An extension library for adding Selenium 3.0 draft and Mobile JSON Wire Protocol Specification draft functionality to the Python language bindings, for use with the mobile testing framework Appium.

The Appium Python Client is fully compliant with the Selenium 3.0 specification draft, with some helpers to make mobile testing in Python easier. The majority of the usage remains as it has been for Selenium 2 (WebDriver), and as the official Selenium Python bindings begins to implement the new specification that implementation will be used underneath, so test code can be written that is utilizable with both bindings.

To use the new functionality now, and to use the superset of functions, instead of including the Selenium webdriver module in your test code, use that from Appium instead.

from appium import webdriver

From there much of your test code will work with no change.

As a base for the following code examples, the following sets up the UnitTest environment:

# Android environment import unittest from appium import webdriver desired_caps = {} desired_caps['platformName'] = 'Android' desired_caps['platformVersion'] = '4.2' desired_caps['deviceName'] = 'Android Emulator' desired_caps['app'] = PATH('../../../apps/selendroid-test-app.apk') self.driver = webdriver.Remote('http://localhost:4723/wd/hub', desired_caps)

# iOS environment import unittest from appium import webdriver desired_caps = {} desired_caps['app'] = PATH('../../apps/UICatalog.app.zip') self.driver = webdriver.Remote('http://localhost:4723/wd/hub', desired_caps)

The methods that do change are...

For mobile testing the Selnium methods for switching between windows was previously commandeered for switching between native applications and webview contexts. Methods explicitly for this have been added to the Selenium 3 specification, so moving forward these 'context' methods are to be used.

To get the current context, rather than calling driver.current_window_handle you use

current = driver.context

The available contexts are not retrieved using driver.window_handles but with

driver.contexts

Finally, to switch to a new context, rather than driver.switch_to.window(name), use the comparable context method

context_name = "WEBVIEW_1" driver.switch_to.context(context_name)

This allows elements in iOS applications to be found using recursive element search using the UIAutomation library. Adds the methods driver.find_element_by_ios_uiautomation and driver.find_elements_by_ios_uiautomation.

el = self.driver.find_element_by_ios_uiautomation('.elements()[0]') self.assertEqual('UICatalog', el.get_attribute('name'))

els = self.driver.find_elements_by_ios_uiautomation('elements()') self.assertIsInstance(els, list)

This allows elements in an Android application to be found using recursive element search using the UIAutomator library. Adds the methods driver.find_element_by_android_uiautomator and driver.find_elements_by_android_uiautomator.

el = self.driver.find_element_by_android_uiautomator('new UiSelector().description("Animation")') self.assertIsNotNone(el)

els = self.driver.find_elements_by_android_uiautomator('new UiSelector().clickable(true)') self.assertIsInstance(els, list)

Allows for elements to be found using the "Accessibility ID". The methods take a string representing the accessibility id or label attached to a given element, e.g., for iOS the accessibility identifier and for Android the content-description. Adds the methods driver.find_element_by_accessibility_id and find_elements_by_accessibility_id.

el = self.driver.find_element_by_accessibility_id('Animation') self.assertIsNotNone(el)

els = self.driver.find_elements_by_accessibility_id('Animation') self.assertIsInstance(els, list)

In order to accomodate mobile touch actions, and touch actions involving multiple pointers, the Selenium 3.0 draft specifies "touch gestures" and "multi actions", which build upon the touch actions.

move_to: note that use keyword arguments if no element

The API is built around TouchAction objects, which are chains of one or more actions to be performed in a sequence. The actions are:

The perform method sends the chain to the server in order to be enacted. It also empties the action chain, so the object can be reused. It will be at the end of all single action chains, but is unused when writing multi-action chains.

The tap method stands alone, being unable to be chained with other methods. If you need a tap-like action that starts a longer chain, use press.

It can take either an element with an optional x-y offset, or absolute x-y coordinates for the tap, and an optional count.

el = self.driver.find_element_by_accessibility_id('Animation') action = TouchAction(self.driver) action.tap(el).perform() el = self.driver.find_element_by_accessibility_id('Bouncing Balls') self.assertIsNotNone(el)

In addition to chains of actions performed with in a single gesture, it is also possible to perform multiple chains at the same time, to simulate multi-finger actions. This is done through building a MultiAction object that comprises a number of individual TouchAction objects, one for each "finger".

Given two lists next to each other, we can scroll them independently but simultaneously:

els = self.driver.find_elements_by_tag_name('listView') a1 = TouchAction() a1.press(els[0]) \ .move_to(x=10, y=0).move_to(x=10, y=-75).move_to(x=10, y=-600).release() a2 = TouchAction() a2.press(els[1]) \ .move_to(x=10, y=10).move_to(x=10, y=-300).move_to(x=10, y=-600).release() ma = MultiAction(self.driver, els[0]) ma.add(a1, a2) ma.perform();

There are a small number of operations that mobile testers need to do quite a bit that can be relatively complicated to build using the Touch and Multi-touch Action API. For these we provide some convenience methods in the Appium client.

This method, on the WedDriver object, allows for tapping with multiple fingers, simply by passing in an array of x-y coordinates to tap.

el = self.driver.find_element_by_name('Touch Paint') action.tap(el).perform() # set up array of two coordinates positions = [] positions.append((100, 200)) positions.append((100, 400)) self.driver.tap(positions)

Swipe from one point to another point.

Zoom in on an element, doing a pinch out operation.

Zoom out on an element, doing a pinch in operation.