75

I am using a third party module to retrieve data from an API. I simply would like to asynchronously await the module to return the data which occasionally takes several seconds and freezes up my app. However, when I try to await a call to that module I receive the TypeError:

TypeError: object dict can't be used in 'await' expression

import thirdPartyAPIwrapper

async def getData():
    retrienveData = await thirdPartyAPIWrapper.data()
    return await retrieveData

def main():
    loop = asncio.get_event_loop()
    data = loop.run_until_complete(getData())
    loop.close
    return data

Why can I not await a type('dict')? Is there a way around this? If async/await with asyncio will not work with a third party module that doesn't return a coroutine then what are my other options?

Riley Hughes
  • 1,344
  • 2
  • 12
  • 22

4 Answers4

83

Only asynchronous (defined with async def) functions can be awaited. Whole idea is that such functions are written special way what makes possible to run (await) them without blocking event loop.

If you want to get result from common (defined with def) function that takes some considerable time to be executed you have these options:

  • rewrite this whole function to be asynchronous
  • call this function in another thread and await for result asynchronously
  • call this function in another process and await for result asynchronously

Usually you want to choose second option.

Here's example of how to do it:

import asyncio
import time
from concurrent.futures import ThreadPoolExecutor


_executor = ThreadPoolExecutor(1)


def sync_blocking():
    time.sleep(2)


async def hello_world():
    # run blocking function in another thread,
    # and wait for it's result:
    await loop.run_in_executor(_executor, sync_blocking)


loop = asyncio.get_event_loop()
loop.run_until_complete(hello_world())
loop.close()

Please, read this answer about how asyncio works. I think it'll help you much.

Mikhail Gerasimov
  • 36,989
  • 16
  • 116
  • 159
  • 6
    Note, that unlike Javascript Python isn't created with asynchronicity in mind. In Python if you do something asynchronous you'll need to do it explicitly: explicitly to define asynchronous functions in particular. Common function can't just become asynchronous without efforts. That's a reason why you'll need thread to run synchronous function asynchronously. – Mikhail Gerasimov Apr 13 '18 at 18:29
8

As thirdPartyAPIWrapper.data() is a normal sync function you should call it in another thread.

There is a helper function for that in a asgiref library.
Assume we've got a blocking function with an argument:

import asyncio
import time

from asgiref.sync import sync_to_async


def blocking_function(seconds: int) -> str:
    time.sleep(seconds)
    return f"Finished in {seconds} seconds"

async def main():
    seconds_to_sleep = 5
    function_message = await sync_to_async(blocking_function)(seconds_to_sleep)
    print(function_message)

loop = asyncio.get_event_loop()
loop.run_until_complete(main())
loop.close()

There is also an async_to_sync helper function in that library.

Serge
  • 1,027
  • 9
  • 8
8

You do not need to await which function is running asynchronously

async def getData():
    retrienveData = thirdPartyAPIWrapper.data()
    return retrieveData
MD SHAYON
  • 7,001
  • 45
  • 38
0

I am writing test cases and I need to mock async functionality. So, you can write a simple helper function like so.

async def resolve(val):
    return val

Now you can await anything

foo = resolve(1) 
await foo # No error!
Souradeep Nanda
  • 3,116
  • 2
  • 30
  • 44