/**
* Extracts text from a file using Apache Tika
*
* @param FileInterface $file
* @return string Text extracted from the input file
*/
public function extractText(FileInterface $file)
{
$extractedContent = NULL;
$tika = TikaServiceFactory::getTika($this->configuration['extractor']);
$extractedContent = $tika->extractText($file);
return $extractedContent;
}
/**
* Extracts meta data from a file using Apache Tika
*
* @param Resource\File $file
* @param array $previousExtractedData Already extracted/existing data
* @return array
*/
public function extractMetaData(Resource\File $file, array $previousExtractedData = array())
{
$metaData = NULL;
$tikaService = TikaServiceFactory::getTika($this->configuration['extractor']);
$extractedMetaData = $tikaService->extractMetaData($file);
$metaData = $this->normalizeMetaData($extractedMetaData);
return $metaData;
}