[Tejas Lamba (TejasLamba2006)]

When stripNulls filters an outer array and assigns the result to obj[k], recursion still visited the old unfiltered array held in value. Nested replacements were disconnected from the returned structure.

The recursive call now uses obj[k] directly instead of the stale value reference, so it always operates on the post-filter array.

Added regression tests for nested arrays (groups: number[][]) and flat arrays to prevent regressions.

Related to #496.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota. Try requesting a review at a lower effort level.

[Copilot]

🔵 Needs a closer look

A maintainer needs to reconcile this change with the duplicate fix in open draft PR #496 and preserve contributor credit before merging.

No open issues 🎉

_{Review effort: Max}

[robgruen]

Thanks Tejas Lamba (@TejasLamba2006) for this focused fix. The nested-array regression catches the stale-reference bug, and the implementation looks sound in local testing.

One improvement I'd recommend is using createTypeScriptJsonValidator in the regression tests instead of a custom validator that only checks Array.isArray. Keep attemptRepair = false and close the validator in finally. That would verify not just the resulting array contents, but also that the cleaned response actually satisfies groups: number[][].

It would be useful to cover both:

• {"groups":[null,[null,1]]} -> {"groups":[[1]]}
• {"groups":[[null,1],null,[null,null]]} -> {"groups":[[1],[]]}

JOJO (@moneymaker26754-crypto), thank you for the earlier work in #496, which already includes these real-validator regression cases and a fix for the same root cause. Please coordinate the overlapping contributions; if the work is consolidated into one PR, let's preserve credit for both authors, including the implementation and regression coverage.

Since #496 is itself an open draft PR rather than an issue, "Related to #496" would describe the relationship more accurately than "Fixes #496".

[robgruen]

Please see my improvement comment.

[robgruen]

Can you two (Tejas Lamba (@TejasLamba2006) , JOJO (@moneymaker26754-crypto)) please consolidate into #497 .

[Tejas Lamba (TejasLamba2006)]

Sounds good, happy to consolidate into #497. JOJO (@moneymaker26754-crypto) if you want to pull the nested-array stripNulls fix into your draft, feel free to cherry-pick from this branch or I can push it there directly. Either way works for me, just want to make sure both fixes land.

Will update the "Fixes #496" reference to "Related to #496" as suggested.