/**
* @return boolean
*/
public function checkFilePermission()
{
return $this->_file->isReadable();
}
/**
* returns the extracted fulltext of the given file or an exception in
* case of errors
*
* @param Opus_File $file
* @throws Opus_SolrSearch_Index_Exception
* @return extracted fulltext
*/
private function getFileContent(Opus_File $file)
{
$this->log->debug('extracting fulltext from ' . $file->getPath());
if (!$file->exists()) {
$this->log->err($file->getPath() . ' does not exist.');
throw new Opus_SolrSearch_Index_Exception($file->getPath() . ' does not exist.');
}
if (!$file->isReadable()) {
$this->log->err($file->getPath() . ' is not readable.');
throw new Opus_SolrSearch_Index_Exception($file->getPath() . ' is not readable.');
}
if (!$this->hasSupportedMimeType($file)) {
$this->log->err($file->getPath() . ' has MIME type ' . $file->getMimeType() . ' which is not supported');
throw new Opus_SolrSearch_Index_Exception($file->getPath() . ' has MIME type ' . $file->getMimeType() . ' which is not supported');
}
// Check for cached ...
$fulltext = $this->getCachedFileContent($file);
if ($fulltext !== false and is_string($fulltext)) {
$this->log->info('Found cached fulltext for file ' . $file->getPath());
return $fulltext;
}
$params = array('extractOnly' => 'true', 'extractFormat' => 'text');
try {
$response = $this->getSolrServer('extract')->extract($file->getPath(), $params);
// TODO add mime type information
$jsonResponse = Zend_Json_Decoder::decode($response->getRawResponse());
if (array_key_exists('', $jsonResponse)) {
$fulltext = trim($jsonResponse['']);
$this->setCachedFileContent($file, $fulltext);
return $fulltext;
// TODO evaluate additional data in json response
}
} catch (Exception $e) {
$this->log->err('error while extracting fulltext from file ' . $file->getPath());
throw new Opus_SolrSearch_Index_Exception('error while extracting fulltext from file ' . $file->getPath(), null, $e);
}
return '';
}
