MQLAlchemy

Query SQLAlchemy models using MongoDB style syntax.

Why?

The need arose for me to be able to pass complex database filters from client side JavaScript to a Python server. I started building some JSON style syntax to do so, then realized such a thing already existed. I’ve never seriously used MongoDB, but the syntax for querying lends itself pretty perfectly to this use case.

That sounds pretty dangerous…

It can be. When using this with any sort of user input, you’ll want to pass in a whitelist of attributes that are ok to query, as well as any required filters for each model class, otherwise you’ll open the possibility of leaked passwords and all sorts of other scary stuff.

How fast is it?

The time it takes to parse should be minimal compared to the actual database query, so this shouldn’t slow your queries down noticably.

Supported Operators

$and
$or
$not
$nor
$in
$nin
$gt
$gte
$lt
$lte
$ne
$mod
$exists

Custom operators added for convenience:

$eq - Explicit equality check.
$like - Search a text field for the given value.

Not yet supported, but would like to add:

Index based relation queries. Album.tracks.0.track_id won’t work.
$regex

Examples

from sqlalchemy import create_engine
from sqlalchemy.orm import sessionmaker
from mqlalchemy import apply_mql_filters
from myapp.mymodels import Album

# get your sqlalchemy db session here
db_engine = create_engine("sqlite+pysqlite:///mydb.sqlite")
DBSession = sessionmaker(bind=db_engine)
db_session = DBSession()

# define which fields of Album are ok to query
whitelist = ["album_id", "artist.name", "tracks.playlists.name"]
# Find all albums that are either by Led Zeppelin or have a track
# that can be found on the "Grunge" playlist.
filters = {
    "$or": [
        {"tracks.playlists.name": "Grunge"},
        {"artist.name": "Led Zeppelin"}
    ]
}
query = select(Album)
query = apply_mql_filters(
    model_class=Album,
    query=select(Album),
    filters=filters,
    whitelist=whitelist)
matching_records = db_session.execute(query).scalars().all()

For more, please see the included tests, as they’re probably the easiest way to get an idea of how the library can be used.

Contributing

Submit a pull request and make sure to include an updated AUTHORS with your name along with an updated CHANGES.rst.

License

MIT