vortex_array/arrays/varbinview/vtable/
mod.rs1use std::hash::Hasher;
5use std::mem::size_of;
6use std::sync::Arc;
7
8use vortex_buffer::Buffer;
9use vortex_error::VortexResult;
10use vortex_error::vortex_bail;
11use vortex_error::vortex_ensure;
12use vortex_error::vortex_err;
13use vortex_error::vortex_panic;
14use vortex_session::VortexSession;
15use vortex_session::registry::CachedId;
16
17use crate::ArrayParts;
18use crate::ArrayRef;
19use crate::EqMode;
20use crate::ExecutionCtx;
21use crate::ExecutionResult;
22use crate::array::Array;
23use crate::array::ArrayId;
24use crate::array::ArrayView;
25use crate::array::VTable;
26use crate::arrays::varbinview::BinaryView;
27use crate::arrays::varbinview::VarBinViewData;
28use crate::arrays::varbinview::array::VarBinViewSlots;
29use crate::arrays::varbinview::compute::rules::PARENT_RULES;
30use crate::buffer::BufferHandle;
31use crate::builders::ArrayBuilder;
32use crate::builders::VarBinViewBuilder;
33use crate::dtype::DType;
34use crate::hash::ArrayEq;
35use crate::hash::ArrayHash;
36use crate::serde::ArrayChildren;
37use crate::validity::Validity;
38mod kernel;
39mod operations;
40mod validity;
41pub type VarBinViewArray = Array<VarBinView>;
43
44pub(crate) fn initialize(session: &VortexSession) {
45 kernel::initialize(session);
46}
47
48#[derive(Clone, Debug)]
49pub struct VarBinView;
50
51impl ArrayHash for VarBinViewData {
52 fn array_hash<H: Hasher>(&self, state: &mut H, accuracy: EqMode) {
53 for buffer in self.buffers.iter() {
54 buffer.array_hash(state, accuracy);
55 }
56 self.views.array_hash(state, accuracy);
57 }
58}
59
60impl ArrayEq for VarBinViewData {
61 fn array_eq(&self, other: &Self, accuracy: EqMode) -> bool {
62 self.buffers.len() == other.buffers.len()
63 && self
64 .buffers
65 .iter()
66 .zip(other.buffers.iter())
67 .all(|(a, b)| a.array_eq(b, accuracy))
68 && self.views.array_eq(&other.views, accuracy)
69 }
70}
71
72impl VTable for VarBinView {
73 type TypedArrayData = VarBinViewData;
74
75 type OperationsVTable = Self;
76 type ValidityVTable = Self;
77
78 fn id(&self) -> ArrayId {
79 static ID: CachedId = CachedId::new("vortex.varbinview");
80 *ID
81 }
82
83 fn nbuffers(array: ArrayView<'_, Self>) -> usize {
84 array.data_buffers().len() + 1
85 }
86
87 fn validate(
88 &self,
89 data: &VarBinViewData,
90 dtype: &DType,
91 len: usize,
92 slots: &[Option<ArrayRef>],
93 ) -> VortexResult<()> {
94 vortex_ensure!(
95 slots.len() == VarBinViewSlots::COUNT,
96 "VarBinViewArray expected {} slots, found {}",
97 VarBinViewSlots::COUNT,
98 slots.len()
99 );
100 vortex_ensure!(
101 data.len() == len,
102 "VarBinViewArray length {} does not match outer length {}",
103 data.len(),
104 len
105 );
106 vortex_ensure!(
107 matches!(dtype, DType::Binary(_) | DType::Utf8(_)),
108 "VarBinViewArray dtype must be binary or utf8, got {dtype}"
109 );
110 Ok(())
111 }
112
113 fn buffer(array: ArrayView<'_, Self>, idx: usize) -> BufferHandle {
114 let ndata = array.data_buffers().len();
115 if idx < ndata {
116 array.data_buffers()[idx].clone()
117 } else if idx == ndata {
118 array.views_handle().clone()
119 } else {
120 vortex_panic!("VarBinViewArray buffer index {idx} out of bounds")
121 }
122 }
123
124 fn buffer_name(array: ArrayView<'_, Self>, idx: usize) -> Option<String> {
125 let ndata = array.data_buffers().len();
126 if idx < ndata {
127 Some(format!("buffer_{idx}"))
128 } else if idx == ndata {
129 Some("views".to_string())
130 } else {
131 vortex_panic!("VarBinViewArray buffer_name index {idx} out of bounds")
132 }
133 }
134
135 fn with_buffers(
136 &self,
137 array: ArrayView<'_, Self>,
138 buffers: &[BufferHandle],
139 ) -> VortexResult<ArrayParts<Self>> {
140 let Some((views, data_buffers)) = buffers.split_last() else {
141 vortex_bail!("Expected at least 1 buffer, got 0");
142 };
143 let data = VarBinViewData::try_new_handle(
144 views.clone(),
145 Arc::from(data_buffers.to_vec()),
146 array.dtype().clone(),
147 array.validity()?,
148 )?;
149 Ok(
150 ArrayParts::new(self.clone(), array.dtype().clone(), array.len(), data)
151 .with_slots(array.slots().iter().cloned().collect()),
152 )
153 }
154
155 fn serialize(
156 _array: ArrayView<'_, Self>,
157 _session: &VortexSession,
158 ) -> VortexResult<Option<Vec<u8>>> {
159 Ok(Some(vec![]))
160 }
161
162 fn deserialize(
163 &self,
164 dtype: &DType,
165 len: usize,
166 metadata: &[u8],
167
168 buffers: &[BufferHandle],
169 children: &dyn ArrayChildren,
170 _session: &VortexSession,
171 ) -> VortexResult<ArrayParts<Self>> {
172 if !metadata.is_empty() {
173 vortex_bail!(
174 "VarBinViewArray expects empty metadata, got {} bytes",
175 metadata.len()
176 );
177 }
178 let Some((views_handle, data_handles)) = buffers.split_last() else {
179 vortex_bail!("Expected at least 1 buffer, got 0");
180 };
181
182 let validity = if children.is_empty() {
183 Validity::from(dtype.nullability())
184 } else if children.len() == 1 {
185 let validity = children.get(0, &Validity::DTYPE, len)?;
186 Validity::Array(validity)
187 } else {
188 vortex_bail!("Expected 0 or 1 children, got {}", children.len());
189 };
190
191 let views_nbytes = views_handle.len();
192 let expected_views_nbytes = len
193 .checked_mul(size_of::<BinaryView>())
194 .ok_or_else(|| vortex_err!("views byte length overflow for len={len}"))?;
195 if views_nbytes != expected_views_nbytes {
196 vortex_bail!(
197 "Expected views buffer length {} bytes, got {} bytes",
198 expected_views_nbytes,
199 views_nbytes
200 );
201 }
202
203 if buffers.iter().any(|b| b.is_on_device()) {
205 let data = VarBinViewData::try_new_handle(
206 views_handle.clone(),
207 Arc::from(data_handles.to_vec()),
208 dtype.clone(),
209 validity.clone(),
210 )?;
211 let slots = VarBinViewData::make_slots(&validity, len);
212 return Ok(ArrayParts::new(self.clone(), dtype.clone(), len, data).with_slots(slots));
213 }
214
215 let data_buffers = data_handles
216 .iter()
217 .map(|b| b.as_host().clone())
218 .collect::<Vec<_>>();
219 let views = Buffer::<BinaryView>::from_byte_buffer(views_handle.clone().as_host().clone());
220
221 let data = VarBinViewData::try_new(
222 views,
223 Arc::from(data_buffers),
224 dtype.clone(),
225 validity.clone(),
226 )?;
227 let slots = VarBinViewData::make_slots(&validity, len);
228 Ok(ArrayParts::new(self.clone(), dtype.clone(), len, data).with_slots(slots))
229 }
230
231 fn slot_name(_array: ArrayView<'_, Self>, idx: usize) -> String {
232 VarBinViewSlots::NAMES[idx].to_string()
233 }
234
235 fn reduce_parent(
236 array: ArrayView<'_, Self>,
237 parent: &ArrayRef,
238 child_idx: usize,
239 ) -> VortexResult<Option<ArrayRef>> {
240 PARENT_RULES.evaluate(array, parent, child_idx)
241 }
242
243 fn append_to_builder(
244 array: ArrayView<'_, Self>,
245 builder: &mut dyn ArrayBuilder,
246 ctx: &mut ExecutionCtx,
247 ) -> VortexResult<()> {
248 let Some(builder) = builder.as_any_mut().downcast_mut::<VarBinViewBuilder>() else {
249 vortex_bail!("append_to_builder for VarBinView requires a VarBinViewBuilder");
250 };
251 builder.append_varbinview_array(&array.into_owned(), ctx)
252 }
253
254 fn execute(array: Array<Self>, _ctx: &mut ExecutionCtx) -> VortexResult<ExecutionResult> {
255 Ok(ExecutionResult::done(array))
256 }
257}
258
259#[cfg(test)]
260mod tests {
261 use vortex_buffer::ByteBufferMut;
262 use vortex_session::registry::ReadContext;
263
264 use super::*;
265 use crate::ArrayContext;
266 use crate::IntoArray;
267 use crate::VortexSessionExecute;
268 use crate::array_session;
269 use crate::assert_arrays_eq;
270 use crate::serde::SerializeOptions;
271 use crate::serde::SerializedArray;
272
273 #[test]
274 fn test_nullable_varbinview_serde_roundtrip() {
275 let array = VarBinViewArray::from_iter_nullable_str([
276 Some("hello"),
277 None,
278 Some("world"),
279 None,
280 Some("a moderately long string for testing"),
281 ]);
282 let dtype = array.dtype().clone();
283 let len = array.len();
284
285 let session = array_session();
286 let mut ctx = session.create_execution_ctx();
287 let array_ctx = ArrayContext::empty();
288 let serialized = array
289 .clone()
290 .into_array()
291 .serialize(&array_ctx, &session, &SerializeOptions::default())
292 .unwrap();
293
294 let mut concat = ByteBufferMut::empty();
295 for buf in serialized {
296 concat.extend_from_slice(buf.as_ref());
297 }
298 let parts = SerializedArray::try_from(concat.freeze()).unwrap();
299 let decoded = parts
300 .decode(&dtype, len, &ReadContext::new(array_ctx.to_ids()), &session)
301 .unwrap();
302
303 assert_arrays_eq!(decoded, array, &mut ctx);
304 }
305}