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 Lesser 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  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with this library; if not, write to the
  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 1997-2000.  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 G_OS_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    "_"
  93    G_CSET_A_2_Z
  94    G_CSET_DIGITS
  95    G_CSET_LATINS
  96    G_CSET_LATINC
  97    )                    /* cset_identifier_nth */,
  98   ( "#\n" )             /* cpair_comment_single */,
  99
 100   FALSE                 /* case_sensitive */,
 101
 102   TRUE                  /* skip_comment_multi */,
 103   TRUE                  /* skip_comment_single */,
 104   TRUE                  /* scan_comment_multi */,
 105   TRUE                  /* scan_identifier */,
 106   FALSE                 /* scan_identifier_1char */,
 107   FALSE                 /* scan_identifier_NULL */,
 108   TRUE                  /* scan_symbols */,
 109   FALSE                 /* scan_binary */,
 110   TRUE                  /* scan_octal */,
 111   TRUE                  /* scan_float */,
 112   TRUE                  /* scan_hex */,
 113   FALSE                 /* scan_hex_dollar */,
 114   TRUE                  /* scan_string_sq */,
 115   TRUE                  /* scan_string_dq */,
 116   TRUE                  /* numbers_2_int */,
 117   FALSE                 /* int_2_float */,
 118   FALSE                 /* identifier_2_string */,
 119   TRUE                  /* char_2_token */,
 120   FALSE                 /* symbol_2_token */,
 121   FALSE                 /* scope_0_fallback */,
 122 };
 123
 124
 125 /* --- prototypes --- */
 126 static inline
 127 GScannerKey*    g_scanner_lookup_internal (GScanner     *scanner,
 128                                            guint         scope_id,
 129                                            const gchar  *symbol);
 130 static gboolean g_scanner_key_equal       (gconstpointer v1,
 131                                            gconstpointer v2);
 132 static guint    g_scanner_key_hash        (gconstpointer v);
 133 static void     g_scanner_get_token_ll    (GScanner     *scanner,
 134                                            GTokenType   *token_p,
 135                                            GTokenValue  *value_p,
 136                                            guint        *line_p,
 137                                            guint        *position_p);
 138 static void     g_scanner_get_token_i     (GScanner     *scanner,
 139                                            GTokenType   *token_p,
 140                                            GTokenValue  *value_p,
 141                                            guint        *line_p,
 142                                            guint        *position_p);
 143
 144 static guchar   g_scanner_peek_next_char  (GScanner     *scanner);
 145 static guchar   g_scanner_get_char        (GScanner     *scanner,
 146                                            guint        *line_p,
 147                                            guint        *position_p);
 148 static void     g_scanner_msg_handler     (GScanner     *scanner,
 149                                            gchar        *message,
 150                                            gint          is_error);
 151
 152
 153 /* --- functions --- */
 154 static inline gint
 155 g_scanner_char_2_num (guchar    c,
 156                       guchar    base)
 157 {
 158   if (c >= '0' && c <= '9')
 159     c -= '0';
 160   else if (c >= 'A' && c <= 'Z')
 161     c -= 'A' - 10;
 162   else if (c >= 'a' && c <= 'z')
 163     c -= 'a' - 10;
 164   else
 165     return -1;
 166
 167   if (c < base)
 168     return c;
 169
 170   return -1;
 171 }
 172
 173 GScanner*
 174 g_scanner_new (const GScannerConfig *config_templ)
 175 {
 176   GScanner *scanner;
 177
 178   if (!config_templ)
 179     config_templ = &g_scanner_config_template;
 180
 181   scanner = g_new0 (GScanner, 1);
 182
 183   scanner->user_data = NULL;
 184   scanner->max_parse_errors = 0;
 185   scanner->parse_errors = 0;
 186   scanner->input_name = NULL;
 187   g_datalist_init (&scanner->qdata);
 188
 189   scanner->config = g_new0 (GScannerConfig, 1);
 190
 191   scanner->config->case_sensitive        = config_templ->case_sensitive;
 192   scanner->config->cset_skip_characters  = config_templ->cset_skip_characters;
 193   if (!scanner->config->cset_skip_characters)
 194     scanner->config->cset_skip_characters = "";
 195   scanner->config->cset_identifier_first = config_templ->cset_identifier_first;
 196   scanner->config->cset_identifier_nth   = config_templ->cset_identifier_nth;
 197   scanner->config->cpair_comment_single  = config_templ->cpair_comment_single;
 198   scanner->config->skip_comment_multi    = config_templ->skip_comment_multi;
 199   scanner->config->skip_comment_single   = config_templ->skip_comment_single;
 200   scanner->config->scan_comment_multi    = config_templ->scan_comment_multi;
 201   scanner->config->scan_identifier       = config_templ->scan_identifier;
 202   scanner->config->scan_identifier_1char = config_templ->scan_identifier_1char;
 203   scanner->config->scan_identifier_NULL  = config_templ->scan_identifier_NULL;
 204   scanner->config->scan_symbols          = config_templ->scan_symbols;
 205   scanner->config->scan_binary           = config_templ->scan_binary;
 206   scanner->config->scan_octal            = config_templ->scan_octal;
 207   scanner->config->scan_float            = config_templ->scan_float;
 208   scanner->config->scan_hex              = config_templ->scan_hex;
 209   scanner->config->scan_hex_dollar       = config_templ->scan_hex_dollar;
 210   scanner->config->scan_string_sq        = config_templ->scan_string_sq;
 211   scanner->config->scan_string_dq        = config_templ->scan_string_dq;
 212   scanner->config->numbers_2_int         = config_templ->numbers_2_int;
 213   scanner->config->int_2_float           = config_templ->int_2_float;
 214   scanner->config->identifier_2_string   = config_templ->identifier_2_string;
 215   scanner->config->char_2_token          = config_templ->char_2_token;
 216   scanner->config->symbol_2_token        = config_templ->symbol_2_token;
 217   scanner->config->scope_0_fallback      = config_templ->scope_0_fallback;
 218
 219   scanner->token = G_TOKEN_NONE;
 220   scanner->value.v_int = 0;
 221   scanner->line = 1;
 222   scanner->position = 0;
 223
 224   scanner->next_token = G_TOKEN_NONE;
 225   scanner->next_value.v_int = 0;
 226   scanner->next_line = 1;
 227   scanner->next_position = 0;
 228
 229   scanner->symbol_table = g_hash_table_new (g_scanner_key_hash, g_scanner_key_equal);
 230   scanner->input_fd = -1;
 231   scanner->text = NULL;
 232   scanner->text_end = NULL;
 233   scanner->buffer = NULL;
 234   scanner->scope_id = 0;
 235
 236   scanner->msg_handler = g_scanner_msg_handler;
 237
 238   return scanner;
 239 }
 240
 241 static inline void
 242 g_scanner_free_value (GTokenType     *token_p,
 243                       GTokenValue     *value_p)
 244 {
 245   switch (*token_p)
 246     {
 247     case G_TOKEN_STRING:
 248     case G_TOKEN_IDENTIFIER:
 249     case G_TOKEN_IDENTIFIER_NULL:
 250     case G_TOKEN_COMMENT_SINGLE:
 251     case G_TOKEN_COMMENT_MULTI:
 252       g_free (value_p->v_string);
 253       break;
 254
 255     default:
 256       break;
 257     }
 258
 259   *token_p = G_TOKEN_NONE;
 260 }
 261
 262 static void
 263 g_scanner_destroy_symbol_table_entry (gpointer _key,
 264                                       gpointer _value,
 265                                       gpointer _data)
 266 {
 267   GScannerKey *key = _key;
 268
 269   g_free (key->symbol);
 270   g_free (key);
 271 }
 272
 273 void
 274 g_scanner_destroy (GScanner     *scanner)
 275 {
 276   g_return_if_fail (scanner != NULL);
 277
 278   g_datalist_clear (&scanner->qdata);
 279   g_hash_table_foreach (scanner->symbol_table,
 280                         g_scanner_destroy_symbol_table_entry, NULL);
 281   g_hash_table_destroy (scanner->symbol_table);
 282   g_scanner_free_value (&scanner->token, &scanner->value);
 283   g_scanner_free_value (&scanner->next_token, &scanner->next_value);
 284   g_free (scanner->config);
 285   g_free (scanner->buffer);
 286   g_free (scanner);
 287 }
 288
 289 static void
 290 g_scanner_msg_handler (GScanner         *scanner,
 291                        gchar            *message,
 292                        gint              is_error)
 293 {
 294   g_return_if_fail (scanner != NULL);
 295
 296   fprintf (stdout, "%s:%d: ", scanner->input_name, scanner->line);
 297   if (is_error)
 298     fprintf (stdout, "error: ");
 299   fprintf (stdout, "%s\n", message);
 300 }
 301
 302 void
 303 g_scanner_error (GScanner       *scanner,
 304                  const gchar    *format,
 305                  ...)
 306 {
 307   g_return_if_fail (scanner != NULL);
 308   g_return_if_fail (format != NULL);
 309
 310   scanner->parse_errors++;
 311
 312   if (scanner->msg_handler)
 313     {
 314       va_list args;
 315       gchar *string;
 316
 317       va_start (args, format);
 318       string = g_strdup_vprintf (format, args);
 319       va_end (args);
 320
 321       scanner->msg_handler (scanner, string, TRUE);
 322
 323       g_free (string);
 324     }
 325 }
 326
 327 void
 328 g_scanner_warn (GScanner       *scanner,
 329                 const gchar    *format,
 330                 ...)
 331 {
 332   g_return_if_fail (scanner != NULL);
 333   g_return_if_fail (format != NULL);
 334
 335   if (scanner->msg_handler)
 336     {
 337       va_list args;
 338       gchar *string;
 339
 340       va_start (args, format);
 341       string = g_strdup_vprintf (format, args);
 342       va_end (args);
 343
 344       scanner->msg_handler (scanner, string, FALSE);
 345
 346       g_free (string);
 347     }
 348 }
 349
 350 static gboolean
 351 g_scanner_key_equal (gconstpointer v1,
 352                      gconstpointer v2)
 353 {
 354   const GScannerKey *key1 = v1;
 355   const GScannerKey *key2 = v2;
 356
 357   return (key1->scope_id == key2->scope_id) && (strcmp (key1->symbol, key2->symbol) == 0);
 358 }
 359
 360 static guint
 361 g_scanner_key_hash (gconstpointer v)
 362 {
 363   const GScannerKey *key = v;
 364   gchar *c;
 365   guint h;
 366
 367   h = key->scope_id;
 368   for (c = key->symbol; *c; c++)
 369     {
 370       guint g;
 371
 372       h = (h << 4) + *c;
 373       g = h & 0xf0000000;
 374       if (g)
 375         {
 376           h = h ^ (g >> 24);
 377           h = h ^ g;
 378         }
 379     }
 380
 381   return h;
 382 }
 383
 384 static inline GScannerKey*
 385 g_scanner_lookup_internal (GScanner     *scanner,
 386                            guint         scope_id,
 387                            const gchar  *symbol)
 388 {
 389   GScannerKey   *key_p;
 390   GScannerKey key;
 391
 392   key.scope_id = scope_id;
 393
 394   if (!scanner->config->case_sensitive)
 395     {
 396       gchar *d;
 397       const gchar *c;
 398
 399       key.symbol = g_new (gchar, strlen (symbol) + 1);
 400       for (d = key.symbol, c = symbol; *c; c++, d++)
 401         *d = to_lower (*c);
 402       *d = 0;
 403       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 404       g_free (key.symbol);
 405     }
 406   else
 407     {
 408       key.symbol = (gchar*) symbol;
 409       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 410     }
 411
 412   return key_p;
 413 }
 414
 415 void
 416 g_scanner_scope_add_symbol (GScanner    *scanner,
 417                             guint        scope_id,
 418                             const gchar *symbol,
 419                             gpointer     value)
 420 {
 421   GScannerKey   *key;
 422
 423   g_return_if_fail (scanner != NULL);
 424   g_return_if_fail (symbol != NULL);
 425
 426   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 427
 428   if (!key)
 429     {
 430       key = g_new (GScannerKey, 1);
 431       key->scope_id = scope_id;
 432       key->symbol = g_strdup (symbol);
 433       key->value = value;
 434       if (!scanner->config->case_sensitive)
 435         {
 436           gchar *c;
 437
 438           c = key->symbol;
 439           while (*c != 0)
 440             {
 441               *c = to_lower (*c);
 442               c++;
 443             }
 444         }
 445       g_hash_table_insert (scanner->symbol_table, key, key);
 446     }
 447   else
 448     key->value = value;
 449 }
 450
 451 void
 452 g_scanner_scope_remove_symbol (GScanner    *scanner,
 453                                guint        scope_id,
 454                                const gchar *symbol)
 455 {
 456   GScannerKey   *key;
 457
 458   g_return_if_fail (scanner != NULL);
 459   g_return_if_fail (symbol != NULL);
 460
 461   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 462
 463   if (key)
 464     {
 465       g_hash_table_remove (scanner->symbol_table, key);
 466       g_free (key->symbol);
 467       g_free (key);
 468     }
 469 }
 470
 471 gpointer
 472 g_scanner_lookup_symbol (GScanner       *scanner,
 473                          const gchar    *symbol)
 474 {
 475   GScannerKey   *key;
 476   guint scope_id;
 477
 478   g_return_val_if_fail (scanner != NULL, NULL);
 479
 480   if (!symbol)
 481     return NULL;
 482
 483   scope_id = scanner->scope_id;
 484   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 485   if (!key && scope_id && scanner->config->scope_0_fallback)
 486     key = g_scanner_lookup_internal (scanner, 0, symbol);
 487
 488   if (key)
 489     return key->value;
 490   else
 491     return NULL;
 492 }
 493
 494 gpointer
 495 g_scanner_scope_lookup_symbol (GScanner       *scanner,
 496                                guint           scope_id,
 497                                const gchar    *symbol)
 498 {
 499   GScannerKey   *key;
 500
 501   g_return_val_if_fail (scanner != NULL, NULL);
 502
 503   if (!symbol)
 504     return NULL;
 505
 506   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 507
 508   if (key)
 509     return key->value;
 510   else
 511     return NULL;
 512 }
 513
 514 guint
 515 g_scanner_set_scope (GScanner       *scanner,
 516                      guint           scope_id)
 517 {
 518   guint old_scope_id;
 519
 520   g_return_val_if_fail (scanner != NULL, 0);
 521
 522   old_scope_id = scanner->scope_id;
 523   scanner->scope_id = scope_id;
 524
 525   return old_scope_id;
 526 }
 527
 528 static void
 529 g_scanner_foreach_internal (gpointer  _key,
 530                             gpointer  _value,
 531                             gpointer  _user_data)
 532 {
 533   GScannerKey *key;
 534   gpointer *d;
 535   GHFunc func;
 536   gpointer user_data;
 537   guint *scope_id;
 538
 539   d = _user_data;
 540   func = (GHFunc) d[0];
 541   user_data = d[1];
 542   scope_id = d[2];
 543   key = _value;
 544
 545   if (key->scope_id == *scope_id)
 546     func (key->symbol, key->value, user_data);
 547 }
 548
 549 void
 550 g_scanner_scope_foreach_symbol (GScanner       *scanner,
 551                                 guint           scope_id,
 552                                 GHFunc          func,
 553                                 gpointer        user_data)
 554 {
 555   gpointer d[3];
 556
 557   g_return_if_fail (scanner != NULL);
 558
 559   d[0] = (gpointer) func;
 560   d[1] = user_data;
 561   d[2] = &scope_id;
 562
 563   g_hash_table_foreach (scanner->symbol_table, g_scanner_foreach_internal, d);
 564 }
 565
 566 void
 567 g_scanner_freeze_symbol_table (GScanner *scanner)
 568 {
 569 #ifdef G_ENABLE_DEBUG
 570   static gboolean first_call = TRUE;
 571
 572   if (first_call)
 573     {
 574       g_message ("g_scanner_freeze_symbol_table() and "
 575                  "g_scanner_thaw_symbol_table() are deprecated.");
 576       first_call = FALSE;
 577     }
 578 #endif /* G_ENABLE_DEBUG */
 579 }
 580
 581 void
 582 g_scanner_thaw_symbol_table (GScanner *scanner)
 583 {
 584   g_scanner_freeze_symbol_table (scanner);
 585 }
 586
 587 GTokenType
 588 g_scanner_peek_next_token (GScanner     *scanner)
 589 {
 590   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 591
 592   if (scanner->next_token == G_TOKEN_NONE)
 593     {
 594       scanner->next_line = scanner->line;
 595       scanner->next_position = scanner->position;
 596       g_scanner_get_token_i (scanner,
 597                              &scanner->next_token,
 598                              &scanner->next_value,
 599                              &scanner->next_line,
 600                              &scanner->next_position);
 601     }
 602
 603   return scanner->next_token;
 604 }
 605
 606 GTokenType
 607 g_scanner_get_next_token (GScanner      *scanner)
 608 {
 609   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 610
 611   if (scanner->next_token != G_TOKEN_NONE)
 612     {
 613       g_scanner_free_value (&scanner->token, &scanner->value);
 614
 615       scanner->token = scanner->next_token;
 616       scanner->value = scanner->next_value;
 617       scanner->line = scanner->next_line;
 618       scanner->position = scanner->next_position;
 619       scanner->next_token = G_TOKEN_NONE;
 620     }
 621   else
 622     g_scanner_get_token_i (scanner,
 623                            &scanner->token,
 624                            &scanner->value,
 625                            &scanner->line,
 626                            &scanner->position);
 627
 628   return scanner->token;
 629 }
 630
 631 GTokenType
 632 g_scanner_cur_token (GScanner *scanner)
 633 {
 634   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 635
 636   return scanner->token;
 637 }
 638
 639 GTokenValue
 640 g_scanner_cur_value (GScanner *scanner)
 641 {
 642   GTokenValue v;
 643
 644   v.v_int = 0;
 645
 646   g_return_val_if_fail (scanner != NULL, v);
 647
 648   /* MSC isn't capable of handling return scanner->value; ? */
 649
 650   v = scanner->value;
 651
 652   return v;
 653 }
 654
 655 guint
 656 g_scanner_cur_line (GScanner *scanner)
 657 {
 658   g_return_val_if_fail (scanner != NULL, 0);
 659
 660   return scanner->line;
 661 }
 662
 663 guint
 664 g_scanner_cur_position (GScanner *scanner)
 665 {
 666   g_return_val_if_fail (scanner != NULL, 0);
 667
 668   return scanner->position;
 669 }
 670
 671 gboolean
 672 g_scanner_eof (GScanner *scanner)
 673 {
 674   g_return_val_if_fail (scanner != NULL, TRUE);
 675
 676   return scanner->token == G_TOKEN_EOF;
 677 }
 678
 679 void
 680 g_scanner_input_file (GScanner *scanner,
 681                       gint      input_fd)
 682 {
 683   g_return_if_fail (scanner != NULL);
 684   g_return_if_fail (input_fd >= 0);
 685
 686   if (scanner->input_fd >= 0)
 687     g_scanner_sync_file_offset (scanner);
 688
 689   scanner->token = G_TOKEN_NONE;
 690   scanner->value.v_int = 0;
 691   scanner->line = 1;
 692   scanner->position = 0;
 693   scanner->next_token = G_TOKEN_NONE;
 694
 695   scanner->input_fd = input_fd;
 696   scanner->text = NULL;
 697   scanner->text_end = NULL;
 698
 699   if (!scanner->buffer)
 700     scanner->buffer = g_new (gchar, READ_BUFFER_SIZE + 1);
 701 }
 702
 703 void
 704 g_scanner_input_text (GScanner    *scanner,
 705                       const gchar *text,
 706                       guint        text_len)
 707 {
 708   g_return_if_fail (scanner != NULL);
 709   if (text_len)
 710     g_return_if_fail (text != NULL);
 711   else
 712     text = NULL;
 713
 714   if (scanner->input_fd >= 0)
 715     g_scanner_sync_file_offset (scanner);
 716
 717   scanner->token = G_TOKEN_NONE;
 718   scanner->value.v_int = 0;
 719   scanner->line = 1;
 720   scanner->position = 0;
 721   scanner->next_token = G_TOKEN_NONE;
 722
 723   scanner->input_fd = -1;
 724   scanner->text = text;
 725   scanner->text_end = text + text_len;
 726
 727   if (scanner->buffer)
 728     {
 729       g_free (scanner->buffer);
 730       scanner->buffer = NULL;
 731     }
 732 }
 733
 734 static guchar
 735 g_scanner_peek_next_char (GScanner *scanner)
 736 {
 737   if (scanner->text < scanner->text_end)
 738     {
 739       return *scanner->text;
 740     }
 741   else if (scanner->input_fd >= 0)
 742     {
 743       gint count;
 744       gchar *buffer;
 745
 746       buffer = scanner->buffer;
 747       do
 748         {
 749           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 750         }
 751       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 752
 753       if (count < 1)
 754         {
 755           scanner->input_fd = -1;
 756
 757           return 0;
 758         }
 759       else
 760         {
 761           scanner->text = buffer;
 762           scanner->text_end = buffer + count;
 763
 764           return *buffer;
 765         }
 766     }
 767   else
 768     return 0;
 769 }
 770
 771 void
 772 g_scanner_sync_file_offset (GScanner *scanner)
 773 {
 774   g_return_if_fail (scanner != NULL);
 775
 776   /* for file input, rewind the filedescriptor to the current
 777    * buffer position and blow the file read ahead buffer. usefull for
 778    * third party uses of our filedescriptor, which hooks onto the current
 779    * scanning position.
 780    */
 781
 782   if (scanner->input_fd >= 0 && scanner->text_end > scanner->text)
 783     {
 784       gint buffered;
 785
 786       buffered = scanner->text_end - scanner->text;
 787       if (lseek (scanner->input_fd, - buffered, SEEK_CUR) >= 0)
 788         {
 789           /* we succeeded, blow our buffer's contents now */
 790           scanner->text = NULL;
 791           scanner->text_end = NULL;
 792         }
 793       else
 794         errno = 0;
 795     }
 796 }
 797
 798 static guchar
 799 g_scanner_get_char (GScanner    *scanner,
 800                     guint       *line_p,
 801                     guint       *position_p)
 802 {
 803   guchar fchar;
 804
 805   if (scanner->text < scanner->text_end)
 806     fchar = *(scanner->text++);
 807   else if (scanner->input_fd >= 0)
 808     {
 809       gint count;
 810       gchar *buffer;
 811
 812       buffer = scanner->buffer;
 813       do
 814         {
 815           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 816         }
 817       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 818
 819       if (count < 1)
 820         {
 821           scanner->input_fd = -1;
 822           fchar = 0;
 823         }
 824       else
 825         {
 826           scanner->text = buffer + 1;
 827           scanner->text_end = buffer + count;
 828           fchar = *buffer;
 829           if (!fchar)
 830             {
 831               g_scanner_sync_file_offset (scanner);
 832               scanner->text_end = scanner->text;
 833               scanner->input_fd = -1;
 834             }
 835         }
 836     }
 837   else
 838     fchar = 0;
 839
 840   if (fchar == '\n')
 841     {
 842       (*position_p) = 0;
 843       (*line_p)++;
 844     }
 845   else if (fchar)
 846     {
 847       (*position_p)++;
 848     }
 849
 850   return fchar;
 851 }
 852
 853 void
 854 g_scanner_unexp_token (GScanner         *scanner,
 855                        GTokenType        expected_token,
 856                        const gchar      *identifier_spec,
 857                        const gchar      *symbol_spec,
 858                        const gchar      *symbol_name,
 859                        const gchar      *message,
 860                        gint              is_error)
 861 {
 862   gchar *token_string;
 863   guint token_string_len;
 864   gchar *expected_string;
 865   guint expected_string_len;
 866   gchar *message_prefix;
 867   gboolean print_unexp;
 868   void (*msg_handler)   (GScanner*, const gchar*, ...);
 869
 870   g_return_if_fail (scanner != NULL);
 871
 872   if (is_error)
 873     msg_handler = g_scanner_error;
 874   else
 875     msg_handler = g_scanner_warn;
 876
 877   if (!identifier_spec)
 878     identifier_spec = "identifier";
 879   if (!symbol_spec)
 880     symbol_spec = "symbol";
 881
 882   token_string_len = 56;
 883   token_string = g_new (gchar, token_string_len + 1);
 884   expected_string_len = 64;
 885   expected_string = g_new (gchar, expected_string_len + 1);
 886   print_unexp = TRUE;
 887
 888   switch (scanner->token)
 889     {
 890     case G_TOKEN_EOF:
 891       g_snprintf (token_string, token_string_len, "end of file");
 892       break;
 893
 894     default:
 895       if (scanner->token >= 1 && scanner->token <= 255)
 896         {
 897           if ((scanner->token >= ' ' && scanner->token <= '~') ||
 898               strchr (scanner->config->cset_identifier_first, scanner->token) ||
 899               strchr (scanner->config->cset_identifier_nth, scanner->token))
 900             g_snprintf (token_string, expected_string_len, "character `%c'", scanner->token);
 901           else
 902             g_snprintf (token_string, expected_string_len, "character `\\%o'", scanner->token);
 903           break;
 904         }
 905       else if (!scanner->config->symbol_2_token)
 906         {
 907           g_snprintf (token_string, token_string_len, "(unknown) token <%d>", scanner->token);
 908           break;
 909         }
 910       /* fall through */
 911     case G_TOKEN_SYMBOL:
 912       if (expected_token == G_TOKEN_SYMBOL ||
 913           (scanner->config->symbol_2_token &&
 914            expected_token > G_TOKEN_LAST))
 915         print_unexp = FALSE;
 916       if (symbol_name)
 917         g_snprintf (token_string,
 918                     token_string_len,
 919                     "%s%s `%s'",
 920                     print_unexp ? "" : "invalid ",
 921                     symbol_spec,
 922                     symbol_name);
 923       else
 924         g_snprintf (token_string,
 925                     token_string_len,
 926                     "%s%s",
 927                     print_unexp ? "" : "invalid ",
 928                     symbol_spec);
 929       break;
 930
 931     case G_TOKEN_ERROR:
 932       print_unexp = FALSE;
 933       expected_token = G_TOKEN_NONE;
 934       switch (scanner->value.v_error)
 935         {
 936         case G_ERR_UNEXP_EOF:
 937           g_snprintf (token_string, token_string_len, "scanner: unexpected end of file");
 938           break;
 939
 940         case G_ERR_UNEXP_EOF_IN_STRING:
 941           g_snprintf (token_string, token_string_len, "scanner: unterminated string constant");
 942           break;
 943
 944         case G_ERR_UNEXP_EOF_IN_COMMENT:
 945           g_snprintf (token_string, token_string_len, "scanner: unterminated comment");
 946           break;
 947
 948         case G_ERR_NON_DIGIT_IN_CONST:
 949           g_snprintf (token_string, token_string_len, "scanner: non digit in constant");
 950           break;
 951
 952         case G_ERR_FLOAT_RADIX:
 953           g_snprintf (token_string, token_string_len, "scanner: invalid radix for floating constant");
 954           break;
 955
 956         case G_ERR_FLOAT_MALFORMED:
 957           g_snprintf (token_string, token_string_len, "scanner: malformed floating constant");
 958           break;
 959
 960         case G_ERR_DIGIT_RADIX:
 961           g_snprintf (token_string, token_string_len, "scanner: digit is beyond radix");
 962           break;
 963
 964         case G_ERR_UNKNOWN:
 965         default:
 966           g_snprintf (token_string, token_string_len, "scanner: unknown error");
 967           break;
 968         }
 969       break;
 970
 971     case G_TOKEN_CHAR:
 972       g_snprintf (token_string, token_string_len, "character `%c'", scanner->value.v_char);
 973       break;
 974
 975     case G_TOKEN_IDENTIFIER:
 976     case G_TOKEN_IDENTIFIER_NULL:
 977       if (expected_token == G_TOKEN_IDENTIFIER ||
 978           expected_token == G_TOKEN_IDENTIFIER_NULL)
 979         print_unexp = FALSE;
 980       g_snprintf (token_string,
 981                   token_string_len,
 982                   "%s%s `%s'",
 983                   print_unexp ? "" : "invalid ",
 984                   identifier_spec,
 985                   scanner->token == G_TOKEN_IDENTIFIER ? scanner->value.v_string : "null");
 986       break;
 987
 988     case G_TOKEN_BINARY:
 989     case G_TOKEN_OCTAL:
 990     case G_TOKEN_INT:
 991     case G_TOKEN_HEX:
 992       g_snprintf (token_string, token_string_len, "number `%ld'", scanner->value.v_int);
 993       break;
 994
 995     case G_TOKEN_FLOAT:
 996       g_snprintf (token_string, token_string_len, "number `%.3f'", scanner->value.v_float);
 997       break;
 998
 999     case G_TOKEN_STRING:
1000       if (expected_token == G_TOKEN_STRING)
1001         print_unexp = FALSE;
1002       g_snprintf (token_string,
1003                   token_string_len,
1004                   "%s%sstring constant \"%s\"",
1005                   print_unexp ? "" : "invalid ",
1006                   scanner->value.v_string[0] == 0 ? "empty " : "",
1007                   scanner->value.v_string);
1008       token_string[token_string_len - 2] = '"';
1009       token_string[token_string_len - 1] = 0;
1010       break;
1011
1012     case G_TOKEN_COMMENT_SINGLE:
1013     case G_TOKEN_COMMENT_MULTI:
1014       g_snprintf (token_string, token_string_len, "comment");
1015       break;
1016
1017     case G_TOKEN_NONE:
1018       /* somehow the user's parsing code is screwed, there isn't much
1019        * we can do about it.
1020        * Note, a common case to trigger this is
1021        * g_scanner_peek_next_token(); g_scanner_unexp_token();
1022        * without an intermediate g_scanner_get_next_token().
1023        */
1024       g_assert_not_reached ();
1025       break;
1026     }
1027
1028
1029   switch (expected_token)
1030     {
1031       gboolean need_valid;
1032
1033     default:
1034       if (expected_token >= 1 && expected_token <= 255)
1035         {
1036           if ((expected_token >= ' ' && expected_token <= '~') ||
1037               strchr (scanner->config->cset_identifier_first, expected_token) ||
1038               strchr (scanner->config->cset_identifier_nth, expected_token))
1039             g_snprintf (expected_string, expected_string_len, "character `%c'", expected_token);
1040           else
1041             g_snprintf (expected_string, expected_string_len, "character `\\%o'", expected_token);
1042           break;
1043         }
1044       else if (!scanner->config->symbol_2_token)
1045         {
1046           g_snprintf (expected_string, expected_string_len, "(unknown) token <%d>", expected_token);
1047           break;
1048         }
1049       /* fall through */
1050     case G_TOKEN_SYMBOL:
1051       need_valid = (scanner->token == G_TOKEN_SYMBOL ||
1052                     (scanner->config->symbol_2_token &&
1053                      scanner->token > G_TOKEN_LAST));
1054       g_snprintf (expected_string,
1055                   expected_string_len,
1056                   "%s%s",
1057                   need_valid ? "valid " : "",
1058                   symbol_spec);
1059       /* FIXME: should we attempt to lookup the symbol_name for symbol_2_token? */
1060       break;
1061
1062     case G_TOKEN_INT:
1063       g_snprintf (expected_string, expected_string_len, "number (integer)");
1064       break;
1065
1066     case G_TOKEN_FLOAT:
1067       g_snprintf (expected_string, expected_string_len, "number (float)");
1068       break;
1069
1070     case G_TOKEN_STRING:
1071       g_snprintf (expected_string,
1072                   expected_string_len,
1073                   "%sstring constant",
1074                   scanner->token == G_TOKEN_STRING ? "valid " : "");
1075       break;
1076
1077     case G_TOKEN_IDENTIFIER:
1078     case G_TOKEN_IDENTIFIER_NULL:
1079       g_snprintf (expected_string,
1080                   expected_string_len,
1081                   "%s%s",
1082                   (scanner->token == G_TOKEN_IDENTIFIER_NULL ||
1083                    scanner->token == G_TOKEN_IDENTIFIER ? "valid " : ""),
1084                   identifier_spec);
1085       break;
1086
1087     case G_TOKEN_EOF:
1088       g_snprintf (expected_string, expected_string_len, "end of file");
1089       break;
1090
1091     case G_TOKEN_NONE:
1092       break;
1093     }
1094
1095   if (message && message[0] != 0)
1096     message_prefix = " - ";
1097   else
1098     {
1099       message_prefix = "";
1100       message = "";
1101     }
1102
1103   if (expected_token != G_TOKEN_NONE)
1104     {
1105       if (print_unexp)
1106         msg_handler (scanner,
1107                      "unexpected %s, expected %s%s%s",
1108                      token_string,
1109                      expected_string,
1110                      message_prefix,
1111                      message);
1112       else
1113         msg_handler (scanner,
1114                      "%s, expected %s%s%s",
1115                      token_string,
1116                      expected_string,
1117                      message_prefix,
1118                      message);
1119     }
1120   else
1121     {
1122       if (print_unexp)
1123         msg_handler (scanner,
1124                      "unexpected %s%s%s",
1125                      token_string,
1126                      message_prefix,
1127                      message);
1128       else
1129         msg_handler (scanner,
1130                      "%s%s%s",
1131                      token_string,
1132                      message_prefix,
1133                      message);
1134     }
1135
1136   g_free (token_string);
1137   g_free (expected_string);
1138 }
1139
1140 gint
1141 g_scanner_stat_mode (const gchar *filename)
1142 {
1143   struct stat  *stat_buf;
1144   gint          st_mode;
1145
1146   stat_buf = g_new0 (struct stat, 1);
1147 #ifdef HAVE_LSTAT
1148   lstat (filename, stat_buf);
1149 #else
1150   stat (filename, stat_buf);
1151 #endif
1152   st_mode = stat_buf->st_mode;
1153
1154   g_free (stat_buf);
1155
1156   return st_mode;
1157 }
1158
1159 static void
1160 g_scanner_get_token_i (GScanner *scanner,
1161                        GTokenType       *token_p,
1162                        GTokenValue      *value_p,
1163                        guint            *line_p,
1164                        guint            *position_p)
1165 {
1166   do
1167     {
1168       g_scanner_free_value (token_p, value_p);
1169       g_scanner_get_token_ll (scanner, token_p, value_p, line_p, position_p);
1170     }
1171   while (((*token_p > 0 && *token_p < 256) &&
1172           strchr (scanner->config->cset_skip_characters, *token_p)) ||
1173          (*token_p == G_TOKEN_CHAR &&
1174           strchr (scanner->config->cset_skip_characters, value_p->v_char)) ||
1175          (*token_p == G_TOKEN_COMMENT_MULTI &&
1176           scanner->config->skip_comment_multi) ||
1177          (*token_p == G_TOKEN_COMMENT_SINGLE &&
1178           scanner->config->skip_comment_single));
1179
1180   switch (*token_p)
1181     {
1182     case G_TOKEN_IDENTIFIER:
1183       if (scanner->config->identifier_2_string)
1184         *token_p = G_TOKEN_STRING;
1185       break;
1186
1187     case G_TOKEN_SYMBOL:
1188       if (scanner->config->symbol_2_token)
1189         *token_p = (GTokenType) value_p->v_symbol;
1190       break;
1191
1192     case G_TOKEN_BINARY:
1193     case G_TOKEN_OCTAL:
1194     case G_TOKEN_HEX:
1195       if (scanner->config->numbers_2_int)
1196         *token_p = G_TOKEN_INT;
1197       break;
1198
1199     default:
1200       break;
1201     }
1202
1203   if (*token_p == G_TOKEN_INT &&
1204       scanner->config->int_2_float)
1205     {
1206       *token_p = G_TOKEN_FLOAT;
1207       value_p->v_float = value_p->v_int;
1208     }
1209
1210   errno = 0;
1211 }
1212
1213 static void
1214 g_scanner_get_token_ll  (GScanner       *scanner,
1215                          GTokenType     *token_p,
1216                          GTokenValue    *value_p,
1217                          guint          *line_p,
1218                          guint          *position_p)
1219 {
1220   GScannerConfig *config;
1221   GTokenType       token;
1222   gboolean         in_comment_multi;
1223   gboolean         in_comment_single;
1224   gboolean         in_string_sq;
1225   gboolean         in_string_dq;
1226   GString         *gstring;
1227   GTokenValue      value;
1228   guchar           ch;
1229
1230   config = scanner->config;
1231   (*value_p).v_int = 0;
1232
1233   if ((scanner->text >= scanner->text_end && scanner->input_fd < 0) ||
1234       scanner->token == G_TOKEN_EOF)
1235     {
1236       *token_p = G_TOKEN_EOF;
1237       return;
1238     }
1239
1240   in_comment_multi = FALSE;
1241   in_comment_single = FALSE;
1242   in_string_sq = FALSE;
1243   in_string_dq = FALSE;
1244   gstring = NULL;
1245
1246   do /* while (ch != 0) */
1247     {
1248       gboolean dotted_float = FALSE;
1249
1250       ch = g_scanner_get_char (scanner, line_p, position_p);
1251
1252       value.v_int = 0;
1253       token = G_TOKEN_NONE;
1254
1255       /* this is *evil*, but needed ;(
1256        * we first check for identifier first character, because  it
1257        * might interfere with other key chars like slashes or numbers
1258        */
1259       if (config->scan_identifier &&
1260           ch && strchr (config->cset_identifier_first, ch))
1261         goto identifier_precedence;
1262
1263       switch (ch)
1264         {
1265         case 0:
1266           token = G_TOKEN_EOF;
1267           (*position_p)++;
1268           /* ch = 0; */
1269           break;
1270
1271         case '/':
1272           if (!config->scan_comment_multi ||
1273               g_scanner_peek_next_char (scanner) != '*')
1274             goto default_case;
1275           g_scanner_get_char (scanner, line_p, position_p);
1276           token = G_TOKEN_COMMENT_MULTI;
1277           in_comment_multi = TRUE;
1278           gstring = g_string_new ("");
1279           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1280             {
1281               if (ch == '*' && g_scanner_peek_next_char (scanner) == '/')
1282                 {
1283                   g_scanner_get_char (scanner, line_p, position_p);
1284                   in_comment_multi = FALSE;
1285                   break;
1286                 }
1287               else
1288                 gstring = g_string_append_c (gstring, ch);
1289             }
1290           ch = 0;
1291           break;
1292
1293         case '\'':
1294           if (!config->scan_string_sq)
1295             goto default_case;
1296           token = G_TOKEN_STRING;
1297           in_string_sq = TRUE;
1298           gstring = g_string_new ("");
1299           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1300             {
1301               if (ch == '\'')
1302                 {
1303                   in_string_sq = FALSE;
1304                   break;
1305                 }
1306               else
1307                 gstring = g_string_append_c (gstring, ch);
1308             }
1309           ch = 0;
1310           break;
1311
1312         case '"':
1313           if (!config->scan_string_dq)
1314             goto default_case;
1315           token = G_TOKEN_STRING;
1316           in_string_dq = TRUE;
1317           gstring = g_string_new ("");
1318           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1319             {
1320               if (ch == '"')
1321                 {
1322                   in_string_dq = FALSE;
1323                   break;
1324                 }
1325               else
1326                 {
1327                   if (ch == '\\')
1328                     {
1329                       ch = g_scanner_get_char (scanner, line_p, position_p);
1330                       switch (ch)
1331                         {
1332                           guint i;
1333                           guint fchar;
1334
1335                         case 0:
1336                           break;
1337
1338                         case '\\':
1339                           gstring = g_string_append_c (gstring, '\\');
1340                           break;
1341
1342                         case 'n':
1343                           gstring = g_string_append_c (gstring, '\n');
1344                           break;
1345
1346                         case 't':
1347                           gstring = g_string_append_c (gstring, '\t');
1348                           break;
1349
1350                         case 'r':
1351                           gstring = g_string_append_c (gstring, '\r');
1352                           break;
1353
1354                         case 'b':
1355                           gstring = g_string_append_c (gstring, '\b');
1356                           break;
1357
1358                         case 'f':
1359                           gstring = g_string_append_c (gstring, '\f');
1360                           break;
1361
1362                         case '0':
1363                         case '1':
1364                         case '2':
1365                         case '3':
1366                         case '4':
1367                         case '5':
1368                         case '6':
1369                         case '7':
1370                           i = ch - '0';
1371                           fchar = g_scanner_peek_next_char (scanner);
1372                           if (fchar >= '0' && fchar <= '7')
1373                             {
1374                               ch = g_scanner_get_char (scanner, line_p, position_p);
1375                               i = i * 8 + ch - '0';
1376                               fchar = g_scanner_peek_next_char (scanner);
1377                               if (fchar >= '0' && fchar <= '7')
1378                                 {
1379                                   ch = g_scanner_get_char (scanner, line_p, position_p);
1380                                   i = i * 8 + ch - '0';
1381                                 }
1382                             }
1383                           gstring = g_string_append_c (gstring, i);
1384                           break;
1385
1386                         default:
1387                           gstring = g_string_append_c (gstring, ch);
1388                           break;
1389                         }
1390                     }
1391                   else
1392                     gstring = g_string_append_c (gstring, ch);
1393                 }
1394             }
1395           ch = 0;
1396           break;
1397
1398         case '.':
1399           if (!config->scan_float)
1400             goto default_case;
1401           token = G_TOKEN_FLOAT;
1402           dotted_float = TRUE;
1403           ch = g_scanner_get_char (scanner, line_p, position_p);
1404           goto number_parsing;
1405
1406         case '$':
1407           if (!config->scan_hex_dollar)
1408             goto default_case;
1409           token = G_TOKEN_HEX;
1410           ch = g_scanner_get_char (scanner, line_p, position_p);
1411           goto number_parsing;
1412
1413         case '0':
1414           if (config->scan_octal)
1415             token = G_TOKEN_OCTAL;
1416           else
1417             token = G_TOKEN_INT;
1418           ch = g_scanner_peek_next_char (scanner);
1419           if (config->scan_hex && (ch == 'x' || ch == 'X'))
1420             {
1421               token = G_TOKEN_HEX;
1422               g_scanner_get_char (scanner, line_p, position_p);
1423               ch = g_scanner_get_char (scanner, line_p, position_p);
1424               if (ch == 0)
1425                 {
1426                   token = G_TOKEN_ERROR;
1427                   value.v_error = G_ERR_UNEXP_EOF;
1428                   (*position_p)++;
1429                   break;
1430                 }
1431               if (g_scanner_char_2_num (ch, 16) < 0)
1432                 {
1433                   token = G_TOKEN_ERROR;
1434                   value.v_error = G_ERR_DIGIT_RADIX;
1435                   ch = 0;
1436                   break;
1437                 }
1438             }
1439           else if (config->scan_binary && (ch == 'b' || ch == 'B'))
1440             {
1441               token = G_TOKEN_BINARY;
1442               g_scanner_get_char (scanner, line_p, position_p);
1443               ch = g_scanner_get_char (scanner, line_p, position_p);
1444               if (ch == 0)
1445                 {
1446                   token = G_TOKEN_ERROR;
1447                   value.v_error = G_ERR_UNEXP_EOF;
1448                   (*position_p)++;
1449                   break;
1450                 }
1451               if (g_scanner_char_2_num (ch, 10) < 0)
1452                 {
1453                   token = G_TOKEN_ERROR;
1454                   value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1455                   ch = 0;
1456                   break;
1457                 }
1458             }
1459           else
1460             ch = '0';
1461           /* fall through */
1462         case '1':
1463         case '2':
1464         case '3':
1465         case '4':
1466         case '5':
1467         case '6':
1468         case '7':
1469         case '8':
1470         case '9':
1471         number_parsing:
1472         {
1473           gboolean in_number = TRUE;
1474           gchar *endptr;
1475
1476           if (token == G_TOKEN_NONE)
1477             token = G_TOKEN_INT;
1478
1479           gstring = g_string_new (dotted_float ? "0." : "");
1480           gstring = g_string_append_c (gstring, ch);
1481
1482           do /* while (in_number) */
1483             {
1484               gboolean is_E;
1485
1486               is_E = token == G_TOKEN_FLOAT && (ch == 'e' || ch == 'E');
1487
1488               ch = g_scanner_peek_next_char (scanner);
1489
1490               if (g_scanner_char_2_num (ch, 36) >= 0 ||
1491                   (config->scan_float && ch == '.') ||
1492                   (is_E && (ch == '+' || ch == '-')))
1493                 {
1494                   ch = g_scanner_get_char (scanner, line_p, position_p);
1495
1496                   switch (ch)
1497                     {
1498                     case '.':
1499                       if (token != G_TOKEN_INT && token != G_TOKEN_OCTAL)
1500                         {
1501                           value.v_error = token == G_TOKEN_FLOAT ? G_ERR_FLOAT_MALFORMED : G_ERR_FLOAT_RADIX;
1502                           token = G_TOKEN_ERROR;
1503                           in_number = FALSE;
1504                         }
1505                       else
1506                         {
1507                           token = G_TOKEN_FLOAT;
1508                           gstring = g_string_append_c (gstring, ch);
1509                         }
1510                       break;
1511
1512                     case '0':
1513                     case '1':
1514                     case '2':
1515                     case '3':
1516                     case '4':
1517                     case '5':
1518                     case '6':
1519                     case '7':
1520                     case '8':
1521                     case '9':
1522                       gstring = g_string_append_c (gstring, ch);
1523                       break;
1524
1525                     case '-':
1526                     case '+':
1527                       if (token != G_TOKEN_FLOAT)
1528                         {
1529                           token = G_TOKEN_ERROR;
1530                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1531                           in_number = FALSE;
1532                         }
1533                       else
1534                         gstring = g_string_append_c (gstring, ch);
1535                       break;
1536
1537                     case 'e':
1538                     case 'E':
1539                       if ((token != G_TOKEN_HEX && !config->scan_float) ||
1540                           (token != G_TOKEN_HEX &&
1541                            token != G_TOKEN_OCTAL &&
1542                            token != G_TOKEN_FLOAT &&
1543                            token != G_TOKEN_INT))
1544                         {
1545                           token = G_TOKEN_ERROR;
1546                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1547                           in_number = FALSE;
1548                         }
1549                       else
1550                         {
1551                           if (token != G_TOKEN_HEX)
1552                             token = G_TOKEN_FLOAT;
1553                           gstring = g_string_append_c (gstring, ch);
1554                         }
1555                       break;
1556
1557                     default:
1558                       if (token != G_TOKEN_HEX)
1559                         {
1560                           token = G_TOKEN_ERROR;
1561                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1562                           in_number = FALSE;
1563                         }
1564                       else
1565                         gstring = g_string_append_c (gstring, ch);
1566                       break;
1567                     }
1568                 }
1569               else
1570                 in_number = FALSE;
1571             }
1572           while (in_number);
1573
1574           endptr = NULL;
1575           switch (token)
1576             {
1577             case G_TOKEN_BINARY:
1578               value.v_binary = strtol (gstring->str, &endptr, 2);
1579               break;
1580
1581             case G_TOKEN_OCTAL:
1582               value.v_octal = strtol (gstring->str, &endptr, 8);
1583               break;
1584
1585             case G_TOKEN_INT:
1586               value.v_int = strtol (gstring->str, &endptr, 10);
1587               break;
1588
1589             case G_TOKEN_FLOAT:
1590               value.v_float = g_strtod (gstring->str, &endptr);
1591               break;
1592
1593             case G_TOKEN_HEX:
1594               value.v_hex = strtol (gstring->str, &endptr, 16);
1595               break;
1596
1597             default:
1598               break;
1599             }
1600           if (endptr && *endptr)
1601             {
1602               token = G_TOKEN_ERROR;
1603               if (*endptr == 'e' || *endptr == 'E')
1604                 value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1605               else
1606                 value.v_error = G_ERR_DIGIT_RADIX;
1607             }
1608           g_string_free (gstring, TRUE);
1609           gstring = NULL;
1610           ch = 0;
1611         } /* number_parsing:... */
1612         break;
1613
1614         default:
1615         default_case:
1616         {
1617           if (config->cpair_comment_single &&
1618               ch == config->cpair_comment_single[0])
1619             {
1620               token = G_TOKEN_COMMENT_SINGLE;
1621               in_comment_single = TRUE;
1622               gstring = g_string_new ("");
1623               ch = g_scanner_get_char (scanner, line_p, position_p);
1624               while (ch != 0)
1625                 {
1626                   if (ch == config->cpair_comment_single[1])
1627                     {
1628                       in_comment_single = FALSE;
1629                       ch = 0;
1630                       break;
1631                     }
1632
1633                   gstring = g_string_append_c (gstring, ch);
1634                   ch = g_scanner_get_char (scanner, line_p, position_p);
1635                 }
1636             }
1637           else if (config->scan_identifier && ch &&
1638                    strchr (config->cset_identifier_first, ch))
1639             {
1640             identifier_precedence:
1641
1642               if (config->cset_identifier_nth && ch &&
1643                   strchr (config->cset_identifier_nth,
1644                           g_scanner_peek_next_char (scanner)))
1645                 {
1646                   token = G_TOKEN_IDENTIFIER;
1647                   gstring = g_string_new ("");
1648                   gstring = g_string_append_c (gstring, ch);
1649                   do
1650                     {
1651                       ch = g_scanner_get_char (scanner, line_p, position_p);
1652                       gstring = g_string_append_c (gstring, ch);
1653                       ch = g_scanner_peek_next_char (scanner);
1654                     }
1655                   while (ch && strchr (config->cset_identifier_nth, ch));
1656                   ch = 0;
1657                 }
1658               else if (config->scan_identifier_1char)
1659                 {
1660                   token = G_TOKEN_IDENTIFIER;
1661                   value.v_identifier = g_new0 (gchar, 2);
1662                   value.v_identifier[0] = ch;
1663                   ch = 0;
1664                 }
1665             }
1666           if (ch)
1667             {
1668               if (config->char_2_token)
1669                 token = ch;
1670               else
1671                 {
1672                   token = G_TOKEN_CHAR;
1673                   value.v_char = ch;
1674                 }
1675               ch = 0;
1676             }
1677         } /* default_case:... */
1678         break;
1679         }
1680       g_assert (ch == 0 && token != G_TOKEN_NONE); /* paranoid */
1681     }
1682   while (ch != 0);
1683
1684   if (in_comment_multi || in_comment_single ||
1685       in_string_sq || in_string_dq)
1686     {
1687       token = G_TOKEN_ERROR;
1688       if (gstring)
1689         {
1690           g_string_free (gstring, TRUE);
1691           gstring = NULL;
1692         }
1693       (*position_p)++;
1694       if (in_comment_multi || in_comment_single)
1695         value.v_error = G_ERR_UNEXP_EOF_IN_COMMENT;
1696       else /* (in_string_sq || in_string_dq) */
1697         value.v_error = G_ERR_UNEXP_EOF_IN_STRING;
1698     }
1699
1700   if (gstring)
1701     {
1702       value.v_string = gstring->str;
1703       g_string_free (gstring, FALSE);
1704       gstring = NULL;
1705     }
1706
1707   if (token == G_TOKEN_IDENTIFIER)
1708     {
1709       if (config->scan_symbols)
1710         {
1711           GScannerKey *key;
1712           guint scope_id;
1713
1714           scope_id = scanner->scope_id;
1715           key = g_scanner_lookup_internal (scanner, scope_id, value.v_identifier);
1716           if (!key && scope_id && scanner->config->scope_0_fallback)
1717             key = g_scanner_lookup_internal (scanner, 0, value.v_identifier);
1718
1719           if (key)
1720             {
1721               g_free (value.v_identifier);
1722               token = G_TOKEN_SYMBOL;
1723               value.v_symbol = key->value;
1724             }
1725         }
1726
1727       if (token == G_TOKEN_IDENTIFIER &&
1728           config->scan_identifier_NULL &&
1729           strlen (value.v_identifier) == 4)
1730         {
1731           gchar *null_upper = "NULL";
1732           gchar *null_lower = "null";
1733
1734           if (scanner->config->case_sensitive)
1735             {
1736               if (value.v_identifier[0] == null_upper[0] &&
1737                   value.v_identifier[1] == null_upper[1] &&
1738                   value.v_identifier[2] == null_upper[2] &&
1739                   value.v_identifier[3] == null_upper[3])
1740                 token = G_TOKEN_IDENTIFIER_NULL;
1741             }
1742           else
1743             {
1744               if ((value.v_identifier[0] == null_upper[0] ||
1745                    value.v_identifier[0] == null_lower[0]) &&
1746                   (value.v_identifier[1] == null_upper[1] ||
1747                    value.v_identifier[1] == null_lower[1]) &&
1748                   (value.v_identifier[2] == null_upper[2] ||
1749                    value.v_identifier[2] == null_lower[2]) &&
1750                   (value.v_identifier[3] == null_upper[3] ||
1751                    value.v_identifier[3] == null_lower[3]))
1752                 token = G_TOKEN_IDENTIFIER_NULL;
1753             }
1754         }
1755     }
1756
1757   *token_p = token;
1758   *value_p = value;
1759 }