feat: support text-fragments (#5140)

---------

Co-authored-by: Divyansh Singh <40380293+brc-dd@users.noreply.github.com>

__tests__/tsconfig.json

@@ -5,8 +5,9 @@
     "baseUrl": ".",
     "types": ["node", "vitest/globals"],
     "paths": {
+      "client/*": ["../src/client/*"],
       "node/*": ["../src/node/*"],
-      "client/*": ["../src/client/*"]
+      "shared/*": ["../src/shared/*"]
     }
   }
 }

__tests__/unit/node/markdown/plugins/link.test.ts

@@ -0,0 +1,51 @@
+import { slugify } from '@mdit-vue/shared'
+import { MarkdownItAsync } from 'markdown-it-async'
+import { linkPlugin } from 'node/markdown/plugins/link'
+
+describe('node/markdown/plugins/link', () => {
+  const md = new MarkdownItAsync()
+  linkPlugin(md, {}, '/', slugify)
+
+  test('preserves text-fragment hashes on markdown links', async () => {
+    const html = await md.renderAsync(
+      '[58-61](/resources/server/user#:~:text=58*,time%20authentication%20token)',
+      { cleanUrls: false }
+    )
+
+    expect(html).toContain(
+      'href="/resources/server/user.html#:~:text=58*,time%20authentication%20token"'
+    )
+  })
+
+  // https://web.dev/articles/text-fragments#mixing_element_and_text_fragments
+  test('preserves mixed element and text-fragment hashes', async () => {
+    const html = await md.renderAsync(
+      '[Section](/guide/getting-started#Hello%20World:~:text=Hello%20World)',
+      { cleanUrls: false }
+    )
+
+    expect(html).toContain(
+      'href="/guide/getting-started.html#hello-world:~:text=Hello%20World"'
+    )
+  })
+
+  test('continues to normalize regular heading hashes', async () => {
+    const html = await md.renderAsync(
+      '[Section](/guide/getting-started#Hello%20World)',
+      { cleanUrls: false }
+    )
+
+    expect(html).toContain('href="/guide/getting-started.html#hello-world"')
+  })
+
+  test('does not break encoding for text-fragments', async () => {
+    const html = await md.renderAsync(
+      '[Section](/foo?title=Cat&oldid=916388819#:~:text=Claws-,Like%20almost,the%20Felidae%2C,-cats)',
+      { cleanUrls: false }
+    )
+
+    expect(html).toContain(
+      'href="/foo.html?title=Cat&oldid=916388819#:~:text=Claws-,Like%20almost,the%20Felidae%2C,-cats"'
+    )
+  })
+})

__tests__/unit/node/markdown/plugins/snippet.test.ts

@@ -3,7 +3,6 @@ import {
   findRegion,
   rawPathToToken
 } from 'node/markdown/plugins/snippet'
-import { expect } from 'vitest'
 
 const removeEmptyKeys = <T extends Record<string, unknown>>(obj: T) => {
   return Object.fromEntries(

__tests__/unit/vitest.config.ts

@@ -12,6 +12,7 @@ export default defineConfig({
       { find: '@siteData', replacement: resolve(dir, './shims.ts') },
       { find: 'client', replacement: resolve(dir, '../../src/client') },
       { find: 'node', replacement: resolve(dir, '../../src/node') },
+      { find: 'shared', replacement: resolve(dir, '../../src/shared') },
       {
         find: /^vitepress$/,
         replacement: resolve(dir, '../../src/client/index.js')

src/client/app/router.ts

@@ -79,9 +79,21 @@ export function createRouter(
   const router: Router = {
     route,
     async go(href, options) {
+      const { hash } = new URL(href, fakeHost)
+      const hasTextFragment =
+        inBrowser && document.fragmentDirective && hash.includes(':~:')
       href = normalizeHref(href)
       if ((await router.onBeforeRouteChange?.(href)) === false) return
-      if (!inBrowser || (await changeRoute(href, options))) await loadPage(href)
+      if (
+        !inBrowser ||
+        (await changeRoute(href, { ...options, hasTextFragment }))
+      ) {
+        await loadPage(href, { initialLoad: !!options?.initialLoad })
+      }
+      if (hasTextFragment) {
+        // this will create a new history entry, but that's almost unavoidable
+        location.hash = hash
+      }
       syncRouteQueryAndHash()
       await router.onAfterRouteChange?.(href)
     }
@@ -89,7 +101,10 @@ export function createRouter(
 
   let latestPendingPath: string | null = null
 
-  async function loadPage(href: string, scrollPosition = 0, isRetry = false) {
+  async function loadPage(
+    href: string,
+    { scrollPosition = 0, isRetry = false, initialLoad = false } = {}
+  ) {
     if ((await router.onBeforePageLoad?.(href)) === false) return
 
     const targetLoc = new URL(href, fakeHost)
@@ -130,7 +145,7 @@ export function createRouter(
               history.replaceState({}, '', href)
             }
 
-            return scrollTo(targetLoc.hash, false, scrollPosition)
+            if (!initialLoad) scrollTo(targetLoc.hash, false, scrollPosition)
           })
         }
       }
@@ -149,7 +164,7 @@ export function createRouter(
         try {
           const res = await fetch(siteDataRef.value.base + 'hashmap.json')
           ;(window as any).__VP_HASH_MAP__ = await res.json()
-          await loadPage(href, scrollPosition, true)
+          await loadPage(href, { scrollPosition, isRetry: true, initialLoad })
           return
         } catch (e) {}
       }
@@ -229,7 +244,7 @@ export function createRouter(
     window.addEventListener('popstate', async (e) => {
       if (e.state === null) return
       const href = normalizeHref(location.href)
-      await loadPage(href, (e.state && e.state.scrollPosition) || 0)
+      await loadPage(href, { scrollPosition: e.state.scrollPosition || 0 })
       syncRouteQueryAndHash()
       await router.onAfterRouteChange?.(href)
     })
@@ -341,12 +356,17 @@ function normalizeHref(href: string): string {
   } else if (!url.pathname.endsWith('/') && !url.pathname.endsWith('.html')) {
     url.pathname += '.html'
   }
-  return url.pathname + url.search + url.hash
+  return url.pathname + url.search + url.hash.split(':~:')[0]
 }
 
 async function changeRoute(
   href: string,
-  { smoothScroll = false, initialLoad = false, replace = false } = {}
+  {
+    smoothScroll = false,
+    initialLoad = false,
+    replace = false,
+    hasTextFragment = false
+  } = {}
 ): Promise<boolean> {
   const loc = normalizeHref(location.href)
   const nextUrl = new URL(href, location.origin)
@@ -354,7 +374,7 @@ async function changeRoute(
 
   if (href === loc) {
     if (!initialLoad) {
-      scrollTo(nextUrl.hash, smoothScroll)
+      if (!hasTextFragment) scrollTo(nextUrl.hash, smoothScroll)
       return false
     }
   } else {
@@ -375,7 +395,7 @@ async function changeRoute(
             newURL: nextUrl.href
           })
         )
-        scrollTo(nextUrl.hash, smoothScroll)
+        if (!hasTextFragment) scrollTo(nextUrl.hash, smoothScroll)
       }
 
       return false

src/node/markdown/plugins/link.ts

@@ -35,7 +35,8 @@ export const linkPlugin = (
       token.attrGet('class') !== 'header-anchor' // header anchors are already normalized
     ) {
       const hrefAttr = token.attrs![hrefIndex]
-      const url = hrefAttr[1]
+      let [url, frag] = hrefAttr[1].split(':~:', 2)
+      hrefAttr[1] = url
       if (isExternal(url)) {
         Object.entries(externalAttrs).forEach(([key, val]) => {
           token.attrSet(key, val)
@@ -66,6 +67,9 @@ export const linkPlugin = (
           hrefAttr[1] = `${base}${hrefAttr[1]}`.replace(/\/+/g, '/')
         }
       }
+      if (frag) {
+        hrefAttr[1] += (hrefAttr[1].includes('#') ? '' : '#') + ':~:' + frag
+      }
     }
     return self.renderToken(tokens, idx, options)
   }

src/node/plugin.ts

@@ -63,8 +63,7 @@ const isPageChunk = (
     chunk.facadeModuleId.endsWith('.md')
   )
 
-const cleanUrl = (url: string): string =>
+const cleanUrl = (url: string): string => url.replace(/[?#].*$/s, '')
-  url.replace(/#.*$/s, '').replace(/\?.*$/s, '')
 
 export async function createVitePressPlugin(
   siteConfig: SiteConfig,

src/shared/shared.ts

@@ -28,7 +28,7 @@ export const APPEARANCE_KEY = 'vitepress-theme-appearance'
 export const VP_SOURCE_KEY = '[VP_SOURCE]'
 const UnpackStackView = Symbol('stack-view:unpack')
 
-const HASH_RE = /#.*$/
+const HASH_WITHOUT_FRAGMENT_RE = /#.*?(?=:~:|$)/
 const HASH_OR_QUERY_RE = /[?#].*$/
 const INDEX_OR_EXT_RE = /(?:(^|\/)index)?\.(?:md|html)$/
 
@@ -64,7 +64,7 @@ export function isActive(
     return false
   }
 
-  const hashMatch = matchPath.match(HASH_RE)
+  const hashMatch = matchPath.match(HASH_WITHOUT_FRAGMENT_RE)
 
   if (hashMatch) {
     return (inBrowser ? location.hash : '') === hashMatch[0]