notify-me-senpy 0.5.1

Simple python tracker

What is it?

Notify Me SenPy is a productivity tool that let's you track your Python scripts' execution from your smartphone and receive a notification whenever the execution reaches certain points.

It is composed of a Python package and a mobile app. The package is used within your Python script to specify which job you want to track and at which points you want to receive notifications. The application is where you can check on your job progress and see the notification history.

Usage

You can send notifications from a script to your app using notify_me

from senpy import notify_me

notify_me("Great job!")

You can track your jobs using the ntm python with statement:

from senpy import ntm
from time import sleep 

with ntm(range(10)) as iterator:
    for i, item in enumerate(iterator):
        sleep(5)

Use cases

1 - Hyperparameter fine-tuning :

Here we try to find the appropriate value for the number of clusters. This computation can be time-consuming and cumbersome to track manually.

from sklearn.cluster import KMeans
from senpy import notify_me, ntm


# random dataset generated by sklearn
X = make_blobs(n_samples=300_000, n_features=100)[0]

inertias = []

with ntm(range(1, 100, 10)) as nb_clusters:
    # Using ntm, SenPy keeps us updated about the current iteration
    for i in nb_clusters:
        km = KMeans(n_clusters=i, init='random')
        km.fit(X)
        inertia.append(km.intertia_)


# At the end of the computation we get notified by SenPy
notify_me(f"KMeans fitted, inertia went from {inertias[0]} to , {inertias[-1]}")

2 - Use notify-me and ntm at the same time:

from senpy import notify_me, ntm
from time import sleep

IDLE_TIME = 3 * 60 * 1000 ## 3 minutes

def very_long_computation():
    time.sleep(IDLE_TIME)
    
    
with ntm(range(0, 100)) as values:
    for value in values:
        very_long_computation()
        if(value % 10 == 0):
            # Get notified whenever something interesting happens
            notify_me(f"We are at iteration {value}")


notify_me("The computation is done !")

Here are a few more examples that illustrate how Notify Me SenPy can be used:

• PyTorch model fitting
• Password bruteforce attack
• Real time Bitcoin updates
• Tensorflow Keras Callbacks

API

ntm

All parameters are optional except iterable.

Parameter	Type	Description
iterable	Iterable	The iterable that ntm is going to wrap. if iterable is a generator then you have to specify the length parameter
name	String	Name of the job. Default is "Job #<job number>".
current_iteration	Integer	Initial iteration number of the job. Default is 0.
update_period	Integer	Time in seconds between each update sent to the server. Default is 5 seconds. NB: the server enforces API request rate limits.
length	Integer	The total number of iterations. Default is len(iterable), as a consequence this argument is required when passing a generator as iterable

notify_me

Parameter	Type	Description
content	String	The content of the notification sent to the app.

Installation

1. Package

First install the Notify Me SenPy Python package

pip install notify-me-senpy

Create an account from the command line

senpy register
>>> ? Username: your_username
>>> ? Password: *********
>>> ? Password confirmation: *********

The existing commands are

senpy {register|login|logout|change_password|delete_account}

You are done with the package setup! 🎉

2. App

Download the Notify Me SenPy app from the store.

Once the app installed, open the app and wait until the configuration is done. You should receive a notification confirming that the configuration has been successfully completed. Otherwise, check your connection, try again or try re-installing the app.

Once done, you simply have to log into your account in the app to complete the installation.

To contribute

Architecture

The directory senpy is currently built as follows :

├───senpy # Folder containing the source code
    ├───account_manager.py # authentication system
    ├───cli.py # cli utility functions
    ├───jobs.py # ntm with statement source code
    ├───notifications.py # notify_me source code
    ├───request_utils.py # HTTP methods
    └───user_token.py # Auth-token-related functions
├───tests # Unittests for each function
    ├───notify_me.py 
    └───ntm.py
└───setup.py # Package configuration file

Development

To get started, you can create a virtual environnement and install the packages listed in requirements.txt:

git clone https://github.com/magantoine/senpy-package
pip install virtualenv
virtualenv venv
source venv/bin/activate
cd senpy-package
pip install -r requirements.txt

Run the following command from the top folder:

pip install --editable .

You only need to run this command once. What it does is that it installs a package, called senpy, whose implementation is in the current folder (and not in the usual site-packages), so every subsequent code modifications while be taken into account.