cleanup_no_backup.py

#!/usr/bin/env python3
"""
Project Cleanup Script - NO BACKUP VERSION
Removes redundant files and directories to reduce project size.
WARNING: This script does NOT create backups. Use with caution.
"""

import os
import shutil

def safe_remove(path):
    """Safely remove a file or directory"""
    try:
        if os.path.isfile(path):
            os.remove(path)
            print(f"  ✅ Removed file: {path}")
            return True
        elif os.path.isdir(path):
            shutil.rmtree(path)
            print(f"  ✅ Removed directory: {path}")
            return True
        else:
            print(f"  ⚠️ Not found: {path}")
            return False
    except Exception as e:
        print(f"  ❌ Error removing {path}: {e}")
        return False

def cleanup_project():
    """Remove redundant files and directories"""
    
    # OLD MODELS TO REMOVE
    old_models = [
        "bert_hallucination_model",
        "bias_fixed_mega_model", 
        "competition_model",
        "enhanced_contradiction_model",
        "enhanced_hallucination_model",
        "enhanced_hallucination_model_v2",
        "enhanced_model",
        "enhanced_v2_model",
        "final_balanced_mega_model",
        "fine_tuned_hallucination_model",
        "fixed_enhanced_model",
        "mega_hallucination_model_20250814_151158",
        "progressive_model_1",
        "progressive_model_2", 
        "progressive_model_3",
        "progressive_stage_1_20250814_155617",
        "progressive_stage_2_20250814_155850",
        "progressive_stage_3_20250814_160640",
        "stable_mega_enhanced_model",
        "stable_mega_model_20250814_155042",
        "cv_fold_0",
        "ensemble_models",
        "evaluation_results",
        "model_checkpoints",
        "training_checkpoints"
    ]
    
    # OLD TRAINING SCRIPTS TO REMOVE
    old_training_scripts = [
        "train_bert_alternative.py",
        "train_bias_fix.py",
        "train_competition_model.py",
        "train_enhanced_contradiction.py",
        "train_enhanced_model.py",
        "train_enhanced_performance.py",
        "train_final_balance.py",
        "train_fixed_enhanced.py",
        "train_incremental_improvement.py",
        "train_mega_model.py",
        "train_progressive_complete.py",
        "train_simple_competition.py",
        "train_simple_model.py",
        "train_simple_production.py",
        "train_simple_working_model.py",
        "train_smart_incremental.py",
        "train_stable_mega.py",
        "train_working_enhanced.py"
    ]
    
    # OLD TEST SCRIPTS TO REMOVE  
    old_test_scripts = [
        "test_all_models.py",
        "test_bias_fixed.py",
        "test_complete_mega_model.py",
        "test_detector_direct.py",
        "test_device_patterns.py",
        "test_enhanced_api.py",
        "test_enhanced_model.py",
        "test_enhanced_v2_comprehensive.py",
        "test_final_balance.py",
        "test_fresh_predictions.py",
        "test_hybrid_accuracy.py",
        "test_mega_final.py",
        "test_mega_model.py",
        "test_model.py",
        "test_raw_model.py",
        "test_regex_fix.py",
        "test_reliability.py",
        "test_stable_mega.py"
    ]
    
    # DEBUG/DEVELOPMENT FILES TO REMOVE
    debug_files = [
        "debug_prediction.py",
        "debug_rules.py", 
        "apply_quick_fix.py",
        "hybrid_detector.py",
        "standalone_test.py",
        "simple_test_server.py",
        "comprehensive_server_test.py",
        "comprehensive_test.py"
    ]
    
    # OLD DATA FILES TO REMOVE
    old_data_files = [
        "advanced_training_data.csv",
        "bias_fix_training.csv",
        "combined_training_data.csv",
        "competition_training_data.csv",
        "edge_cases_training_data.csv",
        "halueval_combined_training.csv",
        "halueval_training_prepared.csv",
        "mega_training_data.csv",
        "sample_training_data.csv"
    ]
    
    # CACHE/TEMP FILES TO REMOVE
    cache_files = [
        "cache_persistent.pkl",
        "app.log",
        "__pycache__",
        ".pytest_cache",
        "logs",
        "tmp", 
        "model_cache",
        "plots"
    ]
    
    # OLD ANALYSIS SCRIPTS TO REMOVE
    analysis_scripts = [
        "analyze_halueval.py",
        "analyze_training_metrics.py",
        "complete_data_audit.py",
        "create_verified_dataset.py",
        "diagnose_bias.py",
        "fact_check_report.py",
        "monitor_training.py",
        "training_summary.py",
        "verify_fixes_and_retrain.py",
        "analyze_cleanup.py"
    ]
    
    # CONFIG FILES TO REMOVE
    config_files = [
        "competition_config.json",
        "monitoring_config.json",
        "setup_report.json",
        "requirements_competition.txt"
    ]
    
    # BATCH FILES TO REMOVE (keeping only essential ones)
    batch_files = [
        "run_simple_training.bat",
        "run_training.bat"
    ]
    
    # DOCUMENTATION TO REMOVE (keeping only README.md)
    docs_files = [
        "COMPETITION_README.md",
        "RELIABILITY_ANALYSIS.md",
        "setup_instructions.md"
    ]
    
    # ALL FILES TO REMOVE
    all_files_to_remove = (
        old_models + old_training_scripts + old_test_scripts + 
        debug_files + old_data_files + cache_files + 
        analysis_scripts + config_files + batch_files + docs_files
    )
    
    print("🧹 STARTING PROJECT CLEANUP (NO BACKUP)")
    print("=" * 60)
    print("⚠️  WARNING: No backup will be created!")
    print("🎯 Target: Remove ~90 redundant files/directories")
    print("✅ Keep: Essential production files only")
    print("=" * 60)
    
    removed_count = 0
    
    for item in all_files_to_remove:
        if os.path.exists(item):
            if safe_remove(item):
                removed_count += 1
        
    print("=" * 60)
    print(f"🎉 Cleanup completed! Removed {removed_count} items.")
    print("\n📁 REMAINING ESSENTIAL FILES:")
    print("   ✅ app/ (core application)")
    print("   ✅ complete_halueval_model/ (working AI model)")
    print("   ✅ comprehensive_training_data.csv (main training data)")
    print("   ✅ training.csv (backup training data)")
    print("   ✅ test_comprehensive_hybrid.py (main test)")
    print("   ✅ test_api.py (API tests)")
    print("   ✅ test_mega_accuracy.py (accuracy test)")
    print("   ✅ requirements.txt (dependencies)")
    print("   ✅ config.yaml (configuration)")
    print("   ✅ README.md (documentation)")
    print("   ✅ Dockerfile & docker-compose.yml (deployment)")
    print("   ✅ prepare_training_data.py (training utility)")
    print("   ✅ combine_training_data.py (data utility)")
    print("   ✅ run_server.bat (server script)")
    print("   ✅ setup_competition.py (setup utility)")
    
    print(f"\n💾 Project size significantly reduced!")
    print(f"🗑️  Removed {removed_count} redundant files")
    
    return removed_count

def main():
    """Main function"""
    print("PROJECT CLEANUP TOOL - NO BACKUP VERSION")
    print("=" * 60)
    print("⚠️  WARNING: This will permanently delete files!")
    print("📦 This will remove ~90 redundant files to save space")
    print("✅ Essential production files will be preserved")
    print("❌ NO BACKUP will be created")
    print()
    
    # Ask for confirmation
    response = input("Are you SURE you want to proceed? (type 'YES' to confirm): ")
    if response != 'YES':
        print("❌ Cleanup cancelled.")
        return
    
    # Perform cleanup
    removed_count = cleanup_project()
    
    print(f"\n🎉 Project cleanup completed successfully!")
    print(f"📊 Removed {removed_count} redundant items")
    print("🔧 All essential files preserved")
    print("💾 Project size significantly reduced")
    
    # Clean up this script too
    try:
        os.remove("cleanup_no_backup.py")
        print("🧹 Cleanup script removed")
    except:
        pass

if __name__ == "__main__":
    main()