vortex_array/arrays/varbin/vtable/
mod.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use vortex_error::VortexExpect;
5use vortex_error::VortexResult;
6use vortex_error::vortex_bail;
7use vortex_error::vortex_err;
8use vortex_error::vortex_panic;
9
10use crate::ArrayRef;
11use crate::DeserializeMetadata;
12use crate::ExecutionCtx;
13use crate::ExecutionStep;
14use crate::IntoArray;
15use crate::ProstMetadata;
16use crate::SerializeMetadata;
17use crate::arrays::VarBinArray;
18use crate::buffer::BufferHandle;
19use crate::dtype::DType;
20use crate::dtype::Nullability;
21use crate::dtype::PType;
22use crate::serde::ArrayChildren;
23use crate::validity::Validity;
24use crate::vtable;
25use crate::vtable::ArrayId;
26use crate::vtable::VTable;
27use crate::vtable::ValidityVTableFromValidityHelper;
28use crate::vtable::validity_nchildren;
29use crate::vtable::validity_to_child;
30mod canonical;
31mod kernel;
32mod operations;
33mod validity;
34use std::hash::Hash;
35
36use canonical::varbin_to_canonical;
37use kernel::PARENT_KERNELS;
38use vortex_session::VortexSession;
39
40use crate::Precision;
41use crate::arrays::varbin::compute::rules::PARENT_RULES;
42use crate::hash::ArrayEq;
43use crate::hash::ArrayHash;
44use crate::stats::StatsSetRef;
45
46vtable!(VarBin);
47
48#[derive(Clone, prost::Message)]
49pub struct VarBinMetadata {
50    #[prost(enumeration = "PType", tag = "1")]
51    pub(crate) offsets_ptype: i32,
52}
53
54impl VTable for VarBinVTable {
55    type Array = VarBinArray;
56
57    type Metadata = ProstMetadata<VarBinMetadata>;
58    type OperationsVTable = Self;
59    type ValidityVTable = ValidityVTableFromValidityHelper;
60    fn id(_array: &Self::Array) -> ArrayId {
61        Self::ID
62    }
63
64    fn len(array: &VarBinArray) -> usize {
65        array.offsets().len().saturating_sub(1)
66    }
67
68    fn dtype(array: &VarBinArray) -> &DType {
69        &array.dtype
70    }
71
72    fn stats(array: &VarBinArray) -> StatsSetRef<'_> {
73        array.stats_set.to_ref(array.as_ref())
74    }
75
76    fn array_hash<H: std::hash::Hasher>(array: &VarBinArray, state: &mut H, precision: Precision) {
77        array.dtype.hash(state);
78        array.bytes().array_hash(state, precision);
79        array.offsets().array_hash(state, precision);
80        array.validity.array_hash(state, precision);
81    }
82
83    fn array_eq(array: &VarBinArray, other: &VarBinArray, precision: Precision) -> bool {
84        array.dtype == other.dtype
85            && array.bytes().array_eq(other.bytes(), precision)
86            && array.offsets().array_eq(other.offsets(), precision)
87            && array.validity.array_eq(&other.validity, precision)
88    }
89
90    fn nbuffers(_array: &VarBinArray) -> usize {
91        1
92    }
93
94    fn buffer(array: &VarBinArray, idx: usize) -> BufferHandle {
95        match idx {
96            0 => array.bytes_handle().clone(),
97            _ => vortex_panic!("VarBinArray buffer index {idx} out of bounds"),
98        }
99    }
100
101    fn buffer_name(_array: &VarBinArray, idx: usize) -> Option<String> {
102        match idx {
103            0 => Some("bytes".to_string()),
104            _ => vortex_panic!("VarBinArray buffer_name index {idx} out of bounds"),
105        }
106    }
107
108    fn nchildren(array: &VarBinArray) -> usize {
109        1 + validity_nchildren(&array.validity)
110    }
111
112    fn child(array: &VarBinArray, idx: usize) -> ArrayRef {
113        match idx {
114            0 => array.offsets().clone(),
115            1 => validity_to_child(&array.validity, array.len())
116                .vortex_expect("VarBinArray validity child out of bounds"),
117            _ => vortex_panic!("VarBinArray child index {idx} out of bounds"),
118        }
119    }
120
121    fn child_name(_array: &VarBinArray, idx: usize) -> String {
122        match idx {
123            0 => "offsets".to_string(),
124            1 => "validity".to_string(),
125            _ => vortex_panic!("VarBinArray child_name index {idx} out of bounds"),
126        }
127    }
128
129    fn metadata(array: &VarBinArray) -> VortexResult<Self::Metadata> {
130        Ok(ProstMetadata(VarBinMetadata {
131            offsets_ptype: PType::try_from(array.offsets().dtype())
132                .vortex_expect("Must be a valid PType") as i32,
133        }))
134    }
135
136    fn serialize(metadata: Self::Metadata) -> VortexResult<Option<Vec<u8>>> {
137        Ok(Some(metadata.serialize()))
138    }
139
140    fn deserialize(
141        bytes: &[u8],
142        _dtype: &DType,
143        _len: usize,
144        _buffers: &[BufferHandle],
145        _session: &VortexSession,
146    ) -> VortexResult<Self::Metadata> {
147        Ok(ProstMetadata(ProstMetadata::<VarBinMetadata>::deserialize(
148            bytes,
149        )?))
150    }
151
152    fn build(
153        dtype: &DType,
154        len: usize,
155        metadata: &Self::Metadata,
156        buffers: &[BufferHandle],
157        children: &dyn ArrayChildren,
158    ) -> VortexResult<VarBinArray> {
159        let validity = if children.len() == 1 {
160            Validity::from(dtype.nullability())
161        } else if children.len() == 2 {
162            let validity = children.get(1, &Validity::DTYPE, len)?;
163            Validity::Array(validity)
164        } else {
165            vortex_bail!("Expected 1 or 2 children, got {}", children.len());
166        };
167
168        let offsets = children.get(
169            0,
170            &DType::Primitive(metadata.offsets_ptype(), Nullability::NonNullable),
171            len + 1,
172        )?;
173
174        if buffers.len() != 1 {
175            vortex_bail!("Expected 1 buffer, got {}", buffers.len());
176        }
177        let bytes = buffers[0].clone().try_to_host_sync()?;
178
179        VarBinArray::try_new(offsets, bytes, dtype.clone(), validity)
180    }
181
182    fn with_children(array: &mut Self::Array, children: Vec<ArrayRef>) -> VortexResult<()> {
183        match children.len() {
184            1 => {
185                let [offsets]: [ArrayRef; 1] = children
186                    .try_into()
187                    .map_err(|_| vortex_err!("Failed to convert children to array"))?;
188                array.offsets = offsets;
189            }
190            2 => {
191                let [offsets, validity]: [ArrayRef; 2] = children
192                    .try_into()
193                    .map_err(|_| vortex_err!("Failed to convert children to array"))?;
194                array.offsets = offsets;
195                array.validity = Validity::Array(validity);
196            }
197            _ => vortex_bail!(
198                "VarBinArray expects 1 or 2 children (offsets, validity?), got {}",
199                children.len()
200            ),
201        }
202        Ok(())
203    }
204
205    fn reduce_parent(
206        array: &Self::Array,
207        parent: &ArrayRef,
208        child_idx: usize,
209    ) -> VortexResult<Option<ArrayRef>> {
210        PARENT_RULES.evaluate(array, parent, child_idx)
211    }
212
213    fn execute_parent(
214        array: &Self::Array,
215        parent: &ArrayRef,
216        child_idx: usize,
217        ctx: &mut ExecutionCtx,
218    ) -> VortexResult<Option<ArrayRef>> {
219        PARENT_KERNELS.execute(array, parent, child_idx, ctx)
220    }
221
222    fn execute(array: &Self::Array, ctx: &mut ExecutionCtx) -> VortexResult<ExecutionStep> {
223        Ok(ExecutionStep::Done(
224            varbin_to_canonical(array, ctx)?.into_array(),
225        ))
226    }
227}
228
229#[derive(Debug)]
230pub struct VarBinVTable;
231
232impl VarBinVTable {
233    pub const ID: ArrayId = ArrayId::new_ref("vortex.varbin");
234}
vortex_array/arrays/varbin/vtable/mod.rs

vortex_array/arrays/varbin/vtable/
mod.rs
