PNG prediction decode only decodes part of the image

[GoogleCodeExporter]

When using PDFs containing PNG images with prediction > 10, the current 
implementation only decodes part of the image (1/3 of each row of the image).

Luckily, I already found the problem and I will attach a patch with a possible 
solution :)

Original issue reported on code.google.com by chirila....@gmail.com on 17 Sep 2013 at 9:55

Attachments:

• png_prediction.patch

[GoogleCodeExporter]

The problem was that in:

for i in range(1, numSamplesPerRow):
    # apply reverse prediction to current scanline (bytes)

numSamplesPerRow contained the number of sample, but the actual number of bytes 
are 
numSamplesPerRow * bytesPerSample + 1, so if bytesPerSample was 3 (as it was in 
my case), only a third of the image would get decoded.

Original comment by chirila....@gmail.com on 17 Sep 2013 at 10:00

[GoogleCodeExporter]

Hi! 

Thanks for the feedback! Could you attach or send me a PDF sample to work with 
it and verify the problem, please?

Thanks!

Original comment by josemigu...@gmail.com on 18 Sep 2013 at 10:01