vortex_array/arrays/varbin/vtable/
mod.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use std::sync::Arc;
5
6use vortex_error::VortexExpect;
7use vortex_error::VortexResult;
8use vortex_error::vortex_bail;
9use vortex_error::vortex_err;
10use vortex_error::vortex_panic;
11
12use crate::ArrayRef;
13use crate::DeserializeMetadata;
14use crate::ExecutionCtx;
15use crate::ExecutionResult;
16use crate::IntoArray;
17use crate::ProstMetadata;
18use crate::SerializeMetadata;
19use crate::arrays::VarBinArray;
20use crate::buffer::BufferHandle;
21use crate::dtype::DType;
22use crate::dtype::Nullability;
23use crate::dtype::PType;
24use crate::serde::ArrayChildren;
25use crate::validity::Validity;
26use crate::vtable;
27use crate::vtable::Array;
28use crate::vtable::ArrayId;
29use crate::vtable::VTable;
30use crate::vtable::ValidityVTableFromValidityHelper;
31use crate::vtable::validity_nchildren;
32use crate::vtable::validity_to_child;
33mod canonical;
34mod kernel;
35mod operations;
36mod validity;
37use std::hash::Hash;
38
39use canonical::varbin_to_canonical;
40use kernel::PARENT_KERNELS;
41use vortex_session::VortexSession;
42
43use crate::Precision;
44use crate::arrays::varbin::compute::rules::PARENT_RULES;
45use crate::hash::ArrayEq;
46use crate::hash::ArrayHash;
47use crate::stats::StatsSetRef;
48
49vtable!(VarBin);
50
51#[derive(Clone, prost::Message)]
52pub struct VarBinMetadata {
53    #[prost(enumeration = "PType", tag = "1")]
54    pub(crate) offsets_ptype: i32,
55}
56
57impl VTable for VarBin {
58    type Array = VarBinArray;
59
60    type Metadata = ProstMetadata<VarBinMetadata>;
61    type OperationsVTable = Self;
62    type ValidityVTable = ValidityVTableFromValidityHelper;
63    fn vtable(_array: &Self::Array) -> &Self {
64        &VarBin
65    }
66
67    fn id(&self) -> ArrayId {
68        Self::ID
69    }
70
71    fn len(array: &VarBinArray) -> usize {
72        array.offsets().len().saturating_sub(1)
73    }
74
75    fn dtype(array: &VarBinArray) -> &DType {
76        &array.dtype
77    }
78
79    fn stats(array: &VarBinArray) -> StatsSetRef<'_> {
80        array.stats_set.to_ref(array.as_ref())
81    }
82
83    fn array_hash<H: std::hash::Hasher>(array: &VarBinArray, state: &mut H, precision: Precision) {
84        array.dtype.hash(state);
85        array.bytes().array_hash(state, precision);
86        array.offsets().array_hash(state, precision);
87        array.validity.array_hash(state, precision);
88    }
89
90    fn array_eq(array: &VarBinArray, other: &VarBinArray, precision: Precision) -> bool {
91        array.dtype == other.dtype
92            && array.bytes().array_eq(other.bytes(), precision)
93            && array.offsets().array_eq(other.offsets(), precision)
94            && array.validity.array_eq(&other.validity, precision)
95    }
96
97    fn nbuffers(_array: &VarBinArray) -> usize {
98        1
99    }
100
101    fn buffer(array: &VarBinArray, idx: usize) -> BufferHandle {
102        match idx {
103            0 => array.bytes_handle().clone(),
104            _ => vortex_panic!("VarBinArray buffer index {idx} out of bounds"),
105        }
106    }
107
108    fn buffer_name(_array: &VarBinArray, idx: usize) -> Option<String> {
109        match idx {
110            0 => Some("bytes".to_string()),
111            _ => vortex_panic!("VarBinArray buffer_name index {idx} out of bounds"),
112        }
113    }
114
115    fn nchildren(array: &VarBinArray) -> usize {
116        1 + validity_nchildren(&array.validity)
117    }
118
119    fn child(array: &VarBinArray, idx: usize) -> ArrayRef {
120        match idx {
121            0 => array.offsets().clone(),
122            1 => validity_to_child(&array.validity, array.len())
123                .vortex_expect("VarBinArray validity child out of bounds"),
124            _ => vortex_panic!("VarBinArray child index {idx} out of bounds"),
125        }
126    }
127
128    fn child_name(_array: &VarBinArray, idx: usize) -> String {
129        match idx {
130            0 => "offsets".to_string(),
131            1 => "validity".to_string(),
132            _ => vortex_panic!("VarBinArray child_name index {idx} out of bounds"),
133        }
134    }
135
136    fn metadata(array: &VarBinArray) -> VortexResult<Self::Metadata> {
137        Ok(ProstMetadata(VarBinMetadata {
138            offsets_ptype: PType::try_from(array.offsets().dtype())
139                .vortex_expect("Must be a valid PType") as i32,
140        }))
141    }
142
143    fn serialize(metadata: Self::Metadata) -> VortexResult<Option<Vec<u8>>> {
144        Ok(Some(metadata.serialize()))
145    }
146
147    fn deserialize(
148        bytes: &[u8],
149        _dtype: &DType,
150        _len: usize,
151        _buffers: &[BufferHandle],
152        _session: &VortexSession,
153    ) -> VortexResult<Self::Metadata> {
154        Ok(ProstMetadata(ProstMetadata::<VarBinMetadata>::deserialize(
155            bytes,
156        )?))
157    }
158
159    fn build(
160        dtype: &DType,
161        len: usize,
162        metadata: &Self::Metadata,
163        buffers: &[BufferHandle],
164        children: &dyn ArrayChildren,
165    ) -> VortexResult<VarBinArray> {
166        let validity = if children.len() == 1 {
167            Validity::from(dtype.nullability())
168        } else if children.len() == 2 {
169            let validity = children.get(1, &Validity::DTYPE, len)?;
170            Validity::Array(validity)
171        } else {
172            vortex_bail!("Expected 1 or 2 children, got {}", children.len());
173        };
174
175        let offsets = children.get(
176            0,
177            &DType::Primitive(metadata.offsets_ptype(), Nullability::NonNullable),
178            len + 1,
179        )?;
180
181        if buffers.len() != 1 {
182            vortex_bail!("Expected 1 buffer, got {}", buffers.len());
183        }
184        let bytes = buffers[0].clone().try_to_host_sync()?;
185
186        VarBinArray::try_new(offsets, bytes, dtype.clone(), validity)
187    }
188
189    fn with_children(array: &mut Self::Array, children: Vec<ArrayRef>) -> VortexResult<()> {
190        match children.len() {
191            1 => {
192                let [offsets]: [ArrayRef; 1] = children
193                    .try_into()
194                    .map_err(|_| vortex_err!("Failed to convert children to array"))?;
195                array.offsets = offsets;
196            }
197            2 => {
198                let [offsets, validity]: [ArrayRef; 2] = children
199                    .try_into()
200                    .map_err(|_| vortex_err!("Failed to convert children to array"))?;
201                array.offsets = offsets;
202                array.validity = Validity::Array(validity);
203            }
204            _ => vortex_bail!(
205                "VarBinArray expects 1 or 2 children (offsets, validity?), got {}",
206                children.len()
207            ),
208        }
209        Ok(())
210    }
211
212    fn reduce_parent(
213        array: &Array<Self>,
214        parent: &ArrayRef,
215        child_idx: usize,
216    ) -> VortexResult<Option<ArrayRef>> {
217        PARENT_RULES.evaluate(array, parent, child_idx)
218    }
219
220    fn execute_parent(
221        array: &Array<Self>,
222        parent: &ArrayRef,
223        child_idx: usize,
224        ctx: &mut ExecutionCtx,
225    ) -> VortexResult<Option<ArrayRef>> {
226        PARENT_KERNELS.execute(array, parent, child_idx, ctx)
227    }
228
229    fn execute(array: Arc<Array<Self>>, ctx: &mut ExecutionCtx) -> VortexResult<ExecutionResult> {
230        Ok(ExecutionResult::done(
231            varbin_to_canonical(&array, ctx)?.into_array(),
232        ))
233    }
234}
235
236#[derive(Clone, Debug)]
237pub struct VarBin;
238
239impl VarBin {
240    pub const ID: ArrayId = ArrayId::new_ref("vortex.varbin");
241}
vortex_array/arrays/varbin/vtable/mod.rs

vortex_array/arrays/varbin/vtable/
mod.rs
