mongo不同库之间多表导入导出

需求:

源库地址: 192.168.10.10

目标地址: 192.168.10.20

需要将源库的多张表复制到目标库

python脚本

#!/usr/bin/python
# -*- coding: UTF-8 -*-
import time
from pymongo import MongoReplicaSetClient
from pymongo import MongoClient
import logging
from  remotessh import sshmongo

# 日志存储目录
logging.basicConfig(filename='D:/info.txt', level=logging.INFO)

# 获取当前时间
NowTime = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(time.time()))
# 数据源
DataSource = MongoClient('mongodb://192.168.10.10:27017')
dbSource = DataSource.company
# 密码认证，没有可不写
dbSource.authenticate('test', 'test')
# 目标库
DataTarget = MongoClient('mongodb://192.168.10.20:27017')
dbTarget = DataTarget.company

# 查看所有库名
# print c.list_database_names()
# 需要导入的三张表
ImportTables = ['cm', 'ccc', 'hehe']
# 查看所有表
# for i in dbTarget.collection_names():
#     print i, '总条数:', dbTarget[i].find().count()

def MainFunc():
    for i in ImportTables:
        if i in dbSource.collection_names():
            SourceNum = dbSource[i].find().count()
            if i in dbTarget.collection_names():
                TargetNum = dbTarget[i].find().count()
                print i, '源库总条数:', SourceNum, i, '目标总条数:', TargetNum
                # 删除目标表中的数据
                dbTarget[i].drop()
                # 插入新的数据
                res = sshmongo.RemoteSer(i)
                if res == 0001:
                    logging.info(NowTime + "-------->" + "无法通过ssh连接远程服务器")
                elif res == 0002:
                    logging.info(NowTime + "-------->" + "连接mongo失败或者插入数据错误!")
                else:
                    pass
            else:
                logging.info(
                    NowTime + "-------->" + "%s 表在源库共有 %d 条数据，目标中没有，直接导入数据" %
                    (i, SourceNum))
                res = sshmongo.RemoteSer(i)
                if res == 0001:
                    logging.info(NowTime + "-------->" + "无法通过ssh连接远程服务器")
                elif res == 0002:
                    logging.info(NowTime + "-------->" + "连接mongo失败或者插入数据错误!")
                else:
                    pass
        else:
            # print "%s 表在源库中不存在，无法导入目标库" % i
            logging.info(NowTime + "-------->" + "%s 表在源库中不存在，无法导入目标库" % i)
    DataSource.close()
    DataTarget.close()

MainFunc()

这个脚本调用了另外一个远程连接服务器执行shell命令的python脚本

#!/usr/bin/python
# -*- coding: UTF-8 -*-
import paramiko
import os
import re

hostname = '192.168.10.20'
username = 'root'
password = '11111111111'
port = int(20001)


def RemoteSer(tables):
    # noinspection PyBroadException
    try:
        ssh = paramiko.SSHClient()
        key = paramiko.AutoAddPolicy()
        ssh.set_missing_host_key_policy(key)
        ssh.connect(hostname, port, username, password, timeout=5)
        # 执行多条shell命令
        # stdin, stdout, stderr = ssh.exec_command('cd /data/; ls -l; touch test.txt',get_pty=True)
        # stdin, stdout, stderr = ssh.exec_command('rm -rf %s/*' % backdir, get_pty=True)
        stdin, stdout, stderr = ssh.exec_command('mongoexport -h 192.168.10.10 -u test -ptest -d company --port 27017 -c'
                                                 ' %s  | mongoimport -h 192.168.10.20 -d company -c %s --port 27017'
                                                 % (tables, tables), get_pty=True)

        for i in stdout.readlines():
            ErrorStr = re.compile(r'error', re.IGNORECASE).findall(i)
            if ErrorStr:
                pass
                return 0002

        ssh.close()

    except Exception:
        # print "连接服务器失败!"
        return 0001

把你需要的表写入到

ImportTables

并且修改成你的地址就OK了!