https://github.com/bryanoliveira/fcm-notifier

A python module to notify about ML model training statuses with Firebase Cloud Messaging.
https://github.com/bryanoliveira/fcm-notifier

Last synced: 7 months ago
JSON representation

A python module to notify about ML model training statuses with Firebase Cloud Messaging.

• Host: GitHub
• URL: https://github.com/bryanoliveira/fcm-notifier
• Owner: bryanoliveira
• License: mit
• Created: 2020-01-18T16:57:32.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2020-01-26T17:01:59.000Z (over 5 years ago)
• Last Synced: 2025-01-03T01:54:44.188Z (9 months ago)
• Language: Python
• Size: 190 KB
• Stars: 2
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Firebase Cloud Messaging Notifier

A python module to notify about Machine Learning model training statuses with Firebase Cloud Messaging. It is best used with [Flutter Firebase Cloud Messaging Receiver](https://github.com/bryanlincoln/flutter-fcm-receiver), a Flutter app I made to export device's tokens and receive notifications.

  

> App's view (left), notification center (center) and smartband (right). This is the way I use it.

The main motivation is that monitoring machine learning models' training status can be time consuming (specially when SSH-ing from the phone). Using this, monitoring becomes passive.

## Requirements

-   Python 3 and `pyfcm` module (`pip install pyfcm`)

-   A Firebase Cloud Messaging project token (see https://firebase.google.com/docs/cloud-messaging)

-   A device token (to receive notifications). You may use [Flutter Firebase Cloud Messaging Receiver](https://github.com/bryanlincoln/flutter-fcm-receiver) for this.

## Usage

-   Install:

    `pip install git+https://github.com/bryanlincoln/fcm-notifier`

-   Set up the environment variables `FCM_API_KEY` and `FCM_DEVICE_TOKEN` with your Firebase project token and your device token. Alternatively, you may pass this values directly to the FCMNotifier constructor (in this order), but it's not recommended for security reasons.

-   In your code, import it with

    `from fcm_notifier import FCMNotifier`

-   Instantiate it:

    `notifier = FCMNotifier()`

-   Push a notification!

    `notifier.notify(title='Hello world')`

## Additional Info

You may use the optional argument `message` to set up a notification body for method `notify`. Aditional arguments are accepted too. They'll be arranged into the notification's body. Examples:

-   Calling `notify(title='Notification Title', message='Notification Body')` generates a notification like:

| Notification Title |

| ------------------ |

| Notification Body  |

-   Calling `notify(Loss=123, Accuracy=0.9)` generates a notification like:

| Train Status  |

| ------------- |

| Loss: 123     |

| Accuracy: 0.9 |

## Contributing

This module (and [app](https://github.com/bryanlincoln/flutter-fcm-receiver)) are far from complete. There are some ways I imagine they could be better:

-   Saving historic data, when new notifications arise

-   Plotting charts with historic data

-   Automatic device token synchronization between phone and reporter machine

-   Easier project configuration, using some service other than Firebase.

Fell free to fork this project and improve it for your needs.