Beispiel #1
0
static void
write_pdf_data (PDFData               data,
                TrackerSparqlBuilder *metadata,
                GPtrArray            *keywords)
{
	if (!tracker_is_empty_string (data.title)) {
		tracker_sparql_builder_predicate (metadata, "nie:title");
		tracker_sparql_builder_object_unvalidated (metadata, data.title);
	}

	if (!tracker_is_empty_string (data.subject)) {
		tracker_sparql_builder_predicate (metadata, "nie:subject");
		tracker_sparql_builder_object_unvalidated (metadata, data.subject);
	}

	if (!tracker_is_empty_string (data.author)) {
		tracker_sparql_builder_predicate (metadata, "nco:creator");
		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, data.author);
		tracker_sparql_builder_object_blank_close (metadata);
	}

	if (!tracker_is_empty_string (data.date)) {
		tracker_sparql_builder_predicate (metadata, "nie:contentCreated");
		tracker_sparql_builder_object_unvalidated (metadata, data.date);
	}

	if (!tracker_is_empty_string (data.keywords)) {
		tracker_keywords_parse (keywords, data.keywords);
	}
}
Beispiel #2
0
/**
 * tracker_guarantee_date_from_file_mtime:
 * @metadata: the metadata object to insert the data into
 * @key: the key to insert into @metadata
 * @current_value: the current data to check before looking at @uri
 * @uri: a string representing a URI to use
 *
 * Checks @current_value to make sure it is sane (i.e. not %NULL or an
 * empty string). If it is, then @uri is parsed to guarantee a
 * metadata value for @key.
 *
 * When parsing @uri, stat() is called on the file to create a
 * date based on the file's mtime.
 *
 * Returns: %TRUE on success, otherwise %FALSE.
 *
 * Since: 0.10
 **/
gboolean
tracker_guarantee_date_from_file_mtime (TrackerSparqlBuilder *metadata,
                                        const gchar          *key,
                                        const gchar          *current_value,
                                        const gchar          *uri)
{
#ifdef GUARANTEE_METADATA
	g_return_val_if_fail (metadata != NULL, FALSE);
	g_return_val_if_fail (key != NULL, FALSE);
	g_return_val_if_fail (uri != NULL, FALSE);

	tracker_sparql_builder_predicate (metadata, key);

	if (current_value && *current_value != '\0') {
		tracker_sparql_builder_object_unvalidated (metadata, current_value);
	} else {
		gchar *value;

		value = get_date_from_file_mtime (uri);
		tracker_sparql_builder_object_unvalidated (metadata, value);
		g_free (value);
	}
#else  /* GUARANTEE_METADATA */
	if (current_value && *current_value != '\0') {
		tracker_sparql_builder_predicate (metadata, key);
		tracker_sparql_builder_object_unvalidated (metadata, current_value);
	}
#endif /* GUARANTEE_METADATA */

	return TRUE;
}
Beispiel #3
0
/**
 * tracker_guarantee_title_from_file:
 * @metadata: the metadata object to insert the data into
 * @key: the key to insert into @metadata
 * @current_value: the current data to check before looking at @uri.
 * @uri: a string representing a URI to use
 * @p_new_value: pointer to a string which receives the new title, or
 *             %NULL
 *
 * Checks @current_value to make sure it is sane (i.e. not %NULL or an
 * empty string). If it is, then @uri is parsed to guarantee a
 * metadata value for @key.
 *
 * Parses the file pointed to by @uri and uses the basename
 * (before the "." and extension of the file) as the title. If the
 * title has any "_" characters, they are also converted into spaces.
 *
 * Returns: %TRUE on success and content was added to @metadata, otherwise %FALSE.
 *
 * Since: 0.10
 **/
gboolean
tracker_guarantee_title_from_file (TrackerSparqlBuilder  *metadata,
                                   const gchar           *key,
                                   const gchar           *current_value,
                                   const gchar           *uri,
                                   gchar                **p_new_value)
{
	gboolean success = TRUE;

#ifdef GUARANTEE_METADATA
	g_return_val_if_fail (metadata != NULL, FALSE);
	g_return_val_if_fail (key != NULL, FALSE);
	g_return_val_if_fail (uri != NULL, FALSE);

	tracker_sparql_builder_predicate (metadata, key);

	if (current_value && *current_value != '\0') {
		tracker_sparql_builder_object_unvalidated (metadata, current_value);

		if (p_new_value != NULL) {
			*p_new_value = g_strdup (current_value);
		}
	} else {
		gchar *value;

		value = get_title_from_file (uri);

		if (value && value[0] != '\0') {
			tracker_sparql_builder_object_unvalidated (metadata, value);
		} else {
			success = FALSE;
		}

		if (p_new_value != NULL) {
			*p_new_value = value;
		} else {
			g_free (value);
		}
	}
#else  /* GUARANTEE_METADATA */
	if (current_value && *current_value != '\0') {
		tracker_sparql_builder_predicate (metadata, key);
		tracker_sparql_builder_object_unvalidated (metadata, current_value);

		if (p_new_value != NULL) {
			*p_new_value = g_strdup (current_value);
		}
	} else {
		success = FALSE;
	}
#endif /* GUARANTEE_METADATA */

	return success;
}
Beispiel #4
0
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;
}
static void
set_value_string (TrackerSparqlBuilder *metadata,
                  const gchar          *predicate,
                  const gchar          *value)
{
	tracker_sparql_builder_predicate (metadata, predicate);
	tracker_sparql_builder_object_unvalidated (metadata, value);
}
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;
}
Beispiel #7
0
static void
add_tuple (TrackerSparqlBuilder *metadata,
           const char           *predicate,
           const char           *object)
{
	if (object) {
		tracker_sparql_builder_predicate (metadata, predicate);
		tracker_sparql_builder_object_unvalidated (metadata, object);
	}
}
Beispiel #8
0
static gboolean
extract_opf (const gchar          *uri,
             const gchar          *opf_path,
             TrackerExtractInfo   *info)
{
	GMarkupParseContext *context;
	OPFData *data = NULL;
	GError *error = NULL;
	gchar *dirname, *contents;
	GMarkupParser opf_parser = {
		opf_xml_start_element_handler,
		opf_xml_end_element_handler,
		opf_xml_text_handler,
		NULL, NULL
	};

	g_debug ("Extracting OPF file contents from EPUB '%s'", uri);

	data = opf_data_new (info);

	tracker_sparql_builder_predicate (data->metadata, "a");
	tracker_sparql_builder_object (data->metadata, "nfo:EBook");

	/* Create parsing context */
	context = g_markup_parse_context_new (&opf_parser, 0, data, NULL);

	/* Load the internal container file from the Zip archive,
	 * and parse it to extract the .opf file to get metadata from
	 */
	tracker_gsf_parse_xml_in_zip (uri, opf_path, context, &error);
	g_markup_parse_context_free (context);

	if (error) {
		g_warning ("Could not get EPUB '%s' file: %s\n", opf_path,
		           (error) ? error->message : "No error provided");
		g_error_free (error);
		opf_data_free (data);
		return FALSE;
	}

	dirname = g_path_get_dirname (opf_path);
	contents = extract_opf_contents (uri, dirname, data->pages);
	g_free (dirname);

	if (contents && *contents) {
		tracker_sparql_builder_predicate (data->metadata, "nie:plainTextContent");
		tracker_sparql_builder_object_unvalidated (data->metadata, contents);
	}

	opf_data_free (data);
	g_free (contents);

	return TRUE;
}
static void
entry_parsed (TotemPlParser *parser,
              gchar         *to_uri,
              GHashTable    *to_metadata,
              gpointer       user_data)
{
	PlaylistMetadata *data;

	data = (PlaylistMetadata *) user_data;
	data->track_counter++;

	if (data->track_counter > 1000) {
		/* limit playlists to 1000 entries for query performance reasons */
		g_message ("Playlist has > 1000 entries. Ignoring for performance reasons.");
		return;
	}

	if (data->track_counter == 1) {
		/* first track, predicate needed */
		tracker_sparql_builder_predicate (data->metadata, "nfo:hasMediaFileListEntry");
	}

	tracker_sparql_builder_object_blank_open (data->metadata);
	tracker_sparql_builder_predicate (data->metadata, "a");
	tracker_sparql_builder_object (data->metadata, "nfo:MediaFileListEntry");

	tracker_sparql_builder_predicate (data->metadata, "nfo:entryUrl");
	tracker_sparql_builder_object_unvalidated (data->metadata, to_uri);

	tracker_sparql_builder_predicate (data->metadata, "nfo:listPosition");
	tracker_sparql_builder_object_int64 (data->metadata, (gint64) data->track_counter);

	tracker_sparql_builder_object_blank_close (data->metadata);

	if (to_metadata) {
		gchar *duration;

		duration = g_hash_table_lookup (to_metadata, TOTEM_PL_PARSER_FIELD_DURATION);

		if (duration == NULL) {
			duration = g_hash_table_lookup (to_metadata, TOTEM_PL_PARSER_FIELD_DURATION_MS);
		}

		if (duration != NULL) {
			gint64 secs = totem_pl_parser_parse_duration (duration, FALSE);

			if (secs > 0) {
				data->total_time += secs;
			}
		}
	}
}
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;
}
Beispiel #11
0
static void
read_outline (PopplerDocument      *document,
              TrackerSparqlBuilder *metadata)
{
	PopplerIndexIter *index;
	GString *toc = NULL;

	index = poppler_index_iter_new (document);

	if (!index) {
		return;
	}

	read_toc (index, &toc);

	if (toc) {
		if (toc->len > 0) {
			tracker_sparql_builder_predicate (metadata, "nfo:tableOfContents");
			tracker_sparql_builder_object_unvalidated (metadata, toc->str);
		}

		g_string_free (toc, TRUE);
	}
}
Beispiel #12
0
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;
}
Beispiel #13
0
static void
opf_xml_text_handler (GMarkupParseContext   *context,
                      const gchar           *text,
                      gsize                  text_len,
                      gpointer               user_data,
                      GError               **error)
{
	OPFData *data = user_data;

	switch (data->element) {
	case OPF_TAG_TYPE_PUBLISHER:
		if (data->has_publisher) {
			g_warning ("Avoiding additional publisher (%s) in EPUB '%s'",
			           text, data->uri);
		} else {
			data->has_publisher = TRUE;
			tracker_sparql_builder_predicate (data->metadata, "nco:publisher");

			tracker_sparql_builder_object_blank_open (data->metadata);
			tracker_sparql_builder_predicate (data->metadata, "a");
			tracker_sparql_builder_object (data->metadata, "nco:Contact");

			tracker_sparql_builder_predicate (data->metadata, "nco:fullname");
			tracker_sparql_builder_object_unvalidated (data->metadata, text);
			tracker_sparql_builder_object_blank_close (data->metadata);
		}
		break;
	case OPF_TAG_TYPE_AUTHOR:
	case OPF_TAG_TYPE_EDITOR:
	case OPF_TAG_TYPE_ILLUSTRATOR:
	case OPF_TAG_TYPE_CONTRIBUTOR: {
		gchar *fname, *gname, *oname;
		const gchar *fullname = NULL;
		gchar *role_uri = NULL;
		const gchar *role_str = NULL;
		gint i, j = 0, len;

		fname = NULL;
		gname = NULL;
		oname = NULL;

		/* parse name.  may not work for dissimilar cultures. */
		if (data->savedstring != NULL) {
			fullname = data->savedstring;

			/* <family name>, <given name> <other name> */
			g_debug ("Parsing 'opf:file-as' attribute:'%s'", data->savedstring);
			len = strlen (data->savedstring);

			for (i = 0; i < len; i++) {
				if (data->savedstring[i] == ',') {
					fname = g_strndup (data->savedstring, i);
					g_debug ("Found family name:'%s'", fname);

					for (; data->savedstring[i] == ',' || data->savedstring[i] == ' '; i++);
					j = i;

					break;
				}
			}

			if (i == len) {
				fname = g_strdup (data->savedstring);
				g_debug ("Found only one name");
			} else {
				for (; i <= len; i++) {
					if (i == len || data->savedstring[i] == ' ') {
						gname = g_strndup (data->savedstring + j, i - j);
						g_debug ("Found given name:'%s'", gname);

						for (; data->savedstring[i] == ',' || data->savedstring[i] == ' '; i++);

						if (i != len) {
							oname = g_strdup (data->savedstring + i);
							g_debug ("Found other name:'%s'", oname);
						}

						break;
					}
				}
			}
		} else {
			fullname = text;

			/* <given name> <other name> <family name> */
			g_debug ("Parsing name, no 'opf:file-as' found: '%s'", text);

			len = strlen (text);

			for (i = 0; i < len; i++) {
				if (text[i] == ' ') {
					gname = g_strndup (text, i);
					g_debug ("Found given name:'%s'", gname);
					j = i + 1;

					break;
				}
			}

			if (j == 0) {
				fname = g_strdup (data->savedstring);
				g_debug ("Found only one name:'%s'", fname);
			} else {
				for (i = len - 1; i >= j - 1; i--) {
					if (text[i] == ' ') {
						fname = g_strdup (text + i + 1);
						g_debug ("Found family name:'%s'", fname);

						if (i > j) {
							oname = g_strndup (text + j, i - j);
							g_debug ("Found other name:'%s'", oname);
						}

						break;
					}
				}
			}
		}

		/* Role details */
		role_uri = tracker_sparql_escape_uri_printf ("urn:artist:%s", fullname);

		if (data->element == OPF_TAG_TYPE_AUTHOR) {
			role_str = "nco:creator";
		} else if (data->element == OPF_TAG_TYPE_EDITOR && !data->has_publisher) {
			/* Should this be nco:contributor ?
			 * 'Editor' is a bit vague here.
			 */
			role_str = "nco:publisher";
		} else if (data->element == OPF_TAG_TYPE_ILLUSTRATOR) {
			/* There is no illustrator class, using contributor */
			role_str = "nco:contributor";
		} else {
			g_assert ("Unknown role");
		}

		if (role_uri) {
			tracker_sparql_builder_insert_open (data->preupdate, NULL);
			if (data->graph) {
				tracker_sparql_builder_graph_open (data->preupdate, data->graph);
			}

			tracker_sparql_builder_subject_iri (data->preupdate, role_uri);
			tracker_sparql_builder_predicate (data->preupdate, "a");
			tracker_sparql_builder_object (data->preupdate, "nmm:Artist");
			tracker_sparql_builder_predicate (data->preupdate, "nmm:artistName");
			tracker_sparql_builder_object_unvalidated (data->preupdate, fullname);

			if (data->graph) {
				tracker_sparql_builder_graph_close (data->preupdate);
			}
			tracker_sparql_builder_insert_close (data->preupdate);
		}

		/* Creator contact details */
		tracker_sparql_builder_predicate (data->metadata, "nco:creator");
		tracker_sparql_builder_object_blank_open (data->metadata);
		tracker_sparql_builder_predicate (data->metadata, "a");
		tracker_sparql_builder_object (data->metadata, "nco:PersonContact");
		tracker_sparql_builder_predicate (data->metadata, "nco:fullname");
		tracker_sparql_builder_object_unvalidated (data->metadata, fullname);

		if (fname) {
			tracker_sparql_builder_predicate (data->metadata, "nco:nameFamily");
			tracker_sparql_builder_object_unvalidated (data->metadata, fname);
			g_free (fname);
		}

		if (gname) {
			tracker_sparql_builder_predicate (data->metadata, "nco:nameGiven");
			tracker_sparql_builder_object_unvalidated (data->metadata, gname);
			g_free (gname);
		}

		if (oname) {
			tracker_sparql_builder_predicate (data->metadata, "nco:nameAdditional");
			tracker_sparql_builder_object_unvalidated (data->metadata, oname);
			g_free (oname);
		}

		if (role_uri) {
			tracker_sparql_builder_predicate (data->metadata, role_str);
			tracker_sparql_builder_object_iri (data->metadata, role_uri);
			g_free (role_uri);
		}

		tracker_sparql_builder_object_blank_close (data->metadata);

		break;
	}
	case OPF_TAG_TYPE_TITLE:
		if (data->has_title) {
			g_warning ("Avoiding additional title (%s) in EPUB '%s'",
			           text, data->uri);
		} else {
			data->has_title = TRUE;
			tracker_sparql_builder_predicate (data->metadata, "nie:title");
			tracker_sparql_builder_object_unvalidated (data->metadata, text);
		}
		break;
	case OPF_TAG_TYPE_CREATED: {
		if (data->has_content_created) {
			g_warning ("Avoiding additional creation time (%s) in EPUB '%s'",
			           text, data->uri);
		} else {
			gchar *date = tracker_date_guess (text);

			if (date) {
				data->has_content_created = TRUE;
				tracker_sparql_builder_predicate (data->metadata, "nie:contentCreated");
				tracker_sparql_builder_object_unvalidated (data->metadata, date);
				g_free (date);
			} else {
				g_warning ("Could not parse creation time (%s) in EPUB '%s'",
				           text, data->uri);
			}
		}
		break;
	}
	case OPF_TAG_TYPE_LANGUAGE:
		if (data->has_language) {
			g_warning ("Avoiding additional language (%s) in EPUB '%s'",
			           text, data->uri);
		} else {
			data->has_language = TRUE;
			tracker_sparql_builder_predicate (data->metadata, "nie:language");
			tracker_sparql_builder_object_unvalidated (data->metadata, text);
		}
		break;
	case OPF_TAG_TYPE_SUBJECT:
		if (data->has_subject) {
			g_warning ("Avoiding additional subject (%s) in EPUB '%s'",
			           text, data->uri);
		} else {
			data->has_subject = TRUE;
			tracker_sparql_builder_predicate (data->metadata, "nie:subject");
			tracker_sparql_builder_object_unvalidated (data->metadata, text);
		}
		break;
	case OPF_TAG_TYPE_DESCRIPTION:
		if (data->has_description) {
			g_warning ("Avoiding additional description (%s) in EPUB '%s'",
			           text, data->uri);
		} else {
			data->has_description = TRUE;
			tracker_sparql_builder_predicate (data->metadata, "nie:description");
			tracker_sparql_builder_object_unvalidated (data->metadata, text);
		}
		break;
	case OPF_TAG_TYPE_UUID:
	case OPF_TAG_TYPE_ISBN:
		if (data->has_identifier) {
			g_warning ("Avoiding additional identifier (%s) in EPUB '%s'",
			           text, data->uri);
		} else {
			data->has_identifier = TRUE;
			tracker_sparql_builder_predicate (data->metadata, "nie:identifier");
			tracker_sparql_builder_object_unvalidated (data->metadata, text);
		}
		break;
	/* case OPF_TAG_TYPE_RATING: */
	case OPF_TAG_TYPE_UNKNOWN:
	default:
		break;
	}

	opf_data_clear_saved_string (data);
}
Beispiel #14
0
static void
read_metadata (TrackerSparqlBuilder *preupdate,
               TrackerSparqlBuilder *metadata,
               GString              *where,
               png_structp           png_ptr,
               png_infop             info_ptr,
               png_infop             end_ptr,
               const gchar          *uri,
               const gchar          *graph)
{
	MergeData md = { 0 };
	PngData pd = { 0 };
	TrackerExifData *ed = NULL;
	TrackerXmpData *xd = NULL;
	png_infop info_ptrs[2];
	png_textp text_ptr;
	gint info_index;
	gint num_text;
	gint i;
	gint found;
	GPtrArray *keywords;

	info_ptrs[0] = info_ptr;
	info_ptrs[1] = end_ptr;

	for (info_index = 0; info_index < 2; info_index++) {
		if ((found = png_get_text (png_ptr, info_ptrs[info_index], &text_ptr, &num_text)) < 1) {
			g_debug ("Calling png_get_text() returned %d (< 1)", found);
			continue;
		}

		for (i = 0; i < num_text; i++) {
			if (!text_ptr[i].key || !text_ptr[i].text || text_ptr[i].text[0] == '\0') {
				continue;
			}

#if defined(HAVE_EXEMPI) && defined(PNG_iTXt_SUPPORTED)
			if (g_strcmp0 ("XML:com.adobe.xmp", text_ptr[i].key) == 0) {
				/* ATM tracker_extract_xmp_read supports setting xd
				 * multiple times, keep it that way as here it's
				 * theoretically possible that the function gets
				 * called multiple times
				 */
				xd = tracker_xmp_new (text_ptr[i].text,
				                      text_ptr[i].itxt_length,
				                      uri);

				continue;
			}

			if (g_strcmp0 ("Raw profile type xmp", text_ptr[i].key) == 0) {
				gchar *xmp_buffer;
				guint xmp_buffer_length = 0;
				guint input_len;

				if (text_ptr[i].text_length) {
					input_len = text_ptr[i].text_length;
				} else {
					input_len = text_ptr[i].itxt_length;
				}

				xmp_buffer = raw_profile_new (text_ptr[i].text,
				                              input_len,
				                              &xmp_buffer_length);

				if (xmp_buffer) {
					xd = tracker_xmp_new (xmp_buffer,
					                      xmp_buffer_length,
					                      uri);
				}

				g_free (xmp_buffer);

				continue;
			}
#endif /*HAVE_EXEMPI && PNG_iTXt_SUPPORTED */

#if defined(HAVE_LIBEXIF) && defined(PNG_iTXt_SUPPORTED)
			if (g_strcmp0 ("Raw profile type exif", text_ptr[i].key) == 0) {
				gchar *exif_buffer;
				guint exif_buffer_length = 0;
				guint input_len;

				if (text_ptr[i].text_length) {
					input_len = text_ptr[i].text_length;
				} else {
					input_len = text_ptr[i].itxt_length;
				}

				exif_buffer = raw_profile_new (text_ptr[i].text,
				                               input_len,
				                               &exif_buffer_length);

				if (exif_buffer) {
					ed = tracker_exif_new (exif_buffer,
					                       exif_buffer_length,
					                       uri);
				}

				g_free (exif_buffer);

				continue;
			}
#endif /* HAVE_LIBEXIF && PNG_iTXt_SUPPORTED */

			if (g_strcmp0 (text_ptr[i].key, "Author") == 0) {
				pd.author = text_ptr[i].text;
				continue;
			}

			if (g_strcmp0 (text_ptr[i].key, "Creator") == 0) {
				pd.creator = text_ptr[i].text;
				continue;
			}

			if (g_strcmp0 (text_ptr[i].key, "Description") == 0) {
				pd.description = text_ptr[i].text;
				continue;
			}

			if (g_strcmp0 (text_ptr[i].key, "Comment") == 0) {
				pd.comment = text_ptr[i].text;
				continue;
			}

			if (g_strcmp0 (text_ptr[i].key, "Copyright") == 0) {
				pd.copyright = text_ptr[i].text;
				continue;
			}

			if (g_strcmp0 (text_ptr[i].key, "Creation Time") == 0) {
				pd.creation_time = rfc1123_to_iso8601_date (text_ptr[i].text);
				continue;
			}

			if (g_strcmp0 (text_ptr[i].key, "Title") == 0) {
				pd.title = text_ptr[i].text;
				continue;
			}

			if (g_strcmp0 (text_ptr[i].key, "Disclaimer") == 0) {
				pd.disclaimer = text_ptr[i].text;
				continue;
			}

			if (g_strcmp0(text_ptr[i].key, "Software") == 0) {
				pd.software = text_ptr[i].text;
				continue;
			}
		}
	}

	if (!ed) {
		ed = g_new0 (TrackerExifData, 1);
	}

	if (!xd) {
		xd = g_new0 (TrackerXmpData, 1);
	}

	md.creator = tracker_coalesce_strip (3, xd->creator, pd.creator, pd.author);
	md.title = tracker_coalesce_strip (5, xd->title, pd.title, ed->document_name, xd->title2, xd->pdf_title);
	md.copyright = tracker_coalesce_strip (3, xd->rights, pd.copyright, ed->copyright);
	md.license = tracker_coalesce_strip (2, xd->license, pd.disclaimer);
	md.description = tracker_coalesce_strip (3, xd->description, pd.description, ed->description);
	md.date = tracker_coalesce_strip (5, xd->date, xd->time_original, pd.creation_time, ed->time, ed->time_original);
	md.comment = tracker_coalesce_strip (2, pd.comment, ed->user_comment);
	md.artist = tracker_coalesce_strip (3, xd->artist, ed->artist, xd->contributor);
	md.orientation = tracker_coalesce_strip (2, xd->orientation, ed->orientation);
	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.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.metering_mode = tracker_coalesce_strip (2, xd->metering_mode, ed->metering_mode);
	md.white_balance = tracker_coalesce_strip (2, xd->white_balance, ed->white_balance);
	md.make = tracker_coalesce_strip (2, xd->make, ed->make);
	md.model = tracker_coalesce_strip (2, xd->model, ed->model);

	keywords = g_ptr_array_new ();

	if (md.comment) {
		tracker_sparql_builder_predicate (metadata, "nie:comment");
		tracker_sparql_builder_object_unvalidated (metadata, md.comment);
	}

	if (md.license) {
		tracker_sparql_builder_predicate (metadata, "nie:license");
		tracker_sparql_builder_object_unvalidated (metadata, md.license);
	}

	/* TODO: add ontology and store this ed->software */

	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);

		tracker_sparql_builder_predicate (metadata, "nco:creator");
		tracker_sparql_builder_object_iri (metadata, uri);
		g_free (uri);
	}

	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.copyright) {
		tracker_sparql_builder_predicate (metadata, "nie:copyright");
		tracker_sparql_builder_object_unvalidated (metadata, md.copyright);
	}

	tracker_guarantee_title_from_file (metadata,
	                                   "nie:title",
	                                   md.title,
	                                   uri,
	                                   NULL);

	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);
	}

	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.orientation) {
		tracker_sparql_builder_predicate (metadata, "nfo:orientation");
		tracker_sparql_builder_object_unvalidated (metadata, md.orientation);
	}

	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);
	}

	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.metering_mode) {
		tracker_sparql_builder_predicate (metadata, "nmm:meteringMode");
		tracker_sparql_builder_object_unvalidated (metadata, md.metering_mode);
	}


	if (xd->keywords) {
		tracker_keywords_parse (keywords, xd->keywords);
	}

	if (xd->pdf_keywords) {
		tracker_keywords_parse (keywords, xd->pdf_keywords);
	}

	if (xd->rating) {
		tracker_sparql_builder_predicate (metadata, "nao:numericRating");
		tracker_sparql_builder_object_unvalidated (metadata, xd->rating);
	}

	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:creator");
		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->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->gps_direction) {
		tracker_sparql_builder_predicate (metadata, "nfo:heading");
		tracker_sparql_builder_object_unvalidated (metadata, xd->gps_direction);
	}

	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);
	}

	if (xd->regions) {
		tracker_xmp_apply_regions (preupdate, metadata, graph, xd);
	}

	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);

	if (g_strcmp0(pd.software, "gnome-screenshot") == 0) {
		tracker_sparql_builder_predicate (metadata, "nie:isPartOf");
		tracker_sparql_builder_object (metadata, "nfo:image-category-screenshot");
	}

	tracker_exif_free (ed);
	tracker_xmp_free (xd);
	g_free (pd.creation_time);
}
Beispiel #15
0
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;
}
Beispiel #16
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;
}
static inline void
process_item (ProcessUserguideData  *data,
              GFileInfo             *file_info,
              gboolean               is_dir,
              GError               **error)
{
	TrackerSparqlBuilder *sparql;
	gchar *uri;
	const gchar *mime_type;
	const gchar *urn;
	const gchar *parent_urn;
	gboolean is_iri;
	guint64 time_;

	sparql = data->sparql;

	uri = g_file_get_uri (data->file);
	mime_type = g_file_info_get_content_type (file_info);

	urn = get_file_urn (data->miner, data->file, &is_iri);

	tracker_sparql_builder_insert_silent_open (sparql, NULL);
	tracker_sparql_builder_graph_open (sparql, TRACKER_MINER_FS_GRAPH_URN);

	if (is_iri) {
		tracker_sparql_builder_subject_iri (sparql, urn);
	} else {
		tracker_sparql_builder_subject (sparql, urn);
	}

	tracker_sparql_builder_predicate (sparql, "a");
	tracker_sparql_builder_object (sparql, "nfo:FileDataObject");
	tracker_sparql_builder_object (sparql, "nie:InformationElement");

	if (is_dir) {
		tracker_sparql_builder_object (sparql, "nfo:Folder");
	} else {
		tracker_sparql_builder_object (sparql, "nfo:HelpDocument");
	}

	parent_urn = tracker_miner_fs_get_parent_urn (TRACKER_MINER_FS (data->miner), data->file);

	if (parent_urn) {
		tracker_sparql_builder_predicate (sparql, "nfo:belongsToContainer");
		tracker_sparql_builder_object_iri (sparql, parent_urn);
	}

	tracker_sparql_builder_predicate (sparql, "nfo:fileName");
	tracker_sparql_builder_object_string (sparql, g_file_info_get_display_name (file_info));

	tracker_sparql_builder_predicate (sparql, "nfo:fileSize");
	tracker_sparql_builder_object_int64 (sparql, g_file_info_get_size (file_info));

	time_ = g_file_info_get_attribute_uint64 (file_info, G_FILE_ATTRIBUTE_TIME_MODIFIED);
	tracker_sparql_builder_predicate (sparql, "nfo:fileLastModified");
	tracker_sparql_builder_object_date (sparql, (time_t *) &time_);

	time_ = g_file_info_get_attribute_uint64 (file_info, G_FILE_ATTRIBUTE_TIME_ACCESS);
	tracker_sparql_builder_predicate (sparql, "nfo:fileLastAccessed");
	tracker_sparql_builder_object_date (sparql, (time_t *) &time_);

	/* Laying the link between the IE and the DO. We use IE = DO */
	tracker_sparql_builder_predicate (sparql, "nie:isStoredAs");
	if (is_iri) {
		tracker_sparql_builder_object_iri (sparql, urn);
	} else {
		tracker_sparql_builder_object (sparql, urn);
	}

	/* The URL of the DataObject (because IE = DO, this is correct) */
	tracker_sparql_builder_predicate (sparql, "nie:url");
	tracker_sparql_builder_object_string (sparql, uri);

	tracker_sparql_builder_predicate (sparql, "nie:mimeType");
	tracker_sparql_builder_object_string (sparql, mime_type);

	/* FIXME: Add nie:dataSource for switching different userguides? */
	tracker_sparql_builder_predicate (sparql, "tracker:available");
	tracker_sparql_builder_object_boolean (sparql, TRUE);

	if (!is_dir) {
		gchar *content = NULL;
		gchar *title = NULL;

		/* Get content */
		parser_get_file_content (uri, MAX_EXTRACT_SIZE, &content, &title);

		g_message ("Adding userguide:'%s', uri:'%s'",
		           title,
		           uri);

		if (title && title[0]) {
			tracker_sparql_builder_predicate (sparql, "nie:title");
			tracker_sparql_builder_object_unvalidated (sparql, title);
		}

		if (content && content[0]) {
			tracker_sparql_builder_predicate (sparql, "nie:plainTextContent");
			tracker_sparql_builder_object_unvalidated (sparql, content);
		}

		g_free (content);
		g_free (title);
	} else {
		g_message ("Adding userguide directory:'%s'", uri);
	}

	tracker_sparql_builder_graph_close (sparql);
	tracker_sparql_builder_insert_close (sparql);

	g_free (uri);
}
Beispiel #18
0
static void
extract_ps_from_filestream (FILE *f,
                            TrackerSparqlBuilder *preupdate,
                            TrackerSparqlBuilder *metadata)
{
	gchar *line;
	gsize length;
	gssize read_char;
	gsize accum;
	gsize max_bytes;

	line = NULL;
	length = 0;

	tracker_sparql_builder_predicate (metadata, "a");
	tracker_sparql_builder_object (metadata, "nfo:PaginatedTextDocument");

	/* 20 MiB should be enough! (original safe limit) */
	accum = 0;
	max_bytes = 20u << 20;

	/* Reuse the same buffer for all lines. Must be dynamically allocated with
	 * malloc family methods as getline() may re-size it with realloc() */
	length = 1024;
	line = g_malloc (length);

	/* Halt the whole when one of these conditions is met:
	 *  a) Reached max bytes to read
	 *  b) No more lines to read
	 */
	while ((accum < max_bytes) &&
	       (read_char = tracker_getline (&line, &length, f)) != -1) {
		gboolean pageno_atend = FALSE;
		gboolean header_finished = FALSE;

		/* Update accumulated bytes read */
		accum += read_char;

		line[read_char - 1] = '\0';  /* overwrite '\n' char */

		if (!header_finished && strncmp (line, "%%Copyright:", 12) == 0) {
			tracker_sparql_builder_predicate (metadata, "nie:copyright");
			tracker_sparql_builder_object_unvalidated (metadata, line + 13);
		} else if (!header_finished && strncmp (line, "%%Title:", 8) == 0) {
			tracker_sparql_builder_predicate (metadata, "nie:title");
			tracker_sparql_builder_object_unvalidated (metadata, line + 9);
		} else if (!header_finished && strncmp (line, "%%Creator:", 10) == 0) {
			tracker_sparql_builder_predicate (metadata, "nco:creator");
			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, line + 11);
			tracker_sparql_builder_object_blank_close (metadata);
		} else if (!header_finished && strncmp (line, "%%CreationDate:", 15) == 0) {
			gchar *date;

			date = date_to_iso8601 (line + 16);
			if (date) {
				tracker_sparql_builder_predicate (metadata, "nie:contentCreated");
				tracker_sparql_builder_object_unvalidated (metadata, date);
				g_free (date);
			}
		} else if (strncmp (line, "%%Pages:", 8) == 0) {
			if (strcmp (line + 9, "(atend)") == 0) {
				pageno_atend = TRUE;
			} else {
				gint64 page_count;

				page_count = g_ascii_strtoll (line + 9, NULL, 10);
				tracker_sparql_builder_predicate (metadata, "nfo:pageCount");
				tracker_sparql_builder_object_int64 (metadata, page_count);
			}
		} else if (strncmp (line, "%%EndComments", 14) == 0) {
			header_finished = TRUE;

			if (!pageno_atend) {
				break;
			}
		}
	}

	/* Deallocate the buffer */
	if (line) {
		g_free (line);
	}
}
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;
}
static void
read_metadata (TrackerSparqlBuilder *preupdate,
               TrackerSparqlBuilder *metadata,
               GString              *where,
               GifFileType          *gifFile,
               const gchar          *uri,
               const gchar          *graph)
{
	GifRecordType RecordType;
	int frameheight;
	int framewidth;
	unsigned char *framedata = NULL;
	GPtrArray *keywords;
	guint i;
	int status;
	MergeData md = { 0 };
	GifData   gd = { 0 };
	TrackerXmpData *xd = NULL;

	do {
		GifByteType *ExtData;
		int ExtCode;
		ExtBlock extBlock;

		if (DGifGetRecordType(gifFile, &RecordType) == GIF_ERROR) {
#if GIFLIB_MAJOR < 5
			PrintGifError ();
#else  /* GIFLIB_MAJOR < 5 */
			gif_error ("Could not read next GIF record type", gifFile->Error);
#endif /* GIFLIB_MAJOR < 5 */
			return;
		}

		switch (RecordType) {
			case IMAGE_DESC_RECORD_TYPE:
			if (DGifGetImageDesc(gifFile) == GIF_ERROR) {
#if GIFLIB_MAJOR < 5
				PrintGifError();
#else  /* GIFLIB_MAJOR < 5 */
				gif_error ("Could not get GIF record information", gifFile->Error);
#endif /* GIFLIB_MAJOR < 5 */
				return;
			}

			framewidth  = gifFile->Image.Width;
			frameheight = gifFile->Image.Height;

			framedata = g_malloc (framewidth*frameheight);

			if (DGifGetLine(gifFile, framedata, framewidth*frameheight)==GIF_ERROR) {
#if GIFLIB_MAJOR < 5
				PrintGifError();
#else  /* GIFLIB_MAJOR < 5 */
				gif_error ("Could not load a block of GIF pixes", gifFile->Error);
#endif /* GIFLIB_MAJOR < 5 */
				return;
			}

			gd.width  = g_strdup_printf ("%d", framewidth);
			gd.height = g_strdup_printf ("%d", frameheight);


			g_free (framedata);

		break;
		case EXTENSION_RECORD_TYPE:
			extBlock.bytes = NULL;
			extBlock.byteCount = 0;

			if ((status = DGifGetExtension (gifFile, &ExtCode, &ExtData)) != GIF_OK) {
				g_warning ("Problem getting the extension");
				return;
			}
#if defined(HAVE_EXEMPI)
			if (ExtData && *ExtData &&
			    strncmp (&ExtData[1],"XMP Data",8) == 0) {
				while (ExtData != NULL && status == GIF_OK ) {
					if ((status = DGifGetExtensionNext (gifFile, &ExtData)) == GIF_OK) {
						if (ExtData != NULL) {
							if (ext_block_append (&extBlock, ExtData[0]+1, (char *) &(ExtData[0])) != GIF_OK) {
								g_warning ("Problem with extension data");
								return;
							}
						}
					}
				}

				xd = tracker_xmp_new (extBlock.bytes,
				                      extBlock.byteCount-XMP_MAGIC_TRAILER_LENGTH,
				                      uri);

				g_free (extBlock.bytes);
			} else
#endif
			/* See Section 24. Comment Extension. in the GIF format definition */
			if (ExtCode == EXTENSION_RECORD_COMMENT_BLOCK_CODE &&
			    ExtData && *ExtData) {
				guint block_count = 0;

				/* Merge all blocks */
				do {
					block_count++;

					g_debug ("Comment Extension block found (#%u, %u bytes)",
					         block_count,
					         ExtData[0]);
					if (ext_block_append (&extBlock, ExtData[0], (char *) &(ExtData[1])) != GIF_OK) {
						g_warning ("Problem with Comment extension data");
						return;
					}
				} while (((status = DGifGetExtensionNext(gifFile, &ExtData)) == GIF_OK) &&
				         ExtData != NULL);

				/* Add last NUL byte */
				g_debug ("Comment Extension blocks found (%u) with %u bytes",
				         block_count,
				         extBlock.byteCount);
				extBlock.bytes = g_realloc (extBlock.bytes, extBlock.byteCount + 1);
				extBlock.bytes[extBlock.byteCount] = '\0';

				/* Set commentt */
				gd.comment = extBlock.bytes;
			} else {
				do {
					status = DGifGetExtensionNext(gifFile, &ExtData);
				} while ( status == GIF_OK && ExtData != NULL);
			}
		break;
		case TERMINATE_RECORD_TYPE:
			break;
		default:
			break;
		}
	} while (RecordType != TERMINATE_RECORD_TYPE);


	if (!xd) {
		xd = g_new0 (TrackerXmpData, 1);
	}

	md.title = tracker_coalesce_strip (3, xd->title, xd->title2, xd->pdf_title);
	md.date = tracker_coalesce_strip (2, xd->date, xd->time_original);
	md.artist = tracker_coalesce_strip (2, xd->artist, xd->contributor);

	if (xd->license) {
		tracker_sparql_builder_predicate (metadata, "nie:license");
		tracker_sparql_builder_object_unvalidated (metadata, xd->license);
	}

	if (xd->creator) {
		gchar *uri = tracker_sparql_escape_uri_printf ("urn:contact:%s", xd->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, xd->creator);

		if (graph) {
			tracker_sparql_builder_graph_close (preupdate);
		}
		tracker_sparql_builder_insert_close (preupdate);

		tracker_sparql_builder_predicate (metadata, "nco:creator");
		tracker_sparql_builder_object_iri (metadata, uri);
		g_free (uri);
	}

	tracker_guarantee_date_from_file_mtime (metadata,
	                                        "nie:contentCreated",
	                                        md.date,
	                                        uri);

	if (xd->description) {
		tracker_sparql_builder_predicate (metadata, "nie:description");
		tracker_sparql_builder_object_unvalidated (metadata, xd->description);
	}

	if (xd->copyright) {
		tracker_sparql_builder_predicate (metadata, "nie:copyright");
		tracker_sparql_builder_object_unvalidated (metadata, xd->copyright);
	}

	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);
	}

	tracker_guarantee_title_from_file (metadata,
	                                   "nie:title",
	                                   md.title,
	                                   uri,
	                                   NULL);

	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 (xd->orientation) {
		tracker_sparql_builder_predicate (metadata, "nfo:orientation");
		tracker_sparql_builder_object_unvalidated (metadata, xd->orientation);
	}

	if (xd->exposure_time) {
		tracker_sparql_builder_predicate (metadata, "nmm:exposureTime");
		tracker_sparql_builder_object_unvalidated (metadata, xd->exposure_time);
	}

	if (xd->iso_speed_ratings) {
		tracker_sparql_builder_predicate (metadata, "nmm:isoSpeed");
		tracker_sparql_builder_object_unvalidated (metadata, xd->iso_speed_ratings);
	}

	if (xd->white_balance) {
		tracker_sparql_builder_predicate (metadata, "nmm:whiteBalance");
		tracker_sparql_builder_object_unvalidated (metadata, xd->white_balance);
	}

	if (xd->fnumber) {
		tracker_sparql_builder_predicate (metadata, "nmm:fnumber");
		tracker_sparql_builder_object_unvalidated (metadata, xd->fnumber);
	}

	if (xd->flash) {
		tracker_sparql_builder_predicate (metadata, "nmm:flash");
		tracker_sparql_builder_object_unvalidated (metadata, xd->flash);
	}

	if (xd->focal_length) {
		tracker_sparql_builder_predicate (metadata, "nmm:focalLength");
		tracker_sparql_builder_object_unvalidated (metadata, xd->focal_length);
	}

	if (xd->metering_mode) {
		tracker_sparql_builder_predicate (metadata, "nmm:meteringMode");
		tracker_sparql_builder_object_unvalidated (metadata, xd->metering_mode);
	}

	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->rating) {
		tracker_sparql_builder_predicate (metadata, "nao:numericRating");
		tracker_sparql_builder_object_unvalidated (metadata, xd->rating);
	}

	if (xd->subject) {
		tracker_keywords_parse (keywords, xd->subject);
	}

        if (xd->regions) {
                tracker_xmp_apply_regions (preupdate, metadata, graph, xd);
        }

	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);

	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:creator");
		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->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->gps_direction) {
		tracker_sparql_builder_predicate (metadata, "nfo:heading");
		tracker_sparql_builder_object_unvalidated (metadata, xd->gps_direction);
	}

	if (gd.width) {
		tracker_sparql_builder_predicate (metadata, "nfo:width");
		tracker_sparql_builder_object_unvalidated (metadata, gd.width);
		g_free (gd.width);
	}

	if (gd.height) {
		tracker_sparql_builder_predicate (metadata, "nfo:height");
		tracker_sparql_builder_object_unvalidated (metadata, gd.height);
		g_free (gd.height);
	}

	if (gd.comment) {
		tracker_sparql_builder_predicate (metadata, "nie:comment");
		tracker_sparql_builder_object_unvalidated (metadata, gd.comment);
		g_free (gd.comment);
	}

	tracker_xmp_free (xd);
}
Beispiel #21
0
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;
}
static void
parser_start_element (void           *data,
                      const xmlChar  *name_,
                      const xmlChar **attrs_)
{
	parser_data *pd = data;
	const gchar *name = (const gchar*) name_;
	const gchar **attrs = (const gchar**) attrs_;

	if (!pd || !name) {
		return;
	}

	/* Look for RDFa triple describing the license */
	if (g_ascii_strcasecmp (name, "a") == 0) {
		/* This tag is a license.  Ignore, however, if it is
		 * referring to another document.
		 */
		if (has_attribute (attrs, "rel", "license") &&
		    has_attribute (attrs, "about", NULL) == FALSE) {
			const xmlChar *href;

			href = lookup_attribute (attrs, "href");

			if (href) {
				tracker_sparql_builder_predicate (pd->metadata, "nie:license");
				tracker_sparql_builder_object_unvalidated (pd->metadata, href);
			}
		}
	} else if (g_ascii_strcasecmp (name, "title") == 0) {
		pd->current = READ_TITLE;
	} else if (g_ascii_strcasecmp (name, "meta") == 0) {
		if (has_attribute (attrs, "name", "author")) {
			const xmlChar *author;

			author = lookup_attribute (attrs, "content");

			if (author) {
				tracker_sparql_builder_predicate (pd->metadata, "nco:creator");
				tracker_sparql_builder_object_blank_open (pd->metadata);
				tracker_sparql_builder_predicate (pd->metadata, "a");
				tracker_sparql_builder_object (pd->metadata, "nco:Contact");
				tracker_sparql_builder_predicate (pd->metadata, "nco:fullname");
				tracker_sparql_builder_object_unvalidated (pd->metadata, author);
				tracker_sparql_builder_object_blank_close (pd->metadata);
			}
		}

		if (has_attribute (attrs, "name", "description")) {
			const xmlChar *desc;

			desc = lookup_attribute (attrs,"content");

			if (desc) {
				tracker_sparql_builder_predicate (pd->metadata, "nie:description");
				tracker_sparql_builder_object_unvalidated (pd->metadata, desc);
			}
		}

		if (has_attribute (attrs, "name", "keywords")) {
			const xmlChar* content = lookup_attribute (attrs, "content");

			if (content) {
				gchar **keywords;
				gint i;

				keywords = g_strsplit (content, ",", -1);
				if (keywords) {
					for (i = 0; keywords[i] != NULL; i++) {
						if (!keywords[i] || keywords[i] == '\0') {
							continue;
						}

						tracker_sparql_builder_predicate (pd->metadata, "nie:keyword");
						tracker_sparql_builder_object_unvalidated (pd->metadata, g_strstrip (keywords[i]));
					}

					g_strfreev (keywords);
				}
			}
		}
	} else if (g_ascii_strcasecmp (name, "body") == 0) {
		pd->in_body = TRUE;
	} else if (g_ascii_strcasecmp (name, "script") == 0) {
		/* Ignore javascript and such */
		pd->current = READ_IGNORE;
	}
}
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;
}