#!/usr/bin/env python

""" Cleans SSTables on S3 """
import eventlet
eventlet.monkey_patch()

import argparse
import boto
import json
import logging
import os
import socket
import sys
from eventlet.greenpool import GreenPool
from eventlet import pools

from datetime import datetime
from dateutil import parser as dtparser

now = None

def days_ago(tstamp):
    global now
    if now is None:
        now = datetime.now()
    # Datetime format from the last_modified property on keys from boto
    backup = datetime.strptime(tstamp, '%Y-%m-%dT%H:%M:%S.%fZ')
    delta = now - backup
    return delta.days


def get_bucket(args):
    try:
        s3conn = boto.s3.connect_to_region(args.aws_region,
                                           aws_access_key_id=args.key,
                                           aws_secret_access_key=args.secret,
                                           security_token=args.token)
        return s3conn.get_bucket(args.bucket)
    except boto.exception.BotoServerError, e:
        log.error('Problem initializing S3 connection: %s', e)
        sys.exit(1)


def clean_backups(args, log):

    if args.debug:
        log.setLevel(logging.DEBUG)

    if not args.name:
        args.name = socket.getfqdn()

    bucket = get_bucket(args)

    try:
        key_list = bucket.list("%s:%s" % (args.name, args.path))
    except boto.exception.BotoServerError, e:
        log.error('Problem getting keys from S3 bucket: %s', e)
        sys.exit(1)

    log.info("Connected to S3, getting keys ...")

    index_keys = []
    data_keys = []
    files_to_delete = set()

    for key in key_list:
        files_to_delete.add(key.name)
        if key.name.endswith('-listdir.json'):
            index_keys.append(key)
        else:
            data_keys.append(key)

    index_keys.sort(key=lambda k: dtparser.parse(k.last_modified),
                    reverse=True)

    log.info("Found %s index files and %s data files", len(index_keys), len(data_keys))

    index_files_to_keep = set()
    index_files_to_keep.add(index_keys[0].name)  # Always keep the most recent index.
    for index_key in index_keys:
        if days_ago(index_key.last_modified) > args.age:
            break
        index_files_to_keep.add(index_key.name)

    log.info("Keeping %s/%s index files", len(index_files_to_keep), len(index_keys))

    files_to_keep = get_files_from_indexes(args, index_files_to_keep)
    log.info("Keeping %s/%s data files", len(files_to_keep), len(data_keys))

    files_to_delete -= files_to_keep
    files_to_delete -= index_files_to_keep

    if args.list_deletes:
        for filename in sorted(files_to_delete):
            log.info('Deleting %s', filename)

    if args.debug:
        log.debug("Would delete %s/%s files", len(files_to_delete), len(index_keys) + len(data_keys))
        log.debug("Test mode, nothing deleted")
        return

    log.info("Deleting %s/%s files", len(files_to_delete), len(index_keys) + len(data_keys))
    try:
        bucket.delete_keys(files_to_delete)
    except Exception as e:
        log.error('S3 delete ERR, will try again later [%s]', e)


def get_files_from_indexes(args, index_files):
    pool = GreenPool(int(args.parallel))
    bucket_pool = pools.Pool(create=lambda: get_bucket(args), max_size=int(args.parallel))

    def worker(index_file):
        with bucket_pool.item() as bucket:
            return get_file_list(args, bucket, index_file)

    res = set()
    for result in pool.imap(worker, index_files):
        res.update(result)
    return res


def get_file_list(args, bucket, index_file):
    index = json.loads(bucket.get_key(index_file).get_contents_as_string())
    if len(index.values()) != 1:
        raise SystemError('-listdir.json file should have a single key/value pair!')
    dirname = index.keys()[0]
    return ["%s:%s" % (args.name, os.path.join(dirname, filename)) for filename in index.values()[0]]


def main(log):
    parser = argparse.ArgumentParser(
        description='Clean SSTables from S3. Scroll backwards through '
        '-listdir.json keys in chronological order collecting a "keeper" '
        'list until it reaches it\'s age cutoff. Deletes all keys not in that '
        'list')
    parser.add_argument(
        '-d',
        '--debug',
        dest='debug',
        action='store_true',
        help='Run in debug mode, will not delete keys. Implies -v')
    parser.add_argument(
        '-l',
        '--list-deletes',
        dest='list_deletes',
        action='store_true',
        help='Log every file being deleted')
    parser.add_argument(
        '-j',
        '--parallel',
        dest='parallel',
        default=5,
        help='Parallelism to fetch index files from S3')
    parser.add_argument(
        '-k',
        '--key',
        dest='key',
        default=os.environ.get('AWS_ACCESS_KEY_ID'),
        help='Amazon S3 Key (default from AWS_ACCESS_KEY_ID in environment)')
    parser.add_argument(
        '-s',
        '--secret',
        dest='secret',
        default=os.environ.get('AWS_SECRET_ACCESS_KEY'),
        help='Amazon S3 Secret (default from AWS_SECRET_ACCESS_KEY in environment)')
    parser.add_argument('--token',
        default=os.environ.get('AWS_SECURITY_TOKEN'),
        help='Amazon S3 Token (default from AWS_SECURITY_TOKEN in environment)')
    parser.add_argument('--aws-region',
        default='us-east-1',
        choices=[region.name for region in boto.s3.regions()],
        help='AWS region to connect to.')
    parser.add_argument(
        '-n',
        '--name',
        dest='name',
        required=False,
        help='Use this name instead of the FQDN to identify the files from '
             'this host')
    parser.add_argument(
        'bucket',
        help='S3 Bucket')
    parser.add_argument(
        'path',
        help='Path portion of key in S3')
    parser.add_argument(
        'age',
        type=int,
        help='How many days worth of backups to keep')
    args = parser.parse_args()
    clean_backups(args, log)

if __name__ == '__main__':

    log = logging.getLogger('tablechop')
    stderr = logging.StreamHandler()
    stderr.setFormatter(logging.Formatter(
        '%(name)s [%(asctime)s] %(levelname)s %(message)s'))
    log.addHandler(stderr)
    if os.environ.get('TDEBUG', False):
        log.setLevel(logging.DEBUG)
    else:
        log.setLevel(logging.INFO)

    main(log)
