intellif-aihub 0.1.5__py3-none-any.whl → 0.1.6__py3-none-any.whl

aihub/__init__.py

@@ -1 +1 @@
-__version__ = "0.1.5"
+__version__ = "0.1.6"

aihub/client.py

@@ -6,45 +6,64 @@ import httpx
 
 from .exceptions import APIError
 from .services import artifact
+from .services import data_warehouse
 from .services import dataset_management
 from .services import document_center
+from .services import eval
 from .services import labelfree
+from .services import model_center
 from .services import model_training_platform
 from .services import quota_schedule_management
 from .services import tag_resource_management
 from .services import task_center
 from .services import user_system
+from .services import workflow_center
 from .services.artifact import ArtifactService
+from .services.data_warehouse import DataWarehouseService
 from .services.dataset_management import DatasetManagementService
 from .services.document_center import DocumentCenterService
+from .services.eval import EvalService
 from .services.labelfree import LabelfreeService
+from .services.model_center import ModelCenterService
 from .services.model_training_platform import ModelTrainingPlatformService
 from .services.quota_schedule_management import QuotaScheduleManagementService
 from .services.tag_resource_management import TagResourceManagementService
 from .services.task_center import TaskCenterService
 from .services.user_system import UserSystemService
+from .services.workflow_center import WorkflowCenterService
 
 
 class Client:
     """AI-HUB python SDK 客户端
 
     Attributes:
+        artifact (ArtifactService): 制品管理服务
+        data_warehouse (DataWarehouseService): 数据仓库服务
         dataset_management (DatasetManagementService): 数据集管理服务
+        document_center (DocumentCenterService): 文档中心服务
+        eval (EvalService): 评测服务
         labelfree (LabelfreeService): 标注服务
-        task_center (TaskCenterService): 任务中心
-        artifact (ArtifactService): 制品管理
-
+        model_center (ModelCenterService): 模型中心服务
+        model_training_platform (ModelTrainingPlatformService): 模型训练服务
+        quota_schedule_management (QuotaScheduleManagementService): 配额调度服务
+        tag_resource_management (TagResourceManagementService): 标签管理服务
+        task_center (TaskCenterService): 任务中心服务
+        workflow_center (WorkflowCenterService): 工作流服务
     """
 
+    artifact: ArtifactService = None
+    data_warehouse: DataWarehouseService = None
+    dataset_management: DatasetManagementService = None
+    document_center: DocumentCenterService = None
+    eval: EvalService = None
     labelfree: LabelfreeService = None
+    model_center: ModelCenterService = None
+    model_training_platform: ModelTrainingPlatformService = None
+    quota_schedule_management: QuotaScheduleManagementService = None
     tag_resource_management: TagResourceManagementService = None
-    document_center: DocumentCenterService = None
     task_center: TaskCenterService = None
-    dataset_management: DatasetManagementService = None
-    artifact: ArtifactService = None
     user_system: UserSystemService = None
-    model_training_platform: ModelTrainingPlatformService = None
-    quota_schedule_management: QuotaScheduleManagementService = None
+    workflow_center: WorkflowCenterService = None
 
     def __init__(self, *, base_url: str, token: str | None = None, timeout: float = 60.0):
         """AI-HUB python SDK 客户端
@@ -71,15 +90,19 @@ class Client:
             headers={"Authorization": f"Bearer {token}"},
             # event_hooks={"response": [self._raise_for_status]},
         )
+        self.artifact = artifact.ArtifactService(self._http)
+        self.data_warehouse = data_warehouse.DataWarehouseService(self._http)
         self.dataset_management = dataset_management.DatasetManagementService(self._http)
+        self.document_center = document_center.DocumentCenterService(self._http)
+        self.eval = eval.EvalService(self._http)
         self.labelfree = labelfree.LabelfreeService(self._http)
+        self.model_center = model_center.ModelCenterService(self._http)
+        self.model_training_platform = model_training_platform.ModelTrainingPlatformService(self._http)
+        self.quota_schedule_management = quota_schedule_management.QuotaScheduleManagementService(self._http)
         self.tag_resource_management = tag_resource_management.TagResourceManagementService(self._http)
-        self.document_center = document_center.DocumentCenterService(self._http)
         self.task_center = task_center.TaskCenterService(self._http)
-        self.artifact = artifact.ArtifactService(self._http)
         self.user_system = user_system.UserSystemService(self._http)
-        self.model_training_platform = model_training_platform.ModelTrainingPlatformService(self._http)
-        self.quota_schedule_management = quota_schedule_management.QuotaScheduleManagementService(self._http)
+        self.workflow_center = workflow_center.WorkflowCenterService(self._http)
 
     @staticmethod
     def _raise_for_status(r: httpx.Response):

aihub/models/data_warehouse.py

@@ -0,0 +1,95 @@
+from __future__ import annotations
+
+from enum import IntEnum
+from typing import List, Optional
+
+from pydantic import BaseModel, Field
+
+
+class SearchType(IntEnum):
+    """搜索类型"""
+    SQL = 1  # SQL
+    IMAGE = 2  # 图片
+    TEXT = 3  # 文本
+
+
+class SearchStatus(IntEnum):
+    """搜索任务状态"""
+    WAITING = 1  # 等待中
+    RUNNING = 2  # 运行中
+    SUCCESS = 3  # 成功
+    FAIL = 4  # 失败
+
+
+class Image(BaseModel):
+    """检索图片对象"""
+    url: str = Field(alias="url", description="URL")
+    box: Optional[List[int]] = Field(default_factory=list, description="检测框坐标 [x1, y1, x2, y2]")
+
+
+class SearchResult(BaseModel):
+    """单条检索结果"""
+    image_url: str = Field(alias="image_url", description="图片URL")
+    label: str = Field(description="标签")
+    box: List[int] = Field(description="检测框坐标 [x1, y1, x2, y2]")
+    score: float = Field(description="分数")
+
+
+class Search(BaseModel):
+    """搜索任务完整信息"""
+    id: int = Field(description="任务ID")
+    type: SearchType = Field(description="搜索类型")
+    name: str = Field(description="任务名称")
+    description: str = Field(description="任务描述")
+    sql: str = Field(description="SQL语句")
+    feature_lib_id: int = Field(alias="feature_lib_id", description="特征库ID")
+    feature_lib_name: str = Field(alias="feature_lib_name", description="特征库名称")
+    images: Optional[List[Image]] = Field(default_factory=list, description="搜索图片列表")
+    keywords: str = Field(description="关键词")
+    top_k: int = Field(alias="top_k", description="返回前K个候选")
+    status: SearchStatus = Field(description="任务状态")
+    message: str = Field(description="信息")
+    result_url: str = Field(alias="result_url", description="结果地址")
+    results: Optional[List[SearchResult]] = Field(default_factory=list, description="检索结果列表")
+    created_at: int = Field(alias="created_at", description="创建时间戳 (ms)")
+    username: str = Field(description="创建人用户名")
+
+    model_config = {"use_enum_values": True}
+
+
+class ListSearchRequest(BaseModel):
+    """分页搜索任务请求"""
+    page_size: int = Field(20, alias="page_size", description="每页数量")
+    page_num: int = Field(1, alias="page_num", description="当前页码，从1开始")
+    name: Optional[str] = Field(None, description="名称过滤")
+    status: Optional[SearchStatus] = Field(None, description="状态过滤")
+    user_id: Optional[int] = Field(None, alias="user_id", description="用户ID过滤")
+
+    model_config = {"use_enum_values": True}
+
+
+class ListSearchResponse(BaseModel):
+    """分页搜索任务返回"""
+    total: int = Field(description="总条数")
+    page_size: int = Field(alias="page_size", description="每页数量")
+    page_num: int = Field(alias="page_num", description="当前页码")
+    data: Optional[List[Search]] = Field(default_factory=list, description="搜索任务列表")
+
+
+class CreateSearchRequest(BaseModel):
+    """创建搜索任务请求"""
+    type: SearchType = Field(description="搜索类型")
+    name: str = Field(description="名称")
+    description: Optional[str] = Field(None, description="描述")
+    sql: Optional[str] = Field(None, description="SQL语句")
+    feature_lib_id: Optional[int] = Field(None, alias="feature_lib_id", description="特征库ID")
+    images: Optional[List[Image]] = Field(default_factory=list, description="搜索图片列表")
+    keywords: Optional[str] = Field(None, description="关键词")
+    top_k: Optional[int] = Field(None, alias="top_k", description="返回前K条")
+
+    model_config = {"use_enum_values": True}
+
+
+class CreateSearchResponse(BaseModel):
+    """创建搜索任务返回"""
+    id: int = Field(description="任务ID")

aihub/models/dataset_management.py

@@ -1,99 +1,137 @@
 from __future__ import annotations
 
+from enum import IntEnum
 from typing import List, Optional
 
 from pydantic import BaseModel, Field
 
 
+class DatasetVersionStatus(IntEnum):
+    """数据集版本状态"""
+    Waiting = 1  # 等待中
+    Running = 2  # 运行中
+    Success = 3  # 成功
+    Fail = 4  # 失败
+    MetaDB = 5  # 加载meta
+    BuildIndex = 6  # 构建index
+
+
+class UploadType(IntEnum):
+    """上传类型"""
+    LOCAL = 1  # 本地上传
+    SERVER_PATH = 3  # 服务器路径上传
+    LABELFREE = 4  # Labelfree
+    DATA_INGEST = 5  # 数据接入
+
+
 class CreateDatasetRequest(BaseModel):
-    name: str
-    description: str
-    tags: List[int]
-    cover_img: Optional[str] = Field(None, alias="cover_img")
-    create_by: Optional[int] = Field(None, alias="create_by")
-    is_private: Optional[bool] = Field(None, alias="is_private")
-    access_user_ids: Optional[List[int]] = Field(None, alias="access_user_ids")
+    """创建数据集请求"""
+    name: str = Field(description="数据集名称")
+    description: str = Field(description="数据集描述")
+    tags: List[int] = Field(description="标签ID列表")
+    cover_img: Optional[str] = Field(None, alias="cover_img", description="封面图片URL")
+    create_by: Optional[int] = Field(None, alias="create_by", description="创建人")
+    is_private: Optional[bool] = Field(None, alias="is_private", description="是否私有")
+    access_user_ids: Optional[List[int]] = Field(None, alias="access_user_ids", description="具有访问权限的用户ID列表")
 
 
 class CreateDatasetResponse(BaseModel):
-    id: int = Field(alias="id")
+    """创建数据集返回"""
+    id: int = Field(alias="id", description="数据集ID")
 
 
 class DatasetVersionBase(BaseModel):
-    id: int
-    version: int
-    status: int
-    parquet_index_path: Optional[str] = Field(None, alias="parquet_index_path")
-    data_count: int = Field(alias="data_count")
+    """数据集版本概要"""
+    id: int = Field(description="版本ID")
+    version: int = Field(description="版本号")
+    status: DatasetVersionStatus = Field(description="版本状态")
+    parquet_index_path: Optional[str] = Field(None, alias="parquet_index_path", description="parquet索引文件路径")
+    data_count: int = Field(alias="data_count", description="数量")
+
+    model_config = {"use_enum_values": True}
 
 
 class DatasetDetail(BaseModel):
-    id: int
-    name: str
-    description: str
-    cover_img: Optional[str] = Field(None, alias="cover_img")
-    created_at: int = Field(alias="created_at")
-    updated_at: int = Field(alias="update_at")
-    user_id: int = Field(alias="user_id")
-    username: str
-    tags: List[int]
-    access_user_ids: Optional[List[int]] = Field(None, alias="access_user_ids")
-    is_private: Optional[bool] = Field(None, alias="is_private")
-    versions: List[DatasetVersionBase]
+    """数据集详情"""
+    id: int = Field(description="数据集 ID")
+    name: str = Field(description="名称")
+    description: str = Field(description="描述")
+    cover_img: Optional[str] = Field(None, alias="cover_img", description="封面图片URL")
+    created_at: int = Field(alias="created_at", description="创建时间戳 (ms)")
+    updated_at: int = Field(alias="update_at", description="更新时间戳 (ms)")
+    user_id: int = Field(alias="user_id", description="创建人ID")
+    username: str = Field(description="创建人用户名")
+    tags: List[int] = Field(description="标签ID列表")
+    access_user_ids: Optional[List[int]] = Field(None, alias="access_user_ids", description="可访问的用户ID列表")
+    is_private: Optional[bool] = Field(None, alias="is_private", description="是否私有")
+    versions: List[DatasetVersionBase] = Field(description="版本列表")
 
 
 class ExtInfo(BaseModel):
-    rec_file_path: Optional[str] = Field(None, alias="rec_file_path")
-    idx_file_path: Optional[str] = Field(None, alias="idx_file_path")
-    json_file_path: Optional[str] = Field(None, alias="json_file_path")
-    image_dir_path: Optional[str] = Field(None, alias="image_dir_path")
+    """扩展信息"""
+    rec_file_path: Optional[str] = Field(None, alias="rec_file_path", description="rec文件路径")
+    idx_file_path: Optional[str] = Field(None, alias="idx_file_path", description="idx文件路径")
+    json_file_path: Optional[str] = Field(None, alias="json_file_path", description="json文件路径")
+    image_dir_path: Optional[str] = Field(None, alias="image_dir_path", description="图片目录路径")
 
 
 class CreateDatasetVersionRequest(BaseModel):
-    upload_path: str = Field(alias="upload_path")
-    description: Optional[str] = None
-    dataset_id: int = Field(alias="dataset_id")
-    object_cnt: Optional[int] = Field(None, alias="object_cnt")
-    data_size: Optional[int] = Field(None, alias="data_size")
-    create_by: Optional[int] = Field(None, alias="create_by")
-    upload_type: Optional[int] = Field(4, alias="upload_type")
-    ext_info: Optional[ExtInfo] = Field(None, alias="ext_info")
+    """创建版本请求"""
+    upload_path: str = Field(alias="upload_path", description="上传路径")
+    description: Optional[str] = Field(None, description="版本描述")
+    dataset_id: int = Field(alias="dataset_id", description="数据集ID")
+    object_cnt: Optional[int] = Field(None, alias="object_cnt", description="对象数")
+    data_size: Optional[int] = Field(None, alias="data_size", description="数据大小")
+    create_by: Optional[int] = Field(None, alias="create_by", description="创建人")
+    upload_type: Optional[UploadType] = Field(alias="upload_type", description="上传类型")
+    ext_info: Optional[ExtInfo] = Field(None, alias="ext_info", description="扩展文件信息")
+
+    model_config = {"use_enum_values": True}
 
 
 class CreateDatasetVersionResponse(BaseModel):
-    id: int = Field(alias="id")
+    """创建版本返回"""
+    id: int = Field(alias="id", description="版本ID")
 
 
 class UploadDatasetVersionRequest(BaseModel):
-    upload_path: str = Field(alias="upload_path")
-    upload_type: int = Field(alias="upload_type")
-    dataset_id: int = Field(alias="dataset_id")
-    parent_version_id: Optional[int] = Field(None, alias="parent_version_id")
-    description: Optional[str] = Field(None, alias="description")
+    """上传数据集版本请求"""
+    upload_path: str = Field(alias="upload_path", description="上传目录")
+    upload_type: UploadType = Field(alias="upload_type", description="上传类型")
+    dataset_id: int = Field(alias="dataset_id", description="数据集ID")
+    parent_version_id: Optional[int] = Field(None, alias="parent_version_id", description="父版本ID")
+    description: Optional[str] = Field(None, alias="description", description="版本描述")
+
+    model_config = {"use_enum_values": True}
 
 
 class UploadDatasetVersionResponse(BaseModel):
-    id: int = Field(alias="id")
+    """上传数据集版本返回"""
+    id: int = Field(alias="id", description="版本ID")
 
 
 class DatasetVersionDetail(BaseModel):
-    id: int
-    version: int
-    dataset_id: int = Field(alias="dataset_id")
-    upload_path: str = Field(alias="upload_path")
-    upload_type: int = Field(alias="upload_type")
-    parent_version_id: Optional[int] = Field(None, alias="parent_version_id")
-    description: Optional[str] = None
-    status: int
-    message: Optional[str] = None
-    created_at: int = Field(alias="created_at")
-    user_id: int = Field(alias="user_id")
-    data_size: Optional[int] = Field(None, alias="data_size")
-    data_count: Optional[int] = Field(None, alias="data_count")
-    parquet_index_path: Optional[str] = Field(None, alias="parquet_index_path")
-    ext_info: Optional[ExtInfo] = Field(None, alias="ext_info")
+    """数据集版本详情"""
+    id: int = Field(description="版本ID")
+    version: int = Field(description="版本号")
+    dataset_id: int = Field(alias="dataset_id", description="数据集ID")
+    upload_path: str = Field(alias="upload_path", description="上传路径")
+    upload_type: UploadType = Field(alias="upload_type", description="上传类型")
+    parent_version_id: Optional[int] = Field(None, alias="parent_version_id", description="父版本ID")
+    description: Optional[str] = Field(None, description="版本描述")
+    status: DatasetVersionStatus = Field(description="状态")
+    message: Optional[str] = Field(None, description="信息")
+    created_at: int = Field(alias="created_at", description="创建时间戳 (ms)")
+    user_id: int = Field(alias="user_id", description="创建人ID")
+    data_size: Optional[int] = Field(None, alias="data_size", description="数据大小")
+    data_count: Optional[int] = Field(None, alias="data_count", description="条数")
+    parquet_index_path: Optional[str] = Field(None, alias="parquet_index_path", description="parquet索引文件路径")
+    ext_info: Optional[ExtInfo] = Field(None, alias="ext_info", description="扩展信息")
+
+    model_config = {"use_enum_values": True}
 
 
 class FileUploadData(BaseModel):
-    path: str
-    url: str
+    """文件上传数据"""
+    path: str = Field(description="路径")
+    url: str = Field(description="URL")

aihub/models/document_center.py

@@ -1,28 +1,36 @@
-# !/usr/bin/env python
-# -*-coding:utf-8 -*-
-
 from __future__ import annotations
 
+from enum import IntEnum
 from typing import List, Optional, Any
 
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
+
+
+class DocumentType(IntEnum):
+    """文档类型"""
+    MARKDOWN = 0  # Markdown
+    PDF = 1  # 飞书文档生成的pdf
 
 
 class Document(BaseModel):
-    id: int
-    title: str
-    type: int
-    edit_time: int
-    need_update: bool
-    content: str
-    username: str
-    user_id: int
-    created_at: int
-    comments: Optional[Any] = None
+    """文档"""
+    id: int = Field(description="文档ID")
+    title: str = Field(description="标题")
+    type: DocumentType = Field(description="文档类型")
+    edit_time: int = Field(alias="edit_time", description="编辑时间")
+    need_update: bool = Field(alias="need_update", description="是否需要更新")
+    content: str = Field(description="内容")
+    username: str = Field(description="用户名")
+    user_id: int = Field(alias="user_id", description="作者ID")
+    created_at: int = Field(alias="created_at", description="创建时间戳 (ms)")
+    comments: Optional[Any] = Field(None, description="评论")
+
+    model_config = {"use_enum_values": True}
 
 
 class GetDocumentsResponse(BaseModel):
-    total: int
-    page_size: int
-    page_num: int
-    data: List[Document]
+    """分页文档返回"""
+    total: int = Field(description="总条数")
+    page_size: int = Field(alias="page_size", description="每页数量")
+    page_num: int = Field(alias="page_num", description="当前页码")
+    data: List[Document] = Field(description="文档列表")

aihub/models/eval.py

@@ -1,17 +1,26 @@
 # !/usr/bin/env python
 # -*-coding:utf-8 -*-
-from pydantic import BaseModel
+from typing import Dict
 
+from pydantic import BaseModel, Field
 
-class CreatEvalReq(BaseModel):
-    dataset_id: int
-    dataset_version_id: int
-    prediction_artifact_path: str
-    evaled_artifact_path: str
-    run_id: str
-    user_id: int = 0
-    report: dict = {}
 
+class CreateEvalReq(BaseModel):
+    """创建评测任务"""
+    dataset_id: int = Field(description="数据集ID")
+    dataset_version_id: int = Field(description="数据集版本ID")
+    prediction_artifact_path: str = Field(description="推理产物的路径")
+    evaled_artifact_path: str = Field(description="评测结果产物的路径")
+    run_id: str = Field(description="运行ID")
+    user_id: int = Field(0, description="用户ID")
+    report: Dict = Field(default_factory=dict, description="评测报告")
 
-class CreatEvalResp(BaseModel):
-    id: int
+
+class EvalRun(BaseModel):
+    """评测任务的运行实体"""
+    id: int = Field(description="评测的运行ID")
+
+
+class CreateEvalResp(BaseModel):
+    """创建评测任务的返回结果"""
+    eval_run: EvalRun = Field(alias="eval_run", description="评测运行信息")

aihub/models/labelfree.py

@@ -7,47 +7,21 @@ from pydantic import BaseModel, Field
 
 class Stats(BaseModel):
     """标注统计信息"""
-
     total_annotations: int = Field(alias="total_annotations", description="总数据量")
-    labeled_annotations: int = Field(
-        alias="labeled_annotations", description="已标注数据量"
-    )
+    labeled_annotations: int = Field(alias="labeled_annotations", description="已标注数据量")
     total_labels: int = Field(alias="total_labels", description="总标签量")
-    total_reviews: Optional[int] = Field(
-        None, alias="total_reviews", description="总质检量"
-    )
-    unlabeled_reviews: Optional[int] = Field(
-        None, alias="unlabeled_reviews", description="未质检量"
-    )
-    labeled_reviews: Optional[int] = Field(
-        None, alias="labeled_reviews", description="已质检量"
-    )
-    accepted_count: Optional[int] = Field(
-        None, alias="accepted_count", description="质检通过量"
-    )
-    rejected_count: Optional[int] = Field(
-        None, alias="rejected_count", description="质检打回量"
-    )
+    total_reviews: Optional[int] = Field(None, alias="total_reviews", description="总质检量")
+    unlabeled_reviews: Optional[int] = Field(None, alias="unlabeled_reviews", description="未质检量")
+    labeled_reviews: Optional[int] = Field(None, alias="labeled_reviews", description="已质检量")
+    accepted_count: Optional[int] = Field(None, alias="accepted_count", description="质检通过量")
+    rejected_count: Optional[int] = Field(None, alias="rejected_count", description="质检打回量")
 
 
 class GetGlobalStatsResponse(BaseModel):
-    """
-    标注统计概况
-    """
-
+    """标注统计概况"""
     global_stats: Stats = Field(alias="global_stats")
-    valid_ten_percent: bool = Field(
-        alias="valid_ten_percent", description="是否完成验收10%"
-    )
-    valid_fifty_percent: bool = Field(
-        alias="valid_fifty_percent", description="是否完成验收50%"
-    )
-    valid_hundred_percent: bool = Field(
-        alias="valid_hundred_percent", description="是否完成验收100%"
-    )
-    data_exported_count: int = Field(
-        alias="data_exported_count", description="已导出数据次数"
-    )
-    exported_dataset_name: str = Field(
-        alias="exported_dataset_name", description="最新数据集名称"
-    )
+    valid_ten_percent: bool = Field(alias="valid_ten_percent", description="是否完成验收10%")
+    valid_fifty_percent: bool = Field(alias="valid_fifty_percent", description="是否完成验收50%")
+    valid_hundred_percent: bool = Field(alias="valid_hundred_percent", description="是否完成验收100%")
+    data_exported_count: int = Field(alias="data_exported_count", description="已导出数据次数")
+    exported_dataset_name: str = Field(alias="exported_dataset_name", description="最新数据集名称")