Panic on image extract

[sanbornm]

I'm getting panic: runtime error: index out of range [45675] with length 45675

on pdfcpu/pdfcpu/pkg/pdfcpu/writeImage.go:346. when trying to extract images from PDF.

Is this potentially an off by one error? Maybe it just needs an length check and if it is out of bounds to use the defaulted alpha := uint8(255)?

Linux Arch
6.8.8-arch1-1 #1 SMP PREEMPT_DYNAMIC Sun, 28 Apr 2024 15:59:47 +0000 x86_64 GNU/Linux

Commenting out alpha = im.softMask[y*im.w+x] makes my PDF extract images without panic

[sanbornm]

I think this is the same bug as: #804 sorry for the additional issue.

[hhrutter]

Perfect!
Could you share this file in order to get this fixed right way?
🙏🏻

[sanbornm]

Sent via email. I hope that's ok.

[hhrutter]

The file seems to be corrupt:

out pdfcpu val 2024-rs-excise-q1.pdf
validating(mode=relaxed) 2024-rs-excise-q1.pdf ...
Read: xRefTable failed: pdfcpu: corrupted last xref section

Neither does it open with Adobe Reader or Mac Preview.

[sanbornm]

Thanks for taking a look. Even though the file is corrupt and only somewhat readable I feel like it shouldn't panic. Especially from using an index out of bounds type error (length should be checked and image skipped?) I could wrap a recover around image extract but would prefer if it didn't panic.

[hhrutter]

My point was that the attached file got damaged probably during or on the way to the upload.
The error above is an indication for it - it crashes right at the beginning when trying to load the xreftable.

...unless you can also not open this file with Adobe Reader - then it would be corrupt for sure and needs fixing anyways.

[hhrutter]

This is fixed with the latest commit!

[sanbornm]

Thank you!!