Strip GitHub alert markers from search index content (#59709)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: heiskr

src/search/scripts/scrape/lib/build-records-from-api.ts

@@ -35,6 +35,12 @@ import type {
   Redirects,
 } from '@/search/scripts/scrape/types'
 
+// GitHub-style alert markers (> [!NOTE], > [!TIP], etc.) that appear in
+// markdown returned by the Article API.  The rehype alerts plugin only runs
+// in the HTML pipeline, so these leak through as literal text when we index
+// the markdown-only output.  Strip them so they don't appear in search results.
+const ALERT_MARKER_REGEXP = /\[!(NOTE|TIP|WARNING|IMPORTANT|CAUTION)\]\n?/gi
+
 // Same ignored headings as the HTML scraping approach
 const IGNORED_HEADING_SLUGS = new Set(['in-this-article', 'further-reading', 'prerequisites'])
 
@@ -190,7 +196,7 @@ export function extractFromMarkdown(markdown: string): { headings: string; conte
 
   // 2. Convert full AST to plain text (code blocks are kept so that terms
   //    appearing only in code examples remain searchable).
-  const content = astToPlainText(ast)
+  const content = astToPlainText(ast).replace(ALERT_MARKER_REGEXP, '')
 
   return { headings: headings.join('\n'), content }
 }

src/search/tests/build-records-from-api.ts

@@ -212,6 +212,36 @@ More text.
     expect(text).toContain('Second paragraph in blockquote.')
   })
 
+  test('strips GitHub alert markers from plain text', () => {
+    const markdown = `> [!NOTE]
+> This is a note.
+
+> [!TIP]
+> This is a tip.
+
+> [!WARNING]
+> This is a warning.
+
+> [!IMPORTANT]
+> This is important.
+
+> [!CAUTION]
+> This is a caution.
+`
+    const text = markdownToPlainText(markdown)
+    expect(text).not.toContain('[!NOTE]')
+    expect(text).not.toContain('[!TIP]')
+    expect(text).not.toContain('[!WARNING]')
+    expect(text).not.toContain('[!IMPORTANT]')
+    expect(text).not.toContain('[!CAUTION]')
+    // The alert body text should still be present
+    expect(text).toContain('This is a note.')
+    expect(text).toContain('This is a tip.')
+    expect(text).toContain('This is a warning.')
+    expect(text).toContain('This is important.')
+    expect(text).toContain('This is a caution.')
+  })
+
   test('handles GFM tables cleanly', () => {
     const markdown = `Some intro.