Sql server 2008 r2 - 冗余行

Question

第一个查询

   --645 rows
    SELECT  *
            FROM    (
                      SELECT DISTINCT
                                cu.*,
                                ROW_NUMBER() OVER ( ORDER BY cu.Id ) AS RowNum
                      FROM      Customers cu
                                LEFT JOIN dbo.CustomerCategories cc
                                ON cc.CustomerId = cu.Id
                                LEFT JOIN dbo.CustomerServices cs
                                 ON cs.CustomerId = cu.Id
                      WHERE     ( @FullName IS NULL
                                  OR cu.FullName LIKE @FullName
                                )
                                AND ( @CategoriesIdXml IS   NULL
                                      OR cc.CategoryId IN ( SELECT  *
                                                            FROM    @CategoriesList )
                                    )
                                AND ( @ServicesIdXml IS NULL
                                      OR cs.ServiceId IN ( SELECT   *
                                                           FROM     @ServicesList )
                                    ) 

                                    ) AS _
            WHERE   RowNum BETWEEN ( @PageIndex - 1 ) * @PageSize + 1
                           AND     @PageIndex * @PageSize

第二次查询

 --41 rows
    SELECT  *
        FROM    (
                  SELECT DISTINCT
                            cu.*,
                            ROW_NUMBER() OVER ( ORDER BY cu.Id ) AS RowNum
                  FROM      Customers cu
                  --          LEFT JOIN dbo.CustomerCategories cc
                  --          ON cc.CustomerId = cu.Id
                  --          LEFT JOIN dbo.CustomerServices cs
                  --           ON cs.CustomerId = cu.Id
                  --WHERE     ( @FullName IS NULL
                  --            OR cu.FullName LIKE @FullName
                  --          )
                  --          AND ( @CategoriesIdXml IS   NULL
                  --                OR cc.CategoryId IN ( SELECT  *
                  --                                      FROM    @CategoriesList )
                  --              )
                  --          AND ( @ServicesIdXml IS NULL
                  --                OR cs.ServiceId IN ( SELECT   *
                  --                                     FROM     @ServicesList )
                  --              ) 

                                ) AS _
        WHERE   RowNum BETWEEN ( @PageIndex - 1 ) * @PageSize + 1
                       AND     @PageIndex * @PageSize

第二个查询返回正确的结果集（41 行），但第一个查询返回错误的 645 行。但是我在两个查询中都使用了DISTINCT，我想知道为什么首先返回太多行。

如何避免？

Answer 1

在创建 ROW_NUMBER() 之后应用 DISTINCT

由于每一行的 ROW_NUMBER() 都不同，因此每一行根据定义都是唯一的。这意味着您似乎有几个选择。

在一个查询中应用 Distinct，然后为 ROW_NUMBER() 包装另一个查询

SELECT
  *
FROM
(
  SELECT
    *,
    ROW_NUMBER() OVER (ORDER BY id) AS row_num
  FROM
  (
    SELECT DISTINCT
      cu.*
    FROM
      <your query>
  )
    AS raw_data
)
  AS ordered_data
WHERE
  RowNum BETWEEN ( @PageIndex - 1 ) * @PageSize + 1
             AND   @PageIndex       * @PageSize

使用 GROUP BY 代替 DISTINCT

SELECT
  *
FROM
(
  SELECT DISTINCT
    cu.*,
    ROW_NUMBER() OVER (ORDER BY id) AS row_num
  FROM
    <your query>
  GROUP BY
    cu.id,
    cu.field1,
    cu.field2,
    etc, etc
)
  AS ordered_data
WHERE
  RowNum BETWEEN ( @PageIndex - 1 ) * @PageSize + 1
             AND   @PageIndex       * @PageSize

Answer 2

ROW_NUMBER 不对，改用DENSE_RANK。

您可以在这里看到不同之处：Difference between ROW_NUMBER, RANK and DENSE_RANK

ROW_NUMBER 将为同一客户提供不同的编号，这不是您想要的，您需要相同的值才能使您的 distinct 起作用。