Hi Marcel, thank you for your reply. I was able to extract text correctly for those encrypted files: I created a class that extends from AbstractParser, I specified this parser in the tika config for the index. Inside that class, I use an instance of AutoDetectParser that will be charged to return the supported types and to extract text from the decrypted file (decryption performed in the parse method).
The lucene index files will contain data that allows someone with access to > the files on disk to reconstruct the text in the document. > You are right, they wanted to preserve search functionality and also wanted that files were stored encrypted so anyone with access, even if they managed to extract the file from DB, it would be useless. Anyway, the have been warned :) I agree, encryption at rest would be best. But I have heard is expensive... Thanks for your help. Regards. Jorge El lun., 30 sept. 2019 a las 5:52, Marcel Reutegger (<[email protected]>) escribió: > Hi, > > On 27.09.19, 14:39, "jorgeeflorez ." <[email protected]> > wrote: > > thank you for your reply. Yes, it seems pretty much impossible to do > > that. Now I am trying something "simpler": I want to encrypt a file > > before adding it to a node. I am able to do that, but I realize I will > > lose text extraction because what FulltextBinaryTextExtractor will > > receive is the encrypted file. Any ideas how can I manage to decrypt > > before passing the file to tika? > > Well, even if you manage to do this, you will have another problem. The > lucene index files will contain data that allows someone with access to > the files on disk to reconstruct the text in the document. I don't think > there's a solution to your problem. I would still suggest the system should > rely on encryption at rest mechanisms as supported by MongoDB. > > Regards > Marcel > >
