public function __construct()
{
$this->mysqli =& $GLOBALS['mysqli_connection'];
if ($GLOBALS['ENV_NAME'] == 'production' && environment_defined('slave')) {
$this->mysqli_slave = load_mysql_environment('slave');
} else {
$this->mysqli_slave =& $this->mysqli;
}
$this->sparql_client = SparqlClient::connection();
$this->published_id = TranslatedResourceStatus::find_or_create_by_label('Published')->id;
$this->trusted_id = Vetted::trusted()->id;
$this->unknown_id = Vetted::unknown()->id;
$this->untrusted_id = Vetted::untrusted()->id;
$this->visible_id = Visibility::visible()->id;
$this->invisible_id = Visibility::invisible()->id;
$this->master_curator_id = CuratorLevel::master_curator()->id;
$this->full_curator_id = CuratorLevel::full_curator()->id;
$this->assistant_curator_id = CuratorLevel::assistant_curator()->id;
$this->curator_ids = CuratorLevel::curator_ids();
$this->data_object_scope = ChangeableObjectType::data_object_scope();
$this->worms_content_partner_id = ContentPartner::find_or_create_by_full_name('World Register of Marine Species')->id;
$this->col_hierarchy_id = Hierarchy::find_or_create_by_label('Species 2000 & ITIS Catalogue of Life: Annual Checklist 2011')->id;
$this->latest_harvest_event_ids();
$this->worms_latest_harvest_event_id();
}
private function add_string_types($taxon_id, $label, $value)
{
$m = new \eol_schema\MeasurementOrFact();
$occurrence = $this->add_occurrence($taxon_id, $label);
$m->occurrenceID = $occurrence->occurrenceID;
if (in_array($label, array("Prey", "Predator", "Pollen", "Visits", "Distribution"))) {
$m->measurementOfTaxon = 'true';
}
if ($label == "Occurrence ID") {
$m->measurementType = "http://rs.tdwg.org/dwc/terms/occurrenceID";
} elseif ($label == "Distribution") {
$m->measurementType = "http://rs.tdwg.org/ontology/voc/SPMInfoItems#Distribution";
} else {
$m->measurementType = "http://hymenoptera.org/" . SparqlClient::to_underscore($label);
}
$m->measurementValue = $value;
$m->contributor = 'Catalog of Hymenoptera in America North of Mexico';
$this->archive_builder->write_object_to_file($m);
}
private function add_string_types($rec, $label, $value, $measurementType, $family)
{
$taxon_id = (string) $rec["taxon_id"];
$object_id = (string) $rec["object_id"];
$m = new \eol_schema\MeasurementOrFact();
$occurrence = $this->add_occurrence($taxon_id, $object_id);
$m->occurrenceID = $occurrence->occurrenceID;
$m->measurementOfTaxon = 'true';
$m->source = @$rec["source"];
if ($val = $measurementType) {
$m->measurementType = $val;
} else {
$m->measurementType = "http://ggbn.org/" . SparqlClient::to_underscore($label);
}
$m->measurementValue = (string) $value;
if (!isset($this->measurement_ids[$m->occurrenceID])) {
$this->archive_builder->write_object_to_file($m);
$this->measurement_ids[$m->occurrenceID] = '';
}
}
private function prepare_turtle($row, $row_class_name, $resource_uri, $entry_uri)
{
$row_type = $row_class_name::ROW_TYPE;
$primary_key = @self::field_decode($row[$row_class_name::PRIMARY_KEY]);
$graph_name = $this->harvest_event->resource->virtuoso_graph_name();
// # TODO: taxon id of the partner will not be globally unique
// # may this is a reason we will need to do represent the data in a custom way
if ($primary_key) {
$node_uri = $graph_name . "/" . $row_class_name::GRAPH_NAME . "/" . SparqlClient::to_underscore($primary_key);
} else {
$node_uri = $graph_name . "/" . $row_class_name::GRAPH_NAME . "/" . md5(serialize($row));
}
$turtle = "<{$node_uri}> a <{$row_type}>";
$simple_turtle = "";
foreach ($row as $key => $value) {
$value = @self::field_decode($value);
if ($value && preg_match("/^http:\\/\\//", $key, $arr)) {
if ($key == $row_class_name::PRIMARY_KEY) {
continue;
}
if ($key == "http://rs.tdwg.org/dwc/terms/taxonID") {
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($graph_name . "/taxa/" . SparqlClient::to_underscore($value)) . "\n";
if ($row_type == 'http://rs.tdwg.org/dwc/terms/Occurrence') {
$this->occurrence_taxon_mapping[$primary_key] = @self::field_decode($value);
}
} elseif ($key == "http://eol.org/schema/targetTaxonID") {
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($graph_name . "/taxa/" . SparqlClient::to_underscore($value)) . "\n";
if ($row_type == 'http://rs.tdwg.org/dwc/terms/targetOccurrenceID') {
$this->occurrence_taxon_mapping[$primary_key] = @self::field_decode($value);
}
} elseif ($key == "http://rs.tdwg.org/dwc/terms/eventID") {
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($graph_name . "/events/" . SparqlClient::to_underscore($value)) . "\n";
} elseif ($key == "http://rs.tdwg.org/dwc/terms/occurrenceID" || $key == "http://eol.org/schema/targetOccurrenceID") {
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($graph_name . "/occurrences/" . SparqlClient::to_underscore($value)) . "\n";
if ($key == "http://rs.tdwg.org/dwc/terms/occurrenceID" && ($row_type == "http://rs.tdwg.org/dwc/terms/MeasurementOrFact" || $row_type == 'http://eol.org/schema/Association')) {
$occurrence_id = $value;
} elseif ($row_type == 'http://eol.org/schema/Association' && $key == "http://eol.org/schema/targetOccurrenceID") {
$object = $value;
}
} elseif ($key == "http://eol.org/schema/associationID") {
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($graph_name . "/associations/" . SparqlClient::to_underscore($value)) . "\n";
} elseif ($key == "http://eol.org/schema/parentMeasurementID") {
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($graph_name . "/measurements/" . SparqlClient::to_underscore($value)) . "\n";
} elseif ($key == "http://eol.org/schema/reference/referenceID") {
$reference_ids = self::get_foreign_keys_from_row($row, 'http://eol.org/schema/reference/referenceID');
foreach ($reference_ids as $reference_id) {
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($graph_name . "/references/" . SparqlClient::to_underscore($reference_id)) . "\n";
}
} elseif ($row_type == "http://rs.tdwg.org/dwc/terms/MeasurementOrFact" && $key == "http://rs.tdwg.org/dwc/terms/measurementType") {
$predicate = $value;
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($value) . "\n";
} elseif ($row_type == "http://eol.org/schema/Association" && $key == "http://eol.org/schema/associationType") {
$predicate = $value;
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($value) . "\n";
} elseif ($row_type == "http://rs.tdwg.org/dwc/terms/MeasurementOrFact" && $key == "http://rs.tdwg.org/dwc/terms/measurementValue") {
$object = $value;
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($value) . "\n";
} elseif ($row_type == "http://rs.tdwg.org/dwc/terms/MeasurementOrFact" && $key == "http://rs.tdwg.org/dwc/terms/measurementUnit") {
$unit_of_measure = $value;
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($value) . "\n";
} else {
$turtle .= "; " . SparqlClient::enclose_value($key) . " " . SparqlClient::enclose_value($value) . "\n";
}
}
}
if ($row_type == "http://rs.tdwg.org/dwc/terms/MeasurementOrFact" || $row_type == "http://eol.org/schema/Association") {
if (isset($predicate) && isset($object)) {
$trait_id = md5(serialize($row));
$trait_uri = "{$resource_uri}/traits/{$trait_id}";
$simple_turtle = "{$trait_uri} a eol:trait ;\n";
$simple_turtle .= " eol:predicate {$predicate} ;\n";
$simple_turtle .= " eol:recognized_as {$entry_uri} ;\n";
$simple_turtle .= " eol:value {$object} ;\n";
$simple_turtle .= " eol:source {$resource_uri} .\n";
if (isset($unit_of_measure)) {
$simple_turtle .= " eol:units {$unit_of_measure} .\n";
}
$simple_turtle .= "{$entry_uri} {$predicate} {$trait_uri} .\n";
}
// prepare data point uri attributes
$data_point_uri = array("uri" => $node_uri, "created_at" => 'NOW()', "updated_at" => 'NOW()', "resource_id" => $this->harvest_event->resource->id);
if (isset($predicate)) {
$data_point_uri["predicate"] = $predicate;
}
if (isset($object)) {
$data_point_uri["object"] = $object;
}
if (isset($unit_of_measure)) {
$data_point_uri["unit_of_measure"] = $unit_of_measure;
}
if (isset($occurrence_id)) {
$data_point_uri["occurrence_id"] = $occurrence_id;
}
} else {
$data_point_uri = NULL;
}
return array($turtle, $data_point_uri, $simple_turtle);
}
private function add_string_types($line_data, $taxon)
{
$string_types = array('Biogeographic Regions', 'Aquatic Biomes', 'Wetlands', 'Other Habitat Features', 'Primary Diet', 'Animal Foods', 'Plant Foods', 'Other Foods');
foreach ($string_types as $label) {
if ($v = trim($line_data[$this->column_indices[$label]])) {
$values = explode(";", $v);
foreach ($values as $value) {
$value = trim($value);
$this_label = $label;
$m = new \eol_schema\MeasurementOrFact();
$m->measurementOfTaxon = 'true';
if ($this_label == 'Biogeographic Regions') {
$qualification = null;
if (preg_match("/^(.+) :: (.+)\$/", $value, $arr)) {
$value = $arr[1];
$qualification = $arr[2];
}
$this_label = trim("{$qualification} range");
} elseif ($this_label == 'Primary Diet') {
if (preg_match("/ :: (.+)\$/", $value, $arr)) {
$value = $arr[1];
}
}
print_r($taxon);
print "\n value = [{$value}]";
exit;
// print "\n [$taxon->taxonID] [$value]";
$occurrence = $this->add_occurrence($taxon, $value);
$m->occurrenceID = $occurrence->occurrenceID;
$m->measurementType = "http://adw.org/" . SparqlClient::to_underscore($this_label);
$m->measurementValue = "http://adw.org/" . SparqlClient::to_underscore($value);
$this->archive_builder->write_object_to_file($m);
}
}
}
}
private function get_distributions($taxon_id, $sciname)
{
$xml = self::create_cache("distribution", $taxon_id);
$xml = simplexml_load_string($xml);
$lines = array();
foreach ($xml->Distribution as $rec) {
if (!isset($rec->Location->CountryName)) {
continue;
}
$line = trim($rec->Location->CountryName) . trim($rec->Location->RegionName) . trim($rec->Location->UpperName);
if (in_array($line, $lines)) {
continue;
}
$lines[] = $line;
$citation = trim($rec->Reference->FullCitation);
$ref_url = TROPICOS_DOMAIN . "/Reference/" . trim($rec->Reference->ReferenceId);
// $agents = array();
// $agents[] = array("role" => "source", "homepage" => "http://www.tropicos.org", "fullName" => "Tropicos");
// $agent_ids = self::create_agents($agents);
$text_id = $rec->Location->LocationID;
$region = "";
if ($RegionName = trim($rec->Location->RegionName)) {
$region .= $RegionName;
}
if ($CountryName = trim($rec->Location->CountryName)) {
$region .= " - " . $CountryName;
}
if ($UpperName = trim($rec->Location->UpperName)) {
$region .= " - " . $UpperName;
}
self::add_string_types($taxon_id, $text_id, "Distribution", $region, "http://rs.tdwg.org/ontology/voc/SPMInfoItems#Distribution", true);
self::add_string_types($taxon_id, $text_id, "Taxon", $sciname, "http://rs.tdwg.org/dwc/terms/scientificName");
self::add_string_types($taxon_id, $text_id, "Country", trim($rec->Location->CountryName), "http://rs.tdwg.org/dwc/terms/country");
self::add_string_types($taxon_id, $text_id, "Continent", trim($rec->Location->RegionName), "http://rs.tdwg.org/dwc/terms/continent");
if ($upper_name = @$rec->Location->UpperName) {
self::add_string_types($taxon_id, $text_id, "Upper name", $upper_name, "http://tropicos.org/" . SparqlClient::to_underscore("Upper name"));
}
}
// $title = "Localities documented in Tropicos sources";
// $subject = "http://rs.tdwg.org/ontology/voc/SPMInfoItems#Distribution";
// $license = "http://creativecommons.org/licenses/by-nc-sa/3.0/";
}
private function add_string_types($measurementOfTaxon, $rec, $label, $value, $mtype, $reference_ids = array())
{
$taxon_id = $rec["schema_taxon_id"];
$catnum = $rec["catnum"];
$m = new \eol_schema\MeasurementOrFact();
$occurrence = $this->add_occurrence($taxon_id, $catnum);
$m->occurrenceID = $occurrence->occurrenceID;
if ($mtype) {
$m->measurementType = $mtype;
} else {
$m->measurementType = "http://domain.org/" . SparqlClient::to_underscore($label);
}
// currently won't pass here
$m->measurementValue = $value;
if ($val = $measurementOfTaxon) {
$m->measurementOfTaxon = $val;
$m->source = $rec["source"];
if ($reference_ids) {
$m->referenceID = implode("; ", $reference_ids);
}
// $m->measurementRemarks = ''; $m->contributor = ''; $m->measurementMethod = '';
}
$this->archive_builder->write_object_to_file($m);
}
private function add_string_types($measurementOfTaxon, $rec, $label, $value, $mtype, $measurementRemarks = null)
{
$taxon_id = $rec["taxon_id"];
$catnum = $rec["catnum"];
$m = new \eol_schema\MeasurementOrFact();
$occurrence = $this->add_occurrence($taxon_id, $catnum);
$m->occurrenceID = $occurrence->occurrenceID;
if ($mtype) {
$m->measurementType = $mtype;
} else {
$m->measurementType = "http://feis.org/" . SparqlClient::to_underscore($label);
}
// currently won't pass here
$m->measurementValue = $value;
if ($val = $measurementOfTaxon) {
$m->measurementOfTaxon = $val;
}
if ($measurementOfTaxon) {
$m->source = $rec["source"];
$m->measurementRemarks = $measurementRemarks;
// not used... $m->contributor, $m->measurementMethod
}
$this->archive_builder->write_object_to_file($m);
}
private function add_associations($line_data)
{
$source_taxon_id = trim($line_data[$this->column_indices['my active taxon id']]);
$target_taxon_id = trim($line_data[$this->column_indices['my passive taxon id']]);
$relationship = trim($line_data[$this->column_indices['active relation']]);
if ($source_taxon_id && $target_taxon_id && $relationship) {
$m = new \eol_schema\Association();
if (!$this->taxon_occurrences[$source_taxon_id]) {
return;
}
if (!$this->taxon_occurrences[$target_taxon_id]) {
return;
}
$m->occurrenceID = $this->taxon_occurrences[$source_taxon_id]->occurrenceID;
$m->associationType = "http://bioinfo.org/" . SparqlClient::to_underscore($relationship);
$m->targetOccurrenceID = $this->taxon_occurrences[$target_taxon_id]->occurrenceID;
$this->archive_builder->write_object_to_file($m);
}
}
private function add_string_types($measurementOfTaxon, $rec, $label, $value, $mtype, $measurementRemarks = null)
{
echo "\n [{$label}]:[{$value}]\n";
$taxon_id = $rec["taxon_id"];
$catnum = $rec["catnum"];
$m = new \eol_schema\MeasurementOrFact();
$occurrence = $this->add_occurrence($taxon_id, $catnum);
$m->occurrenceID = $occurrence->occurrenceID;
if ($mtype) {
$m->measurementType = $mtype;
} else {
$m->measurementType = "http://iucn.org/" . SparqlClient::to_underscore($label);
}
$m->measurementValue = $value;
if ($val = $measurementOfTaxon) {
$m->measurementOfTaxon = $val;
}
if ($measurementOfTaxon) {
$m->source = $rec["source"];
$m->measurementRemarks = $measurementRemarks;
// $m->contributor = '';
// $m->measurementMethod = '';
}
$this->archive_builder->write_object_to_file($m);
}
