Flow Table Extractor

For End Users

This Chrome extension extracts tables from websites and uploads them to Flow Immersive, with optional Claude AI smart parsing for complex tables.

✨ Key Features

• 🔍 Automatically detects all tables on any webpage
• 📊 One-click extraction and upload to Flow
• 🤖 Optional Claude AI smart parsing for multi-row headers
• 🔐 Secure credential storage (enter once, use everywhere)
• 🔄 Dataset versioning (same title = new version)

🚀 Quick Start (5 Minutes)

Step 1: Install Browser Extension

1. Open Chrome and go to chrome://extensions/
2. Enable "Developer mode" (toggle in top-right)
3. Click "Load unpacked"
4. Select the flow-table-extractor folder
5. Look for the green "FT" icon in your toolbar

Step 2: Configure Credentials (One Time!)

1. Click the "FT" extension icon
2. Click "Configure Flow Credentials"
3. Enter your Flow username and password
4. (Optional) Enter Claude API key for smart parsing
5. Click "Save All Credentials"
6. Done! Never need to enter again ✓

Step 3: Extract Your First Table

1. Go to any website with tables (try Wikipedia!)
2. Click the "FT" extension icon
3. Select a table from the dropdown
4. Enter a dataset title
5. Click "Extract & Upload to Flow"
6. Check Flow to see your data!

🔐 Security & Privacy

Where are credentials stored?

• Browser: Chrome's encrypted sync storage
• Backend: Local file (~/.flow_extractor_config.json) with restricted permissions

Is it secure?

• ✅ Chrome automatically encrypts stored credentials
• ✅ File permissions prevent other users from reading
• ✅ Communication is localhost-only or HTTPS
• ✅ No third-party access to your credentials

See SECURITY.md for detailed information.

🤖 Claude AI Smart Parsing (Optional)

For complex tables with multi-row headers (like Wikipedia population tables), enable Claude AI:

Setup:

1. Get API key from https://console.anthropic.com/
2. Enter it in "Configure Flow Credentials" (or set as environment variable)
3. Check ☑️ "Use Claude AI for smart table parsing"

Cost: ~$0.01-0.03 per table (1-3 cents)

What it does:

• Intelligently combines multi-row headers
• Handles merged cells properly
• Creates clear column names
• Perfect for complex Wikipedia/financial tables

See CLAUDE_SETUP.md for details.

📖 Documentation

• START_HERE.md - Welcome guide
• QUICKSTART.md - 5-minute setup
• SECURITY.md - Credential storage explained
• CLAUDE_SETUP.md - Claude AI setup
• README.md - Complete documentation
• VISUAL_GUIDE.md - Step-by-step walkthrough

🧪 Test Files

• test-page.html - Simple tables for testing
• test-complex-tables.html - Complex multi-row header tables

💡 Tips

1. Descriptive titles: Use clear dataset names like "Q4 Sales 2024"
2. Version control: Re-upload with same title creates new version
3. Claude for complex: Only use Claude AI for multi-row header tables
4. Keep backend running: Leave python server.py running in background

🐛 Troubleshooting

"Connection error: Failed to fetch" → Backend isn't running. Start python server.py

"Flow credentials not configured" → Click "Configure Flow Credentials" and save them

"No tables found on this page" → Page doesn't have HTML <table> elements

Tables extracted incorrectly → Try enabling "Claude AI smart parsing" for complex tables

More help: See README.md or check the terminal output

📊 Usage Examples

Wikipedia Data:

1. Go to "List of countries by population"
2. Enable Claude AI (multi-row headers!)
3. Extract → Upload to Flow
4. Analyze in Flow!

Your Company Dashboard:

1. Open dashboard with data tables
2. Select table
3. Upload to Flow for analysis

Financial Reports:

1. Find tables on investor sites
2. Enable Claude for grouped columns
3. Extract clean data

🔄 Updates

To update credentials:

1. Click extension icon
2. Click "Configure Flow Credentials"
3. Update any field
4. Save

To add/change Claude API key: Same as above - it's in the credentials form.

🎯 System Requirements

• Browser: Chrome, Edge, or Brave
• OS: Windows, Mac, or Linux
• Internet: Required for Flow uploads and Claude AI

Enjoy smart table extraction! 🚀