oaas-sdk2-py 0.3.5

No project description provided

OaaS-SDK2

This library helps you develop a runtime that can be run in a Object as a Service (OaaS) serverless platform. For more information on the OaaS model, visit https://github.com/hpcclab/OaaS.

Table of Contents

• Documentation
• Installation
• Features
• Quick Start
• Examples
• API Overview
• Build the project

Documentation

For a comprehensive guide and API reference, please see the docs directory:

• Tutorial: A step-by-step guide to getting started with the OaaS SDK.
• API Reference: A detailed reference of all classes, methods, and functions.

Installation

To install oaas-sdk2-py, you can use pip:

pip install oaas-sdk2-py

Or, if you are using uv:

# For adding/installing packages with uv, the command is 'uv pip install'
uv add oaas-sdk2-py

Features

• Simplified API: Easy-to-use decorators and type-safe method definitions
• Type Safety: Full Pydantic model support with automatic validation
• Async/Sync Support: Built with async/await for non-blocking operations
• Data Persistence: Object data is persisted and can be retrieved
• Remote Procedure Calls (RPC): Invoke methods on objects remotely
• Mocking Framework: Includes a mocking utility for testing your OaaS applications
• Rust-Powered Core: High-performance core components written in Rust for speed and efficiency
• Accessor Methods: Explicit @oaas.getter/@oaas.setter for typed persisted fields (not exported as standalone functions)

Quick Start

Basic Service Definition

from oaas_sdk2_py import oaas, OaasObject, OaasConfig
from pydantic import BaseModel

# Configure OaaS
config = OaasConfig(async_mode=True, mock_mode=False)
oaas.configure(config)

# Define request/response models
class GreetRequest(BaseModel):
    name: str

class GreetResponse(BaseModel):
    message: str

# Define your service
@oaas.service("Greeter", package="example")
class Greeter(OaasObject):
    counter: int = 0
    
    @oaas.method()
    async def greet(self, req: GreetRequest) -> GreetResponse:
        return GreetResponse(message=f"Hello, {req.name}!")
    
    @oaas.method()
    async def count_greetings(self) -> int:
        self.counter += 1
        return self.counter
    
    @oaas.method()
    async def is_popular(self, threshold: int = 10) -> bool:
        return self.counter > threshold

    # Accessor methods (not exported as RPC functions)
    @oaas.getter()
    async def get_counter(self) -> int:
        ...

    @oaas.setter()
    async def set_counter(self, value: int) -> int:
        ...

# Usage
async def main():
    # Create and use locally
    greeter = Greeter.create(local=True)
    
    # Test with different types
    response = await greeter.greet(GreetRequest(name="World"))
    count = await greeter.count_greetings()  # Returns int
    popular = await greeter.is_popular(5)     # Returns bool
    current_counter = await greeter.get_counter()  # Accessor getter
    _ = await greeter.set_counter(42)             # Accessor setter
    
    print(f"{response.message} (Count: {count}, Popular: {popular})")

Server & Agent Management

# Start gRPC server (for external access)
oaas.start_server(port=8080)

# Start agent (for background processing)
agent_id = await oaas.start_agent(Greeter, obj_id=123)

# Check status
print(f"Server running: {oaas.is_server_running()}")
print(f"Agents: {oaas.list_agents()}")

# Cleanup
await oaas.stop_agent(agent_id)
oaas.stop_server()

API Overview

Core Components

• @oaas.service: Decorator to define OaaS services
• @oaas.method: Decorator to expose methods as RPC endpoints
• @oaas.getter/@oaas.setter: Accessor decorators for persisted fields (not exported as RPC functions)
• OaasObject: Base class for all OaaS objects with persistence
• OaasConfig: Configuration for OaaS runtime

Supported Types

The SDK natively supports these Python types:

• Primitives: int, float, bool, str
• Collections: list, dict
• Binary: bytes
• Models: Pydantic BaseModel classes

Examples

System Monitoring Service

import psutil
from oaas_sdk2_py import oaas, OaasObject

@oaas.service("ComputeDevice", package="monitoring")
class ComputeDevice(OaasObject):
    metrics: dict = {}

    @oaas.method()
    async def get_cpu_usage(self) -> float:
        """Get current CPU usage as a percentage."""
        return psutil.cpu_percent(interval=0.1)

    @oaas.method()
    async def get_process_count(self) -> int:
        """Get number of running processes."""
        return len(psutil.pids())

    @oaas.method()
    async def is_healthy(self, cpu_threshold: float = 80.0) -> bool:
        """Check if system is healthy."""
        cpu_usage = await self.get_cpu_usage()
        return cpu_usage < cpu_threshold

    @oaas.method()
    async def monitor_continuously(self, duration: int) -> dict:
        """Monitor for specified duration and return metrics."""
        samples = []
        for _ in range(duration):
            cpu = psutil.cpu_percent(interval=1.0)
            samples.append(cpu)
        
        return {
            "avg_cpu": sum(samples) / len(samples),
            "samples": len(samples),
            "duration": duration
        }

# Usage
device = ComputeDevice.create(local=True)
cpu_usage = await device.get_cpu_usage()      # Returns float
process_count = await device.get_process_count()  # Returns int
is_healthy = await device.is_healthy(75.0)    # Returns bool
metrics = await device.monitor_continuously(5)  # Returns dict

Counter Service with State

from oaas_sdk2_py import oaas, OaasObject

@oaas.service("Counter", package="example")
class Counter(OaasObject):
    count: int = 0
    history: list = []

    @oaas.method()
    async def increment(self, amount: int = 1) -> int:
        """Increment counter by amount."""
        self.count += amount
        self.history.append(f"Added {amount}")
        return self.count

    @oaas.method()
    async def reset(self) -> bool:
        """Reset counter to zero."""
        self.count = 0
        self.history.clear()
        return True

    # Accessors for the state fields
    @oaas.getter()
    async def get_count(self) -> int:
        ...

    @oaas.setter()
    async def set_count(self, value: int) -> int:
        ...

    @oaas.getter("history")
    async def get_history(self) -> list:
        ...

# Usage
counter = Counter.create(local=True)
value = await counter.increment(5)     # Returns int: 5
current = await counter.get_count()    # Accessor getter: 5
history = await counter.get_history()  # Accessor getter: ["Added 5"]
reset = await counter.reset()          # Returns bool: True
await counter.set_count(5)             # Accessor setter: set to 5
value = await counter.get_count()      # Accessor getter: 5

Why use accessors over methods?

Primary goal: avoid redundant RPC for simple state access. A method does RPC → run code → access data → RPC reply; an accessor reads/writes the persisted data directly.

Accessors (@oaas.getter/@oaas.setter) are the preferred way to expose simple reads/writes of persisted fields:

• Smaller external surface: accessors are not exported as standalone RPC functions, reducing accidental public APIs.
• Type-safe binding: validated against the field’s annotation at registration; name-based inference cuts boilerplate.
• Clear semantics: getters don’t have side effects; setters just write. Easy to reason about and review.
• Projection support: getters can return a projected sub-value from structured fields (e.g., dict/model paths).

Before (method):

@oaas.method()
async def get_count(self) -> int:
    return self.count

After (accessor):

@oaas.getter("count")
async def get_count(self) -> int:
    ...  # body not required; semantics = read persisted field

Testing with Mock Mode

import pytest
from oaas_sdk2_py import oaas, OaasConfig

# Configure for testing
@pytest.fixture
def setup_mock():
    config = OaasConfig(mock_mode=True, async_mode=True)
    oaas.configure(config)

@pytest.mark.asyncio
async def test_counter_service(setup_mock):
    counter = Counter.create(local=True)
    
    # Test increment
    result = await counter.increment(10)
    assert result == 10
    assert isinstance(result, int)
    
    # Test history
    history = await counter.get_history()
    assert history == ["Added 10"]
    assert isinstance(history, list)
    
    # Test reset
    reset_result = await counter.reset()
    assert reset_result is True
    assert isinstance(reset_result, bool)

Running as Server/Agent

Create a main module to run your service:

# main.py
import asyncio
import sys
from oaas_sdk2_py import oaas, OaasConfig

async def run_server():
    """Run gRPC server for external access."""
    config = OaasConfig(async_mode=True, mock_mode=False)
    oaas.configure(config)
    
    oaas.start_server(port=8080)
    print("🚀 Server running on port 8080")
    
    try:
        while True:
            await asyncio.sleep(1)
    except KeyboardInterrupt:
        print("🛑 Shutting down...")
    finally:
        oaas.stop_server()

async def run_agent():
    """Run agent for background processing."""
    config = OaasConfig(async_mode=True, mock_mode=False)
    oaas.configure(config)
    
    # Start both server and agent
    oaas.start_server(port=8080)
    agent_id = await oaas.start_agent(Counter, obj_id=1)
    print(f"🤖 Agent started: {agent_id}")
    
    try:
        while True:
            await asyncio.sleep(5)
            print(f"📊 Server: {oaas.is_server_running()}, Agents: {len(oaas.list_agents())}")
    except KeyboardInterrupt:
        print("🛑 Shutting down...")
    finally:
        await oaas.stop_all_agents()
        oaas.stop_server()

if __name__ == "__main__":
    if len(sys.argv) > 1 and sys.argv[1] == "agent":
        asyncio.run(run_agent())
    else:
        asyncio.run(run_server())

Run with:

# Server only
python main.py

# Server + Agent
python main.py agent

Run on OaaS

Prerequisites

• cargo (install via rust)
• oprc-cli cargo install --git https://github.com/pawissanutt/oaas-rs.git oprc-cli
• OaaS Platform (Oparaca)
  • Kubernetes Cluster (e.g., k3d with Docker runtime)

Build the project

You don't need to follow this guide unless you want to build the Python package on your own.

Prerequisites

• Python
• cargo (install via rust)
• uv (python package manager)

Build

uv sync
uv build