General

To markFailedAndStop or Not To markFailedAndStop?

Published December 9, 2018 | By Don Pedro

Based on the criteria of the original tests I set up several months ago, if an object was missing, that was a critical error and needed to be flagged. Since we were making heavy adjustments to the code base, anything missing needed to be checked. Things have changed quite a bit, so if an object is missing, should the situation be handled within an IF statement or does a missing object represent a much larger problem and the test has to be flagged as an Error?

This continues on from the idea of checking for the existence of an object without throwing an error. Using a Custom Keyword is was possible to suppress the error and not fill the log with False Positives.

Taking that a step further, it seems we could check if an object exists and continue on with the test if needed or exit out of the test completely.

Using the example from before, let’s say we want to count the number of Tasks a user has. If a user doesn’t have any Tasks to display, that’s not a critical error. They may not have made any. We could handle that within the IF statement and not try to count something that doesn’t exist.

Conversely, what if we tried to create a Task, but the New Task button was missing. Now we have a problem. We want to perform an action and piece of functionality is missing. And if the New Task button is missing, attempting to run code to fill in a series of input fields, then confirm it was saved won’t work and will lead to multiple errors. It would be better to mark the test as failed when we first notice the New Task button is missing.

We could handle both of those conditions within a single Custom Keyword. We can check for our object and if it’s a critical object, stop the test on failure. We can add a “critical” flag and handle both.

Our previous "verifyObjectPresent" method would be extended to create the following:

@Keyword
//Function to determine if an item exists on the page using WebUiCommonHelper.findWebElement
boolean verifyObjectPresent(String objectReference, boolean critical) {
try {
  WebUiCommonHelper.findWebElement(findTestObject(objectReference),5)
  return true;
} catch (Exception e) {
  if (critical==true){
  log.logWarning("The object with the name, " + objectReference + " was not found. Execution Halted.")
  KeywordUtil.markFailedAndStop("ERROR: The object with the name, " + objectReference + " was not found. Exiting Test.")
} else {
  log.logWarning("The object with the name, " + objectReference + " was not found.")
  return false;
}
}
}

We are now accepting “boolean critical” as a parameter. If the object is missing and this is true, we note the missing object and execute, “markFailedAndStop.” If it’s not a critical object, the name of the object is noted, but execution continues. Control then returns to the Test Case to handle a non-critical missing object.

A call to the Custom Keyword would look like the following:

boolean elementVisible=CustomKeywords.'commonCode.verifyObjects.verifyObjectPresent'('Blog/text-blogTitle', false)
if (elementVisible==true){
  blogTitle = WebUI.getText(findTestObject('Blog/text-blogTitle'))
  log.logWarning('Blog Title: ' + blogTitle)
}
log.logWarning("Test Complete")

If the blog title exists, get the text of that object. If it doesn’t, take the appropriate action based on the critical parameter. If it’s a critical object, mark the tests as failed and exit.

I like this approach because it removes a lot of redundant check, so I plan to make adjustments to replace my standard “verify” based calls with this method. I can now decide how to handle a missing object within a single method and if it’s a real problem, stop the code.

https://github.com/DonPedroQA/qajamboree/blob/master/Katalon/verifyObjectPresent.groovy

Checking for the presence of an object without throwing an error

Published December 8, 2018 | By Don Pedro

One of my goals has to been to make my tests less noisy and not generate errors in the log unless it’s really necessary. For example, if an object is missing from the page, that may or may not be an error. If a tab is missing, the user may not meet the criteria to make it display.

When checking for the existence of an object using the WebUI.verifyElementVisible() keyword, a lot of additional logging that can be misleading. For example, this code block will execute as expected, but generates 2 entries for the Failed log.

@Keyword
//Function to determine if an item exists on the page
boolean verifyObjectVisible(String objectReference) {
try {
  WebUI.verifyElementPresent(findTestObject(objectReference),5)
  return true;
} catch (Exception e) {
  log.logWarning("The object with the name, " + objectReference + " was not found. Exiting Test.")
  return false;
}
}

There is a cleaner way to check for an object using the WebUiCommonHelper.findWebElement command, which is part of the,

import com.kms.katalon.core.webui.common.WebUiCommonHelper library and performs the same function. I like it better because it allows for better handling of errors and logging.

Below is a Custom Keyword that wraps the use of the findWebElement statement within a Try/Catch block to determine if the object exists. The benefit is suppressing an error if the object doesn’t exit.

@Keyword
//Function to determine if an item exists on the page using WebUiCommonHelper.findWebElement
//This does not throw an exception error to the log so it looks cleaner
boolean verifyObjectPresent(String objectReference) {
try {
  WebUiCommonHelper.findWebElement(findTestObject(objectReference),5)
  return true;
} catch (Exception e) {
  return false;
}
}

A call to the Custom Keyword would then look like:

//Check if there are existing notes on the page and count how many
boolean elementVisible=CustomKeywords.'tools.commonCode.verifyObjectPresent'(katalonObject)
if (elementVisible==true){
  originalNumberOfTasks=CustomKeywords.'tools.commonCode.countRowsPerPage'(xpath)
} else {
  log.logWarning('Customer has no Tasks available at this time')
}

This gives control within the code whether or not to display an error. In looking at the code above, if no Tasks are displayed on the page, a notification message is display, not an error. The test hasn’t failed and there is no problem with the site, a user may not have created an tasks which is why they don’t exit.

The verifyElementPresent or verifyElementVisible will perform the same task, but will log multiple errors if an element isn’t found, even when wrapped in a Try/Catch block.

More Katalon Studio and Automation to come

Published December 8, 2018 | By Don Pedro

Although I haven’t posted anything for some time now, I haven’t abandoned automation or Katalon Studio. Quite the opposite. I’ve completed several Udemy courses for Groovy, Java, automation and SQL. Along with that, I’ve used this slower time of the year to review all my code and make improvements so that my tests are more resilient, more modular, more scalable and less “noisy.”

There have also been quite a few changes to Katalon Studio. There is the ability to import and export Custom Keywords, so moving more code to that section of the project seems ideal. The Execution Profiles are a bit easier to work with, so I’ve moved more data out of the tests themselves and set them up as Global Variables. Cucumber BDD is available as well, and while I’m not using it currently, I did take a Udemy course on it so it might be something I try to incorporate into my project in the future.

Of course, after running the tests hundreds of times, I’ve noted areas of improvement such as better handling for transient conditions like an item not being found. And with some new found experience, I see how a test that performs one task can be parameterize, turned into a method and perform the same task multiple times with different input.

I’m quite pleased with the improvements and will share some of my discoveries. To that end, I have created a rather unexciting Github repository where I have shared some code snippets. It’ll get better over time. At least I hope so.

You can access it at: https://github.com/DonPedroQA/qajamboree

One ReplaceAll Statement to Rule Them All!

Published August 9, 2018 | By Don Pedro

Well, sort of.

In reviewing my code to make it more efficient, I noticed I have several variations on the .replaceAll() statement. In some case I'm trying to remove the $ and , from a value. In others I'm trying to remove the newline and CRLF characters. For another the goal was to remove the ( and ) from a value. And yet another was for the \ in a date. On top of this, I called replace several times in a row in order to remove specific characters or a string.

While those work, there is a way to accomplish this in a single call and return just the number I'm really looking for.

For example, I have the value $5,573,127 and want just the integer value. That can be done with:

tempText=tempText.replaceAll("[^0-9]","").toInteger()

Everything that isn't a number will be removed and I'll have 5573127 as an Integer. This includes ()[]$,. and -.

If the value is -$5,573,127.35 and I need to keep the decimal or the negative, it's a simple change to add them at the end as values to keep.

tempText=tempText.replaceAll("[^0-9.-]","").toDouble()

Now, if there is additional text, such as "Showing 1 to 8 of 2,056 entries", that can be handled by passing the text string to remove to the .replaceAll(). We then use the pipe | delimiter to indicate there is more conversion we wish to make.

tempText="Showing 1 to 8 of 2,056 entries"
removeText="Showing 1 to 8 of"
tempText=tempText.replaceAll(removeText + "|[^0-9]","").toInteger()
or
tempText=tempText.replaceAll("1 to 8|[^0-9]","").toInteger()

We now have 2056 as the number of entries. We specifically need to remove the 1 to 8 text or else the 1 and 8 will be retained giving us 182056 as a result, which is incorrect.

This will save quite a few lines of codes and make the .replaceAll() much more consistent across my projects. It pretty much won't matter what I'm reading, it'll be easy to get back just the number with one command.

Call the same Test Case twice with a Custom Keyword

Published August 6, 2018 | By Don Pedro

I recently had cause to run a series of tests where I needed to call the same Test Case multiple times, something Katalon Studio does not inherently support. For my test, I wanted to enter several pieces of information into a form, then perform the same validation steps to check table data. The validation steps are the same each time, so I wanted to reuse the test case. No such luck, Katalon Studio does not allow you to add the same Test Case name to the Test Suite.

My first thought was to duplicate each Test Case multiple times, putting a number in front. That seemed awkward and wasteful. I then noted I could go into the Test Suite XML file itself and repeatedly add the same Test Case name multiple times.

While interesting, each Test Case is tied to a GUID, so I would need to make a fake one and that's by no means to maintain. And if someone makes a change to the Suite, most likely everything will come apart like a house of cards.

Turns out I can do what I want by creating a Custom Keyword that calls all my Test Cases. My Test Case will prepare all the data by populating the fields, then at the end of Test, there will be a call to the Custom Keyword, which will run all the tests. It sounds a little odd, but this is at least easy to maintain and is far cleaner than hacking together an XML file.

The structure of the Custom Keyword looks like this:

@Keyword
    def verifyData(){
        WebUI.callTestCase(findTestCase('Read Table Values'), [:], FailureHandling.STOP_ON_FAILURE)
        WebUI.callTestCase(findTestCase('Read Savings Totals'), [:], FailureHandling.STOP_ON_FAILURE)
        etc
        etc
        etc
    }

The Test Suite will contain calls to the Test Cases that populate the data. The Test Cases will each call the Keyword to validate the data. It's not the most elegant of solutions, but it certainly works and gets the job done.

Sans Pantalones

Wouldn't you prefer a good game of chess?

Recent Posts

Favorite Games

Search

To markFailedAndStop or Not To markFailedAndStop?

Other articles of interest:

Checking for the presence of an object without throwing an error

Other articles of interest:

More Katalon Studio and Automation to come

Other articles of interest:

One ReplaceAll Statement to Rule Them All!

Other articles of interest:

Call the same Test Case twice with a Custom Keyword

Other articles of interest:

Recent Comments

Meta