实现此目的的一种方法是使用队列。
#!/usr/bin/python3
import asyncio
try:
# python 3.4
from asyncio import JoinableQueue as Queue
except:
# python 3.5
from asyncio import Queue
@asyncio.coroutine
def do_work(task_name, work_queue):
while not work_queue.empty():
queue_item = work_queue.get_nowait()
# simulate condition where task is added dynamically
if queue_item % 2 != 0:
work_queue.put_nowait(2)
print('Added additional item to queue')
print('{0} got item: {1}'.format(task_name, queue_item))
yield from asyncio.sleep(queue_item)
print('{0} finished processing item: {1}'.format(task_name, queue_item))
if __name__ == '__main__':
queue = Queue()
# Load initial jobs into queue
[queue.put_nowait(x) for x in range(1, 6)]
# use 3 workers to consume tasks
taskers = [
do_work('task1', queue),
do_work('task2', queue),
do_work('task3', queue)
]
loop = asyncio.get_event_loop()
loop.run_until_complete(asyncio.wait(taskers))
loop.close()
使用来自 asyncio 的队列,您可以确保工作的“单元”与最初提供给 asyncio 事件循环的任务/期货是分开的。基本上,这允许在某些条件下添加额外的工作“单元”。
请注意,在上面的示例中,偶数编号的任务是终端,因此如果是这种情况,则不会添加额外的任务。这最终会导致所有任务的完成,但在您的情况下,您可以轻松地使用另一个条件来确定是否将另一个项目添加到队列中。
输出:
Added additional item to queue
task2 got item: 1
task1 got item: 2
Added additional item to queue
task3 got item: 3
task2 finished processing item: 1
task2 got item: 4
task1 finished processing item: 2
Added additional item to queue
task1 got item: 5
task3 finished processing item: 3
task3 got item: 2
task3 finished processing item: 2
task3 got item: 2
task2 finished processing item: 4
task2 got item: 2
task1 finished processing item: 5
task3 finished processing item: 2
task2 finished processing item: 2