duckdb: Exit earlier on Validity::AllFalse

Signed-off-by: Mikhail Kot <to@myrrc.dev>

Author: myrrc

vortex-duckdb/src/exporter/all_invalid.rs

@@ -3,47 +3,32 @@
 
 use vortex::array::ExecutionCtx;
 use vortex::error::VortexResult;
-use vortex::error::vortex_ensure;
 
-use crate::duckdb::LogicalTypeRef;
-use crate::duckdb::Value;
 use crate::duckdb::VectorRef;
 use crate::exporter::ColumnExporter;
 
-struct AllInvalidExporter {
-    len: usize,
-    null_value: Value,
-}
+struct AllInvalidExporter;
 
-pub(crate) fn new_exporter(len: usize, logical_type: &LogicalTypeRef) -> Box<dyn ColumnExporter> {
-    Box::new(AllInvalidExporter {
-        len,
-        null_value: Value::null(logical_type),
-    })
+pub(crate) fn new_exporter() -> Box<dyn ColumnExporter> {
+    Box::new(AllInvalidExporter {})
 }
 
 impl ColumnExporter for AllInvalidExporter {
     fn export(
         &self,
-        offset: usize,
-        len: usize,
+        _offset: usize,
+        _len: usize,
         vector: &mut VectorRef,
         _ctx: &mut ExecutionCtx,
     ) -> VortexResult<()> {
-        vortex_ensure!(
-            offset + len <= self.len,
-            "invalid exporter: offset + len must be less than or equal to len"
-        );
-
-        vector.reference_value(&self.null_value);
+        vector.set_all_false_validity();
         Ok(())
     }
 }
 
 #[cfg(test)]
 mod tests {
     use vortex::array::VortexSessionExecute;
-    use vortex::array::arrays::PrimitiveArray;
 
     use super::*;
     use crate::SESSION;
@@ -52,12 +37,10 @@ mod tests {
 
     #[test]
     fn all_null_array() {
-        let arr = PrimitiveArray::from_option_iter::<i32, _>([None, None, None]);
         let ltype = LogicalType::int32();
+        let mut chunk = DataChunk::new([ltype]);
 
-        let mut chunk = DataChunk::new([ltype.clone()]);
-
-        new_exporter(arr.len(), &ltype)
+        new_exporter()
             .export(
                 0,
                 3,

vortex-duckdb/src/exporter/bool.rs

@@ -4,11 +4,11 @@
 use vortex::array::ExecutionCtx;
 use vortex::array::arrays::BoolArray;
 use vortex::array::arrays::bool::BoolArrayExt;
+use vortex::array::validity::Validity;
 use vortex::buffer::BitBuffer;
 use vortex::error::VortexResult;
 use vortex::mask::Mask;
 
-use crate::duckdb::LogicalType;
 use crate::duckdb::VectorRef;
 use crate::exporter::ColumnExporter;
 use crate::exporter::all_invalid;
@@ -24,11 +24,12 @@ pub(crate) fn new_exporter(
 ) -> VortexResult<Box<dyn ColumnExporter>> {
     let len = array.len();
     let bits = array.to_bit_buffer();
-    let validity = array.validity()?.to_array(len).execute::<Mask>(ctx)?;
 
-    if validity.all_false() {
-        return Ok(all_invalid::new_exporter(len, &LogicalType::bool()));
+    let validity = array.validity()?;
+    if matches!(validity, Validity::AllInvalid) {
+        return Ok(all_invalid::new_exporter());
     }
+    let validity = validity.to_array(len).execute::<Mask>(ctx)?;
 
     Ok(validity::new_exporter(
         validity,

vortex-duckdb/src/exporter/constant.rs

@@ -65,14 +65,14 @@ impl ColumnExporter for ConstantExporter {
     fn export(
         &self,
         _offset: usize,
-        len: usize,
+        _len: usize,
         vector: &mut VectorRef,
         _ctx: &mut ExecutionCtx,
     ) -> VortexResult<()> {
         match self.value.as_ref() {
             None => {
                 // TODO(ngates): would be good if DuckDB supported constant null vectors.
-                unsafe { vector.set_validity(&Mask::AllFalse(len), 0, len) };
+                vector.set_all_false_validity();
             }
             Some(value) => {
                 vector.reference_value(value);

vortex-duckdb/src/exporter/decimal.rs

@@ -8,9 +8,9 @@ use vortex::array::ExecutionCtx;
 use vortex::array::arrays::DecimalArray;
 use vortex::array::arrays::decimal::DecimalDataParts;
 use vortex::array::match_each_decimal_value_type;
+use vortex::array::validity::Validity;
 use vortex::buffer::Buffer;
 use vortex::dtype::BigCast;
-use vortex::dtype::DType;
 use vortex::dtype::DecimalDType;
 use vortex::dtype::DecimalType;
 use vortex::dtype::NativeDecimalType;
@@ -19,7 +19,6 @@ use vortex::error::VortexResult;
 use vortex::error::vortex_bail;
 use vortex::mask::Mask;
 
-use crate::duckdb::LogicalType;
 use crate::duckdb::VectorBuffer;
 use crate::duckdb::VectorRef;
 use crate::exporter::ColumnExporter;
@@ -49,13 +48,11 @@ pub(crate) fn new_exporter(
         values,
     } = array.into_data_parts();
     let dest_values_type = precision_to_duckdb_storage_size(&decimal_dtype)?;
-    let nullability = validity.nullability();
-    let validity = validity.to_array(len).execute::<Mask>(ctx)?;
 
-    if validity.all_false() {
-        let ltype = LogicalType::try_from(DType::Decimal(decimal_dtype, nullability))?;
-        return Ok(all_invalid::new_exporter(len, &ltype));
+    if matches!(validity, Validity::AllInvalid) {
+        return Ok(all_invalid::new_exporter());
     }
+    let validity = validity.to_array(len).execute::<Mask>(ctx)?;
 
     let exporter = if values_type == dest_values_type {
         match_each_decimal_value_type!(values_type, |D| {

vortex-duckdb/src/exporter/dict.rs

@@ -17,7 +17,6 @@ use vortex::dtype::IntegerPType;
 use vortex::error::VortexResult;
 use vortex::mask::Mask;
 
-use crate::duckdb::LogicalType;
 use crate::duckdb::ReusableDict;
 use crate::duckdb::SelectionVector;
 use crate::duckdb::VectorRef;
@@ -43,7 +42,6 @@ pub(crate) fn new_exporter_with_flatten(
 ) -> VortexResult<Box<dyn ColumnExporter>> {
     // Grab the cache dictionary values.
     let values = array.values();
-    let values_type: LogicalType = values.dtype().try_into()?;
     if let Some(constant) = values.as_opt::<Constant>() {
         return constant::new_exporter_with_mask(
             ConstantArray::new(constant.scalar().clone(), array.codes().len()),
@@ -57,7 +55,7 @@ pub(crate) fn new_exporter_with_flatten(
 
     match codes_mask {
         Mask::AllTrue(_) => {}
-        Mask::AllFalse(len) => return Ok(all_invalid::new_exporter(len, &values_type)),
+        Mask::AllFalse(_) => return Ok(all_invalid::new_exporter()),
         Mask::Values(_) => {
             // duckdb cannot have a dictionary with validity in the codes, so flatten the array and
             // apply the validity mask there.

vortex-duckdb/src/exporter/fixed_size_list.rs

@@ -11,14 +11,14 @@
 use vortex::array::ExecutionCtx;
 use vortex::array::arrays::FixedSizeListArray;
 use vortex::array::arrays::fixed_size_list::FixedSizeListArrayExt;
+use vortex::array::validity::Validity;
 use vortex::error::VortexResult;
 use vortex::mask::Mask;
 
 use super::ConversionCache;
 use super::all_invalid;
 use super::new_array_exporter_with_flatten;
 use super::validity;
-use crate::duckdb::LogicalType;
 use crate::duckdb::VectorRef;
 use crate::exporter::ColumnExporter;
 
@@ -42,15 +42,14 @@ pub(crate) fn new_exporter(
     let parts = array.into_data_parts();
     let elements = parts.elements;
     let validity = parts.validity;
-    let dtype = parts.dtype;
-    let mask = validity.to_array(len).execute::<Mask>(ctx)?;
-    let elements_exporter = new_array_exporter_with_flatten(elements, cache, ctx, true)?;
 
-    if mask.all_false() {
-        let ltype = LogicalType::try_from(dtype)?;
-        return Ok(all_invalid::new_exporter(len, &ltype));
+    if matches!(validity, Validity::AllInvalid) {
+        return Ok(all_invalid::new_exporter());
     }
 
+    let mask = validity.to_array(len).execute::<Mask>(ctx)?;
+    let elements_exporter = new_array_exporter_with_flatten(elements, cache, ctx, true)?;
+
     Ok(validity::new_exporter(
         mask,
         Box::new(FixedSizeListExporter {

vortex-duckdb/src/exporter/list.rs

@@ -10,6 +10,7 @@ use vortex::array::arrays::ListArray;
 use vortex::array::arrays::PrimitiveArray;
 use vortex::array::arrays::list::ListDataParts;
 use vortex::array::match_each_integer_ptype;
+use vortex::array::validity::Validity;
 use vortex::dtype::IntegerPType;
 use vortex::error::VortexResult;
 use vortex::error::vortex_err;
@@ -49,15 +50,14 @@ pub(crate) fn new_exporter(
         elements,
         offsets,
         validity,
-        dtype,
+        dtype: _dtype,
     } = array.into_data_parts();
     let num_elements = elements.len();
-    let validity = validity.to_array(array_len).execute::<Mask>(ctx)?;
 
-    if validity.all_false() {
-        let ltype = LogicalType::try_from(dtype)?;
-        return Ok(all_invalid::new_exporter(array_len, &ltype));
+    if matches!(validity, Validity::AllInvalid) {
+        return Ok(all_invalid::new_exporter());
     }
+    let validity = validity.to_array(array_len).execute::<Mask>(ctx)?;
 
     let values_key = elements.addr();
     // Check if we have a cached vector and extract it if we do.

vortex-duckdb/src/exporter/list_view.rs

@@ -10,7 +10,7 @@ use vortex::array::arrays::ListViewArray;
 use vortex::array::arrays::PrimitiveArray;
 use vortex::array::arrays::listview::ListViewDataParts;
 use vortex::array::match_each_integer_ptype;
-use vortex::dtype::DType;
+use vortex::array::validity::Validity;
 use vortex::dtype::IntegerPType;
 use vortex::error::VortexResult;
 use vortex::error::vortex_err;
@@ -56,13 +56,11 @@ pub(crate) fn new_exporter(
     } = array.into_data_parts();
     // Cache an `elements` vector up front so that future exports can reference it.
     let num_elements = elements.len();
-    let nullability = validity.nullability();
-    let validity = validity.to_array(len).execute::<Mask>(ctx)?;
 
-    if validity.all_false() {
-        let ltype = LogicalType::try_from(DType::List(elements_dtype, nullability))?;
-        return Ok(all_invalid::new_exporter(len, &ltype));
+    if matches!(validity, Validity::AllInvalid) {
+        return Ok(all_invalid::new_exporter());
     }
+    let validity = validity.to_array(len).execute::<Mask>(ctx)?;
 
     let values_key = elements.addr();
     // Check if we have a cached vector and extract it if we do.

vortex-duckdb/src/exporter/mod.rs

@@ -37,7 +37,6 @@ use vortex::error::VortexResult;
 use vortex::error::vortex_bail;
 
 use crate::duckdb::DataChunkRef;
-use crate::duckdb::LogicalType;
 use crate::duckdb::VectorRef;
 use crate::duckdb::duckdb_vector_size;
 
@@ -166,7 +165,7 @@ fn new_array_exporter_with_flatten(
 
     // Otherwise, we fall back to canonical
     match array.execute::<Canonical>(ctx)? {
-        Canonical::Null(array) => Ok(all_invalid::new_exporter(array.len(), &LogicalType::null())),
+        Canonical::Null(_) => Ok(all_invalid::new_exporter()),
         Canonical::Bool(array) => bool::new_exporter(array, ctx),
         Canonical::Primitive(array) => primitive::new_exporter(array, ctx),
         Canonical::Decimal(array) => decimal::new_exporter(array, ctx),

vortex-duckdb/src/exporter/primitive.rs

@@ -6,11 +6,11 @@ use std::marker::PhantomData;
 use vortex::array::ExecutionCtx;
 use vortex::array::arrays::PrimitiveArray;
 use vortex::array::match_each_native_ptype;
+use vortex::array::validity::Validity;
 use vortex::dtype::NativePType;
 use vortex::error::VortexResult;
 use vortex::mask::Mask;
 
-use crate::duckdb::LogicalType;
 use crate::duckdb::VectorBuffer;
 use crate::duckdb::VectorRef;
 use crate::exporter::ColumnExporter;
@@ -28,15 +28,11 @@ pub fn new_exporter(
     array: PrimitiveArray,
     ctx: &mut ExecutionCtx,
 ) -> VortexResult<Box<dyn ColumnExporter>> {
-    let validity = array
-        .validity()?
-        .to_array(array.len())
-        .execute::<Mask>(ctx)?;
-
-    if validity.all_false() {
-        let ltype = LogicalType::try_from(array.ptype())?;
-        return Ok(all_invalid::new_exporter(array.len(), &ltype));
-    }
+    let validity = array.validity()?;
+    if matches!(validity, Validity::AllInvalid) {
+        return Ok(all_invalid::new_exporter());
+    };
+    let validity = validity.to_array(array.len()).execute::<Mask>(ctx)?;
 
     match_each_native_ptype!(array.ptype(), |T| {
         let buffer = array.to_buffer::<T>();