Add routes for /article, /article/body, and /article/meta (#54652)

Author: hectorsector

src/article-api/middleware/article.ts

@@ -0,0 +1,106 @@
+import type { RequestHandler, Response } from 'express'
+import express from 'express'
+
+import { defaultCacheControl } from '@/frame/middleware/cache-control.js'
+import { Context } from '#src/types.js'
+import catchMiddlewareError from '@/observability/middleware/catch-middleware-error.js'
+import { ExtendedRequestWithPageInfo } from '../types'
+import { pageValidationMiddleware, pathValidationMiddleware } from './validation'
+import contextualize from '#src/frame/middleware/context/context.js'
+
+/** START helper functions */
+
+// for now, we're just querying pageinfo, we'll likely replace /api/pageinfo
+// with /api/meta and move or reference that code here
+async function getArticleMetadata(req: ExtendedRequestWithPageInfo) {
+  const queryString = new URLSearchParams(req.query as Record<string, string>).toString()
+  const apiUrl = `${req.protocol}://${req.get('host')}/api/pageinfo${queryString ? `?${queryString}` : ''}`
+
+  // Fetch the data from the pageinfo API
+  const response = await fetch(apiUrl)
+
+  // Check if the response is OK
+  if (!response.ok) {
+    const errorText = await response.text()
+    throw new Error(`Failed to fetch metadata: ${response.status} ${errorText}`)
+  }
+
+  return await response.json()
+}
+
+async function getArticleBody(req: ExtendedRequestWithPageInfo) {
+  // req.pageinfo is set from pageValidationMiddleware and pathValidationMiddleware
+  // and is in the ExtendedRequestWithPageInfo
+  const { page, pathname } = req.pageinfo
+
+  // these parts allow us to render the page
+  const mockedContext: Context = {}
+  const renderingReq = {
+    path: pathname,
+    language: page.languageCode,
+    pagePath: pathname,
+    cookies: {},
+    context: mockedContext,
+    headers: {
+      'content-type': 'text/markdown',
+    },
+  }
+
+  // contextualize and render the page
+  await contextualize(renderingReq as ExtendedRequestWithPageInfo, {} as Response, () => {})
+  renderingReq.context.page = page
+  renderingReq.context.markdownRequested = true
+  return await page.render(renderingReq.context)
+}
+
+/** END helper functions */
+
+/** START routes */
+const router = express.Router()
+
+// For all these routes:
+// - pathValidationMiddleware ensures the path is properly structured and handles errors when it's not
+// - pageValidationMiddleware fetches the page from the pagelist, returns 404 to the user if not found
+
+router.get(
+  '/',
+  pathValidationMiddleware as RequestHandler,
+  pageValidationMiddleware as RequestHandler,
+  catchMiddlewareError(async function (req: ExtendedRequestWithPageInfo, res: Response) {
+    // First, fetch metadata
+    const metaData = await getArticleMetadata(req)
+    const bodyContent = await getArticleBody(req)
+
+    defaultCacheControl(res)
+    return res.json({
+      meta: metaData,
+      body: bodyContent,
+    })
+  }),
+)
+
+router.get(
+  '/body',
+  pathValidationMiddleware as RequestHandler,
+  pageValidationMiddleware as RequestHandler,
+  catchMiddlewareError(async function (req: ExtendedRequestWithPageInfo, res: Response) {
+    const rendered = await getArticleBody(req)
+    defaultCacheControl(res)
+    return res.type('text/markdown').send(rendered)
+  }),
+)
+
+router.get(
+  '/meta',
+  pathValidationMiddleware as RequestHandler,
+  pageValidationMiddleware as RequestHandler,
+  catchMiddlewareError(async function (req: ExtendedRequestWithPageInfo, res: Response) {
+    const metaData = await getArticleMetadata(req)
+    defaultCacheControl(res)
+    return res.json(metaData)
+  }),
+)
+
+/** END routes */
+
+export default router

src/article-api/middleware/validation.ts

@@ -2,6 +2,7 @@ import { ExtendedRequestWithPageInfo } from '../types'
 import type { NextFunction, Response } from 'express'
 import { isArchivedVersionByPath } from '@/archives/lib/is-archived-version'
 import getRedirect from '@/redirects/lib/get-redirect.js'
+import { Page } from '#src/types.js'
 
 export const pathValidationMiddleware = (
   req: ExtendedRequestWithPageInfo,
@@ -24,7 +25,9 @@ export const pathValidationMiddleware = (
   if (/\s/.test(pathname)) {
     return res.status(400).json({ error: `'pathname' cannot contain whitespace` })
   }
-  req.pageinfo = { pathname }
+
+  // req.pageinfo.page will be defined later or it will throw
+  req.pageinfo = { pathname, page: {} as Page }
   return next()
 }
 
@@ -59,6 +62,9 @@ export const pageValidationMiddleware = (
     pathname = `/${req.context.currentLanguage}`
   }
 
+  // Initialize archived property to avoid it being undefined
+  req.pageinfo.archived = { isArchived: false }
+
   if (!(pathname in req.context.pages)) {
     // If a pathname is not a known page, it might *either* be a redirect,
     // or an archived enterprise version, or both.
@@ -76,6 +82,9 @@ export const pageValidationMiddleware = (
 
   // Remember this might yield undefined if the pathname is not a page
   req.pageinfo.page = req.context.pages[pathname]
+  if (!req.pageinfo.page && !req.pageinfo.archived.isArchived) {
+    return res.status(404).json({ error: `No page found for '${pathname}'` })
+  }
   // The pathname might have changed if it was a redirect
   req.pageinfo.pathname = pathname
 

src/article-api/tests/pageinfo.js

@@ -50,7 +50,7 @@ describe('pageinfo api', () => {
 
   test('a pathname that does not exist', async () => {
     const res = await get(makeURL('/en/never/heard/of'))
-    expect(res.statusCode).toBe(400)
+    expect(res.statusCode).toBe(404)
     const { error } = JSON.parse(res.body)
     expect(error).toBe("No page found for '/en/never/heard/of'")
   })

src/article-api/types.ts

@@ -8,7 +8,7 @@ export type ArchivedVersion = {
 export type ExtendedRequestWithPageInfo = ExtendedRequest & {
   pageinfo: {
     pathname: string
-    page?: Page
+    page: Page
     archived?: ArchivedVersion
   }
 }

src/frame/middleware/api.ts

@@ -5,8 +5,9 @@ import events from '@/events/middleware.js'
 import anchorRedirect from '@/rest/api/anchor-redirect.js'
 import aiSearch from '@/search/middleware/ai-search'
 import search from '@/search/middleware/search-routes.js'
-import pageInfo from '#src/article-api/middleware/pageinfo.ts'
+import pageInfo from '@/article-api/middleware/pageinfo'
 import pageList from '@/article-api/middleware/pagelist'
+import article from '@/article-api/middleware/article'
 import webhooks from '@/webhooks/middleware/webhooks.js'
 import { ExtendedRequest } from '@/types'
 import { noCacheControl } from './cache-control'
@@ -30,6 +31,7 @@ router.use('/webhooks', createAPIRateLimiter(internalRoutesRateLimit), webhooks)
 router.use('/anchor-redirect', createAPIRateLimiter(internalRoutesRateLimit), anchorRedirect)
 router.use('/pageinfo', createAPIRateLimiter(3), pageInfo)
 router.use('/pagelist', createAPIRateLimiter(publicRoutesRateLimit), pageList)
+router.use('/article', createAPIRateLimiter(publicRoutesRateLimit), article)
 
 // The purpose of this is for convenience to everyone who runs this code
 // base locally but don't have an Elasticsearch server locally.

src/frame/middleware/context/context.ts

@@ -39,8 +39,6 @@ export default async function contextualize(
   req.context.process = { env: {} }
 
   if (req.pagePath && req.pagePath.endsWith('.md')) {
-    req.context.markdownRequested = true
-
     // req.pagePath is used later in the rendering pipeline to
     // locate the file in the tree so it cannot have .md
     req.pagePath = req.pagePath.replace(/\/index\.md$/, '').replace(/\.md$/, '')

src/shielding/middleware/handle-invalid-paths.ts

@@ -89,7 +89,16 @@ export default function handleInvalidPaths(
     // E.g. `/en/foo.md?bar=baz`
     const newUrl = req.originalUrl.replace(req.path, req.path.replace(/\/index\.md$/, ''))
     return res.redirect(newUrl)
+  } else if (req.path.endsWith('.md')) {
+    // encode the query params but also make them pretty so we can see
+    // them as `/` and `@` in the address bar
+    // e.g. /api/article/body?pathname=/en/[email protected]/admin...
+    // NOT: /api/article/body?pathname=%2Fen%2Fenterprise-server%403.16%2Fadmin...
+    const encodedPath = encodeURIComponent(req.path.replace(/\.md$/, ''))
+      .replace(/%2F/g, '/')
+      .replace(/%40/g, '@')
+    const newUrl = `/api/article/body?pathname=${encodedPath}`
+    return res.redirect(newUrl)
   }
-
   return next()
 }