class UserProfileUpdateWriter(ModeWriter):
def __init__(self):
self.client = PyMongoClient()
self.conn = self.client.getConn()
self.modename = "UserProfile"
def setClient(self, client):
self.client = client
self.conn = self.client.getConn()
def remove(self, appkey, modename, tm):
pass
def write(self, data, appkey, modename, modetools, *args, **kwargs):
modename = self.modename
a = time.time()
ip_loc = kwargs["ip_loc"]
curDay = kwargs["today"].replace("-", "")
# fix_deltaday = getDayDelta(curDay, "20160101")
for uid in data:
for ip in data[uid]["jhd_ip"]:
loc = ip_loc.get(ip, {})
if not loc:
continue
prov = loc.get("prov", "#")
city = loc.get("city", "#")
loc_data = {"prov": prov, "city": city}
data[uid].setdefault("locs", []).append(loc_data)
op = []
for uid in data:
locs = data[uid].get("locs", [])
if not locs:
continue
for loc in locs:
op.append(UpdateOne({"_id": uid}, {"$addToSet": {
"locs": loc
}}))
try:
if op:
self.client.bulkWrite(appkey, modename, op)
except:
print(
"Warn: bulkStore 'UserProfile' Rise a error; Switch to Single Mode"
)
try:
replace_onebyone(data, appkey, modename, self.client)
except:
import traceback
print(traceback.print_exc())
print("UserProfileUpdateWriter cost seconds %.3f" %
((time.time() - a), ))
class UserIP(ModeWriter):
'''
{
_id: [ip],
inctag: [number],
timestamp: [number]
}
'''
def __init__(self, client=None):
if client is None:
self.client = PyMongoClient()
else:
self.client = client
self.dbname = "jh"
self.modename = "UserIP"
def setClient(self, client):
self.client = client
def remove(self, *args, **kwargs):
pass
def write(self, data, appkey, modename, modetools, *args, **kwargs):
modename = self.modename
ip_set = set()
for key in data:
# uid = key
for ip in data[key]["jhd_ip"]:
ip_set.add(ip)
op = []
for ip in ip_set:
query_update = {
"$set": {
"timestamp": time.time()
},
"$inc": {
"inctag": 1
},
"$addToSet": {
"appkey": appkey
}
}
op.append(UpdateOne({"_id": ip}, query_update, upsert=True))
try:
if op:
self.client.bulkWrite(self.dbname, modename, op)
except:
print(
"Warn: bulkStore 'jh.UserIP' Rise a error; Switch to Single Mode"
)
for op_item in op:
try:
self.client.bulkWrite(self.dbname, modename, [op_item])
except:
import traceback
print(traceback.print_exc())
class UserEventWriter(ModeWriter):
def __init__(self):
self.client = PyMongoClient()
self.groupwriter = UserEventGroupWriter()
self.modename = "UserEvent"
def setClient(self, client):
self.client = client
def remove(self, appkey, modename, tm):
modename = self.modename
tm = tm.replace("-", "")
# 格式化
tm = time.strftime(
"%Y%m%d", time.localtime(time.mktime(time.strptime(tm, "%Y%m%d"))))
self.client.remove(appkey, modename, {"partition_date": tm})
self.groupwriter.remove(appkey, modename, tm)
def write(self, data, appkey, modename, modetools, *args, **kwargs):
a = time.time()
modename = self.modename
op = []
for doc in data:
try:
# doc["_id"] = ObjectId()
doc["partition_date"] = kwargs["today"].replace("-", "") \
if "today" in kwargs else time.strftime("%Y%m%d", time.localtime(time.time()-86400))
jhd_userkey = doc["jhd_userkey"]
jhd_ts = doc["jhd_ts"]
jhd_eventId = doc["jhd_eventId"]
_id = "_".join(
map(str, [jhd_userkey, jhd_ts, jhd_eventId[:10]]))
doc["_id"] = _id
# op.append(InsertOne(doc))
op.append(ReplaceOne({"_id": _id}, doc, True))
except:
import traceback
print(traceback.print_exc(), doc)
try:
if op:
op = self.client.bulkWrite(appkey, modename, op)
except:
print(
"Warn: bulkStore 'UserProfile' Rise a error; Switch to Single Mode"
)
for op_item in op:
try:
self.client.bulkWrite(appkey, modename, [op_item])
except:
import traceback
print(traceback.print_exc())
finallyMask(appkey, modename, self.client)
self.groupwriter.write(data, appkey, modename, modetools, *args,
**kwargs)
print("UserEventWriter cost seconds %.3f" % ((time.time() - a), ))
class UserActiveUpdateWriter(ModeWriter):
def __init__(self):
self.client = PyMongoClient()
self.modename = "UserActiveUpdate"
def setClient(self, client):
self.client = client
def remove(self, appkey, modename, tm):
pass
def write(self, data, appkey, modename, modetools, *args, **kwargs):
a = time.time()
modename = self.modename
curDay = kwargs["today"].replace("-", "")
# self.client.getConn()[appkey][modename].remove({})
# data_count = self.client.getConn()[appkey][modename].find({}).count()
# if data_count == 0:
# data_cur = self.client.getConn()[appkey]["UserActive"].find({"partition_date": "20161114"}, {"jh_uid": 1})
# for item in data_cur:
# self.client.getConn()[appkey][modename].insert({"_id": item["jh_uid"]})
activelifeabs = getDayDelta(curDay, "20160101")
update_query = {"$addToSet": {"activelifeabs": activelifeabs}}
op = []
for key in data:
# key is userkey
try:
uid = key
op.append(UpdateOne({"_id": uid}, update_query, True))
except:
import traceback
print(traceback.print_exc(), key, data[data])
try:
if op:
op = self.client.bulkWrite(appkey, modename, op)
except:
import traceback
print(traceback.print_exc())
print(
"Warn: bulkStore 'UserActive' Rise a error; Switch to Single Mode"
)
for op_item in op:
try:
self.client.bulkWrite(appkey, modename, [op_item])
except:
import traceback
print(traceback.print_exc())
finallyMask(appkey, modename, self.client)
print("UserActiveUpdateWriter cost seconds %.10f" %
((time.time() - a), ))
class IPStorageMG(IPStorage):
def __init__(self):
self.client = PyMongoClient()
self.conn = self.client.getConn()
def store(self, _id, **kwargs):
op = UpdateOne({"_id": _id}, {"$set": dict({}, **kwargs)}, False)
self.client.bulkWrite("jh", "UserIP", [op])
def storeItem(self, _id, key, value):
pass
def bulkStore(self, data):
op = []
for key in data:
_id = key
op.append(
UpdateOne({"_id": _id}, {"$set": dict({}, **data[key])},
False))
self.client.bulkWrite("jh", "UserIP", op)
class ModeWriteMongoUserIP(ModeWriter):
def __init__(self, mongo_id=1):
self.client = PyMongoClient(mongo_id=mongo_id)
self.conn = self.client.getConn()
self.dbname = "jh"
self.modename = "UserIP"
def remove(self, *args, **kwargs):
pass
def write(self, *args, **kwargs):
'''
:param args: 保留参数
:param kwargs: today = 当天日期（yyyy-mm-dd）
:return:
'''
today = kwargs["today"]
cur = self.conn[self.dbname][self.modename].find(
{
"timestamp": {
"$gte": time.time() - 100
},
"province": {
"$exists": True
},
"city": {
"$exists": True
},
"appkey": {
"$exists": True
}
}, {
"province": True,
"city": True,
"appkey": True
})
update_appkey = {}
for item in cur:
ip = item["_id"]
province = item["province"]
city = item["city"]
if not province:
continue
if not city:
city = province
appkey = item["appkey"]
for a_appkey in appkey:
update_appkey.setdefault(a_appkey, []).append(
UpdateOne(
OrderedDict([("tm", today),
("jhd_loc", {
"$exista": False
}), ("jhd_ip", ip)]), # 需要配合索引使用提高更新速度
{
"$addToSet": {
"jhd_loc": {
"prov": province,
"city": city
}
}
}))
for a_appkey in update_appkey:
a = time.time()
self.client.bulkWrite(a_appkey, "uvfile", update_appkey[a_appkey])
print(__name__,
time.time() - a, a_appkey, len(update_appkey[a_appkey]))
class UserMapMetaWriter(ModeWriter):
def __init__(self, mongo_id=1):
self.client = PyMongoClient(mongo_id=mongo_id)
self.modename = "UserMapMeta"
def setClient(self, client):
self.client = client
def remove(self, appkey, modename, tm):
modename = "UserMapMeta"
pass
def write(self, data, appkey, modename, modetools, *args, **kwargs):
a = time.time()
# self.client.getConn()[appkey][modename].remove({})
modename = self.modename
op = []
for _data in data:
try:
if not _data:
continue
update_query = {
"$addToSet": {
"fields": {
"$each": _data.pop("fields")
}
}
}
_id = _data.pop("_id")
op.append(UpdateOne({"_id": _id}, update_query, True))
for key in _data["field_elems"].keys():
update_query = {
"$addToSet": {
"field_elems.%s" % key: {
"$each": [
item
for item in _data["field_elems"].pop(key)
]
}
}
}
# import json
# print "-"*100, json.dumps(update_query)
# 元素个数 >100 的不更新
op.append(
UpdateOne(
{
"_id": _id,
"field_elems.%s.100" % (key, ): {
"$exists": False
}
}, update_query, True))
except:
import traceback
print(traceback.print_exc(), op)
try:
if op:
op = self.client.bulkWrite(appkey, modename, op)
except:
print(
"Warn: bulkStore 'UserProfile' Rise a error; Switch to Single Mode"
)
for op_item in op:
try:
self.client.bulkWrite(appkey, modename, [op_item])
except:
import traceback
print(traceback.print_exc())
# self.client.getConn()[appkey][modename].remove({"field_elems": {"$exists": True}})
finallyMask(appkey, modename, self.client)
print("UserMapMetaWriter cost seconds %.10f" % ((time.time() - a), ))
class UserProfileWriter(ModeWriter):
def __init__(self, mongo_id=1):
self.client = PyMongoClient(mongo_id=mongo_id)
self.conn = self.client.getConn()
self.modename = "UserProfile"
def setClient(self, client):
self.client = client
self.conn = self.client.getConn()
def remove(self, appkey, modename, tm):
pass
def write(self, data, appkey, modename, modetools, *args, **kwargs):
a = time.time()
curDay = kwargs["today"].replace("-", "")
fix_deltaday = getDayDelta(curDay, "20160101")
modename = self.modename
uids = data.keys()
docs = self.client.find(appkey, modename, {"_id": {"$in": uids}})
oldusers = set()
for doc in docs:
try:
# UserProfile _id 为 userkey
key = doc["_id"]
oldusers.add(key)
# 如果新添加数据比首次访问时间要早，对历史数据进行修正处理
if "lastLoginTime" in data[key]:
lastLoginTime_new = data[key]["lastLoginTime"][:8]
if "firstLoginTime" in doc and "firstLoginTime" in data[key]:
activelife = doc.get("activelife", [0])
firstLoginTime_new = data[key]["firstLoginTime"][:8]
firstLoginTime_old = doc["firstLoginTime"][:8]
if firstLoginTime_new < firstLoginTime_old:
firstLoginDelta = getDayDelta(firstLoginTime_old,
firstLoginTime_new)
doc["activelife"] = map(lambda i: i + firstLoginDelta,
activelife)
data[key] = modetools.mergeUserProfile(data[key], doc)
# 生成用户生命周期数据
firstLoginDay = data[key]["firstLoginTime"][:8]
lastLoginDay = data[key]["lastLoginTime"][:8]
dayDelta = getDayDelta(lastLoginTime_new, firstLoginDay)
data[key].setdefault("activelife", [0]) # 兼容历史数据
if dayDelta not in data[key]["activelife"]:
data[key]["activelife"].append(dayDelta)
data[key]["activelife"].sort()
# 用户绝对活跃数据，起始 日期为 2016-01-01
try:
firstlogin_deltaday = getDayDelta(firstLoginDay,
"20160101")
data[key]["activelifeabs"] = [
firstlogin_deltaday + remain_day
for remain_day in data[key]["activelife"]
]
except:
import traceback
print(traceback.print_exc())
except:
import traceback
print(traceback.print_exc())
# 设置新增版本
# for key in set(uids)-set([item["_id"] for item in docs]):
for key in set(uids) - oldusers:
data[key]["comever"] = data[key]["ver"]
# 用户绝对活跃数据，起始 日期为 2016-01-01
try:
firstLoginDay = data[key]["lastLoginTime"][:8]
firstlogin_deltaday = getDayDelta(firstLoginDay, "20160101")
data[key]["activelifeabs"] = [
firstlogin_deltaday + remain_day
for remain_day in data[key]["activelife"]
]
except:
import traceback
print(traceback.print_exc())
op = []
for key in data:
op.append(ReplaceOne({"_id": key}, data[key], True))
try:
if op:
self.client.bulkWrite(appkey, modename, op)
except:
print(
"Warn: bulkStore 'UserProfile' Rise a error; Switch to Single Mode"
)
try:
replace_onebyone(data, appkey, modename, self.client)
except:
import traceback
print(traceback.print_exc())
finallyMask(appkey, modename, self.client)
print("UserProfileWriter cost seconds %.3f" % ((time.time() - a), ))
class UserCrumbsWriter(ModeWriter):
def __init__(self, mongo_id=1):
self.client = PyMongoClient(mongo_id=mongo_id)
self.conn = self.client.getConn()
self.modename = "uvfile"
# self.store_attachmode = UserIP()
self.attachmode_storers = []
try:
# self.attachmode_storers = [UserIP(), UserProfileUpdateWriter()]
self.attachmode_storers = [UserIP()]
except:
import traceback
print(traceback.print_exc())
def setClient(self, client):
self.client = client
self.conn = self.client.getConn()
def remove(self, appkey, modename, tm):
modename = self.modename
tm = tm.replace("-", "")
tm = time.strftime(
"%Y-%m-%d", time.localtime(time.mktime(time.strptime(tm,
"%Y%m%d"))))
self.client.remove(appkey, modename, {"tm": tm})
def getMeasure(self, activelifeabs, fix_deltaday):
activelifeabs = [i for i in activelifeabs if i <= fix_deltaday]
measure = {
"last7ActiveNum": 0,
"last14ActiveNum": 0,
"last28ActiveNum": 0,
"last30ActiveNum": 0,
}
for activelifeabs_delta in activelifeabs:
delta = fix_deltaday - activelifeabs_delta
if delta <= 6:
measure["last7ActiveNum"] += 1
if delta <= 13:
measure["last14ActiveNum"] += 1
if delta <= 27:
measure["last28ActiveNum"] += 1
if delta <= 29:
measure["last30ActiveNum"] += 1
return measure
def write(self, data, appkey, modename, modetools, *args, **kwargs):
modename = self.modename
# today = kwargs["today"] if "today" in kwargs else time.strftime("%Y-%m-%d", time.localtime(time.time()-86400))
today = kwargs["today"]
today = today.replace("-", "")
uids = data.keys()
yesterday = time.strftime(
"%Y-%m-%d",
time.localtime(
time.mktime(time.strptime(today, "%Y%m%d")) - 86400))
yyyy_mm_dd = time.strftime(
"%Y-%m-%d",
time.localtime(
time.mktime(time.strptime(yesterday.replace("-", ""),
"%Y%m%d")) + 86400))
uvfile = self.client.find(
appkey, "uvfile",
OrderedDict([("tm", yyyy_mm_dd), ("jhd_userkey", {
"$in": uids
})]))
user_profile = self.client.find(appkey, "UserProfile",
{"_id": {
"$in": uids
}})
ips = set()
ip_loc = {}
try:
for uid in data:
ips = ips.union(data[uid].get("jhd_ip"))
ip_loc_cur = self.conn["jh"]["UserIP"].find(
OrderedDict([("_id", {
"$in": list(ips)
}), ("province", {
"$exists": True
}), ("city", {
"$exists": True
})]), {
"province": True,
"city": True
})
for item in ip_loc_cur:
ip = item["_id"]
province = item["province"]
city = item["city"]
if not province:
continue
if not city:
city = province
# ip_loc.setdefault(ip, "_".join([province, city]))
ip_loc.setdefault(ip, {"prov": province, "city": city})
except:
import traceback
print traceback.print_exc()
# 合并 数据
for doc in uvfile:
uid = doc["jhd_userkey"]
data[uid] = modetools.mergeUserCrumbs(doc, data[uid])
try:
ip_lis = data[uid]["jhd_ip"]
data[uid].setdefault("jhd_loc", [])
for ip in ip_lis:
loc = ip_loc.get(ip, None)
if loc and loc not in data[uid]["jhd_loc"]:
data[uid]["jhd_loc"].append(loc)
# tmp = []
# for item in data[uid]["jhd_loc"]:
# if isinstance(item, dict):
# tmp.append(item)
# data[uid]["jhd_loc"] = tmp
except:
import traceback
print traceback.print_exc()
fix_deltaday = getDayDelta(today, "20160101")
# lastActiveInterval
# firstLoginTime
for doc in user_profile:
# print("doc", doc["_id"], fix_deltaday, doc.get("activelifeabs", []))
key = doc["_id"]
tmp = {}
# 获取用户首次登录时间
tmp["firstLoginTime"] = doc.get("firstLoginTime", "unknown")
firstloginday = tmp["firstLoginTime"][:8]
activelifeabs = doc.get("activelifeabs", [])
# 获取用户最近最近活跃信息
tmp["measure"] = self.getMeasure(activelifeabs, fix_deltaday)
tmp["measure"]["firstLoginTime"] = tmp["firstLoginTime"]
# 更新数据
data[key] = dict(data[key], **tmp)
op = []
for key in data:
if "_id" not in data[key]:
data[key]["_id"] = ObjectId()
_id = data[key]["_id"]
op.append(
ReplaceOne({"_id": _id}, modetools.formatList(data[key]),
True))
try:
if op:
self.client.bulkWrite(appkey, modename, op)
except:
print(
"Warn: bulkStore 'uvfile' Rise a error; Switch to Single Mode")
try:
replace_onebyone(data, appkey, modename, self.client)
except:
import traceback
print(traceback.print_exc())
finallyMask(appkey, modename, self.client)
try:
kwargs["ip_loc"] = ip_loc
self.store_attachmode(data, appkey, modename, modetools, *args,
**kwargs)
except:
import traceback
print traceback.print_exc()