gio/gcontenttype.c

   1 /* -*- mode: C; c-file-style: "gnu"; indent-tabs-mode: nil; -*- */
   2
   3 /* GIO - GLib Input, Output and Streaming Library
   4  *
   5  * Copyright (C) 2006-2007 Red Hat, Inc.
   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
  18  * Public 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  * Author: Alexander Larsson <alexl@redhat.com>
  23  */
  24
  25 #include "config.h"
  26 #include <sys/types.h>
  27 #include <stdlib.h>
  28 #include <string.h>
  29 #include <stdio.h>
  30 #include "gcontenttypeprivate.h"
  31 #include "gthemedicon.h"
  32 #include "gicon.h"
  33 #include "gfile.h"
  34 #include "gfileenumerator.h"
  35 #include "gfileinfo.h"
  36 #include "glibintl.h"
  37
  38
  39 /**
  40  * SECTION:gcontenttype
  41  * @short_description: Platform-specific content typing
  42  * @include: gio/gio.h
  43  *
  44  * A content type is a platform specific string that defines the type
  45  * of a file. On UNIX it is a <ulink url="http://www.wikipedia.org/wiki/Internet_media_type">mime type</ulink> like "text/plain" or "image/png".
  46  * On Win32 it is an extension string like ".doc", ".txt" or a perceived
  47  * string like "audio". Such strings can be looked up in the registry at
  48  * HKEY_CLASSES_ROOT.
  49  **/
  50
  51 #include <dirent.h>
  52
  53 #define XDG_PREFIX _gio_xdg
  54 #include "xdgmime/xdgmime.h"
  55
  56 /* We lock this mutex whenever we modify global state in this module.  */
  57 G_LOCK_DEFINE_STATIC (gio_xdgmime);
  58
  59 gsize
  60 _g_unix_content_type_get_sniff_len (void)
  61 {
  62   gsize size;
  63
  64   G_LOCK (gio_xdgmime);
  65   size = xdg_mime_get_max_buffer_extents ();
  66   G_UNLOCK (gio_xdgmime);
  67
  68   return size;
  69 }
  70
  71 gchar *
  72 _g_unix_content_type_unalias (const gchar *type)
  73 {
  74   gchar *res;
  75
  76   G_LOCK (gio_xdgmime);
  77   res = g_strdup (xdg_mime_unalias_mime_type (type));
  78   G_UNLOCK (gio_xdgmime);
  79
  80   return res;
  81 }
  82
  83 gchar **
  84 _g_unix_content_type_get_parents (const gchar *type)
  85 {
  86   const gchar *umime;
  87   gchar **parents;
  88   GPtrArray *array;
  89   int i;
  90
  91   array = g_ptr_array_new ();
  92
  93   G_LOCK (gio_xdgmime);
  94
  95   umime = xdg_mime_unalias_mime_type (type);
  96
  97   g_ptr_array_add (array, g_strdup (umime));
  98
  99   parents = xdg_mime_list_mime_parents (umime);
 100   for (i = 0; parents && parents[i] != NULL; i++)
 101     g_ptr_array_add (array, g_strdup (parents[i]));
 102
 103   free (parents);
 104
 105   G_UNLOCK (gio_xdgmime);
 106
 107   g_ptr_array_add (array, NULL);
 108
 109   return (gchar **)g_ptr_array_free (array, FALSE);
 110 }
 111
 112 /**
 113  * g_content_type_equals:
 114  * @type1: a content type string
 115  * @type2: a content type string
 116  *
 117  * Compares two content types for equality.
 118  *
 119  * Returns: %TRUE if the two strings are identical or equivalent,
 120  *     %FALSE otherwise.
 121  */
 122 gboolean
 123 g_content_type_equals (const gchar *type1,
 124                        const gchar *type2)
 125 {
 126   gboolean res;
 127
 128   g_return_val_if_fail (type1 != NULL, FALSE);
 129   g_return_val_if_fail (type2 != NULL, FALSE);
 130
 131   G_LOCK (gio_xdgmime);
 132   res = xdg_mime_mime_type_equal (type1, type2);
 133   G_UNLOCK (gio_xdgmime);
 134
 135   return res;
 136 }
 137
 138 /**
 139  * g_content_type_is_a:
 140  * @type: a content type string
 141  * @supertype: a content type string
 142  *
 143  * Determines if @type is a subset of @supertype.
 144  *
 145  * Returns: %TRUE if @type is a kind of @supertype,
 146  *     %FALSE otherwise.
 147  */
 148 gboolean
 149 g_content_type_is_a (const gchar *type,
 150                      const gchar *supertype)
 151 {
 152   gboolean res;
 153
 154   g_return_val_if_fail (type != NULL, FALSE);
 155   g_return_val_if_fail (supertype != NULL, FALSE);
 156
 157   G_LOCK (gio_xdgmime);
 158   res = xdg_mime_mime_type_subclass (type, supertype);
 159   G_UNLOCK (gio_xdgmime);
 160
 161   return res;
 162 }
 163
 164 /**
 165  * g_content_type_is_unknown:
 166  * @type: a content type string
 167  *
 168  * Checks if the content type is the generic "unknown" type.
 169  * On UNIX this is the "application/octet-stream" mimetype,
 170  * while on win32 it is "*".
 171  *
 172  * Returns: %TRUE if the type is the unknown type.
 173  */
 174 gboolean
 175 g_content_type_is_unknown (const gchar *type)
 176 {
 177   g_return_val_if_fail (type != NULL, FALSE);
 178
 179   return strcmp (XDG_MIME_TYPE_UNKNOWN, type) == 0;
 180 }
 181
 182
 183 typedef enum {
 184   MIME_TAG_TYPE_OTHER,
 185   MIME_TAG_TYPE_COMMENT
 186 } MimeTagType;
 187
 188 typedef struct {
 189   int current_type;
 190   int current_lang_level;
 191   int comment_lang_level;
 192   char *comment;
 193 } MimeParser;
 194
 195
 196 static int
 197 language_level (const char *lang)
 198 {
 199   const char * const *lang_list;
 200   int i;
 201
 202   /* The returned list is sorted from most desirable to least
 203      desirable and always contains the default locale "C". */
 204   lang_list = g_get_language_names ();
 205
 206   for (i = 0; lang_list[i]; i++)
 207     if (strcmp (lang_list[i], lang) == 0)
 208       return 1000-i;
 209
 210   return 0;
 211 }
 212
 213 static void
 214 mime_info_start_element (GMarkupParseContext  *context,
 215                          const gchar          *element_name,
 216                          const gchar         **attribute_names,
 217                          const gchar         **attribute_values,
 218                          gpointer              user_data,
 219                          GError              **error)
 220 {
 221   int i;
 222   const char *lang;
 223   MimeParser *parser = user_data;
 224
 225   if (strcmp (element_name, "comment") == 0)
 226     {
 227       lang = "C";
 228       for (i = 0; attribute_names[i]; i++)
 229         if (strcmp (attribute_names[i], "xml:lang") == 0)
 230           {
 231             lang = attribute_values[i];
 232             break;
 233           }
 234
 235       parser->current_lang_level = language_level (lang);
 236       parser->current_type = MIME_TAG_TYPE_COMMENT;
 237     }
 238   else
 239     parser->current_type = MIME_TAG_TYPE_OTHER;
 240 }
 241
 242 static void
 243 mime_info_end_element (GMarkupParseContext  *context,
 244                        const gchar          *element_name,
 245                        gpointer              user_data,
 246                        GError              **error)
 247 {
 248   MimeParser *parser = user_data;
 249
 250   parser->current_type = MIME_TAG_TYPE_OTHER;
 251 }
 252
 253 static void
 254 mime_info_text (GMarkupParseContext  *context,
 255                 const gchar          *text,
 256                 gsize                 text_len,
 257                 gpointer              user_data,
 258                 GError              **error)
 259 {
 260   MimeParser *parser = user_data;
 261
 262   if (parser->current_type == MIME_TAG_TYPE_COMMENT &&
 263       parser->current_lang_level > parser->comment_lang_level)
 264     {
 265       g_free (parser->comment);
 266       parser->comment = g_strndup (text, text_len);
 267       parser->comment_lang_level = parser->current_lang_level;
 268     }
 269 }
 270
 271 static char *
 272 load_comment_for_mime_helper (const char *dir,
 273                               const char *basename)
 274 {
 275   GMarkupParseContext *context;
 276   char *filename, *data;
 277   gsize len;
 278   gboolean res;
 279   MimeParser parse_data = {0};
 280   GMarkupParser parser = {
 281     mime_info_start_element,
 282     mime_info_end_element,
 283     mime_info_text
 284   };
 285
 286   filename = g_build_filename (dir, "mime", basename, NULL);
 287
 288   res = g_file_get_contents (filename,  &data,  &len,  NULL);
 289   g_free (filename);
 290   if (!res)
 291     return NULL;
 292
 293   context = g_markup_parse_context_new   (&parser, 0, &parse_data, NULL);
 294   res = g_markup_parse_context_parse (context, data, len, NULL);
 295   g_free (data);
 296   g_markup_parse_context_free (context);
 297
 298   if (!res)
 299     return NULL;
 300
 301   return parse_data.comment;
 302 }
 303
 304
 305 static char *
 306 load_comment_for_mime (const char *mimetype)
 307 {
 308   const char * const* dirs;
 309   char *basename;
 310   char *comment;
 311   int i;
 312
 313   basename = g_strdup_printf ("%s.xml", mimetype);
 314
 315   comment = load_comment_for_mime_helper (g_get_user_data_dir (), basename);
 316   if (comment)
 317     {
 318       g_free (basename);
 319       return comment;
 320     }
 321
 322   dirs = g_get_system_data_dirs ();
 323
 324   for (i = 0; dirs[i] != NULL; i++)
 325     {
 326       comment = load_comment_for_mime_helper (dirs[i], basename);
 327       if (comment)
 328         {
 329           g_free (basename);
 330           return comment;
 331         }
 332     }
 333   g_free (basename);
 334
 335   return g_strdup_printf (_("%s type"), mimetype);
 336 }
 337
 338 /**
 339  * g_content_type_get_description:
 340  * @type: a content type string
 341  *
 342  * Gets the human readable description of the content type.
 343  *
 344  * Returns: a short description of the content type @type. Free the
 345  *     returned string with g_free()
 346  */
 347 gchar *
 348 g_content_type_get_description (const gchar *type)
 349 {
 350   static GHashTable *type_comment_cache = NULL;
 351   gchar *comment;
 352
 353   g_return_val_if_fail (type != NULL, NULL);
 354
 355   G_LOCK (gio_xdgmime);
 356   type = xdg_mime_unalias_mime_type (type);
 357
 358   if (type_comment_cache == NULL)
 359     type_comment_cache = g_hash_table_new_full (g_str_hash, g_str_equal, g_free, g_free);
 360
 361   comment = g_hash_table_lookup (type_comment_cache, type);
 362   comment = g_strdup (comment);
 363   G_UNLOCK (gio_xdgmime);
 364
 365   if (comment != NULL)
 366     return comment;
 367
 368   comment = load_comment_for_mime (type);
 369
 370   G_LOCK (gio_xdgmime);
 371   g_hash_table_insert (type_comment_cache,
 372                        g_strdup (type),
 373                        g_strdup (comment));
 374   G_UNLOCK (gio_xdgmime);
 375
 376   return comment;
 377 }
 378
 379 /**
 380  * g_content_type_get_mime_type:
 381  * @type: a content type string
 382  *
 383  * Gets the mime type for the content type, if one is registered.
 384  *
 385  * Returns: (allow-none): the registered mime type for the given @type,
 386  *     or %NULL if unknown.
 387  */
 388 char *
 389 g_content_type_get_mime_type (const char *type)
 390 {
 391   g_return_val_if_fail (type != NULL, NULL);
 392
 393   return g_strdup (type);
 394 }
 395
 396
 397 static GIcon *
 398 g_content_type_get_icon_internal (const gchar *type,
 399                                   gboolean     symbolic)
 400 {
 401   char *mimetype_icon;
 402   char *generic_mimetype_icon = NULL;
 403   char *q;
 404   char *xdg_mimetype_icon = NULL;
 405   char *legacy_mimetype_icon;
 406   char *xdg_mimetype_generic_icon;
 407   char *icon_names[5];
 408   int n = 0;
 409   GIcon *themed_icon;
 410   const char *file_template;
 411   const char  *xdg_icon;
 412
 413   g_return_val_if_fail (type != NULL, NULL);
 414
 415   if (symbolic)
 416     {
 417       file_template = "%s-symbolic";
 418     }
 419   else
 420     {
 421       file_template = "%s";
 422     }
 423
 424   G_LOCK (gio_xdgmime);
 425   xdg_icon = xdg_mime_get_icon (type);
 426   G_UNLOCK (gio_xdgmime);
 427   if (xdg_icon != NULL)
 428     xdg_mimetype_icon = g_strdup_printf (file_template, xdg_icon);
 429   xdg_mimetype_generic_icon = g_content_type_get_generic_icon_name (type);
 430
 431   mimetype_icon = g_strdup_printf (file_template, type);
 432   if (xdg_mimetype_generic_icon)
 433     generic_mimetype_icon = g_strdup_printf (file_template, xdg_mimetype_generic_icon);
 434
 435   while ((q = strchr (mimetype_icon, '/')) != NULL)
 436     *q = '-';
 437
 438   /* Not all icons have migrated to the new icon theme spec, look for old names too */
 439   legacy_mimetype_icon = g_strconcat ("gnome-mime-", mimetype_icon, NULL);
 440
 441   if (xdg_mimetype_icon)
 442     icon_names[n++] = xdg_mimetype_icon;
 443
 444   icon_names[n++] = mimetype_icon;
 445   icon_names[n++] = legacy_mimetype_icon;
 446
 447   if (generic_mimetype_icon)
 448     icon_names[n++] = generic_mimetype_icon;
 449
 450   themed_icon = g_themed_icon_new_from_names (icon_names, n);
 451
 452   g_free (xdg_mimetype_icon);
 453   g_free (xdg_mimetype_generic_icon);
 454   g_free (mimetype_icon);
 455   g_free (legacy_mimetype_icon);
 456   g_free (generic_mimetype_icon);
 457
 458   return themed_icon;
 459 }
 460
 461 /**
 462  * g_content_type_get_icon:
 463  * @type: a content type string
 464  *
 465  * Gets the icon for a content type.
 466  *
 467  * Returns: (transfer full): #GIcon corresponding to the content type. Free the returned
 468  *     object with g_object_unref()
 469  */
 470 GIcon *
 471 g_content_type_get_icon (const gchar *type)
 472 {
 473   return g_content_type_get_icon_internal (type, FALSE);
 474 }
 475
 476 /**
 477  * g_content_type_get_symbolic_icon:
 478  * @type: a content type string
 479  *
 480  * Gets the symbolic icon for a content type.
 481  *
 482  * Returns: (transfer full): symbolic #GIcon corresponding to the content type.
 483  *     Free the returned object with g_object_unref()
 484  *
 485  * Since: 2.34
 486  */
 487 GIcon *
 488 g_content_type_get_symbolic_icon (const gchar *type)
 489 {
 490   return g_content_type_get_icon_internal (type, TRUE);
 491 }
 492
 493 /**
 494  * g_content_type_get_generic_icon_name:
 495  * @type: a content type string
 496  *
 497  * Gets the generic icon name for a content type.
 498  *
 499  * See the <ulink url="http://www.freedesktop.org/wiki/Specifications/shared-mime-info-spec">shared-mime-info</ulink>
 500  * specification for more on the generic icon name.
 501  *
 502  * Returns: (allow-none): the registered generic icon name for the given @type,
 503  *     or %NULL if unknown. Free with g_free()
 504  *
 505  * Since: 2.34
 506  */
 507 gchar *
 508 g_content_type_get_generic_icon_name (const gchar *type)
 509 {
 510   const gchar *xdg_icon_name;
 511   gchar *icon_name;
 512
 513   G_LOCK (gio_xdgmime);
 514   xdg_icon_name = xdg_mime_get_generic_icon (type);
 515   G_UNLOCK (gio_xdgmime);
 516
 517   if (!xdg_icon_name)
 518     {
 519       const char *p;
 520       const char *suffix = "-x-generic";
 521
 522       p = strchr (type, '/');
 523       if (p == NULL)
 524         p = type + strlen (type);
 525
 526       icon_name = g_malloc (p - type + strlen (suffix) + 1);
 527       memcpy (icon_name, type, p - type);
 528       memcpy (icon_name + (p - type), suffix, strlen (suffix));
 529       icon_name[(p - type) + strlen (suffix)] = 0;
 530     }
 531   else
 532     {
 533       icon_name = g_strdup (xdg_icon_name);
 534     }
 535
 536   return icon_name;
 537 }
 538
 539 /**
 540  * g_content_type_can_be_executable:
 541  * @type: a content type string
 542  *
 543  * Checks if a content type can be executable. Note that for instance
 544  * things like text files can be executables (i.e. scripts and batch files).
 545  *
 546  * Returns: %TRUE if the file type corresponds to a type that
 547  *     can be executable, %FALSE otherwise.
 548  */
 549 gboolean
 550 g_content_type_can_be_executable (const gchar *type)
 551 {
 552   g_return_val_if_fail (type != NULL, FALSE);
 553
 554   if (g_content_type_is_a (type, "application/x-executable")  ||
 555       g_content_type_is_a (type, "text/plain"))
 556     return TRUE;
 557
 558   return FALSE;
 559 }
 560
 561 static gboolean
 562 looks_like_text (const guchar *data, gsize data_size)
 563 {
 564   gsize i;
 565   char c;
 566
 567   for (i = 0; i < data_size; i++)
 568     {
 569       c = data[i];
 570
 571       if (g_ascii_iscntrl (c) &&
 572           !g_ascii_isspace (c) &&
 573           c != '\b')
 574         return FALSE;
 575     }
 576   return TRUE;
 577 }
 578
 579 /**
 580  * g_content_type_from_mime_type:
 581  * @mime_type: a mime type string
 582  *
 583  * Tries to find a content type based on the mime type name.
 584  *
 585  * Returns: (allow-none): Newly allocated string with content type
 586  *     or %NULL. Free with g_free()
 587  *
 588  * Since: 2.18
 589  **/
 590 gchar *
 591 g_content_type_from_mime_type (const gchar *mime_type)
 592 {
 593   char *umime;
 594
 595   g_return_val_if_fail (mime_type != NULL, NULL);
 596
 597   G_LOCK (gio_xdgmime);
 598   /* mime type and content type are same on unixes */
 599   umime = g_strdup (xdg_mime_unalias_mime_type (mime_type));
 600   G_UNLOCK (gio_xdgmime);
 601
 602   return umime;
 603 }
 604
 605 /**
 606  * g_content_type_guess:
 607  * @filename: (allow-none): a string, or %NULL
 608  * @data: (allow-none) (array length=data_size): a stream of data, or %NULL
 609  * @data_size: the size of @data
 610  * @result_uncertain: (allow-none) (out): return location for the certainty
 611  *     of the result, or %NULL
 612  *
 613  * Guesses the content type based on example data. If the function is
 614  * uncertain, @result_uncertain will be set to %TRUE. Either @filename
 615  * or @data may be %NULL, in which case the guess will be based solely
 616  * on the other argument.
 617  *
 618  * Returns: a string indicating a guessed content type for the
 619  *     given data. Free with g_free()
 620  */
 621 gchar *
 622 g_content_type_guess (const gchar  *filename,
 623                       const guchar *data,
 624                       gsize         data_size,
 625                       gboolean     *result_uncertain)
 626 {
 627   char *basename;
 628   const char *name_mimetypes[10], *sniffed_mimetype;
 629   char *mimetype;
 630   int i;
 631   int n_name_mimetypes;
 632   int sniffed_prio;
 633
 634   sniffed_prio = 0;
 635   n_name_mimetypes = 0;
 636   sniffed_mimetype = XDG_MIME_TYPE_UNKNOWN;
 637
 638   if (result_uncertain)
 639     *result_uncertain = FALSE;
 640
 641   /* our test suite and potentially other code used -1 in the past, which is
 642    * not documented and not allowed; guard against that */
 643   g_return_val_if_fail (data_size != (gsize) -1, g_strdup (XDG_MIME_TYPE_UNKNOWN));
 644
 645   G_LOCK (gio_xdgmime);
 646
 647   if (filename)
 648     {
 649       i = strlen (filename);
 650       if (filename[i - 1] == '/')
 651         {
 652           name_mimetypes[0] = "inode/directory";
 653           name_mimetypes[1] = NULL;
 654           n_name_mimetypes = 1;
 655           if (result_uncertain)
 656             *result_uncertain = TRUE;
 657         }
 658       else
 659         {
 660           basename = g_path_get_basename (filename);
 661           n_name_mimetypes = xdg_mime_get_mime_types_from_file_name (basename, name_mimetypes, 10);
 662           g_free (basename);
 663         }
 664     }
 665
 666   /* Got an extension match, and no conflicts. This is it. */
 667   if (n_name_mimetypes == 1)
 668     {
 669       gchar *s = g_strdup (name_mimetypes[0]);
 670       G_UNLOCK (gio_xdgmime);
 671       return s;
 672     }
 673
 674   if (data)
 675     {
 676       sniffed_mimetype = xdg_mime_get_mime_type_for_data (data, data_size, &sniffed_prio);
 677       if (sniffed_mimetype == XDG_MIME_TYPE_UNKNOWN &&
 678           data &&
 679           looks_like_text (data, data_size))
 680         sniffed_mimetype = "text/plain";
 681
 682       /* For security reasons we don't ever want to sniff desktop files
 683        * where we know the filename and it doesn't have a .desktop extension.
 684        * This is because desktop files allow executing any application and
 685        * we don't want to make it possible to hide them looking like something
 686        * else.
 687        */
 688       if (filename != NULL &&
 689           strcmp (sniffed_mimetype, "application/x-desktop") == 0)
 690         sniffed_mimetype = "text/plain";
 691     }
 692
 693   if (n_name_mimetypes == 0)
 694     {
 695       if (sniffed_mimetype == XDG_MIME_TYPE_UNKNOWN &&
 696           result_uncertain)
 697         *result_uncertain = TRUE;
 698
 699       mimetype = g_strdup (sniffed_mimetype);
 700     }
 701   else
 702     {
 703       mimetype = NULL;
 704       if (sniffed_mimetype != XDG_MIME_TYPE_UNKNOWN)
 705         {
 706           if (sniffed_prio >= 80) /* High priority sniffing match, use that */
 707             mimetype = g_strdup (sniffed_mimetype);
 708           else
 709             {
 710               /* There are conflicts between the name matches and we
 711                * have a sniffed type, use that as a tie breaker.
 712                */
 713               for (i = 0; i < n_name_mimetypes; i++)
 714                 {
 715                   if ( xdg_mime_mime_type_subclass (name_mimetypes[i], sniffed_mimetype))
 716                     {
 717                       /* This nametype match is derived from (or the same as)
 718                        * the sniffed type). This is probably it.
 719                        */
 720                       mimetype = g_strdup (name_mimetypes[i]);
 721                       break;
 722                     }
 723                 }
 724             }
 725         }
 726
 727       if (mimetype == NULL)
 728         {
 729           /* Conflicts, and sniffed type was no help or not there.
 730            * Guess on the first one
 731            */
 732           mimetype = g_strdup (name_mimetypes[0]);
 733           if (result_uncertain)
 734             *result_uncertain = TRUE;
 735         }
 736     }
 737
 738   G_UNLOCK (gio_xdgmime);
 739
 740   return mimetype;
 741 }
 742
 743 static void
 744 enumerate_mimetypes_subdir (const char *dir,
 745                             const char *prefix,
 746                             GHashTable *mimetypes)
 747 {
 748   DIR *d;
 749   struct dirent *ent;
 750   char *mimetype;
 751
 752   d = opendir (dir);
 753   if (d)
 754     {
 755       while ((ent = readdir (d)) != NULL)
 756         {
 757           if (g_str_has_suffix (ent->d_name, ".xml"))
 758             {
 759               mimetype = g_strdup_printf ("%s/%.*s", prefix, (int) strlen (ent->d_name) - 4, ent->d_name);
 760               g_hash_table_replace (mimetypes, mimetype, NULL);
 761             }
 762         }
 763       closedir (d);
 764     }
 765 }
 766
 767 static void
 768 enumerate_mimetypes_dir (const char *dir,
 769                          GHashTable *mimetypes)
 770 {
 771   DIR *d;
 772   struct dirent *ent;
 773   char *mimedir;
 774   char *name;
 775
 776   mimedir = g_build_filename (dir, "mime", NULL);
 777
 778   d = opendir (mimedir);
 779   if (d)
 780     {
 781       while ((ent = readdir (d)) != NULL)
 782         {
 783           if (strcmp (ent->d_name, "packages") != 0)
 784             {
 785               name = g_build_filename (mimedir, ent->d_name, NULL);
 786               if (g_file_test (name, G_FILE_TEST_IS_DIR))
 787                 enumerate_mimetypes_subdir (name, ent->d_name, mimetypes);
 788               g_free (name);
 789             }
 790         }
 791       closedir (d);
 792     }
 793
 794   g_free (mimedir);
 795 }
 796
 797 /**
 798  * g_content_types_get_registered:
 799  *
 800  * Gets a list of strings containing all the registered content types
 801  * known to the system. The list and its data should be freed using
 802  * <programlisting>
 803  * g_list_free_full (list, g_free);
 804  * </programlisting>
 805  *
 806  * Returns: (element-type utf8) (transfer full): #GList of the registered content types
 807  */
 808 GList *
 809 g_content_types_get_registered (void)
 810 {
 811   const char * const* dirs;
 812   GHashTable *mimetypes;
 813   GHashTableIter iter;
 814   gpointer key;
 815   int i;
 816   GList *l;
 817
 818   mimetypes = g_hash_table_new_full (g_str_hash, g_str_equal, g_free, NULL);
 819
 820   enumerate_mimetypes_dir (g_get_user_data_dir (), mimetypes);
 821   dirs = g_get_system_data_dirs ();
 822
 823   for (i = 0; dirs[i] != NULL; i++)
 824     enumerate_mimetypes_dir (dirs[i], mimetypes);
 825
 826   l = NULL;
 827   g_hash_table_iter_init (&iter, mimetypes);
 828   while (g_hash_table_iter_next (&iter, &key, NULL))
 829     {
 830       l = g_list_prepend (l, key);
 831       g_hash_table_iter_steal (&iter);
 832     }
 833
 834   g_hash_table_destroy (mimetypes);
 835
 836   return l;
 837 }
 838
 839
 840 /* tree magic data */
 841 static GList *tree_matches = NULL;
 842 static gboolean need_reload = FALSE;
 843
 844 G_LOCK_DEFINE_STATIC (gio_treemagic);
 845
 846 typedef struct
 847 {
 848   gchar *path;
 849   GFileType type;
 850   guint match_case : 1;
 851   guint executable : 1;
 852   guint non_empty  : 1;
 853   guint on_disc    : 1;
 854   gchar *mimetype;
 855   GList *matches;
 856 } TreeMatchlet;
 857
 858 typedef struct
 859 {
 860   gchar *contenttype;
 861   gint priority;
 862   GList *matches;
 863 } TreeMatch;
 864
 865
 866 static void
 867 tree_matchlet_free (TreeMatchlet *matchlet)
 868 {
 869   g_list_free_full (matchlet->matches, (GDestroyNotify) tree_matchlet_free);
 870   g_free (matchlet->path);
 871   g_free (matchlet->mimetype);
 872   g_slice_free (TreeMatchlet, matchlet);
 873 }
 874
 875 static void
 876 tree_match_free (TreeMatch *match)
 877 {
 878   g_list_free_full (match->matches, (GDestroyNotify) tree_matchlet_free);
 879   g_free (match->contenttype);
 880   g_slice_free (TreeMatch, match);
 881 }
 882
 883 static TreeMatch *
 884 parse_header (gchar *line)
 885 {
 886   gint len;
 887   gchar *s;
 888   TreeMatch *match;
 889
 890   len = strlen (line);
 891
 892   if (line[0] != '[' || line[len - 1] != ']')
 893     return NULL;
 894
 895   line[len - 1] = 0;
 896   s = strchr (line, ':');
 897
 898   match = g_slice_new0 (TreeMatch);
 899   match->priority = atoi (line + 1);
 900   match->contenttype = g_strdup (s + 1);
 901
 902   return match;
 903 }
 904
 905 static TreeMatchlet *
 906 parse_match_line (gchar *line,
 907                   gint  *depth)
 908 {
 909   gchar *s, *p;
 910   TreeMatchlet *matchlet;
 911   gchar **parts;
 912   gint i;
 913
 914   matchlet = g_slice_new0 (TreeMatchlet);
 915
 916   if (line[0] == '>')
 917     {
 918       *depth = 0;
 919       s = line;
 920     }
 921   else
 922     {
 923       *depth = atoi (line);
 924       s = strchr (line, '>');
 925     }
 926   s += 2;
 927   p = strchr (s, '"');
 928   *p = 0;
 929
 930   matchlet->path = g_strdup (s);
 931   s = p + 1;
 932   parts = g_strsplit (s, ",", 0);
 933   if (strcmp (parts[0], "=file") == 0)
 934     matchlet->type = G_FILE_TYPE_REGULAR;
 935   else if (strcmp (parts[0], "=directory") == 0)
 936     matchlet->type = G_FILE_TYPE_DIRECTORY;
 937   else if (strcmp (parts[0], "=link") == 0)
 938     matchlet->type = G_FILE_TYPE_SYMBOLIC_LINK;
 939   else
 940     matchlet->type = G_FILE_TYPE_UNKNOWN;
 941   for (i = 1; parts[i]; i++)
 942     {
 943       if (strcmp (parts[i], "executable") == 0)
 944         matchlet->executable = 1;
 945       else if (strcmp (parts[i], "match-case") == 0)
 946         matchlet->match_case = 1;
 947       else if (strcmp (parts[i], "non-empty") == 0)
 948         matchlet->non_empty = 1;
 949       else if (strcmp (parts[i], "on-disc") == 0)
 950         matchlet->on_disc = 1;
 951       else
 952         matchlet->mimetype = g_strdup (parts[i]);
 953     }
 954
 955   g_strfreev (parts);
 956
 957   return matchlet;
 958 }
 959
 960 static gint
 961 cmp_match (gconstpointer a, gconstpointer b)
 962 {
 963   const TreeMatch *aa = (const TreeMatch *)a;
 964   const TreeMatch *bb = (const TreeMatch *)b;
 965
 966   return bb->priority - aa->priority;
 967 }
 968
 969 static void
 970 insert_match (TreeMatch *match)
 971 {
 972   tree_matches = g_list_insert_sorted (tree_matches, match, cmp_match);
 973 }
 974
 975 static void
 976 insert_matchlet (TreeMatch    *match,
 977                  TreeMatchlet *matchlet,
 978                  gint          depth)
 979 {
 980   if (depth == 0)
 981     match->matches = g_list_append (match->matches, matchlet);
 982   else
 983     {
 984       GList *last;
 985       TreeMatchlet *m;
 986
 987       last = g_list_last (match->matches);
 988       if (!last)
 989         {
 990           tree_matchlet_free (matchlet);
 991           g_warning ("can't insert tree matchlet at depth %d", depth);
 992           return;
 993         }
 994
 995       m = (TreeMatchlet *) last->data;
 996       while (--depth > 0)
 997         {
 998           last = g_list_last (m->matches);
 999           if (!last)
1000             {
1001               tree_matchlet_free (matchlet);
1002               g_warning ("can't insert tree matchlet at depth %d", depth);
1003               return;
1004             }
1005
1006           m = (TreeMatchlet *) last->data;
1007         }
1008       m->matches = g_list_append (m->matches, matchlet);
1009     }
1010 }
1011
1012 static void
1013 read_tree_magic_from_directory (const gchar *prefix)
1014 {
1015   gchar *filename;
1016   gchar *text;
1017   gsize len;
1018   gchar **lines;
1019   gint i;
1020   TreeMatch *match;
1021   TreeMatchlet *matchlet;
1022   gint depth;
1023
1024   filename = g_build_filename (prefix, "mime", "treemagic", NULL);
1025
1026   if (g_file_get_contents (filename, &text, &len, NULL))
1027     {
1028       if (strcmp (text, "MIME-TreeMagic") == 0)
1029         {
1030           lines = g_strsplit (text + strlen ("MIME-TreeMagic") + 2, "\n", 0);
1031           match = NULL;
1032           for (i = 0; lines[i] && lines[i][0]; i++)
1033             {
1034               if (lines[i][0] == '[')
1035                 {
1036                   match = parse_header (lines[i]);
1037                   insert_match (match);
1038                 }
1039               else
1040                 {
1041                   matchlet = parse_match_line (lines[i], &depth);
1042                   insert_matchlet (match, matchlet, depth);
1043                 }
1044             }
1045
1046           g_strfreev (lines);
1047         }
1048       else
1049         g_warning ("%s: header not found, skipping\n", filename);
1050
1051       g_free (text);
1052     }
1053
1054   g_free (filename);
1055 }
1056
1057
1058 static void
1059 xdg_mime_reload (void *user_data)
1060 {
1061   need_reload = TRUE;
1062 }
1063
1064 static void
1065 tree_magic_shutdown (void)
1066 {
1067   g_list_free_full (tree_matches, (GDestroyNotify) tree_match_free);
1068   tree_matches = NULL;
1069 }
1070
1071 static void
1072 tree_magic_init (void)
1073 {
1074   static gboolean initialized = FALSE;
1075   const gchar *dir;
1076   const gchar * const * dirs;
1077   int i;
1078
1079   if (!initialized)
1080     {
1081       initialized = TRUE;
1082
1083       xdg_mime_register_reload_callback (xdg_mime_reload, NULL, NULL);
1084       need_reload = TRUE;
1085     }
1086
1087   if (need_reload)
1088     {
1089       need_reload = FALSE;
1090
1091       tree_magic_shutdown ();
1092
1093       dir = g_get_user_data_dir ();
1094       read_tree_magic_from_directory (dir);
1095       dirs = g_get_system_data_dirs ();
1096       for (i = 0; dirs[i]; i++)
1097         read_tree_magic_from_directory (dirs[i]);
1098     }
1099 }
1100
1101 /* a filtering enumerator */
1102
1103 typedef struct
1104 {
1105   gchar *path;
1106   gint depth;
1107   gboolean ignore_case;
1108   gchar **components;
1109   gchar **case_components;
1110   GFileEnumerator **enumerators;
1111   GFile **children;
1112 } Enumerator;
1113
1114 static gboolean
1115 component_match (Enumerator  *e,
1116                  gint         depth,
1117                  const gchar *name)
1118 {
1119   gchar *case_folded, *key;
1120   gboolean found;
1121
1122   if (strcmp (name, e->components[depth]) == 0)
1123     return TRUE;
1124
1125   if (!e->ignore_case)
1126     return FALSE;
1127
1128   case_folded = g_utf8_casefold (name, -1);
1129   key = g_utf8_collate_key (case_folded, -1);
1130
1131   found = strcmp (key, e->case_components[depth]) == 0;
1132
1133   g_free (case_folded);
1134   g_free (key);
1135
1136   return found;
1137 }
1138
1139 static GFile *
1140 next_match_recurse (Enumerator *e,
1141                     gint        depth)
1142 {
1143   GFile *file;
1144   GFileInfo *info;
1145   const gchar *name;
1146
1147   while (TRUE)
1148     {
1149       if (e->enumerators[depth] == NULL)
1150         {
1151           if (depth > 0)
1152             {
1153               file = next_match_recurse (e, depth - 1);
1154               if (file)
1155                 {
1156                   e->children[depth] = file;
1157                   e->enumerators[depth] = g_file_enumerate_children (file,
1158                                                                      G_FILE_ATTRIBUTE_STANDARD_NAME,
1159                                                                      G_FILE_QUERY_INFO_NONE,
1160                                                                      NULL,
1161                                                                      NULL);
1162                 }
1163             }
1164           if (e->enumerators[depth] == NULL)
1165             return NULL;
1166         }
1167
1168       while ((info = g_file_enumerator_next_file (e->enumerators[depth], NULL, NULL)))
1169         {
1170           name = g_file_info_get_name (info);
1171           if (component_match (e, depth, name))
1172             {
1173               file = g_file_get_child (e->children[depth], name);
1174               g_object_unref (info);
1175               return file;
1176             }
1177           g_object_unref (info);
1178         }
1179
1180       g_object_unref (e->enumerators[depth]);
1181       e->enumerators[depth] = NULL;
1182       g_object_unref (e->children[depth]);
1183       e->children[depth] = NULL;
1184     }
1185 }
1186
1187 static GFile *
1188 enumerator_next (Enumerator *e)
1189 {
1190   return next_match_recurse (e, e->depth - 1);
1191 }
1192
1193 static Enumerator *
1194 enumerator_new (GFile      *root,
1195                 const char *path,
1196                 gboolean    ignore_case)
1197 {
1198   Enumerator *e;
1199   gint i;
1200   gchar *case_folded;
1201
1202   e = g_new0 (Enumerator, 1);
1203   e->path = g_strdup (path);
1204   e->ignore_case = ignore_case;
1205
1206   e->components = g_strsplit (e->path, G_DIR_SEPARATOR_S, -1);
1207   e->depth = g_strv_length (e->components);
1208   if (e->ignore_case)
1209     {
1210       e->case_components = g_new0 (char *, e->depth + 1);
1211       for (i = 0; e->components[i]; i++)
1212         {
1213           case_folded = g_utf8_casefold (e->components[i], -1);
1214           e->case_components[i] = g_utf8_collate_key (case_folded, -1);
1215           g_free (case_folded);
1216         }
1217     }
1218
1219   e->children = g_new0 (GFile *, e->depth);
1220   e->children[0] = g_object_ref (root);
1221   e->enumerators = g_new0 (GFileEnumerator *, e->depth);
1222   e->enumerators[0] = g_file_enumerate_children (root,
1223                                                  G_FILE_ATTRIBUTE_STANDARD_NAME,
1224                                                  G_FILE_QUERY_INFO_NONE,
1225                                                  NULL,
1226                                                  NULL);
1227
1228   return e;
1229 }
1230
1231 static void
1232 enumerator_free (Enumerator *e)
1233 {
1234   gint i;
1235
1236   for (i = 0; i < e->depth; i++)
1237     {
1238       if (e->enumerators[i])
1239         g_object_unref (e->enumerators[i]);
1240       if (e->children[i])
1241         g_object_unref (e->children[i]);
1242     }
1243
1244   g_free (e->enumerators);
1245   g_free (e->children);
1246   g_strfreev (e->components);
1247   if (e->case_components)
1248     g_strfreev (e->case_components);
1249   g_free (e->path);
1250   g_free (e);
1251 }
1252
1253 static gboolean
1254 matchlet_match (TreeMatchlet *matchlet,
1255                 GFile        *root)
1256 {
1257   GFile *file;
1258   GFileInfo *info;
1259   gboolean result;
1260   const gchar *attrs;
1261   Enumerator *e;
1262   GList *l;
1263
1264   e = enumerator_new (root, matchlet->path, !matchlet->match_case);
1265
1266   do
1267     {
1268       file = enumerator_next (e);
1269       if (!file)
1270         {
1271           enumerator_free (e);
1272           return FALSE;
1273         }
1274
1275       if (matchlet->mimetype)
1276         attrs = G_FILE_ATTRIBUTE_STANDARD_TYPE ","
1277                 G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE ","
1278                 G_FILE_ATTRIBUTE_STANDARD_CONTENT_TYPE;
1279       else
1280         attrs = G_FILE_ATTRIBUTE_STANDARD_TYPE ","
1281                 G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE;
1282       info = g_file_query_info (file,
1283                                 attrs,
1284                                 G_FILE_QUERY_INFO_NONE,
1285                                 NULL,
1286                                 NULL);
1287       if (info)
1288         {
1289           result = TRUE;
1290
1291           if (matchlet->type != G_FILE_TYPE_UNKNOWN &&
1292               g_file_info_get_file_type (info) != matchlet->type)
1293             result = FALSE;
1294
1295           if (matchlet->executable &&
1296               !g_file_info_get_attribute_boolean (info, G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE))
1297             result = FALSE;
1298         }
1299       else
1300         result = FALSE;
1301
1302       if (result && matchlet->non_empty)
1303         {
1304           GFileEnumerator *child_enum;
1305           GFileInfo *child_info;
1306
1307           child_enum = g_file_enumerate_children (file,
1308                                                   G_FILE_ATTRIBUTE_STANDARD_NAME,
1309                                                   G_FILE_QUERY_INFO_NONE,
1310                                                   NULL,
1311                                                   NULL);
1312
1313           if (child_enum)
1314             {
1315               child_info = g_file_enumerator_next_file (child_enum, NULL, NULL);
1316               if (child_info)
1317                 g_object_unref (child_info);
1318               else
1319                 result = FALSE;
1320               g_object_unref (child_enum);
1321             }
1322           else
1323             result = FALSE;
1324         }
1325
1326       if (result && matchlet->mimetype)
1327         {
1328           if (strcmp (matchlet->mimetype, g_file_info_get_content_type (info)) != 0)
1329             result = FALSE;
1330         }
1331
1332       g_object_unref (info);
1333       g_object_unref (file);
1334     }
1335   while (!result);
1336
1337   enumerator_free (e);
1338
1339   if (!matchlet->matches)
1340     return TRUE;
1341
1342   for (l = matchlet->matches; l; l = l->next)
1343     {
1344       TreeMatchlet *submatchlet;
1345
1346       submatchlet = l->data;
1347       if (matchlet_match (submatchlet, root))
1348         return TRUE;
1349     }
1350
1351   return FALSE;
1352 }
1353
1354 static void
1355 match_match (TreeMatch    *match,
1356              GFile        *root,
1357              GPtrArray    *types)
1358 {
1359   GList *l;
1360
1361   for (l = match->matches; l; l = l->next)
1362     {
1363       TreeMatchlet *matchlet = l->data;
1364       if (matchlet_match (matchlet, root))
1365         {
1366           g_ptr_array_add (types, g_strdup (match->contenttype));
1367           break;
1368         }
1369     }
1370 }
1371
1372 /**
1373  * g_content_type_guess_for_tree:
1374  * @root: the root of the tree to guess a type for
1375  *
1376  * Tries to guess the type of the tree with root @root, by
1377  * looking at the files it contains. The result is an array
1378  * of content types, with the best guess coming first.
1379  *
1380  * The types returned all have the form x-content/foo, e.g.
1381  * x-content/audio-cdda (for audio CDs) or x-content/image-dcf
1382  * (for a camera memory card). See the <ulink url="http://www.freedesktop.org/wiki/Specifications/shared-mime-info-spec">shared-mime-info</ulink>
1383  * specification for more on x-content types.
1384  *
1385  * This function is useful in the implementation of
1386  * g_mount_guess_content_type().
1387  *
1388  * Returns: (transfer full) (array zero-terminated=1): an %NULL-terminated
1389  *     array of zero or more content types. Free with g_strfreev()
1390  *
1391  * Since: 2.18
1392  */
1393 gchar **
1394 g_content_type_guess_for_tree (GFile *root)
1395 {
1396   GPtrArray *types;
1397   GList *l;
1398
1399   types = g_ptr_array_new ();
1400
1401   G_LOCK (gio_treemagic);
1402
1403   tree_magic_init ();
1404   for (l = tree_matches; l; l = l->next)
1405     {
1406       TreeMatch *match = l->data;
1407       match_match (match, root, types);
1408     }
1409
1410   G_UNLOCK (gio_treemagic);
1411
1412   g_ptr_array_add (types, NULL);
1413
1414   return (gchar **)g_ptr_array_free (types, FALSE);
1415 }