#!/usr/bin/env python
#-*- coding:utf-8 -*-
#
#
# author : cold night
# email  : wh_linux@126.com
#

import pymongo
import re
from StringIO import StringIO

conn = pymongo.Connection()
db = conn.test

def replace_url():
  regex = re.compile(r'([href|src])=["|\']http://.*"|\']')
  results = db['test'].find()
  db_coll = db['test']
  def replace(r):
    content = r.get('content')
    if not content: return
    content = StringIO(content)
    content.seek(0)
    result = StringIO()
    for line in content.readlines():
      t = regex.sub(r'\1="\2"', line)
      result.write(t)

    result.seek(0)
    content = result.read()
    if content:
      r['content'] = content
    _id = r.get('_id')
    db_coll.update({'_id':_id}, r)

  results = [replace(i) for i in results]

if __name__=="__main__":replace_url()