13693252552
13296629989
13640810839
15755106631
15119622732
13904446048
18874791953
13695658500
13695658547
15950179080
15573462779
15217624651
15018485989
13706522482
13666519777
13666515188
18857287528
15575394501

# coding=UTF-8
# get provider information by phoneNumber
from urllib import urlopen
import re
# get html source code for url
def getPageCode(url):
  file = urlopen(url)
  text = file.read()
  file.close()
#  text = text.decode("utf-8")   # depending on coding of source code responded
  return text
# parse html source code to get provider information
def parseString(src, result):
  pat = []
  pat.append('(?<=歸屬地：</span>).+(?=<br />)')
  pat.append('(?<=卡類型：</span>).+(?=<br />)')
  pat.append('(?<=運(yùn)營(yíng)商：</span>).+(?=<br />)')
  pat.append('(?<=區(qū)號(hào)：</span>)\d+(?=<br />)')
  pat.append('(?<=郵編：</span>)\d+(?=<br />)')
  item = []
  for i in range(len(pat)):
    m = re.search(pat[i], src)
    if m:
      v = m.group(0)
      item.append(v)
  return item
# get provider by phoneNum
def getProvider(phoneNum, result):
  url = "http://www.sjgsd.com/n/?q=%s" %phoneNum
  text = getPageCode(url)
  item = parseString(text, result)
  result.append((phoneNum, item))
# write result to file
def writeResult(result):
  f = open("result.log", "w")
  for num, item in result:
    f.write("%s:\t" %num)
    for i in item:
      f.write("%s,\t" %i)
    f.write("\n")
  f.close()
if __name__ == "__main__":
  result = []
  for line in open("test.txt", "r"):
    phoneNum = line.strip(" \t\r\n")
    getProvider(phoneNum, result)
    print("%s is finished" %phoneNum)
  writeResult(result)