process-backup.py

#!/usr/bin/env python

from os import listdir,path,remove
from collections import namedtuple
from time import strptime
import datetime
import string
import shutil

BackupRange = namedtuple('BackupRange', ['daily', 'weekly', 'monthly'])

class BackupFile:
    PREFIX = 'AUTOBACKUP'
    SEP    = '-'
    EXT    = '.tar.gz'
    PATH   = '/srv/backup/'

    def __init__(self, volume, date, time, ext):
        self.volume = volume
        self.date   = date
        self.time   = time
        self.ext    = ext

    def __repr__(self):
        return string.join([
            BackupFile.PREFIX,
            self.volume,
            self.date.isoformat().replace('-', '.'),
            self.time.strftime("%H.%M.%S")
        ], BackupFile.SEP) + self.ext

def retrieve_file_list():
    files = []

    for f in (f for f in listdir(BackupFile.PATH) if path.isfile(path.join(BackupFile.PATH, f)) and f.endswith(BackupFile.EXT)):
        f = f.strip(BackupFile.EXT)
            
        if f.startswith(BackupFile.PREFIX) and f.count(BackupFile.SEP) == 3:
            prefix, fvolume, fdate, ftime = f.split(BackupFile.SEP)
        
            ts    = strptime(fdate + " " + ftime, "%Y.%m.%d %H.%M.%S")
            fdate = datetime.date(ts.tm_year, ts.tm_mon, ts.tm_mday)
            ftime = datetime.time(ts.tm_hour, ts.tm_min, ts.tm_sec)
    
            files.append(BackupFile(fvolume, fdate, ftime, BackupFile.EXT))
        
    return files

def copy_files(to, files):
    for idx,f in enumerate(files):
        print str(idx+1) + ') Copying ' + str(f) + ' ...'
        shutil.copy(path.join(BackupFile.PATH, str(f)), path.join(BackupFile.PATH, to, str(f)))

def delete_files(files):
    for idx,f in enumerate(files):
        print str(idx+1) + ') Deleting ' + str(f) + ' ...'
        remove(path.join(BackupFile.PATH, str(f)))
    
def get_limits():
    today = datetime.date.today()
    last_monday = today - datetime.timedelta(today.weekday())

    daily   = last_monday - datetime.timedelta(7)
    weekly  = daily - datetime.timedelta(0, 0, 0, 0, 0, 0, 4)
    monthly = weekly - datetime.timedelta(0, 0, 0, 0, 0, 0, 53) 
    
    return BackupRange(daily=daily, weekly=weekly, monthly=monthly)
    
def filter_files_by_limits(files, limits):
    daily   = [ f for f in files if f.date >= limits.daily ]
    weekly  = [ f for f in files if f.date >= limits.weekly and f.date < limits.daily and f.date.weekday() == 0 ]
    monthly = [ f for f in files if f.date >= limits.monthly and f.date < limits.weekly and f.date.day == 1 ]

    return BackupRange(daily=daily, weekly=weekly, monthly=monthly)


files  = retrieve_file_list()
limits = get_limits()

need_copy  = filter_files_by_limits(files, limits)
copy_union = need_copy.daily + need_copy.weekly + need_copy.monthly
need_del   = [ f for f in files if f not in copy_union ]

print str(len(copy_union)) + " file(s) need to be copied."
copy_files("daily/", need_copy.daily)
copy_files("weekly/", need_copy.weekly)
copy_files("monthly/", need_copy.monthly)

print
print str(len(files)) + " file(s) need to be deleted."
delete_files(files)