import yaml
import os
import shutil

# Paths
SLOT_FIXES_PATH = "/Users/kempersc/apps/glam/data/fixes/slot_fixes.yaml"
SLOTS_DIR = "/Users/kempersc/apps/glam/schemas/20251121/linkml/modules/slots/"
ARCHIVE_DIR = os.path.join(SLOTS_DIR, "archive")

# Ensure archive directory exists
os.makedirs(ARCHIVE_DIR, exist_ok=True)

def get_processed_slots(fixes_path):
    processed_slots = []
    try:
        with open(fixes_path, 'r') as f:
            data = yaml.safe_load(f)
            
        if 'fixes' in data:
            for fix in data['fixes']:
                if fix.get('processed', {}).get('status') == True:
                    # Extract slot name from URL
                    # e.g., https://nde.nl/ontology/hc/slot/activities_societies -> activities_societies
                    original_id = fix.get('original_slot_id', '')
                    if '/slot/' in original_id:
                        slot_name = original_id.split('/slot/')[-1]
                        processed_slots.append(slot_name)
    except Exception as e:
        print(f"Error reading YAML: {e}")
        # Fallback to simple parsing if yaml load fails (e.g. huge file)
        # This is less robust but works for simple structure
        pass
        
    return processed_slots

def main():
    print(f"Reading fixes from {SLOT_FIXES_PATH}...")
    processed_slots = get_processed_slots(SLOT_FIXES_PATH)
    print(f"Found {len(processed_slots)} processed slots in fixes file.")
    
    # Get current slot files
    current_files = os.listdir(SLOTS_DIR)
    
    moved_count = 0
    for slot_name in processed_slots:
        filename = f"{slot_name}.yaml"
        src_path = os.path.join(SLOTS_DIR, filename)
        dst_path = os.path.join(ARCHIVE_DIR, filename)
        
        if os.path.exists(src_path):
            print(f"Moving {filename} to archive...")
            try:
                shutil.move(src_path, dst_path)
                moved_count += 1
            except Exception as e:
                print(f"Failed to move {filename}: {e}")
                
    print(f"Operation complete. Archived {moved_count} files.")

if __name__ == "__main__":
    main()