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