CaloriSee

AI-powered calorie estimation from WhatsApp chat exports. Classifies food messages and estimates calories/macros using OpenAI (text + vision).

Quick Start

# 1. Install deps
pip install -r requirements.txt
pip install -r backend/requirements.txt
cd frontend && npm install && cd ..

# 2. Set your OpenAI key
cp .env.example .env
# edit .env → OPENAI_API_KEY=sk-proj-...

# 3. Run the web app
uvicorn backend.app:app --reload --port 8000   # terminal 1
cd frontend && npm run dev                      # terminal 2
# Open http://localhost:5173

CLI Usage

# From Google Drive
python -m whatsapp_calorie_bot --gdrive_url "https://drive.google.com/file/d/..." --out_dir ./run_001

# From local file
python -m whatsapp_calorie_bot --local_archive ./archive.zip --out_dir ./run_001

# Batch API (50% cheaper)
python -m whatsapp_calorie_bot --local_archive ./archive.zip --batch --out_dir ./run_001

Flags: --timezone, --food_confidence_threshold, --max_messages, --since, --until, --force_redo, --batch, --verbose.

How It Works

1. Extract WhatsApp export archive (zip/rar/7z)
2. Parse chat .txt file, associate images
3. Classify each message as food/non-food (gpt-4.1-mini)
4. Estimate calories + macros for food messages (gpt-4.1 for images, gpt-4.1-mini for text)
5. Dashboard shows daily calories chart, KPIs, macro breakdown, top items
6. Override AI estimates per-message from the web UI

Results cached in SQLite — re-runs skip already-processed messages.

Privacy

Message text and food images are sent to OpenAI's API. Don't use with sensitive data. Test with --max_messages 5 first.