/politweets

Analysis of U.S. Congressional Tweets

Primary LanguageJavaScript

Politweets

Better name coming soon. 🤷🏻‍♀️

Development

To run the development server, you must have Docker installed.

From the root of the repo, run:

docker-compose up

This will spin up Postgres, Webpack, and Django services.

You can access the application from your web browser at: http://localhost:8000/

You can connect to the PostgreSQL database at postgres://politweets:politweets@localhost:15432/politweets

Getting the Data

To get data, you will need access to both the ProPublica Congress API and the Twitter API.

Once you have your credentials, you will need an .env file in the root of the repository with them. This file is ignored by git to prevent you from publishing your secret keys. You can either create an .env file with this format:

PROPUBLICA_API_KEY='XXXXXXXXXX'
PROPUBLICA_API_BASE='https://api.propublica.org/congress/v1/'

TWITTER_CONSUMER_KEY='XXXXXXXXXX'
TWITTER_CONSUMER_SECRET='XXXXXXXXXX'
TWITTER_ACCESS_TOKEN='XXXXXXXXXX'
TWITTER_ACCESS_TOKEN_SECRET='XXXXXXXXXX'

Or you can run this script to run through the automated setup wizard:

scripts/setup

You may need to restart docker-compose after creating this file.

With the API credentials in place, run these commands to synchronize data:

# Sync members of congress to the database
docker-compose exec django pipenv run ./manage.py sync_members
# Sync new tweets to the database
docker-compose exec django pipenv run ./manage.py sync_tweets

Production Build

To build production assets run:

cd client
npm install
npm run build

To force Django to use the production assets, create a file politweetalyzer/settings/local.py:

import os


# Build paths inside the project like this: os.path.join(BASE_DIR, ...)
BASE_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))


WEBPACK_LOADER = {
    'DEFAULT': {
            'BUNDLE_DIR_NAME': '',
            'STATS_FILE': os.path.join(BASE_DIR, 'client', 'dist', 'webpack-stats.build.json'),
        }
}

Then run the Django server without the Webpack container:

docker-compose up django

Deployment

We use Heroku's container-based deployment. To run deployment commands, you must have Heroku CLI installed and be logged in to Heroku's container registry. Then run:

scripts/deploy

This script will:

  1. Clear precompiled javascript and CSS out of /client/dist
  2. Reinstall node dependencies and run webpack to compile production-ready javascript and Congress
  3. Push a docker image to the Heroku registry
  4. Release that docker image to the Heroku app