Welcome to Dilium - Distributed Selenium project!

What is it

It is python alternative of selenium grid, which uses selenium and ansible under hood.

Why is it

Selenium grid has some disadvantages:

  • Selenium grid doesn’t manage node. It just provides webdriver API. Some additional functionalities, like video capture and xvfb mode, are implemented via hacks on client side (proxies, servlets, etc).
  • Selenium grid requires additional client (minion) on node.
  • Selenium grid is developed with Java. If it’s necessary to expand its functionality we need to learn Java, even if project(s) is developed with other language.

Gridless advantages:

  • Ansible provides flexible and extensible way to manipulate with nodes.
  • Ansible doesn’t require client-side software, just ssh-access.
  • Webdriver can be launched directly on node and listen remote connections as it happens in localhost.

How to launch

Functionality in development

  • Clone project:

    git clone https://github.com/sergeychipiga/dilium

  • Launch ubuntu virtual machine (for ex. via vagrant) with available IP.

  • Add your .ssh/id_rsa.pub to vm .ssh/authorized_keys in order to enable ssh access with key.

  • Install google-chrome, chromedriver, xvfb, libav-tools to vm.

  • Add chromedriver path to $PATH in /etc/environment.

  • Install dilium server:

    pip install -r dilium_server/requirements.txt
pip install -e dilium_server

  • Install dilium client:

    pip install -r dilium_client/requirements.txt
pip install -e dilium_client

  • Launch dilium server:

    DEBUG=1 python dilium_server/dilium_server/bin/app.py

  • Execute code to upload test config:

    import requests
remote_ip = 'ip.of.your.vm'  # for ex. 192.168.1.8
config = {
    remote_ip: {
        'capabilities': [
            {'browserName': 'chrome'}
        ]
    }
}
requests.post('http://127.0.0.1:8888/upload-config/',
              json=config)

  • Execute code on remote node to launch browser in xvfb with video capturing:

    from dilium_client import Client

client = Client('http://127.0.0.1:8888/')
node = client.get_node({'browserName': 'chrome'})

screen_size = (1366, 768)

with node.inside_xvfb(*screen_size):
    with node.capture_video(*screen_size):

        browser = node.get_browser()
        browser.launch()
        # in xvfb direct ``maximize_window``-call doesn't work properly :(
        browser.set_window_size(*screen_size)
        browser.maximize_window()
        browser.get('http://yandex.ru')
        browser.save_screenshot('/path/to/screenshot.png')
        browser.quit()

node.download_video('/path/to/video.mp4')
client.release_host()

Current abilities

  • Hosts distribution
  • Nodes provisioning (verified with ubuntu 14.04)
  • Browser manipulation (verfied with google chrome)
  • Xvfb launching
  • Video capturing

TODO list

  • Check that async call really starts process. (It isn’t verified by default due to shell specific.)
  • Support firefox on linux.
  • Support windows and macos.
  • Expand capabilities to manage nodes from client and server.

Deep to structure