Tika parser may not be parsing embedded documents

[tballison]

If calling Tika's parse() with 4 parameters (using the ParseContext), you need to add a Parser.class to the ParseContext to handle embedded documents.

See TIKA-2096 for a proposal to fix our API in Tika 2.0, and for a list of other projects (including Tika!) that fell victim to this.

I'd open a PR, but I can't quickly see how to test Tika. I'd recommend grabbing our test_recursive_embedded.docx.

[jnioche]

Thanks @tballison, I've implemented your suggestion and created a test environment for Tika, which will be useful for other things as well.

Just to check that I understand it properly : when dealing with embedded docs, is there a way to separate each individual subdoc or does it all get lumped up in a single object? StormCrawler can generate discrete subdocuments from an original one so we could use that.

[tballison]

Well, now that you mention it, :), see #361. Let me know if you have any questions.