Thanks - I tried it, it works! I will test more later today, but so far it works on everything incl patents and patent applications. That was really fast :)

I still get the Google Patents is not working ("Could not save item") error (ReportID: 1345975352), on any patent (I have tried). Example:
http://www.google.com/patents/about?id=KchEAAAAEBAJ

With both the newest stable version of Zotero as well as the beta
Zotero 3.0b2 (de/us), Firefox 6.0.1 (de/us), Windows 7 (de), all translators updated. Also tried the above.

The "Google Patents" translator doesn't extract correctly inventors from patents with several inventors and doesn't extract abstract. Example: http://www.google.fr/patents?id=Nh17AAAAEBAJ
I think also that this translator should set country to "U.S." automatically as all patents in Google database are U.S. patents (see http://www.google.fr/googlepatents/about.html)

Yes, it works like a charm! Thank you very much, Avram.

It works sometimes, mostly not. Which is a medium annoyance for my work, as I work with patents rather often.

Now I started to play around with translators (hoping to write ones for EU and German patens), and looked first into the Google Patents translator.

It appears, the/a problem is: the function detectWeb() looks whether substring "id=" is present in the URL, then return the document type "patent"

In most cases the "id=" is abscent: Google just changed something.

However, on the "Overview" type of patent view, in most (see below for exception) cases I tested the substring "ei=" is present.

I have replaced "id=" by "ei=" in the translator (file "Google Patents.js", line 23), and now both functions (detectWeb and doWeb) appear to work properly from Zotero Scaffold: the first one returns type "patent", the second one parses the page OK.

However, trying to save the data from the Firefox page still results in "Could Not Save Item" error - no idea why. The translator actually used is "Google Patents.js" (hovering the mouse pointer over the "Save to Zotero" icon)

There is another small issue. The URL may be in the form http://www.google.com/patents/US4390992?dq=plasma+Dean+Judd&ei=r6JDT4iDD8fYsgbSmN3qBA
or in the short form
http://www.google.com/patents/US4390992

The data scraping by doWeb looks OK (see below) in both cases, thus detectWeb has to be adjusted.

I could do that and more testing, if somebody explains me, how to find out why I still get the error message (as it apparently works from Scaffold), and probably helps me otherwise.

-------
From Scaffold:

Returned item:
{
"itemType": "patent",
"creators": [
{
"firstName": "William P.",
"lastName": "Arnold",
"creatorType": "inventor"
}
],
"notes": [],
"tags": [],
"seeAlso": [],
"attachments": [],
"abstractNote": "A rapid, semi-automated method for determining dibucaine numbers is disclosed wherein use is made of a unit dosage form of dibucaine or a test pack containing dibucaine in a unit dosage form.",
"country": "United States",
"extra": "U.S. Classification: 435/20\nInternational Classification: : C12Q 146",
"patentNumber": "4340667",
"date": "Jul 20, 1982",
"filingDate": "May 20, 1980",
"assignee": "The University of Virginia",
"title": "Rapid, semi-automated method for determining dibucaine numbers",
"url": "http://www.google.com/patents/US4340667?dq=plasma+oher&ei=raRDT7a3JsXftAa9_JnnBA",
"libraryCatalog": "Google Patents",
"accessDate": "CURRENT_TIMESTAMP"
}

the way to check for this, btw. is to look in your error report after getting a translator error.

OK, I think I've got this, but since I made some substantial changes I'd like some testing before we push this out.
Download the translator from here:
https://github.com/adam3smith/translators/raw/googlepatents/Google%20Patents.js
and place it in the translator folder in your zotero data directory:
http://www.zotero.org/support/zotero_data , replacing the file of the same name.
This should work from all google patent views. It should also (re-?)establish pdf download for patents.
Please let me know how it goes.

It works now much better ;)

Well, from the search results (patent list) it looks like all data are read in OK.

From the "overview" patent view - also.

But from "patent page" view like one below - the mileage varies. Sometimes it works, but in many cases only limited data are saved, e.g. no inventor names, no abstract, no assignee.

If I understand it correctly, for multiple items list, the script fetches the data not directly from the page, but from other URLs.

Would it make sense to apply the same technique for "patent page" views: get the "overview" page, then scrape the data?

BTW I'm not sure why I got the pages 7-8 of the patent in the link below - I think, I just clicked onto the search result.

http://www.google.com/patents?id=jH4DAAAAEBAJ&pg=PA9&dq=weltmann+plasma&hl=en&sa=X&ei=J4RGT4bWIaPS0QXwuPSKDg&ved=0CD0Q6AEwAw#v=onepage&q=weltmann%20plasma&f=false

please try again - same link as above.

thanks, that's very thorough testing - it's those two underscores in the id=y-__AQAAEBAJ
I'll think about whether we can just add those to the regex finding the IDs - seems like it should be OK. If you don't mind try a couple more and see if you can find anything else. Otherwise I'll add the underscore and that should be it.

in the last version of translator, on line 124 there is still
...id=[a-zA-Z0-9]...
which might probably result in failure in some cases, if I understand it correctly

OK, the fix (including for the last bit concerning the URL) is now up. If anyone else experiences any problems, there is no need to manually install the translator for anyone else - by now it should have gotten to all Zotero clients.

confirmed - the translator is currently broken (Zotero imports generic metadata from the page). We'll have a fix out asap.

(thanks - it's a big mess, I had a quick look and then decided it'd take a bunch of time).

@fbennett

Thanks for the information. I can wait patiently as long as I know it's coming. :-)