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 #pragma GCC diagnostic push
 397 #pragma GCC diagnostic ignored "-Wformat-nonliteral"
 398
 399 static GIcon *
 400 g_content_type_get_icon_internal (const gchar *type,
 401                                   gboolean     symbolic)
 402 {
 403   char *mimetype_icon;
 404   char *generic_mimetype_icon = NULL;
 405   char *q;
 406   char *xdg_mimetype_icon = NULL;
 407   char *xdg_mimetype_generic_icon;
 408   char *icon_names[3];
 409   int n = 0;
 410   GIcon *themed_icon;
 411   const char *file_template;
 412   const char  *xdg_icon;
 413
 414   g_return_val_if_fail (type != NULL, NULL);
 415
 416   if (symbolic)
 417     {
 418       file_template = "%s-symbolic";
 419     }
 420   else
 421     {
 422       file_template = "%s";
 423     }
 424
 425   G_LOCK (gio_xdgmime);
 426   xdg_icon = xdg_mime_get_icon (type);
 427   G_UNLOCK (gio_xdgmime);
 428    if (xdg_icon != NULL)
 429     xdg_mimetype_icon = g_strdup_printf (file_template, xdg_icon);
 430
 431   if (xdg_mimetype_icon)
 432     icon_names[n++] = xdg_mimetype_icon;
 433
 434   mimetype_icon = g_strdup_printf (file_template, type);
 435   while ((q = strchr (mimetype_icon, '/')) != NULL)
 436     *q = '-';
 437
 438   icon_names[n++] = mimetype_icon;
 439
 440   xdg_mimetype_generic_icon = g_content_type_get_generic_icon_name (type);
 441   if (xdg_mimetype_generic_icon)
 442     generic_mimetype_icon = g_strdup_printf (file_template, xdg_mimetype_generic_icon);
 443   if (generic_mimetype_icon)
 444     icon_names[n++] = generic_mimetype_icon;
 445
 446   themed_icon = g_themed_icon_new_from_names (icon_names, n);
 447
 448   g_free (xdg_mimetype_icon);
 449   g_free (xdg_mimetype_generic_icon);
 450   g_free (mimetype_icon);
 451   g_free (generic_mimetype_icon);
 452
 453   return themed_icon;
 454 }
 455 #pragma GCC diagnostic pop
 456
 457 /**
 458  * g_content_type_get_icon:
 459  * @type: a content type string
 460  *
 461  * Gets the icon for a content type.
 462  *
 463  * Returns: (transfer full): #GIcon corresponding to the content type. Free the returned
 464  *     object with g_object_unref()
 465  */
 466 GIcon *
 467 g_content_type_get_icon (const gchar *type)
 468 {
 469   return g_content_type_get_icon_internal (type, FALSE);
 470 }
 471
 472 /**
 473  * g_content_type_get_symbolic_icon:
 474  * @type: a content type string
 475  *
 476  * Gets the symbolic icon for a content type.
 477  *
 478  * Returns: (transfer full): symbolic #GIcon corresponding to the content type.
 479  *     Free the returned object with g_object_unref()
 480  *
 481  * Since: 2.34
 482  */
 483 GIcon *
 484 g_content_type_get_symbolic_icon (const gchar *type)
 485 {
 486   return g_content_type_get_icon_internal (type, TRUE);
 487 }
 488
 489 /**
 490  * g_content_type_get_generic_icon_name:
 491  * @type: a content type string
 492  *
 493  * Gets the generic icon name for a content type.
 494  *
 495  * See the <ulink url="http://www.freedesktop.org/wiki/Specifications/shared-mime-info-spec">shared-mime-info</ulink>
 496  * specification for more on the generic icon name.
 497  *
 498  * Returns: (allow-none): the registered generic icon name for the given @type,
 499  *     or %NULL if unknown. Free with g_free()
 500  *
 501  * Since: 2.34
 502  */
 503 gchar *
 504 g_content_type_get_generic_icon_name (const gchar *type)
 505 {
 506   const gchar *xdg_icon_name;
 507   gchar *icon_name;
 508
 509   G_LOCK (gio_xdgmime);
 510   xdg_icon_name = xdg_mime_get_generic_icon (type);
 511   G_UNLOCK (gio_xdgmime);
 512
 513   if (!xdg_icon_name)
 514     {
 515       const char *p;
 516       const char *suffix = "-x-generic";
 517
 518       p = strchr (type, '/');
 519       if (p == NULL)
 520         p = type + strlen (type);
 521
 522       icon_name = g_malloc (p - type + strlen (suffix) + 1);
 523       memcpy (icon_name, type, p - type);
 524       memcpy (icon_name + (p - type), suffix, strlen (suffix));
 525       icon_name[(p - type) + strlen (suffix)] = 0;
 526     }
 527   else
 528     {
 529       icon_name = g_strdup (xdg_icon_name);
 530     }
 531
 532   return icon_name;
 533 }
 534
 535 /**
 536  * g_content_type_can_be_executable:
 537  * @type: a content type string
 538  *
 539  * Checks if a content type can be executable. Note that for instance
 540  * things like text files can be executables (i.e. scripts and batch files).
 541  *
 542  * Returns: %TRUE if the file type corresponds to a type that
 543  *     can be executable, %FALSE otherwise.
 544  */
 545 gboolean
 546 g_content_type_can_be_executable (const gchar *type)
 547 {
 548   g_return_val_if_fail (type != NULL, FALSE);
 549
 550   if (g_content_type_is_a (type, "application/x-executable")  ||
 551       g_content_type_is_a (type, "text/plain"))
 552     return TRUE;
 553
 554   return FALSE;
 555 }
 556
 557 static gboolean
 558 looks_like_text (const guchar *data, gsize data_size)
 559 {
 560   gsize i;
 561   char c;
 562
 563   for (i = 0; i < data_size; i++)
 564     {
 565       c = data[i];
 566
 567       if (g_ascii_iscntrl (c) &&
 568           !g_ascii_isspace (c) &&
 569           c != '\b')
 570         return FALSE;
 571     }
 572   return TRUE;
 573 }
 574
 575 /**
 576  * g_content_type_from_mime_type:
 577  * @mime_type: a mime type string
 578  *
 579  * Tries to find a content type based on the mime type name.
 580  *
 581  * Returns: (allow-none): Newly allocated string with content type
 582  *     or %NULL. Free with g_free()
 583  *
 584  * Since: 2.18
 585  **/
 586 gchar *
 587 g_content_type_from_mime_type (const gchar *mime_type)
 588 {
 589   char *umime;
 590
 591   g_return_val_if_fail (mime_type != NULL, NULL);
 592
 593   G_LOCK (gio_xdgmime);
 594   /* mime type and content type are same on unixes */
 595   umime = g_strdup (xdg_mime_unalias_mime_type (mime_type));
 596   G_UNLOCK (gio_xdgmime);
 597
 598   return umime;
 599 }
 600
 601 /**
 602  * g_content_type_guess:
 603  * @filename: (allow-none): a string, or %NULL
 604  * @data: (allow-none) (array length=data_size): a stream of data, or %NULL
 605  * @data_size: the size of @data
 606  * @result_uncertain: (allow-none) (out): return location for the certainty
 607  *     of the result, or %NULL
 608  *
 609  * Guesses the content type based on example data. If the function is
 610  * uncertain, @result_uncertain will be set to %TRUE. Either @filename
 611  * or @data may be %NULL, in which case the guess will be based solely
 612  * on the other argument.
 613  *
 614  * Returns: a string indicating a guessed content type for the
 615  *     given data. Free with g_free()
 616  */
 617 gchar *
 618 g_content_type_guess (const gchar  *filename,
 619                       const guchar *data,
 620                       gsize         data_size,
 621                       gboolean     *result_uncertain)
 622 {
 623   char *basename;
 624   const char *name_mimetypes[10], *sniffed_mimetype;
 625   char *mimetype;
 626   int i;
 627   int n_name_mimetypes;
 628   int sniffed_prio;
 629
 630   sniffed_prio = 0;
 631   n_name_mimetypes = 0;
 632   sniffed_mimetype = XDG_MIME_TYPE_UNKNOWN;
 633
 634   if (result_uncertain)
 635     *result_uncertain = FALSE;
 636
 637   /* our test suite and potentially other code used -1 in the past, which is
 638    * not documented and not allowed; guard against that */
 639   g_return_val_if_fail (data_size != (gsize) -1, g_strdup (XDG_MIME_TYPE_UNKNOWN));
 640
 641   G_LOCK (gio_xdgmime);
 642
 643   if (filename)
 644     {
 645       i = strlen (filename);
 646       if (filename[i - 1] == '/')
 647         {
 648           name_mimetypes[0] = "inode/directory";
 649           name_mimetypes[1] = NULL;
 650           n_name_mimetypes = 1;
 651           if (result_uncertain)
 652             *result_uncertain = TRUE;
 653         }
 654       else
 655         {
 656           basename = g_path_get_basename (filename);
 657           n_name_mimetypes = xdg_mime_get_mime_types_from_file_name (basename, name_mimetypes, 10);
 658           g_free (basename);
 659         }
 660     }
 661
 662   /* Got an extension match, and no conflicts. This is it. */
 663   if (n_name_mimetypes == 1)
 664     {
 665       gchar *s = g_strdup (name_mimetypes[0]);
 666       G_UNLOCK (gio_xdgmime);
 667       return s;
 668     }
 669
 670   if (data)
 671     {
 672       sniffed_mimetype = xdg_mime_get_mime_type_for_data (data, data_size, &sniffed_prio);
 673       if (sniffed_mimetype == XDG_MIME_TYPE_UNKNOWN &&
 674           data &&
 675           looks_like_text (data, data_size))
 676         sniffed_mimetype = "text/plain";
 677
 678       /* For security reasons we don't ever want to sniff desktop files
 679        * where we know the filename and it doesn't have a .desktop extension.
 680        * This is because desktop files allow executing any application and
 681        * we don't want to make it possible to hide them looking like something
 682        * else.
 683        */
 684       if (filename != NULL &&
 685           strcmp (sniffed_mimetype, "application/x-desktop") == 0)
 686         sniffed_mimetype = "text/plain";
 687     }
 688
 689   if (n_name_mimetypes == 0)
 690     {
 691       if (sniffed_mimetype == XDG_MIME_TYPE_UNKNOWN &&
 692           result_uncertain)
 693         *result_uncertain = TRUE;
 694
 695       mimetype = g_strdup (sniffed_mimetype);
 696     }
 697   else
 698     {
 699       mimetype = NULL;
 700       if (sniffed_mimetype != XDG_MIME_TYPE_UNKNOWN)
 701         {
 702           if (sniffed_prio >= 80) /* High priority sniffing match, use that */
 703             mimetype = g_strdup (sniffed_mimetype);
 704           else
 705             {
 706               /* There are conflicts between the name matches and we
 707                * have a sniffed type, use that as a tie breaker.
 708                */
 709               for (i = 0; i < n_name_mimetypes; i++)
 710                 {
 711                   if ( xdg_mime_mime_type_subclass (name_mimetypes[i], sniffed_mimetype))
 712                     {
 713                       /* This nametype match is derived from (or the same as)
 714                        * the sniffed type). This is probably it.
 715                        */
 716                       mimetype = g_strdup (name_mimetypes[i]);
 717                       break;
 718                     }
 719                 }
 720             }
 721         }
 722
 723       if (mimetype == NULL)
 724         {
 725           /* Conflicts, and sniffed type was no help or not there.
 726            * Guess on the first one
 727            */
 728           mimetype = g_strdup (name_mimetypes[0]);
 729           if (result_uncertain)
 730             *result_uncertain = TRUE;
 731         }
 732     }
 733
 734   G_UNLOCK (gio_xdgmime);
 735
 736   return mimetype;
 737 }
 738
 739 static void
 740 enumerate_mimetypes_subdir (const char *dir,
 741                             const char *prefix,
 742                             GHashTable *mimetypes)
 743 {
 744   DIR *d;
 745   struct dirent *ent;
 746   char *mimetype;
 747
 748   d = opendir (dir);
 749   if (d)
 750     {
 751       while ((ent = readdir (d)) != NULL)
 752         {
 753           if (g_str_has_suffix (ent->d_name, ".xml"))
 754             {
 755               mimetype = g_strdup_printf ("%s/%.*s", prefix, (int) strlen (ent->d_name) - 4, ent->d_name);
 756               g_hash_table_replace (mimetypes, mimetype, NULL);
 757             }
 758         }
 759       closedir (d);
 760     }
 761 }
 762
 763 static void
 764 enumerate_mimetypes_dir (const char *dir,
 765                          GHashTable *mimetypes)
 766 {
 767   DIR *d;
 768   struct dirent *ent;
 769   char *mimedir;
 770   char *name;
 771
 772   mimedir = g_build_filename (dir, "mime", NULL);
 773
 774   d = opendir (mimedir);
 775   if (d)
 776     {
 777       while ((ent = readdir (d)) != NULL)
 778         {
 779           if (strcmp (ent->d_name, "packages") != 0)
 780             {
 781               name = g_build_filename (mimedir, ent->d_name, NULL);
 782               if (g_file_test (name, G_FILE_TEST_IS_DIR))
 783                 enumerate_mimetypes_subdir (name, ent->d_name, mimetypes);
 784               g_free (name);
 785             }
 786         }
 787       closedir (d);
 788     }
 789
 790   g_free (mimedir);
 791 }
 792
 793 /**
 794  * g_content_types_get_registered:
 795  *
 796  * Gets a list of strings containing all the registered content types
 797  * known to the system. The list and its data should be freed using
 798  * <programlisting>
 799  * g_list_free_full (list, g_free);
 800  * </programlisting>
 801  *
 802  * Returns: (element-type utf8) (transfer full): #GList of the registered content types
 803  */
 804 GList *
 805 g_content_types_get_registered (void)
 806 {
 807   const char * const* dirs;
 808   GHashTable *mimetypes;
 809   GHashTableIter iter;
 810   gpointer key;
 811   int i;
 812   GList *l;
 813
 814   mimetypes = g_hash_table_new_full (g_str_hash, g_str_equal, g_free, NULL);
 815
 816   enumerate_mimetypes_dir (g_get_user_data_dir (), mimetypes);
 817   dirs = g_get_system_data_dirs ();
 818
 819   for (i = 0; dirs[i] != NULL; i++)
 820     enumerate_mimetypes_dir (dirs[i], mimetypes);
 821
 822   l = NULL;
 823   g_hash_table_iter_init (&iter, mimetypes);
 824   while (g_hash_table_iter_next (&iter, &key, NULL))
 825     {
 826       l = g_list_prepend (l, key);
 827       g_hash_table_iter_steal (&iter);
 828     }
 829
 830   g_hash_table_destroy (mimetypes);
 831
 832   return l;
 833 }
 834
 835
 836 /* tree magic data */
 837 static GList *tree_matches = NULL;
 838 static gboolean need_reload = FALSE;
 839
 840 G_LOCK_DEFINE_STATIC (gio_treemagic);
 841
 842 typedef struct
 843 {
 844   gchar *path;
 845   GFileType type;
 846   guint match_case : 1;
 847   guint executable : 1;
 848   guint non_empty  : 1;
 849   guint on_disc    : 1;
 850   gchar *mimetype;
 851   GList *matches;
 852 } TreeMatchlet;
 853
 854 typedef struct
 855 {
 856   gchar *contenttype;
 857   gint priority;
 858   GList *matches;
 859 } TreeMatch;
 860
 861
 862 static void
 863 tree_matchlet_free (TreeMatchlet *matchlet)
 864 {
 865   g_list_free_full (matchlet->matches, (GDestroyNotify) tree_matchlet_free);
 866   g_free (matchlet->path);
 867   g_free (matchlet->mimetype);
 868   g_slice_free (TreeMatchlet, matchlet);
 869 }
 870
 871 static void
 872 tree_match_free (TreeMatch *match)
 873 {
 874   g_list_free_full (match->matches, (GDestroyNotify) tree_matchlet_free);
 875   g_free (match->contenttype);
 876   g_slice_free (TreeMatch, match);
 877 }
 878
 879 static TreeMatch *
 880 parse_header (gchar *line)
 881 {
 882   gint len;
 883   gchar *s;
 884   TreeMatch *match;
 885
 886   len = strlen (line);
 887
 888   if (line[0] != '[' || line[len - 1] != ']')
 889     return NULL;
 890
 891   line[len - 1] = 0;
 892   s = strchr (line, ':');
 893
 894   match = g_slice_new0 (TreeMatch);
 895   match->priority = atoi (line + 1);
 896   match->contenttype = g_strdup (s + 1);
 897
 898   return match;
 899 }
 900
 901 static TreeMatchlet *
 902 parse_match_line (gchar *line,
 903                   gint  *depth)
 904 {
 905   gchar *s, *p;
 906   TreeMatchlet *matchlet;
 907   gchar **parts;
 908   gint i;
 909
 910   matchlet = g_slice_new0 (TreeMatchlet);
 911
 912   if (line[0] == '>')
 913     {
 914       *depth = 0;
 915       s = line;
 916     }
 917   else
 918     {
 919       *depth = atoi (line);
 920       s = strchr (line, '>');
 921     }
 922   s += 2;
 923   p = strchr (s, '"');
 924   *p = 0;
 925
 926   matchlet->path = g_strdup (s);
 927   s = p + 1;
 928   parts = g_strsplit (s, ",", 0);
 929   if (strcmp (parts[0], "=file") == 0)
 930     matchlet->type = G_FILE_TYPE_REGULAR;
 931   else if (strcmp (parts[0], "=directory") == 0)
 932     matchlet->type = G_FILE_TYPE_DIRECTORY;
 933   else if (strcmp (parts[0], "=link") == 0)
 934     matchlet->type = G_FILE_TYPE_SYMBOLIC_LINK;
 935   else
 936     matchlet->type = G_FILE_TYPE_UNKNOWN;
 937   for (i = 1; parts[i]; i++)
 938     {
 939       if (strcmp (parts[i], "executable") == 0)
 940         matchlet->executable = 1;
 941       else if (strcmp (parts[i], "match-case") == 0)
 942         matchlet->match_case = 1;
 943       else if (strcmp (parts[i], "non-empty") == 0)
 944         matchlet->non_empty = 1;
 945       else if (strcmp (parts[i], "on-disc") == 0)
 946         matchlet->on_disc = 1;
 947       else
 948         matchlet->mimetype = g_strdup (parts[i]);
 949     }
 950
 951   g_strfreev (parts);
 952
 953   return matchlet;
 954 }
 955
 956 static gint
 957 cmp_match (gconstpointer a, gconstpointer b)
 958 {
 959   const TreeMatch *aa = (const TreeMatch *)a;
 960   const TreeMatch *bb = (const TreeMatch *)b;
 961
 962   return bb->priority - aa->priority;
 963 }
 964
 965 static void
 966 insert_match (TreeMatch *match)
 967 {
 968   tree_matches = g_list_insert_sorted (tree_matches, match, cmp_match);
 969 }
 970
 971 static void
 972 insert_matchlet (TreeMatch    *match,
 973                  TreeMatchlet *matchlet,
 974                  gint          depth)
 975 {
 976   if (depth == 0)
 977     match->matches = g_list_append (match->matches, matchlet);
 978   else
 979     {
 980       GList *last;
 981       TreeMatchlet *m;
 982
 983       last = g_list_last (match->matches);
 984       if (!last)
 985         {
 986           tree_matchlet_free (matchlet);
 987           g_warning ("can't insert tree matchlet at depth %d", depth);
 988           return;
 989         }
 990
 991       m = (TreeMatchlet *) last->data;
 992       while (--depth > 0)
 993         {
 994           last = g_list_last (m->matches);
 995           if (!last)
 996             {
 997               tree_matchlet_free (matchlet);
 998               g_warning ("can't insert tree matchlet at depth %d", depth);
 999               return;
1000             }
1001
1002           m = (TreeMatchlet *) last->data;
1003         }
1004       m->matches = g_list_append (m->matches, matchlet);
1005     }
1006 }
1007
1008 static void
1009 read_tree_magic_from_directory (const gchar *prefix)
1010 {
1011   gchar *filename;
1012   gchar *text;
1013   gsize len;
1014   gchar **lines;
1015   gint i;
1016   TreeMatch *match;
1017   TreeMatchlet *matchlet;
1018   gint depth;
1019
1020   filename = g_build_filename (prefix, "mime", "treemagic", NULL);
1021
1022   if (g_file_get_contents (filename, &text, &len, NULL))
1023     {
1024       if (strcmp (text, "MIME-TreeMagic") == 0)
1025         {
1026           lines = g_strsplit (text + strlen ("MIME-TreeMagic") + 2, "\n", 0);
1027           match = NULL;
1028           for (i = 0; lines[i] && lines[i][0]; i++)
1029             {
1030               if (lines[i][0] == '[')
1031                 {
1032                   match = parse_header (lines[i]);
1033                   insert_match (match);
1034                 }
1035               else
1036                 {
1037                   matchlet = parse_match_line (lines[i], &depth);
1038                   insert_matchlet (match, matchlet, depth);
1039                 }
1040             }
1041
1042           g_strfreev (lines);
1043         }
1044       else
1045         g_warning ("%s: header not found, skipping\n", filename);
1046
1047       g_free (text);
1048     }
1049
1050   g_free (filename);
1051 }
1052
1053
1054 static void
1055 xdg_mime_reload (void *user_data)
1056 {
1057   need_reload = TRUE;
1058 }
1059
1060 static void
1061 tree_magic_shutdown (void)
1062 {
1063   g_list_free_full (tree_matches, (GDestroyNotify) tree_match_free);
1064   tree_matches = NULL;
1065 }
1066
1067 static void
1068 tree_magic_init (void)
1069 {
1070   static gboolean initialized = FALSE;
1071   const gchar *dir;
1072   const gchar * const * dirs;
1073   int i;
1074
1075   if (!initialized)
1076     {
1077       initialized = TRUE;
1078
1079       xdg_mime_register_reload_callback (xdg_mime_reload, NULL, NULL);
1080       need_reload = TRUE;
1081     }
1082
1083   if (need_reload)
1084     {
1085       need_reload = FALSE;
1086
1087       tree_magic_shutdown ();
1088
1089       dir = g_get_user_data_dir ();
1090       read_tree_magic_from_directory (dir);
1091       dirs = g_get_system_data_dirs ();
1092       for (i = 0; dirs[i]; i++)
1093         read_tree_magic_from_directory (dirs[i]);
1094     }
1095 }
1096
1097 /* a filtering enumerator */
1098
1099 typedef struct
1100 {
1101   gchar *path;
1102   gint depth;
1103   gboolean ignore_case;
1104   gchar **components;
1105   gchar **case_components;
1106   GFileEnumerator **enumerators;
1107   GFile **children;
1108 } Enumerator;
1109
1110 static gboolean
1111 component_match (Enumerator  *e,
1112                  gint         depth,
1113                  const gchar *name)
1114 {
1115   gchar *case_folded, *key;
1116   gboolean found;
1117
1118   if (strcmp (name, e->components[depth]) == 0)
1119     return TRUE;
1120
1121   if (!e->ignore_case)
1122     return FALSE;
1123
1124   case_folded = g_utf8_casefold (name, -1);
1125   key = g_utf8_collate_key (case_folded, -1);
1126
1127   found = strcmp (key, e->case_components[depth]) == 0;
1128
1129   g_free (case_folded);
1130   g_free (key);
1131
1132   return found;
1133 }
1134
1135 static GFile *
1136 next_match_recurse (Enumerator *e,
1137                     gint        depth)
1138 {
1139   GFile *file;
1140   GFileInfo *info;
1141   const gchar *name;
1142
1143   while (TRUE)
1144     {
1145       if (e->enumerators[depth] == NULL)
1146         {
1147           if (depth > 0)
1148             {
1149               file = next_match_recurse (e, depth - 1);
1150               if (file)
1151                 {
1152                   e->children[depth] = file;
1153                   e->enumerators[depth] = g_file_enumerate_children (file,
1154                                                                      G_FILE_ATTRIBUTE_STANDARD_NAME,
1155                                                                      G_FILE_QUERY_INFO_NONE,
1156                                                                      NULL,
1157                                                                      NULL);
1158                 }
1159             }
1160           if (e->enumerators[depth] == NULL)
1161             return NULL;
1162         }
1163
1164       while ((info = g_file_enumerator_next_file (e->enumerators[depth], NULL, NULL)))
1165         {
1166           name = g_file_info_get_name (info);
1167           if (component_match (e, depth, name))
1168             {
1169               file = g_file_get_child (e->children[depth], name);
1170               g_object_unref (info);
1171               return file;
1172             }
1173           g_object_unref (info);
1174         }
1175
1176       g_object_unref (e->enumerators[depth]);
1177       e->enumerators[depth] = NULL;
1178       g_object_unref (e->children[depth]);
1179       e->children[depth] = NULL;
1180     }
1181 }
1182
1183 static GFile *
1184 enumerator_next (Enumerator *e)
1185 {
1186   return next_match_recurse (e, e->depth - 1);
1187 }
1188
1189 static Enumerator *
1190 enumerator_new (GFile      *root,
1191                 const char *path,
1192                 gboolean    ignore_case)
1193 {
1194   Enumerator *e;
1195   gint i;
1196   gchar *case_folded;
1197
1198   e = g_new0 (Enumerator, 1);
1199   e->path = g_strdup (path);
1200   e->ignore_case = ignore_case;
1201
1202   e->components = g_strsplit (e->path, G_DIR_SEPARATOR_S, -1);
1203   e->depth = g_strv_length (e->components);
1204   if (e->ignore_case)
1205     {
1206       e->case_components = g_new0 (char *, e->depth + 1);
1207       for (i = 0; e->components[i]; i++)
1208         {
1209           case_folded = g_utf8_casefold (e->components[i], -1);
1210           e->case_components[i] = g_utf8_collate_key (case_folded, -1);
1211           g_free (case_folded);
1212         }
1213     }
1214
1215   e->children = g_new0 (GFile *, e->depth);
1216   e->children[0] = g_object_ref (root);
1217   e->enumerators = g_new0 (GFileEnumerator *, e->depth);
1218   e->enumerators[0] = g_file_enumerate_children (root,
1219                                                  G_FILE_ATTRIBUTE_STANDARD_NAME,
1220                                                  G_FILE_QUERY_INFO_NONE,
1221                                                  NULL,
1222                                                  NULL);
1223
1224   return e;
1225 }
1226
1227 static void
1228 enumerator_free (Enumerator *e)
1229 {
1230   gint i;
1231
1232   for (i = 0; i < e->depth; i++)
1233     {
1234       if (e->enumerators[i])
1235         g_object_unref (e->enumerators[i]);
1236       if (e->children[i])
1237         g_object_unref (e->children[i]);
1238     }
1239
1240   g_free (e->enumerators);
1241   g_free (e->children);
1242   g_strfreev (e->components);
1243   if (e->case_components)
1244     g_strfreev (e->case_components);
1245   g_free (e->path);
1246   g_free (e);
1247 }
1248
1249 static gboolean
1250 matchlet_match (TreeMatchlet *matchlet,
1251                 GFile        *root)
1252 {
1253   GFile *file;
1254   GFileInfo *info;
1255   gboolean result;
1256   const gchar *attrs;
1257   Enumerator *e;
1258   GList *l;
1259
1260   e = enumerator_new (root, matchlet->path, !matchlet->match_case);
1261
1262   do
1263     {
1264       file = enumerator_next (e);
1265       if (!file)
1266         {
1267           enumerator_free (e);
1268           return FALSE;
1269         }
1270
1271       if (matchlet->mimetype)
1272         attrs = G_FILE_ATTRIBUTE_STANDARD_TYPE ","
1273                 G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE ","
1274                 G_FILE_ATTRIBUTE_STANDARD_CONTENT_TYPE;
1275       else
1276         attrs = G_FILE_ATTRIBUTE_STANDARD_TYPE ","
1277                 G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE;
1278       info = g_file_query_info (file,
1279                                 attrs,
1280                                 G_FILE_QUERY_INFO_NONE,
1281                                 NULL,
1282                                 NULL);
1283       if (info)
1284         {
1285           result = TRUE;
1286
1287           if (matchlet->type != G_FILE_TYPE_UNKNOWN &&
1288               g_file_info_get_file_type (info) != matchlet->type)
1289             result = FALSE;
1290
1291           if (matchlet->executable &&
1292               !g_file_info_get_attribute_boolean (info, G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE))
1293             result = FALSE;
1294         }
1295       else
1296         result = FALSE;
1297
1298       if (result && matchlet->non_empty)
1299         {
1300           GFileEnumerator *child_enum;
1301           GFileInfo *child_info;
1302
1303           child_enum = g_file_enumerate_children (file,
1304                                                   G_FILE_ATTRIBUTE_STANDARD_NAME,
1305                                                   G_FILE_QUERY_INFO_NONE,
1306                                                   NULL,
1307                                                   NULL);
1308
1309           if (child_enum)
1310             {
1311               child_info = g_file_enumerator_next_file (child_enum, NULL, NULL);
1312               if (child_info)
1313                 g_object_unref (child_info);
1314               else
1315                 result = FALSE;
1316               g_object_unref (child_enum);
1317             }
1318           else
1319             result = FALSE;
1320         }
1321
1322       if (result && matchlet->mimetype)
1323         {
1324           if (strcmp (matchlet->mimetype, g_file_info_get_content_type (info)) != 0)
1325             result = FALSE;
1326         }
1327
1328       g_object_unref (info);
1329       g_object_unref (file);
1330     }
1331   while (!result);
1332
1333   enumerator_free (e);
1334
1335   if (!matchlet->matches)
1336     return TRUE;
1337
1338   for (l = matchlet->matches; l; l = l->next)
1339     {
1340       TreeMatchlet *submatchlet;
1341
1342       submatchlet = l->data;
1343       if (matchlet_match (submatchlet, root))
1344         return TRUE;
1345     }
1346
1347   return FALSE;
1348 }
1349
1350 static void
1351 match_match (TreeMatch    *match,
1352              GFile        *root,
1353              GPtrArray    *types)
1354 {
1355   GList *l;
1356
1357   for (l = match->matches; l; l = l->next)
1358     {
1359       TreeMatchlet *matchlet = l->data;
1360       if (matchlet_match (matchlet, root))
1361         {
1362           g_ptr_array_add (types, g_strdup (match->contenttype));
1363           break;
1364         }
1365     }
1366 }
1367
1368 /**
1369  * g_content_type_guess_for_tree:
1370  * @root: the root of the tree to guess a type for
1371  *
1372  * Tries to guess the type of the tree with root @root, by
1373  * looking at the files it contains. The result is an array
1374  * of content types, with the best guess coming first.
1375  *
1376  * The types returned all have the form x-content/foo, e.g.
1377  * x-content/audio-cdda (for audio CDs) or x-content/image-dcf
1378  * (for a camera memory card). See the <ulink url="http://www.freedesktop.org/wiki/Specifications/shared-mime-info-spec">shared-mime-info</ulink>
1379  * specification for more on x-content types.
1380  *
1381  * This function is useful in the implementation of
1382  * g_mount_guess_content_type().
1383  *
1384  * Returns: (transfer full) (array zero-terminated=1): an %NULL-terminated
1385  *     array of zero or more content types. Free with g_strfreev()
1386  *
1387  * Since: 2.18
1388  */
1389 gchar **
1390 g_content_type_guess_for_tree (GFile *root)
1391 {
1392   GPtrArray *types;
1393   GList *l;
1394
1395   types = g_ptr_array_new ();
1396
1397   G_LOCK (gio_treemagic);
1398
1399   tree_magic_init ();
1400   for (l = tree_matches; l; l = l->next)
1401     {
1402       TreeMatch *match = l->data;
1403       match_match (match, root, types);
1404     }
1405
1406   G_UNLOCK (gio_treemagic);
1407
1408   g_ptr_array_add (types, NULL);
1409
1410   return (gchar **)g_ptr_array_free (types, FALSE);
1411 }