[英]How can I change this code to use context managers?
我正在嘗試使用aiohttp
和asyncio
多個憑據同時登錄網站。 在create_tasks
函數中,我生成一個要用於每個會話的會話列表。 我不能在login
功能中創建一個sesssion的原因是因為在整個代碼中將使用相同的會話對象。 我正在嘗試做的是設計一種方法,我可以使用上下文管理器來處理會話的關閉(以避免運行時錯誤使其保持打開狀態)。
以下代碼按預期工作(並發收集登錄頁面並在進程池中解析令牌),但它會與任務分開生成會話,並要求我在最后關閉它們。
from bs4 import BeautifulSoup
from concurrent.futures import ProcessPoolExecutor
import aiohttp
import asyncio
#TODO: make this safe, handle exceptions
LOGIN_URL = "http://example.com/login"
CLIENT_CNT = 10
proc_pool = ProcessPoolExecutor(CLIENT_CNT)
def get_key(text):
soup = BeautifulSoup(text, "html.parser")
form = soup.find("form")
key = form.find("input", attrs={"type": "hidden", "name": "authenticityToken"})
return key.get("value", None)
async def login(username:str, password:str, session:aiohttp.ClientSession, sem:asyncio.BoundedSemaphore, loop:asyncio.AbstractEventLoop=None):
loop = loop or asyncio.get_event_loop()
async with sem:
async with session.get(LOGIN_URL) as resp:
x = await asyncio.ensure_future(loop.run_in_executor(proc_pool, get_key, await resp.text()))
print(x)
def create_tasks(usernames, passwords, sem:asyncio.BoundedSemaphore, loop:asyncio.AbstractEventLoop=None):
loop = loop or asyncio.get_event_loop()
tasks = []
sessions = []
for u, p in zip(usernames, passwords):
session = aiohttp.ClientSession(loop=loop)
sessions.append(session)
tasks.append(login(u, p, session, sem, loop))
return tasks, sessions
if __name__ == "__main__":
loop = asyncio.get_event_loop()
sem = asyncio.BoundedSemaphore(CLIENT_CNT)
usernames = ("a", "b", "c", "d", "e", "f", "g")
passwords = ("a", "b", "c", "d", "e", "f", "g")
tasks, sessions = create_tasks(usernames, passwords, sem, loop)
loop.run_until_complete(asyncio.gather(*tasks, loop=loop))
for session in sessions:
session.close()
我之前使create_tasks
成為一個協程,寫了一個包裝類來制作異步迭代,並嘗試使用
async with aiohttp.ClientSession() as session:
tasks.append(login(u, p, session, sem, loop)
但正如我所擔心的,它表示會議在運行時已經關閉。
這是一種使推理更容易的結構:
async def user(u, p, ...):
"""Everything a single user does"""
auth = await login(u, p)
await download_something(auth, ...)
await post_something(auth, ...)
async def login(u, p): ...
async with aiohttp.ClientSession() as session:
async with session.get("http://xxx/login", ...) as r:
data = await r.json()
return data["something"]
async def download_xxx(...): ...
async def post_xxx(...): ...
async def everything():
creds = [("u1", "p1"), ...]
flows = [asyncio.ensure_future(user(cred)) for cred in creds]
for flow in flows:
await flow
警告程序員:默認情況下, aiohttp
出現存儲cookie,確保它不會對您的用戶流進行異花授粉。
獎勵積分:在最后一個異步函數中正確使用asyncio.gather()
。
使用ExitStack 。
from contextlib import ExitStack
def create_tasks(..., context):
tasks = []
for username in usernames:
session = aiohttp.ClientSession()
tasks.append(...)
context.enter_context(session)
return tasks
if __name__ == "__main__":
context = ExitStack()
tasks = create_tasks(..., context)
with context:
loop.run_until_complete(asyncio.gather(*tasks))
你沒有真正解釋你需要什么樣的任務,一個簡單的獲取?
更復雜的東西?
您是否希望每個用戶名/密碼具體?
你需要最后保存所有回復嗎?
對於此代碼,我假設用戶名/密碼無關緊要,但它可以快速更改。
而不是你如何分別啟動會話我使用了消費者/生產者模式。
每個消費者與上下文管理器會話,也不需要信號量(因為隊列)。
import asyncio
from concurrent.futures import ProcessPoolExecutor
from aiohttp import ClientSession
from bs4 import BeautifulSoup
LOGIN_URL = "http://example.com/login"
CLIENT_CNT = 10
proc_pool = ProcessPoolExecutor(CLIENT_CNT)
def get_key(text):
soup = BeautifulSoup(text, "html.parser")
form = soup.find("form")
key = form.find("input", attrs={"type": "hidden", "name": "authenticityToken"})
return key.get("value", None)
async def init_consumer(username: str, password: str, loop, queue):
loop = loop or asyncio.get_event_loop()
async with ClientSession(loop=loop) as session:
# init the session with creds? i you didn't use the username/password
async with session.get(LOGIN_URL) as login_resp:
x = await asyncio.ensure_future(loop.run_in_executor(proc_pool, get_key, await login_resp.text()))
print(x)
url = await queue.get()
while url is not None:
# Do things with session and queue
async with session.get(url) as resp:
rsp_as_txt = await resp.text()
queue.task_done()
url = await queue.get()
async def generate_tasks(queue):
tasks = ["http://www.example.com" for i in range(20)]
# putting all tasks in queue
for task in tasks:
await queue.put(task)
# waiting for all tasks to finish
queue.join()
# Telling consumer to finish process
for i in range(queue.maxsize):
queue.put(None)
async def run(loop):
queue = asyncio.Queue(CLIENT_CNT)
usernames = ("a", "b", "c", "d", "e", "f", "g")
passwords = ("a", "b", "c", "d", "e", "f", "g")
consumers = [asyncio.ensure_future(init_consumer(u, p, loop, queue)) for u, p in zip(usernames, passwords)]
return await generate_tasks(queue)
if __name__ == "__main__":
loop = asyncio.get_event_loop()
loop.run_until_complete(run(loop=loop))
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.