1use std::any::Any;
5use std::mem::MaybeUninit;
6
7use num_traits::AsPrimitive;
8use vortex_buffer::BitBufferMut;
9use vortex_buffer::BufferMut;
10use vortex_buffer::ByteBuffer;
11use vortex_buffer::ByteBufferMut;
12use vortex_error::VortexExpect;
13use vortex_error::VortexResult;
14use vortex_error::vortex_bail;
15use vortex_error::vortex_ensure;
16use vortex_error::vortex_panic;
17use vortex_mask::AllOr;
18use vortex_mask::Mask;
19
20use crate::ArrayRef;
21use crate::ArrayView;
22use crate::Canonical;
23use crate::ExecutionCtx;
24use crate::IntoArray;
25#[cfg(debug_assertions)]
26use crate::VortexSessionExecute;
27use crate::arrays::PrimitiveArray;
28use crate::arrays::VarBin;
29use crate::arrays::VarBinArray;
30use crate::arrays::VarBinView;
31use crate::arrays::varbin::VarBinArrayExt;
32use crate::arrays::varbin::VarBinArraySlotsExt;
33use crate::arrays::varbinview::VarBinViewArrayExt;
34use crate::builders::ArrayBuilder;
35use crate::dtype::DType;
36use crate::dtype::OffsetBuilderPType;
37use crate::expr::stats::Precision;
38use crate::expr::stats::Stat;
39#[cfg(debug_assertions)]
40use crate::legacy_session;
41use crate::match_each_integer_ptype;
42use crate::scalar::Scalar;
43use crate::validity::Validity;
44
45pub struct VarBinBuilder<O: OffsetBuilderPType> {
58 dtype: DType,
59 offsets: BufferMut<O>,
60 data: ByteBufferMut,
61 validity: BitBufferMut,
62}
63
64impl<O: OffsetBuilderPType> VarBinBuilder<O> {
65 pub fn new(dtype: DType) -> Self {
67 Self::with_capacity(dtype, 0)
68 }
69
70 pub fn with_capacity(dtype: DType, capacity: usize) -> Self {
72 assert!(
73 matches!(dtype, DType::Utf8(_) | DType::Binary(_)),
74 "VarBinBuilder dtype must be Utf8 or Binary, got {dtype}"
75 );
76 let mut offsets = BufferMut::with_capacity(capacity + 1);
77 offsets.push(O::zero());
78 Self {
79 dtype,
80 offsets,
81 data: BufferMut::empty(),
82 validity: BitBufferMut::with_capacity(capacity),
83 }
84 }
85
86 pub fn with_capacity_bytes(dtype: DType, capacity: usize, bytes: usize) -> Self {
88 let mut builder = Self::with_capacity(dtype, capacity);
89 builder.reserve_data(bytes);
90 builder
91 }
92
93 pub fn reserve_data(&mut self, additional: usize) {
99 self.data.reserve(additional);
100 }
101
102 #[inline]
109 pub fn append(&mut self, value: Option<&[u8]>) {
110 match value {
111 Some(v) => self.append_value(v),
112 None => self.push_null(),
113 }
114 }
115
116 #[inline]
124 pub fn append_value(&mut self, value: impl AsRef<[u8]>) {
125 self.push_value(value.as_ref());
126 self.validity.append_true();
127 }
128
129 pub fn append_n_values(&mut self, value: impl AsRef<[u8]>, n: usize) -> VortexResult<()> {
136 let value = value.as_ref();
137 let Some(num_bytes) = value.len().checked_mul(n) else {
138 vortex_bail!("Byte count overflow: {} values of {} bytes", n, value.len());
139 };
140 self.check_offset_limit(self.data.len(), num_bytes)?;
142 self.offsets.reserve(n);
143 self.data.reserve(num_bytes);
144 for _ in 0..n {
145 self.push_value(value);
146 }
147 self.validity.append_n(true, n);
148 Ok(())
149 }
150
151 #[inline]
157 pub fn push_null(&mut self) {
158 self.push_nulls(1)
159 }
160
161 #[inline]
163 pub fn push_nulls(&mut self, n: usize) {
164 self.offsets.push_n(self.last_offset(), n);
165 self.validity.append_n(false, n);
166 }
167
168 pub fn append_scalar_repeated(&mut self, scalar: &Scalar, n: usize) -> VortexResult<()> {
175 vortex_ensure!(
176 scalar.dtype() == &self.dtype,
177 "VarBinBuilder expected scalar with dtype {}, got {}",
178 self.dtype,
179 scalar.dtype()
180 );
181 match &self.dtype {
182 DType::Utf8(_) => match scalar.as_utf8().value() {
183 Some(value) => self.append_n_values(value, n)?,
184 None => self.push_nulls(n),
185 },
186 DType::Binary(_) => match scalar.as_binary().value() {
187 Some(value) => self.append_n_values(value, n)?,
188 None => self.push_nulls(n),
189 },
190 dtype => vortex_bail!("VarBinBuilder cannot append scalar of dtype {dtype}"),
191 }
192 Ok(())
193 }
194
195 #[inline]
200 pub fn append_values<P>(
201 &mut self,
202 values: &[u8],
203 end_offsets: impl Iterator<Item = P>,
204 validity: &Mask,
205 ) -> VortexResult<()>
206 where
207 P: AsPrimitive<usize>,
208 usize: AsPrimitive<O>,
209 {
210 self.extend_offsets(values.len(), validity.len(), end_offsets)?;
213 self.data.extend_from_slice(values);
214 self.append_validity(validity);
215 Ok(())
216 }
217
218 pub unsafe fn append_decoded<P>(
248 &mut self,
249 num_bytes: usize,
250 slack: usize,
251 lengths: &[P],
252 validity: &Mask,
253 decode: &mut dyn FnMut(&mut [MaybeUninit<u8>]) -> VortexResult<usize>,
254 ) -> VortexResult<()>
255 where
256 P: AsPrimitive<usize>,
257 usize: AsPrimitive<O>,
258 {
259 let Some(capacity) = num_bytes.checked_add(slack) else {
260 vortex_bail!("Decoded size overflow: {num_bytes} + {slack}");
261 };
262 self.check_offset_limit(self.data.len(), num_bytes)?;
265 self.data.reserve(capacity);
266
267 let data_len = self.data.len();
268 let written = decode(self.data.spare_capacity_mut())?;
269 vortex_ensure!(
270 written == num_bytes,
271 "Decoded {written} bytes, expected {num_bytes}"
272 );
273
274 self.extend_offsets(num_bytes, validity.len(), prefix_sums(lengths))?;
277
278 unsafe { self.data.set_len(data_len + num_bytes) };
281 self.append_validity(validity);
282 Ok(())
283 }
284
285 pub fn append_valid_slices<'a, F>(
296 &mut self,
297 num_bytes: usize,
298 validity: &Mask,
299 value: F,
300 ) -> VortexResult<()>
301 where
302 F: FnMut(usize) -> &'a [u8],
303 usize: AsPrimitive<O>,
304 {
305 let data_start = self.data.len();
306 match self.gather_valid_slices(num_bytes, validity, value) {
307 Ok(()) => {
308 self.append_validity(validity);
309 Ok(())
310 }
311 Err(error) => {
312 self.data.truncate(data_start);
314 Err(error)
315 }
316 }
317 }
318
319 pub fn append_varbin(
321 &mut self,
322 array: ArrayView<'_, VarBin>,
323 ctx: &mut ExecutionCtx,
324 ) -> VortexResult<()>
325 where
326 usize: AsPrimitive<O>,
327 {
328 let offsets = array.offsets().clone().execute::<PrimitiveArray>(ctx)?;
329 let bytes: ByteBuffer = array.sliced_bytes();
330 let validity = array
331 .varbin_validity()
332 .execute_mask(array.as_ref().len(), ctx)?;
333 match_each_integer_ptype!(offsets.ptype(), |P| {
334 let offsets = offsets.as_slice::<P>();
335 let first: usize = offsets[0].as_();
336 self.append_values(
337 bytes.as_slice(),
338 offsets[1..]
341 .iter()
342 .map(|offset| AsPrimitive::<usize>::as_(*offset).wrapping_sub(first)),
343 &validity,
344 )
345 })
346 }
347
348 pub fn append_varbinview(
350 &mut self,
351 array: ArrayView<'_, VarBinView>,
352 ctx: &mut ExecutionCtx,
353 ) -> VortexResult<()>
354 where
355 usize: AsPrimitive<O>,
356 {
357 let len = array.as_ref().len();
358 let validity = array.varbinview_validity().execute_mask(len, ctx)?;
359
360 let views = array.views();
363 let buffers = array
364 .data_buffers()
365 .iter()
366 .map(|buffer| buffer.as_host().as_slice())
367 .collect::<Vec<_>>();
368
369 let num_bytes = match validity.bit_buffer() {
370 AllOr::All => views.iter().map(|view| view.len() as usize).sum(),
371 AllOr::None => 0,
372 AllOr::Some(bits) => {
373 let mut total = 0;
374 bits.for_each_set_index(|index| total += views[index].len() as usize);
375 total
376 }
377 };
378
379 self.append_valid_slices(num_bytes, &validity, |index| views[index].bytes(&buffers))
380 }
381
382 #[allow(clippy::disallowed_methods)]
384 pub fn finish_into_varbin(&mut self) -> VarBinArray {
385 assert_eq!(
386 self.offsets.len() - 1,
387 self.validity.len(),
388 "The offset count must be one more than the validity length"
389 );
390
391 let mut fresh_offsets = BufferMut::with_capacity(1);
392 fresh_offsets.push(O::zero());
393 let offsets = PrimitiveArray::new(
394 std::mem::replace(&mut self.offsets, fresh_offsets).freeze(),
395 Validity::NonNullable,
396 );
397 let data = std::mem::replace(&mut self.data, BufferMut::empty());
398 let nulls = std::mem::replace(&mut self.validity, BitBufferMut::empty()).freeze();
399
400 let validity = Validity::from_bit_buffer(nulls, self.dtype.nullability());
401
402 #[cfg(debug_assertions)]
405 {
406 let offsets_are_sorted = offsets
407 .statistics()
408 .compute_is_sorted(&mut legacy_session().create_execution_ctx())
409 .unwrap_or(false);
410 debug_assert!(offsets_are_sorted, "VarBinBuilder offsets must be sorted");
411 }
412 offsets
413 .statistics()
414 .set(Stat::IsSorted, Precision::Exact(true.into()));
415
416 unsafe {
422 VarBinArray::new_unchecked(
423 offsets.into_array(),
424 data.freeze(),
425 self.dtype.clone(),
426 validity,
427 )
428 }
429 }
430
431 #[inline]
432 fn last_offset(&self) -> O {
433 self.offsets[self.offsets.len() - 1]
434 }
435
436 #[inline]
438 fn push_value(&mut self, value: &[u8]) {
439 self.offsets
440 .push(O::from(self.data.len() + value.len()).unwrap_or_else(|| {
441 vortex_panic!(
442 "Failed to convert sum of {} and {} to offset of type {}",
443 self.data.len(),
444 value.len(),
445 std::any::type_name::<O>()
446 )
447 }));
448 self.data.extend_from_slice(value);
449 }
450
451 fn append_validity(&mut self, validity: &Mask) {
452 match validity {
453 Mask::AllTrue(len) => self.validity.append_n(true, *len),
454 Mask::AllFalse(len) => self.validity.append_n(false, *len),
455 Mask::Values(values) => self.validity.append_buffer(values.bit_buffer()),
456 }
457 }
458
459 fn extend_offsets<P>(
465 &mut self,
466 num_bytes: usize,
467 count: usize,
468 end_offsets: impl Iterator<Item = P>,
469 ) -> VortexResult<()>
470 where
471 P: AsPrimitive<usize>,
472 usize: AsPrimitive<O>,
473 {
474 let data_start = self.data.len();
475 let offsets_len = self.offsets.len();
476 self.check_offset_limit(data_start, num_bytes)?;
477 self.offsets.reserve(count);
478
479 let spare = &mut self.offsets.spare_capacity_mut()[..count];
482 let mut end_offsets = end_offsets;
483 let mut previous = 0usize;
484 for slot in spare.iter_mut() {
485 let Some(end) = end_offsets.next() else {
486 vortex_bail!("End offset count is less than the validity length {count}");
487 };
488 let end = end.as_();
489 vortex_ensure!(
490 end >= previous && end <= num_bytes,
491 "End offsets must be monotonically increasing within {num_bytes} bytes, \
492 got {end} after {previous}"
493 );
494 slot.write((data_start + end).as_());
495 previous = end;
496 }
497 vortex_ensure!(
498 end_offsets.next().is_none(),
499 "End offset count exceeds the validity length {count}"
500 );
501 vortex_ensure!(
502 previous == num_bytes,
503 "Final end offset {previous} does not match the value byte count {num_bytes}"
504 );
505
506 unsafe { self.offsets.set_len(offsets_len + count) };
508 Ok(())
509 }
510
511 fn gather_valid_slices<'a, F>(
515 &mut self,
516 num_bytes: usize,
517 validity: &Mask,
518 mut value: F,
519 ) -> VortexResult<()>
520 where
521 F: FnMut(usize) -> &'a [u8],
522 usize: AsPrimitive<O>,
523 {
524 let count = validity.len();
525 let data_start = self.data.len();
526 let offsets_len = self.offsets.len();
527 self.check_offset_limit(data_start, num_bytes)?;
528 self.offsets.reserve(count);
529 self.data.reserve(num_bytes);
530
531 let Self { offsets, data, .. } = self;
534 let spare = &mut offsets.spare_capacity_mut()[..count];
535
536 match validity.bit_buffer() {
537 AllOr::All => {
538 for (row, slot) in spare.iter_mut().enumerate() {
539 data.extend_from_slice(value(row));
540 slot.write(data.len().as_());
541 }
542 }
543 AllOr::None => {
544 spare.fill(MaybeUninit::new(data_start.as_()));
545 }
546 AllOr::Some(bits) => {
547 let mut row = 0;
548 bits.for_each_set_index(|index| {
549 spare[row..index].fill(MaybeUninit::new(data.len().as_()));
551 data.extend_from_slice(value(index));
552 spare[index].write(data.len().as_());
553 row = index + 1;
554 });
555 spare[row..].fill(MaybeUninit::new(data.len().as_()));
556 }
557 }
558
559 vortex_ensure!(
563 data.len() == data_start + num_bytes,
564 "Value slices total {} bytes, expected {num_bytes}",
565 data.len() - data_start
566 );
567
568 unsafe { self.offsets.set_len(offsets_len + count) };
570 Ok(())
571 }
572
573 fn check_offset_limit(&self, data_start: usize, num_bytes: usize) -> VortexResult<()> {
575 let Some(limit) = data_start.checked_add(num_bytes) else {
576 vortex_bail!("Byte offset overflow: {data_start} + {num_bytes}");
577 };
578 vortex_ensure!(
579 u64::try_from(limit).is_ok_and(|limit| limit <= O::max_value_as_u64()),
580 "Byte offset {limit} does not fit in {}",
581 std::any::type_name::<O>()
582 );
583 Ok(())
584 }
585}
586
587impl<O: OffsetBuilderPType> ArrayBuilder for VarBinBuilder<O> {
588 fn as_any(&self) -> &dyn Any {
589 self
590 }
591
592 fn as_any_mut(&mut self) -> &mut dyn Any {
593 self
594 }
595
596 fn dtype(&self) -> &DType {
597 &self.dtype
598 }
599
600 fn len(&self) -> usize {
601 self.validity.len()
602 }
603
604 fn append_zeros(&mut self, n: usize) {
605 self.offsets.push_n(self.last_offset(), n);
606 self.validity.append_n(true, n);
607 }
608
609 unsafe fn append_nulls_unchecked(&mut self, n: usize) {
610 self.push_nulls(n);
611 }
612
613 fn append_scalar(&mut self, scalar: &Scalar) -> VortexResult<()> {
614 self.append_scalar_repeated(scalar, 1)
615 }
616
617 fn reserve_exact(&mut self, additional: usize) {
618 self.offsets.reserve(additional);
619 self.validity.reserve(additional);
620 }
621
622 fn finish(&mut self) -> ArrayRef {
623 self.finish_into_varbin().into_array()
624 }
625
626 fn finish_into_canonical(&mut self, ctx: &mut ExecutionCtx) -> Canonical {
627 self.finish()
628 .execute::<Canonical>(ctx)
629 .vortex_expect("varbin builder should canonicalize")
630 }
631}
632
633#[macro_export]
646macro_rules! match_each_varbin_builder {
647 ($builder:expr, | $typed:ident | $body:expr) => {
648 $crate::__match_varbin_builder_widths!($builder, |$typed| $body, [u32, u64, i32, i64])
649 };
650}
651
652#[doc(hidden)]
655#[macro_export]
656macro_rules! __match_varbin_builder_widths {
657 ($builder:expr, | $typed:ident | $body:expr, [$($width:ty),+ $(,)?]) => {{
658 let __varbin_builder: &mut dyn $crate::builders::ArrayBuilder = $builder;
659 $crate::__match_varbin_builder_arms!(__varbin_builder, |$typed| $body, [$($width),+])
660 }};
661}
662
663#[doc(hidden)]
665#[macro_export]
666macro_rules! __match_varbin_builder_arms {
667 ($builder:expr, | $typed:ident | $body:expr, []) => {
668 None
669 };
670 ($builder:expr, | $typed:ident | $body:expr, [$head:ty $(, $tail:ty)*]) => {
671 if $builder
672 .as_any()
673 .is::<$crate::builders::VarBinBuilder<$head>>()
674 {
675 let $typed = match $builder
676 .as_any_mut()
677 .downcast_mut::<$crate::builders::VarBinBuilder<$head>>()
678 {
679 Some(typed) => typed,
680 None => unreachable!("builder type checked above"),
681 };
682 Some($body)
683 } else {
684 $crate::__match_varbin_builder_arms!($builder, |$typed| $body, [$($tail),*])
685 }
686 };
687}
688
689#[inline]
692fn prefix_sums<P: AsPrimitive<usize>>(lengths: &[P]) -> impl Iterator<Item = usize> {
693 lengths.iter().scan(0usize, |end, length| {
694 *end = end.wrapping_add(length.as_());
695 Some(*end)
696 })
697}
698
699#[cfg(test)]
700mod tests {
701 use std::mem::MaybeUninit;
702
703 use rstest::rstest;
704 use vortex_error::VortexResult;
705 use vortex_mask::Mask;
706
707 use crate::IntoArray;
708 use crate::VortexSessionExecute;
709 use crate::array_session;
710 use crate::arrays::ChunkedArray;
711 use crate::arrays::ConstantArray;
712 use crate::arrays::VarBinArray;
713 use crate::arrays::VarBinViewArray;
714 use crate::arrays::varbin::VarBinArraySlotsExt;
715 use crate::arrays::varbin::builder::VarBinBuilder;
716 use crate::assert_arrays_eq;
717 use crate::builders::ArrayBuilder;
718 use crate::dtype::DType;
719 use crate::dtype::Nullability::Nullable;
720 use crate::expr::stats::Precision;
721 use crate::expr::stats::Stat;
722 use crate::expr::stats::StatsProviderExt;
723 use crate::scalar::Scalar;
724
725 #[test]
726 fn test_builder() {
727 let mut builder = VarBinBuilder::<i32>::with_capacity(DType::Utf8(Nullable), 0);
728 builder.append(Some(b"hello"));
729 builder.append(None);
730 builder.append(Some(b"world"));
731 let array = builder.finish_into_varbin();
732
733 assert_eq!(array.len(), 3);
734 assert_eq!(array.dtype().nullability(), Nullable);
735 assert_eq!(
736 array
737 .execute_scalar(0, &mut array_session().create_execution_ctx())
738 .unwrap(),
739 Scalar::utf8("hello".to_string(), Nullable)
740 );
741 assert!(
742 array
743 .execute_scalar(1, &mut array_session().create_execution_ctx())
744 .unwrap()
745 .is_null()
746 );
747 }
748
749 #[rstest]
750 #[case(false)]
751 #[case(true)]
752 fn test_append_varbin_to_builder(#[case] large_offsets: bool) -> VortexResult<()> {
753 let source = VarBinArray::from_iter(
754 [
755 Some("prefix"),
756 Some("hello"),
757 None,
758 Some("world"),
759 Some("suffix"),
760 ],
761 DType::Utf8(Nullable),
762 )
763 .into_array()
764 .slice(1..4)?;
765 let mut ctx = array_session().create_execution_ctx();
766
767 let actual = with_offsets(large_offsets, source.dtype().clone(), |builder| {
768 source.append_to_builder(builder, &mut ctx)
769 })?;
770
771 assert_arrays_eq!(actual, source, &mut ctx);
772 Ok(())
773 }
774
775 #[test]
776 fn append_n_values_offset_overflow_returns_error() {
777 let mut builder = VarBinBuilder::<i32>::new(DType::Utf8(Nullable));
778
779 let result = builder.append_n_values(b"hello", i32::MAX as usize / 5 + 1);
781
782 assert!(result.is_err());
783 assert_eq!(builder.offsets.len(), 1);
784 assert!(builder.data.is_empty());
785 assert_eq!(builder.validity.len(), 0);
786 }
787
788 #[test]
789 fn append_values_rejects_a_short_offset_count() {
790 let mut builder = VarBinBuilder::<i32>::new(DType::Utf8(Nullable));
791
792 let result = builder.append_values(b"ab", [1usize, 2].into_iter(), &Mask::new_true(3));
793
794 assert!(result.is_err());
795 assert_eq!(builder.offsets.len(), 1);
796 assert!(builder.data.is_empty());
797 }
798
799 #[test]
800 fn append_values_rejects_non_monotonic_offsets() {
801 let mut builder = VarBinBuilder::<i32>::new(DType::Utf8(Nullable));
802
803 let result = builder.append_values(b"ab", [2usize, 1].into_iter(), &Mask::new_true(2));
804
805 assert!(result.is_err());
806 assert_eq!(builder.offsets.len(), 1);
807 }
808
809 #[test]
810 fn append_decoded_writes_into_the_builder_storage() -> VortexResult<()> {
811 let mut ctx = array_session().create_execution_ctx();
812 let mut builder = VarBinBuilder::<i32>::new(DType::Utf8(Nullable));
813
814 unsafe {
816 builder.append_decoded(
817 6,
818 4,
819 &[3usize, 0, 3],
820 &Mask::from_iter([true, false, true]),
821 &mut |spare: &mut [MaybeUninit<u8>]| {
822 for (slot, byte) in spare.iter_mut().zip(b"foobar") {
823 slot.write(*byte);
824 }
825 Ok(6)
826 },
827 )?;
828 }
829
830 let expected =
831 VarBinViewArray::from_iter([Some("foo"), None, Some("bar")], DType::Utf8(Nullable));
832 assert_arrays_eq!(builder.finish_into_varbin(), expected, &mut ctx);
833 Ok(())
834 }
835
836 #[test]
839 fn append_decoded_rejects_an_offset_overflow_without_decoding() {
840 let mut builder = VarBinBuilder::<i32>::new(DType::Utf8(Nullable));
841 let num_bytes = i32::MAX as usize + 1;
842 let mut decoded = false;
843
844 let result = unsafe {
846 builder.append_decoded(num_bytes, 0, &[num_bytes], &Mask::new_true(1), &mut |_| {
847 decoded = true;
848 Ok(0)
849 })
850 };
851
852 assert!(result.is_err());
853 assert!(!decoded, "decode ran before the offset width was checked");
854 assert_eq!(builder.offsets.len(), 1);
855 assert_eq!(builder.validity.len(), 0);
856 }
857
858 #[test]
861 fn append_scalar_repeated_rejects_an_offset_overflow() {
862 let mut builder = VarBinBuilder::<i32>::new(DType::Utf8(Nullable));
863
864 let result = builder
865 .append_scalar_repeated(&Scalar::utf8("hello", Nullable), i32::MAX as usize / 5 + 1);
866
867 assert!(result.is_err());
868 assert_eq!(builder.offsets.len(), 1);
869 assert!(builder.data.is_empty());
870 assert_eq!(builder.validity.len(), 0);
871 }
872
873 #[test]
874 fn append_decoded_rejects_a_short_decode() {
875 let mut builder = VarBinBuilder::<i32>::new(DType::Utf8(Nullable));
876
877 let result = unsafe {
880 builder.append_decoded(6, 0, &[3usize, 3], &Mask::new_true(2), &mut |spare| {
881 spare[..3].fill(MaybeUninit::new(b'x'));
882 Ok(3)
883 })
884 };
885
886 assert!(result.is_err());
887 assert_eq!(builder.offsets.len(), 1);
888 assert_eq!(builder.validity.len(), 0);
889 }
890
891 #[rstest]
895 #[case::all_valid(Mask::new_true(2))]
896 #[case::some_valid(Mask::from_iter([true, false, true]))]
897 fn append_valid_slices_rejects_a_byte_count_mismatch(#[case] validity: Mask) {
898 let mut builder = VarBinBuilder::<i32>::new(DType::Utf8(Nullable));
899 let values = [b"foo".as_slice(), b"quux".as_slice()];
900
901 let mut next = 0;
902 let result = builder.append_valid_slices(6, &validity, |_| {
903 next += 1;
904 values[next - 1]
905 });
906
907 assert!(result.is_err());
908 assert_eq!(builder.offsets.len(), 1);
909 assert!(builder.data.is_empty());
910 assert_eq!(builder.validity.len(), 0);
911 }
912
913 #[test]
914 #[should_panic(expected = "The offset count must be one more than the validity length")]
915 fn finish_rejects_mismatched_validity() {
916 let mut builder = VarBinBuilder::<i32>::new(DType::Utf8(Nullable));
917 builder.validity.append_true();
918 drop(builder.finish_into_varbin());
919 }
920
921 #[rstest]
922 #[case(false)]
923 #[case(true)]
924 fn test_array_builder_methods(#[case] large_offsets: bool) -> VortexResult<()> {
925 let mut ctx = array_session().create_execution_ctx();
926 let result = with_offsets(large_offsets, DType::Utf8(Nullable), |builder| {
927 builder.reserve_exact(3);
928 builder.append_zero();
929 builder.append_scalar(&Scalar::utf8("hello", Nullable))?;
930 builder.append_null();
931 assert_eq!(builder.len(), 3);
932 Ok(())
933 })?;
934
935 assert_eq!(
936 result.validity()?.execute_mask(3, &mut ctx)?,
937 Mask::from_iter([true, true, false])
938 );
939 Ok(())
940 }
941
942 #[rstest]
943 #[case(false)]
944 #[case(true)]
945 fn test_append_varbinview_validity_to_builder(#[case] large_offsets: bool) -> VortexResult<()> {
946 let long = "a value that does not fit inline";
947 let all_null = VarBinViewArray::from_iter([None::<&str>, None], DType::Utf8(Nullable));
948 let mixed =
949 VarBinViewArray::from_iter([Some("hello"), None, Some(long)], DType::Utf8(Nullable));
950 let expected = VarBinViewArray::from_iter(
951 [None, None, Some("hello"), None, Some(long)],
952 DType::Utf8(Nullable),
953 );
954 let mut ctx = array_session().create_execution_ctx();
955
956 let actual = with_offsets(large_offsets, expected.dtype().clone(), |builder| {
957 all_null
958 .clone()
959 .into_array()
960 .append_to_builder(builder, &mut ctx)?;
961 mixed
962 .clone()
963 .into_array()
964 .append_to_builder(builder, &mut ctx)
965 })?;
966
967 assert_arrays_eq!(actual, expected, &mut ctx);
968 Ok(())
969 }
970
971 #[rstest]
976 #[case::u32(VarBinBuilder::<u32>::new(DType::Utf8(Nullable)))]
977 #[case::u64(VarBinBuilder::<u64>::new(DType::Utf8(Nullable)))]
978 #[case::i32(VarBinBuilder::<i32>::new(DType::Utf8(Nullable)))]
979 #[case::i64(VarBinBuilder::<i64>::new(DType::Utf8(Nullable)))]
980 fn append_to_every_offset_width(#[case] mut builder: impl ArrayBuilder) -> VortexResult<()> {
981 let mut ctx = array_session().create_execution_ctx();
982 let long = "a string that is far too long to be inlined in a view";
983 let values = [Some("hello"), None, Some(long), Some("")];
984
985 let view = VarBinViewArray::from_iter(values, DType::Utf8(Nullable)).into_array();
986 let varbin = VarBinArray::from_iter(values, DType::Utf8(Nullable)).into_array();
987 let constant = ConstantArray::new(Scalar::from("hello").into_nullable(), 2).into_array();
988
989 for chunk in [&view, &varbin, &constant] {
990 chunk.append_to_builder(&mut builder, &mut ctx)?;
991 }
992
993 let expected = ChunkedArray::try_new(vec![view, varbin, constant], DType::Utf8(Nullable))?;
994 assert_arrays_eq!(builder.finish(), expected, &mut ctx);
995 Ok(())
996 }
997
998 #[test]
999 fn offsets_have_is_sorted_stat() -> VortexResult<()> {
1000 let mut builder = VarBinBuilder::<i32>::with_capacity(DType::Utf8(Nullable), 0);
1001 builder.append_value(b"aaa");
1002 builder.push_null();
1003 builder.append_value(b"bbb");
1004 let array = builder.finish_into_varbin();
1005
1006 let is_sorted = array
1007 .offsets()
1008 .statistics()
1009 .with_typed_stats_set(|s| s.get_as::<bool>(Stat::IsSorted));
1010 assert_eq!(is_sorted, Precision::Exact(true));
1011 Ok(())
1012 }
1013
1014 #[test]
1015 fn empty_builder_offsets_have_is_sorted_stat() -> VortexResult<()> {
1016 let mut builder = VarBinBuilder::<i32>::new(DType::Utf8(Nullable));
1017 let array = builder.finish_into_varbin();
1018
1019 let is_sorted = array
1020 .offsets()
1021 .statistics()
1022 .with_typed_stats_set(|s| s.get_as::<bool>(Stat::IsSorted));
1023 assert_eq!(is_sorted, Precision::Exact(true));
1024 Ok(())
1025 }
1026
1027 fn with_offsets(
1029 large_offsets: bool,
1030 dtype: DType,
1031 f: impl FnOnce(&mut dyn ArrayBuilder) -> VortexResult<()>,
1032 ) -> VortexResult<VarBinArray> {
1033 if large_offsets {
1034 let mut builder = VarBinBuilder::<i64>::with_capacity(dtype, 8);
1035 f(&mut builder)?;
1036 Ok(builder.finish_into_varbin())
1037 } else {
1038 let mut builder = VarBinBuilder::<i32>::with_capacity(dtype, 8);
1039 f(&mut builder)?;
1040 Ok(builder.finish_into_varbin())
1041 }
1042 }
1043}