Mix pairs of song lyrics to form new lyrics.

Songs can be picked either by the user or randomly by the application.

Development

Run tests with pipenv run pytest -v

You can enable console print by running pipenv run pytest -s

Code coverage: pipenv run pytest -v --cov=. --cov-report html, then access the coverage report at htmlcov/index.html

Skip slow tests: pipenv run pytest --without-slow-integration (see https://pypi.org/project/pytest-integration/).

Runtime

Standalone scripts

Mix lyrics parsing artists from free text

pipenv run python mixlyrics_parse_free_text.py 'Madonna, Slayer'

REST API

Implemented using Flask

See: lyrics_mixer/rest_api.py and lyrics_mixer/rest_api.py.

Run locally with:

pipenv shell
gunicorn lyrics_mixer.rest_api:app

Endpoints

HTTP GET http://localhost:8000/: gives status

HTTP GET http://localhost:8000/mix/random: mix two random songs

HTTP GET http://localhost:8000/mix/artists/<artist1>/<artist2>: mix random lyrics from two artists

HTTP GET http://localhost:8000/mix/songs/<artist1>/<title1>/<artist2>/<title2>: mix two specific lyrics

Twitter Bot

Implemented using Tweepy.

Deployment

Environment variables

Environment variables for storing auth tokens:

LYRICS_MIXER_TWITTER_CONSUMER_KEY

LYRICS_MIXER_TWITTER_CONSUMER_SECRET

LYRICS_MIXER_TWITTER_ACCESS_TOKEN

LYRICS_MIXER_TWITTER_ACCESS_TOKEN_SECRET

Environment variable for the database URL:

LYRICS_MIXER_DATABASE_URL

Docker

A docker-compose.yml file is available for running the PostgreSQL and Twitter Bot Docker containers.

By default, docker-compose will look for an .env file in this directory with the definitions of the environment variables listed above.

To build the Docker image of the Twitter bot:

docker-compose build

To run the containers:

docker-compose up

To run the containers in the background and support host restart:

docker-compose up -d

Heroku

The file Procfile describes both the REST API app and the Twitter bot (as a worker).

Database setup

Enable PostgreSQL add-on on Heroku dashboard.

twitter_bot.py creates tables on startup.

Verify:

heroku pg:info

should show 1 table (streamcursor table)

Keep awake Heroku instance

Keep awake Heroku instance by running a schedule job.

Cron

Every 30 minutes except between 2 am and 8 am, since Heroku force sleep free instances 6 hours a day:

0/30 0-2,8-23 * * * /usr/bin/curl https://lyricsmixer.herokuapp.com > /tmp/lyricsmixer-ping.log

Systemd Timer

Install unit provided in the heroku directory: heroku_lyricsmixer_ping.timer, heroku_lyricsmixer_ping.service

cd heroku
sudo cp heroku_lyricsmixer_ping.* /etc/systemd/system
sudo systemctl enable heroku_lyricsmixer_ping.timer
sudo systemctl start heroku_lyricsmixer_ping.service

Check timer is installed:

sudo systemctl list-timers --all