#!/usr/bin/env python
"""
gooshparse.py - Google Search History RSS Parser
Author: Sean B. Palmer, inamidst.com
"""

import sys, time, xml.dom.minidom

def text(element): 
   return element.firstChild.wholeText

def main(): 
   f = open(sys.argv[1])
   bytes = f.read()
   f.close()

   doc = xml.dom.minidom.parseString(bytes)

   for item in doc.getElementsByTagName('item'): 
      date = text(item.getElementsByTagName('pubDate')[0])
      date = time.strptime(date, '%a, %d %b %Y %H:%M:%S GMT')
      date = time.strftime('%Y-%m-%dT%H:%M:%SZ', date)

      title = text(item.getElementsByTagName('title')[0])
      title = title.strip(' \t')
      category = text(item.getElementsByTagName('category')[0])
      category = ('Q', 'R')['result' in category]

      if category == 'R': 
         link = text(item.getElementsByTagName('link')[0])
         link = link.strip(' \t').replace(' ', '%20')
         print date, category, link, title.encode('utf-8')
      else: print date, category, title.encode('utf-8')

if __name__ == '__main__': 
   main()