SQL Bigquery - 使用 ARRAY_AGG(Struct) 从 3 个具有主记录且无重复项的表创建数据集

Question

我正在尝试查询 3 个表以创建一个包含主项目数据、交易数据和序列号/位置的数据集。 我的左表是每个项目被提及一次的唯一键，然后交易数据和序列数据有多次重复的键，我希望它们成为项目下的一个集合，以便在 Looker Studio 中使用。

我为 Item to Transaction 和 Item to Serial 工作，但是当我把它们放在一起时，我得到了数据的循环和重复。

这是表格的示例，在此处输入图片描述

这是所需的 output在此处输入图片描述

ItemCode
Manufacturer
Model
Qty on Order
Transaction
    Date
    Qty
Location
    WH-Bin
    Serial
    Qty

这是我用来连接两个表的代码......

with nested as (

  SELECT
    i.ItemCode,i.Manuf,i.ItemName,i.QtyonOrder,i.AvgPrice,
    array_agg(struct (t.Date, t.Quanitty)) as Transactions, 

  FROM
    ItemMasterData i

  Left join
    Transactions t on i.ItemCode = t.ItemCode

  group by i.ItemCode,i.Manuf,i.ItemName,i.QtyonOrder,i.AvgPrice,
)

 
Select * from nested,

当我并列添加另一个

array_agg(struct (l.WHBin, l.Serial, l.Qty)) as Transactions,

Left Join Transactions l on i.ItemCode = l.ItemCode 

我最终用循环填充了库存和交易数据的空白。

Answer 1

我猜您想使用子 select 在表中生成 arrays。

With ItemMasterData as (Select *, rand()*1000 as info_fields  from unnest(["IT0001","IT0002"]) ItemCode ),
Transactions as (Select ItemCode, Date from ItemMasterData,unnest(generate_date_array("2022-01-01","2022-01-03")) as Date),
Location as (Select ItemCode, rand()*100 as loc from unnest(["IT0001","IT0002","IT---"]) ItemCode,unnest(generate_array(1,5)) as Date)

select ItemCode , info_fields,
((Select array_agg(struct(date,"..." as other_fields)) from Transactions t where i.ItemCode=t.ItemCode group by ItemCode )) as Transactions,
((Select array_agg(struct(loc,"---" as another_fields)) from Location L where i.ItemCode=L.ItemCode group by ItemCode )) as Locations,

from
 ItemMasterData i

另一种方法是生成行号并以此合并表格。

With ItemMasterData as (Select *, rand()*1000 as info_fields  from unnest(["IT0001","IT0002"]) ItemCode ),
Transactions as (Select ItemCode, Date from ItemMasterData,unnest(generate_date_array("2022-01-01","2022-01-03")) as Date),
Location as (Select ItemCode, rand()*100 as loc from unnest(["IT0001","IT0002","IT---"]) ItemCode,unnest(generate_array(1,5)) as Date)

select * #except(dummy_row),
from
 ItemMasterData

Left join
  (Select *, row_number() over (partition by ItemCode) as dummy_row from Transactions)
  using(ItemCode) 
full join
  (Select *,row_number() over (partition by ItemCode) as dummy_row from Location)
  using(ItemCode,dummy_row)

group by 1
order by 1,2