How do I invoke an OpenAI function using Python 3.x?

Below is an example of how you might call your generated function in Python 3.x using the OpenAI API. In this example, we use the new ChatCompletion API with function calling:

PYTHONCopied
import openai
import json

# Replace with your actual OpenAI API key
openai.api_key = "YOUR_API_KEY_HERE"

# Define the function schema as provided
function_definition = {
    "name": "generate_jokes",
    "description": "Generate 3 jokes based on input text",
    "strict": True,
    "parameters": {
        "type": "object",
        "required": [
            "input_text",
            "number_of_jokes"
        ],
        "properties": {
            "input_text": {
                "type": "string",
                "description": "The text or context based on which jokes will be generated"
            },
            "number_of_jokes": {
                "type": "number",
                "description": "The number of jokes to generate; default is 3"
            }
        },
        "additionalProperties": False
    }
}

# Prepare the conversation messages; you can adjust the content as needed
messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {
        "role": "user",
        "content": "Can you tell me a few jokes about programming?"
    }
]

# Call the ChatCompletion endpoint with the function definition
response = openai.ChatCompletion.create(
    model="gpt-4-0613",  # or another model supporting function calling
    messages=messages,
    functions=[function_definition],
    function_call="auto"  # automatically decides to call the function based on the conversation
)

# The response message may include a 'function_call' key if a function is called
message = response.choices[0].message

if "function_call" in message:
    function_call = message["function_call"]
    # The function_call contains:
    # - name: the function name to be called
    # - arguments: parameters in JSON string format, which can be parsed
    print("Function call requested:")
    print("Function Name:", function_call.get("name"))
    print("Arguments:", json.loads(function_call.get("arguments", "{}")))
else:
    # If no function call was made, just print the assistant's reply.
    print("Assistant reply:")
    print(message.get("content"))

Explanation

• First, we define the function with its schema as given in your JSON.
• We then build a conversation with messages. Here, the system and user roles are used to set context.
• The openai.ChatCompletion.create() call includes:
  • The conversation messages,
  • The list of functions (in this case, one function defined by your schema),
  • And the function_call parameter set to "auto".
• If the assistant determines that the function should be called based on the conversation, a function_call key will appear in the response. We print the function name and the parsed arguments.
• Otherwise, the assistant's text-based reply is printed.

This code snippet provides a basic example of how to integrate function calling using the OpenAI API in a Python 3.x environment. Adjust the conversation, model name, and API key as needed for your application.