Repeated setContent is 50x slower without goto('about:blank') in between than with it

[phiresky]

I'm using puppeteer just to extract some information from html, setting the content using page.setContent(). This is extremely slow, even though I never navigate to an actual website.

Steps to reproduce

• Puppeteer version: 1.11.0
• Platform / OS version: Ubuntu 18.04
• Node.js version: v11.3.0

Minimal example:

import chrome, { Page } from "puppeteer"

const content = `
<div>
    ${Array(1000)
      .fill("<p>test</p>")
      .join("\n")}
</div>`

const count = 100

async function testNoBlank(tab: Page) {
  for (let i = 0; i < count; i++) {
    await tab.setContent(content)
    const ret = await tab.$$eval("p", ps => ps.map(p => p.textContent))
    console.assert(ret.length === 1000)
  }
}
async function testBlank(tab: Page) {
  for (let i = 0; i < count; i++) {
    await tab.setContent(content)
    const ret = await tab.$$eval("p", ps => ps.map(p => p.textContent))
    console.assert(ret.length === 1000)
    await tab.goto("about:blank")
  }
}

async function both() {
  const browser = await chrome.launch({ headless: true })
  const [tab] = await browser.pages()
  console.time("blank")
  await testBlank(tab)
  console.timeEnd("blank")
  await tab.goto("about:blank")
  console.time("no blank")
  await testNoBlank(tab)
  console.timeEnd("no blank")

  await browser.close()
}

both()

Output:

blank: 3568.199ms
no blank: 191955.917ms
Done in 196.83s.

Without goto("about:blank") in between setContent calls it is 50x! slower.

I found this by accident. If this can't be fixed, it should at least be mentioned in the docs, since setContent seems to be the canonical method to load html from a string.

[aslushnikov]

Can reproduce with this:

const puppeteer = require('puppeteer');

(async () => {
  const browser = await puppeteer.launch({ headless: true })
  const [tab] = await browser.pages()

  console.time('time');
  for (let i = 0; i < 20; ++i)
    await tab.setContent('<div></div>');
  console.timeEnd('time');
  await browser.close()
})();

[devnyxie]

Hey! I have exactly the same issue, can't provide the code since it's not open-sourced tho but it's almost the same as in example above.
I pass the Page to one function, and if page is not redirected to 'about:blank' first and I use .setContent() - it will cause an infinite loading and reach default 30000ms timeout.