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