bitmap_instead_smaller_datatypes

Author: coderfender

datafusion/functions-aggregate-common/src/aggregate/count_distinct/native.rs

@@ -230,9 +230,9 @@ impl Accumulator for BoolArray256DistinctCountAccumulator {
             .filter_map(|(idx, &seen)| if seen { Some(idx as u8) } else { None })
             .collect();
 
-        let arr = Arc::new(PrimitiveArray::<arrow::datatypes::UInt8Type>::from_iter_values(
-            values,
-        ));
+        let arr = Arc::new(
+            PrimitiveArray::<arrow::datatypes::UInt8Type>::from_iter_values(values),
+        );
         Ok(vec![
             SingleRowListArrayBuilder::new(arr).build_list_scalar(),
         ])
@@ -308,18 +308,16 @@ impl Accumulator for BoolArray256DistinctCountAccumulatorI8 {
             .seen
             .iter()
             .enumerate()
-            .filter_map(|(idx, &seen)| {
-                if seen {
-                    Some(idx as u8 as i8)
-                } else {
-                    None
-                }
-            })
+            .filter_map(
+                |(idx, &seen)| {
+                    if seen { Some(idx as u8 as i8) } else { None }
+                },
+            )
             .collect();
 
-        let arr = Arc::new(PrimitiveArray::<arrow::datatypes::Int8Type>::from_iter_values(
-            values,
-        ));
+        let arr = Arc::new(
+            PrimitiveArray::<arrow::datatypes::Int8Type>::from_iter_values(values),
+        );
         Ok(vec![
             SingleRowListArrayBuilder::new(arr).build_list_scalar(),
         ])
@@ -409,9 +407,9 @@ impl Accumulator for Bitmap65536DistinctCountAccumulator {
             }
         }
 
-        let arr = Arc::new(PrimitiveArray::<arrow::datatypes::UInt16Type>::from_iter_values(
-            values,
-        ));
+        let arr = Arc::new(
+            PrimitiveArray::<arrow::datatypes::UInt16Type>::from_iter_values(values),
+        );
         Ok(vec![
             SingleRowListArrayBuilder::new(arr).build_list_scalar(),
         ])
@@ -502,9 +500,9 @@ impl Accumulator for Bitmap65536DistinctCountAccumulatorI16 {
             }
         }
 
-        let arr = Arc::new(PrimitiveArray::<arrow::datatypes::Int16Type>::from_iter_values(
-            values,
-        ));
+        let arr = Arc::new(
+            PrimitiveArray::<arrow::datatypes::Int16Type>::from_iter_values(values),
+        );
         Ok(vec![
             SingleRowListArrayBuilder::new(arr).build_list_scalar(),
         ])

datafusion/functions-aggregate/benches/count_distinct.rs

@@ -17,7 +17,9 @@
 
 use std::sync::Arc;
 
-use arrow::array::{ArrayRef, Int16Array, Int64Array, Int8Array, UInt16Array, UInt8Array};
+use arrow::array::{
+    ArrayRef, Int8Array, Int16Array, Int64Array, UInt8Array, UInt16Array,
+};
 use arrow::datatypes::{DataType, Field, Schema};
 use criterion::{Criterion, criterion_group, criterion_main};
 use datafusion_expr::function::AccumulatorArgs;

datafusion/functions-aggregate/src/count.rs

@@ -21,8 +21,8 @@ use arrow::{
     compute,
     datatypes::{
         DataType, Date32Type, Date64Type, Decimal128Type, Decimal256Type, Field,
-        FieldRef, Float16Type, Float32Type, Float64Type, Int32Type,
-        Int64Type, Time32MillisecondType, Time32SecondType, Time64MicrosecondType,
+        FieldRef, Float16Type, Float32Type, Float64Type, Int32Type, Int64Type,
+        Time32MillisecondType, Time32SecondType, Time64MicrosecondType,
         Time64NanosecondType, TimeUnit, TimestampMicrosecondType,
         TimestampMillisecondType, TimestampNanosecondType, TimestampSecondType,
         UInt32Type, UInt64Type,