1 /* Definitions for data structures and routines for the regular 2 expression library. 3 Copyright (C) 1985, 1989-1993, 1995-1998, 2000-2003, 2005-2013 Free Software 4 Foundation, Inc. 5 This file is part of the GNU C Library. 6 7 The GNU C Library is free software; you can redistribute it and/or 8 modify it under the terms of the GNU General Public 9 License as published by the Free Software Foundation; either 10 version 3 of the License, or (at your option) any later version. 11 12 The GNU C Library is distributed in the hope that it will be useful, 13 but WITHOUT ANY WARRANTY; without even the implied warranty of 14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 15 General Public License for more details. 16 17 You should have received a copy of the GNU General Public 18 License along with the GNU C Library; if not, see 19 <http://www.gnu.org/licenses/>. */ 20 21 #ifndef _REGEX_H 22 #define _REGEX_H 1 23 24 #include <sys/types.h> 25 26 /* Allow the use in C++ code. */ 27 #ifdef __cplusplus 28 extern "C" { 29 #endif 30 31 /* Define __USE_GNU to declare GNU extensions that violate the 32 POSIX name space rules. */ 33 #ifdef _GNU_SOURCE 34 # define __USE_GNU 1 35 #endif 36 37 #ifdef _REGEX_LARGE_OFFSETS 38 39 /* Use types and values that are wide enough to represent signed and 40 unsigned byte offsets in memory. This currently works only when 41 the regex code is used outside of the GNU C library; it is not yet 42 supported within glibc itself, and glibc users should not define 43 _REGEX_LARGE_OFFSETS. */ 44 45 /* The type of nonnegative object indexes. Traditionally, GNU regex 46 uses 'int' for these. Code that uses __re_idx_t should work 47 regardless of whether the type is signed. */ 48 typedef size_t __re_idx_t; 49 50 /* The type of object sizes. */ 51 typedef size_t __re_size_t; 52 53 /* The type of object sizes, in places where the traditional code 54 uses unsigned long int. */ 55 typedef size_t __re_long_size_t; 56 57 #else 58 59 /* The traditional GNU regex implementation mishandles strings longer 60 than INT_MAX. */ 61 typedef int __re_idx_t; 62 typedef unsigned int __re_size_t; 63 typedef unsigned long int __re_long_size_t; 64 65 #endif 66 67 /* The following two types have to be signed and unsigned integer type 68 wide enough to hold a value of a pointer. For most ANSI compilers 69 ptrdiff_t and size_t should be likely OK. Still size of these two 70 types is 2 for Microsoft C. Ugh... */ 71 typedef long int s_reg_t; 72 typedef unsigned long int active_reg_t; 73 74 /* The following bits are used to determine the regexp syntax we 75 recognize. The set/not-set meanings are chosen so that Emacs syntax 76 remains the value 0. The bits are given in alphabetical order, and 77 the definitions shifted by one from the previous bit; thus, when we 78 add or remove a bit, only one other definition need change. */ 79 typedef unsigned long int reg_syntax_t; 80 81 #ifdef __USE_GNU 82 /* If this bit is not set, then \ inside a bracket expression is literal. 83 If set, then such a \ quotes the following character. */ 84 # define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1) 85 86 /* If this bit is not set, then + and ? are operators, and \+ and \? are 87 literals. 88 If set, then \+ and \? are operators and + and ? are literals. */ 89 # define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1) 90 91 /* If this bit is set, then character classes are supported. They are: 92 [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:], 93 [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:]. 94 If not set, then character classes are not supported. */ 95 # define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1) 96 97 /* If this bit is set, then ^ and $ are always anchors (outside bracket 98 expressions, of course). 99 If this bit is not set, then it depends: 100 ^ is an anchor if it is at the beginning of a regular 101 expression or after an open-group or an alternation operator; 102 $ is an anchor if it is at the end of a regular expression, or 103 before a close-group or an alternation operator. 104 105 This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because 106 POSIX draft 11.2 says that * etc. in leading positions is undefined. 107 We already implemented a previous draft which made those constructs 108 invalid, though, so we haven't changed the code back. */ 109 # define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1) 110 111 /* If this bit is set, then special characters are always special 112 regardless of where they are in the pattern. 113 If this bit is not set, then special characters are special only in 114 some contexts; otherwise they are ordinary. Specifically, 115 * + ? and intervals are only special when not after the beginning, 116 open-group, or alternation operator. */ 117 # define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1) 118 119 /* If this bit is set, then *, +, ?, and { cannot be first in an re or 120 immediately after an alternation or begin-group operator. */ 121 # define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1) 122 123 /* If this bit is set, then . matches newline. 124 If not set, then it doesn't. */ 125 # define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1) 126 127 /* If this bit is set, then . doesn't match NUL. 128 If not set, then it does. */ 129 # define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1) 130 131 /* If this bit is set, nonmatching lists [^...] do not match newline. 132 If not set, they do. */ 133 # define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1) 134 135 /* If this bit is set, either \{...\} or {...} defines an 136 interval, depending on RE_NO_BK_BRACES. 137 If not set, \{, \}, {, and } are literals. */ 138 # define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1) 139 140 /* If this bit is set, +, ? and | aren't recognized as operators. 141 If not set, they are. */ 142 # define RE_LIMITED_OPS (RE_INTERVALS << 1) 143 144 /* If this bit is set, newline is an alternation operator. 145 If not set, newline is literal. */ 146 # define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1) 147 148 /* If this bit is set, then '{...}' defines an interval, and \{ and \} 149 are literals. 150 If not set, then '\{...\}' defines an interval. */ 151 # define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1) 152 153 /* If this bit is set, (...) defines a group, and \( and \) are literals. 154 If not set, \(...\) defines a group, and ( and ) are literals. */ 155 # define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1) 156 157 /* If this bit is set, then \<digit> matches <digit>. 158 If not set, then \<digit> is a back-reference. */ 159 # define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1) 160 161 /* If this bit is set, then | is an alternation operator, and \| is literal. 162 If not set, then \| is an alternation operator, and | is literal. */ 163 # define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1) 164 165 /* If this bit is set, then an ending range point collating higher 166 than the starting range point, as in [z-a], is invalid. 167 If not set, then when ending range point collates higher than the 168 starting range point, the range is ignored. */ 169 # define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1) 170 171 /* If this bit is set, then an unmatched ) is ordinary. 172 If not set, then an unmatched ) is invalid. */ 173 # define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1) 174 175 /* If this bit is set, succeed as soon as we match the whole pattern, 176 without further backtracking. */ 177 # define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1) 178 179 /* If this bit is set, do not process the GNU regex operators. 180 If not set, then the GNU regex operators are recognized. */ 181 # define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1) 182 183 /* If this bit is set, turn on internal regex debugging. 184 If not set, and debugging was on, turn it off. 185 This only works if regex.c is compiled -DDEBUG. 186 We define this bit always, so that all that's needed to turn on 187 debugging is to recompile regex.c; the calling code can always have 188 this bit set, and it won't affect anything in the normal case. */ 189 # define RE_DEBUG (RE_NO_GNU_OPS << 1) 190 191 /* If this bit is set, a syntactically invalid interval is treated as 192 a string of ordinary characters. For example, the ERE 'a{1' is 193 treated as 'a\{1'. */ 194 # define RE_INVALID_INTERVAL_ORD (RE_DEBUG << 1) 195 196 /* If this bit is set, then ignore case when matching. 197 If not set, then case is significant. */ 198 # define RE_ICASE (RE_INVALID_INTERVAL_ORD << 1) 199 200 /* This bit is used internally like RE_CONTEXT_INDEP_ANCHORS but only 201 for ^, because it is difficult to scan the regex backwards to find 202 whether ^ should be special. */ 203 # define RE_CARET_ANCHORS_HERE (RE_ICASE << 1) 204 205 /* If this bit is set, then \{ cannot be first in a regex or 206 immediately after an alternation, open-group or \} operator. */ 207 # define RE_CONTEXT_INVALID_DUP (RE_CARET_ANCHORS_HERE << 1) 208 209 /* If this bit is set, then no_sub will be set to 1 during 210 re_compile_pattern. */ 211 # define RE_NO_SUB (RE_CONTEXT_INVALID_DUP << 1) 212 #endif 213 214 /* This global variable defines the particular regexp syntax to use (for 215 some interfaces). When a regexp is compiled, the syntax used is 216 stored in the pattern buffer, so changing this does not affect 217 already-compiled regexps. */ 218 extern reg_syntax_t re_syntax_options; 219 220 #ifdef __USE_GNU 221 /* Define combinations of the above bits for the standard possibilities. 222 (The [[[ comments delimit what gets put into the Texinfo file, so 223 don't delete them!) */ 224 /* [[[begin syntaxes]]] */ 225 # define RE_SYNTAX_EMACS 0 226 227 # define RE_SYNTAX_AWK \ 228 (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \ 229 | RE_NO_BK_PARENS | RE_NO_BK_REFS \ 230 | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \ 231 | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \ 232 | RE_CHAR_CLASSES \ 233 | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS) 234 235 # define RE_SYNTAX_GNU_AWK \ 236 ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \ 237 | RE_INVALID_INTERVAL_ORD) \ 238 & ~(RE_DOT_NOT_NULL | RE_CONTEXT_INDEP_OPS \ 239 | RE_CONTEXT_INVALID_OPS )) 240 241 # define RE_SYNTAX_POSIX_AWK \ 242 (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \ 243 | RE_INTERVALS | RE_NO_GNU_OPS \ 244 | RE_INVALID_INTERVAL_ORD) 245 246 # define RE_SYNTAX_GREP \ 247 (RE_BK_PLUS_QM | RE_CHAR_CLASSES \ 248 | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \ 249 | RE_NEWLINE_ALT) 250 251 # define RE_SYNTAX_EGREP \ 252 (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \ 253 | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \ 254 | RE_NEWLINE_ALT | RE_NO_BK_PARENS \ 255 | RE_NO_BK_VBAR) 256 257 # define RE_SYNTAX_POSIX_EGREP \ 258 (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES \ 259 | RE_INVALID_INTERVAL_ORD) 260 261 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */ 262 # define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC 263 264 # define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC 265 266 /* Syntax bits common to both basic and extended POSIX regex syntax. */ 267 # define _RE_SYNTAX_POSIX_COMMON \ 268 (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \ 269 | RE_INTERVALS | RE_NO_EMPTY_RANGES) 270 271 # define RE_SYNTAX_POSIX_BASIC \ 272 (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM | RE_CONTEXT_INVALID_DUP) 273 274 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes 275 RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this 276 isn't minimal, since other operators, such as \`, aren't disabled. */ 277 # define RE_SYNTAX_POSIX_MINIMAL_BASIC \ 278 (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS) 279 280 # define RE_SYNTAX_POSIX_EXTENDED \ 281 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ 282 | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \ 283 | RE_NO_BK_PARENS | RE_NO_BK_VBAR \ 284 | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD) 285 286 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is 287 removed and RE_NO_BK_REFS is added. */ 288 # define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \ 289 (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ 290 | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \ 291 | RE_NO_BK_PARENS | RE_NO_BK_REFS \ 292 | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD) 293 /* [[[end syntaxes]]] */ 294 295 /* Maximum number of duplicates an interval can allow. POSIX-conforming 296 systems might define this in <limits.h>, but we want our 297 value, so remove any previous define. */ 298 # ifdef _REGEX_INCLUDE_LIMITS_H 299 # include <limits.h> 300 # endif 301 # ifdef RE_DUP_MAX 302 # undef RE_DUP_MAX 303 # endif 304 305 /* RE_DUP_MAX is 2**15 - 1 because an earlier implementation stored 306 the counter as a 2-byte signed integer. This is no longer true, so 307 RE_DUP_MAX could be increased to (INT_MAX / 10 - 1), or to 308 ((SIZE_MAX - 9) / 10) if _REGEX_LARGE_OFFSETS is defined. 309 However, there would be a huge performance problem if someone 310 actually used a pattern like a\{214748363\}, so RE_DUP_MAX retains 311 its historical value. */ 312 # define RE_DUP_MAX (0x7fff) 313 #endif 314 315 316 /* POSIX 'cflags' bits (i.e., information for 'regcomp'). */ 317 318 /* If this bit is set, then use extended regular expression syntax. 319 If not set, then use basic regular expression syntax. */ 320 #define REG_EXTENDED 1 321 322 /* If this bit is set, then ignore case when matching. 323 If not set, then case is significant. */ 324 #define REG_ICASE (1 << 1) 325 326 /* If this bit is set, then anchors do not match at newline 327 characters in the string. 328 If not set, then anchors do match at newlines. */ 329 #define REG_NEWLINE (1 << 2) 330 331 /* If this bit is set, then report only success or fail in regexec. 332 If not set, then returns differ between not matching and errors. */ 333 #define REG_NOSUB (1 << 3) 334 335 336 /* POSIX 'eflags' bits (i.e., information for regexec). */ 337 338 /* If this bit is set, then the beginning-of-line operator doesn't match 339 the beginning of the string (presumably because it's not the 340 beginning of a line). 341 If not set, then the beginning-of-line operator does match the 342 beginning of the string. */ 343 #define REG_NOTBOL 1 344 345 /* Like REG_NOTBOL, except for the end-of-line. */ 346 #define REG_NOTEOL (1 << 1) 347 348 /* Use PMATCH[0] to delimit the start and end of the search in the 349 buffer. */ 350 #define REG_STARTEND (1 << 2) 351 352 353 /* If any error codes are removed, changed, or added, update the 354 '__re_error_msgid' table in regcomp.c. */ 355 356 typedef enum 357 { 358 _REG_ENOSYS = -1, /* This will never happen for this implementation. */ 359 _REG_NOERROR = 0, /* Success. */ 360 _REG_NOMATCH, /* Didn't find a match (for regexec). */ 361 362 /* POSIX regcomp return error codes. (In the order listed in the 363 standard.) */ 364 _REG_BADPAT, /* Invalid pattern. */ 365 _REG_ECOLLATE, /* Invalid collating element. */ 366 _REG_ECTYPE, /* Invalid character class name. */ 367 _REG_EESCAPE, /* Trailing backslash. */ 368 _REG_ESUBREG, /* Invalid back reference. */ 369 _REG_EBRACK, /* Unmatched left bracket. */ 370 _REG_EPAREN, /* Parenthesis imbalance. */ 371 _REG_EBRACE, /* Unmatched \{. */ 372 _REG_BADBR, /* Invalid contents of \{\}. */ 373 _REG_ERANGE, /* Invalid range end. */ 374 _REG_ESPACE, /* Ran out of memory. */ 375 _REG_BADRPT, /* No preceding re for repetition op. */ 376 377 /* Error codes we've added. */ 378 _REG_EEND, /* Premature end. */ 379 _REG_ESIZE, /* Too large (e.g., repeat count too large). */ 380 _REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */ 381 } reg_errcode_t; 382 383 #if defined _XOPEN_SOURCE || defined __USE_XOPEN2K 384 # define REG_ENOSYS _REG_ENOSYS 385 #endif 386 #define REG_NOERROR _REG_NOERROR 387 #define REG_NOMATCH _REG_NOMATCH 388 #define REG_BADPAT _REG_BADPAT 389 #define REG_ECOLLATE _REG_ECOLLATE 390 #define REG_ECTYPE _REG_ECTYPE 391 #define REG_EESCAPE _REG_EESCAPE 392 #define REG_ESUBREG _REG_ESUBREG 393 #define REG_EBRACK _REG_EBRACK 394 #define REG_EPAREN _REG_EPAREN 395 #define REG_EBRACE _REG_EBRACE 396 #define REG_BADBR _REG_BADBR 397 #define REG_ERANGE _REG_ERANGE 398 #define REG_ESPACE _REG_ESPACE 399 #define REG_BADRPT _REG_BADRPT 400 #define REG_EEND _REG_EEND 401 #define REG_ESIZE _REG_ESIZE 402 #define REG_ERPAREN _REG_ERPAREN 403 404 /* This data structure represents a compiled pattern. Before calling 405 the pattern compiler, the fields 'buffer', 'allocated', 'fastmap', 406 and 'translate' can be set. After the pattern has been compiled, 407 the fields 're_nsub', 'not_bol' and 'not_eol' are available. All 408 other fields are private to the regex routines. */ 409 410 #ifndef RE_TRANSLATE_TYPE 411 # define __RE_TRANSLATE_TYPE unsigned char * 412 # ifdef __USE_GNU 413 # define RE_TRANSLATE_TYPE __RE_TRANSLATE_TYPE 414 # endif 415 #endif 416 417 #ifdef __USE_GNU 418 # define __REPB_PREFIX(name) name 419 #else 420 # define __REPB_PREFIX(name) __##name 421 #endif 422 423 struct re_pattern_buffer 424 { 425 /* Space that holds the compiled pattern. The type 426 'struct re_dfa_t' is private and is not declared here. */ 427 struct re_dfa_t *__REPB_PREFIX(buffer); 428 429 /* Number of bytes to which 'buffer' points. */ 430 __re_long_size_t __REPB_PREFIX(allocated); 431 432 /* Number of bytes actually used in 'buffer'. */ 433 __re_long_size_t __REPB_PREFIX(used); 434 435 /* Syntax setting with which the pattern was compiled. */ 436 reg_syntax_t __REPB_PREFIX(syntax); 437 438 /* Pointer to a fastmap, if any, otherwise zero. re_search uses the 439 fastmap, if there is one, to skip over impossible starting points 440 for matches. */ 441 char *__REPB_PREFIX(fastmap); 442 443 /* Either a translate table to apply to all characters before 444 comparing them, or zero for no translation. The translation is 445 applied to a pattern when it is compiled and to a string when it 446 is matched. */ 447 __RE_TRANSLATE_TYPE __REPB_PREFIX(translate); 448 449 /* Number of subexpressions found by the compiler. */ 450 size_t re_nsub; 451 452 /* Zero if this pattern cannot match the empty string, one else. 453 Well, in truth it's used only in 're_search_2', to see whether or 454 not we should use the fastmap, so we don't set this absolutely 455 perfectly; see 're_compile_fastmap' (the "duplicate" case). */ 456 unsigned __REPB_PREFIX(can_be_null) : 1; 457 458 /* If REGS_UNALLOCATED, allocate space in the 'regs' structure 459 for 'max (RE_NREGS, re_nsub + 1)' groups. 460 If REGS_REALLOCATE, reallocate space if necessary. 461 If REGS_FIXED, use what's there. */ 462 #ifdef __USE_GNU 463 # define REGS_UNALLOCATED 0 464 # define REGS_REALLOCATE 1 465 # define REGS_FIXED 2 466 #endif 467 unsigned __REPB_PREFIX(regs_allocated) : 2; 468 469 /* Set to zero when 're_compile_pattern' compiles a pattern; set to 470 one by 're_compile_fastmap' if it updates the fastmap. */ 471 unsigned __REPB_PREFIX(fastmap_accurate) : 1; 472 473 /* If set, 're_match_2' does not return information about 474 subexpressions. */ 475 unsigned __REPB_PREFIX(no_sub) : 1; 476 477 /* If set, a beginning-of-line anchor doesn't match at the beginning 478 of the string. */ 479 unsigned __REPB_PREFIX(not_bol) : 1; 480 481 /* Similarly for an end-of-line anchor. */ 482 unsigned __REPB_PREFIX(not_eol) : 1; 483 484 /* If true, an anchor at a newline matches. */ 485 unsigned __REPB_PREFIX(newline_anchor) : 1; 486 }; 487 488 typedef struct re_pattern_buffer regex_t; 489 490 /* Type for byte offsets within the string. POSIX mandates this. */ 491 #ifdef _REGEX_LARGE_OFFSETS 492 /* POSIX 1003.1-2008 requires that regoff_t be at least as wide as 493 ptrdiff_t and ssize_t. We don't know of any hosts where ptrdiff_t 494 is wider than ssize_t, so ssize_t is safe. */ 495 typedef ssize_t regoff_t; 496 #else 497 /* The traditional GNU regex implementation mishandles strings longer 498 than INT_MAX. */ 499 typedef int regoff_t; 500 #endif 501 502 503 #ifdef __USE_GNU 504 /* This is the structure we store register match data in. See 505 regex.texinfo for a full description of what registers match. */ 506 struct re_registers 507 { 508 __re_size_t num_regs; 509 regoff_t *start; 510 regoff_t *end; 511 }; 512 513 514 /* If 'regs_allocated' is REGS_UNALLOCATED in the pattern buffer, 515 're_match_2' returns information about at least this many registers 516 the first time a 'regs' structure is passed. */ 517 # ifndef RE_NREGS 518 # define RE_NREGS 30 519 # endif 520 #endif 521 522 523 /* POSIX specification for registers. Aside from the different names than 524 're_registers', POSIX uses an array of structures, instead of a 525 structure of arrays. */ 526 typedef struct 527 { 528 regoff_t rm_so; /* Byte offset from string's start to substring's start. */ 529 regoff_t rm_eo; /* Byte offset from string's start to substring's end. */ 530 } regmatch_t; 531 532 /* Declarations for routines. */ 533 534 #ifdef __USE_GNU 535 /* Sets the current default syntax to SYNTAX, and return the old syntax. 536 You can also simply assign to the 're_syntax_options' variable. */ 537 extern reg_syntax_t re_set_syntax (reg_syntax_t __syntax); 538 539 /* Compile the regular expression PATTERN, with length LENGTH 540 and syntax given by the global 're_syntax_options', into the buffer 541 BUFFER. Return NULL if successful, and an error string if not. 542 543 To free the allocated storage, you must call 'regfree' on BUFFER. 544 Note that the translate table must either have been initialised by 545 'regcomp', with a malloc'ed value, or set to NULL before calling 546 'regfree'. */ 547 extern const char *re_compile_pattern (const char *__pattern, size_t __length, 548 struct re_pattern_buffer *__buffer); 549 550 551 /* Compile a fastmap for the compiled pattern in BUFFER; used to 552 accelerate searches. Return 0 if successful and -2 if was an 553 internal error. */ 554 extern int re_compile_fastmap (struct re_pattern_buffer *__buffer); 555 556 557 /* Search in the string STRING (with length LENGTH) for the pattern 558 compiled into BUFFER. Start searching at position START, for RANGE 559 characters. Return the starting position of the match, -1 for no 560 match, or -2 for an internal error. Also return register 561 information in REGS (if REGS and BUFFER->no_sub are nonzero). */ 562 extern regoff_t re_search (struct re_pattern_buffer *__buffer, 563 const char *__string, __re_idx_t __length, 564 __re_idx_t __start, regoff_t __range, 565 struct re_registers *__regs); 566 567 568 /* Like 're_search', but search in the concatenation of STRING1 and 569 STRING2. Also, stop searching at index START + STOP. */ 570 extern regoff_t re_search_2 (struct re_pattern_buffer *__buffer, 571 const char *__string1, __re_idx_t __length1, 572 const char *__string2, __re_idx_t __length2, 573 __re_idx_t __start, regoff_t __range, 574 struct re_registers *__regs, 575 __re_idx_t __stop); 576 577 578 /* Like 're_search', but return how many characters in STRING the regexp 579 in BUFFER matched, starting at position START. */ 580 extern regoff_t re_match (struct re_pattern_buffer *__buffer, 581 const char *__string, __re_idx_t __length, 582 __re_idx_t __start, struct re_registers *__regs); 583 584 585 /* Relates to 're_match' as 're_search_2' relates to 're_search'. */ 586 extern regoff_t re_match_2 (struct re_pattern_buffer *__buffer, 587 const char *__string1, __re_idx_t __length1, 588 const char *__string2, __re_idx_t __length2, 589 __re_idx_t __start, struct re_registers *__regs, 590 __re_idx_t __stop); 591 592 593 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and 594 ENDS. Subsequent matches using BUFFER and REGS will use this memory 595 for recording register information. STARTS and ENDS must be 596 allocated with malloc, and must each be at least 'NUM_REGS * sizeof 597 (regoff_t)' bytes long. 598 599 If NUM_REGS == 0, then subsequent matches should allocate their own 600 register data. 601 602 Unless this function is called, the first search or match using 603 BUFFER will allocate its own register data, without 604 freeing the old data. */ 605 extern void re_set_registers (struct re_pattern_buffer *__buffer, 606 struct re_registers *__regs, 607 __re_size_t __num_regs, 608 regoff_t *__starts, regoff_t *__ends); 609 #endif /* Use GNU */ 610 611 #if defined _REGEX_RE_COMP || (defined _LIBC && defined __USE_BSD) 612 # ifndef _CRAY 613 /* 4.2 bsd compatibility. */ 614 extern char *re_comp (const char *); 615 extern int re_exec (const char *); 616 # endif 617 #endif 618 619 /* GCC 2.95 and later have "__restrict"; C99 compilers have 620 "restrict", and "configure" may have defined "restrict". 621 Other compilers use __restrict, __restrict__, and _Restrict, and 622 'configure' might #define 'restrict' to those words, so pick a 623 different name. */ 624 #ifndef _Restrict_ 625 # if 199901L <= __STDC_VERSION__ 626 # define _Restrict_ restrict 627 # elif 2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__) 628 # define _Restrict_ __restrict 629 # else 630 # define _Restrict_ 631 # endif 632 #endif 633 /* gcc 3.1 and up support the [restrict] syntax. Don't trust 634 sys/cdefs.h's definition of __restrict_arr, though, as it 635 mishandles gcc -ansi -pedantic. */ 636 #ifndef _Restrict_arr_ 637 # if ((199901L <= __STDC_VERSION__ \ 638 || ((3 < __GNUC__ || (3 == __GNUC__ && 1 <= __GNUC_MINOR__)) \ 639 && !defined __STRICT_ANSI__)) \ 640 && !defined __GNUG__) 641 # define _Restrict_arr_ _Restrict_ 642 # else 643 # define _Restrict_arr_ 644 # endif 645 #endif 646 647 /* POSIX compatibility. */ 648 extern int regcomp (regex_t *_Restrict_ __preg, 649 const char *_Restrict_ __pattern, 650 int __cflags); 651 652 extern int regexec (const regex_t *_Restrict_ __preg, 653 const char *_Restrict_ __string, size_t __nmatch, 654 regmatch_t __pmatch[_Restrict_arr_], 655 int __eflags); 656 657 extern size_t regerror (int __errcode, const regex_t *_Restrict_ __preg, 658 char *_Restrict_ __errbuf, size_t __errbuf_size); 659 660 extern void regfree (regex_t *__preg); 661 662 663 #ifdef __cplusplus 664 } 665 #endif /* C++ */ 666 667 #endif /* regex.h */ 668