问题
How do I automatically map all of my fields of the forms my_column
to corresponding columns MyColumn
in the database?
class MyTable(Base):
__tablename__ = 'MyTable'
# what I have now:
my_column_1 = Column('MyColumn1', Integer)
# what I want: how to configure a convention mapping rule for this:
my_column_1 = Column(Integer)
回答1:
You could provide your own metaclass that renames columns, before passing control to DeclarativeMeta
. Another option would be to alter column names in a before_parent_attach event handler for Column, but that would affect all columns globally, compared to just the ones in mapped classes using a specific declarative base.
from sqlalchemy.ext.declarative.api import DeclarativeMeta
from sqlalchemy import Column
class RenamingDeclarativeMeta(DeclarativeMeta):
def __init__(cls, name, bases, namespace):
_rename_declared_columns(namespace)
super().__init__(name, bases, namespace)
def __setattr__(cls, key, value):
if isinstance(value, Column):
_undefer_column_name_only(key, value)
super().__setattr__(key, value)
def to_camelcase(s):
return ''.join([w.title() for w in s.split('_')])
def _undefer_column_name_only(key, column):
if column.name is None:
column.name = to_camelcase(key)
def _rename_declared_columns(namespace):
for key, attr in namespace.items():
if isinstance(attr, Column):
_undefer_column_name_only(key, attr)
This omits some functionality, such as renaming columns contained in ColumnProperty
and CompositeProperty
attributes. In order to use it pass it as the metaclass to declarative_base():
In [2]: from sqlalchemy.ext.declarative import declarative_base
In [3]: Base = declarative_base(metaclass=RenamingDeclarativeMeta)
In [4]: from sqlalchemy import Column, Integer
In [5]: class Foo(Base):
...: __tablename__ = 'foo'
...: id = Column(Integer, primary_key=True)
...: bar_baz_1 = Column(Integer, nullable=False)
...:
In [6]: Foo.id
Out[6]: <sqlalchemy.orm.attributes.InstrumentedAttribute at 0x7ffb7f770e60>
In [7]: Foo.bar_baz_1
Out[7]: <sqlalchemy.orm.attributes.InstrumentedAttribute at 0x7ffb7dc47200>
In [8]: list(Foo.__table__.columns)
Out[8]:
[Column('Id', Integer(), table=<foo>, key='id', primary_key=True, nullable=False),
Column('BarBaz1', Integer(), table=<foo>, key='bar_baz_1', nullable=False)]
The other approach would be to globally rename all columns that are attached to Table objects using the event API:
from sqlalchemy import event
@event.listens_for(Column, 'before_parent_attach')
def receive_column_before_table_attach(column, table):
# As dumb as it gets. Blindly modifies all columns.
column.name = to_camelcase(column.name)
This has a major drawback: it will blindly rename columns in reflected tables as well, which is why I'd recommend using the metaclass approach. It's better contained and is less surprising.
来源:https://stackoverflow.com/questions/51886972/sqlalchemy-orm-column-name-convention