SQL自我參照表中的主要類別分組

Question

我需要獲取按主要類別和Seller分組的總銷售額列表。 請注意，主要類別上可能會有銷售（這是我目前可以想到的最好的例子）。

源表

+--------------------------------------+
|ID   |Name        |Seller|Qty|ParentID|
+--------------------------------------+
|10   |Egg         |John  |5  |NULL    |
|10   |Egg         |Anna  |2  |NULL    |
|10-01|Egg - Small |John  |3  |10      |
|10-01|Egg - Small |Anna  |4  |10      |
|10-02|Egg - Medium|John  |2  |10      |
|10-02|Egg - Medium|Bob   |11 |10      |
|10-03|Egg - Large |Anna  |7  |10      |
+--------------------------------------+

所需的輸出

+------------------+
|ID|Name|Seller|Qty|
+------------------+
|10|Egg |John  |10 | <- SUM of all sales John has made for any type of egg
|10|Egg |Anna  |13 |
|10|Egg |Bob   |11 |
+------------------+

我與這個查詢很接近，但是如果有人沒有在主要類別上進行銷售，當我使用MIN(Name)時，他們將得到錯誤的Name 。

當前查詢

SELECT 
    SUBSTRING(t1.ID, 1, 2) AS 'ID',
    MIN(t1.Name) AS 'Name', 
    t1.Seller, 
    SUM(t1.Qty) AS 'Qty'
FROM EggTest t1
GROUP BY 
    SUBSTRING(t1.ID, 1, 2),
    t1.Seller

電流輸出

+--------------------------+
|ID|Name        |Seller|Qty|
+--------------------------+
|10|Egg         |Anna  |13 |
|10|Egg - Medium|Bob   |11 | <- Bob has not made sales on the main category
|10|Egg         |John  |10 |
+--------------------------+

編輯：看到多個答案已經建議SUBSTRING(Name, 1, 3)對我不起作用。 Name並不總是以“ Egg”開頭。

更新：

現在嘗試此查詢：

WITH report AS(
  SELECT 
    ID = CASE WHEN s.ParentID IS NOT NULL THEN s.ParentID ELSE s.ID END,
    Name = CASE WHEN s.ParentID IS NOT NULL THEN p.Name ELSE s.Name END,
    s.Seller,
    s.Qty
  FROM EggTest s
  LEFT JOIN EggTest p ON p.ID = s.ParentID
)

SELECT ID, Name, Seller, SUM(Qty) AS 'Total'
FROM report
GROUP BY ID, Name, Seller;

但是我得到這個奇怪的結果：

+--------------------+
|ID|Name|Seller|Total|
+--------------------+
|10|Egg |Anna  |24   | <- Wrong (Should be 13)
|10|Egg |Bob   |22   | <- Wrong (Should be 11)
|10|Egg |John  |15   | <- Correct(!!)
+--------------------+

在report -table中，我得到了一些重復項：

+------------------+
|ID|Name|Seller|Qty|
+------------------+
|10|Egg |John  |5  |
|10|Egg |Anna  |2  |
|10|Egg |John  |3  |
|10|Egg |John  |3  |
|10|Egg |Anna  |4  |
|10|Egg |Anna  |4  |
|10|Egg |John  |2  |
|10|Egg |John  |2  |
|10|Egg |Anna  |7  |
|10|Egg |Anna  |7  |
|10|Egg |Bob   |11 |
|10|Egg |Bob   |11 |
+------------------+

Answer 1

我將源表名稱視為[ Sales ]

您可以使用以下內容

with report as(
   select ID = case when s.ParentID is not null then s.ParentID else s.ID end,
          Name= case when s.ParentID is not null then p.Name else s.Name end,
          s.Seller,
          s.Qty
   from Sales s
   left join Sales p on p.ID = s.ParentID and p.Seller = s.Seller
)
select ID,Name,Seller,sum(Qty) as Qty
from report
group by ID,Name,Seller

這是使用Distinct的演示

這是一個通過Seller in the left join包含Seller in the left join的演示 ，這將為您提供賣方Bob的項目名稱為NULL ，如果您具有正確的數據完整性，這意味着應該使用單獨的表來顯示條目和類別，則左側聯接應該可以工作

回復您的最后評論，這里是一個演示如何使您的數據清晰

希望這個能對您有所幫助

Answer 2

試試這個查詢。 如果您需要解釋，請詢問:)但這是相當簡單的查詢:)

SELECT MAX(SUBSTRING(ID, 1, 2)) AS ID, 
       SUBSTRING(Name, 1, 3) AS Name,
       Seller, 
       SUM(Qty) AS Qty 
FROM TABLE_NAME
GROUP BY Seller, SUBSTRING(Name, 1, 3)

Answer 3

我不確定ID是否始終為nn[-nn]格式，並且Name可以處理雞蛋以外的其他內容...

此方法在任何情況下均適用：

;with
m as (
    select *, nullif(charindex('-', ID), 0) div_id, nullif(charindex(' - ', name), 0) div_cat
    from EggTest 
),
c as (
    select *, 
        SUBSTRING(ID, 1, isnull(div_id-1, 1000)) main_ID, 
        SUBSTRING(name, 1, isnull(div_cat-1, 1000)) main_cat, 
        nullif(SUBSTRING(name, isnull(div_cat, 1000)+2, 1000), '') sub_cat
    from m
)
select main_ID ID, main_cat [Name], Seller, sum(qty) Qty
from c
group by main_ID, main_cat, seller

輸出：

ID  Name    Seller  Qty
10  Egg     Anna    13
10  Egg     Bob     11
10  Egg     John    10