Question

首先......我知道在表格中使用逗号分隔值是不好的，我不能改变它。

我有几个包含以下数据的表：

************** Table 1 **********
stock_id   products_id   stock_attributes 
   5271       279           1559,2764

************** Table 2 *********************
products_attributes_id   products_id   options_id   options_values_id
   1559                     279             2               8
   2764                     279             3               63

************** Table 3 ************************
products_options_id     products_options_name
    2                       Size
    3                       Color

************** Table 4 *****************
products_options_values_id      products_options_values_name
    14                                  Pink
    63                                  Mint
    13                                  Black
    8                                   S
    9                                   M
    10                                  L
    11                                  XL

我想要做的是创建一个查询，以获取表1中的字段stock_attributes，并使用表2,3和2中的信息展开它。 4所以我最终得到以下结论：

*********** Resulting Table **********
stock_id    products_id     opt1  opt2  opt3   opt4     
5271        279             Size  S     Color  Mint

我可以在事后以编程方式执行此操作，但我很好奇是否可以在单个SQL查询中完成。我已经找到了类似的问题和答案，如何从逗号分隔的字段中选择一个特定的值，但没有做到这一点。任何帮助表示赞赏。

Answer 1

如果属性数量是固定的。您可以简单地将列拆分为多个。

SELECT stock_id, products_id,
    PARSENAME(REPLACE(stock_attributes,',','.'),2) a1, 
PARSENAME(REPLACE(stock_attributes,',','.'),1) a2
FROM Table1

要进一步扩展并包含其他表中的值，您可以执行简单的JOIN

select pTable.stock_id, pTable.products_id, 
    (select products_options_name from Table3 where products_options_id = t2.options_id) as opt1,
    (select products_options_values_name from Table4 where products_options_values_id = t2.options_values_id) as opt2,
    (select products_options_name from Table3 where products_options_id = t3.options_id) as opt3,
    (select products_options_values_name from Table4 where products_options_values_id = t3.options_values_id) as opt4
from 
(
    SELECT stock_id, products_id,
        PARSENAME(REPLACE(stock_attributes,',','.'),2) a1, 
    PARSENAME(REPLACE(stock_attributes,',','.'),1) a2
    FROM Table1
) pTable
join Table2 t2 on pTable.a1 = t2.products_attributes_id and pTable.products_id = t2.products_id   
join Table2 t3 on pTable.a2 = t3.products_attributes_id and pTable.products_id = t3.products_id

Answer 2

这很长。所以我会停下来，直到数据准备好 Dynamic SQL Pivot

这可以解决任意数量的属性。只需将输出保存到表中o创建存储过程以获取动态数据透视的数据。

tmp： Split 以逗号分隔的行数
product_attributes：只是一个选择来查看递归函数的结果。无法删除
product_details：将属性与其他表连接以获取其值
product_attribute_count：每个属性都需要row_number，因此您可以稍后创建标题。我也意识到你不需要COUNT()
product_pre_pivot：为option_id和option_value_id创建标题和值
product_pivot：将这些对标题分隔为单个列

您可以测试每个步骤以查看结果.... SELECT * FROM [STEP#]

<强> SQL Fiddle Demo

代码：

;WITH tmp([stock_id], [products_id], products_attributes_id, data_r) AS   
(
    SELECT 
        [stock_id], [products_id], 
        LEFT([stock_attributes], Charindex(',', [stock_attributes] + ',') - 1), 
        STUFF([stock_attributes], 1, Charindex(',', [stock_attributes] + ','), '') 
    FROM   
        Table1 

    UNION ALL 

    SELECT 
        [stock_id], [products_id], 
        LEFT(data_r, Charindex(',', data_r + ',') - 1), 
        STUFF(data_r, 1, Charindex(',', data_r + ','), '') 
    FROM   
        tmp 
    WHERE  
        data_r > ''
), product_attributes AS 
(         
    SELECT 
        [stock_id], [products_id], 
        [products_attributes_id]
    FROM   
        tmp 
), product_details AS 
(        
    SELECT 
        pa.*, 
        t2.options_id, t2.options_values_id,
        t3.[products_options_name],
        t4.[products_options_values_name]
    FROM 
        product_attributes pa
    JOIN 
        Table2 t2 ON pa.[products_id] = t2.[products_id]
                  AND pa.[products_attributes_id] = t2.[products_attributes_id]
    JOIN 
        Table3 t3 ON t2.[options_id] = t3.[products_options_id] 
    JOIN 
        Table4 t4 ON t2.[options_values_id] = t4.[products_options_values_id]
), product_attribute_count AS 
(           
    SELECT 
        *,
        row_number() over (PARTITION BY products_id 
                           ORDER BY options_id) as rn,
        count(*) over (partition by products_id) cnt
    FROM 
        product_details
), product_pre_pivot AS 
( 
    SELECT 
        stock_id, products_id,    
        'opt' + CAST(2*rn - 1 as varchar(max))  as header1,
        products_options_name as d_value,
        'opt' + CAST(2*rn  as varchar(max)) as header2,
        products_options_values_name as a_value
    FROM 
        product_attribute_count
), product_pivot AS 
(  
    SELECT 
        stock_id, products_id,
        header1 header, d_value value
    FROM 
        product_pre_pivot

    UNION ALL

    SELECT 
        stock_id, products_id,
        header2 header, a_value value
    FROM 
        product_pre_pivot 
)
SELECT *
FROM product_pivot
ORDER BY header

<强>输出

| stock_id | products_id | header | value |
|----------|-------------|--------|-------|
|     5271 |         279 |   opt1 |  Size |
|     5271 |         279 |   opt2 |     S |
|     5271 |         279 |   opt3 | Color |
|     5271 |         279 |   opt4 |  Mint |

如何编写将逗号分隔的字段扩展为多个字段的SQL查询？

2 个答案: