amazon.co.jp: unfound ISBNs common, not failing with error

A colleague asked me to look into grab failures on amazon.co.jp. Here is one of the sample records:

http://www.amazon.co.jp/災害対策基本法解説-防災法研究会/dp/4421003549/ref=sr_1_3?ie=UTF8&qid=1435796822&sr=8-3&keywords=災害対策基本法解説

The translation attempt just hangs there. I suppose it will eventually time out, but no record is found.

Looking at the flow in the translator, it's detecting that the place of publication is missing, and attempting a search by ISBN - but the ISBNs of Japanese publications are not available from the search sources, so we get no record at all.

A little scratching around has not turned up a source for Japanese ISBN-searchable metadata. The only solution I can see is to use the isAsian flag set in the translator to bypass the ISBN search, and save a record with the limited metadata found in the Amazon record. I'll set that up for the time being in Juris-M. Would it be acceptable to adopt the same solution on Zotero-side? If there is a better solution, that would of course be, ah, better.
  • The Amazon translator is taking its time on this example (because of the ISBN lookup) but it is finishing for me. After the first time, the translation runs now fast. Are you sure that translation is not finishing after some time?
  • How long is "some time"? I've let it run for 10 minutes, with no result:

    http://i.imgur.com/SRndbK1.png

    Maybe the behaviour differs depending on the calling location? If the first download takes >600 seconds, I think it's reasonable for a user to give up.

    When it eventually clears, do you get a place of publication in the record?
  • Now, it only takes a second. However, the first time was longer (~1 minute and I pressed "run detect" in scaffold during waiting). It actually showed the failed debug message, and no it does not a get any place in the end.

    Maybe, someone else can test this example also?
  • works for me on Firefox ~20-30secs.
  • So amazon.co.jp does not seem to be working from inside Japan ...
  • Here's the JS console during the transaction. I don't know which of the search targets uses VeApps, but there is no activity after this point.

    http://i.imgur.com/KXPsiVK.png
  • I can confirm the same behavior as Frank is reporting. No solution... but it's breaking from within US too.
  • Frank, can you try updating your translators and trying this again? There was an issue with Lulu translator, which I fixed and that may have been causing problems. That would still mean that there are further issues with the way we handle failures to fetch invalid URLs though.
  • Yes, that's done it, thanks. Same behaviour as the other reports.
Sign In or Register to comment.