Question

精彩的SQL编写者，

我试图获取某个范围内的所有日期（存储为T1中的记录），以及相关表（T2）中记录值的总和。但是，T2中的某些记录将被第三个表（T3）中的字段值过滤掉。

假设这样的表：

TABLE T1
| MonthYearKey |
|==============|
| 201401       |
| 201402       |
| 201403       |
| 201404       |
| 201405       |
| 201406       |

TABLE T2
| MonthYearKey | NextKey | MyValue |
|==============+=========+=========|
| 201402       | 6       | 10      |
| 201403       | 6       | 10      |
| 201404       | 6       | 10      |
| 201402       | 8       | 10      |
| 201403       | 8       | 10      |
| 201404       | 8       | 10      |
| 201401       | 10      | 10      |
| 201402       | 10      | 10      |
| 201406       | 10      | 10      |

TABLE T3
| NextKey | IsValid |
|=========+=========|
| 6       | 1       |
| 8       | 1       |
| 10      | 0       |

我正在运行的SQL是：

SELECT T1.MonthYearKey, SUM(ISNULL(T2.MyValue, 0)) AS SumOfValues
FROM T1
  LEFT OUTER JOIN T2 ON T1.MonthYearKey = T2.MonthYearKey
  LEFT OUTER JOIN T3 ON T2.NextKey = T3.NextKey
WHERE ISNULL(T3.IsValid, 1) = 1
GROUP BY T1.MonthYearKey

我预期的输出是：

| MonthYearKey | SumOfValues |
|==============+=============|
| 201401       | 0           |
| 201402       | 20          |
| 201403       | 20          |
| 201404       | 20          |
| 201405       | 0           |
| 201406       | 0           |

但是，正如您在SQL Fiddle中看到的那样，201401和201406个月完全退出了结果。我假设这是因为它选择了NextKey = 10的记录，然后由IsValid = 0过滤掉。

问题：如何获取所有MonthYearKeys，即使是那些当前在我的SQL中被过滤掉的那些？

Answer 1

在where子句中应用过滤器时，会丢失行 - 包括用于group by的列。

相反，使用条件聚合：

SELECT T1.MonthYearKey,
       COALESCE(SUM(case when t3.isvalid is null or t3.isvalid = 1
                         then T2.MyValue
                    end), 0
               ) as SumOfValues
FROM T1 LEFT OUTER JOIN
     T2
     ON T1.MonthYearKey = T2.MonthYearKey LEFT OUTER JOIN T3
     ON T2.NextKey = T3.NextKey
GROUP BY T1.MonthYearKey;

Answer 2

试试这个：

SELECT T1.MonthYearKey, SUM(ISNULL(T2.MyValue, 0)) AS SumOfValues
FROM T1
  LEFT OUTER JOIN (
    SELECT T2.MyValue, T2.MonthYearKey
    FROM T2
    JOIN T3 ON T2.NextKey = T3.NextKey AND ISNULL(T3.IsValid, 1) = 1
  ) T2 ON T1.MonthYearKey = T2.MonthYearKey
GROUP BY T1.MonthYearKey

在此尝试：http://www.sqlfiddle.com/#!3/4a333/26

修改：

没有嵌套查询：

SELECT T1.MonthYearKey, SUM(ISNULL(T2.MyValue, 0)) AS SumOfValues FROM T2 JOIN T3 ON T2.NextKey = T3.NextKey AND T3.IsValid = 1 RIGHT OUTER JOIN T1 ON T1.MonthYearKey = T2.MonthYearKey GROUP BY T1.MonthYearKey

通过更改JOIN优先级
，无需嵌套查询
SELECT T1.MonthYearKey, SUM(ISNULL(T2.MyValue, 0)) AS SumOfValues FROM T1 LEFT OUTER JOIN (T2 JOIN T3 ON T2.NextKey = T3.NextKey AND T3.IsValid = 1) ON T1.MonthYearKey = T2.MonthYearKey GROUP BY T1.MonthYearKey

SQL Fidle：http://www.sqlfiddle.com/#!3/4a333/45

LEFT OUTER JOIN与条件在第三表中

2 个答案: