[0bit]

I would recommend using Apache Tika to extract the text from the PDFs and using Solr (or Elasticsearch) to index and search them.