glib/gscanner.h

   1 /* GLIB - Library of useful routines for C programming
   2  * Copyright (C) 1995-1997  Peter Mattis, Spencer Kimball and Josh MacDonald
   3  *
   4  * This library is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU Lesser General Public
   6  * License as published by the Free Software Foundation; either
   7  * version 2 of the License, or (at your option) any later version.
   8  *
   9  * This 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  *
  14  * You should have received a copy of the GNU Lesser General Public
  15  * License along with this library; if not, write to the
  16  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  17  * Boston, MA 02111-1307, USA.
  18  */
  19
  20 /*
  21  * Modified by the GLib Team and others 1997-2000.  See the AUTHORS
  22  * file for a list of people on the GLib Team.  See the ChangeLog
  23  * files for a list of changes.  These files are distributed with
  24  * GLib at ftp://ftp.gtk.org/pub/gtk/.
  25  */
  26
  27 #ifndef __G_SCANNER_H__
  28 #define __G_SCANNER_H__
  29
  30 #include <ghash.h>
  31
  32 G_BEGIN_DECLS
  33
  34 typedef struct _GScanner        GScanner;
  35 typedef struct _GScannerConfig  GScannerConfig;
  36 typedef union  _GTokenValue     GTokenValue;
  37
  38 typedef void            (*GScannerMsgFunc)      (GScanner      *scanner,
  39                                                  gchar         *message,
  40                                                  gint           error);
  41
  42 /* GScanner: Flexible lexical scanner for general purpose.
  43  */
  44
  45 /* Character sets */
  46 #define G_CSET_A_2_Z    "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
  47 #define G_CSET_a_2_z    "abcdefghijklmnopqrstuvwxyz"
  48 #define G_CSET_DIGITS   "0123456789"
  49 #define G_CSET_LATINC   "\300\301\302\303\304\305\306"\
  50                         "\307\310\311\312\313\314\315\316\317\320"\
  51                         "\321\322\323\324\325\326"\
  52                         "\330\331\332\333\334\335\336"
  53 #define G_CSET_LATINS   "\337\340\341\342\343\344\345\346"\
  54                         "\347\350\351\352\353\354\355\356\357\360"\
  55                         "\361\362\363\364\365\366"\
  56                         "\370\371\372\373\374\375\376\377"
  57
  58 /* Error types */
  59 typedef enum
  60 {
  61   G_ERR_UNKNOWN,
  62   G_ERR_UNEXP_EOF,
  63   G_ERR_UNEXP_EOF_IN_STRING,
  64   G_ERR_UNEXP_EOF_IN_COMMENT,
  65   G_ERR_NON_DIGIT_IN_CONST,
  66   G_ERR_DIGIT_RADIX,
  67   G_ERR_FLOAT_RADIX,
  68   G_ERR_FLOAT_MALFORMED
  69 } GErrorType;
  70
  71 /* Token types */
  72 typedef enum
  73 {
  74   G_TOKEN_EOF                   =   0,
  75
  76   G_TOKEN_LEFT_PAREN            = '(',
  77   G_TOKEN_RIGHT_PAREN           = ')',
  78   G_TOKEN_LEFT_CURLY            = '{',
  79   G_TOKEN_RIGHT_CURLY           = '}',
  80   G_TOKEN_LEFT_BRACE            = '[',
  81   G_TOKEN_RIGHT_BRACE           = ']',
  82   G_TOKEN_EQUAL_SIGN            = '=',
  83   G_TOKEN_COMMA                 = ',',
  84
  85   G_TOKEN_NONE                  = 256,
  86
  87   G_TOKEN_ERROR,
  88
  89   G_TOKEN_CHAR,
  90   G_TOKEN_BINARY,
  91   G_TOKEN_OCTAL,
  92   G_TOKEN_INT,
  93   G_TOKEN_HEX,
  94   G_TOKEN_FLOAT,
  95   G_TOKEN_STRING,
  96
  97   G_TOKEN_SYMBOL,
  98   G_TOKEN_IDENTIFIER,
  99   G_TOKEN_IDENTIFIER_NULL,
 100
 101   G_TOKEN_COMMENT_SINGLE,
 102   G_TOKEN_COMMENT_MULTI,
 103   G_TOKEN_LAST
 104 } GTokenType;
 105
 106 union   _GTokenValue
 107 {
 108   gpointer      v_symbol;
 109   gchar         *v_identifier;
 110   gulong        v_binary;
 111   gulong        v_octal;
 112   gulong        v_int;
 113   gdouble       v_float;
 114   gulong        v_hex;
 115   gchar         *v_string;
 116   gchar         *v_comment;
 117   guchar        v_char;
 118   guint         v_error;
 119 };
 120
 121 struct  _GScannerConfig
 122 {
 123   /* Character sets
 124    */
 125   gchar         *cset_skip_characters;          /* default: " \t\n" */
 126   gchar         *cset_identifier_first;
 127   gchar         *cset_identifier_nth;
 128   gchar         *cpair_comment_single;          /* default: "#\n" */
 129
 130   /* Should symbol lookup work case sensitive?
 131    */
 132   guint         case_sensitive : 1;
 133
 134   /* Boolean values to be adjusted "on the fly"
 135    * to configure scanning behaviour.
 136    */
 137   guint         skip_comment_multi : 1;         /* C like comment */
 138   guint         skip_comment_single : 1;        /* single line comment */
 139   guint         scan_comment_multi : 1;         /* scan multi line comments? */
 140   guint         scan_identifier : 1;
 141   guint         scan_identifier_1char : 1;
 142   guint         scan_identifier_NULL : 1;
 143   guint         scan_symbols : 1;
 144   guint         scan_binary : 1;
 145   guint         scan_octal : 1;
 146   guint         scan_float : 1;
 147   guint         scan_hex : 1;                   /* `0x0ff0' */
 148   guint         scan_hex_dollar : 1;            /* `$0ff0' */
 149   guint         scan_string_sq : 1;             /* string: 'anything' */
 150   guint         scan_string_dq : 1;             /* string: "\\-escapes!\n" */
 151   guint         numbers_2_int : 1;              /* bin, octal, hex => int */
 152   guint         int_2_float : 1;                /* int => G_TOKEN_FLOAT? */
 153   guint         identifier_2_string : 1;
 154   guint         char_2_token : 1;               /* return G_TOKEN_CHAR? */
 155   guint         symbol_2_token : 1;
 156   guint         scope_0_fallback : 1;           /* try scope 0 on lookups? */
 157 };
 158
 159 struct  _GScanner
 160 {
 161   /* unused fields */
 162   gpointer              user_data;
 163   guint                 max_parse_errors;
 164
 165   /* g_scanner_error() increments this field */
 166   guint                 parse_errors;
 167
 168   /* name of input stream, featured by the default message handler */
 169   const gchar           *input_name;
 170
 171   /* data pointer for derived structures */
 172   gpointer              derived_data;
 173
 174   /* link into the scanner configuration */
 175   GScannerConfig        *config;
 176
 177   /* fields filled in after g_scanner_get_next_token() */
 178   GTokenType            token;
 179   GTokenValue           value;
 180   guint                 line;
 181   guint                 position;
 182
 183   /* fields filled in after g_scanner_peek_next_token() */
 184   GTokenType            next_token;
 185   GTokenValue           next_value;
 186   guint                 next_line;
 187   guint                 next_position;
 188
 189   /* to be considered private */
 190   GHashTable            *symbol_table;
 191   gint                  input_fd;
 192   const gchar           *text;
 193   const gchar           *text_end;
 194   gchar                 *buffer;
 195   guint                 scope_id;
 196
 197   /* handler function for _warn and _error */
 198   GScannerMsgFunc       msg_handler;
 199 };
 200
 201 GScanner*       g_scanner_new                   (GScannerConfig *config_templ);
 202 void            g_scanner_destroy               (GScanner       *scanner);
 203 void            g_scanner_input_file            (GScanner       *scanner,
 204                                                  gint           input_fd);
 205 void            g_scanner_sync_file_offset      (GScanner       *scanner);
 206 void            g_scanner_input_text            (GScanner       *scanner,
 207                                                  const  gchar   *text,
 208                                                  guint          text_len);
 209 GTokenType      g_scanner_get_next_token        (GScanner       *scanner);
 210 GTokenType      g_scanner_peek_next_token       (GScanner       *scanner);
 211 GTokenType      g_scanner_cur_token             (GScanner       *scanner);
 212 GTokenValue     g_scanner_cur_value             (GScanner       *scanner);
 213 guint           g_scanner_cur_line              (GScanner       *scanner);
 214 guint           g_scanner_cur_position          (GScanner       *scanner);
 215 gboolean        g_scanner_eof                   (GScanner       *scanner);
 216 guint           g_scanner_set_scope             (GScanner       *scanner,
 217                                                  guint           scope_id);
 218 void            g_scanner_scope_add_symbol      (GScanner       *scanner,
 219                                                  guint           scope_id,
 220                                                  const gchar    *symbol,
 221                                                  gpointer       value);
 222 void            g_scanner_scope_remove_symbol   (GScanner       *scanner,
 223                                                  guint           scope_id,
 224                                                  const gchar    *symbol);
 225 gpointer        g_scanner_scope_lookup_symbol   (GScanner       *scanner,
 226                                                  guint           scope_id,
 227                                                  const gchar    *symbol);
 228 void            g_scanner_scope_foreach_symbol  (GScanner       *scanner,
 229                                                  guint           scope_id,
 230                                                  GHFunc          func,
 231                                                  gpointer        user_data);
 232 gpointer        g_scanner_lookup_symbol         (GScanner       *scanner,
 233                                                  const gchar    *symbol);
 234 void            g_scanner_unexp_token           (GScanner       *scanner,
 235                                                  GTokenType     expected_token,
 236                                                  const gchar    *identifier_spec,
 237                                                  const gchar    *symbol_spec,
 238                                                  const gchar    *symbol_name,
 239                                                  const gchar    *message,
 240                                                  gint            is_error);
 241 void            g_scanner_error                 (GScanner       *scanner,
 242                                                  const gchar    *format,
 243                                                  ...) G_GNUC_PRINTF (2,3);
 244 void            g_scanner_warn                  (GScanner       *scanner,
 245                                                  const gchar    *format,
 246                                                  ...) G_GNUC_PRINTF (2,3);
 247 gint            g_scanner_stat_mode             (const gchar    *filename);
 248 /* keep downward source compatibility */
 249 #define         g_scanner_add_symbol( scanner, symbol, value )  G_STMT_START { \
 250   g_scanner_scope_add_symbol ((scanner), 0, (symbol), (value)); \
 251 } G_STMT_END
 252 #define         g_scanner_remove_symbol( scanner, symbol )      G_STMT_START { \
 253   g_scanner_scope_remove_symbol ((scanner), 0, (symbol)); \
 254 } G_STMT_END
 255 #define         g_scanner_foreach_symbol( scanner, func, data ) G_STMT_START { \
 256   g_scanner_scope_foreach_symbol ((scanner), 0, (func), (data)); \
 257 } G_STMT_END
 258
 259 /* The following two functions are deprecated and will be removed in
 260  * the next major release. They do no good. */
 261 void            g_scanner_freeze_symbol_table   (GScanner       *scanner);
 262 void            g_scanner_thaw_symbol_table     (GScanner       *scanner);
 263
 264 G_END_DECLS
 265
 266 #endif /* __G_SCANNER_H__ */
 267