vortex_array/arrays/bool/vtable/
mod.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use std::hash::Hash;
5use std::hash::Hasher;
6
7use prost::Message;
8use vortex_error::VortexExpect;
9use vortex_error::VortexResult;
10use vortex_error::vortex_bail;
11use vortex_error::vortex_ensure;
12use vortex_error::vortex_panic;
13use vortex_session::VortexSession;
14
15use crate::ArrayRef;
16use crate::ExecutionCtx;
17use crate::ExecutionResult;
18use crate::array::Array;
19use crate::array::ArrayParts;
20use crate::array::ArrayView;
21use crate::array::VTable;
22use crate::array::child_to_validity;
23use crate::arrays::bool::BoolData;
24use crate::arrays::bool::array::SLOT_NAMES;
25use crate::buffer::BufferHandle;
26use crate::builders::ArrayBuilder;
27use crate::builders::BoolBuilder;
28use crate::dtype::DType;
29use crate::serde::ArrayChildren;
30use crate::validity::Validity;
31mod canonical;
32mod kernel;
33mod operations;
34mod validity;
35
36use vortex_session::registry::CachedId;
37
38use crate::EqMode;
39use crate::array::ArrayId;
40use crate::arrays::bool::compute::rules::RULES;
41use crate::hash::ArrayEq;
42use crate::hash::ArrayHash;
43
44/// A [`Bool`]-encoded Vortex array.
45pub type BoolArray = Array<Bool>;
46
47pub(crate) fn initialize(session: &VortexSession) {
48    kernel::initialize(session);
49}
50
51#[derive(prost::Message)]
52pub struct BoolMetadata {
53    // The offset in bits must be <8
54    #[prost(uint32, tag = "1")]
55    pub offset: u32,
56}
57
58impl ArrayHash for BoolData {
59    fn array_hash<H: Hasher>(&self, state: &mut H, accuracy: EqMode) {
60        self.bits.array_hash(state, accuracy);
61        self.meta.offset().hash(state);
62    }
63}
64
65impl ArrayEq for BoolData {
66    fn array_eq(&self, other: &Self, accuracy: EqMode) -> bool {
67        self.meta.offset() == other.meta.offset() && self.bits.array_eq(&other.bits, accuracy)
68    }
69}
70
71impl VTable for Bool {
72    type TypedArrayData = BoolData;
73
74    type OperationsVTable = Self;
75    type ValidityVTable = Self;
76
77    fn id(&self) -> ArrayId {
78        static ID: CachedId = CachedId::new("vortex.bool");
79        *ID
80    }
81
82    fn nbuffers(_array: ArrayView<'_, Self>) -> usize {
83        1
84    }
85
86    fn buffer(array: ArrayView<'_, Self>, idx: usize) -> BufferHandle {
87        match idx {
88            0 => array.bits.clone(),
89            _ => vortex_panic!("BoolArray buffer index {idx} out of bounds"),
90        }
91    }
92
93    fn buffer_name(_array: ArrayView<'_, Self>, idx: usize) -> Option<String> {
94        match idx {
95            0 => Some("bits".to_string()),
96            _ => None,
97        }
98    }
99
100    fn with_buffers(
101        &self,
102        array: ArrayView<'_, Self>,
103        buffers: &[BufferHandle],
104    ) -> VortexResult<ArrayParts<Self>> {
105        vortex_ensure!(
106            buffers.len() == 1,
107            "Expected 1 buffer, got {}",
108            buffers.len()
109        );
110        let mut data = array.data().clone();
111        data.bits = buffers[0].clone();
112        Ok(
113            ArrayParts::new(self.clone(), array.dtype().clone(), array.len(), data)
114                .with_slots(array.slots().iter().cloned().collect()),
115        )
116    }
117
118    fn serialize(
119        array: ArrayView<'_, Self>,
120        _session: &VortexSession,
121    ) -> VortexResult<Option<Vec<u8>>> {
122        let offset = array.meta.offset();
123        assert!(offset < 8, "Offset must be <8, got {offset}");
124        Ok(Some(
125            BoolMetadata {
126                offset: u32::try_from(offset).vortex_expect("checked"),
127            }
128            .encode_to_vec(),
129        ))
130    }
131
132    fn validate(
133        &self,
134        data: &BoolData,
135        dtype: &DType,
136        len: usize,
137        slots: &[Option<ArrayRef>],
138    ) -> VortexResult<()> {
139        let DType::Bool(nullability) = dtype else {
140            vortex_bail!("Expected bool dtype, got {dtype:?}");
141        };
142        vortex_ensure!(
143            data.bits.len() * 8 >= data.meta.offset() + len,
144            "BoolArray buffer with offset {} cannot back outer length {} (buffer bits = {})",
145            data.meta.offset(),
146            len,
147            data.bits.len() * 8
148        );
149
150        let validity = child_to_validity(slots[0].as_ref(), *nullability);
151        if let Some(validity_len) = validity.maybe_len() {
152            vortex_ensure!(
153                validity_len == len,
154                "BoolArray validity len {} does not match outer length {}",
155                validity_len,
156                len
157            );
158        }
159
160        Ok(())
161    }
162
163    fn deserialize(
164        &self,
165        dtype: &DType,
166        len: usize,
167        metadata: &[u8],
168        buffers: &[BufferHandle],
169        children: &dyn ArrayChildren,
170        _session: &VortexSession,
171    ) -> VortexResult<ArrayParts<Self>> {
172        let metadata = BoolMetadata::decode(metadata)?;
173        if buffers.len() != 1 {
174            vortex_bail!("Expected 1 buffer, got {}", buffers.len());
175        }
176
177        let validity = if children.is_empty() {
178            Validity::from(dtype.nullability())
179        } else if children.len() == 1 {
180            let validity = children.get(0, &Validity::DTYPE, len)?;
181            Validity::Array(validity)
182        } else {
183            vortex_bail!("Expected 0 or 1 child, got {}", children.len());
184        };
185
186        let buffer = buffers[0].clone();
187        let slots = BoolData::make_slots(&validity, len);
188        let data = BoolData::try_new_from_handle(buffer, metadata.offset as usize, len, validity)?;
189        Ok(ArrayParts::new(self.clone(), dtype.clone(), len, data).with_slots(slots))
190    }
191
192    fn slot_name(_array: ArrayView<'_, Self>, idx: usize) -> String {
193        SLOT_NAMES[idx].to_string()
194    }
195
196    fn append_to_builder(
197        array: ArrayView<'_, Self>,
198        builder: &mut dyn ArrayBuilder,
199        ctx: &mut ExecutionCtx,
200    ) -> VortexResult<()> {
201        if let Some(builder) = builder.as_any_mut().downcast_mut::<BoolBuilder>() {
202            return builder.append_bool_array(&array.into_owned(), ctx);
203        }
204
205        builder.extend_from_array(array.as_ref());
206        Ok(())
207    }
208
209    fn execute(array: Array<Self>, _ctx: &mut ExecutionCtx) -> VortexResult<ExecutionResult> {
210        Ok(ExecutionResult::done(array))
211    }
212
213    fn reduce_parent(
214        array: ArrayView<'_, Self>,
215        parent: &ArrayRef,
216        child_idx: usize,
217    ) -> VortexResult<Option<ArrayRef>> {
218        RULES.evaluate(array, parent, child_idx)
219    }
220}
221
222#[derive(Clone, Debug)]
223pub struct Bool;
224
225#[cfg(test)]
226mod tests {
227    use vortex_buffer::ByteBufferMut;
228    use vortex_session::registry::ReadContext;
229
230    use crate::ArrayContext;
231    use crate::IntoArray;
232    use crate::VortexSessionExecute;
233    use crate::array_session;
234    use crate::arrays::BoolArray;
235    use crate::assert_arrays_eq;
236    use crate::serde::SerializeOptions;
237    use crate::serde::SerializedArray;
238
239    #[test]
240    fn test_nullable_bool_serde_roundtrip() {
241        let session = array_session();
242        let mut ctx = session.create_execution_ctx();
243        let array = BoolArray::from_iter([Some(true), None, Some(false), None]);
244        let dtype = array.dtype().clone();
245        let len = array.len();
246
247        let array_ctx = ArrayContext::empty();
248        let serialized = array
249            .clone()
250            .into_array()
251            .serialize(&array_ctx, &session, &SerializeOptions::default())
252            .unwrap();
253
254        let mut concat = ByteBufferMut::empty();
255        for buf in serialized {
256            concat.extend_from_slice(buf.as_ref());
257        }
258        let parts = SerializedArray::try_from(concat.freeze()).unwrap();
259        let decoded = parts
260            .decode(&dtype, len, &ReadContext::new(array_ctx.to_ids()), &session)
261            .unwrap();
262
263        assert_arrays_eq!(decoded, array, &mut ctx);
264    }
265}
vortex_array/arrays/bool/vtable/mod.rs

vortex_array/arrays/bool/vtable/
mod.rs
