fix: use read schema field names when rebuilding non-partition struct

[Smith-Cruise]

Purpose

It looks like a typo, we need to reconstruct RecordBatch's schema from read_schema, not from data_schema.

Tests

API and Format

Documentation

Generative AI tooling

[lxy-9602]

Good catch! The inconsistency you spotted (using non_partition_data_schema name in one branch vs non_partition_read_schema in the other) indeed exists. However, it won't affect the final result because MappingFields is called right after CastNonPartitionArrayIfNeed and overwrites the field names using non_partition_read_schema[i].Name().

That said, I'd suggest a slightly different fix: change line 110 to:

casted_field_names.push_back(non_partition_info_.non_partition_data_schema[i].Name());

instead of changing line 115.

The reasoning is that CastNonPartitionArrayIfNeed should only be responsible for type casting, not for name mapping. Field name remapping (e.g. due to RENAME COLUMN) should be left entirely to MappingFields, which is the dedicated function for that purpose. This keeps a cleaner separation of concerns.

Could you also add a test case that exercises the interaction between RENAME COLUMN and TYPE CAST to verify correctness? Something like:

TEST_F(FieldMappingReaderTest, TestReadWithSchemaEvolutionRenameOnCastField) {
    std::vector<DataField> data_fields = {
        DataField(0, arrow::field("f0", arrow::utf8())),
        DataField(1, arrow::field("f1", arrow::int32())),
        DataField(2, arrow::field("f2", arrow::int32())),
        DataField(3, arrow::field("f3", arrow::int32())),
    };
    auto data_schema = DataField::ConvertDataFieldsToArrowSchema(data_fields);
    auto data_array = std::dynamic_pointer_cast<arrow::StructArray>(
        arrow::ipc::internal::json::ArrayFromJSON(arrow::struct_(data_schema->fields()),
                                                  R"([
        ["Bob", 100, 10, 1],
        ["Emily", 200, 20, 2],
        ["Alice", 300, 30, 3]
    ])")
            .ValueOrDie());

    std::vector<DataField> read_fields = {
        DataField(0, arrow::field("f0", arrow::utf8())),
        DataField(1, arrow::field("new_f1", arrow::int32())),
        DataField(2, arrow::field("f2", arrow::utf8())),
        DataField(3, arrow::field("new_f3", arrow::utf8())),
    };
    auto read_schema = DataField::ConvertDataFieldsToArrowSchema(read_fields);

    auto expected = std::dynamic_pointer_cast<arrow::StructArray>(
        arrow::ipc::internal::json::ArrayFromJSON(arrow::struct_(read_schema->fields()),
                                                  R"([
        ["Bob", 100, "10", "1"],
        ["Emily", 200, "20", "2"],
        ["Alice", 300, "30", "3"]
    ])")
            .ValueOrDie());

    CheckResult(data_schema, data_array, read_schema, /*predicate=*/nullptr,
                /*partition_keys=*/{}, BinaryRow::EmptyRow(), expected);
}

This covers the case where some fields are both renamed AND cast (f3→new_f3, int32→utf8), some are only renamed (f1→new_f1), and some are only cast (f2, int32→utf8), ensuring the fix doesn't break any combination.

[lxy-9602]

I've already fixed this issue and added the corresponding tests in #293 , which has been merged. Closing this PR to avoid duplication.
Please feel free to comment if you have any questions or concerns.