alexmanno / panther

A browser testing and web crawling library for PHP and Symfony

A browser testing and web scraping library for PHP and Symfony

Panther is a convenient standalone library to scrape websites and to run end-to-end tests using real browsers.

Panther is super powerful, it leverages the W3C's WebDriver protocol to drive native web browsers such as Google Chrome and Firefox.

Panther is very easy to use, because it implements the popular Symfony's BrowserKit and DomCrawler APIs, and contains all features you need to test your apps. It will sound familiar if you have ever created a functional test for a Symfony app: as the API is exactly the same! Keep in mind that Panther can be used in every PHP project, it's a standalone library.

Panther automatically finds your local installation of Chrome and launches it (thanks to ChromeDriver), so you don't need to install anything on your computer, neither Selenium server nor obscure driver.

In test mode, Panther automatically starts your application using the PHP built-in web-server. You can just focus on writing your tests or web-scraping scenario, Panther takes care of everything else.

Install

Use Composer to install Panther in your project. You may want to use the --dev flag if you want to use Panther for testing only and not for web scraping:

composer req symfony/panther

composer req --dev symfony/panther

Basic Usage

<?php

require __DIR__.'/vendor/autoload.php'; // Composer's autoloader

$client = \Symfony\Component\Panther\Client::createChromeClient();
$crawler = $client->request('GET', 'http://api-platform.com'); // Yes, this website is 100% in JavaScript

$link = $crawler->selectLink('Support')->link();
$crawler = $client->click($link);

// Wait for an element to be rendered
$client->waitFor('.support');

echo $crawler->filter('.support')->text();
$client->takeScreenshot('screen.png'); // Yeah, screenshot!

Testing Usage

The PantherTestCase class allows you to easily write E2E tests. It automatically starts your app using the built-in PHP web server and let you crawl it using Panther. It extends PHPUnit's TestCase and provide all testing tools you're used to.

<?php

use Symfony\Component\Panther\PantherTestCase;

class E2eTest extends PantherTestCase
{
    public function testMyApp()
    {
        $client = static::createPantherClient(); // Your app is automatically started using the built-in web server
        $crawler = $client->request('GET', '/mypage');

        $this->assertContains('My Title', $crawler->filter('title')->text()); // You can use any PHPUnit assertion
    }
}

To run this test:

phpunit tests/E2eTest.php

A Polymorph Feline

If you are testing a Symfony application, PantherTestCase automatically extends the WebTestCase class. It means you can easily create functional tests, which can directly execute the kernel of your application and access all your existing services. Unlike the Panther's client, the Symfony's testing client doesn't support JavaScript and screenshots capturing, but it is super-fast!

Alternatively (and even for non-Symfony apps), Panther can also leverage the Goutte web scraping library, which is an intermediate between the Symfony's and the Panther's test clients. Goutte sends real HTTP requests, it is fast and is able to browse any webpage, not only the ones of the application under test. But Goutte doesn't support JavaScript and other advanced features because it is entirely written in PHP.

The fun part is that the 3 libraries implement the exact same API, so you can switch from one to another just by calling the appropriate factory method, and find the good trade off for every single test case (do I need JavaScript, do I need to authenticate to an external SSO server, do I want to access the kernel of the current request...).

<?php

use Symfony\Component\Panther\PantherTestCase;
use Symfony\Component\Panther\Client;

class E2eTest extends PantherTestCase
{
    public function testMyApp()
    {
        $symfonyClient = static::createClient(); // A cute kitty: the Symfony's functional test too
        $goutteClient = static::createGoutteClient(); // An agile lynx: Goutte
        $pantherClient = static::createPantherClient(); // A majestic Panther
        // Both Goutte and Panther benefits from the built-in HTTP server

        $customChromeClient = Client::createChromeClient(null, null, [], 'https://example.com'); // Create a custom Chrome client
        $customSeleniumClient = Client::createSeleniumClient('http://127.0.0.1:4444/wd/hub', null, 'https://example.com'); // Create a custom Selenium client
        // When initializing a custom client, the integrated web server IS NOT started automatically.
        // Use PantherTestCase::startWebServer() or WebServerManager if you want to start it manually.

        // enjoy the same API for the 3 felines
        // $*client->request('GET', '...')

        $kernel = static::createKernel(); // You can also access to the app's kernel

        // ...
    }
}

Features

Unlike testing and web scraping libraries you're used to, Panther:

• executes the JavaScript code contained in webpages
• supports everything that Chrome (or Firefox) implements
• allows screenshots taking
• can wait for the appearance of elements loaded asynchronously
• lets you run your own JS code or XPath queries in the context of the loaded page
• supports custom Selenium server installations
• supports remote browser testing services including SauceLabs and BrowserStack

Documentation

Since Panther implements the API of popular libraries, it already has an extensive documentation:

• For the Client class, read the BrowserKit's documentation
• For the Crawler class, read the DomCrawler's documentation
• For Webdriver, read the Facebook's PHP WebDriver documentation

Notice

• Webdriver returns only the displayed text. When you filter on head tag (like title), the method text() returns an empty string. Use the method html() method to get the complete contents of the tag (including the tag itself).

Environment Variables

The following environment variables can be set to change some Panther behaviors:

• PANTHER_NO_HEADLESS: to disable browsers's headless mode (will display the testing window, useful to debug)
• PANTHER_NO_SANDBOX: to disable Chrome's sandboxing (unsafe, but allows to use Panther in containers)
• PANTHER_WEB_SERVER_DIR: to change the project's document root (default to public/)
• PANTHER_CHROME_DRIVER_BINARY: to use another chromedriver binary, instead of relying on the ones already provided by Panther
• PANTHER_CHROME_ARGUMENTS: to customize chromedriver arguments. You need to set PANTHER_NO_HEADLESS to fully customize.

Docker Integration

Here is a minimal Docker image that can run Panther:

FROM php:latest

RUN apt-get update && apt-get install -y zlib1g-dev chromium && docker-php-ext-install zip
ENV PANTHER_NO_SANDBOX 1

Build it with docker build . -t myproject Run it with docker run -it -v "$PWD":/srv/myproject -w /srv/myproject myproject bin/phpunit

Travis CI Integration

Panther will work out of the box with Travis if you add the Chrome addon. Here is a minimal .travis.yml file to run Panther tests:

language: php
addons:
  chrome: stable

php:
  - 7.2

script:
  - phpunit

AppVeyor Integration

Panther will work out of the box with AppVeyor as long as Google Chrome is installed. Here is a minimal appveyor.yml file to run Panther tests:

build: false
platform: x86
clone_folder: c:\projects\myproject

cache:
  - '%LOCALAPPDATA%\Composer\files'

install:
  - ps: Set-Service wuauserv -StartupType Manual
  - cinst -y php composer googlechrome
  - refreshenv
  - cd c:\tools\php72
  - copy php.ini-production php.ini /Y
  - echo date.timezone="UTC" >> php.ini
  - echo extension_dir=ext >> php.ini
  - echo extension=php_openssl.dll >> php.ini
  - echo extension=php_mbstring.dll >> php.ini
  - echo extension=php_curl.dll >> php.ini
  - echo memory_limit=3G >> php.ini
  - cd %APPVEYOR_BUILD_FOLDER%
  - composer install --no-interaction --no-progress

test_script:
  - cd %APPVEYOR_BUILD_FOLDER%
  - php vendor\phpunit\phpunit\phpunit

Limitations

The following features are not currently supported:

• Crawling XML documents (only HTML is supported)
• Updating existing documents (browsers are mostly used to consume data, not to create webpages)
• Setting form values using the multidimensional PHP array syntax
• Methods returning an instance of \DOMElement (because this library uses WebDriverElement internally)
• Selecting invalid choices in select

Pull Requests are welcome to fill the remaining gaps!

Credits

Created by Kévin Dunglas. Sponsored by Les-Tilleuls.coop.

Panther is built on top of PHP WebDriver and several other FOSS libraries. It has been inspired by Nightwatch.js, a WebDriver-based testing tool for JavaScript.