Writing files asynchronously

Eli Korvigo picture Eli Korvigo · Jan 22, 2017 · Viewed 28.1k times · Source

I've been trying to create a server-process that receives an input file path and an output path from client processes asynchronously. The server does some database-reliant transformations, but for the sake of simplicity let's assume it merely puts everything to the upper case. Here is a toy example of the server:

import asyncio
import aiofiles as aiof
import logging
import sys


ADDRESS = ("localhost", 10000)

logging.basicConfig(level=logging.DEBUG,
                    format="%(name)s: %(message)s",
                    stream=sys.stderr)

log = logging.getLogger("main")
loop = asyncio.get_event_loop()


async def server(reader, writer):
    log = logging.getLogger("process at {}:{}".format(*ADDRESS))
    paths = await reader.read()
    in_fp, out_fp = paths.splitlines()
    log.debug("connection accepted")
    log.debug("processing file {!r}, writing output to {!r}".format(in_fp, out_fp))
    async with aiof.open(in_fp, loop=loop) as inp, aiof.open(out_fp, "w", loop=loop) as out:
        async for line in inp:
            out.write(line.upper())
        out.flush()
    writer.write(b"done")
    await writer.drain()
    log.debug("closing")
    writer.close()
    return


factory = asyncio.start_server(server, *ADDRESS)
server = loop.run_until_complete(factory)
log.debug("starting up on {} port {}".format(*ADDRESS))

try:
    loop.run_forever()
except KeyboardInterrupt:
    pass
finally:
    log.debug("closing server")
    server.close()
    loop.run_until_complete(server.wait_closed())
    log.debug("closing event loop")
    loop.close()

The client:

import asyncio
import logging
import sys
import random

ADDRESS = ("localhost", 10000)
MESSAGES = ["/path/to/a/big/file.txt\n", 
            "/output/file_{}.txt\n".format(random.randint(0, 99999))]

logging.basicConfig(level=logging.DEBUG,
                    format="%(name)s: %(message)s",
                    stream=sys.stderr)

log = logging.getLogger("main")
loop = asyncio.get_event_loop()

async def client(address, messages):
    log = logging.getLogger("client")
    log.debug("connecting to {} port {}".format(*address))
    reader, writer = await asyncio.open_connection(*address)
    writer.writelines([bytes(line, "utf8") for line in messages])
    if writer.can_write_eof():
        writer.write_eof()
    await writer.drain()

    log.debug("waiting for response")
    response = await reader.read()
    log.debug("received {!r}".format(response))
    writer.close()
    return


try:
    loop.run_until_complete(client(ADDRESS, MESSAGES))
finally:
    log.debug("closing event loop")
    loop.close()

I activated the server and several clients at once. The server's logs:

asyncio: Using selector: KqueueSelector
main: starting up on localhost port 10000
process at localhost:10000: connection accepted
process at localhost:10000: processing file b'/path/to/a/big/file.txt', writing output to b'/output/file_79609.txt'
process at localhost:10000: connection accepted
process at localhost:10000: processing file b'/path/to/a/big/file.txt', writing output to b'/output/file_68917.txt'
process at localhost:10000: connection accepted
process at localhost:10000: processing file b'/path/to/a/big/file.txt', writing output to b'/output/file_2439.txt'
process at localhost:10000: closing
process at localhost:10000: closing
process at localhost:10000: closing

All clients print this:

asyncio: Using selector: KqueueSelector
client: connecting to localhost port 10000
client: waiting for response
client: received b'done'
main: closing event loop

The output files are created, but they remain empty. I believe they are not being flushed. Any way I can fix it?

Answer

Udi picture Udi · Jan 24, 2017

You are missing an await before out.write() and out.flush():

import asyncio
from pathlib import Path

import aiofiles as aiof

FILENAME = "foo.txt"


async def bad():
    async with aiof.open(FILENAME, "w") as out:
        out.write("hello world")
        out.flush()
    print("done")
    

async def good():
    async with aiof.open(FILENAME, "w") as out:
        await out.write("hello world")
        await out.flush()
    print("done")


loop = asyncio.get_event_loop()

server = loop.run_until_complete(bad())
print(Path(FILENAME).stat().st_size)  # prints 0

server = loop.run_until_complete(good())
print(Path(FILENAME).stat().st_size)  # prints 11

However, I would strongly recommend trying to skip aiofiles and use regular, synchronized disk I/O, and keep asyncio for network activity:

with open(file, "w") as out:  # regular file I/O
    async for s in network_request():  # asyncio for slow network work. measure it!
        out.write(s) # should be really quick, measure it!