gscanner.c

   1 /* GLIB - Library of useful routines for C programming
   2  * Copyright (C) 1995-1997  Peter Mattis, Spencer Kimball and Josh MacDonald
   3  *
   4  * GScanner: Flexible lexical scanner for general purpose.
   5  * Copyright (C) 1997, 1998 Tim Janik
   6  *
   7  * This library is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Library General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2 of the License, or (at your option) any later version.
  11  *
  12  * This library is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Library General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Library General Public
  18  * License along with this library; if not, write to the
  19  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  20  * Boston, MA 02111-1307, USA.
  21  */
  22
  23 /*
  24  * Modified by the GLib Team and others 1997-1999.  See the AUTHORS
  25  * file for a list of people on the GLib Team.  See the ChangeLog
  26  * files for a list of changes.  These files are distributed with
  27  * GLib at ftp://ftp.gtk.org/pub/gtk/.
  28  */
  29
  30 /*
  31  * MT safe
  32  */
  33
  34 #define         __gscanner_c__
  35
  36 #ifdef HAVE_CONFIG_H
  37 #include <config.h>
  38 #endif
  39
  40 #include        <stdlib.h>
  41 #include        <stdarg.h>
  42 #include        <string.h>
  43 #include        <stdio.h>
  44 #include        "glib.h"
  45 #ifdef HAVE_UNISTD_H
  46 #include        <unistd.h>
  47 #endif
  48 #include        <errno.h>
  49 #include        <sys/types.h>   /* needed for sys/stat.h */
  50 #include        <sys/stat.h>
  51 #ifdef 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   g_return_if_fail (scanner != NULL);
 569
 570   g_hash_table_freeze (scanner->symbol_table);
 571 }
 572
 573 void
 574 g_scanner_thaw_symbol_table (GScanner *scanner)
 575 {
 576   g_return_if_fail (scanner != NULL);
 577
 578   g_hash_table_thaw (scanner->symbol_table);
 579 }
 580
 581 GTokenType
 582 g_scanner_peek_next_token (GScanner     *scanner)
 583 {
 584   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 585
 586   if (scanner->next_token == G_TOKEN_NONE)
 587     {
 588       scanner->next_line = scanner->line;
 589       scanner->next_position = scanner->position;
 590       g_scanner_get_token_i (scanner,
 591                              &scanner->next_token,
 592                              &scanner->next_value,
 593                              &scanner->next_line,
 594                              &scanner->next_position);
 595     }
 596
 597   return scanner->next_token;
 598 }
 599
 600 GTokenType
 601 g_scanner_get_next_token (GScanner      *scanner)
 602 {
 603   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 604
 605   if (scanner->next_token != G_TOKEN_NONE)
 606     {
 607       g_scanner_free_value (&scanner->token, &scanner->value);
 608
 609       scanner->token = scanner->next_token;
 610       scanner->value = scanner->next_value;
 611       scanner->line = scanner->next_line;
 612       scanner->position = scanner->next_position;
 613       scanner->next_token = G_TOKEN_NONE;
 614     }
 615   else
 616     g_scanner_get_token_i (scanner,
 617                            &scanner->token,
 618                            &scanner->value,
 619                            &scanner->line,
 620                            &scanner->position);
 621
 622   return scanner->token;
 623 }
 624
 625 GTokenType
 626 g_scanner_cur_token (GScanner *scanner)
 627 {
 628   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 629
 630   return scanner->token;
 631 }
 632
 633 GTokenValue
 634 g_scanner_cur_value (GScanner *scanner)
 635 {
 636   GTokenValue v;
 637
 638   v.v_int = 0;
 639
 640   g_return_val_if_fail (scanner != NULL, v);
 641
 642   /* MSC isn't capable of handling return scanner->value; ? */
 643
 644   v = scanner->value;
 645
 646   return v;
 647 }
 648
 649 guint
 650 g_scanner_cur_line (GScanner *scanner)
 651 {
 652   g_return_val_if_fail (scanner != NULL, 0);
 653
 654   return scanner->line;
 655 }
 656
 657 guint
 658 g_scanner_cur_position (GScanner *scanner)
 659 {
 660   g_return_val_if_fail (scanner != NULL, 0);
 661
 662   return scanner->position;
 663 }
 664
 665 gboolean
 666 g_scanner_eof (GScanner *scanner)
 667 {
 668   g_return_val_if_fail (scanner != NULL, TRUE);
 669
 670   return scanner->token == G_TOKEN_EOF;
 671 }
 672
 673 void
 674 g_scanner_input_file (GScanner *scanner,
 675                       gint      input_fd)
 676 {
 677   g_return_if_fail (scanner != NULL);
 678   g_return_if_fail (input_fd >= 0);
 679
 680   if (scanner->input_fd >= 0)
 681     g_scanner_sync_file_offset (scanner);
 682
 683   scanner->token = G_TOKEN_NONE;
 684   scanner->value.v_int = 0;
 685   scanner->line = 1;
 686   scanner->position = 0;
 687   scanner->next_token = G_TOKEN_NONE;
 688
 689   scanner->input_fd = input_fd;
 690   scanner->text = NULL;
 691   scanner->text_end = NULL;
 692
 693   if (!scanner->buffer)
 694     scanner->buffer = g_new (gchar, READ_BUFFER_SIZE + 1);
 695 }
 696
 697 void
 698 g_scanner_input_text (GScanner    *scanner,
 699                       const gchar *text,
 700                       guint        text_len)
 701 {
 702   g_return_if_fail (scanner != NULL);
 703   if (text_len)
 704     g_return_if_fail (text != NULL);
 705   else
 706     text = NULL;
 707
 708   if (scanner->input_fd >= 0)
 709     g_scanner_sync_file_offset (scanner);
 710
 711   scanner->token = G_TOKEN_NONE;
 712   scanner->value.v_int = 0;
 713   scanner->line = 1;
 714   scanner->position = 0;
 715   scanner->next_token = G_TOKEN_NONE;
 716
 717   scanner->input_fd = -1;
 718   scanner->text = text;
 719   scanner->text_end = text + text_len;
 720
 721   if (scanner->buffer)
 722     {
 723       g_free (scanner->buffer);
 724       scanner->buffer = NULL;
 725     }
 726 }
 727
 728 static guchar
 729 g_scanner_peek_next_char (GScanner *scanner)
 730 {
 731   if (scanner->text < scanner->text_end)
 732     {
 733       return *scanner->text;
 734     }
 735   else if (scanner->input_fd >= 0)
 736     {
 737       gint count;
 738       gchar *buffer;
 739
 740       buffer = scanner->buffer;
 741       do
 742         {
 743           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 744         }
 745       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 746
 747       if (count < 1)
 748         {
 749           scanner->input_fd = -1;
 750
 751           return 0;
 752         }
 753       else
 754         {
 755           scanner->text = buffer;
 756           scanner->text_end = buffer + count;
 757
 758           return *buffer;
 759         }
 760     }
 761   else
 762     return 0;
 763 }
 764
 765 void
 766 g_scanner_sync_file_offset (GScanner *scanner)
 767 {
 768   g_return_if_fail (scanner != NULL);
 769
 770   /* for file input, rewind the filedescriptor to the current
 771    * buffer position and blow the file read ahead buffer. usefull for
 772    * third party uses of our filedescriptor, which hooks onto the current
 773    * scanning position.
 774    */
 775
 776   if (scanner->input_fd >= 0 && scanner->text_end > scanner->text)
 777     {
 778       gint buffered;
 779
 780       buffered = scanner->text_end - scanner->text;
 781       if (lseek (scanner->input_fd, - buffered, SEEK_CUR) >= 0)
 782         {
 783           /* we succeeded, blow our buffer's contents now */
 784           scanner->text = NULL;
 785           scanner->text_end = NULL;
 786         }
 787       else
 788         errno = 0;
 789     }
 790 }
 791
 792 static guchar
 793 g_scanner_get_char (GScanner    *scanner,
 794                     guint       *line_p,
 795                     guint       *position_p)
 796 {
 797   guchar fchar;
 798
 799   if (scanner->text < scanner->text_end)
 800     fchar = *(scanner->text++);
 801   else if (scanner->input_fd >= 0)
 802     {
 803       gint count;
 804       gchar *buffer;
 805
 806       buffer = scanner->buffer;
 807       do
 808         {
 809           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 810         }
 811       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 812
 813       if (count < 1)
 814         {
 815           scanner->input_fd = -1;
 816           fchar = 0;
 817         }
 818       else
 819         {
 820           scanner->text = buffer + 1;
 821           scanner->text_end = buffer + count;
 822           fchar = *buffer;
 823           if (!fchar)
 824             {
 825               g_scanner_sync_file_offset (scanner);
 826               scanner->text_end = scanner->text;
 827               scanner->input_fd = -1;
 828             }
 829         }
 830     }
 831   else
 832     fchar = 0;
 833
 834   if (fchar == '\n')
 835     {
 836       (*position_p) = 0;
 837       (*line_p)++;
 838     }
 839   else if (fchar)
 840     {
 841       (*position_p)++;
 842     }
 843
 844   return fchar;
 845 }
 846
 847 void
 848 g_scanner_unexp_token (GScanner         *scanner,
 849                        GTokenType        expected_token,
 850                        const gchar      *identifier_spec,
 851                        const gchar      *symbol_spec,
 852                        const gchar      *symbol_name,
 853                        const gchar      *message,
 854                        gint              is_error)
 855 {
 856   gchar *token_string;
 857   guint token_string_len;
 858   gchar *expected_string;
 859   guint expected_string_len;
 860   gchar *message_prefix;
 861   gboolean print_unexp;
 862   void (*msg_handler)   (GScanner*, const gchar*, ...);
 863
 864   g_return_if_fail (scanner != NULL);
 865
 866   if (is_error)
 867     msg_handler = g_scanner_error;
 868   else
 869     msg_handler = g_scanner_warn;
 870
 871   if (!identifier_spec)
 872     identifier_spec = "identifier";
 873   if (!symbol_spec)
 874     symbol_spec = "symbol";
 875
 876   token_string_len = 56;
 877   token_string = g_new (gchar, token_string_len + 1);
 878   expected_string_len = 64;
 879   expected_string = g_new (gchar, expected_string_len + 1);
 880   print_unexp = TRUE;
 881
 882   switch (scanner->token)
 883     {
 884     case G_TOKEN_EOF:
 885       g_snprintf (token_string, token_string_len, "end of file");
 886       break;
 887
 888     default:
 889       if (scanner->token >= 1 && scanner->token <= 255)
 890         {
 891           if ((scanner->token >= ' ' && scanner->token <= '~') ||
 892               strchr (scanner->config->cset_identifier_first, scanner->token) ||
 893               strchr (scanner->config->cset_identifier_nth, scanner->token))
 894             g_snprintf (token_string, expected_string_len, "character `%c'", scanner->token);
 895           else
 896             g_snprintf (token_string, expected_string_len, "character `\\%o'", scanner->token);
 897           break;
 898         }
 899       else if (!scanner->config->symbol_2_token)
 900         {
 901           g_snprintf (token_string, token_string_len, "(unknown) token <%d>", scanner->token);
 902           break;
 903         }
 904       /* fall through */
 905     case G_TOKEN_SYMBOL:
 906       if (expected_token == G_TOKEN_SYMBOL ||
 907           (scanner->config->symbol_2_token &&
 908            expected_token > G_TOKEN_LAST))
 909         print_unexp = FALSE;
 910       if (symbol_name)
 911         g_snprintf (token_string,
 912                     token_string_len,
 913                     "%s%s `%s'",
 914                     print_unexp ? "" : "invalid ",
 915                     symbol_spec,
 916                     symbol_name);
 917       else
 918         g_snprintf (token_string,
 919                     token_string_len,
 920                     "%s%s",
 921                     print_unexp ? "" : "invalid ",
 922                     symbol_spec);
 923       break;
 924
 925     case G_TOKEN_ERROR:
 926       print_unexp = FALSE;
 927       expected_token = G_TOKEN_NONE;
 928       switch (scanner->value.v_error)
 929         {
 930         case G_ERR_UNEXP_EOF:
 931           g_snprintf (token_string, token_string_len, "scanner: unexpected end of file");
 932           break;
 933
 934         case G_ERR_UNEXP_EOF_IN_STRING:
 935           g_snprintf (token_string, token_string_len, "scanner: unterminated string constant");
 936           break;
 937
 938         case G_ERR_UNEXP_EOF_IN_COMMENT:
 939           g_snprintf (token_string, token_string_len, "scanner: unterminated comment");
 940           break;
 941
 942         case G_ERR_NON_DIGIT_IN_CONST:
 943           g_snprintf (token_string, token_string_len, "scanner: non digit in constant");
 944           break;
 945
 946         case G_ERR_FLOAT_RADIX:
 947           g_snprintf (token_string, token_string_len, "scanner: invalid radix for floating constant");
 948           break;
 949
 950         case G_ERR_FLOAT_MALFORMED:
 951           g_snprintf (token_string, token_string_len, "scanner: malformed floating constant");
 952           break;
 953
 954         case G_ERR_DIGIT_RADIX:
 955           g_snprintf (token_string, token_string_len, "scanner: digit is beyond radix");
 956           break;
 957
 958         case G_ERR_UNKNOWN:
 959         default:
 960           g_snprintf (token_string, token_string_len, "scanner: unknown error");
 961           break;
 962         }
 963       break;
 964
 965     case G_TOKEN_CHAR:
 966       g_snprintf (token_string, token_string_len, "character `%c'", scanner->value.v_char);
 967       break;
 968
 969     case G_TOKEN_IDENTIFIER:
 970     case G_TOKEN_IDENTIFIER_NULL:
 971       if (expected_token == G_TOKEN_IDENTIFIER ||
 972           expected_token == G_TOKEN_IDENTIFIER_NULL)
 973         print_unexp = FALSE;
 974       g_snprintf (token_string,
 975                   token_string_len,
 976                   "%s%s `%s'",
 977                   print_unexp ? "" : "invalid ",
 978                   identifier_spec,
 979                   scanner->token == G_TOKEN_IDENTIFIER ? scanner->value.v_string : "null");
 980       break;
 981
 982     case G_TOKEN_BINARY:
 983     case G_TOKEN_OCTAL:
 984     case G_TOKEN_INT:
 985     case G_TOKEN_HEX:
 986       g_snprintf (token_string, token_string_len, "number `%ld'", scanner->value.v_int);
 987       break;
 988
 989     case G_TOKEN_FLOAT:
 990       g_snprintf (token_string, token_string_len, "number `%.3f'", scanner->value.v_float);
 991       break;
 992
 993     case G_TOKEN_STRING:
 994       if (expected_token == G_TOKEN_STRING)
 995         print_unexp = FALSE;
 996       g_snprintf (token_string,
 997                   token_string_len,
 998                   "%s%sstring constant \"%s\"",
 999                   print_unexp ? "" : "invalid ",
1000                   scanner->value.v_string[0] == 0 ? "empty " : "",
1001                   scanner->value.v_string);
1002       token_string[token_string_len - 2] = '"';
1003       token_string[token_string_len - 1] = 0;
1004       break;
1005
1006     case G_TOKEN_COMMENT_SINGLE:
1007     case G_TOKEN_COMMENT_MULTI:
1008       g_snprintf (token_string, token_string_len, "comment");
1009       break;
1010
1011     case G_TOKEN_NONE:
1012       /* somehow the user's parsing code is screwed, there isn't much
1013        * we can do about it.
1014        * Note, a common case to trigger this is
1015        * g_scanner_peek_next_token(); g_scanner_unexp_token();
1016        * without an intermediate g_scanner_get_next_token().
1017        */
1018       g_assert_not_reached ();
1019       break;
1020     }
1021
1022
1023   switch (expected_token)
1024     {
1025       gboolean need_valid;
1026
1027     default:
1028       if (expected_token >= 1 && expected_token <= 255)
1029         {
1030           if ((expected_token >= ' ' && expected_token <= '~') ||
1031               strchr (scanner->config->cset_identifier_first, expected_token) ||
1032               strchr (scanner->config->cset_identifier_nth, expected_token))
1033             g_snprintf (expected_string, expected_string_len, "character `%c'", expected_token);
1034           else
1035             g_snprintf (expected_string, expected_string_len, "character `\\%o'", expected_token);
1036           break;
1037         }
1038       else if (!scanner->config->symbol_2_token)
1039         {
1040           g_snprintf (expected_string, expected_string_len, "(unknown) token <%d>", expected_token);
1041           break;
1042         }
1043       /* fall through */
1044     case G_TOKEN_SYMBOL:
1045       need_valid = (scanner->token == G_TOKEN_SYMBOL ||
1046                     (scanner->config->symbol_2_token &&
1047                      scanner->token > G_TOKEN_LAST));
1048       g_snprintf (expected_string,
1049                   expected_string_len,
1050                   "%s%s",
1051                   need_valid ? "valid " : "",
1052                   symbol_spec);
1053       /* FIXME: should we attempt to lookup the symbol_name for symbol_2_token? */
1054       break;
1055
1056     case G_TOKEN_INT:
1057       g_snprintf (expected_string, expected_string_len, "number (integer)");
1058       break;
1059
1060     case G_TOKEN_FLOAT:
1061       g_snprintf (expected_string, expected_string_len, "number (float)");
1062       break;
1063
1064     case G_TOKEN_STRING:
1065       g_snprintf (expected_string,
1066                   expected_string_len,
1067                   "%sstring constant",
1068                   scanner->token == G_TOKEN_STRING ? "valid " : "");
1069       break;
1070
1071     case G_TOKEN_IDENTIFIER:
1072     case G_TOKEN_IDENTIFIER_NULL:
1073       g_snprintf (expected_string,
1074                   expected_string_len,
1075                   "%s%s",
1076                   (scanner->token == G_TOKEN_IDENTIFIER_NULL ||
1077                    scanner->token == G_TOKEN_IDENTIFIER ? "valid " : ""),
1078                   identifier_spec);
1079       break;
1080
1081     case G_TOKEN_EOF:
1082       g_snprintf (expected_string, expected_string_len, "end of file");
1083       break;
1084
1085     case G_TOKEN_NONE:
1086       break;
1087     }
1088
1089   if (message && message[0] != 0)
1090     message_prefix = " - ";
1091   else
1092     {
1093       message_prefix = "";
1094       message = "";
1095     }
1096
1097   if (expected_token != G_TOKEN_NONE)
1098     {
1099       if (print_unexp)
1100         msg_handler (scanner,
1101                      "unexpected %s, expected %s%s%s",
1102                      token_string,
1103                      expected_string,
1104                      message_prefix,
1105                      message);
1106       else
1107         msg_handler (scanner,
1108                      "%s, expected %s%s%s",
1109                      token_string,
1110                      expected_string,
1111                      message_prefix,
1112                      message);
1113     }
1114   else
1115     {
1116       if (print_unexp)
1117         msg_handler (scanner,
1118                      "unexpected %s%s%s",
1119                      token_string,
1120                      message_prefix,
1121                      message);
1122       else
1123         msg_handler (scanner,
1124                      "%s%s%s",
1125                      token_string,
1126                      message_prefix,
1127                      message);
1128     }
1129
1130   g_free (token_string);
1131   g_free (expected_string);
1132 }
1133
1134 gint
1135 g_scanner_stat_mode (const gchar *filename)
1136 {
1137   struct stat  *stat_buf;
1138   gint          st_mode;
1139
1140   stat_buf = g_new0 (struct stat, 1);
1141 #ifdef HAVE_LSTAT
1142   lstat (filename, stat_buf);
1143 #else
1144   stat (filename, stat_buf);
1145 #endif
1146   st_mode = stat_buf->st_mode;
1147
1148   g_free (stat_buf);
1149
1150   return st_mode;
1151 }
1152
1153 static void
1154 g_scanner_get_token_i (GScanner *scanner,
1155                        GTokenType       *token_p,
1156                        GTokenValue      *value_p,
1157                        guint            *line_p,
1158                        guint            *position_p)
1159 {
1160   do
1161     {
1162       g_scanner_free_value (token_p, value_p);
1163       g_scanner_get_token_ll (scanner, token_p, value_p, line_p, position_p);
1164     }
1165   while (((*token_p > 0 && *token_p < 256) &&
1166           strchr (scanner->config->cset_skip_characters, *token_p)) ||
1167          (*token_p == G_TOKEN_CHAR &&
1168           strchr (scanner->config->cset_skip_characters, value_p->v_char)) ||
1169          (*token_p == G_TOKEN_COMMENT_MULTI &&
1170           scanner->config->skip_comment_multi) ||
1171          (*token_p == G_TOKEN_COMMENT_SINGLE &&
1172           scanner->config->skip_comment_single));
1173
1174   switch (*token_p)
1175     {
1176     case G_TOKEN_IDENTIFIER:
1177       if (scanner->config->identifier_2_string)
1178         *token_p = G_TOKEN_STRING;
1179       break;
1180
1181     case G_TOKEN_SYMBOL:
1182       if (scanner->config->symbol_2_token)
1183         *token_p = (GTokenType) value_p->v_symbol;
1184       break;
1185
1186     case G_TOKEN_BINARY:
1187     case G_TOKEN_OCTAL:
1188     case G_TOKEN_HEX:
1189       if (scanner->config->numbers_2_int)
1190         *token_p = G_TOKEN_INT;
1191       break;
1192
1193     default:
1194       break;
1195     }
1196
1197   if (*token_p == G_TOKEN_INT &&
1198       scanner->config->int_2_float)
1199     {
1200       *token_p = G_TOKEN_FLOAT;
1201       value_p->v_float = value_p->v_int;
1202     }
1203
1204   errno = 0;
1205 }
1206
1207 static void
1208 g_scanner_get_token_ll  (GScanner       *scanner,
1209                          GTokenType     *token_p,
1210                          GTokenValue    *value_p,
1211                          guint          *line_p,
1212                          guint          *position_p)
1213 {
1214   GScannerConfig *config;
1215   GTokenType       token;
1216   gboolean         in_comment_multi;
1217   gboolean         in_comment_single;
1218   gboolean         in_string_sq;
1219   gboolean         in_string_dq;
1220   GString         *gstring;
1221   GTokenValue      value;
1222   guchar           ch;
1223
1224   config = scanner->config;
1225   (*value_p).v_int = 0;
1226
1227   if ((scanner->text >= scanner->text_end && scanner->input_fd < 0) ||
1228       scanner->token == G_TOKEN_EOF)
1229     {
1230       *token_p = G_TOKEN_EOF;
1231       return;
1232     }
1233
1234   in_comment_multi = FALSE;
1235   in_comment_single = FALSE;
1236   in_string_sq = FALSE;
1237   in_string_dq = FALSE;
1238   gstring = NULL;
1239
1240   do /* while (ch != 0) */
1241     {
1242       gboolean dotted_float = FALSE;
1243
1244       ch = g_scanner_get_char (scanner, line_p, position_p);
1245
1246       value.v_int = 0;
1247       token = G_TOKEN_NONE;
1248
1249       /* this is *evil*, but needed ;(
1250        * we first check for identifier first character, because  it
1251        * might interfere with other key chars like slashes or numbers
1252        */
1253       if (config->scan_identifier &&
1254           ch && strchr (config->cset_identifier_first, ch))
1255         goto identifier_precedence;
1256
1257       switch (ch)
1258         {
1259         case 0:
1260           token = G_TOKEN_EOF;
1261           (*position_p)++;
1262           /* ch = 0; */
1263           break;
1264
1265         case '/':
1266           if (!config->scan_comment_multi ||
1267               g_scanner_peek_next_char (scanner) != '*')
1268             goto default_case;
1269           g_scanner_get_char (scanner, line_p, position_p);
1270           token = G_TOKEN_COMMENT_MULTI;
1271           in_comment_multi = TRUE;
1272           gstring = g_string_new ("");
1273           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1274             {
1275               if (ch == '*' && g_scanner_peek_next_char (scanner) == '/')
1276                 {
1277                   g_scanner_get_char (scanner, line_p, position_p);
1278                   in_comment_multi = FALSE;
1279                   break;
1280                 }
1281               else
1282                 gstring = g_string_append_c (gstring, ch);
1283             }
1284           ch = 0;
1285           break;
1286
1287         case '\'':
1288           if (!config->scan_string_sq)
1289             goto default_case;
1290           token = G_TOKEN_STRING;
1291           in_string_sq = TRUE;
1292           gstring = g_string_new ("");
1293           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1294             {
1295               if (ch == '\'')
1296                 {
1297                   in_string_sq = FALSE;
1298                   break;
1299                 }
1300               else
1301                 gstring = g_string_append_c (gstring, ch);
1302             }
1303           ch = 0;
1304           break;
1305
1306         case '"':
1307           if (!config->scan_string_dq)
1308             goto default_case;
1309           token = G_TOKEN_STRING;
1310           in_string_dq = TRUE;
1311           gstring = g_string_new ("");
1312           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1313             {
1314               if (ch == '"')
1315                 {
1316                   in_string_dq = FALSE;
1317                   break;
1318                 }
1319               else
1320                 {
1321                   if (ch == '\\')
1322                     {
1323                       ch = g_scanner_get_char (scanner, line_p, position_p);
1324                       switch (ch)
1325                         {
1326                           guint i;
1327                           guint fchar;
1328
1329                         case 0:
1330                           break;
1331
1332                         case '\\':
1333                           gstring = g_string_append_c (gstring, '\\');
1334                           break;
1335
1336                         case 'n':
1337                           gstring = g_string_append_c (gstring, '\n');
1338                           break;
1339
1340                         case 't':
1341                           gstring = g_string_append_c (gstring, '\t');
1342                           break;
1343
1344                         case 'r':
1345                           gstring = g_string_append_c (gstring, '\r');
1346                           break;
1347
1348                         case 'b':
1349                           gstring = g_string_append_c (gstring, '\b');
1350                           break;
1351
1352                         case 'f':
1353                           gstring = g_string_append_c (gstring, '\f');
1354                           break;
1355
1356                         case '0':
1357                         case '1':
1358                         case '2':
1359                         case '3':
1360                         case '4':
1361                         case '5':
1362                         case '6':
1363                         case '7':
1364                           i = ch - '0';
1365                           fchar = g_scanner_peek_next_char (scanner);
1366                           if (fchar >= '0' && fchar <= '7')
1367                             {
1368                               ch = g_scanner_get_char (scanner, line_p, position_p);
1369                               i = i * 8 + ch - '0';
1370                               fchar = g_scanner_peek_next_char (scanner);
1371                               if (fchar >= '0' && fchar <= '7')
1372                                 {
1373                                   ch = g_scanner_get_char (scanner, line_p, position_p);
1374                                   i = i * 8 + ch - '0';
1375                                 }
1376                             }
1377                           gstring = g_string_append_c (gstring, i);
1378                           break;
1379
1380                         default:
1381                           gstring = g_string_append_c (gstring, ch);
1382                           break;
1383                         }
1384                     }
1385                   else
1386                     gstring = g_string_append_c (gstring, ch);
1387                 }
1388             }
1389           ch = 0;
1390           break;
1391
1392         case '.':
1393           if (!config->scan_float)
1394             goto default_case;
1395           token = G_TOKEN_FLOAT;
1396           dotted_float = TRUE;
1397           ch = g_scanner_get_char (scanner, line_p, position_p);
1398           goto number_parsing;
1399
1400         case '$':
1401           if (!config->scan_hex_dollar)
1402             goto default_case;
1403           token = G_TOKEN_HEX;
1404           ch = g_scanner_get_char (scanner, line_p, position_p);
1405           goto number_parsing;
1406
1407         case '0':
1408           if (config->scan_octal)
1409             token = G_TOKEN_OCTAL;
1410           else
1411             token = G_TOKEN_INT;
1412           ch = g_scanner_peek_next_char (scanner);
1413           if (config->scan_hex && (ch == 'x' || ch == 'X'))
1414             {
1415               token = G_TOKEN_HEX;
1416               g_scanner_get_char (scanner, line_p, position_p);
1417               ch = g_scanner_get_char (scanner, line_p, position_p);
1418               if (ch == 0)
1419                 {
1420                   token = G_TOKEN_ERROR;
1421                   value.v_error = G_ERR_UNEXP_EOF;
1422                   (*position_p)++;
1423                   break;
1424                 }
1425               if (g_scanner_char_2_num (ch, 16) < 0)
1426                 {
1427                   token = G_TOKEN_ERROR;
1428                   value.v_error = G_ERR_DIGIT_RADIX;
1429                   ch = 0;
1430                   break;
1431                 }
1432             }
1433           else if (config->scan_binary && (ch == 'b' || ch == 'B'))
1434             {
1435               token = G_TOKEN_BINARY;
1436               g_scanner_get_char (scanner, line_p, position_p);
1437               ch = g_scanner_get_char (scanner, line_p, position_p);
1438               if (ch == 0)
1439                 {
1440                   token = G_TOKEN_ERROR;
1441                   value.v_error = G_ERR_UNEXP_EOF;
1442                   (*position_p)++;
1443                   break;
1444                 }
1445               if (g_scanner_char_2_num (ch, 10) < 0)
1446                 {
1447                   token = G_TOKEN_ERROR;
1448                   value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1449                   ch = 0;
1450                   break;
1451                 }
1452             }
1453           else
1454             ch = '0';
1455           /* fall through */
1456         case '1':
1457         case '2':
1458         case '3':
1459         case '4':
1460         case '5':
1461         case '6':
1462         case '7':
1463         case '8':
1464         case '9':
1465         number_parsing:
1466         {
1467           gboolean in_number = TRUE;
1468           gchar *endptr;
1469
1470           if (token == G_TOKEN_NONE)
1471             token = G_TOKEN_INT;
1472
1473           gstring = g_string_new (dotted_float ? "0." : "");
1474           gstring = g_string_append_c (gstring, ch);
1475
1476           do /* while (in_number) */
1477             {
1478               gboolean is_E;
1479
1480               is_E = token == G_TOKEN_FLOAT && (ch == 'e' || ch == 'E');
1481
1482               ch = g_scanner_peek_next_char (scanner);
1483
1484               if (g_scanner_char_2_num (ch, 36) >= 0 ||
1485                   (config->scan_float && ch == '.') ||
1486                   (is_E && (ch == '+' || ch == '-')))
1487                 {
1488                   ch = g_scanner_get_char (scanner, line_p, position_p);
1489
1490                   switch (ch)
1491                     {
1492                     case '.':
1493                       if (token != G_TOKEN_INT && token != G_TOKEN_OCTAL)
1494                         {
1495                           value.v_error = token == G_TOKEN_FLOAT ? G_ERR_FLOAT_MALFORMED : G_ERR_FLOAT_RADIX;
1496                           token = G_TOKEN_ERROR;
1497                           in_number = FALSE;
1498                         }
1499                       else
1500                         {
1501                           token = G_TOKEN_FLOAT;
1502                           gstring = g_string_append_c (gstring, ch);
1503                         }
1504                       break;
1505
1506                     case '0':
1507                     case '1':
1508                     case '2':
1509                     case '3':
1510                     case '4':
1511                     case '5':
1512                     case '6':
1513                     case '7':
1514                     case '8':
1515                     case '9':
1516                       gstring = g_string_append_c (gstring, ch);
1517                       break;
1518
1519                     case '-':
1520                     case '+':
1521                       if (token != G_TOKEN_FLOAT)
1522                         {
1523                           token = G_TOKEN_ERROR;
1524                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1525                           in_number = FALSE;
1526                         }
1527                       else
1528                         gstring = g_string_append_c (gstring, ch);
1529                       break;
1530
1531                     case 'e':
1532                     case 'E':
1533                       if ((token != G_TOKEN_HEX && !config->scan_float) ||
1534                           (token != G_TOKEN_HEX &&
1535                            token != G_TOKEN_OCTAL &&
1536                            token != G_TOKEN_FLOAT &&
1537                            token != G_TOKEN_INT))
1538                         {
1539                           token = G_TOKEN_ERROR;
1540                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1541                           in_number = FALSE;
1542                         }
1543                       else
1544                         {
1545                           if (token != G_TOKEN_HEX)
1546                             token = G_TOKEN_FLOAT;
1547                           gstring = g_string_append_c (gstring, ch);
1548                         }
1549                       break;
1550
1551                     default:
1552                       if (token != G_TOKEN_HEX)
1553                         {
1554                           token = G_TOKEN_ERROR;
1555                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1556                           in_number = FALSE;
1557                         }
1558                       else
1559                         gstring = g_string_append_c (gstring, ch);
1560                       break;
1561                     }
1562                 }
1563               else
1564                 in_number = FALSE;
1565             }
1566           while (in_number);
1567
1568           endptr = NULL;
1569           switch (token)
1570             {
1571             case G_TOKEN_BINARY:
1572               value.v_binary = strtol (gstring->str, &endptr, 2);
1573               break;
1574
1575             case G_TOKEN_OCTAL:
1576               value.v_octal = strtol (gstring->str, &endptr, 8);
1577               break;
1578
1579             case G_TOKEN_INT:
1580               value.v_int = strtol (gstring->str, &endptr, 10);
1581               break;
1582
1583             case G_TOKEN_FLOAT:
1584               value.v_float = g_strtod (gstring->str, &endptr);
1585               break;
1586
1587             case G_TOKEN_HEX:
1588               value.v_hex = strtol (gstring->str, &endptr, 16);
1589               break;
1590
1591             default:
1592               break;
1593             }
1594           if (endptr && *endptr)
1595             {
1596               token = G_TOKEN_ERROR;
1597               if (*endptr == 'e' || *endptr == 'E')
1598                 value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1599               else
1600                 value.v_error = G_ERR_DIGIT_RADIX;
1601             }
1602           g_string_free (gstring, TRUE);
1603           gstring = NULL;
1604           ch = 0;
1605         } /* number_parsing:... */
1606         break;
1607
1608         default:
1609         default_case:
1610         {
1611           if (config->cpair_comment_single &&
1612               ch == config->cpair_comment_single[0])
1613             {
1614               token = G_TOKEN_COMMENT_SINGLE;
1615               in_comment_single = TRUE;
1616               gstring = g_string_new ("");
1617               ch = g_scanner_get_char (scanner, line_p, position_p);
1618               while (ch != 0)
1619                 {
1620                   if (ch == config->cpair_comment_single[1])
1621                     {
1622                       in_comment_single = FALSE;
1623                       ch = 0;
1624                       break;
1625                     }
1626
1627                   gstring = g_string_append_c (gstring, ch);
1628                   ch = g_scanner_get_char (scanner, line_p, position_p);
1629                 }
1630             }
1631           else if (config->scan_identifier && ch &&
1632                    strchr (config->cset_identifier_first, ch))
1633             {
1634             identifier_precedence:
1635
1636               if (config->cset_identifier_nth && ch &&
1637                   strchr (config->cset_identifier_nth,
1638                           g_scanner_peek_next_char (scanner)))
1639                 {
1640                   token = G_TOKEN_IDENTIFIER;
1641                   gstring = g_string_new ("");
1642                   gstring = g_string_append_c (gstring, ch);
1643                   do
1644                     {
1645                       ch = g_scanner_get_char (scanner, line_p, position_p);
1646                       gstring = g_string_append_c (gstring, ch);
1647                       ch = g_scanner_peek_next_char (scanner);
1648                     }
1649                   while (ch && strchr (config->cset_identifier_nth, ch));
1650                   ch = 0;
1651                 }
1652               else if (config->scan_identifier_1char)
1653                 {
1654                   token = G_TOKEN_IDENTIFIER;
1655                   value.v_identifier = g_new0 (gchar, 2);
1656                   value.v_identifier[0] = ch;
1657                   ch = 0;
1658                 }
1659             }
1660           if (ch)
1661             {
1662               if (config->char_2_token)
1663                 token = ch;
1664               else
1665                 {
1666                   token = G_TOKEN_CHAR;
1667                   value.v_char = ch;
1668                 }
1669               ch = 0;
1670             }
1671         } /* default_case:... */
1672         break;
1673         }
1674       g_assert (ch == 0 && token != G_TOKEN_NONE); /* paranoid */
1675     }
1676   while (ch != 0);
1677
1678   if (in_comment_multi || in_comment_single ||
1679       in_string_sq || in_string_dq)
1680     {
1681       token = G_TOKEN_ERROR;
1682       if (gstring)
1683         {
1684           g_string_free (gstring, TRUE);
1685           gstring = NULL;
1686         }
1687       (*position_p)++;
1688       if (in_comment_multi || in_comment_single)
1689         value.v_error = G_ERR_UNEXP_EOF_IN_COMMENT;
1690       else /* (in_string_sq || in_string_dq) */
1691         value.v_error = G_ERR_UNEXP_EOF_IN_STRING;
1692     }
1693
1694   if (gstring)
1695     {
1696       value.v_string = gstring->str;
1697       g_string_free (gstring, FALSE);
1698       gstring = NULL;
1699     }
1700
1701   if (token == G_TOKEN_IDENTIFIER)
1702     {
1703       if (config->scan_symbols)
1704         {
1705           GScannerKey *key;
1706           guint scope_id;
1707
1708           scope_id = scanner->scope_id;
1709           key = g_scanner_lookup_internal (scanner, scope_id, value.v_identifier);
1710           if (!key && scope_id && scanner->config->scope_0_fallback)
1711             key = g_scanner_lookup_internal (scanner, 0, value.v_identifier);
1712
1713           if (key)
1714             {
1715               g_free (value.v_identifier);
1716               token = G_TOKEN_SYMBOL;
1717               value.v_symbol = key->value;
1718             }
1719         }
1720
1721       if (token == G_TOKEN_IDENTIFIER &&
1722           config->scan_identifier_NULL &&
1723           strlen (value.v_identifier) == 4)
1724         {
1725           gchar *null_upper = "NULL";
1726           gchar *null_lower = "null";
1727
1728           if (scanner->config->case_sensitive)
1729             {
1730               if (value.v_identifier[0] == null_upper[0] &&
1731                   value.v_identifier[1] == null_upper[1] &&
1732                   value.v_identifier[2] == null_upper[2] &&
1733                   value.v_identifier[3] == null_upper[3])
1734                 token = G_TOKEN_IDENTIFIER_NULL;
1735             }
1736           else
1737             {
1738               if ((value.v_identifier[0] == null_upper[0] ||
1739                    value.v_identifier[0] == null_lower[0]) &&
1740                   (value.v_identifier[1] == null_upper[1] ||
1741                    value.v_identifier[1] == null_lower[1]) &&
1742                   (value.v_identifier[2] == null_upper[2] ||
1743                    value.v_identifier[2] == null_lower[2]) &&
1744                   (value.v_identifier[3] == null_upper[3] ||
1745                    value.v_identifier[3] == null_lower[3]))
1746                 token = G_TOKEN_IDENTIFIER_NULL;
1747             }
1748         }
1749     }
1750
1751   *token_p = token;
1752   *value_p = value;
1753 }