16831011058

Committed 08 Aug 2025 12:58PM UTC coverage: 84.902% (+0.9%) from 83.993%

Build # 16831011058

Build Type

Pull #4155

github

Committed by

web-flow

Commit Message

Merge b27de7bd8 into 1d60e8d6c

Pull Request Pull Request #4155: chore[bench-website]: add back tpc-ds to query_bench

Run Details

50710 of 59728 relevant lines covered (84.9%)

567607.41 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

81.98

/vortex-array/src/arrays/primitive/mod.rs

// SPDX-License-Identifier: Apache-2.0
// SPDX-FileCopyrightText: Copyright the Vortex contributors

use std::fmt::Debug;
use std::iter;

mod accessor;

use arrow_buffer::BooleanBufferBuilder;
use vortex_buffer::{Alignment, Buffer, BufferMut, ByteBuffer, ByteBufferMut};
use vortex_dtype::{DType, NativePType, Nullability, PType, match_each_native_ptype};
use vortex_error::{VortexResult, vortex_panic};

use crate::builders::ArrayBuilder;
use crate::stats::{ArrayStats, StatsSetRef};
use crate::validity::Validity;
use crate::{Array, ArrayRef, Canonical, EncodingId, EncodingRef, IntoArray, vtable};

mod compute;
mod native_value;
mod ops;
mod patch;
mod serde;
mod top_value;

pub use compute::{IS_CONST_LANE_WIDTH, compute_is_constant};
pub use native_value::NativeValue;

use crate::vtable::{
    ArrayVTable, CanonicalVTable, NotSupported, VTable, ValidityHelper,
    ValidityVTableFromValidityHelper,
};

vtable!(Primitive);

impl VTable for PrimitiveVTable {
    type Array = PrimitiveArray;
    type Encoding = PrimitiveEncoding;

    type ArrayVTable = Self;
    type CanonicalVTable = Self;
    type OperationsVTable = Self;
    type ValidityVTable = ValidityVTableFromValidityHelper;
    type VisitorVTable = Self;
    type ComputeVTable = NotSupported;
    type EncodeVTable = NotSupported;
    type SerdeVTable = Self;

    fn id(_encoding: &Self::Encoding) -> EncodingId {
        EncodingId::new_ref("vortex.primitive")
    }

    fn encoding(_array: &Self::Array) -> EncodingRef {
        EncodingRef::new_ref(PrimitiveEncoding.as_ref())
    }
}

/// A primitive array that stores [native types][vortex_dtype::NativePType] in a contiguous buffer
/// of memory, along with an optional validity child.
///
/// This mirrors the Apache Arrow Primitive layout and can be converted into and out of one
/// without allocations or copies.
///
/// The underlying buffer must be natively aligned to the primitive type they are representing.
///
/// Values are stored in their native representation with proper alignment.
/// Null values still occupy space in the buffer but are marked invalid in the validity mask.
///
/// # Examples
///
/// ```
/// use vortex_array::arrays::PrimitiveArray;
/// use vortex_array::compute::sum;
/// ///
/// // Create from iterator using FromIterator impl
/// let array: PrimitiveArray = [1i32, 2, 3, 4, 5].into_iter().collect();
///
/// // Slice the array
/// let sliced = array.slice(1, 3).unwrap();
///
/// // Access individual values
/// let value = sliced.scalar_at(0).unwrap();
/// assert_eq!(value, 2i32.into());
///
/// // Convert into a type-erased array that can be passed to compute functions.
/// let summed = sum(sliced.as_ref()).unwrap().as_primitive().typed_value::<i64>().unwrap();
/// assert_eq!(summed, 5i64);
/// ```
#[derive(Clone, Debug)]
pub struct PrimitiveArray {
    dtype: DType,
    buffer: ByteBuffer,
    validity: Validity,
    stats_set: ArrayStats,
}

#[derive(Clone, Debug)]
pub struct PrimitiveEncoding;

impl PrimitiveArray {
    pub fn new<T: NativePType>(buffer: impl Into<Buffer<T>>, validity: Validity) -> Self {
        let buffer = buffer.into();
        if let Some(len) = validity.maybe_len()
            && buffer.len() != len
        {
            vortex_panic!(
                "Buffer and validity length mismatch: buffer={}, validity={}",
                buffer.len(),
                len
            );
        }

        Self {
            dtype: DType::Primitive(T::PTYPE, validity.nullability()),
            buffer: buffer.into_byte_buffer(),
            validity,
            stats_set: Default::default(),
        }
    }

    pub fn empty<T: NativePType>(nullability: Nullability) -> Self {
        Self::new(Buffer::<T>::empty(), nullability.into())
    }

    pub fn from_byte_buffer(buffer: ByteBuffer, ptype: PType, validity: Validity) -> Self {
        match_each_native_ptype!(ptype, |T| {
            Self::new::<T>(Buffer::from_byte_buffer(buffer), validity)
        })
    }

    /// Create a PrimitiveArray from an iterator of `T`.
    /// NOTE: we cannot impl FromIterator trait since it conflicts with `FromIterator<T>`.
    pub fn from_option_iter<T: NativePType, I: IntoIterator<Item = Option<T>>>(iter: I) -> Self {
        let iter = iter.into_iter();
        let mut values = BufferMut::with_capacity(iter.size_hint().0);
        let mut validity = BooleanBufferBuilder::new(values.capacity());

        for i in iter {
            match i {
                None => {
                    validity.append(false);
                    values.push(T::default());
                }
                Some(e) => {
                    validity.append(true);
                    values.push(e);
                }
            }
        }
        Self::new(values.freeze(), Validity::from(validity.finish()))
    }

    /// Create a PrimitiveArray from a byte buffer containing only the valid elements.
    pub fn from_values_byte_buffer(
        valid_elems_buffer: ByteBuffer,
        ptype: PType,
        validity: Validity,
        n_rows: usize,
    ) -> VortexResult<Self> {
        let byte_width = ptype.byte_width();
        let alignment = Alignment::new(byte_width);
        let buffer = match &validity {
            Validity::AllValid | Validity::NonNullable => valid_elems_buffer.aligned(alignment),
            Validity::AllInvalid => ByteBuffer::zeroed_aligned(n_rows * byte_width, alignment),
            Validity::Array(is_valid) => {
                let bool_array = is_valid.to_canonical()?.into_bool()?;
                let bool_buffer = bool_array.boolean_buffer();
                let mut bytes = ByteBufferMut::zeroed_aligned(n_rows * byte_width, alignment);
                for (i, valid_i) in bool_buffer.set_indices().enumerate() {
                    bytes[valid_i * byte_width..(valid_i + 1) * byte_width]
                        .copy_from_slice(&valid_elems_buffer[i * byte_width..(i + 1) * byte_width])
                }
                bytes.freeze()
            }
        };

        Ok(Self::from_byte_buffer(buffer, ptype, validity))
    }

    pub fn ptype(&self) -> PType {
        self.dtype().as_ptype()
    }

    pub fn byte_buffer(&self) -> &ByteBuffer {
        &self.buffer
    }

    pub fn into_byte_buffer(self) -> ByteBuffer {
        self.buffer
    }

    pub fn buffer<T: NativePType>(&self) -> Buffer<T> {
        if T::PTYPE != self.ptype() {
            vortex_panic!(
                "Attempted to get buffer of type {} from array of type {}",
                T::PTYPE,
                self.ptype()
            )
        }
        Buffer::from_byte_buffer(self.byte_buffer().clone())
    }

    pub fn into_buffer<T: NativePType>(self) -> Buffer<T> {
        if T::PTYPE != self.ptype() {
            vortex_panic!(
                "Attempted to get buffer of type {} from array of type {}",
                T::PTYPE,
                self.ptype()
            )
        }
        Buffer::from_byte_buffer(self.buffer)
    }

    /// Extract a mutable buffer from the PrimitiveArray. Attempts to do this with zero-copy
    /// if the buffer is uniquely owned, otherwise will make a copy.
    pub fn into_buffer_mut<T: NativePType>(self) -> BufferMut<T> {
        if T::PTYPE != self.ptype() {
            vortex_panic!(
                "Attempted to get buffer_mut of type {} from array of type {}",
                T::PTYPE,
                self.ptype()
            )
        }
        self.into_buffer()
            .try_into_mut()
            .unwrap_or_else(|buffer| BufferMut::<T>::copy_from(&buffer))
    }

    /// Try to extract a mutable buffer from the PrimitiveArray with zero copy.
    #[allow(clippy::panic_in_result_fn)]
    pub fn try_into_buffer_mut<T: NativePType>(self) -> Result<BufferMut<T>, PrimitiveArray> {
        if T::PTYPE != self.ptype() {
            vortex_panic!(
                "Attempted to get buffer_mut of type {} from array of type {}",
                T::PTYPE,
                self.ptype()
            )
        }
        let validity = self.validity().clone();
        Buffer::<T>::from_byte_buffer(self.into_byte_buffer())
            .try_into_mut()
            .map_err(|buffer| PrimitiveArray::new(buffer, validity))
    }

    /// Map each element in the array to a new value.
    ///
    /// This ignores validity and maps over all maybe-null elements.
    ///
    /// TODO(ngates): we could be smarter here if validity is sparse and only run the function
    ///   over the valid elements.
    pub fn map_each<T, R, F>(self, f: F) -> PrimitiveArray
    where
        T: NativePType,
        R: NativePType,
        F: FnMut(T) -> R,
    {
        let validity = self.validity().clone();
        let buffer = match self.try_into_buffer_mut() {
            Ok(buffer_mut) => buffer_mut.map_each(f),
            Err(parray) => BufferMut::<R>::from_iter(parray.buffer::<T>().iter().copied().map(f)),
        };
        PrimitiveArray::new(buffer.freeze(), validity)
    }

    /// Map each element in the array to a new value.
    ///
    /// This doesn't ignore validity and maps over all maybe-null elements, with a bool true if
    /// valid and false otherwise.
    pub fn map_each_with_validity<T, R, F>(self, f: F) -> VortexResult<PrimitiveArray>
    where
        T: NativePType,
        R: NativePType,
        F: FnMut((T, bool)) -> R,
    {
        let validity = self.validity();

        let buf_iter = self.buffer::<T>().into_iter();

        let buffer = match &validity {
            Validity::NonNullable | Validity::AllValid => {
                BufferMut::<R>::from_iter(buf_iter.zip(iter::repeat(true)).map(f))
            }
            Validity::AllInvalid => {
                BufferMut::<R>::from_iter(buf_iter.zip(iter::repeat(false)).map(f))
            }
            Validity::Array(val) => {
                let val = val.to_canonical()?.into_bool()?;
                BufferMut::<R>::from_iter(buf_iter.zip(val.boolean_buffer()).map(f))
            }
        };
        Ok(PrimitiveArray::new(buffer.freeze(), validity.clone()))
    }

    /// Return a slice of the array's buffer.
    ///
    /// NOTE: these values may be nonsense if the validity buffer indicates that the value is null.
    pub fn as_slice<T: NativePType>(&self) -> &[T] {
        if T::PTYPE != self.ptype() {
            vortex_panic!(
                "Attempted to get slice of type {} from array of type {}",
                T::PTYPE,
                self.ptype()
            )
        }
        let raw_slice = self.byte_buffer().as_ptr();
        // SAFETY: alignment of Buffer is checked on construction
        unsafe {
            std::slice::from_raw_parts(raw_slice.cast(), self.byte_buffer().len() / size_of::<T>())
        }
    }

    pub fn reinterpret_cast(&self, ptype: PType) -> Self {
        if self.ptype() == ptype {
            return self.clone();
        }

        assert_eq!(
            self.ptype().byte_width(),
            ptype.byte_width(),
            "can't reinterpret cast between integers of two different widths"
        );

        PrimitiveArray::from_byte_buffer(self.byte_buffer().clone(), ptype, self.validity().clone())
    }
}

impl ArrayVTable<PrimitiveVTable> for PrimitiveVTable {
    fn len(array: &PrimitiveArray) -> usize {
        array.byte_buffer().len() / array.ptype().byte_width()
    }

    fn dtype(array: &PrimitiveArray) -> &DType {
        &array.dtype
    }

    fn stats(array: &PrimitiveArray) -> StatsSetRef<'_> {
        array.stats_set.to_ref(array.as_ref())
    }
}

impl ValidityHelper for PrimitiveArray {
    fn validity(&self) -> &Validity {
        &self.validity
    }
}

impl<T: NativePType> FromIterator<T> for PrimitiveArray {
    fn from_iter<I: IntoIterator<Item = T>>(iter: I) -> Self {
        let values = BufferMut::from_iter(iter);
        PrimitiveArray::new(values.freeze(), Validity::NonNullable)
    }
}

impl<T: NativePType> IntoArray for Buffer<T> {
    fn into_array(self) -> ArrayRef {
        PrimitiveArray::new(self, Validity::NonNullable).into_array()
    }
}

impl<T: NativePType> IntoArray for BufferMut<T> {
    fn into_array(self) -> ArrayRef {
        self.freeze().into_array()
    }
}

impl CanonicalVTable<PrimitiveVTable> for PrimitiveVTable {
    fn canonicalize(array: &PrimitiveArray) -> VortexResult<Canonical> {
        Ok(Canonical::Primitive(array.clone()))
    }

    fn append_to_builder(
        array: &PrimitiveArray,
        builder: &mut dyn ArrayBuilder,
    ) -> VortexResult<()> {
        builder.extend_from_array(array.as_ref())
    }
}

#[cfg(test)]
mod tests {
    use vortex_buffer::buffer;
    use vortex_scalar::PValue;

    use crate::arrays::{BoolArray, PrimitiveArray};
    use crate::compute::conformance::filter::test_filter_conformance;
    use crate::compute::conformance::mask::test_mask_conformance;
    use crate::compute::conformance::search_sorted::rstest_reuse::apply;
    use crate::compute::conformance::search_sorted::{search_sorted_conformance, *};
    use crate::search_sorted::{SearchResult, SearchSorted, SearchSortedSide};
    use crate::validity::Validity;
    use crate::{ArrayRef, IntoArray};

    #[apply(search_sorted_conformance)]
    fn test_search_sorted_primitive(
        #[case] array: ArrayRef,
        #[case] value: i32,
        #[case] side: SearchSortedSide,
        #[case] expected: SearchResult,
    ) {
        let res = array
            .as_primitive_typed()
            .search_sorted(&Some(PValue::from(value)), side);
        assert_eq!(res, expected);
    }

    #[test]
    fn test_mask_primitive_array() {
        test_mask_conformance(
            PrimitiveArray::new(buffer![0, 1, 2, 3, 4], Validity::NonNullable).as_ref(),
        );
        test_mask_conformance(
            PrimitiveArray::new(buffer![0, 1, 2, 3, 4], Validity::AllValid).as_ref(),
        );
        test_mask_conformance(
            PrimitiveArray::new(buffer![0, 1, 2, 3, 4], Validity::AllInvalid).as_ref(),
        );
        test_mask_conformance(
            PrimitiveArray::new(
                buffer![0, 1, 2, 3, 4],
                Validity::Array(
                    BoolArray::from_iter([true, false, true, false, true]).into_array(),
                ),
            )
            .as_ref(),
        );
    }

    #[test]
    fn test_filter_primitive_array() {
        // Test various sizes
        test_filter_conformance(
            PrimitiveArray::new(buffer![42i32], Validity::NonNullable).as_ref(),
        );
        test_filter_conformance(PrimitiveArray::new(buffer![0, 1], Validity::NonNullable).as_ref());
        test_filter_conformance(
            PrimitiveArray::new(buffer![0, 1, 2, 3, 4], Validity::NonNullable).as_ref(),
        );
        test_filter_conformance(
            PrimitiveArray::new(buffer![0, 1, 2, 3, 4, 5, 6, 7], Validity::NonNullable).as_ref(),
        );

        // Test with validity
        test_filter_conformance(
            PrimitiveArray::new(buffer![0, 1, 2, 3, 4], Validity::AllValid).as_ref(),
        );
        test_filter_conformance(
            PrimitiveArray::new(
                buffer![0, 1, 2, 3, 4, 5],
                Validity::Array(
                    BoolArray::from_iter([true, false, true, false, true, true]).into_array(),
                ),
            )
            .as_ref(),
        );
    }
}

1	// SPDX-License-Identifier: Apache-2.0
2	// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4	use std::fmt::Debug;
5	use std::iter;
6
7	mod accessor;
8
9	use arrow_buffer::BooleanBufferBuilder;
10	use vortex_buffer::{Alignment, Buffer, BufferMut, ByteBuffer, ByteBufferMut};
11	use vortex_dtype::{DType, NativePType, Nullability, PType, match_each_native_ptype};
12	use vortex_error::{VortexResult, vortex_panic};
13
14	use crate::builders::ArrayBuilder;
15	use crate::stats::{ArrayStats, StatsSetRef};
16	use crate::validity::Validity;
17	use crate::{Array, ArrayRef, Canonical, EncodingId, EncodingRef, IntoArray, vtable};
18
19	mod compute;
20	mod native_value;
21	mod ops;
22	mod patch;
23	mod serde;
24	mod top_value;
25
26	pub use compute::{IS_CONST_LANE_WIDTH, compute_is_constant};
27	pub use native_value::NativeValue;
28
29	use crate::vtable::{
30	ArrayVTable, CanonicalVTable, NotSupported, VTable, ValidityHelper,
31	ValidityVTableFromValidityHelper,
32	};
33
34	vtable!(Primitive);
35
36	impl VTable for PrimitiveVTable {
37	type Array = PrimitiveArray;
38	type Encoding = PrimitiveEncoding;
39
40	type ArrayVTable = Self;
41	type CanonicalVTable = Self;
42	type OperationsVTable = Self;
43	type ValidityVTable = ValidityVTableFromValidityHelper;
44	type VisitorVTable = Self;
45	type ComputeVTable = NotSupported;
46	type EncodeVTable = NotSupported;
47	type SerdeVTable = Self;
48
49	fn id(_encoding: &Self::Encoding) -> EncodingId {	1,757,528✔
50	EncodingId::new_ref("vortex.primitive")	1,757,528✔
51	}	1,757,528✔
52
53	fn encoding(_array: &Self::Array) -> EncodingRef {	868,728✔
54	EncodingRef::new_ref(PrimitiveEncoding.as_ref())	868,728✔
55	}	868,728✔
56	}
57
58	/// A primitive array that stores [native types][vortex_dtype::NativePType] in a contiguous buffer
59	/// of memory, along with an optional validity child.
60	///
61	/// This mirrors the Apache Arrow Primitive layout and can be converted into and out of one
62	/// without allocations or copies.
63	///
64	/// The underlying buffer must be natively aligned to the primitive type they are representing.
65	///
66	/// Values are stored in their native representation with proper alignment.
67	/// Null values still occupy space in the buffer but are marked invalid in the validity mask.
68	///
69	/// # Examples
70	///
71	/// ```
72	/// use vortex_array::arrays::PrimitiveArray;
73	/// use vortex_array::compute::sum;
74	/// ///
75	/// // Create from iterator using FromIterator impl
76	/// let array: PrimitiveArray = [1i32, 2, 3, 4, 5].into_iter().collect();
77	///
78	/// // Slice the array
79	/// let sliced = array.slice(1, 3).unwrap();
80	///
81	/// // Access individual values
82	/// let value = sliced.scalar_at(0).unwrap();
83	/// assert_eq!(value, 2i32.into());
84	///
85	/// // Convert into a type-erased array that can be passed to compute functions.
86	/// let summed = sum(sliced.as_ref()).unwrap().as_primitive().typed_value::<i64>().unwrap();
87	/// assert_eq!(summed, 5i64);
88	/// ```
89	#[derive(Clone, Debug)]
90	pub struct PrimitiveArray {
91	dtype: DType,
92	buffer: ByteBuffer,
93	validity: Validity,
94	stats_set: ArrayStats,
95	}
96
97	#[derive(Clone, Debug)]
98	pub struct PrimitiveEncoding;
99
100	impl PrimitiveArray {
101	pub fn new<T: NativePType>(buffer: impl Into<Buffer<T>>, validity: Validity) -> Self {	4,701,810✔
102	let buffer = buffer.into();	4,701,810✔
103	if let Some(len) = validity.maybe_len()	4,701,810✔
104	&& buffer.len() != len	95,405✔
105	{
106	vortex_panic!(	×
107	"Buffer and validity length mismatch: buffer={}, validity={}",	×
108	buffer.len(),	×
109	len
110	);
111	}	4,701,810✔
112
113	Self {	4,701,810✔
114	dtype: DType::Primitive(T::PTYPE, validity.nullability()),	4,701,810✔
115	buffer: buffer.into_byte_buffer(),	4,701,810✔
116	validity,	4,701,810✔
117	stats_set: Default::default(),	4,701,810✔
118	}	4,701,810✔
119	}	4,701,810✔
120
121	pub fn empty<T: NativePType>(nullability: Nullability) -> Self {	7,198✔
122	Self::new(Buffer::<T>::empty(), nullability.into())	7,198✔
123	}	7,198✔
124
125	pub fn from_byte_buffer(buffer: ByteBuffer, ptype: PType, validity: Validity) -> Self {	536,758✔
126	match_each_native_ptype!(ptype, \|T\| {	536,758✔
127	Self::new::<T>(Buffer::from_byte_buffer(buffer), validity)	2,857✔
128	})
129	}	536,758✔
130
131	/// Create a PrimitiveArray from an iterator of `T`.
132	/// NOTE: we cannot impl FromIterator trait since it conflicts with `FromIterator<T>`.
133	pub fn from_option_iter<T: NativePType, I: IntoIterator<Item = Option<T>>>(iter: I) -> Self {	6,665✔
134	let iter = iter.into_iter();	6,665✔
135	let mut values = BufferMut::with_capacity(iter.size_hint().0);	6,665✔
136	let mut validity = BooleanBufferBuilder::new(values.capacity());	6,665✔
137
138	for i in iter {	35,133✔
139	match i {	28,468✔
140	None => {	9,284✔
141	validity.append(false);	9,284✔
142	values.push(T::default());	9,284✔
143	}	9,284✔
144	Some(e) => {	19,184✔
145	validity.append(true);	19,184✔
146	values.push(e);	19,184✔
147	}	19,184✔
148	}
149	}
150	Self::new(values.freeze(), Validity::from(validity.finish()))	6,665✔
151	}	6,665✔
152
153	/// Create a PrimitiveArray from a byte buffer containing only the valid elements.
154	pub fn from_values_byte_buffer(	475,722✔
155	valid_elems_buffer: ByteBuffer,	475,722✔
156	ptype: PType,	475,722✔
157	validity: Validity,	475,722✔
158	n_rows: usize,	475,722✔
159	) -> VortexResult<Self> {	475,722✔
160	let byte_width = ptype.byte_width();	475,722✔
161	let alignment = Alignment::new(byte_width);	475,722✔
162	let buffer = match &validity {	475,722✔
163	Validity::AllValid \| Validity::NonNullable => valid_elems_buffer.aligned(alignment),	473,070✔
164	Validity::AllInvalid => ByteBuffer::zeroed_aligned(n_rows * byte_width, alignment),	×
165	Validity::Array(is_valid) => {	2,652✔
166	let bool_array = is_valid.to_canonical()?.into_bool()?;	2,652✔
167	let bool_buffer = bool_array.boolean_buffer();	2,652✔
168	let mut bytes = ByteBufferMut::zeroed_aligned(n_rows * byte_width, alignment);	2,652✔
169	for (i, valid_i) in bool_buffer.set_indices().enumerate() {	4,758✔
170	bytes[valid_i * byte_width..(valid_i + 1) * byte_width]	4,758✔
171	.copy_from_slice(&valid_elems_buffer[i * byte_width..(i + 1) * byte_width])	4,758✔
172	}
173	bytes.freeze()	2,652✔
174	}
175	};
176
177	Ok(Self::from_byte_buffer(buffer, ptype, validity))	475,722✔
178	}	475,722✔
179
180	pub fn ptype(&self) -> PType {	220,843,747✔
181	self.dtype().as_ptype()	220,843,747✔
182	}	220,843,747✔
183
184	pub fn byte_buffer(&self) -> &ByteBuffer {	218,312,464✔
185	&self.buffer	218,312,464✔
186	}	218,312,464✔
187
188	pub fn into_byte_buffer(self) -> ByteBuffer {	268,066✔
189	self.buffer	268,066✔
190	}	268,066✔
191
192	pub fn buffer<T: NativePType>(&self) -> Buffer<T> {	2,595,514✔
193	if T::PTYPE != self.ptype() {	2,595,514✔
194	vortex_panic!(	×
195	"Attempted to get buffer of type {} from array of type {}",	×
196	T::PTYPE,
197	self.ptype()	×
198	)
199	}	2,595,514✔
200	Buffer::from_byte_buffer(self.byte_buffer().clone())	2,595,514✔
201	}	2,595,514✔
202
203	pub fn into_buffer<T: NativePType>(self) -> Buffer<T> {	100,780✔
204	if T::PTYPE != self.ptype() {	100,780✔
205	vortex_panic!(	×
206	"Attempted to get buffer of type {} from array of type {}",	×
207	T::PTYPE,
208	self.ptype()	×
209	)
210	}	100,780✔
211	Buffer::from_byte_buffer(self.buffer)	100,780✔
212	}	100,780✔
213
214	/// Extract a mutable buffer from the PrimitiveArray. Attempts to do this with zero-copy
215	/// if the buffer is uniquely owned, otherwise will make a copy.
216	pub fn into_buffer_mut<T: NativePType>(self) -> BufferMut<T> {	82,075✔
217	if T::PTYPE != self.ptype() {	82,075✔
218	vortex_panic!(	×
219	"Attempted to get buffer_mut of type {} from array of type {}",	×
220	T::PTYPE,
221	self.ptype()	×
222	)
223	}	82,075✔
224	self.into_buffer()	82,075✔
225	.try_into_mut()	82,075✔
226	.unwrap_or_else(\|buffer\| BufferMut::<T>::copy_from(&buffer))	82,075✔
227	}	82,075✔
228
229	/// Try to extract a mutable buffer from the PrimitiveArray with zero copy.
230	#[allow(clippy::panic_in_result_fn)]
231	pub fn try_into_buffer_mut<T: NativePType>(self) -> Result<BufferMut<T>, PrimitiveArray> {	×
232	if T::PTYPE != self.ptype() {	×
233	vortex_panic!(	×
234	"Attempted to get buffer_mut of type {} from array of type {}",	×
235	T::PTYPE,
236	self.ptype()	×
237	)
238	}	×
239	let validity = self.validity().clone();	×
240	Buffer::<T>::from_byte_buffer(self.into_byte_buffer())	×
241	.try_into_mut()	×
242	.map_err(\|buffer\| PrimitiveArray::new(buffer, validity))	×
243	}	×
244
245	/// Map each element in the array to a new value.
246	///
247	/// This ignores validity and maps over all maybe-null elements.
248	///
249	/// TODO(ngates): we could be smarter here if validity is sparse and only run the function
250	/// over the valid elements.
251	pub fn map_each<T, R, F>(self, f: F) -> PrimitiveArray	×
252	where	×
253	T: NativePType,	×
254	R: NativePType,	×
255	F: FnMut(T) -> R,	×
256	{
257	let validity = self.validity().clone();	×
258	let buffer = match self.try_into_buffer_mut() {	×
259	Ok(buffer_mut) => buffer_mut.map_each(f),	×
260	Err(parray) => BufferMut::<R>::from_iter(parray.buffer::<T>().iter().copied().map(f)),	×
261	};
262	PrimitiveArray::new(buffer.freeze(), validity)	×
263	}	×
264
265	/// Map each element in the array to a new value.
266	///
267	/// This doesn't ignore validity and maps over all maybe-null elements, with a bool true if
268	/// valid and false otherwise.
269	pub fn map_each_with_validity<T, R, F>(self, f: F) -> VortexResult<PrimitiveArray>	4,496✔
270	where	4,496✔
271	T: NativePType,	4,496✔
272	R: NativePType,	4,496✔
273	F: FnMut((T, bool)) -> R,	4,496✔
274	{
275	let validity = self.validity();	4,496✔
276
277	let buf_iter = self.buffer::<T>().into_iter();	4,496✔
278
279	let buffer = match &validity {	4,496✔
280	Validity::NonNullable \| Validity::AllValid => {
281	BufferMut::<R>::from_iter(buf_iter.zip(iter::repeat(true)).map(f))	4,246✔
282	}
283	Validity::AllInvalid => {
284	BufferMut::<R>::from_iter(buf_iter.zip(iter::repeat(false)).map(f))	×
285	}
286	Validity::Array(val) => {	250✔
287	let val = val.to_canonical()?.into_bool()?;	250✔
288	BufferMut::<R>::from_iter(buf_iter.zip(val.boolean_buffer()).map(f))	250✔
289	}
290	};
291	Ok(PrimitiveArray::new(buffer.freeze(), validity.clone()))	4,496✔
292	}	4,496✔
293
294	/// Return a slice of the array's buffer.
295	///
296	/// NOTE: these values may be nonsense if the validity buffer indicates that the value is null.
297	pub fn as_slice<T: NativePType>(&self) -> &[T] {	39,038,906✔
298	if T::PTYPE != self.ptype() {	39,038,906✔
299	vortex_panic!(	×
300	"Attempted to get slice of type {} from array of type {}",	×
301	T::PTYPE,
302	self.ptype()	×
303	)
304	}	39,038,906✔
305	let raw_slice = self.byte_buffer().as_ptr();	39,038,906✔
306	// SAFETY: alignment of Buffer is checked on construction
307	unsafe {
308	std::slice::from_raw_parts(raw_slice.cast(), self.byte_buffer().len() / size_of::<T>())	39,038,906✔
309	}
310	}	39,038,906✔
311
312	pub fn reinterpret_cast(&self, ptype: PType) -> Self {	78,930✔
313	if self.ptype() == ptype {	78,930✔
314	return self.clone();	41,313✔
315	}	37,617✔
316
317	assert_eq!(	37,617✔
318	self.ptype().byte_width(),	37,617✔
319	ptype.byte_width(),	37,617✔
320	"can't reinterpret cast between integers of two different widths"	×
321	);
322
323	PrimitiveArray::from_byte_buffer(self.byte_buffer().clone(), ptype, self.validity().clone())	37,617✔
324	}	78,930✔
325	}
326
327	impl ArrayVTable<PrimitiveVTable> for PrimitiveVTable {
328	fn len(array: &PrimitiveArray) -> usize {	137,407,484✔
329	array.byte_buffer().len() / array.ptype().byte_width()	137,407,484✔
330	}	137,407,484✔
331
332	fn dtype(array: &PrimitiveArray) -> &DType {	320,466,095✔
333	&array.dtype	320,466,095✔
334	}	320,466,095✔
335
336	fn stats(array: &PrimitiveArray) -> StatsSetRef<'_> {	13,847,758✔
337	array.stats_set.to_ref(array.as_ref())	13,847,758✔
338	}	13,847,758✔
339	}
340
341	impl ValidityHelper for PrimitiveArray {
342	fn validity(&self) -> &Validity {	69,877,550✔
343	&self.validity	69,877,550✔
344	}	69,877,550✔
345	}
346
347	impl<T: NativePType> FromIterator<T> for PrimitiveArray {
348	fn from_iter<I: IntoIterator<Item = T>>(iter: I) -> Self {	39,920✔
349	let values = BufferMut::from_iter(iter);	39,920✔
350	PrimitiveArray::new(values.freeze(), Validity::NonNullable)	39,920✔
351	}	39,920✔
352	}
353
354	impl<T: NativePType> IntoArray for Buffer<T> {
355	fn into_array(self) -> ArrayRef {	37,550✔
356	PrimitiveArray::new(self, Validity::NonNullable).into_array()	37,550✔
357	}	37,550✔
358	}
359
360	impl<T: NativePType> IntoArray for BufferMut<T> {
361	fn into_array(self) -> ArrayRef {	14,563✔
362	self.freeze().into_array()	14,563✔
363	}	14,563✔
364	}
365
366	impl CanonicalVTable<PrimitiveVTable> for PrimitiveVTable {
367	fn canonicalize(array: &PrimitiveArray) -> VortexResult<Canonical> {	3,260,838✔
368	Ok(Canonical::Primitive(array.clone()))	3,260,838✔
369	}	3,260,838✔
370
371	fn append_to_builder(	39,458✔
372	array: &PrimitiveArray,	39,458✔
373	builder: &mut dyn ArrayBuilder,	39,458✔
374	) -> VortexResult<()> {	39,458✔
375	builder.extend_from_array(array.as_ref())	39,458✔
376	}	39,458✔
377	}
378
379	#[cfg(test)]
380	mod tests {
381	use vortex_buffer::buffer;
382	use vortex_scalar::PValue;
383
384	use crate::arrays::{BoolArray, PrimitiveArray};
385	use crate::compute::conformance::filter::test_filter_conformance;
386	use crate::compute::conformance::mask::test_mask_conformance;
387	use crate::compute::conformance::search_sorted::rstest_reuse::apply;
388	use crate::compute::conformance::search_sorted::{search_sorted_conformance, *};
389	use crate::search_sorted::{SearchResult, SearchSorted, SearchSortedSide};
390	use crate::validity::Validity;
391	use crate::{ArrayRef, IntoArray};
392
393	#[apply(search_sorted_conformance)]
394	fn test_search_sorted_primitive(
395	#[case] array: ArrayRef,
396	#[case] value: i32,
397	#[case] side: SearchSortedSide,
398	#[case] expected: SearchResult,
399	) {
400	let res = array
401	.as_primitive_typed()
402	.search_sorted(&Some(PValue::from(value)), side);
403	assert_eq!(res, expected);
404	}
405
406	#[test]
407	fn test_mask_primitive_array() {	1✔
408	test_mask_conformance(	1✔
409	PrimitiveArray::new(buffer![0, 1, 2, 3, 4], Validity::NonNullable).as_ref(),	1✔
410	);
411	test_mask_conformance(	1✔
412	PrimitiveArray::new(buffer![0, 1, 2, 3, 4], Validity::AllValid).as_ref(),	1✔
413	);
414	test_mask_conformance(	1✔
415	PrimitiveArray::new(buffer![0, 1, 2, 3, 4], Validity::AllInvalid).as_ref(),	1✔
416	);
417	test_mask_conformance(	1✔
418	PrimitiveArray::new(	1✔
419	buffer![0, 1, 2, 3, 4],	1✔
420	Validity::Array(	1✔
421	BoolArray::from_iter([true, false, true, false, true]).into_array(),	1✔
422	),	1✔
423	)	1✔
424	.as_ref(),	1✔
425	);
426	}	1✔
427
428	#[test]
429	fn test_filter_primitive_array() {	1✔
430	// Test various sizes
431	test_filter_conformance(	1✔
432	PrimitiveArray::new(buffer![42i32], Validity::NonNullable).as_ref(),	1✔
433	);
434	test_filter_conformance(PrimitiveArray::new(buffer![0, 1], Validity::NonNullable).as_ref());	1✔
435	test_filter_conformance(	1✔
436	PrimitiveArray::new(buffer![0, 1, 2, 3, 4], Validity::NonNullable).as_ref(),	1✔
437	);
438	test_filter_conformance(	1✔
439	PrimitiveArray::new(buffer![0, 1, 2, 3, 4, 5, 6, 7], Validity::NonNullable).as_ref(),	1✔
440	);
441
442	// Test with validity
443	test_filter_conformance(	1✔
444	PrimitiveArray::new(buffer![0, 1, 2, 3, 4], Validity::AllValid).as_ref(),	1✔
445	);
446	test_filter_conformance(	1✔
447	PrimitiveArray::new(	1✔
448	buffer![0, 1, 2, 3, 4, 5],	1✔
449	Validity::Array(	1✔
450	BoolArray::from_iter([true, false, true, false, true, true]).into_array(),	1✔
451	),	1✔
452	)	1✔
453	.as_ref(),	1✔
454	);
455	}	1✔
456	}

vortex-data / vortex / 16831011058

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous