Driftr

1. Tech Stack Overview

Key Dependencies

Package	Version	Purpose
react-native	0.81.5	Core framework
expo	~54.0.33	Managed workflow + native modules
@supabase/supabase-js	^2.95.3	Backend client (Auth, DB, Realtime, Storage)
react-native-purchases	^9.7.6	RevenueCat SDK for subscriptions
react-native-purchases-ui	^9.7.6	RevenueCat pre-built paywall components
zustand	^5.0.11	Global state management
nativewind	^4.2.1	Tailwind CSS for React Native
expo-camera	~17.0.10	Video intro recording
expo-location	~19.0.8	GPS positioning for map & proximity

2. System Architecture

File Structure

driftr/
├── app/
│   ├── _layout.tsx             # Root layout — RevenueCat & auth init
│   ├── (auth)/                 # Welcome, login, signup screens
│   ├── (onboarding)/           # 5-step onboarding (basics → location)
│   ├── (tabs)/                 # Main app: discover, dating, friends, builders, profile
│   └── (screens)/              # Modals: paywall, chat, matches, settings
├── hooks/
│   ├── usePremium.ts           # RevenueCat entitlement hook
│   ├── useWaveLimit.ts         # Daily wave counter (free/premium)
│   └── useNearbyProfiles.ts    # Geo-proximity queries
├── stores/
│   ├── auth-store.ts           # Authentication state (Zustand)
│   └── match-store.ts          # Matches & messaging state
├── lib/
│   ├── supabase.ts             # Supabase client configuration
│   └── constants.ts            # App constants (wave limits, etc.)
├── components/                 # SwipeCard, MapPin, MatchCelebration, etc.
└── supabase/
    ├── schema.sql              # Full database schema + RLS policies
    └── seed.sql                # Test data

3. Database Schema

Core Tables

Table	Purpose	Key Columns
profiles	User profiles (extends auth.users)	name, avatar_url, video_intro_url, van_type, travel_style, latitude, longitude, status, looking_for[], is_builder, is_verified
waves	Directional likes	from_user, to_user, mode (dating/friends). Unique constraint prevents duplicates.
matches	Bilateral connections	user_a, user_b, mode. Auto-created by DB trigger. Ordered pair constraint (a < b).
messages	Chat messages	match_id, sender_id, content. Realtime-enabled via Supabase publication.
builder_reviews	Builder ratings	builder_id, reviewer_id, rating (1-5), comment. One review per pair.
daily_wave_count	Rate limiting	user_id, date (UTC), count. Primary key (user_id, date).

RPC Functions

-- Proximity search using PostGIS earthdistance extension
CREATE FUNCTION nearby_profiles(
  lat DOUBLE PRECISION,
  lng DOUBLE PRECISION,
  radius_km INTEGER,
  filter_mode TEXT
) RETURNS TABLE (...)
-- Uses ll_to_earth() + earth_distance() with GiST index
-- Filters by looking_for[], excludes null coords
-- Returns distance_km sorted ASC, capped at 50 results

-- Atomic wave counter (UPSERT)
CREATE FUNCTION increment_wave_count(
  p_user_id UUID,
  p_date DATE
) RETURNS INTEGER
-- INSERT ... ON CONFLICT (user_id, date) DO UPDATE SET count = count + 1

Database Triggers

• handle_new_user() — Auto-creates a profile row when a user signs up via Supabase Auth.
• check_and_create_match() — Fires on every wave INSERT. Checks if a reciprocal wave exists in the same mode. If yes, auto-creates a match. No polling, no cron jobs.

Row Level Security

Every table has RLS enabled. Profiles are publicly readable; users can only update their own. Waves, matches, and messages are scoped to participating users. Reviews are public-read with owner-only write.

4. RevenueCat Implementation

Configuration

Setting	Value
Entitlement	premium
Offering	default
Monthly Product	driftr_monthly — $7.99/mo
Annual Product	driftr_annual — $59.99/yr (7-day free trial)
User Identity	Supabase Auth user.id

Initialization Flow

RevenueCat is initialized in the root layout (app/_layout.tsx) immediately after Supabase authentication:

// app/_layout.tsx — Root layout
import Purchases from 'react-native-purchases';

// On auth state change:
Purchases.configure({
  apiKey: Platform.OS === 'ios'
    ? process.env.EXPO_PUBLIC_REVENUECAT_IOS_KEY
    : process.env.EXPO_PUBLIC_REVENUECAT_ANDROID_KEY
});
Purchases.logIn(user.id); // Sync with Supabase Auth ID

usePremium Hook

The usePremium hook (hooks/usePremium.ts) provides reactive premium state to the entire app:

export function usePremium() {
  const [isPremium, setIsPremium] = useState(false);

  useEffect(() => {
    // Check initial state
    const info = await Purchases.getCustomerInfo();
    setIsPremium(!!info.entitlements.active['premium']);

    // Listen for real-time changes (purchase, restore, expiry)
    const listener = Purchases.addCustomerInfoUpdateListener(
      (info) => setIsPremium(!!info.entitlements.active['premium'])
    );
    return () => listener.remove();
  }, []);

  return { isPremium };
}

useWaveLimit Hook

The useWaveLimit hook integrates with both RevenueCat (premium check) and Supabase (wave counting):

export function useWaveLimit() {
  const { isPremium } = usePremium();
  const limit = isPremium ? Infinity : FREE_WAVE_LIMIT; // 3

  // Query daily_wave_count from Supabase
  // Increment via increment_wave_count() RPC

  return { wavesRemaining, canWave, incrementWave };
}

Paywall Screen

The paywall (app/(screens)/paywall.tsx) fetches offerings from RevenueCat and handles purchases:

// Fetch current offering
const offerings = await Purchases.getOfferings();
const packages = offerings.current?.availablePackages;

// Purchase flow
const { customerInfo } = await Purchases.purchasePackage(pkg);
if (customerInfo.entitlements.active['premium']) {
  // Auto-dismiss paywall, user is now premium
  router.back();
}

// Restore purchases
const info = await Purchases.restorePurchases();

5 Strategic Paywall Touchpoints

5. Geo-Proximity System

The map is the home screen. Every pin represents a real person. The system uses PostgreSQL's earthdistance extension with cube for efficient spherical distance calculations.

How It Works

-- GiST index for fast proximity queries
CREATE INDEX idx_profiles_location
  ON profiles USING gist (
    ll_to_earth(latitude, longitude)
  );

-- Query: users within 50km, sorted by distance
SELECT *, earth_distance(
  ll_to_earth(lat, lng),
  ll_to_earth(p.latitude, p.longitude)
) / 1000 AS distance_km
FROM profiles p
WHERE earth_box(ll_to_earth(lat, lng), radius_km * 1000)
  @> ll_to_earth(p.latitude, p.longitude)
ORDER BY distance_km LIMIT 50;

6. Wave & Match Engine

Matching is handled entirely at the database level — no polling, no cron jobs, no background workers.

-- Trigger function: auto-create match on bilateral wave
CREATE FUNCTION check_and_create_match() RETURNS trigger AS $$
BEGIN
  IF EXISTS (
    SELECT 1 FROM waves
    WHERE from_user = NEW.to_user
      AND to_user = NEW.from_user
      AND mode = NEW.mode
  ) THEN
    INSERT INTO matches (user_a, user_b, mode)
    VALUES (
      LEAST(NEW.from_user, NEW.to_user),
      GREATEST(NEW.from_user, NEW.to_user),
      NEW.mode
    ) ON CONFLICT DO NOTHING;
  END IF;
  RETURN NEW;
END;
$$ LANGUAGE plpgsql;