Remove redirects from markdown pages (#52355)

Co-authored-by: Kevin Heis <[email protected]>

Author: hectorsector

package-lock.json

@@ -313,6 +313,7 @@
     "remark-parse": "^11.0.0",
     "remark-rehype": "^11.1.0",
     "remark-remove-comments": "^1.0.1",
+    "remark-stringify": "^11.0.0",
     "rss-parser": "^3.13.0",
     "scroll-anchoring": "^0.1.0",
     "semver": "^7.6.2",

package.json

@@ -1,5 +1,5 @@
 import { renderLiquid } from './liquid/index.js'
-import { renderUnified } from './unified/index.js'
+import { renderMarkdown, renderUnified } from './unified/index.js'
 import { engine } from './liquid/engine.js'
 
 const globalCache = new Map()
@@ -26,6 +26,12 @@ export async function renderContent(template = '', context = {}, options = {}) {
   }
   try {
     template = await renderLiquid(template, context)
+    if (context.markdownRequested) {
+      const md = await renderMarkdown(template, context, options)
+
+      return md
+    }
+
     const html = await renderUnified(template, context, options)
     if (cacheKey) {
       globalCache.set(cacheKey, html)

src/content-render/index.js

@@ -1,5 +1,5 @@
 import { fastTextOnly } from './text-only.js'
-import { createProcessor } from './processor.js'
+import { createProcessor, createMarkdownOnlyProcessor } from './processor.js'
 
 export async function renderUnified(template, context, options) {
   const processor = createProcessor(context)
@@ -12,3 +12,11 @@ export async function renderUnified(template, context, options) {
 
   return html.trim()
 }
+
+export async function renderMarkdown(template, context, options) {
+  const processor = createMarkdownOnlyProcessor(context)
+  const vFile = await processor.process(template)
+  let markdown = vFile.toString()
+
+  return markdown.trim()
+}

src/content-render/unified/index.js

@@ -29,6 +29,7 @@ import annotate from './annotate.js'
 import alerts from './alerts.js'
 import replaceDomain from './replace-domain.js'
 import removeHtmlComments from 'remark-remove-comments'
+import remarkStringify from 'remark-stringify'
 
 export function createProcessor(context) {
   return (
@@ -79,6 +80,10 @@ export function createProcessor(context) {
   )
 }
 
+export function createMarkdownOnlyProcessor(context) {
+  return unified().use(remarkParse).use(gfm).use(remarkStringify)
+}
+
 export function createMinimalProcessor(context) {
   return unified()
     .use(remarkParse)

src/content-render/unified/processor.js

@@ -36,7 +36,7 @@ describe('permission statements', () => {
   })
 
   test('page with permission frontmatter and product statement', async () => {
-    const $ = await getDOM('/get-started/foo/page-with-permissions-and-product-callout.md')
+    const $ = await getDOM('/get-started/foo/page-with-permissions-and-product-callout')
     const html = $('[data-testid=permissions-callout] div').html()
     // part of the UI
     expect(html).toMatch('Who can use this feature')

src/fixtures/tests/permissions-callout.js

@@ -38,6 +38,14 @@ 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$/, '').replace(/\.md$/, '')
+  }
+
   // define each context property explicitly for code-search friendliness
   // e.g. searches for "req.context.page" will include results from this file
   req.context.currentLanguage = req.language

src/frame/middleware/context/context.ts

@@ -190,6 +190,15 @@ export default async function renderPage(req: ExtendedRequest, res: Response) {
     }
   }
 
+  if (context.markdownRequested) {
+    if (!page.autogenerated && page.documentType === 'article') {
+      return res.type('text/markdown').send(req.context.renderedPage)
+    } else {
+      const newUrl = req.originalUrl.replace(req.path, req.path.replace(/\.md$/, ''))
+      return res.redirect(newUrl)
+    }
+  }
+
   defaultCacheControl(res)
 
   return nextHandleRequest(req, res)

src/frame/middleware/render-page.ts

@@ -83,14 +83,11 @@ export default function handleInvalidPaths(
     return res.status(404).send('Not found')
   }
 
-  if (req.path.endsWith('/index.md') || req.path.endsWith('.md')) {
+  if (req.path.endsWith('/index.md')) {
     defaultCacheControl(res)
     // The originalUrl is the full URL including query string.
     // E.g. `/en/foo.md?bar=baz`
-    const newUrl = req.originalUrl.replace(
-      req.path,
-      req.path.replace(/\/index\.md$/, '').replace(/\.md$/, ''),
-    )
+    const newUrl = req.originalUrl.replace(req.path, req.path.replace(/\/index\.md$/, ''))
     return res.redirect(newUrl)
   }
 

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

@@ -71,7 +71,8 @@ describe('index.md and .md suffixes', () => {
     }
   })
 
-  test('any URL that ends with /.md redirects', async () => {
+  // TODO-ARTICLEAPI: unskip tests or replace when ready to ship article API
+  test.skip('any URL that ends with /.md redirects', async () => {
     // With language prefix
     {
       const res = await get('/en/get-started/hello.md')