views:

133

answers:

2

I have a table to entities (lets call them people) and properties (one person can have an arbitrary number of properties). Ex:

People

Name  Age
--------
Jane  27
Joe   36
Jim   16

Properties

Name   Property
-----------------
Jane   Smart
Jane   Funny
Jane   Good-looking
Joe    Smart
Joe    Workaholic
Jim    Funny
Jim    Young

I would like to write an efficient select that would select people based on age and return all or some of their properties.

Ex: People older than 26
Name Properties
Jane Smart, Funny, Good-looking
Joe Smart, Workaholic

It's also acceptable to return one of the properties and total property count.

The query should be efficient: there are millions of rows in people table, hundreds of thousands of rows in properties table (so most people have no properties). There are hundreds of rows selected at a time.

Is there any way to do it?

+4  A: 

Use:

   SELECT x.name,
          GROUP_CONCAT(y.property SEPARATOR ', ')
     FROM PEOPLE x
LEFT JOIN PROPERTIES y ON y.name = x.name
    WHERE x.age > 26
 GROUP BY x.name

You want the MySQL function GROUP_CONCAT (documentation) in order to return a comma separated list of the PROPERTIES.property value.

I used a LEFT JOIN rather than a JOIN in order to include PEOPLE records that don't have a value in the PROPERTIES table - if you only want a list of people with values in the PROPERTIES table, use:

   SELECT x.name,
          GROUP_CONCAT(y.property SEPARATOR ', ')
     FROM PEOPLE x
     JOIN PROPERTIES y ON y.name = x.name
    WHERE x.age > 26
 GROUP BY x.name

I realize this is an example, but using a name is a poor choice for referencial integrity when you consider how many "John Smith"s there are. Assigning a user_id, being a numeric value, would be a better choice.

OMG Ponies
I heart GROUP_CONCAT :)
retailevolved
and of course there should be an index on Properties.name
meriton
@meriton: I'd like have both columns in the PROPERTIES table as the primary key.
OMG Ponies
Right, that is even better than just blindly following the "every foreign key should have an index" rule :-)
meriton
A: 

You can use INNER JOIN to link the two tables together. More info on JOINs.

SELECT *
FROM People P
INNER JOIN Properties Pr
  ON Pr.Name = P.Name
WHERE P.Name = 'Joe' -- or a specific age, etc

However, it's often a lot faster to add a unique primary key to tables like these, and to create an index to increase speed.

Say the table People has a field id
And the table Properties has a field peopleId to link them together

Then the query would then look something like this:

SELECT *
FROM People P
INNER JOIN Properties Pr
  ON Pr.id = P.peopleId
WHERE P.Name = 'Joe'
Alec