Skip to main content

A Python(ic) Implementation of the Eclipse Modeling Framework (EMF/Ecore), Python 2.7 backport

Project description

pypi-version master-build coverage license

PyEcore is a Model Driven Engineering (MDE) framework written for Python 2.7 (backport of the PyEcore for Python 3 version). Precisely, it is an implementation of EMF/Ecore for Python, and it tries to give an API which is compatible with the original EMF Java implementation.

PyEcore allows you to handle models and metamodels (structured data model), and gives the key you need for building MDE-based tools and other applications based on a structured data model. It supports out-of-the-box:

  • Data inheritance,

  • Two-ways relationship management (opposite references),

  • XMI (de)serialization,

  • JSON (de)serialization,

  • Notification system,

  • Reflexive API…

Let see how to create on a very simple “dynamic” metamodel (in opposite to static ones, see the documentation for more details):

>>> from pyecore.ecore import EClass, EAttribute, EString, EObject
>>> Graph = EClass('Graph')  # We create a 'Graph' concept
>>> Node = EClass('Node')  # We create a 'Node' concept
>>> # We add a "name" attribute to the Graph concept
>>> Graph.eStructuralFeatures.append(EAttribute('name', EString,
>>> # And one on the 'Node' concept
>>> Node.eStructuralFeatures.append(EAttribute('name', EString))
>>> # We now introduce a containment relation between Graph and Node
>>> contains_nodes = EReference('nodes', Node, upper=-1, containment=True)
>>> Graph.eStructuralFeatures.append(contains_nodes)
>>> # We add an opposite relation between Graph and Node
>>> Node.eStructuralFeatures.append(EReference('owned_by', Graph, eOpposite=contains_nodes))

With this code, we have defined two concepts: Graph and Node. Both have a name, and it exists a containment relationship between them. This relation is bi-directionnal, which means that each time a Node object is added to the nodes relationship of a Graph, the owned_by relation of the Node is updated also (it also work in the other way).

Let’s create some instances of our freshly created metamodel:

>>> # We create a Graph
>>> g1 = Graph(name='Graph 1')
>>> g1
<pyecore.ecore.Graph at 0x7f0055554dd8>
>>> # And two node instances
>>> n1 = Node(name='Node 1')
>>> n2 = Node(name='Node 2')
>>> n1, n2
(<pyecore.ecore.Node at 0x7f0055550588>,
 <pyecore.ecore.Node at 0x7f00555502b0>)
>>> # We add them to the Graph
>>> g1.nodes.extend([n1, n2])
>>> g1.nodes
EOrderedSet([<pyecore.ecore.Node object at 0x7f0055550588>,
             <pyecore.ecore.Node object at 0x7f00555502b0>])
>>> # bi-directional references are updated
>>> n1.owned_by
<pyecore.ecore.Graph at 0x7f0055554dd8>

This example gives a quick overview of some of the features you get for free when using PyEcore.

The project slowly grows and it still requires more love.


PyEcore is available on pypi, you can simply install it using pip:

$ pip install pyecore

The installation can also be performed manually (better in a virtualenv):

$ python install


You can read the documentation (for Python 3), which sums up pretty much everything you can find in this page, and more:

You can also find on this page a part of the Documentation:

Dynamic Metamodels

Dynamic metamodels reflects the ability to create metamodels “on-the-fly”. You can create metaclass hierarchie, add EAttribute and EReference.

In order to create a new metaclass, you need to create an EClass instance:

>>> import pyecore.ecore as Ecore
>>> MyMetaclass = Ecore.EClass('MyMetaclass')

You can then create instances of your metaclass:

>>> instance1 = MyMetaclass()
>>> instance2 = MyMetaclass()
>>> assert instance1 is not instance2

From the created instances, we can go back to the metaclasses:

>>> instance1.eClass
<EClass name="MyMetaclass">

Then, we can add metaproperties to the freshly created metaclass:

>>> instance1.eClass.eAttributes
>>> MyMetaclass.eStructuralFeatures.append(Ecore.EAttribute('name', Ecore.EString))  # We add a 'name' which is a string
>>> instance1.eClass.eAttributes  # Is there a new feature?
[<pyecore.ecore.EAttribute object at 0x7f7da72ba940>]  # Yep, the new feature is here!
>>> str(  # There is a default value for the new attribute
>>> = 'mystuff'
>>> # As the feature exists in the metaclass, the name of the feature can be used in the constructor
>>> instance3 = MyMetaclass(name='myname')

We can also create a new metaclass B and a new meta-references towards B:

>>> B = Ecore.EClass('B')
>>> MyMetaclass.eStructuralFeatures.append(Ecore.EReference('toB', B, containment=True))
>>> b1 = B()
>>> instance1.toB = b1
>>> instance1.toB
<pyecore.ecore.B object at 0x7f7da70531d0>
>>> b1.eContainer() is instance1   # because 'toB' is a containment reference

Opposite and ‘collection’ meta-references are also managed:

>>> C = Ecore.EClass('C')
>>> C.eStructuralFeatures.append(Ecore.EReference('toMy', MyMetaclass))
>>> MyMetaclass.eStructuralFeatures.append(Ecore.EReference('toCs', C, upper=-1, eOpposite=C.eStructuralFeatures[0]))
>>> instance1.toCs
>>> c1 = C()
>>> c1.toMy = instance1
>>> instance1.toCs  # 'toCs' should contain 'c1' because 'toMy' is opposite relation of 'toCs'
[<pyecore.ecore.C object at 0x7f7da7053390>]

Explore Dynamic metamodel/model objects

It is possible, when you are handling an object in the Python console, to ask for all the meta-attributes/meta-references and meta-operations that can be called on it using dir() on, either a dynamic metamodel object or a model instance. This allows you to quickly experiment and find the information you are looking for:

>>> A = EClass('A')
>>> dir(A)
 # ... there is many others
>>> a = A()
>>> dir(a)
>>> A.eStructuralFeatures.append(EAttribute('myname', EString))
>>> dir(a)

Enhance the Dynamic metamodel

Even if you define or use a dynamic metamodel, you can add dedicated methods (e.g: __repr__) to the equivalent Python class. Each EClass instance is linked to a Python class which can be reached using the python_class field:

>>> A = EClass('A')
>>> A.python_class

You can directly add new “non-metamodel” related method to this class:

>>> a = A()
>>> a
<pyecore.ecore.A at 0x7f4f0839b7b8>
>>> A.python_class.__repr__ = lambda self: 'My repr for real'
>>> a
My repr for real

Static Metamodels

The static definition of a metamodel using PyEcore mostly relies on the classical classes definitions in Python. Each Python class is linked to an EClass instance and has a special metaclass. The static code for metamodel also provides a model layer and the ability to easily refer/navigate inside the defined meta-layer.

$ ls library

$ cat library/
# ... stuffs here
class Writer(EObject):
    __metaclass__ = MetaEClass
    name = EAttribute(eType=EString)
    books = EReference(upper=-1)

    def __init__(self, name=None, books=None, **kwargs):
        if kwargs:
            raise AttributeError('unexpected arguments: {}'.format(kwargs))

        if name is not None:
   = name
        if books:
# ... Other stuffs here

$ python
>>> import library
>>> # we can create elements and handle the model level
>>> smith = library.Writer(name='smith')
>>> book1 = library.Book(title='Ye Old Book1')
>>> book1.pages = 100
>>> smith.books.append(book1)
>>> assert book1 in smith.books
>>> assert smith in book1.authors
>>> # ...
>>> # We can also navigate the meta-level
>>> import pyecore.ecore as Ecore  # We import the Ecore metamodel only for tests
>>> assert isinstance(library.Book.authors, Ecore.EReference)
>>> library.Book.authors.upperBound
>>> assert isinstance(, Ecore.EAttribute)

There is two main ways of creating static EClass with PyEcore. The first one relies on automatic code generation while the second one uses manual definition.

The automatic code generator defines a Python package hierarchie instead of only a Python module. This allows more freedom for dedicated operations and references between packages.

How to Generate the Static Metamodel Code

The static code is generated from a .ecore where your metamodel is defined (the EMF .genmodel files are not yet supported (probably in future version).

There is currently two ways of generating the code for your metamodel. The first one is to use a MTL generator (in /generator) and the second one is to use a dedicated command line tool written in Python, using Pymultigen, Jinja and PyEcore.

Using the Accelo/MTL Generator

To use this generator, you need Eclipse and the right Acceleo plugins. Once Eclipse is installed with the right plugins, you need to create a new Acceleo project, copy the PyEcore generator in it, configure a new Acceleo runner, select your .ecore and your good to go. There is plenty of documentation over the Internet for Acceleo/MTL project creation/management…

WARNING: the Acceleo generator is now deprecated, use pyecoregen instead!

Using the Dedicated CLI Generator (PyEcoregen)

For simple generation, the Acceleo generator will still do the job, but for more complex metamodel and a more robust generation, pyecoregen is significantly better. Its use is the prefered solution for the static metamodel code generation. Advantages over the Acceleo generator are the following:

  • it gives the ability to deal with generation from the command line,

  • it gives the ability to launch generation programmatically (and very simply),

  • it introduces into PyEcore a framework for code generation,

  • it allows you to code dedicated behavior in mixin classes,

  • it can be installed from pip.

This generator source code can be found at this address with mode details: and is available on Pypi, so you can install it quite symply using:

$ pip install pyecoregen

This will automatically install all the required dependencies and give you a new CLI tool: pyecoregen.

Using this tool, your static code generation is very simple:

$ pyecoregen -e your_ecore_file.ecore -o your_output_path

The generated code is automatically formatted using autopep8. Once the code is generated, your can import it and use it in your Python code.

Manually defines static EClass

To manually defines static EClass, it is simply a matter of creating a Python class, and adding to it the @EMetaclass class decorator. This decorator will automatically add the righ metaclass to the defined class, and introduce the missing classes in it’s inheritance tree. Defining simple metaclass is thus fairly easy:

class Person(object):
    name = EAttribute(eType=EString)
    age = EAttribute(eType=EInt)
    children = EReference(upper=-1, containment=True)

    def __init__(self, name): = name

Person.children.eType = Person  # As the relation is reflexive, it must be set AFTER the metaclass creation

p1 = Person('Parent')

Without more information, all the created metaclass will be added to a default EPackage, generated on the fly. If the EPackage must be controlled, a global variable of EPackage type, named eClass, must be created in the module.

eClass = EPackage(name='pack', nsURI='http://pack/1.0', nsPrefix='pack')

class TestMeta(object):

assert TestMeta.eClass.ePackage is eClass

However, when @EMetaclass is used, the direct super() call in the __init__ constructor cannot be directly called. Instead, super(x, self) must be called:

class Stuff(object):
    def __init__(self):
        self.stuff = 10

class A(Stuff):
    def __init__(self, tmp):
        super(A, self).__init__()
        self.tmp = tmp

a = A()
assert a.stuff == 10

If you want to directly extends the PyEcore metamodel, the @EMetaclass is not required, and super() can be used.

class MyNamedElement(ENamedElement):
    def __init__(self, tmp=None, **kwargs):
        self.tmp = tmp

Static/Dynamic EOperation

PyEcore also support EOperation definition for static and dynamic metamodel. For static metamodel, the solution is simple, a simple method with the code is added inside the defined class. The corresponding EOperation is created on the fly. Theire is still some “requirements” for this. In order to be understood as an EOperation candidate, the defined method must have at least one parameter and the first parameter must always be named self.

For dynamic metamodels, the simple fact of adding an EOperation instance in the EClass instance, adds an “empty” implementation:

>>> import pyecore.ecore as Ecore
>>> A = Ecore.EClass('A')
>>> operation = Ecore.EOperation('myoperation')
>>> param1 = Ecore.EParameter('param1', eType=Ecore.EString, required=True)
>>> operation.eParameters.append(param1)
>>> A.eOperations.append(operation)
>>> a = A()
>>> help(a.myoperation)
Help on method myoperation:

myoperation(param1) method of pyecore.ecore.A instance
>>> a.myoperation('test')
NotImplementedError: Method myoperation(param1) is not yet implemented

For each EParameter, the required parameter express the fact that the parameter is required or not in the produced operation:

>>> operation2 = Ecore.EOperation('myoperation2')
>>> p1 = Ecore.EParameter('p1', eType=Ecore.EString)
>>> operation2.eParameters.append(p1)
>>> A.eOperations.append(operation2)
>>> a = A()
>>> a.operation2(p1='test')  # Will raise a NotImplementedError exception

You can then create an implementation for the eoperation and link it to the EClass:

>>> def myoperation(self, param1):
...     print(self, param1)
>>> A.python_class.myoperation = myoperation

To be able to propose a dynamic empty implementation of the operation, PyEcore relies on Python code generation at runtime.


PyEcore gives you the ability to listen to modifications performed on an element. The EObserver class provides a basic observer which can receive notifications from the EObject it is register in:

>>> import library as lib  # we use the wikipedia library example
>>> from pyecore.notification import EObserver, Kind
>>> smith = lib.Writer()
>>> b1 = lib.Book()
>>> observer = EObserver(smith, notifyChanged=lambda x: print(x))
>>> b1.authors.append(smith)  # observer receive the notification from smith because 'authors' is eOpposite or 'books'

The EObserver notification method can be set using a lambda as in the previous example, using a regular function or by class inheritance:

>>> def print_notif(notification):
...     print(notification)
>>> observer = EObserver()
>>> observer.observe(b1)
>>> observer.notifyChanged = print_notif
>>> b1.authors.append(smith)  # observer receive the notification from b1

Using inheritance:

>>> class PrintNotification(EObserver):
...     def __init__(self, notifier=None):
...         super().__init__(notifier=notifier)
...     def notifyChanged(self, notification):
...         print(notification)
>>> observer = PrintNotification(b1)
>>> b1.authors.append(smith)  # observer receive the notification from b1

The Notification object contains information about the performed modification:

  • new -> the new added value (can be a collection) or None is remove or unset

  • old -> the replaced value (always None for collections)

  • feature -> the EStructuralFeature modified

  • notifer -> the object that have been modified

  • kind -> the kind of modification performed

The different kind of notifications that can be currently received are:

  • ADD -> when an object is added to a collection

  • ADD_MANY -> when many objects are added to a collection

  • REMOVE -> when an object is removed from a collection

  • SET -> when a value is set in an attribute/reference

  • UNSET -> when a value is removed from an attribute/reference

Deep Journey Inside PyEcore

This section will provide some explanation of how PyEcore works.

EClasse Instances as Factories

The most noticeable difference between PyEcore and Java-EMF implementation is the fact that there is no factories (as you probably already seen). Each EClass instance is in itself a factory. This allows you to do this kind of tricks:

>>> A = EClass('A')
>>> eobject = A()  # We create an A instance
>>> eobject.eClass
<EClass name="A">
>>> eobject2 = eobject.eClass()  # We create another A instance
>>> assert isinstance(eobject2, eobject.__class__)
>>> from pyecore.ecore import EcoreUtils
>>> assert EcoreUtils.isinstance(eobject2, A)

In fact, each EClass instance create a new Python class named after the EClass name and keep a strong relationship towards it. Moreover, EClass implements is a callable and each time () is called on an EClass instance, an instance of the associated Python class is created. Here is a small example:

>>> MyClass = EClass('MyClass')  # We create an EClass instance
>>> type(MyClass)
>>> MyClass.python_class
>>> myclass_instance = MyClass()  # MyClass is callable, creates an instance of the 'python_class' class
>>> myclass_instance
<pyecore.ecore.MyClass at 0x7f64b697df98>
>>> type(myclass_instance)
# We can access the EClass instance from the created instance and go back
>>> myclass_instance.eClass
<EClass name="MyClass">
>>> assert myclass_instance.eClass.python_class is MyClass.python_class
>>> assert myclass_instance.eClass.python_class.eClass is MyClass
>>> assert myclass_instance.__class__ is MyClass.python_class
>>> assert myclass_instance.__class__.eClass is MyClass
>>> assert myclass_instance.__class__.eClass is myclass_instance.eClass

The Python class hierarchie (inheritance tree) associated to the EClass instance

>>> B = EClass('B')  # in complement, we create a new B metaclass
>>> list(B.eAllSuperTypes())
>>> B.eSuperTypes.append(A)  # B inherits from A
>>> list(B.eAllSuperTypes())
{<EClass name="A">}
>>> B.python_class.mro()
>>> b_instance = B()
>>> assert isinstance(b_instance, A.python_class)
>>> assert EcoreUtils.isinstance(b_instance, A)

Importing an Existing XMI Metamodel/Model

XMI support is still a little rough on the edges, but the XMI import is on good tracks. Currently, only basic XMI metamodel (.ecore) and model instances can be loaded:

>>> from pyecore.resources import ResourceSet, URI
>>> rset = ResourceSet()
>>> resource = rset.get_resource(URI('path/to/mm.ecore'))
>>> mm_root = resource.contents[0]
>>> rset.metamodel_registry[mm_root.nsURI] = mm_root
>>> # At this point, the .ecore is loaded in the 'rset' as a metamodel
>>> resource = rset.get_resource(URI('path/to/instance.xmi'))
>>> model_root = resource.contents[0]
>>> # At this point, the model instance is loaded!

The ResourceSet/Resource/URI will evolve in the future. At the moment, only basic operations are enabled: create_resource/get_resource/load/save....

Dynamic Metamodels Helper

Once a metamodel is loaded from an XMI metamodel (from a .ecore file), the metamodel root that is gathered is an EPackage instance. To access each EClass from the loaded resource, a getEClassifier(...) call must be performed:

>>> #...
>>> resource = rset.get_resource(URI('path/to/mm.ecore'))
>>> mm_root = resource.contents[0]
>>> A = mm_root.getEClassifier('A')
>>> a_instance = A()

When a big metamodel is loaded, this operation can be cumbersome. To ease this operation, PyEcore proposes an helper that gives a quick access to each EClass contained in the EPackage and its subpackages. This helper is the DynamicEPackage class. Its creation is performed like so:

>>> #...
>>> resource = rset.get_resource(URI('path/to/mm.ecore'))
>>> mm_root = resource.contents[0]
>>> from pyecore.utils import DynamicEPackage
>>> MyMetamodel = DynamicEPackage(mm_root)  # We create a DynamicEPackage for the loaded root
>>> a_instance = MyMetamodel.A()
>>> b_instance = MyMetamodel.B()

Adding External Metamodel Resources

External resources for metamodel loading should be added in the resource set. For example, some metamodels use the XMLType instead of the Ecore one. The resource creation should be done by hand first:

int_conversion = lambda x: int(x)  # translating str to int durint load()
String = Ecore.EDataType('String', str)
Double = Ecore.EDataType('Double', int, 0, from_string=int_conversion)
Int = Ecore.EDataType('Int', int, from_string=int_conversion)
IntObject = Ecore.EDataType('IntObject', int, None,
Boolean = Ecore.EDataType('Boolean', bool, False,
                          from_string=lambda x: x in ['True', 'true'],
                          to_string=lambda x: str(x).lower())
Long = Ecore.EDataType('Long', int, 0, from_string=int_conversion)
EJavaObject = Ecore.EDataType('EJavaObject', object)
xmltype = Ecore.EPackage()
xmltype.nsURI = ''
xmltype.nsPrefix = 'xmltype' = 'xmltype'
rset.metamodel_registry[xmltype.nsURI] = xmltype

# Then the resource can be loaded (here from an http address)
resource = rset.get_resource(HttpURI('http://myadress.ecore'))
root = resource.contents[0]

Metamodel References by ‘File Path’

If a metamodel references others in a ‘file path’ manner (for example, a metamodel A had some relationship towards a B metamodel like this: ../metamodelb.ecore ), PyEcore requires that the B metamodel is loaded first and registered against the metamodel path URI like (in the example, against the ../metamodelb.ecore URI).

>>> # We suppose that the metamodel A.ecore has references towards B.ecore
... # '../../B.ecore'. Path of A.ecore is 'a/b/A.ecore' and B.ecore is '.'
>>> resource = rset.get_resource(URI('B.ecore'))  # We load B.ecore first
>>> root = resource.contents[0]
>>> rset.metamodel_registry['../../B.ecore'] = root  # We register it against the 'file path' uri
>>> resource = rset.get_resource(URI('a/b/A.ecore'))  # A.ecore now loads just fine

Adding External resources

When a model reference another one, they both need to be added inside the same ResourceSet.

resource = rset.get_resource(URI('uri/towards/my/secon/resource'))

If for some reason, you want to dynamically create the resource which is required for XMI deserialization of another one, you need to create an empty resource first:

# Other model is 'external_model'
resource = rset.create_resource(URI('the/wanted/uri'))

Exporting an Existing XMI Resource

As for the XMI import, the XMI export (serialization) is still somehow very basic. Here is an example of how you could save your objects in a file:

>>> # we suppose we have an already existing model in 'root'
>>> from pyecore.resources.xmi import XMIResource
>>> from pyecore.resources import URI
>>> resource = XMIResource(URI('my/path.xmi'))
>>> resource.append(root)  # We add the root to the resource
>>>  # will save the result in 'my/path.xmi'
>>>'test/path.xmi'))  # save the result in 'test/path.xmi'

You can also use a ResourceSet to deal with this:

>>> # we suppose we have an already existing model in 'root'
>>> from pyecore.resources import ResourceSet, URI
>>> rset = ResourceSet()
>>> resource = rset.create_resource(URI('my/path.xmi'))
>>> resource.append(root)

Dealing with JSON Resources

PyEcore is also able to load/save JSON models/metamodels. The JSON format it uses tries to be close from the one described in the emfjson-jackson project. The way the JSON serialization/deserialization works, on a user point of view, is pretty much the same than the XMI resources, but as the JSON resource factory is not loaded by default (for XMI, it is), you have to manually register it first. The registration can be performed globally or at a ResourceSet level. Here is how to register the JSON resource factory for a given ResourceSet.

>>> from pyecore.resources import ResourceSet
>>> from pyecore.resources.json import JsonResource
>>> rset = ResourceSet()  # We have a resource set
>>> rset.resource_factory['json'] = lambda uri: JsonResource(uri)  # we register the factory for '.json' extensions

And here is how to register the factory at a global level:

>>> from pyecore.resources import ResourceSet
>>> from pyecore.resources.json import JsonResource
>>> ResourceSet.resource_factory['json'] = lambda uri: JsonResource(uri)

Once the resource factory is registered, you can load/save models/metamodels exactly the same way you would have done it for XMI. Check the XMI section to see how to load/save resources using a ResourceSet.

NOTE: Currently, the Json serialization is performed using the defaut Python json lib. It means that your PyEcore model is first translated to a huge dict before the export/import. For huge models, this could implies a memory and a performance cost.

Creating Your own URI

PyEcore uses URI to deal with ‘stream’ opening/reading/writing/closing. An URI is used to give a file-like object to a Resource. By default, the basic URI provides a way to read and write to files on your system (if the path used is a file system path, abstract paths or logical ones are not serialized onto the disk). Another, HttpURI opens a file-like object from a remote URL, but does not give the ability to write to a remote URL.

As example, in this section, we will create a StringURI that gives the resource the ability to read/write from/to a Python String.

class StringURI(URI):
def __init__(self, uri, text=None):
    super(StringURI, self).__init__(uri)
    if text is not None:
        self.__stream = StringIO(text)

def getvalue(self):
    return self.__stream.getvalue()

def create_instream(self):
    return self.__stream

def create_outstream(self):
    self.__stream = StringIO()
    return self.__stream

The StringURI class inherits from URI, and adds a new parameter to the constructor: text. In this class, the __stream attribute is handled in the URI base class, and inherited from it.

The constructor builds a new StringIO instance if a text is passed to this URI. This parameter is used when a string must be decoded. In this context, the create_instream() method is used to provide the __stream to read from. In this case, it only returns the stream created in the constructor.

The create_outstream() methods is used to create the output stream. In this case, a simple StringIO instance is created.

In complement, the getvalue() method provides a way of getting the result of the load/save operation. The following code illustrate how the StringURI can be used:

# we have a model in memory in 'root'
uri = StringURI('myuri')
resource = rset.create_resource(uri)
print(uri.getvalue())  # we get the result of the serialization

mystr = uri.getvalue()  # we assume this is a new string
uri = StringURI('newuri', text=mystr)
resource = rset.create_resource(uri)
root = resource.contents[0]  # we get the root of the loaded resource

Deleting Elements

Deleting elements in EMF is still a sensible point because of all the special model “shape” that can impact the deletion algorithm. PyEcore supports two main way of deleting elements: one is a real kind of deletion, while the other is more less direct.

The delete() method

The first way of deleting element is to use the delete() method which is owned by every EObject/EProxy:

>>> # we suppose we have an already existing element in 'elem'
>>> elem.delete()

This call is also recursive by default: every sub-object of the deleted element is also deleted. This behavior is the one by default as a “containment” reference is a strong constraint.

The behavior of the delete() method can be confusing when there is many EProxy in the game. As the EProxy only gives a partial view of the object while it is not resolved, the delete() can only correctly remove resolved proxies. If a resource or many elements that are referenced in many other resources must be destroyed, in order to be sure to not introduce broken proxies, the best solution is to resolve all the proxies first, then to delete them.

Removing an element from it’s container

You can also, in a way, removing elements from a model using the XMI serialization. If you want to remove an element from a Resource, you have to remove it from its container. PyEcore does not serialize elements that are not contained by a Resource and each reference to this ‘not-contained’ element is not serialized.

Modifying Elements Using Commands

PyEcore objects can be modified as shown previously, using basic Python operators, but these mofifications cannot be undone. To do so, it is required to use Command and a CommandStack. Each command represent a basic action that can be performed on an element (set/add/remove/move/delete):

>>> from pyecore.commands import Set
>>> # we assume have a metamodel with an A EClass that owns a 'name' feature
>>> a = A()
>>> set = Set(owner=a, feature='name', value='myname')
>>> if set.can_execute:
...     set.execute()

If you use a simple command withtout CommandStack, the can_execute call is mandatory! It performs some prior computation before the actual command execution. Each executed command also supports ‘undo’ and ‘redo’:

>>> if set.can_undo:
...     set.undo()
>>> assert is None
>>> set.redo()
>>> assert == 'myname'

As for the execute() method, the can_undo call must be done before calling the undo() method. However, there is no can_redo, the redo() call can be mad right away after an undo.

To compose all of these commands, a Compound can be used. Basically, a Compound acts as a list with extra methods (execute, undo, redo…):

>>> from pyecore.commands import Compound
>>> a = A()  # we use a new A instance
>>> c = Compound(Set(owner=a, feature='name', value='myname'),
...              Set(owner=a, feature='name', value='myname2'))
>>> len(c)
>>> if c.can_execute:
...     c.execute()
>>> if c.can_undo:
...     c.undo()
>>> assert is None

In order to organize and keep a stack of each played command, a CommandStack can be used:

>>> from pyecore.commands import CommandStack
>>> a = A()
>>> stack = CommandStack()
>>> stack.execute(Set(owner=a, feature='name', value='myname'))
>>> stack.execute(Set(owner=a, feature='name', value='myname2'))
>>> stack.undo()
>>> assert == 'myname'
>>> stack.redo()
>>> assert == 'myname2'

Here is a quick review of each command:

  • Set –> sets a feature to a value for an owner

  • Add –> adds a value object to a feature collection from an owner object (Add(owner=a, feature='collection', value=b)). This command can also add a value at a dedicated index (Add(owner=a, feature='collection', value=b, index=0))

  • Remove –> removes a value object from a feature collection from an owner (Remove(owner=a, feature='collection', value=b)). This command can also remove an object located at an index (Remove(owner=a, feature='collection', index=0))

  • Move –> moves a value to a to_index position inside a feature collection (Move(owner=a, feature='collection', value=b, to_index=1)). This command can also move an element from a from_index to a to_index in a collection (Move(owner=a, feature='collection', from_index=0, to_index=1))

  • Delete –> deletes an elements and its contained elements (Delete(owner=a))

Dynamically Extending PyEcore Base Classes

PyEcore is extensible and there is two ways of modifying it: either by extending the basic concepts (as EClass, EStructuralFeature…), or by directly modifying the same concepts.

Extending PyEcore Base Classes

To extend the PyEcore base classes, the only thing to do is to create new EClass instances that have some base classes as superclass. The following excerpt shows how you can create an EClass instance that will add support EAnnotation to each created instance:

>>> from pyecore.ecore import *
>>> A = EClass('A', superclass=(EModelElement.eClass))  # we need to use '.eClass' to stay in the PyEcore EClass instance level
>>> a = A()  # we create an instance that has 'eAnnotations' support
>>> a.eAnnotations
>>> annotation = EAnnotation(source='testSource')
>>> annotation.details['mykey'] = 33
>>> a.eAnnotations.append(annotation)
>>> EOrderedSet([<pyecore.ecore.EAnnotation object at 0x7fb860a99f28>])

If you want to extend EClass, the process is mainly the same, but there is a twist:

>>> from pyecore.ecore import *
>>> NewEClass = EClass('NewEClass', superclass=(EClass.eClass))  # NewEClass is an EClass instance and an EClass
>>> A = NewEClass('A')  # here is the twist, currently, EClass instance MUST be named
>>> a = A()  # we can create 'A' instance
>>> a
<pyecore.ecore.A at 0x7fb85b6c06d8>

Modifying PyEcore Base Classes

PyEcore let you dynamically add new features to the base class and thus introduce new feature for base classes instances:

>>> from pyecore.ecore import *
>>> EClass.new_feature = EAttribute('new_feature', EInt)  # EClass has now a new EInt feature
>>> A = EClass('A')
>>> A.new_feature
>>> A.new_feature = 5
>>> A.new_feature


The dependencies required by pyecore are:

  • ordered-set which is used for the ordered and unique collections expressed in the metamodel,

  • lxml which is used for the XMI parsing.

These dependencies are directly installed if you choose to use pip.

Run the Tests

Tests uses py.test and ‘coverage’. Everything is driven by Tox, so in order to run the tests simply run:

$ tox

Liberty Regarding the Java EMF Implementation

  • There is some meta-property that could be missing inside PyEcore. If you see one missing, please open a new ticket!

  • Proxies are not “removed” once resolved as in the the Java version, instead they acts as transparent proxies and redirect each calls to the ‘proxied’ object.

  • PyEcore is able to automatically load some model/metamodel dependencies on its own.


In the current state, the project implements:

  • the dynamic/static metamodel definitions,

  • reflexive API,

  • inheritance,

  • enumerations,

  • abstract metaclasses,

  • runtime typechecking,

  • attribute/reference creations,

  • collections (attribute/references with upper bound set to -1),

  • reference eopposite,

  • containment reference,

  • introspection,

  • select/reject on collections,

  • Eclipse XMI import (partially, only single root models),

  • Eclipse XMI export (partially, only single root models),

  • simple notification/Event system,

  • EOperations support,

  • code generator for the static part,

  • EMF proxies (first version),

  • object deletion (first version),

  • EMF commands (first version),

  • EMF basic command stack,

  • EMF very basic Editing Domain,

  • JSON import (simple JSON format),

  • JSON export (simple JSON format),

  • introduce behavior @runtime,

  • resources auto-load for some cross-references,

  • derived collections,

  • multiple roots ressources,

  • xsi:schemaLocation support for XMI resources,

  • URI mapper like.

The things that are in the roadmap:

  • new implementation of EOrderedSet, EList, ESet and EBag,

  • new implementation of EStringToStringMapEntry and EFeatureMapEntry,

  • improve documentation,

  • copy/paste (?).

Existing Projects

There is not so much projects proposing to handle model and metamodel in Python. The only projects I found are:

PyEMOF proposes an implementation of the OMG’s EMOF in Python. The project targets Python2, only supports Class/Primitive Types (no Enumeration), XMI import/export and does not provide a reflexion layer. The project didn’t move since 2005.

EMF4CPP proposes a C++ implementation of EMF. This implementation also introduces Python scripts to call the generated C++ code from a Python environment. It seems that the EMF4CPP does not provide a reflexive layer either.

PyEMOFUC proposes, like PyEMOF, a pure Python implementation of the OMG’s EMOF. If we stick to a kind of EMF terminology, PyEMOFUC only supports dynamic metamodels and seems to provide a reflexive layer. The project does not appear seems to have moved since a while.


Thanks for making PyEcore better!

Additional Resources

  • This article on the blog of the Professor Jordi Cabot gives more information and implementations details about PyEcore.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pyecore-py2-0.9.0.tar.gz (81.2 kB view hashes)

Uploaded source

Built Distribution

pyecore_py2-0.9.0-py2-none-any.whl (49.7 kB view hashes)

Uploaded py2

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page