使用 Flask 中的 SQLAlchemy 会话会引发“在线程中创建的 SQLite 对象只能在同一线程中使用”

Question

我有一个 Flask 视图，它使用 SQLAlchemy 来查询和显示一些博客文章。我正在使用 mod_wsgi 运行我的应用程序。此视图在我第一次访问该页面时有效，但下次返回 500 错误。回溯显示错误ProgrammingError: SQLite objects created in a thread can only be used in that same thread. 为什么会出现此错误以及如何解决？

views.py

engine = create_engine('sqlite:////var/www/homepage/blog.db')
Base.metadata.bind = engine
DBSession = sessionmaker(bind = engine)
session = DBSession()

@app.route('/blog')
@app.route('/blog.html')
def blog():
    entrys = session.query(Entry).order_by(desc(Entry.timestamp)).all()
    return render_template('blog.html', blog_entrys = entrys)

models.py:

class Entry(Base):
    __tablename__ = 'entry'

    id = Column(Integer, primary_key = True)

    title = Column(String(100), nullable = False)
    body = Column(String, nullable = False)
    timestamp = Column(DateTime, nullable = False)
    featured = Column(Boolean, nullable = False)

    comments = relationship('Comment')

    def is_featured(self):
        return self.featured


class Comment(Base):
    __tablename__ = 'comment'

    id = Column(Integer, primary_key = True)
    entry_id = Column(Integer, ForeignKey('entry.id'))

    text = Column(String(500), nullable = False)
    name = Column(String(80))


engine = create_engine('sqlite:////var/www/homepage/blog.db')
Base.metadata.create_all(engine)

Exception on /blog.html [GET]
Traceback (most recent call last):
  File "/usr/lib/python2.6/dist-packages/flask/app.py", line 861, in wsgi_app
    rv = self.dispatch_request()
  File "/usr/lib/python2.6/dist-packages/flask/app.py", line 696, in dispatch_request
    return self.view_functions[rule.endpoint](**req.view_args)
  File "/var/www/homepage/webserver.py", line 38, in blog
    entrys = session.query(Entry).order_by(desc(Entry.timestamp)).all()
  File "/usr/lib/python2.6/dist-packages/sqlalchemy/orm/query.py", line 1453, in all
    return list(self)
  File "/usr/lib/python2.6/dist-packages/sqlalchemy/orm/query.py", line 1565, in __iter__
    return self._execute_and_instances(context)
  File "/usr/lib/python2.6/dist-packages/sqlalchemy/orm/query.py", line 1570, in _execute_and_instances
    mapper=self._mapper_zero_or_none())
  File "/usr/lib/python2.6/dist-packages/sqlalchemy/orm/session.py", line 735, in execute
    clause, params or {})
  File "/usr/lib/python2.6/dist-packages/sqlalchemy/engine/base.py", line 1157, in execute
    params)
  File "/usr/lib/python2.6/dist-packages/sqlalchemy/engine/base.py", line 1235, in _execute_clauseelement
    parameters=params
  File "/usr/lib/python2.6/dist-packages/sqlalchemy/engine/base.py", line 1348, in __create_execution_context
    None, None)
  File "/usr/lib/python2.6/dist-packages/sqlalchemy/engine/base.py", line 1343, in __create_execution_context
    connection=self, **kwargs)
  File "/usr/lib/python2.6/dist-packages/sqlalchemy/engine/default.py", line 381, in __init__
    self.cursor = self.create_cursor()
  File "/usr/lib/python2.6/dist-packages/sqlalchemy/engine/default.py", line 523, in create_cursor
    return self._connection.connection.cursor()
  File "/usr/lib/python2.6/dist-packages/sqlalchemy/pool.py", line 383, in cursor
    c = self.connection.cursor(*args, **kwargs)
ProgrammingError: (ProgrammingError) SQLite objects created in a thread can only be used in that same thread.The object was created in thread id 140244498364160 and this is thread id 140244523542272 None [{}]

Answer 1

从this SO answer 得到提示，我搜索了 SA 文档，发现您可以这样做：

engine = create_engine('sqlite:////var/www/homepage/blog.db?check_same_thread=False')

scoped_session 并不适合我的情况，因为 Flask-SQLAlchemy 只接受连接字符串参数：

from flask import Flask
from flask_sqlalchemy import SQLAlchemy


class Config(object):
    SQLALCHEMY_DATABASE_URI = 'sqlite:///app.db?check_same_thread=False'


db = SQLAlchemy()


def create_app():
    app.config.from_object(Config)
    app = Flask(__name__)
    db.init_app(app)
    ...

根据sqlite3.connect：

默认情况下，check_same_thread 是True，只有创建线程可以 使用连接。如果设置False，返回的连接可能是 跨多个线程共享。 当使用多个线程时 用户应将相同的连接写入操作序列化到 避免数据损坏。

Answer 2

如果您跨线程共享会话，SQLAlchemy（在这种情况下也是 SQLite）不起作用。您可能没有明确使用线程，但mod_wsgi 是，并且您已经定义了一个全局session 对象。使用scoped_session 为每个线程创建一个唯一的会话。

session = scoped_session(sessionmaker(bind=engine))

@app.teardown_request
def remove_session(ex=None):
    session.remove()

@app.route('/')
def example():
    item = session.query(MyModel).filter(...).all()
    ...

最好使用Flask-SQLAlchemy，它会为您处理这些事情和其他事情。 SQLAlchemy 文档建议您使用集成库，而不是自己这样做。

db = SQLAlchemy(app)

@app.route('/')
def example():
    item = db.session.query(MyModel).filter(...).all()
    ...

还请注意，您应该只定义一次引擎、会话等并将其导入其他地方，而不是像当前代码那样在每个文件中重新定义它。

Answer 3

我在使用 SQLAlchemy + SQLite + FastAPI 时遇到过类似的问题，我添加了参数
connect_args={"check_same_thread": False}

完整代码：

#SqlAlchemy Setup 
SQLALCHEMY_DATABASE_URL = 'sqlite+pysqlite:///.db.sqlite3:' engine = create_engine(SQLALCHEMY_DATABASE_URL,
                           connect_args={"check_same_thread": False},
                           echo=True,
                           future=True
                           ) SessionLocal = sessionmaker(autocommit=False, autoflush=False, bind=engine) Base = declarative_base()