function add_hierarchy_entry(&$row, $parent_hierarchy_entry_id, $ancestry, $branch_kingdom)
{
self::debug_iterations("Inserting taxon");
self::commit_iterations("Taxa", 500);
if ($this->archive_validator->has_error_by_line('http://rs.tdwg.org/dwc/terms/taxon', $row['archive_file_location'], $row['archive_line_number'])) {
write_to_resource_harvesting_log("ERROR: add_hierarchy_entry: has_error_by_line" . ",file_location:" . $row['archive_file_location'] . ",line_number:" . $row['archive_line_number']);
return false;
}
// make sure this taxon has a name, otherwise skip this branch
$scientific_name = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/scientificName']);
$authorship = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/scientificNameAuthorship']);
$kingdom = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/kingdom']);
$genus = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/genus']);
$rank_label = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/taxonRank']);
// COL exception
if (strtolower($kingdom) == 'viruses') {
if (substr($scientific_name, -1) == ":") {
$scientific_name = substr($scientific_name, 0, -1);
}
if (preg_match("/^(.*) ICTV\$/i", $scientific_name, $arr)) {
$scientific_name = $arr[1];
}
}
// COL exception
if (strtolower($kingdom) == 'viruses' && $genus && strtolower($rank_label) != 'genus') {
if (stripos($scientific_name, $genus) == 0) {
$scientific_name = ucfirst(trim(substr($scientific_name, strlen($genus))));
}
} else {
if ($authorship && stripos($scientific_name, $authorship) === false) {
$scientific_name = trim($scientific_name . " " . $authorship);
}
}
if (!$scientific_name) {
return false;
}
$taxon_id = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/taxonID']);
if (!$taxon_id) {
$taxon_id = @self::field_decode($row['http://purl.org/dc/terms/identifier']);
}
if (!$taxon_id) {
debug("ERROR - no taxon ID for {$scientific_name}, skipping");
return false;
}
if (isset($this->taxon_ids_inserted[$taxon_id])) {
// this taxon_id has already been inserted meaning this tree has a loop in it - so stop
debug("ERROR - taxon ID ({$taxon_id}) for {$scientific_name} already inserted; LOOP?");
return false;
}
$scientific_name = ucfirst($scientific_name);
$name = Name::find_or_create_by_string($scientific_name);
if (@(!$name->id)) {
debug("ERROR - Failed to insert name: {$scientific_name}");
return false;
}
$phylum = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/phylum']);
$class = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/class']);
$order = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/order']);
$family = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/family']);
$rank = Rank::find_or_create_by_translated_label($rank_label);
$dataset_id = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/datasetID']);
$taxonomic_status = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/taxonomicStatus']);
$source_url = @self::field_decode($row['http://rs.tdwg.org/ac/terms/furtherInformationURL']);
if (!$source_url) {
$source_url = @self::field_decode($row['http://purl.org/dc/terms/source']);
}
if (!$source_url) {
$source_url = @self::field_decode($row['http://purl.org/dc/terms/references']);
}
if (!$source_url) {
$source_url = @self::field_decode($row['http://purl.org/dc/terms/isReferencedBy']);
}
if (isset($row['http://rs.tdwg.org/dwc/terms/taxonRemarks'])) {
$taxon_remarks = @self::field_decode($row['http://rs.tdwg.org/dwc/terms/taxonRemarks']);
} else {
$taxon_remarks = NULL;
}
if (!$taxon_remarks && strtolower($taxonomic_status) == 'provisionally accepted name') {
$taxon_remarks = "provisionally accepted name";
}
// TODO: This block is somewhat confusing. Clearly, it's clearing the
// rank that's currently being read, but shouldn't it also clear all of
// the ranks below that?
if (strtolower($rank_label) == 'kingdom') {
$kingdom = null;
}
if (strtolower($rank_label) == 'phylum') {
$phylum = null;
}
if (strtolower($rank_label) == 'class') {
$class = null;
}
if (strtolower($rank_label) == 'order') {
$order = null;
}
if (strtolower($rank_label) == 'family') {
$family = null;
}
if (strtolower($rank_label) == 'genus') {
$genus = null;
}
// these are the taxa using the adjacency list format
if (!$parent_hierarchy_entry_id && ($kingdom || $phylum || $class || $order || $family || $genus)) {
$params = array("identifier" => $taxon_id, "source_url" => $source_url, "kingdom" => ucfirst($kingdom), "phylum" => ucfirst($phylum), "class" => ucfirst($class), "order" => ucfirst($order), "family" => ucfirst($family), "genus" => ucfirst($genus), "scientificName" => ucfirst($scientific_name), "name" => $name, "rank" => $rank, "taxon_remarks" => $taxon_remarks);
$hierarchy_entry = HierarchyEntry::create_entries_for_taxon($params, $this->harvest_event->resource->hierarchy_id);
if (@(!$hierarchy_entry->id)) {
debug("ERROR - unable to insert hierarchy entry for {$scientific_name}");
return;
}
// NOTE: This is NOT adding a hierarchy entry, but a
// harvest_event_hierarchy_entry:
// TODO: I am not sure this adds entries for ancestors!
$this->harvest_event->add_hierarchy_entry($hierarchy_entry, 'inserted');
$this->taxon_ids_inserted[$taxon_id] = array('hierarchy_entry_id' => $hierarchy_entry->id, 'taxon_concept_id' => $hierarchy_entry->taxon_concept_id, 'source_url' => $source_url);
self::compress_array($this->taxon_ids_inserted[$taxon_id]);
} else {
$params = array("identifier" => $taxon_id, "source_url" => $source_url, "name_id" => $name->id, "parent_id" => $parent_hierarchy_entry_id, "hierarchy_id" => $this->harvest_event->resource->hierarchy_id, "rank" => $rank, "ancestry" => $ancestry, "taxon_remarks" => $taxon_remarks);
$hierarchy_entry = HierarchyEntry::find_or_create_by_array($params);
if (@(!$hierarchy_entry->id)) {
return;
}
$this->harvest_event->add_hierarchy_entry($hierarchy_entry, 'inserted');
$this->taxon_ids_inserted[$taxon_id] = array('hierarchy_entry_id' => $hierarchy_entry->id, 'taxon_concept_id' => $hierarchy_entry->taxon_concept_id, 'source_url' => $source_url);
self::compress_array($this->taxon_ids_inserted[$taxon_id]);
}
if (!isset($this->entry_references_deleted[$hierarchy_entry->id])) {
$hierarchy_entry->delete_refs();
$this->entry_references_deleted[$hierarchy_entry->id] = true;
}
if (!isset($this->entry_vernacular_names_deleted[$hierarchy_entry->id])) {
$this->mysqli->delete("DELETE FROM synonyms WHERE hierarchy_entry_id={$hierarchy_entry->id} AND hierarchy_entry_id={$hierarchy_entry->id} AND hierarchy_id=" . $this->harvest_event->resource->hierarchy_id . " AND language_id!=0 AND language_id!=" . Language::find_or_create_for_parser('scientific name')->id);
$this->entry_vernacular_names_deleted[$hierarchy_entry->id] = true;
}
if (!isset($this->entry_synonyms_deleted[$hierarchy_entry->id])) {
$hierarchy_entry->delete_synonyms();
$this->entry_synonyms_deleted[$hierarchy_entry->id] = true;
}
if ($name_published_in = @$row['http://rs.tdwg.org/dwc/terms/namePublishedIn']) {
$individual_references = explode("||", $name_published_in);
foreach ($individual_references as $reference_string) {
$reference = Reference::find_or_create_by_full_reference(trim($reference_string));
if (@$reference->id) {
$hierarchy_entry->add_reference($reference->id);
$this->mysqli->query("UPDATE refs SET published=1, visibility_id=" . Visibility::visible()->id . " WHERE id={$reference->id}");
}
}
}
// keep track of reference foreign keys
self::append_foreign_keys_from_row($row, 'http://eol.org/schema/reference/referenceID', $this->taxon_reference_ids, $hierarchy_entry->id);
if (isset($this->synonyms[$taxon_id])) {
foreach ($this->synonyms[$taxon_id] as $synonym_row) {
self::uncompress_array($synonym_row);
$synonym_scientific_name = @self::field_decode($synonym_row['http://rs.tdwg.org/dwc/terms/scientificName']);
$synonym_authorship = @self::field_decode($synonym_row['http://rs.tdwg.org/dwc/terms/scientificNameAuthorship']);
if ($synonym_authorship && stripos($synonym_scientific_name, $synonym_authorship) === false) {
$synonym_scientific_name = trim($synonym_scientific_name . " " . $synonym_authorship);
}
if (!$synonym_scientific_name) {
continue;
}
$synonym_taxon_id = @self::field_decode($synonym_row['http://rs.tdwg.org/dwc/terms/taxonID']);
if (!$synonym_taxon_id) {
$taxon_id = @self::field_decode($synonym_row['http://purl.org/dc/terms/identifier']);
}
if (!$synonym_taxon_id) {
continue;
}
$synonym_name = Name::find_or_create_by_string(ucfirst($synonym_scientific_name));
if (@(!$synonym_name->id)) {
continue;
}
$taxonomic_status = @self::field_decode($synonym_row['http://rs.tdwg.org/dwc/terms/taxonomicStatus']) ?: 'synonym';
if (isset($synonym_row['http://rs.tdwg.org/dwc/terms/taxonRemarks'])) {
$taxon_remarks = @self::field_decode($synonym_row['http://rs.tdwg.org/dwc/terms/taxonRemarks']);
} else {
$taxon_remarks = NULL;
}
$synonym_relation = SynonymRelation::find_or_create_by_translated_label($taxonomic_status);
$hierarchy_entry->add_synonym($synonym_name->id, @$synonym_relation->id ?: 0, 0, 0, 0, 0, $taxon_remarks);
}
unset($this->synonyms[$taxon_id]);
}
// COL exception
if ($dataset_id && isset($this->dataset_metadata[$dataset_id]) && ($metadata = $this->dataset_metadata[$dataset_id])) {
$hierarchy_entry->delete_agents();
$agent_name = $metadata['title'];
if ($editors = $metadata['editors']) {
$agent_name .= " by {$editors}";
}
$params = array("full_name" => $agent_name, "agent_role" => AgentRole::find_or_create_by_translated_label('Source'));
$agent = Agent::find_or_create($params);
$hierarchy_entry->add_agent($agent->id, @$a['agent_role']->id ?: 0, 0);
$reference = Reference::find_or_create(array("full_reference" => $metadata['citation']));
$this->mysqli->insert("INSERT IGNORE INTO hierarchy_entries_refs (hierarchy_entry_id, ref_id) VALUES ({$hierarchy_entry->id}, {$reference->id})");
$this->mysqli->query("UPDATE refs SET published=1, visibility_id=" . Visibility::visible()->id . " WHERE id={$reference->id}");
}
$parameters = array('archive_table_definition' => (object) array('row_type' => 'http://rs.tdwg.org/dwc/terms/Taxon'));
$this->insert_data($row, $parameters);
if (isset($this->children[$taxon_id])) {
// set the ancestry for its children
if ($ancestry) {
$this_ancestry = $ancestry . "|" . $name->id;
} else {
$this_ancestry = $name->id;
}
foreach ($this->children[$taxon_id] as &$row) {
self::uncompress_array($row);
$this->add_hierarchy_entry($row, $hierarchy_entry->id, $this_ancestry, $branch_kingdom);
}
unset($this->children[$taxon_id]);
}
unset($hierarchy_entry);
unset($row);
}
public static function read_taxon_xml($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;
}
