How to Split Sql Int Value into Multiple Rows - sql

Lets say I have the following table in MS SQL 2000
Id | description | quantity |
-------------------------------
1 my desc 3
2 desc 2 2
I need to display multiple rows based on the quantity, so I need the following output:
Id | description | quantity |
-----------------------------
1 my desc 1
1 my desc 1
1 my desc 1
2 desc 2 1
2 desc 2 1
Any ideas how to accomplish this?

This works just fine, no need for any cursors on this one. It may be possible to fangle something out without a number table as well.
Note if going for this kind of solution I would keep a number table around and not recreate it every time I ran the query.
create table #splitme (Id int, description varchar(255), quantity int)
insert #splitme values (1 ,'my desc', 3)
insert #splitme values (2 ,'desc 2', 2)
create table #numbers (num int identity primary key)
declare #i int
select #i = max(quantity) from #splitme
while #i > 0
begin
insert #numbers default values
set #i = #i - 1
end
select Id, description, 1 from #splitme
join #numbers on num <= quantity

DECLARE #Id INT
DECLARE #Description VARCHAR(32)
DECLARE #Quantity INT
DECLARE #Results TABLE (Id INT, [description] VARCHAR(32), quantity INT)
DECLARE MyCursor CURSOR FOR
SELECT Id, [description], quantity
FROM
MyTable
OPEN MyCursor
FETCH NEXT FROM MyCursor INTO #Id, #Description, #Quantity
WHILE ##FETCH_STATUS = 0
BEGIN
WHILE #Quantity > 0
BEGIN
INSERT INTO #Results (
Id,
[description],
quantity
) VALUES (
#Id,
#Description,
1
)
SET #Quantity = #Quantity - 1
END
FETCH NEXT FROM MyCursor INTO #Id, #Description, #Quantity
END
CLOSE MyCursor
DEALLOCATE MyCursor
SELECT *
FROM
#Results
By the way, cursors are generally considered evil. So I will both recommend against something like this, and thank everyone in advance for their flames ;) (But it should work)

Any one looking for a CTE based solution, here is one:
create table test_splitme (Id int, description varchar(255), quantity int)
insert test_splitme values (1 ,'my desc', 3)
insert test_splitme values (2 ,'desc 2', 2)
;
--TSQL solution
with splitcte as(
select Id,description, 1 as quantity, quantity as orig_quantity, 1 as cnt
from test_splitme
union all
select Id, description, quantity, orig_quantity,cnt+1
from splitcte
where cnt < orig_quantity
)
select Id, description, quantity
from splitcte
order by 1

Related

PARTITION based on a column and GROUP BY another column [duplicate]

This question already has answers here:
Get top 1 row of each group
(19 answers)
Select top 10 records for each category
(14 answers)
Closed 6 months ago.
Please consider the below script:
declare #tbl Table
(
CustomerId INT,
CountryID int,
Amount int
);
insert into #tbl values
(1,1,100),
(1,2,200),
(1,3,300),
(1,4,400),
(2,1,800),
(2,1,1000),
(3,1,500),
(2,4,200),
(2,3,900),
(3,1,3000),
(5,1,100),
(5,2,200),
(5,4,5000),
(6,1,1000),
(6,3,900),
(7,2,400),
(8,3,4000),
(2,1,100),
(1,1,100)
Declare #Result Table
(
CountryID int,
CustomerID int,
SumAmount int
);
Declare #CountryID int;
DECLARE db_cursor CURSOR FOR
SELECT distinct CountryID
FROM #tbl
OPEN db_cursor
FETCH NEXT FROM db_cursor INTO #CountryID
WHILE ##FETCH_STATUS = 0
BEGIN
insert into #Result
select top 2 #CountryID, CustomerID, SUM(Amount)
from #tbl
where CountryID = #CountryID
group by CustomerId
order by 3 desc
FETCH NEXT FROM db_cursor INTO #CountryID
END
CLOSE db_cursor
DEALLOCATE db_cursor
select *
from #Result
It returns this result :
CountryID CustomerID SumAmount
----------------------------------
1 3 3500
1 2 1900
2 7 400
2 5 200
3 8 4000
3 6 900
4 5 5000
4 1 400
In fact I want to get Top 2 customers that have maximum Amount in each Country.
How can I get that result without CURSOR and single query?
Thanks
The solution is :
WITH T AS
(
SELECT CountryID, CustomerId, SUM(Amount) AS SumAmount,
RANK() OVER(PARTITION BY CountryID ORDER BY SUM(Amount) DESC) AS R
FROM #tbl
GROUP BY CountryID, CustomerId
)
SELECT *
FROM T
WHERE R <= 2
But remember that when you want a top n rank , you will not systematically have exactly n rows returning because of ex aequo... You can have more, you can have less, depending of which ranking function you use and how many equal mesure you are ranking...

How to get records that between m records back and n records forward from a reference row - Non-sequential data

My scenario is as follows:
I have a reference record, say, ProductId = 1
The records each have a non-unique ItemTypeId
I would like to fetch records that exists between the following points
START POINT being 2 records BACKWARDS of type ItemTypeId = 1, from record of ProductId =1
END POINT being 3 records FORWARDS of type ItemTypeId = 1, from record of ProductId = 1
The query should get ALL data between the two points, inclusively
Here's a picture that illustrates this better than my words:
How would I structure my query to do this?
Any better way to do it without temp tables?
Thank-you!
Note that for this to work at all, you need that record ID to be an actual column in the table. Rows have no inherent order in a table.
With that in place, you can use LAG and LEAD to get what you want:
CREATE TABLE #t
(
RecordId INT IDENTITY(1,1),
ProductId INT,
ItemType INT
);
INSERT INTO #t(ProductId, ItemType)
VALUES
(5,1),(3,1),(7,3),(6,1),(2,7),
(1,1),(7,3),(8,1),(10,3),(9,5),
(11,1),(19,1),(17,4),(13,3);
WITH c1 AS
(
SELECT ProductId,
RecordId,
LAG(RecordId,2) OVER (ORDER BY RecordId) AS Back2,
LEAD(RecordId,3) OVER (ORDER BY RecordId) AS Forward3
FROM #t
WHERE ItemType = (SELECT ItemType FROM #t WHERE ProductId = 1)
),c2 AS
(
SELECT c1.Back2, c1.Forward3 FROM c1
WHERE c1.ProductId = 1
)
SELECT #t.*
FROM #t
INNER JOIN c2 ON #t.RecordId BETWEEN c2.Back2 AND c2.Forward3;
If you wanna do without using temp tables as you ask, the following solution work.
But it is not very nice i agree.
Well this is what i done :
CREATE DATABASE TEST;
USE TEST
CREATE TABLE PRODUCT
(
ProductId INT,
ItemType INT
)
INSERT INTO PRODUCT
VALUES
(5,1),
(3,1),
(7,3),
(6,1),
(2,7),
(1,1),
(7,3),
(8,1),
(10,3),
(9,5),
(11,1),
(19,1),
(17,4),
(13,3)
DECLARE product_cursor CURSOR FOR
SELECT * FROM PRODUCT;
OPEN product_cursor
DECLARE
#ProductId INT,
#ItemId INT,
#END_FETCH INT,
#countFrom INT,
#countTo INT
DECLARE #TableResult TABLE
(
RProductId INT,
RItemId INT
)
FETCH NEXT FROM product_cursor
INTO #ProductId, #ItemId
SET #END_FETCH = 0
SET #countFrom = 0
SET #countTo = 0
WHILE ##FETCH_STATUS = 0 AND #END_FETCH = 0
BEGIN
IF #ItemId = 1 AND (#countFrom = 0 AND #countTo = 0)
BEGIN
SET #countFrom = 3
SET #countTo = 3
END
ELSE
BEGIN
IF #countFrom > 0
BEGIN
--SELECT 'INSERTION : ' ,#ProductId,#ItemId
INSERT INTO #TableResult VALUES(#ProductId, #ItemId)
IF #ItemId = 1
BEGIN
SET #countFrom -= 1
--SELECT 'CountFrom : ', #countFrom
END
END
ELSE
BEGIN
IF #countTo > 0
BEGIN
--SELECT 'INSERTION : ' ,#ProductId,#ItemId
INSERT INTO #TableResult VALUES(#ProductId, #ItemId)
IF #ItemId = 1
BEGIN
SET #countTo -= 1
--SELECT 'CountTO : ', #countTo
END
END
ELSE
BEGIN
SET #END_FETCH = 1
END
END
END
FETCH NEXT FROM product_cursor
INTO #ProductId, #ItemId
END
CLOSE product_cursor
DEALLOCATE product_cursor
SELECT * FROM #TableResult
And this is the result i got :
RProductId RItemId
3 1
7 3
6 1
2 7
1 1
7 3
8 1
10 3
9 5
11 1
19 1
But i prefer the solution of #James Casey.
By the way, why won't you use temp table ?

Offset & Fetch Next will be ok to Use in a Loop?

I need to write a loop to pass data which are on a SQL temporary table to a stored procedure one by one. Is it OK to use the below approach? Does it cost lots of memory?
DECLARE #SomeData AS TABLE(
Id INT,
Name VARCHAR(MAX),
Age INT
)
INSERT INTO #SomeData
( Id, Name, Age )
VALUES ( 1, 'AAAAA', 10 ),
( 2, 'BBBBB', 12 ),
( 3, 'CCCCC', 13 ),
( 4, 'DDDDD', 14 ),
( 5, 'EEEEE', 15 );
DECLARE #iterator INT = 0
WHILE #iterator< (SELECT COUNT(Id) FROM #SomeData)
BEGIN
SELECT *
FROM #SomeData
ORDER BY Id
OFFSET #iterator ROWS
FETCH NEXT 1 ROWS ONLY
SET #iterator +=1
/*Execute a Stored Procedure*/
END
I think you don't need to (SELECT COUNT(Id) FROM #SomeData) again after while, if table SomeData is a big data it will slow. You can try like this:
DECLARE #iterator INT = 0
declare #n int = (SELECT COUNT(Id) FROM #SomeData)
WHILE #iterator< #n
BEGIN
SELECT *
FROM #SomeData
ORDER BY Id
OFFSET #iterator ROWS
FETCH NEXT 1 ROWS ONLY
SET #iterator +=1
/*Execute a Stored Procedure*/
END

How to create loop based on value of row?

I have problem when I use my query bellow to have a looping inside the cursor.
data in table1 will be like this:
id | data
----|---------
A | 4
B | 2
C | 5
the result in table2 should be like this:
id | data
----|---------
A | 1
A | 1
A | 1
A | 1
B | 1
B | 1
C | 1
C | 1
C | 1
C | 1
C | 1
I have SQL query with cursor like this:
DECLARE #table2 table ( id VARCHAR(500), data INTEGER)
DECLARE Cur CURSOR FOR
SELECT id, data FROM table1
OPEN Cur
WHILE ( ##FETCH_STATUS = 0 )
BEGIN
DECLARE #LoopNum INTEGER
DECLARE #tempID VARCHAR(255)
DECLARE #tempDATA INTEGER
FETCH NEXT FROM Cur INTO #tempID, #tempDATA
set #LoopNum = 0
WHILE #LoopNum < #tempDATA
BEGIN
INSERT INTO table2 (id, data)
VALUES( #tempID, 1)
SET #LoopNum = #LoopNum + 1
END
END
CLOSE Cur
DEALLOCATE Cur
SELECT * FROM table2
but the query didn't work. is there something wrong with my query?
Thank you.
Use this query to the expected result.
CREATE TABLE #test
(id CHAR(1),data INT)
INSERT #test VALUES ('A',4)
INSERT #test VALUES('B',2)
INSERT #test VALUES('C',5);
SELECT s.id, 1 AS data
FROM #test s
INNER JOIN
master.dbo.spt_values t ON t.type='P'
AND t.number BETWEEN 1 AND s.data
Note: Refer this Why (and how) to split column using master..spt_values?
You actually don't need a loop
IF OBJECT_ID('TEMPDB..#TEMP') IS NOT NULL
DROP TABLE #TEMP
SELECT 'A' AS ID, 4 AS DATA
INTO #TEMP UNION
SELECT 'B', 2 UNION
SELECT 'C', 5
;WITH CTE AS
(
SELECT 1 AS NUMBER
UNION ALL
SELECT NUMBER + 1
FROM CTE
WHERE NUMBER < 100
)
SELECT T.ID, 1
FROM CTE C
INNER JOIN #TEMP T
ON C.NUMBER <= T.DATA
ORDER BY T.ID
Carefull that if you want ot generate a large set of numbers in the CTE it may become slower.
Use a Recursive CTE which will help you to loop through the records.
CREATE TABLE #test
(id CHAR(1),data INT)
INSERT #test
VALUES ('A',4),('B',2),('C',5);
WITH cte
AS (SELECT 1 AS da,id,data
FROM #test a
UNION ALL
SELECT da + 1,id,data
FROM cte a
WHERE da < (SELECT data
FROM #test b
WHERE a.id = b.id))
SELECT id,
1 AS data
FROM cte
ORDER BY id
i used two loops
1. for each row
2. for number for duplicate insert
SET NOCOUNT on;
DECLARE #t table(row int IDENTITY(1,1),id varchar(10),data int)
INSERT INTO #t
SELECT * from xyz
DECLARE #x table(id varchar(10),data int) --table to hold the new data
DECLARE #i int=(SELECT count (*) from xyz) --number of rows main table
DECLARE #y int --number of duplicate
DECLARE #p int=1 --number of rows
WHILE #i!=0 --loop until last row of main table
BEGIN
SET #y=(SELECT data FROM #t WHERE row=#p) --set #y for number of 'row duplicate'
WHILE #y!=0
BEGIN
INSERT INTO #x
SELECT id,1
FROM #t
WHERE row=#p
SET #y=#y-1
END
SET #p=#p+1
SET #i=#i-1
END
SELECT * FROM #x

How to implement FIFO in sql

I working on FIFO implementation in sql.
I have Batch number concept in my application.
If suppose I am selling on inventory then my application should tell me that which inventory is the first come.
Lets. Say I purchased Inventory 'A' on 4th-Aug, 5th-Aug & 6th-Aug
On 4th Aug - A Inventory has batch number BT002 - 10 (Qty)
On 5th Aug - A's Inventory has batch number BT003 - 15 (Qty)
On 6th Aug - A's Inventory has batch number BT001 - 10 (Qty)
So, Now I am having stock Now in my hand as following :
A Inventory
BT002 - 10 - 4-Aug
BT003 - 15 - 5-Aug
BT001 - 10 - 6-Aug
Now If I want to sell that Inventory to anybody then my application should tell me that I should sell
BT002 (Batch number) inventory first beacause it came first.
That was the concept I am using in my application.
Now I want to sell 15 Qty from 'A' (Inventory).
Then O/p Should be like this :
BT002 - 10
BT003 - 5
Here's My Query :
SELECT ISNULL(SUM(qty),0) AS Qty,batch_no,accept_date FROM RS_GIN_Master
GROUP BY batch_no,accept_date
HAVING ISNULL(SUM(qty),0) <= 15
ORDER BY accept_date asc
O/p Of Given Query :
How can I get O/P like this :
BT002 - 10
BT003 - 5
Any Help will be appreciated.
Thank you in Advance.
This should work for you:
Working sample on Fiddle
CREATE FUNCTION [dbo].[GetBatchAmounts]
(
#requestedAmount int
)
RETURNS
#tBatchResults TABLE
(
Batch nvarchar(50),
Amount int
)
AS
BEGIN
/*This is just a mock of ersults of your query*/
DECLARE #RS_GIN_Master TABLE(
Qty int,
batch_no NVARCHAR(max),
accept_date DATETIME
)
insert into #RS_GIN_Master(Qty,batch_no,accept_date)
SELECT 10,'BT002', CAST(CAST(2014 AS varchar) + '-' + CAST(8 AS varchar) + '-' + CAST(4 AS varchar) AS DATETIME)
insert into #RS_GIN_Master(Qty,batch_no,accept_date)
SELECT 10,'BT003', CAST(CAST(2014 AS varchar) + '-' + CAST(8 AS varchar) + '-' + CAST(5 AS varchar) AS DATETIME)
insert into #RS_GIN_Master(Qty,batch_no,accept_date)
SELECT 10,'BT001', CAST(CAST(2014 AS varchar) + '-' + CAST(8 AS varchar) + '-' + CAST(6 AS varchar) AS DATETIME)
/*---------------------------*/
DECLARE #Qty int
DECLARE #batch_no NVARCHAR(max)
DECLARE #accept_date DATETIME
DECLARE myCursor CURSOR FOR
SELECT Qty, batch_no, accept_date FROM #RS_GIN_Master ORDER BY accept_date ASC
OPEN myCursor
FETCH NEXT FROM myCursor INTO #Qty, #batch_no,#accept_date
WHILE (##FETCH_STATUS = 0 AND #requestedAmount > 0 )
BEGIN
Declare #actualQty int
IF #requestedAmount > #Qty
SET #actualQty = #Qty
ELSE
SET #actualQty = #requestedAmount
INSERT INTO #tBatchResults (batch, Amount)
SELECT #batch_no, #actualQty
set #requestedAmount = #requestedAmount - #actualQty
FETCH NEXT FROM myCursor INTO #Qty, #batch_no,#accept_date
END /*WHILE*/
CLOSE myCursor
DEALLOCATE myCursor
RETURN
END
Just make sure to replace the marked part of the function with your query...
You need to create a stored procedure in your database and taking the quantity from your stock table. And you should also have the id of each record to update that records from where u have taken that qty.
Alter PROCEDURE sp_UpdateStockForSale
#batchNO varchar(10),
#qty decimal(9,3)
AS
BEGIN
Create Table #tmpOutput(ID int identity(1,1), StockID int, batchNo varchar(10), qty decimal(9,3));
SET NOCOUNT ON;
DECLARE #ID int;
DECLARE #Stock Decimal(9,3);
DECLARE #TEMPID int;
Select #TEMPID=(Max(ID)+1) From RS_GIN_Master Where qty > 0 And batch_no = #batchNO;
While (#qty > 0) BEGIN
Select #ID=ID, #Stock=qty From RS_GIN_Master Where qty > 0 And batch_no = #batchNO AND ID < #TEMPID Order By accept_date Desc;
--If Outward Qty is more than Stock
IF (#Stock < #qty) BEGIN
SET #qty = #qty - #Stock;
SET #Stock = 0;
END
--If Outward Qty is less than Stock
ELSE BEGIN
SET #Stock = #Stock - #qty;
SET #qty = 0;
END
Insert Into #tmpOutput(StockID,batchNo,qty)Values(#ID,#batchNO,#Stock);
SET #TEMPID = #ID;
--This will update that record don't need it now.
--Update RS_GIN_Master Set qty = #Stock Where ID=#ID
END
Select StockID, batchNo, qty From #tmpOutput;
END
GO
The above example is not compiled but, you can get the logic how you can retrieve the records from your stock table according to FIFO method. You can use accept_date instead of ID in RS_GIN_Master table. but, i would prefer to make it unique so, if i want to get a specific record then it can be possible.
One query .. like this
This should be tweaked for you case as you have groups and other stuff, is only for example purposes.
;with qty as (
select 15 as value
)
,l as (
select
ROW_NUMBER () over (order by accept_date desc) rn
,*
from xxx
)
,q as (
select
batch_no
,accept_date
,case when value>qty then value-qty else 0 end as remainder
,case when value>qty then qty else value end as used
,rn
from l
cross join qty
where rn=1
union all
select
r.batch_no
,r.accept_date
,case when q.remainder>r.qty then q.remainder-r.qty else 0 end as remainder
,case when q.remainder>r.qty then r.qty else q.remainder end as used
,r.rn
from q
join l r
on q.rn+1 = r.rn
where q.remainder!=0
)
select *
from q
where used != 0
and the fiffle for it http://sqlfiddle.com/#!6/9b063/34/0
Below should work for you
Create table RS_GIN_Master
(id int,
qty int,
batch_no varchar(5),
accept_date Datetime
)
GO
Insert into RS_GIN_Master (id, qty, batch_no, accept_date)
values(1,10,'BT001','2018-04-06')
Insert into RS_GIN_Master (id, qty, batch_no, accept_date)
values(2,10,'BT002','2018-04-04')
Insert into RS_GIN_Master (id, qty, batch_no, accept_date)
values(3,15,'BT003','2018-05-06')
GO
----------------------------
CREATE PROC FIFO
#TakenQty int
AS
BEGIN
WITH cte AS (SELECT *, SUM(qty) OVER (ORDER BY accept_date, id ASC) as CumQty FROM RS_GIN_Master WHERE qty>0)
SELECT TOP ((SELECT COUNT(*) FROM cte WHERE CumQty <#TakenQty)+1) batch_no, accept_date,
CASE
WHEN CumQty<#TakenQty THEN qty
ELSE #TakenQty -(CumQty-Qty)
END AS TakenOut
FROM cte
END
Result
| batch_no | accept_date | TakenOut |
|----------|----------------------|----------|
| BT002 | 2018-04-04T00:00:00Z | 10 |
| BT001 | 2018-04-06T00:00:00Z | 5 |
http://www.sqlfiddle.com/#!18/f7ee7/1