Produce Better Refactored Code with AI by Giving Mock Data
When you ask an AI to refactor or write a function for you it will have limited knowledge of what data you are using.
By providing it some expected input to a function you want it to write or refactor, it will do a much better job at writing code that gets the job done. Of course you will still want to verify that your code works as you expect.
[00:00] I'm going to select the body of this function, hit command k, and then go over to the input I used to pass into this function. So this, and I'll select it, copy it, and I'm going to tell it this isn't properly parsing out the appearance variable. And then I'll go ahead and paste in what's essentially the input for that [00:19] function. So once I hit enter, it'll start working on it. You can see the dots going down there. It'll show me a diff of what it wants to do. I'm gonna accept this, and then you'll see that before it was only splitting on semicolons. And our input for appearance, the first line here, it would have split here and grabbed both of these lines. And [00:39] now because it knew more about the input, it decided it also needs to split here and here so that it grabs the curlies out, and it only focuses on lines with the semicolons. So when you're using AIs to fix your functions, it's super valuable to have essentially mock data. This is kind of like running an AI unit test against this, and then you'll still want to test [00:59] it yourself. But there's no way that AI could have solved this without me providing the input for the function.