Local HTML -> PDF rendering

[captn3m0]

# Title

- [chapter 1](chapter1.html)
- [chapter 2](chapter2.html)

Convert the HTML to PDFs and merge accordingly.

[captn3m0]

Notes from PDF rendering research in Python:

• The best results come from running pandoc, but that comes with latex dependencies, which we want to avoid.
• The next best tools are:
  • wkhtml2pdf, which also has installation concerns. I don't like the render quality either.
  • xhtml2pdf which uses PyPDF2 (😞) and reportlab underneath.
  • Directly using reportlab, but this has a few native dependencies that it comes bundled with. (libfreetype, libpng, libz)
  • And finally, rinohtype, which is pure-python but is heavily under development and is missing some important features for now (such as remote images)
  • weasyprint, which I haven't tried much yet.

Did some experiments with all of the above to get close-to-pandoc typography. Will add some more details here.

[Vonter]

I tried out Weasyprint and the results look promising to me, but it does have a couple of external dependencies, including Pango. However there do seem to be plans to minimize the external dependencies

[captn3m0]

My recommended options are (not in any order):

• reportlab (dependencies are bundled in the wheel, so not much concerns)
• xhtml2pdf (But with a fix for Switch to a better maintained PyPDF Fork xhtml2pdf/xhtml2pdf#560)
• rinohtype, which is pre-stable currently, but holds good promise. Even without remote images, I think it will make for a good start.

[captn3m0]

I filed a PR to xhtml2pdf for PyPDF3 support: xhtml2pdf/xhtml2pdf#582.

[captn3m0]

xhtml2pdf moved back to pypdf2: https://github.com/xhtml2pdf/xhtml2pdf/pull/656/files which is now merged back to the pypdf project. PyPDF3 is unmaintained.

xhtml2pdf is getting regular updates now. There might even be scope to just let people pick between reportlab, xhtml2pdf, rinohtype, or borb as different "renderers".