Getting "Expected to have result"error after completion of a step

[anooppc07]

Below is my test.But am getting the error ""Expected to have result"" just after typing in username.Can you help me here.

test.only("executes query", async ({ page }) => {
await page.goto("https://www.saucedemo.com/");

const headerText="Playwright";
await auto(`type "${headerText}" in the username`, { page, test },{model: 'gpt-3.5-turbo-1106', debug: true});
await auto(`type "Hello" in the password`, { page, test },{model: 'gpt-3.5-turbo-1106', debug: true});
await page.pause();

});

[SmartALB]

Here the same.

[azertyalex]

Same issues here

[AbhishekAutomationTest]

same here

[srsantmyer]

I bumped into this today while looking to try out auto-playwright on the cheap with GPT-3.5. I was able to get around the issue as follows:

in node_modules/auto-playwright/dist/completeTask.js at line 23 there's an if statement. The GPT-3.5 final role === "assistant" response message doesn't appear to contain a value that satisfies this predicate:
message.function_call?.name.startsWith("result") . It just doesn't contain a message.function_call value.

I updated the if as follows:

        if (message.role === "assistant" &&
            (message.function_call?.name.startsWith("result"))
            || message.content?.indexOf(" successfully ") > -1) {
            lastFunctionResult = JSON.parse(message.function_call?.arguments?? 'true');
        }

GPT-3.5 has been returning a message like "I have successfully ..." or "... as been completed successfully...". I'm not sure why it isn't the same every time, but I've seen a couple variations. In my cursory testing this worked, but YMMV. I'm not sure how it will hold up to more robust situations.