This repository generates JSON and YAML files from the WHATWG HTML Living Standard and Accessible Rich Internet Applications (WAI-ARIA) websites. It queries the Last-Modified header to find when was each source file updated.
This repository was originally a fork of Tawesoft's html5spec repository, which was archived by its owner on Nov 3, 2025.
| Page | Last Updated |
|---|---|
| Semantics, structure, and APIs of HTML documents | Fri, 25 Sep 2026 09:21:50 GMT |
| WHATWG HTML Living Standard Index Page | Fri, 25 Sep 2026 09:21:51 GMT |
| The input element | Fri, 25 Sep 2026 09:21:50 GMT |
| The HTML syntax | Fri, 25 Sep 2026 09:21:51 GMT |
| Accessible Rich Internet Applications (WAI-ARIA) | Thu, 24 Sep 2026 16:59:38 GMT |
Caution
I do not intend to keep any backwards compatibility whatsoever with the old html5spec repository. If you are relying on the old repository conventions, your best bet would be to fork it yourself. Sorry about that ¯\__(ツ)__/¯.
Also, aside from the inherent brittle nature of trying to add structure to data from non-structured data sources, the data structures for this repository's own end files are still work in progress, so do not rely on hardwired keys or data types.
If the following issues are not crossed of, they are not solved by the updates shown above and they need specific workarounds to overcome their effects.
| Location | Description | Last checked at |
|---|---|---|
controls table row |
the "Element(s)" cell is missing a ; between the video and img <code> elements. |
Fri, 25 Sep 2026 09:21:51 GMT |
This repository contains Python code that generates machine-readable JSON from the WHATWG HTML Living Standard and Accessible Rich Internet Applications (WAI-ARIA)
This is a work-in-progress and incomplete without a stable format. Contributions are very welcome. Regardless, even in this undeveloped state, this project is still a good basis for many real-world applications.
Type
maketo download and parse the spec.This repo currently contains a spec as of 2020, and only small updates are needed to support the spec as of December 2024.
Much of HTML5 depends on context and subtle semantics. Checking this may either be extremely computationally expensive, or express the intention of the author in a way that not even a human reader can validate for certain. Additionally, it is challenging to represent complex rules in a language-agnostic way.
This machine-readable specification will, therefore, always be incomplete. Sometimes the best we can practically do is provide human-readable hints in descriptions. Therefore, this machine-readable specification aims to assist a human author and catch obvious errors - but not replace the human author entirely.