python - 通过Asyncio解决阻塞http python函数

Question

我正在实现一个通过 http 请求进行通信的区 block 链 ( inspired by this blogpost )。该区 block 链具有工作量证明方法，根据难度，可以在相当长的时间内阻止其他 http 请求。这就是为什么我尝试从 python 实现新的 asyncio 功能。作品如下:

async def proof_of_work(self, last_proof):
    """
    Simple Proof of Work Algorithm:
     - Find a number p' such that hash(pp') contains leading 4 zeroes, where p is the previous p'
    """
    proof = 0
    while self.valid_proof(last_proof, proof) is False:
        proof += 1
        await asyncio.sleep(1)

    return proof

但是，这使得我的工作量证明极其缓慢，我猜这是因为它在每次迭代后被迫休眠。解决这个问题的更优雅的方法是什么？

    while self.valid_proof(last_proof, proof) is False:
        proof += 1
        if proof % 1000 == 0:
             await asyncio.sleep(1)

可以加快速度，但看起来有点脏。实现这个的正确方法是什么？

Answer 1

如果您想在协程内运行 CPU 阻塞代码，则应使用 run_in_executor() 在单独的执行流中运行它(以避免 asyncio 的事件循环卡住)。

如果您只是想要另一个执行流，或者(我认为更好)使用ProcessPoolExecutor将与CPU相关的工作委托(delegate)给其他核心，则可以使用ThreadPoolExecutor。

import asyncio
from concurrent.futures import ProcessPoolExecutor
import hashlib


# ORIGINAL VERSION:
# https://github.com/dvf/blockchain/blob/master/blockchain.py
def valid_proof(last_proof, proof):
    guess = f'{last_proof}{proof}'.encode()
    guess_hash = hashlib.sha256(guess).hexdigest()
    return guess_hash[:4] == "0000"


def proof_of_work(last_proof):
    proof = 0
    while valid_proof(last_proof, proof) is False:
        proof += 1
    return proof


# ASYNC VERSION:
async def async_proof_of_work(last_proof):
    proof = await loop.run_in_executor(_executor, proof_of_work, last_proof)
    return proof


async def main():
    proof = await async_proof_of_work(0)
    print(proof)


if __name__ ==  '__main__':
    _executor = ProcessPoolExecutor(4)

    loop = asyncio.get_event_loop()
    try:
        loop.run_until_complete(main())
    finally:
        loop.run_until_complete(loop.shutdown_asyncgens())
        loop.close()

输出:

69732