move full text extraction to hydra:derivatives

[elrayle]

It is desirable to have all derivative generation code in hydra:derivatives to create a single API for generating derivatives. This allows for the generation of derivatives to be configured by filetype in #makes_derivatives in file derivatives.rb.

The process for a hydra-works based app to override #makes_derivatives needs to be defined and documented.

This will likely replace the service oriented approach for derivatives including Hydra::Works::GenerateThumbnail and proposed Hydra::Works::GenerateExtractedText (see PR #213).

QUESTIONS:
Looking at Hydra::Works::GenerateThumbnail, object.build_thumbnail is called prior to calling object.create_derivatives.

• Do files need to be created prior to calling #create_derivatives for all potential derivatives?
• How would the app know which files (e.g. #build_thumbnail, #build_extracted_text) to create since it may vary based on the generic file's primary content type?
• Is it possible to do this automatically based on the configuration of setup in #make_derivatives

[elrayle]

Related discussion in issue #215

[jcoyne]

I'm for moving full text extraction to hydra derivatives. Please note that makes_derivatives has already been removed from the API. I think the object.build_* belongs in the output file service that you pass to the Hydra::Derivatives Runner.

[jcoyne]

Closed by #228