ps2pdf compression timeouts for large pdf files

[lenke182]

Using dangerzone on a big pdf file with 243 pages results in a timeout for the compression in ps2pdf. Everything else seems to work fine. I suggest to increase the timeout or calculate it based on the page count, since smaller pdf files wont take that long to compress in contrast to bigger pdf files.

https://github.com/firstlookmedia/dangerzone-converter/blob/master/scripts/pixels-to-pdf-unpriv#L125

Merging 243 pages into a single PDF
Compressing PDF
Error compressing PDF, ps2pdf timed out after 60 seconds

[lenke182]

This seems to work great on most of my tested pdf files.

Some pdf files (also around 250 pages) are causing a timeout of pdfseparate. Problem is, that the script at that point doesn't know the page count yet, so the timeout can't be calculated proportionally.

https://github.com/firstlookmedia/dangerzone-converter/blob/master/scripts/document-to-pixels-unpriv#L151

Separating document into pages
Error separating document into pages, pdfseparate timed out after 60 seconds

sudo: unable to resolve host 9227720a63d0: Temporary failure in name resolution

[yveszoundi]

@lenke182 , I encountered similar issues with large PDF documents of roughly 700 pages. I ended up increasing timeouts (proportional to page count) and switching from pdfseparate to pdftk.

My solution consisted into the following changes:

• Global timeout based on script leveraging pdfinfo: global_timeout = pdfinfo_page_ct * 30
• Changes in dangerzone-converter
  • Introduce a new dependency pdftk and libvips-tools in the Dockerfile
  • In dangerzone-converter update all scripts and set the process run timeout to global_timeout
  • In document-to-pixels-unpriv, switch from pdfseparate to pdftk: args = ["pdftk", pdf_filename, "burst", "output", "/tmp/page-%d.pdf"]
  • In document-to-pixels-unpriv, swap pdftocairo with vips: ["vips", "pdfload", pdf_filename, png_filename]. vips appears faster than pdftocairo but I didn't benchmark it.
• Changes in dangerzone UI project
  • Use my custom Docker image by updating container.py and global_common.py
  • In tasks.py, skip image pulling as I'm using my own Docker local image.

There are still cases where the GUI stops printing command line output (10+MB PDF files)... I'm no Python connoisseur but in such cases I just "watch the contents" of the Docker volume folder and copy the generated safe PDF to its final destination (once the Docker process exits).

[yveszoundi]

@micahflee, I believe that for big PDF files many will face problems. My main use case for dangerzone is reading IT e-books (often 200+ pages). I'm only testing on an iMac, my other system runs Qubes OS.