This is an automated email from the ASF dual-hosted git repository.
github-bot pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/datafusion.git
The following commit(s) were added to refs/heads/main by this push:
new 473cbdab83 Refactor `to_local_time()` signature away from user_defined
(#18707)
473cbdab83 is described below
commit 473cbdab83e03668ce3b1d20170d1f47631ad08f
Author: Jeffrey Vo <[email protected]>
AuthorDate: Wed Nov 26 18:38:13 2025 +1100
Refactor `to_local_time()` signature away from user_defined (#18707)
## Which issue does this PR close?
<!--
We generally require a GitHub issue to be filed for all bug fixes and
enhancements and this helps us generate change logs for our releases.
You can link an issue to this PR using the GitHub syntax. For example
`Closes #123` indicates that this PR will close issue #123.
-->
Part of #12725
## Rationale for this change
<!--
Why are you proposing this change? If this is already explained clearly
in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand
your changes and offer better suggestions for fixes.
-->
Prefer to avoid user_defined for consistency in function definitions.
## What changes are included in this PR?
<!--
There is no need to duplicate the description in the issue here but it
is sometimes worth providing a summary of the individual changes in this
PR.
-->
Refactor signature of `to_local_time` away from user_defined.
Also some other refactors for `to_local_time`.
## Are these changes tested?
<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code
If tests are not included in your PR, please explain why (for example,
are they covered by existing tests)?
-->
Existing tests.
## Are there any user-facing changes?
<!--
If there are user-facing changes then we may require documentation to be
updated before approving the PR.
-->
No.
<!--
If there are any breaking changes to public APIs, please add the `api
change` label.
-->
---
datafusion/functions/src/datetime/to_local_time.rs | 339 ++++++++++-----------
datafusion/sqllogictest/test_files/timestamps.slt | 49 +--
2 files changed, 182 insertions(+), 206 deletions(-)
diff --git a/datafusion/functions/src/datetime/to_local_time.rs
b/datafusion/functions/src/datetime/to_local_time.rs
index 6e0a150b0a..0495a4841f 100644
--- a/datafusion/functions/src/datetime/to_local_time.rs
+++ b/datafusion/functions/src/datetime/to_local_time.rs
@@ -20,7 +20,7 @@ use std::ops::Add;
use std::sync::Arc;
use arrow::array::timezone::Tz;
-use arrow::array::{Array, ArrayRef, PrimitiveBuilder};
+use arrow::array::{ArrayRef, PrimitiveBuilder};
use arrow::datatypes::DataType::Timestamp;
use arrow::datatypes::TimeUnit::{Microsecond, Millisecond, Nanosecond, Second};
use arrow::datatypes::{
@@ -31,11 +31,12 @@ use chrono::{DateTime, MappedLocalTime, Offset, TimeDelta,
TimeZone, Utc};
use datafusion_common::cast::as_primitive_array;
use datafusion_common::{
- exec_err, internal_datafusion_err, plan_err, utils::take_function_args,
Result,
+ exec_err, internal_datafusion_err, internal_err,
utils::take_function_args, Result,
ScalarValue,
};
use datafusion_expr::{
- ColumnarValue, Documentation, ScalarUDFImpl, Signature, Volatility,
+ Coercion, ColumnarValue, Documentation, ScalarUDFImpl, Signature,
TypeSignatureClass,
+ Volatility,
};
use datafusion_macros::user_doc;
@@ -111,133 +112,163 @@ impl Default for ToLocalTimeFunc {
impl ToLocalTimeFunc {
pub fn new() -> Self {
Self {
- signature: Signature::user_defined(Volatility::Immutable),
+ signature: Signature::coercible(
+ vec![Coercion::new_exact(TypeSignatureClass::Timestamp)],
+ Volatility::Immutable,
+ ),
}
}
+}
- fn to_local_time(&self, args: &[ColumnarValue]) -> Result<ColumnarValue> {
- let [time_value] = take_function_args(self.name(), args)?;
+impl ScalarUDFImpl for ToLocalTimeFunc {
+ fn as_any(&self) -> &dyn Any {
+ self
+ }
- let arg_type = time_value.data_type();
- match arg_type {
- Timestamp(_, None) => {
- // if no timezone specified, just return the input
- Ok(time_value.clone())
- }
- // If has timezone, adjust the underlying time value. The current
time value
- // is stored as i64 in UTC, even though the timezone may not be in
UTC. Therefore,
- // we need to adjust the time value to the local time. See
[`adjust_to_local_time`]
- // for more details.
- //
- // Then remove the timezone in return type, i.e. return None
- Timestamp(_, Some(timezone)) => {
- let tz: Tz = timezone.parse()?;
-
- match time_value {
- ColumnarValue::Scalar(ScalarValue::TimestampNanosecond(
- Some(ts),
- Some(_),
- )) => {
- let adjusted_ts =
-
adjust_to_local_time::<TimestampNanosecondType>(*ts, tz)?;
-
Ok(ColumnarValue::Scalar(ScalarValue::TimestampNanosecond(
- Some(adjusted_ts),
- None,
- )))
- }
- ColumnarValue::Scalar(ScalarValue::TimestampMicrosecond(
- Some(ts),
- Some(_),
- )) => {
- let adjusted_ts =
-
adjust_to_local_time::<TimestampMicrosecondType>(*ts, tz)?;
-
Ok(ColumnarValue::Scalar(ScalarValue::TimestampMicrosecond(
- Some(adjusted_ts),
- None,
- )))
- }
- ColumnarValue::Scalar(ScalarValue::TimestampMillisecond(
- Some(ts),
- Some(_),
- )) => {
- let adjusted_ts =
-
adjust_to_local_time::<TimestampMillisecondType>(*ts, tz)?;
-
Ok(ColumnarValue::Scalar(ScalarValue::TimestampMillisecond(
- Some(adjusted_ts),
- None,
- )))
- }
- ColumnarValue::Scalar(ScalarValue::TimestampSecond(
- Some(ts),
- Some(_),
- )) => {
- let adjusted_ts =
- adjust_to_local_time::<TimestampSecondType>(*ts,
tz)?;
- Ok(ColumnarValue::Scalar(ScalarValue::TimestampSecond(
- Some(adjusted_ts),
- None,
- )))
- }
- ColumnarValue::Array(array) => {
- fn transform_array<T: ArrowTimestampType>(
- array: &ArrayRef,
- tz: Tz,
- ) -> Result<ColumnarValue> {
- let mut builder = PrimitiveBuilder::<T>::new();
-
- let primitive_array =
as_primitive_array::<T>(array)?;
- for ts_opt in primitive_array.iter() {
- match ts_opt {
- None => builder.append_null(),
- Some(ts) => {
- let adjusted_ts: i64 =
- adjust_to_local_time::<T>(ts, tz)?;
- builder.append_value(adjusted_ts)
- }
- }
- }
-
-
Ok(ColumnarValue::Array(Arc::new(builder.finish())))
- }
-
- match array.data_type() {
- Timestamp(_, None) => {
- // if no timezone specified, just return the
input
- Ok(time_value.clone())
- }
- Timestamp(Nanosecond, Some(_)) => {
-
transform_array::<TimestampNanosecondType>(array, tz)
- }
- Timestamp(Microsecond, Some(_)) => {
-
transform_array::<TimestampMicrosecondType>(array, tz)
- }
- Timestamp(Millisecond, Some(_)) => {
-
transform_array::<TimestampMillisecondType>(array, tz)
- }
- Timestamp(Second, Some(_)) => {
- transform_array::<TimestampSecondType>(array,
tz)
- }
- _ => {
- exec_err!("to_local_time function requires
timestamp argument in array, got {:?}", array.data_type())
- }
- }
- }
- _ => {
- exec_err!(
- "to_local_time function requires timestamp argument,
got {:?}",
- time_value.data_type()
- )
- }
- }
- }
- _ => {
- exec_err!(
- "to_local_time function requires timestamp argument, got
{:?}",
- arg_type
- )
+ fn name(&self) -> &str {
+ "to_local_time"
+ }
+
+ fn signature(&self) -> &Signature {
+ &self.signature
+ }
+
+ fn return_type(&self, arg_types: &[DataType]) -> Result<DataType> {
+ match &arg_types[0] {
+ DataType::Null => Ok(Timestamp(Nanosecond, None)),
+ Timestamp(timeunit, _) => Ok(Timestamp(*timeunit, None)),
+ dt => internal_err!(
+ "The to_local_time function can only accept timestamp as the
arg, got {dt}"
+ ),
+ }
+ }
+
+ fn invoke_with_args(
+ &self,
+ args: datafusion_expr::ScalarFunctionArgs,
+ ) -> Result<ColumnarValue> {
+ let [time_value] = take_function_args(self.name(), &args.args)?;
+ to_local_time(time_value)
+ }
+
+ fn documentation(&self) -> Option<&Documentation> {
+ self.doc()
+ }
+}
+
+fn transform_array<T: ArrowTimestampType>(
+ array: &ArrayRef,
+ tz: Tz,
+) -> Result<ColumnarValue> {
+ let primitive_array = as_primitive_array::<T>(array)?;
+ let mut builder =
PrimitiveBuilder::<T>::with_capacity(primitive_array.len());
+ for ts_opt in primitive_array.iter() {
+ match ts_opt {
+ None => builder.append_null(),
+ Some(ts) => {
+ let adjusted_ts: i64 = adjust_to_local_time::<T>(ts, tz)?;
+ builder.append_value(adjusted_ts)
}
}
}
+
+ Ok(ColumnarValue::Array(Arc::new(builder.finish())))
+}
+
+fn to_local_time(time_value: &ColumnarValue) -> Result<ColumnarValue> {
+ let arg_type = time_value.data_type();
+
+ let tz: Tz = match &arg_type {
+ Timestamp(_, Some(timezone)) => timezone.parse()?,
+ Timestamp(_, None) => {
+ // if no timezone specified, just return the input
+ return Ok(time_value.clone());
+ }
+ DataType::Null => {
+ return Ok(ColumnarValue::Scalar(ScalarValue::TimestampNanosecond(
+ None, None,
+ )));
+ }
+ dt => {
+ return internal_err!(
+ "to_local_time function requires timestamp argument, got {dt}"
+ )
+ }
+ };
+
+ // If has timezone, adjust the underlying time value. The current time
value
+ // is stored as i64 in UTC, even though the timezone may not be in UTC.
Therefore,
+ // we need to adjust the time value to the local time. See
[`adjust_to_local_time`]
+ // for more details.
+ //
+ // Then remove the timezone in return type, i.e. return None
+ match time_value {
+ ColumnarValue::Scalar(ScalarValue::TimestampSecond(None, Some(_))) =>
Ok(
+ ColumnarValue::Scalar(ScalarValue::TimestampSecond(None, None)),
+ ),
+ ColumnarValue::Scalar(ScalarValue::TimestampMillisecond(None,
Some(_))) => Ok(
+ ColumnarValue::Scalar(ScalarValue::TimestampMillisecond(None,
None)),
+ ),
+ ColumnarValue::Scalar(ScalarValue::TimestampMicrosecond(None,
Some(_))) => Ok(
+ ColumnarValue::Scalar(ScalarValue::TimestampMicrosecond(None,
None)),
+ ),
+ ColumnarValue::Scalar(ScalarValue::TimestampNanosecond(None, Some(_)))
=> Ok(
+ ColumnarValue::Scalar(ScalarValue::TimestampNanosecond(None,
None)),
+ ),
+ ColumnarValue::Scalar(ScalarValue::TimestampNanosecond(Some(ts),
Some(_))) => {
+ let adjusted_ts =
adjust_to_local_time::<TimestampNanosecondType>(*ts, tz)?;
+ Ok(ColumnarValue::Scalar(ScalarValue::TimestampNanosecond(
+ Some(adjusted_ts),
+ None,
+ )))
+ }
+ ColumnarValue::Scalar(ScalarValue::TimestampMicrosecond(Some(ts),
Some(_))) => {
+ let adjusted_ts =
adjust_to_local_time::<TimestampMicrosecondType>(*ts, tz)?;
+ Ok(ColumnarValue::Scalar(ScalarValue::TimestampMicrosecond(
+ Some(adjusted_ts),
+ None,
+ )))
+ }
+ ColumnarValue::Scalar(ScalarValue::TimestampMillisecond(Some(ts),
Some(_))) => {
+ let adjusted_ts =
adjust_to_local_time::<TimestampMillisecondType>(*ts, tz)?;
+ Ok(ColumnarValue::Scalar(ScalarValue::TimestampMillisecond(
+ Some(adjusted_ts),
+ None,
+ )))
+ }
+ ColumnarValue::Scalar(ScalarValue::TimestampSecond(Some(ts), Some(_)))
=> {
+ let adjusted_ts = adjust_to_local_time::<TimestampSecondType>(*ts,
tz)?;
+ Ok(ColumnarValue::Scalar(ScalarValue::TimestampSecond(
+ Some(adjusted_ts),
+ None,
+ )))
+ }
+ ColumnarValue::Array(array)
+ if matches!(array.data_type(), Timestamp(Nanosecond, Some(_))) =>
+ {
+ transform_array::<TimestampNanosecondType>(array, tz)
+ }
+ ColumnarValue::Array(array)
+ if matches!(array.data_type(), Timestamp(Microsecond, Some(_))) =>
+ {
+ transform_array::<TimestampMicrosecondType>(array, tz)
+ }
+ ColumnarValue::Array(array)
+ if matches!(array.data_type(), Timestamp(Millisecond, Some(_))) =>
+ {
+ transform_array::<TimestampMillisecondType>(array, tz)
+ }
+ ColumnarValue::Array(array)
+ if matches!(array.data_type(), Timestamp(Second, Some(_))) =>
+ {
+ transform_array::<TimestampSecondType>(array, tz)
+ }
+ _ => {
+ internal_err!(
+ "to_local_time function requires timestamp argument, got
{arg_type}"
+ )
+ }
+ }
}
/// This function converts a timestamp with a timezone to a timestamp without
a timezone.
@@ -343,68 +374,6 @@ fn adjust_to_local_time<T: ArrowTimestampType>(ts: i64,
tz: Tz) -> Result<i64> {
}
}
-impl ScalarUDFImpl for ToLocalTimeFunc {
- fn as_any(&self) -> &dyn Any {
- self
- }
-
- fn name(&self) -> &str {
- "to_local_time"
- }
-
- fn signature(&self) -> &Signature {
- &self.signature
- }
-
- fn return_type(&self, arg_types: &[DataType]) -> Result<DataType> {
- let [time_value] = take_function_args(self.name(), arg_types)?;
-
- match time_value {
- Timestamp(timeunit, _) => Ok(Timestamp(*timeunit, None)),
- _ => exec_err!(
- "The to_local_time function can only accept timestamp as the
arg, got {:?}", time_value
- )
- }
- }
-
- fn invoke_with_args(
- &self,
- args: datafusion_expr::ScalarFunctionArgs,
- ) -> Result<ColumnarValue> {
- let [time_value] = take_function_args(self.name(), args.args)?;
-
- self.to_local_time(std::slice::from_ref(&time_value))
- }
-
- fn coerce_types(&self, arg_types: &[DataType]) -> Result<Vec<DataType>> {
- if arg_types.len() != 1 {
- return plan_err!(
- "to_local_time function requires 1 argument, got {:?}",
- arg_types.len()
- );
- }
-
- let first_arg = arg_types[0].clone();
- match &first_arg {
- DataType::Null => Ok(vec![Timestamp(Nanosecond, None)]),
- Timestamp(Nanosecond, timezone) => {
- Ok(vec![Timestamp(Nanosecond, timezone.clone())])
- }
- Timestamp(Microsecond, timezone) => {
- Ok(vec![Timestamp(Microsecond, timezone.clone())])
- }
- Timestamp(Millisecond, timezone) => {
- Ok(vec![Timestamp(Millisecond, timezone.clone())])
- }
- Timestamp(Second, timezone) => Ok(vec![Timestamp(Second,
timezone.clone())]),
- _ => plan_err!("The to_local_time function can only accept
Timestamp as the arg got {first_arg}"),
- }
- }
- fn documentation(&self) -> Option<&Documentation> {
- self.doc()
- }
-}
-
#[cfg(test)]
mod tests {
use std::sync::Arc;
diff --git a/datafusion/sqllogictest/test_files/timestamps.slt
b/datafusion/sqllogictest/test_files/timestamps.slt
index 5c365b056d..3a1a26257b 100644
--- a/datafusion/sqllogictest/test_files/timestamps.slt
+++ b/datafusion/sqllogictest/test_files/timestamps.slt
@@ -2250,7 +2250,7 @@ SET TIME ZONE = '+05:00'
statement ok
CREATE TABLE foo (time TIMESTAMPTZ) AS VALUES
- ('2020-01-01T00:00:00+05:00'),
+ ('2020-01-01T00:00:00+05:00'),
('2020-01-01T01:00:00+05:00'),
('2020-01-01T02:00:00+05:00'),
('2020-01-01T03:00:00+05:00')
@@ -2335,17 +2335,17 @@ NULL 1970-01-01T00:00:00 2031-01-19T23:33:25
1970-01-01T00:00:01 1969-12-31T23:5
# verify timestamp syntax styles are consistent
query BBBBBBBBBBBBB
SELECT to_timestamp(null) is null as c1,
- null::timestamp is null as c2,
- cast(null as timestamp) is null as c3,
- to_timestamp(0) = 0::timestamp as c4,
- to_timestamp(1926632005) = 1926632005::timestamp as c5,
- to_timestamp(1) = 1::timestamp as c6,
- to_timestamp(-1) = -1::timestamp as c7,
+ null::timestamp is null as c2,
+ cast(null as timestamp) is null as c3,
+ to_timestamp(0) = 0::timestamp as c4,
+ to_timestamp(1926632005) = 1926632005::timestamp as c5,
+ to_timestamp(1) = 1::timestamp as c6,
+ to_timestamp(-1) = -1::timestamp as c7,
to_timestamp(0-1) = (0-1)::timestamp as c8,
- to_timestamp(0) = cast(0 as timestamp) as c9,
- to_timestamp(1926632005) = cast(1926632005 as timestamp) as c10,
- to_timestamp(1) = cast(1 as timestamp) as c11,
- to_timestamp(-1) = cast(-1 as timestamp) as c12,
+ to_timestamp(0) = cast(0 as timestamp) as c9,
+ to_timestamp(1926632005) = cast(1926632005 as timestamp) as c10,
+ to_timestamp(1) = cast(1 as timestamp) as c11,
+ to_timestamp(-1) = cast(-1 as timestamp) as c12,
to_timestamp(0-1) = cast(0-1 as timestamp) as c13
----
true true true true true true true true true true true true true
@@ -2358,10 +2358,10 @@ Timestamp(ns) Timestamp(ns) Timestamp(ns)
# verify timestamp output types using timestamp literal syntax
query BBBBBB
-SELECT arrow_typeof(to_timestamp(1)) = arrow_typeof(1::timestamp) as c1,
+SELECT arrow_typeof(to_timestamp(1)) = arrow_typeof(1::timestamp) as c1,
arrow_typeof(to_timestamp(null)) = arrow_typeof(null::timestamp) as c2,
arrow_typeof(to_timestamp('2023-01-10 12:34:56.000')) =
arrow_typeof('2023-01-10 12:34:56.000'::timestamp) as c3,
- arrow_typeof(to_timestamp(1)) = arrow_typeof(cast(1 as timestamp)) as
c4,
+ arrow_typeof(to_timestamp(1)) = arrow_typeof(cast(1 as timestamp)) as
c4,
arrow_typeof(to_timestamp(null)) = arrow_typeof(cast(null as
timestamp)) as c5,
arrow_typeof(to_timestamp('2023-01-10 12:34:56.000')) =
arrow_typeof(cast('2023-01-10 12:34:56.000' as timestamp)) as c6
----
@@ -2605,13 +2605,13 @@ drop table table_a
##########
statement ok
-create table table_a (ts timestamp) as values
- ('2020-09-08T11:42:29Z'::timestamp),
+create table table_a (ts timestamp) as values
+ ('2020-09-08T11:42:29Z'::timestamp),
('2020-09-08T12:42:29Z'::timestamp),
('2020-09-08T13:42:29Z'::timestamp)
statement ok
-create table table_b (ts timestamp) as values
+create table table_b (ts timestamp) as values
('2020-09-08T11:42:29.190Z'::timestamp),
('2020-09-08T13:42:29.190Z'::timestamp),
('2020-09-08T12:42:29.190Z'::timestamp)
@@ -2733,8 +2733,8 @@ statement ok
drop table t1
statement ok
-create table table_a (val int, ts1 timestamp, ts2 timestamp) as values
- (1, '2018-07-01T06:00:00'::timestamp, '2018-07-01T07:00:00'::timestamp),
+create table table_a (val int, ts1 timestamp, ts2 timestamp) as values
+ (1, '2018-07-01T06:00:00'::timestamp, '2018-07-01T07:00:00'::timestamp),
(2, '2018-07-01T07:00:00'::timestamp, '2018-07-01T08:00:00'::timestamp)
query I?
@@ -3038,7 +3038,7 @@ NULL
query T
SELECT to_char(date_column, '%Y-%m-%d')
-FROM (VALUES
+FROM (VALUES
(DATE '2020-09-01'),
(NULL)
) AS t(date_column);
@@ -3048,7 +3048,7 @@ NULL
query T
SELECT to_char(date_column, '%Y-%m-%d')
-FROM (VALUES
+FROM (VALUES
(NULL),
(DATE '2020-09-01')
) AS t(date_column);
@@ -3249,7 +3249,7 @@ statement error
select to_local_time('2024-04-01T00:00:20Z'::timestamp, 'some string');
# invalid argument data type
-statement error The to_local_time function can only accept Timestamp as the
arg got Utf8
+statement error DataFusion error: Error during planning: Internal error:
Expect TypeSignatureClass::Timestamp but received NativeType::String, DataType:
Utf8
select to_local_time('2024-04-01T00:00:20Z');
# invalid timezone
@@ -3277,6 +3277,13 @@ select to_local_time(NULL);
----
NULL
+query PT
+select
+ to_local_time(arrow_cast(null, 'Timestamp(s, "Asia/Tokyo")')),
+ arrow_typeof(to_local_time(arrow_cast(null, 'Timestamp(s, "Asia/Tokyo")')));
+----
+NULL Timestamp(s)
+
query PTPT
select
time,
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
