GROUPING函数来排除group by cube数据问题

原由：group by cube的结果中出现两条维度值同样，汇总值不同的数据。

SELECT * FROM(      
            SELECT NVL(T.REG,'0000') AS REGION_ID,
                   NVL(T.PROJECT_TYPE, '0000') AS PROJECT_TYPE_ID,
                   NVL(T.VOLTAGE_LEVEL, '0000') AS VOL_CLASSIFY_ID,
                   NVL(T.CURRENT_NODE, '0000') AS NODE_ID,
                   NVL(T.TOTAL_CONTROL_INFO, '0000') AS TOTAL_CONTROL_TYPE_ID,
                   SUM(T.CN) AS CN,
                   SUM(T.AMT) AMT
              FROM TMP T
             WHERE 1 = 1
             GROUP BY CUBE(T.REG,
                           T.PROJECT_TYPE,
                           T.VOLTAGE_LEVEL,
                           T.CURRENT_NODE,
                           T.TOTAL_CONTROL_INFO)) T 
 WHERE 1=1
       AND T.REGION_ID='0000' 
       AND T.PROJECT_TYPE_ID='0000' 
       AND T.VOL_CLASSIFY_ID='0000' 
       AND T.TOTAL_CONTROL_TYPE_ID='75'
       AND T.NODE_ID='9';

使用GROUPING函数排除缘由：

SELECT * FROM(      
            SELECT NVL(T.REG,'0000') AS REGION_ID,
                   NVL(T.PROJECT_TYPE, '0000') AS PROJECT_TYPE_ID,
                   NVL(T.VOLTAGE_LEVEL, '0000') AS VOL_CLASSIFY_ID,
                   NVL(T.CURRENT_NODE, '0000') AS NODE_ID,
                   NVL(T.TOTAL_CONTROL_INFO, '0000') AS TOTAL_CONTROL_TYPE_ID,
                   grouping(REG) AS REG,
                   grouping(PROJECT_TYPE) AS PROJECT_TYPE,
                   grouping(VOLTAGE_LEVEL) AS VOLTAGE_LEVEL, 
                   grouping(TOTAL_CONTROL_INFO) AS TOTAL_CONTROL_INFO,
                   grouping(TOTAL_CONTROL_INFO) AS TOTAL_CONTROL_INFO,
                   SUM(T.CN) AS CN,
                   SUM(T.AMT) AMT
              FROM TMP T
             WHERE 1 = 1
             GROUP BY CUBE(T.REG,
                           T.PROJECT_TYPE,
                           T.VOLTAGE_LEVEL,
                           T.CURRENT_NODE,
                           T.TOTAL_CONTROL_INFO)) T 
 WHERE 1=1
       AND T.REGION_ID='0000' 
       AND T.PROJECT_TYPE_ID='0000' 
       AND T.VOL_CLASSIFY_ID='0000' 
       AND T.TOTAL_CONTROL_TYPE_ID='75'
       AND T.NODE_ID='9';

经过GROUPING函数能够看出，一条数据使用了reg来进行汇总，一条则没有使用，这个时候就能够看出来，源表中存在reg为0000的脏数据。

解释：

  如上结果中GROUPING函数返回值“0”和“1”的含义。
  若是显示“1”表示CUBE函数对应的列（例如REG字段）是因为CUBE函数所产生的空值对应的信息，即对此列进行汇总计算后的结果。
  若是显示“0”表示此行对应的这列参未与分组汇总活动。