/**
* Takes a file reference and extracts its meta data.
*
* @param \TYPO3\CMS\Core\Resource\File $file
* @return array
*/
public function extractMetaData(File $file)
{
$localTempFilePath = $file->getForLocalProcessing(FALSE);
$query = GeneralUtility::makeInstance('ApacheSolrForTypo3\\Tika\\Service\\SolrCellQuery', $localTempFilePath);
$query->setExtractOnly();
$response = $this->solr->extract($query);
$metaData = $this->solrResponseToArray($response[1]);
$this->cleanupTempFile($localTempFilePath, $file);
$this->log('Meta Data Extraction using Solr', array('file' => $file, 'solr connection' => (array) $this->solr, 'query' => (array) $query, 'response' => $response, 'meta data' => $metaData));
return $metaData;
}
/**
* Creates a single Solr Document for an item in a specific language.
*
* @param Tx_Solr_IndexQueue_Item An index queue item to index.
* @param integer The language to use.
* @return boolean TRUE if item was indexed successfully, FALSE on failure
*/
protected function indexItem(Tx_Solr_IndexQueue_Item $item, $language = 0)
{
$itemIndexed = FALSE;
$documents = array();
$itemDocument = $this->itemToDocument($item, $language);
if (is_null($itemDocument)) {
/*
* If there is no itemDocument, this means there was no translation
* for this record. This should not stop the current item to count as
* being valid because not-indexing not-translated items is perfectly
* fine.
*/
return TRUE;
}
$documents[] = $itemDocument;
$documents = array_merge($documents, $this->getAdditionalDocuments($item, $language, $itemDocument));
$documents = $this->processDocuments($item, $documents);
$documents = $this->preAddModifyDocuments($item, $language, $documents);
$response = $this->solr->addDocuments($documents);
if ($response->getHttpStatus() == 200) {
$itemIndexed = TRUE;
}
$this->log($item, $documents, $response);
return $itemIndexed;
}
/**
* Adds the collected documents to the Solr index.
*
* @param array $documents An array of Apache_Solr_Document objects.
* @return boolean TRUE if documents were added successfully, FALSE otherwise
*/
protected function addDocumentsToSolrIndex(array $documents)
{
$documentsAdded = FALSE;
if (!count($documents)) {
return $documentsAdded;
}
try {
$this->log('Adding ' . count($documents) . ' documents.', 0, $documents);
// chunk adds by 20
$documentChunks = array_chunk($documents, 20);
foreach ($documentChunks as $documentChunk) {
$response = $this->solrConnection->addDocuments($documentChunk);
if ($response->getHttpStatus() != 200) {
$transportException = new Apache_Solr_HttpTransportException($response);
throw new RuntimeException('Solr Request failed.', 1331834983, $transportException);
}
}
$documentsAdded = TRUE;
} catch (Exception $e) {
$this->log($e->getMessage() . ' Error code: ' . $e->getCode(), 2);
if ($GLOBALS['TSFE']->tmpl->setup['plugin.']['tx_solr.']['logging.']['exceptions']) {
GeneralUtility::devLog('Exception while adding documents', 'solr', 3, array($e->__toString()));
}
}
return $documentsAdded;
}
/**
* Sends a ping to the solr server to see whether it is available.
*
* @return boolean Returns TRUE on successful ping.
* @throws Exception Throws an exception in case ping was not successful.
*/
public function ping()
{
$solrAvailable = FALSE;
try {
if (!$this->solr->ping()) {
throw new Exception('Solr Server not responding.', 1237475791);
}
$solrAvailable = TRUE;
} catch (Exception $e) {
if ($GLOBALS['TSFE']->tmpl->setup['plugin.']['tx_solr.']['logging.']['exceptions']) {
GeneralUtility::devLog('exception while trying to ping the solr server', 'solr', 3, array($e->__toString()));
}
}
return $solrAvailable;
}
/**
* Renders the index field table.
*
* Acknowledgement: Some of the code is taken from Drupal's apachesolr module
*
* @param $data
* @return unknown_type
*/
protected function renderData(Apache_Solr_Response $data)
{
$content = '';
$registry = t3lib_div::makeInstance('t3lib_Registry');
$limit = $registry->get('tx_solr', 'luke.limit', 20000);
$numberOfDocuments = $data->index->numDocs;
$content .= '<p>Number of documents in index: ' . $numberOfDocuments . '</p>';
if (isset($data->index->numDocs) && $data->index->numDocs > $limit) {
$notFound = '<em>Omitted</em>';
$content .= '<p>You have more than ' . $limit . ' documents, so term frequencies are being omitted for performance reasons.</p>';
} elseif (isset($data->index->numDocs)) {
$notFound = 'Nothing indexed';
// below limit, so we get more data
// Note: we use 2 since 1 fails on Ubuntu Hardy.
$data = $this->solr->getLukeMetaData(2);
$content .= '<p>Number of terms in index: ' . $data->index->numTerms . '</p>';
}
$fields = (array) $data->fields;
$content .= '<p>Number of fields in index: ' . count($fields) . '</p>';
// getting fields, sorting
$rows = array();
foreach ($fields as $name => $field) {
$rows[$name] = array($name, $field->type, isset($field->distinct) ? $field->distinct : $notFound);
}
ksort($rows);
// Initialise table layout
$tableLayout = array('table' => array('<table border="0" cellspacing="0" cellpadding="2" class="tx_solr_index_list">', '</table>'), '0' => array('tr' => array('<tr class="bgColor2" valign="top">', '</tr>'), 'defCol' => array('<td>', '</td>')), 'defRowOdd' => array('tr' => array('<tr class="bgColor3-20">', '</tr>'), 'defCol' => array('<td>', '</td>')), 'defRowEven' => array('tr' => array('<tr>', '</tr>'), 'defCol' => array('<td>', '</td>')));
$table = array();
// header row
$table[] = array('Field Name', 'Index Type', 'Distinct Terms');
foreach ($rows as $row) {
$table[] = $row;
}
// Render table
$content .= $this->reportsModule->doc->table($table, $tableLayout);
return $content;
}
/**
* Gets the version of the installed plugin.
*
* @param Tx_Solr_SolrService $solrConnection Solr connection to check for the plugin.
* @return string The installed plugin's version number.
*/
public function getInstalledPluginVersion(Tx_Solr_SolrService $solrConnection)
{
$version = '0.0.0';
$pluginsInformation = $solrConnection->getPluginsInformation();
$rawVersion = $pluginsInformation->plugins->OTHER->{self::PLUGIN_CLASS_NAME}->version;
$explodedRawVersion = explode('-', $rawVersion);
$version = $explodedRawVersion[0];
return $version;
}
/**
* Reloads a single Solr core.
*
* @param \Tx_Solr_SolrService $solrServer A Solr server connection
* @param string $coreName Name of the core to reload
* @return bool TRUE if reloading the core was successful, FALSE otherwise
*/
protected function reloadCore(\Tx_Solr_SolrService $solrServer, $coreName)
{
$coreReloaded = FALSE;
$path = $solrServer->getPath();
$pathElements = explode('/', trim($path, '/'));
$coreAdminReloadUrl = $solrServer->getScheme() . '://' . $solrServer->getHost() . ':' . $solrServer->getPort() . '/' . $pathElements[0] . '/' . 'admin/cores?action=reload&core=' . $coreName;
$httpTransport = $solrServer->getHttpTransport();
$httpResponse = $httpTransport->performGetRequest($coreAdminReloadUrl);
$solrResponse = new \Apache_Solr_Response($httpResponse, $solrServer->getCreateDocuments(), $solrServer->getCollapseSingleValueArrays());
if ($solrResponse->getHttpStatus() == 200) {
$coreReloaded = TRUE;
}
return $coreReloaded;
}
