unAPI (+MODS) translator no longer working in Zotero 3.01

monnardj
We have a web site serving references in MODS format via unAPI. With the previous version of Zotero (up to 2.1.10), references could be imported perfectly directly from Firefox. But since updating to Zotero 3.01 (and Firefox 10), it's no longer working.

Steps to reproduce :
- go to http://elearning.unifr.ch/bibup/index.php
- the Zotero icon should appear in the address bar, but it doesn't.
If you want to test it with a single record, you can enter "nte1" under "Tag".

Under "Report errors", there is nothing. I've also added the debug output log at the end of this message, which doesn't help me a lot.

Some additional info :
- the above web page validates correctly with the unAPI validator at http://validator.unapi.info/
- if I copy the MODS for one record generated by our unAPI server, it imports correctly in Zotero via "Import from clipboard".

Since we didn't change anything on our site, I assume there is a problem with the unAPI translator in Zotero 3.01

Thanks for any help on this,
Jacques

--- Debug output when loading the above page

(3)(+0013706): Translators: Looking for translators for http://elearning.unifr.ch/bibup/index.php

(4)(+0000002): Translate: Binding sandbox to http://elearning.unifr.ch/bibup/index.php

(4)(+0000002): Translate: Parsing code for unAPI

(3)(+0000002): HTTP GET http://elearning.unifr.ch/bibup/unapi.php

(3)(+0000099): HTTP GET http://elearning.unifr.ch/bibup/unapi.php?id=3761

(4)(+0000047): Translate: Parsing code for COinS

(4)(+0000001): Translate: Parsing code for DOI

(4)(+0000001): Translate: Parsing code for Embedded Metadata

(3)(+0000001): Translate: All translator detect calls and RPC calls complete

(5)(+0000000): Translate: Running handler 0 for translators

(4)(+0010359): Registering observer for [collection,search,share,group,trash,bucket] in notifier with hash Mw'
  • adamsmith
    I still see the folder in FF 9 - Zotero 3.0.1 - but I get a "not well-formed" error on trying to import (and also if I go to one of the mods displays.
  • Simon
    Detection was apparently broken by Firefox 10, not Zotero 3.0.1. It looks like the way Firefox enumerates the properties of NodeLists has changed. I'll look into both the detection and translation issues.
  • adamsmith
    Simon - while you're at it - people have been reporting that this doesn't work for connectors. Could you check that, too?
  • Simon
    MODS won't work in connectors, because it still uses E4X. Anything else should.
  • Simon
    FWIW, this broke detection.
  • DWL-SDCA
    From the provider of a bibliographic website this is a real problem for me and my database users. (SafetyLit.org is a free service presented without advertising.) I learned about this from a user who needed to download in RIS format and then import the record to ZSA.

    I can confirm that the SafetyLit database no longer displays the Zotero document icon in the address bar when using FF 10 but still does with FF 9 and earlier.

    Is there something that I can do at my website to get around this (without needing to kill my unAPI and replace it with some other metadata visualization)?

    After reading this thread I did vote + on the bug at bugzilla.

    Thanks.
  • Simon
    edited February 3, 2012
    I've added a workaround for the Firefox 10 detect issue to the unAPI translator. It should be automatically updated within 24 hours, or you can press the "Update now" button in the Zotero preferences.

    However, unAPI has additional trouble on the page in the first post because it has to retrieve a lot of files (which takes a little while) and also because at least one of the unAPI responses is not valid XML. (http://elearning.unifr.ch/bibup/unapi.php?id=3700&format=mods has an unescaped ampersand, which is not valid.)
  • DWL-SDCA
    Thank you. We escaped our angle brackets but I guess overlooked other problem characters. We are already working to fix the unescaped characters.

    Do you have any other advice for improving performance when folks download to Zotero? I'm really pushing my students to use Zotero and it would be great if my site integrated in the best possible way.
  • monnardj
    Simon,

    Thanks for the updated translator. Everything is working fine now ! And the code that generated the invalid XML you rightly spotted in one of the references has been fixed.

    Regarding the large number of items on the page : we have a new version of our application coming up, which will only show references for a specific tag. So the list will be shorter and faster to load.

This is an old discussion that has not been active in a long time. Before commenting here, you should strongly consider starting a new discussion instead. If you think the content of this discussion is still relevant, you can link to it from your new discussion.

Sign In or Register to comment.