python linux 打印机,Python抓取打印机硒鼓和组件剩余量的爬虫

#coding:utf-8

import re,os,urllib,string

def getHtml(url):

page = urllib.urlopen(url)

html = page.read()

return html

def getData(html):

cartridge = r'Black Print Cartridge  (.+?%)'

car = re.compile(cartridge)

Cartridge = re.findall(car,html)

rekit = r'Maintenance Kit  (.+?%)'

kit = re.compile(rekit)

Kit = re.findall(kit,html)

print "Cartridge=%s" % Cartridge,

print "Kit=%s" % Kit

def getData_TFNRCPRT02(html):

cartridge = r'  (.+?%)'

car = re.compile(cartridge)

Cartridge = re.findall(car,html)

print "Cartridge=%s" % Cartridge

def IpStatus(ip):

Status = os.system("ping" + " " + ip)

return Status

def getIp(url):

getIp = r'http://(.+?)/'

GetIp = re.compile(getIp)

IpAddress = re.findall(GetIp,url)

ip = ''.join(IpAddress)

return ip

UrlDir = {

'http://10.132.160.171/hp/device/this.LCDispatcher' : 'LegalPrt04',

'http://10.132.160.172/hp/device/this.LCDispatcher' : 'LegalPrt05',

'http://10.153.24.177/hp/device/this.LCDispatcher'  : 'SzPrt11'   ,

'http://10.153.24.172/hp/device/this.LCDispatcher'  : 'SzPrt13'   ,

'http://10.153.26.179/hp/device/this.LCDispatcher'  : 'SzPrt14'   ,

'http://10.186.65.156/hp/device/this.LCDispatcher'  : 'TfnrcPrt01',

'http://10.186.65.101/' : 'TfnrcPrt3390',

'http://10.153.25.200/' : 'SzPrt16'

}

for url in UrlDir:

ip = getIp(url)

Status = IpStatus(ip)

if Status == 0:

if url == 'http://10.186.65.101/':

html = getHtml(url)

print UrlDir[url]+":",

getData_TFNRCPRT02(html)

elif url == 'http://10.153.25.200/':

print "SZPRT16 is OK"

else:

html = getHtml(url)

print UrlDir[url]+":",

getData(html)

else:

print UrlDir[url] + ":"+"Ping fail"