为什么我的消费者在队列中与我的生产者分开工作？

Question

我的目标是异步调用 API，并将结果（来自每次调用）写入文件（1 个调用 -> 1 个文件）。 我认为实现这一点的一种方法是使用队列。 我的目的是在生产者准备好响应后立即将其推送到队列中，然后让消费者在文件可用时立即处理（写入）文件。

困惑：在我运行代码时查看打印语句，我看到首先生产者完成，然后消费者开始使用我的 output。 在 go 看来，这似乎不是我的意图，即消费者在任务可用后立即处理它们。 我也考虑过使用多个进程（1 个用于消费者，1 个用于生产者），但我不确定我是否以这种方式使事情复杂化。

我创建了当前状态的插图：

import aiohttp
import asyncio


async def get_data(session, day):
    async with session.post(url=SOME_URL, json=SOME_FORMAT, headers=HEADERS) as response:
        return await response.text()


async def producer(q, day):
    async with aiohttp.ClientSession() as session:
        result = await get_data(session, day)
        await q.put(result)


async def consumer(q):
    while True:
        outcome = await q.get()
        print("Consumed:", outcome) # assuming I write files here
        q.task_done()


async def main():
    queue = asyncio.Queue()
    days = [day for day in range(20)]  # Here I normally use calendar dates instead of range
    producers = [asyncio.create_task(producer(queue, day) for day in days]
    consumer = asyncio.create_task(consumer(queue)
    await asyncio.gather(*producers)
    await queue.join()
    consumer.cancel()

    if __name__ == '__main__':
        asyncio.run(main())

我在正确的轨道上吗？

Answer 1

您的代码通常很好（除了几个语法错误，我猜这是错误的复制粘贴的结果）。 所有的生产者确实是在消费者开始工作之前创建的，因为他们没有什么可等待的。 但是，如果生产者需要做真正的工作，你会看到他们只有在消费者开始工作后才完成工作，然后事情才开始工作。

这是您的代码的编辑版本，加上 output 证明事情确实有效。

import aiohttp
import asyncio

async def get_data(session, day):
    print(f"get data, day {day}")
    async with session.get(url="https://www.google.com") as response:
        res = await response.text()
    print(f"got data, day {day}")
    return res[:100]

async def producer(q, day):
    async with aiohttp.ClientSession() as session:
        result = await get_data(session, day)
        await q.put(result)

async def consumer(q):
    print("Consumer stated")
    while True:
        outcome = await q.get()
        print("Consumed:", outcome) # assuming I write files here
        asyncio.sleep(1)
        q.task_done()

async def main():
    queue = asyncio.Queue()
    days = [day for day in range(20)]  # Here I normally use calendar dates instead of range
    producers = [asyncio.create_task(producer(queue, day)) for day in days]
    print("main: producer tasks created")
    consumer_task = asyncio.create_task(consumer(queue))
    print("main: consumer task created")
    await asyncio.gather(*producers)
    print("main: gathered producers")
    await queue.join()
    consumer_task.cancel()

if __name__ == '__main__':
    asyncio.run(main())

output：

main: producer tasks created
main: consumer task created
get data, day 0
get data, day 1
get data, day 2
get data, day 3
...
get data, day 19
Consumer stated
got data, day 1
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
queue_so.py:21: RuntimeWarning: coroutine 'sleep' was never awaited
  asyncio.sleep(1)
RuntimeWarning: Enable tracemalloc to get the object allocation traceback
got data, day 10
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
got data, day 19
got data, day 11
got data, day 14
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
got data, day 15
got data, day 17
got data, day 6
got data, day 18
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
got data, day 7
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
got data, day 8
got data, day 9
got data, day 2
got data, day 12
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
got data, day 0
got data, day 5
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
got data, day 4
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
got data, day 3
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
got data, day 13
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
got data, day 16
Consumed: <!doctype html><html itemscope="" itemtype="http://schema.org/WebPage" lang="en"><head><meta content
main: gathered producers

为什么我的消费者在队列中与我的生产者分开工作？

问题描述

1 个解决方案

解决方案1
2 已采纳 2020-06-06 05:44:38

为什么我的消费者在队列中与我的生产者分开工作？

问题描述

1 个解决方案

解决方案1 2 已采纳 2020-06-06 05:44:38

解决方案1
2 已采纳 2020-06-06 05:44:38