u-tools/apps/tar/gnu/regex.h

   1 /* -*- buffer-read-only: t -*- vi: set ro: */
   2 /* DO NOT EDIT! GENERATED AUTOMATICALLY! */
   3 /* Definitions for data structures and routines for the regular
   4    expression library.
   5    Copyright (C) 1985, 1989-1993, 1995-1998, 2000-2003, 2005-2006, 2009-2011
   6    Free Software Foundation, Inc.
   7    This file is part of the GNU C Library.
   8
   9    This program is free software; you can redistribute it and/or modify
  10    it under the terms of the GNU General Public License as published by
  11    the Free Software Foundation; either version 3, or (at your option)
  12    any later version.
  13
  14    This program is distributed in the hope that it will be useful,
  15    but WITHOUT ANY WARRANTY; without even the implied warranty of
  16    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17    GNU General Public License for more details.
  18
  19    You should have received a copy of the GNU General Public License along
  20    with this program; if not, write to the Free Software Foundation,
  21    Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
  22
  23 #ifndef _REGEX_H
  24 #define _REGEX_H 1
  25
  26 #include <sys/types.h>
  27
  28 /* Allow the use in C++ code.  */
  29 #ifdef __cplusplus
  30 extern "C" {
  31 #endif
  32
  33 /* Define __USE_GNU_REGEX to declare GNU extensions that violate the
  34    POSIX name space rules.  */
  35 #undef __USE_GNU_REGEX
  36 #if (defined _GNU_SOURCE                                        \
  37      || (!defined _POSIX_C_SOURCE && !defined _POSIX_SOURCE     \
  38          && !defined _XOPEN_SOURCE))
  39 # define __USE_GNU_REGEX 1
  40 #endif
  41
  42 #ifdef _REGEX_LARGE_OFFSETS
  43
  44 /* Use types and values that are wide enough to represent signed and
  45    unsigned byte offsets in memory.  This currently works only when
  46    the regex code is used outside of the GNU C library; it is not yet
  47    supported within glibc itself, and glibc users should not define
  48    _REGEX_LARGE_OFFSETS.  */
  49
  50 /* The type of the offset of a byte within a string.
  51    For historical reasons POSIX 1003.1-2004 requires that regoff_t be
  52    at least as wide as off_t.  However, many common POSIX platforms set
  53    regoff_t to the more-sensible ssize_t and the Open Group has
  54    signalled its intention to change the requirement to be that
  55    regoff_t be at least as wide as ptrdiff_t and ssize_t; see XBD ERN
  56    60 (2005-08-25).  We don't know of any hosts where ssize_t or
  57    ptrdiff_t is wider than ssize_t, so ssize_t is safe.  */
  58 typedef ssize_t regoff_t;
  59
  60 /* The type of nonnegative object indexes.  Traditionally, GNU regex
  61    uses 'int' for these.  Code that uses __re_idx_t should work
  62    regardless of whether the type is signed.  */
  63 typedef size_t __re_idx_t;
  64
  65 /* The type of object sizes.  */
  66 typedef size_t __re_size_t;
  67
  68 /* The type of object sizes, in places where the traditional code
  69    uses unsigned long int.  */
  70 typedef size_t __re_long_size_t;
  71
  72 #else
  73
  74 /* Use types that are binary-compatible with the traditional GNU regex
  75    implementation, which mishandles strings longer than INT_MAX.  */
  76
  77 typedef int regoff_t;
  78 typedef int __re_idx_t;
  79 typedef unsigned int __re_size_t;
  80 typedef unsigned long int __re_long_size_t;
  81
  82 #endif
  83
  84 /* The following two types have to be signed and unsigned integer type
  85    wide enough to hold a value of a pointer.  For most ANSI compilers
  86    ptrdiff_t and size_t should be likely OK.  Still size of these two
  87    types is 2 for Microsoft C.  Ugh... */
  88 typedef long int s_reg_t;
  89 typedef unsigned long int active_reg_t;
  90
  91 /* The following bits are used to determine the regexp syntax we
  92    recognize.  The set/not-set meanings are chosen so that Emacs syntax
  93    remains the value 0.  The bits are given in alphabetical order, and
  94    the definitions shifted by one from the previous bit; thus, when we
  95    add or remove a bit, only one other definition need change.  */
  96 typedef unsigned long int reg_syntax_t;
  97
  98 #ifdef __USE_GNU_REGEX
  99
 100 /* If this bit is not set, then \ inside a bracket expression is literal.
 101    If set, then such a \ quotes the following character.  */
 102 # define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1)
 103
 104 /* If this bit is not set, then + and ? are operators, and \+ and \? are
 105      literals.
 106    If set, then \+ and \? are operators and + and ? are literals.  */
 107 # define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1)
 108
 109 /* If this bit is set, then character classes are supported.  They are:
 110      [:alpha:], [:upper:], [:lower:],  [:digit:], [:alnum:], [:xdigit:],
 111      [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:].
 112    If not set, then character classes are not supported.  */
 113 # define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1)
 114
 115 /* If this bit is set, then ^ and $ are always anchors (outside bracket
 116      expressions, of course).
 117    If this bit is not set, then it depends:
 118         ^  is an anchor if it is at the beginning of a regular
 119            expression or after an open-group or an alternation operator;
 120         $  is an anchor if it is at the end of a regular expression, or
 121            before a close-group or an alternation operator.
 122
 123    This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because
 124    POSIX draft 11.2 says that * etc. in leading positions is undefined.
 125    We already implemented a previous draft which made those constructs
 126    invalid, though, so we haven't changed the code back.  */
 127 # define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1)
 128
 129 /* If this bit is set, then special characters are always special
 130      regardless of where they are in the pattern.
 131    If this bit is not set, then special characters are special only in
 132      some contexts; otherwise they are ordinary.  Specifically,
 133      * + ? and intervals are only special when not after the beginning,
 134      open-group, or alternation operator.  */
 135 # define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1)
 136
 137 /* If this bit is set, then *, +, ?, and { cannot be first in an re or
 138      immediately after an alternation or begin-group operator.  */
 139 # define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1)
 140
 141 /* If this bit is set, then . matches newline.
 142    If not set, then it doesn't.  */
 143 # define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1)
 144
 145 /* If this bit is set, then . doesn't match NUL.
 146    If not set, then it does.  */
 147 # define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1)
 148
 149 /* If this bit is set, nonmatching lists [^...] do not match newline.
 150    If not set, they do.  */
 151 # define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1)
 152
 153 /* If this bit is set, either \{...\} or {...} defines an
 154      interval, depending on RE_NO_BK_BRACES.
 155    If not set, \{, \}, {, and } are literals.  */
 156 # define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1)
 157
 158 /* If this bit is set, +, ? and | aren't recognized as operators.
 159    If not set, they are.  */
 160 # define RE_LIMITED_OPS (RE_INTERVALS << 1)
 161
 162 /* If this bit is set, newline is an alternation operator.
 163    If not set, newline is literal.  */
 164 # define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1)
 165
 166 /* If this bit is set, then `{...}' defines an interval, and \{ and \}
 167      are literals.
 168   If not set, then `\{...\}' defines an interval.  */
 169 # define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1)
 170
 171 /* If this bit is set, (...) defines a group, and \( and \) are literals.
 172    If not set, \(...\) defines a group, and ( and ) are literals.  */
 173 # define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1)
 174
 175 /* If this bit is set, then \<digit> matches <digit>.
 176    If not set, then \<digit> is a back-reference.  */
 177 # define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1)
 178
 179 /* If this bit is set, then | is an alternation operator, and \| is literal.
 180    If not set, then \| is an alternation operator, and | is literal.  */
 181 # define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1)
 182
 183 /* If this bit is set, then an ending range point collating higher
 184      than the starting range point, as in [z-a], is invalid.
 185    If not set, then when ending range point collates higher than the
 186      starting range point, the range is ignored.  */
 187 # define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1)
 188
 189 /* If this bit is set, then an unmatched ) is ordinary.
 190    If not set, then an unmatched ) is invalid.  */
 191 # define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1)
 192
 193 /* If this bit is set, succeed as soon as we match the whole pattern,
 194    without further backtracking.  */
 195 # define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1)
 196
 197 /* If this bit is set, do not process the GNU regex operators.
 198    If not set, then the GNU regex operators are recognized. */
 199 # define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1)
 200
 201 /* If this bit is set, turn on internal regex debugging.
 202    If not set, and debugging was on, turn it off.
 203    This only works if regex.c is compiled -DDEBUG.
 204    We define this bit always, so that all that's needed to turn on
 205    debugging is to recompile regex.c; the calling code can always have
 206    this bit set, and it won't affect anything in the normal case. */
 207 # define RE_DEBUG (RE_NO_GNU_OPS << 1)
 208
 209 /* If this bit is set, a syntactically invalid interval is treated as
 210    a string of ordinary characters.  For example, the ERE 'a{1' is
 211    treated as 'a\{1'.  */
 212 # define RE_INVALID_INTERVAL_ORD (RE_DEBUG << 1)
 213
 214 /* If this bit is set, then ignore case when matching.
 215    If not set, then case is significant.  */
 216 # define RE_ICASE (RE_INVALID_INTERVAL_ORD << 1)
 217
 218 /* This bit is used internally like RE_CONTEXT_INDEP_ANCHORS but only
 219    for ^, because it is difficult to scan the regex backwards to find
 220    whether ^ should be special.  */
 221 # define RE_CARET_ANCHORS_HERE (RE_ICASE << 1)
 222
 223 /* If this bit is set, then \{ cannot be first in a regex or
 224    immediately after an alternation, open-group or \} operator.  */
 225 # define RE_CONTEXT_INVALID_DUP (RE_CARET_ANCHORS_HERE << 1)
 226
 227 /* If this bit is set, then no_sub will be set to 1 during
 228    re_compile_pattern.  */
 229 # define RE_NO_SUB (RE_CONTEXT_INVALID_DUP << 1)
 230
 231 #endif /* defined __USE_GNU_REGEX */
 232
 233 /* This global variable defines the particular regexp syntax to use (for
 234    some interfaces).  When a regexp is compiled, the syntax used is
 235    stored in the pattern buffer, so changing this does not affect
 236    already-compiled regexps.  */
 237 extern reg_syntax_t re_syntax_options;
 238 \f
 239 #ifdef __USE_GNU_REGEX
 240 /* Define combinations of the above bits for the standard possibilities.
 241    (The [[[ comments delimit what gets put into the Texinfo file, so
 242    don't delete them!)  */
 243 /* [[[begin syntaxes]]] */
 244 # define RE_SYNTAX_EMACS 0
 245
 246 # define RE_SYNTAX_AWK                                                  \
 247   (RE_BACKSLASH_ESCAPE_IN_LISTS   | RE_DOT_NOT_NULL                     \
 248    | RE_NO_BK_PARENS              | RE_NO_BK_REFS                       \
 249    | RE_NO_BK_VBAR                | RE_NO_EMPTY_RANGES                  \
 250    | RE_DOT_NEWLINE               | RE_CONTEXT_INDEP_ANCHORS            \
 251    | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS)
 252
 253 # define RE_SYNTAX_GNU_AWK                                              \
 254   ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DEBUG) \
 255    & ~(RE_DOT_NOT_NULL | RE_INTERVALS | RE_CONTEXT_INDEP_OPS            \
 256        | RE_CONTEXT_INVALID_OPS ))
 257
 258 # define RE_SYNTAX_POSIX_AWK                                            \
 259   (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS              \
 260    | RE_INTERVALS           | RE_NO_GNU_OPS)
 261
 262 # define RE_SYNTAX_GREP                                                 \
 263   (RE_BK_PLUS_QM              | RE_CHAR_CLASSES                         \
 264    | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS                            \
 265    | RE_NEWLINE_ALT)
 266
 267 # define RE_SYNTAX_EGREP                                                \
 268   (RE_CHAR_CLASSES        | RE_CONTEXT_INDEP_ANCHORS                    \
 269    | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE                    \
 270    | RE_NEWLINE_ALT       | RE_NO_BK_PARENS                             \
 271    | RE_NO_BK_VBAR)
 272
 273 # define RE_SYNTAX_POSIX_EGREP                                          \
 274   (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES                     \
 275    | RE_INVALID_INTERVAL_ORD)
 276
 277 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff.  */
 278 # define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
 279
 280 # define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
 281
 282 /* Syntax bits common to both basic and extended POSIX regex syntax.  */
 283 # define _RE_SYNTAX_POSIX_COMMON                                        \
 284   (RE_CHAR_CLASSES | RE_DOT_NEWLINE      | RE_DOT_NOT_NULL              \
 285    | RE_INTERVALS  | RE_NO_EMPTY_RANGES)
 286
 287 # define RE_SYNTAX_POSIX_BASIC                                          \
 288   (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM | RE_CONTEXT_INVALID_DUP)
 289
 290 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
 291    RE_LIMITED_OPS, i.e., \? \+ \| are not recognized.  Actually, this
 292    isn't minimal, since other operators, such as \`, aren't disabled.  */
 293 # define RE_SYNTAX_POSIX_MINIMAL_BASIC                                  \
 294   (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
 295
 296 # define RE_SYNTAX_POSIX_EXTENDED                                       \
 297   (_RE_SYNTAX_POSIX_COMMON  | RE_CONTEXT_INDEP_ANCHORS                  \
 298    | RE_CONTEXT_INDEP_OPS   | RE_NO_BK_BRACES                           \
 299    | RE_NO_BK_PARENS        | RE_NO_BK_VBAR                             \
 300    | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD)
 301
 302 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is
 303    removed and RE_NO_BK_REFS is added.  */
 304 # define RE_SYNTAX_POSIX_MINIMAL_EXTENDED                               \
 305   (_RE_SYNTAX_POSIX_COMMON  | RE_CONTEXT_INDEP_ANCHORS                  \
 306    | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES                           \
 307    | RE_NO_BK_PARENS        | RE_NO_BK_REFS                             \
 308    | RE_NO_BK_VBAR          | RE_UNMATCHED_RIGHT_PAREN_ORD)
 309 /* [[[end syntaxes]]] */
 310
 311 #endif /* defined __USE_GNU_REGEX */
 312 \f
 313 #ifdef __USE_GNU_REGEX
 314
 315 /* Maximum number of duplicates an interval can allow.  POSIX-conforming
 316    systems might define this in <limits.h>, but we want our
 317    value, so remove any previous define.  */
 318 # ifdef RE_DUP_MAX
 319 #  undef RE_DUP_MAX
 320 # endif
 321
 322 /* RE_DUP_MAX is 2**15 - 1 because an earlier implementation stored
 323    the counter as a 2-byte signed integer.  This is no longer true, so
 324    RE_DUP_MAX could be increased to (INT_MAX / 10 - 1), or to
 325    ((SIZE_MAX - 2) / 10 - 1) if _REGEX_LARGE_OFFSETS is defined.
 326    However, there would be a huge performance problem if someone
 327    actually used a pattern like a\{214748363\}, so RE_DUP_MAX retains
 328    its historical value.  */
 329 # define RE_DUP_MAX (0x7fff)
 330
 331 #endif /* defined __USE_GNU_REGEX */
 332
 333
 334 /* POSIX `cflags' bits (i.e., information for `regcomp').  */
 335
 336 /* If this bit is set, then use extended regular expression syntax.
 337    If not set, then use basic regular expression syntax.  */
 338 #define REG_EXTENDED 1
 339
 340 /* If this bit is set, then ignore case when matching.
 341    If not set, then case is significant.  */
 342 #define REG_ICASE (1 << 1)
 343
 344 /* If this bit is set, then anchors do not match at newline
 345      characters in the string.
 346    If not set, then anchors do match at newlines.  */
 347 #define REG_NEWLINE (1 << 2)
 348
 349 /* If this bit is set, then report only success or fail in regexec.
 350    If not set, then returns differ between not matching and errors.  */
 351 #define REG_NOSUB (1 << 3)
 352
 353
 354 /* POSIX `eflags' bits (i.e., information for regexec).  */
 355
 356 /* If this bit is set, then the beginning-of-line operator doesn't match
 357      the beginning of the string (presumably because it's not the
 358      beginning of a line).
 359    If not set, then the beginning-of-line operator does match the
 360      beginning of the string.  */
 361 #define REG_NOTBOL 1
 362
 363 /* Like REG_NOTBOL, except for the end-of-line.  */
 364 #define REG_NOTEOL (1 << 1)
 365
 366 /* Use PMATCH[0] to delimit the start and end of the search in the
 367    buffer.  */
 368 #define REG_STARTEND (1 << 2)
 369
 370
 371 /* If any error codes are removed, changed, or added, update the
 372    `__re_error_msgid' table in regcomp.c.  */
 373
 374 typedef enum
 375 {
 376   _REG_ENOSYS = -1,     /* This will never happen for this implementation.  */
 377   _REG_NOERROR = 0,     /* Success.  */
 378   _REG_NOMATCH,         /* Didn't find a match (for regexec).  */
 379
 380   /* POSIX regcomp return error codes.  (In the order listed in the
 381      standard.)  */
 382   _REG_BADPAT,          /* Invalid pattern.  */
 383   _REG_ECOLLATE,        /* Invalid collating element.  */
 384   _REG_ECTYPE,          /* Invalid character class name.  */
 385   _REG_EESCAPE,         /* Trailing backslash.  */
 386   _REG_ESUBREG,         /* Invalid back reference.  */
 387   _REG_EBRACK,          /* Unmatched left bracket.  */
 388   _REG_EPAREN,          /* Parenthesis imbalance.  */
 389   _REG_EBRACE,          /* Unmatched \{.  */
 390   _REG_BADBR,           /* Invalid contents of \{\}.  */
 391   _REG_ERANGE,          /* Invalid range end.  */
 392   _REG_ESPACE,          /* Ran out of memory.  */
 393   _REG_BADRPT,          /* No preceding re for repetition op.  */
 394
 395   /* Error codes we've added.  */
 396   _REG_EEND,            /* Premature end.  */
 397   _REG_ESIZE,           /* Compiled pattern bigger than 2^16 bytes.  */
 398   _REG_ERPAREN          /* Unmatched ) or \); not returned from regcomp.  */
 399 } reg_errcode_t;
 400
 401 #ifdef _XOPEN_SOURCE
 402 # define REG_ENOSYS     _REG_ENOSYS
 403 #endif
 404 #define REG_NOERROR     _REG_NOERROR
 405 #define REG_NOMATCH     _REG_NOMATCH
 406 #define REG_BADPAT      _REG_BADPAT
 407 #define REG_ECOLLATE    _REG_ECOLLATE
 408 #define REG_ECTYPE      _REG_ECTYPE
 409 #define REG_EESCAPE     _REG_EESCAPE
 410 #define REG_ESUBREG     _REG_ESUBREG
 411 #define REG_EBRACK      _REG_EBRACK
 412 #define REG_EPAREN      _REG_EPAREN
 413 #define REG_EBRACE      _REG_EBRACE
 414 #define REG_BADBR       _REG_BADBR
 415 #define REG_ERANGE      _REG_ERANGE
 416 #define REG_ESPACE      _REG_ESPACE
 417 #define REG_BADRPT      _REG_BADRPT
 418 #define REG_EEND        _REG_EEND
 419 #define REG_ESIZE       _REG_ESIZE
 420 #define REG_ERPAREN     _REG_ERPAREN
 421 \f
 422 /* struct re_pattern_buffer normally uses member names like `buffer'
 423    that POSIX does not allow.  In POSIX mode these members have names
 424    with leading `re_' (e.g., `re_buffer').  */
 425 #ifdef __USE_GNU_REGEX
 426 # define _REG_RE_NAME(id) id
 427 # define _REG_RM_NAME(id) id
 428 #else
 429 # define _REG_RE_NAME(id) re_##id
 430 # define _REG_RM_NAME(id) rm_##id
 431 #endif
 432
 433 /* The user can specify the type of the re_translate member by
 434    defining the macro RE_TRANSLATE_TYPE, which defaults to unsigned
 435    char *.  This pollutes the POSIX name space, so in POSIX mode just
 436    use unsigned char *.  */
 437 #ifdef __USE_GNU_REGEX
 438 # ifndef RE_TRANSLATE_TYPE
 439 #  define RE_TRANSLATE_TYPE unsigned char *
 440 # endif
 441 # define REG_TRANSLATE_TYPE RE_TRANSLATE_TYPE
 442 #else
 443 # define REG_TRANSLATE_TYPE unsigned char *
 444 #endif
 445
 446 /* This data structure represents a compiled pattern.  Before calling
 447    the pattern compiler, the fields `buffer', `allocated', `fastmap',
 448    `translate', and `no_sub' can be set.  After the pattern has been
 449    compiled, the `re_nsub' field is available.  All other fields are
 450    private to the regex routines.  */
 451
 452 struct re_pattern_buffer
 453 {
 454   /* Space that holds the compiled pattern.  It is declared as
 455      `unsigned char *' because its elements are sometimes used as
 456      array indexes.  */
 457   unsigned char *_REG_RE_NAME (buffer);
 458
 459   /* Number of bytes to which `buffer' points.  */
 460   __re_long_size_t _REG_RE_NAME (allocated);
 461
 462   /* Number of bytes actually used in `buffer'.  */
 463   __re_long_size_t _REG_RE_NAME (used);
 464
 465   /* Syntax setting with which the pattern was compiled.  */
 466   reg_syntax_t _REG_RE_NAME (syntax);
 467
 468   /* Pointer to a fastmap, if any, otherwise zero.  re_search uses the
 469      fastmap, if there is one, to skip over impossible starting points
 470      for matches.  */
 471   char *_REG_RE_NAME (fastmap);
 472
 473   /* Either a translate table to apply to all characters before
 474      comparing them, or zero for no translation.  The translation is
 475      applied to a pattern when it is compiled and to a string when it
 476      is matched.  */
 477   REG_TRANSLATE_TYPE _REG_RE_NAME (translate);
 478
 479   /* Number of subexpressions found by the compiler.  */
 480   size_t re_nsub;
 481
 482   /* Zero if this pattern cannot match the empty string, one else.
 483      Well, in truth it's used only in `re_search_2', to see whether or
 484      not we should use the fastmap, so we don't set this absolutely
 485      perfectly; see `re_compile_fastmap' (the `duplicate' case).  */
 486   unsigned int _REG_RE_NAME (can_be_null) : 1;
 487
 488   /* If REGS_UNALLOCATED, allocate space in the `regs' structure
 489      for `max (RE_NREGS, re_nsub + 1)' groups.
 490      If REGS_REALLOCATE, reallocate space if necessary.
 491      If REGS_FIXED, use what's there.  */
 492 #ifdef __USE_GNU_REGEX
 493 # define REGS_UNALLOCATED 0
 494 # define REGS_REALLOCATE 1
 495 # define REGS_FIXED 2
 496 #endif
 497   unsigned int _REG_RE_NAME (regs_allocated) : 2;
 498
 499   /* Set to zero when `re_compile_pattern' compiles a pattern; set to
 500      one by `re_compile_fastmap' if it updates the fastmap.  */
 501   unsigned int _REG_RE_NAME (fastmap_accurate) : 1;
 502
 503   /* If set, `re_match_2' does not return information about
 504      subexpressions.  */
 505   unsigned int _REG_RE_NAME (no_sub) : 1;
 506
 507   /* If set, a beginning-of-line anchor doesn't match at the beginning
 508      of the string.  */
 509   unsigned int _REG_RE_NAME (not_bol) : 1;
 510
 511   /* Similarly for an end-of-line anchor.  */
 512   unsigned int _REG_RE_NAME (not_eol) : 1;
 513
 514   /* If true, an anchor at a newline matches.  */
 515   unsigned int _REG_RE_NAME (newline_anchor) : 1;
 516
 517 /* [[[end pattern_buffer]]] */
 518 };
 519
 520 typedef struct re_pattern_buffer regex_t;
 521 \f
 522 /* This is the structure we store register match data in.  See
 523    regex.texinfo for a full description of what registers match.  */
 524 struct re_registers
 525 {
 526   __re_size_t _REG_RM_NAME (num_regs);
 527   regoff_t *_REG_RM_NAME (start);
 528   regoff_t *_REG_RM_NAME (end);
 529 };
 530
 531
 532 /* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
 533    `re_match_2' returns information about at least this many registers
 534    the first time a `regs' structure is passed.  */
 535 #if !defined RE_NREGS && defined __USE_GNU_REGEX
 536 # define RE_NREGS 30
 537 #endif
 538
 539
 540 /* POSIX specification for registers.  Aside from the different names than
 541    `re_registers', POSIX uses an array of structures, instead of a
 542    structure of arrays.  */
 543 typedef struct
 544 {
 545   regoff_t rm_so;  /* Byte offset from string's start to substring's start.  */
 546   regoff_t rm_eo;  /* Byte offset from string's start to substring's end.  */
 547 } regmatch_t;
 548 \f
 549 /* Declarations for routines.  */
 550
 551 /* Sets the current default syntax to SYNTAX, and return the old syntax.
 552    You can also simply assign to the `re_syntax_options' variable.  */
 553 extern reg_syntax_t re_set_syntax (reg_syntax_t __syntax);
 554
 555 /* Compile the regular expression PATTERN, with length LENGTH
 556    and syntax given by the global `re_syntax_options', into the buffer
 557    BUFFER.  Return NULL if successful, and an error string if not.  */
 558 extern const char *re_compile_pattern (const char *__pattern, size_t __length,
 559                                        struct re_pattern_buffer *__buffer);
 560
 561
 562 /* Compile a fastmap for the compiled pattern in BUFFER; used to
 563    accelerate searches.  Return 0 if successful and -2 if was an
 564    internal error.  */
 565 extern int re_compile_fastmap (struct re_pattern_buffer *__buffer);
 566
 567
 568 /* Search in the string STRING (with length LENGTH) for the pattern
 569    compiled into BUFFER.  Start searching at position START, for RANGE
 570    characters.  Return the starting position of the match, -1 for no
 571    match, or -2 for an internal error.  Also return register
 572    information in REGS (if REGS and BUFFER->no_sub are nonzero).  */
 573 extern regoff_t re_search (struct re_pattern_buffer *__buffer,
 574                            const char *__string, __re_idx_t __length,
 575                            __re_idx_t __start, regoff_t __range,
 576                            struct re_registers *__regs);
 577
 578
 579 /* Like `re_search', but search in the concatenation of STRING1 and
 580    STRING2.  Also, stop searching at index START + STOP.  */
 581 extern regoff_t re_search_2 (struct re_pattern_buffer *__buffer,
 582                              const char *__string1, __re_idx_t __length1,
 583                              const char *__string2, __re_idx_t __length2,
 584                              __re_idx_t __start, regoff_t __range,
 585                              struct re_registers *__regs,
 586                              __re_idx_t __stop);
 587
 588
 589 /* Like `re_search', but return how many characters in STRING the regexp
 590    in BUFFER matched, starting at position START.  */
 591 extern regoff_t re_match (struct re_pattern_buffer *__buffer,
 592                           const char *__string, __re_idx_t __length,
 593                           __re_idx_t __start, struct re_registers *__regs);
 594
 595
 596 /* Relates to `re_match' as `re_search_2' relates to `re_search'.  */
 597 extern regoff_t re_match_2 (struct re_pattern_buffer *__buffer,
 598                             const char *__string1, __re_idx_t __length1,
 599                             const char *__string2, __re_idx_t __length2,
 600                             __re_idx_t __start, struct re_registers *__regs,
 601                             __re_idx_t __stop);
 602
 603
 604 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
 605    ENDS.  Subsequent matches using BUFFER and REGS will use this memory
 606    for recording register information.  STARTS and ENDS must be
 607    allocated with malloc, and must each be at least `NUM_REGS * sizeof
 608    (regoff_t)' bytes long.
 609
 610    If NUM_REGS == 0, then subsequent matches should allocate their own
 611    register data.
 612
 613    Unless this function is called, the first search or match using
 614    BUFFER will allocate its own register data, without freeing the old
 615    data.  */
 616 extern void re_set_registers (struct re_pattern_buffer *__buffer,
 617                               struct re_registers *__regs,
 618                               __re_size_t __num_regs,
 619                               regoff_t *__starts, regoff_t *__ends);
 620
 621 #if defined _REGEX_RE_COMP || defined _LIBC
 622 # ifndef _CRAY
 623 /* 4.2 bsd compatibility.  */
 624 extern char *re_comp (const char *);
 625 extern int re_exec (const char *);
 626 # endif
 627 #endif
 628
 629 /* GCC 2.95 and later have "__restrict"; C99 compilers have
 630    "restrict", and "configure" may have defined "restrict".
 631    Other compilers use __restrict, __restrict__, and _Restrict, and
 632    'configure' might #define 'restrict' to those words, so pick a
 633    different name.  */
 634 #ifndef _Restrict_
 635 # if 199901L <= __STDC_VERSION__
 636 #  define _Restrict_ restrict
 637 # elif 2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__)
 638 #  define _Restrict_ __restrict
 639 # else
 640 #  define _Restrict_
 641 # endif
 642 #endif
 643 /* gcc 3.1 and up support the [restrict] syntax.  Don't trust
 644    sys/cdefs.h's definition of __restrict_arr, though, as it
 645    mishandles gcc -ansi -pedantic.  */
 646 #ifndef _Restrict_arr_
 647 # if ((199901L <= __STDC_VERSION__                                      \
 648        || ((3 < __GNUC__ || (3 == __GNUC__ && 1 <= __GNUC_MINOR__))     \
 649            && !__STRICT_ANSI__))                                        \
 650       && !defined __GNUG__)
 651 #  define _Restrict_arr_ _Restrict_
 652 # else
 653 #  define _Restrict_arr_
 654 # endif
 655 #endif
 656
 657 /* POSIX compatibility.  */
 658 extern int regcomp (regex_t *_Restrict_ __preg,
 659                     const char *_Restrict_ __pattern,
 660                     int __cflags);
 661
 662 extern int regexec (const regex_t *_Restrict_ __preg,
 663                     const char *_Restrict_ __string, size_t __nmatch,
 664                     regmatch_t __pmatch[_Restrict_arr_],
 665                     int __eflags);
 666
 667 extern size_t regerror (int __errcode, const regex_t *_Restrict_ __preg,
 668                         char *_Restrict_ __errbuf, size_t __errbuf_size);
 669
 670 extern void regfree (regex_t *__preg);
 671
 672
 673 #ifdef __cplusplus
 674 }
 675 #endif  /* C++ */
 676
 677 #endif /* regex.h */