NAME

OpenAITests

DESCRIPTION

Using this modul is equivalent to:

use Data::Dumper;
use JSON;
use OpenAPI::Client::OpenAI;
use Test::Most;
use Test::RequiresInternet;
use utf8;
use experimental 'signatures';

run_test_cases

my @test_cases = (
    {
        method      => 'createCompletion',
        description => 'createCompletion with an instruct model',
        params      => {
            model       => 'gpt-3.5-turbo-instruct',
            prompt      => 'What is the capital of France?',
        },
        expected_response => qr{\bParis\b},
    },
    {
        method      => 'createCompletion',
        description => 'createCompletion with an instruct model, using `stop`',
        params      => {
            model       => 'gpt-3.5-turbo-instruct',
            prompt      => 'What is the capital of France?',
            stop        => 'aris',
        },
        expected_response => qr{P$},
    },
);

run_test_cases( \@test_cases );

Runs a set of test cases. The following keys are used for each test case:

• method

  The name of the OpenAI method to call.

• description

  What are we testing? This will be used as the name of the test. If omitted, we use the method argument instead, but it's recommended that you not omit this.

• params

  The parameters which will be passed to the method.

• expected_response

  A regular expression that should match the response outout.

• against

  An optional subreference. It will be passed the OpenAI response as its only argument. Typically, we test the expected_response against $response->{choices}[0]{message}, but you may want to check something else in the response. For example, to test expected_response against the entire response:

  against => sub ($response) {$response},