Convert a complex SQL query to SQLAlchemy

Question

I run out of ideas. I googled now more then one day and I still can't find any useful answer to my question.

What I did until now, I tried to use raw SQL but without luck.

locations = db.session.query(Location, select([text('( 6371 * acos( cos( radians("53.6209798282177") ) * cos( radians( lat ) ) * cos( radians( lng ) - radians("13.96948162900808") ) + sin( radians("53.6209798282177") ) * sin( radians( lat ) ) ) )')]).label('distance')).having('distance' < 25).all()

When using this raw SQL query I get zero results returned, but when running the same query in mysql it returns correct results.

Further more I figured out, that when printing the query to terminal, it does not handle the HAVING() clause correctly.

My query looks like this when printed:

SELECT location.id AS location_id, location.created_date AS location_created_date, location.zip AS location_zip, location.user_id AS location_user_id, location.lat AS location_lat, location.lng AS location_lng, location.city AS location_city 
FROM location 
HAVING false = 1

How do I convert this SQL query to SQLAlchemy

SELECT *, ( 6371 * acos( cos( radians(53.6209798282177) ) * cos( radians( lat ) ) * cos( radians( lng ) - radians(11.96948162900808) ) + sin( radians(53.6209798282177) ) * sin( radians( lat ) ) ) ) AS distance FROM location HAVING distance < 25 ORDER BY distance;

My table looks like this:

+--------------+----------------+------+-----+---------+-------+
| Field        | Type           | Null | Key | Default | Extra |
+--------------+----------------+------+-----+---------+-------+
| id           | varchar(50)    | NO   | PRI | NULL    |       |
| created_date | datetime       | YES  |     | NULL    |       |
| zip          | varchar(5)     | NO   | UNI | NULL    |       |
| user_id      | varchar(50)    | NO   |     | NULL    |       |
| lat          | decimal(15,13) | NO   |     | NULL    |       |
| lng          | decimal(15,13) | NO   |     | NULL    |       |
| city         | text           | NO   |     | NULL    |       |
+--------------+----------------+------+-----+---------+-------+

Any help is appreciated.

Answer 1

Your HAVING is handled correctly, but you're passing it the wrong expression. It seems that you're using Python 2, since the relational comparison between a string and an integer

'distance' < 25

does not raise an exception, but evaluates to False instead. In other words your query is equal to

locations = db.session.query(...).having(False).all()

which explains why you get zero results: all rows are explicitly filtered out by the HAVING clause, as seen in the printed version:

...
HAVING false = 1  -- remove all rows

A solution is to use a suitable construct, such as column(), to produce the expression:

locations = db.session.query(...).having(column('distance') < 25).all()

You shouldn't wrap the complex select list item expression in a select(), which represents a SELECT statement. Either label the text() as is:

text('( 6371 * acos( cos( radians("53.6209798282177") ) * '
     'cos( radians( lat ) ) * cos( radians( lng ) - radians("13.96948162900808") ) + '
     'sin( radians("53.6209798282177") ) * sin( radians( lat ) ) ) ) '
     'AS distance')

or build the expression using the model:

(6371 *
 func.acos(func.cos(func.radians(53.6209798282177)) *
           func.cos(func.radians(Location.lat)) *
           func.cos(func.radians(Location.lng) - func.radians(13.96948162900808)) +
           func.sin(func.radians(53.6209798282177)) *
           func.sin(func.radians(Location.lat)))).label('distance')

You could improve the readability of your query construction by making a function for the great-circle distance, and with a little bit of work you could implement a hybrid method on Location:

import math

def gc_distance(lat1, lng1, lat2, lng2, math=math):
    ang = math.acos(math.cos(math.radians(lat1)) *
                    math.cos(math.radians(lat2)) *
                    math.cos(math.radians(lng2) -
                             math.radians(lng1)) +
                    math.sin(math.radians(lat1)) *
                    math.sin(math.radians(lat2)))

    return 6371 * ang

class Location(db.Model):
    ...
    @hybrid_method
    def distance(self, lat, lng):
        return gc_distance(lat, lng, self.lat, self.lng)

    @distance.expression
    def distance(cls, lat, lng):
        return gc_distance(lat, lng, cls.lat, cls.lng, math=func)

locations = db.session.query(
        Location,
        Location.distance(53.6209798282177,
                          13.96948162900808).label('distance')).\
    having(column('distance') < 25).\
    order_by('distance').\
    all()

---

Note that the way you use HAVING to eliminate non-group rows is not portable. For example in Postgresql the presence of HAVING clause turns a query in to a grouped query, even without a GROUP BY clause. You could use a subquery instead:

stmt = db.session.query(
        Location,
        Location.distance(53.6209798282177,
                          13.96948162900808).label('distance')).\
    subquery()

location_alias = db.aliased(Location, stmt)

locations = db.session.query(location_alias).\
    filter(stmt.c.distance < 25).\
    order_by(stmt.c.distance).\
    all()