Question

我有以下格式的行：

enter image description here

我想要一个SQL查询，它将按EntityId对上述行进行分组，并按以下方式聚合位列：

已提交：如果给定EntityId的任何行为0，则为0，否则为1
已审核：如果给定EntityId的任何行为0，则为0，否则为1
查询：如果给定EntityId 1的任何行，则为1，否则为0

我知道我可以通过将位列转换为int并使用Min / Max来实现这一点，但感觉有点像黑客。我想我的日子很慢，错过了明显的解决方案......

这样做的最佳方式是什么？

我使用的是SQL Server 2008 R2，尽管一般的SQL方法最好。

更新

上述行的所需结果集将为：

enter image description here

Answer 1

我认为转换为int可能是最好的，因为没有按位聚合，其他任何东西也都是“hacky”。

为了好玩，这应该可以在不投射位字段的情况下工作;

select
  EntityId,
  1 ^ count(distinct nullif(Submitted, 1)),
  1 ^ count(distinct nullif(Reviewed, 1)),
  count(distinct nullif(Query, 0))
from t
group by EntityId

Answer 2

依赖隐式数据转换：

select entityId
,max(1*submitted)  as submitted
,max(1*reviewed) as reviewed
,max(1*query) as query
from sd_test
group by entityId
order by entityId;

select entityId
,max(sign(submitted))  as submitted
,max(sign(reviewed)) as reviewed
,max(sign(query)) as query
from sd_test
group by entityId
order by entityId

健康警告：

SIGN：Arguments

numeric_expression是精确数字或表达式   近似数值数据类型类别，位数据类型除外。

它适用于位数据类型。由于位可以是0或1，因此符号将始终返回1或0（但实际上，它可能返回-1,0,1）

select sign(cast(1 as bit)); --returns:1
select sign(cast(0 as bit)); --returns:0

Answer 3

你想要结果中的位，对吗？

SELECT
    t.EntityId,
    CAST(MIN(CAST(t.Submitted AS INT)) AS BIT) AS Submitted,
    CAST(MIN(CAST(t.Reviewed AS INT)) AS BIT) AS Reviewed,
    CAST(MAX(CAST(t.Query AS INT)) AS BIT) AS Query
FROM YourTable t
GROUP BY t.EntityId

Answer 4

如果我理解你是对的，你想要这样的东西：

测试数据：

DECLARE @tbl TABLE(EntityId INT,Submitted BIT, Reviewed BIT,Query BIT)
INSERT INTO @tbl VALUES
        (36,1,0,0),
        (52,1,0,0),
        (52,1,1,0),
        (56,1,0,0),
        (56,1,1,0),
        (57,1,0,0),
        (57,1,1,0)

然后是这样的查询：

SELECT
    table1.EntityId,
    MIN(CAST(Submitted AS INT)) AS Submitted,
    MIN(CAST(Reviewed AS INT)) AS Reviewed,
    MAX(CAST(Query AS INT)) AS Query
FROM
    @tbl as table1
GROUP BY
    table1.EntityId

Answer 5

为了一点乐趣，我把它扔在了一起。与先前的答案不同，它不使用任何铸造或转换......

DECLARE @tbl table (
   EntityId  int
 , Submitted bit
 , Reviewed  bit
 , Query     bit
);

INSERT INTO @tbl
  VALUES (36, 1, 0, 0)
       , (52, 1, 0, 0)
       , (52, 1, 1, 0)
       , (56, 1, 0, 0)
       , (56, 1, 1, 0)
       , (57, 1, 0, 0)
       , (57, 1, 1, 0)
;

SELECT DISTINCT
       base.EntityId
     , Coalesce(submitted.Submitted, 0) As Submitted
     , Coalesce(reviewed.Reviewed  , 0) As Reviewed
     , Coalesce(query.Query        , 1) As Query
FROM   @tbl As base
 LEFT
  JOIN @tbl As submitted
    ON submitted.EntityId = base.EntityId
   AND submitted.Submitted = 1
 LEFT
  JOIN @tbl As reviewed
    ON reviewed.EntityId = base.EntityId
   AND reviewed.Reviewed = 1
 LEFT
  JOIN @tbl As query
    ON query.EntityId = base.EntityId
   AND query.Query = 0
;

结果：

EntityId    Submitted   Reviewed    Query
----------- ----------- ----------- -----------
36          1           0           0
52          1           1           0
56          1           1           0
57          1           1           0

SQL聚合位列到单位结果

5 个答案:

SIGN：Arguments