prepare_for_spaces.py

#!/usr/bin/env python3
"""
Prepare files for HuggingFace Spaces deployment
Copies only the required files to a clean directory
"""

import os
import shutil
from pathlib import Path

# Required files for HuggingFace Spaces
REQUIRED_FILES = [
    # Core application
    'app.py',

    # Processing modules
    'llm.py',
    'extractors.py',
    'tagging.py',
    'chunking.py',
    'validation.py',
    'reporting.py',
    'dashboard.py',
    'production_logger.py',

    # Optional but recommended
    'quote_extractor.py',

    # Configuration
    'requirements.txt',

    # Documentation (optional)
    'README.md',
    'HUGGINGFACE_SPACES_SETUP.md',
]

def prepare_deployment(output_dir='./spaces_deployment'):
    """Copy required files to deployment directory"""

    # Create output directory
    output_path = Path(output_dir)
    if output_path.exists():
        print(f"⚠️  Directory {output_dir} already exists")
        response = input("Delete and recreate? (y/n): ")
        if response.lower() != 'y':
            print("❌ Cancelled")
            return
        shutil.rmtree(output_path)

    output_path.mkdir(exist_ok=True)
    print(f"📁 Created directory: {output_dir}\n")

    # Copy files
    copied = []
    missing = []

    for filename in REQUIRED_FILES:
        src = Path(filename)
        if src.exists():
            dst = output_path / filename
            shutil.copy2(src, dst)
            size_kb = src.stat().st_size / 1024
            print(f"   ✅ {filename} ({size_kb:.1f} KB)")
            copied.append(filename)
        else:
            print(f"   ⚠️  {filename} - NOT FOUND (skipping)")
            missing.append(filename)

    # Summary
    print("\n" + "="*80)
    print("📊 SUMMARY")
    print("="*80)
    print(f"✅ Copied: {len(copied)} files")
    if missing:
        print(f"⚠️  Missing: {len(missing)} files")
        print(f"   {', '.join(missing)}")

    print(f"\n📦 Deployment files ready in: {output_dir}/")
    print("\n📋 Next steps:")
    print("1. Go to https://huggingface.co/new-space")
    print("2. Select Gradio SDK and GPU hardware")
    print("3. Upload all files from the deployment directory")
    print("4. Wait for model download (~2-5 min first time)")
    print("5. Test your Space!")

    # Check for .env file (should not be included)
    if (output_path / '.env').exists():
        print("\n⚠️  WARNING: .env file found in deployment directory!")
        print("   This should NOT be deployed to HuggingFace Spaces")
        os.remove(output_path / '.env')
        print("   ✅ Removed .env file")

    print("\n✨ Deployment package ready!")

if __name__ == '__main__':
    prepare_deployment()