MySQL Select parent row from a many to many relation pivot - mysql

Question

I have three tables, products, ingredients and ingredient_product.
I need to find all products, where the product has the related ingredients.
It also needs to have a matching value on the percentage column.
A product exists with two ingredients related.
+-----+------------+---------------+------------+
| id | product_id | ingredient_id | percentage |
+-----+------------+---------------+------------+
| 1 | 1 | 1 | 50 |
| 2 | 1 | 2 | 50 |
+------------------+--------------+-------------+
SQL to retrieve:
SELECT
products.id
FROM
products,
ingredient_product
WHERE
ingredient_product.product_id = products.id
AND
(ingredient_product.ingredient_id = 1 AND ingredient_product.percentage = 50)
AND
(ingredient_product.ingredient_id = 2 AND ingredient_product.percentage = 50)
But this returns an empty result. Empty set (0.00 sec)
Products:
+-------------------+-----------------------+------+-----+---------+----------------+
| Field | Type | Null | Key | Default | Extra |
+-------------------+-----------------------+------+-----+---------+----------------+
| id | int(10) unsigned | NO | PRI | NULL | auto_increment |
| name | varchar(255) | NO | | NULL | |
| short_description | text | YES | | NULL | |
| long_description | text | YES | | NULL | |
+-------------------+-----------------------+------+-----+---------+----------------+
Ingredients:
+-------------------+-----------------------+------+-----+---------+----------------+
| Field | Type | Null | Key | Default | Extra |
+-------------------+-----------------------+------+-----+---------+----------------+
| id | int(10) unsigned | NO | PRI | NULL | auto_increment |
| short_name | varchar(50) | NO | | NULL | |
| thumbnail | varchar(255) | NO | | NULL | |
+-------------------+-----------------------+------+-----+---------+----------------+
ingredient_product
+---------------+---------------------+------+-----+---------+----------------+
| Field | Type | Null | Key | Default | Extra |
+---------------+---------------------+------+-----+---------+----------------+
| id | int(10) unsigned | NO | PRI | NULL | auto_increment |
| product_id | int(10) unsigned | NO | MUL | NULL | |
| ingredient_id | int(10) unsigned | NO | MUL | NULL | |
| percentage | tinyint(3) unsigned | NO | | NULL | |
+---------------+---------------------+------+-----+---------+----------------+

Answer 1

Based on Strawberries answer:
SELECT product_id
FROM ingredient_product
WHERE ingredient_id IN (1,2) AND percentage = 50
GROUP BY product_id
HAVING COUNT(*) = 2;
However this doesn't work if you need separate percentages, but you can do this:
SELECT product_id FROM
(SELECT product_id
FROM ingredient_product
WHERE ingredient_id = 1 AND percentage = 50
UNION ALL
SELECT product_id
FROM ingredient_product
WHERE ingredient_id = 2 AND percentage = 50) AS tmp
GROUP BY product_id
HAVING COUNT(*) = 2;
Essentially you add a UNION for each requirement (consisting of an id and percentage combo) and then increase the HAVING condition to the number of requirements.
Notes: Since the requirements are mutually exclusive in this case, UNION ALL is quicker than UNION and will give you the same result.
Instead of an UNION in a subquery you could use OR but we found that MySQL seems to like this format better. That however might change from version to version. For completeness sake here that solution as well:
SELECT product_id
FROM ingredient_product
WHERE (ingredient_id = 1 AND percentage = 50) OR (ingredient_id = 2 AND percentage = 50)
GROUP BY product_id
HAVING COUNT(*) = 2;

Answer 2

On the assumption that id is redundant, and that you have a perfectly serviceable natural key on (product_id,ingredient_id)...
SELECT product_id
FROM ingredient_product
WHERE ingredient_id IN (1,2)
GROUP
BY product_id
HAVING COUNT(*) = 2;