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 GTokenType
 567 g_scanner_peek_next_token (GScanner     *scanner)
 568 {
 569   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 570
 571   if (scanner->next_token == G_TOKEN_NONE)
 572     {
 573       scanner->next_line = scanner->line;
 574       scanner->next_position = scanner->position;
 575       g_scanner_get_token_i (scanner,
 576                              &scanner->next_token,
 577                              &scanner->next_value,
 578                              &scanner->next_line,
 579                              &scanner->next_position);
 580     }
 581
 582   return scanner->next_token;
 583 }
 584
 585 GTokenType
 586 g_scanner_get_next_token (GScanner      *scanner)
 587 {
 588   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 589
 590   if (scanner->next_token != G_TOKEN_NONE)
 591     {
 592       g_scanner_free_value (&scanner->token, &scanner->value);
 593
 594       scanner->token = scanner->next_token;
 595       scanner->value = scanner->next_value;
 596       scanner->line = scanner->next_line;
 597       scanner->position = scanner->next_position;
 598       scanner->next_token = G_TOKEN_NONE;
 599     }
 600   else
 601     g_scanner_get_token_i (scanner,
 602                            &scanner->token,
 603                            &scanner->value,
 604                            &scanner->line,
 605                            &scanner->position);
 606
 607   return scanner->token;
 608 }
 609
 610 GTokenType
 611 g_scanner_cur_token (GScanner *scanner)
 612 {
 613   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 614
 615   return scanner->token;
 616 }
 617
 618 GTokenValue
 619 g_scanner_cur_value (GScanner *scanner)
 620 {
 621   GTokenValue v;
 622
 623   v.v_int = 0;
 624
 625   g_return_val_if_fail (scanner != NULL, v);
 626
 627   /* MSC isn't capable of handling return scanner->value; ? */
 628
 629   v = scanner->value;
 630
 631   return v;
 632 }
 633
 634 guint
 635 g_scanner_cur_line (GScanner *scanner)
 636 {
 637   g_return_val_if_fail (scanner != NULL, 0);
 638
 639   return scanner->line;
 640 }
 641
 642 guint
 643 g_scanner_cur_position (GScanner *scanner)
 644 {
 645   g_return_val_if_fail (scanner != NULL, 0);
 646
 647   return scanner->position;
 648 }
 649
 650 gboolean
 651 g_scanner_eof (GScanner *scanner)
 652 {
 653   g_return_val_if_fail (scanner != NULL, TRUE);
 654
 655   return scanner->token == G_TOKEN_EOF;
 656 }
 657
 658 void
 659 g_scanner_input_file (GScanner *scanner,
 660                       gint      input_fd)
 661 {
 662   g_return_if_fail (scanner != NULL);
 663   g_return_if_fail (input_fd >= 0);
 664
 665   if (scanner->input_fd >= 0)
 666     g_scanner_sync_file_offset (scanner);
 667
 668   scanner->token = G_TOKEN_NONE;
 669   scanner->value.v_int = 0;
 670   scanner->line = 1;
 671   scanner->position = 0;
 672   scanner->next_token = G_TOKEN_NONE;
 673
 674   scanner->input_fd = input_fd;
 675   scanner->text = NULL;
 676   scanner->text_end = NULL;
 677
 678   if (!scanner->buffer)
 679     scanner->buffer = g_new (gchar, READ_BUFFER_SIZE + 1);
 680 }
 681
 682 void
 683 g_scanner_input_text (GScanner    *scanner,
 684                       const gchar *text,
 685                       guint        text_len)
 686 {
 687   g_return_if_fail (scanner != NULL);
 688   if (text_len)
 689     g_return_if_fail (text != NULL);
 690   else
 691     text = NULL;
 692
 693   if (scanner->input_fd >= 0)
 694     g_scanner_sync_file_offset (scanner);
 695
 696   scanner->token = G_TOKEN_NONE;
 697   scanner->value.v_int = 0;
 698   scanner->line = 1;
 699   scanner->position = 0;
 700   scanner->next_token = G_TOKEN_NONE;
 701
 702   scanner->input_fd = -1;
 703   scanner->text = text;
 704   scanner->text_end = text + text_len;
 705
 706   if (scanner->buffer)
 707     {
 708       g_free (scanner->buffer);
 709       scanner->buffer = NULL;
 710     }
 711 }
 712
 713 static guchar
 714 g_scanner_peek_next_char (GScanner *scanner)
 715 {
 716   if (scanner->text < scanner->text_end)
 717     {
 718       return *scanner->text;
 719     }
 720   else if (scanner->input_fd >= 0)
 721     {
 722       gint count;
 723       gchar *buffer;
 724
 725       buffer = scanner->buffer;
 726       do
 727         {
 728           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 729         }
 730       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 731
 732       if (count < 1)
 733         {
 734           scanner->input_fd = -1;
 735
 736           return 0;
 737         }
 738       else
 739         {
 740           scanner->text = buffer;
 741           scanner->text_end = buffer + count;
 742
 743           return *buffer;
 744         }
 745     }
 746   else
 747     return 0;
 748 }
 749
 750 void
 751 g_scanner_sync_file_offset (GScanner *scanner)
 752 {
 753   g_return_if_fail (scanner != NULL);
 754
 755   /* for file input, rewind the filedescriptor to the current
 756    * buffer position and blow the file read ahead buffer. usefull for
 757    * third party uses of our filedescriptor, which hooks onto the current
 758    * scanning position.
 759    */
 760
 761   if (scanner->input_fd >= 0 && scanner->text_end > scanner->text)
 762     {
 763       gint buffered;
 764
 765       buffered = scanner->text_end - scanner->text;
 766       if (lseek (scanner->input_fd, - buffered, SEEK_CUR) >= 0)
 767         {
 768           /* we succeeded, blow our buffer's contents now */
 769           scanner->text = NULL;
 770           scanner->text_end = NULL;
 771         }
 772       else
 773         errno = 0;
 774     }
 775 }
 776
 777 static guchar
 778 g_scanner_get_char (GScanner    *scanner,
 779                     guint       *line_p,
 780                     guint       *position_p)
 781 {
 782   guchar fchar;
 783
 784   if (scanner->text < scanner->text_end)
 785     fchar = *(scanner->text++);
 786   else if (scanner->input_fd >= 0)
 787     {
 788       gint count;
 789       gchar *buffer;
 790
 791       buffer = scanner->buffer;
 792       do
 793         {
 794           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 795         }
 796       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 797
 798       if (count < 1)
 799         {
 800           scanner->input_fd = -1;
 801           fchar = 0;
 802         }
 803       else
 804         {
 805           scanner->text = buffer + 1;
 806           scanner->text_end = buffer + count;
 807           fchar = *buffer;
 808           if (!fchar)
 809             {
 810               g_scanner_sync_file_offset (scanner);
 811               scanner->text_end = scanner->text;
 812               scanner->input_fd = -1;
 813             }
 814         }
 815     }
 816   else
 817     fchar = 0;
 818
 819   if (fchar == '\n')
 820     {
 821       (*position_p) = 0;
 822       (*line_p)++;
 823     }
 824   else if (fchar)
 825     {
 826       (*position_p)++;
 827     }
 828
 829   return fchar;
 830 }
 831
 832 void
 833 g_scanner_unexp_token (GScanner         *scanner,
 834                        GTokenType        expected_token,
 835                        const gchar      *identifier_spec,
 836                        const gchar      *symbol_spec,
 837                        const gchar      *symbol_name,
 838                        const gchar      *message,
 839                        gint              is_error)
 840 {
 841   gchar *token_string;
 842   guint token_string_len;
 843   gchar *expected_string;
 844   guint expected_string_len;
 845   gchar *message_prefix;
 846   gboolean print_unexp;
 847   void (*msg_handler)   (GScanner*, const gchar*, ...);
 848
 849   g_return_if_fail (scanner != NULL);
 850
 851   if (is_error)
 852     msg_handler = g_scanner_error;
 853   else
 854     msg_handler = g_scanner_warn;
 855
 856   if (!identifier_spec)
 857     identifier_spec = "identifier";
 858   if (!symbol_spec)
 859     symbol_spec = "symbol";
 860
 861   token_string_len = 56;
 862   token_string = g_new (gchar, token_string_len + 1);
 863   expected_string_len = 64;
 864   expected_string = g_new (gchar, expected_string_len + 1);
 865   print_unexp = TRUE;
 866
 867   switch (scanner->token)
 868     {
 869     case G_TOKEN_EOF:
 870       g_snprintf (token_string, token_string_len, "end of file");
 871       break;
 872
 873     default:
 874       if (scanner->token >= 1 && scanner->token <= 255)
 875         {
 876           if ((scanner->token >= ' ' && scanner->token <= '~') ||
 877               strchr (scanner->config->cset_identifier_first, scanner->token) ||
 878               strchr (scanner->config->cset_identifier_nth, scanner->token))
 879             g_snprintf (token_string, expected_string_len, "character `%c'", scanner->token);
 880           else
 881             g_snprintf (token_string, expected_string_len, "character `\\%o'", scanner->token);
 882           break;
 883         }
 884       else if (!scanner->config->symbol_2_token)
 885         {
 886           g_snprintf (token_string, token_string_len, "(unknown) token <%d>", scanner->token);
 887           break;
 888         }
 889       /* fall through */
 890     case G_TOKEN_SYMBOL:
 891       if (expected_token == G_TOKEN_SYMBOL ||
 892           (scanner->config->symbol_2_token &&
 893            expected_token > G_TOKEN_LAST))
 894         print_unexp = FALSE;
 895       if (symbol_name)
 896         g_snprintf (token_string,
 897                     token_string_len,
 898                     "%s%s `%s'",
 899                     print_unexp ? "" : "invalid ",
 900                     symbol_spec,
 901                     symbol_name);
 902       else
 903         g_snprintf (token_string,
 904                     token_string_len,
 905                     "%s%s",
 906                     print_unexp ? "" : "invalid ",
 907                     symbol_spec);
 908       break;
 909
 910     case G_TOKEN_ERROR:
 911       print_unexp = FALSE;
 912       expected_token = G_TOKEN_NONE;
 913       switch (scanner->value.v_error)
 914         {
 915         case G_ERR_UNEXP_EOF:
 916           g_snprintf (token_string, token_string_len, "scanner: unexpected end of file");
 917           break;
 918
 919         case G_ERR_UNEXP_EOF_IN_STRING:
 920           g_snprintf (token_string, token_string_len, "scanner: unterminated string constant");
 921           break;
 922
 923         case G_ERR_UNEXP_EOF_IN_COMMENT:
 924           g_snprintf (token_string, token_string_len, "scanner: unterminated comment");
 925           break;
 926
 927         case G_ERR_NON_DIGIT_IN_CONST:
 928           g_snprintf (token_string, token_string_len, "scanner: non digit in constant");
 929           break;
 930
 931         case G_ERR_FLOAT_RADIX:
 932           g_snprintf (token_string, token_string_len, "scanner: invalid radix for floating constant");
 933           break;
 934
 935         case G_ERR_FLOAT_MALFORMED:
 936           g_snprintf (token_string, token_string_len, "scanner: malformed floating constant");
 937           break;
 938
 939         case G_ERR_DIGIT_RADIX:
 940           g_snprintf (token_string, token_string_len, "scanner: digit is beyond radix");
 941           break;
 942
 943         case G_ERR_UNKNOWN:
 944         default:
 945           g_snprintf (token_string, token_string_len, "scanner: unknown error");
 946           break;
 947         }
 948       break;
 949
 950     case G_TOKEN_CHAR:
 951       g_snprintf (token_string, token_string_len, "character `%c'", scanner->value.v_char);
 952       break;
 953
 954     case G_TOKEN_IDENTIFIER:
 955     case G_TOKEN_IDENTIFIER_NULL:
 956       if (expected_token == G_TOKEN_IDENTIFIER ||
 957           expected_token == G_TOKEN_IDENTIFIER_NULL)
 958         print_unexp = FALSE;
 959       g_snprintf (token_string,
 960                   token_string_len,
 961                   "%s%s `%s'",
 962                   print_unexp ? "" : "invalid ",
 963                   identifier_spec,
 964                   scanner->token == G_TOKEN_IDENTIFIER ? scanner->value.v_string : "null");
 965       break;
 966
 967     case G_TOKEN_BINARY:
 968     case G_TOKEN_OCTAL:
 969     case G_TOKEN_INT:
 970     case G_TOKEN_HEX:
 971       g_snprintf (token_string, token_string_len, "number `%ld'", scanner->value.v_int);
 972       break;
 973
 974     case G_TOKEN_FLOAT:
 975       g_snprintf (token_string, token_string_len, "number `%.3f'", scanner->value.v_float);
 976       break;
 977
 978     case G_TOKEN_STRING:
 979       if (expected_token == G_TOKEN_STRING)
 980         print_unexp = FALSE;
 981       g_snprintf (token_string,
 982                   token_string_len,
 983                   "%s%sstring constant \"%s\"",
 984                   print_unexp ? "" : "invalid ",
 985                   scanner->value.v_string[0] == 0 ? "empty " : "",
 986                   scanner->value.v_string);
 987       token_string[token_string_len - 2] = '"';
 988       token_string[token_string_len - 1] = 0;
 989       break;
 990
 991     case G_TOKEN_COMMENT_SINGLE:
 992     case G_TOKEN_COMMENT_MULTI:
 993       g_snprintf (token_string, token_string_len, "comment");
 994       break;
 995
 996     case G_TOKEN_NONE:
 997       /* somehow the user's parsing code is screwed, there isn't much
 998        * we can do about it.
 999        * Note, a common case to trigger this is
1000        * g_scanner_peek_next_token(); g_scanner_unexp_token();
1001        * without an intermediate g_scanner_get_next_token().
1002        */
1003       g_assert_not_reached ();
1004       break;
1005     }
1006
1007
1008   switch (expected_token)
1009     {
1010       gboolean need_valid;
1011
1012     default:
1013       if (expected_token >= 1 && expected_token <= 255)
1014         {
1015           if ((expected_token >= ' ' && expected_token <= '~') ||
1016               strchr (scanner->config->cset_identifier_first, expected_token) ||
1017               strchr (scanner->config->cset_identifier_nth, expected_token))
1018             g_snprintf (expected_string, expected_string_len, "character `%c'", expected_token);
1019           else
1020             g_snprintf (expected_string, expected_string_len, "character `\\%o'", expected_token);
1021           break;
1022         }
1023       else if (!scanner->config->symbol_2_token)
1024         {
1025           g_snprintf (expected_string, expected_string_len, "(unknown) token <%d>", expected_token);
1026           break;
1027         }
1028       /* fall through */
1029     case G_TOKEN_SYMBOL:
1030       need_valid = (scanner->token == G_TOKEN_SYMBOL ||
1031                     (scanner->config->symbol_2_token &&
1032                      scanner->token > G_TOKEN_LAST));
1033       g_snprintf (expected_string,
1034                   expected_string_len,
1035                   "%s%s",
1036                   need_valid ? "valid " : "",
1037                   symbol_spec);
1038       /* FIXME: should we attempt to lookup the symbol_name for symbol_2_token? */
1039       break;
1040
1041     case G_TOKEN_INT:
1042       g_snprintf (expected_string, expected_string_len, "number (integer)");
1043       break;
1044
1045     case G_TOKEN_FLOAT:
1046       g_snprintf (expected_string, expected_string_len, "number (float)");
1047       break;
1048
1049     case G_TOKEN_STRING:
1050       g_snprintf (expected_string,
1051                   expected_string_len,
1052                   "%sstring constant",
1053                   scanner->token == G_TOKEN_STRING ? "valid " : "");
1054       break;
1055
1056     case G_TOKEN_IDENTIFIER:
1057     case G_TOKEN_IDENTIFIER_NULL:
1058       g_snprintf (expected_string,
1059                   expected_string_len,
1060                   "%s%s",
1061                   (scanner->token == G_TOKEN_IDENTIFIER_NULL ||
1062                    scanner->token == G_TOKEN_IDENTIFIER ? "valid " : ""),
1063                   identifier_spec);
1064       break;
1065
1066     case G_TOKEN_EOF:
1067       g_snprintf (expected_string, expected_string_len, "end of file");
1068       break;
1069
1070     case G_TOKEN_NONE:
1071       break;
1072     }
1073
1074   if (message && message[0] != 0)
1075     message_prefix = " - ";
1076   else
1077     {
1078       message_prefix = "";
1079       message = "";
1080     }
1081
1082   if (expected_token != G_TOKEN_NONE)
1083     {
1084       if (print_unexp)
1085         msg_handler (scanner,
1086                      "unexpected %s, expected %s%s%s",
1087                      token_string,
1088                      expected_string,
1089                      message_prefix,
1090                      message);
1091       else
1092         msg_handler (scanner,
1093                      "%s, expected %s%s%s",
1094                      token_string,
1095                      expected_string,
1096                      message_prefix,
1097                      message);
1098     }
1099   else
1100     {
1101       if (print_unexp)
1102         msg_handler (scanner,
1103                      "unexpected %s%s%s",
1104                      token_string,
1105                      message_prefix,
1106                      message);
1107       else
1108         msg_handler (scanner,
1109                      "%s%s%s",
1110                      token_string,
1111                      message_prefix,
1112                      message);
1113     }
1114
1115   g_free (token_string);
1116   g_free (expected_string);
1117 }
1118
1119 gint
1120 g_scanner_stat_mode (const gchar *filename)
1121 {
1122   struct stat  *stat_buf;
1123   gint          st_mode;
1124
1125   stat_buf = g_new0 (struct stat, 1);
1126 #ifdef HAVE_LSTAT
1127   lstat (filename, stat_buf);
1128 #else
1129   stat (filename, stat_buf);
1130 #endif
1131   st_mode = stat_buf->st_mode;
1132
1133   g_free (stat_buf);
1134
1135   return st_mode;
1136 }
1137
1138 static void
1139 g_scanner_get_token_i (GScanner *scanner,
1140                        GTokenType       *token_p,
1141                        GTokenValue      *value_p,
1142                        guint            *line_p,
1143                        guint            *position_p)
1144 {
1145   do
1146     {
1147       g_scanner_free_value (token_p, value_p);
1148       g_scanner_get_token_ll (scanner, token_p, value_p, line_p, position_p);
1149     }
1150   while (((*token_p > 0 && *token_p < 256) &&
1151           strchr (scanner->config->cset_skip_characters, *token_p)) ||
1152          (*token_p == G_TOKEN_CHAR &&
1153           strchr (scanner->config->cset_skip_characters, value_p->v_char)) ||
1154          (*token_p == G_TOKEN_COMMENT_MULTI &&
1155           scanner->config->skip_comment_multi) ||
1156          (*token_p == G_TOKEN_COMMENT_SINGLE &&
1157           scanner->config->skip_comment_single));
1158
1159   switch (*token_p)
1160     {
1161     case G_TOKEN_IDENTIFIER:
1162       if (scanner->config->identifier_2_string)
1163         *token_p = G_TOKEN_STRING;
1164       break;
1165
1166     case G_TOKEN_SYMBOL:
1167       if (scanner->config->symbol_2_token)
1168         *token_p = (GTokenType) value_p->v_symbol;
1169       break;
1170
1171     case G_TOKEN_BINARY:
1172     case G_TOKEN_OCTAL:
1173     case G_TOKEN_HEX:
1174       if (scanner->config->numbers_2_int)
1175         *token_p = G_TOKEN_INT;
1176       break;
1177
1178     default:
1179       break;
1180     }
1181
1182   if (*token_p == G_TOKEN_INT &&
1183       scanner->config->int_2_float)
1184     {
1185       *token_p = G_TOKEN_FLOAT;
1186       value_p->v_float = value_p->v_int;
1187     }
1188
1189   errno = 0;
1190 }
1191
1192 static void
1193 g_scanner_get_token_ll  (GScanner       *scanner,
1194                          GTokenType     *token_p,
1195                          GTokenValue    *value_p,
1196                          guint          *line_p,
1197                          guint          *position_p)
1198 {
1199   GScannerConfig *config;
1200   GTokenType       token;
1201   gboolean         in_comment_multi;
1202   gboolean         in_comment_single;
1203   gboolean         in_string_sq;
1204   gboolean         in_string_dq;
1205   GString         *gstring;
1206   GTokenValue      value;
1207   guchar           ch;
1208
1209   config = scanner->config;
1210   (*value_p).v_int = 0;
1211
1212   if ((scanner->text >= scanner->text_end && scanner->input_fd < 0) ||
1213       scanner->token == G_TOKEN_EOF)
1214     {
1215       *token_p = G_TOKEN_EOF;
1216       return;
1217     }
1218
1219   in_comment_multi = FALSE;
1220   in_comment_single = FALSE;
1221   in_string_sq = FALSE;
1222   in_string_dq = FALSE;
1223   gstring = NULL;
1224
1225   do /* while (ch != 0) */
1226     {
1227       gboolean dotted_float = FALSE;
1228
1229       ch = g_scanner_get_char (scanner, line_p, position_p);
1230
1231       value.v_int = 0;
1232       token = G_TOKEN_NONE;
1233
1234       /* this is *evil*, but needed ;(
1235        * we first check for identifier first character, because  it
1236        * might interfere with other key chars like slashes or numbers
1237        */
1238       if (config->scan_identifier &&
1239           ch && strchr (config->cset_identifier_first, ch))
1240         goto identifier_precedence;
1241
1242       switch (ch)
1243         {
1244         case 0:
1245           token = G_TOKEN_EOF;
1246           (*position_p)++;
1247           /* ch = 0; */
1248           break;
1249
1250         case '/':
1251           if (!config->scan_comment_multi ||
1252               g_scanner_peek_next_char (scanner) != '*')
1253             goto default_case;
1254           g_scanner_get_char (scanner, line_p, position_p);
1255           token = G_TOKEN_COMMENT_MULTI;
1256           in_comment_multi = TRUE;
1257           gstring = g_string_new ("");
1258           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1259             {
1260               if (ch == '*' && g_scanner_peek_next_char (scanner) == '/')
1261                 {
1262                   g_scanner_get_char (scanner, line_p, position_p);
1263                   in_comment_multi = FALSE;
1264                   break;
1265                 }
1266               else
1267                 gstring = g_string_append_c (gstring, ch);
1268             }
1269           ch = 0;
1270           break;
1271
1272         case '\'':
1273           if (!config->scan_string_sq)
1274             goto default_case;
1275           token = G_TOKEN_STRING;
1276           in_string_sq = TRUE;
1277           gstring = g_string_new ("");
1278           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1279             {
1280               if (ch == '\'')
1281                 {
1282                   in_string_sq = FALSE;
1283                   break;
1284                 }
1285               else
1286                 gstring = g_string_append_c (gstring, ch);
1287             }
1288           ch = 0;
1289           break;
1290
1291         case '"':
1292           if (!config->scan_string_dq)
1293             goto default_case;
1294           token = G_TOKEN_STRING;
1295           in_string_dq = TRUE;
1296           gstring = g_string_new ("");
1297           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1298             {
1299               if (ch == '"')
1300                 {
1301                   in_string_dq = FALSE;
1302                   break;
1303                 }
1304               else
1305                 {
1306                   if (ch == '\\')
1307                     {
1308                       ch = g_scanner_get_char (scanner, line_p, position_p);
1309                       switch (ch)
1310                         {
1311                           guint i;
1312                           guint fchar;
1313
1314                         case 0:
1315                           break;
1316
1317                         case '\\':
1318                           gstring = g_string_append_c (gstring, '\\');
1319                           break;
1320
1321                         case 'n':
1322                           gstring = g_string_append_c (gstring, '\n');
1323                           break;
1324
1325                         case 't':
1326                           gstring = g_string_append_c (gstring, '\t');
1327                           break;
1328
1329                         case 'r':
1330                           gstring = g_string_append_c (gstring, '\r');
1331                           break;
1332
1333                         case 'b':
1334                           gstring = g_string_append_c (gstring, '\b');
1335                           break;
1336
1337                         case 'f':
1338                           gstring = g_string_append_c (gstring, '\f');
1339                           break;
1340
1341                         case '0':
1342                         case '1':
1343                         case '2':
1344                         case '3':
1345                         case '4':
1346                         case '5':
1347                         case '6':
1348                         case '7':
1349                           i = ch - '0';
1350                           fchar = g_scanner_peek_next_char (scanner);
1351                           if (fchar >= '0' && fchar <= '7')
1352                             {
1353                               ch = g_scanner_get_char (scanner, line_p, position_p);
1354                               i = i * 8 + ch - '0';
1355                               fchar = g_scanner_peek_next_char (scanner);
1356                               if (fchar >= '0' && fchar <= '7')
1357                                 {
1358                                   ch = g_scanner_get_char (scanner, line_p, position_p);
1359                                   i = i * 8 + ch - '0';
1360                                 }
1361                             }
1362                           gstring = g_string_append_c (gstring, i);
1363                           break;
1364
1365                         default:
1366                           gstring = g_string_append_c (gstring, ch);
1367                           break;
1368                         }
1369                     }
1370                   else
1371                     gstring = g_string_append_c (gstring, ch);
1372                 }
1373             }
1374           ch = 0;
1375           break;
1376
1377         case '.':
1378           if (!config->scan_float)
1379             goto default_case;
1380           token = G_TOKEN_FLOAT;
1381           dotted_float = TRUE;
1382           ch = g_scanner_get_char (scanner, line_p, position_p);
1383           goto number_parsing;
1384
1385         case '$':
1386           if (!config->scan_hex_dollar)
1387             goto default_case;
1388           token = G_TOKEN_HEX;
1389           ch = g_scanner_get_char (scanner, line_p, position_p);
1390           goto number_parsing;
1391
1392         case '0':
1393           if (config->scan_octal)
1394             token = G_TOKEN_OCTAL;
1395           else
1396             token = G_TOKEN_INT;
1397           ch = g_scanner_peek_next_char (scanner);
1398           if (config->scan_hex && (ch == 'x' || ch == 'X'))
1399             {
1400               token = G_TOKEN_HEX;
1401               g_scanner_get_char (scanner, line_p, position_p);
1402               ch = g_scanner_get_char (scanner, line_p, position_p);
1403               if (ch == 0)
1404                 {
1405                   token = G_TOKEN_ERROR;
1406                   value.v_error = G_ERR_UNEXP_EOF;
1407                   (*position_p)++;
1408                   break;
1409                 }
1410               if (g_scanner_char_2_num (ch, 16) < 0)
1411                 {
1412                   token = G_TOKEN_ERROR;
1413                   value.v_error = G_ERR_DIGIT_RADIX;
1414                   ch = 0;
1415                   break;
1416                 }
1417             }
1418           else if (config->scan_binary && (ch == 'b' || ch == 'B'))
1419             {
1420               token = G_TOKEN_BINARY;
1421               g_scanner_get_char (scanner, line_p, position_p);
1422               ch = g_scanner_get_char (scanner, line_p, position_p);
1423               if (ch == 0)
1424                 {
1425                   token = G_TOKEN_ERROR;
1426                   value.v_error = G_ERR_UNEXP_EOF;
1427                   (*position_p)++;
1428                   break;
1429                 }
1430               if (g_scanner_char_2_num (ch, 10) < 0)
1431                 {
1432                   token = G_TOKEN_ERROR;
1433                   value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1434                   ch = 0;
1435                   break;
1436                 }
1437             }
1438           else
1439             ch = '0';
1440           /* fall through */
1441         case '1':
1442         case '2':
1443         case '3':
1444         case '4':
1445         case '5':
1446         case '6':
1447         case '7':
1448         case '8':
1449         case '9':
1450         number_parsing:
1451         {
1452           gboolean in_number = TRUE;
1453           gchar *endptr;
1454
1455           if (token == G_TOKEN_NONE)
1456             token = G_TOKEN_INT;
1457
1458           gstring = g_string_new (dotted_float ? "0." : "");
1459           gstring = g_string_append_c (gstring, ch);
1460
1461           do /* while (in_number) */
1462             {
1463               gboolean is_E;
1464
1465               is_E = token == G_TOKEN_FLOAT && (ch == 'e' || ch == 'E');
1466
1467               ch = g_scanner_peek_next_char (scanner);
1468
1469               if (g_scanner_char_2_num (ch, 36) >= 0 ||
1470                   (config->scan_float && ch == '.') ||
1471                   (is_E && (ch == '+' || ch == '-')))
1472                 {
1473                   ch = g_scanner_get_char (scanner, line_p, position_p);
1474
1475                   switch (ch)
1476                     {
1477                     case '.':
1478                       if (token != G_TOKEN_INT && token != G_TOKEN_OCTAL)
1479                         {
1480                           value.v_error = token == G_TOKEN_FLOAT ? G_ERR_FLOAT_MALFORMED : G_ERR_FLOAT_RADIX;
1481                           token = G_TOKEN_ERROR;
1482                           in_number = FALSE;
1483                         }
1484                       else
1485                         {
1486                           token = G_TOKEN_FLOAT;
1487                           gstring = g_string_append_c (gstring, ch);
1488                         }
1489                       break;
1490
1491                     case '0':
1492                     case '1':
1493                     case '2':
1494                     case '3':
1495                     case '4':
1496                     case '5':
1497                     case '6':
1498                     case '7':
1499                     case '8':
1500                     case '9':
1501                       gstring = g_string_append_c (gstring, ch);
1502                       break;
1503
1504                     case '-':
1505                     case '+':
1506                       if (token != G_TOKEN_FLOAT)
1507                         {
1508                           token = G_TOKEN_ERROR;
1509                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1510                           in_number = FALSE;
1511                         }
1512                       else
1513                         gstring = g_string_append_c (gstring, ch);
1514                       break;
1515
1516                     case 'e':
1517                     case 'E':
1518                       if ((token != G_TOKEN_HEX && !config->scan_float) ||
1519                           (token != G_TOKEN_HEX &&
1520                            token != G_TOKEN_OCTAL &&
1521                            token != G_TOKEN_FLOAT &&
1522                            token != G_TOKEN_INT))
1523                         {
1524                           token = G_TOKEN_ERROR;
1525                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1526                           in_number = FALSE;
1527                         }
1528                       else
1529                         {
1530                           if (token != G_TOKEN_HEX)
1531                             token = G_TOKEN_FLOAT;
1532                           gstring = g_string_append_c (gstring, ch);
1533                         }
1534                       break;
1535
1536                     default:
1537                       if (token != G_TOKEN_HEX)
1538                         {
1539                           token = G_TOKEN_ERROR;
1540                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1541                           in_number = FALSE;
1542                         }
1543                       else
1544                         gstring = g_string_append_c (gstring, ch);
1545                       break;
1546                     }
1547                 }
1548               else
1549                 in_number = FALSE;
1550             }
1551           while (in_number);
1552
1553           endptr = NULL;
1554           switch (token)
1555             {
1556             case G_TOKEN_BINARY:
1557               value.v_binary = strtol (gstring->str, &endptr, 2);
1558               break;
1559
1560             case G_TOKEN_OCTAL:
1561               value.v_octal = strtol (gstring->str, &endptr, 8);
1562               break;
1563
1564             case G_TOKEN_INT:
1565               value.v_int = strtol (gstring->str, &endptr, 10);
1566               break;
1567
1568             case G_TOKEN_FLOAT:
1569               value.v_float = g_strtod (gstring->str, &endptr);
1570               break;
1571
1572             case G_TOKEN_HEX:
1573               value.v_hex = strtol (gstring->str, &endptr, 16);
1574               break;
1575
1576             default:
1577               break;
1578             }
1579           if (endptr && *endptr)
1580             {
1581               token = G_TOKEN_ERROR;
1582               if (*endptr == 'e' || *endptr == 'E')
1583                 value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1584               else
1585                 value.v_error = G_ERR_DIGIT_RADIX;
1586             }
1587           g_string_free (gstring, TRUE);
1588           gstring = NULL;
1589           ch = 0;
1590         } /* number_parsing:... */
1591         break;
1592
1593         default:
1594         default_case:
1595         {
1596           if (config->cpair_comment_single &&
1597               ch == config->cpair_comment_single[0])
1598             {
1599               token = G_TOKEN_COMMENT_SINGLE;
1600               in_comment_single = TRUE;
1601               gstring = g_string_new ("");
1602               ch = g_scanner_get_char (scanner, line_p, position_p);
1603               while (ch != 0)
1604                 {
1605                   if (ch == config->cpair_comment_single[1])
1606                     {
1607                       in_comment_single = FALSE;
1608                       ch = 0;
1609                       break;
1610                     }
1611
1612                   gstring = g_string_append_c (gstring, ch);
1613                   ch = g_scanner_get_char (scanner, line_p, position_p);
1614                 }
1615             }
1616           else if (config->scan_identifier && ch &&
1617                    strchr (config->cset_identifier_first, ch))
1618             {
1619             identifier_precedence:
1620
1621               if (config->cset_identifier_nth && ch &&
1622                   strchr (config->cset_identifier_nth,
1623                           g_scanner_peek_next_char (scanner)))
1624                 {
1625                   token = G_TOKEN_IDENTIFIER;
1626                   gstring = g_string_new ("");
1627                   gstring = g_string_append_c (gstring, ch);
1628                   do
1629                     {
1630                       ch = g_scanner_get_char (scanner, line_p, position_p);
1631                       gstring = g_string_append_c (gstring, ch);
1632                       ch = g_scanner_peek_next_char (scanner);
1633                     }
1634                   while (ch && strchr (config->cset_identifier_nth, ch));
1635                   ch = 0;
1636                 }
1637               else if (config->scan_identifier_1char)
1638                 {
1639                   token = G_TOKEN_IDENTIFIER;
1640                   value.v_identifier = g_new0 (gchar, 2);
1641                   value.v_identifier[0] = ch;
1642                   ch = 0;
1643                 }
1644             }
1645           if (ch)
1646             {
1647               if (config->char_2_token)
1648                 token = ch;
1649               else
1650                 {
1651                   token = G_TOKEN_CHAR;
1652                   value.v_char = ch;
1653                 }
1654               ch = 0;
1655             }
1656         } /* default_case:... */
1657         break;
1658         }
1659       g_assert (ch == 0 && token != G_TOKEN_NONE); /* paranoid */
1660     }
1661   while (ch != 0);
1662
1663   if (in_comment_multi || in_comment_single ||
1664       in_string_sq || in_string_dq)
1665     {
1666       token = G_TOKEN_ERROR;
1667       if (gstring)
1668         {
1669           g_string_free (gstring, TRUE);
1670           gstring = NULL;
1671         }
1672       (*position_p)++;
1673       if (in_comment_multi || in_comment_single)
1674         value.v_error = G_ERR_UNEXP_EOF_IN_COMMENT;
1675       else /* (in_string_sq || in_string_dq) */
1676         value.v_error = G_ERR_UNEXP_EOF_IN_STRING;
1677     }
1678
1679   if (gstring)
1680     {
1681       value.v_string = gstring->str;
1682       g_string_free (gstring, FALSE);
1683       gstring = NULL;
1684     }
1685
1686   if (token == G_TOKEN_IDENTIFIER)
1687     {
1688       if (config->scan_symbols)
1689         {
1690           GScannerKey *key;
1691           guint scope_id;
1692
1693           scope_id = scanner->scope_id;
1694           key = g_scanner_lookup_internal (scanner, scope_id, value.v_identifier);
1695           if (!key && scope_id && scanner->config->scope_0_fallback)
1696             key = g_scanner_lookup_internal (scanner, 0, value.v_identifier);
1697
1698           if (key)
1699             {
1700               g_free (value.v_identifier);
1701               token = G_TOKEN_SYMBOL;
1702               value.v_symbol = key->value;
1703             }
1704         }
1705
1706       if (token == G_TOKEN_IDENTIFIER &&
1707           config->scan_identifier_NULL &&
1708           strlen (value.v_identifier) == 4)
1709         {
1710           gchar *null_upper = "NULL";
1711           gchar *null_lower = "null";
1712
1713           if (scanner->config->case_sensitive)
1714             {
1715               if (value.v_identifier[0] == null_upper[0] &&
1716                   value.v_identifier[1] == null_upper[1] &&
1717                   value.v_identifier[2] == null_upper[2] &&
1718                   value.v_identifier[3] == null_upper[3])
1719                 token = G_TOKEN_IDENTIFIER_NULL;
1720             }
1721           else
1722             {
1723               if ((value.v_identifier[0] == null_upper[0] ||
1724                    value.v_identifier[0] == null_lower[0]) &&
1725                   (value.v_identifier[1] == null_upper[1] ||
1726                    value.v_identifier[1] == null_lower[1]) &&
1727                   (value.v_identifier[2] == null_upper[2] ||
1728                    value.v_identifier[2] == null_lower[2]) &&
1729                   (value.v_identifier[3] == null_upper[3] ||
1730                    value.v_identifier[3] == null_lower[3]))
1731                 token = G_TOKEN_IDENTIFIER_NULL;
1732             }
1733         }
1734     }
1735
1736   *token_p = token;
1737   *value_p = value;
1738 }