存储过程中的 CTE 不起作用答案

【问题标题】：CTE in stored procedure not working存储过程中的 CTE 不起作用
【发布时间】：2013-08-02 18:47:21
【问题描述】：

我有一个存储过程正在构建表以便稍后在查询中调用。我不得不添加使用 CTE 的新表，但现在当我使用 ; 运行它时要结束表达式，我收到错误 Incorrect syntax near ';'当我删除它时，我得到了我需要的错误；与 CTE。查询本身在存储过程之外工作。任何帮助将不胜感激......

set @cdq = 'insert #REMOVAL_FLAG
(
    MYKEY,
    REMOVAL_FLAG
)

    WITH 
create_key AS(
        SELECT *,
        (CONVERT(varchar(25), a.NDC11, 101) + CONVERT(varchar(25), a.PharmacyID, 101) + CONVERT(varchar(50),ABS(a.TotalNetCost),101)) as REVERSAL_KEY

        FROM table1 a

),

find_rev AS (

        SELECT *,
            CASE
                WHEN MIN(TotalAmount) OVER (PARTITION BY REVERSAL_KEY) < 0 THEN 1 ELSE NULL
            END as Flag

        FROM create_key

),

primary_flag as (

        SELECT TOP 1500000 *,
            DENSE_RANK() OVER (ORDER BY REVERSAL_KEY) as [GROUP],
            ROW_NUMBER() OVER (PARTITION BY NDC11, PharmacyID, TotalNetCost ORDER BY REVERSAL_KEY) as PK

        FROM find_rev

        WHERE Flag = 1

        ORDER BY [GROUP], PK
),

flagged as (
        SELECT *,
            CASE
                WHEN COUNT (PK) OVER (PARTITION BY [GROUP], PK) > 1 THEN ''REMOVE''
            END as REMOVAL_FLAG

        FROM primary_flag
)

SELECT a1.MYKEY, f.REMOVAL_FLAG

FROM table1 a1
LEFT OUTER JOIN flagged f ON (f.MYKEY = a1.MYKEY)


'

【问题讨论】：

标签： sql stored-procedures common-table-expression

【解决方案1】：

insert 语句可以使用 CTE（with 表达式）。在 SQL Server 中，它位于insert 之前，而不是select 之前：

WITH 
create_key AS(
        SELECT *,
        (CONVERT(varchar(25), a.NDC11, 101) + CONVERT(varchar(25), a.PharmacyID, 101) + CONVERT(varchar(50),ABS(a.TotalNetCost),101)) as REVERSAL_KEY

        FROM table1 a

),
find_rev AS (
        SELECT *,
            CASE
                WHEN MIN(TotalAmount) OVER (PARTITION BY REVERSAL_KEY) < 0 THEN 1 ELSE NULL
            END as Flag
        FROM create_key
),
primary_flag as (

        SELECT TOP 1500000 *,
            DENSE_RANK() OVER (ORDER BY REVERSAL_KEY) as [GROUP],
            ROW_NUMBER() OVER (PARTITION BY NDC11, PharmacyID, TotalNetCost ORDER BY REVERSAL_KEY) as PK
        FROM find_rev
        WHERE Flag = 1
        ORDER BY [GROUP], PK
),
flagged as (
        SELECT *,
            CASE
                WHEN COUNT (PK) OVER (PARTITION BY [GROUP], PK) > 1 THEN ''REMOVE''
            END as REMOVAL_FLAG
        FROM primary_flag
)
insert #REMOVAL_FLAG(MYKEY, REMOVAL_FLAG)
    SELECT a1.MYKEY, f.REMOVAL_FLAG
    FROM table1 a1
    LEFT OUTER JOIN flagged f ON (f.MYKEY = a1.MYKEY)

【讨论】：