---

🤖 RVDiA (Revolusioner, Virtual, Independen)

Revolutionary Virtual Discord Assistant (RVDiA) is a high-performance, Indonesian-oriented Discord entertainment companion and utility bot built entirely on a modern Python stack. It features deep AI-driven personality integration, an advanced local GPU rendering pipeline, professional-grade image processing, a turn-based RPG battle engine, and a premium web dashboard that allows real-time configuration, player interactions, and widget embeddings.

---

Table of Contents

1. Core Features
2. Database Schema & Architecture
3. Local GPU Generation Pipeline
4. Web Dashboard & REST APIs
5. Embeddable Web Chat Widget
6. Local Development Setup
7. Deployment
8. Credits & Contributors

---

Core Features

Turn-Based RPG Battle Engine

• Tactical In-Battle Interface: Features paginated Select Menus (complying with Discord's API limits) for skills and equipments, complete with real-time health trackers and active state monitoring.
• Shop & Inventory: Over 100+ items, skills, and equipments configured in JSON and dynamically translated based on locale settings.
• Boss Fights & Elite Tiers: Challenging bosses with complex phases (e.g., Demi-fiend, Nahobino) and unique mechanisms.

Advanced Image Processing Toolkit

• Mathematical Comparison: Employs OpenCV and Matplotlib to analyze and graph visual histograms side-by-side. Supports multiple metric engines (correl, chisqr, intersect, bhattacharyya).
• Paginated Navigation: Built-in interactive Discord UI button components enabling fast Next/Prev searching for internet images directly inside message threads.

Persistent AI Memories

• Contextual Recall: Integrates Google Gemini API with Prisma Client Python and vector embeddings to search, record, and write persistent user interactions.
• Linguistic Title Casing: Intelligent formatter processing title-casing rules tailored specifically for Indonesian and English grammatical structures.

---

Database Schema & Architecture

RVDiA relies on PostgreSQL for persistence, mapped via schema.prisma. The core models are organized as follows:

erDiagram
    User ||--o| Inventory : "owns"
    User ||--o| Guild : "member of"
    User ||--o| UserSettings : "customizes"
    User ||--o{ Message : "sends"
    User ||--o{ Memory : "associates"

    User {
        BigInt id PK
        Int hp
        Int max_hp
        Json data "Level, EXP, Coins, Karma, Stats"
        DateTime premiumUntil
    }

    Guild {
        Int id PK
        String name
        String tagline
        String iconUrl
        BigInt ownerId
    }

    Inventory {
        Int id PK
        BigInt userId FK
        Json items "IDs & quantities"
        Json skills "learned skills"
        Json equipments "equipped items"
    }

    UserSettings {
        BigInt userId PK
        String lang "en/id"
    }

Loading

• Relations Cascading: Disbanding a guild safely disconnects all user relations in the database. Deleting users cleanly cascade-deletes their related Inventory record.

---

Local GPU Generation Pipeline

To perform heavy Stable Diffusion inference (Meina/MeinaMix_V11) without exhausting memory on a low-VRAM development machine (RTX 3050 4GB), RVDiA offloads rendering to a localized laptop server. The process is gated by interactive Windows notification checks to prevent background disruptions:

sequenceDiagram
    actor User as Discord User
    participant Bot as RVDiA Bot (Railway)
    participant Server as GPU Server (Local Laptop)
    actor Owner as Laptop Owner (Schryzon)

    User->>Bot: Slash Command /generate [prompt]
    Bot->>Server: GET /ping (X-API-Key)
    alt Server Offline
        Bot-->>User: Reply: "Artist Offline"
    else Server Online
        Bot->>Server: POST /generate [prompt, username, is_nsfw]
        Note over Server: Check NSFW blacklist (if SFW channel)
        alt NSFW Blocked
            Server-->>Bot: 400 Bad Request (NSFW Blocked)
            Bot-->>User: Reply: "NSFW content blocked!"
        else Allowed
            Server-->>Bot: 200 OK (request_id)
            Bot->>User: Defer response (loading...)
            Server->>Owner: Winotify / Tkinter gated approval prompt
            alt Owner Declines
                Owner->>Server: Click "Decline"
                Server->>Server: Update status: declined
            else Owner Approves
                Owner->>Server: Click "Approve"
                Server->>Server: Update status: generating
                Note over Server: Load SD model & Move CPU -> CUDA GPU
                Server->>Server: Generate Image (512x512 PNG)
                Note over Server: Move CUDA -> CPU & Aggressive VRAM Clear
                Server->>Server: Save image & Update status: completed
            end
            loop Poll Status (Every 2s, max 15m)
                Bot->>Server: GET /status/[request_id]
                Server-->>Bot: Status (pending/generating/declined/completed)
            end
            alt Status: Completed
                Bot->>Server: GET /image/[request_id]
                Server-->>Bot: Image File
                Bot-->>User: Send generated image!
            else Status: Declined
                Bot-->>User: Reply: "Generation declined by artist."
            else Status: Failed / Timeout
                Bot-->>User: Reply: "Generation failed."
            end
        end
    end

Loading

---

Web Dashboard & REST APIs

The web interface is hosted on aiohttp and Jinja2 with zero Node dependencies, utilizing standard CSS and clean Javascript logic.

Authentication

Authentication utilizes a secure Discord OAuth2 authorization flow. Sessions are persisted client-side using HMAC-signed cookies to prevent hijacking.

Endpoints Definition (in routes.py)

Method	Endpoint	Description	Auth Required
GET	/api/v1/user/profile	Fetches authenticated user's profile metadata and currencies.	Yes
GET	/api/v1/user/inventory	Fetches the user's item, skill, and equipment logs.	Yes
GET	/api/v1/stats	System metrics (uptime, CPU/RAM, command executions).	No
POST	/api/v1/chat	Send a prompt to the context-aware chat engine.	Yes
GET	/api/v1/shop	Fetches the translated list of marketplace items, categorized.	Yes
POST	/api/v1/shop/buy	Purchase a shop item (consumable, skill, equipment).	Yes
GET	/api/v1/leaderboard	Fetches Top Players or Guilds ranked rankings.	Yes
GET	/api/v1/guild	Fetches active guild stats, tagline, icon, and member list.	Yes
POST	/api/v1/guild/create	Spawns a guild for 5,000 Coins.	Yes
POST	/api/v1/guild/edit	Modifies the guild name, tagline, or icon URL.	Yes
POST	/api/v1/guild/kick	Kicks a member (only available to the guild owner).	Yes
POST	/api/v1/guild/leave	Leave/disband the guild.	Yes
POST	/api/v1/public/chat	Session-hashed public endpoint for the web widget.	No

---

Embeddable Web Chat Widget

RVDiA features an embeddable chat script allowing users to interface with the AI assistant from external web pages.

Security & Session-Hashing

To avoid polluting the database with arbitrary guest entries, the public endpoint /api/v1/public/chat handles stateless traffic by mapping incoming client UUIDs:

1. Client generates a random UUID on their first load and saves it to localStorage.
2. The server hashes this string using SHA-256 and extracts the first 8 bytes.
3. This creates a positive 63-bit signed integer representation which serves as a virtual Discord user ID context in PostgreSQL.
4. Allows persistent, isolated chat threads per web visitor without requiring Discord login credentials.

Quickstart Embed

Include this script tag inside your HTML body to launch the chatbot:

<script 
  src="http://localhost:8080/static/js/rvdia-widget.js"
  data-api-url="http://localhost:8080" 
  data-lang="id"
  defer>
</script>

Developer Testbed

Creators can visit the built-in development sandbox at /widget-demo to inspect the embedded script, configure language settings, and view real-time API logs detailing requests, UUID-to-ID hashes, and response packages.

---

Local Development Setup

Follow these steps to set up and run RVDiA on a Windows machine.

Prerequisites

• PowerShell 5.1+ (or modern PowerShell Core)
• Python 3.12 (Scoop package manager is recommended: scoop install python312)
• PostgreSQL database instance.

Step-by-Step Installation

1. Clone the repository:

   git clone https://github.com/Schryzon/RVDiA.git
   cd RVDiA

2. Configure Environment Variables: Copy .env.example to .env and fill out your api credentials, discord developer credentials, and local postgres database link:

   copy .env.example .env

3. Install Requirements:

   pip install -r requirements.txt

4. Initialize Prisma Client: Generate database abstractions and sync models:

   # Generate schema bindings
   prisma generate
   
   # Synchronize models to Postgres
   prisma db push --accept-data-loss

5. Launch the Engine: Run the main bot and web panel:

   python RVDIA.py

   The dashboard will be active at http://localhost:8080.

---

Deployment

RVDiA is configured for seamless deployment on Railway using the provided Dockerfile and railway.toml.

When launching in a production container, start.sh runs automatically, ensuring the prisma database models are pushed and client headers are rebuilt before spawning RVDIA.py.

---

Credits & Contributors

Special Acknowledgments

• Concept Inspiration: Riverdia.
• Advisors & Core Testers: iMaze, Mouchi, Dez, Zenchew, Ismita, Pockii, Kyuu, Kazama, Bcntt, Nateflakes, nathawiguna, opensourze, Shiruto, Satya Yoga.

---

Made with ❤️ and dedication by Jayananda and contributors.

Rebirth: 30/04/2026