Skip to main content

vortex_array/arrow/executor/
byte_view.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use std::sync::Arc;
5
6use arrow_array::ArrayRef as ArrowArrayRef;
7use arrow_array::GenericByteViewArray;
8use arrow_array::types::ByteViewType;
9use arrow_buffer::ScalarBuffer;
10use vortex_error::VortexResult;
11
12use crate::ArrayRef;
13use crate::ExecutionCtx;
14use crate::arrays::VarBinViewArray;
15use crate::arrow::null_buffer::to_null_buffer;
16use crate::builtins::ArrayBuiltins;
17use crate::dtype::DType;
18use crate::dtype::Nullability;
19use crate::dtype::arrow::FromArrowType;
20
21/// Convert a canonical VarBinViewArray directly to Arrow.
22pub fn canonical_varbinview_to_arrow<T: ByteViewType>(
23    array: &VarBinViewArray,
24    ctx: &mut ExecutionCtx,
25) -> VortexResult<ArrowArrayRef> {
26    let views =
27        ScalarBuffer::<u128>::from(array.views_handle().as_host().clone().into_arrow_buffer());
28    let buffers: Vec<_> = array
29        .data_buffers()
30        .iter()
31        .map(|buffer| buffer.as_host().clone().into_arrow_buffer())
32        .collect();
33    let nulls = to_null_buffer(
34        array
35            .as_ref()
36            .validity()?
37            .execute_mask(array.as_ref().len(), ctx)?,
38    );
39
40    // SAFETY: our own VarBinView array is considered safe.
41    Ok(Arc::new(unsafe {
42        GenericByteViewArray::<T>::new_unchecked(views, buffers, nulls)
43    }))
44}
45
46pub fn execute_varbinview_to_arrow<T: ByteViewType>(
47    array: &VarBinViewArray,
48    ctx: &mut ExecutionCtx,
49) -> VortexResult<ArrowArrayRef> {
50    let compacted = array.compact_buffers()?;
51    canonical_varbinview_to_arrow::<T>(&compacted, ctx)
52}
53
54pub(super) fn to_arrow_byte_view<T: ByteViewType>(
55    array: ArrayRef,
56    ctx: &mut ExecutionCtx,
57) -> VortexResult<ArrowArrayRef> {
58    // First we cast the array into the desired ByteView type.
59    // We do this in case the vortex array is Utf8, and we want Binary or vice versa. By casting
60    // first, we may push this down through the Vortex array tree. We choose nullable to be most
61    // flexible since there's no prescribed nullability in Arrow types.
62    let array = array.cast(DType::from_arrow((&T::DATA_TYPE, Nullability::Nullable)))?;
63
64    let array = array.execute::<ArrayRef>(ctx)?;
65    let varbinview = array.execute::<VarBinViewArray>(ctx)?;
66    execute_varbinview_to_arrow::<T>(&varbinview, ctx)
67}