PyUML2: A Pythonic Implementation of the UML Metamodel

PyUML2 is an implementation of the UML2 metamodel for Python 2.7, >=3.4, relying on PyEcore. The goal of this project is to provide an almost-full implementation of the UML-2.5 standard in Python with Profile support and compatibility with the Eclipse UML2 project.

Installation

There is not yet a pypi package, you can manually install the project using:

$ pip install -e .

Documentation

Here is how to load a UML2 model using the implementation.

from pyecore.resources import ResourceSet
import pyuml2.uml as uml

rset = ResourceSet()
rset.metamodel_registry[uml.nsURI] = uml
resource = rset.get_resource('path/to/my/model.uml')
model = resource.contents[0]

print(model.name)
print(model.packagedElement)
print(model.nestedPackage)

Project State

Available:

• Generated UML2 Metamodel (using the awesome pyecoregen)
• Generated Type Metamodel (used by the UML2 metamodel)
• Some derived features implementation

On the roadmap:

• Add default primitive type library
• Derived features implementation
• Profiles support
  • Open an existing Profile (currently, partially supported, some cross-refs are not resolved)
  • Create Metamodel References
  • Define a Profile (as it is done in Eclipse UML2)
  • Apply a Profile
  • Open a profiled Model
  • Apply a Stereotype

Liberties Regarding the Eclipse implementation

The project goal is to be compatible with the Eclipse UML implementation, but it still takes some liberties:

• the Eclipse implementation pluralizes the name of features when required during the code generation. For example, the packagedElement on Package in the UML.ecore metamodel is singular, but the code Eclipse Implementation pluralizes it as packagedElements. PyUML2 does not perform this pluralization (at least not now).

Tests

The project is configured to be used with tox, so, you only need to run:

$ tox

Contributing

All contributions are welcome and are really appreciated. The project is brand new so there is a currently a lot to do. If you want to add the implementation of new derived features or method implementations, the blueprint of each method and feature is located in the pyuml2/uml_mixins.py module and can be filled.

There are some examples that can help you tame the code and show how to add a dedicated implementation. The more representative are:

• the isComposite methods in the PropertyMixin class for "non-many" feature
• the DerivedSuperclass class for "many" feature (manages .superClass for Class)

Obviously, for each added implementation, a test should be written.