Dump MongoDB databases to the zip archives and copy its to the output folder.

mongozip.py

# -*- coding: utf-8 -*-
'''Dump MongoDB databases to the zip archives 
and copy its to the output folder.
Usage: python mongozip.py
'''
import os
import os.path
import datetime
import tempfile
import shutil
import subprocess
import sys
from zipfile import ZipFile, ZIP_DEFLATED

from pymongo import MongoClient
from pymongo.read_preferences import ReadPreference


MONGO_HOST = 'm2.example.com'
MONGO_PORT = 27017
MONGODUMP = 'C:\\mongodb\\bin\\mongodump.exe' # Windows
OUTPUT_DIR = os.path.abspath('./some/secure/published/folder')
MONGO_TARGETS = [
    {
        'database': 'database1', 
        'ignore': ['fs.chunks'], 
        'login':  'database1',
        'password': 'Jfhgd76eHHs'
    },
    {
        'database': 'database2',
        'ignore': [],
        'login':  'database2',
        'password': 'Jdhg613TFg'
    }
]


def get_target_collections(target):
    '''Get list of collections except system collections
    and collections which configured as ignored.
    '''
    mongo = MongoClient(MONGO_HOST, MONGO_PORT)
    dbname = target['database']
    ignores = target['ignore']
    login = target['login']
    password = target['password']
    database = mongo[dbname]
    database.authenticate(login, password)
    database.read_preference = ReadPreference.SECONDARY
    ignore = lambda col: col.startswith('system.') or col in ignores
    return [col for col in database.collection_names() if not ignore(col)]


def dump_collection(target, collection, temp_dir):
    '''Dump target collection to the temporary folder.'''
    dbname = target['database']
    login = target['login']
    password = target['password']
    args = [
        MONGODUMP,
        '-h', MONGO_HOST,
        '--port', str(MONGO_PORT),
        '-u', login,
        '-p', password,
        '-d', dbname,
        '-c', collection,
        '-o', temp_dir
    ]
    subprocess.call(args)

def zip_db_dump(dbname, temp_dir):
    '''Zip database dump folder.'''
    source_zip = os.path.join(temp_dir, dbname)
    date_stamp = datetime.date.today().isoformat()
    zip_name = '{dbname}.{date}.zip'.format(dbname=dbname, date=date_stamp)
    target_zip = os.path.join(temp_dir, zip_name)
    with ZipFile(target_zip, 'w', ZIP_DEFLATED) as myzip:
        for root, _, files in os.walk(source_zip):
            for fname in files:
                absfn = os.path.join(root, fname)
                zfn = absfn[len(temp_dir)+len(os.sep):]
                myzip.write(absfn, zfn)
    return target_zip


def write_to_output(dbname, abszipfn):
    '''Copy archive to the output and rewrite latest database archive.'''
    if not os.path.exists(OUTPUT_DIR):
        os.makedirs(OUTPUT_DIR)
    _, zipfn = os.path.split(abszipfn)
    move_to = os.path.join(OUTPUT_DIR, zipfn)
    shutil.move(abszipfn, move_to)
    latest = '{dbname}.latest.zip'.format(dbname=dbname)
    latest = os.path.join(OUTPUT_DIR, latest)
    shutil.copy2(move_to, latest)


def main():
    '''The entry point of the script.'''
    temp_dir = tempfile.mkdtemp()
    try:
        for target in MONGO_TARGETS:
            cols = get_target_collections(target)
            for col in cols:
                dump_collection(target, col, temp_dir)
            dbname = target['database']
            abszipfn = zip_db_dump(dbname, temp_dir)
            write_to_output(dbname, abszipfn)
    except StandardError as err:
        print >> sys.stderr, str(err)
        return 1
    finally:
        shutil.rmtree(temp_dir)


if __name__ == '__main__':
    sys.exit(main())