I'm trying to do a pivot in SQL Server using a column with a lot of levels. Currently I'm doing a
select '[' + category_1 + '],', count(*)
from myTable
group by '[' + category_1 + '],'
and copying and pasting the output into the text editor.
I was wondering if it's possible to declare a variable and use that as the IN ([],[],[]) information in the pivot command. Has anyone been able to do that?
I think it would look something like this:
#factorLevels = (select '[' + category_1 + '],' from (select category_1, count(*) from myTable) t1)
SELECT *
FROM
(select category_1, item_count from myTable) as t1
PIVOT
(
SUM(item_count)
FOR category_1 IN (#factorLevels)
) as PivotTable;
If I'm understanding you correctly, I think I have a solution. You can use dynamic SQL for this.
You will need to set a variable that contains all the values for the field by which you want to pivot, and then assign the query itself into a variable and execute it to get results:
DECLARE
#values VARCHAR(50),
#querytext VARCHAR(MAX)
;
SET #values = (SELECT '('+(SELECT STUFF(( SELECT ',[' + category_1 +']' FROM
<your table> FOR XML PATH('')), 1, 1, '') )+')')
;
SELECT #querytext =
'SELECT *
FROM (
SELECT
<your fields>
FROM
<your table>
WHERE
<your criteria>
GROUP BY
<your grouping>
) AS Main
PIVOT (
SUM(item_count)
FOR [category_1] IN ' + #values + '
) AS PivotSum'
EXEC sp_executesql #querytext
Related
I have a basic query that looks like this.
SELECT Database_Name,
FilingDate,
SUM(ISNULL([column1], 0) + ISNULL(column2], 0) +
ISNULL([column3], 0) + ISNULL([column4], 0)) AS Total
FROM SomeTable(NOLOCK)
GROUP BY Database_Name,
FilingDate
ORDER BY Database_Name,
FilingDate DESC
This query outputs results that look like this.
I would like to take the dates returned in the FilingDate column and use them as new column headers with the totals for each database and date being used as the row content. The end result should look like this:
My research suggests that a pivot is the best option but I'm struggling to find the right way to execute it as my dates change each day Any assistance would be appreciated.
If this is MS SQL, you can use a dynamic pivot table. Here is a solution using your query (should work, but I don't have the base data to test it).
SELECT Database_Name,
FilingDate,
SUM( ISNULL(column1 ,0) +
ISNULL(column2],0) +
ISNULL([column3],0) +
ISNULL([column4],0)
) AS Total
INTO #T1
FROM SomeTable(NOLOCK)
GROUP BY Database_Name,
FilingDate
DECLARE #PivotColumnHeaders varchar(MAX)
SELECT #PivotColumnHeaders =
COALESCE(
#PivotColumnHeaders + ',[' + CAST(UC.FilingDate AS NVARCHAR(10)) + ']',
'[' + CAST(UC.FilingDate AS NVARCHAR(10)) + ']'
)
FROM (SELECT FilingDate FROM #T1 GROUP BY FilingDate) UC
DECLARE #PQuery varchar(MAX) = '
SELECT * FROM (SELECT Database_Name, FilingDate, Total FROM #T1 T0) T1
PIVOT (SUM([Total]) FOR FilingDate IN (' + #PivotColumnHeaders + ') ) AS P'
EXECUTE (#PQuery)
DROP TABLE #T1
I am working in SQL Server 2014 and below is my database with which I am working on and need some analysis done on it.
Upon inspecting the database sample carefully we can notice a number 8777 in L9 and in L13 column.
Now I want to get only those columns which have 8777 in them and in the end a column named "count" which shows how many times the number appeared means I need in output something like this as shown below:
So far I have written this query which is giving the category and subcategory correct. But it is showing all the columns. I have no idea how to count the occurrences of a number and show its count in a count column.
select *
from Sheet2$
where '8777' IN ([L1],[L2],[L3],[L4],[L5],[L6],[L7],[L8],[L9],[L10],[L11],[L12],[L13]
To dynamically limit the columns, you would need Dynamic SQL
Example
Select *
Into #Temp
From YourTable A
Unpivot ( Value for Item in ([L1], [L2],[ L3], [L4], [L5], [L6], [L7], [L8], [L9], [L10], [L11], [L12], [L13]) ) u
Where Value = 8777
Declare #SQL varchar(max) = Stuff((Select Distinct ',' + QuoteName(Item) From #Temp Order by 1 For XML Path('')),1,1,'')
Select #SQL = '
Select *,[Count] = sum(1) over()
From #Temp A
Pivot (max(Value) For [Item] in (' + #SQL + ') ) p'
Exec(#SQL);
Returns
Category SubCategory L13 L9 Count
C1 SC1 NULL 8777 2
C1 SC3 8777 NULL 2
Hmmm. I think you want the original rows with the count. I think this is:
declare #cols nvarchar(max);
declare #sql nvarchar(max);
set #cols = (select distinct ', ' + v.colname
from t cross apply
(values ('l1', l1),
('l2', l2),
('l3', l3),
('l4', l4),
('l5', l5),
('l6', l6),
('l7', l7),
('l8', l8),
('l9', l9),
('l10', l10),
('l11', l11),
('l12', l12),
('l13', l13)
) v(colname, val)
where v.val = '8777'
for xml path ('')
);
set #sql = '
select category, subcategory' + #cols + ',
count(*) over () as cnt
from t
';
exec sp_executesql #sql;
The only difference from your result is that the count is on every row. That can easily be adjusted using a case expression, but I'm not sure it is necessary.
If you want the count in only one row, then:
set #sql = '
select category, subcategory' + #cols + ',
(case when row_number() over (order by category, subcategory) = 1
then count(*) over ()
end) as cnt
from t
order by category, subcategory
';
You can try this part as a replacement of John's answer's second query to get the proper count, it does not achieve the exact thing you want but can be a work around.
Declare #sql varchar(max) = Stuff((Select Distinct ',' + QuoteName(Item)
From #Temp Order by 1 For XML Path('')),1,1,'')
print #sql;
Select #SQL = '
Select *,value=8777
From #Temp A
Pivot (Count(Value) For [Item] in (' + #sql + ') ) p'
print #sql;
Exec(#SQL);
I just used count function in pivot in place of sum.
I am trying to use POVIT while looking at this example This
Here is the Code:
CREATE DATABASE DEMO
GO
USE DEMO
GO
-- Creating table for demo
IF (object_id('TblPivot','U') > 0)
DROP TABLE TblPivot
CREATE TABLE TblPivot
(
ItemCode int,
ItemName varchar(100),
ItemColour varchar(50)
)
GO
-- Inerting some sample records
INSERT INTO TblPivot
SELECT 1,'Samsung Mobile','Red'
UNION ALL
SELECT 2,'Nokia Mobile','Blue'
UNION ALL
SELECT 3,'Nokia Mobile','Green'
UNION ALL
SELECT 4,'Motorola Mobile','Red'
UNION ALL
SELECT 5,'Samsung Mobile','Green'
UNION ALL
SELECT 2,'Nokia Mobile','Blue'
UNION ALL
SELECT 1,'Samsung Mobile','Red'
UNION ALL
SELECT 2,'Nokia Mobile','Blue'
GO
And here is the PIVOT selection
-- Getting table data
SELECT
ItemCode,
ItemName,
ItemColour
from TblPivot
GO
-- Getting agreegated data using Pivot and converted rows to column
SELECT
*
FROM
(
SELECT
ItemCode,
ItemName,
ItemColour
FROM TblPivot
) AS P
PIVOT
(
Count(ItemName) FOR ItemColour IN (Red, Blue, Green)// Here is the Issue
where it knows what words to give it like Red,Blue,Green what I want is to use
what ever the ItemColur it could be 100s What ever you get
from the data base use that for the `IN(ItemColur)`
) AS pv
GO
It keeps saying that The column name "ItemColur" specified in the PIVOT operator conflicts with the existing column name in the PIVOT argument.
How can I get this working with that I even tried to make a Temp Table didn't work
This is the code for dynamic PIVOT:
DECLARE #columns NVARCHAR(MAX), #sql NVARCHAR(MAX);
SET #columns = N'';
SELECT #columns += N', p.' + QUOTENAME(ItemColour)
FROM (SELECT distinct p.ItemColour FROM dbo.TblPivot AS p
) AS x;
SET #sql = N'
SELECT itemcode, ' + STUFF(#columns, 1, 2, '') + '
FROM
(
SELECT p.itemcode, p.ItemName, p.ItemColour
FROM dbo.TblPivot AS p
) AS j
PIVOT
(
COUNT(ItemName) FOR ItemColour IN ('
+ STUFF(REPLACE(#columns, ', p.[', ',['), 1, 1, '')
+ ')
) AS p;';
PRINT #sql;
EXEC sp_executesql #sql;
The output for above query looks like this:
I have stuck in a select statement, converting rows into columns. I have tried with PIVOT, i was able to convert the single column. But my requirement is little different. I have explained the requirement below.
I have a table structure as below,
I want to select the data as below,
The values in the table are dynamic, which is not a problem for me to deal with that. But i need a way to get the below result.
Could someone please give me a hint on doing it, may be a way to modify the PIVOT below.
select *
from
(
select TSID,AID,Count,BID
from tbl TS
WHERE TS.TPID = 1
) src
pivot
(
sum(Count)
for AID in (AID1,AID2,AID3)
) piv
Thank you..
You may check this fiddle
EDIT
This will work for not previously known column names
DECLARE #Columns AS VARCHAR(MAX)
DECLARE #SQL AS VARCHAR(MAX)
SELECT #Columns = STUFF(( SELECT DISTINCT ',' + AID
FROM Table1
FOR
XML PATH('')
), 1, 1, '')
SET #SQL = '
;WITH MyCTE AS
(
SELECT TSID,
AID,
STUFF(( SELECT '','' + CONVERT(VARCHAR,[Count] )
FROM Table1 I Where I.TSID = O.TSID
FOR
XML PATH('''')
), 1, 1, '''') AS CountList
FROM Table1 O
GROUP BY TSID,
AID
)
SELECT *
FROM MyCTE
PIVOT
(
MAX(CountList)
FOR AID IN
(
' + #Columns + '
)
) AS PivotTable'
EXEC(#SQL)
I'm storing some queries in a table column so I can execute them later passing some parameters.
But it has been really annoying to format the query into an Update sentence, because of the special characters.
For Example:
SELECT * FROM MOUNTAINS WHERE MON_NAME='PALMA' AND MON_DESC LIKE '%TRANVULCANIA%'
Then I need the string just for the udpate query:
UPDATE QUERIES
SET QUE_SEL='SELECT * FROM MOUNTAINS WHERE MON_NAME='''+'PALMA'+''' AND MON_DESC LIKE '''+'%TRANVULCANIA%'+''' '
WHERE QUE_ID=1
as you can see the first ' must be replaced for '''+' but the next door ' must be replaced by '+'''
This is the query I'm working on:
DECLARE #QUERY VARCHAR(MAX)
SELECT #QUERY='SELECT * FROM QUERIES WHERE QUE_NOMBRE='''+'PRUEBA 1'+''' '
SELECT
t.r.value('.', 'varchar(255)') AS token
, ROW_NUMBER() OVER (ORDER BY (SELECT 1)) AS id
FROM (
SELECT myxml = CAST('<t>' + REPLACE(#QUERY, '''', '</t><t>''</t><t>') + '</t>' AS XML)
) p
CROSS APPLY myxml.nodes('/t') t(r)
this is the result:
token id
-------------------------------------------------- --------------------
SELECT * FROM QUERIES WHERE QUE_NOMBRE= 1
' 2
PRUEBA 1 3
' 4
5
Now I want a column that tell me when to open and when to close and then I can set the final replace.
Adapting the solution given by #rivarolle
DECLARE #QUERY VARCHAR(MAX)
DECLARE #FORMATTED varchar(max)
SELECT #QUERY='SELECT * FROM QUERIES WHERE QUE_NOMBRE='''+'PRUEBA 1'+''''
;WITH TOKENS AS(
SELECT
t.r.value('.', 'varchar(MAX)') AS token
, ROW_NUMBER() OVER (ORDER BY (SELECT 1)) AS Id
FROM (
SELECT myxml = CAST('<t>' + REPLACE(#QUERY, '''', '</t><t>''</t><t>') + '</t>' AS XML)
) p
CROSS APPLY myxml.nodes('/t') t(r)
)
,
Tokens2 as (
SELECT
TOKENS.token as token
,quotes.row%2 as tipoapostrofe
from Tokens
left join (select row_number() over( order by Id asc) as row, a.* FROM (SELECT * from Tokens) a where Token = '''') quotes
on quotes.Id = Tokens.Id
)
SELECT #FORMATTED = STUFF((
SELECT ' ' + REPLACE(token,'''',CASE tipoapostrofe WHEN 1 THEN '''''''+''' WHEN 0 THEN '''+''''''' ELSE '' END) AS [text()]
FROM Tokens2
FOR XML PATH('')
), 1, 1, '')
print #FORMATTED
This Works, just need a function for cleaning XML special characters and another for putting back, and the Dynamic queries are printed ready for an update.
I think its not necessary to replace an apostrophe with '''+' to open and '+''' to close, I made some probes and you can exec a query that you replace opening and closing apostrophes with the same.. for example '''+' for open and '''+' for close.
So the query would be:
DECLARE #QUERY VARCHAR(MAX)
DECLARE #FORMATTED varchar(max)
SELECT #QUERY='SELECT * FROM QUERIES WHERE QUE_NOMBRE='''+'PRUEBA 1'+''''
SELECT #FORMATTED= STUFF((
SELECT ' ' +
(SELECT
CASE
WHEN t.r.value('.', 'varchar(250)')='''' THEN REPLACE(t.r.value('.', 'varchar(250)'), '''','''''''+''')
ELSE t.r.value('.', 'varchar(250)')
END
) AS [text()]
-- , ROW_NUMBER() OVER (ORDER BY (SELECT 1)) AS id
FROM (
SELECT myxml = CAST('<t>' + REPLACE(#QUERY, '''', '</t><t>''</t><t>') + '</t>' AS XML)
) p
CROSS APPLY myxml.nodes('/t') t(r)
FOR XML PATH('')
), 1, 1, '')
SET #FORMATTED=REPLACE(#FORMATTED,' ','')
PRINT #FORMATTED
then I get:
SELECT * FROM QUERIES WHERE QUE_NOMBRE= '''+' PRUEBA 1 '''+'
then I copy into a variable and execute
DECLARE #VAR VARCHAR(500)
SET #VAR='SELECT * FROM QUERIES WHERE QUE_NOMBRE='''+'PRUEBA 1'''+' '
EXEC(#VAR)
It Works for very simple queries, but with longer and complicated queries it doesn't works..
Assuming your token table is Tokens(Token, Id, Position):
update Tokens
set position = quotes.row%2
from Tokens
left join (select row_number() over( order by Id asc) as row, a.* FROM (SELECT * from Tokens) a where Token = '''') quotes
on quotes.Id = Tokens.Id
The position column will have a value of 1 for starting quote and 0 for closing quote. NULL for the rest.