Cobra

Cobra is a project creation and management tool for python and git. It creates a conda environment and a github repository based on your requirements.

Requirements

Python >= 3.6
Conda
Github

You can use the base conda to install cobra.

conda config --set auto_activate_base true

Installation

You have three different options to install cobra. Just always keep in mind that the python version need to be > 3.5.

pip

If your pip path is pointing to the pip3 you can also you just pip.

pip3 install cobra-projects

Then execute the command

cobra --install

and follow the instructions.

github

Download the sourcecode from github.
Move into the cobra root directory (folder which contains the setup.py file) cd cobra
Install it pip install .

Then execute the command

cobra --install

and follow the instructions.

Uninstallation

If you have to re-install or uninstall cobra:

Delete the access token from your github page (Settings -> Developer settings -> Personal access tokens -> delete (cobra)
rm -rf ~/.cobra
pip3/pip uninstall cobra-projects

Usage

usage: cobra [-h] [--install] [--test] [--clear] {create,list,remove} ...

Project creation (cobra) command line tool

positional arguments:
  {create,list,remove}  The different functions of cobra
    create              Create a new cobra project
    list                Lists all existing cobra projects
    remove              Remove a cobra project

optional arguments:
  -h, --help            show this help message and exit
  --install             Start the installation wizard.
  --test                Puts cobra into testing mode
  --clear               Clears all user files

Create

To create a new project you can use the create command. The minimum requirements is the -n parameter to give the project a name. You can also lik the project with already existing conda environments or github repos.

CARE: The -gh, --git-hook command is in beta. This command exports before each git commit the conda environment.

usage: cobra create [-h] -n PROJECT_NAME [-p PYTHON_VERSION] [-r REPO_NAME]
                   [-c CONDA_NAME] [-gh]

optional arguments:
  -h, --help            show this help message and exit
  -n PROJECT_NAME, --project-name PROJECT_NAME
                        Name of the project
  -p PYTHON_VERSION, --python-version PYTHON_VERSION
                        Python version for the project. Default: 3.5
  -r REPO_NAME, --repo-name REPO_NAME
                        If you already have a github repo for the project.
                        Enter the repo name.
  -c CONDA_NAME, --conda-name CONDA_NAME
                        If you already have a conda environment for the
                        project. Enter the name.
  -gh, --git-hook       Install a pre-commit git hook which updates the conda
                        environment file before you commit. This is just
                        working, if the current environment is the 
                        environment of the project!

List

This command lists all the cobra project with some additional information.

usage: cobra list [-h]

optional arguments:
  -h, --help  show this help message and exit

Remove

After some time you want to remove a project from your system. This you can do with the remove command. The remove command with out additional options will just delete the cobra project from the system but not the files, environment or the repository. If you want to remove the the files, environment or repository use the below shown parameters.

usage: cobra remove [-h] -n NAME [-r] [-c] [-f] [-a]

optional arguments:
  -h, --help            show this help message and exit
  -n NAME, --name NAME  Name of the cobra project to remove
  -r, --repo            Also removes the remote repository
  -c, --conda           Also removes the conda environment from the system
  -f, --folder          Also removes the local folder
  -a, --remove-all      Remove everything (repo, folder, conda)

Testing

To do testing you have to create the conda environment test_environment.yml and activate it:

conda env create -f test_environment.yml
conda activate cobra_dev

Then you can test the whole program including test coverage (htmlcov/index.html) with:

tox

Cite us

@inproceedings{voegtlin2020cobra,
    address = {Online},
    author = {Voegtlin, Lars and  Pondenkandath, Vinaychandran and Ingold, Rolf},
    title = {{Cobra: A CLI Tool To Create And Share Reproducible Projects}},
    booktitle = {2020 7th Swiss Conference on Data Science (SDS)},
    year = {2020},
    month = {jun},
}

TODO

Create repo with template
Create conda Dockerfile
Automatically build docker container and push it to dockerhub

lvoegtlin/cobra