Advice on migrating from PDF to MathJax/Next.js (keeping .tex as source of truth)

[Vu Hung Nguyen (Hưng)]

Hi everyone,

I am looking to migrate my current website, https://hsc-math-hub.vercel.app/, to a web app using MathJax, while keeping the original .tex files as the sole source of truth.

Currently, the tech stack for HSC-Math-Hub is:

- PDF.js
- Next.js

The app simply reads PDF files—which are compiled directly from the .tex files—and displays the content on the web.

I am considering a new tech stack to improve accessibility and web-native rendering:

- MathJax (server-side rendering)
- Next.js
- make4ht (as the primary conversion tool)

However, I have run into a few challenges during my initial testing:
- TikZ diagrams: I am having trouble successfully converting them, and my curriculum materials rely heavily on them.
- Formula rendering: Some math formulas aren't converting or displaying correctly when parsed and viewed with MathJax.
- Build speed: make4ht feels quite slow, taking about 5 minutes (not too bad) to convert a 100-page document on my Mac.

Has anyone tackled a similar migration or built a pipeline like this?
I would greatly appreciate any advice on handling TikZ diagrams reliably,
or any recommendations for faster alternative tooling (Pandoc, etc.) that integrates well with this stack.

Thanks in advance for your time and insights!

Vu Hung