Python async/await

Summary: in this tutorial, you will learn about Python coroutines and how to use the Python async and await keywords to create and pause coroutines.

Introduction to Python coroutines

A coroutine is a regular function with the ability to pause its execution when encountering an operation that may take a while to complete.

When the long-running operation completes, you can resume the paused coroutine and execute the remaining code in that coroutine.

While the coroutine is waiting for the long-running operation, you can run other code. By doing this, you can run the program asynchronously to improve its performance.

To create and pause a coroutine, you use the Python async and await keywords:

• The async keyword creates a coroutine.
• The await keyword pauses a coroutine.

Defining a coroutine with Python async keyword

The following defines a simple function that returns the square number of an integer:

def square(number: int) -> int: return number*numberCode language: Python (python)

And you can pass an integer to the square() function to get its square number:

def square(number: int) -> int: return number*number result = square(10) print(result)Code language: Python (python)

Output:

100Code language: Python (python)

When you add the async keyword to the function, the function becomes a coroutine:

async def square(number: int) -> int: return number*numberCode language: Python (python)

A calling coroutine returns a coroutine object that will be run later. For example:

async def square(number: int) -> int: return number*number result = square(10) print(result)Code language: Python (python)

Output:

<coroutine object square at 0x00000185C31E7D80> sys:1: RuntimeWarning: coroutine 'square' was never awaitedCode language: Python (python)

In this example, we call the square() coroutine, assign the returned value to the result variable, and print it out.

When you call a coroutine, Python doesn’t execute the code inside the coroutine immediately. Instead, it returns a coroutine object.

The second line in the output also shows an error message indicating that the coroutine was never awaited. More on this in the following await section:

sys:1: RuntimeWarning: coroutine 'square' was never awaitedCode language: JavaScript (javascript)

To run a coroutine, you need to execute it on an event loop. Prior to Python 3.7, you have to manually create an event loop to execute coroutines and close the event loop.

However, since version 3.7, the asyncio library added some functions that simplify the event loop management.

For example, you can use the asyncio.run() function to automatically create an event loop, run a coroutine, and close it.

The following uses the asyncio.run() function to execute the square() coroutine and get the result:

import asyncio async def square(number: int) -> int: return number*number result = asyncio.run(square(10)) print(result)Code language: Python (python)

Output:

100Code language: Python (python)

It’s important to note that the asyncio.run() is designed to be the main entry point of an asyncio program.

Also, the asyncio.run() function only executes one coroutine which may call other coroutines and functions in the program.

Pausing a coroutine with Python await keyword

The await keyword pauses the execution of a coroutine. The await keyword is followed by a call to a coroutine like this:

result = await my_coroutine()Code language: Python (python)

The await keyword causes the my_coroutine() to execute, waits for the code to be completed, and returns a result.

It’s important to note that await keyword is only valid inside a coroutine. In other words, you must use the await keyword inside a coroutine.

This is the reason why you saw an error message in the above example that uses the await keyword outside of a coroutine.

The following example shows how to use the await keyword to pause a coroutine:

import asyncio async def square(number: int) -> int: return number*number async def main() -> None: x = await square(10) print(f'x={x}') y = await square(5) print(f'y={y}') print(f'total={x+y}') if __name__ == '__main__': asyncio.run(main())Code language: Python (python)

Output:

x=100 y=25 total=125Code language: Python (python)

How it works. (we’ll focus on the main() function):

First, call the square() coroutine using the await keyword. The await keyword will pause the execution of the main() coroutine, wait for the square() coroutine to complete, and return the result:

x = await square(10) print(f'x={x}')Code language: Python (python)

Second, call the square() coroutine a second time using the await keyword:

y = await square(5) print(f'y={y}')Code language: Python (python)

Third, display the total:

print(f'total={x+y}')Code language: Python (python)

The following statement uses the run() function to execute the main() coroutine and manage the event loop:

asyncio.run(main())Code language: Python (python)

So far, our program executes like a synchronous program. It doesn’t reveal the power of the asynchronous programming model.