hydra-file_chracterization

Hydra::FileCharacterization as (extracted from Sufia and Hydra::Derivatives)

Purpose

To provide a wrapper for file characterization

How To Use

If you are using Rails add the following to an initializer (./config/initializers/hydra-file_characterization_config.rb):

Hydra::FileCharacterization.configure do |config|
  config.tool_path(:fits, '/path/to/fits')
end

To use the characterizer:

characterization_xml = Hydra.characterize(file.read, file.basename, :fits)

# This does not work at this point
fits_xml, ffprobe_xml  = Hydra.characterize(file.read, file.basename, :fits, :ffprobe)

• Why file.read? To highlight that we want a string. In the case of ActiveFedora, we have a StringIO instead of a file.
• Why file.basename? In the case of Fits, the characterization takes cues from the extension name.

Registering New Characterizers

This is possible by adding a characterizer to the Hydra::FileCharacterization::Characterizers' namespace.

To Consider

How others are using the extract_metadata method

• https://github.com/curationexperts/bawstun/blob/ff8142ac043604c11a6f57b03629284bfd3043ea/app/models/generic_file.rb#L173

Todo Steps

• ~~Given a filename, characterize the file and return a raw XML stream~~
• ~~Provide method for converting a StringIO and original file name to a temp file with comparable, then running the characterizer against the tempfile~~
• ~~Provide a configuration option for the fits path; This would be the default for the characterizer~~
• Update existing Sufia implementation
  • Deprecrate Hydra::Derivatives direct method call
  • Instead call the characterizer with the content
• Allow characterization services to be chained together
  • ~~This would involve renaming the Characterizer to something else (i.e. Characterizers::Fits)~~
• Provide an ActiveFedora Datastream that maps the raw XML stream to a datastructure