SQLAlchemy 连接来自同一个表的多个列

Posted 2023-03-23

【中文标题】SQLAlchemy 连接来自同一个表的多个列

【英文标题】：SQL Alchemy Join Multiple Columns from same table

【发布时间】：2016-02-10 23:01:08

【问题描述】：

class Match(Base):
    __tablename__ = 'matches'

    id = Column(Integer, primary_key=True)
    date = Column(Date, nullable=False)
    time = Column(Time, nullable=True)
    league_id = Column(ForeignKey('leagues.id'), nullable=False, index=True)
    league = relationship('League', backref='matches')
    type = Column(enums.matches_types)
    home_team_id = Column(ForeignKey('teams.id'), nullable=False, index=True)
    home_team = relationship('Team', foreign_keys=[home_team_id], backref='home_matches')
    away_team_id = Column(ForeignKey('teams.id'), nullable=False, index=True)


class Team(Base):
    __tablename__ = 'teams'

    id = Column(Integer, primary_key=True)
    name = Column(String, nullable=False)
    country_id = Column(ForeignKey('countries.id'), nullable=False, index=True)
    country = relationship('Country', backref='teams')

我需要编写一个查询来连接列和团队表，以显示本地和客队的团队信息。

Session.query(Match.date, Match.home_team.name, Match_away_team.name).joins(Team)

这会返回Can't determine join between 'matches' and 'teams'; tables have more than one foreign key constraint relationship between them. Please specify the 'onclause' of this join explicitly

【参考方案1】：

首先，您的代码不起作用的原因是因为 SQLAlchemy 不知道您是想通过home_team 还是away_team 加入Team，所以您必须告诉它。此外，您还需要加入Team 两次，这使事情变得更加复杂。

使用joinedload 可以更轻松地完成此操作：

matches = session.query(Match).options(joinedload(Match.home_team),
                                       joinedload(Match.away_team))
for m in matches:
    print m.date, m.home_team, m.away_team

m.home_team 和 m.away_team 将使用 JOIN 在与 m 相同的查询中加载。

如果您坚持使用显式 .join()，则必须使用 alias Team 实体（未测试）：

home = aliased(Team)
away = aliased(Team)
q = session.query(Match.date, home, away).join(home, Match.home_team) \
                                         .join(away, Match.away_team)
for date, home_team, away_team in q:
    print date, home_team, away_team

【讨论】：

这是如何使用它的完美说明。我测试了你后面的例子，效果很好。