This project pulls the data from DataSF and provides it as a searchable resource online at friscofoodtrucks.com.
On startup and every 24 hours thereafter, the data is pulled as a CSV and since there isn't much data at all, stored in memory, ready to be served.
There are two codebases in this monorepo.
- "api" holds the Phoenix app that gathers, transforms, and exposes the data.
- "web" is the React app that displays the data.
There is only one endpoint on the API that is served by the index action on a trucks controller that serves up the list, taking a filter param - a single string - to use to narrow down the list.
The core of the API is a GenServer that keeps the truck data in state and refreshs it daily. There are also two services to fetch and parse the data. One uses a behaviour to help out with testing.
We only store approved food trucks that have food items and locations listed, filtering out anything else.
There is no database. A future improvement might be to also store the data in a database for use if the DataSF service isn't available, giving us time to figure out what's up without the site being down.
There's also not a lot of error handling to maintain the stability of the site if things go wrong. Having the data backed up in a database would help with that. If the process crashed - probably due to an issue with fetching or parsing the data, such as if their service were down or they changed the fields - we should fall back to pulling the stored data into memory when the process is restarted.
I didn't do much with the client other than throwing it together to actually show the data, test it out, and make it live.
Thanks for looking!