Pubmed translator not working

cavallad

I have been getting an error saving to Zotero 2.0b6.5 (FF 3.0.13) from Pubmed: see 484626930.

[JavaScript Error: "TypeError: text.match(/PMC\d+/) is null" {file: "file:///C:/Documents%20and%20Settings/David_/Application%20Data/Mozilla/Firefox/Profiles/uawroo9x.default/zotero/translators/PubMed%20Central.js" line: 0}]

I wonder if this is related to the line:

newItem.extra = text.match(/PMC\d+/)[0];

and should this change to:

newItem.extra = text.match(/PMCID:\d+/)[0];

??

Also, I have been getting an error from Google patents -- see http://forums.zotero.org/discussion/7489/google-patents-could-not-save-item-error/

Cheers, David

cavallad

Sorry, I mean Pubmed Central not Pubmed

sean

We're aware of the Pubmed Central errors and will have a fix soon.

amacom73

In addition to PMC translators still (at least in my hands) being broken, there appears to be a minor issue with the PubMed translator--it appears to work in all instances *except* the multiple item selector in Abstract view (it works fine in Summary and Abstract Plus view). Seems odd that the metadata structure underneath Abstract and Abstract Plus is different, but there it is. When the address bar icon is clicked, the multiple selector pane doesn't appear, just the "could not save item" error. Can't include a link to replicate the error since was a search result.

Bionatsci

As mentioned here the search results pages for pubmed central appear to be incorrectly treating PMCIDs as PMIDs (this is in addition to the translator for article pages being broken).

Example URL:

http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=367614&tool=pmcentrez

Hope this helps.

Edit: Sorry, I of course mean the Zotero translator for the search results pages is treating the identifiers incorrectly, the pages themselves are fine as far as I know.

amacom73

I hate this about PMID/PMCID--I really like PubMed, but I think whoever had the idea to make PubMed Central a completely different system with a mutually unintelligible set of unique identifiers ought to be fired. Or rather, forced to correct the design, re-code all the PMCIDs as PMIDs and then fired.

Plus a PMID can always be linked to with the simple syntax: www.ncbi.nlm.nih.gov/pubmed/PMID but no such luck with PMCID, for which the best I can figure out is:
www.pubmedcentral.nih.gov/articlerender.fcgi?artid=PMCID&tool=pmcentrez

noksagt

This drifts off-topic, but I thought I'd respond to amacom73:

If an article has a PMCID, it is open access & you know where to find it. Not every reference with a PMID is open access & being in PubMed does not help you locate an article. So: PMCIDs are useful. Hopefully LibX, etc. will link them.

You can use a slightly shorter linkhttp://pubmedcentral.nih.gov/articlerender.fcgi?artid=PMCID & you can create a keyword search in Firefox that types the bulk of the URI for you.

That said, I agree that getting rid of that articlerender.fcgi in the uri would be nice..

amacom73

Rather than drift off topic, we might as well veer!

I use keyword searches all the time for PM, DOI, etc, and I agree it's a reasonable thing to do for PMC as well. Incidentally, PubMed does automatically link to PMCID papers, so in a pinch the PubMed ID gets you to the PMCID page.

But I don't agree that PMCIDs are actually useful unless I'm missing something. Other than the PMC prefix, there's nothing implicit in the number itself that tells you it's a PMCID, it's just a 7 digit string. So why not have an article's PMCID numeric string be identical to the PMID? I think it's reasonable to assume that PMC papers are a fully contained subset of PubMed papers, so every PMC paper ought to map to a PMID, though obviously not vice versa. So a single unique identifier could be searched at either the PM or PMC sites without the profoundly silly PMID<=>PMCID translator.

At the very least, PMC should recognize PMID as a valid search term, which AFAIK it does not.

noksagt

Incidentally, PubMed does automatically link to PMCID papers, so in a pinch the PubMed ID gets you to the PMCID page.

If-and-only-if that paper exists on PMC. Many papers with a PMID are not on PMC. Many are not allowed to be (per copyright).

I think it's reasonable to assume that PMC papers are a fully contained subset of PubMed papers

I'm not certain this is the case. Their FAQ says that only "most" articled on PMC are listed in PM.

cryanking

Is this supposed to be fixed in 2.0b7.2? PMC is still not working for me.

elucify

PubMed and PMC have different id spaces because they don't overlap completely. There is content in PMC that is not in PubMed.

That said, it's a shame that PMC full text isn't just another way of looking at a PubMed record.

An upcoming release of PMC (I don't know when exactly) will support clean URLs for PMC articles, like in PubMed. There will be auxiliary URL forms in PMC for addressing PMC articles by their PubMed IDs, journal-volume-issue numbers, and DOIs.