Skip to main content

vortex_array/arrays/primitive/vtable/
mod.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use kernel::PARENT_KERNELS;
5use vortex_error::VortexResult;
6use vortex_error::vortex_bail;
7use vortex_error::vortex_ensure;
8use vortex_error::vortex_panic;
9
10use crate::ArrayRef;
11use crate::ExecutionCtx;
12use crate::ExecutionResult;
13use crate::array::Array;
14use crate::array::ArrayView;
15use crate::array::VTable;
16use crate::arrays::primitive::PrimitiveData;
17use crate::buffer::BufferHandle;
18use crate::dtype::DType;
19use crate::dtype::PType;
20use crate::serde::ArrayChildren;
21use crate::validity::Validity;
22mod kernel;
23mod operations;
24mod validity;
25
26use std::hash::Hasher;
27
28use vortex_buffer::Alignment;
29use vortex_session::VortexSession;
30use vortex_session::registry::CachedId;
31
32use crate::Precision;
33use crate::array::ArrayId;
34use crate::arrays::primitive::array::SLOT_NAMES;
35use crate::arrays::primitive::compute::rules::RULES;
36use crate::hash::ArrayEq;
37use crate::hash::ArrayHash;
38
39/// A [`Primitive`]-encoded Vortex array.
40pub type PrimitiveArray = Array<Primitive>;
41
42impl ArrayHash for PrimitiveData {
43    fn array_hash<H: Hasher>(&self, state: &mut H, precision: Precision) {
44        self.buffer.array_hash(state, precision);
45    }
46}
47
48impl ArrayEq for PrimitiveData {
49    fn array_eq(&self, other: &Self, precision: Precision) -> bool {
50        self.buffer.array_eq(&other.buffer, precision)
51    }
52}
53
54impl VTable for Primitive {
55    type ArrayData = PrimitiveData;
56
57    type OperationsVTable = Self;
58    type ValidityVTable = Self;
59
60    fn id(&self) -> ArrayId {
61        static ID: CachedId = CachedId::new("vortex.primitive");
62        *ID
63    }
64
65    fn nbuffers(_array: ArrayView<'_, Self>) -> usize {
66        1
67    }
68
69    fn buffer(array: ArrayView<'_, Self>, idx: usize) -> BufferHandle {
70        match idx {
71            0 => array.buffer_handle().clone(),
72            _ => vortex_panic!("PrimitiveArray buffer index {idx} out of bounds"),
73        }
74    }
75
76    fn buffer_name(_array: ArrayView<'_, Self>, idx: usize) -> Option<String> {
77        match idx {
78            0 => Some("values".to_string()),
79            _ => None,
80        }
81    }
82
83    fn serialize(
84        _array: ArrayView<'_, Self>,
85        _session: &VortexSession,
86    ) -> VortexResult<Option<Vec<u8>>> {
87        Ok(Some(vec![]))
88    }
89
90    fn validate(
91        &self,
92        data: &PrimitiveData,
93        dtype: &DType,
94        len: usize,
95        slots: &[Option<ArrayRef>],
96    ) -> VortexResult<()> {
97        let DType::Primitive(_, nullability) = dtype else {
98            vortex_bail!("Expected primitive dtype, got {dtype:?}");
99        };
100        vortex_ensure!(
101            data.len() == len,
102            "PrimitiveArray length {} does not match outer length {}",
103            data.len(),
104            len
105        );
106        let validity = crate::array::child_to_validity(&slots[0], *nullability);
107        if let Some(validity_len) = validity.maybe_len() {
108            vortex_ensure!(
109                validity_len == len,
110                "PrimitiveArray validity len {} does not match outer length {}",
111                validity_len,
112                len
113            );
114        }
115
116        Ok(())
117    }
118
119    fn deserialize(
120        &self,
121        dtype: &DType,
122        len: usize,
123        metadata: &[u8],
124
125        buffers: &[BufferHandle],
126        children: &dyn ArrayChildren,
127        _session: &VortexSession,
128    ) -> VortexResult<crate::array::ArrayParts<Self>> {
129        if !metadata.is_empty() {
130            vortex_bail!(
131                "PrimitiveArray expects empty metadata, got {} bytes",
132                metadata.len()
133            );
134        }
135        if buffers.len() != 1 {
136            vortex_bail!("Expected 1 buffer, got {}", buffers.len());
137        }
138        let buffer = buffers[0].clone();
139
140        let validity = if children.is_empty() {
141            Validity::from(dtype.nullability())
142        } else if children.len() == 1 {
143            let validity = children.get(0, &Validity::DTYPE, len)?;
144            Validity::Array(validity)
145        } else {
146            vortex_bail!("Expected 0 or 1 child, got {}", children.len());
147        };
148
149        let ptype = PType::try_from(dtype)?;
150
151        vortex_ensure!(
152            buffer.is_aligned_to(Alignment::new(ptype.byte_width())),
153            "Misaligned buffer cannot be used to build PrimitiveArray of {ptype}"
154        );
155
156        if buffer.len() != ptype.byte_width() * len {
157            vortex_bail!(
158                "Buffer length {} does not match expected length {} for {}, {}",
159                buffer.len(),
160                ptype.byte_width() * len,
161                ptype.byte_width(),
162                len,
163            );
164        }
165
166        vortex_ensure!(
167            buffer.is_aligned_to(Alignment::new(ptype.byte_width())),
168            "PrimitiveArray::build: Buffer (align={}) must be aligned to {}",
169            buffer.alignment(),
170            ptype.byte_width()
171        );
172
173        // SAFETY: checked ahead of time
174        let slots = PrimitiveData::make_slots(&validity, len);
175        let data = unsafe { PrimitiveData::new_unchecked_from_handle(buffer, ptype, validity) };
176        Ok(crate::array::ArrayParts::new(self.clone(), dtype.clone(), len, data).with_slots(slots))
177    }
178
179    fn slot_name(_array: ArrayView<'_, Self>, idx: usize) -> String {
180        SLOT_NAMES[idx].to_string()
181    }
182
183    fn execute(array: Array<Self>, _ctx: &mut ExecutionCtx) -> VortexResult<ExecutionResult> {
184        Ok(ExecutionResult::done(array))
185    }
186
187    fn reduce_parent(
188        array: ArrayView<'_, Self>,
189        parent: &ArrayRef,
190        child_idx: usize,
191    ) -> VortexResult<Option<ArrayRef>> {
192        RULES.evaluate(array, parent, child_idx)
193    }
194
195    fn execute_parent(
196        array: ArrayView<'_, Self>,
197        parent: &ArrayRef,
198        child_idx: usize,
199        ctx: &mut ExecutionCtx,
200    ) -> VortexResult<Option<ArrayRef>> {
201        PARENT_KERNELS.execute(array, parent, child_idx, ctx)
202    }
203}
204
205#[derive(Clone, Debug)]
206pub struct Primitive;
207
208#[cfg(test)]
209mod tests {
210    use vortex_buffer::ByteBufferMut;
211    use vortex_buffer::buffer;
212    use vortex_session::registry::ReadContext;
213
214    use crate::ArrayContext;
215    use crate::IntoArray;
216    use crate::LEGACY_SESSION;
217    use crate::arrays::PrimitiveArray;
218    use crate::assert_arrays_eq;
219    use crate::serde::SerializeOptions;
220    use crate::serde::SerializedArray;
221    use crate::validity::Validity;
222
223    #[test]
224    fn test_nullable_primitive_serde_roundtrip() {
225        let array = PrimitiveArray::new(
226            buffer![1i32, 2, 3, 4],
227            Validity::from_iter([true, false, true, false]),
228        );
229        let dtype = array.dtype().clone();
230        let len = array.len();
231
232        let ctx = ArrayContext::empty();
233        let serialized = array
234            .clone()
235            .into_array()
236            .serialize(&ctx, &LEGACY_SESSION, &SerializeOptions::default())
237            .unwrap();
238
239        let mut concat = ByteBufferMut::empty();
240        for buf in serialized {
241            concat.extend_from_slice(buf.as_ref());
242        }
243        let parts = SerializedArray::try_from(concat.freeze()).unwrap();
244        let decoded = parts
245            .decode(
246                &dtype,
247                len,
248                &ReadContext::new(ctx.to_ids()),
249                &LEGACY_SESSION,
250            )
251            .unwrap();
252
253        assert_arrays_eq!(decoded, array);
254    }
255}