Last modified: 2013-07-25 07:08:44 UTC
Created attachment 11218 [details] Shows what's broken. For the corrupted pdf file at: https://commons.wikimedia.org/wiki/File:Gerald_Ford_Papers-_Final_Issues_for_Decision,_Army_Corps_of_Engineers-_Puerto_Rico_-_Compact_of_Permanent_Union_%281%29%28Gerald_Ford_Library%29%281554455%29.pdf The portion which normally says: Size of this preview: ... Full resolution ... appears to have the formatting wrong for this portion. It is highlighted in the screenshot.
I can reproduce the issue. I used https://commons.wikimedia.org/wiki/File:Gerald_Ford_Papers-_Final_Issues_for_Decision,_Army_Corps_of_Engineers-_Puerto_Rico_-_Compact_of_Permanent_Union_%281%29%28Gerald_Ford_Library%29%281554455%29.pdf?uselang=qqx to determine that the specific message that isn't being rendered correctly is: (file-info-size: , , (size-megabytes), application/pdf)
Created attachment 12000 [details] Another example showing the import of a PDF
I have the same problem also, it seems to be with all PDFs and not just this one. I'm wondering about my version of gs or pdfinfo (8.70 and 3.03).
So 2 issues here. *when metadata fails to be extracted, dimensions should default to 0×0 so the failure is graceful. * metadata extraction fails for that image (you can tell by looking at iiprop=metadata in the api). This could hust be a broken image but should investigate. Re comment 3 - if it happens for all pdfs for you probably an issue with pdfinfo command
Ah, I think I figured it out. Most of my problems went away when I went from pdfinfo 0.22 to 0.12.4. My next problem was dealing with an out of memory error when running gs, but messing with ulimit helped with that too. - A
Related URL: https://gerrit.wikimedia.org/r/59530 (Gerrit Change I1a49474309e15808928f877dfc29ae366d028928)
Related URL: https://gerrit.wikimedia.org/r/59531 (Gerrit Change I8c956da326e5dc339893a010370d399e97e204fd)
patches were merged.
