verify_hf_spaces_ready.py

#!/usr/bin/env python3
"""
Verification script to check if CogniChat is ready for HuggingFace Spaces deployment.
Run this before deploying to catch any configuration issues.
"""

import os
import sys
from pathlib import Path

def print_header(text):
    """Print a formatted header."""
    print(f"\n{'='*60}")
    print(f"  {text}")
    print(f"{'='*60}")

def print_check(condition, message):
    """Print a check result."""
    status = "PASS" if condition else "FAIL"
    print(f"{status}: {message}")
    return condition

def verify_files_exist():
    """Verify all required files exist."""
    print_header("1. Checking Required Files")
    
    required_files = [
        'app.py',
        'rag_processor.py',
        'Dockerfile',
        'requirements.txt',
        'templates/index.html'
    ]
    
    all_exist = True
    for file in required_files:
        exists = Path(file).exists()
        all_exist = all_exist and print_check(exists, f"File exists: {file}")
    
    return all_exist

def verify_upload_folder_config():
    """Verify upload folder configuration in app.py."""
    print_header("2. Checking Upload Folder Configuration")
    
    with open('app.py', 'r') as f:
        app_content = f.read()
    
    checks = [
        ('SPACE_ID' in app_content, "Detects SPACE_ID environment variable"),
        ("'/tmp/uploads'" in app_content, "Uses /tmp/uploads for HF Spaces"),
        ('is_hf_spaces' in app_content, "Has HF Spaces detection logic"),
        ('os.makedirs(app.config' in app_content, "Creates upload directory"),
    ]
    
    all_pass = True
    for condition, message in checks:
        all_pass = all_pass and print_check(condition, message)
    
    return all_pass

def verify_session_management():
    """Verify session management implementation."""
    print_header("3. Checking Session Management")
    
    # Check backend
    with open('app.py', 'r') as f:
        app_content = f.read()
    
    backend_checks = [
        ("session['session_id']" in app_content, "Stores session_id in Flask session"),
        ("'session_id': session_id" in app_content, "Returns session_id in upload response"),
        ("data.get('session_id')" in app_content, "Accepts session_id from request body"),
    ]
    
    # Check frontend
    with open('templates/index.html', 'r') as f:
        html_content = f.read()
    
    frontend_checks = [
        ('sessionStorage.setItem' in html_content, "Frontend stores session_id"),
        ('sessionStorage.getItem' in html_content, "Frontend retrieves session_id"),
        ('requestBody.session_id' in html_content, "Frontend sends session_id in chat"),
    ]
    
    all_pass = True
    print("  Backend Implementation:")
    for condition, message in backend_checks:
        all_pass = all_pass and print_check(condition, f"  {message}")
    
    print("\n  Frontend Implementation:")
    for condition, message in frontend_checks:
        all_pass = all_pass and print_check(condition, f"  {message}")
    
    return all_pass

def verify_error_handling():
    """Verify robust error handling."""
    print_header("4. Checking Error Handling")
    
    with open('app.py', 'r') as f:
        app_content = f.read()
    
    checks = [
        ('try:' in app_content and 'except' in app_content, "Has try/except blocks"),
        ('load_pdf_with_fallback' in app_content, "Has PDF fallback loading"),
        ('failed_files' in app_content, "Tracks failed file uploads"),
        ('fallback_dir' in app_content, "Has cache directory fallbacks"),
    ]
    
    all_pass = True
    for condition, message in checks:
        all_pass = all_pass and print_check(condition, message)
    
    return all_pass

def verify_environment_variables():
    """Check for environment variable handling."""
    print_header("5. Checking Environment Variables")
    
    with open('app.py', 'r') as f:
        app_content = f.read()
    
    with open('rag_processor.py', 'r') as f:
        rag_content = f.read()
    
    checks = [
        ('GROQ_API_KEY' in rag_content, "RAG processor checks GROQ_API_KEY"),
        ('SPACE_ID' in app_content, "App detects HF Spaces environment"),
        ('HF_HOME' in app_content, "Sets HuggingFace cache paths"),
        ('load_dotenv()' in rag_content, "Loads .env file for local dev"),
    ]
    
    all_pass = True
    for condition, message in checks:
        all_pass = all_pass and print_check(condition, message)
    
    # Check if .env.example exists
    all_pass = all_pass and print_check(
        Path('.env.example').exists() or Path('README.md').exists(),
        "Has documentation for environment variables"
    )
    
    return all_pass

def verify_dockerfile():
    """Verify Dockerfile configuration."""
    print_header("6. Checking Dockerfile")
    
    if not Path('Dockerfile').exists():
        print_check(False, "Dockerfile exists")
        return False
    
    with open('Dockerfile', 'r') as f:
        dockerfile_content = f.read()
    
    checks = [
        ('FROM python' in dockerfile_content, "Uses Python base image"),
        ('WORKDIR /app' in dockerfile_content, "Sets working directory"),
        ('EXPOSE 7860' in dockerfile_content, "Exposes port 7860 (HF requirement)"),
        ('appuser' in dockerfile_content, "Runs as non-root user"),
        ('CMD' in dockerfile_content or 'ENTRYPOINT' in dockerfile_content, "Has startup command"),
    ]
    
    all_pass = True
    for condition, message in checks:
        all_pass = all_pass and print_check(condition, message)
    
    return all_pass

def verify_requirements():
    """Verify requirements.txt has all dependencies."""
    print_header("7. Checking Dependencies")
    
    with open('requirements.txt', 'r') as f:
        requirements = f.read().lower()
    
    critical_deps = [
        'flask',
        'langchain',
        'groq',
        'faiss',
        'sentence-transformers',
        'pypdf',
        'gtts',
        'rank-bm25'
    ]
    
    all_pass = True
    for dep in critical_deps:
        found = dep in requirements
        all_pass = all_pass and print_check(found, f"Has dependency: {dep}")
    
    return all_pass

def verify_no_duplicates():
    """Check for duplicate content in index.html."""
    print_header("8. Checking for Duplicates")
    
    with open('templates/index.html', 'r') as f:
        html_content = f.read()
    
    # Count DOCTYPE declarations
    doctype_count = html_content.count('<!DOCTYPE html>')
    checks = [
        (doctype_count == 1, f"Single HTML document (found {doctype_count} DOCTYPE declarations)"),
        (html_content.count('</html>') == 1, "Single closing </html> tag"),
        (html_content.count('uploadAndProcessFiles') <= 2, "No duplicate JavaScript functions"),
    ]
    
    all_pass = True
    for condition, message in checks:
        all_pass = all_pass and print_check(condition, message)
    
    return all_pass

def main():
    """Run all verification checks."""
    print("\n" + "="*60)
    print("  HuggingFace Spaces Readiness Check for CogniChat")
    print("="*60)
    
    checks = [
        verify_files_exist(),
        verify_upload_folder_config(),
        verify_session_management(),
        verify_error_handling(),
        verify_environment_variables(),
        verify_dockerfile(),
        verify_requirements(),
        verify_no_duplicates(),
    ]
    
    print_header("Summary")
    
    passed = sum(checks)
    total = len(checks)
    
    if all(checks):
        print(f"\n✅ ALL CHECKS PASSED ({passed}/{total})")
        print("\n🚀 Your application is ready for HuggingFace Spaces deployment!")
        print("\nNext steps:")
        print("1. Go to https://huggingface.co/new-space")
        print("2. Select 'Docker' as SDK")
        print("3. Upload all project files")
        print("4. Set GROQ_API_KEY in Space secrets")
        print("5. Wait for build to complete")
        return 0
    else:
        print(f"\n SOME CHECKS FAILED ({total - passed}/{total} issues)")
        print("\n Please fix the issues above before deploying.")
        print("\nFor detailed guidance, see:")
        print("- HF_SPACES_FILE_STORAGE_GUIDE.md")
        print("- DEPLOYMENT.md")
        print("- HF_SPACES_CHECKLIST.md")
        return 1

if __name__ == '__main__':
    exit_code = main()
    sys.exit(exit_code)