perf: replace parse5 with ultrahtml (#336)

2024-07-01 12:06:49 +01:00 · 2022-12-04 22:10:10 +00:00 · 2022-12-04 22:10:10 +00:00 · 1c0f8b0147
parent 04615e443e
commit 1c0f8b0147
8 changed files with 94 additions and 143 deletions
--- a/components/status/StatusActionsMore.vue
+++ b/components/status/StatusActionsMore.vue
@ -68,7 +68,7 @@ const deleteAndRedraft = async () => {
  }

  const { text } = await useMasto().statuses.remove(status.id)
-  openPublishDialog('dialog', getDraftFromStatus(status, text), true)
+  openPublishDialog('dialog', await getDraftFromStatus(status, text), true)
 }

 const reply = () => {
@ -81,9 +81,9 @@ const reply = () => {
  }
 }

-function editStatus() {
+async function editStatus() {
  openPublishDialog(`edit-${status.id}`, {
-    ...getDraftFromStatus(status),
+    ...await getDraftFromStatus(status),
    editingStatus: status,
  })
 }
--- a/composables/content.ts
+++ b/composables/content.ts
@ -1,48 +1,45 @@
 import type { Emoji } from 'masto'
-import type { DefaultTreeAdapterMap } from 'parse5'
-import { parseFragment, serialize } from 'parse5'
+import type { Node } from 'ultrahtml'
+import { TEXT_NODE, parse, render, walkSync } from 'ultrahtml'
 import type { VNode } from 'vue'
 import { Fragment, h, isVNode } from 'vue'
 import { RouterLink } from 'vue-router'
 import ContentCode from '~/components/content/ContentCode.vue'
 import AccountHoverWrapper from '~/components/account/AccountHoverWrapper.vue'

-type Node = DefaultTreeAdapterMap['childNode']
-type Element = DefaultTreeAdapterMap['element']
-
-function handleMention(el: Element) {
+function handleMention(el: Node) {
  // Redirect mentions to the user page
-  if (el.tagName === 'a' && el.attrs.find(i => i.name === 'class' && i.value.includes('mention'))) {
-    const href = el.attrs.find(i => i.name === 'href')
+  if (el.name === 'a' && el.attributes.class?.includes('mention')) {
+    const href = el.attributes.href
    if (href) {
-      const matchUser = href.value.match(UserLinkRE)
+      const matchUser = href.match(UserLinkRE)
      if (matchUser) {
        const [, server, username] = matchUser
        const handle = `@${username}@${server.replace(/(.+\.)(.+\..+)/, '$2')}`
-        href.value = `/${server}/@${username}`
+        el.attributes.href = `/${server}/@${username}`
        return h(AccountHoverWrapper, { handle, class: 'inline-block' }, () => nodeToVNode(el))
      }
-      const matchTag = href.value.match(TagLinkRE)
+      const matchTag = href.match(TagLinkRE)
      if (matchTag) {
        const [, , name] = matchTag
-        href.value = `/${currentServer.value}/tags/${name}`
+        el.attributes.href = `/${currentServer.value}/tags/${name}`
      }
    }
  }
  return undefined
 }

-function handleCodeBlock(el: Element) {
-  if (el.tagName === 'pre' && el.childNodes[0]?.nodeName === 'code') {
-    const codeEl = el.childNodes[0] as Element
-    const classes = codeEl.attrs.find(i => i.name === 'class')?.value
+function handleCodeBlock(el: Node) {
+  if (el.name === 'pre' && el.children[0]?.name === 'code') {
+    const codeEl = el.children[0] as Node
+    const classes = codeEl.attributes.class as string
    const lang = classes?.split(/\s/g).find(i => i.startsWith('language-'))?.replace('language-', '')
-    const code = codeEl.childNodes[0] ? treeToText(codeEl.childNodes[0]) : ''
+    const code = codeEl.children[0] ? treeToText(codeEl.children[0]) : ''
    return h(ContentCode, { lang, code: encodeURIComponent(code) })
  }
 }

-function handleNode(el: Element) {
+function handleNode(el: Node) {
  return handleCodeBlock(el) || handleMention(el) || el
 }

@ -51,7 +48,7 @@ function handleNode(el: Element) {
 * with interop of custom emojis and inline Markdown syntax
 */
 export function parseMastodonHTML(html: string, customEmojis: Record<string, Emoji> = {}) {
-  const processed = html
+  let processed = html
    // custom emojis
    .replace(/:([\w-]+?):/g, (_, name) => {
      const emoji = customEmojis[name]
@ -66,36 +63,36 @@ export function parseMastodonHTML(html: string, customEmojis: Record<string, Emo
      return `><pre><code${classes}>${code}</code></pre>`
    })

-  const tree = parseFragment(processed)
-
-  function walk(node: Node) {
-    if ('childNodes' in node)
-      node.childNodes = node.childNodes.flatMap(n => walk(n))
-
-    if (node.nodeName === '#text') {
-      // @ts-expect-error casing
-      const text = node.value as string
-      const converted = text
-        .replace(/\*\*\*(.*?)\*\*\*/g, '<b><em>$1</em></b>')
-        .replace(/\*\*(.*?)\*\*/g, '<b>$1</b>')
-        .replace(/\*(.*?)\*/g, '<em>$1</em>')
-        .replace(/~~(.*?)~~/g, '<del>$1</del>')
-        .replace(/`([^`]+?)`/g, '<code>$1</code>')
-
-      if (converted !== text)
-        return parseFragment(converted).childNodes
+  walkSync(parse(processed), (node) => {
+    if (node.type !== TEXT_NODE)
+      return
+    const replacements = [
+      [/\*\*\*(.*?)\*\*\*/g, '<b><em>$1</em></b>'],
+      [/\*\*(.*?)\*\*/g, '<b>$1</b>'],
+      [/\*(.*?)\*/g, '<em>$1</em>'],
+      [/~~(.*?)~~/g, '<del>$1</del>'],
+      [/`([^`]+?)`/g, '<code>$1</code>'],
+    ] as const
+    for (const [re, replacement] of replacements) {
+      for (const match of node.value.matchAll(re)) {
+        if (node.loc) {
+          const start = match.index! + node.loc[0].start
+          const end = start + match[0].length + node.loc[0].start
+          processed = processed.slice(0, start) + match[0].replace(re, replacement) + processed.slice(end)
+        }
+        else {
+          processed = processed.replace(match[0], match[0].replace(re, replacement))
+        }
+      }
    }
-    return [node]
-  }
+  })

-  tree.childNodes = tree.childNodes.flatMap(n => walk(n))
-
-  return tree
+  return parse(processed)
 }

-export function convertMastodonHTML(html: string, customEmojis: Record<string, Emoji> = {}) {
+export async function convertMastodonHTML(html: string, customEmojis: Record<string, Emoji> = {}) {
  const tree = parseMastodonHTML(html, customEmojis)
-  return serialize(tree)
+  return await render(tree)
 }

 /**
@ -106,31 +103,28 @@ export function contentToVNode(
  customEmojis: Record<string, Emoji> = {},
 ): VNode {
  const tree = parseMastodonHTML(content, customEmojis)
-  return h(Fragment, tree.childNodes.map(n => treeToVNode(n)))
+  return h(Fragment, (tree.children as Node[]).map(n => treeToVNode(n)))
 }

 function nodeToVNode(node: Node): VNode | string | null {
-  if (node.nodeName === '#text') {
-    // @ts-expect-error casing
-    return input.value as string
-  }
+  if (node.type === TEXT_NODE)
+    return node.value

-  if ('childNodes' in node) {
-    const attrs = Object.fromEntries(node.attrs.map(i => [i.name, i.value]))
-    if (node.nodeName === 'a' && (attrs.href?.startsWith('/') || attrs.href?.startsWith('.'))) {
-      attrs.to = attrs.href
-      delete attrs.href
-      delete attrs.target
+  if ('children' in node) {
+    if (node.name === 'a' && (node.attributes.href?.startsWith('/') || node.attributes.href?.startsWith('.'))) {
+      node.attributes.to = node.attributes.href
+      delete node.attributes.href
+      delete node.attributes.target
      return h(
        RouterLink as any,
-        attrs,
-        () => node.childNodes.map(treeToVNode),
+        node.attributes,
+        () => node.children.map(treeToVNode),
      )
    }
    return h(
-      node.nodeName,
-      attrs,
-      node.childNodes.map(treeToVNode),
+      node.name,
+      node.attributes,
+      node.children.map(treeToVNode),
    )
  }
  return null
@ -139,12 +133,10 @@ function nodeToVNode(node: Node): VNode | string | null {
 function treeToVNode(
  input: Node,
 ): VNode | string | null {
-  if (input.nodeName === '#text') {
-    // @ts-expect-error casing
+  if (input.type === TEXT_NODE)
    return input.value as string
-  }

-  if ('childNodes' in input) {
+  if ('children' in input) {
    const node = handleNode(input)
    if (node == null)
      return null
@ -156,8 +148,8 @@ function treeToVNode(
 }

 export function htmlToText(html: string) {
-  const tree = parseFragment(html)
-  return tree.childNodes.map(n => treeToText(n)).join('').trim()
+  const tree = parse(html)
+  return (tree.children as Node[]).map(n => treeToText(n)).join('').trim()
 }

 export function treeToText(input: Node): string {
@ -165,20 +157,18 @@ export function treeToText(input: Node): string {
  let body = ''
  let post = ''

-  if (input.nodeName === '#text')
-    // @ts-expect-error casing
+  if (input.type === TEXT_NODE)
    return input.value

-  if (input.nodeName === 'br')
+  if (input.name === 'br')
    return '\n'

-  if (['p', 'pre'].includes(input.nodeName))
+  if (['p', 'pre'].includes(input.name))
    pre = '\n'

-  if (input.nodeName === 'code') {
-    if (input.parentNode?.nodeName === 'pre') {
-      const clz = input.attrs.find(attr => attr.name === 'class')
-      const lang = clz?.value.replace('language-', '')
+  if (input.name === 'code') {
+    if (input.parent?.name === 'pre') {
+      const lang = input.attributes.class?.replace('language-', '')

      pre = `\`\`\`${lang || ''}\n`
      post = '\n```'
@ -188,24 +178,24 @@ export function treeToText(input: Node): string {
      post = '`'
    }
  }
-  else if (input.nodeName === 'b' || input.nodeName === 'strong') {
+  else if (input.name === 'b' || input.name === 'strong') {
    pre = '**'
    post = '**'
  }
-  else if (input.nodeName === 'i' || input.nodeName === 'em') {
+  else if (input.name === 'i' || input.name === 'em') {
    pre = '*'
    post = '*'
  }
-  else if (input.nodeName === 'del') {
+  else if (input.name === 'del') {
    pre = '~~'
    post = '~~'
  }

-  if ('childNodes' in input)
-    body = input.childNodes.map(n => treeToText(n)).join('')
+  if ('children' in input)
+    body = (input.children as Node[]).map(n => treeToText(n)).join('')

-  if (input.nodeName === 'img' && input.attrs.some(attr => attr.name === 'class' && attr.value.includes('custom-emoji')))
-    return `:${input.attrs.find(attr => attr.name === 'data-emoji-id')?.value}:`
+  if (input.name === 'img' && input.attributes.class?.includes('custom-emoji'))
+    return `:${input.attributes['data-emoji-id']}:`

  return pre + body + post
 }
--- a/composables/statusDrafts.ts
+++ b/composables/statusDrafts.ts
@ -40,9 +40,9 @@ export function getDefaultDraft(options: Partial<Draft['params'] & Omit<Draft, '
  }
 }

-export function getDraftFromStatus(status: Status, text?: null | string): Draft {
+export async function getDraftFromStatus(status: Status, text?: null | string): Promise<Draft> {
  return getDefaultDraft({
-    status: text || convertMastodonHTML(status.content),
+    status: text || await convertMastodonHTML(status.content),
    mediaIds: status.mediaAttachments.map(att => att.id),
    visibility: status.visibility,
    attachments: status.mediaAttachments,
--- a/package.json
+++ b/package.json
@ -61,7 +61,6 @@
    "lru-cache": "^7.14.1",
    "masto": "^4.7.5",
    "nuxt": "^3.0.0",
-    "parse5": "^7.1.2",
    "pinia": "^2.0.27",
    "postcss-nested": "^6.0.0",
    "prettier": "^2.8.0",
@ -74,6 +73,7 @@
    "tippy.js": "^6.3.7",
    "typescript": "^4.9.3",
    "ufo": "^1.0.1",
+    "ultrahtml": "^1.0.4",
    "unplugin-auto-import": "^0.12.0",
    "vite-plugin-inspect": "^0.7.9",
    "vitest": "^0.25.3",
--- a/pnpm-lock.yaml
+++ b/pnpm-lock.yaml
@ -44,7 +44,6 @@ specifiers:
  lru-cache: ^7.14.1
  masto: ^4.7.5
  nuxt: ^3.0.0
-  parse5: ^7.1.2
  pinia: ^2.0.27
  postcss-nested: ^6.0.0
  prettier: ^2.8.0
@ -57,6 +56,7 @@ specifiers:
  tippy.js: ^6.3.7
  typescript: ^4.9.3
  ufo: ^1.0.1
+  ultrahtml: ^1.0.4
  unplugin-auto-import: ^0.12.0
  vite-plugin-inspect: ^0.7.9
  vitest: ^0.25.3
@ -108,7 +108,6 @@ devDependencies:
  lru-cache: 7.14.1
  masto: 4.7.5
  nuxt: 3.0.0_s5ps7njkmjlaqajutnox5ntcla
-  parse5: 7.1.2
  pinia: 2.0.27_typescript@4.9.3
  postcss-nested: 6.0.0
  prettier: 2.8.0
@ -121,6 +120,7 @@ devDependencies:
  tippy.js: 6.3.7
  typescript: 4.9.3
  ufo: 1.0.1
+  ultrahtml: 1.0.4
  unplugin-auto-import: 0.12.0
  vite-plugin-inspect: 0.7.9
  vitest: 0.25.3
@ -6367,12 +6367,6 @@ packages:
      parse-path: 7.0.0
    dev: true

-  /parse5/7.1.2:
-    resolution: {integrity: sha512-Czj1WaSVpaoj0wbhMzLmWD69anp2WH7FXMB9n1Sy8/ZFF9jolSQVMu1Ij5WIyGmcBmhk7EOndpO4mIpihVqAXw==}
-    dependencies:
-      entities: 4.4.0
-    dev: true
-
  /parseurl/1.3.3:
    resolution: {integrity: sha512-CiyeOxFT/JZyN5m0z9PfXw4SCBJ6Sygz1Dpl0wqjlhDEGGBP1GnsUVEL0p63hoG1fcj3fHynXi9NYO4nWOL+qQ==}
    engines: {node: '>= 0.8'}
--- a/tests/snapshots/content-rich.test.ts.snap
+++ b/tests/snapshots/content-rich.test.ts.snap
@ -1,45 +1,25 @@
 // Vitest Snapshot v1

 exports[`content-rich > code frame 1`] = `
-"<p>Testing code block</p>
-<p></p>
-<pre lang=\\"ts\\">
-import { useMouse, usePreferredDark } from &#39;@vueuse/core&#39;
-
+"<p>Testing code block</p><p><pre lang=\\"ts\\">import { useMouse, usePreferredDark } from &#39;@vueuse/core&#39;
 // tracks mouse position
 const { x, y } = useMouse()
 // is the user prefers dark theme
-const isDark = usePreferredDark()</pre
->
-<p></p>
-"
+const isDark = usePreferredDark()</pre></p>"
 `;

 exports[`content-rich > code frame 2 1`] = `
 "<p>
-  <span class=\\"h-card\\"><a class=\\"u-url mention\\" to=\\"/@antfu@mas.to\\"></a></span>
+  <span class=\\"h-card\\"><a class=\\"u-url mention\\" to=\\"/mas.to/@antfu\\"></a></span>
  Testing<br />
+  <pre lang=\\"ts\\">const a = hello</pre>
 </p>
-<pre lang=\\"ts\\">const a = hello</pre>
-<p></p>
 "
 `;

-exports[`content-rich > code frame empty 1`] = `
-"<p></p>
-<pre></pre>
-<br />
-<p></p>
-"
-`;
+exports[`content-rich > code frame empty 1`] = `"<p><pre></pre><br></p>"`;

-exports[`content-rich > code frame no lang 1`] = `
-"<p></p>
-<pre>hello world</pre>
-<br />no lang
-<p></p>
-"
-`;
+exports[`content-rich > code frame no lang 1`] = `"<p><pre>hello world</pre><br>no lang</p>"`;

 exports[`content-rich > custom emoji 1`] = `
 "Daniel Roe
@ -61,7 +41,7 @@ exports[`content-rich > link + mention 1`] = `
    ><a
      class=\\"u-url mention\\"
      rel=\\"nofollow noopener noreferrer\\"
-      to=\\"/@vitest@mas.to\\"
+      to=\\"/mas.to/@vitest\\"
    ></a
  ></span>
  (migrated from chai+mocha)
--- a/tests/snapshots/html-parse.test.ts.snap
+++ b/tests/snapshots/html-parse.test.ts.snap
@ -1,24 +1,18 @@
 // Vitest Snapshot v1

 exports[`html-parse > code frame > html 1`] = `
-"<p>Testing code block</p>
-<p></p>
-<pre><code class=\\"language-ts\\">import { useMouse, usePreferredDark } from '@vueuse/core'
-
+"<p>Testing code block</p><p><pre><code class=\\"language-ts\\">import { useMouse, usePreferredDark } from &#39;@vueuse/core&#39;
 // tracks mouse position
 const { x, y } = useMouse()
 // is the user prefers dark theme
-const isDark = usePreferredDark()</code></pre>
-<p></p>
-"
+const isDark = usePreferredDark()</code></pre></p>"
 `;

 exports[`html-parse > code frame > text 1`] = `
 "Testing code block

 \`\`\`ts
-import { useMouse, usePreferredDark } from '@vueuse/core'
-
+import { useMouse, usePreferredDark } from &#39;@vueuse/core&#39;
 // tracks mouse position
 const { x, y } = useMouse()
 // is the user prefers dark theme
@ -34,9 +28,8 @@ exports[`html-parse > code frame 2 > html 1`] = `
    ></span
  >
  Testing<br />
+  <pre><code class=\\"language-ts\\">const a = hello</code></pre>
 </p>
-<pre><code class=\\"language-ts\\">const a = hello</code></pre>
-<p></p>
 "
 `;

@ -65,13 +58,7 @@ exports[`html-parse > empty > html 1`] = `""`;

 exports[`html-parse > empty > text 1`] = `""`;

-exports[`html-parse > inline markdown > html 1`] = `
-"<p>text <code>code</code> <b>bold</b> <em>italic</em> <del>del</del></p>
-<p></p>
-<pre><code class=\\"language-js\\">code block</code></pre>
-<p></p>
-"
-`;
+exports[`html-parse > inline markdown > html 1`] = `"<p>text <code>code</code> <b>bold</b> <em>italic</em> <del>del</del></p><p><pre><code class=\\"language-js\\">code block</code></pre></p>"`;

 exports[`html-parse > inline markdown > text 1`] = `
 "text \`code\` **bold** *italic* ~~del~~
--- a/tests/html-parse.test.ts
+++ b/tests/html-parse.test.ts
@ -1,7 +1,7 @@
 import type { Emoji } from 'masto'
 import { describe, expect, it } from 'vitest'
 import { format } from 'prettier'
-import { serialize } from 'parse5'
+import { render as renderTree } from 'ultrahtml'
 import { parseMastodonHTML, treeToText } from '~/composables/content'

 describe('html-parse', () => {
@ -53,9 +53,9 @@ describe('html-parse', () => {

 async function render(input: string, emojis?: Record<string, Emoji>) {
  const tree = parseMastodonHTML(input, emojis)
-  const html = serialize(tree)
+  const html = await renderTree(tree)
  let formatted = ''
-  const serializedText = tree.childNodes.map(n => treeToText(n)).join('').trim()
+  const serializedText = tree.children.map(n => treeToText(n)).join('').trim()

  try {
    formatted = format(html, {