Question

我想与Flask和SQLAlchemy一起使用显式的master-master数据库设置，希望Flask-SQLAlchemy支持这一点。

我希望能够执行类似下面的代码片段，但我不确定Flask-SQLAlchemy是否支持

from flask import Flask
from flask.ext.sqlalchemy import SQLAlchemy

app = Flask(__name__)
SQLALCHEMY_DATABASE_URI = 'default_DB_uri'
SQLALCHEMY_BINDS = { 'master1':'first_master_DB_uri', 'master2': 'second_master_DB_uri' }

app.config['SQLALCHEMY_DATABASE_URI'] = SQLALCHEMY_DATABASE_URI
app.config['SQLALCHEMY_BINDS'] = SQLALCHEMY_BINDS

db = SQLAlchemy(app)

@app.route('/some_endpoint')
def some_endpoint():
    # read some data for the default DB
    readData = db.session.query('select ...')

    m = SomeModel()
    masterSession1 = db.session(bind='master1')
    # persist the data in m into the first master
    masterSession1.add(m)

    masterSession2 = db.session(bind='master2')
    # persist the data into the second master
    masterSession2.add(m)

    return "some return value"

有没有办法使用Flask-SQLAlchemy实现这一点并绑定？我想Flask-SQLAlchemy已经使用绑定处理了多个引擎但是我看不到如何使用它来进行显式数据库选择而不是像这里提到的基于模型的选择：http://pythonhosted.org/Flask-SQLAlchemy/binds.html

感谢您的帮助。

Answer 1

以下代码是我最终拥有此功能的代码。

一些注意事项：

我更改get_table_for_bind以将所有表绑定到所有绑定而没有明确的__bind_key__。这样做是为了能够调用db.create_all()或db.drop_all()并在所有DB中创建/删除表。为了使其工作而不中断默认数据库选择，在未指定特定绑定时，get_binds已更改为在原始实现后再次映射None绑定，以覆盖表 - ＆gt;绑定映射。
如果您未指定using_bind，则应使用默认数据库。
SQLAlchemy映射对象保留对会话和状态的引用，因此您无法将同一对象添加到两个DB。我在添加对象之前制作了该对象的副本，以便将其保存在两个DB中。不确定是否有更好的方法来做到这一点。
我还没有对此进行全面测试，这可能会破坏我未使用或不知道的其他功能。

flask-sqlalchemy覆盖：

from flask_sqlalchemy import SQLAlchemy, SignallingSession, get_state
from flask_sqlalchemy._compat import itervalues

class UsingBindSignallingSession(SignallingSession):
    def get_bind(self, mapper=None, clause=None):
        if self._name:
            _eng = get_state(self.app).db.get_engine(self.app,bind=self._name)
            return _eng
        else:
            return super(UsingBindSignallingSession, self).get_bind(mapper, clause)

    _name = None
    def using_bind(self, name):
        self._name = name
        return self

class UsingBindSQLAlchemy(SQLAlchemy):
    def create_session(self, options):
        return UsingBindSignallingSession(self, **options)

    def get_binds(self, app=None):
        retval = super(UsingBindSQLAlchemy, self).get_binds(app)
        # get the binds for None again in order to make sure that it is the default bind for tables
        # without an explicit bind
        bind = None
        engine = self.get_engine(app, bind)
        tables = self.get_tables_for_bind(bind)
        retval.update(dict((table, engine) for table in tables))
        return retval

    def get_tables_for_bind(self, bind=None):
        """Returns a list of all tables relevant for a bind.
        Tables without an explicit __bind_key__ will be bound to all binds.
        """
        result = []
        for table in itervalues(self.Model.metadata.tables):
            # if we don't have an explicit __bind_key__ bind this table to all databases
            if table.info.get('bind_key') == bind or table.info.get('bind_key') == None:
                result.append(table)
        return result

db = UsingBindSQLAlchemy()

现在你可以这样做：

# This is the default DB
SQLALCHEMY_DATABASE_URI=YOUR_MAIN_DB_URI_CONNECT_STRING
# Master1 and Master2
SQLALCHEMY_BINDS = { 'master1':YOUR_MASTER1_DB_URI_CONNECT_STRING, 'master2':YOUR_MASTER2_DB_URI_CONNECT_STRING }

# Tables without __bind_key__ will be dropped/created on all DBs (default, master1, master2)
db.drop_all()
db.create_all()

s = db.session().using_bind('master1')
s.add(SOME_OBJECT)
s.commit()

s = db.session().using_bind('master2')
s.add(SOME_OBJECT_CLONE) # a clone of the original object, before the first add()
s.commit()

# and the default DB, as always
db.session.add(SOME_OTHER_OBJECT)
db.session.commit()

使用Flask和SQLAlchemy进行显式Master-Master DB设置，希望使用Flask-SQLAlchemy

1 个答案:

flask-sqlalchemy覆盖：

现在你可以这样做：