serverless-twitter-bot

A Twitter bot written in Python that runs on AWS Lambda. The bot has a number of functions, called "modes", that can do the following actions:

• Like tweets from a list of users
• Retweet tweets from a list of users
• Reply with text randomly chosen from a list
• Reply with images randomly chosen from a list
• And more in serverless_twitter_bot/bot_functions

Other features:

• Easy to add new modes for different functionality or behaviour
• Bot state is saved to an S3 bucket
• Images used in replies can be stored on S3 or locally
• Mode can save their own state to persist data after each run
• Rate limiting for the bot, per mode or per user interaction
• Can also just be run locally
• Written in Python 3
• Detailed logging

You will need to create a Twitter developer account and create a twitter application.

Example bot configuration:

name: my-bot

modes:
  reply_to_friends:
    function: reply
    config:
      tweets:
        - 100% agree 💙
        - I miss you
    recipients:
      - BoJackHorseman

  like_tweets:
    function: like
    recipients:
      - NASAPersevere
      - archillect
      - horse_ebooks
    rate_limit:
      type: per_recipient
      time: 2d

  retweet:
    function: retweet
    recipients:
      - dril
      - Coldwar_Steve
    rate_limit:
      type: per_recipient
      time: 3d

  reply_to_idiots:
    function: reply
    config:
      tweets:
        - Are your parents cousins?
        - You are an embarrassment
    recipients:
      - DonaldJTrumpJr

Installation and configuration

Configuration of the modes, rate limiting and other settings is in a simple YAML file. It's easiest to just look at the example: terraform/bot-config.yaml.

To create the AWS Lambda function and all associated resources it's easiest to use the provided Terraform code in terraform.

Running

Some environment variables are required for configuration of the bot:

• BOT_CONFIG_FILE: Path to the bot configuration file
• STATE_S3_BUCKET: The name of the S3 bucket to find/store the bot state
• STATE_S3_KEY: The name of the key in the S3 bucket to find the state file
• AWS_REGION: The AWS region
• TWITTER_CONSUMER_KEY, TWITTER_CONSUMER_SECRET, TWITTER_ACCESS_TOKEN, TWITTER_ACCESS_TOKEN_SECRET: These come from your Twitter app in the Twitter Developer Portal

Locally

Install the required packages:

pip3 install -r requirements.txt

You can run it locally without AWS Lambda like this:

python3 main.py

You can also install it locally:

pip3 install .

Manually using AWS CLI

Run this:

aws lambda invoke --function-name my-bot-name out --log-type Tail --query 'LogResult' --output text |  base64 -d

Manually using AWS Console

You can run a test invocation from the AWS Lambda console tab Test, the payload doesn't matter.

On a schedule

The Terraform code has a enable_schedule variable to enable a schedule for running the AWS Lambda function.

Running tests

To run all tests, from root of repo run:

python3 -m unittest discover

Or to run a specific tests:

python3 -m unittest tests.test_bot.TestBot