from ner_config import RepaymentNERConfig, FlightNERConfig, TrainNERConfig # 脚本:校验非法格式 def check_data_labels(): label_set = set() line_num = 0 with open(FlightNERConfig.DATA_PATH, 'r', encoding='utf-8') as f: for line in f: line_num += 1 line = line.strip() if line: try: _, label = line.split(maxsplit=1) if label not in FlightNERConfig.LABELS: print(f"行 {line_num}: 发现非法标签 '{label}'") label_set.add(label) except Exception as e: print(f"行 {line_num}: 处理出错 - {str(e)}") print("\n发现的所有非法标签:") for label in sorted(label_set): print(f"- {label}") if __name__ == "__main__": check_data_labels()