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

class DescriptionExtractProcessor(FileProcessor):
    def process_content(self, content: str) -> str:
        try:
            data = json.loads(content)
            description = data.get('description', '')
            # Clean the description
            description = re.sub(r'<[^>]+>', '', description)  # Remove HTML
            description = re.sub(r'\s+', ' ', description)     # Normalize whitespace
            return description.strip()
        except json.JSONDecodeError as e:
            self.logger.error(f"Failed to parse JSON: {e}")
            return ""

def main():
    parser = argparse.ArgumentParser(description='Extract descriptions from JSON files')
    parser.add_argument('-d', '--directory', default='.', help='Directory to process')
    parser.add_argument('-v', '--verbose', action='store_true', help='Enable verbose logging')
    args = parser.parse_args()

    options = ProcessorOptions(
        recursive=True,
        dry_run=False,
        debug=args.verbose,
        file_extensions={'.json'}
    )
    
    processor = DescriptionExtractProcessor(options)
    processor.process_directory(Path(args.directory))

if __name__ == "__main__":
    main()