perf(readability): cache html serialization per htmlDom (#823)

Author: Kikobeats

packages/metascraper-readability/src/index.js

@@ -44,9 +44,19 @@ const readability = asyncMemoizeOne(async (url, html, readabilityOpts) => {
   }
 }, memoizeOne.EqualityFirstArgument)
 
+const htmlCache = new WeakMap()
+
+const getHtml = htmlDom => {
+  if (!htmlCache.has(htmlDom)) {
+    htmlCache.set(htmlDom, htmlDom.html())
+  }
+
+  return htmlCache.get(htmlDom)
+}
+
 module.exports = ({ readabilityOpts } = {}) => {
   const getReadbility = composeRule(($, url) =>
-    readability(url, $.html(), readabilityOpts)
+    readability(url, getHtml($), readabilityOpts)
   )
 
   const rules = {

packages/metascraper-readability/test/index.js

@@ -3,6 +3,7 @@
 const test = require('ava')
 const path = require('path')
 const fs = require('fs')
+const { load } = require('cheerio')
 
 const metascraper = require('metascraper')([
   require('metascraper-readability')()
@@ -59,3 +60,23 @@ test('chowhanandsons.com', async t => {
   const metadata = await metascraper({ html, url })
   t.snapshot(metadata)
 })
+
+test('serializes html once per invocation', async t => {
+  const url = 'https://microlink.io'
+  const html = fs.readFileSync(
+    path.resolve(__dirname, 'fixtures/microlink.io.html'),
+    'utf-8'
+  )
+
+  const $ = load(html, { baseURI: url })
+  const originalHtml = $.html.bind($)
+  let htmlCalls = 0
+
+  $.html = (...args) => {
+    if (args.length === 0) htmlCalls++
+    return originalHtml(...args)
+  }
+
+  await metascraper({ htmlDom: $, url })
+  t.is(htmlCalls, 1)
+})