Creating query that returns id if condition is matched in rows from two tables

Question

I\'m learning SQL/dbms and using Postgres. I want to return rows that both have a certain value in a certain column. For example in tables Carpets and Curtain

Answer 1

Have read your question on Meta about this particular question, let me explain why all three answers are indeed correct - as is the way that you worked it out.

I have included examples of all three answers and the schema that they are working on:

Database changed
mysql> create table carpet(id int(3), material varchar(10), color varchar(15));
Query OK, 0 rows affected (0.02 sec)

mysql> create table curtain(id int(3), material varchar(10), color varchar(15));
Query OK, 0 rows affected (0.00 sec)

(bunch of insert statements)

mysql> select * from carpet;
+------+-----------+--------------+
| id   | material  | color        |
+------+-----------+--------------+
|    1 | wool      | Light Yellow |
|    2 | wool      | Beige        |
|    3 | polyester | Light Yellow |
|    4 | polyester | Light Red    |
+------+-----------+--------------+
4 rows in set (0.00 sec)

mysql> select * from curtain;
+------+----------+--------------+
| id   | material | color        |
+------+----------+--------------+
|    1 | Velvet   | Purple       |
|    2 | cotton   | White        |
|    3 | cotton   | Light Yellow |
|    4 | cotton   | Light Blue   |
+------+----------+--------------+
4 rows in set (0.00 sec)

An intersect uses two select statements and brings back matching results. In this case, you are looking for all the rows that have a matching color of 'Light Yellow'.

I can't give you an example in MySQL as it doesn't support it (As you can see below, it's not needed to give the same results).

A union query of two select statements each with a where clause only allowing the color of 'Light Yellow' will return the same data. Although a union can be used to return data that doesn't match, the where clause in each select statement means that it will ondeed only return the rows that you want.

mysql> select id, material, color from carpet
    -> union 
    -> select id, material, color from curtain;
+------+-----------+--------------+
| id   | material  | color        |
+------+-----------+--------------+
|    1 | wool      | Light Yellow |
|    2 | wool      | Beige        |
|    3 | polyester | Light Yellow |
|    4 | polyester | Light Red    |
|    1 | Velvet    | Purple       |
|    2 | cotton    | White        |
|    3 | cotton    | Light Yellow |
|    4 | cotton    | Light Blue   |
+------+-----------+--------------+
8 rows in set (0.00 sec)

Aww, that's bad right? Of course, we didn't specify the where clause:

mysql> select id, material, color from carpet where color='Light Yellow'
    -> union
    -> select id, material, color from curtain where color='Light Yellow';
+------+-----------+--------------+
| id   | material  | color        |
+------+-----------+--------------+
|    1 | wool      | Light Yellow |
|    3 | polyester | Light Yellow |
|    3 | cotton    | Light Yellow |
+------+-----------+--------------+
3 rows in set (0.00 sec)

A join between two tables on the color will allow you to return the rows from both the tables in a single row of data. You can specify the join on the two tables for the item color, and use a where clause to only return the rows that you are looking for.

mysql> select a.id, a.material, a.color, b.id, b.material 
    -> from curtain a join carpet b on a.color=b.color;
+------+----------+--------------+------+-----------+
| id   | material | color        | id   | material  |
+------+----------+--------------+------+-----------+
|    3 | cotton   | Light Yellow |    1 | wool      |
|    3 | cotton   | Light Yellow |    3 | polyester |
+------+----------+--------------+------+-----------+
2 rows in set (0.00 sec)

As you can see, this has returned only the rows with a matching color and allowed you to have columns from both tables in a single row of your resultset.

Now, I clearly didn't plan this very well as I have no other matching results apart from the 'Light Yellow' in both tables, so if I add a few more entries in we get this:

mysql> select * from curtain;
+------+----------+--------------+
| id   | material | color        |
+------+----------+--------------+
|    1 | Velvet   | Purple       |
|    2 | cotton   | White        |
|    3 | cotton   | Light Yellow |
|    4 | cotton   | Light Blue   |
|    5 | Wool     | White        |
|    6 | Fluff    | Beige        |
+------+----------+--------------+
6 rows in set (0.00 sec)

mysql> select * from carpet;
+------+-----------+--------------+
| id   | material  | color        |
+------+-----------+--------------+
|    1 | wool      | Light Yellow |
|    2 | wool      | Beige        |
|    3 | polyester | Light Yellow |
|    4 | polyester | Light Red    |
|    5 | Fluff     | Light Blue   |
+------+-----------+--------------+
5 rows in set (0.00 sec)

Now we can run that again, and this time get:

mysql> select a.id, a.material, a.color, b.id, b.material 
    -> from curtain a join carpet b on a.color=b.color;
+------+----------+--------------+------+-----------+
| id   | material | color        | id   | material  |
+------+----------+--------------+------+-----------+
|    3 | cotton   | Light Yellow |    1 | wool      |
|    3 | cotton   | Light Yellow |    3 | polyester |
|    4 | cotton   | Light Blue   |    5 | Fluff     |
|    6 | Fluff    | Beige        |    2 | wool      |
+------+----------+--------------+------+-----------+
4 rows in set (0.00 sec)

Oh noes!

This is now where we use the join and the where clause together:

mysql> select a.id, a.material, a.color, b.id, b.material 
    -> from curtain a join carpet b on a.color=b.color 
    -> where a.color='Light Yellow';
+------+----------+--------------+------+-----------+
| id   | material | color        | id   | material  |
+------+----------+--------------+------+-----------+
|    3 | cotton   | Light Yellow |    1 | wool      |
|    3 | cotton   | Light Yellow |    3 | polyester |
+------+----------+--------------+------+-----------+
2 rows in set (0.00 sec)

You see, in SQL there are often more ways to get the same result through different means than there are variations of the same data in your tables.

Edit: Okay, so if you only want rows where all the data matches, just include it in the join syntax:

mysql> select a.id, a.material, a.color, b.id, b.material 
    -> from curtain a 
    -> join carpet b on a.color=b.color
    -> and a.id=b.id
    -> where a.color='Light Yellow';
+------+----------+--------------+------+-----------+
| id   | material | color        | id   | material  |
+------+----------+--------------+------+-----------+
|    3 | cotton   | Light Yellow |    3 | polyester |
+------+----------+--------------+------+-----------+
1 row in set (0.00 sec)

As you can see, now we tell the join that both the id and color fields must match between the two tables - and the results speak for themselves. Now, in this case, I technically still didn't match ALL the columns as the material is different. If you wanted to match further, the query would not return any results as I have no matching records where the id, material AND color match, but the syntax would be as follows:

mysql> select a.id, a.material, a.color, b.id, b.material 
    -> from curtain a 
    -> join carpet b on a.color=b.color
    -> and a.id=b.id
    -> and a.material=b.material
    -> where a.color='Light Yellow';
Empty set (0.00 sec)

On that note though, you in most cases, you don't want all the columns to match. Very often tables have an ID that is only used for that table and is an automatically incrementing value. You want to use it to identify a unique row in that table, but not to use it to match unrelated tables. If anything, I would have suggested that you match on material and color - but leave the ID out of it.