#!/usr/bin/python
#
# Author: Tobias Galitzien <tg@trusttheadmin.de>
# Copyright (c) 2012 Tobias Galitzien <tg@trusttheadmin.de>
# Berlin, Germany
#

import sys
import boto.s3.connection
import argparse
import time, datetime, dateutil.parser

now = datetime.datetime.utcnow()

def bailout(msg):
  sys.stderr.write(msg)
  sys.exit(-1)

def is_obsolete(dt):
  try:
    age = (now-dt).days
    if age <= 7: return False
    if age < 35 and dt.day % 7 == 0: return False
    return True
  except:
    return False

def determine_filedate(key, pattern=None):
  if pattern:
    try:
      t = time.strptime(key.name, pattern)
      return datetime.datetime(t.tm_year, t.tm_mon, t.tm_mday, t.tm_hour, t.tm_min, t.tm_sec)
    except ValueError:
      return None
  else:
    return dateutil.parser.parse(key.last_modified).replace(tzinfo=None)

# parse command line
parser = argparse.ArgumentParser(description = 'delete obsolete archives on s3')
parser.add_argument('-b','--bucket', required = True)
parser.add_argument('-p','--pattern', help = 'use strptime() pattern to determine file date')
parser.add_argument('-m','--mtime', action = 'store_true', help = 'use mtime for file date')
parser.add_argument('-v','--verbose', type = int, default = 0, help = 'set boto debug level')
parser.add_argument('-e','--endpoint', default = 's3.amazonaws.com', help = 'set s3 endpoint (s3-eu-west-1.amazonaws.com for EU buckets, see http://docs.amazonwebservices.com/general/latest/gr/rande.html#s3_region)')
parser.add_argument('-d','--dryrun', action = 'store_true', help = 'don\'t actually delete anything, just show what would be done')
args = parser.parse_args()

# check mode
if not (args.pattern or args.mtime):
  print "need either a pattern or the mtime switch"
  parser.print_help()
  sys.exit()

# connect to s3
try:
  s3 = boto.s3.connection.S3Connection(debug = args.verbose, host = args.endpoint)
except boto.exception.S3ResponseError, e:
  bailout("cannot connect to S3: %s\n" % e.code)

# open bucket
try:
  bucket = s3.get_bucket(args.bucket)
except boto.exception.S3ResponseError, e:
  bailout("cannot access bucket %s: %s\n" % (args.bucket,e.code))

for k in bucket.list():
  n = k.name
  dt = determine_filedate(k, args.pattern)
  if dt:
    if is_obsolete(dt):
      if args.dryrun:
        print 'would delete %s, but dryrun mode' % n
      else:
        if args.verbose:
          print 'deleting %s' % n
        k.delete()
  else:
    if args.verbose:
      print 'cannot parse %s' % n