数据透视/逆透视计数

Question

我需要一些有关 UnPivot/Pivot/Cross-Apply 类型操作的帮助。

背景：数据是每个数据库中每个 table 一天两次的行数：早上和晚上。

示例数据：

CREATE TABLE [dbo].[DataCount](
    [DatabaseName] [varchar](255) NULL,
    [TableName] [varchar](255) NULL,
    [RowCount] [bigint] NULL,
    [Date] [date] NULL,
    [DateForAnalysis] [datetime] NULL,
    [Runtime] [varchar](50) NULL
) 



INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableA', 10, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 09:00:00.000' AS DateTime), N'Morning')
GO
INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableB', 10, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 09:00:00.000' AS DateTime), N'Morning')
GO
INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableC', 10, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 09:00:00.000' AS DateTime), N'Morning')
GO
INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableA', 15, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 18:00:00.000' AS DateTime), N'Evening')
GO
INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableB', 15, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 18:00:00.000' AS DateTime), N'Evening')
GO
INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableB', 15, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 18:00:00.000' AS DateTime), N'Evening')
GO
INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableA', 20, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 09:00:00.000' AS DateTime), N'Morning')
GO
INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableB', 20, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 09:00:00.000' AS DateTime), N'Morning')
GO
INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableC', 20, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 09:00:00.000' AS DateTime), N'Morning')
GO
INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableA', 25, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 18:00:00.000' AS DateTime), N'Evening')
GO
INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableB', 25, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 18:00:00.000' AS DateTime), N'Evening')
GO
INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) VALUES (N'DatabaseA', N'TableC', 25, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 18:00:00.000' AS DateTime), N'Evening')
GO

所需的示例数据输出如下图所示 - 但欢迎就如何更好地呈现数据提出建议。

Answer 1

您可以使用单个值语句进行多行操作

INSERT [dbo].[DataCount] ([DatabaseName], [TableName], [RowCount], [Date], [DateForAnalysis], [Runtime]) 
   VALUES (N'DatabaseA', N'TableA', 10, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 09:00:00.000' AS DateTime), N'Morning'),
          (N'DatabaseA', N'TableB', 10, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 09:00:00.000' AS DateTime), N'Morning'),
          (N'DatabaseA', N'TableC', 10, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 09:00:00.000' AS DateTime), N'Morning'),
          (N'DatabaseA', N'TableA', 15, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 18:00:00.000' AS DateTime), N'Evening'),
          (N'DatabaseA', N'TableB', 15, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 18:00:00.000' AS DateTime), N'Evening'),
          (N'DatabaseA', N'TableB', 15, CAST(N'2019-10-01' AS Date), CAST(N'2019-10-01 18:00:00.000' AS DateTime), N'Evening'),
          (N'DatabaseA', N'TableA', 20, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 09:00:00.000' AS DateTime), N'Morning'),
          (N'DatabaseA', N'TableB', 20, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 09:00:00.000' AS DateTime), N'Morning'),
          (N'DatabaseA', N'TableC', 20, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 09:00:00.000' AS DateTime), N'Morning'),
          (N'DatabaseA', N'TableA', 25, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 18:00:00.000' AS DateTime), N'Evening'),
          (N'DatabaseA', N'TableB', 25, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 18:00:00.000' AS DateTime), N'Evening'),
          (N'DatabaseA', N'TableC', 25, CAST(N'2019-10-02' AS Date), CAST(N'2019-10-02 18:00:00.000' AS DateTime), N'Evening')
GO

Answer 2

你需要一个动态枢轴。

DECLARE @ColNames NVARCHAR(MAX) = ''

SELECT @ColNames = @ColNames + ',' + DateAndRuntime
FROM (SELECT DISTINCT QUOTENAME( CONCAT([Date] , ' ',   [Runtime] )) DateAndRuntime FROM [dbo].[DataCount] ) T

SELECT @ColNames = STUFF( @ColNames ,1,1,'')

DECLARE @PilotSql NVARCHAR(MAX) = 'SELECT * FROM (
    SELECT [DatabaseName], [TableName], [RowCount], CONCAT([Date] , '' '', [Runtime] ) AS [DateAndRuntime] FROM [dbo].[DataCount] ) 
AS  SRC
PIVOT (SUM([RowCount]) FOR [DateAndRuntime] IN ('+@ColNames+')) PVT '

EXECUTE sp_executesql @PilotSql

结果：

DatabaseName   TableName       2019-10-01 Evening   2019-10-01 Morning   2019-10-02 Evening   2019-10-02 Morning
-------------- --------------- -------------------- -------------------- -------------------- --------------------
DatabaseA      TableA          15                   10                   25                   20
DatabaseA      TableB          30                   10                   25                   20
DatabaseA      TableC          NULL                 10                   25                   20

数据透视/逆透视计数

Data Pivot / Unpivot Count

sql

sql-server

count

pivot

unpivot