Observer Design Pattern in Python

Observer Design Pattern in Python

Tariq Rafid's photo
Tariq Rafid
ยทSep 19, 2022ยท

5 min read

Play this article

Table of contents

  • What is Observer Pattern? ๐Ÿ”ญ
  • Implementation โš™
  • Code โŒจ
  • Conclusion

Design Patterns are commonly tested solutions for problems developers face, one of the most known design patterns out there is the Observer Pattern, it's pretty common with apps that have GUI. for example in JavaScript observers are offered by the browser API to easily detect changes in the state of the application.

What is Observer Pattern? ๐Ÿ”ญ

The observer pattern is when you have a Subject object and many Observers objects with a one-to-many relation, each Observer is notified whenever the Subject state changes.

All Observers are able to unsubscribe and no longer get the Subject updates

Is it the same as Pub/Sub Pattern? ๐Ÿค”

No, the Publishers/Subscribers Pattern is similar but not the exact thing.

  • Observer Pattern: observers are aware of the Subject, and the Subject maintains a list of Observers.

  • Pub/Sub: Publishers and Subscribers know nothing about each other and communicate using a message passing mechanism which makes it asynchronous.

Implementation โš™


To make it easier to understand the implementation, we will design a simple CPU temperature monitor. Our design will have two abstract classes called Subject and Observer, whereCpuSensor will inherit the Subject class while the CpuController and TempDisplay will inherit the Observer class.

TempDisplay is responsible for displaying temperature, while CpuController will perform dummy actions upon the last temperature read.


First, import ABC and abstractmethod from the built in library abc which allows us to do abstract classes

from abc import ABC, abstractmethod

We will create the class, and make it extends ABC

class Subject(ABC):

in the constructor, we will instantiate a private variable called _observers with an empty array to track the observers, adding the abstractmethod decorator to prevent Subject to be instiniated.

class Subject(ABC):
    def __init__(self) -> None:
        self._observers = []

After that, the register and unregister methods will be added, so the observers can subscribe and unsubscribe from the Subject

class Subject(ABC):
    def __init__(self) -> None:
        self._observers = []

    def register(self, observer):

    def remove(self, observer):

Then, the notify function will be declared, which is responsible for updating observers with the new data, for now, we will just leave it empty.

class Subject(ABC):
    def __init__(self) -> None:
        self._observers = []

    def register(self, observer):

    def remove(self, observer):

    def notify(self):
        for observer in self._observers:


Just like the Subject, the Observer class should extends ABC

class Observer(ABC):

Adding two abstract methods, on_change and update, both have abstractmethod decorator.

  • update: inject data to the observer and call on_change
  • on_change: empty function
class Observer(ABC):
    # data is array of tuples of (dataName, Value)
    def update(self, data):
        for (name, val) in data:
            setattr(self, name, val)


    def on_change(self):

Connect Observer with Subject โš”

Let's get back to the Subject class and complete notify logic.

Since we know that each observer has the update function, we will call that for each observer, but as you know the update function receives data in an array, to do that, we will have a private function called _get_vars which will arrange all public variables of the Subject in array.

class Subject(ABC):
# ...
  def _get_vars(self):
      all_vars = self.__dict__.keys() # get all keys
      # filters all vars that don't start '_'
      # and returns them as an array of tuple that has the name and the value
      return [(key, self.__dict__[key]) for key in all_vars if not key.startswith("_")]
  def notify(self):
          for observer in self._observers:
              observer.update(observer, self._get_vars())


In the CpuSensor class, we will:

  • Extend Subject
  • Call the parent constructor
  • Implement change_measurements
class CpuSensor(Subject):

    # We must override the constructor, because of `abstractmethod` used in the parent 
    def __init__(self):

    def change_measurements(self, temp):
        self.temp = temp

CpuController and TempDisplay

Because we want CpuController and TempDisplay to be observers, we will have them extends Observer, we will also override the on_change function on both observers, so the TempDisplay will print temperature, where the CpuController will print "panic" if the temperature gets more than 90ยฐ.

class CpuController(Observer):
    def on_change(self):
        print("CpuController: ", end='')
        if self.temp > 90:
              print("Everything under control")

class TempDisplay(Observer):
    def on_change(self):
        print("Display: ", end='')
        print(f'{self.temp} Deg')

Driver ๐ŸŽ

It's time now to set up a driver and test our CPU monitor ๐Ÿ˜

We will add a few built-in functions to help us.

from random import randint
from time import sleep


cpu_sensor = CpuSensor() 

for i in range(10000):
    cpu_sensor.change_measurements(randint(50, 110))


CpuController: Everything under control
Display: 56 Deg

CpuController: Everything under control
Display: 74 Deg

CpuController: Panic!
Display: 97 Deg

Code โŒจ

You can check all the of this project on this repo


One drawback of this pattern is that it can lead to a memory leak if the observers were not removed from the Subject class, because it holds a strong reference to the observers, for instance, if these observers were UI components that are no longer appears on the screen but still registered in some Subject, the Subject will keep re-rendering these invisible components and waste CPU cycles on them, this issue known as Lapsed Listener Problem

That was the Observer pattern, with a practical example you can build on top of it, if you are interested in reading real CPU temperatures you can check this cool cross-platform package pyspectator

I'm planning to cover the most common design patterns. if you are interested, you can follow me to get notified of any new articles I publish in the future or check the below list of my published articles about design patterns:

Share this