optimize sql with stat_date

eda/recommended_indexs/cidRate.py

 # -*- coding: UTF-8 -*-
-from utils import con_sql
+from utils import con_sql,get_yesterday_date
 
 
 
 class CidRate(object):
-	def __init__(self, platform, cid_type, ndays=1):
+	def __init__(self, platform, cid_type):
 		"""
-		ndays : 1;2;3;4..  #The number of days from the current time
 		platform : 'all';'ios';'android'
 		cid_type : 'diary';'answer';'question'...
 		"""
-		self.ndays = ndays
 		if platform == "ios":
 			self.platform = "='App Store'"
 		elif platform == "android":
@@ -25,13 +23,13 @@ class CidRate(object):
 		rtype : list
 		"""
 		sql_cid = "select count(cid) from data_feed_click2 \
-		where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
+		where stat_date > '{0}' \
 		and device_type{1} \
-		and cid_type='{2}'".format(self.ndays,self.platform.replace(' ','') if self.platform[-2]=='e' else self.platform,self.cid_type)
+		and cid_type='{2}'".format(get_yesterday_date(),self.platform.replace(' ','') if self.platform[-2]=='e' else self.platform,self.cid_type)
 		cid_clk_count = con_sql(sql_cid)[0][0]
 		sql_all = "select count(cid) from data_feed_click2 \
-		where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
-		and device_type{1}".format(self.ndays, self.platform.replace(' ','') if self.platform[-2]=='e' else self.platform)
+		where stat_date > '{0}' \
+		and device_type{1}".format(get_yesterday_date(), self.platform.replace(' ','') if self.platform[-2]=='e' else self.platform)
 		all_clk_count = con_sql(sql_all)[0][0]
 		cid_clk_rate = round(cid_clk_count/all_clk_count,4)
 		return [platform,cid_clk_count,all_clk_count,cid_clk_rate]
@@ -43,12 +41,12 @@ class CidRate(object):
 		rtype : list
 		"""
 		sql_cid = "select count(cid) from data_feed_exposure2 \
-		where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
-		and device_type{1} and cid_type='{2}'".format(self.ndays,self.platform,self.cid_type)
+		where stat_date > '{0}' \
+		and device_type{1} and cid_type='{2}'".format(get_yesterday_date(),self.platform,self.cid_type)
 		cid_imp_count = con_sql(sql_cid)[0][0]
 		sql_all = "select count(cid) from data_feed_exposure2 \
-		where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
-		and device_type{1}".format(self.ndays, self.platform)
+		where stat_date > '{0}' \
+		and device_type{1}".format(get_yesterday_date(), self.platform)
 		all_imp_count = con_sql(sql_all)[0][0]
 		cid_imp_rate = round(cid_imp_count/all_imp_count,4)
 		return [platform,cid_imp_count,all_imp_count,cid_imp_rate]

eda/recommended_indexs/clkCidUidRate.py

 # -*- coding: UTF-8 -*-
-from utils import con_sql
+from utils import con_sql,get_yesterday_date
 
 
 class ClkCidUidRate(object):
-	def __init__(self, platform, cid_type, ndays=1):
+	def __init__(self, platform, cid_type):
 		"""
-		ndays : 1;2;3;4..  #The number of days from the current time
 		platform : 'all';'ios';'android'
 		cid_type : 'diary';'answer';'question';"everything"...
 		"""
-		self.ndays = ndays
 		if platform == "ios":
 			self.platform = "='App Store'"
 		elif platform == "android":
@@ -27,15 +25,15 @@ class ClkCidUidRate(object):
 		rtype : list
 		"""
 		sql_clk = "select count(distinct(device_id)) from data_feed_click2 \
-		where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
+		where stat_date > '{0}' \
 		and device_type{1} \
-		and cid_type{2}".format(self.ndays,self.platform.replace(' ','') if self.platform[-2]=='e' else self.platform,self.cid_type)
+		and cid_type{2}".format(get_yesterday_date(),self.platform.replace(' ','') if self.platform[-2]=='e' else self.platform,self.cid_type)
 		clk_count = con_sql(sql_clk)[0][0]
 
 		sql_imp = "select count(distinct(device_id)) from data_feed_exposure2 \
-		where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
+		where stat_date > '{0}' \
 		and device_type{1} \
-		and cid_type{2}".format(self.ndays,self.platform,self.cid_type)
+		and cid_type{2}".format(get_yesterday_date(),self.platform,self.cid_type)
 		imp_count = con_sql(sql_imp)[0][0]
 
 		clk_rate = round(clk_count/imp_count,4)

eda/recommended_indexs/func.py

-from utils import con_sql
+from utils import con_sql,get_yesterday_date
 from getClickZeroUidDetail import get_click_zero_uid_count
 from getRegisterUidDetail import get_register_uid_count
 
 
 
 #获取各个平台下的活跃用户点击率
-def get_activate_uid_ctr(platform, ndays=1):
+def get_activate_uid_ctr(platform):
 	"""
-	ndays : 1;2;3;4..  #The number of days from the current time
 	platform : 'all';'ios';'android'
 	rtype : list
 	"""
@@ -18,16 +17,16 @@ def get_activate_uid_ctr(platform, ndays=1):
 	else:
 		platform = " is not null"
 	sql_clk = "select count(device_id) from data_feed_click2 \
-	where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
-	and device_type{1}".format(ndays, platform.replace(' ','') if platform[-2]=='e' else platform)
+	where stat_date > '{0}' \
+	and device_type{1}".format(get_yesterday_date(), platform.replace(' ','') if platform[-2]=='e' else platform)
 	clk_count = con_sql(sql_clk)[0][0]
 	sql_imp = "select count(device_id) from data_feed_exposure2 \
-	where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
+	where stat_date > '{0}' \
 	and device_id in \
 	(select device_id from data_feed_click2 \
-	where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{1} day) \
-	and device_type{2}) \
-	and device_type{3}".format(ndays, ndays, platform.replace(' ','') if platform[-2]=='e' else platform, platform)
+	where stat_date > '{0}' \
+	and device_type{1}) \
+	and device_type{2}".format(get_yesterday_date(), platform.replace(' ','') if platform[-2]=='e' else platform, platform)
 	imp_count = con_sql(sql_imp)[0][0]
 	clk_rate = round(clk_count/imp_count, 4 )
 	if platform == "='App Store'":
@@ -39,9 +38,8 @@ def get_activate_uid_ctr(platform, ndays=1):
 	return [platform, clk_count, imp_count, clk_rate]
 
 #获取活跃用户平均每天曝光次数
-def get_activate_uid_imp_times(city,ndays=1):
+def get_activate_uid_imp_times(city):
 	"""
-	ndays : 1;2;3;4..  #The number of days from the current time
 	city : 'beijing';'all'
 	rtype : list
 	"""
@@ -50,16 +48,16 @@ def get_activate_uid_imp_times(city,ndays=1):
 	else:
 		city = " is not null"
 	sql_uid = "select count(distinct(device_id)) from data_feed_click2 \
-	where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
-	and city_id{1}".format(ndays,city)
+	where stat_date > '{0}' \
+	and city_id{1}".format(get_yesterday_date(),city)
 	sql_uid_count = con_sql(sql_uid)[0][0]
 	sql_imp = "select count(device_id) from data_feed_exposure2 \
 	where device_id in \
 		(select device_id from data_feed_click2 \
-		where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
+		where stat_date > '{0}' \
 		and city_id{1}) \
-	and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
-	and city_id{1}".format(ndays,city)
+	and stat_date > '{0}' \
+	and city_id{1}".format(get_yesterday_date(),city)
 	sql_imp_times = con_sql(sql_imp)[0][0]
 	if city == "='beijing'":
 		city = "北京"

eda/recommended_indexs/topFeatures.py

@@ -4,14 +4,12 @@ from config import DIRECTORY_PATH
 
 
 class TopFeatures(object):
-	def __init__(self, platform, cid_type, top_n=-1, ndays=1):
+	def __init__(self, platform, cid_type, top_n=-1):
 		"""
-		ndays : 1;2;3;4..  #The number of days from the current time
 		platform : 'all';'ios';'android'
 		cid_type : 'diary';'answer';'question'...
 		top_n : the top rows of the result
 		"""
-		self.ndays = ndays
 		if platform == "ios":
 			self.platform = "='App Store'"
 		elif platform == "android":
@@ -25,10 +23,10 @@ class TopFeatures(object):
 	def get_click_times(self):
 		# rtype : dict
 		sql = "select cid,count(cid) from data_feed_click2 \
-		where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
+		where stat_date > '{0}' \
 		and device_type{1} and cid_type='{2}' \
 		group by cid \
-		order by count(cid) desc".format(self.ndays, self.platform.replace(' ','') if self.platform[-2]=='e' else self.platform, self.cid_type)
+		order by count(cid) desc".format(get_yesterday_date(), self.platform.replace(' ','') if self.platform[-2]=='e' else self.platform, self.cid_type)
 		clk_times = tuple2dict(con_sql(sql))
 		return clk_times
 
@@ -36,9 +34,9 @@ class TopFeatures(object):
 	def get_impression_times(self):
 		# rtype : dict
 		sql = "select cid,count(cid) from data_feed_exposure2 \
-		where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
+		where stat_date > '{0}' \
 		and device_type{1} and cid_type='{2}' \
-		group by cid order by count(cid) desc".format(self.ndays, self.platform, self.cid_type)
+		group by cid order by count(cid) desc".format(get_yesterday_date(), self.platform, self.cid_type)
 		imp_times = tuple2dict(con_sql(sql))
 		return imp_times