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 #include "config.h"
  35
  36 #include <errno.h>
  37 #include <stdlib.h>
  38 #include <stdarg.h>
  39 #include <string.h>
  40 #include <stdio.h>
  41 #ifdef HAVE_UNISTD_H
  42 #include <unistd.h>
  43 #endif
  44
  45 #ifdef G_OS_WIN32
  46 #include <io.h> /* For _read() */
  47 #endif
  48
  49 #include "gscanner.h"
  50
  51 #include "gprintfint.h"
  52 #include "gstrfuncs.h"
  53 #include "gstring.h"
  54 #include "gtestutils.h"
  55
  56 /* --- defines --- */
  57 #define to_lower(c)                             ( \
  58         (guchar) (                                                      \
  59           ( (((guchar)(c))>='A' && ((guchar)(c))<='Z') * ('a'-'A') ) |  \
  60           ( (((guchar)(c))>=192 && ((guchar)(c))<=214) * (224-192) ) |  \
  61           ( (((guchar)(c))>=216 && ((guchar)(c))<=222) * (248-216) ) |  \
  62           ((guchar)(c))                                                 \
  63         )                                                               \
  64 )
  65 #define READ_BUFFER_SIZE        (4000)
  66
  67
  68 /* --- typedefs --- */
  69 typedef struct  _GScannerKey    GScannerKey;
  70
  71 struct  _GScannerKey
  72 {
  73   guint          scope_id;
  74   gchar         *symbol;
  75   gpointer       value;
  76 };
  77
  78
  79 /* --- variables --- */
  80 static const GScannerConfig g_scanner_config_template =
  81 {
  82   (
  83    " \t\r\n"
  84    )                    /* cset_skip_characters */,
  85   (
  86    G_CSET_a_2_z
  87    "_"
  88    G_CSET_A_2_Z
  89    )                    /* cset_identifier_first */,
  90   (
  91    G_CSET_a_2_z
  92    "_"
  93    G_CSET_A_2_Z
  94    G_CSET_DIGITS
  95    G_CSET_LATINS
  96    G_CSET_LATINC
  97    )                    /* cset_identifier_nth */,
  98   ( "#\n" )             /* cpair_comment_single */,
  99
 100   FALSE                 /* case_sensitive */,
 101
 102   TRUE                  /* skip_comment_multi */,
 103   TRUE                  /* skip_comment_single */,
 104   TRUE                  /* scan_comment_multi */,
 105   TRUE                  /* scan_identifier */,
 106   FALSE                 /* scan_identifier_1char */,
 107   FALSE                 /* scan_identifier_NULL */,
 108   TRUE                  /* scan_symbols */,
 109   FALSE                 /* scan_binary */,
 110   TRUE                  /* scan_octal */,
 111   TRUE                  /* scan_float */,
 112   TRUE                  /* scan_hex */,
 113   FALSE                 /* scan_hex_dollar */,
 114   TRUE                  /* scan_string_sq */,
 115   TRUE                  /* scan_string_dq */,
 116   TRUE                  /* numbers_2_int */,
 117   FALSE                 /* int_2_float */,
 118   FALSE                 /* identifier_2_string */,
 119   TRUE                  /* char_2_token */,
 120   FALSE                 /* symbol_2_token */,
 121   FALSE                 /* scope_0_fallback */,
 122   FALSE                 /* store_int64 */,
 123 };
 124
 125
 126 /* --- prototypes --- */
 127 static inline
 128 GScannerKey*    g_scanner_lookup_internal (GScanner     *scanner,
 129                                            guint         scope_id,
 130                                            const gchar  *symbol);
 131 static gboolean g_scanner_key_equal       (gconstpointer v1,
 132                                            gconstpointer v2);
 133 static guint    g_scanner_key_hash        (gconstpointer v);
 134 static void     g_scanner_get_token_ll    (GScanner     *scanner,
 135                                            GTokenType   *token_p,
 136                                            GTokenValue  *value_p,
 137                                            guint        *line_p,
 138                                            guint        *position_p);
 139 static void     g_scanner_get_token_i     (GScanner     *scanner,
 140                                            GTokenType   *token_p,
 141                                            GTokenValue  *value_p,
 142                                            guint        *line_p,
 143                                            guint        *position_p);
 144
 145 static guchar   g_scanner_peek_next_char  (GScanner     *scanner);
 146 static guchar   g_scanner_get_char        (GScanner     *scanner,
 147                                            guint        *line_p,
 148                                            guint        *position_p);
 149 static void     g_scanner_msg_handler     (GScanner     *scanner,
 150                                            gchar        *message,
 151                                            gboolean      is_error);
 152
 153
 154 /* --- functions --- */
 155 static inline gint
 156 g_scanner_char_2_num (guchar    c,
 157                       guchar    base)
 158 {
 159   if (c >= '0' && c <= '9')
 160     c -= '0';
 161   else if (c >= 'A' && c <= 'Z')
 162     c -= 'A' - 10;
 163   else if (c >= 'a' && c <= 'z')
 164     c -= 'a' - 10;
 165   else
 166     return -1;
 167
 168   if (c < base)
 169     return c;
 170
 171   return -1;
 172 }
 173
 174 GScanner*
 175 g_scanner_new (const GScannerConfig *config_templ)
 176 {
 177   GScanner *scanner;
 178
 179   if (!config_templ)
 180     config_templ = &g_scanner_config_template;
 181
 182   scanner = g_new0 (GScanner, 1);
 183
 184   scanner->user_data = NULL;
 185   scanner->max_parse_errors = 1;
 186   scanner->parse_errors = 0;
 187   scanner->input_name = NULL;
 188   g_datalist_init (&scanner->qdata);
 189
 190   scanner->config = g_new0 (GScannerConfig, 1);
 191
 192   scanner->config->case_sensitive        = config_templ->case_sensitive;
 193   scanner->config->cset_skip_characters  = config_templ->cset_skip_characters;
 194   if (!scanner->config->cset_skip_characters)
 195     scanner->config->cset_skip_characters = "";
 196   scanner->config->cset_identifier_first = config_templ->cset_identifier_first;
 197   scanner->config->cset_identifier_nth   = config_templ->cset_identifier_nth;
 198   scanner->config->cpair_comment_single  = config_templ->cpair_comment_single;
 199   scanner->config->skip_comment_multi    = config_templ->skip_comment_multi;
 200   scanner->config->skip_comment_single   = config_templ->skip_comment_single;
 201   scanner->config->scan_comment_multi    = config_templ->scan_comment_multi;
 202   scanner->config->scan_identifier       = config_templ->scan_identifier;
 203   scanner->config->scan_identifier_1char = config_templ->scan_identifier_1char;
 204   scanner->config->scan_identifier_NULL  = config_templ->scan_identifier_NULL;
 205   scanner->config->scan_symbols          = config_templ->scan_symbols;
 206   scanner->config->scan_binary           = config_templ->scan_binary;
 207   scanner->config->scan_octal            = config_templ->scan_octal;
 208   scanner->config->scan_float            = config_templ->scan_float;
 209   scanner->config->scan_hex              = config_templ->scan_hex;
 210   scanner->config->scan_hex_dollar       = config_templ->scan_hex_dollar;
 211   scanner->config->scan_string_sq        = config_templ->scan_string_sq;
 212   scanner->config->scan_string_dq        = config_templ->scan_string_dq;
 213   scanner->config->numbers_2_int         = config_templ->numbers_2_int;
 214   scanner->config->int_2_float           = config_templ->int_2_float;
 215   scanner->config->identifier_2_string   = config_templ->identifier_2_string;
 216   scanner->config->char_2_token          = config_templ->char_2_token;
 217   scanner->config->symbol_2_token        = config_templ->symbol_2_token;
 218   scanner->config->scope_0_fallback      = config_templ->scope_0_fallback;
 219   scanner->config->store_int64           = config_templ->store_int64;
 220
 221   scanner->token = G_TOKEN_NONE;
 222   scanner->value.v_int64 = 0;
 223   scanner->line = 1;
 224   scanner->position = 0;
 225
 226   scanner->next_token = G_TOKEN_NONE;
 227   scanner->next_value.v_int64 = 0;
 228   scanner->next_line = 1;
 229   scanner->next_position = 0;
 230
 231   scanner->symbol_table = g_hash_table_new (g_scanner_key_hash, g_scanner_key_equal);
 232   scanner->input_fd = -1;
 233   scanner->text = NULL;
 234   scanner->text_end = NULL;
 235   scanner->buffer = NULL;
 236   scanner->scope_id = 0;
 237
 238   scanner->msg_handler = g_scanner_msg_handler;
 239
 240   return scanner;
 241 }
 242
 243 static inline void
 244 g_scanner_free_value (GTokenType     *token_p,
 245                       GTokenValue     *value_p)
 246 {
 247   switch (*token_p)
 248     {
 249     case G_TOKEN_STRING:
 250     case G_TOKEN_IDENTIFIER:
 251     case G_TOKEN_IDENTIFIER_NULL:
 252     case G_TOKEN_COMMENT_SINGLE:
 253     case G_TOKEN_COMMENT_MULTI:
 254       g_free (value_p->v_string);
 255       break;
 256
 257     default:
 258       break;
 259     }
 260
 261   *token_p = G_TOKEN_NONE;
 262 }
 263
 264 static void
 265 g_scanner_destroy_symbol_table_entry (gpointer _key,
 266                                       gpointer _value,
 267                                       gpointer _data)
 268 {
 269   GScannerKey *key = _key;
 270
 271   g_free (key->symbol);
 272   g_free (key);
 273 }
 274
 275 void
 276 g_scanner_destroy (GScanner     *scanner)
 277 {
 278   g_return_if_fail (scanner != NULL);
 279
 280   g_datalist_clear (&scanner->qdata);
 281   g_hash_table_foreach (scanner->symbol_table,
 282                         g_scanner_destroy_symbol_table_entry, NULL);
 283   g_hash_table_destroy (scanner->symbol_table);
 284   g_scanner_free_value (&scanner->token, &scanner->value);
 285   g_scanner_free_value (&scanner->next_token, &scanner->next_value);
 286   g_free (scanner->config);
 287   g_free (scanner->buffer);
 288   g_free (scanner);
 289 }
 290
 291 static void
 292 g_scanner_msg_handler (GScanner         *scanner,
 293                        gchar            *message,
 294                        gboolean         is_error)
 295 {
 296   g_return_if_fail (scanner != NULL);
 297
 298   _g_fprintf (stderr, "%s:%d: ",
 299               scanner->input_name ? scanner->input_name : "<memory>",
 300               scanner->line);
 301   if (is_error)
 302     _g_fprintf (stderr, "error: ");
 303   _g_fprintf (stderr, "%s\n", message);
 304 }
 305
 306 void
 307 g_scanner_error (GScanner       *scanner,
 308                  const gchar    *format,
 309                  ...)
 310 {
 311   g_return_if_fail (scanner != NULL);
 312   g_return_if_fail (format != NULL);
 313
 314   scanner->parse_errors++;
 315
 316   if (scanner->msg_handler)
 317     {
 318       va_list args;
 319       gchar *string;
 320
 321       va_start (args, format);
 322       string = g_strdup_vprintf (format, args);
 323       va_end (args);
 324
 325       scanner->msg_handler (scanner, string, TRUE);
 326
 327       g_free (string);
 328     }
 329 }
 330
 331 void
 332 g_scanner_warn (GScanner       *scanner,
 333                 const gchar    *format,
 334                 ...)
 335 {
 336   g_return_if_fail (scanner != NULL);
 337   g_return_if_fail (format != NULL);
 338
 339   if (scanner->msg_handler)
 340     {
 341       va_list args;
 342       gchar *string;
 343
 344       va_start (args, format);
 345       string = g_strdup_vprintf (format, args);
 346       va_end (args);
 347
 348       scanner->msg_handler (scanner, string, FALSE);
 349
 350       g_free (string);
 351     }
 352 }
 353
 354 static gboolean
 355 g_scanner_key_equal (gconstpointer v1,
 356                      gconstpointer v2)
 357 {
 358   const GScannerKey *key1 = v1;
 359   const GScannerKey *key2 = v2;
 360
 361   return (key1->scope_id == key2->scope_id) && (strcmp (key1->symbol, key2->symbol) == 0);
 362 }
 363
 364 static guint
 365 g_scanner_key_hash (gconstpointer v)
 366 {
 367   const GScannerKey *key = v;
 368   gchar *c;
 369   guint h;
 370
 371   h = key->scope_id;
 372   for (c = key->symbol; *c; c++)
 373     h = (h << 5) - h + *c;
 374
 375   return h;
 376 }
 377
 378 static inline GScannerKey*
 379 g_scanner_lookup_internal (GScanner     *scanner,
 380                            guint         scope_id,
 381                            const gchar  *symbol)
 382 {
 383   GScannerKey   *key_p;
 384   GScannerKey key;
 385
 386   key.scope_id = scope_id;
 387
 388   if (!scanner->config->case_sensitive)
 389     {
 390       gchar *d;
 391       const gchar *c;
 392
 393       key.symbol = g_new (gchar, strlen (symbol) + 1);
 394       for (d = key.symbol, c = symbol; *c; c++, d++)
 395         *d = to_lower (*c);
 396       *d = 0;
 397       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 398       g_free (key.symbol);
 399     }
 400   else
 401     {
 402       key.symbol = (gchar*) symbol;
 403       key_p = g_hash_table_lookup (scanner->symbol_table, &key);
 404     }
 405
 406   return key_p;
 407 }
 408
 409 void
 410 g_scanner_scope_add_symbol (GScanner    *scanner,
 411                             guint        scope_id,
 412                             const gchar *symbol,
 413                             gpointer     value)
 414 {
 415   GScannerKey   *key;
 416
 417   g_return_if_fail (scanner != NULL);
 418   g_return_if_fail (symbol != NULL);
 419
 420   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 421
 422   if (!key)
 423     {
 424       key = g_new (GScannerKey, 1);
 425       key->scope_id = scope_id;
 426       key->symbol = g_strdup (symbol);
 427       key->value = value;
 428       if (!scanner->config->case_sensitive)
 429         {
 430           gchar *c;
 431
 432           c = key->symbol;
 433           while (*c != 0)
 434             {
 435               *c = to_lower (*c);
 436               c++;
 437             }
 438         }
 439       g_hash_table_insert (scanner->symbol_table, key, key);
 440     }
 441   else
 442     key->value = value;
 443 }
 444
 445 void
 446 g_scanner_scope_remove_symbol (GScanner    *scanner,
 447                                guint        scope_id,
 448                                const gchar *symbol)
 449 {
 450   GScannerKey   *key;
 451
 452   g_return_if_fail (scanner != NULL);
 453   g_return_if_fail (symbol != NULL);
 454
 455   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 456
 457   if (key)
 458     {
 459       g_hash_table_remove (scanner->symbol_table, key);
 460       g_free (key->symbol);
 461       g_free (key);
 462     }
 463 }
 464
 465 gpointer
 466 g_scanner_lookup_symbol (GScanner       *scanner,
 467                          const gchar    *symbol)
 468 {
 469   GScannerKey   *key;
 470   guint scope_id;
 471
 472   g_return_val_if_fail (scanner != NULL, NULL);
 473
 474   if (!symbol)
 475     return NULL;
 476
 477   scope_id = scanner->scope_id;
 478   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 479   if (!key && scope_id && scanner->config->scope_0_fallback)
 480     key = g_scanner_lookup_internal (scanner, 0, symbol);
 481
 482   if (key)
 483     return key->value;
 484   else
 485     return NULL;
 486 }
 487
 488 gpointer
 489 g_scanner_scope_lookup_symbol (GScanner       *scanner,
 490                                guint           scope_id,
 491                                const gchar    *symbol)
 492 {
 493   GScannerKey   *key;
 494
 495   g_return_val_if_fail (scanner != NULL, NULL);
 496
 497   if (!symbol)
 498     return NULL;
 499
 500   key = g_scanner_lookup_internal (scanner, scope_id, symbol);
 501
 502   if (key)
 503     return key->value;
 504   else
 505     return NULL;
 506 }
 507
 508 guint
 509 g_scanner_set_scope (GScanner       *scanner,
 510                      guint           scope_id)
 511 {
 512   guint old_scope_id;
 513
 514   g_return_val_if_fail (scanner != NULL, 0);
 515
 516   old_scope_id = scanner->scope_id;
 517   scanner->scope_id = scope_id;
 518
 519   return old_scope_id;
 520 }
 521
 522 static void
 523 g_scanner_foreach_internal (gpointer  _key,
 524                             gpointer  _value,
 525                             gpointer  _user_data)
 526 {
 527   GScannerKey *key;
 528   gpointer *d;
 529   GHFunc func;
 530   gpointer user_data;
 531   guint *scope_id;
 532
 533   d = _user_data;
 534   func = (GHFunc) d[0];
 535   user_data = d[1];
 536   scope_id = d[2];
 537   key = _value;
 538
 539   if (key->scope_id == *scope_id)
 540     func (key->symbol, key->value, user_data);
 541 }
 542
 543 void
 544 g_scanner_scope_foreach_symbol (GScanner       *scanner,
 545                                 guint           scope_id,
 546                                 GHFunc          func,
 547                                 gpointer        user_data)
 548 {
 549   gpointer d[3];
 550
 551   g_return_if_fail (scanner != NULL);
 552
 553   d[0] = (gpointer) func;
 554   d[1] = user_data;
 555   d[2] = &scope_id;
 556
 557   g_hash_table_foreach (scanner->symbol_table, g_scanner_foreach_internal, d);
 558 }
 559
 560 GTokenType
 561 g_scanner_peek_next_token (GScanner     *scanner)
 562 {
 563   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 564
 565   if (scanner->next_token == G_TOKEN_NONE)
 566     {
 567       scanner->next_line = scanner->line;
 568       scanner->next_position = scanner->position;
 569       g_scanner_get_token_i (scanner,
 570                              &scanner->next_token,
 571                              &scanner->next_value,
 572                              &scanner->next_line,
 573                              &scanner->next_position);
 574     }
 575
 576   return scanner->next_token;
 577 }
 578
 579 GTokenType
 580 g_scanner_get_next_token (GScanner      *scanner)
 581 {
 582   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 583
 584   if (scanner->next_token != G_TOKEN_NONE)
 585     {
 586       g_scanner_free_value (&scanner->token, &scanner->value);
 587
 588       scanner->token = scanner->next_token;
 589       scanner->value = scanner->next_value;
 590       scanner->line = scanner->next_line;
 591       scanner->position = scanner->next_position;
 592       scanner->next_token = G_TOKEN_NONE;
 593     }
 594   else
 595     g_scanner_get_token_i (scanner,
 596                            &scanner->token,
 597                            &scanner->value,
 598                            &scanner->line,
 599                            &scanner->position);
 600
 601   return scanner->token;
 602 }
 603
 604 GTokenType
 605 g_scanner_cur_token (GScanner *scanner)
 606 {
 607   g_return_val_if_fail (scanner != NULL, G_TOKEN_EOF);
 608
 609   return scanner->token;
 610 }
 611
 612 GTokenValue
 613 g_scanner_cur_value (GScanner *scanner)
 614 {
 615   GTokenValue v;
 616
 617   v.v_int64 = 0;
 618
 619   g_return_val_if_fail (scanner != NULL, v);
 620
 621   /* MSC isn't capable of handling return scanner->value; ? */
 622
 623   v = scanner->value;
 624
 625   return v;
 626 }
 627
 628 guint
 629 g_scanner_cur_line (GScanner *scanner)
 630 {
 631   g_return_val_if_fail (scanner != NULL, 0);
 632
 633   return scanner->line;
 634 }
 635
 636 guint
 637 g_scanner_cur_position (GScanner *scanner)
 638 {
 639   g_return_val_if_fail (scanner != NULL, 0);
 640
 641   return scanner->position;
 642 }
 643
 644 gboolean
 645 g_scanner_eof (GScanner *scanner)
 646 {
 647   g_return_val_if_fail (scanner != NULL, TRUE);
 648
 649   return scanner->token == G_TOKEN_EOF || scanner->token == G_TOKEN_ERROR;
 650 }
 651
 652 void
 653 g_scanner_input_file (GScanner *scanner,
 654                       gint      input_fd)
 655 {
 656   g_return_if_fail (scanner != NULL);
 657   g_return_if_fail (input_fd >= 0);
 658
 659   if (scanner->input_fd >= 0)
 660     g_scanner_sync_file_offset (scanner);
 661
 662   scanner->token = G_TOKEN_NONE;
 663   scanner->value.v_int64 = 0;
 664   scanner->line = 1;
 665   scanner->position = 0;
 666   scanner->next_token = G_TOKEN_NONE;
 667
 668   scanner->input_fd = input_fd;
 669   scanner->text = NULL;
 670   scanner->text_end = NULL;
 671
 672   if (!scanner->buffer)
 673     scanner->buffer = g_new (gchar, READ_BUFFER_SIZE + 1);
 674 }
 675
 676 void
 677 g_scanner_input_text (GScanner    *scanner,
 678                       const gchar *text,
 679                       guint        text_len)
 680 {
 681   g_return_if_fail (scanner != NULL);
 682   if (text_len)
 683     g_return_if_fail (text != NULL);
 684   else
 685     text = NULL;
 686
 687   if (scanner->input_fd >= 0)
 688     g_scanner_sync_file_offset (scanner);
 689
 690   scanner->token = G_TOKEN_NONE;
 691   scanner->value.v_int64 = 0;
 692   scanner->line = 1;
 693   scanner->position = 0;
 694   scanner->next_token = G_TOKEN_NONE;
 695
 696   scanner->input_fd = -1;
 697   scanner->text = text;
 698   scanner->text_end = text + text_len;
 699
 700   if (scanner->buffer)
 701     {
 702       g_free (scanner->buffer);
 703       scanner->buffer = NULL;
 704     }
 705 }
 706
 707 static guchar
 708 g_scanner_peek_next_char (GScanner *scanner)
 709 {
 710   if (scanner->text < scanner->text_end)
 711     {
 712       return *scanner->text;
 713     }
 714   else if (scanner->input_fd >= 0)
 715     {
 716       gint count;
 717       gchar *buffer;
 718
 719       buffer = scanner->buffer;
 720       do
 721         {
 722           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 723         }
 724       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 725
 726       if (count < 1)
 727         {
 728           scanner->input_fd = -1;
 729
 730           return 0;
 731         }
 732       else
 733         {
 734           scanner->text = buffer;
 735           scanner->text_end = buffer + count;
 736
 737           return *buffer;
 738         }
 739     }
 740   else
 741     return 0;
 742 }
 743
 744 void
 745 g_scanner_sync_file_offset (GScanner *scanner)
 746 {
 747   g_return_if_fail (scanner != NULL);
 748
 749   /* for file input, rewind the filedescriptor to the current
 750    * buffer position and blow the file read ahead buffer. useful
 751    * for third party uses of our file descriptor, which hooks
 752    * onto the current scanning position.
 753    */
 754
 755   if (scanner->input_fd >= 0 && scanner->text_end > scanner->text)
 756     {
 757       gint buffered;
 758
 759       buffered = scanner->text_end - scanner->text;
 760       if (lseek (scanner->input_fd, - buffered, SEEK_CUR) >= 0)
 761         {
 762           /* we succeeded, blow our buffer's contents now */
 763           scanner->text = NULL;
 764           scanner->text_end = NULL;
 765         }
 766       else
 767         errno = 0;
 768     }
 769 }
 770
 771 static guchar
 772 g_scanner_get_char (GScanner    *scanner,
 773                     guint       *line_p,
 774                     guint       *position_p)
 775 {
 776   guchar fchar;
 777
 778   if (scanner->text < scanner->text_end)
 779     fchar = *(scanner->text++);
 780   else if (scanner->input_fd >= 0)
 781     {
 782       gint count;
 783       gchar *buffer;
 784
 785       buffer = scanner->buffer;
 786       do
 787         {
 788           count = read (scanner->input_fd, buffer, READ_BUFFER_SIZE);
 789         }
 790       while (count == -1 && (errno == EINTR || errno == EAGAIN));
 791
 792       if (count < 1)
 793         {
 794           scanner->input_fd = -1;
 795           fchar = 0;
 796         }
 797       else
 798         {
 799           scanner->text = buffer + 1;
 800           scanner->text_end = buffer + count;
 801           fchar = *buffer;
 802           if (!fchar)
 803             {
 804               g_scanner_sync_file_offset (scanner);
 805               scanner->text_end = scanner->text;
 806               scanner->input_fd = -1;
 807             }
 808         }
 809     }
 810   else
 811     fchar = 0;
 812
 813   if (fchar == '\n')
 814     {
 815       (*position_p) = 0;
 816       (*line_p)++;
 817     }
 818   else if (fchar)
 819     {
 820       (*position_p)++;
 821     }
 822
 823   return fchar;
 824 }
 825
 826 void
 827 g_scanner_unexp_token (GScanner         *scanner,
 828                        GTokenType        expected_token,
 829                        const gchar      *identifier_spec,
 830                        const gchar      *symbol_spec,
 831                        const gchar      *symbol_name,
 832                        const gchar      *message,
 833                        gint              is_error)
 834 {
 835   gchar *token_string;
 836   guint token_string_len;
 837   gchar *expected_string;
 838   guint expected_string_len;
 839   gchar *message_prefix;
 840   gboolean print_unexp;
 841   void (*msg_handler)   (GScanner*, const gchar*, ...);
 842
 843   g_return_if_fail (scanner != NULL);
 844
 845   if (is_error)
 846     msg_handler = g_scanner_error;
 847   else
 848     msg_handler = g_scanner_warn;
 849
 850   if (!identifier_spec)
 851     identifier_spec = "identifier";
 852   if (!symbol_spec)
 853     symbol_spec = "symbol";
 854
 855   token_string_len = 56;
 856   token_string = g_new (gchar, token_string_len + 1);
 857   expected_string_len = 64;
 858   expected_string = g_new (gchar, expected_string_len + 1);
 859   print_unexp = TRUE;
 860
 861   switch (scanner->token)
 862     {
 863     case G_TOKEN_EOF:
 864       _g_snprintf (token_string, token_string_len, "end of file");
 865       break;
 866
 867     default:
 868       if (scanner->token >= 1 && scanner->token <= 255)
 869         {
 870           if ((scanner->token >= ' ' && scanner->token <= '~') ||
 871               strchr (scanner->config->cset_identifier_first, scanner->token) ||
 872               strchr (scanner->config->cset_identifier_nth, scanner->token))
 873             _g_snprintf (token_string, token_string_len, "character `%c'", scanner->token);
 874           else
 875             _g_snprintf (token_string, token_string_len, "character `\\%o'", scanner->token);
 876           break;
 877         }
 878       else if (!scanner->config->symbol_2_token)
 879         {
 880           _g_snprintf (token_string, token_string_len, "(unknown) token <%d>", scanner->token);
 881           break;
 882         }
 883       /* fall through */
 884     case G_TOKEN_SYMBOL:
 885       if (expected_token == G_TOKEN_SYMBOL ||
 886           (scanner->config->symbol_2_token &&
 887            expected_token > G_TOKEN_LAST))
 888         print_unexp = FALSE;
 889       if (symbol_name)
 890         _g_snprintf (token_string,
 891                      token_string_len,
 892                      "%s%s `%s'",
 893                      print_unexp ? "" : "invalid ",
 894                      symbol_spec,
 895                      symbol_name);
 896       else
 897         _g_snprintf (token_string,
 898                      token_string_len,
 899                      "%s%s",
 900                      print_unexp ? "" : "invalid ",
 901                      symbol_spec);
 902       break;
 903
 904     case G_TOKEN_ERROR:
 905       print_unexp = FALSE;
 906       expected_token = G_TOKEN_NONE;
 907       switch (scanner->value.v_error)
 908         {
 909         case G_ERR_UNEXP_EOF:
 910           _g_snprintf (token_string, token_string_len, "scanner: unexpected end of file");
 911           break;
 912
 913         case G_ERR_UNEXP_EOF_IN_STRING:
 914           _g_snprintf (token_string, token_string_len, "scanner: unterminated string constant");
 915           break;
 916
 917         case G_ERR_UNEXP_EOF_IN_COMMENT:
 918           _g_snprintf (token_string, token_string_len, "scanner: unterminated comment");
 919           break;
 920
 921         case G_ERR_NON_DIGIT_IN_CONST:
 922           _g_snprintf (token_string, token_string_len, "scanner: non digit in constant");
 923           break;
 924
 925         case G_ERR_FLOAT_RADIX:
 926           _g_snprintf (token_string, token_string_len, "scanner: invalid radix for floating constant");
 927           break;
 928
 929         case G_ERR_FLOAT_MALFORMED:
 930           _g_snprintf (token_string, token_string_len, "scanner: malformed floating constant");
 931           break;
 932
 933         case G_ERR_DIGIT_RADIX:
 934           _g_snprintf (token_string, token_string_len, "scanner: digit is beyond radix");
 935           break;
 936
 937         case G_ERR_UNKNOWN:
 938         default:
 939           _g_snprintf (token_string, token_string_len, "scanner: unknown error");
 940           break;
 941         }
 942       break;
 943
 944     case G_TOKEN_CHAR:
 945       _g_snprintf (token_string, token_string_len, "character `%c'", scanner->value.v_char);
 946       break;
 947
 948     case G_TOKEN_IDENTIFIER:
 949     case G_TOKEN_IDENTIFIER_NULL:
 950       if (expected_token == G_TOKEN_IDENTIFIER ||
 951           expected_token == G_TOKEN_IDENTIFIER_NULL)
 952         print_unexp = FALSE;
 953       _g_snprintf (token_string,
 954                   token_string_len,
 955                   "%s%s `%s'",
 956                   print_unexp ? "" : "invalid ",
 957                   identifier_spec,
 958                   scanner->token == G_TOKEN_IDENTIFIER ? scanner->value.v_string : "null");
 959       break;
 960
 961     case G_TOKEN_BINARY:
 962     case G_TOKEN_OCTAL:
 963     case G_TOKEN_INT:
 964     case G_TOKEN_HEX:
 965       if (scanner->config->store_int64)
 966         _g_snprintf (token_string, token_string_len, "number `%" G_GUINT64_FORMAT "'", scanner->value.v_int64);
 967       else
 968         _g_snprintf (token_string, token_string_len, "number `%lu'", scanner->value.v_int);
 969       break;
 970
 971     case G_TOKEN_FLOAT:
 972       _g_snprintf (token_string, token_string_len, "number `%.3f'", scanner->value.v_float);
 973       break;
 974
 975     case G_TOKEN_STRING:
 976       if (expected_token == G_TOKEN_STRING)
 977         print_unexp = FALSE;
 978       _g_snprintf (token_string,
 979                    token_string_len,
 980                    "%s%sstring constant \"%s\"",
 981                    print_unexp ? "" : "invalid ",
 982                    scanner->value.v_string[0] == 0 ? "empty " : "",
 983                    scanner->value.v_string);
 984       token_string[token_string_len - 2] = '"';
 985       token_string[token_string_len - 1] = 0;
 986       break;
 987
 988     case G_TOKEN_COMMENT_SINGLE:
 989     case G_TOKEN_COMMENT_MULTI:
 990       _g_snprintf (token_string, token_string_len, "comment");
 991       break;
 992
 993     case G_TOKEN_NONE:
 994       /* somehow the user's parsing code is screwed, there isn't much
 995        * we can do about it.
 996        * Note, a common case to trigger this is
 997        * g_scanner_peek_next_token(); g_scanner_unexp_token();
 998        * without an intermediate g_scanner_get_next_token().
 999        */
1000       g_assert_not_reached ();
1001       break;
1002     }
1003
1004
1005   switch (expected_token)
1006     {
1007       gboolean need_valid;
1008       gchar *tstring;
1009     case G_TOKEN_EOF:
1010       _g_snprintf (expected_string, expected_string_len, "end of file");
1011       break;
1012     default:
1013       if (expected_token >= 1 && expected_token <= 255)
1014         {
1015           if ((expected_token >= ' ' && expected_token <= '~') ||
1016               strchr (scanner->config->cset_identifier_first, expected_token) ||
1017               strchr (scanner->config->cset_identifier_nth, expected_token))
1018             _g_snprintf (expected_string, expected_string_len, "character `%c'", expected_token);
1019           else
1020             _g_snprintf (expected_string, expected_string_len, "character `\\%o'", expected_token);
1021           break;
1022         }
1023       else if (!scanner->config->symbol_2_token)
1024         {
1025           _g_snprintf (expected_string, expected_string_len, "(unknown) token <%d>", expected_token);
1026           break;
1027         }
1028       /* fall through */
1029     case G_TOKEN_SYMBOL:
1030       need_valid = (scanner->token == G_TOKEN_SYMBOL ||
1031                     (scanner->config->symbol_2_token &&
1032                      scanner->token > G_TOKEN_LAST));
1033       _g_snprintf (expected_string,
1034                    expected_string_len,
1035                    "%s%s",
1036                    need_valid ? "valid " : "",
1037                    symbol_spec);
1038       /* FIXME: should we attempt to lookup the symbol_name for symbol_2_token? */
1039       break;
1040     case G_TOKEN_CHAR:
1041       _g_snprintf (expected_string, expected_string_len, "%scharacter",
1042                    scanner->token == G_TOKEN_CHAR ? "valid " : "");
1043       break;
1044     case G_TOKEN_BINARY:
1045       tstring = "binary";
1046       _g_snprintf (expected_string, expected_string_len, "%snumber (%s)",
1047                    scanner->token == expected_token ? "valid " : "", tstring);
1048       break;
1049     case G_TOKEN_OCTAL:
1050       tstring = "octal";
1051       _g_snprintf (expected_string, expected_string_len, "%snumber (%s)",
1052                    scanner->token == expected_token ? "valid " : "", tstring);
1053       break;
1054     case G_TOKEN_INT:
1055       tstring = "integer";
1056       _g_snprintf (expected_string, expected_string_len, "%snumber (%s)",
1057                    scanner->token == expected_token ? "valid " : "", tstring);
1058       break;
1059     case G_TOKEN_HEX:
1060       tstring = "hexadecimal";
1061       _g_snprintf (expected_string, expected_string_len, "%snumber (%s)",
1062                    scanner->token == expected_token ? "valid " : "", tstring);
1063       break;
1064     case G_TOKEN_FLOAT:
1065       tstring = "float";
1066       _g_snprintf (expected_string, expected_string_len, "%snumber (%s)",
1067                    scanner->token == expected_token ? "valid " : "", tstring);
1068       break;
1069     case G_TOKEN_STRING:
1070       _g_snprintf (expected_string,
1071                    expected_string_len,
1072                    "%sstring constant",
1073                    scanner->token == G_TOKEN_STRING ? "valid " : "");
1074       break;
1075     case G_TOKEN_IDENTIFIER:
1076     case G_TOKEN_IDENTIFIER_NULL:
1077       need_valid = (scanner->token == G_TOKEN_IDENTIFIER_NULL ||
1078                     scanner->token == G_TOKEN_IDENTIFIER);
1079       _g_snprintf (expected_string,
1080                    expected_string_len,
1081                    "%s%s",
1082                    need_valid ? "valid " : "",
1083                    identifier_spec);
1084       break;
1085     case G_TOKEN_COMMENT_SINGLE:
1086       tstring = "single-line";
1087       _g_snprintf (expected_string, expected_string_len, "%scomment (%s)",
1088                    scanner->token == expected_token ? "valid " : "", tstring);
1089       break;
1090     case G_TOKEN_COMMENT_MULTI:
1091       tstring = "multi-line";
1092       _g_snprintf (expected_string, expected_string_len, "%scomment (%s)",
1093                    scanner->token == expected_token ? "valid " : "", tstring);
1094       break;
1095     case G_TOKEN_NONE:
1096     case G_TOKEN_ERROR:
1097       /* this is handled upon printout */
1098       break;
1099     }
1100
1101   if (message && message[0] != 0)
1102     message_prefix = " - ";
1103   else
1104     {
1105       message_prefix = "";
1106       message = "";
1107     }
1108   if (expected_token == G_TOKEN_ERROR)
1109     {
1110       msg_handler (scanner,
1111                    "failure around %s%s%s",
1112                    token_string,
1113                    message_prefix,
1114                    message);
1115     }
1116   else if (expected_token == G_TOKEN_NONE)
1117     {
1118       if (print_unexp)
1119         msg_handler (scanner,
1120                      "unexpected %s%s%s",
1121                      token_string,
1122                      message_prefix,
1123                      message);
1124       else
1125         msg_handler (scanner,
1126                      "%s%s%s",
1127                      token_string,
1128                      message_prefix,
1129                      message);
1130     }
1131   else
1132     {
1133       if (print_unexp)
1134         msg_handler (scanner,
1135                      "unexpected %s, expected %s%s%s",
1136                      token_string,
1137                      expected_string,
1138                      message_prefix,
1139                      message);
1140       else
1141         msg_handler (scanner,
1142                      "%s, expected %s%s%s",
1143                      token_string,
1144                      expected_string,
1145                      message_prefix,
1146                      message);
1147     }
1148
1149   g_free (token_string);
1150   g_free (expected_string);
1151 }
1152
1153 static void
1154 g_scanner_get_token_i (GScanner *scanner,
1155                        GTokenType       *token_p,
1156                        GTokenValue      *value_p,
1157                        guint            *line_p,
1158                        guint            *position_p)
1159 {
1160   do
1161     {
1162       g_scanner_free_value (token_p, value_p);
1163       g_scanner_get_token_ll (scanner, token_p, value_p, line_p, position_p);
1164     }
1165   while (((*token_p > 0 && *token_p < 256) &&
1166           strchr (scanner->config->cset_skip_characters, *token_p)) ||
1167          (*token_p == G_TOKEN_CHAR &&
1168           strchr (scanner->config->cset_skip_characters, value_p->v_char)) ||
1169          (*token_p == G_TOKEN_COMMENT_MULTI &&
1170           scanner->config->skip_comment_multi) ||
1171          (*token_p == G_TOKEN_COMMENT_SINGLE &&
1172           scanner->config->skip_comment_single));
1173
1174   switch (*token_p)
1175     {
1176     case G_TOKEN_IDENTIFIER:
1177       if (scanner->config->identifier_2_string)
1178         *token_p = G_TOKEN_STRING;
1179       break;
1180
1181     case G_TOKEN_SYMBOL:
1182       if (scanner->config->symbol_2_token)
1183         *token_p = (GTokenType) value_p->v_symbol;
1184       break;
1185
1186     case G_TOKEN_BINARY:
1187     case G_TOKEN_OCTAL:
1188     case G_TOKEN_HEX:
1189       if (scanner->config->numbers_2_int)
1190         *token_p = G_TOKEN_INT;
1191       break;
1192
1193     default:
1194       break;
1195     }
1196
1197   if (*token_p == G_TOKEN_INT &&
1198       scanner->config->int_2_float)
1199     {
1200       *token_p = G_TOKEN_FLOAT;
1201       if (scanner->config->store_int64)
1202         {
1203 #ifdef _MSC_VER
1204           /* work around error C2520, see gvaluetransform.c */
1205           value_p->v_float = (__int64)value_p->v_int64;
1206 #else
1207           value_p->v_float = value_p->v_int64;
1208 #endif
1209         }
1210       else
1211         value_p->v_float = value_p->v_int;
1212     }
1213
1214   errno = 0;
1215 }
1216
1217 static void
1218 g_scanner_get_token_ll  (GScanner       *scanner,
1219                          GTokenType     *token_p,
1220                          GTokenValue    *value_p,
1221                          guint          *line_p,
1222                          guint          *position_p)
1223 {
1224   GScannerConfig *config;
1225   GTokenType       token;
1226   gboolean         in_comment_multi;
1227   gboolean         in_comment_single;
1228   gboolean         in_string_sq;
1229   gboolean         in_string_dq;
1230   GString         *gstring;
1231   GTokenValue      value;
1232   guchar           ch;
1233
1234   config = scanner->config;
1235   (*value_p).v_int64 = 0;
1236
1237   if ((scanner->text >= scanner->text_end && scanner->input_fd < 0) ||
1238       scanner->token == G_TOKEN_EOF)
1239     {
1240       *token_p = G_TOKEN_EOF;
1241       return;
1242     }
1243
1244   in_comment_multi = FALSE;
1245   in_comment_single = FALSE;
1246   in_string_sq = FALSE;
1247   in_string_dq = FALSE;
1248   gstring = NULL;
1249
1250   do /* while (ch != 0) */
1251     {
1252       gboolean dotted_float = FALSE;
1253
1254       ch = g_scanner_get_char (scanner, line_p, position_p);
1255
1256       value.v_int64 = 0;
1257       token = G_TOKEN_NONE;
1258
1259       /* this is *evil*, but needed ;(
1260        * we first check for identifier first character, because  it
1261        * might interfere with other key chars like slashes or numbers
1262        */
1263       if (config->scan_identifier &&
1264           ch && strchr (config->cset_identifier_first, ch))
1265         goto identifier_precedence;
1266
1267       switch (ch)
1268         {
1269         case 0:
1270           token = G_TOKEN_EOF;
1271           (*position_p)++;
1272           /* ch = 0; */
1273           break;
1274
1275         case '/':
1276           if (!config->scan_comment_multi ||
1277               g_scanner_peek_next_char (scanner) != '*')
1278             goto default_case;
1279           g_scanner_get_char (scanner, line_p, position_p);
1280           token = G_TOKEN_COMMENT_MULTI;
1281           in_comment_multi = TRUE;
1282           gstring = g_string_new (NULL);
1283           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1284             {
1285               if (ch == '*' && g_scanner_peek_next_char (scanner) == '/')
1286                 {
1287                   g_scanner_get_char (scanner, line_p, position_p);
1288                   in_comment_multi = FALSE;
1289                   break;
1290                 }
1291               else
1292                 gstring = g_string_append_c (gstring, ch);
1293             }
1294           ch = 0;
1295           break;
1296
1297         case '\'':
1298           if (!config->scan_string_sq)
1299             goto default_case;
1300           token = G_TOKEN_STRING;
1301           in_string_sq = TRUE;
1302           gstring = g_string_new (NULL);
1303           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1304             {
1305               if (ch == '\'')
1306                 {
1307                   in_string_sq = FALSE;
1308                   break;
1309                 }
1310               else
1311                 gstring = g_string_append_c (gstring, ch);
1312             }
1313           ch = 0;
1314           break;
1315
1316         case '"':
1317           if (!config->scan_string_dq)
1318             goto default_case;
1319           token = G_TOKEN_STRING;
1320           in_string_dq = TRUE;
1321           gstring = g_string_new (NULL);
1322           while ((ch = g_scanner_get_char (scanner, line_p, position_p)) != 0)
1323             {
1324               if (ch == '"')
1325                 {
1326                   in_string_dq = FALSE;
1327                   break;
1328                 }
1329               else
1330                 {
1331                   if (ch == '\\')
1332                     {
1333                       ch = g_scanner_get_char (scanner, line_p, position_p);
1334                       switch (ch)
1335                         {
1336                           guint i;
1337                           guint fchar;
1338
1339                         case 0:
1340                           break;
1341
1342                         case '\\':
1343                           gstring = g_string_append_c (gstring, '\\');
1344                           break;
1345
1346                         case 'n':
1347                           gstring = g_string_append_c (gstring, '\n');
1348                           break;
1349
1350                         case 't':
1351                           gstring = g_string_append_c (gstring, '\t');
1352                           break;
1353
1354                         case 'r':
1355                           gstring = g_string_append_c (gstring, '\r');
1356                           break;
1357
1358                         case 'b':
1359                           gstring = g_string_append_c (gstring, '\b');
1360                           break;
1361
1362                         case 'f':
1363                           gstring = g_string_append_c (gstring, '\f');
1364                           break;
1365
1366                         case '0':
1367                         case '1':
1368                         case '2':
1369                         case '3':
1370                         case '4':
1371                         case '5':
1372                         case '6':
1373                         case '7':
1374                           i = ch - '0';
1375                           fchar = g_scanner_peek_next_char (scanner);
1376                           if (fchar >= '0' && fchar <= '7')
1377                             {
1378                               ch = g_scanner_get_char (scanner, line_p, position_p);
1379                               i = i * 8 + ch - '0';
1380                               fchar = g_scanner_peek_next_char (scanner);
1381                               if (fchar >= '0' && fchar <= '7')
1382                                 {
1383                                   ch = g_scanner_get_char (scanner, line_p, position_p);
1384                                   i = i * 8 + ch - '0';
1385                                 }
1386                             }
1387                           gstring = g_string_append_c (gstring, i);
1388                           break;
1389
1390                         default:
1391                           gstring = g_string_append_c (gstring, ch);
1392                           break;
1393                         }
1394                     }
1395                   else
1396                     gstring = g_string_append_c (gstring, ch);
1397                 }
1398             }
1399           ch = 0;
1400           break;
1401
1402         case '.':
1403           if (!config->scan_float)
1404             goto default_case;
1405           token = G_TOKEN_FLOAT;
1406           dotted_float = TRUE;
1407           ch = g_scanner_get_char (scanner, line_p, position_p);
1408           goto number_parsing;
1409
1410         case '$':
1411           if (!config->scan_hex_dollar)
1412             goto default_case;
1413           token = G_TOKEN_HEX;
1414           ch = g_scanner_get_char (scanner, line_p, position_p);
1415           goto number_parsing;
1416
1417         case '0':
1418           if (config->scan_octal)
1419             token = G_TOKEN_OCTAL;
1420           else
1421             token = G_TOKEN_INT;
1422           ch = g_scanner_peek_next_char (scanner);
1423           if (config->scan_hex && (ch == 'x' || ch == 'X'))
1424             {
1425               token = G_TOKEN_HEX;
1426               g_scanner_get_char (scanner, line_p, position_p);
1427               ch = g_scanner_get_char (scanner, line_p, position_p);
1428               if (ch == 0)
1429                 {
1430                   token = G_TOKEN_ERROR;
1431                   value.v_error = G_ERR_UNEXP_EOF;
1432                   (*position_p)++;
1433                   break;
1434                 }
1435               if (g_scanner_char_2_num (ch, 16) < 0)
1436                 {
1437                   token = G_TOKEN_ERROR;
1438                   value.v_error = G_ERR_DIGIT_RADIX;
1439                   ch = 0;
1440                   break;
1441                 }
1442             }
1443           else if (config->scan_binary && (ch == 'b' || ch == 'B'))
1444             {
1445               token = G_TOKEN_BINARY;
1446               g_scanner_get_char (scanner, line_p, position_p);
1447               ch = g_scanner_get_char (scanner, line_p, position_p);
1448               if (ch == 0)
1449                 {
1450                   token = G_TOKEN_ERROR;
1451                   value.v_error = G_ERR_UNEXP_EOF;
1452                   (*position_p)++;
1453                   break;
1454                 }
1455               if (g_scanner_char_2_num (ch, 10) < 0)
1456                 {
1457                   token = G_TOKEN_ERROR;
1458                   value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1459                   ch = 0;
1460                   break;
1461                 }
1462             }
1463           else
1464             ch = '0';
1465           /* fall through */
1466         case '1':
1467         case '2':
1468         case '3':
1469         case '4':
1470         case '5':
1471         case '6':
1472         case '7':
1473         case '8':
1474         case '9':
1475         number_parsing:
1476         {
1477           gboolean in_number = TRUE;
1478           gchar *endptr;
1479
1480           if (token == G_TOKEN_NONE)
1481             token = G_TOKEN_INT;
1482
1483           gstring = g_string_new (dotted_float ? "0." : "");
1484           gstring = g_string_append_c (gstring, ch);
1485
1486           do /* while (in_number) */
1487             {
1488               gboolean is_E;
1489
1490               is_E = token == G_TOKEN_FLOAT && (ch == 'e' || ch == 'E');
1491
1492               ch = g_scanner_peek_next_char (scanner);
1493
1494               if (g_scanner_char_2_num (ch, 36) >= 0 ||
1495                   (config->scan_float && ch == '.') ||
1496                   (is_E && (ch == '+' || ch == '-')))
1497                 {
1498                   ch = g_scanner_get_char (scanner, line_p, position_p);
1499
1500                   switch (ch)
1501                     {
1502                     case '.':
1503                       if (token != G_TOKEN_INT && token != G_TOKEN_OCTAL)
1504                         {
1505                           value.v_error = token == G_TOKEN_FLOAT ? G_ERR_FLOAT_MALFORMED : G_ERR_FLOAT_RADIX;
1506                           token = G_TOKEN_ERROR;
1507                           in_number = FALSE;
1508                         }
1509                       else
1510                         {
1511                           token = G_TOKEN_FLOAT;
1512                           gstring = g_string_append_c (gstring, ch);
1513                         }
1514                       break;
1515
1516                     case '0':
1517                     case '1':
1518                     case '2':
1519                     case '3':
1520                     case '4':
1521                     case '5':
1522                     case '6':
1523                     case '7':
1524                     case '8':
1525                     case '9':
1526                       gstring = g_string_append_c (gstring, ch);
1527                       break;
1528
1529                     case '-':
1530                     case '+':
1531                       if (token != G_TOKEN_FLOAT)
1532                         {
1533                           token = G_TOKEN_ERROR;
1534                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1535                           in_number = FALSE;
1536                         }
1537                       else
1538                         gstring = g_string_append_c (gstring, ch);
1539                       break;
1540
1541                     case 'e':
1542                     case 'E':
1543                       if ((token != G_TOKEN_HEX && !config->scan_float) ||
1544                           (token != G_TOKEN_HEX &&
1545                            token != G_TOKEN_OCTAL &&
1546                            token != G_TOKEN_FLOAT &&
1547                            token != G_TOKEN_INT))
1548                         {
1549                           token = G_TOKEN_ERROR;
1550                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1551                           in_number = FALSE;
1552                         }
1553                       else
1554                         {
1555                           if (token != G_TOKEN_HEX)
1556                             token = G_TOKEN_FLOAT;
1557                           gstring = g_string_append_c (gstring, ch);
1558                         }
1559                       break;
1560
1561                     default:
1562                       if (token != G_TOKEN_HEX)
1563                         {
1564                           token = G_TOKEN_ERROR;
1565                           value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1566                           in_number = FALSE;
1567                         }
1568                       else
1569                         gstring = g_string_append_c (gstring, ch);
1570                       break;
1571                     }
1572                 }
1573               else
1574                 in_number = FALSE;
1575             }
1576           while (in_number);
1577
1578           endptr = NULL;
1579           if (token == G_TOKEN_FLOAT)
1580             value.v_float = g_strtod (gstring->str, &endptr);
1581           else
1582             {
1583               guint64 ui64 = 0;
1584               switch (token)
1585                 {
1586                 case G_TOKEN_BINARY:
1587                   ui64 = g_ascii_strtoull (gstring->str, &endptr, 2);
1588                   break;
1589                 case G_TOKEN_OCTAL:
1590                   ui64 = g_ascii_strtoull (gstring->str, &endptr, 8);
1591                   break;
1592                 case G_TOKEN_INT:
1593                   ui64 = g_ascii_strtoull (gstring->str, &endptr, 10);
1594                   break;
1595                 case G_TOKEN_HEX:
1596                   ui64 = g_ascii_strtoull (gstring->str, &endptr, 16);
1597                   break;
1598                 default: ;
1599                 }
1600               if (scanner->config->store_int64)
1601                 value.v_int64 = ui64;
1602               else
1603                 value.v_int = ui64;
1604             }
1605           if (endptr && *endptr)
1606             {
1607               token = G_TOKEN_ERROR;
1608               if (*endptr == 'e' || *endptr == 'E')
1609                 value.v_error = G_ERR_NON_DIGIT_IN_CONST;
1610               else
1611                 value.v_error = G_ERR_DIGIT_RADIX;
1612             }
1613           g_string_free (gstring, TRUE);
1614           gstring = NULL;
1615           ch = 0;
1616         } /* number_parsing:... */
1617         break;
1618
1619         default:
1620         default_case:
1621         {
1622           if (config->cpair_comment_single &&
1623               ch == config->cpair_comment_single[0])
1624             {
1625               token = G_TOKEN_COMMENT_SINGLE;
1626               in_comment_single = TRUE;
1627               gstring = g_string_new (NULL);
1628               ch = g_scanner_get_char (scanner, line_p, position_p);
1629               while (ch != 0)
1630                 {
1631                   if (ch == config->cpair_comment_single[1])
1632                     {
1633                       in_comment_single = FALSE;
1634                       ch = 0;
1635                       break;
1636                     }
1637
1638                   gstring = g_string_append_c (gstring, ch);
1639                   ch = g_scanner_get_char (scanner, line_p, position_p);
1640                 }
1641               /* ignore a missing newline at EOF for single line comments */
1642               if (in_comment_single &&
1643                   config->cpair_comment_single[1] == '\n')
1644                 in_comment_single = FALSE;
1645             }
1646           else if (config->scan_identifier && ch &&
1647                    strchr (config->cset_identifier_first, ch))
1648             {
1649             identifier_precedence:
1650
1651               if (config->cset_identifier_nth && ch &&
1652                   strchr (config->cset_identifier_nth,
1653                           g_scanner_peek_next_char (scanner)))
1654                 {
1655                   token = G_TOKEN_IDENTIFIER;
1656                   gstring = g_string_new (NULL);
1657                   gstring = g_string_append_c (gstring, ch);
1658                   do
1659                     {
1660                       ch = g_scanner_get_char (scanner, line_p, position_p);
1661                       gstring = g_string_append_c (gstring, ch);
1662                       ch = g_scanner_peek_next_char (scanner);
1663                     }
1664                   while (ch && strchr (config->cset_identifier_nth, ch));
1665                   ch = 0;
1666                 }
1667               else if (config->scan_identifier_1char)
1668                 {
1669                   token = G_TOKEN_IDENTIFIER;
1670                   value.v_identifier = g_new0 (gchar, 2);
1671                   value.v_identifier[0] = ch;
1672                   ch = 0;
1673                 }
1674             }
1675           if (ch)
1676             {
1677               if (config->char_2_token)
1678                 token = ch;
1679               else
1680                 {
1681                   token = G_TOKEN_CHAR;
1682                   value.v_char = ch;
1683                 }
1684               ch = 0;
1685             }
1686         } /* default_case:... */
1687         break;
1688         }
1689       g_assert (ch == 0 && token != G_TOKEN_NONE); /* paranoid */
1690     }
1691   while (ch != 0);
1692
1693   if (in_comment_multi || in_comment_single ||
1694       in_string_sq || in_string_dq)
1695     {
1696       token = G_TOKEN_ERROR;
1697       if (gstring)
1698         {
1699           g_string_free (gstring, TRUE);
1700           gstring = NULL;
1701         }
1702       (*position_p)++;
1703       if (in_comment_multi || in_comment_single)
1704         value.v_error = G_ERR_UNEXP_EOF_IN_COMMENT;
1705       else /* (in_string_sq || in_string_dq) */
1706         value.v_error = G_ERR_UNEXP_EOF_IN_STRING;
1707     }
1708
1709   if (gstring)
1710     {
1711       value.v_string = g_string_free (gstring, FALSE);
1712       gstring = NULL;
1713     }
1714
1715   if (token == G_TOKEN_IDENTIFIER)
1716     {
1717       if (config->scan_symbols)
1718         {
1719           GScannerKey *key;
1720           guint scope_id;
1721
1722           scope_id = scanner->scope_id;
1723           key = g_scanner_lookup_internal (scanner, scope_id, value.v_identifier);
1724           if (!key && scope_id && scanner->config->scope_0_fallback)
1725             key = g_scanner_lookup_internal (scanner, 0, value.v_identifier);
1726
1727           if (key)
1728             {
1729               g_free (value.v_identifier);
1730               token = G_TOKEN_SYMBOL;
1731               value.v_symbol = key->value;
1732             }
1733         }
1734
1735       if (token == G_TOKEN_IDENTIFIER &&
1736           config->scan_identifier_NULL &&
1737           strlen (value.v_identifier) == 4)
1738         {
1739           gchar *null_upper = "NULL";
1740           gchar *null_lower = "null";
1741
1742           if (scanner->config->case_sensitive)
1743             {
1744               if (value.v_identifier[0] == null_upper[0] &&
1745                   value.v_identifier[1] == null_upper[1] &&
1746                   value.v_identifier[2] == null_upper[2] &&
1747                   value.v_identifier[3] == null_upper[3])
1748                 token = G_TOKEN_IDENTIFIER_NULL;
1749             }
1750           else
1751             {
1752               if ((value.v_identifier[0] == null_upper[0] ||
1753                    value.v_identifier[0] == null_lower[0]) &&
1754                   (value.v_identifier[1] == null_upper[1] ||
1755                    value.v_identifier[1] == null_lower[1]) &&
1756                   (value.v_identifier[2] == null_upper[2] ||
1757                    value.v_identifier[2] == null_lower[2]) &&
1758                   (value.v_identifier[3] == null_upper[3] ||
1759                    value.v_identifier[3] == null_lower[3]))
1760                 token = G_TOKEN_IDENTIFIER_NULL;
1761             }
1762         }
1763     }
1764
1765   *token_p = token;
1766   *value_p = value;
1767 }