How can I generate sql matrix - sql

Question

How I can create this matrix from the tables in sql query? I have table which show how many goods was sold. I want show title of goods instead id of goods. But it's important that fields toy1, toy2 and toy3 have datatype decimal(20,8)
Table 1
id year toy1 toy2 toy3
-------------------------------
-1 1 3 2
1 2016 100 20 50
2 2017 150 120 40
Table 2
id toy
---------------
1 doll
2 car
3 bear
Result matrix
id year toy1 toy2 toy3
-------------------------------
-1 doll bear car
1 2016 100 20 50
2 2017 150 120 40

Answer 1

For SQL Server i will do some left joins
SELECT T.id,
T.year,
CASE
WHEN T1.TOY IS NULL THEN CAST(T.[toy1] AS VARCHAR) ELSE T1.TOY
END [toy1],
CASE
WHEN T2.TOY IS NULL THEN CAST(T.[toy2] AS VARCHAR) ELSE T2.TOY
END [toy2],
CASE
WHEN T3.TOY IS NULL THEN CAST(T.[toy3] AS VARCHAR) ELSE T3.TOY
END [toy3]
FROM <Table1> T
LEFT JOIN <Table2> T1 ON T1.ID = T.toy1
LEFT JOIN <Table2> T2 ON T2.ID = T.toy2
LEFT JOIN <Table2> T3 ON T3.ID = T.toy3;
Result :
id year toy1 toy2 toy3
-1 NULL doll bear car
1 2016 100 20 50
2 2017 150 120 40

Answer 2

This is a very bad data model.
A table should represent an entity, such as a toy, an order, yearly sales, ... A column on the other side represents an attribute of that entity, e.g. the toy's name or price, the order's date, the sale's year. In your model you are mixing this. What is "toy1"? A toy ID? An amount? A mix of both it seems. Then your table 2 allows for many, many toys, whereas your table1 allows for only 3. That doesn't match.
What you should have instead:
table toys
toy_id | name
-------+-----
1 | doll
2 | car
3 | bear
table sales
year | toy_id | amount
2016 | 1 | 100
2016 | 2 | 20
2016 | 3 | 50
2017 | 1 | 150
2017 | 2 | 120
2017 | 3 | 40
And you wouldn't write a query to give you the matrix either. You would select the data (e.g. select * from sales order by year, toy_id) and have your app handle the display. It is easy to fetch the data in a loop with whatever programming language, be it C#, Visual Basic, Java, PHP, whatever, and display it in a grid.
In an SQL query you must know the columns beforehand, but this is not the case with your matrix output, as you would have to know how many toys are in the toys table before running or even writing the query. Everytime you added a toy, you'd have to alter your query. This is not how relational databases are supposed to work.

Answer 3

I am not sure but something like this:
SELECT T1.id,
year,
CASE WHEN T1.Toy1 = T2.id AND T1.id=-1 THEN T2.toy
ELSE T1.Toy1
END AS Toy1,
CASE WHEN T1.Toy2 = T2.id AND T1.id=-1THEN T2.toy
ELSE T1.Toy2
END AS Toy1,
CASE WHEN T1.Toy3 = T2.id AND T1.id=-1 THEN T2.toy
ELSE T1.Toy3
END AS Toy3
FROM Table1 T1
LEFT JOIN Table2 T2 ON T1.id=-1 AND T2.id IN (T1.Toy1, T1.Toy2, T1.Toy3)

Answer 4

One way is to only select the -1 row and apply the algorithms to get the toy names. That can be UNION ALL with the rest of the table that is not the -1 row.
You will have to Cast the output as varchar to mix words and numbers:
DECLARE #table1 TABLE (id int, [year] int, toy1 int, toy2 int, toy3 int)
DECLARE #table2 TABLE (id int, toy varchar(100))
INSERT INTO #table1 VALUES (-1, null, 1, 3, 2), (1, 2016, 100,20,50), (2,2017,150,120,40)
INSERT INTO #table2 VALUES (1,'doll'), (2,'car'), (3,'bear')
-- for the -1 row
SELECT T1.id, COALESCE(CAST(year as varchar), '') AS year
,(SELECT toy FROM #table2 T2 WHERE T2.id = T1.toy1) AS toy1
,(SELECT toy FROM #table2 T2 WHERE T2.id = T1.toy2) AS toy2
,(SELECT toy FROM #table2 T2 WHERE T2.id = T1.toy3) AS toy3
FROM #table1 T1
WHERE T1.id = -1
UNION ALL
--all other rows
SELECT T3.id
,CAST(T3.year as varchar)
,CAST(T3.toy1 as varchar)
,CAST(T3.toy2 as varchar)
,CAST(T3.toy3 as varchar)
FROM #table1 T3
WHERE T3.id <> -1
Gives output:
id year toy1 toy2 toy3
-1 doll bear car
1 2016 100 20 50
2 2017 150 120 40