G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { TrackerSparqlBuilder *preupdate, *metadata; goffset size; const gchar *graph; gchar *filename, *uri; GFile *file; preupdate = tracker_extract_info_get_preupdate_builder (info); metadata = tracker_extract_info_get_metadata_builder (info); graph = tracker_extract_info_get_graph (info); file = tracker_extract_info_get_file (info); filename = g_file_get_path (file); size = tracker_file_get_size (filename); if (size < 14) { /* Smaller than BMP header, can't be a real BMP file */ g_free (filename); return FALSE; } tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:Image"); tracker_sparql_builder_object (metadata, "nmm:Photo"); /* TODO: Add actual metadata extraction for BMP files */ return TRUE; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { TrackerSparqlBuilder *metadata; GFile *file; gchar *filename; DviContext *context; metadata = tracker_extract_info_get_metadata_builder (info); file = tracker_extract_info_get_file (info); filename = g_file_get_path (file); context = mdvi_init_context (filename); if (context == NULL) { g_warning ("Could not open dvi file '%s'\n", filename); g_free (filename); return FALSE; } tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:PaginatedTextDocument"); tracker_sparql_builder_predicate (metadata, "nfo:pageCount"); tracker_sparql_builder_object_int64 (metadata, context->npages); if (context->fileid) { tracker_sparql_builder_predicate (metadata, "nie:comment"); tracker_sparql_builder_object_unvalidated (metadata, context->fileid); } mdvi_destroy_context (context); return TRUE; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { TrackerSparqlBuilder *preupdate, *metadata; const gchar *mimetype; GFile *file; gchar *uri; preupdate = tracker_extract_info_get_preupdate_builder (info); metadata = tracker_extract_info_get_metadata_builder (info); mimetype = tracker_extract_info_get_mimetype (info); file = tracker_extract_info_get_file (info); uri = g_file_get_uri (file); #ifdef USING_UNZIPPSFILES if (strcmp (mimetype, "application/x-gzpostscript") == 0) { extract_ps_gz (uri, preupdate, metadata); } else #endif /* USING_UNZIPPSFILES */ { extract_ps (uri, preupdate, metadata); } g_free (uri); return TRUE; }
static gboolean get_file_metadata (TrackerExtractTask *task, TrackerExtractInfo **info_out) { TrackerExtractInfo *info; GFile *file; gchar *mime_used = NULL; gint items = 0; *info_out = NULL; file = g_file_new_for_uri (task->file); info = tracker_extract_info_new (file, task->mimetype, task->graph); g_object_unref (file); if (task->mimetype && *task->mimetype) { /* We know the mime */ mime_used = g_strdup (task->mimetype); } else { tracker_extract_info_unref (info); return FALSE; } /* Now we have sanity checked everything, actually get the * data we need from the extractors. */ if (mime_used) { if (task->cur_func) { TrackerSparqlBuilder *statements; g_debug ("Using %s...", g_module_name (task->cur_module)); (task->cur_func) (info); statements = tracker_extract_info_get_metadata_builder (info); items = tracker_sparql_builder_get_length (statements); if (items > 0) { tracker_sparql_builder_insert_close (statements); task->success = TRUE; } } g_free (mime_used); } g_debug ("Done (%d objects added)\n", items); if (items == 0) { tracker_extract_info_unref (info); info = NULL; } *info_out = info; return (items > 0); }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { TotemPlParser *pl; TrackerSparqlBuilder *metadata; PlaylistMetadata data; GFile *file; gchar *uri; pl = totem_pl_parser_new (); file = tracker_extract_info_get_file (info); uri = g_file_get_uri (file); metadata = tracker_extract_info_get_metadata_builder (info); data.track_counter = PLAYLIST_DEFAULT_NO_TRACKS; data.total_time = PLAYLIST_DEFAULT_DURATION; data.title = NULL; data.metadata = metadata; g_object_set (pl, "recurse", FALSE, "disable-unsafe", TRUE, NULL); g_signal_connect (G_OBJECT (pl), "playlist-started", G_CALLBACK (playlist_started), &data); g_signal_connect (G_OBJECT (pl), "entry-parsed", G_CALLBACK (entry_parsed), &data); tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nmm:Playlist"); tracker_sparql_builder_object (metadata, "nfo:MediaList"); if (totem_pl_parser_parse (pl, uri, FALSE) == TOTEM_PL_PARSER_RESULT_SUCCESS) { if (data.title != NULL) { g_message ("Playlist title:'%s'", data.title); tracker_sparql_builder_predicate (metadata, "nie:title"); tracker_sparql_builder_object_unvalidated (metadata, data.title); g_free (data.title); } else { g_message ("Playlist has no title, attempting to get one from filename"); tracker_guarantee_title_from_file (metadata, "nie:title", NULL, uri, NULL); } if (data.total_time > 0) { tracker_sparql_builder_predicate (metadata, "nfo:listDuration"); tracker_sparql_builder_object_int64 (metadata, data.total_time); } if (data.track_counter > 0) { tracker_sparql_builder_predicate (metadata, "nfo:entryCounter"); tracker_sparql_builder_object_int64 (metadata, data.track_counter); } } else { g_warning ("Playlist could not be parsed, no error given"); } g_object_unref (pl); g_free (uri); return TRUE; }
static inline OPFData * opf_data_new (TrackerExtractInfo *info) { OPFData *data = g_slice_new0 (OPFData); TrackerSparqlBuilder *builder; builder = tracker_extract_info_get_preupdate_builder (info); data->preupdate = g_object_ref (builder); builder = tracker_extract_info_get_metadata_builder (info); data->metadata = g_object_ref (builder); data->graph = g_strdup (tracker_extract_info_get_graph (info)); return data; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { TrackerSparqlBuilder *metadata; TrackerConfig *config; gchar *content; config = tracker_main_get_config (); metadata = tracker_extract_info_get_metadata_builder (info); content = get_file_content (tracker_extract_info_get_file (info), tracker_config_get_max_bytes (config)); tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:PlainTextDocument"); if (content) { tracker_sparql_builder_predicate (metadata, "nie:plainTextContent"); tracker_sparql_builder_object_unvalidated (metadata, content); g_free (content); } return TRUE; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info_) { /* NOTE: This function has to exist, tracker-extract checks * the symbole table for this function and if it doesn't * exist, the module is not loaded to be used as an extractor. */ /* File information */ GFile *file; GError *error = NULL; gchar *filename; OsinfoLoader *loader = NULL; OsinfoMedia *media; OsinfoDb *db; OsinfoOs *os; OsinfoOsVariantList *variants; /* Data input */ gboolean bootable; const gchar *id; const gchar *name; TrackerSparqlBuilder *metadata; metadata = tracker_extract_info_get_metadata_builder (info_); file = tracker_extract_info_get_file (info_); filename = g_file_get_path (file); media = osinfo_media_create_from_location (filename, NULL, &error); if (error != NULL) { if (error->code != OSINFO_MEDIA_ERROR_NOT_BOOTABLE) { g_message ("Could not extract iso info from '%s': %s", filename, error->message); g_free (filename); g_error_free (error); return FALSE; } bootable = FALSE; goto no_os; } else { bootable = TRUE; } g_free (filename); loader = osinfo_loader_new (); osinfo_loader_process_default_path (loader, &error); if (error != NULL) { g_message ("Error loading libosinfo OS data: %s", error->message); g_error_free (error); goto no_os; } g_warn_if_fail (media != NULL); g_warn_if_fail (loader != NULL); db = osinfo_loader_get_db (loader); osinfo_db_identify_media (db, media); os = osinfo_media_get_os (media); if (os == NULL) goto unknown_os; tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:FilesystemImage"); variants = osinfo_media_get_os_variants (media); if (osinfo_list_get_length (OSINFO_LIST (variants)) > 0) { OsinfoEntity *variant; /* FIXME: Assuming first variant from multivariant medias. */ variant = osinfo_list_get_nth (OSINFO_LIST (variants), 0); name = osinfo_os_variant_get_name (OSINFO_OS_VARIANT (variant)); } else { name = osinfo_product_get_name (OSINFO_PRODUCT (os)); } if (name != NULL) { tracker_sparql_builder_predicate (metadata, "nie:title"); tracker_sparql_builder_object_string (metadata, name); } if (osinfo_media_get_live (media)) { tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:OperatingSystem"); } if (osinfo_media_get_installer (media)) { tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "osinfo:Installer"); } tracker_sparql_builder_predicate (metadata, "nfo:isBootable"); tracker_sparql_builder_object_boolean (metadata, bootable); id = osinfo_entity_get_id (OSINFO_ENTITY (os)); if (id != NULL) { tracker_sparql_builder_predicate (metadata, "osinfo:id"); tracker_sparql_builder_object_string (metadata, id); } id = osinfo_entity_get_id (OSINFO_ENTITY (media)); if (id != NULL) { tracker_sparql_builder_predicate (metadata, "osinfo:mediaId"); tracker_sparql_builder_object_string (metadata, id); } g_object_unref (G_OBJECT (media)); g_object_unref (G_OBJECT (loader)); return TRUE; unknown_os: name = osinfo_media_get_volume_id (media); if (name != NULL) { gchar *stripped = g_strdup (name); g_strstrip (stripped); tracker_sparql_builder_predicate (metadata, "nie:title"); tracker_sparql_builder_object_string (metadata, stripped); g_free (stripped); } no_os: if (media != NULL) { g_object_unref (G_OBJECT (media)); } if (loader != NULL) { g_object_unref (G_OBJECT (loader)); } tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:FilesystemImage"); tracker_sparql_builder_predicate (metadata, "nfo:isBootable"); tracker_sparql_builder_object_boolean (metadata, bootable); return TRUE; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { GFile *file; TrackerSparqlBuilder *metadata; TrackerSparqlBuilder *preupdate; const gchar *graph; gchar *absolute_file_path; gchar *uri; AVFormatContext *format = NULL; AVStream *audio_stream = NULL; AVStream *video_stream = NULL; int audio_stream_index; int video_stream_index; AVDictionaryEntry *tag = NULL; const char *title = NULL; av_register_all (); file = tracker_extract_info_get_file (info); metadata = tracker_extract_info_get_metadata_builder (info); preupdate = tracker_extract_info_get_preupdate_builder (info); graph = tracker_extract_info_get_graph (info); uri = g_file_get_uri (file); absolute_file_path = g_file_get_path (file); if (avformat_open_input (&format, absolute_file_path, NULL, NULL)) { g_free (absolute_file_path); g_free (uri); return FALSE; } g_free (absolute_file_path); avformat_find_stream_info (format, NULL); audio_stream_index = av_find_best_stream (format, AVMEDIA_TYPE_AUDIO, -1, -1, NULL, 0); if (audio_stream_index >= 0) { audio_stream = format->streams[audio_stream_index]; } video_stream_index = av_find_best_stream (format, AVMEDIA_TYPE_VIDEO, -1, -1, NULL, 0); if (video_stream_index >= 0) { video_stream = format->streams[video_stream_index]; } if (!audio_stream && !video_stream) { avformat_free_context (format); g_free (uri); return FALSE; } if (audio_stream) { if (audio_stream->codec->sample_rate > 0) { set_value_int64 (metadata, "nfo:sampleRate", audio_stream->codec->sample_rate); } if (audio_stream->codec->channels > 0) { set_value_int64 (metadata, "nfo:channels", audio_stream->codec->channels); } } if (video_stream) { tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nmm:Video"); if (video_stream->codec->width > 0 && video_stream->codec->height > 0) { set_value_int64 (metadata, "nfo:width", video_stream->codec->width); set_value_int64 (metadata, "nfo:height", video_stream->codec->height); } if (video_stream->avg_frame_rate.num > 0) { gdouble frame_rate = (gdouble) video_stream->avg_frame_rate.num / video_stream->avg_frame_rate.den; set_value_double (metadata, "nfo:frameRate", frame_rate); } if (video_stream->duration > 0) { gint64 duration = av_rescale(video_stream->duration, video_stream->time_base.num, video_stream->time_base.den); set_value_int64 (metadata, "nfo:duration", duration); } if (video_stream->sample_aspect_ratio.num > 0) { gdouble aspect_ratio = (gdouble) video_stream->sample_aspect_ratio.num / video_stream->sample_aspect_ratio.den; set_value_double (metadata, "nfo:aspectRatio", aspect_ratio); } if (video_stream->nb_frames > 0) { set_value_int64 (metadata, "nfo:frameCount", video_stream->nb_frames); } if ((tag = av_dict_get (format->metadata, "synopsis", NULL, 0))) { set_value_string (metadata, "nmm:synopsis", tag->value); } if ((tag = av_dict_get (format->metadata, "episode_sort", NULL, 0))) { set_value_int64 (metadata, "nmm:episodeNumber", atoi(tag->value)); } if ((tag = av_dict_get (format->metadata, "season_number", NULL, 0))) { set_value_int64 (metadata, "nmm:season", atoi(tag->value)); } } else if (audio_stream) { const char *album_title = NULL; const char *album_artist = NULL; gchar *album_artist_uri = NULL; gchar *performer_uri = NULL; tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nmm:MusicPiece"); tracker_sparql_builder_object (metadata, "nfo:Audio"); if (audio_stream->duration > 0) { gint64 duration = av_rescale(audio_stream->duration, audio_stream->time_base.num, audio_stream->time_base.den); set_value_int64 (metadata, "nfo:duration", duration); } if ((tag = find_tag (format, audio_stream, "track"))) { int track = atoi(tag->value); if (track > 0) { set_value_int64 (metadata, "nmm:trackNumber", track); } } if ((tag = find_tag (format, audio_stream, "album"))) { album_title = tag->value; } if (album_title && (tag = find_tag (format, audio_stream, "album_artist"))) { album_artist_uri = create_artist (preupdate, graph, tag->value); album_artist = tag->value; } if ((tag = find_tag (format, audio_stream, "artist"))) { performer_uri = create_artist (preupdate, graph, tag->value); if (!album_artist) { album_artist = tag->value; } } if (!performer_uri && (tag = find_tag (format, audio_stream, "performer"))) { performer_uri = create_artist (preupdate, graph, tag->value); if (!album_artist) { album_artist = tag->value; } } if (performer_uri) { set_value_iri (metadata, "nmm:performer", performer_uri); } else if (album_artist_uri) { set_value_iri (metadata, "nmm:performer", album_artist_uri); } if ((tag = find_tag (format, audio_stream, "composer"))) { gchar *composer_uri = create_artist (preupdate, graph, tag->value); set_value_iri (metadata, "nmm:composer", composer_uri); g_free(composer_uri); } if (album_title) { int disc = 1; gchar *disc_uri; gchar *album_uri = tracker_sparql_escape_uri_printf ("urn:album:%s", album_title); open_insert (preupdate, graph, album_uri, "nmm:MusicAlbum"); set_value_string (preupdate, "nmm:albumTitle", album_title); if (album_artist_uri) { set_value_iri (preupdate, "nmm:albumArtist", album_artist_uri); } else if (performer_uri) { set_value_iri (preupdate, "nmm:albumArtist", performer_uri); } close_insert (preupdate, graph); if ((tag = find_tag (format, audio_stream, "disc"))) { disc = atoi (tag->value); } disc_uri = tracker_sparql_escape_uri_printf ("urn:album-disc:%s:Disc%d", album_title, disc); delete_value (preupdate, disc_uri, "nmm:setNumber", "unknown"); delete_value (preupdate, disc_uri, "nmm:albumDiscAlbum", "unknown"); open_insert (preupdate, graph, disc_uri, "nmm:MusicAlbumDisc"); set_value_int64 (preupdate, "nmm:setNumber", disc); set_value_iri (preupdate, "nmm:albumDiscAlbum", album_uri); close_insert (preupdate, graph); set_value_iri (metadata, "nmm:musicAlbumDisc", disc_uri); set_value_iri (metadata, "nmm:musicAlbum", album_uri); g_free (disc_uri); g_free (album_uri); } #ifdef HAVE_LIBMEDIAART if (album_artist || album_title) { MediaArtProcess *media_art_process; GError *error = NULL; gboolean success; media_art_process = tracker_extract_info_get_media_art_process (info); success = media_art_process_file (media_art_process, MEDIA_ART_ALBUM, MEDIA_ART_PROCESS_FLAGS_NONE, file, album_artist, album_title, &error); if (!success || error) { g_warning ("Could not process media art for '%s', %s", uri, error ? error->message : "No error given"); g_clear_error (&error); } } #endif g_free(performer_uri); } if (format->bit_rate > 0) { set_value_int64 (metadata, "nfo:averageBitrate", format->bit_rate); } if ((tag = av_dict_get (format->metadata, "comment", NULL, 0))) { set_value_string (metadata, "nie:comment", tag->value); } if ((tag = av_dict_get (format->metadata, "copyright", NULL, 0))) { set_value_string (metadata, "nie:copyright", tag->value); } if ((tag = av_dict_get (format->metadata, "creation_time", NULL, 0))) { gchar *content_created = tracker_date_guess (tag->value); if (content_created) { set_value_string (metadata, "nie:contentCreated", content_created); g_free (content_created); } } if ((tag = av_dict_get (format->metadata, "description", NULL, 0))) { set_value_string (metadata, "nie:description", tag->value); } if ((tag = av_dict_get (format->metadata, "genre", NULL, 0))) { set_value_string (metadata, "nfo:genre", tag->value); } if ((tag = av_dict_get (format->metadata, "language", NULL, 0))) { set_value_string (metadata, "nfo:language", tag->value); } if ((tag = av_dict_get (format->metadata, "title", NULL, 0))) { title = tag->value; } tracker_guarantee_title_from_file (metadata, "nie:title", title, uri, NULL); g_free (uri); avformat_free_context (format); return TRUE; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { struct jpeg_decompress_struct cinfo; struct tej_error_mgr tejerr; struct jpeg_marker_struct *marker; TrackerSparqlBuilder *preupdate, *metadata; TrackerXmpData *xd = NULL; TrackerExifData *ed = NULL; TrackerIptcData *id = NULL; MergeData md = { 0 }; GFile *file; FILE *f; goffset size; gchar *filename, *uri; gchar *comment = NULL; const gchar *dlna_profile, *dlna_mimetype, *graph; GPtrArray *keywords; gboolean success = TRUE; GString *where; guint i; metadata = tracker_extract_info_get_metadata_builder (info); preupdate = tracker_extract_info_get_preupdate_builder (info); graph = tracker_extract_info_get_graph (info); file = tracker_extract_info_get_file (info); filename = g_file_get_path (file); size = tracker_file_get_size (filename); if (size < 18) { g_free (filename); return FALSE; } f = tracker_file_open (filename); g_free (filename); if (!f) { return FALSE; } uri = g_file_get_uri (file); tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:Image"); tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nmm:Photo"); cinfo.err = jpeg_std_error (&tejerr.jpeg); tejerr.jpeg.error_exit = extract_jpeg_error_exit; if (setjmp (tejerr.setjmp_buffer)) { success = FALSE; goto fail; } jpeg_create_decompress (&cinfo); jpeg_save_markers (&cinfo, JPEG_COM, 0xFFFF); jpeg_save_markers (&cinfo, JPEG_APP0 + 1, 0xFFFF); jpeg_save_markers (&cinfo, JPEG_APP0 + 13, 0xFFFF); jpeg_stdio_src (&cinfo, f); jpeg_read_header (&cinfo, TRUE); /* FIXME? It is possible that there are markers after SOS, * but there shouldn't be. Should we decompress the whole file? * * jpeg_start_decompress(&cinfo); * jpeg_finish_decompress(&cinfo); * * jpeg_calc_output_dimensions(&cinfo); */ marker = (struct jpeg_marker_struct *) &cinfo.marker_list; while (marker) { gchar *str; gsize len; #ifdef HAVE_LIBIPTCDATA gsize offset; guint sublen; #endif /* HAVE_LIBIPTCDATA */ switch (marker->marker) { case JPEG_COM: g_free (comment); comment = g_strndup ((gchar*) marker->data, marker->data_length); break; case JPEG_APP0 + 1: str = (gchar*) marker->data; len = marker->data_length; #ifdef HAVE_LIBEXIF if (strncmp (EXIF_NAMESPACE, str, EXIF_NAMESPACE_LENGTH) == 0) { ed = tracker_exif_new ((guchar *) marker->data, len, uri); } #endif /* HAVE_LIBEXIF */ #ifdef HAVE_EXEMPI if (strncmp (XMP_NAMESPACE, str, XMP_NAMESPACE_LENGTH) == 0) { xd = tracker_xmp_new (str + XMP_NAMESPACE_LENGTH, len - XMP_NAMESPACE_LENGTH, uri); } #endif /* HAVE_EXEMPI */ break; case JPEG_APP0 + 13: str = (gchar*) marker->data; len = marker->data_length; #ifdef HAVE_LIBIPTCDATA if (len > 0 && strncmp (PS3_NAMESPACE, str, PS3_NAMESPACE_LENGTH) == 0) { offset = iptc_jpeg_ps3_find_iptc (str, len, &sublen); if (offset > 0 && sublen > 0) { id = tracker_iptc_new (str + offset, sublen, uri); } } #endif /* HAVE_LIBIPTCDATA */ break; default: marker = marker->next; continue; } marker = marker->next; } if (!ed) { ed = g_new0 (TrackerExifData, 1); } if (!xd) { xd = g_new0 (TrackerXmpData, 1); } if (!id) { id = g_new0 (TrackerIptcData, 1); } md.title = tracker_coalesce_strip (4, xd->title, ed->document_name, xd->title2, xd->pdf_title); md.orientation = tracker_coalesce_strip (3, xd->orientation, ed->orientation, id->image_orientation); md.copyright = tracker_coalesce_strip (4, xd->copyright, xd->rights, ed->copyright, id->copyright_notice); md.white_balance = tracker_coalesce_strip (2, xd->white_balance, ed->white_balance); md.fnumber = tracker_coalesce_strip (2, xd->fnumber, ed->fnumber); md.flash = tracker_coalesce_strip (2, xd->flash, ed->flash); md.focal_length = tracker_coalesce_strip (2, xd->focal_length, ed->focal_length); md.artist = tracker_coalesce_strip (3, xd->artist, ed->artist, xd->contributor); md.exposure_time = tracker_coalesce_strip (2, xd->exposure_time, ed->exposure_time); md.iso_speed_ratings = tracker_coalesce_strip (2, xd->iso_speed_ratings, ed->iso_speed_ratings); md.date = tracker_coalesce_strip (5, xd->date, xd->time_original, ed->time, id->date_created, ed->time_original); md.description = tracker_coalesce_strip (2, xd->description, ed->description); md.metering_mode = tracker_coalesce_strip (2, xd->metering_mode, ed->metering_mode); md.city = tracker_coalesce_strip (2, xd->city, id->city); md.state = tracker_coalesce_strip (2, xd->state, id->state); md.address = tracker_coalesce_strip (2, xd->address, id->sublocation); md.country = tracker_coalesce_strip (2, xd->country, id->country_name); /* FIXME We are not handling the altitude ref here for xmp */ md.gps_altitude = tracker_coalesce_strip (2, xd->gps_altitude, ed->gps_altitude); md.gps_latitude = tracker_coalesce_strip (2, xd->gps_latitude, ed->gps_latitude); md.gps_longitude = tracker_coalesce_strip (2, xd->gps_longitude, ed->gps_longitude); md.gps_direction = tracker_coalesce_strip (2, xd->gps_direction, ed->gps_direction); md.creator = tracker_coalesce_strip (3, xd->creator, id->byline, id->credit); md.comment = tracker_coalesce_strip (2, comment, ed->user_comment); md.make = tracker_coalesce_strip (2, xd->make, ed->make); md.model = tracker_coalesce_strip (2, xd->model, ed->model); /* Prioritize on native dimention in all cases */ tracker_sparql_builder_predicate (metadata, "nfo:width"); tracker_sparql_builder_object_int64 (metadata, cinfo.image_width); /* TODO: add ontology and store ed->software */ tracker_sparql_builder_predicate (metadata, "nfo:height"); tracker_sparql_builder_object_int64 (metadata, cinfo.image_height); if (guess_dlna_profile (cinfo.image_width, cinfo.image_height, &dlna_profile, &dlna_mimetype)) { tracker_sparql_builder_predicate (metadata, "nmm:dlnaProfile"); tracker_sparql_builder_object_string (metadata, dlna_profile); tracker_sparql_builder_predicate (metadata, "nmm:dlnaMime"); tracker_sparql_builder_object_string (metadata, dlna_mimetype); } if (id->contact) { gchar *uri = tracker_sparql_escape_uri_printf ("urn:contact:%s", id->contact); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nco:Contact"); tracker_sparql_builder_predicate (preupdate, "nco:fullname"); tracker_sparql_builder_object_unvalidated (preupdate, id->contact); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nco:representative"); tracker_sparql_builder_object_iri (metadata, uri); g_free (uri); } keywords = g_ptr_array_new (); if (xd->keywords) { tracker_keywords_parse (keywords, xd->keywords); } if (xd->pdf_keywords) { tracker_keywords_parse (keywords, xd->pdf_keywords); } if (xd->subject) { tracker_keywords_parse (keywords, xd->subject); } if (xd->publisher) { gchar *uri = tracker_sparql_escape_uri_printf ("urn:contact:%s", xd->publisher); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nco:Contact"); tracker_sparql_builder_predicate (preupdate, "nco:fullname"); tracker_sparql_builder_object_unvalidated (preupdate, xd->publisher); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nco:publisher"); tracker_sparql_builder_object_iri (metadata, uri); g_free (uri); } if (xd->type) { tracker_sparql_builder_predicate (metadata, "dc:type"); tracker_sparql_builder_object_unvalidated (metadata, xd->type); } if (xd->rating) { tracker_sparql_builder_predicate (metadata, "nao:numericRating"); tracker_sparql_builder_object_unvalidated (metadata, xd->rating); } if (xd->format) { tracker_sparql_builder_predicate (metadata, "dc:format"); tracker_sparql_builder_object_unvalidated (metadata, xd->format); } if (xd->identifier) { tracker_sparql_builder_predicate (metadata, "dc:identifier"); tracker_sparql_builder_object_unvalidated (metadata, xd->identifier); } if (xd->source) { tracker_sparql_builder_predicate (metadata, "dc:source"); tracker_sparql_builder_object_unvalidated (metadata, xd->source); } if (xd->language) { tracker_sparql_builder_predicate (metadata, "dc:language"); tracker_sparql_builder_object_unvalidated (metadata, xd->language); } if (xd->relation) { tracker_sparql_builder_predicate (metadata, "dc:relation"); tracker_sparql_builder_object_unvalidated (metadata, xd->relation); } if (xd->coverage) { tracker_sparql_builder_predicate (metadata, "dc:coverage"); tracker_sparql_builder_object_unvalidated (metadata, xd->coverage); } if (xd->license) { tracker_sparql_builder_predicate (metadata, "nie:license"); tracker_sparql_builder_object_unvalidated (metadata, xd->license); } if (xd->regions) { tracker_xmp_apply_regions (preupdate, metadata, graph, xd); } if (id->keywords) { tracker_keywords_parse (keywords, id->keywords); } where = g_string_new (""); for (i = 0; i < keywords->len; i++) { gchar *p, *escaped, *var; p = g_ptr_array_index (keywords, i); escaped = tracker_sparql_escape_string (p); var = g_strdup_printf ("tag%d", i + 1); /* ensure tag with specified label exists */ tracker_sparql_builder_append (preupdate, "INSERT { "); if (graph) { tracker_sparql_builder_append (preupdate, "GRAPH <"); tracker_sparql_builder_append (preupdate, graph); tracker_sparql_builder_append (preupdate, "> { "); } tracker_sparql_builder_append (preupdate, "_:tag a nao:Tag ; nao:prefLabel \""); tracker_sparql_builder_append (preupdate, escaped); tracker_sparql_builder_append (preupdate, "\""); if (graph) { tracker_sparql_builder_append (preupdate, " } "); } tracker_sparql_builder_append (preupdate, " }\n"); tracker_sparql_builder_append (preupdate, "WHERE { FILTER (NOT EXISTS { " "?tag a nao:Tag ; nao:prefLabel \""); tracker_sparql_builder_append (preupdate, escaped); tracker_sparql_builder_append (preupdate, "\" }) }\n"); /* associate file with tag */ tracker_sparql_builder_predicate (metadata, "nao:hasTag"); tracker_sparql_builder_object_variable (metadata, var); g_string_append_printf (where, "?%s a nao:Tag ; nao:prefLabel \"%s\" .\n", var, escaped); g_free (var); g_free (escaped); g_free (p); } g_ptr_array_free (keywords, TRUE); tracker_extract_info_set_where_clause (info, where->str); g_string_free (where, TRUE); if (md.make || md.model) { gchar *equip_uri; equip_uri = tracker_sparql_escape_uri_printf ("urn:equipment:%s:%s:", md.make ? md.make : "", md.model ? md.model : ""); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, equip_uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nfo:Equipment"); if (md.make) { tracker_sparql_builder_predicate (preupdate, "nfo:manufacturer"); tracker_sparql_builder_object_unvalidated (preupdate, md.make); } if (md.model) { tracker_sparql_builder_predicate (preupdate, "nfo:model"); tracker_sparql_builder_object_unvalidated (preupdate, md.model); } if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nfo:equipment"); tracker_sparql_builder_object_iri (metadata, equip_uri); g_free (equip_uri); } tracker_guarantee_title_from_file (metadata, "nie:title", md.title, uri, NULL); if (md.orientation) { tracker_sparql_builder_predicate (metadata, "nfo:orientation"); tracker_sparql_builder_object (metadata, md.orientation); } if (md.copyright) { tracker_sparql_builder_predicate (metadata, "nie:copyright"); tracker_sparql_builder_object_unvalidated (metadata, md.copyright); } if (md.white_balance) { tracker_sparql_builder_predicate (metadata, "nmm:whiteBalance"); tracker_sparql_builder_object (metadata, md.white_balance); } if (md.fnumber) { gdouble value; value = g_strtod (md.fnumber, NULL); tracker_sparql_builder_predicate (metadata, "nmm:fnumber"); tracker_sparql_builder_object_double (metadata, value); } if (md.flash) { tracker_sparql_builder_predicate (metadata, "nmm:flash"); tracker_sparql_builder_object (metadata, md.flash); } if (md.focal_length) { gdouble value; value = g_strtod (md.focal_length, NULL); tracker_sparql_builder_predicate (metadata, "nmm:focalLength"); tracker_sparql_builder_object_double (metadata, value); } if (md.artist) { gchar *uri = tracker_sparql_escape_uri_printf ("urn:contact:%s", md.artist); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nco:Contact"); tracker_sparql_builder_predicate (preupdate, "nco:fullname"); tracker_sparql_builder_object_unvalidated (preupdate, md.artist); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nco:contributor"); tracker_sparql_builder_object_iri (metadata, uri); g_free (uri); } if (md.exposure_time) { gdouble value; value = g_strtod (md.exposure_time, NULL); tracker_sparql_builder_predicate (metadata, "nmm:exposureTime"); tracker_sparql_builder_object_double (metadata, value); } if (md.iso_speed_ratings) { gdouble value; value = g_strtod (md.iso_speed_ratings, NULL); tracker_sparql_builder_predicate (metadata, "nmm:isoSpeed"); tracker_sparql_builder_object_double (metadata, value); } tracker_guarantee_date_from_file_mtime (metadata, "nie:contentCreated", md.date, uri); if (md.description) { tracker_sparql_builder_predicate (metadata, "nie:description"); tracker_sparql_builder_object_unvalidated (metadata, md.description); } if (md.metering_mode) { tracker_sparql_builder_predicate (metadata, "nmm:meteringMode"); tracker_sparql_builder_object (metadata, md.metering_mode); } if (md.creator) { gchar *uri = tracker_sparql_escape_uri_printf ("urn:contact:%s", md.creator); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nco:Contact"); tracker_sparql_builder_predicate (preupdate, "nco:fullname"); tracker_sparql_builder_object_unvalidated (preupdate, md.creator); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); /* NOTE: We only have affiliation with * nco:PersonContact and we are using * nco:Contact here. */ /* if (id->byline_title) { */ /* tracker_sparql_builder_insert_open (preupdate, NULL); */ /* tracker_sparql_builder_subject (preupdate, "_:affiliation_by_line"); */ /* tracker_sparql_builder_predicate (preupdate, "a"); */ /* tracker_sparql_builder_object (preupdate, "nco:Affiliation"); */ /* tracker_sparql_builder_predicate (preupdate, "nco:title"); */ /* tracker_sparql_builder_object_unvalidated (preupdate, id->byline_title); */ /* tracker_sparql_builder_insert_close (preupdate); */ /* tracker_sparql_builder_predicate (preupdate, "a"); */ /* tracker_sparql_builder_object (preupdate, "nco:Contact"); */ /* tracker_sparql_builder_predicate (preupdate, "nco:hasAffiliation"); */ /* tracker_sparql_builder_object (preupdate, "_:affiliation_by_line"); */ /* } */ tracker_sparql_builder_predicate (metadata, "nco:creator"); tracker_sparql_builder_object_iri (metadata, uri); g_free (uri); } if (md.comment) { tracker_sparql_builder_predicate (metadata, "nie:comment"); tracker_sparql_builder_object_unvalidated (metadata, md.comment); } if (md.address || md.state || md.country || md.city || md.gps_altitude || md.gps_latitude || md.gps_longitude) { tracker_sparql_builder_predicate (metadata, "slo:location"); tracker_sparql_builder_object_blank_open (metadata); /* GeoPoint */ tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "slo:GeoLocation"); if (md.address || md.state || md.country || md.city) { gchar *addruri; addruri = tracker_sparql_get_uuid_urn (); tracker_sparql_builder_predicate (metadata, "slo:postalAddress"); tracker_sparql_builder_object_iri (metadata, addruri); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, addruri); g_free (addruri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nco:PostalAddress"); if (md.address) { tracker_sparql_builder_predicate (preupdate, "nco:streetAddress"); tracker_sparql_builder_object_unvalidated (preupdate, md.address); } if (md.state) { tracker_sparql_builder_predicate (preupdate, "nco:region"); tracker_sparql_builder_object_unvalidated (preupdate, md.state); } if (md.city) { tracker_sparql_builder_predicate (preupdate, "nco:locality"); tracker_sparql_builder_object_unvalidated (preupdate, md.city); } if (md.country) { tracker_sparql_builder_predicate (preupdate, "nco:country"); tracker_sparql_builder_object_unvalidated (preupdate, md.country); } if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); } if (md.gps_altitude) { tracker_sparql_builder_predicate (metadata, "slo:altitude"); tracker_sparql_builder_object_unvalidated (metadata, md.gps_altitude); } if (md.gps_latitude) { tracker_sparql_builder_predicate (metadata, "slo:latitude"); tracker_sparql_builder_object_unvalidated (metadata, md.gps_latitude); } if (md.gps_longitude) { tracker_sparql_builder_predicate (metadata, "slo:longitude"); tracker_sparql_builder_object_unvalidated (metadata, md.gps_longitude); } tracker_sparql_builder_object_blank_close (metadata); /* GeoLocation */ } if (md.gps_direction) { tracker_sparql_builder_predicate (metadata, "nfo:heading"); tracker_sparql_builder_object_unvalidated (metadata, md.gps_direction); } if (cinfo.density_unit != 0 || ed->x_resolution) { gdouble value; if (cinfo.density_unit == 0) { if (ed->resolution_unit != 3) value = g_strtod (ed->x_resolution, NULL); else value = g_strtod (ed->x_resolution, NULL) * CM_TO_INCH; } else { if (cinfo.density_unit == 1) value = cinfo.X_density; else value = cinfo.X_density * CM_TO_INCH; } tracker_sparql_builder_predicate (metadata, "nfo:horizontalResolution"); tracker_sparql_builder_object_double (metadata, value); } if (cinfo.density_unit != 0 || ed->y_resolution) { gdouble value; if (cinfo.density_unit == 0) { if (ed->resolution_unit != 3) value = g_strtod (ed->y_resolution, NULL); else value = g_strtod (ed->y_resolution, NULL) * CM_TO_INCH; } else { if (cinfo.density_unit == 1) value = cinfo.Y_density; else value = cinfo.Y_density * CM_TO_INCH; } tracker_sparql_builder_predicate (metadata, "nfo:verticalResolution"); tracker_sparql_builder_object_double (metadata, value); } jpeg_destroy_decompress (&cinfo); tracker_exif_free (ed); tracker_xmp_free (xd); tracker_iptc_free (id); g_free (comment); fail: tracker_file_close (f, FALSE); g_free (uri); return success; }
void tracker_extract_get_metadata_by_cmdline (TrackerExtract *object, const gchar *uri, const gchar *mime) { GError *error = NULL; TrackerExtractPrivate *priv; TrackerExtractTask *task; TrackerExtractInfo *info; gboolean no_modules = TRUE; priv = TRACKER_EXTRACT_GET_PRIVATE (object); priv->disable_summary_on_finalize = TRUE; g_return_if_fail (uri != NULL); task = extract_task_new (object, uri, mime, NULL, NULL, NULL, &error); if (error) { g_printerr ("Extraction failed, %s\n", error->message); g_error_free (error); return; } task->mimetype_handlers = tracker_extract_module_manager_get_mimetype_handlers (task->mimetype); task->cur_module = tracker_mimetype_info_get_module (task->mimetype_handlers, &task->cur_func, NULL); while (task->cur_module && task->cur_func) { if (!filter_module (object, task->cur_module) && get_file_metadata (task, &info)) { const gchar *preupdate_str, *postupdate_str, *statements_str, *where; TrackerSparqlBuilder *builder; no_modules = FALSE; preupdate_str = statements_str = postupdate_str = NULL; builder = tracker_extract_info_get_metadata_builder (info); if (tracker_sparql_builder_get_length (builder) > 0) { statements_str = tracker_sparql_builder_get_result (builder); } builder = tracker_extract_info_get_preupdate_builder (info); if (tracker_sparql_builder_get_length (builder) > 0) { preupdate_str = tracker_sparql_builder_get_result (builder); } builder = tracker_extract_info_get_postupdate_builder (info); if (tracker_sparql_builder_get_length (builder) > 0) { postupdate_str = tracker_sparql_builder_get_result (builder); } where = tracker_extract_info_get_where_clause (info); g_print ("\n"); g_print ("SPARQL pre-update:\n--\n%s--\n\n", preupdate_str ? preupdate_str : ""); g_print ("SPARQL item:\n--\n%s--\n\n", statements_str ? statements_str : ""); g_print ("SPARQL where clause:\n--\n%s--\n\n", where ? where : ""); g_print ("SPARQL post-update:\n--\n%s--\n\n", postupdate_str ? postupdate_str : ""); tracker_extract_info_unref (info); break; } else { if (!tracker_mimetype_info_iter_next (task->mimetype_handlers)) { break; } task->cur_module = tracker_mimetype_info_get_module (task->mimetype_handlers, &task->cur_func, NULL); } } if (no_modules) { g_print ("No modules found to handle metadata extraction\n\n"); } extract_task_free (task); }
static gboolean get_file_metadata (TrackerExtractTask *task, TrackerExtractInfo **info_out) { TrackerExtractInfo *info; GFile *file; gchar *mime_used = NULL; #ifdef HAVE_LIBSTREAMANALYZER gchar *content_type = NULL; #endif gint items = 0; g_debug ("Extracting..."); *info_out = NULL; file = g_file_new_for_uri (task->file); info = tracker_extract_info_new (file, task->mimetype, task->graph); g_object_unref (file); #ifdef HAVE_LIBSTREAMANALYZER /* FIXME: This entire section is completely broken, * it doesn't even build these days. It should be removed or fixed. * -mr (05/09/11) */ if (!priv->force_internal_extractors) { g_debug (" Using libstreamanalyzer..."); tracker_topanalyzer_extract (task->file, statements, &content_type); if (tracker_sparql_builder_get_length (statements) > 0) { g_free (content_type); tracker_sparql_builder_insert_close (statements); *info_out = info; return TRUE; } } else { g_debug (" Using internal extractors ONLY..."); } #endif /* HAVE_LIBSTREAMANALYZER */ if (task->mimetype && *task->mimetype) { /* We know the mime */ mime_used = g_strdup (task->mimetype); } #ifdef HAVE_LIBSTREAMANALYZER else if (content_type && *content_type) { /* We know the mime from LSA */ mime_used = content_type; g_strstrip (mime_used); } #endif /* HAVE_LIBSTREAMANALYZER */ else { tracker_extract_info_unref (info); return FALSE; } /* Now we have sanity checked everything, actually get the * data we need from the extractors. */ if (mime_used) { if (task->cur_func) { TrackerSparqlBuilder *statements; g_debug (" Using %s...", g_module_name (task->cur_module)); (task->cur_func) (info); statements = tracker_extract_info_get_metadata_builder (info); items = tracker_sparql_builder_get_length (statements); if (items > 0) { tracker_sparql_builder_insert_close (statements); g_debug ("Done (%d items)", items); task->success = TRUE; } } g_free (mime_used); } if (items == 0) { g_debug ("No extractor or failed"); tracker_extract_info_unref (info); info = NULL; } *info_out = info; return (items > 0); }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { TrackerConfig *config; GTime creation_date; GError *error = NULL; TrackerSparqlBuilder *metadata, *preupdate; const gchar *graph; const gchar *urn; TrackerXmpData *xd = NULL; PDFData pd = { 0 }; /* actual data */ PDFData md = { 0 }; /* for merging */ PopplerDocument *document; gchar *xml = NULL; gchar *content, *uri; guint n_bytes; GPtrArray *keywords; guint i; GFile *file; gchar *filename; int fd; gchar *contents = NULL; gsize len; struct stat st; metadata = tracker_extract_info_get_metadata_builder (info); preupdate = tracker_extract_info_get_preupdate_builder (info); graph = tracker_extract_info_get_graph (info); urn = tracker_extract_info_get_urn (info); file = tracker_extract_info_get_file (info); filename = g_file_get_path (file); fd = tracker_file_open_fd (filename); if (fd == -1) { g_warning ("Could not open pdf file '%s': %s\n", filename, g_strerror (errno)); g_free (filename); return FALSE; } if (fstat (fd, &st) == -1) { g_warning ("Could not fstat pdf file '%s': %s\n", filename, g_strerror (errno)); close (fd); g_free (filename); return FALSE; } if (st.st_size == 0) { contents = NULL; len = 0; } else { contents = (gchar *) mmap (NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0); if (contents == NULL || contents == MAP_FAILED) { g_warning ("Could not mmap pdf file '%s': %s\n", filename, g_strerror (errno)); close (fd); g_free (filename); return FALSE; } len = st.st_size; } g_free (filename); uri = g_file_get_uri (file); document = poppler_document_new_from_data (contents, len, NULL, &error); if (error) { if (error->code == POPPLER_ERROR_ENCRYPTED) { tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:PaginatedTextDocument"); tracker_sparql_builder_predicate (metadata, "nfo:isContentEncrypted"); tracker_sparql_builder_object_boolean (metadata, TRUE); g_error_free (error); g_free (uri); close (fd); return TRUE; } else { g_warning ("Couldn't create PopplerDocument from uri:'%s', %s", uri, error->message ? error->message : "no error given"); g_error_free (error); g_free (uri); close (fd); return FALSE; } } if (!document) { g_warning ("Could not create PopplerDocument from uri:'%s', " "NULL returned without an error", uri); g_free (uri); close (fd); return FALSE; } tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:PaginatedTextDocument"); g_object_get (document, "title", &pd.title, "author", &pd.author, "subject", &pd.subject, "keywords", &pd.keywords, "creation-date", &creation_date, "metadata", &xml, NULL); if (creation_date > 0) { pd.creation_date = tracker_date_to_string ((time_t) creation_date); } keywords = g_ptr_array_new_with_free_func ((GDestroyNotify) g_free); if (xml && *xml && (xd = tracker_xmp_new (xml, strlen (xml), uri)) != NULL) { /* The casts here are well understood and known */ md.title = (gchar *) tracker_coalesce_strip (4, pd.title, xd->title, xd->title2, xd->pdf_title); md.subject = (gchar *) tracker_coalesce_strip (2, pd.subject, xd->subject); md.date = (gchar *) tracker_coalesce_strip (3, pd.creation_date, xd->date, xd->time_original); md.author = (gchar *) tracker_coalesce_strip (2, pd.author, xd->creator); write_pdf_data (md, metadata, keywords); if (xd->keywords) { tracker_keywords_parse (keywords, xd->keywords); } if (xd->pdf_keywords) { tracker_keywords_parse (keywords, xd->pdf_keywords); } if (xd->publisher) { tracker_sparql_builder_predicate (metadata, "nco:publisher"); tracker_sparql_builder_object_blank_open (metadata); tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nco:Contact"); tracker_sparql_builder_predicate (metadata, "nco:fullname"); tracker_sparql_builder_object_unvalidated (metadata, xd->publisher); tracker_sparql_builder_object_blank_close (metadata); } if (xd->type) { tracker_sparql_builder_predicate (metadata, "dc:type"); tracker_sparql_builder_object_unvalidated (metadata, xd->type); } if (xd->format) { tracker_sparql_builder_predicate (metadata, "dc:format"); tracker_sparql_builder_object_unvalidated (metadata, xd->format); } if (xd->identifier) { tracker_sparql_builder_predicate (metadata, "dc:identifier"); tracker_sparql_builder_object_unvalidated (metadata, xd->identifier); } if (xd->source) { tracker_sparql_builder_predicate (metadata, "dc:source"); tracker_sparql_builder_object_unvalidated (metadata, xd->source); } if (xd->language) { tracker_sparql_builder_predicate (metadata, "dc:language"); tracker_sparql_builder_object_unvalidated (metadata, xd->language); } if (xd->relation) { tracker_sparql_builder_predicate (metadata, "dc:relation"); tracker_sparql_builder_object_unvalidated (metadata, xd->relation); } if (xd->coverage) { tracker_sparql_builder_predicate (metadata, "dc:coverage"); tracker_sparql_builder_object_unvalidated (metadata, xd->coverage); } if (xd->license) { tracker_sparql_builder_predicate (metadata, "nie:license"); tracker_sparql_builder_object_unvalidated (metadata, xd->license); } if (xd->make || xd->model) { gchar *equip_uri; equip_uri = tracker_sparql_escape_uri_printf ("urn:equipment:%s:%s:", xd->make ? xd->make : "", xd->model ? xd->model : ""); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, equip_uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nfo:Equipment"); if (xd->make) { tracker_sparql_builder_predicate (preupdate, "nfo:manufacturer"); tracker_sparql_builder_object_unvalidated (preupdate, xd->make); } if (xd->model) { tracker_sparql_builder_predicate (preupdate, "nfo:model"); tracker_sparql_builder_object_unvalidated (preupdate, xd->model); } if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nfo:equipment"); tracker_sparql_builder_object_iri (metadata, equip_uri); g_free (equip_uri); } if (xd->orientation) { tracker_sparql_builder_predicate (metadata, "nfo:orientation"); tracker_sparql_builder_object (metadata, xd->orientation); } if (xd->rights) { tracker_sparql_builder_predicate (metadata, "nie:copyright"); tracker_sparql_builder_object_unvalidated (metadata, xd->rights); } if (xd->white_balance) { tracker_sparql_builder_predicate (metadata, "nmm:whiteBalance"); tracker_sparql_builder_object (metadata, xd->white_balance); } if (xd->fnumber) { gdouble value; value = g_strtod (xd->fnumber, NULL); tracker_sparql_builder_predicate (metadata, "nmm:fnumber"); tracker_sparql_builder_object_double (metadata, value); } if (xd->flash) { tracker_sparql_builder_predicate (metadata, "nmm:flash"); tracker_sparql_builder_object (metadata, xd->flash); } if (xd->focal_length) { gdouble value; value = g_strtod (xd->focal_length, NULL); tracker_sparql_builder_predicate (metadata, "nmm:focalLength"); tracker_sparql_builder_object_double (metadata, value); } /* Question: Shouldn't xd->Artist be merged with md.author instead? */ if (xd->artist || xd->contributor) { const gchar *artist; artist = tracker_coalesce_strip (2, xd->artist, xd->contributor); tracker_sparql_builder_predicate (metadata, "nco:contributor"); tracker_sparql_builder_object_blank_open (metadata); tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nco:Contact"); tracker_sparql_builder_predicate (metadata, "nco:fullname"); tracker_sparql_builder_object_unvalidated (metadata, artist); tracker_sparql_builder_object_blank_close (metadata); } if (xd->exposure_time) { gdouble value; value = g_strtod (xd->exposure_time, NULL); tracker_sparql_builder_predicate (metadata, "nmm:exposureTime"); tracker_sparql_builder_object_double (metadata, value); } if (xd->iso_speed_ratings) { gdouble value; value = g_strtod (xd->iso_speed_ratings, NULL); tracker_sparql_builder_predicate (metadata, "nmm:isoSpeed"); tracker_sparql_builder_object_double (metadata, value); } if (xd->description) { tracker_sparql_builder_predicate (metadata, "nie:description"); tracker_sparql_builder_object_unvalidated (metadata, xd->description); } if (xd->metering_mode) { tracker_sparql_builder_predicate (metadata, "nmm:meteringMode"); tracker_sparql_builder_object (metadata, xd->metering_mode); } if (xd->address || xd->state || xd->country || xd->city || xd->gps_altitude || xd->gps_latitude || xd-> gps_longitude) { tracker_sparql_builder_predicate (metadata, "slo:location"); tracker_sparql_builder_object_blank_open (metadata); /* GeoLocation */ tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "slo:GeoLocation"); if (xd->address || xd->state || xd->country || xd->city) { gchar *addruri; addruri = tracker_sparql_get_uuid_urn (); tracker_sparql_builder_predicate (metadata, "slo:postalAddress"); tracker_sparql_builder_object_iri (metadata, addruri); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, addruri); g_free (addruri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nco:PostalAddress"); if (xd->address) { tracker_sparql_builder_predicate (preupdate, "nco:streetAddress"); tracker_sparql_builder_object_unvalidated (preupdate, xd->address); } if (xd->state) { tracker_sparql_builder_predicate (preupdate, "nco:region"); tracker_sparql_builder_object_unvalidated (preupdate, xd->state); } if (xd->city) { tracker_sparql_builder_predicate (preupdate, "nco:locality"); tracker_sparql_builder_object_unvalidated (preupdate, xd->city); } if (xd->country) { tracker_sparql_builder_predicate (preupdate, "nco:country"); tracker_sparql_builder_object_unvalidated (preupdate, xd->country); } if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); } if (xd->gps_altitude) { tracker_sparql_builder_predicate (metadata, "slo:altitude"); tracker_sparql_builder_object_unvalidated (metadata, xd->gps_altitude); } if (xd->gps_latitude) { tracker_sparql_builder_predicate (metadata, "slo:latitude"); tracker_sparql_builder_object_unvalidated (metadata, xd->gps_latitude); } if (xd->gps_longitude) { tracker_sparql_builder_predicate (metadata, "slo:longitude"); tracker_sparql_builder_object_unvalidated (metadata, xd->gps_longitude); } tracker_sparql_builder_object_blank_close (metadata); /* GeoLocation */ } if (xd->regions) { tracker_xmp_apply_regions (preupdate, metadata, graph, xd); } tracker_xmp_free (xd); } else { /* So if we are here we have NO XMP data and we just * write what we know from Poppler. */ write_pdf_data (pd, metadata, keywords); } for (i = 0; i < keywords->len; i++) { gchar *escaped, *subject; const gchar *p; p = g_ptr_array_index (keywords, i); escaped = tracker_sparql_escape_string (p); subject = g_strdup_printf ("_:tag%d", i + 1); /* ensure tag with specified label exists */ tracker_sparql_builder_insert_open (preupdate, graph); tracker_sparql_builder_subject (preupdate, subject); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nao:Tag"); tracker_sparql_builder_predicate (preupdate, "nao:prefLabel"); tracker_sparql_builder_object_unvalidated (preupdate, escaped); tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_append (preupdate, "WHERE { FILTER (NOT EXISTS { " "?tag a nao:Tag ; nao:prefLabel \""); tracker_sparql_builder_append (preupdate, escaped); tracker_sparql_builder_append (preupdate, "\" }) }\n"); /* associate file with tag */ tracker_sparql_builder_insert_open (preupdate, graph); tracker_sparql_builder_subject_iri (preupdate, urn); tracker_sparql_builder_predicate (preupdate, "nao:hasTag"); tracker_sparql_builder_object (preupdate, "?tag"); tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject (preupdate, "?tag"); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nao:Tag"); tracker_sparql_builder_predicate (preupdate, "nao:prefLabel"); tracker_sparql_builder_object_unvalidated (preupdate, escaped); tracker_sparql_builder_where_close (preupdate); g_free (subject); g_free (escaped); } g_ptr_array_free (keywords, TRUE); tracker_sparql_builder_predicate (metadata, "nfo:pageCount"); tracker_sparql_builder_object_int64 (metadata, poppler_document_get_n_pages (document)); config = tracker_main_get_config (); n_bytes = tracker_config_get_max_bytes (config); content = extract_content_text (document, n_bytes); if (content) { tracker_sparql_builder_predicate (metadata, "nie:plainTextContent"); tracker_sparql_builder_object_unvalidated (metadata, content); g_free (content); } read_outline (document, metadata); g_free (xml); g_free (pd.keywords); g_free (pd.title); g_free (pd.subject); g_free (pd.creation_date); g_free (pd.author); g_free (pd.date); g_free (uri); g_object_unref (document); if (contents) { munmap (contents, len); } close (fd); return TRUE; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { TIFF *image; TrackerXmpData *xd = NULL; TrackerIptcData *id = NULL; TrackerExifData *ed = NULL; MergeData md = { 0 }; TiffData td = { 0 }; gchar *filename, *uri; gchar *date; glong exif_offset; GPtrArray *keywords; guint i; GFile *file; TrackerSparqlBuilder *metadata, *preupdate; const gchar *graph, *urn; int fd; #ifdef HAVE_LIBIPTCDATA gchar *iptc_offset; guint32 iptc_size; #endif #ifdef HAVE_EXEMPI gchar *xmp_offset; guint32 size; #endif /* HAVE_EXEMPI */ file = tracker_extract_info_get_file (info); filename = g_file_get_path (file); preupdate = tracker_extract_info_get_preupdate_builder (info); metadata = tracker_extract_info_get_metadata_builder (info); graph = tracker_extract_info_get_graph (info); urn = tracker_extract_info_get_urn (info); fd = tracker_file_open_fd (filename); if (fd == -1) { g_warning ("Could not open tiff file '%s': %s\n", filename, g_strerror (errno)); g_free (filename); return FALSE; } if ((image = TIFFFdOpen (fd, filename, "r")) == NULL){ g_warning ("Could not open image:'%s'\n", filename); g_free (filename); close (fd); return FALSE; } tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:Image"); tracker_sparql_builder_object (metadata, "nmm:Photo"); uri = g_file_get_uri (file); #ifdef HAVE_LIBIPTCDATA if (TIFFGetField (image, TIFFTAG_RICHTIFFIPTC, &iptc_size, &iptc_offset)) { if (TIFFIsByteSwapped(image) != 0) { TIFFSwabArrayOfLong((uint32*) iptc_offset, (unsigned long) iptc_size); } id = tracker_iptc_new (iptc_offset, 4 * iptc_size, uri); } #endif /* HAVE_LIBIPTCDATA */ if (!id) { id = g_new0 (TrackerIptcData, 1); } /* FIXME There are problems between XMP data embedded with different tools due to bugs in the original spec (type) */ #ifdef HAVE_EXEMPI if (TIFFGetField (image, TIFFTAG_XMLPACKET, &size, &xmp_offset)) { xd = tracker_xmp_new (xmp_offset, size, uri); } #endif /* HAVE_EXEMPI */ if (!xd) { xd = g_new0 (TrackerXmpData, 1); } ed = g_new0 (TrackerExifData, 1); /* Get Tiff specifics */ td.width = tag_to_string (image, TIFFTAG_IMAGEWIDTH, TAG_TYPE_UINT32); td.length = tag_to_string (image, TIFFTAG_IMAGELENGTH, TAG_TYPE_UINT32); td.artist = tag_to_string (image, TIFFTAG_ARTIST, TAG_TYPE_STRING); td.copyright = tag_to_string (image, TIFFTAG_COPYRIGHT, TAG_TYPE_STRING); date = tag_to_string (image, TIFFTAG_DATETIME, TAG_TYPE_STRING); td.date = tracker_date_guess (date); g_free (date); td.title = tag_to_string (image, TIFFTAG_DOCUMENTNAME, TAG_TYPE_STRING); td.description = tag_to_string (image, TIFFTAG_IMAGEDESCRIPTION, TAG_TYPE_STRING); td.make = tag_to_string (image, TIFFTAG_MAKE, TAG_TYPE_STRING); td.model = tag_to_string (image, TIFFTAG_MODEL, TAG_TYPE_STRING); td.orientation = get_orientation (image); /* Get Exif specifics */ if (TIFFGetField (image, TIFFTAG_EXIFIFD, &exif_offset)) { if (TIFFReadEXIFDirectory (image, exif_offset)) { ed->exposure_time = tag_to_string (image, EXIFTAG_EXPOSURETIME, TAG_TYPE_DOUBLE); ed->fnumber = tag_to_string (image, EXIFTAG_FNUMBER, TAG_TYPE_DOUBLE); ed->iso_speed_ratings = tag_to_string (image, EXIFTAG_ISOSPEEDRATINGS, TAG_TYPE_C16_UINT16); date = tag_to_string (image, EXIFTAG_DATETIMEORIGINAL, TAG_TYPE_STRING); ed->time_original = tracker_date_guess (date); g_free (date); ed->metering_mode = get_metering_mode (image); ed->flash = get_flash (image); ed->focal_length = tag_to_string (image, EXIFTAG_DATETIMEORIGINAL, TAG_TYPE_DOUBLE); ed->white_balance = get_white_balance (image); /* ed->software = tag_to_string (image, EXIFTAG_SOFTWARE, TAG_TYPE_STRING); */ } } TIFFClose (image); g_free (filename); md.title = tracker_coalesce_strip (5, xd->title, xd->pdf_title, td.title, ed->document_name, xd->title2); md.orientation = tracker_coalesce_strip (4, xd->orientation, td.orientation, ed->orientation, id->image_orientation); md.copyright = tracker_coalesce_strip (4, xd->rights, td.copyright, ed->copyright, id->copyright_notice); md.white_balance = tracker_coalesce_strip (2, xd->white_balance, ed->white_balance); md.fnumber = tracker_coalesce_strip (2, xd->fnumber, ed->fnumber); md.flash = tracker_coalesce_strip (2, xd->flash, ed->flash); md.focal_length = tracker_coalesce_strip (2, xd->focal_length, ed->focal_length); md.artist = tracker_coalesce_strip (4, xd->artist, td.artist, ed->artist, xd->contributor); md.exposure_time = tracker_coalesce_strip (2, xd->exposure_time, ed->exposure_time); md.iso_speed_ratings = tracker_coalesce_strip (2, xd->iso_speed_ratings, ed->iso_speed_ratings); md.date = tracker_coalesce_strip (6, xd->date, xd->time_original, td.date, ed->time, id->date_created, ed->time_original); md.description = tracker_coalesce_strip (3, xd->description, td.description, ed->description); md.metering_mode = tracker_coalesce_strip (2, xd->metering_mode, ed->metering_mode); md.city = tracker_coalesce_strip (2, xd->city, id->city); md.state = tracker_coalesce_strip (2, xd->state, id->state); md.address = tracker_coalesce_strip (2, xd->address, id->sublocation); md.country = tracker_coalesce_strip (2, xd->country, id->country_name); /* FIXME We are not handling the altitude ref here for xmp */ md.gps_altitude = tracker_coalesce_strip (2, xd->gps_altitude, ed->gps_altitude); md.gps_latitude = tracker_coalesce_strip (2, xd->gps_latitude, ed->gps_latitude); md.gps_longitude = tracker_coalesce_strip (2, xd->gps_longitude, ed->gps_longitude); md.gps_direction = tracker_coalesce_strip (2, xd->gps_direction, ed->gps_direction); md.creator = tracker_coalesce_strip (3, xd->creator, id->byline, id->credit); md.x_dimension = tracker_coalesce_strip (2, td.width, ed->x_dimension); md.y_dimension = tracker_coalesce_strip (2, td.length, ed->y_dimension); md.make = tracker_coalesce_strip (3, xd->make, td.make, ed->make); md.model = tracker_coalesce_strip (3, xd->model, td.model, ed->model); keywords = g_ptr_array_new_with_free_func ((GDestroyNotify) g_free); if (ed->user_comment) { tracker_sparql_builder_predicate (metadata, "nie:comment"); tracker_sparql_builder_object_unvalidated (metadata, ed->user_comment); } if (md.x_dimension) { tracker_sparql_builder_predicate (metadata, "nfo:width"); tracker_sparql_builder_object_unvalidated (metadata, md.x_dimension); } if (md.y_dimension) { tracker_sparql_builder_predicate (metadata, "nfo:height"); tracker_sparql_builder_object_unvalidated (metadata, md.y_dimension); } if (xd->keywords) { tracker_keywords_parse (keywords, xd->keywords); } if (xd->pdf_keywords) { tracker_keywords_parse (keywords, xd->pdf_keywords); } if (xd->subject) { tracker_keywords_parse (keywords, xd->subject); } if (xd->publisher) { gchar *uri = tracker_sparql_escape_uri_printf ("urn:contact:%s", xd->publisher); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nco:Contact"); tracker_sparql_builder_predicate (preupdate, "nco:fullname"); tracker_sparql_builder_object_unvalidated (preupdate, xd->publisher); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nco:publisher"); tracker_sparql_builder_object_iri (metadata, uri); g_free (uri); } if (xd->type) { tracker_sparql_builder_predicate (metadata, "dc:type"); tracker_sparql_builder_object_unvalidated (metadata, xd->type); } if (xd->format) { tracker_sparql_builder_predicate (metadata, "dc:format"); tracker_sparql_builder_object_unvalidated (metadata, xd->format); } if (xd->identifier) { tracker_sparql_builder_predicate (metadata, "dc:identifier"); tracker_sparql_builder_object_unvalidated (metadata, xd->identifier); } if (xd->source) { tracker_sparql_builder_predicate (metadata, "dc:source"); tracker_sparql_builder_object_unvalidated (metadata, xd->source); } if (xd->language) { tracker_sparql_builder_predicate (metadata, "dc:language"); tracker_sparql_builder_object_unvalidated (metadata, xd->language); } if (xd->relation) { tracker_sparql_builder_predicate (metadata, "dc:relation"); tracker_sparql_builder_object_unvalidated (metadata, xd->relation); } if (xd->coverage) { tracker_sparql_builder_predicate (metadata, "dc:coverage"); tracker_sparql_builder_object_unvalidated (metadata, xd->coverage); } if (xd->rating) { tracker_sparql_builder_predicate (metadata, "nao:numericRating"); tracker_sparql_builder_object_unvalidated (metadata, xd->rating); } if (xd->license) { tracker_sparql_builder_predicate (metadata, "nie:license"); tracker_sparql_builder_object_unvalidated (metadata, xd->license); } if (xd->regions) { tracker_xmp_apply_regions (preupdate, metadata, graph, xd); } if (md.address || md.state || md.country || md.city || md.gps_altitude || md.gps_latitude || md.gps_longitude) { tracker_sparql_builder_predicate (metadata, "slo:location"); tracker_sparql_builder_object_blank_open (metadata); /* GeoPoint */ tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "slo:GeoLocation"); if (md.address || md.state || md.country || md.city) { gchar *addruri; addruri = tracker_sparql_get_uuid_urn (); tracker_sparql_builder_predicate (metadata, "slo:postalAddress"); tracker_sparql_builder_object_iri (metadata, addruri); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, addruri); g_free (addruri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nco:PostalAddress"); if (md.address) { tracker_sparql_builder_predicate (preupdate, "nco:streetAddress"); tracker_sparql_builder_object_unvalidated (preupdate, md.address); } if (md.state) { tracker_sparql_builder_predicate (preupdate, "nco:region"); tracker_sparql_builder_object_unvalidated (preupdate, md.state); } if (md.city) { tracker_sparql_builder_predicate (preupdate, "nco:locality"); tracker_sparql_builder_object_unvalidated (preupdate, md.city); } if (md.country) { tracker_sparql_builder_predicate (preupdate, "nco:country"); tracker_sparql_builder_object_unvalidated (preupdate, md.country); } if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); } if (md.gps_altitude) { tracker_sparql_builder_predicate (metadata, "slo:altitude"); tracker_sparql_builder_object_unvalidated (metadata, md.gps_altitude); } if (md.gps_latitude) { tracker_sparql_builder_predicate (metadata, "slo:latitude"); tracker_sparql_builder_object_unvalidated (metadata, md.gps_latitude); } if (md.gps_longitude) { tracker_sparql_builder_predicate (metadata, "slo:longitude"); tracker_sparql_builder_object_unvalidated (metadata, md.gps_longitude); } tracker_sparql_builder_object_blank_close (metadata); /* GeoLocation */ } if (md.gps_direction) { tracker_sparql_builder_predicate (metadata, "nfo:heading"); tracker_sparql_builder_object_unvalidated (metadata, md.gps_direction); } if (id->contact) { gchar *uri = tracker_sparql_escape_uri_printf ("urn:contact:%s", id->contact); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nco:Contact"); tracker_sparql_builder_predicate (preupdate, "nco:fullname"); tracker_sparql_builder_object_unvalidated (preupdate, id->contact); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nco:representative"); tracker_sparql_builder_object_iri (metadata, uri); g_free (uri); } if (id->keywords) { tracker_keywords_parse (keywords, id->keywords); } for (i = 0; i < keywords->len; i++) { gchar *escaped, *subject; const gchar *p; p = g_ptr_array_index (keywords, i); escaped = tracker_sparql_escape_string (p); subject = g_strdup_printf ("_:tag%d", i + 1); /* ensure tag with specified label exists */ tracker_sparql_builder_insert_open (preupdate, graph); tracker_sparql_builder_subject (preupdate, subject); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nao:Tag"); tracker_sparql_builder_predicate (preupdate, "nao:prefLabel"); tracker_sparql_builder_object_unvalidated (preupdate, escaped); tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_append (preupdate, "WHERE { FILTER (NOT EXISTS { " "?tag a nao:Tag ; nao:prefLabel \""); tracker_sparql_builder_append (preupdate, escaped); tracker_sparql_builder_append (preupdate, "\" }) }\n"); /* associate file with tag */ tracker_sparql_builder_insert_open (preupdate, graph); tracker_sparql_builder_subject_iri (preupdate, urn); tracker_sparql_builder_predicate (preupdate, "nao:hasTag"); tracker_sparql_builder_object (preupdate, "?tag"); tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject (preupdate, "?tag"); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nao:Tag"); tracker_sparql_builder_predicate (preupdate, "nao:prefLabel"); tracker_sparql_builder_object_unvalidated (preupdate, escaped); tracker_sparql_builder_where_close (preupdate); g_free (subject); g_free (escaped); } g_ptr_array_free (keywords, TRUE); if (md.make || md.model) { gchar *equip_uri; equip_uri = tracker_sparql_escape_uri_printf ("urn:equipment:%s:%s:", md.make ? md.make : "", md.model ? md.model : ""); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, equip_uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nfo:Equipment"); if (md.make) { tracker_sparql_builder_predicate (preupdate, "nfo:manufacturer"); tracker_sparql_builder_object_unvalidated (preupdate, md.make); } if (md.model) { tracker_sparql_builder_predicate (preupdate, "nfo:model"); tracker_sparql_builder_object_unvalidated (preupdate, md.model); } if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nfo:equipment"); tracker_sparql_builder_object_iri (metadata, equip_uri); g_free (equip_uri); } tracker_guarantee_title_from_file (metadata, "nie:title", md.title, uri, NULL); if (md.orientation) { tracker_sparql_builder_predicate (metadata, "nfo:orientation"); tracker_sparql_builder_object_unvalidated (metadata, md.orientation); } if (md.copyright) { tracker_sparql_builder_predicate (metadata, "nie:copyright"); tracker_sparql_builder_object_unvalidated (metadata, md.copyright); } if (md.white_balance) { tracker_sparql_builder_predicate (metadata, "nmm:whiteBalance"); tracker_sparql_builder_object_unvalidated (metadata, md.white_balance); } if (md.fnumber) { tracker_sparql_builder_predicate (metadata, "nmm:fnumber"); tracker_sparql_builder_object_unvalidated (metadata, md.fnumber); } if (md.flash) { tracker_sparql_builder_predicate (metadata, "nmm:flash"); tracker_sparql_builder_object_unvalidated (metadata, md.flash); } if (md.focal_length) { tracker_sparql_builder_predicate (metadata, "nmm:focalLength"); tracker_sparql_builder_object_unvalidated (metadata, md.focal_length); } if (md.artist) { gchar *uri = tracker_sparql_escape_uri_printf ("urn:contact:%s", md.artist); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nco:Contact"); tracker_sparql_builder_predicate (preupdate, "nco:fullname"); tracker_sparql_builder_object_unvalidated (preupdate, md.artist); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nco:contributor"); tracker_sparql_builder_object_iri (metadata, uri); g_free (uri); } if (md.exposure_time) { tracker_sparql_builder_predicate (metadata, "nmm:exposureTime"); tracker_sparql_builder_object_unvalidated (metadata, md.exposure_time); } if (md.iso_speed_ratings) { tracker_sparql_builder_predicate (metadata, "nmm:isoSpeed"); tracker_sparql_builder_object_unvalidated (metadata, md.iso_speed_ratings); } tracker_guarantee_date_from_file_mtime (metadata, "nie:contentCreated", md.date, uri); if (md.description) { tracker_sparql_builder_predicate (metadata, "nie:description"); tracker_sparql_builder_object_unvalidated (metadata, md.description); } if (md.metering_mode) { tracker_sparql_builder_predicate (metadata, "nmm:meteringMode"); tracker_sparql_builder_object_unvalidated (metadata, md.metering_mode); } if (md.creator) { gchar *uri = tracker_sparql_escape_uri_printf ("urn:contact:%s", md.creator); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nco:Contact"); tracker_sparql_builder_predicate (preupdate, "nco:fullname"); tracker_sparql_builder_object_unvalidated (preupdate, md.creator); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); /* NOTE: We only have affiliation with * nco:PersonContact and we are using * nco:Contact here. */ /* if (id->byline_title) { */ /* tracker_sparql_builder_insert_open (preupdate, NULL); */ /* tracker_sparql_builder_subject (preupdate, "_:affiliation_by_line"); */ /* tracker_sparql_builder_predicate (preupdate, "a"); */ /* tracker_sparql_builder_object (preupdate, "nco:Affiliation"); */ /* tracker_sparql_builder_predicate (preupdate, "nco:title"); */ /* tracker_sparql_builder_object_unvalidated (preupdate, id->byline_title); */ /* tracker_sparql_builder_insert_close (preupdate); */ /* tracker_sparql_builder_predicate (metadata, "a"); */ /* tracker_sparql_builder_object (metadata, "nco:PersonContact"); */ /* tracker_sparql_builder_predicate (metadata, "nco:hasAffiliation"); */ /* tracker_sparql_builder_object (metadata, "_:affiliation_by_line"); */ /* } */ tracker_sparql_builder_predicate (metadata, "nco:creator"); tracker_sparql_builder_object_iri (metadata, uri); g_free (uri); } if (ed->x_resolution) { gdouble value; value = ed->resolution_unit != 3 ? g_strtod (ed->x_resolution, NULL) : g_strtod (ed->x_resolution, NULL) * CM_TO_INCH; tracker_sparql_builder_predicate (metadata, "nfo:horizontalResolution"); tracker_sparql_builder_object_double (metadata, value); } if (ed->y_resolution) { gdouble value; value = ed->resolution_unit != 3 ? g_strtod (ed->y_resolution, NULL) : g_strtod (ed->y_resolution, NULL) * CM_TO_INCH; tracker_sparql_builder_predicate (metadata, "nfo:verticalResolution"); tracker_sparql_builder_object_double (metadata, value); } tiff_data_free (&td); tracker_exif_free (ed); tracker_xmp_free (xd); tracker_iptc_free (id); g_free (uri); close (fd); return TRUE; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { goffset size; FILE *f; png_structp png_ptr; png_infop info_ptr; png_infop end_ptr; png_bytepp row_pointers; guint row; png_uint_32 width, height; gint bit_depth, color_type; gint interlace_type, compression_type, filter_type; const gchar *dlna_profile, *dlna_mimetype, *graph; TrackerSparqlBuilder *preupdate, *metadata; gchar *filename, *uri; GString *where; GFile *file; file = tracker_extract_info_get_file (info); filename = g_file_get_path (file); size = tracker_file_get_size (filename); preupdate = tracker_extract_info_get_preupdate_builder (info); metadata = tracker_extract_info_get_metadata_builder (info); graph = tracker_extract_info_get_graph (info); if (size < 64) { return FALSE; } f = tracker_file_open (filename); g_free (filename); if (!f) { return FALSE; } png_ptr = png_create_read_struct (PNG_LIBPNG_VER_STRING, NULL, NULL, NULL); if (!png_ptr) { tracker_file_close (f, FALSE); return FALSE; } info_ptr = png_create_info_struct (png_ptr); if (!info_ptr) { png_destroy_read_struct (&png_ptr, &info_ptr, NULL); tracker_file_close (f, FALSE); return FALSE; } end_ptr = png_create_info_struct (png_ptr); if (!end_ptr) { png_destroy_read_struct (&png_ptr, &info_ptr, NULL); tracker_file_close (f, FALSE); return FALSE; } if (setjmp (png_jmpbuf (png_ptr))) { png_destroy_read_struct (&png_ptr, &info_ptr, &end_ptr); tracker_file_close (f, FALSE); return FALSE; } png_init_io (png_ptr, f); png_read_info (png_ptr, info_ptr); if (!png_get_IHDR (png_ptr, info_ptr, &width, &height, &bit_depth, &color_type, &interlace_type, &compression_type, &filter_type)) { png_destroy_read_struct (&png_ptr, &info_ptr, &end_ptr); tracker_file_close (f, FALSE); return FALSE; } /* Read the image. FIXME We should be able to skip this step and * just get the info from the end. This causes some errors atm. */ row_pointers = g_new0 (png_bytep, height); for (row = 0; row < height; row++) { row_pointers[row] = png_malloc (png_ptr, png_get_rowbytes (png_ptr,info_ptr)); } png_read_image (png_ptr, row_pointers); for (row = 0; row < height; row++) { png_free (png_ptr, row_pointers[row]); } g_free (row_pointers); png_read_end (png_ptr, end_ptr); tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:Image"); tracker_sparql_builder_object (metadata, "nmm:Photo"); uri = g_file_get_uri (file); where = g_string_new (""); read_metadata (preupdate, metadata, where, png_ptr, info_ptr, end_ptr, uri, graph); tracker_extract_info_set_where_clause (info, where->str); g_string_free (where, TRUE); g_free (uri); tracker_sparql_builder_predicate (metadata, "nfo:width"); tracker_sparql_builder_object_int64 (metadata, width); tracker_sparql_builder_predicate (metadata, "nfo:height"); tracker_sparql_builder_object_int64 (metadata, height); if (guess_dlna_profile (bit_depth, width, height, &dlna_profile, &dlna_mimetype)) { tracker_sparql_builder_predicate (metadata, "nmm:dlnaProfile"); tracker_sparql_builder_object_string (metadata, dlna_profile); tracker_sparql_builder_predicate (metadata, "nmm:dlnaMime"); tracker_sparql_builder_object_string (metadata, dlna_mimetype); } png_destroy_read_struct (&png_ptr, &info_ptr, &end_ptr); tracker_file_close (f, FALSE); return TRUE; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { TrackerSparqlBuilder *preupdate, *metadata; goffset size; GifFileType *gifFile = NULL; GString *where; const gchar *graph; gchar *filename, *uri; GFile *file; int fd; #if GIFLIB_MAJOR >= 5 int err; #endif preupdate = tracker_extract_info_get_preupdate_builder (info); metadata = tracker_extract_info_get_metadata_builder (info); graph = tracker_extract_info_get_graph (info); file = tracker_extract_info_get_file (info); filename = g_file_get_path (file); size = tracker_file_get_size (filename); if (size < 64) { g_free (filename); return FALSE; } fd = tracker_file_open_fd (filename); if (fd == -1) { g_warning ("Could not open GIF file '%s': %s\n", filename, g_strerror (errno)); g_free (filename); return FALSE; } #if GIFLIB_MAJOR < 5 if ((gifFile = DGifOpenFileHandle (fd)) == NULL) { PrintGifError (); #else /* GIFLIB_MAJOR < 5 */ if ((gifFile = DGifOpenFileHandle (fd, &err)) == NULL) { gif_error ("Could not open GIF file with handle", err); #endif /* GIFLIB_MAJOR < 5 */ close (fd); return FALSE; } g_free (filename); tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:Image"); tracker_sparql_builder_object (metadata, "nmm:Photo"); where = g_string_new (""); uri = g_file_get_uri (file); read_metadata (preupdate, metadata, where, gifFile, uri, graph); tracker_extract_info_set_where_clause (info, where->str); g_string_free (where, TRUE); g_free (uri); if (DGifCloseFile (gifFile) != GIF_OK) { #if GIFLIB_MAJOR < 5 PrintGifError (); #else /* GIFLIB_MAJOR < 5 */ gif_error ("Could not close GIF file", gifFile->Error); #endif /* GIFLIB_MAJOR < 5 */ } return TRUE; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { FLAC__Metadata_SimpleIterator *iter; FLAC__StreamMetadata *stream = NULL, *vorbis, *picture; FLAC__bool success; FlacData fd = { 0 }; TrackerSparqlBuilder *preupdate, *metadata; gchar *filename, *uri, *artist_uri = NULL, *album_uri = NULL; const gchar *creator; GFile *file; goffset size; const gchar *graph; graph = tracker_extract_info_get_graph (info); preupdate = tracker_extract_info_get_preupdate_builder (info); metadata = tracker_extract_info_get_metadata_builder (info); file = tracker_extract_info_get_file (info); filename = g_file_get_path (file); size = tracker_file_get_size (filename); if (size < 18) { g_free (filename); return FALSE; } iter = FLAC__metadata_simple_iterator_new (); success = FLAC__metadata_simple_iterator_init (iter, filename, TRUE, TRUE); g_free (filename); if (!success) { FLAC__metadata_simple_iterator_delete (iter); return FALSE; } uri = g_file_get_uri (file); do { switch (FLAC__metadata_simple_iterator_get_block_type (iter)) { case FLAC__METADATA_TYPE_STREAMINFO: stream = FLAC__metadata_simple_iterator_get_block (iter); break; case FLAC__METADATA_TYPE_VORBIS_COMMENT: vorbis = FLAC__metadata_simple_iterator_get_block (iter); parse_vorbis_comments (&(vorbis->data.vorbis_comment), &fd); FLAC__metadata_object_delete (vorbis); break; case FLAC__METADATA_TYPE_PICTURE: picture = FLAC__metadata_simple_iterator_get_block (iter); /* Deal with picture */ FLAC__metadata_object_delete (picture); break; default: break; } } while (FLAC__metadata_simple_iterator_next (iter)); creator = tracker_coalesce_strip (3, fd.artist, fd.albumartist, fd.performer); if (creator) { artist_uri = tracker_sparql_escape_uri_printf ("urn:artist:%s", creator); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, artist_uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nmm:Artist"); tracker_sparql_builder_predicate (preupdate, "nmm:artistName"); tracker_sparql_builder_object_unvalidated (preupdate, creator); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); } if (fd.album) { if (fd.albumartist) { album_uri = tracker_sparql_escape_uri_printf ("urn:album:%s:%s", fd.album, fd.albumartist); } else { album_uri = tracker_sparql_escape_uri_printf ("urn:album:%s", fd.album); } tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, album_uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nmm:MusicAlbum"); /* FIXME: nmm:albumTitle is now deprecated * tracker_sparql_builder_predicate (preupdate, "nie:title"); */ tracker_sparql_builder_predicate (preupdate, "nmm:albumTitle"); tracker_sparql_builder_object_unvalidated (preupdate, fd.album); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); if (fd.trackcount) { tracker_sparql_builder_delete_open (preupdate, NULL); tracker_sparql_builder_subject_iri (preupdate, album_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumTrackCount"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_delete_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject_iri (preupdate, album_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumTrackCount"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_where_close (preupdate); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, album_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumTrackCount"); tracker_sparql_builder_object_unvalidated (preupdate, fd.trackcount); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); } if (fd.albumgain) { tracker_sparql_builder_delete_open (preupdate, NULL); tracker_sparql_builder_subject_iri (preupdate, album_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumGain"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_delete_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject_iri (preupdate, album_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumGain"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_where_close (preupdate); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, album_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumGain"); tracker_sparql_builder_object_double (preupdate, atof (fd.albumgain)); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); } if (fd.albumpeakgain) { tracker_sparql_builder_delete_open (preupdate, NULL); tracker_sparql_builder_subject_iri (preupdate, album_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumPeakGain"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_delete_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject_iri (preupdate, album_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumPeakGain"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_where_close (preupdate); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, album_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumPeakGain"); tracker_sparql_builder_object_double (preupdate, atof (fd.albumpeakgain)); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); } } tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nmm:MusicPiece"); tracker_sparql_builder_object (metadata, "nfo:Audio"); add_tuple (metadata, "nmm:performer", artist_uri); g_free (artist_uri); add_tuple (metadata, "nmm:musicAlbum", album_uri); tracker_guarantee_title_from_file (metadata, "nie:title", fd.title, uri, NULL); add_tuple (metadata, "nmm:trackNumber", fd.tracknumber); if (fd.album && album_uri) { gchar *album_disc_uri; if (fd.albumartist) { album_disc_uri = tracker_sparql_escape_uri_printf ("urn:album-disc:%s:%s:Disc%d", fd.album, fd.albumartist, fd.discno ? atoi(fd.discno) : 1); } else { album_disc_uri = tracker_sparql_escape_uri_printf ("urn:album-disc:%s:Disc%d", fd.album, fd.discno ? atoi(fd.discno) : 1); } tracker_sparql_builder_delete_open (preupdate, NULL); tracker_sparql_builder_subject_iri (preupdate, album_disc_uri); tracker_sparql_builder_predicate (preupdate, "nmm:setNumber"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_delete_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject_iri (preupdate, album_disc_uri); tracker_sparql_builder_predicate (preupdate, "nmm:setNumber"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_where_close (preupdate); tracker_sparql_builder_delete_open (preupdate, NULL); tracker_sparql_builder_subject_iri (preupdate, album_disc_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumDiscAlbum"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_delete_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject_iri (preupdate, album_disc_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumDiscAlbum"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_where_close (preupdate); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, album_disc_uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nmm:MusicAlbumDisc"); tracker_sparql_builder_predicate (preupdate, "nmm:setNumber"); tracker_sparql_builder_object_int64 (preupdate, fd.discno ? atoi (fd.discno) : 1); tracker_sparql_builder_predicate (preupdate, "nmm:albumDiscAlbum"); tracker_sparql_builder_object_iri (preupdate, album_uri); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nmm:musicAlbumDisc"); tracker_sparql_builder_object_iri (metadata, album_disc_uri); g_free (album_disc_uri); } g_free (album_uri); /* FIXME: Trackgain/Trackpeakgain: commented out in vorbis */ add_tuple (metadata, "nie:comment", fd.comment); add_tuple (metadata, "nie:contentCreated", fd.date); add_tuple (metadata, "nfo:genre", fd.genre); add_tuple (metadata, "nie:plainTextContent", fd.lyrics); add_tuple (metadata, "nie:copyright", fd.copyright); add_tuple (metadata, "nie:license", fd.license); if (fd.publisher) { tracker_sparql_builder_predicate (metadata, "dc:publisher"); tracker_sparql_builder_object_blank_open (metadata); tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nco:Contact"); tracker_sparql_builder_predicate (metadata, "nco:fullname"); tracker_sparql_builder_object_unvalidated (metadata, fd.publisher); tracker_sparql_builder_object_blank_close (metadata); } if (stream) { tracker_sparql_builder_predicate (metadata, "nfo:sampleRate"); tracker_sparql_builder_object_int64 (metadata, stream->data.stream_info.sample_rate); tracker_sparql_builder_predicate (metadata, "nfo:channels"); tracker_sparql_builder_object_int64 (metadata, stream->data.stream_info.channels); tracker_sparql_builder_predicate (metadata, "nfo:averageBitrate"); tracker_sparql_builder_object_int64 (metadata, stream->data.stream_info.bits_per_sample); tracker_sparql_builder_predicate (metadata, "nfo:duration"); tracker_sparql_builder_object_int64 (metadata, stream->data.stream_info.total_samples / stream->data.stream_info.sample_rate); } g_free (fd.artist); g_free (fd.album); g_free (fd.albumartist); g_free (fd.performer); g_free (fd.title); g_free (fd.trackcount); g_free (fd.tracknumber); g_free (fd.discno); g_free (fd.trackgain); g_free (fd.trackpeakgain); g_free (fd.albumgain); g_free (fd.albumpeakgain); g_free (fd.date); g_free (fd.comment); g_free (fd.genre); g_free (fd.mbalbumid); g_free (fd.mbartistid); g_free (fd.mbalbumartistid); g_free (fd.mbtrackid); g_free (fd.lyrics); g_free (fd.copyright); g_free (fd.license); g_free (fd.organisation); g_free (fd.location); g_free (fd.publisher); g_free (uri); return TRUE; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { TrackerSparqlBuilder *metadata; GFile *file; TrackerConfig *config; htmlDocPtr doc; parser_data pd; gchar *filename; xmlSAXHandler handler = { NULL, /* internalSubset */ NULL, /* isStandalone */ NULL, /* hasInternalSubset */ NULL, /* hasExternalSubset */ NULL, /* resolveEntity */ NULL, /* getEntity */ NULL, /* entityDecl */ NULL, /* notationDecl */ NULL, /* attributeDecl */ NULL, /* elementDecl */ NULL, /* unparsedEntityDecl */ NULL, /* setDocumentLocator */ NULL, /* startDocument */ NULL, /* endDocument */ parser_start_element, /* startElement */ parser_end_element, /* endElement */ NULL, /* reference */ parser_characters, /* characters */ NULL, /* ignorableWhitespace */ NULL, /* processingInstruction */ NULL, /* comment */ NULL, /* xmlParserWarning */ NULL, /* xmlParserError */ NULL, /* xmlParserError */ NULL, /* getParameterEntity */ NULL, /* cdataBlock */ NULL, /* externalSubset */ 1, /* initialized */ NULL, /* private */ NULL, /* startElementNsSAX2Func */ NULL, /* endElementNsSAX2Func */ NULL /* xmlStructuredErrorFunc */ }; metadata = tracker_extract_info_get_metadata_builder (info); file = tracker_extract_info_get_file (info); tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nfo:HtmlDocument"); pd.metadata = metadata; pd.current = -1; pd.in_body = FALSE; pd.plain_text = g_string_new (NULL); pd.title = g_string_new (NULL); config = tracker_main_get_config (); pd.n_bytes_remaining = tracker_config_get_max_bytes (config); filename = g_file_get_path (file); doc = htmlSAXParseFile (filename, NULL, &handler, &pd); g_free (filename); if (doc) { xmlFreeDoc (doc); } g_strstrip (pd.plain_text->str); g_strstrip (pd.title->str); if (pd.title->str && *pd.title->str != '\0') { tracker_sparql_builder_predicate (metadata, "nie:title"); tracker_sparql_builder_object_unvalidated (metadata, pd.title->str); } if (pd.plain_text->str && *pd.plain_text->str != '\0') { tracker_sparql_builder_predicate (metadata, "nie:plainTextContent"); tracker_sparql_builder_object_unvalidated (metadata, pd.plain_text->str); } g_string_free (pd.plain_text, TRUE); g_string_free (pd.title, TRUE); return TRUE; }
G_MODULE_EXPORT gboolean tracker_extract_get_metadata (TrackerExtractInfo *info) { TrackerSparqlBuilder *preupdate, *metadata; VorbisData vd = { 0 }; MergeData md = { 0 }; FILE *f; gchar *filename; OggVorbis_File vf; vorbis_comment *comment; vorbis_info *vi; unsigned int bitrate; gint time; GFile *file; const gchar *graph; file = tracker_extract_info_get_file (info); filename = g_file_get_path (file); f = tracker_file_open (filename); g_free (filename); preupdate = tracker_extract_info_get_preupdate_builder (info); metadata = tracker_extract_info_get_metadata_builder (info); graph = tracker_extract_info_get_graph (info); if (!f) { return FALSE; } if (ov_open (f, &vf, NULL, 0) < 0) { tracker_file_close (f, FALSE); return FALSE; } tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nmm:MusicPiece"); tracker_sparql_builder_object (metadata, "nfo:Audio"); if ((comment = ov_comment (&vf, -1)) != NULL) { gchar *date; vd.title = ogg_get_comment (comment, "title"); vd.artist = ogg_get_comment (comment, "artist"); vd.album = ogg_get_comment (comment, "album"); vd.album_artist = ogg_get_comment (comment, "albumartist"); vd.track_count = ogg_get_comment (comment, "trackcount"); vd.track_number = ogg_get_comment (comment, "tracknumber"); vd.disc_number = ogg_get_comment (comment, "DiscNo"); vd.performer = ogg_get_comment (comment, "Performer"); vd.track_gain = ogg_get_comment (comment, "TrackGain"); vd.track_peak_gain = ogg_get_comment (comment, "TrackPeakGain"); vd.album_gain = ogg_get_comment (comment, "AlbumGain"); vd.album_peak_gain = ogg_get_comment (comment, "AlbumPeakGain"); date = ogg_get_comment (comment, "date"); vd.date = tracker_date_guess (date); g_free (date); vd.comment = ogg_get_comment (comment, "comment"); vd.genre = ogg_get_comment (comment, "genre"); vd.codec = ogg_get_comment (comment, "Codec"); vd.codec_version = ogg_get_comment (comment, "CodecVersion"); vd.sample_rate = ogg_get_comment (comment, "SampleRate"); vd.channels = ogg_get_comment (comment, "Channels"); vd.mb_album_id = ogg_get_comment (comment, "MBAlbumID"); vd.mb_artist_id = ogg_get_comment (comment, "MBArtistID"); vd.mb_album_artist_id = ogg_get_comment (comment, "MBAlbumArtistID"); vd.mb_track_id = ogg_get_comment (comment, "MBTrackID"); vd.lyrics = ogg_get_comment (comment, "Lyrics"); vd.copyright = ogg_get_comment (comment, "Copyright"); vd.license = ogg_get_comment (comment, "License"); vd.organization = ogg_get_comment (comment, "Organization"); vd.location = ogg_get_comment (comment, "Location"); vd.publisher = ogg_get_comment (comment, "Publisher"); vorbis_comment_clear (comment); } md.creator = tracker_coalesce_strip (3, vd.artist, vd.album_artist, vd.performer); if (md.creator) { /* NOTE: This must be created before vd.album is evaluated */ md.creator_uri = tracker_sparql_escape_uri_printf ("urn:artist:%s", md.creator); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, md.creator_uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nmm:Artist"); tracker_sparql_builder_predicate (preupdate, "nmm:artistName"); tracker_sparql_builder_object_unvalidated (preupdate, md.creator); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nmm:performer"); tracker_sparql_builder_object_iri (metadata, md.creator_uri); } if (vd.album) { gchar *uri; if (vd.album_artist) { uri = tracker_sparql_escape_uri_printf ("urn:album:%s:%s", vd.album, vd.album_artist); } else { uri = tracker_sparql_escape_uri_printf ("urn:album:%s", vd.album); } gchar *album_disc_uri; tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nmm:MusicAlbum"); /* FIXME: nmm:albumTitle is now deprecated * tracker_sparql_builder_predicate (preupdate, "nie:title"); */ tracker_sparql_builder_predicate (preupdate, "nmm:albumTitle"); tracker_sparql_builder_object_unvalidated (preupdate, vd.album); if (md.creator_uri) { tracker_sparql_builder_predicate (preupdate, "nmm:albumArtist"); tracker_sparql_builder_object_iri (preupdate, md.creator_uri); } if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); if (vd.track_count) { tracker_sparql_builder_delete_open (preupdate, NULL); tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumTrackCount"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_delete_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumTrackCount"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_where_close (preupdate); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumTrackCount"); tracker_sparql_builder_object_unvalidated (preupdate, vd.track_count); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); } if (vd.album_gain) { tracker_sparql_builder_delete_open (preupdate, NULL); tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumGain"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_delete_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumGain"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_where_close (preupdate); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumGain"); tracker_sparql_builder_object_double (preupdate, atof (vd.album_gain)); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); } if (vd.album_peak_gain) { tracker_sparql_builder_delete_open (preupdate, NULL); tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumPeakGain"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_delete_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumPeakGain"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_where_close (preupdate); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumPeakGain"); tracker_sparql_builder_object_double (preupdate, atof (vd.album_peak_gain)); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); } if (vd.album_artist) { album_disc_uri = tracker_sparql_escape_uri_printf ("urn:album-disc:%s:%s:Disc%d", vd.album, vd.album_artist, vd.disc_number ? atoi(vd.disc_number) : 1); } else { album_disc_uri = tracker_sparql_escape_uri_printf ("urn:album-disc:%s:Disc%d", vd.album, vd.disc_number ? atoi(vd.disc_number) : 1); } tracker_sparql_builder_delete_open (preupdate, NULL); tracker_sparql_builder_subject_iri (preupdate, album_disc_uri); tracker_sparql_builder_predicate (preupdate, "nmm:setNumber"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_delete_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject_iri (preupdate, album_disc_uri); tracker_sparql_builder_predicate (preupdate, "nmm:setNumber"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_where_close (preupdate); tracker_sparql_builder_delete_open (preupdate, NULL); tracker_sparql_builder_subject_iri (preupdate, album_disc_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumDiscAlbum"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_delete_close (preupdate); tracker_sparql_builder_where_open (preupdate); tracker_sparql_builder_subject_iri (preupdate, album_disc_uri); tracker_sparql_builder_predicate (preupdate, "nmm:albumDiscAlbum"); tracker_sparql_builder_object_variable (preupdate, "unknown"); tracker_sparql_builder_where_close (preupdate); tracker_sparql_builder_insert_open (preupdate, NULL); if (graph) { tracker_sparql_builder_graph_open (preupdate, graph); } tracker_sparql_builder_subject_iri (preupdate, album_disc_uri); tracker_sparql_builder_predicate (preupdate, "a"); tracker_sparql_builder_object (preupdate, "nmm:MusicAlbumDisc"); tracker_sparql_builder_predicate (preupdate, "nmm:setNumber"); tracker_sparql_builder_object_int64 (preupdate, vd.disc_number ? atoi (vd.disc_number) : 1); tracker_sparql_builder_predicate (preupdate, "nmm:albumDiscAlbum"); tracker_sparql_builder_object_iri (preupdate, uri); if (graph) { tracker_sparql_builder_graph_close (preupdate); } tracker_sparql_builder_insert_close (preupdate); tracker_sparql_builder_predicate (metadata, "nmm:musicAlbumDisc"); tracker_sparql_builder_object_iri (metadata, album_disc_uri); g_free (album_disc_uri); g_free (vd.album); tracker_sparql_builder_predicate (metadata, "nmm:musicAlbum"); tracker_sparql_builder_object_iri (metadata, uri); g_free (uri); } g_free (vd.track_count); g_free (vd.album_peak_gain); g_free (vd.album_gain); g_free (vd.disc_number); if (vd.title) { tracker_sparql_builder_predicate (metadata, "nie:title"); tracker_sparql_builder_object_unvalidated (metadata, vd.title); g_free (vd.title); } if (vd.track_number) { tracker_sparql_builder_predicate (metadata, "nmm:trackNumber"); tracker_sparql_builder_object_unvalidated (metadata, vd.track_number); g_free (vd.track_number); } if (vd.track_gain) { /* TODO */ g_free (vd.track_gain); } if (vd.track_peak_gain) { /* TODO */ g_free (vd.track_peak_gain); } if (vd.comment) { tracker_sparql_builder_predicate (metadata, "nie:comment"); tracker_sparql_builder_object_unvalidated (metadata, vd.comment); g_free (vd.comment); } if (vd.date) { tracker_sparql_builder_predicate (metadata, "nie:contentCreated"); tracker_sparql_builder_object_unvalidated (metadata, vd.date); g_free (vd.date); } if (vd.genre) { tracker_sparql_builder_predicate (metadata, "nfo:genre"); tracker_sparql_builder_object_unvalidated (metadata, vd.genre); g_free (vd.genre); } if (vd.codec) { tracker_sparql_builder_predicate (metadata, "nfo:codec"); tracker_sparql_builder_object_unvalidated (metadata, vd.codec); g_free (vd.codec); } if (vd.codec_version) { /* TODO */ g_free (vd.codec_version); } if (vd.sample_rate) { tracker_sparql_builder_predicate (metadata, "nfo:sampleRate"); tracker_sparql_builder_object_unvalidated (metadata, vd.sample_rate); g_free (vd.sample_rate); } if (vd.channels) { tracker_sparql_builder_predicate (metadata, "nfo:channels"); tracker_sparql_builder_object_unvalidated (metadata, vd.channels); g_free (vd.channels); } if (vd.mb_album_id) { /* TODO */ g_free (vd.mb_album_id); } if (vd.mb_artist_id) { /* TODO */ g_free (vd.mb_artist_id); } if (vd.mb_album_artist_id) { /* TODO */ g_free (vd.mb_album_artist_id); } if (vd.mb_track_id) { /* TODO */ g_free (vd.mb_track_id); } if (vd.lyrics) { tracker_sparql_builder_predicate (metadata, "nie:plainTextContent"); tracker_sparql_builder_object_unvalidated (metadata, vd.lyrics); g_free (vd.lyrics); } if (vd.copyright) { tracker_sparql_builder_predicate (metadata, "nie:copyright"); tracker_sparql_builder_object_unvalidated (metadata, vd.copyright); g_free (vd.copyright); } if (vd.license) { tracker_sparql_builder_predicate (metadata, "nie:license"); tracker_sparql_builder_object_unvalidated (metadata, vd.license); g_free (vd.license); } if (vd.organization) { /* TODO */ g_free (vd.organization); } if (vd.location) { /* TODO */ g_free (vd.location); } if (vd.publisher) { tracker_sparql_builder_predicate (metadata, "dc:publisher"); tracker_sparql_builder_object_blank_open (metadata); tracker_sparql_builder_predicate (metadata, "a"); tracker_sparql_builder_object (metadata, "nco:Contact"); tracker_sparql_builder_predicate (metadata, "nco:fullname"); tracker_sparql_builder_object_unvalidated (metadata, vd.publisher); tracker_sparql_builder_object_blank_close (metadata); g_free (vd.publisher); } if ((vi = ov_info (&vf, 0)) != NULL ) { bitrate = vi->bitrate_nominal / 1000; tracker_sparql_builder_predicate (metadata, "nfo:averageBitrate"); tracker_sparql_builder_object_int64 (metadata, (gint64) bitrate); } /* Duration */ if ((time = ov_time_total (&vf, -1)) != OV_EINVAL) { tracker_sparql_builder_predicate (metadata, "nfo:duration"); tracker_sparql_builder_object_int64 (metadata, (gint64) time); } g_free (vd.artist); g_free (vd.album_artist); g_free (vd.performer); g_free (md.creator_uri); #ifdef HAVE_POSIX_FADVISE posix_fadvise (fileno (f), 0, 0, POSIX_FADV_DONTNEED); #endif /* HAVE_POSIX_FADVISE */ /* NOTE: This calls fclose on the file */ ov_clear (&vf); return TRUE; }