To demonstrate the data streaming pipelines by implementing the modern data stack.

• Streaming message data:
  bot -> kafka -> druid -> superset
• Detect hate speech in the Discord Server

• Language: Python 3.9
• Data Source: Discord Bot
• Storage Layer: PostgreSQL
• Streaming Layer: Apache Kafka
• Analytics Layer: Apache Druid
• Visualization Layer: Apache Superset

Due to the scope and purpose of this project, everything will be running in docker containers via simple make commands. Thus, it is essential to have make, docker and docker-compose properly installed.

• make
• docker & docker-compose
• Platform: Mac or Linux (for automatic setup to work)
• Minimum 12GB memory allocated for Docker

• Step 1.
  make setup

• Step 2.1.
  Fill in your Discord Bot Token in .env file

• Step 2.2.

  • Add TALISMAN_ENABLED=False in the superset service as an extra environment variable in Superset's docker-compose.yml
  • Add the below networks block in Superset's docker-compose.yml

  networks:
      discord:
          name: discord_ai_bot_network
          external: true

  • Add the below for all services in Superset's docker-compose.yml

  networks:
    - discord

  • Comment out or remove the below in superset-tests-worker service in Superset's docker-compose.yml

  network_mode: host

• Step 3.
  make kafka
  Note 1: This step will create a messages topic by default
  Note 2: Please make sure all containers are up and running before moving to the next step.

• Step 4.
  make bot
  Note 1: This step will instantiate the bot and make the connection to kafka

• Step 5.
  Install the Bot on your Discord Server and type some texts in the Discord channel for testing. Monitoring the Discord Bot log to see if there are any errors.

• Step 7.
  make druid
  Note 1: kafka host: broker:29092
  Note 2: kafka topic: messages

• Step 8.
  make superset
  Note 1: Druid connection URL: druid://<druid_router_ip>:8888/druid/v2/sql

TODO

• Doc: discord.py API Documentation
• stack overflow: bot cogs usage

• Doc: confluent-kafka API Documentation
• Youtube: Listener and Advertised Listener Explaination
• Github: Confluent Kafka docker-compose Template