17073077835

Committed 19 Aug 2025 02:40PM UTC coverage: 24.083%. First build

Build # 17073077835

Build Type

Pull #4177

github

Committed by

web-flow

Commit Message

Merge b42e5758f into 431a8f2b5

Pull Request Pull Request #4177: feat: ArrayOperations infallible, eager validation + new_unchecked

Run Details

197 of 1455 new or added lines in 154 files covered. (13.54%)

8646 of 35901 relevant lines covered (24.08%)

142.28 hits per line

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

0.0

/vortex-btrblocks/src/float/dictionary.rs

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

//! Float-specific dictionary encoding implementation.

use vortex_array::IntoArray;
use vortex_array::arrays::PrimitiveArray;
use vortex_array::validity::Validity;
use vortex_array::vtable::ValidityHelper;
use vortex_buffer::Buffer;
use vortex_dict::DictArray;
use vortex_dtype::half::f16;

use crate::float::stats::{ErasedDistinctValues, FloatStats};

macro_rules! typed_encode {
    ($stats:ident, $typed:ident, $validity:ident, $typ:ty) => {{
        let values: Buffer<$typ> = $typed.values.iter().map(|x| x.0).collect();

        let max_code = values.len();
        let codes = if max_code <= u8::MAX as usize {
            let buf =
                <DictEncoder as Encode<$typ, u8>>::encode(&values, $stats.src.as_slice::<$typ>());
            PrimitiveArray::new(buf, $validity.clone()).into_array()
        } else if max_code <= u16::MAX as usize {
            let buf =
                <DictEncoder as Encode<$typ, u16>>::encode(&values, $stats.src.as_slice::<$typ>());
            PrimitiveArray::new(buf, $validity.clone()).into_array()
        } else {
            let buf =
                <DictEncoder as Encode<$typ, u32>>::encode(&values, $stats.src.as_slice::<$typ>());
            PrimitiveArray::new(buf, $validity.clone()).into_array()
        };

        let values_validity = match $validity {
            Validity::NonNullable => Validity::NonNullable,
            _ => Validity::AllValid,
        };
        let values = PrimitiveArray::new(values, values_validity).into_array();

        // SAFETY: enforced by the DictEncoder
        unsafe { DictArray::new_unchecked(codes, values) }
    }};
}

pub fn dictionary_encode(stats: &FloatStats) -> DictArray {
    let validity = stats.src.validity();
    match &stats.distinct_values {
        ErasedDistinctValues::F16(typed) => typed_encode!(stats, typed, validity, f16),
        ErasedDistinctValues::F32(typed) => typed_encode!(stats, typed, validity, f32),
        ErasedDistinctValues::F64(typed) => typed_encode!(stats, typed, validity, f64),
    }
}

struct DictEncoder;

trait Encode<T, I> {
    /// Using the distinct value set, turn the values into a set of codes.
    fn encode(distinct: &[T], values: &[T]) -> Buffer<I>;
}

macro_rules! impl_encode {
    ($typ:ty, $utyp:ty) => { impl_encode!($typ, $utyp, u8, u16, u32); };
    ($typ:ty, $utyp:ty, $($ityp:ty),+) => {
        $(
        impl Encode<$typ, $ityp> for DictEncoder {
            #[allow(clippy::cast_possible_truncation)]
            fn encode(distinct: &[$typ], values: &[$typ]) -> Buffer<$ityp> {
                let mut codes =
                    vortex_utils::aliases::hash_map::HashMap::<$utyp, $ityp>::with_capacity(
                        distinct.len(),
                    );
                for (code, &value) in distinct.iter().enumerate() {
                    codes.insert(value.to_bits(), code as $ityp);
                }

                let mut output = vortex_buffer::BufferMut::with_capacity(values.len());
                for value in values {
                    // Any code lookups which fail are for nulls, so their value
                    // does not matter.
                    output.push(codes.get(&value.to_bits()).copied().unwrap_or_default());
                }

                return output.freeze();
            }
        }
        )*
    };
}

impl_encode!(f16, u16);
impl_encode!(f32, u32);
impl_encode!(f64, u64);

#[cfg(test)]
mod tests {
    use vortex_array::arrays::{BoolArray, PrimitiveArray};
    use vortex_array::validity::Validity;
    use vortex_array::{Array, IntoArray, ToCanonical};
    use vortex_buffer::buffer;

    use crate::CompressorStats;
    use crate::float::dictionary::dictionary_encode;
    use crate::float::stats::FloatStats;

    #[test]
    fn test_float_dict_encode() {
        // Create an array that has some nulls
        let values = buffer![1f32, 2f32, 2f32, 0f32, 1f32];
        let validity =
            Validity::Array(BoolArray::from_iter([true, true, true, false, true]).into_array());
        let array = PrimitiveArray::new(values, validity);

        let stats = FloatStats::generate(&array);
        let dict_array = dictionary_encode(&stats);
        assert_eq!(dict_array.values().len(), 2);
        assert_eq!(dict_array.codes().len(), 5);

        let undict = dict_array.to_primitive().unwrap();

        // We just use code zero but it doesn't really matter.
        // We can just shove a whole validity buffer in there instead.
        assert_eq!(undict.as_slice::<f32>(), &[1f32, 2f32, 2f32, 1f32, 1f32]);
    }
}

1	// SPDX-License-Identifier: Apache-2.0
2	// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4	//! Float-specific dictionary encoding implementation.
5
6	use vortex_array::IntoArray;
7	use vortex_array::arrays::PrimitiveArray;
8	use vortex_array::validity::Validity;
9	use vortex_array::vtable::ValidityHelper;
10	use vortex_buffer::Buffer;
11	use vortex_dict::DictArray;
12	use vortex_dtype::half::f16;
13
14	use crate::float::stats::{ErasedDistinctValues, FloatStats};
15
16	macro_rules! typed_encode {
17	($stats:ident, $typed:ident, $validity:ident, $typ:ty) => {{
18	let values: Buffer<$typ> = $typed.values.iter().map(\|x\| x.0).collect();
19
20	let max_code = values.len();
21	let codes = if max_code <= u8::MAX as usize {
22	let buf =
23	<DictEncoder as Encode<$typ, u8>>::encode(&values, $stats.src.as_slice::<$typ>());
24	PrimitiveArray::new(buf, $validity.clone()).into_array()
25	} else if max_code <= u16::MAX as usize {
26	let buf =
27	<DictEncoder as Encode<$typ, u16>>::encode(&values, $stats.src.as_slice::<$typ>());
28	PrimitiveArray::new(buf, $validity.clone()).into_array()
29	} else {
30	let buf =
31	<DictEncoder as Encode<$typ, u32>>::encode(&values, $stats.src.as_slice::<$typ>());
32	PrimitiveArray::new(buf, $validity.clone()).into_array()
33	};
34
35	let values_validity = match $validity {
36	Validity::NonNullable => Validity::NonNullable,
37	_ => Validity::AllValid,
38	};
39	let values = PrimitiveArray::new(values, values_validity).into_array();
40
41	// SAFETY: enforced by the DictEncoder
42	unsafe { DictArray::new_unchecked(codes, values) }
43	}};
44	}
45
NEW 46	pub fn dictionary_encode(stats: &FloatStats) -> DictArray {	×
47	let validity = stats.src.validity();	×
48	match &stats.distinct_values {	×
49	ErasedDistinctValues::F16(typed) => typed_encode!(stats, typed, validity, f16),	×
50	ErasedDistinctValues::F32(typed) => typed_encode!(stats, typed, validity, f32),	×
51	ErasedDistinctValues::F64(typed) => typed_encode!(stats, typed, validity, f64),	×
52	}
53	}	×
54
55	struct DictEncoder;
56
57	trait Encode<T, I> {
58	/// Using the distinct value set, turn the values into a set of codes.
59	fn encode(distinct: &[T], values: &[T]) -> Buffer<I>;
60	}
61
62	macro_rules! impl_encode {
63	($typ:ty, $utyp:ty) => { impl_encode!($typ, $utyp, u8, u16, u32); };
64	($typ:ty, $utyp:ty, $($ityp:ty),+) => {
65	$(
66	impl Encode<$typ, $ityp> for DictEncoder {
67	#[allow(clippy::cast_possible_truncation)]
68	fn encode(distinct: &[$typ], values: &[$typ]) -> Buffer<$ityp> {	×
69	let mut codes =	×
70	vortex_utils::aliases::hash_map::HashMap::<$utyp, $ityp>::with_capacity(	×
71	distinct.len(),	×
72	);
73	for (code, &value) in distinct.iter().enumerate() {	×
74	codes.insert(value.to_bits(), code as $ityp);	×
75	}	×
76
77	let mut output = vortex_buffer::BufferMut::with_capacity(values.len());	×
78	for value in values {	×
79	// Any code lookups which fail are for nulls, so their value	×
80	// does not matter.	×
81	output.push(codes.get(&value.to_bits()).copied().unwrap_or_default());	×
82	}	×
83
84	return output.freeze();	×
85	}	×
86	}
87	)*
88	};
89	}
90
91	impl_encode!(f16, u16);
92	impl_encode!(f32, u32);
93	impl_encode!(f64, u64);
94
95	#[cfg(test)]
96	mod tests {
97	use vortex_array::arrays::{BoolArray, PrimitiveArray};
98	use vortex_array::validity::Validity;
99	use vortex_array::{Array, IntoArray, ToCanonical};
100	use vortex_buffer::buffer;
101
102	use crate::CompressorStats;
103	use crate::float::dictionary::dictionary_encode;
104	use crate::float::stats::FloatStats;
105
106	#[test]
107	fn test_float_dict_encode() {
108	// Create an array that has some nulls
109	let values = buffer![1f32, 2f32, 2f32, 0f32, 1f32];
110	let validity =
111	Validity::Array(BoolArray::from_iter([true, true, true, false, true]).into_array());
112	let array = PrimitiveArray::new(values, validity);
113
114	let stats = FloatStats::generate(&array);
115	let dict_array = dictionary_encode(&stats);
116	assert_eq!(dict_array.values().len(), 2);
117	assert_eq!(dict_array.codes().len(), 5);
118
119	let undict = dict_array.to_primitive().unwrap();
120
121	// We just use code zero but it doesn't really matter.
122	// We can just shove a whole validity buffer in there instead.
123	assert_eq!(undict.as_slice::<f32>(), &[1f32, 2f32, 2f32, 1f32, 1f32]);
124	}
125	}

vortex-data / vortex / 17073077835

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