Concatenate Values from Different T-SQL Queries - sql-server

Question

I have a table in SQL Server with the following data:
+-----------------+-------------------+-------------------+--------+
|Product Family | Product Class | Product | Sales |
|Food | Vegetables | Cauliflower | 24 |
|Food | Prepared Meals | Steak & Patatoes | 54 |
|Food | Fruit | Apples | 76 |
|Food | Fruit | Oranges | 14 |
|Food | Fruit | Pears | 32 |
|Electronics | MP3 Players | Cool Player Z | 57 |
|Electronics | MP3 Players | iStuff 16GB | 45 |
|Electronics | TV's | HD | 96 |
|Electronics | TV's | Ultra HD | 76 |
+-----------------+-------------------+-------------------+--------+
There is a hierarchy in this data:
Product Family
Product Class
Product
I'd like to create a query that will return the sum for each hierarchy level. This union does that:
SELECT 1 as Level, [Product Family] as Item, SUM(SALES) as Sales
FROM [dbo].[HK_Termp_01] GROUP BY [Product Family]
UNION ALL
SELECT 2 as Level, [Product Class] as Item, SUM(SALES) as Sales
FROM [dbo].[HK_Termp_01] GROUP BY [Product Class]
UNION ALL
SELECT 3 as Level, Product as Item, SUM(SALES) as Sales
FROM [dbo].[HK_Termp_01] GROUP BY Product
However, I also require an additional column that will be a concatenation of the 3 string columns, in the order of the hierarchy. The desired output being:
+--------------------------+-----------------------------------------------+--------+
| Level ||Item | Hierarchy | Sales |
| 1 ||Electronics | Electronics | 274 |
| 1 ||Food | Food | 200 |
| 2 ||Fruit | Food > Fruit | 122 |
| 2 ||MP3 Players | Electronics > MP3 Players | 102 |
| 2 ||Prepared Meals | Food > Prepared Meals | 54 |
| 2 ||TV's | Electronics > TV's | 172 |
| 2 ||Vegetables | Food > Vegetables | 24 |
| 3 ||Apples | Food > Fruit > Apples | 76 |
| 3 ||Cauliflower | Food v Vegetables > Cauliflower | 24 |
| 3 ||Cool Player Z | Electronics > MP3 Players > Cool Player Z | 57 |
| 3 ||HD | Electronics > TV's > HD | 96 |
| 3 ||iStuff 16GB | Electronics v MP3 Players > iStuff 16GB | 45 |
| 3 ||Oranges | Food > Fruit > Oranges | 14 |
| 3 ||Pears | Food > Fruit v Pears | 32 |
| 3 ||Steak & Patatoes | Food v Prepared Meals > Steak & Patatoes | 54 |
| 3 ||Ultra HD | Electronics > TV's > Ultra HD | 76 |
+--------------------------+--------------+------+-------------------------+--------+
This is where I get stuck. I can't add all 3 fields to each query in the Union, because then I don't get the right totals by level. But I'm not sure what other avenue to try.
Thanks & Let me know what other info I can supply to clarify the case.

Answer 1

I think you just want a tweak on your query:
SELECT 1 as Level, [Product Family] as Item,
SUM(SALES) as Sales
FROM [dbo].[HK_Termp_01]
GROUP BY [Product Family]
UNION ALL
SELECT 2 as Level, [Product Family] + '>' + [Product Class] as Item,
SUM(SALES) as Sales
FROM [dbo].[HK_Termp_01]
GROUP BY [Product Family] + '>' + [Product Class]
UNION ALL
SELECT 3 as Level, [Product Family] + '>' + [Product Class] + '>' + Product as Item,
SUM(SALES) as Sales
FROM [dbo].[HK_Termp_01]
GROUP BY [Product Family] + '>' + [Product Class] + '>' + Product;
That said, you could do this using GROUPING_SETS:
SELECT [Product Family], [Product Class], Product, SUM(SALES) as Sales
FROM [dbo].[HK_Termp_01]
GROUP BY GROUPING SETS ( ([Product Family], [Product Class], Product),
([Product Family], [Product Class]),
([Product Family])
);
You would then need to fiddle with the names to get the exact output you want.

Answer 2

Just for fun,
Declare #YourTable table ([Product Family] varchar(50),[Product Class] varchar(50),Product varchar(50),Sales int)
Insert Into #YourTable values
('Food','Vegetables','Cauliflower',24),
('Food','Prepared Meals','Steak & Patatoes',54),
('Food','Fruit','Apples',76),
('Food','Fruit','Oranges',14),
('Food','Fruit','Pears',32),
('Electronics','MP3 Players','Cool Player Z',57),
('Electronics','MP3 Players','iStuff 16GB',45),
('Electronics','TV''s','HD',96),
('Electronics','TV''s','Ultra HD',76)
Declare #Top varchar(25) = NULL --<< Sets top of Hier Try ''MP3 Players''
Declare #Nest varchar(25) = '|-----' --<< Optional: Added for readability
;with cte0 as (
Select Distinct ID=Product,Parent=[Product Class],Sales from #YourTable
Union All
Select Distinct ID=[Product Class],Parent=[Product Family],0 from #YourTable
Union All
Select Distinct ID=[Product Family],Parent='Total',0 from #YourTable
Union All
Select Distinct ID='Total',Parent=NULL,0 )
,cteP as (
Select Seq = cast(100000+Row_Number() over (Order by ID) as varchar(500))
,ID
,Parent
,Lvl=1
,Sales = Sales
From cte0
Where IsNull(#Top,'X') = case when #Top is null then isnull(Parent,'X') else ID end
Union All
Select Seq = cast(concat(p.Seq,'.',100000+Row_Number() over (Order by r.ID)) as varchar(500))
,r.ID
,r.Parent
,p.Lvl+1
,r.Sales
From cte0 r
Join cteP p on r.Parent = p.ID)
,cteR1 as (Select *,R1=Row_Number() over (Order By Seq) From cteP)
,cteR2 as (Select A.Seq,A.ID,R2=Max(B.R1) From cteR1 A Join cteR1 B on (B.Seq like A.Seq+'%') Group By A.Seq,A.ID )
Select A.R1
,B.R2
,A.ID
,A.Parent
,A.Lvl
,Title = Replicate(#Nest,A.Lvl-1) + A.ID
,Sales = (Select sum(Sales) from cteR1 S where S.R1 between A.R1 and B.R2)
From cteR1 A
Join cteR2 B on A.ID=B.ID
Group By A.R1,B.R2,A.ID,A.Parent,A.Lvl
Order By A.R1
Returns
Now, If you set #Top = 'MP3 Players' rather than NULL, you'll get :
Just a little narrative:
cte0, we normalize your hierarchy into a Parent/Child relationship
cteP, we build your hierarchy via a recursive cte
cteR1, we generate the sequence/R1 keys
cteR2, we generate the R2 Keys
Now, If yo have slow-moving hierarchies, I tend to store them with the range keys to facilitate navigation and aggregation.