更新sqlalchemy中具有计算列的表

时间:2013-05-06 12:08:37

标签: python orm sqlalchemy

我正在使用SQLalchemy在遗留的MSSQL数据库中工作,我有一个声明性映射 此数据库有几个具有计算列的表。我可以读得很好,但(当然)写入计算列不起作用。但是,当我创建并尝试保存ORM对象时,SQLAlchemy仍尝试在这些列中保存“无”值,从而导致错误。

我在网上发现了一些应该通过使用SQLAlchemy的hybrid_property装饰器使特定列“只读”的示例,但即使在实现之后,仍会出现相同的错误(The column "tlog_real_timehh" cannot be modified because it is either a computed column or is the result of a UNION operator.)。

代码如下 - 映射:

class transactionlog(Base):
    __tablename__ = 'transactionlog'
    tlog_id = Column(VARCHAR(length=36), primary_key=True, nullable=False)
    tlog_ppl_id = Column(VARCHAR(length=36), ForeignKey('people.ppl_id'))
    tlog_evtt_id = Column(VARCHAR(length=5))
    tlog_testrun = Column(BIT())
    tlog_Data = Column(NVARCHAR(length=300))
    tlog_price = Column(DECIMAL(precision=18, scale=2))
    tlog_comment = Column(NVARCHAR(length=1000))
    _tlog_real_timehh = Column('tlog_real_timehh', INTEGER())
    _tlog_real_timemm = Column('tlog_real_timemm', INTEGER())
    _tlog_real_timess = Column('tlog_real_timess', INTEGER())
    _tlog_fin_booking = Column('tlog_fin_booking', BIT())

    @hybrid_property
    def tlog_real_timehh(self):
        return self._tlog_real_timehh

    @tlog_real_timehh.setter
    def tlog_real_timehh(self, tlog_real_timehh):
        self._tlog_real_timehh = tlog_real_timehh

    @hybrid_property
    def tlog_real_timemm(self):
        return self._tlog_real_timemm

    @tlog_real_timemm.setter
    def tlog_real_timemm(self, tlog_real_timemm):
        self._tlog_real_timemm = tlog_real_timemm

    @hybrid_property
    def tlog_real_timess(self):
        return self._tlog_real_timess

    @tlog_real_timess.setter
    def tlog_real_timess(self, tlog_real_timess):
        self._tlog_real_timess = tlog_real_timess

    @hybrid_property
    def tlog_fin_booking(self):
        return self._tlog_fin_booking

    @tlog_fin_booking.setter
    def tlog_fin_booking(self, tlog_fin_booking):
        self._tlog_fin_booking = tlog_fin_booking

以及应添加新记录的代码:

    rem = Transactionlog()
    rem.tlog_testrun = 0
    rem.tlog_evtt_id = 'rem'
    rem.tlog_Data = None
    rem.tlog_comment = 'reminder'
    rem.tlog_price = 0
    db.session.add(rem)
    db.session.flush()

我希望hybrid_property代码使计算字段成为只读,但似乎SQLAlchemy仍然试图根据映射代码在INSERT语句中填充它们。 (当我查看SQL语句时,我可以看到这一点。我无法发布SQL语句,因为我在某种程度上将对象缩写为StackOverflow上没有任何敏感数据)。

问题是,为什么SQLAlchemy仍然尝试为tlog_real_timehh,tlog_real_timemm,tlog_real_timess和tlog_fin_booking插入值,我该如何防止这种情况?

感谢您给我的任何指示。
埃里克

1 个答案:

答案 0 :(得分:3)

使用server-generated columns标记FetchedValue

from sqlalchemy import *
from sqlalchemy.orm import *
from sqlalchemy.ext.declarative import declarative_base

Base = declarative_base()

class A(Base):
    __tablename__ = 'a'

    id = Column(Integer, autoincrement=False, primary_key=True)
    firstname = Column(String(50))
    lastname = Column(String(50))
    fullname = Column(String(100), FetchedValue())

e = create_engine("mssql+pyodbc://scott:tiger@ms_2005", echo=True)
Base.metadata.drop_all(e)

e.execute("""
    CREATE TABLE a (
            id INTEGER PRIMARY KEY,
            firstname VARCHAR(50),
            lastname VARCHAR(50)
        )
""")
e.execute("ALTER TABLE a ADD fullname AS firstname + ' ' + lastname")

sess = Session(e)

sess.add_all([
    A(id=1, firstname='ed', lastname='jones'),
    A(id=2, firstname='wendy', lastname='smith'),
    A(id=3, firstname='jack', lastname='bean')
])
sess.commit()

assert [
    fname for fname, in
    sess.query(A.fullname).order_by(A.id)
] == ['ed jones', 'wendy smith', 'jack bean']


e.execute("DROP TABLE a")