mysql 的分组函数 ROLLUP 语法

• Oracle: GROUP BY ROLLUP(字段A)

• MySQL: GROUP BY 字段A WITH ROLLUP

解决方案1：使用 WITH ROLLUP和 GROUPING()

SELECT 
    CASE 
        WHEN GROUPING(C_JXDM) = 1 THEN 2
        ELSE 0 
    END AS N_GROUPING,
    -- GROUPING(C_JXDM) C_JXDM_GROUPING,
    -- GROUPING(C_PXLB) C_PXLB_GROUPING,
    COALESCE(C_JXDM, '合计') AS C_JXDM,
    COALESCE(C_PXLB, '') AS C_PXLB,
    COUNT(*) AS N_CNT
FROM (
    -- 模拟你的数据表
    SELECT 'name1' AS name, '103001' AS C_JXDM, '1' AS C_PXLB
    UNION ALL
    SELECT 'name2', '103001', '1'
    UNION ALL
    SELECT 'name3', '103001', '2'
    UNION ALL
    SELECT 'name4', '103001', '3'
    UNION ALL
    SELECT 'name5', '103001', '3'
    UNION ALL
    SELECT 'name6', '112001', '1'
    UNION ALL
    SELECT 'name7', '112001', '1'
    UNION ALL
    SELECT 'name8', '138001', '2'
) AS t
GROUP BY C_JXDM, C_PXLB WITH ROLLUP
HAVING GROUPING(C_PXLB) = 0 OR (GROUPING(C_JXDM) = 1 AND GROUPING(C_PXLB) = 1)
ORDER BY 
    GROUPING(C_JXDM) ASC,
    C_JXDM,
    GROUPING(C_PXLB) ASC,
    C_PXLB;

N_GROUPING	C_JXDM	C_PXLB	N_CNT
0	103001	1	2
0	103001	2	1
0	103001	3	2
0	112001	1	2
0	138001	2	1
2	合计		8

解决方案2：更简洁的写法（MySQL 8.0+）

WITH sample_data AS (
    SELECT 'name1' AS name, '103001' AS C_JXDM, '1' AS C_PXLB
    UNION ALL SELECT 'name2', '103001', '1'
    UNION ALL SELECT 'name3', '103001', '2'
    UNION ALL SELECT 'name4', '103001', '3'
    UNION ALL SELECT 'name5', '103001', '3'
    UNION ALL SELECT 'name6', '112001', '1'
    UNION ALL SELECT 'name7', '112001', '1'
    UNION ALL SELECT 'name8', '138001', '2'
)
SELECT 
    CASE 
        WHEN GROUPING(C_JXDM) = 1 THEN 2
        WHEN GROUPING(C_PXLB) = 0 THEN 0
        ELSE 1
    END AS N_GROUPING,
    -- GROUPING(C_JXDM) C_JXDM_GROUPING,
    -- GROUPING(C_PXLB) C_PXLB_GROUPING,
    COALESCE(C_JXDM, '合计') AS C_JXDM,
    COALESCE(C_PXLB, '') AS C_PXLB,
    COUNT(*) AS N_CNT
FROM sample_data
GROUP BY C_JXDM, C_PXLB WITH ROLLUP
-- HAVING GROUPING(C_PXLB) = 0 OR (GROUPING(C_JXDM) = 1 AND GROUPING(C_PXLB) = 1)
ORDER BY 
    GROUPING(C_JXDM),
    C_JXDM,
    GROUPING(C_PXLB),
    C_PXLB;

N_GROUPING	C_JXDM	C_PXLB	N_CNT
0	103001	1	2
0	103001	2	1
0	103001	3	2
1	103001		5
0	112001	1	2
1	112001		2
0	138001	2	1
1	138001		1
2	合计		8

加上"HAVING GROUPING(C_PXLB) = 0 OR (GROUPING(C_JXDM) = 1 AND GROUPING(C_PXLB) = 1)" 就只列出明细和总计行

解决方案3：如果只需要明细和总计（2级汇总）

WITH sample_data AS (
    SELECT 'name1' AS name, '103001' AS C_JXDM, '1' AS C_PXLB
    UNION ALL SELECT 'name2', '103001', '1'
    UNION ALL SELECT 'name3', '103001', '2'
    UNION ALL SELECT 'name4', '103001', '3'
    UNION ALL SELECT 'name5', '103001', '3'
    UNION ALL SELECT 'name6', '112001', '1'
    UNION ALL SELECT 'name7', '112001', '1'
    UNION ALL SELECT 'name8', '138001', '2'
)
SELECT 
    GROUPING(C_JXDM) + GROUPING(C_PXLB) AS N_GROUPING,
    IFNULL(C_JXDM, '合计') AS C_JXDM,
    IFNULL(C_PXLB, '') AS C_PXLB,
    COUNT(*) AS N_CNT
FROM sample_data
GROUP BY C_JXDM, C_PXLB WITH ROLLUP
HAVING (GROUPING(C_JXDM) = 0 AND GROUPING(C_PXLB) = 0)  -- 明细行
    OR (GROUPING(C_JXDM) = 1 AND GROUPING(C_PXLB) = 1)  -- 总计行
ORDER BY GROUPING(C_JXDM), C_JXDM, GROUPING(C_PXLB), C_PXLB;

N_GROUPING	C_JXDM	C_PXLB	N_CNT
0	103001	1	2
0	103001	2	1
0	103001	3	2
0	112001	1	2
0	138001	2	1
2	合计		8

解释：

1. N_GROUPING 值：

   • 0：表示明细行（既有C_JXDM分组，也有C_PXLB分组）

   • 2：表示总计行（两个维度都汇总了）

2. GROUPING() 函数：

   • 返回0表示该列参与分组

   • 返回1表示该列是ROLLUP汇总的结果

3. COALESCE/IFNULL：用于处理ROLLUP产生的NULL值

4. HAVING 子句：

   • 过滤掉中间的汇总行（如每个C_JXDM的小计），只保留明细和最终总计

如果你需要不同级别的汇总（如小计+总计），可以调整HAVING子句的条件。