Skip to content

Commit 64fe229

Browse files
committed
make slice work for nested types
1 parent f1fb2b1 commit 64fe229

File tree

2 files changed

+71
-17
lines changed

2 files changed

+71
-17
lines changed

arrow/src/array/array_struct.rs

Lines changed: 1 addition & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -84,12 +84,7 @@ impl From<ArrayData> for StructArray {
8484
fn from(data: ArrayData) -> Self {
8585
let mut boxed_fields = vec![];
8686
for cd in data.child_data() {
87-
let child_data = if data.offset() != 0 || data.len() != cd.len() {
88-
cd.slice(data.offset(), data.len())
89-
} else {
90-
cd.clone()
91-
};
92-
boxed_fields.push(make_array(child_data));
87+
boxed_fields.push(make_array(cd.clone()));
9388
}
9489
Self { data, boxed_fields }
9590
}

arrow/src/array/data.rs

Lines changed: 70 additions & 11 deletions
Original file line numberDiff line numberDiff line change
@@ -21,14 +21,14 @@
2121
use std::mem;
2222
use std::sync::Arc;
2323

24-
use crate::datatypes::{DataType, IntervalUnit};
24+
use crate::{array::raw_pointer::RawPtrBox, datatypes::{DataType, IntervalUnit}};
2525
use crate::{bitmap::Bitmap, datatypes::ArrowNativeType};
2626
use crate::{
2727
buffer::{Buffer, MutableBuffer},
2828
util::bit_util,
2929
};
3030

31-
use super::equal::equal;
31+
use super::{OffsetSizeTrait, equal::equal};
3232

3333
#[inline]
3434
pub(crate) fn count_nulls(
@@ -383,15 +383,55 @@ impl ArrayData {
383383
pub fn slice(&self, offset: usize, length: usize) -> ArrayData {
384384
assert!((offset + length) <= self.len());
385385

386-
let mut new_data = self.clone();
387-
388-
new_data.len = length;
389-
new_data.offset = offset + self.offset;
390-
391-
new_data.null_count =
392-
count_nulls(new_data.null_buffer(), new_data.offset, new_data.len);
393-
394-
new_data
386+
// If data type is primitive, it's quick to clone array
387+
if self.child_data().is_empty() {
388+
let mut new_data = self.clone();
389+
390+
new_data.len = length;
391+
new_data.offset = offset + self.offset;
392+
393+
new_data.null_count =
394+
count_nulls(new_data.null_buffer(), new_data.offset, new_data.len);
395+
396+
new_data
397+
} else {
398+
// Slice into children
399+
let new_offset = self.offset + offset;
400+
let new_data = ArrayData {
401+
data_type: self.data_type().clone(),
402+
len: length,
403+
null_count: count_nulls(self.null_buffer(), new_offset, length),
404+
offset: new_offset,
405+
buffers: self.buffers.clone(),
406+
child_data: self.child_data().iter().map(|data| {
407+
match self.data_type() {
408+
DataType::List(_) => {
409+
let (start, end) = get_list_child_slice::<i32>(
410+
self.buffers.get(0).unwrap(),
411+
offset,
412+
length
413+
);
414+
data.slice(start, end - start)
415+
}
416+
DataType::LargeList(_) => {
417+
let (start, end) = get_list_child_slice::<i64>(
418+
self.buffers.get(0).unwrap(),
419+
offset,
420+
length
421+
);
422+
data.slice(start, end - start)
423+
}
424+
_ => {
425+
// All other types don't require computing offsets
426+
data.slice(offset, length)
427+
}
428+
}
429+
}).collect(),
430+
null_bitmap: self.null_bitmap().clone(),
431+
};
432+
433+
new_data
434+
}
395435
}
396436

397437
/// Returns the `buffer` as a slice of type `T` starting at self.offset
@@ -465,6 +505,25 @@ impl ArrayData {
465505
}
466506
}
467507

508+
#[inline]
509+
fn get_list_child_slice<OffsetSize: OffsetSizeTrait>(
510+
buffer: &Buffer,
511+
offset: usize,
512+
length: usize
513+
) -> (usize, usize) {
514+
let raw_buffer = buffer.as_ptr();
515+
let value_offsets: &[OffsetSize] = unsafe {
516+
let value_offsets = RawPtrBox::<OffsetSize>::new(raw_buffer);
517+
std::slice::from_raw_parts(
518+
value_offsets.as_ptr().add(offset),
519+
length + 1,
520+
)
521+
};
522+
let start = value_offsets[0];
523+
let end = value_offsets[length - 1];
524+
(start.to_usize().unwrap(), end.to_usize().unwrap())
525+
}
526+
468527
impl PartialEq for ArrayData {
469528
fn eq(&self, other: &Self) -> bool {
470529
equal(self, other)

0 commit comments

Comments
 (0)