用sql的cube和rollup实现汇总小计

作者: 42c64edf12e9 | 来源:发表于2019-08-14 11:03 被阅读0次

用sql的cube和rollup实现汇总小计
二. GROUPING 运算符
HQL聚集计算之进阶篇
Oracle 高级分组用法详解
rollup and cube group by
Hive Grouping Sets多维分析
Hive 多分组函数GROUPING SETS,GROUPING
Hive分析窗口函数(五)grouping sets,cube,
Spark DataFrame中rollup和cube使用
Hive中with cube、with rollup、group

表结构：

CREATE TABLE [dbo].[Students](
    [ID] [int] IDENTITY(1,1) NOT NULL,
    [StudentName] [nvarchar](50) NULL,
    [Sex] [int] NOT NULL,
    [GradeName] [nvarchar](50) NULL,
    [ClassName] [nvarchar](50) NULL,
    [BodyWeight] [decimal](18, 2) NOT NULL,
    [Area] [nvarchar](50) NULL,
 CONSTRAINT [PK_Students] PRIMARY KEY CLUSTERED 
(
    [ID] ASC
)WITH (PAD_INDEX  = OFF, STATISTICS_NORECOMPUTE  = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS  = ON, ALLOW_PAGE_LOCKS  = ON) ON [PRIMARY]
) ON [PRIMARY]

GO

测试数据：

INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (1, N'张三', 1, N'高一', N'1班', CAST(140.00 AS Decimal(18, 2)), N'中国')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (2, N'李四', 1, N'高一', N'1班', CAST(140.00 AS Decimal(18, 2)), N'中国')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (3, N'王五', 1, N'高一', N'1班', CAST(155.00 AS Decimal(18, 2)), N'中国')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (4, N'奥巴马', 1, N'高一', N'2班', CAST(138.00 AS Decimal(18, 2)), N'美国')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (5, N'希拉里', 0, N'高一', N'2班', CAST(113.00 AS Decimal(18, 2)), N'美国')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (6, N'习XX', 1, N'高一', N'1班', CAST(110.00 AS Decimal(18, 2)), N'中国')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (7, N'温宝宝', 1, N'高一', N'1班', CAST(200.00 AS Decimal(18, 2)), N'中国')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (8, N'埃希', 0, N'高一', N'1班', CAST(123.00 AS Decimal(18, 2)), N'澳大利亚')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (9, N'卡特琳娜', 0, N'高二', N'1班', CAST(145.00 AS Decimal(18, 2)), N'澳大利亚')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (10, N'德玛西亚', 1, N'高二', N'2班', CAST(90.00 AS Decimal(18, 2)), N'英国')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (11, N'嘉文', 1, N'高二', N'2班', CAST(95.00 AS Decimal(18, 2)), N'英国')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (12, N'德邦', 1, N'高二', N'2班', CAST(102.00 AS Decimal(18, 2)), N'英国')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (13, N'蛮子', 1, N'高三', N'1班', CAST(160.00 AS Decimal(18, 2)), N'刚果')
INSERT [dbo].[Students] ([ID], [StudentName], [Sex], [GradeName], [ClassName], [BodyWeight], [Area]) VALUES (14, N'易大师', 1, N'高三', N'1班', CAST(120.00 AS Decimal(18, 2)), N'刚果'

情况一：只有一个分类统计列，只需要一个合计。只需要增加with rollup即可。

image.png

SELECT CASE WHEN GROUPING(GradeName)=1 THEN '合计' ELSE GradeName END AS 年级 ,
      SUM(CASE WHEN Sex=1 THEN 1 ELSE 0 END) AS 男生数,
      SUM(CASE WHEN Sex=0 THEN 1 ELSE 0 END) AS 女生数,
      COUNT(Sex) AS 总数
FROM dbo.Students
GROUP BY GradeName WITH ROLLUP
ORDER BY GradeName DESC

情况二：有多个分类汇总列，只需要一个合计。增加rollup之后，需要使用GROUPING函数判断。

GROUPING函数指示是否聚合 GROUP BY 列表中的指定列表达式。在结果集中，如果 GROUPING 返回 1 则指示聚合；返回 0 则指示不聚合。如果指定了 GROUP BY，则 GROUPING 只能用在 SELECT <select> 列表、HAVING 和 ORDER BY 子句中。

image.png

SELECT CASE WHEN GROUPING(GradeName)=1 THEN '合计' ELSE GradeName END AS 年级 ,
       ClassName AS 班级 ,
       Area AS 地区 ,
      SUM(CASE WHEN Sex=1 THEN 1 ELSE 0 END) AS 男生数,
      SUM(CASE WHEN Sex=0 THEN 1 ELSE 0 END) AS 女生数,
      COUNT(Sex) AS 总数
FROM dbo.Students
GROUP BY GradeName,ClassName,Area WITH ROLLUP
HAVING GROUPING(GradeName)=1 OR (GROUPING(ClassName)=0 AND GROUPING(Area) =0)
ORDER BY GradeName DESC

情况三：有多个分类汇总列，需要显示全部的合计和小计。不需要增加判断。

image.png

SELECT CASE WHEN GROUPING(GradeName)=1 THEN '合计' ELSE GradeName END AS 年级 ,
       CASE WHEN GROUPING(GradeName)=0 AND GROUPING(ClassName)=1 THEN '小计' ELSE ClassName END AS 班级 ,
      SUM(CASE WHEN Sex=1 THEN 1 ELSE 0 END) AS 男生数,
      SUM(CASE WHEN Sex=0 THEN 1 ELSE 0 END) AS 女生数,
      COUNT(Sex) AS 总数
FROM dbo.Students
GROUP BY GradeName,ClassName WITH ROLLUP
ORDER BY GradeName DESC

情况四：有多个分类汇总列，需要显示部分的合计和小计。增加rollup之后，需要增加判断

image.png

SELECT CASE WHEN GROUPING(GradeName)=1 THEN '年级合计' ELSE GradeName END AS 年级 ,
       CASE WHEN GROUPING(GradeName)=0 AND GROUPING(ClassName)=1 THEN '班级小计' ELSE ClassName END AS 班级 ,
       CASE WHEN GROUPING(ClassName)=0 AND GROUPING(Area)=1 THEN '地区小计' ELSE Area END AS 地区 ,
      SUM(CASE WHEN Sex=1 THEN 1 ELSE 0 END) AS 男生数,
      SUM(CASE WHEN Sex=0 THEN 1 ELSE 0 END) AS 女生数,
      COUNT(Sex) AS 总数,
      GROUPING(GradeName) AS GradeName_G,
      GROUPING(ClassName) AS ClassName_G,
      GROUPING(Area) AS Area_G
FROM dbo.Students
GROUP BY GradeName,ClassName,Area WITH ROLLUP
HAVING GROUPING(GradeName)=1 OR GROUPING(Area)=0 OR GROUPING(ClassName)=0
ORDER BY GradeName DESC

参考：

使用ROLLUP 汇总数据

GROUPING