How to implement "contain any" in flask/sqlalchemy? - flask

Question

The contains operation in SQLAlchemy only accepts one Model object instead of a list of objects. If I want to create a filter that accepts containing any of a group of objects, is there a more SQL-style way than creating multiple filters using contains and combining them with union?
For example, see the following code:
from flask import Flask
from flask.ext.sqlalchemy import SQLAlchemy
app = Flask(__name__)
db = SQLAlchemy(app)
class User(db.Model):
id = db.Column(db.Integer, primary_key = True)
name = db.Column(db.String(100))
son = db.relationship('Son', backref = 'parent', lazy = 'dynamic')
def __repr__(self):
return '<"%s">' % self.name
class Son(db.Model):
id = db.Column(db.Integer, primary_key = True)
name = db.Column(db.String(1000))
user_id = db.Column(db.Integer, db.ForeignKey('user.id'))
def __repr__(self):
return '<"%s">' % self.name
db.create_all()
son1 = Son(name = '1 a 1')
son2 = Son(name = '1 b 1')
son3 = Son(name = '1 c 1')
son4 = Son(name = '2 a 2')
son5 = Son(name = '2 b 2')
user0 = User(name = 'w1')
user1 = User(name = 'w2')
user2 = User(name = 'w3')
user0.son.append(son1)
user0.son.append(son2)
user1.son.append(son3)
user1.son.append(son4)
user2.son.append(son5)
db.session.add(son1)
db.session.add(son2)
db.session.add(son3)
db.session.add(son4)
db.session.add(son5)
db.session.add(user0)
db.session.add(user1)
db.session.add(user2)
db.session.commit()
son_query = Son.query.filter(Son.name.ilike('%a%'))
son_query_all = son_query.all()
print son_query.all()
user_query = User.query.filter(User.son.contains(son_query_all[0])).union(*[User.query.filter(User.son.contains(query)) for query in son_query_all[1:]])
print user_query.all()
The example firstly creates two models: User and Son, and then creates 3 User instances and 5 Son instances. user0 contains son1 and son2, user1 contains son3 and son4, and user2 contains son5. Note that the name of son1 and son4 are both like %a%. Now I want to select all User instances containing Son instances whose name likes %a%.
The current method is to select all Son instances in son_query_all, and then selects User instances containing individual desired Son instances, and then combines the selecting result using union. Is there a more SQL-style way for SQLAlchemy to select the same? For example, is there anything like contains_any so that the last query can be changed into something like
user_query = User.query.filter(User.son.contains_any(son_query_all))
Note that of course I can define a custom contains_any function for the same purpose using the union and contains operation. My question is whether there is a more efficient way than simply union all contains-ed?

Answer 1

The correct way to solve this kind of filtering is to use JOIN. Basically you join Son table with User table and filter by joined entity's field.
In SQLAlchemy you can express it like this:
User.query.join(Son).filter(Son.name.ilike('%a%'))
And it will produce following SQL:
SELECT * FROM user
JOIN son ON user.id = son.user_id
WHERE lower(son.name) LIKE lower('%a%')