gdg-bu-kg 0.1.2

Python SDK for the Babcock University Knowledge Graph API

gdg_bu_kg: Babcock Knowledge Graph Python SDK

A lightweight, production-ready Python SDK for interacting with the Babcock University Knowledge Graph platform.

Features

• Dual-Mode Querying:
  • client.query(): LLM-as-a-Service with built-in Babcock knowledge.
  • client.graph_query(): Extract raw contextual graph data (nodes & edges).
• LLM Control: Full control over system_prompt, temperature, max_tokens, and top_p.
• Developer-Friendly:
  • Sync and Async clients (httpx-based).
  • Strongly typed Pydantic v2 models.
  • Rich HTML display in Jupyter Notebooks.
  • Automatic retries with exponential backoff on transient errors.
• Minimal Dependencies: Only httpx, pydantic, and tenacity.

Installation

pip install gdg-bu-kg

Quick Start (Sync)

from gdg_bu_kg import KnowledgeGraphClient, AuthenticationError

# 1. Initialize with your API key
client = KnowledgeGraphClient(api_key="your_platform_api_key_here")

try:
    # 2. AI Query: LLM + Knowledge Graph
    response = client.query(
        "Who is the current Vice Chancellor of Babcock University?",
        system_prompt="You are a helpful university administrator.",
        temperature=0.7
    )
    
    # Access the AI text response
    print("AI Answer:", response.data.text_response)
    
    # Access supporting graph data
    for node in response.data.graph.nodes:
        print(f"Verified Entity: {node.label}")

    # 3. Raw Graph Data Query
    graph_resp = client.graph_query("List all departments in the School of Computing")
    print("Context:", graph_resp.data.context)
    
except AuthenticationError:
    print("Invalid API Key. Please regenerate it in the SaaS Dashboard.")
finally:
    client.close()

Async Usage

import asyncio
from kg_client import AsyncKnowledgeGraphClient

async def main():
    async with AsyncKnowledgeGraphClient(api_key="your_key") as client:
        response = await client.query("Tell me about the history of Babcock.")
        print(response.data.text_response)

if __name__ == "__main__":
    asyncio.run(main())

Jupyter Notebook Support

The SDK models render beautifully in Jupyter:

# In a Jupyter cell:
response = client.query("What is the School of Computing?")
response  # Automatically renders as a formatted HTML view

Error Handling

The SDK provides a rich exception hierarchy:

• AuthenticationError (401)
• ForbiddenError (403)
• InvalidQueryError (400)
• RateLimitExceededError (429)
• BackendServiceError (5xx)
• ConnectionError (Network issues)

All inherit from KnowledgeGraphError.

License

MIT