glib/gscanner.c

   1 /* GLIB - Library of useful routines for C programming
   2  * Copyright (C) 1995-1997  Peter Mattis, Spencer Kimball and Josh MacDonald
   3  *
   4  * GScanner: Flexible lexical scanner for general purpose.
   5  * Copyright (C) 1997, 1998 Tim Janik
   6  *
   7  * This library is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Library General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2 of the License, or (at your option) any later version.
  11  *
  12  * This library is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Library General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Library General Public
  18  * License along with this library; if not, write to the
  19  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  20  * Boston, MA 02111-1307, USA.
  21  */
  22
  23 /*
  24  * Modified by the GLib Team and others 1997-1999.  See the AUTHORS
  25  * file for a list of people on the GLib Team.  See the ChangeLog
  26  * files for a list of changes.  These files are distributed with
  27  * GLib at ftp://ftp.gtk.org/pub/gtk/.
  28  */
  29
  30 /*
  31  * MT safe
  32  */
  33
  34 #define         __gscanner_c__
  35
  36 #ifdef HAVE_CONFIG_H
  37 #include <config.h>
  38 #endif
  39
  40 #include        <stdlib.h>
  41 #include        <stdarg.h>
  42 #include        <string.h>
  43 #include        <stdio.h>
  44 #include        "glib.h"
  45 #ifdef HAVE_UNISTD_H
  46 #include        <unistd.h>
  47 #endif
  48 #include        <errno.h>
  49 #include        <sys/types.h>   /* needed for sys/stat.h */
  50 #include        <sys/stat.h>
  51 #ifdef G_OS_WIN32
  52 #include        <io.h>          /* For _read() */
  53 #endif
  54
  55 /* --- defines --- */
  56 #define to_lower(c)                             ( \
  57         (guchar) (                                                      \
  58           ( (((guchar)(c))>='A' && ((guchar)(c))<='Z') * ('a'-'A') ) |  \
  59           ( (((guchar)(c))>=192 && ((guchar)(c))<=214) * (224-192) ) |  \
  60           ( (((guchar)(c))>=216 && ((guchar)(c))<=222) * (248-216) ) |  \
  61           ((guchar)(c))                                                 \
  62         )                                                               \
  63 )
  64 #define READ_BUFFER_SIZE        (4000)
  65
  66
  67 /* --- typedefs --- */
  68 typedef struct  _GScannerKey    GScannerKey;
  69
  70 struct  _GScannerKey
  71 {
  72   guint          scope_id;
  73   gchar         *symbol;
  74   gpointer       value;
  75 };
  76
  77
  78
  79 /* --- variables --- */
  80 static GScannerConfig g_scanner_config_template =
  81 {
  82   (
  83    " \t\r\n"
  84    )                    /* cset_skip_characters */,
  85   (
  86    G_CSET_a_2_z
  87    "_"
  88    G_CSET_A_2_Z
  89    )                    /* cset_identifier_first */,
  90   (
  91    G_CSET_a_2_z
  92    "_0123456789"
  93    G_CSET_A_2_Z
  94    G_CSET_LATINS
  95    G_CSET_LATINC
  96    )                    /* cset_identifier_nth */,
  97   ( "#\n" )             /* cpair_comment_single */,
  98
  99   FALSE                 /* case_sensitive */,
 100
 101   TRUE                  /* skip_comment_multi */,
 102   TRUE                  /* skip_comment_single */,
 103   TRUE                  /* scan_comment_multi */,
 104   TRUE                  /* scan_identifier */,
 105   FALSE                 /* scan_identifier_1char */,
 106   FALSE                 /* scan_identifier_NULL */,
 107   TRUE                  /* scan_symbols */,
 108   FALSE                 /* scan_binary */,
 109   TRUE                  /* scan_octal */,
 110   TRUE                  /* scan_float */,
 111   TRUE                  /* scan_hex */,
 112   FALSE                 /* scan_hex_dollar */,
 113   TRUE                  /* scan_string_sq */,
 114   TRUE                  /* scan_string_dq */,
 115   TRUE                  /* numbers_2_int */,
 116   FALSE                 /* int_2_float */,
 117   FALSE                 /* identifier_2_string */,
 118   TRUE                  /* char_2_token */,
 119   FALSE                 /* symbol_2_token */,
 120   FALSE                 /* scope_0_fallback */,
 121 };
 122
 123
 124 /* --- prototypes --- */
 125 static inline
 126 GScannerKey*    g_scanner_lookup_internal (GScanner     *scanner,
 127                                            guint         scope_id,
 128                                            const gchar  *symbol);
 129 static gint     g_scanner_key_equal       (gconstpointer v1,
 130                                            gconstpointer v2);
 131 static guint    g_scanner_key_hash        (gconstpointer v);
 132 static void     g_scanner_get_token_ll    (GScanner     *scanner,
 133                                            GTokenType   *token_p,
 134                                            GTokenValue  *value_p,
 135                                            guint        *line_p,
 136                                            guint        *position_p);
 137 static void     g_scanner_get_token_i     (GScanner     *scanner,
 138                                            GTokenType   *token_p,
 139                                            GTokenValue  *value_p,
 140                                            guint        *line_p,
 141                                            guint        *position_p);
 142
 143 static guchar   g_scanner_peek_next_char  (GScanner     *scanner);
 144 static guchar   g_scanner_get_char        (GScanner     *scanner,
 145                                            guint        *line_p,
 146                                            guint        *position_p);
 147 static void     g_scanner_msg_handler     (GScanner     *scanner,
 148                                            gchar        *message,
 149                                            gint          is_error);
 150
 151
 152 /* --- functions --- */
 153 static inline gint
 154 g_scanner_char_2_num (guchar    c,
 155                       guchar    base)
 156 {
 157   if (c >= '0' && c <= '9')
 158     c -= '0';
 159   else if (c >= 'A' && c <= 'Z')
 160     c -= 'A' - 10;
 161   else if (c >= 'a' && c <= 'z')
 162     c -= 'a' - 10;
 163   else
 164     return -1;
 165
 166   if (c < base)
 167     return c;
 168
 169   return -1;
 170 }
 171
 172 GScanner*
 173 g_scanner_new (GScannerConfig   *config_templ)
 174 {
 175   GScanner *scanner;
 176
 177   if (!config_templ)
 178     config_templ = &g_scanner_config_template;
 179
 180   scanner = g_new0 (GScanner, 1);
 181
 182   scanner->user_data = NULL;
 183   scanner->max_parse_errors = 0;
 184   scanner->parse_errors = 0;
 185   scanner->input_name = NULL;
 186   scanner->derived_data = NULL;
 187
 188   scanner->config = g_new0 (GScannerConfig, 1);
 189
 190   scanner->config->case_sensitive       = config_templ->case_sensitive;
 191   scanner->config->cset_skip_characters = config_templ->cset_skip_characters;
 192   if (!scanner->config->cset_skip_characters)
 193     scanner->config->cset_skip_characters = "";
 194   scanner->config->cset_identifier_first= config_templ->cset_identifier_first;
 195   scanner->config->cset_identifier_nth  = config_templ->cset_identifier_nth;
 196   scanner->config->cpair_comment_single = config_templ->cpair_comment_single;
 197   scanner->config->skip_comment_multi   = config_templ->skip_comment_multi;
 198   scanner->config->skip_comment_single  = config_templ->skip_comment_single;
 199   scanner->config->scan_comment_multi   = config_templ->scan_comment_multi;
 200   scanner->config->scan_identifier      = config_templ->scan_identifier;
 201   scanner->config->scan_identifier_1char= config_templ->scan_identifier_1char;
 202   scanner->config->scan_identifier_NULL = config_templ->scan_identifier_NULL;
 203   scanner->config->scan_symbols         = config_templ->scan_symbols;
 204   scanner->config->scan_binary          = config_templ->scan_binary;
 205   scanner->config->scan_octal           = config_templ->scan_octal;
 206   scanner->config->scan_float           = config_templ->scan_float;
 207   scanner->config->scan_hex             = config_templ->scan_hex;
 208   scanner->config->scan_hex_dollar      = config_templ->scan_hex_dollar;
 209   scanner->config->scan_string_sq       = config_templ->scan_string_sq;
 210   scanner->config->scan_string_dq       = config_templ->scan_string_dq;
 211   scanner->config->numbers_2_int        = config_templ->numbers_2_int;
 212   scanner->config->int_2_float          = config_templ->int_2_float;
 213   scanner->config->identifier_2_string  = config_templ->identifier_2_string;
 214   scanner->config->char_2_token         = config_templ->char_2_token;
 215   scanner->config->symbol_2_token       = config_templ->symbol_2_token;
 216   scanner->config->scope_0_fallback     = config_templ->scope_0_fallback;
 217
 218   scanner->token = G_TOKEN_NONE;
 219   scanner->value.v_int = 0;
 220   scanner->line = 1;
 221   scanner->position = 0;
 222
 223   scanner->next_token = G_TOKEN_NONE;
 224   scanner->next_value.v_int = 0;
 225   scanner->next_line = 1;
 226   scanner->next_position = 0;
 227
 228   scanner->symbol_table = g_hash_table_new (g_scanner_key_hash, g_scanner_key_equal);
 229   scanner->input_fd = -1;
 230   scanner->text = NULL;
 231   scanner->text_end = NULL;
 232   scanner->buffer = NULL;
 233   scanner->scope_id = 0;
 234
 235   scanner->msg_handler = g_scanner_msg_handler;
 236
 237   return scanner;
 238 }
 239
 240 static inline void
 241 g_scanner_free_value (GTokenType     *token_p,
 242                       GTokenValue     *value_p)
 243 {
 244   switch (*token_p)
 245     {
 246     case G_TOKEN_STRING:
 247     case G_TOKEN_IDENTIFIER:
 248     case G_TOKEN_IDENTIFIER_NULL:
 249     case G_TOKEN_COMMENT_SINGLE:
 250     case G_TOKEN_COMMENT_MULTI:
 251       g_free (value_p->v_string);
 252       break;
 253
 254     default:
 255       break;
 256     }
 257
 258   *token_p = G_TOKEN_NONE;
 259 }
 260
 261 static void
 262 g_scanner_destroy_symbol_table_entry (gpointer _key,
 263                                       gpointer _value,
 264                                       gpointer _data)
 265 {
 266   GScannerKey *key = _key;
 267
 268   g_free (key->symbol);
 269   g_free (key);
 270 }
 271
 272 void
 273 g_scanner_destroy (GScanner     *scanner)
 274 {
 275   g_return_if_fail (scanner != NULL);
 276
 277   g_hash_table_foreach (scanner->symbol_table,
 278                         g_scanner_destroy_symbol_table_entry, NULL);
 279   g_hash_table_destroy (scanner->symbol_table);
 280   g_scanner_free_value (&scanner->token, &scanner->value);
 281   g_scanner_free_value (&scanner->next_token, &scanner->next_value);
 282   g_free (scanner->config);
 283   g_free (scanner->buffer);
 284   g_free (scanner);
 285 }
 286
 287 static void
 288 g_scanner_msg_handler (GScanner         *scanner,
 289                        gchar            *message,
 290                        gint              is_error)
 291 {
 292   g_return_if_fail (scanner != NULL);
 293
 294   fprintf (stdout, "%s:%d: ", scanner->input_name, scanner->line);
 295   if (is_error)
 296     fprintf (stdout, "error: ");
 297   fprintf (stdout, "%s\n", message);
 298 }
 299
 300 void
 301 g_scanner_error (GScanner       *scanner,
 302                  const gchar    *format,
 303                  ...)
 304 {
 305   g_return_if_fail (scanner != NULL);
 306   g_return_if_fail (format != NULL);
 307
 308   scanner->parse_errors++;
 309
 310   if (scanner->msg_handler)
 311     {
 312       va_list args;
 313       gchar *string;
 314
 315       va_start (args, format);
 316       string = g_strdup_vprintf (format, args);
 317       va_end (args);
 318
 319       scanner->msg_handler (scanner, string, TRUE);
 320
 321       g_free (string);
 322     }
 323 }
 324
 325 void
 326 g_scanner_warn (GScanner       *scanner,
 327                 const gchar    *format,
 328                 ...)
 329 {
 330   g_return_if_fail (scanner != NULL);
 331   g_return_if_fail (format != NULL);
 332
 333   if (scanner->msg_handler)
 334     {
 335       va_list args;
 336       gchar *string;
 337
 338       va_start (args, format);
 339       string = g_strdup_vprintf (format, args);
 340       va_end (args);
 341
 342       scanner->msg_handler (scanner, string, FALSE);
 343
 344       g_free (string);
 345     }
 346 }
 347
 348 static gint
 349 g_scanner_key_equal (gconstpointer v1,
 350                      gconstpointer v2)
 351 {
 352   const GScannerKey *key1 = v1;
 353   const GScannerKey *key2 = v2;
 354
 355   return (key1->scope_id == key2->scope_id) && (strcmp (key1->symbol, key2->symbol) == 0);
 356 }
 357
 358 static guint
 359 g_scanner_key_hash (gconstpointer v)
 360 {
 361   const GScannerKey *key = v;
 362   gchar *c;
 363   guint h;
 364
 365   h = key->scope_id;
 366   for (c = key->symbol; *c; c++)
 367     {
 368       guint g;
 369
 370       h = (h << 4) + *c;
 371       g = h & 0xf0000000;
 372       if (g)
 373         {
 374           h = h ^ (g >> 24);
 375           h = h ^ g;
 376         }
 377     }
 378
 379   return h;
 380 }
 381
 382 static inline GScannerKey*
 383 g_scanner_lookup_internal (GScanner     *scanner,
 384                            guint         scope_id,
 385                            const gchar  *symbol)
 386 {
 387   GScannerKey   *key_p;
 388   GScannerKey key;
 389
 390   key.scope_id = scope_id;
 391
 392   if (!scanner->config->case_sensitive)
 393     {
 394       gchar *d;
 395       const gchar *c;
 396
 397       key.symbol = g_new (gchar, strlen (symbol) + 1);
 398       for (d = key.symbol, c = symbol; *c; c++, d++)
 399         *d = to_lower (*c);
 400       *d = 0;
 401       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 402       g_free (key.symbol);
 403     }
 404   else
 405     {
 406       key.symbol = (gchar*) symbol;
 407       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 408     }
 409
 410   return key_p;
 411 }
 412
 413 void
 414 g_scanner_scope_add_symbol (GScanner    *scanner,
 415                             guint        scope_id,
 416                             const gchar *symbol,
 417                             gpointer     value)
 418 {
 419   GScannerKey   *key;
 420
 421   g_return_if_fail (scanner != NULL);
 422   g_return_if_fail (symbol != NULL);
 423
 424   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 425
 426   if (!key)
 427     {
 428       key = g_new (GScannerKey, 1);
 429       key->scope_id = scope_id;
 430       key->symbol = g_strdup (symbol);
 431       key->value = value;
 432       if (!scanner->config->case_sensitive)
 433         {
 434           gchar *c;
 435
 436           c = key->symbol;
 437           while (*c != 0)
 438             {
 439               *c = to_lower (*c);
 440               c++;
 441             }
 442         }
 443       g_hash_table_insert (scanner->symbol_table, key, key);
 444     }
 445   else
 446     key->value = value;
 447 }
 448
 449 void
 450 g_scanner_scope_remove_symbol (GScanner    *scanner,
 451                                guint        scope_id,
 452                                const gchar *symbol)
 453 {
 454   GScannerKey   *key;
 455
 456   g_return_if_fail (scanner != NULL);
 457   g_return_if_fail (symbol != NULL);
 458
 459   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 460
 461   if (key)
 462     {
 463       g_hash_table_remove (scanner->symbol_table, key);
 464       g_free (key->symbol);
 465       g_free (key);
 466     }
 467 }
 468
 469 gpointer
 470 g_scanner_lookup_symbol (GScanner       *scanner,
 471                          const gchar    *symbol)
 472 {
 473   GScannerKey   *key;
 474   guint scope_id;
 475
 476   g_return_val_if_fail (scanner != NULL, NULL);
 477
 478   if (!symbol)
 479     return NULL;
 480
 481   scope_id = scanner->scope_id;
 482   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 483   if (!key && scope_id && scanner->config->scope_0_fallback)
 484     key = g_scanner_lookup_internal (scanner, 0, symbol);
 485
 486   if (key)
 487     return key->value;
 488   else
 489     return NULL;
 490 }
 491
 492 gpointer
 493 g_scanner_scope_lookup_symbol (GScanner       *scanner,
 494                                guint           scope_id,
 495                                const gchar    *symbol)
 496 {
 497   GScannerKey   *key;
 498
 499   g_return_val_if_fail (scanner != NULL, NULL);
 500
 501   if (!symbol)
 502     return NULL;
 503
 504   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 505
 506   if (key)
 507     return key->value;
 508   else
 509     return NULL;
 510 }
 511
 512 guint
 513 g_scanner_set_scope (GScanner       *scanner,
 514                      guint           scope_id)
 515 {
 516   guint old_scope_id;
 517
 518   g_return_val_if_fail (scanner != NULL, 0);
 519
 520   old_scope_id = scanner->scope_id;
 521   scanner->scope_id = scope_id;
 522
 523   return old_scope_id;
 524 }
 525
 526 static void
 527 g_scanner_foreach_internal (gpointer  _key,
 528                             gpointer  _value,
 529                             gpointer  _user_data)
 530 {
 531   GScannerKey *key;
 532   gpointer *d;
 533   GHFunc func;
 534   gpointer user_data;
 535   guint *scope_id;
 536
 537   d = _user_data;
 538   func = (GHFunc) d[0];
 539   user_data = d[1];
 540   scope_id = d[2];
 541   key = _value;
 542
 543   if (key->scope_id == *scope_id)
 544     func (key->symbol, key->value, user_data);
 545 }
 546
 547 void
 548 g_scanner_scope_foreach_symbol (GScanner       *scanner,
 549                                 guint           scope_id,
 550                                 GHFunc          func,
 551                                 gpointer        user_data)
 552 {
 553   gpointer d[3];
 554
 555   g_return_if_fail (scanner != NULL);
 556
 557   d[0] = (gpointer) func;
 558   d[1] = user_data;
 559   d[2] = &scope_id;
 560
 561   g_hash_table_foreach (scanner->symbol_table, g_scanner_foreach_internal, d);
 562 }
 563
 564 void
 565 g_scanner_freeze_symbol_table (GScanner *scanner)
 566 {
 567   g_return_if_fail (scanner != NULL);
 568
 569   g_hash_table_freeze (scanner->symbol_table);
 570 }
 571
 572 void
 573 g_scanner_thaw_symbol_table (GScanner *scanner)
 574 {
 575   g_return_if_fail (scanner != NULL);
 576
 577   g_hash_table_thaw (scanner->symbol_table);
 578 }
 579
 580 GTokenType
 581 g_scanner_peek_next_token (GScanner     *scanner)
 582 {
 583   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 584
 585   if (scanner->next_token == G_TOKEN_NONE)
 586     {
 587       scanner->next_line = scanner->line;
 588       scanner->next_position = scanner->position;
 589       g_scanner_get_token_i (scanner,
 590                              &scanner->next_token,
 591                              &scanner->next_value,
 592                              &scanner->next_line,
 593                              &scanner->next_position);
 594     }
 595
 596   return scanner->next_token;
 597 }
 598
 599 GTokenType
 600 g_scanner_get_next_token (GScanner      *scanner)
 601 {
 602   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 603
 604   if (scanner->next_token != G_TOKEN_NONE)
 605     {
 606       g_scanner_free_value (&scanner->token, &scanner->value);
 607
 608       scanner->token = scanner->next_token;
 609       scanner->value = scanner->next_value;
 610       scanner->line = scanner->next_line;
 611       scanner->position = scanner->next_position;
 612       scanner->next_token = G_TOKEN_NONE;
 613     }
 614   else
 615     g_scanner_get_token_i (scanner,
 616                            &scanner->token,
 617                            &scanner->value,
 618                            &scanner->line,
 619                            &scanner->position);
 620
 621   return scanner->token;
 622 }
 623
 624 GTokenType
 625 g_scanner_cur_token (GScanner *scanner)
 626 {
 627   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 628
 629   return scanner->token;
 630 }
 631
 632 GTokenValue
 633 g_scanner_cur_value (GScanner *scanner)
 634 {
 635   GTokenValue v;
 636
 637   v.v_int = 0;
 638
 639   g_return_val_if_fail (scanner != NULL, v);
 640
 641   /* MSC isn't capable of handling return scanner->value; ? */
 642
 643   v = scanner->value;
 644
 645   return v;
 646 }
 647
 648 guint
 649 g_scanner_cur_line (GScanner *scanner)
 650 {
 651   g_return_val_if_fail (scanner != NULL, 0);
 652
 653   return scanner->line;
 654 }
 655
 656 guint
 657 g_scanner_cur_position (GScanner *scanner)
 658 {
 659   g_return_val_if_fail (scanner != NULL, 0);
 660
 661   return scanner->position;
 662 }
 663
 664 gboolean
 665 g_scanner_eof (GScanner *scanner)
 666 {
 667   g_return_val_if_fail (scanner != NULL, TRUE);
 668
 669   return scanner->token == G_TOKEN_EOF;
 670 }
 671
 672 void
 673 g_scanner_input_file (GScanner *scanner,
 674                       gint      input_fd)
 675 {
 676   g_return_if_fail (scanner != NULL);
 677   g_return_if_fail (input_fd >= 0);
 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 = input_fd;
 689   scanner->text = NULL;
 690   scanner->text_end = NULL;
 691
 692   if (!scanner->buffer)
 693     scanner->buffer = g_new (gchar, READ_BUFFER_SIZE + 1);
 694 }
 695
 696 void
 697 g_scanner_input_text (GScanner    *scanner,
 698                       const gchar *text,
 699                       guint        text_len)
 700 {
 701   g_return_if_fail (scanner != NULL);
 702   if (text_len)
 703     g_return_if_fail (text != NULL);
 704   else
 705     text = NULL;
 706
 707   if (scanner->input_fd >= 0)
 708     g_scanner_sync_file_offset (scanner);
 709
 710   scanner->token = G_TOKEN_NONE;
 711   scanner->value.v_int = 0;
 712   scanner->line = 1;
 713   scanner->position = 0;
 714   scanner->next_token = G_TOKEN_NONE;
 715
 716   scanner->input_fd = -1;
 717   scanner->text = text;
 718   scanner->text_end = text + text_len;
 719
 720   if (scanner->buffer)
 721     {
 722       g_free (scanner->buffer);
 723       scanner->buffer = NULL;
 724     }
 725 }
 726
 727 static guchar
 728 g_scanner_peek_next_char (GScanner *scanner)
 729 {
 730   if (scanner->text < scanner->text_end)
 731     {
 732       return *scanner->text;
 733     }
 734   else if (scanner->input_fd >= 0)
 735     {
 736       gint count;
 737       gchar *buffer;
 738
 739       buffer = scanner->buffer;
 740       do
 741         {
 742           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 743         }
 744       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 745
 746       if (count < 1)
 747         {
 748           scanner->input_fd = -1;
 749
 750           return 0;
 751         }
 752       else
 753         {
 754           scanner->text = buffer;
 755           scanner->text_end = buffer + count;
 756
 757           return *buffer;
 758         }
 759     }
 760   else
 761     return 0;
 762 }
 763
 764 void
 765 g_scanner_sync_file_offset (GScanner *scanner)
 766 {
 767   g_return_if_fail (scanner != NULL);
 768
 769   /* for file input, rewind the filedescriptor to the current
 770    * buffer position and blow the file read ahead buffer. usefull for
 771    * third party uses of our filedescriptor, which hooks onto the current
 772    * scanning position.
 773    */
 774
 775   if (scanner->input_fd >= 0 && scanner->text_end > scanner->text)
 776     {
 777       gint buffered;
 778
 779       buffered = scanner->text_end - scanner->text;
 780       if (lseek (scanner->input_fd, - buffered, SEEK_CUR) >= 0)
 781         {
 782           /* we succeeded, blow our buffer's contents now */
 783           scanner->text = NULL;
 784           scanner->text_end = NULL;
 785         }
 786       else
 787         errno = 0;
 788     }
 789 }
 790
 791 static guchar
 792 g_scanner_get_char (GScanner    *scanner,
 793                     guint       *line_p,
 794                     guint       *position_p)
 795 {
 796   guchar fchar;
 797
 798   if (scanner->text < scanner->text_end)
 799     fchar = *(scanner->text++);
 800   else if (scanner->input_fd >= 0)
 801     {
 802       gint count;
 803       gchar *buffer;
 804
 805       buffer = scanner->buffer;
 806       do
 807         {
 808           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 809         }
 810       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 811
 812       if (count < 1)
 813         {
 814           scanner->input_fd = -1;
 815           fchar = 0;
 816         }
 817       else
 818         {
 819           scanner->text = buffer + 1;
 820           scanner->text_end = buffer + count;
 821           fchar = *buffer;
 822           if (!fchar)
 823             {
 824               g_scanner_sync_file_offset (scanner);
 825               scanner->text_end = scanner->text;
 826               scanner->input_fd = -1;
 827             }
 828         }
 829     }
 830   else
 831     fchar = 0;
 832
 833   if (fchar == '\n')
 834     {
 835       (*position_p) = 0;
 836       (*line_p)++;
 837     }
 838   else if (fchar)
 839     {
 840       (*position_p)++;
 841     }
 842
 843   return fchar;
 844 }
 845
 846 void
 847 g_scanner_unexp_token (GScanner         *scanner,
 848                        GTokenType        expected_token,
 849                        const gchar      *identifier_spec,
 850                        const gchar      *symbol_spec,
 851                        const gchar      *symbol_name,
 852                        const gchar      *message,
 853                        gint              is_error)
 854 {
 855   gchar *token_string;
 856   guint token_string_len;
 857   gchar *expected_string;
 858   guint expected_string_len;
 859   gchar *message_prefix;
 860   gboolean print_unexp;
 861   void (*msg_handler)   (GScanner*, const gchar*, ...);
 862
 863   g_return_if_fail (scanner != NULL);
 864
 865   if (is_error)
 866     msg_handler = g_scanner_error;
 867   else
 868     msg_handler = g_scanner_warn;
 869
 870   if (!identifier_spec)
 871     identifier_spec = "identifier";
 872   if (!symbol_spec)
 873     symbol_spec = "symbol";
 874
 875   token_string_len = 56;
 876   token_string = g_new (gchar, token_string_len + 1);
 877   expected_string_len = 64;
 878   expected_string = g_new (gchar, expected_string_len + 1);
 879   print_unexp = TRUE;
 880
 881   switch (scanner->token)
 882     {
 883     case G_TOKEN_EOF:
 884       g_snprintf (token_string, token_string_len, "end of file");
 885       break;
 886
 887     default:
 888       if (scanner->token >= 1 && scanner->token <= 255)
 889         {
 890           if ((scanner->token >= ' ' && scanner->token <= '~') ||
 891               strchr (scanner->config->cset_identifier_first, scanner->token) ||
 892               strchr (scanner->config->cset_identifier_nth, scanner->token))
 893             g_snprintf (token_string, expected_string_len, "character `%c'", scanner->token);
 894           else
 895             g_snprintf (token_string, expected_string_len, "character `\\%o'", scanner->token);
 896           break;
 897         }
 898       else if (!scanner->config->symbol_2_token)
 899         {
 900           g_snprintf (token_string, token_string_len, "(unknown) token <%d>", scanner->token);
 901           break;
 902         }
 903       /* fall through */
 904     case G_TOKEN_SYMBOL:
 905       if (expected_token == G_TOKEN_SYMBOL ||
 906           (scanner->config->symbol_2_token &&
 907            expected_token > G_TOKEN_LAST))
 908         print_unexp = FALSE;
 909       if (symbol_name)
 910         g_snprintf (token_string,
 911                     token_string_len,
 912                     "%s%s `%s'",
 913                     print_unexp ? "" : "invalid ",
 914                     symbol_spec,
 915                     symbol_name);
 916       else
 917         g_snprintf (token_string,
 918                     token_string_len,
 919                     "%s%s",
 920                     print_unexp ? "" : "invalid ",
 921                     symbol_spec);
 922       break;
 923
 924     case G_TOKEN_ERROR:
 925       print_unexp = FALSE;
 926       expected_token = G_TOKEN_NONE;
 927       switch (scanner->value.v_error)
 928         {
 929         case G_ERR_UNEXP_EOF:
 930           g_snprintf (token_string, token_string_len, "scanner: unexpected end of file");
 931           break;
 932
 933         case G_ERR_UNEXP_EOF_IN_STRING:
 934           g_snprintf (token_string, token_string_len, "scanner: unterminated string constant");
 935           break;
 936
 937         case G_ERR_UNEXP_EOF_IN_COMMENT:
 938           g_snprintf (token_string, token_string_len, "scanner: unterminated comment");
 939           break;
 940
 941         case G_ERR_NON_DIGIT_IN_CONST:
 942           g_snprintf (token_string, token_string_len, "scanner: non digit in constant");
 943           break;
 944
 945         case G_ERR_FLOAT_RADIX:
 946           g_snprintf (token_string, token_string_len, "scanner: invalid radix for floating constant");
 947           break;
 948
 949         case G_ERR_FLOAT_MALFORMED:
 950           g_snprintf (token_string, token_string_len, "scanner: malformed floating constant");
 951           break;
 952
 953         case G_ERR_DIGIT_RADIX:
 954           g_snprintf (token_string, token_string_len, "scanner: digit is beyond radix");
 955           break;
 956
 957         case G_ERR_UNKNOWN:
 958         default:
 959           g_snprintf (token_string, token_string_len, "scanner: unknown error");
 960           break;
 961         }
 962       break;
 963
 964     case G_TOKEN_CHAR:
 965       g_snprintf (token_string, token_string_len, "character `%c'", scanner->value.v_char);
 966       break;
 967
 968     case G_TOKEN_IDENTIFIER:
 969     case G_TOKEN_IDENTIFIER_NULL:
 970       if (expected_token == G_TOKEN_IDENTIFIER ||
 971           expected_token == G_TOKEN_IDENTIFIER_NULL)
 972         print_unexp = FALSE;
 973       g_snprintf (token_string,
 974                   token_string_len,
 975                   "%s%s `%s'",
 976                   print_unexp ? "" : "invalid ",
 977                   identifier_spec,
 978                   scanner->token == G_TOKEN_IDENTIFIER ? scanner->value.v_string : "null");
 979       break;
 980
 981     case G_TOKEN_BINARY:
 982     case G_TOKEN_OCTAL:
 983     case G_TOKEN_INT:
 984     case G_TOKEN_HEX:
 985       g_snprintf (token_string, token_string_len, "number `%ld'", scanner->value.v_int);
 986       break;
 987
 988     case G_TOKEN_FLOAT:
 989       g_snprintf (token_string, token_string_len, "number `%.3f'", scanner->value.v_float);
 990       break;
 991
 992     case G_TOKEN_STRING:
 993       if (expected_token == G_TOKEN_STRING)
 994         print_unexp = FALSE;
 995       g_snprintf (token_string,
 996                   token_string_len,
 997                   "%s%sstring constant \"%s\"",
 998                   print_unexp ? "" : "invalid ",
 999                   scanner->value.v_string[0] == 0 ? "empty " : "",
1000                   scanner->value.v_string);
1001       token_string[token_string_len - 2] = '"';
1002       token_string[token_string_len - 1] = 0;
1003       break;
1004
1005     case G_TOKEN_COMMENT_SINGLE:
1006     case G_TOKEN_COMMENT_MULTI:
1007       g_snprintf (token_string, token_string_len, "comment");
1008       break;
1009
1010     case G_TOKEN_NONE:
1011       /* somehow the user's parsing code is screwed, there isn't much
1012        * we can do about it.
1013        * Note, a common case to trigger this is
1014        * g_scanner_peek_next_token(); g_scanner_unexp_token();
1015        * without an intermediate g_scanner_get_next_token().
1016        */
1017       g_assert_not_reached ();
1018       break;
1019     }
1020
1021
1022   switch (expected_token)
1023     {
1024       gboolean need_valid;
1025
1026     default:
1027       if (expected_token >= 1 && expected_token <= 255)
1028         {
1029           if ((expected_token >= ' ' && expected_token <= '~') ||
1030               strchr (scanner->config->cset_identifier_first, expected_token) ||
1031               strchr (scanner->config->cset_identifier_nth, expected_token))
1032             g_snprintf (expected_string, expected_string_len, "character `%c'", expected_token);
1033           else
1034             g_snprintf (expected_string, expected_string_len, "character `\\%o'", expected_token);
1035           break;
1036         }
1037       else if (!scanner->config->symbol_2_token)
1038         {
1039           g_snprintf (expected_string, expected_string_len, "(unknown) token <%d>", expected_token);
1040           break;
1041         }
1042       /* fall through */
1043     case G_TOKEN_SYMBOL:
1044       need_valid = (scanner->token == G_TOKEN_SYMBOL ||
1045                     (scanner->config->symbol_2_token &&
1046                      scanner->token > G_TOKEN_LAST));
1047       g_snprintf (expected_string,
1048                   expected_string_len,
1049                   "%s%s",
1050                   need_valid ? "valid " : "",
1051                   symbol_spec);
1052       /* FIXME: should we attempt to lookup the symbol_name for symbol_2_token? */
1053       break;
1054
1055     case G_TOKEN_INT:
1056       g_snprintf (expected_string, expected_string_len, "number (integer)");
1057       break;
1058
1059     case G_TOKEN_FLOAT:
1060       g_snprintf (expected_string, expected_string_len, "number (float)");
1061       break;
1062
1063     case G_TOKEN_STRING:
1064       g_snprintf (expected_string,
1065                   expected_string_len,
1066                   "%sstring constant",
1067                   scanner->token == G_TOKEN_STRING ? "valid " : "");
1068       break;
1069
1070     case G_TOKEN_IDENTIFIER:
1071     case G_TOKEN_IDENTIFIER_NULL:
1072       g_snprintf (expected_string,
1073                   expected_string_len,
1074                   "%s%s",
1075                   (scanner->token == G_TOKEN_IDENTIFIER_NULL ||
1076                    scanner->token == G_TOKEN_IDENTIFIER ? "valid " : ""),
1077                   identifier_spec);
1078       break;
1079
1080     case G_TOKEN_EOF:
1081       g_snprintf (expected_string, expected_string_len, "end of file");
1082       break;
1083
1084     case G_TOKEN_NONE:
1085       break;
1086     }
1087
1088   if (message && message[0] != 0)
1089     message_prefix = " - ";
1090   else
1091     {
1092       message_prefix = "";
1093       message = "";
1094     }
1095
1096   if (expected_token != G_TOKEN_NONE)
1097     {
1098       if (print_unexp)
1099         msg_handler (scanner,
1100                      "unexpected %s, expected %s%s%s",
1101                      token_string,
1102                      expected_string,
1103                      message_prefix,
1104                      message);
1105       else
1106         msg_handler (scanner,
1107                      "%s, expected %s%s%s",
1108                      token_string,
1109                      expected_string,
1110                      message_prefix,
1111                      message);
1112     }
1113   else
1114     {
1115       if (print_unexp)
1116         msg_handler (scanner,
1117                      "unexpected %s%s%s",
1118                      token_string,
1119                      message_prefix,
1120                      message);
1121       else
1122         msg_handler (scanner,
1123                      "%s%s%s",
1124                      token_string,
1125                      message_prefix,
1126                      message);
1127     }
1128
1129   g_free (token_string);
1130   g_free (expected_string);
1131 }
1132
1133 gint
1134 g_scanner_stat_mode (const gchar *filename)
1135 {
1136   struct stat  *stat_buf;
1137   gint          st_mode;
1138
1139   stat_buf = g_new0 (struct stat, 1);
1140 #ifdef HAVE_LSTAT
1141   lstat (filename, stat_buf);
1142 #else
1143   stat (filename, stat_buf);
1144 #endif
1145   st_mode = stat_buf->st_mode;
1146
1147   g_free (stat_buf);
1148
1149   return st_mode;
1150 }
1151
1152 static void
1153 g_scanner_get_token_i (GScanner *scanner,
1154                        GTokenType       *token_p,
1155                        GTokenValue      *value_p,
1156                        guint            *line_p,
1157                        guint            *position_p)
1158 {
1159   do
1160     {
1161       g_scanner_free_value (token_p, value_p);
1162       g_scanner_get_token_ll (scanner, token_p, value_p, line_p, position_p);
1163     }
1164   while (((*token_p > 0 && *token_p < 256) &&
1165           strchr (scanner->config->cset_skip_characters, *token_p)) ||
1166          (*token_p == G_TOKEN_CHAR &&
1167           strchr (scanner->config->cset_skip_characters, value_p->v_char)) ||
1168          (*token_p == G_TOKEN_COMMENT_MULTI &&
1169           scanner->config->skip_comment_multi) ||
1170          (*token_p == G_TOKEN_COMMENT_SINGLE &&
1171           scanner->config->skip_comment_single));
1172
1173   switch (*token_p)
1174     {
1175     case G_TOKEN_IDENTIFIER:
1176       if (scanner->config->identifier_2_string)
1177         *token_p = G_TOKEN_STRING;
1178       break;
1179
1180     case G_TOKEN_SYMBOL:
1181       if (scanner->config->symbol_2_token)
1182         *token_p = (GTokenType) value_p->v_symbol;
1183       break;
1184
1185     case G_TOKEN_BINARY:
1186     case G_TOKEN_OCTAL:
1187     case G_TOKEN_HEX:
1188       if (scanner->config->numbers_2_int)
1189         *token_p = G_TOKEN_INT;
1190       break;
1191
1192     default:
1193       break;
1194     }
1195
1196   if (*token_p == G_TOKEN_INT &&
1197       scanner->config->int_2_float)
1198     {
1199       *token_p = G_TOKEN_FLOAT;
1200       value_p->v_float = value_p->v_int;
1201     }
1202
1203   errno = 0;
1204 }
1205
1206 static void
1207 g_scanner_get_token_ll  (GScanner       *scanner,
1208                          GTokenType     *token_p,
1209                          GTokenValue    *value_p,
1210                          guint          *line_p,
1211                          guint          *position_p)
1212 {
1213   GScannerConfig *config;
1214   GTokenType       token;
1215   gboolean         in_comment_multi;
1216   gboolean         in_comment_single;
1217   gboolean         in_string_sq;
1218   gboolean         in_string_dq;
1219   GString         *gstring;
1220   GTokenValue      value;
1221   guchar           ch;
1222
1223   config = scanner->config;
1224   (*value_p).v_int = 0;
1225
1226   if ((scanner->text >= scanner->text_end && scanner->input_fd < 0) ||
1227       scanner->token == G_TOKEN_EOF)
1228     {
1229       *token_p = G_TOKEN_EOF;
1230       return;
1231     }
1232
1233   in_comment_multi = FALSE;
1234   in_comment_single = FALSE;
1235   in_string_sq = FALSE;
1236   in_string_dq = FALSE;
1237   gstring = NULL;
1238
1239   do /* while (ch != 0) */
1240     {
1241       gboolean dotted_float = FALSE;
1242
1243       ch = g_scanner_get_char (scanner, line_p, position_p);
1244
1245       value.v_int = 0;
1246       token = G_TOKEN_NONE;
1247
1248       /* this is *evil*, but needed ;(
1249        * we first check for identifier first character, because  it
1250        * might interfere with other key chars like slashes or numbers
1251        */
1252       if (config->scan_identifier &&
1253           ch && strchr (config->cset_identifier_first, ch))
1254         goto identifier_precedence;
1255
1256       switch (ch)
1257         {
1258         case 0:
1259           token = G_TOKEN_EOF;
1260           (*position_p)++;
1261           /* ch = 0; */
1262           break;
1263
1264         case '/':
1265           if (!config->scan_comment_multi ||
1266               g_scanner_peek_next_char (scanner) != '*')
1267             goto default_case;
1268           g_scanner_get_char (scanner, line_p, position_p);
1269           token = G_TOKEN_COMMENT_MULTI;
1270           in_comment_multi = TRUE;
1271           gstring = g_string_new ("");
1272           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1273             {
1274               if (ch == '*' && g_scanner_peek_next_char (scanner) == '/')
1275                 {
1276                   g_scanner_get_char (scanner, line_p, position_p);
1277                   in_comment_multi = FALSE;
1278                   break;
1279                 }
1280               else
1281                 gstring = g_string_append_c (gstring, ch);
1282             }
1283           ch = 0;
1284           break;
1285
1286         case '\'':
1287           if (!config->scan_string_sq)
1288             goto default_case;
1289           token = G_TOKEN_STRING;
1290           in_string_sq = TRUE;
1291           gstring = g_string_new ("");
1292           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1293             {
1294               if (ch == '\'')
1295                 {
1296                   in_string_sq = FALSE;
1297                   break;
1298                 }
1299               else
1300                 gstring = g_string_append_c (gstring, ch);
1301             }
1302           ch = 0;
1303           break;
1304
1305         case '"':
1306           if (!config->scan_string_dq)
1307             goto default_case;
1308           token = G_TOKEN_STRING;
1309           in_string_dq = TRUE;
1310           gstring = g_string_new ("");
1311           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1312             {
1313               if (ch == '"')
1314                 {
1315                   in_string_dq = FALSE;
1316                   break;
1317                 }
1318               else
1319                 {
1320                   if (ch == '\\')
1321                     {
1322                       ch = g_scanner_get_char (scanner, line_p, position_p);
1323                       switch (ch)
1324                         {
1325                           guint i;
1326                           guint fchar;
1327
1328                         case 0:
1329                           break;
1330
1331                         case '\\':
1332                           gstring = g_string_append_c (gstring, '\\');
1333                           break;
1334
1335                         case 'n':
1336                           gstring = g_string_append_c (gstring, '\n');
1337                           break;
1338
1339                         case 't':
1340                           gstring = g_string_append_c (gstring, '\t');
1341                           break;
1342
1343                         case 'r':
1344                           gstring = g_string_append_c (gstring, '\r');
1345                           break;
1346
1347                         case 'b':
1348                           gstring = g_string_append_c (gstring, '\b');
1349                           break;
1350
1351                         case 'f':
1352                           gstring = g_string_append_c (gstring, '\f');
1353                           break;
1354
1355                         case '0':
1356                         case '1':
1357                         case '2':
1358                         case '3':
1359                         case '4':
1360                         case '5':
1361                         case '6':
1362                         case '7':
1363                           i = ch - '0';
1364                           fchar = g_scanner_peek_next_char (scanner);
1365                           if (fchar >= '0' && fchar <= '7')
1366                             {
1367                               ch = g_scanner_get_char (scanner, line_p, position_p);
1368                               i = i * 8 + ch - '0';
1369                               fchar = g_scanner_peek_next_char (scanner);
1370                               if (fchar >= '0' && fchar <= '7')
1371                                 {
1372                                   ch = g_scanner_get_char (scanner, line_p, position_p);
1373                                   i = i * 8 + ch - '0';
1374                                 }
1375                             }
1376                           gstring = g_string_append_c (gstring, i);
1377                           break;
1378
1379                         default:
1380                           gstring = g_string_append_c (gstring, ch);
1381                           break;
1382                         }
1383                     }
1384                   else
1385                     gstring = g_string_append_c (gstring, ch);
1386                 }
1387             }
1388           ch = 0;
1389           break;
1390
1391         case '.':
1392           if (!config->scan_float)
1393             goto default_case;
1394           token = G_TOKEN_FLOAT;
1395           dotted_float = TRUE;
1396           ch = g_scanner_get_char (scanner, line_p, position_p);
1397           goto number_parsing;
1398
1399         case '$':
1400           if (!config->scan_hex_dollar)
1401             goto default_case;
1402           token = G_TOKEN_HEX;
1403           ch = g_scanner_get_char (scanner, line_p, position_p);
1404           goto number_parsing;
1405
1406         case '0':
1407           if (config->scan_octal)
1408             token = G_TOKEN_OCTAL;
1409           else
1410             token = G_TOKEN_INT;
1411           ch = g_scanner_peek_next_char (scanner);
1412           if (config->scan_hex && (ch == 'x' || ch == 'X'))
1413             {
1414               token = G_TOKEN_HEX;
1415               g_scanner_get_char (scanner, line_p, position_p);
1416               ch = g_scanner_get_char (scanner, line_p, position_p);
1417               if (ch == 0)
1418                 {
1419                   token = G_TOKEN_ERROR;
1420                   value.v_error = G_ERR_UNEXP_EOF;
1421                   (*position_p)++;
1422                   break;
1423                 }
1424               if (g_scanner_char_2_num (ch, 16) < 0)
1425                 {
1426                   token = G_TOKEN_ERROR;
1427                   value.v_error = G_ERR_DIGIT_RADIX;
1428                   ch = 0;
1429                   break;
1430                 }
1431             }
1432           else if (config->scan_binary && (ch == 'b' || ch == 'B'))
1433             {
1434               token = G_TOKEN_BINARY;
1435               g_scanner_get_char (scanner, line_p, position_p);
1436               ch = g_scanner_get_char (scanner, line_p, position_p);
1437               if (ch == 0)
1438                 {
1439                   token = G_TOKEN_ERROR;
1440                   value.v_error = G_ERR_UNEXP_EOF;
1441                   (*position_p)++;
1442                   break;
1443                 }
1444               if (g_scanner_char_2_num (ch, 10) < 0)
1445                 {
1446                   token = G_TOKEN_ERROR;
1447                   value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1448                   ch = 0;
1449                   break;
1450                 }
1451             }
1452           else
1453             ch = '0';
1454           /* fall through */
1455         case '1':
1456         case '2':
1457         case '3':
1458         case '4':
1459         case '5':
1460         case '6':
1461         case '7':
1462         case '8':
1463         case '9':
1464         number_parsing:
1465         {
1466           gboolean in_number = TRUE;
1467           gchar *endptr;
1468
1469           if (token == G_TOKEN_NONE)
1470             token = G_TOKEN_INT;
1471
1472           gstring = g_string_new (dotted_float ? "0." : "");
1473           gstring = g_string_append_c (gstring, ch);
1474
1475           do /* while (in_number) */
1476             {
1477               gboolean is_E;
1478
1479               is_E = token == G_TOKEN_FLOAT && (ch == 'e' || ch == 'E');
1480
1481               ch = g_scanner_peek_next_char (scanner);
1482
1483               if (g_scanner_char_2_num (ch, 36) >= 0 ||
1484                   (config->scan_float && ch == '.') ||
1485                   (is_E && (ch == '+' || ch == '-')))
1486                 {
1487                   ch = g_scanner_get_char (scanner, line_p, position_p);
1488
1489                   switch (ch)
1490                     {
1491                     case '.':
1492                       if (token != G_TOKEN_INT && token != G_TOKEN_OCTAL)
1493                         {
1494                           value.v_error = token == G_TOKEN_FLOAT ? G_ERR_FLOAT_MALFORMED : G_ERR_FLOAT_RADIX;
1495                           token = G_TOKEN_ERROR;
1496                           in_number = FALSE;
1497                         }
1498                       else
1499                         {
1500                           token = G_TOKEN_FLOAT;
1501                           gstring = g_string_append_c (gstring, ch);
1502                         }
1503                       break;
1504
1505                     case '0':
1506                     case '1':
1507                     case '2':
1508                     case '3':
1509                     case '4':
1510                     case '5':
1511                     case '6':
1512                     case '7':
1513                     case '8':
1514                     case '9':
1515                       gstring = g_string_append_c (gstring, ch);
1516                       break;
1517
1518                     case '-':
1519                     case '+':
1520                       if (token != G_TOKEN_FLOAT)
1521                         {
1522                           token = G_TOKEN_ERROR;
1523                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1524                           in_number = FALSE;
1525                         }
1526                       else
1527                         gstring = g_string_append_c (gstring, ch);
1528                       break;
1529
1530                     case 'e':
1531                     case 'E':
1532                       if ((token != G_TOKEN_HEX && !config->scan_float) ||
1533                           (token != G_TOKEN_HEX &&
1534                            token != G_TOKEN_OCTAL &&
1535                            token != G_TOKEN_FLOAT &&
1536                            token != G_TOKEN_INT))
1537                         {
1538                           token = G_TOKEN_ERROR;
1539                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1540                           in_number = FALSE;
1541                         }
1542                       else
1543                         {
1544                           if (token != G_TOKEN_HEX)
1545                             token = G_TOKEN_FLOAT;
1546                           gstring = g_string_append_c (gstring, ch);
1547                         }
1548                       break;
1549
1550                     default:
1551                       if (token != G_TOKEN_HEX)
1552                         {
1553                           token = G_TOKEN_ERROR;
1554                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1555                           in_number = FALSE;
1556                         }
1557                       else
1558                         gstring = g_string_append_c (gstring, ch);
1559                       break;
1560                     }
1561                 }
1562               else
1563                 in_number = FALSE;
1564             }
1565           while (in_number);
1566
1567           endptr = NULL;
1568           switch (token)
1569             {
1570             case G_TOKEN_BINARY:
1571               value.v_binary = strtol (gstring->str, &endptr, 2);
1572               break;
1573
1574             case G_TOKEN_OCTAL:
1575               value.v_octal = strtol (gstring->str, &endptr, 8);
1576               break;
1577
1578             case G_TOKEN_INT:
1579               value.v_int = strtol (gstring->str, &endptr, 10);
1580               break;
1581
1582             case G_TOKEN_FLOAT:
1583               value.v_float = g_strtod (gstring->str, &endptr);
1584               break;
1585
1586             case G_TOKEN_HEX:
1587               value.v_hex = strtol (gstring->str, &endptr, 16);
1588               break;
1589
1590             default:
1591               break;
1592             }
1593           if (endptr && *endptr)
1594             {
1595               token = G_TOKEN_ERROR;
1596               if (*endptr == 'e' || *endptr == 'E')
1597                 value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1598               else
1599                 value.v_error = G_ERR_DIGIT_RADIX;
1600             }
1601           g_string_free (gstring, TRUE);
1602           gstring = NULL;
1603           ch = 0;
1604         } /* number_parsing:... */
1605         break;
1606
1607         default:
1608         default_case:
1609         {
1610           if (config->cpair_comment_single &&
1611               ch == config->cpair_comment_single[0])
1612             {
1613               token = G_TOKEN_COMMENT_SINGLE;
1614               in_comment_single = TRUE;
1615               gstring = g_string_new ("");
1616               while ((ch = g_scanner_get_char (scanner,
1617                                                line_p,
1618                                                position_p)) != 0)
1619                 {
1620                   if (ch == config->cpair_comment_single[1])
1621                     {
1622                       in_comment_single = FALSE;
1623                       ch = 0;
1624                       break;
1625                     }
1626
1627                   gstring = g_string_append_c (gstring, ch);
1628                   ch = 0;
1629                 }
1630             }
1631           else if (config->scan_identifier && ch &&
1632                    strchr (config->cset_identifier_first, ch))
1633             {
1634             identifier_precedence:
1635
1636               if (config->cset_identifier_nth && ch &&
1637                   strchr (config->cset_identifier_nth,
1638                           g_scanner_peek_next_char (scanner)))
1639                 {
1640                   token = G_TOKEN_IDENTIFIER;
1641                   gstring = g_string_new ("");
1642                   gstring = g_string_append_c (gstring, ch);
1643                   do
1644                     {
1645                       ch = g_scanner_get_char (scanner, line_p, position_p);
1646                       gstring = g_string_append_c (gstring, ch);
1647                       ch = g_scanner_peek_next_char (scanner);
1648                     }
1649                   while (ch && strchr (config->cset_identifier_nth, ch));
1650                   ch = 0;
1651                 }
1652               else if (config->scan_identifier_1char)
1653                 {
1654                   token = G_TOKEN_IDENTIFIER;
1655                   value.v_identifier = g_new0 (gchar, 2);
1656                   value.v_identifier[0] = ch;
1657                   ch = 0;
1658                 }
1659             }
1660           if (ch)
1661             {
1662               if (config->char_2_token)
1663                 token = ch;
1664               else
1665                 {
1666                   token = G_TOKEN_CHAR;
1667                   value.v_char = ch;
1668                 }
1669               ch = 0;
1670             }
1671         } /* default_case:... */
1672         break;
1673         }
1674       g_assert (ch == 0 && token != G_TOKEN_NONE); /* paranoid */
1675     }
1676   while (ch != 0);
1677
1678   if (in_comment_multi || in_comment_single ||
1679       in_string_sq || in_string_dq)
1680     {
1681       token = G_TOKEN_ERROR;
1682       if (gstring)
1683         {
1684           g_string_free (gstring, TRUE);
1685           gstring = NULL;
1686         }
1687       (*position_p)++;
1688       if (in_comment_multi || in_comment_single)
1689         value.v_error = G_ERR_UNEXP_EOF_IN_COMMENT;
1690       else /* (in_string_sq || in_string_dq) */
1691         value.v_error = G_ERR_UNEXP_EOF_IN_STRING;
1692     }
1693
1694   if (gstring)
1695     {
1696       value.v_string = gstring->str;
1697       g_string_free (gstring, FALSE);
1698       gstring = NULL;
1699     }
1700
1701   if (token == G_TOKEN_IDENTIFIER)
1702     {
1703       if (config->scan_symbols)
1704         {
1705           GScannerKey *key;
1706           guint scope_id;
1707
1708           scope_id = scanner->scope_id;
1709           key = g_scanner_lookup_internal (scanner, scope_id, value.v_identifier);
1710           if (!key && scope_id && scanner->config->scope_0_fallback)
1711             key = g_scanner_lookup_internal (scanner, 0, value.v_identifier);
1712
1713           if (key)
1714             {
1715               g_free (value.v_identifier);
1716               token = G_TOKEN_SYMBOL;
1717               value.v_symbol = key->value;
1718             }
1719         }
1720
1721       if (token == G_TOKEN_IDENTIFIER &&
1722           config->scan_identifier_NULL &&
1723           strlen (value.v_identifier) == 4)
1724         {
1725           gchar *null_upper = "NULL";
1726           gchar *null_lower = "null";
1727
1728           if (scanner->config->case_sensitive)
1729             {
1730               if (value.v_identifier[0] == null_upper[0] &&
1731                   value.v_identifier[1] == null_upper[1] &&
1732                   value.v_identifier[2] == null_upper[2] &&
1733                   value.v_identifier[3] == null_upper[3])
1734                 token = G_TOKEN_IDENTIFIER_NULL;
1735             }
1736           else
1737             {
1738               if ((value.v_identifier[0] == null_upper[0] ||
1739                    value.v_identifier[0] == null_lower[0]) &&
1740                   (value.v_identifier[1] == null_upper[1] ||
1741                    value.v_identifier[1] == null_lower[1]) &&
1742                   (value.v_identifier[2] == null_upper[2] ||
1743                    value.v_identifier[2] == null_lower[2]) &&
1744                   (value.v_identifier[3] == null_upper[3] ||
1745                    value.v_identifier[3] == null_lower[3]))
1746                 token = G_TOKEN_IDENTIFIER_NULL;
1747             }
1748         }
1749     }
1750
1751   *token_p = token;
1752   *value_p = value;
1753 }