Question

I am using SQLite and I’m not sure if this can be done. Consider I have these table:

======>Table: Total_Table

_id    Name   Ingredients

0     "A"     "table_a"<br>
1     "B"     "table_b"

=====>Table: table_a

 _id    Name

 0      "a_1" <br>
 1      "a_2"

=====>Table: table_b

_id    Name
0      "a_1" <br>

The tables referenced in the rows of Total_Table have a different number of rows and are created as rows in Total_Table are added.

At the moment, I need to structure a query which will return rows from Total_Table.

But will search through the entries of the Ingredients tables.

For example, if I am looking for all rows in Total_Table which have a_1 in their Ingredients table, it would return rows 0,1 from Total_Table.

Currently, I am doing this by querying all rows of Total_Table, then iterating through those entries and querying each table individually, forming an array as I go.

This is pretty bad since the tables I am looking through are relatively large, and hurts performance.

Any ideas? Thanks!

Answer 1

I think you’ll find this easier with a slightly different database schema. SQLite is a relational database, so let’s try to store things accordingly. Basically, we’ll have 3 tables, one for Total_Table, one for your ingredients, and a table to map which ingredients to which instance in the Total_Table.

Logically, it will look like this:

Table: Total_Table

_id    Name
0      "A"
1      "B"

Table: Ingredients

_id    Name

0      "a_1"
1      "a_2"

Table: Total_Ingredients

total_id  ingredient_id
0         0
0         1
1         0

The benefits here are many. First, you don’t need to create any tables at runtime. You’ll need exactly 3 tables, regardless of how many Total_Table entries you have. Second, it’s easy to query. Third, you’re not duplicating the name a_1 anywhere in the revised schema (this is called Database normalization).

Then, your query would look like this. Note that in the sample data, a_1 has an id of 0, which we’ll use for querying purposes.

select _id, Name from Total_Table where id in 
  (select total_id from Total_Ingredients where ingredient_id = 0)

Update: Your other query is pretty similar if I understand correctly. So, you want the recipes which have some or all of a given ingredients. Suppose your subset includes ids 1, 5 and 7 for this example, the only change is to use an in clause.

select _id, Name from Total_Table where id in 
  (select distinct total_id from Total_Ingredients where ingredient_id in (1,5,7))