IndiaForex Market Institutional Terminal

Visit the Deployment at https://indiaforex.vercel.app/

Visit the Repo at https://github.com/indiaforex/indiaforex/

A high-performance financial intelligence platform bridging institutional-grade market data with community-driven insights.

1. System Architecture

The project adopts a Serverless Hybrid Architecture, leveraging Next.js 15 (App Router) for the frontend and Supabase (PostgreSQL) for the backend. It prioritizes "Zero-Click Latency" and real-time data synchronization.

Core Stack

• Frontend: Next.js 15.1, React 19, Tailwind CSS v4, Framer Motion.
• Database: PostgreSQL 15 (Supabase) with uuid-ossp extension.
• Auth: Supabase Auth (JWT + RLS) with Google/GitHub/Twitter OAuth providers.
• State Management: Server Actions for mutations, unstable_cache (ISR) for data fetching, and optimistic UI updates.

Realtime Event Architecture

The application actively utilizes Supabase Realtime (WebSockets) to broadcast state changes instantly to connected clients.

• Pub/Sub Model: The PostgreSQL database acts as the single source of truth, broadcasting INSERT, UPDATE, and DELETE events via the supabase_realtime publication.
• Active Channels:
  • notifications: Users receive instant alerts (toast + bell badge) when mentioned or replied to, without refreshing.
  • forum_comments: Thread discussions update live as new comments are posted by other users.
  • forum_categories: Category structure updates propagate instantly (metadata sync).
• Client-Side Subscription: Implemented via supabase.channel().on('postgres_changes', ...) hooks in React components (NotificationBell.tsx, CommentSection.tsx).

Data Pipeline Strategy

The application employs a dual-pipeline strategy for data delivery:

1. Cold Storage (CMS/Editorial): Using Google Sheets via SheetDB as a headless CMS for the Economic Calendar.

   • Dual-Write Workflow:
     • Direct: Admins can edit the Google Sheet directly.

       

     • Application: Users with the specialized event_analyst role can use the Admin Panel > Events form to append rows to the sheet via API.

       

   • Read Path: Data is cached at the edge using Next.js ISR tags (revalidate: 60), ensuring site resilience even if the SheetDB API is rate-limited.

   

2. Hot Storage (Market Data): Real-time fetching via yahoo-finance2 on the server, protected by a 15-second deduplication cache (unstable_cache) to prevent rate-limiting while serving thousands of concurrent users.

---

2. Technical Features

Command Center & Dashboard

• Global Market Watch: Aggregates real-time indices (NIFTY, SENSEX, NASDAQ) using server-side fetching.

  

• Algorithmic Scanner: "Live Market Scanner" component runs simple heuristics (Gap Up, Volume Shock) on fetched payloads to surface opportunities instantly.

  

• Sector Heatmap: Visualizes relative performance of top 10 constituents using color scales effectively.

  

Community Engine (The "Alpha" Core)

A from-scratch social platform built directly on Postgres.

• Recursive Threading: Supports infinite nesting for deep discussions.

  

• Reputation System:
  • Logic: Implemented via PostgreSQL Triggers, ensuring atomicity. Every Like or Comment fires a database function to update the user's reputation_points.
  • Security: Users cannot "game" the system; points are managed entirely by DB constraints, invalidating points if content is deleted.
• Role-Based Access Control (RBAC):
  • Hierarchy: Guest > User > High Level > Steward > Event Analyst > Admin > Super Admin.
  • Stewards: Can moderate only specific categories (e.g., "Crypto Steward" cannot moderate "Forex").

---

3. Engineering Challenges & Workarounds

A. Role Integration with Supabase RLS

Problem: Supabase Auth handles authentication (identity), but our application relies on complex authorization (roles like super_admin) stored in a public profiles table. Solution:

• We use a Trigger-based Sync (handle_new_user) to auto-create a profile row upon signup.
• Hybrid RLS Policies: Policies don't just check auth.uid(); they perform efficient sub-queries to the profiles table to check roles.

  -- Example Policy: Admins can update any thread
  create policy "Admins can update any thread"
    on forum_threads for update
    using ( exists (select 1 from profiles where id = auth.uid() and role in ('admin', 'super_admin')) );

• Fix Implemented: We had to explicitly DROP and recreate policies to ensure Admin overrides took precedence over standard "User can edit own post" policies, solving a "Forbidden" error for admins.

B. TradingView Integration in React

Problem: The TradingView "Advanced Chart" widget relies on strictly imperative script injection and often throws iframe contentWindow errors when unmounted rapidly (e.g., during React Fast Refresh or tab switching). Solution:

• Wrapped the widget in a memoized component with strictly controlled dependency arrays.
• Implemented a ref-based cleanup specifically targeting the script element to ensure clean unmounting.
• Used a specific container id generation strategy to avoid DOM collisions during re-renders.

C. The "Google Sheets as Backend" Pattern

Problem: Operations team needed to update "Economic Events" faster than a database admin panel could be built/deployed. Solution:

• Implemented sheetdb.ts adapter.
• Optimization: Configured Next.js to cache the SheetDB response for 60 seconds (revalidate: 60). This prevents hitting Google's strict API quotas while keeping the calendar "fresh enough" for macro news.
• Resilience: The adapter includes specific error handling for non-array responses, preventing the dashboard from crashing if the Sheet format is temporarily broken by a human editor.

---

4. Key Security Implementation

• Row Level Security (RLS): 100% of database access is protected by RLS. No server-side "service role" bypass is used for standard user actions.
• XSS Protection: Comments are sanitized using rehype-sanitize before rendering to prevent script injection in the rich text editor.

---

5. Future Roadmap (Technical)

• Migration of sheetdb to a proper Postgres Table once the schema stabilizes or the operators get technically sound.
• Implementation of WebSocket subscriptions for "Live Ticker" pushing (currently polling/cached).