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

vortex_array/arrays/varbinview/vtable/
mod.rs
