In an ideal world, I would like to download PDF articles with an automatically generated field for the bibliographic data. ReferenceMiner for Bookends can do this with PubMed, the Library of Congress, and Amazon. I’m not sure whether this can somehow be integrated with data supplied by e-journals.