fix: Links in blog posts rendered in a feed (rss/atom/json) should be absolute

Author: 政宇 廖

packages/docusaurus-plugin-content-blog/src/feed.ts

@@ -9,7 +9,11 @@ import path from 'path';
 import fs from 'fs-extra';
 import logger from '@docusaurus/logger';
 import {Feed, type Author as FeedAuthor} from 'feed';
-import {normalizeUrl, readOutputHTMLFile} from '@docusaurus/utils';
+import {
+  isValidRelativePath,
+  normalizeUrl,
+  readOutputHTMLFile,
+} from '@docusaurus/utils';
 import {blogPostContainerID} from '@docusaurus/utils-common';
 import {load as cheerioLoad} from 'cheerio';
 import type {DocusaurusConfig} from '@docusaurus/types';
@@ -105,6 +109,20 @@ async function defaultCreateFeedItems({
       );
       const $ = cheerioLoad(content);
 
+      // Prefix the relative paths with url defined in config
+      // leave the anchors and external links untouched
+      // (e.g. href that starts with "#", "http://" or "https://")
+      $(`div#${blogPostContainerID} a`).each((_, elm) => {
+        const {href} = elm.attribs;
+        if (href) {
+          if (isValidRelativePath(href)) {
+            // a valid relative url path might not have leading slash
+            elm.attribs.href =
+              siteConfig.url + href.startsWith('/') ? href : `/${href}`;
+          }
+        }
+      });
+
       const link = normalizeUrl([siteUrl, permalink]);
       const feedItem: BlogFeedItem = {
         title: metadataTitle,

packages/docusaurus-utils/src/__tests__/urlUtils.test.ts

@@ -18,6 +18,7 @@ import {
   buildSshUrl,
   buildHttpsUrl,
   hasSSHProtocol,
+  isValidRelativePath,
 } from '../urlUtils';
 
 describe('normalizeUrl', () => {
@@ -312,3 +313,18 @@ describe('hasSSHProtocol', () => {
     expect(hasSSHProtocol(url)).toBe(false);
   });
 });
+
+describe('isValidRelativePath', () => {
+  it('works', () => {
+    expect(isValidRelativePath('https://docusaurus.io/blog')).toBe(false);
+    expect(isValidRelativePath('notValidhttps://')).toBe(false);
+    expect(isValidRelativePath('#hash')).toBe(false);
+    expect(isValidRelativePath('/#hash')).toBe(true);
+    expect(isValidRelativePath('/blog')).toBe(true);
+    expect(isValidRelativePath('blog/my-first-blog')).toBe(true);
+    expect(isValidRelativePath('/blog///valid//')).toBe(true);
+    expect(isValidRelativePath('/blog/hi#你好')).toBe(true);
+    expect(isValidRelativePath('')).toBe(true);
+    expect(isValidRelativePath('/blog?qs=ho')).toBe(true);
+  });
+});

packages/docusaurus-utils/src/index.ts

@@ -54,6 +54,7 @@ export {
   hasSSHProtocol,
   buildHttpsUrl,
   buildSshUrl,
+  isValidRelativePath,
 } from './urlUtils';
 export {
   type Tag,

packages/docusaurus-utils/src/urlUtils.ts

@@ -232,3 +232,26 @@ export function hasSSHProtocol(sourceRepoUrl: string): boolean {
     return /^(?:[\w-]+@)?[\w.-]+:[\w./-]+/.test(sourceRepoUrl);
   }
 }
+
+/**
+ * Whether the str is a valid relative url path
+ */
+export function isValidRelativePath(str: string): boolean {
+  // When str starts with "#", the parsedPath will be "/#xxx"
+  // and the functions will return true while this is not a relative path
+  // so an early return is required here
+  if (str.startsWith('#')) {
+    return false;
+  }
+  try {
+    const urlObj = new URL(str, 'https://domain.com');
+    const parsedPath = `${urlObj.pathname}${urlObj.hash}`;
+    const strToCompare = (str.startsWith('/') ? str : `/${str}`).split('?')[0];
+    return (
+      parsedPath === strToCompare ||
+      parsedPath === encodeURI(strToCompare ?? '')
+    );
+  } catch {
+    return false;
+  }
+}