… is a self-host TeX compiling service with
- Simple HTTP API
- Result caching
- Missing package detection (WIP)
and other experimental features.
It is designed to use this with [unamed], a simple JavaScript library that finds (La)TeX snippets in a HTML document, turn them into complete (La)TeX documents, feeds into this and then get neatly fitted SVGs.
It is inspired by the karasu project, which allows mixed Markdown and LaTeX in the same document; karasu solved this problem by converting LaTeX contents to SVGs first.
Comparing Markdown and TeX, Markdown is minimalistic, but TeX is far more
powerful than Markdown. Being able to use both in the same document would be
quite interesting. The karasu project managed to do this, but adapting that to
other environments turns out to be tricky because it is written in Haskell.
I managed to write a simple Lua filter for Pandoc that does the same, but I
still don't like the solution. Is there a chance that we can render them on
the fly?
Other notable projects while I was creating these two projects:
- GladTeX, a program with similar approach; however it claims to have only partial Unicode support.
- LaTeX SVG, a Pandoc filter that works in a similar way, written in Haskell.
- TinyTeX, a customized TeX distribution that manages to implement auto-installing required packages.
You will need
- Python3
- pip3
- TeX Live distro. Currently it is recommended to use the full distro, but
once the missing package detection is available, you can use a minimized
distro for faster deployment. At the very minimum, you should have
tex,latex,pdftexanddvisvgmavailable in yourPATH. - Redis, unless you want to use alternative cache backend.
# Setup virtual env.
python3 -m venv venv
# Activate the virtual env.
. venv/bin/acvitave
# Install dependencies.
# Currently, we depend on Flask, Flask-Cache and redis.
# Redis is not required if you choose other caching method.
pip3 install -r requirements.txtWIP
gunicorn -w 4 -b 127.0.0.1:8080 -k gevent app:app