novafarma/scripts/cleanup_duplicates.py

#!/usr/bin/env python3
"""
🧹 CLEANUP DUPLICATES SCRIPT
Keep only GREEN/PROCESSED versions, remove all backups and duplicates
"""

import os
import hashlib
from pathlib import Path
from collections import defaultdict

# Base directory
BASE_DIR = Path("/Users/davidkotnik/repos/novafarma/assets/slike")

# Patterns to DELETE (backup and temporary files)
DELETE_PATTERNS = [
    "assets_BACKUP_",
    "assets__backup_before_greenscreen_",
    "src_assets_library_",
    "_BACKUP_",
    "_ORIGINAL",
    "_ALPHA",
    "_CLEAN",
    "_PROCESSED",
    "_AI",
]

# Patterns to KEEP (prefer these)
KEEP_PATTERNS = [
    "MOJE_SLIKE_KONCNA_",
    "assets_PHASE_PACKS_",
    "assets_sprites_",
    "assets_crops_",
]

def should_delete(filename):
    """Check if file should be deleted based on patterns"""
    for pattern in DELETE_PATTERNS:
        if pattern in filename:
            return True
    return False

def get_priority(filename):
    """Get priority score (higher = keep)"""
    # Prefer MOJE_SLIKE_KONCNA (final processed versions)
    if "MOJE_SLIKE_KONCNA_" in filename:
        return 100
    
    # Then PHASE_PACKS (organized versions)
    if "assets_PHASE_PACKS_" in filename:
        return 50
    
    # Then sprites
    if "assets_sprites_" in filename:
        return 40
    
    # Then crops
    if "assets_crops_" in filename:
        return 30
    
    # Simple names (no prefix) are good
    if not filename.startswith(("assets_", "src_", "MOJE_")):
        return 20
    
    # Everything else
    return 0

def get_base_name(filename):
    """Extract base name without prefixes"""
    # Remove common prefixes
    name = filename
    
    for prefix in ["MOJE_SLIKE_KONCNA_", "assets_BACKUP_20260112_064319_", 
                   "assets_PHASE_PACKS_", "assets__backup_before_greenscreen_",
                   "assets_sprites_", "assets_crops_", "src_assets_library_godot_"]:
        if name.startswith(prefix):
            name = name[len(prefix):]
    
    # Remove numbers at end (timestamps)
    import re
    name = re.sub(r'_\d{13,}', '', name)
    
    return name

def cleanup_directory(directory):
    """Cleanup duplicates in a directory"""
    if not directory.exists():
        return
    
    print(f"\n🔍 Scanning: {directory.relative_to(BASE_DIR)}")
    
    # Group files by base name
    groups = defaultdict(list)
    
    for file_path in directory.rglob("*.png"):
        if file_path.is_file():
            base_name = get_base_name(file_path.name)
            groups[base_name].append(file_path)
    
    deleted_count = 0
    kept_count = 0
    
    # Process each group
    for base_name, files in groups.items():
        if len(files) <= 1:
            kept_count += 1
            continue
        
        # Sort by priority (highest first)
        files_sorted = sorted(files, key=lambda f: get_priority(f.name), reverse=True)
        
        # Keep the highest priority file
        keep_file = files_sorted[0]
        
        # Delete the rest
        for file_path in files_sorted[1:]:
            # Also delete if it matches delete patterns
            if should_delete(file_path.name):
                try:
                    file_path.unlink()
                    deleted_count += 1
                    print(f"  ❌ Deleted: {file_path.name}")
                except Exception as e:
                    print(f"  ⚠️  Error deleting {file_path.name}: {e}")
            else:
                # Check if it's truly a duplicate
                if get_priority(file_path.name) < get_priority(keep_file.name):
                    try:
                        file_path.unlink()
                        deleted_count += 1
                        print(f"  ❌ Deleted duplicate: {file_path.name}")
                    except Exception as e:
                        print(f"  ⚠️  Error: {e}")
        
        kept_count += 1
        print(f"  ✅ Kept: {keep_file.name}")
    
    return deleted_count, kept_count

def cleanup_all():
    """Cleanup entire slike directory"""
    print("🧹 DUPLICATE CLEANUP SCRIPT")
    print("="*60)
    print("Strategy: Keep MOJE_SLIKE_KONCNA (green screen versions)")
    print("         Delete all backups and duplicates")
    print("="*60)
    
    total_deleted = 0
    total_kept = 0
    
    # Process all subdirectories
    for subdir in ["items", "predmeti", "liki", "biomi", "teren", "okolje", "dekoracije"]:
        dir_path = BASE_DIR / subdir
        if dir_path.exists():
            deleted, kept = cleanup_directory(dir_path)
            total_deleted += deleted
            total_kept += kept
    
    print(f"\n{'='*60}")
    print(f"✅ Total kept: {total_kept}")
    print(f"❌ Total deleted: {total_deleted}")
    print(f"💾 Space saved: ~{total_deleted * 300 // 1024} MB (estimated)")
    print(f"{'='*60}\n")

if __name__ == "__main__":
    cleanup_all()
    print("✨ Done!")