERAND_MIN(p) (((long)(p)[3] << 24) + ((long)(p)[4] << 16) \ | |
| 328 + ((long)(p)[5] << 8) + (long)(p)[6]) | |
| 329 /* Obtain a second operand stored as four bytes. */ | |
| 330 #define OPERAND_MAX(p) OPERAND_MIN((p) + 4) | |
| 331 /* Obtain a second single-byte operand stored after a four bytes operand. */ | |
| 332 #define OPERAND_CMP(p) (p)[7] | |
| 333 | |
| 334 /* | |
| 335 * Utility definitions. | |
| 336 */ | |
| 337 #define UCHARAT(p) ((int)*(char_u *)(p)) | |
| 338 | |
| 339 /* Used for an error (down from) vim_regcomp(): give the error message, set | |
| 340 * rc_did_emsg and return NULL */ | |
| 653 | 341 #define EMSG_RET_NULL(m) return (EMSG(m), rc_did_emsg = TRUE, (void *)NULL) |
| 308 | 342 #define EMSG_RET_FAIL(m) return (EMSG(m), rc_did_emsg = TRUE, FAIL) |
| 4444 | 343 #define EMSG2_RET_NULL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, (void *)NULL) |
| 344 #define EMSG2_RET_FAIL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, FAIL) | |
| 345 #define EMSG_ONE_RET_NULL EMSG2_RET_NULL(_("E369: invalid item in %s%%[]"), reg_magic == MAGIC_ALL) | |
| 7 | 346 |
| 347 #define MAX_LIMIT (32767L << 16L) | |
| 348 | |
| 349 static int re_multi_type __ARGS((int)); | |
| 350 static int cstrncmp __ARGS((char_u *s1, char_u *s2, int *n)); | |
| 351 static char_u *cstrchr __ARGS((char_u *, int)); | |
| 352 | |
| 4444 | 353 #ifdef BT_REGEXP_DUMP |
| 354 static void regdump __ARGS((char_u *, bt_regprog_T *)); | |
| 355 #endif | |
| 7 | 356 #ifdef DEBUG |
| 357 static char_u *regprop __ARGS((char_u *)); | |
| 358 #endif | |
| 359 | |
| 4444 | 360 static char_u e_missingbracket[] = N_("E769: Missing ] after %s["); |
| 361 static char_u e_unmatchedpp[] = N_("E53: Unmatched %s%%("); | |
| 362 static char_u e_unmatchedp[] = N_("E54: Unmatched %s("); | |
| 363 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
|
364 #ifdef FEAT_SYN_HL |
|
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
365 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
|
366 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
|
367 #endif |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
368 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
|
369 static char_u e_empty_sb[] = N_("E70: Empty %s%%[]"); |
| 7 | 370 #define NOT_MULTI 0 |
| 371 #define MULTI_ONE 1 | |
| 372 #define MULTI_MULT 2 | |
| 373 /* | |
| 374 * Return NOT_MULTI if c is not a "multi" operator. | |
| 375 * Return MULTI_ONE if c is a single "multi" operator. | |
| 376 * Return MULTI_MULT if c is a multi "multi" operator. | |
| 377 */ | |
| 378 static int | |
| 379 re_multi_type(c) | |
| 380 int c; | |
| 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 |
| 426 static int backslash_trans __ARGS((int c)); | |
| 167 | 427 static int get_char_class __ARGS((char_u **pp)); |
| 428 static int get_equi_class __ARGS((char_u **pp)); | |
| 429 static void reg_equi_class __ARGS((int c)); | |
| 430 static int get_coll_element __ARGS((char_u **pp)); | |
| 7 | 431 static char_u *skip_anyof __ARGS((char_u *p)); |
| 432 static void init_class_tab __ARGS((void)); | |
| 433 | |
| 434 /* | |
| 435 * Translate '\x' to its control character, except "\n", which is Magic. | |
| 436 */ | |
| 437 static int | |
| 438 backslash_trans(c) | |
| 439 int c; | |
| 440 { | |
| 441 switch (c) | |
| 442 { | |
| 443 case 'r': return CAR; | |
| 444 case 't': return TAB; | |
| 445 case 'e': return ESC; | |
| 446 case 'b': return BS; | |
| 447 } | |
| 448 return c; | |
| 449 } | |
| 450 | |
| 451 /* | |
| 167 | 452 * Check for a character class name "[:name:]". "pp" points to the '['. |
| 7 | 453 * Returns one of the CLASS_ items. CLASS_NONE means that no item was |
| 454 * recognized. Otherwise "pp" is advanced to after the item. | |
| 455 */ | |
| 456 static int | |
| 167 | 457 get_char_class(pp) |
| 7 | 458 char_u **pp; |
| 459 { | |
| 460 static const char *(class_names[]) = | |
| 461 { | |
| 462 "alnum:]", | |
| 463 #define CLASS_ALNUM 0 | |
| 464 "alpha:]", | |
| 465 #define CLASS_ALPHA 1 | |
| 466 "blank:]", | |
| 467 #define CLASS_BLANK 2 | |
| 468 "cntrl:]", | |
| 469 #define CLASS_CNTRL 3 | |
| 470 "digit:]", | |
| 471 #define CLASS_DIGIT 4 | |
| 472 "graph:]", | |
| 473 #define CLASS_GRAPH 5 | |
| 474 "lower:]", | |
| 475 #define CLASS_LOWER 6 | |
| 476 "print:]", | |
| 477 #define CLASS_PRINT 7 | |
| 478 "punct:]", | |
| 479 #define CLASS_PUNCT 8 | |
| 480 "space:]", | |
| 481 #define CLASS_SPACE 9 | |
| 482 "upper:]", | |
| 483 #define CLASS_UPPER 10 | |
| 484 "xdigit:]", | |
| 485 #define CLASS_XDIGIT 11 | |
| 486 "tab:]", | |
| 487 #define CLASS_TAB 12 | |
| 488 "return:]", | |
| 489 #define CLASS_RETURN 13 | |
| 490 "backspace:]", | |
| 491 #define CLASS_BACKSPACE 14 | |
| 492 "escape:]", | |
| 493 #define CLASS_ESCAPE 15 | |
| 494 }; | |
| 495 #define CLASS_NONE 99 | |
| 496 int i; | |
| 497 | |
| 498 if ((*pp)[1] == ':') | |
| 499 { | |
| 1877 | 500 for (i = 0; i < (int)(sizeof(class_names) / sizeof(*class_names)); ++i) |
| 7 | 501 if (STRNCMP(*pp + 2, class_names[i], STRLEN(class_names[i])) == 0) |
| 502 { | |
| 503 *pp += STRLEN(class_names[i]) + 2; | |
| 504 return i; | |
| 505 } | |
| 506 } | |
| 507 return CLASS_NONE; | |
| 508 } | |
| 509 | |
| 510 /* | |
| 511 * Specific version of character class functions. | |
| 512 * Using a table to keep this fast. | |
| 513 */ | |
| 514 static short class_tab[256]; | |
| 515 | |
| 516 #define RI_DIGIT 0x01 | |
| 517 #define RI_HEX 0x02 | |
| 518 #define RI_OCTAL 0x04 | |
| 519 #define RI_WORD 0x08 | |
| 520 #define RI_HEAD 0x10 | |
| 521 #define RI_ALPHA 0x20 | |
| 522 #define RI_LOWER 0x40 | |
| 523 #define RI_UPPER 0x80 | |
| 524 #define RI_WHITE 0x100 | |
| 525 | |
| 526 static void | |
| 527 init_class_tab() | |
| 528 { | |
| 529 int i; | |
| 530 static int done = FALSE; | |
| 531 | |
| 532 if (done) | |
| 533 return; | |
| 534 | |
| 535 for (i = 0; i < 256; ++i) | |
| 536 { | |
| 537 if (i >= '0' && i <= '7') | |
| 538 class_tab[i] = RI_DIGIT + RI_HEX + RI_OCTAL + RI_WORD; | |
| 539 else if (i >= '8' && i <= '9') | |
| 540 class_tab[i] = RI_DIGIT + RI_HEX + RI_WORD; | |
| 541 else if (i >= 'a' && i <= 'f') | |
| 542 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
| 543 #ifdef EBCDIC | |
| 544 else if ((i >= 'g' && i <= 'i') || (i >= 'j' && i <= 'r') | |
| 545 || (i >= 's' && i <= 'z')) | |
| 546 #else | |
| 547 else if (i >= 'g' && i <= 'z') | |
| 548 #endif | |
| 549 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
| 550 else if (i >= 'A' && i <= 'F') | |
| 551 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
| 552 #ifdef EBCDIC | |
| 553 else if ((i >= 'G' && i <= 'I') || ( i >= 'J' && i <= 'R') | |
| 554 || (i >= 'S' && i <= 'Z')) | |
| 555 #else | |
| 556 else if (i >= 'G' && i <= 'Z') | |
| 557 #endif | |
| 558 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
| 559 else if (i == '_') | |
| 560 class_tab[i] = RI_WORD + RI_HEAD; | |
| 561 else | |
| 562 class_tab[i] = 0; | |
| 563 } | |
| 564 class_tab[' '] |= RI_WHITE; | |
| 565 class_tab['\t'] |= RI_WHITE; | |
| 566 done = TRUE; | |
| 567 } | |
| 568 | |
| 569 #ifdef FEAT_MBYTE | |
| 570 # define ri_digit(c) (c < 0x100 && (class_tab[c] & RI_DIGIT)) | |
| 571 # define ri_hex(c) (c < 0x100 && (class_tab[c] & RI_HEX)) | |
| 572 # define ri_octal(c) (c < 0x100 && (class_tab[c] & RI_OCTAL)) | |
| 573 # define ri_word(c) (c < 0x100 && (class_tab[c] & RI_WORD)) | |
| 574 # define ri_head(c) (c < 0x100 && (class_tab[c] & RI_HEAD)) | |
| 575 # define ri_alpha(c) (c < 0x100 && (class_tab[c] & RI_ALPHA)) | |
| 576 # define ri_lower(c) (c < 0x100 && (class_tab[c] & RI_LOWER)) | |
| 577 # define ri_upper(c) (c < 0x100 && (class_tab[c] & RI_UPPER)) | |
| 578 # define ri_white(c) (c < 0x100 && (class_tab[c] & RI_WHITE)) | |
| 579 #else | |
| 580 # define ri_digit(c) (class_tab[c] & RI_DIGIT) | |
| 581 # define ri_hex(c) (class_tab[c] & RI_HEX) | |
| 582 # define ri_octal(c) (class_tab[c] & RI_OCTAL) | |
| 583 # define ri_word(c) (class_tab[c] & RI_WORD) | |
| 584 # define ri_head(c) (class_tab[c] & RI_HEAD) | |
| 585 # define ri_alpha(c) (class_tab[c] & RI_ALPHA) | |
| 586 # define ri_lower(c) (class_tab[c] & RI_LOWER) | |
| 587 # define ri_upper(c) (class_tab[c] & RI_UPPER) | |
| 588 # define ri_white(c) (class_tab[c] & RI_WHITE) | |
| 589 #endif | |
| 590 | |
| 591 /* flags for regflags */ | |
| 592 #define RF_ICASE 1 /* ignore case */ | |
| 593 #define RF_NOICASE 2 /* don't ignore case */ | |
| 594 #define RF_HASNL 4 /* can match a NL */ | |
| 595 #define RF_ICOMBINE 8 /* ignore combining characters */ | |
| 596 #define RF_LOOKBH 16 /* uses "\@<=" or "\@<!" */ | |
| 597 | |
| 598 /* | |
| 599 * Global work variables for vim_regcomp(). | |
| 600 */ | |
| 601 | |
| 602 static char_u *regparse; /* Input-scan pointer. */ | |
| 603 static int prevchr_len; /* byte length of previous char */ | |
| 604 static int num_complex_braces; /* Complex \{...} count */ | |
| 605 static int regnpar; /* () count. */ | |
| 606 #ifdef FEAT_SYN_HL | |
| 607 static int regnzpar; /* \z() count. */ | |
| 608 static int re_has_z; /* \z item detected */ | |
| 609 #endif | |
| 610 static char_u *regcode; /* Code-emit pointer, or JUST_CALC_SIZE */ | |
| 611 static long regsize; /* Code size. */ | |
| 2010 | 612 static int reg_toolong; /* TRUE when offset out of range */ |
| 7 | 613 static char_u had_endbrace[NSUBEXP]; /* flags, TRUE if end of () found */ |
| 614 static unsigned regflags; /* RF_ flags for prog */ | |
| 615 static long brace_min[10]; /* Minimums for complex brace repeats */ | |
| 616 static long brace_max[10]; /* Maximums for complex brace repeats */ | |
| 617 static int brace_count[10]; /* Current counts for complex brace repeats */ | |
| 618 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 619 static int had_eol; /* TRUE when EOL found by vim_regcomp() */ | |
| 620 #endif | |
| 621 static int one_exactly = FALSE; /* only do one char for EXACTLY */ | |
| 622 | |
| 623 static int reg_magic; /* magicness of the pattern: */ | |
| 624 #define MAGIC_NONE 1 /* "\V" very unmagic */ | |
| 625 #define MAGIC_OFF 2 /* "\M" or 'magic' off */ | |
| 626 #define MAGIC_ON 3 /* "\m" or 'magic' */ | |
| 627 #define MAGIC_ALL 4 /* "\v" very magic */ | |
| 628 | |
| 629 static int reg_string; /* matching with a string instead of a buffer | |
| 630 line */ | |
| 481 | 631 static int reg_strict; /* "[abc" is illegal */ |
| 7 | 632 |
| 633 /* | |
| 634 * META contains all characters that may be magic, except '^' and '$'. | |
| 635 */ | |
| 636 | |
| 637 #ifdef EBCDIC | |
| 638 static char_u META[] = "%&()*+.123456789<=>?@ACDFHIKLMOPSUVWX[_acdfhiklmnopsuvwxz{|~"; | |
| 639 #else | |
| 640 /* META[] is used often enough to justify turning it into a table. */ | |
| 641 static char_u META_flags[] = { | |
| 642 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| 643 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| 644 /* % & ( ) * + . */ | |
| 645 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0, | |
| 646 /* 1 2 3 4 5 6 7 8 9 < = > ? */ | |
| 647 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, | |
| 648 /* @ A C D F H I K L M O */ | |
| 649 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1, | |
| 650 /* P S U V W X Z [ _ */ | |
| 651 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, | |
| 652 /* a c d f h i k l m n o */ | |
| 653 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, | |
| 654 /* p s u v w x z { | ~ */ | |
| 655 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1 | |
| 656 }; | |
| 657 #endif | |
| 658 | |
| 4444 | 659 static int curchr; /* currently parsed character */ |
| 660 /* Previous character. Note: prevchr is sometimes -1 when we are not at the | |
| 661 * start, eg in /[ ^I]^ the pattern was never found even if it existed, | |
| 662 * because ^ was taken to be magic -- webb */ | |
| 663 static int prevchr; | |
| 664 static int prevprevchr; /* previous-previous character */ | |
| 665 static int nextchr; /* used for ungetchr() */ | |
| 7 | 666 |
| 667 /* arguments for reg() */ | |
| 668 #define REG_NOPAREN 0 /* toplevel reg() */ | |
| 669 #define REG_PAREN 1 /* \(\) */ | |
| 670 #define REG_ZPAREN 2 /* \z(\) */ | |
| 671 #define REG_NPAREN 3 /* \%(\) */ | |
| 672 | |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
673 typedef struct |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
674 { |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
675 char_u *regparse; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
676 int prevchr_len; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
677 int curchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
678 int prevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
679 int prevprevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
680 int nextchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
681 int at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
682 int prev_at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
683 int regnpar; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
684 } parse_state_T; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
685 |
| 7 | 686 /* |
| 687 * Forward declarations for vim_regcomp()'s friends. | |
| 688 */ | |
| 689 static void initchr __ARGS((char_u *)); | |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
690 static void save_parse_state __ARGS((parse_state_T *ps)); |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
691 static void restore_parse_state __ARGS((parse_state_T *ps)); |
| 7 | 692 static int getchr __ARGS((void)); |
| 693 static void skipchr_keepstart __ARGS((void)); | |
| 694 static int peekchr __ARGS((void)); | |
| 695 static void skipchr __ARGS((void)); | |
| 696 static void ungetchr __ARGS((void)); | |
| 24 | 697 static int gethexchrs __ARGS((int maxinputlen)); |
| 698 static int getoctchrs __ARGS((void)); | |
| 699 static int getdecchrs __ARGS((void)); | |
| 700 static int coll_get_char __ARGS((void)); | |
| 7 | 701 static void regcomp_start __ARGS((char_u *expr, int flags)); |
| 702 static char_u *reg __ARGS((int, int *)); | |
| 703 static char_u *regbranch __ARGS((int *flagp)); | |
| 704 static char_u *regconcat __ARGS((int *flagp)); | |
| 705 static char_u *regpiece __ARGS((int *)); | |
| 706 static char_u *regatom __ARGS((int *)); | |
| 707 static char_u *regnode __ARGS((int)); | |
| 714 | 708 #ifdef FEAT_MBYTE |
| 709 static int use_multibytecode __ARGS((int c)); | |
| 710 #endif | |
| 7 | 711 static int prog_magic_wrong __ARGS((void)); |
| 712 static char_u *regnext __ARGS((char_u *)); | |
| 713 static void regc __ARGS((int b)); | |
| 714 #ifdef FEAT_MBYTE | |
| 715 static void regmbc __ARGS((int c)); | |
| 2974 | 716 # define REGMBC(x) regmbc(x); |
| 717 # define CASEMBC(x) case x: | |
| 167 | 718 #else |
| 719 # define regmbc(c) regc(c) | |
| 2974 | 720 # define REGMBC(x) |
| 721 # define CASEMBC(x) | |
| 7 | 722 #endif |
| 723 static void reginsert __ARGS((int, char_u *)); | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
724 static void reginsert_nr __ARGS((int op, long val, char_u *opnd)); |
| 7 | 725 static void reginsert_limits __ARGS((int, long, long, char_u *)); |
| 726 static char_u *re_put_long __ARGS((char_u *pr, long_u val)); | |
| 727 static int read_limits __ARGS((long *, long *)); | |
| 728 static void regtail __ARGS((char_u *, char_u *)); | |
| 729 static void regoptail __ARGS((char_u *, char_u *)); | |
| 730 | |
| 4444 | 731 static regengine_T bt_regengine; |
| 732 static regengine_T nfa_regengine; | |
| 733 | |
| 7 | 734 /* |
| 735 * Return TRUE if compiled regular expression "prog" can match a line break. | |
| 736 */ | |
| 737 int | |
| 738 re_multiline(prog) | |
| 739 regprog_T *prog; | |
| 740 { | |
| 741 return (prog->regflags & RF_HASNL); | |
| 742 } | |
| 743 | |
| 744 /* | |
| 745 * Return TRUE if compiled regular expression "prog" looks before the start | |
| 746 * position (pattern contains "\@<=" or "\@<!"). | |
| 747 */ | |
| 748 int | |
| 749 re_lookbehind(prog) | |
| 750 regprog_T *prog; | |
| 751 { | |
| 752 return (prog->regflags & RF_LOOKBH); | |
| 753 } | |
| 754 | |
| 755 /* | |
| 167 | 756 * Check for an equivalence class name "[=a=]". "pp" points to the '['. |
| 757 * Returns a character representing the class. Zero means that no item was | |
| 758 * recognized. Otherwise "pp" is advanced to after the item. | |
| 759 */ | |
| 760 static int | |
| 761 get_equi_class(pp) | |
| 762 char_u **pp; | |
| 763 { | |
| 764 int c; | |
| 765 int l = 1; | |
| 766 char_u *p = *pp; | |
| 767 | |
| 768 if (p[1] == '=') | |
| 769 { | |
| 770 #ifdef FEAT_MBYTE | |
| 771 if (has_mbyte) | |
| 474 | 772 l = (*mb_ptr2len)(p + 2); |
| 167 | 773 #endif |
| 774 if (p[l + 2] == '=' && p[l + 3] == ']') | |
| 775 { | |
| 776 #ifdef FEAT_MBYTE | |
| 777 if (has_mbyte) | |
| 778 c = mb_ptr2char(p + 2); | |
| 779 else | |
| 780 #endif | |
| 781 c = p[2]; | |
| 782 *pp += l + 4; | |
| 783 return c; | |
| 784 } | |
| 785 } | |
| 786 return 0; | |
| 787 } | |
| 788 | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
789 #ifdef EBCDIC |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
790 /* |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
791 * 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
|
792 */ |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
793 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
|
794 "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
|
795 "C\x68", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
796 "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
|
797 "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
|
798 "N\x69", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
799 "O\xEB\xEC\xED\xEE\xEF", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
800 "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
|
801 "Y\xBA", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
802 "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
|
803 "c\x48", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
804 "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
|
805 "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
|
806 "n\x49", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
807 "o\xCB\xCC\xCD\xCE\xCF", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
808 "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
|
809 "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
|
810 }; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
811 #endif |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
812 |
| 167 | 813 /* |
| 814 * Produce the bytes for equivalence class "c". | |
| 815 * Currently only handles latin1, latin9 and utf-8. | |
| 4444 | 816 * NOTE: When changing this function, also change nfa_emit_equi_class() |
| 167 | 817 */ |
| 818 static void | |
| 819 reg_equi_class(c) | |
| 820 int c; | |
| 821 { | |
| 822 #ifdef FEAT_MBYTE | |
| 823 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
| 492 | 824 || STRCMP(p_enc, "iso-8859-15") == 0) |
| 167 | 825 #endif |
| 826 { | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
827 #ifdef EBCDIC |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
828 int i; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
829 |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
830 /* 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
|
831 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
|
832 { |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
833 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
|
834 { |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
835 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
|
836 |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
837 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
|
838 regmbc(*p++); |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
839 return; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
840 } |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
841 } |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
842 #else |
| 167 | 843 switch (c) |
| 844 { | |
| 236 | 845 case 'A': case '\300': case '\301': case '\302': |
| 2974 | 846 CASEMBC(0x100) CASEMBC(0x102) CASEMBC(0x104) CASEMBC(0x1cd) |
| 847 CASEMBC(0x1de) CASEMBC(0x1e0) CASEMBC(0x1ea2) | |
| 236 | 848 case '\303': case '\304': case '\305': |
| 849 regmbc('A'); regmbc('\300'); regmbc('\301'); | |
| 850 regmbc('\302'); regmbc('\303'); regmbc('\304'); | |
| 851 regmbc('\305'); | |
| 2974 | 852 REGMBC(0x100) REGMBC(0x102) REGMBC(0x104) |
| 853 REGMBC(0x1cd) REGMBC(0x1de) REGMBC(0x1e0) | |
| 854 REGMBC(0x1ea2) | |
| 855 return; | |
| 856 case 'B': CASEMBC(0x1e02) CASEMBC(0x1e06) | |
| 857 regmbc('B'); REGMBC(0x1e02) REGMBC(0x1e06) | |
| 167 | 858 return; |
| 236 | 859 case 'C': case '\307': |
| 2974 | 860 CASEMBC(0x106) CASEMBC(0x108) CASEMBC(0x10a) CASEMBC(0x10c) |
| 236 | 861 regmbc('C'); regmbc('\307'); |
| 2974 | 862 REGMBC(0x106) REGMBC(0x108) REGMBC(0x10a) |
| 863 REGMBC(0x10c) | |
| 864 return; | |
| 865 case 'D': CASEMBC(0x10e) CASEMBC(0x110) CASEMBC(0x1e0a) | |
| 866 CASEMBC(0x1e0e) CASEMBC(0x1e10) | |
| 867 regmbc('D'); REGMBC(0x10e) REGMBC(0x110) | |
| 868 REGMBC(0x1e0a) REGMBC(0x1e0e) REGMBC(0x1e10) | |
| 167 | 869 return; |
| 236 | 870 case 'E': case '\310': case '\311': case '\312': case '\313': |
| 2974 | 871 CASEMBC(0x112) CASEMBC(0x114) CASEMBC(0x116) CASEMBC(0x118) |
| 872 CASEMBC(0x11a) CASEMBC(0x1eba) CASEMBC(0x1ebc) | |
| 236 | 873 regmbc('E'); regmbc('\310'); regmbc('\311'); |
| 874 regmbc('\312'); regmbc('\313'); | |
| 2974 | 875 REGMBC(0x112) REGMBC(0x114) REGMBC(0x116) |
| 876 REGMBC(0x118) REGMBC(0x11a) REGMBC(0x1eba) | |
| 877 REGMBC(0x1ebc) | |
| 878 return; | |
| 879 case 'F': CASEMBC(0x1e1e) | |
| 880 regmbc('F'); REGMBC(0x1e1e) | |
| 881 return; | |
| 882 case 'G': CASEMBC(0x11c) CASEMBC(0x11e) CASEMBC(0x120) | |
| 883 CASEMBC(0x122) CASEMBC(0x1e4) CASEMBC(0x1e6) CASEMBC(0x1f4) | |
| 884 CASEMBC(0x1e20) | |
| 885 regmbc('G'); REGMBC(0x11c) REGMBC(0x11e) | |
| 886 REGMBC(0x120) REGMBC(0x122) REGMBC(0x1e4) | |
| 887 REGMBC(0x1e6) REGMBC(0x1f4) REGMBC(0x1e20) | |
| 888 return; | |
| 889 case 'H': CASEMBC(0x124) CASEMBC(0x126) CASEMBC(0x1e22) | |
| 890 CASEMBC(0x1e26) CASEMBC(0x1e28) | |
| 891 regmbc('H'); REGMBC(0x124) REGMBC(0x126) | |
| 892 REGMBC(0x1e22) REGMBC(0x1e26) REGMBC(0x1e28) | |
| 167 | 893 return; |
| 236 | 894 case 'I': case '\314': case '\315': case '\316': case '\317': |
| 2974 | 895 CASEMBC(0x128) CASEMBC(0x12a) CASEMBC(0x12c) CASEMBC(0x12e) |
| 896 CASEMBC(0x130) CASEMBC(0x1cf) CASEMBC(0x1ec8) | |
| 236 | 897 regmbc('I'); regmbc('\314'); regmbc('\315'); |
| 898 regmbc('\316'); regmbc('\317'); | |
| 2974 | 899 REGMBC(0x128) REGMBC(0x12a) REGMBC(0x12c) |
| 900 REGMBC(0x12e) REGMBC(0x130) REGMBC(0x1cf) | |
| 901 REGMBC(0x1ec8) | |
| 902 return; | |
| 903 case 'J': CASEMBC(0x134) | |
| 904 regmbc('J'); REGMBC(0x134) | |
| 905 return; | |
| 906 case 'K': CASEMBC(0x136) CASEMBC(0x1e8) CASEMBC(0x1e30) | |
| 907 CASEMBC(0x1e34) | |
| 908 regmbc('K'); REGMBC(0x136) REGMBC(0x1e8) | |
| 909 REGMBC(0x1e30) REGMBC(0x1e34) | |
| 910 return; | |
| 911 case 'L': CASEMBC(0x139) CASEMBC(0x13b) CASEMBC(0x13d) | |
| 912 CASEMBC(0x13f) CASEMBC(0x141) CASEMBC(0x1e3a) | |
| 913 regmbc('L'); REGMBC(0x139) REGMBC(0x13b) | |
| 914 REGMBC(0x13d) REGMBC(0x13f) REGMBC(0x141) | |
| 915 REGMBC(0x1e3a) | |
| 916 return; | |
| 917 case 'M': CASEMBC(0x1e3e) CASEMBC(0x1e40) | |
| 918 regmbc('M'); REGMBC(0x1e3e) REGMBC(0x1e40) | |
| 167 | 919 return; |
| 236 | 920 case 'N': case '\321': |
| 2974 | 921 CASEMBC(0x143) CASEMBC(0x145) CASEMBC(0x147) CASEMBC(0x1e44) |
| 922 CASEMBC(0x1e48) | |
| 236 | 923 regmbc('N'); regmbc('\321'); |
| 2974 | 924 REGMBC(0x143) REGMBC(0x145) REGMBC(0x147) |
| 925 REGMBC(0x1e44) REGMBC(0x1e48) | |
| 167 | 926 return; |
| 236 | 927 case 'O': case '\322': case '\323': case '\324': case '\325': |
| 2974 | 928 case '\326': case '\330': |
| 929 CASEMBC(0x14c) CASEMBC(0x14e) CASEMBC(0x150) CASEMBC(0x1a0) | |
| 930 CASEMBC(0x1d1) CASEMBC(0x1ea) CASEMBC(0x1ec) CASEMBC(0x1ece) | |
| 236 | 931 regmbc('O'); regmbc('\322'); regmbc('\323'); |
| 932 regmbc('\324'); regmbc('\325'); regmbc('\326'); | |
| 2974 | 933 regmbc('\330'); |
| 934 REGMBC(0x14c) REGMBC(0x14e) REGMBC(0x150) | |
| 935 REGMBC(0x1a0) REGMBC(0x1d1) REGMBC(0x1ea) | |
| 936 REGMBC(0x1ec) REGMBC(0x1ece) | |
| 937 return; | |
| 938 case 'P': case 0x1e54: case 0x1e56: | |
| 939 regmbc('P'); REGMBC(0x1e54) REGMBC(0x1e56) | |
| 940 return; | |
| 941 case 'R': CASEMBC(0x154) CASEMBC(0x156) CASEMBC(0x158) | |
| 942 CASEMBC(0x1e58) CASEMBC(0x1e5e) | |
| 943 regmbc('R'); REGMBC(0x154) REGMBC(0x156) REGMBC(0x158) | |
| 944 REGMBC(0x1e58) REGMBC(0x1e5e) | |
| 945 return; | |
| 946 case 'S': CASEMBC(0x15a) CASEMBC(0x15c) CASEMBC(0x15e) | |
| 947 CASEMBC(0x160) CASEMBC(0x1e60) | |
| 948 regmbc('S'); REGMBC(0x15a) REGMBC(0x15c) | |
| 949 REGMBC(0x15e) REGMBC(0x160) REGMBC(0x1e60) | |
| 950 return; | |
| 951 case 'T': CASEMBC(0x162) CASEMBC(0x164) CASEMBC(0x166) | |
| 952 CASEMBC(0x1e6a) CASEMBC(0x1e6e) | |
| 953 regmbc('T'); REGMBC(0x162) REGMBC(0x164) | |
| 954 REGMBC(0x166) REGMBC(0x1e6a) REGMBC(0x1e6e) | |
| 167 | 955 return; |
| 236 | 956 case 'U': case '\331': case '\332': case '\333': case '\334': |
| 2974 | 957 CASEMBC(0x168) CASEMBC(0x16a) CASEMBC(0x16c) CASEMBC(0x16e) |
| 958 CASEMBC(0x170) CASEMBC(0x172) CASEMBC(0x1af) CASEMBC(0x1d3) | |
| 959 CASEMBC(0x1ee6) | |
| 236 | 960 regmbc('U'); regmbc('\331'); regmbc('\332'); |
| 961 regmbc('\333'); regmbc('\334'); | |
| 2974 | 962 REGMBC(0x168) REGMBC(0x16a) REGMBC(0x16c) |
| 963 REGMBC(0x16e) REGMBC(0x170) REGMBC(0x172) | |
| 964 REGMBC(0x1af) REGMBC(0x1d3) REGMBC(0x1ee6) | |
| 965 return; | |
| 966 case 'V': CASEMBC(0x1e7c) | |
| 967 regmbc('V'); REGMBC(0x1e7c) | |
| 968 return; | |
| 969 case 'W': CASEMBC(0x174) CASEMBC(0x1e80) CASEMBC(0x1e82) | |
| 970 CASEMBC(0x1e84) CASEMBC(0x1e86) | |
| 971 regmbc('W'); REGMBC(0x174) REGMBC(0x1e80) | |
| 972 REGMBC(0x1e82) REGMBC(0x1e84) REGMBC(0x1e86) | |
| 973 return; | |
| 974 case 'X': CASEMBC(0x1e8a) CASEMBC(0x1e8c) | |
| 975 regmbc('X'); REGMBC(0x1e8a) REGMBC(0x1e8c) | |
| 167 | 976 return; |
| 236 | 977 case 'Y': case '\335': |
| 2974 | 978 CASEMBC(0x176) CASEMBC(0x178) CASEMBC(0x1e8e) CASEMBC(0x1ef2) |
| 979 CASEMBC(0x1ef6) CASEMBC(0x1ef8) | |
| 236 | 980 regmbc('Y'); regmbc('\335'); |
| 2974 | 981 REGMBC(0x176) REGMBC(0x178) REGMBC(0x1e8e) |
| 982 REGMBC(0x1ef2) REGMBC(0x1ef6) REGMBC(0x1ef8) | |
| 983 return; | |
| 984 case 'Z': CASEMBC(0x179) CASEMBC(0x17b) CASEMBC(0x17d) | |
| 985 CASEMBC(0x1b5) CASEMBC(0x1e90) CASEMBC(0x1e94) | |
| 986 regmbc('Z'); REGMBC(0x179) REGMBC(0x17b) | |
| 987 REGMBC(0x17d) REGMBC(0x1b5) REGMBC(0x1e90) | |
| 988 REGMBC(0x1e94) | |
| 167 | 989 return; |
| 236 | 990 case 'a': case '\340': case '\341': case '\342': |
| 991 case '\343': case '\344': case '\345': | |
| 2974 | 992 CASEMBC(0x101) CASEMBC(0x103) CASEMBC(0x105) CASEMBC(0x1ce) |
| 993 CASEMBC(0x1df) CASEMBC(0x1e1) CASEMBC(0x1ea3) | |
| 236 | 994 regmbc('a'); regmbc('\340'); regmbc('\341'); |
| 995 regmbc('\342'); regmbc('\343'); regmbc('\344'); | |
| 996 regmbc('\345'); | |
| 2974 | 997 REGMBC(0x101) REGMBC(0x103) REGMBC(0x105) |
| 998 REGMBC(0x1ce) REGMBC(0x1df) REGMBC(0x1e1) | |
| 999 REGMBC(0x1ea3) | |
| 1000 return; | |
| 1001 case 'b': CASEMBC(0x1e03) CASEMBC(0x1e07) | |
| 1002 regmbc('b'); REGMBC(0x1e03) REGMBC(0x1e07) | |
| 167 | 1003 return; |
| 236 | 1004 case 'c': case '\347': |
| 2974 | 1005 CASEMBC(0x107) CASEMBC(0x109) CASEMBC(0x10b) CASEMBC(0x10d) |
| 236 | 1006 regmbc('c'); regmbc('\347'); |
| 2974 | 1007 REGMBC(0x107) REGMBC(0x109) REGMBC(0x10b) |
| 1008 REGMBC(0x10d) | |
| 1009 return; | |
| 1010 case 'd': CASEMBC(0x10f) CASEMBC(0x111) CASEMBC(0x1d0b) | |
| 1011 CASEMBC(0x1e11) | |
| 1012 regmbc('d'); REGMBC(0x10f) REGMBC(0x111) | |
| 1013 REGMBC(0x1e0b) REGMBC(0x01e0f) REGMBC(0x1e11) | |
| 167 | 1014 return; |
| 236 | 1015 case 'e': case '\350': case '\351': case '\352': case '\353': |
| 2974 | 1016 CASEMBC(0x113) CASEMBC(0x115) CASEMBC(0x117) CASEMBC(0x119) |
| 1017 CASEMBC(0x11b) CASEMBC(0x1ebb) CASEMBC(0x1ebd) | |
| 236 | 1018 regmbc('e'); regmbc('\350'); regmbc('\351'); |
| 1019 regmbc('\352'); regmbc('\353'); | |
| 2974 | 1020 REGMBC(0x113) REGMBC(0x115) REGMBC(0x117) |
| 1021 REGMBC(0x119) REGMBC(0x11b) REGMBC(0x1ebb) | |
| 1022 REGMBC(0x1ebd) | |
| 1023 return; | |
| 1024 case 'f': CASEMBC(0x1e1f) | |
| 1025 regmbc('f'); REGMBC(0x1e1f) | |
| 1026 return; | |
| 1027 case 'g': CASEMBC(0x11d) CASEMBC(0x11f) CASEMBC(0x121) | |
| 1028 CASEMBC(0x123) CASEMBC(0x1e5) CASEMBC(0x1e7) CASEMBC(0x1f5) | |
| 1029 CASEMBC(0x1e21) | |
| 1030 regmbc('g'); REGMBC(0x11d) REGMBC(0x11f) | |
| 1031 REGMBC(0x121) REGMBC(0x123) REGMBC(0x1e5) | |
| 1032 REGMBC(0x1e7) REGMBC(0x1f5) REGMBC(0x1e21) | |
| 1033 return; | |
| 1034 case 'h': CASEMBC(0x125) CASEMBC(0x127) CASEMBC(0x1e23) | |
| 1035 CASEMBC(0x1e27) CASEMBC(0x1e29) CASEMBC(0x1e96) | |
| 1036 regmbc('h'); REGMBC(0x125) REGMBC(0x127) | |
| 1037 REGMBC(0x1e23) REGMBC(0x1e27) REGMBC(0x1e29) | |
| 1038 REGMBC(0x1e96) | |
| 167 | 1039 return; |
| 236 | 1040 case 'i': case '\354': case '\355': case '\356': case '\357': |
| 2974 | 1041 CASEMBC(0x129) CASEMBC(0x12b) CASEMBC(0x12d) CASEMBC(0x12f) |
| 1042 CASEMBC(0x1d0) CASEMBC(0x1ec9) | |
| 236 | 1043 regmbc('i'); regmbc('\354'); regmbc('\355'); |
| 1044 regmbc('\356'); regmbc('\357'); | |
| 2974 | 1045 REGMBC(0x129) REGMBC(0x12b) REGMBC(0x12d) |
| 1046 REGMBC(0x12f) REGMBC(0x1d0) REGMBC(0x1ec9) | |
| 1047 return; | |
| 1048 case 'j': CASEMBC(0x135) CASEMBC(0x1f0) | |
| 1049 regmbc('j'); REGMBC(0x135) REGMBC(0x1f0) | |
| 1050 return; | |
| 1051 case 'k': CASEMBC(0x137) CASEMBC(0x1e9) CASEMBC(0x1e31) | |
| 1052 CASEMBC(0x1e35) | |
| 1053 regmbc('k'); REGMBC(0x137) REGMBC(0x1e9) | |
| 1054 REGMBC(0x1e31) REGMBC(0x1e35) | |
| 1055 return; | |
| 1056 case 'l': CASEMBC(0x13a) CASEMBC(0x13c) CASEMBC(0x13e) | |
| 1057 CASEMBC(0x140) CASEMBC(0x142) CASEMBC(0x1e3b) | |
| 1058 regmbc('l'); REGMBC(0x13a) REGMBC(0x13c) | |
| 1059 REGMBC(0x13e) REGMBC(0x140) REGMBC(0x142) | |
| 1060 REGMBC(0x1e3b) | |
| 1061 return; | |
| 1062 case 'm': CASEMBC(0x1e3f) CASEMBC(0x1e41) | |
| 1063 regmbc('m'); REGMBC(0x1e3f) REGMBC(0x1e41) | |
| 167 | 1064 return; |
| 236 | 1065 case 'n': case '\361': |
| 2974 | 1066 CASEMBC(0x144) CASEMBC(0x146) CASEMBC(0x148) CASEMBC(0x149) |
| 1067 CASEMBC(0x1e45) CASEMBC(0x1e49) | |
| 236 | 1068 regmbc('n'); regmbc('\361'); |
| 2974 | 1069 REGMBC(0x144) REGMBC(0x146) REGMBC(0x148) |
| 1070 REGMBC(0x149) REGMBC(0x1e45) REGMBC(0x1e49) | |
| 167 | 1071 return; |
| 236 | 1072 case 'o': case '\362': case '\363': case '\364': case '\365': |
| 2974 | 1073 case '\366': case '\370': |
| 1074 CASEMBC(0x14d) CASEMBC(0x14f) CASEMBC(0x151) CASEMBC(0x1a1) | |
| 1075 CASEMBC(0x1d2) CASEMBC(0x1eb) CASEMBC(0x1ed) CASEMBC(0x1ecf) | |
| 236 | 1076 regmbc('o'); regmbc('\362'); regmbc('\363'); |
| 1077 regmbc('\364'); regmbc('\365'); regmbc('\366'); | |
| 2974 | 1078 regmbc('\370'); |
| 1079 REGMBC(0x14d) REGMBC(0x14f) REGMBC(0x151) | |
| 1080 REGMBC(0x1a1) REGMBC(0x1d2) REGMBC(0x1eb) | |
| 1081 REGMBC(0x1ed) REGMBC(0x1ecf) | |
| 1082 return; | |
| 1083 case 'p': CASEMBC(0x1e55) CASEMBC(0x1e57) | |
| 1084 regmbc('p'); REGMBC(0x1e55) REGMBC(0x1e57) | |
| 1085 return; | |
| 1086 case 'r': CASEMBC(0x155) CASEMBC(0x157) CASEMBC(0x159) | |
| 1087 CASEMBC(0x1e59) CASEMBC(0x1e5f) | |
| 1088 regmbc('r'); REGMBC(0x155) REGMBC(0x157) REGMBC(0x159) | |
| 1089 REGMBC(0x1e59) REGMBC(0x1e5f) | |
| 1090 return; | |
| 1091 case 's': CASEMBC(0x15b) CASEMBC(0x15d) CASEMBC(0x15f) | |
| 1092 CASEMBC(0x161) CASEMBC(0x1e61) | |
| 1093 regmbc('s'); REGMBC(0x15b) REGMBC(0x15d) | |
| 1094 REGMBC(0x15f) REGMBC(0x161) REGMBC(0x1e61) | |
| 1095 return; | |
| 1096 case 't': CASEMBC(0x163) CASEMBC(0x165) CASEMBC(0x167) | |
| 1097 CASEMBC(0x1e6b) CASEMBC(0x1e6f) CASEMBC(0x1e97) | |
| 1098 regmbc('t'); REGMBC(0x163) REGMBC(0x165) REGMBC(0x167) | |
| 1099 REGMBC(0x1e6b) REGMBC(0x1e6f) REGMBC(0x1e97) | |
| 167 | 1100 return; |
| 236 | 1101 case 'u': case '\371': case '\372': case '\373': case '\374': |
| 2974 | 1102 CASEMBC(0x169) CASEMBC(0x16b) CASEMBC(0x16d) CASEMBC(0x16f) |
| 1103 CASEMBC(0x171) CASEMBC(0x173) CASEMBC(0x1b0) CASEMBC(0x1d4) | |
| 1104 CASEMBC(0x1ee7) | |
| 236 | 1105 regmbc('u'); regmbc('\371'); regmbc('\372'); |
| 1106 regmbc('\373'); regmbc('\374'); | |
| 2974 | 1107 REGMBC(0x169) REGMBC(0x16b) REGMBC(0x16d) |
| 1108 REGMBC(0x16f) REGMBC(0x171) REGMBC(0x173) | |
| 1109 REGMBC(0x1b0) REGMBC(0x1d4) REGMBC(0x1ee7) | |
| 1110 return; | |
| 1111 case 'v': CASEMBC(0x1e7d) | |
| 1112 regmbc('v'); REGMBC(0x1e7d) | |
| 1113 return; | |
| 1114 case 'w': CASEMBC(0x175) CASEMBC(0x1e81) CASEMBC(0x1e83) | |
| 1115 CASEMBC(0x1e85) CASEMBC(0x1e87) CASEMBC(0x1e98) | |
| 1116 regmbc('w'); REGMBC(0x175) REGMBC(0x1e81) | |
| 1117 REGMBC(0x1e83) REGMBC(0x1e85) REGMBC(0x1e87) | |
| 1118 REGMBC(0x1e98) | |
| 1119 return; | |
| 1120 case 'x': CASEMBC(0x1e8b) CASEMBC(0x1e8d) | |
| 1121 regmbc('x'); REGMBC(0x1e8b) REGMBC(0x1e8d) | |
| 167 | 1122 return; |
| 236 | 1123 case 'y': case '\375': case '\377': |
| 2974 | 1124 CASEMBC(0x177) CASEMBC(0x1e8f) CASEMBC(0x1e99) |
| 1125 CASEMBC(0x1ef3) CASEMBC(0x1ef7) CASEMBC(0x1ef9) | |
| 236 | 1126 regmbc('y'); regmbc('\375'); regmbc('\377'); |
| 2974 | 1127 REGMBC(0x177) REGMBC(0x1e8f) REGMBC(0x1e99) |
| 1128 REGMBC(0x1ef3) REGMBC(0x1ef7) REGMBC(0x1ef9) | |
| 1129 return; | |
| 1130 case 'z': CASEMBC(0x17a) CASEMBC(0x17c) CASEMBC(0x17e) | |
| 1131 CASEMBC(0x1b6) CASEMBC(0x1e91) CASEMBC(0x1e95) | |
| 1132 regmbc('z'); REGMBC(0x17a) REGMBC(0x17c) | |
| 1133 REGMBC(0x17e) REGMBC(0x1b6) REGMBC(0x1e91) | |
| 1134 REGMBC(0x1e95) | |
| 167 | 1135 return; |
| 1136 } | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
1137 #endif |
| 167 | 1138 } |
| 1139 regmbc(c); | |
| 1140 } | |
| 1141 | |
| 1142 /* | |
| 1143 * Check for a collating element "[.a.]". "pp" points to the '['. | |
| 1144 * Returns a character. Zero means that no item was recognized. Otherwise | |
| 1145 * "pp" is advanced to after the item. | |
| 1146 * Currently only single characters are recognized! | |
| 1147 */ | |
| 1148 static int | |
| 1149 get_coll_element(pp) | |
| 1150 char_u **pp; | |
| 1151 { | |
| 1152 int c; | |
| 1153 int l = 1; | |
| 1154 char_u *p = *pp; | |
| 1155 | |
| 1156 if (p[1] == '.') | |
| 1157 { | |
| 1158 #ifdef FEAT_MBYTE | |
| 1159 if (has_mbyte) | |
| 474 | 1160 l = (*mb_ptr2len)(p + 2); |
| 167 | 1161 #endif |
| 1162 if (p[l + 2] == '.' && p[l + 3] == ']') | |
| 1163 { | |
| 1164 #ifdef FEAT_MBYTE | |
| 1165 if (has_mbyte) | |
| 1166 c = mb_ptr2char(p + 2); | |
| 1167 else | |
| 1168 #endif | |
| 1169 c = p[2]; | |
| 1170 *pp += l + 4; | |
| 1171 return c; | |
| 1172 } | |
| 1173 } | |
| 1174 return 0; | |
| 1175 } | |
| 1176 | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1177 static void get_cpo_flags __ARGS((void)); |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1178 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
|
1179 static int reg_cpo_bsl; /* 'cpoptions' contains '\' flag */ |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1180 |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1181 static void |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1182 get_cpo_flags() |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1183 { |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1184 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
|
1185 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
|
1186 } |
| 167 | 1187 |
| 1188 /* | |
| 1189 * Skip over a "[]" range. | |
| 1190 * "p" must point to the character after the '['. | |
| 1191 * The returned pointer is on the matching ']', or the terminating NUL. | |
| 1192 */ | |
| 1193 static char_u * | |
| 1194 skip_anyof(p) | |
| 1195 char_u *p; | |
| 1196 { | |
| 1197 #ifdef FEAT_MBYTE | |
| 1198 int l; | |
| 1199 #endif | |
| 1200 | |
| 1201 if (*p == '^') /* Complement of range. */ | |
| 1202 ++p; | |
| 1203 if (*p == ']' || *p == '-') | |
| 1204 ++p; | |
| 1205 while (*p != NUL && *p != ']') | |
| 1206 { | |
| 1207 #ifdef FEAT_MBYTE | |
| 474 | 1208 if (has_mbyte && (l = (*mb_ptr2len)(p)) > 1) |
| 167 | 1209 p += l; |
| 1210 else | |
| 1211 #endif | |
| 1212 if (*p == '-') | |
| 1213 { | |
| 1214 ++p; | |
| 1215 if (*p != ']' && *p != NUL) | |
| 1216 mb_ptr_adv(p); | |
| 1217 } | |
| 1218 else if (*p == '\\' | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1219 && !reg_cpo_bsl |
| 167 | 1220 && (vim_strchr(REGEXP_INRANGE, p[1]) != NULL |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1221 || (!reg_cpo_lit && vim_strchr(REGEXP_ABBR, p[1]) != NULL))) |
| 167 | 1222 p += 2; |
| 1223 else if (*p == '[') | |
| 1224 { | |
| 1225 if (get_char_class(&p) == CLASS_NONE | |
| 1226 && get_equi_class(&p) == 0 | |
| 1227 && get_coll_element(&p) == 0) | |
| 1228 ++p; /* It was not a class name */ | |
| 1229 } | |
| 1230 else | |
| 1231 ++p; | |
| 1232 } | |
| 1233 | |
| 1234 return p; | |
| 1235 } | |
| 1236 | |
| 1237 /* | |
| 7 | 1238 * Skip past regular expression. |
| 153 | 1239 * Stop at end of "startp" or where "dirc" is found ('/', '?', etc). |
| 7 | 1240 * Take care of characters with a backslash in front of it. |
| 1241 * Skip strings inside [ and ]. | |
| 1242 * When "newp" is not NULL and "dirc" is '?', make an allocated copy of the | |
| 1243 * expression and change "\?" to "?". If "*newp" is not NULL the expression | |
| 1244 * is changed in-place. | |
| 1245 */ | |
| 1246 char_u * | |
| 1247 skip_regexp(startp, dirc, magic, newp) | |
| 1248 char_u *startp; | |
| 1249 int dirc; | |
| 1250 int magic; | |
| 1251 char_u **newp; | |
| 1252 { | |
| 1253 int mymagic; | |
| 1254 char_u *p = startp; | |
| 1255 | |
| 1256 if (magic) | |
| 1257 mymagic = MAGIC_ON; | |
| 1258 else | |
| 1259 mymagic = MAGIC_OFF; | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1260 get_cpo_flags(); |
| 7 | 1261 |
| 39 | 1262 for (; p[0] != NUL; mb_ptr_adv(p)) |
| 7 | 1263 { |
| 1264 if (p[0] == dirc) /* found end of regexp */ | |
| 1265 break; | |
| 1266 if ((p[0] == '[' && mymagic >= MAGIC_ON) | |
| 1267 || (p[0] == '\\' && p[1] == '[' && mymagic <= MAGIC_OFF)) | |
| 1268 { | |
| 1269 p = skip_anyof(p + 1); | |
| 1270 if (p[0] == NUL) | |
| 1271 break; | |
| 1272 } | |
| 1273 else if (p[0] == '\\' && p[1] != NUL) | |
| 1274 { | |
| 1275 if (dirc == '?' && newp != NULL && p[1] == '?') | |
| 1276 { | |
| 1277 /* change "\?" to "?", make a copy first. */ | |
| 1278 if (*newp == NULL) | |
| 1279 { | |
| 1280 *newp = vim_strsave(startp); | |
| 1281 if (*newp != NULL) | |
| 1282 p = *newp + (p - startp); | |
| 1283 } | |
| 1284 if (*newp != NULL) | |
| 1621 | 1285 STRMOVE(p, p + 1); |
| 7 | 1286 else |
| 1287 ++p; | |
| 1288 } | |
| 1289 else | |
| 1290 ++p; /* skip next character */ | |
| 1291 if (*p == 'v') | |
| 1292 mymagic = MAGIC_ALL; | |
| 1293 else if (*p == 'V') | |
| 1294 mymagic = MAGIC_NONE; | |
| 1295 } | |
| 1296 } | |
| 1297 return p; | |
| 1298 } | |
| 1299 | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1300 static regprog_T *bt_regcomp __ARGS((char_u *expr, int re_flags)); |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1301 static void bt_regfree __ARGS((regprog_T *prog)); |
| 4444 | 1302 |
| 7 | 1303 /* |
| 4444 | 1304 * bt_regcomp() - compile a regular expression into internal code for the |
| 1305 * traditional back track matcher. | |
| 41 | 1306 * Returns the program in allocated space. Returns NULL for an error. |
| 7 | 1307 * |
| 1308 * We can't allocate space until we know how big the compiled form will be, | |
| 1309 * but we can't compile it (and thus know how big it is) until we've got a | |
| 1310 * place to put the code. So we cheat: we compile it twice, once with code | |
| 1311 * generation turned off and size counting turned on, and once "for real". | |
| 1312 * This also means that we don't allocate space until we are sure that the | |
| 1313 * thing really will compile successfully, and we never have to move the | |
| 1314 * code and thus invalidate pointers into it. (Note that it has to be in | |
| 1315 * one piece because vim_free() must be able to free it all.) | |
| 1316 * | |
| 1317 * Whether upper/lower case is to be ignored is decided when executing the | |
| 1318 * program, it does not matter here. | |
| 1319 * | |
| 1320 * Beware that the optimization-preparation code in here knows about some | |
| 1321 * of the structure of the compiled regexp. | |
| 1322 * "re_flags": RE_MAGIC and/or RE_STRING. | |
| 1323 */ | |
| 4444 | 1324 static regprog_T * |
| 1325 bt_regcomp(expr, re_flags) | |
| 7 | 1326 char_u *expr; |
| 1327 int re_flags; | |
| 1328 { | |
| 4444 | 1329 bt_regprog_T *r; |
| 7 | 1330 char_u *scan; |
| 1331 char_u *longest; | |
| 1332 int len; | |
| 1333 int flags; | |
| 1334 | |
| 1335 if (expr == NULL) | |
| 1336 EMSG_RET_NULL(_(e_null)); | |
| 1337 | |
| 1338 init_class_tab(); | |
| 1339 | |
| 1340 /* | |
| 1341 * First pass: determine size, legality. | |
| 1342 */ | |
| 1343 regcomp_start(expr, re_flags); | |
| 1344 regcode = JUST_CALC_SIZE; | |
| 1345 regc(REGMAGIC); | |
| 1346 if (reg(REG_NOPAREN, &flags) == NULL) | |
| 1347 return NULL; | |
| 1348 | |
| 1349 /* Small enough for pointer-storage convention? */ | |
| 1350 #ifdef SMALL_MALLOC /* 16 bit storage allocation */ | |
| 1351 if (regsize >= 65536L - 256L) | |
| 1352 EMSG_RET_NULL(_("E339: Pattern too long")); | |
| 1353 #endif | |
| 1354 | |
| 1355 /* Allocate space. */ | |
| 4444 | 1356 r = (bt_regprog_T *)lalloc(sizeof(bt_regprog_T) + regsize, TRUE); |
| 7 | 1357 if (r == NULL) |
| 1358 return NULL; | |
| 1359 | |
| 1360 /* | |
| 1361 * Second pass: emit code. | |
| 1362 */ | |
| 1363 regcomp_start(expr, re_flags); | |
| 1364 regcode = r->program; | |
| 1365 regc(REGMAGIC); | |
| 2010 | 1366 if (reg(REG_NOPAREN, &flags) == NULL || reg_toolong) |
| 7 | 1367 { |
| 1368 vim_free(r); | |
| 2010 | 1369 if (reg_toolong) |
| 1370 EMSG_RET_NULL(_("E339: Pattern too long")); | |
| 7 | 1371 return NULL; |
| 1372 } | |
| 1373 | |
| 1374 /* Dig out information for optimizations. */ | |
| 1375 r->regstart = NUL; /* Worst-case defaults. */ | |
| 1376 r->reganch = 0; | |
| 1377 r->regmust = NULL; | |
| 1378 r->regmlen = 0; | |
| 1379 r->regflags = regflags; | |
| 1380 if (flags & HASNL) | |
| 1381 r->regflags |= RF_HASNL; | |
| 1382 if (flags & HASLOOKBH) | |
| 1383 r->regflags |= RF_LOOKBH; | |
| 1384 #ifdef FEAT_SYN_HL | |
| 1385 /* Remember whether this pattern has any \z specials in it. */ | |
| 1386 r->reghasz = re_has_z; | |
| 1387 #endif | |
| 1388 scan = r->program + 1; /* First BRANCH. */ | |
| 1389 if (OP(regnext(scan)) == END) /* Only one top-level choice. */ | |
| 1390 { | |
| 1391 scan = OPERAND(scan); | |
| 1392 | |
| 1393 /* Starting-point info. */ | |
| 1394 if (OP(scan) == BOL || OP(scan) == RE_BOF) | |
| 1395 { | |
| 1396 r->reganch++; | |
| 1397 scan = regnext(scan); | |
| 1398 } | |
| 1399 | |
| 1400 if (OP(scan) == EXACTLY) | |
| 1401 { | |
| 1402 #ifdef FEAT_MBYTE | |
| 1403 if (has_mbyte) | |
| 1404 r->regstart = (*mb_ptr2char)(OPERAND(scan)); | |
| 1405 else | |
| 1406 #endif | |
| 1407 r->regstart = *OPERAND(scan); | |
| 1408 } | |
| 1409 else if ((OP(scan) == BOW | |
| 1410 || OP(scan) == EOW | |
| 1411 || OP(scan) == NOTHING | |
| 1412 || OP(scan) == MOPEN + 0 || OP(scan) == NOPEN | |
| 1413 || OP(scan) == MCLOSE + 0 || OP(scan) == NCLOSE) | |
| 1414 && OP(regnext(scan)) == EXACTLY) | |
| 1415 { | |
| 1416 #ifdef FEAT_MBYTE | |
| 1417 if (has_mbyte) | |
| 1418 r->regstart = (*mb_ptr2char)(OPERAND(regnext(scan))); | |
| 1419 else | |
| 1420 #endif | |
| 1421 r->regstart = *OPERAND(regnext(scan)); | |
| 1422 } | |
| 1423 | |
| 1424 /* | |
| 1425 * If there's something expensive in the r.e., find the longest | |
| 1426 * literal string that must appear and make it the regmust. Resolve | |
| 1427 * ties in favor of later strings, since the regstart check works | |
| 1428 * with the beginning of the r.e. and avoiding duplication | |
| 1429 * strengthens checking. Not a strong reason, but sufficient in the | |
| 1430 * absence of others. | |
| 1431 */ | |
| 1432 /* | |
| 1433 * When the r.e. starts with BOW, it is faster to look for a regmust | |
| 1434 * first. Used a lot for "#" and "*" commands. (Added by mool). | |
| 1435 */ | |
| 1436 if ((flags & SPSTART || OP(scan) == BOW || OP(scan) == EOW) | |
| 1437 && !(flags & HASNL)) | |
| 1438 { | |
| 1439 longest = NULL; | |
| 1440 len = 0; | |
| 1441 for (; scan != NULL; scan = regnext(scan)) | |
| 1442 if (OP(scan) == EXACTLY && STRLEN(OPERAND(scan)) >= (size_t)len) | |
| 1443 { | |
| 1444 longest = OPERAND(scan); | |
| 1445 len = (int)STRLEN(OPERAND(scan)); | |
| 1446 } | |
| 1447 r->regmust = longest; | |
| 1448 r->regmlen = len; | |
| 1449 } | |
| 1450 } | |
| 4444 | 1451 #ifdef BT_REGEXP_DUMP |
| 7 | 1452 regdump(expr, r); |
| 1453 #endif | |
| 4444 | 1454 r->engine = &bt_regengine; |
| 1455 return (regprog_T *)r; | |
| 7 | 1456 } |
| 1457 | |
| 1458 /* | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1459 * 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
|
1460 */ |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1461 static void |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1462 bt_regfree(prog) |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1463 regprog_T *prog; |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1464 { |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1465 vim_free(prog); |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1466 } |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1467 |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1468 /* |
| 7 | 1469 * Setup to parse the regexp. Used once to get the length and once to do it. |
| 1470 */ | |
| 1471 static void | |
| 1472 regcomp_start(expr, re_flags) | |
| 1473 char_u *expr; | |
| 1474 int re_flags; /* see vim_regcomp() */ | |
| 1475 { | |
| 1476 initchr(expr); | |
| 1477 if (re_flags & RE_MAGIC) | |
| 1478 reg_magic = MAGIC_ON; | |
| 1479 else | |
| 1480 reg_magic = MAGIC_OFF; | |
| 1481 reg_string = (re_flags & RE_STRING); | |
| 481 | 1482 reg_strict = (re_flags & RE_STRICT); |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1483 get_cpo_flags(); |
| 7 | 1484 |
| 1485 num_complex_braces = 0; | |
| 1486 regnpar = 1; | |
| 1487 vim_memset(had_endbrace, 0, sizeof(had_endbrace)); | |
| 1488 #ifdef FEAT_SYN_HL | |
| 1489 regnzpar = 1; | |
| 1490 re_has_z = 0; | |
| 1491 #endif | |
| 1492 regsize = 0L; | |
| 2010 | 1493 reg_toolong = FALSE; |
| 7 | 1494 regflags = 0; |
| 1495 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1496 had_eol = FALSE; | |
| 1497 #endif | |
| 1498 } | |
| 1499 | |
| 1500 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1501 /* | |
| 1502 * Check if during the previous call to vim_regcomp the EOL item "$" has been | |
| 1503 * found. This is messy, but it works fine. | |
| 1504 */ | |
| 1505 int | |
| 1506 vim_regcomp_had_eol() | |
| 1507 { | |
| 1508 return had_eol; | |
| 1509 } | |
| 1510 #endif | |
| 1511 | |
| 1512 /* | |
| 4444 | 1513 * Parse regular expression, i.e. main body or parenthesized thing. |
| 7 | 1514 * |
| 1515 * Caller must absorb opening parenthesis. | |
| 1516 * | |
| 1517 * Combining parenthesis handling with the base level of regular expression | |
| 1518 * is a trifle forced, but the need to tie the tails of the branches to what | |
| 1519 * follows makes it hard to avoid. | |
| 1520 */ | |
| 1521 static char_u * | |
| 1522 reg(paren, flagp) | |
| 1523 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
| 1524 int *flagp; | |
| 1525 { | |
| 1526 char_u *ret; | |
| 1527 char_u *br; | |
| 1528 char_u *ender; | |
| 1529 int parno = 0; | |
| 1530 int flags; | |
| 1531 | |
| 1532 *flagp = HASWIDTH; /* Tentatively. */ | |
| 1533 | |
| 1534 #ifdef FEAT_SYN_HL | |
| 1535 if (paren == REG_ZPAREN) | |
| 1536 { | |
| 1537 /* Make a ZOPEN node. */ | |
| 1538 if (regnzpar >= NSUBEXP) | |
| 1539 EMSG_RET_NULL(_("E50: Too many \\z(")); | |
| 1540 parno = regnzpar; | |
| 1541 regnzpar++; | |
| 1542 ret = regnode(ZOPEN + parno); | |
| 1543 } | |
| 1544 else | |
| 1545 #endif | |
| 1546 if (paren == REG_PAREN) | |
| 1547 { | |
| 1548 /* Make a MOPEN node. */ | |
| 1549 if (regnpar >= NSUBEXP) | |
| 4444 | 1550 EMSG2_RET_NULL(_("E51: Too many %s("), reg_magic == MAGIC_ALL); |
| 7 | 1551 parno = regnpar; |
| 1552 ++regnpar; | |
| 1553 ret = regnode(MOPEN + parno); | |
| 1554 } | |
| 1555 else if (paren == REG_NPAREN) | |
| 1556 { | |
| 1557 /* Make a NOPEN node. */ | |
| 1558 ret = regnode(NOPEN); | |
| 1559 } | |
| 1560 else | |
| 1561 ret = NULL; | |
| 1562 | |
| 1563 /* Pick up the branches, linking them together. */ | |
| 1564 br = regbranch(&flags); | |
| 1565 if (br == NULL) | |
| 1566 return NULL; | |
| 1567 if (ret != NULL) | |
| 1568 regtail(ret, br); /* [MZ]OPEN -> first. */ | |
| 1569 else | |
| 1570 ret = br; | |
| 1571 /* If one of the branches can be zero-width, the whole thing can. | |
| 1572 * If one of the branches has * at start or matches a line-break, the | |
| 1573 * whole thing can. */ | |
| 1574 if (!(flags & HASWIDTH)) | |
| 1575 *flagp &= ~HASWIDTH; | |
| 1576 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1577 while (peekchr() == Magic('|')) | |
| 1578 { | |
| 1579 skipchr(); | |
| 1580 br = regbranch(&flags); | |
| 2010 | 1581 if (br == NULL || reg_toolong) |
| 7 | 1582 return NULL; |
| 1583 regtail(ret, br); /* BRANCH -> BRANCH. */ | |
| 1584 if (!(flags & HASWIDTH)) | |
| 1585 *flagp &= ~HASWIDTH; | |
| 1586 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1587 } | |
| 1588 | |
| 1589 /* Make a closing node, and hook it on the end. */ | |
| 1590 ender = regnode( | |
| 1591 #ifdef FEAT_SYN_HL | |
| 1592 paren == REG_ZPAREN ? ZCLOSE + parno : | |
| 1593 #endif | |
| 1594 paren == REG_PAREN ? MCLOSE + parno : | |
| 1595 paren == REG_NPAREN ? NCLOSE : END); | |
| 1596 regtail(ret, ender); | |
| 1597 | |
| 1598 /* Hook the tails of the branches to the closing node. */ | |
| 1599 for (br = ret; br != NULL; br = regnext(br)) | |
| 1600 regoptail(br, ender); | |
| 1601 | |
| 1602 /* Check for proper termination. */ | |
| 1603 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
| 1604 { | |
| 1605 #ifdef FEAT_SYN_HL | |
| 1606 if (paren == REG_ZPAREN) | |
| 308 | 1607 EMSG_RET_NULL(_("E52: Unmatched \\z(")); |
| 7 | 1608 else |
| 1609 #endif | |
| 1610 if (paren == REG_NPAREN) | |
| 4444 | 1611 EMSG2_RET_NULL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); |
| 7 | 1612 else |
| 4444 | 1613 EMSG2_RET_NULL(_(e_unmatchedp), reg_magic == MAGIC_ALL); |
| 7 | 1614 } |
| 1615 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
| 1616 { | |
| 1617 if (curchr == Magic(')')) | |
| 4444 | 1618 EMSG2_RET_NULL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); |
| 7 | 1619 else |
| 308 | 1620 EMSG_RET_NULL(_(e_trailing)); /* "Can't happen". */ |
| 7 | 1621 /* NOTREACHED */ |
| 1622 } | |
| 1623 /* | |
| 1624 * Here we set the flag allowing back references to this set of | |
| 1625 * parentheses. | |
| 1626 */ | |
| 1627 if (paren == REG_PAREN) | |
| 1628 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
| 1629 return ret; | |
| 1630 } | |
| 1631 | |
| 1632 /* | |
| 4444 | 1633 * Parse one alternative of an | operator. |
| 7 | 1634 * Implements the & operator. |
| 1635 */ | |
| 1636 static char_u * | |
| 1637 regbranch(flagp) | |
| 1638 int *flagp; | |
| 1639 { | |
| 1640 char_u *ret; | |
| 1641 char_u *chain = NULL; | |
| 1642 char_u *latest; | |
| 1643 int flags; | |
| 1644 | |
| 1645 *flagp = WORST | HASNL; /* Tentatively. */ | |
| 1646 | |
| 1647 ret = regnode(BRANCH); | |
| 1648 for (;;) | |
| 1649 { | |
| 1650 latest = regconcat(&flags); | |
| 1651 if (latest == NULL) | |
| 1652 return NULL; | |
| 1653 /* If one of the branches has width, the whole thing has. If one of | |
| 1654 * the branches anchors at start-of-line, the whole thing does. | |
| 1655 * If one of the branches uses look-behind, the whole thing does. */ | |
| 1656 *flagp |= flags & (HASWIDTH | SPSTART | HASLOOKBH); | |
| 1657 /* If one of the branches doesn't match a line-break, the whole thing | |
| 1658 * doesn't. */ | |
| 1659 *flagp &= ~HASNL | (flags & HASNL); | |
| 1660 if (chain != NULL) | |
| 1661 regtail(chain, latest); | |
| 1662 if (peekchr() != Magic('&')) | |
| 1663 break; | |
| 1664 skipchr(); | |
| 1665 regtail(latest, regnode(END)); /* operand ends */ | |
| 2010 | 1666 if (reg_toolong) |
| 1667 break; | |
| 7 | 1668 reginsert(MATCH, latest); |
| 1669 chain = latest; | |
| 1670 } | |
| 1671 | |
| 1672 return ret; | |
| 1673 } | |
| 1674 | |
| 1675 /* | |
| 4444 | 1676 * Parse one alternative of an | or & operator. |
| 7 | 1677 * Implements the concatenation operator. |
| 1678 */ | |
| 1679 static char_u * | |
| 1680 regconcat(flagp) | |
| 1681 int *flagp; | |
| 1682 { | |
| 1683 char_u *first = NULL; | |
| 1684 char_u *chain = NULL; | |
| 1685 char_u *latest; | |
| 1686 int flags; | |
| 1687 int cont = TRUE; | |
| 1688 | |
| 1689 *flagp = WORST; /* Tentatively. */ | |
| 1690 | |
| 1691 while (cont) | |
| 1692 { | |
| 1693 switch (peekchr()) | |
| 1694 { | |
| 1695 case NUL: | |
| 1696 case Magic('|'): | |
| 1697 case Magic('&'): | |
| 1698 case Magic(')'): | |
| 1699 cont = FALSE; | |
| 1700 break; | |
| 1701 case Magic('Z'): | |
| 1702 #ifdef FEAT_MBYTE | |
| 1703 regflags |= RF_ICOMBINE; | |
| 1704 #endif | |
| 1705 skipchr_keepstart(); | |
| 1706 break; | |
| 1707 case Magic('c'): | |
| 1708 regflags |= RF_ICASE; | |
| 1709 skipchr_keepstart(); | |
| 1710 break; | |
| 1711 case Magic('C'): | |
| 1712 regflags |= RF_NOICASE; | |
| 1713 skipchr_keepstart(); | |
| 1714 break; | |
| 1715 case Magic('v'): | |
| 1716 reg_magic = MAGIC_ALL; | |
| 1717 skipchr_keepstart(); | |
| 1718 curchr = -1; | |
| 1719 break; | |
| 1720 case Magic('m'): | |
| 1721 reg_magic = MAGIC_ON; | |
| 1722 skipchr_keepstart(); | |
| 1723 curchr = -1; | |
| 1724 break; | |
| 1725 case Magic('M'): | |
| 1726 reg_magic = MAGIC_OFF; | |
| 1727 skipchr_keepstart(); | |
| 1728 curchr = -1; | |
| 1729 break; | |
| 1730 case Magic('V'): | |
| 1731 reg_magic = MAGIC_NONE; | |
| 1732 skipchr_keepstart(); | |
| 1733 curchr = -1; | |
| 1734 break; | |
| 1735 default: | |
| 1736 latest = regpiece(&flags); | |
| 2010 | 1737 if (latest == NULL || reg_toolong) |
| 7 | 1738 return NULL; |
| 1739 *flagp |= flags & (HASWIDTH | HASNL | HASLOOKBH); | |
| 1740 if (chain == NULL) /* First piece. */ | |
| 1741 *flagp |= flags & SPSTART; | |
| 1742 else | |
| 1743 regtail(chain, latest); | |
| 1744 chain = latest; | |
| 1745 if (first == NULL) | |
| 1746 first = latest; | |
| 1747 break; | |
| 1748 } | |
| 1749 } | |
| 1750 if (first == NULL) /* Loop ran zero times. */ | |
| 1751 first = regnode(NOTHING); | |
| 1752 return first; | |
| 1753 } | |
| 1754 | |
| 1755 /* | |
| 4444 | 1756 * Parse something followed by possible [*+=]. |
| 7 | 1757 * |
| 1758 * Note that the branching code sequences used for = and the general cases | |
| 1759 * of * and + are somewhat optimized: they use the same NOTHING node as | |
| 1760 * both the endmarker for their branch list and the body of the last branch. | |
| 1761 * It might seem that this node could be dispensed with entirely, but the | |
| 1762 * endmarker role is not redundant. | |
| 1763 */ | |
| 1764 static char_u * | |
| 1765 regpiece(flagp) | |
| 1766 int *flagp; | |
| 1767 { | |
| 1768 char_u *ret; | |
| 1769 int op; | |
| 1770 char_u *next; | |
| 1771 int flags; | |
| 1772 long minval; | |
| 1773 long maxval; | |
| 1774 | |
| 1775 ret = regatom(&flags); | |
| 1776 if (ret == NULL) | |
| 1777 return NULL; | |
| 1778 | |
| 1779 op = peekchr(); | |
| 1780 if (re_multi_type(op) == NOT_MULTI) | |
| 1781 { | |
| 1782 *flagp = flags; | |
| 1783 return ret; | |
| 1784 } | |
| 1785 /* default flags */ | |
| 1786 *flagp = (WORST | SPSTART | (flags & (HASNL | HASLOOKBH))); | |
| 1787 | |
| 1788 skipchr(); | |
| 1789 switch (op) | |
| 1790 { | |
| 1791 case Magic('*'): | |
| 1792 if (flags & SIMPLE) | |
| 1793 reginsert(STAR, ret); | |
| 1794 else | |
| 1795 { | |
| 1796 /* Emit x* as (x&|), where & means "self". */ | |
| 1797 reginsert(BRANCH, ret); /* Either x */ | |
| 1798 regoptail(ret, regnode(BACK)); /* and loop */ | |
| 1799 regoptail(ret, ret); /* back */ | |
| 1800 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1801 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1802 } | |
| 1803 break; | |
| 1804 | |
| 1805 case Magic('+'): | |
| 1806 if (flags & SIMPLE) | |
| 1807 reginsert(PLUS, ret); | |
| 1808 else | |
| 1809 { | |
| 1810 /* Emit x+ as x(&|), where & means "self". */ | |
| 1811 next = regnode(BRANCH); /* Either */ | |
| 1812 regtail(ret, next); | |
| 233 | 1813 regtail(regnode(BACK), ret); /* loop back */ |
| 7 | 1814 regtail(next, regnode(BRANCH)); /* or */ |
| 1815 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1816 } | |
| 1817 *flagp = (WORST | HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1818 break; | |
| 1819 | |
| 1820 case Magic('@'): | |
| 1821 { | |
| 1822 int lop = END; | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1823 int nr; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1824 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1825 nr = getdecchrs(); |
| 7 | 1826 switch (no_Magic(getchr())) |
| 1827 { | |
| 1828 case '=': lop = MATCH; break; /* \@= */ | |
| 1829 case '!': lop = NOMATCH; break; /* \@! */ | |
| 1830 case '>': lop = SUBPAT; break; /* \@> */ | |
| 1831 case '<': switch (no_Magic(getchr())) | |
| 1832 { | |
| 1833 case '=': lop = BEHIND; break; /* \@<= */ | |
| 1834 case '!': lop = NOBEHIND; break; /* \@<! */ | |
| 1835 } | |
| 1836 } | |
| 1837 if (lop == END) | |
| 4444 | 1838 EMSG2_RET_NULL(_("E59: invalid character after %s@"), |
| 7 | 1839 reg_magic == MAGIC_ALL); |
| 1840 /* Look behind must match with behind_pos. */ | |
| 1841 if (lop == BEHIND || lop == NOBEHIND) | |
| 1842 { | |
| 1843 regtail(ret, regnode(BHPOS)); | |
| 1844 *flagp |= HASLOOKBH; | |
| 1845 } | |
| 1846 regtail(ret, regnode(END)); /* operand ends */ | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1847 if (lop == BEHIND || lop == NOBEHIND) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1848 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1849 if (nr < 0) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1850 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
|
1851 reginsert_nr(lop, nr, ret); |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1852 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1853 else |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1854 reginsert(lop, ret); |
| 7 | 1855 break; |
| 1856 } | |
| 1857 | |
| 1858 case Magic('?'): | |
| 1859 case Magic('='): | |
| 1860 /* Emit x= as (x|) */ | |
| 1861 reginsert(BRANCH, ret); /* Either x */ | |
| 1862 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1863 next = regnode(NOTHING); /* null. */ | |
| 1864 regtail(ret, next); | |
| 1865 regoptail(ret, next); | |
| 1866 break; | |
| 1867 | |
| 1868 case Magic('{'): | |
| 1869 if (!read_limits(&minval, &maxval)) | |
| 1870 return NULL; | |
| 1871 if (flags & SIMPLE) | |
| 1872 { | |
| 1873 reginsert(BRACE_SIMPLE, ret); | |
| 1874 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1875 } | |
| 1876 else | |
| 1877 { | |
| 1878 if (num_complex_braces >= 10) | |
| 4444 | 1879 EMSG2_RET_NULL(_("E60: Too many complex %s{...}s"), |
| 7 | 1880 reg_magic == MAGIC_ALL); |
| 1881 reginsert(BRACE_COMPLEX + num_complex_braces, ret); | |
| 1882 regoptail(ret, regnode(BACK)); | |
| 1883 regoptail(ret, ret); | |
| 1884 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1885 ++num_complex_braces; | |
| 1886 } | |
| 1887 if (minval > 0 && maxval > 0) | |
| 1888 *flagp = (HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1889 break; | |
| 1890 } | |
| 1891 if (re_multi_type(peekchr()) != NOT_MULTI) | |
| 1892 { | |
| 1893 /* Can't have a multi follow a multi. */ | |
| 1894 if (peekchr() == Magic('*')) | |
| 1895 sprintf((char *)IObuff, _("E61: Nested %s*"), | |
| 1896 reg_magic >= MAGIC_ON ? "" : "\\"); | |
| 1897 else | |
| 1898 sprintf((char *)IObuff, _("E62: Nested %s%c"), | |
| 1899 reg_magic == MAGIC_ALL ? "" : "\\", no_Magic(peekchr())); | |
| 1900 EMSG_RET_NULL(IObuff); | |
| 1901 } | |
| 1902 | |
| 1903 return ret; | |
| 1904 } | |
| 1905 | |
| 4444 | 1906 /* When making changes to classchars also change nfa_classcodes. */ |
| 1907 static char_u *classchars = (char_u *)".iIkKfFpPsSdDxXoOwWhHaAlLuU"; | |
| 1908 static int classcodes[] = { | |
| 1909 ANY, IDENT, SIDENT, KWORD, SKWORD, | |
| 1910 FNAME, SFNAME, PRINT, SPRINT, | |
| 1911 WHITE, NWHITE, DIGIT, NDIGIT, | |
| 1912 HEX, NHEX, OCTAL, NOCTAL, | |
| 1913 WORD, NWORD, HEAD, NHEAD, | |
| 1914 ALPHA, NALPHA, LOWER, NLOWER, | |
| 1915 UPPER, NUPPER | |
| 1916 }; | |
| 1917 | |
| 7 | 1918 /* |
| 4444 | 1919 * Parse the lowest level. |
| 7 | 1920 * |
| 1921 * Optimization: gobbles an entire sequence of ordinary characters so that | |
| 1922 * it can turn them into a single node, which is smaller to store and | |
| 1923 * faster to run. Don't do this when one_exactly is set. | |
| 1924 */ | |
| 1925 static char_u * | |
| 1926 regatom(flagp) | |
| 1927 int *flagp; | |
| 1928 { | |
| 1929 char_u *ret; | |
| 1930 int flags; | |
| 1931 int c; | |
| 1932 char_u *p; | |
| 1933 int extra = 0; | |
| 1934 | |
| 1935 *flagp = WORST; /* Tentatively. */ | |
| 1936 | |
| 1937 c = getchr(); | |
| 1938 switch (c) | |
| 1939 { | |
| 1940 case Magic('^'): | |
| 1941 ret = regnode(BOL); | |
| 1942 break; | |
| 1943 | |
| 1944 case Magic('$'): | |
| 1945 ret = regnode(EOL); | |
| 1946 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1947 had_eol = TRUE; | |
| 1948 #endif | |
| 1949 break; | |
| 1950 | |
| 1951 case Magic('<'): | |
| 1952 ret = regnode(BOW); | |
| 1953 break; | |
| 1954 | |
| 1955 case Magic('>'): | |
| 1956 ret = regnode(EOW); | |
| 1957 break; | |
| 1958 | |
| 1959 case Magic('_'): | |
| 1960 c = no_Magic(getchr()); | |
| 1961 if (c == '^') /* "\_^" is start-of-line */ | |
| 1962 { | |
| 1963 ret = regnode(BOL); | |
| 1964 break; | |
| 1965 } | |
| 1966 if (c == '$') /* "\_$" is end-of-line */ | |
| 1967 { | |
| 1968 ret = regnode(EOL); | |
| 1969 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1970 had_eol = TRUE; | |
| 1971 #endif | |
| 1972 break; | |
| 1973 } | |
| 1974 | |
| 1975 extra = ADD_NL; | |
| 1976 *flagp |= HASNL; | |
| 1977 | |
| 1978 /* "\_[" is character range plus newline */ | |
| 1979 if (c == '[') | |
| 1980 goto collection; | |
| 1981 | |
| 1982 /* "\_x" is character class plus newline */ | |
| 1983 /*FALLTHROUGH*/ | |
| 1984 | |
| 1985 /* | |
| 1986 * Character classes. | |
| 1987 */ | |
| 1988 case Magic('.'): | |
| 1989 case Magic('i'): | |
| 1990 case Magic('I'): | |
| 1991 case Magic('k'): | |
| 1992 case Magic('K'): | |
| 1993 case Magic('f'): | |
| 1994 case Magic('F'): | |
| 1995 case Magic('p'): | |
| 1996 case Magic('P'): | |
| 1997 case Magic('s'): | |
| 1998 case Magic('S'): | |
| 1999 case Magic('d'): | |
| 2000 case Magic('D'): | |
| 2001 case Magic('x'): | |
| 2002 case Magic('X'): | |
| 2003 case Magic('o'): | |
| 2004 case Magic('O'): | |
| 2005 case Magic('w'): | |
| 2006 case Magic('W'): | |
| 2007 case Magic('h'): | |
| 2008 case Magic('H'): | |
| 2009 case Magic('a'): | |
| 2010 case Magic('A'): | |
| 2011 case Magic('l'): | |
| 2012 case Magic('L'): | |
| 2013 case Magic('u'): | |
| 2014 case Magic('U'): | |
| 2015 p = vim_strchr(classchars, no_Magic(c)); | |
| 2016 if (p == NULL) | |
| 2017 EMSG_RET_NULL(_("E63: invalid use of \\_")); | |
| 714 | 2018 #ifdef FEAT_MBYTE |
| 2019 /* When '.' is followed by a composing char ignore the dot, so that | |
| 2020 * the composing char is matched here. */ | |
| 2021 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
| 2022 { | |
| 2023 c = getchr(); | |
| 2024 goto do_multibyte; | |
| 2025 } | |
| 2026 #endif | |
| 7 | 2027 ret = regnode(classcodes[p - classchars] + extra); |
| 2028 *flagp |= HASWIDTH | SIMPLE; | |
| 2029 break; | |
| 2030 | |
| 2031 case Magic('n'): | |
| 2032 if (reg_string) | |
| 2033 { | |
| 2034 /* In a string "\n" matches a newline character. */ | |
| 2035 ret = regnode(EXACTLY); | |
| 2036 regc(NL); | |
| 2037 regc(NUL); | |
| 2038 *flagp |= HASWIDTH | SIMPLE; | |
| 2039 } | |
| 2040 else | |
| 2041 { | |
| 2042 /* In buffer text "\n" matches the end of a line. */ | |
| 2043 ret = regnode(NEWL); | |
| 2044 *flagp |= HASWIDTH | HASNL; | |
| 2045 } | |
| 2046 break; | |
| 2047 | |
| 2048 case Magic('('): | |
| 2049 if (one_exactly) | |
| 2050 EMSG_ONE_RET_NULL; | |
| 2051 ret = reg(REG_PAREN, &flags); | |
| 2052 if (ret == NULL) | |
| 2053 return NULL; | |
| 2054 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 2055 break; | |
| 2056 | |
| 2057 case NUL: | |
| 2058 case Magic('|'): | |
| 2059 case Magic('&'): | |
| 2060 case Magic(')'): | |
| 1468 | 2061 if (one_exactly) |
| 2062 EMSG_ONE_RET_NULL; | |
| 7 | 2063 EMSG_RET_NULL(_(e_internal)); /* Supposed to be caught earlier. */ |
| 2064 /* NOTREACHED */ | |
| 2065 | |
| 2066 case Magic('='): | |
| 2067 case Magic('?'): | |
| 2068 case Magic('+'): | |
| 2069 case Magic('@'): | |
| 2070 case Magic('{'): | |
| 2071 case Magic('*'): | |
| 2072 c = no_Magic(c); | |
| 2073 sprintf((char *)IObuff, _("E64: %s%c follows nothing"), | |
| 2074 (c == '*' ? reg_magic >= MAGIC_ON : reg_magic == MAGIC_ALL) | |
| 2075 ? "" : "\\", c); | |
| 2076 EMSG_RET_NULL(IObuff); | |
| 2077 /* NOTREACHED */ | |
| 2078 | |
| 2079 case Magic('~'): /* previous substitute pattern */ | |
| 359 | 2080 if (reg_prev_sub != NULL) |
| 7 | 2081 { |
| 2082 char_u *lp; | |
| 2083 | |
| 2084 ret = regnode(EXACTLY); | |
| 2085 lp = reg_prev_sub; | |
| 2086 while (*lp != NUL) | |
| 2087 regc(*lp++); | |
| 2088 regc(NUL); | |
| 2089 if (*reg_prev_sub != NUL) | |
| 2090 { | |
| 2091 *flagp |= HASWIDTH; | |
| 2092 if ((lp - reg_prev_sub) == 1) | |
| 2093 *flagp |= SIMPLE; | |
| 2094 } | |
| 2095 } | |
| 2096 else | |
| 2097 EMSG_RET_NULL(_(e_nopresub)); | |
| 2098 break; | |
| 2099 | |
| 2100 case Magic('1'): | |
| 2101 case Magic('2'): | |
| 2102 case Magic('3'): | |
| 2103 case Magic('4'): | |
| 2104 case Magic('5'): | |
| 2105 case Magic('6'): | |
| 2106 case Magic('7'): | |
| 2107 case Magic('8'): | |
| 2108 case Magic('9'): | |
| 2109 { | |
| 2110 int refnum; | |
| 2111 | |
| 2112 refnum = c - Magic('0'); | |
| 2113 /* | |
| 2114 * Check if the back reference is legal. We must have seen the | |
| 2115 * close brace. | |
| 2116 * TODO: Should also check that we don't refer to something | |
| 2117 * that is repeated (+*=): what instance of the repetition | |
| 2118 * should we match? | |
| 2119 */ | |
| 2120 if (!had_endbrace[refnum]) | |
| 2121 { | |
| 2122 /* Trick: check if "@<=" or "@<!" follows, in which case | |
| 2123 * the \1 can appear before the referenced match. */ | |
| 2124 for (p = regparse; *p != NUL; ++p) | |
| 2125 if (p[0] == '@' && p[1] == '<' | |
| 2126 && (p[2] == '!' || p[2] == '=')) | |
| 2127 break; | |
| 2128 if (*p == NUL) | |
| 2129 EMSG_RET_NULL(_("E65: Illegal back reference")); | |
| 2130 } | |
| 2131 ret = regnode(BACKREF + refnum); | |
| 2132 } | |
| 2133 break; | |
| 2134 | |
| 2135 case Magic('z'): | |
| 2136 { | |
| 2137 c = no_Magic(getchr()); | |
| 2138 switch (c) | |
| 2139 { | |
| 741 | 2140 #ifdef FEAT_SYN_HL |
| 7 | 2141 case '(': if (reg_do_extmatch != REX_SET) |
|
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
2142 EMSG_RET_NULL(_(e_z_not_allowed)); |
| 7 | 2143 if (one_exactly) |
| 2144 EMSG_ONE_RET_NULL; | |
| 2145 ret = reg(REG_ZPAREN, &flags); | |
| 2146 if (ret == NULL) | |
| 2147 return NULL; | |
| 2148 *flagp |= flags & (HASWIDTH|SPSTART|HASNL|HASLOOKBH); | |
| 2149 re_has_z = REX_SET; | |
| 2150 break; | |
| 2151 | |
| 2152 case '1': | |
| 2153 case '2': | |
| 2154 case '3': | |
| 2155 case '4': | |
| 2156 case '5': | |
| 2157 case '6': | |
| 2158 case '7': | |
| 2159 case '8': | |
| 2160 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
|
2161 EMSG_RET_NULL(_(e_z1_not_allowed)); |
| 7 | 2162 ret = regnode(ZREF + c - '0'); |
| 2163 re_has_z = REX_USE; | |
| 2164 break; | |
| 741 | 2165 #endif |
| 7 | 2166 |
| 2167 case 's': ret = regnode(MOPEN + 0); | |
| 2168 break; | |
| 2169 | |
| 2170 case 'e': ret = regnode(MCLOSE + 0); | |
| 2171 break; | |
| 2172 | |
| 2173 default: EMSG_RET_NULL(_("E68: Invalid character after \\z")); | |
| 2174 } | |
| 2175 } | |
| 2176 break; | |
| 2177 | |
| 2178 case Magic('%'): | |
| 2179 { | |
| 2180 c = no_Magic(getchr()); | |
| 2181 switch (c) | |
| 2182 { | |
| 2183 /* () without a back reference */ | |
| 2184 case '(': | |
| 2185 if (one_exactly) | |
| 2186 EMSG_ONE_RET_NULL; | |
| 2187 ret = reg(REG_NPAREN, &flags); | |
| 2188 if (ret == NULL) | |
| 2189 return NULL; | |
| 2190 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 2191 break; | |
| 2192 | |
| 2193 /* Catch \%^ and \%$ regardless of where they appear in the | |
| 2194 * pattern -- regardless of whether or not it makes sense. */ | |
| 2195 case '^': | |
| 2196 ret = regnode(RE_BOF); | |
| 2197 break; | |
| 2198 | |
| 2199 case '$': | |
| 2200 ret = regnode(RE_EOF); | |
| 2201 break; | |
| 2202 | |
| 2203 case '#': | |
| 2204 ret = regnode(CURSOR); | |
| 2205 break; | |
| 2206 | |
| 639 | 2207 case 'V': |
| 2208 ret = regnode(RE_VISUAL); | |
| 2209 break; | |
| 2210 | |
| 7 | 2211 /* \%[abc]: Emit as a list of branches, all ending at the last |
| 2212 * branch which matches nothing. */ | |
| 2213 case '[': | |
| 2214 if (one_exactly) /* doesn't nest */ | |
| 2215 EMSG_ONE_RET_NULL; | |
| 2216 { | |
| 2217 char_u *lastbranch; | |
| 2218 char_u *lastnode = NULL; | |
| 2219 char_u *br; | |
| 2220 | |
| 2221 ret = NULL; | |
| 2222 while ((c = getchr()) != ']') | |
| 2223 { | |
| 2224 if (c == NUL) | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2225 EMSG2_RET_NULL(_(e_missing_sb), |
| 7 | 2226 reg_magic == MAGIC_ALL); |
| 2227 br = regnode(BRANCH); | |
| 2228 if (ret == NULL) | |
| 2229 ret = br; | |
| 2230 else | |
| 2231 regtail(lastnode, br); | |
| 2232 | |
| 2233 ungetchr(); | |
| 2234 one_exactly = TRUE; | |
| 2235 lastnode = regatom(flagp); | |
| 2236 one_exactly = FALSE; | |
| 2237 if (lastnode == NULL) | |
| 2238 return NULL; | |
| 2239 } | |
| 2240 if (ret == NULL) | |
|
4760
532a9855bd30
updated for version 7.3.1127
Bram Moolenaar <bram@vim.org>
parents:
4746
diff
changeset
|
2241 EMSG2_RET_NULL(_(e_empty_sb), |
| 7 | 2242 reg_magic == MAGIC_ALL); |
| 2243 lastbranch = regnode(BRANCH); | |
| 2244 br = regnode(NOTHING); | |
| 2245 if (ret != JUST_CALC_SIZE) | |
| 2246 { | |
| 2247 regtail(lastnode, br); | |
| 2248 regtail(lastbranch, br); | |
| 2249 /* connect all branches to the NOTHING | |
| 2250 * branch at the end */ | |
| 2251 for (br = ret; br != lastnode; ) | |
| 2252 { | |
| 2253 if (OP(br) == BRANCH) | |
| 2254 { | |
| 2255 regtail(br, lastbranch); | |
| 2256 br = OPERAND(br); | |
| 2257 } | |
| 2258 else | |
| 2259 br = regnext(br); | |
| 2260 } | |
| 2261 } | |
| 1701 | 2262 *flagp &= ~(HASWIDTH | SIMPLE); |
| 7 | 2263 break; |
| 2264 } | |
| 2265 | |
| 24 | 2266 case 'd': /* %d123 decimal */ |
| 2267 case 'o': /* %o123 octal */ | |
| 2268 case 'x': /* %xab hex 2 */ | |
| 2269 case 'u': /* %uabcd hex 4 */ | |
| 2270 case 'U': /* %U1234abcd hex 8 */ | |
| 2271 { | |
| 2272 int i; | |
| 2273 | |
| 2274 switch (c) | |
| 2275 { | |
| 2276 case 'd': i = getdecchrs(); break; | |
| 2277 case 'o': i = getoctchrs(); break; | |
| 2278 case 'x': i = gethexchrs(2); break; | |
| 2279 case 'u': i = gethexchrs(4); break; | |
| 2280 case 'U': i = gethexchrs(8); break; | |
| 2281 default: i = -1; break; | |
| 2282 } | |
| 2283 | |
| 2284 if (i < 0) | |
| 4444 | 2285 EMSG2_RET_NULL( |
| 24 | 2286 _("E678: Invalid character after %s%%[dxouU]"), |
| 2287 reg_magic == MAGIC_ALL); | |
| 714 | 2288 #ifdef FEAT_MBYTE |
| 2289 if (use_multibytecode(i)) | |
| 2290 ret = regnode(MULTIBYTECODE); | |
| 2291 else | |
| 2292 #endif | |
| 2293 ret = regnode(EXACTLY); | |
| 24 | 2294 if (i == 0) |
| 2295 regc(0x0a); | |
| 2296 else | |
| 2297 #ifdef FEAT_MBYTE | |
| 2298 regmbc(i); | |
| 2299 #else | |
| 2300 regc(i); | |
| 2301 #endif | |
| 2302 regc(NUL); | |
| 2303 *flagp |= HASWIDTH; | |
| 2304 break; | |
| 2305 } | |
| 2306 | |
| 7 | 2307 default: |
| 639 | 2308 if (VIM_ISDIGIT(c) || c == '<' || c == '>' |
| 2309 || c == '\'') | |
| 7 | 2310 { |
| 2311 long_u n = 0; | |
| 2312 int cmp; | |
| 2313 | |
| 2314 cmp = c; | |
| 2315 if (cmp == '<' || cmp == '>') | |
| 2316 c = getchr(); | |
| 2317 while (VIM_ISDIGIT(c)) | |
| 2318 { | |
| 2319 n = n * 10 + (c - '0'); | |
| 2320 c = getchr(); | |
| 2321 } | |
| 639 | 2322 if (c == '\'' && n == 0) |
| 2323 { | |
| 2324 /* "\%'m", "\%<'m" and "\%>'m": Mark */ | |
| 2325 c = getchr(); | |
| 2326 ret = regnode(RE_MARK); | |
| 2327 if (ret == JUST_CALC_SIZE) | |
| 2328 regsize += 2; | |
| 2329 else | |
| 2330 { | |
| 2331 *regcode++ = c; | |
| 2332 *regcode++ = cmp; | |
| 2333 } | |
| 2334 break; | |
| 2335 } | |
| 2336 else if (c == 'l' || c == 'c' || c == 'v') | |
| 7 | 2337 { |
| 2338 if (c == 'l') | |
| 2339 ret = regnode(RE_LNUM); | |
| 2340 else if (c == 'c') | |
| 2341 ret = regnode(RE_COL); | |
| 2342 else | |
| 2343 ret = regnode(RE_VCOL); | |
| 2344 if (ret == JUST_CALC_SIZE) | |
| 2345 regsize += 5; | |
| 2346 else | |
| 2347 { | |
| 2348 /* put the number and the optional | |
| 2349 * comparator after the opcode */ | |
| 2350 regcode = re_put_long(regcode, n); | |
| 2351 *regcode++ = cmp; | |
| 2352 } | |
| 2353 break; | |
| 2354 } | |
| 2355 } | |
| 2356 | |
| 4444 | 2357 EMSG2_RET_NULL(_("E71: Invalid character after %s%%"), |
| 7 | 2358 reg_magic == MAGIC_ALL); |
| 2359 } | |
| 2360 } | |
| 2361 break; | |
| 2362 | |
| 2363 case Magic('['): | |
| 2364 collection: | |
| 2365 { | |
| 2366 char_u *lp; | |
| 2367 | |
| 2368 /* | |
| 2369 * If there is no matching ']', we assume the '[' is a normal | |
| 2370 * character. This makes 'incsearch' and ":help [" work. | |
| 2371 */ | |
| 2372 lp = skip_anyof(regparse); | |
| 2373 if (*lp == ']') /* there is a matching ']' */ | |
| 2374 { | |
| 2375 int startc = -1; /* > 0 when next '-' is a range */ | |
| 2376 int endc; | |
| 2377 | |
| 2378 /* | |
| 2379 * In a character class, different parsing rules apply. | |
| 2380 * Not even \ is special anymore, nothing is. | |
| 2381 */ | |
| 2382 if (*regparse == '^') /* Complement of range. */ | |
| 2383 { | |
| 2384 ret = regnode(ANYBUT + extra); | |
| 2385 regparse++; | |
| 2386 } | |
| 2387 else | |
| 2388 ret = regnode(ANYOF + extra); | |
| 2389 | |
| 2390 /* At the start ']' and '-' mean the literal character. */ | |
| 2391 if (*regparse == ']' || *regparse == '-') | |
| 167 | 2392 { |
| 2393 startc = *regparse; | |
| 7 | 2394 regc(*regparse++); |
| 167 | 2395 } |
| 7 | 2396 |
| 2397 while (*regparse != NUL && *regparse != ']') | |
| 2398 { | |
| 2399 if (*regparse == '-') | |
| 2400 { | |
| 2401 ++regparse; | |
| 2402 /* The '-' is not used for a range at the end and | |
| 2403 * after or before a '\n'. */ | |
| 2404 if (*regparse == ']' || *regparse == NUL | |
| 2405 || startc == -1 | |
| 2406 || (regparse[0] == '\\' && regparse[1] == 'n')) | |
| 2407 { | |
| 2408 regc('-'); | |
| 2409 startc = '-'; /* [--x] is a range */ | |
| 2410 } | |
| 2411 else | |
| 2412 { | |
| 167 | 2413 /* Also accept "a-[.z.]" */ |
| 2414 endc = 0; | |
| 2415 if (*regparse == '[') | |
| 2416 endc = get_coll_element(®parse); | |
| 2417 if (endc == 0) | |
| 2418 { | |
| 7 | 2419 #ifdef FEAT_MBYTE |
| 167 | 2420 if (has_mbyte) |
| 2421 endc = mb_ptr2char_adv(®parse); | |
| 2422 else | |
| 7 | 2423 #endif |
| 167 | 2424 endc = *regparse++; |
| 2425 } | |
| 24 | 2426 |
| 2427 /* Handle \o40, \x20 and \u20AC style sequences */ | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2428 if (endc == '\\' && !reg_cpo_lit && !reg_cpo_bsl) |
| 24 | 2429 endc = coll_get_char(); |
| 2430 | |
| 7 | 2431 if (startc > endc) |
| 2432 EMSG_RET_NULL(_(e_invrange)); | |
| 2433 #ifdef FEAT_MBYTE | |
| 2434 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
| 2435 || (*mb_char2len)(endc) > 1)) | |
| 2436 { | |
| 2437 /* Limit to a range of 256 chars */ | |
| 2438 if (endc > startc + 256) | |
| 2439 EMSG_RET_NULL(_(e_invrange)); | |
| 2440 while (++startc <= endc) | |
| 2441 regmbc(startc); | |
| 2442 } | |
| 2443 else | |
| 2444 #endif | |
| 2445 { | |
| 2446 #ifdef EBCDIC | |
| 2447 int alpha_only = FALSE; | |
| 2448 | |
| 2449 /* for alphabetical range skip the gaps | |
| 2450 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
| 2451 if (isalpha(startc) && isalpha(endc)) | |
| 2452 alpha_only = TRUE; | |
| 2453 #endif | |
| 2454 while (++startc <= endc) | |
| 2455 #ifdef EBCDIC | |
| 2456 if (!alpha_only || isalpha(startc)) | |
| 2457 #endif | |
| 2458 regc(startc); | |
| 2459 } | |
| 2460 startc = -1; | |
| 2461 } | |
| 2462 } | |
| 2463 /* | |
| 2464 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
| 2465 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
| 2466 * 'cpoptions' is not included. | |
| 167 | 2467 * Posix doesn't recognize backslash at all. |
| 7 | 2468 */ |
| 2469 else if (*regparse == '\\' | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2470 && !reg_cpo_bsl |
| 7 | 2471 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2472 || (!reg_cpo_lit |
| 7 | 2473 && vim_strchr(REGEXP_ABBR, |
| 2474 regparse[1]) != NULL))) | |
| 2475 { | |
| 2476 regparse++; | |
| 2477 if (*regparse == 'n') | |
| 2478 { | |
| 2479 /* '\n' in range: also match NL */ | |
| 2480 if (ret != JUST_CALC_SIZE) | |
| 2481 { | |
| 4084 | 2482 /* Using \n inside [^] does not change what |
| 2483 * matches. "[^\n]" is the same as ".". */ | |
| 2484 if (*ret == ANYOF) | |
| 2485 { | |
| 7 | 2486 *ret = ANYOF + ADD_NL; |
| 4084 | 2487 *flagp |= HASNL; |
| 2488 } | |
| 7 | 2489 /* else: must have had a \n already */ |
| 2490 } | |
| 2491 regparse++; | |
| 2492 startc = -1; | |
| 2493 } | |
| 24 | 2494 else if (*regparse == 'd' |
| 2495 || *regparse == 'o' | |
| 2496 || *regparse == 'x' | |
| 2497 || *regparse == 'u' | |
| 2498 || *regparse == 'U') | |
| 2499 { | |
| 2500 startc = coll_get_char(); | |
| 2501 if (startc == 0) | |
| 2502 regc(0x0a); | |
| 2503 else | |
| 2504 #ifdef FEAT_MBYTE | |
| 2505 regmbc(startc); | |
| 2506 #else | |
| 2507 regc(startc); | |
| 2508 #endif | |
| 2509 } | |
| 7 | 2510 else |
| 2511 { | |
| 2512 startc = backslash_trans(*regparse++); | |
| 2513 regc(startc); | |
| 2514 } | |
| 2515 } | |
| 2516 else if (*regparse == '[') | |
| 2517 { | |
| 2518 int c_class; | |
| 2519 int cu; | |
| 2520 | |
| 167 | 2521 c_class = get_char_class(®parse); |
| 7 | 2522 startc = -1; |
| 2523 /* Characters assumed to be 8 bits! */ | |
| 2524 switch (c_class) | |
| 2525 { | |
| 2526 case CLASS_NONE: | |
| 167 | 2527 c_class = get_equi_class(®parse); |
| 2528 if (c_class != 0) | |
| 2529 { | |
| 2530 /* produce equivalence class */ | |
| 2531 reg_equi_class(c_class); | |
| 2532 } | |
| 2533 else if ((c_class = | |
| 2534 get_coll_element(®parse)) != 0) | |
| 2535 { | |
| 2536 /* produce a collating element */ | |
| 2537 regmbc(c_class); | |
| 2538 } | |
| 2539 else | |
| 2540 { | |
| 2541 /* literal '[', allow [[-x] as a range */ | |
| 2542 startc = *regparse++; | |
| 2543 regc(startc); | |
| 2544 } | |
| 7 | 2545 break; |
| 2546 case CLASS_ALNUM: | |
| 2547 for (cu = 1; cu <= 255; cu++) | |
| 2548 if (isalnum(cu)) | |
| 2549 regc(cu); | |
| 2550 break; | |
| 2551 case CLASS_ALPHA: | |
| 2552 for (cu = 1; cu <= 255; cu++) | |
| 2553 if (isalpha(cu)) | |
| 2554 regc(cu); | |
| 2555 break; | |
| 2556 case CLASS_BLANK: | |
| 2557 regc(' '); | |
| 2558 regc('\t'); | |
| 2559 break; | |
| 2560 case CLASS_CNTRL: | |
| 2561 for (cu = 1; cu <= 255; cu++) | |
| 2562 if (iscntrl(cu)) | |
| 2563 regc(cu); | |
| 2564 break; | |
| 2565 case CLASS_DIGIT: | |
| 2566 for (cu = 1; cu <= 255; cu++) | |
| 2567 if (VIM_ISDIGIT(cu)) | |
| 2568 regc(cu); | |
| 2569 break; | |
| 2570 case CLASS_GRAPH: | |
| 2571 for (cu = 1; cu <= 255; cu++) | |
| 2572 if (isgraph(cu)) | |
| 2573 regc(cu); | |
| 2574 break; | |
| 2575 case CLASS_LOWER: | |
| 2576 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2577 if (MB_ISLOWER(cu)) |
| 7 | 2578 regc(cu); |
| 2579 break; | |
| 2580 case CLASS_PRINT: | |
| 2581 for (cu = 1; cu <= 255; cu++) | |
| 2582 if (vim_isprintc(cu)) | |
| 2583 regc(cu); | |
| 2584 break; | |
| 2585 case CLASS_PUNCT: | |
| 2586 for (cu = 1; cu <= 255; cu++) | |
| 2587 if (ispunct(cu)) | |
| 2588 regc(cu); | |
| 2589 break; | |
| 2590 case CLASS_SPACE: | |
| 2591 for (cu = 9; cu <= 13; cu++) | |
| 2592 regc(cu); | |
| 2593 regc(' '); | |
| 2594 break; | |
| 2595 case CLASS_UPPER: | |
| 2596 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2597 if (MB_ISUPPER(cu)) |
| 7 | 2598 regc(cu); |
| 2599 break; | |
| 2600 case CLASS_XDIGIT: | |
| 2601 for (cu = 1; cu <= 255; cu++) | |
| 2602 if (vim_isxdigit(cu)) | |
| 2603 regc(cu); | |
| 2604 break; | |
| 2605 case CLASS_TAB: | |
| 2606 regc('\t'); | |
| 2607 break; | |
| 2608 case CLASS_RETURN: | |
| 2609 regc('\r'); | |
| 2610 break; | |
| 2611 case CLASS_BACKSPACE: | |
| 2612 regc('\b'); | |
| 2613 break; | |
| 2614 case CLASS_ESCAPE: | |
| 2615 regc('\033'); | |
| 2616 break; | |
| 2617 } | |
| 2618 } | |
| 2619 else | |
| 2620 { | |
| 2621 #ifdef FEAT_MBYTE | |
| 2622 if (has_mbyte) | |
| 2623 { | |
| 2624 int len; | |
| 2625 | |
| 2626 /* produce a multibyte character, including any | |
| 2627 * following composing characters */ | |
| 2628 startc = mb_ptr2char(regparse); | |
| 474 | 2629 len = (*mb_ptr2len)(regparse); |
| 7 | 2630 if (enc_utf8 && utf_char2len(startc) != len) |
| 2631 startc = -1; /* composing chars */ | |
| 2632 while (--len >= 0) | |
| 2633 regc(*regparse++); | |
| 2634 } | |
| 2635 else | |
| 2636 #endif | |
| 2637 { | |
| 2638 startc = *regparse++; | |
| 2639 regc(startc); | |
| 2640 } | |
| 2641 } | |
| 2642 } | |
| 2643 regc(NUL); | |
| 2644 prevchr_len = 1; /* last char was the ']' */ | |
| 2645 if (*regparse != ']') | |
| 2646 EMSG_RET_NULL(_(e_toomsbra)); /* Cannot happen? */ | |
| 2647 skipchr(); /* let's be friends with the lexer again */ | |
| 2648 *flagp |= HASWIDTH | SIMPLE; | |
| 2649 break; | |
| 2650 } | |
| 481 | 2651 else if (reg_strict) |
| 4444 | 2652 EMSG2_RET_NULL(_(e_missingbracket), reg_magic > MAGIC_OFF); |
| 7 | 2653 } |
| 2654 /* FALLTHROUGH */ | |
| 2655 | |
| 2656 default: | |
| 2657 { | |
| 2658 int len; | |
| 2659 | |
| 2660 #ifdef FEAT_MBYTE | |
| 2661 /* A multi-byte character is handled as a separate atom if it's | |
| 714 | 2662 * before a multi and when it's a composing char. */ |
| 2663 if (use_multibytecode(c)) | |
| 7 | 2664 { |
| 714 | 2665 do_multibyte: |
| 7 | 2666 ret = regnode(MULTIBYTECODE); |
| 2667 regmbc(c); | |
| 2668 *flagp |= HASWIDTH | SIMPLE; | |
| 2669 break; | |
| 2670 } | |
| 2671 #endif | |
| 2672 | |
| 2673 ret = regnode(EXACTLY); | |
| 2674 | |
| 2675 /* | |
| 2676 * Append characters as long as: | |
| 2677 * - there is no following multi, we then need the character in | |
| 2678 * front of it as a single character operand | |
| 2679 * - not running into a Magic character | |
| 2680 * - "one_exactly" is not set | |
| 2681 * But always emit at least one character. Might be a Multi, | |
| 2682 * e.g., a "[" without matching "]". | |
| 2683 */ | |
| 2684 for (len = 0; c != NUL && (len == 0 | |
| 2685 || (re_multi_type(peekchr()) == NOT_MULTI | |
| 2686 && !one_exactly | |
| 2687 && !is_Magic(c))); ++len) | |
| 2688 { | |
| 2689 c = no_Magic(c); | |
| 2690 #ifdef FEAT_MBYTE | |
| 2691 if (has_mbyte) | |
| 2692 { | |
| 2693 regmbc(c); | |
| 2694 if (enc_utf8) | |
| 2695 { | |
| 2696 int l; | |
| 2697 | |
| 714 | 2698 /* Need to get composing character too. */ |
| 7 | 2699 for (;;) |
| 2700 { | |
| 714 | 2701 l = utf_ptr2len(regparse); |
| 2702 if (!UTF_COMPOSINGLIKE(regparse, regparse + l)) | |
| 7 | 2703 break; |
| 714 | 2704 regmbc(utf_ptr2char(regparse)); |
| 2705 skipchr(); | |
| 7 | 2706 } |
| 2707 } | |
| 2708 } | |
| 2709 else | |
| 2710 #endif | |
| 2711 regc(c); | |
| 2712 c = getchr(); | |
| 2713 } | |
| 2714 ungetchr(); | |
| 2715 | |
| 2716 regc(NUL); | |
| 2717 *flagp |= HASWIDTH; | |
| 2718 if (len == 1) | |
| 2719 *flagp |= SIMPLE; | |
| 2720 } | |
| 2721 break; | |
| 2722 } | |
| 2723 | |
| 2724 return ret; | |
| 2725 } | |
| 2726 | |
| 714 | 2727 #ifdef FEAT_MBYTE |
| 2728 /* | |
| 2729 * Return TRUE if MULTIBYTECODE should be used instead of EXACTLY for | |
| 2730 * character "c". | |
| 2731 */ | |
| 2732 static int | |
| 2733 use_multibytecode(c) | |
| 2734 int c; | |
| 2735 { | |
| 2736 return has_mbyte && (*mb_char2len)(c) > 1 | |
| 2737 && (re_multi_type(peekchr()) != NOT_MULTI | |
| 2738 || (enc_utf8 && utf_iscomposing(c))); | |
| 2739 } | |
| 2740 #endif | |
| 2741 | |
| 7 | 2742 /* |
| 4444 | 2743 * Emit a node. |
| 7 | 2744 * Return pointer to generated code. |
| 2745 */ | |
| 2746 static char_u * | |
| 2747 regnode(op) | |
| 2748 int op; | |
| 2749 { | |
| 2750 char_u *ret; | |
| 2751 | |
| 2752 ret = regcode; | |
| 2753 if (ret == JUST_CALC_SIZE) | |
| 2754 regsize += 3; | |
| 2755 else | |
| 2756 { | |
| 2757 *regcode++ = op; | |
| 2758 *regcode++ = NUL; /* Null "next" pointer. */ | |
| 2759 *regcode++ = NUL; | |
| 2760 } | |
| 2761 return ret; | |
| 2762 } | |
| 2763 | |
| 2764 /* | |
| 2765 * Emit (if appropriate) a byte of code | |
| 2766 */ | |
| 2767 static void | |
| 2768 regc(b) | |
| 2769 int b; | |
| 2770 { | |
| 2771 if (regcode == JUST_CALC_SIZE) | |
| 2772 regsize++; | |
| 2773 else | |
| 2774 *regcode++ = b; | |
| 2775 } | |
| 2776 | |
| 2777 #ifdef FEAT_MBYTE | |
| 2778 /* | |
| 2779 * Emit (if appropriate) a multi-byte character of code | |
| 2780 */ | |
| 2781 static void | |
| 2782 regmbc(c) | |
| 2783 int c; | |
| 2784 { | |
| 2974 | 2785 if (!has_mbyte && c > 0xff) |
| 2786 return; | |
| 7 | 2787 if (regcode == JUST_CALC_SIZE) |
| 2788 regsize += (*mb_char2len)(c); | |
| 2789 else | |
| 2790 regcode += (*mb_char2bytes)(c, regcode); | |
| 2791 } | |
| 2792 #endif | |
| 2793 | |
| 2794 /* | |
| 4444 | 2795 * Insert an operator in front of already-emitted operand |
| 7 | 2796 * |
| 2797 * Means relocating the operand. | |
| 2798 */ | |
| 2799 static void | |
| 2800 reginsert(op, opnd) | |
| 2801 int op; | |
| 2802 char_u *opnd; | |
| 2803 { | |
| 2804 char_u *src; | |
| 2805 char_u *dst; | |
| 2806 char_u *place; | |
| 2807 | |
| 2808 if (regcode == JUST_CALC_SIZE) | |
| 2809 { | |
| 2810 regsize += 3; | |
| 2811 return; | |
| 2812 } | |
| 2813 src = regcode; | |
| 2814 regcode += 3; | |
| 2815 dst = regcode; | |
| 2816 while (src > opnd) | |
| 2817 *--dst = *--src; | |
| 2818 | |
| 2819 place = opnd; /* Op node, where operand used to be. */ | |
| 2820 *place++ = op; | |
| 2821 *place++ = NUL; | |
| 2822 *place = NUL; | |
| 2823 } | |
| 2824 | |
| 2825 /* | |
| 4444 | 2826 * 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
|
2827 * Add a number to the operator. |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2828 */ |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2829 static void |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2830 reginsert_nr(op, val, opnd) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2831 int op; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2832 long val; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2833 char_u *opnd; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2834 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2835 char_u *src; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2836 char_u *dst; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2837 char_u *place; |
|
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 if (regcode == JUST_CALC_SIZE) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2840 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2841 regsize += 7; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2842 return; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2843 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2844 src = regcode; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2845 regcode += 7; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2846 dst = regcode; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2847 while (src > opnd) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2848 *--dst = *--src; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2849 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2850 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
|
2851 *place++ = op; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2852 *place++ = NUL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2853 *place++ = NUL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2854 place = re_put_long(place, (long_u)val); |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2855 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2856 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2857 /* |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2858 * Insert an operator in front of already-emitted operand. |
| 7 | 2859 * The operator has the given limit values as operands. Also set next pointer. |
| 2860 * | |
| 2861 * Means relocating the operand. | |
| 2862 */ | |
| 2863 static void | |
| 2864 reginsert_limits(op, minval, maxval, opnd) | |
| 2865 int op; | |
| 2866 long minval; | |
| 2867 long maxval; | |
| 2868 char_u *opnd; | |
| 2869 { | |
| 2870 char_u *src; | |
| 2871 char_u *dst; | |
| 2872 char_u *place; | |
| 2873 | |
| 2874 if (regcode == JUST_CALC_SIZE) | |
| 2875 { | |
| 2876 regsize += 11; | |
| 2877 return; | |
| 2878 } | |
| 2879 src = regcode; | |
| 2880 regcode += 11; | |
| 2881 dst = regcode; | |
| 2882 while (src > opnd) | |
| 2883 *--dst = *--src; | |
| 2884 | |
| 2885 place = opnd; /* Op node, where operand used to be. */ | |
| 2886 *place++ = op; | |
| 2887 *place++ = NUL; | |
| 2888 *place++ = NUL; | |
| 2889 place = re_put_long(place, (long_u)minval); | |
| 2890 place = re_put_long(place, (long_u)maxval); | |
| 2891 regtail(opnd, place); | |
| 2892 } | |
| 2893 | |
| 2894 /* | |
| 2895 * Write a long as four bytes at "p" and return pointer to the next char. | |
| 2896 */ | |
| 2897 static char_u * | |
| 2898 re_put_long(p, val) | |
| 2899 char_u *p; | |
| 2900 long_u val; | |
| 2901 { | |
| 2902 *p++ = (char_u) ((val >> 24) & 0377); | |
| 2903 *p++ = (char_u) ((val >> 16) & 0377); | |
| 2904 *p++ = (char_u) ((val >> 8) & 0377); | |
| 2905 *p++ = (char_u) (val & 0377); | |
| 2906 return p; | |
| 2907 } | |
| 2908 | |
| 2909 /* | |
| 4444 | 2910 * Set the next-pointer at the end of a node chain. |
| 7 | 2911 */ |
| 2912 static void | |
| 2913 regtail(p, val) | |
| 2914 char_u *p; | |
| 2915 char_u *val; | |
| 2916 { | |
| 2917 char_u *scan; | |
| 2918 char_u *temp; | |
| 2919 int offset; | |
| 2920 | |
| 2921 if (p == JUST_CALC_SIZE) | |
| 2922 return; | |
| 2923 | |
| 2924 /* Find last node. */ | |
| 2925 scan = p; | |
| 2926 for (;;) | |
| 2927 { | |
| 2928 temp = regnext(scan); | |
| 2929 if (temp == NULL) | |
| 2930 break; | |
| 2931 scan = temp; | |
| 2932 } | |
| 2933 | |
| 233 | 2934 if (OP(scan) == BACK) |
| 7 | 2935 offset = (int)(scan - val); |
| 2936 else | |
| 2937 offset = (int)(val - scan); | |
| 2010 | 2938 /* When the offset uses more than 16 bits it can no longer fit in the two |
| 2974 | 2939 * bytes available. Use a global flag to avoid having to check return |
| 2010 | 2940 * values in too many places. */ |
| 2941 if (offset > 0xffff) | |
| 2942 reg_toolong = TRUE; | |
| 2943 else | |
| 2944 { | |
| 2945 *(scan + 1) = (char_u) (((unsigned)offset >> 8) & 0377); | |
| 2946 *(scan + 2) = (char_u) (offset & 0377); | |
| 2947 } | |
| 7 | 2948 } |
| 2949 | |
| 2950 /* | |
| 4444 | 2951 * Like regtail, on item after a BRANCH; nop if none. |
| 7 | 2952 */ |
| 2953 static void | |
| 2954 regoptail(p, val) | |
| 2955 char_u *p; | |
| 2956 char_u *val; | |
| 2957 { | |
| 2958 /* When op is neither BRANCH nor BRACE_COMPLEX0-9, it is "operandless" */ | |
| 2959 if (p == NULL || p == JUST_CALC_SIZE | |
| 2960 || (OP(p) != BRANCH | |
| 2961 && (OP(p) < BRACE_COMPLEX || OP(p) > BRACE_COMPLEX + 9))) | |
| 2962 return; | |
| 2963 regtail(OPERAND(p), val); | |
| 2964 } | |
| 2965 | |
| 2966 /* | |
| 4444 | 2967 * Functions for getting characters from the regexp input. |
| 7 | 2968 */ |
| 2969 | |
| 2970 static int at_start; /* True when on the first character */ | |
| 2971 static int prev_at_start; /* True when on the second character */ | |
| 2972 | |
| 4444 | 2973 /* |
| 2974 * Start parsing at "str". | |
| 2975 */ | |
| 7 | 2976 static void |
| 2977 initchr(str) | |
| 2978 char_u *str; | |
| 2979 { | |
| 2980 regparse = str; | |
| 2981 prevchr_len = 0; | |
| 2982 curchr = prevprevchr = prevchr = nextchr = -1; | |
| 2983 at_start = TRUE; | |
| 2984 prev_at_start = FALSE; | |
| 2985 } | |
| 2986 | |
| 4444 | 2987 /* |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2988 * 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
|
2989 * starts in the same state again. |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2990 */ |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2991 static void |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2992 save_parse_state(ps) |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2993 parse_state_T *ps; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2994 { |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2995 ps->regparse = regparse; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2996 ps->prevchr_len = prevchr_len; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2997 ps->curchr = curchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2998 ps->prevchr = prevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2999 ps->prevprevchr = prevprevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3000 ps->nextchr = nextchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3001 ps->at_start = at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3002 ps->prev_at_start = prev_at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3003 ps->regnpar = regnpar; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3004 } |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3005 |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3006 /* |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3007 * Restore a previously saved parse state. |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3008 */ |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3009 static void |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3010 restore_parse_state(ps) |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3011 parse_state_T *ps; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3012 { |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3013 regparse = ps->regparse; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3014 prevchr_len = ps->prevchr_len; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3015 curchr = ps->curchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3016 prevchr = ps->prevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3017 prevprevchr = ps->prevprevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3018 nextchr = ps->nextchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3019 at_start = ps->at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3020 prev_at_start = ps->prev_at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3021 regnpar = ps->regnpar; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3022 } |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3023 |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3024 |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3025 /* |
| 4444 | 3026 * Get the next character without advancing. |
| 3027 */ | |
| 7 | 3028 static int |
| 3029 peekchr() | |
| 3030 { | |
| 167 | 3031 static int after_slash = FALSE; |
| 3032 | |
| 7 | 3033 if (curchr == -1) |
| 3034 { | |
| 3035 switch (curchr = regparse[0]) | |
| 3036 { | |
| 3037 case '.': | |
| 3038 case '[': | |
| 3039 case '~': | |
| 3040 /* magic when 'magic' is on */ | |
| 3041 if (reg_magic >= MAGIC_ON) | |
| 3042 curchr = Magic(curchr); | |
| 3043 break; | |
| 3044 case '(': | |
| 3045 case ')': | |
| 3046 case '{': | |
| 3047 case '%': | |
| 3048 case '+': | |
| 3049 case '=': | |
| 3050 case '?': | |
| 3051 case '@': | |
| 3052 case '!': | |
| 3053 case '&': | |
| 3054 case '|': | |
| 3055 case '<': | |
| 3056 case '>': | |
| 3057 case '#': /* future ext. */ | |
| 3058 case '"': /* future ext. */ | |
| 3059 case '\'': /* future ext. */ | |
| 3060 case ',': /* future ext. */ | |
| 3061 case '-': /* future ext. */ | |
| 3062 case ':': /* future ext. */ | |
| 3063 case ';': /* future ext. */ | |
| 3064 case '`': /* future ext. */ | |
| 3065 case '/': /* Can't be used in / command */ | |
| 3066 /* magic only after "\v" */ | |
| 3067 if (reg_magic == MAGIC_ALL) | |
| 3068 curchr = Magic(curchr); | |
| 3069 break; | |
| 3070 case '*': | |
| 167 | 3071 /* * is not magic as the very first character, eg "?*ptr", when |
| 3072 * after '^', eg "/^*ptr" and when after "\(", "\|", "\&". But | |
| 3073 * "\(\*" is not magic, thus must be magic if "after_slash" */ | |
| 3074 if (reg_magic >= MAGIC_ON | |
| 3075 && !at_start | |
| 3076 && !(prev_at_start && prevchr == Magic('^')) | |
| 3077 && (after_slash | |
| 3078 || (prevchr != Magic('(') | |
| 3079 && prevchr != Magic('&') | |
| 3080 && prevchr != Magic('|')))) | |
| 7 | 3081 curchr = Magic('*'); |
| 3082 break; | |
| 3083 case '^': | |
| 3084 /* '^' is only magic as the very first character and if it's after | |
| 3085 * "\(", "\|", "\&' or "\n" */ | |
| 3086 if (reg_magic >= MAGIC_OFF | |
| 3087 && (at_start | |
| 3088 || reg_magic == MAGIC_ALL | |
| 3089 || prevchr == Magic('(') | |
| 3090 || prevchr == Magic('|') | |
| 3091 || prevchr == Magic('&') | |
| 3092 || prevchr == Magic('n') | |
| 3093 || (no_Magic(prevchr) == '(' | |
| 3094 && prevprevchr == Magic('%')))) | |
| 3095 { | |
| 3096 curchr = Magic('^'); | |
| 3097 at_start = TRUE; | |
| 3098 prev_at_start = FALSE; | |
| 3099 } | |
| 3100 break; | |
| 3101 case '$': | |
| 3102 /* '$' is only magic as the very last char and if it's in front of | |
| 3103 * either "\|", "\)", "\&", or "\n" */ | |
| 3104 if (reg_magic >= MAGIC_OFF) | |
| 3105 { | |
| 3106 char_u *p = regparse + 1; | |
| 3107 | |
| 3108 /* ignore \c \C \m and \M after '$' */ | |
| 3109 while (p[0] == '\\' && (p[1] == 'c' || p[1] == 'C' | |
| 3110 || p[1] == 'm' || p[1] == 'M' || p[1] == 'Z')) | |
| 3111 p += 2; | |
| 3112 if (p[0] == NUL | |
| 3113 || (p[0] == '\\' | |
| 3114 && (p[1] == '|' || p[1] == '&' || p[1] == ')' | |
| 3115 || p[1] == 'n')) | |
| 3116 || reg_magic == MAGIC_ALL) | |
| 3117 curchr = Magic('$'); | |
| 3118 } | |
| 3119 break; | |
| 3120 case '\\': | |
| 3121 { | |
| 3122 int c = regparse[1]; | |
| 3123 | |
| 3124 if (c == NUL) | |
| 3125 curchr = '\\'; /* trailing '\' */ | |
| 3126 else if ( | |
| 3127 #ifdef EBCDIC | |
| 3128 vim_strchr(META, c) | |
| 3129 #else | |
| 3130 c <= '~' && META_flags[c] | |
| 3131 #endif | |
| 3132 ) | |
| 3133 { | |
| 3134 /* | |
| 3135 * META contains everything that may be magic sometimes, | |
| 3136 * except ^ and $ ("\^" and "\$" are only magic after | |
| 3137 * "\v"). We now fetch the next character and toggle its | |
| 3138 * magicness. Therefore, \ is so meta-magic that it is | |
| 3139 * not in META. | |
| 3140 */ | |
| 3141 curchr = -1; | |
| 3142 prev_at_start = at_start; | |
| 3143 at_start = FALSE; /* be able to say "/\*ptr" */ | |
| 3144 ++regparse; | |
| 167 | 3145 ++after_slash; |
| 7 | 3146 peekchr(); |
| 3147 --regparse; | |
| 167 | 3148 --after_slash; |
| 7 | 3149 curchr = toggle_Magic(curchr); |
| 3150 } | |
| 3151 else if (vim_strchr(REGEXP_ABBR, c)) | |
| 3152 { | |
| 3153 /* | |
| 3154 * Handle abbreviations, like "\t" for TAB -- webb | |
| 3155 */ | |
| 3156 curchr = backslash_trans(c); | |
| 3157 } | |
| 3158 else if (reg_magic == MAGIC_NONE && (c == '$' || c == '^')) | |
| 3159 curchr = toggle_Magic(c); | |
| 3160 else | |
| 3161 { | |
| 3162 /* | |
| 3163 * Next character can never be (made) magic? | |
| 3164 * Then backslashing it won't do anything. | |
| 3165 */ | |
| 3166 #ifdef FEAT_MBYTE | |
| 3167 if (has_mbyte) | |
| 3168 curchr = (*mb_ptr2char)(regparse + 1); | |
| 3169 else | |
| 3170 #endif | |
| 3171 curchr = c; | |
| 3172 } | |
| 3173 break; | |
| 3174 } | |
| 3175 | |
| 3176 #ifdef FEAT_MBYTE | |
| 3177 default: | |
| 3178 if (has_mbyte) | |
| 3179 curchr = (*mb_ptr2char)(regparse); | |
| 3180 #endif | |
| 3181 } | |
| 3182 } | |
| 3183 | |
| 3184 return curchr; | |
| 3185 } | |
| 3186 | |
| 3187 /* | |
| 3188 * Eat one lexed character. Do this in a way that we can undo it. | |
| 3189 */ | |
| 3190 static void | |
| 3191 skipchr() | |
| 3192 { | |
| 3193 /* peekchr() eats a backslash, do the same here */ | |
| 3194 if (*regparse == '\\') | |
| 3195 prevchr_len = 1; | |
| 3196 else | |
| 3197 prevchr_len = 0; | |
| 3198 if (regparse[prevchr_len] != NUL) | |
| 3199 { | |
| 3200 #ifdef FEAT_MBYTE | |
| 714 | 3201 if (enc_utf8) |
| 1449 | 3202 /* exclude composing chars that mb_ptr2len does include */ |
| 3203 prevchr_len += utf_ptr2len(regparse + prevchr_len); | |
| 714 | 3204 else if (has_mbyte) |
| 474 | 3205 prevchr_len += (*mb_ptr2len)(regparse + prevchr_len); |
| 7 | 3206 else |
| 3207 #endif | |
| 3208 ++prevchr_len; | |
| 3209 } | |
| 3210 regparse += prevchr_len; | |
| 3211 prev_at_start = at_start; | |
| 3212 at_start = FALSE; | |
| 3213 prevprevchr = prevchr; | |
| 3214 prevchr = curchr; | |
| 3215 curchr = nextchr; /* use previously unget char, or -1 */ | |
| 3216 nextchr = -1; | |
| 3217 } | |
| 3218 | |
| 3219 /* | |
| 3220 * Skip a character while keeping the value of prev_at_start for at_start. | |
| 3221 * prevchr and prevprevchr are also kept. | |
| 3222 */ | |
| 3223 static void | |
| 3224 skipchr_keepstart() | |
| 3225 { | |
| 3226 int as = prev_at_start; | |
| 3227 int pr = prevchr; | |
| 3228 int prpr = prevprevchr; | |
| 3229 | |
| 3230 skipchr(); | |
| 3231 at_start = as; | |
| 3232 prevchr = pr; | |
| 3233 prevprevchr = prpr; | |
| 3234 } | |
| 3235 | |
| 4444 | 3236 /* |
| 3237 * Get the next character from the pattern. We know about magic and such, so | |
| 3238 * therefore we need a lexical analyzer. | |
| 3239 */ | |
| 7 | 3240 static int |
| 3241 getchr() | |
| 3242 { | |
| 3243 int chr = peekchr(); | |
| 3244 | |
| 3245 skipchr(); | |
| 3246 return chr; | |
| 3247 } | |
| 3248 | |
| 3249 /* | |
| 3250 * put character back. Works only once! | |
| 3251 */ | |
| 3252 static void | |
| 3253 ungetchr() | |
| 3254 { | |
| 3255 nextchr = curchr; | |
| 3256 curchr = prevchr; | |
| 3257 prevchr = prevprevchr; | |
| 3258 at_start = prev_at_start; | |
| 3259 prev_at_start = FALSE; | |
| 3260 | |
| 3261 /* Backup regparse, so that it's at the same position as before the | |
| 3262 * getchr(). */ | |
| 3263 regparse -= prevchr_len; | |
| 3264 } | |
| 3265 | |
| 3266 /* | |
| 29 | 3267 * Get and return the value of the hex string at the current position. |
| 3268 * Return -1 if there is no valid hex number. | |
| 3269 * The position is updated: | |
| 24 | 3270 * blahblah\%x20asdf |
| 856 | 3271 * before-^ ^-after |
| 24 | 3272 * The parameter controls the maximum number of input characters. This will be |
| 3273 * 2 when reading a \%x20 sequence and 4 when reading a \%u20AC sequence. | |
| 3274 */ | |
| 3275 static int | |
| 3276 gethexchrs(maxinputlen) | |
| 3277 int maxinputlen; | |
| 3278 { | |
| 3279 int nr = 0; | |
| 3280 int c; | |
| 3281 int i; | |
| 3282 | |
| 3283 for (i = 0; i < maxinputlen; ++i) | |
| 3284 { | |
| 3285 c = regparse[0]; | |
| 3286 if (!vim_isxdigit(c)) | |
| 3287 break; | |
| 3288 nr <<= 4; | |
| 3289 nr |= hex2nr(c); | |
| 3290 ++regparse; | |
| 3291 } | |
| 3292 | |
| 3293 if (i == 0) | |
| 3294 return -1; | |
| 3295 return nr; | |
| 3296 } | |
| 3297 | |
| 3298 /* | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3299 * Get and return the value of the decimal string immediately after the |
| 24 | 3300 * current position. Return -1 for invalid. Consumes all digits. |
| 3301 */ | |
| 3302 static int | |
| 3303 getdecchrs() | |
| 3304 { | |
| 3305 int nr = 0; | |
| 3306 int c; | |
| 3307 int i; | |
| 3308 | |
| 3309 for (i = 0; ; ++i) | |
| 3310 { | |
| 3311 c = regparse[0]; | |
| 3312 if (c < '0' || c > '9') | |
| 3313 break; | |
| 3314 nr *= 10; | |
| 3315 nr += c - '0'; | |
| 3316 ++regparse; | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3317 curchr = -1; /* no longer valid */ |
| 24 | 3318 } |
| 3319 | |
| 3320 if (i == 0) | |
| 3321 return -1; | |
| 3322 return nr; | |
| 3323 } | |
| 3324 | |
| 3325 /* | |
| 3326 * get and return the value of the octal string immediately after the current | |
| 3327 * position. Return -1 for invalid, or 0-255 for valid. Smart enough to handle | |
| 3328 * numbers > 377 correctly (for example, 400 is treated as 40) and doesn't | |
| 3329 * treat 8 or 9 as recognised characters. Position is updated: | |
| 3330 * blahblah\%o210asdf | |
| 856 | 3331 * before-^ ^-after |
| 24 | 3332 */ |
| 3333 static int | |
| 3334 getoctchrs() | |
| 3335 { | |
| 3336 int nr = 0; | |
| 3337 int c; | |
| 3338 int i; | |
| 3339 | |
| 3340 for (i = 0; i < 3 && nr < 040; ++i) | |
| 3341 { | |
| 3342 c = regparse[0]; | |
| 3343 if (c < '0' || c > '7') | |
| 3344 break; | |
| 3345 nr <<= 3; | |
| 3346 nr |= hex2nr(c); | |
| 3347 ++regparse; | |
| 3348 } | |
| 3349 | |
| 3350 if (i == 0) | |
| 3351 return -1; | |
| 3352 return nr; | |
| 3353 } | |
| 3354 | |
| 3355 /* | |
| 3356 * Get a number after a backslash that is inside []. | |
| 3357 * When nothing is recognized return a backslash. | |
| 3358 */ | |
| 3359 static int | |
| 3360 coll_get_char() | |
| 3361 { | |
| 3362 int nr = -1; | |
| 3363 | |
| 3364 switch (*regparse++) | |
| 3365 { | |
| 3366 case 'd': nr = getdecchrs(); break; | |
| 3367 case 'o': nr = getoctchrs(); break; | |
| 3368 case 'x': nr = gethexchrs(2); break; | |
| 3369 case 'u': nr = gethexchrs(4); break; | |
| 3370 case 'U': nr = gethexchrs(8); break; | |
| 3371 } | |
| 3372 if (nr < 0) | |
| 3373 { | |
| 3374 /* If getting the number fails be backwards compatible: the character | |
| 3375 * is a backslash. */ | |
| 3376 --regparse; | |
| 3377 nr = '\\'; | |
| 3378 } | |
| 3379 return nr; | |
| 3380 } | |
| 3381 | |
| 3382 /* | |
| 7 | 3383 * read_limits - Read two integers to be taken as a minimum and maximum. |
| 3384 * If the first character is '-', then the range is reversed. | |
| 3385 * Should end with 'end'. If minval is missing, zero is default, if maxval is | |
| 3386 * missing, a very big number is the default. | |
| 3387 */ | |
| 3388 static int | |
| 3389 read_limits(minval, maxval) | |
| 3390 long *minval; | |
| 3391 long *maxval; | |
| 3392 { | |
| 3393 int reverse = FALSE; | |
| 3394 char_u *first_char; | |
| 3395 long tmp; | |
| 3396 | |
| 3397 if (*regparse == '-') | |
| 3398 { | |
| 3399 /* Starts with '-', so reverse the range later */ | |
| 3400 regparse++; | |
| 3401 reverse = TRUE; | |
| 3402 } | |
| 3403 first_char = regparse; | |
| 3404 *minval = getdigits(®parse); | |
| 3405 if (*regparse == ',') /* There is a comma */ | |
| 3406 { | |
| 3407 if (vim_isdigit(*++regparse)) | |
| 3408 *maxval = getdigits(®parse); | |
| 3409 else | |
| 3410 *maxval = MAX_LIMIT; | |
| 3411 } | |
| 3412 else if (VIM_ISDIGIT(*first_char)) | |
| 3413 *maxval = *minval; /* It was \{n} or \{-n} */ | |
| 3414 else | |
| 3415 *maxval = MAX_LIMIT; /* It was \{} or \{-} */ | |
| 3416 if (*regparse == '\\') | |
| 3417 regparse++; /* Allow either \{...} or \{...\} */ | |
| 167 | 3418 if (*regparse != '}') |
| 7 | 3419 { |
| 3420 sprintf((char *)IObuff, _("E554: Syntax error in %s{...}"), | |
| 3421 reg_magic == MAGIC_ALL ? "" : "\\"); | |
| 3422 EMSG_RET_FAIL(IObuff); | |
| 3423 } | |
| 3424 | |
| 3425 /* | |
| 3426 * Reverse the range if there was a '-', or make sure it is in the right | |
| 3427 * order otherwise. | |
| 3428 */ | |
| 3429 if ((!reverse && *minval > *maxval) || (reverse && *minval < *maxval)) | |
| 3430 { | |
| 3431 tmp = *minval; | |
| 3432 *minval = *maxval; | |
| 3433 *maxval = tmp; | |
| 3434 } | |
| 3435 skipchr(); /* let's be friends with the lexer again */ | |
| 3436 return OK; | |
| 3437 } | |
| 3438 | |
| 3439 /* | |
| 3440 * vim_regexec and friends | |
| 3441 */ | |
| 3442 | |
| 3443 /* | |
| 3444 * Global work variables for vim_regexec(). | |
| 3445 */ | |
| 3446 | |
| 3447 /* The current match-position is remembered with these variables: */ | |
| 3448 static linenr_T reglnum; /* line number, relative to first line */ | |
| 3449 static char_u *regline; /* start of current line */ | |
| 3450 static char_u *reginput; /* current input, points into "regline" */ | |
| 3451 | |
| 3452 static int need_clear_subexpr; /* subexpressions still need to be | |
| 3453 * cleared */ | |
| 3454 #ifdef FEAT_SYN_HL | |
| 3455 static int need_clear_zsubexpr = FALSE; /* extmatch subexpressions | |
| 3456 * still need to be cleared */ | |
| 3457 #endif | |
| 3458 | |
| 3459 /* | |
| 3460 * Structure used to save the current input state, when it needs to be | |
| 3461 * restored after trying a match. Used by reg_save() and reg_restore(). | |
| 233 | 3462 * Also stores the length of "backpos". |
| 7 | 3463 */ |
| 3464 typedef struct | |
| 3465 { | |
| 3466 union | |
| 3467 { | |
| 3468 char_u *ptr; /* reginput pointer, for single-line regexp */ | |
| 3469 lpos_T pos; /* reginput pos, for multi-line regexp */ | |
| 3470 } rs_u; | |
| 233 | 3471 int rs_len; |
| 7 | 3472 } regsave_T; |
| 3473 | |
| 3474 /* struct to save start/end pointer/position in for \(\) */ | |
| 3475 typedef struct | |
| 3476 { | |
| 3477 union | |
| 3478 { | |
| 3479 char_u *ptr; | |
| 3480 lpos_T pos; | |
| 3481 } se_u; | |
| 3482 } save_se_T; | |
| 3483 | |
| 1579 | 3484 /* used for BEHIND and NOBEHIND matching */ |
| 3485 typedef struct regbehind_S | |
| 3486 { | |
| 3487 regsave_T save_after; | |
| 3488 regsave_T save_behind; | |
| 1602 | 3489 int save_need_clear_subexpr; |
| 1579 | 3490 save_se_T save_start[NSUBEXP]; |
| 3491 save_se_T save_end[NSUBEXP]; | |
| 3492 } regbehind_T; | |
| 3493 | |
| 7 | 3494 static char_u *reg_getline __ARGS((linenr_T lnum)); |
| 4444 | 3495 static long bt_regexec_both __ARGS((char_u *line, colnr_T col, proftime_T *tm)); |
| 3496 static long regtry __ARGS((bt_regprog_T *prog, colnr_T col)); | |
| 7 | 3497 static void cleanup_subexpr __ARGS((void)); |
| 3498 #ifdef FEAT_SYN_HL | |
| 3499 static void cleanup_zsubexpr __ARGS((void)); | |
| 3500 #endif | |
| 1579 | 3501 static void save_subexpr __ARGS((regbehind_T *bp)); |
| 3502 static void restore_subexpr __ARGS((regbehind_T *bp)); | |
| 7 | 3503 static void reg_nextline __ARGS((void)); |
| 233 | 3504 static void reg_save __ARGS((regsave_T *save, garray_T *gap)); |
| 3505 static void reg_restore __ARGS((regsave_T *save, garray_T *gap)); | |
| 7 | 3506 static int reg_save_equal __ARGS((regsave_T *save)); |
| 3507 static void save_se_multi __ARGS((save_se_T *savep, lpos_T *posp)); | |
| 3508 static void save_se_one __ARGS((save_se_T *savep, char_u **pp)); | |
| 3509 | |
| 3510 /* Save the sub-expressions before attempting a match. */ | |
| 3511 #define save_se(savep, posp, pp) \ | |
| 3512 REG_MULTI ? save_se_multi((savep), (posp)) : save_se_one((savep), (pp)) | |
| 3513 | |
| 3514 /* After a failed match restore the sub-expressions. */ | |
| 3515 #define restore_se(savep, posp, pp) { \ | |
| 3516 if (REG_MULTI) \ | |
| 3517 *(posp) = (savep)->se_u.pos; \ | |
| 3518 else \ | |
| 3519 *(pp) = (savep)->se_u.ptr; } | |
| 3520 | |
| 3521 static int re_num_cmp __ARGS((long_u val, char_u *scan)); | |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
3522 static int match_with_backref __ARGS((linenr_T start_lnum, colnr_T start_col, linenr_T end_lnum, colnr_T end_col, int *bytelen)); |
| 180 | 3523 static int regmatch __ARGS((char_u *prog)); |
| 7 | 3524 static int regrepeat __ARGS((char_u *p, long maxcount)); |
| 3525 | |
| 3526 #ifdef DEBUG | |
| 3527 int regnarrate = 0; | |
| 3528 #endif | |
| 3529 | |
| 3530 /* | |
| 3531 * Internal copy of 'ignorecase'. It is set at each call to vim_regexec(). | |
| 3532 * Normally it gets the value of "rm_ic" or "rmm_ic", but when the pattern | |
| 3533 * contains '\c' or '\C' the value is overruled. | |
| 3534 */ | |
| 3535 static int ireg_ic; | |
| 3536 | |
| 3537 #ifdef FEAT_MBYTE | |
| 3538 /* | |
| 3539 * Similar to ireg_ic, but only for 'combining' characters. Set with \Z flag | |
| 3540 * in the regexp. Defaults to false, always. | |
| 3541 */ | |
| 3542 static int ireg_icombine; | |
| 3543 #endif | |
| 3544 | |
| 3545 /* | |
| 410 | 3546 * Copy of "rmm_maxcol": maximum column to search for a match. Zero when |
| 3547 * there is no maximum. | |
| 3548 */ | |
| 418 | 3549 static colnr_T ireg_maxcol; |
| 410 | 3550 |
| 3551 /* | |
| 7 | 3552 * Sometimes need to save a copy of a line. Since alloc()/free() is very |
| 3553 * slow, we keep one allocated piece of memory and only re-allocate it when | |
| 4444 | 3554 * it's too small. It's freed in bt_regexec_both() when finished. |
| 7 | 3555 */ |
| 1468 | 3556 static char_u *reg_tofree = NULL; |
| 7 | 3557 static unsigned reg_tofreelen; |
| 3558 | |
| 3559 /* | |
| 3560 * These variables are set when executing a regexp to speed up the execution. | |
| 1209 | 3561 * Which ones are set depends on whether a single-line or multi-line match is |
| 7 | 3562 * done: |
| 3563 * single-line multi-line | |
| 3564 * reg_match ®match_T NULL | |
| 3565 * reg_mmatch NULL ®mmatch_T | |
| 3566 * reg_startp reg_match->startp <invalid> | |
| 3567 * reg_endp reg_match->endp <invalid> | |
| 3568 * reg_startpos <invalid> reg_mmatch->startpos | |
| 3569 * reg_endpos <invalid> reg_mmatch->endpos | |
| 3570 * reg_win NULL window in which to search | |
| 4061 | 3571 * reg_buf curbuf buffer in which to search |
| 7 | 3572 * reg_firstlnum <invalid> first line in which to search |
| 3573 * reg_maxline 0 last line nr | |
| 3574 * reg_line_lbr FALSE or TRUE FALSE | |
| 3575 */ | |
| 3576 static regmatch_T *reg_match; | |
| 3577 static regmmatch_T *reg_mmatch; | |
| 3578 static char_u **reg_startp = NULL; | |
| 3579 static char_u **reg_endp = NULL; | |
| 3580 static lpos_T *reg_startpos = NULL; | |
| 3581 static lpos_T *reg_endpos = NULL; | |
| 3582 static win_T *reg_win; | |
| 3583 static buf_T *reg_buf; | |
| 3584 static linenr_T reg_firstlnum; | |
| 3585 static linenr_T reg_maxline; | |
| 3586 static int reg_line_lbr; /* "\n" in string is line break */ | |
| 3587 | |
| 270 | 3588 /* Values for rs_state in regitem_T. */ |
| 3589 typedef enum regstate_E | |
| 3590 { | |
| 3591 RS_NOPEN = 0 /* NOPEN and NCLOSE */ | |
| 3592 , RS_MOPEN /* MOPEN + [0-9] */ | |
| 3593 , RS_MCLOSE /* MCLOSE + [0-9] */ | |
| 3594 #ifdef FEAT_SYN_HL | |
| 3595 , RS_ZOPEN /* ZOPEN + [0-9] */ | |
| 3596 , RS_ZCLOSE /* ZCLOSE + [0-9] */ | |
| 3597 #endif | |
| 3598 , RS_BRANCH /* BRANCH */ | |
| 3599 , RS_BRCPLX_MORE /* BRACE_COMPLEX and trying one more match */ | |
| 3600 , RS_BRCPLX_LONG /* BRACE_COMPLEX and trying longest match */ | |
| 3601 , RS_BRCPLX_SHORT /* BRACE_COMPLEX and trying shortest match */ | |
| 3602 , RS_NOMATCH /* NOMATCH */ | |
| 3603 , RS_BEHIND1 /* BEHIND / NOBEHIND matching rest */ | |
| 3604 , RS_BEHIND2 /* BEHIND / NOBEHIND matching behind part */ | |
| 3605 , RS_STAR_LONG /* STAR/PLUS/BRACE_SIMPLE longest match */ | |
| 3606 , RS_STAR_SHORT /* STAR/PLUS/BRACE_SIMPLE shortest match */ | |
| 3607 } regstate_T; | |
| 3608 | |
| 3609 /* | |
| 3610 * When there are alternatives a regstate_T is put on the regstack to remember | |
| 3611 * what we are doing. | |
| 3612 * Before it may be another type of item, depending on rs_state, to remember | |
| 3613 * more things. | |
| 3614 */ | |
| 3615 typedef struct regitem_S | |
| 3616 { | |
| 3617 regstate_T rs_state; /* what we are doing, one of RS_ above */ | |
| 3618 char_u *rs_scan; /* current node in program */ | |
| 3619 union | |
| 3620 { | |
| 3621 save_se_T sesave; | |
| 3622 regsave_T regsave; | |
| 3623 } rs_un; /* room for saving reginput */ | |
| 1579 | 3624 short rs_no; /* submatch nr or BEHIND/NOBEHIND */ |
| 270 | 3625 } regitem_T; |
| 3626 | |
| 3627 static regitem_T *regstack_push __ARGS((regstate_T state, char_u *scan)); | |
| 3628 static void regstack_pop __ARGS((char_u **scan)); | |
| 3629 | |
| 3630 /* used for STAR, PLUS and BRACE_SIMPLE matching */ | |
| 3631 typedef struct regstar_S | |
| 3632 { | |
| 3633 int nextb; /* next byte */ | |
| 3634 int nextb_ic; /* next byte reverse case */ | |
| 3635 long count; | |
| 3636 long minval; | |
| 3637 long maxval; | |
| 3638 } regstar_T; | |
| 3639 | |
| 3640 /* used to store input position when a BACK was encountered, so that we now if | |
| 3641 * we made any progress since the last time. */ | |
| 3642 typedef struct backpos_S | |
| 3643 { | |
| 3644 char_u *bp_scan; /* "scan" where BACK was encountered */ | |
| 3645 regsave_T bp_pos; /* last input position */ | |
| 3646 } backpos_T; | |
| 3647 | |
| 3648 /* | |
| 1520 | 3649 * "regstack" and "backpos" are used by regmatch(). They are kept over calls |
| 3650 * to avoid invoking malloc() and free() often. | |
| 3651 * "regstack" is a stack with regitem_T items, sometimes preceded by regstar_T | |
| 3652 * or regbehind_T. | |
| 3653 * "backpos_T" is a table with backpos_T for BACK | |
| 270 | 3654 */ |
| 1520 | 3655 static garray_T regstack = {0, 0, 0, 0, NULL}; |
| 3656 static garray_T backpos = {0, 0, 0, 0, NULL}; | |
| 3657 | |
| 3658 /* | |
| 3659 * Both for regstack and backpos tables we use the following strategy of | |
| 3660 * allocation (to reduce malloc/free calls): | |
| 3661 * - Initial size is fairly small. | |
| 3662 * - When needed, the tables are grown bigger (8 times at first, double after | |
| 3663 * that). | |
| 3664 * - After executing the match we free the memory only if the array has grown. | |
| 3665 * Thus the memory is kept allocated when it's at the initial size. | |
| 3666 * This makes it fast while not keeping a lot of memory allocated. | |
| 3667 * A three times speed increase was observed when using many simple patterns. | |
| 3668 */ | |
| 3669 #define REGSTACK_INITIAL 2048 | |
| 3670 #define BACKPOS_INITIAL 64 | |
| 3671 | |
| 3672 #if defined(EXITFREE) || defined(PROTO) | |
| 3673 void | |
| 3674 free_regexp_stuff() | |
| 3675 { | |
| 3676 ga_clear(®stack); | |
| 3677 ga_clear(&backpos); | |
| 3678 vim_free(reg_tofree); | |
| 3679 vim_free(reg_prev_sub); | |
| 3680 } | |
| 3681 #endif | |
| 270 | 3682 |
| 7 | 3683 /* |
| 3684 * Get pointer to the line "lnum", which is relative to "reg_firstlnum". | |
| 3685 */ | |
| 3686 static char_u * | |
| 3687 reg_getline(lnum) | |
| 3688 linenr_T lnum; | |
| 3689 { | |
| 3690 /* when looking behind for a match/no-match lnum is negative. But we | |
| 3691 * can't go before line 1 */ | |
| 3692 if (reg_firstlnum + lnum < 1) | |
| 3693 return NULL; | |
| 482 | 3694 if (lnum > reg_maxline) |
| 481 | 3695 /* Must have matched the "\n" in the last line. */ |
| 3696 return (char_u *)""; | |
| 7 | 3697 return ml_get_buf(reg_buf, reg_firstlnum + lnum, FALSE); |
| 3698 } | |
| 3699 | |
| 3700 static regsave_T behind_pos; | |
| 3701 | |
| 3702 #ifdef FEAT_SYN_HL | |
| 3703 static char_u *reg_startzp[NSUBEXP]; /* Workspace to mark beginning */ | |
| 3704 static char_u *reg_endzp[NSUBEXP]; /* and end of \z(...\) matches */ | |
| 3705 static lpos_T reg_startzpos[NSUBEXP]; /* idem, beginning pos */ | |
| 3706 static lpos_T reg_endzpos[NSUBEXP]; /* idem, end pos */ | |
| 3707 #endif | |
| 3708 | |
| 3709 /* TRUE if using multi-line regexp. */ | |
| 3710 #define REG_MULTI (reg_match == NULL) | |
| 3711 | |
| 4444 | 3712 static int bt_regexec __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); |
| 3713 | |
| 7 | 3714 /* |
| 3715 * Match a regexp against a string. | |
| 3716 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3717 * Uses curbuf for line count and 'iskeyword'. | |
| 3718 * | |
| 3719 * Return TRUE if there is a match, FALSE if not. | |
| 3720 */ | |
| 4444 | 3721 static int |
| 3722 bt_regexec(rmp, line, col) | |
| 7 | 3723 regmatch_T *rmp; |
| 3724 char_u *line; /* string to match against */ | |
| 3725 colnr_T col; /* column to start looking for match */ | |
| 3726 { | |
| 3727 reg_match = rmp; | |
| 3728 reg_mmatch = NULL; | |
| 3729 reg_maxline = 0; | |
| 3730 reg_line_lbr = FALSE; | |
| 4061 | 3731 reg_buf = curbuf; |
| 7 | 3732 reg_win = NULL; |
| 3733 ireg_ic = rmp->rm_ic; | |
| 3734 #ifdef FEAT_MBYTE | |
| 3735 ireg_icombine = FALSE; | |
| 3736 #endif | |
| 410 | 3737 ireg_maxcol = 0; |
| 4444 | 3738 return (bt_regexec_both(line, col, NULL) != 0); |
| 7 | 3739 } |
| 3740 | |
| 39 | 3741 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ |
| 3742 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 4444 | 3743 |
| 3744 static int bt_regexec_nl __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
| 3745 | |
| 7 | 3746 /* |
| 3747 * Like vim_regexec(), but consider a "\n" in "line" to be a line break. | |
| 3748 */ | |
| 4444 | 3749 static int |
| 3750 bt_regexec_nl(rmp, line, col) | |
| 7 | 3751 regmatch_T *rmp; |
| 3752 char_u *line; /* string to match against */ | |
| 3753 colnr_T col; /* column to start looking for match */ | |
| 3754 { | |
| 3755 reg_match = rmp; | |
| 3756 reg_mmatch = NULL; | |
| 3757 reg_maxline = 0; | |
| 3758 reg_line_lbr = TRUE; | |
| 4061 | 3759 reg_buf = curbuf; |
| 7 | 3760 reg_win = NULL; |
| 3761 ireg_ic = rmp->rm_ic; | |
| 3762 #ifdef FEAT_MBYTE | |
| 3763 ireg_icombine = FALSE; | |
| 3764 #endif | |
| 410 | 3765 ireg_maxcol = 0; |
| 4444 | 3766 return (bt_regexec_both(line, col, NULL) != 0); |
| 7 | 3767 } |
| 3768 #endif | |
| 3769 | |
| 4444 | 3770 static long bt_regexec_multi __ARGS((regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm)); |
| 3771 | |
| 7 | 3772 /* |
| 3773 * Match a regexp against multiple lines. | |
| 3774 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3775 * Uses curbuf for line count and 'iskeyword'. | |
| 3776 * | |
| 3777 * Return zero if there is no match. Return number of lines contained in the | |
| 3778 * match otherwise. | |
| 3779 */ | |
| 4444 | 3780 static long |
| 3781 bt_regexec_multi(rmp, win, buf, lnum, col, tm) | |
| 7 | 3782 regmmatch_T *rmp; |
| 3783 win_T *win; /* window in which to search or NULL */ | |
| 3784 buf_T *buf; /* buffer in which to search */ | |
| 3785 linenr_T lnum; /* nr of line to start looking for match */ | |
| 3786 colnr_T col; /* column to start looking for match */ | |
| 1521 | 3787 proftime_T *tm; /* timeout limit or NULL */ |
| 7 | 3788 { |
| 3789 long r; | |
| 3790 | |
| 3791 reg_match = NULL; | |
| 3792 reg_mmatch = rmp; | |
| 3793 reg_buf = buf; | |
| 3794 reg_win = win; | |
| 3795 reg_firstlnum = lnum; | |
| 3796 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
| 3797 reg_line_lbr = FALSE; | |
| 3798 ireg_ic = rmp->rmm_ic; | |
| 3799 #ifdef FEAT_MBYTE | |
| 3800 ireg_icombine = FALSE; | |
| 3801 #endif | |
| 410 | 3802 ireg_maxcol = rmp->rmm_maxcol; |
| 7 | 3803 |
| 4444 | 3804 r = bt_regexec_both(NULL, col, tm); |
| 7 | 3805 |
| 3806 return r; | |
| 3807 } | |
| 3808 | |
| 3809 /* | |
| 3810 * Match a regexp against a string ("line" points to the string) or multiple | |
| 3811 * lines ("line" is NULL, use reg_getline()). | |
| 3812 */ | |
| 3813 static long | |
| 4444 | 3814 bt_regexec_both(line, col, tm) |
| 7 | 3815 char_u *line; |
| 3816 colnr_T col; /* column to start looking for match */ | |
| 1877 | 3817 proftime_T *tm UNUSED; /* timeout limit or NULL */ |
| 7 | 3818 { |
| 4444 | 3819 bt_regprog_T *prog; |
| 7 | 3820 char_u *s; |
| 189 | 3821 long retval = 0L; |
| 7 | 3822 |
| 1520 | 3823 /* Create "regstack" and "backpos" if they are not allocated yet. |
| 3824 * We allocate *_INITIAL amount of bytes first and then set the grow size | |
| 3825 * to much bigger value to avoid many malloc calls in case of deep regular | |
| 3826 * expressions. */ | |
| 3827 if (regstack.ga_data == NULL) | |
| 3828 { | |
| 3829 /* Use an item size of 1 byte, since we push different things | |
| 3830 * onto the regstack. */ | |
| 3831 ga_init2(®stack, 1, REGSTACK_INITIAL); | |
| 3832 ga_grow(®stack, REGSTACK_INITIAL); | |
| 3833 regstack.ga_growsize = REGSTACK_INITIAL * 8; | |
| 3834 } | |
| 3835 | |
| 3836 if (backpos.ga_data == NULL) | |
| 3837 { | |
| 3838 ga_init2(&backpos, sizeof(backpos_T), BACKPOS_INITIAL); | |
| 3839 ga_grow(&backpos, BACKPOS_INITIAL); | |
| 3840 backpos.ga_growsize = BACKPOS_INITIAL * 8; | |
| 3841 } | |
| 270 | 3842 |
| 7 | 3843 if (REG_MULTI) |
| 3844 { | |
| 4444 | 3845 prog = (bt_regprog_T *)reg_mmatch->regprog; |
| 7 | 3846 line = reg_getline((linenr_T)0); |
| 3847 reg_startpos = reg_mmatch->startpos; | |
| 3848 reg_endpos = reg_mmatch->endpos; | |
| 3849 } | |
| 3850 else | |
| 3851 { | |
| 4444 | 3852 prog = (bt_regprog_T *)reg_match->regprog; |
| 7 | 3853 reg_startp = reg_match->startp; |
| 3854 reg_endp = reg_match->endp; | |
| 3855 } | |
| 3856 | |
| 3857 /* Be paranoid... */ | |
| 3858 if (prog == NULL || line == NULL) | |
| 3859 { | |
| 3860 EMSG(_(e_null)); | |
| 3861 goto theend; | |
| 3862 } | |
| 3863 | |
| 3864 /* Check validity of program. */ | |
| 3865 if (prog_magic_wrong()) | |
| 3866 goto theend; | |
| 3867 | |
| 410 | 3868 /* If the start column is past the maximum column: no need to try. */ |
| 3869 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3870 goto theend; | |
| 3871 | |
| 7 | 3872 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ |
| 3873 if (prog->regflags & RF_ICASE) | |
| 3874 ireg_ic = TRUE; | |
| 3875 else if (prog->regflags & RF_NOICASE) | |
| 3876 ireg_ic = FALSE; | |
| 3877 | |
| 3878 #ifdef FEAT_MBYTE | |
| 3879 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
| 3880 if (prog->regflags & RF_ICOMBINE) | |
| 3881 ireg_icombine = TRUE; | |
| 3882 #endif | |
| 3883 | |
| 3884 /* If there is a "must appear" string, look for it. */ | |
| 3885 if (prog->regmust != NULL) | |
| 3886 { | |
| 3887 int c; | |
| 3888 | |
| 3889 #ifdef FEAT_MBYTE | |
| 3890 if (has_mbyte) | |
| 3891 c = (*mb_ptr2char)(prog->regmust); | |
| 3892 else | |
| 3893 #endif | |
| 3894 c = *prog->regmust; | |
| 3895 s = line + col; | |
| 170 | 3896 |
| 3897 /* | |
| 3898 * This is used very often, esp. for ":global". Use three versions of | |
| 3899 * the loop to avoid overhead of conditions. | |
| 3900 */ | |
| 3901 if (!ireg_ic | |
| 3902 #ifdef FEAT_MBYTE | |
| 3903 && !has_mbyte | |
| 3904 #endif | |
| 3905 ) | |
| 3906 while ((s = vim_strbyte(s, c)) != NULL) | |
| 3907 { | |
| 3908 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3909 break; /* Found it. */ | |
| 3910 ++s; | |
| 3911 } | |
| 3912 #ifdef FEAT_MBYTE | |
| 3913 else if (!ireg_ic || (!enc_utf8 && mb_char2len(c) > 1)) | |
| 3914 while ((s = vim_strchr(s, c)) != NULL) | |
| 3915 { | |
| 3916 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3917 break; /* Found it. */ | |
| 3918 mb_ptr_adv(s); | |
| 3919 } | |
| 3920 #endif | |
| 3921 else | |
| 3922 while ((s = cstrchr(s, c)) != NULL) | |
| 3923 { | |
| 3924 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3925 break; /* Found it. */ | |
| 3926 mb_ptr_adv(s); | |
| 3927 } | |
| 7 | 3928 if (s == NULL) /* Not present. */ |
| 3929 goto theend; | |
| 3930 } | |
| 3931 | |
| 3932 regline = line; | |
| 3933 reglnum = 0; | |
| 2578 | 3934 reg_toolong = FALSE; |
| 7 | 3935 |
| 3936 /* Simplest case: Anchored match need be tried only once. */ | |
| 3937 if (prog->reganch) | |
| 3938 { | |
| 3939 int c; | |
| 3940 | |
| 3941 #ifdef FEAT_MBYTE | |
| 3942 if (has_mbyte) | |
| 3943 c = (*mb_ptr2char)(regline + col); | |
| 3944 else | |
| 3945 #endif | |
| 3946 c = regline[col]; | |
| 3947 if (prog->regstart == NUL | |
| 3948 || prog->regstart == c | |
| 3949 || (ireg_ic && (( | |
| 3950 #ifdef FEAT_MBYTE | |
| 3951 (enc_utf8 && utf_fold(prog->regstart) == utf_fold(c))) | |
| 3952 || (c < 255 && prog->regstart < 255 && | |
| 3953 #endif | |
| 1347 | 3954 MB_TOLOWER(prog->regstart) == MB_TOLOWER(c))))) |
| 7 | 3955 retval = regtry(prog, col); |
| 3956 else | |
| 3957 retval = 0; | |
| 3958 } | |
| 3959 else | |
| 3960 { | |
| 1521 | 3961 #ifdef FEAT_RELTIME |
| 3962 int tm_count = 0; | |
| 3963 #endif | |
| 7 | 3964 /* Messy cases: unanchored match. */ |
| 180 | 3965 while (!got_int) |
| 7 | 3966 { |
| 3967 if (prog->regstart != NUL) | |
| 3968 { | |
| 170 | 3969 /* Skip until the char we know it must start with. |
| 3970 * Used often, do some work to avoid call overhead. */ | |
| 3971 if (!ireg_ic | |
| 3972 #ifdef FEAT_MBYTE | |
| 3973 && !has_mbyte | |
| 3974 #endif | |
| 3975 ) | |
| 3976 s = vim_strbyte(regline + col, prog->regstart); | |
| 3977 else | |
| 3978 s = cstrchr(regline + col, prog->regstart); | |
| 7 | 3979 if (s == NULL) |
| 3980 { | |
| 3981 retval = 0; | |
| 3982 break; | |
| 3983 } | |
| 3984 col = (int)(s - regline); | |
| 3985 } | |
| 3986 | |
| 410 | 3987 /* Check for maximum column to try. */ |
| 3988 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3989 { | |
| 3990 retval = 0; | |
| 3991 break; | |
| 3992 } | |
| 3993 | |
| 7 | 3994 retval = regtry(prog, col); |
| 3995 if (retval > 0) | |
| 3996 break; | |
| 3997 | |
| 3998 /* if not currently on the first line, get it again */ | |
| 3999 if (reglnum != 0) | |
| 4000 { | |
| 481 | 4001 reglnum = 0; |
| 7 | 4002 regline = reg_getline((linenr_T)0); |
| 4003 } | |
| 4004 if (regline[col] == NUL) | |
| 4005 break; | |
| 4006 #ifdef FEAT_MBYTE | |
| 4007 if (has_mbyte) | |
| 474 | 4008 col += (*mb_ptr2len)(regline + col); |
| 7 | 4009 else |
| 4010 #endif | |
| 4011 ++col; | |
| 1521 | 4012 #ifdef FEAT_RELTIME |
| 4013 /* Check for timeout once in a twenty times to avoid overhead. */ | |
| 4014 if (tm != NULL && ++tm_count == 20) | |
| 4015 { | |
| 4016 tm_count = 0; | |
| 4017 if (profile_passed_limit(tm)) | |
| 4018 break; | |
| 4019 } | |
| 4020 #endif | |
| 7 | 4021 } |
| 4022 } | |
| 4023 | |
| 4024 theend: | |
| 1520 | 4025 /* Free "reg_tofree" when it's a bit big. |
| 4026 * Free regstack and backpos if they are bigger than their initial size. */ | |
| 4027 if (reg_tofreelen > 400) | |
| 4028 { | |
| 4029 vim_free(reg_tofree); | |
| 4030 reg_tofree = NULL; | |
| 4031 } | |
| 4032 if (regstack.ga_maxlen > REGSTACK_INITIAL) | |
| 4033 ga_clear(®stack); | |
| 4034 if (backpos.ga_maxlen > BACKPOS_INITIAL) | |
| 4035 ga_clear(&backpos); | |
| 270 | 4036 |
| 7 | 4037 return retval; |
| 4038 } | |
| 4039 | |
| 4040 #ifdef FEAT_SYN_HL | |
| 4041 static reg_extmatch_T *make_extmatch __ARGS((void)); | |
| 4042 | |
| 4043 /* | |
| 4044 * Create a new extmatch and mark it as referenced once. | |
| 4045 */ | |
| 4046 static reg_extmatch_T * | |
| 4047 make_extmatch() | |
| 4048 { | |
| 4049 reg_extmatch_T *em; | |
| 4050 | |
| 4051 em = (reg_extmatch_T *)alloc_clear((unsigned)sizeof(reg_extmatch_T)); | |
| 4052 if (em != NULL) | |
| 4053 em->refcnt = 1; | |
| 4054 return em; | |
| 4055 } | |
| 4056 | |
| 4057 /* | |
| 4058 * Add a reference to an extmatch. | |
| 4059 */ | |
| 4060 reg_extmatch_T * | |
| 4061 ref_extmatch(em) | |
| 4062 reg_extmatch_T *em; | |
| 4063 { | |
| 4064 if (em != NULL) | |
| 4065 em->refcnt++; | |
| 4066 return em; | |
| 4067 } | |
| 4068 | |
| 4069 /* | |
| 4070 * Remove a reference to an extmatch. If there are no references left, free | |
| 4071 * the info. | |
| 4072 */ | |
| 4073 void | |
| 4074 unref_extmatch(em) | |
| 4075 reg_extmatch_T *em; | |
| 4076 { | |
| 4077 int i; | |
| 4078 | |
| 4079 if (em != NULL && --em->refcnt <= 0) | |
| 4080 { | |
| 4081 for (i = 0; i < NSUBEXP; ++i) | |
| 4082 vim_free(em->matches[i]); | |
| 4083 vim_free(em); | |
| 4084 } | |
| 4085 } | |
| 4086 #endif | |
| 4087 | |
| 4088 /* | |
| 4089 * regtry - try match of "prog" with at regline["col"]. | |
| 4090 * Returns 0 for failure, number of lines contained in the match otherwise. | |
| 4091 */ | |
| 4092 static long | |
| 4093 regtry(prog, col) | |
| 4444 | 4094 bt_regprog_T *prog; |
| 7 | 4095 colnr_T col; |
| 4096 { | |
| 4097 reginput = regline + col; | |
| 4098 need_clear_subexpr = TRUE; | |
| 4099 #ifdef FEAT_SYN_HL | |
| 4100 /* Clear the external match subpointers if necessary. */ | |
| 4101 if (prog->reghasz == REX_SET) | |
| 4102 need_clear_zsubexpr = TRUE; | |
| 4103 #endif | |
| 4104 | |
| 189 | 4105 if (regmatch(prog->program + 1) == 0) |
| 4106 return 0; | |
| 4107 | |
| 4108 cleanup_subexpr(); | |
| 4109 if (REG_MULTI) | |
| 7 | 4110 { |
| 189 | 4111 if (reg_startpos[0].lnum < 0) |
| 7 | 4112 { |
| 189 | 4113 reg_startpos[0].lnum = 0; |
| 4114 reg_startpos[0].col = col; | |
| 4115 } | |
| 4116 if (reg_endpos[0].lnum < 0) | |
| 4117 { | |
| 4118 reg_endpos[0].lnum = reglnum; | |
| 4119 reg_endpos[0].col = (int)(reginput - regline); | |
| 7 | 4120 } |
| 4121 else | |
| 189 | 4122 /* Use line number of "\ze". */ |
| 4123 reglnum = reg_endpos[0].lnum; | |
| 4124 } | |
| 4125 else | |
| 4126 { | |
| 4127 if (reg_startp[0] == NULL) | |
| 4128 reg_startp[0] = regline + col; | |
| 4129 if (reg_endp[0] == NULL) | |
| 4130 reg_endp[0] = reginput; | |
| 4131 } | |
| 7 | 4132 #ifdef FEAT_SYN_HL |
| 189 | 4133 /* Package any found \z(...\) matches for export. Default is none. */ |
| 4134 unref_extmatch(re_extmatch_out); | |
| 4135 re_extmatch_out = NULL; | |
| 4136 | |
| 4137 if (prog->reghasz == REX_SET) | |
| 4138 { | |
| 4139 int i; | |
| 4140 | |
| 4141 cleanup_zsubexpr(); | |
| 4142 re_extmatch_out = make_extmatch(); | |
| 4143 for (i = 0; i < NSUBEXP; i++) | |
| 7 | 4144 { |
| 189 | 4145 if (REG_MULTI) |
| 7 | 4146 { |
| 189 | 4147 /* Only accept single line matches. */ |
| 4148 if (reg_startzpos[i].lnum >= 0 | |
| 4149 && reg_endzpos[i].lnum == reg_startzpos[i].lnum) | |
| 4150 re_extmatch_out->matches[i] = | |
| 4151 vim_strnsave(reg_getline(reg_startzpos[i].lnum) | |
| 7 | 4152 + reg_startzpos[i].col, |
| 189 | 4153 reg_endzpos[i].col - reg_startzpos[i].col); |
| 4154 } | |
| 4155 else | |
| 4156 { | |
| 4157 if (reg_startzp[i] != NULL && reg_endzp[i] != NULL) | |
| 4158 re_extmatch_out->matches[i] = | |
| 7 | 4159 vim_strnsave(reg_startzp[i], |
| 189 | 4160 (int)(reg_endzp[i] - reg_startzp[i])); |
| 7 | 4161 } |
| 4162 } | |
| 189 | 4163 } |
| 7 | 4164 #endif |
| 189 | 4165 return 1 + reglnum; |
| 7 | 4166 } |
| 4167 | |
| 4168 #ifdef FEAT_MBYTE | |
| 4169 static int reg_prev_class __ARGS((void)); | |
| 4170 | |
| 4171 /* | |
| 4172 * Get class of previous character. | |
| 4173 */ | |
| 4174 static int | |
| 4175 reg_prev_class() | |
| 4176 { | |
| 4177 if (reginput > regline) | |
| 4069 | 4178 return mb_get_class_buf(reginput - 1 |
| 4179 - (*mb_head_off)(regline, reginput - 1), reg_buf); | |
| 7 | 4180 return -1; |
| 4181 } | |
| 4182 | |
| 4183 #endif | |
|
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4184 #ifdef FEAT_VISUAL |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4185 static int reg_match_visual __ARGS((void)); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4186 |
|
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 * 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
|
4189 */ |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4190 static int |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4191 reg_match_visual() |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4192 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4193 pos_T top, bot; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4194 linenr_T lnum; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4195 colnr_T col; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4196 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
|
4197 int mode; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4198 colnr_T start, end; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4199 colnr_T start2, end2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4200 colnr_T cols; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4201 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4202 /* Check if the buffer is the current buffer. */ |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4203 if (reg_buf != curbuf || VIsual.lnum == 0) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4204 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4205 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4206 if (VIsual_active) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4207 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4208 if (lt(VIsual, wp->w_cursor)) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4209 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4210 top = VIsual; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4211 bot = wp->w_cursor; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4212 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4213 else |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4214 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4215 top = wp->w_cursor; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4216 bot = VIsual; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4217 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4218 mode = VIsual_mode; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4219 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4220 else |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4221 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4222 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
|
4223 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4224 top = curbuf->b_visual.vi_start; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4225 bot = curbuf->b_visual.vi_end; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4226 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4227 else |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4228 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4229 top = curbuf->b_visual.vi_end; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4230 bot = curbuf->b_visual.vi_start; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4231 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4232 mode = curbuf->b_visual.vi_mode; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4233 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4234 lnum = reglnum + reg_firstlnum; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4235 if (lnum < top.lnum || lnum > bot.lnum) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4236 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4237 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4238 if (mode == 'v') |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4239 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4240 col = (colnr_T)(reginput - regline); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4241 if ((lnum == top.lnum && col < top.col) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4242 || (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
|
4243 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4244 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4245 else if (mode == Ctrl_V) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4246 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4247 getvvcol(wp, &top, &start, NULL, &end); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4248 getvvcol(wp, &bot, &start2, NULL, &end2); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4249 if (start2 < start) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4250 start = start2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4251 if (end2 > end) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4252 end = end2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4253 if (top.col == MAXCOL || bot.col == MAXCOL) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4254 end = MAXCOL; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4255 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
|
4256 if (cols < start || cols > end - (*p_sel == 'e')) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4257 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4258 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4259 return TRUE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4260 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4261 #endif |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4262 |
| 39 | 4263 #define ADVANCE_REGINPUT() mb_ptr_adv(reginput) |
| 7 | 4264 |
| 4265 /* | |
| 4266 * The arguments from BRACE_LIMITS are stored here. They are actually local | |
| 4267 * to regmatch(), but they are here to reduce the amount of stack space used | |
| 4268 * (it can be called recursively many times). | |
| 4269 */ | |
| 4270 static long bl_minval; | |
| 4271 static long bl_maxval; | |
| 4272 | |
| 4273 /* | |
| 4274 * regmatch - main matching routine | |
| 4275 * | |
| 180 | 4276 * Conceptually the strategy is simple: Check to see whether the current node |
| 4277 * matches, push an item onto the regstack and loop to see whether the rest | |
| 4278 * matches, and then act accordingly. In practice we make some effort to | |
| 4279 * avoid using the regstack, in particular by going through "ordinary" nodes | |
| 4280 * (that don't need to know whether the rest of the match failed) by a nested | |
| 4281 * loop. | |
| 7 | 4282 * |
| 4283 * Returns TRUE when there is a match. Leaves reginput and reglnum just after | |
| 4284 * the last matched character. | |
| 4285 * Returns FALSE when there is no match. Leaves reginput and reglnum in an | |
| 4286 * undefined state! | |
| 4287 */ | |
| 4288 static int | |
| 180 | 4289 regmatch(scan) |
| 7 | 4290 char_u *scan; /* Current node. */ |
| 4291 { | |
| 180 | 4292 char_u *next; /* Next node. */ |
| 4293 int op; | |
| 4294 int c; | |
| 4295 regitem_T *rp; | |
| 4296 int no; | |
| 4297 int status; /* one of the RA_ values: */ | |
| 4298 #define RA_FAIL 1 /* something failed, abort */ | |
| 4299 #define RA_CONT 2 /* continue in inner loop */ | |
| 4300 #define RA_BREAK 3 /* break inner loop */ | |
| 4301 #define RA_MATCH 4 /* successful match */ | |
| 4302 #define RA_NOMATCH 5 /* didn't match */ | |
| 270 | 4303 |
| 1520 | 4304 /* Make "regstack" and "backpos" empty. They are allocated and freed in |
| 4444 | 4305 * bt_regexec_both() to reduce malloc()/free() calls. */ |
| 270 | 4306 regstack.ga_len = 0; |
| 4307 backpos.ga_len = 0; | |
| 233 | 4308 |
| 180 | 4309 /* |
| 233 | 4310 * Repeat until "regstack" is empty. |
| 180 | 4311 */ |
| 4312 for (;;) | |
| 4313 { | |
| 4444 | 4314 /* Some patterns may cause a long time to match, even though they are not |
| 7 | 4315 * illegal. E.g., "\([a-z]\+\)\+Q". Allow breaking them with CTRL-C. */ |
| 4316 fast_breakcheck(); | |
| 4317 | |
| 4318 #ifdef DEBUG | |
| 4319 if (scan != NULL && regnarrate) | |
| 4320 { | |
| 4444 | 4321 mch_errmsg((char *)regprop(scan)); |
| 7 | 4322 mch_errmsg("(\n"); |
| 4323 } | |
| 4324 #endif | |
| 180 | 4325 |
| 4326 /* | |
| 233 | 4327 * Repeat for items that can be matched sequentially, without using the |
| 180 | 4328 * regstack. |
| 4329 */ | |
| 4330 for (;;) | |
| 7 | 4331 { |
| 180 | 4332 if (got_int || scan == NULL) |
| 4333 { | |
| 4334 status = RA_FAIL; | |
| 4335 break; | |
| 4336 } | |
| 4337 status = RA_CONT; | |
| 4338 | |
| 7 | 4339 #ifdef DEBUG |
| 4340 if (regnarrate) | |
| 4341 { | |
| 4444 | 4342 mch_errmsg((char *)regprop(scan)); |
| 7 | 4343 mch_errmsg("...\n"); |
| 4344 # ifdef FEAT_SYN_HL | |
| 4345 if (re_extmatch_in != NULL) | |
| 4346 { | |
| 4347 int i; | |
| 4348 | |
| 4349 mch_errmsg(_("External submatches:\n")); | |
| 4350 for (i = 0; i < NSUBEXP; i++) | |
| 4351 { | |
| 4352 mch_errmsg(" \""); | |
| 4353 if (re_extmatch_in->matches[i] != NULL) | |
| 4444 | 4354 mch_errmsg((char *)re_extmatch_in->matches[i]); |
| 7 | 4355 mch_errmsg("\"\n"); |
| 4356 } | |
| 4357 } | |
| 4358 # endif | |
| 4359 } | |
| 4360 #endif | |
| 4361 next = regnext(scan); | |
| 4362 | |
| 4363 op = OP(scan); | |
| 4364 /* Check for character class with NL added. */ | |
| 1018 | 4365 if (!reg_line_lbr && WITH_NL(op) && REG_MULTI |
| 4366 && *reginput == NUL && reglnum <= reg_maxline) | |
| 7 | 4367 { |
| 4368 reg_nextline(); | |
| 4369 } | |
| 4370 else if (reg_line_lbr && WITH_NL(op) && *reginput == '\n') | |
| 4371 { | |
| 4372 ADVANCE_REGINPUT(); | |
| 4373 } | |
| 4374 else | |
| 4375 { | |
| 4376 if (WITH_NL(op)) | |
| 180 | 4377 op -= ADD_NL; |
| 7 | 4378 #ifdef FEAT_MBYTE |
| 4379 if (has_mbyte) | |
| 4380 c = (*mb_ptr2char)(reginput); | |
| 4381 else | |
| 4382 #endif | |
| 4383 c = *reginput; | |
| 4384 switch (op) | |
| 4385 { | |
| 4386 case BOL: | |
| 4387 if (reginput != regline) | |
| 180 | 4388 status = RA_NOMATCH; |
| 7 | 4389 break; |
| 4390 | |
| 4391 case EOL: | |
| 4392 if (c != NUL) | |
| 180 | 4393 status = RA_NOMATCH; |
| 7 | 4394 break; |
| 4395 | |
| 4396 case RE_BOF: | |
| 1458 | 4397 /* We're not at the beginning of the file when below the first |
| 4398 * line where we started, not at the start of the line or we | |
| 4399 * didn't start at the first line of the buffer. */ | |
| 7 | 4400 if (reglnum != 0 || reginput != regline |
| 1458 | 4401 || (REG_MULTI && reg_firstlnum > 1)) |
| 180 | 4402 status = RA_NOMATCH; |
| 7 | 4403 break; |
| 4404 | |
| 4405 case RE_EOF: | |
| 4406 if (reglnum != reg_maxline || c != NUL) | |
| 180 | 4407 status = RA_NOMATCH; |
| 7 | 4408 break; |
| 4409 | |
| 4410 case CURSOR: | |
| 4411 /* Check if the buffer is in a window and compare the | |
| 4412 * reg_win->w_cursor position to the match position. */ | |
| 4413 if (reg_win == NULL | |
| 4414 || (reglnum + reg_firstlnum != reg_win->w_cursor.lnum) | |
| 4415 || ((colnr_T)(reginput - regline) != reg_win->w_cursor.col)) | |
| 180 | 4416 status = RA_NOMATCH; |
| 7 | 4417 break; |
| 4418 | |
| 639 | 4419 case RE_MARK: |
|
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4420 /* Compare the mark position to the match position. */ |
| 639 | 4421 { |
| 4422 int mark = OPERAND(scan)[0]; | |
| 4423 int cmp = OPERAND(scan)[1]; | |
| 4424 pos_T *pos; | |
| 4425 | |
| 4043 | 4426 pos = getmark_buf(reg_buf, mark, FALSE); |
| 1148 | 4427 if (pos == NULL /* mark doesn't exist */ |
|
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4428 || pos->lnum <= 0 /* mark isn't set in reg_buf */ |
| 639 | 4429 || (pos->lnum == reglnum + reg_firstlnum |
| 4430 ? (pos->col == (colnr_T)(reginput - regline) | |
| 4431 ? (cmp == '<' || cmp == '>') | |
| 4432 : (pos->col < (colnr_T)(reginput - regline) | |
| 4433 ? cmp != '>' | |
| 4434 : cmp != '<')) | |
| 4435 : (pos->lnum < reglnum + reg_firstlnum | |
| 4436 ? cmp != '>' | |
| 4437 : cmp != '<'))) | |
| 4438 status = RA_NOMATCH; | |
| 4439 } | |
| 4440 break; | |
| 4441 | |
| 4442 case RE_VISUAL: | |
| 4443 #ifdef FEAT_VISUAL | |
|
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4444 if (!reg_match_visual()) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4445 #endif |
| 639 | 4446 status = RA_NOMATCH; |
| 4447 break; | |
| 4448 | |
| 7 | 4449 case RE_LNUM: |
| 4450 if (!REG_MULTI || !re_num_cmp((long_u)(reglnum + reg_firstlnum), | |
| 4451 scan)) | |
| 180 | 4452 status = RA_NOMATCH; |
| 7 | 4453 break; |
| 4454 | |
| 4455 case RE_COL: | |
| 4456 if (!re_num_cmp((long_u)(reginput - regline) + 1, scan)) | |
| 180 | 4457 status = RA_NOMATCH; |
| 7 | 4458 break; |
| 4459 | |
| 4460 case RE_VCOL: | |
| 4461 if (!re_num_cmp((long_u)win_linetabsize( | |
| 4462 reg_win == NULL ? curwin : reg_win, | |
| 4463 regline, (colnr_T)(reginput - regline)) + 1, scan)) | |
| 180 | 4464 status = RA_NOMATCH; |
| 7 | 4465 break; |
| 4466 | |
| 4467 case BOW: /* \<word; reginput points to w */ | |
| 4468 if (c == NUL) /* Can't match at end of line */ | |
| 180 | 4469 status = RA_NOMATCH; |
| 7 | 4470 #ifdef FEAT_MBYTE |
| 180 | 4471 else if (has_mbyte) |
| 7 | 4472 { |
| 4473 int this_class; | |
| 4474 | |
| 4475 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4476 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4477 if (this_class <= 1) |
| 180 | 4478 status = RA_NOMATCH; /* not on a word at all */ |
| 4479 else if (reg_prev_class() == this_class) | |
| 4480 status = RA_NOMATCH; /* previous char is in same word */ | |
| 7 | 4481 } |
| 4482 #endif | |
| 4483 else | |
| 4484 { | |
| 4061 | 4485 if (!vim_iswordc_buf(c, reg_buf) || (reginput > regline |
| 4486 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
| 180 | 4487 status = RA_NOMATCH; |
| 7 | 4488 } |
| 4489 break; | |
| 4490 | |
| 4491 case EOW: /* word\>; reginput points after d */ | |
| 4492 if (reginput == regline) /* Can't match at start of line */ | |
| 180 | 4493 status = RA_NOMATCH; |
| 7 | 4494 #ifdef FEAT_MBYTE |
| 180 | 4495 else if (has_mbyte) |
| 7 | 4496 { |
| 4497 int this_class, prev_class; | |
| 4498 | |
| 4499 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4500 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4501 prev_class = reg_prev_class(); |
| 180 | 4502 if (this_class == prev_class |
| 4503 || prev_class == 0 || prev_class == 1) | |
| 4504 status = RA_NOMATCH; | |
| 7 | 4505 } |
| 180 | 4506 #endif |
| 7 | 4507 else |
| 4508 { | |
| 4043 | 4509 if (!vim_iswordc_buf(reginput[-1], reg_buf) |
| 4510 || (reginput[0] != NUL && vim_iswordc_buf(c, reg_buf))) | |
| 180 | 4511 status = RA_NOMATCH; |
| 7 | 4512 } |
| 4513 break; /* Matched with EOW */ | |
| 4514 | |
| 4515 case ANY: | |
| 4084 | 4516 /* ANY does not match new lines. */ |
| 7 | 4517 if (c == NUL) |
| 180 | 4518 status = RA_NOMATCH; |
| 4519 else | |
| 4520 ADVANCE_REGINPUT(); | |
| 7 | 4521 break; |
| 4522 | |
| 4523 case IDENT: | |
| 4524 if (!vim_isIDc(c)) | |
| 180 | 4525 status = RA_NOMATCH; |
| 4526 else | |
| 4527 ADVANCE_REGINPUT(); | |
| 7 | 4528 break; |
| 4529 | |
| 4530 case SIDENT: | |
| 4531 if (VIM_ISDIGIT(*reginput) || !vim_isIDc(c)) | |
| 180 | 4532 status = RA_NOMATCH; |
| 4533 else | |
| 4534 ADVANCE_REGINPUT(); | |
| 7 | 4535 break; |
| 4536 | |
| 4537 case KWORD: | |
| 4069 | 4538 if (!vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4539 status = RA_NOMATCH; |
| 4540 else | |
| 4541 ADVANCE_REGINPUT(); | |
| 7 | 4542 break; |
| 4543 | |
| 4544 case SKWORD: | |
| 4069 | 4545 if (VIM_ISDIGIT(*reginput) || !vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4546 status = RA_NOMATCH; |
| 4547 else | |
| 4548 ADVANCE_REGINPUT(); | |
| 7 | 4549 break; |
| 4550 | |
| 4551 case FNAME: | |
| 4552 if (!vim_isfilec(c)) | |
| 180 | 4553 status = RA_NOMATCH; |
| 4554 else | |
| 4555 ADVANCE_REGINPUT(); | |
| 7 | 4556 break; |
| 4557 | |
| 4558 case SFNAME: | |
| 4559 if (VIM_ISDIGIT(*reginput) || !vim_isfilec(c)) | |
| 180 | 4560 status = RA_NOMATCH; |
| 4561 else | |
| 4562 ADVANCE_REGINPUT(); | |
| 7 | 4563 break; |
| 4564 | |
| 4565 case PRINT: | |
| 4566 if (ptr2cells(reginput) != 1) | |
| 180 | 4567 status = RA_NOMATCH; |
| 4568 else | |
| 4569 ADVANCE_REGINPUT(); | |
| 7 | 4570 break; |
| 4571 | |
| 4572 case SPRINT: | |
| 4573 if (VIM_ISDIGIT(*reginput) || ptr2cells(reginput) != 1) | |
| 180 | 4574 status = RA_NOMATCH; |
| 4575 else | |
| 4576 ADVANCE_REGINPUT(); | |
| 7 | 4577 break; |
| 4578 | |
| 4579 case WHITE: | |
| 4580 if (!vim_iswhite(c)) | |
| 180 | 4581 status = RA_NOMATCH; |
| 4582 else | |
| 4583 ADVANCE_REGINPUT(); | |
| 7 | 4584 break; |
| 4585 | |
| 4586 case NWHITE: | |
| 4587 if (c == NUL || vim_iswhite(c)) | |
| 180 | 4588 status = RA_NOMATCH; |
| 4589 else | |
| 4590 ADVANCE_REGINPUT(); | |
| 7 | 4591 break; |
| 4592 | |
| 4593 case DIGIT: | |
| 4594 if (!ri_digit(c)) | |
| 180 | 4595 status = RA_NOMATCH; |
| 4596 else | |
| 4597 ADVANCE_REGINPUT(); | |
| 7 | 4598 break; |
| 4599 | |
| 4600 case NDIGIT: | |
| 4601 if (c == NUL || ri_digit(c)) | |
| 180 | 4602 status = RA_NOMATCH; |
| 4603 else | |
| 4604 ADVANCE_REGINPUT(); | |
| 7 | 4605 break; |
| 4606 | |
| 4607 case HEX: | |
| 4608 if (!ri_hex(c)) | |
| 180 | 4609 status = RA_NOMATCH; |
| 4610 else | |
| 4611 ADVANCE_REGINPUT(); | |
| 7 | 4612 break; |
| 4613 | |
| 4614 case NHEX: | |
| 4615 if (c == NUL || ri_hex(c)) | |
| 180 | 4616 status = RA_NOMATCH; |
| 4617 else | |
| 4618 ADVANCE_REGINPUT(); | |
| 7 | 4619 break; |
| 4620 | |
| 4621 case OCTAL: | |
| 4622 if (!ri_octal(c)) | |
| 180 | 4623 status = RA_NOMATCH; |
| 4624 else | |
| 4625 ADVANCE_REGINPUT(); | |
| 7 | 4626 break; |
| 4627 | |
| 4628 case NOCTAL: | |
| 4629 if (c == NUL || ri_octal(c)) | |
| 180 | 4630 status = RA_NOMATCH; |
| 4631 else | |
| 4632 ADVANCE_REGINPUT(); | |
| 7 | 4633 break; |
| 4634 | |
| 4635 case WORD: | |
| 4636 if (!ri_word(c)) | |
| 180 | 4637 status = RA_NOMATCH; |
| 4638 else | |
| 4639 ADVANCE_REGINPUT(); | |
| 7 | 4640 break; |
| 4641 | |
| 4642 case NWORD: | |
| 4643 if (c == NUL || ri_word(c)) | |
| 180 | 4644 status = RA_NOMATCH; |
| 4645 else | |
| 4646 ADVANCE_REGINPUT(); | |
| 7 | 4647 break; |
| 4648 | |
| 4649 case HEAD: | |
| 4650 if (!ri_head(c)) | |
| 180 | 4651 status = RA_NOMATCH; |
| 4652 else | |
| 4653 ADVANCE_REGINPUT(); | |
| 7 | 4654 break; |
| 4655 | |
| 4656 case NHEAD: | |
| 4657 if (c == NUL || ri_head(c)) | |
| 180 | 4658 status = RA_NOMATCH; |
| 4659 else | |
| 4660 ADVANCE_REGINPUT(); | |
| 7 | 4661 break; |
| 4662 | |
| 4663 case ALPHA: | |
| 4664 if (!ri_alpha(c)) | |
| 180 | 4665 status = RA_NOMATCH; |
| 4666 else | |
| 4667 ADVANCE_REGINPUT(); | |
| 7 | 4668 break; |
| 4669 | |
| 4670 case NALPHA: | |
| 4671 if (c == NUL || ri_alpha(c)) | |
| 180 | 4672 status = RA_NOMATCH; |
| 4673 else | |
| 4674 ADVANCE_REGINPUT(); | |
| 7 | 4675 break; |
| 4676 | |
| 4677 case LOWER: | |
| 4678 if (!ri_lower(c)) | |
| 180 | 4679 status = RA_NOMATCH; |
| 4680 else | |
| 4681 ADVANCE_REGINPUT(); | |
| 7 | 4682 break; |
| 4683 | |
| 4684 case NLOWER: | |
| 4685 if (c == NUL || ri_lower(c)) | |
| 180 | 4686 status = RA_NOMATCH; |
| 4687 else | |
| 4688 ADVANCE_REGINPUT(); | |
| 7 | 4689 break; |
| 4690 | |
| 4691 case UPPER: | |
| 4692 if (!ri_upper(c)) | |
| 180 | 4693 status = RA_NOMATCH; |
| 4694 else | |
| 4695 ADVANCE_REGINPUT(); | |
| 7 | 4696 break; |
| 4697 | |
| 4698 case NUPPER: | |
| 4699 if (c == NUL || ri_upper(c)) | |
| 180 | 4700 status = RA_NOMATCH; |
| 4701 else | |
| 4702 ADVANCE_REGINPUT(); | |
| 7 | 4703 break; |
| 4704 | |
| 4705 case EXACTLY: | |
| 4706 { | |
| 4707 int len; | |
| 4708 char_u *opnd; | |
| 4709 | |
| 4710 opnd = OPERAND(scan); | |
| 4711 /* Inline the first byte, for speed. */ | |
| 4712 if (*opnd != *reginput | |
| 4713 && (!ireg_ic || ( | |
| 4714 #ifdef FEAT_MBYTE | |
| 4715 !enc_utf8 && | |
| 4716 #endif | |
| 1347 | 4717 MB_TOLOWER(*opnd) != MB_TOLOWER(*reginput)))) |
| 180 | 4718 status = RA_NOMATCH; |
| 4719 else if (*opnd == NUL) | |
| 7 | 4720 { |
| 4721 /* match empty string always works; happens when "~" is | |
| 4722 * empty. */ | |
| 4723 } | |
| 4724 else if (opnd[1] == NUL | |
| 4725 #ifdef FEAT_MBYTE | |
| 4726 && !(enc_utf8 && ireg_ic) | |
| 4727 #endif | |
| 4728 ) | |
| 4729 ++reginput; /* matched a single char */ | |
| 4730 else | |
| 4731 { | |
| 4732 len = (int)STRLEN(opnd); | |
| 4733 /* Need to match first byte again for multi-byte. */ | |
| 4734 if (cstrncmp(opnd, reginput, &len) != 0) | |
| 180 | 4735 status = RA_NOMATCH; |
| 7 | 4736 #ifdef FEAT_MBYTE |
| 4737 /* Check for following composing character. */ | |
| 180 | 4738 else if (enc_utf8 |
| 4739 && UTF_COMPOSINGLIKE(reginput, reginput + len)) | |
| 7 | 4740 { |
| 4741 /* raaron: This code makes a composing character get | |
| 4742 * ignored, which is the correct behavior (sometimes) | |
| 4743 * for voweled Hebrew texts. */ | |
| 4744 if (!ireg_icombine) | |
| 180 | 4745 status = RA_NOMATCH; |
| 7 | 4746 } |
| 180 | 4747 #endif |
| 7 | 4748 else |
| 180 | 4749 reginput += len; |
| 7 | 4750 } |
| 4751 } | |
| 4752 break; | |
| 4753 | |
| 4754 case ANYOF: | |
| 4755 case ANYBUT: | |
| 4756 if (c == NUL) | |
| 180 | 4757 status = RA_NOMATCH; |
| 4758 else if ((cstrchr(OPERAND(scan), c) == NULL) == (op == ANYOF)) | |
| 4759 status = RA_NOMATCH; | |
| 4760 else | |
| 4761 ADVANCE_REGINPUT(); | |
| 7 | 4762 break; |
| 4763 | |
| 4764 #ifdef FEAT_MBYTE | |
| 4765 case MULTIBYTECODE: | |
| 4766 if (has_mbyte) | |
| 4767 { | |
| 4768 int i, len; | |
| 4769 char_u *opnd; | |
| 944 | 4770 int opndc = 0, inpc; |
| 7 | 4771 |
| 4772 opnd = OPERAND(scan); | |
| 4773 /* Safety check (just in case 'encoding' was changed since | |
| 4774 * compiling the program). */ | |
| 474 | 4775 if ((len = (*mb_ptr2len)(opnd)) < 2) |
| 180 | 4776 { |
| 4777 status = RA_NOMATCH; | |
| 4778 break; | |
| 4779 } | |
| 714 | 4780 if (enc_utf8) |
| 4781 opndc = mb_ptr2char(opnd); | |
| 4782 if (enc_utf8 && utf_iscomposing(opndc)) | |
| 4783 { | |
| 4784 /* When only a composing char is given match at any | |
| 4785 * position where that composing char appears. */ | |
| 4786 status = RA_NOMATCH; | |
| 4787 for (i = 0; reginput[i] != NUL; i += utf_char2len(inpc)) | |
| 180 | 4788 { |
| 714 | 4789 inpc = mb_ptr2char(reginput + i); |
| 4790 if (!utf_iscomposing(inpc)) | |
| 4791 { | |
| 4792 if (i > 0) | |
| 4793 break; | |
| 4794 } | |
| 4795 else if (opndc == inpc) | |
| 4796 { | |
| 4797 /* Include all following composing chars. */ | |
| 4798 len = i + mb_ptr2len(reginput + i); | |
| 4799 status = RA_MATCH; | |
| 4800 break; | |
| 4801 } | |
| 180 | 4802 } |
| 714 | 4803 } |
| 4804 else | |
| 4805 for (i = 0; i < len; ++i) | |
| 4806 if (opnd[i] != reginput[i]) | |
| 4807 { | |
| 4808 status = RA_NOMATCH; | |
| 4809 break; | |
| 4810 } | |
| 7 | 4811 reginput += len; |
| 4812 } | |
| 4813 else | |
| 180 | 4814 status = RA_NOMATCH; |
| 7 | 4815 break; |
| 4816 #endif | |
| 4817 | |
| 4818 case NOTHING: | |
| 4819 break; | |
| 4820 | |
| 4821 case BACK: | |
| 233 | 4822 { |
| 4823 int i; | |
| 4824 backpos_T *bp; | |
| 4825 | |
| 4826 /* | |
| 4827 * When we run into BACK we need to check if we don't keep | |
| 4828 * looping without matching any input. The second and later | |
| 4829 * times a BACK is encountered it fails if the input is still | |
| 4830 * at the same position as the previous time. | |
| 4831 * The positions are stored in "backpos" and found by the | |
| 4832 * current value of "scan", the position in the RE program. | |
| 4833 */ | |
| 4834 bp = (backpos_T *)backpos.ga_data; | |
| 4835 for (i = 0; i < backpos.ga_len; ++i) | |
| 4836 if (bp[i].bp_scan == scan) | |
| 4837 break; | |
| 4838 if (i == backpos.ga_len) | |
| 4839 { | |
| 4840 /* First time at this BACK, make room to store the pos. */ | |
| 4841 if (ga_grow(&backpos, 1) == FAIL) | |
| 4842 status = RA_FAIL; | |
| 4843 else | |
| 4844 { | |
| 4845 /* get "ga_data" again, it may have changed */ | |
| 4846 bp = (backpos_T *)backpos.ga_data; | |
| 4847 bp[i].bp_scan = scan; | |
| 4848 ++backpos.ga_len; | |
| 4849 } | |
| 4850 } | |
| 4851 else if (reg_save_equal(&bp[i].bp_pos)) | |
| 4852 /* Still at same position as last time, fail. */ | |
| 4853 status = RA_NOMATCH; | |
| 4854 | |
| 4855 if (status != RA_FAIL && status != RA_NOMATCH) | |
| 4856 reg_save(&bp[i].bp_pos, &backpos); | |
| 4857 } | |
| 179 | 4858 break; |
| 4859 | |
| 7 | 4860 case MOPEN + 0: /* Match start: \zs */ |
| 4861 case MOPEN + 1: /* \( */ | |
| 4862 case MOPEN + 2: | |
| 4863 case MOPEN + 3: | |
| 4864 case MOPEN + 4: | |
| 4865 case MOPEN + 5: | |
| 4866 case MOPEN + 6: | |
| 4867 case MOPEN + 7: | |
| 4868 case MOPEN + 8: | |
| 4869 case MOPEN + 9: | |
| 4870 { | |
| 4871 no = op - MOPEN; | |
| 4872 cleanup_subexpr(); | |
| 270 | 4873 rp = regstack_push(RS_MOPEN, scan); |
| 180 | 4874 if (rp == NULL) |
| 4875 status = RA_FAIL; | |
| 4876 else | |
| 4877 { | |
| 4878 rp->rs_no = no; | |
| 4879 save_se(&rp->rs_un.sesave, ®_startpos[no], | |
| 4880 ®_startp[no]); | |
| 4881 /* We simply continue and handle the result when done. */ | |
| 4882 } | |
| 7 | 4883 } |
| 180 | 4884 break; |
| 7 | 4885 |
| 4886 case NOPEN: /* \%( */ | |
| 4887 case NCLOSE: /* \) after \%( */ | |
| 270 | 4888 if (regstack_push(RS_NOPEN, scan) == NULL) |
| 180 | 4889 status = RA_FAIL; |
| 4890 /* We simply continue and handle the result when done. */ | |
| 4891 break; | |
| 7 | 4892 |
| 4893 #ifdef FEAT_SYN_HL | |
| 4894 case ZOPEN + 1: | |
| 4895 case ZOPEN + 2: | |
| 4896 case ZOPEN + 3: | |
| 4897 case ZOPEN + 4: | |
| 4898 case ZOPEN + 5: | |
| 4899 case ZOPEN + 6: | |
| 4900 case ZOPEN + 7: | |
| 4901 case ZOPEN + 8: | |
| 4902 case ZOPEN + 9: | |
| 4903 { | |
| 4904 no = op - ZOPEN; | |
| 4905 cleanup_zsubexpr(); | |
| 270 | 4906 rp = regstack_push(RS_ZOPEN, scan); |
| 180 | 4907 if (rp == NULL) |
| 4908 status = RA_FAIL; | |
| 4909 else | |
| 4910 { | |
| 4911 rp->rs_no = no; | |
| 4912 save_se(&rp->rs_un.sesave, ®_startzpos[no], | |
| 4913 ®_startzp[no]); | |
| 4914 /* We simply continue and handle the result when done. */ | |
| 4915 } | |
| 7 | 4916 } |
| 180 | 4917 break; |
| 7 | 4918 #endif |
| 4919 | |
| 4920 case MCLOSE + 0: /* Match end: \ze */ | |
| 4921 case MCLOSE + 1: /* \) */ | |
| 4922 case MCLOSE + 2: | |
| 4923 case MCLOSE + 3: | |
| 4924 case MCLOSE + 4: | |
| 4925 case MCLOSE + 5: | |
| 4926 case MCLOSE + 6: | |
| 4927 case MCLOSE + 7: | |
| 4928 case MCLOSE + 8: | |
| 4929 case MCLOSE + 9: | |
| 4930 { | |
| 4931 no = op - MCLOSE; | |
| 4932 cleanup_subexpr(); | |
| 270 | 4933 rp = regstack_push(RS_MCLOSE, scan); |
| 180 | 4934 if (rp == NULL) |
| 4935 status = RA_FAIL; | |
| 4936 else | |
| 4937 { | |
| 4938 rp->rs_no = no; | |
| 4939 save_se(&rp->rs_un.sesave, ®_endpos[no], ®_endp[no]); | |
| 4940 /* We simply continue and handle the result when done. */ | |
| 4941 } | |
| 7 | 4942 } |
| 180 | 4943 break; |
| 7 | 4944 |
| 4945 #ifdef FEAT_SYN_HL | |
| 4946 case ZCLOSE + 1: /* \) after \z( */ | |
| 4947 case ZCLOSE + 2: | |
| 4948 case ZCLOSE + 3: | |
| 4949 case ZCLOSE + 4: | |
| 4950 case ZCLOSE + 5: | |
| 4951 case ZCLOSE + 6: | |
| 4952 case ZCLOSE + 7: | |
| 4953 case ZCLOSE + 8: | |
| 4954 case ZCLOSE + 9: | |
| 4955 { | |
| 4956 no = op - ZCLOSE; | |
| 4957 cleanup_zsubexpr(); | |
| 270 | 4958 rp = regstack_push(RS_ZCLOSE, scan); |
| 180 | 4959 if (rp == NULL) |
| 4960 status = RA_FAIL; | |
| 4961 else | |
| 4962 { | |
| 4963 rp->rs_no = no; | |
| 4964 save_se(&rp->rs_un.sesave, ®_endzpos[no], | |
| 4965 ®_endzp[no]); | |
| 4966 /* We simply continue and handle the result when done. */ | |
| 4967 } | |
| 7 | 4968 } |
| 180 | 4969 break; |
| 7 | 4970 #endif |
| 4971 | |
| 4972 case BACKREF + 1: | |
| 4973 case BACKREF + 2: | |
| 4974 case BACKREF + 3: | |
| 4975 case BACKREF + 4: | |
| 4976 case BACKREF + 5: | |
| 4977 case BACKREF + 6: | |
| 4978 case BACKREF + 7: | |
| 4979 case BACKREF + 8: | |
| 4980 case BACKREF + 9: | |
| 4981 { | |
| 4982 int len; | |
| 4983 | |
| 4984 no = op - BACKREF; | |
| 4985 cleanup_subexpr(); | |
| 4986 if (!REG_MULTI) /* Single-line regexp */ | |
| 4987 { | |
| 1815 | 4988 if (reg_startp[no] == NULL || reg_endp[no] == NULL) |
| 7 | 4989 { |
| 4990 /* Backref was not set: Match an empty string. */ | |
| 4991 len = 0; | |
| 4992 } | |
| 4993 else | |
| 4994 { | |
| 4995 /* Compare current input with back-ref in the same | |
| 4996 * line. */ | |
| 4997 len = (int)(reg_endp[no] - reg_startp[no]); | |
| 4998 if (cstrncmp(reg_startp[no], reginput, &len) != 0) | |
| 180 | 4999 status = RA_NOMATCH; |
| 7 | 5000 } |
| 5001 } | |
| 5002 else /* Multi-line regexp */ | |
| 5003 { | |
| 1815 | 5004 if (reg_startpos[no].lnum < 0 || reg_endpos[no].lnum < 0) |
| 7 | 5005 { |
| 5006 /* Backref was not set: Match an empty string. */ | |
| 5007 len = 0; | |
| 5008 } | |
| 5009 else | |
| 5010 { | |
| 5011 if (reg_startpos[no].lnum == reglnum | |
| 5012 && reg_endpos[no].lnum == reglnum) | |
| 5013 { | |
| 5014 /* Compare back-ref within the current line. */ | |
| 5015 len = reg_endpos[no].col - reg_startpos[no].col; | |
| 5016 if (cstrncmp(regline + reg_startpos[no].col, | |
| 5017 reginput, &len) != 0) | |
| 180 | 5018 status = RA_NOMATCH; |
| 7 | 5019 } |
| 5020 else | |
| 5021 { | |
| 5022 /* Messy situation: Need to compare between two | |
| 5023 * lines. */ | |
|
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5024 int r = match_with_backref( |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5025 reg_startpos[no].lnum, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5026 reg_startpos[no].col, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5027 reg_endpos[no].lnum, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5028 reg_endpos[no].col, |
|
4899
4837fd61be52
updated for version 7.3.1195
Bram Moolenaar <bram@vim.org>
parents:
4891
diff
changeset
|
5029 &len); |
|
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5030 |
|
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5031 if (r != RA_MATCH) |
|
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5032 status = r; |
| 7 | 5033 } |
| 5034 } | |
| 5035 } | |
| 5036 | |
| 5037 /* Matched the backref, skip over it. */ | |
| 5038 reginput += len; | |
| 5039 } | |
| 5040 break; | |
| 5041 | |
| 5042 #ifdef FEAT_SYN_HL | |
| 5043 case ZREF + 1: | |
| 5044 case ZREF + 2: | |
| 5045 case ZREF + 3: | |
| 5046 case ZREF + 4: | |
| 5047 case ZREF + 5: | |
| 5048 case ZREF + 6: | |
| 5049 case ZREF + 7: | |
| 5050 case ZREF + 8: | |
| 5051 case ZREF + 9: | |
| 5052 { | |
| 5053 int len; | |
| 5054 | |
| 5055 cleanup_zsubexpr(); | |
| 5056 no = op - ZREF; | |
| 5057 if (re_extmatch_in != NULL | |
| 5058 && re_extmatch_in->matches[no] != NULL) | |
| 5059 { | |
| 5060 len = (int)STRLEN(re_extmatch_in->matches[no]); | |
| 5061 if (cstrncmp(re_extmatch_in->matches[no], | |
| 5062 reginput, &len) != 0) | |
| 180 | 5063 status = RA_NOMATCH; |
| 5064 else | |
| 5065 reginput += len; | |
| 7 | 5066 } |
| 5067 else | |
| 5068 { | |
| 5069 /* Backref was not set: Match an empty string. */ | |
| 5070 } | |
| 5071 } | |
| 5072 break; | |
| 5073 #endif | |
| 5074 | |
| 5075 case BRANCH: | |
| 5076 { | |
| 5077 if (OP(next) != BRANCH) /* No choice. */ | |
| 5078 next = OPERAND(scan); /* Avoid recursion. */ | |
| 5079 else | |
| 5080 { | |
| 270 | 5081 rp = regstack_push(RS_BRANCH, scan); |
| 180 | 5082 if (rp == NULL) |
| 5083 status = RA_FAIL; | |
| 5084 else | |
| 5085 status = RA_BREAK; /* rest is below */ | |
| 7 | 5086 } |
| 5087 } | |
| 5088 break; | |
| 5089 | |
| 5090 case BRACE_LIMITS: | |
| 5091 { | |
| 5092 if (OP(next) == BRACE_SIMPLE) | |
| 5093 { | |
| 5094 bl_minval = OPERAND_MIN(scan); | |
| 5095 bl_maxval = OPERAND_MAX(scan); | |
| 5096 } | |
| 5097 else if (OP(next) >= BRACE_COMPLEX | |
| 5098 && OP(next) < BRACE_COMPLEX + 10) | |
| 5099 { | |
| 5100 no = OP(next) - BRACE_COMPLEX; | |
| 5101 brace_min[no] = OPERAND_MIN(scan); | |
| 5102 brace_max[no] = OPERAND_MAX(scan); | |
| 5103 brace_count[no] = 0; | |
| 5104 } | |
| 5105 else | |
| 5106 { | |
| 5107 EMSG(_(e_internal)); /* Shouldn't happen */ | |
| 180 | 5108 status = RA_FAIL; |
| 7 | 5109 } |
| 5110 } | |
| 5111 break; | |
| 5112 | |
| 5113 case BRACE_COMPLEX + 0: | |
| 5114 case BRACE_COMPLEX + 1: | |
| 5115 case BRACE_COMPLEX + 2: | |
| 5116 case BRACE_COMPLEX + 3: | |
| 5117 case BRACE_COMPLEX + 4: | |
| 5118 case BRACE_COMPLEX + 5: | |
| 5119 case BRACE_COMPLEX + 6: | |
| 5120 case BRACE_COMPLEX + 7: | |
| 5121 case BRACE_COMPLEX + 8: | |
| 5122 case BRACE_COMPLEX + 9: | |
| 5123 { | |
| 5124 no = op - BRACE_COMPLEX; | |
| 5125 ++brace_count[no]; | |
| 5126 | |
| 5127 /* If not matched enough times yet, try one more */ | |
| 5128 if (brace_count[no] <= (brace_min[no] <= brace_max[no] | |
| 180 | 5129 ? brace_min[no] : brace_max[no])) |
| 7 | 5130 { |
| 270 | 5131 rp = regstack_push(RS_BRCPLX_MORE, scan); |
| 180 | 5132 if (rp == NULL) |
| 5133 status = RA_FAIL; | |
| 5134 else | |
| 5135 { | |
| 5136 rp->rs_no = no; | |
| 233 | 5137 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5138 next = OPERAND(scan); |
| 5139 /* We continue and handle the result when done. */ | |
| 5140 } | |
| 5141 break; | |
| 7 | 5142 } |
| 5143 | |
| 5144 /* If matched enough times, may try matching some more */ | |
| 5145 if (brace_min[no] <= brace_max[no]) | |
| 5146 { | |
| 5147 /* Range is the normal way around, use longest match */ | |
| 5148 if (brace_count[no] <= brace_max[no]) | |
| 5149 { | |
| 270 | 5150 rp = regstack_push(RS_BRCPLX_LONG, scan); |
| 180 | 5151 if (rp == NULL) |
| 5152 status = RA_FAIL; | |
| 5153 else | |
| 5154 { | |
| 5155 rp->rs_no = no; | |
| 233 | 5156 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5157 next = OPERAND(scan); |
| 5158 /* We continue and handle the result when done. */ | |
| 5159 } | |
| 7 | 5160 } |
| 5161 } | |
| 5162 else | |
| 5163 { | |
| 5164 /* Range is backwards, use shortest match first */ | |
| 5165 if (brace_count[no] <= brace_min[no]) | |
| 5166 { | |
| 270 | 5167 rp = regstack_push(RS_BRCPLX_SHORT, scan); |
| 180 | 5168 if (rp == NULL) |
| 5169 status = RA_FAIL; | |
| 5170 else | |
| 5171 { | |
| 233 | 5172 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5173 /* We continue and handle the result when done. */ |
| 5174 } | |
| 7 | 5175 } |
| 5176 } | |
| 5177 } | |
| 5178 break; | |
| 5179 | |
| 5180 case BRACE_SIMPLE: | |
| 5181 case STAR: | |
| 5182 case PLUS: | |
| 5183 { | |
| 180 | 5184 regstar_T rst; |
| 7 | 5185 |
| 5186 /* | |
| 5187 * Lookahead to avoid useless match attempts when we know | |
| 5188 * what character comes next. | |
| 5189 */ | |
| 5190 if (OP(next) == EXACTLY) | |
| 5191 { | |
| 180 | 5192 rst.nextb = *OPERAND(next); |
| 7 | 5193 if (ireg_ic) |
| 5194 { | |
| 1347 | 5195 if (MB_ISUPPER(rst.nextb)) |
| 5196 rst.nextb_ic = MB_TOLOWER(rst.nextb); | |
| 7 | 5197 else |
| 1347 | 5198 rst.nextb_ic = MB_TOUPPER(rst.nextb); |
| 7 | 5199 } |
| 5200 else | |
| 180 | 5201 rst.nextb_ic = rst.nextb; |
| 7 | 5202 } |
| 5203 else | |
| 5204 { | |
| 180 | 5205 rst.nextb = NUL; |
| 5206 rst.nextb_ic = NUL; | |
| 7 | 5207 } |
| 5208 if (op != BRACE_SIMPLE) | |
| 5209 { | |
| 180 | 5210 rst.minval = (op == STAR) ? 0 : 1; |
| 5211 rst.maxval = MAX_LIMIT; | |
| 7 | 5212 } |
| 5213 else | |
| 5214 { | |
| 180 | 5215 rst.minval = bl_minval; |
| 5216 rst.maxval = bl_maxval; | |
| 7 | 5217 } |
| 5218 | |
| 5219 /* | |
| 5220 * When maxval > minval, try matching as much as possible, up | |
| 5221 * to maxval. When maxval < minval, try matching at least the | |
| 5222 * minimal number (since the range is backwards, that's also | |
| 5223 * maxval!). | |
| 5224 */ | |
| 180 | 5225 rst.count = regrepeat(OPERAND(scan), rst.maxval); |
| 7 | 5226 if (got_int) |
| 180 | 5227 { |
| 5228 status = RA_FAIL; | |
| 5229 break; | |
| 5230 } | |
| 5231 if (rst.minval <= rst.maxval | |
| 5232 ? rst.count >= rst.minval : rst.count >= rst.maxval) | |
| 7 | 5233 { |
| 180 | 5234 /* It could match. Prepare for trying to match what |
| 5235 * follows. The code is below. Parameters are stored in | |
| 5236 * a regstar_T on the regstack. */ | |
| 212 | 5237 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5238 { |
| 5239 EMSG(_(e_maxmempat)); | |
| 5240 status = RA_FAIL; | |
| 5241 } | |
| 5242 else if (ga_grow(®stack, sizeof(regstar_T)) == FAIL) | |
| 180 | 5243 status = RA_FAIL; |
| 5244 else | |
| 7 | 5245 { |
| 180 | 5246 regstack.ga_len += sizeof(regstar_T); |
| 270 | 5247 rp = regstack_push(rst.minval <= rst.maxval |
| 233 | 5248 ? RS_STAR_LONG : RS_STAR_SHORT, scan); |
| 180 | 5249 if (rp == NULL) |
| 5250 status = RA_FAIL; | |
| 5251 else | |
| 7 | 5252 { |
| 180 | 5253 *(((regstar_T *)rp) - 1) = rst; |
| 5254 status = RA_BREAK; /* skip the restore bits */ | |
| 7 | 5255 } |
| 5256 } | |
| 5257 } | |
| 5258 else | |
| 180 | 5259 status = RA_NOMATCH; |
| 5260 | |
| 7 | 5261 } |
| 180 | 5262 break; |
| 7 | 5263 |
| 5264 case NOMATCH: | |
| 5265 case MATCH: | |
| 5266 case SUBPAT: | |
| 270 | 5267 rp = regstack_push(RS_NOMATCH, scan); |
| 180 | 5268 if (rp == NULL) |
| 5269 status = RA_FAIL; | |
| 5270 else | |
| 7 | 5271 { |
| 180 | 5272 rp->rs_no = op; |
| 233 | 5273 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5274 next = OPERAND(scan); |
| 5275 /* We continue and handle the result when done. */ | |
| 7 | 5276 } |
| 5277 break; | |
| 5278 | |
| 5279 case BEHIND: | |
| 5280 case NOBEHIND: | |
| 180 | 5281 /* Need a bit of room to store extra positions. */ |
| 212 | 5282 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5283 { |
| 5284 EMSG(_(e_maxmempat)); | |
| 5285 status = RA_FAIL; | |
| 5286 } | |
| 5287 else if (ga_grow(®stack, sizeof(regbehind_T)) == FAIL) | |
| 180 | 5288 status = RA_FAIL; |
| 5289 else | |
| 7 | 5290 { |
| 180 | 5291 regstack.ga_len += sizeof(regbehind_T); |
| 270 | 5292 rp = regstack_push(RS_BEHIND1, scan); |
| 180 | 5293 if (rp == NULL) |
| 5294 status = RA_FAIL; | |
| 5295 else | |
| 7 | 5296 { |
| 1579 | 5297 /* Need to save the subexpr to be able to restore them |
| 5298 * when there is a match but we don't use it. */ | |
| 5299 save_subexpr(((regbehind_T *)rp) - 1); | |
| 5300 | |
| 180 | 5301 rp->rs_no = op; |
| 233 | 5302 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5303 /* First try if what follows matches. If it does then we |
| 5304 * check the behind match by looping. */ | |
| 7 | 5305 } |
| 5306 } | |
| 180 | 5307 break; |
| 7 | 5308 |
| 5309 case BHPOS: | |
| 5310 if (REG_MULTI) | |
| 5311 { | |
| 5312 if (behind_pos.rs_u.pos.col != (colnr_T)(reginput - regline) | |
| 5313 || behind_pos.rs_u.pos.lnum != reglnum) | |
| 180 | 5314 status = RA_NOMATCH; |
| 7 | 5315 } |
| 5316 else if (behind_pos.rs_u.ptr != reginput) | |
| 180 | 5317 status = RA_NOMATCH; |
| 7 | 5318 break; |
| 5319 | |
| 5320 case NEWL: | |
| 1018 | 5321 if ((c != NUL || !REG_MULTI || reglnum > reg_maxline |
| 5322 || reg_line_lbr) && (c != '\n' || !reg_line_lbr)) | |
| 180 | 5323 status = RA_NOMATCH; |
| 5324 else if (reg_line_lbr) | |
| 7 | 5325 ADVANCE_REGINPUT(); |
| 5326 else | |
| 5327 reg_nextline(); | |
| 5328 break; | |
| 5329 | |
| 5330 case END: | |
| 180 | 5331 status = RA_MATCH; /* Success! */ |
| 5332 break; | |
| 7 | 5333 |
| 5334 default: | |
| 5335 EMSG(_(e_re_corr)); | |
| 5336 #ifdef DEBUG | |
| 5337 printf("Illegal op code %d\n", op); | |
| 5338 #endif | |
| 180 | 5339 status = RA_FAIL; |
| 5340 break; | |
| 7 | 5341 } |
| 5342 } | |
| 5343 | |
| 180 | 5344 /* If we can't continue sequentially, break the inner loop. */ |
| 5345 if (status != RA_CONT) | |
| 5346 break; | |
| 5347 | |
| 5348 /* Continue in inner loop, advance to next item. */ | |
| 7 | 5349 scan = next; |
| 180 | 5350 |
| 5351 } /* end of inner loop */ | |
| 7 | 5352 |
| 5353 /* | |
| 180 | 5354 * If there is something on the regstack execute the code for the state. |
| 233 | 5355 * If the state is popped then loop and use the older state. |
| 7 | 5356 */ |
| 180 | 5357 while (regstack.ga_len > 0 && status != RA_FAIL) |
| 5358 { | |
| 5359 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; | |
| 5360 switch (rp->rs_state) | |
| 5361 { | |
| 5362 case RS_NOPEN: | |
| 5363 /* Result is passed on as-is, simply pop the state. */ | |
| 270 | 5364 regstack_pop(&scan); |
| 180 | 5365 break; |
| 5366 | |
| 5367 case RS_MOPEN: | |
| 5368 /* Pop the state. Restore pointers when there is no match. */ | |
| 5369 if (status == RA_NOMATCH) | |
| 5370 restore_se(&rp->rs_un.sesave, ®_startpos[rp->rs_no], | |
| 5371 ®_startp[rp->rs_no]); | |
| 270 | 5372 regstack_pop(&scan); |
| 180 | 5373 break; |
| 5374 | |
| 5375 #ifdef FEAT_SYN_HL | |
| 5376 case RS_ZOPEN: | |
| 5377 /* Pop the state. Restore pointers when there is no match. */ | |
| 5378 if (status == RA_NOMATCH) | |
| 5379 restore_se(&rp->rs_un.sesave, ®_startzpos[rp->rs_no], | |
| 5380 ®_startzp[rp->rs_no]); | |
| 270 | 5381 regstack_pop(&scan); |
| 180 | 5382 break; |
| 5383 #endif | |
| 5384 | |
| 5385 case RS_MCLOSE: | |
| 5386 /* Pop the state. Restore pointers when there is no match. */ | |
| 5387 if (status == RA_NOMATCH) | |
| 5388 restore_se(&rp->rs_un.sesave, ®_endpos[rp->rs_no], | |
| 5389 ®_endp[rp->rs_no]); | |
| 270 | 5390 regstack_pop(&scan); |
| 180 | 5391 break; |
| 5392 | |
| 5393 #ifdef FEAT_SYN_HL | |
| 5394 case RS_ZCLOSE: | |
| 5395 /* Pop the state. Restore pointers when there is no match. */ | |
| 5396 if (status == RA_NOMATCH) | |
| 5397 restore_se(&rp->rs_un.sesave, ®_endzpos[rp->rs_no], | |
| 5398 ®_endzp[rp->rs_no]); | |
| 270 | 5399 regstack_pop(&scan); |
| 180 | 5400 break; |
| 5401 #endif | |
| 5402 | |
| 5403 case RS_BRANCH: | |
| 5404 if (status == RA_MATCH) | |
| 5405 /* this branch matched, use it */ | |
| 270 | 5406 regstack_pop(&scan); |
| 180 | 5407 else |
| 5408 { | |
| 5409 if (status != RA_BREAK) | |
| 5410 { | |
| 5411 /* After a non-matching branch: try next one. */ | |
| 233 | 5412 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5413 scan = rp->rs_scan; |
| 5414 } | |
| 5415 if (scan == NULL || OP(scan) != BRANCH) | |
| 5416 { | |
| 5417 /* no more branches, didn't find a match */ | |
| 5418 status = RA_NOMATCH; | |
| 270 | 5419 regstack_pop(&scan); |
| 180 | 5420 } |
| 5421 else | |
| 5422 { | |
| 5423 /* Prepare to try a branch. */ | |
| 5424 rp->rs_scan = regnext(scan); | |
| 233 | 5425 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5426 scan = OPERAND(scan); |
| 5427 } | |
| 5428 } | |
| 5429 break; | |
| 5430 | |
| 5431 case RS_BRCPLX_MORE: | |
| 5432 /* Pop the state. Restore pointers when there is no match. */ | |
| 5433 if (status == RA_NOMATCH) | |
| 5434 { | |
| 233 | 5435 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5436 --brace_count[rp->rs_no]; /* decrement match count */ |
| 5437 } | |
| 270 | 5438 regstack_pop(&scan); |
| 180 | 5439 break; |
| 5440 | |
| 5441 case RS_BRCPLX_LONG: | |
| 5442 /* Pop the state. Restore pointers when there is no match. */ | |
| 5443 if (status == RA_NOMATCH) | |
| 5444 { | |
| 5445 /* There was no match, but we did find enough matches. */ | |
| 233 | 5446 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5447 --brace_count[rp->rs_no]; |
| 5448 /* continue with the items after "\{}" */ | |
| 5449 status = RA_CONT; | |
| 5450 } | |
| 270 | 5451 regstack_pop(&scan); |
| 180 | 5452 if (status == RA_CONT) |
| 5453 scan = regnext(scan); | |
| 5454 break; | |
| 5455 | |
| 5456 case RS_BRCPLX_SHORT: | |
| 5457 /* Pop the state. Restore pointers when there is no match. */ | |
| 5458 if (status == RA_NOMATCH) | |
| 5459 /* There was no match, try to match one more item. */ | |
| 233 | 5460 reg_restore(&rp->rs_un.regsave, &backpos); |
| 270 | 5461 regstack_pop(&scan); |
| 180 | 5462 if (status == RA_NOMATCH) |
| 5463 { | |
| 5464 scan = OPERAND(scan); | |
| 5465 status = RA_CONT; | |
| 5466 } | |
| 5467 break; | |
| 5468 | |
| 5469 case RS_NOMATCH: | |
| 5470 /* Pop the state. If the operand matches for NOMATCH or | |
| 5471 * doesn't match for MATCH/SUBPAT, we fail. Otherwise backup, | |
| 5472 * except for SUBPAT, and continue with the next item. */ | |
| 5473 if (status == (rp->rs_no == NOMATCH ? RA_MATCH : RA_NOMATCH)) | |
| 5474 status = RA_NOMATCH; | |
| 5475 else | |
| 5476 { | |
| 5477 status = RA_CONT; | |
| 233 | 5478 if (rp->rs_no != SUBPAT) /* zero-width */ |
| 5479 reg_restore(&rp->rs_un.regsave, &backpos); | |
| 180 | 5480 } |
| 270 | 5481 regstack_pop(&scan); |
| 180 | 5482 if (status == RA_CONT) |
| 5483 scan = regnext(scan); | |
| 5484 break; | |
| 5485 | |
| 5486 case RS_BEHIND1: | |
| 5487 if (status == RA_NOMATCH) | |
| 5488 { | |
| 270 | 5489 regstack_pop(&scan); |
| 180 | 5490 regstack.ga_len -= sizeof(regbehind_T); |
| 5491 } | |
| 5492 else | |
| 5493 { | |
| 5494 /* The stuff after BEHIND/NOBEHIND matches. Now try if | |
| 5495 * the behind part does (not) match before the current | |
| 5496 * position in the input. This must be done at every | |
| 5497 * position in the input and checking if the match ends at | |
| 5498 * the current position. */ | |
| 5499 | |
| 5500 /* save the position after the found match for next */ | |
| 233 | 5501 reg_save(&(((regbehind_T *)rp) - 1)->save_after, &backpos); |
| 180 | 5502 |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5503 /* Start looking for a match with operand at the current |
| 1209 | 5504 * position. Go back one character until we find the |
| 180 | 5505 * result, hitting the start of the line or the previous |
| 5506 * line (for multi-line matching). | |
| 5507 * Set behind_pos to where the match should end, BHPOS | |
| 5508 * will match it. Save the current value. */ | |
| 5509 (((regbehind_T *)rp) - 1)->save_behind = behind_pos; | |
| 5510 behind_pos = rp->rs_un.regsave; | |
| 5511 | |
| 5512 rp->rs_state = RS_BEHIND2; | |
| 5513 | |
| 233 | 5514 reg_restore(&rp->rs_un.regsave, &backpos); |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5515 scan = OPERAND(rp->rs_scan) + 4; |
| 180 | 5516 } |
| 5517 break; | |
| 5518 | |
| 5519 case RS_BEHIND2: | |
| 5520 /* | |
| 5521 * Looping for BEHIND / NOBEHIND match. | |
| 5522 */ | |
| 5523 if (status == RA_MATCH && reg_save_equal(&behind_pos)) | |
| 5524 { | |
| 5525 /* found a match that ends where "next" started */ | |
| 5526 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5527 if (rp->rs_no == BEHIND) | |
| 233 | 5528 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5529 &backpos); | |
| 180 | 5530 else |
| 1579 | 5531 { |
| 5532 /* But we didn't want a match. Need to restore the | |
| 5533 * subexpr, because what follows matched, so they have | |
| 5534 * been set. */ | |
| 180 | 5535 status = RA_NOMATCH; |
| 1579 | 5536 restore_subexpr(((regbehind_T *)rp) - 1); |
| 5537 } | |
| 270 | 5538 regstack_pop(&scan); |
| 180 | 5539 regstack.ga_len -= sizeof(regbehind_T); |
| 5540 } | |
| 5541 else | |
| 5542 { | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5543 long limit; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5544 |
| 1579 | 5545 /* No match or a match that doesn't end where we want it: Go |
| 5546 * back one character. May go to previous line once. */ | |
| 180 | 5547 no = OK; |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5548 limit = OPERAND_MIN(rp->rs_scan); |
| 180 | 5549 if (REG_MULTI) |
| 5550 { | |
|
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5551 if (limit > 0 |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5552 && ((rp->rs_un.regsave.rs_u.pos.lnum |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5553 < behind_pos.rs_u.pos.lnum |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5554 ? (colnr_T)STRLEN(regline) |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5555 : behind_pos.rs_u.pos.col) |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5556 - 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
|
5557 no = FAIL; |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5558 else if (rp->rs_un.regsave.rs_u.pos.col == 0) |
| 180 | 5559 { |
| 5560 if (rp->rs_un.regsave.rs_u.pos.lnum | |
| 5561 < behind_pos.rs_u.pos.lnum | |
| 5562 || reg_getline( | |
| 5563 --rp->rs_un.regsave.rs_u.pos.lnum) | |
| 5564 == NULL) | |
| 5565 no = FAIL; | |
| 5566 else | |
| 5567 { | |
| 233 | 5568 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5569 rp->rs_un.regsave.rs_u.pos.col = |
| 5570 (colnr_T)STRLEN(regline); | |
| 5571 } | |
| 5572 } | |
| 5573 else | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5574 { |
| 4176 | 5575 #ifdef FEAT_MBYTE |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5576 if (has_mbyte) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5577 rp->rs_un.regsave.rs_u.pos.col -= |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5578 (*mb_head_off)(regline, regline |
| 4176 | 5579 + 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
|
5580 else |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5581 #endif |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5582 --rp->rs_un.regsave.rs_u.pos.col; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5583 } |
| 180 | 5584 } |
| 5585 else | |
| 5586 { | |
| 5587 if (rp->rs_un.regsave.rs_u.ptr == regline) | |
| 5588 no = FAIL; | |
| 5589 else | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5590 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5591 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
|
5592 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
|
5593 - rp->rs_un.regsave.rs_u.ptr) > limit) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5594 no = FAIL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5595 } |
| 180 | 5596 } |
| 5597 if (no == OK) | |
| 5598 { | |
| 5599 /* Advanced, prepare for finding match again. */ | |
| 233 | 5600 reg_restore(&rp->rs_un.regsave, &backpos); |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5601 scan = OPERAND(rp->rs_scan) + 4; |
| 1579 | 5602 if (status == RA_MATCH) |
| 5603 { | |
| 5604 /* We did match, so subexpr may have been changed, | |
| 5605 * need to restore them for the next try. */ | |
| 5606 status = RA_NOMATCH; | |
| 5607 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5608 } | |
| 180 | 5609 } |
| 5610 else | |
| 5611 { | |
| 5612 /* Can't advance. For NOBEHIND that's a match. */ | |
| 5613 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5614 if (rp->rs_no == NOBEHIND) | |
| 5615 { | |
| 233 | 5616 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5617 &backpos); | |
| 180 | 5618 status = RA_MATCH; |
| 5619 } | |
| 5620 else | |
| 1579 | 5621 { |
| 5622 /* We do want a proper match. Need to restore the | |
| 5623 * subexpr if we had a match, because they may have | |
| 5624 * been set. */ | |
| 5625 if (status == RA_MATCH) | |
| 5626 { | |
| 5627 status = RA_NOMATCH; | |
| 5628 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5629 } | |
| 5630 } | |
| 270 | 5631 regstack_pop(&scan); |
| 180 | 5632 regstack.ga_len -= sizeof(regbehind_T); |
| 5633 } | |
| 5634 } | |
| 5635 break; | |
| 5636 | |
| 5637 case RS_STAR_LONG: | |
| 5638 case RS_STAR_SHORT: | |
| 5639 { | |
| 5640 regstar_T *rst = ((regstar_T *)rp) - 1; | |
| 5641 | |
| 5642 if (status == RA_MATCH) | |
| 5643 { | |
| 270 | 5644 regstack_pop(&scan); |
| 180 | 5645 regstack.ga_len -= sizeof(regstar_T); |
| 5646 break; | |
| 5647 } | |
| 5648 | |
| 5649 /* Tried once already, restore input pointers. */ | |
| 5650 if (status != RA_BREAK) | |
| 233 | 5651 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5652 |
| 5653 /* Repeat until we found a position where it could match. */ | |
| 5654 for (;;) | |
| 5655 { | |
| 5656 if (status != RA_BREAK) | |
| 5657 { | |
| 5658 /* Tried first position already, advance. */ | |
| 5659 if (rp->rs_state == RS_STAR_LONG) | |
| 5660 { | |
| 685 | 5661 /* Trying for longest match, but couldn't or |
| 5662 * didn't match -- back up one char. */ | |
| 180 | 5663 if (--rst->count < rst->minval) |
| 5664 break; | |
| 5665 if (reginput == regline) | |
| 5666 { | |
| 5667 /* backup to last char of previous line */ | |
| 5668 --reglnum; | |
| 5669 regline = reg_getline(reglnum); | |
| 5670 /* Just in case regrepeat() didn't count | |
| 5671 * right. */ | |
| 5672 if (regline == NULL) | |
| 5673 break; | |
| 5674 reginput = regline + STRLEN(regline); | |
| 5675 fast_breakcheck(); | |
| 5676 } | |
| 5677 else | |
| 5678 mb_ptr_back(regline, reginput); | |
| 5679 } | |
| 5680 else | |
| 5681 { | |
| 5682 /* Range is backwards, use shortest match first. | |
| 5683 * Careful: maxval and minval are exchanged! | |
| 5684 * Couldn't or didn't match: try advancing one | |
| 5685 * char. */ | |
| 5686 if (rst->count == rst->minval | |
| 5687 || regrepeat(OPERAND(rp->rs_scan), 1L) == 0) | |
| 5688 break; | |
| 5689 ++rst->count; | |
| 5690 } | |
| 5691 if (got_int) | |
| 5692 break; | |
| 5693 } | |
| 5694 else | |
| 5695 status = RA_NOMATCH; | |
| 5696 | |
| 5697 /* If it could match, try it. */ | |
| 5698 if (rst->nextb == NUL || *reginput == rst->nextb | |
| 5699 || *reginput == rst->nextb_ic) | |
| 5700 { | |
| 233 | 5701 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5702 scan = regnext(rp->rs_scan); |
| 5703 status = RA_CONT; | |
| 5704 break; | |
| 5705 } | |
| 5706 } | |
| 5707 if (status != RA_CONT) | |
| 5708 { | |
| 5709 /* Failed. */ | |
| 270 | 5710 regstack_pop(&scan); |
| 180 | 5711 regstack.ga_len -= sizeof(regstar_T); |
| 5712 status = RA_NOMATCH; | |
| 5713 } | |
| 5714 } | |
| 5715 break; | |
| 5716 } | |
| 5717 | |
| 685 | 5718 /* If we want to continue the inner loop or didn't pop a state |
| 5719 * continue matching loop */ | |
| 180 | 5720 if (status == RA_CONT || rp == (regitem_T *) |
| 5721 ((char *)regstack.ga_data + regstack.ga_len) - 1) | |
| 5722 break; | |
| 5723 } | |
| 5724 | |
| 189 | 5725 /* May need to continue with the inner loop, starting at "scan". */ |
| 180 | 5726 if (status == RA_CONT) |
| 5727 continue; | |
| 5728 | |
| 5729 /* | |
| 5730 * If the regstack is empty or something failed we are done. | |
| 5731 */ | |
| 5732 if (regstack.ga_len == 0 || status == RA_FAIL) | |
| 5733 { | |
| 5734 if (scan == NULL) | |
| 5735 { | |
| 5736 /* | |
| 5737 * We get here only if there's trouble -- normally "case END" is | |
| 5738 * the terminating point. | |
| 5739 */ | |
| 5740 EMSG(_(e_re_corr)); | |
| 7 | 5741 #ifdef DEBUG |
| 180 | 5742 printf("Premature EOL\n"); |
| 7 | 5743 #endif |
| 180 | 5744 } |
| 189 | 5745 if (status == RA_FAIL) |
| 5746 got_int = TRUE; | |
| 180 | 5747 return (status == RA_MATCH); |
| 5748 } | |
| 5749 | |
| 5750 } /* End of loop until the regstack is empty. */ | |
| 5751 | |
| 5752 /* NOTREACHED */ | |
| 5753 } | |
| 5754 | |
| 5755 /* | |
| 5756 * Push an item onto the regstack. | |
| 5757 * Returns pointer to new item. Returns NULL when out of memory. | |
| 5758 */ | |
| 5759 static regitem_T * | |
| 270 | 5760 regstack_push(state, scan) |
| 180 | 5761 regstate_T state; |
| 5762 char_u *scan; | |
| 5763 { | |
| 5764 regitem_T *rp; | |
| 5765 | |
| 270 | 5766 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5767 { |
| 5768 EMSG(_(e_maxmempat)); | |
| 5769 return NULL; | |
| 5770 } | |
| 270 | 5771 if (ga_grow(®stack, sizeof(regitem_T)) == FAIL) |
| 180 | 5772 return NULL; |
| 5773 | |
| 270 | 5774 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len); |
| 180 | 5775 rp->rs_state = state; |
| 5776 rp->rs_scan = scan; | |
| 5777 | |
| 270 | 5778 regstack.ga_len += sizeof(regitem_T); |
| 180 | 5779 return rp; |
| 5780 } | |
| 5781 | |
| 5782 /* | |
| 5783 * Pop an item from the regstack. | |
| 5784 */ | |
| 5785 static void | |
| 270 | 5786 regstack_pop(scan) |
| 180 | 5787 char_u **scan; |
| 5788 { | |
| 5789 regitem_T *rp; | |
| 5790 | |
| 270 | 5791 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; |
| 180 | 5792 *scan = rp->rs_scan; |
| 5793 | |
| 270 | 5794 regstack.ga_len -= sizeof(regitem_T); |
| 7 | 5795 } |
| 5796 | |
| 5797 /* | |
| 5798 * regrepeat - repeatedly match something simple, return how many. | |
| 5799 * Advances reginput (and reglnum) to just after the matched chars. | |
| 5800 */ | |
| 5801 static int | |
| 5802 regrepeat(p, maxcount) | |
| 5803 char_u *p; | |
| 5804 long maxcount; /* maximum number of matches allowed */ | |
| 5805 { | |
| 5806 long count = 0; | |
| 5807 char_u *scan; | |
| 5808 char_u *opnd; | |
| 5809 int mask; | |
| 5810 int testval = 0; | |
| 5811 | |
| 5812 scan = reginput; /* Make local copy of reginput for speed. */ | |
| 5813 opnd = OPERAND(p); | |
| 5814 switch (OP(p)) | |
| 5815 { | |
| 5816 case ANY: | |
| 5817 case ANY + ADD_NL: | |
| 5818 while (count < maxcount) | |
| 5819 { | |
| 5820 /* Matching anything means we continue until end-of-line (or | |
| 5821 * end-of-file for ANY + ADD_NL), only limited by maxcount. */ | |
| 5822 while (*scan != NUL && count < maxcount) | |
| 5823 { | |
| 5824 ++count; | |
| 39 | 5825 mb_ptr_adv(scan); |
| 7 | 5826 } |
| 1018 | 5827 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5828 || reg_line_lbr || count == maxcount) | |
| 7 | 5829 break; |
| 5830 ++count; /* count the line-break */ | |
| 5831 reg_nextline(); | |
| 5832 scan = reginput; | |
| 5833 if (got_int) | |
| 5834 break; | |
| 5835 } | |
| 5836 break; | |
| 5837 | |
| 5838 case IDENT: | |
| 5839 case IDENT + ADD_NL: | |
| 5840 testval = TRUE; | |
| 5841 /*FALLTHROUGH*/ | |
| 5842 case SIDENT: | |
| 5843 case SIDENT + ADD_NL: | |
| 5844 while (count < maxcount) | |
| 5845 { | |
| 4466 | 5846 if (vim_isIDc(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5847 { |
| 39 | 5848 mb_ptr_adv(scan); |
| 7 | 5849 } |
| 5850 else if (*scan == NUL) | |
| 5851 { | |
| 1018 | 5852 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5853 || reg_line_lbr) | |
| 7 | 5854 break; |
| 5855 reg_nextline(); | |
| 5856 scan = reginput; | |
| 5857 if (got_int) | |
| 5858 break; | |
| 5859 } | |
| 5860 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5861 ++scan; | |
| 5862 else | |
| 5863 break; | |
| 5864 ++count; | |
| 5865 } | |
| 5866 break; | |
| 5867 | |
| 5868 case KWORD: | |
| 5869 case KWORD + ADD_NL: | |
| 5870 testval = TRUE; | |
| 5871 /*FALLTHROUGH*/ | |
| 5872 case SKWORD: | |
| 5873 case SKWORD + ADD_NL: | |
| 5874 while (count < maxcount) | |
| 5875 { | |
| 4069 | 5876 if (vim_iswordp_buf(scan, reg_buf) |
| 5877 && (testval || !VIM_ISDIGIT(*scan))) | |
| 7 | 5878 { |
| 39 | 5879 mb_ptr_adv(scan); |
| 7 | 5880 } |
| 5881 else if (*scan == NUL) | |
| 5882 { | |
| 1018 | 5883 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5884 || reg_line_lbr) | |
| 7 | 5885 break; |
| 5886 reg_nextline(); | |
| 5887 scan = reginput; | |
| 5888 if (got_int) | |
| 5889 break; | |
| 5890 } | |
| 5891 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5892 ++scan; | |
| 5893 else | |
| 5894 break; | |
| 5895 ++count; | |
| 5896 } | |
| 5897 break; | |
| 5898 | |
| 5899 case FNAME: | |
| 5900 case FNAME + ADD_NL: | |
| 5901 testval = TRUE; | |
| 5902 /*FALLTHROUGH*/ | |
| 5903 case SFNAME: | |
| 5904 case SFNAME + ADD_NL: | |
| 5905 while (count < maxcount) | |
| 5906 { | |
| 4466 | 5907 if (vim_isfilec(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5908 { |
| 39 | 5909 mb_ptr_adv(scan); |
| 7 | 5910 } |
| 5911 else if (*scan == NUL) | |
| 5912 { | |
| 1018 | 5913 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5914 || reg_line_lbr) | |
| 7 | 5915 break; |
| 5916 reg_nextline(); | |
| 5917 scan = reginput; | |
| 5918 if (got_int) | |
| 5919 break; | |
| 5920 } | |
| 5921 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5922 ++scan; | |
| 5923 else | |
| 5924 break; | |
| 5925 ++count; | |
| 5926 } | |
| 5927 break; | |
| 5928 | |
| 5929 case PRINT: | |
| 5930 case PRINT + ADD_NL: | |
| 5931 testval = TRUE; | |
| 5932 /*FALLTHROUGH*/ | |
| 5933 case SPRINT: | |
| 5934 case SPRINT + ADD_NL: | |
| 5935 while (count < maxcount) | |
| 5936 { | |
| 5937 if (*scan == NUL) | |
| 5938 { | |
| 1018 | 5939 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5940 || reg_line_lbr) | |
| 7 | 5941 break; |
| 5942 reg_nextline(); | |
| 5943 scan = reginput; | |
| 5944 if (got_int) | |
| 5945 break; | |
| 5946 } | |
| 5947 else if (ptr2cells(scan) == 1 && (testval || !VIM_ISDIGIT(*scan))) | |
| 5948 { | |
| 39 | 5949 mb_ptr_adv(scan); |
| 7 | 5950 } |
| 5951 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5952 ++scan; | |
| 5953 else | |
| 5954 break; | |
| 5955 ++count; | |
| 5956 } | |
| 5957 break; | |
| 5958 | |
| 5959 case WHITE: | |
| 5960 case WHITE + ADD_NL: | |
| 5961 testval = mask = RI_WHITE; | |
| 5962 do_class: | |
| 5963 while (count < maxcount) | |
| 5964 { | |
| 5965 #ifdef FEAT_MBYTE | |
| 5966 int l; | |
| 5967 #endif | |
| 5968 if (*scan == NUL) | |
| 5969 { | |
| 1018 | 5970 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5971 || reg_line_lbr) | |
| 7 | 5972 break; |
| 5973 reg_nextline(); | |
| 5974 scan = reginput; | |
| 5975 if (got_int) | |
| 5976 break; | |
| 5977 } | |
| 5978 #ifdef FEAT_MBYTE | |
| 474 | 5979 else if (has_mbyte && (l = (*mb_ptr2len)(scan)) > 1) |
| 7 | 5980 { |
| 5981 if (testval != 0) | |
| 5982 break; | |
| 5983 scan += l; | |
| 5984 } | |
| 5985 #endif | |
| 5986 else if ((class_tab[*scan] & mask) == testval) | |
| 5987 ++scan; | |
| 5988 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5989 ++scan; | |
| 5990 else | |
| 5991 break; | |
| 5992 ++count; | |
| 5993 } | |
| 5994 break; | |
| 5995 | |
| 5996 case NWHITE: | |
| 5997 case NWHITE + ADD_NL: | |
| 5998 mask = RI_WHITE; | |
| 5999 goto do_class; | |
| 6000 case DIGIT: | |
| 6001 case DIGIT + ADD_NL: | |
| 6002 testval = mask = RI_DIGIT; | |
| 6003 goto do_class; | |
| 6004 case NDIGIT: | |
| 6005 case NDIGIT + ADD_NL: | |
| 6006 mask = RI_DIGIT; | |
| 6007 goto do_class; | |
| 6008 case HEX: | |
| 6009 case HEX + ADD_NL: | |
| 6010 testval = mask = RI_HEX; | |
| 6011 goto do_class; | |
| 6012 case NHEX: | |
| 6013 case NHEX + ADD_NL: | |
| 6014 mask = RI_HEX; | |
| 6015 goto do_class; | |
| 6016 case OCTAL: | |
| 6017 case OCTAL + ADD_NL: | |
| 6018 testval = mask = RI_OCTAL; | |
| 6019 goto do_class; | |
| 6020 case NOCTAL: | |
| 6021 case NOCTAL + ADD_NL: | |
| 6022 mask = RI_OCTAL; | |
| 6023 goto do_class; | |
| 6024 case WORD: | |
| 6025 case WORD + ADD_NL: | |
| 6026 testval = mask = RI_WORD; | |
| 6027 goto do_class; | |
| 6028 case NWORD: | |
| 6029 case NWORD + ADD_NL: | |
| 6030 mask = RI_WORD; | |
| 6031 goto do_class; | |
| 6032 case HEAD: | |
| 6033 case HEAD + ADD_NL: | |
| 6034 testval = mask = RI_HEAD; | |
| 6035 goto do_class; | |
| 6036 case NHEAD: | |
| 6037 case NHEAD + ADD_NL: | |
| 6038 mask = RI_HEAD; | |
| 6039 goto do_class; | |
| 6040 case ALPHA: | |
| 6041 case ALPHA + ADD_NL: | |
| 6042 testval = mask = RI_ALPHA; | |
| 6043 goto do_class; | |
| 6044 case NALPHA: | |
| 6045 case NALPHA + ADD_NL: | |
| 6046 mask = RI_ALPHA; | |
| 6047 goto do_class; | |
| 6048 case LOWER: | |
| 6049 case LOWER + ADD_NL: | |
| 6050 testval = mask = RI_LOWER; | |
| 6051 goto do_class; | |
| 6052 case NLOWER: | |
| 6053 case NLOWER + ADD_NL: | |
| 6054 mask = RI_LOWER; | |
| 6055 goto do_class; | |
| 6056 case UPPER: | |
| 6057 case UPPER + ADD_NL: | |
| 6058 testval = mask = RI_UPPER; | |
| 6059 goto do_class; | |
| 6060 case NUPPER: | |
| 6061 case NUPPER + ADD_NL: | |
| 6062 mask = RI_UPPER; | |
| 6063 goto do_class; | |
| 6064 | |
| 6065 case EXACTLY: | |
| 6066 { | |
| 6067 int cu, cl; | |
| 6068 | |
| 6069 /* This doesn't do a multi-byte character, because a MULTIBYTECODE | |
| 1347 | 6070 * would have been used for it. It does handle single-byte |
| 6071 * characters, such as latin1. */ | |
| 7 | 6072 if (ireg_ic) |
| 6073 { | |
| 1347 | 6074 cu = MB_TOUPPER(*opnd); |
| 6075 cl = MB_TOLOWER(*opnd); | |
| 7 | 6076 while (count < maxcount && (*scan == cu || *scan == cl)) |
| 6077 { | |
| 6078 count++; | |
| 6079 scan++; | |
| 6080 } | |
| 6081 } | |
| 6082 else | |
| 6083 { | |
| 6084 cu = *opnd; | |
| 6085 while (count < maxcount && *scan == cu) | |
| 6086 { | |
| 6087 count++; | |
| 6088 scan++; | |
| 6089 } | |
| 6090 } | |
| 6091 break; | |
| 6092 } | |
| 6093 | |
| 6094 #ifdef FEAT_MBYTE | |
| 6095 case MULTIBYTECODE: | |
| 6096 { | |
| 6097 int i, len, cf = 0; | |
| 6098 | |
| 6099 /* Safety check (just in case 'encoding' was changed since | |
| 6100 * compiling the program). */ | |
| 474 | 6101 if ((len = (*mb_ptr2len)(opnd)) > 1) |
| 7 | 6102 { |
| 6103 if (ireg_ic && enc_utf8) | |
| 6104 cf = utf_fold(utf_ptr2char(opnd)); | |
| 6105 while (count < maxcount) | |
| 6106 { | |
| 6107 for (i = 0; i < len; ++i) | |
| 6108 if (opnd[i] != scan[i]) | |
| 6109 break; | |
| 6110 if (i < len && (!ireg_ic || !enc_utf8 | |
| 6111 || utf_fold(utf_ptr2char(scan)) != cf)) | |
| 6112 break; | |
| 6113 scan += len; | |
| 6114 ++count; | |
| 6115 } | |
| 6116 } | |
| 6117 } | |
| 6118 break; | |
| 6119 #endif | |
| 6120 | |
| 6121 case ANYOF: | |
| 6122 case ANYOF + ADD_NL: | |
| 6123 testval = TRUE; | |
| 6124 /*FALLTHROUGH*/ | |
| 6125 | |
| 6126 case ANYBUT: | |
| 6127 case ANYBUT + ADD_NL: | |
| 6128 while (count < maxcount) | |
| 6129 { | |
| 6130 #ifdef FEAT_MBYTE | |
| 6131 int len; | |
| 6132 #endif | |
| 6133 if (*scan == NUL) | |
| 6134 { | |
| 1018 | 6135 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 6136 || reg_line_lbr) | |
| 7 | 6137 break; |
| 6138 reg_nextline(); | |
| 6139 scan = reginput; | |
| 6140 if (got_int) | |
| 6141 break; | |
| 6142 } | |
| 6143 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 6144 ++scan; | |
| 6145 #ifdef FEAT_MBYTE | |
| 474 | 6146 else if (has_mbyte && (len = (*mb_ptr2len)(scan)) > 1) |
| 7 | 6147 { |
| 6148 if ((cstrchr(opnd, (*mb_ptr2char)(scan)) == NULL) == testval) | |
| 6149 break; | |
| 6150 scan += len; | |
| 6151 } | |
| 6152 #endif | |
| 6153 else | |
| 6154 { | |
| 6155 if ((cstrchr(opnd, *scan) == NULL) == testval) | |
| 6156 break; | |
| 6157 ++scan; | |
| 6158 } | |
| 6159 ++count; | |
| 6160 } | |
| 6161 break; | |
| 6162 | |
| 6163 case NEWL: | |
| 6164 while (count < maxcount | |
| 1018 | 6165 && ((*scan == NUL && reglnum <= reg_maxline && !reg_line_lbr |
| 6166 && REG_MULTI) || (*scan == '\n' && reg_line_lbr))) | |
| 7 | 6167 { |
| 6168 count++; | |
| 6169 if (reg_line_lbr) | |
| 6170 ADVANCE_REGINPUT(); | |
| 6171 else | |
| 6172 reg_nextline(); | |
| 6173 scan = reginput; | |
| 6174 if (got_int) | |
| 6175 break; | |
| 6176 } | |
| 6177 break; | |
| 6178 | |
| 6179 default: /* Oh dear. Called inappropriately. */ | |
| 6180 EMSG(_(e_re_corr)); | |
| 6181 #ifdef DEBUG | |
| 6182 printf("Called regrepeat with op code %d\n", OP(p)); | |
| 6183 #endif | |
| 6184 break; | |
| 6185 } | |
| 6186 | |
| 6187 reginput = scan; | |
| 6188 | |
| 6189 return (int)count; | |
| 6190 } | |
| 6191 | |
| 6192 /* | |
| 6193 * regnext - dig the "next" pointer out of a node | |
| 2010 | 6194 * Returns NULL when calculating size, when there is no next item and when |
| 6195 * there is an error. | |
| 7 | 6196 */ |
| 6197 static char_u * | |
| 6198 regnext(p) | |
| 6199 char_u *p; | |
| 6200 { | |
| 6201 int offset; | |
| 6202 | |
| 2010 | 6203 if (p == JUST_CALC_SIZE || reg_toolong) |
| 7 | 6204 return NULL; |
| 6205 | |
| 6206 offset = NEXT(p); | |
| 6207 if (offset == 0) | |
| 6208 return NULL; | |
| 6209 | |
| 233 | 6210 if (OP(p) == BACK) |
| 7 | 6211 return p - offset; |
| 6212 else | |
| 6213 return p + offset; | |
| 6214 } | |
| 6215 | |
| 6216 /* | |
| 6217 * Check the regexp program for its magic number. | |
| 6218 * Return TRUE if it's wrong. | |
| 6219 */ | |
| 6220 static int | |
| 6221 prog_magic_wrong() | |
| 6222 { | |
| 4444 | 6223 regprog_T *prog; |
| 6224 | |
| 6225 prog = REG_MULTI ? reg_mmatch->regprog : reg_match->regprog; | |
| 6226 if (prog->engine == &nfa_regengine) | |
| 6227 /* For NFA matcher we don't check the magic */ | |
| 6228 return FALSE; | |
| 6229 | |
| 6230 if (UCHARAT(((bt_regprog_T *)prog)->program) != REGMAGIC) | |
| 7 | 6231 { |
| 6232 EMSG(_(e_re_corr)); | |
| 6233 return TRUE; | |
| 6234 } | |
| 6235 return FALSE; | |
| 6236 } | |
| 6237 | |
| 6238 /* | |
| 6239 * Cleanup the subexpressions, if this wasn't done yet. | |
| 6240 * This construction is used to clear the subexpressions only when they are | |
| 6241 * used (to increase speed). | |
| 6242 */ | |
| 6243 static void | |
| 6244 cleanup_subexpr() | |
| 6245 { | |
| 6246 if (need_clear_subexpr) | |
| 6247 { | |
| 6248 if (REG_MULTI) | |
| 6249 { | |
| 6250 /* Use 0xff to set lnum to -1 */ | |
| 6251 vim_memset(reg_startpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6252 vim_memset(reg_endpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6253 } | |
| 6254 else | |
| 6255 { | |
| 6256 vim_memset(reg_startp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6257 vim_memset(reg_endp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6258 } | |
| 6259 need_clear_subexpr = FALSE; | |
| 6260 } | |
| 6261 } | |
| 6262 | |
| 6263 #ifdef FEAT_SYN_HL | |
| 6264 static void | |
| 6265 cleanup_zsubexpr() | |
| 6266 { | |
| 6267 if (need_clear_zsubexpr) | |
| 6268 { | |
| 6269 if (REG_MULTI) | |
| 6270 { | |
| 6271 /* Use 0xff to set lnum to -1 */ | |
| 6272 vim_memset(reg_startzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6273 vim_memset(reg_endzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6274 } | |
| 6275 else | |
| 6276 { | |
| 6277 vim_memset(reg_startzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6278 vim_memset(reg_endzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6279 } | |
| 6280 need_clear_zsubexpr = FALSE; | |
| 6281 } | |
| 6282 } | |
| 6283 #endif | |
| 6284 | |
| 6285 /* | |
| 1579 | 6286 * Save the current subexpr to "bp", so that they can be restored |
| 6287 * later by restore_subexpr(). | |
| 6288 */ | |
| 6289 static void | |
| 6290 save_subexpr(bp) | |
| 6291 regbehind_T *bp; | |
| 6292 { | |
| 6293 int i; | |
| 6294 | |
| 1602 | 6295 /* When "need_clear_subexpr" is set we don't need to save the values, only |
| 6296 * remember that this flag needs to be set again when restoring. */ | |
| 6297 bp->save_need_clear_subexpr = need_clear_subexpr; | |
| 6298 if (!need_clear_subexpr) | |
| 1579 | 6299 { |
| 1602 | 6300 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6301 { |
| 1602 | 6302 if (REG_MULTI) |
| 6303 { | |
| 6304 bp->save_start[i].se_u.pos = reg_startpos[i]; | |
| 6305 bp->save_end[i].se_u.pos = reg_endpos[i]; | |
| 6306 } | |
| 6307 else | |
| 6308 { | |
| 6309 bp->save_start[i].se_u.ptr = reg_startp[i]; | |
| 6310 bp->save_end[i].se_u.ptr = reg_endp[i]; | |
| 6311 } | |
| 1579 | 6312 } |
| 6313 } | |
| 6314 } | |
| 6315 | |
| 6316 /* | |
| 6317 * Restore the subexpr from "bp". | |
| 6318 */ | |
| 6319 static void | |
| 6320 restore_subexpr(bp) | |
| 6321 regbehind_T *bp; | |
| 6322 { | |
| 6323 int i; | |
| 6324 | |
| 1602 | 6325 /* Only need to restore saved values when they are not to be cleared. */ |
| 6326 need_clear_subexpr = bp->save_need_clear_subexpr; | |
| 6327 if (!need_clear_subexpr) | |
| 1579 | 6328 { |
| 1602 | 6329 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6330 { |
| 1602 | 6331 if (REG_MULTI) |
| 6332 { | |
| 6333 reg_startpos[i] = bp->save_start[i].se_u.pos; | |
| 6334 reg_endpos[i] = bp->save_end[i].se_u.pos; | |
| 6335 } | |
| 6336 else | |
| 6337 { | |
| 6338 reg_startp[i] = bp->save_start[i].se_u.ptr; | |
| 6339 reg_endp[i] = bp->save_end[i].se_u.ptr; | |
| 6340 } | |
| 1579 | 6341 } |
| 6342 } | |
| 6343 } | |
| 6344 | |
| 6345 /* | |
| 7 | 6346 * Advance reglnum, regline and reginput to the next line. |
| 6347 */ | |
| 6348 static void | |
| 6349 reg_nextline() | |
| 6350 { | |
| 6351 regline = reg_getline(++reglnum); | |
| 6352 reginput = regline; | |
| 6353 fast_breakcheck(); | |
| 6354 } | |
| 6355 | |
| 6356 /* | |
| 6357 * Save the input line and position in a regsave_T. | |
| 6358 */ | |
| 6359 static void | |
| 233 | 6360 reg_save(save, gap) |
| 7 | 6361 regsave_T *save; |
| 233 | 6362 garray_T *gap; |
| 7 | 6363 { |
| 6364 if (REG_MULTI) | |
| 6365 { | |
| 6366 save->rs_u.pos.col = (colnr_T)(reginput - regline); | |
| 6367 save->rs_u.pos.lnum = reglnum; | |
| 6368 } | |
| 6369 else | |
| 6370 save->rs_u.ptr = reginput; | |
| 233 | 6371 save->rs_len = gap->ga_len; |
| 7 | 6372 } |
| 6373 | |
| 6374 /* | |
| 6375 * Restore the input line and position from a regsave_T. | |
| 6376 */ | |
| 6377 static void | |
| 233 | 6378 reg_restore(save, gap) |
| 7 | 6379 regsave_T *save; |
| 233 | 6380 garray_T *gap; |
| 7 | 6381 { |
| 6382 if (REG_MULTI) | |
| 6383 { | |
| 6384 if (reglnum != save->rs_u.pos.lnum) | |
| 6385 { | |
| 6386 /* only call reg_getline() when the line number changed to save | |
| 6387 * a bit of time */ | |
| 6388 reglnum = save->rs_u.pos.lnum; | |
| 6389 regline = reg_getline(reglnum); | |
| 6390 } | |
| 6391 reginput = regline + save->rs_u.pos.col; | |
| 6392 } | |
| 6393 else | |
| 6394 reginput = save->rs_u.ptr; | |
| 233 | 6395 gap->ga_len = save->rs_len; |
| 7 | 6396 } |
| 6397 | |
| 6398 /* | |
| 6399 * Return TRUE if current position is equal to saved position. | |
| 6400 */ | |
| 6401 static int | |
| 6402 reg_save_equal(save) | |
| 6403 regsave_T *save; | |
| 6404 { | |
| 6405 if (REG_MULTI) | |
| 6406 return reglnum == save->rs_u.pos.lnum | |
| 6407 && reginput == regline + save->rs_u.pos.col; | |
| 6408 return reginput == save->rs_u.ptr; | |
| 6409 } | |
| 6410 | |
| 6411 /* | |
| 6412 * Tentatively set the sub-expression start to the current position (after | |
| 6413 * calling regmatch() they will have changed). Need to save the existing | |
| 6414 * values for when there is no match. | |
| 6415 * Use se_save() to use pointer (save_se_multi()) or position (save_se_one()), | |
| 6416 * depending on REG_MULTI. | |
| 6417 */ | |
| 6418 static void | |
| 6419 save_se_multi(savep, posp) | |
| 6420 save_se_T *savep; | |
| 6421 lpos_T *posp; | |
| 6422 { | |
| 6423 savep->se_u.pos = *posp; | |
| 6424 posp->lnum = reglnum; | |
| 6425 posp->col = (colnr_T)(reginput - regline); | |
| 6426 } | |
| 6427 | |
| 6428 static void | |
| 6429 save_se_one(savep, pp) | |
| 6430 save_se_T *savep; | |
| 6431 char_u **pp; | |
| 6432 { | |
| 6433 savep->se_u.ptr = *pp; | |
| 6434 *pp = reginput; | |
| 6435 } | |
| 6436 | |
| 6437 /* | |
| 6438 * Compare a number with the operand of RE_LNUM, RE_COL or RE_VCOL. | |
| 6439 */ | |
| 6440 static int | |
| 6441 re_num_cmp(val, scan) | |
| 6442 long_u val; | |
| 6443 char_u *scan; | |
| 6444 { | |
| 6445 long_u n = OPERAND_MIN(scan); | |
| 6446 | |
| 6447 if (OPERAND_CMP(scan) == '>') | |
| 6448 return val > n; | |
| 6449 if (OPERAND_CMP(scan) == '<') | |
| 6450 return val < n; | |
| 6451 return val == n; | |
| 6452 } | |
| 6453 | |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6454 /* |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6455 * Check whether a backreference matches. |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6456 * Returns RA_FAIL, RA_NOMATCH or RA_MATCH. |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6457 * If "bytelen" is not NULL, it is set to the bytelength of the whole match. |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6458 */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6459 static int |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6460 match_with_backref(start_lnum, start_col, end_lnum, end_col, bytelen) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6461 linenr_T start_lnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6462 colnr_T start_col; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6463 linenr_T end_lnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6464 colnr_T end_col; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6465 int *bytelen; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6466 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6467 linenr_T clnum = start_lnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6468 colnr_T ccol = start_col; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6469 int len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6470 char_u *p; |
|
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 if (bytelen != NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6473 *bytelen = 0; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6474 for (;;) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6475 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6476 /* 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
|
6477 * Slow! */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6478 if (regline != reg_tofree) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6479 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6480 len = (int)STRLEN(regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6481 if (reg_tofree == NULL || len >= (int)reg_tofreelen) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6482 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6483 len += 50; /* get some extra */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6484 vim_free(reg_tofree); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6485 reg_tofree = alloc(len); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6486 if (reg_tofree == NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6487 return RA_FAIL; /* out of memory!*/ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6488 reg_tofreelen = len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6489 } |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6490 STRCPY(reg_tofree, regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6491 reginput = reg_tofree + (reginput - regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6492 regline = reg_tofree; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6493 } |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6494 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6495 /* Get the line to compare with. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6496 p = reg_getline(clnum); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6497 if (clnum == end_lnum) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6498 len = end_col - ccol; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6499 else |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6500 len = (int)STRLEN(p + ccol); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6501 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6502 if (cstrncmp(p + ccol, reginput, &len) != 0) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6503 return RA_NOMATCH; /* doesn't match */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6504 if (bytelen != NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6505 *bytelen += len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6506 if (clnum == end_lnum) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6507 break; /* match and at end! */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6508 if (reglnum >= reg_maxline) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6509 return RA_NOMATCH; /* text too short */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6510 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6511 /* Advance to next line. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6512 reg_nextline(); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6513 ++clnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6514 ccol = 0; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6515 if (got_int) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6516 return RA_FAIL; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6517 } |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6518 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6519 /* 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
|
6520 * that should not matter. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6521 return RA_MATCH; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6522 } |
| 7 | 6523 |
| 4444 | 6524 #ifdef BT_REGEXP_DUMP |
| 7 | 6525 |
| 6526 /* | |
| 6527 * regdump - dump a regexp onto stdout in vaguely comprehensible form | |
| 6528 */ | |
| 6529 static void | |
| 6530 regdump(pattern, r) | |
| 6531 char_u *pattern; | |
| 4444 | 6532 bt_regprog_T *r; |
| 7 | 6533 { |
| 6534 char_u *s; | |
| 6535 int op = EXACTLY; /* Arbitrary non-END op. */ | |
| 6536 char_u *next; | |
| 6537 char_u *end = NULL; | |
| 4444 | 6538 FILE *f; |
| 6539 | |
| 6540 #ifdef BT_REGEXP_LOG | |
| 6541 f = fopen("bt_regexp_log.log", "a"); | |
| 6542 #else | |
| 6543 f = stdout; | |
| 6544 #endif | |
| 6545 if (f == NULL) | |
| 6546 return; | |
| 6547 fprintf(f, "-------------------------------------\n\r\nregcomp(%s):\r\n", pattern); | |
| 7 | 6548 |
| 6549 s = r->program + 1; | |
| 6550 /* | |
| 6551 * Loop until we find the END that isn't before a referred next (an END | |
| 6552 * can also appear in a NOMATCH operand). | |
| 6553 */ | |
| 6554 while (op != END || s <= end) | |
| 6555 { | |
| 6556 op = OP(s); | |
| 4444 | 6557 fprintf(f, "%2d%s", (int)(s - r->program), regprop(s)); /* Where, what. */ |
| 7 | 6558 next = regnext(s); |
| 6559 if (next == NULL) /* Next ptr. */ | |
| 4444 | 6560 fprintf(f, "(0)"); |
| 7 | 6561 else |
| 4444 | 6562 fprintf(f, "(%d)", (int)((s - r->program) + (next - s))); |
| 7 | 6563 if (end < next) |
| 6564 end = next; | |
| 6565 if (op == BRACE_LIMITS) | |
| 6566 { | |
|
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6567 /* Two ints */ |
| 4444 | 6568 fprintf(f, " minval %ld, maxval %ld", OPERAND_MIN(s), OPERAND_MAX(s)); |
| 7 | 6569 s += 8; |
| 6570 } | |
|
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6571 else if (op == BEHIND || op == NOBEHIND) |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6572 { |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6573 /* one int */ |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6574 fprintf(f, " count %ld", OPERAND_MIN(s)); |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6575 s += 4; |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6576 } |
|
4770
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6577 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
|
6578 { |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6579 /* one int plus comperator */ |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6580 fprintf(f, " count %ld", OPERAND_MIN(s)); |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6581 s += 5; |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6582 } |
| 7 | 6583 s += 3; |
| 6584 if (op == ANYOF || op == ANYOF + ADD_NL | |
| 6585 || op == ANYBUT || op == ANYBUT + ADD_NL | |
| 6586 || op == EXACTLY) | |
| 6587 { | |
| 6588 /* Literal string, where present. */ | |
| 4444 | 6589 fprintf(f, "\nxxxxxxxxx\n"); |
| 7 | 6590 while (*s != NUL) |
| 4444 | 6591 fprintf(f, "%c", *s++); |
| 6592 fprintf(f, "\nxxxxxxxxx\n"); | |
| 7 | 6593 s++; |
| 6594 } | |
| 4444 | 6595 fprintf(f, "\r\n"); |
| 7 | 6596 } |
| 6597 | |
| 6598 /* Header fields of interest. */ | |
| 6599 if (r->regstart != NUL) | |
| 4444 | 6600 fprintf(f, "start `%s' 0x%x; ", r->regstart < 256 |
| 7 | 6601 ? (char *)transchar(r->regstart) |
| 6602 : "multibyte", r->regstart); | |
| 6603 if (r->reganch) | |
| 4444 | 6604 fprintf(f, "anchored; "); |
| 7 | 6605 if (r->regmust != NULL) |
| 4444 | 6606 fprintf(f, "must have \"%s\"", r->regmust); |
| 6607 fprintf(f, "\r\n"); | |
| 6608 | |
| 6609 #ifdef BT_REGEXP_LOG | |
| 6610 fclose(f); | |
| 6611 #endif | |
| 7 | 6612 } |
| 4444 | 6613 #endif /* BT_REGEXP_DUMP */ |
| 6614 | |
| 6615 #ifdef DEBUG | |
| 7 | 6616 /* |
| 6617 * regprop - printable representation of opcode | |
| 6618 */ | |
| 6619 static char_u * | |
| 6620 regprop(op) | |
| 6621 char_u *op; | |
| 6622 { | |
| 4444 | 6623 char *p; |
| 6624 static char buf[50]; | |
| 6625 | |
| 6626 STRCPY(buf, ":"); | |
| 6627 | |
| 6628 switch ((int) OP(op)) | |
| 7 | 6629 { |
| 6630 case BOL: | |
| 6631 p = "BOL"; | |
| 6632 break; | |
| 6633 case EOL: | |
| 6634 p = "EOL"; | |
| 6635 break; | |
| 6636 case RE_BOF: | |
| 6637 p = "BOF"; | |
| 6638 break; | |
| 6639 case RE_EOF: | |
| 6640 p = "EOF"; | |
| 6641 break; | |
| 6642 case CURSOR: | |
| 6643 p = "CURSOR"; | |
| 6644 break; | |
| 639 | 6645 case RE_VISUAL: |
| 6646 p = "RE_VISUAL"; | |
| 6647 break; | |
| 7 | 6648 case RE_LNUM: |
| 6649 p = "RE_LNUM"; | |
| 6650 break; | |
| 639 | 6651 case RE_MARK: |
| 6652 p = "RE_MARK"; | |
| 6653 break; | |
| 7 | 6654 case RE_COL: |
| 6655 p = "RE_COL"; | |
| 6656 break; | |
| 6657 case RE_VCOL: | |
| 6658 p = "RE_VCOL"; | |
| 6659 break; | |
| 6660 case BOW: | |
| 6661 p = "BOW"; | |
| 6662 break; | |
| 6663 case EOW: | |
| 6664 p = "EOW"; | |
| 6665 break; | |
| 6666 case ANY: | |
| 6667 p = "ANY"; | |
| 6668 break; | |
| 6669 case ANY + ADD_NL: | |
| 6670 p = "ANY+NL"; | |
| 6671 break; | |
| 6672 case ANYOF: | |
| 6673 p = "ANYOF"; | |
| 6674 break; | |
| 6675 case ANYOF + ADD_NL: | |
| 6676 p = "ANYOF+NL"; | |
| 6677 break; | |
| 6678 case ANYBUT: | |
| 6679 p = "ANYBUT"; | |
| 6680 break; | |
| 6681 case ANYBUT + ADD_NL: | |
| 6682 p = "ANYBUT+NL"; | |
| 6683 break; | |
| 6684 case IDENT: | |
| 6685 p = "IDENT"; | |
| 6686 break; | |
| 6687 case IDENT + ADD_NL: | |
| 6688 p = "IDENT+NL"; | |
| 6689 break; | |
| 6690 case SIDENT: | |
| 6691 p = "SIDENT"; | |
| 6692 break; | |
| 6693 case SIDENT + ADD_NL: | |
| 6694 p = "SIDENT+NL"; | |
| 6695 break; | |
| 6696 case KWORD: | |
| 6697 p = "KWORD"; | |
| 6698 break; | |
| 6699 case KWORD + ADD_NL: | |
| 6700 p = "KWORD+NL"; | |
| 6701 break; | |
| 6702 case SKWORD: | |
| 6703 p = "SKWORD"; | |
| 6704 break; | |
| 6705 case SKWORD + ADD_NL: | |
| 6706 p = "SKWORD+NL"; | |
| 6707 break; | |
| 6708 case FNAME: | |
| 6709 p = "FNAME"; | |
| 6710 break; | |
| 6711 case FNAME + ADD_NL: | |
| 6712 p = "FNAME+NL"; | |
| 6713 break; | |
| 6714 case SFNAME: | |
| 6715 p = "SFNAME"; | |
| 6716 break; | |
| 6717 case SFNAME + ADD_NL: | |
| 6718 p = "SFNAME+NL"; | |
| 6719 break; | |
| 6720 case PRINT: | |
| 6721 p = "PRINT"; | |
| 6722 break; | |
| 6723 case PRINT + ADD_NL: | |
| 6724 p = "PRINT+NL"; | |
| 6725 break; | |
| 6726 case SPRINT: | |
| 6727 p = "SPRINT"; | |
| 6728 break; | |
| 6729 case SPRINT + ADD_NL: | |
| 6730 p = "SPRINT+NL"; | |
| 6731 break; | |
| 6732 case WHITE: | |
| 6733 p = "WHITE"; | |
| 6734 break; | |
| 6735 case WHITE + ADD_NL: | |
| 6736 p = "WHITE+NL"; | |
| 6737 break; | |
| 6738 case NWHITE: | |
| 6739 p = "NWHITE"; | |
| 6740 break; | |
| 6741 case NWHITE + ADD_NL: | |
| 6742 p = "NWHITE+NL"; | |
| 6743 break; | |
| 6744 case DIGIT: | |
| 6745 p = "DIGIT"; | |
| 6746 break; | |
| 6747 case DIGIT + ADD_NL: | |
| 6748 p = "DIGIT+NL"; | |
| 6749 break; | |
| 6750 case NDIGIT: | |
| 6751 p = "NDIGIT"; | |
| 6752 break; | |
| 6753 case NDIGIT + ADD_NL: | |
| 6754 p = "NDIGIT+NL"; | |
| 6755 break; | |
| 6756 case HEX: | |
| 6757 p = "HEX"; | |
| 6758 break; | |
| 6759 case HEX + ADD_NL: | |
| 6760 p = "HEX+NL"; | |
| 6761 break; | |
| 6762 case NHEX: | |
| 6763 p = "NHEX"; | |
| 6764 break; | |
| 6765 case NHEX + ADD_NL: | |
| 6766 p = "NHEX+NL"; | |
| 6767 break; | |
| 6768 case OCTAL: | |
| 6769 p = "OCTAL"; | |
| 6770 break; | |
| 6771 case OCTAL + ADD_NL: | |
| 6772 p = "OCTAL+NL"; | |
| 6773 break; | |
| 6774 case NOCTAL: | |
| 6775 p = "NOCTAL"; | |
| 6776 break; | |
| 6777 case NOCTAL + ADD_NL: | |
| 6778 p = "NOCTAL+NL"; | |
| 6779 break; | |
| 6780 case WORD: | |
| 6781 p = "WORD"; | |
| 6782 break; | |
| 6783 case WORD + ADD_NL: | |
| 6784 p = "WORD+NL"; | |
| 6785 break; | |
| 6786 case NWORD: | |
| 6787 p = "NWORD"; | |
| 6788 break; | |
| 6789 case NWORD + ADD_NL: | |
| 6790 p = "NWORD+NL"; | |
| 6791 break; | |
| 6792 case HEAD: | |
| 6793 p = "HEAD"; | |
| 6794 break; | |
| 6795 case HEAD + ADD_NL: | |
| 6796 p = "HEAD+NL"; | |
| 6797 break; | |
| 6798 case NHEAD: | |
| 6799 p = "NHEAD"; | |
| 6800 break; | |
| 6801 case NHEAD + ADD_NL: | |
| 6802 p = "NHEAD+NL"; | |
| 6803 break; | |
| 6804 case ALPHA: | |
| 6805 p = "ALPHA"; | |
| 6806 break; | |
| 6807 case ALPHA + ADD_NL: | |
| 6808 p = "ALPHA+NL"; | |
| 6809 break; | |
| 6810 case NALPHA: | |
| 6811 p = "NALPHA"; | |
| 6812 break; | |
| 6813 case NALPHA + ADD_NL: | |
| 6814 p = "NALPHA+NL"; | |
| 6815 break; | |
| 6816 case LOWER: | |
| 6817 p = "LOWER"; | |
| 6818 break; | |
| 6819 case LOWER + ADD_NL: | |
| 6820 p = "LOWER+NL"; | |
| 6821 break; | |
| 6822 case NLOWER: | |
| 6823 p = "NLOWER"; | |
| 6824 break; | |
| 6825 case NLOWER + ADD_NL: | |
| 6826 p = "NLOWER+NL"; | |
| 6827 break; | |
| 6828 case UPPER: | |
| 6829 p = "UPPER"; | |
| 6830 break; | |
| 6831 case UPPER + ADD_NL: | |
| 6832 p = "UPPER+NL"; | |
| 6833 break; | |
| 6834 case NUPPER: | |
| 6835 p = "NUPPER"; | |
| 6836 break; | |
| 6837 case NUPPER + ADD_NL: | |
| 6838 p = "NUPPER+NL"; | |
| 6839 break; | |
| 6840 case BRANCH: | |
| 6841 p = "BRANCH"; | |
| 6842 break; | |
| 6843 case EXACTLY: | |
| 6844 p = "EXACTLY"; | |
| 6845 break; | |
| 6846 case NOTHING: | |
| 6847 p = "NOTHING"; | |
| 6848 break; | |
| 6849 case BACK: | |
| 6850 p = "BACK"; | |
| 6851 break; | |
| 6852 case END: | |
| 6853 p = "END"; | |
| 6854 break; | |
| 6855 case MOPEN + 0: | |
| 6856 p = "MATCH START"; | |
| 6857 break; | |
| 6858 case MOPEN + 1: | |
| 6859 case MOPEN + 2: | |
| 6860 case MOPEN + 3: | |
| 6861 case MOPEN + 4: | |
| 6862 case MOPEN + 5: | |
| 6863 case MOPEN + 6: | |
| 6864 case MOPEN + 7: | |
| 6865 case MOPEN + 8: | |
| 6866 case MOPEN + 9: | |
| 6867 sprintf(buf + STRLEN(buf), "MOPEN%d", OP(op) - MOPEN); | |
| 6868 p = NULL; | |
| 6869 break; | |
| 6870 case MCLOSE + 0: | |
| 6871 p = "MATCH END"; | |
| 6872 break; | |
| 6873 case MCLOSE + 1: | |
| 6874 case MCLOSE + 2: | |
| 6875 case MCLOSE + 3: | |
| 6876 case MCLOSE + 4: | |
| 6877 case MCLOSE + 5: | |
| 6878 case MCLOSE + 6: | |
| 6879 case MCLOSE + 7: | |
| 6880 case MCLOSE + 8: | |
| 6881 case MCLOSE + 9: | |
| 6882 sprintf(buf + STRLEN(buf), "MCLOSE%d", OP(op) - MCLOSE); | |
| 6883 p = NULL; | |
| 6884 break; | |
| 6885 case BACKREF + 1: | |
| 6886 case BACKREF + 2: | |
| 6887 case BACKREF + 3: | |
| 6888 case BACKREF + 4: | |
| 6889 case BACKREF + 5: | |
| 6890 case BACKREF + 6: | |
| 6891 case BACKREF + 7: | |
| 6892 case BACKREF + 8: | |
| 6893 case BACKREF + 9: | |
| 6894 sprintf(buf + STRLEN(buf), "BACKREF%d", OP(op) - BACKREF); | |
| 6895 p = NULL; | |
| 6896 break; | |
| 6897 case NOPEN: | |
| 6898 p = "NOPEN"; | |
| 6899 break; | |
| 6900 case NCLOSE: | |
| 6901 p = "NCLOSE"; | |
| 6902 break; | |
| 6903 #ifdef FEAT_SYN_HL | |
| 6904 case ZOPEN + 1: | |
| 6905 case ZOPEN + 2: | |
| 6906 case ZOPEN + 3: | |
| 6907 case ZOPEN + 4: | |
| 6908 case ZOPEN + 5: | |
| 6909 case ZOPEN + 6: | |
| 6910 case ZOPEN + 7: | |
| 6911 case ZOPEN + 8: | |
| 6912 case ZOPEN + 9: | |
| 6913 sprintf(buf + STRLEN(buf), "ZOPEN%d", OP(op) - ZOPEN); | |
| 6914 p = NULL; | |
| 6915 break; | |
| 6916 case ZCLOSE + 1: | |
| 6917 case ZCLOSE + 2: | |
| 6918 case ZCLOSE + 3: | |
| 6919 case ZCLOSE + 4: | |
| 6920 case ZCLOSE + 5: | |
| 6921 case ZCLOSE + 6: | |
| 6922 case ZCLOSE + 7: | |
| 6923 case ZCLOSE + 8: | |
| 6924 case ZCLOSE + 9: | |
| 6925 sprintf(buf + STRLEN(buf), "ZCLOSE%d", OP(op) - ZCLOSE); | |
| 6926 p = NULL; | |
| 6927 break; | |
| 6928 case ZREF + 1: | |
| 6929 case ZREF + 2: | |
| 6930 case ZREF + 3: | |
| 6931 case ZREF + 4: | |
| 6932 case ZREF + 5: | |
| 6933 case ZREF + 6: | |
| 6934 case ZREF + 7: | |
| 6935 case ZREF + 8: | |
| 6936 case ZREF + 9: | |
| 6937 sprintf(buf + STRLEN(buf), "ZREF%d", OP(op) - ZREF); | |
| 6938 p = NULL; | |
| 6939 break; | |
| 6940 #endif | |
| 6941 case STAR: | |
| 6942 p = "STAR"; | |
| 6943 break; | |
| 6944 case PLUS: | |
| 6945 p = "PLUS"; | |
| 6946 break; | |
| 6947 case NOMATCH: | |
| 6948 p = "NOMATCH"; | |
| 6949 break; | |
| 6950 case MATCH: | |
| 6951 p = "MATCH"; | |
| 6952 break; | |
| 6953 case BEHIND: | |
| 6954 p = "BEHIND"; | |
| 6955 break; | |
| 6956 case NOBEHIND: | |
| 6957 p = "NOBEHIND"; | |
| 6958 break; | |
| 6959 case SUBPAT: | |
| 6960 p = "SUBPAT"; | |
| 6961 break; | |
| 6962 case BRACE_LIMITS: | |
| 6963 p = "BRACE_LIMITS"; | |
| 6964 break; | |
| 6965 case BRACE_SIMPLE: | |
| 6966 p = "BRACE_SIMPLE"; | |
| 6967 break; | |
| 6968 case BRACE_COMPLEX + 0: | |
| 6969 case BRACE_COMPLEX + 1: | |
| 6970 case BRACE_COMPLEX + 2: | |
| 6971 case BRACE_COMPLEX + 3: | |
| 6972 case BRACE_COMPLEX + 4: | |
| 6973 case BRACE_COMPLEX + 5: | |
| 6974 case BRACE_COMPLEX + 6: | |
| 6975 case BRACE_COMPLEX + 7: | |
| 6976 case BRACE_COMPLEX + 8: | |
| 6977 case BRACE_COMPLEX + 9: | |
| 6978 sprintf(buf + STRLEN(buf), "BRACE_COMPLEX%d", OP(op) - BRACE_COMPLEX); | |
| 6979 p = NULL; | |
| 6980 break; | |
| 6981 #ifdef FEAT_MBYTE | |
| 6982 case MULTIBYTECODE: | |
| 6983 p = "MULTIBYTECODE"; | |
| 6984 break; | |
| 6985 #endif | |
| 6986 case NEWL: | |
| 6987 p = "NEWL"; | |
| 6988 break; | |
| 6989 default: | |
| 6990 sprintf(buf + STRLEN(buf), "corrupt %d", OP(op)); | |
| 6991 p = NULL; | |
| 6992 break; | |
| 6993 } | |
| 6994 if (p != NULL) | |
| 4444 | 6995 STRCAT(buf, p); |
| 6996 return (char_u *)buf; | |
| 7 | 6997 } |
| 4444 | 6998 #endif /* DEBUG */ |
| 7 | 6999 |
| 7000 #ifdef FEAT_MBYTE | |
| 7001 static void mb_decompose __ARGS((int c, int *c1, int *c2, int *c3)); | |
| 7002 | |
| 7003 typedef struct | |
| 7004 { | |
| 7005 int a, b, c; | |
| 7006 } decomp_T; | |
| 7007 | |
| 7008 | |
| 7009 /* 0xfb20 - 0xfb4f */ | |
| 297 | 7010 static decomp_T decomp_table[0xfb4f-0xfb20+1] = |
| 7 | 7011 { |
| 7012 {0x5e2,0,0}, /* 0xfb20 alt ayin */ | |
| 7013 {0x5d0,0,0}, /* 0xfb21 alt alef */ | |
| 7014 {0x5d3,0,0}, /* 0xfb22 alt dalet */ | |
| 7015 {0x5d4,0,0}, /* 0xfb23 alt he */ | |
| 7016 {0x5db,0,0}, /* 0xfb24 alt kaf */ | |
| 7017 {0x5dc,0,0}, /* 0xfb25 alt lamed */ | |
| 7018 {0x5dd,0,0}, /* 0xfb26 alt mem-sofit */ | |
| 7019 {0x5e8,0,0}, /* 0xfb27 alt resh */ | |
| 7020 {0x5ea,0,0}, /* 0xfb28 alt tav */ | |
| 7021 {'+', 0, 0}, /* 0xfb29 alt plus */ | |
| 7022 {0x5e9, 0x5c1, 0}, /* 0xfb2a shin+shin-dot */ | |
| 7023 {0x5e9, 0x5c2, 0}, /* 0xfb2b shin+sin-dot */ | |
| 7024 {0x5e9, 0x5c1, 0x5bc}, /* 0xfb2c shin+shin-dot+dagesh */ | |
| 7025 {0x5e9, 0x5c2, 0x5bc}, /* 0xfb2d shin+sin-dot+dagesh */ | |
| 7026 {0x5d0, 0x5b7, 0}, /* 0xfb2e alef+patah */ | |
| 7027 {0x5d0, 0x5b8, 0}, /* 0xfb2f alef+qamats */ | |
| 7028 {0x5d0, 0x5b4, 0}, /* 0xfb30 alef+hiriq */ | |
| 7029 {0x5d1, 0x5bc, 0}, /* 0xfb31 bet+dagesh */ | |
| 7030 {0x5d2, 0x5bc, 0}, /* 0xfb32 gimel+dagesh */ | |
| 7031 {0x5d3, 0x5bc, 0}, /* 0xfb33 dalet+dagesh */ | |
| 7032 {0x5d4, 0x5bc, 0}, /* 0xfb34 he+dagesh */ | |
| 7033 {0x5d5, 0x5bc, 0}, /* 0xfb35 vav+dagesh */ | |
| 7034 {0x5d6, 0x5bc, 0}, /* 0xfb36 zayin+dagesh */ | |
| 7035 {0xfb37, 0, 0}, /* 0xfb37 -- UNUSED */ | |
| 7036 {0x5d8, 0x5bc, 0}, /* 0xfb38 tet+dagesh */ | |
| 7037 {0x5d9, 0x5bc, 0}, /* 0xfb39 yud+dagesh */ | |
| 7038 {0x5da, 0x5bc, 0}, /* 0xfb3a kaf sofit+dagesh */ | |
| 7039 {0x5db, 0x5bc, 0}, /* 0xfb3b kaf+dagesh */ | |
| 7040 {0x5dc, 0x5bc, 0}, /* 0xfb3c lamed+dagesh */ | |
| 7041 {0xfb3d, 0, 0}, /* 0xfb3d -- UNUSED */ | |
| 7042 {0x5de, 0x5bc, 0}, /* 0xfb3e mem+dagesh */ | |
| 7043 {0xfb3f, 0, 0}, /* 0xfb3f -- UNUSED */ | |
| 7044 {0x5e0, 0x5bc, 0}, /* 0xfb40 nun+dagesh */ | |
| 7045 {0x5e1, 0x5bc, 0}, /* 0xfb41 samech+dagesh */ | |
| 7046 {0xfb42, 0, 0}, /* 0xfb42 -- UNUSED */ | |
| 7047 {0x5e3, 0x5bc, 0}, /* 0xfb43 pe sofit+dagesh */ | |
| 7048 {0x5e4, 0x5bc,0}, /* 0xfb44 pe+dagesh */ | |
| 7049 {0xfb45, 0, 0}, /* 0xfb45 -- UNUSED */ | |
| 7050 {0x5e6, 0x5bc, 0}, /* 0xfb46 tsadi+dagesh */ | |
| 7051 {0x5e7, 0x5bc, 0}, /* 0xfb47 qof+dagesh */ | |
| 7052 {0x5e8, 0x5bc, 0}, /* 0xfb48 resh+dagesh */ | |
| 7053 {0x5e9, 0x5bc, 0}, /* 0xfb49 shin+dagesh */ | |
| 7054 {0x5ea, 0x5bc, 0}, /* 0xfb4a tav+dagesh */ | |
| 7055 {0x5d5, 0x5b9, 0}, /* 0xfb4b vav+holam */ | |
| 7056 {0x5d1, 0x5bf, 0}, /* 0xfb4c bet+rafe */ | |
| 7057 {0x5db, 0x5bf, 0}, /* 0xfb4d kaf+rafe */ | |
| 7058 {0x5e4, 0x5bf, 0}, /* 0xfb4e pe+rafe */ | |
| 7059 {0x5d0, 0x5dc, 0} /* 0xfb4f alef-lamed */ | |
| 7060 }; | |
| 7061 | |
| 7062 static void | |
| 7063 mb_decompose(c, c1, c2, c3) | |
| 7064 int c, *c1, *c2, *c3; | |
| 7065 { | |
| 7066 decomp_T d; | |
| 7067 | |
|
4505
d037b9cbdaaa
updated for version 7.3.1000
Bram Moolenaar <bram@vim.org>
parents:
4466
diff
changeset
|
7068 if (c >= 0xfb20 && c <= 0xfb4f) |
| 7 | 7069 { |
| 7070 d = decomp_table[c - 0xfb20]; | |
| 7071 *c1 = d.a; | |
| 7072 *c2 = d.b; | |
| 7073 *c3 = d.c; | |
| 7074 } | |
| 7075 else | |
| 7076 { | |
| 7077 *c1 = c; | |
| 7078 *c2 = *c3 = 0; | |
| 7079 } | |
| 7080 } | |
| 7081 #endif | |
| 7082 | |
| 7083 /* | |
| 7084 * Compare two strings, ignore case if ireg_ic set. | |
| 7085 * Return 0 if strings match, non-zero otherwise. | |
| 7086 * Correct the length "*n" when composing characters are ignored. | |
| 7087 */ | |
| 7088 static int | |
| 7089 cstrncmp(s1, s2, n) | |
| 7090 char_u *s1, *s2; | |
| 7091 int *n; | |
| 7092 { | |
| 7093 int result; | |
| 7094 | |
| 7095 if (!ireg_ic) | |
| 7096 result = STRNCMP(s1, s2, *n); | |
| 7097 else | |
| 7098 result = MB_STRNICMP(s1, s2, *n); | |
| 7099 | |
| 7100 #ifdef FEAT_MBYTE | |
| 7101 /* if it failed and it's utf8 and we want to combineignore: */ | |
| 7102 if (result != 0 && enc_utf8 && ireg_icombine) | |
| 7103 { | |
| 7104 char_u *str1, *str2; | |
| 7105 int c1, c2, c11, c12; | |
| 7106 int junk; | |
| 7107 | |
| 7108 /* we have to handle the strcmp ourselves, since it is necessary to | |
| 7109 * deal with the composing characters by ignoring them: */ | |
| 7110 str1 = s1; | |
| 7111 str2 = s2; | |
| 7112 c1 = c2 = 0; | |
| 507 | 7113 while ((int)(str1 - s1) < *n) |
| 7 | 7114 { |
| 7115 c1 = mb_ptr2char_adv(&str1); | |
| 7116 c2 = mb_ptr2char_adv(&str2); | |
| 7117 | |
| 7118 /* decompose the character if necessary, into 'base' characters | |
| 7119 * because I don't care about Arabic, I will hard-code the Hebrew | |
| 7120 * which I *do* care about! So sue me... */ | |
| 7121 if (c1 != c2 && (!ireg_ic || utf_fold(c1) != utf_fold(c2))) | |
| 7122 { | |
| 7123 /* decomposition necessary? */ | |
| 7124 mb_decompose(c1, &c11, &junk, &junk); | |
| 7125 mb_decompose(c2, &c12, &junk, &junk); | |
| 7126 c1 = c11; | |
| 7127 c2 = c12; | |
| 7128 if (c11 != c12 && (!ireg_ic || utf_fold(c11) != utf_fold(c12))) | |
| 7129 break; | |
| 7130 } | |
| 7131 } | |
| 7132 result = c2 - c1; | |
| 7133 if (result == 0) | |
| 7134 *n = (int)(str2 - s2); | |
| 7135 } | |
| 7136 #endif | |
| 7137 | |
| 7138 return result; | |
| 7139 } | |
| 7140 | |
| 7141 /* | |
| 7142 * cstrchr: This function is used a lot for simple searches, keep it fast! | |
| 7143 */ | |
| 7144 static char_u * | |
| 7145 cstrchr(s, c) | |
| 7146 char_u *s; | |
| 7147 int c; | |
| 7148 { | |
| 7149 char_u *p; | |
| 7150 int cc; | |
| 7151 | |
| 7152 if (!ireg_ic | |
| 7153 #ifdef FEAT_MBYTE | |
| 7154 || (!enc_utf8 && mb_char2len(c) > 1) | |
| 7155 #endif | |
| 7156 ) | |
| 7157 return vim_strchr(s, c); | |
| 7158 | |
| 7159 /* tolower() and toupper() can be slow, comparing twice should be a lot | |
| 7160 * faster (esp. when using MS Visual C++!). | |
| 7161 * For UTF-8 need to use folded case. */ | |
| 7162 #ifdef FEAT_MBYTE | |
| 7163 if (enc_utf8 && c > 0x80) | |
| 7164 cc = utf_fold(c); | |
| 7165 else | |
| 7166 #endif | |
| 1347 | 7167 if (MB_ISUPPER(c)) |
| 7168 cc = MB_TOLOWER(c); | |
| 7169 else if (MB_ISLOWER(c)) | |
| 7170 cc = MB_TOUPPER(c); | |
| 7 | 7171 else |
| 7172 return vim_strchr(s, c); | |
| 7173 | |
| 7174 #ifdef FEAT_MBYTE | |
| 7175 if (has_mbyte) | |
| 7176 { | |
| 474 | 7177 for (p = s; *p != NUL; p += (*mb_ptr2len)(p)) |
| 7 | 7178 { |
| 7179 if (enc_utf8 && c > 0x80) | |
| 7180 { | |
| 7181 if (utf_fold(utf_ptr2char(p)) == cc) | |
| 7182 return p; | |
| 7183 } | |
| 7184 else if (*p == c || *p == cc) | |
| 7185 return p; | |
| 7186 } | |
| 7187 } | |
| 7188 else | |
| 7189 #endif | |
| 7190 /* Faster version for when there are no multi-byte characters. */ | |
| 7191 for (p = s; *p != NUL; ++p) | |
| 7192 if (*p == c || *p == cc) | |
| 7193 return p; | |
| 7194 | |
| 7195 return NULL; | |
| 7196 } | |
| 7197 | |
| 7198 /*************************************************************** | |
| 7199 * regsub stuff * | |
| 7200 ***************************************************************/ | |
| 7201 | |
| 7202 /* This stuff below really confuses cc on an SGI -- webb */ | |
| 7203 #ifdef __sgi | |
| 7204 # undef __ARGS | |
| 7205 # define __ARGS(x) () | |
| 7206 #endif | |
| 7207 | |
| 7208 /* | |
| 7209 * We should define ftpr as a pointer to a function returning a pointer to | |
| 7210 * a function returning a pointer to a function ... | |
| 7211 * This is impossible, so we declare a pointer to a function returning a | |
| 7212 * pointer to a function returning void. This should work for all compilers. | |
| 7213 */ | |
| 772 | 7214 typedef void (*(*fptr_T) __ARGS((int *, int)))(); |
| 7215 | |
| 7216 static fptr_T do_upper __ARGS((int *, int)); | |
| 7217 static fptr_T do_Upper __ARGS((int *, int)); | |
| 7218 static fptr_T do_lower __ARGS((int *, int)); | |
| 7219 static fptr_T do_Lower __ARGS((int *, int)); | |
| 7 | 7220 |
| 7221 static int vim_regsub_both __ARGS((char_u *source, char_u *dest, int copy, int magic, int backslash)); | |
| 7222 | |
| 772 | 7223 static fptr_T |
| 7 | 7224 do_upper(d, c) |
| 772 | 7225 int *d; |
| 7226 int c; | |
| 7 | 7227 { |
| 772 | 7228 *d = MB_TOUPPER(c); |
| 7229 | |
| 7230 return (fptr_T)NULL; | |
| 7 | 7231 } |
| 7232 | |
| 772 | 7233 static fptr_T |
| 7234 do_Upper(d, c) | |
| 7235 int *d; | |
| 7 | 7236 int c; |
| 7237 { | |
| 772 | 7238 *d = MB_TOUPPER(c); |
| 7239 | |
| 7240 return (fptr_T)do_Upper; | |
| 7241 } | |
| 7242 | |
| 7243 static fptr_T | |
| 7244 do_lower(d, c) | |
| 7245 int *d; | |
| 7246 int c; | |
| 7247 { | |
| 7248 *d = MB_TOLOWER(c); | |
| 7249 | |
| 7250 return (fptr_T)NULL; | |
| 7251 } | |
| 7252 | |
| 7253 static fptr_T | |
| 7254 do_Lower(d, c) | |
| 7255 int *d; | |
| 7256 int c; | |
| 7257 { | |
| 7258 *d = MB_TOLOWER(c); | |
| 7259 | |
| 7260 return (fptr_T)do_Lower; | |
| 7 | 7261 } |
| 7262 | |
| 7263 /* | |
| 7264 * regtilde(): Replace tildes in the pattern by the old pattern. | |
| 7265 * | |
| 7266 * Short explanation of the tilde: It stands for the previous replacement | |
| 7267 * pattern. If that previous pattern also contains a ~ we should go back a | |
| 7268 * step further... But we insert the previous pattern into the current one | |
| 7269 * and remember that. | |
| 772 | 7270 * This still does not handle the case where "magic" changes. So require the |
| 7271 * user to keep his hands off of "magic". | |
| 7 | 7272 * |
| 7273 * The tildes are parsed once before the first call to vim_regsub(). | |
| 7274 */ | |
| 7275 char_u * | |
| 7276 regtilde(source, magic) | |
| 7277 char_u *source; | |
| 7278 int magic; | |
| 7279 { | |
| 7280 char_u *newsub = source; | |
| 7281 char_u *tmpsub; | |
| 7282 char_u *p; | |
| 7283 int len; | |
| 7284 int prevlen; | |
| 7285 | |
| 7286 for (p = newsub; *p; ++p) | |
| 7287 { | |
| 7288 if ((*p == '~' && magic) || (*p == '\\' && *(p + 1) == '~' && !magic)) | |
| 7289 { | |
| 7290 if (reg_prev_sub != NULL) | |
| 7291 { | |
| 7292 /* length = len(newsub) - 1 + len(prev_sub) + 1 */ | |
| 7293 prevlen = (int)STRLEN(reg_prev_sub); | |
| 7294 tmpsub = alloc((unsigned)(STRLEN(newsub) + prevlen)); | |
| 7295 if (tmpsub != NULL) | |
| 7296 { | |
| 7297 /* copy prefix */ | |
| 7298 len = (int)(p - newsub); /* not including ~ */ | |
| 7299 mch_memmove(tmpsub, newsub, (size_t)len); | |
| 1209 | 7300 /* interpret tilde */ |
| 7 | 7301 mch_memmove(tmpsub + len, reg_prev_sub, (size_t)prevlen); |
| 7302 /* copy postfix */ | |
| 7303 if (!magic) | |
| 7304 ++p; /* back off \ */ | |
| 7305 STRCPY(tmpsub + len + prevlen, p + 1); | |
| 7306 | |
| 7307 if (newsub != source) /* already allocated newsub */ | |
| 7308 vim_free(newsub); | |
| 7309 newsub = tmpsub; | |
| 7310 p = newsub + len + prevlen; | |
| 7311 } | |
| 7312 } | |
| 7313 else if (magic) | |
| 1621 | 7314 STRMOVE(p, p + 1); /* remove '~' */ |
| 7 | 7315 else |
| 1621 | 7316 STRMOVE(p, p + 2); /* remove '\~' */ |
| 7 | 7317 --p; |
| 7318 } | |
| 7319 else | |
| 7320 { | |
| 7321 if (*p == '\\' && p[1]) /* skip escaped characters */ | |
| 7322 ++p; | |
| 7323 #ifdef FEAT_MBYTE | |
| 7324 if (has_mbyte) | |
| 474 | 7325 p += (*mb_ptr2len)(p) - 1; |
| 7 | 7326 #endif |
| 7327 } | |
| 7328 } | |
| 7329 | |
| 7330 vim_free(reg_prev_sub); | |
| 7331 if (newsub != source) /* newsub was allocated, just keep it */ | |
| 7332 reg_prev_sub = newsub; | |
| 7333 else /* no ~ found, need to save newsub */ | |
| 7334 reg_prev_sub = vim_strsave(newsub); | |
| 7335 return newsub; | |
| 7336 } | |
| 7337 | |
| 7338 #ifdef FEAT_EVAL | |
| 7339 static int can_f_submatch = FALSE; /* TRUE when submatch() can be used */ | |
| 7340 | |
| 7341 /* These pointers are used instead of reg_match and reg_mmatch for | |
| 7342 * reg_submatch(). Needed for when the substitution string is an expression | |
| 7343 * that contains a call to substitute() and submatch(). */ | |
| 7344 static regmatch_T *submatch_match; | |
| 7345 static regmmatch_T *submatch_mmatch; | |
| 2011 | 7346 static linenr_T submatch_firstlnum; |
| 7347 static linenr_T submatch_maxline; | |
| 2904 | 7348 static int submatch_line_lbr; |
| 7 | 7349 #endif |
| 7350 | |
| 7351 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO) | |
| 7352 /* | |
| 7353 * vim_regsub() - perform substitutions after a vim_regexec() or | |
| 7354 * vim_regexec_multi() match. | |
| 7355 * | |
| 7356 * If "copy" is TRUE really copy into "dest". | |
| 7357 * If "copy" is FALSE nothing is copied, this is just to find out the length | |
| 7358 * of the result. | |
| 7359 * | |
| 7360 * If "backslash" is TRUE, a backslash will be removed later, need to double | |
| 7361 * them to keep them, and insert a backslash before a CR to avoid it being | |
| 7362 * replaced with a line break later. | |
| 7363 * | |
| 7364 * Note: The matched text must not change between the call of | |
| 7365 * vim_regexec()/vim_regexec_multi() and vim_regsub()! It would make the back | |
| 7366 * references invalid! | |
| 7367 * | |
| 7368 * Returns the size of the replacement, including terminating NUL. | |
| 7369 */ | |
| 7370 int | |
| 7371 vim_regsub(rmp, source, dest, copy, magic, backslash) | |
| 7372 regmatch_T *rmp; | |
| 7373 char_u *source; | |
| 7374 char_u *dest; | |
| 7375 int copy; | |
| 7376 int magic; | |
| 7377 int backslash; | |
| 7378 { | |
| 7379 reg_match = rmp; | |
| 7380 reg_mmatch = NULL; | |
| 7381 reg_maxline = 0; | |
| 4061 | 7382 reg_buf = curbuf; |
| 7 | 7383 return vim_regsub_both(source, dest, copy, magic, backslash); |
| 7384 } | |
| 7385 #endif | |
| 7386 | |
| 7387 int | |
| 7388 vim_regsub_multi(rmp, lnum, source, dest, copy, magic, backslash) | |
| 7389 regmmatch_T *rmp; | |
| 7390 linenr_T lnum; | |
| 7391 char_u *source; | |
| 7392 char_u *dest; | |
| 7393 int copy; | |
| 7394 int magic; | |
| 7395 int backslash; | |
| 7396 { | |
| 7397 reg_match = NULL; | |
| 7398 reg_mmatch = rmp; | |
| 7399 reg_buf = curbuf; /* always works on the current buffer! */ | |
| 7400 reg_firstlnum = lnum; | |
| 7401 reg_maxline = curbuf->b_ml.ml_line_count - lnum; | |
| 7402 return vim_regsub_both(source, dest, copy, magic, backslash); | |
| 7403 } | |
| 7404 | |
| 7405 static int | |
| 7406 vim_regsub_both(source, dest, copy, magic, backslash) | |
| 7407 char_u *source; | |
| 7408 char_u *dest; | |
| 7409 int copy; | |
| 7410 int magic; | |
| 7411 int backslash; | |
| 7412 { | |
| 7413 char_u *src; | |
| 7414 char_u *dst; | |
| 7415 char_u *s; | |
| 7416 int c; | |
| 772 | 7417 int cc; |
| 7 | 7418 int no = -1; |
| 4244 | 7419 fptr_T func_all = (fptr_T)NULL; |
| 7420 fptr_T func_one = (fptr_T)NULL; | |
| 7 | 7421 linenr_T clnum = 0; /* init for GCC */ |
| 7422 int len = 0; /* init for GCC */ | |
| 7423 #ifdef FEAT_EVAL | |
| 7424 static char_u *eval_result = NULL; | |
| 7425 #endif | |
| 7426 | |
| 7427 /* Be paranoid... */ | |
| 7428 if (source == NULL || dest == NULL) | |
| 7429 { | |
| 7430 EMSG(_(e_null)); | |
| 7431 return 0; | |
| 7432 } | |
| 7433 if (prog_magic_wrong()) | |
| 7434 return 0; | |
| 7435 src = source; | |
| 7436 dst = dest; | |
| 7437 | |
| 7438 /* | |
| 7439 * When the substitute part starts with "\=" evaluate it as an expression. | |
| 7440 */ | |
| 7441 if (source[0] == '\\' && source[1] == '=' | |
| 7442 #ifdef FEAT_EVAL | |
| 7443 && !can_f_submatch /* can't do this recursively */ | |
| 7444 #endif | |
| 7445 ) | |
| 7446 { | |
| 7447 #ifdef FEAT_EVAL | |
| 7448 /* To make sure that the length doesn't change between checking the | |
| 7449 * length and copying the string, and to speed up things, the | |
| 7450 * resulting string is saved from the call with "copy" == FALSE to the | |
| 7451 * call with "copy" == TRUE. */ | |
| 7452 if (copy) | |
| 7453 { | |
| 7454 if (eval_result != NULL) | |
| 7455 { | |
| 7456 STRCPY(dest, eval_result); | |
| 7457 dst += STRLEN(eval_result); | |
| 7458 vim_free(eval_result); | |
| 7459 eval_result = NULL; | |
| 7460 } | |
| 7461 } | |
| 7462 else | |
| 7463 { | |
| 7464 win_T *save_reg_win; | |
| 7465 int save_ireg_ic; | |
| 7466 | |
| 7467 vim_free(eval_result); | |
| 7468 | |
| 7469 /* The expression may contain substitute(), which calls us | |
| 7470 * recursively. Make sure submatch() gets the text from the first | |
| 7471 * level. Don't need to save "reg_buf", because | |
| 7472 * vim_regexec_multi() can't be called recursively. */ | |
| 7473 submatch_match = reg_match; | |
| 7474 submatch_mmatch = reg_mmatch; | |
| 2011 | 7475 submatch_firstlnum = reg_firstlnum; |
| 7476 submatch_maxline = reg_maxline; | |
| 2904 | 7477 submatch_line_lbr = reg_line_lbr; |
| 7 | 7478 save_reg_win = reg_win; |
| 7479 save_ireg_ic = ireg_ic; | |
| 7480 can_f_submatch = TRUE; | |
| 7481 | |
| 714 | 7482 eval_result = eval_to_string(source + 2, NULL, TRUE); |
| 7 | 7483 if (eval_result != NULL) |
| 7484 { | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7485 int had_backslash = FALSE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7486 |
| 39 | 7487 for (s = eval_result; *s != NUL; mb_ptr_adv(s)) |
| 7 | 7488 { |
| 2904 | 7489 /* Change NL to CR, so that it becomes a line break, |
| 7490 * unless called from vim_regexec_nl(). | |
| 7 | 7491 * Skip over a backslashed character. */ |
| 2904 | 7492 if (*s == NL && !submatch_line_lbr) |
| 7 | 7493 *s = CAR; |
| 7494 else if (*s == '\\' && s[1] != NUL) | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7495 { |
| 7 | 7496 ++s; |
| 2173 | 7497 /* Change NL to CR here too, so that this works: |
| 7498 * :s/abc\\\ndef/\="aaa\\\nbbb"/ on text: | |
| 7499 * abc\ | |
| 7500 * def | |
| 2904 | 7501 * Not when called from vim_regexec_nl(). |
| 2173 | 7502 */ |
| 2904 | 7503 if (*s == NL && !submatch_line_lbr) |
| 2173 | 7504 *s = CAR; |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7505 had_backslash = TRUE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7506 } |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7507 } |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7508 if (had_backslash && backslash) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7509 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7510 /* Backslashes will be consumed, need to double them. */ |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7511 s = vim_strsave_escaped(eval_result, (char_u *)"\\"); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7512 if (s != NULL) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7513 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7514 vim_free(eval_result); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7515 eval_result = s; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7516 } |
| 7 | 7517 } |
| 7518 | |
| 7519 dst += STRLEN(eval_result); | |
| 7520 } | |
| 7521 | |
| 7522 reg_match = submatch_match; | |
| 7523 reg_mmatch = submatch_mmatch; | |
| 2011 | 7524 reg_firstlnum = submatch_firstlnum; |
| 7525 reg_maxline = submatch_maxline; | |
| 2904 | 7526 reg_line_lbr = submatch_line_lbr; |
| 7 | 7527 reg_win = save_reg_win; |
| 7528 ireg_ic = save_ireg_ic; | |
| 7529 can_f_submatch = FALSE; | |
| 7530 } | |
| 7531 #endif | |
| 7532 } | |
| 7533 else | |
| 7534 while ((c = *src++) != NUL) | |
| 7535 { | |
| 7536 if (c == '&' && magic) | |
| 7537 no = 0; | |
| 7538 else if (c == '\\' && *src != NUL) | |
| 7539 { | |
| 7540 if (*src == '&' && !magic) | |
| 7541 { | |
| 7542 ++src; | |
| 7543 no = 0; | |
| 7544 } | |
| 7545 else if ('0' <= *src && *src <= '9') | |
| 7546 { | |
| 7547 no = *src++ - '0'; | |
| 7548 } | |
| 7549 else if (vim_strchr((char_u *)"uUlLeE", *src)) | |
| 7550 { | |
| 7551 switch (*src++) | |
| 7552 { | |
| 4244 | 7553 case 'u': func_one = (fptr_T)do_upper; |
| 7 | 7554 continue; |
| 4244 | 7555 case 'U': func_all = (fptr_T)do_Upper; |
| 7 | 7556 continue; |
| 4244 | 7557 case 'l': func_one = (fptr_T)do_lower; |
| 7 | 7558 continue; |
| 4244 | 7559 case 'L': func_all = (fptr_T)do_Lower; |
| 7 | 7560 continue; |
| 7561 case 'e': | |
| 4244 | 7562 case 'E': func_one = func_all = (fptr_T)NULL; |
| 7 | 7563 continue; |
| 7564 } | |
| 7565 } | |
| 7566 } | |
| 7567 if (no < 0) /* Ordinary character. */ | |
| 7568 { | |
| 798 | 7569 if (c == K_SPECIAL && src[0] != NUL && src[1] != NUL) |
| 7570 { | |
| 1209 | 7571 /* Copy a special key as-is. */ |
| 798 | 7572 if (copy) |
| 7573 { | |
| 7574 *dst++ = c; | |
| 7575 *dst++ = *src++; | |
| 7576 *dst++ = *src++; | |
| 7577 } | |
| 7578 else | |
| 7579 { | |
| 7580 dst += 3; | |
| 7581 src += 2; | |
| 7582 } | |
| 7583 continue; | |
| 7584 } | |
| 7585 | |
| 7 | 7586 if (c == '\\' && *src != NUL) |
| 7587 { | |
| 7588 /* Check for abbreviations -- webb */ | |
| 7589 switch (*src) | |
| 7590 { | |
| 7591 case 'r': c = CAR; ++src; break; | |
| 7592 case 'n': c = NL; ++src; break; | |
| 7593 case 't': c = TAB; ++src; break; | |
| 7594 /* Oh no! \e already has meaning in subst pat :-( */ | |
| 7595 /* case 'e': c = ESC; ++src; break; */ | |
| 7596 case 'b': c = Ctrl_H; ++src; break; | |
| 7597 | |
| 7598 /* If "backslash" is TRUE the backslash will be removed | |
| 7599 * later. Used to insert a literal CR. */ | |
| 7600 default: if (backslash) | |
| 7601 { | |
| 7602 if (copy) | |
| 7603 *dst = '\\'; | |
| 7604 ++dst; | |
| 7605 } | |
| 7606 c = *src++; | |
| 7607 } | |
| 7608 } | |
| 798 | 7609 #ifdef FEAT_MBYTE |
| 7610 else if (has_mbyte) | |
| 7611 c = mb_ptr2char(src - 1); | |
| 7612 #endif | |
| 7 | 7613 |
| 7614 /* Write to buffer, if copy is set. */ | |
| 4244 | 7615 if (func_one != (fptr_T)NULL) |
| 7616 /* Turbo C complains without the typecast */ | |
| 7617 func_one = (fptr_T)(func_one(&cc, c)); | |
| 7618 else if (func_all != (fptr_T)NULL) | |
| 7619 /* Turbo C complains without the typecast */ | |
| 7620 func_all = (fptr_T)(func_all(&cc, c)); | |
| 7621 else /* just copy */ | |
| 772 | 7622 cc = c; |
| 7623 | |
| 7624 #ifdef FEAT_MBYTE | |
| 7625 if (has_mbyte) | |
| 7 | 7626 { |
|
2307
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7627 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
|
7628 |
| 7 | 7629 if (copy) |
| 772 | 7630 mb_char2bytes(cc, dst); |
| 7631 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
|
7632 if (enc_utf8) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7633 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7634 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
|
7635 |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7636 /* 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
|
7637 * 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
|
7638 if (clen < totlen) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7639 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7640 if (copy) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7641 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
|
7642 (size_t)(totlen - clen)); |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7643 dst += totlen - clen; |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7644 } |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7645 } |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7646 src += totlen - 1; |
| 7 | 7647 } |
| 7648 else | |
| 7649 #endif | |
| 7650 if (copy) | |
| 772 | 7651 *dst = cc; |
| 7 | 7652 dst++; |
| 7653 } | |
| 7654 else | |
| 7655 { | |
| 7656 if (REG_MULTI) | |
| 7657 { | |
| 7658 clnum = reg_mmatch->startpos[no].lnum; | |
| 7659 if (clnum < 0 || reg_mmatch->endpos[no].lnum < 0) | |
| 7660 s = NULL; | |
| 7661 else | |
| 7662 { | |
| 7663 s = reg_getline(clnum) + reg_mmatch->startpos[no].col; | |
| 7664 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7665 len = reg_mmatch->endpos[no].col | |
| 7666 - reg_mmatch->startpos[no].col; | |
| 7667 else | |
| 7668 len = (int)STRLEN(s); | |
| 7669 } | |
| 7670 } | |
| 7671 else | |
| 7672 { | |
| 7673 s = reg_match->startp[no]; | |
| 7674 if (reg_match->endp[no] == NULL) | |
| 7675 s = NULL; | |
| 7676 else | |
| 7677 len = (int)(reg_match->endp[no] - s); | |
| 7678 } | |
| 7679 if (s != NULL) | |
| 7680 { | |
| 7681 for (;;) | |
| 7682 { | |
| 7683 if (len == 0) | |
| 7684 { | |
| 7685 if (REG_MULTI) | |
| 7686 { | |
| 7687 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7688 break; | |
| 7689 if (copy) | |
| 7690 *dst = CAR; | |
| 7691 ++dst; | |
| 7692 s = reg_getline(++clnum); | |
| 7693 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7694 len = reg_mmatch->endpos[no].col; | |
| 7695 else | |
| 7696 len = (int)STRLEN(s); | |
| 7697 } | |
| 7698 else | |
| 7699 break; | |
| 7700 } | |
| 7701 else if (*s == NUL) /* we hit NUL. */ | |
| 7702 { | |
| 7703 if (copy) | |
| 7704 EMSG(_(e_re_damg)); | |
| 7705 goto exit; | |
| 7706 } | |
| 7707 else | |
| 7708 { | |
| 7709 if (backslash && (*s == CAR || *s == '\\')) | |
| 7710 { | |
| 7711 /* | |
| 7712 * Insert a backslash in front of a CR, otherwise | |
| 7713 * it will be replaced by a line break. | |
| 7714 * Number of backslashes will be halved later, | |
| 7715 * double them here. | |
| 7716 */ | |
| 7717 if (copy) | |
| 7718 { | |
| 7719 dst[0] = '\\'; | |
| 7720 dst[1] = *s; | |
| 7721 } | |
| 7722 dst += 2; | |
| 7723 } | |
| 7724 else | |
| 7725 { | |
| 772 | 7726 #ifdef FEAT_MBYTE |
| 7727 if (has_mbyte) | |
| 7728 c = mb_ptr2char(s); | |
| 7729 else | |
| 7730 #endif | |
| 7731 c = *s; | |
| 7732 | |
| 4244 | 7733 if (func_one != (fptr_T)NULL) |
| 7734 /* Turbo C complains without the typecast */ | |
| 7735 func_one = (fptr_T)(func_one(&cc, c)); | |
| 7736 else if (func_all != (fptr_T)NULL) | |
| 7737 /* Turbo C complains without the typecast */ | |
| 7738 func_all = (fptr_T)(func_all(&cc, c)); | |
| 7739 else /* just copy */ | |
| 772 | 7740 cc = c; |
| 7741 | |
| 7742 #ifdef FEAT_MBYTE | |
| 7743 if (has_mbyte) | |
| 7 | 7744 { |
| 1332 | 7745 int l; |
| 7746 | |
| 7747 /* Copy composing characters separately, one | |
| 7748 * at a time. */ | |
| 7749 if (enc_utf8) | |
| 7750 l = utf_ptr2len(s) - 1; | |
| 7751 else | |
| 7752 l = mb_ptr2len(s) - 1; | |
| 772 | 7753 |
| 7754 s += l; | |
| 7755 len -= l; | |
| 7756 if (copy) | |
| 7757 mb_char2bytes(cc, dst); | |
| 7758 dst += mb_char2len(cc) - 1; | |
| 7 | 7759 } |
| 772 | 7760 else |
| 7761 #endif | |
| 7762 if (copy) | |
| 7763 *dst = cc; | |
| 7764 dst++; | |
| 7 | 7765 } |
| 772 | 7766 |
| 7 | 7767 ++s; |
| 7768 --len; | |
| 7769 } | |
| 7770 } | |
| 7771 } | |
| 7772 no = -1; | |
| 7773 } | |
| 7774 } | |
| 7775 if (copy) | |
| 7776 *dst = NUL; | |
| 7777 | |
| 7778 exit: | |
| 7779 return (int)((dst - dest) + 1); | |
| 7780 } | |
| 7781 | |
| 7782 #ifdef FEAT_EVAL | |
| 2012 | 7783 static char_u *reg_getline_submatch __ARGS((linenr_T lnum)); |
| 7784 | |
| 7 | 7785 /* |
| 2011 | 7786 * Call reg_getline() with the line numbers from the submatch. If a |
| 7787 * substitute() was used the reg_maxline and other values have been | |
| 7788 * overwritten. | |
| 7789 */ | |
| 7790 static char_u * | |
| 7791 reg_getline_submatch(lnum) | |
| 7792 linenr_T lnum; | |
| 7793 { | |
| 7794 char_u *s; | |
| 7795 linenr_T save_first = reg_firstlnum; | |
| 7796 linenr_T save_max = reg_maxline; | |
| 7797 | |
| 7798 reg_firstlnum = submatch_firstlnum; | |
| 7799 reg_maxline = submatch_maxline; | |
| 7800 | |
| 7801 s = reg_getline(lnum); | |
| 7802 | |
| 7803 reg_firstlnum = save_first; | |
| 7804 reg_maxline = save_max; | |
| 7805 return s; | |
| 7806 } | |
| 7807 | |
| 7808 /* | |
| 1209 | 7809 * Used for the submatch() function: get the string from the n'th submatch in |
| 7 | 7810 * allocated memory. |
| 7811 * Returns NULL when not in a ":s" command and for a non-existing submatch. | |
| 7812 */ | |
| 7813 char_u * | |
| 7814 reg_submatch(no) | |
| 7815 int no; | |
| 7816 { | |
| 7817 char_u *retval = NULL; | |
| 7818 char_u *s; | |
| 7819 int len; | |
| 7820 int round; | |
| 7821 linenr_T lnum; | |
| 7822 | |
| 840 | 7823 if (!can_f_submatch || no < 0) |
| 7 | 7824 return NULL; |
| 7825 | |
| 7826 if (submatch_match == NULL) | |
| 7827 { | |
| 7828 /* | |
| 7829 * First round: compute the length and allocate memory. | |
| 7830 * Second round: copy the text. | |
| 7831 */ | |
| 7832 for (round = 1; round <= 2; ++round) | |
| 7833 { | |
| 7834 lnum = submatch_mmatch->startpos[no].lnum; | |
| 7835 if (lnum < 0 || submatch_mmatch->endpos[no].lnum < 0) | |
| 7836 return NULL; | |
| 7837 | |
| 2011 | 7838 s = reg_getline_submatch(lnum) + submatch_mmatch->startpos[no].col; |
| 7 | 7839 if (s == NULL) /* anti-crash check, cannot happen? */ |
| 7840 break; | |
| 7841 if (submatch_mmatch->endpos[no].lnum == lnum) | |
| 7842 { | |
| 7843 /* Within one line: take form start to end col. */ | |
| 7844 len = submatch_mmatch->endpos[no].col | |
| 7845 - submatch_mmatch->startpos[no].col; | |
| 7846 if (round == 2) | |
| 418 | 7847 vim_strncpy(retval, s, len); |
| 7 | 7848 ++len; |
| 7849 } | |
| 7850 else | |
| 7851 { | |
| 7852 /* Multiple lines: take start line from start col, middle | |
| 7853 * lines completely and end line up to end col. */ | |
| 7854 len = (int)STRLEN(s); | |
| 7855 if (round == 2) | |
| 7856 { | |
| 7857 STRCPY(retval, s); | |
| 7858 retval[len] = '\n'; | |
| 7859 } | |
| 7860 ++len; | |
| 7861 ++lnum; | |
| 7862 while (lnum < submatch_mmatch->endpos[no].lnum) | |
| 7863 { | |
| 2011 | 7864 s = reg_getline_submatch(lnum++); |
| 7 | 7865 if (round == 2) |
| 7866 STRCPY(retval + len, s); | |
| 7867 len += (int)STRLEN(s); | |
| 7868 if (round == 2) | |
| 7869 retval[len] = '\n'; | |
| 7870 ++len; | |
| 7871 } | |
| 7872 if (round == 2) | |
| 2011 | 7873 STRNCPY(retval + len, reg_getline_submatch(lnum), |
| 7 | 7874 submatch_mmatch->endpos[no].col); |
| 7875 len += submatch_mmatch->endpos[no].col; | |
| 7876 if (round == 2) | |
| 7877 retval[len] = NUL; | |
| 7878 ++len; | |
| 7879 } | |
| 7880 | |
| 840 | 7881 if (retval == NULL) |
| 7 | 7882 { |
| 7883 retval = lalloc((long_u)len, TRUE); | |
| 840 | 7884 if (retval == NULL) |
| 7 | 7885 return NULL; |
| 7886 } | |
| 7887 } | |
| 7888 } | |
| 7889 else | |
| 7890 { | |
| 1815 | 7891 s = submatch_match->startp[no]; |
| 7892 if (s == NULL || submatch_match->endp[no] == NULL) | |
| 7 | 7893 retval = NULL; |
| 7894 else | |
| 7895 retval = vim_strnsave(s, (int)(submatch_match->endp[no] - s)); | |
| 7896 } | |
| 7897 | |
| 7898 return retval; | |
| 7899 } | |
| 7900 #endif | |
| 4444 | 7901 |
| 7902 static regengine_T bt_regengine = | |
| 7903 { | |
| 7904 bt_regcomp, | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7905 bt_regfree, |
| 4444 | 7906 bt_regexec, |
| 7907 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
| 7908 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 7909 bt_regexec_nl, | |
| 7910 #endif | |
| 7911 bt_regexec_multi | |
| 7912 #ifdef DEBUG | |
| 7913 ,(char_u *)"" | |
| 7914 #endif | |
| 7915 }; | |
| 7916 | |
| 7917 | |
| 7918 #include "regexp_nfa.c" | |
| 7919 | |
| 7920 static regengine_T nfa_regengine = | |
| 7921 { | |
| 7922 nfa_regcomp, | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7923 nfa_regfree, |
| 4444 | 7924 nfa_regexec, |
| 7925 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
| 7926 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 7927 nfa_regexec_nl, | |
| 7928 #endif | |
| 7929 nfa_regexec_multi | |
| 7930 #ifdef DEBUG | |
| 7931 ,(char_u *)"" | |
| 7932 #endif | |
| 7933 }; | |
| 7934 | |
| 7935 /* Which regexp engine to use? Needed for vim_regcomp(). | |
| 7936 * Must match with 'regexpengine'. */ | |
| 7937 static int regexp_engine = 0; | |
| 7938 #define AUTOMATIC_ENGINE 0 | |
| 7939 #define BACKTRACKING_ENGINE 1 | |
| 7940 #define NFA_ENGINE 2 | |
| 7941 #ifdef DEBUG | |
| 7942 static char_u regname[][30] = { | |
| 7943 "AUTOMATIC Regexp Engine", | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
7944 "BACKTRACKING Regexp Engine", |
| 4444 | 7945 "NFA Regexp Engine" |
| 7946 }; | |
| 7947 #endif | |
| 7948 | |
| 7949 /* | |
| 7950 * Compile a regular expression into internal code. | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7951 * Returns the program in allocated memory. |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7952 * Use vim_regfree() to free the memory. |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
7953 * Returns NULL for an error. |
| 4444 | 7954 */ |
| 7955 regprog_T * | |
| 7956 vim_regcomp(expr_arg, re_flags) | |
| 7957 char_u *expr_arg; | |
| 7958 int re_flags; | |
| 7959 { | |
| 7960 regprog_T *prog = NULL; | |
| 7961 char_u *expr = expr_arg; | |
| 7962 | |
| 7963 regexp_engine = p_re; | |
| 7964 | |
| 7965 /* Check for prefix "\%#=", that sets the regexp engine */ | |
| 7966 if (STRNCMP(expr, "\\%#=", 4) == 0) | |
| 7967 { | |
| 7968 int newengine = expr[4] - '0'; | |
| 7969 | |
| 7970 if (newengine == AUTOMATIC_ENGINE | |
| 7971 || newengine == BACKTRACKING_ENGINE | |
| 7972 || newengine == NFA_ENGINE) | |
| 7973 { | |
| 7974 regexp_engine = expr[4] - '0'; | |
| 7975 expr += 5; | |
| 7976 #ifdef DEBUG | |
| 7977 EMSG3("New regexp mode selected (%d): %s", regexp_engine, | |
| 7978 regname[newengine]); | |
| 7979 #endif | |
| 7980 } | |
| 7981 else | |
| 7982 { | |
| 7983 EMSG(_("E864: \\%#= can only be followed by 0, 1, or 2. The automatic engine will be used ")); | |
| 7984 regexp_engine = AUTOMATIC_ENGINE; | |
| 7985 } | |
| 7986 } | |
| 7987 #ifdef DEBUG | |
| 7988 bt_regengine.expr = expr; | |
| 7989 nfa_regengine.expr = expr; | |
| 7990 #endif | |
| 7991 | |
| 7992 /* | |
| 7993 * First try the NFA engine, unless backtracking was requested. | |
| 7994 */ | |
| 7995 if (regexp_engine != BACKTRACKING_ENGINE) | |
| 7996 prog = nfa_regengine.regcomp(expr, re_flags); | |
| 7997 else | |
| 7998 prog = bt_regengine.regcomp(expr, re_flags); | |
| 7999 | |
| 8000 if (prog == NULL) /* error compiling regexp with initial engine */ | |
| 8001 { | |
| 4460 | 8002 #ifdef BT_REGEXP_DEBUG_LOG |
| 4444 | 8003 if (regexp_engine != BACKTRACKING_ENGINE) /* debugging log for NFA */ |
| 8004 { | |
| 8005 FILE *f; | |
| 4460 | 8006 f = fopen(BT_REGEXP_DEBUG_LOG_NAME, "a"); |
| 4444 | 8007 if (f) |
| 8008 { | |
|
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8009 fprintf(f, "Syntax error in \"%s\"\n", expr); |
| 4444 | 8010 fclose(f); |
| 8011 } | |
| 8012 else | |
| 4460 | 8013 EMSG2("(NFA) Could not open \"%s\" to write !!!", |
| 8014 BT_REGEXP_DEBUG_LOG_NAME); | |
| 4444 | 8015 } |
| 8016 #endif | |
| 8017 /* | |
| 8018 * If NFA engine failed, then revert to the backtracking engine. | |
| 8019 * Except when there was a syntax error, which was properly handled by | |
| 8020 * NFA engine. | |
| 8021 */ | |
| 8022 if (regexp_engine == AUTOMATIC_ENGINE) | |
|
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8023 prog = bt_regengine.regcomp(expr, re_flags); |
|
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8024 } |
| 4444 | 8025 |
| 8026 return prog; | |
| 8027 } | |
| 8028 | |
| 8029 /* | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8030 * 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
|
8031 */ |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8032 void |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8033 vim_regfree(prog) |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8034 regprog_T *prog; |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8035 { |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8036 if (prog != NULL) |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8037 prog->engine->regfree(prog); |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8038 } |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8039 |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8040 /* |
| 4444 | 8041 * Match a regexp against a string. |
| 8042 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 8043 * Uses curbuf for line count and 'iskeyword'. | |
| 8044 * | |
| 8045 * Return TRUE if there is a match, FALSE if not. | |
| 8046 */ | |
| 8047 int | |
| 8048 vim_regexec(rmp, line, col) | |
| 8049 regmatch_T *rmp; | |
| 8050 char_u *line; /* string to match against */ | |
| 8051 colnr_T col; /* column to start looking for match */ | |
| 8052 { | |
| 8053 return rmp->regprog->engine->regexec(rmp, line, col); | |
| 8054 } | |
| 8055 | |
| 8056 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
| 8057 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 8058 /* | |
| 8059 * Like vim_regexec(), but consider a "\n" in "line" to be a line break. | |
| 8060 */ | |
| 8061 int | |
| 8062 vim_regexec_nl(rmp, line, col) | |
| 8063 regmatch_T *rmp; | |
| 8064 char_u *line; | |
| 8065 colnr_T col; | |
| 8066 { | |
| 8067 return rmp->regprog->engine->regexec_nl(rmp, line, col); | |
| 8068 } | |
| 8069 #endif | |
| 8070 | |
| 8071 /* | |
| 8072 * Match a regexp against multiple lines. | |
| 8073 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 8074 * Uses curbuf for line count and 'iskeyword'. | |
| 8075 * | |
| 8076 * Return zero if there is no match. Return number of lines contained in the | |
| 8077 * match otherwise. | |
| 8078 */ | |
| 8079 long | |
| 8080 vim_regexec_multi(rmp, win, buf, lnum, col, tm) | |
| 8081 regmmatch_T *rmp; | |
| 8082 win_T *win; /* window in which to search or NULL */ | |
| 8083 buf_T *buf; /* buffer in which to search */ | |
| 8084 linenr_T lnum; /* nr of line to start looking for match */ | |
| 8085 colnr_T col; /* column to start looking for match */ | |
| 8086 proftime_T *tm; /* timeout limit or NULL */ | |
| 8087 { | |
| 8088 return rmp->regprog->engine->regexec_multi(rmp, win, buf, lnum, col, tm); | |
| 8089 } |