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