16937708903

Committed 13 Aug 2025 12:48PM UTC coverage: 87.684% (+0.03%) from 87.658%

Build # 16937708903

Build Type

push

github

Committed by

web-flow

Commit Message

perf: Use sum compute function (#4207)

This patch finishes two todo items to replace with `SUM` compute
function in `Validity`'s `all_valid` and `all_invalid` functions.

Signed-off-by: Liang-Chi Hsieh <viirya@gmail.com>

Run Details

5 of 5 new or added lines in 1 file covered. (100.0%)

11 existing lines in 1 file now uncovered.

56464 of 64395 relevant lines covered (87.68%)

627631.53 hits per line

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

84.1

/vortex-array/src/validity.rs

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

//! Array validity and nullability behavior, used by arrays and compute functions.

use std::fmt::Debug;
use std::ops::{BitAnd, Not};

use arrow_buffer::{BooleanBuffer, NullBuffer};
use vortex_dtype::{DType, Nullability};
use vortex_error::{VortexExpect as _, VortexResult, vortex_bail, vortex_err, vortex_panic};
use vortex_mask::{AllOr, Mask, MaskValues};
use vortex_scalar::Scalar;

use crate::arrays::{BoolArray, ConstantArray};
use crate::compute::{fill_null, filter, sum, take};
use crate::patches::Patches;
use crate::{Array, ArrayRef, IntoArray, ToCanonical};

/// Validity information for an array
#[derive(Clone, Debug)]
pub enum Validity {
    /// Items *can't* be null
    NonNullable,
    /// All items are valid
    AllValid,
    /// All items are null
    AllInvalid,
    /// Specified items are null
    Array(ArrayRef),
}

impl Validity {
    /// The [`DType`] of the underlying validity array (if it exists).
    pub const DTYPE: DType = DType::Bool(Nullability::NonNullable);

    pub fn null_count(&self, length: usize) -> VortexResult<usize> {
        match self {
            Self::NonNullable | Self::AllValid => Ok(0),
            Self::AllInvalid => Ok(length),
            Self::Array(a) => {
                let validity_len = a.len();
                if validity_len != length {
                    vortex_bail!(
                        "Validity array length {} doesn't match array length {}",
                        validity_len,
                        length
                    )
                }
                let true_count = sum(a)?
                    .as_primitive()
                    .as_::<usize>()?
                    .ok_or_else(|| vortex_err!("Failed to compute true count"))?;
                Ok(length - true_count)
            }
        }
    }

    /// If Validity is [`Validity::Array`], returns the array, otherwise returns `None`.
    pub fn into_array(self) -> Option<ArrayRef> {
        match self {
            Self::Array(a) => Some(a),
            _ => None,
        }
    }

    /// If Validity is [`Validity::Array`], returns a reference to the array array, otherwise returns `None`.
    pub fn as_array(&self) -> Option<&ArrayRef> {
        match self {
            Self::Array(a) => Some(a),
            _ => None,
        }
    }

    pub fn nullability(&self) -> Nullability {
        match self {
            Self::NonNullable => Nullability::NonNullable,
            _ => Nullability::Nullable,
        }
    }

    /// The union nullability and validity.
    pub fn union_nullability(self, nullability: Nullability) -> Self {
        match nullability {
            Nullability::NonNullable => self,
            Nullability::Nullable => self.into_nullable(),
        }
    }

    pub fn all_valid(&self) -> VortexResult<bool> {
        Ok(match self {
            Validity::NonNullable | Validity::AllValid => true,
            Validity::AllInvalid => false,
            Validity::Array(array) => {
                // TODO(ngates): replace with SUM compute function
                array.to_bool()?.boolean_buffer().count_set_bits() == array.len()
            }
        })
    }

    pub fn all_invalid(&self) -> VortexResult<bool> {
        Ok(match self {
            Validity::NonNullable | Validity::AllValid => false,
            Validity::AllInvalid => true,
            Validity::Array(array) => {
                // TODO(ngates): replace with SUM compute function
                array.to_bool()?.boolean_buffer().count_set_bits() == 0
            }
        })
    }

    /// Returns whether the `index` item is valid.
    #[inline]
    pub fn is_valid(&self, index: usize) -> VortexResult<bool> {
        Ok(match self {
            Self::NonNullable | Self::AllValid => true,
            Self::AllInvalid => false,
            Self::Array(a) => {
                let scalar = a.scalar_at(index)?;
                scalar
                    .as_bool()
                    .value()
                    .vortex_expect("Validity must be non-nullable")
            }
        })
    }

    #[inline]
    pub fn is_null(&self, index: usize) -> VortexResult<bool> {
        Ok(!self.is_valid(index)?)
    }

    pub fn slice(&self, start: usize, stop: usize) -> VortexResult<Self> {
        match self {
            Self::Array(a) => Ok(Self::Array(a.slice(start, stop)?)),
            _ => Ok(self.clone()),
        }
    }

    pub fn take(&self, indices: &dyn Array) -> VortexResult<Self> {
        match self {
            Self::NonNullable => match indices.validity_mask()?.boolean_buffer() {
                AllOr::All => {
                    if indices.dtype().is_nullable() {
                        Ok(Self::AllValid)
                    } else {
                        Ok(Self::NonNullable)
                    }
                }
                AllOr::None => Ok(Self::AllInvalid),
                AllOr::Some(buf) => Ok(Validity::from(buf.clone())),
            },
            Self::AllValid => match indices.validity_mask()?.boolean_buffer() {
                AllOr::All => Ok(Self::AllValid),
                AllOr::None => Ok(Self::AllInvalid),
                AllOr::Some(buf) => Ok(Validity::from(buf.clone())),
            },
            Self::AllInvalid => Ok(Self::AllInvalid),
            Self::Array(is_valid) => {
                let maybe_is_valid = take(is_valid, indices)?;
                // Null indices invalidate that position.
                let is_valid = fill_null(&maybe_is_valid, &Scalar::from(false))?;
                Ok(Self::Array(is_valid))
            }
        }
    }

    /// Keep only the entries for which the mask is true.
    ///
    /// The result has length equal to the number of true values in mask.
    pub fn filter(&self, mask: &Mask) -> VortexResult<Self> {
        // NOTE(ngates): we take the mask as a reference to avoid the caller cloning unnecessarily
        //  if we happen to be NonNullable, AllValid, or AllInvalid.
        match self {
            v @ (Validity::NonNullable | Validity::AllValid | Validity::AllInvalid) => {
                Ok(v.clone())
            }
            Validity::Array(arr) => Ok(Validity::Array(filter(arr, mask)?)),
        }
    }

    /// Set to false any entries for which the mask is true.
    ///
    /// The result is always nullable. The result has the same length as self.
    pub fn mask(&self, mask: &Mask) -> VortexResult<Self> {
        match mask.boolean_buffer() {
            AllOr::All => Ok(Validity::AllInvalid),
            AllOr::None => Ok(self.clone()),
            AllOr::Some(make_invalid) => Ok(match self {
                Validity::NonNullable | Validity::AllValid => {
                    Validity::Array(BoolArray::from(make_invalid.not()).into_array())
                }
                Validity::AllInvalid => Validity::AllInvalid,
                Validity::Array(is_valid) => {
                    let is_valid = is_valid.to_bool()?;
                    let keep_valid = make_invalid.not();
                    Validity::from(is_valid.boolean_buffer().bitand(&keep_valid))
                }
            }),
        }
    }

    pub fn to_mask(&self, length: usize) -> VortexResult<Mask> {
        Ok(match self {
            Self::NonNullable | Self::AllValid => Mask::AllTrue(length),
            Self::AllInvalid => Mask::AllFalse(length),
            Self::Array(is_valid) => {
                assert_eq!(
                    is_valid.len(),
                    length,
                    "Validity::Array length must equal to_logical's argument: {}, {}.",
                    is_valid.len(),
                    length,
                );
                Mask::try_from(&is_valid.to_bool()?)?
            }
        })
    }

    /// Logically & two Validity values of the same length
    pub fn and(self, rhs: Validity) -> VortexResult<Validity> {
        let validity = match (self, rhs) {
            // Should be pretty clear
            (Validity::NonNullable, Validity::NonNullable) => Validity::NonNullable,
            // Any `AllInvalid` makes the output all invalid values
            (Validity::AllInvalid, _) | (_, Validity::AllInvalid) => Validity::AllInvalid,
            // All truthy values on one side, which makes no effect on an `Array` variant
            (Validity::Array(a), Validity::AllValid)
            | (Validity::Array(a), Validity::NonNullable)
            | (Validity::NonNullable, Validity::Array(a))
            | (Validity::AllValid, Validity::Array(a)) => Validity::Array(a),
            // Both sides are all valid
            (Validity::NonNullable, Validity::AllValid)
            | (Validity::AllValid, Validity::NonNullable)
            | (Validity::AllValid, Validity::AllValid) => Validity::AllValid,
            // Here we actually have to do some work
            (Validity::Array(lhs), Validity::Array(rhs)) => {
                let lhs = lhs.to_bool()?;
                let rhs = rhs.to_bool()?;

                let lhs = lhs.boolean_buffer();
                let rhs = rhs.boolean_buffer();

                Validity::from(lhs.bitand(rhs))
            }
        };

        Ok(validity)
    }

    pub fn patch(
        self,
        len: usize,
        indices_offset: usize,
        indices: &dyn Array,
        patches: &Validity,
    ) -> VortexResult<Self> {
        match (&self, patches) {
            (Validity::NonNullable, Validity::NonNullable) => return Ok(Validity::NonNullable),
            (Validity::NonNullable, _) => {
                vortex_bail!("Can't patch a non-nullable validity with nullable validity")
            }
            (_, Validity::NonNullable) => {
                vortex_bail!("Can't patch a nullable validity with non-nullable validity")
            }
            (Validity::AllValid, Validity::AllValid) => return Ok(Validity::AllValid),
            (Validity::AllInvalid, Validity::AllInvalid) => return Ok(Validity::AllInvalid),
            _ => {}
        };

        let own_nullability = if self == Validity::NonNullable {
            Nullability::NonNullable
        } else {
            Nullability::Nullable
        };

        let source = match self {
            Validity::NonNullable => BoolArray::from(BooleanBuffer::new_set(len)),
            Validity::AllValid => BoolArray::from(BooleanBuffer::new_set(len)),
            Validity::AllInvalid => BoolArray::from(BooleanBuffer::new_unset(len)),
            Validity::Array(a) => a.to_bool()?,
        };

        let patch_values = match patches {
            Validity::NonNullable => BoolArray::from(BooleanBuffer::new_set(indices.len())),
            Validity::AllValid => BoolArray::from(BooleanBuffer::new_set(indices.len())),
            Validity::AllInvalid => BoolArray::from(BooleanBuffer::new_unset(indices.len())),
            Validity::Array(a) => a.to_bool()?,
        };

        let patches = Patches::new(
            len,
            indices_offset,
            indices.to_array(),
            patch_values.into_array(),
        );

        Ok(Self::from_array(
            source.patch(&patches)?.into_array(),
            own_nullability,
        ))
    }

    /// Convert into a nullable variant
    pub fn into_nullable(self) -> Validity {
        match self {
            Self::NonNullable => Self::AllValid,
            _ => self,
        }
    }

    /// Convert into a non-nullable variant
    pub fn into_non_nullable(self) -> Option<Validity> {
        match self {
            Self::NonNullable => Some(Self::NonNullable),
            Self::AllValid => Some(Self::NonNullable),
            Self::AllInvalid => None,
            Self::Array(is_valid) => {
                is_valid
                    .statistics()
                    .compute_min::<bool>()
                    .vortex_expect("validity array must support min")
                    .then(|| {
                        // min true => all true
                        Self::NonNullable
                    })
            }
        }
    }

    /// Convert into a variant compatible with the given nullability, if possible.
    pub fn cast_nullability(self, nullability: Nullability) -> VortexResult<Validity> {
        match nullability {
            Nullability::NonNullable => self.into_non_nullable().ok_or_else(|| {
                vortex_err!("Cannot cast array with invalid values to non-nullable type.")
            }),
            Nullability::Nullable => Ok(self.into_nullable()),
        }
    }

    /// Create Validity by copying the given array's validity.
    pub fn copy_from_array(array: &dyn Array) -> VortexResult<Self> {
        Ok(Validity::from_mask(
            array.validity_mask()?,
            array.dtype().nullability(),
        ))
    }

    /// Create Validity from boolean array with given nullability of the array.
    ///
    /// Note: You want to pass the nullability of parent array and not the nullability of the validity array itself
    ///     as that is always nonnullable
    fn from_array(value: ArrayRef, nullability: Nullability) -> Self {
        if !matches!(value.dtype(), DType::Bool(Nullability::NonNullable)) {
            vortex_panic!("Expected a non-nullable boolean array")
        }
        match nullability {
            Nullability::NonNullable => Self::NonNullable,
            Nullability::Nullable => Self::Array(value),
        }
    }

    /// Returns the length of the validity array, if it exists.
    pub fn maybe_len(&self) -> Option<usize> {
        match self {
            Self::NonNullable | Self::AllValid | Self::AllInvalid => None,
            Self::Array(a) => Some(a.len()),
        }
    }

    pub fn uncompressed_size(&self) -> usize {
        if let Validity::Array(a) = self {
            a.len().div_ceil(8)
        } else {
            0
        }
    }

    pub fn is_array(&self) -> bool {
        matches!(self, Validity::Array(_))
    }
}

impl PartialEq for Validity {
    fn eq(&self, other: &Self) -> bool {
        match (self, other) {
            (Self::NonNullable, Self::NonNullable) => true,
            (Self::AllValid, Self::AllValid) => true,
            (Self::AllInvalid, Self::AllInvalid) => true,
            (Self::Array(a), Self::Array(b)) => {
                let a = a
                    .to_bool()
                    .vortex_expect("Failed to get Validity Array as BoolArray");
                let b = b
                    .to_bool()
                    .vortex_expect("Failed to get Validity Array as BoolArray");
                a.boolean_buffer() == b.boolean_buffer()
            }
            _ => false,
        }
    }
}

impl From<BooleanBuffer> for Validity {
    fn from(value: BooleanBuffer) -> Self {
        if value.count_set_bits() == value.len() {
            Self::AllValid
        } else if value.count_set_bits() == 0 {
            Self::AllInvalid
        } else {
            Self::Array(BoolArray::from(value).into_array())
        }
    }
}

impl From<NullBuffer> for Validity {
    fn from(value: NullBuffer) -> Self {
        value.into_inner().into()
    }
}

impl FromIterator<Mask> for Validity {
    fn from_iter<T: IntoIterator<Item = Mask>>(iter: T) -> Self {
        Validity::from_mask(iter.into_iter().collect(), Nullability::Nullable)
    }
}

impl FromIterator<bool> for Validity {
    fn from_iter<T: IntoIterator<Item = bool>>(iter: T) -> Self {
        Validity::from(BooleanBuffer::from_iter(iter))
    }
}

impl From<Nullability> for Validity {
    fn from(value: Nullability) -> Self {
        match value {
            Nullability::NonNullable => Validity::NonNullable,
            Nullability::Nullable => Validity::AllValid,
        }
    }
}

impl Validity {
    pub fn from_mask(mask: Mask, nullability: Nullability) -> Self {
        assert!(
            nullability == Nullability::Nullable || matches!(mask, Mask::AllTrue(_)),
            "NonNullable validity must be AllValid",
        );
        match mask {
            Mask::AllTrue(_) => match nullability {
                Nullability::NonNullable => Validity::NonNullable,
                Nullability::Nullable => Validity::AllValid,
            },
            Mask::AllFalse(_) => Validity::AllInvalid,
            Mask::Values(values) => Validity::Array(values.into_array()),
        }
    }
}

impl IntoArray for Mask {
    fn into_array(self) -> ArrayRef {
        match self {
            Self::AllTrue(len) => ConstantArray::new(true, len).into_array(),
            Self::AllFalse(len) => ConstantArray::new(false, len).into_array(),
            Self::Values(a) => a.into_array(),
        }
    }
}

impl IntoArray for &MaskValues {
    fn into_array(self) -> ArrayRef {
        BoolArray::new(self.boolean_buffer().clone(), Validity::NonNullable).into_array()
    }
}

#[cfg(test)]
mod tests {
    use rstest::rstest;
    use vortex_buffer::{Buffer, buffer};
    use vortex_dtype::Nullability;
    use vortex_mask::Mask;

    use crate::arrays::{BoolArray, PrimitiveArray};
    use crate::validity::Validity;
    use crate::{ArrayRef, IntoArray};

    #[rstest]
    #[case(Validity::AllValid, 5, &[2, 4], Validity::AllValid, Validity::AllValid)]
    #[case(Validity::AllValid, 5, &[2, 4], Validity::AllInvalid, Validity::Array(BoolArray::from_iter([true, true, false, true, false]).into_array())
    )]
    #[case(Validity::AllValid, 5, &[2, 4], Validity::Array(BoolArray::from_iter([true, false]).into_array()), Validity::Array(BoolArray::from_iter([true, true, true, true, false]).into_array())
    )]
    #[case(Validity::AllInvalid, 5, &[2, 4], Validity::AllValid, Validity::Array(BoolArray::from_iter([false, false, true, false, true]).into_array())
    )]
    #[case(Validity::AllInvalid, 5, &[2, 4], Validity::AllInvalid, Validity::AllInvalid)]
    #[case(Validity::AllInvalid, 5, &[2, 4], Validity::Array(BoolArray::from_iter([true, false]).into_array()), Validity::Array(BoolArray::from_iter([false, false, true, false, false]).into_array())
    )]
    #[case(Validity::Array(BoolArray::from_iter([false, true, false, true, false]).into_array()), 5, &[2, 4], Validity::AllValid, Validity::Array(BoolArray::from_iter([false, true, true, true, true]).into_array())
    )]
    #[case(Validity::Array(BoolArray::from_iter([false, true, false, true, false]).into_array()), 5, &[2, 4], Validity::AllInvalid, Validity::Array(BoolArray::from_iter([false, true, false, true, false]).into_array())
    )]
    #[case(Validity::Array(BoolArray::from_iter([false, true, false, true, false]).into_array()), 5, &[2, 4], Validity::Array(BoolArray::from_iter([true, false]).into_array()), Validity::Array(BoolArray::from_iter([false, true, true, true, false]).into_array())
    )]
    fn patch_validity(
        #[case] validity: Validity,
        #[case] len: usize,
        #[case] positions: &[u64],
        #[case] patches: Validity,
        #[case] expected: Validity,
    ) {
        let indices =
            PrimitiveArray::new(Buffer::copy_from(positions), Validity::NonNullable).into_array();
        assert_eq!(
            validity.patch(len, 0, &indices, &patches).unwrap(),
            expected
        );
    }

    #[test]
    #[should_panic]
    fn out_of_bounds_patch() {
        Validity::NonNullable
            .patch(2, 0, &buffer![4].into_array(), &Validity::AllInvalid)
            .unwrap();
    }

    #[test]
    #[should_panic]
    fn into_validity_nullable() {
        Validity::from_mask(Mask::AllFalse(10), Nullability::NonNullable);
    }

    #[test]
    #[should_panic]
    fn into_validity_nullable_array() {
        Validity::from_mask(Mask::from_iter(vec![true, false]), Nullability::NonNullable);
    }

    #[rstest]
    #[case(Validity::AllValid, PrimitiveArray::new(buffer![0, 1], Validity::from_iter(vec![true, false])).into_array(), Validity::from_iter(vec![true, false]))]
    #[case(Validity::AllValid, buffer![0, 1].into_array(), Validity::AllValid)]
    #[case(Validity::AllValid, PrimitiveArray::new(buffer![0, 1], Validity::AllInvalid).into_array(), Validity::AllInvalid)]
    #[case(Validity::NonNullable, PrimitiveArray::new(buffer![0, 1], Validity::from_iter(vec![true, false])).into_array(), Validity::from_iter(vec![true, false]))]
    #[case(Validity::NonNullable, buffer![0, 1].into_array(), Validity::NonNullable)]
    #[case(Validity::NonNullable, PrimitiveArray::new(buffer![0, 1], Validity::AllInvalid).into_array(), Validity::AllInvalid)]
    fn validity_take(
        #[case] validity: Validity,
        #[case] indices: ArrayRef,
        #[case] expected: Validity,
    ) {
        assert_eq!(validity.take(&indices).unwrap(), expected);
    }
}

1	// SPDX-License-Identifier: Apache-2.0
2	// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4	//! Array validity and nullability behavior, used by arrays and compute functions.
5
6	use std::fmt::Debug;
7	use std::ops::{BitAnd, Not};
8
9	use arrow_buffer::{BooleanBuffer, NullBuffer};
10	use vortex_dtype::{DType, Nullability};
11	use vortex_error::{VortexExpect as _, VortexResult, vortex_bail, vortex_err, vortex_panic};
12	use vortex_mask::{AllOr, Mask, MaskValues};
13	use vortex_scalar::Scalar;
14
15	use crate::arrays::{BoolArray, ConstantArray};
16	use crate::compute::{fill_null, filter, sum, take};
17	use crate::patches::Patches;
18	use crate::{Array, ArrayRef, IntoArray, ToCanonical};
19
20	/// Validity information for an array
21	#[derive(Clone, Debug)]
22	pub enum Validity {
23	/// Items can't be null
24	NonNullable,
25	/// All items are valid
26	AllValid,
27	/// All items are null
28	AllInvalid,
29	/// Specified items are null
30	Array(ArrayRef),
31	}
32
33	impl Validity {
34	/// The [`DType`] of the underlying validity array (if it exists).
35	pub const DTYPE: DType = DType::Bool(Nullability::NonNullable);
36
37	pub fn null_count(&self, length: usize) -> VortexResult<usize> {	×
38	match self {	×
39	Self::NonNullable \| Self::AllValid => Ok(0),	×
40	Self::AllInvalid => Ok(length),	×
41	Self::Array(a) => {	×
42	let validity_len = a.len();	×
43	if validity_len != length {	×
44	vortex_bail!(	×
45	"Validity array length {} doesn't match array length {}",	×
46	validity_len,
47	length
48	)
49	}	×
50	let true_count = sum(a)?	×
51	.as_primitive()	×
52	.as_::<usize>()?	×
53	.ok_or_else(\|\| vortex_err!("Failed to compute true count"))?;	×
54	Ok(length - true_count)	×
55	}
56	}
57	}	×
58
59	/// If Validity is [`Validity::Array`], returns the array, otherwise returns `None`.
60	pub fn into_array(self) -> Option<ArrayRef> {	×
61	match self {	×
62	Self::Array(a) => Some(a),	×
63	_ => None,	×
64	}
65	}	×
66
67	/// If Validity is [`Validity::Array`], returns a reference to the array array, otherwise returns `None`.
68	pub fn as_array(&self) -> Option<&ArrayRef> {	×
69	match self {	×
70	Self::Array(a) => Some(a),	×
71	_ => None,	×
72	}
73	}	×
74
75	pub fn nullability(&self) -> Nullability {	5,090,583✔
76	match self {	5,090,583✔
77	Self::NonNullable => Nullability::NonNullable,	4,104,172✔
78	_ => Nullability::Nullable,	986,411✔
79	}
80	}	5,090,583✔
81
82	/// The union nullability and validity.
83	pub fn union_nullability(self, nullability: Nullability) -> Self {	907✔
84	match nullability {	907✔
85	Nullability::NonNullable => self,	406✔
86	Nullability::Nullable => self.into_nullable(),	501✔
87	}
88	}	907✔
89
90	pub fn all_valid(&self) -> VortexResult<bool> {	19,408,721✔
91	Ok(match self {	19,408,721✔
92	Validity::NonNullable \| Validity::AllValid => true,	19,395,771✔
93	Validity::AllInvalid => false,	46✔
94	Validity::Array(array) => {	12,904✔
95	// TODO(ngates): replace with SUM compute function
96	array.to_bool()?.boolean_buffer().count_set_bits() == array.len()	12,904✔
97	}
98	})
99	}	19,408,673✔
100
101	pub fn all_invalid(&self) -> VortexResult<bool> {	494,023✔
102	Ok(match self {	494,071✔
103	Validity::NonNullable \| Validity::AllValid => false,	461,901✔
104	Validity::AllInvalid => true,	923✔
105	Validity::Array(array) => {	31,515✔
106	// TODO(ngates): replace with SUM compute function	158✔
107	array.to_bool()?.boolean_buffer().count_set_bits() == 0	31,357✔
108	}
109	})
110	}	494,023✔
111
112	/// Returns whether the `index` item is valid.
113	#[inline]
114	pub fn is_valid(&self, index: usize) -> VortexResult<bool> {	73,907,453✔
115	Ok(match self {	73,907,453✔
116	Self::NonNullable \| Self::AllValid => true,	67,523,533✔
117	Self::AllInvalid => false,	173,882✔
118	Self::Array(a) => {	6,210,196✔
119	let scalar = a.scalar_at(index)?;	6,210,196✔
120	scalar	6,210,206✔
121	.as_bool()	6,210,206✔
122	.value()	6,210,206✔
123	.vortex_expect("Validity must be non-nullable")	6,210,196✔
124	}
125	})
126	}	73,907,453✔
127
128	#[inline]
129	pub fn is_null(&self, index: usize) -> VortexResult<bool> {	147,615✔
130	Ok(!self.is_valid(index)?)	147,615✔
131	}	147,615✔
132		10✔
133	pub fn slice(&self, start: usize, stop: usize) -> VortexResult<Self> {	3,942,958✔
134	match self {	3,942,958✔
135	Self::Array(a) => Ok(Self::Array(a.slice(start, stop)?)),	9,885✔
136	_ => Ok(self.clone()),	3,933,073✔
137	}
138	}	3,942,958✔
139		160✔
140	pub fn take(&self, indices: &dyn Array) -> VortexResult<Self> {	76,085✔
141	match self {	75,925✔
142	Self::NonNullable => match indices.validity_mask()?.boolean_buffer() {	59,747✔
143	AllOr::All => {
144	if indices.dtype().is_nullable() {	52,805✔
145	Ok(Self::AllValid)	277✔
146	} else {
147	Ok(Self::NonNullable)	52,368✔
148	}
149	}
150	AllOr::None => Ok(Self::AllInvalid),	1✔
151	AllOr::Some(buf) => Ok(Validity::from(buf.clone())),	6,941✔
152	},
153	Self::AllValid => match indices.validity_mask()?.boolean_buffer() {	6,698✔
154	AllOr::All => Ok(Self::AllValid),	3,262✔
155	AllOr::None => Ok(Self::AllInvalid),	1✔
156	AllOr::Some(buf) => Ok(Validity::from(buf.clone())),	3,435✔
157	},
158	Self::AllInvalid => Ok(Self::AllInvalid),	123✔
159	Self::Array(is_valid) => {	9,517✔
160	let maybe_is_valid = take(is_valid, indices)?;	9,517✔
161	// Null indices invalidate that position.
162	let is_valid = fill_null(&maybe_is_valid, &Scalar::from(false))?;	9,517✔
163	Ok(Self::Array(is_valid))	9,517✔
164	}
165	}
166	}	75,925✔
167
168	/// Keep only the entries for which the mask is true.
169	///
170	/// The result has length equal to the number of true values in mask.
171	pub fn filter(&self, mask: &Mask) -> VortexResult<Self> {	58,746✔
172	// NOTE(ngates): we take the mask as a reference to avoid the caller cloning unnecessarily
173	// if we happen to be NonNullable, AllValid, or AllInvalid.
174	match self {	58,746✔
175	v @ (Validity::NonNullable \| Validity::AllValid \| Validity::AllInvalid) => {	46,899✔
176	Ok(v.clone())	46,899✔
177	}
178	Validity::Array(arr) => Ok(Validity::Array(filter(arr, mask)?)),	11,847✔
179	}
180	}	58,746✔
181
182	/// Set to false any entries for which the mask is true.
183	///
184	/// The result is always nullable. The result has the same length as self.
185	pub fn mask(&self, mask: &Mask) -> VortexResult<Self> {	11,585✔
186	match mask.boolean_buffer() {	11,585✔
187	AllOr::All => Ok(Validity::AllInvalid),
188	AllOr::None => Ok(self.clone()),
189	AllOr::Some(make_invalid) => Ok(match self {	11,585✔
190	Validity::NonNullable \| Validity::AllValid => {
191	Validity::Array(BoolArray::from(make_invalid.not()).into_array())	4,748✔
192	}
193	Validity::AllInvalid => Validity::AllInvalid,	55✔
194	Validity::Array(is_valid) => {	6,782✔
195	let is_valid = is_valid.to_bool()?;	6,782✔
196	let keep_valid = make_invalid.not();	6,782✔
197	Validity::from(is_valid.boolean_buffer().bitand(&keep_valid))	6,782✔
198	}
199	}),
200	}
201	}	11,585✔
202
203	pub fn to_mask(&self, length: usize) -> VortexResult<Mask> {	1,443,984✔
204	Ok(match self {	1,443,984✔
205	Self::NonNullable \| Self::AllValid => Mask::AllTrue(length),	1,372,795✔
206	Self::AllInvalid => Mask::AllFalse(length),	499✔
207	Self::Array(is_valid) => {	70,690✔
208	assert_eq!(	70,690✔
209	is_valid.len(),	71,204✔
210	length,	514✔
211	"Validity::Array length must equal to_logical's argument: {}, {}.",	514✔
UNCOV 212	is_valid.len(),	×
213	length,
214	);
215	Mask::try_from(&is_valid.to_bool()?)?	70,690✔
216	}
217	})
218	}	1,443,984✔
219
220	/// Logically & two Validity values of the same length
221	pub fn and(self, rhs: Validity) -> VortexResult<Validity> {	55✔
222	let validity = match (self, rhs) {	55✔
223	// Should be pretty clear
224	(Validity::NonNullable, Validity::NonNullable) => Validity::NonNullable,	569✔
225	// Any `AllInvalid` makes the output all invalid values
226	(Validity::AllInvalid, _) \| (_, Validity::AllInvalid) => Validity::AllInvalid,
227	// All truthy values on one side, which makes no effect on an `Array` variant
UNCOV 228	(Validity::Array(a), Validity::AllValid)	×
229	\| (Validity::Array(a), Validity::NonNullable)
UNCOV 230	\| (Validity::NonNullable, Validity::Array(a))	×
231	\| (Validity::AllValid, Validity::Array(a)) => Validity::Array(a),
232	// Both sides are all valid
233	(Validity::NonNullable, Validity::AllValid)
234	\| (Validity::AllValid, Validity::NonNullable)
235	\| (Validity::AllValid, Validity::AllValid) => Validity::AllValid,	×
236	// Here we actually have to do some work
237	(Validity::Array(lhs), Validity::Array(rhs)) => {	×
238	let lhs = lhs.to_bool()?;
239	let rhs = rhs.to_bool()?;
240
241	let lhs = lhs.boolean_buffer();	×
242	let rhs = rhs.boolean_buffer();
243
244	Validity::from(lhs.bitand(rhs))	×
245	}
246	};
247
248	Ok(validity)	55✔
249	}	55✔
250
251	pub fn patch(	12,051✔
252	self,	12,051✔
253	len: usize,	12,051✔
254	indices_offset: usize,	12,051✔
255	indices: &dyn Array,	12,051✔
256	patches: &Validity,	12,051✔
257	) -> VortexResult<Self> {	12,051✔
258	match (&self, patches) {	12,051✔
259	(Validity::NonNullable, Validity::NonNullable) => return Ok(Validity::NonNullable),	9,264✔
260	(Validity::NonNullable, _) => {
261	vortex_bail!("Can't patch a non-nullable validity with nullable validity")	1✔
262	}
263	(_, Validity::NonNullable) => {
UNCOV 264	vortex_bail!("Can't patch a nullable validity with non-nullable validity")	×
265	}
266	(Validity::AllValid, Validity::AllValid) => return Ok(Validity::AllValid),	1✔
267	(Validity::AllInvalid, Validity::AllInvalid) => return Ok(Validity::AllInvalid),	1✔
268	_ => {}	2,784✔
269	};
270
271	let own_nullability = if self == Validity::NonNullable {	2,784✔
UNCOV 272	Nullability::NonNullable	×
273	} else {
274	Nullability::Nullable	2,784✔
275	};
276
277	let source = match self {	2,784✔
278	Validity::NonNullable => BoolArray::from(BooleanBuffer::new_set(len)),	×
279	Validity::AllValid => BoolArray::from(BooleanBuffer::new_set(len)),	392✔
280	Validity::AllInvalid => BoolArray::from(BooleanBuffer::new_unset(len)),	2,311✔
281	Validity::Array(a) => a.to_bool()?,	81✔
282	};
283
284	let patch_values = match patches {	2,784✔
UNCOV 285	Validity::NonNullable => BoolArray::from(BooleanBuffer::new_set(indices.len())),	×
286	Validity::AllValid => BoolArray::from(BooleanBuffer::new_set(indices.len())),	1,804✔
287	Validity::AllInvalid => BoolArray::from(BooleanBuffer::new_unset(indices.len())),	2✔
288	Validity::Array(a) => a.to_bool()?,	978✔
289	};
290
291	let patches = Patches::new(	2,784✔
292	len,	2,784✔
293	indices_offset,	2,784✔
294	indices.to_array(),	2,784✔
295	patch_values.into_array(),	2,784✔
296	);
297
298	Ok(Self::from_array(	2,784✔
299	source.patch(&patches)?.into_array(),	2,784✔
300	own_nullability,	2,784✔
301	))
302	}	12,051✔
303
304	/// Convert into a nullable variant
305	pub fn into_nullable(self) -> Validity {	19,228✔
306	match self {	19,228✔
307	Self::NonNullable => Self::AllValid,	18,991✔
308	_ => self,	237✔
309	}
310	}	19,228✔
311
312	/// Convert into a non-nullable variant
313	pub fn into_non_nullable(self) -> Option<Validity> {	2,555✔
314	match self {	2,555✔
315	Self::NonNullable => Some(Self::NonNullable),	508✔
316	Self::AllValid => Some(Self::NonNullable),	1,282✔
317	Self::AllInvalid => None,
318	Self::Array(is_valid) => {	765✔
319	is_valid	765✔
320	.statistics()	765✔
321	.compute_min::<bool>()	765✔
322	.vortex_expect("validity array must support min")	765✔
323	.then(\|\| {	765✔
324	// min true => all true
325	Self::NonNullable	238✔
326	})	238✔
327	}
328	}
329	}	2,555✔
330
331	/// Convert into a variant compatible with the given nullability, if possible.
332	pub fn cast_nullability(self, nullability: Nullability) -> VortexResult<Validity> {	9,010✔
333	match nullability {	9,010✔
334	Nullability::NonNullable => self.into_non_nullable().ok_or_else(\|\| {	2,555✔
335	vortex_err!("Cannot cast array with invalid values to non-nullable type.")	527✔
336	}),	527✔
337	Nullability::Nullable => Ok(self.into_nullable()),	6,455✔
338	}
339	}	9,010✔
340
341	/// Create Validity by copying the given array's validity.
342	pub fn copy_from_array(array: &dyn Array) -> VortexResult<Self> {	23,477✔
343	Ok(Validity::from_mask(	23,477✔
344	array.validity_mask()?,	23,477✔
345	array.dtype().nullability(),	23,477✔
346	))
347	}	23,477✔
348
349	/// Create Validity from boolean array with given nullability of the array.
350	///
351	/// Note: You want to pass the nullability of parent array and not the nullability of the validity array itself
352	/// as that is always nonnullable
353	fn from_array(value: ArrayRef, nullability: Nullability) -> Self {	2,784✔
354	if !matches!(value.dtype(), DType::Bool(Nullability::NonNullable)) {	2,784✔
355	vortex_panic!("Expected a non-nullable boolean array")
356	}	2,784✔
357	match nullability {	2,784✔
358	Nullability::NonNullable => Self::NonNullable,
359	Nullability::Nullable => Self::Array(value),	2,784✔
360	}
361	}	2,784✔
362
363	/// Returns the length of the validity array, if it exists.
364	pub fn maybe_len(&self) -> Option<usize> {	6,203,679✔
365	match self {	6,203,679✔
366	Self::NonNullable \| Self::AllValid \| Self::AllInvalid => None,	6,075,421✔
367	Self::Array(a) => Some(a.len()),	128,258✔
368	}
369	}	6,203,679✔
370		730✔
371	pub fn uncompressed_size(&self) -> usize {	730✔
372	if let Validity::Array(a) = self {	730✔
UNCOV 373	a.len().div_ceil(8)	×
374	} else {
375	0	730✔
376	}
377	}	×
378
379	pub fn is_array(&self) -> bool {	×
380	matches!(self, Validity::Array(_))
381	}	×
382	}
383
384	impl PartialEq for Validity {
385	fn eq(&self, other: &Self) -> bool {	63,757✔
386	match (self, other) {	63,757✔
387	(Self::NonNullable, Self::NonNullable) => true,	36,442✔
388	(Self::AllValid, Self::AllValid) => true,	126✔
389	(Self::AllInvalid, Self::AllInvalid) => true,	123✔
390	(Self::Array(a), Self::Array(b)) => {	526✔
391	let a = a	526✔
392	.to_bool()	526✔
393	.vortex_expect("Failed to get Validity Array as BoolArray");	526✔
394	let b = b	526✔
395	.to_bool()	526✔
396	.vortex_expect("Failed to get Validity Array as BoolArray");	526✔
397	a.boolean_buffer() == b.boolean_buffer()	526✔
398	}
399	_ => false,	26,540✔
400	}
401	}	63,757✔
402	}
403
404	impl From<BooleanBuffer> for Validity {
405	fn from(value: BooleanBuffer) -> Self {	68,346✔
406	if value.count_set_bits() == value.len() {	68,346✔
407	Self::AllValid	374✔
408	} else if value.count_set_bits() == 0 {	67,972✔
409	Self::AllInvalid	2,235✔
410	} else {
411	Self::Array(BoolArray::from(value).into_array())	65,737✔
412	}
413	}	68,346✔
414	}
415
416	impl From<NullBuffer> for Validity {
417	fn from(value: NullBuffer) -> Self {	14,942✔
418	value.into_inner().into()	14,942✔
419	}	14,942✔
420	}
421
422	impl FromIterator<Mask> for Validity {
UNCOV 423	fn from_iter<T: IntoIterator<Item = Mask>>(iter: T) -> Self {	×
UNCOV 424	Validity::from_mask(iter.into_iter().collect(), Nullability::Nullable)	×
UNCOV 425	}	×
426	}
427
428	impl FromIterator<bool> for Validity {
429	fn from_iter<T: IntoIterator<Item = bool>>(iter: T) -> Self {	3,391✔
430	Validity::from(BooleanBuffer::from_iter(iter))	3,391✔
431	}	3,391✔
432	}
433
434	impl From<Nullability> for Validity {
435	fn from(value: Nullability) -> Self {	48,705✔
436	match value {	48,705✔
437	Nullability::NonNullable => Validity::NonNullable,	43,280✔
438	Nullability::Nullable => Validity::AllValid,	5,425✔
439	}
440	}	48,705✔
441	}	6✔
442		6✔
443	impl Validity {	6✔
444	pub fn from_mask(mask: Mask, nullability: Nullability) -> Self {	24,243✔
445	assert!(	24,243✔
446	nullability == Nullability::Nullable \|\| matches!(mask, Mask::AllTrue(_)),	24,249✔
447	"NonNullable validity must be AllValid",	2✔
448	);
449	match mask {	24,241✔
450	Mask::AllTrue(_) => match nullability {	20,018✔
451	Nullability::NonNullable => Validity::NonNullable,	19,740✔
452	Nullability::Nullable => Validity::AllValid,	278✔
453	},
454	Mask::AllFalse(_) => Validity::AllInvalid,	78✔
455	Mask::Values(values) => Validity::Array(values.into_array()),	4,145✔
456	}
457	}	24,241✔
458	}
459
460	impl IntoArray for Mask {
461	fn into_array(self) -> ArrayRef {	40✔
462	match self {	40✔
UNCOV 463	Self::AllTrue(len) => ConstantArray::new(true, len).into_array(),	×
464	Self::AllFalse(len) => ConstantArray::new(false, len).into_array(),
465	Self::Values(a) => a.into_array(),	40✔
466	}
467	}	40✔
468	}
469
470	impl IntoArray for &MaskValues {
471	fn into_array(self) -> ArrayRef {	4,224✔
472	BoolArray::new(self.boolean_buffer().clone(), Validity::NonNullable).into_array()	4,224✔
473	}	4,224✔
474	}
475
476	#[cfg(test)]
477	mod tests {
478	use rstest::rstest;
479	use vortex_buffer::{Buffer, buffer};
480	use vortex_dtype::Nullability;
481	use vortex_mask::Mask;
482
483	use crate::arrays::{BoolArray, PrimitiveArray};
484	use crate::validity::Validity;
485	use crate::{ArrayRef, IntoArray};
486
487	#[rstest]
488	#[case(Validity::AllValid, 5, &[2, 4], Validity::AllValid, Validity::AllValid)]
489	#[case(Validity::AllValid, 5, &[2, 4], Validity::AllInvalid, Validity::Array(BoolArray::from_iter([true, true, false, true, false]).into_array())
490	)]
491	#[case(Validity::AllValid, 5, &[2, 4], Validity::Array(BoolArray::from_iter([true, false]).into_array()), Validity::Array(BoolArray::from_iter([true, true, true, true, false]).into_array())
492	)]
493	#[case(Validity::AllInvalid, 5, &[2, 4], Validity::AllValid, Validity::Array(BoolArray::from_iter([false, false, true, false, true]).into_array())
494	)]
495	#[case(Validity::AllInvalid, 5, &[2, 4], Validity::AllInvalid, Validity::AllInvalid)]
496	#[case(Validity::AllInvalid, 5, &[2, 4], Validity::Array(BoolArray::from_iter([true, false]).into_array()), Validity::Array(BoolArray::from_iter([false, false, true, false, false]).into_array())
497	)]
498	#[case(Validity::Array(BoolArray::from_iter([false, true, false, true, false]).into_array()), 5, &[2, 4], Validity::AllValid, Validity::Array(BoolArray::from_iter([false, true, true, true, true]).into_array())
499	)]
500	#[case(Validity::Array(BoolArray::from_iter([false, true, false, true, false]).into_array()), 5, &[2, 4], Validity::AllInvalid, Validity::Array(BoolArray::from_iter([false, true, false, true, false]).into_array())
501	)]
502	#[case(Validity::Array(BoolArray::from_iter([false, true, false, true, false]).into_array()), 5, &[2, 4], Validity::Array(BoolArray::from_iter([true, false]).into_array()), Validity::Array(BoolArray::from_iter([false, true, true, true, false]).into_array())
503	)]
504	fn patch_validity(
505	#[case] validity: Validity,
506	#[case] len: usize,
507	#[case] positions: &[u64],
508	#[case] patches: Validity,
509	#[case] expected: Validity,
510	) {
511	let indices =
512	PrimitiveArray::new(Buffer::copy_from(positions), Validity::NonNullable).into_array();
513	assert_eq!(
514	validity.patch(len, 0, &indices, &patches).unwrap(),
515	expected
516	);
517	}
518
519	#[test]
520	#[should_panic]
521	fn out_of_bounds_patch() {	1✔
522	Validity::NonNullable	1✔
523	.patch(2, 0, &buffer![4].into_array(), &Validity::AllInvalid)	1✔
524	.unwrap();	1✔
525	}	1✔
526
527	#[test]
528	#[should_panic]
529	fn into_validity_nullable() {	1✔
530	Validity::from_mask(Mask::AllFalse(10), Nullability::NonNullable);	1✔
531	}	1✔
532
533	#[test]
534	#[should_panic]
535	fn into_validity_nullable_array() {	1✔
536	Validity::from_mask(Mask::from_iter(vec![true, false]), Nullability::NonNullable);	1✔
537	}	1✔
538
539	#[rstest]
540	#[case(Validity::AllValid, PrimitiveArray::new(buffer![0, 1], Validity::from_iter(vec![true, false])).into_array(), Validity::from_iter(vec![true, false]))]
541	#[case(Validity::AllValid, buffer![0, 1].into_array(), Validity::AllValid)]
542	#[case(Validity::AllValid, PrimitiveArray::new(buffer![0, 1], Validity::AllInvalid).into_array(), Validity::AllInvalid)]
543	#[case(Validity::NonNullable, PrimitiveArray::new(buffer![0, 1], Validity::from_iter(vec![true, false])).into_array(), Validity::from_iter(vec![true, false]))]
544	#[case(Validity::NonNullable, buffer![0, 1].into_array(), Validity::NonNullable)]
545	#[case(Validity::NonNullable, PrimitiveArray::new(buffer![0, 1], Validity::AllInvalid).into_array(), Validity::AllInvalid)]
546	fn validity_take(
547	#[case] validity: Validity,
548	#[case] indices: ArrayRef,
549	#[case] expected: Validity,
550	) {
551	assert_eq!(validity.take(&indices).unwrap(), expected);
552	}
553	}

vortex-data / vortex / 16937708903

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