chore(helpers): run loadIframe in worker_threads

package.json

@@ -174,8 +174,7 @@
     "postcss": "latest",
     "postcss-focus": "latest",
     "simple-git-hooks": "latest",
-    "standard": "latest",
-    "standard-markdown": "latest"
+    "standard": "latest"
   },
   "engines": {
     "node": ">= 16"
@@ -189,7 +188,7 @@
     "coverage": "c8 report --reporter=text-lcov > coverage/lcov.info",
     "dev": "concurrently \"gulp\" \"npm run dev:server\"",
     "dev:server": "browser-sync start --server --files \"index.html, README.md, static/**/*.(css|js)\"",
-    "lint": "standard-markdown README.md && standard",
+    "lint": "standard",
     "pretest": "npm run lint",
     "release": "lerna publish --yes --sort --conventional-commits -m \"chore(release): %s\" --create-release github",
     "test": "c8 pnpm --recursive test",
@@ -213,9 +212,6 @@
       "prettier-standard",
       "standard --fix"
     ],
-    "*.md": [
-      "standard-markdown"
-    ],
     "package.json": [
       "finepack"
     ]

packages/metascraper-helpers/package.json

@@ -3,7 +3,7 @@
   "description": "Collection of helper functions used by metascraper",
   "homepage": "https://github.com/microlinkhq/metascraper/packages/metascraper-helpers",
   "version": "5.46.11",
-  "main": "index.js",
+  "main": "src/index.js",
   "author": {
     "email": "[email protected]",
     "name": "microlink.io",

packages/metascraper-helpers/index.js → packages/metascraper-helpers/src/index.js

@@ -5,7 +5,6 @@ const debug = require('debug-logfmt')('metascraper:find-rule')
 const condenseWhitespace = require('condense-whitespace')
 const { getExtension: mimeExtension } = require('mime')
 const capitalize = require('microsoft-capitalize')
-const { JSDOM, VirtualConsole } = require('jsdom')
 const isRelativeUrl = require('is-relative-url')
 const fileExtension = require('file-extension')
 const _normalizeUrl = require('normalize-url')
@@ -466,40 +465,6 @@ const composeRule =
 const has = value =>
   value !== undefined && !Number.isNaN(value) && hasValues(value)
 
-const loadIframe = (url, $, { timeout = 5000 } = {}) =>
-  new Promise(resolve => {
-    const dom = new JSDOM($.html(), {
-      url,
-      virtualConsole: new VirtualConsole(),
-      runScripts: 'dangerously',
-      resources: 'usable'
-    })
-
-    const done = (html = '') => resolve($.load(html))
-
-    const listen = (element, method, fn) =>
-      element[`${method}EventListener`]('load', fn, {
-        capture: true,
-        once: true,
-        passive: true
-      })
-
-    const iframe = dom.window.document.querySelector('iframe')
-    if (!iframe) return done()
-
-    const timer = setTimeout(() => {
-      listen(iframe, 'remove', load)
-      done()
-    }, timeout)
-
-    function load () {
-      clearTimeout(timer)
-      done(iframe.contentDocument.documentElement.outerHTML)
-    }
-
-    listen(iframe, 'add', load)
-  })
-
 const getUrls = input => String(input).match(urlRegexForMatch) ?? []
 
 module.exports = {
@@ -536,7 +501,7 @@ module.exports = {
   iso6393,
   jsonld,
   lang,
-  loadIframe,
+  loadIframe: require('./load-iframe'),
   logo,
   memoizeOne,
   mimeExtension,

packages/metascraper-helpers/src/load-iframe/index.js

@@ -0,0 +1,18 @@
+'use strict'
+
+const { Worker } = require('worker_threads')
+const path = require('path')
+
+const SCRIPT_PATH = path.resolve(__dirname, 'worker.js')
+
+module.exports = (url, $, { timeout = 5000 } = {}) => {
+  const worker = new Worker(SCRIPT_PATH, {
+    workerData: { url, html: $.html(), timeout },
+    stdout: true,
+    stderr: true
+  })
+  const { promise, resolve, reject } = Promise.withResolvers()
+  worker.on('message', html => resolve($.load(html || '')))
+  worker.on('error', reject)
+  return promise
+}

packages/metascraper-helpers/src/load-iframe/worker.js

@@ -0,0 +1,37 @@
+'use strict'
+
+const { workerData, parentPort } = require('node:worker_threads')
+const { JSDOM, VirtualConsole } = require('jsdom')
+
+async function main ({ url, html, timeout }) {
+  const dom = new JSDOM(html, {
+    url,
+    virtualConsole: new VirtualConsole(),
+    runScripts: 'dangerously',
+    resources: 'usable'
+  })
+
+  const iframe = dom.window.document.querySelector('iframe')
+  if (!iframe) return
+
+  let timeoutId
+
+  const waitForIframe = new Promise(resolve => {
+    iframe.addEventListener(
+      'load',
+      () => {
+        clearTimeout(timeoutId)
+        resolve(iframe.contentDocument.documentElement.outerHTML)
+      },
+      { once: true }
+    )
+  })
+
+  const timeoutReached = new Promise(
+    resolve => (timeoutId = setTimeout(resolve, timeout))
+  )
+
+  return Promise.race([waitForIframe, timeoutReached])
+}
+
+main(workerData).then(html => parentPort.postMessage(html))

packages/metascraper-helpers/test/index.js

@@ -29,7 +29,7 @@ const {
   parseUrl,
   url,
   video
-} = require('..')
+} = require('../src')
 
 const measure = fn => {
   const time = process.hrtime()

packages/metascraper-helpers/test/load-iframe.js

@@ -22,3 +22,14 @@ test('wait `load` event', async t => {
   const $iframe = await loadIframe(url, $)
   t.true($iframe.html().includes('twitter:player'))
 })
+
+test('markup is correct', async t => {
+  const url =
+    'https://saas.transistor.fm/episodes/paul-jarvis-gaining-freedom-by-building-an-indie-business'
+  const src = 'https://share.transistor.fm/e/e83b42d0'
+  const $ = await loadIframe(
+    url,
+    cheerio.load(`<iframe src="${src}"></iframe>`)
+  )
+  t.snapshot($.html())
+})