1use crate::array::print_long_array;
19use crate::builder::{FixedSizeListBuilder, PrimitiveBuilder};
20use crate::iterator::FixedSizeListIter;
21use crate::{Array, ArrayAccessor, ArrayRef, ArrowPrimitiveType, make_array};
22use arrow_buffer::ArrowNativeType;
23use arrow_buffer::buffer::NullBuffer;
24use arrow_data::{ArrayData, ArrayDataBuilder};
25use arrow_schema::{ArrowError, DataType, FieldRef};
26use std::any::Any;
27use std::sync::Arc;
28
29#[derive(Clone)]
119pub struct FixedSizeListArray {
120 data_type: DataType, values: ArrayRef,
122 nulls: Option<NullBuffer>,
123 value_length: i32,
124 len: usize,
125}
126
127impl FixedSizeListArray {
128 pub fn new(field: FieldRef, size: i32, values: ArrayRef, nulls: Option<NullBuffer>) -> Self {
142 Self::try_new(field, size, values, nulls).unwrap()
143 }
144
145 pub unsafe fn new_unchecked(
153 field: FieldRef,
154 size: i32,
155 values: ArrayRef,
156 nulls: Option<NullBuffer>,
157 len: usize,
158 ) -> Self {
159 if cfg!(feature = "force_validate") {
160 return Self::try_new_with_length(field, size, values, nulls, len).unwrap();
161 }
162 Self {
163 data_type: DataType::FixedSizeList(field, size),
164 values,
165 value_length: size,
166 nulls,
167 len,
168 }
169 }
170
171 pub fn try_new(
188 field: FieldRef,
189 size: i32,
190 values: ArrayRef,
191 nulls: Option<NullBuffer>,
192 ) -> Result<Self, ArrowError> {
193 let s = size.to_usize().ok_or_else(|| {
194 ArrowError::InvalidArgumentError(format!("Size cannot be negative, got {size}"))
195 })?;
196
197 if s == 0 {
198 let len = nulls.as_ref().map(|x| x.len()).unwrap_or_default();
201
202 Self::try_new_with_length(field, size, values, nulls, len)
203 } else {
204 if !values.len().is_multiple_of(s) {
205 return Err(ArrowError::InvalidArgumentError(format!(
206 "Incorrect length of values buffer for FixedSizeListArray, \
207 expected a multiple of {s} got {}",
208 values.len(),
209 )));
210 }
211
212 let len = values.len() / s;
213
214 if let Some(null_buffer) = &nulls
216 && s * null_buffer.len() != values.len()
217 {
218 return Err(ArrowError::InvalidArgumentError(format!(
219 "Incorrect length of values buffer for FixedSizeListArray, \
220 expected {} got {}",
221 s * null_buffer.len(),
222 values.len(),
223 )));
224 }
225
226 Self::try_new_with_length(field, size, values, nulls, len)
227 }
228 }
229
230 pub fn try_new_with_length(
246 field: FieldRef,
247 size: i32,
248 values: ArrayRef,
249 nulls: Option<NullBuffer>,
250 len: usize,
251 ) -> Result<Self, ArrowError> {
252 let s = size.to_usize().ok_or_else(|| {
253 ArrowError::InvalidArgumentError(format!("Size cannot be negative, got {size}"))
254 })?;
255
256 if let Some(null_buffer) = &nulls
257 && null_buffer.len() != len
258 {
259 return Err(ArrowError::InvalidArgumentError(format!(
260 "Invalid null buffer for FixedSizeListArray, expected {len} found {}",
261 null_buffer.len()
262 )));
263 }
264
265 if s == 0 && !values.is_empty() {
266 return Err(ArrowError::InvalidArgumentError(format!(
267 "An degenerate FixedSizeListArray should have no underlying values, found {} values",
268 values.len()
269 )));
270 }
271
272 if values.len() != len * s {
273 return Err(ArrowError::InvalidArgumentError(format!(
274 "Incorrect length of values buffer for FixedSizeListArray, expected {} got {}",
275 len * s,
276 values.len(),
277 )));
278 }
279
280 if field.data_type() != values.data_type() {
281 return Err(ArrowError::InvalidArgumentError(format!(
282 "FixedSizeListArray expected data type {} got {} for {:?}",
283 field.data_type(),
284 values.data_type(),
285 field.name()
286 )));
287 }
288
289 if let Some(a) = values.logical_nulls() {
290 let nulls_valid = field.is_nullable()
291 || nulls
292 .as_ref()
293 .is_some_and(|n| n.expand(size as _).contains(&a))
294 || (nulls.is_none() && a.null_count() == 0);
295
296 if !nulls_valid {
297 return Err(ArrowError::InvalidArgumentError(format!(
298 "Found unmasked nulls for non-nullable FixedSizeListArray field {:?}",
299 field.name()
300 )));
301 }
302 }
303
304 let data_type = DataType::FixedSizeList(field, size);
305 Ok(Self {
306 data_type,
307 values,
308 value_length: size,
309 nulls,
310 len,
311 })
312 }
313
314 pub fn new_null(field: FieldRef, size: i32, len: usize) -> Self {
323 let capacity = size.to_usize().unwrap().checked_mul(len).unwrap();
324 Self {
325 values: make_array(ArrayData::new_null(field.data_type(), capacity)),
326 data_type: DataType::FixedSizeList(field, size),
327 nulls: Some(NullBuffer::new_null(len)),
328 value_length: size,
329 len,
330 }
331 }
332
333 pub fn into_parts(self) -> (FieldRef, i32, ArrayRef, Option<NullBuffer>) {
335 let DataType::FixedSizeList(f, _) = self.data_type else {
336 unreachable!()
337 };
338 (f, self.value_length, self.values, self.nulls)
339 }
340
341 pub fn value_field(&self) -> &FieldRef {
343 match &self.data_type {
344 DataType::FixedSizeList(f, _) => f,
345 _ => unreachable!(),
346 }
347 }
348
349 pub fn values(&self) -> &ArrayRef {
351 &self.values
352 }
353
354 pub fn value_type(&self) -> DataType {
356 self.values.data_type().clone()
357 }
358
359 pub fn value(&self, i: usize) -> ArrayRef {
367 self.values
368 .slice(self.value_offset_at(i), self.value_length() as usize)
369 }
370
371 #[inline]
375 pub fn value_offset(&self, i: usize) -> i32 {
376 self.value_offset_at(i) as i32
377 }
378
379 #[inline]
383 pub const fn value_length(&self) -> i32 {
384 self.value_length
385 }
386
387 #[inline]
388 const fn value_offset_at(&self, i: usize) -> usize {
389 i * self.value_length as usize
390 }
391
392 pub fn slice(&self, offset: usize, len: usize) -> Self {
397 assert!(
398 offset.saturating_add(len) <= self.len,
399 "the length + offset of the sliced FixedSizeListArray cannot exceed the existing length"
400 );
401 let size = self.value_length as usize;
402
403 Self {
404 data_type: self.data_type.clone(),
405 values: self.values.slice(offset * size, len * size),
406 nulls: self.nulls.as_ref().map(|n| n.slice(offset, len)),
407 value_length: self.value_length,
408 len,
409 }
410 }
411
412 pub fn from_iter_primitive<T, P, I>(iter: I, length: i32) -> Self
428 where
429 T: ArrowPrimitiveType,
430 P: IntoIterator<Item = Option<<T as ArrowPrimitiveType>::Native>>,
431 I: IntoIterator<Item = Option<P>>,
432 {
433 let l = length as usize;
434 let iter = iter.into_iter();
435 let size_hint = iter.size_hint().0;
436 let mut builder = FixedSizeListBuilder::with_capacity(
437 PrimitiveBuilder::<T>::with_capacity(size_hint * l),
438 length,
439 size_hint,
440 );
441
442 for i in iter {
443 match i {
444 Some(p) => {
445 for t in p {
446 builder.values().append_option(t);
447 }
448 builder.append(true);
449 }
450 None => {
451 builder.values().append_nulls(l);
452 builder.append(false)
453 }
454 }
455 }
456 builder.finish()
457 }
458
459 pub fn iter(&self) -> FixedSizeListIter<'_> {
461 FixedSizeListIter::new(self)
462 }
463}
464
465impl<'a> IntoIterator for &'a FixedSizeListArray {
466 type Item = Option<ArrayRef>;
467 type IntoIter = FixedSizeListIter<'a>;
468
469 fn into_iter(self) -> Self::IntoIter {
470 FixedSizeListIter::new(self)
471 }
472}
473
474impl From<ArrayData> for FixedSizeListArray {
475 fn from(data: ArrayData) -> Self {
476 let (data_type, len, nulls, offset, _buffers, child_data) = data.into_parts();
477
478 let value_length = match data_type {
479 DataType::FixedSizeList(_, len) => len,
480 data_type => {
481 panic!(
482 "FixedSizeListArray data should contain a FixedSizeList data type, got {data_type}"
483 )
484 }
485 };
486
487 let size = value_length as usize;
488 let values = make_array(child_data[0].slice(offset * size, len * size));
489 Self {
490 data_type,
491 values,
492 nulls,
493 value_length,
494 len,
495 }
496 }
497}
498
499impl From<FixedSizeListArray> for ArrayData {
500 fn from(array: FixedSizeListArray) -> Self {
501 let builder = ArrayDataBuilder::new(array.data_type)
502 .len(array.len)
503 .nulls(array.nulls)
504 .child_data(vec![array.values.to_data()]);
505
506 unsafe { builder.build_unchecked() }
507 }
508}
509
510unsafe impl Array for FixedSizeListArray {
512 fn as_any(&self) -> &dyn Any {
513 self
514 }
515
516 fn to_data(&self) -> ArrayData {
517 self.clone().into()
518 }
519
520 fn into_data(self) -> ArrayData {
521 self.into()
522 }
523
524 fn data_type(&self) -> &DataType {
525 &self.data_type
526 }
527
528 fn slice(&self, offset: usize, length: usize) -> ArrayRef {
529 Arc::new(self.slice(offset, length))
530 }
531
532 fn len(&self) -> usize {
533 self.len
534 }
535
536 fn is_empty(&self) -> bool {
537 self.len == 0
538 }
539
540 fn shrink_to_fit(&mut self) {
541 self.values.shrink_to_fit();
542 if let Some(nulls) = &mut self.nulls {
543 nulls.shrink_to_fit();
544 }
545 }
546
547 fn offset(&self) -> usize {
548 0
549 }
550
551 fn nulls(&self) -> Option<&NullBuffer> {
552 self.nulls.as_ref()
553 }
554
555 fn logical_null_count(&self) -> usize {
556 self.null_count()
558 }
559
560 fn get_buffer_memory_size(&self) -> usize {
561 let mut size = self.values.get_buffer_memory_size();
562 if let Some(n) = self.nulls.as_ref() {
563 size += n.buffer().capacity();
564 }
565 size
566 }
567
568 fn get_array_memory_size(&self) -> usize {
569 let mut size = std::mem::size_of::<Self>() + self.values.get_array_memory_size();
570 if let Some(n) = self.nulls.as_ref() {
571 size += n.buffer().capacity();
572 }
573 size
574 }
575
576 #[cfg(feature = "pool")]
577 fn claim(&self, pool: &dyn arrow_buffer::MemoryPool) {
578 self.values.claim(pool);
579 if let Some(nulls) = &self.nulls {
580 nulls.claim(pool);
581 }
582 }
583}
584
585impl super::ListLikeArray for FixedSizeListArray {
586 fn values(&self) -> &ArrayRef {
587 self.values()
588 }
589
590 fn element_range(&self, index: usize) -> std::ops::Range<usize> {
591 let value_length = self.value_length().as_usize();
592 let offset = index * value_length;
593 offset..(offset + value_length)
594 }
595}
596
597impl ArrayAccessor for FixedSizeListArray {
598 type Item = ArrayRef;
599
600 fn value(&self, index: usize) -> Self::Item {
601 FixedSizeListArray::value(self, index)
602 }
603
604 unsafe fn value_unchecked(&self, index: usize) -> Self::Item {
605 FixedSizeListArray::value(self, index)
606 }
607}
608
609impl std::fmt::Debug for FixedSizeListArray {
610 fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
611 write!(f, "FixedSizeListArray<{}>\n[\n", self.value_length())?;
612 print_long_array(self, f, &mut |index, f| {
613 std::fmt::Debug::fmt(&self.value(index), f)
614 })?;
615 write!(f, "]")
616 }
617}
618
619impl ArrayAccessor for &FixedSizeListArray {
620 type Item = ArrayRef;
621
622 fn value(&self, index: usize) -> Self::Item {
623 FixedSizeListArray::value(self, index)
624 }
625
626 unsafe fn value_unchecked(&self, index: usize) -> Self::Item {
627 FixedSizeListArray::value(self, index)
628 }
629}
630
631#[cfg(test)]
632mod tests {
633 use arrow_buffer::{BooleanBuffer, Buffer, bit_util};
634 use arrow_schema::Field;
635
636 use crate::cast::AsArray;
637 use crate::types::Int32Type;
638 use crate::{Int32Array, new_empty_array};
639
640 use super::*;
641
642 #[test]
643 fn test_fixed_size_list_array() {
644 let value_data = ArrayData::builder(DataType::Int32)
646 .len(9)
647 .add_buffer(Buffer::from_slice_ref([0, 1, 2, 3, 4, 5, 6, 7, 8]))
648 .build()
649 .unwrap();
650
651 let list_data_type =
653 DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int32, false)), 3);
654 let list_data = ArrayData::builder(list_data_type.clone())
655 .len(3)
656 .add_child_data(value_data.clone())
657 .build()
658 .unwrap();
659 let list_array = FixedSizeListArray::from(list_data);
660
661 assert_eq!(value_data, list_array.values().to_data());
662 assert_eq!(DataType::Int32, list_array.value_type());
663 assert_eq!(3, list_array.len());
664 assert_eq!(0, list_array.null_count());
665 assert_eq!(6, list_array.value_offset(2));
666 assert_eq!(3, list_array.value_length());
667 assert_eq!(0, list_array.value(0).as_primitive::<Int32Type>().value(0));
668 for i in 0..3 {
669 assert!(list_array.is_valid(i));
670 assert!(!list_array.is_null(i));
671 }
672
673 let list_data = ArrayData::builder(list_data_type)
675 .len(2)
676 .offset(1)
677 .add_child_data(value_data.clone())
678 .build()
679 .unwrap();
680 let list_array = FixedSizeListArray::from(list_data);
681
682 assert_eq!(value_data.slice(3, 6), list_array.values().to_data());
683 assert_eq!(DataType::Int32, list_array.value_type());
684 assert_eq!(2, list_array.len());
685 assert_eq!(0, list_array.null_count());
686 assert_eq!(3, list_array.value(0).as_primitive::<Int32Type>().value(0));
687 assert_eq!(3, list_array.value_offset(1));
688 assert_eq!(3, list_array.value_length());
689 }
690
691 #[test]
692 #[should_panic(expected = "assertion failed: end <= self.len()")]
693 #[cfg(not(feature = "force_validate"))]
696 fn test_fixed_size_list_array_unequal_children() {
697 let value_data = ArrayData::builder(DataType::Int32)
699 .len(8)
700 .add_buffer(Buffer::from_slice_ref([0, 1, 2, 3, 4, 5, 6, 7]))
701 .build()
702 .unwrap();
703
704 let list_data_type =
706 DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int32, false)), 3);
707 let list_data = unsafe {
708 ArrayData::builder(list_data_type)
709 .len(3)
710 .add_child_data(value_data)
711 .build_unchecked()
712 };
713 drop(FixedSizeListArray::from(list_data));
714 }
715
716 #[test]
717 fn test_fixed_size_list_array_slice() {
718 let value_data = ArrayData::builder(DataType::Int32)
720 .len(10)
721 .add_buffer(Buffer::from_slice_ref([0, 1, 2, 3, 4, 5, 6, 7, 8, 9]))
722 .build()
723 .unwrap();
724
725 let mut null_bits: [u8; 1] = [0; 1];
729 bit_util::set_bit(&mut null_bits, 0);
730 bit_util::set_bit(&mut null_bits, 3);
731 bit_util::set_bit(&mut null_bits, 4);
732
733 let list_data_type =
735 DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int32, false)), 2);
736 let list_data = ArrayData::builder(list_data_type)
737 .len(5)
738 .add_child_data(value_data.clone())
739 .null_bit_buffer(Some(Buffer::from(null_bits)))
740 .build()
741 .unwrap();
742 let list_array = FixedSizeListArray::from(list_data);
743
744 assert_eq!(value_data, list_array.values().to_data());
745 assert_eq!(DataType::Int32, list_array.value_type());
746 assert_eq!(5, list_array.len());
747 assert_eq!(2, list_array.null_count());
748 assert_eq!(6, list_array.value_offset(3));
749 assert_eq!(2, list_array.value_length());
750
751 let sliced_array = list_array.slice(1, 4);
752 assert_eq!(4, sliced_array.len());
753 assert_eq!(2, sliced_array.null_count());
754
755 for i in 0..sliced_array.len() {
756 if bit_util::get_bit(&null_bits, 1 + i) {
757 assert!(sliced_array.is_valid(i));
758 } else {
759 assert!(sliced_array.is_null(i));
760 }
761 }
762
763 let sliced_list_array = sliced_array
765 .as_any()
766 .downcast_ref::<FixedSizeListArray>()
767 .unwrap();
768 assert_eq!(2, sliced_list_array.value_length());
769 assert_eq!(4, sliced_list_array.value_offset(2));
770 assert_eq!(6, sliced_list_array.value_offset(3));
771 }
772
773 #[test]
774 #[should_panic(expected = "the offset of the new Buffer cannot exceed the existing length")]
775 fn test_fixed_size_list_array_index_out_of_bound() {
776 let value_data = ArrayData::builder(DataType::Int32)
778 .len(10)
779 .add_buffer(Buffer::from_slice_ref([0, 1, 2, 3, 4, 5, 6, 7, 8, 9]))
780 .build()
781 .unwrap();
782
783 let mut null_bits: [u8; 1] = [0; 1];
787 bit_util::set_bit(&mut null_bits, 0);
788 bit_util::set_bit(&mut null_bits, 3);
789 bit_util::set_bit(&mut null_bits, 4);
790
791 let list_data_type =
793 DataType::FixedSizeList(Arc::new(Field::new_list_field(DataType::Int32, false)), 2);
794 let list_data = ArrayData::builder(list_data_type)
795 .len(5)
796 .add_child_data(value_data)
797 .null_bit_buffer(Some(Buffer::from(null_bits)))
798 .build()
799 .unwrap();
800 let list_array = FixedSizeListArray::from(list_data);
801
802 list_array.value(10);
803 }
804
805 #[test]
806 fn test_fixed_size_list_constructors() {
807 let values = Arc::new(Int32Array::from_iter([
808 Some(1),
809 Some(2),
810 None,
811 None,
812 Some(3),
813 Some(4),
814 ]));
815
816 let field = Arc::new(Field::new_list_field(DataType::Int32, true));
817 let list = FixedSizeListArray::new(field.clone(), 2, values.clone(), None);
818 assert_eq!(list.len(), 3);
819
820 let nulls = NullBuffer::new_null(3);
821 let list = FixedSizeListArray::new(field.clone(), 2, values.clone(), Some(nulls));
822 assert_eq!(list.len(), 3);
823
824 let list = FixedSizeListArray::new(field.clone(), 3, values.clone(), None);
825 assert_eq!(list.len(), 2);
826
827 let err = FixedSizeListArray::try_new(field.clone(), 4, values.clone(), None).unwrap_err();
828 assert_eq!(
829 err.to_string(),
830 "Invalid argument error: Incorrect length of values buffer for FixedSizeListArray, \
831 expected a multiple of 4 got 6",
832 );
833
834 let err =
835 FixedSizeListArray::try_new_with_length(field.clone(), 4, values.clone(), None, 1)
836 .unwrap_err();
837 assert_eq!(
838 err.to_string(),
839 "Invalid argument error: Incorrect length of values buffer for FixedSizeListArray, expected 4 got 6"
840 );
841
842 let err = FixedSizeListArray::try_new(field.clone(), -1, values.clone(), None).unwrap_err();
843 assert_eq!(
844 err.to_string(),
845 "Invalid argument error: Size cannot be negative, got -1"
846 );
847
848 let nulls = NullBuffer::new_null(2);
849 let err = FixedSizeListArray::try_new(field, 2, values.clone(), Some(nulls)).unwrap_err();
850 assert_eq!(
851 err.to_string(),
852 "Invalid argument error: Incorrect length of values buffer for FixedSizeListArray, expected 4 got 6"
853 );
854
855 let field = Arc::new(Field::new_list_field(DataType::Int32, false));
856 let err = FixedSizeListArray::try_new(field.clone(), 2, values.clone(), None).unwrap_err();
857 assert_eq!(
858 err.to_string(),
859 "Invalid argument error: Found unmasked nulls for non-nullable FixedSizeListArray field \"item\""
860 );
861
862 let nulls = NullBuffer::new(BooleanBuffer::new(Buffer::from([0b0000101]), 0, 3));
864 FixedSizeListArray::new(field, 2, values.clone(), Some(nulls));
865
866 let field = Arc::new(Field::new_list_field(DataType::Int64, true));
867 let err = FixedSizeListArray::try_new(field, 2, values, None).unwrap_err();
868 assert_eq!(
869 err.to_string(),
870 "Invalid argument error: FixedSizeListArray expected data type Int64 got Int32 for \"item\""
871 );
872 }
873
874 #[test]
875 fn degenerate_fixed_size_list() {
876 let field = Arc::new(Field::new_list_field(DataType::Int32, true));
877 let nulls = NullBuffer::new_null(2);
878 let values = new_empty_array(&DataType::Int32);
879 let list = FixedSizeListArray::new(field.clone(), 0, values.clone(), Some(nulls.clone()));
880 assert_eq!(list.len(), 2);
881
882 let err = FixedSizeListArray::try_new_with_length(
884 field.clone(),
885 0,
886 values.clone(),
887 Some(nulls),
888 5,
889 )
890 .unwrap_err();
891 assert_eq!(
892 err.to_string(),
893 "Invalid argument error: Invalid null buffer for FixedSizeListArray, expected 5 found 2"
894 );
895
896 let non_empty_values = Arc::new(Int32Array::from(vec![1, 2, 3]));
898 let err =
899 FixedSizeListArray::try_new_with_length(field.clone(), 0, non_empty_values, None, 3)
900 .unwrap_err();
901 assert_eq!(
902 err.to_string(),
903 "Invalid argument error: An degenerate FixedSizeListArray should have no underlying values, found 3 values"
904 );
905 }
906
907 #[test]
908 fn test_fixed_size_list_new_null_len() {
909 let field = Arc::new(Field::new_list_field(DataType::Int32, true));
910 let array = FixedSizeListArray::new_null(field, 2, 5);
911 assert_eq!(array.len(), 5);
912 }
913}