减去雪花/DBT 中的列

Posted 2023-03-29

【中文标题】减去雪花/DBT 中的列

【英文标题】：Subtracting columns in snowflake/DBT

【发布时间】：2021-09-22 12:21:27

【问题描述】：

我有两列，一列是挑选对象，另一列是每个客户的未挑选对象。

我需要创建一个包含最终对象的列，例如：

Customer_ID	Picked.	Unpicked.	. Final Cart
799235	shirt, pants, glasses	. glasses.	shirt, pants
799246	dress, pants, glasses	. pants, dresses	glasses

我试图在 Snowflake 中执行此操作，但看起来定义的函数都不起作用。

客户可以根据需要多次添加和删除项目，我只对最终值感兴趣。

【参考方案1】：

不幸的是，dbt jinja 对这个问题没有用处，因为它（几乎总是）在 SQL 运行时之前编译，所以它无法访问这些值。

这个可以，但是可以用 Snowflake SQL 解决。基本模式是将值展平，然后进行减法，然后将它们聚合在一起。

这里有一些代码：

with carts as (
    select 799235 as customer_id, array_construct('shirt', 'pants', 'glasses')  as picked, array_construct('glasses') as unpicked
    union
    select 799246, array_construct('dress', 'pants', 'glasses'), array_construct('pants', 'dress')
),

picked_expanded as (
    select
        carts.customer_id,
        picked.value::varchar as item
    from carts, lateral flatten ( input => carts.picked ) as picked
),

unpicked_expanded as (
    select
        carts.customer_id,
        unpicked.value::varchar as item
    from carts, lateral flatten ( input => carts.unpicked ) as unpicked
),

combined as (
    select * from picked_expanded
    except
    select * from unpicked_expanded
),

agg as (
    select
        customer_id,
        array_agg(item) as final_cart
    from combined
    group by 1
)

select * from agg

结果如下：

CUSTOMER_ID	FINAL_CART
799246	["glasses"]
799235	["shirt", "pants"]