Named async Python class

Question

I would like to call a class that does a post for an Endpoint but that this call was not blocking, because it is inside an event that sometimes triggers 5 cohorts in a row and I can’t wait for the Post’s conclusion to continue.

I did it this way, but I’m having a hard time testing if it’s really working

Main class.py

"""
códigos que não interessa no exemplo
"""
from def_api import Api
api = Api('teste')
for item in range(10):   #for ficticio
   api.Send('valores')

def_api.py class

import requests

class Api(object)
   def __init__(self, valor1):
      self.__valor1 = valor1

   async def Send(self, valor2):
     self.__valor2 = valor2
     data = {'valor1': self.__valor1, 'valor2': self.__valor2}
     r = request.post(url, json=data)
     return await r.status_code

I was in doubt: whether the await should be placed on the request, r = await request.post.. or if I should take the await so it doesn’t get blocking..

I ask for help, links, etc.

Answer 1

The requests is a synchronous library - that is, nothing you do will allow requests to run in a non-blocking and collaborative way with asyncio, directly. A path would be to search Pypi for other libraries that emulate requests, but have an asynchronous api - for example the aiohttp-requests.

However, even without exchanging requests for another lib, not everything is lost: it is precisely for these cases that the Python asyncio includes the so-called "run_in_executor" - basically, there is a way to create an object "Executor" of the library concurrent.futures that maintains a set of threads (or processes), and can run the blocking functions on one of these threads, so that the function that needs this result can continue in a non-blocking way on the main thread.

The call loop.run_in_executor in itself takes charge of dispatching the function to the executor, and check the result -the bureaucracy is minimal, just creating the executing object.

Documentation: https://docs.python.org/3/library/asyncio-eventloop.html#asyncio.loop.run_in_executor

I won’t be able to test the code now, but at first the adaptation of your code is something like:

import requests
from concurrent.futures import ThreadPoolExecutor as Executor

# configura 20 requisições máximas em paralelo:

http_executor = Executor(20)

class Api(object)
   def __init__(self, valor1):
      self.__valor1 = valor1

   async def send(self, valor2):
     self.__valor2 = valor2
     data = {'valor1': self.__valor1, 'valor2': self.__valor2}
     loop = asyncio.get_event_loop()
     r = await loop.run_in_executor(http_executor,  request.post, url, json=data)
     return r.status_code

(plus the hint of how will be the layer of run_in_executor, I changed the method name to send in lowercase instead of Send, why is the convention in Python)