public function insert_references($row, $parameters)
{
self::debug_iterations("Inserting reference");
$this->commit_iterations("Reference", 500);
if ($this->archive_validator->has_error_by_line('http://eol.org/schema/reference/reference', $parameters['archive_table_definition']->location, $parameters['archive_line_number'])) {
write_to_resource_harvesting_log("ERROR: insert_references: has_error_by_line" . ",file_location:" . $parameters['archive_table_definition']->location . ",line_number:" . $parameters['archive_line_number']);
return false;
}
$reference_id = @self::field_decode($row['http://purl.org/dc/terms/identifier']);
$reference_taxon_ids = self::get_foreign_keys_from_row($row, 'http://rs.tdwg.org/dwc/terms/taxonID');
$reference_taxon_info = array();
if ($reference_taxon_ids) {
foreach ($reference_taxon_ids as $taxon_id) {
if ($taxon_info = @$this->taxon_ids_inserted[$taxon_id]) {
self::uncompress_array($taxon_info);
$reference_taxon_info[] = $taxon_info;
}
}
}
$full_reference = @self::field_decode($row['http://eol.org/schema/reference/full_reference']);
$title = @self::field_decode($row['http://purl.org/dc/terms/title']);
$pages = @self::field_decode($row['http://purl.org/ontology/bibo/pages']);
$pageStart = @self::field_decode($row['http://purl.org/ontology/bibo/pageStart']);
$pageEnd = @self::field_decode($row['http://purl.org/ontology/bibo/pageEnd']);
$volume = @self::field_decode($row['http://purl.org/ontology/bibo/volume']);
$edition = @self::field_decode($row['http://purl.org/ontology/bibo/edition']);
$publisher = @self::field_decode($row['http://purl.org/dc/terms/publisher']);
$authorList = @self::field_decode($row['http://purl.org/ontology/bibo/authorList']);
$editorList = @self::field_decode($row['http://purl.org/ontology/bibo/editorList']);
$created = @self::field_decode($row['http://purl.org/dc/terms/created']);
$language = Language::find_or_create_for_parser(@self::field_decode($row['http://purl.org/dc/terms/language']));
$uri = @self::field_decode($row['http://purl.org/ontology/bibo/uri']);
$doi = @self::field_decode($row['http://purl.org/ontology/bibo/doi']);
$params = array("provider_mangaed_id" => $reference_id, "full_reference" => $full_reference, "title" => $title, "pages" => $pages, "page_start" => $pageStart, "page_end" => $pageEnd, "volume" => $volume, "edition" => $edition, "publisher" => $publisher, "authors" => $authorList, "editors" => $editorList, "publication_created_at" => @$created ?: '0000-00-00 00:00:00', "language_id" => @$language->id ?: 0, "editors" => $editorList);
$reference = Reference::find_or_create($params);
if ($uri) {
$type = RefIdentifierType::find_or_create_by_label('uri');
$reference->add_ref_identifier(@$type->id ?: 0, $uri);
}
if ($doi) {
$type = RefIdentifierType::find_or_create_by_label('doi');
$reference->add_ref_identifier(@$type->id ?: 0, $doi);
}
if (isset($this->taxon_reference_ids[$reference_id])) {
foreach ($this->taxon_reference_ids[$reference_id] as $hierarchy_entry_id => $val) {
$this->mysqli->insert("INSERT IGNORE INTO hierarchy_entries_refs (hierarchy_entry_id, ref_id) VALUES ({$hierarchy_entry_id}, {$reference->id})");
$this->mysqli->query("UPDATE refs SET published=1, visibility_id=" . Visibility::visible()->id . " WHERE id={$reference->id}");
// TODO: find_or_create doesn't work here because of the dual primary key
// HierarchyEntriesRef::find_or_create(array(
// 'hierarchy_entry_id' => $hierarchy_entry_id,
// 'ref_id' => $reference->id));
}
}
if (isset($this->media_reference_ids[$reference_id])) {
foreach ($this->media_reference_ids[$reference_id] as $data_object_id => $data_object_guid) {
$this->mysqli->insert("INSERT IGNORE INTO data_objects_refs (data_object_id, ref_id) VALUES ({$data_object_id}, {$reference->id})");
$this->mysqli->query("UPDATE refs SET published=1, visibility_id=" . Visibility::visible()->id . " WHERE id={$reference->id}");
// TODO: find_or_create doesn't work here because of the dual primary key - same as above with entries
}
}
foreach ($reference_taxon_info as $taxon_info) {
$he_id = $taxon_info['hierarchy_entry_id'];
$this->mysqli->insert("INSERT IGNORE INTO hierarchy_entries_refs (hierarchy_entry_id, ref_id) VALUES ({$he_id}, {$reference->id})");
$this->mysqli->query("UPDATE refs SET published=1, visibility_id=" . Visibility::visible()->id . " WHERE id={$reference->id}");
// TODO: find_or_create doesn't work here because of the dual primary key - same as above with entries
}
$this->insert_data($row, $parameters);
}
public static function read_taxon_xml($t, $resource)
{
$t_dc = $t->children("http://purl.org/dc/elements/1.1/");
$t_dcterms = $t->children("http://purl.org/dc/terms/");
$t_dwc = $t->children("http://rs.tdwg.org/dwc/dwcore/");
$taxon_parameters = array();
$taxon_parameters["identifier"] = Functions::import_decode($t_dc->identifier);
$taxon_parameters["source_url"] = Functions::import_decode($t_dc->source);
$taxon_parameters["kingdom"] = Functions::import_decode($t_dwc->Kingdom);
$taxon_parameters["phylum"] = Functions::import_decode($t_dwc->Phylum);
$taxon_parameters["class"] = Functions::import_decode($t_dwc->Class);
$taxon_parameters["order"] = Functions::import_decode($t_dwc->Order);
$taxon_parameters["family"] = Functions::import_decode($t_dwc->Family);
$taxon_parameters["genus"] = Functions::import_decode($t_dwc->Genus);
$taxon_parameters["scientific_name"] = Functions::import_decode($t_dwc->ScientificName);
$taxon_parameters["rank"] = Rank::find_or_create_by_translated_label(Functions::import_decode($t->rank));
$taxon_parameters["taxon_created_at"] = trim($t_dcterms->created);
$taxon_parameters["taxon_modified_at"] = trim($t_dcterms->modified);
if ($taxon_parameters["scientific_name"]) {
$taxon_parameters["name"] = Name::find_or_create_by_string($taxon_parameters["scientific_name"]);
} else {
if ($name = $taxon_parameters["genus"]) {
$taxon_parameters["scientific_name"] = $name;
$taxon_parameters["name"] = Name::find_or_create_by_string($name);
$taxon_parameters["genus"] = "";
} elseif ($name = $taxon_parameters["family"]) {
$taxon_parameters["scientific_name"] = $name;
$taxon_parameters["name"] = Name::find_or_create_by_string($name);
$taxon_parameters["family"] = "";
} elseif ($name = $taxon_parameters["order"]) {
$taxon_parameters["scientific_name"] = $name;
$taxon_parameters["name"] = Name::find_or_create_by_string($name);
$taxon_parameters["order"] = "";
} elseif ($name = $taxon_parameters["class"]) {
$taxon_parameters["scientific_name"] = $name;
$taxon_parameters["name"] = Name::find_or_create_by_string($name);
$taxon_parameters["class"] = "";
} elseif ($name = $taxon_parameters["phylum"]) {
$taxon_parameters["scientific_name"] = $name;
$taxon_parameters["name"] = Name::find_or_create_by_string($name);
$taxon_parameters["phylum"] = "";
} elseif ($name = $taxon_parameters["kingdom"]) {
$taxon_parameters["scientific_name"] = $name;
$taxon_parameters["name"] = Name::find_or_create_by_string($name);
$taxon_parameters["kingdom"] = "";
} else {
return;
}
}
$taxon_parameters["common_names"] = array();
foreach ($t->commonName as $c) {
$common_name = Functions::import_decode((string) $c);
if (!$common_name) {
continue;
}
$xml_attr = $c->attributes("http://www.w3.org/XML/1998/namespace");
$params = array("name" => $common_name, "language" => Language::find_or_create_for_parser(@Functions::import_decode($xml_attr["lang"])));
$taxon_parameters["common_names"][] = $params;
}
$taxon_parameters["synonyms"] = array();
foreach ($t->synonym as $s) {
$synonym = Functions::import_decode((string) $s);
if (!$synonym) {
continue;
}
$attr = $s->attributes();
if (!@$attr["relationship"]) {
$attr["relationship"] = 'synonym';
}
$params = array("name" => Name::find_or_create_by_string($synonym), "synonym_relation" => SynonymRelation::find_or_create_by_translated_label(trim($attr["relationship"])));
$taxon_parameters["synonyms"][] = $params;
}
$taxon_parameters["agents"] = array();
foreach ($t->agent as $a) {
$agent_name = Functions::import_decode((string) $a);
if (!$agent_name) {
continue;
}
$attr = $a->attributes();
$params = array("full_name" => Functions::import_decode((string) $a, 0, 0), "homepage" => @Functions::import_decode($attr["homepage"]), "logo_url" => @Functions::import_decode($attr["logoURL"]), "agent_role" => AgentRole::find_or_create_by_translated_label(@trim($attr["role"])));
$taxon_parameters["agents"][] = $params;
unset($params);
}
$taxon_parameters["refs"] = array();
foreach ($t->reference as $r) {
$reference = Functions::import_decode((string) $r, 0, 0);
if (!$reference) {
continue;
}
$ref = Reference::find_or_create_by_full_reference($reference);
$taxon_parameters["refs"][] = $ref;
$id_labels = array("bici", "coden", "doi", "eissn", "handle", "issn", "isbn", "lsid", "oclc", "sici", "url", "urn");
$attr = $r->attributes();
foreach ($id_labels as $label) {
if ($id = @Functions::import_decode($attr[$label], 0, 0)) {
$type = RefIdentifierType::find_or_create_by_label($label);
$ref->add_ref_identifier(@$type->id ?: 0, $id);
}
}
}
$taxon_parameters["data_objects"] = array();
foreach ($t->dataObject as $d) {
$d_dc = $d->children("http://purl.org/dc/elements/1.1/");
$d_dcterms = $d->children("http://purl.org/dc/terms/");
$d_geo = $d->children("http://www.w3.org/2003/01/geo/wgs84_pos#");
$data_object = new DataObject();
$data_object->identifier = Functions::import_decode($d_dc->identifier);
$data_object->data_type = DataType::find_or_create_by_schema_value(Functions::import_decode($d->dataType));
$data_object->mime_type = MimeType::find_or_create_by_translated_label(Functions::import_decode($d->mimeType));
$data_object->object_created_at = Functions::import_decode($d_dcterms->created);
$data_object->object_modified_at = Functions::import_decode($d_dcterms->modified);
$data_object->object_title = Functions::import_decode($d_dc->title, 0, 0);
$data_object->language = Language::find_or_create_for_parser(Functions::import_decode($d_dc->language));
$data_object->license = License::find_or_create_for_parser(Functions::import_decode($d->license));
$data_object->rights_statement = Functions::import_decode($d_dc->rights, 0, 0);
$data_object->rights_holder = Functions::import_decode($d_dcterms->rightsHolder, 0, 0);
$data_object->bibliographic_citation = Functions::import_decode($d_dcterms->bibliographicCitation, 0, 0);
$data_object->source_url = Functions::import_decode($d_dc->source);
$data_object->description = Functions::import_decode($d_dc->description, 0, 0);
$data_object->object_url = Functions::import_decode($d->mediaURL);
$data_object->thumbnail_url = Functions::import_decode($d->thumbnailURL);
$data_object->location = Functions::import_decode($d->location, 0, 0);
if (@$d->additionalInformation) {
$data_object->additional_information = (array) $d->additionalInformation;
}
if ($r = (string) @$d->additionalInformation->rating) {
if (is_numeric($r) && $r > 0 && $r <= 5) {
$data_object->data_rating = $r;
}
}
if ($subtype = @$d->additionalInformation->subtype) {
if ($dt = DataType::find_or_create_by_schema_value(Functions::import_decode($subtype))) {
$data_object->data_subtype_id = $dt->id;
}
}
$data_object_parameters = array();
if (!$data_object->language) {
$xml_attr = $d_dc->description->attributes("http://www.w3.org/XML/1998/namespace");
$data_object->language = Language::find_or_create_for_parser(@Functions::import_decode($xml_attr["lang"]));
}
if (!$data_object->language && $resource->language) {
$data_object->language = $resource->language;
}
//TODO - update this
if ($data_object->mime_type && $data_object->mime_type->equals(MimeType::flash()) && $data_object->is_video()) {
$data_object->data_type = DataType::youtube();
$data_object->data_type_id = DataType::youtube()->id;
}
//take the taxon's source_url if none present
if (!@$data_object->source_url && @$taxon_parameters["source_url"]) {
$data_object->source_url = $taxon_parameters["source_url"];
}
// Turn newlines into paragraphs
$data_object->description = str_replace("\n", "</p><p>", $data_object->description);
/* Checking requirements*/
//if text: must have description
if ($data_object->data_type->equals(DataType::text()) && !$data_object->description) {
continue;
}
//if image, movie or sound: must have object_url
if (($data_object->data_type->equals(DataType::video()) || $data_object->data_type->equals(DataType::sound()) || $data_object->data_type->equals(DataType::image())) && !$data_object->object_url) {
continue;
}
$data_object->latitude = 0;
$data_object->longitude = 0;
$data_object->altitude = 0;
foreach ($d_geo->Point as $p) {
$p_geo = $p->children("http://www.w3.org/2003/01/geo/wgs84_pos#");
$data_object->latitude = Functions::import_decode($p_geo->lat);
$data_object->longitude = Functions::import_decode($p_geo->long);
$data_object->altitude = Functions::import_decode($p_geo->alt);
}
$data_object_parameters["agents"] = array();
foreach ($d->agent as $a) {
$agent_name = Functions::import_decode((string) $a);
if (!$agent_name) {
continue;
}
$attr = $a->attributes();
$params = array("full_name" => Functions::import_decode((string) $a, 0, 0), "homepage" => @Functions::import_decode($attr["homepage"]), "logo_url" => @Functions::import_decode($attr["logoURL"]), "agent_role" => AgentRole::find_or_create_by_translated_label(@trim($attr["role"])));
$data_object_parameters["agents"][] = $params;
unset($params);
}
$data_object_parameters["audiences"] = array();
foreach ($d->audience as $a) {
$data_object_parameters["audiences"][] = Audience::find_or_create_by_translated_label(trim((string) $a));
}
$data_object_parameters["info_items"] = array();
foreach ($d->subject as $s) {
$data_object_parameters["info_items"][] = InfoItem::find_or_create_by_schema_value(trim((string) $s));
}
if ($subject = @$d->additionalInformation->subject) {
if ($ii = InfoItem::find_or_create_by_schema_value(trim((string) $subject))) {
$data_object_parameters["info_items"] = array($ii);
}
}
// EXCEPTIONS
if ($data_object->is_text()) {
if ($resource->title == "BOLD Systems Resource") {
// EXCEPTION - overriding the subject for BOLD
$data_object_parameters["info_items"] = array(InfoItem::find_or_create_by_schema_value('http://www.eol.org/voc/table_of_contents#Barcode'));
} elseif ($resource->title == "Wikipedia") {
// EXCEPTION - overriding the subject for Wikipedia
$data_object_parameters["info_items"] = array(InfoItem::find_or_create_by_schema_value('http://www.eol.org/voc/table_of_contents#Wikipedia'));
} elseif ($resource->title == "IUCN Red List") {
if ($data_object->object_title == "IUCNConservationStatus") {
// EXCEPTION - overriding the data type for IUCN text
$data_object->data_type_id = DataType::iucn()->id;
$data_object->data_type = DataType::iucn();
}
}
}
$data_object_parameters["refs"] = array();
foreach ($d->reference as $r) {
$reference = Functions::import_decode((string) $r, 0, 0);
if (!$reference) {
continue;
}
$ref = Reference::find_or_create_by_full_reference($reference);
$data_object_parameters["refs"][] = $ref;
$id_labels = array("bici", "coden", "doi", "eissn", "handle", "issn", "isbn", "lsid", "oclc", "sici", "url", "urn");
$attr = $r->attributes();
foreach ($id_labels as $label) {
if ($id = @Functions::import_decode($attr[$label], 0, 0)) {
$type = RefIdentifierType::find_or_create_by_label($label);
$ref->add_ref_identifier(@$type->id ?: 0, $id);
}
}
}
$taxon_parameters["data_objects"][] = array($data_object, $data_object_parameters);
unset($data_object);
}
return $taxon_parameters;
}