@browserless/goto/src/adblock.js

Navigate to web pages with built-in ad blocking, device emulation, and optimized loading for faster automation.

'use strict'

const { PuppeteerBlocker } = require('@ghostery/adblocker-puppeteer')
const path = require('path')
const fs = require('fs')

const debug = require('debug-logfmt')('browserless:goto:adblock')

const engine = PuppeteerBlocker.deserialize(
  new Uint8Array(fs.readFileSync(path.resolve(__dirname, './engine.bin')))
)

engine.on('request-blocked', ({ url }) => debug('block', url))
engine.on('request-redirected', ({ url }) => debug('redirect', url))

/**
 * autoconsent.playwright.js is the only browser-injectable IIFE bundle in the package.
 * It is not in the package's "exports" map, so pin @duckduckgo/autoconsent with ~ to
 * avoid breakage from internal restructuring on minor/patch bumps.
 */
const autoconsentPlaywrightScript = fs.readFileSync(
  path.resolve(
    path.dirname(require.resolve('@duckduckgo/autoconsent')),
    'autoconsent.playwright.js'
  ),
  'utf8'
)

/* Configuration passed to autoconsent's `initResp` message.
   See https://github.com/duckduckgo/autoconsent/blob/main/api.md */
const autoconsentConfig = Object.freeze({
  /* activate consent rule matching */
  enabled: true,
  /* automatically reject (opt-out) all cookies */
  autoAction: 'optOut',
  /* hide banners early via CSS before detection finishes */
  enablePrehide: true,
  /* apply CSS-only rules that hide popups lacking a reject button */
  enableCosmeticRules: true,
  /* skip bundled ABP/uBO cosmetic filter list (saves bundle size) */
  enableFilterList: false,
  /* how many times to retry CMP detection (~50 ms apart) */
  detectRetries: 20,
  logs: {
    /* CMP detection / opt-out lifecycle events */
    lifecycle: false,
    /* individual rule step execution */
    rulesteps: false,
    /* eval snippet calls */
    evals: false,
    /* rule errors */
    errors: false,
    /* background ↔ content-script messages */
    messages: false
  }
})

const sendMessage = (page, message) =>
  page
    .evaluate(msg => {
      if (window.autoconsentReceiveMessage) {
        return window.autoconsentReceiveMessage(msg)
      }
    }, message)
    .catch(() => {})

const setupAutoConsent = async page => {
  if (page._autoconsentSetup) return

  await page.exposeFunction('autoconsentSendMessage', async message => {
    if (!message || typeof message !== 'object') return

    if (message.type === 'init') {
      return sendMessage(page, { type: 'initResp', config: autoconsentConfig })
    }

    if (message.type === 'eval') {
      return sendMessage(page, { type: 'evalResp', id: message.id, result: false })
    }
  })

  await page.evaluateOnNewDocument(autoconsentPlaywrightScript)
  page._autoconsentSetup = true
}

const runAutoConsent = page => page.evaluate(autoconsentPlaywrightScript)

const enableBlockingInPage = (page, run, actionTimeout) => {
  page.disableAdblock = () =>
    engine
      .disableBlockingInPage(page, { keepRequestInterception: true })
      .then(() => debug('disabled'))
      .catch(() => {})

  return [
    run({
      fn: setupAutoConsent(page),
      timeout: actionTimeout,
      debug: 'autoconsent:setup'
    }),
    run({
      fn: engine.enableBlockingInPage(page),
      timeout: actionTimeout,
      debug: 'adblock'
    })
  ]
}

module.exports = { enableBlockingInPage, runAutoConsent }