Page MenuHomePhabricator
Create Task
Maniphest T214270

Thumbnails of many scanned PDF books show OCR text instead of scanned pages
Closed, ResolvedPublic
Actions

Description

Many scanned PDF books shows OCR text instead of scanned pages recently. Even those, which render properly some time ago.
Few examples:
https://commons.wikimedia.org/wiki/File:Z_pogrzebu_Mickiewicza_na_Wawelu_4go_Lipca_1890_roku.pdf :

Other examplws of affected books:
https://commons.wikimedia.org/wiki/File:PL_Dumas_-_W_pa%C5%82acu_car%C3%B3w.pdf
https://commons.wikimedia.org/wiki/File:PL_Molier_-_Pocieszne_wykwintnisie.pdf
https://commons.wikimedia.org/wiki/File:David_George_Hogarth._Accidents_of_an_antiquary%27s_life._1910.pdf
https://commons.wikimedia.org/wiki/File:Eddington_A._The_Theory_of_Relativity_and_its_Influence_on_Scientific_Thought._1922.pdf
...
In the above cases old thumbnails display OK, but an attempt to generate a new one gives OCR or an empty page.

Related Objects
Search...

Event Timeline

Ankry created this task.Jan 20 2019, 7:58 PM
Restricted Application added a subscriber: Aklapper. · View Herald TranscriptJan 20 2019, 7:58 PM
Ankry added a comment.Jan 20 2019, 8:02 PM

Please note: PDF rendering is critical for Wikisources where users dgitize books. Providing an OCR layer instead of real page images is misleading contributors who may interpret OCR errors as print errors. Please advice about urgent fix.

Even 404 errors might be better here than "fake" scans.

Matlin added a subscriber: Matlin.Jan 20 2019, 8:17 PM
Aklapper renamed this task from Broken rendering of PDF based books to Thumbnails of many scanned PDF books show OCR text instead of scanned pages.Jan 20 2019, 9:13 PM
Aklapper added a parent task: T43371: Thumbnail/imagescaler (tracking).
Aklapper added a project: Thumbor.
Zdzislaw added a subscriber: Zdzislaw.Jan 20 2019, 9:20 PM
• Gilles added a subscriber: • Gilles.Jan 21 2019, 11:04 AM

This is another occurence of the ancient ghostscript we use in production having issues with specific files. It outputs the following while falling back to the OCR layer:

gilles@thumbor2001:~$ gs -dFirstPage=1 -dLastPage=1 -dNOPAUSE -sDEVICE=jpeg -dJPEG=90 -sOutputFile=foo.jpg -dSAFER -dBATCH -r150 -q Z_pogrzebu_Mickiewicza_na_Wawelu_4go_Lipca_1890_roku.pdf 
unable to decode JPX image data.

   **** Warning: File has insufficient data for an image.

   **** This file had errors that were repaired or ignored.
   **** The file was produced by: 
   **** >>>> ABBYY FineReader 11 <<<<
   **** Please notify the author of the software that produced this
   **** file that it does not conform to Adobe's published PDF
   **** specification.

gilles@thumbor2001:~$ gs -version
GPL Ghostscript 9.06 (2012-08-08)
Copyright (C) 2012 Artifex Software, Inc.  All rights reserved.

Checking on the Beta Stretch Thumbor host, it converts fine and gives the expected output. This will be solved by the upgrade to Stretch.

• Gilles added a parent task: T170817: Upgrade Thumbor servers to Stretch.Jan 21 2019, 11:05 AM
• Gilles closed this task as Resolved.Feb 12 2019, 1:56 PM
• Gilles claimed this task.

Fixed by the ghostscript update. Trying purging affected files (I purged the ones in the task description).

Content licensed under Creative Commons Attribution-ShareAlike 3.0 (CC-BY-SA) unless otherwise noted; code licensed under GNU General Public License (GPL) or other open source licenses. By using this site, you agree to the Terms of Use, Privacy Policy, and Code of Conduct. · Wikimedia Foundation · Privacy Policy · Code of Conduct · Terms of Use · Disclaimer · CC-BY-SA · GPL