Recently, I needed to build a query that would transform data in our database into a format that we posted to ElasticSearch. I’ll use the example of blog posts here since they’re easy for everyone to grasp. Imagine that each post can have many tags and you want one field on ElasticSearch with the tag ids and another field that has the tag names and description.

Here’s what the database might look like:

posts

id title
1 Blogging about dogs is fun!
2 Look at these cute dogs
3 Wow how about these cute dogs

post_tags

post_id tag_id
1 1
1 2
2 1
2 3
3 1

tags

id name description
1 dogs Posts about dogs
2 cute Cute things
3 omg OMG CUTE

And here is the result we’re looking for:

id title tag_ids tag_names_descriptions
1 Blogging about dogs is fun! 1, 2 dogs, cute, Posts about dogs, Cute things

Starting Query

Let’s start by retrieving a join table of our posts and tags. We’ll limit this to one post to keep our tables small for this example.

SELECT post.id,
       post.title,
       tag.id          AS tag_id,
       tag.name        AS tag_name,
       tag.description AS tag_description
FROM   posts post
       INNER JOIN posts_tags pt
               ON pt.post_id = post.id
       INNER JOIN tags tag
               ON tag.id = pt.tag_id
LIMIT  1;
id title tag_id tag_name tag_description
1 Blogging about dogs is fun! 1 dogs Posts about dogs
2 Blogging about dogs is fun! 2 cute Cute things

Ok, great, we have a table with our post and its tags, but we have duplicate rows! We can use GROUP BY to group these row by the post id, but then we’ll lose the tag data in the second row and get a result that looks like this:

id title tag_id tag_name tag_description
1 Blogging about dogs is fun! 1 dogs Posts about dogs

Group_concat

We can use the Group_concat function to concatenate data from multiple rows when we use GROUP BY:

SELECT post.id,
       post.title,
       Group_concat(tag.id) AS tag_ids,
       tag.name             AS tag_name,
       tag.description      AS tag_description
FROM   posts post
       INNER JOIN posts_tags pt
               ON pt.post_id = post.id
       INNER JOIN tags tag
               ON tag.id = pt.tag_id
GROUP  BY post.id
LIMIT  1;  
id title tag_ids tag_name tag_description
1 Blogging about dogs is fun! 1, 2 dogs Posts about dogs

Take a look at the MySQL Docs for Group_concat you can do some cool things with it like ensure the values are unique, sort the values, and choose a custom separator.

But what about our tag_name and tag_description? Here, we need to concatenate two separate columns into one!

Concat_ws

We know that Group_concat gives us a string, so if we look under String Functions in the MySQL docs, we’ll find Concat_ws. Concat_ws lets us concatenate two or more strings with a separator between them. The separator is a comma by default, but for legibility, we want a space too, so let’s use the SEPARATOR option. Here are our steps:

Concatenate the tag name:

Group_concat(tag.name SEPARATOR ", ")

Concatenate the tag description:

Group_concat(tag.description SEPARATOR ", ")

And concatenate both of those!

Concat_ws(
  ", "
  Group_concat(tag.name SEPARATOR ", "),
  Group_concat(tag.description SEPARATOR ", ")
)

Putting it all together

SELECT post.id,
       post.title,
       Group_concat(tag.id) AS tag_ids,
       Concat_ws(
         ", "
         Group_concat(tag.name SEPARATOR ", "),
         Group_concat(tag.description SEPARATOR ", ")
       )                    AS tag_names_descriptions
FROM   posts post
       INNER JOIN posts_tags pt
               ON pt.post_id = post.id
       INNER JOIN tags tag
               ON tag.id = pt.tag_id
GROUP  BY post.id
LIMIT  1;
id title tag_ids tag_names_descriptions
1 Blogging about dogs is fun! 1, 2 dogs, cute, Posts about dogs, Cute things

Nice work!

In this post, we’ve learned how to use Group_concat to concatenate data in a column when using GROUP BY. Group_concat lets us keep data that would otherwise disappear when we group things.

We also learned how to combine data from multiple columns by using Concat_ws together with Group_concat. With this function, we can join multiple columns and rows together into one value.