feat: AI Code Review Panel for DSA Submissions (#857)#1418
feat: AI Code Review Panel for DSA Submissions (#857)#1418ChandanMeher4 wants to merge 7 commits into
Conversation
github-actions
Bot
added
quality:clean
|
Hi @ChandanMeher4, thanks for contributing to InternHack! 🎉 I have automatically:
Our workflows will now analyze your changes to classify:
Tip Ensure your PR description references the issue it resolves (e.g. Happy coding! 🚀 |
github-actions
Bot
added
type:feature
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (2)
🚧 Files skipped from review as they are similar to previous changes (2)
📝 WalkthroughWalkthroughAdds end-to-end AI code review: new type and Zod schema, Prisma enum and admin schema update, DsaService prompt/parse/logging, controller+route for POST review, and client UI (ai-review tab, mutation, DsaAiReviewPanel). ChangesDSA AI Code Review
Sequence DiagramsequenceDiagram
participant Student as Student (Browser)
participant DetailPage as DsaProblemDetailPage
participant API as DsaController / Route
participant Service as DsaService
participant AIProvider as AI Provider (DSA_CODE_REVIEW)
participant ReviewPanel as DsaAiReviewPanel
Student->>DetailPage: Click "Run Code"
DetailPage->>DetailPage: executeMutation runs (Judge0)
DetailPage-->>Student: Test results (instant)
DetailPage->>API: POST /dsa/submissions/{id}/review (aiReviewMutation)
API->>Service: generateCodeReview(submissionId, userId)
Service->>Service: Load submission, verify ownership, load problem context
Service->>AIProvider: Call with buildCodeReviewPrompt(payload)
AIProvider-->>Service: Raw AI response (text)
Service->>Service: parseCodeReviewResponse(text)
Service-->>API: CodeReviewResponse (validated/fallback)
API-->>DetailPage: Review data
DetailPage->>ReviewPanel: Provide review / isLoading / error
ReviewPanel-->>Student: Render complexity, readability, edge cases, suggestions
Student->>ReviewPanel: Click "Re-analyze"
ReviewPanel->>API: POST /dsa/submissions/{id}/review (retry)
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes Suggested labels
Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@server/src/module/dsa/dsa.service.ts`:
- Line 962: The logAIRequest call records success before parseCodeReviewResponse
validates the output; move or update logging so success is only set after
parseCodeReviewResponse completes without throwing. Wrap
parseCodeReviewResponse(response) in a try/catch (or await and check result) and
call logAIRequest("DSA_CODE_REVIEW", response, true, undefined, studentId) only
on successful validation, and call logAIRequest(..., false, error, studentId)
inside the catch to record failures; reference the existing logAIRequest and
parseCodeReviewResponse calls to locate where to adjust logging.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 1c0586e6-3b7a-472e-84e5-7bfba1402c74
📒 Files selected for processing (8)
client/src/lib/types/learning.types.tsclient/src/module/student/dsa/DsaProblemDetailPage.tsxclient/src/module/student/dsa/components/DsaAiReviewPanel.tsxserver/src/database/prisma/schema/base.prismaserver/src/module/dsa/dsa.controller.tsserver/src/module/dsa/dsa.routes.tsserver/src/module/dsa/dsa.service.tsserver/src/module/dsa/dsa.validation.ts
Code Review — PR #1418: AI Code Review Panel for DSA SubmissionsHi @ChandanMeher4, the panel UI is clean and the mutation error handling is solid. Found two issues to address. 🔴 Backend endpoint
|
Sachinchaurasiya360
left a comment
Sachinchaurasiya360
left a comment
There was a problem hiding this comment.
The module pattern, prompt engineering, client component structure, and error handling are all solid. However there are five issues that must be fixed before this can merge.
Issue 1 (Critical) - Missing Prisma migration for DSA_CODE_REVIEW enum value:
The PR adds DSA_CODE_REVIEW to the AIServiceType enum in base.prisma but ships no migration file. Every prior enum addition has a corresponding migration with ALTER TYPE "AIServiceType" ADD VALUE. Without it the production PostgreSQL type is never updated and the getProviderForService("DSA_CODE_REVIEW") call will fail at runtime. Add a migration file with:
ALTER TYPE "AIServiceType" ADD VALUE 'DSA_CODE_REVIEW';
Issue 2 (Critical) - No server-side premium gating:
The /submissions/:submissionId/review endpoint does not check premium status. A non-premium user with a submissionId can call it freely. The claim that "only premium users can run code so it is implicitly gated" is incorrect - the submissionId is not a secret and the endpoint has no isPremiumUser check. Add the same pattern used in latex-chat.controller.ts and ats.controller.ts: check subscriptionPlan (MONTHLY or YEARLY) and subscriptionStatus (ACTIVE), return 403 if not premium.
Issue 3 (Critical) - Usage not logged so the rate limit counter never increments:
The route applies usageLimit("CODE_RUN") as a guard but generateCodeReview in dsa.service.ts never writes a UsageLog entry after a successful review. Compare with executeCode which explicitly calls prisma.usageLog.create after execution. Without the write, the daily cap checks but never charges, allowing unlimited calls after the first one passes.
Issue 4 (Critical) - DSA_CODE_REVIEW missing from admin validation schema:
admin.validation.ts has a hardcoded z.enum([...]) of AIServiceType values for the admin AI config endpoint. DSA_CODE_REVIEW is not included. Admins cannot configure the model or provider for this service through the admin panel and any attempt returns a Zod 400.
Issue 5 (High) - No aiServiceConfig seed row documented or provided:
getProviderForService falls back silently to the default Gemini model if no row exists for DSA_CODE_REVIEW in aiServiceConfig. The PR needs to either add a seed entry or document the manual admin-panel step needed after deploy, so the service does not run silently on an unconfigured fallback.
Minor:
- rounded-full on the loading spinner in DsaAiReviewPanel - confirm with the maintainer whether spinners are exempt from the no-rounded-full rule.
- Raw Error throws in dsa.service.ts for "Submission not found" and "Not authorized" will surface as 500s. Handle them explicitly in the controller for proper 404/403 responses.
Fix the five critical issues and this is ready for another review pass. The AI prompt, Zod schema, client mutation handling, and ownership check are all well done.
|
Update on Issue 1: the project uses Issues 2-5 still need to be addressed before merge:
|
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
server/src/module/dsa/dsa.service.ts (1)
1049-1059:⚠️ Potential issue | 🟠 Major | ⚡ Quick winReturn a fallback response when parsing fails completely.
Lines 1049-1059 still throw on unrecoverable parse failure, which can fail the endpoint instead of degrading safely with a default
CodeReviewResponse.💡 Minimal fallback patch
let parsed: unknown; try { parsed = JSON.parse(jsonStr); } catch { // Try to extract just the JSON object if there's surrounding text const objMatch = jsonStr.match(/\{[\s\S]*\}/); if (objMatch) { const cleaned = objMatch[0].replace(/,\s*([\]}])/g, "$1"); parsed = JSON.parse(cleaned); } else { - throw new Error("Failed to parse AI code review response"); + return { + timeComplexity: "Unable to determine", + spaceComplexity: "Unable to determine", + readability: { score: 5, feedback: "No feedback available" }, + edgeCases: [], + suggestions: [], + }; } }As per coding guidelines, "Always handle AI parse failures with fallbacks in Gemini responses".
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@server/src/module/dsa/dsa.service.ts` around lines 1049 - 1059, The JSON parse catch currently throws on unrecoverable failures; instead, in the catch branch where neither direct JSON.parse nor the object-extraction path succeeds, construct and return a safe default CodeReviewResponse (e.g., an object matching the CodeReviewResponse shape with empty/neutral fields like success: false, summary: "Unable to parse AI response", comments: []) rather than throwing; update the function around variables parsed and jsonStr (in dsa.service.ts) so that when parsed remains undefined you return this fallback CodeReviewResponse to degrade gracefully.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@server/src/module/admin/admin.validation.ts`:
- Line 350: The validation enum for switchAIProviderSchema.service is missing
the "EMAIL_CHAT" value and thus is out of sync with the Prisma AIServiceType;
update the enum in admin.validation.ts (the switchAIProviderSchema.service
definition) to include "EMAIL_CHAT" so it matches AIServiceType and accepts
admin requests for that service.
---
Outside diff comments:
In `@server/src/module/dsa/dsa.service.ts`:
- Around line 1049-1059: The JSON parse catch currently throws on unrecoverable
failures; instead, in the catch branch where neither direct JSON.parse nor the
object-extraction path succeeds, construct and return a safe default
CodeReviewResponse (e.g., an object matching the CodeReviewResponse shape with
empty/neutral fields like success: false, summary: "Unable to parse AI
response", comments: []) rather than throwing; update the function around
variables parsed and jsonStr (in dsa.service.ts) so that when parsed remains
undefined you return this fallback CodeReviewResponse to degrade gracefully.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 64091797-8512-494e-a623-8e6dd27abfa2
📒 Files selected for processing (6)
client/src/module/student/dsa/DsaProblemDetailPage.tsxclient/src/module/student/dsa/components/DsaAiReviewPanel.tsxserver/src/database/prisma/schema/base.prismaserver/src/module/admin/admin.validation.tsserver/src/module/dsa/dsa.controller.tsserver/src/module/dsa/dsa.service.ts
🚧 Files skipped from review as they are similar to previous changes (2)
- client/src/module/student/dsa/DsaProblemDetailPage.tsx
- client/src/module/student/dsa/components/DsaAiReviewPanel.tsx
2 participants
Description
Adds a manual AI code review feature to the DSA problem-solving page. After running the code and seeing the test results, a new "AI Review" tab appears in the right panel. Clicking "Get AI Review" sends the submissionId to the backend, which fetches the submission and problem context from the DB and calls Gemini. The review renders time/space complexity, readability score, edge cases, and improvement suggestions.
The review is manually triggered rather than auto-firing after every run - this keeps API costs minimal. Only Premium users can execute code at all, so the endpoint is already naturally gated behind the existing Premium wall with zero extra logic needed.
Related Issue
Fixes #857
Type of Change
Testing
Screenshots
Checklist
.env, credentials, ornode_modulescommittedSummary by CodeRabbit