aether-framework-sdk 0.1.4

Python SDK for Aether workflow engine

Aether Python SDK

A Python SDK for the Aether workflow engine.

Installation

pip install aether-sdk

Quick Start

from aether import AetherService, step, activity, workflow

class MyService(AetherService):
    service_name = 'my-service'
    group = 'my-group'
    language = ['python']
    
    @step()
    def process_data(self, data: dict) -> dict:
        """Process data step"""
        return {**data, 'processed': True}
    
    @activity(max_attempts=3, timeout=30000)
    def analyze_data(self, data: dict) -> dict:
        """Analyze data with retry"""
        return run_analysis(data)
    
    @workflow()
    async def pipeline(self, ctx, data: dict) -> dict:
        """Complete processing pipeline"""
        processed = await ctx.step('self::process_data', data)
        return await ctx.step('self::analyze_data', processed)

# Start the service
service = MyService()
await service.start(port=50051)

Features

• Decorators: @step(), @activity(), @workflow()
• Service Registration: Automatic registration with Aether server
• Context Methods: ctx.step(), ctx.activity(), ctx.child()
• Retry Logic: Built-in retry with exponential backoff

API Reference

AetherService

class AetherService:
    service_name: str      # Unique service identifier
    group: str             # Logical grouping
    language: List[str]    # Programming languages
    
    async def start(self, host: str = '0.0.0.0', port: int = 50051)
    async def stop(self)

Decorators

@step(name: Optional[str] = None)
@activity(options: Optional[ActivityOptions] = None, name: Optional[str] = None)
@workflow(name: Optional[str] = None)

Context Methods

async ctx.step(name: str, input: Any) -> Any
async ctx.activity(name: str, input: Any, options: Optional[ActivityOptions] = None) -> Any
async ctx.child(workflow: str, args: List[Any]) -> Any

License

MIT