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        <unistd.h>
  29 #include        <errno.h>
  30 #include        <sys/types.h>   /* needed for sys/stat.h */
  31 #include        <sys/stat.h>
  32 #include        "glib.h"
  33
  34
  35
  36 /* --- defines --- */
  37 #define to_lower(c)                             ( \
  38         (guchar) (                                                      \
  39           ( (((guchar)(c))>='A' && ((guchar)(c))<='Z') * ('a'-'A') ) |  \
  40           ( (((guchar)(c))>=192 && ((guchar)(c))<=214) * (224-192) ) |  \
  41           ( (((guchar)(c))>=216 && ((guchar)(c))<=222) * (248-216) ) |  \
  42           ((guchar)(c))                                                 \
  43         )                                                               \
  44 )
  45 #define READ_BUFFER_SIZE        (4000)
  46
  47
  48 /* --- typedefs --- */
  49 typedef struct  _GScannerKey    GScannerKey;
  50
  51 struct  _GScannerKey
  52 {
  53   guint          scope_id;
  54   gchar         *symbol;
  55   gpointer       value;
  56 };
  57
  58
  59
  60 /* --- variables --- */
  61 static GScannerConfig g_scanner_config_template =
  62 {
  63   (
  64    " \t\n"
  65    )                    /* cset_skip_characters */,
  66   (
  67    G_CSET_a_2_z
  68    "_"
  69    G_CSET_A_2_Z
  70    )                    /* cset_identifier_first */,
  71   (
  72    G_CSET_a_2_z
  73    "_0123456789"
  74    G_CSET_A_2_Z
  75    G_CSET_LATINS
  76    G_CSET_LATINC
  77    )                    /* cset_identifier_nth */,
  78   ( "#\n" )             /* cpair_comment_single */,
  79
  80   FALSE                 /* case_sensitive */,
  81
  82   TRUE                  /* skip_comment_multi */,
  83   TRUE                  /* skip_comment_single */,
  84   TRUE                  /* scan_comment_multi */,
  85   TRUE                  /* scan_identifier */,
  86   FALSE                 /* scan_identifier_1char */,
  87   FALSE                 /* scan_identifier_NULL */,
  88   TRUE                  /* scan_symbols */,
  89   FALSE                 /* scan_binary */,
  90   TRUE                  /* scan_octal */,
  91   TRUE                  /* scan_float */,
  92   TRUE                  /* scan_hex */,
  93   FALSE                 /* scan_hex_dollar */,
  94   TRUE                  /* scan_string_sq */,
  95   TRUE                  /* scan_string_dq */,
  96   TRUE                  /* numbers_2_int */,
  97   FALSE                 /* int_2_float */,
  98   FALSE                 /* identifier_2_string */,
  99   TRUE                  /* char_2_token */,
 100   FALSE                 /* symbol_2_token */,
 101   FALSE                 /* scope_0_fallback */,
 102 };
 103
 104
 105 /* --- prototypes --- */
 106 static inline
 107 GScannerKey*    g_scanner_lookup_internal (GScanner     *scanner,
 108                                            guint         scope_id,
 109                                            const gchar  *symbol);
 110 static gint     g_scanner_key_equal       (gconstpointer v1,
 111                                            gconstpointer v2);
 112 static guint    g_scanner_key_hash        (gconstpointer v);
 113 static void     g_scanner_get_token_ll    (GScanner     *scanner,
 114                                            GTokenType   *token_p,
 115                                            GTokenValue  *value_p,
 116                                            guint        *line_p,
 117                                            guint        *position_p);
 118 static void     g_scanner_get_token_i     (GScanner     *scanner,
 119                                            GTokenType   *token_p,
 120                                            GTokenValue  *value_p,
 121                                            guint        *line_p,
 122                                            guint        *position_p);
 123
 124 static guchar   g_scanner_peek_next_char  (GScanner     *scanner);
 125 static guchar   g_scanner_get_char        (GScanner     *scanner,
 126                                            guint        *line_p,
 127                                            guint        *position_p);
 128 static void     g_scanner_msg_handler     (GScanner     *scanner,
 129                                            gchar        *message,
 130                                            gint          is_error);
 131
 132
 133 /* --- functions --- */
 134 static inline gint
 135 g_scanner_char_2_num (guchar    c,
 136                       guchar    base)
 137 {
 138   if (c >= '0' && c <= '9')
 139     c -= '0';
 140   else if (c >= 'A' && c <= 'Z')
 141     c -= 'A' - 10;
 142   else if (c >= 'a' && c <= 'z')
 143     c -= 'a' - 10;
 144   else
 145     return -1;
 146
 147   if (c < base)
 148     return c;
 149
 150   return -1;
 151 }
 152
 153 GScanner*
 154 g_scanner_new (GScannerConfig   *config_templ)
 155 {
 156   register GScanner     *scanner;
 157
 158   if (!config_templ)
 159     config_templ = &g_scanner_config_template;
 160
 161   scanner = g_new0 (GScanner, 1);
 162
 163   scanner->user_data = NULL;
 164   scanner->max_parse_errors = 0;
 165   scanner->parse_errors = 0;
 166   scanner->input_name = NULL;
 167   scanner->derived_data = NULL;
 168
 169   scanner->config = g_new0 (GScannerConfig, 1);
 170
 171   scanner->config->case_sensitive       = config_templ->case_sensitive;
 172   scanner->config->cset_skip_characters = config_templ->cset_skip_characters;
 173   scanner->config->cset_identifier_first= config_templ->cset_identifier_first;
 174   scanner->config->cset_identifier_nth  = config_templ->cset_identifier_nth;
 175   scanner->config->cpair_comment_single = config_templ->cpair_comment_single;
 176   scanner->config->skip_comment_multi   = config_templ->skip_comment_multi;
 177   scanner->config->skip_comment_single  = config_templ->skip_comment_single;
 178   scanner->config->scan_comment_multi   = config_templ->scan_comment_multi;
 179   scanner->config->scan_identifier      = config_templ->scan_identifier;
 180   scanner->config->scan_identifier_1char= config_templ->scan_identifier_1char;
 181   scanner->config->scan_identifier_NULL = config_templ->scan_identifier_NULL;
 182   scanner->config->scan_symbols         = config_templ->scan_symbols;
 183   scanner->config->scan_binary          = config_templ->scan_binary;
 184   scanner->config->scan_octal           = config_templ->scan_octal;
 185   scanner->config->scan_float           = config_templ->scan_float;
 186   scanner->config->scan_hex             = config_templ->scan_hex;
 187   scanner->config->scan_hex_dollar      = config_templ->scan_hex_dollar;
 188   scanner->config->scan_string_sq       = config_templ->scan_string_sq;
 189   scanner->config->scan_string_dq       = config_templ->scan_string_dq;
 190   scanner->config->numbers_2_int        = config_templ->numbers_2_int;
 191   scanner->config->int_2_float          = config_templ->int_2_float;
 192   scanner->config->identifier_2_string  = config_templ->identifier_2_string;
 193   scanner->config->char_2_token         = config_templ->char_2_token;
 194   scanner->config->symbol_2_token       = config_templ->symbol_2_token;
 195   scanner->config->scope_0_fallback     = config_templ->scope_0_fallback;
 196
 197   scanner->token = G_TOKEN_NONE;
 198   scanner->value.v_int = 0;
 199   scanner->line = 1;
 200   scanner->position = 0;
 201
 202   scanner->next_token = G_TOKEN_NONE;
 203   scanner->next_value.v_int = 0;
 204   scanner->next_line = 1;
 205   scanner->next_position = 0;
 206
 207   scanner->symbol_table = g_hash_table_new (g_scanner_key_hash, g_scanner_key_equal);
 208   scanner->input_fd = -1;
 209   scanner->text = NULL;
 210   scanner->text_end = NULL;
 211   scanner->buffer = NULL;
 212   scanner->scope_id = 0;
 213
 214   scanner->msg_handler = g_scanner_msg_handler;
 215
 216   return scanner;
 217 }
 218
 219 static inline void
 220 g_scanner_free_value (GTokenType     *token_p,
 221                       GTokenValue     *value_p)
 222 {
 223   switch (*token_p)
 224     {
 225     case  G_TOKEN_STRING:
 226     case  G_TOKEN_IDENTIFIER:
 227     case  G_TOKEN_IDENTIFIER_NULL:
 228     case  G_TOKEN_COMMENT_SINGLE:
 229     case  G_TOKEN_COMMENT_MULTI:
 230       g_free (value_p->v_string);
 231       break;
 232
 233     default:
 234       break;
 235     }
 236
 237   *token_p = G_TOKEN_NONE;
 238 }
 239
 240 static void
 241 g_scanner_destroy_symbol_table_entry (gpointer _key,
 242                                       gpointer _value,
 243                                       gpointer _data)
 244 {
 245   GScannerKey *key = _key;
 246
 247   g_free (key->symbol);
 248   g_free (key);
 249 }
 250
 251 void
 252 g_scanner_destroy (GScanner     *scanner)
 253 {
 254   g_return_if_fail (scanner != NULL);
 255
 256   g_hash_table_foreach (scanner->symbol_table,
 257                         g_scanner_destroy_symbol_table_entry, NULL);
 258   g_hash_table_destroy (scanner->symbol_table);
 259   g_scanner_free_value (&scanner->token, &scanner->value);
 260   g_scanner_free_value (&scanner->next_token, &scanner->next_value);
 261   g_free (scanner->config);
 262   g_free (scanner->buffer);
 263   g_free (scanner);
 264 }
 265
 266 static void
 267 g_scanner_msg_handler (GScanner         *scanner,
 268                        gchar            *message,
 269                        gint              is_error)
 270 {
 271   g_return_if_fail (scanner != NULL);
 272
 273   fprintf (stdout, "%s:%d: ", scanner->input_name, scanner->line);
 274   if (is_error)
 275     fprintf (stdout, "error: ");
 276   fprintf (stdout, "%s\n", message);
 277 }
 278
 279 void
 280 g_scanner_error (GScanner       *scanner,
 281                  const gchar    *format,
 282                  ...)
 283 {
 284   g_return_if_fail (scanner != NULL);
 285   g_return_if_fail (format != NULL);
 286
 287   scanner->parse_errors++;
 288
 289   if (scanner->msg_handler)
 290     {
 291       va_list args;
 292       gchar *string;
 293
 294       va_start (args, format);
 295       string = g_strdup_vprintf (format, args);
 296       va_end (args);
 297
 298       scanner->msg_handler (scanner, string, TRUE);
 299
 300       g_free (string);
 301     }
 302 }
 303
 304 void
 305 g_scanner_warn (GScanner       *scanner,
 306                 const gchar    *format,
 307                 ...)
 308 {
 309   g_return_if_fail (scanner != NULL);
 310   g_return_if_fail (format != NULL);
 311
 312   if (scanner->msg_handler)
 313     {
 314       va_list args;
 315       gchar *string;
 316
 317       va_start (args, format);
 318       string = g_strdup_vprintf (format, args);
 319       va_end (args);
 320
 321       scanner->msg_handler (scanner, string, FALSE);
 322
 323       g_free (string);
 324     }
 325 }
 326
 327 static gint
 328 g_scanner_key_equal (gconstpointer v1,
 329                      gconstpointer v2)
 330 {
 331   register const GScannerKey *key1 = v1;
 332   register const GScannerKey *key2 = v2;
 333
 334   return (key1->scope_id == key2->scope_id) && (strcmp (key1->symbol, key2->symbol) == 0);
 335 }
 336
 337 static guint
 338 g_scanner_key_hash (gconstpointer v)
 339 {
 340   register const GScannerKey *key = v;
 341   register gchar *c;
 342   register guint h;
 343
 344   h = key->scope_id;
 345   for (c = key->symbol; *c; c++)
 346     {
 347       register guint g;
 348
 349       h = (h << 4) + *c;
 350       g = h & 0xf0000000;
 351       if (g)
 352         {
 353           h = h ^ (g >> 24);
 354           h = h ^ g;
 355         }
 356     }
 357
 358   return h;
 359 }
 360
 361 static inline GScannerKey*
 362 g_scanner_lookup_internal (GScanner     *scanner,
 363                            guint         scope_id,
 364                            const gchar  *symbol)
 365 {
 366   register GScannerKey  *key_p;
 367   GScannerKey key;
 368
 369   key.scope_id = scope_id;
 370
 371   if (!scanner->config->case_sensitive)
 372     {
 373       register gchar *d;
 374       register const gchar *c;
 375
 376       key.symbol = g_new (gchar, strlen (symbol) + 1);
 377       for (d = key.symbol, c = symbol; *c; c++, d++)
 378         *d = to_lower (*c);
 379       *d = 0;
 380       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 381       g_free (key.symbol);
 382     }
 383   else
 384     {
 385       key.symbol = (gchar*) symbol;
 386       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 387     }
 388
 389   return key_p;
 390 }
 391
 392 void
 393 g_scanner_scope_add_symbol (GScanner    *scanner,
 394                             guint        scope_id,
 395                             const gchar *symbol,
 396                             gpointer     value)
 397 {
 398   register GScannerKey  *key;
 399
 400   g_return_if_fail (scanner != NULL);
 401   g_return_if_fail (symbol != NULL);
 402
 403   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 404
 405   if (!key)
 406     {
 407       key = g_new (GScannerKey, 1);
 408       key->scope_id = scope_id;
 409       key->symbol = g_strdup (symbol);
 410       key->value = value;
 411       if (!scanner->config->case_sensitive)
 412         {
 413           register gchar *c;
 414
 415           c = key->symbol;
 416           while (*c != 0)
 417             {
 418               *c = to_lower (*c);
 419               c++;
 420             }
 421         }
 422       g_hash_table_insert (scanner->symbol_table, key, key);
 423     }
 424   else
 425     key->value = value;
 426 }
 427
 428 void
 429 g_scanner_scope_remove_symbol (GScanner    *scanner,
 430                                guint        scope_id,
 431                                const gchar *symbol)
 432 {
 433   register GScannerKey  *key;
 434
 435   g_return_if_fail (scanner != NULL);
 436   g_return_if_fail (symbol != NULL);
 437
 438   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 439
 440   if (key)
 441     {
 442       g_hash_table_remove (scanner->symbol_table, key);
 443       g_free (key->symbol);
 444       g_free (key);
 445     }
 446 }
 447
 448 gpointer
 449 g_scanner_lookup_symbol (GScanner       *scanner,
 450                          const gchar    *symbol)
 451 {
 452   register GScannerKey  *key;
 453   register guint scope_id;
 454
 455   g_return_val_if_fail (scanner != NULL, NULL);
 456
 457   if (!symbol)
 458     return NULL;
 459
 460   scope_id = scanner->scope_id;
 461   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 462   if (!key && scope_id && scanner->config->scope_0_fallback)
 463     key = g_scanner_lookup_internal (scanner, 0, symbol);
 464
 465   if (key)
 466     return key->value;
 467   else
 468     return NULL;
 469 }
 470
 471 gpointer
 472 g_scanner_scope_lookup_symbol (GScanner       *scanner,
 473                                guint           scope_id,
 474                                const gchar    *symbol)
 475 {
 476   register GScannerKey  *key;
 477
 478   g_return_val_if_fail (scanner != NULL, NULL);
 479
 480   if (!symbol)
 481     return NULL;
 482
 483   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 484
 485   if (key)
 486     return key->value;
 487   else
 488     return NULL;
 489 }
 490
 491 guint
 492 g_scanner_set_scope (GScanner       *scanner,
 493                      guint           scope_id)
 494 {
 495   register guint old_scope_id;
 496
 497   g_return_val_if_fail (scanner != NULL, 0);
 498
 499   old_scope_id = scanner->scope_id;
 500   scanner->scope_id = scope_id;
 501
 502   return old_scope_id;
 503 }
 504
 505 static void
 506 g_scanner_foreach_internal (gpointer  _key,
 507                             gpointer  _value,
 508                             gpointer  _user_data)
 509 {
 510   register GScannerKey *key;
 511   register gpointer *d;
 512   register GHFunc func;
 513   register gpointer func_data;
 514   register guint *scope_id;
 515
 516   d = _user_data;
 517   func = (GHFunc) d[0];
 518   func_data = d[1];
 519   scope_id = d[2];
 520   key = _value;
 521
 522   if (key->scope_id == *scope_id)
 523     func (key->symbol, key->value, func_data);
 524 }
 525
 526 void
 527 g_scanner_scope_foreach_symbol (GScanner       *scanner,
 528                                 guint           scope_id,
 529                                 GHFunc          func,
 530                                 gpointer        func_data)
 531 {
 532   gpointer d[3];
 533
 534   g_return_if_fail (scanner != NULL);
 535
 536   d[0] = (gpointer) func;
 537   d[1] = func_data;
 538   d[2] = &scope_id;
 539
 540   g_hash_table_foreach (scanner->symbol_table, g_scanner_foreach_internal, d);
 541 }
 542
 543 void
 544 g_scanner_freeze_symbol_table (GScanner *scanner)
 545 {
 546   g_return_if_fail (scanner != NULL);
 547
 548   g_hash_table_freeze (scanner->symbol_table);
 549 }
 550
 551 void
 552 g_scanner_thaw_symbol_table (GScanner *scanner)
 553 {
 554   g_return_if_fail (scanner != NULL);
 555
 556   g_hash_table_thaw (scanner->symbol_table);
 557 }
 558
 559 GTokenType
 560 g_scanner_peek_next_token (GScanner     *scanner)
 561 {
 562   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 563
 564   if (scanner->next_token == G_TOKEN_NONE)
 565     {
 566       scanner->next_line = scanner->line;
 567       scanner->next_position = scanner->position;
 568       g_scanner_get_token_i (scanner,
 569                              &scanner->next_token,
 570                              &scanner->next_value,
 571                              &scanner->next_line,
 572                              &scanner->next_position);
 573     }
 574
 575   return scanner->next_token;
 576 }
 577
 578 GTokenType
 579 g_scanner_get_next_token (GScanner      *scanner)
 580 {
 581   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 582
 583   if (scanner->next_token != G_TOKEN_NONE)
 584     {
 585       g_scanner_free_value (&scanner->token, &scanner->value);
 586
 587       scanner->token = scanner->next_token;
 588       scanner->value = scanner->next_value;
 589       scanner->line = scanner->next_line;
 590       scanner->position = scanner->next_position;
 591       scanner->next_token = G_TOKEN_NONE;
 592     }
 593   else
 594     g_scanner_get_token_i (scanner,
 595                            &scanner->token,
 596                            &scanner->value,
 597                            &scanner->line,
 598                            &scanner->position);
 599
 600   return scanner->token;
 601 }
 602
 603 GTokenType
 604 g_scanner_cur_token (GScanner *scanner)
 605 {
 606   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 607
 608   return scanner->token;
 609 }
 610
 611 GTokenValue
 612 g_scanner_cur_value (GScanner *scanner)
 613 {
 614   GTokenValue v;
 615
 616   v.v_int = 0;
 617   g_return_val_if_fail (scanner != NULL, v);
 618
 619   return scanner->value;
 620 }
 621
 622 guint
 623 g_scanner_cur_line (GScanner *scanner)
 624 {
 625   g_return_val_if_fail (scanner != NULL, 0);
 626
 627   return scanner->line;
 628 }
 629
 630 guint
 631 g_scanner_cur_position (GScanner *scanner)
 632 {
 633   g_return_val_if_fail (scanner != NULL, 0);
 634
 635   return scanner->position;
 636 }
 637
 638 gboolean
 639 g_scanner_eof (GScanner *scanner)
 640 {
 641   g_return_val_if_fail (scanner != NULL, TRUE);
 642
 643   return scanner->token == G_TOKEN_EOF;
 644 }
 645
 646 void
 647 g_scanner_input_file (GScanner *scanner,
 648                       gint      input_fd)
 649 {
 650   g_return_if_fail (scanner != NULL);
 651   g_return_if_fail (input_fd >= 0);
 652
 653   scanner->token = G_TOKEN_NONE;
 654   scanner->value.v_int = 0;
 655   scanner->line = 1;
 656   scanner->position = 0;
 657   scanner->next_token = G_TOKEN_NONE;
 658
 659   scanner->input_fd = input_fd;
 660   scanner->text = NULL;
 661   scanner->text_end = NULL;
 662
 663   if (!scanner->buffer)
 664     scanner->buffer = g_new (gchar, READ_BUFFER_SIZE + 1);
 665 }
 666
 667 void
 668 g_scanner_input_text (GScanner    *scanner,
 669                       const gchar *text,
 670                       guint        text_len)
 671 {
 672   g_return_if_fail (scanner != NULL);
 673   if (text_len)
 674     g_return_if_fail (text != NULL);
 675   else
 676     text = NULL;
 677
 678   scanner->token = G_TOKEN_NONE;
 679   scanner->value.v_int = 0;
 680   scanner->line = 1;
 681   scanner->position = 0;
 682   scanner->next_token = G_TOKEN_NONE;
 683
 684   scanner->input_fd = -1;
 685   scanner->text = text;
 686   scanner->text_end = text + text_len;
 687
 688   if (scanner->buffer)
 689     {
 690       g_free (scanner->buffer);
 691       scanner->buffer = NULL;
 692     }
 693 }
 694
 695 static guchar
 696 g_scanner_peek_next_char (GScanner *scanner)
 697 {
 698   if (scanner->text < scanner->text_end)
 699     {
 700       return *scanner->text;
 701     }
 702   else if (scanner->input_fd >= 0)
 703     {
 704       register gint count;
 705       register gchar *buffer;
 706
 707       buffer = scanner->buffer;
 708       do
 709         {
 710           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 711         }
 712       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 713
 714       if (count < 1)
 715         {
 716           scanner->input_fd = -1;
 717
 718           return 0;
 719         }
 720       else
 721         {
 722           scanner->text = buffer;
 723           scanner->text_end = buffer + count;
 724
 725           return *buffer;
 726         }
 727     }
 728   else
 729     return 0;
 730 }
 731
 732 static guchar
 733 g_scanner_get_char (GScanner    *scanner,
 734                     guint       *line_p,
 735                     guint       *position_p)
 736 {
 737   register guchar fchar;
 738
 739   if (scanner->text < scanner->text_end)
 740     fchar = *(scanner->text++);
 741   else if (scanner->input_fd >= 0)
 742     {
 743       register gint count;
 744       register gchar *buffer;
 745
 746       buffer = scanner->buffer;
 747       do
 748         {
 749           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 750         }
 751       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 752
 753       if (count < 1)
 754         {
 755           scanner->input_fd = -1;
 756           fchar = 0;
 757         }
 758       else
 759         {
 760           scanner->text = buffer + 1;
 761           scanner->text_end = buffer + count;
 762           fchar = *buffer;
 763         }
 764     }
 765   else
 766     fchar = 0;
 767
 768   if (fchar == '\n')
 769     {
 770       (*position_p) = 0;
 771       (*line_p)++;
 772     }
 773   else if (fchar)
 774     {
 775       (*position_p)++;
 776     }
 777
 778   return fchar;
 779 }
 780
 781 void
 782 g_scanner_unexp_token (GScanner         *scanner,
 783                        GTokenType        expected_token,
 784                        const gchar      *identifier_spec,
 785                        const gchar      *symbol_spec,
 786                        const gchar      *symbol_name,
 787                        const gchar      *message,
 788                        gint              is_error)
 789 {
 790   register gchar        *token_string;
 791   register guint        token_string_len;
 792   register gchar        *expected_string;
 793   register guint        expected_string_len;
 794   register gchar        *message_prefix;
 795   register gboolean     print_unexp;
 796   void (*msg_handler)   (GScanner*, const gchar*, ...);
 797
 798   g_return_if_fail (scanner != NULL);
 799
 800   if (is_error)
 801     msg_handler = g_scanner_error;
 802   else
 803     msg_handler = g_scanner_warn;
 804
 805   if (!identifier_spec)
 806     identifier_spec = "identifier";
 807   if (!symbol_spec)
 808     symbol_spec = "symbol";
 809
 810   token_string_len = 56;
 811   token_string = g_new (gchar, token_string_len + 1);
 812   expected_string_len = 64;
 813   expected_string = g_new (gchar, expected_string_len + 1);
 814   print_unexp = TRUE;
 815
 816   switch (scanner->token)
 817     {
 818     case  G_TOKEN_EOF:
 819       g_snprintf (token_string, token_string_len, "end of file");
 820       break;
 821
 822     default:
 823       if (scanner->token >= 1 && scanner->token <= 255)
 824         {
 825           if ((scanner->token >= ' ' && scanner->token <= '~') ||
 826               strchr (scanner->config->cset_identifier_first, scanner->token) ||
 827               strchr (scanner->config->cset_identifier_nth, scanner->token))
 828             g_snprintf (token_string, expected_string_len, "character `%c'", scanner->token);
 829           else
 830             g_snprintf (token_string, expected_string_len, "character `\\%o'", scanner->token);
 831           break;
 832         }
 833       else if (!scanner->config->symbol_2_token)
 834         {
 835           g_snprintf (token_string, token_string_len, "(unknown) token <%d>", scanner->token);
 836           break;
 837         }
 838       /* fall through */
 839     case  G_TOKEN_SYMBOL:
 840       if (expected_token == G_TOKEN_SYMBOL ||
 841           (scanner->config->symbol_2_token &&
 842            expected_token > G_TOKEN_LAST))
 843         print_unexp = FALSE;
 844       if (symbol_name)
 845         g_snprintf (token_string,
 846                     token_string_len,
 847                     "%s%s `%s'",
 848                     print_unexp ? "" : "invalid ",
 849                     symbol_spec,
 850                     symbol_name);
 851       else
 852         g_snprintf (token_string,
 853                     token_string_len,
 854                     "%s%s",
 855                     print_unexp ? "" : "invalid ",
 856                     symbol_spec);
 857       break;
 858
 859     case  G_TOKEN_ERROR:
 860       print_unexp = FALSE;
 861       expected_token = G_TOKEN_NONE;
 862       switch (scanner->value.v_error)
 863         {
 864         case  G_ERR_UNEXP_EOF:
 865           g_snprintf (token_string, token_string_len, "scanner: unexpected end of file");
 866           break;
 867
 868         case  G_ERR_UNEXP_EOF_IN_STRING:
 869           g_snprintf (token_string, token_string_len, "scanner: unterminated string constant");
 870           break;
 871
 872         case  G_ERR_UNEXP_EOF_IN_COMMENT:
 873           g_snprintf (token_string, token_string_len, "scanner: unterminated comment");
 874           break;
 875
 876         case  G_ERR_NON_DIGIT_IN_CONST:
 877           g_snprintf (token_string, token_string_len, "scanner: non digit in constant");
 878           break;
 879
 880         case  G_ERR_FLOAT_RADIX:
 881           g_snprintf (token_string, token_string_len, "scanner: invalid radix for floating constant");
 882           break;
 883
 884         case  G_ERR_FLOAT_MALFORMED:
 885           g_snprintf (token_string, token_string_len, "scanner: malformed floating constant");
 886           break;
 887
 888         case  G_ERR_DIGIT_RADIX:
 889           g_snprintf (token_string, token_string_len, "scanner: digit is beyond radix");
 890           break;
 891
 892         case  G_ERR_UNKNOWN:
 893         default:
 894           g_snprintf (token_string, token_string_len, "scanner: unknown error");
 895           break;
 896         }
 897       break;
 898
 899     case  G_TOKEN_CHAR:
 900       g_snprintf (token_string, token_string_len, "character `%c'", scanner->value.v_char);
 901       break;
 902
 903     case  G_TOKEN_IDENTIFIER:
 904       if (expected_token == G_TOKEN_IDENTIFIER)
 905         print_unexp = FALSE;
 906       g_snprintf (token_string,
 907                   token_string_len,
 908                   "%s%s `%s'",
 909                   print_unexp ? "" : "invalid ",
 910                   identifier_spec,
 911                   scanner->value.v_string);
 912       break;
 913
 914     case  G_TOKEN_BINARY:
 915     case  G_TOKEN_OCTAL:
 916     case  G_TOKEN_INT:
 917     case  G_TOKEN_HEX:
 918       g_snprintf (token_string, token_string_len, "number `%ld'", scanner->value.v_int);
 919       break;
 920
 921     case  G_TOKEN_FLOAT:
 922       g_snprintf (token_string, token_string_len, "number `%.3f'", scanner->value.v_float);
 923       break;
 924
 925     case  G_TOKEN_STRING:
 926       if (expected_token == G_TOKEN_STRING)
 927         print_unexp = FALSE;
 928       g_snprintf (token_string,
 929                   token_string_len,
 930                   "%s%sstring constant \"%s\"",
 931                   print_unexp ? "" : "invalid ",
 932                   scanner->value.v_string[0] == 0 ? "empty " : "",
 933                   scanner->value.v_string);
 934       token_string[token_string_len - 2] = '"';
 935       token_string[token_string_len - 1] = 0;
 936       break;
 937
 938     case  G_TOKEN_COMMENT_SINGLE:
 939     case  G_TOKEN_COMMENT_MULTI:
 940       g_snprintf (token_string, token_string_len, "comment");
 941       break;
 942
 943     case  G_TOKEN_NONE:
 944       /* somehow the user's parsing code is screwed, there isn't much
 945        * we can do about it.
 946        * Note, a common case to trigger this is
 947        * g_scanner_peek_next_token(); g_scanner_unexp_token();
 948        * without an intermediate g_scanner_get_next_token().
 949        */
 950       g_assert_not_reached ();
 951       break;
 952     }
 953
 954
 955   switch (expected_token)
 956     {
 957       gboolean need_valid;
 958
 959     default:
 960       if (expected_token >= 1 && expected_token <= 255)
 961         {
 962           if ((expected_token >= ' ' && expected_token <= '~') ||
 963               strchr (scanner->config->cset_identifier_first, expected_token) ||
 964               strchr (scanner->config->cset_identifier_nth, expected_token))
 965             g_snprintf (expected_string, expected_string_len, "character `%c'", expected_token);
 966           else
 967             g_snprintf (expected_string, expected_string_len, "character `\\%o'", expected_token);
 968           break;
 969         }
 970       else if (!scanner->config->symbol_2_token)
 971         {
 972           g_snprintf (expected_string, expected_string_len, "(unknown) token <%d>", expected_token);
 973           break;
 974         }
 975       /* fall through */
 976     case  G_TOKEN_SYMBOL:
 977       need_valid = (scanner->token == G_TOKEN_SYMBOL ||
 978                     (scanner->config->symbol_2_token &&
 979                      scanner->token > G_TOKEN_LAST));
 980       g_snprintf (expected_string,
 981                   expected_string_len,
 982                   "%s%s",
 983                   need_valid ? "valid " : "",
 984                   symbol_spec);
 985       /* FIXME: should we attempt to lookup the symbol_name for symbol_2_token? */
 986       break;
 987
 988     case  G_TOKEN_INT:
 989       g_snprintf (expected_string, expected_string_len, "number (integer)");
 990       break;
 991
 992     case  G_TOKEN_FLOAT:
 993       g_snprintf (expected_string, expected_string_len, "number (float)");
 994       break;
 995
 996     case  G_TOKEN_STRING:
 997       g_snprintf (expected_string,
 998                   expected_string_len,
 999                   "%sstring constant",
1000                   scanner->token == G_TOKEN_STRING ? "valid " : "");
1001       break;
1002
1003     case  G_TOKEN_IDENTIFIER:
1004       g_snprintf (expected_string,
1005                   expected_string_len,
1006                   "%s%s",
1007                   scanner->token == G_TOKEN_IDENTIFIER ? "valid " : "",
1008                   identifier_spec);
1009       break;
1010
1011     case  G_TOKEN_NONE:
1012       break;
1013     }
1014
1015   if (message && message[0] != 0)
1016     message_prefix = " - ";
1017   else
1018     {
1019       message_prefix = "";
1020       message = "";
1021     }
1022
1023   if (expected_token != G_TOKEN_NONE)
1024     {
1025       if (print_unexp)
1026         msg_handler (scanner,
1027                      "unexpected %s, expected %s%s%s",
1028                      token_string,
1029                      expected_string,
1030                      message_prefix,
1031                      message);
1032       else
1033         msg_handler (scanner,
1034                      "%s, expected %s%s%s",
1035                      token_string,
1036                      expected_string,
1037                      message_prefix,
1038                      message);
1039     }
1040   else
1041     {
1042       if (print_unexp)
1043         msg_handler (scanner,
1044                      "unexpected %s%s%s",
1045                      token_string,
1046                      message_prefix,
1047                      message);
1048       else
1049         msg_handler (scanner,
1050                      "%s%s%s",
1051                      token_string,
1052                      message_prefix,
1053                      message);
1054     }
1055
1056   g_free (token_string);
1057   g_free (expected_string);
1058 }
1059
1060 gint
1061 g_scanner_stat_mode (const gchar *filename)
1062 {
1063   struct stat  *stat_buf;
1064   gint          st_mode;
1065
1066   stat_buf = g_new0 (struct stat, 1);
1067
1068   lstat (filename, stat_buf);
1069
1070   st_mode = stat_buf->st_mode;
1071
1072   g_free (stat_buf);
1073
1074   return st_mode;
1075 }
1076
1077 static void
1078 g_scanner_get_token_i (GScanner *scanner,
1079                        GTokenType       *token_p,
1080                        GTokenValue      *value_p,
1081                        guint            *line_p,
1082                        guint            *position_p)
1083 {
1084   do
1085     {
1086       g_scanner_free_value (token_p, value_p);
1087       g_scanner_get_token_ll (scanner, token_p, value_p, line_p, position_p);
1088     }
1089   while (((*token_p > 0 && *token_p < 256) &&
1090           strchr (scanner->config->cset_skip_characters, *token_p)) ||
1091          (*token_p == G_TOKEN_CHAR &&
1092           strchr (scanner->config->cset_skip_characters, value_p->v_char)) ||
1093          (*token_p == G_TOKEN_COMMENT_MULTI &&
1094           scanner->config->skip_comment_multi) ||
1095          (*token_p == G_TOKEN_COMMENT_SINGLE &&
1096           scanner->config->skip_comment_single));
1097
1098   switch (*token_p)
1099     {
1100     case G_TOKEN_IDENTIFIER:
1101       if (scanner->config->identifier_2_string)
1102         *token_p = G_TOKEN_STRING;
1103       break;
1104
1105     case G_TOKEN_SYMBOL:
1106       if (scanner->config->symbol_2_token)
1107         *token_p = (GTokenType) value_p->v_symbol;
1108       break;
1109
1110     case G_TOKEN_BINARY:
1111     case G_TOKEN_OCTAL:
1112     case G_TOKEN_HEX:
1113       if (scanner->config->numbers_2_int)
1114         *token_p = G_TOKEN_INT;
1115       break;
1116
1117     default:
1118       break;
1119     }
1120
1121   if (*token_p == G_TOKEN_INT &&
1122       scanner->config->int_2_float)
1123     {
1124       *token_p = G_TOKEN_FLOAT;
1125       value_p->v_float = value_p->v_int;
1126     }
1127
1128   errno = 0;
1129 }
1130
1131 static void
1132 g_scanner_get_token_ll  (GScanner       *scanner,
1133                          GTokenType     *token_p,
1134                          GTokenValue    *value_p,
1135                          guint          *line_p,
1136                          guint          *position_p)
1137 {
1138   register GScannerConfig *config;
1139   register GTokenType      token;
1140   register gboolean        in_comment_multi;
1141   register gboolean        in_comment_single;
1142   register gboolean        in_string_sq;
1143   register gboolean        in_string_dq;
1144   register GString        *gstring;
1145   GTokenValue              value;
1146   guchar                   ch;
1147
1148   config = scanner->config;
1149   (*value_p).v_int = 0;
1150
1151   if ((scanner->text >= scanner->text_end && scanner->input_fd < 0) ||
1152       scanner->token == G_TOKEN_EOF)
1153     {
1154       *token_p = G_TOKEN_EOF;
1155       return;
1156     }
1157
1158   in_comment_multi = FALSE;
1159   in_comment_single = FALSE;
1160   in_string_sq = FALSE;
1161   in_string_dq = FALSE;
1162   gstring = NULL;
1163
1164   do /* while (ch != 0) */
1165     {
1166       register gboolean         dotted_float = FALSE;
1167
1168       ch = g_scanner_get_char (scanner, line_p, position_p);
1169
1170       value.v_int = 0;
1171       token = G_TOKEN_NONE;
1172
1173       /* this is *evil*, but needed ;(
1174        * we first check for identifier first character, because  it
1175        * might interfere with other key chars like slashes or numbers
1176        */
1177       if (config->scan_identifier &&
1178           ch && strchr (config->cset_identifier_first, ch))
1179         goto identifier_precedence;
1180
1181       switch (ch)
1182         {
1183         case  0:
1184           token = G_TOKEN_EOF;
1185           (*position_p)++;
1186           /* ch = 0; */
1187           break;
1188
1189         case  '/':
1190           if (!config->scan_comment_multi ||
1191               g_scanner_peek_next_char (scanner) != '*')
1192             goto default_case;
1193           g_scanner_get_char (scanner, line_p, position_p);
1194           token = G_TOKEN_COMMENT_MULTI;
1195           in_comment_multi = TRUE;
1196           gstring = g_string_new ("");
1197           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1198             {
1199               if (ch == '*' && g_scanner_peek_next_char (scanner) == '/')
1200                 {
1201                   g_scanner_get_char (scanner, line_p, position_p);
1202                   in_comment_multi = FALSE;
1203                   break;
1204                 }
1205               else
1206                 gstring = g_string_append_c (gstring, ch);
1207             }
1208           ch = 0;
1209           break;
1210
1211         case  '\'':
1212           if (!config->scan_string_sq)
1213             goto default_case;
1214           token = G_TOKEN_STRING;
1215           in_string_sq = TRUE;
1216           gstring = g_string_new ("");
1217           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1218             {
1219               if (ch == '\'')
1220                 {
1221                   in_string_sq = FALSE;
1222                   break;
1223                 }
1224               else
1225                 gstring = g_string_append_c (gstring, ch);
1226             }
1227           ch = 0;
1228           break;
1229
1230         case  '"':
1231           if (!config->scan_string_dq)
1232             goto default_case;
1233           token = G_TOKEN_STRING;
1234           in_string_dq = TRUE;
1235           gstring = g_string_new ("");
1236           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1237             {
1238               if (ch == '"')
1239                 {
1240                   in_string_dq = FALSE;
1241                   break;
1242                 }
1243               else
1244                 {
1245                   if (ch == '\\')
1246                     {
1247                       ch = g_scanner_get_char (scanner, line_p, position_p);
1248                       switch (ch)
1249                         {
1250                           register guint        i;
1251                           register guint        fchar;
1252
1253                         case  0:
1254                           break;
1255
1256                         case  '\\':
1257                           gstring = g_string_append_c (gstring, '\\');
1258                           break;
1259
1260                         case  'n':
1261                           gstring = g_string_append_c (gstring, '\n');
1262                           break;
1263
1264                         case  't':
1265                           gstring = g_string_append_c (gstring, '\t');
1266                           break;
1267
1268                         case  'r':
1269                           gstring = g_string_append_c (gstring, '\r');
1270                           break;
1271
1272                         case  'b':
1273                           gstring = g_string_append_c (gstring, '\b');
1274                           break;
1275
1276                         case  'f':
1277                           gstring = g_string_append_c (gstring, '\f');
1278                           break;
1279
1280                         case  '0':
1281                         case  '1':
1282                         case  '2':
1283                         case  '3':
1284                         case  '4':
1285                         case  '5':
1286                         case  '6':
1287                         case  '7':
1288                           i = ch - '0';
1289                           fchar = g_scanner_peek_next_char (scanner);
1290                           if (fchar >= '0' && fchar <= '7')
1291                             {
1292                               ch = g_scanner_get_char (scanner, line_p, position_p);
1293                               i = i * 8 + ch - '0';
1294                               fchar = g_scanner_peek_next_char (scanner);
1295                               if (fchar >= '0' && fchar <= '7')
1296                                 {
1297                                   ch = g_scanner_get_char (scanner, line_p, position_p);
1298                                   i = i * 8 + ch - '0';
1299                                 }
1300                             }
1301                           gstring = g_string_append_c (gstring, i);
1302                           break;
1303
1304                         default:
1305                           gstring = g_string_append_c (gstring, ch);
1306                           break;
1307                         }
1308                     }
1309                   else
1310                     gstring = g_string_append_c (gstring, ch);
1311                 }
1312             }
1313           ch = 0;
1314           break;
1315
1316         case  '.':
1317           if (!config->scan_float)
1318             goto default_case;
1319           token = G_TOKEN_FLOAT;
1320           dotted_float = TRUE;
1321           ch = g_scanner_get_char (scanner, line_p, position_p);
1322           goto number_parsing;
1323
1324         case  '$':
1325           if (!config->scan_hex_dollar)
1326             goto default_case;
1327           token = G_TOKEN_HEX;
1328           ch = g_scanner_get_char (scanner, line_p, position_p);
1329           goto number_parsing;
1330
1331         case  '0':
1332           if (config->scan_octal)
1333             token = G_TOKEN_OCTAL;
1334           else
1335             token = G_TOKEN_INT;
1336           ch = g_scanner_peek_next_char (scanner);
1337           if (config->scan_hex && (ch == 'x' || ch == 'X'))
1338             {
1339               token = G_TOKEN_HEX;
1340               g_scanner_get_char (scanner, line_p, position_p);
1341               ch = g_scanner_get_char (scanner, line_p, position_p);
1342               if (ch == 0)
1343                 {
1344                   token = G_TOKEN_ERROR;
1345                   value.v_error = G_ERR_UNEXP_EOF;
1346                   (*position_p)++;
1347                   break;
1348                 }
1349               if (g_scanner_char_2_num (ch, 16) < 0)
1350                 {
1351                   token = G_TOKEN_ERROR;
1352                   value.v_error = G_ERR_DIGIT_RADIX;
1353                   ch = 0;
1354                   break;
1355                 }
1356             }
1357           else if (config->scan_binary && (ch == 'b' || ch == 'B'))
1358             {
1359               token = G_TOKEN_BINARY;
1360               g_scanner_get_char (scanner, line_p, position_p);
1361               ch = g_scanner_get_char (scanner, line_p, position_p);
1362               if (ch == 0)
1363                 {
1364                   token = G_TOKEN_ERROR;
1365                   value.v_error = G_ERR_UNEXP_EOF;
1366                   (*position_p)++;
1367                   break;
1368                 }
1369               if (g_scanner_char_2_num (ch, 10) < 0)
1370                 {
1371                   token = G_TOKEN_ERROR;
1372                   value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1373                   ch = 0;
1374                   break;
1375                 }
1376             }
1377           else
1378             ch = '0';
1379           /* fall through */
1380         case  '1':
1381         case  '2':
1382         case  '3':
1383         case  '4':
1384         case  '5':
1385         case  '6':
1386         case  '7':
1387         case  '8':
1388         case  '9':
1389         number_parsing:
1390         {
1391           register gboolean in_number = TRUE;
1392           gchar *endptr;
1393
1394           if (token == G_TOKEN_NONE)
1395             token = G_TOKEN_INT;
1396
1397           gstring = g_string_new (dotted_float ? "0." : "");
1398           gstring = g_string_append_c (gstring, ch);
1399
1400           do /* while (in_number) */
1401             {
1402               register gboolean is_E;
1403
1404               is_E = token == G_TOKEN_FLOAT && (ch == 'e' || ch == 'E');
1405
1406               ch = g_scanner_peek_next_char (scanner);
1407
1408               if (g_scanner_char_2_num (ch, 36) >= 0 ||
1409                   (config->scan_float && ch == '.') ||
1410                   (is_E && (ch == '+' || ch == '-')))
1411                 {
1412                   ch = g_scanner_get_char (scanner, line_p, position_p);
1413
1414                   switch (ch)
1415                     {
1416                     case '.':
1417                       if (token != G_TOKEN_INT && token != G_TOKEN_OCTAL)
1418                         {
1419                           value.v_error = token == G_TOKEN_FLOAT ? G_ERR_FLOAT_MALFORMED : G_ERR_FLOAT_RADIX;
1420                           token = G_TOKEN_ERROR;
1421                           in_number = FALSE;
1422                         }
1423                       else
1424                         {
1425                           token = G_TOKEN_FLOAT;
1426                           gstring = g_string_append_c (gstring, ch);
1427                         }
1428                       break;
1429
1430                     case '0':
1431                     case '1':
1432                     case '2':
1433                     case '3':
1434                     case '4':
1435                     case '5':
1436                     case '6':
1437                     case '7':
1438                     case '8':
1439                     case '9':
1440                       gstring = g_string_append_c (gstring, ch);
1441                       break;
1442
1443                     case '-':
1444                     case '+':
1445                       if (token != G_TOKEN_FLOAT)
1446                         {
1447                           token = G_TOKEN_ERROR;
1448                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1449                           in_number = FALSE;
1450                         }
1451                       else
1452                         gstring = g_string_append_c (gstring, ch);
1453                       break;
1454
1455                     case 'e':
1456                     case 'E':
1457                       if ((token != G_TOKEN_HEX && !config->scan_float) ||
1458                           (token != G_TOKEN_HEX &&
1459                            token != G_TOKEN_OCTAL &&
1460                            token != G_TOKEN_FLOAT &&
1461                            token != G_TOKEN_INT))
1462                         {
1463                           token = G_TOKEN_ERROR;
1464                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1465                           in_number = FALSE;
1466                         }
1467                       else
1468                         {
1469                           if (token != G_TOKEN_HEX)
1470                             token = G_TOKEN_FLOAT;
1471                           gstring = g_string_append_c (gstring, ch);
1472                         }
1473                       break;
1474
1475                     default:
1476                       if (token != G_TOKEN_HEX)
1477                         {
1478                           token = G_TOKEN_ERROR;
1479                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1480                           in_number = FALSE;
1481                         }
1482                       else
1483                         gstring = g_string_append_c (gstring, ch);
1484                       break;
1485                     }
1486                 }
1487               else
1488                 in_number = FALSE;
1489             }
1490           while (in_number);
1491
1492           endptr = NULL;
1493           switch (token)
1494             {
1495             case G_TOKEN_BINARY:
1496               value.v_binary = strtol (gstring->str, &endptr, 2);
1497               break;
1498
1499             case G_TOKEN_OCTAL:
1500               value.v_octal = strtol (gstring->str, &endptr, 8);
1501               break;
1502
1503             case G_TOKEN_INT:
1504               value.v_int = strtol (gstring->str, &endptr, 10);
1505               break;
1506
1507             case G_TOKEN_FLOAT:
1508               value.v_float = g_strtod (gstring->str, &endptr);
1509               break;
1510
1511             case G_TOKEN_HEX:
1512               value.v_hex = strtol (gstring->str, &endptr, 16);
1513               break;
1514
1515             default:
1516               break;
1517             }
1518           if (endptr && *endptr)
1519             {
1520               token = G_TOKEN_ERROR;
1521               if (*endptr == 'e' || *endptr == 'E')
1522                 value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1523               else
1524                 value.v_error = G_ERR_DIGIT_RADIX;
1525             }
1526           g_string_free (gstring, TRUE);
1527           gstring = NULL;
1528           ch = 0;
1529         } /* number_parsing:... */
1530         break;
1531
1532         default:
1533         default_case:
1534         {
1535           if (config->cpair_comment_single &&
1536               ch == config->cpair_comment_single[0])
1537             {
1538               token = G_TOKEN_COMMENT_SINGLE;
1539               in_comment_single = TRUE;
1540               gstring = g_string_new ("");
1541               while ((ch = g_scanner_get_char (scanner,
1542                                                line_p,
1543                                                position_p)) != 0)
1544                 {
1545                   if (ch == config->cpair_comment_single[1])
1546                     {
1547                       in_comment_single = FALSE;
1548                       ch = 0;
1549                       break;
1550                     }
1551
1552                   gstring = g_string_append_c (gstring, ch);
1553                   ch = 0;
1554                 }
1555             }
1556           else if (config->scan_identifier && ch &&
1557                    strchr (config->cset_identifier_first, ch))
1558             {
1559             identifier_precedence:
1560
1561               if (config->cset_identifier_nth && ch &&
1562                   strchr (config->cset_identifier_nth,
1563                           g_scanner_peek_next_char (scanner)))
1564                 {
1565                   token = G_TOKEN_IDENTIFIER;
1566                   gstring = g_string_new ("");
1567                   gstring = g_string_append_c (gstring, ch);
1568                   do
1569                     {
1570                       ch = g_scanner_get_char (scanner, line_p, position_p);
1571                       gstring = g_string_append_c (gstring, ch);
1572                       ch = g_scanner_peek_next_char (scanner);
1573                     }
1574                   while (ch && strchr (config->cset_identifier_nth, ch));
1575                   ch = 0;
1576                 }
1577               else if (config->scan_identifier_1char)
1578                 {
1579                   token = G_TOKEN_IDENTIFIER;
1580                   value.v_identifier = g_new0 (gchar, 2);
1581                   value.v_identifier[0] = ch;
1582                   ch = 0;
1583                 }
1584             }
1585           if (ch)
1586             {
1587               if (config->char_2_token)
1588                 token = ch;
1589               else
1590                 {
1591                   token = G_TOKEN_CHAR;
1592                   value.v_char = ch;
1593                 }
1594               ch = 0;
1595             }
1596         } /* default_case:... */
1597         break;
1598         }
1599       g_assert (ch == 0 && token != G_TOKEN_NONE); /* paranoid */
1600     }
1601   while (ch != 0);
1602
1603   if (in_comment_multi || in_comment_single ||
1604       in_string_sq || in_string_dq)
1605     {
1606       token = G_TOKEN_ERROR;
1607       if (gstring)
1608         {
1609           g_string_free (gstring, TRUE);
1610           gstring = NULL;
1611         }
1612       (*position_p)++;
1613       if (in_comment_multi || in_comment_single)
1614         value.v_error = G_ERR_UNEXP_EOF_IN_COMMENT;
1615       else /* (in_string_sq || in_string_dq) */
1616         value.v_error = G_ERR_UNEXP_EOF_IN_STRING;
1617     }
1618
1619   if (gstring)
1620     {
1621       value.v_string = gstring->str;
1622       g_string_free (gstring, FALSE);
1623       gstring = NULL;
1624     }
1625
1626   if (token == G_TOKEN_IDENTIFIER)
1627     {
1628       if (config->scan_symbols)
1629         {
1630           register GScannerKey  *key;
1631           register guint scope_id;
1632
1633           scope_id = scanner->scope_id;
1634           key = g_scanner_lookup_internal (scanner, scope_id, value.v_identifier);
1635           if (!key && scope_id && scanner->config->scope_0_fallback)
1636             key = g_scanner_lookup_internal (scanner, 0, value.v_identifier);
1637
1638           if (key)
1639             {
1640               g_free (value.v_identifier);
1641               token = G_TOKEN_SYMBOL;
1642               value.v_symbol = key->value;
1643             }
1644         }
1645
1646       if (token == G_TOKEN_IDENTIFIER &&
1647           config->scan_identifier_NULL &&
1648           strlen (value.v_identifier) == 4)
1649         {
1650           gchar *null_upper = "NULL";
1651           gchar *null_lower = "null";
1652
1653           if (scanner->config->case_sensitive)
1654             {
1655               if (value.v_identifier[0] == null_upper[0] &&
1656                   value.v_identifier[1] == null_upper[1] &&
1657                   value.v_identifier[2] == null_upper[2] &&
1658                   value.v_identifier[3] == null_upper[3])
1659                 token = G_TOKEN_IDENTIFIER_NULL;
1660             }
1661           else
1662             {
1663               if ((value.v_identifier[0] == null_upper[0] ||
1664                    value.v_identifier[0] == null_lower[0]) &&
1665                   (value.v_identifier[1] == null_upper[1] ||
1666                    value.v_identifier[1] == null_lower[1]) &&
1667                   (value.v_identifier[2] == null_upper[2] ||
1668                    value.v_identifier[2] == null_lower[2]) &&
1669                   (value.v_identifier[3] == null_upper[3] ||
1670                    value.v_identifier[3] == null_lower[3]))
1671                 token = G_TOKEN_IDENTIFIER_NULL;
1672             }
1673         }
1674     }
1675
1676   *token_p = token;
1677   *value_p = value;
1678 }