I am experimenting with WatiN for our UI testing, I can get tests to work, but I can't get IE to close afterwards.
I'm trying to close IE in my class clean up code, using WatiN's example IEStaticInstanceHelper technique.
The problem seems to be attaching to the IE thread, which times out:
_instance = IE.AttachTo<IE>(Find.By("hwnd", _ieHwnd));
(_ieHwnd is the handle to IE stored when IE is first launched.)
This gives the error:
Class Cleanup method
Class1.MyClassCleanup failed. Error
Message:
WatiN.Core.Exceptions.BrowserNotFoundException:
Could not find an IE window matching
constraint: Attribute 'hwnd' equals
'1576084'. Search expired after '30'
seconds.. Stack Trace: at
WatiN.Core.Native.InternetExplorer.AttachToIeHelper.Find(Constraint
findBy, Int32 timeout, Boolean
waitForComplete)
I'm sure I must be missing something obvious, has anyone got any ideas about this one?
Thanks
For completeness, the static helper looks like this:
public class StaticBrowser
{
private IE _instance;
private int _ieThread;
private string _ieHwnd;
public IE Instance
{
get
{
var currentThreadId = GetCurrentThreadId();
if (currentThreadId != _ieThread)
{
_instance = IE.AttachTo<IE>(Find.By("hwnd", _ieHwnd));
_ieThread = currentThreadId;
}
return _instance;
}
set
{
_instance = value;
_ieHwnd = _instance.hWnd.ToString();
_ieThread = GetCurrentThreadId();
}
}
private int GetCurrentThreadId()
{
return Thread.CurrentThread.GetHashCode();
}
}
And the clean up code looks like this:
private static StaticBrowser _staticBrowser;
[ClassCleanup]
public static void MyClassCleanup()
{
_staticBrowser.Instance.Close();
_staticBrowser = null;
}
The problem is that when MSTEST executes the method with the [ClassCleanup] attribute, it will be run on a thread that isn't part of the STA.
If you run the following code it should work:
[ClassCleanup]
public static void MyClassCleanup()
{
var thread = new Thread(() =>
{
_staticBrowser.Instance.Close();
_staticBrowser = null;
});
thread.SetApartmentState(ApartmentState.STA);
thread.Start();
thread.Join();
}
The WatiN website briefly mentions that WatiN won't work with threads not in the STA here but it isn't obvious that [TestMethod]'s run in the STA while methods like [ClassCleanup] and [AssemblyCleanupAttribute] do not.
By default when IE object are destroyed, they autoclose the browser.
Your CleanUp code may try to find a browser already close, which why you have an error.
Fixed this myself by dumping mstest and using mbunit instead. I also found that I didn't need to use any of the IEStaticInstanceHelper stuff either, it just worked.
Related
guys. Today I have done my custom realization for WebDriverEventListener. I need only onException() method which will create screenshot. But I got problem because I am using fluent wait.
new FluentWait<>(webDriver)
.withTimeout(Duration.ofSeconds(10))
.pollingEvery(Duration.ofMillis(500))
.ignoring(NoSuchElementException.class)
.until(someCondition)
So, finally, I have got screen for each ignoring(NoSuchElementException.class) - 20 screenshots for 1 fail ))). Had somebody the such problem or had someone resolve it?
when you use .ignoring(NoSuchElementException.class) you don't avoid that the exception is raised, you are just ignoring that exception. What is happening is that the exception is being raised by your FluentWait, but it is ignored (when you declare .ignoring(NoSuchElementException.class)).
You have three options here:
Capture the screen at the end of your test if the test failed [preferred].
Have a Try-Catch wherever you are using your FluentWait or any other Selenium code.
Use reflection to avoid capture when the event is raised from the method that implements the FluentWait.
This is an idea after what we have discussed:
private void ExceptionThrown(object sender, WebDriverExceptionEventArgs e)
{
if (e.ThrownException is NoSuchElementException)
{
// Get the stack trace from the current exception
StackTrace stackTrace = new StackTrace(e.ThrownException, true);
// Get the method stack frame index.
int stackTraceIndex = stackTrace.FrameCount - 1;
// Get the method name that caused the exception
string methodName = stackTrace.GetFrame(stackTraceIndex).GetMethod().Name;
if(methodName != "MyFindElement")
{
TakeSceenshot();
}
}
else
{
TakeSceenshot();
}
}
// This is an extension method of the DriverHelper interface
public IWebElement MyFindElement(this IWebDriver driver, By by, int timeOut = 0)
{
var wait = new WebDriverWait(driver, TimeSpan.FromSeconds(timeOut));
wait.IgnoreExceptionTypes(typeof(NoSuchElementException));
// I wait until the element exist
IWebElement result = wait.Until(drv => drv.FindElement(by) != null);
// it means that the element doesn't exist, so we throw the exception
if(result == null)
{
MyPersonalException(by);
}
}
// The parameter will help up to generate more accurate log
public void MyPersonalException(By by)
{
throw new NoSuchElementException(by.ToString());
}
This probably require changes in EventFiringWebDriver, because this class is without WebDriverWait instance and events for them. If you want avoid it, create bool variable in your EventFiringWebDriver extended class and check this value in your OnException like:
protected void OnException(WebDriverExceptionEventArgs e) {
if (IsWaitHandler)
return;
Your actions...
}
but this is not perfect solution.
I realize that Selenium has a default value for implicit waits, but how do I get this value if I change it? For example:
driver.implicitly_wait( 13 );
How do I later get the 13 value from the driver?
Unfortunately there's no getter for that.
http://selenium.googlecode.com/git/docs/api/java/org/openqa/selenium/WebDriver.Timeouts.html
There isn't for explicit waits either.
http://selenium.googlecode.com/svn/trunk/docs/api/java/com/thoughtworks/selenium/Wait.html
I know I'm a couple years late, and #JaneGoodall is not wrong -- there is no built-in function for that. But it's not impossible!
It's not very difficult to create your own versions of the WebDriver interface and browser-specific driver class. And then, you can put whatever code you want into the driver!
Example:
MyDriver.java (specialized version of WebDriver, not quite mandatory but a very good idea):
public interface MyDriver extends WebDriver {
void setWait(int timeout);
int getWait();
}
MyChrome.java (specialized version of ChromeDriver -- works the same for any browser)
public class MyChrome extends ChromeDriver implements MyDriver {
int timeout = 0;
public void setWait(int timeout) {
this.timeout = timeout;
this.manage().timeouts().implicitlyWait(timeout, TimeUnit.SECONDS);
}
public int getWait() {
return timeout;
}
}
And now, to use it, MyProgram.java:
public class MyProgram {
MyDriver driver = new MyChrome();
driver.setWait(10);
assert(driver.getWait() == 10);
}
I hope this is helpful!
For those who came here from google. In 2018 it seems like there is a method to get those timeouts at least in javascript (I know question was about java):
const {implicit, pageLoad, script} = await driver.manage().getTimeouts();
Hope this will help.
TL;DR - This is not a solution to get implicit waits. You cannot get the implicit wait in Java even today, without using a workaround like this.
In 2020, selenium 3.141.59 still does not have a getter for any timeouts. The WebDriver interface has a nested interface Timeouts which does not define any getters. RemoteWebDriver, which is the parent of Chrome and Firefox drivers, implements the WebDriver interface and it does not add a getter for timeouts.
RemoteWebDriver implements WebDriver.Timeouts, but it does not store the value of implicit timeout anywhere, as you can see from the code below.
protected class RemoteTimeouts implements Timeouts {
public Timeouts implicitlyWait(long time, TimeUnit unit) {
execute(DriverCommand.SET_TIMEOUT, ImmutableMap.of(
"implicit", TimeUnit.MILLISECONDS.convert(time, unit)));
return this;
}
public Timeouts setScriptTimeout(long time, TimeUnit unit) {
execute(DriverCommand.SET_TIMEOUT, ImmutableMap.of(
"script", TimeUnit.MILLISECONDS.convert(time, unit)));
return this;
}
public Timeouts pageLoadTimeout(long time, TimeUnit unit) {
execute(DriverCommand.SET_TIMEOUT, ImmutableMap.of(
"pageLoad", TimeUnit.MILLISECONDS.convert(time, unit)));
return this;
}
} // timeouts class.
The execute() method in the RemoteWebDriver takes the wait inside a Map of parameters, but it does not make that map or the wait settings accessible to us via a getter.
protected Response execute(String driverCommand, Map<String, ?> parameters)
//Open the source code to see why you can't make your own getter for implicitWait.
This can print real timeout value (plus calculating time, usually within 100ms):
public void getCurrentWaitTimeout() {
long milliseconds = java.time.ZonedDateTime.now().toInstant().toEpochMilli();
driver.findElements(By.cssSelector(".nonExistingElement"));
milliseconds = java.time.ZonedDateTime.now().toInstant().toEpochMilli() - milliseconds;
log.info("Current waiting timeout is {} milliseconds", milliseconds);
}
So you can always call such a method to be sure you know actual timeout, not the value you tried to set.
For Java version of Selenium, org.seleniumhq.selenium:selenium-api:4.0.0-beta-4
allows you to get the current implicit wait duration:
WebDriver.manage().timeouts().getImplicitWaitTimeout()
With this method, it makes possible to temporarily change the timeout to let's say 1 second and restore it afterwards:
final Duration originalTimeout = driver.manage().timeouts().getImplicitWaitTimeout();
driver.manage().timeouts().implicitlyWait(Duration.of(1, ChronoUnit.SECONDS));
... // do something
// restore the original timeout
driver.manage().timeouts().implicitlyWait(originalTimeout);
Probably, this functionality is present even before selenium-api:4.0.0-beta-4.
I get the defined implicitTimeout with:
driver.manage().timeouts().getImplicitWaitTimeout().getSeconds()
Many years later, in Python, using selenium 4.4.3 you can access the timeouts very easily via simply:
print(driver.timeouts.implicit_wait)
print(driver.timeouts.page_load)
print(driver.timeouts.script)
Note that when I call driver.implicitly_wait(60), it changed the implicit_wait value only.
Also, there is no such function as driver.manage() in Python, as in the Java answers above.
For measuring execution time of methods, I've seen suggestions to use
public class PerformanceInterceptor {
#AroundInvoke
Object measureTime(InvocationContext ctx) throws Exception {
long beforeTime = System.currentTimeMillis();
Object obj = null;
try {
obj = ctx.proceed();
return obj;
}
finally {
time = System.currentTimeMillis() - beforeTime;
// Log time
}
}
Then put
#Interceptors(PerformanceInterceptor.class)
before whatever method you want measured.
Anyway I tried this and it seems to work fine.
I also added a
public static long countCalls = 0;
to the PerformanceInterceptor class and a
countCalls++;
to the measureTime() which also seems to work o.k.
With my newby hat on, I will ask if my use of the countCalls is o.k. i.e
that Glassfish/JEE6 is o.k. with me using static variables in a Java class that is
used as an Interceptor.... in particular with regard to thread safety. I know that
normally you are supposed to synchronize setting of class variables in Java, but I
don't know what the case is with JEE6/Glassfish. Any thoughts ?
There is not any additional thread safety provided by container in this case. Each bean instance does have its own instance of interceptor. As a consequence multiple thread can access static countCalls same time.
That's why you have to guard both reads and writes to it as usual. Other possibility is to use AtomicLong:
private static final AtomicLong callCount = new AtomicLong();
private long getCallCount() {
return callCount.get();
}
private void increaseCountCall() {
callCount.getAndIncrement();
}
As expected, these solutions will work only as long as all of the instances are in same JVM, for cluster shared storage is needed.
Is this valid code?
selenium = new DefaultSelenium("localhost", 4444, "*iehta",
"http://www.google.com/");
selenium.start();
...
selenium.stop();
...
selenium.start();
...
selenium.stop();
There's nothing wrong with having multiple browsers open (what you call "seleniums"). In fact, it's the only way you can test certain applications. Imagine an application that has an administrative UI and an end-user UI, where you make changes on the admin side and verify their effects on the user side. You can either write your test to jump back and forth between the two on the same browser session, or you can open two browsers, one for each aspect of the application. The former is the usual technique, but the latter is much cleaner.
And why do you think it shouldn't be safe? unless it works ok it's fine, If it doesn't than recreate the DefaultSelenium object again, it won't slow down your code anyway
You should usually keep start() and stop() as you set up and tear down methods. While using TestNG you can annonate then with #BeforeClass and #AfterClass annonations. Hence browser would be launched and shut down only before and after a test method in a class.
b/w did you support Selenium Proposal on area51 - http://area51.stackexchange.com/proposals/4693/selenium
This proposal is backed by SeleniumHQ and we need more users to commit to it to make it see day of light.
That was my fault.
Unexpected behaviour caused by this code and occurs because I stop selenium two times (selenium object never become null):
public class SeleniumController {
private static Selenium selenium;
public static Selenium startNewSelenium(){
// if already exists stop it and replace with new one
if(selenium != null){
selenium.stop();
}
selenium = createNewSelenium(getCurContext());
return selenium;
}
public static void stopSelenium() {
if(selenium != null){
selenium.stop();
}
}
private static Selenium createNewSelenium(TestContext testContext){
TestProperties testProps = new TestProperties(testContext);
ExtendedSelenium selenium = new ExtendedSelenium("localhost", RemoteControlConfiguration.DEFAULT_PORT,
testProps.getBrowser(), testProps.getServerUrl());
selenium.start();
selenium.useXpathLibrary("javascript-xpath");
selenium.allowNativeXpath("false");
return selenium;
}
}
The correct class code is:
public class SeleniumController {
private static Selenium selenium;
public static Selenium startNewSelenium(){
// if already exists stop it and replace with new one
stopSelenium();
selenium = createNewSelenium(getCurContext());
return selenium;
}
public static void stopSelenium() {
if(selenium != null){
selenium.stop();
selenium = null;
}
}
private static Selenium createNewSelenium(TestContext testContext){
TestProperties testProps = new TestProperties(testContext);
ExtendedSelenium selenium = new ExtendedSelenium("localhost", RemoteControlConfiguration.DEFAULT_PORT,
testProps.getBrowser(), testProps.getServerUrl());
selenium.start();
selenium.useXpathLibrary("javascript-xpath");
selenium.allowNativeXpath("false");
return selenium;
}
}
I currently have a test which tests the presenter I have in the MVP model. On my presenter I have a property which will call into my View, which in my test is mocked out. In the Initialization of my test, after I set my View on the Presenter to be the mocked View, I set my property on the Presenter which will call this method.
In my test I do not have an Expect.Call for the method I invoke, yet when I run I get this Rhino mock exception:
Rhino.Mocks.Exceptions.ExpectationViolationException: IView.MethodToInvoke(); Expected #1, Actual #0..
From what I understand with Rhino mocks, as long as I am invoking on the Mock outside the expecting block it should not be recording this. I would imagine the test to pass. Is there a reason it is not passing?
Below is some code to show my setup.
public class Presenter
{
public IView View;
public Presenter(IView view)
{
View = view
}
private int _property;
public int Property
get { return _property;}
set
{
_property = value;
View.MethodToInvoke();
}
}
... Test Code Below ...
[TestInitialize]
public void Initilize()
{
_mocks = new MockRepository();
_view = _mocks.StrictMock<IView>();
_presenter = new Presenter(_view);
_presenter.Property = 1;
}
[TestMethod]
public void Test()
{
Rhino.Mocks.With.Mocks(_mocks).Expecting(delegate
{
}).Verify(delegate
{
_presenter.SomeOtherMethod();
});
}
Why in the world would you want to test the same thing each time a test is run?
If you want to test that a specific thing happens, you should check that in a single test.
The pattern you are using now implies that you need to
- set up prerequisites for testing
- do behavior
- check that behavior is correct
and then repeat that several times in one test
You need to start testing one thing for each test, and that help make the tests clearer, and make it easier to use the AAA syntax.
There's several things to discuss here, but it certainly would be clearer if you did it something like:
[TestMethod]
ShouldCallInvokedMethodWhenSettingProperty()
{
var viewMock = MockRepository.GenerateMock<IView>()
var presenter = new Presenter(viewMock);
presenter.Property = 1;
viewMock.AssertWasCalled(view => view.InvokedMethod());
}
Read up more on Rhino Mocks 3.5 syntax here: http://ayende.com/Wiki/Rhino+Mocks+3.5.ashx
What exactly are you trying to test in the Test method?
You should try to avoid using strict mocks.
I suggest using the Rhino's AAA syntax (Arrange, Act, Assert).
The problem lied with me not understanding the record/verify that is going on with Strict mocks. In order to fix the issue I was having this is how I changed my TestInitilize function. This basicaly does a quick test on my intial state I'm setting up for all my tests.
[TestInitialize]
public void Initilize()
{
_mocks = new MockRepository();
_view = _mocks.StrictMock<IView>();
_presenter = new Presenter(_view);
Expect.Call(delegate { _presenter.View.InvokedMethod(); });
_mocks.ReplayAll();
_mocks.VerifyAll();
_mocks.BackToRecordAll();
_presenter.Property = 1;
}