See Your Function In Action

This section shows an example of a mock application similar to a real-world setup. The specific functionality isn't important; instead, it demonstrates how you’d use Func Runner in a real application. In production, this mock app could represent something like a REST API.

Start Your First Function

Assuming you completed the previous section, run your Func Runner Application, and it will wait for incoming function requests.

python main.py

[info     ] Registering function: 'greet'
[info     ] Starting Func Runner application...
[info     ] Available functions: greet
[info     ] Updating assistant functions... assistant_id=asst_tliJoe0qJpbcZJsODeolbVJF
[info     ] Generating function spec for function: 'greet'
[debug    ] Generated function spec for function: 'greet' spec={'type': 'function', 'function': {'name': 'greet', 'description': 'generates a greeting with a given name', 'parameters': {'type': 'object', 'properties': {'name': {'type': 'string', 'description': 'The name parameter.'}}, 'required': ['name'], 'additionalProperties': False}, 'strict': True}}
[info     ] Finished updating assistant functions... assistant_id=asst_tliJoe0qJpbcZJsODeolbVJF

Create a Mock Application

This is an example application that sends a request that should invoke function-calling.

import os
from time import sleep

import openai
from dotenv import load_dotenv

# Load environment variables
load_dotenv()

# Initialize the OpenAI client with a proxy base URL and API key
client = openai.Client(
    base_url="https://proxy.funcrunner.com/v1",
    api_key=os.getenv("FUNCRUNNER_API_KEY")
)

message_content = "Greet 5 different influential philosophers."
print(message_content)

# Create a new thread with an initial user message
thread = client.beta.threads.create(
    messages=[
        {
            "role": "user",
            "content": message_content
        }
    ]
)

# Run the assistant on the thread to process the messages
run = client.beta.threads.runs.create(thread_id=thread.id, assistant_id="asst_tliJoe0qJpbcZJsODeolbVJF")

# Wait for the run to complete
while not run.status == "completed":
    sleep(5)
    run = client.beta.threads.runs.retrieve(run_id=run.id, thread_id=thread.id)
    print(f"Current run status: {run.status}")

# Retrieve the messages on the thread. 
messages = client.beta.threads.messages.list(thread_id=thread.id)

print(f"\n{messages.data[0].content[0].text.value}")

You should see results similar to the following:

Greet 5 different influential philosophers.
Current run status: requires_action <-- Func Runner takes over here.
Current run status: in_progress  <-- Func Runner submitted your function results back to OpenAI on your behalf.
Current run status: completed 


- Hello, Aristotle
- Hello, Confucius
- Hello, Immanuel Kant
- Hello, Simone de Beauvoir
- Hello, Friedrich Nietzsche

Did you notice something interesting in our mock application? It uses the official OpenAI library! So adopting Func Runner is incredibly simple as it was designed to allow you to keep using the official OpenAI libraries regardless of the language you choose! 😎

If you look back at your Func Runner application logs you will see additional logs related to this request.

[info     ] Message dequeued               correlation_id= message_id=cade79a1-f13a-431d-ad93-3c0dd9f62373
[info     ] Processing queue message       correlation_id= message_id=cade79a1-f13a-431d-ad93-3c0dd9f62373
[debug    ] Fetching run data from URL: https://proxy.funcrunner.com/v1/threads/thread_DIx9dh9wd9o60mLkxC7pV1Jr/runs/run_BknfyirUdjj991uZgHUUqAND correlation_id= message_id=cade79a1-f13a-431d-ad93-3c0dd9f62373 run_id=run_BknfyirUdjj991uZgHUUqAND
[info     ] Processing function call request function_name=greet tool_call_id=call_rpbwIh1OUmKpn1muIzJvyWg5
[info     ] Processing function call request function_name=greet tool_call_id=call_8W2i6cdtOJhpYebXNiToRkyR
[info     ] Processing function call request function_name=greet tool_call_id=call_bRbKehTgtUFMVNM5lGSGIre4
[info     ] Processing function call request function_name=greet tool_call_id=call_fjxLc2Z9Gb6pJqTgfeitmo38
[info     ] Processing function call request function_name=greet tool_call_id=call_lo4wSY3OS6JVp9xNFcpDgqFS
[info     ] Executing function             arguments={'name': 'Aristotle'} correlation_id= function_name=greet
[info     ] Executing function             arguments={'name': 'Confucius'} correlation_id= function_name=greet
[info     ] Executing function             arguments={'name': 'Immanuel Kant'} correlation_id= function_name=greet
[info     ] Executing function             arguments={'name': 'Simone de Beauvoir'} correlation_id= function_name=greet
[info     ] Executing function             arguments={'name': 'Friedrich Nietzsche'} correlation_id= function_name=greet
[info     ] Submitting function results    correlation_id=thread_DIx9dh9wd9o60mLkxC7pV1Jr run_id=run_BknfyirUdjj991uZgHUUqAND
[info     ] Successfully submitted function results correlation_id=thread_DIx9dh9wd9o60mLkxC7pV1Jr run_id=run_BknfyirUdjj991uZgHUUqAND
[info     ] Deleting message from the queue correlation_id= message_id=cade79a1-f13a-431d-ad93-3c0dd9f62373
[info     ] Successfully deleted queue message correlation_id= message_id=cade79a1-f13a-431d-ad93-3c0dd9f62373

Last modified: 17 December 2024