tamar-file-hub-client 0.1.4__py3-none-any.whl → 0.1.6__py3-none-any.whl

file_hub_client/rpc/protos/file_service.proto

@@ -182,7 +182,8 @@ message BatchDownloadUrlResponse {
 message DownloadUrlInfo {
     string file_id = 1;
     string url = 2;
-    optional string error = 3;
+    string mime_type = 3;
+    optional string error = 4;
 }
 
 message GetGcsUrlResponse {

file_hub_client/schemas/file.py

@@ -1,170 +1,171 @@
-"""
-文件相关数据模型
-"""
-from datetime import datetime
-from typing import Optional, Dict, List, Any
-from pydantic import BaseModel, Field
-
-
-class File(BaseModel):
-    """文件信息模型"""
-    id: str = Field(..., description="文件ID")
-    folder_id: str = Field(..., description="所属文件夹ID")
-    file_name: str = Field(..., description="原始文件名")
-    file_type: str = Field(..., description="文件类型")
-    created_at: datetime = Field(..., description="创建时间")
-    updated_at: datetime = Field(..., description="更新时间")
-
-    class Config:
-        json_encoders = {
-            datetime: lambda v: v.isoformat()
-        }
-
-
-class UploadFile(BaseModel):
-    """上传文件信息模型"""
-    id: str = Field(..., description="上传文件ID")
-    folder_id: str = Field(..., description="所属文件夹ID")
-    storage_type: str = Field(..., description="存储类型")
-    stored_name: str = Field(..., description="存储文件名")
-    stored_path: str = Field(..., description="存储路径")
-    file_id: str = Field(..., description="所属文件ID")
-    file_name: str = Field(..., description="原始文件名")
-    file_size: int = Field(0, description="文件大小（字节）")
-    file_ext: str = Field(..., description="文件后缀")
-    mime_type: str = Field(..., description="MIME类型")
-    created_at: datetime = Field(..., description="创建时间")
-    updated_at: datetime = Field(..., description="更新时间")
-
-    class Config:
-        json_encoders = {
-            datetime: lambda v: v.isoformat()
-        }
-
-
-class FileUploadResponse(BaseModel):
-    """文件上传返回"""
-    file: File = Field(..., description="文件信息")
-    upload_file: UploadFile = Field(..., description="上传文件信息")
-
-
-class UploadUrlResponse(BaseModel):
-    """上传URL响应"""
-    file: File = Field(..., description="文件信息")
-    upload_file: UploadFile = Field(..., description="上传文件信息")
-    upload_url: str = Field(..., description="上传URL")
-
-
-class ShareLinkRequest(BaseModel):
-    """生成分享链接请求"""
-    file_id: str = Field(..., description="文件ID")
-    is_public: bool = Field(True, description="是否公开")
-    access_scope: str = Field("view", description="访问范围")
-    expire_seconds: int = Field(86400, description="过期时间（秒）")
-    max_access: Optional[int] = Field(None, description="最大访问次数")
-    password: Optional[str] = Field(None, description="访问密码")
-
-
-class FileVisitRequest(BaseModel):
-    """文件访问请求"""
-    file_share_id: str = Field(..., description="分享ID")
-    access_type: str = Field(..., description="访问类型")
-    access_duration: int = Field(..., description="访问时长")
-    metadata: Dict[str, Any] = Field(default_factory=dict, description="元数据")
-
-
-class FileListRequest(BaseModel):
-    """文件列表请求"""
-    folder_id: str = Field(..., description="文件夹ID")
-    file_name: Optional[str] = Field(None, description="文件名")
-    file_type: Optional[List[str]] = Field(None, description="文件类型")
-    created_by_role: Optional[str] = Field(None, description="创建者角色")
-    created_by: Optional[str] = Field(None, description="创建者")
-    page_size: int = Field(20, description="每页大小")
-    page: int = Field(1, description="页码")
-
-
-class FileListResponse(BaseModel):
-    """文件列表响应"""
-    files: List[File] = Field(default_factory=list, description="文件列表")
-
-
-class GetFileResponse(BaseModel):
-    """获取文件响应"""
-    file: File = Field(..., description="文件信息")
-    upload_file: Optional[UploadFile] = Field(None, description="上传文件信息")
-
-
-class DownloadUrlInfo(BaseModel):
-    """下载URL信息"""
-    file_id: str = Field(..., description="文件ID")
-    url: str = Field(..., description="下载URL")
-    error: Optional[str] = Field(None, description="错误信息")
-
-
-class BatchDownloadUrlResponse(BaseModel):
-    """批量下载URL响应"""
-    download_urls: List[DownloadUrlInfo] = Field(default_factory=list, description="下载URL列表")
-
-
-class GcsUrlInfo(BaseModel):
-    """GCS URL信息"""
-    file_id: str = Field(..., description="文件ID")
-    gcs_url: str = Field(..., description="GCS URL")
-    mime_type: str = Field(..., description="MIME类型")
-    error: Optional[str] = Field(None, description="错误信息")
-
-
-class GetGcsUrlResponse(BaseModel):
-    """获取GCS URL响应"""
-    gcs_url: str = Field(..., description="GCS URL")
-    mime_type: str = Field(..., description="MIME类型")
-
-
-class BatchGcsUrlResponse(BaseModel):
-    """批量GCS URL响应"""
-    gcs_urls: List[GcsUrlInfo] = Field(default_factory=list, description="GCS URL列表")
-
-
-# ========= 压缩服务相关模型 =========
-
-class CompressedVariant(BaseModel):
-    """压缩变体信息"""
-    variant_name: str = Field(..., description="变体名称")
-    variant_type: str = Field(..., description="变体类型")
-    media_type: str = Field(..., description="媒体类型")
-    width: int = Field(..., description="宽度")
-    height: int = Field(..., description="高度")
-    file_size: int = Field(..., description="文件大小")
-    format: str = Field(..., description="格式")
-    quality: Optional[int] = Field(None, description="质量")
-    duration: Optional[float] = Field(None, description="时长")
-    bitrate: Optional[int] = Field(None, description="比特率")
-    fps: Optional[int] = Field(None, description="帧率")
-    compression_ratio: float = Field(..., description="压缩比")
-    stored_path: str = Field(..., description="存储路径")
-
-
-class CompressionStatusResponse(BaseModel):
-    """压缩状态响应"""
-    status: str = Field(..., description="状态: pending, processing, completed, failed")
-    error_message: Optional[str] = Field(None, description="错误信息")
-    variants: List[CompressedVariant] = Field(default_factory=list, description="压缩变体列表")
-
-
-class GetVariantsResponse(BaseModel):
-    """获取变体响应"""
-    variants: List[CompressedVariant] = Field(default_factory=list, description="压缩变体列表")
-
-
-class RecompressionResponse(BaseModel):
-    """重新压缩响应"""
-    task_id: str = Field(..., description="任务ID")
-    status: str = Field(..., description="状态")
-
-
-class VariantDownloadUrlResponse(BaseModel):
-    """变体下载URL响应"""
-    url: str = Field(..., description="下载URL")
-    error: Optional[str] = Field(None, description="错误信息")
-    variant_info: Optional[CompressedVariant] = Field(None, description="变体详细信息")
+"""
+文件相关数据模型
+"""
+from datetime import datetime
+from typing import Optional, Dict, List, Any
+from pydantic import BaseModel, Field
+
+
+class File(BaseModel):
+    """文件信息模型"""
+    id: str = Field(..., description="文件ID")
+    folder_id: str = Field(..., description="所属文件夹ID")
+    file_name: str = Field(..., description="原始文件名")
+    file_type: str = Field(..., description="文件类型")
+    created_at: datetime = Field(..., description="创建时间")
+    updated_at: datetime = Field(..., description="更新时间")
+
+    class Config:
+        json_encoders = {
+            datetime: lambda v: v.isoformat()
+        }
+
+
+class UploadFile(BaseModel):
+    """上传文件信息模型"""
+    id: str = Field(..., description="上传文件ID")
+    folder_id: str = Field(..., description="所属文件夹ID")
+    storage_type: str = Field(..., description="存储类型")
+    stored_name: str = Field(..., description="存储文件名")
+    stored_path: str = Field(..., description="存储路径")
+    file_id: str = Field(..., description="所属文件ID")
+    file_name: str = Field(..., description="原始文件名")
+    file_size: int = Field(0, description="文件大小（字节）")
+    file_ext: str = Field(..., description="文件后缀")
+    mime_type: str = Field(..., description="MIME类型")
+    created_at: datetime = Field(..., description="创建时间")
+    updated_at: datetime = Field(..., description="更新时间")
+
+    class Config:
+        json_encoders = {
+            datetime: lambda v: v.isoformat()
+        }
+
+
+class FileUploadResponse(BaseModel):
+    """文件上传返回"""
+    file: File = Field(..., description="文件信息")
+    upload_file: UploadFile = Field(..., description="上传文件信息")
+
+
+class UploadUrlResponse(BaseModel):
+    """上传URL响应"""
+    file: File = Field(..., description="文件信息")
+    upload_file: UploadFile = Field(..., description="上传文件信息")
+    upload_url: str = Field(..., description="上传URL")
+
+
+class ShareLinkRequest(BaseModel):
+    """生成分享链接请求"""
+    file_id: str = Field(..., description="文件ID")
+    is_public: bool = Field(True, description="是否公开")
+    access_scope: str = Field("view", description="访问范围")
+    expire_seconds: int = Field(86400, description="过期时间（秒）")
+    max_access: Optional[int] = Field(None, description="最大访问次数")
+    password: Optional[str] = Field(None, description="访问密码")
+
+
+class FileVisitRequest(BaseModel):
+    """文件访问请求"""
+    file_share_id: str = Field(..., description="分享ID")
+    access_type: str = Field(..., description="访问类型")
+    access_duration: int = Field(..., description="访问时长")
+    metadata: Dict[str, Any] = Field(default_factory=dict, description="元数据")
+
+
+class FileListRequest(BaseModel):
+    """文件列表请求"""
+    folder_id: str = Field(..., description="文件夹ID")
+    file_name: Optional[str] = Field(None, description="文件名")
+    file_type: Optional[List[str]] = Field(None, description="文件类型")
+    created_by_role: Optional[str] = Field(None, description="创建者角色")
+    created_by: Optional[str] = Field(None, description="创建者")
+    page_size: int = Field(20, description="每页大小")
+    page: int = Field(1, description="页码")
+
+
+class FileListResponse(BaseModel):
+    """文件列表响应"""
+    files: List[File] = Field(default_factory=list, description="文件列表")
+
+
+class GetFileResponse(BaseModel):
+    """获取文件响应"""
+    file: File = Field(..., description="文件信息")
+    upload_file: Optional[UploadFile] = Field(None, description="上传文件信息")
+
+
+class DownloadUrlInfo(BaseModel):
+    """下载URL信息"""
+    file_id: str = Field(..., description="文件ID")
+    url: str = Field(..., description="下载URL")
+    mime_type: str = Field(..., description="MIME类型")
+    error: Optional[str] = Field(None, description="错误信息")
+
+
+class BatchDownloadUrlResponse(BaseModel):
+    """批量下载URL响应"""
+    download_urls: List[DownloadUrlInfo] = Field(default_factory=list, description="下载URL列表")
+
+
+class GcsUrlInfo(BaseModel):
+    """GCS URL信息"""
+    file_id: str = Field(..., description="文件ID")
+    gcs_url: str = Field(..., description="GCS URL")
+    mime_type: str = Field(..., description="MIME类型")
+    error: Optional[str] = Field(None, description="错误信息")
+
+
+class GetGcsUrlResponse(BaseModel):
+    """获取GCS URL响应"""
+    gcs_url: str = Field(..., description="GCS URL")
+    mime_type: str = Field(..., description="MIME类型")
+
+
+class BatchGcsUrlResponse(BaseModel):
+    """批量GCS URL响应"""
+    gcs_urls: List[GcsUrlInfo] = Field(default_factory=list, description="GCS URL列表")
+
+
+# ========= 压缩服务相关模型 =========
+
+class CompressedVariant(BaseModel):
+    """压缩变体信息"""
+    variant_name: str = Field(..., description="变体名称")
+    variant_type: str = Field(..., description="变体类型")
+    media_type: str = Field(..., description="媒体类型")
+    width: int = Field(..., description="宽度")
+    height: int = Field(..., description="高度")
+    file_size: int = Field(..., description="文件大小")
+    format: str = Field(..., description="格式")
+    quality: Optional[int] = Field(None, description="质量")
+    duration: Optional[float] = Field(None, description="时长")
+    bitrate: Optional[int] = Field(None, description="比特率")
+    fps: Optional[int] = Field(None, description="帧率")
+    compression_ratio: float = Field(..., description="压缩比")
+    stored_path: str = Field(..., description="存储路径")
+
+
+class CompressionStatusResponse(BaseModel):
+    """压缩状态响应"""
+    status: str = Field(..., description="状态: pending, processing, completed, failed")
+    error_message: Optional[str] = Field(None, description="错误信息")
+    variants: List[CompressedVariant] = Field(default_factory=list, description="压缩变体列表")
+
+
+class GetVariantsResponse(BaseModel):
+    """获取变体响应"""
+    variants: List[CompressedVariant] = Field(default_factory=list, description="压缩变体列表")
+
+
+class RecompressionResponse(BaseModel):
+    """重新压缩响应"""
+    task_id: str = Field(..., description="任务ID")
+    status: str = Field(..., description="状态")
+
+
+class VariantDownloadUrlResponse(BaseModel):
+    """变体下载URL响应"""
+    url: str = Field(..., description="下载URL")
+    error: Optional[str] = Field(None, description="错误信息")
+    variant_info: Optional[CompressedVariant] = Field(None, description="变体详细信息")

file_hub_client/services/file/async_blob_service.py

@@ -221,7 +221,7 @@ class AsyncBlobService(BaseFileService):
         upload_url = await self.http_uploader.start_resumable_session(
             url=upload_url_resp.upload_url,
             total_file_size=file_size,
-            mine_type=mime_type,
+            mime_type=mime_type,
         )
 
         # 上传文件到对象存储
@@ -403,6 +403,7 @@ class AsyncBlobService(BaseFileService):
             keep_original_filename: Optional[bool] = False,
             url: Optional[str] = None,
             file_name: Optional[str] = None,
+            mime_type: Optional[str] = None,
             request_id: Optional[str] = None,
             **metadata
     ) -> FileUploadResponse:
@@ -418,6 +419,7 @@ class AsyncBlobService(BaseFileService):
             keep_original_filename: 是否保留原始文件名（默认False）
             url: 要下载并上传的URL（可选）
             file_name: 当使用url参数时指定的文件名（可选）
+            mime_type: MIME类型（可选，用于推断文件扩展名，特别适用于AI生成的字节数据）
             request_id: 请求ID（可选，如果不提供则自动生成）
             **metadata: 额外的元数据
 
@@ -427,6 +429,8 @@ class AsyncBlobService(BaseFileService):
         Note:
             必须提供 file 或 url 参数之一
             
+            当传入bytes或BinaryIO且未提供file_name时，建议提供mime_type以确保正确的文件扩展名推断
+            
             Cache-Control 头在 GCS 直传模式（STREAM/RESUMABLE）下自动设置为 "public, max-age=86400"
         """
         # 参数验证：必须提供 file 或 url 之一
@@ -449,26 +453,38 @@ class AsyncBlobService(BaseFileService):
             # 使用下载的内容作为file参数
             file = downloaded_content
 
-            # 提取文件信息（bytes会返回默认的MIME类型，我们稍后会基于文件名重新计算）
-            _, content, file_size, _, _, file_hash = self._extract_file_info(file)
+            # 基于文件名计算MIME类型
+            mime_type = get_file_mime_type(Path(file_name))
+            
+            # 提取文件信息，传入MIME类型用于推断扩展名
+            _, content, file_size, _, _, file_hash = self._extract_file_info(file, mime_type)
 
             # file_name已经在上面设置了（要么是用户指定的，要么是从URL提取的）
             extracted_file_name = file_name
 
-            # 基于文件名计算文件类型和MIME类型
+            # 基于文件名计算文件类型
             file_type = Path(extracted_file_name).suffix.lstrip('.').lower() if Path(
                 extracted_file_name).suffix else 'dat'
-            mime_type = get_file_mime_type(Path(extracted_file_name))
         else:
             # 解析文件参数，提取文件信息
-            extracted_file_name, content, file_size, extract_mime_type, extract_file_type, file_hash = self._extract_file_info(
-                file)
+            # 如果用户指定了文件名，先从文件名推断MIME类型，然后传给_extract_file_info
             if file_name:
+                # 用户指定了文件名，优先使用用户提供的MIME类型，否则从文件名推断
+                if mime_type:
+                    file_name_mime_type = mime_type
+                else:
+                    file_name_mime_type = get_file_mime_type(Path(file_name))
+                extracted_file_name, content, file_size, extract_mime_type, extract_file_type, file_hash = self._extract_file_info(
+                    file, file_name_mime_type)
+                # 使用用户指定的文件名
                 extracted_file_name = file_name
-                mime_type = get_file_mime_type(file_name)
+                mime_type = file_name_mime_type
                 file_type = Path(extracted_file_name).suffix.lstrip('.').lower() if Path(
                     extracted_file_name).suffix else 'dat'
             else:
+                # 没有指定文件名，传入用户提供的MIME类型（如果有）
+                extracted_file_name, content, file_size, extract_mime_type, extract_file_type, file_hash = self._extract_file_info(
+                    file, mime_type)
                 mime_type = extract_mime_type
                 file_type = extract_file_type
 
@@ -670,6 +686,7 @@ class AsyncBlobService(BaseFileService):
             download_urls.append(DownloadUrlInfo(
                 file_id=url_info.file_id,
                 url=url_info.url,
+                mime_type=url_info.mime_type,
                 error=url_info.error if url_info.HasField('error') else None
             ))