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
  23 /*
  24  * Modified by the GLib Team and others 1997-1999.  See the AUTHORS
  25  * file for a list of people on the GLib Team.  See the ChangeLog
  26  * files for a list of changes.  These files are distributed with
  27  * GLib at ftp://ftp.gtk.org/pub/gtk/.
  28  */
  29
  30 /*
  31  * MT safe
  32  */
  33
  34 #define         __gscanner_c__
  35
  36 #ifdef HAVE_CONFIG_H
  37 #include <config.h>
  38 #endif
  39
  40 #include        <stdlib.h>
  41 #include        <stdarg.h>
  42 #include        <string.h>
  43 #include        <stdio.h>
  44 #include        "glib.h"
  45 #ifdef HAVE_UNISTD_H
  46 #include        <unistd.h>
  47 #endif
  48 #include        <errno.h>
  49 #include        <sys/types.h>   /* needed for sys/stat.h */
  50 #include        <sys/stat.h>
  51 #ifdef NATIVE_WIN32
  52 #include        <io.h>          /* For _read() */
  53 #endif
  54
  55 /* --- defines --- */
  56 #define to_lower(c)                             ( \
  57         (guchar) (                                                      \
  58           ( (((guchar)(c))>='A' && ((guchar)(c))<='Z') * ('a'-'A') ) |  \
  59           ( (((guchar)(c))>=192 && ((guchar)(c))<=214) * (224-192) ) |  \
  60           ( (((guchar)(c))>=216 && ((guchar)(c))<=222) * (248-216) ) |  \
  61           ((guchar)(c))                                                 \
  62         )                                                               \
  63 )
  64 #define READ_BUFFER_SIZE        (4000)
  65
  66
  67 /* --- typedefs --- */
  68 typedef struct  _GScannerKey    GScannerKey;
  69
  70 struct  _GScannerKey
  71 {
  72   guint          scope_id;
  73   gchar         *symbol;
  74   gpointer       value;
  75 };
  76
  77
  78
  79 /* --- variables --- */
  80 static GScannerConfig g_scanner_config_template =
  81 {
  82   (
  83    " \t\r\n"
  84    )                    /* cset_skip_characters */,
  85   (
  86    G_CSET_a_2_z
  87    "_"
  88    G_CSET_A_2_Z
  89    )                    /* cset_identifier_first */,
  90   (
  91    G_CSET_a_2_z
  92    "_0123456789"
  93    G_CSET_A_2_Z
  94    G_CSET_LATINS
  95    G_CSET_LATINC
  96    )                    /* cset_identifier_nth */,
  97   ( "#\n" )             /* cpair_comment_single */,
  98
  99   FALSE                 /* case_sensitive */,
 100
 101   TRUE                  /* skip_comment_multi */,
 102   TRUE                  /* skip_comment_single */,
 103   TRUE                  /* scan_comment_multi */,
 104   TRUE                  /* scan_identifier */,
 105   FALSE                 /* scan_identifier_1char */,
 106   FALSE                 /* scan_identifier_NULL */,
 107   TRUE                  /* scan_symbols */,
 108   FALSE                 /* scan_binary */,
 109   TRUE                  /* scan_octal */,
 110   TRUE                  /* scan_float */,
 111   TRUE                  /* scan_hex */,
 112   FALSE                 /* scan_hex_dollar */,
 113   TRUE                  /* scan_string_sq */,
 114   TRUE                  /* scan_string_dq */,
 115   TRUE                  /* numbers_2_int */,
 116   FALSE                 /* int_2_float */,
 117   FALSE                 /* identifier_2_string */,
 118   TRUE                  /* char_2_token */,
 119   FALSE                 /* symbol_2_token */,
 120   FALSE                 /* scope_0_fallback */,
 121 };
 122
 123
 124 /* --- prototypes --- */
 125 static inline
 126 GScannerKey*    g_scanner_lookup_internal (GScanner     *scanner,
 127                                            guint         scope_id,
 128                                            const gchar  *symbol);
 129 static gint     g_scanner_key_equal       (gconstpointer v1,
 130                                            gconstpointer v2);
 131 static guint    g_scanner_key_hash        (gconstpointer v);
 132 static void     g_scanner_get_token_ll    (GScanner     *scanner,
 133                                            GTokenType   *token_p,
 134                                            GTokenValue  *value_p,
 135                                            guint        *line_p,
 136                                            guint        *position_p);
 137 static void     g_scanner_get_token_i     (GScanner     *scanner,
 138                                            GTokenType   *token_p,
 139                                            GTokenValue  *value_p,
 140                                            guint        *line_p,
 141                                            guint        *position_p);
 142
 143 static guchar   g_scanner_peek_next_char  (GScanner     *scanner);
 144 static guchar   g_scanner_get_char        (GScanner     *scanner,
 145                                            guint        *line_p,
 146                                            guint        *position_p);
 147 static void     g_scanner_msg_handler     (GScanner     *scanner,
 148                                            gchar        *message,
 149                                            gint          is_error);
 150
 151
 152 /* --- functions --- */
 153 static inline gint
 154 g_scanner_char_2_num (guchar    c,
 155                       guchar    base)
 156 {
 157   if (c >= '0' && c <= '9')
 158     c -= '0';
 159   else if (c >= 'A' && c <= 'Z')
 160     c -= 'A' - 10;
 161   else if (c >= 'a' && c <= 'z')
 162     c -= 'a' - 10;
 163   else
 164     return -1;
 165
 166   if (c < base)
 167     return c;
 168
 169   return -1;
 170 }
 171
 172 GScanner*
 173 g_scanner_new (GScannerConfig   *config_templ)
 174 {
 175   GScanner *scanner;
 176
 177   if (!config_templ)
 178     config_templ = &g_scanner_config_template;
 179
 180   scanner = g_new0 (GScanner, 1);
 181
 182   scanner->user_data = NULL;
 183   scanner->max_parse_errors = 0;
 184   scanner->parse_errors = 0;
 185   scanner->input_name = NULL;
 186   scanner->derived_data = NULL;
 187
 188   scanner->config = g_new0 (GScannerConfig, 1);
 189
 190   scanner->config->case_sensitive       = config_templ->case_sensitive;
 191   scanner->config->cset_skip_characters = config_templ->cset_skip_characters;
 192   scanner->config->cset_identifier_first= config_templ->cset_identifier_first;
 193   scanner->config->cset_identifier_nth  = config_templ->cset_identifier_nth;
 194   scanner->config->cpair_comment_single = config_templ->cpair_comment_single;
 195   scanner->config->skip_comment_multi   = config_templ->skip_comment_multi;
 196   scanner->config->skip_comment_single  = config_templ->skip_comment_single;
 197   scanner->config->scan_comment_multi   = config_templ->scan_comment_multi;
 198   scanner->config->scan_identifier      = config_templ->scan_identifier;
 199   scanner->config->scan_identifier_1char= config_templ->scan_identifier_1char;
 200   scanner->config->scan_identifier_NULL = config_templ->scan_identifier_NULL;
 201   scanner->config->scan_symbols         = config_templ->scan_symbols;
 202   scanner->config->scan_binary          = config_templ->scan_binary;
 203   scanner->config->scan_octal           = config_templ->scan_octal;
 204   scanner->config->scan_float           = config_templ->scan_float;
 205   scanner->config->scan_hex             = config_templ->scan_hex;
 206   scanner->config->scan_hex_dollar      = config_templ->scan_hex_dollar;
 207   scanner->config->scan_string_sq       = config_templ->scan_string_sq;
 208   scanner->config->scan_string_dq       = config_templ->scan_string_dq;
 209   scanner->config->numbers_2_int        = config_templ->numbers_2_int;
 210   scanner->config->int_2_float          = config_templ->int_2_float;
 211   scanner->config->identifier_2_string  = config_templ->identifier_2_string;
 212   scanner->config->char_2_token         = config_templ->char_2_token;
 213   scanner->config->symbol_2_token       = config_templ->symbol_2_token;
 214   scanner->config->scope_0_fallback     = config_templ->scope_0_fallback;
 215
 216   scanner->token = G_TOKEN_NONE;
 217   scanner->value.v_int = 0;
 218   scanner->line = 1;
 219   scanner->position = 0;
 220
 221   scanner->next_token = G_TOKEN_NONE;
 222   scanner->next_value.v_int = 0;
 223   scanner->next_line = 1;
 224   scanner->next_position = 0;
 225
 226   scanner->symbol_table = g_hash_table_new (g_scanner_key_hash, g_scanner_key_equal);
 227   scanner->input_fd = -1;
 228   scanner->text = NULL;
 229   scanner->text_end = NULL;
 230   scanner->buffer = NULL;
 231   scanner->scope_id = 0;
 232
 233   scanner->msg_handler = g_scanner_msg_handler;
 234
 235   return scanner;
 236 }
 237
 238 static inline void
 239 g_scanner_free_value (GTokenType     *token_p,
 240                       GTokenValue     *value_p)
 241 {
 242   switch (*token_p)
 243     {
 244     case G_TOKEN_STRING:
 245     case G_TOKEN_IDENTIFIER:
 246     case G_TOKEN_IDENTIFIER_NULL:
 247     case G_TOKEN_COMMENT_SINGLE:
 248     case G_TOKEN_COMMENT_MULTI:
 249       g_free (value_p->v_string);
 250       break;
 251
 252     default:
 253       break;
 254     }
 255
 256   *token_p = G_TOKEN_NONE;
 257 }
 258
 259 static void
 260 g_scanner_destroy_symbol_table_entry (gpointer _key,
 261                                       gpointer _value,
 262                                       gpointer _data)
 263 {
 264   GScannerKey *key = _key;
 265
 266   g_free (key->symbol);
 267   g_free (key);
 268 }
 269
 270 void
 271 g_scanner_destroy (GScanner     *scanner)
 272 {
 273   g_return_if_fail (scanner != NULL);
 274
 275   g_hash_table_foreach (scanner->symbol_table,
 276                         g_scanner_destroy_symbol_table_entry, NULL);
 277   g_hash_table_destroy (scanner->symbol_table);
 278   g_scanner_free_value (&scanner->token, &scanner->value);
 279   g_scanner_free_value (&scanner->next_token, &scanner->next_value);
 280   g_free (scanner->config);
 281   g_free (scanner->buffer);
 282   g_free (scanner);
 283 }
 284
 285 static void
 286 g_scanner_msg_handler (GScanner         *scanner,
 287                        gchar            *message,
 288                        gint              is_error)
 289 {
 290   g_return_if_fail (scanner != NULL);
 291
 292   fprintf (stdout, "%s:%d: ", scanner->input_name, scanner->line);
 293   if (is_error)
 294     fprintf (stdout, "error: ");
 295   fprintf (stdout, "%s\n", message);
 296 }
 297
 298 void
 299 g_scanner_error (GScanner       *scanner,
 300                  const gchar    *format,
 301                  ...)
 302 {
 303   g_return_if_fail (scanner != NULL);
 304   g_return_if_fail (format != NULL);
 305
 306   scanner->parse_errors++;
 307
 308   if (scanner->msg_handler)
 309     {
 310       va_list args;
 311       gchar *string;
 312
 313       va_start (args, format);
 314       string = g_strdup_vprintf (format, args);
 315       va_end (args);
 316
 317       scanner->msg_handler (scanner, string, TRUE);
 318
 319       g_free (string);
 320     }
 321 }
 322
 323 void
 324 g_scanner_warn (GScanner       *scanner,
 325                 const gchar    *format,
 326                 ...)
 327 {
 328   g_return_if_fail (scanner != NULL);
 329   g_return_if_fail (format != NULL);
 330
 331   if (scanner->msg_handler)
 332     {
 333       va_list args;
 334       gchar *string;
 335
 336       va_start (args, format);
 337       string = g_strdup_vprintf (format, args);
 338       va_end (args);
 339
 340       scanner->msg_handler (scanner, string, FALSE);
 341
 342       g_free (string);
 343     }
 344 }
 345
 346 static gint
 347 g_scanner_key_equal (gconstpointer v1,
 348                      gconstpointer v2)
 349 {
 350   const GScannerKey *key1 = v1;
 351   const GScannerKey *key2 = v2;
 352
 353   return (key1->scope_id == key2->scope_id) && (strcmp (key1->symbol, key2->symbol) == 0);
 354 }
 355
 356 static guint
 357 g_scanner_key_hash (gconstpointer v)
 358 {
 359   const GScannerKey *key = v;
 360   gchar *c;
 361   guint h;
 362
 363   h = key->scope_id;
 364   for (c = key->symbol; *c; c++)
 365     {
 366       guint g;
 367
 368       h = (h << 4) + *c;
 369       g = h & 0xf0000000;
 370       if (g)
 371         {
 372           h = h ^ (g >> 24);
 373           h = h ^ g;
 374         }
 375     }
 376
 377   return h;
 378 }
 379
 380 static inline GScannerKey*
 381 g_scanner_lookup_internal (GScanner     *scanner,
 382                            guint         scope_id,
 383                            const gchar  *symbol)
 384 {
 385   GScannerKey   *key_p;
 386   GScannerKey key;
 387
 388   key.scope_id = scope_id;
 389
 390   if (!scanner->config->case_sensitive)
 391     {
 392       gchar *d;
 393       const gchar *c;
 394
 395       key.symbol = g_new (gchar, strlen (symbol) + 1);
 396       for (d = key.symbol, c = symbol; *c; c++, d++)
 397         *d = to_lower (*c);
 398       *d = 0;
 399       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 400       g_free (key.symbol);
 401     }
 402   else
 403     {
 404       key.symbol = (gchar*) symbol;
 405       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 406     }
 407
 408   return key_p;
 409 }
 410
 411 void
 412 g_scanner_scope_add_symbol (GScanner    *scanner,
 413                             guint        scope_id,
 414                             const gchar *symbol,
 415                             gpointer     value)
 416 {
 417   GScannerKey   *key;
 418
 419   g_return_if_fail (scanner != NULL);
 420   g_return_if_fail (symbol != NULL);
 421
 422   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 423
 424   if (!key)
 425     {
 426       key = g_new (GScannerKey, 1);
 427       key->scope_id = scope_id;
 428       key->symbol = g_strdup (symbol);
 429       key->value = value;
 430       if (!scanner->config->case_sensitive)
 431         {
 432           gchar *c;
 433
 434           c = key->symbol;
 435           while (*c != 0)
 436             {
 437               *c = to_lower (*c);
 438               c++;
 439             }
 440         }
 441       g_hash_table_insert (scanner->symbol_table, key, key);
 442     }
 443   else
 444     key->value = value;
 445 }
 446
 447 void
 448 g_scanner_scope_remove_symbol (GScanner    *scanner,
 449                                guint        scope_id,
 450                                const gchar *symbol)
 451 {
 452   GScannerKey   *key;
 453
 454   g_return_if_fail (scanner != NULL);
 455   g_return_if_fail (symbol != NULL);
 456
 457   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 458
 459   if (key)
 460     {
 461       g_hash_table_remove (scanner->symbol_table, key);
 462       g_free (key->symbol);
 463       g_free (key);
 464     }
 465 }
 466
 467 gpointer
 468 g_scanner_lookup_symbol (GScanner       *scanner,
 469                          const gchar    *symbol)
 470 {
 471   GScannerKey   *key;
 472   guint scope_id;
 473
 474   g_return_val_if_fail (scanner != NULL, NULL);
 475
 476   if (!symbol)
 477     return NULL;
 478
 479   scope_id = scanner->scope_id;
 480   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 481   if (!key && scope_id && scanner->config->scope_0_fallback)
 482     key = g_scanner_lookup_internal (scanner, 0, symbol);
 483
 484   if (key)
 485     return key->value;
 486   else
 487     return NULL;
 488 }
 489
 490 gpointer
 491 g_scanner_scope_lookup_symbol (GScanner       *scanner,
 492                                guint           scope_id,
 493                                const gchar    *symbol)
 494 {
 495   GScannerKey   *key;
 496
 497   g_return_val_if_fail (scanner != NULL, NULL);
 498
 499   if (!symbol)
 500     return NULL;
 501
 502   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 503
 504   if (key)
 505     return key->value;
 506   else
 507     return NULL;
 508 }
 509
 510 guint
 511 g_scanner_set_scope (GScanner       *scanner,
 512                      guint           scope_id)
 513 {
 514   guint old_scope_id;
 515
 516   g_return_val_if_fail (scanner != NULL, 0);
 517
 518   old_scope_id = scanner->scope_id;
 519   scanner->scope_id = scope_id;
 520
 521   return old_scope_id;
 522 }
 523
 524 static void
 525 g_scanner_foreach_internal (gpointer  _key,
 526                             gpointer  _value,
 527                             gpointer  _user_data)
 528 {
 529   GScannerKey *key;
 530   gpointer *d;
 531   GHFunc func;
 532   gpointer user_data;
 533   guint *scope_id;
 534
 535   d = _user_data;
 536   func = (GHFunc) d[0];
 537   user_data = d[1];
 538   scope_id = d[2];
 539   key = _value;
 540
 541   if (key->scope_id == *scope_id)
 542     func (key->symbol, key->value, user_data);
 543 }
 544
 545 void
 546 g_scanner_scope_foreach_symbol (GScanner       *scanner,
 547                                 guint           scope_id,
 548                                 GHFunc          func,
 549                                 gpointer        user_data)
 550 {
 551   gpointer d[3];
 552
 553   g_return_if_fail (scanner != NULL);
 554
 555   d[0] = (gpointer) func;
 556   d[1] = user_data;
 557   d[2] = &scope_id;
 558
 559   g_hash_table_foreach (scanner->symbol_table, g_scanner_foreach_internal, d);
 560 }
 561
 562 void
 563 g_scanner_freeze_symbol_table (GScanner *scanner)
 564 {
 565   g_return_if_fail (scanner != NULL);
 566
 567   g_hash_table_freeze (scanner->symbol_table);
 568 }
 569
 570 void
 571 g_scanner_thaw_symbol_table (GScanner *scanner)
 572 {
 573   g_return_if_fail (scanner != NULL);
 574
 575   g_hash_table_thaw (scanner->symbol_table);
 576 }
 577
 578 GTokenType
 579 g_scanner_peek_next_token (GScanner     *scanner)
 580 {
 581   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 582
 583   if (scanner->next_token == G_TOKEN_NONE)
 584     {
 585       scanner->next_line = scanner->line;
 586       scanner->next_position = scanner->position;
 587       g_scanner_get_token_i (scanner,
 588                              &scanner->next_token,
 589                              &scanner->next_value,
 590                              &scanner->next_line,
 591                              &scanner->next_position);
 592     }
 593
 594   return scanner->next_token;
 595 }
 596
 597 GTokenType
 598 g_scanner_get_next_token (GScanner      *scanner)
 599 {
 600   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 601
 602   if (scanner->next_token != G_TOKEN_NONE)
 603     {
 604       g_scanner_free_value (&scanner->token, &scanner->value);
 605
 606       scanner->token = scanner->next_token;
 607       scanner->value = scanner->next_value;
 608       scanner->line = scanner->next_line;
 609       scanner->position = scanner->next_position;
 610       scanner->next_token = G_TOKEN_NONE;
 611     }
 612   else
 613     g_scanner_get_token_i (scanner,
 614                            &scanner->token,
 615                            &scanner->value,
 616                            &scanner->line,
 617                            &scanner->position);
 618
 619   return scanner->token;
 620 }
 621
 622 GTokenType
 623 g_scanner_cur_token (GScanner *scanner)
 624 {
 625   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 626
 627   return scanner->token;
 628 }
 629
 630 GTokenValue
 631 g_scanner_cur_value (GScanner *scanner)
 632 {
 633   GTokenValue v;
 634
 635   v.v_int = 0;
 636
 637   g_return_val_if_fail (scanner != NULL, v);
 638
 639   /* MSC isn't capable of handling return scanner->value; ? */
 640
 641   v = scanner->value;
 642
 643   return v;
 644 }
 645
 646 guint
 647 g_scanner_cur_line (GScanner *scanner)
 648 {
 649   g_return_val_if_fail (scanner != NULL, 0);
 650
 651   return scanner->line;
 652 }
 653
 654 guint
 655 g_scanner_cur_position (GScanner *scanner)
 656 {
 657   g_return_val_if_fail (scanner != NULL, 0);
 658
 659   return scanner->position;
 660 }
 661
 662 gboolean
 663 g_scanner_eof (GScanner *scanner)
 664 {
 665   g_return_val_if_fail (scanner != NULL, TRUE);
 666
 667   return scanner->token == G_TOKEN_EOF;
 668 }
 669
 670 void
 671 g_scanner_input_file (GScanner *scanner,
 672                       gint      input_fd)
 673 {
 674   g_return_if_fail (scanner != NULL);
 675   g_return_if_fail (input_fd >= 0);
 676
 677   if (scanner->input_fd >= 0)
 678     g_scanner_sync_file_offset (scanner);
 679
 680   scanner->token = G_TOKEN_NONE;
 681   scanner->value.v_int = 0;
 682   scanner->line = 1;
 683   scanner->position = 0;
 684   scanner->next_token = G_TOKEN_NONE;
 685
 686   scanner->input_fd = input_fd;
 687   scanner->text = NULL;
 688   scanner->text_end = NULL;
 689
 690   if (!scanner->buffer)
 691     scanner->buffer = g_new (gchar, READ_BUFFER_SIZE + 1);
 692 }
 693
 694 void
 695 g_scanner_input_text (GScanner    *scanner,
 696                       const gchar *text,
 697                       guint        text_len)
 698 {
 699   g_return_if_fail (scanner != NULL);
 700   if (text_len)
 701     g_return_if_fail (text != NULL);
 702   else
 703     text = NULL;
 704
 705   if (scanner->input_fd >= 0)
 706     g_scanner_sync_file_offset (scanner);
 707
 708   scanner->token = G_TOKEN_NONE;
 709   scanner->value.v_int = 0;
 710   scanner->line = 1;
 711   scanner->position = 0;
 712   scanner->next_token = G_TOKEN_NONE;
 713
 714   scanner->input_fd = -1;
 715   scanner->text = text;
 716   scanner->text_end = text + text_len;
 717
 718   if (scanner->buffer)
 719     {
 720       g_free (scanner->buffer);
 721       scanner->buffer = NULL;
 722     }
 723 }
 724
 725 static guchar
 726 g_scanner_peek_next_char (GScanner *scanner)
 727 {
 728   if (scanner->text < scanner->text_end)
 729     {
 730       return *scanner->text;
 731     }
 732   else if (scanner->input_fd >= 0)
 733     {
 734       gint count;
 735       gchar *buffer;
 736
 737       buffer = scanner->buffer;
 738       do
 739         {
 740           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 741         }
 742       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 743
 744       if (count < 1)
 745         {
 746           scanner->input_fd = -1;
 747
 748           return 0;
 749         }
 750       else
 751         {
 752           scanner->text = buffer;
 753           scanner->text_end = buffer + count;
 754
 755           return *buffer;
 756         }
 757     }
 758   else
 759     return 0;
 760 }
 761
 762 void
 763 g_scanner_sync_file_offset (GScanner *scanner)
 764 {
 765   g_return_if_fail (scanner != NULL);
 766
 767   /* for file input, rewind the filedescriptor to the current
 768    * buffer position and blow the file read ahead buffer. usefull for
 769    * third party uses of our filedescriptor, which hooks onto the current
 770    * scanning position.
 771    */
 772
 773   if (scanner->input_fd >= 0 && scanner->text_end > scanner->text)
 774     {
 775       gint buffered;
 776
 777       buffered = scanner->text_end - scanner->text;
 778       if (lseek (scanner->input_fd, - buffered, SEEK_CUR) >= 0)
 779         {
 780           /* we succeeded, blow our buffer's contents now */
 781           scanner->text = NULL;
 782           scanner->text_end = NULL;
 783         }
 784       else
 785         errno = 0;
 786     }
 787 }
 788
 789 static guchar
 790 g_scanner_get_char (GScanner    *scanner,
 791                     guint       *line_p,
 792                     guint       *position_p)
 793 {
 794   guchar fchar;
 795
 796   if (scanner->text < scanner->text_end)
 797     fchar = *(scanner->text++);
 798   else if (scanner->input_fd >= 0)
 799     {
 800       gint count;
 801       gchar *buffer;
 802
 803       buffer = scanner->buffer;
 804       do
 805         {
 806           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 807         }
 808       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 809
 810       if (count < 1)
 811         {
 812           scanner->input_fd = -1;
 813           fchar = 0;
 814         }
 815       else
 816         {
 817           scanner->text = buffer + 1;
 818           scanner->text_end = buffer + count;
 819           fchar = *buffer;
 820           if (!fchar)
 821             {
 822               g_scanner_sync_file_offset (scanner);
 823               scanner->text_end = scanner->text;
 824               scanner->input_fd = -1;
 825             }
 826         }
 827     }
 828   else
 829     fchar = 0;
 830
 831   if (fchar == '\n')
 832     {
 833       (*position_p) = 0;
 834       (*line_p)++;
 835     }
 836   else if (fchar)
 837     {
 838       (*position_p)++;
 839     }
 840
 841   return fchar;
 842 }
 843
 844 void
 845 g_scanner_unexp_token (GScanner         *scanner,
 846                        GTokenType        expected_token,
 847                        const gchar      *identifier_spec,
 848                        const gchar      *symbol_spec,
 849                        const gchar      *symbol_name,
 850                        const gchar      *message,
 851                        gint              is_error)
 852 {
 853   gchar *token_string;
 854   guint token_string_len;
 855   gchar *expected_string;
 856   guint expected_string_len;
 857   gchar *message_prefix;
 858   gboolean print_unexp;
 859   void (*msg_handler)   (GScanner*, const gchar*, ...);
 860
 861   g_return_if_fail (scanner != NULL);
 862
 863   if (is_error)
 864     msg_handler = g_scanner_error;
 865   else
 866     msg_handler = g_scanner_warn;
 867
 868   if (!identifier_spec)
 869     identifier_spec = "identifier";
 870   if (!symbol_spec)
 871     symbol_spec = "symbol";
 872
 873   token_string_len = 56;
 874   token_string = g_new (gchar, token_string_len + 1);
 875   expected_string_len = 64;
 876   expected_string = g_new (gchar, expected_string_len + 1);
 877   print_unexp = TRUE;
 878
 879   switch (scanner->token)
 880     {
 881     case G_TOKEN_EOF:
 882       g_snprintf (token_string, token_string_len, "end of file");
 883       break;
 884
 885     default:
 886       if (scanner->token >= 1 && scanner->token <= 255)
 887         {
 888           if ((scanner->token >= ' ' && scanner->token <= '~') ||
 889               strchr (scanner->config->cset_identifier_first, scanner->token) ||
 890               strchr (scanner->config->cset_identifier_nth, scanner->token))
 891             g_snprintf (token_string, expected_string_len, "character `%c'", scanner->token);
 892           else
 893             g_snprintf (token_string, expected_string_len, "character `\\%o'", scanner->token);
 894           break;
 895         }
 896       else if (!scanner->config->symbol_2_token)
 897         {
 898           g_snprintf (token_string, token_string_len, "(unknown) token <%d>", scanner->token);
 899           break;
 900         }
 901       /* fall through */
 902     case G_TOKEN_SYMBOL:
 903       if (expected_token == G_TOKEN_SYMBOL ||
 904           (scanner->config->symbol_2_token &&
 905            expected_token > G_TOKEN_LAST))
 906         print_unexp = FALSE;
 907       if (symbol_name)
 908         g_snprintf (token_string,
 909                     token_string_len,
 910                     "%s%s `%s'",
 911                     print_unexp ? "" : "invalid ",
 912                     symbol_spec,
 913                     symbol_name);
 914       else
 915         g_snprintf (token_string,
 916                     token_string_len,
 917                     "%s%s",
 918                     print_unexp ? "" : "invalid ",
 919                     symbol_spec);
 920       break;
 921
 922     case G_TOKEN_ERROR:
 923       print_unexp = FALSE;
 924       expected_token = G_TOKEN_NONE;
 925       switch (scanner->value.v_error)
 926         {
 927         case G_ERR_UNEXP_EOF:
 928           g_snprintf (token_string, token_string_len, "scanner: unexpected end of file");
 929           break;
 930
 931         case G_ERR_UNEXP_EOF_IN_STRING:
 932           g_snprintf (token_string, token_string_len, "scanner: unterminated string constant");
 933           break;
 934
 935         case G_ERR_UNEXP_EOF_IN_COMMENT:
 936           g_snprintf (token_string, token_string_len, "scanner: unterminated comment");
 937           break;
 938
 939         case G_ERR_NON_DIGIT_IN_CONST:
 940           g_snprintf (token_string, token_string_len, "scanner: non digit in constant");
 941           break;
 942
 943         case G_ERR_FLOAT_RADIX:
 944           g_snprintf (token_string, token_string_len, "scanner: invalid radix for floating constant");
 945           break;
 946
 947         case G_ERR_FLOAT_MALFORMED:
 948           g_snprintf (token_string, token_string_len, "scanner: malformed floating constant");
 949           break;
 950
 951         case G_ERR_DIGIT_RADIX:
 952           g_snprintf (token_string, token_string_len, "scanner: digit is beyond radix");
 953           break;
 954
 955         case G_ERR_UNKNOWN:
 956         default:
 957           g_snprintf (token_string, token_string_len, "scanner: unknown error");
 958           break;
 959         }
 960       break;
 961
 962     case G_TOKEN_CHAR:
 963       g_snprintf (token_string, token_string_len, "character `%c'", scanner->value.v_char);
 964       break;
 965
 966     case G_TOKEN_IDENTIFIER:
 967     case G_TOKEN_IDENTIFIER_NULL:
 968       if (expected_token == G_TOKEN_IDENTIFIER ||
 969           expected_token == G_TOKEN_IDENTIFIER_NULL)
 970         print_unexp = FALSE;
 971       g_snprintf (token_string,
 972                   token_string_len,
 973                   "%s%s `%s'",
 974                   print_unexp ? "" : "invalid ",
 975                   identifier_spec,
 976                   scanner->token == G_TOKEN_IDENTIFIER ? scanner->value.v_string : "null");
 977       break;
 978
 979     case G_TOKEN_BINARY:
 980     case G_TOKEN_OCTAL:
 981     case G_TOKEN_INT:
 982     case G_TOKEN_HEX:
 983       g_snprintf (token_string, token_string_len, "number `%ld'", scanner->value.v_int);
 984       break;
 985
 986     case G_TOKEN_FLOAT:
 987       g_snprintf (token_string, token_string_len, "number `%.3f'", scanner->value.v_float);
 988       break;
 989
 990     case G_TOKEN_STRING:
 991       if (expected_token == G_TOKEN_STRING)
 992         print_unexp = FALSE;
 993       g_snprintf (token_string,
 994                   token_string_len,
 995                   "%s%sstring constant \"%s\"",
 996                   print_unexp ? "" : "invalid ",
 997                   scanner->value.v_string[0] == 0 ? "empty " : "",
 998                   scanner->value.v_string);
 999       token_string[token_string_len - 2] = '"';
1000       token_string[token_string_len - 1] = 0;
1001       break;
1002
1003     case G_TOKEN_COMMENT_SINGLE:
1004     case G_TOKEN_COMMENT_MULTI:
1005       g_snprintf (token_string, token_string_len, "comment");
1006       break;
1007
1008     case G_TOKEN_NONE:
1009       /* somehow the user's parsing code is screwed, there isn't much
1010        * we can do about it.
1011        * Note, a common case to trigger this is
1012        * g_scanner_peek_next_token(); g_scanner_unexp_token();
1013        * without an intermediate g_scanner_get_next_token().
1014        */
1015       g_assert_not_reached ();
1016       break;
1017     }
1018
1019
1020   switch (expected_token)
1021     {
1022       gboolean need_valid;
1023
1024     default:
1025       if (expected_token >= 1 && expected_token <= 255)
1026         {
1027           if ((expected_token >= ' ' && expected_token <= '~') ||
1028               strchr (scanner->config->cset_identifier_first, expected_token) ||
1029               strchr (scanner->config->cset_identifier_nth, expected_token))
1030             g_snprintf (expected_string, expected_string_len, "character `%c'", expected_token);
1031           else
1032             g_snprintf (expected_string, expected_string_len, "character `\\%o'", expected_token);
1033           break;
1034         }
1035       else if (!scanner->config->symbol_2_token)
1036         {
1037           g_snprintf (expected_string, expected_string_len, "(unknown) token <%d>", expected_token);
1038           break;
1039         }
1040       /* fall through */
1041     case G_TOKEN_SYMBOL:
1042       need_valid = (scanner->token == G_TOKEN_SYMBOL ||
1043                     (scanner->config->symbol_2_token &&
1044                      scanner->token > G_TOKEN_LAST));
1045       g_snprintf (expected_string,
1046                   expected_string_len,
1047                   "%s%s",
1048                   need_valid ? "valid " : "",
1049                   symbol_spec);
1050       /* FIXME: should we attempt to lookup the symbol_name for symbol_2_token? */
1051       break;
1052
1053     case G_TOKEN_INT:
1054       g_snprintf (expected_string, expected_string_len, "number (integer)");
1055       break;
1056
1057     case G_TOKEN_FLOAT:
1058       g_snprintf (expected_string, expected_string_len, "number (float)");
1059       break;
1060
1061     case G_TOKEN_STRING:
1062       g_snprintf (expected_string,
1063                   expected_string_len,
1064                   "%sstring constant",
1065                   scanner->token == G_TOKEN_STRING ? "valid " : "");
1066       break;
1067
1068     case G_TOKEN_IDENTIFIER:
1069     case G_TOKEN_IDENTIFIER_NULL:
1070       g_snprintf (expected_string,
1071                   expected_string_len,
1072                   "%s%s",
1073                   (scanner->token == G_TOKEN_IDENTIFIER_NULL ||
1074                    scanner->token == G_TOKEN_IDENTIFIER ? "valid " : ""),
1075                   identifier_spec);
1076       break;
1077
1078     case G_TOKEN_EOF:
1079       g_snprintf (expected_string, expected_string_len, "end of file");
1080       break;
1081
1082     case G_TOKEN_NONE:
1083       break;
1084     }
1085
1086   if (message && message[0] != 0)
1087     message_prefix = " - ";
1088   else
1089     {
1090       message_prefix = "";
1091       message = "";
1092     }
1093
1094   if (expected_token != G_TOKEN_NONE)
1095     {
1096       if (print_unexp)
1097         msg_handler (scanner,
1098                      "unexpected %s, expected %s%s%s",
1099                      token_string,
1100                      expected_string,
1101                      message_prefix,
1102                      message);
1103       else
1104         msg_handler (scanner,
1105                      "%s, expected %s%s%s",
1106                      token_string,
1107                      expected_string,
1108                      message_prefix,
1109                      message);
1110     }
1111   else
1112     {
1113       if (print_unexp)
1114         msg_handler (scanner,
1115                      "unexpected %s%s%s",
1116                      token_string,
1117                      message_prefix,
1118                      message);
1119       else
1120         msg_handler (scanner,
1121                      "%s%s%s",
1122                      token_string,
1123                      message_prefix,
1124                      message);
1125     }
1126
1127   g_free (token_string);
1128   g_free (expected_string);
1129 }
1130
1131 gint
1132 g_scanner_stat_mode (const gchar *filename)
1133 {
1134   struct stat  *stat_buf;
1135   gint          st_mode;
1136
1137   stat_buf = g_new0 (struct stat, 1);
1138 #ifdef HAVE_LSTAT
1139   lstat (filename, stat_buf);
1140 #else
1141   stat (filename, stat_buf);
1142 #endif
1143   st_mode = stat_buf->st_mode;
1144
1145   g_free (stat_buf);
1146
1147   return st_mode;
1148 }
1149
1150 static void
1151 g_scanner_get_token_i (GScanner *scanner,
1152                        GTokenType       *token_p,
1153                        GTokenValue      *value_p,
1154                        guint            *line_p,
1155                        guint            *position_p)
1156 {
1157   do
1158     {
1159       g_scanner_free_value (token_p, value_p);
1160       g_scanner_get_token_ll (scanner, token_p, value_p, line_p, position_p);
1161     }
1162   while (((*token_p > 0 && *token_p < 256) &&
1163           strchr (scanner->config->cset_skip_characters, *token_p)) ||
1164          (*token_p == G_TOKEN_CHAR &&
1165           strchr (scanner->config->cset_skip_characters, value_p->v_char)) ||
1166          (*token_p == G_TOKEN_COMMENT_MULTI &&
1167           scanner->config->skip_comment_multi) ||
1168          (*token_p == G_TOKEN_COMMENT_SINGLE &&
1169           scanner->config->skip_comment_single));
1170
1171   switch (*token_p)
1172     {
1173     case G_TOKEN_IDENTIFIER:
1174       if (scanner->config->identifier_2_string)
1175         *token_p = G_TOKEN_STRING;
1176       break;
1177
1178     case G_TOKEN_SYMBOL:
1179       if (scanner->config->symbol_2_token)
1180         *token_p = (GTokenType) value_p->v_symbol;
1181       break;
1182
1183     case G_TOKEN_BINARY:
1184     case G_TOKEN_OCTAL:
1185     case G_TOKEN_HEX:
1186       if (scanner->config->numbers_2_int)
1187         *token_p = G_TOKEN_INT;
1188       break;
1189
1190     default:
1191       break;
1192     }
1193
1194   if (*token_p == G_TOKEN_INT &&
1195       scanner->config->int_2_float)
1196     {
1197       *token_p = G_TOKEN_FLOAT;
1198       value_p->v_float = value_p->v_int;
1199     }
1200
1201   errno = 0;
1202 }
1203
1204 static void
1205 g_scanner_get_token_ll  (GScanner       *scanner,
1206                          GTokenType     *token_p,
1207                          GTokenValue    *value_p,
1208                          guint          *line_p,
1209                          guint          *position_p)
1210 {
1211   GScannerConfig *config;
1212   GTokenType       token;
1213   gboolean         in_comment_multi;
1214   gboolean         in_comment_single;
1215   gboolean         in_string_sq;
1216   gboolean         in_string_dq;
1217   GString         *gstring;
1218   GTokenValue      value;
1219   guchar           ch;
1220
1221   config = scanner->config;
1222   (*value_p).v_int = 0;
1223
1224   if ((scanner->text >= scanner->text_end && scanner->input_fd < 0) ||
1225       scanner->token == G_TOKEN_EOF)
1226     {
1227       *token_p = G_TOKEN_EOF;
1228       return;
1229     }
1230
1231   in_comment_multi = FALSE;
1232   in_comment_single = FALSE;
1233   in_string_sq = FALSE;
1234   in_string_dq = FALSE;
1235   gstring = NULL;
1236
1237   do /* while (ch != 0) */
1238     {
1239       gboolean dotted_float = FALSE;
1240
1241       ch = g_scanner_get_char (scanner, line_p, position_p);
1242
1243       value.v_int = 0;
1244       token = G_TOKEN_NONE;
1245
1246       /* this is *evil*, but needed ;(
1247        * we first check for identifier first character, because  it
1248        * might interfere with other key chars like slashes or numbers
1249        */
1250       if (config->scan_identifier &&
1251           ch && strchr (config->cset_identifier_first, ch))
1252         goto identifier_precedence;
1253
1254       switch (ch)
1255         {
1256         case 0:
1257           token = G_TOKEN_EOF;
1258           (*position_p)++;
1259           /* ch = 0; */
1260           break;
1261
1262         case '/':
1263           if (!config->scan_comment_multi ||
1264               g_scanner_peek_next_char (scanner) != '*')
1265             goto default_case;
1266           g_scanner_get_char (scanner, line_p, position_p);
1267           token = G_TOKEN_COMMENT_MULTI;
1268           in_comment_multi = TRUE;
1269           gstring = g_string_new ("");
1270           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1271             {
1272               if (ch == '*' && g_scanner_peek_next_char (scanner) == '/')
1273                 {
1274                   g_scanner_get_char (scanner, line_p, position_p);
1275                   in_comment_multi = FALSE;
1276                   break;
1277                 }
1278               else
1279                 gstring = g_string_append_c (gstring, ch);
1280             }
1281           ch = 0;
1282           break;
1283
1284         case '\'':
1285           if (!config->scan_string_sq)
1286             goto default_case;
1287           token = G_TOKEN_STRING;
1288           in_string_sq = TRUE;
1289           gstring = g_string_new ("");
1290           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1291             {
1292               if (ch == '\'')
1293                 {
1294                   in_string_sq = FALSE;
1295                   break;
1296                 }
1297               else
1298                 gstring = g_string_append_c (gstring, ch);
1299             }
1300           ch = 0;
1301           break;
1302
1303         case '"':
1304           if (!config->scan_string_dq)
1305             goto default_case;
1306           token = G_TOKEN_STRING;
1307           in_string_dq = TRUE;
1308           gstring = g_string_new ("");
1309           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1310             {
1311               if (ch == '"')
1312                 {
1313                   in_string_dq = FALSE;
1314                   break;
1315                 }
1316               else
1317                 {
1318                   if (ch == '\\')
1319                     {
1320                       ch = g_scanner_get_char (scanner, line_p, position_p);
1321                       switch (ch)
1322                         {
1323                           guint i;
1324                           guint fchar;
1325
1326                         case 0:
1327                           break;
1328
1329                         case '\\':
1330                           gstring = g_string_append_c (gstring, '\\');
1331                           break;
1332
1333                         case 'n':
1334                           gstring = g_string_append_c (gstring, '\n');
1335                           break;
1336
1337                         case 't':
1338                           gstring = g_string_append_c (gstring, '\t');
1339                           break;
1340
1341                         case 'r':
1342                           gstring = g_string_append_c (gstring, '\r');
1343                           break;
1344
1345                         case 'b':
1346                           gstring = g_string_append_c (gstring, '\b');
1347                           break;
1348
1349                         case 'f':
1350                           gstring = g_string_append_c (gstring, '\f');
1351                           break;
1352
1353                         case '0':
1354                         case '1':
1355                         case '2':
1356                         case '3':
1357                         case '4':
1358                         case '5':
1359                         case '6':
1360                         case '7':
1361                           i = ch - '0';
1362                           fchar = g_scanner_peek_next_char (scanner);
1363                           if (fchar >= '0' && fchar <= '7')
1364                             {
1365                               ch = g_scanner_get_char (scanner, line_p, position_p);
1366                               i = i * 8 + ch - '0';
1367                               fchar = g_scanner_peek_next_char (scanner);
1368                               if (fchar >= '0' && fchar <= '7')
1369                                 {
1370                                   ch = g_scanner_get_char (scanner, line_p, position_p);
1371                                   i = i * 8 + ch - '0';
1372                                 }
1373                             }
1374                           gstring = g_string_append_c (gstring, i);
1375                           break;
1376
1377                         default:
1378                           gstring = g_string_append_c (gstring, ch);
1379                           break;
1380                         }
1381                     }
1382                   else
1383                     gstring = g_string_append_c (gstring, ch);
1384                 }
1385             }
1386           ch = 0;
1387           break;
1388
1389         case '.':
1390           if (!config->scan_float)
1391             goto default_case;
1392           token = G_TOKEN_FLOAT;
1393           dotted_float = TRUE;
1394           ch = g_scanner_get_char (scanner, line_p, position_p);
1395           goto number_parsing;
1396
1397         case '$':
1398           if (!config->scan_hex_dollar)
1399             goto default_case;
1400           token = G_TOKEN_HEX;
1401           ch = g_scanner_get_char (scanner, line_p, position_p);
1402           goto number_parsing;
1403
1404         case '0':
1405           if (config->scan_octal)
1406             token = G_TOKEN_OCTAL;
1407           else
1408             token = G_TOKEN_INT;
1409           ch = g_scanner_peek_next_char (scanner);
1410           if (config->scan_hex && (ch == 'x' || ch == 'X'))
1411             {
1412               token = G_TOKEN_HEX;
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, 16) < 0)
1423                 {
1424                   token = G_TOKEN_ERROR;
1425                   value.v_error = G_ERR_DIGIT_RADIX;
1426                   ch = 0;
1427                   break;
1428                 }
1429             }
1430           else if (config->scan_binary && (ch == 'b' || ch == 'B'))
1431             {
1432               token = G_TOKEN_BINARY;
1433               g_scanner_get_char (scanner, line_p, position_p);
1434               ch = g_scanner_get_char (scanner, line_p, position_p);
1435               if (ch == 0)
1436                 {
1437                   token = G_TOKEN_ERROR;
1438                   value.v_error = G_ERR_UNEXP_EOF;
1439                   (*position_p)++;
1440                   break;
1441                 }
1442               if (g_scanner_char_2_num (ch, 10) < 0)
1443                 {
1444                   token = G_TOKEN_ERROR;
1445                   value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1446                   ch = 0;
1447                   break;
1448                 }
1449             }
1450           else
1451             ch = '0';
1452           /* fall through */
1453         case '1':
1454         case '2':
1455         case '3':
1456         case '4':
1457         case '5':
1458         case '6':
1459         case '7':
1460         case '8':
1461         case '9':
1462         number_parsing:
1463         {
1464           gboolean in_number = TRUE;
1465           gchar *endptr;
1466
1467           if (token == G_TOKEN_NONE)
1468             token = G_TOKEN_INT;
1469
1470           gstring = g_string_new (dotted_float ? "0." : "");
1471           gstring = g_string_append_c (gstring, ch);
1472
1473           do /* while (in_number) */
1474             {
1475               gboolean is_E;
1476
1477               is_E = token == G_TOKEN_FLOAT && (ch == 'e' || ch == 'E');
1478
1479               ch = g_scanner_peek_next_char (scanner);
1480
1481               if (g_scanner_char_2_num (ch, 36) >= 0 ||
1482                   (config->scan_float && ch == '.') ||
1483                   (is_E && (ch == '+' || ch == '-')))
1484                 {
1485                   ch = g_scanner_get_char (scanner, line_p, position_p);
1486
1487                   switch (ch)
1488                     {
1489                     case '.':
1490                       if (token != G_TOKEN_INT && token != G_TOKEN_OCTAL)
1491                         {
1492                           value.v_error = token == G_TOKEN_FLOAT ? G_ERR_FLOAT_MALFORMED : G_ERR_FLOAT_RADIX;
1493                           token = G_TOKEN_ERROR;
1494                           in_number = FALSE;
1495                         }
1496                       else
1497                         {
1498                           token = G_TOKEN_FLOAT;
1499                           gstring = g_string_append_c (gstring, ch);
1500                         }
1501                       break;
1502
1503                     case '0':
1504                     case '1':
1505                     case '2':
1506                     case '3':
1507                     case '4':
1508                     case '5':
1509                     case '6':
1510                     case '7':
1511                     case '8':
1512                     case '9':
1513                       gstring = g_string_append_c (gstring, ch);
1514                       break;
1515
1516                     case '-':
1517                     case '+':
1518                       if (token != G_TOKEN_FLOAT)
1519                         {
1520                           token = G_TOKEN_ERROR;
1521                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1522                           in_number = FALSE;
1523                         }
1524                       else
1525                         gstring = g_string_append_c (gstring, ch);
1526                       break;
1527
1528                     case 'e':
1529                     case 'E':
1530                       if ((token != G_TOKEN_HEX && !config->scan_float) ||
1531                           (token != G_TOKEN_HEX &&
1532                            token != G_TOKEN_OCTAL &&
1533                            token != G_TOKEN_FLOAT &&
1534                            token != G_TOKEN_INT))
1535                         {
1536                           token = G_TOKEN_ERROR;
1537                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1538                           in_number = FALSE;
1539                         }
1540                       else
1541                         {
1542                           if (token != G_TOKEN_HEX)
1543                             token = G_TOKEN_FLOAT;
1544                           gstring = g_string_append_c (gstring, ch);
1545                         }
1546                       break;
1547
1548                     default:
1549                       if (token != G_TOKEN_HEX)
1550                         {
1551                           token = G_TOKEN_ERROR;
1552                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1553                           in_number = FALSE;
1554                         }
1555                       else
1556                         gstring = g_string_append_c (gstring, ch);
1557                       break;
1558                     }
1559                 }
1560               else
1561                 in_number = FALSE;
1562             }
1563           while (in_number);
1564
1565           endptr = NULL;
1566           switch (token)
1567             {
1568             case G_TOKEN_BINARY:
1569               value.v_binary = strtol (gstring->str, &endptr, 2);
1570               break;
1571
1572             case G_TOKEN_OCTAL:
1573               value.v_octal = strtol (gstring->str, &endptr, 8);
1574               break;
1575
1576             case G_TOKEN_INT:
1577               value.v_int = strtol (gstring->str, &endptr, 10);
1578               break;
1579
1580             case G_TOKEN_FLOAT:
1581               value.v_float = g_strtod (gstring->str, &endptr);
1582               break;
1583
1584             case G_TOKEN_HEX:
1585               value.v_hex = strtol (gstring->str, &endptr, 16);
1586               break;
1587
1588             default:
1589               break;
1590             }
1591           if (endptr && *endptr)
1592             {
1593               token = G_TOKEN_ERROR;
1594               if (*endptr == 'e' || *endptr == 'E')
1595                 value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1596               else
1597                 value.v_error = G_ERR_DIGIT_RADIX;
1598             }
1599           g_string_free (gstring, TRUE);
1600           gstring = NULL;
1601           ch = 0;
1602         } /* number_parsing:... */
1603         break;
1604
1605         default:
1606         default_case:
1607         {
1608           if (config->cpair_comment_single &&
1609               ch == config->cpair_comment_single[0])
1610             {
1611               token = G_TOKEN_COMMENT_SINGLE;
1612               in_comment_single = TRUE;
1613               gstring = g_string_new ("");
1614               while ((ch = g_scanner_get_char (scanner,
1615                                                line_p,
1616                                                position_p)) != 0)
1617                 {
1618                   if (ch == config->cpair_comment_single[1])
1619                     {
1620                       in_comment_single = FALSE;
1621                       ch = 0;
1622                       break;
1623                     }
1624
1625                   gstring = g_string_append_c (gstring, ch);
1626                   ch = 0;
1627                 }
1628             }
1629           else if (config->scan_identifier && ch &&
1630                    strchr (config->cset_identifier_first, ch))
1631             {
1632             identifier_precedence:
1633
1634               if (config->cset_identifier_nth && ch &&
1635                   strchr (config->cset_identifier_nth,
1636                           g_scanner_peek_next_char (scanner)))
1637                 {
1638                   token = G_TOKEN_IDENTIFIER;
1639                   gstring = g_string_new ("");
1640                   gstring = g_string_append_c (gstring, ch);
1641                   do
1642                     {
1643                       ch = g_scanner_get_char (scanner, line_p, position_p);
1644                       gstring = g_string_append_c (gstring, ch);
1645                       ch = g_scanner_peek_next_char (scanner);
1646                     }
1647                   while (ch && strchr (config->cset_identifier_nth, ch));
1648                   ch = 0;
1649                 }
1650               else if (config->scan_identifier_1char)
1651                 {
1652                   token = G_TOKEN_IDENTIFIER;
1653                   value.v_identifier = g_new0 (gchar, 2);
1654                   value.v_identifier[0] = ch;
1655                   ch = 0;
1656                 }
1657             }
1658           if (ch)
1659             {
1660               if (config->char_2_token)
1661                 token = ch;
1662               else
1663                 {
1664                   token = G_TOKEN_CHAR;
1665                   value.v_char = ch;
1666                 }
1667               ch = 0;
1668             }
1669         } /* default_case:... */
1670         break;
1671         }
1672       g_assert (ch == 0 && token != G_TOKEN_NONE); /* paranoid */
1673     }
1674   while (ch != 0);
1675
1676   if (in_comment_multi || in_comment_single ||
1677       in_string_sq || in_string_dq)
1678     {
1679       token = G_TOKEN_ERROR;
1680       if (gstring)
1681         {
1682           g_string_free (gstring, TRUE);
1683           gstring = NULL;
1684         }
1685       (*position_p)++;
1686       if (in_comment_multi || in_comment_single)
1687         value.v_error = G_ERR_UNEXP_EOF_IN_COMMENT;
1688       else /* (in_string_sq || in_string_dq) */
1689         value.v_error = G_ERR_UNEXP_EOF_IN_STRING;
1690     }
1691
1692   if (gstring)
1693     {
1694       value.v_string = gstring->str;
1695       g_string_free (gstring, FALSE);
1696       gstring = NULL;
1697     }
1698
1699   if (token == G_TOKEN_IDENTIFIER)
1700     {
1701       if (config->scan_symbols)
1702         {
1703           GScannerKey *key;
1704           guint scope_id;
1705
1706           scope_id = scanner->scope_id;
1707           key = g_scanner_lookup_internal (scanner, scope_id, value.v_identifier);
1708           if (!key && scope_id && scanner->config->scope_0_fallback)
1709             key = g_scanner_lookup_internal (scanner, 0, value.v_identifier);
1710
1711           if (key)
1712             {
1713               g_free (value.v_identifier);
1714               token = G_TOKEN_SYMBOL;
1715               value.v_symbol = key->value;
1716             }
1717         }
1718
1719       if (token == G_TOKEN_IDENTIFIER &&
1720           config->scan_identifier_NULL &&
1721           strlen (value.v_identifier) == 4)
1722         {
1723           gchar *null_upper = "NULL";
1724           gchar *null_lower = "null";
1725
1726           if (scanner->config->case_sensitive)
1727             {
1728               if (value.v_identifier[0] == null_upper[0] &&
1729                   value.v_identifier[1] == null_upper[1] &&
1730                   value.v_identifier[2] == null_upper[2] &&
1731                   value.v_identifier[3] == null_upper[3])
1732                 token = G_TOKEN_IDENTIFIER_NULL;
1733             }
1734           else
1735             {
1736               if ((value.v_identifier[0] == null_upper[0] ||
1737                    value.v_identifier[0] == null_lower[0]) &&
1738                   (value.v_identifier[1] == null_upper[1] ||
1739                    value.v_identifier[1] == null_lower[1]) &&
1740                   (value.v_identifier[2] == null_upper[2] ||
1741                    value.v_identifier[2] == null_lower[2]) &&
1742                   (value.v_identifier[3] == null_upper[3] ||
1743                    value.v_identifier[3] == null_lower[3]))
1744                 token = G_TOKEN_IDENTIFIER_NULL;
1745             }
1746         }
1747     }
1748
1749   *token_p = token;
1750   *value_p = value;
1751 }