如何創建一個 sql 查詢獲取表中的第 2 個到最大結果？

Question

我有一張雪花表，我想查看最大日期和第二個到最大日期的結果。 我在下面寫了一個查詢，給出了最大日期的結果

SELECT MAX(FILE_DATE)  "MAX_DATE"
    ,UPPER(RETAIL)  "SHOP"
FROM PRODUCTS
GROUP BY UPPER(RETAIL)

結果是這樣的

MAX_DATE   SHOP
2022-09-27 SHOP1
2022-08-01 SHOP2

我在尋找

2nd_MAX_DATE MAX_DATE   SHOP
2022-08-27   2022-09-27 SHOP1
2022-07-01   2022-08-01 SHOP2

我嘗試使用 where 子句 WHERE FILE_DATE < ( SELECT MAX(FILE_DATE) FROM "PRODUCTS" GROUP BY UPPER(RETAILER)) 但這不起作用

Answer 1

一種方法是使用交叉連接。

這是使用 SQL 服務器語法粗略更改為您的架構的完整示例。

CREATE TABLE Products (
  Id INT PRIMARY KEY, 
  Retail VARCHAR(100), 
  FileDate DATETIME
);

INSERT INTO Products (Id, Retail, FileDate)
VALUES (1, 'shop 1', '2022-09-01'), 
       (2, 'shop 1', '2022-09-01'),
       (3, 'shop 1', '2022-09-01'),
       (4, 'shop 2', '2022-09-01'),
       (5, 'shop 2', '2022-09-01'),
       (6, 'shop 2', '2022-09-30'),
       (7, 'shop 3', '2022-09-01'),
       (8, 'shop 3', '2022-09-01'),
       (9, 'shop 3', '2022-09-20');

SELECT DISTINCT
    MAX(p1.FileDate) "MAX_DATE",
    UPPER(p1.Retail) "SHOP"
FROM Products p1
    CROSS JOIN Products p2
WHERE
    p1.FileDate > p2.FileDate
GROUP BY 
    p1.Retail,
    p2.Retail

Output

2022-09-30|SHOP 2
2022-09-20|SHOP 3

Answer 2

因為你想要每個商店的最新兩個 file_dates，你可以使用 WINDOW function

WITH CTE as(
SELECT FILE_DATE  
    ,UPPER(RETAIL)  "SHOP"
    , ROW_NUMBER() OVER(PARTITION BY UPPER(RETAIL) ORDER BY FILE_DATE DESC) rn
FROM PRODUCTS
)
SELECT "SHOP",MIN(FILE_DATE) second_date, MAX(FILE_DATE) first_date
FROM CTE 
WHERE rn < 3 
GROUP BY "SHOP"

Answer 3

創建或替換 TABLE SNOWPARK_QUICKSTART.TELCO.Products ( ID INT, FileDate TIMESTAMP, Retail VARCHAR(16777216) );

INSERT INTO Products (Id, Retail, FileDate)
VALUES (1, 'shop 1', '2022-09-01'), 
   (2, 'shop 1', '2022-09-01'),
   (3, 'shop 1', '2022-09-01'),
   (4, 'shop 2', '2022-09-01'),
   (5, 'shop 2', '2022-09-01'),
   (6, 'shop 2', '2022-09-30'),
   (7, 'shop 3', '2022-09-01'),
   (8, 'shop 3', '2022-09-01'),
   (9, 'shop 3', '2022-09-20');


SELECT RETAIL
      ,FILEDATE
      ,IFF(RANK()OVER(PARTITION BY RETAIL ORDER BY FILEDATE DESC)=2,'SECOND MAX DATE','MAX DATE') DAY_DESC  
FROM 
    Products 
GROUP BY 
    RETAIL,FILEDATE
QUALIFY 
    RANK()OVER(PARTITION BY RETAIL ORDER BY FILEDATE DESC)<3 

Answer 4

使用 PeteGo 提供的數據作為 CTE：

with Products (Id, Retail, FileDate) as 
(
    select * from values
        (1, 'shop 1', '2022-09-01'::date), 
        (2, 'shop 1', '2022-09-01'::date),
        (3, 'shop 1', '2022-09-01'::date),
        (4, 'shop 2', '2022-09-01'::date),
        (5, 'shop 2', '2022-09-01'::date),
        (6, 'shop 2', '2022-09-30'::date),
        (7, 'shop 3', '2022-09-01'::date),
        (8, 'shop 3', '2022-09-01'::date),
        (9, 'shop 3', '2022-09-20'::date)
)

您可以使用 window function NTH_VALUE然后結合 DISTINCT 去重復：

select distinct
    Retail
    ,nth_value(FileDate, 1) over(partition by retail order by filedate desc) as "1st_MAX_DATE"
    ,nth_value(FileDate, 2) over(partition by retail order by filedate desc) as "2nd_MAX_DATE"
from products

給出：

零售	1st_MAX_DATE	2nd_MAX_DATE
店鋪 1	2022-09-01	2022-09-01
店鋪 2	2022-09-30	2022-09-01
店鋪 3	2022-09-20	2022-09-01

但是您可以看到shop 1的2022-09-01的兩個值被選中。

如果您還選擇了 id，則可以顯示：

select distinct
    Retail
    ,nth_value(FileDate, 1) over(partition by retail order by filedate, id desc) as "1st_MAX_DATE"
    ,nth_value(FileDate, 2) over(partition by retail order by filedate, id desc) as "2nd_MAX_DATE"
    ,nth_value(id, 1) over(partition by retail order by filedate, id desc) as "1st_id"
    ,nth_value(id, 2) over(partition by retail order by filedate, id desc) as "2nd_id"
from products

零售	1st_MAX_DATE	2nd_MAX_DATE	1st_id	2nd_id
店鋪 1	2022-09-01	2022-09-01	3個	2個
店鋪 2	2022-09-01	2022-09-01	5個	4個
店鋪 3	2022-09-01	2022-09-01	8個	7

因此，您可以刪除商店/日期的重復數據：

select distinct
    Retail
    ,nth_value(FileDate, 1) over(partition by retail order by filedate desc) as "1st_MAX_DATE"
    ,nth_value(FileDate, 2) over(partition by retail order by filedate desc) as "2nd_MAX_DATE"
from (
    select distinct Retail, FileDate from products     
)

現在按照我最初的預期給出了null ：

零售	1st_MAX_DATE	2nd_MAX_DATE
店鋪 1	2022-09-01	null
店鋪 2	2022-09-30	2022-09-01
店鋪 3	2022-09-20	2022-09-01