Skip to content

Support nested datatype filter pushdown to parquet #18560

@zhuqi-lucas

Description

@zhuqi-lucas

Is your feature request related to a problem or challenge?

Now we don't support pushdown nested datatype pushdown to parquet:

 fn check_single_column(&mut self, column_name: &str) -> Option<TreeNodeRecursion> {
        if let Ok(idx) = self.table_schema.index_of(column_name) {
            self.required_columns.insert(idx);
            if DataType::is_nested(self.table_schema.field(idx).data_type()) {
                self.non_primitive_columns = true;
                return Some(TreeNodeRecursion::Jump);
            }
        } else {
            // If the column does not exist in the (un-projected) table schema then
            // it must be a projected column.
            self.projected_columns = true;
            return Some(TreeNodeRecursion::Jump);
        }

        None
    }

But i think, some nest datatypes will be useful if we support pushdown, for example, the list column a, here is the case:

array_has_all(a, ['c'])

We can pushdown to parquet, so arrow-rs side can filter it? But i am not sure if arrow-rs support it.

Describe the solution you'd like

No response

Describe alternatives you've considered

No response

Additional context

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions