def getGCLogFilesFromData(self,dataDom,cubeName):
gcLogFilesDom = dataDom.getElementsByTagName( 'gcLogFile' )
if len(gcLogFilesDom)==0:
logger.warning('%s: 未配置gcLogFile, 不对服务的gc日志做收集' % cubeName)
return []
gcLogFiles=[]
for gcLogFileDom in gcLogFilesDom:
file = self.getText( gcLogFileDom, 'fileName', False, cubeName )
# 读取wholeFile配置, 若未配置则为False
wholeFile = self.getText( gcLogFileDom, 'wholeFile', False, cubeName )
if wholeFile.lower()=='true':
wholeFile=True
else:
wholeFile=False
# 读取监控的时间间隔, 若未配置则为DEFAULT_GCLOGFILE_TIME_INTERVAL
timeInterval = self.getText( gcLogFileDom, 'timeInterval', False, cubeName )
if timeInterval!='':
self.assertDigit(timeInterval, cubeName, 'timeInterval')
timeInterval = int( timeInterval )
else:
timeInterval=self.DEFAULT_GCLOGFILE_TIME_INTERVAL
# 读取alarm keys
if 'on' in self.getText(gcLogFileDom,'alarm',False,cubeName):
# 可能会有多个gc files, 因此需要deepcopy
keyAlarmers=self.getGCKeyAlarmers()
else:
keyAlarmers={}
statisticTool=GCLogTool(logger,self.gcKeys)
gcLogFile=LogFile(self.ldap,self.product,self.type,file,timeInterval,self.gcKeys,[],statisticTool,wholeFile,keyAlarmers,self.debug,logger,cubeName)
if gcLogFile.validate():
gcLogFiles.append(gcLogFile)
return gcLogFiles
def UpdateExtensionInfo(mssqlDict):
mssqlHandle = MssqlAPI(server = mssqlDict['SERVER'], db = mssqlDict['DB'], user = mssqlDict['USER'], pwd = mssqlDict['PWD'])
logFile = LogFile(name = 'Indirection')
try:
mssqlHandle.sqlExecuteProc('usp_UpdateFrom251', ())
mssqlHandle.sqlCommit()
logFile.logInfo('Execute SQL Proc "usp_UpdateFrom251" succeed.')
except:
logFile.logInfo('Execute SQL Proc "usp_UpdateFrom251" failed.')
print traceback.format_exc()
def getLogFilesFromData(self, dataDom, cubeName):
logFilesDom = dataDom.getElementsByTagName('logFile')
if len(logFilesDom) == 0:
logger.warning('%s: 未配置logFile, 不对服务日志做收集' % cubeName)
return []
logFiles = []
for logFileDom in logFilesDom:
file = self.getText(logFileDom, 'fileName', False, cubeName)
# 读取wholeFile配置, 若未配置则为False
wholeFile = self.getText(logFileDom, 'wholeFile', False, cubeName)
if wholeFile.lower() == 'true':
wholeFile = True
else:
wholeFile = False
# 读取监控的时间间隔, 若未配置则为DEFAULT_LOGFILE_TIME_INTERVAL
timeInterval = self.getText(logFileDom, 'timeInterval', False,
cubeName)
if timeInterval != '':
self.assertDigit(timeInterval, cubeName, 'timeInterval')
timeInterval = int(timeInterval)
else:
timeInterval = self.DEFAULT_LOGFILE_TIME_INTERVAL
# 读取key
numberKeys = self.getText(logFileDom, 'numberKeys', False,
cubeName).split(',')
stringKeys = self.getText(logFileDom, 'stringKeys', False,
cubeName).split(',')
# 去除各字段首尾空格，以及空字段
numberKeys = [
str(item).strip() for item in numberKeys if str(item).strip()
]
stringKeys = [
str(item).strip() for item in stringKeys if str(item).strip()
]
# 读取beforeDays配置, 默认为0
beforeDays = 0
# 读取noTimeStamp配置, 默认为False
noTimeStamp = False
# 读取alarm keys
keyAlarmers = self.parseAlarmKeys(
self.getText(logFileDom, 'alarmKeys', False, cubeName))
statisticTool = KeyValueLogTool(numberKeys + stringKeys,
timeInterval, beforeDays,
noTimeStamp, logger)
logFile = LogFile(self.ldap, self.product, self.type, file,
timeInterval, numberKeys, stringKeys,
statisticTool, wholeFile, keyAlarmers,
self.debug, logger, cubeName)
if logFile.validate():
logFiles.append(logFile)
return logFiles
def main(argv):
logsList = [LogFile(name) for name in fileNamesList]
print(logsList)
gen = SpamGenerator()
while not isAllComplete(logsList):
for log in logsList:
log.addContent(gen.generateMessage(random.randrange(7)))
def dropOriginalData(mongoDict):
logFile = LogFile(name = 'Indirection')
for mongo in mongoDict:
try:
mongoHandle = MongoAPI(server = mongo['SERVER'], port = mongo['PORT'])
#drop table
splitTableList = ['KDayData', 'KWeeklyData', 'KMonthData']
mergeTableList = ['CandleDay', 'CandleWeek', 'CandleMonth']
for splitTable in splitTableList:
mongoHandle.drop(splitTable)
for mergeTable in mergeTableList:
mongoHandle.drop(mergeTable)
#create index
for mergeTable in mergeTableList:
mongoHandle.createIndex(mergeTable, 'data', [('_id.IC', pymongo.ASCENDING)])
mongoHandle.createIndex(mergeTable, 'data', [('TO', pymongo.DESCENDING)])
logFile.logInfo('Drop original data from server: ' + mongo['SERVER'] + ' succeed.')
del mongoHandle
except:
logFile.logInfo('Drop original data from server: ' + mongo['SERVER'] + ' failed.')
print traceback.format_exc()
def getLogFilesFromData(self,dataDom,cubeName):
logFilesDom = dataDom.getElementsByTagName( 'logFile' )
if len(logFilesDom)==0:
logger.warning('%s: 未配置logFile, 不对服务日志做收集' % cubeName)
return []
logFiles=[]
for logFileDom in logFilesDom:
file = self.getText( logFileDom, 'fileName', False, cubeName )
# 读取wholeFile配置, 若未配置则为False
wholeFile = self.getText( logFileDom, 'wholeFile', False, cubeName )
if wholeFile.lower()=='true':
wholeFile=True
else:
wholeFile=False
# 读取监控的时间间隔, 若未配置则为DEFAULT_LOGFILE_TIME_INTERVAL
timeInterval = self.getText( logFileDom, 'timeInterval', False, cubeName )
if timeInterval!='':
self.assertDigit(timeInterval, cubeName, 'timeInterval')
timeInterval = int( timeInterval )
else:
timeInterval=self.DEFAULT_LOGFILE_TIME_INTERVAL
# 读取key
numberKeys = self.getText( logFileDom, 'numberKeys', False, cubeName ).split( ',' )
stringKeys = self.getText( logFileDom, 'stringKeys', False, cubeName ).split( ',' )
# 去除各字段首尾空格，以及空字段
numberKeys = [ str( item ).strip() for item in numberKeys if str( item ).strip() ]
stringKeys = [ str( item ).strip() for item in stringKeys if str( item ).strip() ]
# 读取beforeDays配置, 默认为0
beforeDays=0
# 读取noTimeStamp配置, 默认为False
noTimeStamp=False
# 读取alarm keys
keyAlarmers=self.parseAlarmKeys(self.getText(logFileDom,'alarmKeys',False,cubeName))
statisticTool=KeyValueLogTool(numberKeys+stringKeys,timeInterval,beforeDays,noTimeStamp,logger)
logFile=LogFile(self.ldap,self.product,self.type,file,timeInterval,numberKeys,stringKeys,statisticTool,wholeFile,keyAlarmers,self.debug,logger,cubeName)
if logFile.validate():
logFiles.append(logFile)
return logFiles
def getGCLogFilesFromData(self, dataDom, cubeName):
gcLogFilesDom = dataDom.getElementsByTagName('gcLogFile')
if len(gcLogFilesDom) == 0:
logger.warning('%s: 未配置gcLogFile, 不对服务的gc日志做收集' % cubeName)
return []
gcLogFiles = []
for gcLogFileDom in gcLogFilesDom:
file = self.getText(gcLogFileDom, 'fileName', False, cubeName)
# 读取wholeFile配置, 若未配置则为False
wholeFile = self.getText(gcLogFileDom, 'wholeFile', False,
cubeName)
if wholeFile.lower() == 'true':
wholeFile = True
else:
wholeFile = False
# 读取监控的时间间隔, 若未配置则为DEFAULT_GCLOGFILE_TIME_INTERVAL
timeInterval = self.getText(gcLogFileDom, 'timeInterval', False,
cubeName)
if timeInterval != '':
self.assertDigit(timeInterval, cubeName, 'timeInterval')
timeInterval = int(timeInterval)
else:
timeInterval = self.DEFAULT_GCLOGFILE_TIME_INTERVAL
# 读取alarm keys
if 'on' in self.getText(gcLogFileDom, 'alarm', False, cubeName):
# 可能会有多个gc files, 因此需要deepcopy
keyAlarmers = self.getGCKeyAlarmers()
else:
keyAlarmers = {}
statisticTool = GCLogTool(logger, self.gcKeys)
gcLogFile = LogFile(self.ldap, self.product, self.type, file,
timeInterval, self.gcKeys, [], statisticTool,
wholeFile, keyAlarmers, self.debug, logger,
cubeName)
if gcLogFile.validate():
gcLogFiles.append(gcLogFile)
return gcLogFiles
def __getAnalysisInfo(self, db, condition, queue):
'''
private method
#获取指定证券代码列表所示分析信息，以单个证券代码为单位由queue传递出去
#传出消息按如下字典格式:
{'DB' :'DAY/WEEK/MONTH',
'MSG' :[{'INNER_CODE':<INNER_CODE>,
'LCLOSE':<LCLOSE1>,
'TOPEN':<TOPEN1>,
'TCLOSE':<TCLOSE1>,
'THIGH':<THIGH1>,
'TLOW':<TLOW1>,
'TVOLUME':<TVOLUME1>,
'TVALUE':<TVALUE1>,
... ...
},
{'INNER_CODE':<INNER_CODE>,
'LCLOSE':<LCLOSE2>,
'TOPEN':<TOPEN2>,
'TCLOSE':<TCLOSE2>,
'THIGH':<THIGH2>,
'TLOW':<TLOW2>,
'TVOLUME':<TVOLUME2>,
'TVALUE':<TVALUE2>,
... ...},
... ...]
}
'''
logFile = LogFile(name = 'Indirection')
try:
if ((db == 'WEEK') or (db == 'MONTH')):
field = '''
[SECURITY_ANALYSIS].[FIRST_TRADE_DATE] AS FDATE,
[SECURITY_ANALYSIS].[LAST_TRADE_DATE] AS LDATE,
'''
else:
field = '''
[SECURITY_ANALYSIS].[TRADE_DATE] AS FDATE,
[SECURITY_ANALYSIS].[TRADE_DATE] AS LDATE,
'''
#Instruct SQL command
sqlCmd = '''
SELECT
[SECURITY_ANALYSIS].[INNER_CODE] AS INNER_CODE,
[SECURITY_ANALYSIS].[LCLOSE] AS LCLOSE,
[SECURITY_ANALYSIS].[TOPEN] AS TOPEN,
[SECURITY_ANALYSIS].[TCLOSE] AS TCLOSE,
[SECURITY_ANALYSIS].[THIGH] AS THIGH,
[SECURITY_ANALYSIS].[TLOW] AS TLOW,
[SECURITY_ANALYSIS].[TVOLUME] AS TVOLUME,
[SECURITY_ANALYSIS].[TVALUE] AS TVALUE,
[SECURITY_ANALYSIS].[CHNG] AS CHNG,
[SECURITY_ANALYSIS].[EXCHR] AS EXCHR,
''' + field + '''
''' + dbDict[db]['TRADEDATE'] + '''
FROM
''' + dbDict[db]['TABLE']['ANA'] + ''' AS [SECURITY_ANALYSIS]
WHERE
''' + condition + '''
ORDER BY
INNER_CODE,
''' + dbDict[db]['ORDER']
records = self.mssqlHandle.sqlQuery(sqlCmd)
#records = self.mssqlHandle.sqlQueryProc('p_list_all', (db,))
logFile.logInfo('Get analysis info succeed.')
innerCode = ''
#securityInfoDict = {} 内存较大适用
#单个证券多日衍生数据列表
securityInfoList = list()
#遍历查询记录
for rec in records:
infoDict = {}
#遍历基础信息字段
for column in dbDict[db]['COLUMN']['ANA']:
#查询结果以字段名称为key值存入有序字典
infoDict[column] = rec[column]
#if ((db == 'WEEK') or (db == 'MONTH')):
infoDict['FDATE'] = rec['FDATE']
infoDict['LDATE'] = rec['LDATE']
#不同股票写入不同字典键值中
if ((rec['INNER_CODE'] != innerCode) and (innerCode != '')):
queue.put({'DB':db,'MSG':securityInfoList})
#securityInfoDict[innerCode] = securityInfoList 内存较大适用
securityInfoList = list()
innerCode = rec['INNER_CODE']
elif (innerCode == ''):
innerCode = rec['INNER_CODE']
#将有序字典追加到结果集列表中
securityInfoList.append(infoDict)
else:
#securityInfoDict[innerCode] = securityInfoList 内存较大适用
queue.put({'DB':db,'MSG':securityInfoList})
#securityInfoList = list()
#for innerCode in securityInfoDict: 内存较大适用
# queue.put({'DB':db,'MSG':securityInfoDict[innerCode]})
logFile.logInfo('Standardize analysis info succeed.')
except IndexError:
logFile.logInfo('Invalid table name.')
print 'Invalid table name.'
def CorrectAllSecurityInfo(mssqlDict, mongoDict, processDict, queueDict):
startTime = datetime.now()
print 'Start at:',startTime
print 'DisposeSplitSecurity at',datetime.now()
#分表名获取
splitSecurityDict = DisposeSplitSecurity(mssqlDict, mongoDict)
print 'DisposeAnalysisInfoProc at',datetime.now()
#证券分析数据查询
analysisQueue = Queue(queueDict['ANA'])
monitorProcessDict = dict()
analysisInfoProcessList = list()
for db in dbDict:
process = Process(target=DisposeAnalysisInfoProc, args=(mssqlDict, db, analysisQueue))
process.start()
analysisInfoProcessList.append(process)
monitorProcessDict['Analysis'] = analysisInfoProcessList
#print 'DisposeCalculateExtensionProc at',datetime.now()
#证券衍生数据计算
#extensionQueue = Queue(queueDict['EXT'])
mongoQueue = Queue(queueDict['MONGO'])
excludeRightDividend = ExcludeRightDividend(mssqlDict['SERVER'], mssqlDict['DB'], mssqlDict['USER'], mssqlDict['PWD'])
excludeRightDividendDict = excludeRightDividend.GetExcludeRightDividendInfo()
calculateExtensionProcList = DisposeCalculateExtensionProc(processDict, excludeRightDividendDict, analysisQueue, mongoQueue)
monitorProcessDict['Calculate'] = calculateExtensionProcList
#print 'DisposeExtensionInfoProc at',datetime.now()
#启动写入MSSQL衍生数据进程
#extentionInfoProcList = DisposeExtensionInfoProc(processDict, mssqlDict, extensionQueue, mongoQueue)
#monitorProcessDict['Extension'] = extensionInfoProcList
#print 'DisposeInsertInfoIntoMongoDbProc at',datetime.now()
#启动写入MongoDB数据进程
insertInfoIntoMongoDbProcList = DisposeInsertInfoIntoMongoDbProc(processDict, mongoDict, mongoQueue, splitSecurityDict)
monitorProcessDict['Mongo'] = insertInfoIntoMongoDbProcList
terminateNum = 0
while True:
#移除已终止的进程
for processName in monitorProcessDict:
for process in monitorProcessDict[processName]:
if (process.is_alive() == False):
monitorProcessDict[processName].remove(process)
#检查计算，SQL写入，MongoDB写入进程，如进程意外终止，则重新启动进程
while (len(monitorProcessDict['Calculate']) < processDict['CALCULATE']):
process = Process(target=CalculateExtensionProc, args=(excludeRightDividendDict, analysisQueue, mongoQueue))
process.start()
monitorProcessDict['Calculate'].append(process)
while (len(monitorProcessDict['Mongo']) < processDict['MONGO']):
process = Process(target=InsertInfoIntoMongoDbProc, args=(mongoDict, mongoQueue, splitSecurityDict))
process.start()
monitorProcessDict['Mongo'].append(process)
print datetime.now()
for processName in monitorProcessDict:
print processName,len(monitorProcessDict[processName])
print 'analysisQueue:',analysisQueue.qsize()
#print 'extensionQueue:',extensionQueue.qsize()
print 'mongoQueue:',mongoQueue.qsize()
if ((len(monitorProcessDict['Analysis']) == 0) and
(analysisQueue.qsize() == 0) and
(mongoQueue.qsize() == 0)):
terminateNum += 1
if (terminateNum >= 10):
#关闭所有进程
for processName in monitorProcessDict:
for process in monitorProcessDict[processName]:
if (process.is_alive() == True):
process.terminate()
break
sleep(10)
finalTime = datetime.now()
deltaTime = finalTime - startTime
totalTime = deltaTime.total_seconds()
totalHour = totalTime // 3600
totalMin = (totalTime % 3600) // 60
totalSec = totalTime % 60
print("Total time: %d(h)%d(m)%d(s)" % (totalHour, totalMin, totalSec))
logFile = LogFile(name = 'Indirection')
logFile.logInfo('Correct all security info succeed, total time: ' + str(totalHour) + '(h)' + str(totalMin) + '(m)' + str(totalSec) + '(s)')
