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