The CNS python package eco-system

A repository to check out all packages of our eco-system, to set up a virtual environment, and finally to run all unit tests. The directory names of the submodules was modified intentionally to render the dependency graph manually, as pip is not able to resolve dependencies within a requirements.txt file. If one wishes to add a new git submodule, think of an appropriate prefix. This allows us (i) to generate a requirements.txt file based on all submodules and (ii) to run all pyhton unittests.

HINT: This project can also be used to develop/extend existing packages (submodules).

Overview of the cnspy eco-system

HINT: the diagram was created with DrawIO

Package List

• cnspy_csv2dataframe and the official package
• cnspy_numpy_utils and the official package
• cnspy_rosbag2csv and the official package
• cnspy_rosbag2image and the official package
• cnspy_script_utils and the official package
• cnspy_spatial_csv_formats and the official package
• cnspy_timestamp_association and the official package
• cnspy_trajectory and the official package
• cnspy_trajectory_evaluation and the official package
• cnspy_ranging_evaluation and the official package

Setup

Clone the repository with all submodules:

git clone --recurse-submodules -j8 git@github.com:aau-cns/cnspy_eco_system_test.git

or after cloning

git submodule update --init --recursive

Then run the source setup-env.sh that will

1. generate the requirements.txt based on the submodules
2. install a virtual environment python-venv
3. activate the local interpreter (venv) or launch PyCharm from the root.

Unit-tests

After the setup, simply run run_tests.sh and wait (some windows of plots need to be closed as they are blocking!).

Upload packages

To upload the files to PyPI (or Test PyPI by uncommenting command) is automated by running upload_packages.sh. Follow the official documentation and you might have to configure the API token.

Develop

One can use this repository to modify or extend our package eco-system. A good choice is to the IDE PyCharm. As the dependencies (our new packages) have to be resolved properly, the IDE has to use our virtual Python environment. Therefore, run the application in the root of the repository after the setup step was performed. One will be asked to define a project interpreter, but reject it in the beginning. Then got to File->Settings->Project:<>->Python Interpreter. There none is selected, thus click on the setting button (*) to get the Add dialog. Click on Existing Interpreter and navigate to <root>/python-venv/env/bin/python3. Now everything should be working. A quick check is to run a test in the package trajectory_evaluation (e.g. test_TrajectoryNEES.py).

Develop a new package

1. Create a new git repository containing initial files for the python package (LICENCE, README.md, RELEASE, setup.py, requirements.txt, folders: test, <package_name>) copy and modify from e.g, the [cnspy_spatial_csv_formats]((https://github.com/aau-cns/cnspy_spatial_csv_formats) package.
2. Add it as a submodule under the folder pkgs, with an alphabetical prefix referring to the top-level dependency. E.g. d_csv2dataframe requires b_spatial_csv_formats. The prefix is needed to create automatically the requirements.txt file! E.g: pkgs$ git submodule add git@github.com:aau-cns/XXX.git f_XXX
3. If the have already create a virtual environment in the root of cnspy_eco_system_test, source the python environment (cnspy_eco_system_test$ source python-venv/env/bin/activate), recreate the reqirements.txt by cnspy_eco_system_test$ ./generate_requirements.sh, and install all packages again by (env) <>/cnspy_eco_system_test$ pip install -r requirements.txt
4. Now, the new package is in the environment and can be developed/modified in Pycharm that is launch in the root with the virtual-environment selected.

License

Software License Agreement (MIT License), refer to the LICENSE file.

Sharing is caring! - Roland Jung