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, see <http://www.gnu.org/licenses/>.
  19  *
  20  * Author: Alexander Larsson <alexl@redhat.com>
  21  */
  22
  23 #include "config.h"
  24 #include <sys/types.h>
  25 #include <stdlib.h>
  26 #include <string.h>
  27 #include <stdio.h>
  28 #include "gcontenttypeprivate.h"
  29 #include "gthemedicon.h"
  30 #include "gicon.h"
  31 #include "gfile.h"
  32 #include "gfileenumerator.h"
  33 #include "gfileinfo.h"
  34 #include "glibintl.h"
  35
  36
  37 /**
  38  * SECTION:gcontenttype
  39  * @short_description: Platform-specific content typing
  40  * @include: gio/gio.h
  41  *
  42  * A content type is a platform specific string that defines the type
  43  * of a file. On UNIX it is a
  44  * [mime type](http://www.wikipedia.org/wiki/Internet_media_type)
  45  * 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: (nullable): 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 static GIcon *
 397 g_content_type_get_icon_internal (const gchar *type,
 398                                   gboolean     symbolic)
 399 {
 400   char *mimetype_icon;
 401   char *generic_mimetype_icon = NULL;
 402   char *q;
 403   char *icon_names[6];
 404   int n = 0;
 405   GIcon *themed_icon;
 406   const char  *xdg_icon;
 407   int i;
 408
 409   g_return_val_if_fail (type != NULL, NULL);
 410
 411   G_LOCK (gio_xdgmime);
 412   xdg_icon = xdg_mime_get_icon (type);
 413   G_UNLOCK (gio_xdgmime);
 414
 415   if (xdg_icon)
 416     icon_names[n++] = g_strdup (xdg_icon);
 417
 418   mimetype_icon = g_strdup (type);
 419   while ((q = strchr (mimetype_icon, '/')) != NULL)
 420     *q = '-';
 421
 422   icon_names[n++] = mimetype_icon;
 423
 424   generic_mimetype_icon = g_content_type_get_generic_icon_name (type);
 425   if (generic_mimetype_icon)
 426     icon_names[n++] = generic_mimetype_icon;
 427
 428   if (symbolic)
 429     {
 430       for (i = 0; i < n; i++)
 431         {
 432           icon_names[n + i] = icon_names[i];
 433           icon_names[i] = g_strconcat (icon_names[i], "-symbolic", NULL);
 434         }
 435
 436       n += n;
 437     }
 438
 439   themed_icon = g_themed_icon_new_from_names (icon_names, n);
 440
 441   for (i = 0; i < n; i++)
 442     g_free (icon_names[i]);
 443
 444   return themed_icon;
 445 }
 446
 447 /**
 448  * g_content_type_get_icon:
 449  * @type: a content type string
 450  *
 451  * Gets the icon for a content type.
 452  *
 453  * Returns: (transfer full): #GIcon corresponding to the content type. Free the returned
 454  *     object with g_object_unref()
 455  */
 456 GIcon *
 457 g_content_type_get_icon (const gchar *type)
 458 {
 459   return g_content_type_get_icon_internal (type, FALSE);
 460 }
 461
 462 /**
 463  * g_content_type_get_symbolic_icon:
 464  * @type: a content type string
 465  *
 466  * Gets the symbolic icon for a content type.
 467  *
 468  * Returns: (transfer full): symbolic #GIcon corresponding to the content type.
 469  *     Free the returned object with g_object_unref()
 470  *
 471  * Since: 2.34
 472  */
 473 GIcon *
 474 g_content_type_get_symbolic_icon (const gchar *type)
 475 {
 476   return g_content_type_get_icon_internal (type, TRUE);
 477 }
 478
 479 /**
 480  * g_content_type_get_generic_icon_name:
 481  * @type: a content type string
 482  *
 483  * Gets the generic icon name for a content type.
 484  *
 485  * See the
 486  * [shared-mime-info](http://www.freedesktop.org/wiki/Specifications/shared-mime-info-spec)
 487  * specification for more on the generic icon name.
 488  *
 489  * Returns: (allow-none): the registered generic icon name for the given @type,
 490  *     or %NULL if unknown. Free with g_free()
 491  *
 492  * Since: 2.34
 493  */
 494 gchar *
 495 g_content_type_get_generic_icon_name (const gchar *type)
 496 {
 497   const gchar *xdg_icon_name;
 498   gchar *icon_name;
 499
 500   G_LOCK (gio_xdgmime);
 501   xdg_icon_name = xdg_mime_get_generic_icon (type);
 502   G_UNLOCK (gio_xdgmime);
 503
 504   if (!xdg_icon_name)
 505     {
 506       const char *p;
 507       const char *suffix = "-x-generic";
 508
 509       p = strchr (type, '/');
 510       if (p == NULL)
 511         p = type + strlen (type);
 512
 513       icon_name = g_malloc (p - type + strlen (suffix) + 1);
 514       memcpy (icon_name, type, p - type);
 515       memcpy (icon_name + (p - type), suffix, strlen (suffix));
 516       icon_name[(p - type) + strlen (suffix)] = 0;
 517     }
 518   else
 519     {
 520       icon_name = g_strdup (xdg_icon_name);
 521     }
 522
 523   return icon_name;
 524 }
 525
 526 /**
 527  * g_content_type_can_be_executable:
 528  * @type: a content type string
 529  *
 530  * Checks if a content type can be executable. Note that for instance
 531  * things like text files can be executables (i.e. scripts and batch files).
 532  *
 533  * Returns: %TRUE if the file type corresponds to a type that
 534  *     can be executable, %FALSE otherwise.
 535  */
 536 gboolean
 537 g_content_type_can_be_executable (const gchar *type)
 538 {
 539   g_return_val_if_fail (type != NULL, FALSE);
 540
 541   if (g_content_type_is_a (type, "application/x-executable")  ||
 542       g_content_type_is_a (type, "text/plain"))
 543     return TRUE;
 544
 545   return FALSE;
 546 }
 547
 548 static gboolean
 549 looks_like_text (const guchar *data, gsize data_size)
 550 {
 551   gsize i;
 552   char c;
 553
 554   for (i = 0; i < data_size; i++)
 555     {
 556       c = data[i];
 557
 558       if (g_ascii_iscntrl (c) &&
 559           !g_ascii_isspace (c) &&
 560           c != '\b')
 561         return FALSE;
 562     }
 563   return TRUE;
 564 }
 565
 566 /**
 567  * g_content_type_from_mime_type:
 568  * @mime_type: a mime type string
 569  *
 570  * Tries to find a content type based on the mime type name.
 571  *
 572  * Returns: (nullable): Newly allocated string with content type or
 573  *     %NULL. Free with g_free()
 574  *
 575  * Since: 2.18
 576  **/
 577 gchar *
 578 g_content_type_from_mime_type (const gchar *mime_type)
 579 {
 580   char *umime;
 581
 582   g_return_val_if_fail (mime_type != NULL, NULL);
 583
 584   G_LOCK (gio_xdgmime);
 585   /* mime type and content type are same on unixes */
 586   umime = g_strdup (xdg_mime_unalias_mime_type (mime_type));
 587   G_UNLOCK (gio_xdgmime);
 588
 589   return umime;
 590 }
 591
 592 /**
 593  * g_content_type_guess:
 594  * @filename: (allow-none): a string, or %NULL
 595  * @data: (allow-none) (array length=data_size): a stream of data, or %NULL
 596  * @data_size: the size of @data
 597  * @result_uncertain: (allow-none) (out): return location for the certainty
 598  *     of the result, or %NULL
 599  *
 600  * Guesses the content type based on example data. If the function is
 601  * uncertain, @result_uncertain will be set to %TRUE. Either @filename
 602  * or @data may be %NULL, in which case the guess will be based solely
 603  * on the other argument.
 604  *
 605  * Returns: a string indicating a guessed content type for the
 606  *     given data. Free with g_free()
 607  */
 608 gchar *
 609 g_content_type_guess (const gchar  *filename,
 610                       const guchar *data,
 611                       gsize         data_size,
 612                       gboolean     *result_uncertain)
 613 {
 614   char *basename;
 615   const char *name_mimetypes[10], *sniffed_mimetype;
 616   char *mimetype;
 617   int i;
 618   int n_name_mimetypes;
 619   int sniffed_prio;
 620
 621   sniffed_prio = 0;
 622   n_name_mimetypes = 0;
 623   sniffed_mimetype = XDG_MIME_TYPE_UNKNOWN;
 624
 625   if (result_uncertain)
 626     *result_uncertain = FALSE;
 627
 628   /* our test suite and potentially other code used -1 in the past, which is
 629    * not documented and not allowed; guard against that */
 630   g_return_val_if_fail (data_size != (gsize) -1, g_strdup (XDG_MIME_TYPE_UNKNOWN));
 631
 632   G_LOCK (gio_xdgmime);
 633
 634   if (filename)
 635     {
 636       i = strlen (filename);
 637       if (filename[i - 1] == '/')
 638         {
 639           name_mimetypes[0] = "inode/directory";
 640           name_mimetypes[1] = NULL;
 641           n_name_mimetypes = 1;
 642           if (result_uncertain)
 643             *result_uncertain = TRUE;
 644         }
 645       else
 646         {
 647           basename = g_path_get_basename (filename);
 648           n_name_mimetypes = xdg_mime_get_mime_types_from_file_name (basename, name_mimetypes, 10);
 649           g_free (basename);
 650         }
 651     }
 652
 653   /* Got an extension match, and no conflicts. This is it. */
 654   if (n_name_mimetypes == 1)
 655     {
 656       gchar *s = g_strdup (name_mimetypes[0]);
 657       G_UNLOCK (gio_xdgmime);
 658       return s;
 659     }
 660
 661   if (data)
 662     {
 663       sniffed_mimetype = xdg_mime_get_mime_type_for_data (data, data_size, &sniffed_prio);
 664       if (sniffed_mimetype == XDG_MIME_TYPE_UNKNOWN &&
 665           data &&
 666           looks_like_text (data, data_size))
 667         sniffed_mimetype = "text/plain";
 668
 669       /* For security reasons we don't ever want to sniff desktop files
 670        * where we know the filename and it doesn't have a .desktop extension.
 671        * This is because desktop files allow executing any application and
 672        * we don't want to make it possible to hide them looking like something
 673        * else.
 674        */
 675       if (filename != NULL &&
 676           strcmp (sniffed_mimetype, "application/x-desktop") == 0)
 677         sniffed_mimetype = "text/plain";
 678     }
 679
 680   if (n_name_mimetypes == 0)
 681     {
 682       if (sniffed_mimetype == XDG_MIME_TYPE_UNKNOWN &&
 683           result_uncertain)
 684         *result_uncertain = TRUE;
 685
 686       mimetype = g_strdup (sniffed_mimetype);
 687     }
 688   else
 689     {
 690       mimetype = NULL;
 691       if (sniffed_mimetype != XDG_MIME_TYPE_UNKNOWN)
 692         {
 693           if (sniffed_prio >= 80) /* High priority sniffing match, use that */
 694             mimetype = g_strdup (sniffed_mimetype);
 695           else
 696             {
 697               /* There are conflicts between the name matches and we
 698                * have a sniffed type, use that as a tie breaker.
 699                */
 700               for (i = 0; i < n_name_mimetypes; i++)
 701                 {
 702                   if ( xdg_mime_mime_type_subclass (name_mimetypes[i], sniffed_mimetype))
 703                     {
 704                       /* This nametype match is derived from (or the same as)
 705                        * the sniffed type). This is probably it.
 706                        */
 707                       mimetype = g_strdup (name_mimetypes[i]);
 708                       break;
 709                     }
 710                 }
 711             }
 712         }
 713
 714       if (mimetype == NULL)
 715         {
 716           /* Conflicts, and sniffed type was no help or not there.
 717            * Guess on the first one
 718            */
 719           mimetype = g_strdup (name_mimetypes[0]);
 720           if (result_uncertain)
 721             *result_uncertain = TRUE;
 722         }
 723     }
 724
 725   G_UNLOCK (gio_xdgmime);
 726
 727   return mimetype;
 728 }
 729
 730 static void
 731 enumerate_mimetypes_subdir (const char *dir,
 732                             const char *prefix,
 733                             GHashTable *mimetypes)
 734 {
 735   DIR *d;
 736   struct dirent *ent;
 737   char *mimetype;
 738
 739   d = opendir (dir);
 740   if (d)
 741     {
 742       while ((ent = readdir (d)) != NULL)
 743         {
 744           if (g_str_has_suffix (ent->d_name, ".xml"))
 745             {
 746               mimetype = g_strdup_printf ("%s/%.*s", prefix, (int) strlen (ent->d_name) - 4, ent->d_name);
 747               g_hash_table_replace (mimetypes, mimetype, NULL);
 748             }
 749         }
 750       closedir (d);
 751     }
 752 }
 753
 754 static void
 755 enumerate_mimetypes_dir (const char *dir,
 756                          GHashTable *mimetypes)
 757 {
 758   DIR *d;
 759   struct dirent *ent;
 760   char *mimedir;
 761   char *name;
 762
 763   mimedir = g_build_filename (dir, "mime", NULL);
 764
 765   d = opendir (mimedir);
 766   if (d)
 767     {
 768       while ((ent = readdir (d)) != NULL)
 769         {
 770           if (strcmp (ent->d_name, "packages") != 0)
 771             {
 772               name = g_build_filename (mimedir, ent->d_name, NULL);
 773               if (g_file_test (name, G_FILE_TEST_IS_DIR))
 774                 enumerate_mimetypes_subdir (name, ent->d_name, mimetypes);
 775               g_free (name);
 776             }
 777         }
 778       closedir (d);
 779     }
 780
 781   g_free (mimedir);
 782 }
 783
 784 /**
 785  * g_content_types_get_registered:
 786  *
 787  * Gets a list of strings containing all the registered content types
 788  * known to the system. The list and its data should be freed using
 789  * g_list_free_full (list, g_free).
 790  *
 791  * Returns: (element-type utf8) (transfer full): list of the registered
 792  *     content types
 793  */
 794 GList *
 795 g_content_types_get_registered (void)
 796 {
 797   const char * const* dirs;
 798   GHashTable *mimetypes;
 799   GHashTableIter iter;
 800   gpointer key;
 801   int i;
 802   GList *l;
 803
 804   mimetypes = g_hash_table_new_full (g_str_hash, g_str_equal, g_free, NULL);
 805
 806   enumerate_mimetypes_dir (g_get_user_data_dir (), mimetypes);
 807   dirs = g_get_system_data_dirs ();
 808
 809   for (i = 0; dirs[i] != NULL; i++)
 810     enumerate_mimetypes_dir (dirs[i], mimetypes);
 811
 812   l = NULL;
 813   g_hash_table_iter_init (&iter, mimetypes);
 814   while (g_hash_table_iter_next (&iter, &key, NULL))
 815     {
 816       l = g_list_prepend (l, key);
 817       g_hash_table_iter_steal (&iter);
 818     }
 819
 820   g_hash_table_destroy (mimetypes);
 821
 822   return l;
 823 }
 824
 825
 826 /* tree magic data */
 827 static GList *tree_matches = NULL;
 828 static gboolean need_reload = FALSE;
 829
 830 G_LOCK_DEFINE_STATIC (gio_treemagic);
 831
 832 typedef struct
 833 {
 834   gchar *path;
 835   GFileType type;
 836   guint match_case : 1;
 837   guint executable : 1;
 838   guint non_empty  : 1;
 839   guint on_disc    : 1;
 840   gchar *mimetype;
 841   GList *matches;
 842 } TreeMatchlet;
 843
 844 typedef struct
 845 {
 846   gchar *contenttype;
 847   gint priority;
 848   GList *matches;
 849 } TreeMatch;
 850
 851
 852 static void
 853 tree_matchlet_free (TreeMatchlet *matchlet)
 854 {
 855   g_list_free_full (matchlet->matches, (GDestroyNotify) tree_matchlet_free);
 856   g_free (matchlet->path);
 857   g_free (matchlet->mimetype);
 858   g_slice_free (TreeMatchlet, matchlet);
 859 }
 860
 861 static void
 862 tree_match_free (TreeMatch *match)
 863 {
 864   g_list_free_full (match->matches, (GDestroyNotify) tree_matchlet_free);
 865   g_free (match->contenttype);
 866   g_slice_free (TreeMatch, match);
 867 }
 868
 869 static TreeMatch *
 870 parse_header (gchar *line)
 871 {
 872   gint len;
 873   gchar *s;
 874   TreeMatch *match;
 875
 876   len = strlen (line);
 877
 878   if (line[0] != '[' || line[len - 1] != ']')
 879     return NULL;
 880
 881   line[len - 1] = 0;
 882   s = strchr (line, ':');
 883
 884   match = g_slice_new0 (TreeMatch);
 885   match->priority = atoi (line + 1);
 886   match->contenttype = g_strdup (s + 1);
 887
 888   return match;
 889 }
 890
 891 static TreeMatchlet *
 892 parse_match_line (gchar *line,
 893                   gint  *depth)
 894 {
 895   gchar *s, *p;
 896   TreeMatchlet *matchlet;
 897   gchar **parts;
 898   gint i;
 899
 900   matchlet = g_slice_new0 (TreeMatchlet);
 901
 902   if (line[0] == '>')
 903     {
 904       *depth = 0;
 905       s = line;
 906     }
 907   else
 908     {
 909       *depth = atoi (line);
 910       s = strchr (line, '>');
 911     }
 912   s += 2;
 913   p = strchr (s, '"');
 914   *p = 0;
 915
 916   matchlet->path = g_strdup (s);
 917   s = p + 1;
 918   parts = g_strsplit (s, ",", 0);
 919   if (strcmp (parts[0], "=file") == 0)
 920     matchlet->type = G_FILE_TYPE_REGULAR;
 921   else if (strcmp (parts[0], "=directory") == 0)
 922     matchlet->type = G_FILE_TYPE_DIRECTORY;
 923   else if (strcmp (parts[0], "=link") == 0)
 924     matchlet->type = G_FILE_TYPE_SYMBOLIC_LINK;
 925   else
 926     matchlet->type = G_FILE_TYPE_UNKNOWN;
 927   for (i = 1; parts[i]; i++)
 928     {
 929       if (strcmp (parts[i], "executable") == 0)
 930         matchlet->executable = 1;
 931       else if (strcmp (parts[i], "match-case") == 0)
 932         matchlet->match_case = 1;
 933       else if (strcmp (parts[i], "non-empty") == 0)
 934         matchlet->non_empty = 1;
 935       else if (strcmp (parts[i], "on-disc") == 0)
 936         matchlet->on_disc = 1;
 937       else
 938         matchlet->mimetype = g_strdup (parts[i]);
 939     }
 940
 941   g_strfreev (parts);
 942
 943   return matchlet;
 944 }
 945
 946 static gint
 947 cmp_match (gconstpointer a, gconstpointer b)
 948 {
 949   const TreeMatch *aa = (const TreeMatch *)a;
 950   const TreeMatch *bb = (const TreeMatch *)b;
 951
 952   return bb->priority - aa->priority;
 953 }
 954
 955 static void
 956 insert_match (TreeMatch *match)
 957 {
 958   tree_matches = g_list_insert_sorted (tree_matches, match, cmp_match);
 959 }
 960
 961 static void
 962 insert_matchlet (TreeMatch    *match,
 963                  TreeMatchlet *matchlet,
 964                  gint          depth)
 965 {
 966   if (depth == 0)
 967     match->matches = g_list_append (match->matches, matchlet);
 968   else
 969     {
 970       GList *last;
 971       TreeMatchlet *m;
 972
 973       last = g_list_last (match->matches);
 974       if (!last)
 975         {
 976           tree_matchlet_free (matchlet);
 977           g_warning ("can't insert tree matchlet at depth %d", depth);
 978           return;
 979         }
 980
 981       m = (TreeMatchlet *) last->data;
 982       while (--depth > 0)
 983         {
 984           last = g_list_last (m->matches);
 985           if (!last)
 986             {
 987               tree_matchlet_free (matchlet);
 988               g_warning ("can't insert tree matchlet at depth %d", depth);
 989               return;
 990             }
 991
 992           m = (TreeMatchlet *) last->data;
 993         }
 994       m->matches = g_list_append (m->matches, matchlet);
 995     }
 996 }
 997
 998 static void
 999 read_tree_magic_from_directory (const gchar *prefix)
1000 {
1001   gchar *filename;
1002   gchar *text;
1003   gsize len;
1004   gchar **lines;
1005   gint i;
1006   TreeMatch *match;
1007   TreeMatchlet *matchlet;
1008   gint depth;
1009
1010   filename = g_build_filename (prefix, "mime", "treemagic", NULL);
1011
1012   if (g_file_get_contents (filename, &text, &len, NULL))
1013     {
1014       if (strcmp (text, "MIME-TreeMagic") == 0)
1015         {
1016           lines = g_strsplit (text + strlen ("MIME-TreeMagic") + 2, "\n", 0);
1017           match = NULL;
1018           for (i = 0; lines[i] && lines[i][0]; i++)
1019             {
1020               if (lines[i][0] == '[')
1021                 {
1022                   match = parse_header (lines[i]);
1023                   insert_match (match);
1024                 }
1025               else if (match != NULL)
1026                 {
1027                   matchlet = parse_match_line (lines[i], &depth);
1028                   insert_matchlet (match, matchlet, depth);
1029                 }
1030               else
1031                 {
1032                   g_warning ("%s: header corrupt; skipping\n", filename);
1033                   break;
1034                 }
1035             }
1036
1037           g_strfreev (lines);
1038         }
1039       else
1040         g_warning ("%s: header not found, skipping\n", filename);
1041
1042       g_free (text);
1043     }
1044
1045   g_free (filename);
1046 }
1047
1048
1049 static void
1050 xdg_mime_reload (void *user_data)
1051 {
1052   need_reload = TRUE;
1053 }
1054
1055 static void
1056 tree_magic_shutdown (void)
1057 {
1058   g_list_free_full (tree_matches, (GDestroyNotify) tree_match_free);
1059   tree_matches = NULL;
1060 }
1061
1062 static void
1063 tree_magic_init (void)
1064 {
1065   static gboolean initialized = FALSE;
1066   const gchar *dir;
1067   const gchar * const * dirs;
1068   int i;
1069
1070   if (!initialized)
1071     {
1072       initialized = TRUE;
1073
1074       xdg_mime_register_reload_callback (xdg_mime_reload, NULL, NULL);
1075       need_reload = TRUE;
1076     }
1077
1078   if (need_reload)
1079     {
1080       need_reload = FALSE;
1081
1082       tree_magic_shutdown ();
1083
1084       dir = g_get_user_data_dir ();
1085       read_tree_magic_from_directory (dir);
1086       dirs = g_get_system_data_dirs ();
1087       for (i = 0; dirs[i]; i++)
1088         read_tree_magic_from_directory (dirs[i]);
1089     }
1090 }
1091
1092 /* a filtering enumerator */
1093
1094 typedef struct
1095 {
1096   gchar *path;
1097   gint depth;
1098   gboolean ignore_case;
1099   gchar **components;
1100   gchar **case_components;
1101   GFileEnumerator **enumerators;
1102   GFile **children;
1103 } Enumerator;
1104
1105 static gboolean
1106 component_match (Enumerator  *e,
1107                  gint         depth,
1108                  const gchar *name)
1109 {
1110   gchar *case_folded, *key;
1111   gboolean found;
1112
1113   if (strcmp (name, e->components[depth]) == 0)
1114     return TRUE;
1115
1116   if (!e->ignore_case)
1117     return FALSE;
1118
1119   case_folded = g_utf8_casefold (name, -1);
1120   key = g_utf8_collate_key (case_folded, -1);
1121
1122   found = strcmp (key, e->case_components[depth]) == 0;
1123
1124   g_free (case_folded);
1125   g_free (key);
1126
1127   return found;
1128 }
1129
1130 static GFile *
1131 next_match_recurse (Enumerator *e,
1132                     gint        depth)
1133 {
1134   GFile *file;
1135   GFileInfo *info;
1136   const gchar *name;
1137
1138   while (TRUE)
1139     {
1140       if (e->enumerators[depth] == NULL)
1141         {
1142           if (depth > 0)
1143             {
1144               file = next_match_recurse (e, depth - 1);
1145               if (file)
1146                 {
1147                   e->children[depth] = file;
1148                   e->enumerators[depth] = g_file_enumerate_children (file,
1149                                                                      G_FILE_ATTRIBUTE_STANDARD_NAME,
1150                                                                      G_FILE_QUERY_INFO_NONE,
1151                                                                      NULL,
1152                                                                      NULL);
1153                 }
1154             }
1155           if (e->enumerators[depth] == NULL)
1156             return NULL;
1157         }
1158
1159       while ((info = g_file_enumerator_next_file (e->enumerators[depth], NULL, NULL)))
1160         {
1161           name = g_file_info_get_name (info);
1162           if (component_match (e, depth, name))
1163             {
1164               file = g_file_get_child (e->children[depth], name);
1165               g_object_unref (info);
1166               return file;
1167             }
1168           g_object_unref (info);
1169         }
1170
1171       g_object_unref (e->enumerators[depth]);
1172       e->enumerators[depth] = NULL;
1173       g_object_unref (e->children[depth]);
1174       e->children[depth] = NULL;
1175     }
1176 }
1177
1178 static GFile *
1179 enumerator_next (Enumerator *e)
1180 {
1181   return next_match_recurse (e, e->depth - 1);
1182 }
1183
1184 static Enumerator *
1185 enumerator_new (GFile      *root,
1186                 const char *path,
1187                 gboolean    ignore_case)
1188 {
1189   Enumerator *e;
1190   gint i;
1191   gchar *case_folded;
1192
1193   e = g_new0 (Enumerator, 1);
1194   e->path = g_strdup (path);
1195   e->ignore_case = ignore_case;
1196
1197   e->components = g_strsplit (e->path, G_DIR_SEPARATOR_S, -1);
1198   e->depth = g_strv_length (e->components);
1199   if (e->ignore_case)
1200     {
1201       e->case_components = g_new0 (char *, e->depth + 1);
1202       for (i = 0; e->components[i]; i++)
1203         {
1204           case_folded = g_utf8_casefold (e->components[i], -1);
1205           e->case_components[i] = g_utf8_collate_key (case_folded, -1);
1206           g_free (case_folded);
1207         }
1208     }
1209
1210   e->children = g_new0 (GFile *, e->depth);
1211   e->children[0] = g_object_ref (root);
1212   e->enumerators = g_new0 (GFileEnumerator *, e->depth);
1213   e->enumerators[0] = g_file_enumerate_children (root,
1214                                                  G_FILE_ATTRIBUTE_STANDARD_NAME,
1215                                                  G_FILE_QUERY_INFO_NONE,
1216                                                  NULL,
1217                                                  NULL);
1218
1219   return e;
1220 }
1221
1222 static void
1223 enumerator_free (Enumerator *e)
1224 {
1225   gint i;
1226
1227   for (i = 0; i < e->depth; i++)
1228     {
1229       if (e->enumerators[i])
1230         g_object_unref (e->enumerators[i]);
1231       if (e->children[i])
1232         g_object_unref (e->children[i]);
1233     }
1234
1235   g_free (e->enumerators);
1236   g_free (e->children);
1237   g_strfreev (e->components);
1238   if (e->case_components)
1239     g_strfreev (e->case_components);
1240   g_free (e->path);
1241   g_free (e);
1242 }
1243
1244 static gboolean
1245 matchlet_match (TreeMatchlet *matchlet,
1246                 GFile        *root)
1247 {
1248   GFile *file;
1249   GFileInfo *info;
1250   gboolean result;
1251   const gchar *attrs;
1252   Enumerator *e;
1253   GList *l;
1254
1255   e = enumerator_new (root, matchlet->path, !matchlet->match_case);
1256
1257   do
1258     {
1259       file = enumerator_next (e);
1260       if (!file)
1261         {
1262           enumerator_free (e);
1263           return FALSE;
1264         }
1265
1266       if (matchlet->mimetype)
1267         attrs = G_FILE_ATTRIBUTE_STANDARD_TYPE ","
1268                 G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE ","
1269                 G_FILE_ATTRIBUTE_STANDARD_CONTENT_TYPE;
1270       else
1271         attrs = G_FILE_ATTRIBUTE_STANDARD_TYPE ","
1272                 G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE;
1273       info = g_file_query_info (file,
1274                                 attrs,
1275                                 G_FILE_QUERY_INFO_NONE,
1276                                 NULL,
1277                                 NULL);
1278       if (info)
1279         {
1280           result = TRUE;
1281
1282           if (matchlet->type != G_FILE_TYPE_UNKNOWN &&
1283               g_file_info_get_file_type (info) != matchlet->type)
1284             result = FALSE;
1285
1286           if (matchlet->executable &&
1287               !g_file_info_get_attribute_boolean (info, G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE))
1288             result = FALSE;
1289         }
1290       else
1291         result = FALSE;
1292
1293       if (result && matchlet->non_empty)
1294         {
1295           GFileEnumerator *child_enum;
1296           GFileInfo *child_info;
1297
1298           child_enum = g_file_enumerate_children (file,
1299                                                   G_FILE_ATTRIBUTE_STANDARD_NAME,
1300                                                   G_FILE_QUERY_INFO_NONE,
1301                                                   NULL,
1302                                                   NULL);
1303
1304           if (child_enum)
1305             {
1306               child_info = g_file_enumerator_next_file (child_enum, NULL, NULL);
1307               if (child_info)
1308                 g_object_unref (child_info);
1309               else
1310                 result = FALSE;
1311               g_object_unref (child_enum);
1312             }
1313           else
1314             result = FALSE;
1315         }
1316
1317       if (result && matchlet->mimetype)
1318         {
1319           if (strcmp (matchlet->mimetype, g_file_info_get_content_type (info)) != 0)
1320             result = FALSE;
1321         }
1322
1323       if (info)
1324         g_object_unref (info);
1325       g_object_unref (file);
1326     }
1327   while (!result);
1328
1329   enumerator_free (e);
1330
1331   if (!matchlet->matches)
1332     return TRUE;
1333
1334   for (l = matchlet->matches; l; l = l->next)
1335     {
1336       TreeMatchlet *submatchlet;
1337
1338       submatchlet = l->data;
1339       if (matchlet_match (submatchlet, root))
1340         return TRUE;
1341     }
1342
1343   return FALSE;
1344 }
1345
1346 static void
1347 match_match (TreeMatch    *match,
1348              GFile        *root,
1349              GPtrArray    *types)
1350 {
1351   GList *l;
1352
1353   for (l = match->matches; l; l = l->next)
1354     {
1355       TreeMatchlet *matchlet = l->data;
1356       if (matchlet_match (matchlet, root))
1357         {
1358           g_ptr_array_add (types, g_strdup (match->contenttype));
1359           break;
1360         }
1361     }
1362 }
1363
1364 /**
1365  * g_content_type_guess_for_tree:
1366  * @root: the root of the tree to guess a type for
1367  *
1368  * Tries to guess the type of the tree with root @root, by
1369  * looking at the files it contains. The result is an array
1370  * of content types, with the best guess coming first.
1371  *
1372  * The types returned all have the form x-content/foo, e.g.
1373  * x-content/audio-cdda (for audio CDs) or x-content/image-dcf
1374  * (for a camera memory card). See the
1375  * [shared-mime-info](http://www.freedesktop.org/wiki/Specifications/shared-mime-info-spec)
1376  * specification for more on x-content types.
1377  *
1378  * This function is useful in the implementation of
1379  * g_mount_guess_content_type().
1380  *
1381  * Returns: (transfer full) (array zero-terminated=1): an %NULL-terminated
1382  *     array of zero or more content types. Free with g_strfreev()
1383  *
1384  * Since: 2.18
1385  */
1386 gchar **
1387 g_content_type_guess_for_tree (GFile *root)
1388 {
1389   GPtrArray *types;
1390   GList *l;
1391
1392   types = g_ptr_array_new ();
1393
1394   G_LOCK (gio_treemagic);
1395
1396   tree_magic_init ();
1397   for (l = tree_matches; l; l = l->next)
1398     {
1399       TreeMatch *match = l->data;
1400       match_match (match, root, types);
1401     }
1402
1403   G_UNLOCK (gio_treemagic);
1404
1405   g_ptr_array_add (types, NULL);
1406
1407   return (gchar **)g_ptr_array_free (types, FALSE);
1408 }