sqlalchemy筛选查询中的子级，但不是父级

Question

I have a query which returns a User object. User s have a number of Post s which they have made. When I perform a query, I want to filter the posts found in User.post based on a specific time. For example, to return only the posts within 1 day of a given timestamp.

class User(base):
    __tablename__ = 'User'

class Post(base):
    __tablename__ = 'Post'

    user_id = Column(Integer, ForeignKey('User.uid'))
    user = relationship(User, backref=backref('post')
    time = Column(DateTime, default=func.current_timestamp())

Is there a way to dynamically change the way the children are loaded by the orm, and to specify that at query time? Something like (pseudocode):

User.query.filter_child(Post.time in timespan)

Importantly, I don't want to filter out parents if there are no children which match the criteria. I only want to filter the children loaded by the orm for the given Users.

Answer 1

Loading custom filtered collections is done using contains_eager() . Since the goal is to load all User objects even if they have no Post objects that match the filtering criteria, an outerjoin() should be used and the predicates for posts put in the ON clause of the join:

end = datetime(2019, 4, 10, 12, 0)
start = end - timedelta(days=1)

# If you want to use half closed intervals, replace `between` with
# suitable relational comparisons.
User.query.\
    outerjoin(Post, and_(Post.user_id == User.uid,
                         Post.time.between(start, end))).\
    options(contains_eager(User.post))

Answer 2

I'm not sure this is what you are looking for, but what's wrong with this KISS approach?

Post.query.filter(Post.user == <username>).filter(
    Post.time > START_OF_DAY).filter(Post.time < END_OF_DAY).all()