From d0031fabc8fb2acfa955b6ec161fa4453852c6de Mon Sep 17 00:00:00 2001 From: jason Date: Wed, 6 May 2026 10:42:26 +0800 Subject: [PATCH] feat: add PUT /{file_id}/replace endpoint for clearing file tags Add new API endpoint to fully replace (not merge) file annotation tags. - Empty tag list now clears all tags (partial update kept them) - Supports both simplified and full tag formats with auto-conversion - Addresses bug where tags couldn't be deleted via empty list --- .../app/module/annotation/interface/task.py | 78 +++++++++++++++++++ .../app/module/dataset/service/service.py | 78 +++++++++++++++++++ 2 files changed, 156 insertions(+) diff --git a/runtime/datamate-python/app/module/annotation/interface/task.py b/runtime/datamate-python/app/module/annotation/interface/task.py index 57aeb506..ca2c7a70 100644 --- a/runtime/datamate-python/app/module/annotation/interface/task.py +++ b/runtime/datamate-python/app/module/annotation/interface/task.py @@ -321,3 +321,81 @@ async def update_file_tags( message="标签更新成功", data=response_data ) + + +@router.put( + "/{file_id}/replace", + response_model=StandardResponse[UpdateFileTagsResponse], +) +async def replace_file_tags( + request: UpdateFileTagsRequest, + file_id: str = Path(..., description="文件ID"), + db: AsyncSession = Depends(get_db) +): + """ + Replace File Tags (Full Replacement with Auto Format Conversion) + + 完全替换文件的标签列表。与部分更新不同，此方法会： + - 空列表会清空所有标签 + - 新标签列表完全替换原有标签（不合并） + + 支持两种标签格式： + 1. 简化格式: + [{"from_name": "label", "to_name": "image", "values": ["cat", "dog"]}] + + 2. 完整格式: + [{"id": "...", "from_name": "label", "to_name": "image", "type": "choices", + "value": {"choices": ["cat", "dog"]}}] + + 系统会自动根据数据集关联的模板将简化格式转换为完整格式。 + """ + service = DatasetManagementService(db) + + result = await db.execute( + select(DatasetFiles).where(DatasetFiles.id == file_id) + ) + file_record = result.scalar_one_or_none() + + if not file_record: + raise HTTPException(status_code=404, detail=f"File not found: {file_id}") + + dataset_id = str(file_record.dataset_id) + + mapping_service = DatasetMappingService(db) + template_id = await mapping_service.get_template_id_by_dataset_id(dataset_id) + + if template_id: + logger.info(f"Found template {template_id} for dataset {dataset_id}, will auto-convert tag format") + else: + logger.warning(f"No template found for dataset {dataset_id}, tags must be in full format") + + success, error_msg, updated_at = await service.replace_file_tags( + file_id=file_id, + new_tags=request.tags, + template_id=template_id + ) + + if not success: + if "not found" in (error_msg or "").lower(): + raise HTTPException(status_code=404, detail=error_msg) + raise HTTPException(status_code=500, detail=error_msg or "替换标签失败") + + result = await db.execute( + select(DatasetFiles).where(DatasetFiles.id == file_id) + ) + file_record = result.scalar_one_or_none() + + if not file_record: + raise HTTPException(status_code=404, detail=f"File not found: {file_id}") + + response_data = UpdateFileTagsResponse( + fileId=file_id, + tags=file_record.tags or [], + tagsUpdatedAt=updated_at or datetime.now() + ) + + return StandardResponse( + code="0", + message="标签替换成功", + data=response_data + ) diff --git a/runtime/datamate-python/app/module/dataset/service/service.py b/runtime/datamate-python/app/module/dataset/service/service.py index 04dbbbdb..d120bc62 100644 --- a/runtime/datamate-python/app/module/dataset/service/service.py +++ b/runtime/datamate-python/app/module/dataset/service/service.py @@ -470,6 +470,84 @@ def _index_tag(idx: int, tag: Dict[str, Any]) -> None: await self.db.rollback() return False, str(e), None + async def replace_file_tags( + self, + file_id: str, + new_tags: List[Dict[str, Any]], + template_id: Optional[str] = None + ) -> tuple[bool, Optional[str], Optional[datetime]]: + """ + 完全替换文件标签（非部分更新） + + 与部分更新不同，此方法会完全替换标签列表： + - 空列表会清空所有标签 + - 新标签列表会完全替换原有标签（不合并） + + 如果提供了 template_id，会自动将简化格式的标签转换为完整格式。 + + Args: + file_id: 文件ID + new_tags: 新的标签列表（完全替换），可以是简化格式或完整格式，空列表会清空所有标签 + template_id: 可选的模板ID，用于格式转换 + + Returns: + (成功标志, 错误信息, 更新时间) + """ + try: + logger.info(f"Replacing tags for file: {file_id}, new_tags count: {len(new_tags)}") + + result = await self.db.execute( + select(DatasetFiles).where(DatasetFiles.id == file_id) + ) + file_record = result.scalar_one_or_none() + + if not file_record: + logger.error(f"File not found: {file_id}") + return False, f"File not found: {file_id}", None + + processed_tags = new_tags + if template_id and new_tags: + logger.debug(f"Converting tags using template: {template_id}") + + try: + from app.db.models import AnnotationTemplate + template_result = await self.db.execute( + select(AnnotationTemplate).where( + AnnotationTemplate.id == template_id, + AnnotationTemplate.deleted_at.is_(None) + ) + ) + template = template_result.scalar_one_or_none() + + if not template: + logger.warning(f"Template {template_id} not found, skipping conversion") + else: + from app.module.annotation.utils import create_converter_from_template_config + + converter = create_converter_from_template_config(template.configuration) # type: ignore + processed_tags = converter.convert_if_needed(new_tags) + + logger.info(f"Converted {len(new_tags)} tags to full format") + + except Exception as e: + logger.error(f"Failed to convert tags using template: {e}") + logger.warning("Continuing with original tag format") + + update_time = datetime.utcnow() + file_record.tags = processed_tags # type: ignore + file_record.tags_updated_at = update_time # type: ignore + + await self.db.commit() + await self.db.refresh(file_record) + + logger.info(f"Successfully replaced tags for file: {file_id}, new tags count: {len(processed_tags)}") + return True, None, update_time + + except Exception as e: + logger.error(f"Failed to replace tags for file {file_id}: {e}") + await self.db.rollback() + return False, str(e), None + @staticmethod async def _get_or_create_dataset_directory(dataset: Dataset) -> str: """Get or create dataset directory"""