python将文件备份到数据库（全备、增备）

from Avenger import AvengerAccessor
import os,datetime,subprocess,sys,re
Tmp_dev_50="/home/wqhist/"
Tmp_dev_20="/tmp/DEV20_WqhistLog/wqhist/"
Tmp_dev_60="/tmp/DEV60_WqhistLog/wqhist/"


def scpFileToRemoteNode(user, ip, password, local_dir, remote_dir, port=22):
    os.system("rm -rf local_dir") 
    os.system("mkdir local_dir")  # 继续创建一个用来存放远程主机文件的本地目录
    SCP_CMD_BASE = r"""
        expect -c "
        set timeout 300 ;
        spawn scp -P {port} -r {username}@{host}:{remote_dir} {local_dir};
        expect *assword* {{{{ send {password}\r }}}} ;
        expect *\r ;
        expect \r ;
        expect eof
        "
    """.format(username=user, password=password, host=ip, local_dir=local_dir, remote_dir=remote_dir, port=port)
    SCP_CMD = SCP_CMD_BASE.format(local_dir=local_dir)
    print "execute SCP_CMD: ", SCP_CMD
    p = subprocess.Popen(SCP_CMD, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=True)
    p.communicate()

    os.system(SCP_CMD)


scpFileToRemoteNode("root", "172.16.20.20", "passwd", "/tmp/DEV20_WqhistLog", "/home/wqhist/", 22)
scpFileToRemoteNode("root", "172.16.20.60", "passwd", "/tmp/DEV60_WqhistLog", "/home/wqhist/", 22)
client = AvengerAccessor(None)
client.login("liaogang","passwd")
#登陆Avenger
def walk(file_list,*args):   #遍历出所有日志文件返回列表。
    for walkpath in args:
#        print(walkpath)
        ff = os.walk(walkpath)
        for time_dir, dirs, files, in ff:
            for file in files:
                # print(file)
                file_list.append(os.path.join(time_dir,file))


files = []
walk(files,Tmp_dev_50,Tmp_dev_20,Tmp_dev_60)
#print(files)

def creat_source(aa):#创建Avenger中的source，只需要创建一次，不能重复创建同名的source
#     client.SourceManager.createSource(mySourceName="wqhist", shardKey={}, indexList=[], Some_Description="WQHIST" )
    client.SourceManager.createSource(aa, {}, [], "WQHIST" )
# source=creat_source(‘wqhistory_command‘)
source_name="wqhistory_command"
dataAndStatus = client.SourceManager.listSource()
print dataAndStatus

def full_bak():#全量备份。此处只执行第一次
        full_query = client.queryDataFromSource(dataType="meta",#查询source中是否存在日志文件。
                                       source=source_name,
                                       isPandas=False,
                                       maxRecordNum=None,
                                       volatile=False,
                                       partitionList=None
                                   )
        if len(full_query) == 0:
            for file in files:

                client.insertFile(avengerDataId=file, metaDict={"File_key": "wqhistory"}, metaSource=source_name,
                                  filePath=file, upsert=False, storageType=‘Mongo‘,    #upsert=True会覆盖之前的内容。False是追加
                                  rsaPubPkcs1=None, compressType=None, serialType=None)
                print("Full backup complete!")
full_bak()

def LastWeek_list():#输出前7的所有日期。对应于wqhist下面的日期目录。
    lastweek_list=[]
    for i in range(1,8):
        dt = datetime.date.today() - datetime.timedelta(days=i)
        lastweek_list.append(dt.strftime("%Y-%m-%d"))
    return lastweek_list
sevendays = LastWeek_list()
#print(sevendays)

def Add_bak(*args):#增量备份，备份前7日的日志
    today = datetime.date.today()
    lastweek_date = str(today - datetime.timedelta(days=7))

    for f_path in args:
        lastweek_rootlog=f_path+lastweek_date+‘/‘+‘root_‘+lastweek_date+‘.log‘
        #print(lastweek_rootlog)

        query_lastweek = client.queryDataFromSource(dataType="meta",_id=lastweek_rootlog,#查询source中是否有7天前的数据
                                           source=source_name,
                                           isPandas=False,
                                           maxRecordNum=None,
                                           volatile=False,
                                           partitionList=None
                                           )

        if query_lastweek is  None:
            sevendays_path_list=[]
            for i in sevendays:
                sevendays_path_list.append(f_path+i+‘/‘)
            for sevendays_path in sevendays_path_list:
                sevendays_files = os.listdir(sevendays_path)
                sevenday_file = []
                for file in sevendays_files:
                    sevenday_file.append(sevendays_path+file)
                for file_path in sevenday_file:
                    #print(file_path)
                    client.insertFile(avengerDataId=file_path, metaDict={"File_key":"wqhistory"}, metaSource=source_name,#插入前7天的日志
                                  filePath=file_path, upsert=False, storageType=‘Mongo‘,
                                  rsaPubPkcs1=None, compressType=None, serialType=None)
            print("Incremental backup complete!")
LastWeek_list()
Add_bak(Tmp_dev_50,Tmp_dev_20,Tmp_dev_60)
#需要备份的日志路径作为参数