glib/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 #ifdef HAVE_CONFIG_H
  35 #include <config.h>
  36 #endif
  37
  38 #include        <stdlib.h>
  39 #include        <stdarg.h>
  40 #include        <string.h>
  41 #include        <stdio.h>
  42 #include        "glib.h"
  43 #ifdef HAVE_UNISTD_H
  44 #include        <unistd.h>
  45 #endif
  46 #include        <errno.h>
  47 #ifdef G_OS_WIN32
  48 #include        <io.h>          /* For _read() */
  49 #endif
  50
  51 /* --- defines --- */
  52 #define to_lower(c)                             ( \
  53         (guchar) (                                                      \
  54           ( (((guchar)(c))>='A' && ((guchar)(c))<='Z') * ('a'-'A') ) |  \
  55           ( (((guchar)(c))>=192 && ((guchar)(c))<=214) * (224-192) ) |  \
  56           ( (((guchar)(c))>=216 && ((guchar)(c))<=222) * (248-216) ) |  \
  57           ((guchar)(c))                                                 \
  58         )                                                               \
  59 )
  60 #define READ_BUFFER_SIZE        (4000)
  61
  62
  63 /* --- typedefs --- */
  64 typedef struct  _GScannerKey    GScannerKey;
  65
  66 struct  _GScannerKey
  67 {
  68   guint          scope_id;
  69   gchar         *symbol;
  70   gpointer       value;
  71 };
  72
  73
  74
  75 /* --- variables --- */
  76 static GScannerConfig g_scanner_config_template =
  77 {
  78   (
  79    " \t\r\n"
  80    )                    /* cset_skip_characters */,
  81   (
  82    G_CSET_a_2_z
  83    "_"
  84    G_CSET_A_2_Z
  85    )                    /* cset_identifier_first */,
  86   (
  87    G_CSET_a_2_z
  88    "_"
  89    G_CSET_A_2_Z
  90    G_CSET_DIGITS
  91    G_CSET_LATINS
  92    G_CSET_LATINC
  93    )                    /* cset_identifier_nth */,
  94   ( "#\n" )             /* cpair_comment_single */,
  95
  96   FALSE                 /* case_sensitive */,
  97
  98   TRUE                  /* skip_comment_multi */,
  99   TRUE                  /* skip_comment_single */,
 100   TRUE                  /* scan_comment_multi */,
 101   TRUE                  /* scan_identifier */,
 102   FALSE                 /* scan_identifier_1char */,
 103   FALSE                 /* scan_identifier_NULL */,
 104   TRUE                  /* scan_symbols */,
 105   FALSE                 /* scan_binary */,
 106   TRUE                  /* scan_octal */,
 107   TRUE                  /* scan_float */,
 108   TRUE                  /* scan_hex */,
 109   FALSE                 /* scan_hex_dollar */,
 110   TRUE                  /* scan_string_sq */,
 111   TRUE                  /* scan_string_dq */,
 112   TRUE                  /* numbers_2_int */,
 113   FALSE                 /* int_2_float */,
 114   FALSE                 /* identifier_2_string */,
 115   TRUE                  /* char_2_token */,
 116   FALSE                 /* symbol_2_token */,
 117   FALSE                 /* scope_0_fallback */,
 118 };
 119
 120
 121 /* --- prototypes --- */
 122 static inline
 123 GScannerKey*    g_scanner_lookup_internal (GScanner     *scanner,
 124                                            guint         scope_id,
 125                                            const gchar  *symbol);
 126 static gboolean g_scanner_key_equal       (gconstpointer v1,
 127                                            gconstpointer v2);
 128 static guint    g_scanner_key_hash        (gconstpointer v);
 129 static void     g_scanner_get_token_ll    (GScanner     *scanner,
 130                                            GTokenType   *token_p,
 131                                            GTokenValue  *value_p,
 132                                            guint        *line_p,
 133                                            guint        *position_p);
 134 static void     g_scanner_get_token_i     (GScanner     *scanner,
 135                                            GTokenType   *token_p,
 136                                            GTokenValue  *value_p,
 137                                            guint        *line_p,
 138                                            guint        *position_p);
 139
 140 static guchar   g_scanner_peek_next_char  (GScanner     *scanner);
 141 static guchar   g_scanner_get_char        (GScanner     *scanner,
 142                                            guint        *line_p,
 143                                            guint        *position_p);
 144 static void     g_scanner_msg_handler     (GScanner     *scanner,
 145                                            gchar        *message,
 146                                            gboolean      is_error);
 147
 148
 149 /* --- functions --- */
 150 static inline gint
 151 g_scanner_char_2_num (guchar    c,
 152                       guchar    base)
 153 {
 154   if (c >= '0' && c <= '9')
 155     c -= '0';
 156   else if (c >= 'A' && c <= 'Z')
 157     c -= 'A' - 10;
 158   else if (c >= 'a' && c <= 'z')
 159     c -= 'a' - 10;
 160   else
 161     return -1;
 162
 163   if (c < base)
 164     return c;
 165
 166   return -1;
 167 }
 168
 169 GScanner*
 170 g_scanner_new (const GScannerConfig *config_templ)
 171 {
 172   GScanner *scanner;
 173
 174   if (!config_templ)
 175     config_templ = &g_scanner_config_template;
 176
 177   scanner = g_new0 (GScanner, 1);
 178
 179   scanner->user_data = NULL;
 180   scanner->max_parse_errors = 1;
 181   scanner->parse_errors = 0;
 182   scanner->input_name = NULL;
 183   g_datalist_init (&scanner->qdata);
 184
 185   scanner->config = g_new0 (GScannerConfig, 1);
 186
 187   scanner->config->case_sensitive        = config_templ->case_sensitive;
 188   scanner->config->cset_skip_characters  = config_templ->cset_skip_characters;
 189   if (!scanner->config->cset_skip_characters)
 190     scanner->config->cset_skip_characters = "";
 191   scanner->config->cset_identifier_first = config_templ->cset_identifier_first;
 192   scanner->config->cset_identifier_nth   = config_templ->cset_identifier_nth;
 193   scanner->config->cpair_comment_single  = config_templ->cpair_comment_single;
 194   scanner->config->skip_comment_multi    = config_templ->skip_comment_multi;
 195   scanner->config->skip_comment_single   = config_templ->skip_comment_single;
 196   scanner->config->scan_comment_multi    = config_templ->scan_comment_multi;
 197   scanner->config->scan_identifier       = config_templ->scan_identifier;
 198   scanner->config->scan_identifier_1char = config_templ->scan_identifier_1char;
 199   scanner->config->scan_identifier_NULL  = config_templ->scan_identifier_NULL;
 200   scanner->config->scan_symbols          = config_templ->scan_symbols;
 201   scanner->config->scan_binary           = config_templ->scan_binary;
 202   scanner->config->scan_octal            = config_templ->scan_octal;
 203   scanner->config->scan_float            = config_templ->scan_float;
 204   scanner->config->scan_hex              = config_templ->scan_hex;
 205   scanner->config->scan_hex_dollar       = config_templ->scan_hex_dollar;
 206   scanner->config->scan_string_sq        = config_templ->scan_string_sq;
 207   scanner->config->scan_string_dq        = config_templ->scan_string_dq;
 208   scanner->config->numbers_2_int         = config_templ->numbers_2_int;
 209   scanner->config->int_2_float           = config_templ->int_2_float;
 210   scanner->config->identifier_2_string   = config_templ->identifier_2_string;
 211   scanner->config->char_2_token          = config_templ->char_2_token;
 212   scanner->config->symbol_2_token        = config_templ->symbol_2_token;
 213   scanner->config->scope_0_fallback      = config_templ->scope_0_fallback;
 214
 215   scanner->token = G_TOKEN_NONE;
 216   scanner->value.v_int = 0;
 217   scanner->line = 1;
 218   scanner->position = 0;
 219
 220   scanner->next_token = G_TOKEN_NONE;
 221   scanner->next_value.v_int = 0;
 222   scanner->next_line = 1;
 223   scanner->next_position = 0;
 224
 225   scanner->symbol_table = g_hash_table_new (g_scanner_key_hash, g_scanner_key_equal);
 226   scanner->input_fd = -1;
 227   scanner->text = NULL;
 228   scanner->text_end = NULL;
 229   scanner->buffer = NULL;
 230   scanner->scope_id = 0;
 231
 232   scanner->msg_handler = g_scanner_msg_handler;
 233
 234   return scanner;
 235 }
 236
 237 static inline void
 238 g_scanner_free_value (GTokenType     *token_p,
 239                       GTokenValue     *value_p)
 240 {
 241   switch (*token_p)
 242     {
 243     case G_TOKEN_STRING:
 244     case G_TOKEN_IDENTIFIER:
 245     case G_TOKEN_IDENTIFIER_NULL:
 246     case G_TOKEN_COMMENT_SINGLE:
 247     case G_TOKEN_COMMENT_MULTI:
 248       g_free (value_p->v_string);
 249       break;
 250
 251     default:
 252       break;
 253     }
 254
 255   *token_p = G_TOKEN_NONE;
 256 }
 257
 258 static void
 259 g_scanner_destroy_symbol_table_entry (gpointer _key,
 260                                       gpointer _value,
 261                                       gpointer _data)
 262 {
 263   GScannerKey *key = _key;
 264
 265   g_free (key->symbol);
 266   g_free (key);
 267 }
 268
 269 void
 270 g_scanner_destroy (GScanner     *scanner)
 271 {
 272   g_return_if_fail (scanner != NULL);
 273
 274   g_datalist_clear (&scanner->qdata);
 275   g_hash_table_foreach (scanner->symbol_table,
 276                         g_scanner_destroy_symbol_table_entry, NULL);
 277   g_hash_table_destroy (scanner->symbol_table);
 278   g_scanner_free_value (&scanner->token, &scanner->value);
 279   g_scanner_free_value (&scanner->next_token, &scanner->next_value);
 280   g_free (scanner->config);
 281   g_free (scanner->buffer);
 282   g_free (scanner);
 283 }
 284
 285 static void
 286 g_scanner_msg_handler (GScanner         *scanner,
 287                        gchar            *message,
 288                        gboolean         is_error)
 289 {
 290   g_return_if_fail (scanner != NULL);
 291
 292   fprintf (stderr, "%s:%d: ",
 293            scanner->input_name ? scanner->input_name : "<memory>",
 294            scanner->line);
 295   if (is_error)
 296     fprintf (stderr, "error: ");
 297   fprintf (stderr, "%s\n", message);
 298 }
 299
 300 void
 301 g_scanner_error (GScanner       *scanner,
 302                  const gchar    *format,
 303                  ...)
 304 {
 305   g_return_if_fail (scanner != NULL);
 306   g_return_if_fail (format != NULL);
 307
 308   scanner->parse_errors++;
 309
 310   if (scanner->msg_handler)
 311     {
 312       va_list args;
 313       gchar *string;
 314
 315       va_start (args, format);
 316       string = g_strdup_vprintf (format, args);
 317       va_end (args);
 318
 319       scanner->msg_handler (scanner, string, TRUE);
 320
 321       g_free (string);
 322     }
 323 }
 324
 325 void
 326 g_scanner_warn (GScanner       *scanner,
 327                 const gchar    *format,
 328                 ...)
 329 {
 330   g_return_if_fail (scanner != NULL);
 331   g_return_if_fail (format != NULL);
 332
 333   if (scanner->msg_handler)
 334     {
 335       va_list args;
 336       gchar *string;
 337
 338       va_start (args, format);
 339       string = g_strdup_vprintf (format, args);
 340       va_end (args);
 341
 342       scanner->msg_handler (scanner, string, FALSE);
 343
 344       g_free (string);
 345     }
 346 }
 347
 348 static gboolean
 349 g_scanner_key_equal (gconstpointer v1,
 350                      gconstpointer v2)
 351 {
 352   const GScannerKey *key1 = v1;
 353   const GScannerKey *key2 = v2;
 354
 355   return (key1->scope_id == key2->scope_id) && (strcmp (key1->symbol, key2->symbol) == 0);
 356 }
 357
 358 static guint
 359 g_scanner_key_hash (gconstpointer v)
 360 {
 361   const GScannerKey *key = v;
 362   gchar *c;
 363   guint h;
 364
 365   h = key->scope_id;
 366   for (c = key->symbol; *c; c++)
 367     h = (h << 5) - h + *c;
 368
 369   return h;
 370 }
 371
 372 static inline GScannerKey*
 373 g_scanner_lookup_internal (GScanner     *scanner,
 374                            guint         scope_id,
 375                            const gchar  *symbol)
 376 {
 377   GScannerKey   *key_p;
 378   GScannerKey key;
 379
 380   key.scope_id = scope_id;
 381
 382   if (!scanner->config->case_sensitive)
 383     {
 384       gchar *d;
 385       const gchar *c;
 386
 387       key.symbol = g_new (gchar, strlen (symbol) + 1);
 388       for (d = key.symbol, c = symbol; *c; c++, d++)
 389         *d = to_lower (*c);
 390       *d = 0;
 391       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 392       g_free (key.symbol);
 393     }
 394   else
 395     {
 396       key.symbol = (gchar*) symbol;
 397       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 398     }
 399
 400   return key_p;
 401 }
 402
 403 void
 404 g_scanner_scope_add_symbol (GScanner    *scanner,
 405                             guint        scope_id,
 406                             const gchar *symbol,
 407                             gpointer     value)
 408 {
 409   GScannerKey   *key;
 410
 411   g_return_if_fail (scanner != NULL);
 412   g_return_if_fail (symbol != NULL);
 413
 414   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 415
 416   if (!key)
 417     {
 418       key = g_new (GScannerKey, 1);
 419       key->scope_id = scope_id;
 420       key->symbol = g_strdup (symbol);
 421       key->value = value;
 422       if (!scanner->config->case_sensitive)
 423         {
 424           gchar *c;
 425
 426           c = key->symbol;
 427           while (*c != 0)
 428             {
 429               *c = to_lower (*c);
 430               c++;
 431             }
 432         }
 433       g_hash_table_insert (scanner->symbol_table, key, key);
 434     }
 435   else
 436     key->value = value;
 437 }
 438
 439 void
 440 g_scanner_scope_remove_symbol (GScanner    *scanner,
 441                                guint        scope_id,
 442                                const gchar *symbol)
 443 {
 444   GScannerKey   *key;
 445
 446   g_return_if_fail (scanner != NULL);
 447   g_return_if_fail (symbol != NULL);
 448
 449   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 450
 451   if (key)
 452     {
 453       g_hash_table_remove (scanner->symbol_table, key);
 454       g_free (key->symbol);
 455       g_free (key);
 456     }
 457 }
 458
 459 gpointer
 460 g_scanner_lookup_symbol (GScanner       *scanner,
 461                          const gchar    *symbol)
 462 {
 463   GScannerKey   *key;
 464   guint scope_id;
 465
 466   g_return_val_if_fail (scanner != NULL, NULL);
 467
 468   if (!symbol)
 469     return NULL;
 470
 471   scope_id = scanner->scope_id;
 472   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 473   if (!key && scope_id && scanner->config->scope_0_fallback)
 474     key = g_scanner_lookup_internal (scanner, 0, symbol);
 475
 476   if (key)
 477     return key->value;
 478   else
 479     return NULL;
 480 }
 481
 482 gpointer
 483 g_scanner_scope_lookup_symbol (GScanner       *scanner,
 484                                guint           scope_id,
 485                                const gchar    *symbol)
 486 {
 487   GScannerKey   *key;
 488
 489   g_return_val_if_fail (scanner != NULL, NULL);
 490
 491   if (!symbol)
 492     return NULL;
 493
 494   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 495
 496   if (key)
 497     return key->value;
 498   else
 499     return NULL;
 500 }
 501
 502 guint
 503 g_scanner_set_scope (GScanner       *scanner,
 504                      guint           scope_id)
 505 {
 506   guint old_scope_id;
 507
 508   g_return_val_if_fail (scanner != NULL, 0);
 509
 510   old_scope_id = scanner->scope_id;
 511   scanner->scope_id = scope_id;
 512
 513   return old_scope_id;
 514 }
 515
 516 static void
 517 g_scanner_foreach_internal (gpointer  _key,
 518                             gpointer  _value,
 519                             gpointer  _user_data)
 520 {
 521   GScannerKey *key;
 522   gpointer *d;
 523   GHFunc func;
 524   gpointer user_data;
 525   guint *scope_id;
 526
 527   d = _user_data;
 528   func = (GHFunc) d[0];
 529   user_data = d[1];
 530   scope_id = d[2];
 531   key = _value;
 532
 533   if (key->scope_id == *scope_id)
 534     func (key->symbol, key->value, user_data);
 535 }
 536
 537 void
 538 g_scanner_scope_foreach_symbol (GScanner       *scanner,
 539                                 guint           scope_id,
 540                                 GHFunc          func,
 541                                 gpointer        user_data)
 542 {
 543   gpointer d[3];
 544
 545   g_return_if_fail (scanner != NULL);
 546
 547   d[0] = (gpointer) func;
 548   d[1] = user_data;
 549   d[2] = &scope_id;
 550
 551   g_hash_table_foreach (scanner->symbol_table, g_scanner_foreach_internal, d);
 552 }
 553
 554 GTokenType
 555 g_scanner_peek_next_token (GScanner     *scanner)
 556 {
 557   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 558
 559   if (scanner->next_token == G_TOKEN_NONE)
 560     {
 561       scanner->next_line = scanner->line;
 562       scanner->next_position = scanner->position;
 563       g_scanner_get_token_i (scanner,
 564                              &scanner->next_token,
 565                              &scanner->next_value,
 566                              &scanner->next_line,
 567                              &scanner->next_position);
 568     }
 569
 570   return scanner->next_token;
 571 }
 572
 573 GTokenType
 574 g_scanner_get_next_token (GScanner      *scanner)
 575 {
 576   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 577
 578   if (scanner->next_token != G_TOKEN_NONE)
 579     {
 580       g_scanner_free_value (&scanner->token, &scanner->value);
 581
 582       scanner->token = scanner->next_token;
 583       scanner->value = scanner->next_value;
 584       scanner->line = scanner->next_line;
 585       scanner->position = scanner->next_position;
 586       scanner->next_token = G_TOKEN_NONE;
 587     }
 588   else
 589     g_scanner_get_token_i (scanner,
 590                            &scanner->token,
 591                            &scanner->value,
 592                            &scanner->line,
 593                            &scanner->position);
 594
 595   return scanner->token;
 596 }
 597
 598 GTokenType
 599 g_scanner_cur_token (GScanner *scanner)
 600 {
 601   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 602
 603   return scanner->token;
 604 }
 605
 606 GTokenValue
 607 g_scanner_cur_value (GScanner *scanner)
 608 {
 609   GTokenValue v;
 610
 611   v.v_int = 0;
 612
 613   g_return_val_if_fail (scanner != NULL, v);
 614
 615   /* MSC isn't capable of handling return scanner->value; ? */
 616
 617   v = scanner->value;
 618
 619   return v;
 620 }
 621
 622 guint
 623 g_scanner_cur_line (GScanner *scanner)
 624 {
 625   g_return_val_if_fail (scanner != NULL, 0);
 626
 627   return scanner->line;
 628 }
 629
 630 guint
 631 g_scanner_cur_position (GScanner *scanner)
 632 {
 633   g_return_val_if_fail (scanner != NULL, 0);
 634
 635   return scanner->position;
 636 }
 637
 638 gboolean
 639 g_scanner_eof (GScanner *scanner)
 640 {
 641   g_return_val_if_fail (scanner != NULL, TRUE);
 642
 643   return scanner->token == G_TOKEN_EOF || scanner->token == G_TOKEN_ERROR;
 644 }
 645
 646 void
 647 g_scanner_input_file (GScanner *scanner,
 648                       gint      input_fd)
 649 {
 650   g_return_if_fail (scanner != NULL);
 651   g_return_if_fail (input_fd >= 0);
 652
 653   if (scanner->input_fd >= 0)
 654     g_scanner_sync_file_offset (scanner);
 655
 656   scanner->token = G_TOKEN_NONE;
 657   scanner->value.v_int = 0;
 658   scanner->line = 1;
 659   scanner->position = 0;
 660   scanner->next_token = G_TOKEN_NONE;
 661
 662   scanner->input_fd = input_fd;
 663   scanner->text = NULL;
 664   scanner->text_end = NULL;
 665
 666   if (!scanner->buffer)
 667     scanner->buffer = g_new (gchar, READ_BUFFER_SIZE + 1);
 668 }
 669
 670 void
 671 g_scanner_input_text (GScanner    *scanner,
 672                       const gchar *text,
 673                       guint        text_len)
 674 {
 675   g_return_if_fail (scanner != NULL);
 676   if (text_len)
 677     g_return_if_fail (text != NULL);
 678   else
 679     text = NULL;
 680
 681   if (scanner->input_fd >= 0)
 682     g_scanner_sync_file_offset (scanner);
 683
 684   scanner->token = G_TOKEN_NONE;
 685   scanner->value.v_int = 0;
 686   scanner->line = 1;
 687   scanner->position = 0;
 688   scanner->next_token = G_TOKEN_NONE;
 689
 690   scanner->input_fd = -1;
 691   scanner->text = text;
 692   scanner->text_end = text + text_len;
 693
 694   if (scanner->buffer)
 695     {
 696       g_free (scanner->buffer);
 697       scanner->buffer = NULL;
 698     }
 699 }
 700
 701 static guchar
 702 g_scanner_peek_next_char (GScanner *scanner)
 703 {
 704   if (scanner->text < scanner->text_end)
 705     {
 706       return *scanner->text;
 707     }
 708   else if (scanner->input_fd >= 0)
 709     {
 710       gint count;
 711       gchar *buffer;
 712
 713       buffer = scanner->buffer;
 714       do
 715         {
 716           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 717         }
 718       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 719
 720       if (count < 1)
 721         {
 722           scanner->input_fd = -1;
 723
 724           return 0;
 725         }
 726       else
 727         {
 728           scanner->text = buffer;
 729           scanner->text_end = buffer + count;
 730
 731           return *buffer;
 732         }
 733     }
 734   else
 735     return 0;
 736 }
 737
 738 void
 739 g_scanner_sync_file_offset (GScanner *scanner)
 740 {
 741   g_return_if_fail (scanner != NULL);
 742
 743   /* for file input, rewind the filedescriptor to the current
 744    * buffer position and blow the file read ahead buffer. usefull for
 745    * third party uses of our filedescriptor, which hooks onto the current
 746    * scanning position.
 747    */
 748
 749   if (scanner->input_fd >= 0 && scanner->text_end > scanner->text)
 750     {
 751       gint buffered;
 752
 753       buffered = scanner->text_end - scanner->text;
 754       if (lseek (scanner->input_fd, - buffered, SEEK_CUR) >= 0)
 755         {
 756           /* we succeeded, blow our buffer's contents now */
 757           scanner->text = NULL;
 758           scanner->text_end = NULL;
 759         }
 760       else
 761         errno = 0;
 762     }
 763 }
 764
 765 static guchar
 766 g_scanner_get_char (GScanner    *scanner,
 767                     guint       *line_p,
 768                     guint       *position_p)
 769 {
 770   guchar fchar;
 771
 772   if (scanner->text < scanner->text_end)
 773     fchar = *(scanner->text++);
 774   else if (scanner->input_fd >= 0)
 775     {
 776       gint count;
 777       gchar *buffer;
 778
 779       buffer = scanner->buffer;
 780       do
 781         {
 782           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 783         }
 784       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 785
 786       if (count < 1)
 787         {
 788           scanner->input_fd = -1;
 789           fchar = 0;
 790         }
 791       else
 792         {
 793           scanner->text = buffer + 1;
 794           scanner->text_end = buffer + count;
 795           fchar = *buffer;
 796           if (!fchar)
 797             {
 798               g_scanner_sync_file_offset (scanner);
 799               scanner->text_end = scanner->text;
 800               scanner->input_fd = -1;
 801             }
 802         }
 803     }
 804   else
 805     fchar = 0;
 806
 807   if (fchar == '\n')
 808     {
 809       (*position_p) = 0;
 810       (*line_p)++;
 811     }
 812   else if (fchar)
 813     {
 814       (*position_p)++;
 815     }
 816
 817   return fchar;
 818 }
 819
 820 void
 821 g_scanner_unexp_token (GScanner         *scanner,
 822                        GTokenType        expected_token,
 823                        const gchar      *identifier_spec,
 824                        const gchar      *symbol_spec,
 825                        const gchar      *symbol_name,
 826                        const gchar      *message,
 827                        gint              is_error)
 828 {
 829   gchar *token_string;
 830   guint token_string_len;
 831   gchar *expected_string;
 832   guint expected_string_len;
 833   gchar *message_prefix;
 834   gboolean print_unexp;
 835   void (*msg_handler)   (GScanner*, const gchar*, ...);
 836
 837   g_return_if_fail (scanner != NULL);
 838
 839   if (is_error)
 840     msg_handler = g_scanner_error;
 841   else
 842     msg_handler = g_scanner_warn;
 843
 844   if (!identifier_spec)
 845     identifier_spec = "identifier";
 846   if (!symbol_spec)
 847     symbol_spec = "symbol";
 848
 849   token_string_len = 56;
 850   token_string = g_new (gchar, token_string_len + 1);
 851   expected_string_len = 64;
 852   expected_string = g_new (gchar, expected_string_len + 1);
 853   print_unexp = TRUE;
 854
 855   switch (scanner->token)
 856     {
 857     case G_TOKEN_EOF:
 858       g_snprintf (token_string, token_string_len, "end of file");
 859       break;
 860
 861     default:
 862       if (scanner->token >= 1 && scanner->token <= 255)
 863         {
 864           if ((scanner->token >= ' ' && scanner->token <= '~') ||
 865               strchr (scanner->config->cset_identifier_first, scanner->token) ||
 866               strchr (scanner->config->cset_identifier_nth, scanner->token))
 867             g_snprintf (token_string, token_string_len, "character `%c'", scanner->token);
 868           else
 869             g_snprintf (token_string, token_string_len, "character `\\%o'", scanner->token);
 870           break;
 871         }
 872       else if (!scanner->config->symbol_2_token)
 873         {
 874           g_snprintf (token_string, token_string_len, "(unknown) token <%d>", scanner->token);
 875           break;
 876         }
 877       /* fall through */
 878     case G_TOKEN_SYMBOL:
 879       if (expected_token == G_TOKEN_SYMBOL ||
 880           (scanner->config->symbol_2_token &&
 881            expected_token > G_TOKEN_LAST))
 882         print_unexp = FALSE;
 883       if (symbol_name)
 884         g_snprintf (token_string,
 885                     token_string_len,
 886                     "%s%s `%s'",
 887                     print_unexp ? "" : "invalid ",
 888                     symbol_spec,
 889                     symbol_name);
 890       else
 891         g_snprintf (token_string,
 892                     token_string_len,
 893                     "%s%s",
 894                     print_unexp ? "" : "invalid ",
 895                     symbol_spec);
 896       break;
 897
 898     case G_TOKEN_ERROR:
 899       print_unexp = FALSE;
 900       expected_token = G_TOKEN_NONE;
 901       switch (scanner->value.v_error)
 902         {
 903         case G_ERR_UNEXP_EOF:
 904           g_snprintf (token_string, token_string_len, "scanner: unexpected end of file");
 905           break;
 906
 907         case G_ERR_UNEXP_EOF_IN_STRING:
 908           g_snprintf (token_string, token_string_len, "scanner: unterminated string constant");
 909           break;
 910
 911         case G_ERR_UNEXP_EOF_IN_COMMENT:
 912           g_snprintf (token_string, token_string_len, "scanner: unterminated comment");
 913           break;
 914
 915         case G_ERR_NON_DIGIT_IN_CONST:
 916           g_snprintf (token_string, token_string_len, "scanner: non digit in constant");
 917           break;
 918
 919         case G_ERR_FLOAT_RADIX:
 920           g_snprintf (token_string, token_string_len, "scanner: invalid radix for floating constant");
 921           break;
 922
 923         case G_ERR_FLOAT_MALFORMED:
 924           g_snprintf (token_string, token_string_len, "scanner: malformed floating constant");
 925           break;
 926
 927         case G_ERR_DIGIT_RADIX:
 928           g_snprintf (token_string, token_string_len, "scanner: digit is beyond radix");
 929           break;
 930
 931         case G_ERR_UNKNOWN:
 932         default:
 933           g_snprintf (token_string, token_string_len, "scanner: unknown error");
 934           break;
 935         }
 936       break;
 937
 938     case G_TOKEN_CHAR:
 939       g_snprintf (token_string, token_string_len, "character `%c'", scanner->value.v_char);
 940       break;
 941
 942     case G_TOKEN_IDENTIFIER:
 943     case G_TOKEN_IDENTIFIER_NULL:
 944       if (expected_token == G_TOKEN_IDENTIFIER ||
 945           expected_token == G_TOKEN_IDENTIFIER_NULL)
 946         print_unexp = FALSE;
 947       g_snprintf (token_string,
 948                   token_string_len,
 949                   "%s%s `%s'",
 950                   print_unexp ? "" : "invalid ",
 951                   identifier_spec,
 952                   scanner->token == G_TOKEN_IDENTIFIER ? scanner->value.v_string : "null");
 953       break;
 954
 955     case G_TOKEN_BINARY:
 956     case G_TOKEN_OCTAL:
 957     case G_TOKEN_INT:
 958     case G_TOKEN_HEX:
 959       g_snprintf (token_string, token_string_len, "number `%ld'", scanner->value.v_int);
 960       break;
 961
 962     case G_TOKEN_FLOAT:
 963       g_snprintf (token_string, token_string_len, "number `%.3f'", scanner->value.v_float);
 964       break;
 965
 966     case G_TOKEN_STRING:
 967       if (expected_token == G_TOKEN_STRING)
 968         print_unexp = FALSE;
 969       g_snprintf (token_string,
 970                   token_string_len,
 971                   "%s%sstring constant \"%s\"",
 972                   print_unexp ? "" : "invalid ",
 973                   scanner->value.v_string[0] == 0 ? "empty " : "",
 974                   scanner->value.v_string);
 975       token_string[token_string_len - 2] = '"';
 976       token_string[token_string_len - 1] = 0;
 977       break;
 978
 979     case G_TOKEN_COMMENT_SINGLE:
 980     case G_TOKEN_COMMENT_MULTI:
 981       g_snprintf (token_string, token_string_len, "comment");
 982       break;
 983
 984     case G_TOKEN_NONE:
 985       /* somehow the user's parsing code is screwed, there isn't much
 986        * we can do about it.
 987        * Note, a common case to trigger this is
 988        * g_scanner_peek_next_token(); g_scanner_unexp_token();
 989        * without an intermediate g_scanner_get_next_token().
 990        */
 991       g_assert_not_reached ();
 992       break;
 993     }
 994
 995
 996   switch (expected_token)
 997     {
 998       gboolean need_valid;
 999       gchar *tstring;
1000     case G_TOKEN_EOF:
1001       g_snprintf (expected_string, expected_string_len, "end of file");
1002       break;
1003     default:
1004       if (expected_token >= 1 && expected_token <= 255)
1005         {
1006           if ((expected_token >= ' ' && expected_token <= '~') ||
1007               strchr (scanner->config->cset_identifier_first, expected_token) ||
1008               strchr (scanner->config->cset_identifier_nth, expected_token))
1009             g_snprintf (expected_string, expected_string_len, "character `%c'", expected_token);
1010           else
1011             g_snprintf (expected_string, expected_string_len, "character `\\%o'", expected_token);
1012           break;
1013         }
1014       else if (!scanner->config->symbol_2_token)
1015         {
1016           g_snprintf (expected_string, expected_string_len, "(unknown) token <%d>", expected_token);
1017           break;
1018         }
1019       /* fall through */
1020     case G_TOKEN_SYMBOL:
1021       need_valid = (scanner->token == G_TOKEN_SYMBOL ||
1022                     (scanner->config->symbol_2_token &&
1023                      scanner->token > G_TOKEN_LAST));
1024       g_snprintf (expected_string,
1025                   expected_string_len,
1026                   "%s%s",
1027                   need_valid ? "valid " : "",
1028                   symbol_spec);
1029       /* FIXME: should we attempt to lookup the symbol_name for symbol_2_token? */
1030       break;
1031     case G_TOKEN_CHAR:
1032       g_snprintf (expected_string, expected_string_len, "%scharacter",
1033                   scanner->token == G_TOKEN_CHAR ? "valid " : "");
1034       break;
1035     case G_TOKEN_BINARY:
1036       tstring = "binary";
1037       g_snprintf (expected_string, expected_string_len, "%snumber (%s)",
1038                   scanner->token == expected_token ? "valid " : "", tstring);
1039       break;
1040     case G_TOKEN_OCTAL:
1041       tstring = "octal";
1042       g_snprintf (expected_string, expected_string_len, "%snumber (%s)",
1043                   scanner->token == expected_token ? "valid " : "", tstring);
1044       break;
1045     case G_TOKEN_INT:
1046       tstring = "integer";
1047       g_snprintf (expected_string, expected_string_len, "%snumber (%s)",
1048                   scanner->token == expected_token ? "valid " : "", tstring);
1049       break;
1050     case G_TOKEN_HEX:
1051       tstring = "hexadecimal";
1052       g_snprintf (expected_string, expected_string_len, "%snumber (%s)",
1053                   scanner->token == expected_token ? "valid " : "", tstring);
1054       break;
1055     case G_TOKEN_FLOAT:
1056       tstring = "float";
1057       g_snprintf (expected_string, expected_string_len, "%snumber (%s)",
1058                   scanner->token == expected_token ? "valid " : "", tstring);
1059       break;
1060     case G_TOKEN_STRING:
1061       g_snprintf (expected_string,
1062                   expected_string_len,
1063                   "%sstring constant",
1064                   scanner->token == G_TOKEN_STRING ? "valid " : "");
1065       break;
1066     case G_TOKEN_IDENTIFIER:
1067     case G_TOKEN_IDENTIFIER_NULL:
1068       need_valid = (scanner->token == G_TOKEN_IDENTIFIER_NULL ||
1069                     scanner->token == G_TOKEN_IDENTIFIER);
1070       g_snprintf (expected_string,
1071                   expected_string_len,
1072                   "%s%s",
1073                   need_valid ? "valid " : "",
1074                   identifier_spec);
1075       break;
1076     case G_TOKEN_COMMENT_SINGLE:
1077       tstring = "single-line";
1078       g_snprintf (expected_string, expected_string_len, "%scomment (%s)",
1079                   scanner->token == expected_token ? "valid " : "", tstring);
1080       break;
1081     case G_TOKEN_COMMENT_MULTI:
1082       tstring = "multi-line";
1083       g_snprintf (expected_string, expected_string_len, "%scomment (%s)",
1084                   scanner->token == expected_token ? "valid " : "", tstring);
1085       break;
1086     case G_TOKEN_NONE:
1087     case G_TOKEN_ERROR:
1088       /* this is handled upon printout */
1089       break;
1090     }
1091
1092   if (message && message[0] != 0)
1093     message_prefix = " - ";
1094   else
1095     {
1096       message_prefix = "";
1097       message = "";
1098     }
1099   if (expected_token == G_TOKEN_ERROR)
1100     {
1101       msg_handler (scanner,
1102                    "failure around %s%s%s",
1103                    token_string,
1104                    message_prefix,
1105                    message);
1106     }
1107   else if (expected_token == G_TOKEN_NONE)
1108     {
1109       if (print_unexp)
1110         msg_handler (scanner,
1111                      "unexpected %s%s%s",
1112                      token_string,
1113                      message_prefix,
1114                      message);
1115       else
1116         msg_handler (scanner,
1117                      "%s%s%s",
1118                      token_string,
1119                      message_prefix,
1120                      message);
1121     }
1122   else
1123     {
1124       if (print_unexp)
1125         msg_handler (scanner,
1126                      "unexpected %s, expected %s%s%s",
1127                      token_string,
1128                      expected_string,
1129                      message_prefix,
1130                      message);
1131       else
1132         msg_handler (scanner,
1133                      "%s, expected %s%s%s",
1134                      token_string,
1135                      expected_string,
1136                      message_prefix,
1137                      message);
1138     }
1139
1140   g_free (token_string);
1141   g_free (expected_string);
1142 }
1143
1144 static void
1145 g_scanner_get_token_i (GScanner *scanner,
1146                        GTokenType       *token_p,
1147                        GTokenValue      *value_p,
1148                        guint            *line_p,
1149                        guint            *position_p)
1150 {
1151   do
1152     {
1153       g_scanner_free_value (token_p, value_p);
1154       g_scanner_get_token_ll (scanner, token_p, value_p, line_p, position_p);
1155     }
1156   while (((*token_p > 0 && *token_p < 256) &&
1157           strchr (scanner->config->cset_skip_characters, *token_p)) ||
1158          (*token_p == G_TOKEN_CHAR &&
1159           strchr (scanner->config->cset_skip_characters, value_p->v_char)) ||
1160          (*token_p == G_TOKEN_COMMENT_MULTI &&
1161           scanner->config->skip_comment_multi) ||
1162          (*token_p == G_TOKEN_COMMENT_SINGLE &&
1163           scanner->config->skip_comment_single));
1164
1165   switch (*token_p)
1166     {
1167     case G_TOKEN_IDENTIFIER:
1168       if (scanner->config->identifier_2_string)
1169         *token_p = G_TOKEN_STRING;
1170       break;
1171
1172     case G_TOKEN_SYMBOL:
1173       if (scanner->config->symbol_2_token)
1174         *token_p = (GTokenType) value_p->v_symbol;
1175       break;
1176
1177     case G_TOKEN_BINARY:
1178     case G_TOKEN_OCTAL:
1179     case G_TOKEN_HEX:
1180       if (scanner->config->numbers_2_int)
1181         *token_p = G_TOKEN_INT;
1182       break;
1183
1184     default:
1185       break;
1186     }
1187
1188   if (*token_p == G_TOKEN_INT &&
1189       scanner->config->int_2_float)
1190     {
1191       *token_p = G_TOKEN_FLOAT;
1192       value_p->v_float = value_p->v_int;
1193     }
1194
1195   errno = 0;
1196 }
1197
1198 static void
1199 g_scanner_get_token_ll  (GScanner       *scanner,
1200                          GTokenType     *token_p,
1201                          GTokenValue    *value_p,
1202                          guint          *line_p,
1203                          guint          *position_p)
1204 {
1205   GScannerConfig *config;
1206   GTokenType       token;
1207   gboolean         in_comment_multi;
1208   gboolean         in_comment_single;
1209   gboolean         in_string_sq;
1210   gboolean         in_string_dq;
1211   GString         *gstring;
1212   GTokenValue      value;
1213   guchar           ch;
1214
1215   config = scanner->config;
1216   (*value_p).v_int = 0;
1217
1218   if ((scanner->text >= scanner->text_end && scanner->input_fd < 0) ||
1219       scanner->token == G_TOKEN_EOF)
1220     {
1221       *token_p = G_TOKEN_EOF;
1222       return;
1223     }
1224
1225   in_comment_multi = FALSE;
1226   in_comment_single = FALSE;
1227   in_string_sq = FALSE;
1228   in_string_dq = FALSE;
1229   gstring = NULL;
1230
1231   do /* while (ch != 0) */
1232     {
1233       gboolean dotted_float = FALSE;
1234
1235       ch = g_scanner_get_char (scanner, line_p, position_p);
1236
1237       value.v_int = 0;
1238       token = G_TOKEN_NONE;
1239
1240       /* this is *evil*, but needed ;(
1241        * we first check for identifier first character, because  it
1242        * might interfere with other key chars like slashes or numbers
1243        */
1244       if (config->scan_identifier &&
1245           ch && strchr (config->cset_identifier_first, ch))
1246         goto identifier_precedence;
1247
1248       switch (ch)
1249         {
1250         case 0:
1251           token = G_TOKEN_EOF;
1252           (*position_p)++;
1253           /* ch = 0; */
1254           break;
1255
1256         case '/':
1257           if (!config->scan_comment_multi ||
1258               g_scanner_peek_next_char (scanner) != '*')
1259             goto default_case;
1260           g_scanner_get_char (scanner, line_p, position_p);
1261           token = G_TOKEN_COMMENT_MULTI;
1262           in_comment_multi = TRUE;
1263           gstring = g_string_new ("");
1264           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1265             {
1266               if (ch == '*' && g_scanner_peek_next_char (scanner) == '/')
1267                 {
1268                   g_scanner_get_char (scanner, line_p, position_p);
1269                   in_comment_multi = FALSE;
1270                   break;
1271                 }
1272               else
1273                 gstring = g_string_append_c (gstring, ch);
1274             }
1275           ch = 0;
1276           break;
1277
1278         case '\'':
1279           if (!config->scan_string_sq)
1280             goto default_case;
1281           token = G_TOKEN_STRING;
1282           in_string_sq = TRUE;
1283           gstring = g_string_new ("");
1284           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1285             {
1286               if (ch == '\'')
1287                 {
1288                   in_string_sq = FALSE;
1289                   break;
1290                 }
1291               else
1292                 gstring = g_string_append_c (gstring, ch);
1293             }
1294           ch = 0;
1295           break;
1296
1297         case '"':
1298           if (!config->scan_string_dq)
1299             goto default_case;
1300           token = G_TOKEN_STRING;
1301           in_string_dq = TRUE;
1302           gstring = g_string_new ("");
1303           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1304             {
1305               if (ch == '"')
1306                 {
1307                   in_string_dq = FALSE;
1308                   break;
1309                 }
1310               else
1311                 {
1312                   if (ch == '\\')
1313                     {
1314                       ch = g_scanner_get_char (scanner, line_p, position_p);
1315                       switch (ch)
1316                         {
1317                           guint i;
1318                           guint fchar;
1319
1320                         case 0:
1321                           break;
1322
1323                         case '\\':
1324                           gstring = g_string_append_c (gstring, '\\');
1325                           break;
1326
1327                         case 'n':
1328                           gstring = g_string_append_c (gstring, '\n');
1329                           break;
1330
1331                         case 't':
1332                           gstring = g_string_append_c (gstring, '\t');
1333                           break;
1334
1335                         case 'r':
1336                           gstring = g_string_append_c (gstring, '\r');
1337                           break;
1338
1339                         case 'b':
1340                           gstring = g_string_append_c (gstring, '\b');
1341                           break;
1342
1343                         case 'f':
1344                           gstring = g_string_append_c (gstring, '\f');
1345                           break;
1346
1347                         case '0':
1348                         case '1':
1349                         case '2':
1350                         case '3':
1351                         case '4':
1352                         case '5':
1353                         case '6':
1354                         case '7':
1355                           i = ch - '0';
1356                           fchar = g_scanner_peek_next_char (scanner);
1357                           if (fchar >= '0' && fchar <= '7')
1358                             {
1359                               ch = g_scanner_get_char (scanner, line_p, position_p);
1360                               i = i * 8 + ch - '0';
1361                               fchar = g_scanner_peek_next_char (scanner);
1362                               if (fchar >= '0' && fchar <= '7')
1363                                 {
1364                                   ch = g_scanner_get_char (scanner, line_p, position_p);
1365                                   i = i * 8 + ch - '0';
1366                                 }
1367                             }
1368                           gstring = g_string_append_c (gstring, i);
1369                           break;
1370
1371                         default:
1372                           gstring = g_string_append_c (gstring, ch);
1373                           break;
1374                         }
1375                     }
1376                   else
1377                     gstring = g_string_append_c (gstring, ch);
1378                 }
1379             }
1380           ch = 0;
1381           break;
1382
1383         case '.':
1384           if (!config->scan_float)
1385             goto default_case;
1386           token = G_TOKEN_FLOAT;
1387           dotted_float = TRUE;
1388           ch = g_scanner_get_char (scanner, line_p, position_p);
1389           goto number_parsing;
1390
1391         case '$':
1392           if (!config->scan_hex_dollar)
1393             goto default_case;
1394           token = G_TOKEN_HEX;
1395           ch = g_scanner_get_char (scanner, line_p, position_p);
1396           goto number_parsing;
1397
1398         case '0':
1399           if (config->scan_octal)
1400             token = G_TOKEN_OCTAL;
1401           else
1402             token = G_TOKEN_INT;
1403           ch = g_scanner_peek_next_char (scanner);
1404           if (config->scan_hex && (ch == 'x' || ch == 'X'))
1405             {
1406               token = G_TOKEN_HEX;
1407               g_scanner_get_char (scanner, line_p, position_p);
1408               ch = g_scanner_get_char (scanner, line_p, position_p);
1409               if (ch == 0)
1410                 {
1411                   token = G_TOKEN_ERROR;
1412                   value.v_error = G_ERR_UNEXP_EOF;
1413                   (*position_p)++;
1414                   break;
1415                 }
1416               if (g_scanner_char_2_num (ch, 16) < 0)
1417                 {
1418                   token = G_TOKEN_ERROR;
1419                   value.v_error = G_ERR_DIGIT_RADIX;
1420                   ch = 0;
1421                   break;
1422                 }
1423             }
1424           else if (config->scan_binary && (ch == 'b' || ch == 'B'))
1425             {
1426               token = G_TOKEN_BINARY;
1427               g_scanner_get_char (scanner, line_p, position_p);
1428               ch = g_scanner_get_char (scanner, line_p, position_p);
1429               if (ch == 0)
1430                 {
1431                   token = G_TOKEN_ERROR;
1432                   value.v_error = G_ERR_UNEXP_EOF;
1433                   (*position_p)++;
1434                   break;
1435                 }
1436               if (g_scanner_char_2_num (ch, 10) < 0)
1437                 {
1438                   token = G_TOKEN_ERROR;
1439                   value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1440                   ch = 0;
1441                   break;
1442                 }
1443             }
1444           else
1445             ch = '0';
1446           /* fall through */
1447         case '1':
1448         case '2':
1449         case '3':
1450         case '4':
1451         case '5':
1452         case '6':
1453         case '7':
1454         case '8':
1455         case '9':
1456         number_parsing:
1457         {
1458           gboolean in_number = TRUE;
1459           gchar *endptr;
1460
1461           if (token == G_TOKEN_NONE)
1462             token = G_TOKEN_INT;
1463
1464           gstring = g_string_new (dotted_float ? "0." : "");
1465           gstring = g_string_append_c (gstring, ch);
1466
1467           do /* while (in_number) */
1468             {
1469               gboolean is_E;
1470
1471               is_E = token == G_TOKEN_FLOAT && (ch == 'e' || ch == 'E');
1472
1473               ch = g_scanner_peek_next_char (scanner);
1474
1475               if (g_scanner_char_2_num (ch, 36) >= 0 ||
1476                   (config->scan_float && ch == '.') ||
1477                   (is_E && (ch == '+' || ch == '-')))
1478                 {
1479                   ch = g_scanner_get_char (scanner, line_p, position_p);
1480
1481                   switch (ch)
1482                     {
1483                     case '.':
1484                       if (token != G_TOKEN_INT && token != G_TOKEN_OCTAL)
1485                         {
1486                           value.v_error = token == G_TOKEN_FLOAT ? G_ERR_FLOAT_MALFORMED : G_ERR_FLOAT_RADIX;
1487                           token = G_TOKEN_ERROR;
1488                           in_number = FALSE;
1489                         }
1490                       else
1491                         {
1492                           token = G_TOKEN_FLOAT;
1493                           gstring = g_string_append_c (gstring, ch);
1494                         }
1495                       break;
1496
1497                     case '0':
1498                     case '1':
1499                     case '2':
1500                     case '3':
1501                     case '4':
1502                     case '5':
1503                     case '6':
1504                     case '7':
1505                     case '8':
1506                     case '9':
1507                       gstring = g_string_append_c (gstring, ch);
1508                       break;
1509
1510                     case '-':
1511                     case '+':
1512                       if (token != G_TOKEN_FLOAT)
1513                         {
1514                           token = G_TOKEN_ERROR;
1515                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1516                           in_number = FALSE;
1517                         }
1518                       else
1519                         gstring = g_string_append_c (gstring, ch);
1520                       break;
1521
1522                     case 'e':
1523                     case 'E':
1524                       if ((token != G_TOKEN_HEX && !config->scan_float) ||
1525                           (token != G_TOKEN_HEX &&
1526                            token != G_TOKEN_OCTAL &&
1527                            token != G_TOKEN_FLOAT &&
1528                            token != G_TOKEN_INT))
1529                         {
1530                           token = G_TOKEN_ERROR;
1531                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1532                           in_number = FALSE;
1533                         }
1534                       else
1535                         {
1536                           if (token != G_TOKEN_HEX)
1537                             token = G_TOKEN_FLOAT;
1538                           gstring = g_string_append_c (gstring, ch);
1539                         }
1540                       break;
1541
1542                     default:
1543                       if (token != G_TOKEN_HEX)
1544                         {
1545                           token = G_TOKEN_ERROR;
1546                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1547                           in_number = FALSE;
1548                         }
1549                       else
1550                         gstring = g_string_append_c (gstring, ch);
1551                       break;
1552                     }
1553                 }
1554               else
1555                 in_number = FALSE;
1556             }
1557           while (in_number);
1558
1559           endptr = NULL;
1560           switch (token)
1561             {
1562             case G_TOKEN_BINARY:
1563               value.v_binary = strtol (gstring->str, &endptr, 2);
1564               break;
1565
1566             case G_TOKEN_OCTAL:
1567               value.v_octal = strtol (gstring->str, &endptr, 8);
1568               break;
1569
1570             case G_TOKEN_INT:
1571               value.v_int = strtol (gstring->str, &endptr, 10);
1572               break;
1573
1574             case G_TOKEN_FLOAT:
1575               value.v_float = g_strtod (gstring->str, &endptr);
1576               break;
1577
1578             case G_TOKEN_HEX:
1579               value.v_hex = strtol (gstring->str, &endptr, 16);
1580               break;
1581
1582             default:
1583               break;
1584             }
1585           if (endptr && *endptr)
1586             {
1587               token = G_TOKEN_ERROR;
1588               if (*endptr == 'e' || *endptr == 'E')
1589                 value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1590               else
1591                 value.v_error = G_ERR_DIGIT_RADIX;
1592             }
1593           g_string_free (gstring, TRUE);
1594           gstring = NULL;
1595           ch = 0;
1596         } /* number_parsing:... */
1597         break;
1598
1599         default:
1600         default_case:
1601         {
1602           if (config->cpair_comment_single &&
1603               ch == config->cpair_comment_single[0])
1604             {
1605               token = G_TOKEN_COMMENT_SINGLE;
1606               in_comment_single = TRUE;
1607               gstring = g_string_new ("");
1608               ch = g_scanner_get_char (scanner, line_p, position_p);
1609               while (ch != 0)
1610                 {
1611                   if (ch == config->cpair_comment_single[1])
1612                     {
1613                       in_comment_single = FALSE;
1614                       ch = 0;
1615                       break;
1616                     }
1617
1618                   gstring = g_string_append_c (gstring, ch);
1619                   ch = g_scanner_get_char (scanner, line_p, position_p);
1620                 }
1621             }
1622           else if (config->scan_identifier && ch &&
1623                    strchr (config->cset_identifier_first, ch))
1624             {
1625             identifier_precedence:
1626
1627               if (config->cset_identifier_nth && ch &&
1628                   strchr (config->cset_identifier_nth,
1629                           g_scanner_peek_next_char (scanner)))
1630                 {
1631                   token = G_TOKEN_IDENTIFIER;
1632                   gstring = g_string_new ("");
1633                   gstring = g_string_append_c (gstring, ch);
1634                   do
1635                     {
1636                       ch = g_scanner_get_char (scanner, line_p, position_p);
1637                       gstring = g_string_append_c (gstring, ch);
1638                       ch = g_scanner_peek_next_char (scanner);
1639                     }
1640                   while (ch && strchr (config->cset_identifier_nth, ch));
1641                   ch = 0;
1642                 }
1643               else if (config->scan_identifier_1char)
1644                 {
1645                   token = G_TOKEN_IDENTIFIER;
1646                   value.v_identifier = g_new0 (gchar, 2);
1647                   value.v_identifier[0] = ch;
1648                   ch = 0;
1649                 }
1650             }
1651           if (ch)
1652             {
1653               if (config->char_2_token)
1654                 token = ch;
1655               else
1656                 {
1657                   token = G_TOKEN_CHAR;
1658                   value.v_char = ch;
1659                 }
1660               ch = 0;
1661             }
1662         } /* default_case:... */
1663         break;
1664         }
1665       g_assert (ch == 0 && token != G_TOKEN_NONE); /* paranoid */
1666     }
1667   while (ch != 0);
1668
1669   if (in_comment_multi || in_comment_single ||
1670       in_string_sq || in_string_dq)
1671     {
1672       token = G_TOKEN_ERROR;
1673       if (gstring)
1674         {
1675           g_string_free (gstring, TRUE);
1676           gstring = NULL;
1677         }
1678       (*position_p)++;
1679       if (in_comment_multi || in_comment_single)
1680         value.v_error = G_ERR_UNEXP_EOF_IN_COMMENT;
1681       else /* (in_string_sq || in_string_dq) */
1682         value.v_error = G_ERR_UNEXP_EOF_IN_STRING;
1683     }
1684
1685   if (gstring)
1686     {
1687       value.v_string = gstring->str;
1688       g_string_free (gstring, FALSE);
1689       gstring = NULL;
1690     }
1691
1692   if (token == G_TOKEN_IDENTIFIER)
1693     {
1694       if (config->scan_symbols)
1695         {
1696           GScannerKey *key;
1697           guint scope_id;
1698
1699           scope_id = scanner->scope_id;
1700           key = g_scanner_lookup_internal (scanner, scope_id, value.v_identifier);
1701           if (!key && scope_id && scanner->config->scope_0_fallback)
1702             key = g_scanner_lookup_internal (scanner, 0, value.v_identifier);
1703
1704           if (key)
1705             {
1706               g_free (value.v_identifier);
1707               token = G_TOKEN_SYMBOL;
1708               value.v_symbol = key->value;
1709             }
1710         }
1711
1712       if (token == G_TOKEN_IDENTIFIER &&
1713           config->scan_identifier_NULL &&
1714           strlen (value.v_identifier) == 4)
1715         {
1716           gchar *null_upper = "NULL";
1717           gchar *null_lower = "null";
1718
1719           if (scanner->config->case_sensitive)
1720             {
1721               if (value.v_identifier[0] == null_upper[0] &&
1722                   value.v_identifier[1] == null_upper[1] &&
1723                   value.v_identifier[2] == null_upper[2] &&
1724                   value.v_identifier[3] == null_upper[3])
1725                 token = G_TOKEN_IDENTIFIER_NULL;
1726             }
1727           else
1728             {
1729               if ((value.v_identifier[0] == null_upper[0] ||
1730                    value.v_identifier[0] == null_lower[0]) &&
1731                   (value.v_identifier[1] == null_upper[1] ||
1732                    value.v_identifier[1] == null_lower[1]) &&
1733                   (value.v_identifier[2] == null_upper[2] ||
1734                    value.v_identifier[2] == null_lower[2]) &&
1735                   (value.v_identifier[3] == null_upper[3] ||
1736                    value.v_identifier[3] == null_lower[3]))
1737                 token = G_TOKEN_IDENTIFIER_NULL;
1738             }
1739         }
1740     }
1741
1742   *token_p = token;
1743   *value_p = value;
1744 }