GroupsAccumulator for Duration (apache#15322)

Author: shruti2522

datafusion/functions-aggregate/src/min_max.rs

@@ -32,9 +32,10 @@ use arrow::array::{
 };
 use arrow::compute;
 use arrow::datatypes::{
-    DataType, Decimal128Type, Decimal256Type, Float16Type, Float32Type, Float64Type,
-    Int16Type, Int32Type, Int64Type, Int8Type, IntervalUnit, UInt16Type, UInt32Type,
-    UInt64Type, UInt8Type,
+    DataType, Decimal128Type, Decimal256Type, DurationMicrosecondType,
+    DurationMillisecondType, DurationNanosecondType, DurationSecondType, Float16Type,
+    Float32Type, Float64Type, Int16Type, Int32Type, Int64Type, Int8Type, IntervalUnit,
+    UInt16Type, UInt32Type, UInt64Type, UInt8Type,
 };
 use datafusion_common::stats::Precision;
 use datafusion_common::{
@@ -263,6 +264,7 @@ impl AggregateUDFImpl for Max {
                 | Binary
                 | LargeBinary
                 | BinaryView
+                | Duration(_)
         )
     }
 
@@ -317,6 +319,18 @@ impl AggregateUDFImpl for Max {
             Timestamp(Nanosecond, _) => {
                 primitive_max_accumulator!(data_type, i64, TimestampNanosecondType)
             }
+            Duration(Second) => {
+                primitive_max_accumulator!(data_type, i64, DurationSecondType)
+            }
+            Duration(Millisecond) => {
+                primitive_max_accumulator!(data_type, i64, DurationMillisecondType)
+            }
+            Duration(Microsecond) => {
+                primitive_max_accumulator!(data_type, i64, DurationMicrosecondType)
+            }
+            Duration(Nanosecond) => {
+                primitive_max_accumulator!(data_type, i64, DurationNanosecondType)
+            }
             Decimal128(_, _) => {
                 primitive_max_accumulator!(data_type, i128, Decimal128Type)
             }
@@ -1090,6 +1104,7 @@ impl AggregateUDFImpl for Min {
                 | Binary
                 | LargeBinary
                 | BinaryView
+                | Duration(_)
         )
     }
 
@@ -1144,6 +1159,18 @@ impl AggregateUDFImpl for Min {
             Timestamp(Nanosecond, _) => {
                 primitive_min_accumulator!(data_type, i64, TimestampNanosecondType)
             }
+            Duration(Second) => {
+                primitive_min_accumulator!(data_type, i64, DurationSecondType)
+            }
+            Duration(Millisecond) => {
+                primitive_min_accumulator!(data_type, i64, DurationMillisecondType)
+            }
+            Duration(Microsecond) => {
+                primitive_min_accumulator!(data_type, i64, DurationMicrosecondType)
+            }
+            Duration(Nanosecond) => {
+                primitive_min_accumulator!(data_type, i64, DurationNanosecondType)
+            }
             Decimal128(_, _) => {
                 primitive_min_accumulator!(data_type, i128, Decimal128Type)
             }

datafusion/sqllogictest/test_files/aggregate.slt

@@ -162,7 +162,7 @@ SELECT approx_percentile_cont(c12, 0.95, c5) FROM aggregate_test_100
 
 # Not supported over sliding windows
 query error This feature is not implemented: Aggregate can not be used as a sliding accumulator because `retract_batch` is not implemented
-SELECT approx_percentile_cont(c3, 0.5) OVER (ROWS BETWEEN 4 PRECEDING AND CURRENT ROW) 
+SELECT approx_percentile_cont(c3, 0.5) OVER (ROWS BETWEEN 4 PRECEDING AND CURRENT ROW)
 FROM aggregate_test_100
 
 # array agg can use order by
@@ -2194,7 +2194,7 @@ drop table t;
 
 # test count with largeutf8
 statement ok
-create table t (c string) as values 
+create table t (c string) as values
   (arrow_cast('a', 'LargeUtf8')),
   (arrow_cast('b', 'LargeUtf8')),
   (arrow_cast(null, 'LargeUtf8')),
@@ -3807,6 +3807,52 @@ SELECT MIN(value), MAX(value) FROM timestampmicrosecond
 statement ok
 DROP TABLE timestampmicrosecond;
 
+# min_duration, max_duration
+statement ok
+create table d
+as values
+  (arrow_cast(1, 'Duration(Second)'), arrow_cast(2, 'Duration(Millisecond)'),  arrow_cast(3, 'Duration(Microsecond)'),  arrow_cast(4, 'Duration(Nanosecond)'), 1),
+  (arrow_cast(11, 'Duration(Second)'),arrow_cast(22, 'Duration(Millisecond)'), arrow_cast(33, 'Duration(Microsecond)'), arrow_cast(44, 'Duration(Nanosecond)'), 1);
+
+query ????
+SELECT min(column1), min(column2), min(column3), min(column4) FROM d;
+----
+0 days 0 hours 0 mins 1 secs 0 days 0 hours 0 mins 0.002 secs 0 days 0 hours 0 mins 0.000003 secs 0 days 0 hours 0 mins 0.000000004 secs
+
+query ????
+SELECT max(column1), max(column2), max(column3), max(column4) FROM d;
+----
+0 days 0 hours 0 mins 11 secs 0 days 0 hours 0 mins 0.022 secs 0 days 0 hours 0 mins 0.000033 secs 0 days 0 hours 0 mins 0.000000044 secs
+
+# GROUP BY follows a different code path
+query ????I
+SELECT min(column1), min(column2), min(column3), min(column4), column5 FROM d GROUP BY column5;
+----
+0 days 0 hours 0 mins 1 secs 0 days 0 hours 0 mins 0.002 secs 0 days 0 hours 0 mins 0.000003 secs 0 days 0 hours 0 mins 0.000000004 secs 1
+
+query ????I
+SELECT max(column1), max(column2), max(column3), max(column4), column5 FROM d GROUP BY column5;
+----
+0 days 0 hours 0 mins 11 secs 0 days 0 hours 0 mins 0.022 secs 0 days 0 hours 0 mins 0.000033 secs 0 days 0 hours 0 mins 0.000000044 secs 1
+
+statement ok
+INSERT INTO d VALUES
+  (arrow_cast(3, 'Duration(Second)'), arrow_cast(1, 'Duration(Millisecond)'), arrow_cast(7, 'Duration(Microsecond)'), arrow_cast(2, 'Duration(Nanosecond)'), 1),
+  (arrow_cast(0, 'Duration(Second)'), arrow_cast(9, 'Duration(Millisecond)'), arrow_cast(5, 'Duration(Microsecond)'), arrow_cast(8, 'Duration(Nanosecond)'), 1);
+
+query ????I
+SELECT max(column1), max(column2), max(column3), max(column4), column5 FROM d GROUP BY column5 ORDER BY column5;
+----
+0 days 0 hours 0 mins 11 secs 0 days 0 hours 0 mins 0.022 secs 0 days 0 hours 0 mins 0.000033 secs 0 days 0 hours 0 mins 0.000000044 secs 1
+
+query ????I
+SELECT min(column1), min(column2), min(column3), min(column4), column5 FROM d GROUP BY column5 ORDER BY column5;
+----
+0 days 0 hours 0 mins 0 secs 0 days 0 hours 0 mins 0.001 secs 0 days 0 hours 0 mins 0.000003 secs 0 days 0 hours 0 mins 0.000000002 secs 1
+
+statement ok
+drop table d;
+
 # max_bool
 statement ok
 CREATE TABLE max_bool (value BOOLEAN);
@@ -4647,7 +4693,7 @@ statement ok
 create table t (c1 decimal(10, 0), c2 int) as values (null, null), (null, null), (null, null);
 
 query RTIT
-select 
+select
   sum(c1), arrow_typeof(sum(c1)),
   sum(c2), arrow_typeof(sum(c2))
 from t;
@@ -5299,7 +5345,7 @@ NULL NULL 3 NULL 1 4 0 8 0
 
 # regr_*() basic tests
 query RRIRRRRRR
-select 
+select
     regr_slope(column2, column1),
     regr_intercept(column2, column1),
     regr_count(column2, column1),
@@ -5314,7 +5360,7 @@ from (values (1,2), (2,4), (3,6));
 2 0 3 1 2 4 2 8 4
 
 query RRIRRRRRR
-select 
+select
     regr_slope(c12, c11),
     regr_intercept(c12, c11),
     regr_count(c12, c11),
@@ -5332,7 +5378,7 @@ from aggregate_test_100;
 
 # regr_*() functions ignore NULLs
 query RRIRRRRRR
-select 
+select
     regr_slope(column2, column1),
     regr_intercept(column2, column1),
     regr_count(column2, column1),
@@ -5347,7 +5393,7 @@ from (values (1,NULL), (2,4), (3,6));
 2 0 2 1 2.5 5 0.5 2 1
 
 query RRIRRRRRR
-select 
+select
     regr_slope(column2, column1),
     regr_intercept(column2, column1),
     regr_count(column2, column1),
@@ -5362,7 +5408,7 @@ from (values (1,NULL), (NULL,4), (3,6));
 NULL NULL 1 NULL 3 6 0 0 0
 
 query RRIRRRRRR
-select 
+select
     regr_slope(column2, column1),
     regr_intercept(column2, column1),
     regr_count(column2, column1),
@@ -5377,8 +5423,8 @@ from (values (1,NULL), (NULL,4), (NULL,NULL));
 NULL NULL 0 NULL NULL NULL NULL NULL NULL
 
 query TRRIRRRRRR rowsort
-select 
-    column3, 
+select
+    column3,
     regr_slope(column2, column1),
     regr_intercept(column2, column1),
     regr_count(column2, column1),
@@ -5402,7 +5448,7 @@ statement ok
 set datafusion.execution.batch_size = 1;
 
 query RRIRRRRRR
-select 
+select
     regr_slope(c12, c11),
     regr_intercept(c12, c11),
     regr_count(c12, c11),
@@ -5420,7 +5466,7 @@ statement ok
 set datafusion.execution.batch_size = 2;
 
 query RRIRRRRRR
-select 
+select
     regr_slope(c12, c11),
     regr_intercept(c12, c11),
     regr_count(c12, c11),
@@ -5438,7 +5484,7 @@ statement ok
 set datafusion.execution.batch_size = 3;
 
 query RRIRRRRRR
-select 
+select
     regr_slope(c12, c11),
     regr_intercept(c12, c11),
     regr_count(c12, c11),
@@ -5592,7 +5638,7 @@ CREATE TABLE float_table (
 
 # Test string_agg with largeutf8
 statement ok
-create table string_agg_large_utf8 (c string) as values 
+create table string_agg_large_utf8 (c string) as values
   (arrow_cast('a', 'LargeUtf8')),
   (arrow_cast('b', 'LargeUtf8')),
   (arrow_cast('c', 'LargeUtf8'))
@@ -5647,7 +5693,7 @@ select count(*) from (select count(*) a, count(*) b from (select 1));
 
 # UTF8 string matters for string to &[u8] conversion, add it to prevent regression
 statement ok
-create table distinct_count_string_table as values 
+create table distinct_count_string_table as values
     (1, 'a', 'longstringtest_a', '台灣'),
     (2, 'b', 'longstringtest_b1', '日本'),
     (2, 'b', 'longstringtest_b2', '中國'),
@@ -6625,7 +6671,7 @@ group1 0.0003
 # median with all nulls
 statement ok
 create table group_median_all_nulls(
-  a STRING NOT NULL, 
+  a STRING NOT NULL,
   b INT
 ) AS VALUES
 ( 'group0', NULL),