hareshkh opened a new pull request, #20679:
URL: https://github.com/apache/datafusion/pull/20679
## Which issue does this PR close?
- Closes #.
## Rationale for this change
When the Substrait consumer applies root-level schema renames that involve
struct field renames inside an aggregate measure's return type (e.g.
`List<Struct{c0,c1}>` → `List<Struct{one,two}>`), `rename_expressions` injects
`Expr::Cast` around the aggregate function. The physical planner rejects
Cast-wrapped aggregates, causing a planning error.
## What changes are included in this PR?
- Extracted the root-level rename logic from
`from_substrait_plan_with_consumer` into a dedicated `apply_renames` function
in `plan.rs`.
- Added a new `alias_expressions` utility in `utils.rs` that applies only
top-level column name aliases without injecting `Expr::Cast`, making it safe
for aggregate expressions.
- Changed the `Aggregate` rename strategy: first apply alias-only renames
via `alias_expressions`; if the schema still doesn't match (e.g. nested struct
field renames require casts), wrap the Aggregate in a Projection that carries
those casts.
- Added a test (`aggregate_with_struct_field_rename`) that constructs a
Substrait plan with `array_agg` returning `List<Struct{c0,c1}>` and root names
that rename the struct fields to `one` and `two`, verifying the consumer
produces a valid plan that the physical planner accepts.
## Are these changes tested?
Yes. A new test case `aggregate_with_struct_field_rename` in
`aggregation_tests.rs` builds a Substrait plan with a struct field rename on an
aggregate measure and verifies that:
1. The resulting logical plan wraps the Aggregate in a Projection.
2. The physical planner accepts the plan and can execute it
(`DataFrame::show()`).
## Are there any user-facing changes?
No. This is a bug fix in the Substrait consumer that corrects plan
generation for aggregate expressions with nested struct field renames. There
are no API changes.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
