埋点数据管理类

src/core/manager/user_event_manager.py

+# -*- coding: utf-8 -*-
+
+import pandas as pd
+
+from ydl_ai_recommender.src.core.manager import DatabaseDataManager
+from ydl_ai_recommender.src.core.manager import UserProfileManager
+from ydl_ai_recommender.src.utils.log import create_logger
+
+
+EXPOSE_CLICK_SQL = """
+SELECT
+    t1.session_id,
+    t1.user_ids,
+    t1.duration,
+    t1.cnt AS ae_cnt,
+    t2.cnt AS doctor_card_cnt,
+    t3.cnt AS private_chat_cnt,
+    t1.doctor_ids AS ae_doctor_ids,
+    t2.doctor_ids AS doctor_card_doctor_ids,
+    t3.doctor_ids AS private_chat_doctor_ids
+FROM
+    (
+        SELECT
+            session_id,
+            GROUP_CONCAT(DISTINCT(uid)) AS user_ids,
+            GROUP_CONCAT(DISTINCT(doctor_id)) AS doctor_ids,
+            ROUND((MAX(client_time) - MIN(client_time)) / 1000 / 60, 1) AS duration,
+            COUNT(1) AS cnt
+        FROM
+            dwd.dwd_event_doctor_ae
+        WHERE
+            dt >= '{start_date}'
+            AND dt < '{end_date}'
+            AND uid IS NOT NULL
+            AND uid <> 0
+        GROUP BY
+            session_id
+    ) t1
+    LEFT JOIN (
+        SELECT
+            session_id,
+            GROUP_CONCAT(DISTINCT(doctor_id)) AS doctor_ids,
+            COUNT(1) AS cnt
+        FROM
+            dwd.dwd_event_doctor_card_click
+        WHERE
+            dt >= '{start_date}'
+            AND dt < '{end_date}'
+        GROUP BY
+            session_id
+    ) t2 ON t1.session_id = t2.session_id
+    LEFT JOIN (
+        SELECT
+            session_id,
+            GROUP_CONCAT(DISTINCT(doctor_id)) AS doctor_ids,
+            COUNT(1) AS cnt
+        FROM
+            dwd.dwd_event_private_chat_click
+        WHERE
+            dt >= '{start_date}'
+            AND dt < '{end_date}'
+        GROUP BY
+            session_id
+    ) t3 ON t1.session_id = t3.session_id
+WHERE
+    t2.cnt IS NOT NULL
+    OR t2.cnt IS NOT NULL
+"""
+
+EXPOSE_CLICK_USER_SQL = """
+
+"""
+
+class UserEventManager(DatabaseDataManager):
+    """
+    埋点数据管理
+    """
+
+    def __init__(self, client=None) -> None:
+        super().__init__(client, create_logger(__name__, 'user_event_manager.log'))
+
+    def update_data(self, start_date: str = '2023-01-01', end_date: str = '2023-02-01'):
+        """ 从数据库中拉取曝光点击数据并保存 """
+
+        sql = EXPOSE_CLICK_SQL.format(start_date=start_date, end_date=end_date)
+        _, all_data = self.fetch_data_from_db(sql)
+
+        df = pd.DataFrame(all_data)
+        self.save_xlsx_data(df, f'expose_click_{start_date}_{end_date}.xlsx')
+
+        # 相关用户的画像数据也要下载
+        manager = UserProfileManager()
+        sql = 'SELECT {} FROM ads.ads_register_user_profiles'.format(manager.select_items_str)
+        sql += f" WHERE uid IN (SELECT uid FROM dwd.dwd_event_doctor_ae WHERE dt >= '{start_date}' AND dt < '{end_date}')"
+        _, all_data = self.fetch_data_from_db(sql)
+
+        df = pd.DataFrame(all_data)
+        self.save_xlsx_data(df, f'expose_click_user_{start_date}_{end_date}.xlsx')
+
+    def load_data(self, start_date: str, end_date: str) -> list:
+        return self.load_xlsx_data(f'expose_click_{start_date}_{end_date}.xlsx')
\ No newline at end of file