#!/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 hashlib
import os
import socket
import sys
import time

from multiprocessing import Process, Queue, cpu_count
from subprocess import *

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

import logging
import logging.handlers

import secrets

logger = logging.getLogger(__name__)

sysloghandler = logging.handlers.SysLogHandler('/dev/log',
                facility=logging.handlers.SysLogHandler.LOG_DAEMON)
syslogformatter = logging.Formatter('%(filename)s: %(levelname)s: %(message)s')
sysloghandler.setFormatter(syslogformatter)
logger.addHandler(sysloghandler)

consolehandler = logging.StreamHandler(sys.stdout)
consoleformatter = logging.Formatter('%(asctime)s: %(levelname)s: %(message)s')
consolehandler.setFormatter(consoleformatter)
logger.addHandler(consolehandler)

logger.setLevel(logging.DEBUG)

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

    proc = Popen(cmd, preexec_fn=lambda : os.nice(10), 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
        logger.info('%s shrunk by %.2f%% (%i -> %i bytes)' % (filename, compressed, oldfilesize, newfilesize))
        os.unlink(filename)
        return filename + '.gpg'
    else:
        logger.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:
        logger.info('Creating bucket %s' % mybucketname)
        bucket = conn.create_bucket(mybucketname)
    bucket.set_acl('private')
    return bucket

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

def send_file(bucket, filename):
    basefilename = os.path.basename(filename)
    if bucket.get_key(basefilename):
        logger.error('Duplicate filename %s! I hope that is OK.' % basefilename)
    k = Key(bucket)
    k.key = basefilename
    logger.info('Uploading %s...' % basefilename)
    fd = open(name=filename, mode='rb')
    putHeaders = {'x-amz-storage-class': 'REDUCED_REDUNDANCY'}
    k.set_contents_from_file(fd, headers=putHeaders, cb=handle_progress)
    return k

def encryption_worker(in_q, out_q):
    "Encrypts things from the in_q, puts them in the out_q"
    start_time = time.time()
    counter = 0
    for filename, gpgkey, comppath in iter(in_q.get, 'STOP'):
        counter += 1
        cryptstart_time = time.time()
        logger.info("Beginning encryption of %s.", filename)
        result = encrypt_file(filename, gpgkey, comppath)
        out_q.put(result)
        logger.info("Finished encryption of %s in %i seconds.", filename, time.time()-cryptstart_time)
    logger.info("Finished processing %i items in encryption queue in %i seconds.", counter, time.time()-start_time)
    out_q.put('STOP')

def sending_worker(in_q, accesskey, sharedkey, host):
    "Sends things from the in_q using the send_file method"
    start_time = time.time()
    counter = 0
    bucket = open_s3(accesskey, sharedkey, host)
    for filename in iter(in_q.get, 'STOP'):
        sending_start = time.time()
        counter += 1
        retry_count = 0
        max_retries = 10

        while retry_count <= max_retries:
            try:
                key = send_file(bucket, filename)
                key.set_acl('private')
                key.close()
                retry_count = max_retries+1
            except (boto.exception.S3ResponseError, socket.error), e:
                retry_count += 1
                sleeptime = 2**retry_count
                logger.error('Encountered exception %s, retrying in %i seconds (%i/%i)', e, sleeptime, retry_count, max_retries)
                time.sleep(sleeptime)

        size = os.path.getsize(filename)
        os.unlink(filename)
        sending_seconds = time.time() - sending_start

        bytespersecond = size / sending_seconds

        logger.info("Finished sending of %s in %i seconds, transfer speed %i bytes/second.", filename, sending_seconds, bytespersecond)

if __name__ == '__main__':
    # Read in arguments, verify that they match the BackupPC standard exactly
    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)

    beginning = time.time()

    if share == '*':
        share = '\*'

    # Is there already evidence of this having been done before?
    if glob.glob('%s/%s.*.tar.*' % (outLoc, host)):
        logger.info('Evidence of failed execution run prior!  Finishing it.')
        somefile = os.path.basename(glob.glob('%s/%s.*.tar.*' % (outLoc, host))[0])
        keyparts = somefile.split('.')
        encrypted = split = tarred = final = False
        if keyparts[-1] == 'gpg':
            keyparts.pop()
        if keyparts[-1] != 'tar' and len(keyparts[-1]) is 2:
            keyparts.pop()
        if keyparts[-1] == 'tar':
            keyparts.pop()

        bkupNum = int(keyparts.pop())

        filehead = '%s/%s.%i.tar.' % (outLoc, host, bkupNum)
        fileglob = filehead + '*'

        mesg = "Continuing upload 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'
        logger.info(mesg)
    else:
        mesg = "Writing archive for host %s, backup #%i" % (host, bkupNum)

        tarcmd = [tarCreate, '-t']
        tarcmd.extend(['-h', host])
        tarcmd.extend(['-n', bkupNum])
        tarcmd.extend(['-s', share])
        tarcmd.extend(['.'])

        splitcmd = None
        outfile = '%s/%s.%i.tar' % (outLoc, host, bkupNum)

        if splitSize > 0 and is_exe(splitPath):
            filehead = outfile + '.'
            fileglob = filehead + '*'
            splitcmd = [splitPath, '-b', splitSize, '-', filehead]
            mesg += ', split into %i byte chunks' % splitSize
        else:
            fileglob = outfile
            filehead = fileglob + '.'

        if secrets.gpgsymmetrickey:
            mesg += ', encrypted with secret key'

        logger.info(mesg)
        logger.debug('Executing tarcmd: %s > %s', ' '.join(tarcmd), outfile)

        outfp = open(outfile, 'wb')
        proc = Popen(tarcmd, preexec_fn=lambda : os.nice(10), stdout=outfile)
        proc.communicate()
        outfp.close()

        if splitcmd:
            logger.debug('Splitting file using splitcmd: %s', ' '.join(splitcmd))
            infp = open(outfile, 'rb')
            proc = Popen(splitcmd, preexec_fn=lambda : os.nice(10), stdin=infp)
            proc.communicate()
            infp.close()

    logger.info('Beginning post-processing of %i files from %s #%i' % (len(glob.glob(fileglob)), host, bkupNum))

    # Create queues for handling encryption and file transfers
    gpg_queue = Queue()
    send_queue = Queue()

    # Pre-run to check for artifacts
    for i in glob.glob(fileglob):
        if not i.endswith('.gpg') and os.path.exists(i + '.gpg'):
            logger.info("Orphaned GPG file exists: %s", i + '.gpg')
            os.unlink(i + '.gpg')

    # Run again to send files to the relevant queue
    for i in glob.glob(fileglob):
        if secrets.gpgsymmetrickey and not i.endswith('.gpg'):
            # A tar file, unencrypted, needs encrypted.
            logger.debug("Adding %s to gpg_queue", i)
            gpg_queue.put([i, secrets.gpgsymmetrickey, compPath])
        else:
            # either encryption is off, or the file is already encrypted
            logger.debug("Adding %s to send_queue", i)
            send_queue.put(i)

    # Put a STOP command at the end of the GPG queue.
    gpg_queue.put('STOP')

    # Start some handlers, wait until everything is done
    try:
        process_count = cpu_count()
    except NotImplementedError:
        process_count = 1

    for i in range(process_count):
        Process(target=encryption_worker, args=(gpg_queue, send_queue,)).start()

    send_p = Process(target=sending_worker, args=(send_queue, secrets.accesskey, secrets.sharedkey, host))
    send_p.start()

    while send_p.is_alive():
        # Generate some output so that BackupPC doesn't time out
        mesg = "Checkpoint: %i task(s) in GPG queue, %i task(s) in S3 queue.\n" % (gpg_queue.qsize(), send_queue.qsize())
        logger.info(mesg)
        sys.stdout.write(mesg + '\n')
        sys.stdout.flush()
        send_p.join(300)

    if not gpg_queue.empty():
        raise Exception("GPG queue not empty")
    if not send_queue.empty():
        raise Exception("Send queue not empty")

    logger.info("Finalizing backup.")

    # finalize the backup
    bucket = open_s3(secrets.accesskey, secrets.sharedkey, host)
    key = Key(bucket)
    key.key = '%sCOMPLETE' % os.path.basename(filehead)
    key.set_contents_from_string('%s %s "%s"' % (beginning, time.time(), mesg))
    key.close()