Find BAILEY's BLUE BANDANNA - sql

Question

Imagine dance teams (of varying size) can choose one or more outfits and each outfit can have one or more items. All members of a team must have all same outfits and all the items of each outfit, i.e., incomplete outfits are bad -- we want to find those bad outfits.
The tables below define two teams: APPLE and BANANA. Team APPLE has 3 members, team BANANA two members. The APPLEs have chosen a single outfit which has a single item, the YELLOW PANTSUIT outfit, naturally a crowd favorite. Team BANANA have two outfits: RED, and BLUE; but, BAILEY is missing the BANDANNAS item for the BLUE outfit. Awkward.
Let's get BAILEY out of trouble. Create a query to find missing items in the outfits chosen by the teams.
Thanks to #Brits for the SQL below:
Teams and team-members:
CREATE TABLE team (
id int unique,
name text
);
INSERT INTO team (id, name)
VALUES (1, 'APPLE'),
(2, 'BANANA');
CREATE TABLE team_member (
id int unique,
name text,
team_id int references team (id)
);
INSERT INTO team_member (id, name, team_id)
VALUES (1, 'ADAM', 2),
(2, 'BAILEY', 2),
(3, 'CATE', 1),
(4, 'DAVE', 1),
(5, 'ERIN', 1);
Outfits and outfit items:
CREATE TABLE outfit (
id int unique,
name text
);
INSERT INTO outfit (id, name)
VALUES (1, 'RED'),
(2, 'YELLOW'),
(3, 'BLUE');
CREATE TABLE outfit_item (
id int unique,
name text,
outfit_id int references outfit (id)
);
INSERT INTO outfit_item (id, name, outfit_id)
VALUES (1, 'SHORTS', 1),
(2, 'SHIRT', 1),
(3, 'PANTSUIT', 2),
(4, 'BANDANNA', 3),
(5, 'HAT', 3);
Team member outfit items:
CREATE TABLE member_item (
member_id int references team_member (id),
item_id int references outfit_item (id)
);
INSERT INTO member_item (member_id, item_id)
VALUES (1, 1),
(1, 2),
(1, 4),
(1, 5),
(2, 1),
(2, 2),
(2, 5),
(3, 3),
(4, 3),
(5, 3);
Team APPLE's members all have the YELLOW PANTSUIT outfit so the APPLEs are ready to rumble.
Team BANANA chose the RED and BLUE outfits; sadly, I failed to give team BANANA the YELLOW PANTSUIT outfit - they would have killed it, but whatever. Team BANANA's ADAM and BAILEY have the items for the RED outfit, but BAILEY does not have the BANDANNA for the BLUE outfit; let's not let a bandanna get in the way to team BANANA so we need a query to return just:
BANANA BAILEY BLUE BANDANNAS
To find the number of items in an outfit:
SELECT
o.name
, count(*) AS "number of items"
FROM
outfit_item i
, outfit o
WHERE
i.outfit_id = o.id
GROUP BY
i.outfit_id
, o.name
name | number of items
--------+-----------------
RED | 2
BLUE | 2
YELLOW | 1
Similarly, the number of members in a team:
SELECT
t.name
, count(*) AS "number of members"
FROM
team_member m
, team t
WHERE
m.team_id = t.id
GROUP BY
m.team_id
, t.name
name | number of members
--------+-------------------
BANANA | 2
APPLE | 3
That's all well and good, but how do we combine this information so we can bail BAILEY out of this BANANA blunder?

Answer 1

Taking on board the comment "If any member has an outfit item then all team members must have all items in that outfit" I believe the following will do what you are looking for (I'm using a CTE to work out what outfits are associated with each team).
with teamoutfit as (
-- If any mmber of a team has any part of an outfit then that team is linked to that outfit
select distinct te.id as team_id, oft.id as outfit_id from
team te
inner join team_member tm on tm.team_id = te.id
inner join member_item mi on mi.member_id = tm.id
inner join outfit_item oi on oi.id = mi.item_id
inner join outfit oft on oft.id = oi.outfit_id
)
select te.name as team, tm.name as member, of.name as outfit, oi.name as item from
team te
inner join team_member tm on tm.team_id = te.id
inner join teamoutfit tof on tof.team_id = tm.team_id
inner join outfit of on of.id = tof.outfit_id
inner join outfit_item oi on tof.outfit_id = oi.outfit_id
left join member_item mi on oi.id = mi.item_id and tm.id = mi.member_id
where
mi.member_id is null
Output:
| team | member | outfit | item |
|--------|--------|--------|----------|
| BANANA | BAILEY | BLUE | BANDANNA |
I setup a SQL Fiddle you can use to play with this (and perhaps clarify your question if I misunderstood what you are trying to do).

Answer 2

Confirmed to work on postgres, albeit not pretty and am 85% sure this can be simplified. Also, I was a little confused on how outfit and outfit_item tables are related. In the way you laid out the tables, bandanas and hats are are always blue, pantsuits are always yellow and shorts/shirts are always red when you join outfit to outfit_item (outfit.id = outfit_item.outfit_id). The test data and query has been written with that understanding.
select distinct
tm.name as member_name,
t.name as team_name,
o.name as outfit_color,
oi.name as outfit_item
from team_member tm
join team t
on t.id = tm.team_id
join member_items mi
on mi.member_id = tm.id
join(
with outfit_item_count as
(SELECT outfit_id, count(*) AS items
FROM outfit_items
GROUP BY 1)
select mi.member_id, oi.outfit_id, oic.items as items_needed, count(mi.item_id) as member_item_count
from member_items mi
join outfit_items oi on oi.id = mi.item_id
join outfit_item_count oic on oic.outfit_id = oi.outfit_id
group by 1,2,3)z
on z.member_id = tm.id
join outfit_items oi
on oi.outfit_id = z.outfit_id
join outfit o
on o.id = oi.outfit_id
where z.items_needed > z.member_item_count
and oi.id not in (select item_id from member_items)