vortex_array/arrays/list/vtable/
mod.rs1use 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::ListArraySlotsExt;
30use crate::arrays::list::ListData;
31use crate::arrays::list::ListSlots;
32use crate::arrays::list::compute::rules::PARENT_RULES;
33use crate::arrays::listview::list_view_from_list;
34use crate::buffer::BufferHandle;
35use crate::builders::ArrayBuilder;
36use crate::dtype::DType;
37use crate::dtype::Nullability;
38use crate::dtype::PType;
39use crate::serde::ArrayChildren;
40use crate::validity::Validity;
41mod operations;
42mod validity;
43pub type ListArray = Array<List>;
45
46#[derive(Clone, prost::Message)]
47pub struct ListMetadata {
48 #[prost(uint64, tag = "1")]
49 elements_len: u64,
50 #[prost(enumeration = "PType", tag = "2")]
51 offset_ptype: i32,
52}
53
54impl ArrayHash for ListData {
55 fn array_hash<H: Hasher>(&self, _state: &mut H, _accuracy: EqMode) {}
56}
57
58impl ArrayEq for ListData {
59 fn array_eq(&self, _other: &Self, _accuracy: EqMode) -> bool {
60 true
61 }
62}
63
64impl VTable for List {
65 type TypedArrayData = ListData;
66
67 type OperationsVTable = Self;
68 type ValidityVTable = Self;
69 fn id(&self) -> ArrayId {
70 static ID: CachedId = CachedId::new("vortex.list");
71 *ID
72 }
73
74 fn nbuffers(_array: ArrayView<'_, Self>) -> usize {
75 0
76 }
77
78 fn buffer(_array: ArrayView<'_, Self>, idx: usize) -> BufferHandle {
79 vortex_panic!("ListArray buffer index {idx} out of bounds")
80 }
81
82 fn buffer_name(_array: ArrayView<'_, Self>, idx: usize) -> Option<String> {
83 vortex_panic!("ListArray buffer_name index {idx} out of bounds")
84 }
85
86 fn with_buffers(
87 &self,
88 array: ArrayView<'_, Self>,
89 buffers: &[BufferHandle],
90 ) -> VortexResult<ArrayParts<Self>> {
91 with_empty_buffers(self, array, buffers)
92 }
93
94 fn reduce_parent(
95 array: ArrayView<'_, Self>,
96 parent: &ArrayRef,
97 child_idx: usize,
98 ) -> VortexResult<Option<ArrayRef>> {
99 PARENT_RULES.evaluate(array, parent, child_idx)
100 }
101
102 fn serialize(
103 array: ArrayView<'_, Self>,
104 _session: &VortexSession,
105 ) -> VortexResult<Option<Vec<u8>>> {
106 Ok(Some(
107 ListMetadata {
108 elements_len: array.elements().len() as u64,
109 offset_ptype: PType::try_from(array.offsets().dtype())? as i32,
110 }
111 .encode_to_vec(),
112 ))
113 }
114
115 fn validate(
116 &self,
117 _data: &ListData,
118 dtype: &DType,
119 len: usize,
120 slots: &[Option<ArrayRef>],
121 ) -> VortexResult<()> {
122 vortex_ensure!(
123 slots.len() == ListSlots::COUNT,
124 "ListArray expected {} slots, found {}",
125 ListSlots::COUNT,
126 slots.len()
127 );
128 let elements = slots[ListSlots::ELEMENTS]
129 .as_ref()
130 .vortex_expect("ListArray elements slot");
131 let offsets = slots[ListSlots::OFFSETS]
132 .as_ref()
133 .vortex_expect("ListArray offsets slot");
134 vortex_ensure!(
135 offsets.len().saturating_sub(1) == len,
136 "ListArray length {} does not match outer length {}",
137 offsets.len().saturating_sub(1),
138 len
139 );
140
141 let actual_dtype = DType::List(Arc::new(elements.dtype().clone()), dtype.nullability());
142 vortex_ensure!(
143 &actual_dtype == dtype,
144 "ListArray dtype {} does not match outer dtype {}",
145 actual_dtype,
146 dtype
147 );
148
149 Ok(())
150 }
151
152 fn deserialize(
153 &self,
154 dtype: &DType,
155 len: usize,
156 metadata: &[u8],
157
158 _buffers: &[BufferHandle],
159 children: &dyn ArrayChildren,
160 _session: &VortexSession,
161 ) -> VortexResult<ArrayParts<Self>> {
162 let metadata = ListMetadata::decode(metadata)?;
163 let validity = if children.len() == 2 {
164 Validity::from(dtype.nullability())
165 } else if children.len() == 3 {
166 let validity = children.get(2, &Validity::DTYPE, len)?;
167 Validity::Array(validity)
168 } else {
169 vortex_bail!("Expected 2 or 3 children, got {}", children.len());
170 };
171
172 let DType::List(element_dtype, _) = &dtype else {
173 vortex_bail!("Expected List dtype, got {:?}", dtype);
174 };
175 let elements = children.get(
176 0,
177 element_dtype.as_ref(),
178 usize::try_from(metadata.elements_len)?,
179 )?;
180
181 let offsets = children.get(
182 1,
183 &DType::Primitive(metadata.offset_ptype(), Nullability::NonNullable),
184 len + 1,
185 )?;
186
187 let data = ListData::try_build(elements.clone(), offsets.clone(), validity.clone())?;
188 let slots = ListData::make_slots(&elements, &offsets, &validity, len);
189 Ok(ArrayParts::new(self.clone(), dtype.clone(), len, data).with_slots(slots))
190 }
191
192 fn slot_name(_array: ArrayView<'_, Self>, idx: usize) -> String {
193 ListSlots::NAMES[idx].to_string()
194 }
195
196 fn execute(array: Array<Self>, ctx: &mut ExecutionCtx) -> VortexResult<ExecutionResult> {
197 Ok(ExecutionResult::done(
198 list_view_from_list(array, ctx)?.into_array(),
199 ))
200 }
201
202 fn append_to_builder(
203 array: ArrayView<'_, Self>,
204 builder: &mut dyn ArrayBuilder,
205 ctx: &mut ExecutionCtx,
206 ) -> VortexResult<()> {
207 builder.append_list_array(array, ctx)
208 }
209}
210
211#[derive(Clone, Debug)]
212pub struct List;