glib/gregex.h

   1 /* GRegex -- regular expression API wrapper around PCRE.
   2  *
   3  * Copyright (C) 1999, 2000 Scott Wimer
   4  * Copyright (C) 2004, Matthias Clasen <mclasen@redhat.com>
   5  * Copyright (C) 2005 - 2007, Marco Barisione <marco@barisione.org>
   6  *
   7  * This library is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2.1 of the License, or (at your option) any later version.
  11  *
  12  * This library is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with this library; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  20  */
  21
  22 #if defined(G_DISABLE_SINGLE_INCLUDES) && !defined (__GLIB_H_INSIDE__) && !defined (GLIB_COMPILATION)
  23 #error "Only <glib.h> can be included directly."
  24 #endif
  25
  26 #ifndef __G_REGEX_H__
  27 #define __G_REGEX_H__
  28
  29 #include <glib/gerror.h>
  30 #include <glib/gstring.h>
  31
  32 G_BEGIN_DECLS
  33
  34 typedef enum
  35 {
  36   G_REGEX_ERROR_COMPILE,
  37   G_REGEX_ERROR_OPTIMIZE,
  38   G_REGEX_ERROR_REPLACE,
  39   G_REGEX_ERROR_MATCH,
  40   G_REGEX_ERROR_INTERNAL,
  41
  42   /* These are the error codes from PCRE + 100 */
  43   G_REGEX_ERROR_STRAY_BACKSLASH = 101,
  44   G_REGEX_ERROR_MISSING_CONTROL_CHAR = 102,
  45   G_REGEX_ERROR_UNRECOGNIZED_ESCAPE = 103,
  46   G_REGEX_ERROR_QUANTIFIERS_OUT_OF_ORDER = 104,
  47   G_REGEX_ERROR_QUANTIFIER_TOO_BIG = 105,
  48   G_REGEX_ERROR_UNTERMINATED_CHARACTER_CLASS = 106,
  49   G_REGEX_ERROR_INVALID_ESCAPE_IN_CHARACTER_CLASS = 107,
  50   G_REGEX_ERROR_RANGE_OUT_OF_ORDER = 108,
  51   G_REGEX_ERROR_NOTHING_TO_REPEAT = 109,
  52   G_REGEX_ERROR_UNRECOGNIZED_CHARACTER = 112,
  53   G_REGEX_ERROR_POSIX_NAMED_CLASS_OUTSIDE_CLASS = 113,
  54   G_REGEX_ERROR_UNMATCHED_PARENTHESIS = 114,
  55   G_REGEX_ERROR_INEXISTENT_SUBPATTERN_REFERENCE = 115,
  56   G_REGEX_ERROR_UNTERMINATED_COMMENT = 118,
  57   G_REGEX_ERROR_EXPRESSION_TOO_LARGE = 120,
  58   G_REGEX_ERROR_MEMORY_ERROR = 121,
  59   G_REGEX_ERROR_VARIABLE_LENGTH_LOOKBEHIND = 125,
  60   G_REGEX_ERROR_MALFORMED_CONDITION = 126,
  61   G_REGEX_ERROR_TOO_MANY_CONDITIONAL_BRANCHES = 127,
  62   G_REGEX_ERROR_ASSERTION_EXPECTED = 128,
  63   G_REGEX_ERROR_UNKNOWN_POSIX_CLASS_NAME = 130,
  64   G_REGEX_ERROR_POSIX_COLLATING_ELEMENTS_NOT_SUPPORTED = 131,
  65   G_REGEX_ERROR_HEX_CODE_TOO_LARGE = 134,
  66   G_REGEX_ERROR_INVALID_CONDITION = 135,
  67   G_REGEX_ERROR_SINGLE_BYTE_MATCH_IN_LOOKBEHIND = 136,
  68   G_REGEX_ERROR_INFINITE_LOOP = 140,
  69   G_REGEX_ERROR_MISSING_SUBPATTERN_NAME_TERMINATOR = 142,
  70   G_REGEX_ERROR_DUPLICATE_SUBPATTERN_NAME = 143,
  71   G_REGEX_ERROR_MALFORMED_PROPERTY = 146,
  72   G_REGEX_ERROR_UNKNOWN_PROPERTY = 147,
  73   G_REGEX_ERROR_SUBPATTERN_NAME_TOO_LONG = 148,
  74   G_REGEX_ERROR_TOO_MANY_SUBPATTERNS = 149,
  75   G_REGEX_ERROR_INVALID_OCTAL_VALUE = 151,
  76   G_REGEX_ERROR_TOO_MANY_BRANCHES_IN_DEFINE = 154,
  77   G_REGEX_ERROR_DEFINE_REPETION = 155,
  78   G_REGEX_ERROR_INCONSISTENT_NEWLINE_OPTIONS = 156,
  79   G_REGEX_ERROR_MISSING_BACK_REFERENCE = 157
  80 } GRegexError;
  81
  82 #define G_REGEX_ERROR g_regex_error_quark ()
  83
  84 GQuark g_regex_error_quark (void);
  85
  86 /* Remember to update G_REGEX_COMPILE_MASK in gregex.c after
  87  * adding a new flag. */
  88 typedef enum
  89 {
  90   G_REGEX_CASELESS          = 1 << 0,
  91   G_REGEX_MULTILINE         = 1 << 1,
  92   G_REGEX_DOTALL            = 1 << 2,
  93   G_REGEX_EXTENDED          = 1 << 3,
  94   G_REGEX_ANCHORED          = 1 << 4,
  95   G_REGEX_DOLLAR_ENDONLY    = 1 << 5,
  96   G_REGEX_UNGREEDY          = 1 << 9,
  97   G_REGEX_RAW               = 1 << 11,
  98   G_REGEX_NO_AUTO_CAPTURE   = 1 << 12,
  99   G_REGEX_OPTIMIZE          = 1 << 13,
 100   G_REGEX_DUPNAMES          = 1 << 19,
 101   G_REGEX_NEWLINE_CR        = 1 << 20,
 102   G_REGEX_NEWLINE_LF        = 1 << 21,
 103   G_REGEX_NEWLINE_CRLF      = G_REGEX_NEWLINE_CR | G_REGEX_NEWLINE_LF
 104 } GRegexCompileFlags;
 105
 106 /* Remember to update G_REGEX_MATCH_MASK in gregex.c after
 107  * adding a new flag. */
 108 typedef enum
 109 {
 110   G_REGEX_MATCH_ANCHORED      = 1 << 4,
 111   G_REGEX_MATCH_NOTBOL        = 1 << 7,
 112   G_REGEX_MATCH_NOTEOL        = 1 << 8,
 113   G_REGEX_MATCH_NOTEMPTY      = 1 << 10,
 114   G_REGEX_MATCH_PARTIAL       = 1 << 15,
 115   G_REGEX_MATCH_NEWLINE_CR    = 1 << 20,
 116   G_REGEX_MATCH_NEWLINE_LF    = 1 << 21,
 117   G_REGEX_MATCH_NEWLINE_CRLF  = G_REGEX_MATCH_NEWLINE_CR | G_REGEX_MATCH_NEWLINE_LF,
 118   G_REGEX_MATCH_NEWLINE_ANY   = 1 << 22
 119 } GRegexMatchFlags;
 120
 121 typedef struct _GRegex          GRegex;
 122 typedef struct _GMatchInfo      GMatchInfo;
 123
 124 typedef gboolean (*GRegexEvalCallback)          (const GMatchInfo *match_info,
 125                                                  GString          *result,
 126                                                  gpointer          user_data);
 127
 128
 129 GRegex           *g_regex_new                   (const gchar         *pattern,
 130                                                  GRegexCompileFlags   compile_options,
 131                                                  GRegexMatchFlags     match_options,
 132                                                  GError             **error);
 133 GRegex           *g_regex_ref                   (GRegex              *regex);
 134 void              g_regex_unref                 (GRegex              *regex);
 135 const gchar      *g_regex_get_pattern           (const GRegex        *regex);
 136 gint              g_regex_get_max_backref       (const GRegex        *regex);
 137 gint              g_regex_get_capture_count     (const GRegex        *regex);
 138 gint              g_regex_get_string_number     (const GRegex        *regex,
 139                                                  const gchar         *name);
 140 gchar            *g_regex_escape_string         (const gchar         *string,
 141                                                  gint                 length);
 142
 143 /* Matching. */
 144 gboolean          g_regex_match_simple          (const gchar         *pattern,
 145                                                  const gchar         *string,
 146                                                  GRegexCompileFlags   compile_options,
 147                                                  GRegexMatchFlags     match_options);
 148 gboolean          g_regex_match                 (const GRegex        *regex,
 149                                                  const gchar         *string,
 150                                                  GRegexMatchFlags     match_options,
 151                                                  GMatchInfo         **match_info);
 152 gboolean          g_regex_match_full            (const GRegex        *regex,
 153                                                  const gchar         *string,
 154                                                  gssize               string_len,
 155                                                  gint                 start_position,
 156                                                  GRegexMatchFlags     match_options,
 157                                                  GMatchInfo         **match_info,
 158                                                  GError             **error);
 159 gboolean          g_regex_match_all             (const GRegex        *regex,
 160                                                  const gchar         *string,
 161                                                  GRegexMatchFlags     match_options,
 162                                                  GMatchInfo         **match_info);
 163 gboolean          g_regex_match_all_full        (const GRegex        *regex,
 164                                                  const gchar         *string,
 165                                                  gssize               string_len,
 166                                                  gint                 start_position,
 167                                                  GRegexMatchFlags     match_options,
 168                                                  GMatchInfo         **match_info,
 169                                                  GError             **error);
 170
 171 /* String splitting. */
 172 gchar           **g_regex_split_simple          (const gchar         *pattern,
 173                                                  const gchar         *string,
 174                                                  GRegexCompileFlags   compile_options,
 175                                                  GRegexMatchFlags     match_options);
 176 gchar           **g_regex_split                 (const GRegex        *regex,
 177                                                  const gchar         *string,
 178                                                  GRegexMatchFlags     match_options);
 179 gchar           **g_regex_split_full            (const GRegex        *regex,
 180                                                  const gchar         *string,
 181                                                  gssize               string_len,
 182                                                  gint                 start_position,
 183                                                  GRegexMatchFlags     match_options,
 184                                                  gint                 max_tokens,
 185                                                  GError             **error);
 186
 187 /* String replacement. */
 188 gchar            *g_regex_replace               (const GRegex        *regex,
 189                                                  const gchar         *string,
 190                                                  gssize               string_len,
 191                                                  gint                 start_position,
 192                                                  const gchar         *replacement,
 193                                                  GRegexMatchFlags     match_options,
 194                                                  GError             **error);
 195 gchar            *g_regex_replace_literal       (const GRegex        *regex,
 196                                                  const gchar         *string,
 197                                                  gssize               string_len,
 198                                                  gint                 start_position,
 199                                                  const gchar         *replacement,
 200                                                  GRegexMatchFlags     match_options,
 201                                                  GError             **error);
 202 gchar            *g_regex_replace_eval          (const GRegex        *regex,
 203                                                  const gchar         *string,
 204                                                  gssize               string_len,
 205                                                  gint                 start_position,
 206                                                  GRegexMatchFlags     match_options,
 207                                                  GRegexEvalCallback   eval,
 208                                                  gpointer             user_data,
 209                                                  GError             **error);
 210 gboolean          g_regex_check_replacement     (const gchar         *replacement,
 211                                                  gboolean            *has_references,
 212                                                  GError             **error);
 213
 214 /* Match info */
 215 GRegex           *g_match_info_get_regex        (const GMatchInfo    *match_info);
 216 const gchar      *g_match_info_get_string       (const GMatchInfo    *match_info);
 217
 218 void              g_match_info_free             (GMatchInfo          *match_info);
 219 gboolean          g_match_info_next             (GMatchInfo          *match_info,
 220                                                  GError             **error);
 221 gboolean          g_match_info_matches          (const GMatchInfo    *match_info);
 222 gint              g_match_info_get_match_count  (const GMatchInfo    *match_info);
 223 gboolean          g_match_info_is_partial_match (const GMatchInfo    *match_info);
 224 gchar            *g_match_info_expand_references(const GMatchInfo    *match_info,
 225                                                  const gchar         *string_to_expand,
 226                                                  GError             **error);
 227 gchar            *g_match_info_fetch            (const GMatchInfo    *match_info,
 228                                                  gint                 match_num);
 229 gboolean          g_match_info_fetch_pos        (const GMatchInfo    *match_info,
 230                                                  gint                 match_num,
 231                                                  gint                *start_pos,
 232                                                  gint                *end_pos);
 233 gchar            *g_match_info_fetch_named      (const GMatchInfo    *match_info,
 234                                                  const gchar         *name);
 235 gboolean          g_match_info_fetch_named_pos  (const GMatchInfo    *match_info,
 236                                                  const gchar         *name,
 237                                                  gint                *start_pos,
 238                                                  gint                *end_pos);
 239 gchar           **g_match_info_fetch_all        (const GMatchInfo    *match_info);
 240
 241 G_END_DECLS
 242
 243 #endif  /*  __G_REGEX_H__ */