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 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 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 testDoTask(self):
dayList = DateUtil.splitIntoDays(DateUtil.addDaysOnCurrent(-3),
DateUtil.getCurrentDate())
hourList = DateUtil.splitIntoHours(DateUtil.addHoursOnCurrent(-6),
DateUtil.getCurrentDate())
db = 0
print "----test doTask func begin"
unlinkKeyList = []
unlinkSetKeyList = []
# 删除以day为单位的key
for day in dayList:
unlinkSetKeyList.append('launch_date_' +
day) # 投放日期(Set),存储campaign Id 的set(白名单)
unlinkKeyList.append('launch_date_2_creatives_' +
day) # 投放日期到creative的映射关系,存储的是BitSet
unlinkKeyList.append('fingerprint_2_creativeId_' +
day) # fingerprint与creativeId对应关系
self.__redisClient.sadd(db, 'launch_date_' + day,
*['mem1', 'mem2'])
self.__redisClient.set(db, 'launch_date_2_creatives_' + day,
'launch_date_2_creatives_value1')
self.__redisClient.set(db, 'fingerprint_2_creativeId_' + day,
'fingerprint_2_creativeId_value1')
# 删除以hour为单位的key
for hour in hourList:
unlinkKeyList.append(
'account_unfreeze_time_' +
hour) # 广告商账户级别的冻结(预算不足),该小时冻结,存储的是agentId set
unlinkKeyList.append(
'campaign_unfreeze_time_' +
hour) # 推广计划级别的冻结(预算不足),该小时冻结,存储的是campaign Id set(黑名单)
unlinkKeyList.append('account_unfreeze_time_2_creatives_' +
hour) # 账户级别的冻结时间到creative的映射关系,存储的是BitSet
unlinkKeyList.append('campaign_unfreeze_time_2_creatives_' +
hour) # 推广计划级别的冻结时间到creative的映射关系,存储的是BitSet
self.__redisClient.set(db, 'account_unfreeze_time_' + hour,
'account_unfreeze_time_')
self.__redisClient.set(db, 'campaign_unfreeze_time_' + hour,
'campaign_unfreeze_time_')
self.__redisClient.set(db,
'account_unfreeze_time_2_creatives_' + hour,
'account_unfreeze_time_2_creatives_')
self.__redisClient.set(
db, 'campaign_unfreeze_time_2_creatives_' + hour,
'campaign_unfreeze_time_2_creatives_')
print("init data:")
print(self.__redisClient.mget(db, *unlinkKeyList))
for unlinkSetKey in unlinkSetKeyList:
print(self.__redisClient.sscan(db, unlinkSetKey, 0, None, 3))
self.__keyCleanTask.doTask()
print("after clean:")
print(self.__redisClient.mget(db, *unlinkKeyList))
for unlinkSetKey in unlinkSetKeyList:
print(self.__redisClient.sscan(db, unlinkSetKey, 0, None, 3))
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 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 doTask(self):
'''
清理指定的key
:return:
'''
dayList = DateUtil.splitIntoDays(DateUtil.addDaysOnCurrent(-3),
DateUtil.getCurrentDate())
hourList = DateUtil.splitIntoHours(DateUtil.addHoursOnCurrent(-6),
DateUtil.getCurrentDate())
unlinkKeyList = []
# 删除以day为单位的key
for day in dayList:
unlinkKeyList.append('launch_date_' +
day) # 投放日期(Set),存储campaign Id 的set(白名单)
unlinkKeyList.append('launch_date_2_creatives_' +
day) # 投放日期到creative的映射关系,存储的是BitSet
unlinkKeyList.append('fingerprint_2_creativeId_' +
day) # fingerprint与creativeId对应关系
# 删除以hour为单位的key
for hour in hourList:
unlinkKeyList.append(
'account_unfreeze_time_' +
hour) # 广告商账户级别的冻结(预算不足),该小时冻结,存储的是agentId set
unlinkKeyList.append(
'campaign_unfreeze_time_' +
hour) # 推广计划级别的冻结(预算不足),该小时冻结,存储的是campaign Id set(黑名单)
unlinkKeyList.append('account_unfreeze_time_2_creatives_' +
hour) # 账户级别的冻结时间到creative的映射关系,存储的是BitSet
unlinkKeyList.append('campaign_unfreeze_time_2_creatives_' +
hour) # 推广计划级别的冻结时间到creative的映射关系,存储的是BitSet
self.__redisClient.unlink(0, *unlinkKeyList)
self.__log.info(
"unlink keys:{keys}".format(keys=unlinkKeyList.__str__()))