vortex_array/arrays/list/vtable/
mod.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use std::hash::Hasher;
5use std::sync::Arc;
6
7use prost::Message;
8use vortex_error::VortexExpect;
9use vortex_error::VortexResult;
10use vortex_error::vortex_bail;
11use vortex_error::vortex_ensure;
12use vortex_error::vortex_panic;
13use vortex_session::VortexSession;
14
15use crate::ArrayEq;
16use crate::ArrayHash;
17use crate::ArrayRef;
18use crate::ExecutionCtx;
19use crate::ExecutionResult;
20use crate::IntoArray;
21use crate::Precision;
22use crate::array::Array;
23use crate::array::ArrayId;
24use crate::array::ArrayView;
25use crate::array::VTable;
26use crate::arrays::list::ListArrayExt;
27use crate::arrays::list::ListData;
28use crate::arrays::list::array::NUM_SLOTS;
29use crate::arrays::list::array::SLOT_NAMES;
30use crate::arrays::list::compute::PARENT_KERNELS;
31use crate::arrays::list::compute::rules::PARENT_RULES;
32use crate::arrays::listview::list_view_from_list;
33use crate::buffer::BufferHandle;
34use crate::dtype::DType;
35use crate::dtype::Nullability;
36use crate::dtype::PType;
37use crate::serde::ArrayChildren;
38use crate::validity::Validity;
39mod operations;
40mod validity;
41/// A [`List`]-encoded Vortex array.
42pub type ListArray = Array<List>;
43
44#[derive(Clone, prost::Message)]
45pub struct ListMetadata {
46    #[prost(uint64, tag = "1")]
47    elements_len: u64,
48    #[prost(enumeration = "PType", tag = "2")]
49    offset_ptype: i32,
50}
51
52impl ArrayHash for ListData {
53    fn array_hash<H: Hasher>(&self, _state: &mut H, _precision: Precision) {}
54}
55
56impl ArrayEq for ListData {
57    fn array_eq(&self, _other: &Self, _precision: Precision) -> bool {
58        true
59    }
60}
61
62impl VTable for List {
63    type ArrayData = ListData;
64
65    type OperationsVTable = Self;
66    type ValidityVTable = Self;
67
68    fn id(&self) -> ArrayId {
69        Self::ID
70    }
71
72    fn nbuffers(_array: ArrayView<'_, Self>) -> usize {
73        0
74    }
75
76    fn buffer(_array: ArrayView<'_, Self>, idx: usize) -> BufferHandle {
77        vortex_panic!("ListArray buffer index {idx} out of bounds")
78    }
79
80    fn buffer_name(_array: ArrayView<'_, Self>, idx: usize) -> Option<String> {
81        vortex_panic!("ListArray buffer_name index {idx} out of bounds")
82    }
83
84    fn reduce_parent(
85        array: ArrayView<'_, Self>,
86        parent: &ArrayRef,
87        child_idx: usize,
88    ) -> VortexResult<Option<ArrayRef>> {
89        PARENT_RULES.evaluate(array, parent, child_idx)
90    }
91
92    fn serialize(
93        array: ArrayView<'_, Self>,
94        _session: &VortexSession,
95    ) -> VortexResult<Option<Vec<u8>>> {
96        Ok(Some(
97            ListMetadata {
98                elements_len: array.elements().len() as u64,
99                offset_ptype: PType::try_from(array.offsets().dtype())? as i32,
100            }
101            .encode_to_vec(),
102        ))
103    }
104
105    fn validate(
106        &self,
107        _data: &ListData,
108        dtype: &DType,
109        len: usize,
110        slots: &[Option<ArrayRef>],
111    ) -> VortexResult<()> {
112        vortex_ensure!(
113            slots.len() == NUM_SLOTS,
114            "ListArray expected {NUM_SLOTS} slots, found {}",
115            slots.len()
116        );
117        let elements = slots[crate::arrays::list::array::ELEMENTS_SLOT]
118            .as_ref()
119            .vortex_expect("ListArray elements slot");
120        let offsets = slots[crate::arrays::list::array::OFFSETS_SLOT]
121            .as_ref()
122            .vortex_expect("ListArray offsets slot");
123        vortex_ensure!(
124            offsets.len().saturating_sub(1) == len,
125            "ListArray length {} does not match outer length {}",
126            offsets.len().saturating_sub(1),
127            len
128        );
129
130        let actual_dtype = DType::List(Arc::new(elements.dtype().clone()), dtype.nullability());
131        vortex_ensure!(
132            &actual_dtype == dtype,
133            "ListArray dtype {} does not match outer dtype {}",
134            actual_dtype,
135            dtype
136        );
137
138        Ok(())
139    }
140
141    fn deserialize(
142        &self,
143        dtype: &DType,
144        len: usize,
145        metadata: &[u8],
146
147        _buffers: &[BufferHandle],
148        children: &dyn ArrayChildren,
149        _session: &VortexSession,
150    ) -> VortexResult<crate::array::ArrayParts<Self>> {
151        let metadata = ListMetadata::decode(metadata)?;
152        let validity = if children.len() == 2 {
153            Validity::from(dtype.nullability())
154        } else if children.len() == 3 {
155            let validity = children.get(2, &Validity::DTYPE, len)?;
156            Validity::Array(validity)
157        } else {
158            vortex_bail!("Expected 2 or 3 children, got {}", children.len());
159        };
160
161        let DType::List(element_dtype, _) = &dtype else {
162            vortex_bail!("Expected List dtype, got {:?}", dtype);
163        };
164        let elements = children.get(
165            0,
166            element_dtype.as_ref(),
167            usize::try_from(metadata.elements_len)?,
168        )?;
169
170        let offsets = children.get(
171            1,
172            &DType::Primitive(metadata.offset_ptype(), Nullability::NonNullable),
173            len + 1,
174        )?;
175
176        let data = ListData::try_build(elements.clone(), offsets.clone(), validity.clone())?;
177        let slots = ListData::make_slots(&elements, &offsets, &validity, len);
178        Ok(crate::array::ArrayParts::new(self.clone(), dtype.clone(), len, data).with_slots(slots))
179    }
180
181    fn slot_name(_array: ArrayView<'_, Self>, idx: usize) -> String {
182        SLOT_NAMES[idx].to_string()
183    }
184
185    fn execute(array: Array<Self>, ctx: &mut ExecutionCtx) -> VortexResult<ExecutionResult> {
186        Ok(ExecutionResult::done(
187            list_view_from_list(array, ctx)?.into_array(),
188        ))
189    }
190
191    fn execute_parent(
192        array: ArrayView<'_, Self>,
193        parent: &ArrayRef,
194        child_idx: usize,
195        ctx: &mut ExecutionCtx,
196    ) -> VortexResult<Option<ArrayRef>> {
197        PARENT_KERNELS.execute(array, parent, child_idx, ctx)
198    }
199}
200
201#[derive(Clone, Debug)]
202pub struct List;
203
204impl List {
205    pub const ID: ArrayId = ArrayId::new_ref("vortex.list");
206}
vortex_array/arrays/list/vtable/mod.rs

vortex_array/arrays/list/vtable/
mod.rs
