Double HTML encoding in blog excerpts produces HTML artifacts

[ukusormus]

https://nixos.org/blog/ post excerpts seem to be double HTML encoded, producing visual text artifacts such as ", & and <.

Visual examples:

Note the posts themselves don't contain these artifacts, as special characters seem correctly HTML encoded once (converting " -> &quot, & -> & etc).

Example from page's source code:

[...] in supporting the Nix & NixOS ecosystem.

Double HTML encoding supposedly happens like this:

<!-- Original: -->
[...] in supporting the Nix & NixOS ecosystem.

<!-- 1st pass: -->
[...] in supporting the Nix &amp; NixOS ecosystem.
<!-- Browser renders as "Nix & NixOS" -->

<!-- 2nd pass: -->
[...] in supporting the Nix &amp;amp; NixOS ecosystem.
<!-- Browser renders as "Nix &amp; NixOS" -->

Likely culprits from code (unverified):

• first encoding pass, MarkdownIt for creating excerpts:

  nixos-homepage/core/src/lib/utils.js

  Lines 14 to 15 in aa54c73

  	export function createExcerpt(post) {
  	const parser = new MarkdownIt();

• second encoding pass, Astro template where the excerpt is included in:

  nixos-homepage/core/src/components/pages/blog/BlogListEntry.astro

  Line 26 in 192956d

  <p class="font-extralight">{createExcerpt(post.body)}...</p>