1use crate::array::{get_offsets, print_long_array};
19use crate::builder::GenericByteBuilder;
20use crate::iterator::ArrayIter;
21use crate::types::bytes::ByteArrayNativeType;
22use crate::types::ByteArrayType;
23use crate::{Array, ArrayAccessor, ArrayRef, OffsetSizeTrait, Scalar};
24use arrow_buffer::{ArrowNativeType, Buffer, MutableBuffer};
25use arrow_buffer::{NullBuffer, OffsetBuffer};
26use arrow_data::{ArrayData, ArrayDataBuilder};
27use arrow_schema::{ArrowError, DataType};
28use std::any::Any;
29use std::sync::Arc;
30
31pub struct GenericByteArray<T: ByteArrayType> {
88 data_type: DataType,
89 value_offsets: OffsetBuffer<T::Offset>,
90 value_data: Buffer,
91 nulls: Option<NullBuffer>,
92}
93
94impl<T: ByteArrayType> Clone for GenericByteArray<T> {
95 fn clone(&self) -> Self {
96 Self {
97 data_type: T::DATA_TYPE,
98 value_offsets: self.value_offsets.clone(),
99 value_data: self.value_data.clone(),
100 nulls: self.nulls.clone(),
101 }
102 }
103}
104
105impl<T: ByteArrayType> GenericByteArray<T> {
106 pub const DATA_TYPE: DataType = T::DATA_TYPE;
108
109 pub fn new(
115 offsets: OffsetBuffer<T::Offset>,
116 values: Buffer,
117 nulls: Option<NullBuffer>,
118 ) -> Self {
119 Self::try_new(offsets, values, nulls).unwrap()
120 }
121
122 pub fn try_new(
129 offsets: OffsetBuffer<T::Offset>,
130 values: Buffer,
131 nulls: Option<NullBuffer>,
132 ) -> Result<Self, ArrowError> {
133 let len = offsets.len() - 1;
134
135 T::validate(&offsets, &values)?;
137
138 if let Some(n) = nulls.as_ref() {
139 if n.len() != len {
140 return Err(ArrowError::InvalidArgumentError(format!(
141 "Incorrect length of null buffer for {}{}Array, expected {len} got {}",
142 T::Offset::PREFIX,
143 T::PREFIX,
144 n.len(),
145 )));
146 }
147 }
148
149 Ok(Self {
150 data_type: T::DATA_TYPE,
151 value_offsets: offsets,
152 value_data: values,
153 nulls,
154 })
155 }
156
157 pub unsafe fn new_unchecked(
163 offsets: OffsetBuffer<T::Offset>,
164 values: Buffer,
165 nulls: Option<NullBuffer>,
166 ) -> Self {
167 if cfg!(feature = "force_validate") {
168 return Self::new(offsets, values, nulls);
169 }
170 Self {
171 data_type: T::DATA_TYPE,
172 value_offsets: offsets,
173 value_data: values,
174 nulls,
175 }
176 }
177
178 pub fn new_null(len: usize) -> Self {
180 Self {
181 data_type: T::DATA_TYPE,
182 value_offsets: OffsetBuffer::new_zeroed(len),
183 value_data: MutableBuffer::new(0).into(),
184 nulls: Some(NullBuffer::new_null(len)),
185 }
186 }
187
188 pub fn new_scalar(value: impl AsRef<T::Native>) -> Scalar<Self> {
190 Scalar::new(Self::from_iter_values(std::iter::once(value)))
191 }
192
193 pub fn from_iter_values<Ptr, I>(iter: I) -> Self
195 where
196 Ptr: AsRef<T::Native>,
197 I: IntoIterator<Item = Ptr>,
198 {
199 let iter = iter.into_iter();
200 let (_, data_len) = iter.size_hint();
201 let data_len = data_len.expect("Iterator must be sized"); let mut offsets = MutableBuffer::new((data_len + 1) * std::mem::size_of::<T::Offset>());
204 offsets.push(T::Offset::usize_as(0));
205
206 let mut values = MutableBuffer::new(0);
207 for s in iter {
208 let s: &[u8] = s.as_ref().as_ref();
209 values.extend_from_slice(s);
210 offsets.push(T::Offset::usize_as(values.len()));
211 }
212
213 T::Offset::from_usize(values.len()).expect("offset overflow");
214 let offsets = Buffer::from(offsets);
215
216 let value_offsets = unsafe { OffsetBuffer::new_unchecked(offsets.into()) };
218
219 Self {
220 data_type: T::DATA_TYPE,
221 value_data: values.into(),
222 value_offsets,
223 nulls: None,
224 }
225 }
226
227 pub fn into_parts(self) -> (OffsetBuffer<T::Offset>, Buffer, Option<NullBuffer>) {
229 (self.value_offsets, self.value_data, self.nulls)
230 }
231
232 #[inline]
236 pub fn value_length(&self, i: usize) -> T::Offset {
237 let offsets = self.value_offsets();
238 offsets[i + 1] - offsets[i]
239 }
240
241 #[inline]
246 pub fn offsets(&self) -> &OffsetBuffer<T::Offset> {
247 &self.value_offsets
248 }
249
250 #[inline]
255 pub fn values(&self) -> &Buffer {
256 &self.value_data
257 }
258
259 pub fn value_data(&self) -> &[u8] {
261 self.value_data.as_slice()
262 }
263
264 pub fn is_ascii(&self) -> bool {
266 let offsets = self.value_offsets();
267 let start = offsets.first().unwrap();
268 let end = offsets.last().unwrap();
269 self.value_data()[start.as_usize()..end.as_usize()].is_ascii()
270 }
271
272 #[inline]
274 pub fn value_offsets(&self) -> &[T::Offset] {
275 &self.value_offsets
276 }
277
278 pub unsafe fn value_unchecked(&self, i: usize) -> &T::Native {
286 let end = *self.value_offsets().get_unchecked(i + 1);
287 let start = *self.value_offsets().get_unchecked(i);
288
289 let b = std::slice::from_raw_parts(
299 self.value_data
300 .as_ptr()
301 .offset(start.to_isize().unwrap_unchecked()),
302 (end - start).to_usize().unwrap_unchecked(),
303 );
304
305 T::Native::from_bytes_unchecked(b)
308 }
309
310 pub fn value(&self, i: usize) -> &T::Native {
318 assert!(
319 i < self.len(),
320 "Trying to access an element at index {} from a {}{}Array of length {}",
321 i,
322 T::Offset::PREFIX,
323 T::PREFIX,
324 self.len()
325 );
326 unsafe { self.value_unchecked(i) }
329 }
330
331 pub fn iter(&self) -> ArrayIter<&Self> {
333 ArrayIter::new(self)
334 }
335
336 pub fn slice(&self, offset: usize, length: usize) -> Self {
338 Self {
339 data_type: T::DATA_TYPE,
340 value_offsets: self.value_offsets.slice(offset, length),
341 value_data: self.value_data.clone(),
342 nulls: self.nulls.as_ref().map(|n| n.slice(offset, length)),
343 }
344 }
345
346 pub fn into_builder(self) -> Result<GenericByteBuilder<T>, Self> {
349 let len = self.len();
350 let value_len = T::Offset::as_usize(self.value_offsets()[len] - self.value_offsets()[0]);
351
352 let data = self.into_data();
353 let null_bit_buffer = data.nulls().map(|b| b.inner().sliced());
354
355 let element_len = std::mem::size_of::<T::Offset>();
356 let offset_buffer = data.buffers()[0]
357 .slice_with_length(data.offset() * element_len, (len + 1) * element_len);
358
359 let element_len = std::mem::size_of::<u8>();
360 let value_buffer = data.buffers()[1]
361 .slice_with_length(data.offset() * element_len, value_len * element_len);
362
363 drop(data);
364
365 let try_mutable_null_buffer = match null_bit_buffer {
366 None => Ok(None),
367 Some(null_buffer) => {
368 null_buffer.into_mutable().map(Some)
370 }
371 };
372
373 let try_mutable_buffers = match try_mutable_null_buffer {
374 Ok(mutable_null_buffer) => {
375 let try_mutable_offset_buffer = offset_buffer.into_mutable();
377 let try_mutable_value_buffer = value_buffer.into_mutable();
378
379 match (try_mutable_offset_buffer, try_mutable_value_buffer) {
382 (Ok(mutable_offset_buffer), Ok(mutable_value_buffer)) => unsafe {
383 Ok(GenericByteBuilder::<T>::new_from_buffer(
384 mutable_offset_buffer,
385 mutable_value_buffer,
386 mutable_null_buffer,
387 ))
388 },
389 (Ok(mutable_offset_buffer), Err(value_buffer)) => Err((
390 mutable_offset_buffer.into(),
391 value_buffer,
392 mutable_null_buffer.map(|b| b.into()),
393 )),
394 (Err(offset_buffer), Ok(mutable_value_buffer)) => Err((
395 offset_buffer,
396 mutable_value_buffer.into(),
397 mutable_null_buffer.map(|b| b.into()),
398 )),
399 (Err(offset_buffer), Err(value_buffer)) => Err((
400 offset_buffer,
401 value_buffer,
402 mutable_null_buffer.map(|b| b.into()),
403 )),
404 }
405 }
406 Err(mutable_null_buffer) => {
407 Err((offset_buffer, value_buffer, Some(mutable_null_buffer)))
409 }
410 };
411
412 match try_mutable_buffers {
413 Ok(builder) => Ok(builder),
414 Err((offset_buffer, value_buffer, null_bit_buffer)) => {
415 let builder = ArrayData::builder(T::DATA_TYPE)
416 .len(len)
417 .add_buffer(offset_buffer)
418 .add_buffer(value_buffer)
419 .null_bit_buffer(null_bit_buffer);
420
421 let array_data = unsafe { builder.build_unchecked() };
422 let array = GenericByteArray::<T>::from(array_data);
423
424 Err(array)
425 }
426 }
427 }
428}
429
430impl<T: ByteArrayType> std::fmt::Debug for GenericByteArray<T> {
431 fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
432 write!(f, "{}{}Array\n[\n", T::Offset::PREFIX, T::PREFIX)?;
433 print_long_array(self, f, |array, index, f| {
434 std::fmt::Debug::fmt(&array.value(index), f)
435 })?;
436 write!(f, "]")
437 }
438}
439
440impl<T: ByteArrayType> Array for GenericByteArray<T> {
441 fn as_any(&self) -> &dyn Any {
442 self
443 }
444
445 fn to_data(&self) -> ArrayData {
446 self.clone().into()
447 }
448
449 fn into_data(self) -> ArrayData {
450 self.into()
451 }
452
453 fn data_type(&self) -> &DataType {
454 &self.data_type
455 }
456
457 fn slice(&self, offset: usize, length: usize) -> ArrayRef {
458 Arc::new(self.slice(offset, length))
459 }
460
461 fn len(&self) -> usize {
462 self.value_offsets.len() - 1
463 }
464
465 fn is_empty(&self) -> bool {
466 self.value_offsets.len() <= 1
467 }
468
469 fn shrink_to_fit(&mut self) {
470 self.value_offsets.shrink_to_fit();
471 self.value_data.shrink_to_fit();
472 if let Some(nulls) = &mut self.nulls {
473 nulls.shrink_to_fit();
474 }
475 }
476
477 fn offset(&self) -> usize {
478 0
479 }
480
481 fn nulls(&self) -> Option<&NullBuffer> {
482 self.nulls.as_ref()
483 }
484
485 fn logical_null_count(&self) -> usize {
486 self.null_count()
488 }
489
490 fn get_buffer_memory_size(&self) -> usize {
491 let mut sum = self.value_offsets.inner().inner().capacity();
492 sum += self.value_data.capacity();
493 if let Some(x) = &self.nulls {
494 sum += x.buffer().capacity()
495 }
496 sum
497 }
498
499 fn get_array_memory_size(&self) -> usize {
500 std::mem::size_of::<Self>() + self.get_buffer_memory_size()
501 }
502}
503
504impl<'a, T: ByteArrayType> ArrayAccessor for &'a GenericByteArray<T> {
505 type Item = &'a T::Native;
506
507 fn value(&self, index: usize) -> Self::Item {
508 GenericByteArray::value(self, index)
509 }
510
511 unsafe fn value_unchecked(&self, index: usize) -> Self::Item {
512 GenericByteArray::value_unchecked(self, index)
513 }
514}
515
516impl<T: ByteArrayType> From<ArrayData> for GenericByteArray<T> {
517 fn from(data: ArrayData) -> Self {
518 assert_eq!(
519 data.data_type(),
520 &Self::DATA_TYPE,
521 "{}{}Array expects DataType::{}",
522 T::Offset::PREFIX,
523 T::PREFIX,
524 Self::DATA_TYPE
525 );
526 assert_eq!(
527 data.buffers().len(),
528 2,
529 "{}{}Array data should contain 2 buffers only (offsets and values)",
530 T::Offset::PREFIX,
531 T::PREFIX,
532 );
533 let value_offsets = unsafe { get_offsets(&data) };
536 let value_data = data.buffers()[1].clone();
537 Self {
538 value_offsets,
539 value_data,
540 data_type: T::DATA_TYPE,
541 nulls: data.nulls().cloned(),
542 }
543 }
544}
545
546impl<T: ByteArrayType> From<GenericByteArray<T>> for ArrayData {
547 fn from(array: GenericByteArray<T>) -> Self {
548 let len = array.len();
549
550 let offsets = array.value_offsets.into_inner().into_inner();
551 let builder = ArrayDataBuilder::new(array.data_type)
552 .len(len)
553 .buffers(vec![offsets, array.value_data])
554 .nulls(array.nulls);
555
556 unsafe { builder.build_unchecked() }
557 }
558}
559
560impl<'a, T: ByteArrayType> IntoIterator for &'a GenericByteArray<T> {
561 type Item = Option<&'a T::Native>;
562 type IntoIter = ArrayIter<Self>;
563
564 fn into_iter(self) -> Self::IntoIter {
565 ArrayIter::new(self)
566 }
567}
568
569impl<'a, Ptr, T: ByteArrayType> FromIterator<&'a Option<Ptr>> for GenericByteArray<T>
570where
571 Ptr: AsRef<T::Native> + 'a,
572{
573 fn from_iter<I: IntoIterator<Item = &'a Option<Ptr>>>(iter: I) -> Self {
574 iter.into_iter()
575 .map(|o| o.as_ref().map(|p| p.as_ref()))
576 .collect()
577 }
578}
579
580impl<Ptr, T: ByteArrayType> FromIterator<Option<Ptr>> for GenericByteArray<T>
581where
582 Ptr: AsRef<T::Native>,
583{
584 fn from_iter<I: IntoIterator<Item = Option<Ptr>>>(iter: I) -> Self {
585 let iter = iter.into_iter();
586 let mut builder = GenericByteBuilder::with_capacity(iter.size_hint().0, 1024);
587 builder.extend(iter);
588 builder.finish()
589 }
590}
591
592#[cfg(test)]
593mod tests {
594 use crate::{BinaryArray, StringArray};
595 use arrow_buffer::{Buffer, NullBuffer, OffsetBuffer};
596
597 #[test]
598 fn try_new() {
599 let data = Buffer::from_slice_ref("helloworld");
600 let offsets = OffsetBuffer::new(vec![0, 5, 10].into());
601 StringArray::new(offsets.clone(), data.clone(), None);
602
603 let nulls = NullBuffer::new_null(3);
604 let err =
605 StringArray::try_new(offsets.clone(), data.clone(), Some(nulls.clone())).unwrap_err();
606 assert_eq!(err.to_string(), "Invalid argument error: Incorrect length of null buffer for StringArray, expected 2 got 3");
607
608 let err = BinaryArray::try_new(offsets.clone(), data.clone(), Some(nulls)).unwrap_err();
609 assert_eq!(err.to_string(), "Invalid argument error: Incorrect length of null buffer for BinaryArray, expected 2 got 3");
610
611 let non_utf8_data = Buffer::from_slice_ref(b"he\xFFloworld");
612 let err = StringArray::try_new(offsets.clone(), non_utf8_data.clone(), None).unwrap_err();
613 assert_eq!(err.to_string(), "Invalid argument error: Encountered non UTF-8 data: invalid utf-8 sequence of 1 bytes from index 2");
614
615 BinaryArray::new(offsets, non_utf8_data, None);
616
617 let offsets = OffsetBuffer::new(vec![0, 5, 11].into());
618 let err = StringArray::try_new(offsets.clone(), data.clone(), None).unwrap_err();
619 assert_eq!(
620 err.to_string(),
621 "Invalid argument error: Offset of 11 exceeds length of values 10"
622 );
623
624 let err = BinaryArray::try_new(offsets.clone(), data, None).unwrap_err();
625 assert_eq!(
626 err.to_string(),
627 "Invalid argument error: Maximum offset of 11 is larger than values of length 10"
628 );
629
630 let non_ascii_data = Buffer::from_slice_ref("heìloworld");
631 StringArray::new(offsets.clone(), non_ascii_data.clone(), None);
632 BinaryArray::new(offsets, non_ascii_data.clone(), None);
633
634 let offsets = OffsetBuffer::new(vec![0, 3, 10].into());
635 let err = StringArray::try_new(offsets.clone(), non_ascii_data.clone(), None).unwrap_err();
636 assert_eq!(
637 err.to_string(),
638 "Invalid argument error: Split UTF-8 codepoint at offset 3"
639 );
640
641 BinaryArray::new(offsets, non_ascii_data, None);
642 }
643}