src: automatically rewrite latests and docs links (#39)

Co-authored-by: Claudio W <[email protected]>

src/constants/r2Prefixes.ts

@@ -3,11 +3,23 @@
 //  later on.
 // (e.g. url path `/dist` points to R2 path `nodejs/release`)
 // See https://raw.githubusercontent.com/nodejs/build/main/ansible/www-standalone/resources/config/nodejs.org
+import map from './redirectLinks.json' assert { type: 'json' };
+
+export const REDIRECT_MAP = new Map(map as [string, string][]);
 
 export const DIST_PATH_PREFIX = 'nodejs/release';
 
 export const DOWNLOAD_PATH_PREFIX = 'nodejs';
 
 export const DOCS_PATH_PREFIX = 'nodejs/docs';
 
-export const API_PATH_PREFIX = 'nodejs/docs/latest/api';
+export const API_PATH_PREFIX = `${REDIRECT_MAP.get('nodejs/docs/latest')}/api`;
+
+export const VIRTUAL_DIRS: Record<string, Set<string>> = {
+  'docs/': new Set(
+    [...REDIRECT_MAP]
+      .filter(([key]) => key.startsWith('nodejs/docs/'))
+      .reverse()
+      .map(([key]) => key.substring(12) + '/')
+  ),
+};

src/handlers/get.ts

@@ -1,4 +1,5 @@
 import responses from '../commonResponses';
+import { VIRTUAL_DIRS } from '../constants/r2Prefixes';
 import {
   isCacheEnabled,
   isDirectoryPath,
@@ -7,7 +8,10 @@ import {
   parseUrl,
 } from '../util';
 import { Handler } from './handler';
-import { listDirectory } from './strategies/directoryListing';
+import {
+  listDirectory,
+  renderDirectoryListing,
+} from './strategies/directoryListing';
 import { getFile } from './strategies/serveFile';
 
 const getHandler: Handler = async (request, env, ctx, cache) => {
@@ -45,22 +49,28 @@ const getHandler: Handler = async (request, env, ctx, cache) => {
       return responses.FILE_NOT_FOUND(request);
     }
 
-    if (!hasTrailingSlash(bucketPath)) {
+    if (bucketPath && !hasTrailingSlash(requestUrl.pathname)) {
       // We always want to add trailing slashes to a directory URL
       requestUrl.pathname += '/';
 
       return Response.redirect(requestUrl.toString(), 301);
     }
   }
 
-  // This returns a Promise that returns either a directory listing
-  // or a file response based on the requested URL
-  const responsePromise: Promise<Response> = isPathADirectory
-    ? listDirectory(requestUrl, request, bucketPath, env)
-    : getFile(requestUrl, request, bucketPath, env);
-
-  // waits for the response to be resolved (async R2 request)
-  const response = await responsePromise;
+  let response: Response;
+  if (bucketPath in VIRTUAL_DIRS) {
+    response = renderDirectoryListing(
+      requestUrl,
+      request,
+      VIRTUAL_DIRS[bucketPath],
+      [],
+      env
+    );
+  } else if (isPathADirectory) {
+    response = await listDirectory(requestUrl, request, bucketPath, env);
+  } else {
+    response = await getFile(requestUrl, request, bucketPath, env);
+  }
 
   // Cache response if cache is enabled
   if (shouldServeCache && response.status !== 304 && response.status !== 206) {

src/handlers/strategies/directoryListing.ts

@@ -10,27 +10,23 @@ import htmlTemplate from '../../templates/directoryListing.out.js';
 // Applies the Template into a Handlebars Template Function
 const handleBarsTemplate = Handlebars.template(htmlTemplate);
 
-type DirectoryListingResponse = {
-  html: string;
-  lastModified: string;
-};
-
 /**
  * @TODO: Simplify the iteration logic or make it more readable
  *
  * Renders the html for a directory listing response
  * @param url Parsed url of the request
- * @param bucketPath Path in R2 bucket
+ * @param request Request object itself
  * @param delimitedPrefixes Directories in the bucket
  * @param listingResponse Listing response to render
  * @returns {@link DirectoryListingResponse} instance
  */
-function renderDirectoryListing(
+export function renderDirectoryListing(
   url: URL,
-  bucketPath: string,
+  request: Request,
   delimitedPrefixes: Set<string>,
-  objects: R2Object[]
-): DirectoryListingResponse {
+  objects: R2Object[],
+  env: Env
+): Response {
   // Holds all the html for each directory and file we're listing
   const tableElements = [];
 
@@ -45,9 +41,7 @@ function renderDirectoryListing(
   const urlPathname = `${url.pathname}${url.pathname.endsWith('/') ? '' : '/'}`;
 
   // Renders all the subdirectories within the Directory
-  delimitedPrefixes.forEach(directory => {
-    // R2 sends us back the absolute path of the directory, cut it
-    const name = directory.substring(bucketPath.length);
+  delimitedPrefixes.forEach(name => {
     const extra = encodeURIComponent(name.substring(0, name.length - 1));
 
     tableElements.push({
@@ -63,8 +57,7 @@ function renderDirectoryListing(
 
   // Renders all the Files within the Directory
   objects.forEach(object => {
-    // R2 sends us back the absolute path of the object, cut it
-    const name = object.key.substring(bucketPath.length);
+    const name = object.key;
 
     // Find the most recent date a file in this
     //  directory was modified, we'll use it
@@ -95,7 +88,13 @@ function renderDirectoryListing(
   // Gets an UTC-string on the ISO-8901 format of last modified date
   const lastModifiedUTC = (lastModified ?? new Date()).toUTCString();
 
-  return { html: renderedListing, lastModified: lastModifiedUTC };
+  return new Response(request.method === 'GET' ? renderedListing : null, {
+    headers: {
+      'last-modified': lastModifiedUTC,
+      'content-type': 'text/html',
+      'cache-control': env.DIRECTORY_CACHE_CONTROL || 'no-store',
+    },
+  });
 }
 
 /**
@@ -124,7 +123,10 @@ export async function listDirectory(
       cursor,
     });
 
-    result.delimitedPrefixes.forEach(prefix => delimitedPrefixes.add(prefix));
+    // R2 sends us back the absolute path of the object, cut it
+    result.delimitedPrefixes.forEach(prefix =>
+      delimitedPrefixes.add(prefix.substring(bucketPath.length))
+    );
 
     const hasIndexFile = result.objects.find(object =>
       object.key.endsWith('index.html')
@@ -134,7 +136,13 @@ export async function listDirectory(
       return getFile(url, request, `${bucketPath}index.html`, env);
     }
 
-    objects.push(...result.objects);
+    // R2 sends us back the absolute path of the object, cut it
+    result.objects.forEach(object =>
+      objects.push({
+        ...object,
+        key: object.key.substring(bucketPath.length),
+      } as R2Object)
+    );
 
     truncated = result.truncated;
     cursor = result.truncated ? result.cursor : undefined;
@@ -145,18 +153,5 @@ export async function listDirectory(
     return responses.DIRECTORY_NOT_FOUND(request);
   }
 
-  const response = renderDirectoryListing(
-    url,
-    bucketPath,
-    delimitedPrefixes,
-    objects
-  );
-
-  return new Response(request.method === 'GET' ? response.html : null, {
-    headers: {
-      'last-modified': response.lastModified,
-      'content-type': 'text/html',
-      'cache-control': env.DIRECTORY_CACHE_CONTROL || 'no-store',
-    },
-  });
+  return renderDirectoryListing(url, request, delimitedPrefixes, objects, env);
 }

src/util.ts

@@ -3,6 +3,7 @@ import {
   DIST_PATH_PREFIX,
   DOCS_PATH_PREFIX,
   DOWNLOAD_PATH_PREFIX,
+  REDIRECT_MAP,
 } from './constants/r2Prefixes';
 import { Env } from './env';
 
@@ -52,19 +53,24 @@ export function mapUrlPathToBucketPath(
   const urlToBucketPathMap: Record<string, string> = {
     dist: DIST_PATH_PREFIX + url.pathname.substring(5),
     download: DOWNLOAD_PATH_PREFIX + url.pathname.substring(9),
-    docs: DOCS_PATH_PREFIX + url.pathname.substring(5),
-    api: API_PATH_PREFIX + url.pathname.substring(4),
+    api: API_PATH_PREFIX + (url.pathname.substring(4) || '/'),
     metrics: url.pathname.substring(1), // substring to cut off the /
   };
 
   // Example: /docs/asd/123
   let bucketPath: string | undefined;
 
   const splitPath = url.pathname.split('/'); // ['', 'docs', 'asd', '123']
-
   const basePath = splitPath[1]; // 'docs'
 
-  if (basePath in urlToBucketPathMap) {
+  if (
+    REDIRECT_MAP.has(`${DOWNLOAD_PATH_PREFIX}/${splitPath[1]}/${splitPath[2]}`)
+  ) {
+    // All items in REDIRECT_MAP are three levels deep, that is asserted in tests
+    bucketPath = `${REDIRECT_MAP.get(
+      `${DOWNLOAD_PATH_PREFIX}/${splitPath[1]}/${splitPath[2]}`
+    )}/${splitPath.slice(3).join('/')}`;
+  } else if (basePath in urlToBucketPathMap) {
     bucketPath = urlToBucketPathMap[basePath];
   } else if (env.DIRECTORY_LISTING !== 'restricted') {
     bucketPath = url.pathname.substring(1);
@@ -88,7 +94,10 @@ export function mapBucketPathToUrlPath(
   if (bucketPath.startsWith(DIST_PATH_PREFIX)) {
     const path = bucketPath.substring(15);
     return [`/dist${path}`, `/download/releases${path}`];
-  } else if (bucketPath.startsWith(API_PATH_PREFIX)) {
+  } else if (
+    bucketPath.startsWith(API_PATH_PREFIX) ||
+    bucketPath.startsWith('nodejs/docs/latest/api')
+  ) {
     const path = bucketPath.substring(22);
     return [`/api${path}`, `/docs/latest/api${path}`];
   } else if (bucketPath.startsWith(DOCS_PATH_PREFIX)) {

tests/e2e/directory.test.ts

@@ -63,34 +63,6 @@ describe('Directory Tests (Restricted Directory Listing)', () => {
     assert.strictEqual(res.status, 200);
   });
 
-  it('redirects `/docs` to `/docs/`', async () => {
-    const originalRes = await mf.dispatchFetch(`${url}docs`, {
-      redirect: 'manual',
-    });
-    assert.strictEqual(originalRes.status, 301);
-    const res = await mf.dispatchFetch(originalRes.headers.get('location')!);
-    assert.strictEqual(res.status, 200);
-  });
-
-  it('allows `/docs/`', async () => {
-    const res = await mf.dispatchFetch(`${url}docs/`);
-    assert.strictEqual(res.status, 200);
-  });
-
-  it('redirects `/api` to `/api/`', async () => {
-    const originalRes = await mf.dispatchFetch(`${url}api`, {
-      redirect: 'manual',
-    });
-    assert.strictEqual(originalRes.status, 301);
-    const res = await mf.dispatchFetch(originalRes.headers.get('location')!);
-    assert.strictEqual(res.status, 200);
-  });
-
-  it('allows `/api/`', async () => {
-    const res = await mf.dispatchFetch(`${url}api/`);
-    assert.strictEqual(res.status, 200);
-  });
-
   it('redirects `/metrics` to `/metrics/`', async () => {
     const originalRes = await mf.dispatchFetch(`${url}metrics`, {
       redirect: 'manual',

tests/unit/util.test.ts

@@ -6,8 +6,20 @@ import {
   mapUrlPathToBucketPath,
   niceBytes,
 } from '../../src/util';
+import { REDIRECT_MAP } from '../../src/constants/r2Prefixes';
 
 describe('mapUrlPathToBucketPath', () => {
+  it('expects all items in REDIRECT_MAP to be pathes in the length of 3', () => {
+    // If this test breaks, the code will and we'll need to fix the code
+    REDIRECT_MAP.forEach((val, key) => {
+      assert.strictEqual(
+        key.split('/').length,
+        3,
+        `expected ${key} to be a path with 3 slashes`
+      );
+    });
+  });
+
   it('converts `/unknown-base-path` to undefined when DIRECTORY_LISTING=restricted', () => {
     const result = mapUrlPathToBucketPath(
       new URL('http://localhost/unknown-base-path'),
@@ -65,38 +77,37 @@ describe('mapUrlPathToBucketPath', () => {
     assert.strictEqual(result, 'nodejs/releases');
   });
 
-  it('converts `/docs` to `nodejs/docs`', () => {
-    const result = mapUrlPathToBucketPath(new URL('http://localhost/docs'), {
-      DIRECTORY_LISTING: 'restricted',
-    });
-    assert.strictEqual(result, 'nodejs/docs');
-  });
-
-  it('converts `/docs/latest` to `nodejs/docs/latest`', () => {
+  it('converts `/docs/latest` to `nodejs/release/v.X.X.X/docs/`', () => {
     const result = mapUrlPathToBucketPath(
       new URL('http://localhost/docs/latest'),
       {
         DIRECTORY_LISTING: 'restricted',
       }
     );
-    assert.strictEqual(result, 'nodejs/docs/latest');
+    assert.match(result ?? '', /^nodejs\/release\/v.\d+\.\d+\.\d+\/docs\/$/);
   });
 
-  it('converts `/api` to `nodejs/docs`', () => {
+  it('converts `/api` to `nodejs/release/v.X.X.X/docs/api/`', () => {
     const result = mapUrlPathToBucketPath(new URL('http://localhost/api'), {
       DIRECTORY_LISTING: 'restricted',
     });
-    assert.strictEqual(result, 'nodejs/docs/latest/api');
+    assert.match(
+      result ?? '',
+      /^nodejs\/release\/v.\d+\.\d+\.\d+\/docs\/api\/$/
+    );
   });
 
-  it('converts `/api/assert.html` to `nodejs/docs/latest/api/assert.html`', () => {
+  it('converts `/api/assert.html` to `nodejs/release/v.X.X.X/docs/api/assert.html`', () => {
     const result = mapUrlPathToBucketPath(
       new URL('http://localhost/api/assert.html'),
       {
         DIRECTORY_LISTING: 'restricted',
       }
     );
-    assert.strictEqual(result, 'nodejs/docs/latest/api/assert.html');
+    assert.match(
+      result ?? '',
+      /^nodejs\/release\/v.\d+\.\d+\.\d+\/docs\/api\/assert\.html$/
+    );
   });
 });