CalCount

A personal, self-hosted calorie counter web app with barcode scanning via phone camera and OpenFoodFacts lookup. See SPEC.md for the full spec, data model, API design, and development rules (§8 — read it before writing code).

Stack

Backend: Python ≥3.12, FastAPI, SQLAlchemy 2, Pydantic 2, SQLite, httpx — managed with uv
Frontend: Svelte 5 (runes) + Vite 7, zxing-wasm barcode fallback — managed with npm (Node ≥22.12)

Project layout

backend/    FastAPI app (main.py), SQLAlchemy models, Pydantic schemas,
            routers/, services/, migrations/, tests/
frontend/   Vite + Svelte SPA (src/), lib/ (api, stores, scanner, format),
            components/

Setup

# Backend (uv creates .venv and installs from uv.lock)
cd backend
uv sync

# Frontend
cd frontend
npm install

Run

Single command — starts both servers, cleans up on Ctrl+C:

./dev.sh

Server	URL
Backend	http://localhost:8000
Frontend	http://localhost:5173

Or manually in two terminals:

# Terminal 1 — backend
cd backend && uv run uvicorn main:app --reload --host 0.0.0.0

# Terminal 2 — frontend (exposed on LAN for phone testing)
cd frontend && npm run dev:host

The SQLite database (backend/calcount.db) is created and migrated automatically on backend startup (backend/migrations/).

Test

cd backend && uv run pytest        # backend suite
cd frontend && npm test            # frontend suite (vitest)

Build

cd frontend && npm run build       # production build to frontend/dist/

Agentic dev

Defined subagents, idea is big strong agent tells the little ones what to do. I maybe have overkilled on the "small" agents, v4 pro is till pretty powerful. Play around with it and see what we can get away with.

Future enhancements

fully offline JS version (new project built out of this, i have some notes on this somewhere)
download and use https://world.openfoodfacts.org/data
have this version work offline (pwa or local storage or something)

2.3 KiB Raw Blame History