from utils.file_processor import FileProcessor, ProcessorOptions
import re
from pathlib import Path

class BoyRemovalProcessor(FileProcessor):
    def process_content(self, content: str) -> str:
        # Remove occurrences with comma and space
        content = re.sub(r',\s*([1-9]boy|[1-9]boys|[1-9]girl|[1-9]girls)', '', content)
        # Remove remaining occurrences
        content = re.sub(r'([1-9]boy|[1-9]boys|[1-9]girl|[1-9]girls)', '', content)
        # Clean up commas
        content = re.sub(r',\s*,', ',', content)  # Remove double commas
        content = re.sub(r',\s*$', '', content)   # Remove trailing comma
        content = re.sub(r'^\s*,', '', content)   # Remove leading comma
        content = re.sub(r'\s*,\s*', ', ', content)  # Normalize comma spacing
        return content

def main():
    options = ProcessorOptions(
        recursive=True,
        dry_run=False,
        file_extensions={'.txt'}
    )
    
    processor = BoyRemovalProcessor(options)
    processor.process_directory(Path('.'))

if __name__ == "__main__":
    main()