Selenium driver for WinForms applications
Any contributions are welcomed :)
This project is using
- Windows Automation
- Nancy as the webserver
- XPathParser for parsing XPath expressions (https://github.com/quamotion/XPathParser)
- Selenium.WebDriver.3.141.0 to get the internals of web driver (commands for Nancy endpoints for example)
- windowsphonedriver (https://github.com/forcedotcom/windowsphonedriver) for the basic infrastructure (commands & command handlers)
Why another driver, when there already is https://github.com/Microsoft/WinAppDriver? The app I'm testing is quite large with many elements, and that driver was timing out after 60 seconds on some XPath queries (see this issue), and after putting everything together, the query finished under 2 seconds.
This project relies on MS UI Automation which is a part of the Windows Accessibility API.
The official documentation mentions support for various platforms:
UI Automation is supported on the following operating systems: Windows XP, Windows Server 2003, Windows Server 2003 R2, Windows Vista, Windows 7, Windows 10, Windows Server 2008, Windows Server 2008 R2, Windows Server 2012, Windows Server 2012 R2, Windows Server 2016, and Windows Server 2019.
I can't confirm it will work on those platforms since I only have Win 10 Enterprise on my machine, but all the required libraries are part of the project.
As far as I know developer mode is not required.
Currently, there is no installer. Clone the repository and build the executable from the sources.
Use the executable found under WinAppDriver.Server\bin\Debug\WinAppDriver.Server.exe
Latest client driver version 3.141.0.0 must be used when calling the server.
- acceptAlert
- default captions to locate the accept button are
Ok
andYes
, additional captions can be added by settingacceptAlertButtonCaptions
capability with a semicolon separated list of values
- default captions to locate the accept button are
- actions
- clearElement
- clickElement
- close
- describeElement
- dismissAlert
- default captions to locate the accept button are
Cancel
andNo
, additional captions can be added by settingdismissAlertButtonCaptions
capability with a semicolon separated list of values
- default captions to locate the accept button are
- elementEquals
- executeAsyncScript
- executeScript
- findChildElement
- findChildElements
- findElement
- findElements
- getActiveElement
- getAlertText
- getCurrentWindowHandle
- getElementAttribute
- getElementLocation
- getElementLocationOnceScrolledIntoView
- getElementSize
- getElementTagName
- getElementText
- getElementValueOfCssProperty
- getSessionCapabilities
- getSessionList
- getTitle
- getWindowHandles
- getWindowPosition
- getWindowRect/getWindowSize
- implicitlyWait
- isElementDisplayed
- isElementEnabled
- isElementSelected
- maximizeWindow
- mouseClick
- mouseDoubleClick
- mouseDown
- mouseMoveTo
- mouseUp
- newSession
- quit
- screenshot
- sendKeysToActiveElement
- sendKeysToElement
- setAlertValue
- setTimeout
- setWindowPosition
- setWindowSize
- status
- submitElement
- switchToWindow
- touchDoubleTap
- touchDown
- touchFlick
- touchLongPress
- touchMove
- touchScroll
- touchSingleTap
- touchUp
- uploadFile
Unsupported Selenium commands
- addCookie
- deleteAllCookies
- deleteCookie
- get
- getCookies
- getCurrentUrl
- getOrientation
- getPageSource
- goBack
- goForward
- refresh
- setOrientation
- setScriptTimeout
- switchToFrame
XPath support:
- axes
- Ancestor
- AncestorOrSelf
- Attribute
- Child
- Descendant
- DescendantOrSelf
- Following
- FollowingSibling
- Namespace
- Parent
- Preceding
- PrecedingSibling
- Self
- Root
- predicates
- position predicate
//node[3]
- attribute predicate
//node[@attribute = 'X']
- position predicate
- operators
- Or
- And
- Eq
- Ne
- Lt
- Le
- Gt
- Ge
- Plus
- Minus
- Multiply
- Divide
- Modulo
- UnaryMinus
- Union
- functions (https://developer.mozilla.org/en-US/docs/Web/XPath/Functions)
- boolean()
- ceiling()
- choose()
- concat()
- contains()
- count()
- current() XSLT-specific
- document() XSLT-specific
- element-available()
- false()
- floor()
- format-number() XSLT-specific
- function-available()
- generate-id() XSLT-specific
- id() (partially supported)
- key() XSLT-specific
- lang()
- last()
- local-name()
- name()
- namespace-uri()
- normalize-space()
- not()
- number()
- position()
- round()
- starts-with()
- string()
- string-length()
- substring()
- substring-after()
- substring-before()
- sum()
- system-property() XSLT-specific
- translate()
- true()
- unparsed-entity-url() XSLT-specific (not supported)
Only this Appium command is implemented
- closeApp
Add a reference to Selenium.WebDriver
v3.141.0 (https://www.nuget.org/packages/Selenium.WebDriver/3.141.0) and you are ready to go.
The driver can either start the system under test process or attach to a running process. Use capabilities to define the process to attach to.
When no command-line argument is provided, the server will be launched at default IP address http://127.0.0.1:4444
.
Following capabilities are supported:
mode
-start
(default value) orattach
processId
- id of the process to attach toprocessName
- name of the process to attach toexePath
orapp
- path to the executable to start the process (arguments cannot be provided at the moment)appWorkingDir
- set the working directory of the new processmainWindowTitle
- regular expression to help the WinAppDriver narrow down the process to attach tosendKeyStrategy
(oneByOne
|grouped
|setValue
) - strategy to use to type text into a text field, default value isoneByOne
, currently this capability cannot be changed during the session
public static RemoteWebDriver CreateSessionByStartingTheApplication()
{
DesiredCapabilities desktopCapabilities = new DesiredCapabilities();
desktopCapabilities.SetCapability("app", "<name of the program to start>");
// or "exePath" desktopCapabilities.SetCapability("exePath", "<path to the executable to start the process>");
// following capabilities should be provided for UWP applications like Calculator or Clocks & Alarms
// optional - to set the working directory
desktopCapabilities.SetCapability("appWorkingDir", "<path to run the process in>");
// optional - to identify the process
desktopCapabilities.SetCapability("processName", "<name of the process>");
// optional - to identify the main window
desktopCapabilities.SetCapability("mainWindowTitle", "<name of the process>");
return new RemoteWebDriver(
new CommandExec(new Uri("http://127.0.0.1:4444"),
TimeSpan.FromSeconds(60)),
desktopCapabilities);
}
public static RemoteWebDriver CreateSessionByAttachingToRunningProcess()
{
DesiredCapabilities desktopCapabilities = new DesiredCapabilities();
desktopCapabilities.SetCapability("mode", "attach");
// attach to process using process name
desktopCapabilities.SetCapability("processName", "<name of the process to attach to>");
// with (optional)
desktopCapabilities.SetCapability("windowTitle", "<regular expression to narrow down the list of matching processes>");
// or attach to process using process id
desktopCapabilities.SetCapability("processId", "<id of the process to attach to>");
return new RemoteWebDriver(
new CommandExec(new Uri("http://127.0.0.1:4444"),
TimeSpan.FromSeconds(60)),
desktopCapabilities);
}
Recommended element location is using XPath expression (though with a limited expression support)
var webBrowser = session.FindElement(By.XPath("//Pane[@AutomationId='webBrowser']"));
Element location mechanisms that are supported
- XPath
- Id, i.e. UI automation element automation id
- CSS class (limited)
- tag name, for example
TextBox
- id, for example
#id
- tag name, for example
- Class name, i.e. UI automation element class name
- Tag name
- Accessibility id, i.e. UI automation element automation id
- Name, i.e. UI automation element name
Windows in the Win application are not like windows in the browser. The windows (ControlType.Window
) can be nested inside the control tree, for example in a Tab
element.
Window can be either located using XPath expression var window = session.FindElement(By.XPath("/Window/Pane/Window[@AutomationId='WindowName']"));
or by switching to it session.SwitchTo().Window("WindowName");
, in the first example you will get the element reference, in the other the internal context is switched to new window and elements cached during the following operations can be disposed when window is cloded session.Close()
.
Note that element wrappers like OpenQA.Selenium.Support.UI.SelectElement
do not work because internally select
and option
elements are expected.
There is quite an ugly hack which bypasses the need to switch windows when searching for a child window using XPath expression. By default, the search starts at the root element of a window - the main window or the window the user switched to - and the search for a child window was failing since it was not a direct child of the current root. The fix starts the search for a window at the top level.