# -*- coding: utf-8 -*- """ flaskext.sqlalchemy ~~~~~~~~~~~~~~~~~~~ Adds basic SQLAlchemy support to your application. :copyright: (c) 2014 by Armin Ronacher, Daniel Neuhäuser. :license: BSD, see LICENSE for more details. """ from __future__ import with_statement, absolute_import import os import re import sys import time import functools import warnings import sqlalchemy from math import ceil from functools import partial from flask import _request_ctx_stack, abort, has_request_context, request from flask.signals import Namespace from operator import itemgetter from threading import Lock from sqlalchemy import orm, event, inspect from sqlalchemy.orm.exc import UnmappedClassError from sqlalchemy.orm.session import Session as SessionBase from sqlalchemy.engine.url import make_url from sqlalchemy.ext.declarative import declarative_base, DeclarativeMeta from flask_sqlalchemy._compat import iteritems, itervalues, xrange, \ string_types # the best timer function for the platform if sys.platform == 'win32': _timer = time.clock else: _timer = time.time try: from flask import _app_ctx_stack except ImportError: _app_ctx_stack = None __version__ = '2.1' # Which stack should we use? _app_ctx_stack is new in 0.9 connection_stack = _app_ctx_stack or _request_ctx_stack _camelcase_re = re.compile(r'([A-Z]+)(?=[a-z0-9])') _signals = Namespace() models_committed = _signals.signal('models-committed') before_models_committed = _signals.signal('before-models-committed') def _make_table(db): def _make_table(*args, **kwargs): if len(args) > 1 and isinstance(args[1], db.Column): args = (args[0], db.metadata) + args[1:] info = kwargs.pop('info', None) or {} info.setdefault('bind_key', None) kwargs['info'] = info return sqlalchemy.Table(*args, **kwargs) return _make_table def _set_default_query_class(d): if 'query_class' not in d: d['query_class'] = BaseQuery def _wrap_with_default_query_class(fn): @functools.wraps(fn) def newfn(*args, **kwargs): _set_default_query_class(kwargs) if "backref" in kwargs: backref = kwargs['backref'] if isinstance(backref, string_types): backref = (backref, {}) _set_default_query_class(backref[1]) return fn(*args, **kwargs) return newfn def _include_sqlalchemy(obj): for module in sqlalchemy, sqlalchemy.orm: for key in module.__all__: if not hasattr(obj, key): setattr(obj, key, getattr(module, key)) # Note: obj.Table does not attempt to be a SQLAlchemy Table class. obj.Table = _make_table(obj) obj.relationship = _wrap_with_default_query_class(obj.relationship) obj.relation = _wrap_with_default_query_class(obj.relation) obj.dynamic_loader = _wrap_with_default_query_class(obj.dynamic_loader) obj.event = event class _DebugQueryTuple(tuple): statement = property(itemgetter(0)) parameters = property(itemgetter(1)) start_time = property(itemgetter(2)) end_time = property(itemgetter(3)) context = property(itemgetter(4)) @property def duration(self): return self.end_time - self.start_time def __repr__(self): return '' % ( self.statement, self.parameters, self.duration ) def _calling_context(app_path): frm = sys._getframe(1) while frm.f_back is not None: name = frm.f_globals.get('__name__') if name and (name == app_path or name.startswith(app_path + '.')): funcname = frm.f_code.co_name return '%s:%s (%s)' % ( frm.f_code.co_filename, frm.f_lineno, funcname ) frm = frm.f_back return '' class SignallingSession(SessionBase): """The signalling session is the default session that Flask-SQLAlchemy uses. It extends the default session system with bind selection and modification tracking. If you want to use a different session you can override the :meth:`SQLAlchemy.create_session` function. .. versionadded:: 2.0 .. versionadded:: 2.1 The `binds` option was added, which allows a session to be joined to an external transaction. """ def __init__(self, db, autocommit=False, autoflush=True, app=None, **options): #: The application that this session belongs to. self.app = app = db.get_app() track_modifications = app.config['SQLALCHEMY_TRACK_MODIFICATIONS'] bind = options.pop('bind', None) or db.engine binds = options.pop('binds', None) or db.get_binds(app) if track_modifications is None or track_modifications: _SessionSignalEvents.register(self) SessionBase.__init__( self, autocommit=autocommit, autoflush=autoflush, bind=bind, binds=binds, **options ) def get_bind(self, mapper=None, clause=None): # mapper is None if someone tries to just get a connection if mapper is not None: info = getattr(mapper.mapped_table, 'info', {}) bind_key = info.get('bind_key') if bind_key is not None: state = get_state(self.app) return state.db.get_engine(self.app, bind=bind_key) return SessionBase.get_bind(self, mapper, clause) class _SessionSignalEvents(object): @classmethod def register(cls, session): if not hasattr(session, '_model_changes'): session._model_changes = {} event.listen(session, 'before_flush', cls.record_ops) event.listen(session, 'before_commit', cls.record_ops) event.listen(session, 'before_commit', cls.before_commit) event.listen(session, 'after_commit', cls.after_commit) event.listen(session, 'after_rollback', cls.after_rollback) @classmethod def unregister(cls, session): if hasattr(session, '_model_changes'): del session._model_changes event.remove(session, 'before_flush', cls.record_ops) event.remove(session, 'before_commit', cls.record_ops) event.remove(session, 'before_commit', cls.before_commit) event.remove(session, 'after_commit', cls.after_commit) event.remove(session, 'after_rollback', cls.after_rollback) @staticmethod def record_ops(session, flush_context=None, instances=None): try: d = session._model_changes except AttributeError: return for targets, operation in ((session.new, 'insert'), (session.dirty, 'update'), (session.deleted, 'delete')): for target in targets: state = inspect(target) key = state.identity_key if state.has_identity else id(target) d[key] = (target, operation) @staticmethod def before_commit(session): try: d = session._model_changes except AttributeError: return if d: before_models_committed.send(session.app, changes=list(d.values())) @staticmethod def after_commit(session): try: d = session._model_changes except AttributeError: return if d: models_committed.send(session.app, changes=list(d.values())) d.clear() @staticmethod def after_rollback(session): try: d = session._model_changes except AttributeError: return d.clear() class _EngineDebuggingSignalEvents(object): """Sets up handlers for two events that let us track the execution time of queries.""" def __init__(self, engine, import_name): self.engine = engine self.app_package = import_name def register(self): event.listen(self.engine, 'before_cursor_execute', self.before_cursor_execute) event.listen(self.engine, 'after_cursor_execute', self.after_cursor_execute) def before_cursor_execute(self, conn, cursor, statement, parameters, context, executemany): if connection_stack.top is not None: context._query_start_time = _timer() def after_cursor_execute(self, conn, cursor, statement, parameters, context, executemany): ctx = connection_stack.top if ctx is not None: queries = getattr(ctx, 'sqlalchemy_queries', None) if queries is None: queries = [] setattr(ctx, 'sqlalchemy_queries', queries) queries.append(_DebugQueryTuple(( statement, parameters, context._query_start_time, _timer(), _calling_context(self.app_package)))) def get_debug_queries(): """In debug mode Flask-SQLAlchemy will log all the SQL queries sent to the database. This information is available until the end of request which makes it possible to easily ensure that the SQL generated is the one expected on errors or in unittesting. If you don't want to enable the DEBUG mode for your unittests you can also enable the query recording by setting the ``'SQLALCHEMY_RECORD_QUERIES'`` config variable to `True`. This is automatically enabled if Flask is in testing mode. The value returned will be a list of named tuples with the following attributes: `statement` The SQL statement issued `parameters` The parameters for the SQL statement `start_time` / `end_time` Time the query started / the results arrived. Please keep in mind that the timer function used depends on your platform. These values are only useful for sorting or comparing. They do not necessarily represent an absolute timestamp. `duration` Time the query took in seconds `context` A string giving a rough estimation of where in your application query was issued. The exact format is undefined so don't try to reconstruct filename or function name. """ return getattr(connection_stack.top, 'sqlalchemy_queries', []) class Pagination(object): """Internal helper class returned by :meth:`BaseQuery.paginate`. You can also construct it from any other SQLAlchemy query object if you are working with other libraries. Additionally it is possible to pass `None` as query object in which case the :meth:`prev` and :meth:`next` will no longer work. """ def __init__(self, query, page, per_page, total, items): #: the unlimited query object that was used to create this #: pagination object. self.query = query #: the current page number (1 indexed) self.page = page #: the number of items to be displayed on a page. self.per_page = per_page #: the total number of items matching the query self.total = total #: the items for the current page self.items = items @property def pages(self): """The total number of pages""" if self.per_page == 0: pages = 0 else: pages = int(ceil(self.total / float(self.per_page))) return pages def prev(self, error_out=False): """Returns a :class:`Pagination` object for the previous page.""" assert self.query is not None, 'a query object is required ' \ 'for this method to work' return self.query.paginate(self.page - 1, self.per_page, error_out) @property def prev_num(self): """Number of the previous page.""" return self.page - 1 @property def has_prev(self): """True if a previous page exists""" return self.page > 1 def next(self, error_out=False): """Returns a :class:`Pagination` object for the next page.""" assert self.query is not None, 'a query object is required ' \ 'for this method to work' return self.query.paginate(self.page + 1, self.per_page, error_out) @property def has_next(self): """True if a next page exists.""" return self.page < self.pages @property def next_num(self): """Number of the next page""" return self.page + 1 def iter_pages(self, left_edge=2, left_current=2, right_current=5, right_edge=2): """Iterates over the page numbers in the pagination. The four parameters control the thresholds how many numbers should be produced from the sides. Skipped page numbers are represented as `None`. This is how you could render such a pagination in the templates: .. sourcecode:: html+jinja {% macro render_pagination(pagination, endpoint) %} {% endmacro %} """ last = 0 for num in xrange(1, self.pages + 1): if num <= left_edge or \ (num > self.page - left_current - 1 and \ num < self.page + right_current) or \ num > self.pages - right_edge: if last + 1 != num: yield None yield num last = num class BaseQuery(orm.Query): """The default query object used for models, and exposed as :attr:`~SQLAlchemy.Query`. This can be subclassed and replaced for individual models by setting the :attr:`~Model.query_class` attribute. This is a subclass of a standard SQLAlchemy :class:`~sqlalchemy.orm.query.Query` class and has all the methods of a standard query as well. """ def get_or_404(self, ident): """Like :meth:`get` but aborts with 404 if not found instead of returning `None`. """ rv = self.get(ident) if rv is None: abort(404) return rv def first_or_404(self): """Like :meth:`first` but aborts with 404 if not found instead of returning `None`. """ rv = self.first() if rv is None: abort(404) return rv def paginate(self, page=None, per_page=None, error_out=True): """Returns `per_page` items from page `page`. By default it will abort with 404 if no items were found and the page was larger than 1. This behavor can be disabled by setting `error_out` to `False`. If page or per_page are None, they will be retrieved from the request query. If the values are not ints and ``error_out`` is true, it will abort with 404. If there is no request or they aren't in the query, they default to page 1 and 20 respectively. Returns an :class:`Pagination` object. """ if has_request_context(): if page is None: try: page = int(request.args.get('page', 1)) except (TypeError, ValueError): if error_out: abort(404) page = 1 if per_page is None: try: per_page = int(request.args.get('per_page', 20)) except (TypeError, ValueError): if error_out: abort(404) per_page = 20 else: if page is None: page = 1 if per_page is None: per_page = 20 if error_out and page < 1: abort(404) items = self.limit(per_page).offset((page - 1) * per_page).all() if not items and page != 1 and error_out: abort(404) # No need to count if we're on the first page and there are fewer # items than we expected. if page == 1 and len(items) < per_page: total = len(items) else: total = self.order_by(None).count() return Pagination(self, page, per_page, total, items) class _QueryProperty(object): def __init__(self, sa): self.sa = sa def __get__(self, obj, type): try: mapper = orm.class_mapper(type) if mapper: return type.query_class(mapper, session=self.sa.session()) except UnmappedClassError: return None def _record_queries(app): if app.debug: return True rq = app.config['SQLALCHEMY_RECORD_QUERIES'] if rq is not None: return rq return bool(app.config.get('TESTING')) class _EngineConnector(object): def __init__(self, sa, app, bind=None): self._sa = sa self._app = app self._engine = None self._connected_for = None self._bind = bind self._lock = Lock() def get_uri(self): if self._bind is None: return self._app.config['SQLALCHEMY_DATABASE_URI'] binds = self._app.config.get('SQLALCHEMY_BINDS') or () assert self._bind in binds, \ 'Bind %r is not specified. Set it in the SQLALCHEMY_BINDS ' \ 'configuration variable' % self._bind return binds[self._bind] def get_engine(self): with self._lock: uri = self.get_uri() echo = self._app.config['SQLALCHEMY_ECHO'] if (uri, echo) == self._connected_for: return self._engine info = make_url(uri) options = {'convert_unicode': True} self._sa.apply_pool_defaults(self._app, options) self._sa.apply_driver_hacks(self._app, info, options) if echo: options['echo'] = True self._engine = rv = sqlalchemy.create_engine(info, **options) if _record_queries(self._app): _EngineDebuggingSignalEvents(self._engine, self._app.import_name).register() self._connected_for = (uri, echo) return rv def _should_set_tablename(bases, d): """Check what values are set by a class and its bases to determine if a tablename should be automatically generated. The class and its bases are checked in order of precedence: the class itself then each base in the order they were given at class definition. Abstract classes do not generate a tablename, although they may have set or inherited a tablename elsewhere. If a class defines a tablename or table, a new one will not be generated. Otherwise, if the class defines a primary key, a new name will be generated. This supports: * Joined table inheritance without explicitly naming sub-models. * Single table inheritance. * Inheriting from mixins or abstract models. :param bases: base classes of new class :param d: new class dict :return: True if tablename should be set """ if '__tablename__' in d or '__table__' in d or '__abstract__' in d: return False if any(v.primary_key for v in itervalues(d) if isinstance(v, sqlalchemy.Column)): return True for base in bases: if hasattr(base, '__tablename__') or hasattr(base, '__table__'): return False for name in dir(base): attr = getattr(base, name) if isinstance(attr, sqlalchemy.Column) and attr.primary_key: return True class _BoundDeclarativeMeta(DeclarativeMeta): def __new__(cls, name, bases, d): if _should_set_tablename(bases, d): def _join(match): word = match.group() if len(word) > 1: return ('_%s_%s' % (word[:-1], word[-1])).lower() return '_' + word.lower() d['__tablename__'] = _camelcase_re.sub(_join, name).lstrip('_') return DeclarativeMeta.__new__(cls, name, bases, d) def __init__(self, name, bases, d): bind_key = d.pop('__bind_key__', None) DeclarativeMeta.__init__(self, name, bases, d) if bind_key is not None: self.__table__.info['bind_key'] = bind_key def get_state(app): """Gets the state for the application""" assert 'sqlalchemy' in app.extensions, \ 'The sqlalchemy extension was not registered to the current ' \ 'application. Please make sure to call init_app() first.' return app.extensions['sqlalchemy'] class _SQLAlchemyState(object): """Remembers configuration for the (db, app) tuple.""" def __init__(self, db, app): self.db = db self.app = app self.connectors = {} class Model(object): """Baseclass for custom user models.""" #: the query class used. The :attr:`query` attribute is an instance #: of this class. By default a :class:`BaseQuery` is used. query_class = BaseQuery #: an instance of :attr:`query_class`. Can be used to query the #: database for instances of this model. query = None class SQLAlchemy(object): """This class is used to control the SQLAlchemy integration to one or more Flask applications. Depending on how you initialize the object it is usable right away or will attach as needed to a Flask application. There are two usage modes which work very similarly. One is binding the instance to a very specific Flask application:: app = Flask(__name__) db = SQLAlchemy(app) The second possibility is to create the object once and configure the application later to support it:: db = SQLAlchemy() def create_app(): app = Flask(__name__) db.init_app(app) return app The difference between the two is that in the first case methods like :meth:`create_all` and :meth:`drop_all` will work all the time but in the second case a :meth:`flask.Flask.app_context` has to exist. By default Flask-SQLAlchemy will apply some backend-specific settings to improve your experience with them. As of SQLAlchemy 0.6 SQLAlchemy will probe the library for native unicode support. If it detects unicode it will let the library handle that, otherwise do that itself. Sometimes this detection can fail in which case you might want to set `use_native_unicode` (or the ``SQLALCHEMY_NATIVE_UNICODE`` configuration key) to `False`. Note that the configuration key overrides the value you pass to the constructor. This class also provides access to all the SQLAlchemy functions and classes from the :mod:`sqlalchemy` and :mod:`sqlalchemy.orm` modules. So you can declare models like this:: class User(db.Model): username = db.Column(db.String(80), unique=True) pw_hash = db.Column(db.String(80)) You can still use :mod:`sqlalchemy` and :mod:`sqlalchemy.orm` directly, but note that Flask-SQLAlchemy customizations are available only through an instance of this :class:`SQLAlchemy` class. Query classes default to :class:`BaseQuery` for `db.Query`, `db.Model.query_class`, and the default query_class for `db.relationship` and `db.backref`. If you use these interfaces through :mod:`sqlalchemy` and :mod:`sqlalchemy.orm` directly, the default query class will be that of :mod:`sqlalchemy`. .. admonition:: Check types carefully Don't perform type or `isinstance` checks against `db.Table`, which emulates `Table` behavior but is not a class. `db.Table` exposes the `Table` interface, but is a function which allows omission of metadata. You may also define your own SessionExtension instances as well when defining your SQLAlchemy class instance. You may pass your custom instances to the `session_extensions` keyword. This can be either a single SessionExtension instance, or a list of SessionExtension instances. In the following use case we use the VersionedListener from the SQLAlchemy versioning examples.:: from history_meta import VersionedMeta, VersionedListener app = Flask(__name__) db = SQLAlchemy(app, session_extensions=[VersionedListener()]) class User(db.Model): __metaclass__ = VersionedMeta username = db.Column(db.String(80), unique=True) pw_hash = db.Column(db.String(80)) The `session_options` parameter can be used to override session options. If provided it's a dict of parameters passed to the session's constructor. .. versionadded:: 0.10 The `session_options` parameter was added. .. versionadded:: 0.16 `scopefunc` is now accepted on `session_options`. It allows specifying a custom function which will define the SQLAlchemy session's scoping. .. versionadded:: 2.1 The `metadata` parameter was added. This allows for setting custom naming conventions among other, non-trivial things. """ def __init__(self, app=None, use_native_unicode=True, session_options=None, metadata=None): if session_options is None: session_options = {} session_options.setdefault('scopefunc', connection_stack.__ident_func__) self.use_native_unicode = use_native_unicode self.session = self.create_scoped_session(session_options) self.Model = self.make_declarative_base(metadata) self.Query = BaseQuery self._engine_lock = Lock() self.app = app _include_sqlalchemy(self) if app is not None: self.init_app(app) @property def metadata(self): """Returns the metadata""" return self.Model.metadata def create_scoped_session(self, options=None): """Helper factory method that creates a scoped session. It internally calls :meth:`create_session`. """ if options is None: options = {} scopefunc = options.pop('scopefunc', None) return orm.scoped_session(partial(self.create_session, options), scopefunc=scopefunc) def create_session(self, options): """Creates the session. The default implementation returns a :class:`SignallingSession`. .. versionadded:: 2.0 """ return SignallingSession(self, **options) def make_declarative_base(self, metadata=None): """Creates the declarative base.""" base = declarative_base(cls=Model, name='Model', metadata=metadata, metaclass=_BoundDeclarativeMeta) base.query = _QueryProperty(self) return base def init_app(self, app): """This callback can be used to initialize an application for the use with this database setup. Never use a database in the context of an application not initialized that way or connections will leak. """ app.config.setdefault('SQLALCHEMY_DATABASE_URI', 'sqlite://') app.config.setdefault('SQLALCHEMY_BINDS', None) app.config.setdefault('SQLALCHEMY_NATIVE_UNICODE', None) app.config.setdefault('SQLALCHEMY_ECHO', False) app.config.setdefault('SQLALCHEMY_RECORD_QUERIES', None) app.config.setdefault('SQLALCHEMY_POOL_SIZE', None) app.config.setdefault('SQLALCHEMY_POOL_TIMEOUT', None) app.config.setdefault('SQLALCHEMY_POOL_RECYCLE', None) app.config.setdefault('SQLALCHEMY_MAX_OVERFLOW', None) app.config.setdefault('SQLALCHEMY_COMMIT_ON_TEARDOWN', False) track_modifications = app.config.setdefault('SQLALCHEMY_TRACK_MODIFICATIONS', None) if track_modifications is None: warnings.warn('SQLALCHEMY_TRACK_MODIFICATIONS adds significant overhead and will be disabled by default in the future. Set it to True to suppress this warning.') if not hasattr(app, 'extensions'): app.extensions = {} app.extensions['sqlalchemy'] = _SQLAlchemyState(self, app) # 0.9 and later if hasattr(app, 'teardown_appcontext'): teardown = app.teardown_appcontext # 0.7 to 0.8 elif hasattr(app, 'teardown_request'): teardown = app.teardown_request # Older Flask versions else: if app.config['SQLALCHEMY_COMMIT_ON_TEARDOWN']: raise RuntimeError("Commit on teardown requires Flask >= 0.7") teardown = app.after_request @teardown def shutdown_session(response_or_exc): if app.config['SQLALCHEMY_COMMIT_ON_TEARDOWN']: if response_or_exc is None: self.session.commit() self.session.remove() return response_or_exc def apply_pool_defaults(self, app, options): def _setdefault(optionkey, configkey): value = app.config[configkey] if value is not None: options[optionkey] = value _setdefault('pool_size', 'SQLALCHEMY_POOL_SIZE') _setdefault('pool_timeout', 'SQLALCHEMY_POOL_TIMEOUT') _setdefault('pool_recycle', 'SQLALCHEMY_POOL_RECYCLE') _setdefault('max_overflow', 'SQLALCHEMY_MAX_OVERFLOW') def apply_driver_hacks(self, app, info, options): """This method is called before engine creation and used to inject driver specific hacks into the options. The `options` parameter is a dictionary of keyword arguments that will then be used to call the :func:`sqlalchemy.create_engine` function. The default implementation provides some saner defaults for things like pool sizes for MySQL and sqlite. Also it injects the setting of `SQLALCHEMY_NATIVE_UNICODE`. """ if info.drivername.startswith('mysql'): info.query.setdefault('charset', 'utf8') if info.drivername != 'mysql+gaerdbms': options.setdefault('pool_size', 10) options.setdefault('pool_recycle', 7200) elif info.drivername == 'sqlite': pool_size = options.get('pool_size') detected_in_memory = False # we go to memory and the pool size was explicitly set to 0 # which is fail. Let the user know that if info.database in (None, '', ':memory:'): detected_in_memory = True from sqlalchemy.pool import StaticPool options['poolclass'] = StaticPool if 'connect_args' not in options: options['connect_args'] = {} options['connect_args']['check_same_thread'] = False if pool_size == 0: raise RuntimeError('SQLite in memory database with an ' 'empty queue not possible due to data ' 'loss.') # if pool size is None or explicitly set to 0 we assume the # user did not want a queue for this sqlite connection and # hook in the null pool. elif not pool_size: from sqlalchemy.pool import NullPool options['poolclass'] = NullPool # if it's not an in memory database we make the path absolute. if not detected_in_memory: info.database = os.path.join(app.root_path, info.database) unu = app.config['SQLALCHEMY_NATIVE_UNICODE'] if unu is None: unu = self.use_native_unicode if not unu: options['use_native_unicode'] = False @property def engine(self): """Gives access to the engine. If the database configuration is bound to a specific application (initialized with an application) this will always return a database connection. If however the current application is used this might raise a :exc:`RuntimeError` if no application is active at the moment. """ return self.get_engine(self.get_app()) def make_connector(self, app, bind=None): """Creates the connector for a given state and bind.""" return _EngineConnector(self, app, bind) def get_engine(self, app, bind=None): """Returns a specific engine. .. versionadded:: 0.12 """ with self._engine_lock: state = get_state(app) connector = state.connectors.get(bind) if connector is None: connector = self.make_connector(app, bind) state.connectors[bind] = connector return connector.get_engine() def get_app(self, reference_app=None): """Helper method that implements the logic to look up an application. """ if reference_app is not None: return reference_app if self.app is not None: return self.app ctx = connection_stack.top if ctx is not None: return ctx.app raise RuntimeError('application not registered on db ' 'instance and no application bound ' 'to current context') def get_tables_for_bind(self, bind=None): """Returns a list of all tables relevant for a bind.""" result = [] for table in itervalues(self.Model.metadata.tables): if table.info.get('bind_key') == bind: result.append(table) return result def get_binds(self, app=None): """Returns a dictionary with a table->engine mapping. This is suitable for use of sessionmaker(binds=db.get_binds(app)). """ app = self.get_app(app) binds = [None] + list(app.config.get('SQLALCHEMY_BINDS') or ()) retval = {} for bind in binds: 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 _execute_for_all_tables(self, app, bind, operation, skip_tables=False): app = self.get_app(app) if bind == '__all__': binds = [None] + list(app.config.get('SQLALCHEMY_BINDS') or ()) elif isinstance(bind, string_types) or bind is None: binds = [bind] else: binds = bind for bind in binds: extra = {} if not skip_tables: tables = self.get_tables_for_bind(bind) extra['tables'] = tables op = getattr(self.Model.metadata, operation) op(bind=self.get_engine(app, bind), **extra) def create_all(self, bind='__all__', app=None): """Creates all tables. .. versionchanged:: 0.12 Parameters were added """ self._execute_for_all_tables(app, bind, 'create_all') def drop_all(self, bind='__all__', app=None): """Drops all tables. .. versionchanged:: 0.12 Parameters were added """ self._execute_for_all_tables(app, bind, 'drop_all') def reflect(self, bind='__all__', app=None): """Reflects tables from the database. .. versionchanged:: 0.12 Parameters were added """ self._execute_for_all_tables(app, bind, 'reflect', skip_tables=True) def __repr__(self): app = None if self.app is not None: app = self.app else: ctx = connection_stack.top if ctx is not None: app = ctx.app return '<%s engine=%r>' % ( self.__class__.__name__, app and app.config['SQLALCHEMY_DATABASE_URI'] or None )