fix(orm): preserve array mutation shape for optional typed-JSON fields

[ymc9]

Summary

• makeNullableTypedJsonMutationSchema was calling makeScalarSchema(type, attributes) internally, which discarded the already-built fieldSchema passed by callers — including any array/list mutation shape ({ set } on create, { set, push } on update)
• Changed the function to accept a fieldSchema: z.ZodTypeAny parameter and use it directly in safeParse instead of reconstructing from scratch
• Updated both callers (create schema and update schema) to pass the pre-built fieldSchema

Test plan

• Verify optional typed-JSON array fields accept { set: [...] } on create
• Verify optional typed-JSON array fields accept { set, push } operators on update
• Verify null / DbNull / JsonNull sentinels still pass validation for optional typed-JSON fields
• Run existing regression tests: pnpm test

🤖 Generated with Claude Code

Summary by CodeRabbit

Release Notes

• Refactor
  • Optimized internal schema validation handling to improve efficiency and consistency in data mutation processing.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: c80e36d2-29ad-42b8-892e-a6e086af1e70

📥 Commits

Reviewing files that changed from the base of the PR and between e9482fe and faa5166.

📒 Files selected for processing (1)

• packages/orm/src/client/zod/factory.ts

---

📝 Walkthrough

Walkthrough

The function makeNullableTypedJsonMutationSchema was refactored to accept a prebuilt Zod schema instead of constructing one internally. The validation logic now delegates to the provided schema while maintaining null/undefined handling. Callers were updated to pass precomputed field schemas.

Changes

Cohort / File(s)	Summary
Zod Factory Refactoring packages/orm/src/client/zod/factory.ts	Changed makeNullableTypedJsonMutationSchema to accept a prebuilt fieldSchema parameter instead of constructing the schema internally. Updated validation to use fieldSchema.safeParse() and modified both create/update data paths to pass precomputed field schemas.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Poem

🐰✨ A schema once built, now passed with care,
No more construction, just reuse we share,
The rabbit hops through mutations with grace,
Each validation finding its proper place! 🎯

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main fix: preserving array mutation shape (set, push operators) for optional typed-JSON fields by changing how the schema is constructed.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/nullable-typed-json-array-mutation-schema

---

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.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[claude]

Claude finished @ymc9's task —— View job

---

PR Code Review

Review Tasks

• Read and analyze the modified file
• Review code quality and implementation
• Check for potential bugs or issues
• Assess performance implications
• Review security considerations
• Evaluate test coverage needs
• Provide comprehensive feedback

Summary

This PR fixes a bug in the Zod schema factory where makeNullableTypedJsonMutationSchema was discarding pre-built field schemas that contained array mutation operators ({ set }, { push }). The fix changes the function signature to accept the already-built fieldSchema parameter instead of reconstructing it from scratch.

Code Quality & Implementation ✅

Excellent implementation with clear improvements:

1. Clear problem identification: The comments in packages/orm/src/client/zod/factory.ts:614-617 perfectly explain the issue and solution
2. Minimal, focused change: Only touches the specific problematic function and its two call sites
3. Parameter signature improvement: Changes from (type: string, attributes?: readonly AttributeApplication[]) to (fieldSchema: z.ZodTypeAny) - much cleaner and preserves caller's work
4. Maintains backward compatibility: Internal function change doesn't affect public API

Potential Issues ✅

No significant issues found:

1. Type safety: Uses z.ZodTypeAny which is the correct parent type for all Zod schemas
2. Null handling: Properly preserves existing null value validation logic (packages/orm/src/client/zod/factory.ts:623-628)
3. Error propagation: Correctly forwards Zod validation errors using ctx.addIssue(issue as any) (packages/orm/src/client/zod/factory.ts:632)

Performance Considerations ✅

Performance neutral or slightly improved:

1. Eliminates redundant work: No longer calls makeScalarSchema(type, attributes) to rebuild what callers already built
2. Same validation path: Uses the same safeParse approach, no performance regression
3. Memory efficiency: Reuses existing schema object instead of creating a new one

Security Considerations ✅

No security concerns:

1. Input validation preserved: All existing Zod validation rules still apply
2. No new attack vectors: Changes internal implementation without exposing new inputs
3. Null sentinel handling intact: Still properly handles DbNull, JsonNull, AnyNull classes

Test Coverage Assessment ⚠️

Test coverage exists but could be more comprehensive:

Existing coverage:

• Basic typed JSON field tests in /tests/e2e/orm/client-api/typed-json-fields.test.ts
• General JSON filter tests

Missing coverage for this specific fix:

• No tests specifically for optional typed-JSON array fields with mutation operators
• Missing test cases for { set: [...] } on create operations
• Missing test cases for { set, push } operators on update operations
• No regression test for the bug this PR fixes

Recommendation: Add test cases to verify:

// Create with array mutation shape
await client.model.create({
  data: {
    optionalTypedJsonArray: { set: [item1, item2] }
  }
})

// Update with array mutation operators  
await client.model.update({
  where: { id: 1 },
  data: {
    optionalTypedJsonArray: { push: newItem }
  }
})

Overall Assessment 🎯

High quality fix with excellent implementation:

• ✅ Addresses the root cause correctly
• ✅ Minimal, surgical change
• ✅ Well-documented with clear comments
• ✅ No breaking changes
• ⚠️ Could benefit from more specific test coverage

Recommendation: Approve with suggestion to add regression tests for the specific bug scenario.