askoa commented on code in PR #3622:
URL: https://github.com/apache/arrow-rs/pull/3622#discussion_r1090539035
##########
arrow-select/src/take.rs:
##########
@@ -810,6 +816,70 @@ where
Ok(DictionaryArray::<T>::from(data))
}
+macro_rules! primitive_run_take {
+ ($t:ty, $o:ty, $indices:ident, $value:ident) => {
+ take_primitive_run_values::<$o, $t>(
+ $indices,
+ as_primitive_array::<$t>($value.values()),
+ )
+ };
+}
+
+/// `take` implementation for run arrays
+///
+/// performs binary search on `run_ends` to get physical indices for the given
logical indices.
+/// builds output run array by taking values in the input run array at the
physical indices.
+/// for e.g. an input `RunArray{ run_ends = [2,4,6,8], values=[1,2,1,2] }` and
`indices=[2,7]`
+/// would be converted to `physical_indices=[1,3]` which will be used to build
+/// output `RunArray{ run_ends=[2], values=[2] }`
+
+pub fn take_run<T, I>(
+ run_array: &RunArray<T>,
+ logical_indices: &PrimitiveArray<I>,
+) -> Result<RunArray<T>, ArrowError>
+where
+ T: RunEndIndexType,
+ T::Native: num::Num,
+ I: ArrowPrimitiveType,
+ I::Native: ToPrimitive,
+{
+ match run_array.data_type() {
+ DataType::RunEndEncoded(_, fl) => {
+ let physical_indices =
+ run_array.get_physical_indices(logical_indices.values())?;
+ downcast_primitive! {
+ fl.data_type() => (primitive_run_take, T, physical_indices,
run_array),
+ dt => Err(ArrowError::NotYetImplemented(format!("take_run is
not implemented for {dt:?}")))
+ }
+ }
+ dt => Err(ArrowError::InvalidArgumentError(format!(
+ "Expected DataType::RunEndEncoded found {dt:?}"
+ ))),
+ }
+}
+// Builds a `RunArray` by taking values from given array for the given indices.
+fn take_primitive_run_values<R, V>(
Review Comment:
The approach you are suggesting will lead to inefficient compression in
certain cases. In the below e.g. if we build `new_run_ends` using physical
indices then it'll remain `[1,3]` and the final run array will be `RunArray{
run_ends=[1,2], values=[2,2]}`. The final run array will have two values
instead of one. This issue might have huge effect for filters as they tend to
`take` similar values. If we want to build efficient compression then values
should be compared.
```
/// for e.g. an input `RunArray{ run_ends = [2,4,6,8], values=[1,2,1,2] }`
and `indices=[2,7]`
/// would be converted to `physical_indices=[1,3]` which will be used to
build
/// output `RunArray{ run_ends=[2], values=[2] }`
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
