Asta Summarizer Service

A flexible text summarization service that supports multiple AI providers including OpenAI-compatible APIs and HuggingFace Inference endpoints with a consistent tone used for Asta.

Features

• Multiple Provider Support: Works with both OpenAI-compatible APIs and HuggingFace Inference clients
• Flexible Summarization Types: Support for different summarization styles (default, thread titles, etc.)
• Customizable Length: Configure summary length to meet your needs

Installation

pip install asta-summarizer

Quick Start

Using with OpenAI-compatible API

import asyncio
from asta_summarizer import SummarizerService, SummarizationType

async def main():
    # Initialize with base_url for OpenAI-compatible API
    service = SummarizerService(
        base_url="https://your-api-endpoint.com/v1",
        model="your-model-name",
        api_key="your-api-key"  # or set SUMMARIZER_API_KEY environment variable
    )

    text = "Your long text to summarize..."

    summary = await service.summarize(
        text=text,
        length=100,
        summarization_type=SummarizationType.DEFAULT
    )

    print(summary)

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

Using with HuggingFace Provider

import asyncio
from asta_summarizer import SummarizerService, SummarizationType

async def main():
    # Initialize with provider for HuggingFace Inference
    service = SummarizerService(
        provider="huggingface",  # or other supported providers
        model="your-model-name",
        api_key="your-hf-token"  # or set SUMMARIZER_API_KEY environment variable
    )

    text = "Your long text to summarize..."

    summary = await service.summarize(
        text=text,
        length=200,
        summarization_type=SummarizationType.THREAD_TITLE
    )

    print(summary)

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

Configuration

Initialization Parameters

• model (required): The model name to use for summarization
• base_url (optional): Base URL for OpenAI-compatible APIs (mutually exclusive with provider)
• provider (optional): Provider name for HuggingFace Inference (mutually exclusive with base_url)
• api_key (optional): API key for authentication (defaults to SUMMARIZER_API_KEY environment variable)

Summarization Types

The service supports different summarization styles via the SummarizationType enum:

• SummarizationType.DEFAULT: Standard summarization. Pass in your own prompt alongside the text to summarize in the text field.
• SummarizationType.THREAD_TITLE: Optimized for creating thread titles

Environment Variables

Set your API key as an environment variable:

export SUMMARIZER_API_KEY="your-api-key-here"

Development

Installing Dependencies

For development work:

# Install package in editable mode with development dependencies
pip install -e ".[dev]"

Or install dependencies manually:

# Core dependencies
pip install huggingface-hub>=0.20.0 openai>=1.0.0

# Development dependencies
pip install pytest>=7.0.0 pytest-asyncio>=0.21.0 black>=23.0.0 isort>=5.12.0 mypy>=1.0.0

Code Formatting

black .
isort .

Type Checking

mypy asta_summarizer/

Publication

You can publish a new version from your branch before merging to main, or from main after merging.

Edit the version.txt file with the new version, then run

export AI2_NORA_PYPI_TOKEN=<SECRET IN NORA VAULT>
make publish

This will publish the summarizer with the version number contained in version.txt

License

MIT License - see LICENSE file for details.