sqlalchemy使用ORM创建VIEW

Question

我创建了以下ORM:

from sqlalchemy import Column, Integer, String, UniqueConstraint
from sqlalchemy.dialects.postgresql import JSONB
from sqlalchemy.ext.declarative import declarative_base
Base = declarative_base()


class TableA(Base):
    __tablename__ = 'table_a'

    id = Column(Integer, primary_key=True, nullable=False)
    identifier = Column(String(320))
    internal_id = Column(Integer)
    type = Column(String(32))
    time = Column(DateTime(timezone=True))
    success = Column(Boolean())
    parameters = Column(JSONB())



class TableB(Base):
    __tablename__ = 'table_b'
    __table_args__ = (UniqueConstraint('generate_action',
                                       'print_action',
                                        name='my_action_key'),)

    id = Column(Integer, primary_key=True, autoincrement=True, nullable=False)
    generate_action = Column(Integer)
    print_action = Column(Integer)
    generate_action = Column(Integer)
    coupon_code = Column(String(300))
    number_of_rebought_items = Column(Integer)
    seconds_between_rebuy = Column(Integer)

我试图弄清楚如何view使用sqlalchemy 将以下原始SQL转换为ORM语法.

CREATE VIEW my_view AS
    SELECT table_b.id as table_b_id,
        tb.coupon_code as coupon_code,
        tb.number_of_rebought_items as number_of_rebought_items,
        ta.id as table_a_action_id,
        ta.time as time,
        ta.parameters as parameters,
    FROM table_b tb
    LEFT JOIN table_a ta on
        ta.id = tb.generate_action;  

找不到任何关于如何使用ORM做的好例子.
到目前为止,我的解决方案是只运行原始sql来创建此视图.

任何人都可以指出我正确的方向,或举例说明如何使用sqlalchemy orm创建视图？

是否可以创建视图 metadata.create_all()

Answer 1

库 sqlalchemy-utils 现在包括创建视图的功能，它将视图与 sqlalchemy 的元数据相关联，以便可以使用创建视图 Base.metadata.create_all

例子：

# installation: pip install sqlalchemy-utils
from sqlalchemy_utils import create_view
from sqlalchemy import select, func

# engine Base & Table declaration elided for brevity

stmt = select([
    TableB.id.label('table_b_id'),
    TableB.coupon_code,
    TableB.number_of_rebought_items,
    TableA.id.label('table_a_action_id'),
    TableA.time,
    TableA.parameters
]).select_from(TableB.__table__.outerjoin(TableA, TableB.generate_action == TableA.id))

# attaches the view to the metadata using the select statement
view = create_view('my_view', stmt, Base.metadata)

# provides an ORM interface to the view
class MyView(Base):
    __table__ = view

# will create all tables & views defined with ``create_view``
Base.metadata.create_all()

# At this point running the following yields 0, as expected,
# indicating that the view has been constructed on the server 
engine.execute(select([func.count('*')], from_obj=MyView)).scalar()