static MetadataParsingReturn metadataparse_jpeg_iptc (JpegParseData * jpeg_data, guint8 ** buf, guint32 * bufsize, guint8 ** next_start, guint32 * next_size) { int ret; ret = metadataparse_util_hold_chunk (&jpeg_data->read, buf, bufsize, next_start, next_size, jpeg_data->iptc_adapter); if (ret == META_PARSING_DONE) { const guint8 *buf; guint32 size; unsigned int iptc_len; int res; jpeg_data->state = JPEG_PARSE_READING; size = gst_adapter_available (*jpeg_data->iptc_adapter); buf = gst_adapter_peek (*jpeg_data->iptc_adapter, size); /* FIXME: currently we are throwing away others PhotoShop data */ res = iptc_jpeg_ps3_find_iptc (buf, size, &iptc_len); if (res < 0) { /* error */ ret = META_PARSING_ERROR; } else if (res == 0) { /* no iptc data found */ gst_adapter_clear (*jpeg_data->iptc_adapter); } else { gst_adapter_flush (*jpeg_data->iptc_adapter, res); size = gst_adapter_available (*jpeg_data->iptc_adapter); if (size > iptc_len) { GstBuffer *buf; buf = gst_adapter_take_buffer (*jpeg_data->iptc_adapter, iptc_len); gst_adapter_clear (*jpeg_data->iptc_adapter); gst_adapter_push (*jpeg_data->iptc_adapter, buf); } } /* if there is a second Iptc chunk in the file it will be jumped */ jpeg_data->iptc_adapter = NULL; } return ret; }
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; }