use super::{_MutableArrayData, Extend};
use crate::ArrayData;
use arrow_schema::{ArrowError, DataType};
fn nullable_union_field(fields: &arrow_schema::UnionFields) -> Result<(usize, i8), ArrowError> {
fields
.iter()
.enumerate()
.find_map(|(index, (type_id, field))| field.is_nullable().then_some((index, type_id)))
.ok_or_else(|| {
ArrowError::InvalidArgumentError(if fields.is_empty() {
"Cannot extend a union with no fields with nulls".to_owned()
} else {
"Cannot extend a union with no nullable fields with nulls".to_owned()
})
})
}
pub(super) fn build_extend_sparse(array: &ArrayData) -> Extend<'_> {
let type_ids = array.buffer::<i8>(0);
let offset = array.offset();
Box::new(
move |mutable: &mut _MutableArrayData, index: usize, start: usize, len: usize| {
mutable
.buffer1
.extend_from_slice(&type_ids[start..start + len]);
let child_start = start.checked_add(offset).ok_or_else(|| {
ArrowError::InvalidArgumentError(format!(
"sparse union offset {offset} with start {start} overflows usize"
))
})?;
let child_end = child_start.checked_add(len).ok_or_else(|| {
ArrowError::InvalidArgumentError(format!(
"sparse union start {child_start} with length {len} overflows usize"
))
})?;
for child in &mut mutable.child_data {
child.try_extend(index, child_start, child_end)?;
}
Ok(())
},
)
}
pub(super) fn build_extend_dense(array: &ArrayData) -> Extend<'_> {
let type_ids = array.buffer::<i8>(0);
let offsets = array.buffer::<i32>(1);
let arrow_schema::DataType::Union(src_fields, _) = array.data_type() else {
unreachable!();
};
Box::new(
move |mutable: &mut _MutableArrayData, index: usize, start: usize, len: usize| {
mutable
.buffer1
.extend_from_slice(&type_ids[start..start + len]);
for i in start..start + len {
let type_id = type_ids[i];
let child_index = src_fields
.iter()
.position(|(r, _)| r == type_id)
.expect("invalid union type ID");
let src_offset = offsets[i] as usize;
let child_data = &mut mutable.child_data[child_index];
let dst_offset = child_data.len();
mutable.buffer2.push(dst_offset as i32);
mutable.child_data[child_index].try_extend(index, src_offset, src_offset + 1)?;
}
Ok(())
},
)
}
pub(super) fn extend_nulls_dense(
mutable: &mut _MutableArrayData,
len: usize,
) -> Result<(), ArrowError> {
let DataType::Union(fields, _) = &mutable.data_type else {
unreachable!()
};
let (child_index, null_type_id) = nullable_union_field(fields)?;
mutable
.buffer1
.try_extend_from_slice(&vec![null_type_id; len])
.map_err(|e| ArrowError::MemoryError(e.to_string()))?;
let child_offset = mutable.child_data[child_index].len();
let (start, end) = (child_offset as i32, (child_offset + len) as i32);
mutable.buffer2.extend(start..end);
mutable.child_data[child_index].try_extend_nulls(len)?;
Ok(())
}
pub(super) fn extend_nulls_sparse(
mutable: &mut _MutableArrayData,
len: usize,
) -> Result<(), ArrowError> {
let DataType::Union(fields, _) = &mutable.data_type else {
unreachable!()
};
let (_, null_type_id) = nullable_union_field(fields)?;
mutable
.buffer1
.try_extend_from_slice(&vec![null_type_id; len])
.map_err(|e| ArrowError::MemoryError(e.to_string()))?;
for child in &mut mutable.child_data {
child.try_extend_nulls(len)?;
}
Ok(())
}