SQLserver中cube：多维数据集实例详解

上山打老虎

　　1、cube:生成多维数据集，包含各维度可能组合的交叉表格，使用with 关键字连接 with cube

　　根据需要使用union all 拼接

　　判断 某一列的null值来自源数据还是 cube 使用GROUPING关键字

　　GROUPING([档案号]) = 1 ： null值来自cube（代表所有的档案号）
GROUPING([档案号]) = 0 ： null值来自源数据

　　举例：

SELECT * INTO ##GET
FROM 
  (SELECT *
  FROM ( SELECT
    CASE
    WHEN (GROUPING([档案号]) = 1) THEN
    '合计'
    ELSE [档案号]
    END AS '档案号',
    CASE
    WHEN (GROUPING([系列]) = 1) THEN
    '合计'
    ELSE [系列]
    END AS '系列',
    CASE
    WHEN (GROUPING([店长]) = 1) THEN
    '合计'
    ELSE [店长]
    END AS '店长', SUM (剩余次数) AS '总剩余',
    CASE
    WHEN (GROUPING([店名]) = 1) THEN
    '合计'
    ELSE [店名]
    END AS '店名'
  FROM ##PudianCard
  GROUP BY [档案号], [店名], [店长], [系列]
  WITH cube
  HAVING GROUPING([店名]) != 1
    AND GROUPING([档案号]) = 1 --AND GROUPING([系列]) = 1 ) AS M
  UNION
  ALL 
    (SELECT *
    FROM ( SELECT
      CASE
      WHEN (GROUPING([档案号]) = 1) THEN
      '合计'
      ELSE [档案号]
      END AS '档案号',
      CASE
      WHEN (GROUPING([系列]) = 1) THEN
      '合计'
      ELSE [系列]
      END AS '系列',
      CASE
      WHEN (GROUPING([店长]) = 1) THEN
      '合计'
      ELSE [店长]
      END AS '店长', SUM (剩余次数) AS '总剩余',
      CASE
      WHEN (GROUPING([店名]) = 1) THEN
      '合计'
      ELSE [店名]
      END AS '店名'
    FROM ##PudianCard
    GROUP BY [档案号], [店名], [店长], [系列]
    WITH cube
    HAVING GROUPING([店名]) != 1
      AND GROUPING([店长]) != 1 ) AS P )
    UNION
    ALL 
      (SELECT *
      FROM ( SELECT
        CASE
        WHEN (GROUPING([档案号]) = 1) THEN
        '合计'
        ELSE [档案号]
        END AS '档案号',
        CASE
        WHEN (GROUPING([系列]) = 1) THEN
        '合计'
        ELSE [系列]
        END AS '系列',
        CASE
        WHEN (GROUPING([店长]) = 1) THEN
        '合计'
        ELSE [店长]
        END AS '店长', SUM (剩余次数) AS '总剩余',
        CASE
        WHEN (GROUPING([店名]) = 1) THEN
        '合计'
        ELSE [店名]
        END AS '店名'
      FROM ##PudianCard
      GROUP BY [档案号], [店名], [店长], [系列]
      WITH cube
      HAVING GROUPING([店名]) != 1
        AND GROUPING([店长]) != 1 ) AS W )
      UNION
      ALL 
        (SELECT *
        FROM ( SELECT
          CASE
          WHEN (GROUPING([档案号]) = 1) THEN
          '合计'
          ELSE [档案号]
          END AS '档案号',
          CASE
          WHEN (GROUPING([系列]) = 1) THEN
          '合计'
          ELSE [系列]
          END AS '系列',
          CASE
          WHEN (GROUPING([店长]) = 1) THEN
          '合计'
          ELSE [店长]
          END AS '店长', SUM (剩余次数) AS '总剩余',
          CASE
          WHEN (GROUPING([店名]) = 1) THEN
          '合计'
          ELSE [店名]
          END AS '店名'
        FROM ##PudianCard
        GROUP BY [档案号], [店名], [店长], [系列]
        WITH cube
        HAVING GROUPING([店名]) = 1
          AND GROUPING([店长]) = 1
          AND GROUPING([档案号]) = 1 ) AS K ) ) AS T

　　2、rollup:功能跟cube相似

　　3、将某一列的数据作为列名，动态加载，使用存储过程，拼接字符串

DECLARE @st nvarchar (MAX) = '';SELECT @st =@st + 'max(case when [系列]=''' + CAST ([系列] AS VARCHAR) + ''' then [总剩余] else null end ) as [' + CAST ([系列] AS VARCHAR) + '],'
FROM ##GET
GROUP BY [系列]; print @st;

　　4、根据某一列分组，分别建表

SELECT
'select ROW_NUMBER() over(order by [卡项] desc) as [序号], [会员],[档案号],[卡项],[剩余次数],[员工],[店名] into ' + ltrim([店名]) + ' from 查询 where [店名]=''' + [店名] + ''' ORDER BY [卡项] desc'
FROM
查询
GROUP BY
[店名]

　　总结
　　以上就是本文关于SQLserver中cube：多维数据集实例详解的全部内容，希望对大家有所帮助。感兴趣的朋友可以继续参阅：MYSQL子查询和嵌套查询优化实例解析、几个比较重要的MySQL变量、ORACLE SQL语句优化技术要点解析等，有什么问题可以随时留言，小编会及时回复大家的。感谢各位对本站的支持！