Zotero DOI Manager cannot work

troyprobe

Hello, everyone! I find a problem with "Zotero DOI Manager" plugin.
It cannot get the doi of manay aritcles from the Crossref website. I checked the xpi, I find the OpeanUrl of Crossref is old. The new OpeanUrls have shown on the website https://www.crossref.org/education/retrieve-metadata/openurl/ But I'm not able to update the xpi. Can anyone help me to fix it? Thanks!

bwiernik

Can you give some examples items that it doesn’t retrieve?

troyprobe

Such as this airticle :"A multinational, drug utilisation study to investigate the use of dexmedetomidine (Dexdor) in clinical practice in the EU"
It can find in the Crossref Metadata Search (https://search.crossref.org/), But Zotero cannot get DOI by Zotero DOI Manager

adamsmith

[I haven't checked this out at all, but just to note that CrossRef did some maintenance today, so some APIs may have been temporarily down]

troyprobe

Are there any similar websites that can temporarily replace Crossref? Because I now have a lot of literatures to retrieve DOIs.

adamsmith

The DOI add-on works and it's uses the correct Crossref openurl resolver (see https://www.crossref.org/education/retrieve-metadata/openurl/#openurl-resultsa-id00381-href00381i-classfa-fa-linkia )

The OpenURL search requires more precision than the simple search interface. It finds the DOI for the above referenced title for me when I provide enough information, but I'd guess that the fact that the exact same title exists twice means that it does, in fact, require more detail

bwiernik

Yes, the CrossRef API needs at minimum the title and journal title, as well as either the first author or first page before it will even run a search. In practice, it also needs the publication year before it will return any results.

If I enter the title, journal title, date, and either the first author or first page for this item, DOI Manager will find the DOI.

It should give a more informative error message when it fails. I’ll look into that.

troyprobe

Thank you for your answer. I hope it can be fixed as soon as possible. This plugin is so useful!

bwiernik

To be clear, the only thing that needs to be fixed in the plugin is making a more informative error message. The item you posted finds its DOI fine if you supply enough information (at least title, journal title, date, and author/page).

andrew809

@bwiernik: Reviving an old thread, primarily to say thanks for this great tool. Of course, I also have a question about failed DOI retrievals. For example, I have this article in Zotero: "Zhang, H., ... , 2020. Anaerobic digestion based waste-to-energy technologies can halve the climate impact of China’s fast-growing food waste by 2040. Journal of Cleaner Production"

So I have the title, journal, date, and author (they're all correct: https://www.sciencedirect.com/science/article/abs/pii/S0959652620335356), but DOI Manager is not finding a match.

Do you have any ideas about what the problem could be, and is there anything a regular user can do to troubleshoot?

Many thanks!

andrew809

Although I'd tried systematically eliminating suspect fields in zotero, it looks like a big problem can be the dates, which didn't look to be suspect. The example I gave above was imported to zotero with a date in Month Day Year, which zotero correctly understood (i.e., in the citation above, it gave me just 2020). However, when I changed the date from m-d-y to just year, the DOI lookup worked. So... this might be helpful to others or to bwiernik.

bwiernik

@andrew809 CrossRef lookup of the metadata for that item simply fails. There isn't anything the you can do about it on your end. I am not sure it is fixable by me, either. Most items do retrieve correctly here, so I'm not sure if the problem is with Zotero's OpenURL structure for the item, with CrossRef's API, or with the publisher provided data.

@dstillman I think the issue with this item and the CrossRef OpenURL interface is that these items have a single article number rather than two page numbers. For example, both of these items fail with CrossRef OpenURL lookup:
https://www.sciencedirect.com/science/article/abs/pii/S0959652620335356
https://www.sciencedirect.com/science/article/pii/S000187912030097X

But this one, which is a journal with page numbers, rather than article numbers, works:
https://www.sciencedirect.com/science/article/abs/pii/S2213343715002420

The article-number-having articles fail whether or not I delete extra fields and whether I have or remove either the authors or the page number.

Is there something malformed about the OpenURL request for these items that could be fixed on Zotero's side? Do you have any suggestions?

dstillman

@bwiernik: I don't really know anything about OpenURL. We could experiment, but I suspect someone else knows more.

bwiernik

@dstillman Any suggestions for whom?
@adamsmith @zuphillip

bwiernik

Okay, it looks like the issue might be Zotero's use of the RTF namespace, which CrossRef seems to not accept reliably (anymore?).

For this item:
https://www.sciencedirect.com/science/article/pii/S000187912030097X

The Zotero generated OpenURL (with DOI deleted) fails:
https://www.crossref.org/openurl?url_ver=Z39.88-2004&ctx_ver=Z39.88-2004&rfr_id=info:sid/zotero.org:2&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Addressing common method variance and endogeneity in vocational behavior research: A review of the literature and suggestions for future research&rft.jtitle=Journal of Vocational Behavior&rft.stitle=Journal of Vocational Behavior&rft.volume=121&rft.aufirst=Brian&rft.aulast=Cooper&rft.au=Brian Cooper&rft.au=Nathan Eva&rft.au=Forough Zarea Fazlelahi&rft.au=Alexander Newman&rft.au=Allan Lee&rft.au=Martin Obschonka&rft.date=2020-09-01&rft.pages=103472&rft.issn=0001-8791&rft.language=en&pid=zter:zter321

But this OpenURL works:
https://www.crossref.org/openurl?title=Journal of Vocational Behavior&volume=121&aulast=Cooper&date=2020-09-01&spage=103472&pid=zter:zter321

The second one is based on these guidelines:
https://www.crossref.org/education/retrieve-metadata/openurl/

adamsmith

I actually think andrew809's diagnosis is right:
CrossRef does accept the rft namespace (see https://www.crossref.org/education/retrieve-metadata/openurl/#00388 ) but for some reason it seems to choke on full dates, even if they're in the metadata and work with the non rft version. If I run the exact same query you have above with rft.date=2020 it resolves.

I'm not seeing anything wrong with Zotero's openURL queries here. You could try filing an issue with CrossRef's github (they're not always super responsive, but they're generally very helpful when they do respond), though I seem to recall Geoff Bilder saying they don't do much with the openURL API anymore, of you could just strip the date parts in the openURL queries you're using.

bwiernik

I just call Zotero’s generateContextObject function. The call fails with Zotero’s CrossRef lookup for example. Could we fix the date issue there?

adamsmith

I don't think we'll want to "fix" this in generateContextObject function because that's also used to generate COinS and we don't unnecessarily want to lose data there (I'm fairly certain YYYY-MM-DD is valid rft.date).

I don't know the lookup code well enough to say whether we could do something about that there, but if I understand you correctly that also wouldn't help you?

bwiernik

I need to generate an OpenURL to feed to CrossRef’s API, and this also breaks Zotero’s CrossRef lookup feature. The generated URL could potentially be post-processed to remove month and day parts?

andrew809

If I had more coding capability, I'd offer to help. For now, thanks to you all for the thoughts and time. Both Zotero and the DOI manager are great tools!

adamsmith

bwiernik already proposed a code change, so I hope we can get that fixed quickly.