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
289unsafe impl Array for BooleanArray {
291 fn as_any(&self) -> &dyn Any {
292 self
293 }
294
295 fn to_data(&self) -> ArrayData {
296 self.clone().into()
297 }
298
299 fn into_data(self) -> ArrayData {
300 self.into()
301 }
302
303 fn data_type(&self) -> &DataType {
304 &DataType::Boolean
305 }
306
307 fn slice(&self, offset: usize, length: usize) -> ArrayRef {
308 Arc::new(self.slice(offset, length))
309 }
310
311 fn len(&self) -> usize {
312 self.values.len()
313 }
314
315 fn is_empty(&self) -> bool {
316 self.values.is_empty()
317 }
318
319 fn shrink_to_fit(&mut self) {
320 self.values.shrink_to_fit();
321 if let Some(nulls) = &mut self.nulls {
322 nulls.shrink_to_fit();
323 }
324 }
325
326 fn offset(&self) -> usize {
327 self.values.offset()
328 }
329
330 fn nulls(&self) -> Option<&NullBuffer> {
331 self.nulls.as_ref()
332 }
333
334 fn logical_null_count(&self) -> usize {
335 self.null_count()
336 }
337
338 fn get_buffer_memory_size(&self) -> usize {
339 let mut sum = self.values.inner().capacity();
340 if let Some(x) = &self.nulls {
341 sum += x.buffer().capacity()
342 }
343 sum
344 }
345
346 fn get_array_memory_size(&self) -> usize {
347 std::mem::size_of::<Self>() + self.get_buffer_memory_size()
348 }
349}
350
351impl ArrayAccessor for &BooleanArray {
352 type Item = bool;
353
354 fn value(&self, index: usize) -> Self::Item {
355 BooleanArray::value(self, index)
356 }
357
358 unsafe fn value_unchecked(&self, index: usize) -> Self::Item {
359 unsafe { BooleanArray::value_unchecked(self, index) }
360 }
361}
362
363impl From<Vec<bool>> for BooleanArray {
364 fn from(data: Vec<bool>) -> Self {
365 let mut mut_buf = MutableBuffer::new_null(data.len());
366 {
367 let mut_slice = mut_buf.as_slice_mut();
368 for (i, b) in data.iter().enumerate() {
369 if *b {
370 bit_util::set_bit(mut_slice, i);
371 }
372 }
373 }
374 let array_data = ArrayData::builder(DataType::Boolean)
375 .len(data.len())
376 .add_buffer(mut_buf.into());
377
378 let array_data = unsafe { array_data.build_unchecked() };
379 BooleanArray::from(array_data)
380 }
381}
382
383impl From<Vec<Option<bool>>> for BooleanArray {
384 fn from(data: Vec<Option<bool>>) -> Self {
385 data.iter().collect()
386 }
387}
388
389impl From<ArrayData> for BooleanArray {
390 fn from(data: ArrayData) -> Self {
391 assert_eq!(
392 data.data_type(),
393 &DataType::Boolean,
394 "BooleanArray expected ArrayData with type {} got {}",
395 DataType::Boolean,
396 data.data_type()
397 );
398 assert_eq!(
399 data.buffers().len(),
400 1,
401 "BooleanArray data should contain a single buffer only (values buffer)"
402 );
403 let values = BooleanBuffer::new(data.buffers()[0].clone(), data.offset(), data.len());
404
405 Self {
406 values,
407 nulls: data.nulls().cloned(),
408 }
409 }
410}
411
412impl From<BooleanArray> for ArrayData {
413 fn from(array: BooleanArray) -> Self {
414 let builder = ArrayDataBuilder::new(DataType::Boolean)
415 .len(array.values.len())
416 .offset(array.values.offset())
417 .nulls(array.nulls)
418 .buffers(vec![array.values.into_inner()]);
419
420 unsafe { builder.build_unchecked() }
421 }
422}
423
424impl<'a> IntoIterator for &'a BooleanArray {
425 type Item = Option<bool>;
426 type IntoIter = BooleanIter<'a>;
427
428 fn into_iter(self) -> Self::IntoIter {
429 BooleanIter::<'a>::new(self)
430 }
431}
432
433impl<'a> BooleanArray {
434 pub fn iter(&'a self) -> BooleanIter<'a> {
436 BooleanIter::<'a>::new(self)
437 }
438}
439
440#[derive(Debug)]
449struct BooleanAdapter {
450 pub native: Option<bool>,
452}
453
454impl From<bool> for BooleanAdapter {
455 fn from(value: bool) -> Self {
456 BooleanAdapter {
457 native: Some(value),
458 }
459 }
460}
461
462impl From<&bool> for BooleanAdapter {
463 fn from(value: &bool) -> Self {
464 BooleanAdapter {
465 native: Some(*value),
466 }
467 }
468}
469
470impl From<Option<bool>> for BooleanAdapter {
471 fn from(value: Option<bool>) -> Self {
472 BooleanAdapter { native: value }
473 }
474}
475
476impl From<&Option<bool>> for BooleanAdapter {
477 fn from(value: &Option<bool>) -> Self {
478 BooleanAdapter { native: *value }
479 }
480}
481
482impl<Ptr: Into<BooleanAdapter>> FromIterator<Ptr> for BooleanArray {
483 fn from_iter<I: IntoIterator<Item = Ptr>>(iter: I) -> Self {
484 let iter = iter.into_iter();
485 let capacity = match iter.size_hint() {
486 (lower, Some(upper)) if lower == upper => lower,
487 _ => 0,
488 };
489 let mut builder = BooleanBuilder::with_capacity(capacity);
490 builder.extend(iter.map(|item| item.into().native));
491 builder.finish()
492 }
493}
494
495impl BooleanArray {
496 #[inline]
508 #[allow(
509 private_bounds,
510 reason = "We will expose BooleanAdapter if there is a need"
511 )]
512 pub unsafe fn from_trusted_len_iter<I, P>(iter: I) -> Self
513 where
514 P: Into<BooleanAdapter>,
515 I: ExactSizeIterator<Item = P>,
516 {
517 let data_len = iter.len();
518
519 let num_bytes = bit_util::ceil(data_len, 8);
520 let mut null_builder = MutableBuffer::from_len_zeroed(num_bytes);
521 let mut val_builder = MutableBuffer::from_len_zeroed(num_bytes);
522
523 let data = val_builder.as_slice_mut();
524
525 let null_slice = null_builder.as_slice_mut();
526 iter.enumerate().for_each(|(i, item)| {
527 if let Some(a) = item.into().native {
528 unsafe {
529 bit_util::set_bit_raw(null_slice.as_mut_ptr(), i);
532 if a {
533 bit_util::set_bit_raw(data.as_mut_ptr(), i);
534 }
535 }
536 }
537 });
538
539 let data = unsafe {
540 ArrayData::new_unchecked(
541 DataType::Boolean,
542 data_len,
543 None,
544 Some(null_builder.into()),
545 0,
546 vec![val_builder.into()],
547 vec![],
548 )
549 };
550 BooleanArray::from(data)
551 }
552}
553
554impl From<BooleanBuffer> for BooleanArray {
555 fn from(values: BooleanBuffer) -> Self {
556 Self {
557 values,
558 nulls: None,
559 }
560 }
561}
562
563#[cfg(test)]
564mod tests {
565 use super::*;
566 use arrow_buffer::Buffer;
567 use rand::{Rng, rng};
568
569 #[test]
570 fn test_boolean_fmt_debug() {
571 let arr = BooleanArray::from(vec![true, false, false]);
572 assert_eq!(
573 "BooleanArray\n[\n true,\n false,\n false,\n]",
574 format!("{arr:?}")
575 );
576 }
577
578 #[test]
579 fn test_boolean_with_null_fmt_debug() {
580 let mut builder = BooleanArray::builder(3);
581 builder.append_value(true);
582 builder.append_null();
583 builder.append_value(false);
584 let arr = builder.finish();
585 assert_eq!(
586 "BooleanArray\n[\n true,\n null,\n false,\n]",
587 format!("{arr:?}")
588 );
589 }
590
591 #[test]
592 fn test_boolean_array_from_vec() {
593 let buf = Buffer::from([10_u8]);
594 let arr = BooleanArray::from(vec![false, true, false, true]);
595 assert_eq!(&buf, arr.values().inner());
596 assert_eq!(4, arr.len());
597 assert_eq!(0, arr.offset());
598 assert_eq!(0, arr.null_count());
599 for i in 0..4 {
600 assert!(!arr.is_null(i));
601 assert!(arr.is_valid(i));
602 assert_eq!(i == 1 || i == 3, arr.value(i), "failed at {i}")
603 }
604 }
605
606 #[test]
607 fn test_boolean_array_from_vec_option() {
608 let buf = Buffer::from([10_u8]);
609 let arr = BooleanArray::from(vec![Some(false), Some(true), None, Some(true)]);
610 assert_eq!(&buf, arr.values().inner());
611 assert_eq!(4, arr.len());
612 assert_eq!(0, arr.offset());
613 assert_eq!(1, arr.null_count());
614 for i in 0..4 {
615 if i == 2 {
616 assert!(arr.is_null(i));
617 assert!(!arr.is_valid(i));
618 } else {
619 assert!(!arr.is_null(i));
620 assert!(arr.is_valid(i));
621 assert_eq!(i == 1 || i == 3, arr.value(i), "failed at {i}")
622 }
623 }
624 }
625
626 #[test]
627 fn test_boolean_array_from_packed() {
628 let v = [1_u8, 2_u8, 3_u8];
629 let arr = BooleanArray::new_from_packed(v, 0, 24);
630 assert_eq!(24, arr.len());
631 assert_eq!(0, arr.offset());
632 assert_eq!(0, arr.null_count());
633 assert!(arr.nulls.is_none());
634 for i in 0..24 {
635 assert!(!arr.is_null(i));
636 assert!(arr.is_valid(i));
637 assert_eq!(
638 i == 0 || i == 9 || i == 16 || i == 17,
639 arr.value(i),
640 "failed t {i}"
641 )
642 }
643 }
644
645 #[test]
646 fn test_boolean_array_from_slice_u8() {
647 let v: Vec<u8> = vec![1, 2, 3];
648 let slice = &v[..];
649 let arr = BooleanArray::new_from_u8(slice);
650 assert_eq!(24, arr.len());
651 assert_eq!(0, arr.offset());
652 assert_eq!(0, arr.null_count());
653 assert!(arr.nulls().is_none());
654 for i in 0..24 {
655 assert!(!arr.is_null(i));
656 assert!(arr.is_valid(i));
657 assert_eq!(
658 i == 0 || i == 9 || i == 16 || i == 17,
659 arr.value(i),
660 "failed t {i}"
661 )
662 }
663 }
664
665 #[test]
666 fn test_boolean_array_from_iter() {
667 let v = vec![Some(false), Some(true), Some(false), Some(true)];
668 let arr = v.into_iter().collect::<BooleanArray>();
669 assert_eq!(4, arr.len());
670 assert_eq!(0, arr.offset());
671 assert_eq!(0, arr.null_count());
672 assert!(arr.nulls().is_none());
673 for i in 0..3 {
674 assert!(!arr.is_null(i));
675 assert!(arr.is_valid(i));
676 assert_eq!(i == 1 || i == 3, arr.value(i), "failed at {i}")
677 }
678 }
679
680 #[test]
681 fn test_boolean_array_from_non_nullable_iter() {
682 let v = vec![true, false, true];
683 let arr = v.into_iter().collect::<BooleanArray>();
684 assert_eq!(3, arr.len());
685 assert_eq!(0, arr.offset());
686 assert_eq!(0, arr.null_count());
687 assert!(arr.nulls().is_none());
688
689 assert!(arr.value(0));
690 assert!(!arr.value(1));
691 assert!(arr.value(2));
692 }
693
694 #[test]
695 fn test_boolean_array_from_nullable_iter() {
696 let v = vec![Some(true), None, Some(false), None];
697 let arr = v.into_iter().collect::<BooleanArray>();
698 assert_eq!(4, arr.len());
699 assert_eq!(0, arr.offset());
700 assert_eq!(2, arr.null_count());
701 assert!(arr.nulls().is_some());
702
703 assert!(arr.is_valid(0));
704 assert!(arr.is_null(1));
705 assert!(arr.is_valid(2));
706 assert!(arr.is_null(3));
707
708 assert!(arr.value(0));
709 assert!(!arr.value(2));
710 }
711
712 #[test]
713 fn test_boolean_array_from_nullable_trusted_len_iter() {
714 let v = vec![Some(true), None, Some(false), None];
716 let expected = v.clone().into_iter().collect::<BooleanArray>();
717 let actual = unsafe {
718 BooleanArray::from_trusted_len_iter(v.into_iter())
720 };
721 assert_eq!(expected, actual);
722 }
723
724 #[test]
725 fn test_boolean_array_from_iter_with_larger_upper_bound() {
726 let iterator = vec![Some(true), None, Some(false), None]
729 .into_iter()
730 .filter(Option::is_some);
731 let arr = iterator.collect::<BooleanArray>();
732 assert_eq!(2, arr.len());
733 }
734
735 #[test]
736 fn test_boolean_array_builder() {
737 let buf = Buffer::from([27_u8]);
740 let buf2 = buf.clone();
741 let data = ArrayData::builder(DataType::Boolean)
742 .len(5)
743 .offset(2)
744 .add_buffer(buf)
745 .build()
746 .unwrap();
747 let arr = BooleanArray::from(data);
748 assert_eq!(&buf2, arr.values().inner());
749 assert_eq!(5, arr.len());
750 assert_eq!(2, arr.offset());
751 assert_eq!(0, arr.null_count());
752 for i in 0..3 {
753 assert_eq!(i != 0, arr.value(i), "failed at {i}");
754 }
755 }
756
757 #[test]
758 #[should_panic(
759 expected = "Trying to access an element at index 4 from a BooleanArray of length 3"
760 )]
761 fn test_fixed_size_binary_array_get_value_index_out_of_bound() {
762 let v = vec![Some(true), None, Some(false)];
763 let array = v.into_iter().collect::<BooleanArray>();
764
765 array.value(4);
766 }
767
768 #[test]
769 #[should_panic(expected = "BooleanArray data should contain a single buffer only \
770 (values buffer)")]
771 #[cfg(not(feature = "force_validate"))]
774 fn test_boolean_array_invalid_buffer_len() {
775 let data = unsafe {
776 ArrayData::builder(DataType::Boolean)
777 .len(5)
778 .build_unchecked()
779 };
780 drop(BooleanArray::from(data));
781 }
782
783 #[test]
784 #[should_panic(expected = "BooleanArray expected ArrayData with type Boolean got Int32")]
785 fn test_from_array_data_validation() {
786 let _ = BooleanArray::from(ArrayData::new_empty(&DataType::Int32));
787 }
788
789 #[test]
790 #[cfg_attr(miri, ignore)] fn test_true_false_count() {
792 let mut rng = rng();
793
794 for _ in 0..10 {
795 let d: Vec<_> = (0..2000).map(|_| rng.random_bool(0.5)).collect();
797 let b = BooleanArray::from(d.clone());
798
799 let expected_true = d.iter().filter(|x| **x).count();
800 assert_eq!(b.true_count(), expected_true);
801 assert_eq!(b.false_count(), d.len() - expected_true);
802
803 let d: Vec<_> = (0..2000)
805 .map(|_| rng.random_bool(0.5).then(|| rng.random_bool(0.5)))
806 .collect();
807 let b = BooleanArray::from(d.clone());
808
809 let expected_true = d.iter().filter(|x| matches!(x, Some(true))).count();
810 assert_eq!(b.true_count(), expected_true);
811
812 let expected_false = d.iter().filter(|x| matches!(x, Some(false))).count();
813 assert_eq!(b.false_count(), expected_false);
814 }
815 }
816
817 #[test]
818 fn test_into_parts() {
819 let boolean_array = [Some(true), None, Some(false)]
820 .into_iter()
821 .collect::<BooleanArray>();
822 let (values, nulls) = boolean_array.into_parts();
823 assert_eq!(values.values(), &[0b0000_0001]);
824 assert!(nulls.is_some());
825 assert_eq!(nulls.unwrap().buffer().as_slice(), &[0b0000_0101]);
826
827 let boolean_array =
828 BooleanArray::from(vec![false, false, false, false, false, false, false, true]);
829 let (values, nulls) = boolean_array.into_parts();
830 assert_eq!(values.values(), &[0b1000_0000]);
831 assert!(nulls.is_none());
832 }
833
834 #[test]
835 fn test_new_null_array() {
836 let arr = BooleanArray::new_null(5);
837
838 assert_eq!(arr.len(), 5);
839 assert_eq!(arr.null_count(), 5);
840 assert_eq!(arr.true_count(), 0);
841 assert_eq!(arr.false_count(), 0);
842
843 for i in 0..5 {
844 assert!(arr.is_null(i));
845 assert!(!arr.is_valid(i));
846 }
847 }
848
849 #[test]
850 fn test_slice_with_nulls() {
851 let arr = BooleanArray::from(vec![Some(true), None, Some(false)]);
852 let sliced = arr.slice(1, 2);
853
854 assert_eq!(sliced.len(), 2);
855 assert_eq!(sliced.null_count(), 1);
856
857 assert!(sliced.is_null(0));
858 assert!(sliced.is_valid(1));
859 assert!(!sliced.value(1));
860 }
861}