Select multiple values from same column based on 2 columns - sql

Question

Best explained using an example. My trials and results are bellow the example.
There are two tables (in reality I have multiple tables)
TABLE: Products
ID name
-----------
1 apple
2 orange
3 pear
TABLE: ATTRIBUTES
ID prod_ID attr_id value
----------------------------
1 1 101 20
2 1 102 red
3 1 103 sweet
4 2 101 30
5 2 102 orange
6 2 103 sour
6 3 101 40
7 3 102 green
8 3 103 sweet
DESIRED OUTPUT
name attr_id 101 AS 'price' attr_id 102 AS 'taste'
------------------------------------------------------
apple 20 sweet
orange 30 sour
pear 40 sweet
I have managed SQL till now but recently I have had to call 3 tables and combine column values like shown above. I just can't get my head wrapped around this. Help would be greatly appreciated.

Answer 1

Since you have not mentioned any RDBMS in your question, the query below will work on most RDBMS.
SELECT a.Name,
MAX(CASE WHEN b.attr_ID = 101 THEN b.value END) Price,
MAX(CASE WHEN b.attr_ID = 103 THEN b.value END) Taste
FROM Products a
INNER JOIN Attributes b
ON a.ID = b.prod_ID
GROUP BY a.Name
SQLFiddle Demo

Answer 2

You can use CASE statement for that: (This will work in MySQL)
SELECT p.name
,GROUP_CONCAT(CASE WHEN attr_id = 101 THEN value else NULL END) AS price
,GROUP_CONCAT(CASE WHEN attr_id = 102 THEN value else NULL END) AS color
,GROUP_CONCAT(CASE WHEN attr_id = 103 THEN value else NULL END) AS taste
FROM Products p JOIN Attributes a
ON p.id = a.prod_id
GROUP BY p.name;
Or you can also do the same with join: (This will work in both MySQL and SQL Server)
SELECT Name
,CASE WHEN a.attr_id = 101 THEN a.value ELSE NULL END AS price
,CASE WHEN b.attr_id = 102 THEN b.value else NULL END AS color
,CASE WHEN c.attr_id = 103 THEN c.value else NULL END AS taste
FROM Products p
LEFT JOIN Attributes a ON p.id = a.prod_id
LEFT JOIN Attributes b ON p.id = b.prod_id AND a.attr_id = 101 AND b.attr_id = 102
LEFT JOIN Attributes c ON p.id = c.prod_id AND a.attr_id = 101 AND c.attr_id = 103
WHERE a.attr_id IS NOT NULL and b.attr_id IS NOT NULL AND c.attr_id IS NOT NULL
I have also added a column for color.
Output
| NAME | PRICE | COLOR | TASTE |
-----------------------------------
| apple | 20 | red | sweet |
| orange | 30 | orange | sour |
| pear | 40 | green | sweet |
See this SQLFiddle
Demo for SQL Server

Answer 3

You have to query for each column separately...
SELECT
p.name,
(SELECT value FROM Attributes a WHERE attr_id=101 AND a.prod_ID=p.ID) AS price,
(SELECT value FROM Attributes a WHERE attr_id=102 AND a.prod_ID=p.ID) AS taste
FROM Products p
... or ...
If you are using MSSQL 2008 R2 you can use PIVOT:
http://msdn.microsoft.com/en-us/library/ms177410(v=sql.105).aspx

Answer 4

Using PIVOT, easier to adjust for extra attributes.
SELECT NAME, [101],[102],[103]
FROM (
SELECT P.NAME as NAME, A.ATTR_ID as ATTR_ID, A.VALUE as VALUE
FROM PRODUCTS as P, ATTRIBUTES as A
WHERE A.PROD_ID = P.ID
) SrcTable
PIVOT
(
MAX(VALUE)
for ATTR_ID in ([101],[102],[103])
) PivotTable;
Or a dynamic statement could be generated to adjust for a variable number of attributes (since this is the only thing that changes in the above statement)
The column string could be obtained with somethings like:
DECLARE #columns AS NVARCHAR(MAX);
set #columns = STUFF((SELECT distinct ',' + QUOTENAME(ATTR_ID)
FROM ATTRIBUTES
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)'),1,1,'');
Next, this could variable could be used to generate a string like the static PIVOT
DECALARE #stmt AS NVARCHAR(MAX);
set #stmt ='SELECT NAME, ' + #columns + '
FROM (
SELECT P.NAME as NAME, A.ATTR_ID as ATTR_ID, A.VALUE as VALUE
FROM PRODUCTS as P, ATTRIBUTES as A
WHERE A.PROD_ID = P.ID
) SourceTable
PIVOT (
MAX(VALUE)
for ATTR_ID in (' + #columns + ')
) PivotTable';
EXECUTE(#stmt);