Improve PDF metadata retrieval (Google Scholar)

[Joeran]

• apparently, entering the captcha does not help currently. So, I assume there is a bug?
• maybe we should limit the number of requests to 1 (instead of 3) by default (and make a note in the dialog that users might want to increase the number)?
• maybe there should be a small delay between sending e.g. the title to Google Scholar and requesting the BibTeX data. I assume that if one IP requests multiple BibTeX entries within milliseconds, this is a strong indication for a bot.
• chose a new http header after the captcha was entered?

[silberzwiebel]

Not sure, if this is the right place to report a possible cause of a bug that leads to a non-working google scholar metadata retrieval for some PDFs. I noticed that when I'm in the network of the university, the retrieval works only for some PDFs. For others, the no search result is returned because the metadata search does not finish (there is an information saying "Finished 0 of 1 Request(s)" and an endless rotating circle).

I suggest this problem is related to extra information added to Google Scholar if you are in the university network such as a link to Web of Science or a link that the full text is available through the university library. Somehow docear's metadata scraper gets confused by this and stops working. You may find screenshots and logs of a working and a non-working example under https://kartoffelsalat.ddns.net:8001/d/586327a604/