修复显示11月23号之前的数据

api/api_v1/endpoints/xquery.py

@@ -3,7 +3,7 @@ import mimetypes
 from collections import defaultdict
 import time
 from urllib.parse import quote
-
+import re
 import pandas as pd
 import numpy as np
 from fastapi import APIRouter, Depends, Request
@@ -75,7 +75,42 @@ async def ltv_model_sql(
         split_="""AND 1 """
         news_sql = split_sql[0] + split_+new_sql + split_sql[1] + split_+new_sql+ split_sql[2]+split_+split_sql[3]
         df = await ckdb.query_dataframe(news_sql)
-
+    # 判断11月23号之前的数据
+    list_data_range=analysis.date_range
+    liststr_data_range=[]
+    for i in list_data_range:
+        liststr_data_range.append(str(i))
+    quota = analysis.event_view['quota']
+    #判断是设备LTV则执行下面代码，如是角色实充LTV则不执行
+    if quota == '#distinct_id':
+        if '2021-11-22' in liststr_data_range or '2021-11-22' >=liststr_data_range[-1]:
+            #取搜索最后为11.23号之前的数据
+            if '2021-11-22' >=liststr_data_range[-1]:
+                news_sql=""""""
+                split_sql=sql.split('AND is_new_device = 1')
+                new_sql=split_sql[0]+split_sql[1]+split_sql[2]
+                news_sql+=new_sql
+                df_twenty_three=await ckdb.query_dataframe(news_sql)
+            #取包含有11.23号之前和23号之后的那一段
+            else:
+                start_date=str(list_data_range[0])
+                end_date='2021-11-22'
+                news_sql = """"""
+                split_sql = sql.split('AND is_new_device = 1')
+                for i in split_sql:
+                    news_sql += i
+                #用正则表达式切时间
+                zhengze_time=r'\d{4}-\d{1,2}-\d{1,2}'
+                zhengze_sql=re.split(zhengze_time,news_sql)
+                zz_new_sql=zhengze_sql[0]+start_date+zhengze_sql[1]+end_date+zhengze_sql[2]+start_date+zhengze_sql[3]+end_date+zhengze_sql[4]
+                zz_news_sql=""""""
+                zz_news_sql+=zz_new_sql
+                df_twenty_three = await ckdb.query_dataframe(zz_news_sql)
+            #上下合并两组数据，忽略以前的索引下标
+            df= pd.concat([df,df_twenty_three], axis=0, ignore_index=True)
+            df.sort_values('date', inplace=True)
+            #去重
+            #df.drop_duplicates(inplace=True)
     quota = res['quota']  #字段名
     ltv_n = res['ltv_n']
     #df = await ckdb.query_dataframe(sql)