Currently citoid cannot retrieve metadata from a link to a pdf.
Zotero does this, but there's no support for the feature in translation-server: https://www.zotero.org/support/retrieve_pdf_metadata
Example request that fails: http://query.nytimes.com/mem/archive-free/pdf?res=FB0610FF3F5912738DDDAC0A94D8415B858CF1D3
There is an open bug for this in translation-server here: https://github.com/zotero/translation-server/issues/70