add_normalized_url.py (482B)
1 from gus import constants 2 from gus.lib.db_model import init_db, Page 3 from gus.lib.gemini import GeminiResource, GeminiRobotFileParser 4 5 def main(): 6 db = init_db(f"index.new/{constants.DB_FILENAME}") 7 for page in Page.select(): 8 print(f"\nBefore: {page.normalized_url}") 9 page.normalized_url = GeminiResource(page.url).normalized_url 10 page.save() 11 print(f"After : {page.normalized_url}") 12 print("\nDone!") 13 14 15 if __name__ == "__main__": 16 main()