Skip to main content

vortex_array/builders/
bool.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use std::any::Any;
5use std::mem;
6
7use vortex_buffer::BitBufferMut;
8use vortex_error::VortexResult;
9use vortex_error::vortex_ensure;
10use vortex_mask::Mask;
11
12use crate::ArrayRef;
13use crate::ExecutionCtx;
14use crate::IntoArray;
15use crate::arrays::BoolArray;
16use crate::arrays::bool::BoolArrayExt;
17use crate::builders::ArrayBuilder;
18use crate::builders::DEFAULT_BUILDER_CAPACITY;
19use crate::builders::LazyBitBufferBuilder;
20use crate::canonical::Canonical;
21use crate::dtype::DType;
22use crate::dtype::Nullability;
23use crate::scalar::Scalar;
24
25pub struct BoolBuilder {
26    dtype: DType,
27    inner: BitBufferMut,
28    nulls: LazyBitBufferBuilder,
29}
30
31impl BoolBuilder {
32    pub fn new(nullability: Nullability) -> Self {
33        Self::with_capacity(nullability, DEFAULT_BUILDER_CAPACITY)
34    }
35
36    pub fn with_capacity(nullability: Nullability, capacity: usize) -> Self {
37        Self {
38            inner: BitBufferMut::with_capacity(capacity),
39            nulls: LazyBitBufferBuilder::new(capacity),
40            dtype: DType::Bool(nullability),
41        }
42    }
43
44    /// Appends a boolean value to the builder.
45    pub fn append_value(&mut self, value: bool) {
46        self.append_values(value, 1)
47    }
48
49    /// Appends the same boolean value multiple times to the builder.
50    ///
51    /// This method appends the given boolean value `n` times.
52    pub fn append_values(&mut self, value: bool, n: usize) {
53        self.inner.append_n(value, n);
54        self.nulls.append_n_non_nulls(n)
55    }
56
57    /// Finishes the builder directly into a [`BoolArray`].
58    pub fn finish_into_bool(&mut self) -> BoolArray {
59        assert_eq!(
60            self.nulls.len(),
61            self.inner.len(),
62            "Null count and value count should match when calling BoolBuilder::finish."
63        );
64
65        BoolArray::new(
66            mem::take(&mut self.inner).freeze(),
67            self.nulls.finish_with_nullability(self.dtype.nullability()),
68        )
69    }
70
71    pub(crate) fn append_bool_array(
72        &mut self,
73        array: &BoolArray,
74        ctx: &mut ExecutionCtx,
75    ) -> VortexResult<()> {
76        self.inner.append_buffer(&array.to_bit_buffer());
77        self.nulls
78            .append_validity_mask(&BoolArrayExt::validity(array).execute_mask(array.len(), ctx)?);
79        Ok(())
80    }
81}
82
83impl ArrayBuilder for BoolBuilder {
84    fn as_any(&self) -> &dyn Any {
85        self
86    }
87
88    fn as_any_mut(&mut self) -> &mut dyn Any {
89        self
90    }
91
92    fn dtype(&self) -> &DType {
93        &self.dtype
94    }
95
96    fn len(&self) -> usize {
97        self.inner.len()
98    }
99
100    fn append_zeros(&mut self, n: usize) {
101        self.append_values(false, n)
102    }
103
104    unsafe fn append_nulls_unchecked(&mut self, n: usize) {
105        self.inner.append_n(false, n);
106        self.nulls.append_n_nulls(n)
107    }
108
109    fn append_scalar(&mut self, scalar: &Scalar) -> VortexResult<()> {
110        vortex_ensure!(
111            scalar.dtype() == self.dtype(),
112            "BoolBuilder expected scalar with dtype {}, got {}",
113            self.dtype(),
114            scalar.dtype()
115        );
116
117        match scalar.as_bool().value() {
118            Some(value) => self.append_value(value),
119            None => self.append_null(),
120        }
121
122        Ok(())
123    }
124
125    fn reserve_exact(&mut self, additional: usize) {
126        self.inner.reserve(additional);
127        self.nulls.reserve_exact(additional);
128    }
129
130    unsafe fn set_validity_unchecked(&mut self, validity: Mask) {
131        self.nulls = LazyBitBufferBuilder::from_validity_mask(validity);
132    }
133
134    fn finish(&mut self) -> ArrayRef {
135        self.finish_into_bool().into_array()
136    }
137
138    fn finish_into_canonical(&mut self, _ctx: &mut ExecutionCtx) -> Canonical {
139        Canonical::Bool(self.finish_into_bool())
140    }
141}
142
143#[cfg(test)]
144mod tests {
145    use rand::RngExt;
146    use rand::SeedableRng;
147    use rand::prelude::StdRng;
148    use vortex_error::VortexResult;
149
150    use crate::ArrayRef;
151    use crate::IntoArray;
152    use crate::VortexSessionExecute;
153    use crate::array_session;
154    use crate::arrays::ChunkedArray;
155    use crate::arrays::bool::BoolArrayExt;
156    use crate::assert_arrays_eq;
157    use crate::builders::ArrayBuilder;
158    use crate::builders::BoolBuilder;
159    use crate::builders::bool::BoolArray;
160    use crate::builders::builder_with_capacity;
161    use crate::dtype::DType;
162    use crate::dtype::Nullability;
163    use crate::scalar::Scalar;
164
165    fn make_opt_bool_chunks(len: usize, chunk_count: usize) -> ArrayRef {
166        let mut rng = StdRng::seed_from_u64(0);
167
168        (0..chunk_count)
169            .map(|_| {
170                BoolArray::from_iter((0..len).map(|_| match rng.random_range::<u8, _>(0..=2) {
171                    0 => Some(false),
172                    1 => Some(true),
173                    2 => None,
174                    _ => unreachable!(),
175                }))
176                .into_array()
177            })
178            .collect::<ChunkedArray>()
179            .into_array()
180    }
181
182    #[test]
183    fn tests() -> VortexResult<()> {
184        let len = 1000;
185        let chunk_count = 10;
186        let chunk = make_opt_bool_chunks(len, chunk_count);
187
188        let mut ctx = array_session().create_execution_ctx();
189        let mut builder = builder_with_capacity(chunk.dtype(), len * chunk_count);
190        chunk
191            .clone()
192            .append_to_builder(builder.as_mut(), &mut ctx)?;
193
194        let canon_into = builder.finish().execute::<BoolArray>(&mut ctx)?;
195        let into_canon = chunk.clone().execute::<BoolArray>(&mut ctx)?;
196
197        assert!(canon_into.validity()?.mask_eq(
198            &into_canon.validity()?,
199            canon_into.len(),
200            &mut ctx
201        )?);
202        assert_eq!(canon_into.to_bit_buffer(), into_canon.to_bit_buffer());
203        Ok(())
204    }
205
206    #[test]
207    fn test_append_scalar() {
208        let mut ctx = array_session().create_execution_ctx();
209        let mut builder = BoolBuilder::with_capacity(Nullability::Nullable, 10);
210
211        // Test appending true value.
212        let true_scalar = Scalar::bool(true, Nullability::Nullable);
213        builder.append_scalar(&true_scalar).unwrap();
214
215        // Test appending false value.
216        let false_scalar = Scalar::bool(false, Nullability::Nullable);
217        builder.append_scalar(&false_scalar).unwrap();
218
219        // Test appending null value.
220        let null_scalar = Scalar::null(DType::Bool(Nullability::Nullable));
221        builder.append_scalar(&null_scalar).unwrap();
222
223        let array = builder.finish_into_bool();
224        let expected = BoolArray::from_iter([Some(true), Some(false), None]);
225        assert_arrays_eq!(&array, &expected, &mut ctx);
226
227        // Test wrong dtype error.
228        let mut builder = BoolBuilder::with_capacity(Nullability::NonNullable, 10);
229        let wrong_scalar = Scalar::from(42i32);
230        assert!(builder.append_scalar(&wrong_scalar).is_err());
231    }
232}