public function mutipleSearchWithElasticSearch($searchedTerm, $selectedTerm)
{
if ($selectedTerm == "") {
$selectedTerm = $searchedTerm;
}
function auto_fuzzy($term)
{
if (mb_strlen($term) <= 2) {
$fuzzy_int = 0;
} elseif (mb_strlen($term) < 10) {
$fuzzy_int = 1;
} elseif (mb_strlen($term) < 17) {
$fuzzy_int = 2;
} elseif (mb_strlen($term) <= 23) {
$fuzzy_int = 3;
} else {
$fuzzy_int = 4;
}
return $fuzzy_int;
}
$terms = array();
$finder = $this->container->get('fos_elastica.finder.shlml.word');
// On explose le string d'entrée en tableau, en séparant les mots espacés par un espace
$terms = explode(" ", $searchedTerm);
// Je ne retiens que les trois premiers mots
$terms = array_slice($terms, 0, 3);
// Je trouve les mots "similaires" grâce à Elasticsearch
$found = array();
for ($i = 0; $i < sizeof($terms); $i++) {
$query = new \Elastica\Query\Fuzzy();
$query->setField("content", $terms[$i]);
$fuzzy_int = auto_fuzzy($terms[$i]);
$query->setFieldOption("fuzziness", $fuzzy_int);
array_push($found, $finder->find($query, 10000000));
}
// Choisir le type de recherche en fonction du nombre de combinaisons pour éviter l'explosion combinatoire
$nb_combi = 1;
foreach ($found as $tab) {
$nb_combi = $nb_combi * sizeof($tab);
}
// J'effectue toutes les combinaisons possibles
$combinations = array();
if ($nb_combi < 1000) {
for ($i = 0; $i < sizeof($found[0]); $i++) {
array_push($combinations, $found[0][$i]->getContent());
}
for ($i = 1; $i < sizeof($found); $i++) {
$temp = array();
for ($j = 0; $j < sizeof($found[$i]); $j++) {
for ($k = 0; $k < sizeof($combinations); $k++) {
array_push($temp, $combinations[$k] . " " . $found[$i][$j]->getContent());
}
}
$combinations = $temp;
}
} else {
// Je ne fais varier que un seul mot
for ($i = 0; $i < sizeof($found); $i++) {
for ($j = 0; $j < sizeof($found[$i]); $j++) {
$temp = "";
for ($k = 0; $k < sizeof($found); $k++) {
if ($i != $k) {
$temp = $temp . " " . $terms[$k];
} else {
$temp = $temp . " " . $found[$k][$j]->getContent();
}
}
array_push($combinations, substr($temp, 1));
}
}
}
// On cherche les combinaisons qui sont effectivement dans les documents
$finder = $this->container->get('fos_elastica.finder.shlml.document');
$results = array();
for ($i = 0; $i < sizeof($combinations); $i++) {
$query = new \Elastica\Query\Match();
$query->setFieldQuery('content', $combinations[$i]);
$query->setFieldType('content', 'phrase');
if ($combinations[$i] == $selectedTerm) {
$docs = $finder->find($query, 100);
} else {
$docs = $finder->find($query, 1);
}
if ($docs != null) {
$results[$combinations[$i]] = array();
for ($j = 0; $j < sizeof($docs); $j++) {
if ($this->get('security.context')->isGranted('ROLE_USER') || $docs[$j]->getPublic()) {
array_push($results[$combinations[$i]], $docs[$j]->getPath());
}
}
}
}
$wordList = array_keys($results);
$selectedDocList = null;
//var_dump($results);
if (empty($results)) {
$selectedDocList = array("mot_introuvable.pdf");
$wordList = array($searchedTerm);
if ($searchedTerm != $selectedTerm) {
array_push($wordList, $selectedTerm);
}
} else {
if (array_key_exists($searchedTerm, $results)) {
unset($wordList[array_search($searchedTerm, $wordList)]);
array_unshift($wordList, $searchedTerm);
$selectedDocList = $results[$searchedTerm];
} else {
array_unshift($wordList, $searchedTerm);
}
if (array_key_exists($selectedTerm, $results)) {
$selectedDocList = $results[$selectedTerm];
}
if ($selectedDocList == null) {
$selectedDocList = array("mot_introuvable.pdf");
}
}
$res = array($selectedDocList, $wordList);
return $res;
}
/**
* @param {Integer} libraryID
* @param {String} searchText
* @return {Array<String>|Boolean} An array of item keys, or FALSE if no results
*/
public static function searchInLibrary($libraryID, $searchText)
{
// TEMP: For now, strip double-quotes and make everything a phrase search
$searchText = str_replace('"', '', $searchText);
$type = self::getReadType();
$libraryFilter = new \Elastica\Filter\Term();
$libraryFilter->setTerm("libraryID", $libraryID);
$matchQuery = new \Elastica\Query\Match();
$matchQuery->setFieldQuery('content', $searchText);
$matchQuery->setFieldType('content', 'phrase');
$matchQuery = new \Elastica\Query\Filtered($matchQuery, $libraryFilter);
$start = microtime(true);
$resultSet = $type->search($matchQuery, ['routing' => $libraryID]);
StatsD::timing("elasticsearch.client.item_fulltext.search", (microtime(true) - $start) * 1000);
if ($resultSet->getResponse()->hasError()) {
throw new Exception($resultSet->getResponse()->getError());
}
$results = $resultSet->getResults();
$keys = array();
foreach ($results as $result) {
$keys[] = explode("/", $result->getId())[1];
}
return $keys;
}
