class ExceedDateKeyCleanTask(object):
'''
删除超过N天数据的清理任务
'''
def __init__(self):
self.__config = GetConfig()
self.__log = Logger('exceedDateKeyCleanTask')
self.__redisClient = RedisClient(self.__config.redis_host(), self.__config.redis_port(), self.__config.redis_password())
self.__taskName = 'clean exceed date redis key'
self.__className = self.__class__
def __doStart(self, keyPrefix, db, date):
startTime = time.time()
taskName = self.__taskName + " by doStart"
self.__log.info("start doing task:{task}, className:{className}".format(task=taskName, className=self.__class__))
key = keyPrefix + date;
self.__redisClient.unlink(db, key);
self.__log.info("unlink key:{key}".format(key=key))
costTime = time.time() - startTime
self.__log.info("task:{task} finish, className:{className}, key:{key}, db:{db}, cost:{cost}" \
.format(task=taskName, className=self.__className, key=key, db=db, cost=costTime))
def start(self, keyPrefix, db, validDays):
"""
删除指定前缀的key
:param keyPrefix 待删除的key前缀 key的格式为 keyPrefix_yyyyMMdd,其中前缀为**_
:param db key所在的数据库
:param validDays key在几天内有效
:return:
"""
day = DateUtil.format4Y2m2d(DateUtil.addDaysOnCurrent(-validDays))
self.__doStart(keyPrefix, db, day)
pass
def startHour(self, keyPrefix, db, validDays):
"""
清除时间格式为小时的key，即key的格式为keyPrefix_yyyyMMddHH
:param keyPrefix 待删除的key前缀 key的格式为 keyPrefix_yyyyMMdd,其中前缀为**_
:param db key所在的数据库
:param validDays key在几天内有效
:return:
"""
day = DateUtil.format4Y2m2d(DateUtil.addDaysOnCurrent(-validDays))
for i in xrange(0, 24):
if i < 10:
self.__doStart(keyPrefix, db, day + '0' + str(i))
else:
self.__doStart(keyPrefix, db, day + str(i))
def startBatch(self, keyPrefix, db, validDays):
"""
删除指定前缀的set中保存的所有value对应的key（set中保存的就是要删除的key）
:param keyPrefix 待删除的key前缀 key的格式为 keyPrefix_yyyyMMdd,其中前缀为**_
:param db key所在的数据库
:param validDays key在几天内有效
:return:
"""
startTime = time.time()
taskName = self.__taskName + " by startBatch"
self.__log.info("start doing task:{task}, className:{className}".format(task=taskName, className=self.__className))
day = DateUtil.format4Y2m2d(DateUtil.addDaysOnCurrent(-validDays))
setKey = keyPrefix + day
cursor = 0
# 每次删除1000条记录
while True:
scanResult = self.__redisClient.sscan(db, setKey, cursor, None, 1000)
# 如果获取不到记录，则跳出循环
if len(scanResult) == 0:
break
cursor = scanResult[0]
# 删除记录
if len(scanResult[1]) > 0:
self.__redisClient.unlink(db, *scanResult[1])
self.__redisClient.srem(db, setKey, *scanResult[1])
# 如果已经检索完，则跳出循环
if cursor == 0:
break
costTime = time.time() - startTime
self.__log.info("task:{task} finish, className:{className}, key:{key}, db:{db}, cost:{cost}" \
.format(task=taskName, className=self.__className, key=setKey, db=db, cost=costTime))
def startHscan(self, keyPrefix, db, validDays):
"""
使用hscan删除大数据量的hash key
:param keyPrefix 待删除的key前缀 key的格式为 keyPrefix_yyyyMMdd,其中前缀为**_
:param db key所在的数据库
:param validDays key在几天内有效
:return:
"""
startTime = time.time()
taskName = self.__taskName + " by startHscan"
self.__log.info("start doing task:{task}, className:{className}".format(task=taskName, className=self.__className))
day = DateUtil.format4Y2m2d(DateUtil.addDaysOnCurrent(-validDays))
key = keyPrefix + day
cursor = 0
# 每次删除1000条记录
while True:
scanResult = self.__redisClient.hscan(db, key, cursor, None, 1000)
# 如果获取不到记录，则跳出循环
if len(scanResult) == 0:
break
cursor = scanResult[0]
# 获取map中的所有key
keyList = []
if len(scanResult[1]) > 0:
for (k, v) in scanResult[1].items():
keyList.append(k)
# 批量删除记录
self.__redisClient.hdel(db, key, *keyList)
# 如果已经检索完，则跳出循环
if cursor == 0:
break
costTime = time.time() - startTime
self.__log.info("task:{task} finish, className:{className}, key:{key}, db:{db}, cost:{cost}" \
.format(task=taskName, className=self.__className, key=key, db=db, cost=costTime))
def startSscan(self, keyPrefix, db, validDays):
"""
使用sscan删除大数据量的set key
:param keyPrefix 待删除的key前缀 key的格式为 keyPrefix_yyyyMMdd,其中前缀为**_
:param db key所在的数据库
:param validDays key在几天内有效
:return:
"""
startTime = time.time()
taskName = self.__taskName + " by startSscan"
self.__log.info("start doing task:{task}, className:{className}".format(task=taskName, className=self.__className))
day = DateUtil.format4Y2m2d(DateUtil.addDaysOnCurrent(-validDays))
key = keyPrefix + day
cursor = 0
# 每次删除1000条记录
while True:
scanResult = self.__redisClient.sscan(db, key, cursor, None, 1000)
# 如果获取不到记录，则跳出循环
if len(scanResult) == 0:
break
cursor = scanResult[0]
# 删除记录
if len(scanResult[1]) > 0:
self.__redisClient.srem(db, key, *scanResult[1])
# 如果已经检索完，则跳出循环
if cursor == 0:
break
costTime = time.time() - startTime
self.__log.info("task:{task} finish, className:{className}, key:{key}, db:{db}, cost:{cost}" \
.format(task=taskName, className=self.__className, key=key, db=db, cost=costTime))
class ExceedDateKeyCleanTaskTest(object):
'''
KeyCleanTask 测试类
'''
def __init__(self):
self.__exceedDateKeyCleanTask = ExceedDateKeyCleanTask()
self.__config = GetConfig()
self.__redisClient = RedisClient(self.__config.redis_host(),
self.__config.redis_port(),
self.__config.redis_password())
def testStart(self):
keyPrefix = 'account_cost_'
validDays = 2
db = 1
day = DateUtil.format4Y2m2d(DateUtil.addDaysOnCurrent(-validDays))
key = keyPrefix + day
print "----test start func begin"
self.__redisClient.hset(db, key, 'map1', 'value1')
self.__redisClient.hset(db, key, 'map2', 'value2')
print("init data:")
print(self.__redisClient.hscan(db, key, 0, None, 3))
self.__exceedDateKeyCleanTask.start(keyPrefix, db, validDays)
print("after clean:")
print(self.__redisClient.hscan(db, key, 0, None, 3))
pass
def testStartHour(self):
keyPrefix = 'data_cost_'
validDays = 2
db = 3
day = DateUtil.format4Y2m2d(DateUtil.addDaysOnCurrent(-validDays))
print "----test startHour func begin"
keyList = []
for i in xrange(0, 24):
if i < 10:
key = keyPrefix + day + '0' + str(i)
else:
key = keyPrefix + day + str(i)
keyList.append(key)
self.__redisClient.hset(db, key, 'map1', 'value1')
self.__redisClient.hset(db, key, 'map2', 'value2')
print("init data:")
for key in keyList:
print(self.__redisClient.hgetall(db, key))
self.__exceedDateKeyCleanTask.startHour(keyPrefix, db, validDays)
print("after clean:")
for key in keyList:
print(self.__redisClient.hgetall(db, key))
pass
def testStartBatch(self):
keyPrefix = 'mix_detail_posid_'
validDays = 30
db = 1
day = DateUtil.format4Y2m2d(DateUtil.addDaysOnCurrent(-validDays))
key = keyPrefix + day
print "----test startBatch func begin"
self.__redisClient.sadd(db, key, *['mem1', 'mem2'])
print("init data:")
print(self.__redisClient.sscan(db, key, 0, None, 3))
self.__exceedDateKeyCleanTask.startBatch(keyPrefix, db, validDays)
print("after clean:")
print(self.__redisClient.sscan(db, key, 0, None, 3))
pass
def testStartHscan(self):
keyPrefix = 'view_st_ts_'
validDays = 1
db = 1
day = DateUtil.format4Y2m2d(DateUtil.addDaysOnCurrent(-validDays))
key = keyPrefix + day
print "----test startHscan func begin"
self.__redisClient.hset(db, key, 'map1', 'value1')
self.__redisClient.hset(db, key, 'map2', 'value2')
print("init data:")
print(self.__redisClient.hscan(db, key, 0, None, 3))
self.__exceedDateKeyCleanTask.startHscan(keyPrefix, db, validDays)
print("after clean:")
print(self.__redisClient.hscan(db, key, 0, None, 3))
pass
def testStartSscan(self):
keyPrefix = 'click_st_'
validDays = 1
db = 1
day = DateUtil.format4Y2m2d(DateUtil.addDaysOnCurrent(-validDays))
key = keyPrefix + day
print "----test startSscan func begin"
self.__redisClient.sadd(db, key, *['mem1', 'mem2'])
print("init data:")
print(self.__redisClient.sscan(db, key, 0, None, 3))
self.__exceedDateKeyCleanTask.startSscan(keyPrefix, db, validDays)
print("after clean:")
print(self.__redisClient.sscan(db, key, 0, None, 3))
pass