Return 4xx instead of 500 for malformed API requests

未关闭
#6,387 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
5/5
预计耗时
一周以上
新手友好度
20/100
Issue 类型
缺陷
描述清晰度
基本清楚
活跃度
活跃
技术栈
django, python
领域
api, backend

调研方向

This umbrella issue covers malformed input across DRF viewsets, /api/sync/, plain Django views, kolibri_public, and search; it names shared patterns and regression tests but leaves the individual root causes to sub-issues. Start by checking whether a scoped sub-issue exists, then read its cited view or serializer and reproduce the malformed request in a regression test. Done means that input returns an appropriate 400 or 404, with a test covering it; the issue says contributions are not open here.

由索引模型根据 Issue 内容生成。

描述

DEV: backend

❌ This issue is not open for contribution. Visit Contributing guidelines to learn about the contributing process and how to find suitable issues.

Overview

Studio's API returns HTTP 500 for many malformed requests: a body with the wrong JSON shape, a query parameter of the wrong type, a value out of range, or a malformed id. In every case the request is the client's error and should get a 400 or 404 that says what is wrong. An audit of every endpoint found the remaining cases and grouped them by root cause. Each sub-issue fixes one root cause wherever it occurs, using the validation pattern set out below.

Scope
  • Included: validation of request input (body, query params, URL kwargs) across DRF viewsets, /api/sync/, plain Django views, and the public kolibri_public and search APIs.
  • Included: sync change handlers that error on malformed changes instead of returning a validation error for that change.
  • Excluded: the /api/internal/* endpoints used by ricecooker, which are to be deprecated and removed in favour of the sync endpoint.
  • Excluded: permission checks and data-integrity bugs, which have their own issues.
Guidance
  • Target hotfixes.
  • Each fix needs a regression test that sends the malformed input and asserts the 4xx. In tests, the DRF test client re-raises view exceptions, so the current behaviour shows up as an exception rather than a 500 status.
Strategy
  • Every request body is validated by a DRF serializer with is_valid(raise_exception=True). Plain Django views that parse request.body become DRF @api_view views.
  • Every query parameter is a typed django-filter filter on the viewset's FilterSet, such as UUIDFilter/UUIDInFilter from contentcuration/contentcuration/viewsets/common.py, NumberFilter, BooleanFilter or ChoiceFilter. A CharFilter whose value is parsed by hand doesn't count.
  • Serializer fields carry the model's limits (max_length, min/max value, choices), so a value that passes validation also saves to the database.
  • One custom DRF EXCEPTION_HANDLER maps Django ValidationError to 400 and ObjectDoesNotExist to 404, as a backstop.
  • Studio's ValuesViewset (contentcuration/contentcuration/viewsets/base.py) is an older copy of Kolibri's kolibri/core/api.py. Fixes there follow Kolibri's shape.

AI usage

I scoped the audit, excluded the internal endpoints, and chose the validation strategy. Claude Code audited every endpoint, reproduced each finding with a temporary test, grouped the findings by root cause, and drafted this issue.

主要语言
Python
星标
212
派生
321
平均合并
1 天 7 小时
30 天内合并 PR
14

环境准备

  • 提供 Dockerfile 或 Docker Compose 文件
  • 没有 Pull Request 模板
  • 阅读贡献指南

从这里开始

  1. 先读完整个 Issue,再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库,在一个分支上完成修改。
  4. 提交 Pull Request,并在描述里引用这个 Issue 编号。

learningequality/studio 的其他 Issue

查看 learningequality/studio 的全部 Issue

相似的 Issue

更多 Python Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。