function get_common_names($names)
{
// might need or not need this...
$common = utf8_encode($name['commonName']);
if (Functions::is_utf8($common)) {
$arr_names[] = array("name" => Functions::import_decode($common), "language" => $name['xml_lang']);
}
}
private function add_higher_level_taxa_to_archive()
{
$exclude = array("kingdom", "phylum", "class", "order", "family", "genus", "species", "subspecies");
foreach ($this->name_id as $taxon_id => $rec) {
if (in_array($rec["sciname"], $exclude)) {
continue;
}
if (!Functions::is_utf8($rec["sciname"])) {
continue;
}
$taxon = new \eol_schema\Taxon();
$taxon->taxonID = (string) $taxon_id;
$taxon->taxonRank = (string) @$rec["rank"];
$taxon->scientificName = (string) Functions::import_decode($rec["sciname"]);
$taxon->parentNameUsageID = $rec["parent"];
if (isset($this->taxon_ids[$taxon_id])) {
continue;
}
if (!$taxon->parentNameUsageID && @$rec["rank"] != "kingdom") {
continue;
}
if (is_numeric(stripos($rec["sciname"], "unassigned"))) {
continue;
}
if (!@$rec["rank"]) {
$this->no_rank++;
// echo "\n wrong data: " . $rec["sciname"] . " [$taxon_id]";
} else {
$this->taxa[$taxon->taxonID] = $taxon;
$this->taxon_ids[$taxon->taxonID] = 1;
}
}
}
private function process_fields($records, $class, $allowed_fields)
{
foreach ($records as $rec) {
if ($class == "VernacularName") {
$c = new \eol_schema\VernacularName();
} elseif ($class == "Agent") {
$c = new \eol_schema\Agent();
} elseif ($class == "Reference") {
$c = new \eol_schema\Reference();
} elseif ($class == "Taxon") {
$c = new \eol_schema\Taxon();
} elseif ($class == "MeasurementOrFact") {
$c = new \eol_schema\MeasurementOrFact();
} elseif ($class == "Occurrence") {
$c = new \eol_schema\Occurrence();
} elseif ($class == "Distribution") {
$c = new \eol_schema\MediaResource();
} elseif ($class == "Image") {
$c = new \eol_schema\MediaResource();
}
$keys = array_keys($rec);
$save = true;
foreach ($keys as $key) {
$temp = pathinfo($key);
$field = $temp["basename"];
/* resource specifications */
// if($this->resource_id == "345") //3I Interactive resource
// if(true)
// {
// if($class == "Image" && $field == "license") $field = "UsageTerms";
// if($class == "Distribution" && $field == "locality") $field = "Description";
// }
/* end specifications */
// manual adjustment bec. of a typo in meta.xml, without "s"
if ($field == "measurementRemark") {
$field = "measurementRemarks";
}
/*
// sample way to exclude if field is to be excluded
if($field == "attribution") continue; //not recognized in eol: http://indiabiodiversity.org/terms/attribution
*/
if (!in_array($field, $allowed_fields)) {
$this->debug["undefined"][$class][$field] = '';
continue;
}
// some fields have '#', e.g. "http://schemas.talis.com/2005/address/schema#localityName"
$parts = explode("#", $field);
if ($parts[0]) {
$field = $parts[0];
}
if (@$parts[1]) {
$field = $parts[1];
}
$value = trim((string) $rec[$key]);
$value = trim(Functions::import_decode($value));
if (!Functions::is_utf8($value)) {
$save = false;
}
//special arrangement
if ($class == "Reference") {
if ($field == "identifier") {
$this->reference_ids[$value] = '';
}
if ($field == "full_reference") {
if (!$value) {
$full_ref = "";
if ($val = (string) @$rec["http://eol.org/schema/reference/primaryTitle"]) {
$full_ref .= $val;
} elseif ($val = (string) @$rec["http://purl.org/dc/terms/title"]) {
$full_ref .= $val;
}
if ($val = (string) @$rec["http://purl.org/ontology/bibo/pageStart"]) {
$full_ref .= ". Page(s) " . $val;
}
if ($val = (string) @$rec["http://purl.org/ontology/bibo/pageEnd"]) {
$full_ref .= " - " . $val;
}
if ($val = (string) @$rec["http://purl.org/ontology/bibo/volume"]) {
$full_ref .= ". Vol. " . $val;
}
if ($val = (string) @$rec["http://purl.org/dc/terms/publisher"]) {
$full_ref .= ". Publisher: " . $val;
}
if ($val = (string) @$rec["http://purl.org/ontology/bibo/authorList"]) {
$full_ref .= ". Author: " . $val;
}
if ($val = (string) @$rec["http://purl.org/ontology/bibo/editorList"]) {
$full_ref .= ". Editor: " . $val;
}
if ($val = (string) @$rec["http://purl.org/dc/terms/created"]) {
$full_ref .= ". " . $val;
}
if ($val = $full_ref) {
$value = $full_ref;
} else {
echo " -- still blank";
}
}
}
}
// if($class == "MeasurementOrFact" && $field == "referenceID")
// {
// if($value)
// {
// if(!isset($this->reference_ids[$value])) echo " -- undefined refid:[$value]";
// }
// }
if ($value) {
$c->{$field} = $value;
}
}
/* if($class == "objects") {} // sample way to filter */
if ($save) {
$this->archive_builder->write_object_to_file($c);
}
}
}
public static function eol_schema_validate($uri)
{
if (!$uri) {
return false;
}
$valid = SchemaValidator::validate($uri);
if ($valid !== true) {
return array();
}
$errors = array();
$warnings = array();
$reader = new \XMLReader();
$reader->open($uri);
$i = 0;
while (@$reader->read()) {
if ($reader->nodeType == \XMLReader::ELEMENT && $reader->name == "taxon") {
$taxon_xml = $reader->readOuterXML();
$t = simplexml_load_string($taxon_xml, null, LIBXML_NOCDATA);
$t_dc = $t->children("http://purl.org/dc/elements/1.1/");
$t_dwc = $t->children("http://rs.tdwg.org/dwc/dwcore/");
$identifier = Functions::import_decode($t_dc->identifier);
$source_url = Functions::import_decode($t_dc->source);
$scientific_name = Functions::import_decode($t_dwc->ScientificName);
if (!$identifier) {
$warnings[$scientific_name]["taxon without dc:identifier"] = true;
}
if (!$source_url) {
$warnings[$scientific_name]["taxon without dc:source"] = true;
}
foreach ($t->dataObject as $d) {
$d_dc = $d->children("http://purl.org/dc/elements/1.1/");
$identifier = Functions::import_decode($d_dc->identifier);
/* Checking requirements*/
if (!$identifier) {
$warnings[$scientific_name]["data object without dc:identifier"] = true;
}
}
$xml->taxon[$i] = null;
$i++;
}
}
return array($errors, $warnings);
}
function get_data_object($rec)
{
$data_object_parameters = array();
$data_object_parameters["identifier"] = trim(@$rec["identifier"]);
$data_object_parameters["source"] = $rec["source"];
$data_object_parameters["dataType"] = trim($rec["dataType"]);
$data_object_parameters["mimeType"] = trim($rec["mimeType"]);
$data_object_parameters["mediaURL"] = trim(@$rec["mediaURL"]);
$data_object_parameters["thumbnailURL"] = trim(@$rec["thumbnailURL"]);
$data_object_parameters["created"] = trim(@$rec["created"]);
$data_object_parameters["description"] = Functions::import_decode(@$rec["description"]);
$data_object_parameters["source"] = @$rec["source"];
$data_object_parameters["license"] = @$rec["license"];
$data_object_parameters["rightsHolder"] = @trim($rec["rightsHolder"]);
$data_object_parameters["title"] = @trim($rec["title"]);
$data_object_parameters["language"] = "en";
//==========================================================================================
$agents = array();
foreach (@$rec["agent"] as $agent) {
$agentParameters = array();
$agentParameters["role"] = $agent["role"];
$agentParameters["homepage"] = $agent["homepage"];
$agentParameters["logoURL"] = "";
$agentParameters["fullName"] = $agent["fullName"];
$agents[] = new \SchemaAgent($agentParameters);
}
$data_object_parameters["agents"] = $agents;
//==========================================================================================
return $data_object_parameters;
}
function get_data_object($desc, $do_identifier, $subject, $dataType, $mimeType, $title, $source, $do_agents, $rightsHolder, $mediaURL = NULL)
{
$dataObjectParameters = array();
$dataObjectParameters["identifier"] = $do_identifier;
$dataObjectParameters["dataType"] = $dataType;
$dataObjectParameters["mimeType"] = $mimeType;
$dataObjectParameters["title"] = $title;
$dataObjectParameters["language"] = "en";
$dataObjectParameters["description"] = $desc;
if ($subject != "") {
$dataObjectParameters["subjects"] = array();
$subjectParameters = array();
$subjectParameters["label"] = $subject;
$dataObjectParameters["subjects"][] = new \SchemaSubject($subjectParameters);
}
//if($mimeType == "text/html")
//{
$agents = array();
foreach ($do_agents as $agent) {
$agentParameters = array();
$agentParameters["role"] = $agent["role"];
$agentParameters["homepage"] = "http://emuweb.fieldmuseum.org/botany/botanytaxon.php";
//$agentParameters["logoURL"] = $agent["logoURL"];
//$agentParameters["fullName"] = Functions::import_decode($agent["name"]);
$agentParameters["fullName"] = utf8_encode($agent["name"]);
$agents[] = new \SchemaAgent($agentParameters);
}
$dataObjectParameters["agents"] = $agents;
//}
$dataObjectParameters["license"] = "http://creativecommons.org/licenses/by-nc-sa/3.0/";
$dataObjectParameters["source"] = $source;
$dataObjectParameters["rightsHolder"] = Functions::import_decode($rightsHolder);
/*
$dataObjectParameters["created"] = $do->created;
$dataObjectParameters["modified"] = $do->modified;
$dataObjectParameters["thumbnailURL"] = $do->thumbnailURL;
$dataObjectParameters["location"] = Functions::import_decode($do->location);
*/
if ($mimeType != "text/html") {
$dataObjectParameters["mediaURL"] = $mediaURL;
}
///////////////////////////////////
$dataObjectParameters["audiences"] = array();
$audienceParameters = array();
$audienceParameters["label"] = "Expert users";
$dataObjectParameters["audiences"][] = new \SchemaAudience($audienceParameters);
$audienceParameters["label"] = "General public";
$dataObjectParameters["audiences"][] = new \SchemaAudience($audienceParameters);
///////////////////////////////////
return $dataObjectParameters;
}
private function format_utf8($string)
{
$string = Functions::import_decode($string);
if (!Functions::is_utf8($string)) {
return utf8_encode($string);
}
return $string;
}
function get_data_object($rec)
{
$data_object_parameters = array();
$data_object_parameters["identifier"] = trim(@$rec["identifier"]);
$data_object_parameters["source"] = $rec["source"];
$data_object_parameters["dataType"] = trim($rec["dataType"]);
$data_object_parameters["mimeType"] = trim($rec["mimeType"]);
$data_object_parameters["mediaURL"] = trim(@$rec["mediaURL"]);
$data_object_parameters["created"] = trim(@$rec["created"]);
$data_object_parameters["source"] = $rec["source"];
$data_object_parameters["description"] = Functions::import_decode($rec["description"]);
$data_object_parameters["location"] = Functions::import_decode($rec["location"]);
$data_object_parameters["license"] = $rec["license"];
$data_object_parameters["rightsHolder"] = trim($rec["rightsHolder"]);
$data_object_parameters["title"] = @trim($rec["title"]);
$data_object_parameters["language"] = "en";
//==========================================================================================
if (trim($rec["subject"])) {
$data_object_parameters["subjects"] = array();
$subjectParameters = array();
$subjectParameters["label"] = trim($rec["subject"]);
$data_object_parameters["subjects"][] = new \SchemaSubject($subjectParameters);
}
//==========================================================================================
$agents = array();
foreach (@$rec["agent"] as $agent) {
$agentParameters = array();
$agentParameters["role"] = $agent["role"];
$agentParameters["homepage"] = $agent["homepage"];
$agentParameters["logoURL"] = "";
$agentParameters["fullName"] = $agent[0];
$agents[] = new \SchemaAgent($agentParameters);
}
$data_object_parameters["agents"] = $agents;
//==========================================================================================
$data_object_parameters["references"] = array();
$ref = array();
foreach ($rec["object_refs"] as $r) {
if (!$r["ref"]) {
continue;
}
$referenceParameters = array();
$referenceParameters["fullReference"] = Functions::import_decode($r["ref"]);
if ($r["url"]) {
$referenceParameters["referenceIdentifiers"][] = new \SchemaReferenceIdentifier(array("label" => "url", "value" => trim($r["url"])));
}
$ref[] = new \SchemaReference($referenceParameters);
}
$data_object_parameters["references"] = $ref;
//==========================================================================================
return $data_object_parameters;
}
private static function clean_string($str, $remove_whitespace = false, $decode = true)
{
$str = Functions::import_decode(trim($str), $remove_whitespace, $decode);
$str = str_replace(" ", " ", $str);
$str = str_replace("\t", " ", $str);
return trim($str);
}
function get_data_object($do, $t_dc2, $t_dcterms)
{
$dataObjectParameters = array();
$dataObjectParameters["identifier"] = $t_dc2->identifier;
$dataObjectParameters["dataType"] = $do->dataType;
$dataObjectParameters["mimeType"] = $do->mimeType;
$description = str_ireplace('<strong>', '', $t_dc2->description);
$description = str_ireplace('</strong>', '', $description);
$dataObjectParameters["description"] = "<strong>{$t_dc2->title}</strong> <br> {$description}";
$dataObjectParameters["title"] = "Functional adaptation";
$dataObjectParameters["language"] = $t_dc2->language;
//this overwrites whatever is given by AskNature, currently it is GenDesc
$dataObjectParameters["subjects"] = array();
$subjectParameters = array();
$subjectParameters["label"] = "http://rs.tdwg.org/ontology/voc/SPMInfoItems#Evolution";
$dataObjectParameters["subjects"][] = new \SchemaSubject($subjectParameters);
$dataObjectParameters["additionalInformation"] = "<subject>http://www.eol.org/voc/table_of_contents#FunctionalAdaptations</subject>";
///////////////////////////////////////////////////////////////////////
$agents = array();
foreach ($do->agent as $agent) {
$agentParameters = array();
$agentParameters["role"] = $agent["role"];
$agentParameters["homepage"] = $agent["homepage"];
$agentParameters["logoURL"] = $agent["logoURL"];
$agentParameters["fullName"] = Functions::import_decode($agent);
$agents[] = new \SchemaAgent($agentParameters);
}
$dataObjectParameters["agents"] = $agents;
///////////////////////////////////////////////////////////////////////
$dataObjectParameters["created"] = $do->created;
$dataObjectParameters["modified"] = $do->modified;
$dataObjectParameters["license"] = $do->license;
$dataObjectParameters["rights"] = $do->rights;
$dataObjectParameters["rightsHolder"] = Functions::import_decode($t_dcterms->rightsHolder);
$dataObjectParameters["source"] = $t_dc2->source;
$dataObjectParameters["mediaURL"] = $do->mediaURL;
$dataObjectParameters["thumbnailURL"] = $do->thumbnailURL;
$dataObjectParameters["location"] = Functions::import_decode($do->location);
///////////////////////////////////////////////////////////////////////
$dataObjectParameters["audiences"] = array();
$audienceParameters = array();
$audienceParameters["label"] = $do->audience;
$dataObjectParameters["audiences"][] = new \SchemaAudience($audienceParameters);
///////////////////////////////////////////////////////////////////////
///////////////////////////////////////////////////////////////////////
$dataObjectParameters["references"] = array();
$refs = array();
foreach ($do->reference as $ref) {
$referenceParameters = array();
$referenceParameters["fullReference"] = $ref;
if (@$agent["URL"] || @$agent["ISBN"]) {
$referenceParameters["referenceIdentifiers"][] = new \SchemaReferenceIdentifier(array("label" => "url", "value" => self::format(@$agent["URL"])));
$referenceParameters["referenceIdentifiers"][] = new \SchemaReferenceIdentifier(array("label" => "isbn", "value" => self::format(@$agent["ISBN"])));
}
$refs[] = new \SchemaReference($referenceParameters);
}
$dataObjectParameters["references"] = $refs;
///////////////////////////////////////////////////////////////////////
return $dataObjectParameters;
}
public static function eol_schema_validate($uri)
{
if (!$uri) {
return false;
}
$valid = SchemaValidator::validate($uri);
if ($valid !== true) {
return array();
}
$errors = array();
$warnings = array();
$reader = new \XMLReader();
$reader->open($uri);
$i = 0;
while (@$reader->read()) {
if ($reader->nodeType == \XMLReader::ELEMENT && $reader->name == "taxon") {
$taxon_xml = $reader->readOuterXML();
$t = simplexml_load_string($taxon_xml, null, LIBXML_NOCDATA);
$t_dc = $t->children("http://purl.org/dc/elements/1.1/");
$t_dwc = $t->children("http://rs.tdwg.org/dwc/dwcore/");
$identifier = Functions::import_decode($t_dc->identifier);
$source_url = Functions::import_decode($t_dc->source);
$scientific_name = Functions::import_decode($t_dwc->ScientificName);
if (!$identifier) {
$warnings[$scientific_name]["taxon without dc:identifier"] = true;
}
if (!$source_url) {
$warnings[$scientific_name]["taxon without dc:source"] = true;
}
foreach ($t->dataObject as $d) {
$d_dc = $d->children("http://purl.org/dc/elements/1.1/");
$identifier = Functions::import_decode($d_dc->identifier);
$data_type = Functions::import_decode($d->dataType);
$license = Functions::import_decode($d->license);
$source_url = Functions::import_decode($d_dc->source);
$description = Functions::import_decode($d_dc->description, 0, 0);
$object_url = Functions::import_decode($d->mediaURL);
$subjects = array();
foreach ($d->subject as $s) {
$subjects[] = trim((string) $s);
}
/* Checking requirements*/
if (!$identifier) {
$warnings[$scientific_name]["data object without dc:identifier"] = true;
}
if (!$license) {
$warnings[$scientific_name]["data object without license"] = true;
}
//if text: must have description
if ($data_type == "http://purl.org/dc/dcmitype/Text" && !$description) {
$errors[$scientific_name]["text without dc:description"] = true;
}
//if text: must have subject
if ($data_type == "http://purl.org/dc/dcmitype/Text" && !$subjects) {
$errors[$scientific_name]["text without subject"] = true;
}
//if image, movie or sound: must have object_url
if ($data_type != "http://purl.org/dc/dcmitype/Text" && !$object_url) {
$errors[$scientific_name]["media without mediaURL"] = true;
}
}
//unset($xml->taxon[$i]);
$xml->taxon[$i] = null;
$i++;
//if($i%100==0 && DEBUG) debug("Parsed taxon $i");
//if(defined("DEBUG_PARSE_TAXON_LIMIT") && $i >= DEBUG_PARSE_TAXON_LIMIT) break;
}
}
return array($errors, $warnings);
}
function get_common_names($names)
{
$arr_names = array();
if ($names) {
foreach ($names as $name) {
$common = utf8_encode($name['commonName']);
if (Functions::is_utf8($common)) {
$arr_names[] = array("name" => Functions::import_decode($common), "language" => $name['xml_lang']);
} else {
echo "\n not utf8 common name: [" . $common . "]\n";
}
}
}
return $arr_names;
}
function get_references($file)
{
$str = Functions::get_remote_file($file);
$str = str_replace(array("\n", "\r", "\t", "\\o", "\\xOB"), '', $str);
$str = str_replace(array("<nobr>", "</nobr>", "taxon_view.cfm?mode=bibliography&citation=", "̈"), '', $str);
$str = str_ireplace('‘', "'", $str);
//special char [�]
$str = str_ireplace('’', "'", $str);
//special char [�]
$str = str_ireplace('“', '"', $str);
//special char [�]
$str = str_ireplace('”', '"', $str);
//special char [�]
$str = str_ireplace('–', '-', $str);
//special char [�]
$str = str_ireplace('́', "'", $str);
//special char [?]
$str = str_ireplace('&', "and", $str);
$str = Functions::import_decode($str);
$str = str_ireplace('<p class="biblio">', 'xxx', $str);
$str = str_ireplace('xxx', "&arr[]=", $str);
$arr = array();
parse_str($str);
$arr_ref = array();
$arr_taxon_ref = array();
$i = 1;
foreach ($arr as $r) {
print "\n {$i} of " . count($arr) . "\n";
if (is_numeric(stripos($r, '<a href="'))) {
if (preg_match("/<a href=\"(.*?)\">/", $r, $matches)) {
$ref_num = $matches[1];
}
$file = $GLOBALS['cited_taxa_prefix'] . $ref_num;
$str = Functions::get_remote_file($file);
//<input type="checkbox"
$str = str_ireplace('<input type="checkbox"', 'xxx<input type="checkbox"', $str);
$str = str_ireplace('xxx', "&arr2[]=", $str);
$arr2 = array();
parse_str($str);
foreach ($arr2 as $r2) {
//onclick="reMap('8418','Chlamys gemmulata')"
if (preg_match("/onclick=\"reMap(.*?)\"/", $r2, $matches)) {
$sub_str = $matches[1];
}
//('6577','Galiteuthis glacialis')
if (preg_match("/\\(\\'(.*?)\\'\\,/", $sub_str, $matches)) {
$taxon_id = $matches[1];
}
if (preg_match("/\\,\\'(.*?)\\'\\)/", $sub_str, $matches)) {
$taxon = $matches[1];
}
print "[{$taxon_id}][{$taxon}]\n";
//$arr_taxon_ref[$taxon_id][]=$ref_num;
$arr_taxon_ref["{$taxon_id}"][] = $ref_num;
}
$arr_ref[$ref_num] = $str = trim(str_ireplace('View cited taxa', "", strip_tags($r, "<em>")));
$i++;
//if($i == 3)break;//debug - to limit no. of records
}
}
return array(0 => $arr_ref, 1 => $arr_taxon_ref);
}
private function get_contributor_name($url)
{
$options = array('resource_id' => 'gbif', 'expire_seconds' => false, 'download_wait_time' => 1000000, 'timeout' => 900, 'download_attempts' => 2, 'delay_in_minutes' => 2);
// 15mins timeout
$options = array('resource_id' => 'gbif', 'expire_seconds' => false, 'download_wait_time' => 1000000, 'timeout' => 900);
// 15mins timeout
if ($html = Functions::lookup_with_cache($url, $options)) {
// <title property="dc:title">Herbarium Berolinense - Dataset detail</title>
if (preg_match("/\"dc:title\">(.*?)\\- Dataset detail/ims", $html, $arr)) {
if (!Functions::is_utf8($arr[1])) {
exit("\n culprit is contributor name \n");
}
return Functions::import_decode(trim($arr[1]));
}
}
}
private function process_text($rec, $source_url, $description, $subject, $reference_ids = array())
{
$identifier = self::get_identifier($source_url);
$description = Functions::import_decode($description);
if (!Functions::is_utf8($description)) {
return;
}
$mr = new \eol_schema\MediaResource();
if ($reference_ids) {
$mr->referenceID = implode("; ", $reference_ids);
}
if ($this->agent_ids) {
$mr->agentID = implode("; ", $this->agent_ids);
}
$mr->taxonID = (string) $rec['taxon_id'];
$mr->identifier = $identifier;
$mr->type = "http://purl.org/dc/dcmitype/Text";
$mr->language = 'en';
$mr->format = "text/html";
$mr->furtherInformationURL = (string) $source_url;
$mr->CVterm = $subject;
$mr->Owner = $this->rights_holder;
$mr->title = "";
$mr->UsageTerms = "http://creativecommons.org/licenses/by-nc-sa/3.0/";
$mr->audience = 'Everyone';
$mr->description = (string) $description;
if (!in_array($mr->identifier, $this->object_ids)) {
$this->object_ids[] = $mr->identifier;
$this->archive_builder->write_object_to_file($mr);
}
}
private function process_mushroom_observer_list($wrong_urls)
{
if ($file = Functions::lookup_with_cache($this->mushroom_observer_eol, $this->download_options)) {
$xml = simplexml_load_string($file);
$i = 0;
$total = count($xml->taxon);
foreach ($xml->taxon as $t) {
$i++;
// if($i > 40) break; //debug
$t_dwc = $t->children("http://rs.tdwg.org/dwc/dwcore/");
$t_dc = $t->children("http://purl.org/dc/elements/1.1/");
$sciname = Functions::import_decode($t_dwc->ScientificName);
$sciname = Functions::canonical_form($sciname);
echo "\n{$i} of {$total}: {$sciname}";
$url = "http://en.wikipedia.org/wiki/" . str_replace(" ", "_", $sciname);
self::get_triple($url, $wrong_urls);
}
}
}
public function get_synth_tags($html)
{
$synth_tags = array();
if (preg_match("/<div id=\"tagCloud\">(.*?)<\\/div>/ims", $html, $arr)) {
if (preg_match_all("/aspx\\?q=(.*?)\">/", $arr[1], $tags, PREG_SET_ORDER)) {
foreach ($tags as $tag) {
$synth_tags[] = Functions::import_decode($tag[1]);
}
}
}
return $synth_tags;
}
function get_data_object($do, $t_dc2, $t_dcterms)
{
/*
print $do->agent . " ";
print $do->agent["role"] . " \n";
print $do->agent["homepage"] . " ";
*/
$dataObjectParameters = array();
$dataObjectParameters["identifier"] = $t_dc2->identifier;
$dataObjectParameters["dataType"] = $do->dataType;
$dataObjectParameters["mimeType"] = $do->mimeType;
$dataObjectParameters["description"] = $t_dc2->description;
if ($do->subject != "") {
$dataObjectParameters["subjects"] = array();
$subjectParameters = array();
$subjectParameters["label"] = $do->subject;
$dataObjectParameters["subjects"][] = new \SchemaSubject($subjectParameters);
}
$agents = array();
foreach ($do->agent as $agent) {
$agentParameters = array();
$agentParameters["role"] = $agent["role"];
$agentParameters["homepage"] = $agent["homepage"];
$agentParameters["logoURL"] = $agent["logoURL"];
$agentParameters["fullName"] = Functions::import_decode($agent);
$agents[] = new \SchemaAgent($agentParameters);
}
$dataObjectParameters["agents"] = $agents;
$dataObjectParameters["created"] = $do->created;
$dataObjectParameters["modified"] = $do->modified;
$dataObjectParameters["license"] = $do->license;
$dataObjectParameters["rightsHolder"] = Functions::import_decode($t_dcterms->rightsHolder);
$dataObjectParameters["source"] = $t_dc2->source;
$dataObjectParameters["mediaURL"] = $do->mediaURL;
$dataObjectParameters["thumbnailURL"] = $do->thumbnailURL;
$dataObjectParameters["location"] = Functions::import_decode($do->location);
///////////////////////////////////
$dataObjectParameters["audiences"] = array();
$audienceParameters = array();
$audienceParameters["label"] = "Expert users";
$dataObjectParameters["audiences"][] = new \SchemaAudience($audienceParameters);
$audienceParameters["label"] = "General public";
$dataObjectParameters["audiences"][] = new \SchemaAudience($audienceParameters);
///////////////////////////////////
return $dataObjectParameters;
}
