feat: Support duration type in approx_distinct

datafusion/functions-aggregate/src/approx_distinct.rs

@@ -25,7 +25,8 @@ use arrow::array::{
 use arrow::buffer::NullBuffer;
 use arrow::datatypes::{
     ArrowPrimitiveType, DataType, Date32Type, Date64Type, Decimal32Type, Decimal64Type,
-    Decimal128Type, Decimal256Type, Field, FieldRef, Int32Type, Int64Type,
+    Decimal128Type, Decimal256Type, DurationMicrosecondType, DurationMillisecondType,
+    DurationNanosecondType, DurationSecondType, Field, FieldRef, Int32Type, Int64Type,
     IntervalDayTimeType, IntervalMonthDayNanoType, IntervalUnit, IntervalYearMonthType,
     Time32MillisecondType, Time32SecondType, Time64MicrosecondType, Time64NanosecondType,
     TimeUnit, TimestampMicrosecondType, TimestampMillisecondType,
@@ -781,6 +782,18 @@ impl AggregateUDFImpl for ApproxDistinct {
             DataType::Decimal256(_, _) => {
                 Box::new(NumericHLLAccumulator::<Decimal256Type>::new())
             }
+            DataType::Duration(TimeUnit::Second) => {
+                Box::new(NumericHLLAccumulator::<DurationSecondType>::new())
+            }
+            DataType::Duration(TimeUnit::Millisecond) => {
+                Box::new(NumericHLLAccumulator::<DurationMillisecondType>::new())
+            }
+            DataType::Duration(TimeUnit::Microsecond) => {
+                Box::new(NumericHLLAccumulator::<DurationMicrosecondType>::new())
+            }
+            DataType::Duration(TimeUnit::Nanosecond) => {
+                Box::new(NumericHLLAccumulator::<DurationNanosecondType>::new())
+            }
             DataType::Utf8
             | DataType::LargeUtf8
             | DataType::Utf8View
@@ -848,6 +861,7 @@ fn is_hll_groups_type(data_type: &DataType) -> bool {
             | DataType::Decimal64(_, _)
             | DataType::Decimal128(_, _)
             | DataType::Decimal256(_, _)
+            | DataType::Duration(_)
             | DataType::Utf8
             | DataType::LargeUtf8
             | DataType::Utf8View

datafusion/sqllogictest/test_files/aggregate.slt

@@ -2005,6 +2005,33 @@ FROM approx_distinct_decimal_test GROUP BY g ORDER BY g;
 statement ok
 DROP TABLE approx_distinct_decimal_test;
 
+# This test runs approx_distinct over all four Duration units for the scalar and the grouped path.
+statement ok
+CREATE TABLE approx_distinct_duration_test AS VALUES
+  (1, arrow_cast(1, 'Duration(Second)'),  arrow_cast(1, 'Duration(Millisecond)'),  arrow_cast(1, 'Duration(Microsecond)'),  arrow_cast(1, 'Duration(Nanosecond)')),
+  (1, arrow_cast(2, 'Duration(Second)'),  arrow_cast(2, 'Duration(Millisecond)'),  arrow_cast(2, 'Duration(Microsecond)'),  arrow_cast(2, 'Duration(Nanosecond)')),
+  (1, arrow_cast(2, 'Duration(Second)'),  arrow_cast(2, 'Duration(Millisecond)'),  arrow_cast(2, 'Duration(Microsecond)'),  arrow_cast(2, 'Duration(Nanosecond)')),
+  (2, arrow_cast(3, 'Duration(Second)'),  arrow_cast(3, 'Duration(Millisecond)'),  arrow_cast(3, 'Duration(Microsecond)'),  arrow_cast(3, 'Duration(Nanosecond)')),
+  (2, arrow_cast(0, 'Duration(Second)'),  arrow_cast(0, 'Duration(Millisecond)'),  arrow_cast(0, 'Duration(Microsecond)'),  arrow_cast(0, 'Duration(Nanosecond)')),
+  (2, arrow_cast(0, 'Duration(Second)'),  arrow_cast(0, 'Duration(Millisecond)'),  arrow_cast(0, 'Duration(Microsecond)'),  arrow_cast(0, 'Duration(Nanosecond)'));
+
+# Scalar path
+query IIII
+SELECT approx_distinct(column2), approx_distinct(column3), approx_distinct(column4), approx_distinct(column5) FROM approx_distinct_duration_test;
+----
+4 4 4 4
+
+# Grouped path
+query IIIII
+SELECT column1, approx_distinct(column2), approx_distinct(column3), approx_distinct(column4), approx_distinct(column5)
+FROM approx_distinct_duration_test GROUP BY column1 ORDER BY column1;
+----
+1 2 2 2 2
+2 2 2 2 2
+
+statement ok
+DROP TABLE approx_distinct_duration_test;
+
 
 # This test runs approx_distinct over the intervals YearMonth,
 # DayTime, MonthDayNano for the scalar and the grouped path.