Advance Level Screenshot capturing using Selenium Webdriver and TestNG Listners

In my opinion, you should avoid the following:

• using static drivers (because using static variables is a bad practice)
• using static methods for taking screenshots (because using static methods is a bad practice)
• taking screenshots in the test methods (because you will duplicate code in lots of tests)
• taking screenshots in the @After methods (because you dont need to do this)

I prefer to take the screenshots using a TestNG listener as follows.

1. create a class for taking the screenshots

Every time you need to take a screenshot, create an object for this class and use the capture method.

The capture method will take the screenshot and save it in the screenshots folder.

package framework;


import java.io.File;
import java.io.FileOutputStream;
import org.openqa.selenium.OutputType;
import org.openqa.selenium.TakesScreenshot;
import org.openqa.selenium.WebDriver;
 
public class Screenshot {
 private WebDriver driver; 
 private final String folderPath = "./src/test/resources/screenshots/";
  
 public Screenshot(WebDriver driver) {  
  this.driver = driver;
  validateFolderExists();
 }


 private void validateFolderExists() {
  File folder = new File(folderPath);
  if (!folder.exists())
   folder.mkdir();
 }
  
 public void capture(String fileName) {  
  try {             
            String name = folderPath + fileName + ".png";                          
   FileOutputStream file = new FileOutputStream(name);
   file.write(
    ((TakesScreenshot) driver)
     .getScreenshotAs(OutputType.BYTES));
   file.close();        
  } 
  catch (Exception ex) {
   throw new RuntimeException("cannot create screenshot;", ex);
  }       
 }
}

2. Create a TestNG listener that takes the screenshot if the test method fails or is skipped

The listener gets the driver object from the test class through reflection.

package framework;


import java.lang.reflect.Field;
import org.openqa.selenium.WebDriver;
import org.testng.ITestContext;
import org.testng.ITestListener;
import org.testng.ITestResult;
import org.testng.Reporter;
import framework.BrowserDriver;
import framework.Screenshot;


public class CustomTestListener implements ITestListener {


 @Override
 public void onTestFailure(ITestResult result) {       
  String fileName = "failed_" + result.getName();  
  new Screenshot(driver(result)).capture(fileName);  
 }


 @Override
 public void onTestSkipped(ITestResult result) {    
  String fileName = "skipped_" + result.getName();  
  new Screenshot(driver(result)).capture(fileName);
 }
 
 @SuppressWarnings("unchecked")
 private WebDriver driver(ITestResult testResult) 
 {
  BrowserDriver browserDriver = null;
  try {
   Classextends ITestResult> testClass = 
    (Class extends ITestResult>) 
     testResult.getInstance().getClass();
 
   Field driverField = testClass.getDeclaredField("driver");
   browserDriver = (BrowserDriver)driverField
        .get(testResult.getInstance());    
  } 
  catch (SecurityException | 
               NoSuchFieldException | 
               IllegalArgumentException | 
               IllegalAccessException e) { 
 
   throw new RuntimeException(
    "couldnt get the driver from the test class!");
  }


  return browserDriver.wrappedDriver();
 }


 @Override
 public void onTestFailedButWithinSuccessPercentage(ITestResult result) {
 }
 
 @Override
 public void onStart(ITestContext context) {
 }
 @Override
 public void onFinish(ITestContext context) {
 }
 
 @Override
 public void onTestStart(ITestResult result) {  
 }
 
 @Override
 public void onTestSuccess(ITestResult result) {
 }
}

3. add the listener to the test class using the Listener annotation

@Listeners(CustomTestListener.class)
public class SiteTests {  
 public BrowserDriver driver; 
  …………………………………..

Advantages of this approach:

1. no static driver
2. no static methods for taking the screenshot
3. no “take screenshot” code in the test methods
4. no “take screenshot” code in the @After methods

Disadvantages:

1. using reflection to get the driver; complicated code for this purpose
2. using null in the test listener

GUEST AUTHOR

Alex Siminiuc