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_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::VortexSessionExecute;
23use crate::array::Array;
24use crate::array::ArrayId;
25use crate::array::ArrayView;
26use crate::array::VTable;
27use crate::arrays::varbinview::BinaryView;
28use crate::arrays::varbinview::VarBinViewData;
29use crate::arrays::varbinview::array::VarBinViewSlots;
30use crate::arrays::varbinview::compute::rules::PARENT_RULES;
31use crate::buffer::BufferHandle;
32use crate::builders::ArrayBuilder;
33use crate::builders::VarBinViewBuilder;
34use crate::dtype::DType;
35use crate::hash::ArrayEq;
36use crate::hash::ArrayHash;
37use crate::match_each_varbin_builder;
38use crate::serde::ArrayChildren;
39use crate::validity::Validity;
40mod kernel;
41mod operations;
42mod validity;
43/// A [`VarBinView`]-encoded Vortex array.
44pub type VarBinViewArray = Array<VarBinView>;
45
46pub(crate) fn initialize(session: &VortexSession) {
47    kernel::initialize(session);
48}
49
50#[derive(Clone, Debug)]
51pub struct VarBinView;
52
53impl ArrayHash for VarBinViewData {
54    fn array_hash<H: Hasher>(&self, state: &mut H, accuracy: EqMode) {
55        for buffer in self.buffers.iter() {
56            buffer.array_hash(state, accuracy);
57        }
58        self.views.array_hash(state, accuracy);
59    }
60}
61
62impl ArrayEq for VarBinViewData {
63    fn array_eq(&self, other: &Self, accuracy: EqMode) -> bool {
64        self.buffers.len() == other.buffers.len()
65            && self
66                .buffers
67                .iter()
68                .zip(other.buffers.iter())
69                .all(|(a, b)| a.array_eq(b, accuracy))
70            && self.views.array_eq(&other.views, accuracy)
71    }
72}
73
74impl VTable for VarBinView {
75    type TypedArrayData = VarBinViewData;
76
77    type OperationsVTable = Self;
78    type ValidityVTable = Self;
79
80    fn id(&self) -> ArrayId {
81        static ID: CachedId = CachedId::new("vortex.varbinview");
82        *ID
83    }
84
85    fn nbuffers(array: ArrayView<'_, Self>) -> usize {
86        array.data_buffers().len() + 1
87    }
88
89    fn validate(
90        &self,
91        data: &VarBinViewData,
92        dtype: &DType,
93        len: usize,
94        slots: &[Option<ArrayRef>],
95    ) -> VortexResult<()> {
96        vortex_ensure!(
97            slots.len() == VarBinViewSlots::COUNT,
98            "VarBinViewArray expected {} slots, found {}",
99            VarBinViewSlots::COUNT,
100            slots.len()
101        );
102        vortex_ensure!(
103            data.len() == len,
104            "VarBinViewArray length {} does not match outer length {}",
105            data.len(),
106            len
107        );
108        vortex_ensure!(
109            matches!(dtype, DType::Binary(_) | DType::Utf8(_)),
110            "VarBinViewArray dtype must be binary or utf8, got {dtype}"
111        );
112        Ok(())
113    }
114
115    fn buffer(array: ArrayView<'_, Self>, idx: usize) -> BufferHandle {
116        let ndata = array.data_buffers().len();
117        if idx < ndata {
118            array.data_buffers()[idx].clone()
119        } else if idx == ndata {
120            array.views_handle().clone()
121        } else {
122            vortex_panic!("VarBinViewArray buffer index {idx} out of bounds")
123        }
124    }
125
126    fn buffer_name(array: ArrayView<'_, Self>, idx: usize) -> Option<String> {
127        let ndata = array.data_buffers().len();
128        if idx < ndata {
129            Some(format!("buffer_{idx}"))
130        } else if idx == ndata {
131            Some("views".to_string())
132        } else {
133            vortex_panic!("VarBinViewArray buffer_name index {idx} out of bounds")
134        }
135    }
136
137    fn with_buffers(
138        &self,
139        array: ArrayView<'_, Self>,
140        buffers: &[BufferHandle],
141    ) -> VortexResult<ArrayParts<Self>> {
142        let Some((views, data_buffers)) = buffers.split_last() else {
143            vortex_bail!("Expected at least 1 buffer, got 0");
144        };
145        let data = VarBinViewData::try_new_handle(
146            views.clone(),
147            Arc::from(data_buffers.to_vec()),
148            array.dtype().clone(),
149            array.validity()?,
150        )?;
151        Ok(
152            ArrayParts::new(self.clone(), array.dtype().clone(), array.len(), data)
153                .with_slots(array.slots().iter().cloned().collect()),
154        )
155    }
156
157    fn serialize(
158        _array: ArrayView<'_, Self>,
159        _session: &VortexSession,
160    ) -> VortexResult<Option<Vec<u8>>> {
161        Ok(Some(vec![]))
162    }
163
164    fn deserialize(
165        &self,
166        dtype: &DType,
167        len: usize,
168        metadata: &[u8],
169
170        buffers: &[BufferHandle],
171        children: &dyn ArrayChildren,
172        session: &VortexSession,
173    ) -> VortexResult<ArrayParts<Self>> {
174        if !metadata.is_empty() {
175            vortex_bail!(
176                "VarBinViewArray expects empty metadata, got {} bytes",
177                metadata.len()
178            );
179        }
180        let Some((views_handle, data_handles)) = buffers.split_last() else {
181            vortex_bail!("Expected at least 1 buffer, got 0");
182        };
183
184        let validity = if children.is_empty() {
185            Validity::from(dtype.nullability())
186        } else if children.len() == 1 {
187            let validity = children.get(0, &Validity::DTYPE, len)?;
188            Validity::Array(validity)
189        } else {
190            vortex_bail!("Expected 0 or 1 children, got {}", children.len());
191        };
192
193        let views_nbytes = views_handle.len();
194        let expected_views_nbytes = len
195            .checked_mul(size_of::<BinaryView>())
196            .ok_or_else(|| vortex_err!("views byte length overflow for len={len}"))?;
197        if views_nbytes != expected_views_nbytes {
198            vortex_bail!(
199                "Expected views buffer length {} bytes, got {} bytes",
200                expected_views_nbytes,
201                views_nbytes
202            );
203        }
204
205        // If any buffer is on device, skip host validation and use try_new_handle.
206        if buffers.iter().any(|b| b.is_on_device()) {
207            let data = VarBinViewData::try_new_handle(
208                views_handle.clone(),
209                Arc::from(data_handles.to_vec()),
210                dtype.clone(),
211                validity.clone(),
212            )?;
213            let slots = VarBinViewData::make_slots(&validity, len);
214            return Ok(ArrayParts::new(self.clone(), dtype.clone(), len, data).with_slots(slots));
215        }
216
217        let data_buffers = data_handles
218            .iter()
219            .map(|b| b.as_host().clone())
220            .collect::<Vec<_>>();
221        let views = Buffer::<BinaryView>::from_byte_buffer(views_handle.clone().as_host().clone());
222
223        let data = VarBinViewData::try_new(
224            views,
225            Arc::from(data_buffers),
226            dtype.clone(),
227            validity.clone(),
228            &mut session.create_execution_ctx(),
229        )?;
230        let slots = VarBinViewData::make_slots(&validity, len);
231        Ok(ArrayParts::new(self.clone(), dtype.clone(), len, data).with_slots(slots))
232    }
233
234    fn slot_name(_array: ArrayView<'_, Self>, idx: usize) -> String {
235        VarBinViewSlots::NAMES[idx].to_string()
236    }
237
238    fn reduce_parent(
239        array: ArrayView<'_, Self>,
240        parent: &ArrayRef,
241        child_idx: usize,
242    ) -> VortexResult<Option<ArrayRef>> {
243        PARENT_RULES.evaluate(array, parent, child_idx)
244    }
245
246    fn append_to_builder(
247        array: ArrayView<'_, Self>,
248        builder: &mut dyn ArrayBuilder,
249        ctx: &mut ExecutionCtx,
250    ) -> VortexResult<()> {
251        if let Some(builder) = builder.as_any_mut().downcast_mut::<VarBinViewBuilder>() {
252            return builder.append_varbinview_array(&array.into_owned(), ctx);
253        }
254        if let Some(result) =
255            match_each_varbin_builder!(builder, |builder| builder.append_varbinview(array, ctx))
256        {
257            return result;
258        }
259        vortex_bail!("append_to_builder for VarBinView requires a variable-binary builder")
260    }
261
262    fn execute(array: Array<Self>, _ctx: &mut ExecutionCtx) -> VortexResult<ExecutionResult> {
263        Ok(ExecutionResult::done(array))
264    }
265}
266
267#[cfg(test)]
268mod tests {
269    use vortex_buffer::ByteBufferMut;
270    use vortex_session::registry::ReadContext;
271
272    use super::*;
273    use crate::ArrayContext;
274    use crate::IntoArray;
275    use crate::VortexSessionExecute;
276    use crate::array_session;
277    use crate::assert_arrays_eq;
278    use crate::serde::SerializeOptions;
279    use crate::serde::SerializedArray;
280
281    #[test]
282    fn test_nullable_varbinview_serde_roundtrip() {
283        let array = VarBinViewArray::from_iter_nullable_str([
284            Some("hello"),
285            None,
286            Some("world"),
287            None,
288            Some("a moderately long string for testing"),
289        ]);
290        let dtype = array.dtype().clone();
291        let len = array.len();
292
293        let session = array_session();
294        let mut ctx = session.create_execution_ctx();
295        let array_ctx = ArrayContext::empty();
296        let serialized = array
297            .clone()
298            .into_array()
299            .serialize(&array_ctx, &session, &SerializeOptions::default())
300            .unwrap();
301
302        let mut concat = ByteBufferMut::empty();
303        for buf in serialized {
304            concat.extend_from_slice(buf.as_ref());
305        }
306        let parts = SerializedArray::try_from(concat.freeze()).unwrap();
307        let decoded = parts
308            .decode(&dtype, len, &ReadContext::new(array_ctx.to_ids()), &session)
309            .unwrap();
310
311        assert_arrays_eq!(decoded, array, &mut ctx);
312    }
313}