pygeppetto

Home of the Geeppetto Python API. The API allows to create a Geppetto Model from Python.

Installation

Until pygeppetto is still in development, it is highly recommended to use a virtualenv in order to deploy it. Once you have a dedicated virtualenv, you can simply install pygeppetto:

$ python setup.py install

Pygeppetto API Basic Usage

First, import the pygeppetto API:

import pygeppetto

This will load the pygeppetto API and name it pygeppetto. Then, you can create instances and handle them:

# We create a new lib
flib = pygeppetto.GeppettoLibrary(name='mylib')
# We create a GeppettoModel instance and we set a name a assign a lib
root = pygeppetto.GeppettoModel(name='MyGeppettoModel', libraries=[flib])

The pygeppetto API also allows you to set all attributes in a "classical" fashion:

root = pygeppetto.GeppettoModel()  # We create a GeppettoModel instance
root.name = 'MyGeppettoModel'  # We set a name
flib = pygeppetto.GeppettoLibrary()  # We create a new lib
flib.name = 'mylib'
root.libraries.append(flib)  # We add the new lib to the created root

If you wan to open an existing XMI, you need to use a ResourceSet (not required, but prefered).

# We import the class that will be used to read the XMI from PyEcore
from pyecore.resources import ResourceSet, URI

# We create a new resource set (not required, but better)
rset = ResourceSet()

Using this ResourceSet, we are able to read the Geppetto XMI:

model_url = URI('tests/xmi-data/MediumNet.net.nml.xmi')  # The model URI
resource = rset.get_resource(model_url)  # We load the model
geppettomodel = resource.contents[0]  # We get the root

At the end of this script, geppettomodel contains the model root.

In order to serialize a new version of the modified model, there is two options. The first one is to serialize onto the existing resource (i.e: in the same file), or to serialize in a new one:

# Using the first option
resource.save()

# Using the second option
resource.save(output=URI('my_new_file.xmi'))

Dependencies

Python >= 3.5
pyecore

Contributions

If the geppettoModel.ecore evolves, the static metamodel must be regenerated. The process of adding a new version is the following:

Copy the of the new geppettoModel.ecore inside ecore/ (in order to keep a version from which the static metamodel is generated).
Generate the new version of the static metamodel.
Manually merge modifications between the current and the new version (if there is manual modifications in the current version).
Run the tests

How to Generate a New Version

The pygeppetto API is generated from the geppettoModel.ecore using the PyEcore Acceleo generator (ecore2pyecore.mtl). The .ecore is a copy of the geppettoModel.ecore from org.geppetto.model (development branch).

Generate the code with pyecoregen (suggested)

Install pyecoregen

pip install pyecoregen

Run the following script:

from pyecore.resources import ResourceSet
from pyecoregen.ecore import EcoreGenerator
import pyecore.type  # We register the XML types (generated by pyecoregen)

# We open the metamodel
rset = ResourceSet()
mm_root = rset.get_resource('pygeppetto/ecore/geppettoModel.ecore').contents[0]

# We generate the code using the EcoreGenerator
EcoreGenerator(auto_register_package=True).generate(mm_root, outfolder='pygeppetto')

Then do the following fix replacements:

from model -> from pygeppetto.model
from type -> from pyecore.type

Generate the code with Eclipse

The ecore2pyecore.mtl script can be directly used in Eclipse as a Acceleo generator.

Install the Acceleo plugin into eclipse
Create a new Acceleo project
Add the file ecore2pyecore.mtl to the project main package
run ecore2pyecore.mtl as an acceleo project. The run configuration will popup: specify the geppettoModel.ecore file to use and the generation path. Generate the code in a path different from pygeppetto and merge the code

Patches and Overrides

The model generated from the ecore file contains some code stubs that are implemented manually on Python. The place to make those implementations is not the files themselves though but the overrides.py file. Writing the overrides there keeps the regeneration process easier as the written code does not actually need to be merged.

Merge conflicts

Manual modifications may have been introduced in the version of the static Geppetto metamodel (e.g: implementation of some methods or technical method additions). The generated version must be manually merged with the new generated one (e.g: using meld or other tool).

An easy way to do that is by using git.

1 - Generate the code in a in a directory reproducing the pygeppetto structure, say pygeppetto_new/pygeppetto/model

2 - Initialize a new git local repo inside our new pygeppetto

cd pygeppetto_new
git init

3 - Add the remote for pygeppetto

git remote add origin https://github.com/openworm/pygeppetto.git

4 - Merge with actual geppetto branch

git merge --allow-unrelated-histories origin/development

5 - Resolve conflicts: use incoming changes for clear overrides and implemented methods

Run the Tests

Tests are written using pytest and are run using tox. To launch all the tests the following command is enough:

$ tox

Or, if you want to avoid using tox, you can just:

$ python -m pytest tests/

openworm / pygeppetto