如何为媒体分析引擎设计 RESTful API [英] How to design a REStful API for a media analysis engine

问题描述

我是 Restful 概念的新手，必须为我需要设置的媒体分析服务设计一个简单的 API，以执行各种任务，例如对上传的图片和视频进行人脸分析、区域检测等.

I am new to Restful concept and have to design a simple API for a media analysis service I need to set up, to perform various tasks, e.g. face analysis, region detection, etc. on uploaded images and video.

我最初的设计大纲如下:

Outline of my initial design is as follows:

• 客户端将配置 XML 文件 POST 到 http://manalysis.com/facerecognition.这将创建一个可用于多个分析会话的配置文件.响应 XML 包含一个 ProfileID 以引用此配置文件.客户端可以跳过此步骤使用默认配置参数
• 客户端将要分析的视频数据发布到 http://manalysis.com/facerecognition(以 ProfileID 作为参数，如果已设置).这将创建一个分析会话.返回 XML 具有 SessionID.
• 客户端可以向 http://manalysis.com/facerecognition/SessionID 发送 GET 请求以接收会话状态.

• Client POSTs a configuration XML file to http://manalysis.com/facerecognition. This creates a profile that can be used for multiple analysis sessions. Response XML includes a ProfileID to refer to this profile. Clients can skip this step to use the default config parameters
• Client POSTs video data to be analyzed to http://manalysis.com/facerecognition (with ProfileID as a parameter, if it's set up). This creates an analysis session. Return XML has the SessionID.
• Client can send a GET to http://manalysis.com/facerecognition/SessionID to receive the status of the session.

我在正确的轨道上吗?具体来说，我有以下问题:

Am I on the right track? Specifically, I have the following questions:

• 我应该在 URL 中包含 facerecognition 吗?Roy Fielding 说REST API 不能定义固定的资源名称或层次结构"这是该错误的一个实例吗?
• 分析结果可以在一个大型 XML 文件中返回给客户端，也可以在检测到每个事件时返回给客户端.我应该如何告诉分析引擎返回结果的位置?
• 我是否应该在分析完成后通过 DELETE 调用明确删除配置文件?

• Should I include facerecognition in the URL? Roy Fielding says that "a REST API must not define fixed resource names or hierarchies" Is this an instance of that mistake?
• The analysis results can either be returned to the client in one large XML file or when each event is detected. How should I tell the analysis engine where to return the results?
• Should I explicitly delete a profile when analysis is done, through a DELETE call?

谢谢，

C

推荐答案

可以修复入口点url，

You can fix the entry point url,

GET /facerecognition

<FaceRecognitionService>
  <Profiles href="/facerecognition/profiles"/>
  <AnalysisRequests href="/facerecognition/analysisrequests"/>
</FaceRecognitionService>

通过将 XML 配置文件发布到 Profiles 元素的 href 属性中的 URL 来创建新配置文件

Create a new profile by posting the XML profile to the URL in the href attribute of the Profiles element

POST /facerecognition/profiles
201 - Created
Location: /facerecognition/profile/33

通过创建新的分析请求来启动分析.我会避免使用术语会话，因为它太笼统，并且在 REST 世界中有很多负面关联.

Initiate the analysis by creating a new Analysis Request. I would avoid using the term session as it is too generic and has lots of negative associations in the REST world.

POST /facerecognition/analysisrequests?profileId=33
201 - Created
Location: /facerecognition/analysisrequest/2103

查看进程状态

GET /facerecognition/analysisrequest/2103

<AnalysisRequest>
   <Status>Processing</Status>
   <Cancel Method="DELETE" href="/facerecognition/analysisrequest/2103" />
</AnalysisRequest>

当处理完成时，同样的GET可以返回

when the processing has finished, the same GET could return

<AnalysisRequest>
   <Status>Completed</Status>
   <Results href="/facerecognition/analysisrequest/2103/results" />
</AnalysisRequest>

我选择的具体网址比较随意，你可以用你最清楚的.

The specific URLs that I have chosen are relatively arbitrary, you can use whatever is the clearest to you.

这篇关于如何为媒体分析引擎设计 RESTful API的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！