SQL 查询以获取具有多列的第二个最新结果

Question

我正在尝试将 go 的所有不同 part_id（基于 order_date）的第二个最新结果放入我正在制作的报告中，以将其与最新结果进行比较。

评论部分来自我尝试但未成功的另一种方法。

任何帮助是极大的赞赏！

（旁注：我是在 SO 上发帖的新手，如果在其他地方得到回答，我提前道歉，但我找不到与此问题有关的任何内容）

我正在使用以下查询：

SELECT
                              PURCHASE_ORDER.ORDER_DATE
                            , PURC_ORDER_LINE.PART_ID
                            , PURCHASE_ORDER.VENDOR_ID
                            , PURC_ORDER_LINE.LINE_STATUS
                            , PURC_ORDER_LINE.ORDER_QTY
                            , PURC_ORDER_LINE.UNIT_PRICE
                            --, ROW_NUMBER() over (ORDER BY PURCHASE_ORDER.ORDER_DATE DESC)AS ROW
                            , CAST (PURC_ORDER_LINE.ORDER_QTY * PURC_ORDER_LINE.UNIT_PRICE AS VARCHAR) AS TOTAL_COST
                    FROM
                              PURCHASE_ORDER
                              INNER JOIN
                                        PURC_ORDER_LINE
                              ON
                                        PURCHASE_ORDER.ID = PURC_ORDER_LINE.PURC_ORDER_ID

                    WHERE PURCHASE_ORDER.ORDER_DATE < (SELECT MAX(ORDER_DATE) FROM PURCHASE_ORDER) AND PURC_ORDER_LINE.PART_ID = 'XXXX'
                    ORDER BY ORDER_dATE DESC





                    --WHERE PURC_ORDER_LINE.PART_ID = 'XXXX' and PURCHASE_ORDER.ORDER_DATE = (SELECT MAX(ORDER_DATE) FROM PURCHASE_ORDER WHERE ORDER_DATE < (SELECT MAX(ORDER_DATE) FROM PURCHASE_ORDER))

编辑 5/28 深夜：

可以说下面是我需要每个 part_id 的第二个结果的数据集（第二个基于 ORDER_DATE DESC）

+-------------+---------+-----------+
| ORDER_DATE  | PART_ID | VENDOR_ID |
+-------------+---------+-----------+
| 2020-05-29  | XXXX    | CVVB      |
| 2020-05-27  | XXXX    | CVVB      |
| 2020-05-28  | XXXX    | CVVA      |
| 2020-05-28  | YYYY    | GGNB      |
| 2020-04-12  | YYYY    | GGNB      |
| 2020-02-08  | YYYY    | GGNB      |
| 2020-05-28  | ZZZZ    | LLNB      |
| 2019-10-28  | ZZZZ    | LLNB      |
| 2019-05-27  | ZZZZ    | OKIJ      |
+-------------+---------+-----------+

我希望收到以下 output（超过 3 个不同的零件 ID）：

+------------+---------+-----------+
| ORDER_DATE | PART_ID | VENDOR_ID |
+------------+---------+-----------+
| 2020-05-28 | XXXX    | CVVA      |
| 2020-04-12 | YYYY    | GGNB      |
| 2019-10-28 | ZZZZ    | LLNB      |
+------------+---------+-----------+

查询中还有其他列，但将它们格式化为表会花费更长的时间。 我已经省略了示例中的一些列。

另一个编辑

我不确定这些信息是否有帮助，但我正在尝试将最近的结果与之前的结果进行比较，以显示定价和供应商的差异。 我们正在将数据编译到报表生成器中； 我的方法是创建 2 个单独的数据集，一个是最新的，另一个是第二个最新的，然后在 Report Builder 中合并来自数据集的数据。 如果有更简单的方法并且我正朝着错误的方向前进，请告诉我！

例子：

+------------+---------+-----------+-------------+----------+------------+
| ORDER_DATE | PART_ID | VENDOR_ID | Porder_Date | Ppart_ID | pVendor_id |
+------------+---------+-----------+-------------+----------+------------+
| 2020-05-29 | XXXX    | CVVB      | 2020-05-28  | XXXX     | CVVA       |
| 2020-05-28 | YYYY    | GGNB      | 2020-04-12  | YYYY     | GGNB       |
| 2020-05-28 | ZZZZ    | LLNB      | 2019-10-28  | ZZZZ     | LLNB       |
+------------+---------+-----------+-------------+----------+------------+

编辑第二天早上

感谢大家的帮助。 在 Harry 发布了他的解决方案之后，我继续进行了一些微小的编辑以添加我需要的列。 我用原来的 select 语句交换了他的联合部分。 这里的一切似乎正是我正在寻找的！

代码：

;
WITH mycte AS
          (
                    SELECT
                              PURCHASE_ORDER.ORDER_DATE
                            , PURC_ORDER_LINE.PART_ID
                            , PURCHASE_ORDER.VENDOR_ID
                            , PURC_ORDER_LINE.LINE_STATUS
                            , PURC_ORDER_LINE.ORDER_QTY
                            , PURC_ORDER_LINE.UNIT_PRICE
                            , CAST (PURC_ORDER_LINE.ORDER_QTY * PURC_ORDER_LINE.UNIT_PRICE AS VARCHAR) AS TOTAL_COST
                    FROM
                              PURCHASE_ORDER
                              INNER JOIN
                                        PURC_ORDER_LINE
                              ON
                                        PURCHASE_ORDER.ID = PURC_ORDER_LINE.PURC_ORDER_ID
          )
        , mycte2 AS
          (
                    SELECT
                              CONVERT(DATE,order_date) AS order_date
                            , part_id
                            , vendor_id
                            , order_qty
                            , unit_price
                            , total_cost
                            , ROW_NUMBER() over(
                                      PARTITION BY part_id
                                      ORDER BY
                                                CONVERT(DATE,order_date) DESC) AS row_num
                    FROM
                              mycte
          )
SELECT
          mycte2.order_date
        , mycte2.part_id
        , mycte2.vendor_id
        , mycte2.order_qty
        , mycte2.unit_price
        , mycte2.total_cost
        , previous.order_date porder_date
        , previous.part_id    ppart_id
        , previous.vendor_id  pvendor_id
        , previous.order_qty  poqrder_qty
        , previous.unit_price punit_price
        , previous.total_cost ptotal_cost
FROM
          mycte2
          LEFT JOIN
                    mycte2 previous
          ON
                    previous.row_num   = mycte2.row_num +1
                    AND mycte2.part_id = previous.part_id
WHERE
          mycte2.row_num = 1

Answer 1

我认为这样的事情可以得到第二个最近的订单：

;WITH cteOrders AS (
 SELECT ROW_NUMBER() OVER (ORDER BY Order_Date DESC) AS row_num,
    PURCHASE_ORDER.ORDER_DATE
    , PURC_ORDER_LINE.PART_ID
    , PURCHASE_ORDER.VENDOR_ID
    , PURC_ORDER_LINE.LINE_STATUS
    , PURC_ORDER_LINE.ORDER_QTY
    , PURC_ORDER_LINE.UNIT_PRICE
 FROM PURCHASE_ORDER 
 INNER JOIN PURC_ORDER_LINE ON PURCHASE_ORDER.ID = PURC_ORDER_LINE.PURC_ORDER_ID
 WHERE PURCHASE_ORDER.ORDER_DATE < (SELECT MAX(ORDER_DATE) FROM PURCHASE_ORDER) AND PURC_ORDER_LINE.PART_ID = 'XXXX'
 )
 SELECT * FROM cteOrders WHERE row_num = 2

Answer 2

根据您提供的数据，您可以使用 cte 和行号 function 执行此操作。

注意 - 它总是有助于显示整个画面，而不是仅仅要求您需要帮助的部分（通常）.. 因为如果我们能够理解整个问题，更容易回答！

请参阅下面的代码

;with mycte as (


select 
 '2020-05-29' as order_date  , 'XXXX'  as part_id   , 'CVVB' as vendor_id
 union all select
 '2020-05-27'  , 'XXXX'    , 'CVVB'  
  union all select
 '2020-05-28'  , 'XXXX'    , 'CVVA'      
  union all select
 '2020-05-28'  , 'YYYY'    , 'GGNB'     
  union all select
 '2020-04-12'  , 'YYYY'    , 'GGNB'      
  union all select
 '2020-02-08'  , 'YYYY'    , 'GGNB'     
  union all select
 '2020-05-28'  , 'ZZZZ'    , 'LLNB'      
  union all select
 '2019-10-28' , 'ZZZZ'    , 'LLNB'    
  union all select
 '2019-05-27'  , 'ZZZZ'    , 'OKIJ' 
 )
 , mycte2 as (
 Select 
  convert(date,order_date) as order_date
 ,part_id
 ,vendor_id
 ,ROW_NUMBER() over( partition by part_id order by convert(date,order_date) desc) as row_num
 from mycte
)

Select 
mycte2.order_date
,mycte2.part_id
,mycte2.vendor_id
,previous.order_date porder_date
,previous.part_id ppart_id
,previous.vendor_id pvendor_id
from  mycte2

 left join mycte2 previous
    on previous.row_num = mycte2.row_num +1
    and mycte2.part_id = previous.part_id

where mycte2.row_num = 1

结果