python-traits 4.4.0-1 / usr / share / pyshared / traits / has_traits.py

#------------------------------------------------------------------------------
#
#  Copyright (c) 2005, Enthought, Inc.
#  All rights reserved.
#
#  This software is provided without warranty under the terms of the BSD
#  license included in enthought/LICENSE.txt and may be redistributed only
#  under the conditions described in the aforementioned license.  The license
#  is also available online at http://www.enthought.com/licenses/BSD.txt
#
#  Thanks for using Enthought open source!
#
#  Author:        David C. Morrill
#  Original Date: 06/21/2002
#
#  Rewritten as a C-based type extension: 06/21/2004
#
#------------------------------------------------------------------------------

""" Defines the HasTraits class, along with several useful subclasses and
    associated metaclasses.
"""

#-------------------------------------------------------------------------------
#  Imports:
#-------------------------------------------------------------------------------

from __future__ import absolute_import

import sys
import copy as copy_module
import weakref
import re

from types import FunctionType, MethodType

from . import __version__ as TraitsVersion

from .adaptation.adaptation_error import AdaptationError

from .ctraits import CHasTraits, CTraitMethod, _HasTraits_monitors

from .traits import (CTrait, ForwardProperty, Property, SpecialNames, Trait,
    TraitFactory, __newobj__, generic_trait, trait_factory)

from .trait_types import Any, Bool, Disallow, Enum, Event, Python, This

from .trait_notifiers import (ExtendedTraitChangeNotifyWrapper,
    FastUITraitChangeNotifyWrapper, NewTraitChangeNotifyWrapper,
    StaticAnyTraitChangeNotifyWrapper, StaticTraitChangeNotifyWrapper,
    TraitChangeNotifyWrapper)

from .trait_handlers import TraitType

from .trait_base import (Missing, SequenceTypes, TraitsCache, Undefined,
    add_article, enumerate, is_none, not_event, not_false)

from .trait_errors import TraitError

from .protocols.advice import addClassAdvisor

from .util.deprecated import deprecated

#-------------------------------------------------------------------------------
#  Set CHECK_INTERFACES to one of the following values:
#
#  - 0: Does not check to see if classes implement their declared interfaces.
#  - 1: Ensures that classes implement the interfaces they say they do, and
#       logs a warning if they don't.
#  - 2: Ensures that classes implement the interfaces they say they do, and
#       raises an InterfaceError if they don't.
#-------------------------------------------------------------------------------

CHECK_INTERFACES = 0

#-------------------------------------------------------------------------------
#  Deferred definitions:
#
#  The following classes have a 'chicken and the egg' definition problem. They
#  require Traits to work, because they subclass Traits, but the Traits
#  meta-class programming support uses them, so Traits can't be subclassed
#  until they are defined.
#
#  Note: We need to look at whether the Category support could be used to
#        allow us to implement this better.
#-------------------------------------------------------------------------------

class ViewElement ( object ):
    pass

def ViewElements ( ):
    return None

#-------------------------------------------------------------------------------
#  Constants:
#-------------------------------------------------------------------------------

WrapperTypes   = ( StaticAnyTraitChangeNotifyWrapper,
                   StaticTraitChangeNotifyWrapper )

MethodTypes    = ( MethodType,   CTraitMethod )
FunctionTypes  = ( FunctionType, CTraitMethod )

# Class dictionary entries used to save trait, listener and view information and
# definitions:

BaseTraits      = '__base_traits__'
ClassTraits     = '__class_traits__'
PrefixTraits    = '__prefix_traits__'
ListenerTraits  = '__listener_traits__'
ViewTraits      = '__view_traits__'
InstanceTraits  = '__instance_traits__'

# The default Traits View name
DefaultTraitsView = 'traits_view'

# Trait types which cannot have default values
CantHaveDefaultValue = ( 'event', 'delegate', 'constant' )

# An empty list
EmptyList = []

# The trait types that should be copied last when doing a 'copy_traits':
DeferredCopy = ( 'delegate', 'property' )

# Quick test for normal vs extended trait name
extended_trait_pat = re.compile( r'.*[ :\+\-,\.\*\?\[\]]' )

# Generic 'Any' trait:
any_trait = Any().as_ctrait()

#-------------------------------------------------------------------------------
#  Creates a clone of a specified trait:
#-------------------------------------------------------------------------------

def _clone_trait ( clone, metadata = None ):
    """ Creates a clone of a specified trait.
    """
    trait = CTrait( 0 )
    trait.clone( clone )

    if clone.__dict__ is not None:
        trait.__dict__ = clone.__dict__.copy()

    if metadata is not None:
        trait.__dict__.update( metadata )

    return trait

#-------------------------------------------------------------------------------
#  Gets the definition of a specified method (if any):
#-------------------------------------------------------------------------------

def _get_method ( cls, method ):
    result = getattr( cls, method, None )
    if (result is not None) and is_method_type(result):
        return result
    return None

def _get_def ( class_name, class_dict, bases, method ):
    """ Gets the definition of a specified method (if any).
    """
    if method[0:2] == '__':
        method = '_%s%s' % ( class_name, method )

    result = class_dict.get( method )
    if ((result is not None) and
        is_function_type(result) and
        (getattr( result, 'on_trait_change', None ) is None)):
        return result

    for base in bases:
        result = getattr( base, method, None )
        if ((result is not None) and
            is_method_type(result) and \
            (getattr( result.im_func, 'on_trait_change', None ) is None)):
            return result

    return None


def is_cython_func_or_method(method):
    """ Test if the given input is a Cython method or function. """
    # The only way to get the type from the method with str comparison ...
    return 'cython_function_or_method' in str(type(method))

def is_method_type(method):
    """ Test if the given input is a Python method or a Cython method. """
    return isinstance(method, MethodTypes ) or is_cython_func_or_method(method)

def is_function_type(function):
    """ Test if the given input is a Python function or a Cython method. """
    return isinstance(function, FunctionTypes ) or \
           is_cython_func_or_method(function)

#-------------------------------------------------------------------------------
#  Returns whether or not a specified value is serializable:
#-------------------------------------------------------------------------------

def _is_serializable ( value ):
    """ Returns whether or not a specified value is serializable.
    """
    if isinstance( value, ( list, tuple ) ):
        for item in value:
            if not _is_serializable( item ):
                return False

        return True

    if isinstance( value, dict ):
        for name, item in value.items():
            if ((not _is_serializable( name )) or
                (not _is_serializable( item ))):
                return False

        return True

    return ((not isinstance( value, HasTraits )) or
            value.has_traits_interface( ISerializable ))

#-------------------------------------------------------------------------------
#  Returns a dictionary of potential 'Instance' or 'List(Instance)' handlers:
#-------------------------------------------------------------------------------

def _get_instance_handlers ( class_dict, bases ):
    """ Returns a dictionary of potential 'Instance' or 'List(Instance)'
        handlers.
    """
    # Create the results dictionary:
    instance_traits = {}

    # Merge all of the base class information into the result:
    for base in bases:
        for name, base_arg_lists in base.__dict__.get( InstanceTraits ).items():
            arg_lists = instance_traits.get( name )
            if arg_lists is None:
                instance_traits[ name ] = base_arg_lists[:]
            else:
                for arg_list in base_arg_lists:
                    if arg_list not in arg_lists:
                        arg_lists.append( arg_list )

    # Merge in the information from the class dictionary:
    for name, value in class_dict.items():
        if (name[:1] == '_') and is_function_type(value):
            n   = 13
            col = name.find( '_changed_for_' )
            if col < 2:
                 n   = 11
                 col = name.find( '_fired_for_' )
            if col >= 2:
                key = name[ col + n: ]
                if key != '':
                    arg_list  = ( name, name[ 1: col ] )
                    arg_lists = instance_traits.setdefault( key, [] )
                    if arg_list not in arg_lists:
                        arg_lists.append( arg_list )

    # Return the dictionary of possible arg_lists:
    return instance_traits

#-------------------------------------------------------------------------------
#  Returns the correct 'delegate' listener pattern for a specified name and
#  delegate trait:
#-------------------------------------------------------------------------------

def get_delegate_pattern ( name, trait ):
    """ Returns the correct 'delegate' listener pattern for a specified name and
        delegate trait.
    """
    prefix = trait._prefix
    if prefix == '':
        prefix = name
    elif (len( prefix ) > 1) and (prefix[-1] == '*'):
        prefix = prefix[:-1] + name

    return ' %s:%s' % ( trait._delegate, prefix )

#-------------------------------------------------------------------------------
#  '_SimpleTest' class:
#-------------------------------------------------------------------------------

class _SimpleTest:
    def __init__ ( self, value ): self.value = value
    def __call__ ( self, test  ): return test == self.value

#-------------------------------------------------------------------------------
#  Checks if a function can be converted to a 'trait method' (and converts it if
#  possible):
#-------------------------------------------------------------------------------

def _check_method ( class_dict, name, func ):
    method_name  = name
    return_trait = Any

    col = name.find( '__' )
    if col >= 1:
        type_name    = name[ : col ]
        method_name  = name[ col + 2: ]

        return_trait = globals().get( type_name )
        if not isinstance( return_trait, CTrait ):
            return_trait = SpecialNames.get( type_name.lower() )
            if return_trait is None:
                return_trait = Any
                method_name  = name

    has_traits = (method_name != name)
    arg_traits = []

    defaults = func.func_defaults
    if defaults is not None:
        for trait in defaults:
            trait = _check_trait( trait )
            if isinstance( trait, CTrait ):
                has_traits = True
            else:
                trait = Any( trait ).as_ctrait()
            arg_traits.append( trait )

    if has_traits:
        code       = func.func_code
        var_names  = code.co_varnames
        arg_traits = (([ Missing ] * (code.co_argcount - len( arg_traits ))) +
                      arg_traits)
        traits     = []
        for i, trait in enumerate( arg_traits ):
            traits.append( var_names[i] )
            traits.append( trait )
        del class_dict[ name ]
        class_dict[ method_name ] = CTraitMethod( method_name, func,
                                          tuple( [ return_trait ] + traits ) )

#-------------------------------------------------------------------------------
#  Returns either the original value or a valid CTrait if the value can be
#  converted to a CTrait:
#-------------------------------------------------------------------------------

def _check_trait ( trait ):
    """ Returns either the original value or a valid CTrait if the value can be
        converted to a CTrait.
    """
    if isinstance( trait, CTrait ):
        return trait

    if isinstance( trait, TraitFactory ):
        return trait_factory( trait )

    if isinstance( trait, type ) and issubclass( trait, TraitType ):
        trait = trait()

    if isinstance( trait, TraitType ):
        return trait.as_ctrait()

    return trait

#-------------------------------------------------------------------------------
#  Returns the trait corresponding to a specified value:
#-------------------------------------------------------------------------------

def _trait_for ( trait ):
    """ Returns the trait corresponding to a specified value.
    """
    trait = _check_trait( trait )
    if isinstance( trait, CTrait ):
        return trait

    return Trait( trait )

#-------------------------------------------------------------------------------
#  Returns the 'mapped trait' definition for a mapped trait:
#-------------------------------------------------------------------------------

def _mapped_trait_for ( trait ):
    """ Returns the 'mapped trait' definition for a mapped trait.
    """
    default_value = trait.default_value()[1]
    try:
        default_value = trait.handler.mapped_value( default_value )
    except:
        pass

    return Any( default_value, is_base  = False, transient = True,
                               editable = False ).as_ctrait()

#-------------------------------------------------------------------------------
#  Adds a list of handlers to a specified notifiers list:
#-------------------------------------------------------------------------------

def _add_notifiers ( notifiers, handlers ):
    """ Adds a list of handlers to a specified notifiers list.
    """
    for handler in handlers:
        if not isinstance( handler, WrapperTypes ):
            handler = StaticTraitChangeNotifyWrapper( handler )
        notifiers.append( handler )

#-------------------------------------------------------------------------------
#  Adds any specified event handlers defined for a trait by a class:
#-------------------------------------------------------------------------------

def _add_event_handlers ( trait, cls, handlers ):
    """ Adds any specified event handlers defined for a trait by a class.
    """
    events = trait.event
    if events is not None:
        if isinstance(events, basestring):
            events = [ events ]

        for event in events:
            handlers.append( _get_method( cls, '_%s_changed' % event ) )
            handlers.append( _get_method( cls, '_%s_fired'   % event ) )

#-------------------------------------------------------------------------------
#  Returns the method associated with a particular class property getter/setter:
#-------------------------------------------------------------------------------

def _property_method ( class_dict, name ):
    """ Returns the method associated with a particular class property
    getter/setter.
    """
    return class_dict.get( name )

#-------------------------------------------------------------------------------
#  Defines a factory function for creating type checked methods:
#-------------------------------------------------------------------------------

def trait_method ( func, return_type, **arg_types ):
    """ Factory function for creating type-checked methods.

    Parameters
    ----------
    func : function
        The method to be type-checked.
    return_type :
        The return type of the method, a trait or value that can be converted
        to a trait using Trait().
    **arg_types :
        Zero or more '*keyword* = *trait*' pairs, the argument names and types
        of parameters of the type-checked method. The *trait* portion of each
        pair must be a trait or a value that can be converted to a trait using
        Trait().
    """
    # Make the sure the first argument is a function:
    if type( func ) is not FunctionType:
        if type( return_type ) is not FunctionType:
            raise TypeError, "First or second argument must be a function."
        else:
            func, return_type = return_type, func

    # Make sure the return type is a trait (if not, coerce it to one):
    return_type = _trait_for( return_type )

    # Make up the list of arguments defined by the function we are wrapping:
    code       = func.func_code
    arg_count  = code.co_argcount
    var_names  = code.co_varnames[ : arg_count ]
    defaults   = func.func_defaults or ()
    defaults   = ( Missing, ) * (arg_count - len( defaults )) + defaults
    arg_traits = []
    for i, name in enumerate( var_names ):
        try:
            trait = arg_types[ name ]
            del arg_types[ name ]
        except:
            # fixme: Should this be a hard error (i.e. missing parameter type?)
            trait = Any

        arg_traits.append( name )
        arg_traits.append( Trait( defaults[i], _trait_for( trait ) ) )

    # Make sure there are no unaccounted for type parameters left over:
    if len( arg_types ) > 0:
        names = arg_types.keys()
        if len( names ) == 1:
            raise TraitError, ("The '%s' keyword defines a type for an "
                               "argument which '%s' does not have." % (
                               names[0], func.func_name ))
        else:
            names.sort()
            raise TraitError, ("The %s keywords define types for arguments "
                               "which '%s' does not have." % (
                               ', '.join( [ "'%s'" % name for name in names ] ),
                               func.func_name ))

    # Otherwise, return a method wrapper for the function:
    return CTraitMethod( func.func_name, func,
                                         tuple( [ return_type ] + arg_traits ) )

#-------------------------------------------------------------------------------
#  Defines a method 'decorator' for adding type checking to methods:
#-------------------------------------------------------------------------------

def _add_assignment_advisor ( callback, depth = 2 ):
    """ Defines a method 'decorator' for adding type checking to methods.
    """
    frame      = sys._getframe( depth )
    old_trace  = [ frame.f_trace ]
    old_locals = frame.f_locals.copy()

    def tracer ( frm, event, arg ):

        if event == 'call':
            if old_trace[0]:
                return old_trace[0]( frm, event, arg )
            else:
                return None
        try:
            if frm is frame and event != 'exception':
                for k, v in frm.f_locals.items():
                    if k not in old_locals:
                        del frm.f_locals[k]
                        break
                    elif old_locals[k] is not v:
                        frm.f_locals[k] = old_locals[k]
                        break
                else:
                    return tracer

                callback( frm, k, v )

        finally:
            if old_trace[0]:
                old_trace[0] = old_trace[0]( frm, event, arg )

        frm.f_trace = old_trace[0]
        sys.settrace( old_trace[0] )
        return old_trace[0]

    frame.f_trace = tracer
    sys.settrace( tracer )

def method ( return_type = Any, *arg_types, **kwarg_types ):
    """ Declares that the method defined immediately following a call to this
    function is type-checked.

    Parameters
    ----------
    return_type : type
        The type returned by the type-checked method. Must be either a trait
        or a value that can be converted to a trait using the Trait()
        function. The default of Any means that the return value is not
        type-checked.
    *arg_types :
        Zero or more types of positional parameters of the type-checked method.
        Each value must either a trait or a value that can be converted to a
        trait using the Trait() function.
    **kwarg_types :
        Zero or more *keyword* = *type* pairs, the type names and types of
        keyword parameters of the type-checked method. The *type* portion of
        the parameter must be either a trait or a value that can be converted
        to a trait using the Trait() function.

    Description
    -----------
    Whenever the type-checked method is called, the method() function ensures
    that each parameter passed to the method of the type specified by
    *arg_types* and *kwarg_types*, and that the return value is of the type
    specified by *return_type*. It is an error to specify both positional and
    keyword definitions for the same method parameter. If a parameter defined by
    the type-checked method is not referenced in the method() call, the
    parameter is not type-checked (i.e., its type is implicitly set to Any).
    If the call to method() signature contains an *arg_types* or *kwarg_types*
    parameter that does not correspond to a parameter in the type-checked method
    definition, a TraitError exception is raised.
    """
    # The following is a 'hack' to get around what seems to be a Python bug
    # that does not pass 'return_type' and 'arg_types' through to the scope of
    # 'callback' below:
    kwarg_types[''] = ( return_type, arg_types )

    def callback ( frame, method_name, func ):

        # This undoes the work of the 'hack' described above:
        return_type, arg_types = kwarg_types['']
        del kwarg_types['']

        # Add a 'fake' positional argument as a place holder for 'self':
        arg_types = ( Any, ) + arg_types

        # Make the sure the first argument is a function:
        if type( func ) is not FunctionType:
            raise TypeError, ("'method' must immediately precede a method "
                              "definition.")

        # Make sure the return type is a trait (if not, coerce it to one):
        return_type = _trait_for( return_type )

        # Make up the list of arguments defined by the function we are wrapping:
        code       = func.func_code
        func_name  = func.func_name
        arg_count  = code.co_argcount
        var_names  = code.co_varnames[ : arg_count ]
        defaults   = func.func_defaults or ()
        defaults   = ( Missing, ) * (arg_count - len( defaults )) + defaults
        arg_traits = []
        n          = len( arg_types )
        if n > len( var_names ):
            raise TraitError, ("Too many positional argument types specified "
                               "in the method signature for %s" % func_name)
        for i, name in enumerate( var_names ):
            if (i > 0) and (i < n):
                if name in kwarg_types:
                    raise TraitError, ("The '%s' argument is defined by both "
                                       "a positional and keyword argument in "
                                       "the method signature for %s" %
                                       ( name, func_name ) )
                trait = arg_types[i]
            else:
                try:
                    trait = kwarg_types[ name ]
                    del kwarg_types[ name ]
                except:
                    # fixme: Should this be an error (missing parameter type?)
                    trait = Any
            arg_traits.append( name )
            arg_traits.append( Trait( defaults[i], _trait_for( trait ) ) )

        # Make sure there are no unaccounted for type parameters left over:
        if len( kwarg_types ) > 0:
            names = kwarg_types.keys()
            if len( names ) == 1:
                raise TraitError, ("The '%s' method signature keyword defines "
                                   "a type for an argument which '%s' does not "
                                   "have." % ( names[0], func_name ))
            else:
                names.sort()
                raise TraitError, ("The %s method signature keywords define "
                          "types for arguments which '%s' does not have." % (
                          ', '.join( [ "'%s'" % name for name in names ] ),
                          func_name ))

        # Otherwise, return a method wrapper for the function:
        frame.f_locals[ method_name ] = CTraitMethod( func_name, func,
                                         tuple( [ return_type ] + arg_traits ) )

    _add_assignment_advisor( callback )

#-------------------------------------------------------------------------------
#  'MetaHasTraits' class:
#-------------------------------------------------------------------------------

# This really should be 'HasTraits', but it's not defined yet:
_HasTraits = None

class MetaHasTraits ( type ):
    ### JMS: Need a docstring here.
    # All registered class creation listeners.
    #
    # { Str class_name : Callable listener }
    _listeners = {}

    def __new__ ( cls, class_name, bases, class_dict ):
        mhto = MetaHasTraitsObject( cls, class_name, bases, class_dict, False )

        # Finish building the class using the updated class dictionary:
        klass = type.__new__( cls, class_name, bases, class_dict )

        # Fix up all self referential traits to refer to this class:
        for trait in mhto.self_referential:
            trait.set_validate( ( 11, klass ) )

        # Call all listeners that registered for this specific class:
        name = '%s.%s' % ( klass.__module__, klass.__name__ )
        for listener in MetaHasTraits._listeners.get( name, [] ):
            listener( klass )

        # Call all listeners that registered for ANY class:
        for listener in MetaHasTraits._listeners.get( '', [] ):
            listener( klass )

        return klass

    def add_listener ( cls, listener, class_name = '' ):
        """ Adds a class creation listener.

        If the class name is the empty string then the listener will be called
        when *any* class is created.
        """
        MetaHasTraits._listeners.setdefault( class_name, [] ).append( listener )

    add_listener = classmethod( add_listener )

    def remove_listener ( cls, listener, class_name = '' ):
        """ Removes a class creation listener.
        """
        MetaHasTraits._listeners[ class_name ].remove( listener )

    remove_listener = classmethod( remove_listener )

#-------------------------------------------------------------------------------
#  'MetaHasTraitsObject' class:
#-------------------------------------------------------------------------------

class MetaHasTraitsObject ( object ):
    """ Performs all of the meta-class processing needed to convert any
        subclass of HasTraits into a well-formed traits class.
    """

    def __init__ ( self, cls, class_name, bases, class_dict, is_category ):
        """ Processes all of the traits related data in the class dictionary.
        """
        # Create the various class dictionaries, lists and objects needed to
        # hold trait and view information and definitions:
        base_traits      = {}
        class_traits     = {}
        prefix_traits    = {}
        listeners        = {}
        prefix_list      = []
        override_bases   = bases
        view_elements    = ViewElements()
        self_referential = []

        # Create a list of just those base classes that derive from HasTraits:
        hastraits_bases = [ base for base in bases
                            if base.__dict__.get( ClassTraits ) is not None ]

        # Create a list of all inherited trait dictionaries:
        inherited_class_traits = [ base.__dict__.get( ClassTraits )
                                   for base in hastraits_bases ]

        # Move all trait definitions from the class dictionary to the
        # appropriate trait class dictionaries:
        for name, value in class_dict.items():
            value = _check_trait( value )
            rc    = isinstance( value, CTrait )

            if (not rc) and isinstance( value, ForwardProperty ):
                rc     = True
                getter = _property_method( class_dict, '_get_' + name )
                setter = _property_method( class_dict, '_set_' + name )
                if (setter is None) and (getter is not None):
                    if getattr( getter, 'settable', False ):
                        setter = HasTraits._set_traits_cache
                    elif getattr( getter, 'flushable', False ):
                        setter = HasTraits._flush_traits_cache
                validate = _property_method( class_dict, '_validate_' + name )
                if validate is None:
                    validate = value.validate

                value = Property( getter, setter, validate, True,
                                  value.handler, **value.metadata )
            if rc:
                del class_dict[ name ]
                if name[-1:] != '_':
                    base_traits[ name ] = class_traits[ name ] = value
                    value_type = value.type
                    if value_type == 'trait':
                       handler = value.handler
                       if handler is not None:
                           if handler.has_items:
                               items_trait = _clone_trait(
                                   handler.items_event(), value.__dict__ )

                               if items_trait.instance_handler == \
                                   '_list_changed_handler':
                                   items_trait.instance_handler = \
                                       '_list_items_changed_handler'

                               class_traits[ name + '_items' ] = items_trait

                           if handler.is_mapped:
                               class_traits[ name + '_' ] = _mapped_trait_for(
                                                                         value )

                           if isinstance( handler, This ):
                               handler.info_text = \
                                   add_article( class_name ) + ' instance'
                               self_referential.append( value )

                    elif value_type == 'delegate':
                        # Only add a listener if the trait.listenable metadata
                        # is not False:
                        if value._listenable is not False:
                            listeners[ name ] = ( 'delegate',
                                           get_delegate_pattern( name, value ) )
                    elif value_type == 'event':
                        on_trait_change = value.on_trait_change
                        if isinstance( on_trait_change, basestring ):
                            listeners[ name ] = ( 'event', on_trait_change )
                else:
                    name = name[:-1]
                    prefix_list.append( name )
                    prefix_traits[ name ] = value

            elif isinstance( value, FunctionType ) or is_cython_func_or_method(value):
                pattern = getattr( value, 'on_trait_change', None )
                if pattern is not None:
                    listeners[ name ] = ( 'method', pattern )

                _check_method( class_dict, name, value )

            elif isinstance( value, property ):
                class_traits[ name ] = generic_trait

            # Handle any view elements found in the class:
            elif isinstance( value, ViewElement ):

                # Add the view element to the class's 'ViewElements' if it is
                # not already defined (duplicate definitions are errors):
                if name in view_elements.content:
                    raise TraitError(
                        "Duplicate definition for view element '%s'" % name )

                view_elements.content[ name ] = value

                # Replace all substitutable view sub elements with 'Include'
                # objects, and add the substituted items to the
                # 'ViewElements':
                value.replace_include( view_elements )

                # Remove the view element from the class definition:
                del class_dict[ name ]

            else:
                for ct in inherited_class_traits:
                    if name in ct:
                        # The subclass is providing a default value for the
                        # trait defined in a superclass.
                        ictrait = ct[ name ]
                        if ictrait.type in CantHaveDefaultValue:
                            raise TraitError( "Cannot specify a default value "
                                "for the %s trait '%s'. You must override the "
                                "the trait definition instead." %
                                ( ictrait.type, name ) )

                        default_value = value
                        class_traits[ name ] = value = ictrait( default_value )
                        # Make sure that the trait now has the default value
                        # has the correct initializer.
                        value.default_value(1, value.default)
                        del class_dict[ name ]
                        override_bases = []
                        handler        = value.handler
                        if (handler is not None) and handler.is_mapped:
                            class_traits[ name + '_' ] = _mapped_trait_for(
                                                                         value )
                        break

        # Process all HasTraits base classes:
        migrated_properties = {}
        implements          = []
        for base in hastraits_bases:
            base_dict = base.__dict__

            # Merge listener information:
            for name, value in base_dict.get( ListenerTraits ).items():
                if (name not in class_traits) and (name not in class_dict):
                    listeners[ name ] = value

            # Merge base traits:
            for name, value in base_dict.get( BaseTraits ).items():
                if name not in base_traits:
                    property_info = value.property()
                    if property_info is not None:
                        key = id( value )
                        migrated_properties[ key ] = value = \
                            self.migrate_property( name, value, property_info,
                                                   class_dict )
                    base_traits[ name ] = value

                elif is_category:
                    raise TraitError, ("Cannot override '%s' trait "
                                       "definition in a category" % name)

            # Merge class traits:
            for name, value in base_dict.get( ClassTraits ).items():
                if name not in class_traits:
                    property_info = value.property()
                    if property_info is not None:
                        new_value = migrated_properties.get( id( value ) )
                        if new_value is not None:
                            value = new_value
                        else:
                            value = self.migrate_property( name, value,
                                                     property_info, class_dict )
                    class_traits[ name ] = value

                elif is_category:
                    raise TraitError, ("Cannot override '%s' trait "
                                       "definition in a category" % name)

            # Merge prefix traits:
            base_prefix_traits = base_dict.get( PrefixTraits )
            for name in base_prefix_traits['*']:
                if name not in prefix_list:
                    prefix_list.append( name )
                    prefix_traits[ name ] = base_prefix_traits[ name ]
                elif is_category:
                    raise TraitError, ("Cannot override '%s_' trait "
                                       "definition in a category" % name)

            # If the base class has a 'ViewElements' object defined, add it to
            # the 'parents' list of this class's 'ViewElements':
            parent_view_elements = base_dict.get( ViewTraits )
            if parent_view_elements is not None:
                view_elements.parents.append( parent_view_elements )

        # Make sure there is a definition for 'undefined' traits:
        if (prefix_traits.get( '' ) is None) and (not is_category):
            prefix_list.append( '' )
            prefix_traits[''] = Python().as_ctrait()

        # Save a link to the prefix_list:
        prefix_traits['*'] = prefix_list

        # Make sure the trait prefixes are sorted longest to shortest
        # so that we can easily bind dynamic traits to the longest matching
        # prefix:
        prefix_list.sort( lambda x, y: len( y ) - len( x ) )

        # Get the list of all possible 'Instance'/'List(Instance)' handlers:
        instance_traits = _get_instance_handlers( class_dict, hastraits_bases )

        # If there is an 'anytrait_changed' event handler, wrap it so that
        # it can be attached to all traits in the class:
        anytrait = _get_def( class_name, class_dict, bases,
                             '_anytrait_changed' )
        if anytrait is not None:
            anytrait = StaticAnyTraitChangeNotifyWrapper( anytrait )

            # Save it in the prefix traits dictionary so that any dynamically
            # created traits (e.g. 'prefix traits') can re-use it:
            prefix_traits['@'] = anytrait

        # Make one final pass over the class traits dictionary, making sure
        # all static trait notification handlers are attached to a 'cloned'
        # copy of the original trait:
        cloned = set()
        for name in class_traits.keys():
            trait    = class_traits[ name ]
            handlers = [ anytrait,
                         _get_def( class_name, class_dict, bases,
                                   '_%s_changed' % name ),
                         _get_def( class_name, class_dict, bases,
                                   '_%s_fired' % name ) ]

            # Check for an 'Instance' or 'List(Instance)' trait with defined
            # handlers:
            instance_handler = trait.instance_handler
            if ((instance_handler is not None) and
                (name in instance_traits) or
                ((instance_handler == '_list_items_changed_handler') and
                 (name[-6:] == '_items') and
                 (name[:-6] in instance_traits))):
                handlers.append( getattr( HasTraits, instance_handler ) )

            events = trait.event
            if events is not None:

                if isinstance(events, basestring):
                    events = [ events ]

                for event in events:
                    handlers.append( _get_def( class_name, class_dict, bases,
                                               '_%s_changed' % event ) )
                    handlers.append( _get_def( class_name, class_dict, bases,
                                               '_%s_fired' % event ) )

            handlers = [ h for h in handlers if h is not None ]
            default  = _get_def( class_name, class_dict, [],
                                 '_%s_default' % name )
            if (len( handlers ) > 0) or (default is not None):

                if name not in cloned:
                    cloned.add( name )
                    class_traits[ name ] = trait = _clone_trait( trait )

                if len( handlers ) > 0:
                    _add_notifiers( trait._notifiers( 1 ), handlers )

                if default is not None:
                    trait.default_value( 8, default )

            # Handle the case of properties whose value depends upon the value
            # of other traits:
            if (trait.type == 'property') and (trait.depends_on is not None):

                cached = trait.cached
                if cached is True:
                    cached = TraitsCache + name

                depends_on = trait.depends_on
                if isinstance( depends_on, SequenceTypes ):
                    depends_on = ','.join( depends_on )
                else:
                    # Note: We add the leading blank to force it to be treated
                    # as using the extended trait notation so that it will
                    # automatically add '_items' listeners to lists/dicts:
                    depends_on = ' ' + depends_on

                listeners[ name ] = ( 'property', cached, depends_on )

        # Save the list of self referential traits:
        self.self_referential = self_referential

        # Add the traits meta-data to the class:
        self.add_traits_meta_data(
            bases, class_dict, base_traits, class_traits, instance_traits,
            prefix_traits, listeners, view_elements )

    #---------------------------------------------------------------------------
    #  Adds the traits meta-data to the class:
    #---------------------------------------------------------------------------

    def add_traits_meta_data ( self, bases, class_dict, base_traits,
                               class_traits,  instance_traits, prefix_traits,
                               listeners, view_elements ):
        """ Adds the Traits metadata to the class dictionary.
        """
        class_dict[ BaseTraits      ] = base_traits
        class_dict[ ClassTraits     ] = class_traits
        class_dict[ InstanceTraits  ] = instance_traits
        class_dict[ PrefixTraits    ] = prefix_traits
        class_dict[ ListenerTraits  ] = listeners
        class_dict[ ViewTraits      ] = view_elements

    #---------------------------------------------------------------------------
    #  Migrates an existing property to the class being defined (allowing for
    #  method overrides):
    #---------------------------------------------------------------------------

    def migrate_property ( self, name, property, property_info, class_dict ):
        """ Migrates an existing property to the class being defined
        (allowing for method overrides).
        """
        get = _property_method( class_dict, '_get_' + name )
        set = _property_method( class_dict, '_set_' + name )
        val = _property_method( class_dict,
                                '_validate_' + name )
        if ((get is not None) or (set is not None) or (val is not None)):
            old_get, old_set, old_val = property_info
            return Property( get or old_get, set or old_set, val or old_val,
                             True, **property.__dict__ )

        return property

#-------------------------------------------------------------------------------
#  Manages the list of trait instance monitors:
#-------------------------------------------------------------------------------

def _trait_monitor_index ( cls, handler ):
    global _HasTraits_monitors

    type_handler = type( handler )
    for i, _cls, _handler in enumerate( _HasTraits_monitors ):
        if type_handler is type( _handler ):
            if (((type_handler is MethodType)  or
                'cython_function_or_method' in str(type_handler)) and \
                (handler.im_self is not None)):
                if ((handler.__name__ == _handler.__name__) and
                    (handler.im_self is _handler.im_self)):
                   return i

            elif handler == _handler:
                return i

    return -1

#-------------------------------------------------------------------------------
#  'HasTraits' decorators:
#-------------------------------------------------------------------------------

def on_trait_change ( name, post_init = False, *names ):
    """ Marks the following method definition as being a handler for the
        extended trait change specified by *name(s)*.

        Refer to the documentation for the on_trait_change() method of
        the **HasTraits** class for information on the correct syntax for
        the *name(s)* argument.

        A handler defined using this decorator is normally effective
        immediately. However, if *post_init* is **True**, then the handler only
        become effective after all object constructor arguments have been
        processed. That is, trait values assigned as part of object construction
        will not cause the handler to be invoked.
    """
    def decorator ( function ):
        prefix = '<'
        if post_init:
            prefix = '>'

        function.on_trait_change = prefix + \
                                   (','.join( [ name ] + list( names ) ))

        return function

    return decorator

def cached_property ( function ):
    """ Marks the following method definition as being a "cached property".
        That is, it is a property getter which, for performance reasons, caches
        its most recently computed result in an attribute whose name is of the
        form: *_traits_cache_name*, where *name* is the name of the property. A
        method marked as being a cached property needs only to compute and
        return its result. The @cached_property decorator automatically wraps
        the decorated method in cache management code, eliminating the need to
        write boilerplate cache management code explicitly. For example::

            file_name = File
            file_contents = Property( depends_on = 'file_name' )

            @cached_property
            def _get_file_contents(self):
                fh = open(self.file_name, 'rb')
                result = fh.read()
                fh.close()
                return result

        In this example, accessing the *file_contents* trait calls the
        _get_file_contents() method only once each time after the **file_name**
        trait is modified. In all other cases, the cached value
        **_file_contents**, which maintained by the @cached_property wrapper
        code, is returned.

        Note the use, in the example, of the **depends_on** metadata attribute
        to specify that the value of **file_contents** depends on **file_name**,
        so that _get_file_contents() is called only when **file_name** changes.
        For details, see the traits.traits.Property() function.
    """
    name = TraitsCache + function.__name__[ 5: ]

    def decorator ( self ):
        result = self.__dict__.get( name, Undefined )
        if result is Undefined:
            self.__dict__[ name ] = result = function( self )

        return result

    decorator.cached_property = True

    return decorator

def property_depends_on ( dependency, settable = False, flushable = False ):
    """ Marks the following method definition as being a "cached property"
        that depends on the specified extended trait names. That is, it is a
        property getter which, for performance reasons, caches its most recently
        computed result in an attribute whose name is of the form:
        *_traits_cache_name*, where *name* is the name of the property. A method
        marked as being a cached property needs only to compute and return its
        result. The @property_depends_on decorator automatically wraps the
        decorated method in cache management code that will cache the most
        recently computed value and flush the cache when any of the specified
        dependencies are modified, thus eliminating the need to write
        boilerplate cache management code explicitly. For example::

            file_name = File
            file_contents = Property

            @property_depends_on( 'file_name' )
            def _get_file_contents(self):
                fh = open(self.file_name, 'rb')
                result = fh.read()
                fh.close()
                return result

        In this example, accessing the *file_contents* trait calls the
        _get_file_contents() method only once each time after the **file_name**
        trait is modified. In all other cases, the cached value
        **_file_contents**, which is maintained by the @cached_property wrapper
        code, is returned.
    """
    def decorator ( function ):
        name = TraitsCache + function.__name__[ 5: ]

        def wrapper ( self ):
            result = self.__dict__.get( name, Undefined )
            if result is Undefined:
                self.__dict__[ name ] = result = function( self )

            return result

        wrapper.cached_property = True
        wrapper.depends_on      = dependency
        wrapper.settable        = settable
        wrapper.flushable       = flushable

        return wrapper

    return decorator

def weak_arg(arg):
    """ Create a weak reference to arg and wrap the function so that the
    dereferenced weakref is passed as the first argument. If arg has been
    deleted then the function is not called.
    """
    # Create the weak reference
    weak_arg = weakref.ref(arg)
    def decorator(function):
        # We need multiple wrappers to traits can find the number of arguments.
        # The all just dereference the weak reference and the call the
        # function if it is not None.
        def wrapper0():
            arg = weak_arg()
            if arg is not None:
                return function(arg)
        def wrapper1(arg1):
            arg = weak_arg()
            if arg is not None:
                return function(arg, arg1)
        def wrapper2(arg1, arg2):
            arg = weak_arg()
            if arg is not None:
                return function(arg, arg1, arg2)
        def wrapper3(arg1, arg2, arg3):
            arg = weak_arg()
            if arg is not None:
                return function(arg, arg1, arg2, arg3)
        def wrapper4(arg1, arg2, arg3, arg4):
            arg = weak_arg()
            if arg is not None:
                return function(arg, arg1, arg2, arg3, arg4)
        def wrappern(*args):
            arg = weak_arg()
            if arg is not None:
                function(arg, *args)
        # Return the correct wrapper depending on the arg count
        args = function.func_code.co_argcount-1
        if args == 0:
            return wrapper0
        elif args == 1:
            return wrapper1
        elif args == 2:
            return wrapper2
        elif args == 3:
            return wrapper3
        elif args == 4:
            return wrapper4
        else:
            return wrappern

    return decorator

#-------------------------------------------------------------------------------
#  'HasTraits' class:
#-------------------------------------------------------------------------------

class HasTraits ( CHasTraits ):
    """ Enables any Python class derived from it to have trait attributes.

    Most of the methods of HasTraits operated by default only on the trait
    attributes explicitly defined in the class definition. They do not operate
    on trait attributes defined by way of wildcards or by calling
    **add_trait()**.
    For example::

        >>>class Person(HasTraits):
        ...    name = Str
        ...    age  = Int
        ...    temp_ = Any
        >>>bob = Person()
        >>>bob.temp_lunch = 'sandwich'
        >>>bob.add_trait('favorite_sport', Str('football'))
        >>>print bob.trait_names()
        ['trait_added', 'age', 'name']

    In this example, the trait_names() method returns only the *age* and
    *name* attributes defined on the Person class. (The **trait_added**
    attribute is an explicit trait event defined on the HasTraits class.)
    The wildcard attribute *temp_lunch* and the dynamically-added trait
    attribute *favorite_sport* are not listed.
    """
    __metaclass__ = MetaHasTraits

    #-- Trait Prefix Rules -----------------------------------------------------

    #: Make traits 'property cache' values private with no type checking:
    _traits_cache__ = Any( private = True, transient = True )

    #-- Class Variables --------------------------------------------------------

    #: Mapping from dispatch type to notification wrapper class type
    wrappers = {
        'same':     TraitChangeNotifyWrapper,
        'extended': ExtendedTraitChangeNotifyWrapper,
        'new':      NewTraitChangeNotifyWrapper,
        'fast_ui':  FastUITraitChangeNotifyWrapper,
        'ui':       FastUITraitChangeNotifyWrapper
    }

    #-- Trait Definitions ------------------------------------------------------

    #: An event fired when a new trait is dynamically added to the object
    trait_added = Event( basestring )

    #: An event that can be fired to indicate that the state of the object has
    #: been modified
    trait_modified = Event

    #---------------------------------------------------------------------------
    #  Handles a 'trait_added' event being fired:
    #---------------------------------------------------------------------------

    def _trait_added_changed ( self, name ):
        """ Handles a 'trait_added' event being fired.
        """
        # fixme: This test should be made more comprehensive by also verifying
        # that if the trait name does end in '_items', its base trait is also
        # a list or dictionary (in order to eliminate a false positive on an
        # unfortunately named trait:
        trait = self.trait( name )
        if (trait.type == 'delegate') and (name[-6:] != '_items'):
            self._init_trait_delegate_listener( name, 'delegate',
                                           get_delegate_pattern( name, trait ) )

    #---------------------------------------------------------------------------
    #  Adds/Removes a trait instance creation monitor:
    #---------------------------------------------------------------------------

    def trait_monitor ( cls, handler, remove = False ):
        """Adds or removes the specified *handler* from the list of active
        monitors.

        Parameters
        ----------
        handler : function
            The function to add or remove as a monitor.
        remove : bool
            Flag indicating whether to remove (True) or add the specified
            handler as a monitor for this class.

        Description
        -----------
        If *remove* is omitted or False, the specified handler is added to
        the list of active monitors; if *remove* is True, the handler is
        removed from the active monitor list.

        """
        global _HasTraits_monitors

        index = _trait_monitor_index( cls, handler )
        if remove:
            if index >= 0:
                del _HasTraits_monitors[ index ]
            return

        if index < 0:
            _HasTraits_monitors.append( ( cls, handler ) )

    trait_monitor = classmethod( trait_monitor )

    #---------------------------------------------------------------------------
    #  Add a new class trait (i.e. applies to all instances and subclasses):
    #---------------------------------------------------------------------------

    def add_class_trait ( cls, name, *trait ):
        """ Adds a named trait attribute to this class.

        Parameters
        ----------
        name : str
            Name of the attribute to add.
        *trait :
            A trait or a value that can be converted to a trait using Trait()
            Trait definition of the attribute. It can be a single value or
            a list equivalent to an argument list for the Trait() function.

        """

        # Make sure a trait argument was specified:
        if len( trait ) == 0:
            raise ValueError, 'No trait definition was specified.'

        # Make sure only valid traits get added:
        if len( trait ) > 1:
            trait = Trait( *trait )
        else:
            trait = _trait_for( trait[0] )

        # Add the trait to the class:
        cls._add_class_trait( name, trait, False )

        # Also add the trait to all subclasses of this class:
        for subclass in cls.trait_subclasses( True ):
            subclass._add_class_trait( name, trait, True )

    add_class_trait = classmethod( add_class_trait )

    def _add_class_trait ( cls, name, trait, is_subclass ):
        # Get a reference to the class's dictionary and 'prefix' traits:
        class_dict    = cls.__dict__
        prefix_traits = class_dict[ PrefixTraits ]

        # See if the trait is a 'prefix' trait:
        if name[-1:] == '_':
            name = name[:-1]
            if name in prefix_traits:
                if is_subclass:
                    return
                raise TraitError( "The '%s_' trait is already defined." % name )
            prefix_traits[ name ] = trait

            # Otherwise, add it to the list of known prefixes:
            prefix_list = prefix_traits['*']
            prefix_list.append( name )

            # Resort the list from longest to shortest:
            prefix_list.sort( lambda x, y: len( y ) - len( x ) )

            return

        # Check to see if the trait is already defined:
        class_traits = class_dict[ ClassTraits ]
        if class_traits.get( name ) is not None:
            if is_subclass:
                return
            raise TraitError( "The '%s' trait is already defined." % name )

        # Check to see if the trait has additional sub-traits that need to be
        # defined also:
        handler = trait.handler
        if handler is not None:
            if handler.has_items:
                cls.add_class_trait( name + '_items', handler.items_event() )
            if handler.is_mapped:
                cls.add_class_trait( name + '_', _mapped_trait_for( trait ) )

        # Make the new trait inheritable (if allowed):
        if trait.is_base is not False:
            class_dict[ BaseTraits ][ name ] = trait

        # See if there are any static notifiers defined:
        handlers = [ _get_method( cls, '_%s_changed' % name ),
                     _get_method( cls, '_%s_fired'   % name ) ]

        # Add any special trait defined event handlers:
        _add_event_handlers( trait, cls, handlers )

        # Add the 'anytrait' handler (if any):
        handlers.append( prefix_traits.get( '@' ) )

        # Filter out any 'None' values:
        handlers = [ h for h in handlers if h is not None ]

        # If there are and handlers, add them to the trait's notifier's list:
        if len( handlers ) > 0:
            trait = _clone_trait( trait )
            _add_notifiers( trait._notifiers( 1 ), handlers )

        # Finally, add the new trait to the class trait dictionary:
        class_traits[ name ] = trait

    _add_class_trait = classmethod( _add_class_trait )

    #---------------------------------------------------------------------------
    #  Adds a 'category' to the class:
    #---------------------------------------------------------------------------

    def add_trait_category ( cls, category ):
        """ Adds a trait category to a class.
        """
        if issubclass( category, HasTraits ):
            cls._add_trait_category(
                getattr( category, BaseTraits ),
                getattr( category, ClassTraits ),
                getattr( category, InstanceTraits ),
                getattr( category, PrefixTraits ),
                getattr( category, ListenerTraits ),
                getattr( category, ViewTraits, None ) )

        # Copy all methods that are not already in the class from the category:
        for subcls in category.__mro__:
            for name, value in subcls.__dict__.items():
                if not hasattr( cls, name ):
                    setattr( cls, name, value )

    add_trait_category = classmethod( add_trait_category )

    #---------------------------------------------------------------------------
    #  Adds a 'category' to the class:
    #---------------------------------------------------------------------------

    def _add_trait_category ( cls, base_traits, class_traits, instance_traits,
                              prefix_traits, listeners, view_elements ):
        # Update the class and each of the existing subclasses:
        for subclass in [ cls ] + cls.trait_subclasses( True ):

            # Merge the 'base_traits':
            subclass_traits = getattr( subclass, BaseTraits )
            for name, value in base_traits.items():
                subclass_traits.setdefault( name, value )

            # Merge the 'class_traits':
            subclass_traits = getattr( subclass, ClassTraits )
            for name, value in class_traits.items():
                subclass_traits.setdefault( name, value )

            # Merge the 'instance_traits':
            subclass_traits = getattr( subclass, InstanceTraits )
            for name, arg_lists in instance_traits.items():
                subclass_arg_lists = subclass_traits.get( name )
                if subclass_arg_lists is None:
                    subclass_traits[ name ] = arg_lists[:]
                else:
                    for arg_list in arg_lists:
                        if arg_list not in subclass_arg_lists:
                            subclass_arg_lists.append( arg_list )

            # Merge the 'prefix_traits':
            subclass_traits = getattr( subclass, PrefixTraits )
            subclass_list   = subclass_traits['*']
            changed         = False
            for name, value in prefix_traits.items():
                if name not in subclass_traits:
                    subclass_traits[ name ] = value
                    subclass_list.append( name )
                    changed = True

            # Resort the list from longest to shortest (if necessary):
            if changed:
                subclass_list.sort( lambda x, y: len( y ) - len( x ) )

            # Merge the 'listeners':
            subclass_traits = getattr( subclass, ListenerTraits )
            for name, value in listeners.items():
                subclass_traits.setdefault( name, value )

        # Copy all our new view elements into the base class's ViewElements:
        if view_elements is not None:
            content = view_elements.content
            if len( content ) > 0:
                base_ve = getattr( cls, ViewTraits, None )
                if base_ve is None:
                    base_ve = ViewElements()
                    setattr( cls, ViewTraits, base_ve )
                base_ve_content = base_ve.content
                for name, value in content.items():
                    base_ve_content.setdefault( name, value )

    _add_trait_category = classmethod( _add_trait_category )

    #---------------------------------------------------------------------------
    #  Sets a trait notification dispatch handler:
    #---------------------------------------------------------------------------

    def set_trait_dispatch_handler ( cls, name, klass, override = False ):
        """ Sets a trait notification dispatch handler.
        """
        try:
            if issubclass( klass, TraitChangeNotifyWrapper ):
                if (not override) and (name in cls.wrappers):
                    raise TraitError, ("A dispatch handler called '%s' has "
                                       "already been defined." % name)
                cls.wrappers[ name ] = klass
                return
        except TypeError:
            pass
        raise TraitError, ('%s is not a subclass of TraitChangeNotifyWrapper.' %
                           klass)

    set_trait_dispatch_handler = classmethod( set_trait_dispatch_handler )

    #---------------------------------------------------------------------------
    #  Returns the immediate (or all) subclasses of this class:
    #---------------------------------------------------------------------------

    def trait_subclasses ( cls, all = False ):
        """ Returns a list of the immediate (or all) subclasses of this class.

        Parameters
        ----------
        all : bool
            Indicates whether to return all subclasses of this class. If
            False, only immediate subclasses are returned.

        """
        if not all:
            return cls.__subclasses__()
        return cls._trait_subclasses( [] )

    trait_subclasses = classmethod( trait_subclasses )

    def _trait_subclasses ( cls, subclasses ):
        for subclass in cls.__subclasses__():
            if subclass not in subclasses:
                subclasses.append( subclass )
                subclass._trait_subclasses( subclasses )
        return subclasses

    _trait_subclasses = classmethod( _trait_subclasses )

    #---------------------------------------------------------------------------
    #  Returns whether the object implements a specified traits interface:
    #---------------------------------------------------------------------------

    def has_traits_interface ( self, *interfaces ):
        """Returns whether the object implements a specified traits interface.

           Parameters
           ----------
           *interfaces :
                One or more traits Interface (sub)classes.

           Description
           -----------
           Tests whether the object implements one or more of the interfaces
           specified by *interfaces*. Return **True** if it does, and **False**
           otherwise.
        """
        return isinstance(self, interfaces)

    #---------------------------------------------------------------------------
    #  Prepares an object to be pickled:
    #---------------------------------------------------------------------------

    def __getstate__ ( self ):
        """ Returns a dictionary of traits to pickle.

        In general, avoid overriding __getstate__ in subclasses. Instead, mark
        traits that should not be pickled with 'transient = True' metadata.

        In cases where this strategy is not sufficient, override __getstate__
        in subclasses using the following pattern to remove items that should
        not be persisted::

            def __getstate__(self):
                state = super(X,self).__getstate__()
                for key in ['foo', 'bar']:
                    if state.has_key(key):
                        del state[key]
                return state
        """
        # Save all traits which do not have any 'transient' metadata:
        result = self.trait_get( transient = is_none )

        # Add all delegate traits that explicitly have 'transient = False'
        # metadata:
        dic    = self.__dict__
        result.update( dict( [ ( name, dic[ name ] )
                             for name in self.trait_names( type = 'delegate',
                                                           transient = False )
                             if name in dic ] ) )

        # If this object implements ISerializable, make sure that all
        # contained HasTraits objects in its persisted state also implement
        # ISerializable:
        if self.has_traits_interface( ISerializable ):
            for name, value in result.items():
                if not _is_serializable( value ):
                    raise TraitError( "The '%s' trait of a '%s' instance "
                                      "contains the unserializable value: %s" %
                                      ( name, self.__class__.__name__, value ) )

        # Store the traits version in the state dictionary (if possible):
        result.setdefault( '__traits_version__', TraitsVersion )

        # Return the final state dictionary:
        return result

    def __reduce_ex__ ( self, protocol ):
        return ( __newobj__, ( self.__class__, ), self.__getstate__() )

    #---------------------------------------------------------------------------
    #  Restores the previously pickled state of an object:
    #---------------------------------------------------------------------------

    def __setstate__ ( self, state, trait_change_notify = True ):
        """ Restores the previously pickled state of an object.
        """
        pop = state.pop
        if pop( '__traits_version__', None ) is None:
            # If the state was saved by a version of Traits prior to 3.0, then
            # use Traits 2.0 compatible code to restore it:
            values = [ ( name, pop( name ) )
                       for name in pop( '__HasTraits_restore__', [] ) ]
            self.__dict__.update( state )
            self.trait_set( trait_change_notify=trait_change_notify,
                            **dict( values ) )
        else:
            # Otherwise, apply the Traits 3.0 restore logic:
            self._init_trait_listeners()
            self.trait_set( trait_change_notify = trait_change_notify, **state )
            self._post_init_trait_listeners()
            self.traits_init()

        self.traits_inited( True )

    #---------------------------------------------------------------------------
    #  Shortcut for retrieving the value of a list of traits:
    #---------------------------------------------------------------------------

    def trait_get ( self, *names, **metadata ):
        """ Shortcut for getting object trait attributes.

        Parameters
        ----------
        names : list of strings
            A list of trait attribute names whose values are requested.

        Returns
        -------
        result : dict
            A dictionary whose keys are the names passed as arguments and whose
            values are the corresponding trait values.

        Description
        -----------
        Looks up the value of each trait whose name is passed as an argument
        and returns a dictionary containing the resulting name/value pairs.
        If any name does not correspond to a defined trait, it is not included
        in the result.

        If no names are specified, the result is a dictionary containing
        name/value pairs for *all* traits defined on the object.
        """

        result = {}
        n      = len( names )
        if (n == 1) and (type( names[0] ) in SequenceTypes):
            names = names[0]
        elif n == 0:
            names = self.trait_names( **metadata )

        for name in names:
            value = getattr( self, name, Missing )
            if value is not Missing:
                result[ name ] = value

        return result

    # Defines the deprecated alias for 'trait_get'
    get = trait_get

    #---------------------------------------------------------------------------
    #  Shortcut for setting object traits:
    #---------------------------------------------------------------------------

    def trait_set ( self, trait_change_notify = True, **traits ):
        """ Shortcut for setting object trait attributes.

        Parameters
        ----------
        trait_change_notify : bool
            If **True** (the default), then each value assigned may generate a
            trait change notification. If **False**, then no trait change
            notifications will be generated. (see also: trait_setq)
        **traits :
            Key/value pairs, the trait attributes and their values to be
            set

        Returns
        -------
        self :
            The method returns this object, after setting attributes.

        Description
        -----------
        Treats each keyword argument to the method as the name of a trait
        attribute and sets the corresponding trait attribute to the value
        specified. This is a useful shorthand when a number of trait attributes
        need to be set on an object, or a trait attribute value needs to be set
        in a lambda function. For example, you can write::

            person.trait_set(name='Bill', age=27)

        instead of::

            person.name = 'Bill'
            person.age = 27

        """
        if not trait_change_notify:
            self._trait_change_notify( False )
            try:
                for name, value in traits.items():
                    setattr( self, name, value )
            finally:
                self._trait_change_notify( True )
        else:
            for name, value in traits.items():
                setattr( self, name, value )

        return self

    # Defines the deprecated alias for 'trait_set'
    set = trait_set

    def trait_setq ( self, **traits ):
        """ Shortcut for setting object trait attributes.

        Parameters
        ----------
        **traits :
            Key/value pairs, the trait attributes and their values to be set.
            No trait change notifications will be generated for any values
            assigned (see also: trait_set).

        Returns
        -------
        self :
            The method returns this object, after setting attributes.

        Description
        -----------
        Treats each keyword argument to the method as the name of a trait
        attribute and sets the corresponding trait attribute to the value
        specified. This is a useful shorthand when a number of trait attributes
        need to be set on an object, or a trait attribute value needs to be set
        in a lambda function. For example, you can write::

            person.trait_setq(name='Bill', age=27)

        instead of::

            person.name = 'Bill'
            person.age = 27

        """
        return self.trait_set( trait_change_notify = False, **traits )

    #---------------------------------------------------------------------------
    #  Resets some or all of an object's traits to their default values:
    #---------------------------------------------------------------------------

    def reset_traits ( self, traits = None, **metadata ):
        """ Resets some or all of an object's trait attributes to their default
        values.

        Parameters
        ----------
        traits : list of strings
            Names of trait attributes to reset.

        Returns
        -------
        unresetable : list of strings
            A list of attributes that the method was unable to reset, which is
            empty if all the attributes were successfully reset.

        Description
        -----------
        Resets each of the traits whose names are specified in the *traits*
        list to their default values. If *traits* is None or omitted, the
        method resets all explicitly-defined object trait attributes to their
        default values. Note that this does not affect wildcard trait
        attributes or trait attributes added via add_trait(), unless they are
        explicitly named in *traits*.

        """
        unresetable = []

        if traits is None:
            traits = self.trait_names( **metadata )

        for name in traits:
            try:
                delattr( self, name )
            except ( AttributeError, TraitError ):
                unresetable.append( name )

        return unresetable

    #---------------------------------------------------------------------------
    #  Returns the list of trait names to copy/clone by default:
    #---------------------------------------------------------------------------

    def copyable_trait_names ( self, **metadata ):
        """ Returns the list of trait names to copy or clone by default.
        """

        metadata.setdefault('transient', lambda t: t is not True)
        return self.trait_names( **metadata )

    #---------------------------------------------------------------------------
    #  Returns the list of all trait names, including implicitly defined
    #  traits:
    #---------------------------------------------------------------------------

    def all_trait_names ( self ):
        """ Returns the list of all trait names, including implicitly defined
            traits.
        """
        return self.__class_traits__.keys()

    #---------------------------------------------------------------------------
    #  Copies another object's traits into this one:
    #---------------------------------------------------------------------------

    def copy_traits ( self, other, traits = None, memo = None, copy = None,
                            **metadata ):
        """ Copies another object's trait attributes into this one.

        Parameters
        ----------
        other : object
            The object whose trait attribute values should be copied.
        traits : list of strings
            A list of names of trait attributes to copy. If None or
            unspecified, the set of names returned by trait_names() is used.
            If 'all' or an empty list, the set of names returned by
            all_trait_names() is used.
        memo : dict
            A dictionary of objects that have already been copied.
        copy : None | 'deep' | 'shallow'
            The type of copy to perform on any trait that does not have
            explicit 'copy' metadata. A value of None means 'copy reference'.

        Returns
        -------
        unassignable : list of strings
            A list of attributes that the method was unable to copy, which is
            empty if all the attributes were successfully copied.

        """

        if traits is None:
            traits = self.copyable_trait_names( **metadata )
        elif (traits == 'all') or (len( traits ) == 0):
            traits = self.all_trait_names()
            if memo is not None:
                memo[ 'traits_to_copy' ] = 'all'

        unassignable = []
        deferred     = []
        deep_copy    = (copy == 'deep')
        shallow_copy = (copy == 'shallow')

        for name in traits:
            try:
                trait = self.trait( name )
                if trait.type in DeferredCopy:
                    deferred.append( name )
                    continue

                base_trait = other.base_trait( name )
                if base_trait.type == 'event':
                    continue

                value     = getattr( other, name )
                copy_type = base_trait.copy
                if copy_type == 'shallow':
                    value = copy_module.copy( value )
                elif copy_type == 'ref':
                    pass
                elif (copy_type == 'deep') or deep_copy:
                    if memo is None:
                        value = copy_module.deepcopy( value )
                    else:
                        value = copy_module.deepcopy( value, memo )
                elif shallow_copy:
                    value = copy_module.copy( value )

                setattr( self, name, value )
            except:
                unassignable.append( name )

        for name in deferred:
            try:
                value     = getattr( other, name )
                copy_type = other.base_trait( name ).copy
                if copy_type == 'shallow':
                    value = copy_module.copy( value )
                elif copy_type == 'ref':
                    pass
                elif (copy_type == 'deep') or deep_copy:
                    if memo is None:
                        value = copy_module.deepcopy( value )
                    else:
                        value = copy_module.deepcopy( value, memo )
                elif shallow_copy:
                    value = copy_module.copy( value )

                setattr( self, name, value )
            except:
                unassignable.append( name )

        return unassignable

    #---------------------------------------------------------------------------
    #  Clones a new object from this one, optionally copying only a specified
    #  set of traits:
    #---------------------------------------------------------------------------

    def clone_traits ( self, traits = None, memo = None, copy = None,
                             **metadata ):
        """ Clones a new object from this one, optionally copying only a
        specified set of traits.

        Parameters
        ----------
        traits : list of strings
            The list of names of the trait attributes to copy.
        memo : dict
            A dictionary of objects that have already been copied.
        copy : str
            The type of copy ``deep`` or ``shallow`` to perform on any trait
            that does not have explicit 'copy' metadata. A value of None means
            'copy reference'.

        Returns
        -------
        new :
            The newly cloned object.

        Description
        -----------
        Creates a new object that is a clone of the current object. If *traits*
        is None (the default), then all explicit trait attributes defined
        for this object are cloned. If *traits* is 'all' or an empty list, the
        list of traits returned by all_trait_names() is used; otherwise,
        *traits* must be a list of the names of the trait attributes to be
        cloned.
        """
        if memo is None:
            memo = {}

        if traits is None:
            traits = self.copyable_trait_names( **metadata )
        elif (traits == 'all') or (len( traits ) == 0):
            traits = self.all_trait_names()
            memo[ 'traits_to_copy' ] = 'all'

        memo[ 'traits_copy_mode' ] = copy
        new = self.__new__( self.__class__ )
        memo[ id( self ) ] = new
        new._init_trait_listeners()
        new.copy_traits( self, traits, memo, copy, **metadata )
        new._post_init_trait_listeners()
        new.traits_init()
        new.traits_inited( True )

        return new

    #---------------------------------------------------------------------------
    #  Creates a deep copy of the object:
    #---------------------------------------------------------------------------

    def __deepcopy__ ( self, memo ):
        """ Creates a deep copy of the object.
        """
        id_self = id( self )
        if id_self in memo:
            return memo[ id_self ]

        result = self.clone_traits( memo   = memo,
                                    traits = memo.get( 'traits_to_copy' ),
                                    copy   = memo.get( 'traits_copy_mode' ) )

        return result

    #---------------------------------------------------------------------------
    #  Edits the object's traits:
    #---------------------------------------------------------------------------

    def edit_traits ( self, view       = None, parent  = None,
                            kind       = None, context = None,
                            handler    = None, id      = '',
                            scrollable = None, **args ):
        """ Displays a user interface window for editing trait attribute values.

        Parameters
        ----------
        view : View or string
            A View object (or its name) that defines a user interface for
            editing trait attribute values of the current object. If the view is
            defined as an attribute on this class, use the name of the attribute.
            Otherwise, use a reference to the view object. If this attribute is
            not specified, the View object returned by trait_view() is used.
        parent : toolkit control
            The reference to a user interface component to use as the parent
            window for the object's UI window.
        kind : str
            The type of user interface window to create. See the
            **traitsui.view.kind_trait** trait for values and
            their meanings. If *kind* is unspecified or None, the **kind**
            attribute of the View object is used.
        context : object or dictionary
            A single object or a dictionary of string/object pairs, whose trait
            attributes are to be edited. If not specified, the current object is
            used.
        handler : Handler
            A handler object used for event handling in the dialog box. If
            None, the default handler for Traits UI is used.
        id : str
            A unique ID for persisting preferences about this user interface,
            such as size and position. If not specified, no user preferences
            are saved.
        scrollable : bool
            Indicates whether the dialog box should be scrollable. When set to
            True, scroll bars appear on the dialog box if it is not large enough
            to display all of the items in the view at one time.
        """
        if context is None:
            context = self

        view = self.trait_view( view )

        return view.ui( context, parent, kind, self.trait_view_elements(),
                        handler, id, scrollable, args )

    #---------------------------------------------------------------------------
    #  Returns the default context to use for editing/configuring traits:
    #---------------------------------------------------------------------------

    def trait_context ( self ):
        """ Returns the default context to use for editing or configuring
            traits.
        """
        return { 'object': self }

    #---------------------------------------------------------------------------
    #  Gets or sets a ViewElement associated with an object's class:
    #---------------------------------------------------------------------------

    def trait_view ( self, name = None, view_element = None ):
        """ Gets or sets a ViewElement associated with an object's class.

        Parameters
        ----------
        name : str
            Name of a view element
        view_element : ViewElement
            View element to associate

        Returns
        -------
        A view element.

        Description
        -----------
        If both *name* and *view_element* are specified, the view element is
        associated with *name* for the current object's class. (That is,
        *view_element* is added to the ViewElements object associated with
        the current object's class, indexed by *name*.)

        If only *name* is specified, the function returns the view element
        object associated with *name*, or None if *name* has no associated
        view element. View elements retrieved by this function are those that
        are bound to a class attribute in the class definition, or that are
        associated with a name by a previous call to this method.

        If neither *name* nor *view_element* is specified, the method returns a
        View object, based on the following order of preference:

        1. If there is a View object named ``traits_view`` associated with the
           current object, it is returned.
        2. If there is exactly one View object associated the current
           object, it is returned.
        3. Otherwise, it returns a View object containing items for all the
           non-event trait attributes on the current object.

        """
        return self.__class__._trait_view( name, view_element,
                            self.default_traits_view, self.trait_view_elements,
                            self.editable_traits, self )

    def class_trait_view ( cls, name = None, view_element = None ):
        return cls._trait_view( name, view_element,
                  cls.class_default_traits_view, cls.class_trait_view_elements,
                  cls.class_editable_traits, None )

    class_trait_view = classmethod( class_trait_view )

    #---------------------------------------------------------------------------
    #  Gets or sets a ViewElement associated with an object's class:
    #---------------------------------------------------------------------------

    def _trait_view ( cls, name, view_element, default_name, view_elements,
                           editable_traits, handler ):
        """ Gets or sets a ViewElement associated with an object's class.
        """
        # If a view element was passed instead of a name or None, return it:
        if isinstance( name, ViewElement ):
            return name

        # Get the ViewElements object associated with the class:
        view_elements = view_elements()

        # The following test should only succeed for objects created before
        # traits has been fully initialized (such as the default Handler):
        if view_elements is None:
            return None

        if name:
            if view_element is None:
                # If only a name was specified, return the ViewElement it
                # matches, if any:
                result = view_elements.find( name )
                if (result is None) and (handler is not None):
                    method = getattr( handler, name, None )
                    if callable( method ):
                        result = method()

                return result

            # Otherwise, save the specified ViewElement under the name
            # specified:
            view_elements.content[ name ] = view_element

            return None

        # Get the default view/view name:
        name = default_name()

        # If the default is a View, return it:
        if isinstance( name, ViewElement ):
            return name

        # Otherwise, get all View objects associated with the object's class:
        names = view_elements.filter_by()

        # If the specified default name is in the list, return its View:
        if name in names:
            return view_elements.find( name )

        if handler is not None:
            method = getattr( handler, name, None )
            if callable( method ):
                result = method()
                if isinstance( result, ViewElement ):
                    return result

        # If there is only one View, return it:
        if len( names ) == 1:
            return view_elements.find( names[0] )

        # Otherwise, create and return a View based on the set of editable
        # traits defined for the object:
        from traitsui.api import View

        return View( editable_traits(), buttons = [ 'OK', 'Cancel' ] )

    _trait_view = classmethod( _trait_view )

    #---------------------------------------------------------------------------
    #  Return the default traits view/name:
    #---------------------------------------------------------------------------

    def default_traits_view ( self ):
        """ Returns the name of the default traits view for the object's class.
        """
        return self.__class__.class_default_traits_view()

    #---------------------------------------------------------------------------
    #  Return the default traits view/name:
    #---------------------------------------------------------------------------

    def class_default_traits_view ( cls ):
        """ Returns the name of the default traits view for the class.
        """
        return DefaultTraitsView

    class_default_traits_view = classmethod( class_default_traits_view )

    #---------------------------------------------------------------------------
    #  Gets the list of names of ViewElements associated with the object's
    #  class that are of a specified ViewElement type:
    #---------------------------------------------------------------------------

    def trait_views ( self, klass = None ):
        """ Returns a list of the names of all view elements associated with the
        current object's class.

        Parameters
        ----------
        klass : class
            A class, such that all returned names must correspond to instances
            of this class. Possible values include:

            * Group
            * Item
            * View
            * ViewElement
            * ViewSubElement

        Description
        -----------
        If *klass* is specified, the list of names is filtered such that only
        objects that are instances of the specified class are returned.
        """
        return self.__class__.__dict__[ ViewTraits ].filter_by( klass )

    #---------------------------------------------------------------------------
    #  Returns the ViewElements object associated with the object's class:
    #---------------------------------------------------------------------------

    def trait_view_elements ( self ):
        """ Returns the ViewElements object associated with the object's
        class.

        The returned object can be used to access all the view elements
        associated with the class.
        """
        return self.__class__.class_trait_view_elements()

    def class_trait_view_elements ( cls ):
        """ Returns the ViewElements object associated with the class.

        The returned object can be used to access all the view elements
        associated with the class.
        """
        return cls.__dict__[ ViewTraits ]

    class_trait_view_elements = classmethod( class_trait_view_elements )

    #---------------------------------------------------------------------------
    #  Configure the object's traits:
    #---------------------------------------------------------------------------

    def configure_traits ( self, filename = None, view       = None,
                                 kind     = None, edit       = True,
                                 context  = None, handler    = None,
                                 id       = '',   scrollable = None, **args ):
        ### JMS: Is it correct to assume that non-modal options for 'kind'
        ###      behave modally when called from this method?
        """Creates and displays a dialog box for editing values of trait
        attributes, as if it were a complete, self-contained GUI application.

        Parameters
        ----------
        filename : str
            The name (including path) of a file that contains a pickled
            representation of the current object. When this parameter is
            specified, the method reads the corresponding file (if it exists)
            to restore the saved values of the object's traits before displaying
            them. If the user confirms the dialog box (by clicking **OK**),
            the new values are written to the file. If this parameter is not
            specified, the values are loaded from the in-memory object, and are
            not persisted when the dialog box is closed.
        view : View or str
            A View object (or its name) that defines a user interface for
            editing trait attribute values of the current object. If the view is
            defined as an attribute on this class, use the name of the attribute.
            Otherwise, use a reference to the view object. If this attribute is
            not specified, the View object returned by trait_view() is used.
        kind : str
            The type of user interface window to create. See the
            **traitsui.view.kind_trait** trait for values and
            their meanings. If *kind* is unspecified or None, the **kind**
            attribute of the View object is used.
        edit : bool
            Indicates whether to display a user interface. If *filename*
            specifies an existing file, setting *edit* to False loads the
            saved values from that file into the object without requiring
            user interaction.
        context : object or dictionary
            A single object or a dictionary of string/object pairs, whose trait
            attributes are to be edited. If not specified, the current object is
            used
        handler : Handler
            A handler object used for event handling in the dialog box. If
            None, the default handler for Traits UI is used.
        id : str
            A unique ID for persisting preferences about this user interface,
            such as size and position. If not specified, no user preferences
            are saved.
        scrollable : bool
            Indicates whether the dialog box should be scrollable. When set to
            True, scroll bars appear on the dialog box if it is not large enough
            to display all of the items in the view at one time.

        Description
        -----------
        This method is intended for use in applications that do not normally
        have a GUI. Control does not resume in the calling application until
        the user closes the dialog box.

        The method attempts to open and unpickle the contents of *filename*
        before displaying the dialog box. When editing is complete, the method
        attempts to pickle the updated contents of the object back to *filename*.
        If the file referenced by *filename* does not exist, the object is not
        modified before displaying the dialog box. If *filename* is unspecified
        or None, no pickling or unpickling occurs.

        If *edit* is True (the default), a dialog box for editing the
        current object is displayed. If *edit* is False or None, no
        dialog box is displayed. You can use ``edit=False`` if you want the
        object to be restored from the contents of *filename*, without being
        modified by the user.
        """
        if filename is not None:
            fd = None
            try:
                import cPickle
                fd = open( filename, 'rb' )
                self.copy_traits( cPickle.Unpickler( fd ).load() )
            except:
                if fd is not None:
                    fd.close()

        if edit:
            from traitsui.api import toolkit
            if context is None:
                context = self
            rc = toolkit().view_application( context, self.trait_view( view ),
                                           kind, handler, id, scrollable, args )
            if rc and (filename is not None):
                fd = None
                try:
                    import cPickle
                    fd = open( filename, 'wb' )
                    cPickle.Pickler( fd, True ).dump( self )
                finally:
                    if fd is not None:
                        fd.close()
            return rc

        return True

    #---------------------------------------------------------------------------
    #  Return the list of editable traits:
    #---------------------------------------------------------------------------

    def editable_traits ( self ):
        """Returns an alphabetically sorted list of the names of non-event
        trait attributes associated with the current object.
        """
        names = self.trait_names( type = not_event, editable = not_false )
        names.sort()
        return names

    def class_editable_traits ( cls ):
        """Returns an alphabetically sorted list of the names of non-event
        trait attributes associated with the current class.
        """
        names = cls.class_trait_names( type = not_event, editable = not_false )
        names.sort()
        return names

    class_editable_traits = classmethod( class_editable_traits )

    #---------------------------------------------------------------------------
    #  Pretty print the traits of an object:
    #---------------------------------------------------------------------------

    def print_traits ( self, show_help = False, **metadata ):
        """Prints the values of all explicitly-defined, non-event trait
        attributes on the current object, in an easily readable format.

        Parameters
        ----------
        show_help : bool
            Indicates whether to display additional descriptive information.
        """

        if len( metadata ) > 0:
            names = self.trait_names( **metadata )
        else:
            names = self.trait_names( type = not_event )

        if len( names ) == 0:
            print ''
            return

        result = []
        pad    = max( [ len( x ) for x in names ] ) + 1
        maxval = 78 - pad
        names.sort()

        for name in names:
            try:
                value = repr( getattr( self, name ) ).replace( '\n', '\\n' )
                if len( value ) > maxval:
                    value = '%s...%s' % ( value[: (maxval - 2) / 2 ],
                                          value[ -((maxval - 3) / 2): ] )
            except:
                value = '<undefined>'
            lname = (name + ':').ljust( pad )
            if show_help:
                result.append( '%s %s\n   The value must be %s.' % (
                       lname, value, self.base_trait( name ).setter.info() ) )
            else:
                result.append( '%s %s' % ( lname, value ) )

        print '\n'.join( result )

    #---------------------------------------------------------------------------
    #  Add/Remove a handler for a specified trait being changed:
    #
    #  If no name is specified, the handler will be invoked for any trait
    #  change.
    #---------------------------------------------------------------------------

    def _on_trait_change ( self, handler, name = None, remove = False,
                                 dispatch = 'same', priority = False,
                                 target = None):
        """Causes the object to invoke a handler whenever a trait attribute
        is modified, or removes the association.

        Parameters
        ----------
        handler : function
            A trait notification function for the attribute specified by *name*.
        name : str
            Specifies the trait attribute whose value changes trigger the
            notification.
        remove : bool
            If True, removes the previously-set association between
            *handler* and *name*; if False (the default), creates the
            association.

        Description
        -----------
        Multiple handlers can be defined for the same object, or even for the
        same trait attribute on the same object. If *name* is not specified or
        is None, *handler* is invoked when any trait attribute on the
        object is changed.
        """

        if type( name ) is list:
            for name_i in name:
                self._on_trait_change( handler, name_i, remove, dispatch,
                                       priority, target )

            return

        name = name or 'anytrait'

        if remove:
            if name == 'anytrait':
                notifiers = self._notifiers( 0 )
            else:
                trait = self._trait( name, 1 )
                if trait is None:
                    return
                notifiers = trait._notifiers( 0 )

            if notifiers is not None:
                for i, notifier in enumerate( notifiers ):
                    if notifier.equals( handler ):
                        del notifiers[i]
                        notifier.dispose()
                        break

            return

        if name == 'anytrait':
            notifiers = self._notifiers( 1 )
        else:
            notifiers = self._trait( name, 2 )._notifiers( 1 )

        for notifier in notifiers:
            if notifier.equals( handler ):
                break
        else:
            wrapper = self.wrappers[ dispatch ]( handler, notifiers, target )

            if priority:
                notifiers.insert( 0, wrapper )
            else:
                notifiers.append( wrapper )

    #---------------------------------------------------------------------------
    #  Add/Remove handlers for an extended set of one or more traits being
    #  changed:
    #
    #  If no name is specified, the handler will be invoked for any trait
    #  change.
    #---------------------------------------------------------------------------

    def on_trait_change ( self, handler, name = None, remove = False,
                                dispatch = 'same', priority = False,
                                deferred = False, target = None ):
        """Causes the object to invoke a handler whenever a trait attribute
        matching a specified pattern is modified, or removes the association.

        Parameters
        ----------
        handler : function
            A trait notification function for the *name* trait attribute, with
            one of the signatures described below.
        name : str
            The name of the trait attribute whose value changes trigger the
            notification. The *name* can specify complex patterns of trait
            changes using an extended *name* syntax, which is described below.
        remove : bool
            If True, removes the previously-set association between
            *handler* and *name*; if False (the default), creates the
            association.
        dispatch : str
            A string indicating the thread on which notifications must be run.
            Possible values are:

            =========== =======================================================
            value       dispatch
            =========== =======================================================
            ``same``    Run notifications on the same thread as this one.
            ``ui``      Run notifications on the UI thread. If the current
                        thread is the UI thread, the notifications are executed
                        immediately; otherwise, they are placed on the UI
                        event queue.
            ``fast_ui`` Alias for ``ui``.
            ``new``     Run notifications in a new thread.
            =========== =======================================================

        Description
        -----------
        Multiple handlers can be defined for the same object, or even for the
        same trait attribute on the same object. If *name* is not specified or
        is None, *handler* is invoked when any trait attribute on the
        object is changed.

        The *name* parameter is a single *xname* or a list of *xname* names,
        where an *xname* is an extended name of the form::

            xname2[('.'|':') xname2]*

        An *xname2* is of the form::

            ( xname3 | '['xname3[','xname3]*']' ) ['*']

        An *xname3* is of the form::

             xname | ['+'|'-'][name] | name['?' | ('+'|'-')[name]]

        A *name* is any valid Python attribute name. The semantic meaning of
        this notation is as follows:

        ================================ ======================================
        expression                       meaning
        ================================ ======================================
        ``item1.item2``                  means *item1* is a trait containing an
                                         object (or objects if *item1* is a
                                         list or dict) with a trait called
                                         *item2*. Changes to either *item1* or
                                         *item2* cause a notification to be
                                         generated.
        ``item1:item2``                  means *item1* is a trait containing an
                                         object (or objects if *item1* is a
                                         list or dict) with a trait called
                                         *item2*. Changes to *item2* cause a
                                         notification to be generated, while
                                         changes to *item1* do not (i.e., the
                                         ':' indicates that changes to the
                                         *link* object should not be reported).
        ``[ item1, item2, ..., itemN ]`` A list which matches any of the
                                         specified items. Note that at the
                                         topmost level, the surrounding square
                                         brackets are optional.
        ``name?``                        If the current object does not have an
                                         attribute called *name*, the reference
                                         can be ignored. If the '?' character
                                         is omitted, the current object must
                                         have a trait called *name*, otherwise
                                         an exception will be raised.
        ``prefix+``                      Matches any trait on the object whose
                                         name begins with *prefix*.
        ``+metadata_name``               Matches any trait on the object having
                                         *metadata_name* metadata.
        ``-metadata_name``               Matches any trait on the object which
                                         does not have *metadata_name*
                                         metadata.
        ``prefix+metadata_name``         Matches any trait on the object whose
                                         name begins with *prefix* and which
                                         has *metadata_name* metadata.
        ``prefix-metadata_name``         Matches any trait on the object
                                         whose name begins with *prefix* and
                                         which does not have *metadata_name*
                                         metadata.
        ``+``                            Matches all traits on the object.
        ``pattern*``                     Matches object graphs where *pattern*
                                         occurs one or more times (useful for
                                         setting up listeners on recursive data
                                         structures like trees or linked
                                         lists).
        ================================ ======================================

        Some examples of valid names and their meaning are as follows:

        ======================= ===============================================
        example                 meaning
        ======================= ===============================================
        ``foo,bar,baz``         Listen for trait changes to *object.foo*,
                                *object.bar*, and *object.baz*.
        ``['foo','bar','baz']`` Equivalent to 'foo,bar,baz', but may be more
                                useful in cases where the individual items are
                                computed.
        ``foo.bar.baz``         Listen for trait changes to
                                *object.foo.bar.baz* and report changes to
                                *object.foo*, *object.foo.bar* or
                                *object.foo.bar.baz*.
        ``foo:bar:baz``         Listen for changes to *object.foo.bar.baz*, and
                                only report changes to *object.foo.bar.baz*.
        ``foo.[bar,baz]``       Listen for trait changes to *object.foo.bar*
                                and *object.foo.baz*.
        ``[left,right]*.name``  Listen for trait changes to the *name* trait of
                                each node of a tree having *left* and *right*
                                links to other tree nodes, and where *object*
                                the method is applied to the root node of the
                                tree.
        ``+dirty``              Listen for trait changes on any trait in the
                                *object* which has the 'dirty' metadata set.
        ``foo.+dirty``          Listen for trait changes on any trait in
                                *object.foo* which has the 'dirty' metadata
                                set.
        ``foo.[bar,-dirty]``    Listen for trait changes on *object.foo.bar* or
                                any trait on *object.foo* which does not have
                                'dirty' metadata set.
        ======================= ===============================================


        Note that any of the intermediate (i.e., non-final) links in a
        pattern can be traits of type Instance, List or Dict. In the case
        of List and Dict traits, the subsequent portion of the pattern is
        applied to each item in the list, or value in the dictionary.

        For example, if the self.children is a list, 'children.name'
        listens for trait changes to the *name* trait for each item in the
        self.children list.

        Note that items added to or removed from a list or dictionary in
        the pattern will cause the *handler* routine to be invoked as well,
        since this is treated as an *implied* change to the item's trait
        being monitored.

        The signature of the *handler* supplied also has an effect on
        how changes to intermediate traits are processed. The five valid
        handler signatures are:

        1. handler()
        2. handler(new)
        3. handler(name,new)
        4. handler(object,name,new)
        5. handler(object,name,old,new)

        For signatures 1, 4 and 5, any change to any element of a path
        being listened to invokes the handler with information about the
        particular element that was modified (e.g., if the item being
        monitored is 'foo.bar.baz', a change to 'bar' will call *handler*
        with the following information:

        - object: object.foo
        - name:   bar
        - old:    old value for object.foo.bar
        - new:    new value for object.foo.bar

        If one of the intermediate links is a List or Dict, the call to
        *handler* may report an *_items* changed event. If in the previous
        example, *bar* is a List, and a new item is added to *bar*, then
        the information passed to *handler* would be:

        - object: object.foo
        - name:   bar_items
        - old:    Undefined
        - new:    TraitListEvent whose *added* trait contains the new item
                  added to *bar*.

        For signatures 2 and 3, the *handler* does not receive enough
        information to discern between a change to the final trait being
        listened to and a change to an intermediate link. In this case,
        the event dispatcher will attempt to map a change to an
        intermediate link to its effective change on the final trait. This
        only works if all of the intermediate links are single values (such
        as an Instance or Any trait) and not Lists or Dicts. If the modified
        intermediate trait or any subsequent intermediate trait preceding
        the final trait is a List or Dict, then a TraitError is raised,
        since the effective value for the final trait cannot in general be
        resolved unambiguously. To prevent TraitErrors in this case, use the
        ':' separator to suppress notifications for changes to any of the
        intermediate links.

        Handler signature 1 also has the special characteristic that if a
        final trait is a List or Dict, it will automatically handle '_items'
        changed events for the final trait as well. This can be useful in
        cases where the *handler* only needs to know that some aspect of the
        final trait has been changed. For all other *handler* signatures,
        you must explicitly specify the 'xxx_items' trait if you want to
        be notified of changes to any of the items of the 'xxx' trait.

        """
        # Check to see if we can do a quick exit to the basic trait change
        # handler:
        if ((isinstance( name, basestring ) and
            (extended_trait_pat.match( name ) is None)) or (name is None)):
            self._on_trait_change( handler, name, remove, dispatch, priority, target )

            return

        from .traits_listener \
            import TraitsListener, ListenerParser, ListenerHandler, \
                   ListenerNotifyWrapper

        if isinstance( name, list ):
            for name_i in name:
                self.on_trait_change( handler, name_i, remove, dispatch,
                                      priority, target )

            return

        # Make sure we have a name string:
        name = (name or 'anytrait').strip()

        if remove:
            dict = self.__dict__.get( TraitsListener )
            if dict is not None:
                listeners = dict.get( name )
                if listeners is not None:
                    for i, wrapper in enumerate( listeners ):
                        if wrapper.equals( handler ):
                            del listeners[i]
                            if len( listeners ) == 0:
                                del dict[ name ]
                                if len( dict ) == 0:
                                    del self.__dict__[ TraitsListener ]
                            wrapper.listener.unregister( self )
                            wrapper.dispose()
                            break
        else:
            dict      = self.__dict__.setdefault( TraitsListener, {} )
            listeners = dict.setdefault( name, [] )
            for wrapper in listeners:
                if wrapper.equals( handler ):
                    break
            else:
                listener = ListenerParser( name ).listener
                lnw = ListenerNotifyWrapper( handler, self, name, listener, target )
                listeners.append( lnw )
                listener.set( handler         = ListenerHandler( handler ),
                              wrapped_handler_ref = weakref.ref(lnw),
                              type            = lnw.type,
                              dispatch        = dispatch,
                              priority        = priority,
                              deferred        = deferred )
                listener.register( self )

    # A synonym for 'on_trait_change'
    on_trait_event = on_trait_change

    #---------------------------------------------------------------------------
    #  Synchronize the value of two traits:
    #---------------------------------------------------------------------------

    def sync_trait ( self, trait_name, object, alias = None, mutual = True,
                           remove = False ):
        """Synchronizes the value of a trait attribute on this object with a
        trait attribute on another object.

        Parameters
        ----------
        name : str
            Name of the trait attribute on this object.
        object : object
            The object with which to synchronize.
        alias : str
            Name of the trait attribute on *other*; if None or omitted, same
            as *name*.
        mutual : bool or int
            Indicates whether synchronization is mutual (True or non-zero)
            or one-way (False or zero)
        remove : bool or int
            Indicates whether synchronization is being added (False or zero)
            or removed (True or non-zero)

        Description
        -----------
        In mutual synchronization, any change to the value of the specified
        trait attribute of either object results in the same value being
        assigned to the corresponding trait attribute of the other object.
        In one-way synchronization, any change to the value of the attribute
        on this object causes the corresponding trait attribute of *object* to
        be updated, but not vice versa.
        """
        if alias is None:
            alias = trait_name

        is_list = (self._is_list_trait( trait_name ) and
                   object._is_list_trait( alias ))

        if remove:
            info = self._get_sync_trait_info()
            dic  = info.get( trait_name )
            if dic is not None:
                key = ( id( object ), alias )
                if key in dic:
                    del dic[ key ]

                    if len( dic ) == 0:
                        del info[ trait_name ]
                        self._on_trait_change( self._sync_trait_modified,
                            trait_name, remove = True )

                        if is_list:
                            self._on_trait_change(
                                self._sync_trait_items_modified,
                                trait_name + '_items', remove = True )

            if mutual:
                object.sync_trait( alias, self, trait_name, False, True )

            return

        value = ( weakref.ref( object, self._sync_trait_listener_deleted ),
                  alias )
        dic   = self._get_sync_trait_info().setdefault( trait_name, {} )
        key   = ( id( object ), alias )
        if key not in dic:
            if len( dic ) == 0:
                self._on_trait_change( self._sync_trait_modified, trait_name )
                if is_list:
                    self._on_trait_change( self._sync_trait_items_modified,
                                           trait_name + '_items' )
            dic[ key ] = value
            setattr( object, alias, getattr( self, trait_name ) )

        if mutual:
            object.sync_trait( alias, self, trait_name, False )

    def _get_sync_trait_info ( self ):
        info = getattr( self, '__sync_trait__', None )
        if info is None:
            self.__dict__[ '__sync_trait__' ] = info = {}
            info[ '' ] = {}

        return info

    def _sync_trait_modified ( self, object, name, old, new ):
        info   = self.__sync_trait__
        locked = info[ '' ]
        locked[ name ] = None
        for object, object_name in info[ name ].values():
            object = object()
            if object_name not in object._get_sync_trait_info()[ '' ]:
                try:
                    setattr( object, object_name, new )
                except:
                    pass

        del locked[ name ]

    def _sync_trait_items_modified ( self, object, name, old, event ):
        n0     = event.index
        n1     = n0 + len( event.removed )
        name   = name[:-6]
        info   = self.__sync_trait__
        locked = info[ '' ]
        locked[ name ] = None
        for object, object_name in info[ name ].values():
            object = object()
            if object_name not in object._get_sync_trait_info()[ '' ]:
                try:
                    getattr( object, object_name )[ n0: n1 ] = event.added
                except:
                    pass

        del locked[ name ]

    def _sync_trait_listener_deleted ( self, ref ):
        info = self.__sync_trait__
        for key, dic in info.items():
            if key != '':
                for name, value in dic.items():
                    if ref is value[0]:
                        del dic[ name ]
                        if len( dic ) == 0:
                            del info[ key ]

    def _is_list_trait ( self, trait_name ):
        handler = self.base_trait( trait_name ).handler

        return ((handler is not None) and (handler.default_value_type == 5))

    #---------------------------------------------------------------------------
    #  Add a new trait:
    #---------------------------------------------------------------------------

    def add_trait ( self, name, *trait ):
        """Adds a trait attribute to this object.

        Parameters
        ----------
        name : str
            Name of the attribute to add.
        *trait :
            Trait or a value that can be converted to a trait by Trait().
            Trait definition for *name*. If more than one value is specified,
            it is equivalent to passing the entire list of values to Trait().

        """

        # Make sure a trait argument was specified:
        if len( trait ) == 0:
            raise ValueError, 'No trait definition was specified.'

        # Make sure only valid traits get added:
        if len( trait ) > 1:
            trait = Trait( *trait )
        else:
            trait = _trait_for( trait[0] )

        # Check to see if the trait has additional sub-traits that need to be
        # defined also:
        handler = trait.handler
        if handler is not None:
            if handler.has_items:
                self.add_trait( name + '_items', handler.items_event() )
            if handler.is_mapped:
                self.add_trait( name + '_', _mapped_trait_for( trait ) )

        # See if there already is a class or instance trait with the same name:
        old_trait = self._trait( name, 0 )

        # Get the object's instance trait dictionary and add a clone of the new
        # trait to it:
        itrait_dict = self._instance_traits()
        itrait_dict[ name ] = trait = _clone_trait( trait )

        # If there already was a trait with the same name:
        if old_trait is not None:
            # Copy the old traits notifiers into the new trait:
            old_notifiers = old_trait._notifiers( 0 )
            if old_notifiers is not None:
                trait._notifiers( 1 ).extend( old_notifiers )
        else:
            # Otherwise, see if there are any static notifiers that should be
            # applied to the trait:
            cls      = self.__class__
            handlers = [ _get_method( cls, '_%s_changed' % name ),
                         _get_method( cls, '_%s_fired'   % name ) ]

            # Add any special trait defined event handlers:
            _add_event_handlers( trait, cls, handlers )

            # Add the 'anytrait' handler (if any):
            handlers.append( self.__prefix_traits__.get( '@' ) )

            # Filter out any 'None' values:
            handlers = [ h for h in handlers if h is not None ]

            # If there are any static notifiers, attach them to the trait:
            if len( handlers ) > 0:
                _add_notifiers( trait._notifiers( 1 ), handlers )

        # If this was a new trait, fire the 'trait_added' event:
        if old_trait is None:
            self.trait_added = name

    #---------------------------------------------------------------------------
    #  Remove an existing trait:
    #---------------------------------------------------------------------------

    def remove_trait ( self, name ):
        """Removes a trait attribute from this object.

        Parameters
        ----------
        name : str
            Name of the attribute to remove.
        """
        # Get the trait definition:
        trait = self._trait( name, 0 )
        if trait is not None:

            # Check to see if the trait has additional sub-traits that need to
            # be removed also:
            handler = trait.handler
            if handler is not None:
                if handler.has_items:
                    self.remove_trait( name + '_items' )
                if handler.is_mapped:
                    self.remove_trait( name + '_' )

            # Remove the trait value from the object dictionary as well:
            if name in self.__dict__:
                del self.__dict__[ name ]

            # Get the object's instance trait dictionary and remove the trait
            # from it:
            itrait_dict = self._instance_traits()
            if name in itrait_dict:
                del itrait_dict[ name ]
                return True

        return False

    #---------------------------------------------------------------------------
    #  Returns the trait definition of a specified trait:
    #---------------------------------------------------------------------------

    def trait ( self, name, force = False, copy = False ):
        """Returns the trait definition for the *name* trait attribute.

        Parameters
        ----------
        name : str
            Name of the attribute whose trait definition is to be returned.
        force : bool
            Indicates whether to return a trait definition if *name* is
            not explicitly defined.
        copy : bool
            Indicates whether to return the original trait definition or a
            copy.

        Description
        -----------
        If *force* is False (the default) and *name* is the name of an
        implicitly defined trait attribute that has never been referenced
        explicitly (i.e., has not yet been defined), the result is None. In
        all other cases, the result is the trait definition object associated
        with *name*.

        If *copy* is True, and a valid trait definition is found for *name*,
        a copy of the trait found is returned. In all other cases, the trait
        definition found is returned unmodified (the default).
        """
        mode = 0
        if force:
            mode = -1
        result = self._trait( name, mode )
        if (not copy) or (result is None):
            return result

        return  _clone_trait( result )

    #---------------------------------------------------------------------------
    #  Returns the base trait definition of a specified trait:
    #---------------------------------------------------------------------------

    def base_trait ( self, name ):
        """Returns the base trait definition for a trait attribute.

        Parameters
        ----------
        name : str
            Name of the attribute whose trait definition is returned.

        Description
        -----------
        This method is similar to the trait() method, and returns a
        different result only in the case where the trait attribute defined by
        *name* is a delegate. In this case, the base_trait() method follows the
        delegation chain until a non-delegated trait attribute is reached, and
        returns the definition of that attribute's trait as the result.
        """
        return self._trait( name, -2 )

    #---------------------------------------------------------------------------
    #  Validates whether or not a specified value is legal for a specified
    # trait and returns the validated value if valid:
    #---------------------------------------------------------------------------

    def validate_trait ( self, name, value ):
        """ Validates whether a value is legal for a trait.

        Returns the validated value if it is valid.
        """
        return self.base_trait( name ).validate( self, name, value )

    #---------------------------------------------------------------------------
    #  Return a dictionary of all traits which match a set of metadata:
    #---------------------------------------------------------------------------

    def traits ( self, **metadata ):
        """Returns a dictionary containing the definitions of all of the trait
        attributes of this object that match the set of *metadata* criteria.

        Parameters
        ----------
        **metadata :
            Criteria for selecting trait attributes.

        Description
        -----------
        The keys of the returned dictionary are the trait attribute names, and
        the values are their corresponding trait definition objects.

        If no *metadata* information is specified, then all explicitly defined
        trait attributes defined for the object are returned.

        Otherwise, the *metadata* keyword dictionary is assumed to define a set
        of search criteria for selecting trait attributes of interest. The
        *metadata* dictionary keys correspond to the names of trait metadata
        attributes to examine, and the values correspond to the values the
        metadata attribute must have in order to be included in the search
        results.

        The *metadata* values either may be simple Python values like strings or
        integers, or may be lambda expressions or functions that return True
        if the trait attribute is to be included in the result. A lambda
        expression or function must receive a single argument, which is the
        value of the trait metadata attribute being tested. If more than one
        metadata keyword is specified, a trait attribute must match the metadata
        values of all keywords to be included in the result.
        """
        traits = self.__base_traits__.copy()
        for name in self.__dict__.keys():
            if name not in traits:
                trait = self.trait( name )
                if trait is not None:
                    traits[ name ] = trait

        if len( metadata ) == 0:
            return traits

        for meta_name, meta_eval in metadata.items():
            if type( meta_eval ) is not FunctionType:
                metadata[ meta_name ] = _SimpleTest( meta_eval )

        result = {}
        for name, trait in traits.items():
            for meta_name, meta_eval in metadata.items():
                if not meta_eval( getattr( trait, meta_name ) ):
                    break
            else:
                result[ name ] = trait

        return result

    #---------------------------------------------------------------------------
    #  Return a dictionary of all traits which match a set of metadata:
    #---------------------------------------------------------------------------

    def class_traits ( cls, **metadata ):
        """Returns a dictionary containing the definitions of all of the trait
        attributes of the class that match the set of *metadata* criteria.

        Parameters
        ----------
        **metadata :
            Criteria for selecting trait attributes.

        Description
        -----------
        The keys of the returned dictionary are the trait attribute names, and
        the values are their corresponding trait definition objects.

        If no *metadata* information is specified, then all explicitly defined
        trait attributes defined for the class are returned.

        Otherwise, the *metadata* keyword dictionary is assumed to define a set
        of search criteria for selecting trait attributes of interest. The
        *metadata* dictionary keys correspond to the names of trait metadata
        attributes to examine, and the values correspond to the values the
        metadata attribute must have in order to be included in the search
        results.

        The *metadata* values either may be simple Python values like strings or
        integers, or may be lambda expressions or functions that return **True**
        if the trait attribute is to be included in the result. A lambda
        expression or function must receive a single argument, which is the
        value of the trait metadata attribute being tested. If more than one
        metadata keyword is specified, a trait attribute must match the metadata
        values of all keywords to be included in the result.
        """
        if len( metadata ) == 0:
            return cls.__base_traits__.copy()

        result = {}

        for meta_name, meta_eval in metadata.items():
            if type( meta_eval ) is not FunctionType:
                metadata[ meta_name ] = _SimpleTest( meta_eval )

        for name, trait in cls.__base_traits__.items():
            for meta_name, meta_eval in metadata.items():
                if not meta_eval( getattr( trait, meta_name ) ):
                    break
            else:
                result[ name ] = trait

        return result

    class_traits = classmethod( class_traits )

    #---------------------------------------------------------------------------
    #  Return a list of all trait names which match a set of metadata:
    #---------------------------------------------------------------------------

    def trait_names ( self, **metadata ):
        """Returns a list of the names of all trait attributes whose definitions
        match the set of *metadata* criteria specified.

        Parameters
        ----------
        **metadata :
            Criteria for selecting trait attributes.

        Description
        -----------
        This method is similar to the traits() method, but returns only the
        names of the matching trait attributes, not the trait definitions.
        """
        return self.traits( **metadata ).keys()

    def class_trait_names ( cls, **metadata ):
        """Returns a list of the names of all trait attributes whose definitions
        match the set of *metadata* criteria specified.

        Parameters
        ----------
        **metadata :
            Criteria for selecting trait attributes.

        Description
        -----------
        This method is similar to the traits() method, but returns only the
        names of the matching trait attributes, not the trait definitions.
        """
        return cls.class_traits( **metadata ).keys()

    class_trait_names = classmethod( class_trait_names )

    #---------------------------------------------------------------------------
    #  Explicitly sets the value of a cached property:
    #---------------------------------------------------------------------------

    def _set_traits_cache ( self, name, value ):
        """ Explicitly sets the value of a cached property.
        """
        cached    = TraitsCache + name
        old_value = self.__dict__.get( cached, Undefined )
        self.__dict__[ cached ] = value
        if old_value != value:
            self.trait_property_changed( name, old_value, value )

    #---------------------------------------------------------------------------
    #  Explicitly flushes the value of a cached property:
    #---------------------------------------------------------------------------

    def _flush_traits_cache ( self, name, value ):
        """ Explicitly flushes the value of a cached property.
        """
        self.trait_property_changed(
            name, self.__dict__.pop( TraitsCache + name, Undefined ) )

    #---------------------------------------------------------------------------
    #  Returns the trait definition for a specified name when there is no
    #  explicit definition in the class:
    #---------------------------------------------------------------------------

    def __prefix_trait__ ( self, name, is_set ):
        # Check to see if the name is of the form '__xxx__':
        if (name[:2] == '__') and (name[-2:] == '__'):
            if name == '__class__':
                return generic_trait

            # If this is for purposes of performing a 'setattr', always map the
            # name to an 'Any' trait:
            if is_set:
                return any_trait

            # Otherwise, it is a 'getattr' request, so indicate that no such
            # attribute exists:
            raise AttributeError, "'%s' object has no attribute '%s'" % (
                                  self.__class__.__name__, name )

        # Handle the special case of 'delegated' traits:
        if name[-1:] == '_':
           trait = self._trait( name[:-1], 0 )
           if (trait is not None) and (trait.type == 'delegate'):
               return _clone_trait( trait )

        prefix_traits = self.__prefix_traits__
        for prefix in prefix_traits['*']:
            if prefix == name[ :len( prefix ) ]:
                # If we found a match, use its trait as a template for a new
                # trait:
                trait = prefix_traits[ prefix ]

                # Get any change notifiers that apply to the trait:
                cls      = self.__class__
                handlers = [ _get_method( cls, '_%s_changed' % name ),
                             _get_method( cls, '_%s_fired'   % name ) ]

                # Add any special trait defined event handlers:
                _add_event_handlers( trait, cls, handlers )

                # Add the 'anytrait' handler (if any):
                handlers.append( prefix_traits.get( '@' ) )

                # Filter out any 'None' values:
                handlers = [ h for h in handlers if h is not None ]

                # If there are any handlers, add them to the trait's notifier's
                # list:
                if len( handlers ) > 0:
                    trait = _clone_trait( trait )
                    _add_notifiers( trait._notifiers( 1 ), handlers )

                return trait

        # There should ALWAYS be a prefix match in the trait classes, since ''
        # is at the end of the list, so we should never get here:
        raise SystemError, ("Trait class look-up failed for attribute '%s' "
                            "for an object of type '%s'") % (
                            name, self.__class__.__name__ )

    #---------------------------------------------------------------------------
    #  Adds/Removes (Java-style) event listeners to an object:
    #---------------------------------------------------------------------------

    def add_trait_listener ( self, object, prefix = '' ):
        self._trait_listener( object, prefix, False )

    def remove_trait_listener ( self, object, prefix = '' ):
        self._trait_listener( object, prefix, True )

    def _trait_listener ( self, object, prefix, remove ):
        if prefix[-1:] != '_':
            prefix += '_'
        n      = len( prefix )
        traits = self.__base_traits__
        for name in self._each_trait_method( object ):
            if name[:n] == prefix:
                if name[-8:] == '_changed':
                    short_name = name[n:-8]
                    if short_name in traits:
                        self._on_trait_change( getattr( object, name ),
                                               short_name, remove = remove )
                    elif short_name == 'anytrait':
                        self._on_trait_change( getattr( object, name ),
                                               remove = remove )
                elif name[:-6] == '_fired':
                    short_name = name[n:-6]
                    if short_name in traits:
                        self._on_trait_change( getattr( object, name ),
                                               short_name, remove = remove )
                    elif short_name == 'anytrait':
                        self._on_trait_change( getattr( object, name ),
                                               remove = remove )

    #---------------------------------------------------------------------------
    #  Generates each (name, method) pair for a specified object:
    #---------------------------------------------------------------------------

    def _each_trait_method ( self, object ):
        """ Generates each (name, method) pair for a specified object.
        """
        dic = {}
        for klass in object.__class__.__mro__:
            for name, method in klass.__dict__.items():
                if (type( method ) is FunctionType) and (name not in dic):
                    dic[ name ] = True
                    yield name

    #---------------------------------------------------------------------------
    #  Handles adding/removing listeners for a generic 'Instance' trait:
    #---------------------------------------------------------------------------

    def _instance_changed_handler ( self, name, old, new ):
        """ Handles adding/removing listeners for a generic 'Instance' trait.
        """
        arg_lists = self._get_instance_handlers( name )

        if old is not None:
            for args in arg_lists:
                old.on_trait_change( remove = True, *args )

        if new is not None:
            for args in arg_lists:
                new.on_trait_change( *args )

    #---------------------------------------------------------------------------
    #  Handles adding/removing listeners for a generic 'List( Instance )' trait:
    #---------------------------------------------------------------------------

    def _list_changed_handler ( self, name, old, new ):
        """ Handles adding/removing listeners for a generic 'List( Instance )'
            trait.
        """
        arg_lists = self._get_instance_handlers( name )

        for item in old:
            for args in arg_lists:
                item.on_trait_change( remove = True, *args )

        for item in new:
            for args in arg_lists:
                item.on_trait_change( *args )

    def _list_items_changed_handler ( self, name, not_used, event ):
        """ Handles adding/removing listeners for a generic 'List( Instance )'
            trait.
        """
        arg_lists = self._get_instance_handlers( name[:-6] )

        for item in event.removed:
            for args in arg_lists:
                item.on_trait_change( remove = True, *args )

        for item in event.added:
            for args in arg_lists:
                item.on_trait_change( *args )

    #---------------------------------------------------------------------------
    #  Returns a list of ( name, method ) pairs for a specified 'Instance' or
    #  'List( Instance )' trait name:
    #---------------------------------------------------------------------------

    def _get_instance_handlers ( self, name ):
        """ Returns a list of ( name, method ) pairs for a specified 'Instance'
            or 'List( Instance )' trait name:
        """
        return [ ( getattr( self, method_name ), item_name )
                 for method_name, item_name in
                     self.__class__.__instance_traits__[ name ] ]

    #---------------------------------------------------------------------------
    #  Initializes the object's statically parsed, but dynamically registered,
    #  traits listeners (called at object creation and unpickling times):
    #---------------------------------------------------------------------------

    def _post_init_trait_listeners ( self ):
        """ Initializes the object's statically parsed, but dynamically
            registered, traits listeners (called at object creation and
            unpickling times).
        """
        for name, data in self.__class__.__listener_traits__.items():
            if data[0] == 'method':
                pattern = data[1]
                if pattern[:1] == '>':
                    self.on_trait_change( getattr( self, name ), pattern[1:],
                                          deferred = True )

    def _init_trait_listeners ( self ):
        """ Initializes the object's statically parsed, but dynamically
            registered, traits listeners (called at object creation and
            unpickling times).
        """
        for name, data in self.__class__.__listener_traits__.items():
            getattr( self, '_init_trait_%s_listener' % data[0] )( name, *data )

    def _init_trait_method_listener ( self, name, kind, pattern ):
        """ Sets up the listener for a method with the @on_trait_change
            decorator.
        """
        if pattern[:1] == '<':
            self.on_trait_change( getattr( self, name ), pattern[1:],
                                  deferred = True )

    def _init_trait_event_listener ( self, name, kind, pattern ):
        """ Sets up the listener for an event with on_trait_change metadata.
        """
        @weak_arg(self)
        def notify ( self ):
            setattr( self, name, True )

        self.on_trait_change( notify, pattern, target=self )

    def _init_trait_property_listener ( self, name, kind, cached, pattern ):
        """ Sets up the listener for a property with 'depends_on' metadata.
        """
        if cached is None:
            @weak_arg(self)
            def notify ( self ):
                self.trait_property_changed( name, None )
        else:
            cached_old = cached + ':old'
            @weak_arg(self)
            def pre_notify ( self ):
                dict = self.__dict__
                old  = dict.get( cached_old, Undefined )
                if old is Undefined:
                    dict[ cached_old ] = dict.pop( cached, None )
            self.on_trait_change( pre_notify, pattern, priority = True, target=self )

            @weak_arg(self)
            def notify ( self ):
                old = self.__dict__.pop( cached_old, Undefined )
                if old is not Undefined:
                    self.trait_property_changed( name, old )

        self.on_trait_change( notify, pattern, target=self )

    def _init_trait_delegate_listener ( self, name, kind, pattern ):
        """ Sets up the listener for a delegate trait.
        """
        name_pattern    = self._trait_delegate_name( name, pattern )
        target_name_len = len( name_pattern.split( ':' )[-1] )

        @weak_arg(self)
        def notify ( self, object, notify_name, old, new ):
            self.trait_property_changed( name + notify_name[ target_name_len: ],
                                         old, new )

        self.on_trait_change( notify, name_pattern, target=self )
        self.__dict__.setdefault( ListenerTraits, {} )[ name ] = notify

    def _remove_trait_delegate_listener ( self, name, remove ):
        """ Removes a delegate listener when the local delegate value is set.
        """
        dict = self.__dict__.setdefault( ListenerTraits, {} )

        if remove:
            # Although the name should be in the dict, it may not be if a value
            # was assigned to a delegate in a constructor or setstate:
            if name in dict:
                # Remove the delegate listener:
                self.on_trait_change( dict[ name ], self._trait_delegate_name(
                         name, self.__class__.__listener_traits__[ name ][1] ),
                         remove = True )
                del dict[ name ]
                if len( dict ) == 0:
                    del self.__dict__[ ListenerTraits ]

            return

        # Otherwise the local copy of the delegate value was deleted, restore
        # the delegate listener (unless it's already there):
        if name not in dict:
            self._init_trait_delegate_listener(
                     name, 0, self.__class__.__listener_traits__[ name ][1] )

    def _trait_delegate_name ( self, name, pattern ):
        """ Returns the fully-formed 'on_trait_change' name for a specified
            delegate.
        """
        if pattern[-1] == '*':
            pattern = '%s%s%s' % ( pattern[:-1], self.__class__.__prefix__,
                                    name )

        return pattern

# Patch the definition of _HasTraits to be the real 'HasTraits':
_HasTraits = HasTraits

#-------------------------------------------------------------------------------
#  'HasStrictTraits' class:
#-------------------------------------------------------------------------------

class HasStrictTraits ( HasTraits ):
    """ This class guarantees that any object attribute that does not have an
    explicit or wildcard trait definition results in an exception.

    This feature can be useful in cases where a more rigorous software
    engineering approach is being used than is typical for Python programs. It
    also helps prevent typos and spelling mistakes in attribute names from
    going unnoticed; a misspelled attribute name typically causes an exception.
    """
    _ = Disallow   # Disallow access to any traits not explicitly defined

#-------------------------------------------------------------------------------
#  'HasPrivateTraits' class:
#-------------------------------------------------------------------------------

class HasPrivateTraits ( HasTraits ):
    """ This class ensures that any public object attribute that does not have
    an explicit or wildcard trait definition results in an exception, but
    "private" attributes (whose names start with '_') have an initial value of
    **None**, and are not type-checked.

    This feature is useful in cases where a class needs private attributes to
    keep track of its internal object state, which are not part of the class's
    public API. Such attributes do not need to be type-checked, because they are
    manipulated only by the (presumably correct) methods of the class itself.
    """
    # Make 'private' traits (leading '_') have no type checking:
    __ = Any( private = True, transient = True )

    # Disallow access to all other traits not explicitly defined:
    _  = Disallow


#------------------------------------------------------------------------------
# ABC classes with traits: (where available)
#------------------------------------------------------------------------------
try:

    import abc


    class ABCMetaHasTraits(abc.ABCMeta, MetaHasTraits):
        """ A MetaHasTraits subclass which also inherits from
        abc.ABCMeta.

        .. note:: The ABCMeta class is cooperative and behaves nicely
            with MetaHasTraits, provided it is inherited first.
        """
        pass


    class ABCHasTraits(HasTraits):
        """ A HasTraits subclass which enables the features of Abstract
        Base Classes (ABC). See the 'abc' module in the standard library
        for more information.

        """
        __metaclass__ = ABCMetaHasTraits


    class ABCHasStrictTraits(ABCHasTraits):
        """ A HasTraits subclass which behaves like HasStrictTraits but
        also enables the features of Abstract Base Classes (ABC). See the
        'abc' module in the standard library for more information.

        """
        _ = Disallow

except ImportError:
    pass

#-------------------------------------------------------------------------------
#  Singleton classes with traits:
#
#  This code is based on a recipe taken from:
#      http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/66531
#  Specifically, the implementation of Oren Tirosh is used.
#-------------------------------------------------------------------------------

class SingletonHasTraits ( HasTraits ):
    """ Singleton class that support trait attributes.
    """
    def __new__ ( cls, *args, **traits ):
        if not '_the_instance' in cls.__dict__:
            cls._the_instance = HasTraits.__new__( cls, *args, **traits )
        return cls._the_instance

class SingletonHasStrictTraits ( HasStrictTraits ):
    """ Singleton class that supports strict trait attributes.

        Non-trait attributes generate an exception.
    """
    def __new__ ( cls, *args, **traits ):
        return SingletonHasTraits.__new__( cls, *args, **traits )

class SingletonHasPrivateTraits ( HasPrivateTraits ):
    """ Singleton class that supports trait attributes, with private attributes
        being unchecked.
    """
    def __new__ ( cls, *args, **traits ):
        return SingletonHasTraits.__new__( cls, *args, **traits )

#-------------------------------------------------------------------------------
#  Defines a 'vetoable' request object and an associated event:
#-------------------------------------------------------------------------------

class Vetoable ( HasStrictTraits ):
    """ Defines a 'vetoable' request object and an associated event.
    """
    # Should the request be vetoed? (Can only be set to 'True')
    veto = Bool( False )

    def _veto_changed ( self, state ):
        self._trait_veto_notify( state )

VetoableEvent = Event( Vetoable )

#-------------------------------------------------------------------------------
#  'MetaInterface' class:
#-------------------------------------------------------------------------------

class MetaInterface ( ABCMetaHasTraits ):
    """ Meta class for interfaces.

    Interfaces are simple ABCs with the following features:-

    1) They cannot be instantiated (they are interfaces, not implementations!).
    2) Calling them is equivalent to calling 'adapt'.

    """

    @deprecated('use "adapt(adaptee, protocol)" instead.')
    def __call__ ( self, adaptee, default=AdaptationError ):
        """ Attempt to adapt the adaptee to this interface.

        Note that this means that (intentionally ;^) that interfaces
        cannot be instantiated!

        """

        from traits.adaptation.api import adapt

        return adapt(adaptee, self, default=default)

#-------------------------------------------------------------------------------
#  'Interface' class:
#-------------------------------------------------------------------------------

class Interface ( HasTraits ):
    """ The base class for all interfaces.
    """

    __metaclass__ = MetaInterface

#-------------------------------------------------------------------------------
#  Class decorator to declare the protocols that a class provides.
#-------------------------------------------------------------------------------

def provides( *protocols ):
    """ Class decorator to declare the protocols that a class provides.

    Parameters
    ----------
    *protocols :
        A list of protocols (Interface classes or Python ABCs) that the
        decorated class provides.

    """

    from abc import ABCMeta

    # Exit immediately if there is nothing to do.
    if len(protocols) == 0:
        return lambda klass: klass

    # Verify that each argument is a valid protocol.
    for protocol in protocols:
        if not issubclass(protocol.__metaclass__, ABCMeta):
            raise TraitError(
                "All arguments to 'provides' must be "
                "subclasses of Interface or be a Python ABC."
            )

    def wrapped_class(klass):
        for protocol in protocols:
            # We use 'type(protocol)' in case the 'protocol' implements
            # its own 'register' method that overrides the ABC method.
            type(protocol).register(protocol, klass)

        # Make sure the class does provide the protocols it claims to.
        if CHECK_INTERFACES:
            from .interface_checker import check_implements
            check_implements(klass, protocols, CHECK_INTERFACES)

        return klass

    return wrapped_class

#-------------------------------------------------------------------------------
#  Return True if the class is an Interface.
#-------------------------------------------------------------------------------

def isinterface( klass ):
    """ Return True if the class is an Interface. """

    return isinstance(klass, MetaInterface)

#-------------------------------------------------------------------------------
#  Declares the interfaces that a class implements.
#-------------------------------------------------------------------------------

def implements( *interfaces ):
    """ Declares the interfaces that a class implements.

    Parameters
    ----------
    *interfaces :
        A list of interface classes that the containing class implements.

    Description
    -----------
    Registers each specified interface with the interface manager as an
    interface that the containing class implements. Each specified interface
    must be a subclass of **Interface**. This function should only be
    called from directly within a class body.
    """

    callback = provides(*interfaces)
    callback = deprecated(
        "'the 'implements' class advisor has been deprecated. "
        "Use the 'provides' class decorator."
    )(callback)

    addClassAdvisor(callback)

#-------------------------------------------------------------------------------
#  'ISerializable' interface:
#-------------------------------------------------------------------------------

class ISerializable ( Interface ):
    """ A class that implemented ISerializable requires that all HasTraits
        objects saved as part of its state also implement ISerializable.
    """
#-------------------------------------------------------------------------------
#  'traits_super' class:
#-------------------------------------------------------------------------------

class traits_super ( super ):

    def __getattribute__ ( self, name ):
        try:
            return super( traits_super, self ).__getattribute__( name )
        except:
            return self._noop

    def _noop ( self, *args, **kw ):
        pass