genai-processors

# GenAI Processors Library 📚

[![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](LICENSE)
[![PyPI version](https://img.shields.io/pypi/v/genai-processors.svg)](https://pypi.org/project/genai-processors/)
[![Documentation](https://img.shields.io/badge/View-Documentation-blue?&logo=read-the-docs)](https://google-gemini.github.io/genai-processors/)

**Build Modular, Asynchronous, and Composable AI Pipelines for Generative AI.**

GenAI Processors is a lightweight Python library that enables efficient,
parallel content processing. It addresses the fragmentation of LLM APIs through
three core pillars:

1.  **Unified Content Model**: A single, consistent representation for inputs
    and outputs across models, agents, and tools.
2.  **Processors**: Simple, composable Python classes that transform content
    streams using native `asyncio`.
3.  **Streaming**: Asynchronous streaming capabilities built-in by default,
    without added plumbing complexity.

At the ecosystem's core lies the `Processor`, which encapsulates a unit of work.
Through a "dual-interface" pattern, it handles the complexity of asynchronous,
multimodal data streaming while exposing a simple API to developers:

```python
from typing import AsyncIterable
from genai_processors import content_api
from genai_processors import processor

class EchoProcessor(processor.Processor):
  # The PRODUCER interface (for the processor author):
  # Takes a robust ProcessorStream as input, and yields part types.
  async def call(
      self, content: content_api.ProcessorStream
  ) -> AsyncIterable[content_api.ProcessorPartTypes]:
      # Process content as it streams in!
      async for part in content:
          yield part
```

Applying a `Processor` is just as straightforward. The CONSUMER interface
accepts wide, forgiving input types and returns a powerful stream that can be
awaited entirely or streamed chunk-by-chunk:

```python
# The CONSUMER interface (for the caller):
# Provide input effortless