use crate::builder::null_buffer_builder::NullBufferBuilder;
use crate::builder::{ArrayBuilder, BufferBuilder};
use crate::{ArrayRef, GenericListArray, OffsetSizeTrait};
use arrow_buffer::Buffer;
use arrow_data::ArrayData;
use arrow_schema::Field;
use std::any::Any;
use std::sync::Arc;
#[derive(Debug)]
pub struct GenericListBuilder<OffsetSize: OffsetSizeTrait, T: ArrayBuilder> {
offsets_builder: BufferBuilder<OffsetSize>,
null_buffer_builder: NullBufferBuilder,
values_builder: T,
}
impl<OffsetSize: OffsetSizeTrait, T: ArrayBuilder> GenericListBuilder<OffsetSize, T> {
pub fn new(values_builder: T) -> Self {
let capacity = values_builder.len();
Self::with_capacity(values_builder, capacity)
}
pub fn with_capacity(values_builder: T, capacity: usize) -> Self {
let mut offsets_builder = BufferBuilder::<OffsetSize>::new(capacity + 1);
offsets_builder.append(OffsetSize::zero());
Self {
offsets_builder,
null_buffer_builder: NullBufferBuilder::new(capacity),
values_builder,
}
}
}
impl<OffsetSize: OffsetSizeTrait, T: ArrayBuilder> ArrayBuilder
for GenericListBuilder<OffsetSize, T>
where
T: 'static,
{
fn as_any(&self) -> &dyn Any {
self
}
fn as_any_mut(&mut self) -> &mut dyn Any {
self
}
fn into_box_any(self: Box<Self>) -> Box<dyn Any> {
self
}
fn len(&self) -> usize {
self.null_buffer_builder.len()
}
fn is_empty(&self) -> bool {
self.null_buffer_builder.is_empty()
}
fn finish(&mut self) -> ArrayRef {
Arc::new(self.finish())
}
fn finish_cloned(&self) -> ArrayRef {
Arc::new(self.finish_cloned())
}
}
impl<OffsetSize: OffsetSizeTrait, T: ArrayBuilder> GenericListBuilder<OffsetSize, T>
where
T: 'static,
{
pub fn values(&mut self) -> &mut T {
&mut self.values_builder
}
pub fn values_ref(&self) -> &T {
&self.values_builder
}
#[inline]
pub fn append(&mut self, is_valid: bool) {
self.offsets_builder
.append(OffsetSize::from_usize(self.values_builder.len()).unwrap());
self.null_buffer_builder.append(is_valid);
}
pub fn finish(&mut self) -> GenericListArray<OffsetSize> {
let len = self.len();
let values_arr = self.values_builder.finish();
let values_data = values_arr.data();
let offset_buffer = self.offsets_builder.finish();
let null_bit_buffer = self.null_buffer_builder.finish();
self.offsets_builder.append(OffsetSize::zero());
let field = Box::new(Field::new(
"item",
values_data.data_type().clone(),
true, ));
let data_type = GenericListArray::<OffsetSize>::DATA_TYPE_CONSTRUCTOR(field);
let array_data_builder = ArrayData::builder(data_type)
.len(len)
.add_buffer(offset_buffer)
.add_child_data(values_data.clone())
.null_bit_buffer(null_bit_buffer);
let array_data = unsafe { array_data_builder.build_unchecked() };
GenericListArray::<OffsetSize>::from(array_data)
}
pub fn finish_cloned(&self) -> GenericListArray<OffsetSize> {
let len = self.len();
let values_arr = self.values_builder.finish_cloned();
let values_data = values_arr.data();
let offset_buffer = Buffer::from_slice_ref(self.offsets_builder.as_slice());
let null_bit_buffer = self
.null_buffer_builder
.as_slice()
.map(Buffer::from_slice_ref);
let field = Box::new(Field::new(
"item",
values_data.data_type().clone(),
true, ));
let data_type = GenericListArray::<OffsetSize>::DATA_TYPE_CONSTRUCTOR(field);
let array_data_builder = ArrayData::builder(data_type)
.len(len)
.add_buffer(offset_buffer)
.add_child_data(values_data.clone())
.null_bit_buffer(null_bit_buffer);
let array_data = unsafe { array_data_builder.build_unchecked() };
GenericListArray::<OffsetSize>::from(array_data)
}
pub fn offsets_slice(&self) -> &[OffsetSize] {
self.offsets_builder.as_slice()
}
}
#[cfg(test)]
mod tests {
use super::*;
use crate::builder::{Int32Builder, ListBuilder};
use crate::{Array, Int32Array};
use arrow_buffer::Buffer;
use arrow_schema::DataType;
fn _test_generic_list_array_builder<O: OffsetSizeTrait>() {
let values_builder = Int32Builder::with_capacity(10);
let mut builder = GenericListBuilder::<O, _>::new(values_builder);
builder.values().append_value(0);
builder.values().append_value(1);
builder.values().append_value(2);
builder.append(true);
builder.values().append_value(3);
builder.values().append_value(4);
builder.values().append_value(5);
builder.append(true);
builder.values().append_value(6);
builder.values().append_value(7);
builder.append(true);
let list_array = builder.finish();
let values = list_array.values().data().buffers()[0].clone();
assert_eq!(Buffer::from_slice_ref([0, 1, 2, 3, 4, 5, 6, 7]), values);
assert_eq!(
Buffer::from_slice_ref([0, 3, 6, 8].map(|n| O::from_usize(n).unwrap())),
list_array.data().buffers()[0].clone()
);
assert_eq!(DataType::Int32, list_array.value_type());
assert_eq!(3, list_array.len());
assert_eq!(0, list_array.null_count());
assert_eq!(O::from_usize(6).unwrap(), list_array.value_offsets()[2]);
assert_eq!(O::from_usize(2).unwrap(), list_array.value_length(2));
for i in 0..3 {
assert!(list_array.is_valid(i));
assert!(!list_array.is_null(i));
}
}
#[test]
fn test_list_array_builder() {
_test_generic_list_array_builder::<i32>()
}
#[test]
fn test_large_list_array_builder() {
_test_generic_list_array_builder::<i64>()
}
fn _test_generic_list_array_builder_nulls<O: OffsetSizeTrait>() {
let values_builder = Int32Builder::with_capacity(10);
let mut builder = GenericListBuilder::<O, _>::new(values_builder);
builder.values().append_value(0);
builder.values().append_value(1);
builder.values().append_value(2);
builder.append(true);
builder.append(false);
builder.values().append_value(3);
builder.values().append_null();
builder.values().append_value(5);
builder.append(true);
builder.values().append_value(6);
builder.values().append_value(7);
builder.append(true);
let list_array = builder.finish();
assert_eq!(DataType::Int32, list_array.value_type());
assert_eq!(4, list_array.len());
assert_eq!(1, list_array.null_count());
assert_eq!(O::from_usize(3).unwrap(), list_array.value_offsets()[2]);
assert_eq!(O::from_usize(3).unwrap(), list_array.value_length(2));
}
#[test]
fn test_list_array_builder_nulls() {
_test_generic_list_array_builder_nulls::<i32>()
}
#[test]
fn test_large_list_array_builder_nulls() {
_test_generic_list_array_builder_nulls::<i64>()
}
#[test]
fn test_list_array_builder_finish() {
let values_builder = Int32Array::builder(5);
let mut builder = ListBuilder::new(values_builder);
builder.values().append_slice(&[1, 2, 3]);
builder.append(true);
builder.values().append_slice(&[4, 5, 6]);
builder.append(true);
let mut arr = builder.finish();
assert_eq!(2, arr.len());
assert!(builder.is_empty());
builder.values().append_slice(&[7, 8, 9]);
builder.append(true);
arr = builder.finish();
assert_eq!(1, arr.len());
assert!(builder.is_empty());
}
#[test]
fn test_list_array_builder_finish_cloned() {
let values_builder = Int32Array::builder(5);
let mut builder = ListBuilder::new(values_builder);
builder.values().append_slice(&[1, 2, 3]);
builder.append(true);
builder.values().append_slice(&[4, 5, 6]);
builder.append(true);
let mut arr = builder.finish_cloned();
assert_eq!(2, arr.len());
assert!(!builder.is_empty());
builder.values().append_slice(&[7, 8, 9]);
builder.append(true);
arr = builder.finish();
assert_eq!(3, arr.len());
assert!(builder.is_empty());
}
#[test]
fn test_list_list_array_builder() {
let primitive_builder = Int32Builder::with_capacity(10);
let values_builder = ListBuilder::new(primitive_builder);
let mut builder = ListBuilder::new(values_builder);
builder.values().values().append_value(1);
builder.values().values().append_value(2);
builder.values().append(true);
builder.values().values().append_value(3);
builder.values().values().append_value(4);
builder.values().append(true);
builder.append(true);
builder.values().values().append_value(5);
builder.values().values().append_value(6);
builder.values().values().append_value(7);
builder.values().append(true);
builder.values().append(false);
builder.values().values().append_value(8);
builder.values().append(true);
builder.append(true);
builder.append(false);
builder.values().values().append_value(9);
builder.values().values().append_value(10);
builder.values().append(true);
builder.append(true);
let list_array = builder.finish();
assert_eq!(4, list_array.len());
assert_eq!(1, list_array.null_count());
assert_eq!(
Buffer::from_slice_ref([0, 2, 5, 5, 6]),
list_array.data().buffers()[0].clone()
);
assert_eq!(6, list_array.values().data().len());
assert_eq!(1, list_array.values().data().null_count());
assert_eq!(
Buffer::from_slice_ref([0, 2, 4, 7, 7, 8, 10]),
list_array.values().data().buffers()[0].clone()
);
assert_eq!(10, list_array.values().data().child_data()[0].len());
assert_eq!(0, list_array.values().data().child_data()[0].null_count());
assert_eq!(
Buffer::from_slice_ref([1, 2, 3, 4, 5, 6, 7, 8, 9, 10]),
list_array.values().data().child_data()[0].buffers()[0].clone()
);
}
}