How do I write a group_concat function in sqlalchemy?

I am trying to re-write this mysql query in SQLAlchemy:

Schema:

CREATE TABLE `posts` (
    `post_id` INT UNSIGNED PRIMARY AUTO_INCREMENT,
    `post_name` VARCHAR(255)
) Engine=InnoDB;

CREATE TABLE `post_tags` (
    `tag_id` INT UNSIGNED PRIMARY AUTO_INCREMENT,
    `tag_name` VARCHAR(255)
) Engine=InnoDB;

CREATE TABLE `post_tags_map` (
    `map_id` INT PRIMARY AUTO_INCREMENT,
    `post_id` INT NOT NULL,
    `tags_id` INT NOT NULL,
    FOREIGN KEY `post_id` REFERENCES `posts` (`post_id`),
    FOREIGN KEY `post_id` REFERENCES `post_tags` (`tag_id`)
) Engine=InnoDB;

Query:

SELECT 
    posts.*,
    GROUP_CONCAT( post_tags.tag_name order by post_tags.tag_name ) AS tags

  FROM posts
    LEFT JOIN posts_tags_map
      ON posts_tags_map.post_id = posts.post_id
    LEFT JOIN post_tags
      ON posts_tags_map.tags_id = posts_tags.tag_id

  WHERE posts.post_id = 1
  GROUP BY post_id

Here’s what I have, but I keep getting:

1064, "You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near 'SEPARATOR /))

rows = DBSession.query(Posts,func.group_concat(Post_Tags.tag_name.op('SEPARATOR')(literal_column('/')))).outerjoin(PostsTagsMap,Posts.post_id==PostsTagsMap.post_id).outerjoin(Post_Tags,PostsTagsMap.tags_id==Post_Tags.tag_id).group_by(Posts.post_id)

Answer

Defining a custom function element seems the easiest way to take care of the group_concat method.

from sqlalchemy.sql import expression
import sqlalchemy
from sqlalchemy.ext import compiler


class group_concat(expression.FunctionElement):
    name = "group_concat"


@compiler.compiles(group_concat, 'mysql')
def _group_concat_mysql(element, compiler, **kw):
    if len(element.clauses) == 2:
        separator = compiler.process(element.clauses.clauses[1])
    else:
        separator = ','

    return 'GROUP_CONCAT(%s SEPARATOR %s)'.format(
        compiler.process(element.clauses.clauses[0]),
        separator,
    )

And use it something like this:

query = select([
    table.c.some_column,
    expression.label(
        'grouped column',
        group_concat(
            table.c.some_oter_column,
            ' separator ',
        ),
    ),
]).group_by(table.c.some_column)