public function getSuggestedLinksForRegistryObject(_registry_object $registry_object, $start, $rows)
{
// First, get published records with the same identifier as us
// Note: whilst we can use SOLR to get the linked-to records,
// we shouldn't use SOLR to get our own information, as
// this would mean that DRAFT requests fail (drafts NOT
// in SOLR index).
$suggestions = array();
$sxml = $registry_object->getSimpleXML();
if ($sxml->registryObject) {
$sxml = $sxml->registryObject;
}
// Identifier matches (if another object has the same identifier)
//var_dump($sxml);
$my_identifiers = array();
if ($sxml->{strtolower($registry_object->class)}->identifier) {
foreach ($sxml->{strtolower($registry_object->class)}->identifier as $identifier) {
if ((string) $identifier != '') {
$my_identifiers[] = (string) $identifier;
}
}
}
if (count($my_identifiers) == 0) {
return $suggestions;
}
$identifier_search_query = '';
if (sizeof($my_identifiers) > 0) {
$identifier_search_query = join('","', $my_identifiers);
$identifier_search_query = '+identifier_value:("' . $identifier_search_query . '")';
}
// But exclude already related objects
$my_relationships = array_map(function ($elt) {
return $elt;
}, $registry_object->getRelatedKeys());
$my_relationships[] = $registry_object->key;
$relationship_search_query = '';
if (sizeof($my_relationships) > 0) {
$relationship_search_query = implode('") -key:("', $my_relationships);
$relationship_search_query = '-key:("' . $relationship_search_query . '")';
}
$query = $relationship_search_query;
if ($identifier_search_query != '') {
$query .= ' AND ' . $identifier_search_query;
}
$suggestions = $this->getSuggestionsBySolrQuery($query, $start, $rows);
if (sizeof($suggestions) > 0) {
$suggestions['values'] = $my_identifiers;
}
return $suggestions;
}
public function getSuggestedLinksForRegistryObject(_registry_object $registry_object, $start, $rows)
{
// First, get published records with the same identifier as us
// Note: whilst we can use SOLR to get the linked-to records,
// we shouldn't use SOLR to get our own information, as
// this would mean that DRAFT requests fail (drafts NOT
// in SOLR index).
$suggestions = array();
$sxml = $registry_object->getSimpleXML();
if ($sxml->registryObject) {
$sxml = $sxml->registryObject;
}
// Subject matches
$my_subjects = array();
if ($sxml->{strtolower($registry_object->class)}->subject) {
foreach ($sxml->{strtolower($registry_object->class)}->subject as $subject) {
$my_subjects[] = (string) removeBadValue($subject);
}
}
if (count($my_subjects) == 0) {
return $suggestions;
}
if (sizeof($my_subjects) > 0) {
$subject_search_query = join('" OR subject_value_resolved:"', $my_subjects);
$subject_search_query = "(subject_value_resolved:\"" . $subject_search_query . "\")";
}
// But exclude already related objects
$my_relationships = array_map(function ($elt) {
return $elt;
}, $registry_object->getRelatedKeys());
$my_relationships[] = $registry_object->key;
$relationship_search_query = '';
if (sizeof($my_relationships) > 0) {
$relationship_search_query = join('","', $my_relationships);
$relationship_search_query = '-key:("' . $relationship_search_query . '")';
}
$query = $relationship_search_query;
if ($subject_search_query != '') {
$query .= ' AND ' . $subject_search_query;
}
$suggestions = $this->getSuggestionsBySolrQuery($query, $start, $rows);
if (sizeof($suggestions) > 0) {
$suggestions['values'] = $my_subjects;
}
return $suggestions;
}
public function getSuggestedLinksForRegistryObject(_registry_object $registry_object, $start, $rows)
{
// First, get published records with the same identifier as us
// Note: whilst we can use SOLR to get the linked-to records,
// we shouldn't use SOLR to get our own information, as
// this would mean that DRAFT requests fail (drafts NOT
// in SOLR index).
$sxml = $registry_object->getSimpleXML();
if ($sxml->registryObject) {
$sxml = $sxml->registryObject;
}
// Identifier matches (if another object has the same identifier)
//var_dump($sxml);
$my_identifiers = array('');
if ($sxml->{strtolower($registry_object->class)}->identifier) {
foreach ($sxml->{strtolower($registry_object->class)}->identifier as $identifier) {
$my_identifiers[] = '"' . (string) $identifier . '"';
}
}
$identifier_search_query = implode(" +identifier_value:", $my_identifiers);
$identifier_search_query = " -slug:" . $registry_object->slug . $identifier_search_query;
$suggestions = $this->getSuggestionsBySolrQuery($identifier_search_query);
return $suggestions;
}
/**
* XXX:
* @return array(_data_source) or NULL
*/
function create(_data_source $data_source, $registry_object_key, $class, $title, $status, $slug, $record_owner, $harvestID)
{
$ro = new _registry_object();
$ro->_initAttribute("data_source_id", $data_source->getAttribute('data_source_id'), TRUE);
$ro->_initAttribute("key", $registry_object_key, TRUE);
$ro->_initAttribute("class", $class, TRUE);
$ro->_initAttribute("title", $title, TRUE);
$ro->_initAttribute("status", $status, TRUE);
$ro->_initAttribute("slug", $slug, TRUE);
$ro->_initAttribute("record_owner", $record_owner, TRUE);
$ro->create();
// Some extras
$ro->setAttribute("created", time());
$ro->setAttribute("harvest_id", $harvestID);
$ro->save();
return $ro;
}
public function getSuggestedLinksForRegistryObject(_registry_object $registry_object, $start, $rows)
{
$CI =& get_instance();
$q = preg_replace("/[^A-Za-z0-9\\- ]/", ' ', $registry_object->titleWithoutCommonWords());
// preg_replace("/[^A-Za-z0-9 ]/", ' ',
$significantsArray = $registry_object->splitTitleBySignificance(strtolower($q));
$q = rawurlencode($q);
$q = str_replace("%5C%22", "\"", $q);
//silly encoding
$q = '?q=+title:(' . $significantsArray[0] . ")+description:(" . $significantsArray[0] . ")";
// Result variable
$links = array();
// Build the weighted URL based on datacite:resourceTypeGeneral
// - First Collections/Datasets
// - Then Film, Image, Sound, PhysicalObject, InteractiveResource
// - Then Model, Software, Service
// - Then Event, Text
// - Finally those without a resourceType
$url = self::DATACITE_SOLR_URL;
$urlTail = '&defType=disMax';
$urlTail = '&fq=language_facet%3A"English"';
// only English
$urlTail .= '&qf=resourceTypeGeneral:("Collection","Dataset")^9999';
$urlTail .= '%20+resourceTypeGeneral:("Film","Image","Sound","PhysicalObject","InteractiveResource")^5555';
$urlTail .= '%20+resourceTypeGeneral:("Model","Software","Service")^1777';
$urlTail .= '%20+resourceTypeGeneral:("Event","Text")^111';
$urlTail .= '%20+resourceTypeGeneral:""^1';
$urlTail .= '&fl=*,score';
$urlTail .= '&start=' . $start;
$urlTail .= '&rows=' . $rows;
$urlTail .= '&version=2.2&wt=json';
$content = json_decode(file_get_contents($url . $q . $urlTail), true);
// Check for a valid SOLR response
if (!isset($content['response']['numFound'])) {
return $links;
}
$found = $content['response']['numFound'];
/* If we got no results, then lets be more tolerant and add the non-significant words and do fuzzy match */
$fuzzy_match_url = "";
if ($found < 10) {
if (isset($significantsArray[1]) && $significantsArray[1] != '') {
$q .= '+title:(' . $significantsArray[1] . ")+description:(" . $significantsArray[1] . ")";
}
$fuzzy_match_url = $url . $q . $urlTail . '&defType=dismax&mm=3%3C75%25';
$content = json_decode(file_get_contents($fuzzy_match_url), true);
$found = $content['response']['numFound'];
}
/* Generate the links data */
if (isset($content['response']['docs'])) {
foreach ($content['response']['docs'] as $doc) {
if ($doc['title'][0]) {
$links[] = array("url" => self::DATACITE_URL_PREFIX . $doc[self::DATACITE_URL_FIELD], "title" => ellipsis($doc['title'][0], self::DATACITE_TITLE_LENGTH), "class" => "external", "expanded_html" => $CI->load->view("registry_object/datacite_preview", $doc, true));
}
}
}
if (!$rows) {
$rows = 10;
}
$pagination = array();
if ($start == 0) {
$currentPage = 1;
} else {
$currentPage = ceil($start / $rows) + 1;
}
$totalPage = ceil($content['response']['numFound'] / (int) $rows);
if ($currentPage != 1) {
$prev = $start - $rows;
$next = $start + $rows;
} else {
if ($currentPage == 1 && $totalPage == 1) {
$prev = false;
$next = false;
} else {
if ($currentPage == $totalPage) {
$prev = $start - $rows;
$next = false;
} else {
$prev = false;
$next = $start + $rows;
$debug = '1';
}
}
}
$pagination = array("currentPage" => $currentPage, "totalPage" => $totalPage);
if ($prev) {
$pagination['prev'] = $prev;
}
if ($next) {
$pagination['next'] = $next;
}
if ($start == 10) {
$pagination['prev'] = '0';
}
$response = array("count" => $found, "links" => $links, "pagination" => $pagination, "suggestor" => "datacite");
return $response;
}
