This commit is contained in:
2010-05-07 17:33:49 +00:00
parent c7c7498f19
commit cae7e001e3
127 changed files with 57530 additions and 0 deletions

View File

@@ -0,0 +1 @@

View File

@@ -0,0 +1,878 @@
"""Contain the ``AssociationProxy`` class.
The ``AssociationProxy`` is a Python property object which provides
transparent proxied access to the endpoint of an association object.
See the example ``examples/association/proxied_association.py``.
"""
import itertools
import operator
import weakref
from sqlalchemy import exceptions
from sqlalchemy import orm
from sqlalchemy import util
from sqlalchemy.orm import collections
from sqlalchemy.sql import not_
def association_proxy(target_collection, attr, **kw):
"""Return a Python property implementing a view of *attr* over a collection.
Implements a read/write view over an instance's *target_collection*,
extracting *attr* from each member of the collection. The property acts
somewhat like this list comprehension::
[getattr(member, *attr*)
for member in getattr(instance, *target_collection*)]
Unlike the list comprehension, the collection returned by the property is
always in sync with *target_collection*, and mutations made to either
collection will be reflected in both.
Implements a Python property representing a relationship as a collection of
simpler values. The proxied property will mimic the collection type of
the target (list, dict or set), or, in the case of a one to one relationship,
a simple scalar value.
:param target_collection: Name of the relationship attribute we'll proxy to,
usually created with :func:`~sqlalchemy.orm.relationship`.
:param attr: Attribute on the associated instances we'll proxy for.
For example, given a target collection of [obj1, obj2], a list created
by this proxy property would look like [getattr(obj1, *attr*),
getattr(obj2, *attr*)]
If the relationship is one-to-one or otherwise uselist=False, then simply:
getattr(obj, *attr*)
:param creator: optional.
When new items are added to this proxied collection, new instances of
the class collected by the target collection will be created. For list
and set collections, the target class constructor will be called with
the 'value' for the new instance. For dict types, two arguments are
passed: key and value.
If you want to construct instances differently, supply a *creator*
function that takes arguments as above and returns instances.
For scalar relationships, creator() will be called if the target is None.
If the target is present, set operations are proxied to setattr() on the
associated object.
If you have an associated object with multiple attributes, you may set
up multiple association proxies mapping to different attributes. See
the unit tests for examples, and for examples of how creator() functions
can be used to construct the scalar relationship on-demand in this
situation.
:param \*\*kw: Passes along any other keyword arguments to
:class:`AssociationProxy`.
"""
return AssociationProxy(target_collection, attr, **kw)
class AssociationProxy(object):
"""A descriptor that presents a read/write view of an object attribute."""
def __init__(self, target_collection, attr, creator=None,
getset_factory=None, proxy_factory=None, proxy_bulk_set=None):
"""Arguments are:
target_collection
Name of the collection we'll proxy to, usually created with
'relationship()' in a mapper setup.
attr
Attribute on the collected instances we'll proxy for. For example,
given a target collection of [obj1, obj2], a list created by this
proxy property would look like [getattr(obj1, attr), getattr(obj2,
attr)]
creator
Optional. When new items are added to this proxied collection, new
instances of the class collected by the target collection will be
created. For list and set collections, the target class constructor
will be called with the 'value' for the new instance. For dict
types, two arguments are passed: key and value.
If you want to construct instances differently, supply a 'creator'
function that takes arguments as above and returns instances.
getset_factory
Optional. Proxied attribute access is automatically handled by
routines that get and set values based on the `attr` argument for
this proxy.
If you would like to customize this behavior, you may supply a
`getset_factory` callable that produces a tuple of `getter` and
`setter` functions. The factory is called with two arguments, the
abstract type of the underlying collection and this proxy instance.
proxy_factory
Optional. The type of collection to emulate is determined by
sniffing the target collection. If your collection type can't be
determined by duck typing or you'd like to use a different
collection implementation, you may supply a factory function to
produce those collections. Only applicable to non-scalar relationships.
proxy_bulk_set
Optional, use with proxy_factory. See the _set() method for
details.
"""
self.target_collection = target_collection
self.value_attr = attr
self.creator = creator
self.getset_factory = getset_factory
self.proxy_factory = proxy_factory
self.proxy_bulk_set = proxy_bulk_set
self.scalar = None
self.owning_class = None
self.key = '_%s_%s_%s' % (
type(self).__name__, target_collection, id(self))
self.collection_class = None
def _get_property(self):
return (orm.class_mapper(self.owning_class).
get_property(self.target_collection))
@property
def target_class(self):
"""The class the proxy is attached to."""
return self._get_property().mapper.class_
def _target_is_scalar(self):
return not self._get_property().uselist
def __get__(self, obj, class_):
if self.owning_class is None:
self.owning_class = class_ and class_ or type(obj)
if obj is None:
return self
elif self.scalar is None:
self.scalar = self._target_is_scalar()
if self.scalar:
self._initialize_scalar_accessors()
if self.scalar:
return self._scalar_get(getattr(obj, self.target_collection))
else:
try:
# If the owning instance is reborn (orm session resurrect,
# etc.), refresh the proxy cache.
creator_id, proxy = getattr(obj, self.key)
if id(obj) == creator_id:
return proxy
except AttributeError:
pass
proxy = self._new(_lazy_collection(obj, self.target_collection))
setattr(obj, self.key, (id(obj), proxy))
return proxy
def __set__(self, obj, values):
if self.owning_class is None:
self.owning_class = type(obj)
if self.scalar is None:
self.scalar = self._target_is_scalar()
if self.scalar:
self._initialize_scalar_accessors()
if self.scalar:
creator = self.creator and self.creator or self.target_class
target = getattr(obj, self.target_collection)
if target is None:
setattr(obj, self.target_collection, creator(values))
else:
self._scalar_set(target, values)
else:
proxy = self.__get__(obj, None)
if proxy is not values:
proxy.clear()
self._set(proxy, values)
def __delete__(self, obj):
if self.owning_class is None:
self.owning_class = type(obj)
delattr(obj, self.key)
def _initialize_scalar_accessors(self):
if self.getset_factory:
get, set = self.getset_factory(None, self)
else:
get, set = self._default_getset(None)
self._scalar_get, self._scalar_set = get, set
def _default_getset(self, collection_class):
attr = self.value_attr
getter = operator.attrgetter(attr)
if collection_class is dict:
setter = lambda o, k, v: setattr(o, attr, v)
else:
setter = lambda o, v: setattr(o, attr, v)
return getter, setter
def _new(self, lazy_collection):
creator = self.creator and self.creator or self.target_class
self.collection_class = util.duck_type_collection(lazy_collection())
if self.proxy_factory:
return self.proxy_factory(lazy_collection, creator, self.value_attr, self)
if self.getset_factory:
getter, setter = self.getset_factory(self.collection_class, self)
else:
getter, setter = self._default_getset(self.collection_class)
if self.collection_class is list:
return _AssociationList(lazy_collection, creator, getter, setter, self)
elif self.collection_class is dict:
return _AssociationDict(lazy_collection, creator, getter, setter, self)
elif self.collection_class is set:
return _AssociationSet(lazy_collection, creator, getter, setter, self)
else:
raise exceptions.ArgumentError(
'could not guess which interface to use for '
'collection_class "%s" backing "%s"; specify a '
'proxy_factory and proxy_bulk_set manually' %
(self.collection_class.__name__, self.target_collection))
def _inflate(self, proxy):
creator = self.creator and self.creator or self.target_class
if self.getset_factory:
getter, setter = self.getset_factory(self.collection_class, self)
else:
getter, setter = self._default_getset(self.collection_class)
proxy.creator = creator
proxy.getter = getter
proxy.setter = setter
def _set(self, proxy, values):
if self.proxy_bulk_set:
self.proxy_bulk_set(proxy, values)
elif self.collection_class is list:
proxy.extend(values)
elif self.collection_class is dict:
proxy.update(values)
elif self.collection_class is set:
proxy.update(values)
else:
raise exceptions.ArgumentError(
'no proxy_bulk_set supplied for custom '
'collection_class implementation')
@property
def _comparator(self):
return self._get_property().comparator
def any(self, criterion=None, **kwargs):
return self._comparator.any(getattr(self.target_class, self.value_attr).has(criterion, **kwargs))
def has(self, criterion=None, **kwargs):
return self._comparator.has(getattr(self.target_class, self.value_attr).has(criterion, **kwargs))
def contains(self, obj):
return self._comparator.any(**{self.value_attr: obj})
def __eq__(self, obj):
return self._comparator.has(**{self.value_attr: obj})
def __ne__(self, obj):
return not_(self.__eq__(obj))
class _lazy_collection(object):
def __init__(self, obj, target):
self.ref = weakref.ref(obj)
self.target = target
def __call__(self):
obj = self.ref()
if obj is None:
raise exceptions.InvalidRequestError(
"stale association proxy, parent object has gone out of "
"scope")
return getattr(obj, self.target)
def __getstate__(self):
return {'obj':self.ref(), 'target':self.target}
def __setstate__(self, state):
self.ref = weakref.ref(state['obj'])
self.target = state['target']
class _AssociationCollection(object):
def __init__(self, lazy_collection, creator, getter, setter, parent):
"""Constructs an _AssociationCollection.
This will always be a subclass of either _AssociationList,
_AssociationSet, or _AssociationDict.
lazy_collection
A callable returning a list-based collection of entities (usually an
object attribute managed by a SQLAlchemy relationship())
creator
A function that creates new target entities. Given one parameter:
value. This assertion is assumed::
obj = creator(somevalue)
assert getter(obj) == somevalue
getter
A function. Given an associated object, return the 'value'.
setter
A function. Given an associated object and a value, store that
value on the object.
"""
self.lazy_collection = lazy_collection
self.creator = creator
self.getter = getter
self.setter = setter
self.parent = parent
col = property(lambda self: self.lazy_collection())
def __len__(self):
return len(self.col)
def __nonzero__(self):
return bool(self.col)
def __getstate__(self):
return {'parent':self.parent, 'lazy_collection':self.lazy_collection}
def __setstate__(self, state):
self.parent = state['parent']
self.lazy_collection = state['lazy_collection']
self.parent._inflate(self)
class _AssociationList(_AssociationCollection):
"""Generic, converting, list-to-list proxy."""
def _create(self, value):
return self.creator(value)
def _get(self, object):
return self.getter(object)
def _set(self, object, value):
return self.setter(object, value)
def __getitem__(self, index):
return self._get(self.col[index])
def __setitem__(self, index, value):
if not isinstance(index, slice):
self._set(self.col[index], value)
else:
if index.stop is None:
stop = len(self)
elif index.stop < 0:
stop = len(self) + index.stop
else:
stop = index.stop
step = index.step or 1
rng = range(index.start or 0, stop, step)
if step == 1:
for i in rng:
del self[index.start]
i = index.start
for item in value:
self.insert(i, item)
i += 1
else:
if len(value) != len(rng):
raise ValueError(
"attempt to assign sequence of size %s to "
"extended slice of size %s" % (len(value),
len(rng)))
for i, item in zip(rng, value):
self._set(self.col[i], item)
def __delitem__(self, index):
del self.col[index]
def __contains__(self, value):
for member in self.col:
# testlib.pragma exempt:__eq__
if self._get(member) == value:
return True
return False
def __getslice__(self, start, end):
return [self._get(member) for member in self.col[start:end]]
def __setslice__(self, start, end, values):
members = [self._create(v) for v in values]
self.col[start:end] = members
def __delslice__(self, start, end):
del self.col[start:end]
def __iter__(self):
"""Iterate over proxied values.
For the actual domain objects, iterate over .col instead or
just use the underlying collection directly from its property
on the parent.
"""
for member in self.col:
yield self._get(member)
raise StopIteration
def append(self, value):
item = self._create(value)
self.col.append(item)
def count(self, value):
return sum([1 for _ in
itertools.ifilter(lambda v: v == value, iter(self))])
def extend(self, values):
for v in values:
self.append(v)
def insert(self, index, value):
self.col[index:index] = [self._create(value)]
def pop(self, index=-1):
return self.getter(self.col.pop(index))
def remove(self, value):
for i, val in enumerate(self):
if val == value:
del self.col[i]
return
raise ValueError("value not in list")
def reverse(self):
"""Not supported, use reversed(mylist)"""
raise NotImplementedError
def sort(self):
"""Not supported, use sorted(mylist)"""
raise NotImplementedError
def clear(self):
del self.col[0:len(self.col)]
def __eq__(self, other):
return list(self) == other
def __ne__(self, other):
return list(self) != other
def __lt__(self, other):
return list(self) < other
def __le__(self, other):
return list(self) <= other
def __gt__(self, other):
return list(self) > other
def __ge__(self, other):
return list(self) >= other
def __cmp__(self, other):
return cmp(list(self), other)
def __add__(self, iterable):
try:
other = list(iterable)
except TypeError:
return NotImplemented
return list(self) + other
def __radd__(self, iterable):
try:
other = list(iterable)
except TypeError:
return NotImplemented
return other + list(self)
def __mul__(self, n):
if not isinstance(n, int):
return NotImplemented
return list(self) * n
__rmul__ = __mul__
def __iadd__(self, iterable):
self.extend(iterable)
return self
def __imul__(self, n):
# unlike a regular list *=, proxied __imul__ will generate unique
# backing objects for each copy. *= on proxied lists is a bit of
# a stretch anyhow, and this interpretation of the __imul__ contract
# is more plausibly useful than copying the backing objects.
if not isinstance(n, int):
return NotImplemented
if n == 0:
self.clear()
elif n > 1:
self.extend(list(self) * (n - 1))
return self
def copy(self):
return list(self)
def __repr__(self):
return repr(list(self))
def __hash__(self):
raise TypeError("%s objects are unhashable" % type(self).__name__)
for func_name, func in locals().items():
if (util.callable(func) and func.func_name == func_name and
not func.__doc__ and hasattr(list, func_name)):
func.__doc__ = getattr(list, func_name).__doc__
del func_name, func
_NotProvided = util.symbol('_NotProvided')
class _AssociationDict(_AssociationCollection):
"""Generic, converting, dict-to-dict proxy."""
def _create(self, key, value):
return self.creator(key, value)
def _get(self, object):
return self.getter(object)
def _set(self, object, key, value):
return self.setter(object, key, value)
def __getitem__(self, key):
return self._get(self.col[key])
def __setitem__(self, key, value):
if key in self.col:
self._set(self.col[key], key, value)
else:
self.col[key] = self._create(key, value)
def __delitem__(self, key):
del self.col[key]
def __contains__(self, key):
# testlib.pragma exempt:__hash__
return key in self.col
def has_key(self, key):
# testlib.pragma exempt:__hash__
return key in self.col
def __iter__(self):
return self.col.iterkeys()
def clear(self):
self.col.clear()
def __eq__(self, other):
return dict(self) == other
def __ne__(self, other):
return dict(self) != other
def __lt__(self, other):
return dict(self) < other
def __le__(self, other):
return dict(self) <= other
def __gt__(self, other):
return dict(self) > other
def __ge__(self, other):
return dict(self) >= other
def __cmp__(self, other):
return cmp(dict(self), other)
def __repr__(self):
return repr(dict(self.items()))
def get(self, key, default=None):
try:
return self[key]
except KeyError:
return default
def setdefault(self, key, default=None):
if key not in self.col:
self.col[key] = self._create(key, default)
return default
else:
return self[key]
def keys(self):
return self.col.keys()
def iterkeys(self):
return self.col.iterkeys()
def values(self):
return [ self._get(member) for member in self.col.values() ]
def itervalues(self):
for key in self.col:
yield self._get(self.col[key])
raise StopIteration
def items(self):
return [(k, self._get(self.col[k])) for k in self]
def iteritems(self):
for key in self.col:
yield (key, self._get(self.col[key]))
raise StopIteration
def pop(self, key, default=_NotProvided):
if default is _NotProvided:
member = self.col.pop(key)
else:
member = self.col.pop(key, default)
return self._get(member)
def popitem(self):
item = self.col.popitem()
return (item[0], self._get(item[1]))
def update(self, *a, **kw):
if len(a) > 1:
raise TypeError('update expected at most 1 arguments, got %i' %
len(a))
elif len(a) == 1:
seq_or_map = a[0]
for item in seq_or_map:
if isinstance(item, tuple):
self[item[0]] = item[1]
else:
self[item] = seq_or_map[item]
for key, value in kw:
self[key] = value
def copy(self):
return dict(self.items())
def __hash__(self):
raise TypeError("%s objects are unhashable" % type(self).__name__)
for func_name, func in locals().items():
if (util.callable(func) and func.func_name == func_name and
not func.__doc__ and hasattr(dict, func_name)):
func.__doc__ = getattr(dict, func_name).__doc__
del func_name, func
class _AssociationSet(_AssociationCollection):
"""Generic, converting, set-to-set proxy."""
def _create(self, value):
return self.creator(value)
def _get(self, object):
return self.getter(object)
def _set(self, object, value):
return self.setter(object, value)
def __len__(self):
return len(self.col)
def __nonzero__(self):
if self.col:
return True
else:
return False
def __contains__(self, value):
for member in self.col:
# testlib.pragma exempt:__eq__
if self._get(member) == value:
return True
return False
def __iter__(self):
"""Iterate over proxied values.
For the actual domain objects, iterate over .col instead or just use
the underlying collection directly from its property on the parent.
"""
for member in self.col:
yield self._get(member)
raise StopIteration
def add(self, value):
if value not in self:
self.col.add(self._create(value))
# for discard and remove, choosing a more expensive check strategy rather
# than call self.creator()
def discard(self, value):
for member in self.col:
if self._get(member) == value:
self.col.discard(member)
break
def remove(self, value):
for member in self.col:
if self._get(member) == value:
self.col.discard(member)
return
raise KeyError(value)
def pop(self):
if not self.col:
raise KeyError('pop from an empty set')
member = self.col.pop()
return self._get(member)
def update(self, other):
for value in other:
self.add(value)
def __ior__(self, other):
if not collections._set_binops_check_strict(self, other):
return NotImplemented
for value in other:
self.add(value)
return self
def _set(self):
return set(iter(self))
def union(self, other):
return set(self).union(other)
__or__ = union
def difference(self, other):
return set(self).difference(other)
__sub__ = difference
def difference_update(self, other):
for value in other:
self.discard(value)
def __isub__(self, other):
if not collections._set_binops_check_strict(self, other):
return NotImplemented
for value in other:
self.discard(value)
return self
def intersection(self, other):
return set(self).intersection(other)
__and__ = intersection
def intersection_update(self, other):
want, have = self.intersection(other), set(self)
remove, add = have - want, want - have
for value in remove:
self.remove(value)
for value in add:
self.add(value)
def __iand__(self, other):
if not collections._set_binops_check_strict(self, other):
return NotImplemented
want, have = self.intersection(other), set(self)
remove, add = have - want, want - have
for value in remove:
self.remove(value)
for value in add:
self.add(value)
return self
def symmetric_difference(self, other):
return set(self).symmetric_difference(other)
__xor__ = symmetric_difference
def symmetric_difference_update(self, other):
want, have = self.symmetric_difference(other), set(self)
remove, add = have - want, want - have
for value in remove:
self.remove(value)
for value in add:
self.add(value)
def __ixor__(self, other):
if not collections._set_binops_check_strict(self, other):
return NotImplemented
want, have = self.symmetric_difference(other), set(self)
remove, add = have - want, want - have
for value in remove:
self.remove(value)
for value in add:
self.add(value)
return self
def issubset(self, other):
return set(self).issubset(other)
def issuperset(self, other):
return set(self).issuperset(other)
def clear(self):
self.col.clear()
def copy(self):
return set(self)
def __eq__(self, other):
return set(self) == other
def __ne__(self, other):
return set(self) != other
def __lt__(self, other):
return set(self) < other
def __le__(self, other):
return set(self) <= other
def __gt__(self, other):
return set(self) > other
def __ge__(self, other):
return set(self) >= other
def __repr__(self):
return repr(set(self))
def __hash__(self):
raise TypeError("%s objects are unhashable" % type(self).__name__)
for func_name, func in locals().items():
if (util.callable(func) and func.func_name == func_name and
not func.__doc__ and hasattr(set, func_name)):
func.__doc__ = getattr(set, func_name).__doc__
del func_name, func

194
sqlalchemy/ext/compiler.py Normal file
View File

@@ -0,0 +1,194 @@
"""Provides an API for creation of custom ClauseElements and compilers.
Synopsis
========
Usage involves the creation of one or more :class:`~sqlalchemy.sql.expression.ClauseElement`
subclasses and one or more callables defining its compilation::
from sqlalchemy.ext.compiler import compiles
from sqlalchemy.sql.expression import ColumnClause
class MyColumn(ColumnClause):
pass
@compiles(MyColumn)
def compile_mycolumn(element, compiler, **kw):
return "[%s]" % element.name
Above, ``MyColumn`` extends :class:`~sqlalchemy.sql.expression.ColumnClause`,
the base expression element for named column objects. The ``compiles``
decorator registers itself with the ``MyColumn`` class so that it is invoked
when the object is compiled to a string::
from sqlalchemy import select
s = select([MyColumn('x'), MyColumn('y')])
print str(s)
Produces::
SELECT [x], [y]
Dialect-specific compilation rules
==================================
Compilers can also be made dialect-specific. The appropriate compiler will be
invoked for the dialect in use::
from sqlalchemy.schema import DDLElement
class AlterColumn(DDLElement):
def __init__(self, column, cmd):
self.column = column
self.cmd = cmd
@compiles(AlterColumn)
def visit_alter_column(element, compiler, **kw):
return "ALTER COLUMN %s ..." % element.column.name
@compiles(AlterColumn, 'postgresql')
def visit_alter_column(element, compiler, **kw):
return "ALTER TABLE %s ALTER COLUMN %s ..." % (element.table.name, element.column.name)
The second ``visit_alter_table`` will be invoked when any ``postgresql`` dialect is used.
Compiling sub-elements of a custom expression construct
=======================================================
The ``compiler`` argument is the :class:`~sqlalchemy.engine.base.Compiled`
object in use. This object can be inspected for any information about the
in-progress compilation, including ``compiler.dialect``,
``compiler.statement`` etc. The :class:`~sqlalchemy.sql.compiler.SQLCompiler`
and :class:`~sqlalchemy.sql.compiler.DDLCompiler` both include a ``process()``
method which can be used for compilation of embedded attributes::
from sqlalchemy.sql.expression import Executable, ClauseElement
class InsertFromSelect(Executable, ClauseElement):
def __init__(self, table, select):
self.table = table
self.select = select
@compiles(InsertFromSelect)
def visit_insert_from_select(element, compiler, **kw):
return "INSERT INTO %s (%s)" % (
compiler.process(element.table, asfrom=True),
compiler.process(element.select)
)
insert = InsertFromSelect(t1, select([t1]).where(t1.c.x>5))
print insert
Produces::
"INSERT INTO mytable (SELECT mytable.x, mytable.y, mytable.z FROM mytable WHERE mytable.x > :x_1)"
Cross Compiling between SQL and DDL compilers
---------------------------------------------
SQL and DDL constructs are each compiled using different base compilers - ``SQLCompiler``
and ``DDLCompiler``. A common need is to access the compilation rules of SQL expressions
from within a DDL expression. The ``DDLCompiler`` includes an accessor ``sql_compiler`` for this reason, such as below where we generate a CHECK
constraint that embeds a SQL expression::
@compiles(MyConstraint)
def compile_my_constraint(constraint, ddlcompiler, **kw):
return "CONSTRAINT %s CHECK (%s)" % (
constraint.name,
ddlcompiler.sql_compiler.process(constraint.expression)
)
Changing the default compilation of existing constructs
=======================================================
The compiler extension applies just as well to the existing constructs. When overriding
the compilation of a built in SQL construct, the @compiles decorator is invoked upon
the appropriate class (be sure to use the class, i.e. ``Insert`` or ``Select``, instead of the creation function such as ``insert()`` or ``select()``).
Within the new compilation function, to get at the "original" compilation routine,
use the appropriate visit_XXX method - this because compiler.process() will call upon the
overriding routine and cause an endless loop. Such as, to add "prefix" to all insert statements::
from sqlalchemy.sql.expression import Insert
@compiles(Insert)
def prefix_inserts(insert, compiler, **kw):
return compiler.visit_insert(insert.prefix_with("some prefix"), **kw)
The above compiler will prefix all INSERT statements with "some prefix" when compiled.
Subclassing Guidelines
======================
A big part of using the compiler extension is subclassing SQLAlchemy expression constructs. To make this easier, the expression and schema packages feature a set of "bases" intended for common tasks. A synopsis is as follows:
* :class:`~sqlalchemy.sql.expression.ClauseElement` - This is the root
expression class. Any SQL expression can be derived from this base, and is
probably the best choice for longer constructs such as specialized INSERT
statements.
* :class:`~sqlalchemy.sql.expression.ColumnElement` - The root of all
"column-like" elements. Anything that you'd place in the "columns" clause of
a SELECT statement (as well as order by and group by) can derive from this -
the object will automatically have Python "comparison" behavior.
:class:`~sqlalchemy.sql.expression.ColumnElement` classes want to have a
``type`` member which is expression's return type. This can be established
at the instance level in the constructor, or at the class level if its
generally constant::
class timestamp(ColumnElement):
type = TIMESTAMP()
* :class:`~sqlalchemy.sql.expression.FunctionElement` - This is a hybrid of a
``ColumnElement`` and a "from clause" like object, and represents a SQL
function or stored procedure type of call. Since most databases support
statements along the line of "SELECT FROM <some function>"
``FunctionElement`` adds in the ability to be used in the FROM clause of a
``select()`` construct.
* :class:`~sqlalchemy.schema.DDLElement` - The root of all DDL expressions,
like CREATE TABLE, ALTER TABLE, etc. Compilation of ``DDLElement``
subclasses is issued by a ``DDLCompiler`` instead of a ``SQLCompiler``.
``DDLElement`` also features ``Table`` and ``MetaData`` event hooks via the
``execute_at()`` method, allowing the construct to be invoked during CREATE
TABLE and DROP TABLE sequences.
* :class:`~sqlalchemy.sql.expression.Executable` - This is a mixin which should be
used with any expression class that represents a "standalone" SQL statement that
can be passed directly to an ``execute()`` method. It is already implicit
within ``DDLElement`` and ``FunctionElement``.
"""
def compiles(class_, *specs):
def decorate(fn):
existing = getattr(class_, '_compiler_dispatcher', None)
if not existing:
existing = _dispatcher()
# TODO: why is the lambda needed ?
setattr(class_, '_compiler_dispatch', lambda *arg, **kw: existing(*arg, **kw))
setattr(class_, '_compiler_dispatcher', existing)
if specs:
for s in specs:
existing.specs[s] = fn
else:
existing.specs['default'] = fn
return fn
return decorate
class _dispatcher(object):
def __init__(self):
self.specs = {}
def __call__(self, element, compiler, **kw):
# TODO: yes, this could also switch off of DBAPI in use.
fn = self.specs.get(compiler.dialect.name, None)
if not fn:
fn = self.specs['default']
return fn(element, compiler, **kw)

View File

@@ -0,0 +1,940 @@
"""
Synopsis
========
SQLAlchemy object-relational configuration involves the use of
:class:`~sqlalchemy.schema.Table`, :func:`~sqlalchemy.orm.mapper`, and
class objects to define the three areas of configuration.
:mod:`~sqlalchemy.ext.declarative` allows all three types of
configuration to be expressed declaratively on an individual
mapped class. Regular SQLAlchemy schema elements and ORM constructs
are used in most cases.
As a simple example::
from sqlalchemy.ext.declarative import declarative_base
Base = declarative_base()
class SomeClass(Base):
__tablename__ = 'some_table'
id = Column(Integer, primary_key=True)
name = Column(String(50))
Above, the :func:`declarative_base` callable returns a new base class from which
all mapped classes should inherit. When the class definition is completed, a
new :class:`~sqlalchemy.schema.Table` and
:class:`~sqlalchemy.orm.mapper` will have been generated, accessible
via the ``__table__`` and ``__mapper__`` attributes on the ``SomeClass`` class.
Defining Attributes
===================
In the above example, the :class:`~sqlalchemy.schema.Column` objects are
automatically named with the name of the attribute to which they are
assigned.
They can also be explicitly named, and that name does not have to be
the same as name assigned on the class.
The column will be assigned to the :class:`~sqlalchemy.schema.Table` using the
given name, and mapped to the class using the attribute name::
class SomeClass(Base):
__tablename__ = 'some_table'
id = Column("some_table_id", Integer, primary_key=True)
name = Column("name", String(50))
Attributes may be added to the class after its construction, and they will be
added to the underlying :class:`~sqlalchemy.schema.Table` and
:func:`~sqlalchemy.orm.mapper()` definitions as appropriate::
SomeClass.data = Column('data', Unicode)
SomeClass.related = relationship(RelatedInfo)
Classes which are mapped explicitly using
:func:`~sqlalchemy.orm.mapper()` can interact freely with declarative
classes.
It is recommended, though not required, that all tables
share the same underlying :class:`~sqlalchemy.schema.MetaData` object,
so that string-configured :class:`~sqlalchemy.schema.ForeignKey`
references can be resolved without issue.
Association of Metadata and Engine
==================================
The :func:`declarative_base` base class contains a
:class:`~sqlalchemy.schema.MetaData` object where newly
defined :class:`~sqlalchemy.schema.Table` objects are collected. This
is accessed via the :class:`~sqlalchemy.schema.MetaData` class level
accessor, so to create tables we can say::
engine = create_engine('sqlite://')
Base.metadata.create_all(engine)
The :class:`~sqlalchemy.engine.base.Engine` created above may also be
directly associated with the declarative base class using the ``bind``
keyword argument, where it will be associated with the underlying
:class:`~sqlalchemy.schema.MetaData` object and allow SQL operations
involving that metadata and its tables to make use of that engine
automatically::
Base = declarative_base(bind=create_engine('sqlite://'))
Alternatively, by way of the normal
:class:`~sqlalchemy.schema.MetaData` behaviour, the ``bind`` attribute
of the class level accessor can be assigned at any time as follows::
Base.metadata.bind = create_engine('sqlite://')
The :func:`declarative_base` can also receive a pre-created
:class:`~sqlalchemy.schema.MetaData` object, which allows a
declarative setup to be associated with an already
existing traditional collection of :class:`~sqlalchemy.schema.Table`
objects::
mymetadata = MetaData()
Base = declarative_base(metadata=mymetadata)
Configuring Relationships
=========================
Relationships to other classes are done in the usual way, with the added
feature that the class specified to :func:`~sqlalchemy.orm.relationship`
may be a string name (note that :func:`~sqlalchemy.orm.relationship` is
only available as of SQLAlchemy 0.6beta2, and in all prior versions is known
as :func:`~sqlalchemy.orm.relation`,
including 0.5 and 0.4). The "class registry" associated with ``Base``
is used at mapper compilation time to resolve the name into the actual
class object, which is expected to have been defined once the mapper
configuration is used::
class User(Base):
__tablename__ = 'users'
id = Column(Integer, primary_key=True)
name = Column(String(50))
addresses = relationship("Address", backref="user")
class Address(Base):
__tablename__ = 'addresses'
id = Column(Integer, primary_key=True)
email = Column(String(50))
user_id = Column(Integer, ForeignKey('users.id'))
Column constructs, since they are just that, are immediately usable,
as below where we define a primary join condition on the ``Address``
class using them::
class Address(Base):
__tablename__ = 'addresses'
id = Column(Integer, primary_key=True)
email = Column(String(50))
user_id = Column(Integer, ForeignKey('users.id'))
user = relationship(User, primaryjoin=user_id == User.id)
In addition to the main argument for :func:`~sqlalchemy.orm.relationship`,
other arguments which depend upon the columns present on an as-yet
undefined class may also be specified as strings. These strings are
evaluated as Python expressions. The full namespace available within
this evaluation includes all classes mapped for this declarative base,
as well as the contents of the ``sqlalchemy`` package, including
expression functions like :func:`~sqlalchemy.sql.expression.desc` and
:attr:`~sqlalchemy.sql.expression.func`::
class User(Base):
# ....
addresses = relationship("Address",
order_by="desc(Address.email)",
primaryjoin="Address.user_id==User.id")
As an alternative to string-based attributes, attributes may also be
defined after all classes have been created. Just add them to the target
class after the fact::
User.addresses = relationship(Address,
primaryjoin=Address.user_id==User.id)
Configuring Many-to-Many Relationships
======================================
There's nothing special about many-to-many with declarative. The
``secondary`` argument to :func:`~sqlalchemy.orm.relationship` still
requires a :class:`~sqlalchemy.schema.Table` object, not a declarative
class. The :class:`~sqlalchemy.schema.Table` should share the same
:class:`~sqlalchemy.schema.MetaData` object used by the declarative
base::
keywords = Table(
'keywords', Base.metadata,
Column('author_id', Integer, ForeignKey('authors.id')),
Column('keyword_id', Integer, ForeignKey('keywords.id'))
)
class Author(Base):
__tablename__ = 'authors'
id = Column(Integer, primary_key=True)
keywords = relationship("Keyword", secondary=keywords)
You should generally **not** map a class and also specify its table in
a many-to-many relationship, since the ORM may issue duplicate INSERT and
DELETE statements.
Defining Synonyms
=================
Synonyms are introduced in :ref:`synonyms`. To define a getter/setter
which proxies to an underlying attribute, use
:func:`~sqlalchemy.orm.synonym` with the ``descriptor`` argument::
class MyClass(Base):
__tablename__ = 'sometable'
_attr = Column('attr', String)
def _get_attr(self):
return self._some_attr
def _set_attr(self, attr):
self._some_attr = attr
attr = synonym('_attr', descriptor=property(_get_attr, _set_attr))
The above synonym is then usable as an instance attribute as well as a
class-level expression construct::
x = MyClass()
x.attr = "some value"
session.query(MyClass).filter(MyClass.attr == 'some other value').all()
For simple getters, the :func:`synonym_for` decorator can be used in
conjunction with ``@property``::
class MyClass(Base):
__tablename__ = 'sometable'
_attr = Column('attr', String)
@synonym_for('_attr')
@property
def attr(self):
return self._some_attr
Similarly, :func:`comparable_using` is a front end for the
:func:`~sqlalchemy.orm.comparable_property` ORM function::
class MyClass(Base):
__tablename__ = 'sometable'
name = Column('name', String)
@comparable_using(MyUpperCaseComparator)
@property
def uc_name(self):
return self.name.upper()
Table Configuration
===================
Table arguments other than the name, metadata, and mapped Column
arguments are specified using the ``__table_args__`` class attribute.
This attribute accommodates both positional as well as keyword
arguments that are normally sent to the
:class:`~sqlalchemy.schema.Table` constructor.
The attribute can be specified in one of two forms. One is as a
dictionary::
class MyClass(Base):
__tablename__ = 'sometable'
__table_args__ = {'mysql_engine':'InnoDB'}
The other, a tuple of the form
``(arg1, arg2, ..., {kwarg1:value, ...})``, which allows positional
arguments to be specified as well (usually constraints)::
class MyClass(Base):
__tablename__ = 'sometable'
__table_args__ = (
ForeignKeyConstraint(['id'], ['remote_table.id']),
UniqueConstraint('foo'),
{'autoload':True}
)
Note that the keyword parameters dictionary is required in the tuple
form even if empty.
As an alternative to ``__tablename__``, a direct
:class:`~sqlalchemy.schema.Table` construct may be used. The
:class:`~sqlalchemy.schema.Column` objects, which in this case require
their names, will be added to the mapping just like a regular mapping
to a table::
class MyClass(Base):
__table__ = Table('my_table', Base.metadata,
Column('id', Integer, primary_key=True),
Column('name', String(50))
)
Mapper Configuration
====================
Configuration of mappers is done with the
:func:`~sqlalchemy.orm.mapper` function and all the possible mapper
configuration parameters can be found in the documentation for that
function.
:func:`~sqlalchemy.orm.mapper` is still used by declaratively mapped
classes and keyword parameters to the function can be passed by
placing them in the ``__mapper_args__`` class variable::
class Widget(Base):
__tablename__ = 'widgets'
id = Column(Integer, primary_key=True)
__mapper_args__ = {'extension': MyWidgetExtension()}
Inheritance Configuration
=========================
Declarative supports all three forms of inheritance as intuitively
as possible. The ``inherits`` mapper keyword argument is not needed
as declarative will determine this from the class itself. The various
"polymorphic" keyword arguments are specified using ``__mapper_args__``.
Joined Table Inheritance
~~~~~~~~~~~~~~~~~~~~~~~~
Joined table inheritance is defined as a subclass that defines its own
table::
class Person(Base):
__tablename__ = 'people'
id = Column(Integer, primary_key=True)
discriminator = Column('type', String(50))
__mapper_args__ = {'polymorphic_on': discriminator}
class Engineer(Person):
__tablename__ = 'engineers'
__mapper_args__ = {'polymorphic_identity': 'engineer'}
id = Column(Integer, ForeignKey('people.id'), primary_key=True)
primary_language = Column(String(50))
Note that above, the ``Engineer.id`` attribute, since it shares the
same attribute name as the ``Person.id`` attribute, will in fact
represent the ``people.id`` and ``engineers.id`` columns together, and
will render inside a query as ``"people.id"``.
To provide the ``Engineer`` class with an attribute that represents
only the ``engineers.id`` column, give it a different attribute name::
class Engineer(Person):
__tablename__ = 'engineers'
__mapper_args__ = {'polymorphic_identity': 'engineer'}
engineer_id = Column('id', Integer, ForeignKey('people.id'), primary_key=True)
primary_language = Column(String(50))
Single Table Inheritance
~~~~~~~~~~~~~~~~~~~~~~~~
Single table inheritance is defined as a subclass that does not have
its own table; you just leave out the ``__table__`` and ``__tablename__``
attributes::
class Person(Base):
__tablename__ = 'people'
id = Column(Integer, primary_key=True)
discriminator = Column('type', String(50))
__mapper_args__ = {'polymorphic_on': discriminator}
class Engineer(Person):
__mapper_args__ = {'polymorphic_identity': 'engineer'}
primary_language = Column(String(50))
When the above mappers are configured, the ``Person`` class is mapped
to the ``people`` table *before* the ``primary_language`` column is
defined, and this column will not be included in its own mapping.
When ``Engineer`` then defines the ``primary_language`` column, the
column is added to the ``people`` table so that it is included in the
mapping for ``Engineer`` and is also part of the table's full set of
columns. Columns which are not mapped to ``Person`` are also excluded
from any other single or joined inheriting classes using the
``exclude_properties`` mapper argument. Below, ``Manager`` will have
all the attributes of ``Person`` and ``Manager`` but *not* the
``primary_language`` attribute of ``Engineer``::
class Manager(Person):
__mapper_args__ = {'polymorphic_identity': 'manager'}
golf_swing = Column(String(50))
The attribute exclusion logic is provided by the
``exclude_properties`` mapper argument, and declarative's default
behavior can be disabled by passing an explicit ``exclude_properties``
collection (empty or otherwise) to the ``__mapper_args__``.
Concrete Table Inheritance
~~~~~~~~~~~~~~~~~~~~~~~~~~
Concrete is defined as a subclass which has its own table and sets the
``concrete`` keyword argument to ``True``::
class Person(Base):
__tablename__ = 'people'
id = Column(Integer, primary_key=True)
name = Column(String(50))
class Engineer(Person):
__tablename__ = 'engineers'
__mapper_args__ = {'concrete':True}
id = Column(Integer, primary_key=True)
primary_language = Column(String(50))
name = Column(String(50))
Usage of an abstract base class is a little less straightforward as it
requires usage of :func:`~sqlalchemy.orm.util.polymorphic_union`::
engineers = Table('engineers', Base.metadata,
Column('id', Integer, primary_key=True),
Column('name', String(50)),
Column('primary_language', String(50))
)
managers = Table('managers', Base.metadata,
Column('id', Integer, primary_key=True),
Column('name', String(50)),
Column('golf_swing', String(50))
)
punion = polymorphic_union({
'engineer':engineers,
'manager':managers
}, 'type', 'punion')
class Person(Base):
__table__ = punion
__mapper_args__ = {'polymorphic_on':punion.c.type}
class Engineer(Person):
__table__ = engineers
__mapper_args__ = {'polymorphic_identity':'engineer', 'concrete':True}
class Manager(Person):
__table__ = managers
__mapper_args__ = {'polymorphic_identity':'manager', 'concrete':True}
Mix-in Classes
==============
A common need when using :mod:`~sqlalchemy.ext.declarative` is to
share some functionality, often a set of columns, across many
classes. The normal python idiom would be to put this common code into
a base class and have all the other classes subclass this class.
When using :mod:`~sqlalchemy.ext.declarative`, this need is met by
using a "mix-in class". A mix-in class is one that isn't mapped to a
table and doesn't subclass the declarative :class:`Base`. For example::
class MyMixin(object):
__table_args__ = {'mysql_engine':'InnoDB'}
__mapper_args__=dict(always_refresh=True)
id = Column(Integer, primary_key=True)
def foo(self):
return 'bar'+str(self.id)
class MyModel(Base,MyMixin):
__tablename__='test'
name = Column(String(1000), nullable=False, index=True)
As the above example shows, ``__table_args__`` and ``__mapper_args__``
can both be abstracted out into a mix-in if you use common values for
these across many classes.
However, particularly in the case of ``__table_args__``, you may want
to combine some parameters from several mix-ins with those you wish to
define on the class iteself. To help with this, a
:func:`~sqlalchemy.util.classproperty` decorator is provided that lets
you implement a class property with a function. For example::
from sqlalchemy.util import classproperty
class MySQLSettings:
__table_args__ = {'mysql_engine':'InnoDB'}
class MyOtherMixin:
__table_args__ = {'info':'foo'}
class MyModel(Base,MySQLSettings,MyOtherMixin):
__tablename__='my_model'
@classproperty
def __table_args__(self):
args = dict()
args.update(MySQLSettings.__table_args__)
args.update(MyOtherMixin.__table_args__)
return args
id = Column(Integer, primary_key=True)
Class Constructor
=================
As a convenience feature, the :func:`declarative_base` sets a default
constructor on classes which takes keyword arguments, and assigns them
to the named attributes::
e = Engineer(primary_language='python')
Sessions
========
Note that ``declarative`` does nothing special with sessions, and is
only intended as an easier way to configure mappers and
:class:`~sqlalchemy.schema.Table` objects. A typical application
setup using :func:`~sqlalchemy.orm.scoped_session` might look like::
engine = create_engine('postgresql://scott:tiger@localhost/test')
Session = scoped_session(sessionmaker(autocommit=False,
autoflush=False,
bind=engine))
Base = declarative_base()
Mapped instances then make usage of
:class:`~sqlalchemy.orm.session.Session` in the usual way.
"""
from sqlalchemy.schema import Table, Column, MetaData
from sqlalchemy.orm import synonym as _orm_synonym, mapper, comparable_property, class_mapper
from sqlalchemy.orm.interfaces import MapperProperty
from sqlalchemy.orm.properties import RelationshipProperty, ColumnProperty
from sqlalchemy.orm.util import _is_mapped_class
from sqlalchemy import util, exceptions
from sqlalchemy.sql import util as sql_util
__all__ = 'declarative_base', 'synonym_for', 'comparable_using', 'instrument_declarative'
def instrument_declarative(cls, registry, metadata):
"""Given a class, configure the class declaratively,
using the given registry, which can be any dictionary, and
MetaData object.
"""
if '_decl_class_registry' in cls.__dict__:
raise exceptions.InvalidRequestError(
"Class %r already has been "
"instrumented declaratively" % cls)
cls._decl_class_registry = registry
cls.metadata = metadata
_as_declarative(cls, cls.__name__, cls.__dict__)
def _as_declarative(cls, classname, dict_):
# dict_ will be a dictproxy, which we can't write to, and we need to!
dict_ = dict(dict_)
column_copies = dict()
unmapped_mixins = False
for base in cls.__bases__:
names = dir(base)
if not _is_mapped_class(base):
unmapped_mixins = True
for name in names:
obj = getattr(base,name, None)
if isinstance(obj, Column):
if obj.foreign_keys:
raise exceptions.InvalidRequestError(
"Columns with foreign keys to other columns "
"are not allowed on declarative mixins at this time."
)
dict_[name]=column_copies[obj]=obj.copy()
elif isinstance(obj, RelationshipProperty):
raise exceptions.InvalidRequestError(
"relationships are not allowed on "
"declarative mixins at this time.")
# doing it this way enables these attributes to be descriptors
get_mapper_args = '__mapper_args__' in dict_
get_table_args = '__table_args__' in dict_
if unmapped_mixins:
get_mapper_args = get_mapper_args or getattr(cls,'__mapper_args__',None)
get_table_args = get_table_args or getattr(cls,'__table_args__',None)
tablename = getattr(cls,'__tablename__',None)
if tablename:
# subtle: if tablename is a descriptor here, we actually
# put the wrong value in, but it serves as a marker to get
# the right value value...
dict_['__tablename__']=tablename
# now that we know whether or not to get these, get them from the class
# if we should, enabling them to be decorators
mapper_args = get_mapper_args and cls.__mapper_args__ or {}
table_args = get_table_args and cls.__table_args__ or None
# make sure that column copies are used rather than the original columns
# from any mixins
for k, v in mapper_args.iteritems():
mapper_args[k] = column_copies.get(v,v)
cls._decl_class_registry[classname] = cls
our_stuff = util.OrderedDict()
for k in dict_:
value = dict_[k]
if (isinstance(value, tuple) and len(value) == 1 and
isinstance(value[0], (Column, MapperProperty))):
util.warn("Ignoring declarative-like tuple value of attribute "
"%s: possibly a copy-and-paste error with a comma "
"left at the end of the line?" % k)
continue
if not isinstance(value, (Column, MapperProperty)):
continue
prop = _deferred_relationship(cls, value)
our_stuff[k] = prop
# set up attributes in the order they were created
our_stuff.sort(key=lambda key: our_stuff[key]._creation_order)
# extract columns from the class dict
cols = []
for key, c in our_stuff.iteritems():
if isinstance(c, ColumnProperty):
for col in c.columns:
if isinstance(col, Column) and col.table is None:
_undefer_column_name(key, col)
cols.append(col)
elif isinstance(c, Column):
_undefer_column_name(key, c)
cols.append(c)
# if the column is the same name as the key,
# remove it from the explicit properties dict.
# the normal rules for assigning column-based properties
# will take over, including precedence of columns
# in multi-column ColumnProperties.
if key == c.key:
del our_stuff[key]
table = None
if '__table__' not in dict_:
if '__tablename__' in dict_:
# see above: if __tablename__ is a descriptor, this
# means we get the right value used!
tablename = cls.__tablename__
if isinstance(table_args, dict):
args, table_kw = (), table_args
elif isinstance(table_args, tuple):
args = table_args[0:-1]
table_kw = table_args[-1]
if len(table_args) < 2 or not isinstance(table_kw, dict):
raise exceptions.ArgumentError(
"Tuple form of __table_args__ is "
"(arg1, arg2, arg3, ..., {'kw1':val1, 'kw2':val2, ...})"
)
else:
args, table_kw = (), {}
autoload = dict_.get('__autoload__')
if autoload:
table_kw['autoload'] = True
cls.__table__ = table = Table(tablename, cls.metadata,
*(tuple(cols) + tuple(args)), **table_kw)
else:
table = cls.__table__
if cols:
for c in cols:
if not table.c.contains_column(c):
raise exceptions.ArgumentError(
"Can't add additional column %r when specifying __table__" % key
)
if 'inherits' not in mapper_args:
for c in cls.__bases__:
if _is_mapped_class(c):
mapper_args['inherits'] = cls._decl_class_registry.get(c.__name__, None)
break
if hasattr(cls, '__mapper_cls__'):
mapper_cls = util.unbound_method_to_callable(cls.__mapper_cls__)
else:
mapper_cls = mapper
if table is None and 'inherits' not in mapper_args:
raise exceptions.InvalidRequestError(
"Class %r does not have a __table__ or __tablename__ "
"specified and does not inherit from an existing table-mapped class." % cls
)
elif 'inherits' in mapper_args and not mapper_args.get('concrete', False):
inherited_mapper = class_mapper(mapper_args['inherits'], compile=False)
inherited_table = inherited_mapper.local_table
if 'inherit_condition' not in mapper_args and table is not None:
# figure out the inherit condition with relaxed rules
# about nonexistent tables, to allow for ForeignKeys to
# not-yet-defined tables (since we know for sure that our
# parent table is defined within the same MetaData)
mapper_args['inherit_condition'] = sql_util.join_condition(
mapper_args['inherits'].__table__, table,
ignore_nonexistent_tables=True)
if table is None:
# single table inheritance.
# ensure no table args
if table_args is not None:
raise exceptions.ArgumentError(
"Can't place __table_args__ on an inherited class with no table."
)
# add any columns declared here to the inherited table.
for c in cols:
if c.primary_key:
raise exceptions.ArgumentError(
"Can't place primary key columns on an inherited class with no table."
)
if c.name in inherited_table.c:
raise exceptions.ArgumentError(
"Column '%s' on class %s conflicts with existing column '%s'" %
(c, cls, inherited_table.c[c.name])
)
inherited_table.append_column(c)
# single or joined inheritance
# exclude any cols on the inherited table which are not mapped on the
# parent class, to avoid
# mapping columns specific to sibling/nephew classes
inherited_mapper = class_mapper(mapper_args['inherits'], compile=False)
inherited_table = inherited_mapper.local_table
if 'exclude_properties' not in mapper_args:
mapper_args['exclude_properties'] = exclude_properties = \
set([c.key for c in inherited_table.c
if c not in inherited_mapper._columntoproperty])
exclude_properties.difference_update([c.key for c in cols])
cls.__mapper__ = mapper_cls(cls, table, properties=our_stuff, **mapper_args)
class DeclarativeMeta(type):
def __init__(cls, classname, bases, dict_):
if '_decl_class_registry' in cls.__dict__:
return type.__init__(cls, classname, bases, dict_)
_as_declarative(cls, classname, cls.__dict__)
return type.__init__(cls, classname, bases, dict_)
def __setattr__(cls, key, value):
if '__mapper__' in cls.__dict__:
if isinstance(value, Column):
_undefer_column_name(key, value)
cls.__table__.append_column(value)
cls.__mapper__.add_property(key, value)
elif isinstance(value, ColumnProperty):
for col in value.columns:
if isinstance(col, Column) and col.table is None:
_undefer_column_name(key, col)
cls.__table__.append_column(col)
cls.__mapper__.add_property(key, value)
elif isinstance(value, MapperProperty):
cls.__mapper__.add_property(key, _deferred_relationship(cls, value))
else:
type.__setattr__(cls, key, value)
else:
type.__setattr__(cls, key, value)
class _GetColumns(object):
def __init__(self, cls):
self.cls = cls
def __getattr__(self, key):
mapper = class_mapper(self.cls, compile=False)
if mapper:
prop = mapper.get_property(key)
if not isinstance(prop, ColumnProperty):
raise exceptions.InvalidRequestError(
"Property %r is not an instance of"
" ColumnProperty (i.e. does not correspond"
" directly to a Column)." % key)
return getattr(self.cls, key)
def _deferred_relationship(cls, prop):
def resolve_arg(arg):
import sqlalchemy
def access_cls(key):
if key in cls._decl_class_registry:
return _GetColumns(cls._decl_class_registry[key])
elif key in cls.metadata.tables:
return cls.metadata.tables[key]
else:
return sqlalchemy.__dict__[key]
d = util.PopulateDict(access_cls)
def return_cls():
try:
x = eval(arg, globals(), d)
if isinstance(x, _GetColumns):
return x.cls
else:
return x
except NameError, n:
raise exceptions.InvalidRequestError(
"When compiling mapper %s, expression %r failed to locate a name (%r). "
"If this is a class name, consider adding this relationship() to the %r "
"class after both dependent classes have been defined." % (
prop.parent, arg, n.args[0], cls))
return return_cls
if isinstance(prop, RelationshipProperty):
for attr in ('argument', 'order_by', 'primaryjoin', 'secondaryjoin',
'secondary', '_foreign_keys', 'remote_side'):
v = getattr(prop, attr)
if isinstance(v, basestring):
setattr(prop, attr, resolve_arg(v))
if prop.backref and isinstance(prop.backref, tuple):
key, kwargs = prop.backref
for attr in ('primaryjoin', 'secondaryjoin', 'secondary',
'foreign_keys', 'remote_side', 'order_by'):
if attr in kwargs and isinstance(kwargs[attr], basestring):
kwargs[attr] = resolve_arg(kwargs[attr])
return prop
def synonym_for(name, map_column=False):
"""Decorator, make a Python @property a query synonym for a column.
A decorator version of :func:`~sqlalchemy.orm.synonym`. The function being
decorated is the 'descriptor', otherwise passes its arguments through
to synonym()::
@synonym_for('col')
@property
def prop(self):
return 'special sauce'
The regular ``synonym()`` is also usable directly in a declarative setting
and may be convenient for read/write properties::
prop = synonym('col', descriptor=property(_read_prop, _write_prop))
"""
def decorate(fn):
return _orm_synonym(name, map_column=map_column, descriptor=fn)
return decorate
def comparable_using(comparator_factory):
"""Decorator, allow a Python @property to be used in query criteria.
This is a decorator front end to
:func:`~sqlalchemy.orm.comparable_property` that passes
through the comparator_factory and the function being decorated::
@comparable_using(MyComparatorType)
@property
def prop(self):
return 'special sauce'
The regular ``comparable_property()`` is also usable directly in a
declarative setting and may be convenient for read/write properties::
prop = comparable_property(MyComparatorType)
"""
def decorate(fn):
return comparable_property(comparator_factory, fn)
return decorate
def _declarative_constructor(self, **kwargs):
"""A simple constructor that allows initialization from kwargs.
Sets attributes on the constructed instance using the names and
values in ``kwargs``.
Only keys that are present as
attributes of the instance's class are allowed. These could be,
for example, any mapped columns or relationships.
"""
for k in kwargs:
if not hasattr(type(self), k):
raise TypeError(
"%r is an invalid keyword argument for %s" %
(k, type(self).__name__))
setattr(self, k, kwargs[k])
_declarative_constructor.__name__ = '__init__'
def declarative_base(bind=None, metadata=None, mapper=None, cls=object,
name='Base', constructor=_declarative_constructor,
metaclass=DeclarativeMeta):
"""Construct a base class for declarative class definitions.
The new base class will be given a metaclass that produces
appropriate :class:`~sqlalchemy.schema.Table` objects and makes
the appropriate :func:`~sqlalchemy.orm.mapper` calls based on the
information provided declaratively in the class and any subclasses
of the class.
:param bind: An optional
:class:`~sqlalchemy.engine.base.Connectable`, will be assigned
the ``bind`` attribute on the :class:`~sqlalchemy.MetaData`
instance.
:param metadata:
An optional :class:`~sqlalchemy.MetaData` instance. All
:class:`~sqlalchemy.schema.Table` objects implicitly declared by
subclasses of the base will share this MetaData. A MetaData instance
will be created if none is provided. The
:class:`~sqlalchemy.MetaData` instance will be available via the
`metadata` attribute of the generated declarative base class.
:param mapper:
An optional callable, defaults to :func:`~sqlalchemy.orm.mapper`. Will be
used to map subclasses to their Tables.
:param cls:
Defaults to :class:`object`. A type to use as the base for the generated
declarative base class. May be a class or tuple of classes.
:param name:
Defaults to ``Base``. The display name for the generated
class. Customizing this is not required, but can improve clarity in
tracebacks and debugging.
:param constructor:
Defaults to
:func:`~sqlalchemy.ext.declarative._declarative_constructor`, an
__init__ implementation that assigns \**kwargs for declared
fields and relationships to an instance. If ``None`` is supplied,
no __init__ will be provided and construction will fall back to
cls.__init__ by way of the normal Python semantics.
:param metaclass:
Defaults to :class:`DeclarativeMeta`. A metaclass or __metaclass__
compatible callable to use as the meta type of the generated
declarative base class.
"""
lcl_metadata = metadata or MetaData()
if bind:
lcl_metadata.bind = bind
bases = not isinstance(cls, tuple) and (cls,) or cls
class_dict = dict(_decl_class_registry=dict(),
metadata=lcl_metadata)
if constructor:
class_dict['__init__'] = constructor
if mapper:
class_dict['__mapper_cls__'] = mapper
return metaclass(name, bases, class_dict)
def _undefer_column_name(key, column):
if column.key is None:
column.key = key
if column.name is None:
column.name = key

View File

@@ -0,0 +1,125 @@
# horizontal_shard.py
# Copyright (C) the SQLAlchemy authors and contributors
#
# This module is part of SQLAlchemy and is released under
# the MIT License: http://www.opensource.org/licenses/mit-license.php
"""Horizontal sharding support.
Defines a rudimental 'horizontal sharding' system which allows a Session to
distribute queries and persistence operations across multiple databases.
For a usage example, see the :ref:`examples_sharding` example included in
the source distrbution.
"""
import sqlalchemy.exceptions as sa_exc
from sqlalchemy import util
from sqlalchemy.orm.session import Session
from sqlalchemy.orm.query import Query
__all__ = ['ShardedSession', 'ShardedQuery']
class ShardedSession(Session):
def __init__(self, shard_chooser, id_chooser, query_chooser, shards=None, **kwargs):
"""Construct a ShardedSession.
:param shard_chooser: A callable which, passed a Mapper, a mapped instance, and possibly a
SQL clause, returns a shard ID. This id may be based off of the
attributes present within the object, or on some round-robin
scheme. If the scheme is based on a selection, it should set
whatever state on the instance to mark it in the future as
participating in that shard.
:param id_chooser: A callable, passed a query and a tuple of identity values, which
should return a list of shard ids where the ID might reside. The
databases will be queried in the order of this listing.
:param query_chooser: For a given Query, returns the list of shard_ids where the query
should be issued. Results from all shards returned will be combined
together into a single listing.
:param shards: A dictionary of string shard names to :class:`~sqlalchemy.engine.base.Engine`
objects.
"""
super(ShardedSession, self).__init__(**kwargs)
self.shard_chooser = shard_chooser
self.id_chooser = id_chooser
self.query_chooser = query_chooser
self.__binds = {}
self._mapper_flush_opts = {'connection_callable':self.connection}
self._query_cls = ShardedQuery
if shards is not None:
for k in shards:
self.bind_shard(k, shards[k])
def connection(self, mapper=None, instance=None, shard_id=None, **kwargs):
if shard_id is None:
shard_id = self.shard_chooser(mapper, instance)
if self.transaction is not None:
return self.transaction.connection(mapper, shard_id=shard_id)
else:
return self.get_bind(mapper,
shard_id=shard_id,
instance=instance).contextual_connect(**kwargs)
def get_bind(self, mapper, shard_id=None, instance=None, clause=None, **kw):
if shard_id is None:
shard_id = self.shard_chooser(mapper, instance, clause=clause)
return self.__binds[shard_id]
def bind_shard(self, shard_id, bind):
self.__binds[shard_id] = bind
class ShardedQuery(Query):
def __init__(self, *args, **kwargs):
super(ShardedQuery, self).__init__(*args, **kwargs)
self.id_chooser = self.session.id_chooser
self.query_chooser = self.session.query_chooser
self._shard_id = None
def set_shard(self, shard_id):
"""return a new query, limited to a single shard ID.
all subsequent operations with the returned query will
be against the single shard regardless of other state.
"""
q = self._clone()
q._shard_id = shard_id
return q
def _execute_and_instances(self, context):
if self._shard_id is not None:
result = self.session.connection(
mapper=self._mapper_zero(),
shard_id=self._shard_id).execute(context.statement, self._params)
return self.instances(result, context)
else:
partial = []
for shard_id in self.query_chooser(self):
result = self.session.connection(
mapper=self._mapper_zero(),
shard_id=shard_id).execute(context.statement, self._params)
partial = partial + list(self.instances(result, context))
# if some kind of in memory 'sorting'
# were done, this is where it would happen
return iter(partial)
def get(self, ident, **kwargs):
if self._shard_id is not None:
return super(ShardedQuery, self).get(ident)
else:
ident = util.to_list(ident)
for shard_id in self.id_chooser(self, ident):
o = self.set_shard(shard_id).get(ident, **kwargs)
if o is not None:
return o
else:
return None

View File

@@ -0,0 +1,315 @@
"""A custom list that manages index/position information for its children.
:author: Jason Kirtland
``orderinglist`` is a helper for mutable ordered relationships. It will intercept
list operations performed on a relationship collection and automatically
synchronize changes in list position with an attribute on the related objects.
(See :ref:`advdatamapping_entitycollections` for more information on the general pattern.)
Example: Two tables that store slides in a presentation. Each slide
has a number of bullet points, displayed in order by the 'position'
column on the bullets table. These bullets can be inserted and re-ordered
by your end users, and you need to update the 'position' column of all
affected rows when changes are made.
.. sourcecode:: python+sql
slides_table = Table('Slides', metadata,
Column('id', Integer, primary_key=True),
Column('name', String))
bullets_table = Table('Bullets', metadata,
Column('id', Integer, primary_key=True),
Column('slide_id', Integer, ForeignKey('Slides.id')),
Column('position', Integer),
Column('text', String))
class Slide(object):
pass
class Bullet(object):
pass
mapper(Slide, slides_table, properties={
'bullets': relationship(Bullet, order_by=[bullets_table.c.position])
})
mapper(Bullet, bullets_table)
The standard relationship mapping will produce a list-like attribute on each Slide
containing all related Bullets, but coping with changes in ordering is totally
your responsibility. If you insert a Bullet into that list, there is no
magic- it won't have a position attribute unless you assign it it one, and
you'll need to manually renumber all the subsequent Bullets in the list to
accommodate the insert.
An ``orderinglist`` can automate this and manage the 'position' attribute on all
related bullets for you.
.. sourcecode:: python+sql
mapper(Slide, slides_table, properties={
'bullets': relationship(Bullet,
collection_class=ordering_list('position'),
order_by=[bullets_table.c.position])
})
mapper(Bullet, bullets_table)
s = Slide()
s.bullets.append(Bullet())
s.bullets.append(Bullet())
s.bullets[1].position
>>> 1
s.bullets.insert(1, Bullet())
s.bullets[2].position
>>> 2
Use the ``ordering_list`` function to set up the ``collection_class`` on relationships
(as in the mapper example above). This implementation depends on the list
starting in the proper order, so be SURE to put an order_by on your relationship.
.. warning:: ``ordering_list`` only provides limited functionality when a primary
key column or unique column is the target of the sort. Since changing the order of
entries often means that two rows must trade values, this is not possible when
the value is constrained by a primary key or unique constraint, since one of the rows
would temporarily have to point to a third available value so that the other row
could take its old value. ``ordering_list`` doesn't do any of this for you,
nor does SQLAlchemy itself.
``ordering_list`` takes the name of the related object's ordering attribute as
an argument. By default, the zero-based integer index of the object's
position in the ``ordering_list`` is synchronized with the ordering attribute:
index 0 will get position 0, index 1 position 1, etc. To start numbering at 1
or some other integer, provide ``count_from=1``.
Ordering values are not limited to incrementing integers. Almost any scheme
can implemented by supplying a custom ``ordering_func`` that maps a Python list
index to any value you require.
"""
from sqlalchemy.orm.collections import collection
from sqlalchemy import util
__all__ = [ 'ordering_list' ]
def ordering_list(attr, count_from=None, **kw):
"""Prepares an OrderingList factory for use in mapper definitions.
Returns an object suitable for use as an argument to a Mapper relationship's
``collection_class`` option. Arguments are:
attr
Name of the mapped attribute to use for storage and retrieval of
ordering information
count_from (optional)
Set up an integer-based ordering, starting at ``count_from``. For
example, ``ordering_list('pos', count_from=1)`` would create a 1-based
list in SQL, storing the value in the 'pos' column. Ignored if
``ordering_func`` is supplied.
Passes along any keyword arguments to ``OrderingList`` constructor.
"""
kw = _unsugar_count_from(count_from=count_from, **kw)
return lambda: OrderingList(attr, **kw)
# Ordering utility functions
def count_from_0(index, collection):
"""Numbering function: consecutive integers starting at 0."""
return index
def count_from_1(index, collection):
"""Numbering function: consecutive integers starting at 1."""
return index + 1
def count_from_n_factory(start):
"""Numbering function: consecutive integers starting at arbitrary start."""
def f(index, collection):
return index + start
try:
f.__name__ = 'count_from_%i' % start
except TypeError:
pass
return f
def _unsugar_count_from(**kw):
"""Builds counting functions from keywrod arguments.
Keyword argument filter, prepares a simple ``ordering_func`` from a
``count_from`` argument, otherwise passes ``ordering_func`` on unchanged.
"""
count_from = kw.pop('count_from', None)
if kw.get('ordering_func', None) is None and count_from is not None:
if count_from == 0:
kw['ordering_func'] = count_from_0
elif count_from == 1:
kw['ordering_func'] = count_from_1
else:
kw['ordering_func'] = count_from_n_factory(count_from)
return kw
class OrderingList(list):
"""A custom list that manages position information for its children.
See the module and __init__ documentation for more details. The
``ordering_list`` factory function is used to configure ``OrderingList``
collections in ``mapper`` relationship definitions.
"""
def __init__(self, ordering_attr=None, ordering_func=None,
reorder_on_append=False):
"""A custom list that manages position information for its children.
``OrderingList`` is a ``collection_class`` list implementation that
syncs position in a Python list with a position attribute on the
mapped objects.
This implementation relies on the list starting in the proper order,
so be **sure** to put an ``order_by`` on your relationship.
ordering_attr
Name of the attribute that stores the object's order in the
relationship.
ordering_func
Optional. A function that maps the position in the Python list to a
value to store in the ``ordering_attr``. Values returned are
usually (but need not be!) integers.
An ``ordering_func`` is called with two positional parameters: the
index of the element in the list, and the list itself.
If omitted, Python list indexes are used for the attribute values.
Two basic pre-built numbering functions are provided in this module:
``count_from_0`` and ``count_from_1``. For more exotic examples
like stepped numbering, alphabetical and Fibonacci numbering, see
the unit tests.
reorder_on_append
Default False. When appending an object with an existing (non-None)
ordering value, that value will be left untouched unless
``reorder_on_append`` is true. This is an optimization to avoid a
variety of dangerous unexpected database writes.
SQLAlchemy will add instances to the list via append() when your
object loads. If for some reason the result set from the database
skips a step in the ordering (say, row '1' is missing but you get
'2', '3', and '4'), reorder_on_append=True would immediately
renumber the items to '1', '2', '3'. If you have multiple sessions
making changes, any of whom happen to load this collection even in
passing, all of the sessions would try to "clean up" the numbering
in their commits, possibly causing all but one to fail with a
concurrent modification error. Spooky action at a distance.
Recommend leaving this with the default of False, and just call
``reorder()`` if you're doing ``append()`` operations with
previously ordered instances or when doing some housekeeping after
manual sql operations.
"""
self.ordering_attr = ordering_attr
if ordering_func is None:
ordering_func = count_from_0
self.ordering_func = ordering_func
self.reorder_on_append = reorder_on_append
# More complex serialization schemes (multi column, e.g.) are possible by
# subclassing and reimplementing these two methods.
def _get_order_value(self, entity):
return getattr(entity, self.ordering_attr)
def _set_order_value(self, entity, value):
setattr(entity, self.ordering_attr, value)
def reorder(self):
"""Synchronize ordering for the entire collection.
Sweeps through the list and ensures that each object has accurate
ordering information set.
"""
for index, entity in enumerate(self):
self._order_entity(index, entity, True)
# As of 0.5, _reorder is no longer semi-private
_reorder = reorder
def _order_entity(self, index, entity, reorder=True):
have = self._get_order_value(entity)
# Don't disturb existing ordering if reorder is False
if have is not None and not reorder:
return
should_be = self.ordering_func(index, self)
if have != should_be:
self._set_order_value(entity, should_be)
def append(self, entity):
super(OrderingList, self).append(entity)
self._order_entity(len(self) - 1, entity, self.reorder_on_append)
def _raw_append(self, entity):
"""Append without any ordering behavior."""
super(OrderingList, self).append(entity)
_raw_append = collection.adds(1)(_raw_append)
def insert(self, index, entity):
super(OrderingList, self).insert(index, entity)
self._reorder()
def remove(self, entity):
super(OrderingList, self).remove(entity)
self._reorder()
def pop(self, index=-1):
entity = super(OrderingList, self).pop(index)
self._reorder()
return entity
def __setitem__(self, index, entity):
if isinstance(index, slice):
step = index.step or 1
start = index.start or 0
if start < 0:
start += len(self)
stop = index.stop or len(self)
if stop < 0:
stop += len(self)
for i in xrange(start, stop, step):
self.__setitem__(i, entity[i])
else:
self._order_entity(index, entity, True)
super(OrderingList, self).__setitem__(index, entity)
def __delitem__(self, index):
super(OrderingList, self).__delitem__(index)
self._reorder()
# Py2K
def __setslice__(self, start, end, values):
super(OrderingList, self).__setslice__(start, end, values)
self._reorder()
def __delslice__(self, start, end):
super(OrderingList, self).__delslice__(start, end)
self._reorder()
# end Py2K
for func_name, func in locals().items():
if (util.callable(func) and func.func_name == func_name and
not func.__doc__ and hasattr(list, func_name)):
func.__doc__ = getattr(list, func_name).__doc__
del func_name, func

View File

@@ -0,0 +1,155 @@
"""Serializer/Deserializer objects for usage with SQLAlchemy query structures,
allowing "contextual" deserialization.
Any SQLAlchemy query structure, either based on sqlalchemy.sql.*
or sqlalchemy.orm.* can be used. The mappers, Tables, Columns, Session
etc. which are referenced by the structure are not persisted in serialized
form, but are instead re-associated with the query structure
when it is deserialized.
Usage is nearly the same as that of the standard Python pickle module::
from sqlalchemy.ext.serializer import loads, dumps
metadata = MetaData(bind=some_engine)
Session = scoped_session(sessionmaker())
# ... define mappers
query = Session.query(MyClass).filter(MyClass.somedata=='foo').order_by(MyClass.sortkey)
# pickle the query
serialized = dumps(query)
# unpickle. Pass in metadata + scoped_session
query2 = loads(serialized, metadata, Session)
print query2.all()
Similar restrictions as when using raw pickle apply; mapped classes must be
themselves be pickleable, meaning they are importable from a module-level
namespace.
The serializer module is only appropriate for query structures. It is not
needed for:
* instances of user-defined classes. These contain no references to engines,
sessions or expression constructs in the typical case and can be serialized directly.
* Table metadata that is to be loaded entirely from the serialized structure (i.e. is
not already declared in the application). Regular pickle.loads()/dumps() can
be used to fully dump any ``MetaData`` object, typically one which was reflected
from an existing database at some previous point in time. The serializer module
is specifically for the opposite case, where the Table metadata is already present
in memory.
"""
from sqlalchemy.orm import class_mapper, Query
from sqlalchemy.orm.session import Session
from sqlalchemy.orm.mapper import Mapper
from sqlalchemy.orm.attributes import QueryableAttribute
from sqlalchemy import Table, Column
from sqlalchemy.engine import Engine
from sqlalchemy.util import pickle
import re
import base64
# Py3K
#from io import BytesIO as byte_buffer
# Py2K
from cStringIO import StringIO as byte_buffer
# end Py2K
# Py3K
#def b64encode(x):
# return base64.b64encode(x).decode('ascii')
#def b64decode(x):
# return base64.b64decode(x.encode('ascii'))
# Py2K
b64encode = base64.b64encode
b64decode = base64.b64decode
# end Py2K
__all__ = ['Serializer', 'Deserializer', 'dumps', 'loads']
def Serializer(*args, **kw):
pickler = pickle.Pickler(*args, **kw)
def persistent_id(obj):
#print "serializing:", repr(obj)
if isinstance(obj, QueryableAttribute):
cls = obj.impl.class_
key = obj.impl.key
id = "attribute:" + key + ":" + b64encode(pickle.dumps(cls))
elif isinstance(obj, Mapper) and not obj.non_primary:
id = "mapper:" + b64encode(pickle.dumps(obj.class_))
elif isinstance(obj, Table):
id = "table:" + str(obj)
elif isinstance(obj, Column) and isinstance(obj.table, Table):
id = "column:" + str(obj.table) + ":" + obj.key
elif isinstance(obj, Session):
id = "session:"
elif isinstance(obj, Engine):
id = "engine:"
else:
return None
return id
pickler.persistent_id = persistent_id
return pickler
our_ids = re.compile(r'(mapper|table|column|session|attribute|engine):(.*)')
def Deserializer(file, metadata=None, scoped_session=None, engine=None):
unpickler = pickle.Unpickler(file)
def get_engine():
if engine:
return engine
elif scoped_session and scoped_session().bind:
return scoped_session().bind
elif metadata and metadata.bind:
return metadata.bind
else:
return None
def persistent_load(id):
m = our_ids.match(id)
if not m:
return None
else:
type_, args = m.group(1, 2)
if type_ == 'attribute':
key, clsarg = args.split(":")
cls = pickle.loads(b64decode(clsarg))
return getattr(cls, key)
elif type_ == "mapper":
cls = pickle.loads(b64decode(args))
return class_mapper(cls)
elif type_ == "table":
return metadata.tables[args]
elif type_ == "column":
table, colname = args.split(':')
return metadata.tables[table].c[colname]
elif type_ == "session":
return scoped_session()
elif type_ == "engine":
return get_engine()
else:
raise Exception("Unknown token: %s" % type_)
unpickler.persistent_load = persistent_load
return unpickler
def dumps(obj, protocol=0):
buf = byte_buffer()
pickler = Serializer(buf, protocol)
pickler.dump(obj)
return buf.getvalue()
def loads(data, metadata=None, scoped_session=None, engine=None):
buf = byte_buffer(data)
unpickler = Deserializer(buf, metadata, scoped_session, engine)
return unpickler.load()

551
sqlalchemy/ext/sqlsoup.py Normal file
View File

@@ -0,0 +1,551 @@
"""
Introduction
============
SqlSoup provides a convenient way to access existing database tables without
having to declare table or mapper classes ahead of time. It is built on top of the SQLAlchemy ORM and provides a super-minimalistic interface to an existing database.
Suppose we have a database with users, books, and loans tables
(corresponding to the PyWebOff dataset, if you're curious).
Creating a SqlSoup gateway is just like creating an SQLAlchemy
engine::
>>> from sqlalchemy.ext.sqlsoup import SqlSoup
>>> db = SqlSoup('sqlite:///:memory:')
or, you can re-use an existing engine::
>>> db = SqlSoup(engine)
You can optionally specify a schema within the database for your
SqlSoup::
>>> db.schema = myschemaname
Loading objects
===============
Loading objects is as easy as this::
>>> users = db.users.all()
>>> users.sort()
>>> users
[MappedUsers(name=u'Joe Student',email=u'student@example.edu',password=u'student',classname=None,admin=0), MappedUsers(name=u'Bhargan Basepair',email=u'basepair@example.edu',password=u'basepair',classname=None,admin=1)]
Of course, letting the database do the sort is better::
>>> db.users.order_by(db.users.name).all()
[MappedUsers(name=u'Bhargan Basepair',email=u'basepair@example.edu',password=u'basepair',classname=None,admin=1), MappedUsers(name=u'Joe Student',email=u'student@example.edu',password=u'student',classname=None,admin=0)]
Field access is intuitive::
>>> users[0].email
u'student@example.edu'
Of course, you don't want to load all users very often. Let's add a
WHERE clause. Let's also switch the order_by to DESC while we're at
it::
>>> from sqlalchemy import or_, and_, desc
>>> where = or_(db.users.name=='Bhargan Basepair', db.users.email=='student@example.edu')
>>> db.users.filter(where).order_by(desc(db.users.name)).all()
[MappedUsers(name=u'Joe Student',email=u'student@example.edu',password=u'student',classname=None,admin=0), MappedUsers(name=u'Bhargan Basepair',email=u'basepair@example.edu',password=u'basepair',classname=None,admin=1)]
You can also use .first() (to retrieve only the first object from a query) or
.one() (like .first when you expect exactly one user -- it will raise an
exception if more were returned)::
>>> db.users.filter(db.users.name=='Bhargan Basepair').one()
MappedUsers(name=u'Bhargan Basepair',email=u'basepair@example.edu',password=u'basepair',classname=None,admin=1)
Since name is the primary key, this is equivalent to
>>> db.users.get('Bhargan Basepair')
MappedUsers(name=u'Bhargan Basepair',email=u'basepair@example.edu',password=u'basepair',classname=None,admin=1)
This is also equivalent to
>>> db.users.filter_by(name='Bhargan Basepair').one()
MappedUsers(name=u'Bhargan Basepair',email=u'basepair@example.edu',password=u'basepair',classname=None,admin=1)
filter_by is like filter, but takes kwargs instead of full clause expressions.
This makes it more concise for simple queries like this, but you can't do
complex queries like the or\_ above or non-equality based comparisons this way.
Full query documentation
------------------------
Get, filter, filter_by, order_by, limit, and the rest of the
query methods are explained in detail in :ref:`ormtutorial_querying`.
Modifying objects
=================
Modifying objects is intuitive::
>>> user = _
>>> user.email = 'basepair+nospam@example.edu'
>>> db.commit()
(SqlSoup leverages the sophisticated SQLAlchemy unit-of-work code, so
multiple updates to a single object will be turned into a single
``UPDATE`` statement when you commit.)
To finish covering the basics, let's insert a new loan, then delete
it::
>>> book_id = db.books.filter_by(title='Regional Variation in Moss').first().id
>>> db.loans.insert(book_id=book_id, user_name=user.name)
MappedLoans(book_id=2,user_name=u'Bhargan Basepair',loan_date=None)
>>> loan = db.loans.filter_by(book_id=2, user_name='Bhargan Basepair').one()
>>> db.delete(loan)
>>> db.commit()
You can also delete rows that have not been loaded as objects. Let's
do our insert/delete cycle once more, this time using the loans
table's delete method. (For SQLAlchemy experts: note that no flush()
call is required since this delete acts at the SQL level, not at the
Mapper level.) The same where-clause construction rules apply here as
to the select methods.
::
>>> db.loans.insert(book_id=book_id, user_name=user.name)
MappedLoans(book_id=2,user_name=u'Bhargan Basepair',loan_date=None)
>>> db.loans.delete(db.loans.book_id==2)
You can similarly update multiple rows at once. This will change the
book_id to 1 in all loans whose book_id is 2::
>>> db.loans.update(db.loans.book_id==2, book_id=1)
>>> db.loans.filter_by(book_id=1).all()
[MappedLoans(book_id=1,user_name=u'Joe Student',loan_date=datetime.datetime(2006, 7, 12, 0, 0))]
Joins
=====
Occasionally, you will want to pull out a lot of data from related
tables all at once. In this situation, it is far more efficient to
have the database perform the necessary join. (Here we do not have *a
lot of data* but hopefully the concept is still clear.) SQLAlchemy is
smart enough to recognize that loans has a foreign key to users, and
uses that as the join condition automatically.
::
>>> join1 = db.join(db.users, db.loans, isouter=True)
>>> join1.filter_by(name='Joe Student').all()
[MappedJoin(name=u'Joe Student',email=u'student@example.edu',password=u'student',classname=None,admin=0,book_id=1,user_name=u'Joe Student',loan_date=datetime.datetime(2006, 7, 12, 0, 0))]
If you're unfortunate enough to be using MySQL with the default MyISAM
storage engine, you'll have to specify the join condition manually,
since MyISAM does not store foreign keys. Here's the same join again,
with the join condition explicitly specified::
>>> db.join(db.users, db.loans, db.users.name==db.loans.user_name, isouter=True)
<class 'sqlalchemy.ext.sqlsoup.MappedJoin'>
You can compose arbitrarily complex joins by combining Join objects
with tables or other joins. Here we combine our first join with the
books table::
>>> join2 = db.join(join1, db.books)
>>> join2.all()
[MappedJoin(name=u'Joe Student',email=u'student@example.edu',password=u'student',classname=None,admin=0,book_id=1,user_name=u'Joe Student',loan_date=datetime.datetime(2006, 7, 12, 0, 0),id=1,title=u'Mustards I Have Known',published_year=u'1989',authors=u'Jones')]
If you join tables that have an identical column name, wrap your join
with `with_labels`, to disambiguate columns with their table name
(.c is short for .columns)::
>>> db.with_labels(join1).c.keys()
[u'users_name', u'users_email', u'users_password', u'users_classname', u'users_admin', u'loans_book_id', u'loans_user_name', u'loans_loan_date']
You can also join directly to a labeled object::
>>> labeled_loans = db.with_labels(db.loans)
>>> db.join(db.users, labeled_loans, isouter=True).c.keys()
[u'name', u'email', u'password', u'classname', u'admin', u'loans_book_id', u'loans_user_name', u'loans_loan_date']
Relationships
=============
You can define relationships on SqlSoup classes:
>>> db.users.relate('loans', db.loans)
These can then be used like a normal SA property:
>>> db.users.get('Joe Student').loans
[MappedLoans(book_id=1,user_name=u'Joe Student',loan_date=datetime.datetime(2006, 7, 12, 0, 0))]
>>> db.users.filter(~db.users.loans.any()).all()
[MappedUsers(name=u'Bhargan Basepair',email='basepair+nospam@example.edu',password=u'basepair',classname=None,admin=1)]
relate can take any options that the relationship function accepts in normal mapper definition:
>>> del db._cache['users']
>>> db.users.relate('loans', db.loans, order_by=db.loans.loan_date, cascade='all, delete-orphan')
Advanced Use
============
Sessions, Transations and Application Integration
-------------------------------------------------
**Note:** please read and understand this section thoroughly before using SqlSoup in any web application.
SqlSoup uses a ScopedSession to provide thread-local sessions. You
can get a reference to the current one like this::
>>> session = db.session
The default session is available at the module level in SQLSoup, via::
>>> from sqlalchemy.ext.sqlsoup import Session
The configuration of this session is ``autoflush=True``, ``autocommit=False``.
This means when you work with the SqlSoup object, you need to call ``db.commit()``
in order to have changes persisted. You may also call ``db.rollback()`` to
roll things back.
Since the SqlSoup object's Session automatically enters into a transaction as soon
as it's used, it is *essential* that you call ``commit()`` or ``rollback()``
on it when the work within a thread completes. This means all the guidelines
for web application integration at :ref:`session_lifespan` must be followed.
The SqlSoup object can have any session or scoped session configured onto it.
This is of key importance when integrating with existing code or frameworks
such as Pylons. If your application already has a ``Session`` configured,
pass it to your SqlSoup object::
>>> from myapplication import Session
>>> db = SqlSoup(session=Session)
If the ``Session`` is configured with ``autocommit=True``, use ``flush()``
instead of ``commit()`` to persist changes - in this case, the ``Session``
closes out its transaction immediately and no external management is needed. ``rollback()`` is also not available. Configuring a new SQLSoup object in "autocommit" mode looks like::
>>> from sqlalchemy.orm import scoped_session, sessionmaker
>>> db = SqlSoup('sqlite://', session=scoped_session(sessionmaker(autoflush=False, expire_on_commit=False, autocommit=True)))
Mapping arbitrary Selectables
-----------------------------
SqlSoup can map any SQLAlchemy ``Selectable`` with the map
method. Let's map a ``Select`` object that uses an aggregate function;
we'll use the SQLAlchemy ``Table`` that SqlSoup introspected as the
basis. (Since we're not mapping to a simple table or join, we need to
tell SQLAlchemy how to find the *primary key* which just needs to be
unique within the select, and not necessarily correspond to a *real*
PK in the database.)
::
>>> from sqlalchemy import select, func
>>> b = db.books._table
>>> s = select([b.c.published_year, func.count('*').label('n')], from_obj=[b], group_by=[b.c.published_year])
>>> s = s.alias('years_with_count')
>>> years_with_count = db.map(s, primary_key=[s.c.published_year])
>>> years_with_count.filter_by(published_year='1989').all()
[MappedBooks(published_year=u'1989',n=1)]
Obviously if we just wanted to get a list of counts associated with
book years once, raw SQL is going to be less work. The advantage of
mapping a Select is reusability, both standalone and in Joins. (And if
you go to full SQLAlchemy, you can perform mappings like this directly
to your object models.)
An easy way to save mapped selectables like this is to just hang them on
your db object::
>>> db.years_with_count = years_with_count
Python is flexible like that!
Raw SQL
-------
SqlSoup works fine with SQLAlchemy's text construct, described in :ref:`sqlexpression_text`.
You can also execute textual SQL directly using the `execute()` method,
which corresponds to the `execute()` method on the underlying `Session`.
Expressions here are expressed like ``text()`` constructs, using named parameters
with colons::
>>> rp = db.execute('select name, email from users where name like :name order by name', name='%Bhargan%')
>>> for name, email in rp.fetchall(): print name, email
Bhargan Basepair basepair+nospam@example.edu
Or you can get at the current transaction's connection using `connection()`. This is the
raw connection object which can accept any sort of SQL expression or raw SQL string passed to the database::
>>> conn = db.connection()
>>> conn.execute("'select name, email from users where name like ? order by name'", '%Bhargan%')
Dynamic table names
-------------------
You can load a table whose name is specified at runtime with the entity() method:
>>> tablename = 'loans'
>>> db.entity(tablename) == db.loans
True
entity() also takes an optional schema argument. If none is specified, the
default schema is used.
"""
from sqlalchemy import Table, MetaData, join
from sqlalchemy import schema, sql
from sqlalchemy.engine.base import Engine
from sqlalchemy.orm import scoped_session, sessionmaker, mapper, \
class_mapper, relationship, session,\
object_session
from sqlalchemy.orm.interfaces import MapperExtension, EXT_CONTINUE
from sqlalchemy.exceptions import SQLAlchemyError, InvalidRequestError, ArgumentError
from sqlalchemy.sql import expression
__all__ = ['PKNotFoundError', 'SqlSoup']
Session = scoped_session(sessionmaker(autoflush=True, autocommit=False))
class AutoAdd(MapperExtension):
def __init__(self, scoped_session):
self.scoped_session = scoped_session
def instrument_class(self, mapper, class_):
class_.__init__ = self._default__init__(mapper)
def _default__init__(ext, mapper):
def __init__(self, **kwargs):
for key, value in kwargs.iteritems():
setattr(self, key, value)
return __init__
def init_instance(self, mapper, class_, oldinit, instance, args, kwargs):
session = self.scoped_session()
session._save_without_cascade(instance)
return EXT_CONTINUE
def init_failed(self, mapper, class_, oldinit, instance, args, kwargs):
sess = object_session(instance)
if sess:
sess.expunge(instance)
return EXT_CONTINUE
class PKNotFoundError(SQLAlchemyError):
pass
def _ddl_error(cls):
msg = 'SQLSoup can only modify mapped Tables (found: %s)' \
% cls._table.__class__.__name__
raise InvalidRequestError(msg)
# metaclass is necessary to expose class methods with getattr, e.g.
# we want to pass db.users.select through to users._mapper.select
class SelectableClassType(type):
def insert(cls, **kwargs):
_ddl_error(cls)
def __clause_element__(cls):
return cls._table
def __getattr__(cls, attr):
if attr == '_query':
# called during mapper init
raise AttributeError()
return getattr(cls._query, attr)
class TableClassType(SelectableClassType):
def insert(cls, **kwargs):
o = cls()
o.__dict__.update(kwargs)
return o
def relate(cls, propname, *args, **kwargs):
class_mapper(cls)._configure_property(propname, relationship(*args, **kwargs))
def _is_outer_join(selectable):
if not isinstance(selectable, sql.Join):
return False
if selectable.isouter:
return True
return _is_outer_join(selectable.left) or _is_outer_join(selectable.right)
def _selectable_name(selectable):
if isinstance(selectable, sql.Alias):
return _selectable_name(selectable.element)
elif isinstance(selectable, sql.Select):
return ''.join(_selectable_name(s) for s in selectable.froms)
elif isinstance(selectable, schema.Table):
return selectable.name.capitalize()
else:
x = selectable.__class__.__name__
if x[0] == '_':
x = x[1:]
return x
def _class_for_table(session, engine, selectable, **mapper_kwargs):
selectable = expression._clause_element_as_expr(selectable)
mapname = 'Mapped' + _selectable_name(selectable)
# Py2K
if isinstance(mapname, unicode):
engine_encoding = engine.dialect.encoding
mapname = mapname.encode(engine_encoding)
# end Py2K
if isinstance(selectable, Table):
klass = TableClassType(mapname, (object,), {})
else:
klass = SelectableClassType(mapname, (object,), {})
def _compare(self, o):
L = list(self.__class__.c.keys())
L.sort()
t1 = [getattr(self, k) for k in L]
try:
t2 = [getattr(o, k) for k in L]
except AttributeError:
raise TypeError('unable to compare with %s' % o.__class__)
return t1, t2
# python2/python3 compatible system of
# __cmp__ - __lt__ + __eq__
def __lt__(self, o):
t1, t2 = _compare(self, o)
return t1 < t2
def __eq__(self, o):
t1, t2 = _compare(self, o)
return t1 == t2
def __repr__(self):
L = ["%s=%r" % (key, getattr(self, key, ''))
for key in self.__class__.c.keys()]
return '%s(%s)' % (self.__class__.__name__, ','.join(L))
for m in ['__eq__', '__repr__', '__lt__']:
setattr(klass, m, eval(m))
klass._table = selectable
klass.c = expression.ColumnCollection()
mappr = mapper(klass,
selectable,
extension=AutoAdd(session),
**mapper_kwargs)
for k in mappr.iterate_properties:
klass.c[k.key] = k.columns[0]
klass._query = session.query_property()
return klass
class SqlSoup(object):
def __init__(self, engine_or_metadata, **kw):
"""Initialize a new ``SqlSoup``.
`args` may either be an ``SQLEngine`` or a set of arguments
suitable for passing to ``create_engine``.
"""
self.session = kw.pop('session', Session)
if isinstance(engine_or_metadata, MetaData):
self._metadata = engine_or_metadata
elif isinstance(engine_or_metadata, (basestring, Engine)):
self._metadata = MetaData(engine_or_metadata)
else:
raise ArgumentError("invalid engine or metadata argument %r" % engine_or_metadata)
self._cache = {}
self.schema = None
@property
def engine(self):
return self._metadata.bind
bind = engine
def delete(self, *args, **kwargs):
self.session.delete(*args, **kwargs)
def execute(self, stmt, **params):
return self.session.execute(sql.text(stmt, bind=self.bind), **params)
@property
def _underlying_session(self):
if isinstance(self.session, session.Session):
return self.session
else:
return self.session()
def connection(self):
return self._underlying_session._connection_for_bind(self.bind)
def flush(self):
self.session.flush()
def rollback(self):
self.session.rollback()
def commit(self):
self.session.commit()
def clear(self):
self.session.expunge_all()
def expunge(self, *args, **kw):
self.session.expunge(*args, **kw)
def expunge_all(self):
self.session.expunge_all()
def map(self, selectable, **kwargs):
try:
t = self._cache[selectable]
except KeyError:
t = _class_for_table(self.session, self.engine, selectable, **kwargs)
self._cache[selectable] = t
return t
def with_labels(self, item):
# TODO give meaningful aliases
return self.map(
expression._clause_element_as_expr(item).
select(use_labels=True).
alias('foo'))
def join(self, *args, **kwargs):
j = join(*args, **kwargs)
return self.map(j)
def entity(self, attr, schema=None):
try:
t = self._cache[attr]
except KeyError, ke:
table = Table(attr, self._metadata, autoload=True, autoload_with=self.bind, schema=schema or self.schema)
if not table.primary_key.columns:
raise PKNotFoundError('table %r does not have a primary key defined [columns: %s]' % (attr, ','.join(table.c.keys())))
if table.columns:
t = _class_for_table(self.session, self.engine, table)
else:
t = None
self._cache[attr] = t
return t
def __getattr__(self, attr):
return self.entity(attr)
def __repr__(self):
return 'SqlSoup(%r)' % self._metadata