fix: extract HTMLBlocks before parsing in stripComments

__tests__/lib/stripComments.test.ts

@@ -382,6 +382,75 @@ end"`);
     expect(output).not.toContain('<!-- bottom comment -->');
   });
 
+  describe('HTMLBlock handling in mdxish mode', () => {
+    it('does not crash on multiline HTMLBlock content', async () => {
+      const input = `<HTMLBlock>{\`
+<div>hello</div>
+\`}</HTMLBlock>`;
+
+      const output = await stripComments(input, { mdxish: true });
+      expect(output).toContain('<HTMLBlock>');
+      expect(output).toContain('<div>hello</div>');
+    });
+
+    it('preserves HTMLBlock with attributes', async () => {
+      const input = `<HTMLBlock id="test">{\`
+<p>content</p>
+\`}</HTMLBlock>`;
+
+      const output = await stripComments(input, { mdxish: true });
+      expect(output).toContain('<HTMLBlock');
+      expect(output).toContain('<p>content</p>');
+    });
+
+    it('strips comments outside HTMLBlocks while preserving the block', async () => {
+      const input = `<!-- top comment -->
+
+<HTMLBlock>{\`
+<div>hello</div>
+\`}</HTMLBlock>
+
+<!-- bottom comment -->`;
+
+      const output = await stripComments(input, { mdxish: true });
+      expect(output).toContain('<HTMLBlock>');
+      expect(output).not.toContain('<!-- top comment -->');
+      expect(output).not.toContain('<!-- bottom comment -->');
+    });
+
+    it('strips HTML comments inside HTMLBlock content', async () => {
+      const input = `<HTMLBlock>{\`
+<!-- authored HTML comment -->
+<div>hello</div>
+\`}</HTMLBlock>`;
+
+      const output = await stripComments(input, { mdxish: true });
+      expect(output).not.toContain('<!-- authored HTML comment -->');
+      expect(output).toContain('<div>hello</div>');
+    });
+
+    it('strips all comments including inside HTMLBlocks', async () => {
+      const input = `<div>
+<!-- strip me -->
+<HTMLBlock>{\`<!-- also strip me --><p>hi</p>\`}</HTMLBlock>
+</div>`;
+
+      const output = await stripComments(input, { mdxish: true });
+      expect(output).not.toContain('<!-- also strip me -->');
+      expect(output).not.toContain('<!-- strip me -->');
+      expect(output).toContain('<HTMLBlock>');
+    });
+
+    it('handles nested HTMLBlocks', async () => {
+      const input = `<HTMLBlock>{\`
+<HTMLBlock>{\`<div>nested</div>\`}</HTMLBlock>
+\`}</HTMLBlock>`;
+
+      const output = await stripComments(input, { mdxish: true });
+      expect(output).toContain('<HTMLBlock>');
+    });
+  });
+
   describe('strip comments edge cases', () => {
     it.each([
       ['should return empty for empty string', '', undefined, ''],

lib/mdast-util/html-block-component/index.ts

@@ -0,0 +1,53 @@
+import type { Html } from 'mdast';
+import type { CompileContext, Extension as FromMarkdownExtension, Handle, Token } from 'mdast-util-from-markdown';
+
+const contextMap = new WeakMap<Token, { chunks: string[]; lastEndLine: number }>();
+
+function findHtmlBlockComponentToken(this: CompileContext): Token | undefined {
+  const events = this.tokenStack;
+  for (let i = events.length - 1; i >= 0; i -= 1) {
+    if (events[i][0].type === 'htmlBlockComponent') return events[i][0];
+  }
+  return undefined;
+}
+
+function enterHtmlBlockComponent(this: CompileContext, token: Parameters<Handle>[0]): void {
+  contextMap.set(token, { chunks: [], lastEndLine: token.start.line });
+  this.enter({ type: 'html', value: '' } as Html, token);
+}
+
+function exitHtmlBlockComponentData(this: CompileContext, token: Parameters<Handle>[0]): void {
+  const componentToken = findHtmlBlockComponentToken.call(this);
+  if (!componentToken) return;
+  const ctx = contextMap.get(componentToken);
+  if (ctx) {
+    const gap = token.start.line - ctx.lastEndLine;
+    if (ctx.chunks.length > 0 && gap > 0) {
+      ctx.chunks.push('\n'.repeat(gap));
+    }
+    ctx.chunks.push(this.sliceSerialize(token));
+    ctx.lastEndLine = token.end.line;
+  }
+}
+
+function exitHtmlBlockComponent(this: CompileContext, token: Parameters<Handle>[0]): void {
+  const ctx = contextMap.get(token);
+  const node = this.stack[this.stack.length - 1] as Html;
+  if (ctx) {
+    node.value = ctx.chunks.join('');
+    contextMap.delete(token);
+  }
+  this.exit(token);
+}
+
+export function htmlBlockComponentFromMarkdown(): FromMarkdownExtension {
+  return {
+    enter: {
+      htmlBlockComponent: enterHtmlBlockComponent,
+    },
+    exit: {
+      htmlBlockComponentData: exitHtmlBlockComponentData,
+      htmlBlockComponent: exitHtmlBlockComponent,
+    },
+  };
+}

lib/micromark/html-block-component/index.ts

@@ -0,0 +1 @@
+export { htmlBlockComponent } from './syntax';