manyi/script/db/remove_old_data_by_date.py

# -*- coding: utf-8 -*-
# !/usr/bin/env python

"""
删除已经迁移到历史数据库的数据.date作为过滤条件
"""

import getopt
import os
import sys

try:
    options, args = getopt.getopt(sys.argv[1:], 'l:e:y:m:o:c:',
                                  ['log=', 'env=', 'year=', 'month=', 'model=', 'check='])
except getopt.GetoptError as e:
    print(str(e))
    sys.exit()

log_file = None
platform_env = 'testing'
year = None
month = None
model_name = None
check = True

for name, value in options:
    if name in ('-l', '--log'):
        log_file = value

    if name in ('-e', '--env'):
        platform_env = value

    if name in ('-y', '--year'):
        year = int(value)

    if name in ('-m', '--month'):
        month = int(value)

    if name in ('-o', '--model'):
        model_name = value

    if name in ('-c', '--check'):
        check = True if value in ['y', 'Y'] else False

os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'configs.{env}'.format(env = platform_env))

PROJECT_ROOT = os.path.join(os.path.abspath(os.path.split(os.path.realpath(__file__))[0] + "/.."), '..')
sys.path.insert(0, PROJECT_ROOT)

from script.base import init_env, setup_logger, get_logger

init_env(interactive = False)

if log_file:
    logger = setup_logger(filename = log_file, namespace = __name__)
else:
    logger = get_logger(__name__)

from apps.web.models import ArchivedModelProxyConfig
from apilib.utils_mongo import BulkHandlerEx

item = ArchivedModelProxyConfig.objects(model = model_name).first()  # type: ArchivedModelProxyConfig
if not item:
    item = ArchivedModelProxyConfig.objects(model = 'default').first()  # type: ArchivedModelProxyConfig

his_data_line = '-'.join(item.startDay.split('-')[0:2])
print his_data_line

if not year or not month or not model_name:
    logger.error('year or month or model is not valid.')
    sys.exit(1)

start_day = '%04d-%02d-%02d' % (year, month, 1)

if month < 12:
    next_year = year
    next_month = month + 1
else:
    next_year = (year + 1)
    next_month = 1

end_day = '%04d-%02d-%02d' % (next_year, next_month, 1)

print start_day
print end_day

if check and '%04d-%02d' % (year, month) >= his_data_line:
    logger.error('{} >= {}'.format('%04d-%02d' % (year, month), his_data_line))
    sys.exit(1)

delete_id_list = []

from script.db import MODEL_MAP

model_class = MODEL_MAP.get(model_name)

items = model_class.get_collection().find({
    'date': {'$gte': start_day, '$lt': end_day}
}, {'_id': 1}).batch_size(50000).limit(1000000)

for item in items:
    print item['_id']
    delete_id_list.append(item['_id'])

logger.debug('try to delete {} items.'.format(len(delete_id_list)))

bulker = BulkHandlerEx(model_class.get_collection())  # type: BulkHandlerEx

for _id in delete_id_list:
    bulker.delete(query_dict = {'_id': _id})

    if len(bulker.requests) >= 2000:
        bulker.execute()
        bulker = BulkHandlerEx(model_class.get_collection())  # type: BulkHandlerEx

if len(bulker.requests) > 0:
    bulker.execute()
    bulker = None