Here is another post in PPL series. In this post I will try to explain "Clean Code" concept and some tips on how to apply it.
What is Clean Code?
Most of clean code's popularity in software engineering discipline can be attributed to Robert C. Martin's Clean Code: A Handbook of Agile Software Craftsmanship. This is the de facto resource material you should be reading to learn clean code instead of this post (lol). However, the idea behind clean code might have been around well before Uncle Bob's book. If I have to take an example the classic KISS principle, "Keep It Simple, Stupid", comes to mind.
Enough with the chitchat. This is the best explanation I can find about clean code.
Clean code is code that is easy to understand and easy to change.Carl Vuorinen
I couldn't agree more with this one sentence summary of what clean code is. The horror of incomprehensible black box code that refuses to change has been subject to many programming creepypastas like this one sample below.
Whether such thing is genuine or not, you get the point. This is the very problem clean code tries to solve. On another note, you can also check this goodread's compilation of quotes from the Clean Code book. The quotes sound poetic and have some of the most powerful clean code messages. They would certainly make good wallpaper material wink. Below is one example. You're welcome.
One difference between a smart programmer and a professional programmer is that the professional understands that clarity is king. Professionals use their powers for good and write code that others can understand.Robert C. Martin, Clean Code: A Handbook of Agile Software Craftsmanship
Applying Clean Code
Thanks. But, that still doesn't explain how to apply clean code. Of course you better read the full guides on Uncle Bob's book rather than having me half-assedly paraphrase them for you. Alternatively, I find a quick summary that can help you navigate different topics on clean code. However, for the sake of completeness I will also demonstrate how we apply some of clean code concepts to our project.
There is a time in our development phase where we need to move our backend app's enviroment multiple times. Sure, its 2020 and container technology easily handles some packaging problemes. However, its still a too steep of a learning curve for our teammates. So we still need to handle these settings on project level. Thus, our Django's settings module also grows larger as we need to account for different scenarios such as local development, staging, and production enviroment.
This one is quite simple.
if DEBUG: ALLOWED_HOSTS = ['*'] else: ALLOWED_HOSTS = os.getenv('ALLOWED_HOSTS', 'localhost').split(';')
Mutating state, anybody?
if DEBUG: INSTALLED_APPS.extend([ 'django_extensions', ])
This one is quite dangerous if not properly configured on the production
SECRET_KEY security warning will be silenced by the default value.
SECRET_KEY = os.getenv('SECRET_KEY', '__secret_high_entropy_random_bytes__')
It's apparent that, so many if/else over our settings module became too repetitive and not easy to change. Its not clean. As one of the clean code design rules says:
Prefer polymorphism to if/else or switch/case.
Thus, we need to refactor our settings module.
Thankfully, this is a common problem for many people using Django and their
solution is to use multiple settings files. The
settings.py file is removed
and instead we have settings directory under our project directory.
$ tree dblood dblood ├── asgi.py ├── __init__.py ├── settings │ ├── common.py │ ├── production.py │ └── staging.py ├── urls.py └── wsgi.py
This way we can define things separately and even reuse some codes defined in
common.py. Here is some lines from
from .common import * # SECURITY WARNING: keep the secret key used in production secret! SECRET_KEY = os.getenv('SECRET_KEY') # SECURITY WARNING: don't run with debug turned on in production! DEBUG = os.getenv('DEBUG', 'True') != 'False' ALLOWED_HOSTS = os.getenv('ALLOWED_HOSTS', 'localhost').split(';')
This one comes from the
from .production import * SECRET_KEY = os.getenv('SECRET_KEY', '__secret_high_entropy_random_bytes__') DEBUG = True ALLOWED_HOSTS = ['*'] # Application definition INSTALLED_APPS.extend([ 'django_extensions', ])
Things already looks better. However, there are also some changes needed on
wsgi.py. This is required to let know both the
./manage.py runserver command and the wsgi server where to look for Django's settings
diff --git a/dblood/wsgi.py b/dblood/wsgi.py --- dblood/wsgi.py +++ dblood/wsgi.py @@ -11,6 +11,6 @@ import os from django.core.wsgi import get_wsgi_application -os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'dblood.settings') +os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'dblood.settings.production') application = get_wsgi_application() diff --git a/manage.py b/manage.py --- manage.py +++ manage.py @@ -5,7 +5,7 @@ import sys def main(): - os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'dblood.settings') + os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'dblood.settings.staging') try: from django.core.management import execute_from_command_line except ImportError as exc:
These changes although seemed small, they have already saved us quite some time while we moved our backend app to new enviroment, now behind a troublesome proxy wall.
Linter and Formatter
Besides Uncle Bob's Clean Code teaching, there are also ways you can use to help applying some clean code principle easily. Developer often employs linter and formatter to catch those readability issues early. So, what are they?
lint, or a linter, is a tool to analyze source code to flag programming errors, bugs, stylistic errors, and suspicious constructs.
Yup, that one is a blatant ripoff from Lint wikipedia page. Meanwhile a formatter, or otherwise called prettier, or beautifier is used to automatically format code to some set of rules. I personally think calling them "formatter" is more correct as they are mainly used to avoid pointless discussion about style at all. They give you results that are sensible enough to be easily read while not trying hard to be pretty.
For example, our backend Django app uses flake8 linter with django plugin and no formatter. We use flake8 because Django Framework itself uses it. The additional Django plugin only a few but usefull rules that understand Django's specific semantic. We ditch formatter because having Lint Test job on our CI is enough for now to catch the occasional rule violations.
As an example I added an empty model called
class FooModel(models.Model): pass
Here is the flake8 output.
(venv) $ flake8 ./main/models.py:100:1: DJ08 __str__ method should be present in all db models ./main/models.py:100:1: DJ09 Model: FooModel must define class Meta
Yup, that's all about Clean Code concept and tips I have.