sql - SQL Server：没有循环的排列/组合

Question

我有两个数据集。第一个是产品配方表以及构成配方的产品。第二个数据集包含按产品划分的单独定价（我可以为单个产品设置多个价格）。

我想要实现的是输出一个结果集，其中包含我的每个产品配方的独特排列。只有所有组件在第二个数据集中都有定价的配方才应该在输出中。

假设：一个配方最多可以配置 5 个组件（不再有）。

DECLARE @ProductRecipe TABLE (ProductRecipeID INT, ComponentProductID INT)

INSERT INTO @ProductRecipe (ProductRecipeID, ComponentProductID) 
VALUES (21, 130), (21, 468), (21, 500), 
       (22, 468), (22, 500), 
       (23, 130), (23, 501)

DECLARE @ComponentPricing TABLE (PricingID INT, ProductID INT)

INSERT INTO @ComponentPricing (PricingID, ProductID)
VALUES (314023, 130), (313616, 130), (313071, 130),
       (312865, 130), (316323, 468), (316329, 468), (398864, 500)

我希望我的输出看起来像这样：

输出示例

我已经尝试过 CTE 和自我连接，但我什至无法接近我想要的输出.. :(

我正在使用 SQL Server 2012

标签： sqlsql-serverpermutation

Answer 1

我将假设您正在使用 SQL Server 2008 或更高版本，这是使该dense_rank()功能正常工作所必需的。

下面的解决方案经历了评论中概述的几个步骤。一个提示是我将其中一条@ProductRecipe记录从(22, 130)更改(22, 468)为 因为我认为它是预期的样本数据，因为Component1所需的输出包括PricingID值 316323 和 316329。

回答：

DECLARE @ProductRecipe TABLE (ProductRecipeID INT, ComponentProductID INT)
INSERT INTO @ProductRecipe (ProductRecipeID, ComponentProductID) VALUES (21, 130)
INSERT INTO @ProductRecipe (ProductRecipeID, ComponentProductID) VALUES (21, 468)
INSERT INTO @ProductRecipe (ProductRecipeID, ComponentProductID) VALUES (21, 500)
INSERT INTO @ProductRecipe (ProductRecipeID, ComponentProductID) VALUES (22, 468) --values were (22, 130) in question
INSERT INTO @ProductRecipe (ProductRecipeID, ComponentProductID) VALUES (22, 500)
INSERT INTO @ProductRecipe (ProductRecipeID, ComponentProductID) VALUES (23, 130)
INSERT INTO @ProductRecipe (ProductRecipeID, ComponentProductID) VALUES (23, 501)

DECLARE @ComponentPricing TABLE (PricingID INT, ProductID INT)
INSERT INTO @ComponentPricing (PricingID, ProductID)
VALUES (314023, 130)
 , (313616, 130)
 , (313071, 130)
 , (312865, 130)
 , (316323, 468)
 , (316329, 468)
 , (398864, 500)

; with base as
    (
        --Joining the two datasets together.
        select pr.ProductRecipeID
        , pr.ComponentProductID
        , cp.PricingID
        from @ProductRecipe as pr
        left join @ComponentPricing as cp on pr.ComponentProductID = cp.ProductID   
    )
    , pr_exclude as
    (
        --Identifying that ProductRecipeID 23 should be excluded because of the 501 NULL value
        select distinct b.ProductRecipeID
        from base as b
        where b.PricingID is null   
    )
    , final_base as
    (
        --Assigning Rank to each ComponentProductID
        select b.ProductRecipeID
        , b.ComponentProductID
        , b.PricingID
        , dense_rank() over (partition by b.ProductRecipeID order by b.ComponentProductID asc) as prod_rnk
        from base as b
        left join pr_exclude as p on b.ProductRecipeID = p.ProductRecipeID
        where 1=1
        and p.ProductRecipeID is null
    )
--Joining it all together
select a.ProductRecipeID
, a.PricingID as Component1
, b.PricingID as Component2
, c.PricingID as Component3
, d.PricingID as Component4
, e.PricingID as Component5
from final_base as a
left join final_base as b on a.ProductRecipeID = b.ProductRecipeID and b.prod_rnk = 2
left join final_base as c on b.ProductRecipeID = c.ProductRecipeID and c.prod_rnk = 3
left join final_base as d on c.ProductRecipeID = d.ProductRecipeID and d.prod_rnk = 4
left join final_base as e on d.ProductRecipeID = e.ProductRecipeID and e.prod_rnk = 5
where a.prod_rnk = 1
order by 1, 2, 3, 4, 5, 6