Skip to content
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

Jump to bottom

Implement min/max for interval types #11015

Merged
merged 4 commits into from
Jun 23, 2024
Merged

Implement min/max for interval types #11015

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
e2b061f
Implement min/max for interval types
maxburke Jun 19, 2024
a1abd87
Add sqllogictests for min/max intervals
alamb Jun 20, 2024
a879c54
Add tests for interval min/max
maxburke Jun 22, 2024
bb27987
update sql logic tests
maxburke Jun 22, 2024
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
139 changes: 131 additions & 8 deletions datafusion/physical-expr/src/aggregate/min_max.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,18 +24,20 @@ use crate::aggregate::groups_accumulator::prim_op::PrimitiveGroupsAccumulator;
use crate::{AggregateExpr, PhysicalExpr};
use arrow::compute;
use arrow::datatypes::{
DataType, Date32Type, Date64Type, Time32MillisecondType, Time32SecondType,
Time64MicrosecondType, Time64NanosecondType, TimeUnit, TimestampMicrosecondType,
TimestampMillisecondType, TimestampNanosecondType, TimestampSecondType,
DataType, Date32Type, Date64Type, IntervalUnit, Time32MillisecondType,
Time32SecondType, Time64MicrosecondType, Time64NanosecondType, TimeUnit,
TimestampMicrosecondType, TimestampMillisecondType, TimestampNanosecondType,
TimestampSecondType,
};
use arrow::{
array::{
ArrayRef, BinaryArray, BooleanArray, Date32Array, Date64Array, Float32Array,
Float64Array, Int16Array, Int32Array, Int64Array, Int8Array, LargeBinaryArray,
LargeStringArray, StringArray, Time32MillisecondArray, Time32SecondArray,
Time64MicrosecondArray, Time64NanosecondArray, TimestampMicrosecondArray,
TimestampMillisecondArray, TimestampNanosecondArray, TimestampSecondArray,
UInt16Array, UInt32Array, UInt64Array, UInt8Array,
Float64Array, Int16Array, Int32Array, Int64Array, Int8Array,
IntervalDayTimeArray, IntervalMonthDayNanoArray, IntervalYearMonthArray,
LargeBinaryArray, LargeStringArray, StringArray, Time32MillisecondArray,
Time32SecondArray, Time64MicrosecondArray, Time64NanosecondArray,
TimestampMicrosecondArray, TimestampMillisecondArray, TimestampNanosecondArray,
TimestampSecondArray, UInt16Array, UInt32Array, UInt64Array, UInt8Array,
},
datatypes::Field,
};
Expand Down Expand Up @@ -408,6 +410,25 @@ macro_rules! min_max_batch {
$OP
)
}
DataType::Interval(IntervalUnit::YearMonth) => {
typed_min_max_batch!(
$VALUES,
IntervalYearMonthArray,
IntervalYearMonth,
$OP
)
}
DataType::Interval(IntervalUnit::DayTime) => {
typed_min_max_batch!($VALUES, IntervalDayTimeArray, IntervalDayTime, $OP)
}
DataType::Interval(IntervalUnit::MonthDayNano) => {
typed_min_max_batch!(
$VALUES,
IntervalMonthDayNanoArray,
IntervalMonthDayNano,
$OP
)
}
other => {
// This should have been handled before
return internal_err!(
Expand Down Expand Up @@ -1121,6 +1142,108 @@ impl Accumulator for SlidingMinAccumulator {
#[cfg(test)]
mod tests {
use super::*;
use arrow::datatypes::{
IntervalDayTimeType, IntervalMonthDayNanoType, IntervalYearMonthType,
};

#[test]
fn interval_min_max() {
// IntervalYearMonth
let b = IntervalYearMonthArray::from(vec![
IntervalYearMonthType::make_value(0, 1),
IntervalYearMonthType::make_value(5, 34),
IntervalYearMonthType::make_value(-2, 4),
IntervalYearMonthType::make_value(7, -4),
IntervalYearMonthType::make_value(0, 1),
]);
let b: ArrayRef = Arc::new(b);

let mut min =
MinAccumulator::try_new(&DataType::Interval(IntervalUnit::YearMonth))
.unwrap();
min.update_batch(&[b.clone()]).unwrap();
let min_res = min.evaluate().unwrap();
assert_eq!(
min_res,
ScalarValue::IntervalYearMonth(Some(IntervalYearMonthType::make_value(
-2, 4
)))
);

let mut max =
MaxAccumulator::try_new(&DataType::Interval(IntervalUnit::YearMonth))
.unwrap();
max.update_batch(&[b.clone()]).unwrap();
let max_res = max.evaluate().unwrap();
assert_eq!(
max_res,
ScalarValue::IntervalYearMonth(Some(IntervalYearMonthType::make_value(
5, 34
)))
);

// IntervalDayTime
let b = IntervalDayTimeArray::from(vec![
IntervalDayTimeType::make_value(0, 0),
IntervalDayTimeType::make_value(5, 454000),
IntervalDayTimeType::make_value(-34, 0),
IntervalDayTimeType::make_value(7, -4000),
IntervalDayTimeType::make_value(1, 0),
]);
let b: ArrayRef = Arc::new(b);

let mut min =
MinAccumulator::try_new(&DataType::Interval(IntervalUnit::DayTime)).unwrap();
min.update_batch(&[b.clone()]).unwrap();
let min_res = min.evaluate().unwrap();
assert_eq!(
min_res,
ScalarValue::IntervalDayTime(Some(IntervalDayTimeType::make_value(-34, 0)))
);

let mut max =
MaxAccumulator::try_new(&DataType::Interval(IntervalUnit::DayTime)).unwrap();
max.update_batch(&[b.clone()]).unwrap();
let max_res = max.evaluate().unwrap();
assert_eq!(
max_res,
ScalarValue::IntervalDayTime(Some(IntervalDayTimeType::make_value(7, -4000)))
);

// IntervalMonthDayNano
let b = IntervalMonthDayNanoArray::from(vec![
IntervalMonthDayNanoType::make_value(1, 0, 0),
IntervalMonthDayNanoType::make_value(344, 34, -43_000_000_000),
IntervalMonthDayNanoType::make_value(-593, -33, 13_000_000_000),
IntervalMonthDayNanoType::make_value(5, 2, 493_000_000_000),
IntervalMonthDayNanoType::make_value(1, 0, 0),
]);
let b: ArrayRef = Arc::new(b);

let mut min =
MinAccumulator::try_new(&DataType::Interval(IntervalUnit::MonthDayNano))
.unwrap();
min.update_batch(&[b.clone()]).unwrap();
let min_res = min.evaluate().unwrap();
assert_eq!(
min_res,
ScalarValue::IntervalMonthDayNano(Some(
IntervalMonthDayNanoType::make_value(-593, -33, 13_000_000_000)
))
);

let mut max =
MaxAccumulator::try_new(&DataType::Interval(IntervalUnit::MonthDayNano))
.unwrap();
max.update_batch(&[b.clone()]).unwrap();
let max_res = max.evaluate().unwrap();
assert_eq!(
max_res,
ScalarValue::IntervalMonthDayNano(Some(
IntervalMonthDayNanoType::make_value(344, 34, -43_000_000_000)
))
);
}

#[test]
fn float_min_max_with_nans() {
Expand Down
49 changes: 31 additions & 18 deletions datafusion/sqllogictest/test_files/aggregate.slt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1785,29 +1785,42 @@ select min(t), max(t) from (select '00:00:00' as t union select '00:00:01' unio
----
00:00:00 00:00:02

# aggregate_decimal_min
query RT
select min(c1), arrow_typeof(min(c1)) from d_table
# aggregate Interval(MonthDayNano) min/max
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think these tests needs to be updated now that you have implemented the correct logic

You can do so by running

cargo test --test sqllogictests  -- aggregate --complete

And then verifying and checking in the results

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

updated; I think they look right at a glance, but I'm not well versed in this part of the system. can you verify?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

They looked good to me -- thanks!

query error
select
arrow_typeof(min(column1)), min(column1), max(column1)
from values
(interval '1 month'),
(interval '2 months'),
(interval '2 month 15 days'),
(interval '-2 month')
----
-100.009 Decimal128(10, 3)
DataFusion error: Internal error: Min/Max accumulator not implemented for type Interval(MonthDayNano).
This was likely caused by a bug in DataFusion's code and we would welcome that you file an bug report in our issue tracker

# aggregate_decimal_max
query RT
select max(c1), arrow_typeof(max(c1)) from d_table
----
110.009 Decimal128(10, 3)

# aggregate_decimal_sum
query RT
select sum(c1), arrow_typeof(sum(c1)) from d_table
# aggregate Interval(DayTime) min/max
query T??
select
arrow_typeof(min(column1)), min(column1), max(column1)
from values
(arrow_cast('60 minutes', 'Interval(DayTime)')),
(arrow_cast('-3 minutes', 'Interval(DayTime)')),
(arrow_cast('30 minutes', 'Interval(DayTime)'));
----
100 Decimal128(20, 3)
Interval(DayTime) 0 years 0 mons 0 days 0 hours -3 mins 0.000 secs 0 years 0 mons 0 days 1 hours 0 mins 0.000 secs

# aggregate_decimal_avg
query RT
select avg(c1), arrow_typeof(avg(c1)) from d_table
----
5 Decimal128(14, 7)
# aggregate Interval(YearMonth) min/max
query error
select
arrow_typeof(min(column1)), min(column1), max(column1)
from values
(arrow_cast('-1 year', 'Interval(YearMonth)')),
(arrow_cast('13 months', 'Interval(YearMonth)')),
(arrow_cast('1 year', 'Interval(YearMonth)'));
----
DataFusion error: Internal error: Min/Max accumulator not implemented for type Interval(YearMonth).
This was likely caused by a bug in DataFusion's code and we would welcome that you file an bug report in our issue tracker


# aggregate
Expand Down
Loading