1use crate::array::print_long_array;
19use crate::builder::BooleanBuilder;
20use crate::iterator::BooleanIter;
21use crate::{Array, ArrayAccessor, ArrayRef, Scalar};
22use arrow_buffer::{BooleanBuffer, Buffer, MutableBuffer, NullBuffer, bit_util};
23use arrow_data::{ArrayData, ArrayDataBuilder};
24use arrow_schema::DataType;
25use std::any::Any;
26use std::sync::Arc;
27
28#[derive(Clone)]
68pub struct BooleanArray {
69 values: BooleanBuffer,
70 nulls: Option<NullBuffer>,
71}
72
73impl std::fmt::Debug for BooleanArray {
74 fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
75 write!(f, "BooleanArray\n[\n")?;
76 print_long_array(self, f, |array, index, f| {
77 std::fmt::Debug::fmt(&array.value(index), f)
78 })?;
79 write!(f, "]")
80 }
81}
82
83impl BooleanArray {
84 pub fn new(values: BooleanBuffer, nulls: Option<NullBuffer>) -> Self {
90 if let Some(n) = nulls.as_ref() {
91 assert_eq!(values.len(), n.len());
92 }
93 Self { values, nulls }
94 }
95
96 pub fn new_null(len: usize) -> Self {
98 Self {
99 values: BooleanBuffer::new_unset(len),
100 nulls: Some(NullBuffer::new_null(len)),
101 }
102 }
103
104 pub fn new_scalar(value: bool) -> Scalar<Self> {
106 let values = match value {
107 true => BooleanBuffer::new_set(1),
108 false => BooleanBuffer::new_unset(1),
109 };
110 Scalar::new(Self::new(values, None))
111 }
112
113 pub fn new_from_packed(buffer: impl Into<Buffer>, offset: usize, len: usize) -> Self {
119 BooleanBuffer::new(buffer.into(), offset, len).into()
120 }
121
122 pub fn new_from_u8(value: &[u8]) -> Self {
128 BooleanBuffer::new(Buffer::from(value), 0, value.len() * 8).into()
129 }
130
131 pub fn len(&self) -> usize {
133 self.values.len()
134 }
135
136 pub fn is_empty(&self) -> bool {
138 self.values.is_empty()
139 }
140
141 pub fn slice(&self, offset: usize, length: usize) -> Self {
143 Self {
144 values: self.values.slice(offset, length),
145 nulls: self.nulls.as_ref().map(|n| n.slice(offset, length)),
146 }
147 }
148
149 pub fn builder(capacity: usize) -> BooleanBuilder {
151 BooleanBuilder::with_capacity(capacity)
152 }
153
154 pub fn values(&self) -> &BooleanBuffer {
156 &self.values
157 }
158
159 pub fn true_count(&self) -> usize {
161 match self.nulls() {
162 Some(nulls) => {
163 let null_chunks = nulls.inner().bit_chunks().iter_padded();
164 let value_chunks = self.values().bit_chunks().iter_padded();
165 null_chunks
166 .zip(value_chunks)
167 .map(|(a, b)| (a & b).count_ones() as usize)
168 .sum()
169 }
170 None => self.values().count_set_bits(),
171 }
172 }
173
174 pub fn false_count(&self) -> usize {
176 self.len() - self.null_count() - self.true_count()
177 }
178
179 pub unsafe fn value_unchecked(&self, i: usize) -> bool {
187 unsafe { self.values.value_unchecked(i) }
188 }
189
190 pub fn value(&self, i: usize) -> bool {
198 assert!(
199 i < self.len(),
200 "Trying to access an element at index {} from a BooleanArray of length {}",
201 i,
202 self.len()
203 );
204 unsafe { self.value_unchecked(i) }
207 }
208
209 pub fn take_iter<'a>(
211 &'a self,
212 indexes: impl Iterator<Item = Option<usize>> + 'a,
213 ) -> impl Iterator<Item = Option<bool>> + 'a {
214 indexes.map(|opt_index| opt_index.map(|index| self.value(index)))
215 }
216
217 pub unsafe fn take_iter_unchecked<'a>(
222 &'a self,
223 indexes: impl Iterator<Item = Option<usize>> + 'a,
224 ) -> impl Iterator<Item = Option<bool>> + 'a {
225 indexes.map(|opt_index| opt_index.map(|index| unsafe { self.value_unchecked(index) }))
226 }
227
228 pub fn from_unary<T: ArrayAccessor, F>(left: T, mut op: F) -> Self
239 where
240 F: FnMut(T::Item) -> bool,
241 {
242 let nulls = left.logical_nulls();
243 let values = BooleanBuffer::collect_bool(left.len(), |i| unsafe {
244 op(left.value_unchecked(i))
246 });
247 Self::new(values, nulls)
248 }
249
250 pub fn from_binary<T: ArrayAccessor, S: ArrayAccessor, F>(left: T, right: S, mut op: F) -> Self
267 where
268 F: FnMut(T::Item, S::Item) -> bool,
269 {
270 assert_eq!(left.len(), right.len());
271
272 let nulls = NullBuffer::union(
273 left.logical_nulls().as_ref(),
274 right.logical_nulls().as_ref(),
275 );
276 let values = BooleanBuffer::collect_bool(left.len(), |i| unsafe {
277 op(left.value_unchecked(i), right.value_unchecked(i))
279 });
280 Self::new(values, nulls)
281 }
282
283 pub fn into_parts(self) -> (BooleanBuffer, Option<NullBuffer>) {
285 (self.values, self.nulls)
286 }
287}
288
289impl super::private::Sealed for BooleanArray {}
290
291impl Array for BooleanArray {
292 fn as_any(&self) -> &dyn Any {
293 self
294 }
295
296 fn to_data(&self) -> ArrayData {
297 self.clone().into()
298 }
299
300 fn into_data(self) -> ArrayData {
301 self.into()
302 }
303
304 fn data_type(&self) -> &DataType {
305 &DataType::Boolean
306 }
307
308 fn slice(&self, offset: usize, length: usize) -> ArrayRef {
309 Arc::new(self.slice(offset, length))
310 }
311
312 fn len(&self) -> usize {
313 self.values.len()
314 }
315
316 fn is_empty(&self) -> bool {
317 self.values.is_empty()
318 }
319
320 fn shrink_to_fit(&mut self) {
321 self.values.shrink_to_fit();
322 if let Some(nulls) = &mut self.nulls {
323 nulls.shrink_to_fit();
324 }
325 }
326
327 fn offset(&self) -> usize {
328 self.values.offset()
329 }
330
331 fn nulls(&self) -> Option<&NullBuffer> {
332 self.nulls.as_ref()
333 }
334
335 fn logical_null_count(&self) -> usize {
336 self.null_count()
337 }
338
339 fn get_buffer_memory_size(&self) -> usize {
340 let mut sum = self.values.inner().capacity();
341 if let Some(x) = &self.nulls {
342 sum += x.buffer().capacity()
343 }
344 sum
345 }
346
347 fn get_array_memory_size(&self) -> usize {
348 std::mem::size_of::<Self>() + self.get_buffer_memory_size()
349 }
350}
351
352impl ArrayAccessor for &BooleanArray {
353 type Item = bool;
354
355 fn value(&self, index: usize) -> Self::Item {
356 BooleanArray::value(self, index)
357 }
358
359 unsafe fn value_unchecked(&self, index: usize) -> Self::Item {
360 unsafe { BooleanArray::value_unchecked(self, index) }
361 }
362}
363
364impl From<Vec<bool>> for BooleanArray {
365 fn from(data: Vec<bool>) -> Self {
366 let mut mut_buf = MutableBuffer::new_null(data.len());
367 {
368 let mut_slice = mut_buf.as_slice_mut();
369 for (i, b) in data.iter().enumerate() {
370 if *b {
371 bit_util::set_bit(mut_slice, i);
372 }
373 }
374 }
375 let array_data = ArrayData::builder(DataType::Boolean)
376 .len(data.len())
377 .add_buffer(mut_buf.into());
378
379 let array_data = unsafe { array_data.build_unchecked() };
380 BooleanArray::from(array_data)
381 }
382}
383
384impl From<Vec<Option<bool>>> for BooleanArray {
385 fn from(data: Vec<Option<bool>>) -> Self {
386 data.iter().collect()
387 }
388}
389
390impl From<ArrayData> for BooleanArray {
391 fn from(data: ArrayData) -> Self {
392 assert_eq!(
393 data.data_type(),
394 &DataType::Boolean,
395 "BooleanArray expected ArrayData with type {} got {}",
396 DataType::Boolean,
397 data.data_type()
398 );
399 assert_eq!(
400 data.buffers().len(),
401 1,
402 "BooleanArray data should contain a single buffer only (values buffer)"
403 );
404 let values = BooleanBuffer::new(data.buffers()[0].clone(), data.offset(), data.len());
405
406 Self {
407 values,
408 nulls: data.nulls().cloned(),
409 }
410 }
411}
412
413impl From<BooleanArray> for ArrayData {
414 fn from(array: BooleanArray) -> Self {
415 let builder = ArrayDataBuilder::new(DataType::Boolean)
416 .len(array.values.len())
417 .offset(array.values.offset())
418 .nulls(array.nulls)
419 .buffers(vec![array.values.into_inner()]);
420
421 unsafe { builder.build_unchecked() }
422 }
423}
424
425impl<'a> IntoIterator for &'a BooleanArray {
426 type Item = Option<bool>;
427 type IntoIter = BooleanIter<'a>;
428
429 fn into_iter(self) -> Self::IntoIter {
430 BooleanIter::<'a>::new(self)
431 }
432}
433
434impl<'a> BooleanArray {
435 pub fn iter(&'a self) -> BooleanIter<'a> {
437 BooleanIter::<'a>::new(self)
438 }
439}
440
441#[derive(Debug)]
450struct BooleanAdapter {
451 pub native: Option<bool>,
453}
454
455impl From<bool> for BooleanAdapter {
456 fn from(value: bool) -> Self {
457 BooleanAdapter {
458 native: Some(value),
459 }
460 }
461}
462
463impl From<&bool> for BooleanAdapter {
464 fn from(value: &bool) -> Self {
465 BooleanAdapter {
466 native: Some(*value),
467 }
468 }
469}
470
471impl From<Option<bool>> for BooleanAdapter {
472 fn from(value: Option<bool>) -> Self {
473 BooleanAdapter { native: value }
474 }
475}
476
477impl From<&Option<bool>> for BooleanAdapter {
478 fn from(value: &Option<bool>) -> Self {
479 BooleanAdapter { native: *value }
480 }
481}
482
483impl<Ptr: Into<BooleanAdapter>> FromIterator<Ptr> for BooleanArray {
484 fn from_iter<I: IntoIterator<Item = Ptr>>(iter: I) -> Self {
485 let iter = iter.into_iter();
486 let capacity = match iter.size_hint() {
487 (lower, Some(upper)) if lower == upper => lower,
488 _ => 0,
489 };
490 let mut builder = BooleanBuilder::with_capacity(capacity);
491 builder.extend(iter.map(|item| item.into().native));
492 builder.finish()
493 }
494}
495
496impl BooleanArray {
497 #[inline]
509 #[allow(
510 private_bounds,
511 reason = "We will expose BooleanAdapter if there is a need"
512 )]
513 pub unsafe fn from_trusted_len_iter<I, P>(iter: I) -> Self
514 where
515 P: Into<BooleanAdapter>,
516 I: ExactSizeIterator<Item = P>,
517 {
518 let data_len = iter.len();
519
520 let num_bytes = bit_util::ceil(data_len, 8);
521 let mut null_builder = MutableBuffer::from_len_zeroed(num_bytes);
522 let mut val_builder = MutableBuffer::from_len_zeroed(num_bytes);
523
524 let data = val_builder.as_slice_mut();
525
526 let null_slice = null_builder.as_slice_mut();
527 iter.enumerate().for_each(|(i, item)| {
528 if let Some(a) = item.into().native {
529 unsafe {
530 bit_util::set_bit_raw(null_slice.as_mut_ptr(), i);
533 if a {
534 bit_util::set_bit_raw(data.as_mut_ptr(), i);
535 }
536 }
537 }
538 });
539
540 let data = unsafe {
541 ArrayData::new_unchecked(
542 DataType::Boolean,
543 data_len,
544 None,
545 Some(null_builder.into()),
546 0,
547 vec![val_builder.into()],
548 vec![],
549 )
550 };
551 BooleanArray::from(data)
552 }
553}
554
555impl From<BooleanBuffer> for BooleanArray {
556 fn from(values: BooleanBuffer) -> Self {
557 Self {
558 values,
559 nulls: None,
560 }
561 }
562}
563
564#[cfg(test)]
565mod tests {
566 use super::*;
567 use arrow_buffer::Buffer;
568 use rand::{Rng, rng};
569
570 #[test]
571 fn test_boolean_fmt_debug() {
572 let arr = BooleanArray::from(vec![true, false, false]);
573 assert_eq!(
574 "BooleanArray\n[\n true,\n false,\n false,\n]",
575 format!("{arr:?}")
576 );
577 }
578
579 #[test]
580 fn test_boolean_with_null_fmt_debug() {
581 let mut builder = BooleanArray::builder(3);
582 builder.append_value(true);
583 builder.append_null();
584 builder.append_value(false);
585 let arr = builder.finish();
586 assert_eq!(
587 "BooleanArray\n[\n true,\n null,\n false,\n]",
588 format!("{arr:?}")
589 );
590 }
591
592 #[test]
593 fn test_boolean_array_from_vec() {
594 let buf = Buffer::from([10_u8]);
595 let arr = BooleanArray::from(vec![false, true, false, true]);
596 assert_eq!(&buf, arr.values().inner());
597 assert_eq!(4, arr.len());
598 assert_eq!(0, arr.offset());
599 assert_eq!(0, arr.null_count());
600 for i in 0..4 {
601 assert!(!arr.is_null(i));
602 assert!(arr.is_valid(i));
603 assert_eq!(i == 1 || i == 3, arr.value(i), "failed at {i}")
604 }
605 }
606
607 #[test]
608 fn test_boolean_array_from_vec_option() {
609 let buf = Buffer::from([10_u8]);
610 let arr = BooleanArray::from(vec![Some(false), Some(true), None, Some(true)]);
611 assert_eq!(&buf, arr.values().inner());
612 assert_eq!(4, arr.len());
613 assert_eq!(0, arr.offset());
614 assert_eq!(1, arr.null_count());
615 for i in 0..4 {
616 if i == 2 {
617 assert!(arr.is_null(i));
618 assert!(!arr.is_valid(i));
619 } else {
620 assert!(!arr.is_null(i));
621 assert!(arr.is_valid(i));
622 assert_eq!(i == 1 || i == 3, arr.value(i), "failed at {i}")
623 }
624 }
625 }
626
627 #[test]
628 fn test_boolean_array_from_packed() {
629 let v = [1_u8, 2_u8, 3_u8];
630 let arr = BooleanArray::new_from_packed(v, 0, 24);
631 assert_eq!(24, arr.len());
632 assert_eq!(0, arr.offset());
633 assert_eq!(0, arr.null_count());
634 assert!(arr.nulls.is_none());
635 for i in 0..24 {
636 assert!(!arr.is_null(i));
637 assert!(arr.is_valid(i));
638 assert_eq!(
639 i == 0 || i == 9 || i == 16 || i == 17,
640 arr.value(i),
641 "failed t {i}"
642 )
643 }
644 }
645
646 #[test]
647 fn test_boolean_array_from_slice_u8() {
648 let v: Vec<u8> = vec![1, 2, 3];
649 let slice = &v[..];
650 let arr = BooleanArray::new_from_u8(slice);
651 assert_eq!(24, arr.len());
652 assert_eq!(0, arr.offset());
653 assert_eq!(0, arr.null_count());
654 assert!(arr.nulls().is_none());
655 for i in 0..24 {
656 assert!(!arr.is_null(i));
657 assert!(arr.is_valid(i));
658 assert_eq!(
659 i == 0 || i == 9 || i == 16 || i == 17,
660 arr.value(i),
661 "failed t {i}"
662 )
663 }
664 }
665
666 #[test]
667 fn test_boolean_array_from_iter() {
668 let v = vec![Some(false), Some(true), Some(false), Some(true)];
669 let arr = v.into_iter().collect::<BooleanArray>();
670 assert_eq!(4, arr.len());
671 assert_eq!(0, arr.offset());
672 assert_eq!(0, arr.null_count());
673 assert!(arr.nulls().is_none());
674 for i in 0..3 {
675 assert!(!arr.is_null(i));
676 assert!(arr.is_valid(i));
677 assert_eq!(i == 1 || i == 3, arr.value(i), "failed at {i}")
678 }
679 }
680
681 #[test]
682 fn test_boolean_array_from_non_nullable_iter() {
683 let v = vec![true, false, true];
684 let arr = v.into_iter().collect::<BooleanArray>();
685 assert_eq!(3, arr.len());
686 assert_eq!(0, arr.offset());
687 assert_eq!(0, arr.null_count());
688 assert!(arr.nulls().is_none());
689
690 assert!(arr.value(0));
691 assert!(!arr.value(1));
692 assert!(arr.value(2));
693 }
694
695 #[test]
696 fn test_boolean_array_from_nullable_iter() {
697 let v = vec![Some(true), None, Some(false), None];
698 let arr = v.into_iter().collect::<BooleanArray>();
699 assert_eq!(4, arr.len());
700 assert_eq!(0, arr.offset());
701 assert_eq!(2, arr.null_count());
702 assert!(arr.nulls().is_some());
703
704 assert!(arr.is_valid(0));
705 assert!(arr.is_null(1));
706 assert!(arr.is_valid(2));
707 assert!(arr.is_null(3));
708
709 assert!(arr.value(0));
710 assert!(!arr.value(2));
711 }
712
713 #[test]
714 fn test_boolean_array_from_nullable_trusted_len_iter() {
715 let v = vec![Some(true), None, Some(false), None];
717 let expected = v.clone().into_iter().collect::<BooleanArray>();
718 let actual = unsafe {
719 BooleanArray::from_trusted_len_iter(v.into_iter())
721 };
722 assert_eq!(expected, actual);
723 }
724
725 #[test]
726 fn test_boolean_array_from_iter_with_larger_upper_bound() {
727 let iterator = vec![Some(true), None, Some(false), None]
730 .into_iter()
731 .filter(Option::is_some);
732 let arr = iterator.collect::<BooleanArray>();
733 assert_eq!(2, arr.len());
734 }
735
736 #[test]
737 fn test_boolean_array_builder() {
738 let buf = Buffer::from([27_u8]);
741 let buf2 = buf.clone();
742 let data = ArrayData::builder(DataType::Boolean)
743 .len(5)
744 .offset(2)
745 .add_buffer(buf)
746 .build()
747 .unwrap();
748 let arr = BooleanArray::from(data);
749 assert_eq!(&buf2, arr.values().inner());
750 assert_eq!(5, arr.len());
751 assert_eq!(2, arr.offset());
752 assert_eq!(0, arr.null_count());
753 for i in 0..3 {
754 assert_eq!(i != 0, arr.value(i), "failed at {i}");
755 }
756 }
757
758 #[test]
759 #[should_panic(
760 expected = "Trying to access an element at index 4 from a BooleanArray of length 3"
761 )]
762 fn test_fixed_size_binary_array_get_value_index_out_of_bound() {
763 let v = vec![Some(true), None, Some(false)];
764 let array = v.into_iter().collect::<BooleanArray>();
765
766 array.value(4);
767 }
768
769 #[test]
770 #[should_panic(expected = "BooleanArray data should contain a single buffer only \
771 (values buffer)")]
772 #[cfg(not(feature = "force_validate"))]
775 fn test_boolean_array_invalid_buffer_len() {
776 let data = unsafe {
777 ArrayData::builder(DataType::Boolean)
778 .len(5)
779 .build_unchecked()
780 };
781 drop(BooleanArray::from(data));
782 }
783
784 #[test]
785 #[should_panic(expected = "BooleanArray expected ArrayData with type Boolean got Int32")]
786 fn test_from_array_data_validation() {
787 let _ = BooleanArray::from(ArrayData::new_empty(&DataType::Int32));
788 }
789
790 #[test]
791 #[cfg_attr(miri, ignore)] fn test_true_false_count() {
793 let mut rng = rng();
794
795 for _ in 0..10 {
796 let d: Vec<_> = (0..2000).map(|_| rng.random_bool(0.5)).collect();
798 let b = BooleanArray::from(d.clone());
799
800 let expected_true = d.iter().filter(|x| **x).count();
801 assert_eq!(b.true_count(), expected_true);
802 assert_eq!(b.false_count(), d.len() - expected_true);
803
804 let d: Vec<_> = (0..2000)
806 .map(|_| rng.random_bool(0.5).then(|| rng.random_bool(0.5)))
807 .collect();
808 let b = BooleanArray::from(d.clone());
809
810 let expected_true = d.iter().filter(|x| matches!(x, Some(true))).count();
811 assert_eq!(b.true_count(), expected_true);
812
813 let expected_false = d.iter().filter(|x| matches!(x, Some(false))).count();
814 assert_eq!(b.false_count(), expected_false);
815 }
816 }
817
818 #[test]
819 fn test_into_parts() {
820 let boolean_array = [Some(true), None, Some(false)]
821 .into_iter()
822 .collect::<BooleanArray>();
823 let (values, nulls) = boolean_array.into_parts();
824 assert_eq!(values.values(), &[0b0000_0001]);
825 assert!(nulls.is_some());
826 assert_eq!(nulls.unwrap().buffer().as_slice(), &[0b0000_0101]);
827
828 let boolean_array =
829 BooleanArray::from(vec![false, false, false, false, false, false, false, true]);
830 let (values, nulls) = boolean_array.into_parts();
831 assert_eq!(values.values(), &[0b1000_0000]);
832 assert!(nulls.is_none());
833 }
834
835 #[test]
836 fn test_new_null_array() {
837 let arr = BooleanArray::new_null(5);
838
839 assert_eq!(arr.len(), 5);
840 assert_eq!(arr.null_count(), 5);
841 assert_eq!(arr.true_count(), 0);
842 assert_eq!(arr.false_count(), 0);
843
844 for i in 0..5 {
845 assert!(arr.is_null(i));
846 assert!(!arr.is_valid(i));
847 }
848 }
849
850 #[test]
851 fn test_slice_with_nulls() {
852 let arr = BooleanArray::from(vec![Some(true), None, Some(false)]);
853 let sliced = arr.slice(1, 2);
854
855 assert_eq!(sliced.len(), 2);
856 assert_eq!(sliced.null_count(), 1);
857
858 assert!(sliced.is_null(0));
859 assert!(sliced.is_valid(1));
860 assert!(!sliced.value(1));
861 }
862}