T-SQL 枢轴 table

Question

我有一个 table MachineStatus，它存储机器的状态历史记录。 table 看起来像这样：

| MachineStatusId                        |   From              |   To                |   State |   MachineId                           |
----------------------------------------------------------------------------------------------------------------------------------------
|   B065FC43-DBE7-E611-9BDB-801F02F47041 | 2017-01-30 07:00:00 | 2017-01-30 08:00:00 |       1 | 92649C7B-E962-4EB1-B631-00086EECA98A  |
|   B165FC43-DBE7-E611-9BDB-801F02F47041 | 2017-01-30 08:00:00 | 2017-01-30 09:00:00 |     200 | 92649C7B-E962-4EB1-B631-00086EECA98A  |
|   B265FC43-DBE7-E611-9BDB-801F02F47041 | 2017-01-30 07:00:00 | 2017-01-30 08:00:00 |       1 | A2649C7B-E962-4EB1-B631-00086EECA98A  |
|   B365FC43-DBE7-E611-9BDB-801F02F47041 | 2017-01-30 08:00:00 | 2017-01-30 09:00:00 |     500 | A2649C7B-E962-4EB1-B631-00086EECA98A  |

它为每台机器存储，对于每一个状态变化的记录，当[To]时[From]当某个[State]有效时的信息。我喜欢计算每台机器在每个状态所花费的时间。

结果应该是这样的：

| MachineId                              |   Alias         |   State1 |   State200 |   State500 |
-------------------------------------------------------------------------------------------------
|   92649C7B-E962-4EB1-B631-00086EECA98A | Somename        |       60 |         60 |          0 |
|   A2649C7B-E962-4EB1-B631-00086EECA98A | Some other name |       60 |          0 |         60 |

每个州都应表示为一列。

这是我到目前为止尝试过的方法：

SELECT
    MAX(mState.MachineId),
    MAX(m.Alias),
    SUM(CASE mState.State WHEN 1 THEN mState.Diff ELSE 0 END) AS CritTime,
    SUM(CASE mState.State WHEN 200 THEN mState.Diff ELSE 0 END) AS OpTime,
    SUM(CASE mState.State WHEN 500 THEN mState.Diff ELSE 0 END) AS OtherTime
FROM 
    (
        SELECT
            DATEDIFF(MINUTE, ms.[From], ISNULL(ms.[To], GETDATE())) AS Diff,
            ms.State AS State,
            MachineId
        FROM
            MachineStatus ms
        WHERE
            ms.[From] >= @rangeFrom AND
            (ms.[To] <= @rangeEnd OR ms.[To] IS NULL)
    ) as mState

    INNER JOIN Machines m ON m.MachineId = mState.MachineId
GROUP BY
    mState.MachineId,
    m.Alias,
    mState.State

计算时间并按机器对结果进行分组是可行的，但我不知道如何减少结果集，每台机器只包含一行，但每个状态包含一列。

Answer 1

我在您的子查询中开始时没有对您的计算数据应用任何总和：

    SELECT m.MachineId,
           m.Alias,
           Minutes,
           s.State
    FROM machines m
         INNER JOIN states s ON m.MachineId = s.MachineId

然后你可以pivot()为[State]计算每个州的sum()，形式如下：

WITH Calc AS
(
    SELECT m.MachineId,
           m.Alias,
           Minutes,
           s.State
    FROM machines m
         INNER JOIN states s ON m.MachineId = s.MachineId
)
SELECT MachineId, Alias, [State1], [State2], [State500]
FROM
(SELECT MachineId, Alias, State, Minutes FROM Calc) AS SourceTable
PIVOT
(
    SUM(Minutes) FOR State IN ([State1],[State2],[State500])
) AS PivotTable;

这是结果：

+--------------------------------------+---------+--------+--------+----------+
|               MachineId              |  Alias  | State1 | State2 | State500 |
+--------------------------------------+---------+--------+--------+----------+
| 92649C7B-E962-4EB1-B631-00086EECA98A | Alias 1 | 100    | 100    | 100      |
+--------------------------------------+---------+--------+--------+----------+
| A2649C7B-E962-4EB1-B631-00086EECA98A | Alias 2 | 10     | 20     | 70       |
+--------------------------------------+---------+--------+--------+----------+

请注意，您必须知道您的数据有多少个状态return。

可以在这里查看：http://rextester.com/DHDX77489

T-SQL 枢轴 table

T-SQL Pivot table

tsql

sql-server

pivot