posix/regex.h

   1 /* Definitions for data structures and routines for the regular
   2    expression library, version 0.12.
   3
   4    Copyright (C) 1985, 89, 90, 91, 92, 93, 95 Free Software Foundation, Inc.
   5
   6 This file is part of the GNU C Library.  Its master source is NOT part of
   7 the C library, however.  The master source lives in /gd/gnu/lib.
   8
   9 The GNU C Library is free software; you can redistribute it and/or
  10 modify it under the terms of the GNU Library General Public License as
  11 published by the Free Software Foundation; either version 2 of the
  12 License, or (at your option) any later version.
  13
  14 The GNU C Library 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 GNU
  17 Library General Public License for more details.
  18
  19 You should have received a copy of the GNU Library General Public
  20 License along with the GNU C Library; see the file COPYING.LIB.  If
  21 not, write to the Free Software Foundation, Inc., 675 Mass Ave,
  22 Cambridge, MA 02139, USA.  */
  23
  24 #ifndef __REGEXP_LIBRARY_H__
  25 #define __REGEXP_LIBRARY_H__
  26
  27 /* POSIX says that <sys/types.h> must be included (by the caller) before
  28    <regex.h>.  */
  29
  30 #if !defined (_POSIX_C_SOURCE) && !defined (_POSIX_SOURCE) && defined (VMS)
  31 /* VMS doesn't have `size_t' in <sys/types.h>, even though POSIX says it
  32    should be there.  */
  33 #include <stddef.h>
  34 #endif
  35
  36
  37 /* The following bits are used to determine the regexp syntax we
  38    recognize.  The set/not-set meanings are chosen so that Emacs syntax
  39    remains the value 0.  The bits are given in alphabetical order, and
  40    the definitions shifted by one from the previous bit; thus, when we
  41    add or remove a bit, only one other definition need change.  */
  42 typedef unsigned reg_syntax_t;
  43
  44 /* If this bit is not set, then \ inside a bracket expression is literal.
  45    If set, then such a \ quotes the following character.  */
  46 #define RE_BACKSLASH_ESCAPE_IN_LISTS (1)
  47
  48 /* If this bit is not set, then + and ? are operators, and \+ and \? are
  49      literals.
  50    If set, then \+ and \? are operators and + and ? are literals.  */
  51 #define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1)
  52
  53 /* If this bit is set, then character classes are supported.  They are:
  54      [:alpha:], [:upper:], [:lower:],  [:digit:], [:alnum:], [:xdigit:],
  55      [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:].
  56    If not set, then character classes are not supported.  */
  57 #define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1)
  58
  59 /* If this bit is set, then ^ and $ are always anchors (outside bracket
  60      expressions, of course).
  61    If this bit is not set, then it depends:
  62         ^  is an anchor if it is at the beginning of a regular
  63            expression or after an open-group or an alternation operator;
  64         $  is an anchor if it is at the end of a regular expression, or
  65            before a close-group or an alternation operator.
  66
  67    This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because
  68    POSIX draft 11.2 says that * etc. in leading positions is undefined.
  69    We already implemented a previous draft which made those constructs
  70    invalid, though, so we haven't changed the code back.  */
  71 #define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1)
  72
  73 /* If this bit is set, then special characters are always special
  74      regardless of where they are in the pattern.
  75    If this bit is not set, then special characters are special only in
  76      some contexts; otherwise they are ordinary.  Specifically,
  77      * + ? and intervals are only special when not after the beginning,
  78      open-group, or alternation operator.  */
  79 #define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1)
  80
  81 /* If this bit is set, then *, +, ?, and { cannot be first in an re or
  82      immediately after an alternation or begin-group operator.  */
  83 #define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1)
  84
  85 /* If this bit is set, then . matches newline.
  86    If not set, then it doesn't.  */
  87 #define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1)
  88
  89 /* If this bit is set, then . doesn't match NUL.
  90    If not set, then it does.  */
  91 #define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1)
  92
  93 /* If this bit is set, nonmatching lists [^...] do not match newline.
  94    If not set, they do.  */
  95 #define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1)
  96
  97 /* If this bit is set, either \{...\} or {...} defines an
  98      interval, depending on RE_NO_BK_BRACES.
  99    If not set, \{, \}, {, and } are literals.  */
 100 #define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1)
 101
 102 /* If this bit is set, +, ? and | aren't recognized as operators.
 103    If not set, they are.  */
 104 #define RE_LIMITED_OPS (RE_INTERVALS << 1)
 105
 106 /* If this bit is set, newline is an alternation operator.
 107    If not set, newline is literal.  */
 108 #define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1)
 109
 110 /* If this bit is set, then `{...}' defines an interval, and \{ and \}
 111      are literals.
 112   If not set, then `\{...\}' defines an interval.  */
 113 #define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1)
 114
 115 /* If this bit is set, (...) defines a group, and \( and \) are literals.
 116    If not set, \(...\) defines a group, and ( and ) are literals.  */
 117 #define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1)
 118
 119 /* If this bit is set, then \<digit> matches <digit>.
 120    If not set, then \<digit> is a back-reference.  */
 121 #define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1)
 122
 123 /* If this bit is set, then | is an alternation operator, and \| is literal.
 124    If not set, then \| is an alternation operator, and | is literal.  */
 125 #define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1)
 126
 127 /* If this bit is set, then an ending range point collating higher
 128      than the starting range point, as in [z-a], is invalid.
 129    If not set, then when ending range point collates higher than the
 130      starting range point, the range is ignored.  */
 131 #define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1)
 132
 133 /* If this bit is set, then an unmatched ) is ordinary.
 134    If not set, then an unmatched ) is invalid.  */
 135 #define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1)
 136
 137 /* If this bit is set, succeed as soon as we match the whole pattern,
 138    without further backtracking.  */
 139 #define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1)
 140
 141 /* This global variable defines the particular regexp syntax to use (for
 142    some interfaces).  When a regexp is compiled, the syntax used is
 143    stored in the pattern buffer, so changing this does not affect
 144    already-compiled regexps.  */
 145 extern reg_syntax_t re_syntax_options;
 146 \f
 147 /* Define combinations of the above bits for the standard possibilities.
 148    (The [[[ comments delimit what gets put into the Texinfo file, so
 149    don't delete them!)  */
 150 /* [[[begin syntaxes]]] */
 151 #define RE_SYNTAX_EMACS 0
 152
 153 #define RE_SYNTAX_AWK                                                   \
 154   (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL                       \
 155    | RE_NO_BK_PARENS            | RE_NO_BK_REFS                         \
 156    | RE_NO_BK_VBAR               | RE_NO_EMPTY_RANGES                   \
 157    | RE_UNMATCHED_RIGHT_PAREN_ORD)
 158
 159 #define RE_SYNTAX_POSIX_AWK                                             \
 160   (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS)
 161
 162 #define RE_SYNTAX_GREP                                                  \
 163   (RE_BK_PLUS_QM              | RE_CHAR_CLASSES                         \
 164    | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS                            \
 165    | RE_NEWLINE_ALT)
 166
 167 #define RE_SYNTAX_EGREP                                                 \
 168   (RE_CHAR_CLASSES        | RE_CONTEXT_INDEP_ANCHORS                    \
 169    | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE                    \
 170    | RE_NEWLINE_ALT       | RE_NO_BK_PARENS                             \
 171    | RE_NO_BK_VBAR)
 172
 173 #define RE_SYNTAX_POSIX_EGREP                                           \
 174   (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES)
 175
 176 /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff.  */
 177 #define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
 178
 179 #define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
 180
 181 /* Syntax bits common to both basic and extended POSIX regex syntax.  */
 182 #define _RE_SYNTAX_POSIX_COMMON                                         \
 183   (RE_CHAR_CLASSES | RE_DOT_NEWLINE      | RE_DOT_NOT_NULL              \
 184    | RE_INTERVALS  | RE_NO_EMPTY_RANGES)
 185
 186 #define RE_SYNTAX_POSIX_BASIC                                           \
 187   (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM)
 188
 189 /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
 190    RE_LIMITED_OPS, i.e., \? \+ \| are not recognized.  Actually, this
 191    isn't minimal, since other operators, such as \`, aren't disabled.  */
 192 #define RE_SYNTAX_POSIX_MINIMAL_BASIC                                   \
 193   (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
 194
 195 #define RE_SYNTAX_POSIX_EXTENDED                                        \
 196   (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS                   \
 197    | RE_CONTEXT_INDEP_OPS  | RE_NO_BK_BRACES                            \
 198    | RE_NO_BK_PARENS       | RE_NO_BK_VBAR                              \
 199    | RE_UNMATCHED_RIGHT_PAREN_ORD)
 200
 201 /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INVALID_OPS
 202    replaces RE_CONTEXT_INDEP_OPS and RE_NO_BK_REFS is added.  */
 203 #define RE_SYNTAX_POSIX_MINIMAL_EXTENDED                                \
 204   (_RE_SYNTAX_POSIX_COMMON  | RE_CONTEXT_INDEP_ANCHORS                  \
 205    | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES                           \
 206    | RE_NO_BK_PARENS        | RE_NO_BK_REFS                             \
 207    | RE_NO_BK_VBAR          | RE_UNMATCHED_RIGHT_PAREN_ORD)
 208 /* [[[end syntaxes]]] */
 209 \f
 210 /* Maximum number of duplicates an interval can allow.  Some systems
 211    (erroneously) define this in other header files, but we want our
 212    value, so remove any previous define.  */
 213 #ifdef RE_DUP_MAX
 214 #undef RE_DUP_MAX
 215 #endif
 216 #define RE_DUP_MAX ((1 << 15) - 1)
 217
 218
 219 /* POSIX `cflags' bits (i.e., information for `regcomp').  */
 220
 221 /* If this bit is set, then use extended regular expression syntax.
 222    If not set, then use basic regular expression syntax.  */
 223 #define REG_EXTENDED 1
 224
 225 /* If this bit is set, then ignore case when matching.
 226    If not set, then case is significant.  */
 227 #define REG_ICASE (REG_EXTENDED << 1)
 228
 229 /* If this bit is set, then anchors do not match at newline
 230      characters in the string.
 231    If not set, then anchors do match at newlines.  */
 232 #define REG_NEWLINE (REG_ICASE << 1)
 233
 234 /* If this bit is set, then report only success or fail in regexec.
 235    If not set, then returns differ between not matching and errors.  */
 236 #define REG_NOSUB (REG_NEWLINE << 1)
 237
 238
 239 /* POSIX `eflags' bits (i.e., information for regexec).  */
 240
 241 /* If this bit is set, then the beginning-of-line operator doesn't match
 242      the beginning of the string (presumably because it's not the
 243      beginning of a line).
 244    If not set, then the beginning-of-line operator does match the
 245      beginning of the string.  */
 246 #define REG_NOTBOL 1
 247
 248 /* Like REG_NOTBOL, except for the end-of-line.  */
 249 #define REG_NOTEOL (1 << 1)
 250
 251
 252 /* If any error codes are removed, changed, or added, update the
 253    `re_error_msg' table in regex.c.  */
 254 typedef enum
 255 {
 256   REG_NOERROR = 0,      /* Success.  */
 257   REG_NOMATCH,          /* Didn't find a match (for regexec).  */
 258
 259   /* POSIX regcomp return error codes.  (In the order listed in the
 260      standard.)  */
 261   REG_BADPAT,           /* Invalid pattern.  */
 262   REG_ECOLLATE,         /* Not implemented.  */
 263   REG_ECTYPE,           /* Invalid character class name.  */
 264   REG_EESCAPE,          /* Trailing backslash.  */
 265   REG_ESUBREG,          /* Invalid back reference.  */
 266   REG_EBRACK,           /* Unmatched left bracket.  */
 267   REG_EPAREN,           /* Parenthesis imbalance.  */
 268   REG_EBRACE,           /* Unmatched \{.  */
 269   REG_BADBR,            /* Invalid contents of \{\}.  */
 270   REG_ERANGE,           /* Invalid range end.  */
 271   REG_ESPACE,           /* Ran out of memory.  */
 272   REG_BADRPT,           /* No preceding re for repetition op.  */
 273
 274   /* Error codes we've added.  */
 275   REG_EEND,             /* Premature end.  */
 276   REG_ESIZE,            /* Compiled pattern bigger than 2^16 bytes.  */
 277   REG_ERPAREN           /* Unmatched ) or \); not returned from regcomp.  */
 278 } reg_errcode_t;
 279 \f
 280 /* This data structure represents a compiled pattern.  Before calling
 281    the pattern compiler, the fields `buffer', `allocated', `fastmap',
 282    `translate', and `no_sub' can be set.  After the pattern has been
 283    compiled, the `re_nsub' field is available.  All other fields are
 284    private to the regex routines.  */
 285
 286 struct re_pattern_buffer
 287 {
 288 /* [[[begin pattern_buffer]]] */
 289         /* Space that holds the compiled pattern.  It is declared as
 290           `unsigned char *' because its elements are
 291            sometimes used as array indexes.  */
 292   unsigned char *buffer;
 293
 294         /* Number of bytes to which `buffer' points.  */
 295   unsigned long allocated;
 296
 297         /* Number of bytes actually used in `buffer'.  */
 298   unsigned long used;
 299
 300         /* Syntax setting with which the pattern was compiled.  */
 301   reg_syntax_t syntax;
 302
 303         /* Pointer to a fastmap, if any, otherwise zero.  re_search uses
 304            the fastmap, if there is one, to skip over impossible
 305            starting points for matches.  */
 306   char *fastmap;
 307
 308         /* Either a translate table to apply to all characters before
 309            comparing them, or zero for no translation.  The translation
 310            is applied to a pattern when it is compiled and to a string
 311            when it is matched.  */
 312   char *translate;
 313
 314         /* Number of subexpressions found by the compiler.  */
 315   size_t re_nsub;
 316
 317         /* Zero if this pattern cannot match the empty string, one else.
 318            Well, in truth it's used only in `re_search_2', to see
 319            whether or not we should use the fastmap, so we don't set
 320            this absolutely perfectly; see `re_compile_fastmap' (the
 321            `duplicate' case).  */
 322   unsigned can_be_null : 1;
 323
 324         /* If REGS_UNALLOCATED, allocate space in the `regs' structure
 325              for `max (RE_NREGS, re_nsub + 1)' groups.
 326            If REGS_REALLOCATE, reallocate space if necessary.
 327            If REGS_FIXED, use what's there.  */
 328 #define REGS_UNALLOCATED 0
 329 #define REGS_REALLOCATE 1
 330 #define REGS_FIXED 2
 331   unsigned regs_allocated : 2;
 332
 333         /* Set to zero when `regex_compile' compiles a pattern; set to one
 334            by `re_compile_fastmap' if it updates the fastmap.  */
 335   unsigned fastmap_accurate : 1;
 336
 337         /* If set, `re_match_2' does not return information about
 338            subexpressions.  */
 339   unsigned no_sub : 1;
 340
 341         /* If set, a beginning-of-line anchor doesn't match at the
 342            beginning of the string.  */
 343   unsigned not_bol : 1;
 344
 345         /* Similarly for an end-of-line anchor.  */
 346   unsigned not_eol : 1;
 347
 348         /* If true, an anchor at a newline matches.  */
 349   unsigned newline_anchor : 1;
 350
 351 /* [[[end pattern_buffer]]] */
 352 };
 353
 354 typedef struct re_pattern_buffer regex_t;
 355 \f
 356 /* Type for byte offsets within the string.  POSIX mandates this.  */
 357 typedef int regoff_t;
 358
 359
 360 /* This is the structure we store register match data in.  See
 361    regex.texinfo for a full description of what registers match.  */
 362 struct re_registers
 363 {
 364   unsigned num_regs;
 365   regoff_t *start;
 366   regoff_t *end;
 367 };
 368
 369
 370 /* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
 371    `re_match_2' returns information about at least this many registers
 372    the first time a `regs' structure is passed.  */
 373 #ifndef RE_NREGS
 374 #define RE_NREGS 30
 375 #endif
 376
 377
 378 /* POSIX specification for registers.  Aside from the different names than
 379    `re_registers', POSIX uses an array of structures, instead of a
 380    structure of arrays.  */
 381 typedef struct
 382 {
 383   regoff_t rm_so;  /* Byte offset from string's start to substring's start.  */
 384   regoff_t rm_eo;  /* Byte offset from string's start to substring's end.  */
 385 } regmatch_t;
 386 \f
 387 /* Declarations for routines.  */
 388
 389 /* To avoid duplicating every routine declaration -- once with a
 390    prototype (if we are ANSI), and once without (if we aren't) -- we
 391    use the following macro to declare argument types.  This
 392    unfortunately clutters up the declarations a bit, but I think it's
 393    worth it.  */
 394
 395 #if __STDC__
 396
 397 #define _RE_ARGS(args) args
 398
 399 #else /* not __STDC__ */
 400
 401 #define _RE_ARGS(args) ()
 402
 403 #endif /* not __STDC__ */
 404
 405 /* Sets the current default syntax to SYNTAX, and return the old syntax.
 406    You can also simply assign to the `re_syntax_options' variable.  */
 407 extern reg_syntax_t re_set_syntax _RE_ARGS ((reg_syntax_t syntax));
 408
 409 /* Compile the regular expression PATTERN, with length LENGTH
 410    and syntax given by the global `re_syntax_options', into the buffer
 411    BUFFER.  Return NULL if successful, and an error string if not.  */
 412 extern const char *re_compile_pattern
 413   _RE_ARGS ((const char *pattern, int length,
 414              struct re_pattern_buffer *buffer));
 415
 416
 417 /* Compile a fastmap for the compiled pattern in BUFFER; used to
 418    accelerate searches.  Return 0 if successful and -2 if was an
 419    internal error.  */
 420 extern int re_compile_fastmap _RE_ARGS ((struct re_pattern_buffer *buffer));
 421
 422
 423 /* Search in the string STRING (with length LENGTH) for the pattern
 424    compiled into BUFFER.  Start searching at position START, for RANGE
 425    characters.  Return the starting position of the match, -1 for no
 426    match, or -2 for an internal error.  Also return register
 427    information in REGS (if REGS and BUFFER->no_sub are nonzero).  */
 428 extern int re_search
 429   _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string,
 430             int length, int start, int range, struct re_registers *regs));
 431
 432
 433 /* Like `re_search', but search in the concatenation of STRING1 and
 434    STRING2.  Also, stop searching at index START + STOP.  */
 435 extern int re_search_2
 436   _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1,
 437              int length1, const char *string2, int length2,
 438              int start, int range, struct re_registers *regs, int stop));
 439
 440
 441 /* Like `re_search', but return how many characters in STRING the regexp
 442    in BUFFER matched, starting at position START.  */
 443 extern int re_match
 444   _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string,
 445              int length, int start, struct re_registers *regs));
 446
 447
 448 /* Relates to `re_match' as `re_search_2' relates to `re_search'.  */
 449 extern int re_match_2
 450   _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1,
 451              int length1, const char *string2, int length2,
 452              int start, struct re_registers *regs, int stop));
 453
 454
 455 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
 456    ENDS.  Subsequent matches using BUFFER and REGS will use this memory
 457    for recording register information.  STARTS and ENDS must be
 458    allocated with malloc, and must each be at least `NUM_REGS * sizeof
 459    (regoff_t)' bytes long.
 460
 461    If NUM_REGS == 0, then subsequent matches should allocate their own
 462    register data.
 463
 464    Unless this function is called, the first search or match using
 465    PATTERN_BUFFER will allocate its own register data, without
 466    freeing the old data.  */
 467 extern void re_set_registers
 468   _RE_ARGS ((struct re_pattern_buffer *buffer, struct re_registers *regs,
 469              unsigned num_regs, regoff_t *starts, regoff_t *ends));
 470
 471 #ifdef _REGEX_RE_COMP
 472 /* 4.2 bsd compatibility.  */
 473 extern char *re_comp _RE_ARGS ((const char *));
 474 extern int re_exec _RE_ARGS ((const char *));
 475 #endif
 476
 477 /* POSIX compatibility.  */
 478 extern int regcomp _RE_ARGS ((regex_t *preg, const char *pattern, int cflags));
 479 extern int regexec
 480   _RE_ARGS ((const regex_t *preg, const char *string, size_t nmatch,
 481              regmatch_t pmatch[], int eflags));
 482 extern size_t regerror
 483   _RE_ARGS ((int errcode, const regex_t *preg, char *errbuf,
 484              size_t errbuf_size));
 485 extern void regfree _RE_ARGS ((regex_t *preg));
 486
 487 #endif /* not __REGEXP_LIBRARY_H__ */
 488 \f
 489 /*
 490 Local variables:
 491 make-backup-files: t
 492 version-control: t
 493 trim-versions-without-asking: nil
 494 End:
 495 */