Description:

When running the Substrait tests with DataFusion built without the unicode_expressions feature (in datafusion/substrait/Cargo.toml only nested_expressions is enabled), the cases::roundtrip_logical_plan::simple_scalar_function_substr test fails with:

NotImplemented("Substring could not be planned by registered expr planner. Hint: enable the unicode_expressions")

Concretely:

• The UnicodeFunctionPlanner is not registered because the unicode_expressions feature is disabled.
• As a result, the SUBSTR / substring function cannot be planned under the Substrait logical plan conversion.
• This causes the simple_scalar_function_substr round‐trip logical plan test to fail under those build settings.

Steps to reproduce:

1. Checkout the DataFusion repo.
2. Build without unicode_expressions (so that only nested_expressions is enabled in the Substrait feature set).
3. Run the Substrait tests — cargo test -p datafusion-substrait --test substrait_integration or specifically cases::roundtrip_logical_plan::simple_scalar_function_substr
4. Observe the failure: NotImplemented("Substring could not be planned by registered expr planner. Hint: enable the unicode_expressions").

Expected behavior:

Either

• The Substr / substring function should be plannable even when unicode_expressions is disabled (i.e. provide a fallback planner / expression implementation), or
• At least the test suite should skip or mark this as expected behavior when unicode_expressions is off, rather than failing.

References:

• The error message: Substring could not be planned by registered expr planner. Hint: enable the unicode_expressions
• The failing test: cases::roundtrip_logical_plan::simple_scalar_function_substr

Issue was discovered in this PR comment - #17468 (comment)