MARISSA MOLEJON | OCT2025-1 | Module-Legacy-Code | Sprint 1 | add unfollow functionality - #6

Summary

adds the ability to unfollow a user. Following existed;
there was no way to reverse it.

Investigation first

Before writing anything, searched the codebase for existing
"unfollow" references:

This found apiService.unfollowUser(username) already fully written
in front-end/lib/api.mjs, POSTing to /unfollow/<username> — but
no backend route existed, and no UI element called it. This
meant the actual scope of this issue was smaller than it first
appeared: build the missing backend endpoint, and wire up a button —
no new frontend API logic needed.

Backend

• unfollow(follower, followee) in follows.py — a DELETE,
  naturally idempotent (no UniqueViolation handling needed, unlike
  follow(), since deleting a row that doesn't exist is a no-op).
• do_unfollow(username) in endpoints.py — takes username from
  the URL path (matching the frontend's existing call and the
  pattern used by other_profile). No request body is sent, so no
  field validation applies.
• POST /unfollow/<username> route registered in main.py.

Frontend

• Added an "Unfollow" button to the main profile template (not the
  "Who to follow" suggestion chips — suggested users are, by
  definition, not already followed, so an unfollow button there
  would never apply).
• createProfile() shows the Unfollow button exactly when
  is_following is true — the inverse of the Follow button's
  existing visibility logic, making the two buttons mutually
  exclusive.
• New handleUnfollow(), mirroring handleFollow(), calling the
  already-existing apiService.unfollowUser().

Testing

• Backend (curl): follow → unfollow removes the relationship
  correctly; idempotent re-unfollow attempt succeeds without error;
  unfollowing a nonexistent user returns a 404-style error.
• Frontend (browser): full Follow → Unfollow → Follow round trip
  verified live on a real profile page — button correctly toggles
  each time, follower counts update correctly in both directions.

Scope

5 files changed, no schema migration needed (uses the existing
follows table). No automated tests added (consistent with prior
PRs in this repo — a good candidate for a follow-up test-coverage
pass across all six issues).