Automatically updating the pdf of an item when a new version is available

Is it possible to automatically update the pdf by replacing it with a newer available version? For instance, for services like arxiv authors have the ability to update pdfs by producing newer versions (v1, v2...etc). Since zotero is also saving metadata such as the link when I save something from arxiv is it possible to have the pdf automatically (or some user option) update whenever a newer version goes up on arxiv?
  • Not automatically — you can run Find Available PDF, but only if there's not an existing PDF on the item. If you know a newer PDF is available at the same URL, you can delete the current one and run that (and you can always restore the old one from the trash if necessary).
  • edited October 3, 2019
    @dstillman that's almost good, cause it requires the user to go and check for each pdf if there's a newer version. Imagine having huge number of collections and pdfs, having this automatic feature can save the day or the month. I wonder how much difficult would it be to alter the functionality of Find Available PDF to encompass this? I presume it wouldn't be that much of a work if the metadata/link is already there check and fetch if online version is newer than current one.
  • How would it know it was newer?
  • edited October 3, 2019
    Whenever you visit a file on arxiv.org with an identifier it always preselects the newest version for you.

    For example in this link https://arxiv.org/abs/1312.6120 there are three versions of the pdf named v1,v2,v3 and whenever you visit the link v3 is the one automatically selected, now I suppose this is done on the arxiv end and much less on my browser end (but I could be wrong). So, in other words seems doable to do a check if current file is v1 and the one on the link from the item metadata is v2 then fetch v2, or at least allow the user to manually do via Find Available PDF without having to delete the file so that we can select multiple files across collections and automatically update them, this way all the work we've done through the years won't be staying static as staying static in todays world might as well be considered suicide. Also that would give zotero an edge compared to other reference managers.
  • For the specific case of arxiv.org, given the arxiv ID it's possible to use the arxiv API to retrieve an Atom response that includes the item's element, like so:

    curl http://export.arxiv.org/api/query?id_list=1312.6120

    ...

    http://arxiv.org/abs/1312.6120v3
    2014-02-19T17:26:57Z
    ...

    So the underpinnings for arxiv.org specifically would make a function like this technically feasible (and possibly modular like site translators). But determining the user experience would be the hardest part.

    If a "find updated document" function is implemented, I hope we would have the option of just being notified about the updated document's existence rather than having it silently replaced in the background. I suppose some of the citation data would have to be updated as well; dates would certainly change, as might titles and authors.

    Provenance is important: people need to know what version of the paper they cited.

    In fact, a "found a newer version of this paper" function might have to offer users one of the following options, to avoid disrupting different workflows:

    a) do nothing: keep the existing version
    b) replace the existing citation metadata & captured document in-place
    c) create a new citation & captured document, keeping the original in-place

    Seems complex.
  • The new updated pdf and citation information along with the merged annotations (with an option to merge) would be a killer. I don't know how the latter would be implemented. Zotero already has the ability to extract annotations and import it to zotero format from external annotators like Okular. In case of merging, the difficulty is in finding the exact text in the new version. The an annotation in the new pdf could be kept if the exact text is present, else ignored.
Sign In or Register to comment.