playlist_search

playlist_search is a Python Flask app that helps users find which Spotify playlists a track is in.

Running

Preparation

1. Install Redis locally

2. git clone git@github.com:alexpogue/playlist_search.git

3. cd playlist_search

4. Set up config.py as follows

   SQLALCHEMY_DATABASE_URI="sqlite:///database.db"
   SQLALCHEMY_TRACK_MODIFICATIONS=False
   
   SPOTIFY_CLIENT_ID="<spotify client id>"
   SPOTIFY_CLIENT_SECRET="<spotify client secret>"
   
   CELERY_BROKER_URL="redis://127.0.0.1:6379"
   CELERY_RESULT_BACKEND="redis://127.0.0.1:6379"
   

   Note: change SQLALCHEMY_DATABASE_URI if you want to post to another database.

5. pipenv shell

6. pipenv install --ignore-pipfile

Running

1. Open two terminals
2. cd playlist_search in both
3. pipenv shell in both
4. Run in one of the terminals: celery -A celery_worker.celery worker --loglevel=info
5. To run the app, in the other terminal, either:
   • For Flask dev server: python main.py
   • For gunicorn: gunicorn --bind 0.0.0.0:5000 wsgi

Installing Wordpress plugin

1. cd wordpress_plugin
2. zip -r playlist-search-plugin.zip playlist-search-plugin
3. Navigate to the Wordpress plugins page and install it from the zip file.

Usage

1. To put the tool somewhere in wordpress, insert the shortcode [playlist-search-tool] into any Wordpress post.
2. To update the database with the particledetector and soundsofspotify playlists, navigate to the plugin settings from the main Wordpress dashboard and press the Update db button. A loading bar should appear showing the status.

Setting up as system services (for RHEL 7)

playlist_search as a system service

1. Copy etc/systemd/system/playlist_search.service in the repo to /etc/systemd/system/playlist_search.service
2. In the above file, under [Service], change the WorkingDirectory to the location the repo is cloned to.
3. chmod 644 /etc/systemd/system/playlist_search.service
4. To start, run systemctl start playlist_search
5. To enable on boot:
   1. systemctl enable playlist_search
   2. Ensure systemctl status playlist_search outputs "enabled" as in the following output:

                                                                   vvvvvvv
      Loaded: loaded (/etc/systemd/system/playlist_search.service; enabled; vendor preset: disabled)
      

celery as a system service

2. Copy etc/default/celeryd in the repo to /etc/default/celeryd
3. In the above file, change cd portion of CELERY_BIN to the location the repo is cloned to.
4. chmod 755 /etc/default/celeryd
5. To start, run service celeryd start
6. To enable on boot:
   1. chkconfig --add celeryd
   2. Ensure chkconfig --list celeryd outputs the following:

      celeryd        	0:off	1:off	2:on	3:on	4:on	5:on	6:off
      

   3. If above command returned more things listed off, try running chkconfig celeryd on and try again

Known issues

1. If there are errors in db update, the loading bar will show red, and you should press Update db again. This can happen frequently with the main fork of spotipy that we use (I don't think it refreshes the auth token or something).
2. Only press the button once at a time - each press will start a new celery worker. Eventually we might only allow one celery worker, but I haven't started that work.
3. The Download CSV link takes a second to show up since we run another HTTP request to get the results again. Instead, we can run one HTTP request and generate both the results and the CSV file using the single request. Previously we had the server process the request and send us a CSV download directly, but I don't think that's possible through the Wordpress AJAX proxy.