将SQL转换为SQLAlchemy

Question

c = conn.cursor()
query = 'SELECT status, COUNT(id) FROM bar_baz where not name = 'Foo' and not name = 'Bar' GROUP BY status'
c.execute(query)
print c.fetchall()


class Instance(Base):
    __tablename__ = 'instance'

    id = Column(Integer, primary_key=True)
    name = Column(String)
    status = Column(String)
    type = Column(String)

class Instance2(Base):
    __tablename__ = 'instance_2'

    id = Column(Integer, primary_key=True)
    name = Column(String)
    status = Column(String)
    type = Column(String)
    inc = Column(Integer)

我刚接触Flask的SQLAlchemy；虽然我明白SQLAlchemy可以简化SQL语法，让创建模型变得更简单，但有时候我们希望在前端以特定的方式展示数据。

我有一个查询，希望用SQLAlchemy的session.query和filter来实现，可能还需要分组。

我的查询是：

1. mysql> SELECT status, COUNT(id) FROM bar_baz where not name = 'Foo' and not name = 'Bar' GROUP BY status
2. select (select COUNT(id) FROM instance where not name = 'erf' and not tiername = 'wer' and type='app') as app, (select COUNT(1) FROM instance_2 where not name = 'visq' and not name = 'werf' and type='adc') as adc from dual;

我验证过这些查询在MySQL中可以正常工作；我想知道我们是否有类似的函数可以使用。

感兴趣的查询：

select (select COUNT(id) FROM instance where not name = 'erf' and not tiername = 'wer' and type='app') as app, (select COUNT(1) FROM instance_2 where not name = 'visq' and not name = 'werf' and type='adc') as adc from dual;`

Answer 1

对于第一个查询，使用 db.func.count 来生成计数的表达式。其他的内容应该从 文档 中很明显地看出来。

status_counts = db.session.query(BarBaz.status, db.func.count(BarBaz.id).label('count_id')
).filter(db.not_(db.or_(BarBaz.name == 'Foo', BarBaz.name == 'Bar'))
).group_by(BarBaz.status
).all()

对于第二个查询，使用 subquery() 来生成可以选择的查询。

sub_app = db.session.query(db.func.count(Instance.id).label('app')
).filter(db.not_(db.or_(Instance.name == 'erf', Instance.tiername == 'wer')), Instance.type == 'app'
).subquery()

sub_adc = db.session.query(db.func.count(Instance.id).label('adc')
).filter(db.not_(db.or_(Instance2.name == 'visq', Instance2.name == 'werf')), Instance2.type == 'adc'
).subquery()

out = db.session.query(sub_app.c.app, sub_adc.c.adc).all()