def outputa(lastMethod,paras,paraUs):
allu={}
for p in paras:
print lastMethod+"-"+p+"\t"+str(dictInfo.info(paraUs[p]))
for u in paraUs[p]:
if u not in allu:
allu[u]=paraUs[p][u]
else:
allu[u]+=paraUs[p][u]
paraUs[p]={}
print lastMethod+"\t"+str(dictInfo.info(allu))
def outputa(lastMethod, paras, paraUs):
allu = {}
for p in paras:
print lastMethod + "-" + p + "\t" + str(dictInfo.info(paraUs[p]))
for u in paraUs[p]:
if u not in allu:
allu[u] = paraUs[p][u]
else:
allu[u] += paraUs[p][u]
paraUs[p] = {}
print lastMethod + "\t" + str(dictInfo.info(allu))
def count(f):
lineNum=0
uNum=0
us={}
nus={}
uids={}
for line in f:
cols=line.strip().split("\t")
if len(cols) < APP_LOG_COLUMNS4:
continue
lineNum+=1
u=cols[USER_CID]
uuid=cols[UUID_ID]
if uuid not in uids:
uids[uuid]=1
else:
uids[uuid]+=1
if PositiveNumP.match(u):
if u not in us:
us[u]={}
uNum+=1
if uuid not in us[u]:
us[u][uuid]=1
else:
us[u][uuid]+=1
else:
if uuid not in nus:
nus[uuid]=1
else:
nus[uuid]+=1
sum=0
for u in us:
sum+=len(us[u])
if len(us[u]) >= 3:
print u,us[u]
v=0
for uuid in us[u]:
v+=us[u][uuid]
us[u]=v
print "注册用户请求日志行数统计信息"
print dictInfo.info(us)
print "非注册用户请求日志行数统计信息"
print dictInfo.info(nus)
print "注册用户日志行数\t总日志行数\t注册用户占比"
print "%d\t%d\t%.4f"%(uNum,lineNum,float(uNum)/lineNum)
print "总uuid数\t注册uuid+非注册uuid\t注册与非注册uuid重叠比例"
print "%d\t%d\t%.4f"%(len(uids),len(nus)+sum,float(len(nus)+sum)/len(uids)-1.0)
print "注册用户数\t注册uuid数\t注册用户多个uuid比例"
print "%d\t%d\t%.4f"%(len(us),sum,float(sum)/len(us)-1.0)
print "注册用户数\t注册用户日志行数\t注册用户平均行数"
print "%d\t%d\t%.4f"%(len(us),uNum,float(uNum)/len(us))
print "非注册uuid数\t非注册用户日志行数\t非注册用户平均行数"
print "%d\t%d\t%.4f"%(len(nus),lineNum-uNum,float(lineNum-uNum)/len(nus))
def methodDistribute():
ms = {}
for line in sys.stdin:
cols = line.strip().split("\t")
if len(cols) < APP_LOG_COLUMNS:
continue
method = cols[METHOD_CID]
if method not in ms:
ms[method] = 1
else:
ms[method] += 1
for m in ms:
print m, ms[m]
print dictInfo.info(ms)
def methodDistribute():
ms={}
for line in sys.stdin:
cols=line.strip().split("\t")
if len(cols) < APP_LOG_COLUMNS:
continue
method=cols[METHOD_CID]
if method not in ms:
ms[method]=1
else:
ms[method]+=1
for m in ms:
print m,ms[m]
print dictInfo.info(ms)
def ipDis():
ips = {}
lastIp = {}
wrong = 0
total = 0
ws = {}
for line in sys.stdin:
cols = line.strip().split("\t")
if len(cols) < APP_LOG_COLUMNS:
continue
if line.find("mobiledevice.initandroiddevice") > 0:
continue
u = ActionUser(cols)
if u == None or u == "":
continue
ip = cols[IP_CID]
if ip == "":
continue
if ip not in ips:
ips[ip] = u
else:
v = ips[ip]
if type(v) == dict:
if u not in v:
v[u] = 1
else:
if u != lastIp[ip]:
wrong += 1
if ip not in ws:
ws[ip] = 1
else:
ws[ip] += 1
else:
if v != u:
ips[ip] = {}
ips[ip][u] = 1
ips[ip][v] = 1
lastIp[ip] = u
total += 1
for ip in ips:
v = ips[ip]
if type(v) == dict:
ips[ip] = len(v)
else:
ips[ip] = 1
print dictInfo.info(ws)
print "total", total
print "wrong", wrong
print dictInfo.info(ips)
def ipDis():
ips={}
lastIp={}
wrong=0
total=0
ws={}
for line in sys.stdin:
cols=line.strip().split("\t")
if len(cols) < APP_LOG_COLUMNS:
continue
if line.find("mobiledevice.initandroiddevice") > 0:
continue
u=ActionUser(cols)
if u == None or u=="":
continue
ip=cols[IP_CID]
if ip == "":
continue
if ip not in ips:
ips[ip]=u
else:
v=ips[ip]
if type(v) == dict:
if u not in v:
v[u]=1
else:
if u != lastIp[ip]:
wrong+=1
if ip not in ws:
ws[ip]=1
else:
ws[ip]+=1
else:
if v != u:
ips[ip]={}
ips[ip][u]=1
ips[ip][v]=1
lastIp[ip]=u
total+=1
for ip in ips:
v=ips[ip]
if type(v) == dict:
ips[ip]=len(v)
else:
ips[ip]=1
print dictInfo.info(ws)
print "total",total
print "wrong",wrong
print dictInfo.info(ips)
def join():
ds = {}
ds["divTime"] = 0
ds["divNum"] = 0
ds["wrongTime"] = 0
ds["wrongNum"] = 0
ds["not"] = 0
ds["viewNoReceive"] = 0
divs = {}
lastU = ""
ms = {}
for line in sys.stdin:
cols = line.strip().split("\t")
if len(cols) < 5:
continue
u = cols[0]
if lastU == "":
lastU = u
if lastU != u:
output(lastU, ms, ds, divs)
lastU = u
ts = {}
ms = {}
if cols[4].startswith("4"):
continue
method = cols[1]
message = cols[2]
time = int(cols[3])
if message not in ms:
ms[message] = {}
ms[message][method] = time
if lastU != "":
output(lastU, ms, ds, divs)
wrongTime = ds["wrongTime"]
wrongNum = ds["wrongNum"]
divTime = ds["divTime"]
divNum = ds["divNum"]
print "view but not received:%d\n" % (ds["viewNoReceive"])
print "not:%d\n" % (ds["not"])
print "wrong:%d\t%d\t%f\n" % (wrongTime, wrongNum, float(wrongTime) /
(wrongNum + 1e-16))
print "div:%d\t%d\t%f\n" % (divTime, divNum, float(divTime) /
(divNum + 1e-16))
print dictInfo.info(divs)
print ccinfo.info(divs)
def join():
ds={}
ds["divTime"]=0
ds["divNum"]=0
ds["wrongTime"]=0
ds["wrongNum"]=0
ds["not"]=0
ds["viewNoReceive"]=0
divs={}
lastU=""
ms={}
for line in sys.stdin:
cols=line.strip().split("\t")
if len(cols) < 5:
continue
u=cols[0]
if lastU == "":
lastU = u
if lastU != u:
output(lastU,ms,ds,divs)
lastU=u
ts={}
ms={}
if cols[4].startswith("4"):
continue
method=cols[1]
message=cols[2]
time=int(cols[3])
if message not in ms:
ms[message]={}
ms[message][method]=time
if lastU != "":
output(lastU,ms,ds,divs)
wrongTime=ds["wrongTime"]
wrongNum=ds["wrongNum"]
divTime=ds["divTime"]
divNum=ds["divNum"]
print "view but not received:%d\n"%(ds["viewNoReceive"])
print "not:%d\n"%(ds["not"])
print "wrong:%d\t%d\t%f\n"%(wrongTime,wrongNum,float(wrongTime)/(wrongNum+1e-16))
print "div:%d\t%d\t%f\n"%(divTime,divNum,float(divTime)/(divNum+1e-16))
print dictInfo.info(divs)
print ccinfo.info(divs)
def appleUserId():
right = 0
wrong = 0
did = ""
pds = {}
ds = {}
for line in sys.stdin:
cols = line.strip().split("\t")
if len(cols) <= APP_LOG_COLUMNS:
continue
did = cols[UUID_ID]
if did == "haodou" or did == "" or did == "null" or did == "haodoua1000033d709ec6":
continue
ip = cols[IP_CID]
if cols[MEDIA_CID] == "appstore":
uid = uuid(cols)
if uid == None:
continue
if PositiveNumP.match(uid):
if PositiveNumP.match(cols[USER_CID]):
right += 1
print line
else:
wrong += 1
if did not in pds:
pds[did] = {}
if ip not in pds[did]:
pds[did][ip] = 1
else:
if did not in ds:
ds[did] = {}
if ip not in ds[did]:
ds[did][ip] = 1
for did in pds:
pds[did] = len(pds[did])
for did in ds:
ds[did] = len(ds[did])
import dictInfo
print dictInfo.info(ds)
print dictInfo.info(pds)
print "right", right
print "wrong", wrong
def appleUserId():
right=0
wrong=0
did=""
pds={}
ds={}
for line in sys.stdin:
cols=line.strip().split("\t")
if len(cols) <= APP_LOG_COLUMNS:
continue
did=cols[UUID_ID]
if did == "haodou" or did == "" or did == "null" or did == "haodoua1000033d709ec6":
continue
ip=cols[IP_CID]
if cols[MEDIA_CID] == "appstore":
uid=uuid(cols)
if uid == None:
continue
if PositiveNumP.match(uid):
if PositiveNumP.match(cols[USER_CID]):
right+=1
print line
else:
wrong+=1
if did not in pds:
pds[did]={}
if ip not in pds[did]:
pds[did][ip]=1
else:
if did not in ds:
ds[did]={}
if ip not in ds[did]:
ds[did][ip]=1
for did in pds:
pds[did]=len(pds[did])
for did in ds:
ds[did]=len(ds[did])
import dictInfo
print dictInfo.info(ds)
print dictInfo.info(pds)
print "right",right
print "wrong",wrong
def tagidDis():
rkt = {}
kt = {}
t = {}
knt = {}
ks = {}
names = {}
for line in sys.stdin:
cols = line.strip().split("\t")
if len(cols) < METHOD_CID + 1:
continue
v = cols[VERSION_CID]
if v != "4.0" and v != "v4.0":
continue
method = cols[METHOD_CID]
if not method.startswith("search.getlist"):
continue
para = cols[PARA_ID]
tid = getValue(para, "tagid")
kw = getValue(para, "keyword")
rqid = getValue(para, "return_request_id")
if tid == "null":
if kw != None and kw != "":
if kw not in knt:
knt[kw] = 1
else:
knt[kw] += 1
elif tid == "" or tid == None:
if kw not in ks:
ks[kw] = 1
else:
ks[kw] += 1
if tid != None and tid != "" and tid != "null":
if kw != None and kw != "":
if rqid != None and rqid != "":
if tid not in rkt:
rkt[tid] = 1
else:
rkt[tid] += 1
if tid not in kt:
kt[tid] = 1
else:
kt[tid] += 1
if tid not in names:
names[tid] = kw
else:
if tid not in t:
t[tid] = 1
else:
t[tid] += 1
print "len of rkt", len(rkt)
print "len of kt", len(kt)
print "len of t", len(t)
for tid in rkt:
if tid not in names:
names[tid] = ""
print tid, names[tid], rkt[tid]
for tid in t:
if tid in names:
print tid, names[tid], t[tid]
else:
print tid, t[tid]
for tid in kt:
print "kt", tid, names[tid], kt[tid]
info = dictInfo.info(knt)
print info
for k in info["top10"]:
print k[0], k[1]
info = dictInfo.info(ks)
print info
for k in info["top10"]:
print k[0], k[1]
import dictInfo
if __name__ == "__main__":
d = {2: 3, "ty": 4, 10: 4, 12: 4, 99: 100}
print dictInfo.info(d)
print dictInfo.topn(d, 3)
def tagidDis():
rkt={}
kt={}
t={}
knt={}
ks={}
names={}
for line in sys.stdin:
cols=line.strip().split("\t")
if len(cols) < METHOD_CID+1:
continue
v=cols[VERSION_CID]
if v != "4.0" and v != "v4.0":
continue
method=cols[METHOD_CID]
if not method.startswith("search.getlist"):
continue
para=cols[PARA_ID]
tid=getValue(para,"tagid")
kw=getValue(para,"keyword")
rqid=getValue(para,"return_request_id")
if tid == "null":
if kw != None and kw != "":
if kw not in knt:
knt[kw]=1
else:
knt[kw]+=1
elif tid == "" or tid == None:
if kw not in ks:
ks[kw]=1
else:
ks[kw]+=1
if tid != None and tid != "" and tid != "null":
if kw != None and kw != "":
if rqid != None and rqid != "":
if tid not in rkt:
rkt[tid]=1
else:
rkt[tid]+=1
if tid not in kt:
kt[tid]=1
else:
kt[tid]+=1
if tid not in names:
names[tid]=kw
else:
if tid not in t:
t[tid]=1
else:
t[tid]+=1
print "len of rkt",len(rkt)
print "len of kt",len(kt)
print "len of t",len(t)
for tid in rkt:
if tid not in names:
names[tid]=""
print tid,names[tid],rkt[tid]
for tid in t:
if tid in names:
print tid,names[tid],t[tid]
else:
print tid,t[tid]
for tid in kt:
print "kt",tid,names[tid],kt[tid]
info=dictInfo.info(knt)
print info
for k in info["top10"]:
print k[0],k[1]
info=dictInfo.info(ks)
print info
for k in info["top10"]:
print k[0],k[1]
import dictInfo
if __name__=="__main__":
d={2:3,"ty":4,10:4,12:4,99:100}
print dictInfo.info(d)
print dictInfo.topn(d,3)
def output(lastm, us, vs):
s = lastm
info = dictInfo.info(us)
print lastm + "\t" + str(info)
info = dictInfo.info(vs)
print "v_" + lastm + "\t" + str(info)
def output(lastm,us,vs): s=lastm info=dictInfo.info(us) print lastm+"\t"+str(info) info=dictInfo.info(vs) print "v_"+lastm+"\t"+str(info)