regex/TRE/lib/tre.h

   1 /*
   2   tre.h - TRE public API definitions
   3
   4   This software is released under a BSD-style license.
   5   See the file LICENSE for details and copyright.
   6
   7 */
   8
   9 #ifndef TRE_H
  10 #define TRE_H 1
  11
  12 #include "tre-config.h"
  13
  14 #ifdef HAVE_SYS_TYPES_H
  15 #include <sys/types.h>
  16 #endif /* HAVE_SYS_TYPES_H */
  17
  18 #ifdef HAVE_LIBUTF8_H
  19 #include <libutf8.h>
  20 #endif /* HAVE_LIBUTF8_H */
  21
  22 #ifdef TRE_USE_SYSTEM_REGEX_H
  23 /* Include the system regex.h to make TRE ABI compatible with the
  24    system regex. */
  25 #include TRE_SYSTEM_REGEX_H_PATH
  26 #define tre_regcomp    regcomp
  27 #define tre_regexec    regexec
  28 #define tre_regerror   regerror
  29 #define tre_regfree    regfree
  30
  31 #ifdef TRE_APPROX
  32 #define tre_regaexec   regaexec
  33 #define tre_reganexec  reganexec
  34 #define tre_regawexec  regawexec
  35 #define tre_regawnexec regawnexec
  36 #endif /* TRE_APPROX */
  37 #define tre_regncomp   regncomp
  38 #define tre_regnexec   regnexec
  39 #define tre_regwcomp   regwcomp
  40 #define tre_regwexec   regwexec
  41 #define tre_regwncomp  regwncomp
  42 #define tre_regwnexec  regwnexec
  43
  44 #define tre_regcomp_l    regcomp_l
  45 #define tre_regncomp_l   regncomp_l
  46 #define tre_regwcomp_l   regwcomp_l
  47 #define tre_regwncomp_l  regwncomp_l
  48 #endif /* TRE_USE_SYSTEM_REGEX_H */
  49
  50 #ifdef __cplusplus
  51 extern "C" {
  52 #endif
  53
  54 #ifdef TRE_USE_SYSTEM_REGEX_H
  55
  56 #ifndef REG_OK
  57 #define REG_OK 0
  58 #endif /* !REG_OK */
  59
  60 #ifndef HAVE_REG_ERRCODE_T
  61 typedef int reg_errcode_t;
  62 #endif /* !HAVE_REG_ERRCODE_T */
  63
  64 #if !defined(REG_NOSPEC) && !defined(REG_LITERAL)
  65 #define REG_LITERAL 0x1000
  66 #endif
  67
  68 /* Extra tre_regcomp() flags. */
  69 #ifndef REG_BASIC
  70 #define REG_BASIC       0
  71 #endif /* !REG_BASIC */
  72 #if 0
  73 #define REG_LEFT_ASSOC  (REG_LITERAL << 1)
  74 #define REG_UNGREEDY    (REG_LEFT_ASSOC << 1)
  75
  76 /* Extra tre_regexec() flags. */
  77 #define REG_APPROX_MATCHER       0x1000
  78 #define REG_BACKTRACKING_MATCHER (REG_APPROX_MATCHER << 1)
  79 #endif
  80
  81 #define RE_MAGIC  ((('r'^0200)<<8) | 'e')
  82
  83 #else /* !TRE_USE_SYSTEM_REGEX_H */
  84
  85 /* If the we're not using system regex.h, we need to define the
  86    structs and enums ourselves. */
  87
  88 typedef int regoff_t;
  89 typedef struct {
  90   size_t re_nsub;      /* Number of parenthesized subexpressions. */
  91   const void *re_endp; /* regex string end pointer (REG_PEND) */
  92   void *value;         /* For internal use only. */
  93 } regex_t;
  94
  95 typedef struct {
  96   regoff_t rm_so;
  97   regoff_t rm_eo;
  98 } regmatch_t;
  99
 100
 101 typedef enum {
 102   REG_OK = 0,           /* No error. */
 103   /* POSIX tre_regcomp() return error codes.  (In the order listed in the
 104      standard.)  */
 105   REG_NOMATCH,          /* No match. */
 106   REG_BADPAT,           /* Invalid regexp. */
 107   REG_ECOLLATE,         /* Unknown collating element. */
 108   REG_ECTYPE,           /* Unknown character class name. */
 109   REG_EESCAPE,          /* Trailing backslash. */
 110   REG_ESUBREG,          /* Invalid back reference. */
 111   REG_EBRACK,           /* "[]" imbalance */
 112   REG_EPAREN,           /* "\(\)" or "()" imbalance */
 113   REG_EBRACE,           /* "\{\}" or "{}" imbalance */
 114   REG_BADBR,            /* Invalid content of {} */
 115   REG_ERANGE,           /* Invalid use of range operator */
 116   REG_ESPACE,           /* Out of memory.  */
 117   REG_BADRPT,           /* Invalid use of repetition operators. */
 118   REG_INVARG,           /* Invalid argument, e.g. negative-length string */
 119   REG_ILLSEQ,           /* illegal byte sequence (bad multibyte character) */
 120 } reg_errcode_t;
 121
 122 /* POSIX tre_regcomp() flags. */
 123 #define REG_EXTENDED    1
 124 #define REG_ICASE       (REG_EXTENDED << 1)
 125 #define REG_NEWLINE     (REG_ICASE << 1)
 126 #define REG_NOSUB       (REG_NEWLINE << 1)
 127
 128 /* Extra tre_regcomp() flags. */
 129 #define REG_BASIC       0
 130 #define REG_LITERAL     (REG_NOSUB << 1)
 131 #define REG_LEFT_ASSOC  (REG_LITERAL << 1)
 132 #define REG_UNGREEDY    (REG_LEFT_ASSOC << 1)
 133 #define REG_PEND        (REG_UNGREEDY << 1)
 134
 135 /* POSIX tre_regexec() flags. */
 136 #define REG_NOTBOL 1
 137 #define REG_NOTEOL (REG_NOTBOL << 1)
 138
 139 /* Extra tre_regexec() flags. */
 140 #define REG_APPROX_MATCHER       (REG_NOTEOL << 1)
 141 #define REG_BACKTRACKING_MATCHER (REG_APPROX_MATCHER << 1)
 142 #define REG_STARTEND             (REG_BACKTRACKING_MATCHER << 1)
 143
 144 #endif /* !TRE_USE_SYSTEM_REGEX_H */
 145
 146 /* REG_NOSPEC and REG_LITERAL mean the same thing. */
 147 #if defined(REG_LITERAL) && !defined(REG_NOSPEC)
 148 #define REG_NOSPEC      REG_LITERAL
 149 #elif defined(REG_NOSPEC) && !defined(REG_LITERAL)
 150 #define REG_LITERAL     REG_NOSPEC
 151 #endif /* defined(REG_NOSPEC) */
 152
 153 /* The maximum number of iterations in a bound expression. */
 154 #undef RE_DUP_MAX
 155 #define RE_DUP_MAX 255
 156
 157 /* The POSIX.2 regexp functions */
 158 extern int
 159 tre_regcomp(regex_t * __restrict preg, const char * __restrict regex, int cflags);
 160
 161 extern int
 162 tre_regexec(const regex_t * __restrict preg, const char * __restrict string, size_t nmatch,
 163         regmatch_t pmatch[ __restrict ], int eflags);
 164
 165 extern size_t
 166 tre_regerror(int errcode, const regex_t * __restrict preg, char * __restrict errbuf,
 167          size_t errbuf_size);
 168
 169 extern void
 170 tre_regfree(regex_t *preg);
 171
 172 #ifdef TRE_WCHAR
 173 #ifdef HAVE_WCHAR_H
 174 #include <wchar.h>
 175 #endif /* HAVE_WCHAR_H */
 176
 177 /* Wide character versions (not in POSIX.2). */
 178 extern int
 179 tre_regwcomp(regex_t * __restrict preg, const wchar_t * __restrict regex, int cflags);
 180
 181 extern int
 182 tre_regwexec(const regex_t * __restrict preg, const wchar_t * __restrict string,
 183          size_t nmatch, regmatch_t pmatch[ __restrict ], int eflags);
 184 #endif /* TRE_WCHAR */
 185
 186 /* Versions with a maximum length argument and therefore the capability to
 187    handle null characters in the middle of the strings (not in POSIX.2). */
 188 extern int
 189 tre_regncomp(regex_t * __restrict preg, const char * __restrict regex, size_t len, int cflags);
 190
 191 extern int
 192 tre_regnexec(const regex_t * __restrict preg, const char * __restrict string, size_t len,
 193          size_t nmatch, regmatch_t pmatch[ __restrict ], int eflags);
 194
 195 #ifdef TRE_WCHAR
 196 extern int
 197 tre_regwncomp(regex_t * __restrict preg, const wchar_t * __restrict regex, size_t len, int cflags);
 198
 199 extern int
 200 tre_regwnexec(const regex_t * __restrict preg, const wchar_t * __restrict string, size_t len,
 201           size_t nmatch, regmatch_t pmatch[ __restrict ], int eflags);
 202 #endif /* TRE_WCHAR */
 203
 204 /* Extended locale versions */
 205 #include <xlocale.h>
 206
 207 extern int
 208 tre_regcomp_l(regex_t * __restrict preg, const char * __restrict regex, int cflags, locale_t __restrict loc);
 209
 210 #ifdef TRE_WCHAR
 211 extern int
 212 tre_regwcomp_l(regex_t * __restrict preg, const wchar_t * __restrict regex, int cflags, locale_t __restrict loc);
 213 #endif /* TRE_WCHAR */
 214
 215 extern int
 216 tre_regncomp_l(regex_t * __restrict preg, const char * __restrict regex, size_t len, int cflags, locale_t __restrict loc);
 217
 218 #ifdef TRE_WCHAR
 219 extern int
 220 tre_regwncomp_l(regex_t * __restrict preg, const wchar_t * __restrict regex, size_t len, int cflags, locale_t __restrict loc);
 221 #endif /* TRE_WCHAR */
 222
 223 #ifdef TRE_APPROX
 224
 225 /* Approximate matching parameter struct. */
 226 typedef struct {
 227   int cost_ins;        /* Default cost of an inserted character. */
 228   int cost_del;        /* Default cost of a deleted character. */
 229   int cost_subst;      /* Default cost of a substituted character. */
 230   int max_cost;        /* Maximum allowed cost of a match. */
 231
 232   int max_ins;         /* Maximum allowed number of inserts. */
 233   int max_del;         /* Maximum allowed number of deletes. */
 234   int max_subst;       /* Maximum allowed number of substitutes. */
 235   int max_err;         /* Maximum allowed number of errors total. */
 236 } regaparams_t;
 237
 238 /* Approximate matching result struct. */
 239 typedef struct {
 240   size_t nmatch;       /* Length of pmatch[] array. */
 241   regmatch_t *pmatch;  /* Submatch data. */
 242   int cost;            /* Cost of the match. */
 243   int num_ins;         /* Number of inserts in the match. */
 244   int num_del;         /* Number of deletes in the match. */
 245   int num_subst;       /* Number of substitutes in the match. */
 246 } regamatch_t;
 247
 248
 249 /* Approximate matching functions. */
 250 extern int
 251 tre_regaexec(const regex_t * __restrict preg, const char * __restrict string,
 252          regamatch_t * __restrict match, regaparams_t params, int eflags);
 253
 254 extern int
 255 tre_reganexec(const regex_t * __restrict preg, const char * __restrict string, size_t len,
 256           regamatch_t * __restrict match, regaparams_t params, int eflags);
 257 #ifdef TRE_WCHAR
 258 /* Wide character approximate matching. */
 259 extern int
 260 tre_regawexec(const regex_t * __restrict preg, const wchar_t * __restrict string,
 261           regamatch_t * __restrict match, regaparams_t params, int eflags);
 262
 263 extern int
 264 tre_regawnexec(const regex_t * __restrict preg, const wchar_t * __restrict string, size_t len,
 265            regamatch_t * __restrict match, regaparams_t params, int eflags);
 266 #endif /* TRE_WCHAR */
 267
 268 /* Sets the parameters to default values. */
 269 extern void
 270 tre_regaparams_default(regaparams_t *params);
 271 #endif /* TRE_APPROX */
 272
 273 #ifdef TRE_WCHAR
 274 typedef wchar_t tre_char_t;
 275 #else /* !TRE_WCHAR */
 276 typedef unsigned char tre_char_t;
 277 #endif /* !TRE_WCHAR */
 278
 279 #ifdef TRE_STR_USER
 280 typedef struct {
 281   int (*get_next_char)(tre_char_t *c, unsigned int *pos_add, void *context);
 282   void (*rewind)(size_t pos, void *context);
 283   int (*compare)(size_t pos1, size_t pos2, size_t len, void *context);
 284   void *context;
 285 } tre_str_source;
 286
 287 extern int
 288 tre_reguexec(const regex_t * __restrict preg, const tre_str_source * __restrict string,
 289          size_t nmatch, regmatch_t pmatch[ __restrict ], int eflags);
 290 #endif /* TRE_STR_USER */
 291
 292 #ifndef __LIBC__
 293 /* Returns the version string.  The returned string is static. */
 294 extern char *
 295 tre_version(void);
 296
 297 /* Returns the value for a config parameter.  The type to which `result'
 298    must point to depends of the value of `query', see documentation for
 299    more details. */
 300 extern int
 301 tre_config(int query, void *result);
 302
 303 enum {
 304   TRE_CONFIG_APPROX,
 305   TRE_CONFIG_WCHAR,
 306   TRE_CONFIG_MULTIBYTE,
 307   TRE_CONFIG_SYSTEM_ABI,
 308   TRE_CONFIG_VERSION
 309 };
 310
 311 /* Returns 1 if the compiled pattern has back references, 0 if not. */
 312 extern int
 313 tre_have_backrefs(const regex_t *preg);
 314
 315 #ifdef TRE_APPROX
 316 /* Returns 1 if the compiled pattern uses approximate matching features,
 317    0 if not. */
 318 extern int
 319 tre_have_approx(const regex_t *preg);
 320 #endif /* TRE_APPROX */
 321 #endif /* !__LIBC__ */
 322
 323 #ifdef __cplusplus
 324 }
 325 #endif
 326 #endif                          /* TRE_H */
 327
 328 /* EOF */