hellopymsdl 1.0.1

A Python template inspired from the Maven Standard Directory Layout

PYMSDL_Template

PYMSDL^{(PYthon Maven Standard Directory Layout)} template is a Python Poetry template inspired from the Maven Standard Directory Layout .

WARNING Prerequisites:

• Python >= 3.10.1
• Poetry >= 1.1.0

Since there are some existing limitations, it is strongly advised to read the Project organization part before the Project commands one.

Note: this template is configured in order to work with Poetry.

if you want to use another layout than Maven Standard Directory Layout, you can change the project structure as explained in the pyproject.toml & project.py files part.

Table of contents

• Quickstart
  • Create a new project
    • 1. Click on Use this template
    • 2. Fork this template
    • 3. Clone/download this template
  • Prepare your environment
  • Test your setup
  • Manage your project
  • Let's dev
• Project organization
  • pyproject.toml & project.py files
  • Maven Standard Directory Layout with python
    • Sources & Resources directories configuration
    • WARNING: Limitations
      • A. Use root package names with suffix
      • B. Use namespace package as root package
      • C. Don't fully respect the Maven Standard Directory Layout
• Project commands
  • Clean project
  • Run module
  • Run tox
  • Run linter
  • Run tests
  • Run mutation
  • Build
    • Wheel archive
    • Source Distribution
  • Delivery (on https://pypi.org/)
• Docker
  • Dev environment
    • Build the pymsdl:devenv docker image
    • Run the pymsdl:devenv environment
  • Docker application image delivery
    • Build your docker image
    • Run your docker image
    • Push your docker image
• Workflows
  • CI
  • On_release

Quickstart

Create a new project

There are several ways to use this template for your project:

1. Click on Use this template

Create a new project from the Use this template button is the standard way for a new GitHub project from a template.

See: Creating a repository from a template

Note: You have to be logged in

2. Fork this template

Create a new project by forking the template is the "old way" to work with templates on GitHub.

However, there are some differences between "fork" and "use template" explained here which can help you about your choice.

See: About forks

3. Clone/download this template

This is a way to use if you are creating a new project but, you are not working on GitHub.

Prepare your environment

PYMSDL template requires Python >= 3.10.1 and Poetry >= 1.1.0.

If your environment is not ready yet, a docker ready to work environment is provided (see Dev environment).

In order to simplify your pipeline evolution and local use, this template provides you a python script with basic management commands. See: Project commands.

You have to run this command line in order to download your project dependencies:

./project.py load_deps

Test your setup

PYMSDL template provides a simple sample project named hellopymsdl. You can test your setup by running this command line:

./project.py test

Manage your project

Since this template use Poetry you have to learn each poetry command.

Moreover, this template provides you a python script with basic management commands. See: Project commands

Note Project commands samples should work on the provided hellopymsdl simple sample project.

Let's dev

Once previous steps are done you can start your developments by respecting the content of Project organization part.

Project organization

pyproject.toml & project.py files

The pyproject.toml file is used in order to describe and deliver correctly your project.

See:

• PEP 517
• PEP 518
• PEP 621
• PEP 660
• Poetry pyproject.toml details

Moreover, this template provides the project.py file which is a command line wrapper in order to simplify the project evolution (see Project commands).

When you start a new project from this template, YOU DON'T HAVE TO UPDATE THE PROJECT.PY FILE, until you switch from Poetry to another tool.

In order to set your project properties, you just have to update the pyproject.toml.

Note: If you don't want to use the Maven Standard Directory Layout, you can reconfigure the project structure by replacing each /src/main/... or /src/test/... by your structure in the pyproject.toml file.

Maven Standard Directory Layout with python

Sources & Resources directories configuration

This template attempts to reach the Maven Standard Directory Layout for python.

That's why you will find the following directory structure:

• src/main/python: Should contains your application sources
• src/main/resources: Should contains your application resources
• src/test/python: Should contains your application test sources
• src/test/resources: Should contains your application test resources

Sources and resources directories must be in your PYTHONPATH.

However, if you don't want to configure your PYTHONPATH, you can use the provided Run module command.

Note: if you are using IDE like Pycharm , it means src/main/python and src/main/resources must be marked as Sources Root (Right click on these directories > Mark directory as > Sources Root) and src/test/python and src/test/resources must be marked as Test Sources Root (Right click on these directories > Mark directory as > Test Sources Root).

Note: Due to a Poetry limitation, using installation with edit mode (pip install -e .) in order to avoid the PYTHONPATH configuration will not work and fail with:

build backend is missing the 'build_editable' ... Consider using a build backend that supports PEP 660.

WARNING: Limitations

Python resources MUST be located in a python package (ie: directory containing an __init__.py file. (You can see that like a java classpath).

Since a package sources cannot be split in several directories, each package in src/main/python, src/main/resources, src/test/python and src/test/resources directories must be different. In case of conflict, the package from the first directory found in your PYTHONPATH will be used.

So if you want to respect the Maven Standard Directory Layout there are several suggestions:

A. Use root package names with suffix

Use root package names with suffixes like this:

• <My_Root_Package_Name> for sources (in src/main/python)
• <My_Root_Package_Name>_rsrc for resources (in src/main/resources)
• <My_Root_Package_Name>_test for test sources (in src/test/python)
• <My_Root_Package_Name>_test_rsrc for test resources (in src/test/resources)

Project structure example:

<My_Project>
 |- ...
 |- src
     |- main
     |   |- python
     |   |   |- <MY_ROOT_PACKAGE_NAME>
     |   |       |- __init__.py
     |   |       |- ...
     |   |- resources
     |       |- <MY_ROOT_PACKAGE_NAME>_rsrc
     |           |- __init__.py
     |           |- ...
     |- test
         |- python
         |   |- <MY_ROOT_PACKAGE_NAME>_test
         |       |- __init__.py
         |       |- ...
         |- resources
             |- <MY_ROOT_PACKAGE_NAME>_test_rsrc
                 |- __init__.py
                 |- ...

Note: since root package names are different (_rsrc, _test, _test_rsrc suffixes) it cannot exist conflict between packages in src/main/python, src/main/resources, src/test/python and src/test/resources.

Note: _rsrc, _test, _test_rsrc are just suggested suffixes you can use your own suffixes.

B. Use namespace package as root package

If you really want a common root package name, you can use a namespace package as root. This namespace package must contain packages without conflict between sources and resources directories.

Reminder: A namespace package doesn't contain any source or __init__.py file.

Project structure with namespace package as root package example:

<My_Project>
 |- ...
 |- src
     |- main
     |   |- python
     |   |   |- <MY_ROOT_NAMESPACE_PACKAGE_NAME>
     |   |       |- <MY_PACKAGE_NAME>
     |   |       |   |- __init__.py
     |   |       |   |- ...
     |   |       |- ...
     |   |- resources
     |       |- <MY_ROOT_NAMESPACE_PACKAGE_NAME>
     |           |- <MY_PACKAGE_NAME>_rsrc
     |           |   |- __init__.py
     |           |   |- ...
     |           |- ...
     |- test
         |- python
         |   |- <MY_ROOT_NAMESPACE_PACKAGE_NAME>
         |       |- <MY_PACKAGE_NAME>_test
         |       |   |- __init__.py
         |       |   |- ...
         |       |- ...
         |- resources
             |- <MY_ROOT_NAMESPACE_PACKAGE_NAME>
                 |- <MY_PACKAGE_NAME>_test_rsrc
                 |   |- __init__.py
                 |   |- ...
                 |- ...

C. Don't fully respect the Maven Standard Directory Layout

If you are not agree with the previous suggestions, you can remove the src/main/resources, src/test/python and src/test/resources directories and put your resources and tests directly into the src/main/python directory.

Note: if you do that, your tests will be included during your project/package installation (not only in the source distribution as it is suggested in the best practices).

Project commands

Project commands are available from the project.py python file.

If you take a look at this file, you will see that it contains:

• A shared project_properties variable which contains all project properties.

• A command line wrapper: this wrapper is designed in order to be used locally and in your deployment scripts (like CI/CD).

  Indeed, by using it, if you want to update you "project tools" (sample: switching from poetry to setuptools) you just have to update this script but not all your pipeline.

  Moreover, this wrapper manage the project PYTHONPATH for you.

  Command wrapper man:

  PROJECT COMMANDS WRAPPER:
  
  Usage: python project.py <COMMAND_1> <arg1_1 ...> ... <COMMAND_N> <argN_1 ...>
          Note: In order to get the wrapped command help, you can try python project.py <command> --help
  
  Available commands are:
    load_deps     Install all dependencies (dev included)
    clean         Remove directories generated by the "build" commands (like 'sdist' or 'wheel')
    run           Run module which can be in the project structure without having to configure the PYTHONPATH
    tox           Run tox
    lint          Run linter
    test          Run configured unit tests
    mut           Run mutation tests
    wheel         Build Wheel archive
    sdist         Build sdist archive
    upload        Upload available deliveries
  

  Note: Using --help argument on a command or if an error occurs, the message will be from the wrapped command line.

Load dependencies

./project.py load_deps

or

python project.py load_deps

Clean project

./project.py clean

or

python project.py clean

Note: Remove dist directory and test cache directories.

Run module

PYMSDL template provides you a run python module command line (even in the project packages) without having to configure your PYTHONPATH:

./project.py run -m <MODULE_NAME>

or

python project.py run -m <MODULE_NAME>

Note: the run command is like opening python in your poetry virtual environment and the project structure configured in your PYTHONPATH.

Moreover, your opened python will work from the current directory (not from the project.py one).

Examples:

- Open a python console from the current directory:

./project.py run

Note: <PATH_TO_PROJECT.PY> run will open a python console in your current directory not in the project.py one.

- Run module example:

./project.py run -m hellopymsdl.__main__

Run the __main__.py module from hellopymsdl package.

- Run package example:

./project.py run -m hellopymsdl

Note: your package MUST contains a __main__.py module.

- Run module from path example:

./project.py run src/main/python/hellopymsdl/__main__.py

Run the __main__.py module from hellopymsdl package. WARNING: the path is from the command CWD if not absolute.

- Run package from path example:

./project.py run src/main/python/hellopymsdl

Note: your package MUST contains a __main__.py module. WARNING: the path is from the command CWD if not absolute.

- Run module with arguments example (using --args string parameter):

./project.py run -m hellopymsdl.__main__ --arg1 --arg2=my_arg2 ...

Run tox

PYMSDL template uses tox in order to manage your tests (unit tests, lint, etc.)

./project.py tox

or

python project.py tox

Note: You will find the tox configuration in the pyproject.toml file ([tool.tox] section).

The default tox configuration provides the following envlist:

• pylint: Executes the configured linter
• covclean: Cleans the coverage reports
• py310: Executes the configured test runner under python 3.10
• report: Generates console, html and xml reports
• mutation: Executes mutation testing and generate reports

Moreover, the provided configuration runs each test on INSTALLED packages (Not on the given sources even if obviously the installed packages depends on the given sources). The goal is to test definitive installed versions.

Run linter

./project.py lint

or

python project.py lint

Note: json and txt reports will be available in build/reports/lint.

Note: By default this template uses pylint as linter. You will find the linter configuration in the pyproject.toml file ([tool.pylint....] sections).

Note: This command uses tox but only in order to call the linter.

Run tests

This project is configured to execute tests with Pytest:

./project.py test

or

python project.py test

Note: html, json and xml reports will be available in build/reports/coverage.

Note: If you change the project structure, don't forget to update the [tool.pytest.ini_options] section from the pyproject.toml file.

Note: This command uses tox but only in order to run your tests.

Run mutation

This template provides mutation testing by using mutmut:

./project.py mut

or

python project.py mut

Note: html report will be available in build/reports/mutation.

Note: If you change the project structure, don't forget to update the pyproject.toml file.

Note: This command uses tox but only in order to run your tests.

Build

Wheel archive

./project.py wheel

or

python project.py wheel

Source Distribution archive

If you need a source distribution (sdist) archive:

./project.py sdist

or

python project.py sdist

Delivery (on https://pypi.org/)

./project.py upload

or

python project.py upload

Note: Obviously, this command must be executed after the Build one.

Docker

WARNING This part requires docker to be installed

Dev environment

If you haven't a python/poetry environment installed, PYMSDL template provides you a "ready to work" Dockerfile:

Build the pymsdl:devenv docker image

docker build -t pymsdl:devenv-1.0.1 -f docker/devenv/Dockerfile .

Run the pymsdl:devenv environment

docker run --rm -it --name pymsdl-dev-env -v ${PWD}:/app pymsdl:devenv-1.0.0

It opens a sh on a well configured python/poetry environment.

Note: you have to use the volume (-v) option in order to mount your project files into the container /app directory. This is useful for staying up to date with your updates.

Now, you are ready to work (see Quickstart).

Docker application image delivery

Note: this step is useless if you are working on a python package/Library/Framework to share (ie, not an application) because in these cases you want to publish your sdist or/and wheel archive(s).

Build your docker image

PYMSDL template provides a Dockerfile in order to build an image for your application based on a generated wheel distribution (See: Dockerfile)).

Note: this Dockerfile is a "generic" one in order to provide a starting docker image distribution solution. You can (/should) update (/rewrite) it in order to adapt it to your application (see: Best practices for writing Dockerfiles).

Build image command:

docker build \
  -f docker/app/Dockerfile \
  -t <DOCKER_IMG_NAME>:<VERSION> \
  --build-arg wheel_name=<WHEEL_NAME> \
  --build-arg entrypoint=<ENTRYPOINT_VALUE> \
  --build-arg cmd=<CMD_VALUE> \
  .

Where:

• <DOCKER_IMG_NAME>: Is your docker image name
• <VERSION>: Is your docker image version
• <WHEEL_NAME>: [REQUIRED] Is the wheel archive name to install from the dist directory
• <ENTRYPOINT_VALUE>: [OPTIONAL: default is python] Is the docker file ENTRYPOINT
• <CMD_VALUE>: [OPTIONAL: default cmd is empty] Is the dockerfile CMD

Note: Obviously, this command must be executed after the Wheel archive one.

Note: Using an entrypoint with an empty cmd provides the possibility for the end user to use cmd as your project parameters (argv). Moreover, you can use cmd in order to define your default project parameters.

Example with the provided hellopymsdl sample application:

First we create a wheel archive with ./project.py wheel.

Note: your dist directory can contain several versions (/wheel archives)

- Build docker image from python entry point example:

The hellopymsdl project provides an hello entry point (See: pyproject.toml).

docker build \
  -f docker/app/Dockerfile \
  -t  hellopymsdl:1.0.1 \
  --build-arg wheel_name=hellopymsdl-1.0.1-py3-none-any.whl \
  --build-arg entrypoint=hello \
  .

- Build docker image from module example:

docker build \
  -f docker/app/Dockerfile \
  -t  hellopymsdl:1.0.1 \
  --build-arg wheel_name=hellopymsdl-1.0.1-py3-none-any.whl \
  --build-arg entrypoint="python -m hellopymsdl.__main__" \
  .

Run the __main__.py module from hellopymsdl package.

- Build docker image from package example:

The hellopymsdl project provided an hello entry point (See: pyproject.toml).

docker build \
  -f docker/app/Dockerfile \
  -t  hellopymsdl:1.0.1 \
  --build-arg wheel_name=hellopymsdl-1.0.1-py3-none-any.whl \
  --build-arg entrypoint="python -m hellopymsdl" \
  .

Note: your package MUST contains a __main__.py module.

- Build docker image without entrypoint build-arg:

if you don't define the entrypoint build-arg, the default one is python. So if you run your image without any argument (or cmd build-arg) it will open a python console in an environment where your project is installed

Trick: If you want to build the last updated wheel (not the last version but the last built archive) you can use --build-arg wheel_name=$(ls -1At dist/ | head -1)

Run your docker image

Once your docker image is ready, you can run it

docker run --rm  <DOCKER_IMG_NAME>:<VERSION>

Where:

• <DOCKER_IMG_NAME>: Is your docker image name
• <VERSION>: Is your docker image version

Example with the provided hellopymsdl sample application:

docker run --rm --name hellopymsdl hellopymsdl:1.0.1

Note: Since we used only the entrypoint build-arg in our previous samples, if you have to pass arguments (argv) to your project you can do it directly like this:

docker run --rm --name hellopymsdl hellopymsdl:1.0.1 arg-1 arg-2 ... arg-N

Moreover, you can define default values for your project parameters. To do that, you have to use the cmd build-arg like this:

docker build \
  -f docker/app/Dockerfile \
  -t  hellopymsdl:1.0.1 \
  --build-arg wheel_name=hellopymsdl-1.0.1-py3-none-any.whl \
  --build-arg entrypoint=hello \
  --build-arg cmd="arg-1 arg-2 ... arg-N" \
  .

Note: You can override the default cmd by using the previous command line sample.

Push your docker image

In order to push your image into a repository (like Docker Hub), you will have to use the docker push command line.

See Docker repositories

Workflows

The PYMSDL template provides a simple GitHub Actions workflow in order to check and manage your project (See: .github).

CI

This workflow is executed on pull_request or push on main. It aims to check your project quality. To do that the following steps are executed:

• Build archives
• Lint
• Unit tests
• Mutation tests
• Check dependencies vulnerabilities

See: ci.yml

On_release

This workflow is used in order to publish your project into pypi.

By default, PYMSDL template try to publish your project into pypi when you publish a new release.

Note: This workflow doesn't execute tests again because released version should be on main branch and this branch is configured to be protected (only pull requests can be merged on it which implies the CI workflow is executed first).

Note: A PYPI_API_TOKEN secret must be configured as pipy API key (See: pypi api token).

See: on_release.yml.

You can find the hellopymsdl sample project published here

Note: it means you can run pip install hellopymsdl.