NextExpense/README.md

ReceiptNext — AI-Powered Expense Tracker

A production-ready, mobile-first Progressive Web App (PWA) for expense management with passwordless OTP login, AI receipt extraction (Gemini / OpenAI / Ollama), and CSV/PDF email reporting with receipt images.

Tech Stack: Go 1.23+ · HTMX · SQLite (pure Go) · Google Gemini / OpenAI / Ollama · PWA

---

🚀 One-Command Install

sudo bash -c "$(curl -fsSL https://git.lohmar.co.uk/cclohmar/ReceiptNext/raw/branch/main/install.sh)"

Or from a local clone:

sudo ./install.sh

The installer will:

1. Clone the repo to /opt/receiptnext/

2. Build the binary (pure Go, no CGO, no dependencies)

3. Ask which AI provider to use:

   1) Google Gemini  (cloud API, needs API key)
   2) OpenAI / Compatible  (OpenAI, Perplexity, Groq, etc.)
   3) Ollama  (local — installs Ollama + qwen3.5:2b automatically)
   

4. Prompt for SMTP settings (for OTP emails and report delivery)

5. Create /opt/receiptnext/.env with all configuration

6. Set up a systemd service that auto-starts on boot

7. Start ReceiptNext

Result: A fully configured, always-running expense tracker at http://YOUR_SERVER:8080.

---

🔄 Updating

sudo /opt/receiptnext/install.sh -update

This pulls the latest code, rebuilds the binary, and restarts the service.

---

📦 Binary Download (manual)

Pre-compiled static binaries are on the Releases page:

# Linux amd64
curl -L -o app https://git.lohmar.co.uk/cclohmar/ReceiptNext/releases/download/v1.0.0/app-linux-amd64
chmod +x app
./app

# Linux arm64 (Raspberry Pi, etc.)
curl -L -o app https://git.lohmar.co.uk/cclohmar/ReceiptNext/releases/download/v1.0.0/app-linux-arm64
chmod +x app
./app

No dependencies, no CGO, no libc. Drop it on any Linux box and run.

---

🔧 Building from Source

Prerequisites

• Go 1.23+ — Download
• No GCC, no CGO, no cross-compilers needed

Build

git clone https://git.lohmar.co.uk/cclohmar/ReceiptNext.git
cd ReceiptNext

# Build for your current platform
CGO_ENABLED=0 go build -o app .

# Cross-compile for any platform (no extra tools needed!)
CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build -o app-linux-arm64 .
CGO_ENABLED=0 GOOS=darwin GOARCH=amd64 go build -o app-darwin-amd64 .
CGO_ENABLED=0 GOOS=darwin GOARCH=arm64 go build -o app-darwin-arm64 .

The binary is fully static — zero runtime dependencies.

---

⚙️ Configuration

Configuration is via environment variables in .env. The install script builds this for you interactively.

Variable	Required	Default	Description
PORT	No	8080	HTTP server port
BASE_URL	No	http://localhost:8080	Public URL for email links
AI Provider
AI_PROVIDER	No	gemini	gemini, openai, or ollama
GEMINI_API_KEY	For Gemini	—	Google Gemini API key
OPENAI_API_KEY	For OpenAI	—	OpenAI-compatible API key
AI_MODEL	For OpenAI/Ollama	gpt-4o-mini / qwen3.5:2b	Model name
AI_BASE_URL	For OpenAI/Ollama	https://api.openai.com/v1	API endpoint
SMTP
SMTP_HOST	See note	—	SMTP server hostname
SMTP_PORT	See note	587	SMTP server port
SMTP_USER	See note	—	SMTP username
SMTP_PASS	See note	—	SMTP password

Note: SMTP is optional without it, OTP codes are logged to the server console for testing. Reports cannot be emailed without SMTP.

---

🚀 Usage

1. Start the Server

# If installed via systemd:
systemctl start receiptnext

# Or run directly:
./app

2. Open in Browser

Navigate to http://YOUR_SERVER:8080

3. Full Acceptance Flow

1. Enter your email → click "Send Verification Code"
2. Check your inbox (or server log) for the 6-digit OTP
3. Enter OTP → click "Verify Code"
4. Click "+ New" → name your event → set claim currency → enter a conversion sample
5. Click "Open" on the event
6. Tap "📷 Camera" or "📁 Upload" to add a receipt
7. AI extracts amount, merchant, category, date → form is pre-filled
8. Click "Save Expense"
9. Click "Submit Event" → enter recipient email → choose CSV or PDF
10. Report + receipt images ZIP are emailed → event auto-closes
11. Click "Reopen" to add more receipts

---

📡 API Reference

Public

Method	Path	Description
GET	/	Landing page
POST	/request-otp	Request OTP code
POST	/verify-otp	Verify OTP and create session
POST	/logout	Clear session

Protected (requires session)

Method	Path	Description
GET	/dashboard	Event list
POST	/events	Create event
GET	/events/{id}/expenses	View event + expenses
PUT	/events/{id}/reopen	Reopen closed event
POST	/expenses/upload	Upload receipt image/PDF
POST	/expenses	Save expense
GET	/expenses/{id}/edit	Get edit form
PUT	/expenses/{id}	Update expense
POST	/events/{id}/file	File report (CSV/PDF + images ZIP)

---

🏗️ Project Structure

/opt/receiptnext/
├── receiptnext              # Compiled binary
├── .env                     # Configuration
├── templates/               # Go HTML templates
├── static/                  # CSS, icons, favicon, service worker
│   ├── css/style.css
│   ├── favicon.svg          # Rx logo
│   ├── manifest.json
│   ├── sw.js
│   └── icons/
├── storage/                 # Uploaded receipt images (runtime)
├── install.sh               # Installer script
├── Makefile                 # Build targets
└── contrib/
    └── receiptnext.service  # Systemd service file

---

🧩 Features

🔐 Passwordless OTP Authentication

• Email-based 6-digit code, 5-minute expiry
• 3 failed attempts → 1-minute cooldown
• HTTP-only session cookie, 24h TTL

🤖 AI Receipt Extraction

• Google Gemini (default) — cloud vision API
• OpenAI-compatible — works with OpenAI, Perplexity, Groq, Together AI, etc.
• Ollama — local, offline, no API key needed (uses qwen3.5:2b)
• Supports: JPEG, PNG, WebP, HEIC, PDF (email receipts from Uber, etc.)

💱 Currency Conversion

• Sample-based: enter a real receipt amount and what you were charged
• System computes the rate automatically
• Each expense stores original + converted amount

📧 Email Reporting

• CSV or PDF report
• Receipt images bundled as ZIP (expense-{event}-images.zip)
• Filenames: expense-{event}-report.csv, expense-{event}-report.pdf
• Item numbers match between report rows and ZIP images

📱 Progressive Web App

• Installable on mobile home screen
• Camera capture + gallery upload
• Dark "Terminal Mint" theme (#0F172A base)
• Rx favicon in emerald green

---

🗄️ Database

SQLite, auto-created on first run — 4 tables: users, auth_otps, events, expenses.

---

🔒 Security

Area	Implementation
Sessions	crypto/rand tokens, in-memory, 24h TTL
OTP	crypto/rand codes, 5min expiry, lockout after 3 failures
Cookies	HTTP-only, SameSite=Lax, path-restricted
SQL	Parameterized queries everywhere
Uploads	Magic byte validation, max 10MB, UUID filenames
Config	.env file is chmod 600 — readable only by root

---

Built with Go, HTMX, SQLite and ❤️