Automating mobile gestures

While the Selenium WebDriver spec has support for certain kinds of mobile interaction, its parameters are not always easily mappable to the functionality that the underlying device automation (like UIAutomation in the case of iOS) provides. To that end, Appium implements the new TouchAction / MultiAction API defined in the newest version of the spec (https://dvcs.w3.org/hg/webdriver/raw-file/tip/webdriver-spec.html#multiactions-1). Note that this is different from the earlier version of the TouchAction API in the original JSON Wire Protocol.

These APIs allow you to build up arbitrary gestures with multiple actuators. Please see the Appium client docs for your language in order to find examples of using this API.

An Overview of the TouchAction / MultiAction API

TouchAction

TouchAction objects contain a chain of events.

In all the appium client libraries, touch objects are created and are given a chain of events.

The available events from the spec are: * press * release * moveTo * tap * wait * longPress * cancel * perform

Here's an example of creating an action in pseudocode:

TouchAction().press(el0).moveTo(el1).release()

The above simulates a user pressing down on an element, sliding their finger to another position, and removing their finger from the screen.

Appium performs the events in sequence. You can add a wait event to control the timing of the gesture.

moveTo coordinates are now absolute to the current position. For example, dragging from 100,100 to 200,200 can be achieved by:

.press(100,100) // Start at 100,100
.moveTo(200,200) // Passing absolute values of 200,200 ending up at 200,200

The appium client libraries have different ways of implementing this, for example: you can pass in coordinates or an element to a moveTo event. Passing both coordinates and an element will treat the coordinates as relative to the element's position, rather than absolute.

Calling the perform event sends the entire sequence of events to appium, and the touch gesture is run on your device.

Appium clients also allow one to directly execute a TouchAction through the driver object, rather than calling the perform event on the TouchAction object.

In pseudocode, both of the following are equivalent:

TouchAction().tap(el).perform()

driver.perform(TouchAction().tap(el))

MultiTouch

MultiTouch objects are collections of TouchActions.

MultiTouch gestures only have two methods, add, and perform.

add is used to add another TouchAction to this MultiTouch.

When perform is called, all the TouchActions which were added to the MultiTouch are sent to appium and performed as if they happened at the same time. Appium first performs the first event of all TouchActions together, then the second, etc.

Pseudocode example of tapping with two fingers:

action0 = TouchAction().tap(el)
action1 = TouchAction().tap(el)
MultiAction().add(action0).add(action1).perform()

Bugs and Workarounds

An unfortunate bug exists in the iOS 7.0 - 8.x Simulators where ScrollViews, CollectionViews, and TableViews don't recognize gestures initiated by UIAutomation (which Appium uses under the hood for iOS). To work around this, we have provided access to a different function, scroll, which in many cases allows you to do what you wanted to do with one of these views, namely, scroll it!

Scrolling

To allow access to this special feature, we override the execute or executeScript methods in the driver, and prefix the command with mobile:. See examples below:

To scroll, pass direction in which you intend to scroll as parameter.

// javascript
driver.execute('mobile: scroll', {direction: 'down'});

// java
JavascriptExecutor js = (JavascriptExecutor) driver;
HashMap<String, String> scrollObject = new HashMap<String, String>();
scrollObject.put("direction", "down");
js.executeScript("mobile: scroll", scrollObject);

# ruby
execute_script 'mobile: scroll', direction: 'down'

# python
driver.execute_script("mobile: scroll", {"direction": "down"})

// c#
Dictionary<string, string> scrollObject = new Dictionary<string, string>();
scrollObject.Add("direction", "down");
((IJavaScriptExecutor)driver).ExecuteScript("mobile: scroll", scrollObject));

$params = array(array('direction' => 'down'));
$driver->executeScript("mobile: scroll", $params);

Sample to scroll using direction and element.

// javascript
driver.execute('mobile: scroll', {direction: 'down', element: element.value.ELEMENT});

// java
JavascriptExecutor js = (JavascriptExecutor) driver;
HashMap<String, String> scrollObject = new HashMap<String, String>();
scrollObject.put("direction", "down");
scrollObject.put("element", ((RemoteWebElement) element).getId());
js.executeScript("mobile: scroll", scrollObject);

# ruby
execute_script 'mobile: scroll', direction: 'down', element: element.ref

# python
driver.execute_script("mobile: scroll", {"direction": "down", element: element.getAttribute("id")})

// c#
Dictionary<string, string> scrollObject = new Dictionary<string, string>();
scrollObject.Add("direction", "down");
scrollObject.Add("element", <element_id>);
((IJavaScriptExecutor)driver).ExecuteScript("mobile: scroll", scrollObject));

$params = array(array('direction' => 'down', 'element' => element.GetAttribute("id")));
$driver->executeScript("mobile: scroll", $params);

Swiping

This is an XCUITest driver specific method that is similar to scrolling (for reference, see https://developer.apple.com/reference/xctest/xcuielement).

This method has the same API as Scrolling, just replace "mobile: scroll" with "mobile: swipe"

Automating Sliders

iOS

// java
// slider values can be string representations of numbers between 0 and 1
// e.g., "0.1" is 10%, "1.0" is 100%
WebElement slider =  driver.findElement(By.xpath("//window[1]/slider[1]"));
slider.sendKeys("0.1");

Android

The best way to interact with the slider on Android is with TouchActions.