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($row);
}
static function force_wikipedia_taxon($t)
{
$wikipedia_resource = Resource::wikipedia();
$last_wikipedia_harvest = new HarvestEvent($wikipedia_resource->most_recent_published_harvest_event_id());
$content_manager = new ContentManager();
$hierarchy_entry = HierarchyEntry::create_entries_for_taxon($t, $wikipedia_resource->hierarchy_id);
if (@(!$hierarchy_entry->id)) {
return false;
}
$last_wikipedia_harvest->add_hierarchy_entry($hierarchy_entry, 'inserted');
foreach ($t['data_objects'] as &$d) {
list($data_object, $status) = DataObject::find_and_compare($wikipedia_resource, $d, $content_manager);
if (@(!$data_object->id)) {
return false;
}
$hierarchy_entry->add_data_object($data_object->id);
$last_wikipedia_harvest->add_data_object($data_object, $status);
if (@$d->info_items_ids) {
$data_object->delete_info_items();
foreach ($d->info_items_ids as &$id) {
$data_object->add_info_item($id);
unset($id);
}
}
return array($hierarchy_entry, $data_object);
}
return false;
}