askoa commented on code in PR #3662:
URL: https://github.com/apache/arrow-rs/pull/3662#discussion_r1101527211
##########
arrow-array/src/array/run_array.rs:
##########
@@ -233,6 +275,63 @@ impl<R: RunEndIndexType> RunArray<R> {
}
Ok(physical_indices)
}
+
+ /// Returns a `RunArray` with zero offset and length matching the last
value
+ /// in run_ends array.
+ pub fn into_non_sliced_array(self) -> Result<Self, ArrowError> {
+ if self.data.offset() == 0 && self.data.len() ==
Self::logical_len(&self.run_ends)
+ {
+ return Ok(self);
+ }
+ // The physical index of original run_ends array from which the
`ArrayData`is sliced.
+ let start_physical_index =
Self::get_physical_index_from_run_ends_array(
+ &self.run_ends,
+ self.data.offset(),
+ )
+ .ok_or_else(|| {
+ ArrowError::InvalidArgumentError(format!(
+ "Cannot convert the offset {} to physical index",
+ self.data.offset()
+ ))
+ })?;
+
+ // The logical length of original run_ends array until which the
`ArrayData` is sliced.
+ let end_logical_index = self.data.offset() + self.data.len() - 1;
+ // The physical index of original run_ends array until which the
`ArrayData`is sliced.
+ let end_physical_index =
+ Self::get_physical_index_from_run_ends_array(&self.run_ends,
end_logical_index).ok_or_else(|| {
+ ArrowError::InvalidArgumentError(format!(
+ "Cannot convert the `offset + len - 1` {end_logical_index}
to physical index"
+ ))
+ })?;
+
+ let physical_length = end_physical_index - start_physical_index + 1;
+
+ // build new run_ends array by subtrating offset from run ends.
+ let new_run_ends: PrimitiveArray<R> = self
+ .run_ends
+ .values()
+ .iter()
+ .skip(start_physical_index)
+ .take(physical_length)
+ .map(|f| f.as_usize() - self.data.offset())
+ .map(|f| f.min(self.len()))
Review Comment:
Yes. I thought the current code was more readable than using a Builder. But
it make sense not to do min operation on all values. I'll use a builder.
--
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]
