BackupPC_archive_host_B2/BackupPC_archiveHost_s3

#!/usr/bin/python -W ignore::DeprecationWarning
# A BackupPC script to archive a host's files to Amazon S3.
#
# Point $Conf{ArchiveClientCmd} at me.
# Requires python-boto
#
# Usage: BackupPC_archiveHost tarCreatePath splitPath parPath host bkupNum \
#             compPath fileExt splitSize outLoc parFile share
#
# Create secrets.py such that it has:
# accesskey = 'amazon aws access key'
# sharedkey = 'amazon aws shared key'
# gpgsymmetrickey = 'gpg symmetric key -- make it good, but do not lose it'

import glob
import os
import secrets
import sys
import time

from subprocess import *

from boto.s3.connection import S3Connection
from boto.s3.key import Key
import boto.exception

import logging
import logging.handlers

logger = logging.getLogger('')
loghandler = logging.handlers.SysLogHandler('/dev/log',
                facility=logging.handlers.SysLogHandler.LOG_DAEMON)
logformatter = logging.Formatter('%(filename)s: %(levelname)s: %(message)s')
loghandler.setFormatter(logformatter)
logger.addHandler(loghandler)
logger.setLevel(logging.DEBUG)

# functions
def is_exe(fpath):
    return os.path.exists(fpath) and os.access(fpath, os.X_OK)

def encrypt_file(filename, key, compress='/bin/cat'):
    compressmap = {'cat': 'none', 'gzip': 'ZLIB', 'bzip2': 'BZIP2'}
    if os.path.basename(compress) in compressmap.keys():
        compress_algo = compressmap[os.path.basename(compress)]
    else:
        compress_algo = 'none'

    cmd =  ['/usr/bin/gpg', '--batch', '--no-tty']
    cmd.extend(['--compress-algo', compress_algo])
    cmd.extend(['--output', '%s.gpg' % filename])
    cmd.extend(['--passphrase-fd', '0'])
    cmd.extend(['--symmetric', filename])

    if is_exe(cmd[0]):
        logging.info('Encrypting %s (compression: %s)' % (filename, compress_algo))
        logging.debug(`cmd`)
    else:
        logging.error('%s is not an executable file!' % cmd[0])

    proc = Popen(cmd, stdin=PIPE, stdout=PIPE)
    proc.communicate(key)

    if os.path.exists(filename + '.gpg'):
        oldfilesize = os.path.getsize(filename)
        newfilesize = os.path.getsize(filename + '.gpg')
        compressed = ((oldfilesize - newfilesize) / float(oldfilesize)) * 100
        logging.info('%s shrunk by %.2f%% (%i -> %i bytes)' % (filename, compressed, oldfilesize, newfilesize))
        os.unlink(filename)
        return filename + '.gpg'
    else:
        logging.error('%s.gpg does not exist' % filename)
        raise Exception

def open_s3(accesskey, sharedkey, host):
    conn = S3Connection(accesskey, sharedkey, is_secure=False)
    mybucketname = (accesskey + '-bkup-' + host).lower()
    try:
        bucket = conn.get_bucket(mybucketname)
    except boto.exception.S3ResponseError:
        logging.info('Creating bucket %s' % mybucketname)
        bucket = conn.create_bucket(mybucketname)
    bucket.set_acl('private')
    return bucket

def handle_progress(transmitted, pending):
    logging.debug('%i of %i bytes transmitted (%.2f%%)' % (transmitted, pending, (transmitted/float(pending))*100))

def send_file(bucket, filename):
    if bucket.get_key(filename):
        logging.error('Duplicate filename %s! I hope that is OK.' % filename)
    k = Key(bucket)
    k.key = os.path.basename(filename)
    logging.info('Uploading %s...' % os.path.basename(filename))

    k.set_contents_from_filename(filename, cb=handle_progress)

    return k
 
# Read in arguments
if len(sys.argv) != 12:
    sys.stderr.write("Usage: %s tarCreatePath splitPath parPath host bkupNum \
          compPath fileExt splitSize outLoc parFile share\n" % sys.argv[0])
    sys.exit(1)
else:
    tarCreate   = sys.argv[1]
    splitPath   = sys.argv[2]
    parPath     = sys.argv[3]
    host        = sys.argv[4]
    bkupNum     = int(sys.argv[5])
    compPath    = sys.argv[6]
    fileExt     = sys.argv[7]
    splitSize   = int(sys.argv[8])
    outLoc      = sys.argv[9]
    parfile     = sys.argv[10]
    share       = sys.argv[11]

for i in [tarCreate, compPath, splitPath, parPath]:
    if i is not '' and not is_exe(i):
        sys.stderr.write('Error: %s is not an executable program\n' % i)
        sys.exit(1)

# open s3 connection
bucket = open_s3(secrets.accesskey, secrets.sharedkey, host)

mesg = "Writing archive for host %s, backup #%i" % (host, bkupNum)
if splitSize > 0 and is_exe(splitPath):
    mesg += ', split into %i byte chunks' % splitSize
if secrets.gpgsymmetrickey:
    mesg += ', encrypted with secret key'

logging.info(mesg)
print mesg

# Prepare the pipeline
if share == '*':
    share = '\*'

cmd = '%s -t -h %s -n %i -s %s . ' % (tarCreate, host, bkupNum, share)

if splitSize > 0 and is_exe(splitPath):
    filehead = '%s/%s.%i.tar.' % (outLoc, host, bkupNum)
    fileglob = filehead + '*'
    cmd += '| %s -b %i - %s' % (splitPath, splitSize, filehead)
else:
    fileglob = '%s/%s.%i.tar' % (outLoc, host, bkupNum)
    cmd += '> %s' % fileglob

logging.debug('Executing %s' % cmd)

returncode = os.system(cmd)

if returncode != 0:
    logger.error('%s died with exit code %i' % (cmd, returncode))
    sys.exit(1)

logging.info('Beginning post-processing of %s #%i' % (host, bkupNum))

for i in sorted(glob.glob(fileglob)):
    if secrets.gpgsymmetrickey:
        sendfile = encrypt_file(i, secrets.gpgsymmetrickey, compPath)
    else:
        sendfile = i

    key = send_file(bucket, sendfile)
    key.set_metadata('backuppc_cmd', cmd)
    key.set_metadata('backuppc_mesg', mesg)
    key.set_acl('private')
    key.close()

    os.unlink(sendfile)