16728097825

Committed 04 Aug 2025 04:00PM UTC coverage: 48.355% (-35.1%) from 83.429%

Build # 16728097825

Build Type

Pull #4108

github

Committed by

web-flow

Commit Message

Merge 1b2d27fd8 into 649ba9576

Pull Request Pull Request #4108: perf[vortex-array]: use all_valid instead of `invalid_count() == 0`

Coverage Stats

1 of 1 new or added line in 1 file covered. (100.0%)

11596 existing lines in 378 files now uncovered.

18635 of 38538 relevant lines covered (48.35%)

151786.4 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

50.7

/vortex-array/src/builders/bool.rs

// SPDX-License-Identifier: Apache-2.0
// SPDX-FileCopyrightText: Copyright the Vortex contributors

use std::any::Any;

use arrow_buffer::BooleanBufferBuilder;
use vortex_dtype::{DType, Nullability};
use vortex_error::{VortexResult, vortex_bail};
use vortex_mask::Mask;

use crate::arrays::BoolArray;
use crate::builders::ArrayBuilder;
use crate::builders::lazy_validity_builder::LazyNullBufferBuilder;
use crate::{Array, ArrayRef, Canonical, IntoArray};

pub struct BoolBuilder {
    inner: BooleanBufferBuilder,
    nulls: LazyNullBufferBuilder,
    nullability: Nullability,
    dtype: DType,
}

impl BoolBuilder {
    pub fn new(nullability: Nullability) -> Self {
        Self::with_capacity(nullability, 1024) // Same as Arrow builders
    }

    pub fn with_capacity(nullability: Nullability, capacity: usize) -> Self {
        Self {
            inner: BooleanBufferBuilder::new(capacity),
            nulls: LazyNullBufferBuilder::new(capacity),
            nullability,
            dtype: DType::Bool(nullability),
        }
    }

    pub fn append_value(&mut self, value: bool) {
        self.append_values(value, 1)
    }

    pub fn append_values(&mut self, value: bool, n: usize) {
        self.inner.append_n(n, value);
        self.nulls.append_n_non_nulls(n)
    }

    pub fn append_option(&mut self, value: Option<bool>) {
        match value {
            Some(value) => self.append_value(value),
            None => self.append_null(),
        }
    }
}

impl ArrayBuilder for BoolBuilder {
    fn as_any(&self) -> &dyn Any {
        self
    }

    fn as_any_mut(&mut self) -> &mut dyn Any {
        self
    }

    fn dtype(&self) -> &DType {
        &self.dtype
    }

    fn len(&self) -> usize {
        self.inner.len()
    }

    fn append_zeros(&mut self, n: usize) {
        self.append_values(false, n)
    }

    fn append_nulls(&mut self, n: usize) {
        self.inner.append_n(n, false);
        self.nulls.append_n_nulls(n)
    }

    fn extend_from_array(&mut self, array: &dyn Array) -> VortexResult<()> {
        let array = array.to_canonical()?;
        let Canonical::Bool(array) = array else {
            vortex_bail!("Expected Canonical::Bool, found {:?}", array);
        };

        self.inner.append_buffer(array.boolean_buffer());
        self.nulls.append_validity_mask(array.validity_mask()?);

        Ok(())
    }

    fn ensure_capacity(&mut self, capacity: usize) {
        if capacity > self.inner.capacity() {
            self.nulls.ensure_capacity(capacity);
            self.inner.reserve(capacity - self.inner.capacity());
        }
    }

    fn set_validity(&mut self, validity: Mask) {
        self.nulls = LazyNullBufferBuilder::new(validity.len());
        self.nulls.append_validity_mask(validity);
    }

    fn finish(&mut self) -> ArrayRef {
        assert_eq!(
            self.nulls.len(),
            self.inner.len(),
            "Null count and value count should match when calling BoolBuilder::finish."
        );

        BoolArray::new(
            self.inner.finish(),
            self.nulls.finish_with_nullability(self.nullability),
        )
        .into_array()
    }
}

#[cfg(test)]
mod tests {
    use rand::prelude::StdRng;
    use rand::{Rng, SeedableRng};

    use crate::array::Array;
    use crate::arrays::{BoolArray, ChunkedArray};
    use crate::builders::builder_with_capacity;
    use crate::canonical::ToCanonical;
    use crate::vtable::ValidityHelper;
    use crate::{ArrayRef, IntoArray};
    fn make_opt_bool_chunks(len: usize, chunk_count: usize) -> ArrayRef {
        let mut rng = StdRng::seed_from_u64(0);

        (0..chunk_count)
            .map(|_| {
                BoolArray::from_iter((0..len).map(|_| match rng.random_range::<u8, _>(0..=2) {
                    0 => Some(false),
                    1 => Some(true),
                    2 => None,
                    _ => unreachable!(),
                }))
                .into_array()
            })
            .collect::<ChunkedArray>()
            .into_array()
    }

    #[test]
    fn tests() {
        let len = 1000;
        let chunk_count = 10;
        let chunk = make_opt_bool_chunks(len, chunk_count);

        let mut builder = builder_with_capacity(chunk.dtype(), len * chunk_count);
        chunk.clone().append_to_builder(builder.as_mut()).unwrap();
        let canon_into = builder
            .finish()
            .to_canonical()
            .unwrap()
            .into_bool()
            .unwrap();

        let into_canon = chunk.to_bool().unwrap();

        assert_eq!(canon_into.validity(), into_canon.validity());
        assert_eq!(canon_into.boolean_buffer(), into_canon.boolean_buffer());
    }
}

1	// SPDX-License-Identifier: Apache-2.0
2	// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4	use std::any::Any;
5
6	use arrow_buffer::BooleanBufferBuilder;
7	use vortex_dtype::{DType, Nullability};
8	use vortex_error::{VortexResult, vortex_bail};
9	use vortex_mask::Mask;
10
11	use crate::arrays::BoolArray;
12	use crate::builders::ArrayBuilder;
13	use crate::builders::lazy_validity_builder::LazyNullBufferBuilder;
14	use crate::{Array, ArrayRef, Canonical, IntoArray};
15
16	pub struct BoolBuilder {
17	inner: BooleanBufferBuilder,
18	nulls: LazyNullBufferBuilder,
19	nullability: Nullability,
20	dtype: DType,
21	}
22
23	impl BoolBuilder {
24	pub fn new(nullability: Nullability) -> Self {	×
25	Self::with_capacity(nullability, 1024) // Same as Arrow builders	×
26	}	×
27
28	pub fn with_capacity(nullability: Nullability, capacity: usize) -> Self {	240✔
29	Self {	240✔
30	inner: BooleanBufferBuilder::new(capacity),	240✔
31	nulls: LazyNullBufferBuilder::new(capacity),	240✔
32	nullability,	240✔
33	dtype: DType::Bool(nullability),	240✔
34	}	240✔
35	}	240✔
36
37	pub fn append_value(&mut self, value: bool) {	2,104✔
38	self.append_values(value, 1)	2,104✔
39	}	2,104✔
40
41	pub fn append_values(&mut self, value: bool, n: usize) {	2,104✔
42	self.inner.append_n(n, value);	2,104✔
43	self.nulls.append_n_non_nulls(n)	2,104✔
44	}	2,104✔
45
UNCOV 46	pub fn append_option(&mut self, value: Option<bool>) {	×
UNCOV 47	match value {	×
UNCOV 48	Some(value) => self.append_value(value),	×
49	None => self.append_null(),	×
50	}
UNCOV 51	}	×
52	}
53
54	impl ArrayBuilder for BoolBuilder {
55	fn as_any(&self) -> &dyn Any {	×
56	self	×
57	}	×
58
UNCOV 59	fn as_any_mut(&mut self) -> &mut dyn Any {	×
UNCOV 60	self	×
UNCOV 61	}	×
62
UNCOV 63	fn dtype(&self) -> &DType {	×
UNCOV 64	&self.dtype	×
UNCOV 65	}	×
66
UNCOV 67	fn len(&self) -> usize {	×
UNCOV 68	self.inner.len()	×
UNCOV 69	}	×
70
71	fn append_zeros(&mut self, n: usize) {	×
72	self.append_values(false, n)	×
73	}	×
74
75	fn append_nulls(&mut self, n: usize) {	×
76	self.inner.append_n(n, false);	×
77	self.nulls.append_n_nulls(n)	×
78	}	×
79
80	fn extend_from_array(&mut self, array: &dyn Array) -> VortexResult<()> {	8✔
81	let array = array.to_canonical()?;	8✔
82	let Canonical::Bool(array) = array else {	8✔
83	vortex_bail!("Expected Canonical::Bool, found {:?}", array);	×
84	};
85
86	self.inner.append_buffer(array.boolean_buffer());	8✔
87	self.nulls.append_validity_mask(array.validity_mask()?);	8✔
88
89	Ok(())	8✔
90	}	8✔
91
92	fn ensure_capacity(&mut self, capacity: usize) {	×
93	if capacity > self.inner.capacity() {	×
94	self.nulls.ensure_capacity(capacity);	×
95	self.inner.reserve(capacity - self.inner.capacity());	×
96	}	×
97	}	×
98
99	fn set_validity(&mut self, validity: Mask) {	8✔
100	self.nulls = LazyNullBufferBuilder::new(validity.len());	8✔
101	self.nulls.append_validity_mask(validity);	8✔
102	}	8✔
103
104	fn finish(&mut self) -> ArrayRef {	240✔
105	assert_eq!(	240✔
106	self.nulls.len(),	240✔
107	self.inner.len(),	240✔
108	"Null count and value count should match when calling BoolBuilder::finish."	×
109	);
110
111	BoolArray::new(	240✔
112	self.inner.finish(),	240✔
113	self.nulls.finish_with_nullability(self.nullability),	240✔
114	)	240✔
115	.into_array()	240✔
116	}	240✔
117	}
118
119	#[cfg(test)]
120	mod tests {
121	use rand::prelude::StdRng;
122	use rand::{Rng, SeedableRng};
123
124	use crate::array::Array;
125	use crate::arrays::{BoolArray, ChunkedArray};
126	use crate::builders::builder_with_capacity;
127	use crate::canonical::ToCanonical;
128	use crate::vtable::ValidityHelper;
129	use crate::{ArrayRef, IntoArray};
130	fn make_opt_bool_chunks(len: usize, chunk_count: usize) -> ArrayRef {
131	let mut rng = StdRng::seed_from_u64(0);
132
133	(0..chunk_count)
134	.map(\|_\| {
135	BoolArray::from_iter((0..len).map(\|_\| match rng.random_range::<u8, _>(0..=2) {
136	0 => Some(false),
137	1 => Some(true),
138	2 => None,
139	_ => unreachable!(),
140	}))
141	.into_array()
142	})
143	.collect::<ChunkedArray>()
144	.into_array()
145	}
146
147	#[test]
148	fn tests() {
149	let len = 1000;
150	let chunk_count = 10;
151	let chunk = make_opt_bool_chunks(len, chunk_count);
152
153	let mut builder = builder_with_capacity(chunk.dtype(), len * chunk_count);
154	chunk.clone().append_to_builder(builder.as_mut()).unwrap();
155	let canon_into = builder
156	.finish()
157	.to_canonical()
158	.unwrap()
159	.into_bool()
160	.unwrap();
161
162	let into_canon = chunk.to_bool().unwrap();
163
164	assert_eq!(canon_into.validity(), into_canon.validity());
165	assert_eq!(canon_into.boolean_buffer(), into_canon.boolean_buffer());
166	}
167	}

vortex-data / vortex / 16728097825

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous