Awesome

RSS-lambda

RSS-lambda transforms RSS feeds without RSS client lock-in

Motivation

There are RSS clients that can perform transformations on RSS feeds, e.g. only keep entries with certain keywords, or translate texts of the entries

However, using those features from the RSS clients will create RSS client lock-in that prevents you from moving to another RSS client if you desire

RSS-lambda is an application that perform transformations on the server-side instead so that you can freely move to another RSS client while keeping the transformations. It's also self-hostable so that you don't even need to rely on the official server instance!

Usage

There is an official server instance available at rss-lambda.xyz

From the web UI, you can tell it what you want to do with the RSS feed, and it will generate a RSS feed URL after your specified transformation

Transformations include

• Filter a rss feed by including entries with certain keywords in their titles
• Filter a rss feed by excluding entries with certain keywords in their titles
• Filter a rss feed by including entries with certain keywords in their contents
• Filter a rss feed by excluding entries with certain keywords in their contents
• Filter a rss feed by only including entries with image(s) in their contents
• Merge with other feeds
• (BETA) Filter a rss feed by only including entries with human image(s) in their contents
• (BETA) Filter a rss feed by only including entries with cat image(s) in their contents
• (BETA) Filter a rss feed by only including entries with dog image(s) in their contents

Self-hosting

You can use the following docker-compose.yml to self-host the application

services:
  app:
    restart: always
    ports:
      - "5000:5000"
    image: ghcr.io/sekai-soft/rss-lambda:latest

The web UI will be exposed at port 5000

The image recognition endpoints are not enabled by default. In order to enable them, you need to

1. Create a blobs folder under docker compose root folder

2. Download the yolo v3 inference model and unzip it in the blobs folder, e.g. the folder structure should look like ./blobs/yolov3/1/... after unzipping

3. Use the following docker-compose.yml file instead to run the application

services:
  app:
    restart: always
    ports:
      - "5000:5000"
    image: ghcr.io/sekai-soft/rss-lambda:latest
    volumes:
      - ./blobs:/app/blobs
      - ./cache:/app/cache
      - ./file_cache:/app/file_cache
  tfserving:
    restart: always
    ports:
      - '8501:8501'
    image: bitnami/tensorflow-serving:latest
    volumes:
      - ./blobs/yolov3:/bitnami/model-data
    environment:
      - TENSORFLOW_SERVING_MODEL_NAME=yolov3

Development

Open in VSCode, then run flask run --reload

The webapp will be available at localhost:5000

To enable image recognition endpoints

1. Create a blobs folder under project root folder

2. Download the yolo v3 inference model and unzip it in the blobs folder, e.g. the folder structure should look like ./blobs/yolov3/1/... after unzipping

3. Run docker compose -f docker-compose.tensorflow-serving.yml up in another terminal window

Run unit tests python -m unittest discover