With ChatGPT, you can communicate and instruct it to say virtually anything you want, as long as it remains respectful. It provides original and relevant answers to queries and usually explains the reasoning behind its responses. Additionally, ChatGPT can remember previous statements and engage in logical discussions. People have utilized ChatGPT for a wide range of tasks, from composing beer-themed folk songs to answering questions about the science of soil.
ChatGPT and Automated Testing
One interesting aspect of ChatGPT, especially for software developers, is its ability to generate suggested code based on natural language prompts. It can produce code in various programming languages and leverage built-in tools specific to those languages.
Naturally, one might wonder if ChatGPT can be used to generate code for automated testing. It has been demonstrated that ChatGPT can indeed write automation code, such as Selenium code in multiple languages, better than some human testers.
However, while ChatGPT is capable of generating functional code, it currently lacks the ability to understand the specific context of the website being tested, including its version and intricacies. So, while it has some impressive capabilities, ChatGPT should be seen as a complement to testing rather than a complete replacement for human testers.
Understanding Low-Code Testing
Low-code testing is a methodology that allows individuals to write code even if they have limited or no prior coding experience. It enables writing code using plain English or through the use of visual, drag-and-drop tools. Low-code testing options make it easier for development teams to create automated test code, reducing the dependency on highly technical skills.
By utilizing natural language, ChatGPT facilitates the creation of test cases that require minimal or no coding. In contrast, template-based models often rely on specific grammar patterns or keywords. ChatGPT excels at generating scripts, classes, and methods for test automation while maintaining the flexibility of natural language.
Low-Code Languages: Cucumber
Among the languages and tools ChatGPT can generate code for, Cucumber stands out as one of the most impressive. Cucumber is a testing tool that adopts behavior-driven development. It allows writing test scenarios in plain English using keywords like Given, When, and Then, which are then connected to corresponding code in step definitions.
Cucumber's strength lies in its ability to combine the natural language description of a test scenario with the actual code that executes it, making it easier for testers with limited coding knowledge to understand the relationship between test purpose and code. ChatGPT's power is evident here, as it can generate code that closely resembles natural language.
ChatGPT and Cucumber Test Cases
An example showcases ChatGPT's proficiency in writing code. By making a simple request, ChatGPT can generate both test cases and step outlines in Cucumber. It can infer the script to test the search feature of a website, such as Google, which is typically a critical part of the website. In this example, the "q" element in Google's search form is correctly identified.
Creating User-Friendly Scripts
While ChatGPT can usually generate code, it's important to note that not all websites are as straightforward as Google. Although it got the "q" element correct in the previous example, it may not always be accurate. When using ChatGPT, it becomes necessary to find and modify the element locators in the generated code. However, this task can be simplified by separating the code
for the About page from the test case code. By utilizing a page object model, you can modify the code easily if there are changes in the website's layout or component locators between tests.
This approach simplifies the process of updating the test script. In the following example, we create a general test for a website's login page using Cucumber and Python. We instruct ChatGPT to utilize a page object model and class variables to locate elements.
ChatGPT accurately generates the LoginPage object with appropriate methods, such as adding a username, and the element locators are represented as class variables.
Towards the end of the example, the step descriptions demonstrate how to interact with the website using the elements on the login page. You can automate this process with ChatGPT or manually create test cases to test the website and obtain functional test code.
Making Changes and Fixing Code with ChatGPT
When examining the step code closely, you'll notice that test input values, such as the URL, login, and password, are hard-coded. It's unlikely that the website you're testing will have the same values as "http://www.example.com/login" or use "username" and "password" as login credentials.
Fortunately, Cucumber allows you to retrieve variables from the feature description and use them in the test cases, rather than hard-coding them. But what if you're unfamiliar with how to implement this or you simply prefer not to modify the code manually in multiple places? ChatGPT can assist you in resolving this issue. You can instruct ChatGPT to make the necessary changes in the generated code.
Instead of hard-coding potentially incorrect values, the updated step methods now read in the input values as requested. One of the strengths of ChatGPT is its ability to understand and fulfill your specific instructions. It does an excellent job of listening and executing the tasks you assign.
In Conclusion
ChatGPT is an impressive natural language model with tremendous potential. It offers various possibilities and is likely to become an excellent low-code testing option for many scenarios. While it can be a valuable tool in the testing process, it's important to note that using ChatGPT requires a good understanding of both the programming language and the application you're testing. However, ChatGPT opens up new possibilities for transforming regular language into functional code, a capability that was previously unavailable with other models.
No comments:
Post a Comment