Skip to main content

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::EqMode;
20use crate::ExecutionCtx;
21use crate::ExecutionResult;
22use crate::IntoArray;
23use crate::array::Array;
24use crate::array::ArrayId;
25use crate::array::ArrayParts;
26use crate::array::ArrayView;
27use crate::array::VTable;
28use crate::array::with_empty_buffers;
29use crate::arrays::list::ListArrayExt;
30use crate::arrays::list::ListData;
31use crate::arrays::list::array::ELEMENTS_SLOT;
32use crate::arrays::list::array::NUM_SLOTS;
33use crate::arrays::list::array::OFFSETS_SLOT;
34use crate::arrays::list::array::SLOT_NAMES;
35use crate::arrays::list::compute::rules::PARENT_RULES;
36use crate::arrays::listview::list_view_from_list;
37use crate::buffer::BufferHandle;
38use crate::builders::ArrayBuilder;
39use crate::dtype::DType;
40use crate::dtype::Nullability;
41use crate::dtype::PType;
42use crate::serde::ArrayChildren;
43use crate::validity::Validity;
44mod operations;
45mod validity;
46/// A [`List`]-encoded Vortex array.
47pub type ListArray = Array<List>;
48
49#[derive(Clone, prost::Message)]
50pub struct ListMetadata {
51    #[prost(uint64, tag = "1")]
52    elements_len: u64,
53    #[prost(enumeration = "PType", tag = "2")]
54    offset_ptype: i32,
55}
56
57impl ArrayHash for ListData {
58    fn array_hash<H: Hasher>(&self, _state: &mut H, _accuracy: EqMode) {}
59}
60
61impl ArrayEq for ListData {
62    fn array_eq(&self, _other: &Self, _accuracy: EqMode) -> bool {
63        true
64    }
65}
66
67impl VTable for List {
68    type TypedArrayData = ListData;
69
70    type OperationsVTable = Self;
71    type ValidityVTable = Self;
72    fn id(&self) -> ArrayId {
73        static ID: CachedId = CachedId::new("vortex.list");
74        *ID
75    }
76
77    fn nbuffers(_array: ArrayView<'_, Self>) -> usize {
78        0
79    }
80
81    fn buffer(_array: ArrayView<'_, Self>, idx: usize) -> BufferHandle {
82        vortex_panic!("ListArray buffer index {idx} out of bounds")
83    }
84
85    fn buffer_name(_array: ArrayView<'_, Self>, idx: usize) -> Option<String> {
86        vortex_panic!("ListArray buffer_name index {idx} out of bounds")
87    }
88
89    fn with_buffers(
90        &self,
91        array: ArrayView<'_, Self>,
92        buffers: &[BufferHandle],
93    ) -> VortexResult<ArrayParts<Self>> {
94        with_empty_buffers(self, array, buffers)
95    }
96
97    fn reduce_parent(
98        array: ArrayView<'_, Self>,
99        parent: &ArrayRef,
100        child_idx: usize,
101    ) -> VortexResult<Option<ArrayRef>> {
102        PARENT_RULES.evaluate(array, parent, child_idx)
103    }
104
105    fn serialize(
106        array: ArrayView<'_, Self>,
107        _session: &VortexSession,
108    ) -> VortexResult<Option<Vec<u8>>> {
109        Ok(Some(
110            ListMetadata {
111                elements_len: array.elements().len() as u64,
112                offset_ptype: PType::try_from(array.offsets().dtype())? as i32,
113            }
114            .encode_to_vec(),
115        ))
116    }
117
118    fn validate(
119        &self,
120        _data: &ListData,
121        dtype: &DType,
122        len: usize,
123        slots: &[Option<ArrayRef>],
124    ) -> VortexResult<()> {
125        vortex_ensure!(
126            slots.len() == NUM_SLOTS,
127            "ListArray expected {NUM_SLOTS} slots, found {}",
128            slots.len()
129        );
130        let elements = slots[ELEMENTS_SLOT]
131            .as_ref()
132            .vortex_expect("ListArray elements slot");
133        let offsets = slots[OFFSETS_SLOT]
134            .as_ref()
135            .vortex_expect("ListArray offsets slot");
136        vortex_ensure!(
137            offsets.len().saturating_sub(1) == len,
138            "ListArray length {} does not match outer length {}",
139            offsets.len().saturating_sub(1),
140            len
141        );
142
143        let actual_dtype = DType::List(Arc::new(elements.dtype().clone()), dtype.nullability());
144        vortex_ensure!(
145            &actual_dtype == dtype,
146            "ListArray dtype {} does not match outer dtype {}",
147            actual_dtype,
148            dtype
149        );
150
151        Ok(())
152    }
153
154    fn deserialize(
155        &self,
156        dtype: &DType,
157        len: usize,
158        metadata: &[u8],
159
160        _buffers: &[BufferHandle],
161        children: &dyn ArrayChildren,
162        _session: &VortexSession,
163    ) -> VortexResult<ArrayParts<Self>> {
164        let metadata = ListMetadata::decode(metadata)?;
165        let validity = if children.len() == 2 {
166            Validity::from(dtype.nullability())
167        } else if children.len() == 3 {
168            let validity = children.get(2, &Validity::DTYPE, len)?;
169            Validity::Array(validity)
170        } else {
171            vortex_bail!("Expected 2 or 3 children, got {}", children.len());
172        };
173
174        let DType::List(element_dtype, _) = &dtype else {
175            vortex_bail!("Expected List dtype, got {:?}", dtype);
176        };
177        let elements = children.get(
178            0,
179            element_dtype.as_ref(),
180            usize::try_from(metadata.elements_len)?,
181        )?;
182
183        let offsets = children.get(
184            1,
185            &DType::Primitive(metadata.offset_ptype(), Nullability::NonNullable),
186            len + 1,
187        )?;
188
189        let data = ListData::try_build(elements.clone(), offsets.clone(), validity.clone())?;
190        let slots = ListData::make_slots(&elements, &offsets, &validity, len);
191        Ok(ArrayParts::new(self.clone(), dtype.clone(), len, data).with_slots(slots))
192    }
193
194    fn slot_name(_array: ArrayView<'_, Self>, idx: usize) -> String {
195        SLOT_NAMES[idx].to_string()
196    }
197
198    fn execute(array: Array<Self>, ctx: &mut ExecutionCtx) -> VortexResult<ExecutionResult> {
199        Ok(ExecutionResult::done(
200            list_view_from_list(array, ctx)?.into_array(),
201        ))
202    }
203
204    fn append_to_builder(
205        array: ArrayView<'_, Self>,
206        builder: &mut dyn ArrayBuilder,
207        ctx: &mut ExecutionCtx,
208    ) -> VortexResult<()> {
209        builder.append_list_array(array, ctx)
210    }
211}
212
213#[derive(Clone, Debug)]
214pub struct List;