AI DIAL Python SDK

• Overview
• Environment Variables
• Usage
  • Echo application example
    • Run
    • Check
  • Endpoints
• Development
  • Development Environment
  • Setup
  • IDE configuration
  • Lint
  • Test
  • Git hooks
  • Clean
  • Build
  • Publish

Overview

Framework to create applications and model adapters for AI DIAL.

Applications and model adapters implemented using this framework will be compatible with AI DIAL API that was designed based on Azure OpenAI API.

Environment Variables

Logging-related environment variables (log level, console format, and trace/span correlation) are documented in docs/logging.md.

Usage

Install the library using pip:

pip install aidial-sdk

Echo application example

The echo application example replies to the user by repeating their last message:

# Save this as app.py
import uvicorn

from aidial_sdk import DIALApp
from aidial_sdk.chat_completion import ChatCompletion, Request, Response


# ChatCompletion is an abstract class for applications and model adapters
class EchoApplication(ChatCompletion):
    async def chat_completion(self, request: Request, response: Response) -> None:
        # Get last message (the newest) from the history
        last_user_message = request.messages[-1]

        # Generate response with a single choice
        with response.create_single_choice() as choice:
            # Fill the content of the response with the last user's content
            choice.append_content(last_user_message.text())


# DIALApp extends FastAPI to provide a user-friendly interface for routing requests to your applications
app = DIALApp()
app.add_chat_completion("echo", EchoApplication())

# Run built app
if __name__ == "__main__":
    uvicorn.run(app, port=5000)

Run

python3 app.py

Check

Send the next request:

curl http://127.0.0.1:5000/openai/deployments/echo/chat/completions \
  -H "Content-Type: application/json" \
  -H "Api-Key: DIAL_API_KEY" \
  -d '{
    "messages": [{"role": "user", "content": "Repeat me!"}]
  }'

You will see the JSON response as:

{
    "choices":[
        {
            "index": 0,
            "finish_reason": "stop",
            "message": {
                "role": "assistant",
                "content": "Repeat me!"
            }
        }
    ],
    "usage": null,
    "id": "d08cfda2-d7c8-476f-8b95-424195fcdafe",
    "created": 1695298034,
    "object": "chat.completion"
}

Endpoints

Every deployment is exposed under two base paths:

The following endpoints are served under each of them:

• POST <base path>/chat/completions
• POST <base path>/rate
• POST <base path>/tokenize
• POST <base path>/truncate_prompt
• GET <base path>/configuration
• POST <base path>/embeddings

request.deployment_id is resolved from the first of the following sources that is set:

1. the X-DIAL-OVERRIDE-NAME header,
2. the deployment name from the path, for the /openai/deployments/{deployment_name} requests,
3. the X-DIAL-DEPLOYMENT-ID header, which DIAL Core sets when it calls the /openai/v1 endpoints.

An /openai/v1 request with none of the headers set fails with 500 Internal Server Error.

Development

Development Environment

This project requires Python ≥3.11 and Poetry ≥2.1.1 for dependency management.

Setup

1. Install Poetry. See the official installation guide.

2. (Optional) Specify custom Python or Poetry executables in .env.dev. This is useful if multiple versions are installed. By default, python and poetry are used.

   POETRY_PYTHON=path-to-python-exe
   POETRY=path-to-poetry-exe
   

3. Create and activate the virtual environment:

   make init_env
   source .venv/bin/activate
   

4. Install project dependencies (including linting, formatting, and test tools):

   make install
   

IDE configuration

The recommended IDE is VSCode. Open the project in VSCode and install the recommended extensions. VS Code is configured to use the Ruff formatter.

Alternatively you can use PyCharm that has built-in Ruff support.

Lint

Run the linting before committing:

make lint

To auto-fix formatting issues run:

make format

Test

Run unit tests locally for available python versions:

make test

Run unit tests for the specific python version:

make test PYTHON=3.11

Git hooks

You may optionally install Git hooks that will automatically run the linting step on Git push. You only need to do it once for the given repository.

make install_git_hooks

Clean

To remove the virtual environment and build artifacts run:

make clean

Build

To build the package run:

make build

Publish

To publish the package to PyPI run:

make publish