/**
* Returns all index documents that are associated to the given collection id.
*
* @param int $collectionId
* @return array An array of Opus_SolrSearch_Result objects.
* @throws Opus_SolrSearch_Exception
*/
private function getListItems($collectionId)
{
$query = new Opus_SolrSearch_Query(Opus_SolrSearch_Query::SIMPLE);
$query->setCatchAll('*:*');
$query->addFilterQuery('collection_ids', $collectionId);
$query->setRows(Opus_SolrSearch_Query::MAX_ROWS);
$solrsearch = new Opus_SolrSearch_Searcher();
return $solrsearch->search($query)->getResults();
}
* @version $Id: find_missing_published_docs_in_searchindex.php 12011 2013-07-05 19:39:58Z sszott $
*/
/**
*
* Dieses Skript gibt alle IDs der Dokumente zurück, die im Server State
* published sind, aber aufgrund eines Fehlers nicht im Index repräsentiert sind.
*
* Siehe dazu auch die Story OPUSVIER-2368.
*
*/
$numOfErrors = 0;
$finder = new Opus_DocumentFinder();
$finder->setServerState('published');
foreach ($finder->ids() as $docId) {
// check if document with id $docId is already persisted in search index
$query = new Opus_SolrSearch_Query(Opus_SolrSearch_Query::DOC_ID);
$query->setField('id', $docId);
$query->setReturnIdsOnly(true);
$query->setRows(Opus_SolrSearch_Query::MAX_ROWS);
$searcher = new Opus_SolrSearch_Searcher();
if ($searcher->search($query)->getNumberOfHits() != 1) {
echo "document # {$docId} is not stored in search index\n";
$numOfErrors++;
}
}
if ($numOfErrors > 0) {
echo "{$numOfErrors} missing documents were found\n";
} else {
echo "no errors were found\n";
}
exit;
private function createAllSearchQuery($input)
{
$this->logger->debug("Constructing query for all search.");
$query = new Opus_SolrSearch_Query(Opus_SolrSearch_Query::ALL_DOCS);
$query->setStart($input['start']);
$query->setRows($input['rows']);
$query->setSortField($input['sortField']);
$query->setSortOrder($input['sortOrder']);
$this->addFiltersToQuery($query, $input);
if ($this->export) {
$query->setReturnIdsOnly(true);
}
$this->logger->debug("Query {$query} complete");
return $query;
}
private function getDocsInSearchIndex($checkConsistency = true)
{
$searcher = new Opus_SolrSearch_Searcher();
$query = new Opus_SolrSearch_Query();
$query->setCatchAll("*:*");
$query->setRows(Opus_SolrSearch_Query::MAX_ROWS);
$resultList = $searcher->search($query, $checkConsistency);
return $resultList;
}
/**
* Find documents in Solr index, that are not in database or that are in
* datbase but not in serverState published Remove such documents from Solr
* index.
*
*/
private function checkSearchIndex()
{
$query = new Opus_SolrSearch_Query();
$query->setCatchAll("*:*");
$query->setRows(Opus_SolrSearch_Query::MAX_ROWS);
$resultList = $this->searcher->search($query, $this->validateDocIds);
$results = $resultList->getResults();
foreach ($results as $result) {
$id = $result->getId();
try {
$doc = new Opus_Document($id);
} catch (Opus_Model_NotFoundException $e) {
$this->logger->info("inconsistency found for document {$id}: document is in Solr index, but is not in database.");
$this->numOfInconsistencies++;
if ($this->removeDocumentFromSearchIndex($id)) {
$this->numOfDeletions++;
}
continue;
}
if ($doc->getServerState() != 'published') {
$this->logger->info("inconsistency found for document {$id}: document is in Solr index, but is not in ServerState published.");
$this->numOfInconsistencies++;
if ($this->removeDocumentFromSearchIndex($id)) {
$this->numOfDeletions++;
}
}
}
}
