courseLoader.js Data adapter

Fetches the full course catalog. Tries the bundled static file first. Falls back to a live API if the local file is unavailable. Returns raw JSON — normalization happens in PlannerContext via normalizeCourse().

Parent

data/

Called by

PlannerContext.jsx (once, on mount)

Primary source

public/all-courses.json (6.5 MB, bundled)

Fallback source

husker.vercel.app/courses/all (live API)

Exported function

fetchCourses() → Promise<RawCourse[]>

Attempts to load courses in this order:

Fetch {BASE_URL}all-courses.json (bundled in public/)
If that fails (404, network error), fetch from https://husker.vercel.app/courses/all

Returns the raw JSON array without normalization. PlannerContext calls normalizeCourse(raw) for each element after receiving the array.

Returns: Promise<RawCourse[]> — raw catalog entries

Why bundle + fallback?

Bundling all-courses.json ensures the app works offline and loads fast. The fallback API exists for development scenarios where the bundled file hasn't been updated yet, or for quick testing without a full build.

The BASE_URL is read from import.meta.env.BASE_URL (Vite-injected) so paths resolve correctly on both localhost (/) and GitHub Pages (/nu-map/).

Updating the course catalog

# Full update workflow
npm run scrape         # fetch from catalog.northeastern.edu (~20 min for full rotation)
npm run patch          # apply YAML corrections from data/patches/
npm run nupath         # merge NUPath attribute data
git add public/all-courses.json
git commit -m "data: update course catalog"