CampusFind

A Django + Channels social-media platform with a real-time feed, follow graph, WebSocket chat, 24-hour stories, live notifications, and an AI caption-assist feature powered by a local Ollama model.

Features

• Auth — username signup/login, profile pictures, bio editing.
• Posts — image uploads, captions, likes, comments, edit & delete.
• Profile & social graph — own profile page, follow / unfollow, follower / following lists, per-user post grids.
• Search — find users and posts by name or caption (/search/?q=...), with a search box in the navbar.
• Stories — 24-hour disappearing image posts in a horizontal strip at the top of the feed (/story/create/).
• Real-time chat — Django Channels WebSocket (ws/chat/<id>/) with sidebar updates (ws/sidebar/).
• Live notifications — bell icon in the navbar with a live-updating badge for likes, comments, follows, and messages. Updates push instantly over ws/notifications/.
• AI caption suggestions — "✨ Suggest" on the create-post page returns three caption rewrites, hashtags, and emoji from a local Ollama model. Falls back gracefully if Ollama is offline.

Stack

• Django 6.0 + Channels (Daphne for ASGI)
• SQLite (development)
• Pillow for image handling
• Bootstrap 5 + Bootstrap Icons + Font Awesome (CDN)
• Ollama for AI caption suggestions (local; no API key)

Quick start

# Activate the venv (lives one level up at ../.venv)
source ../.venv/Scripts/activate        # Git Bash
# or: ../.venv/Scripts/activate.bat     # cmd
# or: ..\.venv\Scripts\Activate.ps1      # PowerShell

pip install -r requirements.txt
python manage.py migrate
python manage.py createsuperuser

# Run the dev server (HTTP + WebSocket)
python manage.py runserver

For WebSocket-heavy testing (chat, notifications) prefer Daphne:

daphne campusfind.asgi:application

AI features (Ollama)

The "✨ Suggest" button on the create-post page calls a local Ollama HTTP API. Install and run Ollama once on your machine:

# Install from https://ollama.com, then:
ollama pull llama3.1:8b
ollama serve                # leave running in another terminal

The Django app uses these env vars (defaults shown):

OLLAMA_URL=http://localhost:11434
OLLAMA_MODEL=llama3.1:8b

If Ollama is unreachable or returns junk, the endpoint falls back to a static result so the UI never breaks.

URL map

• /admin/ — Django admin
• /, /profile/, /user/<username>/, /post/... — feed and post management (posts.urls)
• /login/, /signup/, /logout/ — auth (campusfind.views)
• /messages/, /messages/<id>/, /messages/start/<username>/ — chat (chat.urls)
• /search/?q=... — user and post search (posts.urls)
• /notifications/, /notifications/read/ — notifications dropdown + mark-all-read (notifications.urls)
• /story/create/, /story/strip/ — stories (stories.urls)

WebSocket routes (AuthMiddlewareStack in campusfind/asgi.py):

• ws/chat/<conversation_id>/ — chat messages + sidebar updates
• ws/sidebar/ — per-user inbox sidebar updates
• ws/notifications/ — live notification pushes

Project layout

campusfind/          # Django project (settings, urls, asgi)
posts/               # Feed, posts, likes, comments, follow graph
chat/                # Real-time chat (consumers, models)
stories/             # 24-hour stories + strip
notifications/       # Notifications model + WebSocket + dropdown
ai/                  # Ollama client + caption-suggest endpoint
templates/           # Project-level templates (base.html, auth/, components/)
static/css/          # style.css
media/               # User uploads (posts/, profile_images/, stories/)

Notes

• SECRET_KEY and DEBUG in settings.py are committed defaults for local development. Replace with environment variables before deploying.
• The channel layer is in-memory (channels.layers.InMemoryChannelLayer) — fine for a single-process dev server, won't scale across workers.
• Default profile picture lives at media/profile_images/default.png and is required for fresh installs.