[poppler] CMYK JPX image interpreted as RGB (with testcase + example images)

[Harry Roberts]

Hi Poppler folks,

 I encountered a weird bug with some images in a PDF. It seems like CMYK
data is being interpreted as RGB data, or something similar to that - the
pixel stride is off causing the image to be repeated multiple times.

 PDF Page file can be downloaded from: http://derp.ltd.uk/843-21307-16.pdf

Example of problem image, extracted with pdfimages:
http://i.imgur.com/YSTCKkb.jpg
The image shows with `pdfimages` as:
page   num  type   width height color comp bpc  enc interp  object ID
---------------------------------------------------------------------
   1     7 image     759   185  cmyk    4   8  jpx    no        68  0

I have tested 2 versions of Poppler with 2 versions of LCMS and exhibit the
problem. Google Chrome also has similar problems.

   - Evince from Ubuntu 12.10 (Poppler 0.20.4-0ubuntu1.1 with LCMS2
   2.2+git20110628-2ubuntu4) without openjpeg
   - pdftoppm - Poppler 0.22.0 + LCMS 1.19 (custom build) - without openjpeg


Example page images:

   - Ghostscript: http://i.imgur.com/dfLLptb.jpg
   - Photoshop: http://i.imgur.com/BylwJWW.jpg
   - Poppler: http://i.imgur.com/8cderTo.jpg (shows buggy image)

The PDF was processed using SmartCreate by QuadGraphics, combined by Adobe
Acrobat 8.0... unsure of the original software used to create the
individual files.


On a related note I'm unable to build the current Poppler from git,
otherwise I would've tested with it too. Getting link errors like:
  Linearization.cc:(.text+0x4dc): undefined reference to
`error(ErrorCategory, long, char const*, ...)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/poppler/attachments/20130206/52151afb/attachment.html>