public function searchDataset($criteria, $sorts = array(), $offset = 0, $limit = 0, $operator = 'OR')
{
if ($limit <= 0) {
$limit = Yii::app()->params['es_search']['limits']['default'];
}
$keyword = $criteria['keyword'];
$eQueryString = new \Elastica\Query\QueryString();
$eQueryString->setDefaultOperator($operator);
$eQueryString->setQuery($keyword);
$eQuery = new \Elastica\Query();
$eQuery->setQuery($eQueryString);
$eQuery->setFrom($offset);
$eQuery->setLimit($limit);
$filterAnd = new \Elastica\Filter\BoolAnd();
if (isset($criteria['dataset_type'])) {
$filterAnd->addFilter($this->newTerms('typenames', $criteria['dataset_type']));
}
if (isset($criteria['project'])) {
$filterAnd->addFilter($this->newTerms('projects', $criteria['project']));
}
if (isset($criteria['external_link_type'])) {
$filterAnd->addFilter($this->newTerms('link_types', $criteria['external_link_type']));
}
if (isset($criteria['pubdate_from']) && isset($criteria['pubdate_to'])) {
$pubfilter = new \Elastica\Filter\Range();
$pubfilter->addField('publication_date', array('from' => $criteria['pubdate_from'], 'to' => $criteria['pubdate_to']));
$filterAnd->addFilter($pubfilter);
} elseif (isset($criteria['pubdate_from'])) {
$pubfilter = new \Elastica\Filter\Range();
$pubfilter->addField('publication_date', array('from' => $criteria['pubdate_from']));
$filterAnd->addFilter($pubfilter);
} elseif (isset($criteria['pubdate_to'])) {
$pubfilter = new \Elastica\Filter\Range();
$pubfilter->addField('publication_date', array('to' => $criteria['pubdate_to']));
$filterAnd->addFilter($pubfilter);
}
$arrayAnd = $filterAnd;
$arrayAnd = $arrayAnd->toArray();
if (count($arrayAnd['and']) > 0) {
$eQuery->setPostFilter($filterAnd);
}
$sortList = array_merge(array('_score' => array('order' => 'desc')), $sorts);
$eQuery->setSort($sortList);
$index = Yii::app()->elastic->client->getIndex('gigadb');
$type = $index->getType('dataset');
$eResultSet = $type->search($eQuery);
$results = $eResultSet->getResults();
$total = $eResultSet->getTotalHits();
$data = array();
foreach ($results as $result) {
if ($result) {
$data[] = $result->getData();
}
}
$re = array('data' => $data, 'total' => $total);
//Yii::log(print_r($eResultSet, true), 'debug');
return $re;
}
/**
* To respect the design, 3 searches will be executed:
* 1st: retrieves the main facet (Text / Media / Portfolio / Users / Group) and the count for each of them
* 2nd: - retrieves the results of the first non empty facet term for display in the tab
* - retrieves the secondary facet to enable / disable the filter items
* 3nd: - retrieves the results with all filters applied
* @param unknown $query_string
* @param unknown $limit
* @param unknown $offset
* @param unknown $options
* @param unknown $mainfacetterm
* @param unknown $USER
* @return multitype:number boolean unknown Ambigous <boolean, NULL> Ambigous <boolean, unknown> multitype:multitype:string number Ambigous <string, unknown> |multitype:multitype:
*/
public static function search($query_string, $limit, $offset, $options, $mainfacetterm, $USER)
{
$result = array('count' => 0, 'limit' => $limit, 'offset' => $offset, 'data' => false, 'selected' => isset($mainfacetterm) && strlen($mainfacetterm) > 0 ? $mainfacetterm : false, 'totalresults' => 0, 'facets' => array(array('term' => "Text", 'count' => 0, 'display' => "Text"), array('term' => "Media", 'count' => 0, 'display' => "Media"), array('term' => "Portfolio", 'count' => 0, 'display' => "Portfolio"), array('term' => "User", 'count' => 0, 'display' => "Users"), array('term' => "Group", 'count' => 0, 'display' => "Group")), 'content-filter' => array(array('term' => "all", 'count' => 0, 'display' => "All"), array('term' => "Audio", 'count' => 0, 'display' => "Audio"), array('term' => "Comment", 'count' => 0, 'display' => "Comment"), array('term' => "Document", 'count' => 0, 'display' => "Document"), array('term' => "Folder", 'count' => 0, 'display' => "Folder"), array('term' => "Forum", 'count' => 0, 'display' => "Forum"), array('term' => "Forumpost", 'count' => 0, 'display' => "Forum post"), array('term' => "Image", 'count' => 0, 'display' => "Image"), array('term' => "Journal", 'count' => 0, 'display' => "Journal"), array('term' => "Journalentry", 'count' => 0, 'display' => "Journal entry"), array('term' => "Note", 'count' => 0, 'display' => "Note"), array('term' => "Plan", 'count' => 0, 'display' => "Plan"), array('term' => "Profile", 'count' => 0, 'display' => "Profile"), array('term' => "Resume", 'count' => 0, 'display' => "Résumé"), array('term' => "Video", 'count' => 0, 'display' => "Video"), array('term' => "Wallpost", 'count' => 0, 'display' => "Wall post"), array('term' => "Collection", 'count' => 0, 'display' => "Collection"), array('term' => "Page", 'count' => 0, 'display' => "Page")), 'content-filter-selected' => isset($options['secfacetterm']) && strlen($options['secfacetterm']) > 0 ? $options['secfacetterm'] : 'all', 'owner-filter' => array(array('term' => "all", 'count' => 0, 'display' => "All"), array('term' => "me", 'count' => 0, 'display' => "Me"), array('term' => "others", 'count' => 0, 'display' => "Others")), 'owner-filter-selected' => isset($options['owner']) && strlen($options['owner']) > 0 ? $options['owner'] : 'all', 'tagsonly' => isset($options['tagsonly']) && $options['tagsonly'] == true ? true : Null, 'sort' => isset($options['sort']) && strlen($options['sort']) > 0 ? $options['sort'] : 'score', 'license' => isset($options['license']) && strlen($options['license']) > 0 ? $options['license'] : 'all');
if (strlen($query_string) <= 0) {
return $result;
}
// 1- Get main facet
// ------------------------------------------------------------------------------------------
$records = array();
$elasticaClient = PluginSearchElasticsearch::make_client();
$elasticaIndex = $elasticaClient->getIndex(get_config_plugin('search', 'elasticsearch', 'indexname'));
$elasticaQueryString = new \Elastica\Query\QueryString();
$elasticaAnalyzer = get_config_plugin('search', 'elasticsearch', 'analyzer');
$elasticaQueryString->setAnalyzer($elasticaAnalyzer);
$elasticaQueryString->setDefaultOperator('AND');
$elasticaQueryString->setQuery($query_string);
// if tags only => set fields to tags
if ($result['tagsonly'] === true) {
$elasticaQueryString->setFields(array('tags'));
}
// Create the $elasticaQuery object
$elasticaQuery = new \Elastica\Query();
$elasticaQuery->setFrom($offset);
$elasticaQuery->setLimit($limit);
$elasticaQuery->setQuery($elasticaQueryString);
$elasticaFilterAnd = new \Elastica\Filter\BoolAnd();
// Apply ACL filters
$elasticaFilterACL = new ElasticsearchFilterAcl($USER);
$elasticaFilterAnd->addFilter($elasticaFilterACL);
$elasticaQuery->setFilter($elasticaFilterAnd);
// Define a new facet: mainFacetTerm - WARNING: don't forget to apply the same filter to the facet
$elasticaFacet = new \Elastica\Facet\Terms('mainFacetTerm');
$elasticaFacet->setField('mainfacetterm');
$elasticaFacet->setOrder('count');
$elasticaFacet->setFilter($elasticaFilterAnd);
$elasticaQuery->addFacet($elasticaFacet);
$elasticaResultSet = $elasticaIndex->search($elasticaQuery);
$result['totalresults'] = $elasticaResultSet->getTotalHits();
$elasticaFacets = $elasticaResultSet->getFacets();
$facets = self::process_facets($elasticaFacets['mainFacetTerm']['terms']);
if (count($facets) == 0) {
return $result;
}
array_walk($result['facets'], 'self::process_tabs', $facets);
if ($result['selected'] === false || $facets[$result['selected']] == 0) {
$result['selected'] = self::get_selected_facet($result['facets']);
}
// 2- Retrieve results of selected facet
// ------------------------------------------------------------------------------------------
$elasticaFilterType = new \Elastica\Filter\Term(array('mainfacetterm' => $result['selected']));
$elasticaFilterAnd->addFilter($elasticaFilterType);
$elasticaQuery->setFilter($elasticaFilterAnd);
// Define a new facet: secFacetTerm - WARNING: don't forget to apply the same filter to the facet
$elasticaFacet = new \Elastica\Facet\Terms('secFacetTerm');
$elasticaFacet->setField('secfacetterm');
$elasticaFacet->setOrder('count');
$elasticaFacet->setFilter($elasticaFilterAnd);
$elasticaQuery->addFacet($elasticaFacet);
// Sorting
// Sorting is defined on a per field level, so we must make sure the field exists in the mapping
$sort = explode('_', $result['sort']);
if ($sort[0] == 'score') {
$sort[0] = '_score';
}
// set the second column to sort by the score (to break any 'ties').
$elasticaQuery->setSort(array(array($sort[0] => array('order' => isset($sort[1]) ? $sort[1] : 'desc')), array('_score' => array('order' => 'desc'))));
$elasticaResultSet = $elasticaIndex->search($elasticaQuery);
$result['count'] = $elasticaResultSet->getTotalHits();
$elasticaFacets = $elasticaResultSet->getFacets();
$facets = $elasticaFacets['secFacetTerm']['terms'];
$facets = self::process_facets($elasticaFacets['secFacetTerm']['terms']);
array_walk($result['content-filter'], 'self::process_tabs', $facets);
// set the count of "all" to the total hits
$result['content-filter'][0]['count'] = $result['count'];
// 3- Apply filters and retrieve final results
// ------------------------------------------------------------------------------------------
// Apply Content filter if different from "all"
if ($result['content-filter-selected'] != 'all') {
$elasticaFilterContent = new \Elastica\Filter\Term(array('secfacetterm' => $result['content-filter-selected']));
$elasticaFilterAnd->addFilter($elasticaFilterContent);
}
// Apply Owner filter if different from "all"
if ($result['owner-filter-selected'] != 'all') {
$uid = $USER->get('id');
$elasticaFilterOwner = new \Elastica\Filter\Term(array('owner' => $uid));
if ($result['owner-filter-selected'] == 'others') {
$elasticaFilterOwner = new \Elastica\Filter\BoolNot($elasticaFilterOwner);
}
$elasticaFilterAnd->addFilter($elasticaFilterOwner);
}
// Apply license filter if different from "all"
if ($result['license'] != 'all') {
$elasticaFilterLicense = new \Elastica\Filter\Term(array('license' => $result['license']));
$elasticaFilterAnd->addFilter($elasticaFilterLicense);
}
$elasticaQuery->setFilter($elasticaFilterAnd);
$elasticaResultSet = $elasticaIndex->search($elasticaQuery);
$elasticaResults = $elasticaResultSet->getResults();
$result['count'] = $elasticaResultSet->getTotalHits();
foreach ($elasticaResults as $elasticaResult) {
$tmp = array();
$tmp['type'] = $elasticaResult->getType();
$ES_class = 'ElasticsearchType_' . $tmp['type'];
$tmp = $tmp + $elasticaResult->getData();
// Get all the data from the DB table
$dbrec = $ES_class::getRecordDataById($tmp['type'], $tmp['id']);
if ($dbrec) {
$tmp['db'] = $dbrec;
$tmp['db']->deleted = false;
} else {
// If the record has been deleted, so just pass the cached data
// from the search result. Let the template decide how to handle
// it.
$tmp['db'] = (object) $tmp;
$tmp['db']->deleted = true;
}
$records[] = $tmp;
}
$result['data'] = $records;
return $result;
}
/**
* Main method to make a search
*/
public function search()
{
$this->getSearchFields();
$word = JRequest::getString('searchword', null, 'get');
$offset = JRequest::getInt('start', 0, 'get');
$limit = JRequest::getInt('limit', 10, 'get');
if ($limit == 0) {
// no limit
$limit = 10000;
}
$this->getSearchAreas();
$elasticaQueryString = new \Elastica\Query\QueryString();
// Log search word only on the first page
if ($offset == 0) {
SearchHelper::logSearch($word);
}
// Convert accents
$word = htmlentities($word, ENT_NOQUOTES, 'utf-8');
$word = preg_replace('#\\&([A-za-z])(?:uml|circ|tilde|acute|grave|cedil|ring)\\;#', '\\1', $word);
$word = preg_replace('#\\&([A-za-z]{2})(?:lig)\\;#', '\\1', $word);
$word = preg_replace('#\\&[^;]+\\;#', '', $word);
// Check if there are quotes ( for exact search )
$exactSearch = false;
if (strlen($word) > 1 && $word[0] == '"' && $word[strlen($word) - 1] == '"') {
$exactSearch = true;
$word = substr($word, 1, strlen($word) - 2);
// Remove external "
}
$word = Elastica\Util::replaceBooleanWordsAndEscapeTerm($word);
// Escape ElasticSearch specials char
if ($exactSearch) {
$word = '"' . $word . '"';
}
if ($word == "") {
$word = "*";
}
$elasticaQueryString->setQuery($word);
//Create the actual search object with some data.
$elasticaQuery = new Elastica\Query();
$elasticaQuery->setQuery($elasticaQueryString);
if (ElasticSearchConfig::getHighLigthEnable()) {
$fields = $this->getHighlightFields();
$hlfields = array();
foreach ($fields as $field) {
foreach ($field as $highlight) {
$hlfields[] = array($highlight => array('fragment_size' => 1000, 'number_of_fragments' => 1));
}
}
$highlightFields = array('pre_tags' => array(ElasticSearchConfig::getHighLigthPre()), 'post_tags' => array(ElasticSearchConfig::getHighLigthPost()), "order" => "score", 'fields' => $hlfields);
$elasticaQuery->setHighlight($highlightFields);
}
// Set offset and limit for pagination
$elasticaQuery->setFrom($offset);
$elasticaQuery->setLimit($limit);
//Create a filter for _type
$elasticaFilterype = $this->createFilterType($this->areas);
// Add filter to the search object.
$elasticaQuery->setFilter($elasticaFilterype);
//Search on the index.
$elasticaResultSet = $this->elasticaIndex->search($elasticaQuery);
$this->results = $elasticaResultSet->getResults();
$this->totalHits = $elasticaResultSet->getTotalHits();
}
