Skip to content

Correct physical schema inference for DF files with non-logical types. #5976

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
AdamGS merged 5 commits into from
Jan 20, 2026
Merged

Correct physical schema inference for DF files with non-logical types. #5976

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
6b1fabd
vortex-datafusion: pushing down cast expressions loses physical types
asubiotto Jan 15, 2026
3d51732
Fix physical type logic for DF
AdamGS Jan 15, 2026
17e1000
clippy
AdamGS Jan 15, 2026
83ca36e
Assert sorted
AdamGS Jan 16, 2026
da75e59
Rename, move to dedicated file, add tests
AdamGS Jan 19, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
32 changes: 0 additions & 32 deletions vortex-datafusion/src/convert/exprs.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -235,19 +235,6 @@ impl ExpressionConvertor for DefaultExpressionConvertor {
return Ok(TreeNodeRecursion::Stop);
}

// If the projection contains a `CastColumnExpr` that casts to physical types that don't have a 1:1 mapping
// with Vortex's types system, we make sure to pull the input from the file and keep the projection
if let Some(cast_expr) = node.as_any().downcast_ref::<df_expr::CastColumnExpr>()
&& is_dtype_incompatible(cast_expr.target_field().data_type())
{
scan_projection.push((
cast_expr.input_field().name().clone(),
self.convert(cast_expr.expr().as_ref())?,
));
leftover_projection.push(projection_expr.clone());
return Ok(TreeNodeRecursion::Stop);
}

// DataFusion assumes different decimal types can be coerced.
// Vortex expects a perfect match so we don't push it down.
if let Some(binary_expr) = node.as_any().downcast_ref::<df_expr::BinaryExpr>()
Expand Down Expand Up @@ -430,25 +417,6 @@ fn can_scalar_fn_be_pushed_down(scalar_fn: &ScalarFunctionExpr) -> bool {
ScalarFunctionExpr::try_downcast_func::<GetFieldFunc>(scalar_fn).is_some()
}

fn is_dtype_incompatible(dt: &DataType) -> bool {
use DataType::*;

dt.is_run_ends_type()
|| is_decimal(dt)
|| matches!(
dt,
Dictionary(..)
| Utf8
| LargeUtf8
| Binary
| LargeBinary
| FixedSizeBinary(_)
| FixedSizeList(..)
| ListView(..)
| LargeListView(..)
)
}

// TODO(adam): Replace with `DataType::is_decimal` once its released.
fn is_decimal(dt: &DataType) -> bool {
matches!(
Expand Down
1 change: 1 addition & 0 deletions vortex-datafusion/src/convert/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,7 @@ use vortex::error::VortexResult;

pub(crate) mod exprs;
mod scalars;
pub(crate) mod schema;

/// First-party trait for implementing conversion from DataFusion types to Vortex types.
pub(crate) trait FromDataFusion<D: ?Sized>: Sized {
Expand Down
Loading
Loading