Attaching files

bwbryson

I am a new Zotero user and like many aspects of the program but I am having trouble attaching files to my library. When dragging and dropping PDF files into the library or when trying to work within the attachment menu to attach a file from my computer to an existing listing...the spinner appears and never stops.

I wonder if it is possibly related to the word plug-in I just added.

Thanks

dstillman

Again: http://www.zotero.org/support/reporting_bugs

If you can't provide a Report ID, you'll at least have to tell us what OS you're using.

jonscomo

I also am having this problem.

I am on Mac OS X 10.6.5.

More specifically, I believe it is a problem in the indexing of PDF files. With large PDF files (of the 8 I just attached, the three that caused problems were 2.2 MB, 2.8 MB, and 3.9 MB), the browser goes out to lunch (in Mac, it is the rainbow spinner), and you have to force close, then restart the browser. For these files that cause a freeze, when I started Firefox back up and looked at the reference, the PDF file had been imported and associated, but the file had not been indexed. Looks like there might be a problem in PDF indexing? When I subsequently click the arrows next to the "Indexed: No" indicator for the PDF, the indexing completes just fine.

On restarting the browser, there are no error messages.

This also manifests in import of file from EndNote in RIS. I just finished rewriting the import part of RIS.js for myself, and so I did lots of testing, and in the context of imports from EndNote, once you fix it so it can figure out where the PDF files are, the import frequently goes out to lunch in the midst of imports where references have associated PDF files - when I commented out PDF import, it all worked just fine, even for large numbers of files.

If it would help, I can turn on debugging and run files through again until I get an error, and then submit that.

Thanks,

Jonathan Morgan

adamsmith

might be the same as here:
http://forums.zotero.org/discussion/15468/crashing-on-mac-osx/
and also sounds a lot like the problem with indexing in linux:
http://forums.zotero.org/discussion/4935/firefox-freeze-when-attaching-document/

as a start, it would be helpful to test if this goes away by removing pdfinfo - and if that doesn't help, both pdfinfo and pdftotext from the Zotero data folder.

jonscomo

I will try that a little later today. So, to be clear, I first remove pdfinfo-MacIntel from zotero library folder and see if that fixes problem, then, if not, also remove pdftotext-MacIntel?

This does sound similar to 15468 and 4935. Let me know if you want me to move subsequent responses into one of those threads.

Thanks,

Jonathan

jonscomo

With the one large file that was consistently causing problems, moving pdfinfo-MacIntel out of library folder fixed the problem. Indexing doesn't work now (so is that the PDF parsing and indexing program?).

I'll back up my library and try one of the monster imports I performance tuned RIS.js on. When I was importing, I thought this was my javascript code breaking - the current RIS.js essentially doesn't deal with EndNote PDF files well, so it almost never actually imports a PDF unless you massage the RIS file. Once I fixed that, it started crapping out whenever there were attached PDFs. On my machine, it doesn't break on all PDFs, but large PDFs and lots of PDFs in one import increase the likelihood that the import process will go out to lunch, and when that one dies, nothing is actually added to the database, so you have to start over. I took all regular expressions out of RIS.js because they seemed to exacerbate the problem - would be nice to be able to put them back in, though I need to figure out if I can get some of the changes I made moved into the real RIS.js, so that I don't have to merge changes over time.

At any rate, I'll do some more testing and see if batch import works better now, too.

Thanks,

Jonathan

adamsmith

I first remove pdfinfo-MacIntel from zotero library folder and see if that fixes problem, then, if not, also remove pdftotext-MacIntel?

exactly. Actually, I believe if you just move pdfinfo indexing still works, but Zotero doesn't know/show the status of a file as indexed.
In Linux it was necessary to remove both to fix this.

Getting the RIS pdf import fixed is a huge service - as I'm sure you have seen this comes up over and over again. thanks.

jonscomo

My challenging test worked OK, too, with pdfinfo-MacIntel moved out of library - slight pause at the end, but did not crash.

Would you like me to start with a fresh library and try installing my whole export at once (500+ references, probably 400 PDFs)?

Jonathan

jonscomo

It would be good to see if we can get this freezing on PDF import worked out. The import periodically going out to lunch is the only thing that makes me nervous about the revision to RIS.js I did. If it turns out that it is the PDF import and not the javascript code that is causing the freeze, that will make me feel better.

So is the fix right now to just move the pdf* files out of the way? Is there anything I can do to help debug more substantially?

Jonathan

jonscomo

With RIS PDF import, the one thing I don't know how to do is give the user a way to tell the script where their EndNote PDF folder is. I have a variable in RIS.js in which you place the absolute path of the EndNote PDF folder, then it takes care of detecting the tag for EndNote PDFs and using regular expressions (or, right now, less fancy low-level string manipulation, since I stripped all the RE out) to convert it to a file:// link. It also has a map of file extensions to mime types so it can better detect MIME type (I had some attached Word Documents, also), and it parses the file name off the end and uses that instead of the generic string it was attaching as the name of the attachment. In addition, I rewrote the whole RIS.js import, though, to add objects and get the big giant conditional out of the import, so I could fix some of the wierd ways EndNote maps its fields to RIS (tags map to an ImportField object instance now that specifies different types of actions for a given tag - either a direct one-to-one mapping from tag to zotero item field, or a function pointer to function to use to process the tag), so it is pretty different, and so I might have moved it too far from the original to make it easy to use it as-is. The code is linked out of the zotero-dev google group, though, so any pieces of it that you can use are yours for the taking, and because of the function pointers, the code for indivudal tags is compartmentalized in functions, so it should be easy to grab pieces out and re-use.

Let me know if there is more I can do to help.

Jonathan

jonscomo

Also, I did eventually have to remove both pdfinfo-MacIntel and pdftotext-MacIntel to get indexing and imports to work. Any idea when this might be fixed?

Jonathan