SQLAlchemy is undoubtedly very powerful, but the documentation implicitly assumes lots of prior knowledge and on the subject of relationships, mixes between the backref
and the newly-preferred back_populates()
methods, which I find very confusing.
The following model design is pretty much an exact mirror of the guide in the documentation that deals with the Association Objects for many-to-many relationships. You can see that the comments are still identical to those in the original article, and I've only changed the actual code.
class MatchTeams(db.Model):
match_id = db.Column(db.String, db.ForeignKey('match.id'), primary_key=True)
team_id = db.Column(db.String, db.ForeignKey('team.id'), primary_key=True)
team_score = db.Column(db.Integer, nullable="True")
# bidirectional attribute/collection of "user"/"user_keywords"
match = db.relationship("Match",
backref=db.backref("match_teams",
cascade="all, delete-orphan")
)
# reference to the "Keyword" object
team = db.relationship("Team")
class Match(db.Model):
id = db.Column(db.String, primary_key=True)
# Many side of many to one with Round
round_id = db.Column(db.Integer, ForeignKey('round.id'))
round = db.relationship("Round", back_populates="matches")
# Start of M2M
# association proxy of "match_teams" collection
# to "team" attribute
teams = association_proxy('match_teams', 'team')
def __repr__(self):
return '<Match: %r>' % (self.id)
class Team(db.Model):
id = db.Column(db.Integer, primary_key=True)
name = db.Column(db.String, nullable=False)
goals_for = db.Column(db.Integer)
goals_against = db.Column(db.Integer)
wins = db.Column(db.Integer)
losses = db.Column(db.Integer)
points = db.Column(db.Integer)
matches_played = db.Column(db.Integer)
def __repr__(self):
return '<Team %r with ID: %r>' % (self.name, self.id)
But this snippet, which is supposed to associate the team instance find_liverpool
with the match instance find_match
(both boilerplate objects), doesn't work:
find_liverpool = Team.query.filter(Team.id==1).first()
print(find_liverpool)
find_match = Match.query.filter(Match.id=="123").first()
print(find_match)
find_match.teams.append(find_liverpool)
And outputs the following:
Traceback (most recent call last):
File "/REDACT/temp.py", line 12, in <module>
find_match.teams.append(find_liverpool)
File "/REDACT/lib/python3.4/site-packages/sqlalchemy/ext/associationproxy.py", line 609, in append
item = self._create(value)
File "/REDACT/lib/python3.4/site-packages/sqlalchemy/ext/associationproxy.py", line 532, in _create
return self.creator(value)
TypeError: __init__() takes 1 positional argument but 2 were given
<Team 'Liverpool' with ID: 1>
<Match: '123'>
The call to append
is trying to create a new instance of MatchTeams
, as can be seen from the documentation. This is also noted under "Simplifying Association Objects" that you linked to:
Where above, each
.keywords.append()
operation is equivalent to:
>>> user.user_keywords.append(UserKeyword(Keyword('its_heavy')))
Hence your
find_match.teams.append(find_liverpool)
is equivalent to
find_match.match_teams.append(MatchTeams(find_liverpool))
Since MatchTeams
has no explicitly defined __init__
, it's using the _default_constructor()
as constructor (unless you've overridden it), which accepts only keyword arguments in addition to self
, the only positional argument.
To remedy this either pass a creator
factory to your association proxy:
class Match(db.Model):
teams = association_proxy('match_teams', 'team',
creator=lambda team: MatchTeams(team=team))
or define __init__
on MatchTeams
to suit your needs, for example:
class MatchTeams(db.Model):
# Accepts as positional arguments as well
def __init__(self, team=None, match=None):
self.team = team
self.match = match
or create the association object explicitly:
db.session.add(MatchTeams(match=find_match, team=find_liverpool))
# etc.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With