Merge branch 'master' of http://git.wanmeizhensuo.com/ML/ffm-baseline

ctr-56.py

@@ -12,16 +12,29 @@ def get_yesterday_date():
     return yesterday
 
 
+def get_black_user():
+    conn2db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_prod')
+    cursor = conn2db.cursor()
+    sql = "select distinct device_id from blacklist"
+    cursor.execute(sql)
+    result = cursor.fetchall()
+    black_user = pd.DataFrame(list(result))[0].values.tolist()
+    cursor.close()
+    conn2db.close()
+    return black_user
+
+
 def get_data():
     conn2db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='eagle')
     cursor = conn2db.cursor()
     sql = "select distinct device_id from ffm_diary_queue_temp where device_id regexp '[5|6]$'"
     cursor.execute(sql)
     result = cursor.fetchall()
-    device = tuple(pd.DataFrame(list(result))[0].values.tolist())
+    device = pd.DataFrame(list(result))[0].values.tolist()
     cursor.close()
-    print(device[0:2])
+    conn2db.close()
 
+    device = tuple(set(device)-set(black))
     return device
 
 
@@ -29,34 +42,22 @@ def ctr_all():
     db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_prod')
     cursor = db.cursor()
     sql_active = "select distinct device_id from data_feed_exposure " \
-                 "where (cid_type = 'diary' or cid_type = 'diary_video') " \
+                 "where cid_type = 'diary'" \
                  "and device_id regexp'[5|6]$' and stat_date = '{}';".format(date)
     cursor.execute(sql_active)
     result = cursor.fetchall()
     tail56 = pd.DataFrame(list(result))[0].values.tolist()
-
+    tail56 = set(tail56)-set(black)
     print("当天尾号5或6活跃用户总数：")
     print(len(tail56))
 
-    cover = len(set(tail56)&set(device_id))
+    cover = len(tail56&set(device_id))
     print("当天尾号5或6活跃用户覆盖数：")
     print(cover)
     cover_percent = format(cover / len(tail56), ".6f")
     print("当天尾号5或6活跃用户覆盖率：")
     print(cover_percent)
-    # sql_click = "select count(cid) from data_feed_click " \
-    #             "where (cid_type = 'diary' or cid_type = 'diary_video') " \
-    #             "and stat_date = '{}' and device_id regexp '[5|6]$';".format(date)
-    # cursor.execute(sql_click)
-    # click = cursor.fetchone()[0]
-    # print("点击数：" + str(click))
-    # sql_exp = "select count(cid) from data_feed_exposure " \
-    #           "where (cid_type = 'diary' or cid_type = 'diary_video') and stat_date = '{}' and " \
-    #           "device_id regexp '[5|6]$'".format(date)
-    # cursor.execute(sql_exp)
-    # exp = cursor.fetchone()[0]
-    # print("曝光数：" + str(exp))
-    # print("点击率：" + str(click / exp))
+
     return len(tail56),cover,cover_percent
 
 
@@ -91,6 +92,7 @@ def rate2file():
 if __name__ == "__main__":
     DIRECTORY_PATH = "/data/ffm/"
     date = get_yesterday_date()
+    black = get_black_user()
     device_id = get_data()
     temp_data = ctr()
     data = ctr_all()