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;
}
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_int64 (TrackerSparqlBuilder *metadata,
const gchar *predicate,
gint64 value)
{
tracker_sparql_builder_predicate (metadata, predicate);
tracker_sparql_builder_object_int64 (metadata, value);
}
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)
{
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;
}
G_MODULE_EXPORT gboolean
tracker_extract_get_metadata (TrackerExtractInfo *info)
{
TrackerConfig *config;
GTime creation_date;
GError *error = NULL;
TrackerSparqlBuilder *metadata, *preupdate;
const gchar *graph;
const gchar *urn;
TrackerXmpData *xd = NULL;
PDFData pd = { 0 }; /* actual data */
PDFData md = { 0 }; /* for merging */
PopplerDocument *document;
gchar *xml = NULL;
gchar *content, *uri;
guint n_bytes;
GPtrArray *keywords;
guint i;
GFile *file;
gchar *filename;
int fd;
gchar *contents = NULL;
gsize len;
struct stat st;
metadata = tracker_extract_info_get_metadata_builder (info);
preupdate = tracker_extract_info_get_preupdate_builder (info);
graph = tracker_extract_info_get_graph (info);
urn = tracker_extract_info_get_urn (info);
file = tracker_extract_info_get_file (info);
filename = g_file_get_path (file);
fd = tracker_file_open_fd (filename);
if (fd == -1) {
g_warning ("Could not open pdf file '%s': %s\n",
filename,
g_strerror (errno));
g_free (filename);
return FALSE;
}
if (fstat (fd, &st) == -1) {
g_warning ("Could not fstat pdf file '%s': %s\n",
filename,
g_strerror (errno));
close (fd);
g_free (filename);
return FALSE;
}
if (st.st_size == 0) {
contents = NULL;
len = 0;
} else {
contents = (gchar *) mmap (NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
if (contents == NULL || contents == MAP_FAILED) {
g_warning ("Could not mmap pdf file '%s': %s\n",
filename,
g_strerror (errno));
close (fd);
g_free (filename);
return FALSE;
}
len = st.st_size;
}
g_free (filename);
uri = g_file_get_uri (file);
document = poppler_document_new_from_data (contents, len, NULL, &error);
if (error) {
if (error->code == POPPLER_ERROR_ENCRYPTED) {
tracker_sparql_builder_predicate (metadata, "a");
tracker_sparql_builder_object (metadata, "nfo:PaginatedTextDocument");
tracker_sparql_builder_predicate (metadata, "nfo:isContentEncrypted");
tracker_sparql_builder_object_boolean (metadata, TRUE);
g_error_free (error);
g_free (uri);
close (fd);
return TRUE;
} else {
g_warning ("Couldn't create PopplerDocument from uri:'%s', %s",
uri,
error->message ? error->message : "no error given");
g_error_free (error);
g_free (uri);
close (fd);
return FALSE;
}
}
if (!document) {
g_warning ("Could not create PopplerDocument from uri:'%s', "
"NULL returned without an error",
uri);
g_free (uri);
close (fd);
return FALSE;
}
tracker_sparql_builder_predicate (metadata, "a");
tracker_sparql_builder_object (metadata, "nfo:PaginatedTextDocument");
g_object_get (document,
"title", &pd.title,
"author", &pd.author,
"subject", &pd.subject,
"keywords", &pd.keywords,
"creation-date", &creation_date,
"metadata", &xml,
NULL);
if (creation_date > 0) {
pd.creation_date = tracker_date_to_string ((time_t) creation_date);
}
keywords = g_ptr_array_new_with_free_func ((GDestroyNotify) g_free);
if (xml && *xml &&
(xd = tracker_xmp_new (xml, strlen (xml), uri)) != NULL) {
/* The casts here are well understood and known */
md.title = (gchar *) tracker_coalesce_strip (4, pd.title, xd->title, xd->title2, xd->pdf_title);
md.subject = (gchar *) tracker_coalesce_strip (2, pd.subject, xd->subject);
md.date = (gchar *) tracker_coalesce_strip (3, pd.creation_date, xd->date, xd->time_original);
md.author = (gchar *) tracker_coalesce_strip (2, pd.author, xd->creator);
write_pdf_data (md, metadata, keywords);
if (xd->keywords) {
tracker_keywords_parse (keywords, xd->keywords);
}
if (xd->pdf_keywords) {
tracker_keywords_parse (keywords, xd->pdf_keywords);
}
if (xd->publisher) {
tracker_sparql_builder_predicate (metadata, "nco:publisher");
tracker_sparql_builder_object_blank_open (metadata);
tracker_sparql_builder_predicate (metadata, "a");
tracker_sparql_builder_object (metadata, "nco:Contact");
tracker_sparql_builder_predicate (metadata, "nco:fullname");
tracker_sparql_builder_object_unvalidated (metadata, xd->publisher);
tracker_sparql_builder_object_blank_close (metadata);
}
if (xd->type) {
tracker_sparql_builder_predicate (metadata, "dc:type");
tracker_sparql_builder_object_unvalidated (metadata, xd->type);
}
if (xd->format) {
tracker_sparql_builder_predicate (metadata, "dc:format");
tracker_sparql_builder_object_unvalidated (metadata, xd->format);
}
if (xd->identifier) {
tracker_sparql_builder_predicate (metadata, "dc:identifier");
tracker_sparql_builder_object_unvalidated (metadata, xd->identifier);
}
if (xd->source) {
tracker_sparql_builder_predicate (metadata, "dc:source");
tracker_sparql_builder_object_unvalidated (metadata, xd->source);
}
if (xd->language) {
tracker_sparql_builder_predicate (metadata, "dc:language");
tracker_sparql_builder_object_unvalidated (metadata, xd->language);
}
if (xd->relation) {
tracker_sparql_builder_predicate (metadata, "dc:relation");
tracker_sparql_builder_object_unvalidated (metadata, xd->relation);
}
if (xd->coverage) {
tracker_sparql_builder_predicate (metadata, "dc:coverage");
tracker_sparql_builder_object_unvalidated (metadata, xd->coverage);
}
if (xd->license) {
tracker_sparql_builder_predicate (metadata, "nie:license");
tracker_sparql_builder_object_unvalidated (metadata, xd->license);
}
if (xd->make || xd->model) {
gchar *equip_uri;
equip_uri = tracker_sparql_escape_uri_printf ("urn:equipment:%s:%s:",
xd->make ? xd->make : "",
xd->model ? xd->model : "");
tracker_sparql_builder_insert_open (preupdate, NULL);
if (graph) {
tracker_sparql_builder_graph_open (preupdate, graph);
}
tracker_sparql_builder_subject_iri (preupdate, equip_uri);
tracker_sparql_builder_predicate (preupdate, "a");
tracker_sparql_builder_object (preupdate, "nfo:Equipment");
if (xd->make) {
tracker_sparql_builder_predicate (preupdate, "nfo:manufacturer");
tracker_sparql_builder_object_unvalidated (preupdate, xd->make);
}
if (xd->model) {
tracker_sparql_builder_predicate (preupdate, "nfo:model");
tracker_sparql_builder_object_unvalidated (preupdate, xd->model);
}
if (graph) {
tracker_sparql_builder_graph_close (preupdate);
}
tracker_sparql_builder_insert_close (preupdate);
tracker_sparql_builder_predicate (metadata, "nfo:equipment");
tracker_sparql_builder_object_iri (metadata, equip_uri);
g_free (equip_uri);
}
if (xd->orientation) {
tracker_sparql_builder_predicate (metadata, "nfo:orientation");
tracker_sparql_builder_object (metadata, xd->orientation);
}
if (xd->rights) {
tracker_sparql_builder_predicate (metadata, "nie:copyright");
tracker_sparql_builder_object_unvalidated (metadata, xd->rights);
}
if (xd->white_balance) {
tracker_sparql_builder_predicate (metadata, "nmm:whiteBalance");
tracker_sparql_builder_object (metadata, xd->white_balance);
}
if (xd->fnumber) {
gdouble value;
value = g_strtod (xd->fnumber, NULL);
tracker_sparql_builder_predicate (metadata, "nmm:fnumber");
tracker_sparql_builder_object_double (metadata, value);
}
if (xd->flash) {
tracker_sparql_builder_predicate (metadata, "nmm:flash");
tracker_sparql_builder_object (metadata, xd->flash);
}
if (xd->focal_length) {
gdouble value;
value = g_strtod (xd->focal_length, NULL);
tracker_sparql_builder_predicate (metadata, "nmm:focalLength");
tracker_sparql_builder_object_double (metadata, value);
}
/* Question: Shouldn't xd->Artist be merged with md.author instead? */
if (xd->artist || xd->contributor) {
const gchar *artist;
artist = tracker_coalesce_strip (2, xd->artist, xd->contributor);
tracker_sparql_builder_predicate (metadata, "nco:contributor");
tracker_sparql_builder_object_blank_open (metadata);
tracker_sparql_builder_predicate (metadata, "a");
tracker_sparql_builder_object (metadata, "nco:Contact");
tracker_sparql_builder_predicate (metadata, "nco:fullname");
tracker_sparql_builder_object_unvalidated (metadata, artist);
tracker_sparql_builder_object_blank_close (metadata);
}
if (xd->exposure_time) {
gdouble value;
value = g_strtod (xd->exposure_time, NULL);
tracker_sparql_builder_predicate (metadata, "nmm:exposureTime");
tracker_sparql_builder_object_double (metadata, value);
}
if (xd->iso_speed_ratings) {
gdouble value;
value = g_strtod (xd->iso_speed_ratings, NULL);
tracker_sparql_builder_predicate (metadata, "nmm:isoSpeed");
tracker_sparql_builder_object_double (metadata, value);
}
if (xd->description) {
tracker_sparql_builder_predicate (metadata, "nie:description");
tracker_sparql_builder_object_unvalidated (metadata, xd->description);
}
if (xd->metering_mode) {
tracker_sparql_builder_predicate (metadata, "nmm:meteringMode");
tracker_sparql_builder_object (metadata, xd->metering_mode);
}
if (xd->address || xd->state || xd->country || xd->city ||
xd->gps_altitude || xd->gps_latitude || xd-> gps_longitude) {
tracker_sparql_builder_predicate (metadata, "slo:location");
tracker_sparql_builder_object_blank_open (metadata); /* GeoLocation */
tracker_sparql_builder_predicate (metadata, "a");
tracker_sparql_builder_object (metadata, "slo:GeoLocation");
if (xd->address || xd->state || xd->country || xd->city) {
gchar *addruri;
addruri = tracker_sparql_get_uuid_urn ();
tracker_sparql_builder_predicate (metadata, "slo:postalAddress");
tracker_sparql_builder_object_iri (metadata, addruri);
tracker_sparql_builder_insert_open (preupdate, NULL);
if (graph) {
tracker_sparql_builder_graph_open (preupdate, graph);
}
tracker_sparql_builder_subject_iri (preupdate, addruri);
g_free (addruri);
tracker_sparql_builder_predicate (preupdate, "a");
tracker_sparql_builder_object (preupdate, "nco:PostalAddress");
if (xd->address) {
tracker_sparql_builder_predicate (preupdate, "nco:streetAddress");
tracker_sparql_builder_object_unvalidated (preupdate, xd->address);
}
if (xd->state) {
tracker_sparql_builder_predicate (preupdate, "nco:region");
tracker_sparql_builder_object_unvalidated (preupdate, xd->state);
}
if (xd->city) {
tracker_sparql_builder_predicate (preupdate, "nco:locality");
tracker_sparql_builder_object_unvalidated (preupdate, xd->city);
}
if (xd->country) {
tracker_sparql_builder_predicate (preupdate, "nco:country");
tracker_sparql_builder_object_unvalidated (preupdate, xd->country);
}
if (graph) {
tracker_sparql_builder_graph_close (preupdate);
}
tracker_sparql_builder_insert_close (preupdate);
}
if (xd->gps_altitude) {
tracker_sparql_builder_predicate (metadata, "slo:altitude");
tracker_sparql_builder_object_unvalidated (metadata, xd->gps_altitude);
}
if (xd->gps_latitude) {
tracker_sparql_builder_predicate (metadata, "slo:latitude");
tracker_sparql_builder_object_unvalidated (metadata, xd->gps_latitude);
}
if (xd->gps_longitude) {
tracker_sparql_builder_predicate (metadata, "slo:longitude");
tracker_sparql_builder_object_unvalidated (metadata, xd->gps_longitude);
}
tracker_sparql_builder_object_blank_close (metadata); /* GeoLocation */
}
if (xd->regions) {
tracker_xmp_apply_regions (preupdate, metadata, graph, xd);
}
tracker_xmp_free (xd);
} else {
/* So if we are here we have NO XMP data and we just
* write what we know from Poppler.
*/
write_pdf_data (pd, metadata, keywords);
}
for (i = 0; i < keywords->len; i++) {
gchar *escaped, *subject;
const gchar *p;
p = g_ptr_array_index (keywords, i);
escaped = tracker_sparql_escape_string (p);
subject = g_strdup_printf ("_:tag%d", i + 1);
/* ensure tag with specified label exists */
tracker_sparql_builder_insert_open (preupdate, graph);
tracker_sparql_builder_subject (preupdate, subject);
tracker_sparql_builder_predicate (preupdate, "a");
tracker_sparql_builder_object (preupdate, "nao:Tag");
tracker_sparql_builder_predicate (preupdate, "nao:prefLabel");
tracker_sparql_builder_object_unvalidated (preupdate, escaped);
tracker_sparql_builder_insert_close (preupdate);
tracker_sparql_builder_append (preupdate,
"WHERE { FILTER (NOT EXISTS { "
"?tag a nao:Tag ; nao:prefLabel \"");
tracker_sparql_builder_append (preupdate, escaped);
tracker_sparql_builder_append (preupdate,
"\" }) }\n");
/* associate file with tag */
tracker_sparql_builder_insert_open (preupdate, graph);
tracker_sparql_builder_subject_iri (preupdate, urn);
tracker_sparql_builder_predicate (preupdate, "nao:hasTag");
tracker_sparql_builder_object (preupdate, "?tag");
tracker_sparql_builder_insert_close (preupdate);
tracker_sparql_builder_where_open (preupdate);
tracker_sparql_builder_subject (preupdate, "?tag");
tracker_sparql_builder_predicate (preupdate, "a");
tracker_sparql_builder_object (preupdate, "nao:Tag");
tracker_sparql_builder_predicate (preupdate, "nao:prefLabel");
tracker_sparql_builder_object_unvalidated (preupdate, escaped);
tracker_sparql_builder_where_close (preupdate);
g_free (subject);
g_free (escaped);
}
g_ptr_array_free (keywords, TRUE);
tracker_sparql_builder_predicate (metadata, "nfo:pageCount");
tracker_sparql_builder_object_int64 (metadata, poppler_document_get_n_pages (document));
config = tracker_main_get_config ();
n_bytes = tracker_config_get_max_bytes (config);
content = extract_content_text (document, n_bytes);
if (content) {
tracker_sparql_builder_predicate (metadata, "nie:plainTextContent");
tracker_sparql_builder_object_unvalidated (metadata, content);
g_free (content);
}
read_outline (document, metadata);
g_free (xml);
g_free (pd.keywords);
g_free (pd.title);
g_free (pd.subject);
g_free (pd.creation_date);
g_free (pd.author);
g_free (pd.date);
g_free (uri);
g_object_unref (document);
if (contents) {
munmap (contents, len);
}
close (fd);
return TRUE;
}
G_MODULE_EXPORT gboolean
tracker_extract_get_metadata (TrackerExtractInfo *info)
{
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;
}
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 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);
}
G_MODULE_EXPORT gboolean
tracker_extract_get_metadata (TrackerExtractInfo *info)
{
goffset size;
FILE *f;
png_structp png_ptr;
png_infop info_ptr;
png_infop end_ptr;
png_bytepp row_pointers;
guint row;
png_uint_32 width, height;
gint bit_depth, color_type;
gint interlace_type, compression_type, filter_type;
const gchar *dlna_profile, *dlna_mimetype, *graph;
TrackerSparqlBuilder *preupdate, *metadata;
gchar *filename, *uri;
GString *where;
GFile *file;
file = tracker_extract_info_get_file (info);
filename = g_file_get_path (file);
size = tracker_file_get_size (filename);
preupdate = tracker_extract_info_get_preupdate_builder (info);
metadata = tracker_extract_info_get_metadata_builder (info);
graph = tracker_extract_info_get_graph (info);
if (size < 64) {
return FALSE;
}
f = tracker_file_open (filename);
g_free (filename);
if (!f) {
return FALSE;
}
png_ptr = png_create_read_struct (PNG_LIBPNG_VER_STRING,
NULL,
NULL,
NULL);
if (!png_ptr) {
tracker_file_close (f, FALSE);
return FALSE;
}
info_ptr = png_create_info_struct (png_ptr);
if (!info_ptr) {
png_destroy_read_struct (&png_ptr, &info_ptr, NULL);
tracker_file_close (f, FALSE);
return FALSE;
}
end_ptr = png_create_info_struct (png_ptr);
if (!end_ptr) {
png_destroy_read_struct (&png_ptr, &info_ptr, NULL);
tracker_file_close (f, FALSE);
return FALSE;
}
if (setjmp (png_jmpbuf (png_ptr))) {
png_destroy_read_struct (&png_ptr, &info_ptr, &end_ptr);
tracker_file_close (f, FALSE);
return FALSE;
}
png_init_io (png_ptr, f);
png_read_info (png_ptr, info_ptr);
if (!png_get_IHDR (png_ptr,
info_ptr,
&width,
&height,
&bit_depth,
&color_type,
&interlace_type,
&compression_type,
&filter_type)) {
png_destroy_read_struct (&png_ptr, &info_ptr, &end_ptr);
tracker_file_close (f, FALSE);
return FALSE;
}
/* Read the image. FIXME We should be able to skip this step and
* just get the info from the end. This causes some errors atm.
*/
row_pointers = g_new0 (png_bytep, height);
for (row = 0; row < height; row++) {
row_pointers[row] = png_malloc (png_ptr,
png_get_rowbytes (png_ptr,info_ptr));
}
png_read_image (png_ptr, row_pointers);
for (row = 0; row < height; row++) {
png_free (png_ptr, row_pointers[row]);
}
g_free (row_pointers);
png_read_end (png_ptr, end_ptr);
tracker_sparql_builder_predicate (metadata, "a");
tracker_sparql_builder_object (metadata, "nfo:Image");
tracker_sparql_builder_object (metadata, "nmm:Photo");
uri = g_file_get_uri (file);
where = g_string_new ("");
read_metadata (preupdate, metadata, where, png_ptr, info_ptr, end_ptr, uri, graph);
tracker_extract_info_set_where_clause (info, where->str);
g_string_free (where, TRUE);
g_free (uri);
tracker_sparql_builder_predicate (metadata, "nfo:width");
tracker_sparql_builder_object_int64 (metadata, width);
tracker_sparql_builder_predicate (metadata, "nfo:height");
tracker_sparql_builder_object_int64 (metadata, height);
if (guess_dlna_profile (bit_depth, width, height, &dlna_profile, &dlna_mimetype)) {
tracker_sparql_builder_predicate (metadata, "nmm:dlnaProfile");
tracker_sparql_builder_object_string (metadata, dlna_profile);
tracker_sparql_builder_predicate (metadata, "nmm:dlnaMime");
tracker_sparql_builder_object_string (metadata, dlna_mimetype);
}
png_destroy_read_struct (&png_ptr, &info_ptr, &end_ptr);
tracker_file_close (f, FALSE);
return TRUE;
}
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);
}
}
