/**
* These values end up serialized into Avro which has strict typing
* requirements. float !== int !== string.
*
* @param float $took Number of milliseconds the request took
* @return array
*/
private function buildLogContext($took)
{
$client = $this->connection->getClient();
$query = $client->getLastRequest();
$result = $client->getLastResponse();
$params = $this->logContext;
$this->logContext = array();
$params += array('tookMs' => intval($took), 'source' => self::getExecutionContext(), 'executor' => self::getExecutionId(), 'identity' => self::generateIdentToken());
if ($result) {
$queryData = $query->getData();
$resultData = $result->getData();
$index = explode('/', $query->getPath());
$params['index'] = $index[0];
if (isset($resultData['took'])) {
$elasticTook = $resultData['took'];
$params['elasticTookMs'] = intval($elasticTook);
}
if (isset($resultData['hits']['total'])) {
$params['hitsTotal'] = intval($resultData['hits']['total']);
}
if (isset($resultData['hits']['hits'])) {
$num = count($resultData['hits']['hits']);
$offset = isset($queryData['from']) ? $queryData['from'] : 0;
$params['hitsReturned'] = $num;
$params['hitsOffset'] = intval($offset);
}
if ($this->_isset($queryData, array('query', 'filtered', 'filter', 'terms', 'namespace'))) {
$namespaces = $queryData['query']['filtered']['filter']['terms']['namespace'];
$params['namespaces'] = array_map('intval', $namespaces);
}
if (isset($resultData['suggest']['suggest'][0]['options'][0]['text'])) {
$params['suggestion'] = $resultData['suggest']['suggest'][0]['options'][0]['text'];
}
}
if (self::$logContexts === null) {
DeferredUpdates::addCallableUpdate(function () {
ElasticsearchIntermediary::reportLogContexts();
});
self::$logContexts = array();
}
self::$logContexts[] = $params;
return $params;
}
/**
* Try to detect language using langdetect plugin
* See: https://github.com/jprante/elasticsearch-langdetect
* @param string $text
* @return string|NULL Language name or null
*/
public static function detectLanguage($text)
{
$client = Connection::getClient();
try {
$response = $client->request("_langdetect", Request::POST, $text);
} catch (ResponseException $e) {
// This happens when language detection is not configured
LoggerFactory::getInstance('CirrusSearch')->warning("Could not connect to language detector: {exception}", array("exception" => $e->getMessage()));
return null;
}
if ($response->isOk()) {
$value = $response->getData();
if ($value && !empty($value['languages'])) {
$langs = $value['languages'];
if (count($langs) == 1) {
// TODO: add minimal threshold
return $langs[0]['language'];
}
// FIXME: here I'm just winging it, should be something
// that makes sense for multiple languages
if (count($langs) == 2) {
if ($langs[0]['probability'] > 2 * $langs[1]['probability']) {
return $langs[0]['language'];
}
}
}
}
return null;
}
