regex.h 24 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658
  1. /* Definitions for data structures and routines for the regular
  2. expression library.
  3. Copyright (C) 1985, 1989-2020 Free Software Foundation, Inc.
  4. This file is part of the GNU C Library.
  5. The GNU C Library is free software; you can redistribute it and/or
  6. modify it under the terms of the GNU Lesser General Public
  7. License as published by the Free Software Foundation; either
  8. version 2.1 of the License, or (at your option) any later version.
  9. The GNU C Library is distributed in the hope that it will be useful,
  10. but WITHOUT ANY WARRANTY; without even the implied warranty of
  11. MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  12. Lesser General Public License for more details.
  13. You should have received a copy of the GNU Lesser General Public
  14. License along with the GNU C Library; if not, see
  15. <https://www.gnu.org/licenses/>. */
  16. #ifndef _REGEX_H
  17. #define _REGEX_H 1
  18. #include <sys/types.h>
  19. /* Allow the use in C++ code. */
  20. #ifdef __cplusplus
  21. extern "C" {
  22. #endif
  23. /* Define __USE_GNU to declare GNU extensions that violate the
  24. POSIX name space rules. */
  25. #ifdef _GNU_SOURCE
  26. # define __USE_GNU 1
  27. #endif
  28. #ifdef _REGEX_LARGE_OFFSETS
  29. /* Use types and values that are wide enough to represent signed and
  30. unsigned byte offsets in memory. This currently works only when
  31. the regex code is used outside of the GNU C library; it is not yet
  32. supported within glibc itself, and glibc users should not define
  33. _REGEX_LARGE_OFFSETS. */
  34. /* The type of object sizes. */
  35. typedef size_t __re_size_t;
  36. /* The type of object sizes, in places where the traditional code
  37. uses unsigned long int. */
  38. typedef size_t __re_long_size_t;
  39. #else
  40. /* The traditional GNU regex implementation mishandles strings longer
  41. than INT_MAX. */
  42. typedef unsigned int __re_size_t;
  43. typedef unsigned long int __re_long_size_t;
  44. #endif
  45. /* The following two types have to be signed and unsigned integer type
  46. wide enough to hold a value of a pointer. For most ANSI compilers
  47. ptrdiff_t and size_t should be likely OK. Still size of these two
  48. types is 2 for Microsoft C. Ugh... */
  49. typedef long int s_reg_t;
  50. typedef unsigned long int active_reg_t;
  51. /* The following bits are used to determine the regexp syntax we
  52. recognize. The set/not-set meanings are chosen so that Emacs syntax
  53. remains the value 0. The bits are given in alphabetical order, and
  54. the definitions shifted by one from the previous bit; thus, when we
  55. add or remove a bit, only one other definition need change. */
  56. typedef unsigned long int reg_syntax_t;
  57. #ifdef __USE_GNU
  58. /* If this bit is not set, then \ inside a bracket expression is literal.
  59. If set, then such a \ quotes the following character. */
  60. # define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1)
  61. /* If this bit is not set, then + and ? are operators, and \+ and \? are
  62. literals.
  63. If set, then \+ and \? are operators and + and ? are literals. */
  64. # define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1)
  65. /* If this bit is set, then character classes are supported. They are:
  66. [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:],
  67. [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:].
  68. If not set, then character classes are not supported. */
  69. # define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1)
  70. /* If this bit is set, then ^ and $ are always anchors (outside bracket
  71. expressions, of course).
  72. If this bit is not set, then it depends:
  73. ^ is an anchor if it is at the beginning of a regular
  74. expression or after an open-group or an alternation operator;
  75. $ is an anchor if it is at the end of a regular expression, or
  76. before a close-group or an alternation operator.
  77. This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because
  78. POSIX draft 11.2 says that * etc. in leading positions is undefined.
  79. We already implemented a previous draft which made those constructs
  80. invalid, though, so we haven't changed the code back. */
  81. # define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1)
  82. /* If this bit is set, then special characters are always special
  83. regardless of where they are in the pattern.
  84. If this bit is not set, then special characters are special only in
  85. some contexts; otherwise they are ordinary. Specifically,
  86. * + ? and intervals are only special when not after the beginning,
  87. open-group, or alternation operator. */
  88. # define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1)
  89. /* If this bit is set, then *, +, ?, and { cannot be first in an re or
  90. immediately after an alternation or begin-group operator. */
  91. # define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1)
  92. /* If this bit is set, then . matches newline.
  93. If not set, then it doesn't. */
  94. # define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1)
  95. /* If this bit is set, then . doesn't match NUL.
  96. If not set, then it does. */
  97. # define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1)
  98. /* If this bit is set, nonmatching lists [^...] do not match newline.
  99. If not set, they do. */
  100. # define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1)
  101. /* If this bit is set, either \{...\} or {...} defines an
  102. interval, depending on RE_NO_BK_BRACES.
  103. If not set, \{, \}, {, and } are literals. */
  104. # define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1)
  105. /* If this bit is set, +, ? and | aren't recognized as operators.
  106. If not set, they are. */
  107. # define RE_LIMITED_OPS (RE_INTERVALS << 1)
  108. /* If this bit is set, newline is an alternation operator.
  109. If not set, newline is literal. */
  110. # define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1)
  111. /* If this bit is set, then '{...}' defines an interval, and \{ and \}
  112. are literals.
  113. If not set, then '\{...\}' defines an interval. */
  114. # define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1)
  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. /* If this bit is set, then \<digit> matches <digit>.
  119. If not set, then \<digit> is a back-reference. */
  120. # define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1)
  121. /* If this bit is set, then | is an alternation operator, and \| is literal.
  122. If not set, then \| is an alternation operator, and | is literal. */
  123. # define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1)
  124. /* If this bit is set, then an ending range point collating higher
  125. than the starting range point, as in [z-a], is invalid.
  126. If not set, then when ending range point collates higher than the
  127. starting range point, the range is ignored. */
  128. # define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1)
  129. /* If this bit is set, then an unmatched ) is ordinary.
  130. If not set, then an unmatched ) is invalid. */
  131. # define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1)
  132. /* If this bit is set, succeed as soon as we match the whole pattern,
  133. without further backtracking. */
  134. # define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1)
  135. /* If this bit is set, do not process the GNU regex operators.
  136. If not set, then the GNU regex operators are recognized. */
  137. # define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1)
  138. /* If this bit is set, turn on internal regex debugging.
  139. If not set, and debugging was on, turn it off.
  140. This only works if regex.c is compiled -DDEBUG.
  141. We define this bit always, so that all that's needed to turn on
  142. debugging is to recompile regex.c; the calling code can always have
  143. this bit set, and it won't affect anything in the normal case. */
  144. # define RE_DEBUG (RE_NO_GNU_OPS << 1)
  145. /* If this bit is set, a syntactically invalid interval is treated as
  146. a string of ordinary characters. For example, the ERE 'a{1' is
  147. treated as 'a\{1'. */
  148. # define RE_INVALID_INTERVAL_ORD (RE_DEBUG << 1)
  149. /* If this bit is set, then ignore case when matching.
  150. If not set, then case is significant. */
  151. # define RE_ICASE (RE_INVALID_INTERVAL_ORD << 1)
  152. /* This bit is used internally like RE_CONTEXT_INDEP_ANCHORS but only
  153. for ^, because it is difficult to scan the regex backwards to find
  154. whether ^ should be special. */
  155. # define RE_CARET_ANCHORS_HERE (RE_ICASE << 1)
  156. /* If this bit is set, then \{ cannot be first in a regex or
  157. immediately after an alternation, open-group or \} operator. */
  158. # define RE_CONTEXT_INVALID_DUP (RE_CARET_ANCHORS_HERE << 1)
  159. /* If this bit is set, then no_sub will be set to 1 during
  160. re_compile_pattern. */
  161. # define RE_NO_SUB (RE_CONTEXT_INVALID_DUP << 1)
  162. #endif
  163. /* This global variable defines the particular regexp syntax to use (for
  164. some interfaces). When a regexp is compiled, the syntax used is
  165. stored in the pattern buffer, so changing this does not affect
  166. already-compiled regexps. */
  167. extern reg_syntax_t re_syntax_options;
  168. #ifdef __USE_GNU
  169. /* Define combinations of the above bits for the standard possibilities.
  170. (The [[[ comments delimit what gets put into the Texinfo file, so
  171. don't delete them!) */
  172. /* [[[begin syntaxes]]] */
  173. # define RE_SYNTAX_EMACS 0
  174. # define RE_SYNTAX_AWK \
  175. (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \
  176. | RE_NO_BK_PARENS | RE_NO_BK_REFS \
  177. | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \
  178. | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \
  179. | RE_CHAR_CLASSES \
  180. | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS)
  181. # define RE_SYNTAX_GNU_AWK \
  182. ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
  183. | RE_INVALID_INTERVAL_ORD) \
  184. & ~(RE_DOT_NOT_NULL | RE_CONTEXT_INDEP_OPS \
  185. | RE_CONTEXT_INVALID_OPS ))
  186. # define RE_SYNTAX_POSIX_AWK \
  187. (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \
  188. | RE_INTERVALS | RE_NO_GNU_OPS \
  189. | RE_INVALID_INTERVAL_ORD)
  190. # define RE_SYNTAX_GREP \
  191. ((RE_SYNTAX_POSIX_BASIC | RE_NEWLINE_ALT) \
  192. & ~(RE_CONTEXT_INVALID_DUP | RE_DOT_NOT_NULL))
  193. # define RE_SYNTAX_EGREP \
  194. ((RE_SYNTAX_POSIX_EXTENDED | RE_INVALID_INTERVAL_ORD | RE_NEWLINE_ALT) \
  195. & ~(RE_CONTEXT_INVALID_OPS | RE_DOT_NOT_NULL))
  196. /* POSIX grep -E behavior is no longer incompatible with GNU. */
  197. # define RE_SYNTAX_POSIX_EGREP \
  198. RE_SYNTAX_EGREP
  199. /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */
  200. # define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC
  201. # define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC
  202. /* Syntax bits common to both basic and extended POSIX regex syntax. */
  203. # define _RE_SYNTAX_POSIX_COMMON \
  204. (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \
  205. | RE_INTERVALS | RE_NO_EMPTY_RANGES)
  206. # define RE_SYNTAX_POSIX_BASIC \
  207. (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM | RE_CONTEXT_INVALID_DUP)
  208. /* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes
  209. RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this
  210. isn't minimal, since other operators, such as \`, aren't disabled. */
  211. # define RE_SYNTAX_POSIX_MINIMAL_BASIC \
  212. (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS)
  213. # define RE_SYNTAX_POSIX_EXTENDED \
  214. (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
  215. | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \
  216. | RE_NO_BK_PARENS | RE_NO_BK_VBAR \
  217. | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD)
  218. /* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is
  219. removed and RE_NO_BK_REFS is added. */
  220. # define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \
  221. (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \
  222. | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \
  223. | RE_NO_BK_PARENS | RE_NO_BK_REFS \
  224. | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD)
  225. /* [[[end syntaxes]]] */
  226. /* Maximum number of duplicates an interval can allow. POSIX-conforming
  227. systems might define this in <limits.h>, but we want our
  228. value, so remove any previous define. */
  229. # ifdef _REGEX_INCLUDE_LIMITS_H
  230. # include <limits.h>
  231. # endif
  232. # ifdef RE_DUP_MAX
  233. # undef RE_DUP_MAX
  234. # endif
  235. /* RE_DUP_MAX is 2**15 - 1 because an earlier implementation stored
  236. the counter as a 2-byte signed integer. This is no longer true, so
  237. RE_DUP_MAX could be increased to (INT_MAX / 10 - 1), or to
  238. ((SIZE_MAX - 9) / 10) if _REGEX_LARGE_OFFSETS is defined.
  239. However, there would be a huge performance problem if someone
  240. actually used a pattern like a\{214748363\}, so RE_DUP_MAX retains
  241. its historical value. */
  242. # define RE_DUP_MAX (0x7fff)
  243. #endif
  244. /* POSIX 'cflags' bits (i.e., information for 'regcomp'). */
  245. /* If this bit is set, then use extended regular expression syntax.
  246. If not set, then use basic regular expression syntax. */
  247. #define REG_EXTENDED 1
  248. /* If this bit is set, then ignore case when matching.
  249. If not set, then case is significant. */
  250. #define REG_ICASE (1 << 1)
  251. /* If this bit is set, then anchors do not match at newline
  252. characters in the string.
  253. If not set, then anchors do match at newlines. */
  254. #define REG_NEWLINE (1 << 2)
  255. /* If this bit is set, then report only success or fail in regexec.
  256. If not set, then returns differ between not matching and errors. */
  257. #define REG_NOSUB (1 << 3)
  258. /* POSIX 'eflags' bits (i.e., information for regexec). */
  259. /* If this bit is set, then the beginning-of-line operator doesn't match
  260. the beginning of the string (presumably because it's not the
  261. beginning of a line).
  262. If not set, then the beginning-of-line operator does match the
  263. beginning of the string. */
  264. #define REG_NOTBOL 1
  265. /* Like REG_NOTBOL, except for the end-of-line. */
  266. #define REG_NOTEOL (1 << 1)
  267. /* Use PMATCH[0] to delimit the start and end of the search in the
  268. buffer. */
  269. #define REG_STARTEND (1 << 2)
  270. /* If any error codes are removed, changed, or added, update the
  271. '__re_error_msgid' table in regcomp.c. */
  272. typedef enum
  273. {
  274. _REG_ENOSYS = -1, /* This will never happen for this implementation. */
  275. _REG_NOERROR = 0, /* Success. */
  276. _REG_NOMATCH, /* Didn't find a match (for regexec). */
  277. /* POSIX regcomp return error codes. (In the order listed in the
  278. standard.) */
  279. _REG_BADPAT, /* Invalid pattern. */
  280. _REG_ECOLLATE, /* Invalid collating element. */
  281. _REG_ECTYPE, /* Invalid character class name. */
  282. _REG_EESCAPE, /* Trailing backslash. */
  283. _REG_ESUBREG, /* Invalid back reference. */
  284. _REG_EBRACK, /* Unmatched left bracket. */
  285. _REG_EPAREN, /* Parenthesis imbalance. */
  286. _REG_EBRACE, /* Unmatched \{. */
  287. _REG_BADBR, /* Invalid contents of \{\}. */
  288. _REG_ERANGE, /* Invalid range end. */
  289. _REG_ESPACE, /* Ran out of memory. */
  290. _REG_BADRPT, /* No preceding re for repetition op. */
  291. /* Error codes we've added. */
  292. _REG_EEND, /* Premature end. */
  293. _REG_ESIZE, /* Too large (e.g., repeat count too large). */
  294. _REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */
  295. } reg_errcode_t;
  296. #if defined _XOPEN_SOURCE || defined __USE_XOPEN2K
  297. # define REG_ENOSYS _REG_ENOSYS
  298. #endif
  299. #define REG_NOERROR _REG_NOERROR
  300. #define REG_NOMATCH _REG_NOMATCH
  301. #define REG_BADPAT _REG_BADPAT
  302. #define REG_ECOLLATE _REG_ECOLLATE
  303. #define REG_ECTYPE _REG_ECTYPE
  304. #define REG_EESCAPE _REG_EESCAPE
  305. #define REG_ESUBREG _REG_ESUBREG
  306. #define REG_EBRACK _REG_EBRACK
  307. #define REG_EPAREN _REG_EPAREN
  308. #define REG_EBRACE _REG_EBRACE
  309. #define REG_BADBR _REG_BADBR
  310. #define REG_ERANGE _REG_ERANGE
  311. #define REG_ESPACE _REG_ESPACE
  312. #define REG_BADRPT _REG_BADRPT
  313. #define REG_EEND _REG_EEND
  314. #define REG_ESIZE _REG_ESIZE
  315. #define REG_ERPAREN _REG_ERPAREN
  316. /* This data structure represents a compiled pattern. Before calling
  317. the pattern compiler, the fields 'buffer', 'allocated', 'fastmap',
  318. and 'translate' can be set. After the pattern has been compiled,
  319. the fields 're_nsub', 'not_bol' and 'not_eol' are available. All
  320. other fields are private to the regex routines. */
  321. #ifndef RE_TRANSLATE_TYPE
  322. # define __RE_TRANSLATE_TYPE unsigned char *
  323. # ifdef __USE_GNU
  324. # define RE_TRANSLATE_TYPE __RE_TRANSLATE_TYPE
  325. # endif
  326. #endif
  327. #ifdef __USE_GNU
  328. # define __REPB_PREFIX(name) name
  329. #else
  330. # define __REPB_PREFIX(name) __##name
  331. #endif
  332. struct re_pattern_buffer
  333. {
  334. /* Space that holds the compiled pattern. The type
  335. 'struct re_dfa_t' is private and is not declared here. */
  336. struct re_dfa_t *__REPB_PREFIX(buffer);
  337. /* Number of bytes to which 'buffer' points. */
  338. __re_long_size_t __REPB_PREFIX(allocated);
  339. /* Number of bytes actually used in 'buffer'. */
  340. __re_long_size_t __REPB_PREFIX(used);
  341. /* Syntax setting with which the pattern was compiled. */
  342. reg_syntax_t __REPB_PREFIX(syntax);
  343. /* Pointer to a fastmap, if any, otherwise zero. re_search uses the
  344. fastmap, if there is one, to skip over impossible starting points
  345. for matches. */
  346. char *__REPB_PREFIX(fastmap);
  347. /* Either a translate table to apply to all characters before
  348. comparing them, or zero for no translation. The translation is
  349. applied to a pattern when it is compiled and to a string when it
  350. is matched. */
  351. __RE_TRANSLATE_TYPE __REPB_PREFIX(translate);
  352. /* Number of subexpressions found by the compiler. */
  353. size_t re_nsub;
  354. /* Zero if this pattern cannot match the empty string, one else.
  355. Well, in truth it's used only in 're_search_2', to see whether or
  356. not we should use the fastmap, so we don't set this absolutely
  357. perfectly; see 're_compile_fastmap' (the "duplicate" case). */
  358. unsigned __REPB_PREFIX(can_be_null) : 1;
  359. /* If REGS_UNALLOCATED, allocate space in the 'regs' structure
  360. for 'max (RE_NREGS, re_nsub + 1)' groups.
  361. If REGS_REALLOCATE, reallocate space if necessary.
  362. If REGS_FIXED, use what's there. */
  363. #ifdef __USE_GNU
  364. # define REGS_UNALLOCATED 0
  365. # define REGS_REALLOCATE 1
  366. # define REGS_FIXED 2
  367. #endif
  368. unsigned __REPB_PREFIX(regs_allocated) : 2;
  369. /* Set to zero when 're_compile_pattern' compiles a pattern; set to
  370. one by 're_compile_fastmap' if it updates the fastmap. */
  371. unsigned __REPB_PREFIX(fastmap_accurate) : 1;
  372. /* If set, 're_match_2' does not return information about
  373. subexpressions. */
  374. unsigned __REPB_PREFIX(no_sub) : 1;
  375. /* If set, a beginning-of-line anchor doesn't match at the beginning
  376. of the string. */
  377. unsigned __REPB_PREFIX(not_bol) : 1;
  378. /* Similarly for an end-of-line anchor. */
  379. unsigned __REPB_PREFIX(not_eol) : 1;
  380. /* If true, an anchor at a newline matches. */
  381. unsigned __REPB_PREFIX(newline_anchor) : 1;
  382. };
  383. typedef struct re_pattern_buffer regex_t;
  384. /* Type for byte offsets within the string. POSIX mandates this. */
  385. #ifdef _REGEX_LARGE_OFFSETS
  386. /* POSIX 1003.1-2008 requires that regoff_t be at least as wide as
  387. ptrdiff_t and ssize_t. We don't know of any hosts where ptrdiff_t
  388. is wider than ssize_t, so ssize_t is safe. ptrdiff_t is not
  389. visible here, so use ssize_t. */
  390. typedef ssize_t regoff_t;
  391. #else
  392. /* The traditional GNU regex implementation mishandles strings longer
  393. than INT_MAX. */
  394. typedef int regoff_t;
  395. #endif
  396. #ifdef __USE_GNU
  397. /* This is the structure we store register match data in. See
  398. regex.texinfo for a full description of what registers match. */
  399. struct re_registers
  400. {
  401. __re_size_t num_regs;
  402. regoff_t *start;
  403. regoff_t *end;
  404. };
  405. /* If 'regs_allocated' is REGS_UNALLOCATED in the pattern buffer,
  406. 're_match_2' returns information about at least this many registers
  407. the first time a 'regs' structure is passed. */
  408. # ifndef RE_NREGS
  409. # define RE_NREGS 30
  410. # endif
  411. #endif
  412. /* POSIX specification for registers. Aside from the different names than
  413. 're_registers', POSIX uses an array of structures, instead of a
  414. structure of arrays. */
  415. typedef struct
  416. {
  417. regoff_t rm_so; /* Byte offset from string's start to substring's start. */
  418. regoff_t rm_eo; /* Byte offset from string's start to substring's end. */
  419. } regmatch_t;
  420. /* Declarations for routines. */
  421. #ifdef __USE_GNU
  422. /* Sets the current default syntax to SYNTAX, and return the old syntax.
  423. You can also simply assign to the 're_syntax_options' variable. */
  424. extern reg_syntax_t re_set_syntax (reg_syntax_t __syntax);
  425. /* Compile the regular expression PATTERN, with length LENGTH
  426. and syntax given by the global 're_syntax_options', into the buffer
  427. BUFFER. Return NULL if successful, and an error string if not.
  428. To free the allocated storage, you must call 'regfree' on BUFFER.
  429. Note that the translate table must either have been initialized by
  430. 'regcomp', with a malloc'ed value, or set to NULL before calling
  431. 'regfree'. */
  432. extern const char *re_compile_pattern (const char *__pattern, size_t __length,
  433. struct re_pattern_buffer *__buffer);
  434. /* Compile a fastmap for the compiled pattern in BUFFER; used to
  435. accelerate searches. Return 0 if successful and -2 if was an
  436. internal error. */
  437. extern int re_compile_fastmap (struct re_pattern_buffer *__buffer);
  438. /* Search in the string STRING (with length LENGTH) for the pattern
  439. compiled into BUFFER. Start searching at position START, for RANGE
  440. characters. Return the starting position of the match, -1 for no
  441. match, or -2 for an internal error. Also return register
  442. information in REGS (if REGS and BUFFER->no_sub are nonzero). */
  443. extern regoff_t re_search (struct re_pattern_buffer *__buffer,
  444. const char *__String, regoff_t __length,
  445. regoff_t __start, regoff_t __range,
  446. struct re_registers *__regs);
  447. /* Like 're_search', but search in the concatenation of STRING1 and
  448. STRING2. Also, stop searching at index START + STOP. */
  449. extern regoff_t re_search_2 (struct re_pattern_buffer *__buffer,
  450. const char *__string1, regoff_t __length1,
  451. const char *__string2, regoff_t __length2,
  452. regoff_t __start, regoff_t __range,
  453. struct re_registers *__regs,
  454. regoff_t __stop);
  455. /* Like 're_search', but return how many characters in STRING the regexp
  456. in BUFFER matched, starting at position START. */
  457. extern regoff_t re_match (struct re_pattern_buffer *__buffer,
  458. const char *__String, regoff_t __length,
  459. regoff_t __start, struct re_registers *__regs);
  460. /* Relates to 're_match' as 're_search_2' relates to 're_search'. */
  461. extern regoff_t re_match_2 (struct re_pattern_buffer *__buffer,
  462. const char *__string1, regoff_t __length1,
  463. const char *__string2, regoff_t __length2,
  464. regoff_t __start, struct re_registers *__regs,
  465. regoff_t __stop);
  466. /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
  467. ENDS. Subsequent matches using BUFFER and REGS will use this memory
  468. for recording register information. STARTS and ENDS must be
  469. allocated with malloc, and must each be at least 'NUM_REGS * sizeof
  470. (regoff_t)' bytes long.
  471. If NUM_REGS == 0, then subsequent matches should allocate their own
  472. register data.
  473. Unless this function is called, the first search or match using
  474. BUFFER will allocate its own register data, without
  475. freeing the old data. */
  476. extern void re_set_registers (struct re_pattern_buffer *__buffer,
  477. struct re_registers *__regs,
  478. __re_size_t __num_regs,
  479. regoff_t *__starts, regoff_t *__ends);
  480. #endif /* Use GNU */
  481. #if defined _REGEX_RE_COMP || (defined _LIBC && defined __USE_MISC)
  482. # ifndef _CRAY
  483. /* 4.2 bsd compatibility. */
  484. extern char *re_comp (const char *);
  485. extern int re_exec (const char *);
  486. # endif
  487. #endif
  488. /* For plain 'restrict', use glibc's __restrict if defined.
  489. Otherwise, GCC 2.95 and later have "__restrict"; C99 compilers have
  490. "restrict", and "configure" may have defined "restrict".
  491. Other compilers use __restrict, __restrict__, and _Restrict, and
  492. 'configure' might #define 'restrict' to those words, so pick a
  493. different name. */
  494. #ifndef _Restrict_
  495. # if defined __restrict || 2 < __GNUC__ + (95 <= __GNUC_MINOR__)
  496. # define _Restrict_ __restrict
  497. # elif 199901L <= __STDC_VERSION__ || defined restrict
  498. # define _Restrict_ restrict
  499. # else
  500. # define _Restrict_
  501. # endif
  502. #endif
  503. /* For [restrict], use glibc's __restrict_arr if available.
  504. Otherwise, GCC 3.1 (not in C++ mode) and C99 support [restrict]. */
  505. #ifndef _Restrict_arr_
  506. # ifdef __restrict_arr
  507. # define _Restrict_arr_ __restrict_arr
  508. # elif ((199901L <= __STDC_VERSION__ || 3 < __GNUC__ + (1 <= __GNUC_MINOR__)) \
  509. && !defined __GNUG__)
  510. # define _Restrict_arr_ _Restrict_
  511. # else
  512. # define _Restrict_arr_
  513. # endif
  514. #endif
  515. /* POSIX compatibility. */
  516. extern int regcomp (regex_t *_Restrict_ __preg,
  517. const char *_Restrict_ __pattern,
  518. int __cflags);
  519. extern int regexec (const regex_t *_Restrict_ __preg,
  520. const char *_Restrict_ __String, size_t __nmatch,
  521. regmatch_t __pmatch[_Restrict_arr_],
  522. int __eflags);
  523. extern size_t regerror (int __errcode, const regex_t *_Restrict_ __preg,
  524. char *_Restrict_ __errbuf, size_t __errbuf_size);
  525. extern void regfree (regex_t *__preg);
  526. #ifdef __cplusplus
  527. }
  528. #endif /* C++ */
  529. #endif /* regex.h */