[Docs Site] Decode HTML entities and strip Markdown from page descriptions (#19350)

Author: KianNH

package-lock.json

@@ -88,11 +88,13 @@
 		"rehype-external-links": "3.0.0",
 		"rehype-mermaid": "3.0.0",
 		"rehype-title-figure": "0.1.2",
+		"remark": "15.0.1",
 		"sharp": "0.33.5",
 		"solarflare-theme": "0.0.2",
 		"starlight-image-zoom": "0.9.0",
 		"starlight-links-validator": "0.14.1",
 		"starlight-package-managers": "0.9.0",
+		"strip-markdown": "6.0.0",
 		"svgo": "3.3.2",
 		"tailwindcss": "3.4.17",
 		"tippy.js": "6.3.7",

package.json

@@ -139,7 +139,7 @@ if (currentSection) {
 	}
 }
 
-Astro.props.entry.data.description ??= await getPageDescription(
+Astro.props.entry.data.description = await getPageDescription(
 	// @ts-expect-error TODO: improve types
 	Astro.props.entry,
 	Astro.locals,

src/components/overrides/Head.astro

@@ -8,7 +8,7 @@ sidebar:
 head:
   - tag: title
     content: Cloudflare Workers — Rust language support
-description: Write Workers in 100% Rust using the [`workers-rs`crate](https://github.com/cloudflare/workers-rs)
+description: Write Workers in 100% Rust using the [`workers-rs` crate](https://github.com/cloudflare/workers-rs)
 ---
 
 Cloudflare Workers provides support for Rust via the [`workers-rs` crate](https://github.com/cloudflare/workers-rs), which makes [Runtime APIs](/workers/runtime-apis) and [bindings](/workers/runtime-apis/bindings/) to developer platform products, such as [Workers KV](/kv/concepts/how-kv-works/), [R2](/r2/), and [Queues](/queues/), available directly from your Rust code.

src/content/docs/workers/languages/rust/index.mdx

@@ -1,25 +1,37 @@
 import type { CollectionEntry } from "astro:content";
 import { parse } from "node-html-parser";
 import { entryToString } from "./container";
-/*
-    1. If there is a `description` property in the frontmatter, return that.
-    2. If there is a `<p>...</p>` element in the HTML, return that.
-    3. Return `undefined` to signal to consumers there is no suitable description.
-*/
+import { remark } from "remark";
+import strip from "strip-markdown";
+import he from "he";
+
+/**
+ * Generates a plain-text description for use in the `description` and `og:description` meta tags.
+ *
+ * 1. If there is a `description` property in the frontmatter, strip any Markdown tokens and return.
+ * 2. If there is a `<p>...</p>` element in the HTML, decode any HTML entities and return that.
+ * 3. Return `undefined` to signal to consumers there is no suitable description.
+ */
 export async function getPageDescription(
 	entry: CollectionEntry<"docs">,
 	locals: any,
 ) {
-	if (entry.data.description) return entry.data.description;
+	let description = undefined;
+
+	if (entry.data.description) {
+		const file = await remark().use(strip).process(entry.data.description);
 
-	const html = await entryToString(entry, locals);
+		description = file.toString();
+	} else {
+		const html = await entryToString(entry, locals);
 
-	if (!html) return undefined;
+		if (!html) return undefined;
 
-	const dom = parse(html);
-	const description = dom.querySelector(":root > p");
+		const dom = parse(html);
+		const paragraph = dom.querySelector(":root > p");
 
-	if (description) return description.innerText;
+		if (paragraph) description = he.decode(paragraph.innerText);
+	}
 
-	return undefined;
+	return description?.replaceAll(" ↗", "").trim();
 }