#!/usr/bin/python

import urllib
import re

url = "http://realestate.sfgate.com/RealEstate/openhouses/SearchResults.asp?RoS=S&RegionIds=14&RegionIdCount=1&SaveField=&NewLflag=&BRooms=60&LPrice=750000&HPrice=950000&BType=0&Amenities=0&cid=999&SCountyId=0&AdTagCountyId=&sRegionId=&CountyId=&CountyIdCount=&CountyIds=10%2C15%2C11%2C16%2C12%2C18%2C14&p=C&SOrder=1&KeyWord=&SearchFor=&SearchBy=&SearchFlag=&SearchName=Enter+Search+Name&EmailAlert=&EmailType=&ReportSection=O&WebSearches=Y"


nextPageFormRegEx = re.compile("<form method=\"POST\" name=\"TheNextPage\" action=\"SearchResults.asp\">")
nextPageFormInputRegEx = re.compile("<input type=\"hidden\" name=\"(?P<name>[^\"]+)\" value=\"(?P<value>[^\"]+)\">")

pageFormState = 0

page = urllib.urlopen(url)
for line in page:
#  print line

  if pageFormState == 0:
    if nextPageFormRegEx.match(line):
      print "**********************************"
      print line
      pageFormState = 1

  elif pageFormState == 1:
    print line
    result = nextPageFormInputRegEx.match(line)
    if result:
      print "xxxxxxxxxxxxxxxxxxx"
      print result.group("name")
      print result.group("value")
      
  else:
    none