Mercurial > vim
annotate src/regexp.c @ 4176:1778d51fcbc3 v7.3.840
updated for version 7.3.840
Problem: "\@<!" in regexp does not work correctly with multi-byte
characters, especially cp932.
Solution: Move column to start of multi-byte character. (Yasuhiro Matsumoto)
| author | Bram Moolenaar <bram@vim.org> |
|---|---|
| date | Tue, 26 Feb 2013 18:46:01 +0100 |
| parents | 2c12c4e2fae8 |
| children | 7faeece39228 |
| 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 | |
| 41 #include "vim.h" | |
| 42 | |
| 43 #undef DEBUG | |
| 44 | |
| 45 /* | |
| 46 * The "internal use only" fields in regexp.h are present to pass info from | |
| 47 * compile to execute that permits the execute phase to run lots faster on | |
| 48 * simple cases. They are: | |
| 49 * | |
| 50 * regstart char that must begin a match; NUL if none obvious; Can be a | |
| 51 * multi-byte character. | |
| 52 * reganch is the match anchored (at beginning-of-line only)? | |
| 53 * regmust string (pointer into program) that match must include, or NULL | |
| 54 * regmlen length of regmust string | |
| 55 * regflags RF_ values or'ed together | |
| 56 * | |
| 57 * Regstart and reganch permit very fast decisions on suitable starting points | |
| 58 * for a match, cutting down the work a lot. Regmust permits fast rejection | |
| 59 * of lines that cannot possibly match. The regmust tests are costly enough | |
| 60 * that vim_regcomp() supplies a regmust only if the r.e. contains something | |
| 61 * potentially expensive (at present, the only such thing detected is * or + | |
| 62 * at the start of the r.e., which can involve a lot of backup). Regmlen is | |
| 63 * supplied because the test in vim_regexec() needs it and vim_regcomp() is | |
| 64 * computing it anyway. | |
| 65 */ | |
| 66 | |
| 67 /* | |
| 68 * Structure for regexp "program". This is essentially a linear encoding | |
| 69 * of a nondeterministic finite-state machine (aka syntax charts or | |
| 70 * "railroad normal form" in parsing technology). Each node is an opcode | |
| 71 * plus a "next" pointer, possibly plus an operand. "Next" pointers of | |
| 72 * all nodes except BRANCH and BRACES_COMPLEX implement concatenation; a "next" | |
| 73 * pointer with a BRANCH on both ends of it is connecting two alternatives. | |
| 74 * (Here we have one of the subtle syntax dependencies: an individual BRANCH | |
| 75 * (as opposed to a collection of them) is never concatenated with anything | |
| 76 * because of operator precedence). The "next" pointer of a BRACES_COMPLEX | |
| 167 | 77 * node points to the node after the stuff to be repeated. |
| 78 * The operand of some types of node is a literal string; for others, it is a | |
| 79 * node leading into a sub-FSM. In particular, the operand of a BRANCH node | |
| 80 * is the first node of the branch. | |
| 81 * (NB this is *not* a tree structure: the tail of the branch connects to the | |
| 82 * thing following the set of BRANCHes.) | |
| 7 | 83 * |
| 84 * pattern is coded like: | |
| 85 * | |
| 86 * +-----------------+ | |
| 87 * | V | |
| 88 * <aa>\|<bb> BRANCH <aa> BRANCH <bb> --> END | |
| 89 * | ^ | ^ | |
| 90 * +------+ +----------+ | |
| 91 * | |
| 92 * | |
| 93 * +------------------+ | |
| 94 * V | | |
| 95 * <aa>* BRANCH BRANCH <aa> --> BACK BRANCH --> NOTHING --> END | |
| 96 * | | ^ ^ | |
| 97 * | +---------------+ | | |
| 98 * +---------------------------------------------+ | |
| 99 * | |
| 100 * | |
| 167 | 101 * +----------------------+ |
| 102 * V | | |
| 233 | 103 * <aa>\+ BRANCH <aa> --> BRANCH --> BACK BRANCH --> NOTHING --> END |
| 856 | 104 * | | ^ ^ |
| 105 * | +-----------+ | | |
| 179 | 106 * +--------------------------------------------------+ |
| 167 | 107 * |
| 108 * | |
| 7 | 109 * +-------------------------+ |
| 110 * V | | |
| 111 * <aa>\{} BRANCH BRACE_LIMITS --> BRACE_COMPLEX <aa> --> BACK END | |
| 112 * | | ^ | |
| 113 * | +----------------+ | |
| 114 * +-----------------------------------------------+ | |
| 115 * | |
| 116 * | |
| 117 * <aa>\@!<bb> BRANCH NOMATCH <aa> --> END <bb> --> END | |
| 118 * | | ^ ^ | |
| 119 * | +----------------+ | | |
| 120 * +--------------------------------+ | |
| 121 * | |
| 122 * +---------+ | |
| 123 * | V | |
| 124 * \z[abc] BRANCH BRANCH a BRANCH b BRANCH c BRANCH NOTHING --> END | |
| 125 * | | | | ^ ^ | |
| 126 * | | | +-----+ | | |
| 127 * | | +----------------+ | | |
| 128 * | +---------------------------+ | | |
| 129 * +------------------------------------------------------+ | |
| 130 * | |
| 1209 | 131 * They all start with a BRANCH for "\|" alternatives, even when there is only |
| 7 | 132 * one alternative. |
| 133 */ | |
| 134 | |
| 135 /* | |
| 136 * The opcodes are: | |
| 137 */ | |
| 138 | |
| 139 /* definition number opnd? meaning */ | |
| 140 #define END 0 /* End of program or NOMATCH operand. */ | |
| 141 #define BOL 1 /* Match "" at beginning of line. */ | |
| 142 #define EOL 2 /* Match "" at end of line. */ | |
| 143 #define BRANCH 3 /* node Match this alternative, or the | |
| 144 * next... */ | |
| 145 #define BACK 4 /* Match "", "next" ptr points backward. */ | |
| 146 #define EXACTLY 5 /* str Match this string. */ | |
| 147 #define NOTHING 6 /* Match empty string. */ | |
| 148 #define STAR 7 /* node Match this (simple) thing 0 or more | |
| 149 * times. */ | |
| 150 #define PLUS 8 /* node Match this (simple) thing 1 or more | |
| 151 * times. */ | |
| 152 #define MATCH 9 /* node match the operand zero-width */ | |
| 153 #define NOMATCH 10 /* node check for no match with operand */ | |
| 154 #define BEHIND 11 /* node look behind for a match with operand */ | |
| 155 #define NOBEHIND 12 /* node look behind for no match with operand */ | |
| 156 #define SUBPAT 13 /* node match the operand here */ | |
| 157 #define BRACE_SIMPLE 14 /* node Match this (simple) thing between m and | |
| 158 * n times (\{m,n\}). */ | |
| 159 #define BOW 15 /* Match "" after [^a-zA-Z0-9_] */ | |
| 160 #define EOW 16 /* Match "" at [^a-zA-Z0-9_] */ | |
| 161 #define BRACE_LIMITS 17 /* nr nr define the min & max for BRACE_SIMPLE | |
| 162 * and BRACE_COMPLEX. */ | |
| 163 #define NEWL 18 /* Match line-break */ | |
| 164 #define BHPOS 19 /* End position for BEHIND or NOBEHIND */ | |
| 165 | |
| 166 | |
| 167 /* character classes: 20-48 normal, 50-78 include a line-break */ | |
| 168 #define ADD_NL 30 | |
| 169 #define FIRST_NL ANY + ADD_NL | |
| 170 #define ANY 20 /* Match any one character. */ | |
| 171 #define ANYOF 21 /* str Match any character in this string. */ | |
| 172 #define ANYBUT 22 /* str Match any character not in this | |
| 173 * string. */ | |
| 174 #define IDENT 23 /* Match identifier char */ | |
| 175 #define SIDENT 24 /* Match identifier char but no digit */ | |
| 176 #define KWORD 25 /* Match keyword char */ | |
| 177 #define SKWORD 26 /* Match word char but no digit */ | |
| 178 #define FNAME 27 /* Match file name char */ | |
| 179 #define SFNAME 28 /* Match file name char but no digit */ | |
| 180 #define PRINT 29 /* Match printable char */ | |
| 181 #define SPRINT 30 /* Match printable char but no digit */ | |
| 182 #define WHITE 31 /* Match whitespace char */ | |
| 183 #define NWHITE 32 /* Match non-whitespace char */ | |
| 184 #define DIGIT 33 /* Match digit char */ | |
| 185 #define NDIGIT 34 /* Match non-digit char */ | |
| 186 #define HEX 35 /* Match hex char */ | |
| 187 #define NHEX 36 /* Match non-hex char */ | |
| 188 #define OCTAL 37 /* Match octal char */ | |
| 189 #define NOCTAL 38 /* Match non-octal char */ | |
| 190 #define WORD 39 /* Match word char */ | |
| 191 #define NWORD 40 /* Match non-word char */ | |
| 192 #define HEAD 41 /* Match head char */ | |
| 193 #define NHEAD 42 /* Match non-head char */ | |
| 194 #define ALPHA 43 /* Match alpha char */ | |
| 195 #define NALPHA 44 /* Match non-alpha char */ | |
| 196 #define LOWER 45 /* Match lowercase char */ | |
| 197 #define NLOWER 46 /* Match non-lowercase char */ | |
| 198 #define UPPER 47 /* Match uppercase char */ | |
| 199 #define NUPPER 48 /* Match non-uppercase char */ | |
| 200 #define LAST_NL NUPPER + ADD_NL | |
| 201 #define WITH_NL(op) ((op) >= FIRST_NL && (op) <= LAST_NL) | |
| 202 | |
| 203 #define MOPEN 80 /* -89 Mark this point in input as start of | |
| 204 * \( subexpr. MOPEN + 0 marks start of | |
| 205 * match. */ | |
| 206 #define MCLOSE 90 /* -99 Analogous to MOPEN. MCLOSE + 0 marks | |
| 207 * end of match. */ | |
| 208 #define BACKREF 100 /* -109 node Match same string again \1-\9 */ | |
| 209 | |
| 210 #ifdef FEAT_SYN_HL | |
| 211 # define ZOPEN 110 /* -119 Mark this point in input as start of | |
| 212 * \z( subexpr. */ | |
| 213 # define ZCLOSE 120 /* -129 Analogous to ZOPEN. */ | |
| 214 # define ZREF 130 /* -139 node Match external submatch \z1-\z9 */ | |
| 215 #endif | |
| 216 | |
| 217 #define BRACE_COMPLEX 140 /* -149 node Match nodes between m & n times */ | |
| 218 | |
| 219 #define NOPEN 150 /* Mark this point in input as start of | |
| 220 \%( subexpr. */ | |
| 221 #define NCLOSE 151 /* Analogous to NOPEN. */ | |
| 222 | |
| 223 #define MULTIBYTECODE 200 /* mbc Match one multi-byte character */ | |
| 224 #define RE_BOF 201 /* Match "" at beginning of file. */ | |
| 225 #define RE_EOF 202 /* Match "" at end of file. */ | |
| 226 #define CURSOR 203 /* Match location of cursor. */ | |
| 227 | |
| 228 #define RE_LNUM 204 /* nr cmp Match line number */ | |
| 229 #define RE_COL 205 /* nr cmp Match column number */ | |
| 230 #define RE_VCOL 206 /* nr cmp Match virtual column number */ | |
| 231 | |
| 639 | 232 #define RE_MARK 207 /* mark cmp Match mark position */ |
| 233 #define RE_VISUAL 208 /* Match Visual area */ | |
| 234 | |
| 7 | 235 /* |
| 236 * Magic characters have a special meaning, they don't match literally. | |
| 237 * Magic characters are negative. This separates them from literal characters | |
| 238 * (possibly multi-byte). Only ASCII characters can be Magic. | |
| 239 */ | |
| 240 #define Magic(x) ((int)(x) - 256) | |
| 241 #define un_Magic(x) ((x) + 256) | |
| 242 #define is_Magic(x) ((x) < 0) | |
| 243 | |
| 244 static int no_Magic __ARGS((int x)); | |
| 245 static int toggle_Magic __ARGS((int x)); | |
| 246 | |
| 247 static int | |
| 248 no_Magic(x) | |
| 249 int x; | |
| 250 { | |
| 251 if (is_Magic(x)) | |
| 252 return un_Magic(x); | |
| 253 return x; | |
| 254 } | |
| 255 | |
| 256 static int | |
| 257 toggle_Magic(x) | |
| 258 int x; | |
| 259 { | |
| 260 if (is_Magic(x)) | |
| 261 return un_Magic(x); | |
| 262 return Magic(x); | |
| 263 } | |
| 264 | |
| 265 /* | |
| 266 * The first byte of the regexp internal "program" is actually this magic | |
| 267 * number; the start node begins in the second byte. It's used to catch the | |
| 268 * most severe mutilation of the program by the caller. | |
| 269 */ | |
| 270 | |
| 271 #define REGMAGIC 0234 | |
| 272 | |
| 273 /* | |
| 274 * Opcode notes: | |
| 275 * | |
| 276 * BRANCH The set of branches constituting a single choice are hooked | |
| 277 * together with their "next" pointers, since precedence prevents | |
| 278 * anything being concatenated to any individual branch. The | |
| 279 * "next" pointer of the last BRANCH in a choice points to the | |
| 280 * thing following the whole choice. This is also where the | |
| 281 * final "next" pointer of each individual branch points; each | |
| 282 * branch starts with the operand node of a BRANCH node. | |
| 283 * | |
| 284 * BACK Normal "next" pointers all implicitly point forward; BACK | |
| 285 * exists to make loop structures possible. | |
| 286 * | |
| 287 * STAR,PLUS '=', and complex '*' and '+', are implemented as circular | |
| 288 * BRANCH structures using BACK. Simple cases (one character | |
| 289 * per match) are implemented with STAR and PLUS for speed | |
| 290 * and to minimize recursive plunges. | |
| 291 * | |
| 292 * BRACE_LIMITS This is always followed by a BRACE_SIMPLE or BRACE_COMPLEX | |
| 293 * node, and defines the min and max limits to be used for that | |
| 294 * node. | |
| 295 * | |
| 296 * MOPEN,MCLOSE ...are numbered at compile time. | |
| 297 * ZOPEN,ZCLOSE ...ditto | |
| 298 */ | |
| 299 | |
| 300 /* | |
| 301 * A node is one char of opcode followed by two chars of "next" pointer. | |
| 302 * "Next" pointers are stored as two 8-bit bytes, high order first. The | |
| 303 * value is a positive offset from the opcode of the node containing it. | |
| 304 * An operand, if any, simply follows the node. (Note that much of the | |
| 305 * code generation knows about this implicit relationship.) | |
| 306 * | |
| 307 * Using two bytes for the "next" pointer is vast overkill for most things, | |
| 308 * but allows patterns to get big without disasters. | |
| 309 */ | |
| 310 #define OP(p) ((int)*(p)) | |
| 311 #define NEXT(p) (((*((p) + 1) & 0377) << 8) + (*((p) + 2) & 0377)) | |
| 312 #define OPERAND(p) ((p) + 3) | |
| 313 /* Obtain an operand that was stored as four bytes, MSB first. */ | |
| 314 #define OPERAND_MIN(p) (((long)(p)[3] << 24) + ((long)(p)[4] << 16) \ | |
| 315 + ((long)(p)[5] << 8) + (long)(p)[6]) | |
| 316 /* Obtain a second operand stored as four bytes. */ | |
| 317 #define OPERAND_MAX(p) OPERAND_MIN((p) + 4) | |
| 318 /* Obtain a second single-byte operand stored after a four bytes operand. */ | |
| 319 #define OPERAND_CMP(p) (p)[7] | |
| 320 | |
| 321 /* | |
| 322 * Utility definitions. | |
| 323 */ | |
| 324 #define UCHARAT(p) ((int)*(char_u *)(p)) | |
| 325 | |
| 326 /* Used for an error (down from) vim_regcomp(): give the error message, set | |
| 327 * rc_did_emsg and return NULL */ | |
| 653 | 328 #define EMSG_RET_NULL(m) return (EMSG(m), rc_did_emsg = TRUE, (void *)NULL) |
| 329 #define EMSG_M_RET_NULL(m, c) return (EMSG2((m), (c) ? "" : "\\"), rc_did_emsg = TRUE, (void *)NULL) | |
| 308 | 330 #define EMSG_RET_FAIL(m) return (EMSG(m), rc_did_emsg = TRUE, FAIL) |
| 7 | 331 #define EMSG_ONE_RET_NULL EMSG_M_RET_NULL(_("E369: invalid item in %s%%[]"), reg_magic == MAGIC_ALL) |
| 332 | |
| 333 #define MAX_LIMIT (32767L << 16L) | |
| 334 | |
| 335 static int re_multi_type __ARGS((int)); | |
| 336 static int cstrncmp __ARGS((char_u *s1, char_u *s2, int *n)); | |
| 337 static char_u *cstrchr __ARGS((char_u *, int)); | |
| 338 | |
| 339 #ifdef DEBUG | |
| 340 static void regdump __ARGS((char_u *, regprog_T *)); | |
| 341 static char_u *regprop __ARGS((char_u *)); | |
| 342 #endif | |
| 343 | |
| 344 #define NOT_MULTI 0 | |
| 345 #define MULTI_ONE 1 | |
| 346 #define MULTI_MULT 2 | |
| 347 /* | |
| 348 * Return NOT_MULTI if c is not a "multi" operator. | |
| 349 * Return MULTI_ONE if c is a single "multi" operator. | |
| 350 * Return MULTI_MULT if c is a multi "multi" operator. | |
| 351 */ | |
| 352 static int | |
| 353 re_multi_type(c) | |
| 354 int c; | |
| 355 { | |
| 356 if (c == Magic('@') || c == Magic('=') || c == Magic('?')) | |
| 357 return MULTI_ONE; | |
| 358 if (c == Magic('*') || c == Magic('+') || c == Magic('{')) | |
| 359 return MULTI_MULT; | |
| 360 return NOT_MULTI; | |
| 361 } | |
| 362 | |
| 363 /* | |
| 364 * Flags to be passed up and down. | |
| 365 */ | |
| 366 #define HASWIDTH 0x1 /* Known never to match null string. */ | |
| 367 #define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */ | |
| 368 #define SPSTART 0x4 /* Starts with * or +. */ | |
| 369 #define HASNL 0x8 /* Contains some \n. */ | |
| 370 #define HASLOOKBH 0x10 /* Contains "\@<=" or "\@<!". */ | |
| 371 #define WORST 0 /* Worst case. */ | |
| 372 | |
| 373 /* | |
| 374 * When regcode is set to this value, code is not emitted and size is computed | |
| 375 * instead. | |
| 376 */ | |
| 377 #define JUST_CALC_SIZE ((char_u *) -1) | |
| 378 | |
| 359 | 379 static char_u *reg_prev_sub = NULL; |
| 380 | |
| 7 | 381 /* |
| 382 * REGEXP_INRANGE contains all characters which are always special in a [] | |
| 383 * range after '\'. | |
| 384 * REGEXP_ABBR contains all characters which act as abbreviations after '\'. | |
| 385 * These are: | |
| 386 * \n - New line (NL). | |
| 387 * \r - Carriage Return (CR). | |
| 388 * \t - Tab (TAB). | |
| 389 * \e - Escape (ESC). | |
| 390 * \b - Backspace (Ctrl_H). | |
| 24 | 391 * \d - Character code in decimal, eg \d123 |
| 392 * \o - Character code in octal, eg \o80 | |
| 393 * \x - Character code in hex, eg \x4a | |
| 394 * \u - Multibyte character code, eg \u20ac | |
| 395 * \U - Long multibyte character code, eg \U12345678 | |
| 7 | 396 */ |
| 397 static char_u REGEXP_INRANGE[] = "]^-n\\"; | |
| 24 | 398 static char_u REGEXP_ABBR[] = "nrtebdoxuU"; |
| 7 | 399 |
| 400 static int backslash_trans __ARGS((int c)); | |
| 167 | 401 static int get_char_class __ARGS((char_u **pp)); |
| 402 static int get_equi_class __ARGS((char_u **pp)); | |
| 403 static void reg_equi_class __ARGS((int c)); | |
| 404 static int get_coll_element __ARGS((char_u **pp)); | |
| 7 | 405 static char_u *skip_anyof __ARGS((char_u *p)); |
| 406 static void init_class_tab __ARGS((void)); | |
| 407 | |
| 408 /* | |
| 409 * Translate '\x' to its control character, except "\n", which is Magic. | |
| 410 */ | |
| 411 static int | |
| 412 backslash_trans(c) | |
| 413 int c; | |
| 414 { | |
| 415 switch (c) | |
| 416 { | |
| 417 case 'r': return CAR; | |
| 418 case 't': return TAB; | |
| 419 case 'e': return ESC; | |
| 420 case 'b': return BS; | |
| 421 } | |
| 422 return c; | |
| 423 } | |
| 424 | |
| 425 /* | |
| 167 | 426 * Check for a character class name "[:name:]". "pp" points to the '['. |
| 7 | 427 * Returns one of the CLASS_ items. CLASS_NONE means that no item was |
| 428 * recognized. Otherwise "pp" is advanced to after the item. | |
| 429 */ | |
| 430 static int | |
| 167 | 431 get_char_class(pp) |
| 7 | 432 char_u **pp; |
| 433 { | |
| 434 static const char *(class_names[]) = | |
| 435 { | |
| 436 "alnum:]", | |
| 437 #define CLASS_ALNUM 0 | |
| 438 "alpha:]", | |
| 439 #define CLASS_ALPHA 1 | |
| 440 "blank:]", | |
| 441 #define CLASS_BLANK 2 | |
| 442 "cntrl:]", | |
| 443 #define CLASS_CNTRL 3 | |
| 444 "digit:]", | |
| 445 #define CLASS_DIGIT 4 | |
| 446 "graph:]", | |
| 447 #define CLASS_GRAPH 5 | |
| 448 "lower:]", | |
| 449 #define CLASS_LOWER 6 | |
| 450 "print:]", | |
| 451 #define CLASS_PRINT 7 | |
| 452 "punct:]", | |
| 453 #define CLASS_PUNCT 8 | |
| 454 "space:]", | |
| 455 #define CLASS_SPACE 9 | |
| 456 "upper:]", | |
| 457 #define CLASS_UPPER 10 | |
| 458 "xdigit:]", | |
| 459 #define CLASS_XDIGIT 11 | |
| 460 "tab:]", | |
| 461 #define CLASS_TAB 12 | |
| 462 "return:]", | |
| 463 #define CLASS_RETURN 13 | |
| 464 "backspace:]", | |
| 465 #define CLASS_BACKSPACE 14 | |
| 466 "escape:]", | |
| 467 #define CLASS_ESCAPE 15 | |
| 468 }; | |
| 469 #define CLASS_NONE 99 | |
| 470 int i; | |
| 471 | |
| 472 if ((*pp)[1] == ':') | |
| 473 { | |
| 1877 | 474 for (i = 0; i < (int)(sizeof(class_names) / sizeof(*class_names)); ++i) |
| 7 | 475 if (STRNCMP(*pp + 2, class_names[i], STRLEN(class_names[i])) == 0) |
| 476 { | |
| 477 *pp += STRLEN(class_names[i]) + 2; | |
| 478 return i; | |
| 479 } | |
| 480 } | |
| 481 return CLASS_NONE; | |
| 482 } | |
| 483 | |
| 484 /* | |
| 485 * Specific version of character class functions. | |
| 486 * Using a table to keep this fast. | |
| 487 */ | |
| 488 static short class_tab[256]; | |
| 489 | |
| 490 #define RI_DIGIT 0x01 | |
| 491 #define RI_HEX 0x02 | |
| 492 #define RI_OCTAL 0x04 | |
| 493 #define RI_WORD 0x08 | |
| 494 #define RI_HEAD 0x10 | |
| 495 #define RI_ALPHA 0x20 | |
| 496 #define RI_LOWER 0x40 | |
| 497 #define RI_UPPER 0x80 | |
| 498 #define RI_WHITE 0x100 | |
| 499 | |
| 500 static void | |
| 501 init_class_tab() | |
| 502 { | |
| 503 int i; | |
| 504 static int done = FALSE; | |
| 505 | |
| 506 if (done) | |
| 507 return; | |
| 508 | |
| 509 for (i = 0; i < 256; ++i) | |
| 510 { | |
| 511 if (i >= '0' && i <= '7') | |
| 512 class_tab[i] = RI_DIGIT + RI_HEX + RI_OCTAL + RI_WORD; | |
| 513 else if (i >= '8' && i <= '9') | |
| 514 class_tab[i] = RI_DIGIT + RI_HEX + RI_WORD; | |
| 515 else if (i >= 'a' && i <= 'f') | |
| 516 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
| 517 #ifdef EBCDIC | |
| 518 else if ((i >= 'g' && i <= 'i') || (i >= 'j' && i <= 'r') | |
| 519 || (i >= 's' && i <= 'z')) | |
| 520 #else | |
| 521 else if (i >= 'g' && i <= 'z') | |
| 522 #endif | |
| 523 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_LOWER; | |
| 524 else if (i >= 'A' && i <= 'F') | |
| 525 class_tab[i] = RI_HEX + RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
| 526 #ifdef EBCDIC | |
| 527 else if ((i >= 'G' && i <= 'I') || ( i >= 'J' && i <= 'R') | |
| 528 || (i >= 'S' && i <= 'Z')) | |
| 529 #else | |
| 530 else if (i >= 'G' && i <= 'Z') | |
| 531 #endif | |
| 532 class_tab[i] = RI_WORD + RI_HEAD + RI_ALPHA + RI_UPPER; | |
| 533 else if (i == '_') | |
| 534 class_tab[i] = RI_WORD + RI_HEAD; | |
| 535 else | |
| 536 class_tab[i] = 0; | |
| 537 } | |
| 538 class_tab[' '] |= RI_WHITE; | |
| 539 class_tab['\t'] |= RI_WHITE; | |
| 540 done = TRUE; | |
| 541 } | |
| 542 | |
| 543 #ifdef FEAT_MBYTE | |
| 544 # define ri_digit(c) (c < 0x100 && (class_tab[c] & RI_DIGIT)) | |
| 545 # define ri_hex(c) (c < 0x100 && (class_tab[c] & RI_HEX)) | |
| 546 # define ri_octal(c) (c < 0x100 && (class_tab[c] & RI_OCTAL)) | |
| 547 # define ri_word(c) (c < 0x100 && (class_tab[c] & RI_WORD)) | |
| 548 # define ri_head(c) (c < 0x100 && (class_tab[c] & RI_HEAD)) | |
| 549 # define ri_alpha(c) (c < 0x100 && (class_tab[c] & RI_ALPHA)) | |
| 550 # define ri_lower(c) (c < 0x100 && (class_tab[c] & RI_LOWER)) | |
| 551 # define ri_upper(c) (c < 0x100 && (class_tab[c] & RI_UPPER)) | |
| 552 # define ri_white(c) (c < 0x100 && (class_tab[c] & RI_WHITE)) | |
| 553 #else | |
| 554 # define ri_digit(c) (class_tab[c] & RI_DIGIT) | |
| 555 # define ri_hex(c) (class_tab[c] & RI_HEX) | |
| 556 # define ri_octal(c) (class_tab[c] & RI_OCTAL) | |
| 557 # define ri_word(c) (class_tab[c] & RI_WORD) | |
| 558 # define ri_head(c) (class_tab[c] & RI_HEAD) | |
| 559 # define ri_alpha(c) (class_tab[c] & RI_ALPHA) | |
| 560 # define ri_lower(c) (class_tab[c] & RI_LOWER) | |
| 561 # define ri_upper(c) (class_tab[c] & RI_UPPER) | |
| 562 # define ri_white(c) (class_tab[c] & RI_WHITE) | |
| 563 #endif | |
| 564 | |
| 565 /* flags for regflags */ | |
| 566 #define RF_ICASE 1 /* ignore case */ | |
| 567 #define RF_NOICASE 2 /* don't ignore case */ | |
| 568 #define RF_HASNL 4 /* can match a NL */ | |
| 569 #define RF_ICOMBINE 8 /* ignore combining characters */ | |
| 570 #define RF_LOOKBH 16 /* uses "\@<=" or "\@<!" */ | |
| 571 | |
| 572 /* | |
| 573 * Global work variables for vim_regcomp(). | |
| 574 */ | |
| 575 | |
| 576 static char_u *regparse; /* Input-scan pointer. */ | |
| 577 static int prevchr_len; /* byte length of previous char */ | |
| 578 static int num_complex_braces; /* Complex \{...} count */ | |
| 579 static int regnpar; /* () count. */ | |
| 580 #ifdef FEAT_SYN_HL | |
| 581 static int regnzpar; /* \z() count. */ | |
| 582 static int re_has_z; /* \z item detected */ | |
| 583 #endif | |
| 584 static char_u *regcode; /* Code-emit pointer, or JUST_CALC_SIZE */ | |
| 585 static long regsize; /* Code size. */ | |
| 2010 | 586 static int reg_toolong; /* TRUE when offset out of range */ |
| 7 | 587 static char_u had_endbrace[NSUBEXP]; /* flags, TRUE if end of () found */ |
| 588 static unsigned regflags; /* RF_ flags for prog */ | |
| 589 static long brace_min[10]; /* Minimums for complex brace repeats */ | |
| 590 static long brace_max[10]; /* Maximums for complex brace repeats */ | |
| 591 static int brace_count[10]; /* Current counts for complex brace repeats */ | |
| 592 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 593 static int had_eol; /* TRUE when EOL found by vim_regcomp() */ | |
| 594 #endif | |
| 595 static int one_exactly = FALSE; /* only do one char for EXACTLY */ | |
| 596 | |
| 597 static int reg_magic; /* magicness of the pattern: */ | |
| 598 #define MAGIC_NONE 1 /* "\V" very unmagic */ | |
| 599 #define MAGIC_OFF 2 /* "\M" or 'magic' off */ | |
| 600 #define MAGIC_ON 3 /* "\m" or 'magic' */ | |
| 601 #define MAGIC_ALL 4 /* "\v" very magic */ | |
| 602 | |
| 603 static int reg_string; /* matching with a string instead of a buffer | |
| 604 line */ | |
| 481 | 605 static int reg_strict; /* "[abc" is illegal */ |
| 7 | 606 |
| 607 /* | |
| 608 * META contains all characters that may be magic, except '^' and '$'. | |
| 609 */ | |
| 610 | |
| 611 #ifdef EBCDIC | |
| 612 static char_u META[] = "%&()*+.123456789<=>?@ACDFHIKLMOPSUVWX[_acdfhiklmnopsuvwxz{|~"; | |
| 613 #else | |
| 614 /* META[] is used often enough to justify turning it into a table. */ | |
| 615 static char_u META_flags[] = { | |
| 616 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| 617 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, | |
| 618 /* % & ( ) * + . */ | |
| 619 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0, | |
| 620 /* 1 2 3 4 5 6 7 8 9 < = > ? */ | |
| 621 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, | |
| 622 /* @ A C D F H I K L M O */ | |
| 623 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1, | |
| 624 /* P S U V W X Z [ _ */ | |
| 625 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1, | |
| 626 /* a c d f h i k l m n o */ | |
| 627 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, | |
| 628 /* p s u v w x z { | ~ */ | |
| 629 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1 | |
| 630 }; | |
| 631 #endif | |
| 632 | |
| 633 static int curchr; | |
| 634 | |
| 635 /* arguments for reg() */ | |
| 636 #define REG_NOPAREN 0 /* toplevel reg() */ | |
| 637 #define REG_PAREN 1 /* \(\) */ | |
| 638 #define REG_ZPAREN 2 /* \z(\) */ | |
| 639 #define REG_NPAREN 3 /* \%(\) */ | |
| 640 | |
| 641 /* | |
| 642 * Forward declarations for vim_regcomp()'s friends. | |
| 643 */ | |
| 644 static void initchr __ARGS((char_u *)); | |
| 645 static int getchr __ARGS((void)); | |
| 646 static void skipchr_keepstart __ARGS((void)); | |
| 647 static int peekchr __ARGS((void)); | |
| 648 static void skipchr __ARGS((void)); | |
| 649 static void ungetchr __ARGS((void)); | |
| 24 | 650 static int gethexchrs __ARGS((int maxinputlen)); |
| 651 static int getoctchrs __ARGS((void)); | |
| 652 static int getdecchrs __ARGS((void)); | |
| 653 static int coll_get_char __ARGS((void)); | |
| 7 | 654 static void regcomp_start __ARGS((char_u *expr, int flags)); |
| 655 static char_u *reg __ARGS((int, int *)); | |
| 656 static char_u *regbranch __ARGS((int *flagp)); | |
| 657 static char_u *regconcat __ARGS((int *flagp)); | |
| 658 static char_u *regpiece __ARGS((int *)); | |
| 659 static char_u *regatom __ARGS((int *)); | |
| 660 static char_u *regnode __ARGS((int)); | |
| 714 | 661 #ifdef FEAT_MBYTE |
| 662 static int use_multibytecode __ARGS((int c)); | |
| 663 #endif | |
| 7 | 664 static int prog_magic_wrong __ARGS((void)); |
| 665 static char_u *regnext __ARGS((char_u *)); | |
| 666 static void regc __ARGS((int b)); | |
| 667 #ifdef FEAT_MBYTE | |
| 668 static void regmbc __ARGS((int c)); | |
| 2974 | 669 # define REGMBC(x) regmbc(x); |
| 670 # define CASEMBC(x) case x: | |
| 167 | 671 #else |
| 672 # define regmbc(c) regc(c) | |
| 2974 | 673 # define REGMBC(x) |
| 674 # define CASEMBC(x) | |
| 7 | 675 #endif |
| 676 static void reginsert __ARGS((int, char_u *)); | |
| 677 static void reginsert_limits __ARGS((int, long, long, char_u *)); | |
| 678 static char_u *re_put_long __ARGS((char_u *pr, long_u val)); | |
| 679 static int read_limits __ARGS((long *, long *)); | |
| 680 static void regtail __ARGS((char_u *, char_u *)); | |
| 681 static void regoptail __ARGS((char_u *, char_u *)); | |
| 682 | |
| 683 /* | |
| 684 * Return TRUE if compiled regular expression "prog" can match a line break. | |
| 685 */ | |
| 686 int | |
| 687 re_multiline(prog) | |
| 688 regprog_T *prog; | |
| 689 { | |
| 690 return (prog->regflags & RF_HASNL); | |
| 691 } | |
| 692 | |
| 693 /* | |
| 694 * Return TRUE if compiled regular expression "prog" looks before the start | |
| 695 * position (pattern contains "\@<=" or "\@<!"). | |
| 696 */ | |
| 697 int | |
| 698 re_lookbehind(prog) | |
| 699 regprog_T *prog; | |
| 700 { | |
| 701 return (prog->regflags & RF_LOOKBH); | |
| 702 } | |
| 703 | |
| 704 /* | |
| 167 | 705 * Check for an equivalence class name "[=a=]". "pp" points to the '['. |
| 706 * Returns a character representing the class. Zero means that no item was | |
| 707 * recognized. Otherwise "pp" is advanced to after the item. | |
| 708 */ | |
| 709 static int | |
| 710 get_equi_class(pp) | |
| 711 char_u **pp; | |
| 712 { | |
| 713 int c; | |
| 714 int l = 1; | |
| 715 char_u *p = *pp; | |
| 716 | |
| 717 if (p[1] == '=') | |
| 718 { | |
| 719 #ifdef FEAT_MBYTE | |
| 720 if (has_mbyte) | |
| 474 | 721 l = (*mb_ptr2len)(p + 2); |
| 167 | 722 #endif |
| 723 if (p[l + 2] == '=' && p[l + 3] == ']') | |
| 724 { | |
| 725 #ifdef FEAT_MBYTE | |
| 726 if (has_mbyte) | |
| 727 c = mb_ptr2char(p + 2); | |
| 728 else | |
| 729 #endif | |
| 730 c = p[2]; | |
| 731 *pp += l + 4; | |
| 732 return c; | |
| 733 } | |
| 734 } | |
| 735 return 0; | |
| 736 } | |
| 737 | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
738 #ifdef EBCDIC |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
739 /* |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
740 * 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
|
741 */ |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
742 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
|
743 "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
|
744 "C\x68", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
745 "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
|
746 "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
|
747 "N\x69", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
748 "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
|
749 "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
|
750 "Y\xBA", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
751 "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
|
752 "c\x48", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
753 "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
|
754 "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
|
755 "n\x49", |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
756 "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
|
757 "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
|
758 "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
|
759 }; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
760 #endif |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
761 |
| 167 | 762 /* |
| 763 * Produce the bytes for equivalence class "c". | |
| 764 * Currently only handles latin1, latin9 and utf-8. | |
| 765 */ | |
| 766 static void | |
| 767 reg_equi_class(c) | |
| 768 int c; | |
| 769 { | |
| 770 #ifdef FEAT_MBYTE | |
| 771 if (enc_utf8 || STRCMP(p_enc, "latin1") == 0 | |
| 492 | 772 || STRCMP(p_enc, "iso-8859-15") == 0) |
| 167 | 773 #endif |
| 774 { | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
775 #ifdef EBCDIC |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
776 int i; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
777 |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
778 /* 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
|
779 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
|
780 { |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
781 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
|
782 { |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
783 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
|
784 |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
785 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
|
786 regmbc(*p++); |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
787 return; |
|
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
788 } |
|
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 #else |
| 167 | 791 switch (c) |
| 792 { | |
| 236 | 793 case 'A': case '\300': case '\301': case '\302': |
| 2974 | 794 CASEMBC(0x100) CASEMBC(0x102) CASEMBC(0x104) CASEMBC(0x1cd) |
| 795 CASEMBC(0x1de) CASEMBC(0x1e0) CASEMBC(0x1ea2) | |
| 236 | 796 case '\303': case '\304': case '\305': |
| 797 regmbc('A'); regmbc('\300'); regmbc('\301'); | |
| 798 regmbc('\302'); regmbc('\303'); regmbc('\304'); | |
| 799 regmbc('\305'); | |
| 2974 | 800 REGMBC(0x100) REGMBC(0x102) REGMBC(0x104) |
| 801 REGMBC(0x1cd) REGMBC(0x1de) REGMBC(0x1e0) | |
| 802 REGMBC(0x1ea2) | |
| 803 return; | |
| 804 case 'B': CASEMBC(0x1e02) CASEMBC(0x1e06) | |
| 805 regmbc('B'); REGMBC(0x1e02) REGMBC(0x1e06) | |
| 167 | 806 return; |
| 236 | 807 case 'C': case '\307': |
| 2974 | 808 CASEMBC(0x106) CASEMBC(0x108) CASEMBC(0x10a) CASEMBC(0x10c) |
| 236 | 809 regmbc('C'); regmbc('\307'); |
| 2974 | 810 REGMBC(0x106) REGMBC(0x108) REGMBC(0x10a) |
| 811 REGMBC(0x10c) | |
| 812 return; | |
| 813 case 'D': CASEMBC(0x10e) CASEMBC(0x110) CASEMBC(0x1e0a) | |
| 814 CASEMBC(0x1e0e) CASEMBC(0x1e10) | |
| 815 regmbc('D'); REGMBC(0x10e) REGMBC(0x110) | |
| 816 REGMBC(0x1e0a) REGMBC(0x1e0e) REGMBC(0x1e10) | |
| 167 | 817 return; |
| 236 | 818 case 'E': case '\310': case '\311': case '\312': case '\313': |
| 2974 | 819 CASEMBC(0x112) CASEMBC(0x114) CASEMBC(0x116) CASEMBC(0x118) |
| 820 CASEMBC(0x11a) CASEMBC(0x1eba) CASEMBC(0x1ebc) | |
| 236 | 821 regmbc('E'); regmbc('\310'); regmbc('\311'); |
| 822 regmbc('\312'); regmbc('\313'); | |
| 2974 | 823 REGMBC(0x112) REGMBC(0x114) REGMBC(0x116) |
| 824 REGMBC(0x118) REGMBC(0x11a) REGMBC(0x1eba) | |
| 825 REGMBC(0x1ebc) | |
| 826 return; | |
| 827 case 'F': CASEMBC(0x1e1e) | |
| 828 regmbc('F'); REGMBC(0x1e1e) | |
| 829 return; | |
| 830 case 'G': CASEMBC(0x11c) CASEMBC(0x11e) CASEMBC(0x120) | |
| 831 CASEMBC(0x122) CASEMBC(0x1e4) CASEMBC(0x1e6) CASEMBC(0x1f4) | |
| 832 CASEMBC(0x1e20) | |
| 833 regmbc('G'); REGMBC(0x11c) REGMBC(0x11e) | |
| 834 REGMBC(0x120) REGMBC(0x122) REGMBC(0x1e4) | |
| 835 REGMBC(0x1e6) REGMBC(0x1f4) REGMBC(0x1e20) | |
| 836 return; | |
| 837 case 'H': CASEMBC(0x124) CASEMBC(0x126) CASEMBC(0x1e22) | |
| 838 CASEMBC(0x1e26) CASEMBC(0x1e28) | |
| 839 regmbc('H'); REGMBC(0x124) REGMBC(0x126) | |
| 840 REGMBC(0x1e22) REGMBC(0x1e26) REGMBC(0x1e28) | |
| 167 | 841 return; |
| 236 | 842 case 'I': case '\314': case '\315': case '\316': case '\317': |
| 2974 | 843 CASEMBC(0x128) CASEMBC(0x12a) CASEMBC(0x12c) CASEMBC(0x12e) |
| 844 CASEMBC(0x130) CASEMBC(0x1cf) CASEMBC(0x1ec8) | |
| 236 | 845 regmbc('I'); regmbc('\314'); regmbc('\315'); |
| 846 regmbc('\316'); regmbc('\317'); | |
| 2974 | 847 REGMBC(0x128) REGMBC(0x12a) REGMBC(0x12c) |
| 848 REGMBC(0x12e) REGMBC(0x130) REGMBC(0x1cf) | |
| 849 REGMBC(0x1ec8) | |
| 850 return; | |
| 851 case 'J': CASEMBC(0x134) | |
| 852 regmbc('J'); REGMBC(0x134) | |
| 853 return; | |
| 854 case 'K': CASEMBC(0x136) CASEMBC(0x1e8) CASEMBC(0x1e30) | |
| 855 CASEMBC(0x1e34) | |
| 856 regmbc('K'); REGMBC(0x136) REGMBC(0x1e8) | |
| 857 REGMBC(0x1e30) REGMBC(0x1e34) | |
| 858 return; | |
| 859 case 'L': CASEMBC(0x139) CASEMBC(0x13b) CASEMBC(0x13d) | |
| 860 CASEMBC(0x13f) CASEMBC(0x141) CASEMBC(0x1e3a) | |
| 861 regmbc('L'); REGMBC(0x139) REGMBC(0x13b) | |
| 862 REGMBC(0x13d) REGMBC(0x13f) REGMBC(0x141) | |
| 863 REGMBC(0x1e3a) | |
| 864 return; | |
| 865 case 'M': CASEMBC(0x1e3e) CASEMBC(0x1e40) | |
| 866 regmbc('M'); REGMBC(0x1e3e) REGMBC(0x1e40) | |
| 167 | 867 return; |
| 236 | 868 case 'N': case '\321': |
| 2974 | 869 CASEMBC(0x143) CASEMBC(0x145) CASEMBC(0x147) CASEMBC(0x1e44) |
| 870 CASEMBC(0x1e48) | |
| 236 | 871 regmbc('N'); regmbc('\321'); |
| 2974 | 872 REGMBC(0x143) REGMBC(0x145) REGMBC(0x147) |
| 873 REGMBC(0x1e44) REGMBC(0x1e48) | |
| 167 | 874 return; |
| 236 | 875 case 'O': case '\322': case '\323': case '\324': case '\325': |
| 2974 | 876 case '\326': case '\330': |
| 877 CASEMBC(0x14c) CASEMBC(0x14e) CASEMBC(0x150) CASEMBC(0x1a0) | |
| 878 CASEMBC(0x1d1) CASEMBC(0x1ea) CASEMBC(0x1ec) CASEMBC(0x1ece) | |
| 236 | 879 regmbc('O'); regmbc('\322'); regmbc('\323'); |
| 880 regmbc('\324'); regmbc('\325'); regmbc('\326'); | |
| 2974 | 881 regmbc('\330'); |
| 882 REGMBC(0x14c) REGMBC(0x14e) REGMBC(0x150) | |
| 883 REGMBC(0x1a0) REGMBC(0x1d1) REGMBC(0x1ea) | |
| 884 REGMBC(0x1ec) REGMBC(0x1ece) | |
| 885 return; | |
| 886 case 'P': case 0x1e54: case 0x1e56: | |
| 887 regmbc('P'); REGMBC(0x1e54) REGMBC(0x1e56) | |
| 888 return; | |
| 889 case 'R': CASEMBC(0x154) CASEMBC(0x156) CASEMBC(0x158) | |
| 890 CASEMBC(0x1e58) CASEMBC(0x1e5e) | |
| 891 regmbc('R'); REGMBC(0x154) REGMBC(0x156) REGMBC(0x158) | |
| 892 REGMBC(0x1e58) REGMBC(0x1e5e) | |
| 893 return; | |
| 894 case 'S': CASEMBC(0x15a) CASEMBC(0x15c) CASEMBC(0x15e) | |
| 895 CASEMBC(0x160) CASEMBC(0x1e60) | |
| 896 regmbc('S'); REGMBC(0x15a) REGMBC(0x15c) | |
| 897 REGMBC(0x15e) REGMBC(0x160) REGMBC(0x1e60) | |
| 898 return; | |
| 899 case 'T': CASEMBC(0x162) CASEMBC(0x164) CASEMBC(0x166) | |
| 900 CASEMBC(0x1e6a) CASEMBC(0x1e6e) | |
| 901 regmbc('T'); REGMBC(0x162) REGMBC(0x164) | |
| 902 REGMBC(0x166) REGMBC(0x1e6a) REGMBC(0x1e6e) | |
| 167 | 903 return; |
| 236 | 904 case 'U': case '\331': case '\332': case '\333': case '\334': |
| 2974 | 905 CASEMBC(0x168) CASEMBC(0x16a) CASEMBC(0x16c) CASEMBC(0x16e) |
| 906 CASEMBC(0x170) CASEMBC(0x172) CASEMBC(0x1af) CASEMBC(0x1d3) | |
| 907 CASEMBC(0x1ee6) | |
| 236 | 908 regmbc('U'); regmbc('\331'); regmbc('\332'); |
| 909 regmbc('\333'); regmbc('\334'); | |
| 2974 | 910 REGMBC(0x168) REGMBC(0x16a) REGMBC(0x16c) |
| 911 REGMBC(0x16e) REGMBC(0x170) REGMBC(0x172) | |
| 912 REGMBC(0x1af) REGMBC(0x1d3) REGMBC(0x1ee6) | |
| 913 return; | |
| 914 case 'V': CASEMBC(0x1e7c) | |
| 915 regmbc('V'); REGMBC(0x1e7c) | |
| 916 return; | |
| 917 case 'W': CASEMBC(0x174) CASEMBC(0x1e80) CASEMBC(0x1e82) | |
| 918 CASEMBC(0x1e84) CASEMBC(0x1e86) | |
| 919 regmbc('W'); REGMBC(0x174) REGMBC(0x1e80) | |
| 920 REGMBC(0x1e82) REGMBC(0x1e84) REGMBC(0x1e86) | |
| 921 return; | |
| 922 case 'X': CASEMBC(0x1e8a) CASEMBC(0x1e8c) | |
| 923 regmbc('X'); REGMBC(0x1e8a) REGMBC(0x1e8c) | |
| 167 | 924 return; |
| 236 | 925 case 'Y': case '\335': |
| 2974 | 926 CASEMBC(0x176) CASEMBC(0x178) CASEMBC(0x1e8e) CASEMBC(0x1ef2) |
| 927 CASEMBC(0x1ef6) CASEMBC(0x1ef8) | |
| 236 | 928 regmbc('Y'); regmbc('\335'); |
| 2974 | 929 REGMBC(0x176) REGMBC(0x178) REGMBC(0x1e8e) |
| 930 REGMBC(0x1ef2) REGMBC(0x1ef6) REGMBC(0x1ef8) | |
| 931 return; | |
| 932 case 'Z': CASEMBC(0x179) CASEMBC(0x17b) CASEMBC(0x17d) | |
| 933 CASEMBC(0x1b5) CASEMBC(0x1e90) CASEMBC(0x1e94) | |
| 934 regmbc('Z'); REGMBC(0x179) REGMBC(0x17b) | |
| 935 REGMBC(0x17d) REGMBC(0x1b5) REGMBC(0x1e90) | |
| 936 REGMBC(0x1e94) | |
| 167 | 937 return; |
| 236 | 938 case 'a': case '\340': case '\341': case '\342': |
| 939 case '\343': case '\344': case '\345': | |
| 2974 | 940 CASEMBC(0x101) CASEMBC(0x103) CASEMBC(0x105) CASEMBC(0x1ce) |
| 941 CASEMBC(0x1df) CASEMBC(0x1e1) CASEMBC(0x1ea3) | |
| 236 | 942 regmbc('a'); regmbc('\340'); regmbc('\341'); |
| 943 regmbc('\342'); regmbc('\343'); regmbc('\344'); | |
| 944 regmbc('\345'); | |
| 2974 | 945 REGMBC(0x101) REGMBC(0x103) REGMBC(0x105) |
| 946 REGMBC(0x1ce) REGMBC(0x1df) REGMBC(0x1e1) | |
| 947 REGMBC(0x1ea3) | |
| 948 return; | |
| 949 case 'b': CASEMBC(0x1e03) CASEMBC(0x1e07) | |
| 950 regmbc('b'); REGMBC(0x1e03) REGMBC(0x1e07) | |
| 167 | 951 return; |
| 236 | 952 case 'c': case '\347': |
| 2974 | 953 CASEMBC(0x107) CASEMBC(0x109) CASEMBC(0x10b) CASEMBC(0x10d) |
| 236 | 954 regmbc('c'); regmbc('\347'); |
| 2974 | 955 REGMBC(0x107) REGMBC(0x109) REGMBC(0x10b) |
| 956 REGMBC(0x10d) | |
| 957 return; | |
| 958 case 'd': CASEMBC(0x10f) CASEMBC(0x111) CASEMBC(0x1d0b) | |
| 959 CASEMBC(0x1e11) | |
| 960 regmbc('d'); REGMBC(0x10f) REGMBC(0x111) | |
| 961 REGMBC(0x1e0b) REGMBC(0x01e0f) REGMBC(0x1e11) | |
| 167 | 962 return; |
| 236 | 963 case 'e': case '\350': case '\351': case '\352': case '\353': |
| 2974 | 964 CASEMBC(0x113) CASEMBC(0x115) CASEMBC(0x117) CASEMBC(0x119) |
| 965 CASEMBC(0x11b) CASEMBC(0x1ebb) CASEMBC(0x1ebd) | |
| 236 | 966 regmbc('e'); regmbc('\350'); regmbc('\351'); |
| 967 regmbc('\352'); regmbc('\353'); | |
| 2974 | 968 REGMBC(0x113) REGMBC(0x115) REGMBC(0x117) |
| 969 REGMBC(0x119) REGMBC(0x11b) REGMBC(0x1ebb) | |
| 970 REGMBC(0x1ebd) | |
| 971 return; | |
| 972 case 'f': CASEMBC(0x1e1f) | |
| 973 regmbc('f'); REGMBC(0x1e1f) | |
| 974 return; | |
| 975 case 'g': CASEMBC(0x11d) CASEMBC(0x11f) CASEMBC(0x121) | |
| 976 CASEMBC(0x123) CASEMBC(0x1e5) CASEMBC(0x1e7) CASEMBC(0x1f5) | |
| 977 CASEMBC(0x1e21) | |
| 978 regmbc('g'); REGMBC(0x11d) REGMBC(0x11f) | |
| 979 REGMBC(0x121) REGMBC(0x123) REGMBC(0x1e5) | |
| 980 REGMBC(0x1e7) REGMBC(0x1f5) REGMBC(0x1e21) | |
| 981 return; | |
| 982 case 'h': CASEMBC(0x125) CASEMBC(0x127) CASEMBC(0x1e23) | |
| 983 CASEMBC(0x1e27) CASEMBC(0x1e29) CASEMBC(0x1e96) | |
| 984 regmbc('h'); REGMBC(0x125) REGMBC(0x127) | |
| 985 REGMBC(0x1e23) REGMBC(0x1e27) REGMBC(0x1e29) | |
| 986 REGMBC(0x1e96) | |
| 167 | 987 return; |
| 236 | 988 case 'i': case '\354': case '\355': case '\356': case '\357': |
| 2974 | 989 CASEMBC(0x129) CASEMBC(0x12b) CASEMBC(0x12d) CASEMBC(0x12f) |
| 990 CASEMBC(0x1d0) CASEMBC(0x1ec9) | |
| 236 | 991 regmbc('i'); regmbc('\354'); regmbc('\355'); |
| 992 regmbc('\356'); regmbc('\357'); | |
| 2974 | 993 REGMBC(0x129) REGMBC(0x12b) REGMBC(0x12d) |
| 994 REGMBC(0x12f) REGMBC(0x1d0) REGMBC(0x1ec9) | |
| 995 return; | |
| 996 case 'j': CASEMBC(0x135) CASEMBC(0x1f0) | |
| 997 regmbc('j'); REGMBC(0x135) REGMBC(0x1f0) | |
| 998 return; | |
| 999 case 'k': CASEMBC(0x137) CASEMBC(0x1e9) CASEMBC(0x1e31) | |
| 1000 CASEMBC(0x1e35) | |
| 1001 regmbc('k'); REGMBC(0x137) REGMBC(0x1e9) | |
| 1002 REGMBC(0x1e31) REGMBC(0x1e35) | |
| 1003 return; | |
| 1004 case 'l': CASEMBC(0x13a) CASEMBC(0x13c) CASEMBC(0x13e) | |
| 1005 CASEMBC(0x140) CASEMBC(0x142) CASEMBC(0x1e3b) | |
| 1006 regmbc('l'); REGMBC(0x13a) REGMBC(0x13c) | |
| 1007 REGMBC(0x13e) REGMBC(0x140) REGMBC(0x142) | |
| 1008 REGMBC(0x1e3b) | |
| 1009 return; | |
| 1010 case 'm': CASEMBC(0x1e3f) CASEMBC(0x1e41) | |
| 1011 regmbc('m'); REGMBC(0x1e3f) REGMBC(0x1e41) | |
| 167 | 1012 return; |
| 236 | 1013 case 'n': case '\361': |
| 2974 | 1014 CASEMBC(0x144) CASEMBC(0x146) CASEMBC(0x148) CASEMBC(0x149) |
| 1015 CASEMBC(0x1e45) CASEMBC(0x1e49) | |
| 236 | 1016 regmbc('n'); regmbc('\361'); |
| 2974 | 1017 REGMBC(0x144) REGMBC(0x146) REGMBC(0x148) |
| 1018 REGMBC(0x149) REGMBC(0x1e45) REGMBC(0x1e49) | |
| 167 | 1019 return; |
| 236 | 1020 case 'o': case '\362': case '\363': case '\364': case '\365': |
| 2974 | 1021 case '\366': case '\370': |
| 1022 CASEMBC(0x14d) CASEMBC(0x14f) CASEMBC(0x151) CASEMBC(0x1a1) | |
| 1023 CASEMBC(0x1d2) CASEMBC(0x1eb) CASEMBC(0x1ed) CASEMBC(0x1ecf) | |
| 236 | 1024 regmbc('o'); regmbc('\362'); regmbc('\363'); |
| 1025 regmbc('\364'); regmbc('\365'); regmbc('\366'); | |
| 2974 | 1026 regmbc('\370'); |
| 1027 REGMBC(0x14d) REGMBC(0x14f) REGMBC(0x151) | |
| 1028 REGMBC(0x1a1) REGMBC(0x1d2) REGMBC(0x1eb) | |
| 1029 REGMBC(0x1ed) REGMBC(0x1ecf) | |
| 1030 return; | |
| 1031 case 'p': CASEMBC(0x1e55) CASEMBC(0x1e57) | |
| 1032 regmbc('p'); REGMBC(0x1e55) REGMBC(0x1e57) | |
| 1033 return; | |
| 1034 case 'r': CASEMBC(0x155) CASEMBC(0x157) CASEMBC(0x159) | |
| 1035 CASEMBC(0x1e59) CASEMBC(0x1e5f) | |
| 1036 regmbc('r'); REGMBC(0x155) REGMBC(0x157) REGMBC(0x159) | |
| 1037 REGMBC(0x1e59) REGMBC(0x1e5f) | |
| 1038 return; | |
| 1039 case 's': CASEMBC(0x15b) CASEMBC(0x15d) CASEMBC(0x15f) | |
| 1040 CASEMBC(0x161) CASEMBC(0x1e61) | |
| 1041 regmbc('s'); REGMBC(0x15b) REGMBC(0x15d) | |
| 1042 REGMBC(0x15f) REGMBC(0x161) REGMBC(0x1e61) | |
| 1043 return; | |
| 1044 case 't': CASEMBC(0x163) CASEMBC(0x165) CASEMBC(0x167) | |
| 1045 CASEMBC(0x1e6b) CASEMBC(0x1e6f) CASEMBC(0x1e97) | |
| 1046 regmbc('t'); REGMBC(0x163) REGMBC(0x165) REGMBC(0x167) | |
| 1047 REGMBC(0x1e6b) REGMBC(0x1e6f) REGMBC(0x1e97) | |
| 167 | 1048 return; |
| 236 | 1049 case 'u': case '\371': case '\372': case '\373': case '\374': |
| 2974 | 1050 CASEMBC(0x169) CASEMBC(0x16b) CASEMBC(0x16d) CASEMBC(0x16f) |
| 1051 CASEMBC(0x171) CASEMBC(0x173) CASEMBC(0x1b0) CASEMBC(0x1d4) | |
| 1052 CASEMBC(0x1ee7) | |
| 236 | 1053 regmbc('u'); regmbc('\371'); regmbc('\372'); |
| 1054 regmbc('\373'); regmbc('\374'); | |
| 2974 | 1055 REGMBC(0x169) REGMBC(0x16b) REGMBC(0x16d) |
| 1056 REGMBC(0x16f) REGMBC(0x171) REGMBC(0x173) | |
| 1057 REGMBC(0x1b0) REGMBC(0x1d4) REGMBC(0x1ee7) | |
| 1058 return; | |
| 1059 case 'v': CASEMBC(0x1e7d) | |
| 1060 regmbc('v'); REGMBC(0x1e7d) | |
| 1061 return; | |
| 1062 case 'w': CASEMBC(0x175) CASEMBC(0x1e81) CASEMBC(0x1e83) | |
| 1063 CASEMBC(0x1e85) CASEMBC(0x1e87) CASEMBC(0x1e98) | |
| 1064 regmbc('w'); REGMBC(0x175) REGMBC(0x1e81) | |
| 1065 REGMBC(0x1e83) REGMBC(0x1e85) REGMBC(0x1e87) | |
| 1066 REGMBC(0x1e98) | |
| 1067 return; | |
| 1068 case 'x': CASEMBC(0x1e8b) CASEMBC(0x1e8d) | |
| 1069 regmbc('x'); REGMBC(0x1e8b) REGMBC(0x1e8d) | |
| 167 | 1070 return; |
| 236 | 1071 case 'y': case '\375': case '\377': |
| 2974 | 1072 CASEMBC(0x177) CASEMBC(0x1e8f) CASEMBC(0x1e99) |
| 1073 CASEMBC(0x1ef3) CASEMBC(0x1ef7) CASEMBC(0x1ef9) | |
| 236 | 1074 regmbc('y'); regmbc('\375'); regmbc('\377'); |
| 2974 | 1075 REGMBC(0x177) REGMBC(0x1e8f) REGMBC(0x1e99) |
| 1076 REGMBC(0x1ef3) REGMBC(0x1ef7) REGMBC(0x1ef9) | |
| 1077 return; | |
| 1078 case 'z': CASEMBC(0x17a) CASEMBC(0x17c) CASEMBC(0x17e) | |
| 1079 CASEMBC(0x1b6) CASEMBC(0x1e91) CASEMBC(0x1e95) | |
| 1080 regmbc('z'); REGMBC(0x17a) REGMBC(0x17c) | |
| 1081 REGMBC(0x17e) REGMBC(0x1b6) REGMBC(0x1e91) | |
| 1082 REGMBC(0x1e95) | |
| 167 | 1083 return; |
| 1084 } | |
|
2247
c40cd9aad546
Add patch to improve support of z/OS (OS/390). (Ralf Schandl)
Bram Moolenaar <bram@vim.org>
parents:
2173
diff
changeset
|
1085 #endif |
| 167 | 1086 } |
| 1087 regmbc(c); | |
| 1088 } | |
| 1089 | |
| 1090 /* | |
| 1091 * Check for a collating element "[.a.]". "pp" points to the '['. | |
| 1092 * Returns a character. Zero means that no item was recognized. Otherwise | |
| 1093 * "pp" is advanced to after the item. | |
| 1094 * Currently only single characters are recognized! | |
| 1095 */ | |
| 1096 static int | |
| 1097 get_coll_element(pp) | |
| 1098 char_u **pp; | |
| 1099 { | |
| 1100 int c; | |
| 1101 int l = 1; | |
| 1102 char_u *p = *pp; | |
| 1103 | |
| 1104 if (p[1] == '.') | |
| 1105 { | |
| 1106 #ifdef FEAT_MBYTE | |
| 1107 if (has_mbyte) | |
| 474 | 1108 l = (*mb_ptr2len)(p + 2); |
| 167 | 1109 #endif |
| 1110 if (p[l + 2] == '.' && p[l + 3] == ']') | |
| 1111 { | |
| 1112 #ifdef FEAT_MBYTE | |
| 1113 if (has_mbyte) | |
| 1114 c = mb_ptr2char(p + 2); | |
| 1115 else | |
| 1116 #endif | |
| 1117 c = p[2]; | |
| 1118 *pp += l + 4; | |
| 1119 return c; | |
| 1120 } | |
| 1121 } | |
| 1122 return 0; | |
| 1123 } | |
| 1124 | |
| 1125 | |
| 1126 /* | |
| 1127 * Skip over a "[]" range. | |
| 1128 * "p" must point to the character after the '['. | |
| 1129 * The returned pointer is on the matching ']', or the terminating NUL. | |
| 1130 */ | |
| 1131 static char_u * | |
| 1132 skip_anyof(p) | |
| 1133 char_u *p; | |
| 1134 { | |
| 1135 int cpo_lit; /* 'cpoptions' contains 'l' flag */ | |
| 1136 int cpo_bsl; /* 'cpoptions' contains '\' flag */ | |
| 1137 #ifdef FEAT_MBYTE | |
| 1138 int l; | |
| 1139 #endif | |
| 1140 | |
| 410 | 1141 cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; |
| 1142 cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; | |
| 167 | 1143 |
| 1144 if (*p == '^') /* Complement of range. */ | |
| 1145 ++p; | |
| 1146 if (*p == ']' || *p == '-') | |
| 1147 ++p; | |
| 1148 while (*p != NUL && *p != ']') | |
| 1149 { | |
| 1150 #ifdef FEAT_MBYTE | |
| 474 | 1151 if (has_mbyte && (l = (*mb_ptr2len)(p)) > 1) |
| 167 | 1152 p += l; |
| 1153 else | |
| 1154 #endif | |
| 1155 if (*p == '-') | |
| 1156 { | |
| 1157 ++p; | |
| 1158 if (*p != ']' && *p != NUL) | |
| 1159 mb_ptr_adv(p); | |
| 1160 } | |
| 1161 else if (*p == '\\' | |
| 1162 && !cpo_bsl | |
| 1163 && (vim_strchr(REGEXP_INRANGE, p[1]) != NULL | |
| 1164 || (!cpo_lit && vim_strchr(REGEXP_ABBR, p[1]) != NULL))) | |
| 1165 p += 2; | |
| 1166 else if (*p == '[') | |
| 1167 { | |
| 1168 if (get_char_class(&p) == CLASS_NONE | |
| 1169 && get_equi_class(&p) == 0 | |
| 1170 && get_coll_element(&p) == 0) | |
| 1171 ++p; /* It was not a class name */ | |
| 1172 } | |
| 1173 else | |
| 1174 ++p; | |
| 1175 } | |
| 1176 | |
| 1177 return p; | |
| 1178 } | |
| 1179 | |
| 1180 /* | |
| 7 | 1181 * Skip past regular expression. |
| 153 | 1182 * Stop at end of "startp" or where "dirc" is found ('/', '?', etc). |
| 7 | 1183 * Take care of characters with a backslash in front of it. |
| 1184 * Skip strings inside [ and ]. | |
| 1185 * When "newp" is not NULL and "dirc" is '?', make an allocated copy of the | |
| 1186 * expression and change "\?" to "?". If "*newp" is not NULL the expression | |
| 1187 * is changed in-place. | |
| 1188 */ | |
| 1189 char_u * | |
| 1190 skip_regexp(startp, dirc, magic, newp) | |
| 1191 char_u *startp; | |
| 1192 int dirc; | |
| 1193 int magic; | |
| 1194 char_u **newp; | |
| 1195 { | |
| 1196 int mymagic; | |
| 1197 char_u *p = startp; | |
| 1198 | |
| 1199 if (magic) | |
| 1200 mymagic = MAGIC_ON; | |
| 1201 else | |
| 1202 mymagic = MAGIC_OFF; | |
| 1203 | |
| 39 | 1204 for (; p[0] != NUL; mb_ptr_adv(p)) |
| 7 | 1205 { |
| 1206 if (p[0] == dirc) /* found end of regexp */ | |
| 1207 break; | |
| 1208 if ((p[0] == '[' && mymagic >= MAGIC_ON) | |
| 1209 || (p[0] == '\\' && p[1] == '[' && mymagic <= MAGIC_OFF)) | |
| 1210 { | |
| 1211 p = skip_anyof(p + 1); | |
| 1212 if (p[0] == NUL) | |
| 1213 break; | |
| 1214 } | |
| 1215 else if (p[0] == '\\' && p[1] != NUL) | |
| 1216 { | |
| 1217 if (dirc == '?' && newp != NULL && p[1] == '?') | |
| 1218 { | |
| 1219 /* change "\?" to "?", make a copy first. */ | |
| 1220 if (*newp == NULL) | |
| 1221 { | |
| 1222 *newp = vim_strsave(startp); | |
| 1223 if (*newp != NULL) | |
| 1224 p = *newp + (p - startp); | |
| 1225 } | |
| 1226 if (*newp != NULL) | |
| 1621 | 1227 STRMOVE(p, p + 1); |
| 7 | 1228 else |
| 1229 ++p; | |
| 1230 } | |
| 1231 else | |
| 1232 ++p; /* skip next character */ | |
| 1233 if (*p == 'v') | |
| 1234 mymagic = MAGIC_ALL; | |
| 1235 else if (*p == 'V') | |
| 1236 mymagic = MAGIC_NONE; | |
| 1237 } | |
| 1238 } | |
| 1239 return p; | |
| 1240 } | |
| 1241 | |
| 1242 /* | |
| 41 | 1243 * vim_regcomp() - compile a regular expression into internal code |
| 1244 * Returns the program in allocated space. Returns NULL for an error. | |
| 7 | 1245 * |
| 1246 * We can't allocate space until we know how big the compiled form will be, | |
| 1247 * but we can't compile it (and thus know how big it is) until we've got a | |
| 1248 * place to put the code. So we cheat: we compile it twice, once with code | |
| 1249 * generation turned off and size counting turned on, and once "for real". | |
| 1250 * This also means that we don't allocate space until we are sure that the | |
| 1251 * thing really will compile successfully, and we never have to move the | |
| 1252 * code and thus invalidate pointers into it. (Note that it has to be in | |
| 1253 * one piece because vim_free() must be able to free it all.) | |
| 1254 * | |
| 1255 * Whether upper/lower case is to be ignored is decided when executing the | |
| 1256 * program, it does not matter here. | |
| 1257 * | |
| 1258 * Beware that the optimization-preparation code in here knows about some | |
| 1259 * of the structure of the compiled regexp. | |
| 1260 * "re_flags": RE_MAGIC and/or RE_STRING. | |
| 1261 */ | |
| 1262 regprog_T * | |
| 1263 vim_regcomp(expr, re_flags) | |
| 1264 char_u *expr; | |
| 1265 int re_flags; | |
| 1266 { | |
| 1267 regprog_T *r; | |
| 1268 char_u *scan; | |
| 1269 char_u *longest; | |
| 1270 int len; | |
| 1271 int flags; | |
| 1272 | |
| 1273 if (expr == NULL) | |
| 1274 EMSG_RET_NULL(_(e_null)); | |
| 1275 | |
| 1276 init_class_tab(); | |
| 1277 | |
| 1278 /* | |
| 1279 * First pass: determine size, legality. | |
| 1280 */ | |
| 1281 regcomp_start(expr, re_flags); | |
| 1282 regcode = JUST_CALC_SIZE; | |
| 1283 regc(REGMAGIC); | |
| 1284 if (reg(REG_NOPAREN, &flags) == NULL) | |
| 1285 return NULL; | |
| 1286 | |
| 1287 /* Small enough for pointer-storage convention? */ | |
| 1288 #ifdef SMALL_MALLOC /* 16 bit storage allocation */ | |
| 1289 if (regsize >= 65536L - 256L) | |
| 1290 EMSG_RET_NULL(_("E339: Pattern too long")); | |
| 1291 #endif | |
| 1292 | |
| 1293 /* Allocate space. */ | |
| 1294 r = (regprog_T *)lalloc(sizeof(regprog_T) + regsize, TRUE); | |
| 1295 if (r == NULL) | |
| 1296 return NULL; | |
| 1297 | |
| 1298 /* | |
| 1299 * Second pass: emit code. | |
| 1300 */ | |
| 1301 regcomp_start(expr, re_flags); | |
| 1302 regcode = r->program; | |
| 1303 regc(REGMAGIC); | |
| 2010 | 1304 if (reg(REG_NOPAREN, &flags) == NULL || reg_toolong) |
| 7 | 1305 { |
| 1306 vim_free(r); | |
| 2010 | 1307 if (reg_toolong) |
| 1308 EMSG_RET_NULL(_("E339: Pattern too long")); | |
| 7 | 1309 return NULL; |
| 1310 } | |
| 1311 | |
| 1312 /* Dig out information for optimizations. */ | |
| 1313 r->regstart = NUL; /* Worst-case defaults. */ | |
| 1314 r->reganch = 0; | |
| 1315 r->regmust = NULL; | |
| 1316 r->regmlen = 0; | |
| 1317 r->regflags = regflags; | |
| 1318 if (flags & HASNL) | |
| 1319 r->regflags |= RF_HASNL; | |
| 1320 if (flags & HASLOOKBH) | |
| 1321 r->regflags |= RF_LOOKBH; | |
| 1322 #ifdef FEAT_SYN_HL | |
| 1323 /* Remember whether this pattern has any \z specials in it. */ | |
| 1324 r->reghasz = re_has_z; | |
| 1325 #endif | |
| 1326 scan = r->program + 1; /* First BRANCH. */ | |
| 1327 if (OP(regnext(scan)) == END) /* Only one top-level choice. */ | |
| 1328 { | |
| 1329 scan = OPERAND(scan); | |
| 1330 | |
| 1331 /* Starting-point info. */ | |
| 1332 if (OP(scan) == BOL || OP(scan) == RE_BOF) | |
| 1333 { | |
| 1334 r->reganch++; | |
| 1335 scan = regnext(scan); | |
| 1336 } | |
| 1337 | |
| 1338 if (OP(scan) == EXACTLY) | |
| 1339 { | |
| 1340 #ifdef FEAT_MBYTE | |
| 1341 if (has_mbyte) | |
| 1342 r->regstart = (*mb_ptr2char)(OPERAND(scan)); | |
| 1343 else | |
| 1344 #endif | |
| 1345 r->regstart = *OPERAND(scan); | |
| 1346 } | |
| 1347 else if ((OP(scan) == BOW | |
| 1348 || OP(scan) == EOW | |
| 1349 || OP(scan) == NOTHING | |
| 1350 || OP(scan) == MOPEN + 0 || OP(scan) == NOPEN | |
| 1351 || OP(scan) == MCLOSE + 0 || OP(scan) == NCLOSE) | |
| 1352 && OP(regnext(scan)) == EXACTLY) | |
| 1353 { | |
| 1354 #ifdef FEAT_MBYTE | |
| 1355 if (has_mbyte) | |
| 1356 r->regstart = (*mb_ptr2char)(OPERAND(regnext(scan))); | |
| 1357 else | |
| 1358 #endif | |
| 1359 r->regstart = *OPERAND(regnext(scan)); | |
| 1360 } | |
| 1361 | |
| 1362 /* | |
| 1363 * If there's something expensive in the r.e., find the longest | |
| 1364 * literal string that must appear and make it the regmust. Resolve | |
| 1365 * ties in favor of later strings, since the regstart check works | |
| 1366 * with the beginning of the r.e. and avoiding duplication | |
| 1367 * strengthens checking. Not a strong reason, but sufficient in the | |
| 1368 * absence of others. | |
| 1369 */ | |
| 1370 /* | |
| 1371 * When the r.e. starts with BOW, it is faster to look for a regmust | |
| 1372 * first. Used a lot for "#" and "*" commands. (Added by mool). | |
| 1373 */ | |
| 1374 if ((flags & SPSTART || OP(scan) == BOW || OP(scan) == EOW) | |
| 1375 && !(flags & HASNL)) | |
| 1376 { | |
| 1377 longest = NULL; | |
| 1378 len = 0; | |
| 1379 for (; scan != NULL; scan = regnext(scan)) | |
| 1380 if (OP(scan) == EXACTLY && STRLEN(OPERAND(scan)) >= (size_t)len) | |
| 1381 { | |
| 1382 longest = OPERAND(scan); | |
| 1383 len = (int)STRLEN(OPERAND(scan)); | |
| 1384 } | |
| 1385 r->regmust = longest; | |
| 1386 r->regmlen = len; | |
| 1387 } | |
| 1388 } | |
| 1389 #ifdef DEBUG | |
| 1390 regdump(expr, r); | |
| 1391 #endif | |
| 1392 return r; | |
| 1393 } | |
| 1394 | |
| 1395 /* | |
| 1396 * Setup to parse the regexp. Used once to get the length and once to do it. | |
| 1397 */ | |
| 1398 static void | |
| 1399 regcomp_start(expr, re_flags) | |
| 1400 char_u *expr; | |
| 1401 int re_flags; /* see vim_regcomp() */ | |
| 1402 { | |
| 1403 initchr(expr); | |
| 1404 if (re_flags & RE_MAGIC) | |
| 1405 reg_magic = MAGIC_ON; | |
| 1406 else | |
| 1407 reg_magic = MAGIC_OFF; | |
| 1408 reg_string = (re_flags & RE_STRING); | |
| 481 | 1409 reg_strict = (re_flags & RE_STRICT); |
| 7 | 1410 |
| 1411 num_complex_braces = 0; | |
| 1412 regnpar = 1; | |
| 1413 vim_memset(had_endbrace, 0, sizeof(had_endbrace)); | |
| 1414 #ifdef FEAT_SYN_HL | |
| 1415 regnzpar = 1; | |
| 1416 re_has_z = 0; | |
| 1417 #endif | |
| 1418 regsize = 0L; | |
| 2010 | 1419 reg_toolong = FALSE; |
| 7 | 1420 regflags = 0; |
| 1421 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1422 had_eol = FALSE; | |
| 1423 #endif | |
| 1424 } | |
| 1425 | |
| 1426 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1427 /* | |
| 1428 * Check if during the previous call to vim_regcomp the EOL item "$" has been | |
| 1429 * found. This is messy, but it works fine. | |
| 1430 */ | |
| 1431 int | |
| 1432 vim_regcomp_had_eol() | |
| 1433 { | |
| 1434 return had_eol; | |
| 1435 } | |
| 1436 #endif | |
| 1437 | |
| 1438 /* | |
| 1439 * reg - regular expression, i.e. main body or parenthesized thing | |
| 1440 * | |
| 1441 * Caller must absorb opening parenthesis. | |
| 1442 * | |
| 1443 * Combining parenthesis handling with the base level of regular expression | |
| 1444 * is a trifle forced, but the need to tie the tails of the branches to what | |
| 1445 * follows makes it hard to avoid. | |
| 1446 */ | |
| 1447 static char_u * | |
| 1448 reg(paren, flagp) | |
| 1449 int paren; /* REG_NOPAREN, REG_PAREN, REG_NPAREN or REG_ZPAREN */ | |
| 1450 int *flagp; | |
| 1451 { | |
| 1452 char_u *ret; | |
| 1453 char_u *br; | |
| 1454 char_u *ender; | |
| 1455 int parno = 0; | |
| 1456 int flags; | |
| 1457 | |
| 1458 *flagp = HASWIDTH; /* Tentatively. */ | |
| 1459 | |
| 1460 #ifdef FEAT_SYN_HL | |
| 1461 if (paren == REG_ZPAREN) | |
| 1462 { | |
| 1463 /* Make a ZOPEN node. */ | |
| 1464 if (regnzpar >= NSUBEXP) | |
| 1465 EMSG_RET_NULL(_("E50: Too many \\z(")); | |
| 1466 parno = regnzpar; | |
| 1467 regnzpar++; | |
| 1468 ret = regnode(ZOPEN + parno); | |
| 1469 } | |
| 1470 else | |
| 1471 #endif | |
| 1472 if (paren == REG_PAREN) | |
| 1473 { | |
| 1474 /* Make a MOPEN node. */ | |
| 1475 if (regnpar >= NSUBEXP) | |
| 1476 EMSG_M_RET_NULL(_("E51: Too many %s("), reg_magic == MAGIC_ALL); | |
| 1477 parno = regnpar; | |
| 1478 ++regnpar; | |
| 1479 ret = regnode(MOPEN + parno); | |
| 1480 } | |
| 1481 else if (paren == REG_NPAREN) | |
| 1482 { | |
| 1483 /* Make a NOPEN node. */ | |
| 1484 ret = regnode(NOPEN); | |
| 1485 } | |
| 1486 else | |
| 1487 ret = NULL; | |
| 1488 | |
| 1489 /* Pick up the branches, linking them together. */ | |
| 1490 br = regbranch(&flags); | |
| 1491 if (br == NULL) | |
| 1492 return NULL; | |
| 1493 if (ret != NULL) | |
| 1494 regtail(ret, br); /* [MZ]OPEN -> first. */ | |
| 1495 else | |
| 1496 ret = br; | |
| 1497 /* If one of the branches can be zero-width, the whole thing can. | |
| 1498 * If one of the branches has * at start or matches a line-break, the | |
| 1499 * whole thing can. */ | |
| 1500 if (!(flags & HASWIDTH)) | |
| 1501 *flagp &= ~HASWIDTH; | |
| 1502 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1503 while (peekchr() == Magic('|')) | |
| 1504 { | |
| 1505 skipchr(); | |
| 1506 br = regbranch(&flags); | |
| 2010 | 1507 if (br == NULL || reg_toolong) |
| 7 | 1508 return NULL; |
| 1509 regtail(ret, br); /* BRANCH -> BRANCH. */ | |
| 1510 if (!(flags & HASWIDTH)) | |
| 1511 *flagp &= ~HASWIDTH; | |
| 1512 *flagp |= flags & (SPSTART | HASNL | HASLOOKBH); | |
| 1513 } | |
| 1514 | |
| 1515 /* Make a closing node, and hook it on the end. */ | |
| 1516 ender = regnode( | |
| 1517 #ifdef FEAT_SYN_HL | |
| 1518 paren == REG_ZPAREN ? ZCLOSE + parno : | |
| 1519 #endif | |
| 1520 paren == REG_PAREN ? MCLOSE + parno : | |
| 1521 paren == REG_NPAREN ? NCLOSE : END); | |
| 1522 regtail(ret, ender); | |
| 1523 | |
| 1524 /* Hook the tails of the branches to the closing node. */ | |
| 1525 for (br = ret; br != NULL; br = regnext(br)) | |
| 1526 regoptail(br, ender); | |
| 1527 | |
| 1528 /* Check for proper termination. */ | |
| 1529 if (paren != REG_NOPAREN && getchr() != Magic(')')) | |
| 1530 { | |
| 1531 #ifdef FEAT_SYN_HL | |
| 1532 if (paren == REG_ZPAREN) | |
| 308 | 1533 EMSG_RET_NULL(_("E52: Unmatched \\z(")); |
| 7 | 1534 else |
| 1535 #endif | |
| 1536 if (paren == REG_NPAREN) | |
| 308 | 1537 EMSG_M_RET_NULL(_("E53: Unmatched %s%%("), reg_magic == MAGIC_ALL); |
| 7 | 1538 else |
| 308 | 1539 EMSG_M_RET_NULL(_("E54: Unmatched %s("), reg_magic == MAGIC_ALL); |
| 7 | 1540 } |
| 1541 else if (paren == REG_NOPAREN && peekchr() != NUL) | |
| 1542 { | |
| 1543 if (curchr == Magic(')')) | |
| 308 | 1544 EMSG_M_RET_NULL(_("E55: Unmatched %s)"), reg_magic == MAGIC_ALL); |
| 7 | 1545 else |
| 308 | 1546 EMSG_RET_NULL(_(e_trailing)); /* "Can't happen". */ |
| 7 | 1547 /* NOTREACHED */ |
| 1548 } | |
| 1549 /* | |
| 1550 * Here we set the flag allowing back references to this set of | |
| 1551 * parentheses. | |
| 1552 */ | |
| 1553 if (paren == REG_PAREN) | |
| 1554 had_endbrace[parno] = TRUE; /* have seen the close paren */ | |
| 1555 return ret; | |
| 1556 } | |
| 1557 | |
| 1558 /* | |
| 1468 | 1559 * Handle one alternative of an | operator. |
| 7 | 1560 * Implements the & operator. |
| 1561 */ | |
| 1562 static char_u * | |
| 1563 regbranch(flagp) | |
| 1564 int *flagp; | |
| 1565 { | |
| 1566 char_u *ret; | |
| 1567 char_u *chain = NULL; | |
| 1568 char_u *latest; | |
| 1569 int flags; | |
| 1570 | |
| 1571 *flagp = WORST | HASNL; /* Tentatively. */ | |
| 1572 | |
| 1573 ret = regnode(BRANCH); | |
| 1574 for (;;) | |
| 1575 { | |
| 1576 latest = regconcat(&flags); | |
| 1577 if (latest == NULL) | |
| 1578 return NULL; | |
| 1579 /* If one of the branches has width, the whole thing has. If one of | |
| 1580 * the branches anchors at start-of-line, the whole thing does. | |
| 1581 * If one of the branches uses look-behind, the whole thing does. */ | |
| 1582 *flagp |= flags & (HASWIDTH | SPSTART | HASLOOKBH); | |
| 1583 /* If one of the branches doesn't match a line-break, the whole thing | |
| 1584 * doesn't. */ | |
| 1585 *flagp &= ~HASNL | (flags & HASNL); | |
| 1586 if (chain != NULL) | |
| 1587 regtail(chain, latest); | |
| 1588 if (peekchr() != Magic('&')) | |
| 1589 break; | |
| 1590 skipchr(); | |
| 1591 regtail(latest, regnode(END)); /* operand ends */ | |
| 2010 | 1592 if (reg_toolong) |
| 1593 break; | |
| 7 | 1594 reginsert(MATCH, latest); |
| 1595 chain = latest; | |
| 1596 } | |
| 1597 | |
| 1598 return ret; | |
| 1599 } | |
| 1600 | |
| 1601 /* | |
| 1468 | 1602 * Handle one alternative of an | or & operator. |
| 7 | 1603 * Implements the concatenation operator. |
| 1604 */ | |
| 1605 static char_u * | |
| 1606 regconcat(flagp) | |
| 1607 int *flagp; | |
| 1608 { | |
| 1609 char_u *first = NULL; | |
| 1610 char_u *chain = NULL; | |
| 1611 char_u *latest; | |
| 1612 int flags; | |
| 1613 int cont = TRUE; | |
| 1614 | |
| 1615 *flagp = WORST; /* Tentatively. */ | |
| 1616 | |
| 1617 while (cont) | |
| 1618 { | |
| 1619 switch (peekchr()) | |
| 1620 { | |
| 1621 case NUL: | |
| 1622 case Magic('|'): | |
| 1623 case Magic('&'): | |
| 1624 case Magic(')'): | |
| 1625 cont = FALSE; | |
| 1626 break; | |
| 1627 case Magic('Z'): | |
| 1628 #ifdef FEAT_MBYTE | |
| 1629 regflags |= RF_ICOMBINE; | |
| 1630 #endif | |
| 1631 skipchr_keepstart(); | |
| 1632 break; | |
| 1633 case Magic('c'): | |
| 1634 regflags |= RF_ICASE; | |
| 1635 skipchr_keepstart(); | |
| 1636 break; | |
| 1637 case Magic('C'): | |
| 1638 regflags |= RF_NOICASE; | |
| 1639 skipchr_keepstart(); | |
| 1640 break; | |
| 1641 case Magic('v'): | |
| 1642 reg_magic = MAGIC_ALL; | |
| 1643 skipchr_keepstart(); | |
| 1644 curchr = -1; | |
| 1645 break; | |
| 1646 case Magic('m'): | |
| 1647 reg_magic = MAGIC_ON; | |
| 1648 skipchr_keepstart(); | |
| 1649 curchr = -1; | |
| 1650 break; | |
| 1651 case Magic('M'): | |
| 1652 reg_magic = MAGIC_OFF; | |
| 1653 skipchr_keepstart(); | |
| 1654 curchr = -1; | |
| 1655 break; | |
| 1656 case Magic('V'): | |
| 1657 reg_magic = MAGIC_NONE; | |
| 1658 skipchr_keepstart(); | |
| 1659 curchr = -1; | |
| 1660 break; | |
| 1661 default: | |
| 1662 latest = regpiece(&flags); | |
| 2010 | 1663 if (latest == NULL || reg_toolong) |
| 7 | 1664 return NULL; |
| 1665 *flagp |= flags & (HASWIDTH | HASNL | HASLOOKBH); | |
| 1666 if (chain == NULL) /* First piece. */ | |
| 1667 *flagp |= flags & SPSTART; | |
| 1668 else | |
| 1669 regtail(chain, latest); | |
| 1670 chain = latest; | |
| 1671 if (first == NULL) | |
| 1672 first = latest; | |
| 1673 break; | |
| 1674 } | |
| 1675 } | |
| 1676 if (first == NULL) /* Loop ran zero times. */ | |
| 1677 first = regnode(NOTHING); | |
| 1678 return first; | |
| 1679 } | |
| 1680 | |
| 1681 /* | |
| 1682 * regpiece - something followed by possible [*+=] | |
| 1683 * | |
| 1684 * Note that the branching code sequences used for = and the general cases | |
| 1685 * of * and + are somewhat optimized: they use the same NOTHING node as | |
| 1686 * both the endmarker for their branch list and the body of the last branch. | |
| 1687 * It might seem that this node could be dispensed with entirely, but the | |
| 1688 * endmarker role is not redundant. | |
| 1689 */ | |
| 1690 static char_u * | |
| 1691 regpiece(flagp) | |
| 1692 int *flagp; | |
| 1693 { | |
| 1694 char_u *ret; | |
| 1695 int op; | |
| 1696 char_u *next; | |
| 1697 int flags; | |
| 1698 long minval; | |
| 1699 long maxval; | |
| 1700 | |
| 1701 ret = regatom(&flags); | |
| 1702 if (ret == NULL) | |
| 1703 return NULL; | |
| 1704 | |
| 1705 op = peekchr(); | |
| 1706 if (re_multi_type(op) == NOT_MULTI) | |
| 1707 { | |
| 1708 *flagp = flags; | |
| 1709 return ret; | |
| 1710 } | |
| 1711 /* default flags */ | |
| 1712 *flagp = (WORST | SPSTART | (flags & (HASNL | HASLOOKBH))); | |
| 1713 | |
| 1714 skipchr(); | |
| 1715 switch (op) | |
| 1716 { | |
| 1717 case Magic('*'): | |
| 1718 if (flags & SIMPLE) | |
| 1719 reginsert(STAR, ret); | |
| 1720 else | |
| 1721 { | |
| 1722 /* Emit x* as (x&|), where & means "self". */ | |
| 1723 reginsert(BRANCH, ret); /* Either x */ | |
| 1724 regoptail(ret, regnode(BACK)); /* and loop */ | |
| 1725 regoptail(ret, ret); /* back */ | |
| 1726 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1727 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1728 } | |
| 1729 break; | |
| 1730 | |
| 1731 case Magic('+'): | |
| 1732 if (flags & SIMPLE) | |
| 1733 reginsert(PLUS, ret); | |
| 1734 else | |
| 1735 { | |
| 1736 /* Emit x+ as x(&|), where & means "self". */ | |
| 1737 next = regnode(BRANCH); /* Either */ | |
| 1738 regtail(ret, next); | |
| 233 | 1739 regtail(regnode(BACK), ret); /* loop back */ |
| 7 | 1740 regtail(next, regnode(BRANCH)); /* or */ |
| 1741 regtail(ret, regnode(NOTHING)); /* null. */ | |
| 1742 } | |
| 1743 *flagp = (WORST | HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1744 break; | |
| 1745 | |
| 1746 case Magic('@'): | |
| 1747 { | |
| 1748 int lop = END; | |
| 1749 | |
| 1750 switch (no_Magic(getchr())) | |
| 1751 { | |
| 1752 case '=': lop = MATCH; break; /* \@= */ | |
| 1753 case '!': lop = NOMATCH; break; /* \@! */ | |
| 1754 case '>': lop = SUBPAT; break; /* \@> */ | |
| 1755 case '<': switch (no_Magic(getchr())) | |
| 1756 { | |
| 1757 case '=': lop = BEHIND; break; /* \@<= */ | |
| 1758 case '!': lop = NOBEHIND; break; /* \@<! */ | |
| 1759 } | |
| 1760 } | |
| 1761 if (lop == END) | |
| 1762 EMSG_M_RET_NULL(_("E59: invalid character after %s@"), | |
| 1763 reg_magic == MAGIC_ALL); | |
| 1764 /* Look behind must match with behind_pos. */ | |
| 1765 if (lop == BEHIND || lop == NOBEHIND) | |
| 1766 { | |
| 1767 regtail(ret, regnode(BHPOS)); | |
| 1768 *flagp |= HASLOOKBH; | |
| 1769 } | |
| 1770 regtail(ret, regnode(END)); /* operand ends */ | |
| 1771 reginsert(lop, ret); | |
| 1772 break; | |
| 1773 } | |
| 1774 | |
| 1775 case Magic('?'): | |
| 1776 case Magic('='): | |
| 1777 /* Emit x= as (x|) */ | |
| 1778 reginsert(BRANCH, ret); /* Either x */ | |
| 1779 regtail(ret, regnode(BRANCH)); /* or */ | |
| 1780 next = regnode(NOTHING); /* null. */ | |
| 1781 regtail(ret, next); | |
| 1782 regoptail(ret, next); | |
| 1783 break; | |
| 1784 | |
| 1785 case Magic('{'): | |
| 1786 if (!read_limits(&minval, &maxval)) | |
| 1787 return NULL; | |
| 1788 if (flags & SIMPLE) | |
| 1789 { | |
| 1790 reginsert(BRACE_SIMPLE, ret); | |
| 1791 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1792 } | |
| 1793 else | |
| 1794 { | |
| 1795 if (num_complex_braces >= 10) | |
| 1796 EMSG_M_RET_NULL(_("E60: Too many complex %s{...}s"), | |
| 1797 reg_magic == MAGIC_ALL); | |
| 1798 reginsert(BRACE_COMPLEX + num_complex_braces, ret); | |
| 1799 regoptail(ret, regnode(BACK)); | |
| 1800 regoptail(ret, ret); | |
| 1801 reginsert_limits(BRACE_LIMITS, minval, maxval, ret); | |
| 1802 ++num_complex_braces; | |
| 1803 } | |
| 1804 if (minval > 0 && maxval > 0) | |
| 1805 *flagp = (HASWIDTH | (flags & (HASNL | HASLOOKBH))); | |
| 1806 break; | |
| 1807 } | |
| 1808 if (re_multi_type(peekchr()) != NOT_MULTI) | |
| 1809 { | |
| 1810 /* Can't have a multi follow a multi. */ | |
| 1811 if (peekchr() == Magic('*')) | |
| 1812 sprintf((char *)IObuff, _("E61: Nested %s*"), | |
| 1813 reg_magic >= MAGIC_ON ? "" : "\\"); | |
| 1814 else | |
| 1815 sprintf((char *)IObuff, _("E62: Nested %s%c"), | |
| 1816 reg_magic == MAGIC_ALL ? "" : "\\", no_Magic(peekchr())); | |
| 1817 EMSG_RET_NULL(IObuff); | |
| 1818 } | |
| 1819 | |
| 1820 return ret; | |
| 1821 } | |
| 1822 | |
| 1823 /* | |
| 1824 * regatom - the lowest level | |
| 1825 * | |
| 1826 * Optimization: gobbles an entire sequence of ordinary characters so that | |
| 1827 * it can turn them into a single node, which is smaller to store and | |
| 1828 * faster to run. Don't do this when one_exactly is set. | |
| 1829 */ | |
| 1830 static char_u * | |
| 1831 regatom(flagp) | |
| 1832 int *flagp; | |
| 1833 { | |
| 1834 char_u *ret; | |
| 1835 int flags; | |
| 1836 int cpo_lit; /* 'cpoptions' contains 'l' flag */ | |
| 167 | 1837 int cpo_bsl; /* 'cpoptions' contains '\' flag */ |
| 7 | 1838 int c; |
| 1839 static char_u *classchars = (char_u *)".iIkKfFpPsSdDxXoOwWhHaAlLuU"; | |
| 1840 static int classcodes[] = {ANY, IDENT, SIDENT, KWORD, SKWORD, | |
| 1841 FNAME, SFNAME, PRINT, SPRINT, | |
| 1842 WHITE, NWHITE, DIGIT, NDIGIT, | |
| 1843 HEX, NHEX, OCTAL, NOCTAL, | |
| 1844 WORD, NWORD, HEAD, NHEAD, | |
| 1845 ALPHA, NALPHA, LOWER, NLOWER, | |
| 1846 UPPER, NUPPER | |
| 1847 }; | |
| 1848 char_u *p; | |
| 1849 int extra = 0; | |
| 1850 | |
| 1851 *flagp = WORST; /* Tentatively. */ | |
| 410 | 1852 cpo_lit = vim_strchr(p_cpo, CPO_LITERAL) != NULL; |
| 1853 cpo_bsl = vim_strchr(p_cpo, CPO_BACKSL) != NULL; | |
| 7 | 1854 |
| 1855 c = getchr(); | |
| 1856 switch (c) | |
| 1857 { | |
| 1858 case Magic('^'): | |
| 1859 ret = regnode(BOL); | |
| 1860 break; | |
| 1861 | |
| 1862 case Magic('$'): | |
| 1863 ret = regnode(EOL); | |
| 1864 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1865 had_eol = TRUE; | |
| 1866 #endif | |
| 1867 break; | |
| 1868 | |
| 1869 case Magic('<'): | |
| 1870 ret = regnode(BOW); | |
| 1871 break; | |
| 1872 | |
| 1873 case Magic('>'): | |
| 1874 ret = regnode(EOW); | |
| 1875 break; | |
| 1876 | |
| 1877 case Magic('_'): | |
| 1878 c = no_Magic(getchr()); | |
| 1879 if (c == '^') /* "\_^" is start-of-line */ | |
| 1880 { | |
| 1881 ret = regnode(BOL); | |
| 1882 break; | |
| 1883 } | |
| 1884 if (c == '$') /* "\_$" is end-of-line */ | |
| 1885 { | |
| 1886 ret = regnode(EOL); | |
| 1887 #if defined(FEAT_SYN_HL) || defined(PROTO) | |
| 1888 had_eol = TRUE; | |
| 1889 #endif | |
| 1890 break; | |
| 1891 } | |
| 1892 | |
| 1893 extra = ADD_NL; | |
| 1894 *flagp |= HASNL; | |
| 1895 | |
| 1896 /* "\_[" is character range plus newline */ | |
| 1897 if (c == '[') | |
| 1898 goto collection; | |
| 1899 | |
| 1900 /* "\_x" is character class plus newline */ | |
| 1901 /*FALLTHROUGH*/ | |
| 1902 | |
| 1903 /* | |
| 1904 * Character classes. | |
| 1905 */ | |
| 1906 case Magic('.'): | |
| 1907 case Magic('i'): | |
| 1908 case Magic('I'): | |
| 1909 case Magic('k'): | |
| 1910 case Magic('K'): | |
| 1911 case Magic('f'): | |
| 1912 case Magic('F'): | |
| 1913 case Magic('p'): | |
| 1914 case Magic('P'): | |
| 1915 case Magic('s'): | |
| 1916 case Magic('S'): | |
| 1917 case Magic('d'): | |
| 1918 case Magic('D'): | |
| 1919 case Magic('x'): | |
| 1920 case Magic('X'): | |
| 1921 case Magic('o'): | |
| 1922 case Magic('O'): | |
| 1923 case Magic('w'): | |
| 1924 case Magic('W'): | |
| 1925 case Magic('h'): | |
| 1926 case Magic('H'): | |
| 1927 case Magic('a'): | |
| 1928 case Magic('A'): | |
| 1929 case Magic('l'): | |
| 1930 case Magic('L'): | |
| 1931 case Magic('u'): | |
| 1932 case Magic('U'): | |
| 1933 p = vim_strchr(classchars, no_Magic(c)); | |
| 1934 if (p == NULL) | |
| 1935 EMSG_RET_NULL(_("E63: invalid use of \\_")); | |
| 714 | 1936 #ifdef FEAT_MBYTE |
| 1937 /* When '.' is followed by a composing char ignore the dot, so that | |
| 1938 * the composing char is matched here. */ | |
| 1939 if (enc_utf8 && c == Magic('.') && utf_iscomposing(peekchr())) | |
| 1940 { | |
| 1941 c = getchr(); | |
| 1942 goto do_multibyte; | |
| 1943 } | |
| 1944 #endif | |
| 7 | 1945 ret = regnode(classcodes[p - classchars] + extra); |
| 1946 *flagp |= HASWIDTH | SIMPLE; | |
| 1947 break; | |
| 1948 | |
| 1949 case Magic('n'): | |
| 1950 if (reg_string) | |
| 1951 { | |
| 1952 /* In a string "\n" matches a newline character. */ | |
| 1953 ret = regnode(EXACTLY); | |
| 1954 regc(NL); | |
| 1955 regc(NUL); | |
| 1956 *flagp |= HASWIDTH | SIMPLE; | |
| 1957 } | |
| 1958 else | |
| 1959 { | |
| 1960 /* In buffer text "\n" matches the end of a line. */ | |
| 1961 ret = regnode(NEWL); | |
| 1962 *flagp |= HASWIDTH | HASNL; | |
| 1963 } | |
| 1964 break; | |
| 1965 | |
| 1966 case Magic('('): | |
| 1967 if (one_exactly) | |
| 1968 EMSG_ONE_RET_NULL; | |
| 1969 ret = reg(REG_PAREN, &flags); | |
| 1970 if (ret == NULL) | |
| 1971 return NULL; | |
| 1972 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 1973 break; | |
| 1974 | |
| 1975 case NUL: | |
| 1976 case Magic('|'): | |
| 1977 case Magic('&'): | |
| 1978 case Magic(')'): | |
| 1468 | 1979 if (one_exactly) |
| 1980 EMSG_ONE_RET_NULL; | |
| 7 | 1981 EMSG_RET_NULL(_(e_internal)); /* Supposed to be caught earlier. */ |
| 1982 /* NOTREACHED */ | |
| 1983 | |
| 1984 case Magic('='): | |
| 1985 case Magic('?'): | |
| 1986 case Magic('+'): | |
| 1987 case Magic('@'): | |
| 1988 case Magic('{'): | |
| 1989 case Magic('*'): | |
| 1990 c = no_Magic(c); | |
| 1991 sprintf((char *)IObuff, _("E64: %s%c follows nothing"), | |
| 1992 (c == '*' ? reg_magic >= MAGIC_ON : reg_magic == MAGIC_ALL) | |
| 1993 ? "" : "\\", c); | |
| 1994 EMSG_RET_NULL(IObuff); | |
| 1995 /* NOTREACHED */ | |
| 1996 | |
| 1997 case Magic('~'): /* previous substitute pattern */ | |
| 359 | 1998 if (reg_prev_sub != NULL) |
| 7 | 1999 { |
| 2000 char_u *lp; | |
| 2001 | |
| 2002 ret = regnode(EXACTLY); | |
| 2003 lp = reg_prev_sub; | |
| 2004 while (*lp != NUL) | |
| 2005 regc(*lp++); | |
| 2006 regc(NUL); | |
| 2007 if (*reg_prev_sub != NUL) | |
| 2008 { | |
| 2009 *flagp |= HASWIDTH; | |
| 2010 if ((lp - reg_prev_sub) == 1) | |
| 2011 *flagp |= SIMPLE; | |
| 2012 } | |
| 2013 } | |
| 2014 else | |
| 2015 EMSG_RET_NULL(_(e_nopresub)); | |
| 2016 break; | |
| 2017 | |
| 2018 case Magic('1'): | |
| 2019 case Magic('2'): | |
| 2020 case Magic('3'): | |
| 2021 case Magic('4'): | |
| 2022 case Magic('5'): | |
| 2023 case Magic('6'): | |
| 2024 case Magic('7'): | |
| 2025 case Magic('8'): | |
| 2026 case Magic('9'): | |
| 2027 { | |
| 2028 int refnum; | |
| 2029 | |
| 2030 refnum = c - Magic('0'); | |
| 2031 /* | |
| 2032 * Check if the back reference is legal. We must have seen the | |
| 2033 * close brace. | |
| 2034 * TODO: Should also check that we don't refer to something | |
| 2035 * that is repeated (+*=): what instance of the repetition | |
| 2036 * should we match? | |
| 2037 */ | |
| 2038 if (!had_endbrace[refnum]) | |
| 2039 { | |
| 2040 /* Trick: check if "@<=" or "@<!" follows, in which case | |
| 2041 * the \1 can appear before the referenced match. */ | |
| 2042 for (p = regparse; *p != NUL; ++p) | |
| 2043 if (p[0] == '@' && p[1] == '<' | |
| 2044 && (p[2] == '!' || p[2] == '=')) | |
| 2045 break; | |
| 2046 if (*p == NUL) | |
| 2047 EMSG_RET_NULL(_("E65: Illegal back reference")); | |
| 2048 } | |
| 2049 ret = regnode(BACKREF + refnum); | |
| 2050 } | |
| 2051 break; | |
| 2052 | |
| 2053 case Magic('z'): | |
| 2054 { | |
| 2055 c = no_Magic(getchr()); | |
| 2056 switch (c) | |
| 2057 { | |
| 741 | 2058 #ifdef FEAT_SYN_HL |
| 7 | 2059 case '(': if (reg_do_extmatch != REX_SET) |
| 2060 EMSG_RET_NULL(_("E66: \\z( not allowed here")); | |
| 2061 if (one_exactly) | |
| 2062 EMSG_ONE_RET_NULL; | |
| 2063 ret = reg(REG_ZPAREN, &flags); | |
| 2064 if (ret == NULL) | |
| 2065 return NULL; | |
| 2066 *flagp |= flags & (HASWIDTH|SPSTART|HASNL|HASLOOKBH); | |
| 2067 re_has_z = REX_SET; | |
| 2068 break; | |
| 2069 | |
| 2070 case '1': | |
| 2071 case '2': | |
| 2072 case '3': | |
| 2073 case '4': | |
| 2074 case '5': | |
| 2075 case '6': | |
| 2076 case '7': | |
| 2077 case '8': | |
| 2078 case '9': if (reg_do_extmatch != REX_USE) | |
| 2079 EMSG_RET_NULL(_("E67: \\z1 et al. not allowed here")); | |
| 2080 ret = regnode(ZREF + c - '0'); | |
| 2081 re_has_z = REX_USE; | |
| 2082 break; | |
| 741 | 2083 #endif |
| 7 | 2084 |
| 2085 case 's': ret = regnode(MOPEN + 0); | |
| 2086 break; | |
| 2087 | |
| 2088 case 'e': ret = regnode(MCLOSE + 0); | |
| 2089 break; | |
| 2090 | |
| 2091 default: EMSG_RET_NULL(_("E68: Invalid character after \\z")); | |
| 2092 } | |
| 2093 } | |
| 2094 break; | |
| 2095 | |
| 2096 case Magic('%'): | |
| 2097 { | |
| 2098 c = no_Magic(getchr()); | |
| 2099 switch (c) | |
| 2100 { | |
| 2101 /* () without a back reference */ | |
| 2102 case '(': | |
| 2103 if (one_exactly) | |
| 2104 EMSG_ONE_RET_NULL; | |
| 2105 ret = reg(REG_NPAREN, &flags); | |
| 2106 if (ret == NULL) | |
| 2107 return NULL; | |
| 2108 *flagp |= flags & (HASWIDTH | SPSTART | HASNL | HASLOOKBH); | |
| 2109 break; | |
| 2110 | |
| 2111 /* Catch \%^ and \%$ regardless of where they appear in the | |
| 2112 * pattern -- regardless of whether or not it makes sense. */ | |
| 2113 case '^': | |
| 2114 ret = regnode(RE_BOF); | |
| 2115 break; | |
| 2116 | |
| 2117 case '$': | |
| 2118 ret = regnode(RE_EOF); | |
| 2119 break; | |
| 2120 | |
| 2121 case '#': | |
| 2122 ret = regnode(CURSOR); | |
| 2123 break; | |
| 2124 | |
| 639 | 2125 case 'V': |
| 2126 ret = regnode(RE_VISUAL); | |
| 2127 break; | |
| 2128 | |
| 7 | 2129 /* \%[abc]: Emit as a list of branches, all ending at the last |
| 2130 * branch which matches nothing. */ | |
| 2131 case '[': | |
| 2132 if (one_exactly) /* doesn't nest */ | |
| 2133 EMSG_ONE_RET_NULL; | |
| 2134 { | |
| 2135 char_u *lastbranch; | |
| 2136 char_u *lastnode = NULL; | |
| 2137 char_u *br; | |
| 2138 | |
| 2139 ret = NULL; | |
| 2140 while ((c = getchr()) != ']') | |
| 2141 { | |
| 2142 if (c == NUL) | |
| 2143 EMSG_M_RET_NULL(_("E69: Missing ] after %s%%["), | |
| 2144 reg_magic == MAGIC_ALL); | |
| 2145 br = regnode(BRANCH); | |
| 2146 if (ret == NULL) | |
| 2147 ret = br; | |
| 2148 else | |
| 2149 regtail(lastnode, br); | |
| 2150 | |
| 2151 ungetchr(); | |
| 2152 one_exactly = TRUE; | |
| 2153 lastnode = regatom(flagp); | |
| 2154 one_exactly = FALSE; | |
| 2155 if (lastnode == NULL) | |
| 2156 return NULL; | |
| 2157 } | |
| 2158 if (ret == NULL) | |
| 2159 EMSG_M_RET_NULL(_("E70: Empty %s%%[]"), | |
| 2160 reg_magic == MAGIC_ALL); | |
| 2161 lastbranch = regnode(BRANCH); | |
| 2162 br = regnode(NOTHING); | |
| 2163 if (ret != JUST_CALC_SIZE) | |
| 2164 { | |
| 2165 regtail(lastnode, br); | |
| 2166 regtail(lastbranch, br); | |
| 2167 /* connect all branches to the NOTHING | |
| 2168 * branch at the end */ | |
| 2169 for (br = ret; br != lastnode; ) | |
| 2170 { | |
| 2171 if (OP(br) == BRANCH) | |
| 2172 { | |
| 2173 regtail(br, lastbranch); | |
| 2174 br = OPERAND(br); | |
| 2175 } | |
| 2176 else | |
| 2177 br = regnext(br); | |
| 2178 } | |
| 2179 } | |
| 1701 | 2180 *flagp &= ~(HASWIDTH | SIMPLE); |
| 7 | 2181 break; |
| 2182 } | |
| 2183 | |
| 24 | 2184 case 'd': /* %d123 decimal */ |
| 2185 case 'o': /* %o123 octal */ | |
| 2186 case 'x': /* %xab hex 2 */ | |
| 2187 case 'u': /* %uabcd hex 4 */ | |
| 2188 case 'U': /* %U1234abcd hex 8 */ | |
| 2189 { | |
| 2190 int i; | |
| 2191 | |
| 2192 switch (c) | |
| 2193 { | |
| 2194 case 'd': i = getdecchrs(); break; | |
| 2195 case 'o': i = getoctchrs(); break; | |
| 2196 case 'x': i = gethexchrs(2); break; | |
| 2197 case 'u': i = gethexchrs(4); break; | |
| 2198 case 'U': i = gethexchrs(8); break; | |
| 2199 default: i = -1; break; | |
| 2200 } | |
| 2201 | |
| 2202 if (i < 0) | |
| 856 | 2203 EMSG_M_RET_NULL( |
| 24 | 2204 _("E678: Invalid character after %s%%[dxouU]"), |
| 2205 reg_magic == MAGIC_ALL); | |
| 714 | 2206 #ifdef FEAT_MBYTE |
| 2207 if (use_multibytecode(i)) | |
| 2208 ret = regnode(MULTIBYTECODE); | |
| 2209 else | |
| 2210 #endif | |
| 2211 ret = regnode(EXACTLY); | |
| 24 | 2212 if (i == 0) |
| 2213 regc(0x0a); | |
| 2214 else | |
| 2215 #ifdef FEAT_MBYTE | |
| 2216 regmbc(i); | |
| 2217 #else | |
| 2218 regc(i); | |
| 2219 #endif | |
| 2220 regc(NUL); | |
| 2221 *flagp |= HASWIDTH; | |
| 2222 break; | |
| 2223 } | |
| 2224 | |
| 7 | 2225 default: |
| 639 | 2226 if (VIM_ISDIGIT(c) || c == '<' || c == '>' |
| 2227 || c == '\'') | |
| 7 | 2228 { |
| 2229 long_u n = 0; | |
| 2230 int cmp; | |
| 2231 | |
| 2232 cmp = c; | |
| 2233 if (cmp == '<' || cmp == '>') | |
| 2234 c = getchr(); | |
| 2235 while (VIM_ISDIGIT(c)) | |
| 2236 { | |
| 2237 n = n * 10 + (c - '0'); | |
| 2238 c = getchr(); | |
| 2239 } | |
| 639 | 2240 if (c == '\'' && n == 0) |
| 2241 { | |
| 2242 /* "\%'m", "\%<'m" and "\%>'m": Mark */ | |
| 2243 c = getchr(); | |
| 2244 ret = regnode(RE_MARK); | |
| 2245 if (ret == JUST_CALC_SIZE) | |
| 2246 regsize += 2; | |
| 2247 else | |
| 2248 { | |
| 2249 *regcode++ = c; | |
| 2250 *regcode++ = cmp; | |
| 2251 } | |
| 2252 break; | |
| 2253 } | |
| 2254 else if (c == 'l' || c == 'c' || c == 'v') | |
| 7 | 2255 { |
| 2256 if (c == 'l') | |
| 2257 ret = regnode(RE_LNUM); | |
| 2258 else if (c == 'c') | |
| 2259 ret = regnode(RE_COL); | |
| 2260 else | |
| 2261 ret = regnode(RE_VCOL); | |
| 2262 if (ret == JUST_CALC_SIZE) | |
| 2263 regsize += 5; | |
| 2264 else | |
| 2265 { | |
| 2266 /* put the number and the optional | |
| 2267 * comparator after the opcode */ | |
| 2268 regcode = re_put_long(regcode, n); | |
| 2269 *regcode++ = cmp; | |
| 2270 } | |
| 2271 break; | |
| 2272 } | |
| 2273 } | |
| 2274 | |
| 2275 EMSG_M_RET_NULL(_("E71: Invalid character after %s%%"), | |
| 2276 reg_magic == MAGIC_ALL); | |
| 2277 } | |
| 2278 } | |
| 2279 break; | |
| 2280 | |
| 2281 case Magic('['): | |
| 2282 collection: | |
| 2283 { | |
| 2284 char_u *lp; | |
| 2285 | |
| 2286 /* | |
| 2287 * If there is no matching ']', we assume the '[' is a normal | |
| 2288 * character. This makes 'incsearch' and ":help [" work. | |
| 2289 */ | |
| 2290 lp = skip_anyof(regparse); | |
| 2291 if (*lp == ']') /* there is a matching ']' */ | |
| 2292 { | |
| 2293 int startc = -1; /* > 0 when next '-' is a range */ | |
| 2294 int endc; | |
| 2295 | |
| 2296 /* | |
| 2297 * In a character class, different parsing rules apply. | |
| 2298 * Not even \ is special anymore, nothing is. | |
| 2299 */ | |
| 2300 if (*regparse == '^') /* Complement of range. */ | |
| 2301 { | |
| 2302 ret = regnode(ANYBUT + extra); | |
| 2303 regparse++; | |
| 2304 } | |
| 2305 else | |
| 2306 ret = regnode(ANYOF + extra); | |
| 2307 | |
| 2308 /* At the start ']' and '-' mean the literal character. */ | |
| 2309 if (*regparse == ']' || *regparse == '-') | |
| 167 | 2310 { |
| 2311 startc = *regparse; | |
| 7 | 2312 regc(*regparse++); |
| 167 | 2313 } |
| 7 | 2314 |
| 2315 while (*regparse != NUL && *regparse != ']') | |
| 2316 { | |
| 2317 if (*regparse == '-') | |
| 2318 { | |
| 2319 ++regparse; | |
| 2320 /* The '-' is not used for a range at the end and | |
| 2321 * after or before a '\n'. */ | |
| 2322 if (*regparse == ']' || *regparse == NUL | |
| 2323 || startc == -1 | |
| 2324 || (regparse[0] == '\\' && regparse[1] == 'n')) | |
| 2325 { | |
| 2326 regc('-'); | |
| 2327 startc = '-'; /* [--x] is a range */ | |
| 2328 } | |
| 2329 else | |
| 2330 { | |
| 167 | 2331 /* Also accept "a-[.z.]" */ |
| 2332 endc = 0; | |
| 2333 if (*regparse == '[') | |
| 2334 endc = get_coll_element(®parse); | |
| 2335 if (endc == 0) | |
| 2336 { | |
| 7 | 2337 #ifdef FEAT_MBYTE |
| 167 | 2338 if (has_mbyte) |
| 2339 endc = mb_ptr2char_adv(®parse); | |
| 2340 else | |
| 7 | 2341 #endif |
| 167 | 2342 endc = *regparse++; |
| 2343 } | |
| 24 | 2344 |
| 2345 /* Handle \o40, \x20 and \u20AC style sequences */ | |
| 167 | 2346 if (endc == '\\' && !cpo_lit && !cpo_bsl) |
| 24 | 2347 endc = coll_get_char(); |
| 2348 | |
| 7 | 2349 if (startc > endc) |
| 2350 EMSG_RET_NULL(_(e_invrange)); | |
| 2351 #ifdef FEAT_MBYTE | |
| 2352 if (has_mbyte && ((*mb_char2len)(startc) > 1 | |
| 2353 || (*mb_char2len)(endc) > 1)) | |
| 2354 { | |
| 2355 /* Limit to a range of 256 chars */ | |
| 2356 if (endc > startc + 256) | |
| 2357 EMSG_RET_NULL(_(e_invrange)); | |
| 2358 while (++startc <= endc) | |
| 2359 regmbc(startc); | |
| 2360 } | |
| 2361 else | |
| 2362 #endif | |
| 2363 { | |
| 2364 #ifdef EBCDIC | |
| 2365 int alpha_only = FALSE; | |
| 2366 | |
| 2367 /* for alphabetical range skip the gaps | |
| 2368 * 'i'-'j', 'r'-'s', 'I'-'J' and 'R'-'S'. */ | |
| 2369 if (isalpha(startc) && isalpha(endc)) | |
| 2370 alpha_only = TRUE; | |
| 2371 #endif | |
| 2372 while (++startc <= endc) | |
| 2373 #ifdef EBCDIC | |
| 2374 if (!alpha_only || isalpha(startc)) | |
| 2375 #endif | |
| 2376 regc(startc); | |
| 2377 } | |
| 2378 startc = -1; | |
| 2379 } | |
| 2380 } | |
| 2381 /* | |
| 2382 * Only "\]", "\^", "\]" and "\\" are special in Vi. Vim | |
| 2383 * accepts "\t", "\e", etc., but only when the 'l' flag in | |
| 2384 * 'cpoptions' is not included. | |
| 167 | 2385 * Posix doesn't recognize backslash at all. |
| 7 | 2386 */ |
| 2387 else if (*regparse == '\\' | |
| 167 | 2388 && !cpo_bsl |
| 7 | 2389 && (vim_strchr(REGEXP_INRANGE, regparse[1]) != NULL |
| 2390 || (!cpo_lit | |
| 2391 && vim_strchr(REGEXP_ABBR, | |
| 2392 regparse[1]) != NULL))) | |
| 2393 { | |
| 2394 regparse++; | |
| 2395 if (*regparse == 'n') | |
| 2396 { | |
| 2397 /* '\n' in range: also match NL */ | |
| 2398 if (ret != JUST_CALC_SIZE) | |
| 2399 { | |
| 4084 | 2400 /* Using \n inside [^] does not change what |
| 2401 * matches. "[^\n]" is the same as ".". */ | |
| 2402 if (*ret == ANYOF) | |
| 2403 { | |
| 7 | 2404 *ret = ANYOF + ADD_NL; |
| 4084 | 2405 *flagp |= HASNL; |
| 2406 } | |
| 7 | 2407 /* else: must have had a \n already */ |
| 2408 } | |
| 2409 regparse++; | |
| 2410 startc = -1; | |
| 2411 } | |
| 24 | 2412 else if (*regparse == 'd' |
| 2413 || *regparse == 'o' | |
| 2414 || *regparse == 'x' | |
| 2415 || *regparse == 'u' | |
| 2416 || *regparse == 'U') | |
| 2417 { | |
| 2418 startc = coll_get_char(); | |
| 2419 if (startc == 0) | |
| 2420 regc(0x0a); | |
| 2421 else | |
| 2422 #ifdef FEAT_MBYTE | |
| 2423 regmbc(startc); | |
| 2424 #else | |
| 2425 regc(startc); | |
| 2426 #endif | |
| 2427 } | |
| 7 | 2428 else |
| 2429 { | |
| 2430 startc = backslash_trans(*regparse++); | |
| 2431 regc(startc); | |
| 2432 } | |
| 2433 } | |
| 2434 else if (*regparse == '[') | |
| 2435 { | |
| 2436 int c_class; | |
| 2437 int cu; | |
| 2438 | |
| 167 | 2439 c_class = get_char_class(®parse); |
| 7 | 2440 startc = -1; |
| 2441 /* Characters assumed to be 8 bits! */ | |
| 2442 switch (c_class) | |
| 2443 { | |
| 2444 case CLASS_NONE: | |
| 167 | 2445 c_class = get_equi_class(®parse); |
| 2446 if (c_class != 0) | |
| 2447 { | |
| 2448 /* produce equivalence class */ | |
| 2449 reg_equi_class(c_class); | |
| 2450 } | |
| 2451 else if ((c_class = | |
| 2452 get_coll_element(®parse)) != 0) | |
| 2453 { | |
| 2454 /* produce a collating element */ | |
| 2455 regmbc(c_class); | |
| 2456 } | |
| 2457 else | |
| 2458 { | |
| 2459 /* literal '[', allow [[-x] as a range */ | |
| 2460 startc = *regparse++; | |
| 2461 regc(startc); | |
| 2462 } | |
| 7 | 2463 break; |
| 2464 case CLASS_ALNUM: | |
| 2465 for (cu = 1; cu <= 255; cu++) | |
| 2466 if (isalnum(cu)) | |
| 2467 regc(cu); | |
| 2468 break; | |
| 2469 case CLASS_ALPHA: | |
| 2470 for (cu = 1; cu <= 255; cu++) | |
| 2471 if (isalpha(cu)) | |
| 2472 regc(cu); | |
| 2473 break; | |
| 2474 case CLASS_BLANK: | |
| 2475 regc(' '); | |
| 2476 regc('\t'); | |
| 2477 break; | |
| 2478 case CLASS_CNTRL: | |
| 2479 for (cu = 1; cu <= 255; cu++) | |
| 2480 if (iscntrl(cu)) | |
| 2481 regc(cu); | |
| 2482 break; | |
| 2483 case CLASS_DIGIT: | |
| 2484 for (cu = 1; cu <= 255; cu++) | |
| 2485 if (VIM_ISDIGIT(cu)) | |
| 2486 regc(cu); | |
| 2487 break; | |
| 2488 case CLASS_GRAPH: | |
| 2489 for (cu = 1; cu <= 255; cu++) | |
| 2490 if (isgraph(cu)) | |
| 2491 regc(cu); | |
| 2492 break; | |
| 2493 case CLASS_LOWER: | |
| 2494 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2495 if (MB_ISLOWER(cu)) |
| 7 | 2496 regc(cu); |
| 2497 break; | |
| 2498 case CLASS_PRINT: | |
| 2499 for (cu = 1; cu <= 255; cu++) | |
| 2500 if (vim_isprintc(cu)) | |
| 2501 regc(cu); | |
| 2502 break; | |
| 2503 case CLASS_PUNCT: | |
| 2504 for (cu = 1; cu <= 255; cu++) | |
| 2505 if (ispunct(cu)) | |
| 2506 regc(cu); | |
| 2507 break; | |
| 2508 case CLASS_SPACE: | |
| 2509 for (cu = 9; cu <= 13; cu++) | |
| 2510 regc(cu); | |
| 2511 regc(' '); | |
| 2512 break; | |
| 2513 case CLASS_UPPER: | |
| 2514 for (cu = 1; cu <= 255; cu++) | |
| 1347 | 2515 if (MB_ISUPPER(cu)) |
| 7 | 2516 regc(cu); |
| 2517 break; | |
| 2518 case CLASS_XDIGIT: | |
| 2519 for (cu = 1; cu <= 255; cu++) | |
| 2520 if (vim_isxdigit(cu)) | |
| 2521 regc(cu); | |
| 2522 break; | |
| 2523 case CLASS_TAB: | |
| 2524 regc('\t'); | |
| 2525 break; | |
| 2526 case CLASS_RETURN: | |
| 2527 regc('\r'); | |
| 2528 break; | |
| 2529 case CLASS_BACKSPACE: | |
| 2530 regc('\b'); | |
| 2531 break; | |
| 2532 case CLASS_ESCAPE: | |
| 2533 regc('\033'); | |
| 2534 break; | |
| 2535 } | |
| 2536 } | |
| 2537 else | |
| 2538 { | |
| 2539 #ifdef FEAT_MBYTE | |
| 2540 if (has_mbyte) | |
| 2541 { | |
| 2542 int len; | |
| 2543 | |
| 2544 /* produce a multibyte character, including any | |
| 2545 * following composing characters */ | |
| 2546 startc = mb_ptr2char(regparse); | |
| 474 | 2547 len = (*mb_ptr2len)(regparse); |
| 7 | 2548 if (enc_utf8 && utf_char2len(startc) != len) |
| 2549 startc = -1; /* composing chars */ | |
| 2550 while (--len >= 0) | |
| 2551 regc(*regparse++); | |
| 2552 } | |
| 2553 else | |
| 2554 #endif | |
| 2555 { | |
| 2556 startc = *regparse++; | |
| 2557 regc(startc); | |
| 2558 } | |
| 2559 } | |
| 2560 } | |
| 2561 regc(NUL); | |
| 2562 prevchr_len = 1; /* last char was the ']' */ | |
| 2563 if (*regparse != ']') | |
| 2564 EMSG_RET_NULL(_(e_toomsbra)); /* Cannot happen? */ | |
| 2565 skipchr(); /* let's be friends with the lexer again */ | |
| 2566 *flagp |= HASWIDTH | SIMPLE; | |
| 2567 break; | |
| 2568 } | |
| 481 | 2569 else if (reg_strict) |
| 2570 EMSG_M_RET_NULL(_("E769: Missing ] after %s["), | |
| 2571 reg_magic > MAGIC_OFF); | |
| 7 | 2572 } |
| 2573 /* FALLTHROUGH */ | |
| 2574 | |
| 2575 default: | |
| 2576 { | |
| 2577 int len; | |
| 2578 | |
| 2579 #ifdef FEAT_MBYTE | |
| 2580 /* A multi-byte character is handled as a separate atom if it's | |
| 714 | 2581 * before a multi and when it's a composing char. */ |
| 2582 if (use_multibytecode(c)) | |
| 7 | 2583 { |
| 714 | 2584 do_multibyte: |
| 7 | 2585 ret = regnode(MULTIBYTECODE); |
| 2586 regmbc(c); | |
| 2587 *flagp |= HASWIDTH | SIMPLE; | |
| 2588 break; | |
| 2589 } | |
| 2590 #endif | |
| 2591 | |
| 2592 ret = regnode(EXACTLY); | |
| 2593 | |
| 2594 /* | |
| 2595 * Append characters as long as: | |
| 2596 * - there is no following multi, we then need the character in | |
| 2597 * front of it as a single character operand | |
| 2598 * - not running into a Magic character | |
| 2599 * - "one_exactly" is not set | |
| 2600 * But always emit at least one character. Might be a Multi, | |
| 2601 * e.g., a "[" without matching "]". | |
| 2602 */ | |
| 2603 for (len = 0; c != NUL && (len == 0 | |
| 2604 || (re_multi_type(peekchr()) == NOT_MULTI | |
| 2605 && !one_exactly | |
| 2606 && !is_Magic(c))); ++len) | |
| 2607 { | |
| 2608 c = no_Magic(c); | |
| 2609 #ifdef FEAT_MBYTE | |
| 2610 if (has_mbyte) | |
| 2611 { | |
| 2612 regmbc(c); | |
| 2613 if (enc_utf8) | |
| 2614 { | |
| 2615 int l; | |
| 2616 | |
| 714 | 2617 /* Need to get composing character too. */ |
| 7 | 2618 for (;;) |
| 2619 { | |
| 714 | 2620 l = utf_ptr2len(regparse); |
| 2621 if (!UTF_COMPOSINGLIKE(regparse, regparse + l)) | |
| 7 | 2622 break; |
| 714 | 2623 regmbc(utf_ptr2char(regparse)); |
| 2624 skipchr(); | |
| 7 | 2625 } |
| 2626 } | |
| 2627 } | |
| 2628 else | |
| 2629 #endif | |
| 2630 regc(c); | |
| 2631 c = getchr(); | |
| 2632 } | |
| 2633 ungetchr(); | |
| 2634 | |
| 2635 regc(NUL); | |
| 2636 *flagp |= HASWIDTH; | |
| 2637 if (len == 1) | |
| 2638 *flagp |= SIMPLE; | |
| 2639 } | |
| 2640 break; | |
| 2641 } | |
| 2642 | |
| 2643 return ret; | |
| 2644 } | |
| 2645 | |
| 714 | 2646 #ifdef FEAT_MBYTE |
| 2647 /* | |
| 2648 * Return TRUE if MULTIBYTECODE should be used instead of EXACTLY for | |
| 2649 * character "c". | |
| 2650 */ | |
| 2651 static int | |
| 2652 use_multibytecode(c) | |
| 2653 int c; | |
| 2654 { | |
| 2655 return has_mbyte && (*mb_char2len)(c) > 1 | |
| 2656 && (re_multi_type(peekchr()) != NOT_MULTI | |
| 2657 || (enc_utf8 && utf_iscomposing(c))); | |
| 2658 } | |
| 2659 #endif | |
| 2660 | |
| 7 | 2661 /* |
| 2662 * emit a node | |
| 2663 * Return pointer to generated code. | |
| 2664 */ | |
| 2665 static char_u * | |
| 2666 regnode(op) | |
| 2667 int op; | |
| 2668 { | |
| 2669 char_u *ret; | |
| 2670 | |
| 2671 ret = regcode; | |
| 2672 if (ret == JUST_CALC_SIZE) | |
| 2673 regsize += 3; | |
| 2674 else | |
| 2675 { | |
| 2676 *regcode++ = op; | |
| 2677 *regcode++ = NUL; /* Null "next" pointer. */ | |
| 2678 *regcode++ = NUL; | |
| 2679 } | |
| 2680 return ret; | |
| 2681 } | |
| 2682 | |
| 2683 /* | |
| 2684 * Emit (if appropriate) a byte of code | |
| 2685 */ | |
| 2686 static void | |
| 2687 regc(b) | |
| 2688 int b; | |
| 2689 { | |
| 2690 if (regcode == JUST_CALC_SIZE) | |
| 2691 regsize++; | |
| 2692 else | |
| 2693 *regcode++ = b; | |
| 2694 } | |
| 2695 | |
| 2696 #ifdef FEAT_MBYTE | |
| 2697 /* | |
| 2698 * Emit (if appropriate) a multi-byte character of code | |
| 2699 */ | |
| 2700 static void | |
| 2701 regmbc(c) | |
| 2702 int c; | |
| 2703 { | |
| 2974 | 2704 if (!has_mbyte && c > 0xff) |
| 2705 return; | |
| 7 | 2706 if (regcode == JUST_CALC_SIZE) |
| 2707 regsize += (*mb_char2len)(c); | |
| 2708 else | |
| 2709 regcode += (*mb_char2bytes)(c, regcode); | |
| 2710 } | |
| 2711 #endif | |
| 2712 | |
| 2713 /* | |
| 2714 * reginsert - insert an operator in front of already-emitted operand | |
| 2715 * | |
| 2716 * Means relocating the operand. | |
| 2717 */ | |
| 2718 static void | |
| 2719 reginsert(op, opnd) | |
| 2720 int op; | |
| 2721 char_u *opnd; | |
| 2722 { | |
| 2723 char_u *src; | |
| 2724 char_u *dst; | |
| 2725 char_u *place; | |
| 2726 | |
| 2727 if (regcode == JUST_CALC_SIZE) | |
| 2728 { | |
| 2729 regsize += 3; | |
| 2730 return; | |
| 2731 } | |
| 2732 src = regcode; | |
| 2733 regcode += 3; | |
| 2734 dst = regcode; | |
| 2735 while (src > opnd) | |
| 2736 *--dst = *--src; | |
| 2737 | |
| 2738 place = opnd; /* Op node, where operand used to be. */ | |
| 2739 *place++ = op; | |
| 2740 *place++ = NUL; | |
| 2741 *place = NUL; | |
| 2742 } | |
| 2743 | |
| 2744 /* | |
| 2745 * reginsert_limits - insert an operator in front of already-emitted operand. | |
| 2746 * The operator has the given limit values as operands. Also set next pointer. | |
| 2747 * | |
| 2748 * Means relocating the operand. | |
| 2749 */ | |
| 2750 static void | |
| 2751 reginsert_limits(op, minval, maxval, opnd) | |
| 2752 int op; | |
| 2753 long minval; | |
| 2754 long maxval; | |
| 2755 char_u *opnd; | |
| 2756 { | |
| 2757 char_u *src; | |
| 2758 char_u *dst; | |
| 2759 char_u *place; | |
| 2760 | |
| 2761 if (regcode == JUST_CALC_SIZE) | |
| 2762 { | |
| 2763 regsize += 11; | |
| 2764 return; | |
| 2765 } | |
| 2766 src = regcode; | |
| 2767 regcode += 11; | |
| 2768 dst = regcode; | |
| 2769 while (src > opnd) | |
| 2770 *--dst = *--src; | |
| 2771 | |
| 2772 place = opnd; /* Op node, where operand used to be. */ | |
| 2773 *place++ = op; | |
| 2774 *place++ = NUL; | |
| 2775 *place++ = NUL; | |
| 2776 place = re_put_long(place, (long_u)minval); | |
| 2777 place = re_put_long(place, (long_u)maxval); | |
| 2778 regtail(opnd, place); | |
| 2779 } | |
| 2780 | |
| 2781 /* | |
| 2782 * Write a long as four bytes at "p" and return pointer to the next char. | |
| 2783 */ | |
| 2784 static char_u * | |
| 2785 re_put_long(p, val) | |
| 2786 char_u *p; | |
| 2787 long_u val; | |
| 2788 { | |
| 2789 *p++ = (char_u) ((val >> 24) & 0377); | |
| 2790 *p++ = (char_u) ((val >> 16) & 0377); | |
| 2791 *p++ = (char_u) ((val >> 8) & 0377); | |
| 2792 *p++ = (char_u) (val & 0377); | |
| 2793 return p; | |
| 2794 } | |
| 2795 | |
| 2796 /* | |
| 2797 * regtail - set the next-pointer at the end of a node chain | |
| 2798 */ | |
| 2799 static void | |
| 2800 regtail(p, val) | |
| 2801 char_u *p; | |
| 2802 char_u *val; | |
| 2803 { | |
| 2804 char_u *scan; | |
| 2805 char_u *temp; | |
| 2806 int offset; | |
| 2807 | |
| 2808 if (p == JUST_CALC_SIZE) | |
| 2809 return; | |
| 2810 | |
| 2811 /* Find last node. */ | |
| 2812 scan = p; | |
| 2813 for (;;) | |
| 2814 { | |
| 2815 temp = regnext(scan); | |
| 2816 if (temp == NULL) | |
| 2817 break; | |
| 2818 scan = temp; | |
| 2819 } | |
| 2820 | |
| 233 | 2821 if (OP(scan) == BACK) |
| 7 | 2822 offset = (int)(scan - val); |
| 2823 else | |
| 2824 offset = (int)(val - scan); | |
| 2010 | 2825 /* When the offset uses more than 16 bits it can no longer fit in the two |
| 2974 | 2826 * bytes available. Use a global flag to avoid having to check return |
| 2010 | 2827 * values in too many places. */ |
| 2828 if (offset > 0xffff) | |
| 2829 reg_toolong = TRUE; | |
| 2830 else | |
| 2831 { | |
| 2832 *(scan + 1) = (char_u) (((unsigned)offset >> 8) & 0377); | |
| 2833 *(scan + 2) = (char_u) (offset & 0377); | |
| 2834 } | |
| 7 | 2835 } |
| 2836 | |
| 2837 /* | |
| 2838 * regoptail - regtail on item after a BRANCH; nop if none | |
| 2839 */ | |
| 2840 static void | |
| 2841 regoptail(p, val) | |
| 2842 char_u *p; | |
| 2843 char_u *val; | |
| 2844 { | |
| 2845 /* When op is neither BRANCH nor BRACE_COMPLEX0-9, it is "operandless" */ | |
| 2846 if (p == NULL || p == JUST_CALC_SIZE | |
| 2847 || (OP(p) != BRANCH | |
| 2848 && (OP(p) < BRACE_COMPLEX || OP(p) > BRACE_COMPLEX + 9))) | |
| 2849 return; | |
| 2850 regtail(OPERAND(p), val); | |
| 2851 } | |
| 2852 | |
| 2853 /* | |
| 2854 * getchr() - get the next character from the pattern. We know about | |
| 2855 * magic and such, so therefore we need a lexical analyzer. | |
| 2856 */ | |
| 2857 | |
| 2858 /* static int curchr; */ | |
| 2859 static int prevprevchr; | |
| 2860 static int prevchr; | |
| 2861 static int nextchr; /* used for ungetchr() */ | |
| 2862 /* | |
| 2863 * Note: prevchr is sometimes -1 when we are not at the start, | |
| 2864 * eg in /[ ^I]^ the pattern was never found even if it existed, because ^ was | |
| 2865 * taken to be magic -- webb | |
| 2866 */ | |
| 2867 static int at_start; /* True when on the first character */ | |
| 2868 static int prev_at_start; /* True when on the second character */ | |
| 2869 | |
| 2870 static void | |
| 2871 initchr(str) | |
| 2872 char_u *str; | |
| 2873 { | |
| 2874 regparse = str; | |
| 2875 prevchr_len = 0; | |
| 2876 curchr = prevprevchr = prevchr = nextchr = -1; | |
| 2877 at_start = TRUE; | |
| 2878 prev_at_start = FALSE; | |
| 2879 } | |
| 2880 | |
| 2881 static int | |
| 2882 peekchr() | |
| 2883 { | |
| 167 | 2884 static int after_slash = FALSE; |
| 2885 | |
| 7 | 2886 if (curchr == -1) |
| 2887 { | |
| 2888 switch (curchr = regparse[0]) | |
| 2889 { | |
| 2890 case '.': | |
| 2891 case '[': | |
| 2892 case '~': | |
| 2893 /* magic when 'magic' is on */ | |
| 2894 if (reg_magic >= MAGIC_ON) | |
| 2895 curchr = Magic(curchr); | |
| 2896 break; | |
| 2897 case '(': | |
| 2898 case ')': | |
| 2899 case '{': | |
| 2900 case '%': | |
| 2901 case '+': | |
| 2902 case '=': | |
| 2903 case '?': | |
| 2904 case '@': | |
| 2905 case '!': | |
| 2906 case '&': | |
| 2907 case '|': | |
| 2908 case '<': | |
| 2909 case '>': | |
| 2910 case '#': /* future ext. */ | |
| 2911 case '"': /* future ext. */ | |
| 2912 case '\'': /* future ext. */ | |
| 2913 case ',': /* future ext. */ | |
| 2914 case '-': /* future ext. */ | |
| 2915 case ':': /* future ext. */ | |
| 2916 case ';': /* future ext. */ | |
| 2917 case '`': /* future ext. */ | |
| 2918 case '/': /* Can't be used in / command */ | |
| 2919 /* magic only after "\v" */ | |
| 2920 if (reg_magic == MAGIC_ALL) | |
| 2921 curchr = Magic(curchr); | |
| 2922 break; | |
| 2923 case '*': | |
| 167 | 2924 /* * is not magic as the very first character, eg "?*ptr", when |
| 2925 * after '^', eg "/^*ptr" and when after "\(", "\|", "\&". But | |
| 2926 * "\(\*" is not magic, thus must be magic if "after_slash" */ | |
| 2927 if (reg_magic >= MAGIC_ON | |
| 2928 && !at_start | |
| 2929 && !(prev_at_start && prevchr == Magic('^')) | |
| 2930 && (after_slash | |
| 2931 || (prevchr != Magic('(') | |
| 2932 && prevchr != Magic('&') | |
| 2933 && prevchr != Magic('|')))) | |
| 7 | 2934 curchr = Magic('*'); |
| 2935 break; | |
| 2936 case '^': | |
| 2937 /* '^' is only magic as the very first character and if it's after | |
| 2938 * "\(", "\|", "\&' or "\n" */ | |
| 2939 if (reg_magic >= MAGIC_OFF | |
| 2940 && (at_start | |
| 2941 || reg_magic == MAGIC_ALL | |
| 2942 || prevchr == Magic('(') | |
| 2943 || prevchr == Magic('|') | |
| 2944 || prevchr == Magic('&') | |
| 2945 || prevchr == Magic('n') | |
| 2946 || (no_Magic(prevchr) == '(' | |
| 2947 && prevprevchr == Magic('%')))) | |
| 2948 { | |
| 2949 curchr = Magic('^'); | |
| 2950 at_start = TRUE; | |
| 2951 prev_at_start = FALSE; | |
| 2952 } | |
| 2953 break; | |
| 2954 case '$': | |
| 2955 /* '$' is only magic as the very last char and if it's in front of | |
| 2956 * either "\|", "\)", "\&", or "\n" */ | |
| 2957 if (reg_magic >= MAGIC_OFF) | |
| 2958 { | |
| 2959 char_u *p = regparse + 1; | |
| 2960 | |
| 2961 /* ignore \c \C \m and \M after '$' */ | |
| 2962 while (p[0] == '\\' && (p[1] == 'c' || p[1] == 'C' | |
| 2963 || p[1] == 'm' || p[1] == 'M' || p[1] == 'Z')) | |
| 2964 p += 2; | |
| 2965 if (p[0] == NUL | |
| 2966 || (p[0] == '\\' | |
| 2967 && (p[1] == '|' || p[1] == '&' || p[1] == ')' | |
| 2968 || p[1] == 'n')) | |
| 2969 || reg_magic == MAGIC_ALL) | |
| 2970 curchr = Magic('$'); | |
| 2971 } | |
| 2972 break; | |
| 2973 case '\\': | |
| 2974 { | |
| 2975 int c = regparse[1]; | |
| 2976 | |
| 2977 if (c == NUL) | |
| 2978 curchr = '\\'; /* trailing '\' */ | |
| 2979 else if ( | |
| 2980 #ifdef EBCDIC | |
| 2981 vim_strchr(META, c) | |
| 2982 #else | |
| 2983 c <= '~' && META_flags[c] | |
| 2984 #endif | |
| 2985 ) | |
| 2986 { | |
| 2987 /* | |
| 2988 * META contains everything that may be magic sometimes, | |
| 2989 * except ^ and $ ("\^" and "\$" are only magic after | |
| 2990 * "\v"). We now fetch the next character and toggle its | |
| 2991 * magicness. Therefore, \ is so meta-magic that it is | |
| 2992 * not in META. | |
| 2993 */ | |
| 2994 curchr = -1; | |
| 2995 prev_at_start = at_start; | |
| 2996 at_start = FALSE; /* be able to say "/\*ptr" */ | |
| 2997 ++regparse; | |
| 167 | 2998 ++after_slash; |
| 7 | 2999 peekchr(); |
| 3000 --regparse; | |
| 167 | 3001 --after_slash; |
| 7 | 3002 curchr = toggle_Magic(curchr); |
| 3003 } | |
| 3004 else if (vim_strchr(REGEXP_ABBR, c)) | |
| 3005 { | |
| 3006 /* | |
| 3007 * Handle abbreviations, like "\t" for TAB -- webb | |
| 3008 */ | |
| 3009 curchr = backslash_trans(c); | |
| 3010 } | |
| 3011 else if (reg_magic == MAGIC_NONE && (c == '$' || c == '^')) | |
| 3012 curchr = toggle_Magic(c); | |
| 3013 else | |
| 3014 { | |
| 3015 /* | |
| 3016 * Next character can never be (made) magic? | |
| 3017 * Then backslashing it won't do anything. | |
| 3018 */ | |
| 3019 #ifdef FEAT_MBYTE | |
| 3020 if (has_mbyte) | |
| 3021 curchr = (*mb_ptr2char)(regparse + 1); | |
| 3022 else | |
| 3023 #endif | |
| 3024 curchr = c; | |
| 3025 } | |
| 3026 break; | |
| 3027 } | |
| 3028 | |
| 3029 #ifdef FEAT_MBYTE | |
| 3030 default: | |
| 3031 if (has_mbyte) | |
| 3032 curchr = (*mb_ptr2char)(regparse); | |
| 3033 #endif | |
| 3034 } | |
| 3035 } | |
| 3036 | |
| 3037 return curchr; | |
| 3038 } | |
| 3039 | |
| 3040 /* | |
| 3041 * Eat one lexed character. Do this in a way that we can undo it. | |
| 3042 */ | |
| 3043 static void | |
| 3044 skipchr() | |
| 3045 { | |
| 3046 /* peekchr() eats a backslash, do the same here */ | |
| 3047 if (*regparse == '\\') | |
| 3048 prevchr_len = 1; | |
| 3049 else | |
| 3050 prevchr_len = 0; | |
| 3051 if (regparse[prevchr_len] != NUL) | |
| 3052 { | |
| 3053 #ifdef FEAT_MBYTE | |
| 714 | 3054 if (enc_utf8) |
| 1449 | 3055 /* exclude composing chars that mb_ptr2len does include */ |
| 3056 prevchr_len += utf_ptr2len(regparse + prevchr_len); | |
| 714 | 3057 else if (has_mbyte) |
| 474 | 3058 prevchr_len += (*mb_ptr2len)(regparse + prevchr_len); |
| 7 | 3059 else |
| 3060 #endif | |
| 3061 ++prevchr_len; | |
| 3062 } | |
| 3063 regparse += prevchr_len; | |
| 3064 prev_at_start = at_start; | |
| 3065 at_start = FALSE; | |
| 3066 prevprevchr = prevchr; | |
| 3067 prevchr = curchr; | |
| 3068 curchr = nextchr; /* use previously unget char, or -1 */ | |
| 3069 nextchr = -1; | |
| 3070 } | |
| 3071 | |
| 3072 /* | |
| 3073 * Skip a character while keeping the value of prev_at_start for at_start. | |
| 3074 * prevchr and prevprevchr are also kept. | |
| 3075 */ | |
| 3076 static void | |
| 3077 skipchr_keepstart() | |
| 3078 { | |
| 3079 int as = prev_at_start; | |
| 3080 int pr = prevchr; | |
| 3081 int prpr = prevprevchr; | |
| 3082 | |
| 3083 skipchr(); | |
| 3084 at_start = as; | |
| 3085 prevchr = pr; | |
| 3086 prevprevchr = prpr; | |
| 3087 } | |
| 3088 | |
| 3089 static int | |
| 3090 getchr() | |
| 3091 { | |
| 3092 int chr = peekchr(); | |
| 3093 | |
| 3094 skipchr(); | |
| 3095 return chr; | |
| 3096 } | |
| 3097 | |
| 3098 /* | |
| 3099 * put character back. Works only once! | |
| 3100 */ | |
| 3101 static void | |
| 3102 ungetchr() | |
| 3103 { | |
| 3104 nextchr = curchr; | |
| 3105 curchr = prevchr; | |
| 3106 prevchr = prevprevchr; | |
| 3107 at_start = prev_at_start; | |
| 3108 prev_at_start = FALSE; | |
| 3109 | |
| 3110 /* Backup regparse, so that it's at the same position as before the | |
| 3111 * getchr(). */ | |
| 3112 regparse -= prevchr_len; | |
| 3113 } | |
| 3114 | |
| 3115 /* | |
| 29 | 3116 * Get and return the value of the hex string at the current position. |
| 3117 * Return -1 if there is no valid hex number. | |
| 3118 * The position is updated: | |
| 24 | 3119 * blahblah\%x20asdf |
| 856 | 3120 * before-^ ^-after |
| 24 | 3121 * The parameter controls the maximum number of input characters. This will be |
| 3122 * 2 when reading a \%x20 sequence and 4 when reading a \%u20AC sequence. | |
| 3123 */ | |
| 3124 static int | |
| 3125 gethexchrs(maxinputlen) | |
| 3126 int maxinputlen; | |
| 3127 { | |
| 3128 int nr = 0; | |
| 3129 int c; | |
| 3130 int i; | |
| 3131 | |
| 3132 for (i = 0; i < maxinputlen; ++i) | |
| 3133 { | |
| 3134 c = regparse[0]; | |
| 3135 if (!vim_isxdigit(c)) | |
| 3136 break; | |
| 3137 nr <<= 4; | |
| 3138 nr |= hex2nr(c); | |
| 3139 ++regparse; | |
| 3140 } | |
| 3141 | |
| 3142 if (i == 0) | |
| 3143 return -1; | |
| 3144 return nr; | |
| 3145 } | |
| 3146 | |
| 3147 /* | |
| 3148 * get and return the value of the decimal string immediately after the | |
| 3149 * current position. Return -1 for invalid. Consumes all digits. | |
| 3150 */ | |
| 3151 static int | |
| 3152 getdecchrs() | |
| 3153 { | |
| 3154 int nr = 0; | |
| 3155 int c; | |
| 3156 int i; | |
| 3157 | |
| 3158 for (i = 0; ; ++i) | |
| 3159 { | |
| 3160 c = regparse[0]; | |
| 3161 if (c < '0' || c > '9') | |
| 3162 break; | |
| 3163 nr *= 10; | |
| 3164 nr += c - '0'; | |
| 3165 ++regparse; | |
| 3166 } | |
| 3167 | |
| 3168 if (i == 0) | |
| 3169 return -1; | |
| 3170 return nr; | |
| 3171 } | |
| 3172 | |
| 3173 /* | |
| 3174 * get and return the value of the octal string immediately after the current | |
| 3175 * position. Return -1 for invalid, or 0-255 for valid. Smart enough to handle | |
| 3176 * numbers > 377 correctly (for example, 400 is treated as 40) and doesn't | |
| 3177 * treat 8 or 9 as recognised characters. Position is updated: | |
| 3178 * blahblah\%o210asdf | |
| 856 | 3179 * before-^ ^-after |
| 24 | 3180 */ |
| 3181 static int | |
| 3182 getoctchrs() | |
| 3183 { | |
| 3184 int nr = 0; | |
| 3185 int c; | |
| 3186 int i; | |
| 3187 | |
| 3188 for (i = 0; i < 3 && nr < 040; ++i) | |
| 3189 { | |
| 3190 c = regparse[0]; | |
| 3191 if (c < '0' || c > '7') | |
| 3192 break; | |
| 3193 nr <<= 3; | |
| 3194 nr |= hex2nr(c); | |
| 3195 ++regparse; | |
| 3196 } | |
| 3197 | |
| 3198 if (i == 0) | |
| 3199 return -1; | |
| 3200 return nr; | |
| 3201 } | |
| 3202 | |
| 3203 /* | |
| 3204 * Get a number after a backslash that is inside []. | |
| 3205 * When nothing is recognized return a backslash. | |
| 3206 */ | |
| 3207 static int | |
| 3208 coll_get_char() | |
| 3209 { | |
| 3210 int nr = -1; | |
| 3211 | |
| 3212 switch (*regparse++) | |
| 3213 { | |
| 3214 case 'd': nr = getdecchrs(); break; | |
| 3215 case 'o': nr = getoctchrs(); break; | |
| 3216 case 'x': nr = gethexchrs(2); break; | |
| 3217 case 'u': nr = gethexchrs(4); break; | |
| 3218 case 'U': nr = gethexchrs(8); break; | |
| 3219 } | |
| 3220 if (nr < 0) | |
| 3221 { | |
| 3222 /* If getting the number fails be backwards compatible: the character | |
| 3223 * is a backslash. */ | |
| 3224 --regparse; | |
| 3225 nr = '\\'; | |
| 3226 } | |
| 3227 return nr; | |
| 3228 } | |
| 3229 | |
| 3230 /* | |
| 7 | 3231 * read_limits - Read two integers to be taken as a minimum and maximum. |
| 3232 * If the first character is '-', then the range is reversed. | |
| 3233 * Should end with 'end'. If minval is missing, zero is default, if maxval is | |
| 3234 * missing, a very big number is the default. | |
| 3235 */ | |
| 3236 static int | |
| 3237 read_limits(minval, maxval) | |
| 3238 long *minval; | |
| 3239 long *maxval; | |
| 3240 { | |
| 3241 int reverse = FALSE; | |
| 3242 char_u *first_char; | |
| 3243 long tmp; | |
| 3244 | |
| 3245 if (*regparse == '-') | |
| 3246 { | |
| 3247 /* Starts with '-', so reverse the range later */ | |
| 3248 regparse++; | |
| 3249 reverse = TRUE; | |
| 3250 } | |
| 3251 first_char = regparse; | |
| 3252 *minval = getdigits(®parse); | |
| 3253 if (*regparse == ',') /* There is a comma */ | |
| 3254 { | |
| 3255 if (vim_isdigit(*++regparse)) | |
| 3256 *maxval = getdigits(®parse); | |
| 3257 else | |
| 3258 *maxval = MAX_LIMIT; | |
| 3259 } | |
| 3260 else if (VIM_ISDIGIT(*first_char)) | |
| 3261 *maxval = *minval; /* It was \{n} or \{-n} */ | |
| 3262 else | |
| 3263 *maxval = MAX_LIMIT; /* It was \{} or \{-} */ | |
| 3264 if (*regparse == '\\') | |
| 3265 regparse++; /* Allow either \{...} or \{...\} */ | |
| 167 | 3266 if (*regparse != '}') |
| 7 | 3267 { |
| 3268 sprintf((char *)IObuff, _("E554: Syntax error in %s{...}"), | |
| 3269 reg_magic == MAGIC_ALL ? "" : "\\"); | |
| 3270 EMSG_RET_FAIL(IObuff); | |
| 3271 } | |
| 3272 | |
| 3273 /* | |
| 3274 * Reverse the range if there was a '-', or make sure it is in the right | |
| 3275 * order otherwise. | |
| 3276 */ | |
| 3277 if ((!reverse && *minval > *maxval) || (reverse && *minval < *maxval)) | |
| 3278 { | |
| 3279 tmp = *minval; | |
| 3280 *minval = *maxval; | |
| 3281 *maxval = tmp; | |
| 3282 } | |
| 3283 skipchr(); /* let's be friends with the lexer again */ | |
| 3284 return OK; | |
| 3285 } | |
| 3286 | |
| 3287 /* | |
| 3288 * vim_regexec and friends | |
| 3289 */ | |
| 3290 | |
| 3291 /* | |
| 3292 * Global work variables for vim_regexec(). | |
| 3293 */ | |
| 3294 | |
| 3295 /* The current match-position is remembered with these variables: */ | |
| 3296 static linenr_T reglnum; /* line number, relative to first line */ | |
| 3297 static char_u *regline; /* start of current line */ | |
| 3298 static char_u *reginput; /* current input, points into "regline" */ | |
| 3299 | |
| 3300 static int need_clear_subexpr; /* subexpressions still need to be | |
| 3301 * cleared */ | |
| 3302 #ifdef FEAT_SYN_HL | |
| 3303 static int need_clear_zsubexpr = FALSE; /* extmatch subexpressions | |
| 3304 * still need to be cleared */ | |
| 3305 #endif | |
| 3306 | |
| 3307 /* | |
| 3308 * Structure used to save the current input state, when it needs to be | |
| 3309 * restored after trying a match. Used by reg_save() and reg_restore(). | |
| 233 | 3310 * Also stores the length of "backpos". |
| 7 | 3311 */ |
| 3312 typedef struct | |
| 3313 { | |
| 3314 union | |
| 3315 { | |
| 3316 char_u *ptr; /* reginput pointer, for single-line regexp */ | |
| 3317 lpos_T pos; /* reginput pos, for multi-line regexp */ | |
| 3318 } rs_u; | |
| 233 | 3319 int rs_len; |
| 7 | 3320 } regsave_T; |
| 3321 | |
| 3322 /* struct to save start/end pointer/position in for \(\) */ | |
| 3323 typedef struct | |
| 3324 { | |
| 3325 union | |
| 3326 { | |
| 3327 char_u *ptr; | |
| 3328 lpos_T pos; | |
| 3329 } se_u; | |
| 3330 } save_se_T; | |
| 3331 | |
| 1579 | 3332 /* used for BEHIND and NOBEHIND matching */ |
| 3333 typedef struct regbehind_S | |
| 3334 { | |
| 3335 regsave_T save_after; | |
| 3336 regsave_T save_behind; | |
| 1602 | 3337 int save_need_clear_subexpr; |
| 1579 | 3338 save_se_T save_start[NSUBEXP]; |
| 3339 save_se_T save_end[NSUBEXP]; | |
| 3340 } regbehind_T; | |
| 3341 | |
| 7 | 3342 static char_u *reg_getline __ARGS((linenr_T lnum)); |
| 1521 | 3343 static long vim_regexec_both __ARGS((char_u *line, colnr_T col, proftime_T *tm)); |
| 7 | 3344 static long regtry __ARGS((regprog_T *prog, colnr_T col)); |
| 3345 static void cleanup_subexpr __ARGS((void)); | |
| 3346 #ifdef FEAT_SYN_HL | |
| 3347 static void cleanup_zsubexpr __ARGS((void)); | |
| 3348 #endif | |
| 1579 | 3349 static void save_subexpr __ARGS((regbehind_T *bp)); |
| 3350 static void restore_subexpr __ARGS((regbehind_T *bp)); | |
| 7 | 3351 static void reg_nextline __ARGS((void)); |
| 233 | 3352 static void reg_save __ARGS((regsave_T *save, garray_T *gap)); |
| 3353 static void reg_restore __ARGS((regsave_T *save, garray_T *gap)); | |
| 7 | 3354 static int reg_save_equal __ARGS((regsave_T *save)); |
| 3355 static void save_se_multi __ARGS((save_se_T *savep, lpos_T *posp)); | |
| 3356 static void save_se_one __ARGS((save_se_T *savep, char_u **pp)); | |
| 3357 | |
| 3358 /* Save the sub-expressions before attempting a match. */ | |
| 3359 #define save_se(savep, posp, pp) \ | |
| 3360 REG_MULTI ? save_se_multi((savep), (posp)) : save_se_one((savep), (pp)) | |
| 3361 | |
| 3362 /* After a failed match restore the sub-expressions. */ | |
| 3363 #define restore_se(savep, posp, pp) { \ | |
| 3364 if (REG_MULTI) \ | |
| 3365 *(posp) = (savep)->se_u.pos; \ | |
| 3366 else \ | |
| 3367 *(pp) = (savep)->se_u.ptr; } | |
| 3368 | |
| 3369 static int re_num_cmp __ARGS((long_u val, char_u *scan)); | |
| 180 | 3370 static int regmatch __ARGS((char_u *prog)); |
| 7 | 3371 static int regrepeat __ARGS((char_u *p, long maxcount)); |
| 3372 | |
| 3373 #ifdef DEBUG | |
| 3374 int regnarrate = 0; | |
| 3375 #endif | |
| 3376 | |
| 3377 /* | |
| 3378 * Internal copy of 'ignorecase'. It is set at each call to vim_regexec(). | |
| 3379 * Normally it gets the value of "rm_ic" or "rmm_ic", but when the pattern | |
| 3380 * contains '\c' or '\C' the value is overruled. | |
| 3381 */ | |
| 3382 static int ireg_ic; | |
| 3383 | |
| 3384 #ifdef FEAT_MBYTE | |
| 3385 /* | |
| 3386 * Similar to ireg_ic, but only for 'combining' characters. Set with \Z flag | |
| 3387 * in the regexp. Defaults to false, always. | |
| 3388 */ | |
| 3389 static int ireg_icombine; | |
| 3390 #endif | |
| 3391 | |
| 3392 /* | |
| 410 | 3393 * Copy of "rmm_maxcol": maximum column to search for a match. Zero when |
| 3394 * there is no maximum. | |
| 3395 */ | |
| 418 | 3396 static colnr_T ireg_maxcol; |
| 410 | 3397 |
| 3398 /* | |
| 7 | 3399 * Sometimes need to save a copy of a line. Since alloc()/free() is very |
| 3400 * slow, we keep one allocated piece of memory and only re-allocate it when | |
| 3401 * it's too small. It's freed in vim_regexec_both() when finished. | |
| 3402 */ | |
| 1468 | 3403 static char_u *reg_tofree = NULL; |
| 7 | 3404 static unsigned reg_tofreelen; |
| 3405 | |
| 3406 /* | |
| 3407 * These variables are set when executing a regexp to speed up the execution. | |
| 1209 | 3408 * Which ones are set depends on whether a single-line or multi-line match is |
| 7 | 3409 * done: |
| 3410 * single-line multi-line | |
| 3411 * reg_match ®match_T NULL | |
| 3412 * reg_mmatch NULL ®mmatch_T | |
| 3413 * reg_startp reg_match->startp <invalid> | |
| 3414 * reg_endp reg_match->endp <invalid> | |
| 3415 * reg_startpos <invalid> reg_mmatch->startpos | |
| 3416 * reg_endpos <invalid> reg_mmatch->endpos | |
| 3417 * reg_win NULL window in which to search | |
| 4061 | 3418 * reg_buf curbuf buffer in which to search |
| 7 | 3419 * reg_firstlnum <invalid> first line in which to search |
| 3420 * reg_maxline 0 last line nr | |
| 3421 * reg_line_lbr FALSE or TRUE FALSE | |
| 3422 */ | |
| 3423 static regmatch_T *reg_match; | |
| 3424 static regmmatch_T *reg_mmatch; | |
| 3425 static char_u **reg_startp = NULL; | |
| 3426 static char_u **reg_endp = NULL; | |
| 3427 static lpos_T *reg_startpos = NULL; | |
| 3428 static lpos_T *reg_endpos = NULL; | |
| 3429 static win_T *reg_win; | |
| 3430 static buf_T *reg_buf; | |
| 3431 static linenr_T reg_firstlnum; | |
| 3432 static linenr_T reg_maxline; | |
| 3433 static int reg_line_lbr; /* "\n" in string is line break */ | |
| 3434 | |
| 270 | 3435 /* Values for rs_state in regitem_T. */ |
| 3436 typedef enum regstate_E | |
| 3437 { | |
| 3438 RS_NOPEN = 0 /* NOPEN and NCLOSE */ | |
| 3439 , RS_MOPEN /* MOPEN + [0-9] */ | |
| 3440 , RS_MCLOSE /* MCLOSE + [0-9] */ | |
| 3441 #ifdef FEAT_SYN_HL | |
| 3442 , RS_ZOPEN /* ZOPEN + [0-9] */ | |
| 3443 , RS_ZCLOSE /* ZCLOSE + [0-9] */ | |
| 3444 #endif | |
| 3445 , RS_BRANCH /* BRANCH */ | |
| 3446 , RS_BRCPLX_MORE /* BRACE_COMPLEX and trying one more match */ | |
| 3447 , RS_BRCPLX_LONG /* BRACE_COMPLEX and trying longest match */ | |
| 3448 , RS_BRCPLX_SHORT /* BRACE_COMPLEX and trying shortest match */ | |
| 3449 , RS_NOMATCH /* NOMATCH */ | |
| 3450 , RS_BEHIND1 /* BEHIND / NOBEHIND matching rest */ | |
| 3451 , RS_BEHIND2 /* BEHIND / NOBEHIND matching behind part */ | |
| 3452 , RS_STAR_LONG /* STAR/PLUS/BRACE_SIMPLE longest match */ | |
| 3453 , RS_STAR_SHORT /* STAR/PLUS/BRACE_SIMPLE shortest match */ | |
| 3454 } regstate_T; | |
| 3455 | |
| 3456 /* | |
| 3457 * When there are alternatives a regstate_T is put on the regstack to remember | |
| 3458 * what we are doing. | |
| 3459 * Before it may be another type of item, depending on rs_state, to remember | |
| 3460 * more things. | |
| 3461 */ | |
| 3462 typedef struct regitem_S | |
| 3463 { | |
| 3464 regstate_T rs_state; /* what we are doing, one of RS_ above */ | |
| 3465 char_u *rs_scan; /* current node in program */ | |
| 3466 union | |
| 3467 { | |
| 3468 save_se_T sesave; | |
| 3469 regsave_T regsave; | |
| 3470 } rs_un; /* room for saving reginput */ | |
| 1579 | 3471 short rs_no; /* submatch nr or BEHIND/NOBEHIND */ |
| 270 | 3472 } regitem_T; |
| 3473 | |
| 3474 static regitem_T *regstack_push __ARGS((regstate_T state, char_u *scan)); | |
| 3475 static void regstack_pop __ARGS((char_u **scan)); | |
| 3476 | |
| 3477 /* used for STAR, PLUS and BRACE_SIMPLE matching */ | |
| 3478 typedef struct regstar_S | |
| 3479 { | |
| 3480 int nextb; /* next byte */ | |
| 3481 int nextb_ic; /* next byte reverse case */ | |
| 3482 long count; | |
| 3483 long minval; | |
| 3484 long maxval; | |
| 3485 } regstar_T; | |
| 3486 | |
| 3487 /* used to store input position when a BACK was encountered, so that we now if | |
| 3488 * we made any progress since the last time. */ | |
| 3489 typedef struct backpos_S | |
| 3490 { | |
| 3491 char_u *bp_scan; /* "scan" where BACK was encountered */ | |
| 3492 regsave_T bp_pos; /* last input position */ | |
| 3493 } backpos_T; | |
| 3494 | |
| 3495 /* | |
| 1520 | 3496 * "regstack" and "backpos" are used by regmatch(). They are kept over calls |
| 3497 * to avoid invoking malloc() and free() often. | |
| 3498 * "regstack" is a stack with regitem_T items, sometimes preceded by regstar_T | |
| 3499 * or regbehind_T. | |
| 3500 * "backpos_T" is a table with backpos_T for BACK | |
| 270 | 3501 */ |
| 1520 | 3502 static garray_T regstack = {0, 0, 0, 0, NULL}; |
| 3503 static garray_T backpos = {0, 0, 0, 0, NULL}; | |
| 3504 | |
| 3505 /* | |
| 3506 * Both for regstack and backpos tables we use the following strategy of | |
| 3507 * allocation (to reduce malloc/free calls): | |
| 3508 * - Initial size is fairly small. | |
| 3509 * - When needed, the tables are grown bigger (8 times at first, double after | |
| 3510 * that). | |
| 3511 * - After executing the match we free the memory only if the array has grown. | |
| 3512 * Thus the memory is kept allocated when it's at the initial size. | |
| 3513 * This makes it fast while not keeping a lot of memory allocated. | |
| 3514 * A three times speed increase was observed when using many simple patterns. | |
| 3515 */ | |
| 3516 #define REGSTACK_INITIAL 2048 | |
| 3517 #define BACKPOS_INITIAL 64 | |
| 3518 | |
| 3519 #if defined(EXITFREE) || defined(PROTO) | |
| 3520 void | |
| 3521 free_regexp_stuff() | |
| 3522 { | |
| 3523 ga_clear(®stack); | |
| 3524 ga_clear(&backpos); | |
| 3525 vim_free(reg_tofree); | |
| 3526 vim_free(reg_prev_sub); | |
| 3527 } | |
| 3528 #endif | |
| 270 | 3529 |
| 7 | 3530 /* |
| 3531 * Get pointer to the line "lnum", which is relative to "reg_firstlnum". | |
| 3532 */ | |
| 3533 static char_u * | |
| 3534 reg_getline(lnum) | |
| 3535 linenr_T lnum; | |
| 3536 { | |
| 3537 /* when looking behind for a match/no-match lnum is negative. But we | |
| 3538 * can't go before line 1 */ | |
| 3539 if (reg_firstlnum + lnum < 1) | |
| 3540 return NULL; | |
| 482 | 3541 if (lnum > reg_maxline) |
| 481 | 3542 /* Must have matched the "\n" in the last line. */ |
| 3543 return (char_u *)""; | |
| 7 | 3544 return ml_get_buf(reg_buf, reg_firstlnum + lnum, FALSE); |
| 3545 } | |
| 3546 | |
| 3547 static regsave_T behind_pos; | |
| 3548 | |
| 3549 #ifdef FEAT_SYN_HL | |
| 3550 static char_u *reg_startzp[NSUBEXP]; /* Workspace to mark beginning */ | |
| 3551 static char_u *reg_endzp[NSUBEXP]; /* and end of \z(...\) matches */ | |
| 3552 static lpos_T reg_startzpos[NSUBEXP]; /* idem, beginning pos */ | |
| 3553 static lpos_T reg_endzpos[NSUBEXP]; /* idem, end pos */ | |
| 3554 #endif | |
| 3555 | |
| 3556 /* TRUE if using multi-line regexp. */ | |
| 3557 #define REG_MULTI (reg_match == NULL) | |
| 3558 | |
| 3559 /* | |
| 3560 * Match a regexp against a string. | |
| 3561 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3562 * Uses curbuf for line count and 'iskeyword'. | |
| 3563 * | |
| 3564 * Return TRUE if there is a match, FALSE if not. | |
| 3565 */ | |
| 3566 int | |
| 3567 vim_regexec(rmp, line, col) | |
| 3568 regmatch_T *rmp; | |
| 3569 char_u *line; /* string to match against */ | |
| 3570 colnr_T col; /* column to start looking for match */ | |
| 3571 { | |
| 3572 reg_match = rmp; | |
| 3573 reg_mmatch = NULL; | |
| 3574 reg_maxline = 0; | |
| 3575 reg_line_lbr = FALSE; | |
| 4061 | 3576 reg_buf = curbuf; |
| 7 | 3577 reg_win = NULL; |
| 3578 ireg_ic = rmp->rm_ic; | |
| 3579 #ifdef FEAT_MBYTE | |
| 3580 ireg_icombine = FALSE; | |
| 3581 #endif | |
| 410 | 3582 ireg_maxcol = 0; |
| 1521 | 3583 return (vim_regexec_both(line, col, NULL) != 0); |
| 7 | 3584 } |
| 3585 | |
| 39 | 3586 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) \ |
| 3587 || defined(FIND_REPLACE_DIALOG) || defined(PROTO) | |
| 7 | 3588 /* |
| 3589 * Like vim_regexec(), but consider a "\n" in "line" to be a line break. | |
| 3590 */ | |
| 3591 int | |
| 3592 vim_regexec_nl(rmp, line, col) | |
| 3593 regmatch_T *rmp; | |
| 3594 char_u *line; /* string to match against */ | |
| 3595 colnr_T col; /* column to start looking for match */ | |
| 3596 { | |
| 3597 reg_match = rmp; | |
| 3598 reg_mmatch = NULL; | |
| 3599 reg_maxline = 0; | |
| 3600 reg_line_lbr = TRUE; | |
| 4061 | 3601 reg_buf = curbuf; |
| 7 | 3602 reg_win = NULL; |
| 3603 ireg_ic = rmp->rm_ic; | |
| 3604 #ifdef FEAT_MBYTE | |
| 3605 ireg_icombine = FALSE; | |
| 3606 #endif | |
| 410 | 3607 ireg_maxcol = 0; |
| 1521 | 3608 return (vim_regexec_both(line, col, NULL) != 0); |
| 7 | 3609 } |
| 3610 #endif | |
| 3611 | |
| 3612 /* | |
| 3613 * Match a regexp against multiple lines. | |
| 3614 * "rmp->regprog" is a compiled regexp as returned by vim_regcomp(). | |
| 3615 * Uses curbuf for line count and 'iskeyword'. | |
| 3616 * | |
| 3617 * Return zero if there is no match. Return number of lines contained in the | |
| 3618 * match otherwise. | |
| 3619 */ | |
| 3620 long | |
| 1521 | 3621 vim_regexec_multi(rmp, win, buf, lnum, col, tm) |
| 7 | 3622 regmmatch_T *rmp; |
| 3623 win_T *win; /* window in which to search or NULL */ | |
| 3624 buf_T *buf; /* buffer in which to search */ | |
| 3625 linenr_T lnum; /* nr of line to start looking for match */ | |
| 3626 colnr_T col; /* column to start looking for match */ | |
| 1521 | 3627 proftime_T *tm; /* timeout limit or NULL */ |
| 7 | 3628 { |
| 3629 long r; | |
| 3630 | |
| 3631 reg_match = NULL; | |
| 3632 reg_mmatch = rmp; | |
| 3633 reg_buf = buf; | |
| 3634 reg_win = win; | |
| 3635 reg_firstlnum = lnum; | |
| 3636 reg_maxline = reg_buf->b_ml.ml_line_count - lnum; | |
| 3637 reg_line_lbr = FALSE; | |
| 3638 ireg_ic = rmp->rmm_ic; | |
| 3639 #ifdef FEAT_MBYTE | |
| 3640 ireg_icombine = FALSE; | |
| 3641 #endif | |
| 410 | 3642 ireg_maxcol = rmp->rmm_maxcol; |
| 7 | 3643 |
| 1521 | 3644 r = vim_regexec_both(NULL, col, tm); |
| 7 | 3645 |
| 3646 return r; | |
| 3647 } | |
| 3648 | |
| 3649 /* | |
| 3650 * Match a regexp against a string ("line" points to the string) or multiple | |
| 3651 * lines ("line" is NULL, use reg_getline()). | |
| 3652 */ | |
| 3653 static long | |
| 1521 | 3654 vim_regexec_both(line, col, tm) |
| 7 | 3655 char_u *line; |
| 3656 colnr_T col; /* column to start looking for match */ | |
| 1877 | 3657 proftime_T *tm UNUSED; /* timeout limit or NULL */ |
| 7 | 3658 { |
| 3659 regprog_T *prog; | |
| 3660 char_u *s; | |
| 189 | 3661 long retval = 0L; |
| 7 | 3662 |
| 1520 | 3663 /* Create "regstack" and "backpos" if they are not allocated yet. |
| 3664 * We allocate *_INITIAL amount of bytes first and then set the grow size | |
| 3665 * to much bigger value to avoid many malloc calls in case of deep regular | |
| 3666 * expressions. */ | |
| 3667 if (regstack.ga_data == NULL) | |
| 3668 { | |
| 3669 /* Use an item size of 1 byte, since we push different things | |
| 3670 * onto the regstack. */ | |
| 3671 ga_init2(®stack, 1, REGSTACK_INITIAL); | |
| 3672 ga_grow(®stack, REGSTACK_INITIAL); | |
| 3673 regstack.ga_growsize = REGSTACK_INITIAL * 8; | |
| 3674 } | |
| 3675 | |
| 3676 if (backpos.ga_data == NULL) | |
| 3677 { | |
| 3678 ga_init2(&backpos, sizeof(backpos_T), BACKPOS_INITIAL); | |
| 3679 ga_grow(&backpos, BACKPOS_INITIAL); | |
| 3680 backpos.ga_growsize = BACKPOS_INITIAL * 8; | |
| 3681 } | |
| 270 | 3682 |
| 7 | 3683 if (REG_MULTI) |
| 3684 { | |
| 3685 prog = reg_mmatch->regprog; | |
| 3686 line = reg_getline((linenr_T)0); | |
| 3687 reg_startpos = reg_mmatch->startpos; | |
| 3688 reg_endpos = reg_mmatch->endpos; | |
| 3689 } | |
| 3690 else | |
| 3691 { | |
| 3692 prog = reg_match->regprog; | |
| 3693 reg_startp = reg_match->startp; | |
| 3694 reg_endp = reg_match->endp; | |
| 3695 } | |
| 3696 | |
| 3697 /* Be paranoid... */ | |
| 3698 if (prog == NULL || line == NULL) | |
| 3699 { | |
| 3700 EMSG(_(e_null)); | |
| 3701 goto theend; | |
| 3702 } | |
| 3703 | |
| 3704 /* Check validity of program. */ | |
| 3705 if (prog_magic_wrong()) | |
| 3706 goto theend; | |
| 3707 | |
| 410 | 3708 /* If the start column is past the maximum column: no need to try. */ |
| 3709 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3710 goto theend; | |
| 3711 | |
| 7 | 3712 /* If pattern contains "\c" or "\C": overrule value of ireg_ic */ |
| 3713 if (prog->regflags & RF_ICASE) | |
| 3714 ireg_ic = TRUE; | |
| 3715 else if (prog->regflags & RF_NOICASE) | |
| 3716 ireg_ic = FALSE; | |
| 3717 | |
| 3718 #ifdef FEAT_MBYTE | |
| 3719 /* If pattern contains "\Z" overrule value of ireg_icombine */ | |
| 3720 if (prog->regflags & RF_ICOMBINE) | |
| 3721 ireg_icombine = TRUE; | |
| 3722 #endif | |
| 3723 | |
| 3724 /* If there is a "must appear" string, look for it. */ | |
| 3725 if (prog->regmust != NULL) | |
| 3726 { | |
| 3727 int c; | |
| 3728 | |
| 3729 #ifdef FEAT_MBYTE | |
| 3730 if (has_mbyte) | |
| 3731 c = (*mb_ptr2char)(prog->regmust); | |
| 3732 else | |
| 3733 #endif | |
| 3734 c = *prog->regmust; | |
| 3735 s = line + col; | |
| 170 | 3736 |
| 3737 /* | |
| 3738 * This is used very often, esp. for ":global". Use three versions of | |
| 3739 * the loop to avoid overhead of conditions. | |
| 3740 */ | |
| 3741 if (!ireg_ic | |
| 3742 #ifdef FEAT_MBYTE | |
| 3743 && !has_mbyte | |
| 3744 #endif | |
| 3745 ) | |
| 3746 while ((s = vim_strbyte(s, c)) != NULL) | |
| 3747 { | |
| 3748 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3749 break; /* Found it. */ | |
| 3750 ++s; | |
| 3751 } | |
| 3752 #ifdef FEAT_MBYTE | |
| 3753 else if (!ireg_ic || (!enc_utf8 && mb_char2len(c) > 1)) | |
| 3754 while ((s = vim_strchr(s, c)) != NULL) | |
| 3755 { | |
| 3756 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3757 break; /* Found it. */ | |
| 3758 mb_ptr_adv(s); | |
| 3759 } | |
| 3760 #endif | |
| 3761 else | |
| 3762 while ((s = cstrchr(s, c)) != NULL) | |
| 3763 { | |
| 3764 if (cstrncmp(s, prog->regmust, &prog->regmlen) == 0) | |
| 3765 break; /* Found it. */ | |
| 3766 mb_ptr_adv(s); | |
| 3767 } | |
| 7 | 3768 if (s == NULL) /* Not present. */ |
| 3769 goto theend; | |
| 3770 } | |
| 3771 | |
| 3772 regline = line; | |
| 3773 reglnum = 0; | |
| 2578 | 3774 reg_toolong = FALSE; |
| 7 | 3775 |
| 3776 /* Simplest case: Anchored match need be tried only once. */ | |
| 3777 if (prog->reganch) | |
| 3778 { | |
| 3779 int c; | |
| 3780 | |
| 3781 #ifdef FEAT_MBYTE | |
| 3782 if (has_mbyte) | |
| 3783 c = (*mb_ptr2char)(regline + col); | |
| 3784 else | |
| 3785 #endif | |
| 3786 c = regline[col]; | |
| 3787 if (prog->regstart == NUL | |
| 3788 || prog->regstart == c | |
| 3789 || (ireg_ic && (( | |
| 3790 #ifdef FEAT_MBYTE | |
| 3791 (enc_utf8 && utf_fold(prog->regstart) == utf_fold(c))) | |
| 3792 || (c < 255 && prog->regstart < 255 && | |
| 3793 #endif | |
| 1347 | 3794 MB_TOLOWER(prog->regstart) == MB_TOLOWER(c))))) |
| 7 | 3795 retval = regtry(prog, col); |
| 3796 else | |
| 3797 retval = 0; | |
| 3798 } | |
| 3799 else | |
| 3800 { | |
| 1521 | 3801 #ifdef FEAT_RELTIME |
| 3802 int tm_count = 0; | |
| 3803 #endif | |
| 7 | 3804 /* Messy cases: unanchored match. */ |
| 180 | 3805 while (!got_int) |
| 7 | 3806 { |
| 3807 if (prog->regstart != NUL) | |
| 3808 { | |
| 170 | 3809 /* Skip until the char we know it must start with. |
| 3810 * Used often, do some work to avoid call overhead. */ | |
| 3811 if (!ireg_ic | |
| 3812 #ifdef FEAT_MBYTE | |
| 3813 && !has_mbyte | |
| 3814 #endif | |
| 3815 ) | |
| 3816 s = vim_strbyte(regline + col, prog->regstart); | |
| 3817 else | |
| 3818 s = cstrchr(regline + col, prog->regstart); | |
| 7 | 3819 if (s == NULL) |
| 3820 { | |
| 3821 retval = 0; | |
| 3822 break; | |
| 3823 } | |
| 3824 col = (int)(s - regline); | |
| 3825 } | |
| 3826 | |
| 410 | 3827 /* Check for maximum column to try. */ |
| 3828 if (ireg_maxcol > 0 && col >= ireg_maxcol) | |
| 3829 { | |
| 3830 retval = 0; | |
| 3831 break; | |
| 3832 } | |
| 3833 | |
| 7 | 3834 retval = regtry(prog, col); |
| 3835 if (retval > 0) | |
| 3836 break; | |
| 3837 | |
| 3838 /* if not currently on the first line, get it again */ | |
| 3839 if (reglnum != 0) | |
| 3840 { | |
| 481 | 3841 reglnum = 0; |
| 7 | 3842 regline = reg_getline((linenr_T)0); |
| 3843 } | |
| 3844 if (regline[col] == NUL) | |
| 3845 break; | |
| 3846 #ifdef FEAT_MBYTE | |
| 3847 if (has_mbyte) | |
| 474 | 3848 col += (*mb_ptr2len)(regline + col); |
| 7 | 3849 else |
| 3850 #endif | |
| 3851 ++col; | |
| 1521 | 3852 #ifdef FEAT_RELTIME |
| 3853 /* Check for timeout once in a twenty times to avoid overhead. */ | |
| 3854 if (tm != NULL && ++tm_count == 20) | |
| 3855 { | |
| 3856 tm_count = 0; | |
| 3857 if (profile_passed_limit(tm)) | |
| 3858 break; | |
| 3859 } | |
| 3860 #endif | |
| 7 | 3861 } |
| 3862 } | |
| 3863 | |
| 3864 theend: | |
| 1520 | 3865 /* Free "reg_tofree" when it's a bit big. |
| 3866 * Free regstack and backpos if they are bigger than their initial size. */ | |
| 3867 if (reg_tofreelen > 400) | |
| 3868 { | |
| 3869 vim_free(reg_tofree); | |
| 3870 reg_tofree = NULL; | |
| 3871 } | |
| 3872 if (regstack.ga_maxlen > REGSTACK_INITIAL) | |
| 3873 ga_clear(®stack); | |
| 3874 if (backpos.ga_maxlen > BACKPOS_INITIAL) | |
| 3875 ga_clear(&backpos); | |
| 270 | 3876 |
| 7 | 3877 return retval; |
| 3878 } | |
| 3879 | |
| 3880 #ifdef FEAT_SYN_HL | |
| 3881 static reg_extmatch_T *make_extmatch __ARGS((void)); | |
| 3882 | |
| 3883 /* | |
| 3884 * Create a new extmatch and mark it as referenced once. | |
| 3885 */ | |
| 3886 static reg_extmatch_T * | |
| 3887 make_extmatch() | |
| 3888 { | |
| 3889 reg_extmatch_T *em; | |
| 3890 | |
| 3891 em = (reg_extmatch_T *)alloc_clear((unsigned)sizeof(reg_extmatch_T)); | |
| 3892 if (em != NULL) | |
| 3893 em->refcnt = 1; | |
| 3894 return em; | |
| 3895 } | |
| 3896 | |
| 3897 /* | |
| 3898 * Add a reference to an extmatch. | |
| 3899 */ | |
| 3900 reg_extmatch_T * | |
| 3901 ref_extmatch(em) | |
| 3902 reg_extmatch_T *em; | |
| 3903 { | |
| 3904 if (em != NULL) | |
| 3905 em->refcnt++; | |
| 3906 return em; | |
| 3907 } | |
| 3908 | |
| 3909 /* | |
| 3910 * Remove a reference to an extmatch. If there are no references left, free | |
| 3911 * the info. | |
| 3912 */ | |
| 3913 void | |
| 3914 unref_extmatch(em) | |
| 3915 reg_extmatch_T *em; | |
| 3916 { | |
| 3917 int i; | |
| 3918 | |
| 3919 if (em != NULL && --em->refcnt <= 0) | |
| 3920 { | |
| 3921 for (i = 0; i < NSUBEXP; ++i) | |
| 3922 vim_free(em->matches[i]); | |
| 3923 vim_free(em); | |
| 3924 } | |
| 3925 } | |
| 3926 #endif | |
| 3927 | |
| 3928 /* | |
| 3929 * regtry - try match of "prog" with at regline["col"]. | |
| 3930 * Returns 0 for failure, number of lines contained in the match otherwise. | |
| 3931 */ | |
| 3932 static long | |
| 3933 regtry(prog, col) | |
| 3934 regprog_T *prog; | |
| 3935 colnr_T col; | |
| 3936 { | |
| 3937 reginput = regline + col; | |
| 3938 need_clear_subexpr = TRUE; | |
| 3939 #ifdef FEAT_SYN_HL | |
| 3940 /* Clear the external match subpointers if necessary. */ | |
| 3941 if (prog->reghasz == REX_SET) | |
| 3942 need_clear_zsubexpr = TRUE; | |
| 3943 #endif | |
| 3944 | |
| 189 | 3945 if (regmatch(prog->program + 1) == 0) |
| 3946 return 0; | |
| 3947 | |
| 3948 cleanup_subexpr(); | |
| 3949 if (REG_MULTI) | |
| 7 | 3950 { |
| 189 | 3951 if (reg_startpos[0].lnum < 0) |
| 7 | 3952 { |
| 189 | 3953 reg_startpos[0].lnum = 0; |
| 3954 reg_startpos[0].col = col; | |
| 3955 } | |
| 3956 if (reg_endpos[0].lnum < 0) | |
| 3957 { | |
| 3958 reg_endpos[0].lnum = reglnum; | |
| 3959 reg_endpos[0].col = (int)(reginput - regline); | |
| 7 | 3960 } |
| 3961 else | |
| 189 | 3962 /* Use line number of "\ze". */ |
| 3963 reglnum = reg_endpos[0].lnum; | |
| 3964 } | |
| 3965 else | |
| 3966 { | |
| 3967 if (reg_startp[0] == NULL) | |
| 3968 reg_startp[0] = regline + col; | |
| 3969 if (reg_endp[0] == NULL) | |
| 3970 reg_endp[0] = reginput; | |
| 3971 } | |
| 7 | 3972 #ifdef FEAT_SYN_HL |
| 189 | 3973 /* Package any found \z(...\) matches for export. Default is none. */ |
| 3974 unref_extmatch(re_extmatch_out); | |
| 3975 re_extmatch_out = NULL; | |
| 3976 | |
| 3977 if (prog->reghasz == REX_SET) | |
| 3978 { | |
| 3979 int i; | |
| 3980 | |
| 3981 cleanup_zsubexpr(); | |
| 3982 re_extmatch_out = make_extmatch(); | |
| 3983 for (i = 0; i < NSUBEXP; i++) | |
| 7 | 3984 { |
| 189 | 3985 if (REG_MULTI) |
| 7 | 3986 { |
| 189 | 3987 /* Only accept single line matches. */ |
| 3988 if (reg_startzpos[i].lnum >= 0 | |
| 3989 && reg_endzpos[i].lnum == reg_startzpos[i].lnum) | |
| 3990 re_extmatch_out->matches[i] = | |
| 3991 vim_strnsave(reg_getline(reg_startzpos[i].lnum) | |
| 7 | 3992 + reg_startzpos[i].col, |
| 189 | 3993 reg_endzpos[i].col - reg_startzpos[i].col); |
| 3994 } | |
| 3995 else | |
| 3996 { | |
| 3997 if (reg_startzp[i] != NULL && reg_endzp[i] != NULL) | |
| 3998 re_extmatch_out->matches[i] = | |
| 7 | 3999 vim_strnsave(reg_startzp[i], |
| 189 | 4000 (int)(reg_endzp[i] - reg_startzp[i])); |
| 7 | 4001 } |
| 4002 } | |
| 189 | 4003 } |
| 7 | 4004 #endif |
| 189 | 4005 return 1 + reglnum; |
| 7 | 4006 } |
| 4007 | |
| 4008 #ifdef FEAT_MBYTE | |
| 4009 static int reg_prev_class __ARGS((void)); | |
| 4010 | |
| 4011 /* | |
| 4012 * Get class of previous character. | |
| 4013 */ | |
| 4014 static int | |
| 4015 reg_prev_class() | |
| 4016 { | |
| 4017 if (reginput > regline) | |
| 4069 | 4018 return mb_get_class_buf(reginput - 1 |
| 4019 - (*mb_head_off)(regline, reginput - 1), reg_buf); | |
| 7 | 4020 return -1; |
| 4021 } | |
| 4022 | |
| 4023 #endif | |
| 39 | 4024 #define ADVANCE_REGINPUT() mb_ptr_adv(reginput) |
| 7 | 4025 |
| 4026 /* | |
| 4027 * The arguments from BRACE_LIMITS are stored here. They are actually local | |
| 4028 * to regmatch(), but they are here to reduce the amount of stack space used | |
| 4029 * (it can be called recursively many times). | |
| 4030 */ | |
| 4031 static long bl_minval; | |
| 4032 static long bl_maxval; | |
| 4033 | |
| 4034 /* | |
| 4035 * regmatch - main matching routine | |
| 4036 * | |
| 180 | 4037 * Conceptually the strategy is simple: Check to see whether the current node |
| 4038 * matches, push an item onto the regstack and loop to see whether the rest | |
| 4039 * matches, and then act accordingly. In practice we make some effort to | |
| 4040 * avoid using the regstack, in particular by going through "ordinary" nodes | |
| 4041 * (that don't need to know whether the rest of the match failed) by a nested | |
| 4042 * loop. | |
| 7 | 4043 * |
| 4044 * Returns TRUE when there is a match. Leaves reginput and reglnum just after | |
| 4045 * the last matched character. | |
| 4046 * Returns FALSE when there is no match. Leaves reginput and reglnum in an | |
| 4047 * undefined state! | |
| 4048 */ | |
| 4049 static int | |
| 180 | 4050 regmatch(scan) |
| 7 | 4051 char_u *scan; /* Current node. */ |
| 4052 { | |
| 180 | 4053 char_u *next; /* Next node. */ |
| 4054 int op; | |
| 4055 int c; | |
| 4056 regitem_T *rp; | |
| 4057 int no; | |
| 4058 int status; /* one of the RA_ values: */ | |
| 4059 #define RA_FAIL 1 /* something failed, abort */ | |
| 4060 #define RA_CONT 2 /* continue in inner loop */ | |
| 4061 #define RA_BREAK 3 /* break inner loop */ | |
| 4062 #define RA_MATCH 4 /* successful match */ | |
| 4063 #define RA_NOMATCH 5 /* didn't match */ | |
| 270 | 4064 |
| 1520 | 4065 /* Make "regstack" and "backpos" empty. They are allocated and freed in |
| 4066 * vim_regexec_both() to reduce malloc()/free() calls. */ | |
| 270 | 4067 regstack.ga_len = 0; |
| 4068 backpos.ga_len = 0; | |
| 233 | 4069 |
| 180 | 4070 /* |
| 233 | 4071 * Repeat until "regstack" is empty. |
| 180 | 4072 */ |
| 4073 for (;;) | |
| 4074 { | |
| 7 | 4075 /* Some patterns my cause a long time to match, even though they are not |
| 4076 * illegal. E.g., "\([a-z]\+\)\+Q". Allow breaking them with CTRL-C. */ | |
| 4077 fast_breakcheck(); | |
| 4078 | |
| 4079 #ifdef DEBUG | |
| 4080 if (scan != NULL && regnarrate) | |
| 4081 { | |
| 4082 mch_errmsg(regprop(scan)); | |
| 4083 mch_errmsg("(\n"); | |
| 4084 } | |
| 4085 #endif | |
| 180 | 4086 |
| 4087 /* | |
| 233 | 4088 * Repeat for items that can be matched sequentially, without using the |
| 180 | 4089 * regstack. |
| 4090 */ | |
| 4091 for (;;) | |
| 7 | 4092 { |
| 180 | 4093 if (got_int || scan == NULL) |
| 4094 { | |
| 4095 status = RA_FAIL; | |
| 4096 break; | |
| 4097 } | |
| 4098 status = RA_CONT; | |
| 4099 | |
| 7 | 4100 #ifdef DEBUG |
| 4101 if (regnarrate) | |
| 4102 { | |
| 4103 mch_errmsg(regprop(scan)); | |
| 4104 mch_errmsg("...\n"); | |
| 4105 # ifdef FEAT_SYN_HL | |
| 4106 if (re_extmatch_in != NULL) | |
| 4107 { | |
| 4108 int i; | |
| 4109 | |
| 4110 mch_errmsg(_("External submatches:\n")); | |
| 4111 for (i = 0; i < NSUBEXP; i++) | |
| 4112 { | |
| 4113 mch_errmsg(" \""); | |
| 4114 if (re_extmatch_in->matches[i] != NULL) | |
| 4115 mch_errmsg(re_extmatch_in->matches[i]); | |
| 4116 mch_errmsg("\"\n"); | |
| 4117 } | |
| 4118 } | |
| 4119 # endif | |
| 4120 } | |
| 4121 #endif | |
| 4122 next = regnext(scan); | |
| 4123 | |
| 4124 op = OP(scan); | |
| 4125 /* Check for character class with NL added. */ | |
| 1018 | 4126 if (!reg_line_lbr && WITH_NL(op) && REG_MULTI |
| 4127 && *reginput == NUL && reglnum <= reg_maxline) | |
| 7 | 4128 { |
| 4129 reg_nextline(); | |
| 4130 } | |
| 4131 else if (reg_line_lbr && WITH_NL(op) && *reginput == '\n') | |
| 4132 { | |
| 4133 ADVANCE_REGINPUT(); | |
| 4134 } | |
| 4135 else | |
| 4136 { | |
| 4137 if (WITH_NL(op)) | |
| 180 | 4138 op -= ADD_NL; |
| 7 | 4139 #ifdef FEAT_MBYTE |
| 4140 if (has_mbyte) | |
| 4141 c = (*mb_ptr2char)(reginput); | |
| 4142 else | |
| 4143 #endif | |
| 4144 c = *reginput; | |
| 4145 switch (op) | |
| 4146 { | |
| 4147 case BOL: | |
| 4148 if (reginput != regline) | |
| 180 | 4149 status = RA_NOMATCH; |
| 7 | 4150 break; |
| 4151 | |
| 4152 case EOL: | |
| 4153 if (c != NUL) | |
| 180 | 4154 status = RA_NOMATCH; |
| 7 | 4155 break; |
| 4156 | |
| 4157 case RE_BOF: | |
| 1458 | 4158 /* We're not at the beginning of the file when below the first |
| 4159 * line where we started, not at the start of the line or we | |
| 4160 * didn't start at the first line of the buffer. */ | |
| 7 | 4161 if (reglnum != 0 || reginput != regline |
| 1458 | 4162 || (REG_MULTI && reg_firstlnum > 1)) |
| 180 | 4163 status = RA_NOMATCH; |
| 7 | 4164 break; |
| 4165 | |
| 4166 case RE_EOF: | |
| 4167 if (reglnum != reg_maxline || c != NUL) | |
| 180 | 4168 status = RA_NOMATCH; |
| 7 | 4169 break; |
| 4170 | |
| 4171 case CURSOR: | |
| 4172 /* Check if the buffer is in a window and compare the | |
| 4173 * reg_win->w_cursor position to the match position. */ | |
| 4174 if (reg_win == NULL | |
| 4175 || (reglnum + reg_firstlnum != reg_win->w_cursor.lnum) | |
| 4176 || ((colnr_T)(reginput - regline) != reg_win->w_cursor.col)) | |
| 180 | 4177 status = RA_NOMATCH; |
| 7 | 4178 break; |
| 4179 | |
| 639 | 4180 case RE_MARK: |
| 4181 /* Compare the mark position to the match position. NOTE: Always | |
| 4182 * uses the current buffer. */ | |
| 4183 { | |
| 4184 int mark = OPERAND(scan)[0]; | |
| 4185 int cmp = OPERAND(scan)[1]; | |
| 4186 pos_T *pos; | |
| 4187 | |
| 4043 | 4188 pos = getmark_buf(reg_buf, mark, FALSE); |
| 1148 | 4189 if (pos == NULL /* mark doesn't exist */ |
| 639 | 4190 || pos->lnum <= 0 /* mark isn't set (in curbuf) */ |
| 4191 || (pos->lnum == reglnum + reg_firstlnum | |
| 4192 ? (pos->col == (colnr_T)(reginput - regline) | |
| 4193 ? (cmp == '<' || cmp == '>') | |
| 4194 : (pos->col < (colnr_T)(reginput - regline) | |
| 4195 ? cmp != '>' | |
| 4196 : cmp != '<')) | |
| 4197 : (pos->lnum < reglnum + reg_firstlnum | |
| 4198 ? cmp != '>' | |
| 4199 : cmp != '<'))) | |
| 4200 status = RA_NOMATCH; | |
| 4201 } | |
| 4202 break; | |
| 4203 | |
| 4204 case RE_VISUAL: | |
| 4205 #ifdef FEAT_VISUAL | |
| 4206 /* Check if the buffer is the current buffer. and whether the | |
| 4207 * position is inside the Visual area. */ | |
| 4208 if (reg_buf != curbuf || VIsual.lnum == 0) | |
| 4209 status = RA_NOMATCH; | |
| 4210 else | |
| 4211 { | |
| 4212 pos_T top, bot; | |
| 4213 linenr_T lnum; | |
| 4214 colnr_T col; | |
| 4215 win_T *wp = reg_win == NULL ? curwin : reg_win; | |
| 4216 int mode; | |
| 4217 | |
| 4218 if (VIsual_active) | |
| 4219 { | |
| 4220 if (lt(VIsual, wp->w_cursor)) | |
| 4221 { | |
| 4222 top = VIsual; | |
| 4223 bot = wp->w_cursor; | |
| 4224 } | |
| 4225 else | |
| 4226 { | |
| 4227 top = wp->w_cursor; | |
| 4228 bot = VIsual; | |
| 4229 } | |
| 4230 mode = VIsual_mode; | |
| 4231 } | |
| 4232 else | |
| 4233 { | |
| 692 | 4234 if (lt(curbuf->b_visual.vi_start, curbuf->b_visual.vi_end)) |
| 685 | 4235 { |
| 692 | 4236 top = curbuf->b_visual.vi_start; |
| 4237 bot = curbuf->b_visual.vi_end; | |
| 685 | 4238 } |
| 4239 else | |
| 4240 { | |
| 692 | 4241 top = curbuf->b_visual.vi_end; |
| 4242 bot = curbuf->b_visual.vi_start; | |
| 685 | 4243 } |
| 692 | 4244 mode = curbuf->b_visual.vi_mode; |
| 639 | 4245 } |
| 4246 lnum = reglnum + reg_firstlnum; | |
| 4247 col = (colnr_T)(reginput - regline); | |
| 4248 if (lnum < top.lnum || lnum > bot.lnum) | |
| 4249 status = RA_NOMATCH; | |
| 4250 else if (mode == 'v') | |
| 4251 { | |
| 4252 if ((lnum == top.lnum && col < top.col) | |
| 4253 || (lnum == bot.lnum | |
| 4254 && col >= bot.col + (*p_sel != 'e'))) | |
| 4255 status = RA_NOMATCH; | |
| 4256 } | |
| 4257 else if (mode == Ctrl_V) | |
| 4258 { | |
| 4259 colnr_T start, end; | |
| 4260 colnr_T start2, end2; | |
| 944 | 4261 colnr_T cols; |
| 639 | 4262 |
| 4263 getvvcol(wp, &top, &start, NULL, &end); | |
| 4264 getvvcol(wp, &bot, &start2, NULL, &end2); | |
| 4265 if (start2 < start) | |
| 4266 start = start2; | |
| 4267 if (end2 > end) | |
| 4268 end = end2; | |
| 4269 if (top.col == MAXCOL || bot.col == MAXCOL) | |
| 4270 end = MAXCOL; | |
| 944 | 4271 cols = win_linetabsize(wp, |
| 639 | 4272 regline, (colnr_T)(reginput - regline)); |
| 944 | 4273 if (cols < start || cols > end - (*p_sel == 'e')) |
| 639 | 4274 status = RA_NOMATCH; |
| 4275 } | |
| 4276 } | |
| 4277 #else | |
| 4278 status = RA_NOMATCH; | |
| 4279 #endif | |
| 4280 break; | |
| 4281 | |
| 7 | 4282 case RE_LNUM: |
| 4283 if (!REG_MULTI || !re_num_cmp((long_u)(reglnum + reg_firstlnum), | |
| 4284 scan)) | |
| 180 | 4285 status = RA_NOMATCH; |
| 7 | 4286 break; |
| 4287 | |
| 4288 case RE_COL: | |
| 4289 if (!re_num_cmp((long_u)(reginput - regline) + 1, scan)) | |
| 180 | 4290 status = RA_NOMATCH; |
| 7 | 4291 break; |
| 4292 | |
| 4293 case RE_VCOL: | |
| 4294 if (!re_num_cmp((long_u)win_linetabsize( | |
| 4295 reg_win == NULL ? curwin : reg_win, | |
| 4296 regline, (colnr_T)(reginput - regline)) + 1, scan)) | |
| 180 | 4297 status = RA_NOMATCH; |
| 7 | 4298 break; |
| 4299 | |
| 4300 case BOW: /* \<word; reginput points to w */ | |
| 4301 if (c == NUL) /* Can't match at end of line */ | |
| 180 | 4302 status = RA_NOMATCH; |
| 7 | 4303 #ifdef FEAT_MBYTE |
| 180 | 4304 else if (has_mbyte) |
| 7 | 4305 { |
| 4306 int this_class; | |
| 4307 | |
| 4308 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4309 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4310 if (this_class <= 1) |
| 180 | 4311 status = RA_NOMATCH; /* not on a word at all */ |
| 4312 else if (reg_prev_class() == this_class) | |
| 4313 status = RA_NOMATCH; /* previous char is in same word */ | |
| 7 | 4314 } |
| 4315 #endif | |
| 4316 else | |
| 4317 { | |
| 4061 | 4318 if (!vim_iswordc_buf(c, reg_buf) || (reginput > regline |
| 4319 && vim_iswordc_buf(reginput[-1], reg_buf))) | |
| 180 | 4320 status = RA_NOMATCH; |
| 7 | 4321 } |
| 4322 break; | |
| 4323 | |
| 4324 case EOW: /* word\>; reginput points after d */ | |
| 4325 if (reginput == regline) /* Can't match at start of line */ | |
| 180 | 4326 status = RA_NOMATCH; |
| 7 | 4327 #ifdef FEAT_MBYTE |
| 180 | 4328 else if (has_mbyte) |
| 7 | 4329 { |
| 4330 int this_class, prev_class; | |
| 4331 | |
| 4332 /* Get class of current and previous char (if it exists). */ | |
| 4069 | 4333 this_class = mb_get_class_buf(reginput, reg_buf); |
| 7 | 4334 prev_class = reg_prev_class(); |
| 180 | 4335 if (this_class == prev_class |
| 4336 || prev_class == 0 || prev_class == 1) | |
| 4337 status = RA_NOMATCH; | |
| 7 | 4338 } |
| 180 | 4339 #endif |
| 7 | 4340 else |
| 4341 { | |
| 4043 | 4342 if (!vim_iswordc_buf(reginput[-1], reg_buf) |
| 4343 || (reginput[0] != NUL && vim_iswordc_buf(c, reg_buf))) | |
| 180 | 4344 status = RA_NOMATCH; |
| 7 | 4345 } |
| 4346 break; /* Matched with EOW */ | |
| 4347 | |
| 4348 case ANY: | |
| 4084 | 4349 /* ANY does not match new lines. */ |
| 7 | 4350 if (c == NUL) |
| 180 | 4351 status = RA_NOMATCH; |
| 4352 else | |
| 4353 ADVANCE_REGINPUT(); | |
| 7 | 4354 break; |
| 4355 | |
| 4356 case IDENT: | |
| 4357 if (!vim_isIDc(c)) | |
| 180 | 4358 status = RA_NOMATCH; |
| 4359 else | |
| 4360 ADVANCE_REGINPUT(); | |
| 7 | 4361 break; |
| 4362 | |
| 4363 case SIDENT: | |
| 4364 if (VIM_ISDIGIT(*reginput) || !vim_isIDc(c)) | |
| 180 | 4365 status = RA_NOMATCH; |
| 4366 else | |
| 4367 ADVANCE_REGINPUT(); | |
| 7 | 4368 break; |
| 4369 | |
| 4370 case KWORD: | |
| 4069 | 4371 if (!vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4372 status = RA_NOMATCH; |
| 4373 else | |
| 4374 ADVANCE_REGINPUT(); | |
| 7 | 4375 break; |
| 4376 | |
| 4377 case SKWORD: | |
| 4069 | 4378 if (VIM_ISDIGIT(*reginput) || !vim_iswordp_buf(reginput, reg_buf)) |
| 180 | 4379 status = RA_NOMATCH; |
| 4380 else | |
| 4381 ADVANCE_REGINPUT(); | |
| 7 | 4382 break; |
| 4383 | |
| 4384 case FNAME: | |
| 4385 if (!vim_isfilec(c)) | |
| 180 | 4386 status = RA_NOMATCH; |
| 4387 else | |
| 4388 ADVANCE_REGINPUT(); | |
| 7 | 4389 break; |
| 4390 | |
| 4391 case SFNAME: | |
| 4392 if (VIM_ISDIGIT(*reginput) || !vim_isfilec(c)) | |
| 180 | 4393 status = RA_NOMATCH; |
| 4394 else | |
| 4395 ADVANCE_REGINPUT(); | |
| 7 | 4396 break; |
| 4397 | |
| 4398 case PRINT: | |
| 4399 if (ptr2cells(reginput) != 1) | |
| 180 | 4400 status = RA_NOMATCH; |
| 4401 else | |
| 4402 ADVANCE_REGINPUT(); | |
| 7 | 4403 break; |
| 4404 | |
| 4405 case SPRINT: | |
| 4406 if (VIM_ISDIGIT(*reginput) || ptr2cells(reginput) != 1) | |
| 180 | 4407 status = RA_NOMATCH; |
| 4408 else | |
| 4409 ADVANCE_REGINPUT(); | |
| 7 | 4410 break; |
| 4411 | |
| 4412 case WHITE: | |
| 4413 if (!vim_iswhite(c)) | |
| 180 | 4414 status = RA_NOMATCH; |
| 4415 else | |
| 4416 ADVANCE_REGINPUT(); | |
| 7 | 4417 break; |
| 4418 | |
| 4419 case NWHITE: | |
| 4420 if (c == NUL || vim_iswhite(c)) | |
| 180 | 4421 status = RA_NOMATCH; |
| 4422 else | |
| 4423 ADVANCE_REGINPUT(); | |
| 7 | 4424 break; |
| 4425 | |
| 4426 case DIGIT: | |
| 4427 if (!ri_digit(c)) | |
| 180 | 4428 status = RA_NOMATCH; |
| 4429 else | |
| 4430 ADVANCE_REGINPUT(); | |
| 7 | 4431 break; |
| 4432 | |
| 4433 case NDIGIT: | |
| 4434 if (c == NUL || ri_digit(c)) | |
| 180 | 4435 status = RA_NOMATCH; |
| 4436 else | |
| 4437 ADVANCE_REGINPUT(); | |
| 7 | 4438 break; |
| 4439 | |
| 4440 case HEX: | |
| 4441 if (!ri_hex(c)) | |
| 180 | 4442 status = RA_NOMATCH; |
| 4443 else | |
| 4444 ADVANCE_REGINPUT(); | |
| 7 | 4445 break; |
| 4446 | |
| 4447 case NHEX: | |
| 4448 if (c == NUL || ri_hex(c)) | |
| 180 | 4449 status = RA_NOMATCH; |
| 4450 else | |
| 4451 ADVANCE_REGINPUT(); | |
| 7 | 4452 break; |
| 4453 | |
| 4454 case OCTAL: | |
| 4455 if (!ri_octal(c)) | |
| 180 | 4456 status = RA_NOMATCH; |
| 4457 else | |
| 4458 ADVANCE_REGINPUT(); | |
| 7 | 4459 break; |
| 4460 | |
| 4461 case NOCTAL: | |
| 4462 if (c == NUL || ri_octal(c)) | |
| 180 | 4463 status = RA_NOMATCH; |
| 4464 else | |
| 4465 ADVANCE_REGINPUT(); | |
| 7 | 4466 break; |
| 4467 | |
| 4468 case WORD: | |
| 4469 if (!ri_word(c)) | |
| 180 | 4470 status = RA_NOMATCH; |
| 4471 else | |
| 4472 ADVANCE_REGINPUT(); | |
| 7 | 4473 break; |
| 4474 | |
| 4475 case NWORD: | |
| 4476 if (c == NUL || ri_word(c)) | |
| 180 | 4477 status = RA_NOMATCH; |
| 4478 else | |
| 4479 ADVANCE_REGINPUT(); | |
| 7 | 4480 break; |
| 4481 | |
| 4482 case HEAD: | |
| 4483 if (!ri_head(c)) | |
| 180 | 4484 status = RA_NOMATCH; |
| 4485 else | |
| 4486 ADVANCE_REGINPUT(); | |
| 7 | 4487 break; |
| 4488 | |
| 4489 case NHEAD: | |
| 4490 if (c == NUL || ri_head(c)) | |
| 180 | 4491 status = RA_NOMATCH; |
| 4492 else | |
| 4493 ADVANCE_REGINPUT(); | |
| 7 | 4494 break; |
| 4495 | |
| 4496 case ALPHA: | |
| 4497 if (!ri_alpha(c)) | |
| 180 | 4498 status = RA_NOMATCH; |
| 4499 else | |
| 4500 ADVANCE_REGINPUT(); | |
| 7 | 4501 break; |
| 4502 | |
| 4503 case NALPHA: | |
| 4504 if (c == NUL || ri_alpha(c)) | |
| 180 | 4505 status = RA_NOMATCH; |
| 4506 else | |
| 4507 ADVANCE_REGINPUT(); | |
| 7 | 4508 break; |
| 4509 | |
| 4510 case LOWER: | |
| 4511 if (!ri_lower(c)) | |
| 180 | 4512 status = RA_NOMATCH; |
| 4513 else | |
| 4514 ADVANCE_REGINPUT(); | |
| 7 | 4515 break; |
| 4516 | |
| 4517 case NLOWER: | |
| 4518 if (c == NUL || ri_lower(c)) | |
| 180 | 4519 status = RA_NOMATCH; |
| 4520 else | |
| 4521 ADVANCE_REGINPUT(); | |
| 7 | 4522 break; |
| 4523 | |
| 4524 case UPPER: | |
| 4525 if (!ri_upper(c)) | |
| 180 | 4526 status = RA_NOMATCH; |
| 4527 else | |
| 4528 ADVANCE_REGINPUT(); | |
| 7 | 4529 break; |
| 4530 | |
| 4531 case NUPPER: | |
| 4532 if (c == NUL || ri_upper(c)) | |
| 180 | 4533 status = RA_NOMATCH; |
| 4534 else | |
| 4535 ADVANCE_REGINPUT(); | |
| 7 | 4536 break; |
| 4537 | |
| 4538 case EXACTLY: | |
| 4539 { | |
| 4540 int len; | |
| 4541 char_u *opnd; | |
| 4542 | |
| 4543 opnd = OPERAND(scan); | |
| 4544 /* Inline the first byte, for speed. */ | |
| 4545 if (*opnd != *reginput | |
| 4546 && (!ireg_ic || ( | |
| 4547 #ifdef FEAT_MBYTE | |
| 4548 !enc_utf8 && | |
| 4549 #endif | |
| 1347 | 4550 MB_TOLOWER(*opnd) != MB_TOLOWER(*reginput)))) |
| 180 | 4551 status = RA_NOMATCH; |
| 4552 else if (*opnd == NUL) | |
| 7 | 4553 { |
| 4554 /* match empty string always works; happens when "~" is | |
| 4555 * empty. */ | |
| 4556 } | |
| 4557 else if (opnd[1] == NUL | |
| 4558 #ifdef FEAT_MBYTE | |
| 4559 && !(enc_utf8 && ireg_ic) | |
| 4560 #endif | |
| 4561 ) | |
| 4562 ++reginput; /* matched a single char */ | |
| 4563 else | |
| 4564 { | |
| 4565 len = (int)STRLEN(opnd); | |
| 4566 /* Need to match first byte again for multi-byte. */ | |
| 4567 if (cstrncmp(opnd, reginput, &len) != 0) | |
| 180 | 4568 status = RA_NOMATCH; |
| 7 | 4569 #ifdef FEAT_MBYTE |
| 4570 /* Check for following composing character. */ | |
| 180 | 4571 else if (enc_utf8 |
| 4572 && UTF_COMPOSINGLIKE(reginput, reginput + len)) | |
| 7 | 4573 { |
| 4574 /* raaron: This code makes a composing character get | |
| 4575 * ignored, which is the correct behavior (sometimes) | |
| 4576 * for voweled Hebrew texts. */ | |
| 4577 if (!ireg_icombine) | |
| 180 | 4578 status = RA_NOMATCH; |
| 7 | 4579 } |
| 180 | 4580 #endif |
| 7 | 4581 else |
| 180 | 4582 reginput += len; |
| 7 | 4583 } |
| 4584 } | |
| 4585 break; | |
| 4586 | |
| 4587 case ANYOF: | |
| 4588 case ANYBUT: | |
| 4589 if (c == NUL) | |
| 180 | 4590 status = RA_NOMATCH; |
| 4591 else if ((cstrchr(OPERAND(scan), c) == NULL) == (op == ANYOF)) | |
| 4592 status = RA_NOMATCH; | |
| 4593 else | |
| 4594 ADVANCE_REGINPUT(); | |
| 7 | 4595 break; |
| 4596 | |
| 4597 #ifdef FEAT_MBYTE | |
| 4598 case MULTIBYTECODE: | |
| 4599 if (has_mbyte) | |
| 4600 { | |
| 4601 int i, len; | |
| 4602 char_u *opnd; | |
| 944 | 4603 int opndc = 0, inpc; |
| 7 | 4604 |
| 4605 opnd = OPERAND(scan); | |
| 4606 /* Safety check (just in case 'encoding' was changed since | |
| 4607 * compiling the program). */ | |
| 474 | 4608 if ((len = (*mb_ptr2len)(opnd)) < 2) |
| 180 | 4609 { |
| 4610 status = RA_NOMATCH; | |
| 4611 break; | |
| 4612 } | |
| 714 | 4613 if (enc_utf8) |
| 4614 opndc = mb_ptr2char(opnd); | |
| 4615 if (enc_utf8 && utf_iscomposing(opndc)) | |
| 4616 { | |
| 4617 /* When only a composing char is given match at any | |
| 4618 * position where that composing char appears. */ | |
| 4619 status = RA_NOMATCH; | |
| 4620 for (i = 0; reginput[i] != NUL; i += utf_char2len(inpc)) | |
| 180 | 4621 { |
| 714 | 4622 inpc = mb_ptr2char(reginput + i); |
| 4623 if (!utf_iscomposing(inpc)) | |
| 4624 { | |
| 4625 if (i > 0) | |
| 4626 break; | |
| 4627 } | |
| 4628 else if (opndc == inpc) | |
| 4629 { | |
| 4630 /* Include all following composing chars. */ | |
| 4631 len = i + mb_ptr2len(reginput + i); | |
| 4632 status = RA_MATCH; | |
| 4633 break; | |
| 4634 } | |
| 180 | 4635 } |
| 714 | 4636 } |
| 4637 else | |
| 4638 for (i = 0; i < len; ++i) | |
| 4639 if (opnd[i] != reginput[i]) | |
| 4640 { | |
| 4641 status = RA_NOMATCH; | |
| 4642 break; | |
| 4643 } | |
| 7 | 4644 reginput += len; |
| 4645 } | |
| 4646 else | |
| 180 | 4647 status = RA_NOMATCH; |
| 7 | 4648 break; |
| 4649 #endif | |
| 4650 | |
| 4651 case NOTHING: | |
| 4652 break; | |
| 4653 | |
| 4654 case BACK: | |
| 233 | 4655 { |
| 4656 int i; | |
| 4657 backpos_T *bp; | |
| 4658 | |
| 4659 /* | |
| 4660 * When we run into BACK we need to check if we don't keep | |
| 4661 * looping without matching any input. The second and later | |
| 4662 * times a BACK is encountered it fails if the input is still | |
| 4663 * at the same position as the previous time. | |
| 4664 * The positions are stored in "backpos" and found by the | |
| 4665 * current value of "scan", the position in the RE program. | |
| 4666 */ | |
| 4667 bp = (backpos_T *)backpos.ga_data; | |
| 4668 for (i = 0; i < backpos.ga_len; ++i) | |
| 4669 if (bp[i].bp_scan == scan) | |
| 4670 break; | |
| 4671 if (i == backpos.ga_len) | |
| 4672 { | |
| 4673 /* First time at this BACK, make room to store the pos. */ | |
| 4674 if (ga_grow(&backpos, 1) == FAIL) | |
| 4675 status = RA_FAIL; | |
| 4676 else | |
| 4677 { | |
| 4678 /* get "ga_data" again, it may have changed */ | |
| 4679 bp = (backpos_T *)backpos.ga_data; | |
| 4680 bp[i].bp_scan = scan; | |
| 4681 ++backpos.ga_len; | |
| 4682 } | |
| 4683 } | |
| 4684 else if (reg_save_equal(&bp[i].bp_pos)) | |
| 4685 /* Still at same position as last time, fail. */ | |
| 4686 status = RA_NOMATCH; | |
| 4687 | |
| 4688 if (status != RA_FAIL && status != RA_NOMATCH) | |
| 4689 reg_save(&bp[i].bp_pos, &backpos); | |
| 4690 } | |
| 179 | 4691 break; |
| 4692 | |
| 7 | 4693 case MOPEN + 0: /* Match start: \zs */ |
| 4694 case MOPEN + 1: /* \( */ | |
| 4695 case MOPEN + 2: | |
| 4696 case MOPEN + 3: | |
| 4697 case MOPEN + 4: | |
| 4698 case MOPEN + 5: | |
| 4699 case MOPEN + 6: | |
| 4700 case MOPEN + 7: | |
| 4701 case MOPEN + 8: | |
| 4702 case MOPEN + 9: | |
| 4703 { | |
| 4704 no = op - MOPEN; | |
| 4705 cleanup_subexpr(); | |
| 270 | 4706 rp = regstack_push(RS_MOPEN, scan); |
| 180 | 4707 if (rp == NULL) |
| 4708 status = RA_FAIL; | |
| 4709 else | |
| 4710 { | |
| 4711 rp->rs_no = no; | |
| 4712 save_se(&rp->rs_un.sesave, ®_startpos[no], | |
| 4713 ®_startp[no]); | |
| 4714 /* We simply continue and handle the result when done. */ | |
| 4715 } | |
| 7 | 4716 } |
| 180 | 4717 break; |
| 7 | 4718 |
| 4719 case NOPEN: /* \%( */ | |
| 4720 case NCLOSE: /* \) after \%( */ | |
| 270 | 4721 if (regstack_push(RS_NOPEN, scan) == NULL) |
| 180 | 4722 status = RA_FAIL; |
| 4723 /* We simply continue and handle the result when done. */ | |
| 4724 break; | |
| 7 | 4725 |
| 4726 #ifdef FEAT_SYN_HL | |
| 4727 case ZOPEN + 1: | |
| 4728 case ZOPEN + 2: | |
| 4729 case ZOPEN + 3: | |
| 4730 case ZOPEN + 4: | |
| 4731 case ZOPEN + 5: | |
| 4732 case ZOPEN + 6: | |
| 4733 case ZOPEN + 7: | |
| 4734 case ZOPEN + 8: | |
| 4735 case ZOPEN + 9: | |
| 4736 { | |
| 4737 no = op - ZOPEN; | |
| 4738 cleanup_zsubexpr(); | |
| 270 | 4739 rp = regstack_push(RS_ZOPEN, scan); |
| 180 | 4740 if (rp == NULL) |
| 4741 status = RA_FAIL; | |
| 4742 else | |
| 4743 { | |
| 4744 rp->rs_no = no; | |
| 4745 save_se(&rp->rs_un.sesave, ®_startzpos[no], | |
| 4746 ®_startzp[no]); | |
| 4747 /* We simply continue and handle the result when done. */ | |
| 4748 } | |
| 7 | 4749 } |
| 180 | 4750 break; |
| 7 | 4751 #endif |
| 4752 | |
| 4753 case MCLOSE + 0: /* Match end: \ze */ | |
| 4754 case MCLOSE + 1: /* \) */ | |
| 4755 case MCLOSE + 2: | |
| 4756 case MCLOSE + 3: | |
| 4757 case MCLOSE + 4: | |
| 4758 case MCLOSE + 5: | |
| 4759 case MCLOSE + 6: | |
| 4760 case MCLOSE + 7: | |
| 4761 case MCLOSE + 8: | |
| 4762 case MCLOSE + 9: | |
| 4763 { | |
| 4764 no = op - MCLOSE; | |
| 4765 cleanup_subexpr(); | |
| 270 | 4766 rp = regstack_push(RS_MCLOSE, scan); |
| 180 | 4767 if (rp == NULL) |
| 4768 status = RA_FAIL; | |
| 4769 else | |
| 4770 { | |
| 4771 rp->rs_no = no; | |
| 4772 save_se(&rp->rs_un.sesave, ®_endpos[no], ®_endp[no]); | |
| 4773 /* We simply continue and handle the result when done. */ | |
| 4774 } | |
| 7 | 4775 } |
| 180 | 4776 break; |
| 7 | 4777 |
| 4778 #ifdef FEAT_SYN_HL | |
| 4779 case ZCLOSE + 1: /* \) after \z( */ | |
| 4780 case ZCLOSE + 2: | |
| 4781 case ZCLOSE + 3: | |
| 4782 case ZCLOSE + 4: | |
| 4783 case ZCLOSE + 5: | |
| 4784 case ZCLOSE + 6: | |
| 4785 case ZCLOSE + 7: | |
| 4786 case ZCLOSE + 8: | |
| 4787 case ZCLOSE + 9: | |
| 4788 { | |
| 4789 no = op - ZCLOSE; | |
| 4790 cleanup_zsubexpr(); | |
| 270 | 4791 rp = regstack_push(RS_ZCLOSE, scan); |
| 180 | 4792 if (rp == NULL) |
| 4793 status = RA_FAIL; | |
| 4794 else | |
| 4795 { | |
| 4796 rp->rs_no = no; | |
| 4797 save_se(&rp->rs_un.sesave, ®_endzpos[no], | |
| 4798 ®_endzp[no]); | |
| 4799 /* We simply continue and handle the result when done. */ | |
| 4800 } | |
| 7 | 4801 } |
| 180 | 4802 break; |
| 7 | 4803 #endif |
| 4804 | |
| 4805 case BACKREF + 1: | |
| 4806 case BACKREF + 2: | |
| 4807 case BACKREF + 3: | |
| 4808 case BACKREF + 4: | |
| 4809 case BACKREF + 5: | |
| 4810 case BACKREF + 6: | |
| 4811 case BACKREF + 7: | |
| 4812 case BACKREF + 8: | |
| 4813 case BACKREF + 9: | |
| 4814 { | |
| 4815 int len; | |
| 4816 linenr_T clnum; | |
| 4817 colnr_T ccol; | |
| 4818 char_u *p; | |
| 4819 | |
| 4820 no = op - BACKREF; | |
| 4821 cleanup_subexpr(); | |
| 4822 if (!REG_MULTI) /* Single-line regexp */ | |
| 4823 { | |
| 1815 | 4824 if (reg_startp[no] == NULL || reg_endp[no] == NULL) |
| 7 | 4825 { |
| 4826 /* Backref was not set: Match an empty string. */ | |
| 4827 len = 0; | |
| 4828 } | |
| 4829 else | |
| 4830 { | |
| 4831 /* Compare current input with back-ref in the same | |
| 4832 * line. */ | |
| 4833 len = (int)(reg_endp[no] - reg_startp[no]); | |
| 4834 if (cstrncmp(reg_startp[no], reginput, &len) != 0) | |
| 180 | 4835 status = RA_NOMATCH; |
| 7 | 4836 } |
| 4837 } | |
| 4838 else /* Multi-line regexp */ | |
| 4839 { | |
| 1815 | 4840 if (reg_startpos[no].lnum < 0 || reg_endpos[no].lnum < 0) |
| 7 | 4841 { |
| 4842 /* Backref was not set: Match an empty string. */ | |
| 4843 len = 0; | |
| 4844 } | |
| 4845 else | |
| 4846 { | |
| 4847 if (reg_startpos[no].lnum == reglnum | |
| 4848 && reg_endpos[no].lnum == reglnum) | |
| 4849 { | |
| 4850 /* Compare back-ref within the current line. */ | |
| 4851 len = reg_endpos[no].col - reg_startpos[no].col; | |
| 4852 if (cstrncmp(regline + reg_startpos[no].col, | |
| 4853 reginput, &len) != 0) | |
| 180 | 4854 status = RA_NOMATCH; |
| 7 | 4855 } |
| 4856 else | |
| 4857 { | |
| 4858 /* Messy situation: Need to compare between two | |
| 4859 * lines. */ | |
| 4860 ccol = reg_startpos[no].col; | |
| 4861 clnum = reg_startpos[no].lnum; | |
| 4862 for (;;) | |
| 4863 { | |
| 4864 /* Since getting one line may invalidate | |
| 4865 * the other, need to make copy. Slow! */ | |
| 4866 if (regline != reg_tofree) | |
| 4867 { | |
| 4868 len = (int)STRLEN(regline); | |
| 4869 if (reg_tofree == NULL | |
| 4870 || len >= (int)reg_tofreelen) | |
| 4871 { | |
| 4872 len += 50; /* get some extra */ | |
| 4873 vim_free(reg_tofree); | |
| 4874 reg_tofree = alloc(len); | |
| 4875 if (reg_tofree == NULL) | |
| 180 | 4876 { |
| 4877 status = RA_FAIL; /* outof memory!*/ | |
| 4878 break; | |
| 4879 } | |
| 7 | 4880 reg_tofreelen = len; |
| 4881 } | |
| 4882 STRCPY(reg_tofree, regline); | |
| 4883 reginput = reg_tofree | |
| 4884 + (reginput - regline); | |
| 4885 regline = reg_tofree; | |
| 4886 } | |
| 4887 | |
| 4888 /* Get the line to compare with. */ | |
| 4889 p = reg_getline(clnum); | |
| 4890 if (clnum == reg_endpos[no].lnum) | |
| 4891 len = reg_endpos[no].col - ccol; | |
| 4892 else | |
| 4893 len = (int)STRLEN(p + ccol); | |
| 4894 | |
| 4895 if (cstrncmp(p + ccol, reginput, &len) != 0) | |
| 180 | 4896 { |
| 4897 status = RA_NOMATCH; /* doesn't match */ | |
| 4898 break; | |
| 4899 } | |
| 7 | 4900 if (clnum == reg_endpos[no].lnum) |
| 4901 break; /* match and at end! */ | |
| 481 | 4902 if (reglnum >= reg_maxline) |
| 180 | 4903 { |
| 4904 status = RA_NOMATCH; /* text too short */ | |
| 4905 break; | |
| 4906 } | |
| 7 | 4907 |
| 4908 /* Advance to next line. */ | |
| 4909 reg_nextline(); | |
| 4910 ++clnum; | |
| 4911 ccol = 0; | |
| 180 | 4912 if (got_int) |
| 4913 { | |
| 4914 status = RA_FAIL; | |
| 4915 break; | |
| 4916 } | |
| 7 | 4917 } |
| 4918 | |
| 4919 /* found a match! Note that regline may now point | |
| 4920 * to a copy of the line, that should not matter. */ | |
| 4921 } | |
| 4922 } | |
| 4923 } | |
| 4924 | |
| 4925 /* Matched the backref, skip over it. */ | |
| 4926 reginput += len; | |
| 4927 } | |
| 4928 break; | |
| 4929 | |
| 4930 #ifdef FEAT_SYN_HL | |
| 4931 case ZREF + 1: | |
| 4932 case ZREF + 2: | |
| 4933 case ZREF + 3: | |
| 4934 case ZREF + 4: | |
| 4935 case ZREF + 5: | |
| 4936 case ZREF + 6: | |
| 4937 case ZREF + 7: | |
| 4938 case ZREF + 8: | |
| 4939 case ZREF + 9: | |
| 4940 { | |
| 4941 int len; | |
| 4942 | |
| 4943 cleanup_zsubexpr(); | |
| 4944 no = op - ZREF; | |
| 4945 if (re_extmatch_in != NULL | |
| 4946 && re_extmatch_in->matches[no] != NULL) | |
| 4947 { | |
| 4948 len = (int)STRLEN(re_extmatch_in->matches[no]); | |
| 4949 if (cstrncmp(re_extmatch_in->matches[no], | |
| 4950 reginput, &len) != 0) | |
| 180 | 4951 status = RA_NOMATCH; |
| 4952 else | |
| 4953 reginput += len; | |
| 7 | 4954 } |
| 4955 else | |
| 4956 { | |
| 4957 /* Backref was not set: Match an empty string. */ | |
| 4958 } | |
| 4959 } | |
| 4960 break; | |
| 4961 #endif | |
| 4962 | |
| 4963 case BRANCH: | |
| 4964 { | |
| 4965 if (OP(next) != BRANCH) /* No choice. */ | |
| 4966 next = OPERAND(scan); /* Avoid recursion. */ | |
| 4967 else | |
| 4968 { | |
| 270 | 4969 rp = regstack_push(RS_BRANCH, scan); |
| 180 | 4970 if (rp == NULL) |
| 4971 status = RA_FAIL; | |
| 4972 else | |
| 4973 status = RA_BREAK; /* rest is below */ | |
| 7 | 4974 } |
| 4975 } | |
| 4976 break; | |
| 4977 | |
| 4978 case BRACE_LIMITS: | |
| 4979 { | |
| 4980 if (OP(next) == BRACE_SIMPLE) | |
| 4981 { | |
| 4982 bl_minval = OPERAND_MIN(scan); | |
| 4983 bl_maxval = OPERAND_MAX(scan); | |
| 4984 } | |
| 4985 else if (OP(next) >= BRACE_COMPLEX | |
| 4986 && OP(next) < BRACE_COMPLEX + 10) | |
| 4987 { | |
| 4988 no = OP(next) - BRACE_COMPLEX; | |
| 4989 brace_min[no] = OPERAND_MIN(scan); | |
| 4990 brace_max[no] = OPERAND_MAX(scan); | |
| 4991 brace_count[no] = 0; | |
| 4992 } | |
| 4993 else | |
| 4994 { | |
| 4995 EMSG(_(e_internal)); /* Shouldn't happen */ | |
| 180 | 4996 status = RA_FAIL; |
| 7 | 4997 } |
| 4998 } | |
| 4999 break; | |
| 5000 | |
| 5001 case BRACE_COMPLEX + 0: | |
| 5002 case BRACE_COMPLEX + 1: | |
| 5003 case BRACE_COMPLEX + 2: | |
| 5004 case BRACE_COMPLEX + 3: | |
| 5005 case BRACE_COMPLEX + 4: | |
| 5006 case BRACE_COMPLEX + 5: | |
| 5007 case BRACE_COMPLEX + 6: | |
| 5008 case BRACE_COMPLEX + 7: | |
| 5009 case BRACE_COMPLEX + 8: | |
| 5010 case BRACE_COMPLEX + 9: | |
| 5011 { | |
| 5012 no = op - BRACE_COMPLEX; | |
| 5013 ++brace_count[no]; | |
| 5014 | |
| 5015 /* If not matched enough times yet, try one more */ | |
| 5016 if (brace_count[no] <= (brace_min[no] <= brace_max[no] | |
| 180 | 5017 ? brace_min[no] : brace_max[no])) |
| 7 | 5018 { |
| 270 | 5019 rp = regstack_push(RS_BRCPLX_MORE, scan); |
| 180 | 5020 if (rp == NULL) |
| 5021 status = RA_FAIL; | |
| 5022 else | |
| 5023 { | |
| 5024 rp->rs_no = no; | |
| 233 | 5025 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5026 next = OPERAND(scan); |
| 5027 /* We continue and handle the result when done. */ | |
| 5028 } | |
| 5029 break; | |
| 7 | 5030 } |
| 5031 | |
| 5032 /* If matched enough times, may try matching some more */ | |
| 5033 if (brace_min[no] <= brace_max[no]) | |
| 5034 { | |
| 5035 /* Range is the normal way around, use longest match */ | |
| 5036 if (brace_count[no] <= brace_max[no]) | |
| 5037 { | |
| 270 | 5038 rp = regstack_push(RS_BRCPLX_LONG, scan); |
| 180 | 5039 if (rp == NULL) |
| 5040 status = RA_FAIL; | |
| 5041 else | |
| 5042 { | |
| 5043 rp->rs_no = no; | |
| 233 | 5044 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5045 next = OPERAND(scan); |
| 5046 /* We continue and handle the result when done. */ | |
| 5047 } | |
| 7 | 5048 } |
| 5049 } | |
| 5050 else | |
| 5051 { | |
| 5052 /* Range is backwards, use shortest match first */ | |
| 5053 if (brace_count[no] <= brace_min[no]) | |
| 5054 { | |
| 270 | 5055 rp = regstack_push(RS_BRCPLX_SHORT, scan); |
| 180 | 5056 if (rp == NULL) |
| 5057 status = RA_FAIL; | |
| 5058 else | |
| 5059 { | |
| 233 | 5060 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5061 /* We continue and handle the result when done. */ |
| 5062 } | |
| 7 | 5063 } |
| 5064 } | |
| 5065 } | |
| 5066 break; | |
| 5067 | |
| 5068 case BRACE_SIMPLE: | |
| 5069 case STAR: | |
| 5070 case PLUS: | |
| 5071 { | |
| 180 | 5072 regstar_T rst; |
| 7 | 5073 |
| 5074 /* | |
| 5075 * Lookahead to avoid useless match attempts when we know | |
| 5076 * what character comes next. | |
| 5077 */ | |
| 5078 if (OP(next) == EXACTLY) | |
| 5079 { | |
| 180 | 5080 rst.nextb = *OPERAND(next); |
| 7 | 5081 if (ireg_ic) |
| 5082 { | |
| 1347 | 5083 if (MB_ISUPPER(rst.nextb)) |
| 5084 rst.nextb_ic = MB_TOLOWER(rst.nextb); | |
| 7 | 5085 else |
| 1347 | 5086 rst.nextb_ic = MB_TOUPPER(rst.nextb); |
| 7 | 5087 } |
| 5088 else | |
| 180 | 5089 rst.nextb_ic = rst.nextb; |
| 7 | 5090 } |
| 5091 else | |
| 5092 { | |
| 180 | 5093 rst.nextb = NUL; |
| 5094 rst.nextb_ic = NUL; | |
| 7 | 5095 } |
| 5096 if (op != BRACE_SIMPLE) | |
| 5097 { | |
| 180 | 5098 rst.minval = (op == STAR) ? 0 : 1; |
| 5099 rst.maxval = MAX_LIMIT; | |
| 7 | 5100 } |
| 5101 else | |
| 5102 { | |
| 180 | 5103 rst.minval = bl_minval; |
| 5104 rst.maxval = bl_maxval; | |
| 7 | 5105 } |
| 5106 | |
| 5107 /* | |
| 5108 * When maxval > minval, try matching as much as possible, up | |
| 5109 * to maxval. When maxval < minval, try matching at least the | |
| 5110 * minimal number (since the range is backwards, that's also | |
| 5111 * maxval!). | |
| 5112 */ | |
| 180 | 5113 rst.count = regrepeat(OPERAND(scan), rst.maxval); |
| 7 | 5114 if (got_int) |
| 180 | 5115 { |
| 5116 status = RA_FAIL; | |
| 5117 break; | |
| 5118 } | |
| 5119 if (rst.minval <= rst.maxval | |
| 5120 ? rst.count >= rst.minval : rst.count >= rst.maxval) | |
| 7 | 5121 { |
| 180 | 5122 /* It could match. Prepare for trying to match what |
| 5123 * follows. The code is below. Parameters are stored in | |
| 5124 * a regstar_T on the regstack. */ | |
| 212 | 5125 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5126 { |
| 5127 EMSG(_(e_maxmempat)); | |
| 5128 status = RA_FAIL; | |
| 5129 } | |
| 5130 else if (ga_grow(®stack, sizeof(regstar_T)) == FAIL) | |
| 180 | 5131 status = RA_FAIL; |
| 5132 else | |
| 7 | 5133 { |
| 180 | 5134 regstack.ga_len += sizeof(regstar_T); |
| 270 | 5135 rp = regstack_push(rst.minval <= rst.maxval |
| 233 | 5136 ? RS_STAR_LONG : RS_STAR_SHORT, scan); |
| 180 | 5137 if (rp == NULL) |
| 5138 status = RA_FAIL; | |
| 5139 else | |
| 7 | 5140 { |
| 180 | 5141 *(((regstar_T *)rp) - 1) = rst; |
| 5142 status = RA_BREAK; /* skip the restore bits */ | |
| 7 | 5143 } |
| 5144 } | |
| 5145 } | |
| 5146 else | |
| 180 | 5147 status = RA_NOMATCH; |
| 5148 | |
| 7 | 5149 } |
| 180 | 5150 break; |
| 7 | 5151 |
| 5152 case NOMATCH: | |
| 5153 case MATCH: | |
| 5154 case SUBPAT: | |
| 270 | 5155 rp = regstack_push(RS_NOMATCH, scan); |
| 180 | 5156 if (rp == NULL) |
| 5157 status = RA_FAIL; | |
| 5158 else | |
| 7 | 5159 { |
| 180 | 5160 rp->rs_no = op; |
| 233 | 5161 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5162 next = OPERAND(scan); |
| 5163 /* We continue and handle the result when done. */ | |
| 7 | 5164 } |
| 5165 break; | |
| 5166 | |
| 5167 case BEHIND: | |
| 5168 case NOBEHIND: | |
| 180 | 5169 /* Need a bit of room to store extra positions. */ |
| 212 | 5170 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5171 { |
| 5172 EMSG(_(e_maxmempat)); | |
| 5173 status = RA_FAIL; | |
| 5174 } | |
| 5175 else if (ga_grow(®stack, sizeof(regbehind_T)) == FAIL) | |
| 180 | 5176 status = RA_FAIL; |
| 5177 else | |
| 7 | 5178 { |
| 180 | 5179 regstack.ga_len += sizeof(regbehind_T); |
| 270 | 5180 rp = regstack_push(RS_BEHIND1, scan); |
| 180 | 5181 if (rp == NULL) |
| 5182 status = RA_FAIL; | |
| 5183 else | |
| 7 | 5184 { |
| 1579 | 5185 /* Need to save the subexpr to be able to restore them |
| 5186 * when there is a match but we don't use it. */ | |
| 5187 save_subexpr(((regbehind_T *)rp) - 1); | |
| 5188 | |
| 180 | 5189 rp->rs_no = op; |
| 233 | 5190 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5191 /* First try if what follows matches. If it does then we |
| 5192 * check the behind match by looping. */ | |
| 7 | 5193 } |
| 5194 } | |
| 180 | 5195 break; |
| 7 | 5196 |
| 5197 case BHPOS: | |
| 5198 if (REG_MULTI) | |
| 5199 { | |
| 5200 if (behind_pos.rs_u.pos.col != (colnr_T)(reginput - regline) | |
| 5201 || behind_pos.rs_u.pos.lnum != reglnum) | |
| 180 | 5202 status = RA_NOMATCH; |
| 7 | 5203 } |
| 5204 else if (behind_pos.rs_u.ptr != reginput) | |
| 180 | 5205 status = RA_NOMATCH; |
| 7 | 5206 break; |
| 5207 | |
| 5208 case NEWL: | |
| 1018 | 5209 if ((c != NUL || !REG_MULTI || reglnum > reg_maxline |
| 5210 || reg_line_lbr) && (c != '\n' || !reg_line_lbr)) | |
| 180 | 5211 status = RA_NOMATCH; |
| 5212 else if (reg_line_lbr) | |
| 7 | 5213 ADVANCE_REGINPUT(); |
| 5214 else | |
| 5215 reg_nextline(); | |
| 5216 break; | |
| 5217 | |
| 5218 case END: | |
| 180 | 5219 status = RA_MATCH; /* Success! */ |
| 5220 break; | |
| 7 | 5221 |
| 5222 default: | |
| 5223 EMSG(_(e_re_corr)); | |
| 5224 #ifdef DEBUG | |
| 5225 printf("Illegal op code %d\n", op); | |
| 5226 #endif | |
| 180 | 5227 status = RA_FAIL; |
| 5228 break; | |
| 7 | 5229 } |
| 5230 } | |
| 5231 | |
| 180 | 5232 /* If we can't continue sequentially, break the inner loop. */ |
| 5233 if (status != RA_CONT) | |
| 5234 break; | |
| 5235 | |
| 5236 /* Continue in inner loop, advance to next item. */ | |
| 7 | 5237 scan = next; |
| 180 | 5238 |
| 5239 } /* end of inner loop */ | |
| 7 | 5240 |
| 5241 /* | |
| 180 | 5242 * If there is something on the regstack execute the code for the state. |
| 233 | 5243 * If the state is popped then loop and use the older state. |
| 7 | 5244 */ |
| 180 | 5245 while (regstack.ga_len > 0 && status != RA_FAIL) |
| 5246 { | |
| 5247 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; | |
| 5248 switch (rp->rs_state) | |
| 5249 { | |
| 5250 case RS_NOPEN: | |
| 5251 /* Result is passed on as-is, simply pop the state. */ | |
| 270 | 5252 regstack_pop(&scan); |
| 180 | 5253 break; |
| 5254 | |
| 5255 case RS_MOPEN: | |
| 5256 /* Pop the state. Restore pointers when there is no match. */ | |
| 5257 if (status == RA_NOMATCH) | |
| 5258 restore_se(&rp->rs_un.sesave, ®_startpos[rp->rs_no], | |
| 5259 ®_startp[rp->rs_no]); | |
| 270 | 5260 regstack_pop(&scan); |
| 180 | 5261 break; |
| 5262 | |
| 5263 #ifdef FEAT_SYN_HL | |
| 5264 case RS_ZOPEN: | |
| 5265 /* Pop the state. Restore pointers when there is no match. */ | |
| 5266 if (status == RA_NOMATCH) | |
| 5267 restore_se(&rp->rs_un.sesave, ®_startzpos[rp->rs_no], | |
| 5268 ®_startzp[rp->rs_no]); | |
| 270 | 5269 regstack_pop(&scan); |
| 180 | 5270 break; |
| 5271 #endif | |
| 5272 | |
| 5273 case RS_MCLOSE: | |
| 5274 /* Pop the state. Restore pointers when there is no match. */ | |
| 5275 if (status == RA_NOMATCH) | |
| 5276 restore_se(&rp->rs_un.sesave, ®_endpos[rp->rs_no], | |
| 5277 ®_endp[rp->rs_no]); | |
| 270 | 5278 regstack_pop(&scan); |
| 180 | 5279 break; |
| 5280 | |
| 5281 #ifdef FEAT_SYN_HL | |
| 5282 case RS_ZCLOSE: | |
| 5283 /* Pop the state. Restore pointers when there is no match. */ | |
| 5284 if (status == RA_NOMATCH) | |
| 5285 restore_se(&rp->rs_un.sesave, ®_endzpos[rp->rs_no], | |
| 5286 ®_endzp[rp->rs_no]); | |
| 270 | 5287 regstack_pop(&scan); |
| 180 | 5288 break; |
| 5289 #endif | |
| 5290 | |
| 5291 case RS_BRANCH: | |
| 5292 if (status == RA_MATCH) | |
| 5293 /* this branch matched, use it */ | |
| 270 | 5294 regstack_pop(&scan); |
| 180 | 5295 else |
| 5296 { | |
| 5297 if (status != RA_BREAK) | |
| 5298 { | |
| 5299 /* After a non-matching branch: try next one. */ | |
| 233 | 5300 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5301 scan = rp->rs_scan; |
| 5302 } | |
| 5303 if (scan == NULL || OP(scan) != BRANCH) | |
| 5304 { | |
| 5305 /* no more branches, didn't find a match */ | |
| 5306 status = RA_NOMATCH; | |
| 270 | 5307 regstack_pop(&scan); |
| 180 | 5308 } |
| 5309 else | |
| 5310 { | |
| 5311 /* Prepare to try a branch. */ | |
| 5312 rp->rs_scan = regnext(scan); | |
| 233 | 5313 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5314 scan = OPERAND(scan); |
| 5315 } | |
| 5316 } | |
| 5317 break; | |
| 5318 | |
| 5319 case RS_BRCPLX_MORE: | |
| 5320 /* Pop the state. Restore pointers when there is no match. */ | |
| 5321 if (status == RA_NOMATCH) | |
| 5322 { | |
| 233 | 5323 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5324 --brace_count[rp->rs_no]; /* decrement match count */ |
| 5325 } | |
| 270 | 5326 regstack_pop(&scan); |
| 180 | 5327 break; |
| 5328 | |
| 5329 case RS_BRCPLX_LONG: | |
| 5330 /* Pop the state. Restore pointers when there is no match. */ | |
| 5331 if (status == RA_NOMATCH) | |
| 5332 { | |
| 5333 /* There was no match, but we did find enough matches. */ | |
| 233 | 5334 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5335 --brace_count[rp->rs_no]; |
| 5336 /* continue with the items after "\{}" */ | |
| 5337 status = RA_CONT; | |
| 5338 } | |
| 270 | 5339 regstack_pop(&scan); |
| 180 | 5340 if (status == RA_CONT) |
| 5341 scan = regnext(scan); | |
| 5342 break; | |
| 5343 | |
| 5344 case RS_BRCPLX_SHORT: | |
| 5345 /* Pop the state. Restore pointers when there is no match. */ | |
| 5346 if (status == RA_NOMATCH) | |
| 5347 /* There was no match, try to match one more item. */ | |
| 233 | 5348 reg_restore(&rp->rs_un.regsave, &backpos); |
| 270 | 5349 regstack_pop(&scan); |
| 180 | 5350 if (status == RA_NOMATCH) |
| 5351 { | |
| 5352 scan = OPERAND(scan); | |
| 5353 status = RA_CONT; | |
| 5354 } | |
| 5355 break; | |
| 5356 | |
| 5357 case RS_NOMATCH: | |
| 5358 /* Pop the state. If the operand matches for NOMATCH or | |
| 5359 * doesn't match for MATCH/SUBPAT, we fail. Otherwise backup, | |
| 5360 * except for SUBPAT, and continue with the next item. */ | |
| 5361 if (status == (rp->rs_no == NOMATCH ? RA_MATCH : RA_NOMATCH)) | |
| 5362 status = RA_NOMATCH; | |
| 5363 else | |
| 5364 { | |
| 5365 status = RA_CONT; | |
| 233 | 5366 if (rp->rs_no != SUBPAT) /* zero-width */ |
| 5367 reg_restore(&rp->rs_un.regsave, &backpos); | |
| 180 | 5368 } |
| 270 | 5369 regstack_pop(&scan); |
| 180 | 5370 if (status == RA_CONT) |
| 5371 scan = regnext(scan); | |
| 5372 break; | |
| 5373 | |
| 5374 case RS_BEHIND1: | |
| 5375 if (status == RA_NOMATCH) | |
| 5376 { | |
| 270 | 5377 regstack_pop(&scan); |
| 180 | 5378 regstack.ga_len -= sizeof(regbehind_T); |
| 5379 } | |
| 5380 else | |
| 5381 { | |
| 5382 /* The stuff after BEHIND/NOBEHIND matches. Now try if | |
| 5383 * the behind part does (not) match before the current | |
| 5384 * position in the input. This must be done at every | |
| 5385 * position in the input and checking if the match ends at | |
| 5386 * the current position. */ | |
| 5387 | |
| 5388 /* save the position after the found match for next */ | |
| 233 | 5389 reg_save(&(((regbehind_T *)rp) - 1)->save_after, &backpos); |
| 180 | 5390 |
| 5391 /* start looking for a match with operand at the current | |
| 1209 | 5392 * position. Go back one character until we find the |
| 180 | 5393 * result, hitting the start of the line or the previous |
| 5394 * line (for multi-line matching). | |
| 5395 * Set behind_pos to where the match should end, BHPOS | |
| 5396 * will match it. Save the current value. */ | |
| 5397 (((regbehind_T *)rp) - 1)->save_behind = behind_pos; | |
| 5398 behind_pos = rp->rs_un.regsave; | |
| 5399 | |
| 5400 rp->rs_state = RS_BEHIND2; | |
| 5401 | |
| 233 | 5402 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5403 scan = OPERAND(rp->rs_scan); |
| 5404 } | |
| 5405 break; | |
| 5406 | |
| 5407 case RS_BEHIND2: | |
| 5408 /* | |
| 5409 * Looping for BEHIND / NOBEHIND match. | |
| 5410 */ | |
| 5411 if (status == RA_MATCH && reg_save_equal(&behind_pos)) | |
| 5412 { | |
| 5413 /* found a match that ends where "next" started */ | |
| 5414 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5415 if (rp->rs_no == BEHIND) | |
| 233 | 5416 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5417 &backpos); | |
| 180 | 5418 else |
| 1579 | 5419 { |
| 5420 /* But we didn't want a match. Need to restore the | |
| 5421 * subexpr, because what follows matched, so they have | |
| 5422 * been set. */ | |
| 180 | 5423 status = RA_NOMATCH; |
| 1579 | 5424 restore_subexpr(((regbehind_T *)rp) - 1); |
| 5425 } | |
| 270 | 5426 regstack_pop(&scan); |
| 180 | 5427 regstack.ga_len -= sizeof(regbehind_T); |
| 5428 } | |
| 5429 else | |
| 5430 { | |
| 1579 | 5431 /* No match or a match that doesn't end where we want it: Go |
| 5432 * back one character. May go to previous line once. */ | |
| 180 | 5433 no = OK; |
| 5434 if (REG_MULTI) | |
| 5435 { | |
| 5436 if (rp->rs_un.regsave.rs_u.pos.col == 0) | |
| 5437 { | |
| 5438 if (rp->rs_un.regsave.rs_u.pos.lnum | |
| 5439 < behind_pos.rs_u.pos.lnum | |
| 5440 || reg_getline( | |
| 5441 --rp->rs_un.regsave.rs_u.pos.lnum) | |
| 5442 == NULL) | |
| 5443 no = FAIL; | |
| 5444 else | |
| 5445 { | |
| 233 | 5446 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5447 rp->rs_un.regsave.rs_u.pos.col = |
| 5448 (colnr_T)STRLEN(regline); | |
| 5449 } | |
| 5450 } | |
| 5451 else | |
| 4176 | 5452 #ifdef FEAT_MBYTE |
| 5453 if (has_mbyte) | |
| 5454 rp->rs_un.regsave.rs_u.pos.col -= | |
| 5455 (*mb_head_off)(regline, regline | |
| 5456 + rp->rs_un.regsave.rs_u.pos.col - 1) + 1; | |
| 5457 else | |
| 5458 #endif | |
| 180 | 5459 --rp->rs_un.regsave.rs_u.pos.col; |
| 5460 } | |
| 5461 else | |
| 5462 { | |
| 5463 if (rp->rs_un.regsave.rs_u.ptr == regline) | |
| 5464 no = FAIL; | |
| 5465 else | |
| 5466 --rp->rs_un.regsave.rs_u.ptr; | |
| 5467 } | |
| 5468 if (no == OK) | |
| 5469 { | |
| 5470 /* Advanced, prepare for finding match again. */ | |
| 233 | 5471 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5472 scan = OPERAND(rp->rs_scan); |
| 1579 | 5473 if (status == RA_MATCH) |
| 5474 { | |
| 5475 /* We did match, so subexpr may have been changed, | |
| 5476 * need to restore them for the next try. */ | |
| 5477 status = RA_NOMATCH; | |
| 5478 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5479 } | |
| 180 | 5480 } |
| 5481 else | |
| 5482 { | |
| 5483 /* Can't advance. For NOBEHIND that's a match. */ | |
| 5484 behind_pos = (((regbehind_T *)rp) - 1)->save_behind; | |
| 5485 if (rp->rs_no == NOBEHIND) | |
| 5486 { | |
| 233 | 5487 reg_restore(&(((regbehind_T *)rp) - 1)->save_after, |
| 5488 &backpos); | |
| 180 | 5489 status = RA_MATCH; |
| 5490 } | |
| 5491 else | |
| 1579 | 5492 { |
| 5493 /* We do want a proper match. Need to restore the | |
| 5494 * subexpr if we had a match, because they may have | |
| 5495 * been set. */ | |
| 5496 if (status == RA_MATCH) | |
| 5497 { | |
| 5498 status = RA_NOMATCH; | |
| 5499 restore_subexpr(((regbehind_T *)rp) - 1); | |
| 5500 } | |
| 5501 } | |
| 270 | 5502 regstack_pop(&scan); |
| 180 | 5503 regstack.ga_len -= sizeof(regbehind_T); |
| 5504 } | |
| 5505 } | |
| 5506 break; | |
| 5507 | |
| 5508 case RS_STAR_LONG: | |
| 5509 case RS_STAR_SHORT: | |
| 5510 { | |
| 5511 regstar_T *rst = ((regstar_T *)rp) - 1; | |
| 5512 | |
| 5513 if (status == RA_MATCH) | |
| 5514 { | |
| 270 | 5515 regstack_pop(&scan); |
| 180 | 5516 regstack.ga_len -= sizeof(regstar_T); |
| 5517 break; | |
| 5518 } | |
| 5519 | |
| 5520 /* Tried once already, restore input pointers. */ | |
| 5521 if (status != RA_BREAK) | |
| 233 | 5522 reg_restore(&rp->rs_un.regsave, &backpos); |
| 180 | 5523 |
| 5524 /* Repeat until we found a position where it could match. */ | |
| 5525 for (;;) | |
| 5526 { | |
| 5527 if (status != RA_BREAK) | |
| 5528 { | |
| 5529 /* Tried first position already, advance. */ | |
| 5530 if (rp->rs_state == RS_STAR_LONG) | |
| 5531 { | |
| 685 | 5532 /* Trying for longest match, but couldn't or |
| 5533 * didn't match -- back up one char. */ | |
| 180 | 5534 if (--rst->count < rst->minval) |
| 5535 break; | |
| 5536 if (reginput == regline) | |
| 5537 { | |
| 5538 /* backup to last char of previous line */ | |
| 5539 --reglnum; | |
| 5540 regline = reg_getline(reglnum); | |
| 5541 /* Just in case regrepeat() didn't count | |
| 5542 * right. */ | |
| 5543 if (regline == NULL) | |
| 5544 break; | |
| 5545 reginput = regline + STRLEN(regline); | |
| 5546 fast_breakcheck(); | |
| 5547 } | |
| 5548 else | |
| 5549 mb_ptr_back(regline, reginput); | |
| 5550 } | |
| 5551 else | |
| 5552 { | |
| 5553 /* Range is backwards, use shortest match first. | |
| 5554 * Careful: maxval and minval are exchanged! | |
| 5555 * Couldn't or didn't match: try advancing one | |
| 5556 * char. */ | |
| 5557 if (rst->count == rst->minval | |
| 5558 || regrepeat(OPERAND(rp->rs_scan), 1L) == 0) | |
| 5559 break; | |
| 5560 ++rst->count; | |
| 5561 } | |
| 5562 if (got_int) | |
| 5563 break; | |
| 5564 } | |
| 5565 else | |
| 5566 status = RA_NOMATCH; | |
| 5567 | |
| 5568 /* If it could match, try it. */ | |
| 5569 if (rst->nextb == NUL || *reginput == rst->nextb | |
| 5570 || *reginput == rst->nextb_ic) | |
| 5571 { | |
| 233 | 5572 reg_save(&rp->rs_un.regsave, &backpos); |
| 180 | 5573 scan = regnext(rp->rs_scan); |
| 5574 status = RA_CONT; | |
| 5575 break; | |
| 5576 } | |
| 5577 } | |
| 5578 if (status != RA_CONT) | |
| 5579 { | |
| 5580 /* Failed. */ | |
| 270 | 5581 regstack_pop(&scan); |
| 180 | 5582 regstack.ga_len -= sizeof(regstar_T); |
| 5583 status = RA_NOMATCH; | |
| 5584 } | |
| 5585 } | |
| 5586 break; | |
| 5587 } | |
| 5588 | |
| 685 | 5589 /* If we want to continue the inner loop or didn't pop a state |
| 5590 * continue matching loop */ | |
| 180 | 5591 if (status == RA_CONT || rp == (regitem_T *) |
| 5592 ((char *)regstack.ga_data + regstack.ga_len) - 1) | |
| 5593 break; | |
| 5594 } | |
| 5595 | |
| 189 | 5596 /* May need to continue with the inner loop, starting at "scan". */ |
| 180 | 5597 if (status == RA_CONT) |
| 5598 continue; | |
| 5599 | |
| 5600 /* | |
| 5601 * If the regstack is empty or something failed we are done. | |
| 5602 */ | |
| 5603 if (regstack.ga_len == 0 || status == RA_FAIL) | |
| 5604 { | |
| 5605 if (scan == NULL) | |
| 5606 { | |
| 5607 /* | |
| 5608 * We get here only if there's trouble -- normally "case END" is | |
| 5609 * the terminating point. | |
| 5610 */ | |
| 5611 EMSG(_(e_re_corr)); | |
| 7 | 5612 #ifdef DEBUG |
| 180 | 5613 printf("Premature EOL\n"); |
| 7 | 5614 #endif |
| 180 | 5615 } |
| 189 | 5616 if (status == RA_FAIL) |
| 5617 got_int = TRUE; | |
| 180 | 5618 return (status == RA_MATCH); |
| 5619 } | |
| 5620 | |
| 5621 } /* End of loop until the regstack is empty. */ | |
| 5622 | |
| 5623 /* NOTREACHED */ | |
| 5624 } | |
| 5625 | |
| 5626 /* | |
| 5627 * Push an item onto the regstack. | |
| 5628 * Returns pointer to new item. Returns NULL when out of memory. | |
| 5629 */ | |
| 5630 static regitem_T * | |
| 270 | 5631 regstack_push(state, scan) |
| 180 | 5632 regstate_T state; |
| 5633 char_u *scan; | |
| 5634 { | |
| 5635 regitem_T *rp; | |
| 5636 | |
| 270 | 5637 if ((long)((unsigned)regstack.ga_len >> 10) >= p_mmp) |
| 189 | 5638 { |
| 5639 EMSG(_(e_maxmempat)); | |
| 5640 return NULL; | |
| 5641 } | |
| 270 | 5642 if (ga_grow(®stack, sizeof(regitem_T)) == FAIL) |
| 180 | 5643 return NULL; |
| 5644 | |
| 270 | 5645 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len); |
| 180 | 5646 rp->rs_state = state; |
| 5647 rp->rs_scan = scan; | |
| 5648 | |
| 270 | 5649 regstack.ga_len += sizeof(regitem_T); |
| 180 | 5650 return rp; |
| 5651 } | |
| 5652 | |
| 5653 /* | |
| 5654 * Pop an item from the regstack. | |
| 5655 */ | |
| 5656 static void | |
| 270 | 5657 regstack_pop(scan) |
| 180 | 5658 char_u **scan; |
| 5659 { | |
| 5660 regitem_T *rp; | |
| 5661 | |
| 270 | 5662 rp = (regitem_T *)((char *)regstack.ga_data + regstack.ga_len) - 1; |
| 180 | 5663 *scan = rp->rs_scan; |
| 5664 | |
| 270 | 5665 regstack.ga_len -= sizeof(regitem_T); |
| 7 | 5666 } |
| 5667 | |
| 5668 /* | |
| 5669 * regrepeat - repeatedly match something simple, return how many. | |
| 5670 * Advances reginput (and reglnum) to just after the matched chars. | |
| 5671 */ | |
| 5672 static int | |
| 5673 regrepeat(p, maxcount) | |
| 5674 char_u *p; | |
| 5675 long maxcount; /* maximum number of matches allowed */ | |
| 5676 { | |
| 5677 long count = 0; | |
| 5678 char_u *scan; | |
| 5679 char_u *opnd; | |
| 5680 int mask; | |
| 5681 int testval = 0; | |
| 5682 | |
| 5683 scan = reginput; /* Make local copy of reginput for speed. */ | |
| 5684 opnd = OPERAND(p); | |
| 5685 switch (OP(p)) | |
| 5686 { | |
| 5687 case ANY: | |
| 5688 case ANY + ADD_NL: | |
| 5689 while (count < maxcount) | |
| 5690 { | |
| 5691 /* Matching anything means we continue until end-of-line (or | |
| 5692 * end-of-file for ANY + ADD_NL), only limited by maxcount. */ | |
| 5693 while (*scan != NUL && count < maxcount) | |
| 5694 { | |
| 5695 ++count; | |
| 39 | 5696 mb_ptr_adv(scan); |
| 7 | 5697 } |
| 1018 | 5698 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5699 || reg_line_lbr || count == maxcount) | |
| 7 | 5700 break; |
| 5701 ++count; /* count the line-break */ | |
| 5702 reg_nextline(); | |
| 5703 scan = reginput; | |
| 5704 if (got_int) | |
| 5705 break; | |
| 5706 } | |
| 5707 break; | |
| 5708 | |
| 5709 case IDENT: | |
| 5710 case IDENT + ADD_NL: | |
| 5711 testval = TRUE; | |
| 5712 /*FALLTHROUGH*/ | |
| 5713 case SIDENT: | |
| 5714 case SIDENT + ADD_NL: | |
| 5715 while (count < maxcount) | |
| 5716 { | |
| 5717 if (vim_isIDc(*scan) && (testval || !VIM_ISDIGIT(*scan))) | |
| 5718 { | |
| 39 | 5719 mb_ptr_adv(scan); |
| 7 | 5720 } |
| 5721 else if (*scan == NUL) | |
| 5722 { | |
| 1018 | 5723 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5724 || reg_line_lbr) | |
| 7 | 5725 break; |
| 5726 reg_nextline(); | |
| 5727 scan = reginput; | |
| 5728 if (got_int) | |
| 5729 break; | |
| 5730 } | |
| 5731 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5732 ++scan; | |
| 5733 else | |
| 5734 break; | |
| 5735 ++count; | |
| 5736 } | |
| 5737 break; | |
| 5738 | |
| 5739 case KWORD: | |
| 5740 case KWORD + ADD_NL: | |
| 5741 testval = TRUE; | |
| 5742 /*FALLTHROUGH*/ | |
| 5743 case SKWORD: | |
| 5744 case SKWORD + ADD_NL: | |
| 5745 while (count < maxcount) | |
| 5746 { | |
| 4069 | 5747 if (vim_iswordp_buf(scan, reg_buf) |
| 5748 && (testval || !VIM_ISDIGIT(*scan))) | |
| 7 | 5749 { |
| 39 | 5750 mb_ptr_adv(scan); |
| 7 | 5751 } |
| 5752 else if (*scan == NUL) | |
| 5753 { | |
| 1018 | 5754 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5755 || reg_line_lbr) | |
| 7 | 5756 break; |
| 5757 reg_nextline(); | |
| 5758 scan = reginput; | |
| 5759 if (got_int) | |
| 5760 break; | |
| 5761 } | |
| 5762 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5763 ++scan; | |
| 5764 else | |
| 5765 break; | |
| 5766 ++count; | |
| 5767 } | |
| 5768 break; | |
| 5769 | |
| 5770 case FNAME: | |
| 5771 case FNAME + ADD_NL: | |
| 5772 testval = TRUE; | |
| 5773 /*FALLTHROUGH*/ | |
| 5774 case SFNAME: | |
| 5775 case SFNAME + ADD_NL: | |
| 5776 while (count < maxcount) | |
| 5777 { | |
| 5778 if (vim_isfilec(*scan) && (testval || !VIM_ISDIGIT(*scan))) | |
| 5779 { | |
| 39 | 5780 mb_ptr_adv(scan); |
| 7 | 5781 } |
| 5782 else if (*scan == NUL) | |
| 5783 { | |
| 1018 | 5784 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5785 || reg_line_lbr) | |
| 7 | 5786 break; |
| 5787 reg_nextline(); | |
| 5788 scan = reginput; | |
| 5789 if (got_int) | |
| 5790 break; | |
| 5791 } | |
| 5792 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5793 ++scan; | |
| 5794 else | |
| 5795 break; | |
| 5796 ++count; | |
| 5797 } | |
| 5798 break; | |
| 5799 | |
| 5800 case PRINT: | |
| 5801 case PRINT + ADD_NL: | |
| 5802 testval = TRUE; | |
| 5803 /*FALLTHROUGH*/ | |
| 5804 case SPRINT: | |
| 5805 case SPRINT + ADD_NL: | |
| 5806 while (count < maxcount) | |
| 5807 { | |
| 5808 if (*scan == NUL) | |
| 5809 { | |
| 1018 | 5810 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5811 || reg_line_lbr) | |
| 7 | 5812 break; |
| 5813 reg_nextline(); | |
| 5814 scan = reginput; | |
| 5815 if (got_int) | |
| 5816 break; | |
| 5817 } | |
| 5818 else if (ptr2cells(scan) == 1 && (testval || !VIM_ISDIGIT(*scan))) | |
| 5819 { | |
| 39 | 5820 mb_ptr_adv(scan); |
| 7 | 5821 } |
| 5822 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5823 ++scan; | |
| 5824 else | |
| 5825 break; | |
| 5826 ++count; | |
| 5827 } | |
| 5828 break; | |
| 5829 | |
| 5830 case WHITE: | |
| 5831 case WHITE + ADD_NL: | |
| 5832 testval = mask = RI_WHITE; | |
| 5833 do_class: | |
| 5834 while (count < maxcount) | |
| 5835 { | |
| 5836 #ifdef FEAT_MBYTE | |
| 5837 int l; | |
| 5838 #endif | |
| 5839 if (*scan == NUL) | |
| 5840 { | |
| 1018 | 5841 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 5842 || reg_line_lbr) | |
| 7 | 5843 break; |
| 5844 reg_nextline(); | |
| 5845 scan = reginput; | |
| 5846 if (got_int) | |
| 5847 break; | |
| 5848 } | |
| 5849 #ifdef FEAT_MBYTE | |
| 474 | 5850 else if (has_mbyte && (l = (*mb_ptr2len)(scan)) > 1) |
| 7 | 5851 { |
| 5852 if (testval != 0) | |
| 5853 break; | |
| 5854 scan += l; | |
| 5855 } | |
| 5856 #endif | |
| 5857 else if ((class_tab[*scan] & mask) == testval) | |
| 5858 ++scan; | |
| 5859 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 5860 ++scan; | |
| 5861 else | |
| 5862 break; | |
| 5863 ++count; | |
| 5864 } | |
| 5865 break; | |
| 5866 | |
| 5867 case NWHITE: | |
| 5868 case NWHITE + ADD_NL: | |
| 5869 mask = RI_WHITE; | |
| 5870 goto do_class; | |
| 5871 case DIGIT: | |
| 5872 case DIGIT + ADD_NL: | |
| 5873 testval = mask = RI_DIGIT; | |
| 5874 goto do_class; | |
| 5875 case NDIGIT: | |
| 5876 case NDIGIT + ADD_NL: | |
| 5877 mask = RI_DIGIT; | |
| 5878 goto do_class; | |
| 5879 case HEX: | |
| 5880 case HEX + ADD_NL: | |
| 5881 testval = mask = RI_HEX; | |
| 5882 goto do_class; | |
| 5883 case NHEX: | |
| 5884 case NHEX + ADD_NL: | |
| 5885 mask = RI_HEX; | |
| 5886 goto do_class; | |
| 5887 case OCTAL: | |
| 5888 case OCTAL + ADD_NL: | |
| 5889 testval = mask = RI_OCTAL; | |
| 5890 goto do_class; | |
| 5891 case NOCTAL: | |
| 5892 case NOCTAL + ADD_NL: | |
| 5893 mask = RI_OCTAL; | |
| 5894 goto do_class; | |
| 5895 case WORD: | |
| 5896 case WORD + ADD_NL: | |
| 5897 testval = mask = RI_WORD; | |
| 5898 goto do_class; | |
| 5899 case NWORD: | |
| 5900 case NWORD + ADD_NL: | |
| 5901 mask = RI_WORD; | |
| 5902 goto do_class; | |
| 5903 case HEAD: | |
| 5904 case HEAD + ADD_NL: | |
| 5905 testval = mask = RI_HEAD; | |
| 5906 goto do_class; | |
| 5907 case NHEAD: | |
| 5908 case NHEAD + ADD_NL: | |
| 5909 mask = RI_HEAD; | |
| 5910 goto do_class; | |
| 5911 case ALPHA: | |
| 5912 case ALPHA + ADD_NL: | |
| 5913 testval = mask = RI_ALPHA; | |
| 5914 goto do_class; | |
| 5915 case NALPHA: | |
| 5916 case NALPHA + ADD_NL: | |
| 5917 mask = RI_ALPHA; | |
| 5918 goto do_class; | |
| 5919 case LOWER: | |
| 5920 case LOWER + ADD_NL: | |
| 5921 testval = mask = RI_LOWER; | |
| 5922 goto do_class; | |
| 5923 case NLOWER: | |
| 5924 case NLOWER + ADD_NL: | |
| 5925 mask = RI_LOWER; | |
| 5926 goto do_class; | |
| 5927 case UPPER: | |
| 5928 case UPPER + ADD_NL: | |
| 5929 testval = mask = RI_UPPER; | |
| 5930 goto do_class; | |
| 5931 case NUPPER: | |
| 5932 case NUPPER + ADD_NL: | |
| 5933 mask = RI_UPPER; | |
| 5934 goto do_class; | |
| 5935 | |
| 5936 case EXACTLY: | |
| 5937 { | |
| 5938 int cu, cl; | |
| 5939 | |
| 5940 /* This doesn't do a multi-byte character, because a MULTIBYTECODE | |
| 1347 | 5941 * would have been used for it. It does handle single-byte |
| 5942 * characters, such as latin1. */ | |
| 7 | 5943 if (ireg_ic) |
| 5944 { | |
| 1347 | 5945 cu = MB_TOUPPER(*opnd); |
| 5946 cl = MB_TOLOWER(*opnd); | |
| 7 | 5947 while (count < maxcount && (*scan == cu || *scan == cl)) |
| 5948 { | |
| 5949 count++; | |
| 5950 scan++; | |
| 5951 } | |
| 5952 } | |
| 5953 else | |
| 5954 { | |
| 5955 cu = *opnd; | |
| 5956 while (count < maxcount && *scan == cu) | |
| 5957 { | |
| 5958 count++; | |
| 5959 scan++; | |
| 5960 } | |
| 5961 } | |
| 5962 break; | |
| 5963 } | |
| 5964 | |
| 5965 #ifdef FEAT_MBYTE | |
| 5966 case MULTIBYTECODE: | |
| 5967 { | |
| 5968 int i, len, cf = 0; | |
| 5969 | |
| 5970 /* Safety check (just in case 'encoding' was changed since | |
| 5971 * compiling the program). */ | |
| 474 | 5972 if ((len = (*mb_ptr2len)(opnd)) > 1) |
| 7 | 5973 { |
| 5974 if (ireg_ic && enc_utf8) | |
| 5975 cf = utf_fold(utf_ptr2char(opnd)); | |
| 5976 while (count < maxcount) | |
| 5977 { | |
| 5978 for (i = 0; i < len; ++i) | |
| 5979 if (opnd[i] != scan[i]) | |
| 5980 break; | |
| 5981 if (i < len && (!ireg_ic || !enc_utf8 | |
| 5982 || utf_fold(utf_ptr2char(scan)) != cf)) | |
| 5983 break; | |
| 5984 scan += len; | |
| 5985 ++count; | |
| 5986 } | |
| 5987 } | |
| 5988 } | |
| 5989 break; | |
| 5990 #endif | |
| 5991 | |
| 5992 case ANYOF: | |
| 5993 case ANYOF + ADD_NL: | |
| 5994 testval = TRUE; | |
| 5995 /*FALLTHROUGH*/ | |
| 5996 | |
| 5997 case ANYBUT: | |
| 5998 case ANYBUT + ADD_NL: | |
| 5999 while (count < maxcount) | |
| 6000 { | |
| 6001 #ifdef FEAT_MBYTE | |
| 6002 int len; | |
| 6003 #endif | |
| 6004 if (*scan == NUL) | |
| 6005 { | |
| 1018 | 6006 if (!REG_MULTI || !WITH_NL(OP(p)) || reglnum > reg_maxline |
| 6007 || reg_line_lbr) | |
| 7 | 6008 break; |
| 6009 reg_nextline(); | |
| 6010 scan = reginput; | |
| 6011 if (got_int) | |
| 6012 break; | |
| 6013 } | |
| 6014 else if (reg_line_lbr && *scan == '\n' && WITH_NL(OP(p))) | |
| 6015 ++scan; | |
| 6016 #ifdef FEAT_MBYTE | |
| 474 | 6017 else if (has_mbyte && (len = (*mb_ptr2len)(scan)) > 1) |
| 7 | 6018 { |
| 6019 if ((cstrchr(opnd, (*mb_ptr2char)(scan)) == NULL) == testval) | |
| 6020 break; | |
| 6021 scan += len; | |
| 6022 } | |
| 6023 #endif | |
| 6024 else | |
| 6025 { | |
| 6026 if ((cstrchr(opnd, *scan) == NULL) == testval) | |
| 6027 break; | |
| 6028 ++scan; | |
| 6029 } | |
| 6030 ++count; | |
| 6031 } | |
| 6032 break; | |
| 6033 | |
| 6034 case NEWL: | |
| 6035 while (count < maxcount | |
| 1018 | 6036 && ((*scan == NUL && reglnum <= reg_maxline && !reg_line_lbr |
| 6037 && REG_MULTI) || (*scan == '\n' && reg_line_lbr))) | |
| 7 | 6038 { |
| 6039 count++; | |
| 6040 if (reg_line_lbr) | |
| 6041 ADVANCE_REGINPUT(); | |
| 6042 else | |
| 6043 reg_nextline(); | |
| 6044 scan = reginput; | |
| 6045 if (got_int) | |
| 6046 break; | |
| 6047 } | |
| 6048 break; | |
| 6049 | |
| 6050 default: /* Oh dear. Called inappropriately. */ | |
| 6051 EMSG(_(e_re_corr)); | |
| 6052 #ifdef DEBUG | |
| 6053 printf("Called regrepeat with op code %d\n", OP(p)); | |
| 6054 #endif | |
| 6055 break; | |
| 6056 } | |
| 6057 | |
| 6058 reginput = scan; | |
| 6059 | |
| 6060 return (int)count; | |
| 6061 } | |
| 6062 | |
| 6063 /* | |
| 6064 * regnext - dig the "next" pointer out of a node | |
| 2010 | 6065 * Returns NULL when calculating size, when there is no next item and when |
| 6066 * there is an error. | |
| 7 | 6067 */ |
| 6068 static char_u * | |
| 6069 regnext(p) | |
| 6070 char_u *p; | |
| 6071 { | |
| 6072 int offset; | |
| 6073 | |
| 2010 | 6074 if (p == JUST_CALC_SIZE || reg_toolong) |
| 7 | 6075 return NULL; |
| 6076 | |
| 6077 offset = NEXT(p); | |
| 6078 if (offset == 0) | |
| 6079 return NULL; | |
| 6080 | |
| 233 | 6081 if (OP(p) == BACK) |
| 7 | 6082 return p - offset; |
| 6083 else | |
| 6084 return p + offset; | |
| 6085 } | |
| 6086 | |
| 6087 /* | |
| 6088 * Check the regexp program for its magic number. | |
| 6089 * Return TRUE if it's wrong. | |
| 6090 */ | |
| 6091 static int | |
| 6092 prog_magic_wrong() | |
| 6093 { | |
| 6094 if (UCHARAT(REG_MULTI | |
| 6095 ? reg_mmatch->regprog->program | |
| 6096 : reg_match->regprog->program) != REGMAGIC) | |
| 6097 { | |
| 6098 EMSG(_(e_re_corr)); | |
| 6099 return TRUE; | |
| 6100 } | |
| 6101 return FALSE; | |
| 6102 } | |
| 6103 | |
| 6104 /* | |
| 6105 * Cleanup the subexpressions, if this wasn't done yet. | |
| 6106 * This construction is used to clear the subexpressions only when they are | |
| 6107 * used (to increase speed). | |
| 6108 */ | |
| 6109 static void | |
| 6110 cleanup_subexpr() | |
| 6111 { | |
| 6112 if (need_clear_subexpr) | |
| 6113 { | |
| 6114 if (REG_MULTI) | |
| 6115 { | |
| 6116 /* Use 0xff to set lnum to -1 */ | |
| 6117 vim_memset(reg_startpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6118 vim_memset(reg_endpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6119 } | |
| 6120 else | |
| 6121 { | |
| 6122 vim_memset(reg_startp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6123 vim_memset(reg_endp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6124 } | |
| 6125 need_clear_subexpr = FALSE; | |
| 6126 } | |
| 6127 } | |
| 6128 | |
| 6129 #ifdef FEAT_SYN_HL | |
| 6130 static void | |
| 6131 cleanup_zsubexpr() | |
| 6132 { | |
| 6133 if (need_clear_zsubexpr) | |
| 6134 { | |
| 6135 if (REG_MULTI) | |
| 6136 { | |
| 6137 /* Use 0xff to set lnum to -1 */ | |
| 6138 vim_memset(reg_startzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6139 vim_memset(reg_endzpos, 0xff, sizeof(lpos_T) * NSUBEXP); | |
| 6140 } | |
| 6141 else | |
| 6142 { | |
| 6143 vim_memset(reg_startzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6144 vim_memset(reg_endzp, 0, sizeof(char_u *) * NSUBEXP); | |
| 6145 } | |
| 6146 need_clear_zsubexpr = FALSE; | |
| 6147 } | |
| 6148 } | |
| 6149 #endif | |
| 6150 | |
| 6151 /* | |
| 1579 | 6152 * Save the current subexpr to "bp", so that they can be restored |
| 6153 * later by restore_subexpr(). | |
| 6154 */ | |
| 6155 static void | |
| 6156 save_subexpr(bp) | |
| 6157 regbehind_T *bp; | |
| 6158 { | |
| 6159 int i; | |
| 6160 | |
| 1602 | 6161 /* When "need_clear_subexpr" is set we don't need to save the values, only |
| 6162 * remember that this flag needs to be set again when restoring. */ | |
| 6163 bp->save_need_clear_subexpr = need_clear_subexpr; | |
| 6164 if (!need_clear_subexpr) | |
| 1579 | 6165 { |
| 1602 | 6166 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6167 { |
| 1602 | 6168 if (REG_MULTI) |
| 6169 { | |
| 6170 bp->save_start[i].se_u.pos = reg_startpos[i]; | |
| 6171 bp->save_end[i].se_u.pos = reg_endpos[i]; | |
| 6172 } | |
| 6173 else | |
| 6174 { | |
| 6175 bp->save_start[i].se_u.ptr = reg_startp[i]; | |
| 6176 bp->save_end[i].se_u.ptr = reg_endp[i]; | |
| 6177 } | |
| 1579 | 6178 } |
| 6179 } | |
| 6180 } | |
| 6181 | |
| 6182 /* | |
| 6183 * Restore the subexpr from "bp". | |
| 6184 */ | |
| 6185 static void | |
| 6186 restore_subexpr(bp) | |
| 6187 regbehind_T *bp; | |
| 6188 { | |
| 6189 int i; | |
| 6190 | |
| 1602 | 6191 /* Only need to restore saved values when they are not to be cleared. */ |
| 6192 need_clear_subexpr = bp->save_need_clear_subexpr; | |
| 6193 if (!need_clear_subexpr) | |
| 1579 | 6194 { |
| 1602 | 6195 for (i = 0; i < NSUBEXP; ++i) |
| 1579 | 6196 { |
| 1602 | 6197 if (REG_MULTI) |
| 6198 { | |
| 6199 reg_startpos[i] = bp->save_start[i].se_u.pos; | |
| 6200 reg_endpos[i] = bp->save_end[i].se_u.pos; | |
| 6201 } | |
| 6202 else | |
| 6203 { | |
| 6204 reg_startp[i] = bp->save_start[i].se_u.ptr; | |
| 6205 reg_endp[i] = bp->save_end[i].se_u.ptr; | |
| 6206 } | |
| 1579 | 6207 } |
| 6208 } | |
| 6209 } | |
| 6210 | |
| 6211 /* | |
| 7 | 6212 * Advance reglnum, regline and reginput to the next line. |
| 6213 */ | |
| 6214 static void | |
| 6215 reg_nextline() | |
| 6216 { | |
| 6217 regline = reg_getline(++reglnum); | |
| 6218 reginput = regline; | |
| 6219 fast_breakcheck(); | |
| 6220 } | |
| 6221 | |
| 6222 /* | |
| 6223 * Save the input line and position in a regsave_T. | |
| 6224 */ | |
| 6225 static void | |
| 233 | 6226 reg_save(save, gap) |
| 7 | 6227 regsave_T *save; |
| 233 | 6228 garray_T *gap; |
| 7 | 6229 { |
| 6230 if (REG_MULTI) | |
| 6231 { | |
| 6232 save->rs_u.pos.col = (colnr_T)(reginput - regline); | |
| 6233 save->rs_u.pos.lnum = reglnum; | |
| 6234 } | |
| 6235 else | |
| 6236 save->rs_u.ptr = reginput; | |
| 233 | 6237 save->rs_len = gap->ga_len; |
| 7 | 6238 } |
| 6239 | |
| 6240 /* | |
| 6241 * Restore the input line and position from a regsave_T. | |
| 6242 */ | |
| 6243 static void | |
| 233 | 6244 reg_restore(save, gap) |
| 7 | 6245 regsave_T *save; |
| 233 | 6246 garray_T *gap; |
| 7 | 6247 { |
| 6248 if (REG_MULTI) | |
| 6249 { | |
| 6250 if (reglnum != save->rs_u.pos.lnum) | |
| 6251 { | |
| 6252 /* only call reg_getline() when the line number changed to save | |
| 6253 * a bit of time */ | |
| 6254 reglnum = save->rs_u.pos.lnum; | |
| 6255 regline = reg_getline(reglnum); | |
| 6256 } | |
| 6257 reginput = regline + save->rs_u.pos.col; | |
| 6258 } | |
| 6259 else | |
| 6260 reginput = save->rs_u.ptr; | |
| 233 | 6261 gap->ga_len = save->rs_len; |
| 7 | 6262 } |
| 6263 | |
| 6264 /* | |
| 6265 * Return TRUE if current position is equal to saved position. | |
| 6266 */ | |
| 6267 static int | |
| 6268 reg_save_equal(save) | |
| 6269 regsave_T *save; | |
| 6270 { | |
| 6271 if (REG_MULTI) | |
| 6272 return reglnum == save->rs_u.pos.lnum | |
| 6273 && reginput == regline + save->rs_u.pos.col; | |
| 6274 return reginput == save->rs_u.ptr; | |
| 6275 } | |
| 6276 | |
| 6277 /* | |
| 6278 * Tentatively set the sub-expression start to the current position (after | |
| 6279 * calling regmatch() they will have changed). Need to save the existing | |
| 6280 * values for when there is no match. | |
| 6281 * Use se_save() to use pointer (save_se_multi()) or position (save_se_one()), | |
| 6282 * depending on REG_MULTI. | |
| 6283 */ | |
| 6284 static void | |
| 6285 save_se_multi(savep, posp) | |
| 6286 save_se_T *savep; | |
| 6287 lpos_T *posp; | |
| 6288 { | |
| 6289 savep->se_u.pos = *posp; | |
| 6290 posp->lnum = reglnum; | |
| 6291 posp->col = (colnr_T)(reginput - regline); | |
| 6292 } | |
| 6293 | |
| 6294 static void | |
| 6295 save_se_one(savep, pp) | |
| 6296 save_se_T *savep; | |
| 6297 char_u **pp; | |
| 6298 { | |
| 6299 savep->se_u.ptr = *pp; | |
| 6300 *pp = reginput; | |
| 6301 } | |
| 6302 | |
| 6303 /* | |
| 6304 * Compare a number with the operand of RE_LNUM, RE_COL or RE_VCOL. | |
| 6305 */ | |
| 6306 static int | |
| 6307 re_num_cmp(val, scan) | |
| 6308 long_u val; | |
| 6309 char_u *scan; | |
| 6310 { | |
| 6311 long_u n = OPERAND_MIN(scan); | |
| 6312 | |
| 6313 if (OPERAND_CMP(scan) == '>') | |
| 6314 return val > n; | |
| 6315 if (OPERAND_CMP(scan) == '<') | |
| 6316 return val < n; | |
| 6317 return val == n; | |
| 6318 } | |
| 6319 | |
| 6320 | |
| 6321 #ifdef DEBUG | |
| 6322 | |
| 6323 /* | |
| 6324 * regdump - dump a regexp onto stdout in vaguely comprehensible form | |
| 6325 */ | |
| 6326 static void | |
| 6327 regdump(pattern, r) | |
| 6328 char_u *pattern; | |
| 6329 regprog_T *r; | |
| 6330 { | |
| 6331 char_u *s; | |
| 6332 int op = EXACTLY; /* Arbitrary non-END op. */ | |
| 6333 char_u *next; | |
| 6334 char_u *end = NULL; | |
| 6335 | |
| 6336 printf("\r\nregcomp(%s):\r\n", pattern); | |
| 6337 | |
| 6338 s = r->program + 1; | |
| 6339 /* | |
| 6340 * Loop until we find the END that isn't before a referred next (an END | |
| 6341 * can also appear in a NOMATCH operand). | |
| 6342 */ | |
| 6343 while (op != END || s <= end) | |
| 6344 { | |
| 6345 op = OP(s); | |
| 6346 printf("%2d%s", (int)(s - r->program), regprop(s)); /* Where, what. */ | |
| 6347 next = regnext(s); | |
| 6348 if (next == NULL) /* Next ptr. */ | |
| 6349 printf("(0)"); | |
| 6350 else | |
| 6351 printf("(%d)", (int)((s - r->program) + (next - s))); | |
| 6352 if (end < next) | |
| 6353 end = next; | |
| 6354 if (op == BRACE_LIMITS) | |
| 6355 { | |
| 6356 /* Two short ints */ | |
| 6357 printf(" minval %ld, maxval %ld", OPERAND_MIN(s), OPERAND_MAX(s)); | |
| 6358 s += 8; | |
| 6359 } | |
| 6360 s += 3; | |
| 6361 if (op == ANYOF || op == ANYOF + ADD_NL | |
| 6362 || op == ANYBUT || op == ANYBUT + ADD_NL | |
| 6363 || op == EXACTLY) | |
| 6364 { | |
| 6365 /* Literal string, where present. */ | |
| 6366 while (*s != NUL) | |
| 6367 printf("%c", *s++); | |
| 6368 s++; | |
| 6369 } | |
| 6370 printf("\r\n"); | |
| 6371 } | |
| 6372 | |
| 6373 /* Header fields of interest. */ | |
| 6374 if (r->regstart != NUL) | |
| 6375 printf("start `%s' 0x%x; ", r->regstart < 256 | |
| 6376 ? (char *)transchar(r->regstart) | |
| 6377 : "multibyte", r->regstart); | |
| 6378 if (r->reganch) | |
| 6379 printf("anchored; "); | |
| 6380 if (r->regmust != NULL) | |
| 6381 printf("must have \"%s\"", r->regmust); | |
| 6382 printf("\r\n"); | |
| 6383 } | |
| 6384 | |
| 6385 /* | |
| 6386 * regprop - printable representation of opcode | |
| 6387 */ | |
| 6388 static char_u * | |
| 6389 regprop(op) | |
| 6390 char_u *op; | |
| 6391 { | |
| 6392 char_u *p; | |
| 6393 static char_u buf[50]; | |
| 6394 | |
| 6395 (void) strcpy(buf, ":"); | |
| 6396 | |
| 6397 switch (OP(op)) | |
| 6398 { | |
| 6399 case BOL: | |
| 6400 p = "BOL"; | |
| 6401 break; | |
| 6402 case EOL: | |
| 6403 p = "EOL"; | |
| 6404 break; | |
| 6405 case RE_BOF: | |
| 6406 p = "BOF"; | |
| 6407 break; | |
| 6408 case RE_EOF: | |
| 6409 p = "EOF"; | |
| 6410 break; | |
| 6411 case CURSOR: | |
| 6412 p = "CURSOR"; | |
| 6413 break; | |
| 639 | 6414 case RE_VISUAL: |
| 6415 p = "RE_VISUAL"; | |
| 6416 break; | |
| 7 | 6417 case RE_LNUM: |
| 6418 p = "RE_LNUM"; | |
| 6419 break; | |
| 639 | 6420 case RE_MARK: |
| 6421 p = "RE_MARK"; | |
| 6422 break; | |
| 7 | 6423 case RE_COL: |
| 6424 p = "RE_COL"; | |
| 6425 break; | |
| 6426 case RE_VCOL: | |
| 6427 p = "RE_VCOL"; | |
| 6428 break; | |
| 6429 case BOW: | |
| 6430 p = "BOW"; | |
| 6431 break; | |
| 6432 case EOW: | |
| 6433 p = "EOW"; | |
| 6434 break; | |
| 6435 case ANY: | |
| 6436 p = "ANY"; | |
| 6437 break; | |
| 6438 case ANY + ADD_NL: | |
| 6439 p = "ANY+NL"; | |
| 6440 break; | |
| 6441 case ANYOF: | |
| 6442 p = "ANYOF"; | |
| 6443 break; | |
| 6444 case ANYOF + ADD_NL: | |
| 6445 p = "ANYOF+NL"; | |
| 6446 break; | |
| 6447 case ANYBUT: | |
| 6448 p = "ANYBUT"; | |
| 6449 break; | |
| 6450 case ANYBUT + ADD_NL: | |
| 6451 p = "ANYBUT+NL"; | |
| 6452 break; | |
| 6453 case IDENT: | |
| 6454 p = "IDENT"; | |
| 6455 break; | |
| 6456 case IDENT + ADD_NL: | |
| 6457 p = "IDENT+NL"; | |
| 6458 break; | |
| 6459 case SIDENT: | |
| 6460 p = "SIDENT"; | |
| 6461 break; | |
| 6462 case SIDENT + ADD_NL: | |
| 6463 p = "SIDENT+NL"; | |
| 6464 break; | |
| 6465 case KWORD: | |
| 6466 p = "KWORD"; | |
| 6467 break; | |
| 6468 case KWORD + ADD_NL: | |
| 6469 p = "KWORD+NL"; | |
| 6470 break; | |
| 6471 case SKWORD: | |
| 6472 p = "SKWORD"; | |
| 6473 break; | |
| 6474 case SKWORD + ADD_NL: | |
| 6475 p = "SKWORD+NL"; | |
| 6476 break; | |
| 6477 case FNAME: | |
| 6478 p = "FNAME"; | |
| 6479 break; | |
| 6480 case FNAME + ADD_NL: | |
| 6481 p = "FNAME+NL"; | |
| 6482 break; | |
| 6483 case SFNAME: | |
| 6484 p = "SFNAME"; | |
| 6485 break; | |
| 6486 case SFNAME + ADD_NL: | |
| 6487 p = "SFNAME+NL"; | |
| 6488 break; | |
| 6489 case PRINT: | |
| 6490 p = "PRINT"; | |
| 6491 break; | |
| 6492 case PRINT + ADD_NL: | |
| 6493 p = "PRINT+NL"; | |
| 6494 break; | |
| 6495 case SPRINT: | |
| 6496 p = "SPRINT"; | |
| 6497 break; | |
| 6498 case SPRINT + ADD_NL: | |
| 6499 p = "SPRINT+NL"; | |
| 6500 break; | |
| 6501 case WHITE: | |
| 6502 p = "WHITE"; | |
| 6503 break; | |
| 6504 case WHITE + ADD_NL: | |
| 6505 p = "WHITE+NL"; | |
| 6506 break; | |
| 6507 case NWHITE: | |
| 6508 p = "NWHITE"; | |
| 6509 break; | |
| 6510 case NWHITE + ADD_NL: | |
| 6511 p = "NWHITE+NL"; | |
| 6512 break; | |
| 6513 case DIGIT: | |
| 6514 p = "DIGIT"; | |
| 6515 break; | |
| 6516 case DIGIT + ADD_NL: | |
| 6517 p = "DIGIT+NL"; | |
| 6518 break; | |
| 6519 case NDIGIT: | |
| 6520 p = "NDIGIT"; | |
| 6521 break; | |
| 6522 case NDIGIT + ADD_NL: | |
| 6523 p = "NDIGIT+NL"; | |
| 6524 break; | |
| 6525 case HEX: | |
| 6526 p = "HEX"; | |
| 6527 break; | |
| 6528 case HEX + ADD_NL: | |
| 6529 p = "HEX+NL"; | |
| 6530 break; | |
| 6531 case NHEX: | |
| 6532 p = "NHEX"; | |
| 6533 break; | |
| 6534 case NHEX + ADD_NL: | |
| 6535 p = "NHEX+NL"; | |
| 6536 break; | |
| 6537 case OCTAL: | |
| 6538 p = "OCTAL"; | |
| 6539 break; | |
| 6540 case OCTAL + ADD_NL: | |
| 6541 p = "OCTAL+NL"; | |
| 6542 break; | |
| 6543 case NOCTAL: | |
| 6544 p = "NOCTAL"; | |
| 6545 break; | |
| 6546 case NOCTAL + ADD_NL: | |
| 6547 p = "NOCTAL+NL"; | |
| 6548 break; | |
| 6549 case WORD: | |
| 6550 p = "WORD"; | |
| 6551 break; | |
| 6552 case WORD + ADD_NL: | |
| 6553 p = "WORD+NL"; | |
| 6554 break; | |
| 6555 case NWORD: | |
| 6556 p = "NWORD"; | |
| 6557 break; | |
| 6558 case NWORD + ADD_NL: | |
| 6559 p = "NWORD+NL"; | |
| 6560 break; | |
| 6561 case HEAD: | |
| 6562 p = "HEAD"; | |
| 6563 break; | |
| 6564 case HEAD + ADD_NL: | |
| 6565 p = "HEAD+NL"; | |
| 6566 break; | |
| 6567 case NHEAD: | |
| 6568 p = "NHEAD"; | |
| 6569 break; | |
| 6570 case NHEAD + ADD_NL: | |
| 6571 p = "NHEAD+NL"; | |
| 6572 break; | |
| 6573 case ALPHA: | |
| 6574 p = "ALPHA"; | |
| 6575 break; | |
| 6576 case ALPHA + ADD_NL: | |
| 6577 p = "ALPHA+NL"; | |
| 6578 break; | |
| 6579 case NALPHA: | |
| 6580 p = "NALPHA"; | |
| 6581 break; | |
| 6582 case NALPHA + ADD_NL: | |
| 6583 p = "NALPHA+NL"; | |
| 6584 break; | |
| 6585 case LOWER: | |
| 6586 p = "LOWER"; | |
| 6587 break; | |
| 6588 case LOWER + ADD_NL: | |
| 6589 p = "LOWER+NL"; | |
| 6590 break; | |
| 6591 case NLOWER: | |
| 6592 p = "NLOWER"; | |
| 6593 break; | |
| 6594 case NLOWER + ADD_NL: | |
| 6595 p = "NLOWER+NL"; | |
| 6596 break; | |
| 6597 case UPPER: | |
| 6598 p = "UPPER"; | |
| 6599 break; | |
| 6600 case UPPER + ADD_NL: | |
| 6601 p = "UPPER+NL"; | |
| 6602 break; | |
| 6603 case NUPPER: | |
| 6604 p = "NUPPER"; | |
| 6605 break; | |
| 6606 case NUPPER + ADD_NL: | |
| 6607 p = "NUPPER+NL"; | |
| 6608 break; | |
| 6609 case BRANCH: | |
| 6610 p = "BRANCH"; | |
| 6611 break; | |
| 6612 case EXACTLY: | |
| 6613 p = "EXACTLY"; | |
| 6614 break; | |
| 6615 case NOTHING: | |
| 6616 p = "NOTHING"; | |
| 6617 break; | |
| 6618 case BACK: | |
| 6619 p = "BACK"; | |
| 6620 break; | |
| 6621 case END: | |
| 6622 p = "END"; | |
| 6623 break; | |
| 6624 case MOPEN + 0: | |
| 6625 p = "MATCH START"; | |
| 6626 break; | |
| 6627 case MOPEN + 1: | |
| 6628 case MOPEN + 2: | |
| 6629 case MOPEN + 3: | |
| 6630 case MOPEN + 4: | |
| 6631 case MOPEN + 5: | |
| 6632 case MOPEN + 6: | |
| 6633 case MOPEN + 7: | |
| 6634 case MOPEN + 8: | |
| 6635 case MOPEN + 9: | |
| 6636 sprintf(buf + STRLEN(buf), "MOPEN%d", OP(op) - MOPEN); | |
| 6637 p = NULL; | |
| 6638 break; | |
| 6639 case MCLOSE + 0: | |
| 6640 p = "MATCH END"; | |
| 6641 break; | |
| 6642 case MCLOSE + 1: | |
| 6643 case MCLOSE + 2: | |
| 6644 case MCLOSE + 3: | |
| 6645 case MCLOSE + 4: | |
| 6646 case MCLOSE + 5: | |
| 6647 case MCLOSE + 6: | |
| 6648 case MCLOSE + 7: | |
| 6649 case MCLOSE + 8: | |
| 6650 case MCLOSE + 9: | |
| 6651 sprintf(buf + STRLEN(buf), "MCLOSE%d", OP(op) - MCLOSE); | |
| 6652 p = NULL; | |
| 6653 break; | |
| 6654 case BACKREF + 1: | |
| 6655 case BACKREF + 2: | |
| 6656 case BACKREF + 3: | |
| 6657 case BACKREF + 4: | |
| 6658 case BACKREF + 5: | |
| 6659 case BACKREF + 6: | |
| 6660 case BACKREF + 7: | |
| 6661 case BACKREF + 8: | |
| 6662 case BACKREF + 9: | |
| 6663 sprintf(buf + STRLEN(buf), "BACKREF%d", OP(op) - BACKREF); | |
| 6664 p = NULL; | |
| 6665 break; | |
| 6666 case NOPEN: | |
| 6667 p = "NOPEN"; | |
| 6668 break; | |
| 6669 case NCLOSE: | |
| 6670 p = "NCLOSE"; | |
| 6671 break; | |
| 6672 #ifdef FEAT_SYN_HL | |
| 6673 case ZOPEN + 1: | |
| 6674 case ZOPEN + 2: | |
| 6675 case ZOPEN + 3: | |
| 6676 case ZOPEN + 4: | |
| 6677 case ZOPEN + 5: | |
| 6678 case ZOPEN + 6: | |
| 6679 case ZOPEN + 7: | |
| 6680 case ZOPEN + 8: | |
| 6681 case ZOPEN + 9: | |
| 6682 sprintf(buf + STRLEN(buf), "ZOPEN%d", OP(op) - ZOPEN); | |
| 6683 p = NULL; | |
| 6684 break; | |
| 6685 case ZCLOSE + 1: | |
| 6686 case ZCLOSE + 2: | |
| 6687 case ZCLOSE + 3: | |
| 6688 case ZCLOSE + 4: | |
| 6689 case ZCLOSE + 5: | |
| 6690 case ZCLOSE + 6: | |
| 6691 case ZCLOSE + 7: | |
| 6692 case ZCLOSE + 8: | |
| 6693 case ZCLOSE + 9: | |
| 6694 sprintf(buf + STRLEN(buf), "ZCLOSE%d", OP(op) - ZCLOSE); | |
| 6695 p = NULL; | |
| 6696 break; | |
| 6697 case ZREF + 1: | |
| 6698 case ZREF + 2: | |
| 6699 case ZREF + 3: | |
| 6700 case ZREF + 4: | |
| 6701 case ZREF + 5: | |
| 6702 case ZREF + 6: | |
| 6703 case ZREF + 7: | |
| 6704 case ZREF + 8: | |
| 6705 case ZREF + 9: | |
| 6706 sprintf(buf + STRLEN(buf), "ZREF%d", OP(op) - ZREF); | |
| 6707 p = NULL; | |
| 6708 break; | |
| 6709 #endif | |
| 6710 case STAR: | |
| 6711 p = "STAR"; | |
| 6712 break; | |
| 6713 case PLUS: | |
| 6714 p = "PLUS"; | |
| 6715 break; | |
| 6716 case NOMATCH: | |
| 6717 p = "NOMATCH"; | |
| 6718 break; | |
| 6719 case MATCH: | |
| 6720 p = "MATCH"; | |
| 6721 break; | |
| 6722 case BEHIND: | |
| 6723 p = "BEHIND"; | |
| 6724 break; | |
| 6725 case NOBEHIND: | |
| 6726 p = "NOBEHIND"; | |
| 6727 break; | |
| 6728 case SUBPAT: | |
| 6729 p = "SUBPAT"; | |
| 6730 break; | |
| 6731 case BRACE_LIMITS: | |
| 6732 p = "BRACE_LIMITS"; | |
| 6733 break; | |
| 6734 case BRACE_SIMPLE: | |
| 6735 p = "BRACE_SIMPLE"; | |
| 6736 break; | |
| 6737 case BRACE_COMPLEX + 0: | |
| 6738 case BRACE_COMPLEX + 1: | |
| 6739 case BRACE_COMPLEX + 2: | |
| 6740 case BRACE_COMPLEX + 3: | |
| 6741 case BRACE_COMPLEX + 4: | |
| 6742 case BRACE_COMPLEX + 5: | |
| 6743 case BRACE_COMPLEX + 6: | |
| 6744 case BRACE_COMPLEX + 7: | |
| 6745 case BRACE_COMPLEX + 8: | |
| 6746 case BRACE_COMPLEX + 9: | |
| 6747 sprintf(buf + STRLEN(buf), "BRACE_COMPLEX%d", OP(op) - BRACE_COMPLEX); | |
| 6748 p = NULL; | |
| 6749 break; | |
| 6750 #ifdef FEAT_MBYTE | |
| 6751 case MULTIBYTECODE: | |
| 6752 p = "MULTIBYTECODE"; | |
| 6753 break; | |
| 6754 #endif | |
| 6755 case NEWL: | |
| 6756 p = "NEWL"; | |
| 6757 break; | |
| 6758 default: | |
| 6759 sprintf(buf + STRLEN(buf), "corrupt %d", OP(op)); | |
| 6760 p = NULL; | |
| 6761 break; | |
| 6762 } | |
| 6763 if (p != NULL) | |
| 6764 (void) strcat(buf, p); | |
| 6765 return buf; | |
| 6766 } | |
| 6767 #endif | |
| 6768 | |
| 6769 #ifdef FEAT_MBYTE | |
| 6770 static void mb_decompose __ARGS((int c, int *c1, int *c2, int *c3)); | |
| 6771 | |
| 6772 typedef struct | |
| 6773 { | |
| 6774 int a, b, c; | |
| 6775 } decomp_T; | |
| 6776 | |
| 6777 | |
| 6778 /* 0xfb20 - 0xfb4f */ | |
| 297 | 6779 static decomp_T decomp_table[0xfb4f-0xfb20+1] = |
| 7 | 6780 { |
| 6781 {0x5e2,0,0}, /* 0xfb20 alt ayin */ | |
| 6782 {0x5d0,0,0}, /* 0xfb21 alt alef */ | |
| 6783 {0x5d3,0,0}, /* 0xfb22 alt dalet */ | |
| 6784 {0x5d4,0,0}, /* 0xfb23 alt he */ | |
| 6785 {0x5db,0,0}, /* 0xfb24 alt kaf */ | |
| 6786 {0x5dc,0,0}, /* 0xfb25 alt lamed */ | |
| 6787 {0x5dd,0,0}, /* 0xfb26 alt mem-sofit */ | |
| 6788 {0x5e8,0,0}, /* 0xfb27 alt resh */ | |
| 6789 {0x5ea,0,0}, /* 0xfb28 alt tav */ | |
| 6790 {'+', 0, 0}, /* 0xfb29 alt plus */ | |
| 6791 {0x5e9, 0x5c1, 0}, /* 0xfb2a shin+shin-dot */ | |
| 6792 {0x5e9, 0x5c2, 0}, /* 0xfb2b shin+sin-dot */ | |
| 6793 {0x5e9, 0x5c1, 0x5bc}, /* 0xfb2c shin+shin-dot+dagesh */ | |
| 6794 {0x5e9, 0x5c2, 0x5bc}, /* 0xfb2d shin+sin-dot+dagesh */ | |
| 6795 {0x5d0, 0x5b7, 0}, /* 0xfb2e alef+patah */ | |
| 6796 {0x5d0, 0x5b8, 0}, /* 0xfb2f alef+qamats */ | |
| 6797 {0x5d0, 0x5b4, 0}, /* 0xfb30 alef+hiriq */ | |
| 6798 {0x5d1, 0x5bc, 0}, /* 0xfb31 bet+dagesh */ | |
| 6799 {0x5d2, 0x5bc, 0}, /* 0xfb32 gimel+dagesh */ | |
| 6800 {0x5d3, 0x5bc, 0}, /* 0xfb33 dalet+dagesh */ | |
| 6801 {0x5d4, 0x5bc, 0}, /* 0xfb34 he+dagesh */ | |
| 6802 {0x5d5, 0x5bc, 0}, /* 0xfb35 vav+dagesh */ | |
| 6803 {0x5d6, 0x5bc, 0}, /* 0xfb36 zayin+dagesh */ | |
| 6804 {0xfb37, 0, 0}, /* 0xfb37 -- UNUSED */ | |
| 6805 {0x5d8, 0x5bc, 0}, /* 0xfb38 tet+dagesh */ | |
| 6806 {0x5d9, 0x5bc, 0}, /* 0xfb39 yud+dagesh */ | |
| 6807 {0x5da, 0x5bc, 0}, /* 0xfb3a kaf sofit+dagesh */ | |
| 6808 {0x5db, 0x5bc, 0}, /* 0xfb3b kaf+dagesh */ | |
| 6809 {0x5dc, 0x5bc, 0}, /* 0xfb3c lamed+dagesh */ | |
| 6810 {0xfb3d, 0, 0}, /* 0xfb3d -- UNUSED */ | |
| 6811 {0x5de, 0x5bc, 0}, /* 0xfb3e mem+dagesh */ | |
| 6812 {0xfb3f, 0, 0}, /* 0xfb3f -- UNUSED */ | |
| 6813 {0x5e0, 0x5bc, 0}, /* 0xfb40 nun+dagesh */ | |
| 6814 {0x5e1, 0x5bc, 0}, /* 0xfb41 samech+dagesh */ | |
| 6815 {0xfb42, 0, 0}, /* 0xfb42 -- UNUSED */ | |
| 6816 {0x5e3, 0x5bc, 0}, /* 0xfb43 pe sofit+dagesh */ | |
| 6817 {0x5e4, 0x5bc,0}, /* 0xfb44 pe+dagesh */ | |
| 6818 {0xfb45, 0, 0}, /* 0xfb45 -- UNUSED */ | |
| 6819 {0x5e6, 0x5bc, 0}, /* 0xfb46 tsadi+dagesh */ | |
| 6820 {0x5e7, 0x5bc, 0}, /* 0xfb47 qof+dagesh */ | |
| 6821 {0x5e8, 0x5bc, 0}, /* 0xfb48 resh+dagesh */ | |
| 6822 {0x5e9, 0x5bc, 0}, /* 0xfb49 shin+dagesh */ | |
| 6823 {0x5ea, 0x5bc, 0}, /* 0xfb4a tav+dagesh */ | |
| 6824 {0x5d5, 0x5b9, 0}, /* 0xfb4b vav+holam */ | |
| 6825 {0x5d1, 0x5bf, 0}, /* 0xfb4c bet+rafe */ | |
| 6826 {0x5db, 0x5bf, 0}, /* 0xfb4d kaf+rafe */ | |
| 6827 {0x5e4, 0x5bf, 0}, /* 0xfb4e pe+rafe */ | |
| 6828 {0x5d0, 0x5dc, 0} /* 0xfb4f alef-lamed */ | |
| 6829 }; | |
| 6830 | |
| 6831 static void | |
| 6832 mb_decompose(c, c1, c2, c3) | |
| 6833 int c, *c1, *c2, *c3; | |
| 6834 { | |
| 6835 decomp_T d; | |
| 6836 | |
| 6837 if (c >= 0x4b20 && c <= 0xfb4f) | |
| 6838 { | |
| 6839 d = decomp_table[c - 0xfb20]; | |
| 6840 *c1 = d.a; | |
| 6841 *c2 = d.b; | |
| 6842 *c3 = d.c; | |
| 6843 } | |
| 6844 else | |
| 6845 { | |
| 6846 *c1 = c; | |
| 6847 *c2 = *c3 = 0; | |
| 6848 } | |
| 6849 } | |
| 6850 #endif | |
| 6851 | |
| 6852 /* | |
| 6853 * Compare two strings, ignore case if ireg_ic set. | |
| 6854 * Return 0 if strings match, non-zero otherwise. | |
| 6855 * Correct the length "*n" when composing characters are ignored. | |
| 6856 */ | |
| 6857 static int | |
| 6858 cstrncmp(s1, s2, n) | |
| 6859 char_u *s1, *s2; | |
| 6860 int *n; | |
| 6861 { | |
| 6862 int result; | |
| 6863 | |
| 6864 if (!ireg_ic) | |
| 6865 result = STRNCMP(s1, s2, *n); | |
| 6866 else | |
| 6867 result = MB_STRNICMP(s1, s2, *n); | |
| 6868 | |
| 6869 #ifdef FEAT_MBYTE | |
| 6870 /* if it failed and it's utf8 and we want to combineignore: */ | |
| 6871 if (result != 0 && enc_utf8 && ireg_icombine) | |
| 6872 { | |
| 6873 char_u *str1, *str2; | |
| 6874 int c1, c2, c11, c12; | |
| 6875 int junk; | |
| 6876 | |
| 6877 /* we have to handle the strcmp ourselves, since it is necessary to | |
| 6878 * deal with the composing characters by ignoring them: */ | |
| 6879 str1 = s1; | |
| 6880 str2 = s2; | |
| 6881 c1 = c2 = 0; | |
| 507 | 6882 while ((int)(str1 - s1) < *n) |
| 7 | 6883 { |
| 6884 c1 = mb_ptr2char_adv(&str1); | |
| 6885 c2 = mb_ptr2char_adv(&str2); | |
| 6886 | |
| 6887 /* decompose the character if necessary, into 'base' characters | |
| 6888 * because I don't care about Arabic, I will hard-code the Hebrew | |
| 6889 * which I *do* care about! So sue me... */ | |
| 6890 if (c1 != c2 && (!ireg_ic || utf_fold(c1) != utf_fold(c2))) | |
| 6891 { | |
| 6892 /* decomposition necessary? */ | |
| 6893 mb_decompose(c1, &c11, &junk, &junk); | |
| 6894 mb_decompose(c2, &c12, &junk, &junk); | |
| 6895 c1 = c11; | |
| 6896 c2 = c12; | |
| 6897 if (c11 != c12 && (!ireg_ic || utf_fold(c11) != utf_fold(c12))) | |
| 6898 break; | |
| 6899 } | |
| 6900 } | |
| 6901 result = c2 - c1; | |
| 6902 if (result == 0) | |
| 6903 *n = (int)(str2 - s2); | |
| 6904 } | |
| 6905 #endif | |
| 6906 | |
| 6907 return result; | |
| 6908 } | |
| 6909 | |
| 6910 /* | |
| 6911 * cstrchr: This function is used a lot for simple searches, keep it fast! | |
| 6912 */ | |
| 6913 static char_u * | |
| 6914 cstrchr(s, c) | |
| 6915 char_u *s; | |
| 6916 int c; | |
| 6917 { | |
| 6918 char_u *p; | |
| 6919 int cc; | |
| 6920 | |
| 6921 if (!ireg_ic | |
| 6922 #ifdef FEAT_MBYTE | |
| 6923 || (!enc_utf8 && mb_char2len(c) > 1) | |
| 6924 #endif | |
| 6925 ) | |
| 6926 return vim_strchr(s, c); | |
| 6927 | |
| 6928 /* tolower() and toupper() can be slow, comparing twice should be a lot | |
| 6929 * faster (esp. when using MS Visual C++!). | |
| 6930 * For UTF-8 need to use folded case. */ | |
| 6931 #ifdef FEAT_MBYTE | |
| 6932 if (enc_utf8 && c > 0x80) | |
| 6933 cc = utf_fold(c); | |
| 6934 else | |
| 6935 #endif | |
| 1347 | 6936 if (MB_ISUPPER(c)) |
| 6937 cc = MB_TOLOWER(c); | |
| 6938 else if (MB_ISLOWER(c)) | |
| 6939 cc = MB_TOUPPER(c); | |
| 7 | 6940 else |
| 6941 return vim_strchr(s, c); | |
| 6942 | |
| 6943 #ifdef FEAT_MBYTE | |
| 6944 if (has_mbyte) | |
| 6945 { | |
| 474 | 6946 for (p = s; *p != NUL; p += (*mb_ptr2len)(p)) |
| 7 | 6947 { |
| 6948 if (enc_utf8 && c > 0x80) | |
| 6949 { | |
| 6950 if (utf_fold(utf_ptr2char(p)) == cc) | |
| 6951 return p; | |
| 6952 } | |
| 6953 else if (*p == c || *p == cc) | |
| 6954 return p; | |
| 6955 } | |
| 6956 } | |
| 6957 else | |
| 6958 #endif | |
| 6959 /* Faster version for when there are no multi-byte characters. */ | |
| 6960 for (p = s; *p != NUL; ++p) | |
| 6961 if (*p == c || *p == cc) | |
| 6962 return p; | |
| 6963 | |
| 6964 return NULL; | |
| 6965 } | |
| 6966 | |
| 6967 /*************************************************************** | |
| 6968 * regsub stuff * | |
| 6969 ***************************************************************/ | |
| 6970 | |
| 6971 /* This stuff below really confuses cc on an SGI -- webb */ | |
| 6972 #ifdef __sgi | |
| 6973 # undef __ARGS | |
| 6974 # define __ARGS(x) () | |
| 6975 #endif | |
| 6976 | |
| 6977 /* | |
| 6978 * We should define ftpr as a pointer to a function returning a pointer to | |
| 6979 * a function returning a pointer to a function ... | |
| 6980 * This is impossible, so we declare a pointer to a function returning a | |
| 6981 * pointer to a function returning void. This should work for all compilers. | |
| 6982 */ | |
| 772 | 6983 typedef void (*(*fptr_T) __ARGS((int *, int)))(); |
| 6984 | |
| 6985 static fptr_T do_upper __ARGS((int *, int)); | |
| 6986 static fptr_T do_Upper __ARGS((int *, int)); | |
| 6987 static fptr_T do_lower __ARGS((int *, int)); | |
| 6988 static fptr_T do_Lower __ARGS((int *, int)); | |
| 7 | 6989 |
| 6990 static int vim_regsub_both __ARGS((char_u *source, char_u *dest, int copy, int magic, int backslash)); | |
| 6991 | |
| 772 | 6992 static fptr_T |
| 7 | 6993 do_upper(d, c) |
| 772 | 6994 int *d; |
| 6995 int c; | |
| 7 | 6996 { |
| 772 | 6997 *d = MB_TOUPPER(c); |
| 6998 | |
| 6999 return (fptr_T)NULL; | |
| 7 | 7000 } |
| 7001 | |
| 772 | 7002 static fptr_T |
| 7003 do_Upper(d, c) | |
| 7004 int *d; | |
| 7 | 7005 int c; |
| 7006 { | |
| 772 | 7007 *d = MB_TOUPPER(c); |
| 7008 | |
| 7009 return (fptr_T)do_Upper; | |
| 7010 } | |
| 7011 | |
| 7012 static fptr_T | |
| 7013 do_lower(d, c) | |
| 7014 int *d; | |
| 7015 int c; | |
| 7016 { | |
| 7017 *d = MB_TOLOWER(c); | |
| 7018 | |
| 7019 return (fptr_T)NULL; | |
| 7020 } | |
| 7021 | |
| 7022 static fptr_T | |
| 7023 do_Lower(d, c) | |
| 7024 int *d; | |
| 7025 int c; | |
| 7026 { | |
| 7027 *d = MB_TOLOWER(c); | |
| 7028 | |
| 7029 return (fptr_T)do_Lower; | |
| 7 | 7030 } |
| 7031 | |
| 7032 /* | |
| 7033 * regtilde(): Replace tildes in the pattern by the old pattern. | |
| 7034 * | |
| 7035 * Short explanation of the tilde: It stands for the previous replacement | |
| 7036 * pattern. If that previous pattern also contains a ~ we should go back a | |
| 7037 * step further... But we insert the previous pattern into the current one | |
| 7038 * and remember that. | |
| 772 | 7039 * This still does not handle the case where "magic" changes. So require the |
| 7040 * user to keep his hands off of "magic". | |
| 7 | 7041 * |
| 7042 * The tildes are parsed once before the first call to vim_regsub(). | |
| 7043 */ | |
| 7044 char_u * | |
| 7045 regtilde(source, magic) | |
| 7046 char_u *source; | |
| 7047 int magic; | |
| 7048 { | |
| 7049 char_u *newsub = source; | |
| 7050 char_u *tmpsub; | |
| 7051 char_u *p; | |
| 7052 int len; | |
| 7053 int prevlen; | |
| 7054 | |
| 7055 for (p = newsub; *p; ++p) | |
| 7056 { | |
| 7057 if ((*p == '~' && magic) || (*p == '\\' && *(p + 1) == '~' && !magic)) | |
| 7058 { | |
| 7059 if (reg_prev_sub != NULL) | |
| 7060 { | |
| 7061 /* length = len(newsub) - 1 + len(prev_sub) + 1 */ | |
| 7062 prevlen = (int)STRLEN(reg_prev_sub); | |
| 7063 tmpsub = alloc((unsigned)(STRLEN(newsub) + prevlen)); | |
| 7064 if (tmpsub != NULL) | |
| 7065 { | |
| 7066 /* copy prefix */ | |
| 7067 len = (int)(p - newsub); /* not including ~ */ | |
| 7068 mch_memmove(tmpsub, newsub, (size_t)len); | |
| 1209 | 7069 /* interpret tilde */ |
| 7 | 7070 mch_memmove(tmpsub + len, reg_prev_sub, (size_t)prevlen); |
| 7071 /* copy postfix */ | |
| 7072 if (!magic) | |
| 7073 ++p; /* back off \ */ | |
| 7074 STRCPY(tmpsub + len + prevlen, p + 1); | |
| 7075 | |
| 7076 if (newsub != source) /* already allocated newsub */ | |
| 7077 vim_free(newsub); | |
| 7078 newsub = tmpsub; | |
| 7079 p = newsub + len + prevlen; | |
| 7080 } | |
| 7081 } | |
| 7082 else if (magic) | |
| 1621 | 7083 STRMOVE(p, p + 1); /* remove '~' */ |
| 7 | 7084 else |
| 1621 | 7085 STRMOVE(p, p + 2); /* remove '\~' */ |
| 7 | 7086 --p; |
| 7087 } | |
| 7088 else | |
| 7089 { | |
| 7090 if (*p == '\\' && p[1]) /* skip escaped characters */ | |
| 7091 ++p; | |
| 7092 #ifdef FEAT_MBYTE | |
| 7093 if (has_mbyte) | |
| 474 | 7094 p += (*mb_ptr2len)(p) - 1; |
| 7 | 7095 #endif |
| 7096 } | |
| 7097 } | |
| 7098 | |
| 7099 vim_free(reg_prev_sub); | |
| 7100 if (newsub != source) /* newsub was allocated, just keep it */ | |
| 7101 reg_prev_sub = newsub; | |
| 7102 else /* no ~ found, need to save newsub */ | |
| 7103 reg_prev_sub = vim_strsave(newsub); | |
| 7104 return newsub; | |
| 7105 } | |
| 7106 | |
| 7107 #ifdef FEAT_EVAL | |
| 7108 static int can_f_submatch = FALSE; /* TRUE when submatch() can be used */ | |
| 7109 | |
| 7110 /* These pointers are used instead of reg_match and reg_mmatch for | |
| 7111 * reg_submatch(). Needed for when the substitution string is an expression | |
| 7112 * that contains a call to substitute() and submatch(). */ | |
| 7113 static regmatch_T *submatch_match; | |
| 7114 static regmmatch_T *submatch_mmatch; | |
| 2011 | 7115 static linenr_T submatch_firstlnum; |
| 7116 static linenr_T submatch_maxline; | |
| 2904 | 7117 static int submatch_line_lbr; |
| 7 | 7118 #endif |
| 7119 | |
| 7120 #if defined(FEAT_MODIFY_FNAME) || defined(FEAT_EVAL) || defined(PROTO) | |
| 7121 /* | |
| 7122 * vim_regsub() - perform substitutions after a vim_regexec() or | |
| 7123 * vim_regexec_multi() match. | |
| 7124 * | |
| 7125 * If "copy" is TRUE really copy into "dest". | |
| 7126 * If "copy" is FALSE nothing is copied, this is just to find out the length | |
| 7127 * of the result. | |
| 7128 * | |
| 7129 * If "backslash" is TRUE, a backslash will be removed later, need to double | |
| 7130 * them to keep them, and insert a backslash before a CR to avoid it being | |
| 7131 * replaced with a line break later. | |
| 7132 * | |
| 7133 * Note: The matched text must not change between the call of | |
| 7134 * vim_regexec()/vim_regexec_multi() and vim_regsub()! It would make the back | |
| 7135 * references invalid! | |
| 7136 * | |
| 7137 * Returns the size of the replacement, including terminating NUL. | |
| 7138 */ | |
| 7139 int | |
| 7140 vim_regsub(rmp, source, dest, copy, magic, backslash) | |
| 7141 regmatch_T *rmp; | |
| 7142 char_u *source; | |
| 7143 char_u *dest; | |
| 7144 int copy; | |
| 7145 int magic; | |
| 7146 int backslash; | |
| 7147 { | |
| 7148 reg_match = rmp; | |
| 7149 reg_mmatch = NULL; | |
| 7150 reg_maxline = 0; | |
| 4061 | 7151 reg_buf = curbuf; |
| 7 | 7152 return vim_regsub_both(source, dest, copy, magic, backslash); |
| 7153 } | |
| 7154 #endif | |
| 7155 | |
| 7156 int | |
| 7157 vim_regsub_multi(rmp, lnum, source, dest, copy, magic, backslash) | |
| 7158 regmmatch_T *rmp; | |
| 7159 linenr_T lnum; | |
| 7160 char_u *source; | |
| 7161 char_u *dest; | |
| 7162 int copy; | |
| 7163 int magic; | |
| 7164 int backslash; | |
| 7165 { | |
| 7166 reg_match = NULL; | |
| 7167 reg_mmatch = rmp; | |
| 7168 reg_buf = curbuf; /* always works on the current buffer! */ | |
| 7169 reg_firstlnum = lnum; | |
| 7170 reg_maxline = curbuf->b_ml.ml_line_count - lnum; | |
| 7171 return vim_regsub_both(source, dest, copy, magic, backslash); | |
| 7172 } | |
| 7173 | |
| 7174 static int | |
| 7175 vim_regsub_both(source, dest, copy, magic, backslash) | |
| 7176 char_u *source; | |
| 7177 char_u *dest; | |
| 7178 int copy; | |
| 7179 int magic; | |
| 7180 int backslash; | |
| 7181 { | |
| 7182 char_u *src; | |
| 7183 char_u *dst; | |
| 7184 char_u *s; | |
| 7185 int c; | |
| 772 | 7186 int cc; |
| 7 | 7187 int no = -1; |
| 772 | 7188 fptr_T func = (fptr_T)NULL; |
| 7 | 7189 linenr_T clnum = 0; /* init for GCC */ |
| 7190 int len = 0; /* init for GCC */ | |
| 7191 #ifdef FEAT_EVAL | |
| 7192 static char_u *eval_result = NULL; | |
| 7193 #endif | |
| 7194 | |
| 7195 /* Be paranoid... */ | |
| 7196 if (source == NULL || dest == NULL) | |
| 7197 { | |
| 7198 EMSG(_(e_null)); | |
| 7199 return 0; | |
| 7200 } | |
| 7201 if (prog_magic_wrong()) | |
| 7202 return 0; | |
| 7203 src = source; | |
| 7204 dst = dest; | |
| 7205 | |
| 7206 /* | |
| 7207 * When the substitute part starts with "\=" evaluate it as an expression. | |
| 7208 */ | |
| 7209 if (source[0] == '\\' && source[1] == '=' | |
| 7210 #ifdef FEAT_EVAL | |
| 7211 && !can_f_submatch /* can't do this recursively */ | |
| 7212 #endif | |
| 7213 ) | |
| 7214 { | |
| 7215 #ifdef FEAT_EVAL | |
| 7216 /* To make sure that the length doesn't change between checking the | |
| 7217 * length and copying the string, and to speed up things, the | |
| 7218 * resulting string is saved from the call with "copy" == FALSE to the | |
| 7219 * call with "copy" == TRUE. */ | |
| 7220 if (copy) | |
| 7221 { | |
| 7222 if (eval_result != NULL) | |
| 7223 { | |
| 7224 STRCPY(dest, eval_result); | |
| 7225 dst += STRLEN(eval_result); | |
| 7226 vim_free(eval_result); | |
| 7227 eval_result = NULL; | |
| 7228 } | |
| 7229 } | |
| 7230 else | |
| 7231 { | |
| 7232 win_T *save_reg_win; | |
| 7233 int save_ireg_ic; | |
| 7234 | |
| 7235 vim_free(eval_result); | |
| 7236 | |
| 7237 /* The expression may contain substitute(), which calls us | |
| 7238 * recursively. Make sure submatch() gets the text from the first | |
| 7239 * level. Don't need to save "reg_buf", because | |
| 7240 * vim_regexec_multi() can't be called recursively. */ | |
| 7241 submatch_match = reg_match; | |
| 7242 submatch_mmatch = reg_mmatch; | |
| 2011 | 7243 submatch_firstlnum = reg_firstlnum; |
| 7244 submatch_maxline = reg_maxline; | |
| 2904 | 7245 submatch_line_lbr = reg_line_lbr; |
| 7 | 7246 save_reg_win = reg_win; |
| 7247 save_ireg_ic = ireg_ic; | |
| 7248 can_f_submatch = TRUE; | |
| 7249 | |
| 714 | 7250 eval_result = eval_to_string(source + 2, NULL, TRUE); |
| 7 | 7251 if (eval_result != NULL) |
| 7252 { | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7253 int had_backslash = FALSE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7254 |
| 39 | 7255 for (s = eval_result; *s != NUL; mb_ptr_adv(s)) |
| 7 | 7256 { |
| 2904 | 7257 /* Change NL to CR, so that it becomes a line break, |
| 7258 * unless called from vim_regexec_nl(). | |
| 7 | 7259 * Skip over a backslashed character. */ |
| 2904 | 7260 if (*s == NL && !submatch_line_lbr) |
| 7 | 7261 *s = CAR; |
| 7262 else if (*s == '\\' && s[1] != NUL) | |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7263 { |
| 7 | 7264 ++s; |
| 2173 | 7265 /* Change NL to CR here too, so that this works: |
| 7266 * :s/abc\\\ndef/\="aaa\\\nbbb"/ on text: | |
| 7267 * abc\ | |
| 7268 * def | |
| 2904 | 7269 * Not when called from vim_regexec_nl(). |
| 2173 | 7270 */ |
| 2904 | 7271 if (*s == NL && !submatch_line_lbr) |
| 2173 | 7272 *s = CAR; |
|
2125
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7273 had_backslash = TRUE; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7274 } |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7275 } |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7276 if (had_backslash && backslash) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7277 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7278 /* Backslashes will be consumed, need to double them. */ |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7279 s = vim_strsave_escaped(eval_result, (char_u *)"\\"); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7280 if (s != NULL) |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7281 { |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7282 vim_free(eval_result); |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7283 eval_result = s; |
|
b8744d1982d1
updated for version 7.2.407
Bram Moolenaar <bram@zimbu.org>
parents:
2012
diff
changeset
|
7284 } |
| 7 | 7285 } |
| 7286 | |
| 7287 dst += STRLEN(eval_result); | |
| 7288 } | |
| 7289 | |
| 7290 reg_match = submatch_match; | |
| 7291 reg_mmatch = submatch_mmatch; | |
| 2011 | 7292 reg_firstlnum = submatch_firstlnum; |
| 7293 reg_maxline = submatch_maxline; | |
| 2904 | 7294 reg_line_lbr = submatch_line_lbr; |
| 7 | 7295 reg_win = save_reg_win; |
| 7296 ireg_ic = save_ireg_ic; | |
| 7297 can_f_submatch = FALSE; | |
| 7298 } | |
| 7299 #endif | |
| 7300 } | |
| 7301 else | |
| 7302 while ((c = *src++) != NUL) | |
| 7303 { | |
| 7304 if (c == '&' && magic) | |
| 7305 no = 0; | |
| 7306 else if (c == '\\' && *src != NUL) | |
| 7307 { | |
| 7308 if (*src == '&' && !magic) | |
| 7309 { | |
| 7310 ++src; | |
| 7311 no = 0; | |
| 7312 } | |
| 7313 else if ('0' <= *src && *src <= '9') | |
| 7314 { | |
| 7315 no = *src++ - '0'; | |
| 7316 } | |
| 7317 else if (vim_strchr((char_u *)"uUlLeE", *src)) | |
| 7318 { | |
| 7319 switch (*src++) | |
| 7320 { | |
| 772 | 7321 case 'u': func = (fptr_T)do_upper; |
| 7 | 7322 continue; |
| 772 | 7323 case 'U': func = (fptr_T)do_Upper; |
| 7 | 7324 continue; |
| 772 | 7325 case 'l': func = (fptr_T)do_lower; |
| 7 | 7326 continue; |
| 772 | 7327 case 'L': func = (fptr_T)do_Lower; |
| 7 | 7328 continue; |
| 7329 case 'e': | |
| 772 | 7330 case 'E': func = (fptr_T)NULL; |
| 7 | 7331 continue; |
| 7332 } | |
| 7333 } | |
| 7334 } | |
| 7335 if (no < 0) /* Ordinary character. */ | |
| 7336 { | |
| 798 | 7337 if (c == K_SPECIAL && src[0] != NUL && src[1] != NUL) |
| 7338 { | |
| 1209 | 7339 /* Copy a special key as-is. */ |
| 798 | 7340 if (copy) |
| 7341 { | |
| 7342 *dst++ = c; | |
| 7343 *dst++ = *src++; | |
| 7344 *dst++ = *src++; | |
| 7345 } | |
| 7346 else | |
| 7347 { | |
| 7348 dst += 3; | |
| 7349 src += 2; | |
| 7350 } | |
| 7351 continue; | |
| 7352 } | |
| 7353 | |
| 7 | 7354 if (c == '\\' && *src != NUL) |
| 7355 { | |
| 7356 /* Check for abbreviations -- webb */ | |
| 7357 switch (*src) | |
| 7358 { | |
| 7359 case 'r': c = CAR; ++src; break; | |
| 7360 case 'n': c = NL; ++src; break; | |
| 7361 case 't': c = TAB; ++src; break; | |
| 7362 /* Oh no! \e already has meaning in subst pat :-( */ | |
| 7363 /* case 'e': c = ESC; ++src; break; */ | |
| 7364 case 'b': c = Ctrl_H; ++src; break; | |
| 7365 | |
| 7366 /* If "backslash" is TRUE the backslash will be removed | |
| 7367 * later. Used to insert a literal CR. */ | |
| 7368 default: if (backslash) | |
| 7369 { | |
| 7370 if (copy) | |
| 7371 *dst = '\\'; | |
| 7372 ++dst; | |
| 7373 } | |
| 7374 c = *src++; | |
| 7375 } | |
| 7376 } | |
| 798 | 7377 #ifdef FEAT_MBYTE |
| 7378 else if (has_mbyte) | |
| 7379 c = mb_ptr2char(src - 1); | |
| 7380 #endif | |
| 7 | 7381 |
| 7382 /* Write to buffer, if copy is set. */ | |
| 772 | 7383 if (func == (fptr_T)NULL) /* just copy */ |
| 7384 cc = c; | |
| 7385 else | |
| 7386 /* Turbo C complains without the typecast */ | |
| 7387 func = (fptr_T)(func(&cc, c)); | |
| 7388 | |
| 7389 #ifdef FEAT_MBYTE | |
| 7390 if (has_mbyte) | |
| 7 | 7391 { |
|
2307
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7392 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
|
7393 |
| 7 | 7394 if (copy) |
| 772 | 7395 mb_char2bytes(cc, dst); |
| 7396 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
|
7397 if (enc_utf8) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7398 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7399 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
|
7400 |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7401 /* 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
|
7402 * 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
|
7403 if (clen < totlen) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7404 { |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7405 if (copy) |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7406 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
|
7407 (size_t)(totlen - clen)); |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7408 dst += totlen - clen; |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7409 } |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7410 } |
|
81527f127fb1
Fix: Composing characters in :s substitute text were dropped.
Bram Moolenaar <bram@vim.org>
parents:
2247
diff
changeset
|
7411 src += totlen - 1; |
| 7 | 7412 } |
| 7413 else | |
| 7414 #endif | |
| 7415 if (copy) | |
| 772 | 7416 *dst = cc; |
| 7 | 7417 dst++; |
| 7418 } | |
| 7419 else | |
| 7420 { | |
| 7421 if (REG_MULTI) | |
| 7422 { | |
| 7423 clnum = reg_mmatch->startpos[no].lnum; | |
| 7424 if (clnum < 0 || reg_mmatch->endpos[no].lnum < 0) | |
| 7425 s = NULL; | |
| 7426 else | |
| 7427 { | |
| 7428 s = reg_getline(clnum) + reg_mmatch->startpos[no].col; | |
| 7429 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7430 len = reg_mmatch->endpos[no].col | |
| 7431 - reg_mmatch->startpos[no].col; | |
| 7432 else | |
| 7433 len = (int)STRLEN(s); | |
| 7434 } | |
| 7435 } | |
| 7436 else | |
| 7437 { | |
| 7438 s = reg_match->startp[no]; | |
| 7439 if (reg_match->endp[no] == NULL) | |
| 7440 s = NULL; | |
| 7441 else | |
| 7442 len = (int)(reg_match->endp[no] - s); | |
| 7443 } | |
| 7444 if (s != NULL) | |
| 7445 { | |
| 7446 for (;;) | |
| 7447 { | |
| 7448 if (len == 0) | |
| 7449 { | |
| 7450 if (REG_MULTI) | |
| 7451 { | |
| 7452 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7453 break; | |
| 7454 if (copy) | |
| 7455 *dst = CAR; | |
| 7456 ++dst; | |
| 7457 s = reg_getline(++clnum); | |
| 7458 if (reg_mmatch->endpos[no].lnum == clnum) | |
| 7459 len = reg_mmatch->endpos[no].col; | |
| 7460 else | |
| 7461 len = (int)STRLEN(s); | |
| 7462 } | |
| 7463 else | |
| 7464 break; | |
| 7465 } | |
| 7466 else if (*s == NUL) /* we hit NUL. */ | |
| 7467 { | |
| 7468 if (copy) | |
| 7469 EMSG(_(e_re_damg)); | |
| 7470 goto exit; | |
| 7471 } | |
| 7472 else | |
| 7473 { | |
| 7474 if (backslash && (*s == CAR || *s == '\\')) | |
| 7475 { | |
| 7476 /* | |
| 7477 * Insert a backslash in front of a CR, otherwise | |
| 7478 * it will be replaced by a line break. | |
| 7479 * Number of backslashes will be halved later, | |
| 7480 * double them here. | |
| 7481 */ | |
| 7482 if (copy) | |
| 7483 { | |
| 7484 dst[0] = '\\'; | |
| 7485 dst[1] = *s; | |
| 7486 } | |
| 7487 dst += 2; | |
| 7488 } | |
| 7489 else | |
| 7490 { | |
| 772 | 7491 #ifdef FEAT_MBYTE |
| 7492 if (has_mbyte) | |
| 7493 c = mb_ptr2char(s); | |
| 7494 else | |
| 7495 #endif | |
| 7496 c = *s; | |
| 7497 | |
| 7498 if (func == (fptr_T)NULL) /* just copy */ | |
| 7499 cc = c; | |
| 7500 else | |
| 7501 /* Turbo C complains without the typecast */ | |
| 7502 func = (fptr_T)(func(&cc, c)); | |
| 7503 | |
| 7504 #ifdef FEAT_MBYTE | |
| 7505 if (has_mbyte) | |
| 7 | 7506 { |
| 1332 | 7507 int l; |
| 7508 | |
| 7509 /* Copy composing characters separately, one | |
| 7510 * at a time. */ | |
| 7511 if (enc_utf8) | |
| 7512 l = utf_ptr2len(s) - 1; | |
| 7513 else | |
| 7514 l = mb_ptr2len(s) - 1; | |
| 772 | 7515 |
| 7516 s += l; | |
| 7517 len -= l; | |
| 7518 if (copy) | |
| 7519 mb_char2bytes(cc, dst); | |
| 7520 dst += mb_char2len(cc) - 1; | |
| 7 | 7521 } |
| 772 | 7522 else |
| 7523 #endif | |
| 7524 if (copy) | |
| 7525 *dst = cc; | |
| 7526 dst++; | |
| 7 | 7527 } |
| 772 | 7528 |
| 7 | 7529 ++s; |
| 7530 --len; | |
| 7531 } | |
| 7532 } | |
| 7533 } | |
| 7534 no = -1; | |
| 7535 } | |
| 7536 } | |
| 7537 if (copy) | |
| 7538 *dst = NUL; | |
| 7539 | |
| 7540 exit: | |
| 7541 return (int)((dst - dest) + 1); | |
| 7542 } | |
| 7543 | |
| 7544 #ifdef FEAT_EVAL | |
| 2012 | 7545 static char_u *reg_getline_submatch __ARGS((linenr_T lnum)); |
| 7546 | |
| 7 | 7547 /* |
| 2011 | 7548 * Call reg_getline() with the line numbers from the submatch. If a |
| 7549 * substitute() was used the reg_maxline and other values have been | |
| 7550 * overwritten. | |
| 7551 */ | |
| 7552 static char_u * | |
| 7553 reg_getline_submatch(lnum) | |
| 7554 linenr_T lnum; | |
| 7555 { | |
| 7556 char_u *s; | |
| 7557 linenr_T save_first = reg_firstlnum; | |
| 7558 linenr_T save_max = reg_maxline; | |
| 7559 | |
| 7560 reg_firstlnum = submatch_firstlnum; | |
| 7561 reg_maxline = submatch_maxline; | |
| 7562 | |
| 7563 s = reg_getline(lnum); | |
| 7564 | |
| 7565 reg_firstlnum = save_first; | |
| 7566 reg_maxline = save_max; | |
| 7567 return s; | |
| 7568 } | |
| 7569 | |
| 7570 /* | |
| 1209 | 7571 * Used for the submatch() function: get the string from the n'th submatch in |
| 7 | 7572 * allocated memory. |
| 7573 * Returns NULL when not in a ":s" command and for a non-existing submatch. | |
| 7574 */ | |
| 7575 char_u * | |
| 7576 reg_submatch(no) | |
| 7577 int no; | |
| 7578 { | |
| 7579 char_u *retval = NULL; | |
| 7580 char_u *s; | |
| 7581 int len; | |
| 7582 int round; | |
| 7583 linenr_T lnum; | |
| 7584 | |
| 840 | 7585 if (!can_f_submatch || no < 0) |
| 7 | 7586 return NULL; |
| 7587 | |
| 7588 if (submatch_match == NULL) | |
| 7589 { | |
| 7590 /* | |
| 7591 * First round: compute the length and allocate memory. | |
| 7592 * Second round: copy the text. | |
| 7593 */ | |
| 7594 for (round = 1; round <= 2; ++round) | |
| 7595 { | |
| 7596 lnum = submatch_mmatch->startpos[no].lnum; | |
| 7597 if (lnum < 0 || submatch_mmatch->endpos[no].lnum < 0) | |
| 7598 return NULL; | |
| 7599 | |
| 2011 | 7600 s = reg_getline_submatch(lnum) + submatch_mmatch->startpos[no].col; |
| 7 | 7601 if (s == NULL) /* anti-crash check, cannot happen? */ |
| 7602 break; | |
| 7603 if (submatch_mmatch->endpos[no].lnum == lnum) | |
| 7604 { | |
| 7605 /* Within one line: take form start to end col. */ | |
| 7606 len = submatch_mmatch->endpos[no].col | |
| 7607 - submatch_mmatch->startpos[no].col; | |
| 7608 if (round == 2) | |
| 418 | 7609 vim_strncpy(retval, s, len); |
| 7 | 7610 ++len; |
| 7611 } | |
| 7612 else | |
| 7613 { | |
| 7614 /* Multiple lines: take start line from start col, middle | |
| 7615 * lines completely and end line up to end col. */ | |
| 7616 len = (int)STRLEN(s); | |
| 7617 if (round == 2) | |
| 7618 { | |
| 7619 STRCPY(retval, s); | |
| 7620 retval[len] = '\n'; | |
| 7621 } | |
| 7622 ++len; | |
| 7623 ++lnum; | |
| 7624 while (lnum < submatch_mmatch->endpos[no].lnum) | |
| 7625 { | |
| 2011 | 7626 s = reg_getline_submatch(lnum++); |
| 7 | 7627 if (round == 2) |
| 7628 STRCPY(retval + len, s); | |
| 7629 len += (int)STRLEN(s); | |
| 7630 if (round == 2) | |
| 7631 retval[len] = '\n'; | |
| 7632 ++len; | |
| 7633 } | |
| 7634 if (round == 2) | |
| 2011 | 7635 STRNCPY(retval + len, reg_getline_submatch(lnum), |
| 7 | 7636 submatch_mmatch->endpos[no].col); |
| 7637 len += submatch_mmatch->endpos[no].col; | |
| 7638 if (round == 2) | |
| 7639 retval[len] = NUL; | |
| 7640 ++len; | |
| 7641 } | |
| 7642 | |
| 840 | 7643 if (retval == NULL) |
| 7 | 7644 { |
| 7645 retval = lalloc((long_u)len, TRUE); | |
| 840 | 7646 if (retval == NULL) |
| 7 | 7647 return NULL; |
| 7648 } | |
| 7649 } | |
| 7650 } | |
| 7651 else | |
| 7652 { | |
| 1815 | 7653 s = submatch_match->startp[no]; |
| 7654 if (s == NULL || submatch_match->endp[no] == NULL) | |
| 7 | 7655 retval = NULL; |
| 7656 else | |
| 7657 retval = vim_strnsave(s, (int)(submatch_match->endp[no] - s)); | |
| 7658 } | |
| 7659 | |
| 7660 return retval; | |
| 7661 } | |
| 7662 #endif |