condy 0.2.3

Python client for the Condensation.ai API

Condy: Python Client for Condensation.ai

A modern, async Python client library for interacting with the Condensation.ai API, providing seamless access to document processing and RAG (Retrieval-Augmented Generation) capabilities.

Features

• 🚀 Fully async implementation using httpx
• 📄 Document processing and chunking
• 🔍 RAG (Retrieval-Augmented Generation) querying
• 💪 Type hints and modern Python practices
• ⚡ Efficient markdown processing and page management
• 🛡️ Comprehensive error handling

Installation

pip install condy

Quick Start

import asyncio
from condy import CondyClient

async def main():
    # Initialize the client
    client = CondyClient(api_key="your-api-key")
    
    # Process a document from URL
    doc_output = await client.process_document("https://example.com/document.md") | await client.process_document("https://example.com/document.pdf")
    
    # Query the document using RAG
    response = await client.query_rag(
        question="What are the key points?",
        document_id=doc_output.document_id
    )
    
    print(response.answer)

asyncio.run(main())

Core Features

Document Processing

# Upload pages directly
pages = {
    1: "Page 1 content",
    2: "Page 2 content"
}
doc_output = await client.upload_pages(pages)

# Or process markdown from URL
doc_output = await client.process_document("https://example.com/document.md")

RAG Queries

# Query a document
response = await client.query_rag(
    question="What does the document say about X?",
    document_id="doc-id",
    max_chunks=5
)

# Access the response
print(response.answer)
print(response.source_pages)  # List of source pages used

Chunk Management

# Fetch document chunks
chunks = await client.fetch_chunks(
    document_id="doc-id",
    include_embeddings=False  # Set to True to include vector embeddings
)

Configuration

The client can be configured with custom settings:

client = CondyClient(
    api_key="your-api-key",
    base_url="https://custom-url.example.com",  # Optional
    timeout=60.0  # Custom timeout in seconds
)

Error Handling

The library provides specific exceptions for different error cases:

• NetworkError: For connection and network-related issues
• TimeoutError: For request timeout issues
• APIError: For API-specific errors with status codes and details

from condy import NetworkError, TimeoutError, APIError

try:
    response = await client.query_rag(question="...", document_id="...")
except TimeoutError:
    print("Request timed out")
except NetworkError:
    print("Network error occurred")
except APIError as e:
    print(f"API error: {e.status_code} - {e.detail}")

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

[License information here]

Support

For support, please [contact information here]