我有两个型号,Foo和Bar。 Bar与Foo有一对多的关系,有一个“foo_id”列和一个“foo”关系。我想查询具有foo_id = 1的Bar行。
据我所知,有两种工作方法可以做到这一点:
query.filter(Bar.foo_id == 1)
query.filter(Bar.foo == instance)
我想要替代这两个更像query.filter(Bar.foo == 1)
的方法 - 也就是说,使用关系列和普通整数而不是实例。目前这已失败,AttributeError: 'int' object has no attribute '_sa_instance_state'
我想避免上述两种方法的原因:
第一种方法在实际应用程序中不可行,因为我不知道这个foo_id
列的名称 - 所有表元数据都是通过反射生成的,列名是一个实现细节我宁愿不依赖(它们目前非常不一致,并且在使用alembic迁移重命名的过程中,因此不希望依赖这些名称)
第二种方法在我能够进行实际查询之前添加额外的SQL查询往返来获取Foo的实例,该实际查询除了我已经拥有的整数ID之外不使用任何其他东西(和身份地图没有以我使用它的方式缓存它。我承认这部分危险地接近过早优化,但仍然感觉很浪费,应该有更好的方法。
另一种表达此问题的方法是如何从RelationshipProperty开始到达外键列,可能使用内省
我也可以使用一种方法来获取某种延迟加载的Foo实例,该实例足以通过ID查询但实际上并不发送SQL查询
以下是一些说明问题的独立测试代码:
from sqlalchemy import create_engine, Column, Integer, ForeignKey
from sqlalchemy.orm import relationship, Session
from sqlalchemy.ext.declarative import declarative_base
def method_1(session, some_foo_id):
"""Works, but i don't actually know foo_id"""
session.query(Bar).filter(Bar.foo_id == some_foo_id).first()
def method_2(session, some_foo_id):
"""Works, but adds a pointless roundtrip"""
some_foo = session.query(Foo).get(some_foo_id)
session.query(Bar).filter(Bar.foo == some_foo).first()
def method_3(session, some_foo_id):
"""Throws an exception, passing int instead of instance"""
session.query(Bar).filter(Bar.foo == some_foo_id).first()
# database setup follows
Base = declarative_base()
class Foo(Base):
__tablename__ = 'foo'
id = Column(Integer, primary_key=True)
value = Column(Integer)
class Bar(Base):
__tablename__ = 'bar'
id = Column(Integer, primary_key=True)
foo_id = Column(Integer, ForeignKey('foo.id'))
foo = relationship(Foo)
if __name__ == '__main__':
engine = create_engine('sqlite://')
Base.metadata.create_all(engine)
session = Session(bind=engine)
foo1, foo2 = Foo(value=1), Foo(value=2)
bar1, bar2 = Bar(foo=foo1), Bar(foo=foo2)
session.add_all([foo1, foo2, bar1, bar2])
session.commit()
engine.echo = True
for fun in [method_1, method_2, method_3]:
print("\n---> %s (%s)\n" % (fun.__name__, fun.__doc__))
fun(session, 1)
session.rollback()
输出:
---> method_1 (Works, but i don't actually know foo_id
BEGIN (implicit)
SELECT bar.id AS bar_id, bar.foo_id AS bar_foo_id
FROM bar
WHERE bar.foo_id = ?
LIMIT ? OFFSET ?
(1, 1, 0)
ROLLBACK
---> method_2 (Works, but adds a pointless roundtrip)
BEGIN (implicit)
SELECT foo.id AS foo_id, foo.value AS foo_value
FROM foo
WHERE foo.id = ?
(1,)
SELECT bar.id AS bar_id, bar.foo_id AS bar_foo_id
FROM bar
WHERE ? = bar.foo_id
LIMIT ? OFFSET ?
(1, 1, 0)
ROLLBACK
---> method_3 (Throws an exception, passing int instead of instance)
Traceback (most recent call last):
File "asd.py", line 51, in <module>
fun(session, 1)
File "asd.py", line 19, in method_3
session.query(Bar).filter(Bar.foo == some_foo_id).first()
File "/usr/lib/python2.7/site-packages/sqlalchemy/sql/operators.py", line 304, in __eq__
return self.operate(eq, other)
File "/usr/lib/python2.7/site-packages/sqlalchemy/orm/attributes.py", line 175, in operate
return op(self.comparator, *other, **kwargs)
File "/usr/lib/python2.7/site-packages/sqlalchemy/orm/relationships.py", line 1042, in __eq__
other, adapt_source=self.adapter))
File "/usr/lib/python2.7/site-packages/sqlalchemy/orm/relationships.py", line 1369, in _optimized_compare
state = attributes.instance_state(state)
AttributeError: 'int' object has no attribute '_sa_instance_state'
答案 0 :(得分:0)
找到一种方法,Bar.foo.prop.local_columns
是RelationshipProperty的一个属性,它返回一个通常有一个项目的集合(对于像这样的简单关系)。集合是无序的,您不能只获得第一个项目,因此list(...)[0]
。完整代码:
def method_4(session, some_foo_id):
foo_col = list(Bar.foo.prop.local_columns)[0]
session.query(Bar).filter(foo_col == some_foo_id).first()