在 T-SQL 中使用逻辑函数进行分组

Question

有没有办法在 t-SQL 中使用组来评估 OR 或 AND 逻辑函数？

DECLARE @TimeClock TABLE 
                   (
                       EName  NVARCHAR(14),
                       Time   NVARCHAR(5),
                       OnTime TINYINT
                   )
    
INSERT INTO @TimeClock VALUES ('Adam', '07:55', 1)
INSERT INTO @TimeClock VALUES ('Adam', '08:03', 0)
INSERT INTO @TimeClock VALUES ('Bob',  '07:57', 1)
INSERT INTO @TimeClock VALUES ('Bob',  '07:59', 1)
INSERT INTO @TimeClock VALUES ('Carl', '08:01', 0)
INSERT INTO @TimeClock VALUES ('Carl', '08:02', 0)

SELECT    
    EName,
    OR(OnTime) AS SometimesLate,
    AND(OnTime) AS AlwaysLate,
    NOT(OR(OnTime)) AS NeverLate
FROM 
    @TimeClock
GROUP BY 
    EName

对于一个非常简单的案例，我可以这样做：

SELECT    
    tc.EName,
    IIF(n.EName IS NOT NULL, 1, 0) AS SometimesLate,
    IIF(o.EName IS NULL, 1, 0) AS AlwaysLate,
    IIF(n.EName IS NULL, 1, 0) AS NeverLate
FROM 
    (SELECT DISTINCT EName FROM @TimeClock) tc
LEFT JOIN 
    (SELECT DISTINCT EName FROM @TimeClock WHERE OnTime = 1) o ON o.EName = tc.EName
LEFT JOIN 
    (SELECT DISTINCT EName FROM @TimeClock WHERE OnTime = 0) n ON n.EName = tc.EName

但如果我有很多不同的列（例如，OnTime、迟到 5 分钟、迟到 10 分钟等），就会变得非常密集。

Answer 1

一些数据库具有布尔聚合函数（如 Postgres），但 SQL Server 不是其中之一。

但是，这里不需要那种复杂的逻辑。我们可以使用MIN()、MAX() 和基本算术来模拟布尔聚合，如下所示：

select    
    ename,
    1 - min(ontime) as sometimeslate,
    1 - max(ontime) as alwayslate,
    min(ontime) as neverlate
from @timeclock
group by ename