refactor(mock): emit strict mock types once from structured names

[Hypenate]

Summary

Follow-up refactor from strict mock types (#3525). Aligns the faker schemas: true path with MSW/operation mocks: generators only emit factories, and finalizeMockImplementation prepends strict helpers and {Schema}Mock aliases once from strictMockSchemaTypeNames.

• generateFakerForSchemas returns strictMockSchemaTypeNames and no longer inlines strict type blocks
• writeFakerSchemaMocks calls finalizeMockImplementation (same as other writers)
• dedupeStrictMockTypeDeclarations only prepends from the structured set (removed regex scraping/strip logic)

Closes #3542

Test plan

• @orval/mock unit tests
• orval-tests build (mock client typecheck)
• CI green on PR

Summary by CodeRabbit

• Improvements

  • Faker mock generation now exposes strict schema type names separately, producing leaner mock implementations and enabling predictable header insertion and ordering.

• Bug Fixes

  • Finalized mock output is used when computing imports and emitting files so generated files reflect finalized implementation.

• New Features

  • Generated mock types and factory functions added for additional schemas (e.g., Status, PetList, Score).

• Tests

  • Updated/added tests covering strict-type exposure, deduplication/header behavior, and finalized-output flow.

[coderabbitai]

📝 Walkthrough

Walkthrough

generateFakerForSchemas no longer inlines strict-mock helper/type declarations; it returns an optional strictMockSchemaTypeNames array. Regex scraping helpers were removed, deduplication now uses structured names, writers accept the names for finalization, and tests/exports/spec snapshots were updated.

Changes

Strict Mock Schema Type Name Extraction

Layer / File(s)	Summary
Faker generation refactor packages/mock/src/faker/index.ts, packages/mock/src/faker/index.test.ts, tests/__snapshots__/mock/issue-3525/model/*	generateFakerForSchemas omits inline strict-mock helper/type declarations and returns strictMockSchemaTypeNames?: string[]; tests and generated snapshots updated to match new output shape.
Mock types utilities refactor packages/mock/src/mock-types.ts, packages/mock/src/mock-types.test.ts	Removed regex-based scraping helpers (collectStrictMockSchemaTypeNames, collectStrictMockSchemaNamesFromUsage). dedupeStrictMockTypeDeclarations now uses options.strictSchemaTypeNames (deduped), early-returns when empty, trims body, and prepends the generated strict-mock header.
Integration, writers, exports, and specs packages/orval/src/write-specs.ts, packages/mock/src/index.ts, tests/specifications/issue-3525.yaml, tests/__snapshots__/mock/issue-3525/*	writeFakerSchemaMocks captures strictMockSchemaTypeNames and optionally passes them to builder.finalizeMockImplementation; finalized implementation is used for dependency import generation and file emission. Public re-export for scraping helper removed; spec and snapshot files updated to include additional model modules.

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~45 minutes

Possibly related PRs

• orval-labs/orval#3529: Related changes to strict-mock type/emission handling in packages/mock/src/faker/index.ts.
• orval-labs/orval#3426: Prior changes to faker schema generator API and related emitters.

Suggested reviewers

• melloware

Poem

🐰 I hopped through code with tidy paws,

strict names now leave no hidden cause.
No regex dust, no duplicated art,
a cleaner file — a happier heart. 🥕

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The PR title accurately summarizes the main refactoring: moving strict mock type emission from inline generation to a structured output that is finalized once at file write time.
Linked Issues check	✅ Passed	All coding requirements from #3542 are met: generateFakerForSchemas returns strictMockSchemaTypeNames, inline helpers are removed, dedupeStrictMockTypeDeclarations uses structured names without regex, and non-overridable schemas are now included.
Out of Scope Changes check	✅ Passed	All changes are directly scoped to the stated objectives: refactoring strict mock type emission in faker/MSW paths, updating corresponding tests, and fixing the non-overridable schema bug. No unrelated changes detected.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

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

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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.}

[coderabbitai]

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

packages/mock/src/faker/index.ts (1)

168-170: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Track strict names for every strict schema factory.

Line 168 only records typeName for overridable schemas, but getMockFactorySignatureParts() in packages/mock/src/mock-types.ts still returns {Schema}Mock for strict non-overridable factories. That means a top-level array/scalar/enum schema can generate getFooMock(): FooMock while writeFakerSchemaMocks() never receives Foo in strictSchemaTypeNames, so the finalizer will skip export type FooMock = … and the emitted file stops typechecking.

Suggested fix

-    if (isStrictMock(mockOptions) && isOverridable) {
+    if (isStrictMock(mockOptions)) {
       strictMockTypeNames.add(typeName);
     }

Please add a regression test for a strict non-object schema when fixing this.

🤖 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 `@packages/mock/src/faker/index.ts` around lines 168 - 170, The code only adds
typeName to strictMockTypeNames when isOverridable is true, causing
non-overridable strict schemas to be omitted; update the logic in
packages/mock/src/faker/index.ts (around isStrictMock and isOverridable) to
always add typeName to strictMockTypeNames when isStrictMock(mockOptions) is
true, regardless of isOverridable, so writeFakerSchemaMocks and
strictSchemaTypeNames receive the strict names expected by
getMockFactorySignatureParts (which returns {Schema}Mock for strict
non-overridable factories); after fixing, add a regression test that creates a
strict non-object schema (e.g., top-level enum/array/scalar) and asserts the
emitted file includes the corresponding exported type alias (e.g., FooMock) so
the emitted types typecheck.

🤖 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.

Outside diff comments:
In `@packages/mock/src/faker/index.ts`:
- Around line 168-170: The code only adds typeName to strictMockTypeNames when
isOverridable is true, causing non-overridable strict schemas to be omitted;
update the logic in packages/mock/src/faker/index.ts (around isStrictMock and
isOverridable) to always add typeName to strictMockTypeNames when
isStrictMock(mockOptions) is true, regardless of isOverridable, so
writeFakerSchemaMocks and strictSchemaTypeNames receive the strict names
expected by getMockFactorySignatureParts (which returns {Schema}Mock for strict
non-overridable factories); after fixing, add a regression test that creates a
strict non-object schema (e.g., top-level enum/array/scalar) and asserts the
emitted file includes the corresponding exported type alias (e.g., FooMock) so
the emitted types typecheck.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: d0b9435d-6b6a-4cf3-ae6d-12a08d809da1

📥 Commits

Reviewing files that changed from the base of the PR and between 1739cdf and eaa2714.

📒 Files selected for processing (6)

• packages/mock/src/faker/index.test.ts
• packages/mock/src/faker/index.ts
• packages/mock/src/index.ts
• packages/mock/src/mock-types.test.ts
• packages/mock/src/mock-types.ts
• packages/orval/src/write-specs.ts

💤 Files with no reviewable changes (1)

• packages/mock/src/index.ts

[Hypenate]

@wadakatu We merged the previous PR.
This PR contains the simplification you wanted earlier (I hope)
If you have time, can you review it?

[wadakatu]

Review — LGTM ✅

Beyond being a clean refactor, this actually fixes a latent TS2304 bug for non-overridable strict faker schemas. I regenerated and typechecked locally against a spec with top-level enum/array/scalar schemas.

The one real behavioral change: dropping && isOverridable (= a latent bug fix)

getMockFactorySignatureParts returns ${Type}Mock as the return type even for non-overridable schemas (top-level enum/array/scalar) under strict mode, but on master the guard means the XMock type declaration is never emitted:

• master: error TS2304: Cannot find name 'StatusMock' / 'PetListMock' / 'ScoreMock' → doesn't compile
• this PR: the same spec typechecks ✅

Note that keeping the guard and only doing the refactor would have left this just as broken as master, so commit 3 is an intentional, standalone bug fix rather than a side effect of the refactor. Dropping the guard makes "every XMock referenced by a strict factory" line up exactly with strictMockSchemaTypeNames, which is why removing the scraping safety nets (collectStrictMockSchemaNamesFromUsage, etc.) is safe here. The MSW path is sound for the same reason: applyStrictMockReturnType and strictMockSchemaTypeNames both derive from the same source (getSchemaTypeNamesFromResponses), so completeness is guaranteed by construction.

Suggestions (non-blocking)

1. [Recommended] Add an e2e fixture for the non-overridable strict faker case. This bug fix isn't guarded by the current CI gates (snapshot / typecheck-generated): the existing strict fixtures (issue-3525 family) only contain Pet (an object = overridable). Adding a top-level enum + array schema to the issue-3525 spec would lock the regression down in CI.
2. dedupeStrictMockTypeDeclarations is no longer idempotent (it prepends unconditionally without stripping an existing header). A one-line comment noting the "called once per file" invariant would make that safe to rely on.
3. Nit: .replace(/^\n+/, '').trimStart() — .trimStart() alone already covers leading newlines. And the header ? ... : trimmedBody false branch is effectively dead since we early-return when there are no names.

Nice cleanup 👍

[Hypenate]

@wadakatu I summon you again :D
I've implemented the 3 suggestions you gave.

[wadakatu]

All three are addressed cleanly — thanks for the quick turnaround! 🙌

Verified locally on the latest commit:

• Suggestion 1 (fixture): regenerated issue-3525 and the new Status / PetList / Score schemas now emit StatusMock / PetListMock / ScoreMock and typecheck (incl. the odd-but-valid ScoreMock over number). The non-overridable strict-faker path is now locked down by both the snapshot and typecheck-generated gates.
• Suggestion 2 (invariant): the "not idempotent / call once per file" note matches how the writers and writeFakerSchemaMocks use it. 👍
• Suggestion 3 (cleanup): ${header}\n\n${implementation.trimStart()} is output-equivalent (header is always non-empty once we're past the empty-names early return, and .trimStart() subsumes the old .replace(/^\n+/, '')).

Full check: @orval/mock units (219) green, snapshot suite green after a fresh regen, and tsc --noEmit over all of generated/mock is clean — no output changes anywhere outside issue-3525, so no regressions.

LGTM 🚀