Dependencies
Skelebot offers the ability to list your Python dependencies directly in the skelebot.yaml file and have these dependencies installed into your Docker image automatically.
...
dependencies:
- numpy=1.11.2
- pandas
- ...
By default dependencies are installed using pip install.
Versions for packages can be specified by appending ={version}, =={version}, >={version}, etc. to the end of the dependency name. Skelebot supports most of the standard python version specifiers.
Skelebot also supports dependencies to be installed from the local file system, GitHub repo, requirements file, or pyproject.toml file:
dependencies:
- req:requirement.txt
- proj:pyproject.toml
- file:libs/myPackage.tgz
- req:requirements.txt
- github:myGitHub/fakeRepo
- proj:pyproject.toml
When using a pyproject file, all specified required and optional set(s) of dependencies will be installed.
CodeArtifact Python Packages
Skelebot also supports pulling Python packages that are stored in AWS CodeArtifact. This requires a good deal of information in order to authenticate and pull the correct asset for the package.
dependencies:
- ca_file:{domain}:{owner}:{repo}:{pkg}:{version}:{profile}
These values are needed in order to preperly obtain the package and include it in the Docker image.
- domain - The domain name of the AWS CodeArtifact repository (ex: my_domain)
- owner - The owner of the AWS CodeArtifact repository (ex: 111122223333)
- repo - The repository in CodeArtifact where the package is located (ex: my_repo)
- pkg - The name of the Python package to be installed (ex: my_package)
- version - The version of the package to be pulled (ex: 1.0.0)
- profile - [OPTIONAL] The AWS profile on your machine that has access to this CodeArtifact repository (ex: dev)
NOTES:
- When installing via
file:orgithub:the ability to specify a version is not available. - Python
githubdependencies may optionally specify a protocol like https (more info here).