How can I accomplish the following in SQL? - mysql

Question

I have two tables.
table_a:
id | data_x | data_y
--------------------
1 person joe
2 person bob
3 amount 200
4 addres philville
tableB:
map_id | table_a_id
-------------------
7 1
7 3
7 4
8 4
8 2
The result I want is the map_id if it has an entry in table_a for both data_x = 'person' and data_y = '200'
So with the above table B, the result should be
map_id
------
7
How can I write that query in SQL?

Answer 1

This situation is a perfect fit for an unusual SQL operator: INTERSECT. It is a very declarative, efficient and elegant solution for this problem.
SELECT Map.map_id
FROM Table_B AS Map JOIN Table_A AS Person ON (Person.id = Map.table_a_id) AND (Person.data_x = 'person')
INTERSECT
SELECT Map.map_id
FROM Table_B AS Map JOIN Table_A AS Amount ON (Amount.id = Map.table_a_id) AND (Amount.data_y = '200')
Formally what you are asking for is exactly the intersection of two disjoint sets: the set of map id's that are persons and the set of map id's that have a value of 200.
Please note the INTERSECT operator does not exists in MySQL, but it does in almost all advanced relational DBMS, including PostgreSQL.

Answer 2

This is less elegant than the INTERSECT solution #Malta posted, but it works with the limited capabilities of MySQL as well:
SELECT b1.map_id
FROM table_a a1
JOIN tableb b1 ON a1.id = b1.table_a_id AND a1.data_x = 'person'
JOIN tableb b2 ON b2.map_id = b1.map_id AND b2.table_a_id <> b1.table_a_id
JOIN table_a a2 ON a2.id = b2.table_a_id AND a2.data_y = '200';
SQL Fiddle for MySQL.
SQL Fiddle for Postgres.

Answer 3

Based on your input, the following should get you started using MySQL:
SELECT
map_id
FROM TableB
JOIN Table_A
ON TableB.table_a_id = Table_A.id
AND
((Table_A.data_x = 'person')
OR
(Table_A.data_y = '200')
)
GROUP BY map_id
HAVING COUNT(table_a_id) = 2
;
See it in action: SQL Fiddle.
Update
As Erwin Brandstetter made explicit: If the data can't be trusted to be inherently consistent (along the lines of your inquiry), one option is:
SELECT map_id FROM (
SELECT map_id, 'data_x' t
FROM TableB B JOIN Table_A A ON B.table_a_id = A.id AND A.data_x = 'person'
UNION
SELECT map_id, 'data_y'
FROM TableB B JOIN Table_A A ON B.table_a_id = A.id AND A.data_y = '200'
) T
GROUP BY map_id
HAVING COUNT(DISTINCT t) = 2
;
This should ensure "at least one each". (Alternatives have been suggested by others.) To get "exactly one each", you could try
SELECT map_id FROM (
SELECT map_id, 'data_x' t, data_y
FROM TableB B JOIN Table_A A ON B.table_a_id = A.id AND A.data_x = 'person'
UNION
SELECT map_id, 'data_y', data_y
FROM TableB B JOIN Table_A A ON B.table_a_id = A.id AND A.data_y = '200'
) T
GROUP BY map_id
HAVING COUNT(DISTINCT t) = 2 AND COUNT(DISTINCT data_y) = 2
;
See it in action (with additional test data): SQL Fiddle.
And it works in PostgreSQL as well: SQL Fiddle
Please comment if and as this requires adjustment / further detail.

Answer 4

Join the 2 tables, group by map_id, use conditional counting with either count() or sum(), and filter in having clause (I use mysql syntax below):
select map_id,
sum(
case
when a.data_x='person' or a.data_y='200' then 1
else 0
end
) as matches
from a
inner join b on a.id=b.a_id
group by b.map_id
having matches=2
The above query assumes that you cannot have more than one record for any map_id where data_x is person or data_y is 200. If this assumption is incorrect, then you need to use either exists subqueries or 2 derived tables.

Answer 5

Sounds like you want a standard INNER JOIN.
But I do beg to differ on your result:
map_id if it has an entry in table_a for both data_x = 'person' and data_y = '200'
There is not a record in your data set that has both 'person' and data_y = '200' and therefore no mp_id can be returned
Here is a typical INNER JOIN relating to your narrative.
SELECT DISTINCT
b.map_id
FROM
TableA a
INNER JOIN TableB b
ON a.id = b.table_a_id
WHERE
a.data_x = 'person'
AND a.data_y = '200'
If more than one map_id exists with data_x = 'person' and data_y = '200' then you will get multiple results but only 1 row per map_id
If you want the map_id(s) for records with data_x = 'person' or data_y = '200' then switch the and in the where statement to or and you will receive map_id 7 & 8.
SELECT DISTINCT
b.map_id
FROM
TableA a
INNER JOIN TableB b
ON a.id = b.table_a_id
WHERE
a.data_x = 'person'
OR a.data_y = '200'
Note this encompasses (7,1)(8,2) because 1 & 2 both have data_x = 'person' and then (7,3) because 3 has data_y = '200' therefore it would return map_id 7 & 8.

Answer 6

select map_id from
table_b b
left outer join table_a a1 on (b.table_a_id = a1.id and a1.data_x = 'person')
left outer join table_a a2 on (b.table_a_id = a2.id and a2.data_y = '200')
group by map_id
having count(a1.id) > 0 and count(a2.id) > 0

Answer 7

Lets do it simple:
SELECT * FROM
(
SELECT map_id
FROM table_a a1
inner join TableB b1 ON a1.id = b1.table_a_id
where a1.data_x = 'person'
) as p
inner join
(
SELECT map_id
FROM table_a a1
inner join TableB b1 ON a1.id = b1.table_a_id
where a1.data_y = '200'
) as q
on p.map_id = q.map_id
You may replace SELECT * FROM with SELECT p.map_id FROM.
You may add more sub-set-joins to have more conditions.
sql-fiddle