Skip to content

maestromaximo/OpenAiUnlimitedFuncWrapper

Repository files navigation

PyPI License Documentation Status

OpenAI Unlimited Function Wrapper

The openaiunlimitedfuncwrapper is a Python package that simplifies interaction with the OpenAI API, providing easy access to various models of GPT, including conversational capabilities, dynamic function calling, and pseudo-function execution to elicit specific responses based on parameter modification.

Features

  • Single-question querying to any GPT model and receiving a response.
  • Engaging in a conversation with context management.
  • Dynamically adding callable functions within the code.
  • Forcing execution of real or pseudo-functions to steer responses.
  • Automatic and manual creation of JSON schemas for function descriptions.
  • Setting the OpenAI API key via code for environment preparation.

Installation

To install openaiunlimitedfuncwrapper, simply run:

pip install openaiunlimitedfun

Setting Up Your OpenAI API Key

Before you start using the package, you need to set your OpenAI API key. You can do this by running:

from openaiunlimitedfun import set_openai_api_key

set_openai_api_key('your-api-key-here')

This will create or append to a .env file in your current directory, storing your API key.

Managing Available Functions

To make custom functions available for the OpenAI API to call during a conversation, use the manage_available_functions function:

from openaiunlimitedfun import manage_available_functions

# To save current module's functions
manage_available_functions(retrieve=False)

# To retrieve available functions
functions = manage_available_functions()

Adding Functions to the Function List

If you want to add specific functions to be accessible during the conversation, use manage_function_list:

from openaiunlimitedfun import manage_function_list

# To add a function to the list
manage_function_list(function_to_add='your_function_name')

# To retrieve the list of functions
function_list = manage_function_list(retrieve=True)

  • After running the above, chat_context_function_bank will have the functions availible to run

Generating JSON Schemas for Functions

You can create JSON schemas for your functions automatically or manually. This can be used to generate function descriptions for use within the wrapper.

Automatic JSON Schema Generation

Automatically generate a JSON schema based on user input:

from openaiunlimitedfun import create_json_autoagent

schema = create_json_autoagent('Describe a function that calculates the sum of two numbers.')
print(schema)

Manual JSON Schema Creation

Manually create a JSON schema through an interactive prompt:

from openaiunlimitedfun import create_function_json_manual

create_function_json_manual()
# Follow the interactive prompts to create your function JSON schema.

Usage Examples

Single Question

Query a single question and get a response:

from openaiunlimitedfun import single_question

response = single_question("What is the capital of France?")
print(response)

Conversational Context

Engage in a conversation with the ability to maintain context:

from openaiunlimitedfun import chat_context_function_bank

question = "Who wrote the play Hamlet?"
context = []  # This should be a list of previous messages if you have them

response, updated_context = chat_context_function_bank(question, context)
print(response)

Pseudo-Function Execution

Force the execution of a pseudo-function to get a desired response:

from openaiunlimitedfun import single_turn_pseudofunction

# Define a pseudo-function
pseudo_function = {
    "name": "calculate_sum",
    "parameters": {
        "number1": 5,
        "number2": 3
    }
}

# Use the pseudo-function in a prompt
response = single_turn_pseudofunction("What is the sum of the numbers?", pseudo_function)
print(response)

Contributing

Contributions are welcome! Please feel free to submit pull requests, report bugs, and suggest features.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Build

  • If you want to run a new pip build make sure to actiavte an enviroment, run pip install -r requirement.txt
  • Then be sure to run Remove-Item -Recurse -Force build, dist, *.egg-info if you had run a build before, else it would return an error
  • Then run python setup.py sdist bdist_wheel
  • If wanted you can then run a pip of the wheel file path ".whl" under the dist folder that was just created

Star History

  • If you liked the project star it to help it reach other people :)
  • More people = more possible features, lets grow together
Star History Chart

About

This is a project that simplifies the use of Openai API's with the capability of easily implementing an unlimited amount of functions

Topics

Resources

License

Code of conduct

Security policy

Stars

Watchers

Forks

Packages

No packages published

Languages