Selecting multiple columns to one string (by date) - sql

Question

My pricelist table looks like this:
ItemCode VendorCode UnitCost StartingDate
333 362 2.31 2016-08-19 00:00:00.0
333 362 2.16 2018-02-22 00:00:00.0
444 362 12.96 2014-01-09 00:00:00.0
444 362 13.10 2015-01-09 00:00:00.0
444 430 13.05 2017-04-01 00:00:00.0
444 550 13.30 2018-02-01 00:00:00.0
I would like to have query result following:
333:(362,2.16,2018-02-22)
444:(362,13.10,2015-01-09),(430,13.05,2017-04-01),(550,13.30,2018-02-01)
So all different Vendors and their prices should be listed and only latest by date.
I got this far:
SELECT
Pricelist.ItemCode + ': '+ temp.data
FROM Pricelist
INNER JOIN (SELECT p1.ItemCode,
STUFF((SELECT '; ' + p2.VendorCode
FROM Pricelist p2
WHERE p2.ItemCode = p1.ItemCode
ORDER BY VendorCode
FOR XML PATH('')), 1, 1, '') AS data
FROM Pricelist p1) as temp
ON Pricelist.ItemCode = temp.ItemCode
GROUP BY Pricelist.ItemCode, temp.data
ORDER BY 1
But not even close to the result I need.

Answer 1

I would use row_number() function :
select concat(itemcode, ':',
stuff( ( select top (1) with ties ',(' +concat(VendorCode, ',', UnitCost ,',', cast(StartingDate as date)) +')'
from Pricelist
where itemcode = p.itemcode
order by row_number() over (partition by VendorCode order by StartingDate desc)
for xml path('')
), 1, 1, ''
))
from Pricelist p
group by itemcode;

Answer 2

Try with this query:
create table #t (ItemCode int, VendorCode int, UnitCost decimal (10,2), StartingDate datetime)
insert into #t values
(333,362, 2.31,'2016-08-19 00:00:00.0'),
(333,362, 2.16,'2018-02-22 00:00:00.0'),
(444,362,12.96,'2014-01-09 00:00:00.0'),
(444,362,13.10,'2015-01-09 00:00:00.0'),
(444,430,13.05,'2017-04-01 00:00:00.0'),
(444,550,13.30,'2018-02-01 00:00:00.0')
;with tr1 as (
select
convert(varchar(100),ItemCode) + ':' as ItemCode,
'(' + convert(varchar(100),VendorCode) + ',' + convert(varchar(100),UnitCost) + ',' + convert(varchar(19),StartingDate,121) + ')' as Vals,
row_number() over (partition by ItemCode,VendorCode order by StartingDate desc) rn
from #t
)
select distinct ItemCode,
stuff((
select ',' + Vals
from tr1 b
where b.ItemCode=tr1.ItemCode
and rn=1
for xml path ('')
),1,1,'')
from tr1
where rn=1

Answer 3

First grouping the itemcodes, and then linking that to a string with the vendor details might be quite performant.
Linking those itemcodes to an OUTER APPLY with a FOR XML works well.
For example :
declare #Pricelist table (ItemCode int, VendorCode int, UnitCost decimal (10,2), StartingDate datetime)
insert into #Pricelist values
(333,362,02.31,'2016-08-19T00:01:00'),
(333,362,02.16,'2018-02-22T00:02:00'),
(444,362,12.96,'2014-01-09T00:03:00'),
(444,362,13.10,'2015-01-09T00:04:00'),
(444,430,13.05,'2017-04-01T00:05:00'),
(444,550,13.30,'2018-02-01T00:06:00');
select concat(itemcode,':',stuff(x.details,1,1,'')) as ItemVendorDetails
from (select distinct itemcode from #Pricelist) i
outer apply
(
select top 1 with ties
concat(',(',VendorCode,',',UnitCost,',',convert(date,StartingDate),')')
from #Pricelist p
where p.ItemCode = i.ItemCode
order by row_number() over (partition by ItemCode, VendorCode order by StartingDate desc)
for xml path('')
) x(details);
Result:
ItemVendorDetails
------------------------------------------------------------------------
333:(362,2.16,2018-02-22)
444:(362,13.10,2015-01-09),(430,13.05,2017-04-01),(550,13.30,2018-02-01)

Answer 4

The following query works with the sample data provided:
;WITH VendorPerItemCTE AS (
SELECT ItemCode,
VendorCode,
UnitCost,
StartingDate,
ROW_NUMBER() OVER (PARTITION BY ItemCode, VendorCode
ORDER BY StartingDate DESC) AS seq
FROM PriceList
)
SELECT CAST(ItemCode AS VARCHAR(12)) + ':' + STUFF(ItemData , 1, 1, '')
FROM (
SELECT DISTINCT p.ItemCode,
(SELECT ', (' + CAST(VendorCode AS VARCHAR(10)) + ', ' +
CAST(UnitCost AS VARCHAR(10)) + ', ' +
CONVERT(VARCHAR(12), StartingDate, 102 ) +
')'
FROM VendorPerItemCTE AS c
WHERE p.ItemCode = c.ItemCode AND c.seq = 1
FOR XML PATH('')) AS ItemData
FROM PriceList AS p) AS t
Demo here