PythonAwesomeApp

A source-code template for Python3 applications.

Quick-Start

Please note that, as of now, these instructions are applicable to linux based systems (Ubuntu). Support for MacOS & Windows will be coming soon.

Creating a Python project based on this template

• Install cookiecutter

  pip3 install --user cookiecutter

• Create a python project

  python3 -m cookiecutter https://github.com/afaquejam/PythonAwesomeAppTemplate.git

Running this app

• Clone this source-code repository:

  git clone git@github.com:afaquejam/PythonAwesomeApp.git

• Set up the Python’s Virtual Env.

  python3 -m venv virtualenv
  source virtualenv/bin/activate
  

• Install dependencies:

  pip3 install -r requirements.txt

• Run the app:

  make

• Run the tests:

  make test

• Generate, view test-reports & code-coverage reports (assuming firefox is pre-installed):

  make test_report

• Turn off the Python virtualenv.

  deactivate

Source-Code Organization

• Source-code organization guidelines are here. Please note that in this project, __init__.py has been used because it was a requirement of the pyreverse tool.
• The source-code organization of this project looks as show below. Please note that some directories have not been shown for readability purposes.

  .
  ├── __init__.py
  ├── awesome_app
  ├── package_one
  │   ├── __init__.py
  │   ├── module_one.py
  │   ├── config
  │   │   └── integer_adder.cfg
  │   └── tests
  │       └── test_integer_adder.py
  .
  .
  .
  ├── makefile
  ├── LICENSE
  ├── CHANGELOG.md
  ├── README.md
  └── requirements.txt
  

Testing & Mocking

• Use Pytest.
  • Use pytest fixtures for set-up & teardown functions.
  • Use parameterized tests.
  • Use pytest-cov to generate test reports.
  • Use pytest plugins to enhance your test reports.
• Mocking
  • Use unittest.mock

Development Work-flow

• Assumptions:

  • The application needs few run-time dependencies & dev-dependencies.

  • During the development phase, you need both run-time dependencies & dev-dependencies.

  • When the source-code is pushed to the remote git repo, the CI/CD would run the tests & if successful, package this app. In this case, the CI/CD would need to install both type of dependencies.

  • The application needs to be installed on the target runtime environment. In this case, only run-time dependencies need to be installed.

• In this case, our work-flow would be as follows:

  • Dependencies (both types) should be specified in a requirements-to-freeze.txt file.

  • Install the dependencies:

    pip3 install -r requirements-to-freeze.txt

  • When you’re about to commit your changes, freeze your dependency versions by running the following command:

    pip3 freeze > requirements.txt

  • Dependencies will be installed on the CI/CD or on the target run-time environment using the command:

    pip3 install -r requirements.txt

  • You can upgrade your dependencies by running the following commands:

    pip3 install -r requirements-to-freeze.txt --upgrade
    pip3 freeze > requirements.txt
    

• A mechanism of separating the dev-dependencies from run-time dependencies must be found, such that only run-time dependencies are installed on the target run-time environment.

• In this project, an effort is on-going to simplify the developer work-flow: https://docs.pipenv.org/

Profiling

• Install the profiling softwares:

  pip3 install snakeviz memory_profiler matplotlib
  sudo apt-get install nethogs iotop
  

• Measure running time of a program:

  python3 -m cProfile -o app.prof awesome_app
  snakeviz app.prof
  

• Measure memory consumption of a program:

  mprof run ./awesome_app
  mprof plot
  

• Measure I/O:
  • Network: nethogs
  • Disk: iotop
  • Both provide rudimentary interface to view the I/O information.

Linting

• PEP 8 style guide (https://bit.ly/1ARqSBt) is recommended.
• Automate linting in your favorite code-editor.
• Configuring VSCode for Python programming: https://gist.github.com/afaquejam/af1c3a2d3ab00aa8528d945d98bf5a57

Documentation

• Changelog, git-tagging & release notes.
  • Changelog guidelines by Keep a Changelog are recommended: https://keepachangelog.com/en/1.0.0/
  • Use git-tags to tag your git repository & keep it consistent with the changelog.
  • Maintaining release notes is up-to you how you maintain them. Example release notes: https://atom.io/releases
  • You might argue that what’s the difference between between changelog & release notes? In my opinion, changelog is much more verbose than release notes. However it’s up-to you how you differentiate between them.
• User documentation
  • Please make sure you have instructions how to install your software, it’s dependencies & run it.
• Developer documentation
  • Make sure you have relevant documentation for developers such that they can navigate, read, understand & contribute to the project.

Packaging & Release

• There are different ways to package a Python app depending on how it’s going to be used. Here’s a nice introduction on the topic: https://docs.python-guide.org/shipping/packaging/
• If the app you’re developing is a library, then packaging it as a wheel distribution &/or uploading it to PyPI is an option. Here’s a good tutorial on it: https://packaging.python.org/tutorials/packaging-projects/
• If you’re distributing your app as a standalone application, then pyinstaller is the magical tool for you. This tool can produce a self-contained single app file, which contains the Python interpreter, the app & it’s dependencies, for all Mac, Windows & Linux. Steps to create a distribution package for your app (it’ll produce binaries for the OS on which it’s built):
  • Install pyinstaller: pip3 install pyinstaller
  • pyinstaller awesome_app -n awesome_app --onefile
  • In this case, awesome_app is the main script which gets executed & -n option is the name of the app.
  • This will create the app binary dist folder. Copy & distribute. Simple as that!

Concurrency & Parallelism

• Concurrent programs can be written using Python’s threading module. However, please note that due to the Global Interpretor Lock (GIL) in CPython & PyPy, threads will not execute in parallel.
• If you want to achieve true parallelism on a multi-core processor system, then multiprocessing module provides a way to spawn multiple processes, which can run parallely.
• Python provides good thread & process synchronization mechanisms. You can check their respective sync. mechanisms.
• Queues are recommended ways for inter-thread & inter-process communication.
• Relevant resources:
  • threading: https://pymotw.com/3/threading/index.html
  • multiprocessing: https://pymotw.com/3/multiprocessing/index.html
  • Thread sync mechanisms: https://docs.python.org/3/library/threading.html#lock-objects
  • Process sync mechanisms: https://docs.python.org/3.7/library/multiprocessing.html#synchronization-between-processes
  • Queue: https://pymotw.com/3/queue/index.html

Async.

• async & await is the way to go (v3.5 & above). However, the documentation on this topic is not good.
• A talk on this topic from its creator: https://www.youtube.com/watch?v=m28fiN9y_r8
• Other resource: https://pymotw.com/3/asyncio/

Improving Performance: PyPy

• Pypy claims to be approx. 7 times faster than CPython. It’s generally useful in case your app runs for a long time, in contrast to a Python script which just does one-off operation.
• Tread carefully, since we’re in the realm of performance optimization. I agree with taking the following steps before performing any optimization:
  • Measure app performance.
  • Profile your code.
  • Optimize.
• Also, please identify if your app’s core logic is compute-bound or I/O bound. With PyPy, compute-bound code can be optimized.

PyReverse

• Using PyReverse, you can generate UML diagrams for your app. This can be helpful in understanding & refactoring your code.
• Install pyreverse: sudo apt-get install pylint
• Usage (run this command from outside the project’s directory): pyreverse -o png -p PythonAwesomeApp PythonAwesomeApp/

• Class diagram of this app