path: root/lib/sqlalchemy/orm/mapper.py

# orm/mapper.py
# Copyright (C) 2005, 2006, 2007 Michael Bayer mike_mp@zzzcomputing.com
#
# This module is part of SQLAlchemy and is released under
# the MIT License: http://www.opensource.org/licenses/mit-license.php

import weakref, warnings, operator
from sqlalchemy import sql, util, exceptions, logging
from sqlalchemy.sql import expression
from sqlalchemy.sql import util as sqlutil
from sqlalchemy.orm import util as mapperutil
from sqlalchemy.orm.util import ExtensionCarrier
from sqlalchemy.orm import sync
from sqlalchemy.orm.interfaces import MapperProperty, EXT_CONTINUE, SynonymProperty, PropComparator
deferred_load = None

__all__ = ['Mapper', 'class_mapper', 'object_mapper', 'mapper_registry']

# a dictionary mapping classes to their primary mappers
mapper_registry = weakref.WeakKeyDictionary()

# a list of MapperExtensions that will be installed in all mappers by default
global_extensions = []

# a constant returned by get_attr_by_column to indicate
# this mapper is not handling an attribute for a particular
# column
NO_ATTRIBUTE = object()

# lock used to synchronize the "mapper compile" step
_COMPILE_MUTEX = util.threading.Lock()

# initialize these two lazily
attribute_manager = None
ColumnProperty = None

class Mapper(object):
    """Define the correlation of class attributes to database table
    columns.

    Instances of this class should be constructed via the
    ``sqlalchemy.orm.mapper()`` function.
    """

    def __init__(self,
                 class_,
                 local_table,
                 properties = None,
                 primary_key = None,
                 non_primary = False,
                 inherits = None,
                 inherit_condition = None,
                 inherit_foreign_keys = None,
                 extension = None,
                 order_by = False,
                 allow_column_override = False,
                 entity_name = None,
                 always_refresh = False,
                 version_id_col = None,
                 polymorphic_on=None,
                 _polymorphic_map=None,
                 polymorphic_identity=None,
                 polymorphic_fetch=None,
                 concrete=False,
                 select_table=None,
                 allow_null_pks=False,
                 batch=True,
                 column_prefix=None,
                 include_properties=None,
                 exclude_properties=None):
        """Construct a new mapper.

        Mappers are normally constructed via the [sqlalchemy.orm#mapper()] 
        function.  See for details.
        """

        if not issubclass(class_, object):
            raise exceptions.ArgumentError("Class '%s' is not a new-style class" % class_.__name__)

        for table in (local_table, select_table):
            if table is not None and isinstance(table, expression._SelectBaseMixin):
                # some db's, noteably postgres, dont want to select from a select
                # without an alias.  also if we make our own alias internally, then
                # the configured properties on the mapper are not matched against the alias
                # we make, theres workarounds but it starts to get really crazy (its crazy enough
                # the SQL that gets generated) so just require an alias
                raise exceptions.ArgumentError("Mapping against a Select object requires that it has a name.  Use an alias to give it a name, i.e. s = select(...).alias('myselect')")

        self.class_ = class_
        self.entity_name = entity_name
        self.class_key = ClassKey(class_, entity_name)
        self.primary_key_argument = primary_key
        self.non_primary = non_primary
        self.order_by = order_by
        self.always_refresh = always_refresh
        self.version_id_col = version_id_col
        self.concrete = concrete
        self.single = False
        self.inherits = inherits
        self.select_table = select_table
        self.local_table = local_table
        self.inherit_condition = inherit_condition
        self.inherit_foreign_keys = inherit_foreign_keys
        self.extension = extension
        self.properties = properties or {}
        self.allow_column_override = allow_column_override
        self.allow_null_pks = allow_null_pks
        self.delete_orphans = []
        self.batch = batch
        self.column_prefix = column_prefix
        self.polymorphic_on = polymorphic_on
        self._eager_loaders = util.Set()

        # our 'polymorphic identity', a string name that when located in a result set row
        # indicates this Mapper should be used to construct the object instance for that row.
        self.polymorphic_identity = polymorphic_identity

        if polymorphic_fetch not in (None, 'union', 'select', 'deferred'):
            raise exceptions.ArgumentError("Invalid option for 'polymorphic_fetch': '%s'" % polymorphic_fetch)
        if polymorphic_fetch is None:
            self.polymorphic_fetch = (self.select_table is None) and 'select' or 'union'
        else:
            self.polymorphic_fetch = polymorphic_fetch
        
        # a dictionary of 'polymorphic identity' names, associating those names with
        # Mappers that will be used to construct object instances upon a select operation.
        if _polymorphic_map is None:
            self.polymorphic_map = {}
        else:
            self.polymorphic_map = _polymorphic_map

        self.columns = self.c = util.OrderedProperties()

        self.include_properties = include_properties
        self.exclude_properties = exclude_properties

        # a set of all mappers which inherit from this one.
        self._inheriting_mappers = util.Set()

        # a second mapper that is used for selecting, if the "select_table" argument
        # was sent to this mapper.
        self.__surrogate_mapper = None

        self.__props_init = False

        self.__should_log_info = logging.is_info_enabled(self.logger)
        self.__should_log_debug = logging.is_debug_enabled(self.logger)
        
        self._compile_class()
        self._compile_extensions()
        self._compile_inheritance()
        self._compile_tables()
        self._compile_properties()
        self._compile_selectable()

        self.__log("constructed")

    def __log(self, msg):
        if self.__should_log_info:
            self.logger.info("(" + self.class_.__name__ + "|" + (self.entity_name is not None and "/%s" % self.entity_name or "") + (self.local_table and self.local_table.name or str(self.local_table)) + (not self._is_primary_mapper() and "|non-primary" or "") + ") " + msg)

    def __log_debug(self, msg):
        if self.__should_log_debug:
            self.logger.debug("(" + self.class_.__name__ + "|" + (self.entity_name is not None and "/%s" % self.entity_name or "") + (self.local_table and self.local_table.name or str(self.local_table)) + (not self._is_primary_mapper() and "|non-primary" or "") + ") " + msg)

    def _is_orphan(self, obj):
        optimistic = has_identity(obj)
        for (key,klass) in self.delete_orphans:
            if getattr(klass, key).hasparent(obj, optimistic=optimistic):
               return False
        else:
            if self.delete_orphans:
                if not has_identity(obj):
                    raise exceptions.FlushError("instance %s is an unsaved, pending instance and is an orphan (is not attached to %s)" %
                    (
                        obj,
                        ", nor ".join(["any parent '%s' instance via that classes' '%s' attribute" % (klass.__name__, key) for (key,klass) in self.delete_orphans])
                    ))
                else:
                    return True
            else:
                return False

    def get_property(self, key, resolve_synonyms=False, raiseerr=True):
        """return MapperProperty with the given key."""
        prop = self.__props.get(key, None)
        if resolve_synonyms:
            while isinstance(prop, SynonymProperty):
                prop = self.__props.get(prop.name, None)
        if prop is None and raiseerr:
            raise exceptions.InvalidRequestError("Mapper '%s' has no property '%s'" % (str(self), key))
        return prop
    
    def iterate_properties(self):
        return self.__props.itervalues()
    iterate_properties = property(iterate_properties, doc="returns an iterator of all MapperProperty objects.")
    
    def dispose(self):
        # disaable any attribute-based compilation
        self.__props_init = True
        attribute_manager.reset_class_managed(self.class_)
        if hasattr(self.class_, 'c'):
            del self.class_.c
        if hasattr(self.class_, '__init__') and hasattr(self.class_.__init__, '_oldinit'):
            if self.class_.__init__._oldinit is not None:
                self.class_.__init__ = self.class_.__init__._oldinit
            else:
                delattr(self.class_, '__init__')
        
    def compile(self):
        """Compile this mapper into its final internal format.
        """

        if self.__props_init:
            return self
        _COMPILE_MUTEX.acquire()
        try:
            # double-check inside mutex
            if self.__props_init:
                return self
            # initialize properties on all mappers
            for mapper in mapper_registry.values():
                if not mapper.__props_init:
                    mapper.__initialize_properties()

            # if we're not primary, compile us
            if self.non_primary:
                self.__initialize_properties()

            return self
        finally:
            _COMPILE_MUTEX.release()

    def _check_compile(self):
        if self.non_primary and not self.__props_init:
            self.__initialize_properties()
        return self
        
    def __initialize_properties(self):
        """Call the ``init()`` method on all ``MapperProperties``
        attached to this mapper.

        This happens after all mappers have completed compiling
        everything else up until this point, so that all dependencies
        are fully available.
        """

        self.__log("_initialize_properties() started")
        l = [(key, prop) for key, prop in self.__props.iteritems()]
        for key, prop in l:
            if getattr(prop, 'key', None) is None:
                prop.init(key, self)
        self.__log("_initialize_properties() complete")
        self.__props_init = True


    def _compile_extensions(self):
        """Go through the global_extensions list as well as the list
        of ``MapperExtensions`` specified for this ``Mapper`` and
        creates a linked list of those extensions.
        """

        extlist = util.OrderedSet()

        extension = self.extension
        if extension is not None:
            for ext_obj in util.to_list(extension):
                # local MapperExtensions have already instrumented the class
                extlist.add(ext_obj)

        for ext in global_extensions:
            if isinstance(ext, type):
                ext = ext()
            extlist.add(ext)
            ext.instrument_class(self, self.class_)
            
        self.extension = ExtensionCarrier()
        for ext in extlist:
            self.extension.append(ext)
        
    def _compile_inheritance(self):
        """Determine if this Mapper inherits from another mapper, and
        if so calculates the mapped_table for this Mapper taking the
        inherited mapper into account.

        For joined table inheritance, creates a ``SyncRule`` that will
        synchronize column values between the joined tables. also
        initializes polymorphic variables used in polymorphic loads.
        """

        if self.inherits is not None:
            if isinstance(self.inherits, type):
                self.inherits = class_mapper(self.inherits, compile=False)
            else:
                self.inherits = self.inherits
            if not issubclass(self.class_, self.inherits.class_):
                raise exceptions.ArgumentError("Class '%s' does not inherit from '%s'" % (self.class_.__name__, self.inherits.class_.__name__))
            if self._is_primary_mapper() != self.inherits._is_primary_mapper():
                np = self._is_primary_mapper() and "primary" or "non-primary"
                raise exceptions.ArgumentError("Inheritance of %s mapper for class '%s' is only allowed from a %s mapper" % (np, self.class_.__name__, np))
            # inherit_condition is optional.
            if self.local_table is None:
                self.local_table = self.inherits.local_table
                self.single = True
            if not self.local_table is self.inherits.local_table:
                if self.concrete:
                    self._synchronizer= None
                    self.mapped_table = self.local_table
                else:
                    if self.inherit_condition is None:
                        # figure out inherit condition from our table to the immediate table
                        # of the inherited mapper, not its full table which could pull in other
                        # stuff we dont want (allows test/inheritance.InheritTest4 to pass)
                        self.inherit_condition = sql.join(self.inherits.local_table, self.local_table).onclause
                    self.mapped_table = sql.join(self.inherits.mapped_table, self.local_table, self.inherit_condition)
                    # generate sync rules.  similarly to creating the on clause, specify a
                    # stricter set of tables to create "sync rules" by,based on the immediate
                    # inherited table, rather than all inherited tables
                    self._synchronizer = sync.ClauseSynchronizer(self, self, sync.ONETOMANY)
                    if self.inherit_foreign_keys:
                        fks = util.Set(self.inherit_foreign_keys)
                    else:
                        fks = None
                    self._synchronizer.compile(self.mapped_table.onclause, foreign_keys=fks)
            else:
                self._synchronizer = None
                self.mapped_table = self.local_table
            if self.polymorphic_identity is not None:
                self.inherits._add_polymorphic_mapping(self.polymorphic_identity, self)
                if self.polymorphic_on is None:
                    if self.inherits.polymorphic_on is not None:
                        self.polymorphic_on = self.mapped_table.corresponding_column(self.inherits.polymorphic_on, keys_ok=True, raiseerr=False)
                    else:
                        raise exceptions.ArgumentError("Mapper '%s' specifies a polymorphic_identity of '%s', but no mapper in it's hierarchy specifies the 'polymorphic_on' column argument" % (str(self), self.polymorphic_identity))

            if self.polymorphic_identity is not None and not self.concrete:
                self._identity_class = self.inherits._identity_class
            else:
                self._identity_class = self.class_
                
            if self.order_by is False:
                self.order_by = self.inherits.order_by
            self.polymorphic_map = self.inherits.polymorphic_map
            self.batch = self.inherits.batch
            self.inherits._inheriting_mappers.add(self)
            self.base_mapper = self.inherits.base_mapper
            self._all_tables = self.inherits._all_tables
        else:
            self._all_tables = util.Set()
            self.base_mapper = self
            self._synchronizer = None
            self.mapped_table = self.local_table
            if self.polymorphic_identity is not None:
                if self.polymorphic_on is None:
                    raise exceptions.ArgumentError("Mapper '%s' specifies a polymorphic_identity of '%s', but no mapper in it's hierarchy specifies the 'polymorphic_on' column argument" % (str(self), self.polymorphic_identity))
                self._add_polymorphic_mapping(self.polymorphic_identity, self)
            self._identity_class = self.class_
            
        if self.mapped_table is None:
            raise exceptions.ArgumentError("Mapper '%s' does not have a mapped_table specified.  (Are you using the return value of table.create()?  It no longer has a return value.)" % str(self))

        # convert polymorphic class associations to mappers
        for key in self.polymorphic_map.keys():
            if isinstance(self.polymorphic_map[key], type):
                self.polymorphic_map[key] = class_mapper(self.polymorphic_map[key])

    def _add_polymorphic_mapping(self, key, class_or_mapper, entity_name=None):
        """Add a Mapper to our *polymorphic map*."""

        if isinstance(class_or_mapper, type):
            class_or_mapper = class_mapper(class_or_mapper, entity_name=entity_name)
        self.polymorphic_map[key] = class_or_mapper

    def _compile_tables(self):
        """After the inheritance relationships have been reconciled,
        set up some more table-based instance variables and determine
        the *primary key* columns for all tables represented by this
        ``Mapper``.
        """

        # summary of the various Selectable units:
        # mapped_table - the Selectable that represents a join of the underlying Tables to be saved (or just the Table)
        # local_table - the Selectable that was passed to this Mapper's constructor, if any
        # select_table - the Selectable that will be used during queries.  if this is specified
        # as a constructor keyword argument, it takes precendence over mapped_table, otherwise its mapped_table
        # this is either select_table if it was given explicitly, or in the case of a mapper that inherits
        # its local_table
        # tables - a collection of underlying Table objects pulled from mapped_table

        if self.select_table is None:
            self.select_table = self.mapped_table

        # locate all tables contained within the "table" passed in, which
        # may be a join or other construct
        self.tables = sqlutil.TableFinder(self.mapped_table)

        if not self.tables:
            raise exceptions.InvalidRequestError("Could not find any Table objects in mapped table '%s'" % str(self.mapped_table))

        # determine primary key columns
        self.pks_by_table = {}

        # go through all of our represented tables
        # and assemble primary key columns
        for t in self.tables + [self.mapped_table]:
            self._all_tables.add(t)
            try:
                l = self.pks_by_table[t]
            except KeyError:
                l = self.pks_by_table.setdefault(t, util.OrderedSet())
            for k in t.primary_key:
                l.add(k)
                
        if self.primary_key_argument is not None:
            for k in self.primary_key_argument:
                self.pks_by_table.setdefault(k.table, util.OrderedSet()).add(k)
                
        if len(self.pks_by_table[self.mapped_table]) == 0:
            raise exceptions.ArgumentError("Could not assemble any primary key columns for mapped table '%s'" % (self.mapped_table.name))

        if self.inherits is not None and not self.concrete and not self.primary_key_argument:
            self.primary_key = self.inherits.primary_key
            self._get_clause = self.inherits._get_clause
        else:
            # create the "primary_key" for this mapper.  this will flatten "equivalent" primary key columns
            # into one column, where "equivalent" means that one column references the other via foreign key, or
            # multiple columns that all reference a common parent column.  it will also resolve the column
            # against the "mapped_table" of this mapper.
            equivalent_columns = self._get_equivalent_columns()
        
            primary_key = expression.ColumnSet()

            for col in (self.primary_key_argument or self.pks_by_table[self.mapped_table]):
                c = self.mapped_table.corresponding_column(col, raiseerr=False)
                if c is None:
                    for cc in equivalent_columns[col]:
                        c = self.mapped_table.corresponding_column(cc, raiseerr=False)
                        if c is not None:
                            break
                    else:
                        raise exceptions.ArgumentError("Cant resolve column " + str(col))

                # this step attempts to resolve the column to an equivalent which is not
                # a foreign key elsewhere.  this helps with joined table inheritance
                # so that PKs are expressed in terms of the base table which is always
                # present in the initial select
                # TODO: this is a little hacky right now, the "tried" list is to prevent
                # endless loops between cyclical FKs, try to make this cleaner/work better/etc.,
                # perhaps via topological sort (pick the leftmost item)
                tried = util.Set()
                while True:
                    if not len(c.foreign_keys) or c in tried:
                        break
                    for cc in c.foreign_keys:
                        cc = cc.column
                        c2 = self.mapped_table.corresponding_column(cc, raiseerr=False)
                        if c2 is not None:
                            c = c2
                            tried.add(c)
                            break
                    else:
                        break
                primary_key.add(c)
                
            if len(primary_key) == 0:
                raise exceptions.ArgumentError("Could not assemble any primary key columns for mapped table '%s'" % (self.mapped_table.name))

            self.primary_key = primary_key
            self.__log("Identified primary key columns: " + str(primary_key))
        
            _get_clause = sql.and_()
            for primary_key in self.primary_key:
                _get_clause.clauses.append(primary_key == sql.bindparam(primary_key._label, type_=primary_key.type, unique=True))
            self._get_clause = _get_clause

    def _get_equivalent_columns(self):
        """Create a map of all *equivalent* columns, based on
        the determination of column pairs that are equated to
        one another either by an established foreign key relationship
        or by a joined-table inheritance join.

        This is used to determine the minimal set of primary key 
        columns for the mapper, as well as when relating 
        columns to those of a polymorphic selectable (i.e. a UNION of
        several mapped tables), as that selectable usually only contains
        one column in its columns clause out of a group of several which
        are equated to each other.

        The resulting structure is a dictionary of columns mapped
        to lists of equivalent columns, i.e.

        {
            tablea.col1: 
                set([tableb.col1, tablec.col1]),
            tablea.col2:
                set([tabled.col2])
        }
        
        this method is called repeatedly during the compilation process as 
        the resulting dictionary contains more equivalents as more inheriting 
        mappers are compiled.  the repetition process may be open to some optimization.
        """

        result = {}
        def visit_binary(binary):
            if binary.operator == operator.eq:
                if binary.left in result:
                    result[binary.left].add(binary.right)
                else:
                    result[binary.left] = util.Set([binary.right])
                if binary.right in result:
                    result[binary.right].add(binary.left)
                else:
                    result[binary.right] = util.Set([binary.left])
        vis = mapperutil.BinaryVisitor(visit_binary)

        for mapper in self.base_mapper.polymorphic_iterator():
            if mapper.inherit_condition is not None:
                vis.traverse(mapper.inherit_condition)

        for col in (self.primary_key_argument or self.pks_by_table[self.mapped_table]):
            if not col.foreign_keys:
                result.setdefault(col, util.Set()).add(col)
            else:
                for fk in col.foreign_keys:
                    result.setdefault(fk.column, util.Set()).add(col)

        return result
    
    class _CompileOnAttr(PropComparator):
        """placeholder class attribute which fires mapper compilation on access"""
        def __init__(self, class_, key):
            self.class_ = class_
            self.key = key
        def __getattribute__(self, key):
            cls = object.__getattribute__(self, 'class_')
            clskey = object.__getattribute__(self, 'key')

            # get the class' mapper; will compile all mappers
            class_mapper(cls)
            
            if cls.__dict__.get(clskey) is self:
                # FIXME: there should not be any scenarios where 
                # a mapper compile leaves this CompileOnAttr in 
                # place.  
                warnings.warn(RuntimeWarning("Attribute '%s' on class '%s' was not replaced during mapper compilation operation" % (clskey, cls.__name__)))
                # clean us up explicitly
                delattr(cls, clskey)
                
            return getattr(getattr(cls, clskey), key)
            
    def _compile_properties(self):
        """Inspect the properties dictionary sent to the Mapper's
        constructor as well as the mapped_table, and create
        ``MapperProperty`` objects corresponding to each mapped column
        and relation.
        """

        # object attribute names mapped to MapperProperty objects
        self.__props = util.OrderedDict()

        # table columns mapped to lists of MapperProperty objects
        # using a list allows a single column to be defined as
        # populating multiple object attributes
        self.columntoproperty = mapperutil.TranslatingDict(self.mapped_table)

        # load custom properties
        if self.properties is not None:
            for key, prop in self.properties.iteritems():
                self._compile_property(key, prop, False)

        if self.inherits is not None:
            for key, prop in self.inherits.__props.iteritems():
                if key not in self.__props:
                    self._adapt_inherited_property(key, prop)

        # load properties from the main table object,
        # not overriding those set up in the 'properties' argument
        for column in self.mapped_table.columns:
            if column in self.columntoproperty:
                continue
            if column.key not in self.columns:
                self.columns[column.key] = self.select_table.corresponding_column(column, keys_ok=True, raiseerr=True)

            column_key = (self.column_prefix or '') + column.key
            prop = self.__props.get(column_key, None)

            if prop is None:
                if (self.include_properties is not None and
                    column.key not in self.include_properties):
                    self.__log("not including property %s" % (column.key))
                    continue
                
                if (self.exclude_properties is not None and
                    column.key in self.exclude_properties):
                    self.__log("excluding property %s" % (column.key))
                    continue
                
                prop = ColumnProperty(column)
                self.__props[column_key] = prop

                # TODO: centralize _CompileOnAttr logic, move into MapperProperty classes
                if not hasattr(self.class_, column_key) and not self.non_primary:
                    setattr(self.class_, column_key, Mapper._CompileOnAttr(self.class_, column_key))
                
                prop.set_parent(self)
                self.__log("adding ColumnProperty %s" % (column_key))
            elif isinstance(prop, ColumnProperty):
                if prop.parent is not self:
                    prop = prop.copy()
                    prop.set_parent(self)
                    self.__props[column_key] = prop
                if column in self.primary_key and prop.columns[-1] in self.primary_key:
                    warnings.warn(RuntimeWarning("On mapper %s, primary key column '%s' is being combined with distinct primary key column '%s' in attribute '%s'.  Use explicit properties to give each column its own mapped attribute name." % (str(self), str(column), str(prop.columns[-1]), column_key)))
                prop.columns.append(column)
                self.__log("appending to existing ColumnProperty %s" % (column_key))
            else:
                if not self.allow_column_override:
                    raise exceptions.ArgumentError("WARNING: column '%s' not being added due to property '%s'.  Specify 'allow_column_override=True' to mapper() to ignore this condition." % (column.key, repr(prop)))
                else:
                    continue

            # its a ColumnProperty - match the ultimate table columns
            # back to the property
            self.columntoproperty.setdefault(column, []).append(prop)


    def _compile_selectable(self):
        """If the 'select_table' keyword argument was specified, set
        up a second *surrogate mapper* that will be used for select
        operations.

        The columns of `select_table` should encompass all the columns
        of the `mapped_table` either directly or through proxying
        relationships. Currently, non-column properties are **not**
        copied.  This implies that a polymorphic mapper can't do any
        eager loading right now.
        """

        if self.select_table is not self.mapped_table:
            props = {}
            if self.properties is not None:
                for key, prop in self.properties.iteritems():
                    if expression.is_column(prop):
                        props[key] = self.select_table.corresponding_column(prop)
                    elif (isinstance(prop, list) and expression.is_column(prop[0])):
                        props[key] = [self.select_table.corresponding_column(c) for c in prop]
            self.__surrogate_mapper = Mapper(self.class_, self.select_table, non_primary=True, properties=props, _polymorphic_map=self.polymorphic_map, polymorphic_on=self.select_table.corresponding_column(self.polymorphic_on), primary_key=self.primary_key_argument)

    def _compile_class(self):
        """If this mapper is to be a primary mapper (i.e. the
        non_primary flag is not set), associate this Mapper with the
        given class_ and entity name.

        Subsequent calls to ``class_mapper()`` for the class_/entity
        name combination will return this mapper.  Also decorate the
        `__init__` method on the mapped class to include optional
        auto-session attachment logic.
        """

        if self.non_primary:
            return

        if not self.non_primary and (self.class_key in mapper_registry):
             raise exceptions.ArgumentError("Class '%s' already has a primary mapper defined with entity name '%s'.  Use non_primary=True to create a non primary Mapper, or to create a new primary mapper, remove this mapper first via sqlalchemy.orm.clear_mapper(mapper), or preferably sqlalchemy.orm.clear_mappers() to clear all mappers." % (self.class_, self.entity_name))

        attribute_manager.reset_class_managed(self.class_)

        oldinit = self.class_.__init__
        doinit = oldinit is not None and oldinit is not object.__init__
            
        def init(instance, *args, **kwargs):
            self.compile()
            self.extension.init_instance(self, self.class_, oldinit, instance, args, kwargs)

            if doinit:
                try:
                    oldinit(instance, *args, **kwargs)
                except:
                    # call init_failed but suppress exceptions into warnings so that original __init__ 
                    # exception is raised
                    util.warn_exception(self.extension.init_failed, self, self.class_, oldinit, instance, args, kwargs)
                    raise

        # override oldinit, ensuring that its not already a Mapper-decorated init method
        if oldinit is None or not hasattr(oldinit, '_oldinit'):
            try:
                init.__name__ = oldinit.__name__
                init.__doc__ = oldinit.__doc__
            except:
                # cant set __name__ in py 2.3 !
                pass
            init._oldinit = oldinit
            self.class_.__init__ = init

        _COMPILE_MUTEX.acquire()
        try:
            mapper_registry[self.class_key] = self
        finally:
            _COMPILE_MUTEX.release()

        for ext in util.to_list(self.extension, []):
            ext.instrument_class(self, self.class_)

        if self.entity_name is None:
            self.class_.c = self.c

    def common_parent(self, other):
        """Return true if the given mapper shares a common inherited parent as this mapper."""

        return self.base_mapper is other.base_mapper

    def isa(self, other):
        """Return True if the given mapper inherits from this mapper."""

        m = other
        while m is not self and m.inherits is not None:
            m = m.inherits
        return m is self

    def iterate_to_root(self):
        m = self
        while m is not None:
            yield m
            m = m.inherits

    def polymorphic_iterator(self):
        """Iterate through the collection including this mapper and
        all descendant mappers.

        This includes not just the immediately inheriting mappers but
        all their inheriting mappers as well.

        To iterate through an entire hierarchy, use
        ``mapper.base_mapper.polymorphic_iterator()``."""

        yield self
        for mapper in self._inheriting_mappers:
            for m in mapper.polymorphic_iterator():
                yield m


    def add_properties(self, dict_of_properties):
        """Add the given dictionary of properties to this mapper,
        using `add_property`.
        """

        for key, value in dict_of_properties.iteritems():
            self.add_property(key, value)

    def add_property(self, key, prop):
        """Add an indiviual MapperProperty to this mapper.

        If the mapper has not been compiled yet, just adds the
        property to the initial properties dictionary sent to the
        constructor.  If this Mapper has already been compiled, then
        the given MapperProperty is compiled immediately.
        """

        self.properties[key] = prop
        self._compile_property(key, prop, init=self.__props_init)

    def _create_prop_from_column(self, column):
        column = util.to_list(column)
        if not expression.is_column(column[0]):
            return None
        mapped_column = []
        for c in column:
            mc = self.mapped_table.corresponding_column(c, raiseerr=False)
            if not mc:
                raise exceptions.ArgumentError("Column '%s' is not represented in mapper's table.  Use the `column_property()` function to force this column to be mapped as a read-only attribute." % str(c))
            mapped_column.append(mc)
        return ColumnProperty(*mapped_column)

    def _adapt_inherited_property(self, key, prop):
        if not self.concrete:
            self._compile_property(key, prop, init=False, setparent=False)
        # TODO: concrete properties dont adapt at all right now....will require copies of relations() etc.

    def _compile_property(self, key, prop, init=True, setparent=True):
        """Add a ``MapperProperty`` to this or another ``Mapper``,
        including configuration of the property.

        The properties' parent attribute will be set, and the property
        will also be copied amongst the mappers which inherit from
        this one.

        If the given `prop` is a ``Column`` or list of Columns, a
        ``ColumnProperty`` will be created.
        """

        self.__log("_compile_property(%s, %s)" % (key, prop.__class__.__name__))

        if not isinstance(prop, MapperProperty):
            col = self._create_prop_from_column(prop)
            if col is None:
                raise exceptions.ArgumentError("%s=%r is not an instance of MapperProperty or Column" % (key, prop))
            prop = col

        self.__props[key] = prop
        

        if setparent:
            prop.set_parent(self)

            # TODO: centralize _CompileOnAttr logic, move into MapperProperty classes
            if (not isinstance(prop, SynonymProperty) or prop.proxy) and not self.non_primary and not hasattr(self.class_, key):
                setattr(self.class_, key, Mapper._CompileOnAttr(self.class_, key))

        if isinstance(prop, ColumnProperty):
            # relate the mapper's "select table" to the given ColumnProperty
            col = self.select_table.corresponding_column(prop.columns[0], keys_ok=True, raiseerr=False)
            # col might not be present! the selectable given to the mapper need not include "deferred"
            # columns (included in zblog tests)
            if col is None:
                col = prop.columns[0]
            self.columns[key] = col
            for col in prop.columns:
                proplist = self.columntoproperty.setdefault(col, [])
                proplist.append(prop)

        if init:
            prop.init(key, self)

        for mapper in self._inheriting_mappers:
            mapper._adapt_inherited_property(key, prop)

    def __str__(self):
        return "Mapper|" + self.class_.__name__ + "|" + (self.entity_name is not None and "/%s" % self.entity_name or "") + (self.local_table and self.local_table.encodedname or str(self.local_table)) + (not self._is_primary_mapper() and "|non-primary" or "")

    def _is_primary_mapper(self):
        """Return True if this mapper is the primary mapper for its class key (class + entity_name)."""
        # FIXME: cant we just look at "non_primary" flag ?
        return mapper_registry.get(self.class_key, None) is self

    def primary_mapper(self):
        """Return the primary mapper corresponding to this mapper's class key (class + entity_name)."""
        return mapper_registry[self.class_key]

    def is_assigned(self, instance):
        """Return True if this mapper handles the given instance.

        This is dependent not only on class assignment but the
        optional `entity_name` parameter as well.
        """

        return instance.__class__ is self.class_ and getattr(instance, '_entity_name', None) == self.entity_name

    def _assign_entity_name(self, instance):
        """Assign this Mapper's entity name to the given instance.

        Subsequent Mapper lookups for this instance will return the
        primary mapper corresponding to this Mapper's class and entity
        name.
        """

        instance._entity_name = self.entity_name

    def get_session(self):
        """Return the contextual session provided by the mapper
        extension chain, if any.

        Raise ``InvalidRequestError`` if a session cannot be retrieved
        from the extension chain.
        """

        s = self.extension.get_session()
        if s is EXT_CONTINUE:
            raise exceptions.InvalidRequestError("No contextual Session is established.  Use a MapperExtension that implements get_session or use 'import sqlalchemy.mods.threadlocal' to establish a default thread-local contextual session.")
        return s

    def has_eager(self):
        """Return True if one of the properties attached to this
        Mapper is eager loading.
        """

        return len(self._eager_loaders) > 0

    def instances(self, cursor, session, *mappers, **kwargs):
        """Return a list of mapped instances corresponding to the rows
        in a given ResultProxy.
        """

        import sqlalchemy.orm.query
        return sqlalchemy.orm.Query(self, session).instances(cursor, *mappers, **kwargs)

    def identity_key_from_row(self, row):
        """Return an identity-map key for use in storing/retrieving an
        item from the identity map.

        row
          A ``sqlalchemy.engine.base.RowProxy`` instance or a
          dictionary corresponding result-set ``ColumnElement``
          instances to their values within a row.
        """
        return (self._identity_class, tuple([row[column] for column in self.primary_key]), self.entity_name)

    def identity_key_from_primary_key(self, primary_key):
        """Return an identity-map key for use in storing/retrieving an
        item from an identity map.

        primary_key
          A list of values indicating the identifier.
        """
        return (self._identity_class, tuple(util.to_list(primary_key)), self.entity_name)

    def identity_key_from_instance(self, instance):
        """Return the identity key for the given instance, based on
        its primary key attributes.

        This value is typically also found on the instance itself
        under the attribute name `_instance_key`.
        """
        return self.identity_key_from_primary_key(self.primary_key_from_instance(instance))

    def primary_key_from_instance(self, instance):
        """Return the list of primary key values for the given
        instance.
        """

        return [self.get_attr_by_column(instance, column) for column in self.primary_key]

    def canload(self, instance):
        """return true if this mapper is capable of loading the given instance"""
        if self.polymorphic_on is not None:
            return isinstance(instance, self.class_)
        else:
            return instance.__class__ is self.class_
        
    def _getpropbycolumn(self, column, raiseerror=True):
        try:
            prop = self.columntoproperty[column]
        except KeyError:
            try:
                prop = self.__props[column.key]
                if not raiseerror:
                    return None
                raise exceptions.InvalidRequestError("Column '%s.%s' is not available, due to conflicting property '%s':%s" % (column.table.name, column.name, column.key, repr(prop)))
            except KeyError:
                if not raiseerror:
                    return None
                raise exceptions.InvalidRequestError("No column %s.%s is configured on mapper %s..." % (column.table.name, column.name, str(self)))
        return prop[0]

    def get_attr_by_column(self, obj, column, raiseerror=True):
        """Return an instance attribute using a Column as the key."""

        prop = self._getpropbycolumn(column, raiseerror)
        if prop is None:
            return NO_ATTRIBUTE
        return prop.getattr(obj, column)

    def set_attr_by_column(self, obj, column, value):
        """Set the value of an instance attribute using a Column as the key."""

        self.columntoproperty[column][0].setattr(obj, value, column)

    def save_obj(self, objects, uowtransaction, postupdate=False, post_update_cols=None, single=False):
        """Issue ``INSERT`` and/or ``UPDATE`` statements for a list of objects.

        This is called within the context of a UOWTransaction during a
        flush operation.

        `save_obj` issues SQL statements not just for instances mapped
        directly by this mapper, but for instances mapped by all
        inheriting mappers as well.  This is to maintain proper insert
        ordering among a polymorphic chain of instances. Therefore
        save_obj is typically called only on a *base mapper*, or a
        mapper which does not inherit from any other mapper.
        """

        if self.__should_log_debug:
            self.__log_debug("save_obj() start, " + (single and "non-batched" or "batched"))

        # if batch=false, call save_obj separately for each object
        if not single and not self.batch:
            for obj in objects:
                self.save_obj([obj], uowtransaction, postupdate=postupdate, post_update_cols=post_update_cols, single=True)
            return

        if 'connection_callable' in uowtransaction.mapper_flush_opts:
            connection_callable = uowtransaction.mapper_flush_opts['connection_callable']
            tups = [(obj, connection_callable(self, obj)) for obj in objects]
        else:
            connection = uowtransaction.transaction.connection(self)
            tups = [(obj, connection) for obj in objects]
            
        if not postupdate:
            for obj, connection in tups:
                if not has_identity(obj):
                    for mapper in object_mapper(obj).iterate_to_root():
                        mapper.extension.before_insert(mapper, connection, obj)
                else:
                    for mapper in object_mapper(obj).iterate_to_root():
                        mapper.extension.before_update(mapper, connection, obj)

        for obj, connection in tups:
            # detect if we have a "pending" instance (i.e. has no instance_key attached to it),
            # and another instance with the same identity key already exists as persistent.  convert to an
            # UPDATE if so.
            mapper = object_mapper(obj)
            instance_key = mapper.identity_key_from_instance(obj)
            is_row_switch = not postupdate and not has_identity(obj) and instance_key in uowtransaction.uow.identity_map
            if is_row_switch:
                existing = uowtransaction.uow.identity_map[instance_key]
                if not uowtransaction.is_deleted(existing):
                    raise exceptions.FlushError("New instance %s with identity key %s conflicts with persistent instance %s" % (mapperutil.instance_str(obj), str(instance_key), mapperutil.instance_str(existing)))
                if self.__should_log_debug:
                    self.__log_debug("detected row switch for identity %s.  will update %s, remove %s from transaction" % (instance_key, mapperutil.instance_str(obj), mapperutil.instance_str(existing)))
                uowtransaction.unregister_object(existing)
            if has_identity(obj):
                if obj._instance_key != instance_key:
                    raise exceptions.FlushError("Can't change the identity of instance %s in session (existing identity: %s; new identity: %s)" % (mapperutil.instance_str(obj), obj._instance_key, instance_key))

        inserted_objects = util.Set()
        updated_objects = util.Set()

        table_to_mapper = {}
        for mapper in self.base_mapper.polymorphic_iterator():
            for t in mapper.tables:
                table_to_mapper.setdefault(t, mapper)

        for table in sqlutil.TableCollection(list(table_to_mapper.keys())).sort(reverse=False):
            # two lists to store parameters for each table/object pair located
            insert = []
            update = []

            for obj, connection in tups:
                mapper = object_mapper(obj)
                if table not in mapper.tables or not mapper._has_pks(table):
                    continue
                instance_key = mapper.identity_key_from_instance(obj)
                if self.__should_log_debug:
                    self.__log_debug("save_obj() table '%s' instance %s identity %s" % (table.name, mapperutil.instance_str(obj), str(instance_key)))

                isinsert = not instance_key in uowtransaction.uow.identity_map and not postupdate and not has_identity(obj)
                params = {}
                value_params = {}
                hasdata = False
                for col in table.columns:
                    if col is mapper.version_id_col:
                        if not isinsert:
                            params[col._label] = mapper.get_attr_by_column(obj, col)
                            params[col.key] = params[col._label] + 1
                        else:
                            params[col.key] = 1
                    elif col in mapper.pks_by_table[table]:
                        # column is a primary key ?
                        if not isinsert:
                            # doing an UPDATE?  put primary key values as "WHERE" parameters
                            # matching the bindparam we are creating below, i.e. "<tablename>_<colname>"
                            params[col._label] = mapper.get_attr_by_column(obj, col)
                        else:
                            # doing an INSERT, primary key col ?
                            # if the primary key values are not populated,
                            # leave them out of the INSERT altogether, since PostGres doesn't want
                            # them to be present for SERIAL to take effect.  A SQLEngine that uses
                            # explicit sequences will put them back in if they are needed
                            value = mapper.get_attr_by_column(obj, col)
                            if value is not None:
                                params[col.key] = value
                    elif mapper.polymorphic_on is not None and mapper.polymorphic_on.shares_lineage(col):
                        if isinsert:
                            if self.__should_log_debug:
                                self.__log_debug("Using polymorphic identity '%s' for insert column '%s'" % (mapper.polymorphic_identity, col.key))
                            value = mapper.polymorphic_identity
                            if col.default is None or value is not None:
                                params[col.key] = value
                    else:
                        # column is not a primary key ?
                        if not isinsert:
                            # doing an UPDATE ? get the history for the attribute, with "passive"
                            # so as not to trigger any deferred loads.  if there is a new
                            # value, add it to the bind parameters
                            if post_update_cols is not None and col not in post_update_cols:
                                continue
                            elif is_row_switch:
                                params[col.key] = self.get_attr_by_column(obj, col)
                                hasdata = True
                                continue
                            prop = mapper._getpropbycolumn(col, False)
                            if prop is None:
                                continue
                            history = prop.get_history(obj, passive=True)
                            if history:
                                a = history.added_items()
                                if a:
                                    if isinstance(a[0], sql.ClauseElement):
                                        value_params[col] = a[0]
                                    else:
                                        params[col.key] = prop.get_col_value(col, a[0])
                                    hasdata = True
                        else:
                            # doing an INSERT, non primary key col ?
                            # add the attribute's value to the
                            # bind parameters, unless its None and the column has a
                            # default.  if its None and theres no default, we still might
                            # not want to put it in the col list but SQLIte doesnt seem to like that
                            # if theres no columns at all
                            value = mapper.get_attr_by_column(obj, col, False)
                            if value is NO_ATTRIBUTE:
                                continue
                            if col.default is None or value is not None:
                                # TODO: clauseelments as bind params should 
                                # be handled by Insert/Update expression upon execute()
                                if isinstance(value, sql.ClauseElement):
                                    value_params[col] = value
                                else:
                                    params[col.key] = value

                if not isinsert:
                    if hasdata:
                        # if none of the attributes changed, dont even
                        # add the row to be updated.
                        update.append((obj, params, mapper, connection, value_params))
                else:
                    insert.append((obj, params, mapper, connection, value_params))

            if update:
                mapper = table_to_mapper[table]
                clause = sql.and_()
                for col in mapper.pks_by_table[table]:
                    clause.clauses.append(col == sql.bindparam(col._label, type_=col.type, unique=True))
                if mapper.version_id_col is not None:
                    clause.clauses.append(mapper.version_id_col == sql.bindparam(mapper.version_id_col._label, type_=col.type, unique=True))
                statement = table.update(clause)
                rows = 0
                supports_sane_rowcount = True
                def comparator(a, b):
                    for col in mapper.pks_by_table[table]:
                        x = cmp(a[1][col._label],b[1][col._label])
                        if x != 0:
                            return x
                    return 0
                update.sort(comparator)
                for rec in update:
                    (obj, params, mapper, connection, value_params) = rec
                    c = connection.execute(statement.values(value_params), params)
                    mapper._postfetch(connection, table, obj, c, c.last_updated_params(), value_params)

                    updated_objects.add((obj, connection))
                    rows += c.rowcount

                if c.supports_sane_rowcount() and rows != len(update):
                    raise exceptions.ConcurrentModificationError("Updated rowcount %d does not match number of objects updated %d" % (rows, len(update)))

            if insert:
                statement = table.insert()
                def comparator(a, b):
                    return cmp(a[0]._sa_insert_order, b[0]._sa_insert_order)
                insert.sort(comparator)
                for rec in insert:
                    (obj, params, mapper, connection, value_params) = rec
                    c = connection.execute(statement.values(value_params), params)
                    primary_key = c.last_inserted_ids()

                    if primary_key is not None:
                        i = 0
                        for col in mapper.pks_by_table[table]:
                            if mapper.get_attr_by_column(obj, col) is None and len(primary_key) > i:
                                mapper.set_attr_by_column(obj, col, primary_key[i])
                            i+=1
                    mapper._postfetch(connection, table, obj, c, c.last_inserted_params(), value_params)

                    # synchronize newly inserted ids from one table to the next
                    # TODO: this fires off more than needed, try to organize syncrules
                    # per table
                    def sync(mapper):
                        inherit = mapper.inherits
                        if inherit is not None:
                            sync(inherit)
                        if mapper._synchronizer is not None:
                            mapper._synchronizer.execute(obj, obj)
                    sync(mapper)

                    inserted_objects.add((obj, connection))
        if not postupdate:
            for obj, connection in inserted_objects:
                for mapper in object_mapper(obj).iterate_to_root():
                    mapper.extension.after_insert(mapper, connection, obj)
            for obj, connection in updated_objects:
                for mapper in object_mapper(obj).iterate_to_root():
                    mapper.extension.after_update(mapper, connection, obj)

    def _postfetch(self, connection, table, obj, resultproxy, params, value_params):
        """After an ``INSERT`` or ``UPDATE``, assemble newly generated
        values on an instance.  For columns which are marked as being generated
        on the database side, set up a group-based "deferred" loader 
        which will populate those attributes in one query when next accessed.
        """

        postfetch_cols = resultproxy.postfetch_cols().union(util.Set(value_params.keys())) 
        deferred_props = []

        for c in table.c:
            if c in postfetch_cols and (not c.key in params or c in value_params):
                prop = self._getpropbycolumn(c, raiseerror=False)
                if prop is None:
                    continue
                deferred_props.append(prop)
                continue
            if c.primary_key or not c.key in params:
                continue
            v = self.get_attr_by_column(obj, c, False)
            if v is NO_ATTRIBUTE:
                continue
            elif v != params.get_original(c.key):
                self.set_attr_by_column(obj, c, params.get_original(c.key))
        
        if deferred_props:
            deferred_load(obj, props=deferred_props)

    def delete_obj(self, objects, uowtransaction):
        """Issue ``DELETE`` statements for a list of objects.

        This is called within the context of a UOWTransaction during a
        flush operation.
        """

        if self.__should_log_debug:
            self.__log_debug("delete_obj() start")

        if 'connection_callable' in uowtransaction.mapper_flush_opts:
            connection_callable = uowtransaction.mapper_flush_opts['connection_callable']
            tups = [(obj, connection_callable(self, obj)) for obj in objects]
        else:
            connection = uowtransaction.transaction.connection(self)
            tups = [(obj, connection) for obj in objects]

        for (obj, connection) in tups:
            for mapper in object_mapper(obj).iterate_to_root():
                mapper.extension.before_delete(mapper, connection, obj)
        
        deleted_objects = util.Set()
        table_to_mapper = {}
        for mapper in self.base_mapper.polymorphic_iterator():
            for t in mapper.tables:
                table_to_mapper.setdefault(t, mapper)

        for table in sqlutil.TableCollection(list(table_to_mapper.keys())).sort(reverse=True):
            delete = {}
            for (obj, connection) in tups:
                mapper = object_mapper(obj)
                if table not in mapper.tables or not mapper._has_pks(table):
                    continue

                params = {}
                if not hasattr(obj, '_instance_key'):
                    continue
                else:
                    delete.setdefault(connection, []).append(params)
                for col in mapper.pks_by_table[table]:
                    params[col.key] = mapper.get_attr_by_column(obj, col)
                if mapper.version_id_col is not None:
                    params[mapper.version_id_col.key] = mapper.get_attr_by_column(obj, mapper.version_id_col)
                deleted_objects.add((obj, connection))
            for connection, del_objects in delete.iteritems():
                mapper = table_to_mapper[table]
                def comparator(a, b):
                    for col in mapper.pks_by_table[table]:
                        x = cmp(a[col.key],b[col.key])
                        if x != 0:
                            return x
                    return 0
                del_objects.sort(comparator)
                clause = sql.and_()
                for col in mapper.pks_by_table[table]:
                    clause.clauses.append(col == sql.bindparam(col.key, type_=col.type, unique=True))
                if mapper.version_id_col is not None:
                    clause.clauses.append(mapper.version_id_col == sql.bindparam(mapper.version_id_col.key, type_=mapper.version_id_col.type, unique=True))
                statement = table.delete(clause)
                c = connection.execute(statement, del_objects)
                if c.supports_sane_multi_rowcount() and c.rowcount != len(del_objects):
                    raise exceptions.ConcurrentModificationError("Deleted rowcount %d does not match number of objects deleted %d" % (c.rowcount, len(del_objects)))

        for obj, connection in deleted_objects:
            for mapper in object_mapper(obj).iterate_to_root():
                mapper.extension.after_delete(mapper, connection, obj)

    def _has_pks(self, table):
        try:
            pk = self.pks_by_table[table]
            if not pk:
                return False
            for k in pk:
                if k not in self.columntoproperty:
                    return False
            else:
                return True
        except KeyError:
            return False

    def register_dependencies(self, uowcommit, *args, **kwargs):
        """Register ``DependencyProcessor`` instances with a
        ``unitofwork.UOWTransaction``.

        This call `register_dependencies` on all attached
        ``MapperProperty`` instances.
        """

        for prop in self.__props.values():
            prop.register_dependencies(uowcommit, *args, **kwargs)

    def cascade_iterator(self, type, object, recursive=None, halt_on=None):
        """Iterate each element in an object graph, for all relations
        taht meet the given cascade rule.

        type
          The name of the cascade rule (i.e. save-update, delete,
          etc.)

        object
          The lead object instance.  child items will be processed per
          the relations defined for this object's mapper.

        recursive
          Used by the function for internal context during recursive
          calls, leave as None.
        """

        if recursive is None:
            recursive=util.Set()
        for prop in self.__props.values():
            for c in prop.cascade_iterator(type, object, recursive, halt_on=halt_on):
                yield c

    def cascade_callable(self, type, object, callable_, recursive=None, halt_on=None):
        """Execute a callable for each element in an object graph, for
        all relations that meet the given cascade rule.

        type
          The name of the cascade rule (i.e. save-update, delete, etc.)

        object
          The lead object instance.  child items will be processed per
          the relations defined for this object's mapper.

        callable\_
          The callable function.

        recursive
          Used by the function for internal context during recursive
          calls, leave as None.
          
        """

        if recursive is None:
            recursive=util.Set()
        for prop in self.__props.values():
            prop.cascade_callable(type, object, callable_, recursive, halt_on=halt_on)

    def get_select_mapper(self):
        """Return the mapper used for issuing selects.

        This mapper is the same mapper as `self` unless the
        select_table argument was specified for this mapper.
        """

        return self.__surrogate_mapper or self

    def _instance(self, context, row, result = None, skip_polymorphic=False):
        """Pull an object instance from the given row and append it to
        the given result list.

        If the instance already exists in the given identity map, its
        not added.  In either case, execute all the property loaders
        on the instance to also process extra information in the row.
        """

        # apply ExtensionOptions applied to the Query to this mapper,
        # but only if our mapper matches.
        # TODO: what if our mapper inherits from the mapper (i.e. as in a polymorphic load?)
        if context.mapper is self:
            extension = context.extension
        else:
            extension = self.extension

        ret = extension.translate_row(self, context, row)
        if ret is not EXT_CONTINUE:
            row = ret

        if not skip_polymorphic and self.polymorphic_on is not None:
            discriminator = row[self.polymorphic_on]
            if discriminator is not None:
                mapper = self.polymorphic_map[discriminator]
                if mapper is not self:
                    if ('polymorphic_fetch', mapper) not in context.attributes:
                        context.attributes[('polymorphic_fetch', mapper)] = (self, [t for t in mapper.tables if t not in self.tables])
                    row = self.translate_row(mapper, row)
                    return mapper._instance(context, row, result=result, skip_polymorphic=True)
        
        # look in main identity map.  if its there, we dont do anything to it,
        # including modifying any of its related items lists, as its already
        # been exposed to being modified by the application.

        identitykey = self.identity_key_from_row(row)
        populate_existing = context.populate_existing or self.always_refresh
        if identitykey in context.session.identity_map:
            instance = context.session._get(identitykey)
            if self.__should_log_debug:
                self.__log_debug("_instance(): using existing instance %s identity %s" % (mapperutil.instance_str(instance), str(identitykey)))
            isnew = False
            if context.version_check and self.version_id_col is not None and self.get_attr_by_column(instance, self.version_id_col) != row[self.version_id_col]:
                raise exceptions.ConcurrentModificationError("Instance '%s' version of %s does not match %s" % (instance, self.get_attr_by_column(instance, self.version_id_col), row[self.version_id_col]))

            if populate_existing or context.session.is_expired(instance, unexpire=True):
                if identitykey not in context.identity_map:
                    context.identity_map[identitykey] = instance
                    isnew = True
                if extension.populate_instance(self, context, row, instance, instancekey=identitykey, isnew=isnew) is EXT_CONTINUE:
                    self.populate_instance(context, instance, row, instancekey=identitykey, isnew=isnew)
            if extension.append_result(self, context, row, instance, result, instancekey=identitykey, isnew=isnew) is EXT_CONTINUE:
                if result is not None:
                    result.append(instance)
            return instance
        else:
            if self.__should_log_debug:
                self.__log_debug("_instance(): identity key %s not in session" % str(identitykey))
        # look in result-local identitymap for it.
        exists = identitykey in context.identity_map
        if not exists:
            if self.allow_null_pks:
                # check if *all* primary key cols in the result are None - this indicates
                # an instance of the object is not present in the row.
                for x in identitykey[1]:
                    if x is not None:
                        break
                else:
                    return None
            else:
                # otherwise, check if *any* primary key cols in the result are None - this indicates
                # an instance of the object is not present in the row.
                if None in identitykey[1]:
                    return None

            # plugin point
            instance = extension.create_instance(self, context, row, self.class_)
            if instance is EXT_CONTINUE:
                instance = self._create_instance(context.session)
            instance._entity_name = self.entity_name
            if self.__should_log_debug:
                self.__log_debug("_instance(): created new instance %s identity %s" % (mapperutil.instance_str(instance), str(identitykey)))
            context.identity_map[identitykey] = instance
            isnew = True
        else:
            instance = context.identity_map[identitykey]
            isnew = False

        # call further mapper properties on the row, to pull further
        # instances from the row and possibly populate this item.
        flags = {'instancekey':identitykey, 'isnew':isnew}
        if extension.populate_instance(self, context, row, instance, **flags) is EXT_CONTINUE:
            self.populate_instance(context, instance, row, **flags)
        if extension.append_result(self, context, row, instance, result, **flags) is EXT_CONTINUE:
            if result is not None:
                result.append(instance)
        return instance

    def _create_instance(self, session):
        obj = self.class_.__new__(self.class_)
        obj._entity_name = self.entity_name

        return obj

    def _deferred_inheritance_condition(self, needs_tables):
        cond = self.inherit_condition

        param_names = []
        def visit_binary(binary):
            leftcol = binary.left
            rightcol = binary.right
            if leftcol is None or rightcol is None:
                return
            if leftcol.table not in needs_tables:
                binary.left = sql.bindparam(leftcol.name, None, type_=binary.right.type, unique=True)
                param_names.append(leftcol)
            elif rightcol not in needs_tables:
                binary.right = sql.bindparam(rightcol.name, None, type_=binary.right.type, unique=True)
                param_names.append(rightcol)
        cond = mapperutil.BinaryVisitor(visit_binary).traverse(cond, clone=True)
        return cond, param_names

    def translate_row(self, tomapper, row):
        """Translate the column keys of a row into a new or proxied
        row that can be understood by another mapper.

        This can be used in conjunction with populate_instance to
        populate an instance using an alternate mapper.
        """

        newrow = util.DictDecorator(row)
        for c in tomapper.mapped_table.c:
            c2 = self.mapped_table.corresponding_column(c, keys_ok=True, raiseerr=False)
            if c2 and c2 in row:
                newrow[c] = row[c2]
        return newrow

    def populate_instance(self, selectcontext, instance, row, ispostselect=None, **flags):
        """populate an instance from a result row."""

        selectcontext.stack.push_mapper(self)
        # retrieve a set of "row population" functions derived from the MapperProperties attached
        # to this Mapper.  These are keyed in the select context based primarily off the 
        # "snapshot" of the stack, which represents a path from the lead mapper in the query to this one,
        # including relation() names.  the key also includes "self", and allows us to distinguish between
        # other mappers within our inheritance hierarchy
        populators = selectcontext.attributes.get(('instance_populators', self, selectcontext.stack.snapshot(), ispostselect), None)
        if populators is None:
            # no populators; therefore this is the first time we are receiving a row for
            # this result set.  issue create_row_processor() on all MapperProperty objects
            # and cache in the select context.
            populators = []
            post_processors = []
            for prop in self.__props.values():
                (pop, post_proc) = prop.create_row_processor(selectcontext, self, row)
                if pop is not None:
                    populators.append(pop)
                if post_proc is not None:
                    post_processors.append(post_proc)
                    
            poly_select_loader = self._get_poly_select_loader(selectcontext, row)
            if poly_select_loader is not None:
                post_processors.append(poly_select_loader)
                
            selectcontext.attributes[('instance_populators', self, selectcontext.stack.snapshot(), ispostselect)] = populators
            selectcontext.attributes[('post_processors', self, ispostselect)] = post_processors

        for p in populators:
            p(instance, row, ispostselect=ispostselect, **flags)
        
        selectcontext.stack.pop()
            
        if self.non_primary:
            selectcontext.attributes[('populating_mapper', instance)] = self
        
    def _post_instance(self, selectcontext, instance):
        post_processors = selectcontext.attributes[('post_processors', self, None)]
        for p in post_processors:
            p(instance)

    def _get_poly_select_loader(self, selectcontext, row):
        # 'select' or 'union'+col not present
        (hosted_mapper, needs_tables) = selectcontext.attributes.get(('polymorphic_fetch', self), (None, None))
        if hosted_mapper is None or len(needs_tables)==0 or hosted_mapper.polymorphic_fetch == 'deferred':
            return
        
        cond, param_names = self._deferred_inheritance_condition(needs_tables)
        statement = sql.select(needs_tables, cond, use_labels=True)
        def post_execute(instance, **flags):
            self.__log_debug("Post query loading instance " + mapperutil.instance_str(instance))

            identitykey = self.identity_key_from_instance(instance)

            params = {}
            for c in param_names:
                params[c.name] = self.get_attr_by_column(instance, c)
            row = selectcontext.session.connection(self).execute(statement, params).fetchone()
            self.populate_instance(selectcontext, instance, row, isnew=False, instancekey=identitykey, ispostselect=True)

        return post_execute
            
Mapper.logger = logging.class_logger(Mapper)


class ClassKey(object):
    """Key a class and an entity name to a mapper, via the mapper_registry."""

    __metaclass__ = util.ArgSingleton

    def __init__(self, class_, entity_name):
        self.class_ = class_
        self.entity_name = entity_name

    def __hash__(self):
        return hash((self.class_, self.entity_name))

    def __eq__(self, other):
        return self is other

    def __repr__(self):
        return "ClassKey(%s, %s)" % (repr(self.class_), repr(self.entity_name))

    
def has_identity(object):
    return hasattr(object, '_instance_key')

def has_mapper(object):
    """Return True if the given object has had a mapper association
    set up, either through loading, or via insertion in a session.
    """

    return hasattr(object, '_entity_name')

def object_mapper(object, entity_name=None, raiseerror=True):
    """Given an object, return the primary Mapper associated with the object instance.
    
        object
            The object instance.
            
        entity_name
            Entity name of the mapper to retrieve, if the given instance is 
            transient.  Otherwise uses the entity name already associated 
            with the instance.
            
        raiseerror
            Defaults to True: raise an ``InvalidRequestError`` if no mapper can
            be located.  If False, return None.
            
    """

    try:
        mapper = mapper_registry[ClassKey(object.__class__, getattr(object, '_entity_name', entity_name))]
    except (KeyError, AttributeError):
        if raiseerror:
            raise exceptions.InvalidRequestError("Class '%s' entity name '%s' has no mapper associated with it" % (object.__class__.__name__, getattr(object, '_entity_name', entity_name)))
        else:
            return None
    return mapper.compile()

def class_mapper(class_, entity_name=None, compile=True):
    """Given a class and optional entity_name, return the primary Mapper associated with the key.
    
    If no mapper can be located, raises ``InvalidRequestError``.
    """

    try:
        mapper = mapper_registry[ClassKey(class_, entity_name)]
    except (KeyError, AttributeError):
        raise exceptions.InvalidRequestError("Class '%s' entity name '%s' has no mapper associated with it" % (class_.__name__, entity_name))
    if compile:
        return mapper.compile()
    else:
        return mapper