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