chat-syncer

A Tampermonkey userscript for syncing ChatGPT conversations to Supabase database.

Overview

This userscript allows you to capture and sync ChatGPT conversations directly from the web interface to your Supabase database with a single click. It uses ChatGPT's backend API for reliable data extraction and includes fallback DOM scraping.

Features

ChatGPT Sync

• 🚀 One-click sync to Supabase
• 📚 Batch sync up to 20 recent conversations from homepage
• 🔄 API-first approach with DOM fallback
• ⌨️ Keyboard shortcut support (Ctrl/⌘+Shift+S)
• 🔒 Duplicate detection and prevention
• 📱 Support for regular and shared ChatGPT links
• 🎯 Clean data extraction (removes UI clutter)

Page Upload (New in v1.5.0)

• 📄 Upload any web page to Supabase as Markdown
• 🔄 HTML to Markdown conversion with structure preservation
• 🌐 Works on all websites via Tampermonkey menu
• 🔁 Auto UPSERT based on page URL

General

• 🌙 Dark mode support
• 🔧 Easy configuration via Supabase dashboard

Quick Start

1. Set up Supabase database - See SETUP.md for detailed instructions
2. Install the userscript in Tampermonkey/Violentmonkey
3. Configure your Supabase credentials on first use
4. Start syncing conversations or uploading pages

Installation

Direct Install (Recommended)

Install directly from the raw GitHub URL in your userscript manager:

• Direct install: https://raw.githubusercontent.com/chyx/chat-syncer/main/chat-syncer-unified.user.js
• Auto-updates: Enabled via @updateURL and @downloadURL directives

The unified script includes both ChatGPT syncing and Supabase config helper functionality.

Usage

Easy Configuration (Recommended)

1. Install the unified userscript
2. Go to your Supabase project → Settings → API
3. Click the "🚀 配置 ChatGPT Syncer" button (top-right)
4. Click "🚀 直接保存配置" - no copying needed!
5. Go to ChatGPT and start syncing

Single Conversation Sync

• Navigate to any ChatGPT conversation
• Click the "Sync → Supabase" button (bottom-right corner)
• Or use keyboard shortcut: Ctrl/⌘ + Shift + S

Batch Sync (20 Recent Conversations)

• Go to ChatGPT homepage (https://chatgpt.com/)
• Click the "📚 批量同步最近20条" button (bottom-right corner)
• Monitor progress in the popup modal
• Duplicate conversations are automatically skipped

Upload Any Web Page

• Navigate to any web page
• Click Tampermonkey icon → "Upload Page"
• Page content is converted to Markdown and uploaded to Supabase
• Same URL = update existing record (UPSERT)

Retrieve Uploaded Pages

To get the latest uploaded pages from Supabase:

-- Get the most recent 10 uploads
SELECT page_url, page_title, updated_at, page_content
FROM page_uploads
ORDER BY updated_at DESC
LIMIT 10;

-- Get a specific page by URL
SELECT * FROM page_uploads
WHERE page_url = 'https://example.com/page';

-- Search pages by title or content
SELECT page_url, page_title, updated_at
FROM page_uploads
WHERE page_title ILIKE '%search term%'
   OR page_content ILIKE '%search term%'
ORDER BY updated_at DESC;

Or use Supabase client library:

// Get latest uploads
const { data, error } = await supabase
  .from('page_uploads')
  .select('page_url, page_title, updated_at, page_content')
  .order('updated_at', { ascending: false })
  .limit(10);

Manual Configuration

If auto-detection fails, you can manually configure:

• Click sync button on any ChatGPT page
• Fill in Supabase URL, API key, and table name in the modal

Database Setup

📖 See SETUP.md for complete database setup instructions

Quick links:

• ChatGPT Sync Setup - chat_logs table
• Page Upload Setup - page_uploads table
• Troubleshooting - Common issues and solutions
• Query Examples - Useful SQL queries

Data Format

ChatGPT Conversations (chat_logs)

Each synced conversation is stored as a single row containing:

• Basic info: chat_id, chat_url, chat_title, timestamps
• Messages array: [{idx, role, text, html}, ...]
• Metadata: user agent, source method (api/dom), viewport info

Uploaded Pages (page_uploads)

Each uploaded web page contains:

• Page info: page_url (unique), page_title, created_at, updated_at
• Content: page_content (Markdown format)
• Metadata: user agent, viewport, content length

See SETUP.md for detailed table schemas and security policies.

Development

Project Structure

src/
├── config.js         - Theme, CONFIG, PageDetector
├── chatgpt.js        - ChatGPT module (UI, batch sync, data extraction)
├── supabase.js       - Supabase module (config helper, auto-detection)
├── page-uploader.js  - Page upload module (HTML→Markdown, upload)
└── main.js           - Initialization logic

build.js              - Build script to generate chat-syncer-unified.user.js
SETUP.md              - Complete database setup guide

Building

npm run build       # Generate chat-syncer-unified.user.js from source files
npm test           # Run test suite

The chat-syncer-unified.user.js file is generated from source files in src/. Always edit source files, not the generated file.

Version Management

Version is managed in package.json and automatically injected into the userscript header during build.

License

MIT