I'm currently making a program to store data in database sql server.
and the data is a cost that divided into max 5 column cost details.
As for the first row it labeled with 1 and second with 2, up to 5 row with same id.
So i'm making the table like the picture below. And Now i want to select the table from row to column like in the picture.
i'm making the query like this. I get the result like i want, but the problem is that the query cost like 5/more times the execution plan by just selecting the same table.
My question is, is there a way to make the result like in the picture just by selecting once the table or is there any other way to make like the result with better performance, i hear of using pivot for transposing row to column, but in my case i don't know how to do it. Thanks for reply
this is the execution plan https://www.brentozar.com/pastetheplan/?id=SJGJdaCB7
AND this is the query i used
select * FROM TblKecelakaanBiaya;
with tbl AS (
select *, ROW_NUMBER() OVER(PARTITION by id_kasus ORDER BY id_kasus) rn FROM TblKecelakaanBiaya
)
SELECT A.id_kasus, A.ket_biaya1, A.jlh_biaya1, C.ket_biaya2, C.jlh_biaya2,
D.ket_biaya3, D.jlh_biaya3, E.ket_biaya4, E.jlh_biaya4, F.ket_biaya5, F.jlh_biaya5
FROM (SELECT id_kasus, ket_biaya ket_biaya1, jlh_biaya jlh_biaya1 FROM tbl WHERE rn = 1) A
LEFT JOIN (SELECT id_kasus, ket_biaya ket_biaya2, jlh_biaya jlh_biaya2 FROM tbl WHERE rn = 2) C ON A.id_kasus = C.id_kasus
LEFT JOIN (SELECT id_kasus, ket_biaya ket_biaya3, jlh_biaya jlh_biaya3 FROM tbl WHERE rn = 3) D ON A.id_kasus = D.id_kasus
LEFT JOIN (SELECT id_kasus, ket_biaya ket_biaya4, jlh_biaya jlh_biaya4 FROM tbl WHERE rn = 4) E ON A.id_kasus = E.id_kasus
LEFT JOIN (SELECT id_kasus, ket_biaya ket_biaya5, jlh_biaya jlh_biaya5 FROM tbl WHERE rn = 5) F ON A.id_kasus = F.id_kasus
The execution plan for query 1 is just 1% and the other one is 99%.
Perhaps a Pivot inconcert with a Cross Apply.
Example
Select *
From (
Select id_kasus,item,value
From (Select * ,RN = Row_Number() over (Partition By id_kasus Order By id_kasus) From TblKecekakaanBiaya ) A
Cross Apply (values (concat('jih_biaya',RN),convert(varchar(150),jih_biaya))
,(concat('ket_biaya',RN),ket_biaya)
) b(item,value)
) src
Pivot (max(value) for item in ([ket_biaya1],[jih_biaya1],[ket_biaya2],[jih_biaya2],[ket_biaya3],[jih_biaya3],[ket_biaya4],[jih_biaya4],[ket_biaya5],[jih_biaya5]) ) pvt
Returns
Related
So I've done some looking around and wasn't unable to find quite what I was looking for. I have two tables.
1.) Table where general user information is stored
2.) Where a status is generated and stored.
The problem is, is that there are multiple rows for the same users and querying these results in multiple returns. I can't just merge them because they aren't all the same status. I need just the newest status from that table.
Example of the table:
SELECT DISTINCT
TOP(50) cam.UserID AS PatientID,
mppi.DisplayName AS Surgeon,
ISNULL(sci.IOPStatus, 'N/A') AS Status,
tkstat.TrackerStatusID AS Stat_2
FROM
Main AS cam
INNER JOIN
Providers AS rap
ON cam.VisitID = rap.VisitID
INNER JOIN
ProviderInfo AS mppi
ON rap.UnvUserID = mppi.UnvUserID
LEFT OUTER JOIN
Inop AS sci
ON cam.CwsID = sci.CwsID
LEFT OUTER JOIN
TrackerStatus AS tkstat
ON cam.CwsID = tkstat.CwsID
WHERE
(
cam.Location_ID IN
(
'SURG'
)
)
AND
(
rap.IsAttending = 'Y'
)
AND
(
cam.DateTime BETWEEN CONCAT(CAST(GETDATE() AS DATE), ' 00:00:00') AND CONCAT(CAST(GETDATE() AS DATE), ' 23:59:59')
)
AND
(
cam.Status_StatusID != 'Cancelled'
)
ORDER BY
cam.UserID ASC
So I need to grab only the newest Stat_2 from each ID so they aren't returning multiple rows. Each Stat_2 also has an update time meaning I can sort by the time/date that column is : StatusDateTime
One way to handle this is to create a calculated row_number for the table where you need the newest record.
Easiest way to do that is to change your TKSTAT join to a derived table with the row_number calculation and then add a constraint to your join where the RN =1
SELECT DISTINCT TOP (50)
cam.UserID AS PatientID, mppi.DisplayName AS Surgeon, ISNULL(sci.IOPStatus, 'N/A') AS Status, tkstat.TrackerStatusID AS Stat_2
FROM Main AS cam
INNER JOIN Providers AS rap ON cam.VisitID = rap.VisitID
INNER JOIN ProviderInfo AS mppi ON rap.UnvUserID = mppi.UnvUserID
LEFT OUTER JOIN Inop AS sci ON cam.CwsID = sci.CwsID
LEFT OUTER JOIN (SELECT tk.CwsID, tk.TrackerStatusId, ROW_NUMBER() OVER (PARTITION BY tk.cwsId ORDER BY tk.CreationDate DESC) AS rn FROM TrackerStatus tk)AS tkstat ON cam.CwsID = tkstat.CwsID
AND tkstat.rn = 1
WHERE (cam.Location_ID IN ( 'SURG' )) AND (rap.IsAttending = 'Y')
AND (cam.DateTime BETWEEN CONCAT(CAST(GETDATE() AS DATE), ' 00:00:00') AND CONCAT(CAST(GETDATE() AS DATE), ' 23:59:59'))
AND (cam.Status_StatusID != 'Cancelled')
ORDER BY cam.UserID ASC;
Note you need a way to derive what the "newest" status is; I assume there is a created_date or something; you'll need to enter the correct colum name
ROW_NUMBER() OVER (PARTITION BY tk.cwsId ORDER BY tk.CreationDate DESC) AS rn
SQL Server doesn't offer a FIRST function, but you can reproduce the functionality with ROW_NUMBER() like this:
With Qry1 (
Select <other columns>,
ROW_NUMBER() OVER(
PARTITION BY <group by columns>
ORDER BY <time stamp column*> DESC
) As Seq
From <the rest of your select statement>
)
Select *
From Qry1
Where Seq = 1
* for the "newest" record.
I am trying to join two tables to the following query:
SELECT "NUMBER",
"U_ANALYZED_DATE",
"DV_SALES_ACCOUNT",
"U_USD_TOTAL_POTENTIAL_NNACV"
FROM
(select *, row_number() over ( partition by "DV_SALES_ACCOUNT" order by "U_ANALYZED_DATE" desc ) rownum
from "SURF_RT"."SALES_REQUEST")
WHERE rownum = 1
AND "DV_SALES_CATEGORY" = 'Compliance'
AND "DV_STATE" NOT IN ('Closed Canceled')
AND (YEAR("U_ANALYZED_DATE") = '2019' AND MONTH("U_ANALYZED_DATE") IN ('10','11','12')
OR YEAR("U_ANALYZED_DATE") = '2020' AND MONTH("U_ANALYZED_DATE") IN ('1','2','3'))
AND "U_USD_TOTAL_POTENTIAL_NNACV" > 0
ORDER BY "U_ANALYZED_DATE" desc
The tables should be joined as follows:
JOIN "SURF_RT"."SALES_ACCOUNT" on "SURF_RT"."SALES_ACCOUNT"."NAME" = "SURF_RT"."SALES_REQUEST"."DV_SALES_ACCOUNT"
JOIN "SURF_RT"."SALES_CONTRACT" on "SURF_RT"."SALES_CONTRACT"."DV_ACCOUNT" = "SURF_RT"."SALES_REQUEST"."DV_SALES_ACCOUNT"
I am getting an error no matter what I try and it has to be because of the partition. Does anyone know the solution here?
I suspect that you just need to alias the derived table so you can then refer to join it. In many databases this is mandatory anyway, but apparently not in Hana (otherwise your original query would not run).
But to join on a derived table (the resultset that is generated by the subquery), alias do help:
SELECT
...
FROM
(
select
*,
row_number() over ( partition by "DV_SALES_ACCOUNT" order by "U_ANALYZED_DATE" desc ) rownum
from "SURF_RT"."SALES_REQUEST"
) sr -- table alias
JOIN "SURF_RT"."SALES_ACCOUNT"
ON "SURF_RT"."SALES_ACCOUNT"."NAME" = "SURF_RT"."SALES_REQUEST"."DV_SALES_ACCOUNT"
JOIN "SURF_RT"."SALES_CONTRACT"
ON "SURF_RT"."SALES_CONTRACT"."DV_ACCOUNT" = sr."DV_SALES_ACCOUNT" --reference to the derived table
WHERE
...
Side notes:
you should use table aliases for other tables involved in the query too, in order to make the code more readable
you should also prefix each column in the query with the identifier of the table it belongs to, so the query is unambiguous (and easier to maintain)
The problem with this query is not "that the where rownum = 1 needs to be at the end" but that the OP got confused by the bracketing of SQL expression.
More specifically, trying to reference the sub-query data by specifying join conditions against the base table that is used in the sub-query:
JOIN "SURF_RT"."SALES_ACCOUNT"
on "SURF_RT"."SALES_ACCOUNT"."NAME" = "SURF_RT"."SALES_REQUEST"."DV_SALES_ACCOUNT"
JOIN "SURF_RT"."SALES_CONTRACT"
on "SURF_RT"."SALES_CONTRACT"."DV_ACCOUNT" = "SURF_RT"."SALES_REQUEST"."DV_SALES_ACCOUNT"
Since the sub-query (derived table) is used in the query and should be used for the join, it needs to be referred in the join condition, instead.
So, yes, it needs a table alias here and the join conditions need to refer to it.
SELECT
...
FROM
(select *
, row_number() over
(partition by "DV_SALES_ACCOUNT"
order by "U_ANALYZED_DATE" desc) rownum
from
"SURF_RT"."SALES_REQUEST") sr
INNER JOIN "SURF_RT"."SALES_ACCOUNT" sa
on sa."NAME" = sr."DV_SALES_ACCOUNT"
INNER JOIN "SURF_RT"."SALES_CONTRACT" sc
on sc."DV_ACCOUNT" = sr."DV_SALES_ACCOUNT"
WHERE
sr.rownum = 1
AND "DV_SALES_CATEGORY" = 'Compliance'
AND "DV_STATE" NOT IN ('Closed Canceled')
AND (YEAR("U_ANALYZED_DATE") = '2019'
AND MONTH("U_ANALYZED_DATE") IN ('10','11','12')
OR YEAR("U_ANALYZED_DATE") = '2020'
AND MONTH("U_ANALYZED_DATE") IN ('1','2','3'))
AND "U_USD_TOTAL_POTENTIAL_NNACV" > 0
ORDER BY
"U_ANALYZED_DATE" desc;
With just this little bit of standard SQL syntax and formatting of code the query got a lot easier to understand.
Now it's even obvious that the IN conditions for MONTH and YEAR should in fact be integers, not strings as these functions return integers.
SELECT
...
FROM
(SELECT *
, row_number() over
(partition by "DV_SALES_ACCOUNT"
order by "U_ANALYZED_DATE" desc) rownum
FROM
"SURF_RT"."SALES_REQUEST") sr
INNER JOIN "SURF_RT"."SALES_ACCOUNT" sa
on sa."NAME" = sr."DV_SALES_ACCOUNT"
INNER JOIN "SURF_RT"."SALES_CONTRACT" sc
on sc."DV_ACCOUNT" = sr."DV_SALES_ACCOUNT"
WHERE
sr.rownum = 1
AND "DV_SALES_CATEGORY" = 'Compliance'
AND "DV_STATE" NOT IN ('Closed Canceled')
AND ( YEAR("U_ANALYZED_DATE") = 2019
AND MONTH("U_ANALYZED_DATE") IN (10, 11, 12)
OR YEAR("U_ANALYZED_DATE") = '2020'
AND MONTH("U_ANALYZED_DATE") IN (1 , 2 , 3 )
)
AND "U_USD_TOTAL_POTENTIAL_NNACV" > 0
ORDER BY
"U_ANALYZED_DATE" DESC
My system doesn't allow CTE in coding and the first word has to always be SELECT.
Could someone explain (ideally in a Oracle SQL for dummies format) how I would go about changing code written with CTE into INLINE code?
As an example of CTE code:
WITH table1
AS (SELECT enquiry_status_log.enquiry_number,
enquiry_status_log.enquiry_log_number,
follow_up.follow_up_name,
enquiry_status_log.follow_up_date,
RANK ()
OVER (PARTITION BY enquiry_status_log.enquiry_number
ORDER BY enquiry_status_log.enquiry_log_number DESC)
rn
FROM enquiry_status_log
JOIN follow_up
ON enquiry_status_log.follow_up_code =
follow_up.follow_up_code)
SELECT enquiry_number,
enquiry_log_number,
follow_up_name,
follow_up_date
FROM table1
WHERE rn = 1
ORDER BY enquiry_number, enquiry_log_number
How would I write this (please, VERY easy to understand) so that SELECT would be the first word in the code and CTE wasn't used?
Your CTE is used only once in your query, so you can simply turn it to a subquery:
SELECT enquiry_number,
enquiry_log_number,
follow_up_name,
follow_up_date
FROM (
SELECT
e.enquiry_number,
e.enquiry_log_number,
f.follow_up_name,
e.follow_up_date,
RANK () OVER (
PARTITION BY e.enquiry_number
ORDER BY e.enquiry_log_number DESC
) rn
FROM enquiry_status_log e
JOIN follow_up f ON e.follow_up_code = f.follow_up_code
) table1
WHERE rn = 1
ORDER BY enquiry_number, enquiry_log_number
Note: table aliases make the query shorter and easier to read. I modified you query to use them.
Also, it is possible that your query could be rewritten to use a correlated subquery for filtering instead of rank(). In Oracle, this could lead to better performance:
SELECT
e.enquiry_number,
e.enquiry_log_number,
f.follow_up_name,
e.follow_up_date
FROM enquiry_status_log e
JOIN follow_up f ON e.follow_up_code = f.follow_up_code
WHERE e.enquiry_log_number = (
SELECT MAX(enquiry_log_number)
FROM enquiry_status_log e1
WHERE e.enquiry_number = e1.enquiry_number
)
ORDER BY e.enquiry_number, e.enquiry_log_number
When I made this select it shows as result all “rut” with it phone numbers, my problem
is that each “rut” has almost 10 phone numbers, and I need just 3 phone numbers for
each “rut”, I tryied using TOP, but just shows the first 3 rows of all table and not the
first rows by “rut”, how can I use TOP just for the row “rut” and not in the all the
table
Select Distinct
t1.rut_cliente as rut_cliente,
t1.nro_fono as numero
--Into #tmp_numeros
From dat_clientes_sucursales_contactos_telefonos t1,dat_rut_clientes t2
where t1.rut_cliente = t2.rut_cliente
and cod_prioridad = 1
this is what I get with this query:
Rut_cliente Nro_fono
60506000-5 2046840
60506000-5 3507935
60506000-5 4106886
60506000-5 5440000
60506000-5 5445000
81698900-0 2373281
81698900-0 3541342
81698900-0 3541438
81698900-0 3541518
81698900-0 3542101
and this is what I want:
Rut_cliente Nro_fono
60506000-5 2046840
60506000-5 3507935
60506000-5 4106886
81698900-0 2373281
81698900-0 3541342
81698900-0 3541438
thanks in advance.
The question was originally tagged SQL Server 2008, where you can do this using a common table expression:
;WITH x AS
(
SELECT Rut_cliente, Nro_fono, rn = ROW_NUMBER()
OVER (PARTITION BY Rut_cliente ORDER BY Nro_fono)
FROM dbo.dat_clientes_sucursales_contactos_telefonos AS t1
INNER JOIN dbo.dat_rut_clientes AS t2
ON t1.rut_cliente = t2.rut_cliente
WHERE cod_prioridad = 1
)
SELECT Rut_cliente, Nro_fono FROM x
WHERE rn <= 3
ORDER BY Rut_cliente, Nro_fono;
Other comments:
Please don't use table, table syntax. Use proper INNER JOINs. I explain why here.
Please add proper aliases to the inner query, so people know which columns come from t1 and which columns come from t2.
But now we learn the user is actually using SQL Server 2000. This is how I would do it there, I think, but performance is going to be horrible. I'm not 100% sure this works (because again I'm making guesses about which columns come from which table).
SELECT x.rut_cliente, x.nro_fono, COUNT(*) FROM
(
SELECT t1.rut_cliente, t1.nro_fono
FROM dat_clientes_sucursales_contactos_telefonos AS t1
INNER JOIN dat_rut_clientes AS t2
ON t1.rut_cliente = t2.rut_cliente
WHERE cod_prioridad = 1
) AS x
INNER JOIN dat_clientes_sucursales_contactos_telefonos AS b
ON b.rut_cliente = x.rut_cliente
AND b.nro_fono <= x.nro_fono
GROUP BY x.rut_cliente, x.nro_fono
HAVING COUNT(*) <= 3
ORDER BY x.rut_cliente, x.Nro_fono;
For example with ROW_NUMBER which is a window function and returns a row-number for each partition(similar to group by) determined by the order by.:
WITH CTE AS(
SELECT t1.rut_cliente as rut_cliente, t1.nro_fono as numero,
RN = ROW_NUMBER()OVER(PARTITION BY Rut_cliente ORDER BY Nro_fono)
FROM dbo.dat_clientes_sucursales_contactos_telefonos AS t1
INNER JOIN dbo.dat_rut_clientes AS t2 ON t1.rut_cliente = t2.rut_cliente
WHERE cod_prioridad = 1
)
SELECT rut_cliente as rut_cliente, nro_fono as numero
FROM CTE
WHERE RN <= 3
Try this
;with CTE AS (
select Rut_cliente, Nro_fono , ROW_NUMBER() OVER
(PARTITION BY Rut_cliente ORDER BY Nro_fono) rn
FROM dbo.dat_clientes_sucursales_contactos_telefonos AS a
INNER JOIN dbo.dat_rut_clientes AS b
ON a.rut_cliente = b.rut_cliente
WHERE cod_prioridad = 1
)
SELECT * FROM CTE where rn <=3
Basically I am trying to do a query to a page that holds just an image. I need to provide the tripID number and then the ROWID (as there could be multiple images) to receive a single image. I will be looping until each image is in its corresponding image box in html.
This code doesn't seem to work(I get Invalid column name 'ROWID'), but if I remove the AND ROWID='1' it returns all the images and its row id like this:
ROWID PHOTO
1 32jjr3h2jh23hj4h32jh42ll23j42
2 HU8308DJAOID9ASIDJI32C89EE29
-
Select ROW_NUMBER() OVER (ORDER BY Photo ASC) AS ROWID, TBL_Photo.Photo
From TBL_Photo
left join TBL_TripDetails
ON TBL_Photo.TripID=TBL_TripDetails.pkiTripID
Where pkiTripID = '121' AND ROWID = '1'
You can't reference a column alias in the WHERE clause -- you need to use a subquery or a CTE:
Subquery Example:
SELECT x.rowid,
x.photo
FROM (SELECT ROW_NUMBER() OVER (ORDER BY p.photo) AS ROWID,
p.photo
FROM TBL_PHOTO p
LEFT JOIN TBL_TRIPDETAILS td ON td.pkitripid = p.tripid
WHERE td.pkiTripID = '121') x
WHERE x.rowid = 1
CTE example:
WITH example AS (
SELECT ROW_NUMBER() OVER (ORDER BY p.photo) AS ROWID,
p.photo
FROM TBL_PHOTO p
LEFT JOIN TBL_TRIPDETAILS td ON td.pkitripid = p.tripid
WHERE td.pkiTripID = '121')
SELECT x.rowid,
x.photo
FROM example x
WHERE x.rowid = 1
Performance
There's no performance difference between the two options, but the WITH syntax isn't supported on all databases.