add log and rm black device_id

eda/gray_stat/node2vec_ctr.py

@@ -24,7 +24,10 @@ class GrayStat(object):
 			and device_id in \
 				(select device_id \
 				from nd_device_cid_similarity_matrix_tmp \
-				where stat_date='{0}')".format(self.ndays,self.cid_type,self.uid_type)
+				where stat_date='{0}') \
+			and device_id not in (select distinct(device_id) from jerry_test.bl_device_list) \
+			and device_id not in (select distinct(device_id) from jerry_prod.blacklist)".format(self.ndays,\
+				self.cid_type,self.uid_type)
 		uid_count = con_sql(sql)[0][0]
 		return uid_count
 
@@ -36,7 +39,10 @@ class GrayStat(object):
 			and device_id in \
 				(select device_id \
 				from nd_device_cid_similarity_matrix_tmp \
-				where stat_date='{0}')".format(self.ndays,self.cid_type,self.uid_type)
+				where stat_date='{0}') \
+			and device_id not in (select distinct(device_id) from jerry_test.bl_device_list) \
+			and device_id not in (select distinct(device_id) from jerry_prod.blacklist)".format(self.ndays,\
+				self.cid_type,self.uid_type)
 		uid_clk_times = con_sql(sql)[0][0]
 		return uid_clk_times
 
@@ -48,7 +54,10 @@ class GrayStat(object):
 			and device_id in \
 				(select device_id \
 				from nd_device_cid_similarity_matrix_tmp \
-				where stat_date='{0}')".format(self.ndays,self.cid_type,self.uid_type)
+				where stat_date='{0}') \
+			and device_id not in (select distinct(device_id) from jerry_test.bl_device_list) \
+			and device_id not in (select distinct(device_id) from jerry_prod.blacklist)".format(self.ndays,\
+				self.cid_type,self.uid_type)
 		uid_imp_times = con_sql(sql)[0][0]
 		return uid_imp_times
 
@@ -69,7 +78,10 @@ class AllStat(object):
 		sql = "select count(distinct(device_id)) from data_feed_click \
 			where stat_date='{0}' \
 			and (cid_type='{1}' or cid_type='diary_video') \
-			and device_id regexp '[{2}]$'".format(self.ndays,self.cid_type,self.uid_type)
+			and device_id regexp '[{2}]$' \
+			and device_id not in (select distinct(device_id) from jerry_test.bl_device_list) \
+			and device_id not in (select distinct(device_id) from jerry_prod.blacklist)".format(self.ndays,\
+				self.cid_type,self.uid_type)
 		uid_count = con_sql(sql)[0][0]
 		return uid_count
 
@@ -77,7 +89,10 @@ class AllStat(object):
 		sql = "select count(device_id) from data_feed_click \
 			where stat_date='{0}' \
 			and (cid_type='{1}' or cid_type='diary_video') \
-			and device_id regexp '[{2}]$'".format(self.ndays,self.cid_type,self.uid_type)
+			and device_id regexp '[{2}]$' \
+			and device_id not in (select distinct(device_id) from jerry_test.bl_device_list) \
+			and device_id not in (select distinct(device_id) from jerry_prod.blacklist)".format(self.ndays,\
+				self.cid_type,self.uid_type)
 		uid_clk_times = con_sql(sql)[0][0]
 		return uid_clk_times
 
@@ -85,7 +100,10 @@ class AllStat(object):
 		sql = "select count(device_id) from data_feed_exposure \
 			where stat_date='{0}' \
 			and cid_type='{1}' \
-			and device_id regexp '[{2}]$'".format(self.ndays,self.cid_type,self.uid_type)
+			and device_id regexp '[{2}]$' \
+			and device_id not in (select distinct(device_id) from jerry_test.bl_device_list) \
+			and device_id not in (select distinct(device_id) from jerry_prod.blacklist)".format(self.ndays,\
+				self.cid_type,self.uid_type)
 		uid_imp_times = con_sql(sql)[0][0]
 		return uid_imp_times
 
@@ -95,6 +113,7 @@ def main():
 	output = OUTPUT_PATH + "ctr.csv"
 	result = []
 	for my_date in date_list:
+		print("stat" + "  " + my_date)
 		g_class = GrayStat("diary","3|4",my_date)
 		a_class = AllStat("diary","3|4",my_date)
 		line1 = str(g_class.get_uid_count())+"\t"+str(g_class.get_uid_imp_times())+"\t"+str(g_class.get_uid_clk_times())