Support exceptions (allowlist/whitelist) in adblocker

packages/puppeteer-extra-plugin-adblocker/src/index.test.ts

@@ -38,3 +38,35 @@ test('will block ads', async t => {
 
   await browser.close()
 })
+
+test("will adhere to allowlist rules", async t => {
+  const puppeteer = require('puppeteer-extra')
+  const adblockerPlugin = AdblockerPlugin({
+    allowlist: [
+      // Allow everything
+      /(.*?)/,
+    ]
+  })
+
+  puppeteer.use(adblockerPlugin)
+
+  const browser = await puppeteer.launch({
+    args: PUPPETEER_ARGS,
+    headless: true
+  })
+
+  const blocker = await adblockerPlugin.getBlocker()
+
+  const page = await browser.newPage()
+
+  let blockedRequests = 0
+  blocker.on('request-blocked', (req) => {
+    blockedRequests += 1
+  })
+
+  await page.goto("https://www.google.com/search?q=rent%20a%20car", { waitUntil: 'networkidle0' })
+
+  t.is(blockedRequests, 0)
+
+  await browser.close();
+})

packages/puppeteer-extra-plugin-adblocker/src/index.ts

@@ -2,7 +2,7 @@ import { promises as fs } from 'fs'
 import os from 'os'
 import path from 'path'
 
-import { PuppeteerBlocker } from '@cliqz/adblocker-puppeteer'
+import { NetworkFilter, PuppeteerBlocker } from '@cliqz/adblocker-puppeteer'
 import fetch from 'node-fetch'
 import { PuppeteerExtraPlugin } from 'puppeteer-extra-plugin'
 
@@ -11,6 +11,8 @@ const engineCacheFilename = `${pkg.name}-${pkg.version}-engine.bin`
 
 /** Available plugin options */
 export interface PluginOptions {
+  /** Optional custom list of sites to allow. Default: [] */
+  allowlist: RegExp[]
   /** Whether or not to block trackers (in addition to ads). Default: false */
   blockTrackers: boolean
   /** Whether or not to block trackers and other annoyances, including cookie
@@ -41,6 +43,7 @@ export class PuppeteerExtraPluginAdblocker extends PuppeteerExtraPlugin {
 
   get defaults(): PluginOptions {
     return {
+      allowlist: [],
       blockTrackers: false,
       blockTrackersAndAnnoyances: false,
       useCache: true,
@@ -67,6 +70,18 @@ export class PuppeteerExtraPluginAdblocker extends PuppeteerExtraPlugin {
     await fs.writeFile(this.engineCacheFile, blocker.serialize())
   }
 
+  /**
+   * Return a list of exceptions to the blocking rules.
+   */
+  private getExceptions(): NetworkFilter[] {
+    let list: RegExp[] = this.opts.allowlist || [];
+    return list
+      // Turn the list of regexps into an array of `NetworkFilter` instances.
+      .map((regex) => NetworkFilter.parse("@@" + regex.toString()))
+      // Remove any invalid filters.
+      .filter((filter): filter is NetworkFilter => filter !== undefined)
+  }
+
   /**
    * Initialize instance of `PuppeteerBlocker` from cache if possible.
    * Otherwise, it throws and we will try to initialize it from remote instead.
@@ -118,6 +133,9 @@ export class PuppeteerExtraPluginAdblocker extends PuppeteerExtraPlugin {
         await this.persistToCache(this.blocker)
       }
     }
+    this.blocker.update({
+      newNetworkFilters: this.getExceptions(),
+    })
     return this.blocker
   }