Question

我有一个查询，该查询返回通道名称存在于另一个表中的数据，因此我只返回一组特定的通道。

SELECT DISTINCT 
    ReadingDate, SerialNumber, ChannelName, ChannelValue
FROM 
    [Staging].[UriData]
WHERE 
    ChannelName IN (SELECT ChannelName FROM [dbo].[Channels])
ORDER BY 
    ReadingDate DESC, ChannelName

此查询的重要性在于它只会带回存储移动数据而不是固定诊断数据的通道。

结果（摘要）：

ReadingDate         | SerialNumber | ChannelName |  ChannelValue
2018-09-09 20:30:00     2209            m1            461
2018-09-09 20:30:00     2209            m2            0
2018-09-09 20:30:00     2209            m3            50
2018-09-09 20:30:00     2209            m4            15631
2018-09-09 20:30:00     2209            m5            1
2018-09-09 20:30:00     2209            m6            8150
2018-09-09 20:30:00     2209            m7            0
2018-09-09 20:30:00     2209            m8            2790
2018-09-09 20:30:00     2209            m9            0
2018-09-09 20:15:00     2209            m1            2452
2018-09-09 20:15:00     2209            m2            0
2018-09-09 20:15:00     2209            m3            50
2018-09-09 20:15:00     2209            m4            15629
2018-09-09 20:15:00     2209            m5            1
2018-09-09 20:15:00     2209            m6            8100
2018-09-09 20:15:00     2209            m7            0
2018-09-09 20:15:00     2209            m8            2780

然后我想将此数据透视成列，以便按日期（日期）将它们分组，然后从该日期开始计算时间。

必填输出：

DATE        | SERIAL | ChannelName |   00:15 |  00:30   | ..... | 23:45
2018-09-06    2209         m1          Value    Value    .....    Value
2018-09-06    2209         m2          Value    Value    .....    Value

我一直在研究数据透视，但是我没有获得所需的格式的数据。

Answer 1

您可以尝试使用 cte递归制作时间日历表，然后通过time制作行号。

然后将动态SQL与数据透视一起使用。

这是给您的样品。

DECLARE @cols AS NVARCHAR(MAX),
        @query  AS NVARCHAR(MAX);


;WITH CTE AS (
    SELECT CAST('00:00' AS TIME) startDt, CAST('23:45' AS TIME) endDt
    UNION ALL 
    SELECT DATEADD(MINUTE, 15, startDt),endDt
    FROM CTE
    WHERE DATEADD(MINUTE, 15, startDt) <endDt
), TimeTable AS (
    select *,ROW_NUMBER() OVER (ORDER BY startDt) rn 
    FROM (
       SELECT  startDt,endDt
       FROM CTE
       UNION ALL 
       SELECT CAST('23:45' AS TIME) startDt, CAST('23:45' AS TIME) endDt
    ) t1
)



select @cols = CONCAT(@cols,'MAX(CASE WHEN '''+CAST(startDt AS VARCHAR(5))+''' = CAST(ReadingDate AS TIME) THEN ChannelValue ELSE 0 end) AS ',QUOTENAME(CAST(startDt AS VARCHAR(5))),', ')
from TimeTable
WHERE startDt <= endDt
ORDER BY rn 

SET @cols = left(@cols, len(@cols) - 1)


set @query = '
;WITH CTE AS ( SELECT CAST('''+ cast('00:00' as varchar(5))+''' AS TIME) startDt, CAST('''+ cast('23:45' as varchar(5))+''' AS TIME) endDt
    UNION ALL 
    SELECT DATEADD(MINUTE, 15, startDt),endDt
    FROM CTE
    WHERE DATEADD(MINUTE, 15, startDt) <endDt
), TimeTable AS (
    select *,ROW_NUMBER() OVER (ORDER BY startDt) rn 
    FROM (
       SELECT  startDt,endDt
       FROM CTE
       UNION ALL 
       SELECT CAST('''+ cast('23:45' as varchar(5))+''' AS TIME) startDt, CAST('''+ cast('23:45' as varchar(5))+''' AS TIME) endDt
    ) t1
)

SELECT CONVERT(VARCHAR(10),ReadingDate, 126),
      SerialNumber,
      ChannelName,
       '+ @cols +'
FROM  T
GROUP BY CONVERT(VARCHAR(10),ReadingDate, 126) ,SerialNumber,ChannelName
'

exec(@query)

sqlfiddle

Answer 2

您还可以使用PIVOT构造实现相同的结果，如下所示：

SELECT [Serial], [Channel], [ReadingDate], [00:15], [00:30], ....
FROM(
    SELECT 
        SerialNumber AS [Serial],
        ChannelName AS [Channel],
        CAST(ReadingDate AS DATE) AS [ReadingDate],
        CAST(ReadingDate AS TIME(0)) AS [ReadingTime],
        ChannelValue AS [Value]
    FROM [Staging].[UriData]
    WHERE
        ChannelName IN (SELECT ChannelName FROM Staging.ActiveChannels )
        AND Processed = 0
    ) AS [Raw]
PIVOT
(
    MAX( [Value] ) FOR [ReadingTime] IN( [00:15], [00:30], .... )
) AS pvt
ORDER BY ReadingDate DESC, Channel, [Serial]

要生成“时间”列（并使您不必手动键入），请运行以下查询：

DECLARE @cols AS NVARCHAR(MAX)
;WITH Times AS (
    SELECT CAST( '00:00' AS TIME) AS [Time]
    UNION ALL 
    SELECT DATEADD( MINUTE, 15, [Time] )
    FROM Times
    WHERE [Time] < CAST( '23:45' AS TIME )
)
--SELECT * FROM Times
SELECT @cols = CONCAT( @cols, QUOTENAME( CAST( [Time] AS VARCHAR( 5 ) )), ', ' )
FROM Times

SET @cols = LEFT( @cols, LEN( @cols ) - 1 )
PRINT @cols

将结果复制/粘贴到上面的“ PIVOT”查询中。

按日期，名称和值分类的PIVOT SQL表

2 个答案: