update

api/api_v1/endpoints/query.py

@@ -212,6 +212,15 @@ async def funnel_model(
     groupby = res['groupby']
 
     df = await ckdb.query_dataframe(sql)
+    # 补齐level数据
+    concat_data = []
+    for key, tmp_df in df.groupby(['date'] + groupby):
+        not_exists_level = {i for i in range(1, len(cond_level) + 1)} - set(tmp_df['level'])
+        for item in not_exists_level:
+            key = key if isinstance(key, tuple) else (key,)
+            concat_data.append((*key, item, 0))
+    df = pd.concat([df, pd.DataFrame(concat_data, columns=df.columns)])
+
     # df.set_index('date',inplace=True)
     data_list = []
     date_data = {}
@@ -231,6 +240,34 @@ async def funnel_model(
         tmp['p1'].append(round(v * 100 / tmp_df.loc[1, 'values'], 2))
         # tmp['p2'].append(round(v*100 / tmp_df.loc[i - 1, 'values'], 2))
     data_list.append(tmp)
+
+    # 补齐日期
+    all_idx = {(dt, lv) for dt in date_range for lv in range(1, len(cond_level) + 1)}
+    concat_data = []
+    for i in all_idx - set(df.set_index(['date', 'level']).index):
+        concat_data.append((*i, 0))
+    summary_df = pd.concat(
+        [df[['date', 'level', 'values']], pd.DataFrame(concat_data, columns=['date', 'level', 'values'])])
+    for key, tmp_df in summary_df.groupby('date'):
+        tmp_df = tmp_df.groupby('level').sum()
+        tmp_df.sort_index(inplace=True)
+        for i in tmp_df.index:
+            tmp_df.loc[i, 'values'] = tmp_df.loc[i:]['values'].sum()
+
+        tmp = dict()
+
+        tmp['n'] = tmp_df['values'].to_list()
+        tmp['p1'] = [100]
+        # tmp['p2'] = []
+        for i, v in tmp_df.loc[2:, 'values'].items():
+            var = round(v * 100 / tmp_df.loc[1, 'values'], 2)
+            var = 0 if np.isnan(var) else var
+            tmp['p1'].append(var)
+            # tmp['p2'].append(round(v*100 / tmp_df.loc[i - 1, 'values'], 2))
+
+        _ = date_data.setdefault(key.strftime('%Y-%m-%d'), {})
+        _['总体'] = tmp
+
     if groupby:
         # 补齐数据
         concat_data = []
@@ -253,7 +290,9 @@ async def funnel_model(
             tmp['p1'] = [100]
             # tmp['p2'] = []
             for i, v in tmp_df.loc[2:, 'values'].items():
-                tmp['p1'].append(round(v * 100 / tmp_df.loc[1, 'values'], 2))
+                var = round(v * 100 / tmp_df.loc[1, 'values'], 2)
+                var = 0 if np.isnan(var) else var
+                tmp['p1'].append(var)
                 # tmp['p2'].append(round(v*100 / tmp_df.loc[i - 1, 'values'], 2))
             data_list.append(tmp)
 
@@ -270,13 +309,15 @@ async def funnel_model(
             tmp['p1'] = [100]
             # tmp['p2'] = []
             for i, v in tmp_df.loc[2:, 'values'].items():
-                tmp['p1'].append(round(v * 100 / tmp_df.loc[1, 'values'], 2))
+                var = round(v * 100 / tmp_df.loc[1, 'values'], 2)
+                var = 0 if np.isnan(var) else var
+                tmp['p1'].append(var)
                 # tmp['p2'].append(round(v*100 / tmp_df.loc[i - 1, 'values'], 2))
 
             _ = date_data.setdefault(key[0].strftime('%Y-%m-%d'), {})
             _[key[1]] = tmp
-
+    title = (groupby or ['总体']) + cond_level
-    resp = {'list': data_list, 'date_data': date_data, 'title': groupby + cond_level, 'level': cond_level
+    resp = {'list': data_list, 'date_data': date_data, 'title': title, 'level': cond_level
             }
     return schemas.Msg(code=0, msg='ok', data=resp)
 

models/behavior_analysis.py

@@ -273,7 +273,8 @@ ORDER BY level
             .group_by(sa.Column('date'), *[sa.Column(i.key) for i in self.groupby], sa.Column('level')) \
             .order_by(sa.Column('date'), *[sa.Column(i.key) for i in self.groupby], sa.Column('level'))
         sql = str(qry.compile(compile_kwargs={"literal_binds": True}))
-        sql = sql.replace('_windows_gap_', f"({windows_gap},'strict_increase')")
+        # sql = sql.replace('_windows_gap_', f"({windows_gap},'strict_increase')")
+        sql = sql.replace('_windows_gap_', f"({windows_gap})")
         print(sql)
         return {'sql': sql,
                 'groupby': [i.key for i in self.groupby],