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;
 403   char *q;
 404   char *xdg_mimetype_icon;
 405   char *legacy_mimetype_icon;
 406   char *xdg_mimetype_generic_icon;
 407   char *icon_names[5];
 408   int n = 0;
 409   const char *p;
 410   GIcon *themed_icon;
 411   const char *file_template;
 412   const char *generic_suffix;
 413
 414   g_return_val_if_fail (type != NULL, NULL);
 415
 416   if (symbolic)
 417     {
 418       file_template = "%s-symbolic";
 419       generic_suffix = "-x-generic-symbolic";
 420     }
 421   else
 422     {
 423       file_template = "%s";
 424       generic_suffix = "-x-generic";
 425     }
 426
 427   G_LOCK (gio_xdgmime);
 428   xdg_mimetype_icon = g_strdup_printf (file_template, xdg_mime_get_icon (type));
 429   xdg_mimetype_generic_icon = g_strdup_printf (file_template, xdg_mime_get_generic_icon (type));
 430   G_UNLOCK (gio_xdgmime);
 431
 432   mimetype_icon = g_strdup_printf (file_template, type);
 433
 434   while ((q = strchr (mimetype_icon, '/')) != NULL)
 435     *q = '-';
 436
 437   p = strchr (type, '/');
 438   if (p == NULL)
 439     p = type + strlen (type);
 440
 441   /* Not all icons have migrated to the new icon theme spec, look for old names too */
 442   legacy_mimetype_icon = g_strconcat ("gnome-mime-", mimetype_icon, NULL);
 443
 444   generic_mimetype_icon = g_malloc (p - type + strlen (generic_suffix) + 1);
 445   memcpy (generic_mimetype_icon, type, p - type);
 446   memcpy (generic_mimetype_icon + (p - type), generic_suffix, strlen (generic_suffix));
 447   generic_mimetype_icon[(p - type) + strlen (generic_suffix)] = 0;
 448
 449   if (xdg_mimetype_icon)
 450     icon_names[n++] = xdg_mimetype_icon;
 451
 452   icon_names[n++] = mimetype_icon;
 453   icon_names[n++] = legacy_mimetype_icon;
 454
 455   if (xdg_mimetype_generic_icon)
 456     icon_names[n++] = xdg_mimetype_generic_icon;
 457
 458   icon_names[n++] = generic_mimetype_icon;
 459
 460   themed_icon = g_themed_icon_new_from_names (icon_names, n);
 461
 462   g_free (xdg_mimetype_icon);
 463   g_free (xdg_mimetype_generic_icon);
 464   g_free (mimetype_icon);
 465   g_free (legacy_mimetype_icon);
 466   g_free (generic_mimetype_icon);
 467
 468   return themed_icon;
 469 }
 470
 471 /**
 472  * g_content_type_get_icon:
 473  * @type: a content type string
 474  *
 475  * Gets the icon for a content type.
 476  *
 477  * Returns: (transfer full): #GIcon corresponding to the content type. Free the returned
 478  *     object with g_object_unref()
 479  */
 480 GIcon *
 481 g_content_type_get_icon (const gchar *type)
 482 {
 483   return g_content_type_get_icon_internal (type, FALSE);
 484 }
 485
 486 /**
 487  * g_content_type_get_symbolic_icon:
 488  * @type: a content type string
 489  *
 490  * Gets the symbolic icon for a content type.
 491  *
 492  * Returns: (transfer full): symbolic #GIcon corresponding to the content type.
 493  *     Free the returned object with g_object_unref()
 494  *
 495  * Since: 2.34
 496  */
 497 GIcon *
 498 g_content_type_get_symbolic_icon (const gchar *type)
 499 {
 500   return g_content_type_get_icon_internal (type, TRUE);
 501 }
 502
 503 /**
 504  * g_content_type_can_be_executable:
 505  * @type: a content type string
 506  *
 507  * Checks if a content type can be executable. Note that for instance
 508  * things like text files can be executables (i.e. scripts and batch files).
 509  *
 510  * Returns: %TRUE if the file type corresponds to a type that
 511  *     can be executable, %FALSE otherwise.
 512  */
 513 gboolean
 514 g_content_type_can_be_executable (const gchar *type)
 515 {
 516   g_return_val_if_fail (type != NULL, FALSE);
 517
 518   if (g_content_type_is_a (type, "application/x-executable")  ||
 519       g_content_type_is_a (type, "text/plain"))
 520     return TRUE;
 521
 522   return FALSE;
 523 }
 524
 525 static gboolean
 526 looks_like_text (const guchar *data, gsize data_size)
 527 {
 528   gsize i;
 529   char c;
 530
 531   for (i = 0; i < data_size; i++)
 532     {
 533       c = data[i];
 534
 535       if (g_ascii_iscntrl (c) &&
 536           !g_ascii_isspace (c) &&
 537           c != '\b')
 538         return FALSE;
 539     }
 540   return TRUE;
 541 }
 542
 543 /**
 544  * g_content_type_from_mime_type:
 545  * @mime_type: a mime type string
 546  *
 547  * Tries to find a content type based on the mime type name.
 548  *
 549  * Returns: (allow-none): Newly allocated string with content type
 550  *     or %NULL. Free with g_free()
 551  *
 552  * Since: 2.18
 553  **/
 554 gchar *
 555 g_content_type_from_mime_type (const gchar *mime_type)
 556 {
 557   char *umime;
 558
 559   g_return_val_if_fail (mime_type != NULL, NULL);
 560
 561   G_LOCK (gio_xdgmime);
 562   /* mime type and content type are same on unixes */
 563   umime = g_strdup (xdg_mime_unalias_mime_type (mime_type));
 564   G_UNLOCK (gio_xdgmime);
 565
 566   return umime;
 567 }
 568
 569 /**
 570  * g_content_type_guess:
 571  * @filename: (allow-none): a string, or %NULL
 572  * @data: (allow-none) (array length=data_size): a stream of data, or %NULL
 573  * @data_size: the size of @data
 574  * @result_uncertain: (allow-none) (out): return location for the certainty
 575  *     of the result, or %NULL
 576  *
 577  * Guesses the content type based on example data. If the function is
 578  * uncertain, @result_uncertain will be set to %TRUE. Either @filename
 579  * or @data may be %NULL, in which case the guess will be based solely
 580  * on the other argument.
 581  *
 582  * Returns: a string indicating a guessed content type for the
 583  *     given data. Free with g_free()
 584  */
 585 gchar *
 586 g_content_type_guess (const gchar  *filename,
 587                       const guchar *data,
 588                       gsize         data_size,
 589                       gboolean     *result_uncertain)
 590 {
 591   char *basename;
 592   const char *name_mimetypes[10], *sniffed_mimetype;
 593   char *mimetype;
 594   int i;
 595   int n_name_mimetypes;
 596   int sniffed_prio;
 597
 598   sniffed_prio = 0;
 599   n_name_mimetypes = 0;
 600   sniffed_mimetype = XDG_MIME_TYPE_UNKNOWN;
 601
 602   if (result_uncertain)
 603     *result_uncertain = FALSE;
 604
 605   /* our test suite and potentially other code used -1 in the past, which is
 606    * not documented and not allowed; guard against that */
 607   g_return_val_if_fail (data_size != (gsize) -1, g_strdup (XDG_MIME_TYPE_UNKNOWN));
 608
 609   G_LOCK (gio_xdgmime);
 610
 611   if (filename)
 612     {
 613       i = strlen (filename);
 614       if (filename[i - 1] == '/')
 615         {
 616           name_mimetypes[0] = "inode/directory";
 617           name_mimetypes[1] = NULL;
 618           n_name_mimetypes = 1;
 619           if (result_uncertain)
 620             *result_uncertain = TRUE;
 621         }
 622       else
 623         {
 624           basename = g_path_get_basename (filename);
 625           n_name_mimetypes = xdg_mime_get_mime_types_from_file_name (basename, name_mimetypes, 10);
 626           g_free (basename);
 627         }
 628     }
 629
 630   /* Got an extension match, and no conflicts. This is it. */
 631   if (n_name_mimetypes == 1)
 632     {
 633       G_UNLOCK (gio_xdgmime);
 634       return g_strdup (name_mimetypes[0]);
 635     }
 636
 637   if (data)
 638     {
 639       sniffed_mimetype = xdg_mime_get_mime_type_for_data (data, data_size, &sniffed_prio);
 640       if (sniffed_mimetype == XDG_MIME_TYPE_UNKNOWN &&
 641           data &&
 642           looks_like_text (data, data_size))
 643         sniffed_mimetype = "text/plain";
 644
 645       /* For security reasons we don't ever want to sniff desktop files
 646        * where we know the filename and it doesn't have a .desktop extension.
 647        * This is because desktop files allow executing any application and
 648        * we don't want to make it possible to hide them looking like something
 649        * else.
 650        */
 651       if (filename != NULL &&
 652           strcmp (sniffed_mimetype, "application/x-desktop") == 0)
 653         sniffed_mimetype = "text/plain";
 654     }
 655
 656   if (n_name_mimetypes == 0)
 657     {
 658       if (sniffed_mimetype == XDG_MIME_TYPE_UNKNOWN &&
 659           result_uncertain)
 660         *result_uncertain = TRUE;
 661
 662       mimetype = g_strdup (sniffed_mimetype);
 663     }
 664   else
 665     {
 666       mimetype = NULL;
 667       if (sniffed_mimetype != XDG_MIME_TYPE_UNKNOWN)
 668         {
 669           if (sniffed_prio >= 80) /* High priority sniffing match, use that */
 670             mimetype = g_strdup (sniffed_mimetype);
 671           else
 672             {
 673               /* There are conflicts between the name matches and we
 674                * have a sniffed type, use that as a tie breaker.
 675                */
 676               for (i = 0; i < n_name_mimetypes; i++)
 677                 {
 678                   if ( xdg_mime_mime_type_subclass (name_mimetypes[i], sniffed_mimetype))
 679                     {
 680                       /* This nametype match is derived from (or the same as)
 681                        * the sniffed type). This is probably it.
 682                        */
 683                       mimetype = g_strdup (name_mimetypes[i]);
 684                       break;
 685                     }
 686                 }
 687             }
 688         }
 689
 690       if (mimetype == NULL)
 691         {
 692           /* Conflicts, and sniffed type was no help or not there.
 693            * Guess on the first one
 694            */
 695           mimetype = g_strdup (name_mimetypes[0]);
 696           if (result_uncertain)
 697             *result_uncertain = TRUE;
 698         }
 699     }
 700
 701   G_UNLOCK (gio_xdgmime);
 702
 703   return mimetype;
 704 }
 705
 706 static void
 707 enumerate_mimetypes_subdir (const char *dir,
 708                             const char *prefix,
 709                             GHashTable *mimetypes)
 710 {
 711   DIR *d;
 712   struct dirent *ent;
 713   char *mimetype;
 714
 715   d = opendir (dir);
 716   if (d)
 717     {
 718       while ((ent = readdir (d)) != NULL)
 719         {
 720           if (g_str_has_suffix (ent->d_name, ".xml"))
 721             {
 722               mimetype = g_strdup_printf ("%s/%.*s", prefix, (int) strlen (ent->d_name) - 4, ent->d_name);
 723               g_hash_table_replace (mimetypes, mimetype, NULL);
 724             }
 725         }
 726       closedir (d);
 727     }
 728 }
 729
 730 static void
 731 enumerate_mimetypes_dir (const char *dir,
 732                          GHashTable *mimetypes)
 733 {
 734   DIR *d;
 735   struct dirent *ent;
 736   char *mimedir;
 737   char *name;
 738
 739   mimedir = g_build_filename (dir, "mime", NULL);
 740
 741   d = opendir (mimedir);
 742   if (d)
 743     {
 744       while ((ent = readdir (d)) != NULL)
 745         {
 746           if (strcmp (ent->d_name, "packages") != 0)
 747             {
 748               name = g_build_filename (mimedir, ent->d_name, NULL);
 749               if (g_file_test (name, G_FILE_TEST_IS_DIR))
 750                 enumerate_mimetypes_subdir (name, ent->d_name, mimetypes);
 751               g_free (name);
 752             }
 753         }
 754       closedir (d);
 755     }
 756
 757   g_free (mimedir);
 758 }
 759
 760 /**
 761  * g_content_types_get_registered:
 762  *
 763  * Gets a list of strings containing all the registered content types
 764  * known to the system. The list and its data should be freed using
 765  * <programlisting>
 766  * g_list_free_full (list, g_free);
 767  * </programlisting>
 768  *
 769  * Returns: (element-type utf8) (transfer full): #GList of the registered content types
 770  */
 771 GList *
 772 g_content_types_get_registered (void)
 773 {
 774   const char * const* dirs;
 775   GHashTable *mimetypes;
 776   GHashTableIter iter;
 777   gpointer key;
 778   int i;
 779   GList *l;
 780
 781   mimetypes = g_hash_table_new_full (g_str_hash, g_str_equal, g_free, NULL);
 782
 783   enumerate_mimetypes_dir (g_get_user_data_dir (), mimetypes);
 784   dirs = g_get_system_data_dirs ();
 785
 786   for (i = 0; dirs[i] != NULL; i++)
 787     enumerate_mimetypes_dir (dirs[i], mimetypes);
 788
 789   l = NULL;
 790   g_hash_table_iter_init (&iter, mimetypes);
 791   while (g_hash_table_iter_next (&iter, &key, NULL))
 792     {
 793       l = g_list_prepend (l, key);
 794       g_hash_table_iter_steal (&iter);
 795     }
 796
 797   g_hash_table_destroy (mimetypes);
 798
 799   return l;
 800 }
 801
 802
 803 /* tree magic data */
 804 static GList *tree_matches = NULL;
 805 static gboolean need_reload = FALSE;
 806
 807 G_LOCK_DEFINE_STATIC (gio_treemagic);
 808
 809 typedef struct
 810 {
 811   gchar *path;
 812   GFileType type;
 813   guint match_case : 1;
 814   guint executable : 1;
 815   guint non_empty  : 1;
 816   guint on_disc    : 1;
 817   gchar *mimetype;
 818   GList *matches;
 819 } TreeMatchlet;
 820
 821 typedef struct
 822 {
 823   gchar *contenttype;
 824   gint priority;
 825   GList *matches;
 826 } TreeMatch;
 827
 828
 829 static void
 830 tree_matchlet_free (TreeMatchlet *matchlet)
 831 {
 832   g_list_free_full (matchlet->matches, (GDestroyNotify) tree_matchlet_free);
 833   g_free (matchlet->path);
 834   g_free (matchlet->mimetype);
 835   g_slice_free (TreeMatchlet, matchlet);
 836 }
 837
 838 static void
 839 tree_match_free (TreeMatch *match)
 840 {
 841   g_list_free_full (match->matches, (GDestroyNotify) tree_matchlet_free);
 842   g_free (match->contenttype);
 843   g_slice_free (TreeMatch, match);
 844 }
 845
 846 static TreeMatch *
 847 parse_header (gchar *line)
 848 {
 849   gint len;
 850   gchar *s;
 851   TreeMatch *match;
 852
 853   len = strlen (line);
 854
 855   if (line[0] != '[' || line[len - 1] != ']')
 856     return NULL;
 857
 858   line[len - 1] = 0;
 859   s = strchr (line, ':');
 860
 861   match = g_slice_new0 (TreeMatch);
 862   match->priority = atoi (line + 1);
 863   match->contenttype = g_strdup (s + 1);
 864
 865   return match;
 866 }
 867
 868 static TreeMatchlet *
 869 parse_match_line (gchar *line,
 870                   gint  *depth)
 871 {
 872   gchar *s, *p;
 873   TreeMatchlet *matchlet;
 874   gchar **parts;
 875   gint i;
 876
 877   matchlet = g_slice_new0 (TreeMatchlet);
 878
 879   if (line[0] == '>')
 880     {
 881       *depth = 0;
 882       s = line;
 883     }
 884   else
 885     {
 886       *depth = atoi (line);
 887       s = strchr (line, '>');
 888     }
 889   s += 2;
 890   p = strchr (s, '"');
 891   *p = 0;
 892
 893   matchlet->path = g_strdup (s);
 894   s = p + 1;
 895   parts = g_strsplit (s, ",", 0);
 896   if (strcmp (parts[0], "=file") == 0)
 897     matchlet->type = G_FILE_TYPE_REGULAR;
 898   else if (strcmp (parts[0], "=directory") == 0)
 899     matchlet->type = G_FILE_TYPE_DIRECTORY;
 900   else if (strcmp (parts[0], "=link") == 0)
 901     matchlet->type = G_FILE_TYPE_SYMBOLIC_LINK;
 902   else
 903     matchlet->type = G_FILE_TYPE_UNKNOWN;
 904   for (i = 1; parts[i]; i++)
 905     {
 906       if (strcmp (parts[i], "executable") == 0)
 907         matchlet->executable = 1;
 908       else if (strcmp (parts[i], "match-case") == 0)
 909         matchlet->match_case = 1;
 910       else if (strcmp (parts[i], "non-empty") == 0)
 911         matchlet->non_empty = 1;
 912       else if (strcmp (parts[i], "on-disc") == 0)
 913         matchlet->on_disc = 1;
 914       else
 915         matchlet->mimetype = g_strdup (parts[i]);
 916     }
 917
 918   g_strfreev (parts);
 919
 920   return matchlet;
 921 }
 922
 923 static gint
 924 cmp_match (gconstpointer a, gconstpointer b)
 925 {
 926   const TreeMatch *aa = (const TreeMatch *)a;
 927   const TreeMatch *bb = (const TreeMatch *)b;
 928
 929   return bb->priority - aa->priority;
 930 }
 931
 932 static void
 933 insert_match (TreeMatch *match)
 934 {
 935   tree_matches = g_list_insert_sorted (tree_matches, match, cmp_match);
 936 }
 937
 938 static void
 939 insert_matchlet (TreeMatch    *match,
 940                  TreeMatchlet *matchlet,
 941                  gint          depth)
 942 {
 943   if (depth == 0)
 944     match->matches = g_list_append (match->matches, matchlet);
 945   else
 946     {
 947       GList *last;
 948       TreeMatchlet *m;
 949
 950       last = g_list_last (match->matches);
 951       if (!last)
 952         {
 953           tree_matchlet_free (matchlet);
 954           g_warning ("can't insert tree matchlet at depth %d", depth);
 955           return;
 956         }
 957
 958       m = (TreeMatchlet *) last->data;
 959       while (--depth > 0)
 960         {
 961           last = g_list_last (m->matches);
 962           if (!last)
 963             {
 964               tree_matchlet_free (matchlet);
 965               g_warning ("can't insert tree matchlet at depth %d", depth);
 966               return;
 967             }
 968
 969           m = (TreeMatchlet *) last->data;
 970         }
 971       m->matches = g_list_append (m->matches, matchlet);
 972     }
 973 }
 974
 975 static void
 976 read_tree_magic_from_directory (const gchar *prefix)
 977 {
 978   gchar *filename;
 979   gchar *text;
 980   gsize len;
 981   gchar **lines;
 982   gint i;
 983   TreeMatch *match;
 984   TreeMatchlet *matchlet;
 985   gint depth;
 986
 987   filename = g_build_filename (prefix, "mime", "treemagic", NULL);
 988
 989   if (g_file_get_contents (filename, &text, &len, NULL))
 990     {
 991       if (strcmp (text, "MIME-TreeMagic") == 0)
 992         {
 993           lines = g_strsplit (text + strlen ("MIME-TreeMagic") + 2, "\n", 0);
 994           match = NULL;
 995           for (i = 0; lines[i] && lines[i][0]; i++)
 996             {
 997               if (lines[i][0] == '[')
 998                 {
 999                   match = parse_header (lines[i]);
1000                   insert_match (match);
1001                 }
1002               else
1003                 {
1004                   matchlet = parse_match_line (lines[i], &depth);
1005                   insert_matchlet (match, matchlet, depth);
1006                 }
1007             }
1008
1009           g_strfreev (lines);
1010         }
1011       else
1012         g_warning ("%s: header not found, skipping\n", filename);
1013
1014       g_free (text);
1015     }
1016
1017   g_free (filename);
1018 }
1019
1020
1021 static void
1022 xdg_mime_reload (void *user_data)
1023 {
1024   need_reload = TRUE;
1025 }
1026
1027 static void
1028 tree_magic_shutdown (void)
1029 {
1030   g_list_free_full (tree_matches, (GDestroyNotify) tree_match_free);
1031   tree_matches = NULL;
1032 }
1033
1034 static void
1035 tree_magic_init (void)
1036 {
1037   static gboolean initialized = FALSE;
1038   const gchar *dir;
1039   const gchar * const * dirs;
1040   int i;
1041
1042   if (!initialized)
1043     {
1044       initialized = TRUE;
1045
1046       xdg_mime_register_reload_callback (xdg_mime_reload, NULL, NULL);
1047       need_reload = TRUE;
1048     }
1049
1050   if (need_reload)
1051     {
1052       need_reload = FALSE;
1053
1054       tree_magic_shutdown ();
1055
1056       dir = g_get_user_data_dir ();
1057       read_tree_magic_from_directory (dir);
1058       dirs = g_get_system_data_dirs ();
1059       for (i = 0; dirs[i]; i++)
1060         read_tree_magic_from_directory (dirs[i]);
1061     }
1062 }
1063
1064 /* a filtering enumerator */
1065
1066 typedef struct
1067 {
1068   gchar *path;
1069   gint depth;
1070   gboolean ignore_case;
1071   gchar **components;
1072   gchar **case_components;
1073   GFileEnumerator **enumerators;
1074   GFile **children;
1075 } Enumerator;
1076
1077 static gboolean
1078 component_match (Enumerator  *e,
1079                  gint         depth,
1080                  const gchar *name)
1081 {
1082   gchar *case_folded, *key;
1083   gboolean found;
1084
1085   if (strcmp (name, e->components[depth]) == 0)
1086     return TRUE;
1087
1088   if (!e->ignore_case)
1089     return FALSE;
1090
1091   case_folded = g_utf8_casefold (name, -1);
1092   key = g_utf8_collate_key (case_folded, -1);
1093
1094   found = strcmp (key, e->case_components[depth]) == 0;
1095
1096   g_free (case_folded);
1097   g_free (key);
1098
1099   return found;
1100 }
1101
1102 static GFile *
1103 next_match_recurse (Enumerator *e,
1104                     gint        depth)
1105 {
1106   GFile *file;
1107   GFileInfo *info;
1108   const gchar *name;
1109
1110   while (TRUE)
1111     {
1112       if (e->enumerators[depth] == NULL)
1113         {
1114           if (depth > 0)
1115             {
1116               file = next_match_recurse (e, depth - 1);
1117               if (file)
1118                 {
1119                   e->children[depth] = file;
1120                   e->enumerators[depth] = g_file_enumerate_children (file,
1121                                                                      G_FILE_ATTRIBUTE_STANDARD_NAME,
1122                                                                      G_FILE_QUERY_INFO_NONE,
1123                                                                      NULL,
1124                                                                      NULL);
1125                 }
1126             }
1127           if (e->enumerators[depth] == NULL)
1128             return NULL;
1129         }
1130
1131       while ((info = g_file_enumerator_next_file (e->enumerators[depth], NULL, NULL)))
1132         {
1133           name = g_file_info_get_name (info);
1134           if (component_match (e, depth, name))
1135             {
1136               file = g_file_get_child (e->children[depth], name);
1137               g_object_unref (info);
1138               return file;
1139             }
1140           g_object_unref (info);
1141         }
1142
1143       g_object_unref (e->enumerators[depth]);
1144       e->enumerators[depth] = NULL;
1145       g_object_unref (e->children[depth]);
1146       e->children[depth] = NULL;
1147     }
1148 }
1149
1150 static GFile *
1151 enumerator_next (Enumerator *e)
1152 {
1153   return next_match_recurse (e, e->depth - 1);
1154 }
1155
1156 static Enumerator *
1157 enumerator_new (GFile      *root,
1158                 const char *path,
1159                 gboolean    ignore_case)
1160 {
1161   Enumerator *e;
1162   gint i;
1163   gchar *case_folded;
1164
1165   e = g_new0 (Enumerator, 1);
1166   e->path = g_strdup (path);
1167   e->ignore_case = ignore_case;
1168
1169   e->components = g_strsplit (e->path, G_DIR_SEPARATOR_S, -1);
1170   e->depth = g_strv_length (e->components);
1171   if (e->ignore_case)
1172     {
1173       e->case_components = g_new0 (char *, e->depth + 1);
1174       for (i = 0; e->components[i]; i++)
1175         {
1176           case_folded = g_utf8_casefold (e->components[i], -1);
1177           e->case_components[i] = g_utf8_collate_key (case_folded, -1);
1178           g_free (case_folded);
1179         }
1180     }
1181
1182   e->children = g_new0 (GFile *, e->depth);
1183   e->children[0] = g_object_ref (root);
1184   e->enumerators = g_new0 (GFileEnumerator *, e->depth);
1185   e->enumerators[0] = g_file_enumerate_children (root,
1186                                                  G_FILE_ATTRIBUTE_STANDARD_NAME,
1187                                                  G_FILE_QUERY_INFO_NONE,
1188                                                  NULL,
1189                                                  NULL);
1190
1191   return e;
1192 }
1193
1194 static void
1195 enumerator_free (Enumerator *e)
1196 {
1197   gint i;
1198
1199   for (i = 0; i < e->depth; i++)
1200     {
1201       if (e->enumerators[i])
1202         g_object_unref (e->enumerators[i]);
1203       if (e->children[i])
1204         g_object_unref (e->children[i]);
1205     }
1206
1207   g_free (e->enumerators);
1208   g_free (e->children);
1209   g_strfreev (e->components);
1210   if (e->case_components)
1211     g_strfreev (e->case_components);
1212   g_free (e->path);
1213   g_free (e);
1214 }
1215
1216 static gboolean
1217 matchlet_match (TreeMatchlet *matchlet,
1218                 GFile        *root)
1219 {
1220   GFile *file;
1221   GFileInfo *info;
1222   gboolean result;
1223   const gchar *attrs;
1224   Enumerator *e;
1225   GList *l;
1226
1227   e = enumerator_new (root, matchlet->path, !matchlet->match_case);
1228
1229   do
1230     {
1231       file = enumerator_next (e);
1232       if (!file)
1233         {
1234           enumerator_free (e);
1235           return FALSE;
1236         }
1237
1238       if (matchlet->mimetype)
1239         attrs = G_FILE_ATTRIBUTE_STANDARD_TYPE ","
1240                 G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE ","
1241                 G_FILE_ATTRIBUTE_STANDARD_CONTENT_TYPE;
1242       else
1243         attrs = G_FILE_ATTRIBUTE_STANDARD_TYPE ","
1244                 G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE;
1245       info = g_file_query_info (file,
1246                                 attrs,
1247                                 G_FILE_QUERY_INFO_NONE,
1248                                 NULL,
1249                                 NULL);
1250       if (info)
1251         {
1252           result = TRUE;
1253
1254           if (matchlet->type != G_FILE_TYPE_UNKNOWN &&
1255               g_file_info_get_file_type (info) != matchlet->type)
1256             result = FALSE;
1257
1258           if (matchlet->executable &&
1259               !g_file_info_get_attribute_boolean (info, G_FILE_ATTRIBUTE_ACCESS_CAN_EXECUTE))
1260             result = FALSE;
1261         }
1262       else
1263         result = FALSE;
1264
1265       if (result && matchlet->non_empty)
1266         {
1267           GFileEnumerator *child_enum;
1268           GFileInfo *child_info;
1269
1270           child_enum = g_file_enumerate_children (file,
1271                                                   G_FILE_ATTRIBUTE_STANDARD_NAME,
1272                                                   G_FILE_QUERY_INFO_NONE,
1273                                                   NULL,
1274                                                   NULL);
1275
1276           if (child_enum)
1277             {
1278               child_info = g_file_enumerator_next_file (child_enum, NULL, NULL);
1279               if (child_info)
1280                 g_object_unref (child_info);
1281               else
1282                 result = FALSE;
1283               g_object_unref (child_enum);
1284             }
1285           else
1286             result = FALSE;
1287         }
1288
1289       if (result && matchlet->mimetype)
1290         {
1291           if (strcmp (matchlet->mimetype, g_file_info_get_content_type (info)) != 0)
1292             result = FALSE;
1293         }
1294
1295       g_object_unref (info);
1296       g_object_unref (file);
1297     }
1298   while (!result);
1299
1300   enumerator_free (e);
1301
1302   if (!matchlet->matches)
1303     return TRUE;
1304
1305   for (l = matchlet->matches; l; l = l->next)
1306     {
1307       TreeMatchlet *submatchlet;
1308
1309       submatchlet = l->data;
1310       if (matchlet_match (submatchlet, root))
1311         return TRUE;
1312     }
1313
1314   return FALSE;
1315 }
1316
1317 static void
1318 match_match (TreeMatch    *match,
1319              GFile        *root,
1320              GPtrArray    *types)
1321 {
1322   GList *l;
1323
1324   for (l = match->matches; l; l = l->next)
1325     {
1326       TreeMatchlet *matchlet = l->data;
1327       if (matchlet_match (matchlet, root))
1328         {
1329           g_ptr_array_add (types, g_strdup (match->contenttype));
1330           break;
1331         }
1332     }
1333 }
1334
1335 /**
1336  * g_content_type_guess_for_tree:
1337  * @root: the root of the tree to guess a type for
1338  *
1339  * Tries to guess the type of the tree with root @root, by
1340  * looking at the files it contains. The result is an array
1341  * of content types, with the best guess coming first.
1342  *
1343  * The types returned all have the form x-content/foo, e.g.
1344  * x-content/audio-cdda (for audio CDs) or x-content/image-dcf
1345  * (for a camera memory card). See the <ulink url="http://www.freedesktop.org/wiki/Specifications/shared-mime-info-spec">shared-mime-info</ulink>
1346  * specification for more on x-content types.
1347  *
1348  * This function is useful in the implementation of
1349  * g_mount_guess_content_type().
1350  *
1351  * Returns: (transfer full) (array zero-terminated=1): an %NULL-terminated
1352  *     array of zero or more content types. Free with g_strfreev()
1353  *
1354  * Since: 2.18
1355  */
1356 gchar **
1357 g_content_type_guess_for_tree (GFile *root)
1358 {
1359   GPtrArray *types;
1360   GList *l;
1361
1362   types = g_ptr_array_new ();
1363
1364   G_LOCK (gio_treemagic);
1365
1366   tree_magic_init ();
1367   for (l = tree_matches; l; l = l->next)
1368     {
1369       TreeMatch *match = l->data;
1370       match_match (match, root, types);
1371     }
1372
1373   G_UNLOCK (gio_treemagic);
1374
1375   g_ptr_array_add (types, NULL);
1376
1377   return (gchar **)g_ptr_array_free (types, FALSE);
1378 }