如何通过 BigQuery 中 STRUCT 类型的表达式解决分组问题

Posted 2023-03-25

【中文标题】如何通过 BigQuery 中 STRUCT 类型的表达式解决分组问题

【英文标题】：How to solve group by expressions of type STRUCT in BigQuery

【发布时间】：2021-09-30 14:52:26

【问题描述】：

我有一张如下表：

-------------------------------------------------------------------------------------
|             Name                    |                      Singer          |   ID |
-------------------------------------------------------------------------------------
|   Stay,Love Yourself,Crazy In Love  |  Justin Bieber,Justin Bieber,Beyonce |   1  |
-------------------------------------------------------------------------------------

我尝试将名称列拆分为多行并将其另存为另一个表：

WITH TABLE2 AS (
    SELECT ID, 
           NAME
    FROM TABLE
    UNNEST (SPLIT(NAME)) AS NAME
)

我明白了：

------------------------
|     Name       |  ID |
------------------------
|     Stay       |  1  |
------------------------
| Love Yourself  |  1  |
------------------------
| Crazy In Love  |  1  |
------------------------

我想统计收听每首歌曲的用户数，但它显示“不允许按 STRUCT 类型的表达式分组”。谁能帮帮我？

SELECT t2.Name,
       COUNT(DISTINCT ID) as num_users
FROM TABLE2 t2
GROUP BY t2.Name.

【参考方案1】：

考虑下面

select song, count(distinct id) users
from table, unnest(split(name)) song
group by song