RTL Fix for ChatGPT / DeepSeek

This userscript intelligently enforces Right-to-Left (RTL) text direction
for chat content on modern SPA-based AI chat platforms such as ChatGPT and
DeepSeek, making Persian, Arabic, Hebrew, and other RTL languages fully
readable and properly aligned.

🔍 How It Works

• Automatically analyzes text content to decide RTL vs LTR
• Uses an optimized MutationObserver for full SPA support
• Processes only real text elements (no UI breakage)
• Preserves correct LTR behavior for:
  • Code blocks and preformatted text
  • Numbers
  • Emojis and symbols

✅ Features

• Smart RTL detection (applied only when needed)
• Works with dynamically loaded messages
• Built-in RTL ON / OFF toggle button
• Remembers user preference via LocalStorage
• Lightweight, fast, and non-intrusive
• No ads, no tracking, no data collection

🧠 Technical Highlights

• Prevents redundant processing using WeakSet
• Debounced DOM scanning for performance
• Accurate filtering of real text elements
• Unified font stack optimized for mixed RTL / LTR content

🌍 Supported Languages

• Persian (Farsi)
• Arabic
• Hebrew
• Urdu
• Mixed RTL / LTR content

📦 Installation

Install via OpenUserJS using:

• Tampermonkey
• Violentmonkey

🐞 Issues & Feedback

Please report bugs or suggestions on GitHub:
https://github.com/sadat0063/rtl-fix/issues