Skip to content

Latest commit

 

History

History
190 lines (144 loc) · 8.01 KB

python.md

File metadata and controls

190 lines (144 loc) · 8.01 KB

Python

PEP - Python Enhancement Proposal
Documents written by community of python language developers describing recommended conventions about writing maintainable python code. Source here

  1. Installing Python
  2. Using pip - installing 3rd party packages
  3. Python project setup
    • repository structure
    • virtual environments
    • dependencies
  4. Jupyter notebooks
  5. Naming style

Installing Python

Official downloads page

Custom installation

  • pip
  • python test suite
  • create shortcuts for applications
  • debugging symbols and debug binaries
  • ! customize install location
    choose something simple like C:\python, C:\python310, C:\pythons\py37, C:\pythons\py310

If you install python into a folder at C:\python, you will have to put this path into the Path environment variable. Path is just a list of paths to executable files you want to run easily without having to specify the full path to the executable that actually starts your application. For example, you can open up the command prompt and type notepad.exe to run the notepad program because by default notepad.exe is stored at C:\Windows (for example) which is included in the Path environment variable. Notepad.exe runs a GUI program, python.exe runs a CLI program, that executes your python scripts (.py files).

The python installer offers to modify the Path variable for you. If you want to use multiple versions of python, you can rename the installation directory and the .exe file to some appropriate name e.g. python310.exe, python38.exe. We use (or try to do so) one version and that's python 3.8 that is supported until 2024.

Using pip

pip = package installer for python. Example usage:

  • install latest version of numpy package: python -m pip install numpy
  • check that numpy is installed, show information about it: python -m pip show numpy
  • remove numpy: python -m pip uninstall numpy

-m flag = run standard library module as a script. This syntax may seem more verbose, but it's clearly visible which python interpreter is used. So if I had a folder C:\python310 (added to Path) containing python310.exe, I could call python310 -m pip [...].

The official documentation of pip can be found here. Note on unix vs windows syntax used in the official docs: In Windows, you have the option to install something called python launcher which is a tool used to manage different version of python on one computer and calling pip directly. But it's not necessary to run python programs.

Project setup

Repository structure

.
+-- .gitignore
+-- README.md
+-- src
|   +-- main.py
+-- tests
|	+-- unit
|	|   +-- main_test.py
+-- samples
|   +-- sample_request.json
+-- requirements.txt
+-- env

Virtual environment

Part of the python standard library is a package called venv: docs
To create a virtual environment:

python -m venv my_environment

This creates a new directory structure with the root folder called my_environment. The goal is to create a lightweight isolated environment where you can install packages that won't conflict with the globally installed packages. Three key components:

  1. copy of the python executable
    (this is NOT 'the whole python' it's just a copy of the .exe file that runs the whole python program)
  2. pyenv.cfg file
    Config file for this environment.
    Simple key = value definitions: 
    home = C:\python310
include-system-site-packages = false
version = 3.10.4
    The home settings point to the base python interpreter used in this environment. The standard library modules are taken from there.
  3. site-packages directory
    The packages specific to this environment.

To activate the environment called my_environment, open cmd in the folder where it's located and:

call my_environment/scripts/activate.bat

Among other things, the activation means that it puts whatever .exe is in the scripts folder to the beginning of Path. So if you used python310.exe to create this environment, and the pyenv.cfq file contains path to this file, you can rename the scripts/python310.exe to scripts/python.exe for convenience because what matters is the path under home in pyenv.cfg.

Note: To see what is in Path in cmd: echo %PATH%.

Dependencies

3rd party packages used in the project and their versions are specified in the requirements.txt file. You can specify version of the dependency as:

  • Specific version: some_package==1.20.1
  • Range: some_package>=1.0,<=2.0

To install all requirements at once: python -m pip install -r requirements.txt
explanation:

  • python -m pip run pip as a script with the specific version that python points to (PATH...)
  • install is the command supplied to pip, -r flag should be followed by filename where dependencies are listed, it means install everything from this file.

todo ? wheels - packages that use C-extensions are pre-compiled for specific architecture

Note on IntelliJ

File -> Project Structure -> Project Settings This is where you configure the SDK (general term meaning Software Development Kit) this is either a virtual environment or the global interpreter. For some reason it works better to create environment from cmd and then add it to idea as an existing SDK - if you use more python versions with something like python310.exe.

Jupyter Notebooks

Jupyter notebook has the extension .ipynb and the file contains both code and other elements (charts, latex math symbols, markdown formatted text, etc...) Kernel is some language's interpreter that runs the human written code in the notebook. For python, we have ipython kernel - installed automatically with the python package jupyter.

Jupyter can be listed in requirements.txt and when the server is started from inside the environment, a kernel that is available only in this environment is created. The Jupyter project has more frontends available for notebook management, or we can use the (paid) plugin in intellij.

Other option is to use the static directory on disk where jupyter can store kernels that should be globally accessible.
Example - I have python (3.8) and python310 (3.10) installed, and I want to add python 3.10 to available kernels.

  • python310 -m pip install jupyter - install jupyter in the 'global environment' that is the 3.10 python
  • python310 -m ipykernel install --name "python310" --user - ipykernel is a tool that comes with jupyter for managing python kernels
  • python -m jupyter kernelspec list - output of this will be the static list of kernels, meaning output would be the same if called with python310.

Naming style

snake_case

Lowercase letters, words are separated by underscore. Usage: functions, variables, methods, names of modules (module = file with .py extension) Special in Diribet - also used for package names even though official recommendation is something like mypackage instead of my_package.

PascalCase

Each word is capitalized, words are not separated. Use when naming classes.

SCREAMING_SNAKE_CASE

Use exclusively for variables that are intended to be constants. The value assigned to this variable does not change. Python doesn't provide an easy way to "freeze" a variable, it is possible though usually we rely on the naming style. Constants are usually declared in a specific file named 'constants'.

Miscellaneous

Imports
At the top of the file, first standard library imports, then 3rd party packages, and then your local modules.

Data for analysis
Shared on Google drive. Data that represent a sample (typically sample of a json request for web servers) or data for testing the software are included in the repository.

None as a falsy value
Explicit check for None for variables that are not boolean. if my_variable is not None: vs if not my_variable