waka-next

This is a rewrite of Waka. The notable difference is that it now uses sqlite on both client and server to speed up queries and run offline.

https://next.waka.app

Setup

npm ci
npm run dev for dev
npm run build to build
npm run preview for a preview server

Configuration

The orchestrator will start without any configuration and serve the sample configuration. However, if you do not set up AWS credentials (S3), imports on the server will not work but you can still use the Waka dev tools in the client to test imports locally. If you're not using S3 (e.g Cloudflare R2) you can set AWS_S3_ENDPOINT.

To configure the orchestrator, set WAKA_ORCHESTRATOR_CONFIG to something like:

database:
  region: ap-southeast-2
  bucketName: your-s3-bucket-name
  publicUrl: https://your-bucket-public-url.com/regions
regions:
  nz-akl:
    name: 'Auckland, New Zealand'
    gtfsZipUrl: https://gtfs.at.govt.nz/gtfs.zip
    shouldCache: false # defaults to true. false means the orchestrator won't load the region on launch.
  zz-sample1:
    name: Sample Region
    gtfsZipUrl: https://next.waka.app/sample-feed-1.zip
    gtfsZipDisableEtag: true # set to true if you want to force an import every time
    gtfsZipDisableHead: true # set to true if the upstream server doesn't suport http HEAD
    gtfsZipHeaders:
      Authorization: SECRET.whatever # you can use WAKA_ORCHESTRATOR_SECRETS to substitute secrets
    gtfsTidyOptions: false # set to false if gtfstidy is not installed

You will also need to set a WAKA_ORCHESTRATOR_ACCESS_TOKEN to the administrator bearer token that you want to use. Multiple tokens are not currently supported.

The WAKA_ORCHESTRATOR_CACHE_PERIOD variable can also be set to a number in milliseconds to reduce the number of reads from S3.

The WAKA_ORCHESTRATOR_NO_CACHE variable can be used to set shouldCache to false globally.

Scheduling Updates

fly machine run alpine --entrypoint "sh -c 'apk add --no-cache curl && curl -X POST -H \"Authorization: Bearer \$WAKA_ORCHESTRATOR_ACCESS_TOKEN\" https://waka-next-orchestrator.fly.dev:5000/admin/schedule-update'" --schedule hourly

Application Structure

There's three apps in this repository.

lib: The library for importing, manipulating, and querying GTFS. This will eventually be published as a NPM module that runs on both client and server.
apps/orchestrator: The server that downloads, parses GTFS files and serves them to users. It also provides a HTTP API so queries can be run without downloading the whole database.
apps/web: The new web frontend. It can download and query the sqlite database directly, or use server side rendering to grab the data from the orchestrator's HTTP API.

Notes

Data Loading

This is designed to eventually run fully offline - this is how static GTFS data is loaded in apps/web:

If the sqlite db is already downloaded to the client, query from that (provider: client)
If it's not, query the data via HTTP from apps/orchestrator - this is also run when server side rendering (provider: server)
If the HTTP request fails, wait until the sqlite db is downloaded, and then query on that (provider: static)
If that fails, the request is rejected