Mercurial > vim
annotate src/regexp.c @ 4505:d037b9cbdaaa v7.3.1000
updated for version 7.3.1000
Problem: Typo in char value causes out of bounds access.
Solution: Fix character value. (Klemens Baum)
| author | Bram Moolenaar <bram@vim.org> |
|---|---|
| date | Tue, 21 May 2013 21:37:20 +0200 |
| parents | 4ff8fa416a0d |
| children | 7a2be4a39423 |
| rev | line source |
|---|---|
| 7 | 1 /* vi:set ts=8 sts=4 sw=4: |
| 2 * | |
| 3 * Handling of regular expressions: vim_regcomp(), vim_regexec(), vim_regsub() | |
| 4 * | |
| 5 * NOTICE: | |
| 6 * | |
| 7 * This is NOT the original regular expression code as written by Henry | |
| 8 * Spencer. This code has been modified specifically for use with the VIM | |
| 9 * editor, and should not be used separately from Vim. If you want a good | |
| 10 * regular expression library, get the original code. The copyright notice | |
| 11 * that follows is from the original. | |
| 12 * | |
| 13 * END NOTICE | |
| 14 * | |
| 15 * Copyright (c) 1986 by University of Toronto. | |
| 16 * Written by Henry Spencer. Not derived from licensed software. | |
| 17 * | |
| 18 * Permission is granted to anyone to use this software for any | |
| 19 * purpose on any computer system, and to redistribute it freely, | |
| 20 * subject to the following restrictions: | |
| 21 * | |
| 22 * 1. The author is not responsible for the consequences of use of | |
| 23 * this software, no matter how awful, even if they arise | |
| 24 * from defects in it. | |
| 25 * | |
| 26 * 2. The origin of this software must not be misrepresented, either | |
| 27 * by explicit claim or by omission. | |
| 28 * | |
| 29 * 3. Altered versions must be plainly marked as such, and must not | |
| 30 * be misrepresented as being the original software. | |
| 31 * | |
| 32 * Beware that some of this code is subtly aware of the way operator | |
| 33 * precedence is structured in regular expressions. Serious changes in | |
| 34 * regular-expression syntax might require a total rethink. | |
| 35 * | |
| 24 | 36 * Changes have been made by Tony Andrews, Olaf 'Rhialto' Seibert, Robert |
| 37 * Webb, Ciaran McCreesh and Bram Moolenaar. | |
| 7 | 38 * Named character class support added by Walter Briscoe (1998 Jul 01) |
| 39 */ | |
| 40 | |
| 4444 | 41 /* Uncomment the first if you do not want to see debugging logs or files |
| 42 * related to regular expressions, even when compiling with -DDEBUG. | |
| 43 * Uncomment the second to get the regexp debugging. */ | |
| 44 /* #undef DEBUG */ | |
| 45 /* #define DEBUG */ | |
| 46 | |
| 7 | 47 #include "vim.h" |
| 48 | |
| 4444 | 49 #ifdef DEBUG |
| 50 /* show/save debugging data when BT engine is used */ | |
| 51 # define BT_REGEXP_DUMP | |
| 52 /* save the debugging data to a file instead of displaying it */ | |
| 53 # define BT_REGEXP_LOG | |
| 4460 | 54 # define BT_REGEXP_DEBUG_LOG |
| 55 # define BT_REGEXP_DEBUG_LOG_NAME "bt_regexp_debug.log" | |
| 4444 | 56 #endif |
| 7 | 57 |
| 58 /* | |
| 59 * The "internal use only" fields in regexp.h are present to pass info from | |
| 60 * compile to execute that permits the execute phase to run lots faster on | |
| 61 * simple cases. They are: | |
| 62 * | |
| 63 * regstart char that must begin a match; NUL if none obvious; Can be a | |
| 64 * multi-byte character. | |
| 65 * reganch is the match anchored (at beginning-of-line only)? | |
| 66 * regmust string (pointer into program) that match must include, or NULL | |
| 67 * regmlen length of regmust string | |
| 68 * regflags RF_ values or'ed together | |
| 69 * | |
| 70 * Regstart and reganch permit very fast decisions on suitable starting points | |
| 71 * for a match, cutting down the work a lot. Regmust permits fast rejection | |
| 72 * of lines that cannot possibly match. The regmust tests are costly enough | |
| 73 * that vim_regcomp() supplies a regmust only if the r.e. contains something | |
| 74 * potentially expensive (at present, the only such thing detected is * or + | |
| 75 * at the start of the r.e., which can involve a lot of backup). Regmlen is | |
| 76 * supplied because the test in vim_regexec() needs it and vim_regcomp() is | |
| 77 * computing it anyway. | |
| 78 */ | |
| 79 | |
| 80 /* | |
| 81 * Structure for regexp "program". This is essentially a linear encoding | |
| 82 * of a nondeterministic finite-state machine (aka syntax charts or | |
| 83 * "railroad normal form" in parsing technology). Each node is an opcode | |
| 84 * plus a "next" pointer, possibly plus an operand. "Next" pointers of | |
| 85 * all nodes except BRANCH and BRACES_COMPLEX implement concatenation; a "next" | |
| 86 * pointer with a BRANCH on both ends of it is connecting two alternatives. | |
| 87 * (Here we have one of the subtle syntax dependencies: an individual BRANCH | |
| 88 * (as opposed to a collection of them) is never concatenated with anything | |
| 89 * because of operator precedence). The "next" pointer of a BRACES_COMPLEX | |
| 167 | 90 * node points to the node after the stuff to be repeated. |
| 91 * The operand of some types of node is a literal string; for others, it is a | |
| 92 * node leading into a sub-FSM. In particular, the operand of a BRANCH node | |
| 93 * is the first node of the branch. | |
| 94 * (NB this is *not* a tree structure: the tail of the branch connects to the | |
| 95 * thing following the set of BRANCHes.) | |
| 7 | 96 * |
| 97 * pattern is coded like: | |
| 98 * | |
| 99 * +-----------------+ | |
| 100 * | V | |
| 101 * <aa>\|<bb> BRANCH <aa> BRANCH <bb> --> END | |
| 102 * | ^ | ^ | |
| 103 * +------+ +----------+ | |
| 104 * | |
| 105 * | |
| 106 * +------------------+ | |
| 107 * V | | |
| 108 * <aa>* BRANCH BRANCH <aa> --> BACK BRANCH --> NOTHING --> END | |
| 109 * | | ^ ^ | |
| 110 * | +---------------+ | | |
| 111 * +---------------------------------------------+ | |
| 112 * | |
| 113 * | |
| 167 | 114 * +----------------------+ |
| 115 * V | | |
| 233 | 116 * <aa>\+ BRANCH <aa> --> BRANCH --> BACK BRANCH --> NOTHING --> END |
| 856 | 117 * | | ^ ^ |
| 118 * | +-----------+ | | |
| 179 | 119 * +--------------------------------------------------+ |
| 167 | 120 * |
| 121 * | |
| 7 | 122 * +-------------------------+ |
| 123 * V | | |
| 124 * <aa>\{} BRANCH BRACE_LIMITS --> BRACE_COMPLEX <aa> --> BACK END | |
| 125 * | | ^ | |
| 126 * | +----------------+ | |
| 127 * +-----------------------------------------------+ | |
| 128 * | |
| 129 * | |
| 130 * <aa>\@!<bb> BRANCH NOMATCH <aa> --> END <bb> --> END | |
| 131 * | | ^ ^ | |
| 132 * | +----------------+ | | |
| 133 * +--------------------------------+ | |
| 134 * | |
| 135 * +---------+ | |
| 136 * | V | |
| 137 * \z[abc] BRANCH BRANCH a BRANCH b BRANCH c BRANCH NOTHING --> END | |
| 138 * | | | | ^ ^ | |
| 139 * | | | +-----+ | | |
| 140 * | | +----------------+ | | |
| 141 * | +---------------------------+ | | |
| 142 * +------------------------------------------------------+ | |
| 143 * | |
| 1209 | 144 * They all start with a BRANCH for "\|" alternatives, even when there is only |
| 7 | 145 * one alternative. |
| 146 */ | |
| 147 | |
| 148 /* | |
| 149 * The opcodes are: | |
| 150 */ | |
| 151 | |
| 152 /* definition number opnd? meaning */ | |
| 153 #define END 0 /* End of program or NOMATCH operand. */ | |
| 154 #define BOL 1 /* Match "" at beginning of line. */ | |
| 155 #define EOL 2 /* Match "" at end of line. */ | |
| 156 #define BRANCH 3 /* node Match this alternative, or the | |
| 157 * next... */ | |
| 158 #define BACK 4 /* Match "", "next" ptr points backward. */ | |
| 159 #define EXACTLY 5 /* str Match this string. */ | |
| 160 #define NOTHING 6 /* Match empty string. */ | |
| 161 #define STAR 7 /* node Match this (simple) thing 0 or more | |
| 162 * times. */ | |
| 163 #define PLUS 8 /* node Match this (simple) thing 1 or more | |
| 164 * times. */ | |
| 165 #define MATCH 9 /* node match the operand zero-width */ | |
| 166 #define NOMATCH 10 /* node check for no match with operand */ | |
| 167 #define BEHIND 11 /* node look behind for a match with operand */ | |
| 168 #define NOBEHIND 12 /* node look behind for no match with operand */ | |
| 169 #define SUBPAT 13 /* node match the operand here */ | |
| 170 #define BRACE_SIMPLE 14 /* node Match this (simple) thing between m and | |
| 171 * n times (\{m,n\}). */ | |
| 172 #define BOW 15 /* Match "" after [^a-zA-Z0-9_] */ | |
| 173 #define EOW 16 /* Match "" at [^a-zA-Z0-9_] */ | |
| 174 #define BRACE_LIMITS 17 /* nr nr define the min & max for BRACE_SIMPLE | |
| 175 * and BRACE_COMPLEX. */ | |
| 176 #define NEWL 18 /* Match line-break */ | |
| 177 #define BHPOS 19 /* End position for BEHIND or NOBEHIND */ | |
| 178 | |
| 179 | |
| 180 /* character classes: 20-48 normal, 50-78 include a line-break */ | |
| 181 #define ADD_NL 30 | |
| 182 #define FIRST_NL ANY + ADD_NL | |
| 183 #define ANY 20 /* Match any one character. */ | |
| 184 #define ANYOF 21 /* str Match any character in this string. */ | |
| 185 #define ANYBUT 22 /* str Match any character not in this | |
| 186 * string. */ | |
| 187 #define IDENT 23 /* Match identifier char */ | |
| 188 #define SIDENT 24 /* Match identifier char but no digit */ | |
| 189 #define KWORD 25 /* Match keyword char */ | |
| 190 #define SKWORD 26 /* Match word char but no digit */ | |
| 191 #define FNAME 27 /* Match file name char */ | |
| 192 #define SFNAME 28 /* Match file name char but no digit */ | |
| 193 #define PRINT 29 /* Match printable char */ | |
| 194 #define SPRINT 30 /* Match printable char but no digit */ | |
| 195 #define WHITE 31 /* Match whitespace char */ | |
| 196 #define NWHITE 32 /* Match non-whitespace char */ | |
| 197 #define DIGIT 33 /* Match digit char */ | |
| 198 #define NDIGIT 34 /* Match non-digit char */ | |
| 199 #define HEX 35 /* Match hex char */ | |
| 200 #define NHEX 36 /* Match non-hex char */ | |
| 201 #define OCTAL 37 /* Match octal char */ | |
| 202 #define NOCTAL 38 /* Match non-octal char */ | |
| 203 #define WORD 39 /* Match word char */ | |
| 204 #define NWORD 40 /* Match non-word char */ | |
| 205 #define HEAD 41 /* Match head char */ | |
| 206 #define NHEAD 42 /* Match non-head char */ | |
| 207 #define ALPHA 43 /* Match alpha char */ | |
| 208 #define NALPHA 44 /* Match non-alpha char */ | |
| 209 #define LOWER 45 /* Match lowercase char */ | |
| 210 #define NLOWER 46 /* Match non-lowercase char */ | |
| 211 #define UPPER 47 /* Match uppercase char */ | |
| 212 #define NUPPER 48 /* Match non-uppercase char */ | |
| 213 #define LAST_NL NUPPER + ADD_NL | |
| 214 #define WITH_NL(op) ((op) >= FIRST_NL && (op) <= LAST_NL) | |
| 215 | |
| 216 #define MOPEN 80 /* -89 Mark this point in input as start of | |
| 217 * \( subexpr. MOPEN + 0 marks start of | |
| 218 * match. */ | |
| 219 #define MCLOSE 90 /* -99 Analogous to MOPEN. MCLOSE + 0 marks | |
| 220 * end of match. */ | |
| 221 #define BACKREF 100 /* -109 node Match same string again \1-\9 */ | |
| 222 | |
| 223 #ifdef FEAT_SYN_HL | |
| 224 # define ZOPEN 110 /* -119 Mark this point in input as start of | |
| 225 * \z( subexpr. */ | |
| 226 # define ZCLOSE 120 /* -129 Analogous to ZOPEN. */ | |
| 227 # define ZREF 130 /* -139 node Match external submatch \z1-\z9 */ | |
| 228 #endif | |
| 229 | |
| 230 #define BRACE_COMPLEX 140 /* -149 node Match nodes between m & n times */ | |
| 231 | |
| 232 #define NOPEN 150 /* Mark this point in input as start of | |
| 233 \%( subexpr. */ | |
| 234 #define NCLOSE 151 /* Analogous to NOPEN. */ | |
| 235 | |
| 236 #define MULTIBYTECODE 200 /* mbc Match one multi-byte character */ | |
| 237 #define RE_BOF 201 /* Match "" at beginning of file. */ | |
| 238 #define RE_EOF 202 /* Match "" at end of file. */ | |
| 239 #define CURSOR 203 /* Match location of cursor. */ | |
| 240 | |
| 241 #define RE_LNUM 204 /* nr cmp Match line number */ | |
| 242 #define RE_COL 205 /* nr cmp Match column number */ | |
| 243 #define RE_VCOL 206 /* nr cmp Match virtual column number */ | |
| 244 | |
| 639 | 245 #define RE_MARK 207 /* mark cmp Match mark position */ |
| 246 #define RE_VISUAL 208 /* Match Visual area */ | |
| 247 | |
| 7 | 248 /* |
| 249 * Magic characters have a special meaning, they don't match literally. | |
| 250 * Magic characters are negative. This separates them from literal characters | |
| 251 * (possibly multi-byte). Only ASCII characters can be Magic. | |
| 252 */ | |
| 253 #define Magic(x) ((int)(x) - 256) | |
| 254 #define un_Magic(x) ((x) + 256) | |
| 255 #define is_Magic(x) ((x) < 0) | |
| 256 | |
| 257 static int no_Magic __ARGS((int x)); | |
| 258 static int toggle_Magic __ARGS((int x)); | |
| 259 | |
| 260 static int | |
| 261 no_Magic(x) | |
| 262 int x; | |
| 263 { | |
| 264 if (is_Magic(x)) | |
| 265 return un_Magic(x); | |
| 266 return x; | |
| 267 } | |
| 268 | |
| 269 static int | |
| 270 toggle_Magic(x) | |
| 271 int x; | |
| 272 { | |
| 273 if (is_Magic(x)) | |
| 274 return un_Magic(x); | |
| 275 return Magic(x); | |
| 276 } | |
| 277 | |
| 278 /* | |
| 279 * The first byte of the regexp internal "program" is actually this magic | |
| 280 * number; the start node begins in the second byte. It's used to catch the | |
| 281 * most severe mutilation of the program by the caller. | |
| 282 */ | |
| 283 | |
| 284 #define REGMAGIC 0234 | |
| 285 | |
| 286 /* | |
| 287 * Opcode notes: | |
| 288 * | |
| 289 * BRANCH The set of branches constituting a single choice are hooked | |
| 290 * together with their "next" pointers, since precedence prevents | |
| 291 * anything being concatenated to any individual branch. The | |
| 292 * "next" pointer of the last BRANCH in a choice points to the | |
| 293 * thing following the whole choice. This is also where the | |
| 294 * final "next" pointer of each individual branch points; each | |
| 295 * branch starts with the operand node of a BRANCH node. | |
| 296 * | |
| 297 * BACK Normal "next" pointers all implicitly point forward; BACK | |
| 298 * exists to make loop structures possible. | |
| 299 * | |
| 300 * STAR,PLUS '=', and complex '*' and '+', are implemented as circular | |
| 301 * BRANCH structures using BACK. Simple cases (one character | |
| 302 * per match) are implemented with STAR and PLUS for speed | |
| 303 * and to minimize recursive plunges. | |
| 304 * | |
| 305 * BRACE_LIMITS This is always followed by a BRACE_SIMPLE or BRACE_COMPLEX | |
| 306 * node, and defines the min and max limits to be used for that | |
| 307 * node. | |
| 308 * | |
| 309 * MOPEN,MCLOSE ...are numbered at compile time. | |
| 310 * ZOPEN,ZCLOSE ...ditto | |
| 311 */ | |
| 312 | |
| 313 /* | |
| 314 * A node is one char of opcode followed by two chars of "next" pointer. | |
| 315 * "Next" pointers are stored as two 8-bit bytes, high order first. The | |
| 316 * value is a positive offset from the opcode of the node containing it. | |
| 317 * An operand, if any, simply follows the node. (Note that much of the | |
| 318 * code generation knows about this implicit relationship.) | |
| 319 * | |
| 320 * Using two bytes for the "next" pointer is vast overkill for most things, | |
| 321 * but allows patterns to get big without disasters. | |
| 322 */ | |
| 323 #define OP(p) ((int)*(p)) | |
| 324 #define NEXT(p) (((*((p) + 1) & 0377) << 8) + (*((p) + 2) & 0377)) | |
| 325 #define OPERAND(p) ((p) + 3) | |
| 326 /* Obtain an operand that was stored as four bytes, MSB first. */ | |
| 327 #define OPERAND_MIN(p) (((long)(p)[3] << 24) + ((long)(p)[4] << 16) \ | |
| 328 + ((long)(p)[5] << 8) + (long)(p)[6]) | |
| 329 /* Obtain a second operand stored as four bytes. */ | |
| 330 #define OPERAND_MAX(p) OPERAND_MIN((p) + 4) | |
| 331 /* Obtain a second single-byte operand stored after a four bytes operand. */ | |
| 332 #define OPERAND_CMP(p) (p)[7] | |
| 333 | |
| 334 /* | |
| 335 * Utility definitions. | |
| 336 */ | |
| 337 #define UCHARAT(p) ((int)*(char_u *)(p)) | |
| 338 | |
| 339 /* Used for an error (down from) vim_regcomp(): give the error message, set | |
| 340 * rc_did_emsg and return NULL */ | |
| 653 | 341 #define EMSG_RET_NULL(m) return (EMSG(m), rc_did_emsg = TRUE, (void *)NULL) |
| 308 | 342 #define EMSG_RET_FAIL(m) return (EMSG(m), rc_did_emsg = TRUE, FAIL) |
| 4444 | 343 #define EMSG2_RET_NULL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, (void *)NULL) |
| 344 #define EMSG2_RET_FAIL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, FAIL) | |
| 345 #define EMSG_ONE_RET_NULL EMSG2_RET_NULL(_("E369: invalid item in %s%%[]"), reg_magic == MAGIC_ALL) | |
| 7 | 346 |
| 347 #define MAX_LIMIT (32767L << 16L) | |
| 348 | |
| 349 static int re_multi_type __ARGS((int)); | |
| 350 static int cstrncmp __ARGS((char_u *s1, char_u *s2, int *n)); | |
| 351 static char_u *cstrchr __ARGS((char_u *, int)); | |
| 352 | |
| 4444 | 353 #ifdef BT_REGEXP_DUMP |
| 354 static void regdump __ARGS((char_u *, bt_regprog_T *)); | |
| 355 #endif | |
| 7 | 356 #ifdef DEBUG |
| 357 static char_u *regprop __ARGS((char_u *)); | |
| 358 #endif | |
| 359 | |
| 4444 | 360 static char_u e_missingbracket[] = N_("E769: Missing ] after %s["); |
| 361 static char_u e_unmatchedpp[] = N_("E53: Unmatched %s%%("); | |
| 362 static char_u e_unmatchedp[] = N_("E54: Unmatched %s("); | |
| 363 static char_u e_unmatchedpar[] = N_("E55: Unmatched %s)"); | |
| 364 | |
| 7 | 365 #define NOT_MULTI 0 |
| 366 #define MULTI_ONE 1 | |
| 367 #define MULTI_MULT 2 | |
| 368 /* | |
| 369 * Return NOT_MULTI if c is not a "multi" operator. | |
| 370 * Return MULTI_ONE if c is a single "multi" operator. | |
| 371 * Return MULTI_MULT if c is a multi "multi" operator. | |
| 372 */ | |
| 373 static int | |
| 374 re_multi_type(c) | |
| 375 int c; | |
| 376 { | |
| 377 if (c == Magic('@') || c == Magic('=') || c == Magic('?')) | |
| 378 return MULTI_ONE; | |
| 379 if (c == Magic('*') || c == Magic('+') || c == Magic('{')) | |
| 380 return MULTI_MULT; | |
| 381 return NOT_MULTI; | |
| 382 } | |
| 383 | |
| 384 /* | |
| 385 * Flags to be passed up and down. | |
| 386 */ | |
| 387 #define HASWIDTH 0x1 /* Known never to match null string. */ | |
| 388 #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ | |
| 389 #define SPSTART 0x4 /* Starts with * or +. */ | |
| 390 #define HASNL 0x8 /* Contains some \n. */ | |
| 391 #define HASLOOKBH 0x10 /* Contains "\@<=" or "\@<!". */ | |
| 392 #define WORST 0 /* Worst case. */ | |
| 393 | |
| 394 /* | |
| 395 * When regcode is set to this value, code is not emitted and size is computed | |
| 396 * instead. | |
| 397 */ | |
| 398 #define JUST_CALC_SIZE ((char_u *) -1) | |
| 399 | |
| 359 | 400 static char_u *reg_prev_sub = NULL; |
| 401 | |
| 7 | 402 /* |
| 403 * REGEXP_INRANGE contains all characters which are always special in a [] | |
| 404 * range after '\'. | |
| 405 * REGEXP_ABBR contains all characters which act as abbreviations after '\'. | |
| 406 * These are: | |
| 407 * \n - New line (NL). | |
| 408 * \r - Carriage Return (CR). | |
| 409 * \t - Tab (TAB). | |
| 410 * \e - Escape (ESC). | |
| 411 * \b - Backspace (Ctrl_H). | |
| 24 | 412 * \d - Character code in decimal, eg \d123 |
| 413 * \o - Character code in octal, eg \o80 | |
| 414 * \x - Character code in hex, eg \x4a | |
| 415 * \u - Multibyte character code, eg \u20ac | |
| 416 * \U - Long multibyte character code, eg \U12345678 | |
| 7 | 417 */ |
| 418 static char_u REGEXP_INRANGE[] = "]^-n\\"; | |
| 24 | 419 static char_u REGEXP_ABBR[] = "nrtebdoxuU"; |
| 7 | 420 |
| 421 static int backslash_trans __ARGS((int c)); | |
| 167 | 422 static int get_char_class __ARGS((char_u **pp)); |
| 423 static int get_equi_class __ARGS((char_u **pp)); | |
| 424 static void reg_equi_class __ARGS((int c)); | |
| 425 static int get_coll_element __ARGS((char_u **pp)); | |
| 7 | 426 static char_u *skip_anyof __ARGS((char_u *p)); |
| 427 static void init_class_tab __ARGS((void)); | |
| 428 | |
| 429 /* | |
| 430 * Translate '\x' to its control character, except "\n", which is Magic. | |
| 431 */ | |
| 432 static int | |
| 433 backslash_trans(c) | |
| 434 int c; | |
| 435 { | |
| 436 switch (c) | |
| 437 { | |
| 438 case 'r': return CAR; | |
| 439 case 't': return TAB; | |
| 440 case 'e': return ESC; | |
| 441 case 'b': return BS; | |
| 442 } | |
| 443 return c; | |
| 444 } | |
| 445 | |
| 446 /* | |
| 167 | 447 * Check for a character class name "[:name:]". "pp" points to the '['. |
| 7 | 448 * Returns one of the CLASS_ items. CLASS_NONE means that no item was |
| 449 * recognized. Otherwise "pp" is advanced to after the item. | |
| 450 */ | |
| 451 static int | |
| 167 | 452 get_char_class(pp) |
| 7 | 453 char_u **pp; |
| 454 { | |
| 455 static const char *(class_names[]) = | |
| 456 { | |
| 457 "alnum:]", | |
| 458 #define CLASS_ALNUM 0 | |
| 459 "alpha:]", | |
| 460 #define CLASS_ALPHA 1 | |
| 461 "blank:]", | |
| 462 #define CLASS_BLANK 2 | |
| 463 "cntrl:]", | |
| 464 #define CLASS_CNTRL 3 | |
| 465 "digit:]", | |
| 466 #define CLASS_DIGIT 4 | |
| 467 "graph:]", | |
| 468 #define CLASS_GRAPH 5 | |
| 469 "lower:]", | |
| 470 #define CLASS_LOWER 6 | |
| 471 "print:]", | |
| 472 #define CLASS_PRINT 7 | |
| 473 "punct:]", | |
| 474 #define CLASS_PUNCT 8 | |
| 475 "space:]", | |
| 476 #define CLASS_SPACE 9 | |
| 477 "upper:]", | |
| 478 #define CLASS_UPPER 10 | |
| 479 "xdigit:]", | |
| 480 #define CLASS_XDIGIT 11 | |
| 481 "tab:]", | |
| 482 #define CLASS_TAB 12 | |
| 483 "return:]", | |
| 484 #define CLASS_RETURN 13 | |
| 485 "backspace:]", | |
| 486 #define CLASS_BACKSPACE 14 | |
| 487 "escape:]", | |
| 488 #define CLASS_ESCAPE 15 | |
| 489 }; | |
| 490 #define CLASS_NONE 99 | |
| 491 int i; | |
| 492 | |
| 493 if ((*pp)[1] == ':') | |
| 494 { | |
| 1877 | 495 for (i = 0; i < (int)(sizeof(class_names) / sizeof(*class_names)); ++i) |
| 7 | 496 if (STRNCMP(*pp + 2, class_names[i], STRLEN(class_names[i])) == 0) |
| 497 { | |
| 498 *pp += STRLEN(class_names[i]) + 2; | |
| 499 return i; | |
| 500 } | |
| 501 } | |
| 502 return CLASS_NONE; | |
| 503 } | |
| 504 | |
| 505 /* | |
| 506 * Specific version of character class functions. | |
| 507 * Using a table to keep this fast. | |
| 508 */ | |
| 509 static short class_tab[256]; | |
| 510 | |
| 511 #define RI_DIGIT 0x01 | |
| 512 #define RI_HEX 0x02 | |
| 513 #define RI_OCTAL 0x04 | |
| 514 #define RI_WORD 0x08 | |
| 515 #define RI_HEAD 0x10 | |
| 516 #define RI_ALPHA 0x20 | |
| 517 #define RI_LOWER 0x40 | |
| 518 #define RI_UPPER 0x80 | |
| 519 #define RI_WHITE 0x100 | |
| 520 | |
| 521 static void | |
| 522 init_class_tab() | |
| 523 { | |
| 524 int i; | |
| 525 static int done = FALSE; | |
| 526 | |
| 527 if (done) | |
| 528 return; | |
| 529 | |
| 530 for (i = 0; i < 256; ++i) | |
| 531 { | |
| 532 if (i >= '0' && i <= '7') | |
| 533 class_tab[i] = RI_DIGIT + RI_HEX + RI_OCTAL + RI_WORD; | |
| 534 else if (i >= '8' && i <= '9') | |
| 535 class_tab[i] = RI_DIGIT + RI_HEX + RI_WORD; | |
| 536 else if (i >= 'a' && i <= 'f') | |
| 537 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
| 538 #ifdef EBCDIC | |
| 539 else if ((i >= 'g' && i <= 'i') || (i >= 'j' && i <= 'r') | |
| 540 || (i >= 's' && i <= 'z')) | |
| 541 #else | |
| 542 else if (i >= 'g' && i <= 'z') | |
| 543 #endif | |
| 544 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
| 545 else if (i >= 'A' && i <= 'F') | |
| 546 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
| 547 #ifdef EBCDIC | |
| 548 else if ((i >= 'G' && i <= 'I') || ( i >= 'J' && i <= 'R') | |
| 549 || (i >= 'S' && i <= 'Z')) | |
| 550 #else | |
| 551 else if (i >= 'G' && i <= 'Z') | |
| 552 #endif | |
| 553 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
| 554 else if (i == '_') | |
| 555 class_tab[i] = RI_WORD + RI_HEAD; | |
| 556 else | |
| 557 class_tab[i] = 0; | |
| 558 } | |
| 559 class_tab[' '] |= RI_WHITE; | |
| 560 class_tab['\t'] |= RI_WHITE; | |
| 561 done = TRUE; | |
| 562 } | |
| 563 | |
| 564 #ifdef FEAT_MBYTE | |
| 565 # define ri_digit(c) (c < 0x100 && (class_tab[c] & RI_DIGIT)) | |
| 566 # define ri_hex(c) (c < 0x100 && (class_tab[c] & RI_HEX)) | |
| 567 # define ri_octal(c) (c < 0x100 && (class_tab[c] & RI_OCTAL)) | |
| 568 # define ri_word(c) (c < 0x100 && (class_tab[c] & RI_WORD)) | |
| 569 # define ri_head(c) (c < 0x100 && (class_tab[c] & RI_HEAD)) | |
| 570 # define ri_alpha(c) (c < 0x100 && (class_tab[c] & RI_ALPHA)) | |
| 571 # define ri_lower(c) (c < 0x100 && (class_tab[c] & RI_LOWER)) | |
| 572 # define ri_upper(c) (c < 0x100 && (class_tab[c] & RI_UPPER)) | |
| 573 # define ri_white(c) (c < 0x100 && (class_tab[c] & RI_WHITE)) | |
| 574 #else | |
| 575 # define ri_digit(c) (class_tab[c] & RI_DIGIT) | |
| 576 # define ri_hex(c) (class_tab[c] & RI_HEX) | |
| 577 # define ri_octal(c) (class_tab[c] & RI_OCTAL) | |
| 578 # define ri_word(c) (class_tab[c] & RI_WORD) | |
| 579 # define ri_head(c) (class_tab[c] & RI_HEAD) | |
| 580 # define ri_alpha(c) (class_tab[c] & RI_ALPHA) | |
| 581 # define ri_lower(c) (class_tab[c] & RI_LOWER) | |
| 582 # define ri_upper(c) (class_tab[c] & RI_UPPER) | |
| 583 # define ri_white(c) (class_tab[c] & RI_WHITE) | |
| 584 #endif | |
| 585 | |
| 586 /* flags for regflags */ | |
| 587 #define RF_ICASE 1 /* ignore case */ | |
| 588 #define RF_NOICASE 2 /* don't ignore case */ | |
| 589 #define RF_HASNL 4 /* can match a NL */ | |
| 590 #define RF_ICOMBINE 8 /* ignore combining characters */ | |
| 591 #define RF_LOOKBH 16 /* uses "\@<=" or "\@<!" */ | |
| 592 | |
| 593 /* | |
| 594 * Global work variables for vim_regcomp(). | |
| 595 */ | |
| 596 | |
| 597 static char_u *regparse; /* Input-scan pointer. */ | |
| 598 static int prevchr_len; /* byte length of previous char */ | |
| 599 static int num_complex_braces; /* Complex \{...} count */ | |
| 600 static int regnpar; /* () count. */ | |
| 601 #ifdef FEAT_SYN_HL | |
| 602 static int regnzpar; /* \z() count. */ | |
| 603 static int re_has_z; /* \z item detected */ | |
| 604 #endif | |
| 605 static char_u *regcode; /* Code-emit pointer, or JUST_CALC_SIZE */ | |
| 606 static long regsize; /* Code size. */ | |
| 2010 | 607 static int reg_toolong; /* TRUE when offset out of range */ |
| 7 | 608 static char_u had_endbrace[NSUBEXP]; /* flags, TRUE if end of () found */ |
| 609 static unsigned regflags; /* RF_ flags for prog */ | |
| 610 static long brace_min[10]; /* Minimums for complex brace repeats */ | |
| 611 static long brace_max[10]; /* Maximums for complex brace repeats */ | |
| 612 static int brace_count[10]; /* Current counts for complex brace repeats */ | |
| 613 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 614 static int had_eol; /* TRUE when EOL found by vim_regcomp() */ | |
| 615 #endif | |
| 616 static int one_exactly = FALSE; /* only do one char for EXACTLY */ | |
| 617 | |
| 618 static int reg_magic; /* magicness of the pattern: */ | |
| 619 #define MAGIC_NONE 1 /* "\V" very unmagic */ | |
| 620 #define MAGIC_OFF 2 /* "\M" or 'magic' off */ | |
| 621 #define MAGIC_ON 3 /* "\m" or 'magic' */ | |
| 622 #define MAGIC_ALL 4 /* "\v" very magic */ | |
| 623 | |
| 624 static int reg_string; /* matching with a string instead of a buffer | |
| 625 line */ | |
| 481 | 626 static int reg_strict; /* "[abc" is illegal */ |
| 7 | 627 |
| 628 /* | |
| 629 * META contains all characters that may be magic, except '^' and '$'. | |
| 630 */ | |
| 631 | |
| 632 #ifdef EBCDIC | |
| 633 static char_u META[] = "%&()*+.123456789<=>?@ACDFHIKLMOPSUVWX[_acdfhiklmnopsuvwxz{|~"; | |
| 634 #else | |
| 635 /* META[] is used often enough to justify turning it into a table. */ | |
| 636 static char_u META_flags[] = { | |
| 637 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| 638 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| 639 /* % & ( ) * + . */ | |
| 640 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0, | |
| 641 /* 1 2 3 4 5 6 7 8 9 < = > ? */ | |
| 642 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, | |
| 643 /* @ A C D F H I K L M O */ | |
| 644 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1, | |
| 645 /* P S U V W X Z [ _ */ | |
| 646 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, | |
| 647 /* a c d f h i k l m n o */ | |
| 648 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, | |
| 649 /* p s u v w x z { | ~ */ | |
| 650 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1 | |
| 651 }; | |
| 652 #endif | |
| 653 | |
| 4444 | 654 static int curchr; /* currently parsed character */ |
| 655 /* Previous character. Note: prevchr is sometimes -1 when we are not at the | |
| 656 * start, eg in /[ ^I]^ the pattern was never found even if it existed, | |
| 657 * because ^ was taken to be magic -- webb */ | |
| 658 static int prevchr; | |
| 659 static int prevprevchr; /* previous-previous character */ | |
| 660 static int nextchr; /* used for ungetchr() */ | |
| 7 | 661 |
| 662 /* arguments for reg() */ | |
| 663 #define REG_NOPAREN 0 /* toplevel reg() */ | |
| 664 #define REG_PAREN 1 /* \(\) */ | |
| 665 #define REG_ZPAREN 2 /* \z(\) */ | |
| 666 #define REG_NPAREN 3 /* \%(\) */ | |
| 667 | |
| 668 /* | |
| 669 * Forward declarations for vim_regcomp()'s friends. | |
| 670 */ | |
| 671 static void initchr __ARGS((char_u *)); | |
| 672 static int getchr __ARGS((void)); | |
| 673 static void skipchr_keepstart __ARGS((void)); | |
| 674 static int peekchr __ARGS((void)); | |
| 675 static void skipchr __ARGS((void)); | |
| 676 static void ungetchr __ARGS((void)); | |
| 24 | 677 static int gethexchrs __ARGS((int maxinputlen)); |
| 678 static int getoctchrs __ARGS((void)); | |
| 679 static int getdecchrs __ARGS((void)); | |
| 680 static int coll_get_char __ARGS((void)); | |
| 7 | 681 static void regcomp_start __ARGS((char_u *expr, int flags)); |
| 682 static char_u *reg __ARGS((int, int *)); | |
| 683 static char_u *regbranch __ARGS((int *flagp)); | |
| 684 static char_u *regconcat __ARGS((int *flagp)); | |
| 685 static char_u *regpiece __ARGS((int *)); | |
| 686 static char_u *regatom __ARGS((int *)); | |
| 687 static char_u *regnode __ARGS((int)); | |
| 714 | 688 #ifdef FEAT_MBYTE |
| 689 static int use_multibytecode __ARGS((int c)); | |
| 690 #endif | |
| 7 | 691 static int prog_magic_wrong __ARGS((void)); |
| 692 static char_u *regnext __ARGS((char_u *)); | |
| 693 static void regc __ARGS((int b)); | |
| 694 #ifdef FEAT_MBYTE | |
| 695 static void regmbc __ARGS((int c)); | |
| 2974 | 696 # define REGMBC(x) regmbc(x); |
| 697 # define CASEMBC(x) case x: | |
| 167 | 698 #else |
| 699 # define regmbc(c) regc(c) | |
| 2974 | 700 # define REGMBC(x) |
| 701 # define CASEMBC(x) | |
| 7 | 702 #endif |
| 703 static void reginsert __ARGS((int, char_u *)); | |
| 704 static void reginsert_limits __ARGS((int, long, long, char_u *)); | |
| 705 static char_u *re_put_long __ARGS((char_u *pr, long_u val)); | |
| 706 static int read_limits __ARGS((long *, long *)); | |
| 707 static void regtail __ARGS((char_u *, char_u *)); | |
| 708 static void regoptail __ARGS((char_u *, char_u *)); | |
| 709 | |
| 4444 | 710 static regengine_T bt_regengine; |
| 711 static regengine_T nfa_regengine; | |
| 712 | |
| 7 | 713 /* |
| 714 * Return TRUE if compiled regular expression "prog" can match a line break. | |
| 715 */ | |
| 716 int | |
| 717 re_multiline(prog) | |
| 718 regprog_T *prog; | |
| 719 { | |
| 720 return (prog->regflags & RF_HASNL); | |
| 721 } | |
| 722 | |
| 723 /* | |
| 724 * Return TRUE if compiled regular expression "prog" looks before the start | |
| 725 * position (pattern contains "\@<=" or "\@<!"). | |
| 726 */ | |
| 727 int | |
| 728 re_lookbehind(prog) | |
| 729 regprog_T *prog; | |
| 730 { | |
| 731 return (prog->regflags & RF_LOOKBH); | |
| 732 } | |
| 733 | |
| 734 /* | |
| 167 | 735 * Check for an equivalence class name "[=a=]". "pp" points to the '['. |
| 736 * Returns a character representing the class. Zero means that no item was | |
| 737 * recognized. Otherwise "pp" is advanced to after the item. | |
| 738 */ | |
| 739 static int | |
| 740 get_equi_class(pp) | |
| 741 char_u **pp; | |
| 742 { | |
| 743 int c; | |
| 744 int l = 1; | |
| 745 char_u *p = *pp; | |
| 746 | |
| 747 if (p[1] == '=') | |
| 748 { | |
| 749 #ifdef FEAT_MBYTE | |
| 750 if (has_mbyte) | |
| 474 | 751 l = (*mb_ptr2len)(p + 2); |
| 167 | 752 #endif |
| 753 if (p[l + 2] == '=' && p[l + 3] == ']') | |
| 754 { | |
| 755 #ifdef FEAT_MBYTE | |
| 756 if (has_mbyte) | |
| 757 c = mb_ptr2char(p + 2); | |
| 758 else | |
| 759 #endif | |
| 760 c = p[2]; | |
| 761 *pp += l + 4; | |
| 762 return c; | |
| 763 } | |
| 764 } | |
| 765 return 0; | |
| 766 } | |
| 767 | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
768 #ifdef EBCDIC |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
769 /* |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
770 * 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
|
771 */ |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
772 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
|
773 "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
|
774 "C\x68", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
775 "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
|
776 "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
|
777 "N\x69", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
778 "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
|
779 "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
|
780 "Y\xBA", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
781 "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
|
782 "c\x48", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
783 "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
|
784 "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
|
785 "n\x49", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
786 "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
|
787 "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
|
788 "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
|
789 }; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
790 #endif |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
791 |
| 167 | 792 /* |
| 793 * Produce the bytes for equivalence class "c". | |
| 794 * Currently only handles latin1, latin9 and utf-8. | |
| 4444 | 795 * NOTE: When changing this function, also change nfa_emit_equi_class() |
| 167 | 796 */ |
| 797 static void | |
| 798 reg_equi_class(c) | |
| 799 int c; | |
| 800 { | |
| 801 #ifdef FEAT_MBYTE | |
| 802 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
| 492 | 803 || STRCMP(p_enc, "iso-8859-15") == 0) |
| 167 | 804 #endif |
| 805 { | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
806 #ifdef EBCDIC |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
807 int i; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
808 |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
809 /* 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
|
810 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
|
811 { |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
812 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
|
813 { |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
814 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
|
815 |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
816 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
|
817 regmbc(*p++); |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
818 return; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
819 } |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
820 } |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
821 #else |
| 167 | 822 switch (c) |
| 823 { | |
| 236 | 824 case 'A': case '\300': case '\301': case '\302': |
| 2974 | 825 CASEMBC(0x100) CASEMBC(0x102) CASEMBC(0x104) CASEMBC(0x1cd) |
| 826 CASEMBC(0x1de) CASEMBC(0x1e0) CASEMBC(0x1ea2) | |
| 236 | 827 case '\303': case '\304': case '\305': |
| 828 regmbc('A'); regmbc('\300'); regmbc('\301'); | |
| 829 regmbc('\302'); regmbc('\303'); regmbc('\304'); | |
| 830 regmbc('\305'); | |
| 2974 | 831 REGMBC(0x100) REGMBC(0x102) REGMBC(0x104) |
| 832 REGMBC(0x1cd) REGMBC(0x1de) REGMBC(0x1e0) | |
| 833 REGMBC(0x1ea2) | |
| 834 return; | |
| 835 case 'B': CASEMBC(0x1e02) CASEMBC(0x1e06) | |
| 836 regmbc('B'); REGMBC(0x1e02) REGMBC(0x1e06) | |
| 167 | 837 return; |
| 236 | 838 case 'C': case '\307': |
| 2974 | 839 CASEMBC(0x106) CASEMBC(0x108) CASEMBC(0x10a) CASEMBC(0x10c) |
| 236 | 840 regmbc('C'); regmbc('\307'); |
| 2974 | 841 REGMBC(0x106) REGMBC(0x108) REGMBC(0x10a) |
| 842 REGMBC(0x10c) | |
| 843 return; | |
| 844 case 'D': CASEMBC(0x10e) CASEMBC(0x110) CASEMBC(0x1e0a) | |
| 845 CASEMBC(0x1e0e) CASEMBC(0x1e10) | |
| 846 regmbc('D'); REGMBC(0x10e) REGMBC(0x110) | |
| 847 REGMBC(0x1e0a) REGMBC(0x1e0e) REGMBC(0x1e10) | |
| 167 | 848 return; |
| 236 | 849 case 'E': case '\310': case '\311': case '\312': case '\313': |
| 2974 | 850 CASEMBC(0x112) CASEMBC(0x114) CASEMBC(0x116) CASEMBC(0x118) |
| 851 CASEMBC(0x11a) CASEMBC(0x1eba) CASEMBC(0x1ebc) | |
| 236 | 852 regmbc('E'); regmbc('\310'); regmbc('\311'); |
| 853 regmbc('\312'); regmbc('\313'); | |
| 2974 | 854 REGMBC(0x112) REGMBC(0x114) REGMBC(0x116) |
| 855 REGMBC(0x118) REGMBC(0x11a) REGMBC(0x1eba) | |
| 856 REGMBC(0x1ebc) | |
| 857 return; | |
| 858 case 'F': CASEMBC(0x1e1e) | |
| 859 regmbc('F'); REGMBC(0x1e1e) | |
| 860 return; | |
| 861 case 'G': CASEMBC(0x11c) CASEMBC(0x11e) CASEMBC(0x120) | |
| 862 CASEMBC(0x122) CASEMBC(0x1e4) CASEMBC(0x1e6) CASEMBC(0x1f4) | |
| 863 CASEMBC(0x1e20) | |
| 864 regmbc('G'); REGMBC(0x11c) REGMBC(0x11e) | |
| 865 REGMBC(0x120) REGMBC(0x122) REGMBC(0x1e4) | |
| 866 REGMBC(0x1e6) REGMBC(0x1f4) REGMBC(0x1e20) | |
| 867 return; | |
| 868 case 'H': CASEMBC(0x124) CASEMBC(0x126) CASEMBC(0x1e22) | |
| 869 CASEMBC(0x1e26) CASEMBC(0x1e28) | |
| 870 regmbc('H'); REGMBC(0x124) REGMBC(0x126) | |
| 871 REGMBC(0x1e22) REGMBC(0x1e26) REGMBC(0x1e28) | |
| 167 | 872 return; |
| 236 | 873 case 'I': case '\314': case '\315': case '\316': case '\317': |
| 2974 | 874 CASEMBC(0x128) CASEMBC(0x12a) CASEMBC(0x12c) CASEMBC(0x12e) |
| 875 CASEMBC(0x130) CASEMBC(0x1cf) CASEMBC(0x1ec8) | |
| 236 | 876 regmbc('I'); regmbc('\314'); regmbc('\315'); |
| 877 regmbc('\316'); regmbc('\317'); | |
| 2974 | 878 REGMBC(0x128) REGMBC(0x12a) REGMBC(0x12c) |
| 879 REGMBC(0x12e) REGMBC(0x130) REGMBC(0x1cf) | |
| 880 REGMBC(0x1ec8) | |
| 881 return; | |
| 882 case 'J': CASEMBC(0x134) | |
| 883 regmbc('J'); REGMBC(0x134) | |
| 884 return; | |
| 885 case 'K': CASEMBC(0x136) CASEMBC(0x1e8) CASEMBC(0x1e30) | |
| 886 CASEMBC(0x1e34) | |
| 887 regmbc('K'); REGMBC(0x136) REGMBC(0x1e8) | |
| 888 REGMBC(0x1e30) REGMBC(0x1e34) | |
| 889 return; | |
| 890 case 'L': CASEMBC(0x139) CASEMBC(0x13b) CASEMBC(0x13d) | |
| 891 CASEMBC(0x13f) CASEMBC(0x141) CASEMBC(0x1e3a) | |
| 892 regmbc('L'); REGMBC(0x139) REGMBC(0x13b) | |
| 893 REGMBC(0x13d) REGMBC(0x13f) REGMBC(0x141) | |
| 894 REGMBC(0x1e3a) | |
| 895 return; | |
| 896 case 'M': CASEMBC(0x1e3e) CASEMBC(0x1e40) | |
| 897 regmbc('M'); REGMBC(0x1e3e) REGMBC(0x1e40) | |
| 167 | 898 return; |
| 236 | 899 case 'N': case '\321': |
| 2974 | 900 CASEMBC(0x143) CASEMBC(0x145) CASEMBC(0x147) CASEMBC(0x1e44) |
| 901 CASEMBC(0x1e48) | |
| 236 | 902 regmbc('N'); regmbc('\321'); |
| 2974 | 903 REGMBC(0x143) REGMBC(0x145) REGMBC(0x147) |
| 904 REGMBC(0x1e44) REGMBC(0x1e48) | |
| 167 | 905 return; |
| 236 | 906 case 'O': case '\322': case '\323': case '\324': case '\325': |
| 2974 | 907 case '\326': case '\330': |
| 908 CASEMBC(0x14c) CASEMBC(0x14e) CASEMBC(0x150) CASEMBC(0x1a0) | |
| 909 CASEMBC(0x1d1) CASEMBC(0x1ea) CASEMBC(0x1ec) CASEMBC(0x1ece) | |
| 236 | 910 regmbc('O'); regmbc('\322'); regmbc('\323'); |
| 911 regmbc('\324'); regmbc('\325'); regmbc('\326'); | |
| 2974 | 912 regmbc('\330'); |
| 913 REGMBC(0x14c) REGMBC(0x14e) REGMBC(0x150) | |
| 914 REGMBC(0x1a0) REGMBC(0x1d1) REGMBC(0x1ea) | |
| 915 REGMBC(0x1ec) REGMBC(0x1ece) | |
| 916 return; | |
| 917 case 'P': case 0x1e54: case 0x1e56: | |
| 918 regmbc('P'); REGMBC(0x1e54) REGMBC(0x1e56) | |
| 919 return; | |
| 920 case 'R': CASEMBC(0x154) CASEMBC(0x156) CASEMBC(0x158) | |
| 921 CASEMBC(0x1e58) CASEMBC(0x1e5e) | |
| 922 regmbc('R'); REGMBC(0x154) REGMBC(0x156) REGMBC(0x158) | |
| 923 REGMBC(0x1e58) REGMBC(0x1e5e) | |
| 924 return; | |
| 925 case 'S': CASEMBC(0x15a) CASEMBC(0x15c) CASEMBC(0x15e) | |
| 926 CASEMBC(0x160) CASEMBC(0x1e60) | |
| 927 regmbc('S'); REGMBC(0x15a) REGMBC(0x15c) | |
| 928 REGMBC(0x15e) REGMBC(0x160) REGMBC(0x1e60) | |
| 929 return; | |
| 930 case 'T': CASEMBC(0x162) CASEMBC(0x164) CASEMBC(0x166) | |
| 931 CASEMBC(0x1e6a) CASEMBC(0x1e6e) | |
| 932 regmbc('T'); REGMBC(0x162) REGMBC(0x164) | |
| 933 REGMBC(0x166) REGMBC(0x1e6a) REGMBC(0x1e6e) | |
| 167 | 934 return; |
| 236 | 935 case 'U': case '\331': case '\332': case '\333': case '\334': |
| 2974 | 936 CASEMBC(0x168) CASEMBC(0x16a) CASEMBC(0x16c) CASEMBC(0x16e) |
| 937 CASEMBC(0x170) CASEMBC(0x172) CASEMBC(0x1af) CASEMBC(0x1d3) | |
| 938 CASEMBC(0x1ee6) | |
| 236 | 939 regmbc('U'); regmbc('\331'); regmbc('\332'); |
| 940 regmbc('\333'); regmbc('\334'); | |
| 2974 | 941 REGMBC(0x168) REGMBC(0x16a) REGMBC(0x16c) |
| 942 REGMBC(0x16e) REGMBC(0x170) REGMBC(0x172) | |
| 943 REGMBC(0x1af) REGMBC(0x1d3) REGMBC(0x1ee6) | |
| 944 return; | |
| 945 case 'V': CASEMBC(0x1e7c) | |
| 946 regmbc('V'); REGMBC(0x1e7c) | |
| 947 return; | |
| 948 case 'W': CASEMBC(0x174) CASEMBC(0x1e80) CASEMBC(0x1e82) | |
| 949 CASEMBC(0x1e84) CASEMBC(0x1e86) | |
| 950 regmbc('W'); REGMBC(0x174) REGMBC(0x1e80) | |
| 951 REGMBC(0x1e82) REGMBC(0x1e84) REGMBC(0x1e86) | |
| 952 return; | |
| 953 case 'X': CASEMBC(0x1e8a) CASEMBC(0x1e8c) | |
| 954 regmbc('X'); REGMBC(0x1e8a) REGMBC(0x1e8c) | |
| 167 | 955 return; |
| 236 | 956 case 'Y': case '\335': |
| 2974 | 957 CASEMBC(0x176) CASEMBC(0x178) CASEMBC(0x1e8e) CASEMBC(0x1ef2) |
| 958 CASEMBC(0x1ef6) CASEMBC(0x1ef8) | |
| 236 | 959 regmbc('Y'); regmbc('\335'); |
| 2974 | 960 REGMBC(0x176) REGMBC(0x178) REGMBC(0x1e8e) |
| 961 REGMBC(0x1ef2) REGMBC(0x1ef6) REGMBC(0x1ef8) | |
| 962 return; | |
| 963 case 'Z': CASEMBC(0x179) CASEMBC(0x17b) CASEMBC(0x17d) | |
| 964 CASEMBC(0x1b5) CASEMBC(0x1e90) CASEMBC(0x1e94) | |
| 965 regmbc('Z'); REGMBC(0x179) REGMBC(0x17b) | |
| 966 REGMBC(0x17d) REGMBC(0x1b5) REGMBC(0x1e90) | |
| 967 REGMBC(0x1e94) | |
| 167 | 968 return; |
| 236 | 969 case 'a': case '\340': case '\341': case '\342': |
| 970 case '\343': case '\344': case '\345': | |
| 2974 | 971 CASEMBC(0x101) CASEMBC(0x103) CASEMBC(0x105) CASEMBC(0x1ce) |
| 972 CASEMBC(0x1df) CASEMBC(0x1e1) CASEMBC(0x1ea3) | |
| 236 | 973 regmbc('a'); regmbc('\340'); regmbc('\341'); |
| 974 regmbc('\342'); regmbc('\343'); regmbc('\344'); | |
| 975 regmbc('\345'); | |
| 2974 | 976 REGMBC(0x101) REGMBC(0x103) REGMBC(0x105) |
| 977 REGMBC(0x1ce) REGMBC(0x1df) REGMBC(0x1e1) | |
| 978 REGMBC(0x1ea3) | |
| 979 return; | |
| 980 case 'b': CASEMBC(0x1e03) CASEMBC(0x1e07) | |
| 981 regmbc('b'); REGMBC(0x1e03) REGMBC(0x1e07) | |
| 167 | 982 return; |
| 236 | 983 case 'c': case '\347': |
| 2974 | 984 CASEMBC(0x107) CASEMBC(0x109) CASEMBC(0x10b) CASEMBC(0x10d) |
| 236 | 985 regmbc('c'); regmbc('\347'); |
| 2974 | 986 REGMBC(0x107) REGMBC(0x109) REGMBC(0x10b) |
| 987 REGMBC(0x10d) | |
| 988 return; | |
| 989 case 'd': CASEMBC(0x10f) CASEMBC(0x111) CASEMBC(0x1d0b) | |
| 990 CASEMBC(0x1e11) | |
| 991 regmbc('d'); REGMBC(0x10f) REGMBC(0x111) | |
| 992 REGMBC(0x1e0b) REGMBC(0x01e0f) REGMBC(0x1e11) | |
| 167 | 993 return; |
| 236 | 994 case 'e': case '\350': case '\351': case '\352': case '\353': |
| 2974 | 995 CASEMBC(0x113) CASEMBC(0x115) CASEMBC(0x117) CASEMBC(0x119) |
| 996 CASEMBC(0x11b) CASEMBC(0x1ebb) CASEMBC(0x1ebd) | |
| 236 | 997 regmbc('e'); regmbc('\350'); regmbc('\351'); |
| 998 regmbc('\352'); regmbc('\353'); | |
| 2974 | 999 REGMBC(0x113) REGMBC(0x115) REGMBC(0x117) |
| 1000 REGMBC(0x119) REGMBC(0x11b) REGMBC(0x1ebb) | |
| 1001 REGMBC(0x1ebd) | |
| 1002 return; | |
| 1003 case 'f': CASEMBC(0x1e1f) | |
| 1004 regmbc('f'); REGMBC(0x1e1f) | |
| 1005 return; | |
| 1006 case 'g': CASEMBC(0x11d) CASEMBC(0x11f) CASEMBC(0x121) | |
| 1007 CASEMBC(0x123) CASEMBC(0x1e5) CASEMBC(0x1e7) CASEMBC(0x1f5) | |
| 1008 CASEMBC(0x1e21) | |
| 1009 regmbc('g'); REGMBC(0x11d) REGMBC(0x11f) | |
| 1010 REGMBC(0x121) REGMBC(0x123) REGMBC(0x1e5) | |
| 1011 REGMBC(0x1e7) REGMBC(0x1f5) REGMBC(0x1e21) | |
| 1012 return; | |
| 1013 case 'h': CASEMBC(0x125) CASEMBC(0x127) CASEMBC(0x1e23) | |
| 1014 CASEMBC(0x1e27) CASEMBC(0x1e29) CASEMBC(0x1e96) | |
| 1015 regmbc('h'); REGMBC(0x125) REGMBC(0x127) | |
| 1016 REGMBC(0x1e23) REGMBC(0x1e27) REGMBC(0x1e29) | |
| 1017 REGMBC(0x1e96) | |
| 167 | 1018 return; |
| 236 | 1019 case 'i': case '\354': case '\355': case '\356': case '\357': |
| 2974 | 1020 CASEMBC(0x129) CASEMBC(0x12b) CASEMBC(0x12d) CASEMBC(0x12f) |
| 1021 CASEMBC(0x1d0) CASEMBC(0x1ec9) | |
| 236 | 1022 regmbc('i'); regmbc('\354'); regmbc('\355'); |
| 1023 regmbc('\356'); regmbc('\357'); | |
| 2974 | 1024 REGMBC(0x129) REGMBC(0x12b) REGMBC(0x12d) |
| 1025 REGMBC(0x12f) REGMBC(0x1d0) REGMBC(0x1ec9) | |
| 1026 return; | |
| 1027 case 'j': CASEMBC(0x135) CASEMBC(0x1f0) | |
| 1028 regmbc('j'); REGMBC(0x135) REGMBC(0x1f0) | |
| 1029 return; | |
| 1030 case 'k': CASEMBC(0x137) CASEMBC(0x1e9) CASEMBC(0x1e31) | |
| 1031 CASEMBC(0x1e35) | |
| 1032 regmbc('k'); REGMBC(0x137) REGMBC(0x1e9) | |
| 1033 REGMBC(0x1e31) REGMBC(0x1e35) | |
| 1034 return; | |
| 1035 case 'l': CASEMBC(0x13a) CASEMBC(0x13c) CASEMBC(0x13e) | |
| 1036 CASEMBC(0x140) CASEMBC(0x142) CASEMBC(0x1e3b) | |
| 1037 regmbc('l'); REGMBC(0x13a) REGMBC(0x13c) | |
| 1038 REGMBC(0x13e) REGMBC(0x140) REGMBC(0x142) | |
| 1039 REGMBC(0x1e3b) | |
| 1040 return; | |
| 1041 case 'm': CASEMBC(0x1e3f) CASEMBC(0x1e41) | |
| 1042 regmbc('m'); REGMBC(0x1e3f) REGMBC(0x1e41) | |
| 167 | 1043 return; |
| 236 | 1044 case 'n': case '\361': |
| 2974 | 1045 CASEMBC(0x144) CASEMBC(0x146) CASEMBC(0x148) CASEMBC(0x149) |
| 1046 CASEMBC(0x1e45) CASEMBC(0x1e49) | |
| 236 | 1047 regmbc('n'); regmbc('\361'); |
| 2974 | 1048 REGMBC(0x144) REGMBC(0x146) REGMBC(0x148) |
| 1049 REGMBC(0x149) REGMBC(0x1e45) REGMBC(0x1e49) | |
| 167 | 1050 return; |
| 236 | 1051 case 'o': case '\362': case '\363': case '\364': case '\365': |
| 2974 | 1052 case '\366': case '\370': |
| 1053 CASEMBC(0x14d) CASEMBC(0x14f) CASEMBC(0x151) CASEMBC(0x1a1) | |
| 1054 CASEMBC(0x1d2) CASEMBC(0x1eb) CASEMBC(0x1ed) CASEMBC(0x1ecf) | |
| 236 | 1055 regmbc('o'); regmbc('\362'); regmbc('\363'); |
| 1056 regmbc('\364'); regmbc('\365'); regmbc('\366'); | |
| 2974 | 1057 regmbc('\370'); |
| 1058 REGMBC(0x14d) REGMBC(0x14f) REGMBC(0x151) | |
| 1059 REGMBC(0x1a1) REGMBC(0x1d2) REGMBC(0x1eb) | |
| 1060 REGMBC(0x1ed) REGMBC(0x1ecf) | |
| 1061 return; | |
| 1062 case 'p': CASEMBC(0x1e55) CASEMBC(0x1e57) | |
| 1063 regmbc('p'); REGMBC(0x1e55) REGMBC(0x1e57) | |
| 1064 return; | |
| 1065 case 'r': CASEMBC(0x155) CASEMBC(0x157) CASEMBC(0x159) | |
| 1066 CASEMBC(0x1e59) CASEMBC(0x1e5f) | |
| 1067 regmbc('r'); REGMBC(0x155) REGMBC(0x157) REGMBC(0x159) | |
| 1068 REGMBC(0x1e59) REGMBC(0x1e5f) | |
| 1069 return; | |
| 1070 case 's': CASEMBC(0x15b) CASEMBC(0x15d) CASEMBC(0x15f) | |
| 1071 CASEMBC(0x161) CASEMBC(0x1e61) | |
| 1072 regmbc('s'); REGMBC(0x15b) REGMBC(0x15d) | |
| 1073 REGMBC(0x15f) REGMBC(0x161) REGMBC(0x1e61) | |
| 1074 return; | |
| 1075 case 't': CASEMBC(0x163) CASEMBC(0x165) CASEMBC(0x167) | |
| 1076 CASEMBC(0x1e6b) CASEMBC(0x1e6f) CASEMBC(0x1e97) | |
| 1077 regmbc('t'); REGMBC(0x163) REGMBC(0x165) REGMBC(0x167) | |
| 1078 REGMBC(0x1e6b) REGMBC(0x1e6f) REGMBC(0x1e97) | |
| 167 | 1079 return; |
| 236 | 1080 case 'u': case '\371': case '\372': case '\373': case '\374': |
| 2974 | 1081 CASEMBC(0x169) CASEMBC(0x16b) CASEMBC(0x16d) CASEMBC(0x16f) |
| 1082 CASEMBC(0x171) CASEMBC(0x173) CASEMBC(0x1b0) CASEMBC(0x1d4) | |
| 1083 CASEMBC(0x1ee7) | |
| 236 | 1084 regmbc('u'); regmbc('\371'); regmbc('\372'); |
| 1085 regmbc('\373'); regmbc('\374'); | |
| 2974 | 1086 REGMBC(0x169) REGMBC(0x16b) REGMBC(0x16d) |
| 1087 REGMBC(0x16f) REGMBC(0x171) REGMBC(0x173) | |
| 1088 REGMBC(0x1b0) REGMBC(0x1d4) REGMBC(0x1ee7) | |
| 1089 return; | |
| 1090 case 'v': CASEMBC(0x1e7d) | |
| 1091 regmbc('v'); REGMBC(0x1e7d) | |
| 1092 return; | |
| 1093 case 'w': CASEMBC(0x175) CASEMBC(0x1e81) CASEMBC(0x1e83) | |
| 1094 CASEMBC(0x1e85) CASEMBC(0x1e87) CASEMBC(0x1e98) | |
| 1095 regmbc('w'); REGMBC(0x175) REGMBC(0x1e81) | |
| 1096 REGMBC(0x1e83) REGMBC(0x1e85) REGMBC(0x1e87) | |
| 1097 REGMBC(0x1e98) | |
| 1098 return; | |
| 1099 case 'x': CASEMBC(0x1e8b) CASEMBC(0x1e8d) | |
| 1100 regmbc('x'); REGMBC(0x1e8b) REGMBC(0x1e8d) | |
| 167 | 1101 return; |
| 236 | 1102 case 'y': case '\375': case '\377': |
| 2974 | 1103 CASEMBC(0x177) CASEMBC(0x1e8f) CASEMBC(0x1e99) |
| 1104 CASEMBC(0x1ef3) CASEMBC(0x1ef7) CASEMBC(0x1ef9) | |
| 236 | 1105 regmbc('y'); regmbc('\375'); regmbc('\377'); |
| 2974 | 1106 REGMBC(0x177) REGMBC(0x1e8f) REGMBC(0x1e99) |
| 1107 REGMBC(0x1ef3) REGMBC(0x1ef7) REGMBC(0x1ef9) | |
| 1108 return; | |
| 1109 case 'z': CASEMBC(0x17a) CASEMBC(0x17c) CASEMBC(0x17e) | |
| 1110 CASEMBC(0x1b6) CASEMBC(0x1e91) CASEMBC(0x1e95) | |
| 1111 regmbc('z'); REGMBC(0x17a) REGMBC(0x17c) | |
| 1112 REGMBC(0x17e) REGMBC(0x1b6) REGMBC(0x1e91) | |
| 1113 REGMBC(0x1e95) | |
| 167 | 1114 return; |
| 1115 } | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
1116 #endif |
| 167 | 1117 } |
| 1118 regmbc(c); | |
| 1119 } | |
| 1120 | |
| 1121 /* | |
| 1122 * Check for a collating element "[.a.]". "pp" points to the '['. | |
| 1123 * Returns a character. Zero means that no item was recognized. Otherwise | |
| 1124 * "pp" is advanced to after the item. | |
| 1125 * Currently only single characters are recognized! | |
| 1126 */ | |
| 1127 static int | |
| 1128 get_coll_element(pp) | |
| 1129 char_u **pp; | |
| 1130 { | |
| 1131 int c; | |
| 1132 int l = 1; | |
| 1133 char_u *p = *pp; | |
| 1134 | |
| 1135 if (p[1] == '.') | |
| 1136 { | |
| 1137 #ifdef FEAT_MBYTE | |
| 1138 if (has_mbyte) | |
| 474 | 1139 l = (*mb_ptr2len)(p + 2); |
| 167 | 1140 #endif |
| 1141 if (p[l + 2] == '.' && p[l + 3] == ']') | |
| 1142 { | |
| 1143 #ifdef FEAT_MBYTE | |
| 1144 if (has_mbyte) | |
| 1145 c = mb_ptr2char(p + 2); | |
| 1146 else | |
| 1147 #endif | |
| 1148 c = p[2]; | |
| 1149 *pp += l + 4; | |
| 1150 return c; | |
| 1151 } | |
| 1152 } | |
| 1153 return 0; | |
| 1154 } | |
| 1155 | |
| 1156 | |
| 1157 /* | |
| 1158 * Skip over a "[]" range. | |
| 1159 * "p" must point to the character after the '['. | |
| 1160 * The returned pointer is on the matching ']', or the terminating NUL. | |
| 1161 */ | |
| 1162 static char_u * | |
| 1163 skip_anyof(p) | |
| 1164 char_u *p; | |
| 1165 { | |
| 1166 int cpo_lit; /* 'cpoptions' contains 'l' flag */ | |
| 1167 int cpo_bsl; /* 'cpoptions' contains '\' flag */ | |
| 1168 #ifdef FEAT_MBYTE | |
| 1169 int l; | |
| 1170 #endif | |
| 1171 | |
| 410 | 1172 cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; |
| 1173 cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; | |
| 167 | 1174 |
| 1175 if (*p == '^') /* Complement of range. */ | |
| 1176 ++p; | |
| 1177 if (*p == ']' || *p == '-') | |
| 1178 ++p; | |
| 1179 while (*p != NUL && *p != ']') | |
| 1180 { | |
| 1181 #ifdef FEAT_MBYTE | |
| 474 | 1182 if (has_mbyte && (l = (*mb_ptr2len)(p)) > 1) |
| 167 | 1183 p += l; |
| 1184 else | |
| 1185 #endif | |
| 1186 if (*p == '-') | |
| 1187 { | |
| 1188 ++p; | |
| 1189 if (*p != ']' && *p != NUL) | |
| 1190 mb_ptr_adv(p); | |
| 1191 } | |
| 1192 else if (*p == '\\' | |
| 1193 && !cpo_bsl | |
| 1194 && (vim_strchr(REGEXP_INRANGE, p[1]) != NULL | |
| 1195 || (!cpo_lit && vim_strchr(REGEXP_ABBR, p[1]) != NULL))) | |
| 1196 p += 2; | |
| 1197 else if (*p == '[') | |
| 1198 { | |
| 1199 if (get_char_class(&p) == CLASS_NONE | |
| 1200 && get_equi_class(&p) == 0 | |
| 1201 && get_coll_element(&p) == 0) | |
| 1202 ++p; /* It was not a class name */ | |
| 1203 } | |
| 1204 else | |
| 1205 ++p; | |
| 1206 } | |
| 1207 | |
| 1208 return p; | |
| 1209 } | |
| 1210 | |
| 1211 /* | |
| 7 | 1212 * Skip past regular expression. |
| 153 | 1213 * Stop at end of "startp" or where "dirc" is found ('/', '?', etc). |
| 7 | 1214 * Take care of characters with a backslash in front of it. |
| 1215 * Skip strings inside [ and ]. | |
| 1216 * When "newp" is not NULL and "dirc" is '?', make an allocated copy of the | |
| 1217 * expression and change "\?" to "?". If "*newp" is not NULL the expression | |
| 1218 * is changed in-place. | |
| 1219 */ | |
| 1220 char_u * | |
| 1221 skip_regexp(startp, dirc, magic, newp) | |
| 1222 char_u *startp; | |
| 1223 int dirc; | |
| 1224 int magic; | |
| 1225 char_u **newp; | |
| 1226 { | |
| 1227 int mymagic; | |
| 1228 char_u *p = startp; | |
| 1229 | |
| 1230 if (magic) | |
| 1231 mymagic = MAGIC_ON; | |
| 1232 else | |
| 1233 mymagic = MAGIC_OFF; | |
| 1234 | |
| 39 | 1235 for (; p[0] != NUL; mb_ptr_adv(p)) |
| 7 | 1236 { |
| 1237 if (p[0] == dirc) /* found end of regexp */ | |
| 1238 break; | |
| 1239 if ((p[0] == '[' && mymagic >= MAGIC_ON) | |
| 1240 || (p[0] == '\\' && p[1] == '[' && mymagic <= MAGIC_OFF)) | |
| 1241 { | |
| 1242 p = skip_anyof(p + 1); | |
| 1243 if (p[0] == NUL) | |
| 1244 break; | |
| 1245 } | |
| 1246 else if (p[0] == '\\' && p[1] != NUL) | |
| 1247 { | |
| 1248 if (dirc == '?' && newp != NULL && p[1] == '?') | |
| 1249 { | |
| 1250 /* change "\?" to "?", make a copy first. */ | |
| 1251 if (*newp == NULL) | |
| 1252 { | |
| 1253 *newp = vim_strsave(startp); | |
| 1254 if (*newp != NULL) | |
| 1255 p = *newp + (p - startp); | |
| 1256 } | |
| 1257 if (*newp != NULL) | |
| 1621 | 1258 STRMOVE(p, p + 1); |
| 7 | 1259 else |
| 1260 ++p; | |
| 1261 } | |
| 1262 else | |
| 1263 ++p; /* skip next character */ | |
| 1264 if (*p == 'v') | |
| 1265 mymagic = MAGIC_ALL; | |
| 1266 else if (*p == 'V') | |
| 1267 mymagic = MAGIC_NONE; | |
| 1268 } | |
| 1269 } | |
| 1270 return p; | |
| 1271 } | |
| 1272 | |
| 4444 | 1273 static regprog_T *bt_regcomp __ARGS((char_u *expr, int re_flags)); |
| 1274 | |
| 7 | 1275 /* |
| 4444 | 1276 * bt_regcomp() - compile a regular expression into internal code for the |
| 1277 * traditional back track matcher. | |
| 41 | 1278 * Returns the program in allocated space. Returns NULL for an error. |
| 7 | 1279 * |
| 1280 * We can't allocate space until we know how big the compiled form will be, | |
| 1281 * but we can't compile it (and thus know how big it is) until we've got a | |
| 1282 * place to put the code. So we cheat: we compile it twice, once with code | |
| 1283 * generation turned off and size counting turned on, and once "for real". | |
| 1284 * This also means that we don't allocate space until we are sure that the | |
| 1285 * thing really will compile successfully, and we never have to move the | |
| 1286 * code and thus invalidate pointers into it. (Note that it has to be in | |
| 1287 * one piece because vim_free() must be able to free it all.) | |
| 1288 * | |
| 1289 * Whether upper/lower case is to be ignored is decided when executing the | |
| 1290 * program, it does not matter here. | |
| 1291 * | |
| 1292 * Beware that the optimization-preparation code in here knows about some | |
| 1293 * of the structure of the compiled regexp. | |
| 1294 * "re_flags": RE_MAGIC and/or RE_STRING. | |
| 1295 */ | |
| 4444 | 1296 static regprog_T * |
| 1297 bt_regcomp(expr, re_flags) | |
| 7 | 1298 char_u *expr; |
| 1299 int re_flags; | |
| 1300 { | |
| 4444 | 1301 bt_regprog_T *r; |
| 7 | 1302 char_u *scan; |
| 1303 char_u *longest; | |
| 1304 int len; | |
| 1305 int flags; | |
| 1306 | |
| 1307 if (expr == NULL) | |
| 1308 EMSG_RET_NULL(_(e_null)); | |
| 1309 | |
| 1310 init_class_tab(); | |
| 1311 | |
| 1312 /* | |
| 1313 * First pass: determine size, legality. | |
| 1314 */ | |
| 1315 regcomp_start(expr, re_flags); | |
| 1316 regcode = JUST_CALC_SIZE; | |
| 1317 regc(REGMAGIC); | |
| 1318 if (reg(REG_NOPAREN, &flags) == NULL) | |
| 1319 return NULL; | |
| 1320 | |
| 1321 /* Small enough for pointer-storage convention? */ | |
| 1322 #ifdef SMALL_MALLOC /* 16 bit storage allocation */ | |
| 1323 if (regsize >= 65536L - 256L) | |
| 1324 EMSG_RET_NULL(_("E339: Pattern too long")); | |
| 1325 #endif | |
| 1326 | |
| 1327 /* Allocate space. */ | |
| 4444 | 1328 r = (bt_regprog_T *)lalloc(sizeof(bt_regprog_T) + regsize, TRUE); |
| 7 | 1329 if (r == NULL) |
| 1330 return NULL; | |
| 1331 | |
| 1332 /* | |
| 1333 * Second pass: emit code. | |
| 1334 */ | |
| 1335 regcomp_start(expr, re_flags); | |
| 1336 regcode = r->program; | |
| 1337 regc(REGMAGIC); | |
| 2010 | 1338 if (reg(REG_NOPAREN, &flags) == NULL || reg_toolong) |
| 7 | 1339 { |
| 1340 vim_free(r); | |
| 2010 | 1341 if (reg_toolong) |
| 1342 EMSG_RET_NULL(_("E339: Pattern too long")); | |
| 7 | 1343 return NULL; |
| 1344 } | |
| 1345 | |
| 1346 /* Dig out information for optimizations. */ | |
| 1347 r->regstart = NUL; /* Worst-case defaults. */ | |
| 1348 r->reganch = 0; | |
| 1349 r->regmust = NULL; | |
| 1350 r->regmlen = 0; | |
| 1351 r->regflags = regflags; | |
| 1352 if (flags & HASNL) | |
| 1353 r->regflags |= RF_HASNL; | |
| 1354 if (flags & HASLOOKBH) | |
| 1355 r->regflags |= RF_LOOKBH; | |
| 1356 #ifdef FEAT_SYN_HL | |
| 1357 /* Remember whether this pattern has any \z specials in it. */ | |
| 1358 r->reghasz = re_has_z; | |
| 1359 #endif | |
| 1360 scan = r->program + 1; /* First BRANCH. */ | |
| 1361 if (OP(regnext(scan)) == END) /* Only one top-level choice. */ | |
| 1362 { | |
| 1363 scan = OPERAND(scan); | |
| 1364 | |
| 1365 /* Starting-point info. */ | |
| 1366 if (OP(scan) == BOL || OP(scan) == RE_BOF) | |
| 1367 { | |
| 1368 r->reganch++; | |
| 1369 scan = regnext(scan); | |
| 1370 } | |
| 1371 | |
| 1372 if (OP(scan) == EXACTLY) | |
| 1373 { | |
| 1374 #ifdef FEAT_MBYTE | |
| 1375 if (has_mbyte) | |
| 1376 r->regstart = (*mb_ptr2char)(OPERAND(scan)); | |
| 1377 else | |
| 1378 #endif | |
| 1379 r->regstart = *OPERAND(scan); | |
| 1380 } | |
| 1381 else if ((OP(scan) == BOW | |
| 1382 || OP(scan) == EOW | |
| 1383 || OP(scan) == NOTHING | |
| 1384 || OP(scan) == MOPEN + 0 || OP(scan) == NOPEN | |
| 1385 || OP(scan) == MCLOSE + 0 || OP(scan) == NCLOSE) | |
| 1386 && OP(regnext(scan)) == EXACTLY) | |
| 1387 { | |
| 1388 #ifdef FEAT_MBYTE | |
| 1389 if (has_mbyte) | |
| 1390 r->regstart = (*mb_ptr2char)(OPERAND(regnext(scan))); | |
| 1391 else | |
| 1392 #endif | |
| 1393 r->regstart = *OPERAND(regnext(scan)); | |
| 1394 } | |
| 1395 | |
| 1396 /* | |
| 1397 * If there's something expensive in the r.e., find the longest | |
| 1398 * literal string that must appear and make it the regmust. Resolve | |
| 1399 * ties in favor of later strings, since the regstart check works | |
| 1400 * with the beginning of the r.e. and avoiding duplication | |
| 1401 * strengthens checking. Not a strong reason, but sufficient in the | |
| 1402 * absence of others. | |
| 1403 */ | |
| 1404 /* | |
| 1405 * When the r.e. starts with BOW, it is faster to look for a regmust | |
| 1406 * first. Used a lot for "#" and "*" commands. (Added by mool). | |
| 1407 */ | |
| 1408 if ((flags & SPSTART || OP(scan) == BOW || OP(scan) == EOW) | |
| 1409 && !(flags & HASNL)) | |
| 1410 { | |
| 1411 longest = NULL; | |
| 1412 len = 0; | |
| 1413 for (; scan != NULL; scan = regnext(scan)) | |
| 1414 if (OP(scan) == EXACTLY && STRLEN(OPERAND(scan)) >= (size_t)len) | |
| 1415 { | |
| 1416 longest = OPERAND(scan); | |
| 1417 len = (int)STRLEN(OPERAND(scan)); | |
| 1418 } | |
| 1419 r->regmust = longest; | |
| 1420 r->regmlen = len; | |
| 1421 } | |
| 1422 } | |
| 4444 | 1423 #ifdef BT_REGEXP_DUMP |
| 7 | 1424 regdump(expr, r); |
| 1425 #endif | |
| 4444 | 1426 r->engine = &bt_regengine; |
| 1427 return (regprog_T *)r; | |
| 7 | 1428 } |
| 1429 | |
| 1430 /* | |
| 1431 * Setup to parse the regexp. Used once to get the length and once to do it. | |
| 1432 */ | |
| 1433 static void | |
| 1434 regcomp_start(expr, re_flags) | |
| 1435 char_u *expr; | |
| 1436 int re_flags; /* see vim_regcomp() */ | |
| 1437 { | |
| 1438 initchr(expr); | |
| 1439 if (re_flags & RE_MAGIC) | |
| 1440 reg_magic = MAGIC_ON; | |
| 1441 else | |
| 1442 reg_magic = MAGIC_OFF; | |
| 1443 reg_string = (re_flags & RE_STRING); | |
| 481 | 1444 reg_strict = (re_flags & RE_STRICT); |
| 7 | 1445 |
| 1446 num_complex_braces = 0; | |
| 1447 regnpar = 1; | |
| 1448 vim_memset(had_endbrace, 0, sizeof(had_endbrace)); | |
| 1449 #ifdef FEAT_SYN_HL | |
| 1450 regnzpar = 1; | |
| 1451 re_has_z = 0; | |
| 1452 #endif | |
| 1453 regsize = 0L; | |
| 2010 | 1454 reg_toolong = FALSE; |
| 7 | 1455 regflags = 0; |
| 1456 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1457 had_eol = FALSE; | |
| 1458 #endif | |
| 1459 } | |
| 1460 | |
| 1461 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1462 /* | |
| 1463 * Check if during the previous call to vim_regcomp the EOL item "$" has been | |
| 1464 * found. This is messy, but it works fine. | |
| 1465 */ | |
| 1466 int | |
| 1467 vim_regcomp_had_eol() | |
| 1468 { | |
| 1469 return had_eol; | |
| 1470 } | |
| 1471 #endif | |
| 1472 | |
| 1473 /* | |
| 4444 | 1474 * Parse regular expression, i.e. main body or parenthesized thing. |
| 7 | 1475 * |
| 1476 * Caller must absorb opening parenthesis. | |
| 1477 * | |
| 1478 * Combining parenthesis handling with the base level of regular expression | |
| 1479 * is a trifle forced, but the need to tie the tails of the branches to what | |
| 1480 * follows makes it hard to avoid. | |
| 1481 */ | |
| 1482 static char_u * | |
| 1483 reg(paren, flagp) | |
| 1484 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
| 1485 int *flagp; | |
| 1486 { | |
| 1487 char_u *ret; | |
| 1488 char_u *br; | |
| 1489 char_u *ender; | |
| 1490 int parno = 0; | |
| 1491 int flags; | |
| 1492 | |
| 1493 *flagp = HASWIDTH; /* Tentatively. */ | |
| 1494 | |
| 1495 #ifdef FEAT_SYN_HL | |
| 1496 if (paren == REG_ZPAREN) | |
| 1497 { | |
| 1498 /* Make a ZOPEN node. */ | |
| 1499 if (regnzpar >= NSUBEXP) | |
| 1500 EMSG_RET_NULL(_("E50: Too many \\z(")); | |
| 1501 parno = regnzpar; | |
| 1502 regnzpar++; | |
| 1503 ret = regnode(ZOPEN + parno); | |
| 1504 } | |
| 1505 else | |
| 1506 #endif | |
| 1507 if (paren == REG_PAREN) | |
| 1508 { | |
| 1509 /* Make a MOPEN node. */ | |
| 1510 if (regnpar >= NSUBEXP) | |
| 4444 | 1511 EMSG2_RET_NULL(_("E51: Too many %s("), reg_magic == MAGIC_ALL); |
| 7 | 1512 parno = regnpar; |
| 1513 ++regnpar; | |
| 1514 ret = regnode(MOPEN + parno); | |
| 1515 } | |
| 1516 else if (paren == REG_NPAREN) | |
| 1517 { | |
| 1518 /* Make a NOPEN node. */ | |
| 1519 ret = regnode(NOPEN); | |
| 1520 } | |
| 1521 else | |
| 1522 ret = NULL; | |
| 1523 | |
| 1524 /* Pick up the branches, linking them together. */ | |
| 1525 br = regbranch(&flags); | |
| 1526 if (br == NULL) | |
| 1527 return NULL; | |
| 1528 if (ret != NULL) | |
| 1529 regtail(ret, br); /* [MZ]OPEN -> first. */ | |
| 1530 else | |
| 1531 ret = br; | |
| 1532 /* If one of the branches can be zero-width, the whole thing can. | |
| 1533 * If one of the branches has * at start or matches a line-break, the | |
| 1534 * whole thing can. */ | |
| 1535 if (!(flags & HASWIDTH)) | |
| 1536 *flagp &= ~HASWIDTH; | |
| 1537 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1538 while (peekchr() == Magic('|')) | |
| 1539 { | |
| 1540 skipchr(); | |
| 1541 br = regbranch(&flags); | |
| 2010 | 1542 if (br == NULL || reg_toolong) |
| 7 | 1543 return NULL; |
| 1544 regtail(ret, br); /* BRANCH -> BRANCH. */ | |
| 1545 if (!(flags & HASWIDTH)) | |
| 1546 *flagp &= ~HASWIDTH; | |
| 1547 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1548 } | |
| 1549 | |
| 1550 /* Make a closing node, and hook it on the end. */ | |
| 1551 ender = regnode( | |
| 1552 #ifdef FEAT_SYN_HL | |
| 1553 paren == REG_ZPAREN ? ZCLOSE + parno : | |
| 1554 #endif | |
| 1555 paren == REG_PAREN ? MCLOSE + parno : | |
| 1556 paren == REG_NPAREN ? NCLOSE : END); | |
| 1557 regtail(ret, ender); | |
| 1558 | |
| 1559 /* Hook the tails of the branches to the closing node. */ | |
| 1560 for (br = ret; br != NULL; br = regnext(br)) | |
| 1561 regoptail(br, ender); | |
| 1562 | |
| 1563 /* Check for proper termination. */ | |
| 1564 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
| 1565 { | |
| 1566 #ifdef FEAT_SYN_HL | |
| 1567 if (paren == REG_ZPAREN) | |
| 308 | 1568 EMSG_RET_NULL(_("E52: Unmatched \\z(")); |
| 7 | 1569 else |
| 1570 #endif | |
| 1571 if (paren == REG_NPAREN) | |
| 4444 | 1572 EMSG2_RET_NULL(_(e_unmatchedpp), reg_magic == MAGIC_ALL); |
| 7 | 1573 else |
| 4444 | 1574 EMSG2_RET_NULL(_(e_unmatchedp), reg_magic == MAGIC_ALL); |
| 7 | 1575 } |
| 1576 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
| 1577 { | |
| 1578 if (curchr == Magic(')')) | |
| 4444 | 1579 EMSG2_RET_NULL(_(e_unmatchedpar), reg_magic == MAGIC_ALL); |
| 7 | 1580 else |
| 308 | 1581 EMSG_RET_NULL(_(e_trailing)); /* "Can't happen". */ |
| 7 | 1582 /* NOTREACHED */ |
| 1583 } | |
| 1584 /* | |
| 1585 * Here we set the flag allowing back references to this set of | |
| 1586 * parentheses. | |
| 1587 */ | |
| 1588 if (paren == REG_PAREN) | |
| 1589 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
| 1590 return ret; | |
| 1591 } | |
| 1592 | |
| 1593 /* | |
| 4444 | 1594 * Parse one alternative of an | operator. |
| 7 | 1595 * Implements the & operator. |
| 1596 */ | |
| 1597 static char_u * | |
| 1598 regbranch(flagp) | |
| 1599 int *flagp; | |
| 1600 { | |
| 1601 char_u *ret; | |
| 1602 char_u *chain = NULL; | |
| 1603 char_u *latest; | |
| 1604 int flags; | |
| 1605 | |
| 1606 *flagp = WORST | HASNL; /* Tentatively. */ | |
| 1607 | |
| 1608 ret = regnode(BRANCH); | |
| 1609 for (;;) | |
| 1610 { | |
| 1611 latest = regconcat(&flags); | |
| 1612 if (latest == NULL) | |
| 1613 return NULL; | |
| 1614 /* If one of the branches has width, the whole thing has. If one of | |
| 1615 * the branches anchors at start-of-line, the whole thing does. | |
| 1616 * If one of the branches uses look-behind, the whole thing does. */ | |
| 1617 *flagp |= flags & (HASWIDTH | SPSTART | HASLOOKBH); | |
| 1618 /* If one of the branches doesn't match a line-break, the whole thing | |
| 1619 * doesn't. */ | |
| 1620 *flagp &= ~HASNL | (flags & HASNL); | |
| 1621 if (chain != NULL) | |
| 1622 regtail(chain, latest); | |
| 1623 if (peekchr() != Magic('&')) | |
| 1624 break; | |
| 1625 skipchr(); | |
| 1626 regtail(latest, regnode(END)); /* operand ends */ | |
| 2010 | 1627 if (reg_toolong) |
| 1628 break; | |
| 7 | 1629 reginsert(MATCH, latest); |
| 1630 chain = latest; | |
| 1631 } | |
| 1632 | |
| 1633 return ret; | |
| 1634 } | |
| 1635 | |
| 1636 /* | |
| 4444 | 1637 * Parse one alternative of an | or & operator. |
| 7 | 1638 * Implements the concatenation operator. |
| 1639 */ | |
| 1640 static char_u * | |
| 1641 regconcat(flagp) | |
| 1642 int *flagp; | |
| 1643 { | |
| 1644 char_u *first = NULL; | |
| 1645 char_u *chain = NULL; | |
| 1646 char_u *latest; | |
| 1647 int flags; | |
| 1648 int cont = TRUE; | |
| 1649 | |
| 1650 *flagp = WORST; /* Tentatively. */ | |
| 1651 | |
| 1652 while (cont) | |
| 1653 { | |
| 1654 switch (peekchr()) | |
| 1655 { | |
| 1656 case NUL: | |
| 1657 case Magic('|'): | |
| 1658 case Magic('&'): | |
| 1659 case Magic(')'): | |
| 1660 cont = FALSE; | |
| 1661 break; | |
| 1662 case Magic('Z'): | |
| 1663 #ifdef FEAT_MBYTE | |
| 1664 regflags |= RF_ICOMBINE; | |
| 1665 #endif | |
| 1666 skipchr_keepstart(); | |
| 1667 break; | |
| 1668 case Magic('c'): | |
| 1669 regflags |= RF_ICASE; | |
| 1670 skipchr_keepstart(); | |
| 1671 break; | |
| 1672 case Magic('C'): | |
| 1673 regflags |= RF_NOICASE; | |
| 1674 skipchr_keepstart(); | |
| 1675 break; | |
| 1676 case Magic('v'): | |
| 1677 reg_magic = MAGIC_ALL; | |
| 1678 skipchr_keepstart(); | |
| 1679 curchr = -1; | |
| 1680 break; | |
| 1681 case Magic('m'): | |
| 1682 reg_magic = MAGIC_ON; | |
| 1683 skipchr_keepstart(); | |
| 1684 curchr = -1; | |
| 1685 break; | |
| 1686 case Magic('M'): | |
| 1687 reg_magic = MAGIC_OFF; | |
| 1688 skipchr_keepstart(); | |
| 1689 curchr = -1; | |
| 1690 break; | |
| 1691 case Magic('V'): | |
| 1692 reg_magic = MAGIC_NONE; | |
| 1693 skipchr_keepstart(); | |
| 1694 curchr = -1; | |
| 1695 break; | |
| 1696 default: | |
| 1697 latest = regpiece(&flags); | |
| 2010 | 1698 if (latest == NULL || reg_toolong) |
| 7 | 1699 return NULL; |
| 1700 *flagp |= flags & (HASWIDTH | HASNL | HASLOOKBH); | |
| 1701 if (chain == NULL) /* First piece. */ | |
| 1702 *flagp |= flags & SPSTART; | |
| 1703 else | |
| 1704 regtail(chain, latest); | |
| 1705 chain = latest; | |
| 1706 if (first == NULL) | |
| 1707 first = latest; | |
| 1708 break; | |
| 1709 } | |
| 1710 } | |
| 1711 if (first == NULL) /* Loop ran zero times. */ | |
| 1712 first = regnode(NOTHING); | |
| 1713 return first; | |
| 1714 } | |
| 1715 | |
| 1716 /* | |
| 4444 | 1717 * Parse something followed by possible [*+=]. |
| 7 | 1718 * |
| 1719 * Note that the branching code sequences used for = and the general cases | |
| 1720 * of * and + are somewhat optimized: they use the same NOTHING node as | |
| 1721 * both the endmarker for their branch list and the body of the last branch. | |
| 1722 * It might seem that this node could be dispensed with entirely, but the | |
| 1723 * endmarker role is not redundant. | |
| 1724 */ | |
| 1725 static char_u * | |
| 1726 regpiece(flagp) | |
| 1727 int *flagp; | |
| 1728 { | |
| 1729 char_u *ret; | |
| 1730 int op; | |
| 1731 char_u *next; | |
| 1732 int flags; | |
| 1733 long minval; | |
| 1734 long maxval; | |
| 1735 | |
| 1736 ret = regatom(&flags); | |
| 1737 if (ret == NULL) | |
| 1738 return NULL; | |
| 1739 | |
| 1740 op = peekchr(); | |
| 1741 if (re_multi_type(op) == NOT_MULTI) | |
| 1742 { | |
| 1743 *flagp = flags; | |
| 1744 return ret; | |
| 1745 } | |
| 1746 /* default flags */ | |
| 1747 *flagp = (WORST | SPSTART | (flags & (HASNL | HASLOOKBH))); | |
| 1748 | |
| 1749 skipchr(); | |
| 1750 switch (op) | |
| 1751 { | |
| 1752 case Magic('*'): | |
| 1753 if (flags & SIMPLE) | |
| 1754 reginsert(STAR, ret); | |
| 1755 else | |
| 1756 { | |
| 1757 /* Emit x* as (x&|), where & means "self". */ | |
| 1758 reginsert(BRANCH, ret); /* Either x */ | |
| 1759 regoptail(ret, regnode(BACK)); /* and loop */ | |
| 1760 regoptail(ret, ret); /* back */ | |
| 1761 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1762 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1763 } | |
| 1764 break; | |
| 1765 | |
| 1766 case Magic('+'): | |
| 1767 if (flags & SIMPLE) | |
| 1768 reginsert(PLUS, ret); | |
| 1769 else | |
| 1770 { | |
| 1771 /* Emit x+ as x(&|), where & means "self". */ | |
| 1772 next = regnode(BRANCH); /* Either */ | |
| 1773 regtail(ret, next); | |
| 233 | 1774 regtail(regnode(BACK), ret); /* loop back */ |
| 7 | 1775 regtail(next, regnode(BRANCH)); /* or */ |
| 1776 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1777 } | |
| 1778 *flagp = (WORST | HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1779 break; | |
| 1780 | |
| 1781 case Magic('@'): | |
| 1782 { | |
| 1783 int lop = END; | |
| 1784 | |
| 1785 switch (no_Magic(getchr())) | |
| 1786 { | |
| 1787 case '=': lop = MATCH; break; /* \@= */ | |
| 1788 case '!': lop = NOMATCH; break; /* \@! */ | |
| 1789 case '>': lop = SUBPAT; break; /* \@> */ | |
| 1790 case '<': switch (no_Magic(getchr())) | |
| 1791 { | |
| 1792 case '=': lop = BEHIND; break; /* \@<= */ | |
| 1793 case '!': lop = NOBEHIND; break; /* \@<! */ | |
| 1794 } | |
| 1795 } | |
| 1796 if (lop == END) | |
| 4444 | 1797 EMSG2_RET_NULL(_("E59: invalid character after %s@"), |
| 7 | 1798 reg_magic == MAGIC_ALL); |
| 1799 /* Look behind must match with behind_pos. */ | |
| 1800 if (lop == BEHIND || lop == NOBEHIND) | |
| 1801 { | |
| 1802 regtail(ret, regnode(BHPOS)); | |
| 1803 *flagp |= HASLOOKBH; | |
| 1804 } | |
| 1805 regtail(ret, regnode(END)); /* operand ends */ | |
| 1806 reginsert(lop, ret); | |
| 1807 break; | |
| 1808 } | |
| 1809 | |
| 1810 case Magic('?'): | |
| 1811 case Magic('='): | |
| 1812 /* Emit x= as (x|) */ | |
| 1813 reginsert(BRANCH, ret); /* Either x */ | |
| 1814 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1815 next = regnode(NOTHING); /* null. */ | |
| 1816 regtail(ret, next); | |
| 1817 regoptail(ret, next); | |
| 1818 break; | |
| 1819 | |
| 1820 case Magic('{'): | |
| 1821 if (!read_limits(&minval, &maxval)) | |
| 1822 return NULL; | |
| 1823 if (flags & SIMPLE) | |
| 1824 { | |
| 1825 reginsert(BRACE_SIMPLE, ret); | |
| 1826 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1827 } | |
| 1828 else | |
| 1829 { | |
| 1830 if (num_complex_braces >= 10) | |
| 4444 | 1831 EMSG2_RET_NULL(_("E60: Too many complex %s{...}s"), |
| 7 | 1832 reg_magic == MAGIC_ALL); |
| 1833 reginsert(BRACE_COMPLEX + num_complex_braces, ret); | |
| 1834 regoptail(ret, regnode(BACK)); | |
| 1835 regoptail(ret, ret); | |
| 1836 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1837 ++num_complex_braces; | |
| 1838 } | |
| 1839 if (minval > 0 && maxval > 0) | |
| 1840 *flagp = (HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1841 break; | |
| 1842 } | |
| 1843 if (re_multi_type(peekchr()) != NOT_MULTI) | |
| 1844 { | |
| 1845 /* Can't have a multi follow a multi. */ | |
| 1846 if (peekchr() == Magic('*')) | |
| 1847 sprintf((char *)IObuff, _("E61: Nested %s*"), | |
| 1848 reg_magic >= MAGIC_ON ? "" : "\\"); | |
| 1849 else | |
| 1850 sprintf((char *)IObuff, _("E62: Nested %s%c"), | |
| 1851 reg_magic == MAGIC_ALL ? "" : "\\", no_Magic(peekchr())); | |
| 1852 EMSG_RET_NULL(IObuff); | |
| 1853 } | |
| 1854 | |
| 1855 return ret; | |
| 1856 } | |
| 1857 | |
| 4444 | 1858 /* When making changes to classchars also change nfa_classcodes. */ |
| 1859 static char_u *classchars = (char_u *)".iIkKfFpPsSdDxXoOwWhHaAlLuU"; | |
| 1860 static int classcodes[] = { | |
| 1861 ANY, IDENT, SIDENT, KWORD, SKWORD, | |
| 1862 FNAME, SFNAME, PRINT, SPRINT, | |
| 1863 WHITE, NWHITE, DIGIT, NDIGIT, | |
| 1864 HEX, NHEX, OCTAL, NOCTAL, | |
| 1865 WORD, NWORD, HEAD, NHEAD, | |
| 1866 ALPHA, NALPHA, LOWER, NLOWER, | |
| 1867 UPPER, NUPPER | |
| 1868 }; | |
| 1869 | |
| 7 | 1870 /* |
| 4444 | 1871 * Parse the lowest level. |
| 7 | 1872 * |
| 1873 * Optimization: gobbles an entire sequence of ordinary characters so that | |
| 1874 * it can turn them into a single node, which is smaller to store and | |
| 1875 * faster to run. Don't do this when one_exactly is set. | |
| 1876 */ | |
| 1877 static char_u * | |
| 1878 regatom(flagp) | |
| 1879 int *flagp; | |
| 1880 { | |
| 1881 char_u *ret; | |
| 1882 int flags; | |
| 1883 int cpo_lit; /* 'cpoptions' contains 'l' flag */ | |
| 167 | 1884 int cpo_bsl; /* 'cpoptions' contains '\' flag */ |
| 7 | 1885 int c; |
| 1886 char_u *p; | |
| 1887 int extra = 0; | |
| 1888 | |
| 1889 *flagp = WORST; /* Tentatively. */ | |
| 410 | 1890 cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; |
| 1891 cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; | |
| 7 | 1892 |
| 1893 c = getchr(); | |
| 1894 switch (c) | |
| 1895 { | |
| 1896 case Magic('^'): | |
| 1897 ret = regnode(BOL); | |
| 1898 break; | |
| 1899 | |
| 1900 case Magic('$'): | |
| 1901 ret = regnode(EOL); | |
| 1902 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1903 had_eol = TRUE; | |
| 1904 #endif | |
| 1905 break; | |
| 1906 | |
| 1907 case Magic('<'): | |
| 1908 ret = regnode(BOW); | |
| 1909 break; | |
| 1910 | |
| 1911 case Magic('>'): | |
| 1912 ret = regnode(EOW); | |
| 1913 break; | |
| 1914 | |
| 1915 case Magic('_'): | |
| 1916 c = no_Magic(getchr()); | |
| 1917 if (c == '^') /* "\_^" is start-of-line */ | |
| 1918 { | |
| 1919 ret = regnode(BOL); | |
| 1920 break; | |
| 1921 } | |
| 1922 if (c == '$') /* "\_$" is end-of-line */ | |
| 1923 { | |
| 1924 ret = regnode(EOL); | |
| 1925 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1926 had_eol = TRUE; | |
| 1927 #endif | |
| 1928 break; | |
| 1929 } | |
| 1930 | |
| 1931 extra = ADD_NL; | |
| 1932 *flagp |= HASNL; | |
| 1933 | |
| 1934 /* "\_[" is character range plus newline */ | |
| 1935 if (c == '[') | |
| 1936 goto collection; | |
| 1937 | |
| 1938 /* "\_x" is character class plus newline */ | |
| 1939 /*FALLTHROUGH*/ | |
| 1940 | |
| 1941 /* | |
| 1942 * Character classes. | |
| 1943 */ | |
| 1944 case Magic('.'): | |
| 1945 case Magic('i'): | |
| 1946 case Magic('I'): | |
| 1947 case Magic('k'): | |
| 1948 case Magic('K'): | |
| 1949 case Magic('f'): | |
| 1950 case Magic('F'): | |
| 1951 case Magic('p'): | |
| 1952 case Magic('P'): | |
| 1953 case Magic('s'): | |
| 1954 case Magic('S'): | |
| 1955 case Magic('d'): | |
| 1956 case Magic('D'): | |
| 1957 case Magic('x'): | |
| 1958 case Magic('X'): | |
| 1959 case Magic('o'): | |
| 1960 case Magic('O'): | |
| 1961 case Magic('w'): | |
| 1962 case Magic('W'): | |
| 1963 case Magic('h'): | |
| 1964 case Magic('H'): | |
| 1965 case Magic('a'): | |
| 1966 case Magic('A'): | |
| 1967 case Magic('l'): | |
| 1968 case Magic('L'): | |
| 1969 case Magic('u'): | |
| 1970 case Magic('U'): | |
| 1971 p = vim_strchr(classchars, no_Magic(c)); | |
| 1972 if (p == NULL) | |
| 1973 EMSG_RET_NULL(_("E63: invalid use of \\_")); | |
| 714 | 1974 #ifdef FEAT_MBYTE |
| 1975 /* When '.' is followed by a composing char ignore the dot, so that | |
| 1976 * the composing char is matched here. */ | |
| 1977 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
| 1978 { | |
| 1979 c = getchr(); | |
| 1980 goto do_multibyte; | |
| 1981 } | |
| 1982 #endif | |
| 7 | 1983 ret = regnode(classcodes[p - classchars] + extra); |
| 1984 *flagp |= HASWIDTH | SIMPLE; | |
| 1985 break; | |
| 1986 | |
| 1987 case Magic('n'): | |
| 1988 if (reg_string) | |
| 1989 { | |
| 1990 /* In a string "\n" matches a newline character. */ | |
| 1991 ret = regnode(EXACTLY); | |
| 1992 regc(NL); | |
| 1993 regc(NUL); | |
| 1994 *flagp |= HASWIDTH | SIMPLE; | |
| 1995 } | |
| 1996 else | |
| 1997 { | |
| 1998 /* In buffer text "\n" matches the end of a line. */ | |
| 1999 ret = regnode(NEWL); | |
| 2000 *flagp |= HASWIDTH | HASNL; | |
| 2001 } | |
| 2002 break; | |
| 2003 | |
| 2004 case Magic('('): | |
| 2005 if (one_exactly) | |
| 2006 EMSG_ONE_RET_NULL; | |
| 2007 ret = reg(REG_PAREN, &flags); | |
| 2008 if (ret == NULL) | |
| 2009 return NULL; | |
| 2010 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 2011 break; | |
| 2012 | |
| 2013 case NUL: | |
| 2014 case Magic('|'): | |
| 2015 case Magic('&'): | |
| 2016 case Magic(')'): | |
| 1468 | 2017 if (one_exactly) |
| 2018 EMSG_ONE_RET_NULL; | |
| 7 | 2019 EMSG_RET_NULL(_(e_internal)); /* Supposed to be caught earlier. */ |
| 2020 /* NOTREACHED */ | |
| 2021 | |
| 2022 case Magic('='): | |
| 2023 case Magic('?'): | |
| 2024 case Magic('+'): | |
| 2025 case Magic('@'): | |
| 2026 case Magic('{'): | |
| 2027 case Magic('*'): | |
| 2028 c = no_Magic(c); | |
| 2029 sprintf((char *)IObuff, _("E64: %s%c follows nothing"), | |
| 2030 (c == '*' ? reg_magic >= MAGIC_ON : reg_magic == MAGIC_ALL) | |
| 2031 ? "" : "\\", c); | |
| 2032 EMSG_RET_NULL(IObuff); | |
| 2033 /* NOTREACHED */ | |
| 2034 | |
| 2035 case Magic('~'): /* previous substitute pattern */ | |
| 359 | 2036 if (reg_prev_sub != NULL) |
| 7 | 2037 { |
| 2038 char_u *lp; | |
| 2039 | |
| 2040 ret = regnode(EXACTLY); | |
| 2041 lp = reg_prev_sub; | |
| 2042 while (*lp != NUL) | |
| 2043 regc(*lp++); | |
| 2044 regc(NUL); | |
| 2045 if (*reg_prev_sub != NUL) | |
| 2046 { | |
| 2047 *flagp |= HASWIDTH; | |
| 2048 if ((lp - reg_prev_sub) == 1) | |
| 2049 *flagp |= SIMPLE; | |
| 2050 } | |
| 2051 } | |
| 2052 else | |
| 2053 EMSG_RET_NULL(_(e_nopresub)); | |
| 2054 break; | |
| 2055 | |
| 2056 case Magic('1'): | |
| 2057 case Magic('2'): | |
| 2058 case Magic('3'): | |
| 2059 case Magic('4'): | |
| 2060 case Magic('5'): | |
| 2061 case Magic('6'): | |
| 2062 case Magic('7'): | |
| 2063 case Magic('8'): | |
| 2064 case Magic('9'): | |
| 2065 { | |
| 2066 int refnum; | |
| 2067 | |
| 2068 refnum = c - Magic('0'); | |
| 2069 /* | |
| 2070 * Check if the back reference is legal. We must have seen the | |
| 2071 * close brace. | |
| 2072 * TODO: Should also check that we don't refer to something | |
| 2073 * that is repeated (+*=): what instance of the repetition | |
| 2074 * should we match? | |
| 2075 */ | |
| 2076 if (!had_endbrace[refnum]) | |
| 2077 { | |
| 2078 /* Trick: check if "@<=" or "@<!" follows, in which case | |
| 2079 * the \1 can appear before the referenced match. */ | |
| 2080 for (p = regparse; *p != NUL; ++p) | |
| 2081 if (p[0] == '@' && p[1] == '<' | |
| 2082 && (p[2] == '!' || p[2] == '=')) | |
| 2083 break; | |
| 2084 if (*p == NUL) | |
| 2085 EMSG_RET_NULL(_("E65: Illegal back reference")); | |
| 2086 } | |
| 2087 ret = regnode(BACKREF + refnum); | |
| 2088 } | |
| 2089 break; | |
| 2090 | |
| 2091 case Magic('z'): | |
| 2092 { | |
| 2093 c = no_Magic(getchr()); | |
| 2094 switch (c) | |
| 2095 { | |
| 741 | 2096 #ifdef FEAT_SYN_HL |
| 7 | 2097 case '(': if (reg_do_extmatch != REX_SET) |
| 2098 EMSG_RET_NULL(_("E66: \\z( not allowed here")); | |
| 2099 if (one_exactly) | |
| 2100 EMSG_ONE_RET_NULL; | |
| 2101 ret = reg(REG_ZPAREN, &flags); | |
| 2102 if (ret == NULL) | |
| 2103 return NULL; | |
| 2104 *flagp |= flags & (HASWIDTH|SPSTART|HASNL|HASLOOKBH); | |
| 2105 re_has_z = REX_SET; | |
| 2106 break; | |
| 2107 | |
| 2108 case '1': | |
| 2109 case '2': | |
| 2110 case '3': | |
| 2111 case '4': | |
| 2112 case '5': | |
| 2113 case '6': | |
| 2114 case '7': | |
| 2115 case '8': | |
| 2116 case '9': if (reg_do_extmatch != REX_USE) | |
| 2117 EMSG_RET_NULL(_("E67: \\z1 et al. not allowed here")); | |
| 2118 ret = regnode(ZREF + c - '0'); | |
| 2119 re_has_z = REX_USE; | |
| 2120 break; | |
| 741 | 2121 #endif |
| 7 | 2122 |
| 2123 case 's': ret = regnode(MOPEN + 0); | |
| 2124 break; | |
| 2125 | |
| 2126 case 'e': ret = regnode(MCLOSE + 0); | |
| 2127 break; | |
| 2128 | |
| 2129 default: EMSG_RET_NULL(_("E68: Invalid character after \\z")); | |
| 2130 } | |
| 2131 } | |
| 2132 break; | |
| 2133 | |
| 2134 case Magic('%'): | |
| 2135 { | |
| 2136 c = no_Magic(getchr()); | |
| 2137 switch (c) | |
| 2138 { | |
| 2139 /* () without a back reference */ | |
| 2140 case '(': | |
| 2141 if (one_exactly) | |
| 2142 EMSG_ONE_RET_NULL; | |
| 2143 ret = reg(REG_NPAREN, &flags); | |
| 2144 if (ret == NULL) | |
| 2145 return NULL; | |
| 2146 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 2147 break; | |
| 2148 | |
| 2149 /* Catch \%^ and \%$ regardless of where they appear in the | |
| 2150 * pattern -- regardless of whether or not it makes sense. */ | |
| 2151 case '^': | |
| 2152 ret = regnode(RE_BOF); | |
| 2153 break; | |
| 2154 | |
| 2155 case '$': | |
| 2156 ret = regnode(RE_EOF); | |
| 2157 break; | |
| 2158 | |
| 2159 case '#': | |
| 2160 ret = regnode(CURSOR); | |
| 2161 break; | |
| 2162 | |
| 639 | 2163 case 'V': |
| 2164 ret = regnode(RE_VISUAL); | |
| 2165 break; | |
| 2166 | |
| 7 | 2167 /* \%[abc]: Emit as a list of branches, all ending at the last |
| 2168 * branch which matches nothing. */ | |
| 2169 case '[': | |
| 2170 if (one_exactly) /* doesn't nest */ | |
| 2171 EMSG_ONE_RET_NULL; | |
| 2172 { | |
| 2173 char_u *lastbranch; | |
| 2174 char_u *lastnode = NULL; | |
| 2175 char_u *br; | |
| 2176 | |
| 2177 ret = NULL; | |
| 2178 while ((c = getchr()) != ']') | |
| 2179 { | |
| 2180 if (c == NUL) | |
| 4444 | 2181 EMSG2_RET_NULL(_("E69: Missing ] after %s%%["), |
| 7 | 2182 reg_magic == MAGIC_ALL); |
| 2183 br = regnode(BRANCH); | |
| 2184 if (ret == NULL) | |
| 2185 ret = br; | |
| 2186 else | |
| 2187 regtail(lastnode, br); | |
| 2188 | |
| 2189 ungetchr(); | |
| 2190 one_exactly = TRUE; | |
| 2191 lastnode = regatom(flagp); | |
| 2192 one_exactly = FALSE; | |
| 2193 if (lastnode == NULL) | |
| 2194 return NULL; | |
| 2195 } | |
| 2196 if (ret == NULL) | |
| 4444 | 2197 EMSG2_RET_NULL(_("E70: Empty %s%%[]"), |
| 7 | 2198 reg_magic == MAGIC_ALL); |
| 2199 lastbranch = regnode(BRANCH); | |
| 2200 br = regnode(NOTHING); | |
| 2201 if (ret != JUST_CALC_SIZE) | |
| 2202 { | |
| 2203 regtail(lastnode, br); | |
| 2204 regtail(lastbranch, br); | |
| 2205 /* connect all branches to the NOTHING | |
| 2206 * branch at the end */ | |
| 2207 for (br = ret; br != lastnode; ) | |
| 2208 { | |
| 2209 if (OP(br) == BRANCH) | |
| 2210 { | |
| 2211 regtail(br, lastbranch); | |
| 2212 br = OPERAND(br); | |
| 2213 } | |
| 2214 else | |
| 2215 br = regnext(br); | |
| 2216 } | |
| 2217 } | |
| 1701 | 2218 *flagp &= ~(HASWIDTH | SIMPLE); |
| 7 | 2219 break; |
| 2220 } | |
| 2221 | |
| 24 | 2222 case 'd': /* %d123 decimal */ |
| 2223 case 'o': /* %o123 octal */ | |
| 2224 case 'x': /* %xab hex 2 */ | |
| 2225 case 'u': /* %uabcd hex 4 */ | |
| 2226 case 'U': /* %U1234abcd hex 8 */ | |
| 2227 { | |
| 2228 int i; | |
| 2229 | |
| 2230 switch (c) | |
| 2231 { | |
| 2232 case 'd': i = getdecchrs(); break; | |
| 2233 case 'o': i = getoctchrs(); break; | |
| 2234 case 'x': i = gethexchrs(2); break; | |
| 2235 case 'u': i = gethexchrs(4); break; | |
| 2236 case 'U': i = gethexchrs(8); break; | |
| 2237 default: i = -1; break; | |
| 2238 } | |
| 2239 | |
| 2240 if (i < 0) | |
| 4444 | 2241 EMSG2_RET_NULL( |
| 24 | 2242 _("E678: Invalid character after %s%%[dxouU]"), |
| 2243 reg_magic == MAGIC_ALL); | |
| 714 | 2244 #ifdef FEAT_MBYTE |
| 2245 if (use_multibytecode(i)) | |
| 2246 ret = regnode(MULTIBYTECODE); | |
| 2247 else | |
| 2248 #endif | |
| 2249 ret = regnode(EXACTLY); | |
| 24 | 2250 if (i == 0) |
| 2251 regc(0x0a); | |
| 2252 else | |
| 2253 #ifdef FEAT_MBYTE | |
| 2254 regmbc(i); | |
| 2255 #else | |
| 2256 regc(i); | |
| 2257 #endif | |
| 2258 regc(NUL); | |
| 2259 *flagp |= HASWIDTH; | |
| 2260 break; | |
| 2261 } | |
| 2262 | |
| 7 | 2263 default: |
| 639 | 2264 if (VIM_ISDIGIT(c) || c == '<' || c == '>' |
| 2265 || c == '\'') | |
| 7 | 2266 { |
| 2267 long_u n = 0; | |
| 2268 int cmp; | |
| 2269 | |
| 2270 cmp = c; | |
| 2271 if (cmp == '<' || cmp == '>') | |
| 2272 c = getchr(); | |
| 2273 while (VIM_ISDIGIT(c)) | |
| 2274 { | |
| 2275 n = n * 10 + (c - '0'); | |
| 2276 c = getchr(); | |
| 2277 } | |
| 639 | 2278 if (c == '\'' && n == 0) |
| 2279 { | |
| 2280 /* "\%'m", "\%<'m" and "\%>'m": Mark */ | |
| 2281 c = getchr(); | |
| 2282 ret = regnode(RE_MARK); | |
| 2283 if (ret == JUST_CALC_SIZE) | |
| 2284 regsize += 2; | |
| 2285 else | |
| 2286 { | |
| 2287 *regcode++ = c; | |
| 2288 *regcode++ = cmp; | |
| 2289 } | |
| 2290 break; | |
| 2291 } | |
| 2292 else if (c == 'l' || c == 'c' || c == 'v') | |
| 7 | 2293 { |
| 2294 if (c == 'l') | |
| 2295 ret = regnode(RE_LNUM); | |
| 2296 else if (c == 'c') | |
| 2297 ret = regnode(RE_COL); | |
| 2298 else | |
| 2299 ret = regnode(RE_VCOL); | |
| 2300 if (ret == JUST_CALC_SIZE) | |
| 2301 regsize += 5; | |
| 2302 else | |
| 2303 { | |
| 2304 /* put the number and the optional | |
| 2305 * comparator after the opcode */ | |
| 2306 regcode = re_put_long(regcode, n); | |
| 2307 *regcode++ = cmp; | |
| 2308 } | |
| 2309 break; | |
| 2310 } | |
| 2311 } | |
| 2312 | |
| 4444 | 2313 EMSG2_RET_NULL(_("E71: Invalid character after %s%%"), |
| 7 | 2314 reg_magic == MAGIC_ALL); |
| 2315 } | |
| 2316 } | |
| 2317 break; | |
| 2318 | |
| 2319 case Magic('['): | |
| 2320 collection: | |
| 2321 { | |
| 2322 char_u *lp; | |
| 2323 | |
| 2324 /* | |
| 2325 * If there is no matching ']', we assume the '[' is a normal | |
| 2326 * character. This makes 'incsearch' and ":help [" work. | |
| 2327 */ | |
| 2328 lp = skip_anyof(regparse); | |
| 2329 if (*lp == ']') /* there is a matching ']' */ | |
| 2330 { | |
| 2331 int startc = -1; /* > 0 when next '-' is a range */ | |
| 2332 int endc; | |
| 2333 | |
| 2334 /* | |
| 2335 * In a character class, different parsing rules apply. | |
| 2336 * Not even \ is special anymore, nothing is. | |
| 2337 */ | |
| 2338 if (*regparse == '^') /* Complement of range. */ | |
| 2339 { | |
| 2340 ret = regnode(ANYBUT + extra); | |
| 2341 regparse++; | |
| 2342 } | |
| 2343 else | |
| 2344 ret = regnode(ANYOF + extra); | |
| 2345 | |
| 2346 /* At the start ']' and '-' mean the literal character. */ | |
| 2347 if (*regparse == ']' || *regparse == '-') | |
| 167 | 2348 { |
| 2349 startc = *regparse; | |
| 7 | 2350 regc(*regparse++); |
| 167 | 2351 } |
| 7 | 2352 |
| 2353 while (*regparse != NUL && *regparse != ']') | |
| 2354 { | |
| 2355 if (*regparse == '-') | |
| 2356 { | |
| 2357 ++regparse; | |
| 2358 /* The '-' is not used for a range at the end and | |
| 2359 * after or before a '\n'. */ | |
| 2360 if (*regparse == ']' || *regparse == NUL | |
| 2361 || startc == -1 | |
| 2362 || (regparse[0] == '\\' && regparse[1] == 'n')) | |
| 2363 { | |
| 2364 regc('-'); | |
| 2365 startc = '-'; /* [--x] is a range */ | |
| 2366 } | |
| 2367 else | |
| 2368 { | |
| 167 | 2369 /* Also accept "a-[.z.]" */ |
| 2370 endc = 0; | |
| 2371 if (*regparse == '[') | |
| 2372 endc = get_coll_element(®parse); | |
| 2373 if (endc == 0) | |
| 2374 { | |
| 7 | 2375 #ifdef FEAT_MBYTE |
| 167 | 2376 if (has_mbyte) |
| 2377 endc = mb_ptr2char_adv(®parse); | |
| 2378 else | |
| 7 | 2379 #endif |
| 167 | 2380 endc = *regparse++; |
| 2381 } | |
| 24 | 2382 |
| 2383 /* Handle \o40, \x20 and \u20AC style sequences */ | |
| 167 | 2384 if (endc == '\\' && !cpo_lit && !cpo_bsl) |
| 24 | 2385 endc = coll_get_char(); |
| 2386 | |
| 7 | 2387 if (startc > endc) |
| 2388 EMSG_RET_NULL(_(e_invrange)); | |
| 2389 #ifdef FEAT_MBYTE | |
| 2390 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
| 2391 || (*mb_char2len)(endc) > 1)) | |
| 2392 { | |
| 2393 /* Limit to a range of 256 chars */ | |
| 2394 if (endc > startc + 256) | |
| 2395 EMSG_RET_NULL(_(e_invrange)); | |
| 2396 while (++startc <= endc) | |
| 2397 regmbc(startc); | |
| 2398 } | |
| 2399 else | |
| 2400 #endif | |
| 2401 { | |
| 2402 #ifdef EBCDIC | |
| 2403 int alpha_only = FALSE; | |
| 2404 | |
| 2405 /* for alphabetical range skip the gaps | |
| 2406 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
| 2407 if (isalpha(startc) && isalpha(endc)) | |
| 2408 alpha_only = TRUE; | |
| 2409 #endif | |
| 2410 while (++startc <= endc) | |
| 2411 #ifdef EBCDIC | |
| 2412 if (!alpha_only || isalpha(startc)) | |
| 2413 #endif | |
| 2414 regc(startc); | |
| 2415 } | |
| 2416 startc = -1; | |
| 2417 } | |
| 2418 } | |
| 2419 /* | |
| 2420 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
| 2421 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
| 2422 * 'cpoptions' is not included. | |
| 167 | 2423 * Posix doesn't recognize backslash at all. |
| 7 | 2424 */ |
| 2425 else if (*regparse == '\\' | |
| 167 | 2426 && !cpo_bsl |
| 7 | 2427 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL |
| 2428 || (!cpo_lit | |
| 2429 && vim_strchr(REGEXP_ABBR, | |
| 2430 regparse[1]) != NULL))) | |
| 2431 { | |
| 2432 regparse++; | |
| 2433 if (*regparse == 'n') | |
| 2434 { | |
| 2435 /* '\n' in range: also match NL */ | |
| 2436 if (ret != JUST_CALC_SIZE) | |
| 2437 { | |
| 4084 | 2438 /* Using \n inside [^] does not change what |
| 2439 * matches. "[^\n]" is the same as ".". */ | |
| 2440 if (*ret == ANYOF) | |
| 2441 { | |
| 7 | 2442 *ret = ANYOF + ADD_NL; |
| 4084 | 2443 *flagp |= HASNL; |
| 2444 } | |
| 7 | 2445 /* else: must have had a \n already */ |
| 2446 } | |
| 2447 regparse++; | |
| 2448 startc = -1; | |
| 2449 } | |
| 24 | 2450 else if (*regparse == 'd' |
| 2451 || *regparse == 'o' | |
| 2452 || *regparse == 'x' | |
| 2453 || *regparse == 'u' | |
| 2454 || *regparse == 'U') | |
| 2455 { | |
| 2456 startc = coll_get_char(); | |
| 2457 if (startc == 0) | |
| 2458 regc(0x0a); | |
| 2459 else | |
| 2460 #ifdef FEAT_MBYTE | |
| 2461 regmbc(startc); | |
| 2462 #else | |
| 2463 regc(startc); | |
| 2464 #endif | |
| 2465 } | |
| 7 | 2466 else |
| 2467 { | |
| 2468 startc = backslash_trans(*regparse++); | |
| 2469 regc(startc); | |
| 2470 } | |
| 2471 } | |
| 2472 else if (*regparse == '[') | |
| 2473 { | |
| 2474 int c_class; | |
| 2475 int cu; | |
| 2476 | |
| 167 | 2477 c_class = get_char_class(®parse); |
| 7 | 2478 startc = -1; |
| 2479 /* Characters assumed to be 8 bits! */ | |
| 2480 switch (c_class) | |
| 2481 { | |
| 2482 case CLASS_NONE: | |
| 167 | 2483 c_class = get_equi_class(®parse); |
| 2484 if (c_class != 0) | |
| 2485 { | |
| 2486 /* produce equivalence class */ | |
| 2487 reg_equi_class(c_class); | |
| 2488 } | |
| 2489 else if ((c_class = | |
| 2490 get_coll_element(®parse)) != 0) | |
| 2491 { | |
| 2492 /* produce a collating element */ | |
| 2493 regmbc(c_class); | |
| 2494 } | |
| 2495 else | |
| 2496 { | |
| 2497 /* literal '[', allow [[-x] as a range */ | |
| 2498 startc = *regparse++; | |
| 2499 regc(startc); | |
| 2500 } | |
| 7 | 2501 break; |
| 2502 case CLASS_ALNUM: | |
| 2503 for (cu = 1; cu <= 255; cu++) | |
| 2504 if (isalnum(cu)) | |
| 2505 regc(cu); | |
| 2506 break; | |
| 2507 case CLASS_ALPHA: | |
| 2508 for (cu = 1; cu <= 255; cu++) | |
| 2509 if (isalpha(cu)) | |
| 2510 regc(cu); | |
| 2511 break; | |
| 2512 case CLASS_BLANK: | |
| 2513 regc(' '); | |
| 2514 regc('\t'); | |
| 2515 break; | |
| 2516 case CLASS_CNTRL: | |
| 2517 for (cu = 1; cu <= 255; cu++) | |
| 2518 if (iscntrl(cu)) | |
| 2519 regc(cu); | |
| 2520 break; | |
| 2521 case CLASS_DIGIT: | |
| 2522 for (cu = 1; cu <= 255; cu++) | |
| 2523 if (VIM_ISDIGIT(cu)) | |
| 2524 regc(cu); | |
| 2525 break; | |
| 2526 case CLASS_GRAPH: | |
| 2527 for (cu = 1; cu <= 255; cu++) | |
| 2528 if (isgraph(cu)) | |
| 2529 regc(cu); | |
| 2530 break; | |
| 2531 case CLASS_LOWER: | |
| 2532 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2533 if (MB_ISLOWER(cu)) |
| 7 | 2534 regc(cu); |
| 2535 break; | |
| 2536 case CLASS_PRINT: | |
| 2537 for (cu = 1; cu <= 255; cu++) | |
| 2538 if (vim_isprintc(cu)) | |
| 2539 regc(cu); | |
| 2540 break; | |
| 2541 case CLASS_PUNCT: | |
| 2542 for (cu = 1; cu <= 255; cu++) | |
| 2543 if (ispunct(cu)) | |
| 2544 regc(cu); | |
| 2545 break; | |
| 2546 case CLASS_SPACE: | |
| 2547 for (cu = 9; cu <= 13; cu++) | |
| 2548 regc(cu); | |
| 2549 regc(' '); | |
| 2550 break; | |
| 2551 case CLASS_UPPER: | |
| 2552 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2553 if (MB_ISUPPER(cu)) |
| 7 | 2554 regc(cu); |
| 2555 break; | |
| 2556 case CLASS_XDIGIT: | |
| 2557 for (cu = 1; cu <= 255; cu++) | |
| 2558 if (vim_isxdigit(cu)) | |
| 2559 regc(cu); | |
| 2560 break; | |
| 2561 case CLASS_TAB: | |
| 2562 regc('\t'); | |
| 2563 break; | |
| 2564 case CLASS_RETURN: | |
| 2565 regc('\r'); | |
| 2566 break; | |
| 2567 case CLASS_BACKSPACE: | |
| 2568 regc('\b'); | |
| 2569 break; | |
| 2570 case CLASS_ESCAPE: | |
| 2571 regc('\033'); | |
| 2572 break; | |
| 2573 } | |
| 2574 } | |
| 2575 else | |
| 2576 { | |
| 2577 #ifdef FEAT_MBYTE | |
| 2578 if (has_mbyte) | |
| 2579 { | |
| 2580 int len; | |
| 2581 | |
| 2582 /* produce a multibyte character, including any | |
| 2583 * following composing characters */ | |
| 2584 startc = mb_ptr2char(regparse); | |
| 474 | 2585 len = (*mb_ptr2len)(regparse); |
| 7 | 2586 if (enc_utf8 && utf_char2len(startc) != len) |
| 2587 startc = -1; /* composing chars */ | |
| 2588 while (--len >= 0) | |
| 2589 regc(*regparse++); | |
| 2590 } | |
| 2591 else | |
| 2592 #endif | |
| 2593 { | |
| 2594 startc = *regparse++; | |
| 2595 regc(startc); | |
| 2596 } | |
| 2597 } | |
| 2598 } | |
| 2599 regc(NUL); | |
| 2600 prevchr_len = 1; /* last char was the ']' */ | |
| 2601 if (*regparse != ']') | |
| 2602 EMSG_RET_NULL(_(e_toomsbra)); /* Cannot happen? */ | |
| 2603 skipchr(); /* let's be friends with the lexer again */ | |
| 2604 *flagp |= HASWIDTH | SIMPLE; | |
| 2605 break; | |
| 2606 } | |
| 481 | 2607 else if (reg_strict) |
| 4444 | 2608 EMSG2_RET_NULL(_(e_missingbracket), reg_magic > MAGIC_OFF); |
| 7 | 2609 } |
| 2610 /* FALLTHROUGH */ | |
| 2611 | |
| 2612 default: | |
| 2613 { | |
| 2614 int len; | |
| 2615 | |
| 2616 #ifdef FEAT_MBYTE | |
| 2617 /* A multi-byte character is handled as a separate atom if it's | |
| 714 | 2618 * before a multi and when it's a composing char. */ |
| 2619 if (use_multibytecode(c)) | |
| 7 | 2620 { |
| 714 | 2621 do_multibyte: |
| 7 | 2622 ret = regnode(MULTIBYTECODE); |
| 2623 regmbc(c); | |
| 2624 *flagp |= HASWIDTH | SIMPLE; | |
| 2625 break; | |
| 2626 } | |
| 2627 #endif | |
| 2628 | |
| 2629 ret = regnode(EXACTLY); | |
| 2630 | |
| 2631 /* | |
| 2632 * Append characters as long as: | |
| 2633 * - there is no following multi, we then need the character in | |
| 2634 * front of it as a single character operand | |
| 2635 * - not running into a Magic character | |
| 2636 * - "one_exactly" is not set | |
| 2637 * But always emit at least one character. Might be a Multi, | |
| 2638 * e.g., a "[" without matching "]". | |
| 2639 */ | |
| 2640 for (len = 0; c != NUL && (len == 0 | |
| 2641 || (re_multi_type(peekchr()) == NOT_MULTI | |
| 2642 && !one_exactly | |
| 2643 && !is_Magic(c))); ++len) | |
| 2644 { | |
| 2645 c = no_Magic(c); | |
| 2646 #ifdef FEAT_MBYTE | |
| 2647 if (has_mbyte) | |
| 2648 { | |
| 2649 regmbc(c); | |
| 2650 if (enc_utf8) | |
| 2651 { | |
| 2652 int l; | |
| 2653 | |
| 714 | 2654 /* Need to get composing character too. */ |
| 7 | 2655 for (;;) |
| 2656 { | |
| 714 | 2657 l = utf_ptr2len(regparse); |
| 2658 if (!UTF_COMPOSINGLIKE(regparse, regparse + l)) | |
| 7 | 2659 break; |
| 714 | 2660 regmbc(utf_ptr2char(regparse)); |
| 2661 skipchr(); | |
| 7 | 2662 } |
| 2663 } | |
| 2664 } | |
| 2665 else | |
| 2666 #endif | |
| 2667 regc(c); | |
| 2668 c = getchr(); | |
| 2669 } | |
| 2670 ungetchr(); | |
| 2671 | |
| 2672 regc(NUL); | |
| 2673 *flagp |= HASWIDTH; | |
| 2674 if (len == 1) | |
| 2675 *flagp |= SIMPLE; | |
| 2676 } | |
| 2677 break; | |
| 2678 } | |
| 2679 | |
| 2680 return ret; | |
| 2681 } | |
| 2682 | |
| 714 | 2683 #ifdef FEAT_MBYTE |
| 2684 /* | |
| 2685 * Return TRUE if MULTIBYTECODE should be used instead of EXACTLY for | |
| 2686 * character "c". | |
| 2687 */ | |
| 2688 static int | |
| 2689 use_multibytecode(c) | |
| 2690 int c; | |
| 2691 { | |
| 2692 return has_mbyte && (*mb_char2len)(c) > 1 | |
| 2693 && (re_multi_type(peekchr()) != NOT_MULTI | |
| 2694 || (enc_utf8 && utf_iscomposing(c))); | |
| 2695 } | |
| 2696 #endif | |
| 2697 | |
| 7 | 2698 /* |
| 4444 | 2699 * Emit a node. |
| 7 | 2700 * Return pointer to generated code. |
| 2701 */ | |
| 2702 static char_u * | |
| 2703 regnode(op) | |
| 2704 int op; | |
| 2705 { | |
| 2706 char_u *ret; | |
| 2707 | |
| 2708 ret = regcode; | |
| 2709 if (ret == JUST_CALC_SIZE) | |
| 2710 regsize += 3; | |
| 2711 else | |
| 2712 { | |
| 2713 *regcode++ = op; | |
| 2714 *regcode++ = NUL; /* Null "next" pointer. */ | |
| 2715 *regcode++ = NUL; | |
| 2716 } | |
| 2717 return ret; | |
| 2718 } | |
| 2719 | |
| 2720 /* | |
| 2721 * Emit (if appropriate) a byte of code | |
| 2722 */ | |
| 2723 static void | |
| 2724 regc(b) | |
| 2725 int b; | |
| 2726 { | |
| 2727 if (regcode == JUST_CALC_SIZE) | |
| 2728 regsize++; | |
| 2729 else | |
| 2730 *regcode++ = b; | |
| 2731 } | |
| 2732 | |
| 2733 #ifdef FEAT_MBYTE | |
| 2734 /* | |
| 2735 * Emit (if appropriate) a multi-byte character of code | |
| 2736 */ | |
| 2737 static void | |
| 2738 regmbc(c) | |
| 2739 int c; | |
| 2740 { | |
| 2974 | 2741 if (!has_mbyte && c > 0xff) |
| 2742 return; | |
| 7 | 2743 if (regcode == JUST_CALC_SIZE) |
| 2744 regsize += (*mb_char2len)(c); | |
| 2745 else | |
| 2746 regcode += (*mb_char2bytes)(c, regcode); | |
| 2747 } | |
| 2748 #endif | |
| 2749 | |
| 2750 /* | |
| 4444 | 2751 * Insert an operator in front of already-emitted operand |
| 7 | 2752 * |
| 2753 * Means relocating the operand. | |
| 2754 */ | |
| 2755 static void | |
| 2756 reginsert(op, opnd) | |
| 2757 int op; | |
| 2758 char_u *opnd; | |
| 2759 { | |
| 2760 char_u *src; | |
| 2761 char_u *dst; | |
| 2762 char_u *place; | |
| 2763 | |
| 2764 if (regcode == JUST_CALC_SIZE) | |
| 2765 { | |
| 2766 regsize += 3; | |
| 2767 return; | |
| 2768 } | |
| 2769 src = regcode; | |
| 2770 regcode += 3; | |
| 2771 dst = regcode; | |
| 2772 while (src > opnd) | |
| 2773 *--dst = *--src; | |
| 2774 | |
| 2775 place = opnd; /* Op node, where operand used to be. */ | |
| 2776 *place++ = op; | |
| 2777 *place++ = NUL; | |
| 2778 *place = NUL; | |
| 2779 } | |
| 2780 | |
| 2781 /* | |
| 4444 | 2782 * Insert an operator in front of already-emitted operand. |
| 7 | 2783 * The operator has the given limit values as operands. Also set next pointer. |
| 2784 * | |
| 2785 * Means relocating the operand. | |
| 2786 */ | |
| 2787 static void | |
| 2788 reginsert_limits(op, minval, maxval, opnd) | |
| 2789 int op; | |
| 2790 long minval; | |
| 2791 long maxval; | |
| 2792 char_u *opnd; | |
| 2793 { | |
| 2794 char_u *src; | |
| 2795 char_u *dst; | |
| 2796 char_u *place; | |
| 2797 | |
| 2798 if (regcode == JUST_CALC_SIZE) | |
| 2799 { | |
| 2800 regsize += 11; | |
| 2801 return; | |
| 2802 } | |
| 2803 src = regcode; | |
| 2804 regcode += 11; | |
| 2805 dst = regcode; | |
| 2806 while (src > opnd) | |
| 2807 *--dst = *--src; | |
| 2808 | |
| 2809 place = opnd; /* Op node, where operand used to be. */ | |
| 2810 *place++ = op; | |
| 2811 *place++ = NUL; | |
| 2812 *place++ = NUL; | |
| 2813 place = re_put_long(place, (long_u)minval); | |
| 2814 place = re_put_long(place, (long_u)maxval); | |
| 2815 regtail(opnd, place); | |
| 2816 } | |
| 2817 | |
| 2818 /* | |
| 2819 * Write a long as four bytes at "p" and return pointer to the next char. | |
| 2820 */ | |
| 2821 static char_u * | |
| 2822 re_put_long(p, val) | |
| 2823 char_u *p; | |
| 2824 long_u val; | |
| 2825 { | |
| 2826 *p++ = (char_u) ((val >> 24) & 0377); | |
| 2827 *p++ = (char_u) ((val >> 16) & 0377); | |
| 2828 *p++ = (char_u) ((val >> 8) & 0377); | |
| 2829 *p++ = (char_u) (val & 0377); | |
| 2830 return p; | |
| 2831 } | |
| 2832 | |
| 2833 /* | |
| 4444 | 2834 * Set the next-pointer at the end of a node chain. |
| 7 | 2835 */ |
| 2836 static void | |
| 2837 regtail(p, val) | |
| 2838 char_u *p; | |
| 2839 char_u *val; | |
| 2840 { | |
| 2841 char_u *scan; | |
| 2842 char_u *temp; | |
| 2843 int offset; | |
| 2844 | |
| 2845 if (p == JUST_CALC_SIZE) | |
| 2846 return; | |
| 2847 | |
| 2848 /* Find last node. */ | |
| 2849 scan = p; | |
| 2850 for (;;) | |
| 2851 { | |
| 2852 temp = regnext(scan); | |
| 2853 if (temp == NULL) | |
| 2854 break; | |
| 2855 scan = temp; | |
| 2856 } | |
| 2857 | |
| 233 | 2858 if (OP(scan) == BACK) |
| 7 | 2859 offset = (int)(scan - val); |
| 2860 else | |
| 2861 offset = (int)(val - scan); | |
| 2010 | 2862 /* When the offset uses more than 16 bits it can no longer fit in the two |
| 2974 | 2863 * bytes available. Use a global flag to avoid having to check return |
| 2010 | 2864 * values in too many places. */ |
| 2865 if (offset > 0xffff) | |
| 2866 reg_toolong = TRUE; | |
| 2867 else | |
| 2868 { | |
| 2869 *(scan + 1) = (char_u) (((unsigned)offset >> 8) & 0377); | |
| 2870 *(scan + 2) = (char_u) (offset & 0377); | |
| 2871 } | |
| 7 | 2872 } |
| 2873 | |
| 2874 /* | |
| 4444 | 2875 * Like regtail, on item after a BRANCH; nop if none. |
| 7 | 2876 */ |
| 2877 static void | |
| 2878 regoptail(p, val) | |
| 2879 char_u *p; | |
| 2880 char_u *val; | |
| 2881 { | |
| 2882 /* When op is neither BRANCH nor BRACE_COMPLEX0-9, it is "operandless" */ | |
| 2883 if (p == NULL || p == JUST_CALC_SIZE | |
| 2884 || (OP(p) != BRANCH | |
| 2885 && (OP(p) < BRACE_COMPLEX || OP(p) > BRACE_COMPLEX + 9))) | |
| 2886 return; | |
| 2887 regtail(OPERAND(p), val); | |
| 2888 } | |
| 2889 | |
| 2890 /* | |
| 4444 | 2891 * Functions for getting characters from the regexp input. |
| 7 | 2892 */ |
| 2893 | |
| 2894 static int at_start; /* True when on the first character */ | |
| 2895 static int prev_at_start; /* True when on the second character */ | |
| 2896 | |
| 4444 | 2897 /* |
| 2898 * Start parsing at "str". | |
| 2899 */ | |
| 7 | 2900 static void |
| 2901 initchr(str) | |
| 2902 char_u *str; | |
| 2903 { | |
| 2904 regparse = str; | |
| 2905 prevchr_len = 0; | |
| 2906 curchr = prevprevchr = prevchr = nextchr = -1; | |
| 2907 at_start = TRUE; | |
| 2908 prev_at_start = FALSE; | |
| 2909 } | |
| 2910 | |
| 4444 | 2911 /* |
| 2912 * Get the next character without advancing. | |
| 2913 */ | |
| 7 | 2914 static int |
| 2915 peekchr() | |
| 2916 { | |
| 167 | 2917 static int after_slash = FALSE; |
| 2918 | |
| 7 | 2919 if (curchr == -1) |
| 2920 { | |
| 2921 switch (curchr = regparse[0]) | |
| 2922 { | |
| 2923 case '.': | |
| 2924 case '[': | |
| 2925 case '~': | |
| 2926 /* magic when 'magic' is on */ | |
| 2927 if (reg_magic >= MAGIC_ON) | |
| 2928 curchr = Magic(curchr); | |
| 2929 break; | |
| 2930 case '(': | |
| 2931 case ')': | |
| 2932 case '{': | |
| 2933 case '%': | |
| 2934 case '+': | |
| 2935 case '=': | |
| 2936 case '?': | |
| 2937 case '@': | |
| 2938 case '!': | |
| 2939 case '&': | |
| 2940 case '|': | |
| 2941 case '<': | |
| 2942 case '>': | |
| 2943 case '#': /* future ext. */ | |
| 2944 case '"': /* future ext. */ | |
| 2945 case '\'': /* future ext. */ | |
| 2946 case ',': /* future ext. */ | |
| 2947 case '-': /* future ext. */ | |
| 2948 case ':': /* future ext. */ | |
| 2949 case ';': /* future ext. */ | |
| 2950 case '`': /* future ext. */ | |
| 2951 case '/': /* Can't be used in / command */ | |
| 2952 /* magic only after "\v" */ | |
| 2953 if (reg_magic == MAGIC_ALL) | |
| 2954 curchr = Magic(curchr); | |
| 2955 break; | |
| 2956 case '*': | |
| 167 | 2957 /* * is not magic as the very first character, eg "?*ptr", when |
| 2958 * after '^', eg "/^*ptr" and when after "\(", "\|", "\&". But | |
| 2959 * "\(\*" is not magic, thus must be magic if "after_slash" */ | |
| 2960 if (reg_magic >= MAGIC_ON | |
| 2961 && !at_start | |
| 2962 && !(prev_at_start && prevchr == Magic('^')) | |
| 2963 && (after_slash | |
| 2964 || (prevchr != Magic('(') | |
| 2965 && prevchr != Magic('&') | |
| 2966 && prevchr != Magic('|')))) | |
| 7 | 2967 curchr = Magic('*'); |
| 2968 break; | |
| 2969 case '^': | |
| 2970 /* '^' is only magic as the very first character and if it's after | |
| 2971 * "\(", "\|", "\&' or "\n" */ | |
| 2972 if (reg_magic >= MAGIC_OFF | |
| 2973 && (at_start | |
| 2974 || reg_magic == MAGIC_ALL | |
| 2975 || prevchr == Magic('(') | |
| 2976 || prevchr == Magic('|') | |
| 2977 || prevchr == Magic('&') | |
| 2978 || prevchr == Magic('n') | |
| 2979 || (no_Magic(prevchr) == '(' | |
| 2980 && prevprevchr == Magic('%')))) | |
| 2981 { | |
| 2982 curchr = Magic('^'); | |
| 2983 at_start = TRUE; | |
| 2984 prev_at_start = FALSE; | |
| 2985 } | |
| 2986 break; | |
| 2987 case '$': | |
| 2988 /* '$' is only magic as the very last char and if it's in front of | |
| 2989 * either "\|", "\)", "\&", or "\n" */ | |
| 2990 if (reg_magic >= MAGIC_OFF) | |
| 2991 { | |
| 2992 char_u *p = regparse + 1; | |
| 2993 | |
| 2994 /* ignore \c \C \m and \M after '$' */ | |
| 2995 while (p[0] == '\\' && (p[1] == 'c' || p[1] == 'C' | |
| 2996 || p[1] == 'm' || p[1] == 'M' || p[1] == 'Z')) | |
| 2997 p += 2; | |
| 2998 if (p[0] == NUL | |
| 2999 || (p[0] == '\\' | |
| 3000 && (p[1] == '|' || p[1] == '&' || p[1] == ')' | |
| 3001 || p[1] == 'n')) | |
| 3002 || reg_magic == MAGIC_ALL) | |
| 3003 curchr = Magic('$'); | |
| 3004 } | |
| 3005 break; | |
| 3006 case '\\': | |
| 3007 { | |
| 3008 int c = regparse[1]; | |
| 3009 | |
| 3010 if (c == NUL) | |
| 3011 curchr = '\\'; /* trailing '\' */ | |
| 3012 else if ( | |
| 3013 #ifdef EBCDIC | |
| 3014 vim_strchr(META, c) | |
| 3015 #else | |
| 3016 c <= '~' && META_flags[c] | |
| 3017 #endif | |
| 3018 ) | |
| 3019 { | |
| 3020 /* | |
| 3021 * META contains everything that may be magic sometimes, | |
| 3022 * except ^ and $ ("\^" and "\$" are only magic after | |
| 3023 * "\v"). We now fetch the next character and toggle its | |
| 3024 * magicness. Therefore, \ is so meta-magic that it is | |
| 3025 * not in META. | |
| 3026 */ | |
| 3027 curchr = -1; | |
| 3028 prev_at_start = at_start; | |
| 3029 at_start = FALSE; /* be able to say "/\*ptr" */ | |
| 3030 ++regparse; | |
| 167 | 3031 ++after_slash; |
| 7 | 3032 peekchr(); |
| 3033 --regparse; | |
| 167 | 3034 --after_slash; |
| 7 | 3035 curchr = toggle_Magic(curchr); |
| 3036 } | |
| 3037 else if (vim_strchr(REGEXP_ABBR, c)) | |
| 3038 { | |
| 3039 /* | |
| 3040 * Handle abbreviations, like "\t" for TAB -- webb | |
| 3041 */ | |
| 3042 curchr = backslash_trans(c); | |
| 3043 } | |
| 3044 else if (reg_magic == MAGIC_NONE && (c == '$' || c == '^')) | |
| 3045 curchr = toggle_Magic(c); | |
| 3046 else | |
| 3047 { | |
| 3048 /* | |
| 3049 * Next character can never be (made) magic? | |
| 3050 * Then backslashing it won't do anything. | |
| 3051 */ | |
| 3052 #ifdef FEAT_MBYTE | |
| 3053 if (has_mbyte) | |
| 3054 curchr = (*mb_ptr2char)(regparse + 1); | |
| 3055 else | |
| 3056 #endif | |
| 3057 curchr = c; | |
| 3058 } | |
| 3059 break; | |
| 3060 } | |
| 3061 | |
| 3062 #ifdef FEAT_MBYTE | |
| 3063 default: | |
| 3064 if (has_mbyte) | |
| 3065 curchr = (*mb_ptr2char)(regparse); | |
| 3066 #endif | |
| 3067 } | |
| 3068 } | |
| 3069 | |
| 3070 return curchr; | |
| 3071 } | |
| 3072 | |
| 3073 /* | |
| 3074 * Eat one lexed character. Do this in a way that we can undo it. | |
| 3075 */ | |
| 3076 static void | |
| 3077 skipchr() | |
| 3078 { | |
| 3079 /* peekchr() eats a backslash, do the same here */ | |
| 3080 if (*regparse == '\\') | |
| 3081 prevchr_len = 1; | |
| 3082 else | |
| 3083 prevchr_len = 0; | |
| 3084 if (regparse[prevchr_len] != NUL) | |
| 3085 { | |
| 3086 #ifdef FEAT_MBYTE | |
| 714 | 3087 if (enc_utf8) |
| 1449 | 3088 /* exclude composing chars that mb_ptr2len does include */ |
| 3089 prevchr_len += utf_ptr2len(regparse + prevchr_len); | |
| 714 | 3090 else if (has_mbyte) |
| 474 | 3091 prevchr_len += (*mb_ptr2len)(regparse + prevchr_len); |
| 7 | 3092 else |
| 3093 #endif | |
| 3094 ++prevchr_len; | |
| 3095 } | |
| 3096 regparse += prevchr_len; | |
| 3097 prev_at_start = at_start; | |
| 3098 at_start = FALSE; | |
| 3099 prevprevchr = prevchr; | |
| 3100 prevchr = curchr; | |
| 3101 curchr = nextchr; /* use previously unget char, or -1 */ | |
| 3102 nextchr = -1; | |
| 3103 } | |
| 3104 | |
| 3105 /* | |
| 3106 * Skip a character while keeping the value of prev_at_start for at_start. | |
| 3107 * prevchr and prevprevchr are also kept. | |
| 3108 */ | |
| 3109 static void | |
| 3110 skipchr_keepstart() | |
| 3111 { | |
| 3112 int as = prev_at_start; | |
| 3113 int pr = prevchr; | |
| 3114 int prpr = prevprevchr; | |
| 3115 | |
| 3116 skipchr(); | |
| 3117 at_start = as; | |
| 3118 prevchr = pr; | |
| 3119 prevprevchr = prpr; | |
| 3120 } | |
| 3121 | |
| 4444 | 3122 /* |
| 3123 * Get the next character from the pattern. We know about magic and such, so | |
| 3124 * therefore we need a lexical analyzer. | |
| 3125 */ | |
| 7 | 3126 static int |
| 3127 getchr() | |
| 3128 { | |
| 3129 int chr = peekchr(); | |
| 3130 | |
| 3131 skipchr(); | |
| 3132 return chr; | |
| 3133 } | |
| 3134 | |
| 3135 /* | |
| 3136 * put character back. Works only once! | |
| 3137 */ | |
| 3138 static void | |
| 3139 ungetchr() | |
| 3140 { | |
| 3141 nextchr = curchr; | |
| 3142 curchr = prevchr; | |
| 3143 prevchr = prevprevchr; | |
| 3144 at_start = prev_at_start; | |
| 3145 prev_at_start = FALSE; | |
| 3146 | |
| 3147 /* Backup regparse, so that it's at the same position as before the | |
| 3148 * getchr(). */ | |
| 3149 regparse -= prevchr_len; | |
| 3150 } | |
| 3151 | |
| 3152 /* | |
| 29 | 3153 * Get and return the value of the hex string at the current position. |
| 3154 * Return -1 if there is no valid hex number. | |
| 3155 * The position is updated: | |
| 24 | 3156 * blahblah\%x20asdf |
| 856 | 3157 * before-^ ^-after |
| 24 | 3158 * The parameter controls the maximum number of input characters. This will be |
| 3159 * 2 when reading a \%x20 sequence and 4 when reading a \%u20AC sequence. | |
| 3160 */ | |
| 3161 static int | |
| 3162 gethexchrs(maxinputlen) | |
| 3163 int maxinputlen; | |
| 3164 { | |
| 3165 int nr = 0; | |
| 3166 int c; | |
| 3167 int i; | |
| 3168 | |
| 3169 for (i = 0; i < maxinputlen; ++i) | |
| 3170 { | |
| 3171 c = regparse[0]; | |
| 3172 if (!vim_isxdigit(c)) | |
| 3173 break; | |
| 3174 nr <<= 4; | |
| 3175 nr |= hex2nr(c); | |
| 3176 ++regparse; | |
| 3177 } | |
| 3178 | |
| 3179 if (i == 0) | |
| 3180 return -1; | |
| 3181 return nr; | |
| 3182 } | |
| 3183 | |
| 3184 /* | |
| 3185 * get and return the value of the decimal string immediately after the | |
| 3186 * current position. Return -1 for invalid. Consumes all digits. | |
| 3187 */ | |
| 3188 static int | |
| 3189 getdecchrs() | |
| 3190 { | |
| 3191 int nr = 0; | |
| 3192 int c; | |
| 3193 int i; | |
| 3194 | |
| 3195 for (i = 0; ; ++i) | |
| 3196 { | |
| 3197 c = regparse[0]; | |
| 3198 if (c < '0' || c > '9') | |
| 3199 break; | |
| 3200 nr *= 10; | |
| 3201 nr += c - '0'; | |
| 3202 ++regparse; | |
| 3203 } | |
| 3204 | |
| 3205 if (i == 0) | |
| 3206 return -1; | |
| 3207 return nr; | |
| 3208 } | |
| 3209 | |
| 3210 /* | |
| 3211 * get and return the value of the octal string immediately after the current | |
| 3212 * position. Return -1 for invalid, or 0-255 for valid. Smart enough to handle | |
| 3213 * numbers > 377 correctly (for example, 400 is treated as 40) and doesn't | |
| 3214 * treat 8 or 9 as recognised characters. Position is updated: | |
| 3215 * blahblah\%o210asdf | |
| 856 | 3216 * before-^ ^-after |
| 24 | 3217 */ |
| 3218 static int | |
| 3219 getoctchrs() | |
| 3220 { | |
| 3221 int nr = 0; | |
| 3222 int c; | |
| 3223 int i; | |
| 3224 | |
| 3225 for (i = 0; i < 3 && nr < 040; ++i) | |
| 3226 { | |
| 3227 c = regparse[0]; | |
| 3228 if (c < '0' || c > '7') | |
| 3229 break; | |
| 3230 nr <<= 3; | |
| 3231 nr |= hex2nr(c); | |
| 3232 ++regparse; | |
| 3233 } | |
| 3234 | |
| 3235 if (i == 0) | |
| 3236 return -1; | |
| 3237 return nr; | |
| 3238 } | |
| 3239 | |
| 3240 /* | |
| 3241 * Get a number after a backslash that is inside []. | |
| 3242 * When nothing is recognized return a backslash. | |
| 3243 */ | |
| 3244 static int | |
| 3245 coll_get_char() | |
| 3246 { | |
| 3247 int nr = -1; | |
| 3248 | |
| 3249 switch (*regparse++) | |
| 3250 { | |
| 3251 case 'd': nr = getdecchrs(); break; | |
| 3252 case 'o': nr = getoctchrs(); break; | |
| 3253 case 'x': nr = gethexchrs(2); break; | |
| 3254 case 'u': nr = gethexchrs(4); break; | |
| 3255 case 'U': nr = gethexchrs(8); break; | |
| 3256 } | |
| 3257 if (nr < 0) | |
| 3258 { | |
| 3259 /* If getting the number fails be backwards compatible: the character | |
| 3260 * is a backslash. */ | |
| 3261 --regparse; | |
| 3262 nr = '\\'; | |
| 3263 } | |
| 3264 return nr; | |
| 3265 } | |
| 3266 | |
| 3267 /* | |
| 7 | 3268 * read_limits - Read two integers to be taken as a minimum and maximum. |
| 3269 * If the first character is '-', then the range is reversed. | |
| 3270 * Should end with 'end'. If minval is missing, zero is default, if maxval is | |
| 3271 * missing, a very big number is the default. | |
| 3272 */ | |
| 3273 static int | |
| 3274 read_limits(minval, maxval) | |
| 3275 long *minval; | |
| 3276 long *maxval; | |
| 3277 { | |
| 3278 int reverse = FALSE; | |
| 3279 char_u *first_char; | |
| 3280 long tmp; | |
| 3281 | |
| 3282 if (*regparse == '-') | |
| 3283 { | |
| 3284 /* Starts with '-', so reverse the range later */ | |
| 3285 regparse++; | |
| 3286 reverse = TRUE; | |
| 3287 } | |
| 3288 first_char = regparse; | |
| 3289 *minval = getdigits(®parse); | |
| 3290 if (*regparse == ',') /* There is a comma */ | |
| 3291 { | |
| 3292 if (vim_isdigit(*++regparse)) | |
| 3293 *maxval = getdigits(®parse); | |
| 3294 else | |
| 3295 *maxval = MAX_LIMIT; | |
| 3296 } | |
| 3297 else if (VIM_ISDIGIT(*first_char)) | |
| 3298 *maxval = *minval; /* It was \{n} or \{-n} */ | |
| 3299 else | |
| 3300 *maxval = MAX_LIMIT; /* It was \{} or \{-} */ | |
| 3301 if (*regparse == '\\') | |
| 3302 regparse++; /* Allow either \{...} or \{...\} */ | |
| 167 | 3303 if (*regparse != '}') |
| 7 | 3304 { |
| 3305 sprintf((char *)IObuff, _("E554: Syntax error in %s{...}"), | |
| 3306 reg_magic == MAGIC_ALL ? "" : "\\"); | |
| 3307 EMSG_RET_FAIL(IObuff); | |
| 3308 } | |
| 3309 | |
| 3310 /* | |
| 3311 * Reverse the range if there was a '-', or make sure it is in the right | |
| 3312 * order otherwise. | |
| 3313 */ | |
| 3314 if ((!reverse && *minval > *maxval) || (reverse && *minval < *maxval)) | |
| 3315 { | |
| 3316 tmp = *minval; | |
| 3317 *minval = *maxval; | |
| 3318 *maxval = tmp; | |
| 3319 } | |
| 3320 skipchr(); /* let's be friends with the lexer again */ | |
| 3321 return OK; | |
| 3322 } | |
| 3323 | |
| 3324 /* | |
| 3325 * vim_regexec and friends | |
| 3326 */ | |
| 3327 | |
| 3328 /* | |
| 3329 * Global work variables for vim_regexec(). | |
| 3330 */ | |
| 3331 | |
| 3332 /* The current match-position is remembered with these variables: */ | |
| 3333 static linenr_T reglnum; /* line number, relative to first line */ | |
| 3334 static char_u *regline; /* start of current line */ | |
| 3335 static char_u *reginput; /* current input, points into "regline" */ | |
| 3336 | |
| 3337 static int need_clear_subexpr; /* subexpressions still need to be | |
| 3338 * cleared */ | |
| 3339 #ifdef FEAT_SYN_HL | |
| 3340 static int need_clear_zsubexpr = FALSE; /* extmatch subexpressions | |
| 3341 * still need to be cleared */ | |
| 3342 #endif | |
| 3343 | |
| 3344 /* | |
| 3345 * Structure used to save the current input state, when it needs to be | |
| 3346 * restored after trying a match. Used by reg_save() and reg_restore(). | |
| 233 | 3347 * Also stores the length of "backpos". |
| 7 | 3348 */ |
| 3349 typedef struct | |
| 3350 { | |
| 3351 union | |
| 3352 { | |
| 3353 char_u *ptr; /* reginput pointer, for single-line regexp */ | |
| 3354 lpos_T pos; /* reginput pos, for multi-line regexp */ | |
| 3355 } rs_u; | |
| 233 | 3356 int rs_len; |
| 7 | 3357 } regsave_T; |
| 3358 | |
| 3359 /* struct to save start/end pointer/position in for \(\) */ | |
| 3360 typedef struct | |
| 3361 { | |
| 3362 union | |
| 3363 { | |
| 3364 char_u *ptr; | |
| 3365 lpos_T pos; | |
| 3366 } se_u; | |
| 3367 } save_se_T; | |
| 3368 | |
| 1579 | 3369 /* used for BEHIND and NOBEHIND matching */ |
| 3370 typedef struct regbehind_S | |
| 3371 { | |
| 3372 regsave_T save_after; | |
| 3373 regsave_T save_behind; | |
| 1602 | 3374 int save_need_clear_subexpr; |
| 1579 | 3375 save_se_T save_start[NSUBEXP]; |
| 3376 save_se_T save_end[NSUBEXP]; | |
| 3377 } regbehind_T; | |
| 3378 | |
| 7 | 3379 static char_u *reg_getline __ARGS((linenr_T lnum)); |
| 4444 | 3380 static long bt_regexec_both __ARGS((char_u *line, colnr_T col, proftime_T *tm)); |
| 3381 static long regtry __ARGS((bt_regprog_T *prog, colnr_T col)); | |
| 7 | 3382 static void cleanup_subexpr __ARGS((void)); |
| 3383 #ifdef FEAT_SYN_HL | |
| 3384 static void cleanup_zsubexpr __ARGS((void)); | |
| 3385 #endif | |
| 1579 | 3386 static void save_subexpr __ARGS((regbehind_T *bp)); |
| 3387 static void restore_subexpr __ARGS((regbehind_T *bp)); | |
| 7 | 3388 static void reg_nextline __ARGS((void)); |
| 233 | 3389 static void reg_save __ARGS((regsave_T *save, garray_T *gap)); |
| 3390 static void reg_restore __ARGS((regsave_T *save, garray_T *gap)); | |
| 7 | 3391 static int reg_save_equal __ARGS((regsave_T *save)); |
| 3392 static void save_se_multi __ARGS((save_se_T *savep, lpos_T *posp)); | |
| 3393 static void save_se_one __ARGS((save_se_T *savep, char_u **pp)); | |
| 3394 | |
| 3395 /* Save the sub-expressions before attempting a match. */ | |
| 3396 #define save_se(savep, posp, pp) \ | |
| 3397 REG_MULTI ? save_se_multi((savep), (posp)) : save_se_one((savep), (pp)) | |
| 3398 | |
| 3399 /* After a failed match restore the sub-expressions. */ | |
| 3400 #define restore_se(savep, posp, pp) { \ | |
| 3401 if (REG_MULTI) \ | |
| 3402 *(posp) = (savep)->se_u.pos; \ | |
| 3403 else \ | |
| 3404 *(pp) = (savep)->se_u.ptr; } | |
| 3405 | |
| 3406 static int re_num_cmp __ARGS((long_u val, char_u *scan)); | |
| 180 | 3407 static int regmatch __ARGS((char_u *prog)); |
| 7 | 3408 static int regrepeat __ARGS((char_u *p, long maxcount)); |
| 3409 | |
| 3410 #ifdef DEBUG | |
| 3411 int regnarrate = 0; | |
| 3412 #endif | |
| 3413 | |
| 3414 /* | |
| 3415 * Internal copy of 'ignorecase'. It is set at each call to vim_regexec(). | |
| 3416 * Normally it gets the value of "rm_ic" or "rmm_ic", but when the pattern | |
| 3417 * contains '\c' or '\C' the value is overruled. | |
| 3418 */ | |
| 3419 static int ireg_ic; | |
| 3420 | |
| 3421 #ifdef FEAT_MBYTE | |
| 3422 /* | |
| 3423 * Similar to ireg_ic, but only for 'combining' characters. Set with \Z flag | |
| 3424 * in the regexp. Defaults to false, always. | |
| 3425 */ | |
| 3426 static int ireg_icombine; | |
| 3427 #endif | |
| 3428 | |
| 3429 /* | |
| 410 | 3430 * Copy of "rmm_maxcol": maximum column to search for a match. Zero when |
| 3431 * there is no maximum. | |
| 3432 */ | |
| 418 | 3433 static colnr_T ireg_maxcol; |
| 410 | 3434 |
| 3435 /* | |
| 7 | 3436 * Sometimes need to save a copy of a line. Since alloc()/free() is very |
| 3437 * slow, we keep one allocated piece of memory and only re-allocate it when | |
| 4444 | 3438 * it's too small. It's freed in bt_regexec_both() when finished. |
| 7 | 3439 */ |
| 1468 | 3440 static char_u *reg_tofree = NULL; |
| 7 | 3441 static unsigned reg_tofreelen; |
| 3442 | |
| 3443 /* | |
| 3444 * These variables are set when executing a regexp to speed up the execution. | |
| 1209 | 3445 * Which ones are set depends on whether a single-line or multi-line match is |
| 7 | 3446 * done: |
| 3447 * single-line multi-line | |
| 3448 * reg_match ®match_T NULL | |
| 3449 * reg_mmatch NULL ®mmatch_T | |
| 3450 * reg_startp reg_match->startp <invalid> | |
| 3451 * reg_endp reg_match->endp <invalid> | |
| 3452 * reg_startpos <invalid> reg_mmatch->startpos | |
| 3453 * reg_endpos <invalid> reg_mmatch->endpos | |
| 3454 * reg_win NULL window in which to search | |
| 4061 | 3455 * reg_buf curbuf buffer in which to search |
| 7 | 3456 * reg_firstlnum <invalid> first line in which to search |
| 3457 * reg_maxline 0 last line nr | |
| 3458 * reg_line_lbr FALSE or TRUE FALSE | |
| 3459 */ | |
| 3460 static regmatch_T *reg_match; | |
| 3461 static regmmatch_T *reg_mmatch; | |
| 3462 static char_u **reg_startp = NULL; | |
| 3463 static char_u **reg_endp = NULL; | |
| 3464 static lpos_T *reg_startpos = NULL; | |
| 3465 static lpos_T *reg_endpos = NULL; | |
| 3466 static win_T *reg_win; | |
| 3467 static buf_T *reg_buf; | |
| 3468 static linenr_T reg_firstlnum; | |
| 3469 static linenr_T reg_maxline; | |
| 3470 static int reg_line_lbr; /* "\n" in string is line break */ | |
| 3471 | |
| 270 | 3472 /* Values for rs_state in regitem_T. */ |
| 3473 typedef enum regstate_E | |
| 3474 { | |
| 3475 RS_NOPEN = 0 /* NOPEN and NCLOSE */ | |
| 3476 , RS_MOPEN /* MOPEN + [0-9] */ | |
| 3477 , RS_MCLOSE /* MCLOSE + [0-9] */ | |
| 3478 #ifdef FEAT_SYN_HL | |
| 3479 , RS_ZOPEN /* ZOPEN + [0-9] */ | |
| 3480 , RS_ZCLOSE /* ZCLOSE + [0-9] */ | |
| 3481 #endif | |
| 3482 , RS_BRANCH /* BRANCH */ | |
| 3483 , RS_BRCPLX_MORE /* BRACE_COMPLEX and trying one more match */ | |
| 3484 , RS_BRCPLX_LONG /* BRACE_COMPLEX and trying longest match */ | |
| 3485 , RS_BRCPLX_SHORT /* BRACE_COMPLEX and trying shortest match */ | |
| 3486 , RS_NOMATCH /* NOMATCH */ | |
| 3487 , RS_BEHIND1 /* BEHIND / NOBEHIND matching rest */ | |
| 3488 , RS_BEHIND2 /* BEHIND / NOBEHIND matching behind part */ | |
| 3489 , RS_STAR_LONG /* STAR/PLUS/BRACE_SIMPLE longest match */ | |
| 3490 , RS_STAR_SHORT /* STAR/PLUS/BRACE_SIMPLE shortest match */ | |
| 3491 } regstate_T; | |
| 3492 | |
| 3493 /* | |
| 3494 * When there are alternatives a regstate_T is put on the regstack to remember | |
| 3495 * what we are doing. | |
| 3496 * Before it may be another type of item, depending on rs_state, to remember | |
| 3497 * more things. | |
| 3498 */ | |
| 3499 typedef struct regitem_S | |
| 3500 { | |
| 3501 regstate_T rs_state; /* what we are doing, one of RS_ above */ | |
| 3502 char_u *rs_scan; /* current node in program */ | |
| 3503 union | |
| 3504 { | |
| 3505 save_se_T sesave; | |
| 3506 regsave_T regsave; | |
| 3507 } rs_un; /* room for saving reginput */ | |
| 1579 | 3508 short rs_no; /* submatch nr or BEHIND/NOBEHIND */ |
| 270 | 3509 } regitem_T; |
| 3510 | |
| 3511 static regitem_T *regstack_push __ARGS((regstate_T state, char_u *scan)); | |
| 3512 static void regstack_pop __ARGS((char_u **scan)); | |
| 3513 | |
| 3514 /* used for STAR, PLUS and BRACE_SIMPLE matching */ | |
| 3515 typedef struct regstar_S | |
| 3516 { | |
| 3517 int nextb; /* next byte */ | |
| 3518 int nextb_ic; /* next byte reverse case */ | |
| 3519 long count; | |
| 3520 long minval; | |
| 3521 long maxval; | |
| 3522 } regstar_T; | |
| 3523 | |
| 3524 /* used to store input position when a BACK was encountered, so that we now if | |
| 3525 * we made any progress since the last time. */ | |
| 3526 typedef struct backpos_S | |
| 3527 { | |
| 3528 char_u *bp_scan; /* "scan" where BACK was encountered */ | |
| 3529 regsave_T bp_pos; /* last input position */ | |
| 3530 } backpos_T; | |
| 3531 | |
| 3532 /* | |
| 1520 | 3533 * "regstack" and "backpos" are used by regmatch(). They are kept over calls |
| 3534 * to avoid invoking malloc() and free() often. | |
| 3535 * "regstack" is a stack with regitem_T items, sometimes preceded by regstar_T | |
| 3536 * or regbehind_T. | |
| 3537 * "backpos_T" is a table with backpos_T for BACK | |
| 270 | 3538 */ |
| 1520 | 3539 static garray_T regstack = {0, 0, 0, 0, NULL}; |
| 3540 static garray_T backpos = {0, 0, 0, 0, NULL}; | |
| 3541 | |
| 3542 /* | |
| 3543 * Both for regstack and backpos tables we use the following strategy of | |
| 3544 * allocation (to reduce malloc/free calls): | |
| 3545 * - Initial size is fairly small. | |
| 3546 * - When needed, the tables are grown bigger (8 times at first, double after | |
| 3547 * that). | |
| 3548 * - After executing the match we free the memory only if the array has grown. | |
| 3549 * Thus the memory is kept allocated when it's at the initial size. | |
| 3550 * This makes it fast while not keeping a lot of memory allocated. | |
| 3551 * A three times speed increase was observed when using many simple patterns. | |
| 3552 */ | |
| 3553 #define REGSTACK_INITIAL 2048 | |
| 3554 #define BACKPOS_INITIAL 64 | |
| 3555 | |
| 3556 #if defined(EXITFREE) || defined(PROTO) | |
| 3557 void | |
| 3558 free_regexp_stuff() | |
| 3559 { | |
| 3560 ga_clear(®stack); | |
| 3561 ga_clear(&backpos); | |
| 3562 vim_free(reg_tofree); | |
| 3563 vim_free(reg_prev_sub); | |
| 3564 } | |
| 3565 #endif | |
| 270 | 3566 |
| 7 | 3567 /* |
| 3568 * Get pointer to the line "lnum", which is relative to "reg_firstlnum". | |
| 3569 */ | |
| 3570 static char_u * | |
| 3571 reg_getline(lnum) | |
| 3572 linenr_T lnum; | |
| 3573 { | |
| 3574 /* when looking behind for a match/no-match lnum is negative. But we | |
| 3575 * can't go before line 1 */ | |
| 3576 if (reg_firstlnum + lnum < 1) | |
| 3577 return NULL; | |
| 482 | 3578 if (lnum > reg_maxline) |
| 481 | 3579 /* Must have matched the "\n" in the last line. */ |
| 3580 return (char_u *)""; | |
| 7 | 3581 return ml_get_buf(reg_buf, reg_firstlnum + lnum, FALSE); |
| 3582 } | |
| 3583 | |
| 3584 static regsave_T behind_pos; | |
| 3585 | |
| 3586 #ifdef FEAT_SYN_HL | |
| 3587 static char_u *reg_startzp[NSUBEXP]; /* Workspace to mark beginning */ | |
| 3588 static char_u *reg_endzp[NSUBEXP]; /* and end of \z(...\) matches */ | |
| 3589 static lpos_T reg_startzpos[NSUBEXP]; /* idem, beginning pos */ | |
| 3590 static lpos_T reg_endzpos[NSUBEXP]; /* idem, end pos */ | |
| 3591 #endif | |
| 3592 | |
| 3593 /* TRUE if using multi-line regexp. */ | |
| 3594 #define REG_MULTI (reg_match == NULL) | |
| 3595 | |
| 4444 | 3596 static int bt_regexec __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); |
| 3597 | |
| 7 | 3598 /* |
| 3599 * Match a regexp against a string. | |
| 3600 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3601 * Uses curbuf for line count and 'iskeyword'. | |
| 3602 * | |
| 3603 * Return TRUE if there is a match, FALSE if not. | |
| 3604 */ | |
| 4444 | 3605 static int |
| 3606 bt_regexec(rmp, line, col) | |
| 7 | 3607 regmatch_T *rmp; |
| 3608 char_u *line; /* string to match against */ | |
| 3609 colnr_T col; /* column to start looking for match */ | |
| 3610 { | |
| 3611 reg_match = rmp; | |
| 3612 reg_mmatch = NULL; | |
| 3613 reg_maxline = 0; | |
| 3614 reg_line_lbr = FALSE; | |
| 4061 | 3615 reg_buf = curbuf; |
| 7 | 3616 reg_win = NULL; |
| 3617 ireg_ic = rmp->rm_ic; | |
| 3618 #ifdef FEAT_MBYTE | |
| 3619 ireg_icombine = FALSE; | |
| 3620 #endif | |
| 410 | 3621 ireg_maxcol = 0; |
| 4444 | 3622 return (bt_regexec_both(line, col, NULL) != 0); |
| 7 | 3623 } |
| 3624 | |
| 39 | 3625 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ |
| 3626 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 4444 | 3627 |
| 3628 static int bt_regexec_nl __ARGS((regmatch_T *rmp, char_u *line, colnr_T col)); | |
| 3629 | |
| 7 | 3630 /* |
| 3631 * Like vim_regexec(), but consider a "\n" in "line" to be a line break. | |
| 3632 */ | |
| 4444 | 3633 static int |
| 3634 bt_regexec_nl(rmp, line, col) | |
| 7 | 3635 regmatch_T *rmp; |
| 3636 char_u *line; /* string to match against */ | |
| 3637 colnr_T col; /* column to start looking for match */ | |
| 3638 { | |
| 3639 reg_match = rmp; | |
| 3640 reg_mmatch = NULL; | |
| 3641 reg_maxline = 0; | |
| 3642 reg_line_lbr = TRUE; | |
| 4061 | 3643 reg_buf = curbuf; |
| 7 | 3644 reg_win = NULL; |
| 3645 ireg_ic = rmp->rm_ic; | |
| 3646 #ifdef FEAT_MBYTE | |
| 3647 ireg_icombine = FALSE; | |
| 3648 #endif | |
| 410 | 3649 ireg_maxcol = 0; |
| 4444 | 3650 return (bt_regexec_both(line, col, NULL) != 0); |
| 7 | 3651 } |
| 3652 #endif | |
| 3653 | |
| 4444 | 3654 static long bt_regexec_multi __ARGS((regmmatch_T *rmp, win_T *win, buf_T *buf, linenr_T lnum, colnr_T col, proftime_T *tm)); |
| 3655 | |
| 7 | 3656 /* |
| 3657 * Match a regexp against multiple lines. | |
| 3658 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3659 * Uses curbuf for line count and 'iskeyword'. | |
| 3660 * | |
| 3661 * Return zero if there is no match. Return number of lines contained in the | |
| 3662 * match otherwise. | |
| 3663 */ | |
| 4444 | 3664 static long |
| 3665 bt_regexec_multi(rmp, win, buf, lnum, col, tm) | |
| 7 | 3666 regmmatch_T *rmp; |
| 3667 win_T *win; /* window in which to search or NULL */ | |
| 3668 buf_T *buf; /* buffer in which to search */ | |
| 3669 linenr_T lnum; /* nr of line to start looking for match */ | |
| 3670 colnr_T col; /* column to start looking for match */ | |
| 1521 | 3671 proftime_T *tm; /* timeout limit or NULL */ |
| 7 | 3672 { |
| 3673 long r; | |
| 3674 | |
| 3675 reg_match = NULL; | |
| 3676 reg_mmatch = rmp; | |
| 3677 reg_buf = buf; | |
| 3678 reg_win = win; | |
| 3679 reg_firstlnum = lnum; | |
| 3680 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
| 3681 reg_line_lbr = FALSE; | |
| 3682 ireg_ic = rmp->rmm_ic; | |
| 3683 #ifdef FEAT_MBYTE | |
| 3684 ireg_icombine = FALSE; | |
| 3685 #endif | |
| 410 | 3686 ireg_maxcol = rmp->rmm_maxcol; |
| 7 | 3687 |
| 4444 | 3688 r = bt_regexec_both(NULL, col, tm); |
| 7 | 3689 |
| 3690 return r; | |
| 3691 } | |
| 3692 | |
| 3693 /* | |
| 3694 * Match a regexp against a string ("line" points to the string) or multiple | |
| 3695 * lines ("line" is NULL, use reg_getline()). | |
| 3696 */ | |
| 3697 static long | |
| 4444 | 3698 bt_regexec_both(line, col, tm) |
| 7 | 3699 char_u *line; |
| 3700 colnr_T col; /* column to start looking for match */ | |
| 1877 | 3701 proftime_T *tm UNUSED; /* timeout limit or NULL */ |
| 7 | 3702 { |
| 4444 | 3703 bt_regprog_T *prog; |
| 7 | 3704 char_u *s; |
| 189 | 3705 long retval = 0L; |
| 7 | 3706 |
| 1520 | 3707 /* Create "regstack" and "backpos" if they are not allocated yet. |
| 3708 * We allocate *_INITIAL amount of bytes first and then set the grow size | |
| 3709 * to much bigger value to avoid many malloc calls in case of deep regular | |
| 3710 * expressions. */ | |
| 3711 if (regstack.ga_data == NULL) | |
| 3712 { | |
| 3713 /* Use an item size of 1 byte, since we push different things | |
| 3714 * onto the regstack. */ | |
| 3715 ga_init2(®stack, 1, REGSTACK_INITIAL); | |
| 3716 ga_grow(®stack, REGSTACK_INITIAL); | |
| 3717 regstack.ga_growsize = REGSTACK_INITIAL * 8; | |
| 3718 } | |
| 3719 | |
| 3720 if (backpos.ga_data == NULL) | |
| 3721 { | |
| 3722 ga_init2(&backpos, sizeof(backpos_T), BACKPOS_INITIAL); | |
| 3723 ga_grow(&backpos, BACKPOS_INITIAL); | |
| 3724 backpos.ga_growsize = BACKPOS_INITIAL * 8; | |
| 3725 } | |
| 270 | 3726 |
| 7 | 3727 if (REG_MULTI) |
| 3728 { | |
| 4444 | 3729 prog = (bt_regprog_T *)reg_mmatch->regprog; |
| 7 | 3730 line = reg_getline((linenr_T)0); |
| 3731 reg_startpos = reg_mmatch->startpos; | |
| 3732 reg_endpos = reg_mmatch->endpos; | |
| 3733 } | |
| 3734 else | |
| 3735 { | |
| 4444 | 3736 prog = (bt_regprog_T *)reg_match->regprog; |
| 7 | 3737 reg_startp = reg_match->startp; |
| 3738 reg_endp = reg_match->endp; | |
| 3739 } | |
| 3740 | |
| 3741 /* Be paranoid... */ | |
| 3742 if (prog == NULL || line == NULL) | |
| 3743 { | |
| 3744 EMSG(_(e_null)); | |
| 3745 goto theend; | |
| 3746 } | |
| 3747 | |
| 3748 /* Check validity of program. */ | |
| 3749 if (prog_magic_wrong()) | |
| 3750 goto theend; | |
| 3751 | |
| 410 | 3752 /* If the start column is past the maximum column: no need to try. */ |
| 3753 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3754 goto theend; | |
| 3755 | |
| 7 | 3756 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ |
| 3757 if (prog->regflags & RF_ICASE) | |
| 3758 ireg_ic = TRUE; | |
| 3759 else if (prog->regflags & RF_NOICASE) | |
| 3760 ireg_ic = FALSE; | |
| 3761 | |
| 3762 #ifdef FEAT_MBYTE | |
| 3763 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
| 3764 if (prog->regflags & RF_ICOMBINE) | |
| 3765 ireg_icombine = TRUE; | |
| 3766 #endif | |
| 3767 | |
| 3768 /* If there is a "must appear" string, look for it. */ | |
| 3769 if (prog->regmust != NULL) | |
| 3770 { | |
| 3771 int c; | |
| 3772 | |
| 3773 #ifdef FEAT_MBYTE | |
| 3774 if (has_mbyte) | |
| 3775 c = (*mb_ptr2char)(prog->regmust); | |
| 3776 else | |
| 3777 #endif | |
| 3778 c = *prog->regmust; | |
| 3779 s = line + col; | |
| 170 | 3780 |
| 3781 /* | |
| 3782 * This is used very often, esp. for ":global". Use three versions of | |
| 3783 * the loop to avoid overhead of conditions. | |
| 3784 */ | |
| 3785 if (!ireg_ic | |
| 3786 #ifdef FEAT_MBYTE | |
| 3787 && !has_mbyte | |
| 3788 #endif | |
| 3789 ) | |
| 3790 while ((s = vim_strbyte(s, c)) != NULL) | |
| 3791 { | |
| 3792 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3793 break; /* Found it. */ | |
| 3794 ++s; | |
| 3795 } | |
| 3796 #ifdef FEAT_MBYTE | |
| 3797 else if (!ireg_ic || (!enc_utf8 && mb_char2len(c) > 1)) | |
| 3798 while ((s = vim_strchr(s, c)) != NULL) | |
| 3799 { | |
| 3800 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3801 break; /* Found it. */ | |
| 3802 mb_ptr_adv(s); | |
| 3803 } | |
| 3804 #endif | |
| 3805 else | |
| 3806 while ((s = cstrchr(s, c)) != NULL) | |
| 3807 { | |
| 3808 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3809 break; /* Found it. */ | |
| 3810 mb_ptr_adv(s); | |
| 3811 } | |
| 7 | 3812 if (s == NULL) /* Not present. */ |
| 3813 goto theend; | |
| 3814 } | |
| 3815 | |
| 3816 regline = line; | |
| 3817 reglnum = 0; | |
| 2578 | 3818 reg_toolong = FALSE; |
| 7 | 3819 |
| 3820 /* Simplest case: Anchored match need be tried only once. */ | |
| 3821 if (prog->reganch) | |
| 3822 { | |
| 3823 int c; | |
| 3824 | |
| 3825 #ifdef FEAT_MBYTE | |
| 3826 if (has_mbyte) | |
| 3827 c = (*mb_ptr2char)(regline + col); | |
| 3828 else | |
| 3829 #endif | |
| 3830 c = regline[col]; | |
| 3831 if (prog->regstart == NUL | |
| 3832 || prog->regstart == c | |
| 3833 || (ireg_ic && (( | |
| 3834 #ifdef FEAT_MBYTE | |
| 3835 (enc_utf8 && utf_fold(prog->regstart) == utf_fold(c))) | |
| 3836 || (c < 255 && prog->regstart < 255 && | |
| 3837 #endif | |
| 1347 | 3838 MB_TOLOWER(prog->regstart) == MB_TOLOWER(c))))) |
| 7 | 3839 retval = regtry(prog, col); |
| 3840 else | |
| 3841 retval = 0; | |
| 3842 } | |
| 3843 else | |
| 3844 { | |
| 1521 | 3845 #ifdef FEAT_RELTIME |
| 3846 int tm_count = 0; | |
| 3847 #endif | |
| 7 | 3848 /* Messy cases: unanchored match. */ |
| 180 | 3849 while (!got_int) |
| 7 | 3850 { |
| 3851 if (prog->regstart != NUL) | |
| 3852 { | |
| 170 | 3853 /* Skip until the char we know it must start with. |
| 3854 * Used often, do some work to avoid call overhead. */ | |
| 3855 if (!ireg_ic | |
| 3856 #ifdef FEAT_MBYTE | |
| 3857 && !has_mbyte | |
| 3858 #endif | |
| 3859 ) | |
| 3860 s = vim_strbyte(regline + col, prog->regstart); | |
| 3861 else | |
| 3862 s = cstrchr(regline + col, prog->regstart); | |
| 7 | 3863 if (s == NULL) |
| 3864 { | |
| 3865 retval = 0; | |
| 3866 break; | |
| 3867 } | |
| 3868 col = (int)(s - regline); | |
| 3869 } | |
| 3870 | |
| 410 | 3871 /* Check for maximum column to try. */ |
| 3872 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3873 { | |
| 3874 retval = 0; | |
| 3875 break; | |
| 3876 } | |
| 3877 | |
| 7 | 3878 retval = regtry(prog, col); |
| 3879 if (retval > 0) | |
| 3880 break; | |
| 3881 | |
| 3882 /* if not currently on the first line, get it again */ | |
| 3883 if (reglnum != 0) | |
| 3884 { | |
| 481 | 3885 reglnum = 0; |
| 7 | 3886 regline = reg_getline((linenr_T)0); |
| 3887 } | |
| 3888 if (regline[col] == NUL) | |
| 3889 break; | |
| 3890 #ifdef FEAT_MBYTE | |
| 3891 if (has_mbyte) | |
| 474 | 3892 col += (*mb_ptr2len)(regline + col); |
| 7 | 3893 else |
| 3894 #endif | |
| 3895 ++col; | |
| 1521 | 3896 #ifdef FEAT_RELTIME |
| 3897 /* Check for timeout once in a twenty times to avoid overhead. */ | |
| 3898 if (tm != NULL && ++tm_count == 20) | |
| 3899 { | |
| 3900 tm_count = 0; | |
| 3901 if (profile_passed_limit(tm)) | |
| 3902 break; | |
| 3903 } | |
| 3904 #endif | |
| 7 | 3905 } |
| 3906 } | |
| 3907 | |
| 3908 theend: | |
| 1520 | 3909 /* Free "reg_tofree" when it's a bit big. |
| 3910 * Free regstack and backpos if they are bigger than their initial size. */ | |
| 3911 if (reg_tofreelen > 400) | |
| 3912 { | |
| 3913 vim_free(reg_tofree); | |
| 3914 reg_tofree = NULL; | |
| 3915 } | |
| 3916 if (regstack.ga_maxlen > REGSTACK_INITIAL) | |
| 3917 ga_clear(®stack); | |
| 3918 if (backpos.ga_maxlen > BACKPOS_INITIAL) | |
| 3919 ga_clear(&backpos); | |
| 270 | 3920 |
| 7 | 3921 return retval; |
| 3922 } | |
| 3923 | |
| 3924 #ifdef FEAT_SYN_HL | |
| 3925 static reg_extmatch_T *make_extmatch __ARGS((void)); | |
| 3926 | |
| 3927 /* | |
| 3928 * Create a new extmatch and mark it as referenced once. | |
| 3929 */ | |
| 3930 static reg_extmatch_T * | |
| 3931 make_extmatch() | |
| 3932 { | |
| 3933 reg_extmatch_T *em; | |
| 3934 | |
| 3935 em = (reg_extmatch_T *)alloc_clear((unsigned)sizeof(reg_extmatch_T)); | |
| 3936 if (em != NULL) | |
| 3937 em->refcnt = 1; | |
| 3938 return em; | |
| 3939 } | |
| 3940 | |
| 3941 /* | |
| 3942 * Add a reference to an extmatch. | |
| 3943 */ | |
| 3944 reg_extmatch_T * | |
| 3945 ref_extmatch(em) | |
| 3946 reg_extmatch_T *em; | |
| 3947 { | |
| 3948 if (em != NULL) | |
| 3949 em->refcnt++; | |
| 3950 return em; | |
| 3951 } | |
| 3952 | |
| 3953 /* | |
| 3954 * Remove a reference to an extmatch. If there are no references left, free | |
| 3955 * the info. | |
| 3956 */ | |
| 3957 void | |
| 3958 unref_extmatch(em) | |
| 3959 reg_extmatch_T *em; | |
| 3960 { | |
| 3961 int i; | |
| 3962 | |
| 3963 if (em != NULL && --em->refcnt <= 0) | |
| 3964 { | |
| 3965 for (i = 0; i < NSUBEXP; ++i) | |
| 3966 vim_free(em->matches[i]); | |
| 3967 vim_free(em); | |
| 3968 } | |
| 3969 } | |
| 3970 #endif | |
| 3971 | |
| 3972 /* | |
| 3973 * regtry - try match of "prog" with at regline["col"]. | |
| 3974 * Returns 0 for failure, number of lines contained in the match otherwise. | |
| 3975 */ | |
| 3976 static long | |
| 3977 regtry(prog, col) | |
| 4444 | 3978 bt_regprog_T *prog; |
| 7 | 3979 colnr_T col; |
| 3980 { | |
| 3981 reginput = regline + col; | |
| 3982 need_clear_subexpr = TRUE; | |
| 3983 #ifdef FEAT_SYN_HL | |
| 3984 /* Clear the external match subpointers if necessary. */ | |
| 3985 if (prog->reghasz == REX_SET) | |
| 3986 need_clear_zsubexpr = TRUE; | |
| 3987 #endif | |
| 3988 | |
| 189 | 3989 if (regmatch(prog->program + 1) == 0) |
| 3990 return 0; | |
| 3991 | |
| 3992 cleanup_subexpr(); | |
| 3993 if (REG_MULTI) | |
| 7 | 3994 { |
| 189 | 3995 if (reg_startpos[0].lnum < 0) |
| 7 | 3996 { |
| 189 | 3997 reg_startpos[0].lnum = 0; |
| 3998 reg_startpos[0].col = col; | |
| 3999 } | |
| 4000 if (reg_endpos[0].lnum < 0) | |
| 4001 { | |
| 4002 reg_endpos[0].lnum = reglnum; | |
| 4003 reg_endpos[0].col = (int)(reginput - regline); | |
| 7 | 4004 } |
| 4005 else | |
| 189 | 4006 /* Use line number of "\ze". */ |
| 4007 reglnum = reg_endpos[0].lnum; | |
| 4008 } | |
| 4009 else | |
| 4010 { | |
| 4011 if (reg_startp[0] == NULL) | |
| 4012 reg_startp[0] = regline + col; | |
| 4013 if (reg_endp[0] == NULL) | |
| 4014 reg_endp[0] = reginput; | |
| 4015 } | |
| 7 | 4016 #ifdef FEAT_SYN_HL |
| 189 | 4017 /* Package any found \z(...\) matches for export. Default is none. */ |
| 4018 unref_extmatch(re_extmatch_out); | |
| 4019 re_extmatch_out = NULL; | |
| 4020 | |
| 4021 if (prog->reghasz == REX_SET) | |
| 4022 { | |
| 4023 int i; | |
| 4024 | |
| 4025 cleanup_zsubexpr(); | |
| 4026 re_extmatch_out = make_extmatch(); | |
| 4027 for (i = 0; i < NSUBEXP; i++) | |
| 7 | 4028 { |
| 189 | 4029 if (REG_MULTI) |
| 7 | 4030 { |
| 189 | 4031 /* Only accept single line matches. */ |
| 4032 if (reg_startzpos[i].lnum >= 0 | |
| 4033 && reg_endzpos[i].lnum == reg_startzpos[i].lnum) | |
| 4034 re_extmatch_out->matches[i] = | |
| 4035 vim_strnsave(reg_getline(reg_startzpos[i].lnum) | |
| 7 | 4036 + reg_startzpos[i].col, |
| 189 | 4037 reg_endzpos[i].col - reg_startzpos[i].col); |
| 4038 } | |
| 4039 else | |
| 4040 { | |
| 4041 if (reg_startzp[i] != NULL && reg_endzp[i] != NULL) | |
| 4042 re_extmatch_out->matches[i] = | |
| 7 | 4043 vim_strnsave(reg_startzp[i], |
| 189 | 4044 (int)(reg_endzp[i] - reg_startzp[i])); |
| 7 | 4045 } |
| 4046 } | |
| 189 | 4047 } |
| 7 | 4048 #endif |
| 189 | 4049 return 1 + reglnum; |
| 7 | 4050 } |
| 4051 | |
| 4052 #ifdef FEAT_MBYTE | |
| 4053 static int reg_prev_class __ARGS((void)); | |
| 4054 | |
| 4055 /* | |
| 4056 * Get class of previous character. | |
| 4057 */ | |
| 4058 static int | |
| 4059 reg_prev_class() | |
| 4060 { | |
| 4061 if (reginput > regline) | |
| 4069 | 4062 return mb_get_class_buf(reginput - 1 |
| 4063 - (*mb_head_off)(regline, reginput - 1), reg_buf); | |
| 7 | 4064 return -1; |
| 4065 } | |
| 4066 | |
| 4067 #endif | |
| 39 | 4068 #define ADVANCE_REGINPUT() mb_ptr_adv(reginput) |
| 7 | 4069 |
| 4070 /* | |
| 4071 * The arguments from BRACE_LIMITS are stored here. They are actually local | |
| 4072 * to regmatch(), but they are here to reduce the amount of stack space used | |
| 4073 * (it can be called recursively many times). | |
| 4074 */ | |
| 4075 static long bl_minval; | |
| 4076 static long bl_maxval; | |
| 4077 | |
| 4078 /* | |
| 4079 * regmatch - main matching routine | |
| 4080 * | |
| 180 | 4081 * Conceptually the strategy is simple: Check to see whether the current node |
| 4082 * matches, push an item onto the regstack and loop to see whether the rest | |
| 4083 * matches, and then act accordingly. In practice we make some effort to | |
| 4084 * avoid using the regstack, in particular by going through "ordinary" nodes | |
| 4085 * (that don't need to know whether the rest of the match failed) by a nested | |
| 4086 * loop. | |
| 7 | 4087 * |
| 4088 * Returns TRUE when there is a match. Leaves reginput and reglnum just after | |
| 4089 * the last matched character. | |
| 4090 * Returns FALSE when there is no match. Leaves reginput and reglnum in an | |
| 4091 * undefined state! | |
| 4092 */ | |
| 4093 static int | |
| 180 | 4094 regmatch(scan) |
| 7 | 4095 char_u *scan; /* Current node. */ |
| 4096 { | |
| 180 | 4097 char_u *next; /* Next node. */ |
| 4098 int op; | |
| 4099 int c; | |
| 4100 regitem_T *rp; | |
| 4101 int no; | |
| 4102 int status; /* one of the RA_ values: */ | |
| 4103 #define RA_FAIL 1 /* something failed, abort */ | |
| 4104 #define RA_CONT 2 /* continue in inner loop */ | |
| 4105 #define RA_BREAK 3 /* break inner loop */ | |
| 4106 #define RA_MATCH 4 /* successful match */ | |
| 4107 #define RA_NOMATCH 5 /* didn't match */ | |
| 270 | 4108 |
| 1520 | 4109 /* Make "regstack" and "backpos" empty. They are allocated and freed in |
| 4444 | 4110 * bt_regexec_both() to reduce malloc()/free() calls. */ |
| 270 | 4111 regstack.ga_len = 0; |
| 4112 backpos.ga_len = 0; | |
| 233 | 4113 |
| 180 | 4114 /* |
| 233 | 4115 * Repeat until "regstack" is empty. |
| 180 | 4116 */ |
| 4117 for (;;) | |
| 4118 { | |
| 4444 | 4119 /* Some patterns may cause a long time to match, even though they are not |
| 7 | 4120 * illegal. E.g., "\([a-z]\+\)\+Q". Allow breaking them with CTRL-C. */ |
| 4121 fast_breakcheck(); | |
| 4122 | |
| 4123 #ifdef DEBUG | |
| 4124 if (scan != NULL && regnarrate) | |
| 4125 { | |
| 4444 | 4126 mch_errmsg((char *)regprop(scan)); |
| 7 | 4127 mch_errmsg("(\n"); |
| 4128 } | |
| 4129 #endif | |
| 180 | 4130 |
| 4131 /* | |
| 233 | 4132 * Repeat for items that can be matched sequentially, without using the |
| 180 | 4133 * regstack. |
| 4134 */ | |
| 4135 for (;;) | |
| 7 | 4136 { |
| 180 | 4137 if (got_int || scan == NULL) |
| 4138 { | |
| 4139 status = RA_FAIL; | |
| 4140 break; | |
| 4141 } | |
| 4142 status = RA_CONT; | |
| 4143 | |
| 7 | 4144 #ifdef DEBUG |
| 4145 if (regnarrate) | |
| 4146 { | |
| 4444 | 4147 mch_errmsg((char *)regprop(scan)); |
| 7 | 4148 mch_errmsg("...\n"); |
| 4149 # ifdef FEAT_SYN_HL | |
| 4150 if (re_extmatch_in != NULL) | |
| 4151 { | |
| 4152 int i; | |
| 4153 | |
| 4154 mch_errmsg(_("External submatches:\n")); | |
| 4155 for (i = 0; i < NSUBEXP; i++) | |
| 4156 { | |
| 4157 mch_errmsg(" \""); | |
| 4158 if (re_extmatch_in->matches[i] != NULL) | |
| 4444 | 4159 mch_errmsg((char *)re_extmatch_in->matches[i]); |
| 7 | 4160 mch_errmsg("\"\n"); |
| 4161 } | |
| 4162 } | |
| 4163 # endif | |
| 4164 } | |
| 4165 #endif | |
| 4166 next = regnext(scan); | |
| 4167 | |
| 4168 op = OP(scan); | |
| 4169 /* Check for character class with NL added. */ | |
| 1018 | 4170 if (!reg_line_lbr && WITH_NL(op) && REG_MULTI |
| 4171 && *reginput == NUL && reglnum <= reg_maxline) | |
| 7 | 4172 { |
| 4173 reg_nextline(); | |
| 4174 } | |
| 4175 else if (reg_line_lbr && WITH_NL(op) && *reginput == '\n') | |
| 4176 { | |
| 4177 ADVANCE_REGINPUT(); | |
| 4178 } | |
| 4179 else | |
| 4180 { | |
| 4181 if (WITH_NL(op)) | |
| 180 | 4182 op -= ADD_NL; |
| 7 | 4183 #ifdef FEAT_MBYTE |
| 4184 if (has_mbyte) | |
| 4185 c = (*mb_ptr2char)(reginput); | |
| 4186 else | |
| 4187 #endif | |
| 4188 c = *reginput; | |
| 4189 switch (op) | |
| 4190 { | |
| 4191 case BOL: | |
| 4192 if (reginput != regline) | |
| 180 | 4193 status = RA_NOMATCH; |
| 7 | 4194 break; |
| 4195 | |
| 4196 case EOL: | |
| 4197 if (c != NUL) | |
| 180 | 4198 status = RA_NOMATCH; |
| 7 | 4199 break; |
| 4200 | |
| 4201 case RE_BOF: | |
| 1458 | 4202 /* We're not at the beginning of the file when below the first |
| 4203 * line where we started, not at the start of the line or we | |
| 4204 * didn't start at the first line of the buffer. */ | |
| 7 | 4205 if (reglnum != 0 || reginput != regline |
| 1458 | 4206 || (REG_MULTI && reg_firstlnum > 1)) |
| 180 | 4207 status = RA_NOMATCH; |
| 7 | 4208 break; |
| 4209 | |
| 4210 case RE_EOF: | |
| 4211 if (reglnum != reg_maxline || c != NUL) | |
| 180 | 4212 status = RA_NOMATCH; |
| 7 | 4213 break; |
| 4214 | |
| 4215 case CURSOR: | |
| 4216 /* Check if the buffer is in a window and compare the | |
| 4217 * reg_win->w_cursor position to the match position. */ | |
| 4218 if (reg_win == NULL | |
| 4219 || (reglnum + reg_firstlnum != reg_win->w_cursor.lnum) | |
| 4220 || ((colnr_T)(reginput - regline) != reg_win->w_cursor.col)) | |
| 180 | 4221 status = RA_NOMATCH; |
| 7 | 4222 break; |
| 4223 | |
| 639 | 4224 case RE_MARK: |
| 4225 /* Compare the mark position to the match position. NOTE: Always | |
| 4226 * uses the current buffer. */ | |
| 4227 { | |
| 4228 int mark = OPERAND(scan)[0]; | |
| 4229 int cmp = OPERAND(scan)[1]; | |
| 4230 pos_T *pos; | |
| 4231 | |
| 4043 | 4232 pos = getmark_buf(reg_buf, mark, FALSE); |
| 1148 | 4233 if (pos == NULL /* mark doesn't exist */ |
| 639 | 4234 || pos->lnum <= 0 /* mark isn't set (in curbuf) */ |
| 4235 || (pos->lnum == reglnum + reg_firstlnum | |
| 4236 ? (pos->col == (colnr_T)(reginput - regline) | |
| 4237 ? (cmp == '<' || cmp == '>') | |
| 4238 : (pos->col < (colnr_T)(reginput - regline) | |
| 4239 ? cmp != '>' | |
| 4240 : cmp != '<')) | |
| 4241 : (pos->lnum < reglnum + reg_firstlnum | |
| 4242 ? cmp != '>' | |
| 4243 : cmp != '<'))) | |
| 4244 status = RA_NOMATCH; | |
| 4245 } | |
| 4246 break; | |
| 4247 | |
| 4248 case RE_VISUAL: | |
| 4249 #ifdef FEAT_VISUAL | |
| 4250 /* Check if the buffer is the current buffer. and whether the | |
| 4251 * position is inside the Visual area. */ | |
| 4252 if (reg_buf != curbuf || VIsual.lnum == 0) | |
| 4253 status = RA_NOMATCH; | |
| 4254 else | |
| 4255 { | |
| 4256 pos_T top, bot; | |
| 4257 linenr_T lnum; | |
| 4258 colnr_T col; | |
| 4259 win_T *wp = reg_win == NULL ? curwin : reg_win; | |
| 4260 int mode; | |
| 4261 | |
| 4262 if (VIsual_active) | |
| 4263 { | |
| 4264 if (lt(VIsual, wp->w_cursor)) | |
| 4265 { | |
| 4266 top = VIsual; | |
| 4267 bot = wp->w_cursor; | |
| 4268 } | |
| 4269 else | |
| 4270 { | |
| 4271 top = wp->w_cursor; | |
| 4272 bot = VIsual; | |
| 4273 } | |
| 4274 mode = VIsual_mode; | |
| 4275 } | |
| 4276 else | |
| 4277 { | |
| 692 | 4278 if (lt(curbuf->b_visual.vi_start, curbuf->b_visual.vi_end)) |
| 685 | 4279 { |
| 692 | 4280 top = curbuf->b_visual.vi_start; |
| 4281 bot = curbuf->b_visual.vi_end; | |
| 685 | 4282 } |
| 4283 else | |
| 4284 { | |
| 692 | 4285 top = curbuf->b_visual.vi_end; |
| 4286 bot = curbuf->b_visual.vi_start; | |
| 685 | 4287 } |
| 692 | 4288 mode = curbuf->b_visual.vi_mode; |
| 639 | 4289 } |
| 4290 lnum = reglnum + reg_firstlnum; | |
| 4291 col = (colnr_T)(reginput - regline); | |
| 4292 if (lnum < top.lnum || lnum > bot.lnum) | |
| 4293 status = RA_NOMATCH; | |
| 4294 else if (mode == 'v') | |
| 4295 { | |
| 4296 if ((lnum == top.lnum && col < top.col) | |
| 4297 || (lnum == bot.lnum | |
| 4298 && col >= bot.col + (*p_sel != 'e'))) | |
| 4299 status = RA_NOMATCH; | |
| 4300 } | |
| 4301 else if (mode == Ctrl_V) | |
| 4302 { | |
| 4303 colnr_T start, end; | |
| 4304 colnr_T start2, end2; | |
| 944 | 4305 colnr_T cols; |
| 639 | 4306 |
| 4307 getvvcol(wp, &top, &start, NULL, &end); | |
| 4308 getvvcol(wp, &bot, &start2, NULL, &end2); | |
| 4309 if (start2 < start) | |
| 4310 start = start2; | |
| 4311 if (end2 > end) | |
| 4312 end = end2; | |
| 4313 if (top.col == MAXCOL || bot.col == MAXCOL) | |
| 4314 end = MAXCOL; | |
| 944 | 4315 cols = win_linetabsize(wp, |
| 639 | 4316 regline, (colnr_T)(reginput - regline)); |
| 944 | 4317 if (cols < start || cols > end - (*p_sel == 'e')) |
| 639 | 4318 status = RA_NOMATCH; |
| 4319 } | |
| 4320 } | |
| 4321 #else | |
| 4322 status = RA_NOMATCH; | |
| 4323 #endif | |
| 4324 break; | |
| 4325 | |
| 7 | 4326 case RE_LNUM: |
| 4327 if (!REG_MULTI || !re_num_cmp((long_u)(reglnum + reg_firstlnum), | |
| 4328 scan)) | |
| 180 | 4329 status = RA_NOMATCH; |
| 7 | 4330 break; |
| 4331 | |
| 4332 case RE_COL: | |
| 4333 if (!re_num_cmp((long_u)(reginput - regline) + 1, scan)) | |
| 180 | 4334 status = RA_NOMATCH; |
| 7 | 4335 break; |
| 4336 | |
| 4337 case RE_VCOL: | |
| 4338 if (!re_num_cmp((long_u)win_linetabsize( | |
| 4339 reg_win == NULL ? curwin : reg_win, | |
| 4340 regline, (colnr_T)(reginput - regline)) + 1, scan)) | |
| 180 | 4341 status = RA_NOMATCH; |
| 7 | 4342 break; |
| 4343 | |
| 4344 case BOW: /* \<word; reginput points to w */ | |
| 4345 if (c == NUL) /* Can't match at end of line */ | |
| 180 | 4346 status = RA_NOMATCH; |
| 7 | 4347 #ifdef FEAT_MBYTE |
| 180 | 4348 else if (has_mbyte) |
| 7 | 4349 { |
| 4350 int this_class; | |
| 4351 | |
| 4352 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4353 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4354 if (this_class <= 1) |
| 180 | 4355 status = RA_NOMATCH; /* not on a word at all */ |
| 4356 else if (reg_prev_class() == this_class) | |
| 4357 status = RA_NOMATCH; /* previous char is in same word */ | |
| 7 | 4358 } |
| 4359 #endif | |
| 4360 else | |
| 4361 { | |
| 4061 | 4362 if (!vim_iswordc_buf(c, reg_buf) || (reginput > regline |
| 4363 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
| 180 | 4364 status = RA_NOMATCH; |
| 7 | 4365 } |
| 4366 break; | |
| 4367 | |
| 4368 case EOW: /* word\>; reginput points after d */ | |
| 4369 if (reginput == regline) /* Can't match at start of line */ | |
| 180 | 4370 status = RA_NOMATCH; |
| 7 | 4371 #ifdef FEAT_MBYTE |
| 180 | 4372 else if (has_mbyte) |
| 7 | 4373 { |
| 4374 int this_class, prev_class; | |
| 4375 | |
| 4376 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4377 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4378 prev_class = reg_prev_class(); |
| 180 | 4379 if (this_class == prev_class |
| 4380 || prev_class == 0 || prev_class == 1) | |
| 4381 status = RA_NOMATCH; | |
| 7 | 4382 } |
| 180 | 4383 #endif |
| 7 | 4384 else |
| 4385 { | |
| 4043 | 4386 if (!vim_iswordc_buf(reginput[-1], reg_buf) |
| 4387 || (reginput[0] != NUL && vim_iswordc_buf(c, reg_buf))) | |
| 180 | 4388 status = RA_NOMATCH; |
| 7 | 4389 } |
| 4390 break; /* Matched with EOW */ | |
| 4391 | |
| 4392 case ANY: | |
| 4084 | 4393 /* ANY does not match new lines. */ |
| 7 | 4394 if (c == NUL) |
| 180 | 4395 status = RA_NOMATCH; |
| 4396 else | |
| 4397 ADVANCE_REGINPUT(); | |
| 7 | 4398 break; |
| 4399 | |
| 4400 case IDENT: | |
| 4401 if (!vim_isIDc(c)) | |
| 180 | 4402 status = RA_NOMATCH; |
| 4403 else | |
| 4404 ADVANCE_REGINPUT(); | |
| 7 | 4405 break; |
| 4406 | |
| 4407 case SIDENT: | |
| 4408 if (VIM_ISDIGIT(*reginput) || !vim_isIDc(c)) | |
| 180 | 4409 status = RA_NOMATCH; |
| 4410 else | |
| 4411 ADVANCE_REGINPUT(); | |
| 7 | 4412 break; |
| 4413 | |
| 4414 case KWORD: | |
| 4069 | 4415 if (!vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4416 status = RA_NOMATCH; |
| 4417 else | |
| 4418 ADVANCE_REGINPUT(); | |
| 7 | 4419 break; |
| 4420 | |
| 4421 case SKWORD: | |
| 4069 | 4422 if (VIM_ISDIGIT(*reginput) || !vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4423 status = RA_NOMATCH; |
| 4424 else | |
| 4425 ADVANCE_REGINPUT(); | |
| 7 | 4426 break; |
| 4427 | |
| 4428 case FNAME: | |
| 4429 if (!vim_isfilec(c)) | |
| 180 | 4430 status = RA_NOMATCH; |
| 4431 else | |
| 4432 ADVANCE_REGINPUT(); | |
| 7 | 4433 break; |
| 4434 | |
| 4435 case SFNAME: | |
| 4436 if (VIM_ISDIGIT(*reginput) || !vim_isfilec(c)) | |
| 180 | 4437 status = RA_NOMATCH; |
| 4438 else | |
| 4439 ADVANCE_REGINPUT(); | |
| 7 | 4440 break; |
| 4441 | |
| 4442 case PRINT: | |
| 4443 if (ptr2cells(reginput) != 1) | |
| 180 | 4444 status = RA_NOMATCH; |
| 4445 else | |
| 4446 ADVANCE_REGINPUT(); | |
| 7 | 4447 break; |
| 4448 | |
| 4449 case SPRINT: | |
| 4450 if (VIM_ISDIGIT(*reginput) || ptr2cells(reginput) != 1) | |
| 180 | 4451 status = RA_NOMATCH; |
| 4452 else | |
| 4453 ADVANCE_REGINPUT(); | |
| 7 | 4454 break; |
| 4455 | |
| 4456 case WHITE: | |
| 4457 if (!vim_iswhite(c)) | |
| 180 | 4458 status = RA_NOMATCH; |
| 4459 else | |
| 4460 ADVANCE_REGINPUT(); | |
| 7 | 4461 break; |
| 4462 | |
| 4463 case NWHITE: | |
| 4464 if (c == NUL || vim_iswhite(c)) | |
| 180 | 4465 status = RA_NOMATCH; |
| 4466 else | |
| 4467 ADVANCE_REGINPUT(); | |
| 7 | 4468 break; |
| 4469 | |
| 4470 case DIGIT: | |
| 4471 if (!ri_digit(c)) | |
| 180 | 4472 status = RA_NOMATCH; |
| 4473 else | |
| 4474 ADVANCE_REGINPUT(); | |
| 7 | 4475 break; |
| 4476 | |
| 4477 case NDIGIT: | |
| 4478 if (c == NUL || ri_digit(c)) | |
| 180 | 4479 status = RA_NOMATCH; |
| 4480 else | |
| 4481 ADVANCE_REGINPUT(); | |
| 7 | 4482 break; |
| 4483 | |
| 4484 case HEX: | |
| 4485 if (!ri_hex(c)) | |
| 180 | 4486 status = RA_NOMATCH; |
| 4487 else | |
| 4488 ADVANCE_REGINPUT(); | |
| 7 | 4489 break; |
| 4490 | |
| 4491 case NHEX: | |
| 4492 if (c == NUL || ri_hex(c)) | |
| 180 | 4493 status = RA_NOMATCH; |
| 4494 else | |
| 4495 ADVANCE_REGINPUT(); | |
| 7 | 4496 break; |
| 4497 | |
| 4498 case OCTAL: | |
| 4499 if (!ri_octal(c)) | |
| 180 | 4500 status = RA_NOMATCH; |
| 4501 else | |
| 4502 ADVANCE_REGINPUT(); | |
| 7 | 4503 break; |
| 4504 | |
| 4505 case NOCTAL: | |
| 4506 if (c == NUL || ri_octal(c)) | |
| 180 | 4507 status = RA_NOMATCH; |
| 4508 else | |
| 4509 ADVANCE_REGINPUT(); | |
| 7 | 4510 break; |
| 4511 | |
| 4512 case WORD: | |
| 4513 if (!ri_word(c)) | |
| 180 | 4514 status = RA_NOMATCH; |
| 4515 else | |
| 4516 ADVANCE_REGINPUT(); | |
| 7 | 4517 break; |
| 4518 | |
| 4519 case NWORD: | |
| 4520 if (c == NUL || ri_word(c)) | |
| 180 | 4521 status = RA_NOMATCH; |
| 4522 else | |
| 4523 ADVANCE_REGINPUT(); | |
| 7 | 4524 break; |
| 4525 | |
| 4526 case HEAD: | |
| 4527 if (!ri_head(c)) | |
| 180 | 4528 status = RA_NOMATCH; |
| 4529 else | |
| 4530 ADVANCE_REGINPUT(); | |
| 7 | 4531 break; |
| 4532 | |
| 4533 case NHEAD: | |
| 4534 if (c == NUL || ri_head(c)) | |
| 180 | 4535 status = RA_NOMATCH; |
| 4536 else | |
| 4537 ADVANCE_REGINPUT(); | |
| 7 | 4538 break; |
| 4539 | |
| 4540 case ALPHA: | |
| 4541 if (!ri_alpha(c)) | |
| 180 | 4542 status = RA_NOMATCH; |
| 4543 else | |
| 4544 ADVANCE_REGINPUT(); | |
| 7 | 4545 break; |
| 4546 | |
| 4547 case NALPHA: | |
| 4548 if (c == NUL || ri_alpha(c)) | |
| 180 | 4549 status = RA_NOMATCH; |
| 4550 else | |
| 4551 ADVANCE_REGINPUT(); | |
| 7 | 4552 break; |
| 4553 | |
| 4554 case LOWER: | |
| 4555 if (!ri_lower(c)) | |
| 180 | 4556 status = RA_NOMATCH; |
| 4557 else | |
| 4558 ADVANCE_REGINPUT(); | |
| 7 | 4559 break; |
| 4560 | |
| 4561 case NLOWER: | |
| 4562 if (c == NUL || ri_lower(c)) | |
| 180 | 4563 status = RA_NOMATCH; |
| 4564 else | |
| 4565 ADVANCE_REGINPUT(); | |
| 7 | 4566 break; |
| 4567 | |
| 4568 case UPPER: | |
| 4569 if (!ri_upper(c)) | |
| 180 | 4570 status = RA_NOMATCH; |
| 4571 else | |
| 4572 ADVANCE_REGINPUT(); | |
| 7 | 4573 break; |
| 4574 | |
| 4575 case NUPPER: | |
| 4576 if (c == NUL || ri_upper(c)) | |
| 180 | 4577 status = RA_NOMATCH; |
| 4578 else | |
| 4579 ADVANCE_REGINPUT(); | |
| 7 | 4580 break; |
| 4581 | |
| 4582 case EXACTLY: | |
| 4583 { | |
| 4584 int len; | |
| 4585 char_u *opnd; | |
| 4586 | |
| 4587 opnd = OPERAND(scan); | |
| 4588 /* Inline the first byte, for speed. */ | |
| 4589 if (*opnd != *reginput | |
| 4590 && (!ireg_ic || ( | |
| 4591 #ifdef FEAT_MBYTE | |
| 4592 !enc_utf8 && | |
| 4593 #endif | |
| 1347 | 4594 MB_TOLOWER(*opnd) != MB_TOLOWER(*reginput)))) |
| 180 | 4595 status = RA_NOMATCH; |
| 4596 else if (*opnd == NUL) | |
| 7 | 4597 { |
| 4598 /* match empty string always works; happens when "~" is | |
| 4599 * empty. */ | |
| 4600 } | |
| 4601 else if (opnd[1] == NUL | |
| 4602 #ifdef FEAT_MBYTE | |
| 4603 && !(enc_utf8 && ireg_ic) | |
| 4604 #endif | |
| 4605 ) | |
| 4606 ++reginput; /* matched a single char */ | |
| 4607 else | |
| 4608 { | |
| 4609 len = (int)STRLEN(opnd); | |
| 4610 /* Need to match first byte again for multi-byte. */ | |
| 4611 if (cstrncmp(opnd, reginput, &len) != 0) | |
| 180 | 4612 status = RA_NOMATCH; |
| 7 | 4613 #ifdef FEAT_MBYTE |
| 4614 /* Check for following composing character. */ | |
| 180 | 4615 else if (enc_utf8 |
| 4616 && UTF_COMPOSINGLIKE(reginput, reginput + len)) | |
| 7 | 4617 { |
| 4618 /* raaron: This code makes a composing character get | |
| 4619 * ignored, which is the correct behavior (sometimes) | |
| 4620 * for voweled Hebrew texts. */ | |
| 4621 if (!ireg_icombine) | |
| 180 | 4622 status = RA_NOMATCH; |
| 7 | 4623 } |
| 180 | 4624 #endif |
| 7 | 4625 else |
| 180 | 4626 reginput += len; |
| 7 | 4627 } |
| 4628 } | |
| 4629 break; | |
| 4630 | |
| 4631 case ANYOF: | |
| 4632 case ANYBUT: | |
| 4633 if (c == NUL) | |
| 180 | 4634 status = RA_NOMATCH; |
| 4635 else if ((cstrchr(OPERAND(scan), c) == NULL) == (op == ANYOF)) | |
| 4636 status = RA_NOMATCH; | |
| 4637 else | |
| 4638 ADVANCE_REGINPUT(); | |
| 7 | 4639 break; |
| 4640 | |
| 4641 #ifdef FEAT_MBYTE | |
| 4642 case MULTIBYTECODE: | |
| 4643 if (has_mbyte) | |
| 4644 { | |
| 4645 int i, len; | |
| 4646 char_u *opnd; | |
| 944 | 4647 int opndc = 0, inpc; |
| 7 | 4648 |
| 4649 opnd = OPERAND(scan); | |
| 4650 /* Safety check (just in case 'encoding' was changed since | |
| 4651 * compiling the program). */ | |
| 474 | 4652 if ((len = (*mb_ptr2len)(opnd)) < 2) |
| 180 | 4653 { |
| 4654 status = RA_NOMATCH; | |
| 4655 break; | |
| 4656 } | |
| 714 | 4657 if (enc_utf8) |
| 4658 opndc = mb_ptr2char(opnd); | |
| 4659 if (enc_utf8 && utf_iscomposing(opndc)) | |
| 4660 { | |
| 4661 /* When only a composing char is given match at any | |
| 4662 * position where that composing char appears. */ | |
| 4663 status = RA_NOMATCH; | |
| 4664 for (i = 0; reginput[i] != NUL; i += utf_char2len(inpc)) | |
| 180 | 4665 { |
| 714 | 4666 inpc = mb_ptr2char(reginput + i); |
| 4667 if (!utf_iscomposing(inpc)) | |
| 4668 { | |
| 4669 if (i > 0) | |
| 4670 break; | |
| 4671 } | |
| 4672 else if (opndc == inpc) | |
| 4673 { | |
| 4674 /* Include all following composing chars. */ | |
| 4675 len = i + mb_ptr2len(reginput + i); | |
| 4676 status = RA_MATCH; | |
| 4677 break; | |
| 4678 } | |
| 180 | 4679 } |
| 714 | 4680 } |
| 4681 else | |
| 4682 for (i = 0; i < len; ++i) | |
| 4683 if (opnd[i] != reginput[i]) | |
| 4684 { | |
| 4685 status = RA_NOMATCH; | |
| 4686 break; | |
| 4687 } | |
| 7 | 4688 reginput += len; |
| 4689 } | |
| 4690 else | |
| 180 | 4691 status = RA_NOMATCH; |
| 7 | 4692 break; |
| 4693 #endif | |
| 4694 | |
| 4695 case NOTHING: | |
| 4696 break; | |
| 4697 | |
| 4698 case BACK: | |
| 233 | 4699 { |
| 4700 int i; | |
| 4701 backpos_T *bp; | |
| 4702 | |
| 4703 /* | |
| 4704 * When we run into BACK we need to check if we don't keep | |
| 4705 * looping without matching any input. The second and later | |
| 4706 * times a BACK is encountered it fails if the input is still | |
| 4707 * at the same position as the previous time. | |
| 4708 * The positions are stored in "backpos" and found by the | |
| 4709 * current value of "scan", the position in the RE program. | |
| 4710 */ | |
| 4711 bp = (backpos_T *)backpos.ga_data; | |
| 4712 for (i = 0; i < backpos.ga_len; ++i) | |
| 4713 if (bp[i].bp_scan == scan) | |
| 4714 break; | |
| 4715 if (i == backpos.ga_len) | |
| 4716 { | |
| 4717 /* First time at this BACK, make room to store the pos. */ | |
| 4718 if (ga_grow(&backpos, 1) == FAIL) | |
| 4719 status = RA_FAIL; | |
| 4720 else | |
| 4721 { | |
| 4722 /* get "ga_data" again, it may have changed */ | |
| 4723 bp = (backpos_T *)backpos.ga_data; | |
| 4724 bp[i].bp_scan = scan; | |
| 4725 ++backpos.ga_len; | |
| 4726 } | |
| 4727 } | |
| 4728 else if (reg_save_equal(&bp[i].bp_pos)) | |
| 4729 /* Still at same position as last time, fail. */ | |
| 4730 status = RA_NOMATCH; | |
| 4731 | |
| 4732 if (status != RA_FAIL && status != RA_NOMATCH) | |
| 4733 reg_save(&bp[i].bp_pos, &backpos); | |
| 4734 } | |
| 179 | 4735 break; |
| 4736 | |
| 7 | 4737 case MOPEN + 0: /* Match start: \zs */ |
| 4738 case MOPEN + 1: /* \( */ | |
| 4739 case MOPEN + 2: | |
| 4740 case MOPEN + 3: | |
| 4741 case MOPEN + 4: | |
| 4742 case MOPEN + 5: | |
| 4743 case MOPEN + 6: | |
| 4744 case MOPEN + 7: | |
| 4745 case MOPEN + 8: | |
| 4746 case MOPEN + 9: | |
| 4747 { | |
| 4748 no = op - MOPEN; | |
| 4749 cleanup_subexpr(); | |
| 270 | 4750 rp = regstack_push(RS_MOPEN, scan); |
| 180 | 4751 if (rp == NULL) |
| 4752 status = RA_FAIL; | |
| 4753 else | |
| 4754 { | |
| 4755 rp->rs_no = no; | |
| 4756 save_se(&rp->rs_un.sesave, ®_startpos[no], | |
| 4757 ®_startp[no]); | |
| 4758 /* We simply continue and handle the result when done. */ | |
| 4759 } | |
| 7 | 4760 } |
| 180 | 4761 break; |
| 7 | 4762 |
| 4763 case NOPEN: /* \%( */ | |
| 4764 case NCLOSE: /* \) after \%( */ | |
| 270 | 4765 if (regstack_push(RS_NOPEN, scan) == NULL) |
| 180 | 4766 status = RA_FAIL; |
| 4767 /* We simply continue and handle the result when done. */ | |
| 4768 break; | |
| 7 | 4769 |
| 4770 #ifdef FEAT_SYN_HL | |
| 4771 case ZOPEN + 1: | |
| 4772 case ZOPEN + 2: | |
| 4773 case ZOPEN + 3: | |
| 4774 case ZOPEN + 4: | |
| 4775 case ZOPEN + 5: | |
| 4776 case ZOPEN + 6: | |
| 4777 case ZOPEN + 7: | |
| 4778 case ZOPEN + 8: | |
| 4779 case ZOPEN + 9: | |
| 4780 { | |
| 4781 no = op - ZOPEN; | |
| 4782 cleanup_zsubexpr(); | |
| 270 | 4783 rp = regstack_push(RS_ZOPEN, scan); |
| 180 | 4784 if (rp == NULL) |
| 4785 status = RA_FAIL; | |
| 4786 else | |
| 4787 { | |
| 4788 rp->rs_no = no; | |
| 4789 save_se(&rp->rs_un.sesave, ®_startzpos[no], | |
| 4790 ®_startzp[no]); | |
| 4791 /* We simply continue and handle the result when done. */ | |
| 4792 } | |
| 7 | 4793 } |
| 180 | 4794 break; |
| 7 | 4795 #endif |
| 4796 | |
| 4797 case MCLOSE + 0: /* Match end: \ze */ | |
| 4798 case MCLOSE + 1: /* \) */ | |
| 4799 case MCLOSE + 2: | |
| 4800 case MCLOSE + 3: | |
| 4801 case MCLOSE + 4: | |
| 4802 case MCLOSE + 5: | |
| 4803 case MCLOSE + 6: | |
| 4804 case MCLOSE + 7: | |
| 4805 case MCLOSE + 8: | |
| 4806 case MCLOSE + 9: | |
| 4807 { | |
| 4808 no = op - MCLOSE; | |
| 4809 cleanup_subexpr(); | |
| 270 | 4810 rp = regstack_push(RS_MCLOSE, scan); |
| 180 | 4811 if (rp == NULL) |
| 4812 status = RA_FAIL; | |
| 4813 else | |
| 4814 { | |
| 4815 rp->rs_no = no; | |
| 4816 save_se(&rp->rs_un.sesave, ®_endpos[no], ®_endp[no]); | |
| 4817 /* We simply continue and handle the result when done. */ | |
| 4818 } | |
| 7 | 4819 } |
| 180 | 4820 break; |
| 7 | 4821 |
| 4822 #ifdef FEAT_SYN_HL | |
| 4823 case ZCLOSE + 1: /* \) after \z( */ | |
| 4824 case ZCLOSE + 2: | |
| 4825 case ZCLOSE + 3: | |
| 4826 case ZCLOSE + 4: | |
| 4827 case ZCLOSE + 5: | |
| 4828 case ZCLOSE + 6: | |
| 4829 case ZCLOSE + 7: | |
| 4830 case ZCLOSE + 8: | |
| 4831 case ZCLOSE + 9: | |
| 4832 { | |
| 4833 no = op - ZCLOSE; | |
| 4834 cleanup_zsubexpr(); | |
| 270 | 4835 rp = regstack_push(RS_ZCLOSE, scan); |
| 180 | 4836 if (rp == NULL) |
| 4837 status = RA_FAIL; | |
| 4838 else | |
| 4839 { | |
| 4840 rp->rs_no = no; | |
| 4841 save_se(&rp->rs_un.sesave, ®_endzpos[no], | |
| 4842 ®_endzp[no]); | |
| 4843 /* We simply continue and handle the result when done. */ | |
| 4844 } | |
| 7 | 4845 } |
| 180 | 4846 break; |
| 7 | 4847 #endif |
| 4848 | |
| 4849 case BACKREF + 1: | |
| 4850 case BACKREF + 2: | |
| 4851 case BACKREF + 3: | |
| 4852 case BACKREF + 4: | |
| 4853 case BACKREF + 5: | |
| 4854 case BACKREF + 6: | |
| 4855 case BACKREF + 7: | |
| 4856 case BACKREF + 8: | |
| 4857 case BACKREF + 9: | |
| 4858 { | |
| 4859 int len; | |
| 4860 linenr_T clnum; | |
| 4861 colnr_T ccol; | |
| 4862 char_u *p; | |
| 4863 | |
| 4864 no = op - BACKREF; | |
| 4865 cleanup_subexpr(); | |
| 4866 if (!REG_MULTI) /* Single-line regexp */ | |
| 4867 { | |
| 1815 | 4868 if (reg_startp[no] == NULL || reg_endp[no] == NULL) |
| 7 | 4869 { |
| 4870 /* Backref was not set: Match an empty string. */ | |
| 4871 len = 0; | |
| 4872 } | |
| 4873 else | |
| 4874 { | |
| 4875 /* Compare current input with back-ref in the same | |
| 4876 * line. */ | |
| 4877 len = (int)(reg_endp[no] - reg_startp[no]); | |
| 4878 if (cstrncmp(reg_startp[no], reginput, &len) != 0) | |
| 180 | 4879 status = RA_NOMATCH; |
| 7 | 4880 } |
| 4881 } | |
| 4882 else /* Multi-line regexp */ | |
| 4883 { | |
| 1815 | 4884 if (reg_startpos[no].lnum < 0 || reg_endpos[no].lnum < 0) |
| 7 | 4885 { |
| 4886 /* Backref was not set: Match an empty string. */ | |
| 4887 len = 0; | |
| 4888 } | |
| 4889 else | |
| 4890 { | |
| 4891 if (reg_startpos[no].lnum == reglnum | |
| 4892 && reg_endpos[no].lnum == reglnum) | |
| 4893 { | |
| 4894 /* Compare back-ref within the current line. */ | |
| 4895 len = reg_endpos[no].col - reg_startpos[no].col; | |
| 4896 if (cstrncmp(regline + reg_startpos[no].col, | |
| 4897 reginput, &len) != 0) | |
| 180 | 4898 status = RA_NOMATCH; |
| 7 | 4899 } |
| 4900 else | |
| 4901 { | |
| 4902 /* Messy situation: Need to compare between two | |
| 4903 * lines. */ | |
| 4904 ccol = reg_startpos[no].col; | |
| 4905 clnum = reg_startpos[no].lnum; | |
| 4906 for (;;) | |
| 4907 { | |
| 4908 /* Since getting one line may invalidate | |
| 4909 * the other, need to make copy. Slow! */ | |
| 4910 if (regline != reg_tofree) | |
| 4911 { | |
| 4912 len = (int)STRLEN(regline); | |
| 4913 if (reg_tofree == NULL | |
| 4914 || len >= (int)reg_tofreelen) | |
| 4915 { | |
| 4916 len += 50; /* get some extra */ | |
| 4917 vim_free(reg_tofree); | |
| 4918 reg_tofree = alloc(len); | |
| 4919 if (reg_tofree == NULL) | |
| 180 | 4920 { |
| 4921 status = RA_FAIL; /* outof memory!*/ | |
| 4922 break; | |
| 4923 } | |
| 7 | 4924 reg_tofreelen = len; |
| 4925 } | |
| 4926 STRCPY(reg_tofree, regline); | |
| 4927 reginput = reg_tofree | |
| 4928 + (reginput - regline); | |
| 4929 regline = reg_tofree; | |
| 4930 } | |
| 4931 | |
| 4932 /* Get the line to compare with. */ | |
| 4933 p = reg_getline(clnum); | |
| 4934 if (clnum == reg_endpos[no].lnum) | |
| 4935 len = reg_endpos[no].col - ccol; | |
| 4936 else | |
| 4937 len = (int)STRLEN(p + ccol); | |
| 4938 | |
| 4939 if (cstrncmp(p + ccol, reginput, &len) != 0) | |
| 180 | 4940 { |
| 4941 status = RA_NOMATCH; /* doesn't match */ | |
| 4942 break; | |
| 4943 } | |
| 7 | 4944 if (clnum == reg_endpos[no].lnum) |
| 4945 break; /* match and at end! */ | |
| 481 | 4946 if (reglnum >= reg_maxline) |
| 180 | 4947 { |
| 4948 status = RA_NOMATCH; /* text too short */ | |
| 4949 break; | |
| 4950 } | |
| 7 | 4951 |
| 4952 /* Advance to next line. */ | |
| 4953 reg_nextline(); | |
| 4954 ++clnum; | |
| 4955 ccol = 0; | |
| 180 | 4956 if (got_int) |
| 4957 { | |
| 4958 status = RA_FAIL; | |
| 4959 break; | |
| 4960 } | |
| 7 | 4961 } |
| 4962 | |
| 4963 /* found a match! Note that regline may now point | |
| 4964 * to a copy of the line, that should not matter. */ | |
| 4965 } | |
| 4966 } | |
| 4967 } | |
| 4968 | |
| 4969 /* Matched the backref, skip over it. */ | |
| 4970 reginput += len; | |
| 4971 } | |
| 4972 break; | |
| 4973 | |
| 4974 #ifdef FEAT_SYN_HL | |
| 4975 case ZREF + 1: | |
| 4976 case ZREF + 2: | |
| 4977 case ZREF + 3: | |
| 4978 case ZREF + 4: | |
| 4979 case ZREF + 5: | |
| 4980 case ZREF + 6: | |
| 4981 case ZREF + 7: | |
| 4982 case ZREF + 8: | |
| 4983 case ZREF + 9: | |
| 4984 { | |
| 4985 int len; | |
| 4986 | |
| 4987 cleanup_zsubexpr(); | |
| 4988 no = op - ZREF; | |
| 4989 if (re_extmatch_in != NULL | |
| 4990 && re_extmatch_in->matches[no] != NULL) | |
| 4991 { | |
| 4992 len = (int)STRLEN(re_extmatch_in->matches[no]); | |
| 4993 if (cstrncmp(re_extmatch_in->matches[no], | |
| 4994 reginput, &len) != 0) | |
| 180 | 4995 status = RA_NOMATCH; |
| 4996 else | |
| 4997 reginput += len; | |
| 7 | 4998 } |
| 4999 else | |
| 5000 { | |
| 5001 /* Backref was not set: Match an empty string. */ | |
| 5002 } | |
| 5003 } | |
| 5004 break; | |
| 5005 #endif | |
| 5006 | |
| 5007 case BRANCH: | |
| 5008 { | |
| 5009 if (OP(next) != BRANCH) /* No choice. */ | |
| 5010 next = OPERAND(scan); /* Avoid recursion. */ | |
| 5011 else | |
| 5012 { | |
| 270 | 5013 rp = regstack_push(RS_BRANCH, scan); |
| 180 | 5014 if (rp == NULL) |
| 5015 status = RA_FAIL; | |
| 5016 else | |
| 5017 status = RA_BREAK; /* rest is below */ | |
| 7 | 5018 } |
| 5019 } | |
| 5020 break; | |
| 5021 | |
| 5022 case BRACE_LIMITS: | |
| 5023 { | |
| 5024 if (OP(next) == BRACE_SIMPLE) | |
| 5025 { | |
| 5026 bl_minval = OPERAND_MIN(scan); | |
| 5027 bl_maxval = OPERAND_MAX(scan); | |
| 5028 } | |
| 5029 else if (OP(next) >= BRACE_COMPLEX | |
| 5030 && OP(next) < BRACE_COMPLEX + 10) | |
| 5031 { | |
| 5032 no = OP(next) - BRACE_COMPLEX; | |
| 5033 brace_min[no] = OPERAND_MIN(scan); | |
| 5034 brace_max[no] = OPERAND_MAX(scan); | |
| 5035 brace_count[no] = 0; | |
| 5036 } | |
| 5037 else | |
| 5038 { | |
| 5039 EMSG(_(e_internal)); /* Shouldn't happen */ | |
| 180 | 5040 status = RA_FAIL; |
| 7 | 5041 } |
| 5042 } | |
| 5043 break; | |
| 5044 | |
| 5045 case BRACE_COMPLEX + 0: | |
| 5046 case BRACE_COMPLEX + 1: | |
| 5047 case BRACE_COMPLEX + 2: | |
| 5048 case BRACE_COMPLEX + 3: | |
| 5049 case BRACE_COMPLEX + 4: | |
| 5050 case BRACE_COMPLEX + 5: | |
| 5051 case BRACE_COMPLEX + 6: | |
| 5052 case BRACE_COMPLEX + 7: | |
| 5053 case BRACE_COMPLEX + 8: | |
| 5054 case BRACE_COMPLEX + 9: | |
| 5055 { | |
| 5056 no = op - BRACE_COMPLEX; | |
| 5057 ++brace_count[no]; | |
| 5058 | |
| 5059 /* If not matched enough times yet, try one more */ | |
| 5060 if (brace_count[no] <= (brace_min[no] <= brace_max[no] | |
| 180 | 5061 ? brace_min[no] : brace_max[no])) |
| 7 | 5062 { |
| 270 | 5063 rp = regstack_push(RS_BRCPLX_MORE, scan); |
| 180 | 5064 if (rp == NULL) |
| 5065 status = RA_FAIL; | |
| 5066 else | |
| 5067 { | |
| 5068 rp->rs_no = no; | |
| 233 | 5069 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5070 next = OPERAND(scan); |
| 5071 /* We continue and handle the result when done. */ | |
| 5072 } | |
| 5073 break; | |
| 7 | 5074 } |
| 5075 | |
| 5076 /* If matched enough times, may try matching some more */ | |
| 5077 if (brace_min[no] <= brace_max[no]) | |
| 5078 { | |
| 5079 /* Range is the normal way around, use longest match */ | |
| 5080 if (brace_count[no] <= brace_max[no]) | |
| 5081 { | |
| 270 | 5082 rp = regstack_push(RS_BRCPLX_LONG, scan); |
| 180 | 5083 if (rp == NULL) |
| 5084 status = RA_FAIL; | |
| 5085 else | |
| 5086 { | |
| 5087 rp->rs_no = no; | |
| 233 | 5088 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5089 next = OPERAND(scan); |
| 5090 /* We continue and handle the result when done. */ | |
| 5091 } | |
| 7 | 5092 } |
| 5093 } | |
| 5094 else | |
| 5095 { | |
| 5096 /* Range is backwards, use shortest match first */ | |
| 5097 if (brace_count[no] <= brace_min[no]) | |
| 5098 { | |
| 270 | 5099 rp = regstack_push(RS_BRCPLX_SHORT, scan); |
| 180 | 5100 if (rp == NULL) |
| 5101 status = RA_FAIL; | |
| 5102 else | |
| 5103 { | |
| 233 | 5104 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5105 /* We continue and handle the result when done. */ |
| 5106 } | |
| 7 | 5107 } |
| 5108 } | |
| 5109 } | |
| 5110 break; | |
| 5111 | |
| 5112 case BRACE_SIMPLE: | |
| 5113 case STAR: | |
| 5114 case PLUS: | |
| 5115 { | |
| 180 | 5116 regstar_T rst; |
| 7 | 5117 |
| 5118 /* | |
| 5119 * Lookahead to avoid useless match attempts when we know | |
| 5120 * what character comes next. | |
| 5121 */ | |
| 5122 if (OP(next) == EXACTLY) | |
| 5123 { | |
| 180 | 5124 rst.nextb = *OPERAND(next); |
| 7 | 5125 if (ireg_ic) |
| 5126 { | |
| 1347 | 5127 if (MB_ISUPPER(rst.nextb)) |
| 5128 rst.nextb_ic = MB_TOLOWER(rst.nextb); | |
| 7 | 5129 else |
| 1347 | 5130 rst.nextb_ic = MB_TOUPPER(rst.nextb); |
| 7 | 5131 } |
| 5132 else | |
| 180 | 5133 rst.nextb_ic = rst.nextb; |
| 7 | 5134 } |
| 5135 else | |
| 5136 { | |
| 180 | 5137 rst.nextb = NUL; |
| 5138 rst.nextb_ic = NUL; | |
| 7 | 5139 } |
| 5140 if (op != BRACE_SIMPLE) | |
| 5141 { | |
| 180 | 5142 rst.minval = (op == STAR) ? 0 : 1; |
| 5143 rst.maxval = MAX_LIMIT; | |
| 7 | 5144 } |
| 5145 else | |
| 5146 { | |
| 180 | 5147 rst.minval = bl_minval; |
| 5148 rst.maxval = bl_maxval; | |
| 7 | 5149 } |
| 5150 | |
| 5151 /* | |
| 5152 * When maxval > minval, try matching as much as possible, up | |
| 5153 * to maxval. When maxval < minval, try matching at least the | |
| 5154 * minimal number (since the range is backwards, that's also | |
| 5155 * maxval!). | |
| 5156 */ | |
| 180 | 5157 rst.count = regrepeat(OPERAND(scan), rst.maxval); |
| 7 | 5158 if (got_int) |
| 180 | 5159 { |
| 5160 status = RA_FAIL; | |
| 5161 break; | |
| 5162 } | |
| 5163 if (rst.minval <= rst.maxval | |
| 5164 ? rst.count >= rst.minval : rst.count >= rst.maxval) | |
| 7 | 5165 { |
| 180 | 5166 /* It could match. Prepare for trying to match what |
| 5167 * follows. The code is below. Parameters are stored in | |
| 5168 * a regstar_T on the regstack. */ | |
| 212 | 5169 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5170 { |
| 5171 EMSG(_(e_maxmempat)); | |
| 5172 status = RA_FAIL; | |
| 5173 } | |
| 5174 else if (ga_grow(®stack, sizeof(regstar_T)) == FAIL) | |
| 180 | 5175 status = RA_FAIL; |
| 5176 else | |
| 7 | 5177 { |
| 180 | 5178 regstack.ga_len += sizeof(regstar_T); |
| 270 | 5179 rp = regstack_push(rst.minval <= rst.maxval |
| 233 | 5180 ? RS_STAR_LONG : RS_STAR_SHORT, scan); |
| 180 | 5181 if (rp == NULL) |
| 5182 status = RA_FAIL; | |
| 5183 else | |
| 7 | 5184 { |
| 180 | 5185 *(((regstar_T *)rp) - 1) = rst; |
| 5186 status = RA_BREAK; /* skip the restore bits */ | |
| 7 | 5187 } |
| 5188 } | |
| 5189 } | |
| 5190 else | |
| 180 | 5191 status = RA_NOMATCH; |
| 5192 | |
| 7 | 5193 } |
| 180 | 5194 break; |
| 7 | 5195 |
| 5196 case NOMATCH: | |
| 5197 case MATCH: | |
| 5198 case SUBPAT: | |
| 270 | 5199 rp = regstack_push(RS_NOMATCH, scan); |
| 180 | 5200 if (rp == NULL) |
| 5201 status = RA_FAIL; | |
| 5202 else | |
| 7 | 5203 { |
| 180 | 5204 rp->rs_no = op; |
| 233 | 5205 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5206 next = OPERAND(scan); |
| 5207 /* We continue and handle the result when done. */ | |
| 7 | 5208 } |
| 5209 break; | |
| 5210 | |
| 5211 case BEHIND: | |
| 5212 case NOBEHIND: | |
| 180 | 5213 /* Need a bit of room to store extra positions. */ |
| 212 | 5214 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5215 { |
| 5216 EMSG(_(e_maxmempat)); | |
| 5217 status = RA_FAIL; | |
| 5218 } | |
| 5219 else if (ga_grow(®stack, sizeof(regbehind_T)) == FAIL) | |
| 180 | 5220 status = RA_FAIL; |
| 5221 else | |
| 7 | 5222 { |
| 180 | 5223 regstack.ga_len += sizeof(regbehind_T); |
| 270 | 5224 rp = regstack_push(RS_BEHIND1, scan); |
| 180 | 5225 if (rp == NULL) |
| 5226 status = RA_FAIL; | |
| 5227 else | |
| 7 | 5228 { |
| 1579 | 5229 /* Need to save the subexpr to be able to restore them |
| 5230 * when there is a match but we don't use it. */ | |
| 5231 save_subexpr(((regbehind_T *)rp) - 1); | |
| 5232 | |
| 180 | 5233 rp->rs_no = op; |
| 233 | 5234 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5235 /* First try if what follows matches. If it does then we |
| 5236 * check the behind match by looping. */ | |
| 7 | 5237 } |
| 5238 } | |
| 180 | 5239 break; |
| 7 | 5240 |
| 5241 case BHPOS: | |
| 5242 if (REG_MULTI) | |
| 5243 { | |
| 5244 if (behind_pos.rs_u.pos.col != (colnr_T)(reginput - regline) | |
| 5245 || behind_pos.rs_u.pos.lnum != reglnum) | |
| 180 | 5246 status = RA_NOMATCH; |
| 7 | 5247 } |
| 5248 else if (behind_pos.rs_u.ptr != reginput) | |
| 180 | 5249 status = RA_NOMATCH; |
| 7 | 5250 break; |
| 5251 | |
| 5252 case NEWL: | |
| 1018 | 5253 if ((c != NUL || !REG_MULTI || reglnum > reg_maxline |
| 5254 || reg_line_lbr) && (c != '\n' || !reg_line_lbr)) | |
| 180 | 5255 status = RA_NOMATCH; |
| 5256 else if (reg_line_lbr) | |
| 7 | 5257 ADVANCE_REGINPUT(); |
| 5258 else | |
| 5259 reg_nextline(); | |
| 5260 break; | |
| 5261 | |
| 5262 case END: | |
| 180 | 5263 status = RA_MATCH; /* Success! */ |
| 5264 break; | |
| 7 | 5265 |
| 5266 default: | |
| 5267 EMSG(_(e_re_corr)); | |
| 5268 #ifdef DEBUG | |
| 5269 printf("Illegal op code %d\n", op); | |
| 5270 #endif | |
| 180 | 5271 status = RA_FAIL; |
| 5272 break; | |
| 7 | 5273 } |
| 5274 } | |
| 5275 | |
| 180 | 5276 /* If we can't continue sequentially, break the inner loop. */ |
| 5277 if (status != RA_CONT) | |
| 5278 break; | |
| 5279 | |
| 5280 /* Continue in inner loop, advance to next item. */ | |
| 7 | 5281 scan = next; |
| 180 | 5282 |
| 5283 } /* end of inner loop */ | |
| 7 | 5284 |
| 5285 /* | |
| 180 | 5286 * If there is something on the regstack execute the code for the state. |
| 233 | 5287 * If the state is popped then loop and use the older state. |
| 7 | 5288 */ |
| 180 | 5289 while (regstack.ga_len > 0 && status != RA_FAIL) |
| 5290 { | |
| 5291 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; | |
| 5292 switch (rp->rs_state) | |
| 5293 { | |
| 5294 case RS_NOPEN: | |
| 5295 /* Result is passed on as-is, simply pop the state. */ | |
| 270 | 5296 regstack_pop(&scan); |
| 180 | 5297 break; |
| 5298 | |
| 5299 case RS_MOPEN: | |
| 5300 /* Pop the state. Restore pointers when there is no match. */ | |
| 5301 if (status == RA_NOMATCH) | |
| 5302 restore_se(&rp->rs_un.sesave, ®_startpos[rp->rs_no], | |
| 5303 ®_startp[rp->rs_no]); | |
| 270 | 5304 regstack_pop(&scan); |
| 180 | 5305 break; |
| 5306 | |
| 5307 #ifdef FEAT_SYN_HL | |
| 5308 case RS_ZOPEN: | |
| 5309 /* Pop the state. Restore pointers when there is no match. */ | |
| 5310 if (status == RA_NOMATCH) | |
| 5311 restore_se(&rp->rs_un.sesave, ®_startzpos[rp->rs_no], | |
| 5312 ®_startzp[rp->rs_no]); | |
| 270 | 5313 regstack_pop(&scan); |
| 180 | 5314 break; |
| 5315 #endif | |
| 5316 | |
| 5317 case RS_MCLOSE: | |
| 5318 /* Pop the state. Restore pointers when there is no match. */ | |
| 5319 if (status == RA_NOMATCH) | |
| 5320 restore_se(&rp->rs_un.sesave, ®_endpos[rp->rs_no], | |
| 5321 ®_endp[rp->rs_no]); | |
| 270 | 5322 regstack_pop(&scan); |
| 180 | 5323 break; |
| 5324 | |
| 5325 #ifdef FEAT_SYN_HL | |
| 5326 case RS_ZCLOSE: | |
| 5327 /* Pop the state. Restore pointers when there is no match. */ | |
| 5328 if (status == RA_NOMATCH) | |
| 5329 restore_se(&rp->rs_un.sesave, ®_endzpos[rp->rs_no], | |
| 5330 ®_endzp[rp->rs_no]); | |
| 270 | 5331 regstack_pop(&scan); |
| 180 | 5332 break; |
| 5333 #endif | |
| 5334 | |
| 5335 case RS_BRANCH: | |
| 5336 if (status == RA_MATCH) | |
| 5337 /* this branch matched, use it */ | |
| 270 | 5338 regstack_pop(&scan); |
| 180 | 5339 else |
| 5340 { | |
| 5341 if (status != RA_BREAK) | |
| 5342 { | |
| 5343 /* After a non-matching branch: try next one. */ | |
| 233 | 5344 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5345 scan = rp->rs_scan; |
| 5346 } | |
| 5347 if (scan == NULL || OP(scan) != BRANCH) | |
| 5348 { | |
| 5349 /* no more branches, didn't find a match */ | |
| 5350 status = RA_NOMATCH; | |
| 270 | 5351 regstack_pop(&scan); |
| 180 | 5352 } |
| 5353 else | |
| 5354 { | |
| 5355 /* Prepare to try a branch. */ | |
| 5356 rp->rs_scan = regnext(scan); | |
| 233 | 5357 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5358 scan = OPERAND(scan); |
| 5359 } | |
| 5360 } | |
| 5361 break; | |
| 5362 | |
| 5363 case RS_BRCPLX_MORE: | |
| 5364 /* Pop the state. Restore pointers when there is no match. */ | |
| 5365 if (status == RA_NOMATCH) | |
| 5366 { | |
| 233 | 5367 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5368 --brace_count[rp->rs_no]; /* decrement match count */ |
| 5369 } | |
| 270 | 5370 regstack_pop(&scan); |
| 180 | 5371 break; |
| 5372 | |
| 5373 case RS_BRCPLX_LONG: | |
| 5374 /* Pop the state. Restore pointers when there is no match. */ | |
| 5375 if (status == RA_NOMATCH) | |
| 5376 { | |
| 5377 /* There was no match, but we did find enough matches. */ | |
| 233 | 5378 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5379 --brace_count[rp->rs_no]; |
| 5380 /* continue with the items after "\{}" */ | |
| 5381 status = RA_CONT; | |
| 5382 } | |
| 270 | 5383 regstack_pop(&scan); |
| 180 | 5384 if (status == RA_CONT) |
| 5385 scan = regnext(scan); | |
| 5386 break; | |
| 5387 | |
| 5388 case RS_BRCPLX_SHORT: | |
| 5389 /* Pop the state. Restore pointers when there is no match. */ | |
| 5390 if (status == RA_NOMATCH) | |
| 5391 /* There was no match, try to match one more item. */ | |
| 233 | 5392 reg_restore(&rp->rs_un.regsave, &backpos); |
| 270 | 5393 regstack_pop(&scan); |
| 180 | 5394 if (status == RA_NOMATCH) |
| 5395 { | |
| 5396 scan = OPERAND(scan); | |
| 5397 status = RA_CONT; | |
| 5398 } | |
| 5399 break; | |
| 5400 | |
| 5401 case RS_NOMATCH: | |
| 5402 /* Pop the state. If the operand matches for NOMATCH or | |
| 5403 * doesn't match for MATCH/SUBPAT, we fail. Otherwise backup, | |
| 5404 * except for SUBPAT, and continue with the next item. */ | |
| 5405 if (status == (rp->rs_no == NOMATCH ? RA_MATCH : RA_NOMATCH)) | |
| 5406 status = RA_NOMATCH; | |
| 5407 else | |
| 5408 { | |
| 5409 status = RA_CONT; | |
| 233 | 5410 if (rp->rs_no != SUBPAT) /* zero-width */ |
| 5411 reg_restore(&rp->rs_un.regsave, &backpos); | |
| 180 | 5412 } |
| 270 | 5413 regstack_pop(&scan); |
| 180 | 5414 if (status == RA_CONT) |
| 5415 scan = regnext(scan); | |
| 5416 break; | |
| 5417 | |
| 5418 case RS_BEHIND1: | |
| 5419 if (status == RA_NOMATCH) | |
| 5420 { | |
| 270 | 5421 regstack_pop(&scan); |
| 180 | 5422 regstack.ga_len -= sizeof(regbehind_T); |
| 5423 } | |
| 5424 else | |
| 5425 { | |
| 5426 /* The stuff after BEHIND/NOBEHIND matches. Now try if | |
| 5427 * the behind part does (not) match before the current | |
| 5428 * position in the input. This must be done at every | |
| 5429 * position in the input and checking if the match ends at | |
| 5430 * the current position. */ | |
| 5431 | |
| 5432 /* save the position after the found match for next */ | |
| 233 | 5433 reg_save(&(((regbehind_T *)rp) - 1)->save_after, &backpos); |
| 180 | 5434 |
| 5435 /* start looking for a match with operand at the current | |
| 1209 | 5436 * position. Go back one character until we find the |
| 180 | 5437 * result, hitting the start of the line or the previous |
| 5438 * line (for multi-line matching). | |
| 5439 * Set behind_pos to where the match should end, BHPOS | |
| 5440 * will match it. Save the current value. */ | |
| 5441 (((regbehind_T *)rp) - 1)->save_behind = behind_pos; | |
| 5442 behind_pos = rp->rs_un.regsave; | |
| 5443 | |
| 5444 rp->rs_state = RS_BEHIND2; | |
| 5445 | |
| 233 | 5446 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5447 scan = OPERAND(rp->rs_scan); |
| 5448 } | |
| 5449 break; | |
| 5450 | |
| 5451 case RS_BEHIND2: | |
| 5452 /* | |
| 5453 * Looping for BEHIND / NOBEHIND match. | |
| 5454 */ | |
| 5455 if (status == RA_MATCH && reg_save_equal(&behind_pos)) | |
| 5456 { | |
| 5457 /* found a match that ends where "next" started */ | |
| 5458 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5459 if (rp->rs_no == BEHIND) | |
| 233 | 5460 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5461 &backpos); | |
| 180 | 5462 else |
| 1579 | 5463 { |
| 5464 /* But we didn't want a match. Need to restore the | |
| 5465 * subexpr, because what follows matched, so they have | |
| 5466 * been set. */ | |
| 180 | 5467 status = RA_NOMATCH; |
| 1579 | 5468 restore_subexpr(((regbehind_T *)rp) - 1); |
| 5469 } | |
| 270 | 5470 regstack_pop(&scan); |
| 180 | 5471 regstack.ga_len -= sizeof(regbehind_T); |
| 5472 } | |
| 5473 else | |
| 5474 { | |
| 1579 | 5475 /* No match or a match that doesn't end where we want it: Go |
| 5476 * back one character. May go to previous line once. */ | |
| 180 | 5477 no = OK; |
| 5478 if (REG_MULTI) | |
| 5479 { | |
| 5480 if (rp->rs_un.regsave.rs_u.pos.col == 0) | |
| 5481 { | |
| 5482 if (rp->rs_un.regsave.rs_u.pos.lnum | |
| 5483 < behind_pos.rs_u.pos.lnum | |
| 5484 || reg_getline( | |
| 5485 --rp->rs_un.regsave.rs_u.pos.lnum) | |
| 5486 == NULL) | |
| 5487 no = FAIL; | |
| 5488 else | |
| 5489 { | |
| 233 | 5490 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5491 rp->rs_un.regsave.rs_u.pos.col = |
| 5492 (colnr_T)STRLEN(regline); | |
| 5493 } | |
| 5494 } | |
| 5495 else | |
| 4176 | 5496 #ifdef FEAT_MBYTE |
| 5497 if (has_mbyte) | |
| 5498 rp->rs_un.regsave.rs_u.pos.col -= | |
| 5499 (*mb_head_off)(regline, regline | |
| 5500 + rp->rs_un.regsave.rs_u.pos.col - 1) + 1; | |
| 5501 else | |
| 5502 #endif | |
| 180 | 5503 --rp->rs_un.regsave.rs_u.pos.col; |
| 5504 } | |
| 5505 else | |
| 5506 { | |
| 5507 if (rp->rs_un.regsave.rs_u.ptr == regline) | |
| 5508 no = FAIL; | |
| 5509 else | |
| 5510 --rp->rs_un.regsave.rs_u.ptr; | |
| 5511 } | |
| 5512 if (no == OK) | |
| 5513 { | |
| 5514 /* Advanced, prepare for finding match again. */ | |
| 233 | 5515 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5516 scan = OPERAND(rp->rs_scan); |
| 1579 | 5517 if (status == RA_MATCH) |
| 5518 { | |
| 5519 /* We did match, so subexpr may have been changed, | |
| 5520 * need to restore them for the next try. */ | |
| 5521 status = RA_NOMATCH; | |
| 5522 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5523 } | |
| 180 | 5524 } |
| 5525 else | |
| 5526 { | |
| 5527 /* Can't advance. For NOBEHIND that's a match. */ | |
| 5528 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5529 if (rp->rs_no == NOBEHIND) | |
| 5530 { | |
| 233 | 5531 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5532 &backpos); | |
| 180 | 5533 status = RA_MATCH; |
| 5534 } | |
| 5535 else | |
| 1579 | 5536 { |
| 5537 /* We do want a proper match. Need to restore the | |
| 5538 * subexpr if we had a match, because they may have | |
| 5539 * been set. */ | |
| 5540 if (status == RA_MATCH) | |
| 5541 { | |
| 5542 status = RA_NOMATCH; | |
| 5543 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5544 } | |
| 5545 } | |
| 270 | 5546 regstack_pop(&scan); |
| 180 | 5547 regstack.ga_len -= sizeof(regbehind_T); |
| 5548 } | |
| 5549 } | |
| 5550 break; | |
| 5551 | |
| 5552 case RS_STAR_LONG: | |
| 5553 case RS_STAR_SHORT: | |
| 5554 { | |
| 5555 regstar_T *rst = ((regstar_T *)rp) - 1; | |
| 5556 | |
| 5557 if (status == RA_MATCH) | |
| 5558 { | |
| 270 | 5559 regstack_pop(&scan); |
| 180 | 5560 regstack.ga_len -= sizeof(regstar_T); |
| 5561 break; | |
| 5562 } | |
| 5563 | |
| 5564 /* Tried once already, restore input pointers. */ | |
| 5565 if (status != RA_BREAK) | |
| 233 | 5566 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5567 |
| 5568 /* Repeat until we found a position where it could match. */ | |
| 5569 for (;;) | |
| 5570 { | |
| 5571 if (status != RA_BREAK) | |
| 5572 { | |
| 5573 /* Tried first position already, advance. */ | |
| 5574 if (rp->rs_state == RS_STAR_LONG) | |
| 5575 { | |
| 685 | 5576 /* Trying for longest match, but couldn't or |
| 5577 * didn't match -- back up one char. */ | |
| 180 | 5578 if (--rst->count < rst->minval) |
| 5579 break; | |
| 5580 if (reginput == regline) | |
| 5581 { | |
| 5582 /* backup to last char of previous line */ | |
| 5583 --reglnum; | |
| 5584 regline = reg_getline(reglnum); | |
| 5585 /* Just in case regrepeat() didn't count | |
| 5586 * right. */ | |
| 5587 if (regline == NULL) | |
| 5588 break; | |
| 5589 reginput = regline + STRLEN(regline); | |
| 5590 fast_breakcheck(); | |
| 5591 } | |
| 5592 else | |
| 5593 mb_ptr_back(regline, reginput); | |
| 5594 } | |
| 5595 else | |
| 5596 { | |
| 5597 /* Range is backwards, use shortest match first. | |
| 5598 * Careful: maxval and minval are exchanged! | |
| 5599 * Couldn't or didn't match: try advancing one | |
| 5600 * char. */ | |
| 5601 if (rst->count == rst->minval | |
| 5602 || regrepeat(OPERAND(rp->rs_scan), 1L) == 0) | |
| 5603 break; | |
| 5604 ++rst->count; | |
| 5605 } | |
| 5606 if (got_int) | |
| 5607 break; | |
| 5608 } | |
| 5609 else | |
| 5610 status = RA_NOMATCH; | |
| 5611 | |
| 5612 /* If it could match, try it. */ | |
| 5613 if (rst->nextb == NUL || *reginput == rst->nextb | |
| 5614 || *reginput == rst->nextb_ic) | |
| 5615 { | |
| 233 | 5616 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5617 scan = regnext(rp->rs_scan); |
| 5618 status = RA_CONT; | |
| 5619 break; | |
| 5620 } | |
| 5621 } | |
| 5622 if (status != RA_CONT) | |
| 5623 { | |
| 5624 /* Failed. */ | |
| 270 | 5625 regstack_pop(&scan); |
| 180 | 5626 regstack.ga_len -= sizeof(regstar_T); |
| 5627 status = RA_NOMATCH; | |
| 5628 } | |
| 5629 } | |
| 5630 break; | |
| 5631 } | |
| 5632 | |
| 685 | 5633 /* If we want to continue the inner loop or didn't pop a state |
| 5634 * continue matching loop */ | |
| 180 | 5635 if (status == RA_CONT || rp == (regitem_T *) |
| 5636 ((char *)regstack.ga_data + regstack.ga_len) - 1) | |
| 5637 break; | |
| 5638 } | |
| 5639 | |
| 189 | 5640 /* May need to continue with the inner loop, starting at "scan". */ |
| 180 | 5641 if (status == RA_CONT) |
| 5642 continue; | |
| 5643 | |
| 5644 /* | |
| 5645 * If the regstack is empty or something failed we are done. | |
| 5646 */ | |
| 5647 if (regstack.ga_len == 0 || status == RA_FAIL) | |
| 5648 { | |
| 5649 if (scan == NULL) | |
| 5650 { | |
| 5651 /* | |
| 5652 * We get here only if there's trouble -- normally "case END" is | |
| 5653 * the terminating point. | |
| 5654 */ | |
| 5655 EMSG(_(e_re_corr)); | |
| 7 | 5656 #ifdef DEBUG |
| 180 | 5657 printf("Premature EOL\n"); |
| 7 | 5658 #endif |
| 180 | 5659 } |
| 189 | 5660 if (status == RA_FAIL) |
| 5661 got_int = TRUE; | |
| 180 | 5662 return (status == RA_MATCH); |
| 5663 } | |
| 5664 | |
| 5665 } /* End of loop until the regstack is empty. */ | |
| 5666 | |
| 5667 /* NOTREACHED */ | |
| 5668 } | |
| 5669 | |
| 5670 /* | |
| 5671 * Push an item onto the regstack. | |
| 5672 * Returns pointer to new item. Returns NULL when out of memory. | |
| 5673 */ | |
| 5674 static regitem_T * | |
| 270 | 5675 regstack_push(state, scan) |
| 180 | 5676 regstate_T state; |
| 5677 char_u *scan; | |
| 5678 { | |
| 5679 regitem_T *rp; | |
| 5680 | |
| 270 | 5681 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5682 { |
| 5683 EMSG(_(e_maxmempat)); | |
| 5684 return NULL; | |
| 5685 } | |
| 270 | 5686 if (ga_grow(®stack, sizeof(regitem_T)) == FAIL) |
| 180 | 5687 return NULL; |
| 5688 | |
| 270 | 5689 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len); |
| 180 | 5690 rp->rs_state = state; |
| 5691 rp->rs_scan = scan; | |
| 5692 | |
| 270 | 5693 regstack.ga_len += sizeof(regitem_T); |
| 180 | 5694 return rp; |
| 5695 } | |
| 5696 | |
| 5697 /* | |
| 5698 * Pop an item from the regstack. | |
| 5699 */ | |
| 5700 static void | |
| 270 | 5701 regstack_pop(scan) |
| 180 | 5702 char_u **scan; |
| 5703 { | |
| 5704 regitem_T *rp; | |
| 5705 | |
| 270 | 5706 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; |
| 180 | 5707 *scan = rp->rs_scan; |
| 5708 | |
| 270 | 5709 regstack.ga_len -= sizeof(regitem_T); |
| 7 | 5710 } |
| 5711 | |
| 5712 /* | |
| 5713 * regrepeat - repeatedly match something simple, return how many. | |
| 5714 * Advances reginput (and reglnum) to just after the matched chars. | |
| 5715 */ | |
| 5716 static int | |
| 5717 regrepeat(p, maxcount) | |
| 5718 char_u *p; | |
| 5719 long maxcount; /* maximum number of matches allowed */ | |
| 5720 { | |
| 5721 long count = 0; | |
| 5722 char_u *scan; | |
| 5723 char_u *opnd; | |
| 5724 int mask; | |
| 5725 int testval = 0; | |
| 5726 | |
| 5727 scan = reginput; /* Make local copy of reginput for speed. */ | |
| 5728 opnd = OPERAND(p); | |
| 5729 switch (OP(p)) | |
| 5730 { | |
| 5731 case ANY: | |
| 5732 case ANY + ADD_NL: | |
| 5733 while (count < maxcount) | |
| 5734 { | |
| 5735 /* Matching anything means we continue until end-of-line (or | |
| 5736 * end-of-file for ANY + ADD_NL), only limited by maxcount. */ | |
| 5737 while (*scan != NUL && count < maxcount) | |
| 5738 { | |
| 5739 ++count; | |
| 39 | 5740 mb_ptr_adv(scan); |
| 7 | 5741 } |
| 1018 | 5742 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5743 || reg_line_lbr || count == maxcount) | |
| 7 | 5744 break; |
| 5745 ++count; /* count the line-break */ | |
| 5746 reg_nextline(); | |
| 5747 scan = reginput; | |
| 5748 if (got_int) | |
| 5749 break; | |
| 5750 } | |
| 5751 break; | |
| 5752 | |
| 5753 case IDENT: | |
| 5754 case IDENT + ADD_NL: | |
| 5755 testval = TRUE; | |
| 5756 /*FALLTHROUGH*/ | |
| 5757 case SIDENT: | |
| 5758 case SIDENT + ADD_NL: | |
| 5759 while (count < maxcount) | |
| 5760 { | |
| 4466 | 5761 if (vim_isIDc(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5762 { |
| 39 | 5763 mb_ptr_adv(scan); |
| 7 | 5764 } |
| 5765 else if (*scan == NUL) | |
| 5766 { | |
| 1018 | 5767 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5768 || reg_line_lbr) | |
| 7 | 5769 break; |
| 5770 reg_nextline(); | |
| 5771 scan = reginput; | |
| 5772 if (got_int) | |
| 5773 break; | |
| 5774 } | |
| 5775 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5776 ++scan; | |
| 5777 else | |
| 5778 break; | |
| 5779 ++count; | |
| 5780 } | |
| 5781 break; | |
| 5782 | |
| 5783 case KWORD: | |
| 5784 case KWORD + ADD_NL: | |
| 5785 testval = TRUE; | |
| 5786 /*FALLTHROUGH*/ | |
| 5787 case SKWORD: | |
| 5788 case SKWORD + ADD_NL: | |
| 5789 while (count < maxcount) | |
| 5790 { | |
| 4069 | 5791 if (vim_iswordp_buf(scan, reg_buf) |
| 5792 && (testval || !VIM_ISDIGIT(*scan))) | |
| 7 | 5793 { |
| 39 | 5794 mb_ptr_adv(scan); |
| 7 | 5795 } |
| 5796 else if (*scan == NUL) | |
| 5797 { | |
| 1018 | 5798 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5799 || reg_line_lbr) | |
| 7 | 5800 break; |
| 5801 reg_nextline(); | |
| 5802 scan = reginput; | |
| 5803 if (got_int) | |
| 5804 break; | |
| 5805 } | |
| 5806 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5807 ++scan; | |
| 5808 else | |
| 5809 break; | |
| 5810 ++count; | |
| 5811 } | |
| 5812 break; | |
| 5813 | |
| 5814 case FNAME: | |
| 5815 case FNAME + ADD_NL: | |
| 5816 testval = TRUE; | |
| 5817 /*FALLTHROUGH*/ | |
| 5818 case SFNAME: | |
| 5819 case SFNAME + ADD_NL: | |
| 5820 while (count < maxcount) | |
| 5821 { | |
| 4466 | 5822 if (vim_isfilec(PTR2CHAR(scan)) && (testval || !VIM_ISDIGIT(*scan))) |
| 7 | 5823 { |
| 39 | 5824 mb_ptr_adv(scan); |
| 7 | 5825 } |
| 5826 else if (*scan == NUL) | |
| 5827 { | |
| 1018 | 5828 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5829 || reg_line_lbr) | |
| 7 | 5830 break; |
| 5831 reg_nextline(); | |
| 5832 scan = reginput; | |
| 5833 if (got_int) | |
| 5834 break; | |
| 5835 } | |
| 5836 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5837 ++scan; | |
| 5838 else | |
| 5839 break; | |
| 5840 ++count; | |
| 5841 } | |
| 5842 break; | |
| 5843 | |
| 5844 case PRINT: | |
| 5845 case PRINT + ADD_NL: | |
| 5846 testval = TRUE; | |
| 5847 /*FALLTHROUGH*/ | |
| 5848 case SPRINT: | |
| 5849 case SPRINT + ADD_NL: | |
| 5850 while (count < maxcount) | |
| 5851 { | |
| 5852 if (*scan == NUL) | |
| 5853 { | |
| 1018 | 5854 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5855 || reg_line_lbr) | |
| 7 | 5856 break; |
| 5857 reg_nextline(); | |
| 5858 scan = reginput; | |
| 5859 if (got_int) | |
| 5860 break; | |
| 5861 } | |
| 5862 else if (ptr2cells(scan) == 1 && (testval || !VIM_ISDIGIT(*scan))) | |
| 5863 { | |
| 39 | 5864 mb_ptr_adv(scan); |
| 7 | 5865 } |
| 5866 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5867 ++scan; | |
| 5868 else | |
| 5869 break; | |
| 5870 ++count; | |
| 5871 } | |
| 5872 break; | |
| 5873 | |
| 5874 case WHITE: | |
| 5875 case WHITE + ADD_NL: | |
| 5876 testval = mask = RI_WHITE; | |
| 5877 do_class: | |
| 5878 while (count < maxcount) | |
| 5879 { | |
| 5880 #ifdef FEAT_MBYTE | |
| 5881 int l; | |
| 5882 #endif | |
| 5883 if (*scan == NUL) | |
| 5884 { | |
| 1018 | 5885 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5886 || reg_line_lbr) | |
| 7 | 5887 break; |
| 5888 reg_nextline(); | |
| 5889 scan = reginput; | |
| 5890 if (got_int) | |
| 5891 break; | |
| 5892 } | |
| 5893 #ifdef FEAT_MBYTE | |
| 474 | 5894 else if (has_mbyte && (l = (*mb_ptr2len)(scan)) > 1) |
| 7 | 5895 { |
| 5896 if (testval != 0) | |
| 5897 break; | |
| 5898 scan += l; | |
| 5899 } | |
| 5900 #endif | |
| 5901 else if ((class_tab[*scan] & mask) == testval) | |
| 5902 ++scan; | |
| 5903 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5904 ++scan; | |
| 5905 else | |
| 5906 break; | |
| 5907 ++count; | |
| 5908 } | |
| 5909 break; | |
| 5910 | |
| 5911 case NWHITE: | |
| 5912 case NWHITE + ADD_NL: | |
| 5913 mask = RI_WHITE; | |
| 5914 goto do_class; | |
| 5915 case DIGIT: | |
| 5916 case DIGIT + ADD_NL: | |
| 5917 testval = mask = RI_DIGIT; | |
| 5918 goto do_class; | |
| 5919 case NDIGIT: | |
| 5920 case NDIGIT + ADD_NL: | |
| 5921 mask = RI_DIGIT; | |
| 5922 goto do_class; | |
| 5923 case HEX: | |
| 5924 case HEX + ADD_NL: | |
| 5925 testval = mask = RI_HEX; | |
| 5926 goto do_class; | |
| 5927 case NHEX: | |
| 5928 case NHEX + ADD_NL: | |
| 5929 mask = RI_HEX; | |
| 5930 goto do_class; | |
| 5931 case OCTAL: | |
| 5932 case OCTAL + ADD_NL: | |
| 5933 testval = mask = RI_OCTAL; | |
| 5934 goto do_class; | |
| 5935 case NOCTAL: | |
| 5936 case NOCTAL + ADD_NL: | |
| 5937 mask = RI_OCTAL; | |
| 5938 goto do_class; | |
| 5939 case WORD: | |
| 5940 case WORD + ADD_NL: | |
| 5941 testval = mask = RI_WORD; | |
| 5942 goto do_class; | |
| 5943 case NWORD: | |
| 5944 case NWORD + ADD_NL: | |
| 5945 mask = RI_WORD; | |
| 5946 goto do_class; | |
| 5947 case HEAD: | |
| 5948 case HEAD + ADD_NL: | |
| 5949 testval = mask = RI_HEAD; | |
| 5950 goto do_class; | |
| 5951 case NHEAD: | |
| 5952 case NHEAD + ADD_NL: | |
| 5953 mask = RI_HEAD; | |
| 5954 goto do_class; | |
| 5955 case ALPHA: | |
| 5956 case ALPHA + ADD_NL: | |
| 5957 testval = mask = RI_ALPHA; | |
| 5958 goto do_class; | |
| 5959 case NALPHA: | |
| 5960 case NALPHA + ADD_NL: | |
| 5961 mask = RI_ALPHA; | |
| 5962 goto do_class; | |
| 5963 case LOWER: | |
| 5964 case LOWER + ADD_NL: | |
| 5965 testval = mask = RI_LOWER; | |
| 5966 goto do_class; | |
| 5967 case NLOWER: | |
| 5968 case NLOWER + ADD_NL: | |
| 5969 mask = RI_LOWER; | |
| 5970 goto do_class; | |
| 5971 case UPPER: | |
| 5972 case UPPER + ADD_NL: | |
| 5973 testval = mask = RI_UPPER; | |
| 5974 goto do_class; | |
| 5975 case NUPPER: | |
| 5976 case NUPPER + ADD_NL: | |
| 5977 mask = RI_UPPER; | |
| 5978 goto do_class; | |
| 5979 | |
| 5980 case EXACTLY: | |
| 5981 { | |
| 5982 int cu, cl; | |
| 5983 | |
| 5984 /* This doesn't do a multi-byte character, because a MULTIBYTECODE | |
| 1347 | 5985 * would have been used for it. It does handle single-byte |
| 5986 * characters, such as latin1. */ | |
| 7 | 5987 if (ireg_ic) |
| 5988 { | |
| 1347 | 5989 cu = MB_TOUPPER(*opnd); |
| 5990 cl = MB_TOLOWER(*opnd); | |
| 7 | 5991 while (count < maxcount && (*scan == cu || *scan == cl)) |
| 5992 { | |
| 5993 count++; | |
| 5994 scan++; | |
| 5995 } | |
| 5996 } | |
| 5997 else | |
| 5998 { | |
| 5999 cu = *opnd; | |
| 6000 while (count < maxcount && *scan == cu) | |
| 6001 { | |
| 6002 count++; | |
| 6003 scan++; | |
| 6004 } | |
| 6005 } | |
| 6006 break; | |
| 6007 } | |
| 6008 | |
| 6009 #ifdef FEAT_MBYTE | |
| 6010 case MULTIBYTECODE: | |
| 6011 { | |
| 6012 int i, len, cf = 0; | |
| 6013 | |
| 6014 /* Safety check (just in case 'encoding' was changed since | |
| 6015 * compiling the program). */ | |
| 474 | 6016 if ((len = (*mb_ptr2len)(opnd)) > 1) |
| 7 | 6017 { |
| 6018 if (ireg_ic && enc_utf8) | |
| 6019 cf = utf_fold(utf_ptr2char(opnd)); | |
| 6020 while (count < maxcount) | |
| 6021 { | |
| 6022 for (i = 0; i < len; ++i) | |
| 6023 if (opnd[i] != scan[i]) | |
| 6024 break; | |
| 6025 if (i < len && (!ireg_ic || !enc_utf8 | |
| 6026 || utf_fold(utf_ptr2char(scan)) != cf)) | |
| 6027 break; | |
| 6028 scan += len; | |
| 6029 ++count; | |
| 6030 } | |
| 6031 } | |
| 6032 } | |
| 6033 break; | |
| 6034 #endif | |
| 6035 | |
| 6036 case ANYOF: | |
| 6037 case ANYOF + ADD_NL: | |
| 6038 testval = TRUE; | |
| 6039 /*FALLTHROUGH*/ | |
| 6040 | |
| 6041 case ANYBUT: | |
| 6042 case ANYBUT + ADD_NL: | |
| 6043 while (count < maxcount) | |
| 6044 { | |
| 6045 #ifdef FEAT_MBYTE | |
| 6046 int len; | |
| 6047 #endif | |
| 6048 if (*scan == NUL) | |
| 6049 { | |
| 1018 | 6050 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 6051 || reg_line_lbr) | |
| 7 | 6052 break; |
| 6053 reg_nextline(); | |
| 6054 scan = reginput; | |
| 6055 if (got_int) | |
| 6056 break; | |
| 6057 } | |
| 6058 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 6059 ++scan; | |
| 6060 #ifdef FEAT_MBYTE | |
| 474 | 6061 else if (has_mbyte && (len = (*mb_ptr2len)(scan)) > 1) |
| 7 | 6062 { |
| 6063 if ((cstrchr(opnd, (*mb_ptr2char)(scan)) == NULL) == testval) | |
| 6064 break; | |
| 6065 scan += len; | |
| 6066 } | |
| 6067 #endif | |
| 6068 else | |
| 6069 { | |
| 6070 if ((cstrchr(opnd, *scan) == NULL) == testval) | |
| 6071 break; | |
| 6072 ++scan; | |
| 6073 } | |
| 6074 ++count; | |
| 6075 } | |
| 6076 break; | |
| 6077 | |
| 6078 case NEWL: | |
| 6079 while (count < maxcount | |
| 1018 | 6080 && ((*scan == NUL && reglnum <= reg_maxline && !reg_line_lbr |
| 6081 && REG_MULTI) || (*scan == '\n' && reg_line_lbr))) | |
| 7 | 6082 { |
| 6083 count++; | |
| 6084 if (reg_line_lbr) | |
| 6085 ADVANCE_REGINPUT(); | |
| 6086 else | |
| 6087 reg_nextline(); | |
| 6088 scan = reginput; | |
| 6089 if (got_int) | |
| 6090 break; | |
| 6091 } | |
| 6092 break; | |
| 6093 | |
| 6094 default: /* Oh dear. Called inappropriately. */ | |
| 6095 EMSG(_(e_re_corr)); | |
| 6096 #ifdef DEBUG | |
| 6097 printf("Called regrepeat with op code %d\n", OP(p)); | |
| 6098 #endif | |
| 6099 break; | |
| 6100 } | |
| 6101 | |
| 6102 reginput = scan; | |
| 6103 | |
| 6104 return (int)count; | |
| 6105 } | |
| 6106 | |
| 6107 /* | |
| 6108 * regnext - dig the "next" pointer out of a node | |
| 2010 | 6109 * Returns NULL when calculating size, when there is no next item and when |
| 6110 * there is an error. | |
| 7 | 6111 */ |
| 6112 static char_u * | |
| 6113 regnext(p) | |
| 6114 char_u *p; | |
| 6115 { | |
| 6116 int offset; | |
| 6117 | |
| 2010 | 6118 if (p == JUST_CALC_SIZE || reg_toolong) |
| 7 | 6119 return NULL; |
| 6120 | |
| 6121 offset = NEXT(p); | |
| 6122 if (offset == 0) | |
| 6123 return NULL; | |
| 6124 | |
| 233 | 6125 if (OP(p) == BACK) |
| 7 | 6126 return p - offset; |
| 6127 else | |
| 6128 return p + offset; | |
| 6129 } | |
| 6130 | |
| 6131 /* | |
| 6132 * Check the regexp program for its magic number. | |
| 6133 * Return TRUE if it's wrong. | |
| 6134 */ | |
| 6135 static int | |
| 6136 prog_magic_wrong() | |
| 6137 { | |
| 4444 | 6138 regprog_T *prog; |
| 6139 | |
| 6140 prog = REG_MULTI ? reg_mmatch->regprog : reg_match->regprog; | |
| 6141 if (prog->engine == &nfa_regengine) | |
| 6142 /* For NFA matcher we don't check the magic */ | |
| 6143 return FALSE; | |
| 6144 | |
| 6145 if (UCHARAT(((bt_regprog_T *)prog)->program) != REGMAGIC) | |
| 7 | 6146 { |
| 6147 EMSG(_(e_re_corr)); | |
| 6148 return TRUE; | |
| 6149 } | |
| 6150 return FALSE; | |
| 6151 } | |
| 6152 | |
| 6153 /* | |
| 6154 * Cleanup the subexpressions, if this wasn't done yet. | |
| 6155 * This construction is used to clear the subexpressions only when they are | |
| 6156 * used (to increase speed). | |
| 6157 */ | |
| 6158 static void | |
| 6159 cleanup_subexpr() | |
| 6160 { | |
| 6161 if (need_clear_subexpr) | |
| 6162 { | |
| 6163 if (REG_MULTI) | |
| 6164 { | |
| 6165 /* Use 0xff to set lnum to -1 */ | |
| 6166 vim_memset(reg_startpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6167 vim_memset(reg_endpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6168 } | |
| 6169 else | |
| 6170 { | |
| 6171 vim_memset(reg_startp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6172 vim_memset(reg_endp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6173 } | |
| 6174 need_clear_subexpr = FALSE; | |
| 6175 } | |
| 6176 } | |
| 6177 | |
| 6178 #ifdef FEAT_SYN_HL | |
| 6179 static void | |
| 6180 cleanup_zsubexpr() | |
| 6181 { | |
| 6182 if (need_clear_zsubexpr) | |
| 6183 { | |
| 6184 if (REG_MULTI) | |
| 6185 { | |
| 6186 /* Use 0xff to set lnum to -1 */ | |
| 6187 vim_memset(reg_startzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6188 vim_memset(reg_endzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6189 } | |
| 6190 else | |
| 6191 { | |
| 6192 vim_memset(reg_startzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6193 vim_memset(reg_endzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6194 } | |
| 6195 need_clear_zsubexpr = FALSE; | |
| 6196 } | |
| 6197 } | |
| 6198 #endif | |
| 6199 | |
| 6200 /* | |
| 1579 | 6201 * Save the current subexpr to "bp", so that they can be restored |
| 6202 * later by restore_subexpr(). | |
| 6203 */ | |
| 6204 static void | |
| 6205 save_subexpr(bp) | |
| 6206 regbehind_T *bp; | |
| 6207 { | |
| 6208 int i; | |
| 6209 | |
| 1602 | 6210 /* When "need_clear_subexpr" is set we don't need to save the values, only |
| 6211 * remember that this flag needs to be set again when restoring. */ | |
| 6212 bp->save_need_clear_subexpr = need_clear_subexpr; | |
| 6213 if (!need_clear_subexpr) | |
| 1579 | 6214 { |
| 1602 | 6215 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6216 { |
| 1602 | 6217 if (REG_MULTI) |
| 6218 { | |
| 6219 bp->save_start[i].se_u.pos = reg_startpos[i]; | |
| 6220 bp->save_end[i].se_u.pos = reg_endpos[i]; | |
| 6221 } | |
| 6222 else | |
| 6223 { | |
| 6224 bp->save_start[i].se_u.ptr = reg_startp[i]; | |
| 6225 bp->save_end[i].se_u.ptr = reg_endp[i]; | |
| 6226 } | |
| 1579 | 6227 } |
| 6228 } | |
| 6229 } | |
| 6230 | |
| 6231 /* | |
| 6232 * Restore the subexpr from "bp". | |
| 6233 */ | |
| 6234 static void | |
| 6235 restore_subexpr(bp) | |
| 6236 regbehind_T *bp; | |
| 6237 { | |
| 6238 int i; | |
| 6239 | |
| 1602 | 6240 /* Only need to restore saved values when they are not to be cleared. */ |
| 6241 need_clear_subexpr = bp->save_need_clear_subexpr; | |
| 6242 if (!need_clear_subexpr) | |
| 1579 | 6243 { |
| 1602 | 6244 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6245 { |
| 1602 | 6246 if (REG_MULTI) |
| 6247 { | |
| 6248 reg_startpos[i] = bp->save_start[i].se_u.pos; | |
| 6249 reg_endpos[i] = bp->save_end[i].se_u.pos; | |
| 6250 } | |
| 6251 else | |
| 6252 { | |
| 6253 reg_startp[i] = bp->save_start[i].se_u.ptr; | |
| 6254 reg_endp[i] = bp->save_end[i].se_u.ptr; | |
| 6255 } | |
| 1579 | 6256 } |
| 6257 } | |
| 6258 } | |
| 6259 | |
| 6260 /* | |
| 7 | 6261 * Advance reglnum, regline and reginput to the next line. |
| 6262 */ | |
| 6263 static void | |
| 6264 reg_nextline() | |
| 6265 { | |
| 6266 regline = reg_getline(++reglnum); | |
| 6267 reginput = regline; | |
| 6268 fast_breakcheck(); | |
| 6269 } | |
| 6270 | |
| 6271 /* | |
| 6272 * Save the input line and position in a regsave_T. | |
| 6273 */ | |
| 6274 static void | |
| 233 | 6275 reg_save(save, gap) |
| 7 | 6276 regsave_T *save; |
| 233 | 6277 garray_T *gap; |
| 7 | 6278 { |
| 6279 if (REG_MULTI) | |
| 6280 { | |
| 6281 save->rs_u.pos.col = (colnr_T)(reginput - regline); | |
| 6282 save->rs_u.pos.lnum = reglnum; | |
| 6283 } | |
| 6284 else | |
| 6285 save->rs_u.ptr = reginput; | |
| 233 | 6286 save->rs_len = gap->ga_len; |
| 7 | 6287 } |
| 6288 | |
| 6289 /* | |
| 6290 * Restore the input line and position from a regsave_T. | |
| 6291 */ | |
| 6292 static void | |
| 233 | 6293 reg_restore(save, gap) |
| 7 | 6294 regsave_T *save; |
| 233 | 6295 garray_T *gap; |
| 7 | 6296 { |
| 6297 if (REG_MULTI) | |
| 6298 { | |
| 6299 if (reglnum != save->rs_u.pos.lnum) | |
| 6300 { | |
| 6301 /* only call reg_getline() when the line number changed to save | |
| 6302 * a bit of time */ | |
| 6303 reglnum = save->rs_u.pos.lnum; | |
| 6304 regline = reg_getline(reglnum); | |
| 6305 } | |
| 6306 reginput = regline + save->rs_u.pos.col; | |
| 6307 } | |
| 6308 else | |
| 6309 reginput = save->rs_u.ptr; | |
| 233 | 6310 gap->ga_len = save->rs_len; |
| 7 | 6311 } |
| 6312 | |
| 6313 /* | |
| 6314 * Return TRUE if current position is equal to saved position. | |
| 6315 */ | |
| 6316 static int | |
| 6317 reg_save_equal(save) | |
| 6318 regsave_T *save; | |
| 6319 { | |
| 6320 if (REG_MULTI) | |
| 6321 return reglnum == save->rs_u.pos.lnum | |
| 6322 && reginput == regline + save->rs_u.pos.col; | |
| 6323 return reginput == save->rs_u.ptr; | |
| 6324 } | |
| 6325 | |
| 6326 /* | |
| 6327 * Tentatively set the sub-expression start to the current position (after | |
| 6328 * calling regmatch() they will have changed). Need to save the existing | |
| 6329 * values for when there is no match. | |
| 6330 * Use se_save() to use pointer (save_se_multi()) or position (save_se_one()), | |
| 6331 * depending on REG_MULTI. | |
| 6332 */ | |
| 6333 static void | |
| 6334 save_se_multi(savep, posp) | |
| 6335 save_se_T *savep; | |
| 6336 lpos_T *posp; | |
| 6337 { | |
| 6338 savep->se_u.pos = *posp; | |
| 6339 posp->lnum = reglnum; | |
| 6340 posp->col = (colnr_T)(reginput - regline); | |
| 6341 } | |
| 6342 | |
| 6343 static void | |
| 6344 save_se_one(savep, pp) | |
| 6345 save_se_T *savep; | |
| 6346 char_u **pp; | |
| 6347 { | |
| 6348 savep->se_u.ptr = *pp; | |
| 6349 *pp = reginput; | |
| 6350 } | |
| 6351 | |
| 6352 /* | |
| 6353 * Compare a number with the operand of RE_LNUM, RE_COL or RE_VCOL. | |
| 6354 */ | |
| 6355 static int | |
| 6356 re_num_cmp(val, scan) | |
| 6357 long_u val; | |
| 6358 char_u *scan; | |
| 6359 { | |
| 6360 long_u n = OPERAND_MIN(scan); | |
| 6361 | |
| 6362 if (OPERAND_CMP(scan) == '>') | |
| 6363 return val > n; | |
| 6364 if (OPERAND_CMP(scan) == '<') | |
| 6365 return val < n; | |
| 6366 return val == n; | |
| 6367 } | |
| 6368 | |
| 6369 | |
| 4444 | 6370 #ifdef BT_REGEXP_DUMP |
| 7 | 6371 |
| 6372 /* | |
| 6373 * regdump - dump a regexp onto stdout in vaguely comprehensible form | |
| 6374 */ | |
| 6375 static void | |
| 6376 regdump(pattern, r) | |
| 6377 char_u *pattern; | |
| 4444 | 6378 bt_regprog_T *r; |
| 7 | 6379 { |
| 6380 char_u *s; | |
| 6381 int op = EXACTLY; /* Arbitrary non-END op. */ | |
| 6382 char_u *next; | |
| 6383 char_u *end = NULL; | |
| 4444 | 6384 FILE *f; |
| 6385 | |
| 6386 #ifdef BT_REGEXP_LOG | |
| 6387 f = fopen("bt_regexp_log.log", "a"); | |
| 6388 #else | |
| 6389 f = stdout; | |
| 6390 #endif | |
| 6391 if (f == NULL) | |
| 6392 return; | |
| 6393 fprintf(f, "-------------------------------------\n\r\nregcomp(%s):\r\n", pattern); | |
| 7 | 6394 |
| 6395 s = r->program + 1; | |
| 6396 /* | |
| 6397 * Loop until we find the END that isn't before a referred next (an END | |
| 6398 * can also appear in a NOMATCH operand). | |
| 6399 */ | |
| 6400 while (op != END || s <= end) | |
| 6401 { | |
| 6402 op = OP(s); | |
| 4444 | 6403 fprintf(f, "%2d%s", (int)(s - r->program), regprop(s)); /* Where, what. */ |
| 7 | 6404 next = regnext(s); |
| 6405 if (next == NULL) /* Next ptr. */ | |
| 4444 | 6406 fprintf(f, "(0)"); |
| 7 | 6407 else |
| 4444 | 6408 fprintf(f, "(%d)", (int)((s - r->program) + (next - s))); |
| 7 | 6409 if (end < next) |
| 6410 end = next; | |
| 6411 if (op == BRACE_LIMITS) | |
| 6412 { | |
| 6413 /* Two short ints */ | |
| 4444 | 6414 fprintf(f, " minval %ld, maxval %ld", OPERAND_MIN(s), OPERAND_MAX(s)); |
| 7 | 6415 s += 8; |
| 6416 } | |
| 6417 s += 3; | |
| 6418 if (op == ANYOF || op == ANYOF + ADD_NL | |
| 6419 || op == ANYBUT || op == ANYBUT + ADD_NL | |
| 6420 || op == EXACTLY) | |
| 6421 { | |
| 6422 /* Literal string, where present. */ | |
| 4444 | 6423 fprintf(f, "\nxxxxxxxxx\n"); |
| 7 | 6424 while (*s != NUL) |
| 4444 | 6425 fprintf(f, "%c", *s++); |
| 6426 fprintf(f, "\nxxxxxxxxx\n"); | |
| 7 | 6427 s++; |
| 6428 } | |
| 4444 | 6429 fprintf(f, "\r\n"); |
| 7 | 6430 } |
| 6431 | |
| 6432 /* Header fields of interest. */ | |
| 6433 if (r->regstart != NUL) | |
| 4444 | 6434 fprintf(f, "start `%s' 0x%x; ", r->regstart < 256 |
| 7 | 6435 ? (char *)transchar(r->regstart) |
| 6436 : "multibyte", r->regstart); | |
| 6437 if (r->reganch) | |
| 4444 | 6438 fprintf(f, "anchored; "); |
| 7 | 6439 if (r->regmust != NULL) |
| 4444 | 6440 fprintf(f, "must have \"%s\"", r->regmust); |
| 6441 fprintf(f, "\r\n"); | |
| 6442 | |
| 6443 #ifdef BT_REGEXP_LOG | |
| 6444 fclose(f); | |
| 6445 #endif | |
| 7 | 6446 } |
| 4444 | 6447 #endif /* BT_REGEXP_DUMP */ |
| 6448 | |
| 6449 #ifdef DEBUG | |
| 7 | 6450 /* |
| 6451 * regprop - printable representation of opcode | |
| 6452 */ | |
| 6453 static char_u * | |
| 6454 regprop(op) | |
| 6455 char_u *op; | |
| 6456 { | |
| 4444 | 6457 char *p; |
| 6458 static char buf[50]; | |
| 6459 | |
| 6460 STRCPY(buf, ":"); | |
| 6461 | |
| 6462 switch ((int) OP(op)) | |
| 7 | 6463 { |
| 6464 case BOL: | |
| 6465 p = "BOL"; | |
| 6466 break; | |
| 6467 case EOL: | |
| 6468 p = "EOL"; | |
| 6469 break; | |
| 6470 case RE_BOF: | |
| 6471 p = "BOF"; | |
| 6472 break; | |
| 6473 case RE_EOF: | |
| 6474 p = "EOF"; | |
| 6475 break; | |
| 6476 case CURSOR: | |
| 6477 p = "CURSOR"; | |
| 6478 break; | |
| 639 | 6479 case RE_VISUAL: |
| 6480 p = "RE_VISUAL"; | |
| 6481 break; | |
| 7 | 6482 case RE_LNUM: |
| 6483 p = "RE_LNUM"; | |
| 6484 break; | |
| 639 | 6485 case RE_MARK: |
| 6486 p = "RE_MARK"; | |
| 6487 break; | |
| 7 | 6488 case RE_COL: |
| 6489 p = "RE_COL"; | |
| 6490 break; | |
| 6491 case RE_VCOL: | |
| 6492 p = "RE_VCOL"; | |
| 6493 break; | |
| 6494 case BOW: | |
| 6495 p = "BOW"; | |
| 6496 break; | |
| 6497 case EOW: | |
| 6498 p = "EOW"; | |
| 6499 break; | |
| 6500 case ANY: | |
| 6501 p = "ANY"; | |
| 6502 break; | |
| 6503 case ANY + ADD_NL: | |
| 6504 p = "ANY+NL"; | |
| 6505 break; | |
| 6506 case ANYOF: | |
| 6507 p = "ANYOF"; | |
| 6508 break; | |
| 6509 case ANYOF + ADD_NL: | |
| 6510 p = "ANYOF+NL"; | |
| 6511 break; | |
| 6512 case ANYBUT: | |
| 6513 p = "ANYBUT"; | |
| 6514 break; | |
| 6515 case ANYBUT + ADD_NL: | |
| 6516 p = "ANYBUT+NL"; | |
| 6517 break; | |
| 6518 case IDENT: | |
| 6519 p = "IDENT"; | |
| 6520 break; | |
| 6521 case IDENT + ADD_NL: | |
| 6522 p = "IDENT+NL"; | |
| 6523 break; | |
| 6524 case SIDENT: | |
| 6525 p = "SIDENT"; | |
| 6526 break; | |
| 6527 case SIDENT + ADD_NL: | |
| 6528 p = "SIDENT+NL"; | |
| 6529 break; | |
| 6530 case KWORD: | |
| 6531 p = "KWORD"; | |
| 6532 break; | |
| 6533 case KWORD + ADD_NL: | |
| 6534 p = "KWORD+NL"; | |
| 6535 break; | |
| 6536 case SKWORD: | |
| 6537 p = "SKWORD"; | |
| 6538 break; | |
| 6539 case SKWORD + ADD_NL: | |
| 6540 p = "SKWORD+NL"; | |
| 6541 break; | |
| 6542 case FNAME: | |
| 6543 p = "FNAME"; | |
| 6544 break; | |
| 6545 case FNAME + ADD_NL: | |
| 6546 p = "FNAME+NL"; | |
| 6547 break; | |
| 6548 case SFNAME: | |
| 6549 p = "SFNAME"; | |
| 6550 break; | |
| 6551 case SFNAME + ADD_NL: | |
| 6552 p = "SFNAME+NL"; | |
| 6553 break; | |
| 6554 case PRINT: | |
| 6555 p = "PRINT"; | |
| 6556 break; | |
| 6557 case PRINT + ADD_NL: | |
| 6558 p = "PRINT+NL"; | |
| 6559 break; | |
| 6560 case SPRINT: | |
| 6561 p = "SPRINT"; | |
| 6562 break; | |
| 6563 case SPRINT + ADD_NL: | |
| 6564 p = "SPRINT+NL"; | |
| 6565 break; | |
| 6566 case WHITE: | |
| 6567 p = "WHITE"; | |
| 6568 break; | |
| 6569 case WHITE + ADD_NL: | |
| 6570 p = "WHITE+NL"; | |
| 6571 break; | |
| 6572 case NWHITE: | |
| 6573 p = "NWHITE"; | |
| 6574 break; | |
| 6575 case NWHITE + ADD_NL: | |
| 6576 p = "NWHITE+NL"; | |
| 6577 break; | |
| 6578 case DIGIT: | |
| 6579 p = "DIGIT"; | |
| 6580 break; | |
| 6581 case DIGIT + ADD_NL: | |
| 6582 p = "DIGIT+NL"; | |
| 6583 break; | |
| 6584 case NDIGIT: | |
| 6585 p = "NDIGIT"; | |
| 6586 break; | |
| 6587 case NDIGIT + ADD_NL: | |
| 6588 p = "NDIGIT+NL"; | |
| 6589 break; | |
| 6590 case HEX: | |
| 6591 p = "HEX"; | |
| 6592 break; | |
| 6593 case HEX + ADD_NL: | |
| 6594 p = "HEX+NL"; | |
| 6595 break; | |
| 6596 case NHEX: | |
| 6597 p = "NHEX"; | |
| 6598 break; | |
| 6599 case NHEX + ADD_NL: | |
| 6600 p = "NHEX+NL"; | |
| 6601 break; | |
| 6602 case OCTAL: | |
| 6603 p = "OCTAL"; | |
| 6604 break; | |
| 6605 case OCTAL + ADD_NL: | |
| 6606 p = "OCTAL+NL"; | |
| 6607 break; | |
| 6608 case NOCTAL: | |
| 6609 p = "NOCTAL"; | |
| 6610 break; | |
| 6611 case NOCTAL + ADD_NL: | |
| 6612 p = "NOCTAL+NL"; | |
| 6613 break; | |
| 6614 case WORD: | |
| 6615 p = "WORD"; | |
| 6616 break; | |
| 6617 case WORD + ADD_NL: | |
| 6618 p = "WORD+NL"; | |
| 6619 break; | |
| 6620 case NWORD: | |
| 6621 p = "NWORD"; | |
| 6622 break; | |
| 6623 case NWORD + ADD_NL: | |
| 6624 p = "NWORD+NL"; | |
| 6625 break; | |
| 6626 case HEAD: | |
| 6627 p = "HEAD"; | |
| 6628 break; | |
| 6629 case HEAD + ADD_NL: | |
| 6630 p = "HEAD+NL"; | |
| 6631 break; | |
| 6632 case NHEAD: | |
| 6633 p = "NHEAD"; | |
| 6634 break; | |
| 6635 case NHEAD + ADD_NL: | |
| 6636 p = "NHEAD+NL"; | |
| 6637 break; | |
| 6638 case ALPHA: | |
| 6639 p = "ALPHA"; | |
| 6640 break; | |
| 6641 case ALPHA + ADD_NL: | |
| 6642 p = "ALPHA+NL"; | |
| 6643 break; | |
| 6644 case NALPHA: | |
| 6645 p = "NALPHA"; | |
| 6646 break; | |
| 6647 case NALPHA + ADD_NL: | |
| 6648 p = "NALPHA+NL"; | |
| 6649 break; | |
| 6650 case LOWER: | |
| 6651 p = "LOWER"; | |
| 6652 break; | |
| 6653 case LOWER + ADD_NL: | |
| 6654 p = "LOWER+NL"; | |
| 6655 break; | |
| 6656 case NLOWER: | |
| 6657 p = "NLOWER"; | |
| 6658 break; | |
| 6659 case NLOWER + ADD_NL: | |
| 6660 p = "NLOWER+NL"; | |
| 6661 break; | |
| 6662 case UPPER: | |
| 6663 p = "UPPER"; | |
| 6664 break; | |
| 6665 case UPPER + ADD_NL: | |
| 6666 p = "UPPER+NL"; | |
| 6667 break; | |
| 6668 case NUPPER: | |
| 6669 p = "NUPPER"; | |
| 6670 break; | |
| 6671 case NUPPER + ADD_NL: | |
| 6672 p = "NUPPER+NL"; | |
| 6673 break; | |
| 6674 case BRANCH: | |
| 6675 p = "BRANCH"; | |
| 6676 break; | |
| 6677 case EXACTLY: | |
| 6678 p = "EXACTLY"; | |
| 6679 break; | |
| 6680 case NOTHING: | |
| 6681 p = "NOTHING"; | |
| 6682 break; | |
| 6683 case BACK: | |
| 6684 p = "BACK"; | |
| 6685 break; | |
| 6686 case END: | |
| 6687 p = "END"; | |
| 6688 break; | |
| 6689 case MOPEN + 0: | |
| 6690 p = "MATCH START"; | |
| 6691 break; | |
| 6692 case MOPEN + 1: | |
| 6693 case MOPEN + 2: | |
| 6694 case MOPEN + 3: | |
| 6695 case MOPEN + 4: | |
| 6696 case MOPEN + 5: | |
| 6697 case MOPEN + 6: | |
| 6698 case MOPEN + 7: | |
| 6699 case MOPEN + 8: | |
| 6700 case MOPEN + 9: | |
| 6701 sprintf(buf + STRLEN(buf), "MOPEN%d", OP(op) - MOPEN); | |
| 6702 p = NULL; | |
| 6703 break; | |
| 6704 case MCLOSE + 0: | |
| 6705 p = "MATCH END"; | |
| 6706 break; | |
| 6707 case MCLOSE + 1: | |
| 6708 case MCLOSE + 2: | |
| 6709 case MCLOSE + 3: | |
| 6710 case MCLOSE + 4: | |
| 6711 case MCLOSE + 5: | |
| 6712 case MCLOSE + 6: | |
| 6713 case MCLOSE + 7: | |
| 6714 case MCLOSE + 8: | |
| 6715 case MCLOSE + 9: | |
| 6716 sprintf(buf + STRLEN(buf), "MCLOSE%d", OP(op) - MCLOSE); | |
| 6717 p = NULL; | |
| 6718 break; | |
| 6719 case BACKREF + 1: | |
| 6720 case BACKREF + 2: | |
| 6721 case BACKREF + 3: | |
| 6722 case BACKREF + 4: | |
| 6723 case BACKREF + 5: | |
| 6724 case BACKREF + 6: | |
| 6725 case BACKREF + 7: | |
| 6726 case BACKREF + 8: | |
| 6727 case BACKREF + 9: | |
| 6728 sprintf(buf + STRLEN(buf), "BACKREF%d", OP(op) - BACKREF); | |
| 6729 p = NULL; | |
| 6730 break; | |
| 6731 case NOPEN: | |
| 6732 p = "NOPEN"; | |
| 6733 break; | |
| 6734 case NCLOSE: | |
| 6735 p = "NCLOSE"; | |
| 6736 break; | |
| 6737 #ifdef FEAT_SYN_HL | |
| 6738 case ZOPEN + 1: | |
| 6739 case ZOPEN + 2: | |
| 6740 case ZOPEN + 3: | |
| 6741 case ZOPEN + 4: | |
| 6742 case ZOPEN + 5: | |
| 6743 case ZOPEN + 6: | |
| 6744 case ZOPEN + 7: | |
| 6745 case ZOPEN + 8: | |
| 6746 case ZOPEN + 9: | |
| 6747 sprintf(buf + STRLEN(buf), "ZOPEN%d", OP(op) - ZOPEN); | |
| 6748 p = NULL; | |
| 6749 break; | |
| 6750 case ZCLOSE + 1: | |
| 6751 case ZCLOSE + 2: | |
| 6752 case ZCLOSE + 3: | |
| 6753 case ZCLOSE + 4: | |
| 6754 case ZCLOSE + 5: | |
| 6755 case ZCLOSE + 6: | |
| 6756 case ZCLOSE + 7: | |
| 6757 case ZCLOSE + 8: | |
| 6758 case ZCLOSE + 9: | |
| 6759 sprintf(buf + STRLEN(buf), "ZCLOSE%d", OP(op) - ZCLOSE); | |
| 6760 p = NULL; | |
| 6761 break; | |
| 6762 case ZREF + 1: | |
| 6763 case ZREF + 2: | |
| 6764 case ZREF + 3: | |
| 6765 case ZREF + 4: | |
| 6766 case ZREF + 5: | |
| 6767 case ZREF + 6: | |
| 6768 case ZREF + 7: | |
| 6769 case ZREF + 8: | |
| 6770 case ZREF + 9: | |
| 6771 sprintf(buf + STRLEN(buf), "ZREF%d", OP(op) - ZREF); | |
| 6772 p = NULL; | |
| 6773 break; | |
| 6774 #endif | |
| 6775 case STAR: | |
| 6776 p = "STAR"; | |
| 6777 break; | |
| 6778 case PLUS: | |
| 6779 p = "PLUS"; | |
| 6780 break; | |
| 6781 case NOMATCH: | |
| 6782 p = "NOMATCH"; | |
| 6783 break; | |
| 6784 case MATCH: | |
| 6785 p = "MATCH"; | |
| 6786 break; | |
| 6787 case BEHIND: | |
| 6788 p = "BEHIND"; | |
| 6789 break; | |
| 6790 case NOBEHIND: | |
| 6791 p = "NOBEHIND"; | |
| 6792 break; | |
| 6793 case SUBPAT: | |
| 6794 p = "SUBPAT"; | |
| 6795 break; | |
| 6796 case BRACE_LIMITS: | |
| 6797 p = "BRACE_LIMITS"; | |
| 6798 break; | |
| 6799 case BRACE_SIMPLE: | |
| 6800 p = "BRACE_SIMPLE"; | |
| 6801 break; | |
| 6802 case BRACE_COMPLEX + 0: | |
| 6803 case BRACE_COMPLEX + 1: | |
| 6804 case BRACE_COMPLEX + 2: | |
| 6805 case BRACE_COMPLEX + 3: | |
| 6806 case BRACE_COMPLEX + 4: | |
| 6807 case BRACE_COMPLEX + 5: | |
| 6808 case BRACE_COMPLEX + 6: | |
| 6809 case BRACE_COMPLEX + 7: | |
| 6810 case BRACE_COMPLEX + 8: | |
| 6811 case BRACE_COMPLEX + 9: | |
| 6812 sprintf(buf + STRLEN(buf), "BRACE_COMPLEX%d", OP(op) - BRACE_COMPLEX); | |
| 6813 p = NULL; | |
| 6814 break; | |
| 6815 #ifdef FEAT_MBYTE | |
| 6816 case MULTIBYTECODE: | |
| 6817 p = "MULTIBYTECODE"; | |
| 6818 break; | |
| 6819 #endif | |
| 6820 case NEWL: | |
| 6821 p = "NEWL"; | |
| 6822 break; | |
| 6823 default: | |
| 6824 sprintf(buf + STRLEN(buf), "corrupt %d", OP(op)); | |
| 6825 p = NULL; | |
| 6826 break; | |
| 6827 } | |
| 6828 if (p != NULL) | |
| 4444 | 6829 STRCAT(buf, p); |
| 6830 return (char_u *)buf; | |
| 7 | 6831 } |
| 4444 | 6832 #endif /* DEBUG */ |
| 7 | 6833 |
| 6834 #ifdef FEAT_MBYTE | |
| 6835 static void mb_decompose __ARGS((int c, int *c1, int *c2, int *c3)); | |
| 6836 | |
| 6837 typedef struct | |
| 6838 { | |
| 6839 int a, b, c; | |
| 6840 } decomp_T; | |
| 6841 | |
| 6842 | |
| 6843 /* 0xfb20 - 0xfb4f */ | |
| 297 | 6844 static decomp_T decomp_table[0xfb4f-0xfb20+1] = |
| 7 | 6845 { |
| 6846 {0x5e2,0,0}, /* 0xfb20 alt ayin */ | |
| 6847 {0x5d0,0,0}, /* 0xfb21 alt alef */ | |
| 6848 {0x5d3,0,0}, /* 0xfb22 alt dalet */ | |
| 6849 {0x5d4,0,0}, /* 0xfb23 alt he */ | |
| 6850 {0x5db,0,0}, /* 0xfb24 alt kaf */ | |
| 6851 {0x5dc,0,0}, /* 0xfb25 alt lamed */ | |
| 6852 {0x5dd,0,0}, /* 0xfb26 alt mem-sofit */ | |
| 6853 {0x5e8,0,0}, /* 0xfb27 alt resh */ | |
| 6854 {0x5ea,0,0}, /* 0xfb28 alt tav */ | |
| 6855 {'+', 0, 0}, /* 0xfb29 alt plus */ | |
| 6856 {0x5e9, 0x5c1, 0}, /* 0xfb2a shin+shin-dot */ | |
| 6857 {0x5e9, 0x5c2, 0}, /* 0xfb2b shin+sin-dot */ | |
| 6858 {0x5e9, 0x5c1, 0x5bc}, /* 0xfb2c shin+shin-dot+dagesh */ | |
| 6859 {0x5e9, 0x5c2, 0x5bc}, /* 0xfb2d shin+sin-dot+dagesh */ | |
| 6860 {0x5d0, 0x5b7, 0}, /* 0xfb2e alef+patah */ | |
| 6861 {0x5d0, 0x5b8, 0}, /* 0xfb2f alef+qamats */ | |
| 6862 {0x5d0, 0x5b4, 0}, /* 0xfb30 alef+hiriq */ | |
| 6863 {0x5d1, 0x5bc, 0}, /* 0xfb31 bet+dagesh */ | |
| 6864 {0x5d2, 0x5bc, 0}, /* 0xfb32 gimel+dagesh */ | |
| 6865 {0x5d3, 0x5bc, 0}, /* 0xfb33 dalet+dagesh */ | |
| 6866 {0x5d4, 0x5bc, 0}, /* 0xfb34 he+dagesh */ | |
| 6867 {0x5d5, 0x5bc, 0}, /* 0xfb35 vav+dagesh */ | |
| 6868 {0x5d6, 0x5bc, 0}, /* 0xfb36 zayin+dagesh */ | |
| 6869 {0xfb37, 0, 0}, /* 0xfb37 -- UNUSED */ | |
| 6870 {0x5d8, 0x5bc, 0}, /* 0xfb38 tet+dagesh */ | |
| 6871 {0x5d9, 0x5bc, 0}, /* 0xfb39 yud+dagesh */ | |
| 6872 {0x5da, 0x5bc, 0}, /* 0xfb3a kaf sofit+dagesh */ | |
| 6873 {0x5db, 0x5bc, 0}, /* 0xfb3b kaf+dagesh */ | |
| 6874 {0x5dc, 0x5bc, 0}, /* 0xfb3c lamed+dagesh */ | |
| 6875 {0xfb3d, 0, 0}, /* 0xfb3d -- UNUSED */ | |
| 6876 {0x5de, 0x5bc, 0}, /* 0xfb3e mem+dagesh */ | |
| 6877 {0xfb3f, 0, 0}, /* 0xfb3f -- UNUSED */ | |
| 6878 {0x5e0, 0x5bc, 0}, /* 0xfb40 nun+dagesh */ | |
| 6879 {0x5e1, 0x5bc, 0}, /* 0xfb41 samech+dagesh */ | |
| 6880 {0xfb42, 0, 0}, /* 0xfb42 -- UNUSED */ | |
| 6881 {0x5e3, 0x5bc, 0}, /* 0xfb43 pe sofit+dagesh */ | |
| 6882 {0x5e4, 0x5bc,0}, /* 0xfb44 pe+dagesh */ | |
| 6883 {0xfb45, 0, 0}, /* 0xfb45 -- UNUSED */ | |
| 6884 {0x5e6, 0x5bc, 0}, /* 0xfb46 tsadi+dagesh */ | |
| 6885 {0x5e7, 0x5bc, 0}, /* 0xfb47 qof+dagesh */ | |
| 6886 {0x5e8, 0x5bc, 0}, /* 0xfb48 resh+dagesh */ | |
| 6887 {0x5e9, 0x5bc, 0}, /* 0xfb49 shin+dagesh */ | |
| 6888 {0x5ea, 0x5bc, 0}, /* 0xfb4a tav+dagesh */ | |
| 6889 {0x5d5, 0x5b9, 0}, /* 0xfb4b vav+holam */ | |
| 6890 {0x5d1, 0x5bf, 0}, /* 0xfb4c bet+rafe */ | |
| 6891 {0x5db, 0x5bf, 0}, /* 0xfb4d kaf+rafe */ | |
| 6892 {0x5e4, 0x5bf, 0}, /* 0xfb4e pe+rafe */ | |
| 6893 {0x5d0, 0x5dc, 0} /* 0xfb4f alef-lamed */ | |
| 6894 }; | |
| 6895 | |
| 6896 static void | |
| 6897 mb_decompose(c, c1, c2, c3) | |
| 6898 int c, *c1, *c2, *c3; | |
| 6899 { | |
| 6900 decomp_T d; | |
| 6901 | |
|
4505
d037b9cbdaaa
updated for version 7.3.1000
Bram Moolenaar <bram@vim.org>
parents:
4466
diff
changeset
|
6902 if (c >= 0xfb20 && c <= 0xfb4f) |
| 7 | 6903 { |
| 6904 d = decomp_table[c - 0xfb20]; | |
| 6905 *c1 = d.a; | |
| 6906 *c2 = d.b; | |
| 6907 *c3 = d.c; | |
| 6908 } | |
| 6909 else | |
| 6910 { | |
| 6911 *c1 = c; | |
| 6912 *c2 = *c3 = 0; | |
| 6913 } | |
| 6914 } | |
| 6915 #endif | |
| 6916 | |
| 6917 /* | |
| 6918 * Compare two strings, ignore case if ireg_ic set. | |
| 6919 * Return 0 if strings match, non-zero otherwise. | |
| 6920 * Correct the length "*n" when composing characters are ignored. | |
| 6921 */ | |
| 6922 static int | |
| 6923 cstrncmp(s1, s2, n) | |
| 6924 char_u *s1, *s2; | |
| 6925 int *n; | |
| 6926 { | |
| 6927 int result; | |
| 6928 | |
| 6929 if (!ireg_ic) | |
| 6930 result = STRNCMP(s1, s2, *n); | |
| 6931 else | |
| 6932 result = MB_STRNICMP(s1, s2, *n); | |
| 6933 | |
| 6934 #ifdef FEAT_MBYTE | |
| 6935 /* if it failed and it's utf8 and we want to combineignore: */ | |
| 6936 if (result != 0 && enc_utf8 && ireg_icombine) | |
| 6937 { | |
| 6938 char_u *str1, *str2; | |
| 6939 int c1, c2, c11, c12; | |
| 6940 int junk; | |
| 6941 | |
| 6942 /* we have to handle the strcmp ourselves, since it is necessary to | |
| 6943 * deal with the composing characters by ignoring them: */ | |
| 6944 str1 = s1; | |
| 6945 str2 = s2; | |
| 6946 c1 = c2 = 0; | |
| 507 | 6947 while ((int)(str1 - s1) < *n) |
| 7 | 6948 { |
| 6949 c1 = mb_ptr2char_adv(&str1); | |
| 6950 c2 = mb_ptr2char_adv(&str2); | |
| 6951 | |
| 6952 /* decompose the character if necessary, into 'base' characters | |
| 6953 * because I don't care about Arabic, I will hard-code the Hebrew | |
| 6954 * which I *do* care about! So sue me... */ | |
| 6955 if (c1 != c2 && (!ireg_ic || utf_fold(c1) != utf_fold(c2))) | |
| 6956 { | |
| 6957 /* decomposition necessary? */ | |
| 6958 mb_decompose(c1, &c11, &junk, &junk); | |
| 6959 mb_decompose(c2, &c12, &junk, &junk); | |
| 6960 c1 = c11; | |
| 6961 c2 = c12; | |
| 6962 if (c11 != c12 && (!ireg_ic || utf_fold(c11) != utf_fold(c12))) | |
| 6963 break; | |
| 6964 } | |
| 6965 } | |
| 6966 result = c2 - c1; | |
| 6967 if (result == 0) | |
| 6968 *n = (int)(str2 - s2); | |
| 6969 } | |
| 6970 #endif | |
| 6971 | |
| 6972 return result; | |
| 6973 } | |
| 6974 | |
| 6975 /* | |
| 6976 * cstrchr: This function is used a lot for simple searches, keep it fast! | |
| 6977 */ | |
| 6978 static char_u * | |
| 6979 cstrchr(s, c) | |
| 6980 char_u *s; | |
| 6981 int c; | |
| 6982 { | |
| 6983 char_u *p; | |
| 6984 int cc; | |
| 6985 | |
| 6986 if (!ireg_ic | |
| 6987 #ifdef FEAT_MBYTE | |
| 6988 || (!enc_utf8 && mb_char2len(c) > 1) | |
| 6989 #endif | |
| 6990 ) | |
| 6991 return vim_strchr(s, c); | |
| 6992 | |
| 6993 /* tolower() and toupper() can be slow, comparing twice should be a lot | |
| 6994 * faster (esp. when using MS Visual C++!). | |
| 6995 * For UTF-8 need to use folded case. */ | |
| 6996 #ifdef FEAT_MBYTE | |
| 6997 if (enc_utf8 && c > 0x80) | |
| 6998 cc = utf_fold(c); | |
| 6999 else | |
| 7000 #endif | |
| 1347 | 7001 if (MB_ISUPPER(c)) |
| 7002 cc = MB_TOLOWER(c); | |
| 7003 else if (MB_ISLOWER(c)) | |
| 7004 cc = MB_TOUPPER(c); | |
| 7 | 7005 else |
| 7006 return vim_strchr(s, c); | |
| 7007 | |
| 7008 #ifdef FEAT_MBYTE | |
| 7009 if (has_mbyte) | |
| 7010 { | |
| 474 | 7011 for (p = s; *p != NUL; p += (*mb_ptr2len)(p)) |
| 7 | 7012 { |
| 7013 if (enc_utf8 && c > 0x80) | |
| 7014 { | |
| 7015 if (utf_fold(utf_ptr2char(p)) == cc) | |
| 7016 return p; | |
| 7017 } | |
| 7018 else if (*p == c || *p == cc) | |
| 7019 return p; | |
| 7020 } | |
| 7021 } | |
| 7022 else | |
| 7023 #endif | |
| 7024 /* Faster version for when there are no multi-byte characters. */ | |
| 7025 for (p = s; *p != NUL; ++p) | |
| 7026 if (*p == c || *p == cc) | |
| 7027 return p; | |
| 7028 | |
| 7029 return NULL; | |
| 7030 } | |
| 7031 | |
| 7032 /*************************************************************** | |
| 7033 * regsub stuff * | |
| 7034 ***************************************************************/ | |
| 7035 | |
| 7036 /* This stuff below really confuses cc on an SGI -- webb */ | |
| 7037 #ifdef __sgi | |
| 7038 # undef __ARGS | |
| 7039 # define __ARGS(x) () | |
| 7040 #endif | |
| 7041 | |
| 7042 /* | |
| 7043 * We should define ftpr as a pointer to a function returning a pointer to | |
| 7044 * a function returning a pointer to a function ... | |
| 7045 * This is impossible, so we declare a pointer to a function returning a | |
| 7046 * pointer to a function returning void. This should work for all compilers. | |
| 7047 */ | |
| 772 | 7048 typedef void (*(*fptr_T) __ARGS((int *, int)))(); |
| 7049 | |
| 7050 static fptr_T do_upper __ARGS((int *, int)); | |
| 7051 static fptr_T do_Upper __ARGS((int *, int)); | |
| 7052 static fptr_T do_lower __ARGS((int *, int)); | |
| 7053 static fptr_T do_Lower __ARGS((int *, int)); | |
| 7 | 7054 |
| 7055 static int vim_regsub_both __ARGS((char_u *source, char_u *dest, int copy, int magic, int backslash)); | |
| 7056 | |
| 772 | 7057 static fptr_T |
| 7 | 7058 do_upper(d, c) |
| 772 | 7059 int *d; |
| 7060 int c; | |
| 7 | 7061 { |
| 772 | 7062 *d = MB_TOUPPER(c); |
| 7063 | |
| 7064 return (fptr_T)NULL; | |
| 7 | 7065 } |
| 7066 | |
| 772 | 7067 static fptr_T |
| 7068 do_Upper(d, c) | |
| 7069 int *d; | |
| 7 | 7070 int c; |
| 7071 { | |
| 772 | 7072 *d = MB_TOUPPER(c); |
| 7073 | |
| 7074 return (fptr_T)do_Upper; | |
| 7075 } | |
| 7076 | |
| 7077 static fptr_T | |
| 7078 do_lower(d, c) | |
| 7079 int *d; | |
| 7080 int c; | |
| 7081 { | |
| 7082 *d = MB_TOLOWER(c); | |
| 7083 | |
| 7084 return (fptr_T)NULL; | |
| 7085 } | |
| 7086 | |
| 7087 static fptr_T | |
| 7088 do_Lower(d, c) | |
| 7089 int *d; | |
| 7090 int c; | |
| 7091 { | |
| 7092 *d = MB_TOLOWER(c); | |
| 7093 | |
| 7094 return (fptr_T)do_Lower; | |
| 7 | 7095 } |
| 7096 | |
| 7097 /* | |
| 7098 * regtilde(): Replace tildes in the pattern by the old pattern. | |
| 7099 * | |
| 7100 * Short explanation of the tilde: It stands for the previous replacement | |
| 7101 * pattern. If that previous pattern also contains a ~ we should go back a | |
| 7102 * step further... But we insert the previous pattern into the current one | |
| 7103 * and remember that. | |
| 772 | 7104 * This still does not handle the case where "magic" changes. So require the |
| 7105 * user to keep his hands off of "magic". | |
| 7 | 7106 * |
| 7107 * The tildes are parsed once before the first call to vim_regsub(). | |
| 7108 */ | |
| 7109 char_u * | |
| 7110 regtilde(source, magic) | |
| 7111 char_u *source; | |
| 7112 int magic; | |
| 7113 { | |
| 7114 char_u *newsub = source; | |
| 7115 char_u *tmpsub; | |
| 7116 char_u *p; | |
| 7117 int len; | |
| 7118 int prevlen; | |
| 7119 | |
| 7120 for (p = newsub; *p; ++p) | |
| 7121 { | |
| 7122 if ((*p == '~' && magic) || (*p == '\\' && *(p + 1) == '~' && !magic)) | |
| 7123 { | |
| 7124 if (reg_prev_sub != NULL) | |
| 7125 { | |
| 7126 /* length = len(newsub) - 1 + len(prev_sub) + 1 */ | |
| 7127 prevlen = (int)STRLEN(reg_prev_sub); | |
| 7128 tmpsub = alloc((unsigned)(STRLEN(newsub) + prevlen)); | |
| 7129 if (tmpsub != NULL) | |
| 7130 { | |
| 7131 /* copy prefix */ | |
| 7132 len = (int)(p - newsub); /* not including ~ */ | |
| 7133 mch_memmove(tmpsub, newsub, (size_t)len); | |
| 1209 | 7134 /* interpret tilde */ |
| 7 | 7135 mch_memmove(tmpsub + len, reg_prev_sub, (size_t)prevlen); |
| 7136 /* copy postfix */ | |
| 7137 if (!magic) | |
| 7138 ++p; /* back off \ */ | |
| 7139 STRCPY(tmpsub + len + prevlen, p + 1); | |
| 7140 | |
| 7141 if (newsub != source) /* already allocated newsub */ | |
| 7142 vim_free(newsub); | |
| 7143 newsub = tmpsub; | |
| 7144 p = newsub + len + prevlen; | |
| 7145 } | |
| 7146 } | |
| 7147 else if (magic) | |
| 1621 | 7148 STRMOVE(p, p + 1); /* remove '~' */ |
| 7 | 7149 else |
| 1621 | 7150 STRMOVE(p, p + 2); /* remove '\~' */ |
| 7 | 7151 --p; |
| 7152 } | |
| 7153 else | |
| 7154 { | |
| 7155 if (*p == '\\' && p[1]) /* skip escaped characters */ | |
| 7156 ++p; | |
| 7157 #ifdef FEAT_MBYTE | |
| 7158 if (has_mbyte) | |
| 474 | 7159 p += (*mb_ptr2len)(p) - 1; |
| 7 | 7160 #endif |
| 7161 } | |
| 7162 } | |
| 7163 | |
| 7164 vim_free(reg_prev_sub); | |
| 7165 if (newsub != source) /* newsub was allocated, just keep it */ | |
| 7166 reg_prev_sub = newsub; | |
| 7167 else /* no ~ found, need to save newsub */ | |
| 7168 reg_prev_sub = vim_strsave(newsub); | |
| 7169 return newsub; | |
| 7170 } | |
| 7171 | |
| 7172 #ifdef FEAT_EVAL | |
| 7173 static int can_f_submatch = FALSE; /* TRUE when submatch() can be used */ | |
| 7174 | |
| 7175 /* These pointers are used instead of reg_match and reg_mmatch for | |
| 7176 * reg_submatch(). Needed for when the substitution string is an expression | |
| 7177 * that contains a call to substitute() and submatch(). */ | |
| 7178 static regmatch_T *submatch_match; | |
| 7179 static regmmatch_T *submatch_mmatch; | |
| 2011 | 7180 static linenr_T submatch_firstlnum; |
| 7181 static linenr_T submatch_maxline; | |
| 2904 | 7182 static int submatch_line_lbr; |
| 7 | 7183 #endif |
| 7184 | |
| 7185 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO) | |
| 7186 /* | |
| 7187 * vim_regsub() - perform substitutions after a vim_regexec() or | |
| 7188 * vim_regexec_multi() match. | |
| 7189 * | |
| 7190 * If "copy" is TRUE really copy into "dest". | |
| 7191 * If "copy" is FALSE nothing is copied, this is just to find out the length | |
| 7192 * of the result. | |
| 7193 * | |
| 7194 * If "backslash" is TRUE, a backslash will be removed later, need to double | |
| 7195 * them to keep them, and insert a backslash before a CR to avoid it being | |
| 7196 * replaced with a line break later. | |
| 7197 * | |
| 7198 * Note: The matched text must not change between the call of | |
| 7199 * vim_regexec()/vim_regexec_multi() and vim_regsub()! It would make the back | |
| 7200 * references invalid! | |
| 7201 * | |
| 7202 * Returns the size of the replacement, including terminating NUL. | |
| 7203 */ | |
| 7204 int | |
| 7205 vim_regsub(rmp, source, dest, copy, magic, backslash) | |
| 7206 regmatch_T *rmp; | |
| 7207 char_u *source; | |
| 7208 char_u *dest; | |
| 7209 int copy; | |
| 7210 int magic; | |
| 7211 int backslash; | |
| 7212 { | |
| 7213 reg_match = rmp; | |
| 7214 reg_mmatch = NULL; | |
| 7215 reg_maxline = 0; | |
| 4061 | 7216 reg_buf = curbuf; |
| 7 | 7217 return vim_regsub_both(source, dest, copy, magic, backslash); |
| 7218 } | |
| 7219 #endif | |
| 7220 | |
| 7221 int | |
| 7222 vim_regsub_multi(rmp, lnum, source, dest, copy, magic, backslash) | |
| 7223 regmmatch_T *rmp; | |
| 7224 linenr_T lnum; | |
| 7225 char_u *source; | |
| 7226 char_u *dest; | |
| 7227 int copy; | |
| 7228 int magic; | |
| 7229 int backslash; | |
| 7230 { | |
| 7231 reg_match = NULL; | |
| 7232 reg_mmatch = rmp; | |
| 7233 reg_buf = curbuf; /* always works on the current buffer! */ | |
| 7234 reg_firstlnum = lnum; | |
| 7235 reg_maxline = curbuf->b_ml.ml_line_count - lnum; | |
| 7236 return vim_regsub_both(source, dest, copy, magic, backslash); | |
| 7237 } | |
| 7238 | |
| 7239 static int | |
| 7240 vim_regsub_both(source, dest, copy, magic, backslash) | |
| 7241 char_u *source; | |
| 7242 char_u *dest; | |
| 7243 int copy; | |
| 7244 int magic; | |
| 7245 int backslash; | |
| 7246 { | |
| 7247 char_u *src; | |
| 7248 char_u *dst; | |
| 7249 char_u *s; | |
| 7250 int c; | |
| 772 | 7251 int cc; |
| 7 | 7252 int no = -1; |
| 4244 | 7253 fptr_T func_all = (fptr_T)NULL; |
| 7254 fptr_T func_one = (fptr_T)NULL; | |
| 7 | 7255 linenr_T clnum = 0; /* init for GCC */ |
| 7256 int len = 0; /* init for GCC */ | |
| 7257 #ifdef FEAT_EVAL | |
| 7258 static char_u *eval_result = NULL; | |
| 7259 #endif | |
| 7260 | |
| 7261 /* Be paranoid... */ | |
| 7262 if (source == NULL || dest == NULL) | |
| 7263 { | |
| 7264 EMSG(_(e_null)); | |
| 7265 return 0; | |
| 7266 } | |
| 7267 if (prog_magic_wrong()) | |
| 7268 return 0; | |
| 7269 src = source; | |
| 7270 dst = dest; | |
| 7271 | |
| 7272 /* | |
| 7273 * When the substitute part starts with "\=" evaluate it as an expression. | |
| 7274 */ | |
| 7275 if (source[0] == '\\' && source[1] == '=' | |
| 7276 #ifdef FEAT_EVAL | |
| 7277 && !can_f_submatch /* can't do this recursively */ | |
| 7278 #endif | |
| 7279 ) | |
| 7280 { | |
| 7281 #ifdef FEAT_EVAL | |
| 7282 /* To make sure that the length doesn't change between checking the | |
| 7283 * length and copying the string, and to speed up things, the | |
| 7284 * resulting string is saved from the call with "copy" == FALSE to the | |
| 7285 * call with "copy" == TRUE. */ | |
| 7286 if (copy) | |
| 7287 { | |
| 7288 if (eval_result != NULL) | |
| 7289 { | |
| 7290 STRCPY(dest, eval_result); | |
| 7291 dst += STRLEN(eval_result); | |
| 7292 vim_free(eval_result); | |
| 7293 eval_result = NULL; | |
| 7294 } | |
| 7295 } | |
| 7296 else | |
| 7297 { | |
| 7298 win_T *save_reg_win; | |
| 7299 int save_ireg_ic; | |
| 7300 | |
| 7301 vim_free(eval_result); | |
| 7302 | |
| 7303 /* The expression may contain substitute(), which calls us | |
| 7304 * recursively. Make sure submatch() gets the text from the first | |
| 7305 * level. Don't need to save "reg_buf", because | |
| 7306 * vim_regexec_multi() can't be called recursively. */ | |
| 7307 submatch_match = reg_match; | |
| 7308 submatch_mmatch = reg_mmatch; | |
| 2011 | 7309 submatch_firstlnum = reg_firstlnum; |
| 7310 submatch_maxline = reg_maxline; | |
| 2904 | 7311 submatch_line_lbr = reg_line_lbr; |
| 7 | 7312 save_reg_win = reg_win; |
| 7313 save_ireg_ic = ireg_ic; | |
| 7314 can_f_submatch = TRUE; | |
| 7315 | |
| 714 | 7316 eval_result = eval_to_string(source + 2, NULL, TRUE); |
| 7 | 7317 if (eval_result != NULL) |
| 7318 { | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7319 int had_backslash = FALSE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7320 |
| 39 | 7321 for (s = eval_result; *s != NUL; mb_ptr_adv(s)) |
| 7 | 7322 { |
| 2904 | 7323 /* Change NL to CR, so that it becomes a line break, |
| 7324 * unless called from vim_regexec_nl(). | |
| 7 | 7325 * Skip over a backslashed character. */ |
| 2904 | 7326 if (*s == NL && !submatch_line_lbr) |
| 7 | 7327 *s = CAR; |
| 7328 else if (*s == '\\' && s[1] != NUL) | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7329 { |
| 7 | 7330 ++s; |
| 2173 | 7331 /* Change NL to CR here too, so that this works: |
| 7332 * :s/abc\\\ndef/\="aaa\\\nbbb"/ on text: | |
| 7333 * abc\ | |
| 7334 * def | |
| 2904 | 7335 * Not when called from vim_regexec_nl(). |
| 2173 | 7336 */ |
| 2904 | 7337 if (*s == NL && !submatch_line_lbr) |
| 2173 | 7338 *s = CAR; |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7339 had_backslash = TRUE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7340 } |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7341 } |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7342 if (had_backslash && backslash) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7343 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7344 /* Backslashes will be consumed, need to double them. */ |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7345 s = vim_strsave_escaped(eval_result, (char_u *)"\\"); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7346 if (s != NULL) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7347 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7348 vim_free(eval_result); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7349 eval_result = s; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7350 } |
| 7 | 7351 } |
| 7352 | |
| 7353 dst += STRLEN(eval_result); | |
| 7354 } | |
| 7355 | |
| 7356 reg_match = submatch_match; | |
| 7357 reg_mmatch = submatch_mmatch; | |
| 2011 | 7358 reg_firstlnum = submatch_firstlnum; |
| 7359 reg_maxline = submatch_maxline; | |
| 2904 | 7360 reg_line_lbr = submatch_line_lbr; |
| 7 | 7361 reg_win = save_reg_win; |
| 7362 ireg_ic = save_ireg_ic; | |
| 7363 can_f_submatch = FALSE; | |
| 7364 } | |
| 7365 #endif | |
| 7366 } | |
| 7367 else | |
| 7368 while ((c = *src++) != NUL) | |
| 7369 { | |
| 7370 if (c == '&' && magic) | |
| 7371 no = 0; | |
| 7372 else if (c == '\\' && *src != NUL) | |
| 7373 { | |
| 7374 if (*src == '&' && !magic) | |
| 7375 { | |
| 7376 ++src; | |
| 7377 no = 0; | |
| 7378 } | |
| 7379 else if ('0' <= *src && *src <= '9') | |
| 7380 { | |
| 7381 no = *src++ - '0'; | |
| 7382 } | |
| 7383 else if (vim_strchr((char_u *)"uUlLeE", *src)) | |
| 7384 { | |
| 7385 switch (*src++) | |
| 7386 { | |
| 4244 | 7387 case 'u': func_one = (fptr_T)do_upper; |
| 7 | 7388 continue; |
| 4244 | 7389 case 'U': func_all = (fptr_T)do_Upper; |
| 7 | 7390 continue; |
| 4244 | 7391 case 'l': func_one = (fptr_T)do_lower; |
| 7 | 7392 continue; |
| 4244 | 7393 case 'L': func_all = (fptr_T)do_Lower; |
| 7 | 7394 continue; |
| 7395 case 'e': | |
| 4244 | 7396 case 'E': func_one = func_all = (fptr_T)NULL; |
| 7 | 7397 continue; |
| 7398 } | |
| 7399 } | |
| 7400 } | |
| 7401 if (no < 0) /* Ordinary character. */ | |
| 7402 { | |
| 798 | 7403 if (c == K_SPECIAL && src[0] != NUL && src[1] != NUL) |
| 7404 { | |
| 1209 | 7405 /* Copy a special key as-is. */ |
| 798 | 7406 if (copy) |
| 7407 { | |
| 7408 *dst++ = c; | |
| 7409 *dst++ = *src++; | |
| 7410 *dst++ = *src++; | |
| 7411 } | |
| 7412 else | |
| 7413 { | |
| 7414 dst += 3; | |
| 7415 src += 2; | |
| 7416 } | |
| 7417 continue; | |
| 7418 } | |
| 7419 | |
| 7 | 7420 if (c == '\\' && *src != NUL) |
| 7421 { | |
| 7422 /* Check for abbreviations -- webb */ | |
| 7423 switch (*src) | |
| 7424 { | |
| 7425 case 'r': c = CAR; ++src; break; | |
| 7426 case 'n': c = NL; ++src; break; | |
| 7427 case 't': c = TAB; ++src; break; | |
| 7428 /* Oh no! \e already has meaning in subst pat :-( */ | |
| 7429 /* case 'e': c = ESC; ++src; break; */ | |
| 7430 case 'b': c = Ctrl_H; ++src; break; | |
| 7431 | |
| 7432 /* If "backslash" is TRUE the backslash will be removed | |
| 7433 * later. Used to insert a literal CR. */ | |
| 7434 default: if (backslash) | |
| 7435 { | |
| 7436 if (copy) | |
| 7437 *dst = '\\'; | |
| 7438 ++dst; | |
| 7439 } | |
| 7440 c = *src++; | |
| 7441 } | |
| 7442 } | |
| 798 | 7443 #ifdef FEAT_MBYTE |
| 7444 else if (has_mbyte) | |
| 7445 c = mb_ptr2char(src - 1); | |
| 7446 #endif | |
| 7 | 7447 |
| 7448 /* Write to buffer, if copy is set. */ | |
| 4244 | 7449 if (func_one != (fptr_T)NULL) |
| 7450 /* Turbo C complains without the typecast */ | |
| 7451 func_one = (fptr_T)(func_one(&cc, c)); | |
| 7452 else if (func_all != (fptr_T)NULL) | |
| 7453 /* Turbo C complains without the typecast */ | |
| 7454 func_all = (fptr_T)(func_all(&cc, c)); | |
| 7455 else /* just copy */ | |
| 772 | 7456 cc = c; |
| 7457 | |
| 7458 #ifdef FEAT_MBYTE | |
| 7459 if (has_mbyte) | |
| 7 | 7460 { |
|
2307
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7461 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
|
7462 |
| 7 | 7463 if (copy) |
| 772 | 7464 mb_char2bytes(cc, dst); |
| 7465 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
|
7466 if (enc_utf8) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7467 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7468 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
|
7469 |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7470 /* 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
|
7471 * 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
|
7472 if (clen < totlen) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7473 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7474 if (copy) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7475 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
|
7476 (size_t)(totlen - clen)); |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7477 dst += totlen - clen; |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7478 } |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7479 } |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7480 src += totlen - 1; |
| 7 | 7481 } |
| 7482 else | |
| 7483 #endif | |
| 7484 if (copy) | |
| 772 | 7485 *dst = cc; |
| 7 | 7486 dst++; |
| 7487 } | |
| 7488 else | |
| 7489 { | |
| 7490 if (REG_MULTI) | |
| 7491 { | |
| 7492 clnum = reg_mmatch->startpos[no].lnum; | |
| 7493 if (clnum < 0 || reg_mmatch->endpos[no].lnum < 0) | |
| 7494 s = NULL; | |
| 7495 else | |
| 7496 { | |
| 7497 s = reg_getline(clnum) + reg_mmatch->startpos[no].col; | |
| 7498 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7499 len = reg_mmatch->endpos[no].col | |
| 7500 - reg_mmatch->startpos[no].col; | |
| 7501 else | |
| 7502 len = (int)STRLEN(s); | |
| 7503 } | |
| 7504 } | |
| 7505 else | |
| 7506 { | |
| 7507 s = reg_match->startp[no]; | |
| 7508 if (reg_match->endp[no] == NULL) | |
| 7509 s = NULL; | |
| 7510 else | |
| 7511 len = (int)(reg_match->endp[no] - s); | |
| 7512 } | |
| 7513 if (s != NULL) | |
| 7514 { | |
| 7515 for (;;) | |
| 7516 { | |
| 7517 if (len == 0) | |
| 7518 { | |
| 7519 if (REG_MULTI) | |
| 7520 { | |
| 7521 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7522 break; | |
| 7523 if (copy) | |
| 7524 *dst = CAR; | |
| 7525 ++dst; | |
| 7526 s = reg_getline(++clnum); | |
| 7527 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7528 len = reg_mmatch->endpos[no].col; | |
| 7529 else | |
| 7530 len = (int)STRLEN(s); | |
| 7531 } | |
| 7532 else | |
| 7533 break; | |
| 7534 } | |
| 7535 else if (*s == NUL) /* we hit NUL. */ | |
| 7536 { | |
| 7537 if (copy) | |
| 7538 EMSG(_(e_re_damg)); | |
| 7539 goto exit; | |
| 7540 } | |
| 7541 else | |
| 7542 { | |
| 7543 if (backslash && (*s == CAR || *s == '\\')) | |
| 7544 { | |
| 7545 /* | |
| 7546 * Insert a backslash in front of a CR, otherwise | |
| 7547 * it will be replaced by a line break. | |
| 7548 * Number of backslashes will be halved later, | |
| 7549 * double them here. | |
| 7550 */ | |
| 7551 if (copy) | |
| 7552 { | |
| 7553 dst[0] = '\\'; | |
| 7554 dst[1] = *s; | |
| 7555 } | |
| 7556 dst += 2; | |
| 7557 } | |
| 7558 else | |
| 7559 { | |
| 772 | 7560 #ifdef FEAT_MBYTE |
| 7561 if (has_mbyte) | |
| 7562 c = mb_ptr2char(s); | |
| 7563 else | |
| 7564 #endif | |
| 7565 c = *s; | |
| 7566 | |
| 4244 | 7567 if (func_one != (fptr_T)NULL) |
| 7568 /* Turbo C complains without the typecast */ | |
| 7569 func_one = (fptr_T)(func_one(&cc, c)); | |
| 7570 else if (func_all != (fptr_T)NULL) | |
| 7571 /* Turbo C complains without the typecast */ | |
| 7572 func_all = (fptr_T)(func_all(&cc, c)); | |
| 7573 else /* just copy */ | |
| 772 | 7574 cc = c; |
| 7575 | |
| 7576 #ifdef FEAT_MBYTE | |
| 7577 if (has_mbyte) | |
| 7 | 7578 { |
| 1332 | 7579 int l; |
| 7580 | |
| 7581 /* Copy composing characters separately, one | |
| 7582 * at a time. */ | |
| 7583 if (enc_utf8) | |
| 7584 l = utf_ptr2len(s) - 1; | |
| 7585 else | |
| 7586 l = mb_ptr2len(s) - 1; | |
| 772 | 7587 |
| 7588 s += l; | |
| 7589 len -= l; | |
| 7590 if (copy) | |
| 7591 mb_char2bytes(cc, dst); | |
| 7592 dst += mb_char2len(cc) - 1; | |
| 7 | 7593 } |
| 772 | 7594 else |
| 7595 #endif | |
| 7596 if (copy) | |
| 7597 *dst = cc; | |
| 7598 dst++; | |
| 7 | 7599 } |
| 772 | 7600 |
| 7 | 7601 ++s; |
| 7602 --len; | |
| 7603 } | |
| 7604 } | |
| 7605 } | |
| 7606 no = -1; | |
| 7607 } | |
| 7608 } | |
| 7609 if (copy) | |
| 7610 *dst = NUL; | |
| 7611 | |
| 7612 exit: | |
| 7613 return (int)((dst - dest) + 1); | |
| 7614 } | |
| 7615 | |
| 7616 #ifdef FEAT_EVAL | |
| 2012 | 7617 static char_u *reg_getline_submatch __ARGS((linenr_T lnum)); |
| 7618 | |
| 7 | 7619 /* |
| 2011 | 7620 * Call reg_getline() with the line numbers from the submatch. If a |
| 7621 * substitute() was used the reg_maxline and other values have been | |
| 7622 * overwritten. | |
| 7623 */ | |
| 7624 static char_u * | |
| 7625 reg_getline_submatch(lnum) | |
| 7626 linenr_T lnum; | |
| 7627 { | |
| 7628 char_u *s; | |
| 7629 linenr_T save_first = reg_firstlnum; | |
| 7630 linenr_T save_max = reg_maxline; | |
| 7631 | |
| 7632 reg_firstlnum = submatch_firstlnum; | |
| 7633 reg_maxline = submatch_maxline; | |
| 7634 | |
| 7635 s = reg_getline(lnum); | |
| 7636 | |
| 7637 reg_firstlnum = save_first; | |
| 7638 reg_maxline = save_max; | |
| 7639 return s; | |
| 7640 } | |
| 7641 | |
| 7642 /* | |
| 1209 | 7643 * Used for the submatch() function: get the string from the n'th submatch in |
| 7 | 7644 * allocated memory. |
| 7645 * Returns NULL when not in a ":s" command and for a non-existing submatch. | |
| 7646 */ | |
| 7647 char_u * | |
| 7648 reg_submatch(no) | |
| 7649 int no; | |
| 7650 { | |
| 7651 char_u *retval = NULL; | |
| 7652 char_u *s; | |
| 7653 int len; | |
| 7654 int round; | |
| 7655 linenr_T lnum; | |
| 7656 | |
| 840 | 7657 if (!can_f_submatch || no < 0) |
| 7 | 7658 return NULL; |
| 7659 | |
| 7660 if (submatch_match == NULL) | |
| 7661 { | |
| 7662 /* | |
| 7663 * First round: compute the length and allocate memory. | |
| 7664 * Second round: copy the text. | |
| 7665 */ | |
| 7666 for (round = 1; round <= 2; ++round) | |
| 7667 { | |
| 7668 lnum = submatch_mmatch->startpos[no].lnum; | |
| 7669 if (lnum < 0 || submatch_mmatch->endpos[no].lnum < 0) | |
| 7670 return NULL; | |
| 7671 | |
| 2011 | 7672 s = reg_getline_submatch(lnum) + submatch_mmatch->startpos[no].col; |
| 7 | 7673 if (s == NULL) /* anti-crash check, cannot happen? */ |
| 7674 break; | |
| 7675 if (submatch_mmatch->endpos[no].lnum == lnum) | |
| 7676 { | |
| 7677 /* Within one line: take form start to end col. */ | |
| 7678 len = submatch_mmatch->endpos[no].col | |
| 7679 - submatch_mmatch->startpos[no].col; | |
| 7680 if (round == 2) | |
| 418 | 7681 vim_strncpy(retval, s, len); |
| 7 | 7682 ++len; |
| 7683 } | |
| 7684 else | |
| 7685 { | |
| 7686 /* Multiple lines: take start line from start col, middle | |
| 7687 * lines completely and end line up to end col. */ | |
| 7688 len = (int)STRLEN(s); | |
| 7689 if (round == 2) | |
| 7690 { | |
| 7691 STRCPY(retval, s); | |
| 7692 retval[len] = '\n'; | |
| 7693 } | |
| 7694 ++len; | |
| 7695 ++lnum; | |
| 7696 while (lnum < submatch_mmatch->endpos[no].lnum) | |
| 7697 { | |
| 2011 | 7698 s = reg_getline_submatch(lnum++); |
| 7 | 7699 if (round == 2) |
| 7700 STRCPY(retval + len, s); | |
| 7701 len += (int)STRLEN(s); | |
| 7702 if (round == 2) | |
| 7703 retval[len] = '\n'; | |
| 7704 ++len; | |
| 7705 } | |
| 7706 if (round == 2) | |
| 2011 | 7707 STRNCPY(retval + len, reg_getline_submatch(lnum), |
| 7 | 7708 submatch_mmatch->endpos[no].col); |
| 7709 len += submatch_mmatch->endpos[no].col; | |
| 7710 if (round == 2) | |
| 7711 retval[len] = NUL; | |
| 7712 ++len; | |
| 7713 } | |
| 7714 | |
| 840 | 7715 if (retval == NULL) |
| 7 | 7716 { |
| 7717 retval = lalloc((long_u)len, TRUE); | |
| 840 | 7718 if (retval == NULL) |
| 7 | 7719 return NULL; |
| 7720 } | |
| 7721 } | |
| 7722 } | |
| 7723 else | |
| 7724 { | |
| 1815 | 7725 s = submatch_match->startp[no]; |
| 7726 if (s == NULL || submatch_match->endp[no] == NULL) | |
| 7 | 7727 retval = NULL; |
| 7728 else | |
| 7729 retval = vim_strnsave(s, (int)(submatch_match->endp[no] - s)); | |
| 7730 } | |
| 7731 | |
| 7732 return retval; | |
| 7733 } | |
| 7734 #endif | |
| 4444 | 7735 |
| 7736 static regengine_T bt_regengine = | |
| 7737 { | |
| 7738 bt_regcomp, | |
| 7739 bt_regexec, | |
| 7740 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
| 7741 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 7742 bt_regexec_nl, | |
| 7743 #endif | |
| 7744 bt_regexec_multi | |
| 7745 #ifdef DEBUG | |
| 7746 ,(char_u *)"" | |
| 7747 #endif | |
| 7748 }; | |
| 7749 | |
| 7750 | |
| 7751 #include "regexp_nfa.c" | |
| 7752 | |
| 7753 static regengine_T nfa_regengine = | |
| 7754 { | |
| 7755 nfa_regcomp, | |
| 7756 nfa_regexec, | |
| 7757 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
| 7758 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 7759 nfa_regexec_nl, | |
| 7760 #endif | |
| 7761 nfa_regexec_multi | |
| 7762 #ifdef DEBUG | |
| 7763 ,(char_u *)"" | |
| 7764 #endif | |
| 7765 }; | |
| 7766 | |
| 7767 /* Which regexp engine to use? Needed for vim_regcomp(). | |
| 7768 * Must match with 'regexpengine'. */ | |
| 7769 static int regexp_engine = 0; | |
| 7770 #define AUTOMATIC_ENGINE 0 | |
| 7771 #define BACKTRACKING_ENGINE 1 | |
| 7772 #define NFA_ENGINE 2 | |
| 7773 #ifdef DEBUG | |
| 7774 static char_u regname[][30] = { | |
| 7775 "AUTOMATIC Regexp Engine", | |
| 7776 "BACKTACKING Regexp Engine", | |
| 7777 "NFA Regexp Engine" | |
| 7778 }; | |
| 7779 #endif | |
| 7780 | |
| 7781 /* | |
| 7782 * Compile a regular expression into internal code. | |
| 7783 * Returns the program in allocated memory. Returns NULL for an error. | |
| 7784 */ | |
| 7785 regprog_T * | |
| 7786 vim_regcomp(expr_arg, re_flags) | |
| 7787 char_u *expr_arg; | |
| 7788 int re_flags; | |
| 7789 { | |
| 7790 regprog_T *prog = NULL; | |
| 7791 char_u *expr = expr_arg; | |
| 7792 | |
| 7793 syntax_error = FALSE; | |
| 7794 regexp_engine = p_re; | |
| 7795 | |
| 7796 /* Check for prefix "\%#=", that sets the regexp engine */ | |
| 7797 if (STRNCMP(expr, "\\%#=", 4) == 0) | |
| 7798 { | |
| 7799 int newengine = expr[4] - '0'; | |
| 7800 | |
| 7801 if (newengine == AUTOMATIC_ENGINE | |
| 7802 || newengine == BACKTRACKING_ENGINE | |
| 7803 || newengine == NFA_ENGINE) | |
| 7804 { | |
| 7805 regexp_engine = expr[4] - '0'; | |
| 7806 expr += 5; | |
| 7807 #ifdef DEBUG | |
| 7808 EMSG3("New regexp mode selected (%d): %s", regexp_engine, | |
| 7809 regname[newengine]); | |
| 7810 #endif | |
| 7811 } | |
| 7812 else | |
| 7813 { | |
| 7814 EMSG(_("E864: \\%#= can only be followed by 0, 1, or 2. The automatic engine will be used ")); | |
| 7815 regexp_engine = AUTOMATIC_ENGINE; | |
| 7816 } | |
| 7817 } | |
| 7818 #ifdef DEBUG | |
| 7819 bt_regengine.expr = expr; | |
| 7820 nfa_regengine.expr = expr; | |
| 7821 #endif | |
| 7822 | |
| 7823 /* | |
| 7824 * First try the NFA engine, unless backtracking was requested. | |
| 7825 */ | |
| 7826 if (regexp_engine != BACKTRACKING_ENGINE) | |
| 7827 prog = nfa_regengine.regcomp(expr, re_flags); | |
| 7828 else | |
| 7829 prog = bt_regengine.regcomp(expr, re_flags); | |
| 7830 | |
| 7831 if (prog == NULL) /* error compiling regexp with initial engine */ | |
| 7832 { | |
| 4460 | 7833 #ifdef BT_REGEXP_DEBUG_LOG |
| 4444 | 7834 if (regexp_engine != BACKTRACKING_ENGINE) /* debugging log for NFA */ |
| 7835 { | |
| 7836 FILE *f; | |
| 4460 | 7837 f = fopen(BT_REGEXP_DEBUG_LOG_NAME, "a"); |
| 4444 | 7838 if (f) |
| 7839 { | |
| 7840 if (!syntax_error) | |
| 7841 fprintf(f, "NFA engine could not handle \"%s\"\n", expr); | |
| 7842 else | |
| 7843 fprintf(f, "Syntax error in \"%s\"\n", expr); | |
| 7844 fclose(f); | |
| 7845 } | |
| 7846 else | |
| 4460 | 7847 EMSG2("(NFA) Could not open \"%s\" to write !!!", |
| 7848 BT_REGEXP_DEBUG_LOG_NAME); | |
| 4444 | 7849 /* |
| 7850 if (syntax_error) | |
| 7851 EMSG("NFA Regexp: Syntax Error !"); | |
| 7852 */ | |
| 7853 } | |
| 7854 #endif | |
| 7855 /* | |
| 7856 * If NFA engine failed, then revert to the backtracking engine. | |
| 7857 * Except when there was a syntax error, which was properly handled by | |
| 7858 * NFA engine. | |
| 7859 */ | |
| 7860 if (regexp_engine == AUTOMATIC_ENGINE) | |
| 7861 if (!syntax_error) | |
| 7862 prog = bt_regengine.regcomp(expr, re_flags); | |
| 7863 | |
| 7864 } /* endif prog==NULL */ | |
| 7865 | |
| 7866 | |
| 7867 return prog; | |
| 7868 } | |
| 7869 | |
| 7870 /* | |
| 7871 * Match a regexp against a string. | |
| 7872 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 7873 * Uses curbuf for line count and 'iskeyword'. | |
| 7874 * | |
| 7875 * Return TRUE if there is a match, FALSE if not. | |
| 7876 */ | |
| 7877 int | |
| 7878 vim_regexec(rmp, line, col) | |
| 7879 regmatch_T *rmp; | |
| 7880 char_u *line; /* string to match against */ | |
| 7881 colnr_T col; /* column to start looking for match */ | |
| 7882 { | |
| 7883 return rmp->regprog->engine->regexec(rmp, line, col); | |
| 7884 } | |
| 7885 | |
| 7886 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ | |
| 7887 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 7888 /* | |
| 7889 * Like vim_regexec(), but consider a "\n" in "line" to be a line break. | |
| 7890 */ | |
| 7891 int | |
| 7892 vim_regexec_nl(rmp, line, col) | |
| 7893 regmatch_T *rmp; | |
| 7894 char_u *line; | |
| 7895 colnr_T col; | |
| 7896 { | |
| 7897 return rmp->regprog->engine->regexec_nl(rmp, line, col); | |
| 7898 } | |
| 7899 #endif | |
| 7900 | |
| 7901 /* | |
| 7902 * Match a regexp against multiple lines. | |
| 7903 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 7904 * Uses curbuf for line count and 'iskeyword'. | |
| 7905 * | |
| 7906 * Return zero if there is no match. Return number of lines contained in the | |
| 7907 * match otherwise. | |
| 7908 */ | |
| 7909 long | |
| 7910 vim_regexec_multi(rmp, win, buf, lnum, col, tm) | |
| 7911 regmmatch_T *rmp; | |
| 7912 win_T *win; /* window in which to search or NULL */ | |
| 7913 buf_T *buf; /* buffer in which to search */ | |
| 7914 linenr_T lnum; /* nr of line to start looking for match */ | |
| 7915 colnr_T col; /* column to start looking for match */ | |
| 7916 proftime_T *tm; /* timeout limit or NULL */ | |
| 7917 { | |
| 7918 return rmp->regprog->engine->regexec_multi(rmp, win, buf, lnum, col, tm); | |
| 7919 } |