Function batch_create_training

backend/apps/data_training/curd/data_training.py:262–398

批量创建数据训练记录（复用单条插入逻辑）

(session: SessionDep, info_list: List[DataTrainingInfo], oid: int, trans: Trans)

def batch_create_training(session: SessionDep, info_list: List[DataTrainingInfo], oid: int, trans: Trans):
    """
    批量创建数据训练记录（复用单条插入逻辑）
    """
    if not info_list:
        return {
            'success_count': 0,
            'failed_records': [],
            'duplicate_count': 0,
            'original_count': 0,
            'deduplicated_count': 0
        }

    failed_records = []
    success_count = 0
    inserted_ids = []

    # 第一步：数据去重
    unique_records = {}
    duplicate_records = []

    for info in info_list:
        # 创建唯一标识
        unique_key = (
            info.question.strip().lower() if info.question else "",
            info.datasource_name.strip().lower() if info.datasource_name else "",
            info.advanced_application_name.strip().lower() if info.advanced_application_name else ""
        )

        if unique_key in unique_records:
            duplicate_records.append(info)
        else:
            unique_records[unique_key] = info

    # 将去重后的数据转换为列表
    deduplicated_list = list(unique_records.values())

    # 预加载数据源和高级应用名称到ID的映射
    datasource_name_to_id = {}
    datasource_stmt = select(CoreDatasource.id, CoreDatasource.name).where(CoreDatasource.oid == oid)
    datasource_result = session.execute(datasource_stmt).all()
    for ds in datasource_result:
        datasource_name_to_id[ds.name.strip()] = ds.id

    assistant_name_to_id = {}

    assistant_stmt = select(AssistantModel.id, AssistantModel.name).where(and_(AssistantModel.type == 1, AssistantModel.oid == oid))
    assistant_result = session.execute(assistant_stmt).all()
    for assistant in assistant_result:
        assistant_name_to_id[assistant.name.strip()] = assistant.id

    # 验证和转换数据
    valid_records = []
    for info in deduplicated_list:
        error_messages = []

        # 基本验证
        if not info.question or not info.question.strip():