We Keep Coding

Webdriver instances - how to run multiple tests using one test base

I'm learning automation testing in Java using Selenium and Testng. I've came across the issue that I cannot really solve, maybe someone here knows the solution.
I'm trying to run multiple tests from one class using multisession.xml. That works only if I add separate instance to each test, something like this:
System.setProperty("webdriver.chrome.driver", "src/main/resources/chromedriver.exe"); WebDriver driver; driver = new ChromeDriver();
However, I want to use separate class (test base) that extends my class with tests in order to not duplicate code. If I do that, then my multisession.xml can only perform one test at the time and returns following exception:
org.openqa.selenium.StaleElementReferenceException: stale element reference: element is not attached to the page document
My guess there is a problem with instances of webdriver. In my base initialization code looks like that:
public class TestBase {
public static WebDriver driver;
public static WebDriverWait wait;
public static Properties config;
public static Properties testdata;
Then setting up separate files with test data and initialization:
public static void initialization() {
String url = config.getProperty("URL");
String browser = config.getProperty("browser");
String pageLoadTimeout = config.getProperty("pageLoadTimeout");
String windowsMaximize = config.getProperty("windowsMaximize");
String deleteAllCookies = config.getProperty("deleteAllCookies");
String waitTimeout = config.getProperty("waitTimeout");
switch (browser) {
case "chrome":
System.setProperty("webdriver.chrome.driver", System.getProperty("user.dir") +
"/src/main/resources/chromedriver.exe");
ChromeOptions options = new ChromeOptions();
options.addArguments("--incognito");
driver = new ChromeDriver(options);
break;
case "firefox":
System.setProperty("webdriver.gecko.driver", System.getProperty("user.dir") +
"/src/main/resources/geckodriver.exe");
driver = new FirefoxDriver();
break;
case "msedge":
System.setProperty("webdriver.edge.driver", System.getProperty("user.dir") +
"/src/main/resources/msedgedriver.exe");
driver = new EdgeDriver();
break;
case "opera":
System.setProperty("webdriver.opera.driver", System.getProperty("user.dir") +
"/src/main/resources/operadriver.exe");
driver = new OperaDriver();
break;
I will appreciate any help!

StaleElementReferenceException is thrown when the element you're looking for is no longer attached to the DOM. Most probably you're trying to .findElement first, then DOM gets refreshed and you're trying to access it. This often happens with parallel tests execution if your page-class instances with elements are not thread-safe.
Try to use:
object _lock = new object();
lock (_lock)
{
//your driver initialization/login
}
You should also define how many threads will be executing your tests by modifying thread-count="5" parallel="methods"

When trying to run Selenium Cucumber test in two browsers, test runs in only one browser

I am stuck whilst trying to run a selenium cucumber java script in two browsers (Chrome, Firefox). The test works fine when I use a single browser. I use selenium PageFactory class to initialise the web elements.
The problem is when I run the test for two browsers, the first browser opens, navigates to the URL and then nothing happens. Web elements are not initialised. The test moves on to the second browser, navigates to URL, web elements are initialised, subsequent test methods (testMethod1) run as expected. Why is the test not running on the first browser?
This is the PageFactory BasePage class holding the web elements:
public class BasePage {
private final WebDriver driver;
public BasePage(WebDriver driver) {this.driver = driver;} //constructor
#FindBy(id = "cc-amount")
public WebElement amountField;
This is the test class and how I have tried to run the test in two browsers:
public class Convert {
private static WebDriver driver;
private final BaseUtil baseUtil = new BaseUtil();
private static BasePage basePage;
private static int browser;
public void navigateToUrl(String url) throws InterruptedException {
for (browser = 1; browser <= 2; browser++) {
if (browser == 1) {
WebDriverManager.chromedriver().setup();
driver = new ChromeDriver();
} else if (browser == 2) {
WebDriverManager.firefoxdriver().setup();
driver = new FirefoxDriver();
}
driver.get(baseUtil.getMyUrl()); //Url coming from a utility class
driver.manage().window().maximize();
driver.manage().timeouts().implicitlyWait(30, TimeUnit.SECONDS);
basePage = PageFactory.initElements(driver, BasePage.class); //initialisation of the BasePage class conttaining the pagefactory web elements
After the browser initialisation, this method runs ok in the second browser (Firefox) but did not run at all in the first browser (Chrome):
public void testMetod1(String amount) throws InterruptedException {
basePage.amountField.click();
My suspicion is that PageFacory could not be initiated for both browsers in a single run but I do not know how to diagnose this further. It could also be a for loop error.

I believe it's a synchronisation issue. Just you need to create a class for each browser, then you can create another base class which contains all the shared code.
Creating a class for each browser will give you some advantages:
Cleaner code.
Your tests will run in parallel.
Specific test cases for a specific browsers (Specially for IE).
Solve this sync issue.
So basically instead of using Converter, you can create:
ChromeConverter
FirefoxConverter
And inside your TestExecuter class, just create instance for each one and run your tests.
And to be more detailed, you are using one class Convert and inside it you have two static variables:
private static WebDriver driver;
private static BasePage basePage;
Removing the static keyword from BasePage class won't solve the root cause as you are using BasePage which is a static class and can't be instantiated as instance class and it's inside an external library so you need to separate the classes as kind of wrapper.
Here you are trying to use the same exact initialisation with Chrome then Firefox which won't work in parallel or even in for loop until you can wait to the first test then dispose it and initialise all the stuff related to the second test again.
So the correct approach is to create a separate Convert for each browser like that:
ChromeConverter
FirefoxConverter
And In your TestExecuter you just need to initialise each one and use it, something like that:
#Given("...")
public void navigateToUrl(String url) throws InterruptedException {
chromeConverterApp.navigateToUrl(url);
firefoxConverterApp.navigateToUrl(url);
}
#When("^..")
public void enterCurrencies(String amount, String from, String to) throws InterruptedException {
chromeConverterApp.enterCurrencies(amount, from, to);
firefoxConverterApp.enterCurrencies(amount,from,to);
}

The for loop opens Chrome on the first run (browser=1), then opens Firefox on the second run (browser=2) and then continues with the test on the last opened browser (Firefox).
If you wish to run the complete test on both browsers, you need to put the rest of the test inside of the for loop.
Basically, you have something like:
public void navigateToUrl(String url) throws InterruptedException {
for (browser = 1; browser <= 2; browser++) {
//body of the for loop
}
//some more code
} //end of navigateToUrl
You need to change it to:
public void navigateToUrl(String url) throws InterruptedException {
for (browser = 1; browser <= 2; browser++) {
//body of the for loop
}
//some more code
testMethod1()
} //end of navigateToUrl

Your approach is not correct, if you need to run your tests in parallel you should go for either Selenium Grid or consider implementing other multi-threading option, i.e.:
Plain Java: ExecutorService
JUnit or TestNG: Maven Surefire Plugin Fork Options and Parallel Test Execution
In any case you will need to refactor your test to follow Parallel Tests - Best Practices at least:
Remove static modifier from the WebDriver declaration
Consider putting your WebDriver instance into the Thread Local Storage otherwise you will have clashes

How to pass selenium webdriver instance to another class

I browsed through the site but did not find the answer I am looking.
I have
Superbase class- here I just create object of a webdriver
Baseclass- In this class I extend Superbase class, invoke the driver, and open the URL.
Clicklink class- In this class, I again extend the Superbase Class but only to find a null pointer exception. I think I am getting the exception as the driver object is not initialized.
I am just a beginner, and have not tried the browserfactory and other options, as I want to start with simple flow.
Superclass
Public class Superclass
{
public webdriver Driver;
}
Baseclass
public class Baseclass extends Superclass
{
setting capabilities and launching the browser
}
ClickLink
public class Clicklink extends Superclass
{
here I want to click on a link
driver.findelement(by.xpath("xpath").click());
// after this statement I get a null pointer exception
}
Can you please guide me here? how can I achieve the same.
Thanks much!

SuperClass and BaseClass are very poor names. Do not use the language of coding to name your classes use the language of the problem. In this case, web site application testing, use LoginPage, CartPage, ProfilePage, etc. Use the Page Object Pattern.
I suggest you use the Factory Pattern to provide the instances of WebDriver for each test. Since all those fit the idea is a page, use class extension from a standard PageObject to provide this capability. When navigating, have the current page construct an instance of the new page and pass it the current webDriver connection instance. Then any manipulations you apply to that PageObject will be automatically applied to that webDriver instance and its associated browser instance. You should also use a PageFactory to provide instances of the pageObject.
public abstract class PageObject {
public WebDriver driver;
PageObject() {
// Page can initialise its self
this.driver = BrowserFactory.webDriver();
}
PageObject(final WebDriver webDriver) {
this.driver = webDriver;
}
}

This is lot of guesswork done from my side, but please make sure, that your Superclass actually sets the driver and returns it. You can actually make it in both methods:
public class Superclass
{
public WebDriver driver;
public Superclass(){
driver = new FirefoxDriver();
}
public WebDriver getdriver(){
if (driver == null){
driver = new FirefoxDriver();
return driver;
}else{
return driver;
}
}
}
And later in methods you call it by:
public class Clicklink extends Superclass
{
getdriver().findelement(by.xpath("xpath").click());
}

If you doesn't want pass driver instance to Page Objects constructor you could create some container class for driver and put it before test and remove it after run. For example:
class Driver {
public static ThreadLocal<IWebDriver> driverInstance = new ThreadLocal<IWebDriver>();
public static IWebDriver GetDriver() {
return driverInstance.Value;
}
public static void SetDriver(IWebDriver driver) {
driverInstance.Value = driver;
}
}
and make this container field ThreadLocal to avoid problems with parallel run.

I have taken a slightly different approach than most on this thread. When I start a test session, I pass the browser name as an argument (i.e. -Dbrowser=chrome) in order to be able to test my web application with different browsers. Then I used the "browser" system property to obtain the browser name when setup() is called by my test framework. In my case, I use JUnit annotations in order to JUnit to setup all needed dependencies prior to running any tests.
#BeforeClass
public static void setup() throws Exception {
// Set up other stuff
String browser = System.getProperty("browser");
try {
SessionDataProvider.driver = TestUtils.createDriver(browser);
} catch (Exception e) {
...
}
}
The createDriver(String) is a factory method that instantiates the correct driver.
public static WebDriver createDriver(String browserName) throws Exception {
WebDriver driver = null;
try {
switch(browserName) {
case "firefox":
// code to system props and instantiate the driver
break;
case "chrome":
// code to system props and instantiate the driver
break;
case "ibrowser":
// code to system props and instantiate the driver
break;
case "edge":
// code to system props and instantiate the driver
break;
case "safari":
// code to system props and instantiate the driver
break;
default:
throw new Exception("Unsupported browser: " + browserName);
}
return driver;
}
Then, when I execute a step definition, I simply obtain the driver from the data provider class:
#And("(I click)/Click on {string}")
public void click(String arg) {
// Parse String arg and create web element locator...
try {
By locator = ...;
WebElement element = new WebDriverWait(SessionDataProvider.driver, 2)
.until(ExpectedConditions.elementToBeClickable(locator));
element.click();
} catch (Exception e) {
// handle exception
}
}

I did use below code in utility class like below
public static WebDriver setup(WebDriver driver)
{
if(driver == null) {
driver = new FirefoxDriver();
driver.manage().window().maximize();
driver.manage().timeouts().implicitlyWait(10, TimeUnit.SECONDS);
return driver;
}else
return driver;
//System.out.println("in method "+ driver.getTitle() );
}

Stabilizing test methods in one test class

I have a test class using testNG, with 4 test methods, testing one website for 4 different browsers, like the following code:
public class MyTest {
#Test
public void acceptanceFFTest()
{
}
#Test
public void acceptanceChromeTest()
{
}
#Test
public void acceptanceIETest()
{
}
#Test
public void acceptanceSafariTest()
{
}
#AfterMethod
public void tearDown()
{
webDriver.quit();
}
}
After one test method finished, the webdriver for that browser is killed in the #AfterMethod, the next browser starts with new webdriver respectively, and so on.
My problem is: if I run these test methods one by one, then they are always passed. But if I run from the test class level as a test suite with 4 test methods, then sometimes one or two of them are failed with NoElementException, But the exception is not always at one page or one element, sometimes here, sometimes there. If I run the failed tests alone again, then they are passed.
I think for the test class level, maybe the http signal sometimes is fast and sometimes is slow, the test class level is not stable as single test method, I am not sure.
Is there anyway to stabilize them? Thank you.

I would rewrite the tests so that you have a single test that runs on all browsers. You may run into some slight differences in the browser that you may have to handle but I think this overall approach is much better and more manageable. Here's a simple example of how to do this. For this example I just hardcoded a string browserType. I'm assuming you would pass this in or read it from a file or whatever. The basic idea is that you define your driver variable as a generic WebDriver. Once you determine what browser you want for that run, you instantiate the specific driver and execute the test(s).
String browserType = "firefox"; // hardcoded for the example
WebDriver driver;
switch (browserType)
{
case "firefox":
driver = new FirefoxDriver();
break;
case "ie":
driver = new InternetExplorerDriver();
break;
case "chrome":
driver = new ChromeDriver();
break;
default:
throw new Exception("browserType: " + browserType + " not defined.");
}
// do test case
driver.get("http://www.google.com");
// ... and so on

how to test multiple browser(versions) with selenium and junit

I just discovered selenium - a great tool!
I plan to run/use selenium-ide generated junit4 code. But I need it to run with many browsers/web drivers.
Is there a junit/java-pattern for this use case? My first idea was to use #RunWith(Parameterized.class) and provide a List of WebDrivers (the parameter for the class - probably provided as an external file listing browsers and versions?!). Is this a good idea? Is it possible to provide a central #Parameters -method to be used by all my Selenium-tests?
What alternatives are there?
Probably it is possible to change the "Format" that Selenium exports to minimize manual changes?

Well, I do need to switch drivers from time to time, so I did this:
I initialize selenium related stuff in my own Class - called by name of the application and the driver is approached by the getters. When calling my class constructor, I use enum type of driver to initialize with:
private WebDriver driver;
public TestUI(Environment.DriverToUse drv){
switch (drv){
case CHROME:{
ChromeDriverService service = ChromeDriverService.createDefaultService();
File file = new File(TestUI.class.getResource("/chromedriver.exe").toURI());
System.setProperty(ChromeDriverService.CHROME_DRIVER_EXE_PROPERTY, file.getAbsolutePath());
ChromeOptions options = new ChromeOptions();
options.addArguments("--start-maximized");
driver = new ChromeDriver(service,options);
driver.manage().timeouts().implicitlyWait(30, TimeUnit.SECONDS);
break;
}
case FIREFOX:{
FirefoxProfile ffProfile = new FirefoxProfile();
ffProfile.setPreference("browser.safebrowsing.malware.enabled", false);
driver = new FirefoxDriver(ffProfile);
driver.manage().window().setPosition(new Point(0, 0));
java.awt.Dimension screenSize = Toolkit.getDefaultToolkit().getScreenSize();
Dimension dim = new Dimension((int) screenSize.getWidth(), (int) screenSize.getHeight());
driver.manage().window().setSize(dim);
driver.manage().timeouts().implicitlyWait(30, TimeUnit.SECONDS);
break;
}
public WebDriver getDriver(){
return driver;
}
of course my Environment class looks like this
public class Environment {
public enum DriverToUse {FIREFOX, CHROME};
// .. and some other stuff, because I need to test on different environments, so I store here Environment URL for example
And my test class looks something like this
#Before
public static final Environment.DriverToUse USED_DRIVER = Environment.DriverToUse.FIREFOX;
#Test
public void testVersionNumber() throws Exception{
TestUI testUI= new TestUI(USED_DRIVER);
WebElement version = testUI.getDriver().findElement(By.id("the Id of element"));
version.click();
//...
}

Use Selenium RC/Selenium Server. These come with the API's you will need to run remote tests in multiple browsers simply. Happy Hunting!

Check out the Selenide library. It's an open source wrapper for selenium that makes UI testing a breeze. Here's an example test.
#Test
public void userCanLoginByUsername() {
open("/login");
$(By.name("user.name")).setValue("johny");
$("#submit").click();
$(".loading_progress").should(disappear); // Waits until element disappears
$("#username").shouldHave(text("Hello, Johny!")); // Waits until element gets text
}