Both scripts require setting the following environment variable:
MONGODB_URI
python gtfs.py [daysRelativeToToday]
Downloads GTFS data and loads one day's schedule into the database. Realtime data from HRT is relative to schedule data, so this data is necessary to correct interpret realtime data.
If daysRelativeToToday is not specified, it defaults to 0, which means that the schedule for the current day is loaded.
python ftp.py
Downloads realtime data from HRT, matches to scheduled data and stores results in database.
The current deployment strategy is to use a Heroku custom clock process. Below are details for deploying to AWS Lambda, where we used to run the scripts.
The scripts can be deployed to AWS Lambda using Serverless where they will run periodically.
Set environment variables in serverless.yml
Note that HRT's FTP server only allows whitelisted IPs, so you'll need to set up a VPC. Here is one guide on how to do that.
Install Serverless
npm install -g serverless
Install python requirements plugin
npm install --save serverless-python-requirements
Configure AWS credentials
Create and activate a virtual environment
virtualenv venv
source venv/bin/activate
Deploy
serverless deploy -v