genai-rest-builder 0.2.1

A tool to generate python code with GenAI RestAPIs.

GenAI REST Builder

The GenAI REST Builder is a tool for rapidly creating RESTful services around AI models, allowing you to define service configurations in a YAML file and generate a FastAPI-based project structure to deploy AI-powered endpoints.

Key Features

• Service Configuration with Custom Prompts: Easily define each service's prompt, model configuration, and provider-specific details in YAML format.
• Cloud Provider Compatibility: Supports major providers, including AWS and Azure, to integrate directly with cloud-based AI services.
• Automated Code and Endpoint Generation: Generates FastAPI service endpoints, configuration files, and utility scripts based on YAML specifications.

Installation

1. Install the GenAI REST Builder Tool:

   Begin by installing the genai-rest-builder package:

   pip install pyyaml genai-rest-builder
   

2. Generate Project Structure:

   Use the command below to generate the project structure with REST APIs for all defined services based on the YAML configuration file:

   genai-rest-proj-build
   

   This reads from prompt_service.yaml (to be created by the user with service definitions) and generates the required project files, configurations, and the main FastAPI application.

3. Install Project Dependencies:

   After generating the project structure, install the necessary dependencies:

   pip install -r requirements.txt
   

4. Run the Server:

   Launch the FastAPI server with serve_app.py:

   python serve_app.py
   

   By default, the server runs on localhost:8080 unless overridden in the .env file.

Example YAML - prompt_service.yaml Configuration File

Below is an example structure for defining services in the YAML configuration file:

PromptServices:
  - <servicename>:
      prompt: <prompt template>
      model:
        provider: [aws, azure]
        modelId: <model id>
        temperature: <temperature>
        maxTokens: <maximum tokens>
        apiVersion: <api version>  # Required for Azure provider

YAML Configuration Details

• servicename: Unique identifier for the service.
• prompt: Template text for the AI model’s prompt.
• model: Contains model details:
  • provider: Cloud provider (aws or azure).
  • modelId: Identifier for the specific model.
  • temperature: Controls the randomness of the model's output.
  • maxTokens: Maximum tokens to be generated by the model.
  • apiVersion: (Azure only) Specifies the API version of the Azure model.

Project Structure

After running the project build command, the following structure will be created:

.  
├── service_chains/  
│   ├── base_chains/                     # Base and provider-specific chain definitions  
│   │   ├── base_chain.py                # Abstract base class for chains  
│   │   ├── aws_chain.py                 # AWS-specific chain implementation  
│   │   ├── azure_chain.py               # Azure-specific chain implementation  
│   │   ├── __init__.py                  # Exports for base_chains module  
│   ├── <servicename>_chain.py           # Service-specific chain implementation  
│   ├── utils.py                         # Utility functions for service configurations  
│   ├── __init__.py                      # Package initialization file for service_chains  
├── serve_app.py                         # Main FastAPI application  
├── .env                                 # Environment configuration file with host/port settings  
├── requirements.txt                     # Project dependencies file  
├── prompt_service.yaml                  # YAML configuration file (user-defined)  

Accessing API Documentation

Once the server is running, OpenAPI documentation is available at:

http://<GENAI_SERVER_HOST>:<GENAI_SERVER_PORT>/docs

This page provides an interactive interface to test and explore the API endpoints.

Invoking Services

Each service can be accessed via a unique REST API path based on the service name:

http://<GENAI_SERVER_HOST>:<GENAI_SERVER_PORT>/<service>

Replace <service> with the name specified in the YAML configuration file to send requests to the corresponding service.