CODEWITHSUNDEEP
sqlsqlitejoin
Shaun
Shaun

Reputation: 5531

SQLite query with multiple joins

This is pretty much just a simple join statemnet I believe. I've not worked with SQL much lately and seem to have forgotten how to do this. What I have is an item with few columns in it that reference another table for the name of that field. Like this:

id, name, effect1, effect2, effect3, effect4

The effects reference another table that only has a, id, and name columns. What I'm trying to do is run a query that will pull those names for each of those effects.

Something like: 

SELECT i.name,e.name AS effect1, e.name AS effect2, e.name AS effect3, 
       e.name AS effect4 
FROM item i, effects e 
WHERE i.effect1 = e.name 
 AND i.effect2 = e.name 
 AND i.effect3 = e.name 
 AND i.effect4 = e.name

So, say I have an item that has values like this:

Toast, 1, 2, 3, 4

and the effects are:

1, burned
2, untoasted
3, wet
4, texas

I want it to display toast, burned, untoasted, wet, texas

And ideas?

update:

Table items
id, name, weight, value, effect1,effect2,effect3,effect4

Table effects
id, name

In the effect1,... columns are the id number for the corresponding item in the effect table. A lot ofitems are going to share the same effects, so instead of inflating this already large database with redundant data, I decided to use a join to save space. At the same time I managed how to forget to do it, lol

Update #2 This is the effect I'm going for, but on more than one of the effect columns

SELECT i.name, i.weight,i.value, e.name AS 'effect 1' 
FROM ingredients i JOIN effects e ON effects._id=i.effect1

This works for 1, but if I try to do multiple it just crashes. Any ideas how I can get that effect for all 4 effects?

Upvotes: 14

Views: 35155

Answers (1)

Erwin Brandstetter
Erwin Brandstetter

Reputation: 656381

You need a distinct join for every column:

SELECT i.name
     , i.weight
     , i.value
     , e1.name AS effect1
     , e2.name AS effect2
     , e3.name AS effect3
     , e4.name AS effect4
FROM   ingredients i 
LEFT   JOIN effects e1 ON e1.id = i.effect1
LEFT   JOIN effects e2 ON e2.id = i.effect2
LEFT   JOIN effects e3 ON e3.id = i.effect3
LEFT   JOIN effects e4 ON e4.id = i.effect4;

LEFT JOIN still keeps the ingredient if any of the effects is missing.
The query depends on effects.id being unique.

You can achieve the same with correlated subqueries:

SELECT i.name
     , i.weight
     , i.value
     , (SELECT e.name FROM effects e WHERE e.id = i.effect1) AS effect1
     , (SELECT e.name FROM effects e WHERE e.id = i.effect2) AS effect2
     , (SELECT e.name FROM effects e WHERE e.id = i.effect3) AS effect3
     , (SELECT e.name FROM effects e WHERE e.id = i.effect4) AS effect4
FROM   ingredients i;

Database design

If every ingredient has 4 effects your db design is fine. If the number of effects vary or you have additional information per effect, you might consider an n:m relationship between ingredients and effects, implemented by an additional table. (Replacing the four effect* columns.) Could look like this:

CREATE TABLE ingredients_effects (
  ingredients_id integer REFERENCES ingredients(id)
, effects_id     integer REFERENCES effects(id)
  -- additional data like quantity or notes?
, PRIMARY KEY (ingredients_id, effects_id)
);

More details in the fine manual.

Upvotes: 30

Related Questions