What is HaTeX?
html + latex = hatex
HaTeX is a LaTeX renderer for research writing on the web, aimed at computer science and AI. Papers in
these fields are written in a handful of conference templates, and HaTeX reads them as they are: the
template's title block, its two columns, natbib citations, and the bibliography from your .bib
or from the .bbl that comes with every arXiv source. A paper goes onto a blog or a project page
straight from its source, without being rewritten for the web.
Readers get the paper, not a PDF: equations, figures, tables and theorems numbered as LaTeX numbers them, every reference and citation a link, text that can be searched and copied, and a layout that fits any screen, in two balanced columns when there is room and one on a phone. The same source keeps working for the rest of your writing: posts, notes and beamer slides.
The work happens in the reader's browser. HaTeX parses the .tex source into HTML,
typesets the maths with KaTeX and compiles TikZ with real TeX running in WebAssembly, so there is
nothing to install and nothing to build. If you'd rather ship plain HTML, the same parser runs in Node at build time.
It covers the LaTeX that papers use, and a command it doesn't know degrades to its text instead of breaking the page. Papers in the manual lists the templates and shows how to bring over an arXiv source.