Python3并发检验代理池地址

#encoding=utf-8
#author: walker
#date: 2016-04-14
#summary: 用线程池并发检验代理有效性

import os, sys, time
import requests
from concurrent import futures

cur_dir_fullpath = os.path.dirname(os.path.abspath(__file__))

Headers = {
'Accept': '*/*',
'User-Agent': 'Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 6.1; WOW64; Trident/4.0; SLCC2; .NET CLR 2.0.50727; .NET CLR 3.5.30729; .NET CLR 3.0.30729; .NET4.0C; .NET4.0E)',
}

#检验单个代理的有效性
#如果有效，返回该proxy；否则，返回空字符串
def Check(desturl, proxy, feature):
proxies = {'http': 'http://' + proxy}
proxies = {
'http': proxy,
'https': proxy
}

r = None	#声明
exMsg = None
try:
r = requests.get(url=desturl, headers=Headers, proxies=proxies, timeout=3)
except:
exMsg = '* ' + traceback.format_exc()
#print(exMsg)
finally:
if 'r' in locals() and r:
r.close()

if exMsg:
return ''

if r.status_code != 200:
return ''

if r.content.decode('utf8').find(feature) < 0:
return ''

return proxy

#输入代理列表(set/list)，返回有效代理列表
def GetValidProxyPool(rawProxyPool, desturl, feature):
validProxyList = list()	#有效代理列表

pool = futures.ThreadPoolExecutor(8)
futureList = list()
for proxy in rawProxyPool:
futureList.append(pool.submit(Check, desturl, proxy, feature))

print('\n submit done, waiting for responses\n')

for future in futures.as_completed(futureList):
proxy = future.result()
print('proxy:' + proxy)
if proxy:	#有效代理
validProxyList.append(proxy)

print('validProxyList size:' + str(len(validProxyList)))

return validProxyList

#获取原始代理池
def GetRawProxyPool():
rawProxyPool = set()

#通过某种方式获取原始代理池......

return rawProxyPool

if __name__ == "__main__":
rawProxyPool = GetRawProxyPool()
desturl = 'http://...'		#需要通过代理访问的目标地址
feature = 'xxx'		#目标网页的特征码
validProxyPool = GetValidProxyPool(rawProxyPool, desturl, feature)

#encoding=utf-8
#author: walker
#date: 2017-03-28
#summary: 用协程并发检验代理有效性
#Python sys.version：3.6.1 (v3.6.1:69c0db5, Mar 21 2017, 18:41:36) [MSC v.1900 64 bit (AMD64)]

import os, sys, time
import requests
import aiohttp
import asyncio
import traceback

Headers = {
'Accept': '*/*',
'User-Agent': 'Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 6.1; WOW64; Trident/4.0; SLCC2; .NET CLR 2.0.50727; .NET CLR 3.5.30729; .NET CLR 3.0.30729; .NET4.0C; .NET4.0E)',
}

#检验单个代理的有效性
#如果有效，返回该proxy；否则，返回空字符串
async def Check(desturl, proxy, feature):
proxy = 'http://' + proxy
#print('proxy:' + proxy)

exMsg = None
try:
async with aiohttp.ClientSession() as session:
async with session.get(desturl, headers=Headers, proxy=proxy, timeout=10) as resp:
#print(resp.status)
assert resp.status == 200
#print(await resp.text())
html = await resp.text(encoding='utf-8')

except:
exMsg = '* ' + traceback.format_exc()
#print(exMsg)

if exMsg:
return ''

if html.find(feature) < 0:
return ''

return proxy

#输入代理列表(set/list)，返回有效代理列表
async def GetValidProxyPool(rawProxyPool, desturl, feature):
print('GetValidProxyPool ...')
validProxyList = list()	#有效代理列表

coroList = list()
for proxy in rawProxyPool:
coroList.append(asyncio.ensure_future((Check(desturl, proxy, feature))))

totalSleepTime = 0
for f in asyncio.as_completed(coroList):
proxy = await f
#print('rtn proxy:' + proxy)
if proxy:
validProxyList.append(proxy)

print('validProxyList size: %d' % len(validProxyList))

return validProxyList

#获取原始代理池
def GetRawProxyPool():
rawProxyPool = set()

#通过某种方式获取原始代理池......

return rawProxyPool

if __name__ == "__main__":
startTime = time.time()
rawProxyPool = GetRawProxyPool()
desturl = 'http://...'        #需要通过代理访问的目标地址
feature = 'xxx'       #目标网页的特征码

print('rawProxyPool size:%d' % len(rawProxyPool))
loop = asyncio.get_event_loop()
validProxyList = loop.run_until_complete(GetValidProxyPool(rawProxyPool, desturl, feature))
loop.close()
print('rawProxyPool size:%d' % len(validProxyList))
print('time cost:%.2fs' % (time.time()-startTime))