Add redirectedFrom metadata to article API responses (#60536)

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

Author: heiskr

src/article-api/README.md

@@ -24,7 +24,11 @@ The `/api/article` endpoints return information about a page by `pathname`.
 
 `api/article/meta` is highly cached, in JSON format.
 
-### Autogenerated Content Transformers
+### Redirects and the pagelist
+
+The pagelist (`/api/pagelist/:lang/:version`) returns only **canonical permalinks**. The article API (`/api/article`, `/api/article/body`, `/api/article/meta`) transparently follows redirects—so URLs that don't appear in the pagelist (such as `redirect_from` aliases or old paths) may still return content.
+
+When the article API resolves a redirect through the redirect table, the JSON response includes a `redirectedFrom` field containing the normalized pathname that was looked up (after trailing-slash removal and other standard normalization, not the raw originally-requested pathname). This field is only set for redirect-table lookups; it is not set for the bare `/` to `/<lang>` language rewrite. This lets consumers detect that the URL they requested is not canonical. The `/api/article/body` endpoint returns plain text, so `redirectedFrom` is not included in its response.
 
 For autogenerated pages (REST, GraphQL, webhooks, landing pages, audit logs, etc), the Article API uses specialized transformers to convert the rendered content into markdown format. These transformers are located in `src/article-api/transformers/` and use an extensible architecture.
 

src/article-api/middleware/article-pageinfo.ts

@@ -134,7 +134,7 @@ export async function getMetadata(req: ExtendedRequestWithPageInfo) {
   // /articles or '/en/enterprise-server@latest/foo/bar)
   // So by the time we get here, the pathname should be one of the
   // page's valid permalinks.
-  const { page, pathname, archived } = req.pageinfo
+  const { page, pathname, archived, redirectedFrom } = req.pageinfo
   const documentType = page?.documentType ?? null
 
   if (archived && archived.isArchived) {
@@ -157,5 +157,8 @@ export async function getMetadata(req: ExtendedRequestWithPageInfo) {
   const fromCache = await getPageInfoFromCache(page, pathname)
   const { cacheInfo, ...meta } = fromCache
 
-  return { meta: { ...meta, documentType }, cacheInfo }
+  return {
+    meta: { ...meta, documentType, ...(redirectedFrom && { redirectedFrom }) },
+    cacheInfo,
+  }
 }

src/article-api/middleware/validation.ts

@@ -109,6 +109,7 @@ export const pageValidationMiddleware = (
     if (!req.pageinfo.archived.isArchived) {
       const redirect = getRedirect(pathname, redirectsContext)
       if (redirect) {
+        req.pageinfo.redirectedFrom = pathname
         pathname = redirect
       }
     }

src/article-api/tests/pageinfo.ts

@@ -12,6 +12,7 @@ interface PageMetadata {
   title: string
   intro: string
   documentType: string | null
+  redirectedFrom?: string
 }
 
 interface ErrorResponse {
@@ -46,6 +47,8 @@ describe('pageinfo api', () => {
       'Get started using HubGit to manage Git repositories and collaborate with others.',
     )
     expect(meta.documentType).toBe('category')
+    // Canonical URLs should not have redirectedFrom
+    expect(meta.redirectedFrom).toBeUndefined()
     // Check that it can be cached at the CDN
     expect(res.headers['set-cookie']).toBeUndefined()
     expect(res.headers['cache-control']).toContain('public')
@@ -90,13 +93,15 @@ describe('pageinfo api', () => {
       expect(res.statusCode).toBe(200)
       const meta = JSON.parse(res.body) as PageMetadata
       expect(meta.title).toBe('HubGit.com Fixture Documentation')
+      expect(meta.redirectedFrom).toBe('/en/olden-days')
     }
     // Trailing slashes are always removed
     {
       const res = await get(makeURL('/en/olden-days/'))
       expect(res.statusCode).toBe(200)
       const meta = JSON.parse(res.body) as PageMetadata
       expect(meta.title).toBe('HubGit.com Fixture Documentation')
+      expect(meta.redirectedFrom).toBe('/en/olden-days')
     }
     // Short code for latest version
     {

src/article-api/types.ts

@@ -10,5 +10,6 @@ export type ExtendedRequestWithPageInfo = ExtendedRequest & {
     pathname: string
     page: Page
     archived?: ArchivedVersion
+    redirectedFrom?: string
   }
 }