對於python多線程與單線程之間有多少差距呢?今天用一個小例子比較一下。 說明:爬取代理ip網站ip並用代理ip訪問某網站,看執行後的用時多少。 單線程版
# -*- coding: utf-8 -*- import requests from bs4 import BeautifulSoup as bs import re import time def proxy_list(mbUrl): headers={'User-Agent':"Mozilla/5.0 (Windows NT 6.3; WOW64; rv:51.0) Gecko/20100101 Firefox/51.0"} url='http://www.xicidaili.com/nn/' r=requests.get(url=url,headers=headers) soup=bs(r.content) datas=soup.find_all(name='tr',attrs={'class':re.compile('|[^odd]')}) i = 0 for data in datas: soup_proxy_content=bs(str(data)) soup_proxys=soup_proxy_content.find_all(name='td') #for i in[1,2,5]: # print soup_proxys[i].string ip=str(soup_proxys[1].string) port=str(soup_proxys[2].string) types=str(soup_proxys[5].string) proxy_test(mbUrl,ip,port,types) #thread.start_new_thread(proxy_test,(mbUrl,ip,port,types)) def proxy_test(url,ip,port,types): proxy={} proxy[types.lower()]='%s:%s'%(ip,port) #proxy={'http':'106.46.136.24:808'} try: r=requests.get(url,proxies=proxy,timeout=3) #會把每個代理ip都測試一遍,超時設置爲六秒 ip_content=re.findall(r'\[(.*?)\]',r.text)[0]#匹配[]中的ip #print r.text if ip==ip_content:#判斷代理是否測試成功 #lock.acquire()#線程鎖 print proxy #lock.release() except Exception,e: #print e pass #測試一下 if __name__=='__main__': time_start=time.time() #lock = thread.allocate_lock() # 定義一個線程鎖 proxy_list("http://1212.ip138.com/ic.asp") time_finished=time.time()-time_start print time_finished
運行結果:
{'http': '39.85.13.253:9999'}
{'http': '113.124.9.229:8998'}{'http': '124.88.67.52:843'}
{'http': '123.132.170.185:9999'}
{'http': '123.132.179.153:9999'}
{'http': '106.46.136.145:808'}
{'http': '111.124.245.121:9999'}
{'http': '218.17.43.228:3128'}
{'http': '58.23.103.106:9999'}
{'http': '42.84.103.140:9999'}
{'http': '140.240.245.167:9999'}
{'http': '106.46.136.96:808'}
85.9160001278
-------------------------------------------------------------------
多線程版 # -*- coding: utf-8 -*- import requests from bs4 import BeautifulSoup as bs import re import threading,thread import time def proxy_list(mbUrl): headers={'User-Agent':"Mozilla/5.0 (Windows NT 6.3; WOW64; rv:51.0) Gecko/20100101 Firefox/51.0"} url='http://www.xicidaili.com/nn/' r=requests.get(url=url,headers=headers) soup=bs(r.content) datas=soup.find_all(name='tr',attrs={'class':re.compile('|[^odd]')}) datalen=len(datas) threads=[] #定義一個線程隊列 ip = [] port = [] types = [] for i in range(datalen): soup_proxy_content=bs(str(datas[i])) soup_proxys=soup_proxy_content.find_all(name='td') ip.append(str(soup_proxys[1].string)) port.append(str(soup_proxys[2].string)) types.append(str(soup_proxys[5].string)) for i in range(datalen): t=threading.Thread(target=proxy_test,args=(mbUrl,ip[i],port[i],types[i],)) threads.append(t) for i in range(datalen): threads[i].start() #time.sleep(0.2) for i in range(datalen): threads[i].join() def proxy_test(url,ip,port,types): proxy={} proxy[types.lower()]='%s:%s'%(ip,port) try: r=requests.get(url,proxies=proxy,timeout=3) #會把每個代理ip都測試一遍,超時設置爲六秒 ip_content=re.findall(r'\[(.*?)\]',r.text)[0]#匹配[]中的ip if ip==ip_content:#判斷代理是否測試成功 lock.acquire() # 線程鎖 print proxy lock.release() except Exception,e: #print e pass #測試一下 if __name__=='__main__': time_start=time.time() lock = thread.allocate_lock() # 定義一個線程鎖 proxy_list("http://1212.ip138.com/ic.asp") time_finished=time.time()-time_start print time_finished 運行結果: {'http': '123.132.170.185:9999'} {'http': '180.107.249.162:9999'} {'http': '110.244.202.201:9999'} {'http': '123.52.220.243:9999'} {'http': '121.31.143.237:8123'} {'http': '110.73.2.236:8123'} {'http': '106.46.136.24:808'} {'http': '110.72.41.108:8123'} {'http': '140.240.245.167:9999'} {'http': '218.17.43.228:3128'} {'http': '220.166.241.213:8118'} 3.36299991608
多線程版的僅用了3.3秒,而單線程版的卻用了86秒,足足差了26倍。這之間的差距一目瞭然。