Python crawlers crawling printer cartridges and component surplus

# 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 & nbsp; (. +? % )'

Car = re. compile (cartridge)

Cartridge = re. findall (car, html)

Rekit = r'maintenance Kit & nbsp; (. +? % )'

Kit = re. compile (rekit)

Kit = re. findall (kit, html)

Print "Cartridge = % s" % Cartridge,

Print "Kit = % s" % Kit

Def getData_TFNRCPRT02 (html ):

Cartridge = R' & nbsp; (. +? % )'

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: // 10.135690.171/hp/device/this. LCDispatcher ': 'galprt04 ',

'Http: // 10.135690.172/hp/device/this. LCDispatcher ': 'galprt05 ',

'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"