/**
* @param string $localSite The wikiId to add/remove from local_sites_with_dupe
* @param string $indexName The name of the index to perform updates to
* @param aray $otherActions A list of arrays each containing the id within elasticsearch ('id') and the article id within $localSite ('articleId')
* @return Status
*/
public function sendOtherIndexUpdates($localSite, $indexName, array $otherActions)
{
if (!$this->areIndexesAvailableForWrites(array($indexName), true)) {
return Status::newFatal('cirrussearch-indexes-frozen');
}
$client = $this->connection->getClient();
$status = Status::newGood();
foreach (array_chunk($otherActions, 30) as $updates) {
$bulk = new \Elastica\Bulk($client);
$articleIDs = array();
foreach ($updates as $update) {
$title = Title::makeTitle($update['ns'], $update['dbKey']);
$action = $this->decideRequiredSetAction($title);
$this->log->debug("Performing `{$action}` for {$update['dbKey']} in ns {$update['ns']} on {$localSite} against id {$update['id']} in {$indexName}");
$script = new \Elastica\Script('super_detect_noop', array('source' => array('local_sites_with_dupe' => array($action => $localSite)), 'handlers' => array('local_sites_with_dupe' => 'set')), 'native');
$script->setId($update['id']);
$script->setParam('_type', 'page');
$script->setParam('_index', $indexName);
$bulk->addScript($script, 'update');
$titles[] = $title;
}
// Execute the bulk update
$exception = null;
try {
$this->start("updating {numBulk} documents in other indexes", array('numBulk' => count($updates)));
$bulk->send();
} catch (\Elastica\Exception\Bulk\ResponseException $e) {
if (!$this->bulkResponseExceptionIsJustDocumentMissing($e)) {
$exception = $e;
}
} catch (\Elastica\Exception\ExceptionInterface $e) {
$exception = $e;
}
if ($exception === null) {
$this->success();
} else {
$this->failure($e);
$this->failedLog->warning("OtherIndex update for articles: " . implode(',', $titles), array('exception' => $e));
$status->error('cirrussearch-failed-update-otherindex');
}
}
return $status;
}
/**
* @group functional
*/
public function testRetry()
{
$index = $this->_createIndex();
$type = $index->getType('bulk_test');
$client = $index->getClient();
$doc1 = $type->createDocument(1, array('name' => 'Mister Fantastic'));
$doc1->setOpType(Action::OP_TYPE_UPDATE);
$doc1->setRetryOnConflict(5);
$bulk = new Bulk($client);
$bulk->addDocument($doc1);
$actions = $bulk->getActions();
$metadata = $actions[0]->getMetadata();
$this->assertEquals(5, $metadata['_retry_on_conflict']);
$script = new \Elastica\Script('');
$script->setRetryOnConflict(5);
$bulk = new Bulk($client);
$bulk->addScript($script);
$actions = $bulk->getActions();
$metadata = $actions[0]->getMetadata();
$this->assertEquals(5, $metadata['_retry_on_conflict']);
}
/**
* Converts a document into a call to super_detect_noop from the wikimedia-extra plugin.
*/
private function docToSuperDetectNoopScript($doc)
{
$params = $doc->getParams();
$params['source'] = $doc->getData();
$params['detectors'] = array('incoming_links' => 'within 20%');
$script = new \Elastica\Script('super_detect_noop', $params, 'native');
if ($doc->getDocAsUpsert()) {
$script->setUpsert($doc);
}
return $script;
}
/**
* We can only do this if dynamic scripting is enabled. In elasticsearch.yml:
* script.disable_dynamic: false
* @see vendor/ruffin/elastica/test/bin/run_elasticsearch.sh
*
* @param array $terms
* @return \Elastica\Aggregation\Sum
*/
protected function termsAggregation(array $terms)
{
$terms = str_replace('"', '\\"', $terms);
$script = '
keywords = ["' . implode('","', $terms) . '"]
total = 0
for (term in keywords) {
total += _index["revisions.text"][term].tf()
}
return total';
$script = new \Elastica\Script($script, null, 'groovy');
$aggregation = new \Elastica\Aggregation\Sum('ttf');
// $aggregation->setScript() doesn't seem to properly set 'lang': 'groovy'
// see https://github.com/ruflin/Elastica/pull/748
// $aggregation->setScript( $script );
$aggregation->setParams(array('lang' => 'groovy'));
$aggregation->setParam('script', $script->getScript());
return $aggregation;
}