[AURON #1845] Fix StringSplit conversion

[Flyangz]

Which issue does this PR close?

Closes #1845

Rationale for this change

before: ProjectExec [UDFWrapper(split(col1#1, ,, -1)[0]) AS #5], schema=[#5:Utf8;N]
after: ProjectExec [(spark_ext_function_Spark_StringSplit(#1@0, ,)).[1] AS #5], schema=[#5:Utf8;N]

What changes are included in this PR?

Are there any user-facing changes?

How was this patch tested?

[richox]

one concern: UDFWrapper(split, [1]) fallbacks one item from the result array. but UDFWrapper(split)[1] fallbacks the whole array, increasing a lot of fallback costs.

[Flyangz]

@richox Yes. I'll now break down the original issue into two steps. This PR fixes the problem of StringSplit failing to convert to Native. The other issue #1902 is the ext function caching problem. Completing these two steps will better resolve the original problem.

[Copilot]

Pull request overview

This PR fixes the StringSplit conversion to correctly specify that it returns an array of strings rather than a single string. The issue was that the native function conversion was incorrectly declaring the return type as StringType when it should be ArrayType(StringType).

Changes:

• Added ArrayType to imports and consolidated type imports
• Added defensive .toString calls when accessing Literal.value
• Fixed return type from StringType to ArrayType(StringType) for the native StringSplit function

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.