Bug#911446: diffoscope: pdf comparator no longer diffing metadata

[Chris Lamb]

tags 911446 + pending
thanks

Hi Reiner,

> I just noticed on some packages on tests.r-b.o with PDF differences that
> metadata like /CreationDate and /ID are no longer shown as a plaintext
> diff, but as a binary/hex diff instead.

This was dropped via #893702 in:

  https://salsa.debian.org/reproducible-builds/diffoscope/commit/5530623e9672c8afa4f186303bba29232b28ccd6

… due to the dependency on GCJ which is due to be be removed from
Debian.

In the aforementioned bug, Vagrant mentions that there is a pdftk
fork that does not require GCJ but packaging and maintaining this
would probably outside the scope of the Reproducible Builds team.

However, doko also mentioned PyPDF (thanks), which I've integrated
and implemented in:

  https://salsa.debian.org/reproducible-builds/diffoscope/commit/4e7ba71d40384c7bddc3365814cb91fd6e551790

  debian/control                        |  1 +
  debian/tests/control                  |  2 +-
  diffoscope/comparators/pdf.py         | 35 ++++++++++++++++++++++++++++++++++-
  setup.py                              |  1 +
  tests/comparators/test_pdf.py         | 13 ++++++++++++-
  tests/data/pdf_metadata_expected_diff | 13 +++++++++++++
  6 files changed, 62 insertions(+), 3 deletions(-)

Whilst I was doing this, I found & filed #911649.


Regards,

-- 
      ,''`.
     : :'  :     Chris Lamb
     `. `'`      lamby at debian.org / chris-lamb.co.uk
       `-