gscanner.c

   1 /* GLIB - Library of useful routines for C programming
   2  * Copyright (C) 1995-1997  Peter Mattis, Spencer Kimball and Josh MacDonald
   3  *
   4  * GScanner: Flexible lexical scanner for general purpose.
   5  * Copyright (C) 1997, 1998 Tim Janik
   6  *
   7  * This library is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Library General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2 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  * Library General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Library General Public
  18  * License along with this library; if not, write to the
  19  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  20  * Boston, MA 02111-1307, USA.
  21  */
  22 #define         __gscanner_c__
  23
  24 #ifdef HAVE_CONFIG_H
  25 #include <config.h>
  26 #endif
  27
  28 #include        <stdlib.h>
  29 #include        <stdarg.h>
  30 #include        <string.h>
  31 #include        <stdio.h>
  32 #include        "glib.h"
  33 #ifdef HAVE_UNISTD_H
  34 #include        <unistd.h>
  35 #endif
  36 #include        <errno.h>
  37 #include        <sys/types.h>   /* needed for sys/stat.h */
  38 #include        <sys/stat.h>
  39 #ifdef _MSC_VER
  40 #include        <io.h>          /* For _read() */
  41 #endif
  42
  43 /* --- defines --- */
  44 #define to_lower(c)                             ( \
  45         (guchar) (                                                      \
  46           ( (((guchar)(c))>='A' && ((guchar)(c))<='Z') * ('a'-'A') ) |  \
  47           ( (((guchar)(c))>=192 && ((guchar)(c))<=214) * (224-192) ) |  \
  48           ( (((guchar)(c))>=216 && ((guchar)(c))<=222) * (248-216) ) |  \
  49           ((guchar)(c))                                                 \
  50         )                                                               \
  51 )
  52 #define READ_BUFFER_SIZE        (4000)
  53
  54
  55 /* --- typedefs --- */
  56 typedef struct  _GScannerKey    GScannerKey;
  57
  58 struct  _GScannerKey
  59 {
  60   guint          scope_id;
  61   gchar         *symbol;
  62   gpointer       value;
  63 };
  64
  65
  66
  67 /* --- variables --- */
  68 static GScannerConfig g_scanner_config_template =
  69 {
  70   (
  71    " \t\r\n"
  72    )                    /* cset_skip_characters */,
  73   (
  74    G_CSET_a_2_z
  75    "_"
  76    G_CSET_A_2_Z
  77    )                    /* cset_identifier_first */,
  78   (
  79    G_CSET_a_2_z
  80    "_0123456789"
  81    G_CSET_A_2_Z
  82    G_CSET_LATINS
  83    G_CSET_LATINC
  84    )                    /* cset_identifier_nth */,
  85   ( "#\n" )             /* cpair_comment_single */,
  86
  87   FALSE                 /* case_sensitive */,
  88
  89   TRUE                  /* skip_comment_multi */,
  90   TRUE                  /* skip_comment_single */,
  91   TRUE                  /* scan_comment_multi */,
  92   TRUE                  /* scan_identifier */,
  93   FALSE                 /* scan_identifier_1char */,
  94   FALSE                 /* scan_identifier_NULL */,
  95   TRUE                  /* scan_symbols */,
  96   FALSE                 /* scan_binary */,
  97   TRUE                  /* scan_octal */,
  98   TRUE                  /* scan_float */,
  99   TRUE                  /* scan_hex */,
 100   FALSE                 /* scan_hex_dollar */,
 101   TRUE                  /* scan_string_sq */,
 102   TRUE                  /* scan_string_dq */,
 103   TRUE                  /* numbers_2_int */,
 104   FALSE                 /* int_2_float */,
 105   FALSE                 /* identifier_2_string */,
 106   TRUE                  /* char_2_token */,
 107   FALSE                 /* symbol_2_token */,
 108   FALSE                 /* scope_0_fallback */,
 109 };
 110
 111
 112 /* --- prototypes --- */
 113 static inline
 114 GScannerKey*    g_scanner_lookup_internal (GScanner     *scanner,
 115                                            guint         scope_id,
 116                                            const gchar  *symbol);
 117 static gint     g_scanner_key_equal       (gconstpointer v1,
 118                                            gconstpointer v2);
 119 static guint    g_scanner_key_hash        (gconstpointer v);
 120 static void     g_scanner_get_token_ll    (GScanner     *scanner,
 121                                            GTokenType   *token_p,
 122                                            GTokenValue  *value_p,
 123                                            guint        *line_p,
 124                                            guint        *position_p);
 125 static void     g_scanner_get_token_i     (GScanner     *scanner,
 126                                            GTokenType   *token_p,
 127                                            GTokenValue  *value_p,
 128                                            guint        *line_p,
 129                                            guint        *position_p);
 130
 131 static guchar   g_scanner_peek_next_char  (GScanner     *scanner);
 132 static guchar   g_scanner_get_char        (GScanner     *scanner,
 133                                            guint        *line_p,
 134                                            guint        *position_p);
 135 static void     g_scanner_msg_handler     (GScanner     *scanner,
 136                                            gchar        *message,
 137                                            gint          is_error);
 138
 139
 140 /* --- functions --- */
 141 static inline gint
 142 g_scanner_char_2_num (guchar    c,
 143                       guchar    base)
 144 {
 145   if (c >= '0' && c <= '9')
 146     c -= '0';
 147   else if (c >= 'A' && c <= 'Z')
 148     c -= 'A' - 10;
 149   else if (c >= 'a' && c <= 'z')
 150     c -= 'a' - 10;
 151   else
 152     return -1;
 153
 154   if (c < base)
 155     return c;
 156
 157   return -1;
 158 }
 159
 160 GScanner*
 161 g_scanner_new (GScannerConfig   *config_templ)
 162 {
 163   register GScanner     *scanner;
 164
 165   if (!config_templ)
 166     config_templ = &g_scanner_config_template;
 167
 168   scanner = g_new0 (GScanner, 1);
 169
 170   scanner->user_data = NULL;
 171   scanner->max_parse_errors = 0;
 172   scanner->parse_errors = 0;
 173   scanner->input_name = NULL;
 174   scanner->derived_data = NULL;
 175
 176   scanner->config = g_new0 (GScannerConfig, 1);
 177
 178   scanner->config->case_sensitive       = config_templ->case_sensitive;
 179   scanner->config->cset_skip_characters = config_templ->cset_skip_characters;
 180   scanner->config->cset_identifier_first= config_templ->cset_identifier_first;
 181   scanner->config->cset_identifier_nth  = config_templ->cset_identifier_nth;
 182   scanner->config->cpair_comment_single = config_templ->cpair_comment_single;
 183   scanner->config->skip_comment_multi   = config_templ->skip_comment_multi;
 184   scanner->config->skip_comment_single  = config_templ->skip_comment_single;
 185   scanner->config->scan_comment_multi   = config_templ->scan_comment_multi;
 186   scanner->config->scan_identifier      = config_templ->scan_identifier;
 187   scanner->config->scan_identifier_1char= config_templ->scan_identifier_1char;
 188   scanner->config->scan_identifier_NULL = config_templ->scan_identifier_NULL;
 189   scanner->config->scan_symbols         = config_templ->scan_symbols;
 190   scanner->config->scan_binary          = config_templ->scan_binary;
 191   scanner->config->scan_octal           = config_templ->scan_octal;
 192   scanner->config->scan_float           = config_templ->scan_float;
 193   scanner->config->scan_hex             = config_templ->scan_hex;
 194   scanner->config->scan_hex_dollar      = config_templ->scan_hex_dollar;
 195   scanner->config->scan_string_sq       = config_templ->scan_string_sq;
 196   scanner->config->scan_string_dq       = config_templ->scan_string_dq;
 197   scanner->config->numbers_2_int        = config_templ->numbers_2_int;
 198   scanner->config->int_2_float          = config_templ->int_2_float;
 199   scanner->config->identifier_2_string  = config_templ->identifier_2_string;
 200   scanner->config->char_2_token         = config_templ->char_2_token;
 201   scanner->config->symbol_2_token       = config_templ->symbol_2_token;
 202   scanner->config->scope_0_fallback     = config_templ->scope_0_fallback;
 203
 204   scanner->token = G_TOKEN_NONE;
 205   scanner->value.v_int = 0;
 206   scanner->line = 1;
 207   scanner->position = 0;
 208
 209   scanner->next_token = G_TOKEN_NONE;
 210   scanner->next_value.v_int = 0;
 211   scanner->next_line = 1;
 212   scanner->next_position = 0;
 213
 214   scanner->symbol_table = g_hash_table_new (g_scanner_key_hash, g_scanner_key_equal);
 215   scanner->input_fd = -1;
 216   scanner->text = NULL;
 217   scanner->text_end = NULL;
 218   scanner->buffer = NULL;
 219   scanner->scope_id = 0;
 220
 221   scanner->msg_handler = g_scanner_msg_handler;
 222
 223   return scanner;
 224 }
 225
 226 static inline void
 227 g_scanner_free_value (GTokenType     *token_p,
 228                       GTokenValue     *value_p)
 229 {
 230   switch (*token_p)
 231     {
 232     case  G_TOKEN_STRING:
 233     case  G_TOKEN_IDENTIFIER:
 234     case  G_TOKEN_IDENTIFIER_NULL:
 235     case  G_TOKEN_COMMENT_SINGLE:
 236     case  G_TOKEN_COMMENT_MULTI:
 237       g_free (value_p->v_string);
 238       break;
 239
 240     default:
 241       break;
 242     }
 243
 244   *token_p = G_TOKEN_NONE;
 245 }
 246
 247 static void
 248 g_scanner_destroy_symbol_table_entry (gpointer _key,
 249                                       gpointer _value,
 250                                       gpointer _data)
 251 {
 252   GScannerKey *key = _key;
 253
 254   g_free (key->symbol);
 255   g_free (key);
 256 }
 257
 258 void
 259 g_scanner_destroy (GScanner     *scanner)
 260 {
 261   g_return_if_fail (scanner != NULL);
 262
 263   g_hash_table_foreach (scanner->symbol_table,
 264                         g_scanner_destroy_symbol_table_entry, NULL);
 265   g_hash_table_destroy (scanner->symbol_table);
 266   g_scanner_free_value (&scanner->token, &scanner->value);
 267   g_scanner_free_value (&scanner->next_token, &scanner->next_value);
 268   g_free (scanner->config);
 269   g_free (scanner->buffer);
 270   g_free (scanner);
 271 }
 272
 273 static void
 274 g_scanner_msg_handler (GScanner         *scanner,
 275                        gchar            *message,
 276                        gint              is_error)
 277 {
 278   g_return_if_fail (scanner != NULL);
 279
 280   fprintf (stdout, "%s:%d: ", scanner->input_name, scanner->line);
 281   if (is_error)
 282     fprintf (stdout, "error: ");
 283   fprintf (stdout, "%s\n", message);
 284 }
 285
 286 void
 287 g_scanner_error (GScanner       *scanner,
 288                  const gchar    *format,
 289                  ...)
 290 {
 291   g_return_if_fail (scanner != NULL);
 292   g_return_if_fail (format != NULL);
 293
 294   scanner->parse_errors++;
 295
 296   if (scanner->msg_handler)
 297     {
 298       va_list args;
 299       gchar *string;
 300
 301       va_start (args, format);
 302       string = g_strdup_vprintf (format, args);
 303       va_end (args);
 304
 305       scanner->msg_handler (scanner, string, TRUE);
 306
 307       g_free (string);
 308     }
 309 }
 310
 311 void
 312 g_scanner_warn (GScanner       *scanner,
 313                 const gchar    *format,
 314                 ...)
 315 {
 316   g_return_if_fail (scanner != NULL);
 317   g_return_if_fail (format != NULL);
 318
 319   if (scanner->msg_handler)
 320     {
 321       va_list args;
 322       gchar *string;
 323
 324       va_start (args, format);
 325       string = g_strdup_vprintf (format, args);
 326       va_end (args);
 327
 328       scanner->msg_handler (scanner, string, FALSE);
 329
 330       g_free (string);
 331     }
 332 }
 333
 334 static gint
 335 g_scanner_key_equal (gconstpointer v1,
 336                      gconstpointer v2)
 337 {
 338   register const GScannerKey *key1 = v1;
 339   register const GScannerKey *key2 = v2;
 340
 341   return (key1->scope_id == key2->scope_id) && (strcmp (key1->symbol, key2->symbol) == 0);
 342 }
 343
 344 static guint
 345 g_scanner_key_hash (gconstpointer v)
 346 {
 347   register const GScannerKey *key = v;
 348   register gchar *c;
 349   register guint h;
 350
 351   h = key->scope_id;
 352   for (c = key->symbol; *c; c++)
 353     {
 354       register guint g;
 355
 356       h = (h << 4) + *c;
 357       g = h & 0xf0000000;
 358       if (g)
 359         {
 360           h = h ^ (g >> 24);
 361           h = h ^ g;
 362         }
 363     }
 364
 365   return h;
 366 }
 367
 368 static inline GScannerKey*
 369 g_scanner_lookup_internal (GScanner     *scanner,
 370                            guint         scope_id,
 371                            const gchar  *symbol)
 372 {
 373   register GScannerKey  *key_p;
 374   GScannerKey key;
 375
 376   key.scope_id = scope_id;
 377
 378   if (!scanner->config->case_sensitive)
 379     {
 380       register gchar *d;
 381       register const gchar *c;
 382
 383       key.symbol = g_new (gchar, strlen (symbol) + 1);
 384       for (d = key.symbol, c = symbol; *c; c++, d++)
 385         *d = to_lower (*c);
 386       *d = 0;
 387       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 388       g_free (key.symbol);
 389     }
 390   else
 391     {
 392       key.symbol = (gchar*) symbol;
 393       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 394     }
 395
 396   return key_p;
 397 }
 398
 399 void
 400 g_scanner_scope_add_symbol (GScanner    *scanner,
 401                             guint        scope_id,
 402                             const gchar *symbol,
 403                             gpointer     value)
 404 {
 405   register GScannerKey  *key;
 406
 407   g_return_if_fail (scanner != NULL);
 408   g_return_if_fail (symbol != NULL);
 409
 410   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 411
 412   if (!key)
 413     {
 414       key = g_new (GScannerKey, 1);
 415       key->scope_id = scope_id;
 416       key->symbol = g_strdup (symbol);
 417       key->value = value;
 418       if (!scanner->config->case_sensitive)
 419         {
 420           register gchar *c;
 421
 422           c = key->symbol;
 423           while (*c != 0)
 424             {
 425               *c = to_lower (*c);
 426               c++;
 427             }
 428         }
 429       g_hash_table_insert (scanner->symbol_table, key, key);
 430     }
 431   else
 432     key->value = value;
 433 }
 434
 435 void
 436 g_scanner_scope_remove_symbol (GScanner    *scanner,
 437                                guint        scope_id,
 438                                const gchar *symbol)
 439 {
 440   register GScannerKey  *key;
 441
 442   g_return_if_fail (scanner != NULL);
 443   g_return_if_fail (symbol != NULL);
 444
 445   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 446
 447   if (key)
 448     {
 449       g_hash_table_remove (scanner->symbol_table, key);
 450       g_free (key->symbol);
 451       g_free (key);
 452     }
 453 }
 454
 455 gpointer
 456 g_scanner_lookup_symbol (GScanner       *scanner,
 457                          const gchar    *symbol)
 458 {
 459   register GScannerKey  *key;
 460   register guint scope_id;
 461
 462   g_return_val_if_fail (scanner != NULL, NULL);
 463
 464   if (!symbol)
 465     return NULL;
 466
 467   scope_id = scanner->scope_id;
 468   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 469   if (!key && scope_id && scanner->config->scope_0_fallback)
 470     key = g_scanner_lookup_internal (scanner, 0, symbol);
 471
 472   if (key)
 473     return key->value;
 474   else
 475     return NULL;
 476 }
 477
 478 gpointer
 479 g_scanner_scope_lookup_symbol (GScanner       *scanner,
 480                                guint           scope_id,
 481                                const gchar    *symbol)
 482 {
 483   register GScannerKey  *key;
 484
 485   g_return_val_if_fail (scanner != NULL, NULL);
 486
 487   if (!symbol)
 488     return NULL;
 489
 490   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 491
 492   if (key)
 493     return key->value;
 494   else
 495     return NULL;
 496 }
 497
 498 guint
 499 g_scanner_set_scope (GScanner       *scanner,
 500                      guint           scope_id)
 501 {
 502   register guint old_scope_id;
 503
 504   g_return_val_if_fail (scanner != NULL, 0);
 505
 506   old_scope_id = scanner->scope_id;
 507   scanner->scope_id = scope_id;
 508
 509   return old_scope_id;
 510 }
 511
 512 static void
 513 g_scanner_foreach_internal (gpointer  _key,
 514                             gpointer  _value,
 515                             gpointer  _user_data)
 516 {
 517   register GScannerKey *key;
 518   register gpointer *d;
 519   register GHFunc func;
 520   register gpointer func_data;
 521   register guint *scope_id;
 522
 523   d = _user_data;
 524   func = (GHFunc) d[0];
 525   func_data = d[1];
 526   scope_id = d[2];
 527   key = _value;
 528
 529   if (key->scope_id == *scope_id)
 530     func (key->symbol, key->value, func_data);
 531 }
 532
 533 void
 534 g_scanner_scope_foreach_symbol (GScanner       *scanner,
 535                                 guint           scope_id,
 536                                 GHFunc          func,
 537                                 gpointer        func_data)
 538 {
 539   gpointer d[3];
 540
 541   g_return_if_fail (scanner != NULL);
 542
 543   d[0] = (gpointer) func;
 544   d[1] = func_data;
 545   d[2] = &scope_id;
 546
 547   g_hash_table_foreach (scanner->symbol_table, g_scanner_foreach_internal, d);
 548 }
 549
 550 void
 551 g_scanner_freeze_symbol_table (GScanner *scanner)
 552 {
 553   g_return_if_fail (scanner != NULL);
 554
 555   g_hash_table_freeze (scanner->symbol_table);
 556 }
 557
 558 void
 559 g_scanner_thaw_symbol_table (GScanner *scanner)
 560 {
 561   g_return_if_fail (scanner != NULL);
 562
 563   g_hash_table_thaw (scanner->symbol_table);
 564 }
 565
 566 GTokenType
 567 g_scanner_peek_next_token (GScanner     *scanner)
 568 {
 569   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 570
 571   if (scanner->next_token == G_TOKEN_NONE)
 572     {
 573       scanner->next_line = scanner->line;
 574       scanner->next_position = scanner->position;
 575       g_scanner_get_token_i (scanner,
 576                              &scanner->next_token,
 577                              &scanner->next_value,
 578                              &scanner->next_line,
 579                              &scanner->next_position);
 580     }
 581
 582   return scanner->next_token;
 583 }
 584
 585 GTokenType
 586 g_scanner_get_next_token (GScanner      *scanner)
 587 {
 588   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 589
 590   if (scanner->next_token != G_TOKEN_NONE)
 591     {
 592       g_scanner_free_value (&scanner->token, &scanner->value);
 593
 594       scanner->token = scanner->next_token;
 595       scanner->value = scanner->next_value;
 596       scanner->line = scanner->next_line;
 597       scanner->position = scanner->next_position;
 598       scanner->next_token = G_TOKEN_NONE;
 599     }
 600   else
 601     g_scanner_get_token_i (scanner,
 602                            &scanner->token,
 603                            &scanner->value,
 604                            &scanner->line,
 605                            &scanner->position);
 606
 607   return scanner->token;
 608 }
 609
 610 GTokenType
 611 g_scanner_cur_token (GScanner *scanner)
 612 {
 613   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 614
 615   return scanner->token;
 616 }
 617
 618 GTokenValue
 619 g_scanner_cur_value (GScanner *scanner)
 620 {
 621   GTokenValue v;
 622
 623   g_return_val_if_fail (scanner != NULL, v);
 624
 625   /* MSC isn't capable of handling return scanner->value; ? */
 626
 627   v.v_int = 0;
 628
 629   v = scanner->value;
 630
 631   return v;
 632 }
 633
 634 guint
 635 g_scanner_cur_line (GScanner *scanner)
 636 {
 637   g_return_val_if_fail (scanner != NULL, 0);
 638
 639   return scanner->line;
 640 }
 641
 642 guint
 643 g_scanner_cur_position (GScanner *scanner)
 644 {
 645   g_return_val_if_fail (scanner != NULL, 0);
 646
 647   return scanner->position;
 648 }
 649
 650 gboolean
 651 g_scanner_eof (GScanner *scanner)
 652 {
 653   g_return_val_if_fail (scanner != NULL, TRUE);
 654
 655   return scanner->token == G_TOKEN_EOF;
 656 }
 657
 658 void
 659 g_scanner_input_file (GScanner *scanner,
 660                       gint      input_fd)
 661 {
 662   g_return_if_fail (scanner != NULL);
 663   g_return_if_fail (input_fd >= 0);
 664
 665   if (scanner->input_fd >= 0)
 666     g_scanner_sync_file_offset (scanner);
 667
 668   scanner->token = G_TOKEN_NONE;
 669   scanner->value.v_int = 0;
 670   scanner->line = 1;
 671   scanner->position = 0;
 672   scanner->next_token = G_TOKEN_NONE;
 673
 674   scanner->input_fd = input_fd;
 675   scanner->text = NULL;
 676   scanner->text_end = NULL;
 677
 678   if (!scanner->buffer)
 679     scanner->buffer = g_new (gchar, READ_BUFFER_SIZE + 1);
 680 }
 681
 682 void
 683 g_scanner_input_text (GScanner    *scanner,
 684                       const gchar *text,
 685                       guint        text_len)
 686 {
 687   g_return_if_fail (scanner != NULL);
 688   if (text_len)
 689     g_return_if_fail (text != NULL);
 690   else
 691     text = NULL;
 692
 693   if (scanner->input_fd >= 0)
 694     g_scanner_sync_file_offset (scanner);
 695
 696   scanner->token = G_TOKEN_NONE;
 697   scanner->value.v_int = 0;
 698   scanner->line = 1;
 699   scanner->position = 0;
 700   scanner->next_token = G_TOKEN_NONE;
 701
 702   scanner->input_fd = -1;
 703   scanner->text = text;
 704   scanner->text_end = text + text_len;
 705
 706   if (scanner->buffer)
 707     {
 708       g_free (scanner->buffer);
 709       scanner->buffer = NULL;
 710     }
 711 }
 712
 713 static guchar
 714 g_scanner_peek_next_char (GScanner *scanner)
 715 {
 716   if (scanner->text < scanner->text_end)
 717     {
 718       return *scanner->text;
 719     }
 720   else if (scanner->input_fd >= 0)
 721     {
 722       register gint count;
 723       register gchar *buffer;
 724
 725       buffer = scanner->buffer;
 726       do
 727         {
 728           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 729         }
 730       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 731
 732       if (count < 1)
 733         {
 734           scanner->input_fd = -1;
 735
 736           return 0;
 737         }
 738       else
 739         {
 740           scanner->text = buffer;
 741           scanner->text_end = buffer + count;
 742
 743           return *buffer;
 744         }
 745     }
 746   else
 747     return 0;
 748 }
 749
 750 void
 751 g_scanner_sync_file_offset (GScanner *scanner)
 752 {
 753   g_return_if_fail (scanner != NULL);
 754
 755   /* for file input, rewind the filedescriptor to the current
 756    * buffer position and blow the file read ahead buffer. usefull for
 757    * third party uses of our filedescriptor, which hooks onto the current
 758    * scanning position.
 759    */
 760
 761   if (scanner->input_fd >= 0 && scanner->text_end > scanner->text)
 762     {
 763       gint buffered;
 764
 765       buffered = scanner->text_end - scanner->text;
 766       if (lseek (scanner->input_fd, - buffered, SEEK_CUR) >= 0)
 767         {
 768           /* we succeeded, blow our buffer's contents now */
 769           scanner->text = NULL;
 770           scanner->text_end = NULL;
 771         }
 772       else
 773         errno = 0;
 774     }
 775 }
 776
 777 static guchar
 778 g_scanner_get_char (GScanner    *scanner,
 779                     guint       *line_p,
 780                     guint       *position_p)
 781 {
 782   register guchar fchar;
 783
 784   if (scanner->text < scanner->text_end)
 785     fchar = *(scanner->text++);
 786   else if (scanner->input_fd >= 0)
 787     {
 788       register gint count;
 789       register gchar *buffer;
 790
 791       buffer = scanner->buffer;
 792       do
 793         {
 794           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 795         }
 796       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 797
 798       if (count < 1)
 799         {
 800           scanner->input_fd = -1;
 801           fchar = 0;
 802         }
 803       else
 804         {
 805           scanner->text = buffer + 1;
 806           scanner->text_end = buffer + count;
 807           fchar = *buffer;
 808           if (!fchar)
 809             {
 810               g_scanner_sync_file_offset (scanner);
 811               scanner->text_end = scanner->text;
 812               scanner->input_fd = -1;
 813             }
 814         }
 815     }
 816   else
 817     fchar = 0;
 818
 819   if (fchar == '\n')
 820     {
 821       (*position_p) = 0;
 822       (*line_p)++;
 823     }
 824   else if (fchar)
 825     {
 826       (*position_p)++;
 827     }
 828
 829   return fchar;
 830 }
 831
 832 void
 833 g_scanner_unexp_token (GScanner         *scanner,
 834                        GTokenType        expected_token,
 835                        const gchar      *identifier_spec,
 836                        const gchar      *symbol_spec,
 837                        const gchar      *symbol_name,
 838                        const gchar      *message,
 839                        gint              is_error)
 840 {
 841   register gchar        *token_string;
 842   register guint        token_string_len;
 843   register gchar        *expected_string;
 844   register guint        expected_string_len;
 845   register gchar        *message_prefix;
 846   register gboolean     print_unexp;
 847   void (*msg_handler)   (GScanner*, const gchar*, ...);
 848
 849   g_return_if_fail (scanner != NULL);
 850
 851   if (is_error)
 852     msg_handler = g_scanner_error;
 853   else
 854     msg_handler = g_scanner_warn;
 855
 856   if (!identifier_spec)
 857     identifier_spec = "identifier";
 858   if (!symbol_spec)
 859     symbol_spec = "symbol";
 860
 861   token_string_len = 56;
 862   token_string = g_new (gchar, token_string_len + 1);
 863   expected_string_len = 64;
 864   expected_string = g_new (gchar, expected_string_len + 1);
 865   print_unexp = TRUE;
 866
 867   switch (scanner->token)
 868     {
 869     case  G_TOKEN_EOF:
 870       g_snprintf (token_string, token_string_len, "end of file");
 871       break;
 872
 873     default:
 874       if (scanner->token >= 1 && scanner->token <= 255)
 875         {
 876           if ((scanner->token >= ' ' && scanner->token <= '~') ||
 877               strchr (scanner->config->cset_identifier_first, scanner->token) ||
 878               strchr (scanner->config->cset_identifier_nth, scanner->token))
 879             g_snprintf (token_string, expected_string_len, "character `%c'", scanner->token);
 880           else
 881             g_snprintf (token_string, expected_string_len, "character `\\%o'", scanner->token);
 882           break;
 883         }
 884       else if (!scanner->config->symbol_2_token)
 885         {
 886           g_snprintf (token_string, token_string_len, "(unknown) token <%d>", scanner->token);
 887           break;
 888         }
 889       /* fall through */
 890     case  G_TOKEN_SYMBOL:
 891       if (expected_token == G_TOKEN_SYMBOL ||
 892           (scanner->config->symbol_2_token &&
 893            expected_token > G_TOKEN_LAST))
 894         print_unexp = FALSE;
 895       if (symbol_name)
 896         g_snprintf (token_string,
 897                     token_string_len,
 898                     "%s%s `%s'",
 899                     print_unexp ? "" : "invalid ",
 900                     symbol_spec,
 901                     symbol_name);
 902       else
 903         g_snprintf (token_string,
 904                     token_string_len,
 905                     "%s%s",
 906                     print_unexp ? "" : "invalid ",
 907                     symbol_spec);
 908       break;
 909
 910     case  G_TOKEN_ERROR:
 911       print_unexp = FALSE;
 912       expected_token = G_TOKEN_NONE;
 913       switch (scanner->value.v_error)
 914         {
 915         case  G_ERR_UNEXP_EOF:
 916           g_snprintf (token_string, token_string_len, "scanner: unexpected end of file");
 917           break;
 918
 919         case  G_ERR_UNEXP_EOF_IN_STRING:
 920           g_snprintf (token_string, token_string_len, "scanner: unterminated string constant");
 921           break;
 922
 923         case  G_ERR_UNEXP_EOF_IN_COMMENT:
 924           g_snprintf (token_string, token_string_len, "scanner: unterminated comment");
 925           break;
 926
 927         case  G_ERR_NON_DIGIT_IN_CONST:
 928           g_snprintf (token_string, token_string_len, "scanner: non digit in constant");
 929           break;
 930
 931         case  G_ERR_FLOAT_RADIX:
 932           g_snprintf (token_string, token_string_len, "scanner: invalid radix for floating constant");
 933           break;
 934
 935         case  G_ERR_FLOAT_MALFORMED:
 936           g_snprintf (token_string, token_string_len, "scanner: malformed floating constant");
 937           break;
 938
 939         case  G_ERR_DIGIT_RADIX:
 940           g_snprintf (token_string, token_string_len, "scanner: digit is beyond radix");
 941           break;
 942
 943         case  G_ERR_UNKNOWN:
 944         default:
 945           g_snprintf (token_string, token_string_len, "scanner: unknown error");
 946           break;
 947         }
 948       break;
 949
 950     case  G_TOKEN_CHAR:
 951       g_snprintf (token_string, token_string_len, "character `%c'", scanner->value.v_char);
 952       break;
 953
 954     case  G_TOKEN_IDENTIFIER:
 955       if (expected_token == G_TOKEN_IDENTIFIER)
 956         print_unexp = FALSE;
 957       g_snprintf (token_string,
 958                   token_string_len,
 959                   "%s%s `%s'",
 960                   print_unexp ? "" : "invalid ",
 961                   identifier_spec,
 962                   scanner->value.v_string);
 963       break;
 964
 965     case  G_TOKEN_BINARY:
 966     case  G_TOKEN_OCTAL:
 967     case  G_TOKEN_INT:
 968     case  G_TOKEN_HEX:
 969       g_snprintf (token_string, token_string_len, "number `%ld'", scanner->value.v_int);
 970       break;
 971
 972     case  G_TOKEN_FLOAT:
 973       g_snprintf (token_string, token_string_len, "number `%.3f'", scanner->value.v_float);
 974       break;
 975
 976     case  G_TOKEN_STRING:
 977       if (expected_token == G_TOKEN_STRING)
 978         print_unexp = FALSE;
 979       g_snprintf (token_string,
 980                   token_string_len,
 981                   "%s%sstring constant \"%s\"",
 982                   print_unexp ? "" : "invalid ",
 983                   scanner->value.v_string[0] == 0 ? "empty " : "",
 984                   scanner->value.v_string);
 985       token_string[token_string_len - 2] = '"';
 986       token_string[token_string_len - 1] = 0;
 987       break;
 988
 989     case  G_TOKEN_COMMENT_SINGLE:
 990     case  G_TOKEN_COMMENT_MULTI:
 991       g_snprintf (token_string, token_string_len, "comment");
 992       break;
 993
 994     case  G_TOKEN_NONE:
 995       /* somehow the user's parsing code is screwed, there isn't much
 996        * we can do about it.
 997        * Note, a common case to trigger this is
 998        * g_scanner_peek_next_token(); g_scanner_unexp_token();
 999        * without an intermediate g_scanner_get_next_token().
1000        */
1001       g_assert_not_reached ();
1002       break;
1003     }
1004
1005
1006   switch (expected_token)
1007     {
1008       gboolean need_valid;
1009
1010     default:
1011       if (expected_token >= 1 && expected_token <= 255)
1012         {
1013           if ((expected_token >= ' ' && expected_token <= '~') ||
1014               strchr (scanner->config->cset_identifier_first, expected_token) ||
1015               strchr (scanner->config->cset_identifier_nth, expected_token))
1016             g_snprintf (expected_string, expected_string_len, "character `%c'", expected_token);
1017           else
1018             g_snprintf (expected_string, expected_string_len, "character `\\%o'", expected_token);
1019           break;
1020         }
1021       else if (!scanner->config->symbol_2_token)
1022         {
1023           g_snprintf (expected_string, expected_string_len, "(unknown) token <%d>", expected_token);
1024           break;
1025         }
1026       /* fall through */
1027     case  G_TOKEN_SYMBOL:
1028       need_valid = (scanner->token == G_TOKEN_SYMBOL ||
1029                     (scanner->config->symbol_2_token &&
1030                      scanner->token > G_TOKEN_LAST));
1031       g_snprintf (expected_string,
1032                   expected_string_len,
1033                   "%s%s",
1034                   need_valid ? "valid " : "",
1035                   symbol_spec);
1036       /* FIXME: should we attempt to lookup the symbol_name for symbol_2_token? */
1037       break;
1038
1039     case  G_TOKEN_INT:
1040       g_snprintf (expected_string, expected_string_len, "number (integer)");
1041       break;
1042
1043     case  G_TOKEN_FLOAT:
1044       g_snprintf (expected_string, expected_string_len, "number (float)");
1045       break;
1046
1047     case  G_TOKEN_STRING:
1048       g_snprintf (expected_string,
1049                   expected_string_len,
1050                   "%sstring constant",
1051                   scanner->token == G_TOKEN_STRING ? "valid " : "");
1052       break;
1053
1054     case  G_TOKEN_IDENTIFIER:
1055       g_snprintf (expected_string,
1056                   expected_string_len,
1057                   "%s%s",
1058                   scanner->token == G_TOKEN_IDENTIFIER ? "valid " : "",
1059                   identifier_spec);
1060       break;
1061
1062     case  G_TOKEN_NONE:
1063       break;
1064     }
1065
1066   if (message && message[0] != 0)
1067     message_prefix = " - ";
1068   else
1069     {
1070       message_prefix = "";
1071       message = "";
1072     }
1073
1074   if (expected_token != G_TOKEN_NONE)
1075     {
1076       if (print_unexp)
1077         msg_handler (scanner,
1078                      "unexpected %s, expected %s%s%s",
1079                      token_string,
1080                      expected_string,
1081                      message_prefix,
1082                      message);
1083       else
1084         msg_handler (scanner,
1085                      "%s, expected %s%s%s",
1086                      token_string,
1087                      expected_string,
1088                      message_prefix,
1089                      message);
1090     }
1091   else
1092     {
1093       if (print_unexp)
1094         msg_handler (scanner,
1095                      "unexpected %s%s%s",
1096                      token_string,
1097                      message_prefix,
1098                      message);
1099       else
1100         msg_handler (scanner,
1101                      "%s%s%s",
1102                      token_string,
1103                      message_prefix,
1104                      message);
1105     }
1106
1107   g_free (token_string);
1108   g_free (expected_string);
1109 }
1110
1111 gint
1112 g_scanner_stat_mode (const gchar *filename)
1113 {
1114   struct stat  *stat_buf;
1115   gint          st_mode;
1116
1117   stat_buf = g_new0 (struct stat, 1);
1118 #ifdef HAVE_LSTAT
1119   lstat (filename, stat_buf);
1120 #else
1121   stat (filename, stat_buf);
1122 #endif
1123   st_mode = stat_buf->st_mode;
1124
1125   g_free (stat_buf);
1126
1127   return st_mode;
1128 }
1129
1130 static void
1131 g_scanner_get_token_i (GScanner *scanner,
1132                        GTokenType       *token_p,
1133                        GTokenValue      *value_p,
1134                        guint            *line_p,
1135                        guint            *position_p)
1136 {
1137   do
1138     {
1139       g_scanner_free_value (token_p, value_p);
1140       g_scanner_get_token_ll (scanner, token_p, value_p, line_p, position_p);
1141     }
1142   while (((*token_p > 0 && *token_p < 256) &&
1143           strchr (scanner->config->cset_skip_characters, *token_p)) ||
1144          (*token_p == G_TOKEN_CHAR &&
1145           strchr (scanner->config->cset_skip_characters, value_p->v_char)) ||
1146          (*token_p == G_TOKEN_COMMENT_MULTI &&
1147           scanner->config->skip_comment_multi) ||
1148          (*token_p == G_TOKEN_COMMENT_SINGLE &&
1149           scanner->config->skip_comment_single));
1150
1151   switch (*token_p)
1152     {
1153     case G_TOKEN_IDENTIFIER:
1154       if (scanner->config->identifier_2_string)
1155         *token_p = G_TOKEN_STRING;
1156       break;
1157
1158     case G_TOKEN_SYMBOL:
1159       if (scanner->config->symbol_2_token)
1160         *token_p = (GTokenType) value_p->v_symbol;
1161       break;
1162
1163     case G_TOKEN_BINARY:
1164     case G_TOKEN_OCTAL:
1165     case G_TOKEN_HEX:
1166       if (scanner->config->numbers_2_int)
1167         *token_p = G_TOKEN_INT;
1168       break;
1169
1170     default:
1171       break;
1172     }
1173
1174   if (*token_p == G_TOKEN_INT &&
1175       scanner->config->int_2_float)
1176     {
1177       *token_p = G_TOKEN_FLOAT;
1178       value_p->v_float = value_p->v_int;
1179     }
1180
1181   errno = 0;
1182 }
1183
1184 static void
1185 g_scanner_get_token_ll  (GScanner       *scanner,
1186                          GTokenType     *token_p,
1187                          GTokenValue    *value_p,
1188                          guint          *line_p,
1189                          guint          *position_p)
1190 {
1191   register GScannerConfig *config;
1192   register GTokenType      token;
1193   register gboolean        in_comment_multi;
1194   register gboolean        in_comment_single;
1195   register gboolean        in_string_sq;
1196   register gboolean        in_string_dq;
1197   register GString        *gstring;
1198   GTokenValue              value;
1199   guchar                   ch;
1200
1201   config = scanner->config;
1202   (*value_p).v_int = 0;
1203
1204   if ((scanner->text >= scanner->text_end && scanner->input_fd < 0) ||
1205       scanner->token == G_TOKEN_EOF)
1206     {
1207       *token_p = G_TOKEN_EOF;
1208       return;
1209     }
1210
1211   in_comment_multi = FALSE;
1212   in_comment_single = FALSE;
1213   in_string_sq = FALSE;
1214   in_string_dq = FALSE;
1215   gstring = NULL;
1216
1217   do /* while (ch != 0) */
1218     {
1219       register gboolean         dotted_float = FALSE;
1220
1221       ch = g_scanner_get_char (scanner, line_p, position_p);
1222
1223       value.v_int = 0;
1224       token = G_TOKEN_NONE;
1225
1226       /* this is *evil*, but needed ;(
1227        * we first check for identifier first character, because  it
1228        * might interfere with other key chars like slashes or numbers
1229        */
1230       if (config->scan_identifier &&
1231           ch && strchr (config->cset_identifier_first, ch))
1232         goto identifier_precedence;
1233
1234       switch (ch)
1235         {
1236         case  0:
1237           token = G_TOKEN_EOF;
1238           (*position_p)++;
1239           /* ch = 0; */
1240           break;
1241
1242         case  '/':
1243           if (!config->scan_comment_multi ||
1244               g_scanner_peek_next_char (scanner) != '*')
1245             goto default_case;
1246           g_scanner_get_char (scanner, line_p, position_p);
1247           token = G_TOKEN_COMMENT_MULTI;
1248           in_comment_multi = TRUE;
1249           gstring = g_string_new ("");
1250           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1251             {
1252               if (ch == '*' && g_scanner_peek_next_char (scanner) == '/')
1253                 {
1254                   g_scanner_get_char (scanner, line_p, position_p);
1255                   in_comment_multi = FALSE;
1256                   break;
1257                 }
1258               else
1259                 gstring = g_string_append_c (gstring, ch);
1260             }
1261           ch = 0;
1262           break;
1263
1264         case  '\'':
1265           if (!config->scan_string_sq)
1266             goto default_case;
1267           token = G_TOKEN_STRING;
1268           in_string_sq = TRUE;
1269           gstring = g_string_new ("");
1270           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1271             {
1272               if (ch == '\'')
1273                 {
1274                   in_string_sq = FALSE;
1275                   break;
1276                 }
1277               else
1278                 gstring = g_string_append_c (gstring, ch);
1279             }
1280           ch = 0;
1281           break;
1282
1283         case  '"':
1284           if (!config->scan_string_dq)
1285             goto default_case;
1286           token = G_TOKEN_STRING;
1287           in_string_dq = TRUE;
1288           gstring = g_string_new ("");
1289           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1290             {
1291               if (ch == '"')
1292                 {
1293                   in_string_dq = FALSE;
1294                   break;
1295                 }
1296               else
1297                 {
1298                   if (ch == '\\')
1299                     {
1300                       ch = g_scanner_get_char (scanner, line_p, position_p);
1301                       switch (ch)
1302                         {
1303                           register guint        i;
1304                           register guint        fchar;
1305
1306                         case  0:
1307                           break;
1308
1309                         case  '\\':
1310                           gstring = g_string_append_c (gstring, '\\');
1311                           break;
1312
1313                         case  'n':
1314                           gstring = g_string_append_c (gstring, '\n');
1315                           break;
1316
1317                         case  't':
1318                           gstring = g_string_append_c (gstring, '\t');
1319                           break;
1320
1321                         case  'r':
1322                           gstring = g_string_append_c (gstring, '\r');
1323                           break;
1324
1325                         case  'b':
1326                           gstring = g_string_append_c (gstring, '\b');
1327                           break;
1328
1329                         case  'f':
1330                           gstring = g_string_append_c (gstring, '\f');
1331                           break;
1332
1333                         case  '0':
1334                         case  '1':
1335                         case  '2':
1336                         case  '3':
1337                         case  '4':
1338                         case  '5':
1339                         case  '6':
1340                         case  '7':
1341                           i = ch - '0';
1342                           fchar = g_scanner_peek_next_char (scanner);
1343                           if (fchar >= '0' && fchar <= '7')
1344                             {
1345                               ch = g_scanner_get_char (scanner, line_p, position_p);
1346                               i = i * 8 + ch - '0';
1347                               fchar = g_scanner_peek_next_char (scanner);
1348                               if (fchar >= '0' && fchar <= '7')
1349                                 {
1350                                   ch = g_scanner_get_char (scanner, line_p, position_p);
1351                                   i = i * 8 + ch - '0';
1352                                 }
1353                             }
1354                           gstring = g_string_append_c (gstring, i);
1355                           break;
1356
1357                         default:
1358                           gstring = g_string_append_c (gstring, ch);
1359                           break;
1360                         }
1361                     }
1362                   else
1363                     gstring = g_string_append_c (gstring, ch);
1364                 }
1365             }
1366           ch = 0;
1367           break;
1368
1369         case  '.':
1370           if (!config->scan_float)
1371             goto default_case;
1372           token = G_TOKEN_FLOAT;
1373           dotted_float = TRUE;
1374           ch = g_scanner_get_char (scanner, line_p, position_p);
1375           goto number_parsing;
1376
1377         case  '$':
1378           if (!config->scan_hex_dollar)
1379             goto default_case;
1380           token = G_TOKEN_HEX;
1381           ch = g_scanner_get_char (scanner, line_p, position_p);
1382           goto number_parsing;
1383
1384         case  '0':
1385           if (config->scan_octal)
1386             token = G_TOKEN_OCTAL;
1387           else
1388             token = G_TOKEN_INT;
1389           ch = g_scanner_peek_next_char (scanner);
1390           if (config->scan_hex && (ch == 'x' || ch == 'X'))
1391             {
1392               token = G_TOKEN_HEX;
1393               g_scanner_get_char (scanner, line_p, position_p);
1394               ch = g_scanner_get_char (scanner, line_p, position_p);
1395               if (ch == 0)
1396                 {
1397                   token = G_TOKEN_ERROR;
1398                   value.v_error = G_ERR_UNEXP_EOF;
1399                   (*position_p)++;
1400                   break;
1401                 }
1402               if (g_scanner_char_2_num (ch, 16) < 0)
1403                 {
1404                   token = G_TOKEN_ERROR;
1405                   value.v_error = G_ERR_DIGIT_RADIX;
1406                   ch = 0;
1407                   break;
1408                 }
1409             }
1410           else if (config->scan_binary && (ch == 'b' || ch == 'B'))
1411             {
1412               token = G_TOKEN_BINARY;
1413               g_scanner_get_char (scanner, line_p, position_p);
1414               ch = g_scanner_get_char (scanner, line_p, position_p);
1415               if (ch == 0)
1416                 {
1417                   token = G_TOKEN_ERROR;
1418                   value.v_error = G_ERR_UNEXP_EOF;
1419                   (*position_p)++;
1420                   break;
1421                 }
1422               if (g_scanner_char_2_num (ch, 10) < 0)
1423                 {
1424                   token = G_TOKEN_ERROR;
1425                   value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1426                   ch = 0;
1427                   break;
1428                 }
1429             }
1430           else
1431             ch = '0';
1432           /* fall through */
1433         case  '1':
1434         case  '2':
1435         case  '3':
1436         case  '4':
1437         case  '5':
1438         case  '6':
1439         case  '7':
1440         case  '8':
1441         case  '9':
1442         number_parsing:
1443         {
1444           register gboolean in_number = TRUE;
1445           gchar *endptr;
1446
1447           if (token == G_TOKEN_NONE)
1448             token = G_TOKEN_INT;
1449
1450           gstring = g_string_new (dotted_float ? "0." : "");
1451           gstring = g_string_append_c (gstring, ch);
1452
1453           do /* while (in_number) */
1454             {
1455               register gboolean is_E;
1456
1457               is_E = token == G_TOKEN_FLOAT && (ch == 'e' || ch == 'E');
1458
1459               ch = g_scanner_peek_next_char (scanner);
1460
1461               if (g_scanner_char_2_num (ch, 36) >= 0 ||
1462                   (config->scan_float && ch == '.') ||
1463                   (is_E && (ch == '+' || ch == '-')))
1464                 {
1465                   ch = g_scanner_get_char (scanner, line_p, position_p);
1466
1467                   switch (ch)
1468                     {
1469                     case '.':
1470                       if (token != G_TOKEN_INT && token != G_TOKEN_OCTAL)
1471                         {
1472                           value.v_error = token == G_TOKEN_FLOAT ? G_ERR_FLOAT_MALFORMED : G_ERR_FLOAT_RADIX;
1473                           token = G_TOKEN_ERROR;
1474                           in_number = FALSE;
1475                         }
1476                       else
1477                         {
1478                           token = G_TOKEN_FLOAT;
1479                           gstring = g_string_append_c (gstring, ch);
1480                         }
1481                       break;
1482
1483                     case '0':
1484                     case '1':
1485                     case '2':
1486                     case '3':
1487                     case '4':
1488                     case '5':
1489                     case '6':
1490                     case '7':
1491                     case '8':
1492                     case '9':
1493                       gstring = g_string_append_c (gstring, ch);
1494                       break;
1495
1496                     case '-':
1497                     case '+':
1498                       if (token != G_TOKEN_FLOAT)
1499                         {
1500                           token = G_TOKEN_ERROR;
1501                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1502                           in_number = FALSE;
1503                         }
1504                       else
1505                         gstring = g_string_append_c (gstring, ch);
1506                       break;
1507
1508                     case 'e':
1509                     case 'E':
1510                       if ((token != G_TOKEN_HEX && !config->scan_float) ||
1511                           (token != G_TOKEN_HEX &&
1512                            token != G_TOKEN_OCTAL &&
1513                            token != G_TOKEN_FLOAT &&
1514                            token != G_TOKEN_INT))
1515                         {
1516                           token = G_TOKEN_ERROR;
1517                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1518                           in_number = FALSE;
1519                         }
1520                       else
1521                         {
1522                           if (token != G_TOKEN_HEX)
1523                             token = G_TOKEN_FLOAT;
1524                           gstring = g_string_append_c (gstring, ch);
1525                         }
1526                       break;
1527
1528                     default:
1529                       if (token != G_TOKEN_HEX)
1530                         {
1531                           token = G_TOKEN_ERROR;
1532                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1533                           in_number = FALSE;
1534                         }
1535                       else
1536                         gstring = g_string_append_c (gstring, ch);
1537                       break;
1538                     }
1539                 }
1540               else
1541                 in_number = FALSE;
1542             }
1543           while (in_number);
1544
1545           endptr = NULL;
1546           switch (token)
1547             {
1548             case G_TOKEN_BINARY:
1549               value.v_binary = strtol (gstring->str, &endptr, 2);
1550               break;
1551
1552             case G_TOKEN_OCTAL:
1553               value.v_octal = strtol (gstring->str, &endptr, 8);
1554               break;
1555
1556             case G_TOKEN_INT:
1557               value.v_int = strtol (gstring->str, &endptr, 10);
1558               break;
1559
1560             case G_TOKEN_FLOAT:
1561               value.v_float = g_strtod (gstring->str, &endptr);
1562               break;
1563
1564             case G_TOKEN_HEX:
1565               value.v_hex = strtol (gstring->str, &endptr, 16);
1566               break;
1567
1568             default:
1569               break;
1570             }
1571           if (endptr && *endptr)
1572             {
1573               token = G_TOKEN_ERROR;
1574               if (*endptr == 'e' || *endptr == 'E')
1575                 value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1576               else
1577                 value.v_error = G_ERR_DIGIT_RADIX;
1578             }
1579           g_string_free (gstring, TRUE);
1580           gstring = NULL;
1581           ch = 0;
1582         } /* number_parsing:... */
1583         break;
1584
1585         default:
1586         default_case:
1587         {
1588           if (config->cpair_comment_single &&
1589               ch == config->cpair_comment_single[0])
1590             {
1591               token = G_TOKEN_COMMENT_SINGLE;
1592               in_comment_single = TRUE;
1593               gstring = g_string_new ("");
1594               while ((ch = g_scanner_get_char (scanner,
1595                                                line_p,
1596                                                position_p)) != 0)
1597                 {
1598                   if (ch == config->cpair_comment_single[1])
1599                     {
1600                       in_comment_single = FALSE;
1601                       ch = 0;
1602                       break;
1603                     }
1604
1605                   gstring = g_string_append_c (gstring, ch);
1606                   ch = 0;
1607                 }
1608             }
1609           else if (config->scan_identifier && ch &&
1610                    strchr (config->cset_identifier_first, ch))
1611             {
1612             identifier_precedence:
1613
1614               if (config->cset_identifier_nth && ch &&
1615                   strchr (config->cset_identifier_nth,
1616                           g_scanner_peek_next_char (scanner)))
1617                 {
1618                   token = G_TOKEN_IDENTIFIER;
1619                   gstring = g_string_new ("");
1620                   gstring = g_string_append_c (gstring, ch);
1621                   do
1622                     {
1623                       ch = g_scanner_get_char (scanner, line_p, position_p);
1624                       gstring = g_string_append_c (gstring, ch);
1625                       ch = g_scanner_peek_next_char (scanner);
1626                     }
1627                   while (ch && strchr (config->cset_identifier_nth, ch));
1628                   ch = 0;
1629                 }
1630               else if (config->scan_identifier_1char)
1631                 {
1632                   token = G_TOKEN_IDENTIFIER;
1633                   value.v_identifier = g_new0 (gchar, 2);
1634                   value.v_identifier[0] = ch;
1635                   ch = 0;
1636                 }
1637             }
1638           if (ch)
1639             {
1640               if (config->char_2_token)
1641                 token = ch;
1642               else
1643                 {
1644                   token = G_TOKEN_CHAR;
1645                   value.v_char = ch;
1646                 }
1647               ch = 0;
1648             }
1649         } /* default_case:... */
1650         break;
1651         }
1652       g_assert (ch == 0 && token != G_TOKEN_NONE); /* paranoid */
1653     }
1654   while (ch != 0);
1655
1656   if (in_comment_multi || in_comment_single ||
1657       in_string_sq || in_string_dq)
1658     {
1659       token = G_TOKEN_ERROR;
1660       if (gstring)
1661         {
1662           g_string_free (gstring, TRUE);
1663           gstring = NULL;
1664         }
1665       (*position_p)++;
1666       if (in_comment_multi || in_comment_single)
1667         value.v_error = G_ERR_UNEXP_EOF_IN_COMMENT;
1668       else /* (in_string_sq || in_string_dq) */
1669         value.v_error = G_ERR_UNEXP_EOF_IN_STRING;
1670     }
1671
1672   if (gstring)
1673     {
1674       value.v_string = gstring->str;
1675       g_string_free (gstring, FALSE);
1676       gstring = NULL;
1677     }
1678
1679   if (token == G_TOKEN_IDENTIFIER)
1680     {
1681       if (config->scan_symbols)
1682         {
1683           register GScannerKey  *key;
1684           register guint scope_id;
1685
1686           scope_id = scanner->scope_id;
1687           key = g_scanner_lookup_internal (scanner, scope_id, value.v_identifier);
1688           if (!key && scope_id && scanner->config->scope_0_fallback)
1689             key = g_scanner_lookup_internal (scanner, 0, value.v_identifier);
1690
1691           if (key)
1692             {
1693               g_free (value.v_identifier);
1694               token = G_TOKEN_SYMBOL;
1695               value.v_symbol = key->value;
1696             }
1697         }
1698
1699       if (token == G_TOKEN_IDENTIFIER &&
1700           config->scan_identifier_NULL &&
1701           strlen (value.v_identifier) == 4)
1702         {
1703           gchar *null_upper = "NULL";
1704           gchar *null_lower = "null";
1705
1706           if (scanner->config->case_sensitive)
1707             {
1708               if (value.v_identifier[0] == null_upper[0] &&
1709                   value.v_identifier[1] == null_upper[1] &&
1710                   value.v_identifier[2] == null_upper[2] &&
1711                   value.v_identifier[3] == null_upper[3])
1712                 token = G_TOKEN_IDENTIFIER_NULL;
1713             }
1714           else
1715             {
1716               if ((value.v_identifier[0] == null_upper[0] ||
1717                    value.v_identifier[0] == null_lower[0]) &&
1718                   (value.v_identifier[1] == null_upper[1] ||
1719                    value.v_identifier[1] == null_lower[1]) &&
1720                   (value.v_identifier[2] == null_upper[2] ||
1721                    value.v_identifier[2] == null_lower[2]) &&
1722                   (value.v_identifier[3] == null_upper[3] ||
1723                    value.v_identifier[3] == null_lower[3]))
1724                 token = G_TOKEN_IDENTIFIER_NULL;
1725             }
1726         }
1727     }
1728
1729   *token_p = token;
1730   *value_p = value;
1731 }