[AI Search] Fix bug with #extractProductFromDocsUrl (#55447)

Author: ashishkeshan

src/search/components/helpers/ai-search-links-json.ts

@@ -67,7 +67,11 @@ function extractMarkdownLinks(markdownResponse: string) {
 
 // Given a Docs URL, extract the product name
 function extractProductFromDocsUrl(url: string): string {
-  const pathname = new URL(url).pathname
+  const urlObject = new URL(url)
+  if (urlObject.hostname !== 'docs.github.com') {
+    return ''
+  }
+  const pathname = urlObject.pathname
 
   const segments = pathname.split('/').filter((segment) => segment)
 
@@ -78,7 +82,14 @@ function extractProductFromDocsUrl(url: string): string {
   }
 
   if (segments[0].length === 2) {
-    return segments[1] || ''
+    if (segments.length < 2) {
+      return ''
+    }
+    // if second segment is a version, then product is the third segment
+    if (segments[1].includes('@')) {
+      return segments[2] || ''
+    }
+    return segments[1]
   }
 
   return segments[0]

src/search/tests/ai-search-links-json.ts

@@ -0,0 +1,142 @@
+import { expect, test, describe } from 'vitest'
+import { generateAISearchLinksJson } from '@/search/components/helpers/ai-search-links-json'
+
+describe('generateAISearchLinksJson', () => {
+  test('should return empty array JSON for no links', () => {
+    const sources: { url: string }[] = []
+    const aiResponse = 'This is a response with no links.'
+    const result = generateAISearchLinksJson(sources, aiResponse)
+    expect(JSON.parse(result)).toEqual([])
+  })
+
+  test('should handle only reference links', () => {
+    const sources = [
+      { url: 'https://docs.github.com/en/codespaces/overview' },
+      { url: 'https://docs.github.com/en/actions/learn-github-actions' },
+    ]
+    const aiResponse = 'Response text.'
+    const result = generateAISearchLinksJson(sources, aiResponse)
+    expect(JSON.parse(result)).toEqual([
+      {
+        type: 'reference',
+        url: 'https://docs.github.com/en/codespaces/overview',
+        product: 'codespaces',
+      },
+      {
+        type: 'reference',
+        url: 'https://docs.github.com/en/actions/learn-github-actions',
+        product: 'actions',
+      },
+    ])
+  })
+
+  test('should handle only inline links', () => {
+    const sources: { url: string }[] = []
+    const aiResponse =
+      'Check out [Codespaces](https://docs.github.com/en/codespaces/overview) and [Actions](https://docs.github.com/en/actions/learn-github-actions).'
+    const result = generateAISearchLinksJson(sources, aiResponse)
+    expect(JSON.parse(result)).toEqual([
+      {
+        type: 'inline',
+        url: 'https://docs.github.com/en/codespaces/overview',
+        product: 'codespaces',
+      },
+      {
+        type: 'inline',
+        url: 'https://docs.github.com/en/actions/learn-github-actions',
+        product: 'actions',
+      },
+    ])
+  })
+
+  test('should handle both reference and inline links', () => {
+    const sources = [{ url: 'https://docs.github.com/en/billing/managing-billing' }]
+    const aiResponse = 'Learn about [Billing](https://docs.github.com/en/billing/managing-billing).'
+    const result = generateAISearchLinksJson(sources, aiResponse)
+    // Note: The inline link appears first because it's processed first
+    expect(JSON.parse(result)).toEqual([
+      {
+        type: 'inline',
+        url: 'https://docs.github.com/en/billing/managing-billing',
+        product: 'billing',
+      },
+      {
+        type: 'reference',
+        url: 'https://docs.github.com/en/billing/managing-billing',
+        product: 'billing',
+      },
+    ])
+  })
+
+  test('should handle versioned URLs correctly', () => {
+    const sources = [
+      { url: 'https://docs.github.com/en/enterprise-cloud@latest/codespaces/overview' },
+    ]
+    const aiResponse =
+      'See [Codespaces for Enterprise](https://docs.github.com/en/[email protected]/codespaces/overview).'
+    const result = generateAISearchLinksJson(sources, aiResponse)
+    expect(JSON.parse(result)).toEqual([
+      {
+        type: 'inline',
+        url: 'https://docs.github.com/en/[email protected]/codespaces/overview',
+        product: 'codespaces',
+      },
+      {
+        type: 'reference',
+        url: 'https://docs.github.com/en/enterprise-cloud@latest/codespaces/overview',
+        product: 'codespaces',
+      },
+    ])
+  })
+
+  test('should handle non-docs URLs with empty product', () => {
+    const sources = [{ url: 'https://github.com/features/actions' }]
+    const aiResponse = 'Visit [GitHub](https://github.com/).'
+    const result = generateAISearchLinksJson(sources, aiResponse)
+    expect(JSON.parse(result)).toEqual([
+      { type: 'inline', url: 'https://github.com/', product: '' }, // Non-docs inline link
+      { type: 'reference', url: 'https://github.com/features/actions', product: '' }, // Non-docs reference link
+    ])
+  })
+
+  test('should ignore invalid URLs in markdown', () => {
+    const sources = [{ url: 'https://docs.github.com/en/copilot/overview' }]
+    const aiResponse =
+      'See [Copilot](https://docs.github.com/en/copilot/overview) and [invalid](not-a-valid-url).'
+    const result = generateAISearchLinksJson(sources, aiResponse)
+    expect(JSON.parse(result)).toEqual([
+      { type: 'inline', url: 'https://docs.github.com/en/copilot/overview', product: 'copilot' },
+      { type: 'reference', url: 'https://docs.github.com/en/copilot/overview', product: 'copilot' },
+    ])
+  })
+
+  test('should handle URLs without language code', () => {
+    const sources: { url: string }[] = []
+    const aiResponse = 'Link: [Copilot](https://docs.github.com/copilot/overview)'
+    const result = generateAISearchLinksJson(sources, aiResponse)
+    expect(JSON.parse(result)).toEqual([
+      { type: 'inline', url: 'https://docs.github.com/copilot/overview', product: 'copilot' },
+    ])
+  })
+
+  test('should handle URLs with only language code', () => {
+    const sources = [{ url: 'https://docs.github.com/en' }]
+    const aiResponse = 'Visit [English Docs](https://docs.github.com/en).'
+    const result = generateAISearchLinksJson(sources, aiResponse)
+    expect(JSON.parse(result)).toEqual([
+      { type: 'inline', url: 'https://docs.github.com/en', product: '' },
+      { type: 'reference', url: 'https://docs.github.com/en', product: '' },
+    ])
+  })
+
+  test('should handle URLs with language code and version but no product', () => {
+    const sources = [{ url: 'https://docs.github.com/en/[email protected]' }]
+    const aiResponse =
+      'Visit [Enterprise Server Docs](https://docs.github.com/en/[email protected]).'
+    const result = generateAISearchLinksJson(sources, aiResponse)
+    expect(JSON.parse(result)).toEqual([
+      { type: 'inline', url: 'https://docs.github.com/en/[email protected]', product: '' },
+      { type: 'reference', url: 'https://docs.github.com/en/[email protected]', product: '' },
+    ])
+  })
+})