问题
I have posted a pivot question before and it was answered. However, this particular case requires dates as the headers and the dates will change (and be unknown) each time the pivot is generated. I believe I need dynamic pivot/unpivot based on a few samples but I cannot figure the syntax out.
Here is the table:
CREATE TABLE [dbo].[PhaseFlowChart](
[pfckey] [int] NULL,
[hourlykey] [bigint] NULL,
[daykey] [bigint] NULL,
[weekkey] [int] NULL,
[monthkey] [int] NULL,
[bbkey] [int] NULL,
[Day] [varchar](100) NULL,
[Date] [varchar](100) NULL,
[Bull Bear Gap] [varchar](100) NULL,
[Monthly] [varchar](100) NULL,
[Weekly] [varchar](100) NULL,
[Daily] [varchar](100) NULL,
[Hour 1] [varchar](100) NULL,
[Hour 2] [varchar](100) NULL,
[Hour 3] [varchar](100) NULL,
[Hour 4] [varchar](100) NULL,
[Hour 5] [varchar](100) NULL,
[Hour 6] [varchar](100) NULL,
[Hour 7] [varchar](100) NULL
) ON [PRIMARY]
I do not need any of the 'key' columns in the output. Therefore here is a simple select on the table:
select [DAY],[Date],[Bull Bear Gap],[Monthly],[Weekly],[Daily],[Hour 1],[Hour 2],[Hour 3],[Hour 4],[Hour 5],[Hour 6],[Hour 7]
from PhaseFlowChart
order by pfckey asc
Here is the output from the above:
I want the output (dynamic pivot/unpivot?) to look like this (and again, the dates will be different each time this is generated which is daily):
Finally, I found this sql on the web but after days of manipulating the syntax I cannot figure out how to port it to meet my needs.
CREATE TABLE #yt
([ID] int, [expense] int, [revenue] int, [date] datetime)
;
INSERT INTO #yt
([ID], [expense], [revenue], [date])
VALUES
(1, 43, 45, '2012-12-31 00:00:00'),
(2, 32, 32, '2013-01-01 00:00:00'),
(3, 64, 56, '2013-01-31 00:00:00'),
(4, 31, 32, '2013-03-03 00:00:00')
DECLARE @cols AS NVARCHAR(MAX),
@query AS NVARCHAR(MAX)
select @cols = STUFF((SELECT ',' + QUOTENAME(c.col+'_'+convert(varchar(10), #yt.date, 110))
from #yt
cross apply
(
select 'expense' col union all
select 'revenue'
) c
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)')
,1,1,'')
set @query = 'SELECT id,' + @cols + '
from
(
select id,
col+''_''+convert(varchar(10), date, 110) new_col,
value
from #yt
unpivot
(
value
for col in (expense, revenue)
) un
) src
pivot
(
sum(value)
for new_col in (' + @cols + ')
) p '
execute(@query);
Any/all help is greatly appreciated.
回答1:
You first need to UNPIVOT
our columns and use DENSE_RANK()
to give you something to pivot on later. Put this result into a temp table so that you can get a comma seperated list of the DESNSE_RANK
columns. Then create a UNION
so that Day
and Date
are part of the same column as what was unpivoted. This can be put in a global temp table for use in the dynamic SQL. Create a variable that will store the list of columns and build the dynamic SQL and execute it.
A full example (obviously don't go dropping the PhaseFlowChart
table)
-- pre-cleanup
IF OBJECT_ID('[dbo].[PhaseFlowChart]') IS NOT NULL
DROP TABLE [dbo].[PhaseFlowChart]
GO
IF OBJECT_ID('tempdb..#tmp') IS NOT NULL
DROP TABLE #tmp
GO
IF OBJECT_ID('tempdb..##tmp') IS NOT NULL
DROP TABLE ##tmp
GO
-- setup table and data
CREATE TABLE [dbo].[PhaseFlowChart](
[pfckey] [int] NULL,
[hourlykey] [bigint] NULL,
[daykey] [bigint] NULL,
[weekkey] [int] NULL,
[monthkey] [int] NULL,
[bbkey] [int] NULL,
[Day] [varchar](100) NULL,
[Date] [varchar](100) NULL,
[Bull Bear Gap] [varchar](100) NULL,
[Monthly] [varchar](100) NULL,
[Weekly] [varchar](100) NULL,
[Daily] [varchar](100) NULL,
[Hour 1] [varchar](100) NULL,
[Hour 2] [varchar](100) NULL,
[Hour 3] [varchar](100) NULL,
[Hour 4] [varchar](100) NULL,
[Hour 5] [varchar](100) NULL,
[Hour 6] [varchar](100) NULL,
[Hour 7] [varchar](100) NULL
) ON [PRIMARY]
INSERT INTO [dbo].PhaseFlowChart
([Day], [Date], [Bull Bear Gap], Monthly, Weekly, Daily,
[Hour 1], [Hour 2], [Hour 3], [Hour 4], [Hour 5], [Hour 6], [Hour 7])
VALUES
('MON', '20130101', 'P1', 'P1', 'P1', 'P1', 'P1', 'P1', 'P1', 'P1', 'P1', 'P1', 'P1'),
('TUE', '20130102', 'P2', 'P2', 'P2', 'P2', 'P2', 'P2', 'P2', 'P2', 'P2', 'P2', 'P2'),
('WED', '20130103', 'P3', 'P3', 'P3', 'P3', 'P3', 'P3', 'P3', 'P3', 'P3', 'P3', 'P3'),
('THU', '20130104', 'P4', 'P4', 'P4', 'P4', 'P4', 'P4', 'P4', 'P4', 'P4', 'P4', 'P4'),
('FRI', '20130105', 'P5', 'P5', 'P5', 'P5', 'P5', 'P5', 'P5', 'P5', 'P5', 'P5', 'P5'),
('SAT', '20130106', 'P6', 'P6', 'P6', 'P6', 'P6', 'P6', 'P6', 'P6', 'P6', 'P6', 'P6'),
('SUN', '20130107', 'P7', 'P7', 'P7', 'P7', 'P7', 'P7', 'P7', 'P7', 'P7', 'P7', 'P7')
GO
-- unpivot the columns into 'categories'
SELECT [Day], [Date], [Value], [Category],
DENSE_RANK() OVER (ORDER BY CAST([Date] AS DATE)) dr
INTO #tmp
FROM PhaseFlowChart pfc
UNPIVOT (
Value FOR Category IN ([Bull Bear Gap], Monthly, Weekly, Daily,
[Hour 1], [Hour 2], [Hour 3], [Hour 4],
[Hour 5], [Hour 6], [Hour 7])
) upiv
-- create a global temp table for use later
SELECT *
INTO ##tmp
FROM (
-- union data into single category column
SELECT 'Day' Category, [Day] Value, dr, 1 o FROM #tmp
UNION ALL
SELECT 'Date' Category, [Date] Value, dr, 2 o FROM #tmp
UNION ALL
SELECT [Category], Value, dr, 3 o FROM #tmp
) t
-- get a comma seperated list of columns for the PIVOT
DECLARE @cols VARCHAR(MAX) =
STUFF(CAST((SELECT ',' + QUOTENAME(dr)
FROM (
SELECT DISTINCT dr
FROM #tmp
) t
ORDER BY dr
FOR XML PATH(''), TYPE
) AS VARCHAR(MAX)),1,1,'')
-- create and execute the sql
DECLARE @sql VARCHAR(MAX) = '
SELECT Category, ' + @cols + '
FROM ##tmp
PIVOT (
MAX([Value])
FOR dr IN (' + @cols + ')
) piv
ORDER BY o, CASE Category
WHEN ''Daily'' THEN 4
WHEN ''Weekly'' THEN 3
WHEN ''Monthly'' THEN 2
WHEN ''Bull Bear Gap'' THEN 1
ELSE 5 END, Category
'
EXEC(@sql)
来源:https://stackoverflow.com/questions/19614856/sql-server-2008-r2-dynamic-pivot-unpivot-with-moving-dates