Mercurial > vim
annotate src/regexp.c @ 6677:77172f5bac73 v7.4.663
updated for version 7.4.663
Problem: When using netbeans a buffer is not found in another tab.
Solution: When 'switchbuf' is set to "usetab" then switch to another tab
when possible. (Xavier de Gaye)
| author | Bram Moolenaar <bram@vim.org> |
|---|---|
| date | Sat, 14 Mar 2015 15:35:52 +0100 |
| parents | bdc8e71633e4 |
| children | eefee03a37fe |
| rev | line source |
|---|---|
| 7 | 1 /* vi:set ts=8 sts=4 sw=4: |
| 2 * | |
| 3 * Handling of regular expressions: vim_regcomp(), vim_regexec(), vim_regsub() | |
| 4 * | |
| 5 * NOTICE: | |
| 6 * | |
| 7 * This is NOT the original regular expression code as written by Henry | |
| 8 * Spencer. This code has been modified specifically for use with the VIM | |
| 9 * editor, and should not be used separately from Vim. If you want a good | |
| 10 * regular expression library, get the original code. The copyright notice | |
| 11 * that follows is from the original. | |
| 12 * | |
| 13 * END NOTICE | |
| 14 * | |
| 15 * Copyright (c) 1986 by University of Toronto. | |
| 16 * Written by Henry Spencer. Not derived from licensed software. | |
| 17 * | |
| 18 * Permission is granted to anyone to use this software for any | |
| 19 * purpose on any computer system, and to redistribute it freely, | |
| 20 * subject to the following restrictions: | |
| 21 * | |
| 22 * 1. The author is not responsible for the consequences of use of | |
| 23 * this software, no matter how awful, even if they arise | |
| 24 * from defects in it. | |
| 25 * | |
| 26 * 2. The origin of this software must not be misrepresented, either | |
| 27 * by explicit claim or by omission. | |
| 28 * | |
| 29 * 3. Altered versions must be plainly marked as such, and must not | |
| 30 * be misrepresented as being the original software. | |
| 31 * | |
| 32 * Beware that some of this code is subtly aware of the way operator | |
| 33 * precedence is structured in regular expressions. Serious changes in | |
| 34 * regular-expression syntax might require a total rethink. | |
| 35 * | |
| 24 | 36 * Changes have been made by Tony Andrews, Olaf 'Rhialto' Seibert, Robert |
| 37 * Webb, Ciaran McCreesh and Bram Moolenaar. | |
| 7 | 38 * Named character class support added by Walter Briscoe (1998 Jul 01) |
| 39 */ | |
| 40 | |
| 4444 | 41 /* Uncomment the first if you do not want to see debugging logs or files |
| 42 * related to regular expressions, even when compiling with -DDEBUG. | |
| 43 * Uncomment the second to get the regexp debugging. */ | |
| 44 /* #undef DEBUG */ | |
| 45 /* #define DEBUG */ | |
| 46 | |
| 7 | 47 #include "vim.h" |
| 48 | |
| 4444 | 49 #ifdef DEBUG |
| 50 /* show/save debugging data when BT engine is used */ | |
| 51 # define BT_REGEXP_DUMP | |
| 52 /* save the debugging data to a file instead of displaying it */ | |
| 53 # define BT_REGEXP_LOG | |
| 4460 | 54 # define BT_REGEXP_DEBUG_LOG |
| 55 # define BT_REGEXP_DEBUG_LOG_NAME "bt_regexp_debug.log" | |
| 4444 | 56 #endif |
| 7 | 57 |
| 58 /* | |
| 59 * The "internal use only" fields in regexp.h are present to pass info from | |
| 60 * compile to execute that permits the execute phase to run lots faster on | |
| 61 * simple cases. They are: | |
| 62 * | |
| 63 * regstart char that must begin a match; NUL if none obvious; Can be a | |
| 64 * multi-byte character. | |
| 65 * reganch is the match anchored (at beginning-of-line only)? | |
| 66 * regmust string (pointer into program) that match must include, or NULL | |
| 67 * regmlen length of regmust string | |
| 68 * regflags RF_ values or'ed together | |
| 69 * | |
| 70 * Regstart and reganch permit very fast decisions on suitable starting points | |
| 71 * for a match, cutting down the work a lot. Regmust permits fast rejection | |
| 72 * of lines that cannot possibly match. The regmust tests are costly enough | |
| 73 * that vim_regcomp() supplies a regmust only if the r.e. contains something | |
| 74 * potentially expensive (at present, the only such thing detected is * or + | |
| 75 * at the start of the r.e., which can involve a lot of backup). Regmlen is | |
| 76 * supplied because the test in vim_regexec() needs it and vim_regcomp() is | |
| 77 * computing it anyway. | |
| 78 */ | |
| 79 | |
| 80 /* | |
| 81 * Structure for regexp "program". This is essentially a linear encoding | |
| 82 * of a nondeterministic finite-state machine (aka syntax charts or | |
| 83 * "railroad normal form" in parsing technology). Each node is an opcode | |
| 84 * plus a "next" pointer, possibly plus an operand. "Next" pointers of | |
| 85 * all nodes except BRANCH and BRACES_COMPLEX implement concatenation; a "next" | |
| 86 * pointer with a BRANCH on both ends of it is connecting two alternatives. | |
| 87 * (Here we have one of the subtle syntax dependencies: an individual BRANCH | |
| 88 * (as opposed to a collection of them) is never concatenated with anything | |
| 89 * because of operator precedence). The "next" pointer of a BRACES_COMPLEX | |
| 167 | 90 * node points to the node after the stuff to be repeated. |
| 91 * The operand of some types of node is a literal string; for others, it is a | |
| 92 * node leading into a sub-FSM. In particular, the operand of a BRANCH node | |
| 93 * is the first node of the branch. | |
| 94 * (NB this is *not* a tree structure: the tail of the branch connects to the | |
| 95 * thing following the set of BRANCHes.) | |
| 7 | 96 * |
| 97 * pattern is coded like: | |
| 98 * | |
| 99 * +-----------------+ | |
| 100 * | V | |
| 101 * <aa>\|<bb> BRANCH <aa> BRANCH <bb> --> END | |
| 102 * | ^ | ^ | |
| 103 * +------+ +----------+ | |
| 104 * | |
| 105 * | |
| 106 * +------------------+ | |
| 107 * V | | |
| 108 * <aa>* BRANCH BRANCH <aa> --> BACK BRANCH --> NOTHING --> END | |
| 109 * | | ^ ^ | |
| 110 * | +---------------+ | | |
| 111 * +---------------------------------------------+ | |
| 112 * | |
| 113 * | |
| 167 | 114 * +----------------------+ |
| 115 * V | | |
| 233 | 116 * <aa>\+ BRANCH <aa> --> BRANCH --> BACK BRANCH --> NOTHING --> END |
| 856 | 117 * | | ^ ^ |
| 118 * | +-----------+ | | |
| 179 | 119 * +--------------------------------------------------+ |
| 167 | 120 * |
| 121 * | |
| 7 | 122 * +-------------------------+ |
| 123 * V | | |
| 124 * <aa>\{} BRANCH BRACE_LIMITS --> BRACE_COMPLEX <aa> --> BACK END | |
| 125 * | | ^ | |
| 126 * | +----------------+ | |
| 127 * +-----------------------------------------------+ | |
| 128 * | |
| 129 * | |
| 130 * <aa>\@!<bb> BRANCH NOMATCH <aa> --> END <bb> --> END | |
| 131 * | | ^ ^ | |
| 132 * | +----------------+ | | |
| 133 * +--------------------------------+ | |
| 134 * | |
| 135 * +---------+ | |
| 136 * | V | |
| 137 * \z[abc] BRANCH BRANCH a BRANCH b BRANCH c BRANCH NOTHING --> END | |
| 138 * | | | | ^ ^ | |
| 139 * | | | +-----+ | | |
| 140 * | | +----------------+ | | |
| 141 * | +---------------------------+ | | |
| 142 * +------------------------------------------------------+ | |
| 143 * | |
| 1209 | 144 * They all start with a BRANCH for "\|" alternatives, even when there is only |
| 7 | 145 * one alternative. |
| 146 */ | |
| 147 | |
| 148 /* | |
| 149 * The opcodes are: | |
| 150 */ | |
| 151 | |
| 152 /* definition number opnd? meaning */ | |
| 153 #define END 0 /* End of program or NOMATCH operand. */ | |
| 154 #define BOL 1 /* Match "" at beginning of line. */ | |
| 155 #define EOL 2 /* Match "" at end of line. */ | |
| 156 #define BRANCH 3 /* node Match this alternative, or the | |
| 157 * next... */ | |
| 158 #define BACK 4 /* Match "", "next" ptr points backward. */ | |
| 159 #define EXACTLY 5 /* str Match this string. */ | |
| 160 #define NOTHING 6 /* Match empty string. */ | |
| 161 #define STAR 7 /* node Match this (simple) thing 0 or more | |
| 162 * times. */ | |
| 163 #define PLUS 8 /* node Match this (simple) thing 1 or more | |
| 164 * times. */ | |
| 165 #define MATCH 9 /* node match the operand zero-width */ | |
| 166 #define NOMATCH 10 /* node check for no match with operand */ | |
| 167 #define BEHIND 11 /* node look behind for a match with operand */ | |
| 168 #define NOBEHIND 12 /* node look behind for no match with operand */ | |
| 169 #define SUBPAT 13 /* node match the operand here */ | |
| 170 #define BRACE_SIMPLE 14 /* node Match this (simple) thing between m and | |
| 171 * n times (\{m,n\}). */ | |
| 172 #define BOW 15 /* Match "" after [^a-zA-Z0-9_] */ | |
| 173 #define EOW 16 /* Match "" at [^a-zA-Z0-9_] */ | |
| 174 #define BRACE_LIMITS 17 /* nr nr define the min & max for BRACE_SIMPLE | |
| 175 * and BRACE_COMPLEX. */ | |
| 176 #define NEWL 18 /* Match line-break */ | |
| 177 #define BHPOS 19 /* End position for BEHIND or NOBEHIND */ | |
| 178 | |
| 179 | |
| 180 /* character classes: 20-48 normal, 50-78 include a line-break */ | |
| 181 #define ADD_NL 30 | |
| 182 #define FIRST_NL ANY + ADD_NL | |
| 183 #define ANY 20 /* Match any one character. */ | |
| 184 #define ANYOF 21 /* str Match any character in this string. */ | |
| 185 #define ANYBUT 22 /* str Match any character not in this | |
| 186 * string. */ | |
| 187 #define IDENT 23 /* Match identifier char */ | |
| 188 #define SIDENT 24 /* Match identifier char but no digit */ | |
| 189 #define KWORD 25 /* Match keyword char */ | |
| 190 #define SKWORD 26 /* Match word char but no digit */ | |
| 191 #define FNAME 27 /* Match file name char */ | |
| 192 #define SFNAME 28 /* Match file name char but no digit */ | |
| 193 #define PRINT 29 /* Match printable char */ | |
| 194 #define SPRINT 30 /* Match printable char but no digit */ | |
| 195 #define WHITE 31 /* Match whitespace char */ | |
| 196 #define NWHITE 32 /* Match non-whitespace char */ | |
| 197 #define DIGIT 33 /* Match digit char */ | |
| 198 #define NDIGIT 34 /* Match non-digit char */ | |
| 199 #define HEX 35 /* Match hex char */ | |
| 200 #define NHEX 36 /* Match non-hex char */ | |
| 201 #define OCTAL 37 /* Match octal char */ | |
| 202 #define NOCTAL 38 /* Match non-octal char */ | |
| 203 #define WORD 39 /* Match word char */ | |
| 204 #define NWORD 40 /* Match non-word char */ | |
| 205 #define HEAD 41 /* Match head char */ | |
| 206 #define NHEAD 42 /* Match non-head char */ | |
| 207 #define ALPHA 43 /* Match alpha char */ | |
| 208 #define NALPHA 44 /* Match non-alpha char */ | |
| 209 #define LOWER 45 /* Match lowercase char */ | |
| 210 #define NLOWER 46 /* Match non-lowercase char */ | |
| 211 #define UPPER 47 /* Match uppercase char */ | |
| 212 #define NUPPER 48 /* Match non-uppercase char */ | |
| 213 #define LAST_NL NUPPER + ADD_NL | |
| 214 #define WITH_NL(op) ((op) >= FIRST_NL && (op) <= LAST_NL) | |
| 215 | |
| 216 #define MOPEN 80 /* -89 Mark this point in input as start of | |
| 217 * \( subexpr. MOPEN + 0 marks start of | |
| 218 * match. */ | |
| 219 #define MCLOSE 90 /* -99 Analogous to MOPEN. MCLOSE + 0 marks | |
| 220 * end of match. */ | |
| 221 #define BACKREF 100 /* -109 node Match same string again \1-\9 */ | |
| 222 | |
| 223 #ifdef FEAT_SYN_HL | |
| 224 # define ZOPEN 110 /* -119 Mark this point in input as start of | |
| 225 * \z( subexpr. */ | |
| 226 # define ZCLOSE 120 /* -129 Analogous to ZOPEN. */ | |
| 227 # define ZREF 130 /* -139 node Match external submatch \z1-\z9 */ | |
| 228 #endif | |
| 229 | |
| 230 #define BRACE_COMPLEX 140 /* -149 node Match nodes between m & n times */ | |
| 231 | |
| 232 #define NOPEN 150 /* Mark this point in input as start of | |
| 233 \%( subexpr. */ | |
| 234 #define NCLOSE 151 /* Analogous to NOPEN. */ | |
| 235 | |
| 236 #define MULTIBYTECODE 200 /* mbc Match one multi-byte character */ | |
| 237 #define RE_BOF 201 /* Match "" at beginning of file. */ | |
| 238 #define RE_EOF 202 /* Match "" at end of file. */ | |
| 239 #define CURSOR 203 /* Match location of cursor. */ | |
| 240 | |
| 241 #define RE_LNUM 204 /* nr cmp Match line number */ | |
| 242 #define RE_COL 205 /* nr cmp Match column number */ | |
| 243 #define RE_VCOL 206 /* nr cmp Match virtual column number */ | |
| 244 | |
| 639 | 245 #define RE_MARK 207 /* mark cmp Match mark position */ |
| 246 #define RE_VISUAL 208 /* Match Visual area */ | |
| 5901 | 247 #define RE_COMPOSING 209 /* any composing characters */ |
| 639 | 248 |
| 7 | 249 /* |
| 250 * Magic characters have a special meaning, they don't match literally. | |
| 251 * Magic characters are negative. This separates them from literal characters | |
| 252 * (possibly multi-byte). Only ASCII characters can be Magic. | |
| 253 */ | |
| 254 #define Magic(x) ((int)(x) - 256) | |
| 255 #define un_Magic(x) ((x) + 256) | |
| 256 #define is_Magic(x) ((x) < 0) | |
| 257 | |
| 258 static int no_Magic __ARGS((int x)); | |
| 259 static int toggle_Magic __ARGS((int x)); | |
| 260 | |
| 261 static int | |
| 262 no_Magic(x) | |
| 263 int x; | |
| 264 { | |
| 265 if (is_Magic(x)) | |
| 266 return un_Magic(x); | |
| 267 return x; | |
| 268 } | |
| 269 | |
| 270 static int | |
| 271 toggle_Magic(x) | |
| 272 int x; | |
| 273 { | |
| 274 if (is_Magic(x)) | |
| 275 return un_Magic(x); | |
| 276 return Magic(x); | |
| 277 } | |
| 278 | |
| 279 /* | |
| 280 * The first byte of the regexp internal "program" is actually this magic | |
| 281 * number; the start node begins in the second byte. It's used to catch the | |
| 282 * most severe mutilation of the program by the caller. | |
| 283 */ | |
| 284 | |
| 285 #define REGMAGIC 0234 | |
| 286 | |
| 287 /* | |
| 288 * Opcode notes: | |
| 289 * | |
| 290 * BRANCH The set of branches constituting a single choice are hooked | |
| 291 * together with their "next" pointers, since precedence prevents | |
| 292 * anything being concatenated to any individual branch. The | |
| 293 * "next" pointer of the last BRANCH in a choice points to the | |
| 294 * thing following the whole choice. This is also where the | |
| 295 * final "next" pointer of each individual branch points; each | |
| 296 * branch starts with the operand node of a BRANCH node. | |
| 297 * | |
| 298 * BACK Normal "next" pointers all implicitly point forward; BACK | |
| 299 * exists to make loop structures possible. | |
| 300 * | |
| 301 * STAR,PLUS '=', and complex '*' and '+', are implemented as circular | |
| 302 * BRANCH structures using BACK. Simple cases (one character | |
| 303 * per match) are implemented with STAR and PLUS for speed | |
| 304 * and to minimize recursive plunges. | |
| 305 * | |
| 306 * BRACE_LIMITS This is always followed by a BRACE_SIMPLE or BRACE_COMPLEX | |
| 307 * node, and defines the min and max limits to be used for that | |
| 308 * node. | |
| 309 * | |
| 310 * MOPEN,MCLOSE ...are numbered at compile time. | |
| 311 * ZOPEN,ZCLOSE ...ditto | |
| 312 */ | |
| 313 | |
| 314 /* | |
| 315 * A node is one char of opcode followed by two chars of "next" pointer. | |
| 316 * "Next" pointers are stored as two 8-bit bytes, high order first. The | |
| 317 * value is a positive offset from the opcode of the node containing it. | |
| 318 * An operand, if any, simply follows the node. (Note that much of the | |
| 319 * code generation knows about this implicit relationship.) | |
| 320 * | |
| 321 * Using two bytes for the "next" pointer is vast overkill for most things, | |
| 322 * but allows patterns to get big without disasters. | |
| 323 */ | |
| 324 #define OP(p) ((int)*(p)) | |
| 325 #define NEXT(p) (((*((p) + 1) & 0377) << 8) + (*((p) + 2) & 0377)) | |
| 326 #define OPERAND(p) ((p) + 3) | |
| 327 /* Obtain an operand that was stored as four bytes, MSB first. */ | |
| 328 #define OPERAND_MIN(p) (((long)(p)[3] << 24) + ((long)(p)[4] << 16) \ | |
| 329 + ((long)(p)[5] << 8) + (long)(p)[6]) | |
| 330 /* Obtain a second operand stored as four bytes. */ | |
| 331 #define OPERAND_MAX(p) OPERAND_MIN((p) + 4) | |
| 332 /* Obtain a second single-byte operand stored after a four bytes operand. */ | |
| 333 #define OPERAND_CMP(p) (p)[7] | |
| 334 | |
| 335 /* | |
| 336 * Utility definitions. | |
| 337 */ | |
| 338 #define UCHARAT(p) ((int)*(char_u *)(p)) | |
| 339 | |
| 340 /* Used for an error (down from) vim_regcomp(): give the error message, set | |
| 341 * rc_did_emsg and return NULL */ | |
| 653 | 342 #define EMSG_RET_NULL(m) return (EMSG(m), rc_did_emsg = TRUE, (void *)NULL) |
| 308 | 343 #define EMSG_RET_FAIL(m) return (EMSG(m), rc_did_emsg = TRUE, FAIL) |
| 4444 | 344 #define EMSG2_RET_NULL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, (void *)NULL) |
| 345 #define EMSG2_RET_FAIL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, FAIL) | |
| 346 #define EMSG_ONE_RET_NULL EMSG2_RET_NULL(_("E369: invalid item in %s%%[]"), reg_magic == MAGIC_ALL) | |
| 7 | 347 |
| 348 #define MAX_LIMIT (32767L << 16L) | |
| 349 | |
| 350 static int re_multi_type __ARGS((int)); | |
| 351 static int cstrncmp __ARGS((char_u *s1, char_u *s2, int *n)); | |
| 352 static char_u *cstrchr __ARGS((char_u *, int)); | |
| 353 | |
| 4444 | 354 #ifdef BT_REGEXP_DUMP |
| 355 static void regdump __ARGS((char_u *, bt_regprog_T *)); | |
| 356 #endif | |
| 7 | 357 #ifdef DEBUG |
| 358 static char_u *regprop __ARGS((char_u *)); | |
| 359 #endif | |
| 360 | |
| 6203 | 361 static int re_mult_next __ARGS((char *what)); |
| 362 | |
| 4444 | 363 static char_u e_missingbracket[] = N_("E769: Missing ] after %s["); |
| 364 static char_u e_unmatchedpp[] = N_("E53: Unmatched %s%%("); | |
| 365 static char_u e_unmatchedp[] = N_("E54: Unmatched %s("); | |
| 366 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
|
367 #ifdef FEAT_SYN_HL |
|
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
368 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
|
369 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
|
370 #endif |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
371 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
|
372 static char_u e_empty_sb[] = N_("E70: Empty %s%%[]"); |
| 7 | 373 #define NOT_MULTI 0 |
| 374 #define MULTI_ONE 1 | |
| 375 #define MULTI_MULT 2 | |
| 376 /* | |
| 377 * Return NOT_MULTI if c is not a "multi" operator. | |
| 378 * Return MULTI_ONE if c is a single "multi" operator. | |
| 379 * Return MULTI_MULT if c is a multi "multi" operator. | |
| 380 */ | |
| 381 static int | |
| 382 re_multi_type(c) | |
| 383 int c; | |
| 384 { | |
| 385 if (c == Magic('@') || c == Magic('=') || c == Magic('?')) | |
| 386 return MULTI_ONE; | |
| 387 if (c == Magic('*') || c == Magic('+') || c == Magic('{')) | |
| 388 return MULTI_MULT; | |
| 389 return NOT_MULTI; | |
| 390 } | |
| 391 | |
| 392 /* | |
| 393 * Flags to be passed up and down. | |
| 394 */ | |
| 395 #define HASWIDTH 0x1 /* Known never to match null string. */ | |
| 396 #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ | |
| 397 #define SPSTART 0x4 /* Starts with * or +. */ | |
| 398 #define HASNL 0x8 /* Contains some \n. */ | |
| 399 #define HASLOOKBH 0x10 /* Contains "\@<=" or "\@<!". */ | |
| 400 #define WORST 0 /* Worst case. */ | |
| 401 | |
| 402 /* | |
| 403 * When regcode is set to this value, code is not emitted and size is computed | |
| 404 * instead. | |
| 405 */ | |
| 406 #define JUST_CALC_SIZE ((char_u *) -1) | |
| 407 | |
| 359 | 408 static char_u *reg_prev_sub = NULL; |
| 409 | |
| 7 | 410 /* |
| 411 * REGEXP_INRANGE contains all characters which are always special in a [] | |
| 412 * range after '\'. | |
| 413 * REGEXP_ABBR contains all characters which act as abbreviations after '\'. | |
| 414 * These are: | |
| 415 * \n - New line (NL). | |
| 416 * \r - Carriage Return (CR). | |
| 417 * \t - Tab (TAB). | |
| 418 * \e - Escape (ESC). | |
| 419 * \b - Backspace (Ctrl_H). | |
| 24 | 420 * \d - Character code in decimal, eg \d123 |
| 421 * \o - Character code in octal, eg \o80 | |
| 422 * \x - Character code in hex, eg \x4a | |
| 423 * \u - Multibyte character code, eg \u20ac | |
| 424 * \U - Long multibyte character code, eg \U12345678 | |
| 7 | 425 */ |
| 426 static char_u REGEXP_INRANGE[] = "]^-n\\"; | |
| 24 | 427 static char_u REGEXP_ABBR[] = "nrtebdoxuU"; |
| 7 | 428 |
| 429 static int backslash_trans __ARGS((int c)); | |
| 167 | 430 static int get_char_class __ARGS((char_u **pp)); |
| 431 static int get_equi_class __ARGS((char_u **pp)); | |
| 432 static void reg_equi_class __ARGS((int c)); | |
| 433 static int get_coll_element __ARGS((char_u **pp)); | |
| 7 | 434 static char_u *skip_anyof __ARGS((char_u *p)); |
| 435 static void init_class_tab __ARGS((void)); | |
| 436 | |
| 437 /* | |
| 438 * Translate '\x' to its control character, except "\n", which is Magic. | |
| 439 */ | |
| 440 static int | |
| 441 backslash_trans(c) | |
| 442 int c; | |
| 443 { | |
| 444 switch (c) | |
| 445 { | |
| 446 case 'r': return CAR; | |
| 447 case 't': return TAB; | |
| 448 case 'e': return ESC; | |
| 449 case 'b': return BS; | |
| 450 } | |
| 451 return c; | |
| 452 } | |
| 453 | |
| 454 /* | |
| 167 | 455 * Check for a character class name "[:name:]". "pp" points to the '['. |
| 7 | 456 * Returns one of the CLASS_ items. CLASS_NONE means that no item was |
| 457 * recognized. Otherwise "pp" is advanced to after the item. | |
| 458 */ | |
| 459 static int | |
| 167 | 460 get_char_class(pp) |
| 7 | 461 char_u **pp; |
| 462 { | |
| 463 static const char *(class_names[]) = | |
| 464 { | |
| 465 "alnum:]", | |
| 466 #define CLASS_ALNUM 0 | |
| 467 "alpha:]", | |
| 468 #define CLASS_ALPHA 1 | |
| 469 "blank:]", | |
| 470 #define CLASS_BLANK 2 | |
| 471 "cntrl:]", | |
| 472 #define CLASS_CNTRL 3 | |
| 473 "digit:]", | |
| 474 #define CLASS_DIGIT 4 | |
| 475 "graph:]", | |
| 476 #define CLASS_GRAPH 5 | |
| 477 "lower:]", | |
| 478 #define CLASS_LOWER 6 | |
| 479 "print:]", | |
| 480 #define CLASS_PRINT 7 | |
| 481 "punct:]", | |
| 482 #define CLASS_PUNCT 8 | |
| 483 "space:]", | |
| 484 #define CLASS_SPACE 9 | |
| 485 "upper:]", | |
| 486 #define CLASS_UPPER 10 | |
| 487 "xdigit:]", | |
| 488 #define CLASS_XDIGIT 11 | |
| 489 "tab:]", | |
| 490 #define CLASS_TAB 12 | |
| 491 "return:]", | |
| 492 #define CLASS_RETURN 13 | |
| 493 "backspace:]", | |
| 494 #define CLASS_BACKSPACE 14 | |
| 495 "escape:]", | |
| 496 #define CLASS_ESCAPE 15 | |
| 497 }; | |
| 498 #define CLASS_NONE 99 | |
| 499 int i; | |
| 500 | |
| 501 if ((*pp)[1] == ':') | |
| 502 { | |
| 1877 | 503 for (i = 0; i < (int)(sizeof(class_names) / sizeof(*class_names)); ++i) |
| 7 | 504 if (STRNCMP(*pp + 2, class_names[i], STRLEN(class_names[i])) == 0) |
| 505 { | |
| 506 *pp += STRLEN(class_names[i]) + 2; | |
| 507 return i; | |
| 508 } | |
| 509 } | |
| 510 return CLASS_NONE; | |
| 511 } | |
| 512 | |
| 513 /* | |
| 514 * Specific version of character class functions. | |
| 515 * Using a table to keep this fast. | |
| 516 */ | |
| 517 static short class_tab[256]; | |
| 518 | |
| 519 #define RI_DIGIT 0x01 | |
| 520 #define RI_HEX 0x02 | |
| 521 #define RI_OCTAL 0x04 | |
| 522 #define RI_WORD 0x08 | |
| 523 #define RI_HEAD 0x10 | |
| 524 #define RI_ALPHA 0x20 | |
| 525 #define RI_LOWER 0x40 | |
| 526 #define RI_UPPER 0x80 | |
| 527 #define RI_WHITE 0x100 | |
| 528 | |
| 529 static void | |
| 530 init_class_tab() | |
| 531 { | |
| 532 int i; | |
| 533 static int done = FALSE; | |
| 534 | |
| 535 if (done) | |
| 536 return; | |
| 537 | |
| 538 for (i = 0; i < 256; ++i) | |
| 539 { | |
| 540 if (i >= '0' && i <= '7') | |
| 541 class_tab[i] = RI_DIGIT + RI_HEX + RI_OCTAL + RI_WORD; | |
| 542 else if (i >= '8' && i <= '9') | |
| 543 class_tab[i] = RI_DIGIT + RI_HEX + RI_WORD; | |
| 544 else if (i >= 'a' && i <= 'f') | |
| 545 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
| 546 #ifdef EBCDIC | |
| 547 else if ((i >= 'g' && i <= 'i') || (i >= 'j' && i <= 'r') | |
| 548 || (i >= 's' && i <= 'z')) | |
| 549 #else | |
| 550 else if (i >= 'g' && i <= 'z') | |
| 551 #endif | |
| 552 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
| 553 else if (i >= 'A' && i <= 'F') | |
| 554 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
| 555 #ifdef EBCDIC | |
| 556 else if ((i >= 'G' && i <= 'I') || ( i >= 'J' && i <= 'R') | |
| 557 || (i >= 'S' && i <= 'Z')) | |
| 558 #else | |
| 559 else if (i >= 'G' && i <= 'Z') | |
| 560 #endif | |
| 561 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
| 562 else if (i == '_') | |
| 563 class_tab[i] = RI_WORD + RI_HEAD; | |
| 564 else | |
| 565 class_tab[i] = 0; | |
| 566 } | |
| 567 class_tab[' '] |= RI_WHITE; | |
| 568 class_tab['\t'] |= RI_WHITE; | |
| 569 done = TRUE; | |
| 570 } | |
| 571 | |
| 572 #ifdef FEAT_MBYTE | |
| 573 # define ri_digit(c) (c < 0x100 && (class_tab[c] & RI_DIGIT)) | |
| 574 # define ri_hex(c) (c < 0x100 && (class_tab[c] & RI_HEX)) | |
| 575 # define ri_octal(c) (c < 0x100 && (class_tab[c] & RI_OCTAL)) | |
| 576 # define ri_word(c) (c < 0x100 && (class_tab[c] & RI_WORD)) | |
| 577 # define ri_head(c) (c < 0x100 && (class_tab[c] & RI_HEAD)) | |
| 578 # define ri_alpha(c) (c < 0x100 && (class_tab[c] & RI_ALPHA)) | |
| 579 # define ri_lower(c) (c < 0x100 && (class_tab[c] & RI_LOWER)) | |
| 580 # define ri_upper(c) (c < 0x100 && (class_tab[c] & RI_UPPER)) | |
| 581 # define ri_white(c) (c < 0x100 && (class_tab[c] & RI_WHITE)) | |
| 582 #else | |
| 583 # define ri_digit(c) (class_tab[c] & RI_DIGIT) | |
| 584 # define ri_hex(c) (class_tab[c] & RI_HEX) | |
| 585 # define ri_octal(c) (class_tab[c] & RI_OCTAL) | |
| 586 # define ri_word(c) (class_tab[c] & RI_WORD) | |
| 587 # define ri_head(c) (class_tab[c] & RI_HEAD) | |
| 588 # define ri_alpha(c) (class_tab[c] & RI_ALPHA) | |
| 589 # define ri_lower(c) (class_tab[c] & RI_LOWER) | |
| 590 # define ri_upper(c) (class_tab[c] & RI_UPPER) | |
| 591 # define ri_white(c) (class_tab[c] & RI_WHITE) | |
| 592 #endif | |
| 593 | |
| 594 /* flags for regflags */ | |
| 595 #define RF_ICASE 1 /* ignore case */ | |
| 596 #define RF_NOICASE 2 /* don't ignore case */ | |
| 597 #define RF_HASNL 4 /* can match a NL */ | |
| 598 #define RF_ICOMBINE 8 /* ignore combining characters */ | |
| 599 #define RF_LOOKBH 16 /* uses "\@<=" or "\@<!" */ | |
| 600 | |
| 601 /* | |
| 602 * Global work variables for vim_regcomp(). | |
| 603 */ | |
| 604 | |
| 605 static char_u *regparse; /* Input-scan pointer. */ | |
| 606 static int prevchr_len; /* byte length of previous char */ | |
| 607 static int num_complex_braces; /* Complex \{...} count */ | |
| 608 static int regnpar; /* () count. */ | |
| 609 #ifdef FEAT_SYN_HL | |
| 610 static int regnzpar; /* \z() count. */ | |
| 611 static int re_has_z; /* \z item detected */ | |
| 612 #endif | |
| 613 static char_u *regcode; /* Code-emit pointer, or JUST_CALC_SIZE */ | |
| 614 static long regsize; /* Code size. */ | |
| 2010 | 615 static int reg_toolong; /* TRUE when offset out of range */ |
| 7 | 616 static char_u had_endbrace[NSUBEXP]; /* flags, TRUE if end of () found */ |
| 617 static unsigned regflags; /* RF_ flags for prog */ | |
| 618 static long brace_min[10]; /* Minimums for complex brace repeats */ | |
| 619 static long brace_max[10]; /* Maximums for complex brace repeats */ | |
| 620 static int brace_count[10]; /* Current counts for complex brace repeats */ | |
| 621 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 622 static int had_eol; /* TRUE when EOL found by vim_regcomp() */ | |
| 623 #endif | |
| 624 static int one_exactly = FALSE; /* only do one char for EXACTLY */ | |
| 625 | |
| 626 static int reg_magic; /* magicness of the pattern: */ | |
| 627 #define MAGIC_NONE 1 /* "\V" very unmagic */ | |
| 628 #define MAGIC_OFF 2 /* "\M" or 'magic' off */ | |
| 629 #define MAGIC_ON 3 /* "\m" or 'magic' */ | |
| 630 #define MAGIC_ALL 4 /* "\v" very magic */ | |
| 631 | |
| 632 static int reg_string; /* matching with a string instead of a buffer | |
| 633 line */ | |
| 481 | 634 static int reg_strict; /* "[abc" is illegal */ |
| 7 | 635 |
| 636 /* | |
| 637 * META contains all characters that may be magic, except '^' and '$'. | |
| 638 */ | |
| 639 | |
| 640 #ifdef EBCDIC | |
| 641 static char_u META[] = "%&()*+.123456789<=>?@ACDFHIKLMOPSUVWX[_acdfhiklmnopsuvwxz{|~"; | |
| 642 #else | |
| 643 /* META[] is used often enough to justify turning it into a table. */ | |
| 644 static char_u META_flags[] = { | |
| 645 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| 646 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| 647 /* % & ( ) * + . */ | |
| 648 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0, | |
| 649 /* 1 2 3 4 5 6 7 8 9 < = > ? */ | |
| 650 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, | |
| 651 /* @ A C D F H I K L M O */ | |
| 652 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1, | |
| 653 /* P S U V W X Z [ _ */ | |
| 654 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, | |
| 655 /* a c d f h i k l m n o */ | |
| 656 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, | |
| 657 /* p s u v w x z { | ~ */ | |
| 658 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1 | |
| 659 }; | |
| 660 #endif | |
| 661 | |
| 4444 | 662 static int curchr; /* currently parsed character */ |
| 663 /* Previous character. Note: prevchr is sometimes -1 when we are not at the | |
| 664 * start, eg in /[ ^I]^ the pattern was never found even if it existed, | |
| 665 * because ^ was taken to be magic -- webb */ | |
| 666 static int prevchr; | |
| 667 static int prevprevchr; /* previous-previous character */ | |
| 668 static int nextchr; /* used for ungetchr() */ | |
| 7 | 669 |
| 670 /* arguments for reg() */ | |
| 671 #define REG_NOPAREN 0 /* toplevel reg() */ | |
| 672 #define REG_PAREN 1 /* \(\) */ | |
| 673 #define REG_ZPAREN 2 /* \z(\) */ | |
| 674 #define REG_NPAREN 3 /* \%(\) */ | |
| 675 | |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
676 typedef struct |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
677 { |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
678 char_u *regparse; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
679 int prevchr_len; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
680 int curchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
681 int prevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
682 int prevprevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
683 int nextchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
684 int at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
685 int prev_at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
686 int regnpar; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
687 } parse_state_T; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
688 |
| 7 | 689 /* |
| 690 * Forward declarations for vim_regcomp()'s friends. | |
| 691 */ | |
| 692 static void initchr __ARGS((char_u *)); | |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
693 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
|
694 static void restore_parse_state __ARGS((parse_state_T *ps)); |
| 7 | 695 static int getchr __ARGS((void)); |
| 696 static void skipchr_keepstart __ARGS((void)); | |
| 697 static int peekchr __ARGS((void)); | |
| 698 static void skipchr __ARGS((void)); | |
| 699 static void ungetchr __ARGS((void)); | |
| 24 | 700 static int gethexchrs __ARGS((int maxinputlen)); |
| 701 static int getoctchrs __ARGS((void)); | |
| 702 static int getdecchrs __ARGS((void)); | |
| 703 static int coll_get_char __ARGS((void)); | |
| 7 | 704 static void regcomp_start __ARGS((char_u *expr, int flags)); |
| 705 static char_u *reg __ARGS((int, int *)); | |
| 706 static char_u *regbranch __ARGS((int *flagp)); | |
| 707 static char_u *regconcat __ARGS((int *flagp)); | |
| 708 static char_u *regpiece __ARGS((int *)); | |
| 709 static char_u *regatom __ARGS((int *)); | |
| 710 static char_u *regnode __ARGS((int)); | |
| 714 | 711 #ifdef FEAT_MBYTE |
| 712 static int use_multibytecode __ARGS((int c)); | |
| 713 #endif | |
| 7 | 714 static int prog_magic_wrong __ARGS((void)); |
| 715 static char_u *regnext __ARGS((char_u *)); | |
| 716 static void regc __ARGS((int b)); | |
| 717 #ifdef FEAT_MBYTE | |
| 718 static void regmbc __ARGS((int c)); | |
| 2974 | 719 # define REGMBC(x) regmbc(x); |
| 720 # define CASEMBC(x) case x: | |
| 167 | 721 #else |
| 722 # define regmbc(c) regc(c) | |
| 2974 | 723 # define REGMBC(x) |
| 724 # define CASEMBC(x) | |
| 7 | 725 #endif |
| 726 static void reginsert __ARGS((int, char_u *)); | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
727 static void reginsert_nr __ARGS((int op, long val, char_u *opnd)); |
| 7 | 728 static void reginsert_limits __ARGS((int, long, long, char_u *)); |
| 729 static char_u *re_put_long __ARGS((char_u *pr, long_u val)); | |
| 730 static int read_limits __ARGS((long *, long *)); | |
| 731 static void regtail __ARGS((char_u *, char_u *)); | |
| 732 static void regoptail __ARGS((char_u *, char_u *)); | |
| 733 | |
| 4444 | 734 static regengine_T bt_regengine; |
| 735 static regengine_T nfa_regengine; | |
| 736 | |
| 7 | 737 /* |
| 738 * Return TRUE if compiled regular expression "prog" can match a line break. | |
| 739 */ | |
| 740 int | |
| 741 re_multiline(prog) | |
| 742 regprog_T *prog; | |
| 743 { | |
| 744 return (prog->regflags & RF_HASNL); | |
| 745 } | |
| 746 | |
| 747 /* | |
| 748 * Return TRUE if compiled regular expression "prog" looks before the start | |
| 749 * position (pattern contains "\@<=" or "\@<!"). | |
| 750 */ | |
| 751 int | |
| 752 re_lookbehind(prog) | |
| 753 regprog_T *prog; | |
| 754 { | |
| 755 return (prog->regflags & RF_LOOKBH); | |
| 756 } | |
| 757 | |
| 758 /* | |
| 167 | 759 * Check for an equivalence class name "[=a=]". "pp" points to the '['. |
| 760 * Returns a character representing the class. Zero means that no item was | |
| 761 * recognized. Otherwise "pp" is advanced to after the item. | |
| 762 */ | |
| 763 static int | |
| 764 get_equi_class(pp) | |
| 765 char_u **pp; | |
| 766 { | |
| 767 int c; | |
| 768 int l = 1; | |
| 769 char_u *p = *pp; | |
| 770 | |
| 771 if (p[1] == '=') | |
| 772 { | |
| 773 #ifdef FEAT_MBYTE | |
| 774 if (has_mbyte) | |
| 474 | 775 l = (*mb_ptr2len)(p + 2); |
| 167 | 776 #endif |
| 777 if (p[l + 2] == '=' && p[l + 3] == ']') | |
| 778 { | |
| 779 #ifdef FEAT_MBYTE | |
| 780 if (has_mbyte) | |
| 781 c = mb_ptr2char(p + 2); | |
| 782 else | |
| 783 #endif | |
| 784 c = p[2]; | |
| 785 *pp += l + 4; | |
| 786 return c; | |
| 787 } | |
| 788 } | |
| 789 return 0; | |
| 790 } | |
| 791 | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
792 #ifdef EBCDIC |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
793 /* |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
794 * 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
|
795 */ |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
796 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
|
797 "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
|
798 "C\x68", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
799 "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
|
800 "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
|
801 "N\x69", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
802 "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
|
803 "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
|
804 "Y\xBA", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
805 "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
|
806 "c\x48", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
807 "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
|
808 "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
|
809 "n\x49", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
810 "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
|
811 "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
|
812 "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
|
813 }; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
814 #endif |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
815 |
| 167 | 816 /* |
| 817 * Produce the bytes for equivalence class "c". | |
| 818 * Currently only handles latin1, latin9 and utf-8. | |
| 4444 | 819 * NOTE: When changing this function, also change nfa_emit_equi_class() |
| 167 | 820 */ |
| 821 static void | |
| 822 reg_equi_class(c) | |
| 823 int c; | |
| 824 { | |
| 825 #ifdef FEAT_MBYTE | |
| 826 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
| 492 | 827 || STRCMP(p_enc, "iso-8859-15") == 0) |
| 167 | 828 #endif |
| 829 { | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
830 #ifdef EBCDIC |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
831 int 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 /* 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
|
834 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
|
835 { |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
836 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
|
837 { |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
838 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
|
839 |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
840 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
|
841 regmbc(*p++); |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
842 return; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
843 } |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
844 } |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
845 #else |
| 167 | 846 switch (c) |
| 847 { | |
| 236 | 848 case 'A': case '\300': case '\301': case '\302': |
| 2974 | 849 CASEMBC(0x100) CASEMBC(0x102) CASEMBC(0x104) CASEMBC(0x1cd) |
| 850 CASEMBC(0x1de) CASEMBC(0x1e0) CASEMBC(0x1ea2) | |
| 236 | 851 case '\303': case '\304': case '\305': |
| 852 regmbc('A'); regmbc('\300'); regmbc('\301'); | |
| 853 regmbc('\302'); regmbc('\303'); regmbc('\304'); | |
| 854 regmbc('\305'); | |
| 2974 | 855 REGMBC(0x100) REGMBC(0x102) REGMBC(0x104) |
| 856 REGMBC(0x1cd) REGMBC(0x1de) REGMBC(0x1e0) | |
| 857 REGMBC(0x1ea2) | |
| 858 return; | |
| 859 case 'B': CASEMBC(0x1e02) CASEMBC(0x1e06) | |
| 860 regmbc('B'); REGMBC(0x1e02) REGMBC(0x1e06) | |
| 167 | 861 return; |
| 236 | 862 case 'C': case '\307': |
| 2974 | 863 CASEMBC(0x106) CASEMBC(0x108) CASEMBC(0x10a) CASEMBC(0x10c) |
| 236 | 864 regmbc('C'); regmbc('\307'); |
| 2974 | 865 REGMBC(0x106) REGMBC(0x108) REGMBC(0x10a) |
| 866 REGMBC(0x10c) | |
| 867 return; | |
| 868 case 'D': CASEMBC(0x10e) CASEMBC(0x110) CASEMBC(0x1e0a) | |
| 869 CASEMBC(0x1e0e) CASEMBC(0x1e10) | |
| 870 regmbc('D'); REGMBC(0x10e) REGMBC(0x110) | |
| 871 REGMBC(0x1e0a) REGMBC(0x1e0e) REGMBC(0x1e10) | |
| 167 | 872 return; |
| 236 | 873 case 'E': case '\310': case '\311': case '\312': case '\313': |
| 2974 | 874 CASEMBC(0x112) CASEMBC(0x114) CASEMBC(0x116) CASEMBC(0x118) |
| 875 CASEMBC(0x11a) CASEMBC(0x1eba) CASEMBC(0x1ebc) | |
| 236 | 876 regmbc('E'); regmbc('\310'); regmbc('\311'); |
| 877 regmbc('\312'); regmbc('\313'); | |
| 2974 | 878 REGMBC(0x112) REGMBC(0x114) REGMBC(0x116) |
| 879 REGMBC(0x118) REGMBC(0x11a) REGMBC(0x1eba) | |
| 880 REGMBC(0x1ebc) | |
| 881 return; | |
| 882 case 'F': CASEMBC(0x1e1e) | |
| 883 regmbc('F'); REGMBC(0x1e1e) | |
| 884 return; | |
| 885 case 'G': CASEMBC(0x11c) CASEMBC(0x11e) CASEMBC(0x120) | |
| 886 CASEMBC(0x122) CASEMBC(0x1e4) CASEMBC(0x1e6) CASEMBC(0x1f4) | |
| 887 CASEMBC(0x1e20) | |
| 888 regmbc('G'); REGMBC(0x11c) REGMBC(0x11e) | |
| 889 REGMBC(0x120) REGMBC(0x122) REGMBC(0x1e4) | |
| 890 REGMBC(0x1e6) REGMBC(0x1f4) REGMBC(0x1e20) | |
| 891 return; | |
| 892 case 'H': CASEMBC(0x124) CASEMBC(0x126) CASEMBC(0x1e22) | |
| 893 CASEMBC(0x1e26) CASEMBC(0x1e28) | |
| 894 regmbc('H'); REGMBC(0x124) REGMBC(0x126) | |
| 895 REGMBC(0x1e22) REGMBC(0x1e26) REGMBC(0x1e28) | |
| 167 | 896 return; |
| 236 | 897 case 'I': case '\314': case '\315': case '\316': case '\317': |
| 2974 | 898 CASEMBC(0x128) CASEMBC(0x12a) CASEMBC(0x12c) CASEMBC(0x12e) |
| 899 CASEMBC(0x130) CASEMBC(0x1cf) CASEMBC(0x1ec8) | |
| 236 | 900 regmbc('I'); regmbc('\314'); regmbc('\315'); |
| 901 regmbc('\316'); regmbc('\317'); | |
| 2974 | 902 REGMBC(0x128) REGMBC(0x12a) REGMBC(0x12c) |
| 903 REGMBC(0x12e) REGMBC(0x130) REGMBC(0x1cf) | |
| 904 REGMBC(0x1ec8) | |
| 905 return; | |
| 906 case 'J': CASEMBC(0x134) | |
| 907 regmbc('J'); REGMBC(0x134) | |
| 908 return; | |
| 909 case 'K': CASEMBC(0x136) CASEMBC(0x1e8) CASEMBC(0x1e30) | |
| 910 CASEMBC(0x1e34) | |
| 911 regmbc('K'); REGMBC(0x136) REGMBC(0x1e8) | |
| 912 REGMBC(0x1e30) REGMBC(0x1e34) | |
| 913 return; | |
| 914 case 'L': CASEMBC(0x139) CASEMBC(0x13b) CASEMBC(0x13d) | |
| 915 CASEMBC(0x13f) CASEMBC(0x141) CASEMBC(0x1e3a) | |
| 916 regmbc('L'); REGMBC(0x139) REGMBC(0x13b) | |
| 917 REGMBC(0x13d) REGMBC(0x13f) REGMBC(0x141) | |
| 918 REGMBC(0x1e3a) | |
| 919 return; | |
| 920 case 'M': CASEMBC(0x1e3e) CASEMBC(0x1e40) | |
| 921 regmbc('M'); REGMBC(0x1e3e) REGMBC(0x1e40) | |
| 167 | 922 return; |
| 236 | 923 case 'N': case '\321': |
| 2974 | 924 CASEMBC(0x143) CASEMBC(0x145) CASEMBC(0x147) CASEMBC(0x1e44) |
| 925 CASEMBC(0x1e48) | |
| 236 | 926 regmbc('N'); regmbc('\321'); |
| 2974 | 927 REGMBC(0x143) REGMBC(0x145) REGMBC(0x147) |
| 928 REGMBC(0x1e44) REGMBC(0x1e48) | |
| 167 | 929 return; |
| 236 | 930 case 'O': case '\322': case '\323': case '\324': case '\325': |
| 2974 | 931 case '\326': case '\330': |
| 932 CASEMBC(0x14c) CASEMBC(0x14e) CASEMBC(0x150) CASEMBC(0x1a0) | |
| 933 CASEMBC(0x1d1) CASEMBC(0x1ea) CASEMBC(0x1ec) CASEMBC(0x1ece) | |
| 236 | 934 regmbc('O'); regmbc('\322'); regmbc('\323'); |
| 935 regmbc('\324'); regmbc('\325'); regmbc('\326'); | |
| 2974 | 936 regmbc('\330'); |
| 937 REGMBC(0x14c) REGMBC(0x14e) REGMBC(0x150) | |
| 938 REGMBC(0x1a0) REGMBC(0x1d1) REGMBC(0x1ea) | |
| 939 REGMBC(0x1ec) REGMBC(0x1ece) | |
| 940 return; | |
| 941 case 'P': case 0x1e54: case 0x1e56: | |
| 942 regmbc('P'); REGMBC(0x1e54) REGMBC(0x1e56) | |
| 943 return; | |
| 944 case 'R': CASEMBC(0x154) CASEMBC(0x156) CASEMBC(0x158) | |
| 945 CASEMBC(0x1e58) CASEMBC(0x1e5e) | |
| 946 regmbc('R'); REGMBC(0x154) REGMBC(0x156) REGMBC(0x158) | |
| 947 REGMBC(0x1e58) REGMBC(0x1e5e) | |
| 948 return; | |
| 949 case 'S': CASEMBC(0x15a) CASEMBC(0x15c) CASEMBC(0x15e) | |
| 950 CASEMBC(0x160) CASEMBC(0x1e60) | |
| 951 regmbc('S'); REGMBC(0x15a) REGMBC(0x15c) | |
| 952 REGMBC(0x15e) REGMBC(0x160) REGMBC(0x1e60) | |
| 953 return; | |
| 954 case 'T': CASEMBC(0x162) CASEMBC(0x164) CASEMBC(0x166) | |
| 955 CASEMBC(0x1e6a) CASEMBC(0x1e6e) | |
| 956 regmbc('T'); REGMBC(0x162) REGMBC(0x164) | |
| 957 REGMBC(0x166) REGMBC(0x1e6a) REGMBC(0x1e6e) | |
| 167 | 958 return; |
| 236 | 959 case 'U': case '\331': case '\332': case '\333': case '\334': |
| 2974 | 960 CASEMBC(0x168) CASEMBC(0x16a) CASEMBC(0x16c) CASEMBC(0x16e) |
| 961 CASEMBC(0x170) CASEMBC(0x172) CASEMBC(0x1af) CASEMBC(0x1d3) | |
| 962 CASEMBC(0x1ee6) | |
| 236 | 963 regmbc('U'); regmbc('\331'); regmbc('\332'); |
| 964 regmbc('\333'); regmbc('\334'); | |
| 2974 | 965 REGMBC(0x168) REGMBC(0x16a) REGMBC(0x16c) |
| 966 REGMBC(0x16e) REGMBC(0x170) REGMBC(0x172) | |
| 967 REGMBC(0x1af) REGMBC(0x1d3) REGMBC(0x1ee6) | |
| 968 return; | |
| 969 case 'V': CASEMBC(0x1e7c) | |
| 970 regmbc('V'); REGMBC(0x1e7c) | |
| 971 return; | |
| 972 case 'W': CASEMBC(0x174) CASEMBC(0x1e80) CASEMBC(0x1e82) | |
| 973 CASEMBC(0x1e84) CASEMBC(0x1e86) | |
| 974 regmbc('W'); REGMBC(0x174) REGMBC(0x1e80) | |
| 975 REGMBC(0x1e82) REGMBC(0x1e84) REGMBC(0x1e86) | |
| 976 return; | |
| 977 case 'X': CASEMBC(0x1e8a) CASEMBC(0x1e8c) | |
| 978 regmbc('X'); REGMBC(0x1e8a) REGMBC(0x1e8c) | |
| 167 | 979 return; |
| 236 | 980 case 'Y': case '\335': |
| 2974 | 981 CASEMBC(0x176) CASEMBC(0x178) CASEMBC(0x1e8e) CASEMBC(0x1ef2) |
| 982 CASEMBC(0x1ef6) CASEMBC(0x1ef8) | |
| 236 | 983 regmbc('Y'); regmbc('\335'); |
| 2974 | 984 REGMBC(0x176) REGMBC(0x178) REGMBC(0x1e8e) |
| 985 REGMBC(0x1ef2) REGMBC(0x1ef6) REGMBC(0x1ef8) | |
| 986 return; | |
| 987 case 'Z': CASEMBC(0x179) CASEMBC(0x17b) CASEMBC(0x17d) | |
| 988 CASEMBC(0x1b5) CASEMBC(0x1e90) CASEMBC(0x1e94) | |
| 989 regmbc('Z'); REGMBC(0x179) REGMBC(0x17b) | |
| 990 REGMBC(0x17d) REGMBC(0x1b5) REGMBC(0x1e90) | |
| 991 REGMBC(0x1e94) | |
| 167 | 992 return; |
| 236 | 993 case 'a': case '\340': case '\341': case '\342': |
| 994 case '\343': case '\344': case '\345': | |
| 2974 | 995 CASEMBC(0x101) CASEMBC(0x103) CASEMBC(0x105) CASEMBC(0x1ce) |
| 996 CASEMBC(0x1df) CASEMBC(0x1e1) CASEMBC(0x1ea3) | |
| 236 | 997 regmbc('a'); regmbc('\340'); regmbc('\341'); |
| 998 regmbc('\342'); regmbc('\343'); regmbc('\344'); | |
| 999 regmbc('\345'); | |
| 2974 | 1000 REGMBC(0x101) REGMBC(0x103) REGMBC(0x105) |
| 1001 REGMBC(0x1ce) REGMBC(0x1df) REGMBC(0x1e1) | |
| 1002 REGMBC(0x1ea3) | |
| 1003 return; | |
| 1004 case 'b': CASEMBC(0x1e03) CASEMBC(0x1e07) | |
| 1005 regmbc('b'); REGMBC(0x1e03) REGMBC(0x1e07) | |
| 167 | 1006 return; |
| 236 | 1007 case 'c': case '\347': |
| 2974 | 1008 CASEMBC(0x107) CASEMBC(0x109) CASEMBC(0x10b) CASEMBC(0x10d) |
| 236 | 1009 regmbc('c'); regmbc('\347'); |
| 2974 | 1010 REGMBC(0x107) REGMBC(0x109) REGMBC(0x10b) |
| 1011 REGMBC(0x10d) | |
| 1012 return; | |
| 1013 case 'd': CASEMBC(0x10f) CASEMBC(0x111) CASEMBC(0x1d0b) | |
| 1014 CASEMBC(0x1e11) | |
| 1015 regmbc('d'); REGMBC(0x10f) REGMBC(0x111) | |
| 1016 REGMBC(0x1e0b) REGMBC(0x01e0f) REGMBC(0x1e11) | |
| 167 | 1017 return; |
| 236 | 1018 case 'e': case '\350': case '\351': case '\352': case '\353': |
| 2974 | 1019 CASEMBC(0x113) CASEMBC(0x115) CASEMBC(0x117) CASEMBC(0x119) |
| 1020 CASEMBC(0x11b) CASEMBC(0x1ebb) CASEMBC(0x1ebd) | |
| 236 | 1021 regmbc('e'); regmbc('\350'); regmbc('\351'); |
| 1022 regmbc('\352'); regmbc('\353'); | |
| 2974 | 1023 REGMBC(0x113) REGMBC(0x115) REGMBC(0x117) |
| 1024 REGMBC(0x119) REGMBC(0x11b) REGMBC(0x1ebb) | |
| 1025 REGMBC(0x1ebd) | |
| 1026 return; | |
| 1027 case 'f': CASEMBC(0x1e1f) | |
| 1028 regmbc('f'); REGMBC(0x1e1f) | |
| 1029 return; | |
| 1030 case 'g': CASEMBC(0x11d) CASEMBC(0x11f) CASEMBC(0x121) | |
| 1031 CASEMBC(0x123) CASEMBC(0x1e5) CASEMBC(0x1e7) CASEMBC(0x1f5) | |
| 1032 CASEMBC(0x1e21) | |
| 1033 regmbc('g'); REGMBC(0x11d) REGMBC(0x11f) | |
| 1034 REGMBC(0x121) REGMBC(0x123) REGMBC(0x1e5) | |
| 1035 REGMBC(0x1e7) REGMBC(0x1f5) REGMBC(0x1e21) | |
| 1036 return; | |
| 1037 case 'h': CASEMBC(0x125) CASEMBC(0x127) CASEMBC(0x1e23) | |
| 1038 CASEMBC(0x1e27) CASEMBC(0x1e29) CASEMBC(0x1e96) | |
| 1039 regmbc('h'); REGMBC(0x125) REGMBC(0x127) | |
| 1040 REGMBC(0x1e23) REGMBC(0x1e27) REGMBC(0x1e29) | |
| 1041 REGMBC(0x1e96) | |
| 167 | 1042 return; |
| 236 | 1043 case 'i': case '\354': case '\355': case '\356': case '\357': |
| 2974 | 1044 CASEMBC(0x129) CASEMBC(0x12b) CASEMBC(0x12d) CASEMBC(0x12f) |
| 1045 CASEMBC(0x1d0) CASEMBC(0x1ec9) | |
| 236 | 1046 regmbc('i'); regmbc('\354'); regmbc('\355'); |
| 1047 regmbc('\356'); regmbc('\357'); | |
| 2974 | 1048 REGMBC(0x129) REGMBC(0x12b) REGMBC(0x12d) |
| 1049 REGMBC(0x12f) REGMBC(0x1d0) REGMBC(0x1ec9) | |
| 1050 return; | |
| 1051 case 'j': CASEMBC(0x135) CASEMBC(0x1f0) | |
| 1052 regmbc('j'); REGMBC(0x135) REGMBC(0x1f0) | |
| 1053 return; | |
| 1054 case 'k': CASEMBC(0x137) CASEMBC(0x1e9) CASEMBC(0x1e31) | |
| 1055 CASEMBC(0x1e35) | |
| 1056 regmbc('k'); REGMBC(0x137) REGMBC(0x1e9) | |
| 1057 REGMBC(0x1e31) REGMBC(0x1e35) | |
| 1058 return; | |
| 1059 case 'l': CASEMBC(0x13a) CASEMBC(0x13c) CASEMBC(0x13e) | |
| 1060 CASEMBC(0x140) CASEMBC(0x142) CASEMBC(0x1e3b) | |
| 1061 regmbc('l'); REGMBC(0x13a) REGMBC(0x13c) | |
| 1062 REGMBC(0x13e) REGMBC(0x140) REGMBC(0x142) | |
| 1063 REGMBC(0x1e3b) | |
| 1064 return; | |
| 1065 case 'm': CASEMBC(0x1e3f) CASEMBC(0x1e41) | |
| 1066 regmbc('m'); REGMBC(0x1e3f) REGMBC(0x1e41) | |
| 167 | 1067 return; |
| 236 | 1068 case 'n': case '\361': |
| 2974 | 1069 CASEMBC(0x144) CASEMBC(0x146) CASEMBC(0x148) CASEMBC(0x149) |
| 1070 CASEMBC(0x1e45) CASEMBC(0x1e49) | |
| 236 | 1071 regmbc('n'); regmbc('\361'); |
| 2974 | 1072 REGMBC(0x144) REGMBC(0x146) REGMBC(0x148) |
| 1073 REGMBC(0x149) REGMBC(0x1e45) REGMBC(0x1e49) | |
| 167 | 1074 return; |
| 236 | 1075 case 'o': case '\362': case '\363': case '\364': case '\365': |
| 2974 | 1076 case '\366': case '\370': |
| 1077 CASEMBC(0x14d) CASEMBC(0x14f) CASEMBC(0x151) CASEMBC(0x1a1) | |
| 1078 CASEMBC(0x1d2) CASEMBC(0x1eb) CASEMBC(0x1ed) CASEMBC(0x1ecf) | |
| 236 | 1079 regmbc('o'); regmbc('\362'); regmbc('\363'); |
| 1080 regmbc('\364'); regmbc('\365'); regmbc('\366'); | |
| 2974 | 1081 regmbc('\370'); |
| 1082 REGMBC(0x14d) REGMBC(0x14f) REGMBC(0x151) | |
| 1083 REGMBC(0x1a1) REGMBC(0x1d2) REGMBC(0x1eb) | |
| 1084 REGMBC(0x1ed) REGMBC(0x1ecf) | |
| 1085 return; | |
| 1086 case 'p': CASEMBC(0x1e55) CASEMBC(0x1e57) | |
| 1087 regmbc('p'); REGMBC(0x1e55) REGMBC(0x1e57) | |
| 1088 return; | |
| 1089 case 'r': CASEMBC(0x155) CASEMBC(0x157) CASEMBC(0x159) | |
| 1090 CASEMBC(0x1e59) CASEMBC(0x1e5f) | |
| 1091 regmbc('r'); REGMBC(0x155) REGMBC(0x157) REGMBC(0x159) | |
| 1092 REGMBC(0x1e59) REGMBC(0x1e5f) | |
| 1093 return; | |
| 1094 case 's': CASEMBC(0x15b) CASEMBC(0x15d) CASEMBC(0x15f) | |
| 1095 CASEMBC(0x161) CASEMBC(0x1e61) | |
| 1096 regmbc('s'); REGMBC(0x15b) REGMBC(0x15d) | |
| 1097 REGMBC(0x15f) REGMBC(0x161) REGMBC(0x1e61) | |
| 1098 return; | |
| 1099 case 't': CASEMBC(0x163) CASEMBC(0x165) CASEMBC(0x167) | |
| 1100 CASEMBC(0x1e6b) CASEMBC(0x1e6f) CASEMBC(0x1e97) | |
| 1101 regmbc('t'); REGMBC(0x163) REGMBC(0x165) REGMBC(0x167) | |
| 1102 REGMBC(0x1e6b) REGMBC(0x1e6f) REGMBC(0x1e97) | |
| 167 | 1103 return; |
| 236 | 1104 case 'u': case '\371': case '\372': case '\373': case '\374': |
| 2974 | 1105 CASEMBC(0x169) CASEMBC(0x16b) CASEMBC(0x16d) CASEMBC(0x16f) |
| 1106 CASEMBC(0x171) CASEMBC(0x173) CASEMBC(0x1b0) CASEMBC(0x1d4) | |
| 1107 CASEMBC(0x1ee7) | |
| 236 | 1108 regmbc('u'); regmbc('\371'); regmbc('\372'); |
| 1109 regmbc('\373'); regmbc('\374'); | |
| 2974 | 1110 REGMBC(0x169) REGMBC(0x16b) REGMBC(0x16d) |
| 1111 REGMBC(0x16f) REGMBC(0x171) REGMBC(0x173) | |
| 1112 REGMBC(0x1b0) REGMBC(0x1d4) REGMBC(0x1ee7) | |
| 1113 return; | |
| 1114 case 'v': CASEMBC(0x1e7d) | |
| 1115 regmbc('v'); REGMBC(0x1e7d) | |
| 1116 return; | |
| 1117 case 'w': CASEMBC(0x175) CASEMBC(0x1e81) CASEMBC(0x1e83) | |
| 1118 CASEMBC(0x1e85) CASEMBC(0x1e87) CASEMBC(0x1e98) | |
| 1119 regmbc('w'); REGMBC(0x175) REGMBC(0x1e81) | |
| 1120 REGMBC(0x1e83) REGMBC(0x1e85) REGMBC(0x1e87) | |
| 1121 REGMBC(0x1e98) | |
| 1122 return; | |
| 1123 case 'x': CASEMBC(0x1e8b) CASEMBC(0x1e8d) | |
| 1124 regmbc('x'); REGMBC(0x1e8b) REGMBC(0x1e8d) | |
| 167 | 1125 return; |
| 236 | 1126 case 'y': case '\375': case '\377': |
| 2974 | 1127 CASEMBC(0x177) CASEMBC(0x1e8f) CASEMBC(0x1e99) |
| 1128 CASEMBC(0x1ef3) CASEMBC(0x1ef7) CASEMBC(0x1ef9) | |
| 236 | 1129 regmbc('y'); regmbc('\375'); regmbc('\377'); |
| 2974 | 1130 REGMBC(0x177) REGMBC(0x1e8f) REGMBC(0x1e99) |
| 1131 REGMBC(0x1ef3) REGMBC(0x1ef7) REGMBC(0x1ef9) | |
| 1132 return; | |
| 1133 case 'z': CASEMBC(0x17a) CASEMBC(0x17c) CASEMBC(0x17e) | |
| 1134 CASEMBC(0x1b6) CASEMBC(0x1e91) CASEMBC(0x1e95) | |
| 1135 regmbc('z'); REGMBC(0x17a) REGMBC(0x17c) | |
| 1136 REGMBC(0x17e) REGMBC(0x1b6) REGMBC(0x1e91) | |
| 1137 REGMBC(0x1e95) | |
| 167 | 1138 return; |
| 1139 } | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
1140 #endif |
| 167 | 1141 } |
| 1142 regmbc(c); | |
| 1143 } | |
| 1144 | |
| 1145 /* | |
| 1146 * Check for a collating element "[.a.]". "pp" points to the '['. | |
| 1147 * Returns a character. Zero means that no item was recognized. Otherwise | |
| 1148 * "pp" is advanced to after the item. | |
| 1149 * Currently only single characters are recognized! | |
| 1150 */ | |
| 1151 static int | |
| 1152 get_coll_element(pp) | |
| 1153 char_u **pp; | |
| 1154 { | |
| 1155 int c; | |
| 1156 int l = 1; | |
| 1157 char_u *p = *pp; | |
| 1158 | |
| 1159 if (p[1] == '.') | |
| 1160 { | |
| 1161 #ifdef FEAT_MBYTE | |
| 1162 if (has_mbyte) | |
| 474 | 1163 l = (*mb_ptr2len)(p + 2); |
| 167 | 1164 #endif |
| 1165 if (p[l + 2] == '.' && p[l + 3] == ']') | |
| 1166 { | |
| 1167 #ifdef FEAT_MBYTE | |
| 1168 if (has_mbyte) | |
| 1169 c = mb_ptr2char(p + 2); | |
| 1170 else | |
| 1171 #endif | |
| 1172 c = p[2]; | |
| 1173 *pp += l + 4; | |
| 1174 return c; | |
| 1175 } | |
| 1176 } | |
| 1177 return 0; | |
| 1178 } | |
| 1179 | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1180 static void get_cpo_flags __ARGS((void)); |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1181 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
|
1182 static int reg_cpo_bsl; /* 'cpoptions' contains '\' flag */ |
|
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 static void |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1185 get_cpo_flags() |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1186 { |
|
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1187 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
|
1188 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
|
1189 } |
| 167 | 1190 |
| 1191 /* | |
| 1192 * Skip over a "[]" range. | |
| 1193 * "p" must point to the character after the '['. | |
| 1194 * The returned pointer is on the matching ']', or the terminating NUL. | |
| 1195 */ | |
| 1196 static char_u * | |
| 1197 skip_anyof(p) | |
| 1198 char_u *p; | |
| 1199 { | |
| 1200 #ifdef FEAT_MBYTE | |
| 1201 int l; | |
| 1202 #endif | |
| 1203 | |
| 1204 if (*p == '^') /* Complement of range. */ | |
| 1205 ++p; | |
| 1206 if (*p == ']' || *p == '-') | |
| 1207 ++p; | |
| 1208 while (*p != NUL && *p != ']') | |
| 1209 { | |
| 1210 #ifdef FEAT_MBYTE | |
| 474 | 1211 if (has_mbyte && (l = (*mb_ptr2len)(p)) > 1) |
| 167 | 1212 p += l; |
| 1213 else | |
| 1214 #endif | |
| 1215 if (*p == '-') | |
| 1216 { | |
| 1217 ++p; | |
| 1218 if (*p != ']' && *p != NUL) | |
| 1219 mb_ptr_adv(p); | |
| 1220 } | |
| 1221 else if (*p == '\\' | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1222 && !reg_cpo_bsl |
| 167 | 1223 && (vim_strchr(REGEXP_INRANGE, p[1]) != NULL |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1224 || (!reg_cpo_lit && vim_strchr(REGEXP_ABBR, p[1]) != NULL))) |
| 167 | 1225 p += 2; |
| 1226 else if (*p == '[') | |
| 1227 { | |
| 1228 if (get_char_class(&p) == CLASS_NONE | |
| 1229 && get_equi_class(&p) == 0 | |
| 1230 && get_coll_element(&p) == 0) | |
| 1231 ++p; /* It was not a class name */ | |
| 1232 } | |
| 1233 else | |
| 1234 ++p; | |
| 1235 } | |
| 1236 | |
| 1237 return p; | |
| 1238 } | |
| 1239 | |
| 1240 /* | |
| 7 | 1241 * Skip past regular expression. |
| 153 | 1242 * Stop at end of "startp" or where "dirc" is found ('/', '?', etc). |
| 7 | 1243 * Take care of characters with a backslash in front of it. |
| 1244 * Skip strings inside [ and ]. | |
| 1245 * When "newp" is not NULL and "dirc" is '?', make an allocated copy of the | |
| 1246 * expression and change "\?" to "?". If "*newp" is not NULL the expression | |
| 1247 * is changed in-place. | |
| 1248 */ | |
| 1249 char_u * | |
| 1250 skip_regexp(startp, dirc, magic, newp) | |
| 1251 char_u *startp; | |
| 1252 int dirc; | |
| 1253 int magic; | |
| 1254 char_u **newp; | |
| 1255 { | |
| 1256 int mymagic; | |
| 1257 char_u *p = startp; | |
| 1258 | |
| 1259 if (magic) | |
| 1260 mymagic = MAGIC_ON; | |
| 1261 else | |
| 1262 mymagic = MAGIC_OFF; | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1263 get_cpo_flags(); |
| 7 | 1264 |
| 39 | 1265 for (; p[0] != NUL; mb_ptr_adv(p)) |
| 7 | 1266 { |
| 1267 if (p[0] == dirc) /* found end of regexp */ | |
| 1268 break; | |
| 1269 if ((p[0] == '[' && mymagic >= MAGIC_ON) | |
| 1270 || (p[0] == '\\' && p[1] == '[' && mymagic <= MAGIC_OFF)) | |
| 1271 { | |
| 1272 p = skip_anyof(p + 1); | |
| 1273 if (p[0] == NUL) | |
| 1274 break; | |
| 1275 } | |
| 1276 else if (p[0] == '\\' && p[1] != NUL) | |
| 1277 { | |
| 1278 if (dirc == '?' && newp != NULL && p[1] == '?') | |
| 1279 { | |
| 1280 /* change "\?" to "?", make a copy first. */ | |
| 1281 if (*newp == NULL) | |
| 1282 { | |
| 1283 *newp = vim_strsave(startp); | |
| 1284 if (*newp != NULL) | |
| 1285 p = *newp + (p - startp); | |
| 1286 } | |
| 1287 if (*newp != NULL) | |
| 1621 | 1288 STRMOVE(p, p + 1); |
| 7 | 1289 else |
| 1290 ++p; | |
| 1291 } | |
| 1292 else | |
| 1293 ++p; /* skip next character */ | |
| 1294 if (*p == 'v') | |
| 1295 mymagic = MAGIC_ALL; | |
| 1296 else if (*p == 'V') | |
| 1297 mymagic = MAGIC_NONE; | |
| 1298 } | |
| 1299 } | |
| 1300 return p; | |
| 1301 } | |
| 1302 | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1303 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
|
1304 static void bt_regfree __ARGS((regprog_T *prog)); |
| 4444 | 1305 |
| 7 | 1306 /* |
| 4444 | 1307 * bt_regcomp() - compile a regular expression into internal code for the |
| 1308 * traditional back track matcher. | |
| 41 | 1309 * Returns the program in allocated space. Returns NULL for an error. |
| 7 | 1310 * |
| 1311 * We can't allocate space until we know how big the compiled form will be, | |
| 1312 * but we can't compile it (and thus know how big it is) until we've got a | |
| 1313 * place to put the code. So we cheat: we compile it twice, once with code | |
| 1314 * generation turned off and size counting turned on, and once "for real". | |
| 1315 * This also means that we don't allocate space until we are sure that the | |
| 1316 * thing really will compile successfully, and we never have to move the | |
| 1317 * code and thus invalidate pointers into it. (Note that it has to be in | |
| 1318 * one piece because vim_free() must be able to free it all.) | |
| 1319 * | |
| 1320 * Whether upper/lower case is to be ignored is decided when executing the | |
| 1321 * program, it does not matter here. | |
| 1322 * | |
| 1323 * Beware that the optimization-preparation code in here knows about some | |
| 1324 * of the structure of the compiled regexp. | |
| 1325 * "re_flags": RE_MAGIC and/or RE_STRING. | |
| 1326 */ | |
| 4444 | 1327 static regprog_T * |
| 1328 bt_regcomp(expr, re_flags) | |
| 7 | 1329 char_u *expr; |
| 1330 int re_flags; | |
| 1331 { | |
| 4444 | 1332 bt_regprog_T *r; |
| 7 | 1333 char_u *scan; |
| 1334 char_u *longest; | |
| 1335 int len; | |
| 1336 int flags; | |
| 1337 | |
| 1338 if (expr == NULL) | |
| 1339 EMSG_RET_NULL(_(e_null)); | |
| 1340 | |
| 1341 init_class_tab(); | |
| 1342 | |
| 1343 /* | |
| 1344 * First pass: determine size, legality. | |
| 1345 */ | |
| 1346 regcomp_start(expr, re_flags); | |
| 1347 regcode = JUST_CALC_SIZE; | |
| 1348 regc(REGMAGIC); | |
| 1349 if (reg(REG_NOPAREN, &flags) == NULL) | |
| 1350 return NULL; | |
| 1351 | |
| 1352 /* Small enough for pointer-storage convention? */ | |
| 1353 #ifdef SMALL_MALLOC /* 16 bit storage allocation */ | |
| 1354 if (regsize >= 65536L - 256L) | |
| 1355 EMSG_RET_NULL(_("E339: Pattern too long")); | |
| 1356 #endif | |
| 1357 | |
| 1358 /* Allocate space. */ | |
| 4444 | 1359 r = (bt_regprog_T *)lalloc(sizeof(bt_regprog_T) + regsize, TRUE); |
| 7 | 1360 if (r == NULL) |
| 1361 return NULL; | |
| 1362 | |
| 1363 /* | |
| 1364 * Second pass: emit code. | |
| 1365 */ | |
| 1366 regcomp_start(expr, re_flags); | |
| 1367 regcode = r->program; | |
| 1368 regc(REGMAGIC); | |
| 2010 | 1369 if (reg(REG_NOPAREN, &flags) == NULL || reg_toolong) |
| 7 | 1370 { |
| 1371 vim_free(r); | |
| 2010 | 1372 if (reg_toolong) |
| 1373 EMSG_RET_NULL(_("E339: Pattern too long")); | |
| 7 | 1374 return NULL; |
| 1375 } | |
| 1376 | |
| 1377 /* Dig out information for optimizations. */ | |
| 1378 r->regstart = NUL; /* Worst-case defaults. */ | |
| 1379 r->reganch = 0; | |
| 1380 r->regmust = NULL; | |
| 1381 r->regmlen = 0; | |
| 1382 r->regflags = regflags; | |
| 1383 if (flags & HASNL) | |
| 1384 r->regflags |= RF_HASNL; | |
| 1385 if (flags & HASLOOKBH) | |
| 1386 r->regflags |= RF_LOOKBH; | |
| 1387 #ifdef FEAT_SYN_HL | |
| 1388 /* Remember whether this pattern has any \z specials in it. */ | |
| 1389 r->reghasz = re_has_z; | |
| 1390 #endif | |
| 1391 scan = r->program + 1; /* First BRANCH. */ | |
| 1392 if (OP(regnext(scan)) == END) /* Only one top-level choice. */ | |
| 1393 { | |
| 1394 scan = OPERAND(scan); | |
| 1395 | |
| 1396 /* Starting-point info. */ | |
| 1397 if (OP(scan) == BOL || OP(scan) == RE_BOF) | |
| 1398 { | |
| 1399 r->reganch++; | |
| 1400 scan = regnext(scan); | |
| 1401 } | |
| 1402 | |
| 1403 if (OP(scan) == EXACTLY) | |
| 1404 { | |
| 1405 #ifdef FEAT_MBYTE | |
| 1406 if (has_mbyte) | |
| 1407 r->regstart = (*mb_ptr2char)(OPERAND(scan)); | |
| 1408 else | |
| 1409 #endif | |
| 1410 r->regstart = *OPERAND(scan); | |
| 1411 } | |
| 1412 else if ((OP(scan) == BOW | |
| 1413 || OP(scan) == EOW | |
| 1414 || OP(scan) == NOTHING | |
| 1415 || OP(scan) == MOPEN + 0 || OP(scan) == NOPEN | |
| 1416 || OP(scan) == MCLOSE + 0 || OP(scan) == NCLOSE) | |
| 1417 && OP(regnext(scan)) == EXACTLY) | |
| 1418 { | |
| 1419 #ifdef FEAT_MBYTE | |
| 1420 if (has_mbyte) | |
| 1421 r->regstart = (*mb_ptr2char)(OPERAND(regnext(scan))); | |
| 1422 else | |
| 1423 #endif | |
| 1424 r->regstart = *OPERAND(regnext(scan)); | |
| 1425 } | |
| 1426 | |
| 1427 /* | |
| 1428 * If there's something expensive in the r.e., find the longest | |
| 1429 * literal string that must appear and make it the regmust. Resolve | |
| 1430 * ties in favor of later strings, since the regstart check works | |
| 1431 * with the beginning of the r.e. and avoiding duplication | |
| 1432 * strengthens checking. Not a strong reason, but sufficient in the | |
| 1433 * absence of others. | |
| 1434 */ | |
| 1435 /* | |
| 1436 * When the r.e. starts with BOW, it is faster to look for a regmust | |
| 1437 * first. Used a lot for "#" and "*" commands. (Added by mool). | |
| 1438 */ | |
| 1439 if ((flags & SPSTART || OP(scan) == BOW || OP(scan) == EOW) | |
| 1440 && !(flags & HASNL)) | |
| 1441 { | |
| 1442 longest = NULL; | |
| 1443 len = 0; | |
| 1444 for (; scan != NULL; scan = regnext(scan)) | |
| 1445 if (OP(scan) == EXACTLY && STRLEN(OPERAND(scan)) >= (size_t)len) | |
| 1446 { | |
| 1447 longest = OPERAND(scan); | |
| 1448 len = (int)STRLEN(OPERAND(scan)); | |
| 1449 } | |
| 1450 r->regmust = longest; | |
| 1451 r->regmlen = len; | |
| 1452 } | |
| 1453 } | |
| 4444 | 1454 #ifdef BT_REGEXP_DUMP |
| 7 | 1455 regdump(expr, r); |
| 1456 #endif | |
| 4444 | 1457 r->engine = &bt_regengine; |
| 1458 return (regprog_T *)r; | |
| 7 | 1459 } |
| 1460 | |
| 1461 /* | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1462 * 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
|
1463 */ |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1464 static void |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1465 bt_regfree(prog) |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1466 regprog_T *prog; |
|
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 vim_free(prog); |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1469 } |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1470 |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
1471 /* |
| 7 | 1472 * Setup to parse the regexp. Used once to get the length and once to do it. |
| 1473 */ | |
| 1474 static void | |
| 1475 regcomp_start(expr, re_flags) | |
| 1476 char_u *expr; | |
| 1477 int re_flags; /* see vim_regcomp() */ | |
| 1478 { | |
| 1479 initchr(expr); | |
| 1480 if (re_flags & RE_MAGIC) | |
| 1481 reg_magic = MAGIC_ON; | |
| 1482 else | |
| 1483 reg_magic = MAGIC_OFF; | |
| 1484 reg_string = (re_flags & RE_STRING); | |
| 481 | 1485 reg_strict = (re_flags & RE_STRICT); |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
1486 get_cpo_flags(); |
| 7 | 1487 |
| 1488 num_complex_braces = 0; | |
| 1489 regnpar = 1; | |
| 1490 vim_memset(had_endbrace, 0, sizeof(had_endbrace)); | |
| 1491 #ifdef FEAT_SYN_HL | |
| 1492 regnzpar = 1; | |
| 1493 re_has_z = 0; | |
| 1494 #endif | |
| 1495 regsize = 0L; | |
| 2010 | 1496 reg_toolong = FALSE; |
| 7 | 1497 regflags = 0; |
| 1498 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1499 had_eol = FALSE; | |
| 1500 #endif | |
| 1501 } | |
| 1502 | |
| 1503 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1504 /* | |
| 1505 * Check if during the previous call to vim_regcomp the EOL item "$" has been | |
| 1506 * found. This is messy, but it works fine. | |
| 1507 */ | |
| 1508 int | |
| 1509 vim_regcomp_had_eol() | |
| 1510 { | |
| 1511 return had_eol; | |
| 1512 } | |
| 1513 #endif | |
| 1514 | |
| 1515 /* | |
| 4444 | 1516 * Parse regular expression, i.e. main body or parenthesized thing. |
| 7 | 1517 * |
| 1518 * Caller must absorb opening parenthesis. | |
| 1519 * | |
| 1520 * Combining parenthesis handling with the base level of regular expression | |
| 1521 * is a trifle forced, but the need to tie the tails of the branches to what | |
| 1522 * follows makes it hard to avoid. | |
| 1523 */ | |
| 1524 static char_u * | |
| 1525 reg(paren, flagp) | |
| 1526 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
| 1527 int *flagp; | |
| 1528 { | |
| 1529 char_u *ret; | |
| 1530 char_u *br; | |
| 1531 char_u *ender; | |
| 1532 int parno = 0; | |
| 1533 int flags; | |
| 1534 | |
| 1535 *flagp = HASWIDTH; /* Tentatively. */ | |
| 1536 | |
| 1537 #ifdef FEAT_SYN_HL | |
| 1538 if (paren == REG_ZPAREN) | |
| 1539 { | |
| 1540 /* Make a ZOPEN node. */ | |
| 1541 if (regnzpar >= NSUBEXP) | |
| 1542 EMSG_RET_NULL(_("E50: Too many \\z(")); | |
| 1543 parno = regnzpar; | |
| 1544 regnzpar++; | |
| 1545 ret = regnode(ZOPEN + parno); | |
| 1546 } | |
| 1547 else | |
| 1548 #endif | |
| 1549 if (paren == REG_PAREN) | |
| 1550 { | |
| 1551 /* Make a MOPEN node. */ | |
| 1552 if (regnpar >= NSUBEXP) | |
| 4444 | 1553 EMSG2_RET_NULL(_("E51: Too many %s("), reg_magic == MAGIC_ALL); |
| 7 | 1554 parno = regnpar; |
| 1555 ++regnpar; | |
| 1556 ret = regnode(MOPEN + parno); | |
| 1557 } | |
| 1558 else if (paren == REG_NPAREN) | |
| 1559 { | |
| 1560 /* Make a NOPEN node. */ | |
| 1561 ret = regnode(NOPEN); | |
| 1562 } | |
| 1563 else | |
| 1564 ret = NULL; | |
| 1565 | |
| 1566 /* Pick up the branches, linking them together. */ | |
| 1567 br = regbranch(&flags); | |
| 1568 if (br == NULL) | |
| 1569 return NULL; | |
| 1570 if (ret != NULL) | |
| 1571 regtail(ret, br); /* [MZ]OPEN -> first. */ | |
| 1572 else | |
| 1573 ret = br; | |
| 1574 /* If one of the branches can be zero-width, the whole thing can. | |
| 1575 * If one of the branches has * at start or matches a line-break, the | |
| 1576 * whole thing can. */ | |
| 1577 if (!(flags & HASWIDTH)) | |
| 1578 *flagp &= ~HASWIDTH; | |
| 1579 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1580 while (peekchr() == Magic('|')) | |
| 1581 { | |
| 1582 skipchr(); | |
| 1583 br = regbranch(&flags); | |
| 2010 | 1584 if (br == NULL || reg_toolong) |
| 7 | 1585 return NULL; |
| 1586 regtail(ret, br); /* BRANCH -> BRANCH. */ | |
| 1587 if (!(flags & HASWIDTH)) | |
| 1588 *flagp &= ~HASWIDTH; | |
| 1589 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1590 } | |
| 1591 | |
| 1592 /* Make a closing node, and hook it on the end. */ | |
| 1593 ender = regnode( | |
| 1594 #ifdef FEAT_SYN_HL | |
| 1595 paren == REG_ZPAREN ? ZCLOSE + parno : | |
| 1596 #endif | |
| 1597 paren == REG_PAREN ? MCLOSE + parno : | |
| 1598 paren == REG_NPAREN ? NCLOSE : END); | |
| 1599 regtail(ret, ender); | |
| 1600 | |
| 1601 /* Hook the tails of the branches to the closing node. */ | |
| 1602 for (br = ret; br != NULL; br = regnext(br)) | |
| 1603 regoptail(br, ender); | |
| 1604 | |
| 1605 /* Check for proper termination. */ | |
| 1606 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
| 1607 { | |
| 1608 #ifdef FEAT_SYN_HL | |
| 1609 if (paren == REG_ZPAREN) | |
| 308 | 1610 EMSG_RET_NULL(_("E52: Unmatched \\z(")); |
| 7 | 1611 else |
| 1612 #endif | |
| 1613 if (paren == REG_NPAREN) | |
| 4444 | 1614 EMSG2_RET_NULL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); |
| 7 | 1615 else |
| 4444 | 1616 EMSG2_RET_NULL(_(e_unmatchedp), reg_magic == MAGIC_ALL); |
| 7 | 1617 } |
| 1618 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
| 1619 { | |
| 1620 if (curchr == Magic(')')) | |
| 4444 | 1621 EMSG2_RET_NULL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); |
| 7 | 1622 else |
| 308 | 1623 EMSG_RET_NULL(_(e_trailing)); /* "Can't happen". */ |
| 7 | 1624 /* NOTREACHED */ |
| 1625 } | |
| 1626 /* | |
| 1627 * Here we set the flag allowing back references to this set of | |
| 1628 * parentheses. | |
| 1629 */ | |
| 1630 if (paren == REG_PAREN) | |
| 1631 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
| 1632 return ret; | |
| 1633 } | |
| 1634 | |
| 1635 /* | |
| 4444 | 1636 * Parse one alternative of an | operator. |
| 7 | 1637 * Implements the & operator. |
| 1638 */ | |
| 1639 static char_u * | |
| 1640 regbranch(flagp) | |
| 1641 int *flagp; | |
| 1642 { | |
| 1643 char_u *ret; | |
| 1644 char_u *chain = NULL; | |
| 1645 char_u *latest; | |
| 1646 int flags; | |
| 1647 | |
| 1648 *flagp = WORST | HASNL; /* Tentatively. */ | |
| 1649 | |
| 1650 ret = regnode(BRANCH); | |
| 1651 for (;;) | |
| 1652 { | |
| 1653 latest = regconcat(&flags); | |
| 1654 if (latest == NULL) | |
| 1655 return NULL; | |
| 1656 /* If one of the branches has width, the whole thing has. If one of | |
| 1657 * the branches anchors at start-of-line, the whole thing does. | |
| 1658 * If one of the branches uses look-behind, the whole thing does. */ | |
| 1659 *flagp |= flags & (HASWIDTH | SPSTART | HASLOOKBH); | |
| 1660 /* If one of the branches doesn't match a line-break, the whole thing | |
| 1661 * doesn't. */ | |
| 1662 *flagp &= ~HASNL | (flags & HASNL); | |
| 1663 if (chain != NULL) | |
| 1664 regtail(chain, latest); | |
| 1665 if (peekchr() != Magic('&')) | |
| 1666 break; | |
| 1667 skipchr(); | |
| 1668 regtail(latest, regnode(END)); /* operand ends */ | |
| 2010 | 1669 if (reg_toolong) |
| 1670 break; | |
| 7 | 1671 reginsert(MATCH, latest); |
| 1672 chain = latest; | |
| 1673 } | |
| 1674 | |
| 1675 return ret; | |
| 1676 } | |
| 1677 | |
| 1678 /* | |
| 4444 | 1679 * Parse one alternative of an | or & operator. |
| 7 | 1680 * Implements the concatenation operator. |
| 1681 */ | |
| 1682 static char_u * | |
| 1683 regconcat(flagp) | |
| 1684 int *flagp; | |
| 1685 { | |
| 1686 char_u *first = NULL; | |
| 1687 char_u *chain = NULL; | |
| 1688 char_u *latest; | |
| 1689 int flags; | |
| 1690 int cont = TRUE; | |
| 1691 | |
| 1692 *flagp = WORST; /* Tentatively. */ | |
| 1693 | |
| 1694 while (cont) | |
| 1695 { | |
| 1696 switch (peekchr()) | |
| 1697 { | |
| 1698 case NUL: | |
| 1699 case Magic('|'): | |
| 1700 case Magic('&'): | |
| 1701 case Magic(')'): | |
| 1702 cont = FALSE; | |
| 1703 break; | |
| 1704 case Magic('Z'): | |
| 1705 #ifdef FEAT_MBYTE | |
| 1706 regflags |= RF_ICOMBINE; | |
| 1707 #endif | |
| 1708 skipchr_keepstart(); | |
| 1709 break; | |
| 1710 case Magic('c'): | |
| 1711 regflags |= RF_ICASE; | |
| 1712 skipchr_keepstart(); | |
| 1713 break; | |
| 1714 case Magic('C'): | |
| 1715 regflags |= RF_NOICASE; | |
| 1716 skipchr_keepstart(); | |
| 1717 break; | |
| 1718 case Magic('v'): | |
| 1719 reg_magic = MAGIC_ALL; | |
| 1720 skipchr_keepstart(); | |
| 1721 curchr = -1; | |
| 1722 break; | |
| 1723 case Magic('m'): | |
| 1724 reg_magic = MAGIC_ON; | |
| 1725 skipchr_keepstart(); | |
| 1726 curchr = -1; | |
| 1727 break; | |
| 1728 case Magic('M'): | |
| 1729 reg_magic = MAGIC_OFF; | |
| 1730 skipchr_keepstart(); | |
| 1731 curchr = -1; | |
| 1732 break; | |
| 1733 case Magic('V'): | |
| 1734 reg_magic = MAGIC_NONE; | |
| 1735 skipchr_keepstart(); | |
| 1736 curchr = -1; | |
| 1737 break; | |
| 1738 default: | |
| 1739 latest = regpiece(&flags); | |
| 2010 | 1740 if (latest == NULL || reg_toolong) |
| 7 | 1741 return NULL; |
| 1742 *flagp |= flags & (HASWIDTH | HASNL | HASLOOKBH); | |
| 1743 if (chain == NULL) /* First piece. */ | |
| 1744 *flagp |= flags & SPSTART; | |
| 1745 else | |
| 1746 regtail(chain, latest); | |
| 1747 chain = latest; | |
| 1748 if (first == NULL) | |
| 1749 first = latest; | |
| 1750 break; | |
| 1751 } | |
| 1752 } | |
| 1753 if (first == NULL) /* Loop ran zero times. */ | |
| 1754 first = regnode(NOTHING); | |
| 1755 return first; | |
| 1756 } | |
| 1757 | |
| 1758 /* | |
| 4444 | 1759 * Parse something followed by possible [*+=]. |
| 7 | 1760 * |
| 1761 * Note that the branching code sequences used for = and the general cases | |
| 1762 * of * and + are somewhat optimized: they use the same NOTHING node as | |
| 1763 * both the endmarker for their branch list and the body of the last branch. | |
| 1764 * It might seem that this node could be dispensed with entirely, but the | |
| 1765 * endmarker role is not redundant. | |
| 1766 */ | |
| 1767 static char_u * | |
| 1768 regpiece(flagp) | |
| 1769 int *flagp; | |
| 1770 { | |
| 1771 char_u *ret; | |
| 1772 int op; | |
| 1773 char_u *next; | |
| 1774 int flags; | |
| 1775 long minval; | |
| 1776 long maxval; | |
| 1777 | |
| 1778 ret = regatom(&flags); | |
| 1779 if (ret == NULL) | |
| 1780 return NULL; | |
| 1781 | |
| 1782 op = peekchr(); | |
| 1783 if (re_multi_type(op) == NOT_MULTI) | |
| 1784 { | |
| 1785 *flagp = flags; | |
| 1786 return ret; | |
| 1787 } | |
| 1788 /* default flags */ | |
| 1789 *flagp = (WORST | SPSTART | (flags & (HASNL | HASLOOKBH))); | |
| 1790 | |
| 1791 skipchr(); | |
| 1792 switch (op) | |
| 1793 { | |
| 1794 case Magic('*'): | |
| 1795 if (flags & SIMPLE) | |
| 1796 reginsert(STAR, ret); | |
| 1797 else | |
| 1798 { | |
| 1799 /* Emit x* as (x&|), where & means "self". */ | |
| 1800 reginsert(BRANCH, ret); /* Either x */ | |
| 1801 regoptail(ret, regnode(BACK)); /* and loop */ | |
| 1802 regoptail(ret, ret); /* back */ | |
| 1803 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1804 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1805 } | |
| 1806 break; | |
| 1807 | |
| 1808 case Magic('+'): | |
| 1809 if (flags & SIMPLE) | |
| 1810 reginsert(PLUS, ret); | |
| 1811 else | |
| 1812 { | |
| 1813 /* Emit x+ as x(&|), where & means "self". */ | |
| 1814 next = regnode(BRANCH); /* Either */ | |
| 1815 regtail(ret, next); | |
| 233 | 1816 regtail(regnode(BACK), ret); /* loop back */ |
| 7 | 1817 regtail(next, regnode(BRANCH)); /* or */ |
| 1818 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1819 } | |
| 1820 *flagp = (WORST | HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1821 break; | |
| 1822 | |
| 1823 case Magic('@'): | |
| 1824 { | |
| 1825 int lop = END; | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1826 int nr; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1827 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1828 nr = getdecchrs(); |
| 7 | 1829 switch (no_Magic(getchr())) |
| 1830 { | |
| 1831 case '=': lop = MATCH; break; /* \@= */ | |
| 1832 case '!': lop = NOMATCH; break; /* \@! */ | |
| 1833 case '>': lop = SUBPAT; break; /* \@> */ | |
| 1834 case '<': switch (no_Magic(getchr())) | |
| 1835 { | |
| 1836 case '=': lop = BEHIND; break; /* \@<= */ | |
| 1837 case '!': lop = NOBEHIND; break; /* \@<! */ | |
| 1838 } | |
| 1839 } | |
| 1840 if (lop == END) | |
| 4444 | 1841 EMSG2_RET_NULL(_("E59: invalid character after %s@"), |
| 7 | 1842 reg_magic == MAGIC_ALL); |
| 1843 /* Look behind must match with behind_pos. */ | |
| 1844 if (lop == BEHIND || lop == NOBEHIND) | |
| 1845 { | |
| 1846 regtail(ret, regnode(BHPOS)); | |
| 1847 *flagp |= HASLOOKBH; | |
| 1848 } | |
| 1849 regtail(ret, regnode(END)); /* operand ends */ | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1850 if (lop == BEHIND || lop == NOBEHIND) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1851 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1852 if (nr < 0) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1853 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
|
1854 reginsert_nr(lop, nr, ret); |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1855 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1856 else |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
1857 reginsert(lop, ret); |
| 7 | 1858 break; |
| 1859 } | |
| 1860 | |
| 1861 case Magic('?'): | |
| 1862 case Magic('='): | |
| 1863 /* Emit x= as (x|) */ | |
| 1864 reginsert(BRANCH, ret); /* Either x */ | |
| 1865 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1866 next = regnode(NOTHING); /* null. */ | |
| 1867 regtail(ret, next); | |
| 1868 regoptail(ret, next); | |
| 1869 break; | |
| 1870 | |
| 1871 case Magic('{'): | |
| 1872 if (!read_limits(&minval, &maxval)) | |
| 1873 return NULL; | |
| 1874 if (flags & SIMPLE) | |
| 1875 { | |
| 1876 reginsert(BRACE_SIMPLE, ret); | |
| 1877 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1878 } | |
| 1879 else | |
| 1880 { | |
| 1881 if (num_complex_braces >= 10) | |
| 4444 | 1882 EMSG2_RET_NULL(_("E60: Too many complex %s{...}s"), |
| 7 | 1883 reg_magic == MAGIC_ALL); |
| 1884 reginsert(BRACE_COMPLEX + num_complex_braces, ret); | |
| 1885 regoptail(ret, regnode(BACK)); | |
| 1886 regoptail(ret, ret); | |
| 1887 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1888 ++num_complex_braces; | |
| 1889 } | |
| 1890 if (minval > 0 && maxval > 0) | |
| 1891 *flagp = (HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1892 break; | |
| 1893 } | |
| 1894 if (re_multi_type(peekchr()) != NOT_MULTI) | |
| 1895 { | |
| 1896 /* Can't have a multi follow a multi. */ | |
| 1897 if (peekchr() == Magic('*')) | |
| 1898 sprintf((char *)IObuff, _("E61: Nested %s*"), | |
| 1899 reg_magic >= MAGIC_ON ? "" : "\\"); | |
| 1900 else | |
| 1901 sprintf((char *)IObuff, _("E62: Nested %s%c"), | |
| 1902 reg_magic == MAGIC_ALL ? "" : "\\", no_Magic(peekchr())); | |
| 1903 EMSG_RET_NULL(IObuff); | |
| 1904 } | |
| 1905 | |
| 1906 return ret; | |
| 1907 } | |
| 1908 | |
| 4444 | 1909 /* When making changes to classchars also change nfa_classcodes. */ |
| 1910 static char_u *classchars = (char_u *)".iIkKfFpPsSdDxXoOwWhHaAlLuU"; | |
| 1911 static int classcodes[] = { | |
| 1912 ANY, IDENT, SIDENT, KWORD, SKWORD, | |
| 1913 FNAME, SFNAME, PRINT, SPRINT, | |
| 1914 WHITE, NWHITE, DIGIT, NDIGIT, | |
| 1915 HEX, NHEX, OCTAL, NOCTAL, | |
| 1916 WORD, NWORD, HEAD, NHEAD, | |
| 1917 ALPHA, NALPHA, LOWER, NLOWER, | |
| 1918 UPPER, NUPPER | |
| 1919 }; | |
| 1920 | |
| 7 | 1921 /* |
| 4444 | 1922 * Parse the lowest level. |
| 7 | 1923 * |
| 1924 * Optimization: gobbles an entire sequence of ordinary characters so that | |
| 1925 * it can turn them into a single node, which is smaller to store and | |
| 1926 * faster to run. Don't do this when one_exactly is set. | |
| 1927 */ | |
| 1928 static char_u * | |
| 1929 regatom(flagp) | |
| 1930 int *flagp; | |
| 1931 { | |
| 1932 char_u *ret; | |
| 1933 int flags; | |
| 1934 int c; | |
| 1935 char_u *p; | |
| 1936 int extra = 0; | |
| 1937 | |
| 1938 *flagp = WORST; /* Tentatively. */ | |
| 1939 | |
| 1940 c = getchr(); | |
| 1941 switch (c) | |
| 1942 { | |
| 1943 case Magic('^'): | |
| 1944 ret = regnode(BOL); | |
| 1945 break; | |
| 1946 | |
| 1947 case Magic('$'): | |
| 1948 ret = regnode(EOL); | |
| 1949 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1950 had_eol = TRUE; | |
| 1951 #endif | |
| 1952 break; | |
| 1953 | |
| 1954 case Magic('<'): | |
| 1955 ret = regnode(BOW); | |
| 1956 break; | |
| 1957 | |
| 1958 case Magic('>'): | |
| 1959 ret = regnode(EOW); | |
| 1960 break; | |
| 1961 | |
| 1962 case Magic('_'): | |
| 1963 c = no_Magic(getchr()); | |
| 1964 if (c == '^') /* "\_^" is start-of-line */ | |
| 1965 { | |
| 1966 ret = regnode(BOL); | |
| 1967 break; | |
| 1968 } | |
| 1969 if (c == '$') /* "\_$" is end-of-line */ | |
| 1970 { | |
| 1971 ret = regnode(EOL); | |
| 1972 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1973 had_eol = TRUE; | |
| 1974 #endif | |
| 1975 break; | |
| 1976 } | |
| 1977 | |
| 1978 extra = ADD_NL; | |
| 1979 *flagp |= HASNL; | |
| 1980 | |
| 1981 /* "\_[" is character range plus newline */ | |
| 1982 if (c == '[') | |
| 1983 goto collection; | |
| 1984 | |
| 1985 /* "\_x" is character class plus newline */ | |
| 1986 /*FALLTHROUGH*/ | |
| 1987 | |
| 1988 /* | |
| 1989 * Character classes. | |
| 1990 */ | |
| 1991 case Magic('.'): | |
| 1992 case Magic('i'): | |
| 1993 case Magic('I'): | |
| 1994 case Magic('k'): | |
| 1995 case Magic('K'): | |
| 1996 case Magic('f'): | |
| 1997 case Magic('F'): | |
| 1998 case Magic('p'): | |
| 1999 case Magic('P'): | |
| 2000 case Magic('s'): | |
| 2001 case Magic('S'): | |
| 2002 case Magic('d'): | |
| 2003 case Magic('D'): | |
| 2004 case Magic('x'): | |
| 2005 case Magic('X'): | |
| 2006 case Magic('o'): | |
| 2007 case Magic('O'): | |
| 2008 case Magic('w'): | |
| 2009 case Magic('W'): | |
| 2010 case Magic('h'): | |
| 2011 case Magic('H'): | |
| 2012 case Magic('a'): | |
| 2013 case Magic('A'): | |
| 2014 case Magic('l'): | |
| 2015 case Magic('L'): | |
| 2016 case Magic('u'): | |
| 2017 case Magic('U'): | |
| 2018 p = vim_strchr(classchars, no_Magic(c)); | |
| 2019 if (p == NULL) | |
| 2020 EMSG_RET_NULL(_("E63: invalid use of \\_")); | |
| 714 | 2021 #ifdef FEAT_MBYTE |
| 2022 /* When '.' is followed by a composing char ignore the dot, so that | |
| 2023 * the composing char is matched here. */ | |
| 2024 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
| 2025 { | |
| 2026 c = getchr(); | |
| 2027 goto do_multibyte; | |
| 2028 } | |
| 2029 #endif | |
| 7 | 2030 ret = regnode(classcodes[p - classchars] + extra); |
| 2031 *flagp |= HASWIDTH | SIMPLE; | |
| 2032 break; | |
| 2033 | |
| 2034 case Magic('n'): | |
| 2035 if (reg_string) | |
| 2036 { | |
| 2037 /* In a string "\n" matches a newline character. */ | |
| 2038 ret = regnode(EXACTLY); | |
| 2039 regc(NL); | |
| 2040 regc(NUL); | |
| 2041 *flagp |= HASWIDTH | SIMPLE; | |
| 2042 } | |
| 2043 else | |
| 2044 { | |
| 2045 /* In buffer text "\n" matches the end of a line. */ | |
| 2046 ret = regnode(NEWL); | |
| 2047 *flagp |= HASWIDTH | HASNL; | |
| 2048 } | |
| 2049 break; | |
| 2050 | |
| 2051 case Magic('('): | |
| 2052 if (one_exactly) | |
| 2053 EMSG_ONE_RET_NULL; | |
| 2054 ret = reg(REG_PAREN, &flags); | |
| 2055 if (ret == NULL) | |
| 2056 return NULL; | |
| 2057 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 2058 break; | |
| 2059 | |
| 2060 case NUL: | |
| 2061 case Magic('|'): | |
| 2062 case Magic('&'): | |
| 2063 case Magic(')'): | |
| 1468 | 2064 if (one_exactly) |
| 2065 EMSG_ONE_RET_NULL; | |
| 7 | 2066 EMSG_RET_NULL(_(e_internal)); /* Supposed to be caught earlier. */ |
| 2067 /* NOTREACHED */ | |
| 2068 | |
| 2069 case Magic('='): | |
| 2070 case Magic('?'): | |
| 2071 case Magic('+'): | |
| 2072 case Magic('@'): | |
| 2073 case Magic('{'): | |
| 2074 case Magic('*'): | |
| 2075 c = no_Magic(c); | |
| 2076 sprintf((char *)IObuff, _("E64: %s%c follows nothing"), | |
| 2077 (c == '*' ? reg_magic >= MAGIC_ON : reg_magic == MAGIC_ALL) | |
| 2078 ? "" : "\\", c); | |
| 2079 EMSG_RET_NULL(IObuff); | |
| 2080 /* NOTREACHED */ | |
| 2081 | |
| 2082 case Magic('~'): /* previous substitute pattern */ | |
| 359 | 2083 if (reg_prev_sub != NULL) |
| 7 | 2084 { |
| 2085 char_u *lp; | |
| 2086 | |
| 2087 ret = regnode(EXACTLY); | |
| 2088 lp = reg_prev_sub; | |
| 2089 while (*lp != NUL) | |
| 2090 regc(*lp++); | |
| 2091 regc(NUL); | |
| 2092 if (*reg_prev_sub != NUL) | |
| 2093 { | |
| 2094 *flagp |= HASWIDTH; | |
| 2095 if ((lp - reg_prev_sub) == 1) | |
| 2096 *flagp |= SIMPLE; | |
| 2097 } | |
| 2098 } | |
| 2099 else | |
| 2100 EMSG_RET_NULL(_(e_nopresub)); | |
| 2101 break; | |
| 2102 | |
| 2103 case Magic('1'): | |
| 2104 case Magic('2'): | |
| 2105 case Magic('3'): | |
| 2106 case Magic('4'): | |
| 2107 case Magic('5'): | |
| 2108 case Magic('6'): | |
| 2109 case Magic('7'): | |
| 2110 case Magic('8'): | |
| 2111 case Magic('9'): | |
| 2112 { | |
| 2113 int refnum; | |
| 2114 | |
| 2115 refnum = c - Magic('0'); | |
| 2116 /* | |
| 2117 * Check if the back reference is legal. We must have seen the | |
| 2118 * close brace. | |
| 2119 * TODO: Should also check that we don't refer to something | |
| 2120 * that is repeated (+*=): what instance of the repetition | |
| 2121 * should we match? | |
| 2122 */ | |
| 2123 if (!had_endbrace[refnum]) | |
| 2124 { | |
| 2125 /* Trick: check if "@<=" or "@<!" follows, in which case | |
| 2126 * the \1 can appear before the referenced match. */ | |
| 2127 for (p = regparse; *p != NUL; ++p) | |
| 2128 if (p[0] == '@' && p[1] == '<' | |
| 2129 && (p[2] == '!' || p[2] == '=')) | |
| 2130 break; | |
| 2131 if (*p == NUL) | |
| 2132 EMSG_RET_NULL(_("E65: Illegal back reference")); | |
| 2133 } | |
| 2134 ret = regnode(BACKREF + refnum); | |
| 2135 } | |
| 2136 break; | |
| 2137 | |
| 2138 case Magic('z'): | |
| 2139 { | |
| 2140 c = no_Magic(getchr()); | |
| 2141 switch (c) | |
| 2142 { | |
| 741 | 2143 #ifdef FEAT_SYN_HL |
| 7 | 2144 case '(': if (reg_do_extmatch != REX_SET) |
|
4688
371cc0c44097
updated for version 7.3.1091
Bram Moolenaar <bram@vim.org>
parents:
4682
diff
changeset
|
2145 EMSG_RET_NULL(_(e_z_not_allowed)); |
| 7 | 2146 if (one_exactly) |
| 2147 EMSG_ONE_RET_NULL; | |
| 2148 ret = reg(REG_ZPAREN, &flags); | |
| 2149 if (ret == NULL) | |
| 2150 return NULL; | |
| 2151 *flagp |= flags & (HASWIDTH|SPSTART|HASNL|HASLOOKBH); | |
| 2152 re_has_z = REX_SET; | |
| 2153 break; | |
| 2154 | |
| 2155 case '1': | |
| 2156 case '2': | |
| 2157 case '3': | |
| 2158 case '4': | |
| 2159 case '5': | |
| 2160 case '6': | |
| 2161 case '7': | |
| 2162 case '8': | |
| 2163 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
|
2164 EMSG_RET_NULL(_(e_z1_not_allowed)); |
| 7 | 2165 ret = regnode(ZREF + c - '0'); |
| 2166 re_has_z = REX_USE; | |
| 2167 break; | |
| 741 | 2168 #endif |
| 7 | 2169 |
| 2170 case 's': ret = regnode(MOPEN + 0); | |
| 6203 | 2171 if (re_mult_next("\\zs") == FAIL) |
| 2172 return NULL; | |
| 7 | 2173 break; |
| 2174 | |
| 2175 case 'e': ret = regnode(MCLOSE + 0); | |
| 6203 | 2176 if (re_mult_next("\\ze") == FAIL) |
| 2177 return NULL; | |
| 7 | 2178 break; |
| 2179 | |
| 2180 default: EMSG_RET_NULL(_("E68: Invalid character after \\z")); | |
| 2181 } | |
| 2182 } | |
| 2183 break; | |
| 2184 | |
| 2185 case Magic('%'): | |
| 2186 { | |
| 2187 c = no_Magic(getchr()); | |
| 2188 switch (c) | |
| 2189 { | |
| 2190 /* () without a back reference */ | |
| 2191 case '(': | |
| 2192 if (one_exactly) | |
| 2193 EMSG_ONE_RET_NULL; | |
| 2194 ret = reg(REG_NPAREN, &flags); | |
| 2195 if (ret == NULL) | |
| 2196 return NULL; | |
| 2197 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 2198 break; | |
| 2199 | |
| 2200 /* Catch \%^ and \%$ regardless of where they appear in the | |
| 2201 * pattern -- regardless of whether or not it makes sense. */ | |
| 2202 case '^': | |
| 2203 ret = regnode(RE_BOF); | |
| 2204 break; | |
| 2205 | |
| 2206 case '$': | |
| 2207 ret = regnode(RE_EOF); | |
| 2208 break; | |
| 2209 | |
| 2210 case '#': | |
| 2211 ret = regnode(CURSOR); | |
| 2212 break; | |
| 2213 | |
| 639 | 2214 case 'V': |
| 2215 ret = regnode(RE_VISUAL); | |
| 2216 break; | |
| 2217 | |
| 5901 | 2218 case 'C': |
| 2219 ret = regnode(RE_COMPOSING); | |
| 2220 break; | |
| 2221 | |
| 7 | 2222 /* \%[abc]: Emit as a list of branches, all ending at the last |
| 2223 * branch which matches nothing. */ | |
| 2224 case '[': | |
| 2225 if (one_exactly) /* doesn't nest */ | |
| 2226 EMSG_ONE_RET_NULL; | |
| 2227 { | |
| 2228 char_u *lastbranch; | |
| 2229 char_u *lastnode = NULL; | |
| 2230 char_u *br; | |
| 2231 | |
| 2232 ret = NULL; | |
| 2233 while ((c = getchr()) != ']') | |
| 2234 { | |
| 2235 if (c == NUL) | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2236 EMSG2_RET_NULL(_(e_missing_sb), |
| 7 | 2237 reg_magic == MAGIC_ALL); |
| 2238 br = regnode(BRANCH); | |
| 2239 if (ret == NULL) | |
| 2240 ret = br; | |
| 2241 else | |
| 2242 regtail(lastnode, br); | |
| 2243 | |
| 2244 ungetchr(); | |
| 2245 one_exactly = TRUE; | |
| 2246 lastnode = regatom(flagp); | |
| 2247 one_exactly = FALSE; | |
| 2248 if (lastnode == NULL) | |
| 2249 return NULL; | |
| 2250 } | |
| 2251 if (ret == NULL) | |
|
4760
532a9855bd30
updated for version 7.3.1127
Bram Moolenaar <bram@vim.org>
parents:
4746
diff
changeset
|
2252 EMSG2_RET_NULL(_(e_empty_sb), |
| 7 | 2253 reg_magic == MAGIC_ALL); |
| 2254 lastbranch = regnode(BRANCH); | |
| 2255 br = regnode(NOTHING); | |
| 2256 if (ret != JUST_CALC_SIZE) | |
| 2257 { | |
| 2258 regtail(lastnode, br); | |
| 2259 regtail(lastbranch, br); | |
| 2260 /* connect all branches to the NOTHING | |
| 2261 * branch at the end */ | |
| 2262 for (br = ret; br != lastnode; ) | |
| 2263 { | |
| 2264 if (OP(br) == BRANCH) | |
| 2265 { | |
| 2266 regtail(br, lastbranch); | |
| 2267 br = OPERAND(br); | |
| 2268 } | |
| 2269 else | |
| 2270 br = regnext(br); | |
| 2271 } | |
| 2272 } | |
| 1701 | 2273 *flagp &= ~(HASWIDTH | SIMPLE); |
| 7 | 2274 break; |
| 2275 } | |
| 2276 | |
| 24 | 2277 case 'd': /* %d123 decimal */ |
| 2278 case 'o': /* %o123 octal */ | |
| 2279 case 'x': /* %xab hex 2 */ | |
| 2280 case 'u': /* %uabcd hex 4 */ | |
| 2281 case 'U': /* %U1234abcd hex 8 */ | |
| 2282 { | |
| 2283 int i; | |
| 2284 | |
| 2285 switch (c) | |
| 2286 { | |
| 2287 case 'd': i = getdecchrs(); break; | |
| 2288 case 'o': i = getoctchrs(); break; | |
| 2289 case 'x': i = gethexchrs(2); break; | |
| 2290 case 'u': i = gethexchrs(4); break; | |
| 2291 case 'U': i = gethexchrs(8); break; | |
| 2292 default: i = -1; break; | |
| 2293 } | |
| 2294 | |
| 2295 if (i < 0) | |
| 4444 | 2296 EMSG2_RET_NULL( |
| 24 | 2297 _("E678: Invalid character after %s%%[dxouU]"), |
| 2298 reg_magic == MAGIC_ALL); | |
| 714 | 2299 #ifdef FEAT_MBYTE |
| 2300 if (use_multibytecode(i)) | |
| 2301 ret = regnode(MULTIBYTECODE); | |
| 2302 else | |
| 2303 #endif | |
| 2304 ret = regnode(EXACTLY); | |
| 24 | 2305 if (i == 0) |
| 2306 regc(0x0a); | |
| 2307 else | |
| 2308 #ifdef FEAT_MBYTE | |
| 2309 regmbc(i); | |
| 2310 #else | |
| 2311 regc(i); | |
| 2312 #endif | |
| 2313 regc(NUL); | |
| 2314 *flagp |= HASWIDTH; | |
| 2315 break; | |
| 2316 } | |
| 2317 | |
| 7 | 2318 default: |
| 639 | 2319 if (VIM_ISDIGIT(c) || c == '<' || c == '>' |
| 2320 || c == '\'') | |
| 7 | 2321 { |
| 2322 long_u n = 0; | |
| 2323 int cmp; | |
| 2324 | |
| 2325 cmp = c; | |
| 2326 if (cmp == '<' || cmp == '>') | |
| 2327 c = getchr(); | |
| 2328 while (VIM_ISDIGIT(c)) | |
| 2329 { | |
| 2330 n = n * 10 + (c - '0'); | |
| 2331 c = getchr(); | |
| 2332 } | |
| 639 | 2333 if (c == '\'' && n == 0) |
| 2334 { | |
| 2335 /* "\%'m", "\%<'m" and "\%>'m": Mark */ | |
| 2336 c = getchr(); | |
| 2337 ret = regnode(RE_MARK); | |
| 2338 if (ret == JUST_CALC_SIZE) | |
| 2339 regsize += 2; | |
| 2340 else | |
| 2341 { | |
| 2342 *regcode++ = c; | |
| 2343 *regcode++ = cmp; | |
| 2344 } | |
| 2345 break; | |
| 2346 } | |
| 2347 else if (c == 'l' || c == 'c' || c == 'v') | |
| 7 | 2348 { |
| 2349 if (c == 'l') | |
| 2350 ret = regnode(RE_LNUM); | |
| 2351 else if (c == 'c') | |
| 2352 ret = regnode(RE_COL); | |
| 2353 else | |
| 2354 ret = regnode(RE_VCOL); | |
| 2355 if (ret == JUST_CALC_SIZE) | |
| 2356 regsize += 5; | |
| 2357 else | |
| 2358 { | |
| 2359 /* put the number and the optional | |
| 2360 * comparator after the opcode */ | |
| 2361 regcode = re_put_long(regcode, n); | |
| 2362 *regcode++ = cmp; | |
| 2363 } | |
| 2364 break; | |
| 2365 } | |
| 2366 } | |
| 2367 | |
| 4444 | 2368 EMSG2_RET_NULL(_("E71: Invalid character after %s%%"), |
| 7 | 2369 reg_magic == MAGIC_ALL); |
| 2370 } | |
| 2371 } | |
| 2372 break; | |
| 2373 | |
| 2374 case Magic('['): | |
| 2375 collection: | |
| 2376 { | |
| 2377 char_u *lp; | |
| 2378 | |
| 2379 /* | |
| 2380 * If there is no matching ']', we assume the '[' is a normal | |
| 2381 * character. This makes 'incsearch' and ":help [" work. | |
| 2382 */ | |
| 2383 lp = skip_anyof(regparse); | |
| 2384 if (*lp == ']') /* there is a matching ']' */ | |
| 2385 { | |
| 2386 int startc = -1; /* > 0 when next '-' is a range */ | |
| 2387 int endc; | |
| 2388 | |
| 2389 /* | |
| 2390 * In a character class, different parsing rules apply. | |
| 2391 * Not even \ is special anymore, nothing is. | |
| 2392 */ | |
| 2393 if (*regparse == '^') /* Complement of range. */ | |
| 2394 { | |
| 2395 ret = regnode(ANYBUT + extra); | |
| 2396 regparse++; | |
| 2397 } | |
| 2398 else | |
| 2399 ret = regnode(ANYOF + extra); | |
| 2400 | |
| 2401 /* At the start ']' and '-' mean the literal character. */ | |
| 2402 if (*regparse == ']' || *regparse == '-') | |
| 167 | 2403 { |
| 2404 startc = *regparse; | |
| 7 | 2405 regc(*regparse++); |
| 167 | 2406 } |
| 7 | 2407 |
| 2408 while (*regparse != NUL && *regparse != ']') | |
| 2409 { | |
| 2410 if (*regparse == '-') | |
| 2411 { | |
| 2412 ++regparse; | |
| 2413 /* The '-' is not used for a range at the end and | |
| 2414 * after or before a '\n'. */ | |
| 2415 if (*regparse == ']' || *regparse == NUL | |
| 2416 || startc == -1 | |
| 2417 || (regparse[0] == '\\' && regparse[1] == 'n')) | |
| 2418 { | |
| 2419 regc('-'); | |
| 2420 startc = '-'; /* [--x] is a range */ | |
| 2421 } | |
| 2422 else | |
| 2423 { | |
| 167 | 2424 /* Also accept "a-[.z.]" */ |
| 2425 endc = 0; | |
| 2426 if (*regparse == '[') | |
| 2427 endc = get_coll_element(®parse); | |
| 2428 if (endc == 0) | |
| 2429 { | |
| 7 | 2430 #ifdef FEAT_MBYTE |
| 167 | 2431 if (has_mbyte) |
| 2432 endc = mb_ptr2char_adv(®parse); | |
| 2433 else | |
| 7 | 2434 #endif |
| 167 | 2435 endc = *regparse++; |
| 2436 } | |
| 24 | 2437 |
| 2438 /* Handle \o40, \x20 and \u20AC style sequences */ | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2439 if (endc == '\\' && !reg_cpo_lit && !reg_cpo_bsl) |
| 24 | 2440 endc = coll_get_char(); |
| 2441 | |
| 7 | 2442 if (startc > endc) |
| 2443 EMSG_RET_NULL(_(e_invrange)); | |
| 2444 #ifdef FEAT_MBYTE | |
| 2445 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
| 2446 || (*mb_char2len)(endc) > 1)) | |
| 2447 { | |
| 2448 /* Limit to a range of 256 chars */ | |
| 2449 if (endc > startc + 256) | |
| 2450 EMSG_RET_NULL(_(e_invrange)); | |
| 2451 while (++startc <= endc) | |
| 2452 regmbc(startc); | |
| 2453 } | |
| 2454 else | |
| 2455 #endif | |
| 2456 { | |
| 2457 #ifdef EBCDIC | |
| 2458 int alpha_only = FALSE; | |
| 2459 | |
| 2460 /* for alphabetical range skip the gaps | |
| 2461 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
| 2462 if (isalpha(startc) && isalpha(endc)) | |
| 2463 alpha_only = TRUE; | |
| 2464 #endif | |
| 2465 while (++startc <= endc) | |
| 2466 #ifdef EBCDIC | |
| 2467 if (!alpha_only || isalpha(startc)) | |
| 2468 #endif | |
| 2469 regc(startc); | |
| 2470 } | |
| 2471 startc = -1; | |
| 2472 } | |
| 2473 } | |
| 2474 /* | |
| 2475 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
| 2476 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
| 2477 * 'cpoptions' is not included. | |
| 167 | 2478 * Posix doesn't recognize backslash at all. |
| 7 | 2479 */ |
| 2480 else if (*regparse == '\\' | |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2481 && !reg_cpo_bsl |
| 7 | 2482 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL |
|
4744
a62695305e03
updated for version 7.3.1119
Bram Moolenaar <bram@vim.org>
parents:
4732
diff
changeset
|
2483 || (!reg_cpo_lit |
| 7 | 2484 && vim_strchr(REGEXP_ABBR, |
| 2485 regparse[1]) != NULL))) | |
| 2486 { | |
| 2487 regparse++; | |
| 2488 if (*regparse == 'n') | |
| 2489 { | |
| 2490 /* '\n' in range: also match NL */ | |
| 2491 if (ret != JUST_CALC_SIZE) | |
| 2492 { | |
| 4084 | 2493 /* Using \n inside [^] does not change what |
| 2494 * matches. "[^\n]" is the same as ".". */ | |
| 2495 if (*ret == ANYOF) | |
| 2496 { | |
| 7 | 2497 *ret = ANYOF + ADD_NL; |
| 4084 | 2498 *flagp |= HASNL; |
| 2499 } | |
| 7 | 2500 /* else: must have had a \n already */ |
| 2501 } | |
| 2502 regparse++; | |
| 2503 startc = -1; | |
| 2504 } | |
| 24 | 2505 else if (*regparse == 'd' |
| 2506 || *regparse == 'o' | |
| 2507 || *regparse == 'x' | |
| 2508 || *regparse == 'u' | |
| 2509 || *regparse == 'U') | |
| 2510 { | |
| 2511 startc = coll_get_char(); | |
| 2512 if (startc == 0) | |
| 2513 regc(0x0a); | |
| 2514 else | |
| 2515 #ifdef FEAT_MBYTE | |
| 2516 regmbc(startc); | |
| 2517 #else | |
| 2518 regc(startc); | |
| 2519 #endif | |
| 2520 } | |
| 7 | 2521 else |
| 2522 { | |
| 2523 startc = backslash_trans(*regparse++); | |
| 2524 regc(startc); | |
| 2525 } | |
| 2526 } | |
| 2527 else if (*regparse == '[') | |
| 2528 { | |
| 2529 int c_class; | |
| 2530 int cu; | |
| 2531 | |
| 167 | 2532 c_class = get_char_class(®parse); |
| 7 | 2533 startc = -1; |
| 2534 /* Characters assumed to be 8 bits! */ | |
| 2535 switch (c_class) | |
| 2536 { | |
| 2537 case CLASS_NONE: | |
| 167 | 2538 c_class = get_equi_class(®parse); |
| 2539 if (c_class != 0) | |
| 2540 { | |
| 2541 /* produce equivalence class */ | |
| 2542 reg_equi_class(c_class); | |
| 2543 } | |
| 2544 else if ((c_class = | |
| 2545 get_coll_element(®parse)) != 0) | |
| 2546 { | |
| 2547 /* produce a collating element */ | |
| 2548 regmbc(c_class); | |
| 2549 } | |
| 2550 else | |
| 2551 { | |
| 2552 /* literal '[', allow [[-x] as a range */ | |
| 2553 startc = *regparse++; | |
| 2554 regc(startc); | |
| 2555 } | |
| 7 | 2556 break; |
| 2557 case CLASS_ALNUM: | |
| 2558 for (cu = 1; cu <= 255; cu++) | |
| 2559 if (isalnum(cu)) | |
| 2560 regc(cu); | |
| 2561 break; | |
| 2562 case CLASS_ALPHA: | |
| 2563 for (cu = 1; cu <= 255; cu++) | |
| 2564 if (isalpha(cu)) | |
| 2565 regc(cu); | |
| 2566 break; | |
| 2567 case CLASS_BLANK: | |
| 2568 regc(' '); | |
| 2569 regc('\t'); | |
| 2570 break; | |
| 2571 case CLASS_CNTRL: | |
| 2572 for (cu = 1; cu <= 255; cu++) | |
| 2573 if (iscntrl(cu)) | |
| 2574 regc(cu); | |
| 2575 break; | |
| 2576 case CLASS_DIGIT: | |
| 2577 for (cu = 1; cu <= 255; cu++) | |
| 2578 if (VIM_ISDIGIT(cu)) | |
| 2579 regc(cu); | |
| 2580 break; | |
| 2581 case CLASS_GRAPH: | |
| 2582 for (cu = 1; cu <= 255; cu++) | |
| 2583 if (isgraph(cu)) | |
| 2584 regc(cu); | |
| 2585 break; | |
| 2586 case CLASS_LOWER: | |
| 2587 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2588 if (MB_ISLOWER(cu)) |
| 7 | 2589 regc(cu); |
| 2590 break; | |
| 2591 case CLASS_PRINT: | |
| 2592 for (cu = 1; cu <= 255; cu++) | |
| 2593 if (vim_isprintc(cu)) | |
| 2594 regc(cu); | |
| 2595 break; | |
| 2596 case CLASS_PUNCT: | |
| 2597 for (cu = 1; cu <= 255; cu++) | |
| 2598 if (ispunct(cu)) | |
| 2599 regc(cu); | |
| 2600 break; | |
| 2601 case CLASS_SPACE: | |
| 2602 for (cu = 9; cu <= 13; cu++) | |
| 2603 regc(cu); | |
| 2604 regc(' '); | |
| 2605 break; | |
| 2606 case CLASS_UPPER: | |
| 2607 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2608 if (MB_ISUPPER(cu)) |
| 7 | 2609 regc(cu); |
| 2610 break; | |
| 2611 case CLASS_XDIGIT: | |
| 2612 for (cu = 1; cu <= 255; cu++) | |
| 2613 if (vim_isxdigit(cu)) | |
| 2614 regc(cu); | |
| 2615 break; | |
| 2616 case CLASS_TAB: | |
| 2617 regc('\t'); | |
| 2618 break; | |
| 2619 case CLASS_RETURN: | |
| 2620 regc('\r'); | |
| 2621 break; | |
| 2622 case CLASS_BACKSPACE: | |
| 2623 regc('\b'); | |
| 2624 break; | |
| 2625 case CLASS_ESCAPE: | |
| 2626 regc('\033'); | |
| 2627 break; | |
| 2628 } | |
| 2629 } | |
| 2630 else | |
| 2631 { | |
| 2632 #ifdef FEAT_MBYTE | |
| 2633 if (has_mbyte) | |
| 2634 { | |
| 2635 int len; | |
| 2636 | |
| 2637 /* produce a multibyte character, including any | |
| 2638 * following composing characters */ | |
| 2639 startc = mb_ptr2char(regparse); | |
| 474 | 2640 len = (*mb_ptr2len)(regparse); |
| 7 | 2641 if (enc_utf8 && utf_char2len(startc) != len) |
| 2642 startc = -1; /* composing chars */ | |
| 2643 while (--len >= 0) | |
| 2644 regc(*regparse++); | |
| 2645 } | |
| 2646 else | |
| 2647 #endif | |
| 2648 { | |
| 2649 startc = *regparse++; | |
| 2650 regc(startc); | |
| 2651 } | |
| 2652 } | |
| 2653 } | |
| 2654 regc(NUL); | |
| 2655 prevchr_len = 1; /* last char was the ']' */ | |
| 2656 if (*regparse != ']') | |
| 2657 EMSG_RET_NULL(_(e_toomsbra)); /* Cannot happen? */ | |
| 2658 skipchr(); /* let's be friends with the lexer again */ | |
| 2659 *flagp |= HASWIDTH | SIMPLE; | |
| 2660 break; | |
| 2661 } | |
| 481 | 2662 else if (reg_strict) |
| 4444 | 2663 EMSG2_RET_NULL(_(e_missingbracket), reg_magic > MAGIC_OFF); |
| 7 | 2664 } |
| 2665 /* FALLTHROUGH */ | |
| 2666 | |
| 2667 default: | |
| 2668 { | |
| 2669 int len; | |
| 2670 | |
| 2671 #ifdef FEAT_MBYTE | |
| 2672 /* A multi-byte character is handled as a separate atom if it's | |
| 714 | 2673 * before a multi and when it's a composing char. */ |
| 2674 if (use_multibytecode(c)) | |
| 7 | 2675 { |
| 714 | 2676 do_multibyte: |
| 7 | 2677 ret = regnode(MULTIBYTECODE); |
| 2678 regmbc(c); | |
| 2679 *flagp |= HASWIDTH | SIMPLE; | |
| 2680 break; | |
| 2681 } | |
| 2682 #endif | |
| 2683 | |
| 2684 ret = regnode(EXACTLY); | |
| 2685 | |
| 2686 /* | |
| 2687 * Append characters as long as: | |
| 2688 * - there is no following multi, we then need the character in | |
| 2689 * front of it as a single character operand | |
| 2690 * - not running into a Magic character | |
| 2691 * - "one_exactly" is not set | |
| 2692 * But always emit at least one character. Might be a Multi, | |
| 2693 * e.g., a "[" without matching "]". | |
| 2694 */ | |
| 2695 for (len = 0; c != NUL && (len == 0 | |
| 2696 || (re_multi_type(peekchr()) == NOT_MULTI | |
| 2697 && !one_exactly | |
| 2698 && !is_Magic(c))); ++len) | |
| 2699 { | |
| 2700 c = no_Magic(c); | |
| 2701 #ifdef FEAT_MBYTE | |
| 2702 if (has_mbyte) | |
| 2703 { | |
| 2704 regmbc(c); | |
| 2705 if (enc_utf8) | |
| 2706 { | |
| 2707 int l; | |
| 2708 | |
| 714 | 2709 /* Need to get composing character too. */ |
| 7 | 2710 for (;;) |
| 2711 { | |
| 714 | 2712 l = utf_ptr2len(regparse); |
| 2713 if (!UTF_COMPOSINGLIKE(regparse, regparse + l)) | |
| 7 | 2714 break; |
| 714 | 2715 regmbc(utf_ptr2char(regparse)); |
| 2716 skipchr(); | |
| 7 | 2717 } |
| 2718 } | |
| 2719 } | |
| 2720 else | |
| 2721 #endif | |
| 2722 regc(c); | |
| 2723 c = getchr(); | |
| 2724 } | |
| 2725 ungetchr(); | |
| 2726 | |
| 2727 regc(NUL); | |
| 2728 *flagp |= HASWIDTH; | |
| 2729 if (len == 1) | |
| 2730 *flagp |= SIMPLE; | |
| 2731 } | |
| 2732 break; | |
| 2733 } | |
| 2734 | |
| 2735 return ret; | |
| 2736 } | |
| 2737 | |
| 714 | 2738 #ifdef FEAT_MBYTE |
| 2739 /* | |
| 2740 * Return TRUE if MULTIBYTECODE should be used instead of EXACTLY for | |
| 2741 * character "c". | |
| 2742 */ | |
| 2743 static int | |
| 2744 use_multibytecode(c) | |
| 2745 int c; | |
| 2746 { | |
| 2747 return has_mbyte && (*mb_char2len)(c) > 1 | |
| 2748 && (re_multi_type(peekchr()) != NOT_MULTI | |
| 2749 || (enc_utf8 && utf_iscomposing(c))); | |
| 2750 } | |
| 2751 #endif | |
| 2752 | |
| 7 | 2753 /* |
| 4444 | 2754 * Emit a node. |
| 7 | 2755 * Return pointer to generated code. |
| 2756 */ | |
| 2757 static char_u * | |
| 2758 regnode(op) | |
| 2759 int op; | |
| 2760 { | |
| 2761 char_u *ret; | |
| 2762 | |
| 2763 ret = regcode; | |
| 2764 if (ret == JUST_CALC_SIZE) | |
| 2765 regsize += 3; | |
| 2766 else | |
| 2767 { | |
| 2768 *regcode++ = op; | |
| 2769 *regcode++ = NUL; /* Null "next" pointer. */ | |
| 2770 *regcode++ = NUL; | |
| 2771 } | |
| 2772 return ret; | |
| 2773 } | |
| 2774 | |
| 2775 /* | |
| 2776 * Emit (if appropriate) a byte of code | |
| 2777 */ | |
| 2778 static void | |
| 2779 regc(b) | |
| 2780 int b; | |
| 2781 { | |
| 2782 if (regcode == JUST_CALC_SIZE) | |
| 2783 regsize++; | |
| 2784 else | |
| 2785 *regcode++ = b; | |
| 2786 } | |
| 2787 | |
| 2788 #ifdef FEAT_MBYTE | |
| 2789 /* | |
| 2790 * Emit (if appropriate) a multi-byte character of code | |
| 2791 */ | |
| 2792 static void | |
| 2793 regmbc(c) | |
| 2794 int c; | |
| 2795 { | |
| 2974 | 2796 if (!has_mbyte && c > 0xff) |
| 2797 return; | |
| 7 | 2798 if (regcode == JUST_CALC_SIZE) |
| 2799 regsize += (*mb_char2len)(c); | |
| 2800 else | |
| 2801 regcode += (*mb_char2bytes)(c, regcode); | |
| 2802 } | |
| 2803 #endif | |
| 2804 | |
| 2805 /* | |
| 4444 | 2806 * Insert an operator in front of already-emitted operand |
| 7 | 2807 * |
| 2808 * Means relocating the operand. | |
| 2809 */ | |
| 2810 static void | |
| 2811 reginsert(op, opnd) | |
| 2812 int op; | |
| 2813 char_u *opnd; | |
| 2814 { | |
| 2815 char_u *src; | |
| 2816 char_u *dst; | |
| 2817 char_u *place; | |
| 2818 | |
| 2819 if (regcode == JUST_CALC_SIZE) | |
| 2820 { | |
| 2821 regsize += 3; | |
| 2822 return; | |
| 2823 } | |
| 2824 src = regcode; | |
| 2825 regcode += 3; | |
| 2826 dst = regcode; | |
| 2827 while (src > opnd) | |
| 2828 *--dst = *--src; | |
| 2829 | |
| 2830 place = opnd; /* Op node, where operand used to be. */ | |
| 2831 *place++ = op; | |
| 2832 *place++ = NUL; | |
| 2833 *place = NUL; | |
| 2834 } | |
| 2835 | |
| 2836 /* | |
| 4444 | 2837 * 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
|
2838 * Add a number to the operator. |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2839 */ |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2840 static void |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2841 reginsert_nr(op, val, opnd) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2842 int op; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2843 long val; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2844 char_u *opnd; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2845 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2846 char_u *src; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2847 char_u *dst; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2848 char_u *place; |
|
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 if (regcode == JUST_CALC_SIZE) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2851 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2852 regsize += 7; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2853 return; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2854 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2855 src = regcode; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2856 regcode += 7; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2857 dst = regcode; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2858 while (src > opnd) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2859 *--dst = *--src; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2860 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2861 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
|
2862 *place++ = op; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2863 *place++ = NUL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2864 *place++ = NUL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2865 place = re_put_long(place, (long_u)val); |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2866 } |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2867 |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2868 /* |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
2869 * Insert an operator in front of already-emitted operand. |
| 7 | 2870 * The operator has the given limit values as operands. Also set next pointer. |
| 2871 * | |
| 2872 * Means relocating the operand. | |
| 2873 */ | |
| 2874 static void | |
| 2875 reginsert_limits(op, minval, maxval, opnd) | |
| 2876 int op; | |
| 2877 long minval; | |
| 2878 long maxval; | |
| 2879 char_u *opnd; | |
| 2880 { | |
| 2881 char_u *src; | |
| 2882 char_u *dst; | |
| 2883 char_u *place; | |
| 2884 | |
| 2885 if (regcode == JUST_CALC_SIZE) | |
| 2886 { | |
| 2887 regsize += 11; | |
| 2888 return; | |
| 2889 } | |
| 2890 src = regcode; | |
| 2891 regcode += 11; | |
| 2892 dst = regcode; | |
| 2893 while (src > opnd) | |
| 2894 *--dst = *--src; | |
| 2895 | |
| 2896 place = opnd; /* Op node, where operand used to be. */ | |
| 2897 *place++ = op; | |
| 2898 *place++ = NUL; | |
| 2899 *place++ = NUL; | |
| 2900 place = re_put_long(place, (long_u)minval); | |
| 2901 place = re_put_long(place, (long_u)maxval); | |
| 2902 regtail(opnd, place); | |
| 2903 } | |
| 2904 | |
| 2905 /* | |
| 2906 * Write a long as four bytes at "p" and return pointer to the next char. | |
| 2907 */ | |
| 2908 static char_u * | |
| 2909 re_put_long(p, val) | |
| 2910 char_u *p; | |
| 2911 long_u val; | |
| 2912 { | |
| 2913 *p++ = (char_u) ((val >> 24) & 0377); | |
| 2914 *p++ = (char_u) ((val >> 16) & 0377); | |
| 2915 *p++ = (char_u) ((val >> 8) & 0377); | |
| 2916 *p++ = (char_u) (val & 0377); | |
| 2917 return p; | |
| 2918 } | |
| 2919 | |
| 2920 /* | |
| 4444 | 2921 * Set the next-pointer at the end of a node chain. |
| 7 | 2922 */ |
| 2923 static void | |
| 2924 regtail(p, val) | |
| 2925 char_u *p; | |
| 2926 char_u *val; | |
| 2927 { | |
| 2928 char_u *scan; | |
| 2929 char_u *temp; | |
| 2930 int offset; | |
| 2931 | |
| 2932 if (p == JUST_CALC_SIZE) | |
| 2933 return; | |
| 2934 | |
| 2935 /* Find last node. */ | |
| 2936 scan = p; | |
| 2937 for (;;) | |
| 2938 { | |
| 2939 temp = regnext(scan); | |
| 2940 if (temp == NULL) | |
| 2941 break; | |
| 2942 scan = temp; | |
| 2943 } | |
| 2944 | |
| 233 | 2945 if (OP(scan) == BACK) |
| 7 | 2946 offset = (int)(scan - val); |
| 2947 else | |
| 2948 offset = (int)(val - scan); | |
| 2010 | 2949 /* When the offset uses more than 16 bits it can no longer fit in the two |
| 2974 | 2950 * bytes available. Use a global flag to avoid having to check return |
| 2010 | 2951 * values in too many places. */ |
| 2952 if (offset > 0xffff) | |
| 2953 reg_toolong = TRUE; | |
| 2954 else | |
| 2955 { | |
| 2956 *(scan + 1) = (char_u) (((unsigned)offset >> 8) & 0377); | |
| 2957 *(scan + 2) = (char_u) (offset & 0377); | |
| 2958 } | |
| 7 | 2959 } |
| 2960 | |
| 2961 /* | |
| 4444 | 2962 * Like regtail, on item after a BRANCH; nop if none. |
| 7 | 2963 */ |
| 2964 static void | |
| 2965 regoptail(p, val) | |
| 2966 char_u *p; | |
| 2967 char_u *val; | |
| 2968 { | |
| 2969 /* When op is neither BRANCH nor BRACE_COMPLEX0-9, it is "operandless" */ | |
| 2970 if (p == NULL || p == JUST_CALC_SIZE | |
| 2971 || (OP(p) != BRANCH | |
| 2972 && (OP(p) < BRACE_COMPLEX || OP(p) > BRACE_COMPLEX + 9))) | |
| 2973 return; | |
| 2974 regtail(OPERAND(p), val); | |
| 2975 } | |
| 2976 | |
| 2977 /* | |
| 4444 | 2978 * Functions for getting characters from the regexp input. |
| 7 | 2979 */ |
| 2980 | |
| 2981 static int at_start; /* True when on the first character */ | |
| 2982 static int prev_at_start; /* True when on the second character */ | |
| 2983 | |
| 4444 | 2984 /* |
| 2985 * Start parsing at "str". | |
| 2986 */ | |
| 7 | 2987 static void |
| 2988 initchr(str) | |
| 2989 char_u *str; | |
| 2990 { | |
| 2991 regparse = str; | |
| 2992 prevchr_len = 0; | |
| 2993 curchr = prevprevchr = prevchr = nextchr = -1; | |
| 2994 at_start = TRUE; | |
| 2995 prev_at_start = FALSE; | |
| 2996 } | |
| 2997 | |
| 4444 | 2998 /* |
|
4679
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
2999 * 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
|
3000 * starts in the same state again. |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3001 */ |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3002 static void |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3003 save_parse_state(ps) |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3004 parse_state_T *ps; |
|
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 ps->regparse = regparse; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3007 ps->prevchr_len = prevchr_len; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3008 ps->curchr = curchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3009 ps->prevchr = prevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3010 ps->prevprevchr = prevprevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3011 ps->nextchr = nextchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3012 ps->at_start = at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3013 ps->prev_at_start = prev_at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3014 ps->regnpar = regnpar; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3015 } |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3016 |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3017 /* |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3018 * Restore a previously saved parse state. |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3019 */ |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3020 static void |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3021 restore_parse_state(ps) |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3022 parse_state_T *ps; |
|
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 regparse = ps->regparse; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3025 prevchr_len = ps->prevchr_len; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3026 curchr = ps->curchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3027 prevchr = ps->prevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3028 prevprevchr = ps->prevprevchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3029 nextchr = ps->nextchr; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3030 at_start = ps->at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3031 prev_at_start = ps->prev_at_start; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3032 regnpar = ps->regnpar; |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3033 } |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3034 |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3035 |
|
4d92b873acef
updated for version 7.3.1087
Bram Moolenaar <bram@vim.org>
parents:
4579
diff
changeset
|
3036 /* |
| 4444 | 3037 * Get the next character without advancing. |
| 3038 */ | |
| 7 | 3039 static int |
| 3040 peekchr() | |
| 3041 { | |
| 167 | 3042 static int after_slash = FALSE; |
| 3043 | |
| 7 | 3044 if (curchr == -1) |
| 3045 { | |
| 3046 switch (curchr = regparse[0]) | |
| 3047 { | |
| 3048 case '.': | |
| 3049 case '[': | |
| 3050 case '~': | |
| 3051 /* magic when 'magic' is on */ | |
| 3052 if (reg_magic >= MAGIC_ON) | |
| 3053 curchr = Magic(curchr); | |
| 3054 break; | |
| 3055 case '(': | |
| 3056 case ')': | |
| 3057 case '{': | |
| 3058 case '%': | |
| 3059 case '+': | |
| 3060 case '=': | |
| 3061 case '?': | |
| 3062 case '@': | |
| 3063 case '!': | |
| 3064 case '&': | |
| 3065 case '|': | |
| 3066 case '<': | |
| 3067 case '>': | |
| 3068 case '#': /* future ext. */ | |
| 3069 case '"': /* future ext. */ | |
| 3070 case '\'': /* future ext. */ | |
| 3071 case ',': /* future ext. */ | |
| 3072 case '-': /* future ext. */ | |
| 3073 case ':': /* future ext. */ | |
| 3074 case ';': /* future ext. */ | |
| 3075 case '`': /* future ext. */ | |
| 3076 case '/': /* Can't be used in / command */ | |
| 3077 /* magic only after "\v" */ | |
| 3078 if (reg_magic == MAGIC_ALL) | |
| 3079 curchr = Magic(curchr); | |
| 3080 break; | |
| 3081 case '*': | |
| 167 | 3082 /* * is not magic as the very first character, eg "?*ptr", when |
| 3083 * after '^', eg "/^*ptr" and when after "\(", "\|", "\&". But | |
| 3084 * "\(\*" is not magic, thus must be magic if "after_slash" */ | |
| 3085 if (reg_magic >= MAGIC_ON | |
| 3086 && !at_start | |
| 3087 && !(prev_at_start && prevchr == Magic('^')) | |
| 3088 && (after_slash | |
| 3089 || (prevchr != Magic('(') | |
| 3090 && prevchr != Magic('&') | |
| 3091 && prevchr != Magic('|')))) | |
| 7 | 3092 curchr = Magic('*'); |
| 3093 break; | |
| 3094 case '^': | |
| 3095 /* '^' is only magic as the very first character and if it's after | |
| 3096 * "\(", "\|", "\&' or "\n" */ | |
| 3097 if (reg_magic >= MAGIC_OFF | |
| 3098 && (at_start | |
| 3099 || reg_magic == MAGIC_ALL | |
| 3100 || prevchr == Magic('(') | |
| 3101 || prevchr == Magic('|') | |
| 3102 || prevchr == Magic('&') | |
| 3103 || prevchr == Magic('n') | |
| 3104 || (no_Magic(prevchr) == '(' | |
| 3105 && prevprevchr == Magic('%')))) | |
| 3106 { | |
| 3107 curchr = Magic('^'); | |
| 3108 at_start = TRUE; | |
| 3109 prev_at_start = FALSE; | |
| 3110 } | |
| 3111 break; | |
| 3112 case '$': | |
| 3113 /* '$' is only magic as the very last char and if it's in front of | |
| 3114 * either "\|", "\)", "\&", or "\n" */ | |
| 3115 if (reg_magic >= MAGIC_OFF) | |
| 3116 { | |
| 3117 char_u *p = regparse + 1; | |
| 6041 | 3118 int is_magic_all = (reg_magic == MAGIC_ALL); |
| 3119 | |
| 3120 /* ignore \c \C \m \M \v \V and \Z after '$' */ | |
| 7 | 3121 while (p[0] == '\\' && (p[1] == 'c' || p[1] == 'C' |
| 6041 | 3122 || p[1] == 'm' || p[1] == 'M' |
| 3123 || p[1] == 'v' || p[1] == 'V' || p[1] == 'Z')) | |
| 3124 { | |
| 3125 if (p[1] == 'v') | |
| 3126 is_magic_all = TRUE; | |
| 3127 else if (p[1] == 'm' || p[1] == 'M' || p[1] == 'V') | |
| 3128 is_magic_all = FALSE; | |
| 7 | 3129 p += 2; |
| 6041 | 3130 } |
| 7 | 3131 if (p[0] == NUL |
| 3132 || (p[0] == '\\' | |
| 3133 && (p[1] == '|' || p[1] == '&' || p[1] == ')' | |
| 3134 || p[1] == 'n')) | |
| 6041 | 3135 || (is_magic_all |
| 3136 && (p[0] == '|' || p[0] == '&' || p[0] == ')')) | |
| 7 | 3137 || reg_magic == MAGIC_ALL) |
| 3138 curchr = Magic('$'); | |
| 3139 } | |
| 3140 break; | |
| 3141 case '\\': | |
| 3142 { | |
| 3143 int c = regparse[1]; | |
| 3144 | |
| 3145 if (c == NUL) | |
| 3146 curchr = '\\'; /* trailing '\' */ | |
| 3147 else if ( | |
| 3148 #ifdef EBCDIC | |
| 3149 vim_strchr(META, c) | |
| 3150 #else | |
| 3151 c <= '~' && META_flags[c] | |
| 3152 #endif | |
| 3153 ) | |
| 3154 { | |
| 3155 /* | |
| 3156 * META contains everything that may be magic sometimes, | |
| 3157 * except ^ and $ ("\^" and "\$" are only magic after | |
| 3158 * "\v"). We now fetch the next character and toggle its | |
| 3159 * magicness. Therefore, \ is so meta-magic that it is | |
| 3160 * not in META. | |
| 3161 */ | |
| 3162 curchr = -1; | |
| 3163 prev_at_start = at_start; | |
| 3164 at_start = FALSE; /* be able to say "/\*ptr" */ | |
| 3165 ++regparse; | |
| 167 | 3166 ++after_slash; |
| 7 | 3167 peekchr(); |
| 3168 --regparse; | |
| 167 | 3169 --after_slash; |
| 7 | 3170 curchr = toggle_Magic(curchr); |
| 3171 } | |
| 3172 else if (vim_strchr(REGEXP_ABBR, c)) | |
| 3173 { | |
| 3174 /* | |
| 3175 * Handle abbreviations, like "\t" for TAB -- webb | |
| 3176 */ | |
| 3177 curchr = backslash_trans(c); | |
| 3178 } | |
| 3179 else if (reg_magic == MAGIC_NONE && (c == '$' || c == '^')) | |
| 3180 curchr = toggle_Magic(c); | |
| 3181 else | |
| 3182 { | |
| 3183 /* | |
| 3184 * Next character can never be (made) magic? | |
| 3185 * Then backslashing it won't do anything. | |
| 3186 */ | |
| 3187 #ifdef FEAT_MBYTE | |
| 3188 if (has_mbyte) | |
| 3189 curchr = (*mb_ptr2char)(regparse + 1); | |
| 3190 else | |
| 3191 #endif | |
| 3192 curchr = c; | |
| 3193 } | |
| 3194 break; | |
| 3195 } | |
| 3196 | |
| 3197 #ifdef FEAT_MBYTE | |
| 3198 default: | |
| 3199 if (has_mbyte) | |
| 3200 curchr = (*mb_ptr2char)(regparse); | |
| 3201 #endif | |
| 3202 } | |
| 3203 } | |
| 3204 | |
| 3205 return curchr; | |
| 3206 } | |
| 3207 | |
| 3208 /* | |
| 3209 * Eat one lexed character. Do this in a way that we can undo it. | |
| 3210 */ | |
| 3211 static void | |
| 3212 skipchr() | |
| 3213 { | |
| 3214 /* peekchr() eats a backslash, do the same here */ | |
| 3215 if (*regparse == '\\') | |
| 3216 prevchr_len = 1; | |
| 3217 else | |
| 3218 prevchr_len = 0; | |
| 3219 if (regparse[prevchr_len] != NUL) | |
| 3220 { | |
| 3221 #ifdef FEAT_MBYTE | |
| 714 | 3222 if (enc_utf8) |
| 1449 | 3223 /* exclude composing chars that mb_ptr2len does include */ |
| 3224 prevchr_len += utf_ptr2len(regparse + prevchr_len); | |
| 714 | 3225 else if (has_mbyte) |
| 474 | 3226 prevchr_len += (*mb_ptr2len)(regparse + prevchr_len); |
| 7 | 3227 else |
| 3228 #endif | |
| 3229 ++prevchr_len; | |
| 3230 } | |
| 3231 regparse += prevchr_len; | |
| 3232 prev_at_start = at_start; | |
| 3233 at_start = FALSE; | |
| 3234 prevprevchr = prevchr; | |
| 3235 prevchr = curchr; | |
| 3236 curchr = nextchr; /* use previously unget char, or -1 */ | |
| 3237 nextchr = -1; | |
| 3238 } | |
| 3239 | |
| 3240 /* | |
| 3241 * Skip a character while keeping the value of prev_at_start for at_start. | |
| 3242 * prevchr and prevprevchr are also kept. | |
| 3243 */ | |
| 3244 static void | |
| 3245 skipchr_keepstart() | |
| 3246 { | |
| 3247 int as = prev_at_start; | |
| 3248 int pr = prevchr; | |
| 3249 int prpr = prevprevchr; | |
| 3250 | |
| 3251 skipchr(); | |
| 3252 at_start = as; | |
| 3253 prevchr = pr; | |
| 3254 prevprevchr = prpr; | |
| 3255 } | |
| 3256 | |
| 4444 | 3257 /* |
| 3258 * Get the next character from the pattern. We know about magic and such, so | |
| 3259 * therefore we need a lexical analyzer. | |
| 3260 */ | |
| 7 | 3261 static int |
| 3262 getchr() | |
| 3263 { | |
| 3264 int chr = peekchr(); | |
| 3265 | |
| 3266 skipchr(); | |
| 3267 return chr; | |
| 3268 } | |
| 3269 | |
| 3270 /* | |
| 3271 * put character back. Works only once! | |
| 3272 */ | |
| 3273 static void | |
| 3274 ungetchr() | |
| 3275 { | |
| 3276 nextchr = curchr; | |
| 3277 curchr = prevchr; | |
| 3278 prevchr = prevprevchr; | |
| 3279 at_start = prev_at_start; | |
| 3280 prev_at_start = FALSE; | |
| 3281 | |
| 3282 /* Backup regparse, so that it's at the same position as before the | |
| 3283 * getchr(). */ | |
| 3284 regparse -= prevchr_len; | |
| 3285 } | |
| 3286 | |
| 3287 /* | |
| 29 | 3288 * Get and return the value of the hex string at the current position. |
| 3289 * Return -1 if there is no valid hex number. | |
| 3290 * The position is updated: | |
| 24 | 3291 * blahblah\%x20asdf |
| 856 | 3292 * before-^ ^-after |
| 24 | 3293 * The parameter controls the maximum number of input characters. This will be |
| 3294 * 2 when reading a \%x20 sequence and 4 when reading a \%u20AC sequence. | |
| 3295 */ | |
| 3296 static int | |
| 3297 gethexchrs(maxinputlen) | |
| 3298 int maxinputlen; | |
| 3299 { | |
| 3300 int nr = 0; | |
| 3301 int c; | |
| 3302 int i; | |
| 3303 | |
| 3304 for (i = 0; i < maxinputlen; ++i) | |
| 3305 { | |
| 3306 c = regparse[0]; | |
| 3307 if (!vim_isxdigit(c)) | |
| 3308 break; | |
| 3309 nr <<= 4; | |
| 3310 nr |= hex2nr(c); | |
| 3311 ++regparse; | |
| 3312 } | |
| 3313 | |
| 3314 if (i == 0) | |
| 3315 return -1; | |
| 3316 return nr; | |
| 3317 } | |
| 3318 | |
| 3319 /* | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3320 * Get and return the value of the decimal string immediately after the |
| 24 | 3321 * current position. Return -1 for invalid. Consumes all digits. |
| 3322 */ | |
| 3323 static int | |
| 3324 getdecchrs() | |
| 3325 { | |
| 3326 int nr = 0; | |
| 3327 int c; | |
| 3328 int i; | |
| 3329 | |
| 3330 for (i = 0; ; ++i) | |
| 3331 { | |
| 3332 c = regparse[0]; | |
| 3333 if (c < '0' || c > '9') | |
| 3334 break; | |
| 3335 nr *= 10; | |
| 3336 nr += c - '0'; | |
| 3337 ++regparse; | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
3338 curchr = -1; /* no longer valid */ |
| 24 | 3339 } |
| 3340 | |
| 3341 if (i == 0) | |
| 3342 return -1; | |
| 3343 return nr; | |
| 3344 } | |
| 3345 | |
| 3346 /* | |
| 3347 * get and return the value of the octal string immediately after the current | |
| 3348 * position. Return -1 for invalid, or 0-255 for valid. Smart enough to handle | |
| 3349 * numbers > 377 correctly (for example, 400 is treated as 40) and doesn't | |
| 3350 * treat 8 or 9 as recognised characters. Position is updated: | |
| 3351 * blahblah\%o210asdf | |
| 856 | 3352 * before-^ ^-after |
| 24 | 3353 */ |
| 3354 static int | |
| 3355 getoctchrs() | |
| 3356 { | |
| 3357 int nr = 0; | |
| 3358 int c; | |
| 3359 int i; | |
| 3360 | |
| 3361 for (i = 0; i < 3 && nr < 040; ++i) | |
| 3362 { | |
| 3363 c = regparse[0]; | |
| 3364 if (c < '0' || c > '7') | |
| 3365 break; | |
| 3366 nr <<= 3; | |
| 3367 nr |= hex2nr(c); | |
| 3368 ++regparse; | |
| 3369 } | |
| 3370 | |
| 3371 if (i == 0) | |
| 3372 return -1; | |
| 3373 return nr; | |
| 3374 } | |
| 3375 | |
| 3376 /* | |
| 3377 * Get a number after a backslash that is inside []. | |
| 3378 * When nothing is recognized return a backslash. | |
| 3379 */ | |
| 3380 static int | |
| 3381 coll_get_char() | |
| 3382 { | |
| 3383 int nr = -1; | |
| 3384 | |
| 3385 switch (*regparse++) | |
| 3386 { | |
| 3387 case 'd': nr = getdecchrs(); break; | |
| 3388 case 'o': nr = getoctchrs(); break; | |
| 3389 case 'x': nr = gethexchrs(2); break; | |
| 3390 case 'u': nr = gethexchrs(4); break; | |
| 3391 case 'U': nr = gethexchrs(8); break; | |
| 3392 } | |
| 3393 if (nr < 0) | |
| 3394 { | |
| 3395 /* If getting the number fails be backwards compatible: the character | |
| 3396 * is a backslash. */ | |
| 3397 --regparse; | |
| 3398 nr = '\\'; | |
| 3399 } | |
| 3400 return nr; | |
| 3401 } | |
| 3402 | |
| 3403 /* | |
| 7 | 3404 * read_limits - Read two integers to be taken as a minimum and maximum. |
| 3405 * If the first character is '-', then the range is reversed. | |
| 3406 * Should end with 'end'. If minval is missing, zero is default, if maxval is | |
| 3407 * missing, a very big number is the default. | |
| 3408 */ | |
| 3409 static int | |
| 3410 read_limits(minval, maxval) | |
| 3411 long *minval; | |
| 3412 long *maxval; | |
| 3413 { | |
| 3414 int reverse = FALSE; | |
| 3415 char_u *first_char; | |
| 3416 long tmp; | |
| 3417 | |
| 3418 if (*regparse == '-') | |
| 3419 { | |
| 3420 /* Starts with '-', so reverse the range later */ | |
| 3421 regparse++; | |
| 3422 reverse = TRUE; | |
| 3423 } | |
| 3424 first_char = regparse; | |
| 3425 *minval = getdigits(®parse); | |
| 3426 if (*regparse == ',') /* There is a comma */ | |
| 3427 { | |
| 3428 if (vim_isdigit(*++regparse)) | |
| 3429 *maxval = getdigits(®parse); | |
| 3430 else | |
| 3431 *maxval = MAX_LIMIT; | |
| 3432 } | |
| 3433 else if (VIM_ISDIGIT(*first_char)) | |
| 3434 *maxval = *minval; /* It was \{n} or \{-n} */ | |
| 3435 else | |
| 3436 *maxval = MAX_LIMIT; /* It was \{} or \{-} */ | |
| 3437 if (*regparse == '\\') | |
| 3438 regparse++; /* Allow either \{...} or \{...\} */ | |
| 167 | 3439 if (*regparse != '}') |
| 7 | 3440 { |
| 3441 sprintf((char *)IObuff, _("E554: Syntax error in %s{...}"), | |
| 3442 reg_magic == MAGIC_ALL ? "" : "\\"); | |
| 3443 EMSG_RET_FAIL(IObuff); | |
| 3444 } | |
| 3445 | |
| 3446 /* | |
| 3447 * Reverse the range if there was a '-', or make sure it is in the right | |
| 3448 * order otherwise. | |
| 3449 */ | |
| 3450 if ((!reverse && *minval > *maxval) || (reverse && *minval < *maxval)) | |
| 3451 { | |
| 3452 tmp = *minval; | |
| 3453 *minval = *maxval; | |
| 3454 *maxval = tmp; | |
| 3455 } | |
| 3456 skipchr(); /* let's be friends with the lexer again */ | |
| 3457 return OK; | |
| 3458 } | |
| 3459 | |
| 3460 /* | |
| 3461 * vim_regexec and friends | |
| 3462 */ | |
| 3463 | |
| 3464 /* | |
| 3465 * Global work variables for vim_regexec(). | |
| 3466 */ | |
| 3467 | |
| 3468 /* The current match-position is remembered with these variables: */ | |
| 3469 static linenr_T reglnum; /* line number, relative to first line */ | |
| 3470 static char_u *regline; /* start of current line */ | |
| 3471 static char_u *reginput; /* current input, points into "regline" */ | |
| 3472 | |
| 3473 static int need_clear_subexpr; /* subexpressions still need to be | |
| 3474 * cleared */ | |
| 3475 #ifdef FEAT_SYN_HL | |
| 3476 static int need_clear_zsubexpr = FALSE; /* extmatch subexpressions | |
| 3477 * still need to be cleared */ | |
| 3478 #endif | |
| 3479 | |
| 3480 /* | |
| 3481 * Structure used to save the current input state, when it needs to be | |
| 3482 * restored after trying a match. Used by reg_save() and reg_restore(). | |
| 233 | 3483 * Also stores the length of "backpos". |
| 7 | 3484 */ |
| 3485 typedef struct | |
| 3486 { | |
| 3487 union | |
| 3488 { | |
| 3489 char_u *ptr; /* reginput pointer, for single-line regexp */ | |
| 3490 lpos_T pos; /* reginput pos, for multi-line regexp */ | |
| 3491 } rs_u; | |
| 233 | 3492 int rs_len; |
| 7 | 3493 } regsave_T; |
| 3494 | |
| 3495 /* struct to save start/end pointer/position in for \(\) */ | |
| 3496 typedef struct | |
| 3497 { | |
| 3498 union | |
| 3499 { | |
| 3500 char_u *ptr; | |
| 3501 lpos_T pos; | |
| 3502 } se_u; | |
| 3503 } save_se_T; | |
| 3504 | |
| 1579 | 3505 /* used for BEHIND and NOBEHIND matching */ |
| 3506 typedef struct regbehind_S | |
| 3507 { | |
| 3508 regsave_T save_after; | |
| 3509 regsave_T save_behind; | |
| 1602 | 3510 int save_need_clear_subexpr; |
| 1579 | 3511 save_se_T save_start[NSUBEXP]; |
| 3512 save_se_T save_end[NSUBEXP]; | |
| 3513 } regbehind_T; | |
| 3514 | |
| 7 | 3515 static char_u *reg_getline __ARGS((linenr_T lnum)); |
| 4444 | 3516 static long bt_regexec_both __ARGS((char_u *line, colnr_T col, proftime_T *tm)); |
| 3517 static long regtry __ARGS((bt_regprog_T *prog, colnr_T col)); | |
| 7 | 3518 static void cleanup_subexpr __ARGS((void)); |
| 3519 #ifdef FEAT_SYN_HL | |
| 3520 static void cleanup_zsubexpr __ARGS((void)); | |
| 3521 #endif | |
| 1579 | 3522 static void save_subexpr __ARGS((regbehind_T *bp)); |
| 3523 static void restore_subexpr __ARGS((regbehind_T *bp)); | |
| 7 | 3524 static void reg_nextline __ARGS((void)); |
| 233 | 3525 static void reg_save __ARGS((regsave_T *save, garray_T *gap)); |
| 3526 static void reg_restore __ARGS((regsave_T *save, garray_T *gap)); | |
| 7 | 3527 static int reg_save_equal __ARGS((regsave_T *save)); |
| 3528 static void save_se_multi __ARGS((save_se_T *savep, lpos_T *posp)); | |
| 3529 static void save_se_one __ARGS((save_se_T *savep, char_u **pp)); | |
| 3530 | |
| 3531 /* Save the sub-expressions before attempting a match. */ | |
| 3532 #define save_se(savep, posp, pp) \ | |
| 3533 REG_MULTI ? save_se_multi((savep), (posp)) : save_se_one((savep), (pp)) | |
| 3534 | |
| 3535 /* After a failed match restore the sub-expressions. */ | |
| 3536 #define restore_se(savep, posp, pp) { \ | |
| 3537 if (REG_MULTI) \ | |
| 3538 *(posp) = (savep)->se_u.pos; \ | |
| 3539 else \ | |
| 3540 *(pp) = (savep)->se_u.ptr; } | |
| 3541 | |
| 3542 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
|
3543 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 | 3544 static int regmatch __ARGS((char_u *prog)); |
| 7 | 3545 static int regrepeat __ARGS((char_u *p, long maxcount)); |
| 3546 | |
| 3547 #ifdef DEBUG | |
| 3548 int regnarrate = 0; | |
| 3549 #endif | |
| 3550 | |
| 3551 /* | |
| 3552 * Internal copy of 'ignorecase'. It is set at each call to vim_regexec(). | |
| 3553 * Normally it gets the value of "rm_ic" or "rmm_ic", but when the pattern | |
| 3554 * contains '\c' or '\C' the value is overruled. | |
| 3555 */ | |
| 3556 static int ireg_ic; | |
| 3557 | |
| 3558 #ifdef FEAT_MBYTE | |
| 3559 /* | |
| 3560 * Similar to ireg_ic, but only for 'combining' characters. Set with \Z flag | |
| 3561 * in the regexp. Defaults to false, always. | |
| 3562 */ | |
| 3563 static int ireg_icombine; | |
| 3564 #endif | |
| 3565 | |
| 3566 /* | |
| 410 | 3567 * Copy of "rmm_maxcol": maximum column to search for a match. Zero when |
| 3568 * there is no maximum. | |
| 3569 */ | |
| 418 | 3570 static colnr_T ireg_maxcol; |
| 410 | 3571 |
| 3572 /* | |
| 7 | 3573 * Sometimes need to save a copy of a line. Since alloc()/free() is very |
| 3574 * slow, we keep one allocated piece of memory and only re-allocate it when | |
| 4444 | 3575 * it's too small. It's freed in bt_regexec_both() when finished. |
| 7 | 3576 */ |
| 1468 | 3577 static char_u *reg_tofree = NULL; |
| 7 | 3578 static unsigned reg_tofreelen; |
| 3579 | |
| 3580 /* | |
| 3581 * These variables are set when executing a regexp to speed up the execution. | |
| 1209 | 3582 * Which ones are set depends on whether a single-line or multi-line match is |
| 7 | 3583 * done: |
| 3584 * single-line multi-line | |
| 3585 * reg_match ®match_T NULL | |
| 3586 * reg_mmatch NULL ®mmatch_T | |
| 3587 * reg_startp reg_match->startp <invalid> | |
| 3588 * reg_endp reg_match->endp <invalid> | |
| 3589 * reg_startpos <invalid> reg_mmatch->startpos | |
| 3590 * reg_endpos <invalid> reg_mmatch->endpos | |
| 3591 * reg_win NULL window in which to search | |
| 4061 | 3592 * reg_buf curbuf buffer in which to search |
| 7 | 3593 * reg_firstlnum <invalid> first line in which to search |
| 3594 * reg_maxline 0 last line nr | |
| 3595 * reg_line_lbr FALSE or TRUE FALSE | |
| 3596 */ | |
| 3597 static regmatch_T *reg_match; | |
| 3598 static regmmatch_T *reg_mmatch; | |
| 3599 static char_u **reg_startp = NULL; | |
| 3600 static char_u **reg_endp = NULL; | |
| 3601 static lpos_T *reg_startpos = NULL; | |
| 3602 static lpos_T *reg_endpos = NULL; | |
| 3603 static win_T *reg_win; | |
| 3604 static buf_T *reg_buf; | |
| 3605 static linenr_T reg_firstlnum; | |
| 3606 static linenr_T reg_maxline; | |
| 3607 static int reg_line_lbr; /* "\n" in string is line break */ | |
| 3608 | |
| 270 | 3609 /* Values for rs_state in regitem_T. */ |
| 3610 typedef enum regstate_E | |
| 3611 { | |
| 3612 RS_NOPEN = 0 /* NOPEN and NCLOSE */ | |
| 3613 , RS_MOPEN /* MOPEN + [0-9] */ | |
| 3614 , RS_MCLOSE /* MCLOSE + [0-9] */ | |
| 3615 #ifdef FEAT_SYN_HL | |
| 3616 , RS_ZOPEN /* ZOPEN + [0-9] */ | |
| 3617 , RS_ZCLOSE /* ZCLOSE + [0-9] */ | |
| 3618 #endif | |
| 3619 , RS_BRANCH /* BRANCH */ | |
| 3620 , RS_BRCPLX_MORE /* BRACE_COMPLEX and trying one more match */ | |
| 3621 , RS_BRCPLX_LONG /* BRACE_COMPLEX and trying longest match */ | |
| 3622 , RS_BRCPLX_SHORT /* BRACE_COMPLEX and trying shortest match */ | |
| 3623 , RS_NOMATCH /* NOMATCH */ | |
| 3624 , RS_BEHIND1 /* BEHIND / NOBEHIND matching rest */ | |
| 3625 , RS_BEHIND2 /* BEHIND / NOBEHIND matching behind part */ | |
| 3626 , RS_STAR_LONG /* STAR/PLUS/BRACE_SIMPLE longest match */ | |
| 3627 , RS_STAR_SHORT /* STAR/PLUS/BRACE_SIMPLE shortest match */ | |
| 3628 } regstate_T; | |
| 3629 | |
| 3630 /* | |
| 3631 * When there are alternatives a regstate_T is put on the regstack to remember | |
| 3632 * what we are doing. | |
| 3633 * Before it may be another type of item, depending on rs_state, to remember | |
| 3634 * more things. | |
| 3635 */ | |
| 3636 typedef struct regitem_S | |
| 3637 { | |
| 3638 regstate_T rs_state; /* what we are doing, one of RS_ above */ | |
| 3639 char_u *rs_scan; /* current node in program */ | |
| 3640 union | |
| 3641 { | |
| 3642 save_se_T sesave; | |
| 3643 regsave_T regsave; | |
| 3644 } rs_un; /* room for saving reginput */ | |
| 1579 | 3645 short rs_no; /* submatch nr or BEHIND/NOBEHIND */ |
| 270 | 3646 } regitem_T; |
| 3647 | |
| 3648 static regitem_T *regstack_push __ARGS((regstate_T state, char_u *scan)); | |
| 3649 static void regstack_pop __ARGS((char_u **scan)); | |
| 3650 | |
| 3651 /* used for STAR, PLUS and BRACE_SIMPLE matching */ | |
| 3652 typedef struct regstar_S | |
| 3653 { | |
| 3654 int nextb; /* next byte */ | |
| 3655 int nextb_ic; /* next byte reverse case */ | |
| 3656 long count; | |
| 3657 long minval; | |
| 3658 long maxval; | |
| 3659 } regstar_T; | |
| 3660 | |
| 3661 /* used to store input position when a BACK was encountered, so that we now if | |
| 3662 * we made any progress since the last time. */ | |
| 3663 typedef struct backpos_S | |
| 3664 { | |
| 3665 char_u *bp_scan; /* "scan" where BACK was encountered */ | |
| 3666 regsave_T bp_pos; /* last input position */ | |
| 3667 } backpos_T; | |
| 3668 | |
| 3669 /* | |
| 1520 | 3670 * "regstack" and "backpos" are used by regmatch(). They are kept over calls |
| 3671 * to avoid invoking malloc() and free() often. | |
| 3672 * "regstack" is a stack with regitem_T items, sometimes preceded by regstar_T | |
| 3673 * or regbehind_T. | |
| 3674 * "backpos_T" is a table with backpos_T for BACK | |
| 270 | 3675 */ |
| 1520 | 3676 static garray_T regstack = {0, 0, 0, 0, NULL}; |
| 3677 static garray_T backpos = {0, 0, 0, 0, NULL}; | |
| 3678 | |
| 3679 /* | |
| 3680 * Both for regstack and backpos tables we use the following strategy of | |
| 3681 * allocation (to reduce malloc/free calls): | |
| 3682 * - Initial size is fairly small. | |
| 3683 * - When needed, the tables are grown bigger (8 times at first, double after | |
| 3684 * that). | |
| 3685 * - After executing the match we free the memory only if the array has grown. | |
| 3686 * Thus the memory is kept allocated when it's at the initial size. | |
| 3687 * This makes it fast while not keeping a lot of memory allocated. | |
| 3688 * A three times speed increase was observed when using many simple patterns. | |
| 3689 */ | |
| 3690 #define REGSTACK_INITIAL 2048 | |
| 3691 #define BACKPOS_INITIAL 64 | |
| 3692 | |
| 3693 #if defined(EXITFREE) || defined(PROTO) | |
| 3694 void | |
| 3695 free_regexp_stuff() | |
| 3696 { | |
| 3697 ga_clear(®stack); | |
| 3698 ga_clear(&backpos); | |
| 3699 vim_free(reg_tofree); | |
| 3700 vim_free(reg_prev_sub); | |
| 3701 } | |
| 3702 #endif | |
| 270 | 3703 |
| 7 | 3704 /* |
| 3705 * Get pointer to the line "lnum", which is relative to "reg_firstlnum". | |
| 3706 */ | |
| 3707 static char_u * | |
| 3708 reg_getline(lnum) | |
| 3709 linenr_T lnum; | |
| 3710 { | |
| 3711 /* when looking behind for a match/no-match lnum is negative. But we | |
| 3712 * can't go before line 1 */ | |
| 3713 if (reg_firstlnum + lnum < 1) | |
| 3714 return NULL; | |
| 482 | 3715 if (lnum > reg_maxline) |
| 481 | 3716 /* Must have matched the "\n" in the last line. */ |
| 3717 return (char_u *)""; | |
| 7 | 3718 return ml_get_buf(reg_buf, reg_firstlnum + lnum, FALSE); |
| 3719 } | |
| 3720 | |
| 3721 static regsave_T behind_pos; | |
| 3722 | |
| 3723 #ifdef FEAT_SYN_HL | |
| 3724 static char_u *reg_startzp[NSUBEXP]; /* Workspace to mark beginning */ | |
| 3725 static char_u *reg_endzp[NSUBEXP]; /* and end of \z(...\) matches */ | |
| 3726 static lpos_T reg_startzpos[NSUBEXP]; /* idem, beginning pos */ | |
| 3727 static lpos_T reg_endzpos[NSUBEXP]; /* idem, end pos */ | |
| 3728 #endif | |
| 3729 | |
| 3730 /* TRUE if using multi-line regexp. */ | |
| 3731 #define REG_MULTI (reg_match == NULL) | |
| 3732 | |
| 5838 | 3733 static int bt_regexec_nl __ARGS((regmatch_T *rmp, char_u *line, colnr_T col, int line_lbr)); |
| 3734 | |
| 4444 | 3735 |
| 7 | 3736 /* |
| 3737 * Match a regexp against a string. | |
| 3738 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3739 * Uses curbuf for line count and 'iskeyword'. | |
| 5838 | 3740 * if "line_lbr" is TRUE consider a "\n" in "line" to be a line break. |
| 7 | 3741 * |
| 6390 | 3742 * Returns 0 for failure, number of lines contained in the match otherwise. |
| 7 | 3743 */ |
| 4444 | 3744 static int |
| 5838 | 3745 bt_regexec_nl(rmp, line, col, line_lbr) |
| 7 | 3746 regmatch_T *rmp; |
| 3747 char_u *line; /* string to match against */ | |
| 3748 colnr_T col; /* column to start looking for match */ | |
| 5838 | 3749 int line_lbr; |
| 7 | 3750 { |
| 3751 reg_match = rmp; | |
| 3752 reg_mmatch = NULL; | |
| 3753 reg_maxline = 0; | |
| 5838 | 3754 reg_line_lbr = line_lbr; |
| 4061 | 3755 reg_buf = curbuf; |
| 7 | 3756 reg_win = NULL; |
| 3757 ireg_ic = rmp->rm_ic; | |
| 3758 #ifdef FEAT_MBYTE | |
| 3759 ireg_icombine = FALSE; | |
| 3760 #endif | |
| 410 | 3761 ireg_maxcol = 0; |
| 6390 | 3762 |
| 3763 return bt_regexec_both(line, col, NULL); | |
| 7 | 3764 } |
| 3765 | |
| 4444 | 3766 static long bt_regexec_multi __ARGS((regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm)); |
| 3767 | |
| 7 | 3768 /* |
| 3769 * Match a regexp against multiple lines. | |
| 3770 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3771 * Uses curbuf for line count and 'iskeyword'. | |
| 3772 * | |
| 3773 * Return zero if there is no match. Return number of lines contained in the | |
| 3774 * match otherwise. | |
| 3775 */ | |
| 4444 | 3776 static long |
| 3777 bt_regexec_multi(rmp, win, buf, lnum, col, tm) | |
| 7 | 3778 regmmatch_T *rmp; |
| 3779 win_T *win; /* window in which to search or NULL */ | |
| 3780 buf_T *buf; /* buffer in which to search */ | |
| 3781 linenr_T lnum; /* nr of line to start looking for match */ | |
| 3782 colnr_T col; /* column to start looking for match */ | |
| 1521 | 3783 proftime_T *tm; /* timeout limit or NULL */ |
| 7 | 3784 { |
| 3785 reg_match = NULL; | |
| 3786 reg_mmatch = rmp; | |
| 3787 reg_buf = buf; | |
| 3788 reg_win = win; | |
| 3789 reg_firstlnum = lnum; | |
| 3790 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
| 3791 reg_line_lbr = FALSE; | |
| 3792 ireg_ic = rmp->rmm_ic; | |
| 3793 #ifdef FEAT_MBYTE | |
| 3794 ireg_icombine = FALSE; | |
| 3795 #endif | |
| 410 | 3796 ireg_maxcol = rmp->rmm_maxcol; |
| 7 | 3797 |
| 6390 | 3798 return bt_regexec_both(NULL, col, tm); |
| 7 | 3799 } |
| 3800 | |
| 3801 /* | |
| 3802 * Match a regexp against a string ("line" points to the string) or multiple | |
| 3803 * lines ("line" is NULL, use reg_getline()). | |
| 6390 | 3804 * Returns 0 for failure, number of lines contained in the match otherwise. |
| 7 | 3805 */ |
| 3806 static long | |
| 4444 | 3807 bt_regexec_both(line, col, tm) |
| 7 | 3808 char_u *line; |
| 3809 colnr_T col; /* column to start looking for match */ | |
| 1877 | 3810 proftime_T *tm UNUSED; /* timeout limit or NULL */ |
| 7 | 3811 { |
| 6390 | 3812 bt_regprog_T *prog; |
| 3813 char_u *s; | |
| 3814 long retval = 0L; | |
| 7 | 3815 |
| 1520 | 3816 /* Create "regstack" and "backpos" if they are not allocated yet. |
| 3817 * We allocate *_INITIAL amount of bytes first and then set the grow size | |
| 3818 * to much bigger value to avoid many malloc calls in case of deep regular | |
| 3819 * expressions. */ | |
| 3820 if (regstack.ga_data == NULL) | |
| 3821 { | |
| 3822 /* Use an item size of 1 byte, since we push different things | |
| 3823 * onto the regstack. */ | |
| 3824 ga_init2(®stack, 1, REGSTACK_INITIAL); | |
| 3825 ga_grow(®stack, REGSTACK_INITIAL); | |
| 3826 regstack.ga_growsize = REGSTACK_INITIAL * 8; | |
| 3827 } | |
| 3828 | |
| 3829 if (backpos.ga_data == NULL) | |
| 3830 { | |
| 3831 ga_init2(&backpos, sizeof(backpos_T), BACKPOS_INITIAL); | |
| 3832 ga_grow(&backpos, BACKPOS_INITIAL); | |
| 3833 backpos.ga_growsize = BACKPOS_INITIAL * 8; | |
| 3834 } | |
| 270 | 3835 |
| 7 | 3836 if (REG_MULTI) |
| 3837 { | |
| 4444 | 3838 prog = (bt_regprog_T *)reg_mmatch->regprog; |
| 7 | 3839 line = reg_getline((linenr_T)0); |
| 3840 reg_startpos = reg_mmatch->startpos; | |
| 3841 reg_endpos = reg_mmatch->endpos; | |
| 3842 } | |
| 3843 else | |
| 3844 { | |
| 4444 | 3845 prog = (bt_regprog_T *)reg_match->regprog; |
| 7 | 3846 reg_startp = reg_match->startp; |
| 3847 reg_endp = reg_match->endp; | |
| 3848 } | |
| 3849 | |
| 3850 /* Be paranoid... */ | |
| 3851 if (prog == NULL || line == NULL) | |
| 3852 { | |
| 3853 EMSG(_(e_null)); | |
| 3854 goto theend; | |
| 3855 } | |
| 3856 | |
| 3857 /* Check validity of program. */ | |
| 3858 if (prog_magic_wrong()) | |
| 3859 goto theend; | |
| 3860 | |
| 410 | 3861 /* If the start column is past the maximum column: no need to try. */ |
| 3862 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3863 goto theend; | |
| 3864 | |
| 7 | 3865 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ |
| 3866 if (prog->regflags & RF_ICASE) | |
| 3867 ireg_ic = TRUE; | |
| 3868 else if (prog->regflags & RF_NOICASE) | |
| 3869 ireg_ic = FALSE; | |
| 3870 | |
| 3871 #ifdef FEAT_MBYTE | |
| 3872 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
| 3873 if (prog->regflags & RF_ICOMBINE) | |
| 3874 ireg_icombine = TRUE; | |
| 3875 #endif | |
| 3876 | |
| 3877 /* If there is a "must appear" string, look for it. */ | |
| 3878 if (prog->regmust != NULL) | |
| 3879 { | |
| 3880 int c; | |
| 3881 | |
| 3882 #ifdef FEAT_MBYTE | |
| 3883 if (has_mbyte) | |
| 3884 c = (*mb_ptr2char)(prog->regmust); | |
| 3885 else | |
| 3886 #endif | |
| 3887 c = *prog->regmust; | |
| 3888 s = line + col; | |
| 170 | 3889 |
| 3890 /* | |
| 3891 * This is used very often, esp. for ":global". Use three versions of | |
| 3892 * the loop to avoid overhead of conditions. | |
| 3893 */ | |
| 3894 if (!ireg_ic | |
| 3895 #ifdef FEAT_MBYTE | |
| 3896 && !has_mbyte | |
| 3897 #endif | |
| 3898 ) | |
| 3899 while ((s = vim_strbyte(s, c)) != NULL) | |
| 3900 { | |
| 3901 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3902 break; /* Found it. */ | |
| 3903 ++s; | |
| 3904 } | |
| 3905 #ifdef FEAT_MBYTE | |
| 3906 else if (!ireg_ic || (!enc_utf8 && mb_char2len(c) > 1)) | |
| 3907 while ((s = vim_strchr(s, c)) != NULL) | |
| 3908 { | |
| 3909 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3910 break; /* Found it. */ | |
| 3911 mb_ptr_adv(s); | |
| 3912 } | |
| 3913 #endif | |
| 3914 else | |
| 3915 while ((s = cstrchr(s, c)) != NULL) | |
| 3916 { | |
| 3917 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3918 break; /* Found it. */ | |
| 3919 mb_ptr_adv(s); | |
| 3920 } | |
| 7 | 3921 if (s == NULL) /* Not present. */ |
| 3922 goto theend; | |
| 3923 } | |
| 3924 | |
| 3925 regline = line; | |
| 3926 reglnum = 0; | |
| 2578 | 3927 reg_toolong = FALSE; |
| 7 | 3928 |
| 3929 /* Simplest case: Anchored match need be tried only once. */ | |
| 3930 if (prog->reganch) | |
| 3931 { | |
| 3932 int c; | |
| 3933 | |
| 3934 #ifdef FEAT_MBYTE | |
| 3935 if (has_mbyte) | |
| 3936 c = (*mb_ptr2char)(regline + col); | |
| 3937 else | |
| 3938 #endif | |
| 3939 c = regline[col]; | |
| 3940 if (prog->regstart == NUL | |
| 3941 || prog->regstart == c | |
| 3942 || (ireg_ic && (( | |
| 3943 #ifdef FEAT_MBYTE | |
| 3944 (enc_utf8 && utf_fold(prog->regstart) == utf_fold(c))) | |
| 3945 || (c < 255 && prog->regstart < 255 && | |
| 3946 #endif | |
| 1347 | 3947 MB_TOLOWER(prog->regstart) == MB_TOLOWER(c))))) |
| 7 | 3948 retval = regtry(prog, col); |
| 3949 else | |
| 3950 retval = 0; | |
| 3951 } | |
| 3952 else | |
| 3953 { | |
| 1521 | 3954 #ifdef FEAT_RELTIME |
| 3955 int tm_count = 0; | |
| 3956 #endif | |
| 7 | 3957 /* Messy cases: unanchored match. */ |
| 180 | 3958 while (!got_int) |
| 7 | 3959 { |
| 3960 if (prog->regstart != NUL) | |
| 3961 { | |
| 170 | 3962 /* Skip until the char we know it must start with. |
| 3963 * Used often, do some work to avoid call overhead. */ | |
| 3964 if (!ireg_ic | |
| 3965 #ifdef FEAT_MBYTE | |
| 3966 && !has_mbyte | |
| 3967 #endif | |
| 3968 ) | |
| 3969 s = vim_strbyte(regline + col, prog->regstart); | |
| 3970 else | |
| 3971 s = cstrchr(regline + col, prog->regstart); | |
| 7 | 3972 if (s == NULL) |
| 3973 { | |
| 3974 retval = 0; | |
| 3975 break; | |
| 3976 } | |
| 3977 col = (int)(s - regline); | |
| 3978 } | |
| 3979 | |
| 410 | 3980 /* Check for maximum column to try. */ |
| 3981 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3982 { | |
| 3983 retval = 0; | |
| 3984 break; | |
| 3985 } | |
| 3986 | |
| 7 | 3987 retval = regtry(prog, col); |
| 3988 if (retval > 0) | |
| 3989 break; | |
| 3990 | |
| 3991 /* if not currently on the first line, get it again */ | |
| 3992 if (reglnum != 0) | |
| 3993 { | |
| 481 | 3994 reglnum = 0; |
| 7 | 3995 regline = reg_getline((linenr_T)0); |
| 3996 } | |
| 3997 if (regline[col] == NUL) | |
| 3998 break; | |
| 3999 #ifdef FEAT_MBYTE | |
| 4000 if (has_mbyte) | |
| 474 | 4001 col += (*mb_ptr2len)(regline + col); |
| 7 | 4002 else |
| 4003 #endif | |
| 4004 ++col; | |
| 1521 | 4005 #ifdef FEAT_RELTIME |
| 4006 /* Check for timeout once in a twenty times to avoid overhead. */ | |
| 4007 if (tm != NULL && ++tm_count == 20) | |
| 4008 { | |
| 4009 tm_count = 0; | |
| 4010 if (profile_passed_limit(tm)) | |
| 4011 break; | |
| 4012 } | |
| 4013 #endif | |
| 7 | 4014 } |
| 4015 } | |
| 4016 | |
| 4017 theend: | |
| 1520 | 4018 /* Free "reg_tofree" when it's a bit big. |
| 4019 * Free regstack and backpos if they are bigger than their initial size. */ | |
| 4020 if (reg_tofreelen > 400) | |
| 4021 { | |
| 4022 vim_free(reg_tofree); | |
| 4023 reg_tofree = NULL; | |
| 4024 } | |
| 4025 if (regstack.ga_maxlen > REGSTACK_INITIAL) | |
| 4026 ga_clear(®stack); | |
| 4027 if (backpos.ga_maxlen > BACKPOS_INITIAL) | |
| 4028 ga_clear(&backpos); | |
| 270 | 4029 |
| 7 | 4030 return retval; |
| 4031 } | |
| 4032 | |
| 4033 #ifdef FEAT_SYN_HL | |
| 4034 static reg_extmatch_T *make_extmatch __ARGS((void)); | |
| 4035 | |
| 4036 /* | |
| 4037 * Create a new extmatch and mark it as referenced once. | |
| 4038 */ | |
| 4039 static reg_extmatch_T * | |
| 4040 make_extmatch() | |
| 4041 { | |
| 4042 reg_extmatch_T *em; | |
| 4043 | |
| 4044 em = (reg_extmatch_T *)alloc_clear((unsigned)sizeof(reg_extmatch_T)); | |
| 4045 if (em != NULL) | |
| 4046 em->refcnt = 1; | |
| 4047 return em; | |
| 4048 } | |
| 4049 | |
| 4050 /* | |
| 4051 * Add a reference to an extmatch. | |
| 4052 */ | |
| 4053 reg_extmatch_T * | |
| 4054 ref_extmatch(em) | |
| 4055 reg_extmatch_T *em; | |
| 4056 { | |
| 4057 if (em != NULL) | |
| 4058 em->refcnt++; | |
| 4059 return em; | |
| 4060 } | |
| 4061 | |
| 4062 /* | |
| 4063 * Remove a reference to an extmatch. If there are no references left, free | |
| 4064 * the info. | |
| 4065 */ | |
| 4066 void | |
| 4067 unref_extmatch(em) | |
| 4068 reg_extmatch_T *em; | |
| 4069 { | |
| 4070 int i; | |
| 4071 | |
| 4072 if (em != NULL && --em->refcnt <= 0) | |
| 4073 { | |
| 4074 for (i = 0; i < NSUBEXP; ++i) | |
| 4075 vim_free(em->matches[i]); | |
| 4076 vim_free(em); | |
| 4077 } | |
| 4078 } | |
| 4079 #endif | |
| 4080 | |
| 4081 /* | |
| 4082 * regtry - try match of "prog" with at regline["col"]. | |
| 4083 * Returns 0 for failure, number of lines contained in the match otherwise. | |
| 4084 */ | |
| 4085 static long | |
| 4086 regtry(prog, col) | |
| 4444 | 4087 bt_regprog_T *prog; |
| 7 | 4088 colnr_T col; |
| 4089 { | |
| 4090 reginput = regline + col; | |
| 4091 need_clear_subexpr = TRUE; | |
| 4092 #ifdef FEAT_SYN_HL | |
| 4093 /* Clear the external match subpointers if necessary. */ | |
| 4094 if (prog->reghasz == REX_SET) | |
| 4095 need_clear_zsubexpr = TRUE; | |
| 4096 #endif | |
| 4097 | |
| 189 | 4098 if (regmatch(prog->program + 1) == 0) |
| 4099 return 0; | |
| 4100 | |
| 4101 cleanup_subexpr(); | |
| 4102 if (REG_MULTI) | |
| 7 | 4103 { |
| 189 | 4104 if (reg_startpos[0].lnum < 0) |
| 7 | 4105 { |
| 189 | 4106 reg_startpos[0].lnum = 0; |
| 4107 reg_startpos[0].col = col; | |
| 4108 } | |
| 4109 if (reg_endpos[0].lnum < 0) | |
| 4110 { | |
| 4111 reg_endpos[0].lnum = reglnum; | |
| 4112 reg_endpos[0].col = (int)(reginput - regline); | |
| 7 | 4113 } |
| 4114 else | |
| 189 | 4115 /* Use line number of "\ze". */ |
| 4116 reglnum = reg_endpos[0].lnum; | |
| 4117 } | |
| 4118 else | |
| 4119 { | |
| 4120 if (reg_startp[0] == NULL) | |
| 4121 reg_startp[0] = regline + col; | |
| 4122 if (reg_endp[0] == NULL) | |
| 4123 reg_endp[0] = reginput; | |
| 4124 } | |
| 7 | 4125 #ifdef FEAT_SYN_HL |
| 189 | 4126 /* Package any found \z(...\) matches for export. Default is none. */ |
| 4127 unref_extmatch(re_extmatch_out); | |
| 4128 re_extmatch_out = NULL; | |
| 4129 | |
| 4130 if (prog->reghasz == REX_SET) | |
| 4131 { | |
| 4132 int i; | |
| 4133 | |
| 4134 cleanup_zsubexpr(); | |
| 4135 re_extmatch_out = make_extmatch(); | |
| 4136 for (i = 0; i < NSUBEXP; i++) | |
| 7 | 4137 { |
| 189 | 4138 if (REG_MULTI) |
| 7 | 4139 { |
| 189 | 4140 /* Only accept single line matches. */ |
| 4141 if (reg_startzpos[i].lnum >= 0 | |
| 5820 | 4142 && reg_endzpos[i].lnum == reg_startzpos[i].lnum |
| 4143 && reg_endzpos[i].col >= reg_startzpos[i].col) | |
| 189 | 4144 re_extmatch_out->matches[i] = |
| 4145 vim_strnsave(reg_getline(reg_startzpos[i].lnum) | |
| 7 | 4146 + reg_startzpos[i].col, |
| 189 | 4147 reg_endzpos[i].col - reg_startzpos[i].col); |
| 4148 } | |
| 4149 else | |
| 4150 { | |
| 4151 if (reg_startzp[i] != NULL && reg_endzp[i] != NULL) | |
| 4152 re_extmatch_out->matches[i] = | |
| 7 | 4153 vim_strnsave(reg_startzp[i], |
| 189 | 4154 (int)(reg_endzp[i] - reg_startzp[i])); |
| 7 | 4155 } |
| 4156 } | |
| 189 | 4157 } |
| 7 | 4158 #endif |
| 189 | 4159 return 1 + reglnum; |
| 7 | 4160 } |
| 4161 | |
| 4162 #ifdef FEAT_MBYTE | |
| 4163 static int reg_prev_class __ARGS((void)); | |
| 4164 | |
| 4165 /* | |
| 4166 * Get class of previous character. | |
| 4167 */ | |
| 4168 static int | |
| 4169 reg_prev_class() | |
| 4170 { | |
| 4171 if (reginput > regline) | |
| 4069 | 4172 return mb_get_class_buf(reginput - 1 |
| 4173 - (*mb_head_off)(regline, reginput - 1), reg_buf); | |
| 7 | 4174 return -1; |
| 4175 } | |
| 5735 | 4176 #endif |
| 4177 | |
|
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4178 static int reg_match_visual __ARGS((void)); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4179 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4180 /* |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4181 * 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
|
4182 */ |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4183 static int |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4184 reg_match_visual() |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4185 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4186 pos_T top, bot; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4187 linenr_T lnum; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4188 colnr_T col; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4189 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
|
4190 int mode; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4191 colnr_T start, end; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4192 colnr_T start2, end2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4193 colnr_T cols; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4194 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4195 /* Check if the buffer is the current buffer. */ |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4196 if (reg_buf != curbuf || VIsual.lnum == 0) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4197 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4198 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4199 if (VIsual_active) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4200 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4201 if (lt(VIsual, wp->w_cursor)) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4202 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4203 top = VIsual; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4204 bot = wp->w_cursor; |
|
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 else |
|
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 top = wp->w_cursor; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4209 bot = VIsual; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4210 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4211 mode = VIsual_mode; |
|
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 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
|
4216 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4217 top = curbuf->b_visual.vi_start; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4218 bot = curbuf->b_visual.vi_end; |
|
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 top = curbuf->b_visual.vi_end; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4223 bot = curbuf->b_visual.vi_start; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4224 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4225 mode = curbuf->b_visual.vi_mode; |
|
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 lnum = reglnum + reg_firstlnum; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4228 if (lnum < top.lnum || lnum > bot.lnum) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4229 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4230 |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4231 if (mode == 'v') |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4232 { |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4233 col = (colnr_T)(reginput - regline); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4234 if ((lnum == top.lnum && col < top.col) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4235 || (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
|
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 else if (mode == Ctrl_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 getvvcol(wp, &top, &start, NULL, &end); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4241 getvvcol(wp, &bot, &start2, NULL, &end2); |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4242 if (start2 < start) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4243 start = start2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4244 if (end2 > end) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4245 end = end2; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4246 if (top.col == MAXCOL || bot.col == MAXCOL) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4247 end = MAXCOL; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4248 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
|
4249 if (cols < start || cols > end - (*p_sel == 'e')) |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4250 return FALSE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4251 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4252 return TRUE; |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4253 } |
|
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4254 |
| 39 | 4255 #define ADVANCE_REGINPUT() mb_ptr_adv(reginput) |
| 7 | 4256 |
| 4257 /* | |
| 4258 * The arguments from BRACE_LIMITS are stored here. They are actually local | |
| 4259 * to regmatch(), but they are here to reduce the amount of stack space used | |
| 4260 * (it can be called recursively many times). | |
| 4261 */ | |
| 4262 static long bl_minval; | |
| 4263 static long bl_maxval; | |
| 4264 | |
| 4265 /* | |
| 4266 * regmatch - main matching routine | |
| 4267 * | |
| 180 | 4268 * Conceptually the strategy is simple: Check to see whether the current node |
| 4269 * matches, push an item onto the regstack and loop to see whether the rest | |
| 4270 * matches, and then act accordingly. In practice we make some effort to | |
| 4271 * avoid using the regstack, in particular by going through "ordinary" nodes | |
| 4272 * (that don't need to know whether the rest of the match failed) by a nested | |
| 4273 * loop. | |
| 7 | 4274 * |
| 4275 * Returns TRUE when there is a match. Leaves reginput and reglnum just after | |
| 4276 * the last matched character. | |
| 4277 * Returns FALSE when there is no match. Leaves reginput and reglnum in an | |
| 4278 * undefined state! | |
| 4279 */ | |
| 4280 static int | |
| 180 | 4281 regmatch(scan) |
| 7 | 4282 char_u *scan; /* Current node. */ |
| 4283 { | |
| 180 | 4284 char_u *next; /* Next node. */ |
| 4285 int op; | |
| 4286 int c; | |
| 4287 regitem_T *rp; | |
| 4288 int no; | |
| 4289 int status; /* one of the RA_ values: */ | |
| 4290 #define RA_FAIL 1 /* something failed, abort */ | |
| 4291 #define RA_CONT 2 /* continue in inner loop */ | |
| 4292 #define RA_BREAK 3 /* break inner loop */ | |
| 4293 #define RA_MATCH 4 /* successful match */ | |
| 4294 #define RA_NOMATCH 5 /* didn't match */ | |
| 270 | 4295 |
| 1520 | 4296 /* Make "regstack" and "backpos" empty. They are allocated and freed in |
| 4444 | 4297 * bt_regexec_both() to reduce malloc()/free() calls. */ |
| 270 | 4298 regstack.ga_len = 0; |
| 4299 backpos.ga_len = 0; | |
| 233 | 4300 |
| 180 | 4301 /* |
| 233 | 4302 * Repeat until "regstack" is empty. |
| 180 | 4303 */ |
| 4304 for (;;) | |
| 4305 { | |
| 5310 | 4306 /* Some patterns may take a long time to match, e.g., "\([a-z]\+\)\+Q". |
| 4307 * Allow interrupting them with CTRL-C. */ | |
| 7 | 4308 fast_breakcheck(); |
| 4309 | |
| 4310 #ifdef DEBUG | |
| 4311 if (scan != NULL && regnarrate) | |
| 4312 { | |
| 4444 | 4313 mch_errmsg((char *)regprop(scan)); |
| 7 | 4314 mch_errmsg("(\n"); |
| 4315 } | |
| 4316 #endif | |
| 180 | 4317 |
| 4318 /* | |
| 233 | 4319 * Repeat for items that can be matched sequentially, without using the |
| 180 | 4320 * regstack. |
| 4321 */ | |
| 4322 for (;;) | |
| 7 | 4323 { |
| 180 | 4324 if (got_int || scan == NULL) |
| 4325 { | |
| 4326 status = RA_FAIL; | |
| 4327 break; | |
| 4328 } | |
| 4329 status = RA_CONT; | |
| 4330 | |
| 7 | 4331 #ifdef DEBUG |
| 4332 if (regnarrate) | |
| 4333 { | |
| 4444 | 4334 mch_errmsg((char *)regprop(scan)); |
| 7 | 4335 mch_errmsg("...\n"); |
| 4336 # ifdef FEAT_SYN_HL | |
| 4337 if (re_extmatch_in != NULL) | |
| 4338 { | |
| 4339 int i; | |
| 4340 | |
| 4341 mch_errmsg(_("External submatches:\n")); | |
| 4342 for (i = 0; i < NSUBEXP; i++) | |
| 4343 { | |
| 4344 mch_errmsg(" \""); | |
| 4345 if (re_extmatch_in->matches[i] != NULL) | |
| 4444 | 4346 mch_errmsg((char *)re_extmatch_in->matches[i]); |
| 7 | 4347 mch_errmsg("\"\n"); |
| 4348 } | |
| 4349 } | |
| 4350 # endif | |
| 4351 } | |
| 4352 #endif | |
| 4353 next = regnext(scan); | |
| 4354 | |
| 4355 op = OP(scan); | |
| 4356 /* Check for character class with NL added. */ | |
| 1018 | 4357 if (!reg_line_lbr && WITH_NL(op) && REG_MULTI |
| 4358 && *reginput == NUL && reglnum <= reg_maxline) | |
| 7 | 4359 { |
| 4360 reg_nextline(); | |
| 4361 } | |
| 4362 else if (reg_line_lbr && WITH_NL(op) && *reginput == '\n') | |
| 4363 { | |
| 4364 ADVANCE_REGINPUT(); | |
| 4365 } | |
| 4366 else | |
| 4367 { | |
| 4368 if (WITH_NL(op)) | |
| 180 | 4369 op -= ADD_NL; |
| 7 | 4370 #ifdef FEAT_MBYTE |
| 4371 if (has_mbyte) | |
| 4372 c = (*mb_ptr2char)(reginput); | |
| 4373 else | |
| 4374 #endif | |
| 4375 c = *reginput; | |
| 4376 switch (op) | |
| 4377 { | |
| 4378 case BOL: | |
| 4379 if (reginput != regline) | |
| 180 | 4380 status = RA_NOMATCH; |
| 7 | 4381 break; |
| 4382 | |
| 4383 case EOL: | |
| 4384 if (c != NUL) | |
| 180 | 4385 status = RA_NOMATCH; |
| 7 | 4386 break; |
| 4387 | |
| 4388 case RE_BOF: | |
| 1458 | 4389 /* We're not at the beginning of the file when below the first |
| 4390 * line where we started, not at the start of the line or we | |
| 4391 * didn't start at the first line of the buffer. */ | |
| 7 | 4392 if (reglnum != 0 || reginput != regline |
| 1458 | 4393 || (REG_MULTI && reg_firstlnum > 1)) |
| 180 | 4394 status = RA_NOMATCH; |
| 7 | 4395 break; |
| 4396 | |
| 4397 case RE_EOF: | |
| 4398 if (reglnum != reg_maxline || c != NUL) | |
| 180 | 4399 status = RA_NOMATCH; |
| 7 | 4400 break; |
| 4401 | |
| 4402 case CURSOR: | |
| 4403 /* Check if the buffer is in a window and compare the | |
| 4404 * reg_win->w_cursor position to the match position. */ | |
| 4405 if (reg_win == NULL | |
| 4406 || (reglnum + reg_firstlnum != reg_win->w_cursor.lnum) | |
| 4407 || ((colnr_T)(reginput - regline) != reg_win->w_cursor.col)) | |
| 180 | 4408 status = RA_NOMATCH; |
| 7 | 4409 break; |
| 4410 | |
| 639 | 4411 case RE_MARK: |
|
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4412 /* Compare the mark position to the match position. */ |
| 639 | 4413 { |
| 4414 int mark = OPERAND(scan)[0]; | |
| 4415 int cmp = OPERAND(scan)[1]; | |
| 4416 pos_T *pos; | |
| 4417 | |
| 4043 | 4418 pos = getmark_buf(reg_buf, mark, FALSE); |
| 1148 | 4419 if (pos == NULL /* mark doesn't exist */ |
|
4732
0798b096bab3
updated for version 7.3.1113
Bram Moolenaar <bram@vim.org>
parents:
4730
diff
changeset
|
4420 || pos->lnum <= 0 /* mark isn't set in reg_buf */ |
| 639 | 4421 || (pos->lnum == reglnum + reg_firstlnum |
| 4422 ? (pos->col == (colnr_T)(reginput - regline) | |
| 4423 ? (cmp == '<' || cmp == '>') | |
| 4424 : (pos->col < (colnr_T)(reginput - regline) | |
| 4425 ? cmp != '>' | |
| 4426 : cmp != '<')) | |
| 4427 : (pos->lnum < reglnum + reg_firstlnum | |
| 4428 ? cmp != '>' | |
| 4429 : cmp != '<'))) | |
| 4430 status = RA_NOMATCH; | |
| 4431 } | |
| 4432 break; | |
| 4433 | |
| 4434 case RE_VISUAL: | |
|
4730
749e2b2755d5
updated for version 7.3.1112
Bram Moolenaar <bram@vim.org>
parents:
4720
diff
changeset
|
4435 if (!reg_match_visual()) |
| 639 | 4436 status = RA_NOMATCH; |
| 4437 break; | |
| 4438 | |
| 7 | 4439 case RE_LNUM: |
| 4440 if (!REG_MULTI || !re_num_cmp((long_u)(reglnum + reg_firstlnum), | |
| 4441 scan)) | |
| 180 | 4442 status = RA_NOMATCH; |
| 7 | 4443 break; |
| 4444 | |
| 4445 case RE_COL: | |
| 4446 if (!re_num_cmp((long_u)(reginput - regline) + 1, scan)) | |
| 180 | 4447 status = RA_NOMATCH; |
| 7 | 4448 break; |
| 4449 | |
| 4450 case RE_VCOL: | |
| 4451 if (!re_num_cmp((long_u)win_linetabsize( | |
| 4452 reg_win == NULL ? curwin : reg_win, | |
| 4453 regline, (colnr_T)(reginput - regline)) + 1, scan)) | |
| 180 | 4454 status = RA_NOMATCH; |
| 7 | 4455 break; |
| 4456 | |
| 4457 case BOW: /* \<word; reginput points to w */ | |
| 4458 if (c == NUL) /* Can't match at end of line */ | |
| 180 | 4459 status = RA_NOMATCH; |
| 7 | 4460 #ifdef FEAT_MBYTE |
| 180 | 4461 else if (has_mbyte) |
| 7 | 4462 { |
| 4463 int this_class; | |
| 4464 | |
| 4465 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4466 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4467 if (this_class <= 1) |
| 180 | 4468 status = RA_NOMATCH; /* not on a word at all */ |
| 4469 else if (reg_prev_class() == this_class) | |
| 4470 status = RA_NOMATCH; /* previous char is in same word */ | |
| 7 | 4471 } |
| 4472 #endif | |
| 4473 else | |
| 4474 { | |
| 4061 | 4475 if (!vim_iswordc_buf(c, reg_buf) || (reginput > regline |
| 4476 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
| 180 | 4477 status = RA_NOMATCH; |
| 7 | 4478 } |
| 4479 break; | |
| 4480 | |
| 4481 case EOW: /* word\>; reginput points after d */ | |
| 4482 if (reginput == regline) /* Can't match at start of line */ | |
| 180 | 4483 status = RA_NOMATCH; |
| 7 | 4484 #ifdef FEAT_MBYTE |
| 180 | 4485 else if (has_mbyte) |
| 7 | 4486 { |
| 4487 int this_class, prev_class; | |
| 4488 | |
| 4489 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4490 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4491 prev_class = reg_prev_class(); |
| 180 | 4492 if (this_class == prev_class |
| 4493 || prev_class == 0 || prev_class == 1) | |
| 4494 status = RA_NOMATCH; | |
| 7 | 4495 } |
| 180 | 4496 #endif |
| 7 | 4497 else |
| 4498 { | |
| 4043 | 4499 if (!vim_iswordc_buf(reginput[-1], reg_buf) |
| 4500 || (reginput[0] != NUL && vim_iswordc_buf(c, reg_buf))) | |
| 180 | 4501 status = RA_NOMATCH; |
| 7 | 4502 } |
| 4503 break; /* Matched with EOW */ | |
| 4504 | |
| 4505 case ANY: | |
| 4084 | 4506 /* ANY does not match new lines. */ |
| 7 | 4507 if (c == NUL) |
| 180 | 4508 status = RA_NOMATCH; |
| 4509 else | |
| 4510 ADVANCE_REGINPUT(); | |
| 7 | 4511 break; |
| 4512 | |
| 4513 case IDENT: | |
| 4514 if (!vim_isIDc(c)) | |
| 180 | 4515 status = RA_NOMATCH; |
| 4516 else | |
| 4517 ADVANCE_REGINPUT(); | |
| 7 | 4518 break; |
| 4519 | |
| 4520 case SIDENT: | |
| 4521 if (VIM_ISDIGIT(*reginput) || !vim_isIDc(c)) | |
| 180 | 4522 status = RA_NOMATCH; |
| 4523 else | |
| 4524 ADVANCE_REGINPUT(); | |
| 7 | 4525 break; |
| 4526 | |
| 4527 case KWORD: | |
| 4069 | 4528 if (!vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4529 status = RA_NOMATCH; |
| 4530 else | |
| 4531 ADVANCE_REGINPUT(); | |
| 7 | 4532 break; |
| 4533 | |
| 4534 case SKWORD: | |
| 4069 | 4535 if (VIM_ISDIGIT(*reginput) || !vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4536 status = RA_NOMATCH; |
| 4537 else | |
| 4538 ADVANCE_REGINPUT(); | |
| 7 | 4539 break; |
| 4540 | |
| 4541 case FNAME: | |
| 4542 if (!vim_isfilec(c)) | |
| 180 | 4543 status = RA_NOMATCH; |
| 4544 else | |
| 4545 ADVANCE_REGINPUT(); | |
| 7 | 4546 break; |
| 4547 | |
| 4548 case SFNAME: | |
| 4549 if (VIM_ISDIGIT(*reginput) || !vim_isfilec(c)) | |
| 180 | 4550 status = RA_NOMATCH; |
| 4551 else | |
| 4552 ADVANCE_REGINPUT(); | |
| 7 | 4553 break; |
| 4554 | |
| 4555 case PRINT: | |
|
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
4556 if (!vim_isprintc(PTR2CHAR(reginput))) |
| 180 | 4557 status = RA_NOMATCH; |
| 4558 else | |
| 4559 ADVANCE_REGINPUT(); | |
| 7 | 4560 break; |
| 4561 | |
| 4562 case SPRINT: | |
|
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
4563 if (VIM_ISDIGIT(*reginput) || !vim_isprintc(PTR2CHAR(reginput))) |
| 180 | 4564 status = RA_NOMATCH; |
| 4565 else | |
| 4566 ADVANCE_REGINPUT(); | |
| 7 | 4567 break; |
| 4568 | |
| 4569 case WHITE: | |
| 4570 if (!vim_iswhite(c)) | |
| 180 | 4571 status = RA_NOMATCH; |
| 4572 else | |
| 4573 ADVANCE_REGINPUT(); | |
| 7 | 4574 break; |
| 4575 | |
| 4576 case NWHITE: | |
| 4577 if (c == NUL || vim_iswhite(c)) | |
| 180 | 4578 status = RA_NOMATCH; |
| 4579 else | |
| 4580 ADVANCE_REGINPUT(); | |
| 7 | 4581 break; |
| 4582 | |
| 4583 case DIGIT: | |
| 4584 if (!ri_digit(c)) | |
| 180 | 4585 status = RA_NOMATCH; |
| 4586 else | |
| 4587 ADVANCE_REGINPUT(); | |
| 7 | 4588 break; |
| 4589 | |
| 4590 case NDIGIT: | |
| 4591 if (c == NUL || ri_digit(c)) | |
| 180 | 4592 status = RA_NOMATCH; |
| 4593 else | |
| 4594 ADVANCE_REGINPUT(); | |
| 7 | 4595 break; |
| 4596 | |
| 4597 case HEX: | |
| 4598 if (!ri_hex(c)) | |
| 180 | 4599 status = RA_NOMATCH; |
| 4600 else | |
| 4601 ADVANCE_REGINPUT(); | |
| 7 | 4602 break; |
| 4603 | |
| 4604 case NHEX: | |
| 4605 if (c == NUL || ri_hex(c)) | |
| 180 | 4606 status = RA_NOMATCH; |
| 4607 else | |
| 4608 ADVANCE_REGINPUT(); | |
| 7 | 4609 break; |
| 4610 | |
| 4611 case OCTAL: | |
| 4612 if (!ri_octal(c)) | |
| 180 | 4613 status = RA_NOMATCH; |
| 4614 else | |
| 4615 ADVANCE_REGINPUT(); | |
| 7 | 4616 break; |
| 4617 | |
| 4618 case NOCTAL: | |
| 4619 if (c == NUL || ri_octal(c)) | |
| 180 | 4620 status = RA_NOMATCH; |
| 4621 else | |
| 4622 ADVANCE_REGINPUT(); | |
| 7 | 4623 break; |
| 4624 | |
| 4625 case WORD: | |
| 4626 if (!ri_word(c)) | |
| 180 | 4627 status = RA_NOMATCH; |
| 4628 else | |
| 4629 ADVANCE_REGINPUT(); | |
| 7 | 4630 break; |
| 4631 | |
| 4632 case NWORD: | |
| 4633 if (c == NUL || ri_word(c)) | |
| 180 | 4634 status = RA_NOMATCH; |
| 4635 else | |
| 4636 ADVANCE_REGINPUT(); | |
| 7 | 4637 break; |
| 4638 | |
| 4639 case HEAD: | |
| 4640 if (!ri_head(c)) | |
| 180 | 4641 status = RA_NOMATCH; |
| 4642 else | |
| 4643 ADVANCE_REGINPUT(); | |
| 7 | 4644 break; |
| 4645 | |
| 4646 case NHEAD: | |
| 4647 if (c == NUL || ri_head(c)) | |
| 180 | 4648 status = RA_NOMATCH; |
| 4649 else | |
| 4650 ADVANCE_REGINPUT(); | |
| 7 | 4651 break; |
| 4652 | |
| 4653 case ALPHA: | |
| 4654 if (!ri_alpha(c)) | |
| 180 | 4655 status = RA_NOMATCH; |
| 4656 else | |
| 4657 ADVANCE_REGINPUT(); | |
| 7 | 4658 break; |
| 4659 | |
| 4660 case NALPHA: | |
| 4661 if (c == NUL || ri_alpha(c)) | |
| 180 | 4662 status = RA_NOMATCH; |
| 4663 else | |
| 4664 ADVANCE_REGINPUT(); | |
| 7 | 4665 break; |
| 4666 | |
| 4667 case LOWER: | |
| 4668 if (!ri_lower(c)) | |
| 180 | 4669 status = RA_NOMATCH; |
| 4670 else | |
| 4671 ADVANCE_REGINPUT(); | |
| 7 | 4672 break; |
| 4673 | |
| 4674 case NLOWER: | |
| 4675 if (c == NUL || ri_lower(c)) | |
| 180 | 4676 status = RA_NOMATCH; |
| 4677 else | |
| 4678 ADVANCE_REGINPUT(); | |
| 7 | 4679 break; |
| 4680 | |
| 4681 case UPPER: | |
| 4682 if (!ri_upper(c)) | |
| 180 | 4683 status = RA_NOMATCH; |
| 4684 else | |
| 4685 ADVANCE_REGINPUT(); | |
| 7 | 4686 break; |
| 4687 | |
| 4688 case NUPPER: | |
| 4689 if (c == NUL || ri_upper(c)) | |
| 180 | 4690 status = RA_NOMATCH; |
| 4691 else | |
| 4692 ADVANCE_REGINPUT(); | |
| 7 | 4693 break; |
| 4694 | |
| 4695 case EXACTLY: | |
| 4696 { | |
| 4697 int len; | |
| 4698 char_u *opnd; | |
| 4699 | |
| 4700 opnd = OPERAND(scan); | |
| 4701 /* Inline the first byte, for speed. */ | |
| 4702 if (*opnd != *reginput | |
| 4703 && (!ireg_ic || ( | |
| 4704 #ifdef FEAT_MBYTE | |
| 4705 !enc_utf8 && | |
| 4706 #endif | |
| 1347 | 4707 MB_TOLOWER(*opnd) != MB_TOLOWER(*reginput)))) |
| 180 | 4708 status = RA_NOMATCH; |
| 4709 else if (*opnd == NUL) | |
| 7 | 4710 { |
| 4711 /* match empty string always works; happens when "~" is | |
| 4712 * empty. */ | |
| 4713 } | |
| 5899 | 4714 else |
| 4715 { | |
| 4716 if (opnd[1] == NUL | |
| 7 | 4717 #ifdef FEAT_MBYTE |
| 4718 && !(enc_utf8 && ireg_ic) | |
| 4719 #endif | |
| 4720 ) | |
| 5899 | 4721 { |
| 4722 len = 1; /* matched a single byte above */ | |
| 4723 } | |
| 4724 else | |
| 4725 { | |
| 4726 /* Need to match first byte again for multi-byte. */ | |
| 4727 len = (int)STRLEN(opnd); | |
| 4728 if (cstrncmp(opnd, reginput, &len) != 0) | |
| 4729 status = RA_NOMATCH; | |
| 4730 } | |
| 7 | 4731 #ifdef FEAT_MBYTE |
| 5901 | 4732 /* Check for following composing character, unless %C |
| 4733 * follows (skips over all composing chars). */ | |
| 5899 | 4734 if (status != RA_NOMATCH |
| 4735 && enc_utf8 | |
| 4736 && UTF_COMPOSINGLIKE(reginput, reginput + len) | |
| 5901 | 4737 && !ireg_icombine |
| 4738 && OP(next) != RE_COMPOSING) | |
| 7 | 4739 { |
| 4740 /* raaron: This code makes a composing character get | |
| 4741 * ignored, which is the correct behavior (sometimes) | |
| 4742 * for voweled Hebrew texts. */ | |
| 5899 | 4743 status = RA_NOMATCH; |
| 7 | 4744 } |
| 180 | 4745 #endif |
| 5899 | 4746 if (status != RA_NOMATCH) |
| 180 | 4747 reginput += len; |
| 7 | 4748 } |
| 4749 } | |
| 4750 break; | |
| 4751 | |
| 4752 case ANYOF: | |
| 4753 case ANYBUT: | |
| 4754 if (c == NUL) | |
| 180 | 4755 status = RA_NOMATCH; |
| 4756 else if ((cstrchr(OPERAND(scan), c) == NULL) == (op == ANYOF)) | |
| 4757 status = RA_NOMATCH; | |
| 4758 else | |
| 4759 ADVANCE_REGINPUT(); | |
| 7 | 4760 break; |
| 4761 | |
| 4762 #ifdef FEAT_MBYTE | |
| 4763 case MULTIBYTECODE: | |
| 4764 if (has_mbyte) | |
| 4765 { | |
| 4766 int i, len; | |
| 4767 char_u *opnd; | |
| 944 | 4768 int opndc = 0, inpc; |
| 7 | 4769 |
| 4770 opnd = OPERAND(scan); | |
| 4771 /* Safety check (just in case 'encoding' was changed since | |
| 4772 * compiling the program). */ | |
| 474 | 4773 if ((len = (*mb_ptr2len)(opnd)) < 2) |
| 180 | 4774 { |
| 4775 status = RA_NOMATCH; | |
| 4776 break; | |
| 4777 } | |
| 714 | 4778 if (enc_utf8) |
| 4779 opndc = mb_ptr2char(opnd); | |
| 4780 if (enc_utf8 && utf_iscomposing(opndc)) | |
| 4781 { | |
| 4782 /* When only a composing char is given match at any | |
| 4783 * position where that composing char appears. */ | |
| 4784 status = RA_NOMATCH; | |
| 4785 for (i = 0; reginput[i] != NUL; i += utf_char2len(inpc)) | |
| 180 | 4786 { |
| 714 | 4787 inpc = mb_ptr2char(reginput + i); |
| 4788 if (!utf_iscomposing(inpc)) | |
| 4789 { | |
| 4790 if (i > 0) | |
| 4791 break; | |
| 4792 } | |
| 4793 else if (opndc == inpc) | |
| 4794 { | |
| 4795 /* Include all following composing chars. */ | |
| 4796 len = i + mb_ptr2len(reginput + i); | |
| 4797 status = RA_MATCH; | |
| 4798 break; | |
| 4799 } | |
| 180 | 4800 } |
| 714 | 4801 } |
| 4802 else | |
| 4803 for (i = 0; i < len; ++i) | |
| 4804 if (opnd[i] != reginput[i]) | |
| 4805 { | |
| 4806 status = RA_NOMATCH; | |
| 4807 break; | |
| 4808 } | |
| 7 | 4809 reginput += len; |
| 4810 } | |
| 4811 else | |
| 180 | 4812 status = RA_NOMATCH; |
| 7 | 4813 break; |
| 4814 #endif | |
| 5901 | 4815 case RE_COMPOSING: |
| 4816 #ifdef FEAT_MBYTE | |
| 4817 if (enc_utf8) | |
| 4818 { | |
| 4819 /* Skip composing characters. */ | |
| 4820 while (utf_iscomposing(utf_ptr2char(reginput))) | |
| 4821 mb_cptr_adv(reginput); | |
| 4822 } | |
| 4823 #endif | |
| 4824 break; | |
| 7 | 4825 |
| 4826 case NOTHING: | |
| 4827 break; | |
| 4828 | |
| 4829 case BACK: | |
| 233 | 4830 { |
| 4831 int i; | |
| 4832 backpos_T *bp; | |
| 4833 | |
| 4834 /* | |
| 4835 * When we run into BACK we need to check if we don't keep | |
| 4836 * looping without matching any input. The second and later | |
| 4837 * times a BACK is encountered it fails if the input is still | |
| 4838 * at the same position as the previous time. | |
| 4839 * The positions are stored in "backpos" and found by the | |
| 4840 * current value of "scan", the position in the RE program. | |
| 4841 */ | |
| 4842 bp = (backpos_T *)backpos.ga_data; | |
| 4843 for (i = 0; i < backpos.ga_len; ++i) | |
| 4844 if (bp[i].bp_scan == scan) | |
| 4845 break; | |
| 4846 if (i == backpos.ga_len) | |
| 4847 { | |
| 4848 /* First time at this BACK, make room to store the pos. */ | |
| 4849 if (ga_grow(&backpos, 1) == FAIL) | |
| 4850 status = RA_FAIL; | |
| 4851 else | |
| 4852 { | |
| 4853 /* get "ga_data" again, it may have changed */ | |
| 4854 bp = (backpos_T *)backpos.ga_data; | |
| 4855 bp[i].bp_scan = scan; | |
| 4856 ++backpos.ga_len; | |
| 4857 } | |
| 4858 } | |
| 4859 else if (reg_save_equal(&bp[i].bp_pos)) | |
| 4860 /* Still at same position as last time, fail. */ | |
| 4861 status = RA_NOMATCH; | |
| 4862 | |
| 4863 if (status != RA_FAIL && status != RA_NOMATCH) | |
| 4864 reg_save(&bp[i].bp_pos, &backpos); | |
| 4865 } | |
| 179 | 4866 break; |
| 4867 | |
| 7 | 4868 case MOPEN + 0: /* Match start: \zs */ |
| 4869 case MOPEN + 1: /* \( */ | |
| 4870 case MOPEN + 2: | |
| 4871 case MOPEN + 3: | |
| 4872 case MOPEN + 4: | |
| 4873 case MOPEN + 5: | |
| 4874 case MOPEN + 6: | |
| 4875 case MOPEN + 7: | |
| 4876 case MOPEN + 8: | |
| 4877 case MOPEN + 9: | |
| 4878 { | |
| 4879 no = op - MOPEN; | |
| 4880 cleanup_subexpr(); | |
| 270 | 4881 rp = regstack_push(RS_MOPEN, scan); |
| 180 | 4882 if (rp == NULL) |
| 4883 status = RA_FAIL; | |
| 4884 else | |
| 4885 { | |
| 4886 rp->rs_no = no; | |
| 4887 save_se(&rp->rs_un.sesave, ®_startpos[no], | |
| 4888 ®_startp[no]); | |
| 4889 /* We simply continue and handle the result when done. */ | |
| 4890 } | |
| 7 | 4891 } |
| 180 | 4892 break; |
| 7 | 4893 |
| 4894 case NOPEN: /* \%( */ | |
| 4895 case NCLOSE: /* \) after \%( */ | |
| 270 | 4896 if (regstack_push(RS_NOPEN, scan) == NULL) |
| 180 | 4897 status = RA_FAIL; |
| 4898 /* We simply continue and handle the result when done. */ | |
| 4899 break; | |
| 7 | 4900 |
| 4901 #ifdef FEAT_SYN_HL | |
| 4902 case ZOPEN + 1: | |
| 4903 case ZOPEN + 2: | |
| 4904 case ZOPEN + 3: | |
| 4905 case ZOPEN + 4: | |
| 4906 case ZOPEN + 5: | |
| 4907 case ZOPEN + 6: | |
| 4908 case ZOPEN + 7: | |
| 4909 case ZOPEN + 8: | |
| 4910 case ZOPEN + 9: | |
| 4911 { | |
| 4912 no = op - ZOPEN; | |
| 4913 cleanup_zsubexpr(); | |
| 270 | 4914 rp = regstack_push(RS_ZOPEN, scan); |
| 180 | 4915 if (rp == NULL) |
| 4916 status = RA_FAIL; | |
| 4917 else | |
| 4918 { | |
| 4919 rp->rs_no = no; | |
| 4920 save_se(&rp->rs_un.sesave, ®_startzpos[no], | |
| 4921 ®_startzp[no]); | |
| 4922 /* We simply continue and handle the result when done. */ | |
| 4923 } | |
| 7 | 4924 } |
| 180 | 4925 break; |
| 7 | 4926 #endif |
| 4927 | |
| 4928 case MCLOSE + 0: /* Match end: \ze */ | |
| 4929 case MCLOSE + 1: /* \) */ | |
| 4930 case MCLOSE + 2: | |
| 4931 case MCLOSE + 3: | |
| 4932 case MCLOSE + 4: | |
| 4933 case MCLOSE + 5: | |
| 4934 case MCLOSE + 6: | |
| 4935 case MCLOSE + 7: | |
| 4936 case MCLOSE + 8: | |
| 4937 case MCLOSE + 9: | |
| 4938 { | |
| 4939 no = op - MCLOSE; | |
| 4940 cleanup_subexpr(); | |
| 270 | 4941 rp = regstack_push(RS_MCLOSE, scan); |
| 180 | 4942 if (rp == NULL) |
| 4943 status = RA_FAIL; | |
| 4944 else | |
| 4945 { | |
| 4946 rp->rs_no = no; | |
| 4947 save_se(&rp->rs_un.sesave, ®_endpos[no], ®_endp[no]); | |
| 4948 /* We simply continue and handle the result when done. */ | |
| 4949 } | |
| 7 | 4950 } |
| 180 | 4951 break; |
| 7 | 4952 |
| 4953 #ifdef FEAT_SYN_HL | |
| 4954 case ZCLOSE + 1: /* \) after \z( */ | |
| 4955 case ZCLOSE + 2: | |
| 4956 case ZCLOSE + 3: | |
| 4957 case ZCLOSE + 4: | |
| 4958 case ZCLOSE + 5: | |
| 4959 case ZCLOSE + 6: | |
| 4960 case ZCLOSE + 7: | |
| 4961 case ZCLOSE + 8: | |
| 4962 case ZCLOSE + 9: | |
| 4963 { | |
| 4964 no = op - ZCLOSE; | |
| 4965 cleanup_zsubexpr(); | |
| 270 | 4966 rp = regstack_push(RS_ZCLOSE, scan); |
| 180 | 4967 if (rp == NULL) |
| 4968 status = RA_FAIL; | |
| 4969 else | |
| 4970 { | |
| 4971 rp->rs_no = no; | |
| 4972 save_se(&rp->rs_un.sesave, ®_endzpos[no], | |
| 4973 ®_endzp[no]); | |
| 4974 /* We simply continue and handle the result when done. */ | |
| 4975 } | |
| 7 | 4976 } |
| 180 | 4977 break; |
| 7 | 4978 #endif |
| 4979 | |
| 4980 case BACKREF + 1: | |
| 4981 case BACKREF + 2: | |
| 4982 case BACKREF + 3: | |
| 4983 case BACKREF + 4: | |
| 4984 case BACKREF + 5: | |
| 4985 case BACKREF + 6: | |
| 4986 case BACKREF + 7: | |
| 4987 case BACKREF + 8: | |
| 4988 case BACKREF + 9: | |
| 4989 { | |
| 4990 int len; | |
| 4991 | |
| 4992 no = op - BACKREF; | |
| 4993 cleanup_subexpr(); | |
| 4994 if (!REG_MULTI) /* Single-line regexp */ | |
| 4995 { | |
| 1815 | 4996 if (reg_startp[no] == NULL || reg_endp[no] == NULL) |
| 7 | 4997 { |
| 4998 /* Backref was not set: Match an empty string. */ | |
| 4999 len = 0; | |
| 5000 } | |
| 5001 else | |
| 5002 { | |
| 5003 /* Compare current input with back-ref in the same | |
| 5004 * line. */ | |
| 5005 len = (int)(reg_endp[no] - reg_startp[no]); | |
| 5006 if (cstrncmp(reg_startp[no], reginput, &len) != 0) | |
| 180 | 5007 status = RA_NOMATCH; |
| 7 | 5008 } |
| 5009 } | |
| 5010 else /* Multi-line regexp */ | |
| 5011 { | |
| 1815 | 5012 if (reg_startpos[no].lnum < 0 || reg_endpos[no].lnum < 0) |
| 7 | 5013 { |
| 5014 /* Backref was not set: Match an empty string. */ | |
| 5015 len = 0; | |
| 5016 } | |
| 5017 else | |
| 5018 { | |
| 5019 if (reg_startpos[no].lnum == reglnum | |
| 5020 && reg_endpos[no].lnum == reglnum) | |
| 5021 { | |
| 5022 /* Compare back-ref within the current line. */ | |
| 5023 len = reg_endpos[no].col - reg_startpos[no].col; | |
| 5024 if (cstrncmp(regline + reg_startpos[no].col, | |
| 5025 reginput, &len) != 0) | |
| 180 | 5026 status = RA_NOMATCH; |
| 7 | 5027 } |
| 5028 else | |
| 5029 { | |
| 5030 /* Messy situation: Need to compare between two | |
| 5031 * lines. */ | |
|
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5032 int r = match_with_backref( |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5033 reg_startpos[no].lnum, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5034 reg_startpos[no].col, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5035 reg_endpos[no].lnum, |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
5036 reg_endpos[no].col, |
|
4899
4837fd61be52
updated for version 7.3.1195
Bram Moolenaar <bram@vim.org>
parents:
4891
diff
changeset
|
5037 &len); |
|
4901
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5038 |
|
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5039 if (r != RA_MATCH) |
|
56fbf60e9476
updated for version 7.3.1196
Bram Moolenaar <bram@vim.org>
parents:
4899
diff
changeset
|
5040 status = r; |
| 7 | 5041 } |
| 5042 } | |
| 5043 } | |
| 5044 | |
| 5045 /* Matched the backref, skip over it. */ | |
| 5046 reginput += len; | |
| 5047 } | |
| 5048 break; | |
| 5049 | |
| 5050 #ifdef FEAT_SYN_HL | |
| 5051 case ZREF + 1: | |
| 5052 case ZREF + 2: | |
| 5053 case ZREF + 3: | |
| 5054 case ZREF + 4: | |
| 5055 case ZREF + 5: | |
| 5056 case ZREF + 6: | |
| 5057 case ZREF + 7: | |
| 5058 case ZREF + 8: | |
| 5059 case ZREF + 9: | |
| 5060 { | |
| 5061 int len; | |
| 5062 | |
| 5063 cleanup_zsubexpr(); | |
| 5064 no = op - ZREF; | |
| 5065 if (re_extmatch_in != NULL | |
| 5066 && re_extmatch_in->matches[no] != NULL) | |
| 5067 { | |
| 5068 len = (int)STRLEN(re_extmatch_in->matches[no]); | |
| 5069 if (cstrncmp(re_extmatch_in->matches[no], | |
| 5070 reginput, &len) != 0) | |
| 180 | 5071 status = RA_NOMATCH; |
| 5072 else | |
| 5073 reginput += len; | |
| 7 | 5074 } |
| 5075 else | |
| 5076 { | |
| 5077 /* Backref was not set: Match an empty string. */ | |
| 5078 } | |
| 5079 } | |
| 5080 break; | |
| 5081 #endif | |
| 5082 | |
| 5083 case BRANCH: | |
| 5084 { | |
| 5085 if (OP(next) != BRANCH) /* No choice. */ | |
| 5086 next = OPERAND(scan); /* Avoid recursion. */ | |
| 5087 else | |
| 5088 { | |
| 270 | 5089 rp = regstack_push(RS_BRANCH, scan); |
| 180 | 5090 if (rp == NULL) |
| 5091 status = RA_FAIL; | |
| 5092 else | |
| 5093 status = RA_BREAK; /* rest is below */ | |
| 7 | 5094 } |
| 5095 } | |
| 5096 break; | |
| 5097 | |
| 5098 case BRACE_LIMITS: | |
| 5099 { | |
| 5100 if (OP(next) == BRACE_SIMPLE) | |
| 5101 { | |
| 5102 bl_minval = OPERAND_MIN(scan); | |
| 5103 bl_maxval = OPERAND_MAX(scan); | |
| 5104 } | |
| 5105 else if (OP(next) >= BRACE_COMPLEX | |
| 5106 && OP(next) < BRACE_COMPLEX + 10) | |
| 5107 { | |
| 5108 no = OP(next) - BRACE_COMPLEX; | |
| 5109 brace_min[no] = OPERAND_MIN(scan); | |
| 5110 brace_max[no] = OPERAND_MAX(scan); | |
| 5111 brace_count[no] = 0; | |
| 5112 } | |
| 5113 else | |
| 5114 { | |
| 5115 EMSG(_(e_internal)); /* Shouldn't happen */ | |
| 180 | 5116 status = RA_FAIL; |
| 7 | 5117 } |
| 5118 } | |
| 5119 break; | |
| 5120 | |
| 5121 case BRACE_COMPLEX + 0: | |
| 5122 case BRACE_COMPLEX + 1: | |
| 5123 case BRACE_COMPLEX + 2: | |
| 5124 case BRACE_COMPLEX + 3: | |
| 5125 case BRACE_COMPLEX + 4: | |
| 5126 case BRACE_COMPLEX + 5: | |
| 5127 case BRACE_COMPLEX + 6: | |
| 5128 case BRACE_COMPLEX + 7: | |
| 5129 case BRACE_COMPLEX + 8: | |
| 5130 case BRACE_COMPLEX + 9: | |
| 5131 { | |
| 5132 no = op - BRACE_COMPLEX; | |
| 5133 ++brace_count[no]; | |
| 5134 | |
| 5135 /* If not matched enough times yet, try one more */ | |
| 5136 if (brace_count[no] <= (brace_min[no] <= brace_max[no] | |
| 180 | 5137 ? brace_min[no] : brace_max[no])) |
| 7 | 5138 { |
| 270 | 5139 rp = regstack_push(RS_BRCPLX_MORE, scan); |
| 180 | 5140 if (rp == NULL) |
| 5141 status = RA_FAIL; | |
| 5142 else | |
| 5143 { | |
| 5144 rp->rs_no = no; | |
| 233 | 5145 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5146 next = OPERAND(scan); |
| 5147 /* We continue and handle the result when done. */ | |
| 5148 } | |
| 5149 break; | |
| 7 | 5150 } |
| 5151 | |
| 5152 /* If matched enough times, may try matching some more */ | |
| 5153 if (brace_min[no] <= brace_max[no]) | |
| 5154 { | |
| 5155 /* Range is the normal way around, use longest match */ | |
| 5156 if (brace_count[no] <= brace_max[no]) | |
| 5157 { | |
| 270 | 5158 rp = regstack_push(RS_BRCPLX_LONG, scan); |
| 180 | 5159 if (rp == NULL) |
| 5160 status = RA_FAIL; | |
| 5161 else | |
| 5162 { | |
| 5163 rp->rs_no = no; | |
| 233 | 5164 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5165 next = OPERAND(scan); |
| 5166 /* We continue and handle the result when done. */ | |
| 5167 } | |
| 7 | 5168 } |
| 5169 } | |
| 5170 else | |
| 5171 { | |
| 5172 /* Range is backwards, use shortest match first */ | |
| 5173 if (brace_count[no] <= brace_min[no]) | |
| 5174 { | |
| 270 | 5175 rp = regstack_push(RS_BRCPLX_SHORT, scan); |
| 180 | 5176 if (rp == NULL) |
| 5177 status = RA_FAIL; | |
| 5178 else | |
| 5179 { | |
| 233 | 5180 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5181 /* We continue and handle the result when done. */ |
| 5182 } | |
| 7 | 5183 } |
| 5184 } | |
| 5185 } | |
| 5186 break; | |
| 5187 | |
| 5188 case BRACE_SIMPLE: | |
| 5189 case STAR: | |
| 5190 case PLUS: | |
| 5191 { | |
| 180 | 5192 regstar_T rst; |
| 7 | 5193 |
| 5194 /* | |
| 5195 * Lookahead to avoid useless match attempts when we know | |
| 5196 * what character comes next. | |
| 5197 */ | |
| 5198 if (OP(next) == EXACTLY) | |
| 5199 { | |
| 180 | 5200 rst.nextb = *OPERAND(next); |
| 7 | 5201 if (ireg_ic) |
| 5202 { | |
| 1347 | 5203 if (MB_ISUPPER(rst.nextb)) |
| 5204 rst.nextb_ic = MB_TOLOWER(rst.nextb); | |
| 7 | 5205 else |
| 1347 | 5206 rst.nextb_ic = MB_TOUPPER(rst.nextb); |
| 7 | 5207 } |
| 5208 else | |
| 180 | 5209 rst.nextb_ic = rst.nextb; |
| 7 | 5210 } |
| 5211 else | |
| 5212 { | |
| 180 | 5213 rst.nextb = NUL; |
| 5214 rst.nextb_ic = NUL; | |
| 7 | 5215 } |
| 5216 if (op != BRACE_SIMPLE) | |
| 5217 { | |
| 180 | 5218 rst.minval = (op == STAR) ? 0 : 1; |
| 5219 rst.maxval = MAX_LIMIT; | |
| 7 | 5220 } |
| 5221 else | |
| 5222 { | |
| 180 | 5223 rst.minval = bl_minval; |
| 5224 rst.maxval = bl_maxval; | |
| 7 | 5225 } |
| 5226 | |
| 5227 /* | |
| 5228 * When maxval > minval, try matching as much as possible, up | |
| 5229 * to maxval. When maxval < minval, try matching at least the | |
| 5230 * minimal number (since the range is backwards, that's also | |
| 5231 * maxval!). | |
| 5232 */ | |
| 180 | 5233 rst.count = regrepeat(OPERAND(scan), rst.maxval); |
| 7 | 5234 if (got_int) |
| 180 | 5235 { |
| 5236 status = RA_FAIL; | |
| 5237 break; | |
| 5238 } | |
| 5239 if (rst.minval <= rst.maxval | |
| 5240 ? rst.count >= rst.minval : rst.count >= rst.maxval) | |
| 7 | 5241 { |
| 180 | 5242 /* It could match. Prepare for trying to match what |
| 5243 * follows. The code is below. Parameters are stored in | |
| 5244 * a regstar_T on the regstack. */ | |
| 212 | 5245 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5246 { |
| 5247 EMSG(_(e_maxmempat)); | |
| 5248 status = RA_FAIL; | |
| 5249 } | |
| 5250 else if (ga_grow(®stack, sizeof(regstar_T)) == FAIL) | |
| 180 | 5251 status = RA_FAIL; |
| 5252 else | |
| 7 | 5253 { |
| 180 | 5254 regstack.ga_len += sizeof(regstar_T); |
| 270 | 5255 rp = regstack_push(rst.minval <= rst.maxval |
| 233 | 5256 ? RS_STAR_LONG : RS_STAR_SHORT, scan); |
| 180 | 5257 if (rp == NULL) |
| 5258 status = RA_FAIL; | |
| 5259 else | |
| 7 | 5260 { |
| 180 | 5261 *(((regstar_T *)rp) - 1) = rst; |
| 5262 status = RA_BREAK; /* skip the restore bits */ | |
| 7 | 5263 } |
| 5264 } | |
| 5265 } | |
| 5266 else | |
| 180 | 5267 status = RA_NOMATCH; |
| 5268 | |
| 7 | 5269 } |
| 180 | 5270 break; |
| 7 | 5271 |
| 5272 case NOMATCH: | |
| 5273 case MATCH: | |
| 5274 case SUBPAT: | |
| 270 | 5275 rp = regstack_push(RS_NOMATCH, scan); |
| 180 | 5276 if (rp == NULL) |
| 5277 status = RA_FAIL; | |
| 5278 else | |
| 7 | 5279 { |
| 180 | 5280 rp->rs_no = op; |
| 233 | 5281 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5282 next = OPERAND(scan); |
| 5283 /* We continue and handle the result when done. */ | |
| 7 | 5284 } |
| 5285 break; | |
| 5286 | |
| 5287 case BEHIND: | |
| 5288 case NOBEHIND: | |
| 180 | 5289 /* Need a bit of room to store extra positions. */ |
| 212 | 5290 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5291 { |
| 5292 EMSG(_(e_maxmempat)); | |
| 5293 status = RA_FAIL; | |
| 5294 } | |
| 5295 else if (ga_grow(®stack, sizeof(regbehind_T)) == FAIL) | |
| 180 | 5296 status = RA_FAIL; |
| 5297 else | |
| 7 | 5298 { |
| 180 | 5299 regstack.ga_len += sizeof(regbehind_T); |
| 270 | 5300 rp = regstack_push(RS_BEHIND1, scan); |
| 180 | 5301 if (rp == NULL) |
| 5302 status = RA_FAIL; | |
| 5303 else | |
| 7 | 5304 { |
| 1579 | 5305 /* Need to save the subexpr to be able to restore them |
| 5306 * when there is a match but we don't use it. */ | |
| 5307 save_subexpr(((regbehind_T *)rp) - 1); | |
| 5308 | |
| 180 | 5309 rp->rs_no = op; |
| 233 | 5310 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5311 /* First try if what follows matches. If it does then we |
| 5312 * check the behind match by looping. */ | |
| 7 | 5313 } |
| 5314 } | |
| 180 | 5315 break; |
| 7 | 5316 |
| 5317 case BHPOS: | |
| 5318 if (REG_MULTI) | |
| 5319 { | |
| 5320 if (behind_pos.rs_u.pos.col != (colnr_T)(reginput - regline) | |
| 5321 || behind_pos.rs_u.pos.lnum != reglnum) | |
| 180 | 5322 status = RA_NOMATCH; |
| 7 | 5323 } |
| 5324 else if (behind_pos.rs_u.ptr != reginput) | |
| 180 | 5325 status = RA_NOMATCH; |
| 7 | 5326 break; |
| 5327 | |
| 5328 case NEWL: | |
| 1018 | 5329 if ((c != NUL || !REG_MULTI || reglnum > reg_maxline |
| 5330 || reg_line_lbr) && (c != '\n' || !reg_line_lbr)) | |
| 180 | 5331 status = RA_NOMATCH; |
| 5332 else if (reg_line_lbr) | |
| 7 | 5333 ADVANCE_REGINPUT(); |
| 5334 else | |
| 5335 reg_nextline(); | |
| 5336 break; | |
| 5337 | |
| 5338 case END: | |
| 180 | 5339 status = RA_MATCH; /* Success! */ |
| 5340 break; | |
| 7 | 5341 |
| 5342 default: | |
| 5343 EMSG(_(e_re_corr)); | |
| 5344 #ifdef DEBUG | |
| 5345 printf("Illegal op code %d\n", op); | |
| 5346 #endif | |
| 180 | 5347 status = RA_FAIL; |
| 5348 break; | |
| 7 | 5349 } |
| 5350 } | |
| 5351 | |
| 180 | 5352 /* If we can't continue sequentially, break the inner loop. */ |
| 5353 if (status != RA_CONT) | |
| 5354 break; | |
| 5355 | |
| 5356 /* Continue in inner loop, advance to next item. */ | |
| 7 | 5357 scan = next; |
| 180 | 5358 |
| 5359 } /* end of inner loop */ | |
| 7 | 5360 |
| 5361 /* | |
| 180 | 5362 * If there is something on the regstack execute the code for the state. |
| 233 | 5363 * If the state is popped then loop and use the older state. |
| 7 | 5364 */ |
| 180 | 5365 while (regstack.ga_len > 0 && status != RA_FAIL) |
| 5366 { | |
| 5367 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; | |
| 5368 switch (rp->rs_state) | |
| 5369 { | |
| 5370 case RS_NOPEN: | |
| 5371 /* Result is passed on as-is, simply pop the state. */ | |
| 270 | 5372 regstack_pop(&scan); |
| 180 | 5373 break; |
| 5374 | |
| 5375 case RS_MOPEN: | |
| 5376 /* Pop the state. Restore pointers when there is no match. */ | |
| 5377 if (status == RA_NOMATCH) | |
| 5378 restore_se(&rp->rs_un.sesave, ®_startpos[rp->rs_no], | |
| 5379 ®_startp[rp->rs_no]); | |
| 270 | 5380 regstack_pop(&scan); |
| 180 | 5381 break; |
| 5382 | |
| 5383 #ifdef FEAT_SYN_HL | |
| 5384 case RS_ZOPEN: | |
| 5385 /* Pop the state. Restore pointers when there is no match. */ | |
| 5386 if (status == RA_NOMATCH) | |
| 5387 restore_se(&rp->rs_un.sesave, ®_startzpos[rp->rs_no], | |
| 5388 ®_startzp[rp->rs_no]); | |
| 270 | 5389 regstack_pop(&scan); |
| 180 | 5390 break; |
| 5391 #endif | |
| 5392 | |
| 5393 case RS_MCLOSE: | |
| 5394 /* Pop the state. Restore pointers when there is no match. */ | |
| 5395 if (status == RA_NOMATCH) | |
| 5396 restore_se(&rp->rs_un.sesave, ®_endpos[rp->rs_no], | |
| 5397 ®_endp[rp->rs_no]); | |
| 270 | 5398 regstack_pop(&scan); |
| 180 | 5399 break; |
| 5400 | |
| 5401 #ifdef FEAT_SYN_HL | |
| 5402 case RS_ZCLOSE: | |
| 5403 /* Pop the state. Restore pointers when there is no match. */ | |
| 5404 if (status == RA_NOMATCH) | |
| 5405 restore_se(&rp->rs_un.sesave, ®_endzpos[rp->rs_no], | |
| 5406 ®_endzp[rp->rs_no]); | |
| 270 | 5407 regstack_pop(&scan); |
| 180 | 5408 break; |
| 5409 #endif | |
| 5410 | |
| 5411 case RS_BRANCH: | |
| 5412 if (status == RA_MATCH) | |
| 5413 /* this branch matched, use it */ | |
| 270 | 5414 regstack_pop(&scan); |
| 180 | 5415 else |
| 5416 { | |
| 5417 if (status != RA_BREAK) | |
| 5418 { | |
| 5419 /* After a non-matching branch: try next one. */ | |
| 233 | 5420 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5421 scan = rp->rs_scan; |
| 5422 } | |
| 5423 if (scan == NULL || OP(scan) != BRANCH) | |
| 5424 { | |
| 5425 /* no more branches, didn't find a match */ | |
| 5426 status = RA_NOMATCH; | |
| 270 | 5427 regstack_pop(&scan); |
| 180 | 5428 } |
| 5429 else | |
| 5430 { | |
| 5431 /* Prepare to try a branch. */ | |
| 5432 rp->rs_scan = regnext(scan); | |
| 233 | 5433 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5434 scan = OPERAND(scan); |
| 5435 } | |
| 5436 } | |
| 5437 break; | |
| 5438 | |
| 5439 case RS_BRCPLX_MORE: | |
| 5440 /* Pop the state. Restore pointers when there is no match. */ | |
| 5441 if (status == RA_NOMATCH) | |
| 5442 { | |
| 233 | 5443 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5444 --brace_count[rp->rs_no]; /* decrement match count */ |
| 5445 } | |
| 270 | 5446 regstack_pop(&scan); |
| 180 | 5447 break; |
| 5448 | |
| 5449 case RS_BRCPLX_LONG: | |
| 5450 /* Pop the state. Restore pointers when there is no match. */ | |
| 5451 if (status == RA_NOMATCH) | |
| 5452 { | |
| 5453 /* There was no match, but we did find enough matches. */ | |
| 233 | 5454 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5455 --brace_count[rp->rs_no]; |
| 5456 /* continue with the items after "\{}" */ | |
| 5457 status = RA_CONT; | |
| 5458 } | |
| 270 | 5459 regstack_pop(&scan); |
| 180 | 5460 if (status == RA_CONT) |
| 5461 scan = regnext(scan); | |
| 5462 break; | |
| 5463 | |
| 5464 case RS_BRCPLX_SHORT: | |
| 5465 /* Pop the state. Restore pointers when there is no match. */ | |
| 5466 if (status == RA_NOMATCH) | |
| 5467 /* There was no match, try to match one more item. */ | |
| 233 | 5468 reg_restore(&rp->rs_un.regsave, &backpos); |
| 270 | 5469 regstack_pop(&scan); |
| 180 | 5470 if (status == RA_NOMATCH) |
| 5471 { | |
| 5472 scan = OPERAND(scan); | |
| 5473 status = RA_CONT; | |
| 5474 } | |
| 5475 break; | |
| 5476 | |
| 5477 case RS_NOMATCH: | |
| 5478 /* Pop the state. If the operand matches for NOMATCH or | |
| 5479 * doesn't match for MATCH/SUBPAT, we fail. Otherwise backup, | |
| 5480 * except for SUBPAT, and continue with the next item. */ | |
| 5481 if (status == (rp->rs_no == NOMATCH ? RA_MATCH : RA_NOMATCH)) | |
| 5482 status = RA_NOMATCH; | |
| 5483 else | |
| 5484 { | |
| 5485 status = RA_CONT; | |
| 233 | 5486 if (rp->rs_no != SUBPAT) /* zero-width */ |
| 5487 reg_restore(&rp->rs_un.regsave, &backpos); | |
| 180 | 5488 } |
| 270 | 5489 regstack_pop(&scan); |
| 180 | 5490 if (status == RA_CONT) |
| 5491 scan = regnext(scan); | |
| 5492 break; | |
| 5493 | |
| 5494 case RS_BEHIND1: | |
| 5495 if (status == RA_NOMATCH) | |
| 5496 { | |
| 270 | 5497 regstack_pop(&scan); |
| 180 | 5498 regstack.ga_len -= sizeof(regbehind_T); |
| 5499 } | |
| 5500 else | |
| 5501 { | |
| 5502 /* The stuff after BEHIND/NOBEHIND matches. Now try if | |
| 5503 * the behind part does (not) match before the current | |
| 5504 * position in the input. This must be done at every | |
| 5505 * position in the input and checking if the match ends at | |
| 5506 * the current position. */ | |
| 5507 | |
| 5508 /* save the position after the found match for next */ | |
| 233 | 5509 reg_save(&(((regbehind_T *)rp) - 1)->save_after, &backpos); |
| 180 | 5510 |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5511 /* Start looking for a match with operand at the current |
| 1209 | 5512 * position. Go back one character until we find the |
| 180 | 5513 * result, hitting the start of the line or the previous |
| 5514 * line (for multi-line matching). | |
| 5515 * Set behind_pos to where the match should end, BHPOS | |
| 5516 * will match it. Save the current value. */ | |
| 5517 (((regbehind_T *)rp) - 1)->save_behind = behind_pos; | |
| 5518 behind_pos = rp->rs_un.regsave; | |
| 5519 | |
| 5520 rp->rs_state = RS_BEHIND2; | |
| 5521 | |
| 233 | 5522 reg_restore(&rp->rs_un.regsave, &backpos); |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5523 scan = OPERAND(rp->rs_scan) + 4; |
| 180 | 5524 } |
| 5525 break; | |
| 5526 | |
| 5527 case RS_BEHIND2: | |
| 5528 /* | |
| 5529 * Looping for BEHIND / NOBEHIND match. | |
| 5530 */ | |
| 5531 if (status == RA_MATCH && reg_save_equal(&behind_pos)) | |
| 5532 { | |
| 5533 /* found a match that ends where "next" started */ | |
| 5534 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5535 if (rp->rs_no == BEHIND) | |
| 233 | 5536 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5537 &backpos); | |
| 180 | 5538 else |
| 1579 | 5539 { |
| 5540 /* But we didn't want a match. Need to restore the | |
| 5541 * subexpr, because what follows matched, so they have | |
| 5542 * been set. */ | |
| 180 | 5543 status = RA_NOMATCH; |
| 1579 | 5544 restore_subexpr(((regbehind_T *)rp) - 1); |
| 5545 } | |
| 270 | 5546 regstack_pop(&scan); |
| 180 | 5547 regstack.ga_len -= sizeof(regbehind_T); |
| 5548 } | |
| 5549 else | |
| 5550 { | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5551 long limit; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5552 |
| 1579 | 5553 /* No match or a match that doesn't end where we want it: Go |
| 5554 * back one character. May go to previous line once. */ | |
| 180 | 5555 no = OK; |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5556 limit = OPERAND_MIN(rp->rs_scan); |
| 180 | 5557 if (REG_MULTI) |
| 5558 { | |
|
4682
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5559 if (limit > 0 |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5560 && ((rp->rs_un.regsave.rs_u.pos.lnum |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5561 < behind_pos.rs_u.pos.lnum |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5562 ? (colnr_T)STRLEN(regline) |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5563 : behind_pos.rs_u.pos.col) |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5564 - 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
|
5565 no = FAIL; |
|
2f51ee8825db
updated for version 7.3.1088
Bram Moolenaar <bram@vim.org>
parents:
4679
diff
changeset
|
5566 else if (rp->rs_un.regsave.rs_u.pos.col == 0) |
| 180 | 5567 { |
| 5568 if (rp->rs_un.regsave.rs_u.pos.lnum | |
| 5569 < behind_pos.rs_u.pos.lnum | |
| 5570 || reg_getline( | |
| 5571 --rp->rs_un.regsave.rs_u.pos.lnum) | |
| 5572 == NULL) | |
| 5573 no = FAIL; | |
| 5574 else | |
| 5575 { | |
| 233 | 5576 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5577 rp->rs_un.regsave.rs_u.pos.col = |
| 5578 (colnr_T)STRLEN(regline); | |
| 5579 } | |
| 5580 } | |
| 5581 else | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5582 { |
| 4176 | 5583 #ifdef FEAT_MBYTE |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5584 if (has_mbyte) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5585 rp->rs_un.regsave.rs_u.pos.col -= |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5586 (*mb_head_off)(regline, regline |
| 4176 | 5587 + 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
|
5588 else |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5589 #endif |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5590 --rp->rs_un.regsave.rs_u.pos.col; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5591 } |
| 180 | 5592 } |
| 5593 else | |
| 5594 { | |
| 5595 if (rp->rs_un.regsave.rs_u.ptr == regline) | |
| 5596 no = FAIL; | |
| 5597 else | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5598 { |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5599 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
|
5600 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
|
5601 - rp->rs_un.regsave.rs_u.ptr) > limit) |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5602 no = FAIL; |
|
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5603 } |
| 180 | 5604 } |
| 5605 if (no == OK) | |
| 5606 { | |
| 5607 /* Advanced, prepare for finding match again. */ | |
| 233 | 5608 reg_restore(&rp->rs_un.regsave, &backpos); |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
5609 scan = OPERAND(rp->rs_scan) + 4; |
| 1579 | 5610 if (status == RA_MATCH) |
| 5611 { | |
| 5612 /* We did match, so subexpr may have been changed, | |
| 5613 * need to restore them for the next try. */ | |
| 5614 status = RA_NOMATCH; | |
| 5615 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5616 } | |
| 180 | 5617 } |
| 5618 else | |
| 5619 { | |
| 5620 /* Can't advance. For NOBEHIND that's a match. */ | |
| 5621 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5622 if (rp->rs_no == NOBEHIND) | |
| 5623 { | |
| 233 | 5624 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5625 &backpos); | |
| 180 | 5626 status = RA_MATCH; |
| 5627 } | |
| 5628 else | |
| 1579 | 5629 { |
| 5630 /* We do want a proper match. Need to restore the | |
| 5631 * subexpr if we had a match, because they may have | |
| 5632 * been set. */ | |
| 5633 if (status == RA_MATCH) | |
| 5634 { | |
| 5635 status = RA_NOMATCH; | |
| 5636 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5637 } | |
| 5638 } | |
| 270 | 5639 regstack_pop(&scan); |
| 180 | 5640 regstack.ga_len -= sizeof(regbehind_T); |
| 5641 } | |
| 5642 } | |
| 5643 break; | |
| 5644 | |
| 5645 case RS_STAR_LONG: | |
| 5646 case RS_STAR_SHORT: | |
| 5647 { | |
| 5648 regstar_T *rst = ((regstar_T *)rp) - 1; | |
| 5649 | |
| 5650 if (status == RA_MATCH) | |
| 5651 { | |
| 270 | 5652 regstack_pop(&scan); |
| 180 | 5653 regstack.ga_len -= sizeof(regstar_T); |
| 5654 break; | |
| 5655 } | |
| 5656 | |
| 5657 /* Tried once already, restore input pointers. */ | |
| 5658 if (status != RA_BREAK) | |
| 233 | 5659 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5660 |
| 5661 /* Repeat until we found a position where it could match. */ | |
| 5662 for (;;) | |
| 5663 { | |
| 5664 if (status != RA_BREAK) | |
| 5665 { | |
| 5666 /* Tried first position already, advance. */ | |
| 5667 if (rp->rs_state == RS_STAR_LONG) | |
| 5668 { | |
| 685 | 5669 /* Trying for longest match, but couldn't or |
| 5670 * didn't match -- back up one char. */ | |
| 180 | 5671 if (--rst->count < rst->minval) |
| 5672 break; | |
| 5673 if (reginput == regline) | |
| 5674 { | |
| 5675 /* backup to last char of previous line */ | |
| 5676 --reglnum; | |
| 5677 regline = reg_getline(reglnum); | |
| 5678 /* Just in case regrepeat() didn't count | |
| 5679 * right. */ | |
| 5680 if (regline == NULL) | |
| 5681 break; | |
| 5682 reginput = regline + STRLEN(regline); | |
| 5683 fast_breakcheck(); | |
| 5684 } | |
| 5685 else | |
| 5686 mb_ptr_back(regline, reginput); | |
| 5687 } | |
| 5688 else | |
| 5689 { | |
| 5690 /* Range is backwards, use shortest match first. | |
| 5691 * Careful: maxval and minval are exchanged! | |
| 5692 * Couldn't or didn't match: try advancing one | |
| 5693 * char. */ | |
| 5694 if (rst->count == rst->minval | |
| 5695 || regrepeat(OPERAND(rp->rs_scan), 1L) == 0) | |
| 5696 break; | |
| 5697 ++rst->count; | |
| 5698 } | |
| 5699 if (got_int) | |
| 5700 break; | |
| 5701 } | |
| 5702 else | |
| 5703 status = RA_NOMATCH; | |
| 5704 | |
| 5705 /* If it could match, try it. */ | |
| 5706 if (rst->nextb == NUL || *reginput == rst->nextb | |
| 5707 || *reginput == rst->nextb_ic) | |
| 5708 { | |
| 233 | 5709 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5710 scan = regnext(rp->rs_scan); |
| 5711 status = RA_CONT; | |
| 5712 break; | |
| 5713 } | |
| 5714 } | |
| 5715 if (status != RA_CONT) | |
| 5716 { | |
| 5717 /* Failed. */ | |
| 270 | 5718 regstack_pop(&scan); |
| 180 | 5719 regstack.ga_len -= sizeof(regstar_T); |
| 5720 status = RA_NOMATCH; | |
| 5721 } | |
| 5722 } | |
| 5723 break; | |
| 5724 } | |
| 5725 | |
| 685 | 5726 /* If we want to continue the inner loop or didn't pop a state |
| 5727 * continue matching loop */ | |
| 180 | 5728 if (status == RA_CONT || rp == (regitem_T *) |
| 5729 ((char *)regstack.ga_data + regstack.ga_len) - 1) | |
| 5730 break; | |
| 5731 } | |
| 5732 | |
| 189 | 5733 /* May need to continue with the inner loop, starting at "scan". */ |
| 180 | 5734 if (status == RA_CONT) |
| 5735 continue; | |
| 5736 | |
| 5737 /* | |
| 5738 * If the regstack is empty or something failed we are done. | |
| 5739 */ | |
| 5740 if (regstack.ga_len == 0 || status == RA_FAIL) | |
| 5741 { | |
| 5742 if (scan == NULL) | |
| 5743 { | |
| 5744 /* | |
| 5745 * We get here only if there's trouble -- normally "case END" is | |
| 5746 * the terminating point. | |
| 5747 */ | |
| 5748 EMSG(_(e_re_corr)); | |
| 7 | 5749 #ifdef DEBUG |
| 180 | 5750 printf("Premature EOL\n"); |
| 7 | 5751 #endif |
| 180 | 5752 } |
| 189 | 5753 if (status == RA_FAIL) |
| 5754 got_int = TRUE; | |
| 180 | 5755 return (status == RA_MATCH); |
| 5756 } | |
| 5757 | |
| 5758 } /* End of loop until the regstack is empty. */ | |
| 5759 | |
| 5760 /* NOTREACHED */ | |
| 5761 } | |
| 5762 | |
| 5763 /* | |
| 5764 * Push an item onto the regstack. | |
| 5765 * Returns pointer to new item. Returns NULL when out of memory. | |
| 5766 */ | |
| 5767 static regitem_T * | |
| 270 | 5768 regstack_push(state, scan) |
| 180 | 5769 regstate_T state; |
| 5770 char_u *scan; | |
| 5771 { | |
| 5772 regitem_T *rp; | |
| 5773 | |
| 270 | 5774 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5775 { |
| 5776 EMSG(_(e_maxmempat)); | |
| 5777 return NULL; | |
| 5778 } | |
| 270 | 5779 if (ga_grow(®stack, sizeof(regitem_T)) == FAIL) |
| 180 | 5780 return NULL; |
| 5781 | |
| 270 | 5782 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len); |
| 180 | 5783 rp->rs_state = state; |
| 5784 rp->rs_scan = scan; | |
| 5785 | |
| 270 | 5786 regstack.ga_len += sizeof(regitem_T); |
| 180 | 5787 return rp; |
| 5788 } | |
| 5789 | |
| 5790 /* | |
| 5791 * Pop an item from the regstack. | |
| 5792 */ | |
| 5793 static void | |
| 270 | 5794 regstack_pop(scan) |
| 180 | 5795 char_u **scan; |
| 5796 { | |
| 5797 regitem_T *rp; | |
| 5798 | |
| 270 | 5799 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; |
| 180 | 5800 *scan = rp->rs_scan; |
| 5801 | |
| 270 | 5802 regstack.ga_len -= sizeof(regitem_T); |
| 7 | 5803 } |
| 5804 | |
| 5805 /* | |
| 5806 * regrepeat - repeatedly match something simple, return how many. | |
| 5807 * Advances reginput (and reglnum) to just after the matched chars. | |
| 5808 */ | |
| 5809 static int | |
| 5810 regrepeat(p, maxcount) | |
| 5811 char_u *p; | |
| 5812 long maxcount; /* maximum number of matches allowed */ | |
| 5813 { | |
| 5814 long count = 0; | |
| 5815 char_u *scan; | |
| 5816 char_u *opnd; | |
| 5817 int mask; | |
| 5818 int testval = 0; | |
| 5819 | |
| 5820 scan = reginput; /* Make local copy of reginput for speed. */ | |
| 5821 opnd = OPERAND(p); | |
| 5822 switch (OP(p)) | |
| 5823 { | |
| 5824 case ANY: | |
| 5825 case ANY + ADD_NL: | |
| 5826 while (count < maxcount) | |
| 5827 { | |
| 5828 /* Matching anything means we continue until end-of-line (or | |
| 5829 * end-of-file for ANY + ADD_NL), only limited by maxcount. */ | |
| 5830 while (*scan != NUL && count < maxcount) | |
| 5831 { | |
| 5832 ++count; | |
| 39 | 5833 mb_ptr_adv(scan); |
| 7 | 5834 } |
| 1018 | 5835 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5836 || reg_line_lbr || count == maxcount) | |
| 7 | 5837 break; |
| 5838 ++count; /* count the line-break */ | |
| 5839 reg_nextline(); | |
| 5840 scan = reginput; | |
| 5841 if (got_int) | |
| 5842 break; | |
| 5843 } | |
| 5844 break; | |
| 5845 | |
| 5846 case IDENT: | |
| 5847 case IDENT + ADD_NL: | |
| 5848 testval = TRUE; | |
| 5849 /*FALLTHROUGH*/ | |
| 5850 case SIDENT: | |
| 5851 case SIDENT + ADD_NL: | |
| 5852 while (count < maxcount) | |
| 5853 { | |
| 4466 | 5854 if (vim_isIDc(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5855 { |
| 39 | 5856 mb_ptr_adv(scan); |
| 7 | 5857 } |
| 5858 else if (*scan == NUL) | |
| 5859 { | |
| 1018 | 5860 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5861 || reg_line_lbr) | |
| 7 | 5862 break; |
| 5863 reg_nextline(); | |
| 5864 scan = reginput; | |
| 5865 if (got_int) | |
| 5866 break; | |
| 5867 } | |
| 5868 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5869 ++scan; | |
| 5870 else | |
| 5871 break; | |
| 5872 ++count; | |
| 5873 } | |
| 5874 break; | |
| 5875 | |
| 5876 case KWORD: | |
| 5877 case KWORD + ADD_NL: | |
| 5878 testval = TRUE; | |
| 5879 /*FALLTHROUGH*/ | |
| 5880 case SKWORD: | |
| 5881 case SKWORD + ADD_NL: | |
| 5882 while (count < maxcount) | |
| 5883 { | |
| 4069 | 5884 if (vim_iswordp_buf(scan, reg_buf) |
| 5885 && (testval || !VIM_ISDIGIT(*scan))) | |
| 7 | 5886 { |
| 39 | 5887 mb_ptr_adv(scan); |
| 7 | 5888 } |
| 5889 else if (*scan == NUL) | |
| 5890 { | |
| 1018 | 5891 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5892 || reg_line_lbr) | |
| 7 | 5893 break; |
| 5894 reg_nextline(); | |
| 5895 scan = reginput; | |
| 5896 if (got_int) | |
| 5897 break; | |
| 5898 } | |
| 5899 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5900 ++scan; | |
| 5901 else | |
| 5902 break; | |
| 5903 ++count; | |
| 5904 } | |
| 5905 break; | |
| 5906 | |
| 5907 case FNAME: | |
| 5908 case FNAME + ADD_NL: | |
| 5909 testval = TRUE; | |
| 5910 /*FALLTHROUGH*/ | |
| 5911 case SFNAME: | |
| 5912 case SFNAME + ADD_NL: | |
| 5913 while (count < maxcount) | |
| 5914 { | |
| 4466 | 5915 if (vim_isfilec(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5916 { |
| 39 | 5917 mb_ptr_adv(scan); |
| 7 | 5918 } |
| 5919 else if (*scan == NUL) | |
| 5920 { | |
| 1018 | 5921 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5922 || reg_line_lbr) | |
| 7 | 5923 break; |
| 5924 reg_nextline(); | |
| 5925 scan = reginput; | |
| 5926 if (got_int) | |
| 5927 break; | |
| 5928 } | |
| 5929 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5930 ++scan; | |
| 5931 else | |
| 5932 break; | |
| 5933 ++count; | |
| 5934 } | |
| 5935 break; | |
| 5936 | |
| 5937 case PRINT: | |
| 5938 case PRINT + ADD_NL: | |
| 5939 testval = TRUE; | |
| 5940 /*FALLTHROUGH*/ | |
| 5941 case SPRINT: | |
| 5942 case SPRINT + ADD_NL: | |
| 5943 while (count < maxcount) | |
| 5944 { | |
| 5945 if (*scan == NUL) | |
| 5946 { | |
| 1018 | 5947 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5948 || reg_line_lbr) | |
| 7 | 5949 break; |
| 5950 reg_nextline(); | |
| 5951 scan = reginput; | |
| 5952 if (got_int) | |
| 5953 break; | |
| 5954 } | |
|
5221
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
5955 else if (vim_isprintc(PTR2CHAR(scan)) == 1 |
|
9982ec574beb
updated for version 7.4a.036
Bram Moolenaar <bram@vim.org>
parents:
4901
diff
changeset
|
5956 && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5957 { |
| 39 | 5958 mb_ptr_adv(scan); |
| 7 | 5959 } |
| 5960 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5961 ++scan; | |
| 5962 else | |
| 5963 break; | |
| 5964 ++count; | |
| 5965 } | |
| 5966 break; | |
| 5967 | |
| 5968 case WHITE: | |
| 5969 case WHITE + ADD_NL: | |
| 5970 testval = mask = RI_WHITE; | |
| 5971 do_class: | |
| 5972 while (count < maxcount) | |
| 5973 { | |
| 5974 #ifdef FEAT_MBYTE | |
| 5975 int l; | |
| 5976 #endif | |
| 5977 if (*scan == NUL) | |
| 5978 { | |
| 1018 | 5979 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5980 || reg_line_lbr) | |
| 7 | 5981 break; |
| 5982 reg_nextline(); | |
| 5983 scan = reginput; | |
| 5984 if (got_int) | |
| 5985 break; | |
| 5986 } | |
| 5987 #ifdef FEAT_MBYTE | |
| 474 | 5988 else if (has_mbyte && (l = (*mb_ptr2len)(scan)) > 1) |
| 7 | 5989 { |
| 5990 if (testval != 0) | |
| 5991 break; | |
| 5992 scan += l; | |
| 5993 } | |
| 5994 #endif | |
| 5995 else if ((class_tab[*scan] & mask) == testval) | |
| 5996 ++scan; | |
| 5997 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5998 ++scan; | |
| 5999 else | |
| 6000 break; | |
| 6001 ++count; | |
| 6002 } | |
| 6003 break; | |
| 6004 | |
| 6005 case NWHITE: | |
| 6006 case NWHITE + ADD_NL: | |
| 6007 mask = RI_WHITE; | |
| 6008 goto do_class; | |
| 6009 case DIGIT: | |
| 6010 case DIGIT + ADD_NL: | |
| 6011 testval = mask = RI_DIGIT; | |
| 6012 goto do_class; | |
| 6013 case NDIGIT: | |
| 6014 case NDIGIT + ADD_NL: | |
| 6015 mask = RI_DIGIT; | |
| 6016 goto do_class; | |
| 6017 case HEX: | |
| 6018 case HEX + ADD_NL: | |
| 6019 testval = mask = RI_HEX; | |
| 6020 goto do_class; | |
| 6021 case NHEX: | |
| 6022 case NHEX + ADD_NL: | |
| 6023 mask = RI_HEX; | |
| 6024 goto do_class; | |
| 6025 case OCTAL: | |
| 6026 case OCTAL + ADD_NL: | |
| 6027 testval = mask = RI_OCTAL; | |
| 6028 goto do_class; | |
| 6029 case NOCTAL: | |
| 6030 case NOCTAL + ADD_NL: | |
| 6031 mask = RI_OCTAL; | |
| 6032 goto do_class; | |
| 6033 case WORD: | |
| 6034 case WORD + ADD_NL: | |
| 6035 testval = mask = RI_WORD; | |
| 6036 goto do_class; | |
| 6037 case NWORD: | |
| 6038 case NWORD + ADD_NL: | |
| 6039 mask = RI_WORD; | |
| 6040 goto do_class; | |
| 6041 case HEAD: | |
| 6042 case HEAD + ADD_NL: | |
| 6043 testval = mask = RI_HEAD; | |
| 6044 goto do_class; | |
| 6045 case NHEAD: | |
| 6046 case NHEAD + ADD_NL: | |
| 6047 mask = RI_HEAD; | |
| 6048 goto do_class; | |
| 6049 case ALPHA: | |
| 6050 case ALPHA + ADD_NL: | |
| 6051 testval = mask = RI_ALPHA; | |
| 6052 goto do_class; | |
| 6053 case NALPHA: | |
| 6054 case NALPHA + ADD_NL: | |
| 6055 mask = RI_ALPHA; | |
| 6056 goto do_class; | |
| 6057 case LOWER: | |
| 6058 case LOWER + ADD_NL: | |
| 6059 testval = mask = RI_LOWER; | |
| 6060 goto do_class; | |
| 6061 case NLOWER: | |
| 6062 case NLOWER + ADD_NL: | |
| 6063 mask = RI_LOWER; | |
| 6064 goto do_class; | |
| 6065 case UPPER: | |
| 6066 case UPPER + ADD_NL: | |
| 6067 testval = mask = RI_UPPER; | |
| 6068 goto do_class; | |
| 6069 case NUPPER: | |
| 6070 case NUPPER + ADD_NL: | |
| 6071 mask = RI_UPPER; | |
| 6072 goto do_class; | |
| 6073 | |
| 6074 case EXACTLY: | |
| 6075 { | |
| 6076 int cu, cl; | |
| 6077 | |
| 6078 /* This doesn't do a multi-byte character, because a MULTIBYTECODE | |
| 1347 | 6079 * would have been used for it. It does handle single-byte |
| 6080 * characters, such as latin1. */ | |
| 7 | 6081 if (ireg_ic) |
| 6082 { | |
| 1347 | 6083 cu = MB_TOUPPER(*opnd); |
| 6084 cl = MB_TOLOWER(*opnd); | |
| 7 | 6085 while (count < maxcount && (*scan == cu || *scan == cl)) |
| 6086 { | |
| 6087 count++; | |
| 6088 scan++; | |
| 6089 } | |
| 6090 } | |
| 6091 else | |
| 6092 { | |
| 6093 cu = *opnd; | |
| 6094 while (count < maxcount && *scan == cu) | |
| 6095 { | |
| 6096 count++; | |
| 6097 scan++; | |
| 6098 } | |
| 6099 } | |
| 6100 break; | |
| 6101 } | |
| 6102 | |
| 6103 #ifdef FEAT_MBYTE | |
| 6104 case MULTIBYTECODE: | |
| 6105 { | |
| 6106 int i, len, cf = 0; | |
| 6107 | |
| 6108 /* Safety check (just in case 'encoding' was changed since | |
| 6109 * compiling the program). */ | |
| 474 | 6110 if ((len = (*mb_ptr2len)(opnd)) > 1) |
| 7 | 6111 { |
| 6112 if (ireg_ic && enc_utf8) | |
| 6113 cf = utf_fold(utf_ptr2char(opnd)); | |
| 6114 while (count < maxcount) | |
| 6115 { | |
| 6116 for (i = 0; i < len; ++i) | |
| 6117 if (opnd[i] != scan[i]) | |
| 6118 break; | |
| 6119 if (i < len && (!ireg_ic || !enc_utf8 | |
| 6120 || utf_fold(utf_ptr2char(scan)) != cf)) | |
| 6121 break; | |
| 6122 scan += len; | |
| 6123 ++count; | |
| 6124 } | |
| 6125 } | |
| 6126 } | |
| 6127 break; | |
| 6128 #endif | |
| 6129 | |
| 6130 case ANYOF: | |
| 6131 case ANYOF + ADD_NL: | |
| 6132 testval = TRUE; | |
| 6133 /*FALLTHROUGH*/ | |
| 6134 | |
| 6135 case ANYBUT: | |
| 6136 case ANYBUT + ADD_NL: | |
| 6137 while (count < maxcount) | |
| 6138 { | |
| 6139 #ifdef FEAT_MBYTE | |
| 6140 int len; | |
| 6141 #endif | |
| 6142 if (*scan == NUL) | |
| 6143 { | |
| 1018 | 6144 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 6145 || reg_line_lbr) | |
| 7 | 6146 break; |
| 6147 reg_nextline(); | |
| 6148 scan = reginput; | |
| 6149 if (got_int) | |
| 6150 break; | |
| 6151 } | |
| 6152 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 6153 ++scan; | |
| 6154 #ifdef FEAT_MBYTE | |
| 474 | 6155 else if (has_mbyte && (len = (*mb_ptr2len)(scan)) > 1) |
| 7 | 6156 { |
| 6157 if ((cstrchr(opnd, (*mb_ptr2char)(scan)) == NULL) == testval) | |
| 6158 break; | |
| 6159 scan += len; | |
| 6160 } | |
| 6161 #endif | |
| 6162 else | |
| 6163 { | |
| 6164 if ((cstrchr(opnd, *scan) == NULL) == testval) | |
| 6165 break; | |
| 6166 ++scan; | |
| 6167 } | |
| 6168 ++count; | |
| 6169 } | |
| 6170 break; | |
| 6171 | |
| 6172 case NEWL: | |
| 6173 while (count < maxcount | |
| 1018 | 6174 && ((*scan == NUL && reglnum <= reg_maxline && !reg_line_lbr |
| 6175 && REG_MULTI) || (*scan == '\n' && reg_line_lbr))) | |
| 7 | 6176 { |
| 6177 count++; | |
| 6178 if (reg_line_lbr) | |
| 6179 ADVANCE_REGINPUT(); | |
| 6180 else | |
| 6181 reg_nextline(); | |
| 6182 scan = reginput; | |
| 6183 if (got_int) | |
| 6184 break; | |
| 6185 } | |
| 6186 break; | |
| 6187 | |
| 6188 default: /* Oh dear. Called inappropriately. */ | |
| 6189 EMSG(_(e_re_corr)); | |
| 6190 #ifdef DEBUG | |
| 6191 printf("Called regrepeat with op code %d\n", OP(p)); | |
| 6192 #endif | |
| 6193 break; | |
| 6194 } | |
| 6195 | |
| 6196 reginput = scan; | |
| 6197 | |
| 6198 return (int)count; | |
| 6199 } | |
| 6200 | |
| 6201 /* | |
| 6202 * regnext - dig the "next" pointer out of a node | |
| 2010 | 6203 * Returns NULL when calculating size, when there is no next item and when |
| 6204 * there is an error. | |
| 7 | 6205 */ |
| 6206 static char_u * | |
| 6207 regnext(p) | |
| 6208 char_u *p; | |
| 6209 { | |
| 6210 int offset; | |
| 6211 | |
| 2010 | 6212 if (p == JUST_CALC_SIZE || reg_toolong) |
| 7 | 6213 return NULL; |
| 6214 | |
| 6215 offset = NEXT(p); | |
| 6216 if (offset == 0) | |
| 6217 return NULL; | |
| 6218 | |
| 233 | 6219 if (OP(p) == BACK) |
| 7 | 6220 return p - offset; |
| 6221 else | |
| 6222 return p + offset; | |
| 6223 } | |
| 6224 | |
| 6225 /* | |
| 6226 * Check the regexp program for its magic number. | |
| 6227 * Return TRUE if it's wrong. | |
| 6228 */ | |
| 6229 static int | |
| 6230 prog_magic_wrong() | |
| 6231 { | |
| 4444 | 6232 regprog_T *prog; |
| 6233 | |
| 6234 prog = REG_MULTI ? reg_mmatch->regprog : reg_match->regprog; | |
| 6235 if (prog->engine == &nfa_regengine) | |
| 6236 /* For NFA matcher we don't check the magic */ | |
| 6237 return FALSE; | |
| 6238 | |
| 6239 if (UCHARAT(((bt_regprog_T *)prog)->program) != REGMAGIC) | |
| 7 | 6240 { |
| 6241 EMSG(_(e_re_corr)); | |
| 6242 return TRUE; | |
| 6243 } | |
| 6244 return FALSE; | |
| 6245 } | |
| 6246 | |
| 6247 /* | |
| 6248 * Cleanup the subexpressions, if this wasn't done yet. | |
| 6249 * This construction is used to clear the subexpressions only when they are | |
| 6250 * used (to increase speed). | |
| 6251 */ | |
| 6252 static void | |
| 6253 cleanup_subexpr() | |
| 6254 { | |
| 6255 if (need_clear_subexpr) | |
| 6256 { | |
| 6257 if (REG_MULTI) | |
| 6258 { | |
| 6259 /* Use 0xff to set lnum to -1 */ | |
| 6260 vim_memset(reg_startpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6261 vim_memset(reg_endpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6262 } | |
| 6263 else | |
| 6264 { | |
| 6265 vim_memset(reg_startp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6266 vim_memset(reg_endp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6267 } | |
| 6268 need_clear_subexpr = FALSE; | |
| 6269 } | |
| 6270 } | |
| 6271 | |
| 6272 #ifdef FEAT_SYN_HL | |
| 6273 static void | |
| 6274 cleanup_zsubexpr() | |
| 6275 { | |
| 6276 if (need_clear_zsubexpr) | |
| 6277 { | |
| 6278 if (REG_MULTI) | |
| 6279 { | |
| 6280 /* Use 0xff to set lnum to -1 */ | |
| 6281 vim_memset(reg_startzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6282 vim_memset(reg_endzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6283 } | |
| 6284 else | |
| 6285 { | |
| 6286 vim_memset(reg_startzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6287 vim_memset(reg_endzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6288 } | |
| 6289 need_clear_zsubexpr = FALSE; | |
| 6290 } | |
| 6291 } | |
| 6292 #endif | |
| 6293 | |
| 6294 /* | |
| 1579 | 6295 * Save the current subexpr to "bp", so that they can be restored |
| 6296 * later by restore_subexpr(). | |
| 6297 */ | |
| 6298 static void | |
| 6299 save_subexpr(bp) | |
| 6300 regbehind_T *bp; | |
| 6301 { | |
| 6302 int i; | |
| 6303 | |
| 1602 | 6304 /* When "need_clear_subexpr" is set we don't need to save the values, only |
| 6305 * remember that this flag needs to be set again when restoring. */ | |
| 6306 bp->save_need_clear_subexpr = need_clear_subexpr; | |
| 6307 if (!need_clear_subexpr) | |
| 1579 | 6308 { |
| 1602 | 6309 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6310 { |
| 1602 | 6311 if (REG_MULTI) |
| 6312 { | |
| 6313 bp->save_start[i].se_u.pos = reg_startpos[i]; | |
| 6314 bp->save_end[i].se_u.pos = reg_endpos[i]; | |
| 6315 } | |
| 6316 else | |
| 6317 { | |
| 6318 bp->save_start[i].se_u.ptr = reg_startp[i]; | |
| 6319 bp->save_end[i].se_u.ptr = reg_endp[i]; | |
| 6320 } | |
| 1579 | 6321 } |
| 6322 } | |
| 6323 } | |
| 6324 | |
| 6325 /* | |
| 6326 * Restore the subexpr from "bp". | |
| 6327 */ | |
| 6328 static void | |
| 6329 restore_subexpr(bp) | |
| 6330 regbehind_T *bp; | |
| 6331 { | |
| 6332 int i; | |
| 6333 | |
| 1602 | 6334 /* Only need to restore saved values when they are not to be cleared. */ |
| 6335 need_clear_subexpr = bp->save_need_clear_subexpr; | |
| 6336 if (!need_clear_subexpr) | |
| 1579 | 6337 { |
| 1602 | 6338 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6339 { |
| 1602 | 6340 if (REG_MULTI) |
| 6341 { | |
| 6342 reg_startpos[i] = bp->save_start[i].se_u.pos; | |
| 6343 reg_endpos[i] = bp->save_end[i].se_u.pos; | |
| 6344 } | |
| 6345 else | |
| 6346 { | |
| 6347 reg_startp[i] = bp->save_start[i].se_u.ptr; | |
| 6348 reg_endp[i] = bp->save_end[i].se_u.ptr; | |
| 6349 } | |
| 1579 | 6350 } |
| 6351 } | |
| 6352 } | |
| 6353 | |
| 6354 /* | |
| 7 | 6355 * Advance reglnum, regline and reginput to the next line. |
| 6356 */ | |
| 6357 static void | |
| 6358 reg_nextline() | |
| 6359 { | |
| 6360 regline = reg_getline(++reglnum); | |
| 6361 reginput = regline; | |
| 6362 fast_breakcheck(); | |
| 6363 } | |
| 6364 | |
| 6365 /* | |
| 6366 * Save the input line and position in a regsave_T. | |
| 6367 */ | |
| 6368 static void | |
| 233 | 6369 reg_save(save, gap) |
| 7 | 6370 regsave_T *save; |
| 233 | 6371 garray_T *gap; |
| 7 | 6372 { |
| 6373 if (REG_MULTI) | |
| 6374 { | |
| 6375 save->rs_u.pos.col = (colnr_T)(reginput - regline); | |
| 6376 save->rs_u.pos.lnum = reglnum; | |
| 6377 } | |
| 6378 else | |
| 6379 save->rs_u.ptr = reginput; | |
| 233 | 6380 save->rs_len = gap->ga_len; |
| 7 | 6381 } |
| 6382 | |
| 6383 /* | |
| 6384 * Restore the input line and position from a regsave_T. | |
| 6385 */ | |
| 6386 static void | |
| 233 | 6387 reg_restore(save, gap) |
| 7 | 6388 regsave_T *save; |
| 233 | 6389 garray_T *gap; |
| 7 | 6390 { |
| 6391 if (REG_MULTI) | |
| 6392 { | |
| 6393 if (reglnum != save->rs_u.pos.lnum) | |
| 6394 { | |
| 6395 /* only call reg_getline() when the line number changed to save | |
| 6396 * a bit of time */ | |
| 6397 reglnum = save->rs_u.pos.lnum; | |
| 6398 regline = reg_getline(reglnum); | |
| 6399 } | |
| 6400 reginput = regline + save->rs_u.pos.col; | |
| 6401 } | |
| 6402 else | |
| 6403 reginput = save->rs_u.ptr; | |
| 233 | 6404 gap->ga_len = save->rs_len; |
| 7 | 6405 } |
| 6406 | |
| 6407 /* | |
| 6408 * Return TRUE if current position is equal to saved position. | |
| 6409 */ | |
| 6410 static int | |
| 6411 reg_save_equal(save) | |
| 6412 regsave_T *save; | |
| 6413 { | |
| 6414 if (REG_MULTI) | |
| 6415 return reglnum == save->rs_u.pos.lnum | |
| 6416 && reginput == regline + save->rs_u.pos.col; | |
| 6417 return reginput == save->rs_u.ptr; | |
| 6418 } | |
| 6419 | |
| 6420 /* | |
| 6421 * Tentatively set the sub-expression start to the current position (after | |
| 6422 * calling regmatch() they will have changed). Need to save the existing | |
| 6423 * values for when there is no match. | |
| 6424 * Use se_save() to use pointer (save_se_multi()) or position (save_se_one()), | |
| 6425 * depending on REG_MULTI. | |
| 6426 */ | |
| 6427 static void | |
| 6428 save_se_multi(savep, posp) | |
| 6429 save_se_T *savep; | |
| 6430 lpos_T *posp; | |
| 6431 { | |
| 6432 savep->se_u.pos = *posp; | |
| 6433 posp->lnum = reglnum; | |
| 6434 posp->col = (colnr_T)(reginput - regline); | |
| 6435 } | |
| 6436 | |
| 6437 static void | |
| 6438 save_se_one(savep, pp) | |
| 6439 save_se_T *savep; | |
| 6440 char_u **pp; | |
| 6441 { | |
| 6442 savep->se_u.ptr = *pp; | |
| 6443 *pp = reginput; | |
| 6444 } | |
| 6445 | |
| 6446 /* | |
| 6447 * Compare a number with the operand of RE_LNUM, RE_COL or RE_VCOL. | |
| 6448 */ | |
| 6449 static int | |
| 6450 re_num_cmp(val, scan) | |
| 6451 long_u val; | |
| 6452 char_u *scan; | |
| 6453 { | |
| 6454 long_u n = OPERAND_MIN(scan); | |
| 6455 | |
| 6456 if (OPERAND_CMP(scan) == '>') | |
| 6457 return val > n; | |
| 6458 if (OPERAND_CMP(scan) == '<') | |
| 6459 return val < n; | |
| 6460 return val == n; | |
| 6461 } | |
| 6462 | |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6463 /* |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6464 * Check whether a backreference matches. |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6465 * Returns RA_FAIL, RA_NOMATCH or RA_MATCH. |
| 5504 | 6466 * If "bytelen" is not NULL, it is set to the byte length of the match in the |
| 6467 * last line. | |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6468 */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6469 static int |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6470 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
|
6471 linenr_T start_lnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6472 colnr_T start_col; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6473 linenr_T end_lnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6474 colnr_T end_col; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6475 int *bytelen; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6476 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6477 linenr_T clnum = start_lnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6478 colnr_T ccol = start_col; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6479 int len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6480 char_u *p; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6481 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6482 if (bytelen != NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6483 *bytelen = 0; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6484 for (;;) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6485 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6486 /* 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
|
6487 * Slow! */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6488 if (regline != reg_tofree) |
|
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 len = (int)STRLEN(regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6491 if (reg_tofree == NULL || len >= (int)reg_tofreelen) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6492 { |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6493 len += 50; /* get some extra */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6494 vim_free(reg_tofree); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6495 reg_tofree = alloc(len); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6496 if (reg_tofree == NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6497 return RA_FAIL; /* out of memory!*/ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6498 reg_tofreelen = len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6499 } |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6500 STRCPY(reg_tofree, regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6501 reginput = reg_tofree + (reginput - regline); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6502 regline = reg_tofree; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6503 } |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6504 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6505 /* Get the line to compare with. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6506 p = reg_getline(clnum); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6507 if (clnum == end_lnum) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6508 len = end_col - ccol; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6509 else |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6510 len = (int)STRLEN(p + ccol); |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6511 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6512 if (cstrncmp(p + ccol, reginput, &len) != 0) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6513 return RA_NOMATCH; /* doesn't match */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6514 if (bytelen != NULL) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6515 *bytelen += len; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6516 if (clnum == end_lnum) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6517 break; /* match and at end! */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6518 if (reglnum >= reg_maxline) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6519 return RA_NOMATCH; /* text too short */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6520 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6521 /* Advance to next line. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6522 reg_nextline(); |
| 5504 | 6523 if (bytelen != NULL) |
| 6524 *bytelen = 0; | |
|
4891
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6525 ++clnum; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6526 ccol = 0; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6527 if (got_int) |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6528 return RA_FAIL; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6529 } |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6530 |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6531 /* 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
|
6532 * that should not matter. */ |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6533 return RA_MATCH; |
|
4c42efb4c098
updated for version 7.3.1191
Bram Moolenaar <bram@vim.org>
parents:
4805
diff
changeset
|
6534 } |
| 7 | 6535 |
| 4444 | 6536 #ifdef BT_REGEXP_DUMP |
| 7 | 6537 |
| 6538 /* | |
| 6539 * regdump - dump a regexp onto stdout in vaguely comprehensible form | |
| 6540 */ | |
| 6541 static void | |
| 6542 regdump(pattern, r) | |
| 6543 char_u *pattern; | |
| 4444 | 6544 bt_regprog_T *r; |
| 7 | 6545 { |
| 6546 char_u *s; | |
| 6547 int op = EXACTLY; /* Arbitrary non-END op. */ | |
| 6548 char_u *next; | |
| 6549 char_u *end = NULL; | |
| 4444 | 6550 FILE *f; |
| 6551 | |
| 6552 #ifdef BT_REGEXP_LOG | |
| 6553 f = fopen("bt_regexp_log.log", "a"); | |
| 6554 #else | |
| 6555 f = stdout; | |
| 6556 #endif | |
| 6557 if (f == NULL) | |
| 6558 return; | |
| 6559 fprintf(f, "-------------------------------------\n\r\nregcomp(%s):\r\n", pattern); | |
| 7 | 6560 |
| 6561 s = r->program + 1; | |
| 6562 /* | |
| 6563 * Loop until we find the END that isn't before a referred next (an END | |
| 6564 * can also appear in a NOMATCH operand). | |
| 6565 */ | |
| 6566 while (op != END || s <= end) | |
| 6567 { | |
| 6568 op = OP(s); | |
| 4444 | 6569 fprintf(f, "%2d%s", (int)(s - r->program), regprop(s)); /* Where, what. */ |
| 7 | 6570 next = regnext(s); |
| 6571 if (next == NULL) /* Next ptr. */ | |
| 4444 | 6572 fprintf(f, "(0)"); |
| 7 | 6573 else |
| 4444 | 6574 fprintf(f, "(%d)", (int)((s - r->program) + (next - s))); |
| 7 | 6575 if (end < next) |
| 6576 end = next; | |
| 6577 if (op == BRACE_LIMITS) | |
| 6578 { | |
|
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6579 /* Two ints */ |
| 4444 | 6580 fprintf(f, " minval %ld, maxval %ld", OPERAND_MIN(s), OPERAND_MAX(s)); |
| 7 | 6581 s += 8; |
| 6582 } | |
|
4746
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6583 else if (op == BEHIND || op == NOBEHIND) |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6584 { |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6585 /* one int */ |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6586 fprintf(f, " count %ld", OPERAND_MIN(s)); |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6587 s += 4; |
|
d1376091d18b
updated for version 7.3.1120
Bram Moolenaar <bram@vim.org>
parents:
4744
diff
changeset
|
6588 } |
|
4770
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6589 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
|
6590 { |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6591 /* one int plus comperator */ |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6592 fprintf(f, " count %ld", OPERAND_MIN(s)); |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6593 s += 5; |
|
b20dbf3a5370
updated for version 7.3.1132
Bram Moolenaar <bram@vim.org>
parents:
4762
diff
changeset
|
6594 } |
| 7 | 6595 s += 3; |
| 6596 if (op == ANYOF || op == ANYOF + ADD_NL | |
| 6597 || op == ANYBUT || op == ANYBUT + ADD_NL | |
| 6598 || op == EXACTLY) | |
| 6599 { | |
| 6600 /* Literal string, where present. */ | |
| 4444 | 6601 fprintf(f, "\nxxxxxxxxx\n"); |
| 7 | 6602 while (*s != NUL) |
| 4444 | 6603 fprintf(f, "%c", *s++); |
| 6604 fprintf(f, "\nxxxxxxxxx\n"); | |
| 7 | 6605 s++; |
| 6606 } | |
| 4444 | 6607 fprintf(f, "\r\n"); |
| 7 | 6608 } |
| 6609 | |
| 6610 /* Header fields of interest. */ | |
| 6611 if (r->regstart != NUL) | |
| 4444 | 6612 fprintf(f, "start `%s' 0x%x; ", r->regstart < 256 |
| 7 | 6613 ? (char *)transchar(r->regstart) |
| 6614 : "multibyte", r->regstart); | |
| 6615 if (r->reganch) | |
| 4444 | 6616 fprintf(f, "anchored; "); |
| 7 | 6617 if (r->regmust != NULL) |
| 4444 | 6618 fprintf(f, "must have \"%s\"", r->regmust); |
| 6619 fprintf(f, "\r\n"); | |
| 6620 | |
| 6621 #ifdef BT_REGEXP_LOG | |
| 6622 fclose(f); | |
| 6623 #endif | |
| 7 | 6624 } |
| 4444 | 6625 #endif /* BT_REGEXP_DUMP */ |
| 6626 | |
| 6627 #ifdef DEBUG | |
| 7 | 6628 /* |
| 6629 * regprop - printable representation of opcode | |
| 6630 */ | |
| 6631 static char_u * | |
| 6632 regprop(op) | |
| 6633 char_u *op; | |
| 6634 { | |
| 4444 | 6635 char *p; |
| 6636 static char buf[50]; | |
| 6637 | |
| 6638 STRCPY(buf, ":"); | |
| 6639 | |
| 6640 switch ((int) OP(op)) | |
| 7 | 6641 { |
| 6642 case BOL: | |
| 6643 p = "BOL"; | |
| 6644 break; | |
| 6645 case EOL: | |
| 6646 p = "EOL"; | |
| 6647 break; | |
| 6648 case RE_BOF: | |
| 6649 p = "BOF"; | |
| 6650 break; | |
| 6651 case RE_EOF: | |
| 6652 p = "EOF"; | |
| 6653 break; | |
| 6654 case CURSOR: | |
| 6655 p = "CURSOR"; | |
| 6656 break; | |
| 639 | 6657 case RE_VISUAL: |
| 6658 p = "RE_VISUAL"; | |
| 6659 break; | |
| 7 | 6660 case RE_LNUM: |
| 6661 p = "RE_LNUM"; | |
| 6662 break; | |
| 639 | 6663 case RE_MARK: |
| 6664 p = "RE_MARK"; | |
| 6665 break; | |
| 7 | 6666 case RE_COL: |
| 6667 p = "RE_COL"; | |
| 6668 break; | |
| 6669 case RE_VCOL: | |
| 6670 p = "RE_VCOL"; | |
| 6671 break; | |
| 6672 case BOW: | |
| 6673 p = "BOW"; | |
| 6674 break; | |
| 6675 case EOW: | |
| 6676 p = "EOW"; | |
| 6677 break; | |
| 6678 case ANY: | |
| 6679 p = "ANY"; | |
| 6680 break; | |
| 6681 case ANY + ADD_NL: | |
| 6682 p = "ANY+NL"; | |
| 6683 break; | |
| 6684 case ANYOF: | |
| 6685 p = "ANYOF"; | |
| 6686 break; | |
| 6687 case ANYOF + ADD_NL: | |
| 6688 p = "ANYOF+NL"; | |
| 6689 break; | |
| 6690 case ANYBUT: | |
| 6691 p = "ANYBUT"; | |
| 6692 break; | |
| 6693 case ANYBUT + ADD_NL: | |
| 6694 p = "ANYBUT+NL"; | |
| 6695 break; | |
| 6696 case IDENT: | |
| 6697 p = "IDENT"; | |
| 6698 break; | |
| 6699 case IDENT + ADD_NL: | |
| 6700 p = "IDENT+NL"; | |
| 6701 break; | |
| 6702 case SIDENT: | |
| 6703 p = "SIDENT"; | |
| 6704 break; | |
| 6705 case SIDENT + ADD_NL: | |
| 6706 p = "SIDENT+NL"; | |
| 6707 break; | |
| 6708 case KWORD: | |
| 6709 p = "KWORD"; | |
| 6710 break; | |
| 6711 case KWORD + ADD_NL: | |
| 6712 p = "KWORD+NL"; | |
| 6713 break; | |
| 6714 case SKWORD: | |
| 6715 p = "SKWORD"; | |
| 6716 break; | |
| 6717 case SKWORD + ADD_NL: | |
| 6718 p = "SKWORD+NL"; | |
| 6719 break; | |
| 6720 case FNAME: | |
| 6721 p = "FNAME"; | |
| 6722 break; | |
| 6723 case FNAME + ADD_NL: | |
| 6724 p = "FNAME+NL"; | |
| 6725 break; | |
| 6726 case SFNAME: | |
| 6727 p = "SFNAME"; | |
| 6728 break; | |
| 6729 case SFNAME + ADD_NL: | |
| 6730 p = "SFNAME+NL"; | |
| 6731 break; | |
| 6732 case PRINT: | |
| 6733 p = "PRINT"; | |
| 6734 break; | |
| 6735 case PRINT + ADD_NL: | |
| 6736 p = "PRINT+NL"; | |
| 6737 break; | |
| 6738 case SPRINT: | |
| 6739 p = "SPRINT"; | |
| 6740 break; | |
| 6741 case SPRINT + ADD_NL: | |
| 6742 p = "SPRINT+NL"; | |
| 6743 break; | |
| 6744 case WHITE: | |
| 6745 p = "WHITE"; | |
| 6746 break; | |
| 6747 case WHITE + ADD_NL: | |
| 6748 p = "WHITE+NL"; | |
| 6749 break; | |
| 6750 case NWHITE: | |
| 6751 p = "NWHITE"; | |
| 6752 break; | |
| 6753 case NWHITE + ADD_NL: | |
| 6754 p = "NWHITE+NL"; | |
| 6755 break; | |
| 6756 case DIGIT: | |
| 6757 p = "DIGIT"; | |
| 6758 break; | |
| 6759 case DIGIT + ADD_NL: | |
| 6760 p = "DIGIT+NL"; | |
| 6761 break; | |
| 6762 case NDIGIT: | |
| 6763 p = "NDIGIT"; | |
| 6764 break; | |
| 6765 case NDIGIT + ADD_NL: | |
| 6766 p = "NDIGIT+NL"; | |
| 6767 break; | |
| 6768 case HEX: | |
| 6769 p = "HEX"; | |
| 6770 break; | |
| 6771 case HEX + ADD_NL: | |
| 6772 p = "HEX+NL"; | |
| 6773 break; | |
| 6774 case NHEX: | |
| 6775 p = "NHEX"; | |
| 6776 break; | |
| 6777 case NHEX + ADD_NL: | |
| 6778 p = "NHEX+NL"; | |
| 6779 break; | |
| 6780 case OCTAL: | |
| 6781 p = "OCTAL"; | |
| 6782 break; | |
| 6783 case OCTAL + ADD_NL: | |
| 6784 p = "OCTAL+NL"; | |
| 6785 break; | |
| 6786 case NOCTAL: | |
| 6787 p = "NOCTAL"; | |
| 6788 break; | |
| 6789 case NOCTAL + ADD_NL: | |
| 6790 p = "NOCTAL+NL"; | |
| 6791 break; | |
| 6792 case WORD: | |
| 6793 p = "WORD"; | |
| 6794 break; | |
| 6795 case WORD + ADD_NL: | |
| 6796 p = "WORD+NL"; | |
| 6797 break; | |
| 6798 case NWORD: | |
| 6799 p = "NWORD"; | |
| 6800 break; | |
| 6801 case NWORD + ADD_NL: | |
| 6802 p = "NWORD+NL"; | |
| 6803 break; | |
| 6804 case HEAD: | |
| 6805 p = "HEAD"; | |
| 6806 break; | |
| 6807 case HEAD + ADD_NL: | |
| 6808 p = "HEAD+NL"; | |
| 6809 break; | |
| 6810 case NHEAD: | |
| 6811 p = "NHEAD"; | |
| 6812 break; | |
| 6813 case NHEAD + ADD_NL: | |
| 6814 p = "NHEAD+NL"; | |
| 6815 break; | |
| 6816 case ALPHA: | |
| 6817 p = "ALPHA"; | |
| 6818 break; | |
| 6819 case ALPHA + ADD_NL: | |
| 6820 p = "ALPHA+NL"; | |
| 6821 break; | |
| 6822 case NALPHA: | |
| 6823 p = "NALPHA"; | |
| 6824 break; | |
| 6825 case NALPHA + ADD_NL: | |
| 6826 p = "NALPHA+NL"; | |
| 6827 break; | |
| 6828 case LOWER: | |
| 6829 p = "LOWER"; | |
| 6830 break; | |
| 6831 case LOWER + ADD_NL: | |
| 6832 p = "LOWER+NL"; | |
| 6833 break; | |
| 6834 case NLOWER: | |
| 6835 p = "NLOWER"; | |
| 6836 break; | |
| 6837 case NLOWER + ADD_NL: | |
| 6838 p = "NLOWER+NL"; | |
| 6839 break; | |
| 6840 case UPPER: | |
| 6841 p = "UPPER"; | |
| 6842 break; | |
| 6843 case UPPER + ADD_NL: | |
| 6844 p = "UPPER+NL"; | |
| 6845 break; | |
| 6846 case NUPPER: | |
| 6847 p = "NUPPER"; | |
| 6848 break; | |
| 6849 case NUPPER + ADD_NL: | |
| 6850 p = "NUPPER+NL"; | |
| 6851 break; | |
| 6852 case BRANCH: | |
| 6853 p = "BRANCH"; | |
| 6854 break; | |
| 6855 case EXACTLY: | |
| 6856 p = "EXACTLY"; | |
| 6857 break; | |
| 6858 case NOTHING: | |
| 6859 p = "NOTHING"; | |
| 6860 break; | |
| 6861 case BACK: | |
| 6862 p = "BACK"; | |
| 6863 break; | |
| 6864 case END: | |
| 6865 p = "END"; | |
| 6866 break; | |
| 6867 case MOPEN + 0: | |
| 6868 p = "MATCH START"; | |
| 6869 break; | |
| 6870 case MOPEN + 1: | |
| 6871 case MOPEN + 2: | |
| 6872 case MOPEN + 3: | |
| 6873 case MOPEN + 4: | |
| 6874 case MOPEN + 5: | |
| 6875 case MOPEN + 6: | |
| 6876 case MOPEN + 7: | |
| 6877 case MOPEN + 8: | |
| 6878 case MOPEN + 9: | |
| 6879 sprintf(buf + STRLEN(buf), "MOPEN%d", OP(op) - MOPEN); | |
| 6880 p = NULL; | |
| 6881 break; | |
| 6882 case MCLOSE + 0: | |
| 6883 p = "MATCH END"; | |
| 6884 break; | |
| 6885 case MCLOSE + 1: | |
| 6886 case MCLOSE + 2: | |
| 6887 case MCLOSE + 3: | |
| 6888 case MCLOSE + 4: | |
| 6889 case MCLOSE + 5: | |
| 6890 case MCLOSE + 6: | |
| 6891 case MCLOSE + 7: | |
| 6892 case MCLOSE + 8: | |
| 6893 case MCLOSE + 9: | |
| 6894 sprintf(buf + STRLEN(buf), "MCLOSE%d", OP(op) - MCLOSE); | |
| 6895 p = NULL; | |
| 6896 break; | |
| 6897 case BACKREF + 1: | |
| 6898 case BACKREF + 2: | |
| 6899 case BACKREF + 3: | |
| 6900 case BACKREF + 4: | |
| 6901 case BACKREF + 5: | |
| 6902 case BACKREF + 6: | |
| 6903 case BACKREF + 7: | |
| 6904 case BACKREF + 8: | |
| 6905 case BACKREF + 9: | |
| 6906 sprintf(buf + STRLEN(buf), "BACKREF%d", OP(op) - BACKREF); | |
| 6907 p = NULL; | |
| 6908 break; | |
| 6909 case NOPEN: | |
| 6910 p = "NOPEN"; | |
| 6911 break; | |
| 6912 case NCLOSE: | |
| 6913 p = "NCLOSE"; | |
| 6914 break; | |
| 6915 #ifdef FEAT_SYN_HL | |
| 6916 case ZOPEN + 1: | |
| 6917 case ZOPEN + 2: | |
| 6918 case ZOPEN + 3: | |
| 6919 case ZOPEN + 4: | |
| 6920 case ZOPEN + 5: | |
| 6921 case ZOPEN + 6: | |
| 6922 case ZOPEN + 7: | |
| 6923 case ZOPEN + 8: | |
| 6924 case ZOPEN + 9: | |
| 6925 sprintf(buf + STRLEN(buf), "ZOPEN%d", OP(op) - ZOPEN); | |
| 6926 p = NULL; | |
| 6927 break; | |
| 6928 case ZCLOSE + 1: | |
| 6929 case ZCLOSE + 2: | |
| 6930 case ZCLOSE + 3: | |
| 6931 case ZCLOSE + 4: | |
| 6932 case ZCLOSE + 5: | |
| 6933 case ZCLOSE + 6: | |
| 6934 case ZCLOSE + 7: | |
| 6935 case ZCLOSE + 8: | |
| 6936 case ZCLOSE + 9: | |
| 6937 sprintf(buf + STRLEN(buf), "ZCLOSE%d", OP(op) - ZCLOSE); | |
| 6938 p = NULL; | |
| 6939 break; | |
| 6940 case ZREF + 1: | |
| 6941 case ZREF + 2: | |
| 6942 case ZREF + 3: | |
| 6943 case ZREF + 4: | |
| 6944 case ZREF + 5: | |
| 6945 case ZREF + 6: | |
| 6946 case ZREF + 7: | |
| 6947 case ZREF + 8: | |
| 6948 case ZREF + 9: | |
| 6949 sprintf(buf + STRLEN(buf), "ZREF%d", OP(op) - ZREF); | |
| 6950 p = NULL; | |
| 6951 break; | |
| 6952 #endif | |
| 6953 case STAR: | |
| 6954 p = "STAR"; | |
| 6955 break; | |
| 6956 case PLUS: | |
| 6957 p = "PLUS"; | |
| 6958 break; | |
| 6959 case NOMATCH: | |
| 6960 p = "NOMATCH"; | |
| 6961 break; | |
| 6962 case MATCH: | |
| 6963 p = "MATCH"; | |
| 6964 break; | |
| 6965 case BEHIND: | |
| 6966 p = "BEHIND"; | |
| 6967 break; | |
| 6968 case NOBEHIND: | |
| 6969 p = "NOBEHIND"; | |
| 6970 break; | |
| 6971 case SUBPAT: | |
| 6972 p = "SUBPAT"; | |
| 6973 break; | |
| 6974 case BRACE_LIMITS: | |
| 6975 p = "BRACE_LIMITS"; | |
| 6976 break; | |
| 6977 case BRACE_SIMPLE: | |
| 6978 p = "BRACE_SIMPLE"; | |
| 6979 break; | |
| 6980 case BRACE_COMPLEX + 0: | |
| 6981 case BRACE_COMPLEX + 1: | |
| 6982 case BRACE_COMPLEX + 2: | |
| 6983 case BRACE_COMPLEX + 3: | |
| 6984 case BRACE_COMPLEX + 4: | |
| 6985 case BRACE_COMPLEX + 5: | |
| 6986 case BRACE_COMPLEX + 6: | |
| 6987 case BRACE_COMPLEX + 7: | |
| 6988 case BRACE_COMPLEX + 8: | |
| 6989 case BRACE_COMPLEX + 9: | |
| 6990 sprintf(buf + STRLEN(buf), "BRACE_COMPLEX%d", OP(op) - BRACE_COMPLEX); | |
| 6991 p = NULL; | |
| 6992 break; | |
| 6993 #ifdef FEAT_MBYTE | |
| 6994 case MULTIBYTECODE: | |
| 6995 p = "MULTIBYTECODE"; | |
| 6996 break; | |
| 6997 #endif | |
| 6998 case NEWL: | |
| 6999 p = "NEWL"; | |
| 7000 break; | |
| 7001 default: | |
| 7002 sprintf(buf + STRLEN(buf), "corrupt %d", OP(op)); | |
| 7003 p = NULL; | |
| 7004 break; | |
| 7005 } | |
| 7006 if (p != NULL) | |
| 4444 | 7007 STRCAT(buf, p); |
| 7008 return (char_u *)buf; | |
| 7 | 7009 } |
| 4444 | 7010 #endif /* DEBUG */ |
| 7 | 7011 |
| 6203 | 7012 /* |
| 7013 * Used in a place where no * or \+ can follow. | |
| 7014 */ | |
| 7015 static int | |
| 7016 re_mult_next(what) | |
| 7017 char *what; | |
| 7018 { | |
| 7019 if (re_multi_type(peekchr()) == MULTI_MULT) | |
| 7020 EMSG2_RET_FAIL(_("E888: (NFA regexp) cannot repeat %s"), what); | |
| 7021 return OK; | |
| 7022 } | |
| 7023 | |
| 7 | 7024 #ifdef FEAT_MBYTE |
| 7025 static void mb_decompose __ARGS((int c, int *c1, int *c2, int *c3)); | |
| 7026 | |
| 7027 typedef struct | |
| 7028 { | |
| 7029 int a, b, c; | |
| 7030 } decomp_T; | |
| 7031 | |
| 7032 | |
| 7033 /* 0xfb20 - 0xfb4f */ | |
| 297 | 7034 static decomp_T decomp_table[0xfb4f-0xfb20+1] = |
| 7 | 7035 { |
| 7036 {0x5e2,0,0}, /* 0xfb20 alt ayin */ | |
| 7037 {0x5d0,0,0}, /* 0xfb21 alt alef */ | |
| 7038 {0x5d3,0,0}, /* 0xfb22 alt dalet */ | |
| 7039 {0x5d4,0,0}, /* 0xfb23 alt he */ | |
| 7040 {0x5db,0,0}, /* 0xfb24 alt kaf */ | |
| 7041 {0x5dc,0,0}, /* 0xfb25 alt lamed */ | |
| 7042 {0x5dd,0,0}, /* 0xfb26 alt mem-sofit */ | |
| 7043 {0x5e8,0,0}, /* 0xfb27 alt resh */ | |
| 7044 {0x5ea,0,0}, /* 0xfb28 alt tav */ | |
| 7045 {'+', 0, 0}, /* 0xfb29 alt plus */ | |
| 7046 {0x5e9, 0x5c1, 0}, /* 0xfb2a shin+shin-dot */ | |
| 7047 {0x5e9, 0x5c2, 0}, /* 0xfb2b shin+sin-dot */ | |
| 7048 {0x5e9, 0x5c1, 0x5bc}, /* 0xfb2c shin+shin-dot+dagesh */ | |
| 7049 {0x5e9, 0x5c2, 0x5bc}, /* 0xfb2d shin+sin-dot+dagesh */ | |
| 7050 {0x5d0, 0x5b7, 0}, /* 0xfb2e alef+patah */ | |
| 7051 {0x5d0, 0x5b8, 0}, /* 0xfb2f alef+qamats */ | |
| 7052 {0x5d0, 0x5b4, 0}, /* 0xfb30 alef+hiriq */ | |
| 7053 {0x5d1, 0x5bc, 0}, /* 0xfb31 bet+dagesh */ | |
| 7054 {0x5d2, 0x5bc, 0}, /* 0xfb32 gimel+dagesh */ | |
| 7055 {0x5d3, 0x5bc, 0}, /* 0xfb33 dalet+dagesh */ | |
| 7056 {0x5d4, 0x5bc, 0}, /* 0xfb34 he+dagesh */ | |
| 7057 {0x5d5, 0x5bc, 0}, /* 0xfb35 vav+dagesh */ | |
| 7058 {0x5d6, 0x5bc, 0}, /* 0xfb36 zayin+dagesh */ | |
| 7059 {0xfb37, 0, 0}, /* 0xfb37 -- UNUSED */ | |
| 7060 {0x5d8, 0x5bc, 0}, /* 0xfb38 tet+dagesh */ | |
| 7061 {0x5d9, 0x5bc, 0}, /* 0xfb39 yud+dagesh */ | |
| 7062 {0x5da, 0x5bc, 0}, /* 0xfb3a kaf sofit+dagesh */ | |
| 7063 {0x5db, 0x5bc, 0}, /* 0xfb3b kaf+dagesh */ | |
| 7064 {0x5dc, 0x5bc, 0}, /* 0xfb3c lamed+dagesh */ | |
| 7065 {0xfb3d, 0, 0}, /* 0xfb3d -- UNUSED */ | |
| 7066 {0x5de, 0x5bc, 0}, /* 0xfb3e mem+dagesh */ | |
| 7067 {0xfb3f, 0, 0}, /* 0xfb3f -- UNUSED */ | |
| 7068 {0x5e0, 0x5bc, 0}, /* 0xfb40 nun+dagesh */ | |
| 7069 {0x5e1, 0x5bc, 0}, /* 0xfb41 samech+dagesh */ | |
| 7070 {0xfb42, 0, 0}, /* 0xfb42 -- UNUSED */ | |
| 7071 {0x5e3, 0x5bc, 0}, /* 0xfb43 pe sofit+dagesh */ | |
| 7072 {0x5e4, 0x5bc,0}, /* 0xfb44 pe+dagesh */ | |
| 7073 {0xfb45, 0, 0}, /* 0xfb45 -- UNUSED */ | |
| 7074 {0x5e6, 0x5bc, 0}, /* 0xfb46 tsadi+dagesh */ | |
| 7075 {0x5e7, 0x5bc, 0}, /* 0xfb47 qof+dagesh */ | |
| 7076 {0x5e8, 0x5bc, 0}, /* 0xfb48 resh+dagesh */ | |
| 7077 {0x5e9, 0x5bc, 0}, /* 0xfb49 shin+dagesh */ | |
| 7078 {0x5ea, 0x5bc, 0}, /* 0xfb4a tav+dagesh */ | |
| 7079 {0x5d5, 0x5b9, 0}, /* 0xfb4b vav+holam */ | |
| 7080 {0x5d1, 0x5bf, 0}, /* 0xfb4c bet+rafe */ | |
| 7081 {0x5db, 0x5bf, 0}, /* 0xfb4d kaf+rafe */ | |
| 7082 {0x5e4, 0x5bf, 0}, /* 0xfb4e pe+rafe */ | |
| 7083 {0x5d0, 0x5dc, 0} /* 0xfb4f alef-lamed */ | |
| 7084 }; | |
| 7085 | |
| 7086 static void | |
| 7087 mb_decompose(c, c1, c2, c3) | |
| 7088 int c, *c1, *c2, *c3; | |
| 7089 { | |
| 7090 decomp_T d; | |
| 7091 | |
|
4505
d037b9cbdaaa
updated for version 7.3.1000
Bram Moolenaar <bram@vim.org>
parents:
4466
diff
changeset
|
7092 if (c >= 0xfb20 && c <= 0xfb4f) |
| 7 | 7093 { |
| 7094 d = decomp_table[c - 0xfb20]; | |
| 7095 *c1 = d.a; | |
| 7096 *c2 = d.b; | |
| 7097 *c3 = d.c; | |
| 7098 } | |
| 7099 else | |
| 7100 { | |
| 7101 *c1 = c; | |
| 7102 *c2 = *c3 = 0; | |
| 7103 } | |
| 7104 } | |
| 7105 #endif | |
| 7106 | |
| 7107 /* | |
| 7108 * Compare two strings, ignore case if ireg_ic set. | |
| 7109 * Return 0 if strings match, non-zero otherwise. | |
| 7110 * Correct the length "*n" when composing characters are ignored. | |
| 7111 */ | |
| 7112 static int | |
| 7113 cstrncmp(s1, s2, n) | |
| 7114 char_u *s1, *s2; | |
| 7115 int *n; | |
| 7116 { | |
| 7117 int result; | |
| 7118 | |
| 7119 if (!ireg_ic) | |
| 7120 result = STRNCMP(s1, s2, *n); | |
| 7121 else | |
| 7122 result = MB_STRNICMP(s1, s2, *n); | |
| 7123 | |
| 7124 #ifdef FEAT_MBYTE | |
| 7125 /* if it failed and it's utf8 and we want to combineignore: */ | |
| 7126 if (result != 0 && enc_utf8 && ireg_icombine) | |
| 7127 { | |
| 7128 char_u *str1, *str2; | |
| 7129 int c1, c2, c11, c12; | |
| 7130 int junk; | |
| 7131 | |
| 7132 /* we have to handle the strcmp ourselves, since it is necessary to | |
| 7133 * deal with the composing characters by ignoring them: */ | |
| 7134 str1 = s1; | |
| 7135 str2 = s2; | |
| 7136 c1 = c2 = 0; | |
| 507 | 7137 while ((int)(str1 - s1) < *n) |
| 7 | 7138 { |
| 7139 c1 = mb_ptr2char_adv(&str1); | |
| 7140 c2 = mb_ptr2char_adv(&str2); | |
| 7141 | |
| 7142 /* decompose the character if necessary, into 'base' characters | |
| 7143 * because I don't care about Arabic, I will hard-code the Hebrew | |
| 7144 * which I *do* care about! So sue me... */ | |
| 7145 if (c1 != c2 && (!ireg_ic || utf_fold(c1) != utf_fold(c2))) | |
| 7146 { | |
| 7147 /* decomposition necessary? */ | |
| 7148 mb_decompose(c1, &c11, &junk, &junk); | |
| 7149 mb_decompose(c2, &c12, &junk, &junk); | |
| 7150 c1 = c11; | |
| 7151 c2 = c12; | |
| 7152 if (c11 != c12 && (!ireg_ic || utf_fold(c11) != utf_fold(c12))) | |
| 7153 break; | |
| 7154 } | |
| 7155 } | |
| 7156 result = c2 - c1; | |
| 7157 if (result == 0) | |
| 7158 *n = (int)(str2 - s2); | |
| 7159 } | |
| 7160 #endif | |
| 7161 | |
| 7162 return result; | |
| 7163 } | |
| 7164 | |
| 7165 /* | |
| 7166 * cstrchr: This function is used a lot for simple searches, keep it fast! | |
| 7167 */ | |
| 7168 static char_u * | |
| 7169 cstrchr(s, c) | |
| 7170 char_u *s; | |
| 7171 int c; | |
| 7172 { | |
| 7173 char_u *p; | |
| 7174 int cc; | |
| 7175 | |
| 7176 if (!ireg_ic | |
| 7177 #ifdef FEAT_MBYTE | |
| 7178 || (!enc_utf8 && mb_char2len(c) > 1) | |
| 7179 #endif | |
| 7180 ) | |
| 7181 return vim_strchr(s, c); | |
| 7182 | |
| 7183 /* tolower() and toupper() can be slow, comparing twice should be a lot | |
| 7184 * faster (esp. when using MS Visual C++!). | |
| 7185 * For UTF-8 need to use folded case. */ | |
| 7186 #ifdef FEAT_MBYTE | |
| 7187 if (enc_utf8 && c > 0x80) | |
| 7188 cc = utf_fold(c); | |
| 7189 else | |
| 7190 #endif | |
| 1347 | 7191 if (MB_ISUPPER(c)) |
| 7192 cc = MB_TOLOWER(c); | |
| 7193 else if (MB_ISLOWER(c)) | |
| 7194 cc = MB_TOUPPER(c); | |
| 7 | 7195 else |
| 7196 return vim_strchr(s, c); | |
| 7197 | |
| 7198 #ifdef FEAT_MBYTE | |
| 7199 if (has_mbyte) | |
| 7200 { | |
| 474 | 7201 for (p = s; *p != NUL; p += (*mb_ptr2len)(p)) |
| 7 | 7202 { |
| 7203 if (enc_utf8 && c > 0x80) | |
| 7204 { | |
| 7205 if (utf_fold(utf_ptr2char(p)) == cc) | |
| 7206 return p; | |
| 7207 } | |
| 7208 else if (*p == c || *p == cc) | |
| 7209 return p; | |
| 7210 } | |
| 7211 } | |
| 7212 else | |
| 7213 #endif | |
| 7214 /* Faster version for when there are no multi-byte characters. */ | |
| 7215 for (p = s; *p != NUL; ++p) | |
| 7216 if (*p == c || *p == cc) | |
| 7217 return p; | |
| 7218 | |
| 7219 return NULL; | |
| 7220 } | |
| 7221 | |
| 7222 /*************************************************************** | |
| 7223 * regsub stuff * | |
| 7224 ***************************************************************/ | |
| 7225 | |
| 7226 /* This stuff below really confuses cc on an SGI -- webb */ | |
| 7227 #ifdef __sgi | |
| 7228 # undef __ARGS | |
| 7229 # define __ARGS(x) () | |
| 7230 #endif | |
| 7231 | |
| 7232 /* | |
| 7233 * We should define ftpr as a pointer to a function returning a pointer to | |
| 7234 * a function returning a pointer to a function ... | |
| 7235 * This is impossible, so we declare a pointer to a function returning a | |
| 7236 * pointer to a function returning void. This should work for all compilers. | |
| 7237 */ | |
| 772 | 7238 typedef void (*(*fptr_T) __ARGS((int *, int)))(); |
| 7239 | |
| 7240 static fptr_T do_upper __ARGS((int *, int)); | |
| 7241 static fptr_T do_Upper __ARGS((int *, int)); | |
| 7242 static fptr_T do_lower __ARGS((int *, int)); | |
| 7243 static fptr_T do_Lower __ARGS((int *, int)); | |
| 7 | 7244 |
| 7245 static int vim_regsub_both __ARGS((char_u *source, char_u *dest, int copy, int magic, int backslash)); | |
| 7246 | |
| 772 | 7247 static fptr_T |
| 7 | 7248 do_upper(d, c) |
| 772 | 7249 int *d; |
| 7250 int c; | |
| 7 | 7251 { |
| 772 | 7252 *d = MB_TOUPPER(c); |
| 7253 | |
| 7254 return (fptr_T)NULL; | |
| 7 | 7255 } |
| 7256 | |
| 772 | 7257 static fptr_T |
| 7258 do_Upper(d, c) | |
| 7259 int *d; | |
| 7 | 7260 int c; |
| 7261 { | |
| 772 | 7262 *d = MB_TOUPPER(c); |
| 7263 | |
| 7264 return (fptr_T)do_Upper; | |
| 7265 } | |
| 7266 | |
| 7267 static fptr_T | |
| 7268 do_lower(d, c) | |
| 7269 int *d; | |
| 7270 int c; | |
| 7271 { | |
| 7272 *d = MB_TOLOWER(c); | |
| 7273 | |
| 7274 return (fptr_T)NULL; | |
| 7275 } | |
| 7276 | |
| 7277 static fptr_T | |
| 7278 do_Lower(d, c) | |
| 7279 int *d; | |
| 7280 int c; | |
| 7281 { | |
| 7282 *d = MB_TOLOWER(c); | |
| 7283 | |
| 7284 return (fptr_T)do_Lower; | |
| 7 | 7285 } |
| 7286 | |
| 7287 /* | |
| 7288 * regtilde(): Replace tildes in the pattern by the old pattern. | |
| 7289 * | |
| 7290 * Short explanation of the tilde: It stands for the previous replacement | |
| 7291 * pattern. If that previous pattern also contains a ~ we should go back a | |
| 7292 * step further... But we insert the previous pattern into the current one | |
| 7293 * and remember that. | |
| 772 | 7294 * This still does not handle the case where "magic" changes. So require the |
| 7295 * user to keep his hands off of "magic". | |
| 7 | 7296 * |
| 7297 * The tildes are parsed once before the first call to vim_regsub(). | |
| 7298 */ | |
| 7299 char_u * | |
| 7300 regtilde(source, magic) | |
| 7301 char_u *source; | |
| 7302 int magic; | |
| 7303 { | |
| 7304 char_u *newsub = source; | |
| 7305 char_u *tmpsub; | |
| 7306 char_u *p; | |
| 7307 int len; | |
| 7308 int prevlen; | |
| 7309 | |
| 7310 for (p = newsub; *p; ++p) | |
| 7311 { | |
| 7312 if ((*p == '~' && magic) || (*p == '\\' && *(p + 1) == '~' && !magic)) | |
| 7313 { | |
| 7314 if (reg_prev_sub != NULL) | |
| 7315 { | |
| 7316 /* length = len(newsub) - 1 + len(prev_sub) + 1 */ | |
| 7317 prevlen = (int)STRLEN(reg_prev_sub); | |
| 7318 tmpsub = alloc((unsigned)(STRLEN(newsub) + prevlen)); | |
| 7319 if (tmpsub != NULL) | |
| 7320 { | |
| 7321 /* copy prefix */ | |
| 7322 len = (int)(p - newsub); /* not including ~ */ | |
| 7323 mch_memmove(tmpsub, newsub, (size_t)len); | |
| 1209 | 7324 /* interpret tilde */ |
| 7 | 7325 mch_memmove(tmpsub + len, reg_prev_sub, (size_t)prevlen); |
| 7326 /* copy postfix */ | |
| 7327 if (!magic) | |
| 7328 ++p; /* back off \ */ | |
| 7329 STRCPY(tmpsub + len + prevlen, p + 1); | |
| 7330 | |
| 7331 if (newsub != source) /* already allocated newsub */ | |
| 7332 vim_free(newsub); | |
| 7333 newsub = tmpsub; | |
| 7334 p = newsub + len + prevlen; | |
| 7335 } | |
| 7336 } | |
| 7337 else if (magic) | |
| 1621 | 7338 STRMOVE(p, p + 1); /* remove '~' */ |
| 7 | 7339 else |
| 1621 | 7340 STRMOVE(p, p + 2); /* remove '\~' */ |
| 7 | 7341 --p; |
| 7342 } | |
| 7343 else | |
| 7344 { | |
| 7345 if (*p == '\\' && p[1]) /* skip escaped characters */ | |
| 7346 ++p; | |
| 7347 #ifdef FEAT_MBYTE | |
| 7348 if (has_mbyte) | |
| 474 | 7349 p += (*mb_ptr2len)(p) - 1; |
| 7 | 7350 #endif |
| 7351 } | |
| 7352 } | |
| 7353 | |
| 7354 vim_free(reg_prev_sub); | |
| 7355 if (newsub != source) /* newsub was allocated, just keep it */ | |
| 7356 reg_prev_sub = newsub; | |
| 7357 else /* no ~ found, need to save newsub */ | |
| 7358 reg_prev_sub = vim_strsave(newsub); | |
| 7359 return newsub; | |
| 7360 } | |
| 7361 | |
| 7362 #ifdef FEAT_EVAL | |
| 7363 static int can_f_submatch = FALSE; /* TRUE when submatch() can be used */ | |
| 7364 | |
| 7365 /* These pointers are used instead of reg_match and reg_mmatch for | |
| 7366 * reg_submatch(). Needed for when the substitution string is an expression | |
| 7367 * that contains a call to substitute() and submatch(). */ | |
| 7368 static regmatch_T *submatch_match; | |
| 7369 static regmmatch_T *submatch_mmatch; | |
| 2011 | 7370 static linenr_T submatch_firstlnum; |
| 7371 static linenr_T submatch_maxline; | |
| 2904 | 7372 static int submatch_line_lbr; |
| 7 | 7373 #endif |
| 7374 | |
| 7375 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO) | |
| 7376 /* | |
| 7377 * vim_regsub() - perform substitutions after a vim_regexec() or | |
| 7378 * vim_regexec_multi() match. | |
| 7379 * | |
| 7380 * If "copy" is TRUE really copy into "dest". | |
| 7381 * If "copy" is FALSE nothing is copied, this is just to find out the length | |
| 7382 * of the result. | |
| 7383 * | |
| 7384 * If "backslash" is TRUE, a backslash will be removed later, need to double | |
| 7385 * them to keep them, and insert a backslash before a CR to avoid it being | |
| 7386 * replaced with a line break later. | |
| 7387 * | |
| 7388 * Note: The matched text must not change between the call of | |
| 7389 * vim_regexec()/vim_regexec_multi() and vim_regsub()! It would make the back | |
| 7390 * references invalid! | |
| 7391 * | |
| 7392 * Returns the size of the replacement, including terminating NUL. | |
| 7393 */ | |
| 7394 int | |
| 7395 vim_regsub(rmp, source, dest, copy, magic, backslash) | |
| 7396 regmatch_T *rmp; | |
| 7397 char_u *source; | |
| 7398 char_u *dest; | |
| 7399 int copy; | |
| 7400 int magic; | |
| 7401 int backslash; | |
| 7402 { | |
| 7403 reg_match = rmp; | |
| 7404 reg_mmatch = NULL; | |
| 7405 reg_maxline = 0; | |
| 4061 | 7406 reg_buf = curbuf; |
| 5836 | 7407 reg_line_lbr = TRUE; |
| 7 | 7408 return vim_regsub_both(source, dest, copy, magic, backslash); |
| 7409 } | |
| 7410 #endif | |
| 7411 | |
| 7412 int | |
| 7413 vim_regsub_multi(rmp, lnum, source, dest, copy, magic, backslash) | |
| 7414 regmmatch_T *rmp; | |
| 7415 linenr_T lnum; | |
| 7416 char_u *source; | |
| 7417 char_u *dest; | |
| 7418 int copy; | |
| 7419 int magic; | |
| 7420 int backslash; | |
| 7421 { | |
| 7422 reg_match = NULL; | |
| 7423 reg_mmatch = rmp; | |
| 7424 reg_buf = curbuf; /* always works on the current buffer! */ | |
| 7425 reg_firstlnum = lnum; | |
| 7426 reg_maxline = curbuf->b_ml.ml_line_count - lnum; | |
| 5836 | 7427 reg_line_lbr = FALSE; |
| 7 | 7428 return vim_regsub_both(source, dest, copy, magic, backslash); |
| 7429 } | |
| 7430 | |
| 7431 static int | |
| 7432 vim_regsub_both(source, dest, copy, magic, backslash) | |
| 7433 char_u *source; | |
| 7434 char_u *dest; | |
| 7435 int copy; | |
| 7436 int magic; | |
| 7437 int backslash; | |
| 7438 { | |
| 7439 char_u *src; | |
| 7440 char_u *dst; | |
| 7441 char_u *s; | |
| 7442 int c; | |
| 772 | 7443 int cc; |
| 7 | 7444 int no = -1; |
| 4244 | 7445 fptr_T func_all = (fptr_T)NULL; |
| 7446 fptr_T func_one = (fptr_T)NULL; | |
| 7 | 7447 linenr_T clnum = 0; /* init for GCC */ |
| 7448 int len = 0; /* init for GCC */ | |
| 7449 #ifdef FEAT_EVAL | |
| 7450 static char_u *eval_result = NULL; | |
| 7451 #endif | |
| 7452 | |
| 7453 /* Be paranoid... */ | |
| 7454 if (source == NULL || dest == NULL) | |
| 7455 { | |
| 7456 EMSG(_(e_null)); | |
| 7457 return 0; | |
| 7458 } | |
| 7459 if (prog_magic_wrong()) | |
| 7460 return 0; | |
| 7461 src = source; | |
| 7462 dst = dest; | |
| 7463 | |
| 7464 /* | |
| 7465 * When the substitute part starts with "\=" evaluate it as an expression. | |
| 7466 */ | |
| 7467 if (source[0] == '\\' && source[1] == '=' | |
| 7468 #ifdef FEAT_EVAL | |
| 7469 && !can_f_submatch /* can't do this recursively */ | |
| 7470 #endif | |
| 7471 ) | |
| 7472 { | |
| 7473 #ifdef FEAT_EVAL | |
| 7474 /* To make sure that the length doesn't change between checking the | |
| 7475 * length and copying the string, and to speed up things, the | |
| 7476 * resulting string is saved from the call with "copy" == FALSE to the | |
| 7477 * call with "copy" == TRUE. */ | |
| 7478 if (copy) | |
| 7479 { | |
| 7480 if (eval_result != NULL) | |
| 7481 { | |
| 7482 STRCPY(dest, eval_result); | |
| 7483 dst += STRLEN(eval_result); | |
| 7484 vim_free(eval_result); | |
| 7485 eval_result = NULL; | |
| 7486 } | |
| 7487 } | |
| 7488 else | |
| 7489 { | |
| 7490 win_T *save_reg_win; | |
| 7491 int save_ireg_ic; | |
| 7492 | |
| 7493 vim_free(eval_result); | |
| 7494 | |
| 7495 /* The expression may contain substitute(), which calls us | |
| 7496 * recursively. Make sure submatch() gets the text from the first | |
| 7497 * level. Don't need to save "reg_buf", because | |
| 7498 * vim_regexec_multi() can't be called recursively. */ | |
| 7499 submatch_match = reg_match; | |
| 7500 submatch_mmatch = reg_mmatch; | |
| 2011 | 7501 submatch_firstlnum = reg_firstlnum; |
| 7502 submatch_maxline = reg_maxline; | |
| 2904 | 7503 submatch_line_lbr = reg_line_lbr; |
| 7 | 7504 save_reg_win = reg_win; |
| 7505 save_ireg_ic = ireg_ic; | |
| 7506 can_f_submatch = TRUE; | |
| 7507 | |
| 714 | 7508 eval_result = eval_to_string(source + 2, NULL, TRUE); |
| 7 | 7509 if (eval_result != NULL) |
| 7510 { | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7511 int had_backslash = FALSE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7512 |
| 39 | 7513 for (s = eval_result; *s != NUL; mb_ptr_adv(s)) |
| 7 | 7514 { |
| 2904 | 7515 /* Change NL to CR, so that it becomes a line break, |
| 7516 * unless called from vim_regexec_nl(). | |
| 7 | 7517 * Skip over a backslashed character. */ |
| 2904 | 7518 if (*s == NL && !submatch_line_lbr) |
| 7 | 7519 *s = CAR; |
| 7520 else if (*s == '\\' && s[1] != NUL) | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7521 { |
| 7 | 7522 ++s; |
| 2173 | 7523 /* Change NL to CR here too, so that this works: |
| 7524 * :s/abc\\\ndef/\="aaa\\\nbbb"/ on text: | |
| 7525 * abc\ | |
| 7526 * def | |
| 2904 | 7527 * Not when called from vim_regexec_nl(). |
| 2173 | 7528 */ |
| 2904 | 7529 if (*s == NL && !submatch_line_lbr) |
| 2173 | 7530 *s = CAR; |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7531 had_backslash = TRUE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7532 } |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7533 } |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7534 if (had_backslash && backslash) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7535 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7536 /* Backslashes will be consumed, need to double them. */ |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7537 s = vim_strsave_escaped(eval_result, (char_u *)"\\"); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7538 if (s != NULL) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7539 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7540 vim_free(eval_result); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7541 eval_result = s; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7542 } |
| 7 | 7543 } |
| 7544 | |
| 7545 dst += STRLEN(eval_result); | |
| 7546 } | |
| 7547 | |
| 7548 reg_match = submatch_match; | |
| 7549 reg_mmatch = submatch_mmatch; | |
| 2011 | 7550 reg_firstlnum = submatch_firstlnum; |
| 7551 reg_maxline = submatch_maxline; | |
| 2904 | 7552 reg_line_lbr = submatch_line_lbr; |
| 7 | 7553 reg_win = save_reg_win; |
| 7554 ireg_ic = save_ireg_ic; | |
| 7555 can_f_submatch = FALSE; | |
| 7556 } | |
| 7557 #endif | |
| 7558 } | |
| 7559 else | |
| 7560 while ((c = *src++) != NUL) | |
| 7561 { | |
| 7562 if (c == '&' && magic) | |
| 7563 no = 0; | |
| 7564 else if (c == '\\' && *src != NUL) | |
| 7565 { | |
| 7566 if (*src == '&' && !magic) | |
| 7567 { | |
| 7568 ++src; | |
| 7569 no = 0; | |
| 7570 } | |
| 7571 else if ('0' <= *src && *src <= '9') | |
| 7572 { | |
| 7573 no = *src++ - '0'; | |
| 7574 } | |
| 7575 else if (vim_strchr((char_u *)"uUlLeE", *src)) | |
| 7576 { | |
| 7577 switch (*src++) | |
| 7578 { | |
| 4244 | 7579 case 'u': func_one = (fptr_T)do_upper; |
| 7 | 7580 continue; |
| 4244 | 7581 case 'U': func_all = (fptr_T)do_Upper; |
| 7 | 7582 continue; |
| 4244 | 7583 case 'l': func_one = (fptr_T)do_lower; |
| 7 | 7584 continue; |
| 4244 | 7585 case 'L': func_all = (fptr_T)do_Lower; |
| 7 | 7586 continue; |
| 7587 case 'e': | |
| 4244 | 7588 case 'E': func_one = func_all = (fptr_T)NULL; |
| 7 | 7589 continue; |
| 7590 } | |
| 7591 } | |
| 7592 } | |
| 7593 if (no < 0) /* Ordinary character. */ | |
| 7594 { | |
| 798 | 7595 if (c == K_SPECIAL && src[0] != NUL && src[1] != NUL) |
| 7596 { | |
| 1209 | 7597 /* Copy a special key as-is. */ |
| 798 | 7598 if (copy) |
| 7599 { | |
| 7600 *dst++ = c; | |
| 7601 *dst++ = *src++; | |
| 7602 *dst++ = *src++; | |
| 7603 } | |
| 7604 else | |
| 7605 { | |
| 7606 dst += 3; | |
| 7607 src += 2; | |
| 7608 } | |
| 7609 continue; | |
| 7610 } | |
| 7611 | |
| 7 | 7612 if (c == '\\' && *src != NUL) |
| 7613 { | |
| 7614 /* Check for abbreviations -- webb */ | |
| 7615 switch (*src) | |
| 7616 { | |
| 7617 case 'r': c = CAR; ++src; break; | |
| 7618 case 'n': c = NL; ++src; break; | |
| 7619 case 't': c = TAB; ++src; break; | |
| 7620 /* Oh no! \e already has meaning in subst pat :-( */ | |
| 7621 /* case 'e': c = ESC; ++src; break; */ | |
| 7622 case 'b': c = Ctrl_H; ++src; break; | |
| 7623 | |
| 7624 /* If "backslash" is TRUE the backslash will be removed | |
| 7625 * later. Used to insert a literal CR. */ | |
| 7626 default: if (backslash) | |
| 7627 { | |
| 7628 if (copy) | |
| 7629 *dst = '\\'; | |
| 7630 ++dst; | |
| 7631 } | |
| 7632 c = *src++; | |
| 7633 } | |
| 7634 } | |
| 798 | 7635 #ifdef FEAT_MBYTE |
| 7636 else if (has_mbyte) | |
| 7637 c = mb_ptr2char(src - 1); | |
| 7638 #endif | |
| 7 | 7639 |
| 7640 /* Write to buffer, if copy is set. */ | |
| 4244 | 7641 if (func_one != (fptr_T)NULL) |
| 7642 /* Turbo C complains without the typecast */ | |
| 7643 func_one = (fptr_T)(func_one(&cc, c)); | |
| 7644 else if (func_all != (fptr_T)NULL) | |
| 7645 /* Turbo C complains without the typecast */ | |
| 7646 func_all = (fptr_T)(func_all(&cc, c)); | |
| 7647 else /* just copy */ | |
| 772 | 7648 cc = c; |
| 7649 | |
| 7650 #ifdef FEAT_MBYTE | |
| 7651 if (has_mbyte) | |
| 7 | 7652 { |
|
2307
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7653 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
|
7654 |
| 7 | 7655 if (copy) |
| 772 | 7656 mb_char2bytes(cc, dst); |
| 7657 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
|
7658 if (enc_utf8) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7659 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7660 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
|
7661 |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7662 /* 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
|
7663 * 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
|
7664 if (clen < totlen) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7665 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7666 if (copy) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7667 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
|
7668 (size_t)(totlen - clen)); |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7669 dst += totlen - clen; |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7670 } |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7671 } |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7672 src += totlen - 1; |
| 7 | 7673 } |
| 7674 else | |
| 7675 #endif | |
| 7676 if (copy) | |
| 772 | 7677 *dst = cc; |
| 7 | 7678 dst++; |
| 7679 } | |
| 7680 else | |
| 7681 { | |
| 7682 if (REG_MULTI) | |
| 7683 { | |
| 7684 clnum = reg_mmatch->startpos[no].lnum; | |
| 7685 if (clnum < 0 || reg_mmatch->endpos[no].lnum < 0) | |
| 7686 s = NULL; | |
| 7687 else | |
| 7688 { | |
| 7689 s = reg_getline(clnum) + reg_mmatch->startpos[no].col; | |
| 7690 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7691 len = reg_mmatch->endpos[no].col | |
| 7692 - reg_mmatch->startpos[no].col; | |
| 7693 else | |
| 7694 len = (int)STRLEN(s); | |
| 7695 } | |
| 7696 } | |
| 7697 else | |
| 7698 { | |
| 7699 s = reg_match->startp[no]; | |
| 7700 if (reg_match->endp[no] == NULL) | |
| 7701 s = NULL; | |
| 7702 else | |
| 7703 len = (int)(reg_match->endp[no] - s); | |
| 7704 } | |
| 7705 if (s != NULL) | |
| 7706 { | |
| 7707 for (;;) | |
| 7708 { | |
| 7709 if (len == 0) | |
| 7710 { | |
| 7711 if (REG_MULTI) | |
| 7712 { | |
| 7713 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7714 break; | |
| 7715 if (copy) | |
| 7716 *dst = CAR; | |
| 7717 ++dst; | |
| 7718 s = reg_getline(++clnum); | |
| 7719 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7720 len = reg_mmatch->endpos[no].col; | |
| 7721 else | |
| 7722 len = (int)STRLEN(s); | |
| 7723 } | |
| 7724 else | |
| 7725 break; | |
| 7726 } | |
| 7727 else if (*s == NUL) /* we hit NUL. */ | |
| 7728 { | |
| 7729 if (copy) | |
| 7730 EMSG(_(e_re_damg)); | |
| 7731 goto exit; | |
| 7732 } | |
| 7733 else | |
| 7734 { | |
| 7735 if (backslash && (*s == CAR || *s == '\\')) | |
| 7736 { | |
| 7737 /* | |
| 7738 * Insert a backslash in front of a CR, otherwise | |
| 7739 * it will be replaced by a line break. | |
| 7740 * Number of backslashes will be halved later, | |
| 7741 * double them here. | |
| 7742 */ | |
| 7743 if (copy) | |
| 7744 { | |
| 7745 dst[0] = '\\'; | |
| 7746 dst[1] = *s; | |
| 7747 } | |
| 7748 dst += 2; | |
| 7749 } | |
| 7750 else | |
| 7751 { | |
| 772 | 7752 #ifdef FEAT_MBYTE |
| 7753 if (has_mbyte) | |
| 7754 c = mb_ptr2char(s); | |
| 7755 else | |
| 7756 #endif | |
| 7757 c = *s; | |
| 7758 | |
| 4244 | 7759 if (func_one != (fptr_T)NULL) |
| 7760 /* Turbo C complains without the typecast */ | |
| 7761 func_one = (fptr_T)(func_one(&cc, c)); | |
| 7762 else if (func_all != (fptr_T)NULL) | |
| 7763 /* Turbo C complains without the typecast */ | |
| 7764 func_all = (fptr_T)(func_all(&cc, c)); | |
| 7765 else /* just copy */ | |
| 772 | 7766 cc = c; |
| 7767 | |
| 7768 #ifdef FEAT_MBYTE | |
| 7769 if (has_mbyte) | |
| 7 | 7770 { |
| 1332 | 7771 int l; |
| 7772 | |
| 7773 /* Copy composing characters separately, one | |
| 7774 * at a time. */ | |
| 7775 if (enc_utf8) | |
| 7776 l = utf_ptr2len(s) - 1; | |
| 7777 else | |
| 7778 l = mb_ptr2len(s) - 1; | |
| 772 | 7779 |
| 7780 s += l; | |
| 7781 len -= l; | |
| 7782 if (copy) | |
| 7783 mb_char2bytes(cc, dst); | |
| 7784 dst += mb_char2len(cc) - 1; | |
| 7 | 7785 } |
| 772 | 7786 else |
| 7787 #endif | |
| 7788 if (copy) | |
| 7789 *dst = cc; | |
| 7790 dst++; | |
| 7 | 7791 } |
| 772 | 7792 |
| 7 | 7793 ++s; |
| 7794 --len; | |
| 7795 } | |
| 7796 } | |
| 7797 } | |
| 7798 no = -1; | |
| 7799 } | |
| 7800 } | |
| 7801 if (copy) | |
| 7802 *dst = NUL; | |
| 7803 | |
| 7804 exit: | |
| 7805 return (int)((dst - dest) + 1); | |
| 7806 } | |
| 7807 | |
| 7808 #ifdef FEAT_EVAL | |
| 2012 | 7809 static char_u *reg_getline_submatch __ARGS((linenr_T lnum)); |
| 7810 | |
| 7 | 7811 /* |
| 2011 | 7812 * Call reg_getline() with the line numbers from the submatch. If a |
| 7813 * substitute() was used the reg_maxline and other values have been | |
| 7814 * overwritten. | |
| 7815 */ | |
| 7816 static char_u * | |
| 7817 reg_getline_submatch(lnum) | |
| 7818 linenr_T lnum; | |
| 7819 { | |
| 7820 char_u *s; | |
| 7821 linenr_T save_first = reg_firstlnum; | |
| 7822 linenr_T save_max = reg_maxline; | |
| 7823 | |
| 7824 reg_firstlnum = submatch_firstlnum; | |
| 7825 reg_maxline = submatch_maxline; | |
| 7826 | |
| 7827 s = reg_getline(lnum); | |
| 7828 | |
| 7829 reg_firstlnum = save_first; | |
| 7830 reg_maxline = save_max; | |
| 7831 return s; | |
| 7832 } | |
| 7833 | |
| 7834 /* | |
| 1209 | 7835 * Used for the submatch() function: get the string from the n'th submatch in |
| 7 | 7836 * allocated memory. |
| 7837 * Returns NULL when not in a ":s" command and for a non-existing submatch. | |
| 7838 */ | |
| 7839 char_u * | |
| 7840 reg_submatch(no) | |
| 7841 int no; | |
| 7842 { | |
| 7843 char_u *retval = NULL; | |
| 7844 char_u *s; | |
| 7845 int len; | |
| 7846 int round; | |
| 7847 linenr_T lnum; | |
| 7848 | |
| 840 | 7849 if (!can_f_submatch || no < 0) |
| 7 | 7850 return NULL; |
| 7851 | |
| 7852 if (submatch_match == NULL) | |
| 7853 { | |
| 7854 /* | |
| 7855 * First round: compute the length and allocate memory. | |
| 7856 * Second round: copy the text. | |
| 7857 */ | |
| 7858 for (round = 1; round <= 2; ++round) | |
| 7859 { | |
| 7860 lnum = submatch_mmatch->startpos[no].lnum; | |
| 7861 if (lnum < 0 || submatch_mmatch->endpos[no].lnum < 0) | |
| 7862 return NULL; | |
| 7863 | |
| 2011 | 7864 s = reg_getline_submatch(lnum) + submatch_mmatch->startpos[no].col; |
| 7 | 7865 if (s == NULL) /* anti-crash check, cannot happen? */ |
| 7866 break; | |
| 7867 if (submatch_mmatch->endpos[no].lnum == lnum) | |
| 7868 { | |
| 7869 /* Within one line: take form start to end col. */ | |
| 7870 len = submatch_mmatch->endpos[no].col | |
| 7871 - submatch_mmatch->startpos[no].col; | |
| 7872 if (round == 2) | |
| 418 | 7873 vim_strncpy(retval, s, len); |
| 7 | 7874 ++len; |
| 7875 } | |
| 7876 else | |
| 7877 { | |
| 7878 /* Multiple lines: take start line from start col, middle | |
| 7879 * lines completely and end line up to end col. */ | |
| 7880 len = (int)STRLEN(s); | |
| 7881 if (round == 2) | |
| 7882 { | |
| 7883 STRCPY(retval, s); | |
| 7884 retval[len] = '\n'; | |
| 7885 } | |
| 7886 ++len; | |
| 7887 ++lnum; | |
| 7888 while (lnum < submatch_mmatch->endpos[no].lnum) | |
| 7889 { | |
| 2011 | 7890 s = reg_getline_submatch(lnum++); |
| 7 | 7891 if (round == 2) |
| 7892 STRCPY(retval + len, s); | |
| 7893 len += (int)STRLEN(s); | |
| 7894 if (round == 2) | |
| 7895 retval[len] = '\n'; | |
| 7896 ++len; | |
| 7897 } | |
| 7898 if (round == 2) | |
| 2011 | 7899 STRNCPY(retval + len, reg_getline_submatch(lnum), |
| 7 | 7900 submatch_mmatch->endpos[no].col); |
| 7901 len += submatch_mmatch->endpos[no].col; | |
| 7902 if (round == 2) | |
| 7903 retval[len] = NUL; | |
| 7904 ++len; | |
| 7905 } | |
| 7906 | |
| 840 | 7907 if (retval == NULL) |
| 7 | 7908 { |
| 7909 retval = lalloc((long_u)len, TRUE); | |
| 840 | 7910 if (retval == NULL) |
| 7 | 7911 return NULL; |
| 7912 } | |
| 7913 } | |
| 7914 } | |
| 7915 else | |
| 7916 { | |
| 1815 | 7917 s = submatch_match->startp[no]; |
| 7918 if (s == NULL || submatch_match->endp[no] == NULL) | |
| 7 | 7919 retval = NULL; |
| 7920 else | |
| 7921 retval = vim_strnsave(s, (int)(submatch_match->endp[no] - s)); | |
| 7922 } | |
| 7923 | |
| 7924 return retval; | |
| 7925 } | |
| 5794 | 7926 |
| 7927 /* | |
| 7928 * Used for the submatch() function with the optional non-zero argument: get | |
| 7929 * the list of strings from the n'th submatch in allocated memory with NULs | |
| 7930 * represented in NLs. | |
| 7931 * Returns a list of allocated strings. Returns NULL when not in a ":s" | |
| 7932 * command, for a non-existing submatch and for any error. | |
| 7933 */ | |
| 7934 list_T * | |
| 7935 reg_submatch_list(no) | |
| 7936 int no; | |
| 7937 { | |
| 7938 char_u *s; | |
| 7939 linenr_T slnum; | |
| 7940 linenr_T elnum; | |
| 7941 colnr_T scol; | |
| 7942 colnr_T ecol; | |
| 7943 int i; | |
| 7944 list_T *list; | |
| 7945 int error = FALSE; | |
| 7946 | |
| 7947 if (!can_f_submatch || no < 0) | |
| 7948 return NULL; | |
| 7949 | |
| 7950 if (submatch_match == NULL) | |
| 7951 { | |
| 7952 slnum = submatch_mmatch->startpos[no].lnum; | |
| 7953 elnum = submatch_mmatch->endpos[no].lnum; | |
| 7954 if (slnum < 0 || elnum < 0) | |
| 7955 return NULL; | |
| 7956 | |
| 7957 scol = submatch_mmatch->startpos[no].col; | |
| 7958 ecol = submatch_mmatch->endpos[no].col; | |
| 7959 | |
| 7960 list = list_alloc(); | |
| 7961 if (list == NULL) | |
| 7962 return NULL; | |
| 7963 | |
| 7964 s = reg_getline_submatch(slnum) + scol; | |
| 7965 if (slnum == elnum) | |
| 7966 { | |
| 7967 if (list_append_string(list, s, ecol - scol) == FAIL) | |
| 7968 error = TRUE; | |
| 7969 } | |
| 7970 else | |
| 7971 { | |
| 7972 if (list_append_string(list, s, -1) == FAIL) | |
| 7973 error = TRUE; | |
| 7974 for (i = 1; i < elnum - slnum; i++) | |
| 7975 { | |
| 7976 s = reg_getline_submatch(slnum + i); | |
| 7977 if (list_append_string(list, s, -1) == FAIL) | |
| 7978 error = TRUE; | |
| 7979 } | |
| 7980 s = reg_getline_submatch(elnum); | |
| 7981 if (list_append_string(list, s, ecol) == FAIL) | |
| 7982 error = TRUE; | |
| 7983 } | |
| 7984 } | |
| 7985 else | |
| 7986 { | |
| 7987 s = submatch_match->startp[no]; | |
| 7988 if (s == NULL || submatch_match->endp[no] == NULL) | |
| 7989 return NULL; | |
| 7990 list = list_alloc(); | |
| 7991 if (list == NULL) | |
| 7992 return NULL; | |
| 7993 if (list_append_string(list, s, | |
| 7994 (int)(submatch_match->endp[no] - s)) == FAIL) | |
| 7995 error = TRUE; | |
| 7996 } | |
| 7997 | |
| 7998 if (error) | |
| 7999 { | |
| 8000 list_free(list, TRUE); | |
| 8001 return NULL; | |
| 8002 } | |
| 8003 return list; | |
| 8004 } | |
| 7 | 8005 #endif |
| 4444 | 8006 |
| 8007 static regengine_T bt_regengine = | |
| 8008 { | |
| 8009 bt_regcomp, | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8010 bt_regfree, |
| 4444 | 8011 bt_regexec_nl, |
| 6328 | 8012 bt_regexec_multi, |
| 8013 (char_u *)"" | |
| 4444 | 8014 }; |
| 8015 | |
| 8016 #include "regexp_nfa.c" | |
| 8017 | |
| 8018 static regengine_T nfa_regengine = | |
| 8019 { | |
| 8020 nfa_regcomp, | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8021 nfa_regfree, |
| 4444 | 8022 nfa_regexec_nl, |
| 6328 | 8023 nfa_regexec_multi, |
| 8024 (char_u *)"" | |
| 4444 | 8025 }; |
| 8026 | |
| 8027 /* Which regexp engine to use? Needed for vim_regcomp(). | |
| 8028 * Must match with 'regexpengine'. */ | |
| 8029 static int regexp_engine = 0; | |
| 6328 | 8030 |
| 4444 | 8031 #ifdef DEBUG |
| 8032 static char_u regname[][30] = { | |
| 8033 "AUTOMATIC Regexp Engine", | |
|
4579
7a2be4a39423
updated for version 7.3.1037
Bram Moolenaar <bram@vim.org>
parents:
4505
diff
changeset
|
8034 "BACKTRACKING Regexp Engine", |
| 4444 | 8035 "NFA Regexp Engine" |
| 8036 }; | |
| 8037 #endif | |
| 8038 | |
| 8039 /* | |
| 8040 * Compile a regular expression into internal code. | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8041 * Returns the program in allocated memory. |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8042 * Use vim_regfree() to free the memory. |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8043 * Returns NULL for an error. |
| 4444 | 8044 */ |
| 8045 regprog_T * | |
| 8046 vim_regcomp(expr_arg, re_flags) | |
| 8047 char_u *expr_arg; | |
| 8048 int re_flags; | |
| 8049 { | |
| 8050 regprog_T *prog = NULL; | |
| 8051 char_u *expr = expr_arg; | |
| 8052 | |
| 8053 regexp_engine = p_re; | |
| 8054 | |
| 8055 /* Check for prefix "\%#=", that sets the regexp engine */ | |
| 8056 if (STRNCMP(expr, "\\%#=", 4) == 0) | |
| 8057 { | |
| 8058 int newengine = expr[4] - '0'; | |
| 8059 | |
| 8060 if (newengine == AUTOMATIC_ENGINE | |
| 8061 || newengine == BACKTRACKING_ENGINE | |
| 8062 || newengine == NFA_ENGINE) | |
| 8063 { | |
| 8064 regexp_engine = expr[4] - '0'; | |
| 8065 expr += 5; | |
| 8066 #ifdef DEBUG | |
| 5897 | 8067 smsg((char_u *)"New regexp mode selected (%d): %s", |
| 8068 regexp_engine, regname[newengine]); | |
| 4444 | 8069 #endif |
| 8070 } | |
| 8071 else | |
| 8072 { | |
| 8073 EMSG(_("E864: \\%#= can only be followed by 0, 1, or 2. The automatic engine will be used ")); | |
| 8074 regexp_engine = AUTOMATIC_ENGINE; | |
| 8075 } | |
| 8076 } | |
| 8077 bt_regengine.expr = expr; | |
| 8078 nfa_regengine.expr = expr; | |
| 8079 | |
| 8080 /* | |
| 8081 * First try the NFA engine, unless backtracking was requested. | |
| 8082 */ | |
| 8083 if (regexp_engine != BACKTRACKING_ENGINE) | |
| 6533 | 8084 prog = nfa_regengine.regcomp(expr, |
| 8085 re_flags + (regexp_engine == AUTOMATIC_ENGINE ? RE_AUTO : 0)); | |
| 4444 | 8086 else |
| 8087 prog = bt_regengine.regcomp(expr, re_flags); | |
| 8088 | |
| 6328 | 8089 /* Check for error compiling regexp with initial engine. */ |
| 8090 if (prog == NULL) | |
| 4444 | 8091 { |
| 4460 | 8092 #ifdef BT_REGEXP_DEBUG_LOG |
| 4444 | 8093 if (regexp_engine != BACKTRACKING_ENGINE) /* debugging log for NFA */ |
| 8094 { | |
| 8095 FILE *f; | |
| 4460 | 8096 f = fopen(BT_REGEXP_DEBUG_LOG_NAME, "a"); |
| 4444 | 8097 if (f) |
| 8098 { | |
|
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8099 fprintf(f, "Syntax error in \"%s\"\n", expr); |
| 4444 | 8100 fclose(f); |
| 8101 } | |
| 8102 else | |
| 4460 | 8103 EMSG2("(NFA) Could not open \"%s\" to write !!!", |
| 8104 BT_REGEXP_DEBUG_LOG_NAME); | |
| 4444 | 8105 } |
| 8106 #endif | |
| 8107 /* | |
| 6328 | 8108 * If the NFA engine failed, try the backtracking engine. |
| 6533 | 8109 * The NFA engine also fails for patterns that it can't handle well |
| 8110 * but are still valid patterns, thus a retry should work. | |
| 8111 */ | |
| 4444 | 8112 if (regexp_engine == AUTOMATIC_ENGINE) |
| 6328 | 8113 { |
| 6533 | 8114 regexp_engine = BACKTRACKING_ENGINE; |
|
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8115 prog = bt_regengine.regcomp(expr, re_flags); |
| 6328 | 8116 } |
|
4762
47906f888725
updated for version 7.3.1128
Bram Moolenaar <bram@vim.org>
parents:
4760
diff
changeset
|
8117 } |
| 4444 | 8118 |
| 6328 | 8119 if (prog != NULL) |
| 8120 { | |
| 8121 /* Store the info needed to call regcomp() again when the engine turns | |
| 8122 * out to be very slow when executing it. */ | |
| 8123 prog->re_engine = regexp_engine; | |
| 8124 prog->re_flags = re_flags; | |
| 8125 } | |
| 8126 | |
| 4444 | 8127 return prog; |
| 8128 } | |
| 8129 | |
| 8130 /* | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8131 * 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
|
8132 */ |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8133 void |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8134 vim_regfree(prog) |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8135 regprog_T *prog; |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8136 { |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8137 if (prog != NULL) |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8138 prog->engine->regfree(prog); |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8139 } |
|
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8140 |
| 6328 | 8141 #ifdef FEAT_EVAL |
| 8142 static void report_re_switch __ARGS((char_u *pat)); | |
| 8143 | |
| 8144 static void | |
| 8145 report_re_switch(pat) | |
| 8146 char_u *pat; | |
| 8147 { | |
| 8148 if (p_verbose > 0) | |
| 8149 { | |
| 8150 verbose_enter(); | |
| 8151 MSG_PUTS(_("Switching to backtracking RE engine for pattern: ")); | |
| 8152 MSG_PUTS(pat); | |
| 8153 verbose_leave(); | |
| 8154 } | |
| 8155 } | |
| 8156 #endif | |
| 8157 | |
| 8158 static int vim_regexec_both __ARGS((regmatch_T *rmp, char_u *line, colnr_T col, int nl)); | |
| 8159 | |
|
4805
66803af09906
updated for version 7.3.1149
Bram Moolenaar <bram@vim.org>
parents:
4770
diff
changeset
|
8160 /* |
| 4444 | 8161 * Match a regexp against a string. |
| 8162 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 6375 | 8163 * Note: "rmp->regprog" may be freed and changed. |
| 4444 | 8164 * Uses curbuf for line count and 'iskeyword'. |
| 6328 | 8165 * When "nl" is TRUE consider a "\n" in "line" to be a line break. |
| 4444 | 8166 * |
| 8167 * Return TRUE if there is a match, FALSE if not. | |
| 8168 */ | |
| 6328 | 8169 static int |
| 8170 vim_regexec_both(rmp, line, col, nl) | |
| 8171 regmatch_T *rmp; | |
| 8172 char_u *line; /* string to match against */ | |
| 8173 colnr_T col; /* column to start looking for match */ | |
| 8174 int nl; | |
| 8175 { | |
| 8176 int result = rmp->regprog->engine->regexec_nl(rmp, line, col, nl); | |
| 8177 | |
| 8178 /* NFA engine aborted because it's very slow. */ | |
| 8179 if (rmp->regprog->re_engine == AUTOMATIC_ENGINE | |
| 8180 && result == NFA_TOO_EXPENSIVE) | |
| 8181 { | |
| 8182 int save_p_re = p_re; | |
| 8183 int re_flags = rmp->regprog->re_flags; | |
| 8184 char_u *pat = vim_strsave(((nfa_regprog_T *)rmp->regprog)->pattern); | |
| 8185 | |
| 8186 p_re = BACKTRACKING_ENGINE; | |
| 8187 vim_regfree(rmp->regprog); | |
| 8188 if (pat != NULL) | |
| 8189 { | |
| 8190 #ifdef FEAT_EVAL | |
| 8191 report_re_switch(pat); | |
| 8192 #endif | |
| 8193 rmp->regprog = vim_regcomp(pat, re_flags); | |
| 8194 if (rmp->regprog != NULL) | |
| 8195 result = rmp->regprog->engine->regexec_nl(rmp, line, col, nl); | |
| 8196 vim_free(pat); | |
| 8197 } | |
| 8198 | |
| 8199 p_re = save_p_re; | |
| 8200 } | |
| 6390 | 8201 return result > 0; |
| 6328 | 8202 } |
| 8203 | |
| 6375 | 8204 /* |
| 8205 * Note: "*prog" may be freed and changed. | |
| 6390 | 8206 * Return TRUE if there is a match, FALSE if not. |
| 6375 | 8207 */ |
| 8208 int | |
| 8209 vim_regexec_prog(prog, ignore_case, line, col) | |
| 8210 regprog_T **prog; | |
| 8211 int ignore_case; | |
| 8212 char_u *line; | |
| 8213 colnr_T col; | |
| 8214 { | |
| 8215 int r; | |
| 8216 regmatch_T regmatch; | |
| 8217 | |
| 8218 regmatch.regprog = *prog; | |
| 8219 regmatch.rm_ic = ignore_case; | |
| 8220 r = vim_regexec_both(®match, line, col, FALSE); | |
| 8221 *prog = regmatch.regprog; | |
| 8222 return r; | |
| 8223 } | |
| 8224 | |
| 8225 /* | |
| 8226 * Note: "rmp->regprog" may be freed and changed. | |
| 6390 | 8227 * Return TRUE if there is a match, FALSE if not. |
| 6375 | 8228 */ |
| 4444 | 8229 int |
| 8230 vim_regexec(rmp, line, col) | |
| 6328 | 8231 regmatch_T *rmp; |
| 8232 char_u *line; | |
| 8233 colnr_T col; | |
| 4444 | 8234 { |
| 6328 | 8235 return vim_regexec_both(rmp, line, col, FALSE); |
| 4444 | 8236 } |
| 8237 | |
| 8238 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
| 8239 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 8240 /* | |
| 8241 * Like vim_regexec(), but consider a "\n" in "line" to be a line break. | |
| 6375 | 8242 * Note: "rmp->regprog" may be freed and changed. |
| 6390 | 8243 * Return TRUE if there is a match, FALSE if not. |
| 4444 | 8244 */ |
| 8245 int | |
| 8246 vim_regexec_nl(rmp, line, col) | |
| 6328 | 8247 regmatch_T *rmp; |
| 8248 char_u *line; | |
| 8249 colnr_T col; | |
| 4444 | 8250 { |
| 6328 | 8251 return vim_regexec_both(rmp, line, col, TRUE); |
| 4444 | 8252 } |
| 8253 #endif | |
| 8254 | |
| 8255 /* | |
| 8256 * Match a regexp against multiple lines. | |
| 8257 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 6375 | 8258 * Note: "rmp->regprog" may be freed and changed. |
| 4444 | 8259 * Uses curbuf for line count and 'iskeyword'. |
| 8260 * | |
| 8261 * Return zero if there is no match. Return number of lines contained in the | |
| 8262 * match otherwise. | |
| 8263 */ | |
| 8264 long | |
| 8265 vim_regexec_multi(rmp, win, buf, lnum, col, tm) | |
| 8266 regmmatch_T *rmp; | |
| 8267 win_T *win; /* window in which to search or NULL */ | |
| 8268 buf_T *buf; /* buffer in which to search */ | |
| 8269 linenr_T lnum; /* nr of line to start looking for match */ | |
| 8270 colnr_T col; /* column to start looking for match */ | |
| 8271 proftime_T *tm; /* timeout limit or NULL */ | |
| 8272 { | |
| 6328 | 8273 int result = rmp->regprog->engine->regexec_multi( |
| 8274 rmp, win, buf, lnum, col, tm); | |
| 8275 | |
| 8276 /* NFA engine aborted because it's very slow. */ | |
| 8277 if (rmp->regprog->re_engine == AUTOMATIC_ENGINE | |
| 8278 && result == NFA_TOO_EXPENSIVE) | |
| 8279 { | |
| 8280 int save_p_re = p_re; | |
| 8281 int re_flags = rmp->regprog->re_flags; | |
| 8282 char_u *pat = vim_strsave(((nfa_regprog_T *)rmp->regprog)->pattern); | |
| 8283 | |
| 8284 p_re = BACKTRACKING_ENGINE; | |
| 8285 vim_regfree(rmp->regprog); | |
| 8286 if (pat != NULL) | |
| 8287 { | |
| 8288 #ifdef FEAT_EVAL | |
| 8289 report_re_switch(pat); | |
| 8290 #endif | |
| 8291 rmp->regprog = vim_regcomp(pat, re_flags); | |
| 8292 if (rmp->regprog != NULL) | |
| 8293 result = rmp->regprog->engine->regexec_multi( | |
| 8294 rmp, win, buf, lnum, col, tm); | |
| 8295 vim_free(pat); | |
| 8296 } | |
| 8297 p_re = save_p_re; | |
| 8298 } | |
| 8299 | |
| 6390 | 8300 return result <= 0 ? 0 : result; |
| 4444 | 8301 } |