gscanner.c

   1 /* GLIB - Library of useful routines for C programming
   2  * Copyright (C) 1995-1997  Peter Mattis, Spencer Kimball and Josh MacDonald
   3  *
   4  * GScanner: Flexible lexical scanner for general purpose.
   5  * Copyright (C) 1997, 1998 Tim Janik
   6  *
   7  * This library is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2 of the License, or (at your option) any later version.
  11  *
  12  * This library is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with this library; if not, write to the
  19  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  20  * Boston, MA 02111-1307, USA.
  21  */
  22
  23 /*
  24  * Modified by the GLib Team and others 1997-2000.  See the AUTHORS
  25  * file for a list of people on the GLib Team.  See the ChangeLog
  26  * files for a list of changes.  These files are distributed with
  27  * GLib at ftp://ftp.gtk.org/pub/gtk/.
  28  */
  29
  30 /*
  31  * MT safe
  32  */
  33
  34 #define         __gscanner_c__
  35
  36 #ifdef HAVE_CONFIG_H
  37 #include <config.h>
  38 #endif
  39
  40 #include        <stdlib.h>
  41 #include        <stdarg.h>
  42 #include        <string.h>
  43 #include        <stdio.h>
  44 #include        "glib.h"
  45 #ifdef HAVE_UNISTD_H
  46 #include        <unistd.h>
  47 #endif
  48 #include        <errno.h>
  49 #include        <sys/types.h>   /* needed for sys/stat.h */
  50 #include        <sys/stat.h>
  51 #ifdef G_OS_WIN32
  52 #include        <io.h>          /* For _read() */
  53 #endif
  54
  55 /* --- defines --- */
  56 #define to_lower(c)                             ( \
  57         (guchar) (                                                      \
  58           ( (((guchar)(c))>='A' && ((guchar)(c))<='Z') * ('a'-'A') ) |  \
  59           ( (((guchar)(c))>=192 && ((guchar)(c))<=214) * (224-192) ) |  \
  60           ( (((guchar)(c))>=216 && ((guchar)(c))<=222) * (248-216) ) |  \
  61           ((guchar)(c))                                                 \
  62         )                                                               \
  63 )
  64 #define READ_BUFFER_SIZE        (4000)
  65
  66
  67 /* --- typedefs --- */
  68 typedef struct  _GScannerKey    GScannerKey;
  69
  70 struct  _GScannerKey
  71 {
  72   guint          scope_id;
  73   gchar         *symbol;
  74   gpointer       value;
  75 };
  76
  77
  78
  79 /* --- variables --- */
  80 static GScannerConfig g_scanner_config_template =
  81 {
  82   (
  83    " \t\r\n"
  84    )                    /* cset_skip_characters */,
  85   (
  86    G_CSET_a_2_z
  87    "_"
  88    G_CSET_A_2_Z
  89    )                    /* cset_identifier_first */,
  90   (
  91    G_CSET_a_2_z
  92    "_"
  93    G_CSET_A_2_Z
  94    G_CSET_DIGITS
  95    G_CSET_LATINS
  96    G_CSET_LATINC
  97    )                    /* cset_identifier_nth */,
  98   ( "#\n" )             /* cpair_comment_single */,
  99
 100   FALSE                 /* case_sensitive */,
 101
 102   TRUE                  /* skip_comment_multi */,
 103   TRUE                  /* skip_comment_single */,
 104   TRUE                  /* scan_comment_multi */,
 105   TRUE                  /* scan_identifier */,
 106   FALSE                 /* scan_identifier_1char */,
 107   FALSE                 /* scan_identifier_NULL */,
 108   TRUE                  /* scan_symbols */,
 109   FALSE                 /* scan_binary */,
 110   TRUE                  /* scan_octal */,
 111   TRUE                  /* scan_float */,
 112   TRUE                  /* scan_hex */,
 113   FALSE                 /* scan_hex_dollar */,
 114   TRUE                  /* scan_string_sq */,
 115   TRUE                  /* scan_string_dq */,
 116   TRUE                  /* numbers_2_int */,
 117   FALSE                 /* int_2_float */,
 118   FALSE                 /* identifier_2_string */,
 119   TRUE                  /* char_2_token */,
 120   FALSE                 /* symbol_2_token */,
 121   FALSE                 /* scope_0_fallback */,
 122 };
 123
 124
 125 /* --- prototypes --- */
 126 static inline
 127 GScannerKey*    g_scanner_lookup_internal (GScanner     *scanner,
 128                                            guint         scope_id,
 129                                            const gchar  *symbol);
 130 static gboolean g_scanner_key_equal       (gconstpointer v1,
 131                                            gconstpointer v2);
 132 static guint    g_scanner_key_hash        (gconstpointer v);
 133 static void     g_scanner_get_token_ll    (GScanner     *scanner,
 134                                            GTokenType   *token_p,
 135                                            GTokenValue  *value_p,
 136                                            guint        *line_p,
 137                                            guint        *position_p);
 138 static void     g_scanner_get_token_i     (GScanner     *scanner,
 139                                            GTokenType   *token_p,
 140                                            GTokenValue  *value_p,
 141                                            guint        *line_p,
 142                                            guint        *position_p);
 143
 144 static guchar   g_scanner_peek_next_char  (GScanner     *scanner);
 145 static guchar   g_scanner_get_char        (GScanner     *scanner,
 146                                            guint        *line_p,
 147                                            guint        *position_p);
 148 static void     g_scanner_msg_handler     (GScanner     *scanner,
 149                                            gchar        *message,
 150                                            gint          is_error);
 151
 152
 153 /* --- functions --- */
 154 static inline gint
 155 g_scanner_char_2_num (guchar    c,
 156                       guchar    base)
 157 {
 158   if (c >= '0' && c <= '9')
 159     c -= '0';
 160   else if (c >= 'A' && c <= 'Z')
 161     c -= 'A' - 10;
 162   else if (c >= 'a' && c <= 'z')
 163     c -= 'a' - 10;
 164   else
 165     return -1;
 166
 167   if (c < base)
 168     return c;
 169
 170   return -1;
 171 }
 172
 173 GScanner*
 174 g_scanner_new (GScannerConfig   *config_templ)
 175 {
 176   GScanner *scanner;
 177
 178   if (!config_templ)
 179     config_templ = &g_scanner_config_template;
 180
 181   scanner = g_new0 (GScanner, 1);
 182
 183   scanner->user_data = NULL;
 184   scanner->max_parse_errors = 0;
 185   scanner->parse_errors = 0;
 186   scanner->input_name = NULL;
 187   scanner->derived_data = NULL;
 188
 189   scanner->config = g_new0 (GScannerConfig, 1);
 190
 191   scanner->config->case_sensitive       = config_templ->case_sensitive;
 192   scanner->config->cset_skip_characters = config_templ->cset_skip_characters;
 193   if (!scanner->config->cset_skip_characters)
 194     scanner->config->cset_skip_characters = "";
 195   scanner->config->cset_identifier_first= config_templ->cset_identifier_first;
 196   scanner->config->cset_identifier_nth  = config_templ->cset_identifier_nth;
 197   scanner->config->cpair_comment_single = config_templ->cpair_comment_single;
 198   scanner->config->skip_comment_multi   = config_templ->skip_comment_multi;
 199   scanner->config->skip_comment_single  = config_templ->skip_comment_single;
 200   scanner->config->scan_comment_multi   = config_templ->scan_comment_multi;
 201   scanner->config->scan_identifier      = config_templ->scan_identifier;
 202   scanner->config->scan_identifier_1char= config_templ->scan_identifier_1char;
 203   scanner->config->scan_identifier_NULL = config_templ->scan_identifier_NULL;
 204   scanner->config->scan_symbols         = config_templ->scan_symbols;
 205   scanner->config->scan_binary          = config_templ->scan_binary;
 206   scanner->config->scan_octal           = config_templ->scan_octal;
 207   scanner->config->scan_float           = config_templ->scan_float;
 208   scanner->config->scan_hex             = config_templ->scan_hex;
 209   scanner->config->scan_hex_dollar      = config_templ->scan_hex_dollar;
 210   scanner->config->scan_string_sq       = config_templ->scan_string_sq;
 211   scanner->config->scan_string_dq       = config_templ->scan_string_dq;
 212   scanner->config->numbers_2_int        = config_templ->numbers_2_int;
 213   scanner->config->int_2_float          = config_templ->int_2_float;
 214   scanner->config->identifier_2_string  = config_templ->identifier_2_string;
 215   scanner->config->char_2_token         = config_templ->char_2_token;
 216   scanner->config->symbol_2_token       = config_templ->symbol_2_token;
 217   scanner->config->scope_0_fallback     = config_templ->scope_0_fallback;
 218
 219   scanner->token = G_TOKEN_NONE;
 220   scanner->value.v_int = 0;
 221   scanner->line = 1;
 222   scanner->position = 0;
 223
 224   scanner->next_token = G_TOKEN_NONE;
 225   scanner->next_value.v_int = 0;
 226   scanner->next_line = 1;
 227   scanner->next_position = 0;
 228
 229   scanner->symbol_table = g_hash_table_new (g_scanner_key_hash, g_scanner_key_equal);
 230   scanner->input_fd = -1;
 231   scanner->text = NULL;
 232   scanner->text_end = NULL;
 233   scanner->buffer = NULL;
 234   scanner->scope_id = 0;
 235
 236   scanner->msg_handler = g_scanner_msg_handler;
 237
 238   return scanner;
 239 }
 240
 241 static inline void
 242 g_scanner_free_value (GTokenType     *token_p,
 243                       GTokenValue     *value_p)
 244 {
 245   switch (*token_p)
 246     {
 247     case G_TOKEN_STRING:
 248     case G_TOKEN_IDENTIFIER:
 249     case G_TOKEN_IDENTIFIER_NULL:
 250     case G_TOKEN_COMMENT_SINGLE:
 251     case G_TOKEN_COMMENT_MULTI:
 252       g_free (value_p->v_string);
 253       break;
 254
 255     default:
 256       break;
 257     }
 258
 259   *token_p = G_TOKEN_NONE;
 260 }
 261
 262 static void
 263 g_scanner_destroy_symbol_table_entry (gpointer _key,
 264                                       gpointer _value,
 265                                       gpointer _data)
 266 {
 267   GScannerKey *key = _key;
 268
 269   g_free (key->symbol);
 270   g_free (key);
 271 }
 272
 273 void
 274 g_scanner_destroy (GScanner     *scanner)
 275 {
 276   g_return_if_fail (scanner != NULL);
 277
 278   g_hash_table_foreach (scanner->symbol_table,
 279                         g_scanner_destroy_symbol_table_entry, NULL);
 280   g_hash_table_destroy (scanner->symbol_table);
 281   g_scanner_free_value (&scanner->token, &scanner->value);
 282   g_scanner_free_value (&scanner->next_token, &scanner->next_value);
 283   g_free (scanner->config);
 284   g_free (scanner->buffer);
 285   g_free (scanner);
 286 }
 287
 288 static void
 289 g_scanner_msg_handler (GScanner         *scanner,
 290                        gchar            *message,
 291                        gint              is_error)
 292 {
 293   g_return_if_fail (scanner != NULL);
 294
 295   fprintf (stdout, "%s:%d: ", scanner->input_name, scanner->line);
 296   if (is_error)
 297     fprintf (stdout, "error: ");
 298   fprintf (stdout, "%s\n", message);
 299 }
 300
 301 void
 302 g_scanner_error (GScanner       *scanner,
 303                  const gchar    *format,
 304                  ...)
 305 {
 306   g_return_if_fail (scanner != NULL);
 307   g_return_if_fail (format != NULL);
 308
 309   scanner->parse_errors++;
 310
 311   if (scanner->msg_handler)
 312     {
 313       va_list args;
 314       gchar *string;
 315
 316       va_start (args, format);
 317       string = g_strdup_vprintf (format, args);
 318       va_end (args);
 319
 320       scanner->msg_handler (scanner, string, TRUE);
 321
 322       g_free (string);
 323     }
 324 }
 325
 326 void
 327 g_scanner_warn (GScanner       *scanner,
 328                 const gchar    *format,
 329                 ...)
 330 {
 331   g_return_if_fail (scanner != NULL);
 332   g_return_if_fail (format != NULL);
 333
 334   if (scanner->msg_handler)
 335     {
 336       va_list args;
 337       gchar *string;
 338
 339       va_start (args, format);
 340       string = g_strdup_vprintf (format, args);
 341       va_end (args);
 342
 343       scanner->msg_handler (scanner, string, FALSE);
 344
 345       g_free (string);
 346     }
 347 }
 348
 349 static gboolean
 350 g_scanner_key_equal (gconstpointer v1,
 351                      gconstpointer v2)
 352 {
 353   const GScannerKey *key1 = v1;
 354   const GScannerKey *key2 = v2;
 355
 356   return (key1->scope_id == key2->scope_id) && (strcmp (key1->symbol, key2->symbol) == 0);
 357 }
 358
 359 static guint
 360 g_scanner_key_hash (gconstpointer v)
 361 {
 362   const GScannerKey *key = v;
 363   gchar *c;
 364   guint h;
 365
 366   h = key->scope_id;
 367   for (c = key->symbol; *c; c++)
 368     {
 369       guint g;
 370
 371       h = (h << 4) + *c;
 372       g = h & 0xf0000000;
 373       if (g)
 374         {
 375           h = h ^ (g >> 24);
 376           h = h ^ g;
 377         }
 378     }
 379
 380   return h;
 381 }
 382
 383 static inline GScannerKey*
 384 g_scanner_lookup_internal (GScanner     *scanner,
 385                            guint         scope_id,
 386                            const gchar  *symbol)
 387 {
 388   GScannerKey   *key_p;
 389   GScannerKey key;
 390
 391   key.scope_id = scope_id;
 392
 393   if (!scanner->config->case_sensitive)
 394     {
 395       gchar *d;
 396       const gchar *c;
 397
 398       key.symbol = g_new (gchar, strlen (symbol) + 1);
 399       for (d = key.symbol, c = symbol; *c; c++, d++)
 400         *d = to_lower (*c);
 401       *d = 0;
 402       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 403       g_free (key.symbol);
 404     }
 405   else
 406     {
 407       key.symbol = (gchar*) symbol;
 408       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 409     }
 410
 411   return key_p;
 412 }
 413
 414 void
 415 g_scanner_scope_add_symbol (GScanner    *scanner,
 416                             guint        scope_id,
 417                             const gchar *symbol,
 418                             gpointer     value)
 419 {
 420   GScannerKey   *key;
 421
 422   g_return_if_fail (scanner != NULL);
 423   g_return_if_fail (symbol != NULL);
 424
 425   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 426
 427   if (!key)
 428     {
 429       key = g_new (GScannerKey, 1);
 430       key->scope_id = scope_id;
 431       key->symbol = g_strdup (symbol);
 432       key->value = value;
 433       if (!scanner->config->case_sensitive)
 434         {
 435           gchar *c;
 436
 437           c = key->symbol;
 438           while (*c != 0)
 439             {
 440               *c = to_lower (*c);
 441               c++;
 442             }
 443         }
 444       g_hash_table_insert (scanner->symbol_table, key, key);
 445     }
 446   else
 447     key->value = value;
 448 }
 449
 450 void
 451 g_scanner_scope_remove_symbol (GScanner    *scanner,
 452                                guint        scope_id,
 453                                const gchar *symbol)
 454 {
 455   GScannerKey   *key;
 456
 457   g_return_if_fail (scanner != NULL);
 458   g_return_if_fail (symbol != NULL);
 459
 460   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 461
 462   if (key)
 463     {
 464       g_hash_table_remove (scanner->symbol_table, key);
 465       g_free (key->symbol);
 466       g_free (key);
 467     }
 468 }
 469
 470 gpointer
 471 g_scanner_lookup_symbol (GScanner       *scanner,
 472                          const gchar    *symbol)
 473 {
 474   GScannerKey   *key;
 475   guint scope_id;
 476
 477   g_return_val_if_fail (scanner != NULL, NULL);
 478
 479   if (!symbol)
 480     return NULL;
 481
 482   scope_id = scanner->scope_id;
 483   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 484   if (!key && scope_id && scanner->config->scope_0_fallback)
 485     key = g_scanner_lookup_internal (scanner, 0, symbol);
 486
 487   if (key)
 488     return key->value;
 489   else
 490     return NULL;
 491 }
 492
 493 gpointer
 494 g_scanner_scope_lookup_symbol (GScanner       *scanner,
 495                                guint           scope_id,
 496                                const gchar    *symbol)
 497 {
 498   GScannerKey   *key;
 499
 500   g_return_val_if_fail (scanner != NULL, NULL);
 501
 502   if (!symbol)
 503     return NULL;
 504
 505   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 506
 507   if (key)
 508     return key->value;
 509   else
 510     return NULL;
 511 }
 512
 513 guint
 514 g_scanner_set_scope (GScanner       *scanner,
 515                      guint           scope_id)
 516 {
 517   guint old_scope_id;
 518
 519   g_return_val_if_fail (scanner != NULL, 0);
 520
 521   old_scope_id = scanner->scope_id;
 522   scanner->scope_id = scope_id;
 523
 524   return old_scope_id;
 525 }
 526
 527 static void
 528 g_scanner_foreach_internal (gpointer  _key,
 529                             gpointer  _value,
 530                             gpointer  _user_data)
 531 {
 532   GScannerKey *key;
 533   gpointer *d;
 534   GHFunc func;
 535   gpointer user_data;
 536   guint *scope_id;
 537
 538   d = _user_data;
 539   func = (GHFunc) d[0];
 540   user_data = d[1];
 541   scope_id = d[2];
 542   key = _value;
 543
 544   if (key->scope_id == *scope_id)
 545     func (key->symbol, key->value, user_data);
 546 }
 547
 548 void
 549 g_scanner_scope_foreach_symbol (GScanner       *scanner,
 550                                 guint           scope_id,
 551                                 GHFunc          func,
 552                                 gpointer        user_data)
 553 {
 554   gpointer d[3];
 555
 556   g_return_if_fail (scanner != NULL);
 557
 558   d[0] = (gpointer) func;
 559   d[1] = user_data;
 560   d[2] = &scope_id;
 561
 562   g_hash_table_foreach (scanner->symbol_table, g_scanner_foreach_internal, d);
 563 }
 564
 565 void
 566 g_scanner_freeze_symbol_table (GScanner *scanner)
 567 {
 568 #ifdef G_ENABLE_DEBUG
 569   static gboolean first_call = TRUE;
 570
 571   if (first_call)
 572     {
 573       g_message ("g_scanner_freeze_symbol_table() and "
 574                  "g_scanner_thaw_symbol_table() are deprecated.");
 575       first_call = FALSE;
 576     }
 577 #endif /* G_ENABLE_DEBUG */
 578 }
 579
 580 void
 581 g_scanner_thaw_symbol_table (GScanner *scanner)
 582 {
 583   g_scanner_freeze_symbol_table (scanner);
 584 }
 585
 586 GTokenType
 587 g_scanner_peek_next_token (GScanner     *scanner)
 588 {
 589   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 590
 591   if (scanner->next_token == G_TOKEN_NONE)
 592     {
 593       scanner->next_line = scanner->line;
 594       scanner->next_position = scanner->position;
 595       g_scanner_get_token_i (scanner,
 596                              &scanner->next_token,
 597                              &scanner->next_value,
 598                              &scanner->next_line,
 599                              &scanner->next_position);
 600     }
 601
 602   return scanner->next_token;
 603 }
 604
 605 GTokenType
 606 g_scanner_get_next_token (GScanner      *scanner)
 607 {
 608   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 609
 610   if (scanner->next_token != G_TOKEN_NONE)
 611     {
 612       g_scanner_free_value (&scanner->token, &scanner->value);
 613
 614       scanner->token = scanner->next_token;
 615       scanner->value = scanner->next_value;
 616       scanner->line = scanner->next_line;
 617       scanner->position = scanner->next_position;
 618       scanner->next_token = G_TOKEN_NONE;
 619     }
 620   else
 621     g_scanner_get_token_i (scanner,
 622                            &scanner->token,
 623                            &scanner->value,
 624                            &scanner->line,
 625                            &scanner->position);
 626
 627   return scanner->token;
 628 }
 629
 630 GTokenType
 631 g_scanner_cur_token (GScanner *scanner)
 632 {
 633   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 634
 635   return scanner->token;
 636 }
 637
 638 GTokenValue
 639 g_scanner_cur_value (GScanner *scanner)
 640 {
 641   GTokenValue v;
 642
 643   v.v_int = 0;
 644
 645   g_return_val_if_fail (scanner != NULL, v);
 646
 647   /* MSC isn't capable of handling return scanner->value; ? */
 648
 649   v = scanner->value;
 650
 651   return v;
 652 }
 653
 654 guint
 655 g_scanner_cur_line (GScanner *scanner)
 656 {
 657   g_return_val_if_fail (scanner != NULL, 0);
 658
 659   return scanner->line;
 660 }
 661
 662 guint
 663 g_scanner_cur_position (GScanner *scanner)
 664 {
 665   g_return_val_if_fail (scanner != NULL, 0);
 666
 667   return scanner->position;
 668 }
 669
 670 gboolean
 671 g_scanner_eof (GScanner *scanner)
 672 {
 673   g_return_val_if_fail (scanner != NULL, TRUE);
 674
 675   return scanner->token == G_TOKEN_EOF;
 676 }
 677
 678 void
 679 g_scanner_input_file (GScanner *scanner,
 680                       gint      input_fd)
 681 {
 682   g_return_if_fail (scanner != NULL);
 683   g_return_if_fail (input_fd >= 0);
 684
 685   if (scanner->input_fd >= 0)
 686     g_scanner_sync_file_offset (scanner);
 687
 688   scanner->token = G_TOKEN_NONE;
 689   scanner->value.v_int = 0;
 690   scanner->line = 1;
 691   scanner->position = 0;
 692   scanner->next_token = G_TOKEN_NONE;
 693
 694   scanner->input_fd = input_fd;
 695   scanner->text = NULL;
 696   scanner->text_end = NULL;
 697
 698   if (!scanner->buffer)
 699     scanner->buffer = g_new (gchar, READ_BUFFER_SIZE + 1);
 700 }
 701
 702 void
 703 g_scanner_input_text (GScanner    *scanner,
 704                       const gchar *text,
 705                       guint        text_len)
 706 {
 707   g_return_if_fail (scanner != NULL);
 708   if (text_len)
 709     g_return_if_fail (text != NULL);
 710   else
 711     text = NULL;
 712
 713   if (scanner->input_fd >= 0)
 714     g_scanner_sync_file_offset (scanner);
 715
 716   scanner->token = G_TOKEN_NONE;
 717   scanner->value.v_int = 0;
 718   scanner->line = 1;
 719   scanner->position = 0;
 720   scanner->next_token = G_TOKEN_NONE;
 721
 722   scanner->input_fd = -1;
 723   scanner->text = text;
 724   scanner->text_end = text + text_len;
 725
 726   if (scanner->buffer)
 727     {
 728       g_free (scanner->buffer);
 729       scanner->buffer = NULL;
 730     }
 731 }
 732
 733 static guchar
 734 g_scanner_peek_next_char (GScanner *scanner)
 735 {
 736   if (scanner->text < scanner->text_end)
 737     {
 738       return *scanner->text;
 739     }
 740   else if (scanner->input_fd >= 0)
 741     {
 742       gint count;
 743       gchar *buffer;
 744
 745       buffer = scanner->buffer;
 746       do
 747         {
 748           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 749         }
 750       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 751
 752       if (count < 1)
 753         {
 754           scanner->input_fd = -1;
 755
 756           return 0;
 757         }
 758       else
 759         {
 760           scanner->text = buffer;
 761           scanner->text_end = buffer + count;
 762
 763           return *buffer;
 764         }
 765     }
 766   else
 767     return 0;
 768 }
 769
 770 void
 771 g_scanner_sync_file_offset (GScanner *scanner)
 772 {
 773   g_return_if_fail (scanner != NULL);
 774
 775   /* for file input, rewind the filedescriptor to the current
 776    * buffer position and blow the file read ahead buffer. usefull for
 777    * third party uses of our filedescriptor, which hooks onto the current
 778    * scanning position.
 779    */
 780
 781   if (scanner->input_fd >= 0 && scanner->text_end > scanner->text)
 782     {
 783       gint buffered;
 784
 785       buffered = scanner->text_end - scanner->text;
 786       if (lseek (scanner->input_fd, - buffered, SEEK_CUR) >= 0)
 787         {
 788           /* we succeeded, blow our buffer's contents now */
 789           scanner->text = NULL;
 790           scanner->text_end = NULL;
 791         }
 792       else
 793         errno = 0;
 794     }
 795 }
 796
 797 static guchar
 798 g_scanner_get_char (GScanner    *scanner,
 799                     guint       *line_p,
 800                     guint       *position_p)
 801 {
 802   guchar fchar;
 803
 804   if (scanner->text < scanner->text_end)
 805     fchar = *(scanner->text++);
 806   else if (scanner->input_fd >= 0)
 807     {
 808       gint count;
 809       gchar *buffer;
 810
 811       buffer = scanner->buffer;
 812       do
 813         {
 814           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 815         }
 816       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 817
 818       if (count < 1)
 819         {
 820           scanner->input_fd = -1;
 821           fchar = 0;
 822         }
 823       else
 824         {
 825           scanner->text = buffer + 1;
 826           scanner->text_end = buffer + count;
 827           fchar = *buffer;
 828           if (!fchar)
 829             {
 830               g_scanner_sync_file_offset (scanner);
 831               scanner->text_end = scanner->text;
 832               scanner->input_fd = -1;
 833             }
 834         }
 835     }
 836   else
 837     fchar = 0;
 838
 839   if (fchar == '\n')
 840     {
 841       (*position_p) = 0;
 842       (*line_p)++;
 843     }
 844   else if (fchar)
 845     {
 846       (*position_p)++;
 847     }
 848
 849   return fchar;
 850 }
 851
 852 void
 853 g_scanner_unexp_token (GScanner         *scanner,
 854                        GTokenType        expected_token,
 855                        const gchar      *identifier_spec,
 856                        const gchar      *symbol_spec,
 857                        const gchar      *symbol_name,
 858                        const gchar      *message,
 859                        gint              is_error)
 860 {
 861   gchar *token_string;
 862   guint token_string_len;
 863   gchar *expected_string;
 864   guint expected_string_len;
 865   gchar *message_prefix;
 866   gboolean print_unexp;
 867   void (*msg_handler)   (GScanner*, const gchar*, ...);
 868
 869   g_return_if_fail (scanner != NULL);
 870
 871   if (is_error)
 872     msg_handler = g_scanner_error;
 873   else
 874     msg_handler = g_scanner_warn;
 875
 876   if (!identifier_spec)
 877     identifier_spec = "identifier";
 878   if (!symbol_spec)
 879     symbol_spec = "symbol";
 880
 881   token_string_len = 56;
 882   token_string = g_new (gchar, token_string_len + 1);
 883   expected_string_len = 64;
 884   expected_string = g_new (gchar, expected_string_len + 1);
 885   print_unexp = TRUE;
 886
 887   switch (scanner->token)
 888     {
 889     case G_TOKEN_EOF:
 890       g_snprintf (token_string, token_string_len, "end of file");
 891       break;
 892
 893     default:
 894       if (scanner->token >= 1 && scanner->token <= 255)
 895         {
 896           if ((scanner->token >= ' ' && scanner->token <= '~') ||
 897               strchr (scanner->config->cset_identifier_first, scanner->token) ||
 898               strchr (scanner->config->cset_identifier_nth, scanner->token))
 899             g_snprintf (token_string, expected_string_len, "character `%c'", scanner->token);
 900           else
 901             g_snprintf (token_string, expected_string_len, "character `\\%o'", scanner->token);
 902           break;
 903         }
 904       else if (!scanner->config->symbol_2_token)
 905         {
 906           g_snprintf (token_string, token_string_len, "(unknown) token <%d>", scanner->token);
 907           break;
 908         }
 909       /* fall through */
 910     case G_TOKEN_SYMBOL:
 911       if (expected_token == G_TOKEN_SYMBOL ||
 912           (scanner->config->symbol_2_token &&
 913            expected_token > G_TOKEN_LAST))
 914         print_unexp = FALSE;
 915       if (symbol_name)
 916         g_snprintf (token_string,
 917                     token_string_len,
 918                     "%s%s `%s'",
 919                     print_unexp ? "" : "invalid ",
 920                     symbol_spec,
 921                     symbol_name);
 922       else
 923         g_snprintf (token_string,
 924                     token_string_len,
 925                     "%s%s",
 926                     print_unexp ? "" : "invalid ",
 927                     symbol_spec);
 928       break;
 929
 930     case G_TOKEN_ERROR:
 931       print_unexp = FALSE;
 932       expected_token = G_TOKEN_NONE;
 933       switch (scanner->value.v_error)
 934         {
 935         case G_ERR_UNEXP_EOF:
 936           g_snprintf (token_string, token_string_len, "scanner: unexpected end of file");
 937           break;
 938
 939         case G_ERR_UNEXP_EOF_IN_STRING:
 940           g_snprintf (token_string, token_string_len, "scanner: unterminated string constant");
 941           break;
 942
 943         case G_ERR_UNEXP_EOF_IN_COMMENT:
 944           g_snprintf (token_string, token_string_len, "scanner: unterminated comment");
 945           break;
 946
 947         case G_ERR_NON_DIGIT_IN_CONST:
 948           g_snprintf (token_string, token_string_len, "scanner: non digit in constant");
 949           break;
 950
 951         case G_ERR_FLOAT_RADIX:
 952           g_snprintf (token_string, token_string_len, "scanner: invalid radix for floating constant");
 953           break;
 954
 955         case G_ERR_FLOAT_MALFORMED:
 956           g_snprintf (token_string, token_string_len, "scanner: malformed floating constant");
 957           break;
 958
 959         case G_ERR_DIGIT_RADIX:
 960           g_snprintf (token_string, token_string_len, "scanner: digit is beyond radix");
 961           break;
 962
 963         case G_ERR_UNKNOWN:
 964         default:
 965           g_snprintf (token_string, token_string_len, "scanner: unknown error");
 966           break;
 967         }
 968       break;
 969
 970     case G_TOKEN_CHAR:
 971       g_snprintf (token_string, token_string_len, "character `%c'", scanner->value.v_char);
 972       break;
 973
 974     case G_TOKEN_IDENTIFIER:
 975     case G_TOKEN_IDENTIFIER_NULL:
 976       if (expected_token == G_TOKEN_IDENTIFIER ||
 977           expected_token == G_TOKEN_IDENTIFIER_NULL)
 978         print_unexp = FALSE;
 979       g_snprintf (token_string,
 980                   token_string_len,
 981                   "%s%s `%s'",
 982                   print_unexp ? "" : "invalid ",
 983                   identifier_spec,
 984                   scanner->token == G_TOKEN_IDENTIFIER ? scanner->value.v_string : "null");
 985       break;
 986
 987     case G_TOKEN_BINARY:
 988     case G_TOKEN_OCTAL:
 989     case G_TOKEN_INT:
 990     case G_TOKEN_HEX:
 991       g_snprintf (token_string, token_string_len, "number `%ld'", scanner->value.v_int);
 992       break;
 993
 994     case G_TOKEN_FLOAT:
 995       g_snprintf (token_string, token_string_len, "number `%.3f'", scanner->value.v_float);
 996       break;
 997
 998     case G_TOKEN_STRING:
 999       if (expected_token == G_TOKEN_STRING)
1000         print_unexp = FALSE;
1001       g_snprintf (token_string,
1002                   token_string_len,
1003                   "%s%sstring constant \"%s\"",
1004                   print_unexp ? "" : "invalid ",
1005                   scanner->value.v_string[0] == 0 ? "empty " : "",
1006                   scanner->value.v_string);
1007       token_string[token_string_len - 2] = '"';
1008       token_string[token_string_len - 1] = 0;
1009       break;
1010
1011     case G_TOKEN_COMMENT_SINGLE:
1012     case G_TOKEN_COMMENT_MULTI:
1013       g_snprintf (token_string, token_string_len, "comment");
1014       break;
1015
1016     case G_TOKEN_NONE:
1017       /* somehow the user's parsing code is screwed, there isn't much
1018        * we can do about it.
1019        * Note, a common case to trigger this is
1020        * g_scanner_peek_next_token(); g_scanner_unexp_token();
1021        * without an intermediate g_scanner_get_next_token().
1022        */
1023       g_assert_not_reached ();
1024       break;
1025     }
1026
1027
1028   switch (expected_token)
1029     {
1030       gboolean need_valid;
1031
1032     default:
1033       if (expected_token >= 1 && expected_token <= 255)
1034         {
1035           if ((expected_token >= ' ' && expected_token <= '~') ||
1036               strchr (scanner->config->cset_identifier_first, expected_token) ||
1037               strchr (scanner->config->cset_identifier_nth, expected_token))
1038             g_snprintf (expected_string, expected_string_len, "character `%c'", expected_token);
1039           else
1040             g_snprintf (expected_string, expected_string_len, "character `\\%o'", expected_token);
1041           break;
1042         }
1043       else if (!scanner->config->symbol_2_token)
1044         {
1045           g_snprintf (expected_string, expected_string_len, "(unknown) token <%d>", expected_token);
1046           break;
1047         }
1048       /* fall through */
1049     case G_TOKEN_SYMBOL:
1050       need_valid = (scanner->token == G_TOKEN_SYMBOL ||
1051                     (scanner->config->symbol_2_token &&
1052                      scanner->token > G_TOKEN_LAST));
1053       g_snprintf (expected_string,
1054                   expected_string_len,
1055                   "%s%s",
1056                   need_valid ? "valid " : "",
1057                   symbol_spec);
1058       /* FIXME: should we attempt to lookup the symbol_name for symbol_2_token? */
1059       break;
1060
1061     case G_TOKEN_INT:
1062       g_snprintf (expected_string, expected_string_len, "number (integer)");
1063       break;
1064
1065     case G_TOKEN_FLOAT:
1066       g_snprintf (expected_string, expected_string_len, "number (float)");
1067       break;
1068
1069     case G_TOKEN_STRING:
1070       g_snprintf (expected_string,
1071                   expected_string_len,
1072                   "%sstring constant",
1073                   scanner->token == G_TOKEN_STRING ? "valid " : "");
1074       break;
1075
1076     case G_TOKEN_IDENTIFIER:
1077     case G_TOKEN_IDENTIFIER_NULL:
1078       g_snprintf (expected_string,
1079                   expected_string_len,
1080                   "%s%s",
1081                   (scanner->token == G_TOKEN_IDENTIFIER_NULL ||
1082                    scanner->token == G_TOKEN_IDENTIFIER ? "valid " : ""),
1083                   identifier_spec);
1084       break;
1085
1086     case G_TOKEN_EOF:
1087       g_snprintf (expected_string, expected_string_len, "end of file");
1088       break;
1089
1090     case G_TOKEN_NONE:
1091       break;
1092     }
1093
1094   if (message && message[0] != 0)
1095     message_prefix = " - ";
1096   else
1097     {
1098       message_prefix = "";
1099       message = "";
1100     }
1101
1102   if (expected_token != G_TOKEN_NONE)
1103     {
1104       if (print_unexp)
1105         msg_handler (scanner,
1106                      "unexpected %s, expected %s%s%s",
1107                      token_string,
1108                      expected_string,
1109                      message_prefix,
1110                      message);
1111       else
1112         msg_handler (scanner,
1113                      "%s, expected %s%s%s",
1114                      token_string,
1115                      expected_string,
1116                      message_prefix,
1117                      message);
1118     }
1119   else
1120     {
1121       if (print_unexp)
1122         msg_handler (scanner,
1123                      "unexpected %s%s%s",
1124                      token_string,
1125                      message_prefix,
1126                      message);
1127       else
1128         msg_handler (scanner,
1129                      "%s%s%s",
1130                      token_string,
1131                      message_prefix,
1132                      message);
1133     }
1134
1135   g_free (token_string);
1136   g_free (expected_string);
1137 }
1138
1139 gint
1140 g_scanner_stat_mode (const gchar *filename)
1141 {
1142   struct stat  *stat_buf;
1143   gint          st_mode;
1144
1145   stat_buf = g_new0 (struct stat, 1);
1146 #ifdef HAVE_LSTAT
1147   lstat (filename, stat_buf);
1148 #else
1149   stat (filename, stat_buf);
1150 #endif
1151   st_mode = stat_buf->st_mode;
1152
1153   g_free (stat_buf);
1154
1155   return st_mode;
1156 }
1157
1158 static void
1159 g_scanner_get_token_i (GScanner *scanner,
1160                        GTokenType       *token_p,
1161                        GTokenValue      *value_p,
1162                        guint            *line_p,
1163                        guint            *position_p)
1164 {
1165   do
1166     {
1167       g_scanner_free_value (token_p, value_p);
1168       g_scanner_get_token_ll (scanner, token_p, value_p, line_p, position_p);
1169     }
1170   while (((*token_p > 0 && *token_p < 256) &&
1171           strchr (scanner->config->cset_skip_characters, *token_p)) ||
1172          (*token_p == G_TOKEN_CHAR &&
1173           strchr (scanner->config->cset_skip_characters, value_p->v_char)) ||
1174          (*token_p == G_TOKEN_COMMENT_MULTI &&
1175           scanner->config->skip_comment_multi) ||
1176          (*token_p == G_TOKEN_COMMENT_SINGLE &&
1177           scanner->config->skip_comment_single));
1178
1179   switch (*token_p)
1180     {
1181     case G_TOKEN_IDENTIFIER:
1182       if (scanner->config->identifier_2_string)
1183         *token_p = G_TOKEN_STRING;
1184       break;
1185
1186     case G_TOKEN_SYMBOL:
1187       if (scanner->config->symbol_2_token)
1188         *token_p = (GTokenType) value_p->v_symbol;
1189       break;
1190
1191     case G_TOKEN_BINARY:
1192     case G_TOKEN_OCTAL:
1193     case G_TOKEN_HEX:
1194       if (scanner->config->numbers_2_int)
1195         *token_p = G_TOKEN_INT;
1196       break;
1197
1198     default:
1199       break;
1200     }
1201
1202   if (*token_p == G_TOKEN_INT &&
1203       scanner->config->int_2_float)
1204     {
1205       *token_p = G_TOKEN_FLOAT;
1206       value_p->v_float = value_p->v_int;
1207     }
1208
1209   errno = 0;
1210 }
1211
1212 static void
1213 g_scanner_get_token_ll  (GScanner       *scanner,
1214                          GTokenType     *token_p,
1215                          GTokenValue    *value_p,
1216                          guint          *line_p,
1217                          guint          *position_p)
1218 {
1219   GScannerConfig *config;
1220   GTokenType       token;
1221   gboolean         in_comment_multi;
1222   gboolean         in_comment_single;
1223   gboolean         in_string_sq;
1224   gboolean         in_string_dq;
1225   GString         *gstring;
1226   GTokenValue      value;
1227   guchar           ch;
1228
1229   config = scanner->config;
1230   (*value_p).v_int = 0;
1231
1232   if ((scanner->text >= scanner->text_end && scanner->input_fd < 0) ||
1233       scanner->token == G_TOKEN_EOF)
1234     {
1235       *token_p = G_TOKEN_EOF;
1236       return;
1237     }
1238
1239   in_comment_multi = FALSE;
1240   in_comment_single = FALSE;
1241   in_string_sq = FALSE;
1242   in_string_dq = FALSE;
1243   gstring = NULL;
1244
1245   do /* while (ch != 0) */
1246     {
1247       gboolean dotted_float = FALSE;
1248
1249       ch = g_scanner_get_char (scanner, line_p, position_p);
1250
1251       value.v_int = 0;
1252       token = G_TOKEN_NONE;
1253
1254       /* this is *evil*, but needed ;(
1255        * we first check for identifier first character, because  it
1256        * might interfere with other key chars like slashes or numbers
1257        */
1258       if (config->scan_identifier &&
1259           ch && strchr (config->cset_identifier_first, ch))
1260         goto identifier_precedence;
1261
1262       switch (ch)
1263         {
1264         case 0:
1265           token = G_TOKEN_EOF;
1266           (*position_p)++;
1267           /* ch = 0; */
1268           break;
1269
1270         case '/':
1271           if (!config->scan_comment_multi ||
1272               g_scanner_peek_next_char (scanner) != '*')
1273             goto default_case;
1274           g_scanner_get_char (scanner, line_p, position_p);
1275           token = G_TOKEN_COMMENT_MULTI;
1276           in_comment_multi = TRUE;
1277           gstring = g_string_new ("");
1278           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1279             {
1280               if (ch == '*' && g_scanner_peek_next_char (scanner) == '/')
1281                 {
1282                   g_scanner_get_char (scanner, line_p, position_p);
1283                   in_comment_multi = FALSE;
1284                   break;
1285                 }
1286               else
1287                 gstring = g_string_append_c (gstring, ch);
1288             }
1289           ch = 0;
1290           break;
1291
1292         case '\'':
1293           if (!config->scan_string_sq)
1294             goto default_case;
1295           token = G_TOKEN_STRING;
1296           in_string_sq = TRUE;
1297           gstring = g_string_new ("");
1298           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1299             {
1300               if (ch == '\'')
1301                 {
1302                   in_string_sq = FALSE;
1303                   break;
1304                 }
1305               else
1306                 gstring = g_string_append_c (gstring, ch);
1307             }
1308           ch = 0;
1309           break;
1310
1311         case '"':
1312           if (!config->scan_string_dq)
1313             goto default_case;
1314           token = G_TOKEN_STRING;
1315           in_string_dq = TRUE;
1316           gstring = g_string_new ("");
1317           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1318             {
1319               if (ch == '"')
1320                 {
1321                   in_string_dq = FALSE;
1322                   break;
1323                 }
1324               else
1325                 {
1326                   if (ch == '\\')
1327                     {
1328                       ch = g_scanner_get_char (scanner, line_p, position_p);
1329                       switch (ch)
1330                         {
1331                           guint i;
1332                           guint fchar;
1333
1334                         case 0:
1335                           break;
1336
1337                         case '\\':
1338                           gstring = g_string_append_c (gstring, '\\');
1339                           break;
1340
1341                         case 'n':
1342                           gstring = g_string_append_c (gstring, '\n');
1343                           break;
1344
1345                         case 't':
1346                           gstring = g_string_append_c (gstring, '\t');
1347                           break;
1348
1349                         case 'r':
1350                           gstring = g_string_append_c (gstring, '\r');
1351                           break;
1352
1353                         case 'b':
1354                           gstring = g_string_append_c (gstring, '\b');
1355                           break;
1356
1357                         case 'f':
1358                           gstring = g_string_append_c (gstring, '\f');
1359                           break;
1360
1361                         case '0':
1362                         case '1':
1363                         case '2':
1364                         case '3':
1365                         case '4':
1366                         case '5':
1367                         case '6':
1368                         case '7':
1369                           i = ch - '0';
1370                           fchar = g_scanner_peek_next_char (scanner);
1371                           if (fchar >= '0' && fchar <= '7')
1372                             {
1373                               ch = g_scanner_get_char (scanner, line_p, position_p);
1374                               i = i * 8 + ch - '0';
1375                               fchar = g_scanner_peek_next_char (scanner);
1376                               if (fchar >= '0' && fchar <= '7')
1377                                 {
1378                                   ch = g_scanner_get_char (scanner, line_p, position_p);
1379                                   i = i * 8 + ch - '0';
1380                                 }
1381                             }
1382                           gstring = g_string_append_c (gstring, i);
1383                           break;
1384
1385                         default:
1386                           gstring = g_string_append_c (gstring, ch);
1387                           break;
1388                         }
1389                     }
1390                   else
1391                     gstring = g_string_append_c (gstring, ch);
1392                 }
1393             }
1394           ch = 0;
1395           break;
1396
1397         case '.':
1398           if (!config->scan_float)
1399             goto default_case;
1400           token = G_TOKEN_FLOAT;
1401           dotted_float = TRUE;
1402           ch = g_scanner_get_char (scanner, line_p, position_p);
1403           goto number_parsing;
1404
1405         case '$':
1406           if (!config->scan_hex_dollar)
1407             goto default_case;
1408           token = G_TOKEN_HEX;
1409           ch = g_scanner_get_char (scanner, line_p, position_p);
1410           goto number_parsing;
1411
1412         case '0':
1413           if (config->scan_octal)
1414             token = G_TOKEN_OCTAL;
1415           else
1416             token = G_TOKEN_INT;
1417           ch = g_scanner_peek_next_char (scanner);
1418           if (config->scan_hex && (ch == 'x' || ch == 'X'))
1419             {
1420               token = G_TOKEN_HEX;
1421               g_scanner_get_char (scanner, line_p, position_p);
1422               ch = g_scanner_get_char (scanner, line_p, position_p);
1423               if (ch == 0)
1424                 {
1425                   token = G_TOKEN_ERROR;
1426                   value.v_error = G_ERR_UNEXP_EOF;
1427                   (*position_p)++;
1428                   break;
1429                 }
1430               if (g_scanner_char_2_num (ch, 16) < 0)
1431                 {
1432                   token = G_TOKEN_ERROR;
1433                   value.v_error = G_ERR_DIGIT_RADIX;
1434                   ch = 0;
1435                   break;
1436                 }
1437             }
1438           else if (config->scan_binary && (ch == 'b' || ch == 'B'))
1439             {
1440               token = G_TOKEN_BINARY;
1441               g_scanner_get_char (scanner, line_p, position_p);
1442               ch = g_scanner_get_char (scanner, line_p, position_p);
1443               if (ch == 0)
1444                 {
1445                   token = G_TOKEN_ERROR;
1446                   value.v_error = G_ERR_UNEXP_EOF;
1447                   (*position_p)++;
1448                   break;
1449                 }
1450               if (g_scanner_char_2_num (ch, 10) < 0)
1451                 {
1452                   token = G_TOKEN_ERROR;
1453                   value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1454                   ch = 0;
1455                   break;
1456                 }
1457             }
1458           else
1459             ch = '0';
1460           /* fall through */
1461         case '1':
1462         case '2':
1463         case '3':
1464         case '4':
1465         case '5':
1466         case '6':
1467         case '7':
1468         case '8':
1469         case '9':
1470         number_parsing:
1471         {
1472           gboolean in_number = TRUE;
1473           gchar *endptr;
1474
1475           if (token == G_TOKEN_NONE)
1476             token = G_TOKEN_INT;
1477
1478           gstring = g_string_new (dotted_float ? "0." : "");
1479           gstring = g_string_append_c (gstring, ch);
1480
1481           do /* while (in_number) */
1482             {
1483               gboolean is_E;
1484
1485               is_E = token == G_TOKEN_FLOAT && (ch == 'e' || ch == 'E');
1486
1487               ch = g_scanner_peek_next_char (scanner);
1488
1489               if (g_scanner_char_2_num (ch, 36) >= 0 ||
1490                   (config->scan_float && ch == '.') ||
1491                   (is_E && (ch == '+' || ch == '-')))
1492                 {
1493                   ch = g_scanner_get_char (scanner, line_p, position_p);
1494
1495                   switch (ch)
1496                     {
1497                     case '.':
1498                       if (token != G_TOKEN_INT && token != G_TOKEN_OCTAL)
1499                         {
1500                           value.v_error = token == G_TOKEN_FLOAT ? G_ERR_FLOAT_MALFORMED : G_ERR_FLOAT_RADIX;
1501                           token = G_TOKEN_ERROR;
1502                           in_number = FALSE;
1503                         }
1504                       else
1505                         {
1506                           token = G_TOKEN_FLOAT;
1507                           gstring = g_string_append_c (gstring, ch);
1508                         }
1509                       break;
1510
1511                     case '0':
1512                     case '1':
1513                     case '2':
1514                     case '3':
1515                     case '4':
1516                     case '5':
1517                     case '6':
1518                     case '7':
1519                     case '8':
1520                     case '9':
1521                       gstring = g_string_append_c (gstring, ch);
1522                       break;
1523
1524                     case '-':
1525                     case '+':
1526                       if (token != G_TOKEN_FLOAT)
1527                         {
1528                           token = G_TOKEN_ERROR;
1529                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1530                           in_number = FALSE;
1531                         }
1532                       else
1533                         gstring = g_string_append_c (gstring, ch);
1534                       break;
1535
1536                     case 'e':
1537                     case 'E':
1538                       if ((token != G_TOKEN_HEX && !config->scan_float) ||
1539                           (token != G_TOKEN_HEX &&
1540                            token != G_TOKEN_OCTAL &&
1541                            token != G_TOKEN_FLOAT &&
1542                            token != G_TOKEN_INT))
1543                         {
1544                           token = G_TOKEN_ERROR;
1545                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1546                           in_number = FALSE;
1547                         }
1548                       else
1549                         {
1550                           if (token != G_TOKEN_HEX)
1551                             token = G_TOKEN_FLOAT;
1552                           gstring = g_string_append_c (gstring, ch);
1553                         }
1554                       break;
1555
1556                     default:
1557                       if (token != G_TOKEN_HEX)
1558                         {
1559                           token = G_TOKEN_ERROR;
1560                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1561                           in_number = FALSE;
1562                         }
1563                       else
1564                         gstring = g_string_append_c (gstring, ch);
1565                       break;
1566                     }
1567                 }
1568               else
1569                 in_number = FALSE;
1570             }
1571           while (in_number);
1572
1573           endptr = NULL;
1574           switch (token)
1575             {
1576             case G_TOKEN_BINARY:
1577               value.v_binary = strtol (gstring->str, &endptr, 2);
1578               break;
1579
1580             case G_TOKEN_OCTAL:
1581               value.v_octal = strtol (gstring->str, &endptr, 8);
1582               break;
1583
1584             case G_TOKEN_INT:
1585               value.v_int = strtol (gstring->str, &endptr, 10);
1586               break;
1587
1588             case G_TOKEN_FLOAT:
1589               value.v_float = g_strtod (gstring->str, &endptr);
1590               break;
1591
1592             case G_TOKEN_HEX:
1593               value.v_hex = strtol (gstring->str, &endptr, 16);
1594               break;
1595
1596             default:
1597               break;
1598             }
1599           if (endptr && *endptr)
1600             {
1601               token = G_TOKEN_ERROR;
1602               if (*endptr == 'e' || *endptr == 'E')
1603                 value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1604               else
1605                 value.v_error = G_ERR_DIGIT_RADIX;
1606             }
1607           g_string_free (gstring, TRUE);
1608           gstring = NULL;
1609           ch = 0;
1610         } /* number_parsing:... */
1611         break;
1612
1613         default:
1614         default_case:
1615         {
1616           if (config->cpair_comment_single &&
1617               ch == config->cpair_comment_single[0])
1618             {
1619               token = G_TOKEN_COMMENT_SINGLE;
1620               in_comment_single = TRUE;
1621               gstring = g_string_new ("");
1622               ch = g_scanner_get_char (scanner, line_p, position_p);
1623               while (ch != 0)
1624                 {
1625                   if (ch == config->cpair_comment_single[1])
1626                     {
1627                       in_comment_single = FALSE;
1628                       ch = 0;
1629                       break;
1630                     }
1631
1632                   gstring = g_string_append_c (gstring, ch);
1633                   ch = g_scanner_get_char (scanner, line_p, position_p);
1634                 }
1635             }
1636           else if (config->scan_identifier && ch &&
1637                    strchr (config->cset_identifier_first, ch))
1638             {
1639             identifier_precedence:
1640
1641               if (config->cset_identifier_nth && ch &&
1642                   strchr (config->cset_identifier_nth,
1643                           g_scanner_peek_next_char (scanner)))
1644                 {
1645                   token = G_TOKEN_IDENTIFIER;
1646                   gstring = g_string_new ("");
1647                   gstring = g_string_append_c (gstring, ch);
1648                   do
1649                     {
1650                       ch = g_scanner_get_char (scanner, line_p, position_p);
1651                       gstring = g_string_append_c (gstring, ch);
1652                       ch = g_scanner_peek_next_char (scanner);
1653                     }
1654                   while (ch && strchr (config->cset_identifier_nth, ch));
1655                   ch = 0;
1656                 }
1657               else if (config->scan_identifier_1char)
1658                 {
1659                   token = G_TOKEN_IDENTIFIER;
1660                   value.v_identifier = g_new0 (gchar, 2);
1661                   value.v_identifier[0] = ch;
1662                   ch = 0;
1663                 }
1664             }
1665           if (ch)
1666             {
1667               if (config->char_2_token)
1668                 token = ch;
1669               else
1670                 {
1671                   token = G_TOKEN_CHAR;
1672                   value.v_char = ch;
1673                 }
1674               ch = 0;
1675             }
1676         } /* default_case:... */
1677         break;
1678         }
1679       g_assert (ch == 0 && token != G_TOKEN_NONE); /* paranoid */
1680     }
1681   while (ch != 0);
1682
1683   if (in_comment_multi || in_comment_single ||
1684       in_string_sq || in_string_dq)
1685     {
1686       token = G_TOKEN_ERROR;
1687       if (gstring)
1688         {
1689           g_string_free (gstring, TRUE);
1690           gstring = NULL;
1691         }
1692       (*position_p)++;
1693       if (in_comment_multi || in_comment_single)
1694         value.v_error = G_ERR_UNEXP_EOF_IN_COMMENT;
1695       else /* (in_string_sq || in_string_dq) */
1696         value.v_error = G_ERR_UNEXP_EOF_IN_STRING;
1697     }
1698
1699   if (gstring)
1700     {
1701       value.v_string = gstring->str;
1702       g_string_free (gstring, FALSE);
1703       gstring = NULL;
1704     }
1705
1706   if (token == G_TOKEN_IDENTIFIER)
1707     {
1708       if (config->scan_symbols)
1709         {
1710           GScannerKey *key;
1711           guint scope_id;
1712
1713           scope_id = scanner->scope_id;
1714           key = g_scanner_lookup_internal (scanner, scope_id, value.v_identifier);
1715           if (!key && scope_id && scanner->config->scope_0_fallback)
1716             key = g_scanner_lookup_internal (scanner, 0, value.v_identifier);
1717
1718           if (key)
1719             {
1720               g_free (value.v_identifier);
1721               token = G_TOKEN_SYMBOL;
1722               value.v_symbol = key->value;
1723             }
1724         }
1725
1726       if (token == G_TOKEN_IDENTIFIER &&
1727           config->scan_identifier_NULL &&
1728           strlen (value.v_identifier) == 4)
1729         {
1730           gchar *null_upper = "NULL";
1731           gchar *null_lower = "null";
1732
1733           if (scanner->config->case_sensitive)
1734             {
1735               if (value.v_identifier[0] == null_upper[0] &&
1736                   value.v_identifier[1] == null_upper[1] &&
1737                   value.v_identifier[2] == null_upper[2] &&
1738                   value.v_identifier[3] == null_upper[3])
1739                 token = G_TOKEN_IDENTIFIER_NULL;
1740             }
1741           else
1742             {
1743               if ((value.v_identifier[0] == null_upper[0] ||
1744                    value.v_identifier[0] == null_lower[0]) &&
1745                   (value.v_identifier[1] == null_upper[1] ||
1746                    value.v_identifier[1] == null_lower[1]) &&
1747                   (value.v_identifier[2] == null_upper[2] ||
1748                    value.v_identifier[2] == null_lower[2]) &&
1749                   (value.v_identifier[3] == null_upper[3] ||
1750                    value.v_identifier[3] == null_lower[3]))
1751                 token = G_TOKEN_IDENTIFIER_NULL;
1752             }
1753         }
1754     }
1755
1756   *token_p = token;
1757   *value_p = value;
1758 }