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 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 #define         __gscanner_c__
  23
  24 #include        <stdlib.h>
  25 #include        <stdarg.h>
  26 #include        <string.h>
  27 #include        <stdio.h>
  28 #include        "glib.h"
  29 #ifdef HAVE_UNISTD_H
  30 #include        <unistd.h>
  31 #endif
  32 #include        <errno.h>
  33 #include        <sys/types.h>   /* needed for sys/stat.h */
  34 #include        <sys/stat.h>
  35 #ifdef _MSC_VER
  36 #include        <io.h>          /* For _read() */
  37 #endif
  38
  39 /* --- defines --- */
  40 #define to_lower(c)                             ( \
  41         (guchar) (                                                      \
  42           ( (((guchar)(c))>='A' && ((guchar)(c))<='Z') * ('a'-'A') ) |  \
  43           ( (((guchar)(c))>=192 && ((guchar)(c))<=214) * (224-192) ) |  \
  44           ( (((guchar)(c))>=216 && ((guchar)(c))<=222) * (248-216) ) |  \
  45           ((guchar)(c))                                                 \
  46         )                                                               \
  47 )
  48 #define READ_BUFFER_SIZE        (4000)
  49
  50
  51 /* --- typedefs --- */
  52 typedef struct  _GScannerKey    GScannerKey;
  53
  54 struct  _GScannerKey
  55 {
  56   guint          scope_id;
  57   gchar         *symbol;
  58   gpointer       value;
  59 };
  60
  61
  62
  63 /* --- variables --- */
  64 static GScannerConfig g_scanner_config_template =
  65 {
  66   (
  67    " \t\r\n"
  68    )                    /* cset_skip_characters */,
  69   (
  70    G_CSET_a_2_z
  71    "_"
  72    G_CSET_A_2_Z
  73    )                    /* cset_identifier_first */,
  74   (
  75    G_CSET_a_2_z
  76    "_0123456789"
  77    G_CSET_A_2_Z
  78    G_CSET_LATINS
  79    G_CSET_LATINC
  80    )                    /* cset_identifier_nth */,
  81   ( "#\n" )             /* cpair_comment_single */,
  82
  83   FALSE                 /* case_sensitive */,
  84
  85   TRUE                  /* skip_comment_multi */,
  86   TRUE                  /* skip_comment_single */,
  87   TRUE                  /* scan_comment_multi */,
  88   TRUE                  /* scan_identifier */,
  89   FALSE                 /* scan_identifier_1char */,
  90   FALSE                 /* scan_identifier_NULL */,
  91   TRUE                  /* scan_symbols */,
  92   FALSE                 /* scan_binary */,
  93   TRUE                  /* scan_octal */,
  94   TRUE                  /* scan_float */,
  95   TRUE                  /* scan_hex */,
  96   FALSE                 /* scan_hex_dollar */,
  97   TRUE                  /* scan_string_sq */,
  98   TRUE                  /* scan_string_dq */,
  99   TRUE                  /* numbers_2_int */,
 100   FALSE                 /* int_2_float */,
 101   FALSE                 /* identifier_2_string */,
 102   TRUE                  /* char_2_token */,
 103   FALSE                 /* symbol_2_token */,
 104   FALSE                 /* scope_0_fallback */,
 105 };
 106
 107
 108 /* --- prototypes --- */
 109 static inline
 110 GScannerKey*    g_scanner_lookup_internal (GScanner     *scanner,
 111                                            guint         scope_id,
 112                                            const gchar  *symbol);
 113 static gint     g_scanner_key_equal       (gconstpointer v1,
 114                                            gconstpointer v2);
 115 static guint    g_scanner_key_hash        (gconstpointer v);
 116 static void     g_scanner_get_token_ll    (GScanner     *scanner,
 117                                            GTokenType   *token_p,
 118                                            GTokenValue  *value_p,
 119                                            guint        *line_p,
 120                                            guint        *position_p);
 121 static void     g_scanner_get_token_i     (GScanner     *scanner,
 122                                            GTokenType   *token_p,
 123                                            GTokenValue  *value_p,
 124                                            guint        *line_p,
 125                                            guint        *position_p);
 126
 127 static guchar   g_scanner_peek_next_char  (GScanner     *scanner);
 128 static guchar   g_scanner_get_char        (GScanner     *scanner,
 129                                            guint        *line_p,
 130                                            guint        *position_p);
 131 static void     g_scanner_msg_handler     (GScanner     *scanner,
 132                                            gchar        *message,
 133                                            gint          is_error);
 134
 135
 136 /* --- functions --- */
 137 static inline gint
 138 g_scanner_char_2_num (guchar    c,
 139                       guchar    base)
 140 {
 141   if (c >= '0' && c <= '9')
 142     c -= '0';
 143   else if (c >= 'A' && c <= 'Z')
 144     c -= 'A' - 10;
 145   else if (c >= 'a' && c <= 'z')
 146     c -= 'a' - 10;
 147   else
 148     return -1;
 149
 150   if (c < base)
 151     return c;
 152
 153   return -1;
 154 }
 155
 156 GScanner*
 157 g_scanner_new (GScannerConfig   *config_templ)
 158 {
 159   register GScanner     *scanner;
 160
 161   if (!config_templ)
 162     config_templ = &g_scanner_config_template;
 163
 164   scanner = g_new0 (GScanner, 1);
 165
 166   scanner->user_data = NULL;
 167   scanner->max_parse_errors = 0;
 168   scanner->parse_errors = 0;
 169   scanner->input_name = NULL;
 170   scanner->derived_data = NULL;
 171
 172   scanner->config = g_new0 (GScannerConfig, 1);
 173
 174   scanner->config->case_sensitive       = config_templ->case_sensitive;
 175   scanner->config->cset_skip_characters = config_templ->cset_skip_characters;
 176   scanner->config->cset_identifier_first= config_templ->cset_identifier_first;
 177   scanner->config->cset_identifier_nth  = config_templ->cset_identifier_nth;
 178   scanner->config->cpair_comment_single = config_templ->cpair_comment_single;
 179   scanner->config->skip_comment_multi   = config_templ->skip_comment_multi;
 180   scanner->config->skip_comment_single  = config_templ->skip_comment_single;
 181   scanner->config->scan_comment_multi   = config_templ->scan_comment_multi;
 182   scanner->config->scan_identifier      = config_templ->scan_identifier;
 183   scanner->config->scan_identifier_1char= config_templ->scan_identifier_1char;
 184   scanner->config->scan_identifier_NULL = config_templ->scan_identifier_NULL;
 185   scanner->config->scan_symbols         = config_templ->scan_symbols;
 186   scanner->config->scan_binary          = config_templ->scan_binary;
 187   scanner->config->scan_octal           = config_templ->scan_octal;
 188   scanner->config->scan_float           = config_templ->scan_float;
 189   scanner->config->scan_hex             = config_templ->scan_hex;
 190   scanner->config->scan_hex_dollar      = config_templ->scan_hex_dollar;
 191   scanner->config->scan_string_sq       = config_templ->scan_string_sq;
 192   scanner->config->scan_string_dq       = config_templ->scan_string_dq;
 193   scanner->config->numbers_2_int        = config_templ->numbers_2_int;
 194   scanner->config->int_2_float          = config_templ->int_2_float;
 195   scanner->config->identifier_2_string  = config_templ->identifier_2_string;
 196   scanner->config->char_2_token         = config_templ->char_2_token;
 197   scanner->config->symbol_2_token       = config_templ->symbol_2_token;
 198   scanner->config->scope_0_fallback     = config_templ->scope_0_fallback;
 199
 200   scanner->token = G_TOKEN_NONE;
 201   scanner->value.v_int = 0;
 202   scanner->line = 1;
 203   scanner->position = 0;
 204
 205   scanner->next_token = G_TOKEN_NONE;
 206   scanner->next_value.v_int = 0;
 207   scanner->next_line = 1;
 208   scanner->next_position = 0;
 209
 210   scanner->symbol_table = g_hash_table_new (g_scanner_key_hash, g_scanner_key_equal);
 211   scanner->input_fd = -1;
 212   scanner->text = NULL;
 213   scanner->text_end = NULL;
 214   scanner->buffer = NULL;
 215   scanner->scope_id = 0;
 216
 217   scanner->msg_handler = g_scanner_msg_handler;
 218
 219   return scanner;
 220 }
 221
 222 static inline void
 223 g_scanner_free_value (GTokenType     *token_p,
 224                       GTokenValue     *value_p)
 225 {
 226   switch (*token_p)
 227     {
 228     case  G_TOKEN_STRING:
 229     case  G_TOKEN_IDENTIFIER:
 230     case  G_TOKEN_IDENTIFIER_NULL:
 231     case  G_TOKEN_COMMENT_SINGLE:
 232     case  G_TOKEN_COMMENT_MULTI:
 233       g_free (value_p->v_string);
 234       break;
 235
 236     default:
 237       break;
 238     }
 239
 240   *token_p = G_TOKEN_NONE;
 241 }
 242
 243 static void
 244 g_scanner_destroy_symbol_table_entry (gpointer _key,
 245                                       gpointer _value,
 246                                       gpointer _data)
 247 {
 248   GScannerKey *key = _key;
 249
 250   g_free (key->symbol);
 251   g_free (key);
 252 }
 253
 254 void
 255 g_scanner_destroy (GScanner     *scanner)
 256 {
 257   g_return_if_fail (scanner != NULL);
 258
 259   g_hash_table_foreach (scanner->symbol_table,
 260                         g_scanner_destroy_symbol_table_entry, NULL);
 261   g_hash_table_destroy (scanner->symbol_table);
 262   g_scanner_free_value (&scanner->token, &scanner->value);
 263   g_scanner_free_value (&scanner->next_token, &scanner->next_value);
 264   g_free (scanner->config);
 265   g_free (scanner->buffer);
 266   g_free (scanner);
 267 }
 268
 269 static void
 270 g_scanner_msg_handler (GScanner         *scanner,
 271                        gchar            *message,
 272                        gint              is_error)
 273 {
 274   g_return_if_fail (scanner != NULL);
 275
 276   fprintf (stdout, "%s:%d: ", scanner->input_name, scanner->line);
 277   if (is_error)
 278     fprintf (stdout, "error: ");
 279   fprintf (stdout, "%s\n", message);
 280 }
 281
 282 void
 283 g_scanner_error (GScanner       *scanner,
 284                  const gchar    *format,
 285                  ...)
 286 {
 287   g_return_if_fail (scanner != NULL);
 288   g_return_if_fail (format != NULL);
 289
 290   scanner->parse_errors++;
 291
 292   if (scanner->msg_handler)
 293     {
 294       va_list args;
 295       gchar *string;
 296
 297       va_start (args, format);
 298       string = g_strdup_vprintf (format, args);
 299       va_end (args);
 300
 301       scanner->msg_handler (scanner, string, TRUE);
 302
 303       g_free (string);
 304     }
 305 }
 306
 307 void
 308 g_scanner_warn (GScanner       *scanner,
 309                 const gchar    *format,
 310                 ...)
 311 {
 312   g_return_if_fail (scanner != NULL);
 313   g_return_if_fail (format != NULL);
 314
 315   if (scanner->msg_handler)
 316     {
 317       va_list args;
 318       gchar *string;
 319
 320       va_start (args, format);
 321       string = g_strdup_vprintf (format, args);
 322       va_end (args);
 323
 324       scanner->msg_handler (scanner, string, FALSE);
 325
 326       g_free (string);
 327     }
 328 }
 329
 330 static gint
 331 g_scanner_key_equal (gconstpointer v1,
 332                      gconstpointer v2)
 333 {
 334   register const GScannerKey *key1 = v1;
 335   register const GScannerKey *key2 = v2;
 336
 337   return (key1->scope_id == key2->scope_id) && (strcmp (key1->symbol, key2->symbol) == 0);
 338 }
 339
 340 static guint
 341 g_scanner_key_hash (gconstpointer v)
 342 {
 343   register const GScannerKey *key = v;
 344   register gchar *c;
 345   register guint h;
 346
 347   h = key->scope_id;
 348   for (c = key->symbol; *c; c++)
 349     {
 350       register guint g;
 351
 352       h = (h << 4) + *c;
 353       g = h & 0xf0000000;
 354       if (g)
 355         {
 356           h = h ^ (g >> 24);
 357           h = h ^ g;
 358         }
 359     }
 360
 361   return h;
 362 }
 363
 364 static inline GScannerKey*
 365 g_scanner_lookup_internal (GScanner     *scanner,
 366                            guint         scope_id,
 367                            const gchar  *symbol)
 368 {
 369   register GScannerKey  *key_p;
 370   GScannerKey key;
 371
 372   key.scope_id = scope_id;
 373
 374   if (!scanner->config->case_sensitive)
 375     {
 376       register gchar *d;
 377       register const gchar *c;
 378
 379       key.symbol = g_new (gchar, strlen (symbol) + 1);
 380       for (d = key.symbol, c = symbol; *c; c++, d++)
 381         *d = to_lower (*c);
 382       *d = 0;
 383       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 384       g_free (key.symbol);
 385     }
 386   else
 387     {
 388       key.symbol = (gchar*) symbol;
 389       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 390     }
 391
 392   return key_p;
 393 }
 394
 395 void
 396 g_scanner_scope_add_symbol (GScanner    *scanner,
 397                             guint        scope_id,
 398                             const gchar *symbol,
 399                             gpointer     value)
 400 {
 401   register GScannerKey  *key;
 402
 403   g_return_if_fail (scanner != NULL);
 404   g_return_if_fail (symbol != NULL);
 405
 406   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 407
 408   if (!key)
 409     {
 410       key = g_new (GScannerKey, 1);
 411       key->scope_id = scope_id;
 412       key->symbol = g_strdup (symbol);
 413       key->value = value;
 414       if (!scanner->config->case_sensitive)
 415         {
 416           register gchar *c;
 417
 418           c = key->symbol;
 419           while (*c != 0)
 420             {
 421               *c = to_lower (*c);
 422               c++;
 423             }
 424         }
 425       g_hash_table_insert (scanner->symbol_table, key, key);
 426     }
 427   else
 428     key->value = value;
 429 }
 430
 431 void
 432 g_scanner_scope_remove_symbol (GScanner    *scanner,
 433                                guint        scope_id,
 434                                const gchar *symbol)
 435 {
 436   register GScannerKey  *key;
 437
 438   g_return_if_fail (scanner != NULL);
 439   g_return_if_fail (symbol != NULL);
 440
 441   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 442
 443   if (key)
 444     {
 445       g_hash_table_remove (scanner->symbol_table, key);
 446       g_free (key->symbol);
 447       g_free (key);
 448     }
 449 }
 450
 451 gpointer
 452 g_scanner_lookup_symbol (GScanner       *scanner,
 453                          const gchar    *symbol)
 454 {
 455   register GScannerKey  *key;
 456   register guint scope_id;
 457
 458   g_return_val_if_fail (scanner != NULL, NULL);
 459
 460   if (!symbol)
 461     return NULL;
 462
 463   scope_id = scanner->scope_id;
 464   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 465   if (!key && scope_id && scanner->config->scope_0_fallback)
 466     key = g_scanner_lookup_internal (scanner, 0, symbol);
 467
 468   if (key)
 469     return key->value;
 470   else
 471     return NULL;
 472 }
 473
 474 gpointer
 475 g_scanner_scope_lookup_symbol (GScanner       *scanner,
 476                                guint           scope_id,
 477                                const gchar    *symbol)
 478 {
 479   register GScannerKey  *key;
 480
 481   g_return_val_if_fail (scanner != NULL, NULL);
 482
 483   if (!symbol)
 484     return NULL;
 485
 486   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 487
 488   if (key)
 489     return key->value;
 490   else
 491     return NULL;
 492 }
 493
 494 guint
 495 g_scanner_set_scope (GScanner       *scanner,
 496                      guint           scope_id)
 497 {
 498   register guint old_scope_id;
 499
 500   g_return_val_if_fail (scanner != NULL, 0);
 501
 502   old_scope_id = scanner->scope_id;
 503   scanner->scope_id = scope_id;
 504
 505   return old_scope_id;
 506 }
 507
 508 static void
 509 g_scanner_foreach_internal (gpointer  _key,
 510                             gpointer  _value,
 511                             gpointer  _user_data)
 512 {
 513   register GScannerKey *key;
 514   register gpointer *d;
 515   register GHFunc func;
 516   register gpointer func_data;
 517   register guint *scope_id;
 518
 519   d = _user_data;
 520   func = (GHFunc) d[0];
 521   func_data = d[1];
 522   scope_id = d[2];
 523   key = _value;
 524
 525   if (key->scope_id == *scope_id)
 526     func (key->symbol, key->value, func_data);
 527 }
 528
 529 void
 530 g_scanner_scope_foreach_symbol (GScanner       *scanner,
 531                                 guint           scope_id,
 532                                 GHFunc          func,
 533                                 gpointer        func_data)
 534 {
 535   gpointer d[3];
 536
 537   g_return_if_fail (scanner != NULL);
 538
 539   d[0] = (gpointer) func;
 540   d[1] = func_data;
 541   d[2] = &scope_id;
 542
 543   g_hash_table_foreach (scanner->symbol_table, g_scanner_foreach_internal, d);
 544 }
 545
 546 void
 547 g_scanner_freeze_symbol_table (GScanner *scanner)
 548 {
 549   g_return_if_fail (scanner != NULL);
 550
 551   g_hash_table_freeze (scanner->symbol_table);
 552 }
 553
 554 void
 555 g_scanner_thaw_symbol_table (GScanner *scanner)
 556 {
 557   g_return_if_fail (scanner != NULL);
 558
 559   g_hash_table_thaw (scanner->symbol_table);
 560 }
 561
 562 GTokenType
 563 g_scanner_peek_next_token (GScanner     *scanner)
 564 {
 565   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 566
 567   if (scanner->next_token == G_TOKEN_NONE)
 568     {
 569       scanner->next_line = scanner->line;
 570       scanner->next_position = scanner->position;
 571       g_scanner_get_token_i (scanner,
 572                              &scanner->next_token,
 573                              &scanner->next_value,
 574                              &scanner->next_line,
 575                              &scanner->next_position);
 576     }
 577
 578   return scanner->next_token;
 579 }
 580
 581 GTokenType
 582 g_scanner_get_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       g_scanner_free_value (&scanner->token, &scanner->value);
 589
 590       scanner->token = scanner->next_token;
 591       scanner->value = scanner->next_value;
 592       scanner->line = scanner->next_line;
 593       scanner->position = scanner->next_position;
 594       scanner->next_token = G_TOKEN_NONE;
 595     }
 596   else
 597     g_scanner_get_token_i (scanner,
 598                            &scanner->token,
 599                            &scanner->value,
 600                            &scanner->line,
 601                            &scanner->position);
 602
 603   return scanner->token;
 604 }
 605
 606 GTokenType
 607 g_scanner_cur_token (GScanner *scanner)
 608 {
 609   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 610
 611   return scanner->token;
 612 }
 613
 614 GTokenValue
 615 g_scanner_cur_value (GScanner *scanner)
 616 {
 617   GTokenValue v;
 618
 619   g_return_val_if_fail (scanner != NULL, v);
 620
 621   /* MSC isn't capable of handling return scanner->value; ? */
 622
 623   v.v_int = 0;
 624
 625   v = scanner->value;
 626
 627   return v;
 628 }
 629
 630 guint
 631 g_scanner_cur_line (GScanner *scanner)
 632 {
 633   g_return_val_if_fail (scanner != NULL, 0);
 634
 635   return scanner->line;
 636 }
 637
 638 guint
 639 g_scanner_cur_position (GScanner *scanner)
 640 {
 641   g_return_val_if_fail (scanner != NULL, 0);
 642
 643   return scanner->position;
 644 }
 645
 646 gboolean
 647 g_scanner_eof (GScanner *scanner)
 648 {
 649   g_return_val_if_fail (scanner != NULL, TRUE);
 650
 651   return scanner->token == G_TOKEN_EOF;
 652 }
 653
 654 void
 655 g_scanner_input_file (GScanner *scanner,
 656                       gint      input_fd)
 657 {
 658   g_return_if_fail (scanner != NULL);
 659   g_return_if_fail (input_fd >= 0);
 660
 661   scanner->token = G_TOKEN_NONE;
 662   scanner->value.v_int = 0;
 663   scanner->line = 1;
 664   scanner->position = 0;
 665   scanner->next_token = G_TOKEN_NONE;
 666
 667   scanner->input_fd = input_fd;
 668   scanner->text = NULL;
 669   scanner->text_end = NULL;
 670
 671   if (!scanner->buffer)
 672     scanner->buffer = g_new (gchar, READ_BUFFER_SIZE + 1);
 673 }
 674
 675 void
 676 g_scanner_input_text (GScanner    *scanner,
 677                       const gchar *text,
 678                       guint        text_len)
 679 {
 680   g_return_if_fail (scanner != NULL);
 681   if (text_len)
 682     g_return_if_fail (text != NULL);
 683   else
 684     text = NULL;
 685
 686   scanner->token = G_TOKEN_NONE;
 687   scanner->value.v_int = 0;
 688   scanner->line = 1;
 689   scanner->position = 0;
 690   scanner->next_token = G_TOKEN_NONE;
 691
 692   scanner->input_fd = -1;
 693   scanner->text = text;
 694   scanner->text_end = text + text_len;
 695
 696   if (scanner->buffer)
 697     {
 698       g_free (scanner->buffer);
 699       scanner->buffer = NULL;
 700     }
 701 }
 702
 703 static guchar
 704 g_scanner_peek_next_char (GScanner *scanner)
 705 {
 706   if (scanner->text < scanner->text_end)
 707     {
 708       return *scanner->text;
 709     }
 710   else if (scanner->input_fd >= 0)
 711     {
 712       register gint count;
 713       register gchar *buffer;
 714
 715       buffer = scanner->buffer;
 716       do
 717         {
 718           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 719         }
 720       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 721
 722       if (count < 1)
 723         {
 724           scanner->input_fd = -1;
 725
 726           return 0;
 727         }
 728       else
 729         {
 730           scanner->text = buffer;
 731           scanner->text_end = buffer + count;
 732
 733           return *buffer;
 734         }
 735     }
 736   else
 737     return 0;
 738 }
 739
 740 static guchar
 741 g_scanner_get_char (GScanner    *scanner,
 742                     guint       *line_p,
 743                     guint       *position_p)
 744 {
 745   register guchar fchar;
 746
 747   if (scanner->text < scanner->text_end)
 748     fchar = *(scanner->text++);
 749   else if (scanner->input_fd >= 0)
 750     {
 751       register gint count;
 752       register gchar *buffer;
 753
 754       buffer = scanner->buffer;
 755       do
 756         {
 757           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 758         }
 759       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 760
 761       if (count < 1)
 762         {
 763           scanner->input_fd = -1;
 764           fchar = 0;
 765         }
 766       else
 767         {
 768           scanner->text = buffer + 1;
 769           scanner->text_end = buffer + count;
 770           fchar = *buffer;
 771         }
 772     }
 773   else
 774     fchar = 0;
 775
 776   if (fchar == '\n')
 777     {
 778       (*position_p) = 0;
 779       (*line_p)++;
 780     }
 781   else if (fchar)
 782     {
 783       (*position_p)++;
 784     }
 785
 786   return fchar;
 787 }
 788
 789 void
 790 g_scanner_unexp_token (GScanner         *scanner,
 791                        GTokenType        expected_token,
 792                        const gchar      *identifier_spec,
 793                        const gchar      *symbol_spec,
 794                        const gchar      *symbol_name,
 795                        const gchar      *message,
 796                        gint              is_error)
 797 {
 798   register gchar        *token_string;
 799   register guint        token_string_len;
 800   register gchar        *expected_string;
 801   register guint        expected_string_len;
 802   register gchar        *message_prefix;
 803   register gboolean     print_unexp;
 804   void (*msg_handler)   (GScanner*, const gchar*, ...);
 805
 806   g_return_if_fail (scanner != NULL);
 807
 808   if (is_error)
 809     msg_handler = g_scanner_error;
 810   else
 811     msg_handler = g_scanner_warn;
 812
 813   if (!identifier_spec)
 814     identifier_spec = "identifier";
 815   if (!symbol_spec)
 816     symbol_spec = "symbol";
 817
 818   token_string_len = 56;
 819   token_string = g_new (gchar, token_string_len + 1);
 820   expected_string_len = 64;
 821   expected_string = g_new (gchar, expected_string_len + 1);
 822   print_unexp = TRUE;
 823
 824   switch (scanner->token)
 825     {
 826     case  G_TOKEN_EOF:
 827       g_snprintf (token_string, token_string_len, "end of file");
 828       break;
 829
 830     default:
 831       if (scanner->token >= 1 && scanner->token <= 255)
 832         {
 833           if ((scanner->token >= ' ' && scanner->token <= '~') ||
 834               strchr (scanner->config->cset_identifier_first, scanner->token) ||
 835               strchr (scanner->config->cset_identifier_nth, scanner->token))
 836             g_snprintf (token_string, expected_string_len, "character `%c'", scanner->token);
 837           else
 838             g_snprintf (token_string, expected_string_len, "character `\\%o'", scanner->token);
 839           break;
 840         }
 841       else if (!scanner->config->symbol_2_token)
 842         {
 843           g_snprintf (token_string, token_string_len, "(unknown) token <%d>", scanner->token);
 844           break;
 845         }
 846       /* fall through */
 847     case  G_TOKEN_SYMBOL:
 848       if (expected_token == G_TOKEN_SYMBOL ||
 849           (scanner->config->symbol_2_token &&
 850            expected_token > G_TOKEN_LAST))
 851         print_unexp = FALSE;
 852       if (symbol_name)
 853         g_snprintf (token_string,
 854                     token_string_len,
 855                     "%s%s `%s'",
 856                     print_unexp ? "" : "invalid ",
 857                     symbol_spec,
 858                     symbol_name);
 859       else
 860         g_snprintf (token_string,
 861                     token_string_len,
 862                     "%s%s",
 863                     print_unexp ? "" : "invalid ",
 864                     symbol_spec);
 865       break;
 866
 867     case  G_TOKEN_ERROR:
 868       print_unexp = FALSE;
 869       expected_token = G_TOKEN_NONE;
 870       switch (scanner->value.v_error)
 871         {
 872         case  G_ERR_UNEXP_EOF:
 873           g_snprintf (token_string, token_string_len, "scanner: unexpected end of file");
 874           break;
 875
 876         case  G_ERR_UNEXP_EOF_IN_STRING:
 877           g_snprintf (token_string, token_string_len, "scanner: unterminated string constant");
 878           break;
 879
 880         case  G_ERR_UNEXP_EOF_IN_COMMENT:
 881           g_snprintf (token_string, token_string_len, "scanner: unterminated comment");
 882           break;
 883
 884         case  G_ERR_NON_DIGIT_IN_CONST:
 885           g_snprintf (token_string, token_string_len, "scanner: non digit in constant");
 886           break;
 887
 888         case  G_ERR_FLOAT_RADIX:
 889           g_snprintf (token_string, token_string_len, "scanner: invalid radix for floating constant");
 890           break;
 891
 892         case  G_ERR_FLOAT_MALFORMED:
 893           g_snprintf (token_string, token_string_len, "scanner: malformed floating constant");
 894           break;
 895
 896         case  G_ERR_DIGIT_RADIX:
 897           g_snprintf (token_string, token_string_len, "scanner: digit is beyond radix");
 898           break;
 899
 900         case  G_ERR_UNKNOWN:
 901         default:
 902           g_snprintf (token_string, token_string_len, "scanner: unknown error");
 903           break;
 904         }
 905       break;
 906
 907     case  G_TOKEN_CHAR:
 908       g_snprintf (token_string, token_string_len, "character `%c'", scanner->value.v_char);
 909       break;
 910
 911     case  G_TOKEN_IDENTIFIER:
 912       if (expected_token == G_TOKEN_IDENTIFIER)
 913         print_unexp = FALSE;
 914       g_snprintf (token_string,
 915                   token_string_len,
 916                   "%s%s `%s'",
 917                   print_unexp ? "" : "invalid ",
 918                   identifier_spec,
 919                   scanner->value.v_string);
 920       break;
 921
 922     case  G_TOKEN_BINARY:
 923     case  G_TOKEN_OCTAL:
 924     case  G_TOKEN_INT:
 925     case  G_TOKEN_HEX:
 926       g_snprintf (token_string, token_string_len, "number `%ld'", scanner->value.v_int);
 927       break;
 928
 929     case  G_TOKEN_FLOAT:
 930       g_snprintf (token_string, token_string_len, "number `%.3f'", scanner->value.v_float);
 931       break;
 932
 933     case  G_TOKEN_STRING:
 934       if (expected_token == G_TOKEN_STRING)
 935         print_unexp = FALSE;
 936       g_snprintf (token_string,
 937                   token_string_len,
 938                   "%s%sstring constant \"%s\"",
 939                   print_unexp ? "" : "invalid ",
 940                   scanner->value.v_string[0] == 0 ? "empty " : "",
 941                   scanner->value.v_string);
 942       token_string[token_string_len - 2] = '"';
 943       token_string[token_string_len - 1] = 0;
 944       break;
 945
 946     case  G_TOKEN_COMMENT_SINGLE:
 947     case  G_TOKEN_COMMENT_MULTI:
 948       g_snprintf (token_string, token_string_len, "comment");
 949       break;
 950
 951     case  G_TOKEN_NONE:
 952       /* somehow the user's parsing code is screwed, there isn't much
 953        * we can do about it.
 954        * Note, a common case to trigger this is
 955        * g_scanner_peek_next_token(); g_scanner_unexp_token();
 956        * without an intermediate g_scanner_get_next_token().
 957        */
 958       g_assert_not_reached ();
 959       break;
 960     }
 961
 962
 963   switch (expected_token)
 964     {
 965       gboolean need_valid;
 966
 967     default:
 968       if (expected_token >= 1 && expected_token <= 255)
 969         {
 970           if ((expected_token >= ' ' && expected_token <= '~') ||
 971               strchr (scanner->config->cset_identifier_first, expected_token) ||
 972               strchr (scanner->config->cset_identifier_nth, expected_token))
 973             g_snprintf (expected_string, expected_string_len, "character `%c'", expected_token);
 974           else
 975             g_snprintf (expected_string, expected_string_len, "character `\\%o'", expected_token);
 976           break;
 977         }
 978       else if (!scanner->config->symbol_2_token)
 979         {
 980           g_snprintf (expected_string, expected_string_len, "(unknown) token <%d>", expected_token);
 981           break;
 982         }
 983       /* fall through */
 984     case  G_TOKEN_SYMBOL:
 985       need_valid = (scanner->token == G_TOKEN_SYMBOL ||
 986                     (scanner->config->symbol_2_token &&
 987                      scanner->token > G_TOKEN_LAST));
 988       g_snprintf (expected_string,
 989                   expected_string_len,
 990                   "%s%s",
 991                   need_valid ? "valid " : "",
 992                   symbol_spec);
 993       /* FIXME: should we attempt to lookup the symbol_name for symbol_2_token? */
 994       break;
 995
 996     case  G_TOKEN_INT:
 997       g_snprintf (expected_string, expected_string_len, "number (integer)");
 998       break;
 999
1000     case  G_TOKEN_FLOAT:
1001       g_snprintf (expected_string, expected_string_len, "number (float)");
1002       break;
1003
1004     case  G_TOKEN_STRING:
1005       g_snprintf (expected_string,
1006                   expected_string_len,
1007                   "%sstring constant",
1008                   scanner->token == G_TOKEN_STRING ? "valid " : "");
1009       break;
1010
1011     case  G_TOKEN_IDENTIFIER:
1012       g_snprintf (expected_string,
1013                   expected_string_len,
1014                   "%s%s",
1015                   scanner->token == G_TOKEN_IDENTIFIER ? "valid " : "",
1016                   identifier_spec);
1017       break;
1018
1019     case  G_TOKEN_NONE:
1020       break;
1021     }
1022
1023   if (message && message[0] != 0)
1024     message_prefix = " - ";
1025   else
1026     {
1027       message_prefix = "";
1028       message = "";
1029     }
1030
1031   if (expected_token != G_TOKEN_NONE)
1032     {
1033       if (print_unexp)
1034         msg_handler (scanner,
1035                      "unexpected %s, expected %s%s%s",
1036                      token_string,
1037                      expected_string,
1038                      message_prefix,
1039                      message);
1040       else
1041         msg_handler (scanner,
1042                      "%s, expected %s%s%s",
1043                      token_string,
1044                      expected_string,
1045                      message_prefix,
1046                      message);
1047     }
1048   else
1049     {
1050       if (print_unexp)
1051         msg_handler (scanner,
1052                      "unexpected %s%s%s",
1053                      token_string,
1054                      message_prefix,
1055                      message);
1056       else
1057         msg_handler (scanner,
1058                      "%s%s%s",
1059                      token_string,
1060                      message_prefix,
1061                      message);
1062     }
1063
1064   g_free (token_string);
1065   g_free (expected_string);
1066 }
1067
1068 gint
1069 g_scanner_stat_mode (const gchar *filename)
1070 {
1071   struct stat  *stat_buf;
1072   gint          st_mode;
1073
1074   stat_buf = g_new0 (struct stat, 1);
1075 #ifdef HAVE_LSTAT
1076   lstat (filename, stat_buf);
1077 #else
1078   stat (filename, stat_buf);
1079 #endif
1080   st_mode = stat_buf->st_mode;
1081
1082   g_free (stat_buf);
1083
1084   return st_mode;
1085 }
1086
1087 static void
1088 g_scanner_get_token_i (GScanner *scanner,
1089                        GTokenType       *token_p,
1090                        GTokenValue      *value_p,
1091                        guint            *line_p,
1092                        guint            *position_p)
1093 {
1094   do
1095     {
1096       g_scanner_free_value (token_p, value_p);
1097       g_scanner_get_token_ll (scanner, token_p, value_p, line_p, position_p);
1098     }
1099   while (((*token_p > 0 && *token_p < 256) &&
1100           strchr (scanner->config->cset_skip_characters, *token_p)) ||
1101          (*token_p == G_TOKEN_CHAR &&
1102           strchr (scanner->config->cset_skip_characters, value_p->v_char)) ||
1103          (*token_p == G_TOKEN_COMMENT_MULTI &&
1104           scanner->config->skip_comment_multi) ||
1105          (*token_p == G_TOKEN_COMMENT_SINGLE &&
1106           scanner->config->skip_comment_single));
1107
1108   switch (*token_p)
1109     {
1110     case G_TOKEN_IDENTIFIER:
1111       if (scanner->config->identifier_2_string)
1112         *token_p = G_TOKEN_STRING;
1113       break;
1114
1115     case G_TOKEN_SYMBOL:
1116       if (scanner->config->symbol_2_token)
1117         *token_p = (GTokenType) value_p->v_symbol;
1118       break;
1119
1120     case G_TOKEN_BINARY:
1121     case G_TOKEN_OCTAL:
1122     case G_TOKEN_HEX:
1123       if (scanner->config->numbers_2_int)
1124         *token_p = G_TOKEN_INT;
1125       break;
1126
1127     default:
1128       break;
1129     }
1130
1131   if (*token_p == G_TOKEN_INT &&
1132       scanner->config->int_2_float)
1133     {
1134       *token_p = G_TOKEN_FLOAT;
1135       value_p->v_float = value_p->v_int;
1136     }
1137
1138   errno = 0;
1139 }
1140
1141 static void
1142 g_scanner_get_token_ll  (GScanner       *scanner,
1143                          GTokenType     *token_p,
1144                          GTokenValue    *value_p,
1145                          guint          *line_p,
1146                          guint          *position_p)
1147 {
1148   register GScannerConfig *config;
1149   register GTokenType      token;
1150   register gboolean        in_comment_multi;
1151   register gboolean        in_comment_single;
1152   register gboolean        in_string_sq;
1153   register gboolean        in_string_dq;
1154   register GString        *gstring;
1155   GTokenValue              value;
1156   guchar                   ch;
1157
1158   config = scanner->config;
1159   (*value_p).v_int = 0;
1160
1161   if ((scanner->text >= scanner->text_end && scanner->input_fd < 0) ||
1162       scanner->token == G_TOKEN_EOF)
1163     {
1164       *token_p = G_TOKEN_EOF;
1165       return;
1166     }
1167
1168   in_comment_multi = FALSE;
1169   in_comment_single = FALSE;
1170   in_string_sq = FALSE;
1171   in_string_dq = FALSE;
1172   gstring = NULL;
1173
1174   do /* while (ch != 0) */
1175     {
1176       register gboolean         dotted_float = FALSE;
1177
1178       ch = g_scanner_get_char (scanner, line_p, position_p);
1179
1180       value.v_int = 0;
1181       token = G_TOKEN_NONE;
1182
1183       /* this is *evil*, but needed ;(
1184        * we first check for identifier first character, because  it
1185        * might interfere with other key chars like slashes or numbers
1186        */
1187       if (config->scan_identifier &&
1188           ch && strchr (config->cset_identifier_first, ch))
1189         goto identifier_precedence;
1190
1191       switch (ch)
1192         {
1193         case  0:
1194           token = G_TOKEN_EOF;
1195           (*position_p)++;
1196           /* ch = 0; */
1197           break;
1198
1199         case  '/':
1200           if (!config->scan_comment_multi ||
1201               g_scanner_peek_next_char (scanner) != '*')
1202             goto default_case;
1203           g_scanner_get_char (scanner, line_p, position_p);
1204           token = G_TOKEN_COMMENT_MULTI;
1205           in_comment_multi = TRUE;
1206           gstring = g_string_new ("");
1207           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1208             {
1209               if (ch == '*' && g_scanner_peek_next_char (scanner) == '/')
1210                 {
1211                   g_scanner_get_char (scanner, line_p, position_p);
1212                   in_comment_multi = FALSE;
1213                   break;
1214                 }
1215               else
1216                 gstring = g_string_append_c (gstring, ch);
1217             }
1218           ch = 0;
1219           break;
1220
1221         case  '\'':
1222           if (!config->scan_string_sq)
1223             goto default_case;
1224           token = G_TOKEN_STRING;
1225           in_string_sq = TRUE;
1226           gstring = g_string_new ("");
1227           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1228             {
1229               if (ch == '\'')
1230                 {
1231                   in_string_sq = FALSE;
1232                   break;
1233                 }
1234               else
1235                 gstring = g_string_append_c (gstring, ch);
1236             }
1237           ch = 0;
1238           break;
1239
1240         case  '"':
1241           if (!config->scan_string_dq)
1242             goto default_case;
1243           token = G_TOKEN_STRING;
1244           in_string_dq = TRUE;
1245           gstring = g_string_new ("");
1246           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1247             {
1248               if (ch == '"')
1249                 {
1250                   in_string_dq = FALSE;
1251                   break;
1252                 }
1253               else
1254                 {
1255                   if (ch == '\\')
1256                     {
1257                       ch = g_scanner_get_char (scanner, line_p, position_p);
1258                       switch (ch)
1259                         {
1260                           register guint        i;
1261                           register guint        fchar;
1262
1263                         case  0:
1264                           break;
1265
1266                         case  '\\':
1267                           gstring = g_string_append_c (gstring, '\\');
1268                           break;
1269
1270                         case  'n':
1271                           gstring = g_string_append_c (gstring, '\n');
1272                           break;
1273
1274                         case  't':
1275                           gstring = g_string_append_c (gstring, '\t');
1276                           break;
1277
1278                         case  'r':
1279                           gstring = g_string_append_c (gstring, '\r');
1280                           break;
1281
1282                         case  'b':
1283                           gstring = g_string_append_c (gstring, '\b');
1284                           break;
1285
1286                         case  'f':
1287                           gstring = g_string_append_c (gstring, '\f');
1288                           break;
1289
1290                         case  '0':
1291                         case  '1':
1292                         case  '2':
1293                         case  '3':
1294                         case  '4':
1295                         case  '5':
1296                         case  '6':
1297                         case  '7':
1298                           i = ch - '0';
1299                           fchar = g_scanner_peek_next_char (scanner);
1300                           if (fchar >= '0' && fchar <= '7')
1301                             {
1302                               ch = g_scanner_get_char (scanner, line_p, position_p);
1303                               i = i * 8 + ch - '0';
1304                               fchar = g_scanner_peek_next_char (scanner);
1305                               if (fchar >= '0' && fchar <= '7')
1306                                 {
1307                                   ch = g_scanner_get_char (scanner, line_p, position_p);
1308                                   i = i * 8 + ch - '0';
1309                                 }
1310                             }
1311                           gstring = g_string_append_c (gstring, i);
1312                           break;
1313
1314                         default:
1315                           gstring = g_string_append_c (gstring, ch);
1316                           break;
1317                         }
1318                     }
1319                   else
1320                     gstring = g_string_append_c (gstring, ch);
1321                 }
1322             }
1323           ch = 0;
1324           break;
1325
1326         case  '.':
1327           if (!config->scan_float)
1328             goto default_case;
1329           token = G_TOKEN_FLOAT;
1330           dotted_float = TRUE;
1331           ch = g_scanner_get_char (scanner, line_p, position_p);
1332           goto number_parsing;
1333
1334         case  '$':
1335           if (!config->scan_hex_dollar)
1336             goto default_case;
1337           token = G_TOKEN_HEX;
1338           ch = g_scanner_get_char (scanner, line_p, position_p);
1339           goto number_parsing;
1340
1341         case  '0':
1342           if (config->scan_octal)
1343             token = G_TOKEN_OCTAL;
1344           else
1345             token = G_TOKEN_INT;
1346           ch = g_scanner_peek_next_char (scanner);
1347           if (config->scan_hex && (ch == 'x' || ch == 'X'))
1348             {
1349               token = G_TOKEN_HEX;
1350               g_scanner_get_char (scanner, line_p, position_p);
1351               ch = g_scanner_get_char (scanner, line_p, position_p);
1352               if (ch == 0)
1353                 {
1354                   token = G_TOKEN_ERROR;
1355                   value.v_error = G_ERR_UNEXP_EOF;
1356                   (*position_p)++;
1357                   break;
1358                 }
1359               if (g_scanner_char_2_num (ch, 16) < 0)
1360                 {
1361                   token = G_TOKEN_ERROR;
1362                   value.v_error = G_ERR_DIGIT_RADIX;
1363                   ch = 0;
1364                   break;
1365                 }
1366             }
1367           else if (config->scan_binary && (ch == 'b' || ch == 'B'))
1368             {
1369               token = G_TOKEN_BINARY;
1370               g_scanner_get_char (scanner, line_p, position_p);
1371               ch = g_scanner_get_char (scanner, line_p, position_p);
1372               if (ch == 0)
1373                 {
1374                   token = G_TOKEN_ERROR;
1375                   value.v_error = G_ERR_UNEXP_EOF;
1376                   (*position_p)++;
1377                   break;
1378                 }
1379               if (g_scanner_char_2_num (ch, 10) < 0)
1380                 {
1381                   token = G_TOKEN_ERROR;
1382                   value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1383                   ch = 0;
1384                   break;
1385                 }
1386             }
1387           else
1388             ch = '0';
1389           /* fall through */
1390         case  '1':
1391         case  '2':
1392         case  '3':
1393         case  '4':
1394         case  '5':
1395         case  '6':
1396         case  '7':
1397         case  '8':
1398         case  '9':
1399         number_parsing:
1400         {
1401           register gboolean in_number = TRUE;
1402           gchar *endptr;
1403
1404           if (token == G_TOKEN_NONE)
1405             token = G_TOKEN_INT;
1406
1407           gstring = g_string_new (dotted_float ? "0." : "");
1408           gstring = g_string_append_c (gstring, ch);
1409
1410           do /* while (in_number) */
1411             {
1412               register gboolean is_E;
1413
1414               is_E = token == G_TOKEN_FLOAT && (ch == 'e' || ch == 'E');
1415
1416               ch = g_scanner_peek_next_char (scanner);
1417
1418               if (g_scanner_char_2_num (ch, 36) >= 0 ||
1419                   (config->scan_float && ch == '.') ||
1420                   (is_E && (ch == '+' || ch == '-')))
1421                 {
1422                   ch = g_scanner_get_char (scanner, line_p, position_p);
1423
1424                   switch (ch)
1425                     {
1426                     case '.':
1427                       if (token != G_TOKEN_INT && token != G_TOKEN_OCTAL)
1428                         {
1429                           value.v_error = token == G_TOKEN_FLOAT ? G_ERR_FLOAT_MALFORMED : G_ERR_FLOAT_RADIX;
1430                           token = G_TOKEN_ERROR;
1431                           in_number = FALSE;
1432                         }
1433                       else
1434                         {
1435                           token = G_TOKEN_FLOAT;
1436                           gstring = g_string_append_c (gstring, ch);
1437                         }
1438                       break;
1439
1440                     case '0':
1441                     case '1':
1442                     case '2':
1443                     case '3':
1444                     case '4':
1445                     case '5':
1446                     case '6':
1447                     case '7':
1448                     case '8':
1449                     case '9':
1450                       gstring = g_string_append_c (gstring, ch);
1451                       break;
1452
1453                     case '-':
1454                     case '+':
1455                       if (token != G_TOKEN_FLOAT)
1456                         {
1457                           token = G_TOKEN_ERROR;
1458                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1459                           in_number = FALSE;
1460                         }
1461                       else
1462                         gstring = g_string_append_c (gstring, ch);
1463                       break;
1464
1465                     case 'e':
1466                     case 'E':
1467                       if ((token != G_TOKEN_HEX && !config->scan_float) ||
1468                           (token != G_TOKEN_HEX &&
1469                            token != G_TOKEN_OCTAL &&
1470                            token != G_TOKEN_FLOAT &&
1471                            token != G_TOKEN_INT))
1472                         {
1473                           token = G_TOKEN_ERROR;
1474                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1475                           in_number = FALSE;
1476                         }
1477                       else
1478                         {
1479                           if (token != G_TOKEN_HEX)
1480                             token = G_TOKEN_FLOAT;
1481                           gstring = g_string_append_c (gstring, ch);
1482                         }
1483                       break;
1484
1485                     default:
1486                       if (token != G_TOKEN_HEX)
1487                         {
1488                           token = G_TOKEN_ERROR;
1489                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1490                           in_number = FALSE;
1491                         }
1492                       else
1493                         gstring = g_string_append_c (gstring, ch);
1494                       break;
1495                     }
1496                 }
1497               else
1498                 in_number = FALSE;
1499             }
1500           while (in_number);
1501
1502           endptr = NULL;
1503           switch (token)
1504             {
1505             case G_TOKEN_BINARY:
1506               value.v_binary = strtol (gstring->str, &endptr, 2);
1507               break;
1508
1509             case G_TOKEN_OCTAL:
1510               value.v_octal = strtol (gstring->str, &endptr, 8);
1511               break;
1512
1513             case G_TOKEN_INT:
1514               value.v_int = strtol (gstring->str, &endptr, 10);
1515               break;
1516
1517             case G_TOKEN_FLOAT:
1518               value.v_float = g_strtod (gstring->str, &endptr);
1519               break;
1520
1521             case G_TOKEN_HEX:
1522               value.v_hex = strtol (gstring->str, &endptr, 16);
1523               break;
1524
1525             default:
1526               break;
1527             }
1528           if (endptr && *endptr)
1529             {
1530               token = G_TOKEN_ERROR;
1531               if (*endptr == 'e' || *endptr == 'E')
1532                 value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1533               else
1534                 value.v_error = G_ERR_DIGIT_RADIX;
1535             }
1536           g_string_free (gstring, TRUE);
1537           gstring = NULL;
1538           ch = 0;
1539         } /* number_parsing:... */
1540         break;
1541
1542         default:
1543         default_case:
1544         {
1545           if (config->cpair_comment_single &&
1546               ch == config->cpair_comment_single[0])
1547             {
1548               token = G_TOKEN_COMMENT_SINGLE;
1549               in_comment_single = TRUE;
1550               gstring = g_string_new ("");
1551               while ((ch = g_scanner_get_char (scanner,
1552                                                line_p,
1553                                                position_p)) != 0)
1554                 {
1555                   if (ch == config->cpair_comment_single[1])
1556                     {
1557                       in_comment_single = FALSE;
1558                       ch = 0;
1559                       break;
1560                     }
1561
1562                   gstring = g_string_append_c (gstring, ch);
1563                   ch = 0;
1564                 }
1565             }
1566           else if (config->scan_identifier && ch &&
1567                    strchr (config->cset_identifier_first, ch))
1568             {
1569             identifier_precedence:
1570
1571               if (config->cset_identifier_nth && ch &&
1572                   strchr (config->cset_identifier_nth,
1573                           g_scanner_peek_next_char (scanner)))
1574                 {
1575                   token = G_TOKEN_IDENTIFIER;
1576                   gstring = g_string_new ("");
1577                   gstring = g_string_append_c (gstring, ch);
1578                   do
1579                     {
1580                       ch = g_scanner_get_char (scanner, line_p, position_p);
1581                       gstring = g_string_append_c (gstring, ch);
1582                       ch = g_scanner_peek_next_char (scanner);
1583                     }
1584                   while (ch && strchr (config->cset_identifier_nth, ch));
1585                   ch = 0;
1586                 }
1587               else if (config->scan_identifier_1char)
1588                 {
1589                   token = G_TOKEN_IDENTIFIER;
1590                   value.v_identifier = g_new0 (gchar, 2);
1591                   value.v_identifier[0] = ch;
1592                   ch = 0;
1593                 }
1594             }
1595           if (ch)
1596             {
1597               if (config->char_2_token)
1598                 token = ch;
1599               else
1600                 {
1601                   token = G_TOKEN_CHAR;
1602                   value.v_char = ch;
1603                 }
1604               ch = 0;
1605             }
1606         } /* default_case:... */
1607         break;
1608         }
1609       g_assert (ch == 0 && token != G_TOKEN_NONE); /* paranoid */
1610     }
1611   while (ch != 0);
1612
1613   if (in_comment_multi || in_comment_single ||
1614       in_string_sq || in_string_dq)
1615     {
1616       token = G_TOKEN_ERROR;
1617       if (gstring)
1618         {
1619           g_string_free (gstring, TRUE);
1620           gstring = NULL;
1621         }
1622       (*position_p)++;
1623       if (in_comment_multi || in_comment_single)
1624         value.v_error = G_ERR_UNEXP_EOF_IN_COMMENT;
1625       else /* (in_string_sq || in_string_dq) */
1626         value.v_error = G_ERR_UNEXP_EOF_IN_STRING;
1627     }
1628
1629   if (gstring)
1630     {
1631       value.v_string = gstring->str;
1632       g_string_free (gstring, FALSE);
1633       gstring = NULL;
1634     }
1635
1636   if (token == G_TOKEN_IDENTIFIER)
1637     {
1638       if (config->scan_symbols)
1639         {
1640           register GScannerKey  *key;
1641           register guint scope_id;
1642
1643           scope_id = scanner->scope_id;
1644           key = g_scanner_lookup_internal (scanner, scope_id, value.v_identifier);
1645           if (!key && scope_id && scanner->config->scope_0_fallback)
1646             key = g_scanner_lookup_internal (scanner, 0, value.v_identifier);
1647
1648           if (key)
1649             {
1650               g_free (value.v_identifier);
1651               token = G_TOKEN_SYMBOL;
1652               value.v_symbol = key->value;
1653             }
1654         }
1655
1656       if (token == G_TOKEN_IDENTIFIER &&
1657           config->scan_identifier_NULL &&
1658           strlen (value.v_identifier) == 4)
1659         {
1660           gchar *null_upper = "NULL";
1661           gchar *null_lower = "null";
1662
1663           if (scanner->config->case_sensitive)
1664             {
1665               if (value.v_identifier[0] == null_upper[0] &&
1666                   value.v_identifier[1] == null_upper[1] &&
1667                   value.v_identifier[2] == null_upper[2] &&
1668                   value.v_identifier[3] == null_upper[3])
1669                 token = G_TOKEN_IDENTIFIER_NULL;
1670             }
1671           else
1672             {
1673               if ((value.v_identifier[0] == null_upper[0] ||
1674                    value.v_identifier[0] == null_lower[0]) &&
1675                   (value.v_identifier[1] == null_upper[1] ||
1676                    value.v_identifier[1] == null_lower[1]) &&
1677                   (value.v_identifier[2] == null_upper[2] ||
1678                    value.v_identifier[2] == null_lower[2]) &&
1679                   (value.v_identifier[3] == null_upper[3] ||
1680                    value.v_identifier[3] == null_lower[3]))
1681                 token = G_TOKEN_IDENTIFIER_NULL;
1682             }
1683         }
1684     }
1685
1686   *token_p = token;
1687   *value_p = value;
1688 }