fix(utils): handle '#' inside heading text when creating excerpt

packages/docusaurus-utils/src/__tests__/markdownUtils.test.ts

@@ -47,6 +47,38 @@ describe('createExcerpt', () => {
     );
   });
 
+  it('skips H1 heading whose text contains a hash character (e.g. "C#", "F#")', () => {
+    expect(
+      createExcerpt(dedent`
+
+          # C# Programming Guide
+
+          This paragraph should become the description.
+        `),
+    ).toBe('This paragraph should become the description.');
+  });
+
+  it('skips H1 heading with trailing closing hashes and interior hash', () => {
+    expect(
+      createExcerpt(dedent`
+
+          # F# Programming Guide #
+
+          This paragraph should become the description.
+        `),
+    ).toBe('This paragraph should become the description.');
+  });
+
+  it('keeps hash characters inside non-heading text', () => {
+    expect(
+      createExcerpt(dedent`
+          The language C# is used with .NET, and F# is a functional alternative.
+
+          Nunc porttitor libero nec vulputate venenatis.
+        `),
+    ).toBe('The language C# is used with .NET, and F# is a functional alternative.');
+  });
+
   it('creates excerpt for regular content with alternate title', () => {
     expect(
       createExcerpt(dedent`

packages/docusaurus-utils/src/markdownUtils.ts

@@ -145,10 +145,15 @@ export function createExcerpt(fileString: string): string | undefined {
     const cleanedLine = fileLine
       // Remove HTML tags.
       .replace(/<[^>]*>/g, '')
-      // Remove Title headers
-      .replace(/^#[^#]+#?/gm, '')
+      // Remove Title headers (single-# ATX headings are skipped entirely)
+      // Note: the heading text may legitimately contain "#" (e.g. "C#", "F#"),
+      // so we must match the whole line instead of using a negated class
+      .replace(/^#.*$/gm, '')
       // Remove Markdown + ATX-style headers
-      .replace(/^#{1,6}\s*(?<text>[^#]*?)\s*#{0,6}/gm, '$1')
+      // Note: we must not forbid "#" inside the heading text (e.g. "C#", "F#")
+      // Only the leading marker and an optional trailing closing sequence are
+      // stripped, anchored to the end of the line to avoid eating interior "#"
+      .replace(/^#{1,6}\s*(?<text>.*?)\s*#*$/gm, '$1')
       // Remove emphasis.
       .replace(/(?<opening>[*_]{1,3})(?<text>.*?)\1/g, '$2')
       // Remove strikethroughs.