def anayAct():
f = open(tempDataDir,"wb")
for i in range(10000,900000,10000):
con = Condition("userid between "+str(i)+" and "+str(i+10000))
actdenseList = dictToOrderList(con.actDense)
datedenseList = dictToOrderList(con.dateDense)
f.write(str(i)+","+str(actdenseList[0])+ "," + str(datedenseList[0]) +"\n")
def calActSummary(sql):
res = dbconn.query(sql)
summary = {}
for r in res:
actSummaryString = r.actSummaryString
actStrings = actSummaryString[1:].split(u"$")
for actString in actStrings:
act,count = actString.split(u"|")[0], actString.split(u"|")[1]
if summary.has_key(act):
summary[act] = summary[act] + 1
else:
summary.update({act:1})
summaryOrder = dictToOrderList(summary)
return summaryOrder
def ana(condition):
cons = [
condition,
condition + " and sumcount = 1",
condition + " and sumcount = 2",
condition + " and sumcount between 3 and 10",
condition + " and sumcount > 10"
]
f = open(tempDataDir,"wb")
for c in cons:
con = Condition(c)
flist = dictToOrderList(con.firstact)
##mlist = dictToOrderList(con.mostact)
f.write(c+"\n")
for v in flist:
f.write(str(v[0])+","+str(v[1])+"\n")