Question

我正在尝试创建一个具有连接之间的关系。以下是我正在尝试做的一个简短示例：

#!/usr/bin/env python
import sqlalchemy as sa
from sqlalchemy import orm
from sqlalchemy.ext.declarative import declarative_base


metadata = sa.MetaData()
Base = declarative_base(metadata=metadata)

engine = sa.create_engine('sqlite:///:memory:')

class Network(Base):
    __tablename__ =  "network"
    id = sa.Column(sa.Integer, primary_key=True)
    ip_net_addr_db = sa.Column('ip_net_addr', sa.Integer, index=True)
    ip_broadcast_addr_db = sa.Column('ip_broadcast_addr', sa.Integer, index=True)
    # This can be determined from the net address and the net mask, but we store
    # it in the db so that we can join with the address table.
    ip_net_mask_len = sa.Column(sa.SmallInteger)

class Address(Base):
    __tablename__ =  "address"
    ip_addr_db = sa.Column('ip_addr', sa.Integer, primary_key=True,
              index=True, unique=True)

Network.addresses = orm.relation(Address, 
    primaryjoin=Address.ip_addr_db.between(
                Network.ip_net_addr_db,
                Network.ip_broadcast_addr_db),
    foreign_keys=[Address.ip_addr_db])    

metadata.create_all(engine) 

Session = orm.sessionmaker(bind=engine)

Network()

如果你运行它，你会收到这个错误：

ArgumentError: Could not determine relation direction for primaryjoin condition 
'address.ip_addr BETWEEN network.ip_net_addr AND network.ip_broadcast_addr', on relation Network.addresses. 
Do the columns in 'foreign_keys' represent only the 'foreign' columns in this join condition ?

这个问题的答案是肯定的，但我无法弄清楚如何告诉它

Answer 1

SQLAlchemy遍历条件以在其中查找本地 - 远程对，以确定关系的列和基数。该算法仅适用于二元运算符。解决问题的简单方法是用两个运算符重写BETWEEN。而不是，他们不是“平等”运算符，所以你不能使用这种关系来附加新地址，这就是使用viewonly=True的原因：

Network.addresses = orm.relation(Address,
    viewonly=True,
    primaryjoin=(
        (Address.ip_addr_db>=Network.ip_net_addr_db) &
        (Address.ip_addr_db<=Network.ip_broadcast_addr_db)
    ),
    foreign_keys=[Address.ip_addr_db]
)

sqlalchemy orm关系的join之间的问题

1 个答案: