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