Fix autolinking of URLs inside links in Markdown (#3564)

.changeset/afraid-stingrays-sell.md

@@ -0,0 +1,5 @@
+---
+'@astrojs/markdown-remark': patch
+---
+
+Fix autolinking of URLs inside links

packages/markdown/remark/src/index.ts

@@ -99,14 +99,13 @@ export async function renderMarkdown(
 	parser
 		.use(isMDX ? [rehypeJsx, rehypeExpressions] : [rehypeRaw])
 		.use(rehypeEscape)
-		.use(rehypeIslands);
+		.use(rehypeIslands)
+		.use([rehypeCollectHeaders])
+		.use(rehypeStringify, { allowDangerousHtml: true })
 
 	let result: string;
 	try {
-		const vfile = await parser
-			.use([rehypeCollectHeaders])
-			.use(rehypeStringify, { allowDangerousHtml: true })
-			.process(input);
+		const vfile = await parser.process(input);
 		result = vfile.toString();
 	} catch (err) {
 		// Ensure that the error message contains the input filename

packages/markdown/remark/src/rehype-jsx.ts

@@ -1,15 +1,14 @@
+import type { RehypePlugin } from './types.js';
 import { visit } from 'unist-util-visit';
 
 const MDX_ELEMENTS = ['mdxJsxFlowElement', 'mdxJsxTextElement'];
-export default function rehypeJsx(): any {
-	return function (node: any): any {
-		visit(node, 'element', (child: any) => {
-			child.tagName = `${child.tagName}`;
-		});
-		visit(node, MDX_ELEMENTS, (child: any, index: number | null, parent: any) => {
+
+export default function rehypeJsx(): ReturnType<RehypePlugin> {
+	return function (tree) {
+		visit(tree, MDX_ELEMENTS, (node: any, index: number | null, parent: any) => {
 			if (index === null || !Boolean(parent)) return;
 
-			const attrs = child.attributes.reduce((acc: any[], entry: any) => {
+			const attrs = node.attributes.reduce((acc: any[], entry: any) => {
 				let attr = entry.value;
 				if (attr && typeof attr === 'object') {
 					attr = `{${attr.value}}`;
@@ -26,23 +25,42 @@ export default function rehypeJsx(): any {
 				return acc + ` ${entry.name}${attr ? '=' : ''}${attr}`;
 			}, '');
 
-			if (child.children.length === 0) {
-				child.type = 'raw';
-				child.value = `<${child.name}${attrs} />`;
+			if (node.children.length === 0) {
+				node.type = 'raw';
+				node.value = `<${node.name}${attrs} />`;
 				return;
 			}
 
-			// Replace the current child node with its children
+			// If the current node is a JSX <a> element, remove autolinks from its children
+			// to prevent Markdown code like `<a href="/">**Go to www.example.com now!**</a>`
+			// from creating a nested link to `www.example.com`
+			if (node.name === 'a') {
+				visit(node, 'element', (el, elIndex, elParent) => {
+					const isAutolink = (
+						el.tagName === 'a' &&
+						el.children.length === 1 &&
+						el.children[0].type === 'text' &&
+						el.children[0].value.match(/^(https?:\/\/|www\.)/i)
+					);
+
+					// If we found an autolink, remove it by replacing it with its text-only child
+					if (isAutolink) {
+						elParent.children.splice(elIndex, 1, el.children[0]);
+					}
+				});
+			}
+
+			// Replace the current node with its children
 			// wrapped by raw opening and closing tags
 			const openingTag = {
 				type: 'raw',
-				value: `\n<${child.name}${attrs}>`,
+				value: `\n<${node.name}${attrs}>`,
 			};
 			const closingTag = {
 				type: 'raw',
-				value: `</${child.name}>\n`,
+				value: `</${node.name}>\n`,
 			};
-			parent.children.splice(index, 1, openingTag, ...child.children, closingTag);
+			parent.children.splice(index, 1, openingTag, ...node.children, closingTag);
 		});
 	};
 }

packages/markdown/remark/test/autolinking.test.js

@@ -0,0 +1,96 @@
+import { renderMarkdown } from '../dist/index.js';
+import chai from 'chai';
+
+describe('autolinking', () => {
+	it('autolinks URLs starting with a protocol in plain text', async () => {
+		const { code } = await renderMarkdown(
+			`See https://example.com for more.`,
+			{}
+		);
+
+		chai
+			.expect(code.replace(/\n/g, ''))
+			.to.equal(`<p>See <a href="https://example.com">https://example.com</a> for more.</p>`);
+	});
+
+	it('autolinks URLs starting with "www." in plain text', async () => {
+		const { code } = await renderMarkdown(
+			`See www.example.com for more.`,
+			{}
+		);
+
+		chai
+			.expect(code.trim())
+			.to.equal(`<p>See <a href="http://www.example.com">www.example.com</a> for more.</p>`);
+	});
+
+	it('does not autolink URLs in code blocks', async () => {
+		const { code } = await renderMarkdown(
+			'See `https://example.com` or `www.example.com` for more.',
+			{}
+		);
+
+		chai
+			.expect(code.trim())
+			.to.equal(`<p>See <code is:raw>https://example.com</code> or ` +
+				`<code is:raw>www.example.com</code> for more.</p>`);
+	});
+
+	it('does not autolink URLs in fenced code blocks', async () => {
+		const { code } = await renderMarkdown(
+			'Example:\n```\nGo to https://example.com or www.example.com now.\n```',
+			{}
+		);
+
+		chai
+			.expect(code)
+			.to.contain(`<pre is:raw`)
+			.to.contain(`Go to https://example.com or www.example.com now.`);
+	});
+
+	it('does not autolink URLs starting with a protocol when nested inside links', async () => {
+		const { code } = await renderMarkdown(
+			`See [http://example.com](http://example.com) or ` +
+			`<a test href="https://example.com">https://example.com</a>`,
+			{}
+		);
+
+		chai
+			.expect(code.replace(/\n/g, ''))
+			.to.equal(
+				`<p>See <a href="http://example.com">http://example.com</a> or ` +
+				`<a test href="https://example.com">https://example.com</a></p>`
+			);
+	});
+
+	it('does not autolink URLs starting with "www." when nested inside links', async () => {
+		const { code } = await renderMarkdown(
+			`See [www.example.com](https://www.example.com) or ` +
+			`<a test href="https://www.example.com">www.example.com</a>`,
+			{}
+		);
+
+		chai
+			.expect(code.replace(/\n/g, ''))
+			.to.equal(
+				`<p>See <a href="https://www.example.com">www.example.com</a> or ` +
+				`<a test href="https://www.example.com">www.example.com</a></p>`
+			);
+	});
+
+	it('does not autolink URLs when nested several layers deep inside links', async () => {
+		const { code } = await renderMarkdown(
+			`<a href="https://www.example.com">**Visit _our www.example.com or ` +
+			`http://localhost pages_ for more!**</a>`,
+			{}
+		);
+
+		chai
+			.expect(code.replace(/\n/g, ''))
+			.to.equal(
+				`<a href="https://www.example.com"><strong>` +
+				`Visit <em>our www.example.com or http://localhost pages</em> for more!` +
+				`</strong></a>`
+			);
+	});
+});