Skip to content

Latest commit

 

History

History
110 lines (78 loc) · 2.61 KB

README.md

File metadata and controls

110 lines (78 loc) · 2.61 KB

awaiter

codecov Quality Gate Status

Makes your blocking functions awaitable.

awaiter.ThreadExecutor() represents a single thread that you can use to execute blocking functions in a FIFO manner. It does not use a thread pool like asyncio.to_thread() or loop.run_in_executor(), to keep it minimal and predictable.

Usage

import asyncio
import time

from awaiter import ThreadExecutor


def blocking_function(name):
    time.sleep(1)
    return f'Hello, {name}!'

async def main():
    async with ThreadExecutor() as executor:
        result = await executor(blocking_function)('World')
        print(result)

    # out of context, the thread is closed here

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

Or use the decorator style:

import asyncio
import time

from awaiter import ThreadExecutor

executor = ThreadExecutor()


@executor
def blocking_function(name):
    time.sleep(1)
    return f'Hello, {name}!'

async def main():
    executor.start()

    result = await blocking_function('World')
    print(result)

    executor.shutdown()
    # the thread will be closed.
    # if you want to wait until all queued tasks are completed:
    # await executor.shutdown()

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

If you want to execute multiple tasks at once without waiting in the main thread, use executor.submit():

# ...

    fut1 = executor.submit(blocking_function, 'World')
    fut2 = executor.submit(blocking_function, 'Foo')
    fut3 = executor.submit(blocking_function, 'Bar')

# ...

Last but not least, it also supports generator functions:

# ...

@executor
def generator_function(name):
    yield 'Hello, '
    time.sleep(1)
    yield name
    yield '!'

# ...

    async for data in generator_function('World'):
        print(data)

# ...

But I want a thread pool?

We provide the awaiter.MultiThreadExecutor helper.

It has a thread pool-like approach and is more suitable for use as a single, persistent object:

executor = MultiThreadExecutor(size=10)

How to use is the same, as the interface is identical to awaiter.ThreadExecutor.

Install

python3 -m pip install --upgrade awaiter

License

MIT