gzhu-biyesheji/schema/dataset.py
carry 41447c5ed4 feat(dataset): 添加数据集来源文档字段
- 在 dataset 模型中增加 source_doc 字段,用于记录数据集的来源文档
- 新增字段为可选列表,包含 doc 类型的元素
2025-04-09 17:37:24 +08:00

29 lines
1.4 KiB
Python

from typing import Optional
from pydantic import BaseModel, Field
from datetime import datetime, timezone
class doc(BaseModel):
id: Optional[int] = Field(default=None, description="文档ID")
name: str = Field(default="", description="文档名称")
path: str = Field(default="", description="文档路径")
markdown_files: list[str] = Field(default_factory=list, description="文档路径列表")
class Q_A(BaseModel):
question: str = Field(default="", min_length=1,description="问题")
answer: str = Field(default="", min_length=1, description="答案")
class dataset_item(BaseModel):
id: Optional[int] = Field(default=None, description="数据集项ID")
message: list[Q_A] = Field(description="数据集项内容")
class dataset(BaseModel):
id: Optional[int] = Field(default=None, description="数据集ID")
name: Optional[str] = Field(default=None, description="数据集名称")
model_id: Optional[list[str]] = Field(default=None, description="数据集使用的模型ID")
source_doc: Optional[list[doc]] = Field(default=None, description="数据集来源文档")
description: Optional[str] = Field(default="", description="数据集描述")
created_at: datetime = Field(
default_factory=lambda: datetime.now(timezone.utc),
description="记录创建时间"
)
dataset_items: list[dataset_item] = Field(default_factory=list, description="数据集项列表")