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 let (data_type, len, nulls, offset, mut buffers, _child_data) = data.into_parts();
392 assert_eq!(
393 data_type,
394 DataType::Boolean,
395 "BooleanArray expected ArrayData with type Boolean got {data_type:?}",
396 );
397 assert_eq!(
398 buffers.len(),
399 1,
400 "BooleanArray data should contain a single buffer only (values buffer)"
401 );
402 let buffer = buffers.pop().expect("checked above");
403 let values = BooleanBuffer::new(buffer, offset, len);
404
405 Self { values, nulls }
406 }
407}
408
409impl From<BooleanArray> for ArrayData {
410 fn from(array: BooleanArray) -> Self {
411 let builder = ArrayDataBuilder::new(DataType::Boolean)
412 .len(array.values.len())
413 .offset(array.values.offset())
414 .nulls(array.nulls)
415 .buffers(vec![array.values.into_inner()]);
416
417 unsafe { builder.build_unchecked() }
418 }
419}
420
421impl<'a> IntoIterator for &'a BooleanArray {
422 type Item = Option<bool>;
423 type IntoIter = BooleanIter<'a>;
424
425 fn into_iter(self) -> Self::IntoIter {
426 BooleanIter::<'a>::new(self)
427 }
428}
429
430impl<'a> BooleanArray {
431 pub fn iter(&'a self) -> BooleanIter<'a> {
433 BooleanIter::<'a>::new(self)
434 }
435}
436
437#[derive(Debug)]
446struct BooleanAdapter {
447 pub native: Option<bool>,
449}
450
451impl From<bool> for BooleanAdapter {
452 fn from(value: bool) -> Self {
453 BooleanAdapter {
454 native: Some(value),
455 }
456 }
457}
458
459impl From<&bool> for BooleanAdapter {
460 fn from(value: &bool) -> Self {
461 BooleanAdapter {
462 native: Some(*value),
463 }
464 }
465}
466
467impl From<Option<bool>> for BooleanAdapter {
468 fn from(value: Option<bool>) -> Self {
469 BooleanAdapter { native: value }
470 }
471}
472
473impl From<&Option<bool>> for BooleanAdapter {
474 fn from(value: &Option<bool>) -> Self {
475 BooleanAdapter { native: *value }
476 }
477}
478
479impl<Ptr: Into<BooleanAdapter>> FromIterator<Ptr> for BooleanArray {
480 fn from_iter<I: IntoIterator<Item = Ptr>>(iter: I) -> Self {
481 let iter = iter.into_iter();
482 let capacity = match iter.size_hint() {
483 (lower, Some(upper)) if lower == upper => lower,
484 _ => 0,
485 };
486 let mut builder = BooleanBuilder::with_capacity(capacity);
487 builder.extend(iter.map(|item| item.into().native));
488 builder.finish()
489 }
490}
491
492impl BooleanArray {
493 #[inline]
505 #[allow(
506 private_bounds,
507 reason = "We will expose BooleanAdapter if there is a need"
508 )]
509 pub unsafe fn from_trusted_len_iter<I, P>(iter: I) -> Self
510 where
511 P: Into<BooleanAdapter>,
512 I: ExactSizeIterator<Item = P>,
513 {
514 let data_len = iter.len();
515
516 let num_bytes = bit_util::ceil(data_len, 8);
517 let mut null_builder = MutableBuffer::from_len_zeroed(num_bytes);
518 let mut val_builder = MutableBuffer::from_len_zeroed(num_bytes);
519
520 let data = val_builder.as_slice_mut();
521
522 let null_slice = null_builder.as_slice_mut();
523 iter.enumerate().for_each(|(i, item)| {
524 if let Some(a) = item.into().native {
525 unsafe {
526 bit_util::set_bit_raw(null_slice.as_mut_ptr(), i);
529 if a {
530 bit_util::set_bit_raw(data.as_mut_ptr(), i);
531 }
532 }
533 }
534 });
535
536 let values = BooleanBuffer::new(val_builder.into(), 0, data_len);
537 let nulls = Some(NullBuffer::new(BooleanBuffer::new(
538 null_builder.into(),
539 0,
540 data_len,
541 )))
542 .filter(|n| n.null_count() > 0);
543 BooleanArray::new(values, nulls)
544 }
545}
546
547impl From<BooleanBuffer> for BooleanArray {
548 fn from(values: BooleanBuffer) -> Self {
549 Self {
550 values,
551 nulls: None,
552 }
553 }
554}
555
556#[cfg(test)]
557mod tests {
558 use super::*;
559 use arrow_buffer::Buffer;
560 use rand::{Rng, rng};
561
562 #[test]
563 fn test_boolean_fmt_debug() {
564 let arr = BooleanArray::from(vec![true, false, false]);
565 assert_eq!(
566 "BooleanArray\n[\n true,\n false,\n false,\n]",
567 format!("{arr:?}")
568 );
569 }
570
571 #[test]
572 fn test_boolean_with_null_fmt_debug() {
573 let mut builder = BooleanArray::builder(3);
574 builder.append_value(true);
575 builder.append_null();
576 builder.append_value(false);
577 let arr = builder.finish();
578 assert_eq!(
579 "BooleanArray\n[\n true,\n null,\n false,\n]",
580 format!("{arr:?}")
581 );
582 }
583
584 #[test]
585 fn test_boolean_array_from_vec() {
586 let buf = Buffer::from([10_u8]);
587 let arr = BooleanArray::from(vec![false, true, false, true]);
588 assert_eq!(&buf, arr.values().inner());
589 assert_eq!(4, arr.len());
590 assert_eq!(0, arr.offset());
591 assert_eq!(0, arr.null_count());
592 for i in 0..4 {
593 assert!(!arr.is_null(i));
594 assert!(arr.is_valid(i));
595 assert_eq!(i == 1 || i == 3, arr.value(i), "failed at {i}")
596 }
597 }
598
599 #[test]
600 fn test_boolean_array_from_vec_option() {
601 let buf = Buffer::from([10_u8]);
602 let arr = BooleanArray::from(vec![Some(false), Some(true), None, Some(true)]);
603 assert_eq!(&buf, arr.values().inner());
604 assert_eq!(4, arr.len());
605 assert_eq!(0, arr.offset());
606 assert_eq!(1, arr.null_count());
607 for i in 0..4 {
608 if i == 2 {
609 assert!(arr.is_null(i));
610 assert!(!arr.is_valid(i));
611 } else {
612 assert!(!arr.is_null(i));
613 assert!(arr.is_valid(i));
614 assert_eq!(i == 1 || i == 3, arr.value(i), "failed at {i}")
615 }
616 }
617 }
618
619 #[test]
620 fn test_boolean_array_from_packed() {
621 let v = [1_u8, 2_u8, 3_u8];
622 let arr = BooleanArray::new_from_packed(v, 0, 24);
623 assert_eq!(24, arr.len());
624 assert_eq!(0, arr.offset());
625 assert_eq!(0, arr.null_count());
626 assert!(arr.nulls.is_none());
627 for i in 0..24 {
628 assert!(!arr.is_null(i));
629 assert!(arr.is_valid(i));
630 assert_eq!(
631 i == 0 || i == 9 || i == 16 || i == 17,
632 arr.value(i),
633 "failed t {i}"
634 )
635 }
636 }
637
638 #[test]
639 fn test_boolean_array_from_slice_u8() {
640 let v: Vec<u8> = vec![1, 2, 3];
641 let slice = &v[..];
642 let arr = BooleanArray::new_from_u8(slice);
643 assert_eq!(24, arr.len());
644 assert_eq!(0, arr.offset());
645 assert_eq!(0, arr.null_count());
646 assert!(arr.nulls().is_none());
647 for i in 0..24 {
648 assert!(!arr.is_null(i));
649 assert!(arr.is_valid(i));
650 assert_eq!(
651 i == 0 || i == 9 || i == 16 || i == 17,
652 arr.value(i),
653 "failed t {i}"
654 )
655 }
656 }
657
658 #[test]
659 fn test_boolean_array_from_iter() {
660 let v = vec![Some(false), Some(true), Some(false), Some(true)];
661 let arr = v.into_iter().collect::<BooleanArray>();
662 assert_eq!(4, arr.len());
663 assert_eq!(0, arr.offset());
664 assert_eq!(0, arr.null_count());
665 assert!(arr.nulls().is_none());
666 for i in 0..3 {
667 assert!(!arr.is_null(i));
668 assert!(arr.is_valid(i));
669 assert_eq!(i == 1 || i == 3, arr.value(i), "failed at {i}")
670 }
671 }
672
673 #[test]
674 fn test_boolean_array_from_non_nullable_iter() {
675 let v = vec![true, false, true];
676 let arr = v.into_iter().collect::<BooleanArray>();
677 assert_eq!(3, arr.len());
678 assert_eq!(0, arr.offset());
679 assert_eq!(0, arr.null_count());
680 assert!(arr.nulls().is_none());
681
682 assert!(arr.value(0));
683 assert!(!arr.value(1));
684 assert!(arr.value(2));
685 }
686
687 #[test]
688 fn test_boolean_array_from_nullable_iter() {
689 let v = vec![Some(true), None, Some(false), None];
690 let arr = v.into_iter().collect::<BooleanArray>();
691 assert_eq!(4, arr.len());
692 assert_eq!(0, arr.offset());
693 assert_eq!(2, arr.null_count());
694 assert!(arr.nulls().is_some());
695
696 assert!(arr.is_valid(0));
697 assert!(arr.is_null(1));
698 assert!(arr.is_valid(2));
699 assert!(arr.is_null(3));
700
701 assert!(arr.value(0));
702 assert!(!arr.value(2));
703 }
704
705 #[test]
706 fn test_boolean_array_from_nullable_trusted_len_iter() {
707 let v = vec![Some(true), None, Some(false), None];
709 let expected = v.clone().into_iter().collect::<BooleanArray>();
710 let actual = unsafe {
711 BooleanArray::from_trusted_len_iter(v.into_iter())
713 };
714 assert_eq!(expected, actual);
715 }
716
717 #[test]
718 fn test_boolean_array_from_iter_with_larger_upper_bound() {
719 let iterator = vec![Some(true), None, Some(false), None]
722 .into_iter()
723 .filter(Option::is_some);
724 let arr = iterator.collect::<BooleanArray>();
725 assert_eq!(2, arr.len());
726 }
727
728 #[test]
729 fn test_boolean_array_builder() {
730 let buf = Buffer::from([27_u8]);
733 let buf2 = buf.clone();
734 let data = ArrayData::builder(DataType::Boolean)
735 .len(5)
736 .offset(2)
737 .add_buffer(buf)
738 .build()
739 .unwrap();
740 let arr = BooleanArray::from(data);
741 assert_eq!(&buf2, arr.values().inner());
742 assert_eq!(5, arr.len());
743 assert_eq!(2, arr.offset());
744 assert_eq!(0, arr.null_count());
745 for i in 0..3 {
746 assert_eq!(i != 0, arr.value(i), "failed at {i}");
747 }
748 }
749
750 #[test]
751 #[should_panic(
752 expected = "Trying to access an element at index 4 from a BooleanArray of length 3"
753 )]
754 fn test_fixed_size_binary_array_get_value_index_out_of_bound() {
755 let v = vec![Some(true), None, Some(false)];
756 let array = v.into_iter().collect::<BooleanArray>();
757
758 array.value(4);
759 }
760
761 #[test]
762 #[should_panic(expected = "BooleanArray data should contain a single buffer only \
763 (values buffer)")]
764 #[cfg(not(feature = "force_validate"))]
767 fn test_boolean_array_invalid_buffer_len() {
768 let data = unsafe {
769 ArrayData::builder(DataType::Boolean)
770 .len(5)
771 .build_unchecked()
772 };
773 drop(BooleanArray::from(data));
774 }
775
776 #[test]
777 #[should_panic(expected = "BooleanArray expected ArrayData with type Boolean got Int32")]
778 fn test_from_array_data_validation() {
779 let _ = BooleanArray::from(ArrayData::new_empty(&DataType::Int32));
780 }
781
782 #[test]
783 #[cfg_attr(miri, ignore)] fn test_true_false_count() {
785 let mut rng = rng();
786
787 for _ in 0..10 {
788 let d: Vec<_> = (0..2000).map(|_| rng.random_bool(0.5)).collect();
790 let b = BooleanArray::from(d.clone());
791
792 let expected_true = d.iter().filter(|x| **x).count();
793 assert_eq!(b.true_count(), expected_true);
794 assert_eq!(b.false_count(), d.len() - expected_true);
795
796 let d: Vec<_> = (0..2000)
798 .map(|_| rng.random_bool(0.5).then(|| rng.random_bool(0.5)))
799 .collect();
800 let b = BooleanArray::from(d.clone());
801
802 let expected_true = d.iter().filter(|x| matches!(x, Some(true))).count();
803 assert_eq!(b.true_count(), expected_true);
804
805 let expected_false = d.iter().filter(|x| matches!(x, Some(false))).count();
806 assert_eq!(b.false_count(), expected_false);
807 }
808 }
809
810 #[test]
811 fn test_into_parts() {
812 let boolean_array = [Some(true), None, Some(false)]
813 .into_iter()
814 .collect::<BooleanArray>();
815 let (values, nulls) = boolean_array.into_parts();
816 assert_eq!(values.values(), &[0b0000_0001]);
817 assert!(nulls.is_some());
818 assert_eq!(nulls.unwrap().buffer().as_slice(), &[0b0000_0101]);
819
820 let boolean_array =
821 BooleanArray::from(vec![false, false, false, false, false, false, false, true]);
822 let (values, nulls) = boolean_array.into_parts();
823 assert_eq!(values.values(), &[0b1000_0000]);
824 assert!(nulls.is_none());
825 }
826
827 #[test]
828 fn test_new_null_array() {
829 let arr = BooleanArray::new_null(5);
830
831 assert_eq!(arr.len(), 5);
832 assert_eq!(arr.null_count(), 5);
833 assert_eq!(arr.true_count(), 0);
834 assert_eq!(arr.false_count(), 0);
835
836 for i in 0..5 {
837 assert!(arr.is_null(i));
838 assert!(!arr.is_valid(i));
839 }
840 }
841
842 #[test]
843 fn test_slice_with_nulls() {
844 let arr = BooleanArray::from(vec![Some(true), None, Some(false)]);
845 let sliced = arr.slice(1, 2);
846
847 assert_eq!(sliced.len(), 2);
848 assert_eq!(sliced.null_count(), 1);
849
850 assert!(sliced.is_null(0));
851 assert!(sliced.is_valid(1));
852 assert!(!sliced.value(1));
853 }
854}