behaving is a web application testing framework for Behavior-Driven-Development, based on behave and splinter.

behave is written in Python and is similar to Cucumber. behaving adds the step-libraries for multi-user web/email/sms/gcm interactions, and provides the Python behaving namespace so that independent step-libraries can work together.

Please refer to behave's excellent documentation for a guide on how to use it, how to write your custom steps and make it possible to extend behaving.

Starting to use behaving is pretty easy. Inside some python module, add your features consisting each of one or more scenarios. These features are Gherkin language files with an extension of .feature. In the same directory you should have a steps module which imports the behaving steps as well as your own custom steps (more on that later in the setup_ section) . Here's a basic example:

Feature: Text presence

    Background:
        Given a browser

    Scenario: Search for BDD
        When I visit "http://www.wikipedia.org/"
        And I fill in "search" with "BDD"
        And I press "go"
        Then I should see "Behavior-driven development" within 5 seconds

While the web is the focus of behaving, it also includes simple mocks for a mail, SMS and a GCM server. These come with a small collection of steps allowing you to do things like:

Feature: Email & SMS

    Scenario: Click link in an email
        Given a browser
        When I send an email to "foo@bar.com" with subject "Hello" and body "Try out this website at http://google.com"
        And I click the link in the email I received at "foo@bar.com"
        Then the browser's URL should be "http://google.com/"

    Scenario: Receive SMS with body
        When I send an sms to "+4745690001" with body "Hello world"
        Then I should receive an sms at "+4745690001" containing "world"

    Scenario: Receive GCM Notification
        When I send a gcm message "{"to":"deviceID", "data": {"message": "Foo Bar", "badge": 6}}"
        Then I should receive a gcm notification at "deviceID" containing "{'data': {'message': 'Foo Bar'}}"

Typically, it will be your web application that sends email/sms/notifications and testing it comes down to configuring the application to send email/sms/notifications to the mock servers.

A lot of web apps today rely on multi-user interactions. To help you with those interactions, behaving uses the notion of personas. A persona within a test runs in its own instance of a browser and you can have more than one persona (and its browser instance) running concurrently. You switch among personas by calling

Given "PersonaName" as the persona

Personas are also typically implemented as simple dictionaries allowing them to carry state, save and reuse variables inside a scenario. When a persona is first invoked it is created as an empty dictionary. You can predefine personas though with set values.

Let's take the familiar LOTR characters as our test users. On setting up the test environment (details later in the setup_ section), we set up the characters basic variables we might be needing in the tests as such:

PERSONAS = {
    'Frodo': dict(
        fullname=u'Frodo Baggins',
        email=u'frodo@shire.com',
        password=u'frodopass',
        mobile='+4745690001',
        address: {
            street: "The Shire",
            zip: "4321"
        }
    ),
    'Gandalf': dict(
        fullname=u'Gandalf the Grey',
        email=u'gandalf@wizardry.com',
        password=u'gandalfpass',
        mobile='+4745690004',
        address: {
            street: "Rivendell street 1",
            zip: "1234"
        }
  ),
  ...
}
def before_scenario(context, scenario):
    ...
    context.personas = PERSONAS

Within a test and given a persona, you can now use $var_name to access a variable of a persona. You can also set new variables on personas. So the following,

Given "Gandalf" as the persona
When I fill in "name" with "$fullname"
And I fill in "street" with "$address.street"
And I set "title" to the text of "document-title"
And I fill in "delete" with "$title"
And I set "address.country" to the text of "country"
And I set "postaddress" to:
"""
$fullname
$address.street, $address.zip, $address.country
"""

would fill in the field with id name with Gandalf the Grey, street with Rivendell street 1 set the variable title to the text of the element with id document-title and reuse the variable title to fill in the field with id delete. It would also store the value of the field with id "country" in address[country]. The $var_name pattern is also usable in the text received by steps that expect a body of text, which means that the postaddress persona variable will contain Gandalf's complete snail-mail postage address nicely formatted on multiple lines.

Let us assume the following (coming from a real example) scenario. Crypho, is an online messaging/sharing site that provides users with end-to-end encrypted real-time communications. behaving was written to help test Crypho.

In Crypho, teams collaborate in spaces. To invite somebody in a space the invitee has to share a token with an invitor, so both can verify each other's identity.

Feature: Frodo invites Gandalf to The Shire space

    Given state "the-shire"

    Scenario: Frodo invites Gandalf to The Shire

        Given "Gandalf" as the persona
        When I log in

Before the scenarios start, the custom step Given state "the-shire" executes. This preloads the db with data, sets up the server etc. Then the scenario executes:

First Gandalf logs in. The step Given "Gandalf" as the persona, fires up a browser that belongs to the persona Gandalf. The following step, When I log in is a custom step defined as follows:

@when('I log in')
def log_in(context):

    assert context.persona
    context.execute_steps(u"""
        When I go to Home
            Then I should see an element with id "email" within 2 seconds
        When I fill in "email" with "$email"
        And I press "send-sms"
            Then I should see "We have sent you an SMS with a security code" within 2 seconds
            And I should receive an sms at "$mobile"
            And "token" should be enabled
        When I parse the sms I received at "$mobile" and set "Your Crypho code is {token}"
        And I fill in "token" with "$token"
        And I fill in "password" with "$password"
        And I press "login"
            Then I should see "Crypho" within 5 seconds
    """)

Observe above how the current persona (Gandalf) parses the sms it receives and saves it as "token". Later Gandalf reuses it to fill in the two-factor authentication field.

Now that Gandalf is logged in, the test proceeds with Frodo. Frodo will log in, and invite Gandalf to a private space.

Given "Frodo" as the persona
When I log in
And I click the link with text that contains "My spaces"
And I click the link with text that contains "The Shire"
And I press "invite-members"
    Then I should see "Invite members" within 1 seconds
When I fill in "invitees" with "gandalf@wizardry.com"
And I fill in "invitation-message" with "Come and join us!"
And I press "send-invitations"
    Then I should see "Your invitations have been sent" within 2 seconds

Once the invitations are sent we switch back to Gandalf's browser, who should have received a notification in his browser, as well as an email. He then proceeds to send an sms to Frodo with the token who completes the invitation.

Given "Gandalf" as the persona
Then I should see "Your invitations have been updated" within 2 seconds
And I should receive an email at "gandalf@wizardry.com" containing "Frodo Baggins has invited you to join a private workspace in Crypho"
When I click the link with text that contains "Invitations"
And I click the link with text that contains "Pending invitations"
    Then I should see "Come and join us!"
When I set "token" to the text of "invitation-token"
And I send an sms to "45699900" with body "$token"

Given "Frodo" as the persona
    Then I should receive an sms at "45699900"
When I set "FrodoToken" to the body of the sms I received at "45699900"
And I click the link with text that contains "Invitations"
And I click the link with text that contains "Enter authorization token"
And I fill in "auth-token" with "$FrodoToken"
And I press "Submit"
    Then I should see "The invitation has been accepted." within 5 seconds
    And I should see "Gandalf the Grey has joined the space, invited by Frodo Baggins" within 10 seconds

You can see the test in action on video here.

Start by installing behaving by using either pip or easy_install. This will also install dependencies and create the behave script with which you invoke your tests. If you prefer using buildout, clone the package itself from its repository, it contains already a buildout configuration.

Typically you will be having a folder containing all your features and steps. For example a directory structure like the following:

features/
features/mytest.feature
features/myothertest.feature
features/environment.py
features/steps/
features/steps/steps.py

In the steps directory you will need to import the behaving steps you need. You can also define your own steps. So steps.py might look like:

from behave import when
from behaving.web.steps import *
from behaving.sms.steps import *
from behaving.mail.steps import *
from behaving.notifications.gcm.steps import *
from behaving.personas.steps import *

@when('I go to home')
def go_to_home(context):
    context.browser.visit('https://web/')

In environment.py you specify settings as well the things that need to happen at various stages of testing, i.e. before and after everything, a feature run, or a scenario run. For convenience you can import and reuse behaving.environment which will perform default actions like closing all browsers after a scenario, clean the email folder etc.

It is also possible to use behaving.web.environment, behaving.mail.environment, behaving.sms.environment and behaving.personas.environment on their own, if you don't have need for SMS for example.

An example of an environment that does simply set some variables and then rely on default actions for the various stages, might look like the following:

import os
from behaving import environment as benv

PERSONAS = {}

def before_all(context):
    import mypackage
    context.attachment_dir = os.path.join(os.path.dirname(mypackage.__file__), 'tests/data')
    context.sms_path = os.path.join(os.path.dirname(mypackage.__file__), '../../var/sms/')
    context.gcm_path = os.path.join(os.path.dirname(mypackage.__file__), '../../var/gcm/')
    context.mail_path = os.path.join(os.path.dirname(mypackage.__file__), '../../var/mail/')
    benv.before_all(context)


def after_all(context):
    benv.after_all(context)


def before_feature(context, feature):
    benv.before_feature(context, feature)


def after_feature(context, feature):
    benv.after_feature(context, feature)


def before_scenario(context, scenario):
    benv.before_scenario(context, scenario)
    context.personas = PERSONAS

def after_scenario(context, scenario):
    benv.after_scenario(context, scenario)

The following variables are supported and can be set to override defaults:

• screenshots_dir (the path where screenshots will be saved. If it is set, any failure in a scenario will result in a screenshot of the browser at the time when the failure happened.)
• attachment_dir (the path where file attachments can be found)
• sms_path (the path to be used by smsmock to save sms. Defaults to current_dir/sms )
• gcm_path (the path to be used by gcmmock to save gcm notifications. Defaults to current_dir/gcm )
• mail_path (the path to be used by mailmock to save mail. Defaults to current_dir/mail )
• default_browser
• default_browser_size (tuple (width, height), applied to each browser as it's created)
• max_browser_attempts (how many times to retry creating the browser if it fails)
• remote_webdriver_url (points to your selenium hub url or remote webdriver. Defaults to None)
• browser_args (a dict of additional keyword arguments used when creating a browser)
• base_url (the base url for a browser, allows you to use relative paths)
• accept_ssl_certs (setting to True will accept self-signed/invalid certificates. Defaults to None)

You can run the tests simply by issuing

./bin/behave ./features

For chrome and docker issues, the code below is useful

from selenium.webdriver.chrome.options import Options
chrome_options = Options()
chrome_options.add_argument('--no-sandbox')
context.browser_args = {
    'options': chrome_options
}

When behaving is installed, it creates three scripts to help you test mail, gcm and sms, mailmock, gcmmock and smsmock respectively. You can directly invoke them before running your tests, they all take a port as well as the directory to output data as parameters. For example,

./bin/smsmock -p 8081 -o ./var/sms
./bin/gcmmock -p 8082 -o ./var/notifications/gcm
./bin/mailmock -p 8083 -o ./var/mail [--no-stdout]

• Browsers

  • Given a browser [opens the default browser, i.e. Firefox]
  • Given brand as the default browser [sets the default browser to be brand, this is the browser name when using the remote webdriver or Firefox, Chrome, Safari]
  • Given the electron app "app_path" [for use with electron-based desktop apps]
  • Given browser "name" [opens the browser named name]
  • When I reload
  • When I go back
  • When I go forward
  • When I resize the browser to widthxheight
  • When I resize the viewport to widthxheight
  • When I take a screenshot [will save a screenshot of the browser if screenshots_dir is set on the environment. Also, if screenshots_dir is set, all failing tests will result in a screenshot.]
  • When I execute the script "script"
  • When I set the cookie "key" to "value"
  • When I delete the cookie "key"
  • When I delete all cookies
  • When I close the browser "name"

• Frames

  • When I switch to frame with css "css"
  • When I switch back to the main page

• Windows

  • When I open a new window named "name" at "url"
  • When I name the current window "name"
  • When I switch to the window named "name"

• URLs

  • Given the base url "url" [sets the base url to url, alternatively set context.base_url directly in environment.py]
  • When I visit "url"
  • When I go to "url"
  • When I parse the url path and set "{expression}"
  • Then the browser's URL should be "url"
  • Then the browser's URL should contain "text"
  • Then the browser's URL should not contain "text"

• Links

  • When I click the link to "url"
  • When I click the link to a url that contains "url"
  • When I click the link with text "text"
  • When I click the link with text that contains "text"

• Text, element & class presence

  • When I wait for timeout seconds

  • When I show the element with id "id"

  • When I hide the element with id "id"

  • Text

    • Then I should see "text"
    • Then I should not see "text"
    • Then I should see "text" within timeout seconds
    • Then I should not see "text" within timeout seconds

  • ID

    • Then I should see an element with id "id"
    • Then I should not see an element with id "id"
    • Then I should see an element with id "id" within timeout seconds
    • Then I should not see an element with id "id" within timeout seconds

• CSS

  • Existence
    • Then I should see an element with the css selector "selector"
    • Then I should not see an element with the css selector "selector"
    • Then I should see an element with the css selector "selector" within timeout seconds
    • Then I should not see an element with the css selector "selector" within timeout seconds
    • Then I should see n elements with the css selector "css"
    • Then I should see at least n elements with the css selector "css" within timeout seconds
  • Visibility
    • Then the element with the css selector "css" should be visible
    • Then the element with the css selector "css" should be visible within timeout seconds
    • Then the element with the css selector "css" should not be visible
    • Then the element with the css selector "css" should be visible within timeout seconds
    • Then {n:d} elements with the css selector "css" should be visible
    • Then {n:d} elements with the css selector "css" should be visible within timeout seconds
    • Then at least {n:d} elements with the css selector "css" should be visible
    • Then at least {n:d} elements with the css selector "css" should be visible within timeout seconds
  • Existence of a class on an element
    • Then the element with xpath "xpath" should have the class "cls"
    • Then the element with xpath "xpath" should not have the class "cls"
    • Then the element with xpath "xpath" should have the class "cls" within timeout seconds
    • Then the element with xpath "xpath" should not have the class "cls" within timeout seconds
    • Then "name" should have the class "cls"
    • Then "name" should not have the class "cls"
    • Then "name" should have the class "cls" within timeout seconds
    • Then "name" should not have the class "cls" within timeout:d seconds
  • XPath
    • Then I should see an element with xpath "xpath"
    • Then I should not see an element with xpath "xpath"
    • Then I should see an element with xpath "xpath" within timeout seconds
    • Then I should not see an element with xpath "xpath" within timeout seconds

• Forms

  • When I fill in "name|id" with "value"
  • When I clear field "name|id"
  • When I type "value" to "name|id" [same as fill, but happens slowly triggering keyboard events]
  • When I choose "value" from "name"
  • When I check "name|id"
  • When I uncheck "name|id"
  • When I toggle "name|id"
  • When I select "value" from "name""
  • When I select by text "text" from "name""
  • When I press "name|id|text|innerText"
  • When I press the element with xpath "xpath"
  • When I attach the file "path" to "name"
  • When I set the innner HTML of the element with id "id" to "contents" [Sets html on a contenteditable element with id id to contents]
  • When I set the innner HTML of the element with class "class" to "contents"
  • When I set the innner HTML of the element with class "class" to "contents"
  • When I send "KEY" to "name"
  • When I focus on "name"
  • Then field "name" should have the value "value"
  • Then field "name" should have the value "value" within timeout seconds
  • Then the selection "name" should have the options "valueA, valueB" selected
  • Then "name" should be enabled
  • Then "name" should be disabled
  • Then "name" should not be enabled
  • Then "name" should be valid
  • Then "name" should be invalid
  • Then "name" should not be valid
  • Then "name" should be required
  • Then "name" should not be required

• HTML tables

  • Then the table with id "id" should be
    | header1 | header2 | ... | header(m) |
    | cell00 | cell01 | ... | cell0m |
    | cell10 | cell11 | ... | cell1m |
    ...
    | celln0 | celln1 | ... | cellnm |

  • Then the table with xpath "xpath" should be
    | header1 | header2 | ... | header(m) |
    | cell00 | cell01 | ... | cell0m |
    | cell10 | cell11 | ... | cell1m |
    ...
    | celln0 | celln1 | ... | cellnm |

  • Then the table with id "id" should contain the rows
    | cell00 | cell01 | ... | cell0m |
    | cell10 | cell11 | ... | cell1m |

  • Then the table with xpath "xpath" should contain the rows
    | cell00 | cell01 | ... | cell0m |
    | cell10 | cell11 | ... | cell1m |

  • Then the table with id "id" should not contain the rows
    | cell00 | cell01 | ... | cell0m |
    | cell10 | cell11 | ... | cell1m |

  • Then the table with xpath "xpath" should not contain the rows
    | cell00 | cell01 | ... | cell0m |
    | cell10 | cell11 | ... | cell1m |

  • Then row row_no in the table with id "id" should be
    | cell00 | cell01 | ... | cell0m |

  • Then row row_no in the table with xpath "xpath" should be
    | cell00 | cell01 | ... | cell0m |

  • Then the value of the cell in row row_no, column col_no in the table with id "id" should be "value"

  • Then the value of the cell in row row_no, column col_no in the table with xpath "xpath" should be "value"

  • Then the value of the cell in row row_no, column "col_header" in the table with id "id" should be "value"

  • Then the value of the cell in row row_no, column "col_header" in the table with xpath "xpath" should be "value"

• Alerts & prompts

  • When I enter "text" to the alert - When I accept the alert - When I dismiss the alert - Then I should see an alert - Then I should see an alert within timeout seconds - Then I should see an alert containing "text" - Then I should see an alert containing "text" within timeout seconds

• Mouse

  • When I mouse over the element with xpath "xpath"
  • When I mouse out of the element with xpath "xpath"

• Downloads

  • Then the file "filename" with contents "text" should have been downloaded within timeout seconds
  • Then the file "filename" should have been downloaded within timeout seconds

• Persona interaction & variables

  • When I set "key" to the text of "id|name"
  • When I set "key" to the attribute "attr" of the element with xpath "xpath"
  • When I evaluate the script "script" and assign the result to "key"

• When I click the link in the email I received at "address"
• When I parse the email I received at "address" and set "expression"
• When I clear the email messages
• Then I should receive an email at "address"
• Then I should receive an email at "address" with subject "subject"
• Then I should receive an email at "address" containing "text"
• Then I should receive an email at "address" with attachment "filename"
• Then I should not have received any emails at "address"

• When I set "key" to the body of the sms I received at "number"
• When I parse the sms I received at "number" and set "expression"
• Then I should receive an sms at "number"
• Then I should receive an sms at "number" containing "text"

• When I send a gcm message "{"to":"deviceID", "data": {"message":"Foo Bar", "badge": 6}}"
• Then I should receive a gcm notification at "deviceID" containing "{'data': {'message': 'Foo Bar'}}"
• Then I should have received any gcm notifications at "deviceID"

• Given "name" as the persona
• When I set "key" to "value"
• When I set "key" to:
  """ some longer body of text
usually multiline
  """
• When I clone persona "source" to "target"
• Then "key" is set to "value"

• When I pause the tests

A Dockerfile as well as a complete setup using docker-compose are provided to help you create selenium grid configurations that run your tests. In addition dev container configuration is included if VSCode is your thing.

In addition we provide pre-build images on docker hub for the linux/amd64 and linux/arm64 platforms. Use

docker pull behaving/behaving:latest

to pull the image.

You can run all behaving tests as follows:

Start docker compose:

docker-compose up

Open a shell in the behaving container:

docker-compose exec behaving bash

Run behaving tests:

behave tests/features