This request came out of a discussion I had with Anurag Acharya and Darcy Darpa at Google / Google Scholar.
Anurag mentioned that the more hints we can provide the Google Scholar crawler around the primary file, the better. Currently, Google Scholar has to make some guesses around what file seems to be the primary file (This may or may not be related to DS-1387 – unconfirmed as of yet).
Anurag also mentioned that the url in the "citation_pdf_url" NEED NOT BE A PDF. We could just add in logic to ensure the "primary bitstream" (if one is marked as primary) is linked to from that 'citation_pdf_url'.
So, it may be possible to simplify our logic around what bitstream to link to in the "citation_pdf_url" in the GoogleMetadata class:
This is just an initial idea, but I wanted to track the discussion here.