#!/usr/bin/env python
"""
sarc - Archive Safari Browser History
Author: Sean B. Palmer, inamidst.com

Usage: 
   $ sarc yyyy-mm-dd
   $ sarc today

Output is: date, uri, title, written in plain utf-8 
encoded text to a file named history-yyyy-mm-dd.txt
"""

from __future__ import with_statement

import sys, datetime, subprocess
import os.path
import xml.parsers.expat

apple_epoch = 978307200

def history(): 
   fn = os.path.expanduser('~/Library/Safari/History.plist')
   command = ['PlistBuddy', '-x', '-c', 'Print', fn]
   p = subprocess.Popen(command, stdout=subprocess.PIPE, bufsize=1)
   return p.stdout

def parse(f, extents): 
   state = type('State', (), {})

   state.mode = None
   state.item = {}
   state.key = None
   state.string = None
   state.data = ''
   state.history = {}

   def register(uri, date, title): 
      uri = uri.replace(' ', '%20')
      apple_date = int(float(date))

      if (extents[0] <= apple_date) and (apple_date < extents[1]): 
         pydate = datetime.datetime.fromtimestamp(apple_epoch + apple_date)
         isodate = pydate.strftime('%Y-%m-%dT%H:%M:%SZ')
         state.history.setdefault(isodate, set()).add((uri, title))

   def start_element(name, attrs): 
      if (name == 'key') or (name == 'string'): 
         state.mode = name

   def end_element(name): 
      if name == state.mode: 
         setattr(state, name, state.data.strip())
         state.mode = None
         state.data = ''

      if name == 'string': 
         state.item[state.key] = state.string

      if state.item.has_key('') and \
           state.item.has_key('lastVisitedDate') and \
           state.item.has_key('title'): 
         uri = state.item[''].encode('utf-8')
         date = state.item['lastVisitedDate'].encode('utf-8')
         title = state.item['title'].encode('utf-8')
         register(uri, date, title)
         state.item = {}

   def character_data(data): 
      if state.mode: 
         state.data += data

   p = xml.parsers.expat.ParserCreate()
   p.StartElementHandler = start_element
   p.EndElementHandler = end_element
   p.CharacterDataHandler = character_data

   for line in f: 
      p.Parse(line, False)
   p.Parse('', True)

   return state

def date(arg): 
   if arg == 'today': 
      today = datetime.date.today()
      arg = today.strftime('%Y-%m-%d')

   def date_error(): 
      print 'Error: Date must be in format yyyy-mm-dd'
      sys.exit(1)

   if arg.count('-') != 2: 
      date_error()
   y, m, d = arg.split('-')

   try: y, m, d = int(y), int(m), int(d)
   except ValueError: 
      date_error()

   day = datetime.date(y, m, d)
   next_day = day + datetime.timedelta(1)

   a = int(day.strftime('%s')) - apple_epoch
   b = int(next_day.strftime('%s')) - apple_epoch
   return arg, (a, b)

def main(): 
   try: arg = sys.argv[1]
   except IndexError: 
      print __doc__.strip()
      sys.exit()

   day, extents = date(arg)
   fn = 'history-%s.txt' % day
   if os.path.exists(fn): 
      print 'Error: %s already exists, please move or delete it' % fn
      sys.exit(1)

   with history() as f: 
      state = parse(f, extents)

   lines = 0
   o = open(fn, 'w')
   for isodate in sorted(state.history.keys()): 
      for uri, title in state.history[isodate]: 
         print >> o, isodate, uri, title
         lines += 1
   o.close()

   bytes = os.path.getsize(fn)
   print 'Wrote %s lines in %s bytes to %s' % (lines, bytes, fn)

if __name__ == '__main__': 
   main()