一、Asyncio核心概念

Python 3.4引入的asyncio模块提供了原生的异步I/O支持：

import asyncio

# 定义一个协程
async def fetch_data(url):
    print(f"开始获取 {url}")
    await asyncio.sleep(2)  # 模拟I/O操作
    print(f"完成获取 {url}")
    return f"{url} 的数据"

# 事件循环执行
async def main():
    task1 = asyncio.create_task(fetch_data("https://api1"))
    task2 = asyncio.create_task(fetch_data("https://api2"))
    
    await task1
    await task2

asyncio.run(main())

三大核心组件：事件循环、协程、任务

二、协程与任务管理

# 方式1：create_task
task = asyncio.create_task(coro())

# 方式2：ensure_future (兼容旧版)
task = asyncio.ensure_future(coro())

# 方式3：直接await
result = await coro()

# 同时运行多个协程
async def multi_tasks():
    # gather按顺序返回结果
    results = await asyncio.gather(
        task1(),
        task2(),
        task3()
    )
    
    # wait不保证顺序
    done, pending = await asyncio.wait(
        [task1(), task2()],
        timeout=1.5
    )

三、高级应用场景

import aiohttp

async def fetch_page(session, url):
    async with session.get(url) as response:
        return await response.text()

async def crawl(urls):
    async with aiohttp.ClientSession() as session:
        tasks = [fetch_page(session, url) for url in urls]
        return await asyncio.gather(*tasks)

urls = ["https://example.com/page1", 
        "https://example.com/page2"]
pages = asyncio.run(crawl(urls))

async def handle_websocket(websocket, path):
    async for message in websocket:
        if message == "ping":
            await websocket.send("pong")
        elif message == "time":
            await websocket.send(str(datetime.now())

start_server = websockets.serve(
    handle_websocket, "localhost", 8765)

asyncio.get_event_loop().run_until_complete(start_server)
asyncio.get_event_loop().run_forever()

四、性能优化技巧

• 限制并发量：使用Semaphore控制
• 超时处理：asyncio.wait_for设置超时
• 任务取消：task.cancel()优雅终止
• 资源复用：共享Session对象

# 并发限制示例
sem = asyncio.Semaphore(10)

async def limited_fetch(url):
    async with sem:
        return await fetch(url)

# 超时处理示例
try:
    result = await asyncio.wait_for(
        fetch_data(), 
        timeout=3.0
    )
except asyncio.TimeoutError:
    print("请求超时")

五、与传统多线程对比

测试环境：Python 3.9/100并发请求

六、常见问题解决方案

# 在同步环境中运行协程
def sync_call():
    return asyncio.get_event_loop().run_until_complete(async_func())

# 或者使用nest_asyncio
import nest_asyncio
nest_asyncio.apply()

# 使用run_in_executor
async def run_blocking():
    loop = asyncio.get_event_loop()
    await loop.run_in_executor(
        None,  # 使用默认线程池
        blocking_func
    )