fix(mdx-loader): improve mdxJsxTextElementToHtml (#9262)

packages/docusaurus-mdx-loader/src/remark/toc/__tests__/__fixtures__/non-text-content.md

@@ -7,3 +7,5 @@
 ## <i>HTML</i>
 
 ## `inline.code()`
+
+## some <span className="some-class" style={{border: "solid"}}>styled</span> <strong>heading</strong> <span class="myClass" className="myClassName <> weird char" data-random-attr="456"/> test

packages/docusaurus-mdx-loader/src/remark/toc/__tests__/__snapshots__/index.test.ts.snap

@@ -171,6 +171,11 @@ exports[`toc remark plugin works on non text phrasing content 1`] = `
 		value: '<code>inline.code()</code>',
 		id: 'inlinecode',
 		level: 2
+	},
+	{
+		value: 'some <span class="some-class">styled</span> <strong>heading</strong> <span class="myClassName &lt;&gt; weird char"></span> test',
+		id: 'some-styled-heading--test',
+		level: 2
 	}
 ]
 
@@ -183,6 +188,8 @@ exports[`toc remark plugin works on non text phrasing content 1`] = `
 ## <i>HTML</i>
 
 ## \`inline.code()\`
+
+## some <span className="some-class" style={{border: "solid"}}>styled</span> <strong>heading</strong> <span class="myClass" className="myClassName <> weird char" data-random-attr="456" /> test
 "
 `;
 

packages/docusaurus-mdx-loader/src/remark/utils/index.ts

@@ -8,26 +8,57 @@
 import escapeHtml from 'escape-html';
 import type {Parent} from 'unist';
 import type {PhrasingContent, Heading} from 'mdast';
-// @ts-expect-error: TODO see https://github.com/microsoft/TypeScript/issues/49721
+import type {
-import type {MdxJsxAttributeValueExpression} from 'mdast-util-mdx';
+  MdxJsxAttribute,
+  MdxJsxAttributeValueExpression,
+  MdxJsxTextElement,
+  // @ts-expect-error: TODO see https://github.com/microsoft/TypeScript/issues/49721
+} from 'mdast-util-mdx';
 
 export function stringifyContent(
   node: Parent,
-  toString: (param: unknown) => string, // TODO weird but works): string {
+  toString: (param: unknown) => string, // TODO weird but works
 ): string {
   return (node.children as PhrasingContent[])
     .map((item) => toValue(item, toString))
     .join('');
 }
 
+// TODO This is really a workaround, and not super reliable
+// For now we only support serializing tagName, className and content
+// Can we implement the TOC with real JSX nodes instead of html strings later?
+function mdxJsxTextElementToHtml(
+  element: MdxJsxTextElement,
+  toString: (param: unknown) => string, // TODO weird but works
+): string {
+  const tag = element.name;
+
+  const attributes = element.attributes.filter(
+    (child): child is MdxJsxAttribute => child.type === 'mdxJsxAttribute',
+  );
+
+  const classAttribute =
+    attributes.find((attr) => attr.name === 'className') ??
+    attributes.find((attr) => attr.name === 'class');
+
+  const classAttributeString = classAttribute
+    ? `class="${escapeHtml(String(classAttribute.value))}"`
+    : ``;
+
+  const allAttributes = classAttributeString ? ` ${classAttributeString}` : '';
+
+  const content = stringifyContent(element, toString);
+
+  return `<${tag}${allAttributes}>${content}</${tag}>`;
+}
+
 export function toValue(
   node: PhrasingContent | Heading,
   toString: (param: unknown) => string, // TODO weird but works
 ): string {
   switch (node.type) {
     case 'mdxJsxTextElement': {
-      const tag = node.name;
+      return mdxJsxTextElementToHtml(node as MdxJsxTextElement, toString);
-      return `<${tag}>${stringifyContent(node, toString)}</${tag}>`;
     }
     case 'text':
       return escapeHtml(node.value);