Skip to main content

vortex_array/arrays/varbinview/vtable/
mod.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use 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_ensure_eq;
13use vortex_error::vortex_err;
14use vortex_error::vortex_panic;
15use vortex_session::VortexSession;
16use vortex_session::registry::CachedId;
17
18use crate::ArrayParts;
19use crate::ArrayRef;
20use crate::EqMode;
21use crate::ExecutionCtx;
22use crate::ExecutionResult;
23use crate::VortexSessionExecute;
24use crate::array::Array;
25use crate::array::ArrayId;
26use crate::array::ArrayView;
27use crate::array::VTable;
28use crate::arrays::varbinview::BinaryView;
29use crate::arrays::varbinview::VarBinViewData;
30use crate::arrays::varbinview::array::VarBinViewSlots;
31use crate::arrays::varbinview::compute::rules::PARENT_RULES;
32use crate::buffer::BufferHandle;
33use crate::builders::ArrayBuilder;
34use crate::builders::VarBinViewBuilder;
35use crate::dtype::DType;
36use crate::hash::ArrayEq;
37use crate::hash::ArrayHash;
38use crate::match_each_varbin_builder;
39use crate::serde::ArrayChildren;
40use crate::validity::Validity;
41mod kernel;
42mod operations;
43mod validity;
44/// A [`VarBinView`]-encoded Vortex array.
45pub type VarBinViewArray = Array<VarBinView>;
46
47pub(crate) fn initialize(session: &VortexSession) {
48    kernel::initialize(session);
49}
50
51#[derive(Clone, Debug)]
52pub struct VarBinView;
53
54impl ArrayHash for VarBinViewData {
55    fn array_hash<H: Hasher>(&self, state: &mut H, accuracy: EqMode) {
56        for buffer in self.buffers.iter() {
57            buffer.array_hash(state, accuracy);
58        }
59        self.views.array_hash(state, accuracy);
60    }
61}
62
63impl ArrayEq for VarBinViewData {
64    fn array_eq(&self, other: &Self, accuracy: EqMode) -> bool {
65        self.buffers.len() == other.buffers.len()
66            && self
67                .buffers
68                .iter()
69                .zip(other.buffers.iter())
70                .all(|(a, b)| a.array_eq(b, accuracy))
71            && self.views.array_eq(&other.views, accuracy)
72    }
73}
74
75impl VTable for VarBinView {
76    type TypedArrayData = VarBinViewData;
77
78    type OperationsVTable = Self;
79    type ValidityVTable = Self;
80
81    fn id(&self) -> ArrayId {
82        static ID: CachedId = CachedId::new("vortex.varbinview");
83        *ID
84    }
85
86    fn nbuffers(array: ArrayView<'_, Self>) -> usize {
87        array.data_buffers().len() + 1
88    }
89
90    fn validate(
91        &self,
92        data: &VarBinViewData,
93        dtype: &DType,
94        len: usize,
95        slots: &[Option<ArrayRef>],
96    ) -> VortexResult<()> {
97        vortex_ensure_eq!(slots.len(), VarBinViewSlots::COUNT);
98        vortex_ensure_eq!(
99            data.len(),
100            len,
101            "VarBinViewArray length does not match outer length",
102        );
103        vortex_ensure!(
104            matches!(dtype, DType::Binary(_) | DType::Utf8(_)),
105            "VarBinViewArray dtype must be binary or utf8, got {dtype}"
106        );
107        Ok(())
108    }
109
110    fn buffer(array: ArrayView<'_, Self>, idx: usize) -> BufferHandle {
111        let ndata = array.data_buffers().len();
112        if idx < ndata {
113            array.data_buffers()[idx].clone()
114        } else if idx == ndata {
115            array.views_handle().clone()
116        } else {
117            vortex_panic!("VarBinViewArray buffer index {idx} out of bounds")
118        }
119    }
120
121    fn buffer_name(array: ArrayView<'_, Self>, idx: usize) -> Option<String> {
122        let ndata = array.data_buffers().len();
123        if idx < ndata {
124            Some(format!("buffer_{idx}"))
125        } else if idx == ndata {
126            Some("views".to_string())
127        } else {
128            vortex_panic!("VarBinViewArray buffer_name index {idx} out of bounds")
129        }
130    }
131
132    fn with_buffers(
133        &self,
134        array: ArrayView<'_, Self>,
135        buffers: &[BufferHandle],
136    ) -> VortexResult<ArrayParts<Self>> {
137        let Some((views, data_buffers)) = buffers.split_last() else {
138            vortex_bail!("Expected at least 1 buffer, got 0");
139        };
140        let data = VarBinViewData::try_new_handle(
141            views.clone(),
142            Arc::from(data_buffers.to_vec()),
143            array.dtype().clone(),
144            array.validity()?,
145        )?;
146        Ok(ArrayParts::new(
147            self.clone(),
148            array.dtype().clone(),
149            array.len(),
150            data,
151            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 any buffer is on device, skip host validation and use try_new_handle.
204        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(
213                self.clone(),
214                dtype.clone(),
215                len,
216                data,
217                slots,
218            ));
219        }
220
221        let data_buffers = data_handles
222            .iter()
223            .map(|b| b.as_host().clone())
224            .collect::<Vec<_>>();
225        let views = Buffer::<BinaryView>::from_byte_buffer(views_handle.clone().as_host().clone());
226
227        let data = VarBinViewData::try_new(
228            views,
229            Arc::from(data_buffers),
230            dtype.clone(),
231            validity.clone(),
232            &mut session.create_execution_ctx(),
233        )?;
234        let slots = VarBinViewData::make_slots(&validity, len);
235        Ok(ArrayParts::new(
236            self.clone(),
237            dtype.clone(),
238            len,
239            data,
240            slots,
241        ))
242    }
243
244    fn slot_name(_array: ArrayView<'_, Self>, idx: usize) -> String {
245        VarBinViewSlots::NAMES[idx].to_string()
246    }
247
248    fn reduce_parent(
249        array: ArrayView<'_, Self>,
250        parent: &ArrayRef,
251        child_idx: usize,
252    ) -> VortexResult<Option<ArrayRef>> {
253        PARENT_RULES.evaluate(array, parent, child_idx)
254    }
255
256    fn append_to_builder(
257        array: ArrayView<'_, Self>,
258        builder: &mut dyn ArrayBuilder,
259        ctx: &mut ExecutionCtx,
260    ) -> VortexResult<()> {
261        if let Some(builder) = builder.as_any_mut().downcast_mut::<VarBinViewBuilder>() {
262            return builder.append_varbinview_array(&array.into_owned(), ctx);
263        }
264        if let Some(result) =
265            match_each_varbin_builder!(builder, |builder| builder.append_varbinview(array, ctx))
266        {
267            return result;
268        }
269        vortex_bail!("append_to_builder for VarBinView requires a variable-binary builder")
270    }
271
272    fn execute(array: Array<Self>, _ctx: &mut ExecutionCtx) -> VortexResult<ExecutionResult> {
273        Ok(ExecutionResult::done(array))
274    }
275}
276
277#[cfg(test)]
278mod tests {
279    use vortex_buffer::ByteBufferMut;
280    use vortex_session::registry::ReadContext;
281
282    use super::*;
283    use crate::ArrayContext;
284    use crate::IntoArray;
285    use crate::VortexSessionExecute;
286    use crate::array_session;
287    use crate::assert_arrays_eq;
288    use crate::serde::SerializeOptions;
289    use crate::serde::SerializedArray;
290
291    #[test]
292    fn test_nullable_varbinview_serde_roundtrip() {
293        let array = VarBinViewArray::from_iter_nullable_str([
294            Some("hello"),
295            None,
296            Some("world"),
297            None,
298            Some("a moderately long string for testing"),
299        ]);
300        let dtype = array.dtype().clone();
301        let len = array.len();
302
303        let session = array_session();
304        let mut ctx = session.create_execution_ctx();
305        let array_ctx = ArrayContext::empty();
306        let serialized = array
307            .clone()
308            .into_array()
309            .serialize(&array_ctx, &session, &SerializeOptions::default())
310            .unwrap();
311
312        let mut concat = ByteBufferMut::empty();
313        for buf in serialized {
314            concat.extend_from_slice(buf.as_ref());
315        }
316        let parts = SerializedArray::try_from(concat.freeze()).unwrap();
317        let decoded = parts
318            .decode(&dtype, len, &ReadContext::new(array_ctx.to_ids()), &session)
319            .unwrap();
320
321        assert_arrays_eq!(decoded, array, &mut ctx);
322    }
323}