测试播放器P2P统计中对sid个数的统计，这里的问题是，所有日志中sid集合的个数与warning类型的日志中sid集合个数不同

testSidCountInPlayerP2PStats.py

#!/usr/bin/env python

'''

This script is used to test sid stats in player p2p stats.

Created on 2012-11-14

@author: caoxudong
'''

import sys
import datetime
import os
import getopt
import json


today = datetime.date.today()
targetDateStringParam = str(today)
try:
    #check parameter
    opts, args = getopt.getopt(sys.argv[1:], "d:o:h:", ["date=", "offset=", "hour="])
    for opt, arg in opts:
        if opt in ("-o", "-offset"):
            targetDateStringParam = str(today - datetime.timedelta(days = int(arg)))
        if opt in ("-d", "--date"):
            targetDateStringParam = arg
except Exception,data:
    print Exception,":",data


logPath = '/opt/workspace/logs/original/p2p/1hour'
logPrefix = 'ykp2pdata-'
sidDictInAllLog = {}
sidDictInWarningLog = {}
sidDictInWarningLogWithCumulusGetID = {}  


fileNames = os.listdir(logPath);
qulifiedFileNames = []
for f in fileNames:
    try:
        f.index(logPrefix + targetDateStringParam)
        qulifiedFileNames.append(f)
    except ValueError:
        pass
for fileName in qulifiedFileNames:
    fileObject = open(logPath + os.sep + fileName, 'r')
    for line in fileObject:
        elements = line.split(' ')
        targetString = ''.join(elements[3:])
        try:
            tempDict = json.loads(targetString)
        except:
            sys.exit(1)
        logType = tempDict.get('logType')
        sid = tempDict.get('sid')
        sidDictInAllLog[sid] = 1
        if (logType == 'warning') :
            sidDictInWarningLog[sid] = 1
            errorType = tempDict.get('errorType')
            if errorType == 'CumulusGetIDError' or errorType == 'CumulusGetIDSuccess':
                sidDictInWarningLogWithCumulusGetID[sid] = 1

print 'sidDictInAllLog : ',len(sidDictInAllLog)
print 'sidDictInWarningLog ： ', len(sidDictInWarningLog)
print 'sidDictInWarningLogWithCumulusGetID : ', len(sidDictInWarningLogWithCumulusGetID)