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

vortex_array/arrays/list/vtable/
mod.rs
