I'm writing an application that needs to take in events from two sources. Some events will come from a Websocket client connection, and some will come from incoming HTTP requests. Is it possible with aiohttp to have both of these listeners in the same application, or do I need two separate executables?
I currently have a function that takes websocket events, and calls a callback on each event
async def callback(msg):
print(msg)
async def websocket():
session = aiohttp.ClientSession()
async with session.ws_connect('http://example.org/websocket') as ws:
async for msg in ws:
if msg.type == aiohttp.WSMsgType.TEXT:
await callback(msg.data)
elif msg.type == aiohttp.WSMsgType.CLOSED:
break
elif msg.type == aiohttp.WSMsgType.ERROR:
break
The problem is that this blocks forever, so I don't know how I could integrate this with the aiohttp server. If I do:
await websocket()
aiohttp.web.run_app(app)
then web.run_app
is never called. Is there some way to accomplish this? It seems like the ideal case for asyncio, multiple things handling multiple events asynchronously. How do I do it?
I was able to accomplish my goals by using the on_startup
handler of my aiohttp app.
async def callback(msg):
print(msg)
async def websocket(session):
async with session.ws_connect('http://example.org/websocket') as ws:
async for msg in ws:
if msg.type == aiohttp.WSMsgType.TEXT:
await callback(msg.data)
elif msg.type == aiohttp.WSMsgType.CLOSED:
break
elif msg.type == aiohttp.WSMsgType.ERROR:
break
async def init(app):
session = aiohttp.ClientSession()
app['websocket_task'] = app.loop.create_task(websocket(session))
app = web.Application()
app.on_startup.append(init)