Linux教程之PY IE 主动化点击东西 - 仓酷云

#coding:gbkimportsys,os,shutilimportre,_winreg,globimportcookielibimporturllibimporturllib2fromurllib2importURLError,HTTPErrorfromPAM30importPAMIEfromtimeimportsleepfromrandomimportrandintDEBUG=Truekeyword=raw_input("请输出准确关头词")#注:在利用IE7以上的扫瞄器须将是不是封闭一切选项卡的勾点上不然会形成步伐逝世轮回#================扫除COOKIE================不然百度不管帐算会见,请在步伐第67行修正本人电脑路劲defCleanDir(Dir):ifos.path.isdir(Dir):paths=os.listdir(Dir)forpathinpaths:filePath=os.path.join(Dir,path)ifos.path.isfile(filePath):try:os.remove(filePath)exceptos.error:print"OK"elifos.path.isdir(filePath):iffilePath[-4:].lower()==".svn".lower():continueshutil.rmtree(filePath,True)returnTrue#改换代办署理IP。间接修正WINDOWS注册表defalterip(ip=None):key=_winreg.OpenKey(_winreg.HKEY_CURRENT_USER,r"SoftwareMicrosoftWindowsCurrentVersionInternetSettings",0,_winreg.KEY_ALL_ACCESS)value,type=_winreg.QueryValueEx(key,"ProxyServer")ifip!=None:_winreg.SetValueEx(key,ProxyEnable,0,_winreg.REG_SZ,"1")if_winreg.SetValueEx(key,ProxyServer,0,_winreg.REG_SZ,ip):print"IP改换失利"else:print"改换乐成"_winreg.CloseKey(key)returnTrue#==============IE点击模块#defiepam(keyword=None):ie=PAMIE(timeOut=300)ie._ie.Navigate(http://www.baidu.com)#==============关头词模块#ifie.setTextBox("wd",keyword)!=False:ifie.clickButton("su")!=None:sleep(3)ifie.getLink(unicode("灵动收集","cp936"))!=None:ie.clickLink(ie.getLink(unicode("灵动收集","cp936")))#else:#ie.clickLink(ie.getLink(unicode("[%s]"%i,"cp936")))#ie.clickLink(ie.getLink(unicode("[%s]"%p,"cp936")))#ie.clickLink(ie.getLink(unicode("灵动收集-办事器租用托管,VPS,假造主机,域名注册,企业邮局,收集...","cp936")))#i+=1#=============随机停止工夫=======sleep(randint(5,20))#=============准确填写本人电脑的COOKIE路劲#CleanDir("C:UsersliyunAppDataRoamingMicrosoftWindowsCookies")os.system(taskkill/IMiexplore.exe)#html页面下载函数defgetHtml(url,post_data=None,cookie=None):"""Fetchthetargethtmlurl-URLtofetchpost_data-POSTEntitycookie-CookieHeader"""ifDEBUG:print"getHtml:",urlresult=try:#createarequestrequest=urllib2.Request(url)cj=cookielib.CookieJar()#changeUser-Agentrequest.add_header("accept","*/*")request.add_header(User-Agent,Mozilla/5.0)#changeReferrerrequest.add_header(referer,http://www.51proxied.com/index.html)#ifhascookie,addcookieheaderifcookie:request.add_header(Cookie,cookie)#createaopeneropener=urllib2.build_opener(urllib2.HTTPCookieProcessor(cj))#ifhaspostentityifpost_data:#encodepostdatapost_data=urllib.urlencode(post_data)response=opener.open(request,post_data)else:response=opener.open(request)result=response.read()response.close()#nocontent,dontsaveifnotresultorlen(result)==0:returnreturnresultexceptHTTPError,e:ifDEBUG:printErrorretrievingdata:,eprintServererrordocumentfollows:
#printe.read()returnexceptURLError,e:ifhasattr(e,reason):ifDEBUG:printFailedtoreachaserver.printReason:,e.reasonreturnelifhasattr(e,code):ifDEBUG:printTheservercouldn fulfilltherequest.printErrorcode:,e.codereturnexceptException,e:ifDEBUG:printereturn#必要考证的代办署理列表proxy_urls=[]proxy_urls.append({url:http://www.51proxied.com/http_anonymous.html,type:http_anonymous})#proxy_urls.append({url:http://www.51proxied.com/http_non_anonymous.html,type:http_transparent})importreimportsocketimporttimeimportthreadingresult=[]#线程同步锁lock=threading.Lock()defsynchronous(f):defcall(*args,**kwargs):lock.acquire()try:returnf(*args,**kwargs)finally:lock.release()returncall#先猎取一切待考证的代办署理proxies=[]forproxy_urlinproxy_urls:html=getHtml(proxy_url[url])#正则婚配猎取每代办署理rs=re.compile(r<tr.*?>[sS]*?<td.*?>d+?</td>[sS]*?<td>(S+?)</td>[sS]*?<td.*?>(S+?)</td>[sS]*?<td.*?>(S+?)</td>[sS]*?</tr>,re.DOTALL).findall(html)forrinrs:proxy={}#代办署理域名proxy[domain]=r[0]#代办署理端口proxy[port]=r[1]#代办署理国度proxy[state]=r[2]#代办署理范例proxy[type]=proxy_url[type]#呼应工夫proxy[time]=0ifnot(proxyinproxies):proxies.append(proxy)#猎取一个待考证代办署理@synchronousdefgetproxy():globalproxiesiflen(proxies)>0:returnproxies.pop()else:return#保留考证了局@synchronousdefsaveresult(proxy):globalresultifnot(proxyinresult):result.append(proxy)#线程函数defverify():while1:proxy=getproxy()#一切代办署理均已考证终了iflen(proxy)==0:returnprint"正在考证：%s,%s"%(proxy[domain],proxy[port])#考证代办署理的可用性#创立一个TCP毗连套接字sock=socket.socket(socket.AF_INET,socket.SOCK_STREAM)#设置10超时sock.settimeout(10)try:start=time.clock()#毗连代办署理办事器sock.connect((proxy[domain],int(proxy[port])))proxy[time]=int((time.clock()-start)*1000)sock.close()saveresult(proxy)print"%s,%s考证经由过程，呼应工夫：%dms."%(proxy[domain],proxy[port],proxy[time])exceptException,e:ifDEBUG:printeprint"%s,%s考证失利."%(proxy[domain],proxy[port])#initthread_poolthread_pool=[]foriinrange(20):th=threading.Thread(target=verify,args=());thread_pool.append(th)#startthreadsonebyoneforthreadinthread_pool:thread.start()#collectallthreadsforthreadinthread_pool:threading.Thread.join(thread)#了局按呼应工夫从小到年夜排序result.sort(lambdax,y:cmp(x[time],y[time]))foriteminresult:ifitem[time]<60:ip=%s:%s%(item[domain],item[port])printipifalterip(ip):iepam(keyword)

复制代码