arrow_array/array/
byte_array.rs

1// Licensed to the Apache Software Foundation (ASF) under one
2// or more contributor license agreements.  See the NOTICE file
3// distributed with this work for additional information
4// regarding copyright ownership.  The ASF licenses this file
5// to you under the Apache License, Version 2.0 (the
6// "License"); you may not use this file except in compliance
7// with the License.  You may obtain a copy of the License at
8//
9//   http://www.apache.org/licenses/LICENSE-2.0
10//
11// Unless required by applicable law or agreed to in writing,
12// software distributed under the License is distributed on an
13// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14// KIND, either express or implied.  See the License for the
15// specific language governing permissions and limitations
16// under the License.
17
18use 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
31/// An array of [variable length byte arrays](https://arrow.apache.org/docs/format/Columnar.html#variable-size-binary-layout)
32///
33/// See [`StringArray`] and [`LargeStringArray`] for storing utf8 encoded string data
34///
35/// See [`BinaryArray`] and [`LargeBinaryArray`] for storing arbitrary bytes
36///
37/// # Example: From a Vec
38///
39/// ```
40/// # use arrow_array::{Array, GenericByteArray, types::Utf8Type};
41/// let arr: GenericByteArray<Utf8Type> = vec!["hello", "world", ""].into();
42/// assert_eq!(arr.value_data(), b"helloworld");
43/// assert_eq!(arr.value_offsets(), &[0, 5, 10, 10]);
44/// let values: Vec<_> = arr.iter().collect();
45/// assert_eq!(values, &[Some("hello"), Some("world"), Some("")]);
46/// ```
47///
48/// # Example: From an optional Vec
49///
50/// ```
51/// # use arrow_array::{Array, GenericByteArray, types::Utf8Type};
52/// let arr: GenericByteArray<Utf8Type> = vec![Some("hello"), Some("world"), Some(""), None].into();
53/// assert_eq!(arr.value_data(), b"helloworld");
54/// assert_eq!(arr.value_offsets(), &[0, 5, 10, 10, 10]);
55/// let values: Vec<_> = arr.iter().collect();
56/// assert_eq!(values, &[Some("hello"), Some("world"), Some(""), None]);
57/// ```
58///
59/// # Example: From an iterator of option
60///
61/// ```
62/// # use arrow_array::{Array, GenericByteArray, types::Utf8Type};
63/// let arr: GenericByteArray<Utf8Type> = (0..5).map(|x| (x % 2 == 0).then(|| x.to_string())).collect();
64/// let values: Vec<_> = arr.iter().collect();
65/// assert_eq!(values, &[Some("0"), None, Some("2"), None, Some("4")]);
66/// ```
67///
68/// # Example: Using Builder
69///
70/// ```
71/// # use arrow_array::Array;
72/// # use arrow_array::builder::GenericByteBuilder;
73/// # use arrow_array::types::Utf8Type;
74/// let mut builder = GenericByteBuilder::<Utf8Type>::new();
75/// builder.append_value("hello");
76/// builder.append_null();
77/// builder.append_value("world");
78/// let array = builder.finish();
79/// let values: Vec<_> = array.iter().collect();
80/// assert_eq!(values, &[Some("hello"), None, Some("world")]);
81/// ```
82///
83/// [`StringArray`]: crate::StringArray
84/// [`LargeStringArray`]: crate::LargeStringArray
85/// [`BinaryArray`]: crate::BinaryArray
86/// [`LargeBinaryArray`]: crate::LargeBinaryArray
87pub 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    /// Data type of the array.
107    pub const DATA_TYPE: DataType = T::DATA_TYPE;
108
109    /// Create a new [`GenericByteArray`] from the provided parts, panicking on failure
110    ///
111    /// # Panics
112    ///
113    /// Panics if [`GenericByteArray::try_new`] returns an error
114    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    /// Create a new [`GenericByteArray`] from the provided parts, returning an error on failure
123    ///
124    /// # Errors
125    ///
126    /// * `offsets.len() - 1 != nulls.len()`
127    /// * Any consecutive pair of `offsets` does not denote a valid slice of `values`
128    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        // Verify that each pair of offsets is a valid slices of values
136        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    /// Create a new [`GenericByteArray`] from the provided parts, without validation
158    ///
159    /// # Safety
160    ///
161    /// Safe if [`Self::try_new`] would not error
162    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    /// Create a new [`GenericByteArray`] of length `len` where all values are null
179    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    /// Create a new [`Scalar`] from `v`
189    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    /// Creates a [`GenericByteArray`] based on an iterator of values without nulls
194    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"); // panic if no upper bound.
202
203        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        // Safety: valid by construction
217        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    /// Deconstruct this array into its constituent parts
228    pub fn into_parts(self) -> (OffsetBuffer<T::Offset>, Buffer, Option<NullBuffer>) {
229        (self.value_offsets, self.value_data, self.nulls)
230    }
231
232    /// Returns the length for value at index `i`.
233    /// # Panics
234    /// Panics if index `i` is out of bounds.
235    #[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    /// Returns a reference to the offsets of this array
242    ///
243    /// Unlike [`Self::value_offsets`] this returns the [`OffsetBuffer`]
244    /// allowing for zero-copy cloning
245    #[inline]
246    pub fn offsets(&self) -> &OffsetBuffer<T::Offset> {
247        &self.value_offsets
248    }
249
250    /// Returns the values of this array
251    ///
252    /// Unlike [`Self::value_data`] this returns the [`Buffer`]
253    /// allowing for zero-copy cloning
254    #[inline]
255    pub fn values(&self) -> &Buffer {
256        &self.value_data
257    }
258
259    /// Returns the raw value data
260    pub fn value_data(&self) -> &[u8] {
261        self.value_data.as_slice()
262    }
263
264    /// Returns true if all data within this array is ASCII
265    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    /// Returns the offset values in the offsets buffer
273    #[inline]
274    pub fn value_offsets(&self) -> &[T::Offset] {
275        &self.value_offsets
276    }
277
278    /// Returns the element at index `i`
279    /// # Safety
280    /// Caller is responsible for ensuring that the index is within the bounds of the array
281    pub unsafe fn value_unchecked(&self, i: usize) -> &T::Native {
282        let end = *self.value_offsets().get_unchecked(i + 1);
283        let start = *self.value_offsets().get_unchecked(i);
284
285        // Soundness
286        // pointer alignment & location is ensured by RawPtrBox
287        // buffer bounds/offset is ensured by the value_offset invariants
288
289        // Safety of `to_isize().unwrap()`
290        // `start` and `end` are &OffsetSize, which is a generic type that implements the
291        // OffsetSizeTrait. Currently, only i32 and i64 implement OffsetSizeTrait,
292        // both of which should cleanly cast to isize on an architecture that supports
293        // 32/64-bit offsets
294        let b = std::slice::from_raw_parts(
295            self.value_data
296                .as_ptr()
297                .offset(start.to_isize().unwrap_unchecked()),
298            (end - start).to_usize().unwrap_unchecked(),
299        );
300
301        // SAFETY:
302        // ArrayData is valid
303        T::Native::from_bytes_unchecked(b)
304    }
305
306    /// Returns the element at index `i`
307    /// # Panics
308    /// Panics if index `i` is out of bounds.
309    pub fn value(&self, i: usize) -> &T::Native {
310        assert!(
311            i < self.len(),
312            "Trying to access an element at index {} from a {}{}Array of length {}",
313            i,
314            T::Offset::PREFIX,
315            T::PREFIX,
316            self.len()
317        );
318        // SAFETY:
319        // Verified length above
320        unsafe { self.value_unchecked(i) }
321    }
322
323    /// constructs a new iterator
324    pub fn iter(&self) -> ArrayIter<&Self> {
325        ArrayIter::new(self)
326    }
327
328    /// Returns a zero-copy slice of this array with the indicated offset and length.
329    pub fn slice(&self, offset: usize, length: usize) -> Self {
330        Self {
331            data_type: T::DATA_TYPE,
332            value_offsets: self.value_offsets.slice(offset, length),
333            value_data: self.value_data.clone(),
334            nulls: self.nulls.as_ref().map(|n| n.slice(offset, length)),
335        }
336    }
337
338    /// Returns `GenericByteBuilder` of this byte array for mutating its values if the underlying
339    /// offset and data buffers are not shared by others.
340    pub fn into_builder(self) -> Result<GenericByteBuilder<T>, Self> {
341        let len = self.len();
342        let value_len = T::Offset::as_usize(self.value_offsets()[len] - self.value_offsets()[0]);
343
344        let data = self.into_data();
345        let null_bit_buffer = data.nulls().map(|b| b.inner().sliced());
346
347        let element_len = std::mem::size_of::<T::Offset>();
348        let offset_buffer = data.buffers()[0]
349            .slice_with_length(data.offset() * element_len, (len + 1) * element_len);
350
351        let element_len = std::mem::size_of::<u8>();
352        let value_buffer = data.buffers()[1]
353            .slice_with_length(data.offset() * element_len, value_len * element_len);
354
355        drop(data);
356
357        let try_mutable_null_buffer = match null_bit_buffer {
358            None => Ok(None),
359            Some(null_buffer) => {
360                // Null buffer exists, tries to make it mutable
361                null_buffer.into_mutable().map(Some)
362            }
363        };
364
365        let try_mutable_buffers = match try_mutable_null_buffer {
366            Ok(mutable_null_buffer) => {
367                // Got mutable null buffer, tries to get mutable value buffer
368                let try_mutable_offset_buffer = offset_buffer.into_mutable();
369                let try_mutable_value_buffer = value_buffer.into_mutable();
370
371                // try_mutable_offset_buffer.map(...).map_err(...) doesn't work as the compiler complains
372                // mutable_null_buffer is moved into map closure.
373                match (try_mutable_offset_buffer, try_mutable_value_buffer) {
374                    (Ok(mutable_offset_buffer), Ok(mutable_value_buffer)) => unsafe {
375                        Ok(GenericByteBuilder::<T>::new_from_buffer(
376                            mutable_offset_buffer,
377                            mutable_value_buffer,
378                            mutable_null_buffer,
379                        ))
380                    },
381                    (Ok(mutable_offset_buffer), Err(value_buffer)) => Err((
382                        mutable_offset_buffer.into(),
383                        value_buffer,
384                        mutable_null_buffer.map(|b| b.into()),
385                    )),
386                    (Err(offset_buffer), Ok(mutable_value_buffer)) => Err((
387                        offset_buffer,
388                        mutable_value_buffer.into(),
389                        mutable_null_buffer.map(|b| b.into()),
390                    )),
391                    (Err(offset_buffer), Err(value_buffer)) => Err((
392                        offset_buffer,
393                        value_buffer,
394                        mutable_null_buffer.map(|b| b.into()),
395                    )),
396                }
397            }
398            Err(mutable_null_buffer) => {
399                // Unable to get mutable null buffer
400                Err((offset_buffer, value_buffer, Some(mutable_null_buffer)))
401            }
402        };
403
404        match try_mutable_buffers {
405            Ok(builder) => Ok(builder),
406            Err((offset_buffer, value_buffer, null_bit_buffer)) => {
407                let builder = ArrayData::builder(T::DATA_TYPE)
408                    .len(len)
409                    .add_buffer(offset_buffer)
410                    .add_buffer(value_buffer)
411                    .null_bit_buffer(null_bit_buffer);
412
413                let array_data = unsafe { builder.build_unchecked() };
414                let array = GenericByteArray::<T>::from(array_data);
415
416                Err(array)
417            }
418        }
419    }
420}
421
422impl<T: ByteArrayType> std::fmt::Debug for GenericByteArray<T> {
423    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
424        write!(f, "{}{}Array\n[\n", T::Offset::PREFIX, T::PREFIX)?;
425        print_long_array(self, f, |array, index, f| {
426            std::fmt::Debug::fmt(&array.value(index), f)
427        })?;
428        write!(f, "]")
429    }
430}
431
432impl<T: ByteArrayType> Array for GenericByteArray<T> {
433    fn as_any(&self) -> &dyn Any {
434        self
435    }
436
437    fn to_data(&self) -> ArrayData {
438        self.clone().into()
439    }
440
441    fn into_data(self) -> ArrayData {
442        self.into()
443    }
444
445    fn data_type(&self) -> &DataType {
446        &self.data_type
447    }
448
449    fn slice(&self, offset: usize, length: usize) -> ArrayRef {
450        Arc::new(self.slice(offset, length))
451    }
452
453    fn len(&self) -> usize {
454        self.value_offsets.len() - 1
455    }
456
457    fn is_empty(&self) -> bool {
458        self.value_offsets.len() <= 1
459    }
460
461    fn shrink_to_fit(&mut self) {
462        self.value_offsets.shrink_to_fit();
463        self.value_data.shrink_to_fit();
464        if let Some(nulls) = &mut self.nulls {
465            nulls.shrink_to_fit();
466        }
467    }
468
469    fn offset(&self) -> usize {
470        0
471    }
472
473    fn nulls(&self) -> Option<&NullBuffer> {
474        self.nulls.as_ref()
475    }
476
477    fn logical_null_count(&self) -> usize {
478        // More efficient that the default implementation
479        self.null_count()
480    }
481
482    fn get_buffer_memory_size(&self) -> usize {
483        let mut sum = self.value_offsets.inner().inner().capacity();
484        sum += self.value_data.capacity();
485        if let Some(x) = &self.nulls {
486            sum += x.buffer().capacity()
487        }
488        sum
489    }
490
491    fn get_array_memory_size(&self) -> usize {
492        std::mem::size_of::<Self>() + self.get_buffer_memory_size()
493    }
494}
495
496impl<'a, T: ByteArrayType> ArrayAccessor for &'a GenericByteArray<T> {
497    type Item = &'a T::Native;
498
499    fn value(&self, index: usize) -> Self::Item {
500        GenericByteArray::value(self, index)
501    }
502
503    unsafe fn value_unchecked(&self, index: usize) -> Self::Item {
504        GenericByteArray::value_unchecked(self, index)
505    }
506}
507
508impl<T: ByteArrayType> From<ArrayData> for GenericByteArray<T> {
509    fn from(data: ArrayData) -> Self {
510        assert_eq!(
511            data.data_type(),
512            &Self::DATA_TYPE,
513            "{}{}Array expects DataType::{}",
514            T::Offset::PREFIX,
515            T::PREFIX,
516            Self::DATA_TYPE
517        );
518        assert_eq!(
519            data.buffers().len(),
520            2,
521            "{}{}Array data should contain 2 buffers only (offsets and values)",
522            T::Offset::PREFIX,
523            T::PREFIX,
524        );
525        // SAFETY:
526        // ArrayData is valid, and verified type above
527        let value_offsets = unsafe { get_offsets(&data) };
528        let value_data = data.buffers()[1].clone();
529        Self {
530            value_offsets,
531            value_data,
532            data_type: T::DATA_TYPE,
533            nulls: data.nulls().cloned(),
534        }
535    }
536}
537
538impl<T: ByteArrayType> From<GenericByteArray<T>> for ArrayData {
539    fn from(array: GenericByteArray<T>) -> Self {
540        let len = array.len();
541
542        let offsets = array.value_offsets.into_inner().into_inner();
543        let builder = ArrayDataBuilder::new(array.data_type)
544            .len(len)
545            .buffers(vec![offsets, array.value_data])
546            .nulls(array.nulls);
547
548        unsafe { builder.build_unchecked() }
549    }
550}
551
552impl<'a, T: ByteArrayType> IntoIterator for &'a GenericByteArray<T> {
553    type Item = Option<&'a T::Native>;
554    type IntoIter = ArrayIter<Self>;
555
556    fn into_iter(self) -> Self::IntoIter {
557        ArrayIter::new(self)
558    }
559}
560
561impl<'a, Ptr, T: ByteArrayType> FromIterator<&'a Option<Ptr>> for GenericByteArray<T>
562where
563    Ptr: AsRef<T::Native> + 'a,
564{
565    fn from_iter<I: IntoIterator<Item = &'a Option<Ptr>>>(iter: I) -> Self {
566        iter.into_iter()
567            .map(|o| o.as_ref().map(|p| p.as_ref()))
568            .collect()
569    }
570}
571
572impl<Ptr, T: ByteArrayType> FromIterator<Option<Ptr>> for GenericByteArray<T>
573where
574    Ptr: AsRef<T::Native>,
575{
576    fn from_iter<I: IntoIterator<Item = Option<Ptr>>>(iter: I) -> Self {
577        let iter = iter.into_iter();
578        let mut builder = GenericByteBuilder::with_capacity(iter.size_hint().0, 1024);
579        builder.extend(iter);
580        builder.finish()
581    }
582}
583
584#[cfg(test)]
585mod tests {
586    use crate::{BinaryArray, StringArray};
587    use arrow_buffer::{Buffer, NullBuffer, OffsetBuffer};
588
589    #[test]
590    fn try_new() {
591        let data = Buffer::from_slice_ref("helloworld");
592        let offsets = OffsetBuffer::new(vec![0, 5, 10].into());
593        StringArray::new(offsets.clone(), data.clone(), None);
594
595        let nulls = NullBuffer::new_null(3);
596        let err =
597            StringArray::try_new(offsets.clone(), data.clone(), Some(nulls.clone())).unwrap_err();
598        assert_eq!(err.to_string(), "Invalid argument error: Incorrect length of null buffer for StringArray, expected 2 got 3");
599
600        let err = BinaryArray::try_new(offsets.clone(), data.clone(), Some(nulls)).unwrap_err();
601        assert_eq!(err.to_string(), "Invalid argument error: Incorrect length of null buffer for BinaryArray, expected 2 got 3");
602
603        let non_utf8_data = Buffer::from_slice_ref(b"he\xFFloworld");
604        let err = StringArray::try_new(offsets.clone(), non_utf8_data.clone(), None).unwrap_err();
605        assert_eq!(err.to_string(), "Invalid argument error: Encountered non UTF-8 data: invalid utf-8 sequence of 1 bytes from index 2");
606
607        BinaryArray::new(offsets, non_utf8_data, None);
608
609        let offsets = OffsetBuffer::new(vec![0, 5, 11].into());
610        let err = StringArray::try_new(offsets.clone(), data.clone(), None).unwrap_err();
611        assert_eq!(
612            err.to_string(),
613            "Invalid argument error: Offset of 11 exceeds length of values 10"
614        );
615
616        let err = BinaryArray::try_new(offsets.clone(), data, None).unwrap_err();
617        assert_eq!(
618            err.to_string(),
619            "Invalid argument error: Maximum offset of 11 is larger than values of length 10"
620        );
621
622        let non_ascii_data = Buffer::from_slice_ref("heìloworld");
623        StringArray::new(offsets.clone(), non_ascii_data.clone(), None);
624        BinaryArray::new(offsets, non_ascii_data.clone(), None);
625
626        let offsets = OffsetBuffer::new(vec![0, 3, 10].into());
627        let err = StringArray::try_new(offsets.clone(), non_ascii_data.clone(), None).unwrap_err();
628        assert_eq!(
629            err.to_string(),
630            "Invalid argument error: Split UTF-8 codepoint at offset 3"
631        );
632
633        BinaryArray::new(offsets, non_ascii_data, None);
634    }
635}
arrow_array/array/byte_array.rs

arrow_array/array/
byte_array.rs
