MySQL JOIN the most recent row only?

Viewed 147653

I have a table customer that stores a customer_id, email and reference. There is an additional table customer_data that stores a historical record of the changes made to the customer, i.e. when there's a change made a new row is inserted.

In order to display the customer information in a table, the two tables need to be joined, however only the most recent row from customer_data should be joined to the customer table.

It gets a little more complicated in that the query is paginated, so has a limit and an offset.

How can I do this with MySQL? I think I'm wanting to put a DISTINCT in there somewhere...

The query at the minute is like this-

SELECT *, CONCAT(title,' ',forename,' ',surname) AS name
FROM customer c
INNER JOIN customer_data d on c.customer_id=d.customer_id
WHERE name LIKE '%Smith%' LIMIT 10, 20

Additionaly, am I right in thinking I can use CONCAT with LIKE in this way?

(I appreciate that INNER JOIN might be the wrong type of JOIN to use. I actually have no clue what the difference is between the different JOINs. I'm going to look into that now!)

9 Answers

I know this question is old, but it's got a lot of attention over the years and I think it's missing a concept which may help someone in a similar case. I'm adding it here for completeness sake.

If you cannot modify your original database schema, then a lot of good answers have been provided and solve the problem just fine.

If you can, however, modify your schema, I would advise to add a field in your customer table that holds the id of the latest customer_data record for this customer:

CREATE TABLE customer (
  id INT UNSIGNED NOT NULL AUTO_INCREMENT PRIMARY KEY,
  current_data_id INT UNSIGNED NULL DEFAULT NULL
);

CREATE TABLE customer_data (
   id INT UNSIGNED NOT NULL AUTO_INCREMENT PRIMARY KEY,
   customer_id INT UNSIGNED NOT NULL, 
   title VARCHAR(10) NOT NULL,
   forename VARCHAR(10) NOT NULL,
   surname VARCHAR(10) NOT NULL
);

Querying customers

Querying is as easy and fast as it can be:

SELECT c.*, d.title, d.forename, d.surname
FROM customer c
INNER JOIN customer_data d on d.id = c.current_data_id
WHERE ...;

The drawback is the extra complexity when creating or updating a customer.

Updating a customer

Whenever you want to update a customer, you insert a new record in the customer_data table, and update the customer record.

INSERT INTO customer_data (customer_id, title, forename, surname) VALUES(2, 'Mr', 'John', 'Smith');
UPDATE customer SET current_data_id = LAST_INSERT_ID() WHERE id = 2;

Creating a customer

Creating a customer is just a matter of inserting the customer entry, then running the same statements:

INSERT INTO customer () VALUES ();

SET @customer_id = LAST_INSERT_ID();
INSERT INTO customer_data (customer_id, title, forename, surname) VALUES(@customer_id, 'Mr', 'John', 'Smith');
UPDATE customer SET current_data_id = LAST_INSERT_ID() WHERE id = @customer_id;

Wrapping up

The extra complexity for creating/updating a customer might be fearsome, but it can easily be automated with triggers.

Finally, if you're using an ORM, this can be really easy to manage. The ORM can take care of inserting the values, updating the ids, and joining the two tables automatically for you.

Here is how your mutable Customer model would look like:

class Customer
{
    private int id;
    private CustomerData currentData;

    public Customer(String title, String forename, String surname)
    {
        this.update(title, forename, surname);
    }

    public void update(String title, String forename, String surname)
    {
        this.currentData = new CustomerData(this, title, forename, surname);
    }

    public String getTitle()
    {
        return this.currentData.getTitle();
    }

    public String getForename()
    {
        return this.currentData.getForename();
    }

    public String getSurname()
    {
        return this.currentData.getSurname();
    }
}

And your immutable CustomerData model, that contains only getters:

class CustomerData
{
    private int id;
    private Customer customer;
    private String title;
    private String forename;
    private String surname;

    public CustomerData(Customer customer, String title, String forename, String surname)
    {
        this.customer = customer;
        this.title    = title;
        this.forename = forename;
        this.surname  = surname;
    }

    public String getTitle()
    {
        return this.title;
    }

    public String getForename()
    {
        return this.forename;
    }

    public String getSurname()
    {
        return this.surname;
    }
}

Easy solution to left join the 1 most/least recent row is using select over ON phrase

SELECT *
FROM A
LEFT JOIN B
ON A.id = (SELECT MAX(id) FROM B WHERE id = A.id)

Where A.id is the auto-incremental primary key.

Related