17075133033

Committed 19 Aug 2025 04:01PM UTC coverage: 87.949% (+0.09%) from 87.856%

Build # 17075133033

Build Type

push

github

Committed by

web-flow

Commit Message

feat: ArrayOperations infallible, eager validation + new_unchecked (#4177)

ArrayOperations currently return VortexResult<>, but they really should
just be infallible. A failed array op is generally indicative of
programmer or encoding error. There's really nothing interesting we can
do to handle an out-of-bounds slice() or scalar_at.

There's a lot that falls out of this, like fixing a bunch of tests,
tweaking our scalar value casting to return Option instead of Result,
etc.

---------

Signed-off-by: Andrew Duffy <andrew@a10y.dev>

Run Details

1744 of 1985 new or added lines in 195 files covered. (87.86%)

36 existing lines in 27 files now uncovered.

56745 of 64520 relevant lines covered (87.95%)

624082.56 hits per line

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

79.19

/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) => sum(array)
                .map(|v| {
                    v.as_primitive()
                        .typed_value::<u64>()
                        .map(|count| count == array.len() as u64)
                })?
                .ok_or_else(|| vortex_err!("Failed to compute sum for validity array"))?,
        })
    }

    pub fn all_invalid(&self) -> VortexResult<bool> {
        Ok(match self {
            Validity::NonNullable | Validity::AllValid => false,
            Validity::AllInvalid => true,
            Validity::Array(array) => sum(array)
                .map(|v| {
                    v.as_primitive()
                        .typed_value::<u64>()
                        .map(|count| count == 0u64)
                })?
                .ok_or_else(|| vortex_err!("Failed to compute sum for validity array"))?,
        })
    }

    /// 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) -> Self {
        match self {
            Self::Array(a) => Self::Array(a.slice(start, stop)),
            _ => 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()	×
NEW 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 {	5,042,627✔
76	match self {	5,042,627✔
77	Self::NonNullable => Nullability::NonNullable,	4,061,710✔
78	_ => Nullability::Nullable,	980,917✔
79	}
80	}	5,042,627✔
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> {	19,425,584✔
91	Ok(match self {	19,425,584✔
92	Validity::NonNullable \| Validity::AllValid => true,	19,413,970✔
93	Validity::AllInvalid => false,	46✔
94	Validity::Array(array) => sum(array)	11,568✔
95	.map(\|v\| {	11,568✔
96	v.as_primitive()	11,568✔
97	.typed_value::<u64>()	11,568✔
98	.map(\|count\| count == array.len() as u64)	11,568✔
99	})?	11,568✔
100	.ok_or_else(\|\| vortex_err!("Failed to compute sum for validity array"))?,	11,568✔
101	})
102	}	19,425,584✔
103
104	pub fn all_invalid(&self) -> VortexResult<bool> {	488,532✔
105	Ok(match self {	488,532✔
106	Validity::NonNullable \| Validity::AllValid => false,	458,098✔
107	Validity::AllInvalid => true,	764✔
108	Validity::Array(array) => sum(array)	29,670✔
109	.map(\|v\| {	29,670✔
110	v.as_primitive()	29,670✔
111	.typed_value::<u64>()	29,670✔
112	.map(\|count\| count == 0u64)	29,670✔
113	})?	29,670✔
114	.ok_or_else(\|\| vortex_err!("Failed to compute sum for validity array"))?,	29,670✔
115	})
116	}	488,532✔
117
118	/// Returns whether the `index` item is valid.
119	#[inline]
120	pub fn is_valid(&self, index: usize) -> VortexResult<bool> {	86,274,870✔
121	Ok(match self {	86,274,870✔
122	Self::NonNullable \| Self::AllValid => true,	79,946,783✔
123	Self::AllInvalid => false,	173,888✔
124	Self::Array(a) => {	6,154,199✔
125	let scalar = a.scalar_at(index);	6,154,199✔
126	scalar	6,154,199✔
127	.as_bool()	6,154,199✔
128	.value()	6,154,199✔
129	.vortex_expect("Validity must be non-nullable")	6,154,199✔
130	}
131	})
132	}	86,274,870✔
133
134	#[inline]
135	pub fn is_null(&self, index: usize) -> VortexResult<bool> {	12,224,500✔
136	Ok(!self.is_valid(index)?)	12,224,500✔
137	}	12,224,500✔
138
139	pub fn slice(&self, start: usize, stop: usize) -> Self {	3,937,998✔
140	match self {	3,937,998✔
141	Self::Array(a) => Self::Array(a.slice(start, stop)),	9,885✔
142	_ => self.clone(),	3,928,113✔
143	}
144	}	3,937,998✔
145
146	pub fn take(&self, indices: &dyn Array) -> VortexResult<Self> {	75,925✔
147	match self {	75,925✔
148	Self::NonNullable => match indices.validity_mask()?.boolean_buffer() {	59,587✔
149	AllOr::All => {
150	if indices.dtype().is_nullable() {	52,645✔
151	Ok(Self::AllValid)	277✔
152	} else {
153	Ok(Self::NonNullable)	52,368✔
154	}
155	}
156	AllOr::None => Ok(Self::AllInvalid),	1✔
157	AllOr::Some(buf) => Ok(Validity::from(buf.clone())),	6,941✔
158	},
159	Self::AllValid => match indices.validity_mask()?.boolean_buffer() {	6,698✔
160	AllOr::All => Ok(Self::AllValid),	3,262✔
161	AllOr::None => Ok(Self::AllInvalid),	1✔
162	AllOr::Some(buf) => Ok(Validity::from(buf.clone())),	3,435✔
163	},
164	Self::AllInvalid => Ok(Self::AllInvalid),	123✔
165	Self::Array(is_valid) => {	9,517✔
166	let maybe_is_valid = take(is_valid, indices)?;	9,517✔
167	// Null indices invalidate that position.
168	let is_valid = fill_null(&maybe_is_valid, &Scalar::from(false))?;	9,517✔
169	Ok(Self::Array(is_valid))	9,517✔
170	}
171	}
172	}	75,925✔
173
174	/// Keep only the entries for which the mask is true.
175	///
176	/// The result has length equal to the number of true values in mask.
177	pub fn filter(&self, mask: &Mask) -> VortexResult<Self> {	58,707✔
178	// NOTE(ngates): we take the mask as a reference to avoid the caller cloning unnecessarily
179	// if we happen to be NonNullable, AllValid, or AllInvalid.
180	match self {	58,707✔
181	v @ (Validity::NonNullable \| Validity::AllValid \| Validity::AllInvalid) => {	46,860✔
182	Ok(v.clone())	46,860✔
183	}
184	Validity::Array(arr) => Ok(Validity::Array(filter(arr, mask)?)),	11,847✔
185	}
186	}	58,707✔
187
188	/// Set to false any entries for which the mask is true.
189	///
190	/// The result is always nullable. The result has the same length as self.
191	pub fn mask(&self, mask: &Mask) -> VortexResult<Self> {	11,585✔
192	match mask.boolean_buffer() {	11,585✔
193	AllOr::All => Ok(Validity::AllInvalid),	×
194	AllOr::None => Ok(self.clone()),	×
195	AllOr::Some(make_invalid) => Ok(match self {	11,585✔
196	Validity::NonNullable \| Validity::AllValid => {
197	Validity::Array(BoolArray::from(make_invalid.not()).into_array())	4,748✔
198	}
199	Validity::AllInvalid => Validity::AllInvalid,	55✔
200	Validity::Array(is_valid) => {	6,782✔
201	let is_valid = is_valid.to_bool()?;	6,782✔
202	let keep_valid = make_invalid.not();	6,782✔
203	Validity::from(is_valid.boolean_buffer().bitand(&keep_valid))	6,782✔
204	}
205	}),
206	}
207	}	11,585✔
208
209	pub fn to_mask(&self, length: usize) -> VortexResult<Mask> {	1,469,879✔
210	Ok(match self {	1,469,879✔
211	Self::NonNullable \| Self::AllValid => Mask::AllTrue(length),	1,401,927✔
212	Self::AllInvalid => Mask::AllFalse(length),	421✔
213	Self::Array(is_valid) => {	67,531✔
214	assert_eq!(	67,531✔
215	is_valid.len(),	67,531✔
216	length,
217	"Validity::Array length must equal to_logical's argument: {}, {}.",	×
218	is_valid.len(),	×
219	length,
220	);
221	Mask::try_from(&is_valid.to_bool()?)?	67,531✔
222	}
223	})
224	}	1,469,879✔
225
226	/// Logically & two Validity values of the same length
227	pub fn and(self, rhs: Validity) -> VortexResult<Validity> {	55✔
228	let validity = match (self, rhs) {	55✔
229	// Should be pretty clear
230	(Validity::NonNullable, Validity::NonNullable) => Validity::NonNullable,	55✔
231	// Any `AllInvalid` makes the output all invalid values
232	(Validity::AllInvalid, _) \| (_, Validity::AllInvalid) => Validity::AllInvalid,	×
233	// All truthy values on one side, which makes no effect on an `Array` variant
234	(Validity::Array(a), Validity::AllValid)	×
235	\| (Validity::Array(a), Validity::NonNullable)	×
236	\| (Validity::NonNullable, Validity::Array(a))	×
237	\| (Validity::AllValid, Validity::Array(a)) => Validity::Array(a),	×
238	// Both sides are all valid
239	(Validity::NonNullable, Validity::AllValid)
240	\| (Validity::AllValid, Validity::NonNullable)
241	\| (Validity::AllValid, Validity::AllValid) => Validity::AllValid,	×
242	// Here we actually have to do some work
243	(Validity::Array(lhs), Validity::Array(rhs)) => {	×
244	let lhs = lhs.to_bool()?;	×
245	let rhs = rhs.to_bool()?;	×
246
247	let lhs = lhs.boolean_buffer();	×
248	let rhs = rhs.boolean_buffer();	×
249
250	Validity::from(lhs.bitand(rhs))	×
251	}
252	};
253
254	Ok(validity)	55✔
255	}	55✔
256
257	pub fn patch(	12,009✔
258	self,	12,009✔
259	len: usize,	12,009✔
260	indices_offset: usize,	12,009✔
261	indices: &dyn Array,	12,009✔
262	patches: &Validity,	12,009✔
263	) -> VortexResult<Self> {	12,009✔
264	match (&self, patches) {	12,009✔
265	(Validity::NonNullable, Validity::NonNullable) => return Ok(Validity::NonNullable),	9,222✔
266	(Validity::NonNullable, _) => {
267	vortex_bail!("Can't patch a non-nullable validity with nullable validity")	1✔
268	}
269	(_, Validity::NonNullable) => {
270	vortex_bail!("Can't patch a nullable validity with non-nullable validity")	×
271	}
272	(Validity::AllValid, Validity::AllValid) => return Ok(Validity::AllValid),	1✔
273	(Validity::AllInvalid, Validity::AllInvalid) => return Ok(Validity::AllInvalid),	1✔
274	_ => {}	2,784✔
275	};
276
277	let own_nullability = if self == Validity::NonNullable {	2,784✔
278	Nullability::NonNullable	×
279	} else {
280	Nullability::Nullable	2,784✔
281	};
282
283	let source = match self {	2,784✔
284	Validity::NonNullable => BoolArray::from(BooleanBuffer::new_set(len)),	×
285	Validity::AllValid => BoolArray::from(BooleanBuffer::new_set(len)),	392✔
286	Validity::AllInvalid => BoolArray::from(BooleanBuffer::new_unset(len)),	2,311✔
287	Validity::Array(a) => a.to_bool()?,	81✔
288	};
289
290	let patch_values = match patches {	2,784✔
291	Validity::NonNullable => BoolArray::from(BooleanBuffer::new_set(indices.len())),	×
292	Validity::AllValid => BoolArray::from(BooleanBuffer::new_set(indices.len())),	1,804✔
293	Validity::AllInvalid => BoolArray::from(BooleanBuffer::new_unset(indices.len())),	2✔
294	Validity::Array(a) => a.to_bool()?,	978✔
295	};
296
297	let patches = Patches::new(	2,784✔
298	len,	2,784✔
299	indices_offset,	2,784✔
300	indices.to_array(),	2,784✔
301	patch_values.into_array(),	2,784✔
302	);
303
304	Ok(Self::from_array(	2,784✔
305	source.patch(&patches)?.into_array(),	2,784✔
306	own_nullability,	2,784✔
307	))
308	}	12,009✔
309
310	/// Convert into a nullable variant
311	pub fn into_nullable(self) -> Validity {	19,267✔
312	match self {	19,267✔
313	Self::NonNullable => Self::AllValid,	19,030✔
314	_ => self,	237✔
315	}
316	}	19,267✔
317
318	/// Convert into a non-nullable variant
319	pub fn into_non_nullable(self) -> Option<Validity> {	2,555✔
320	match self {	2,555✔
321	Self::NonNullable => Some(Self::NonNullable),	508✔
322	Self::AllValid => Some(Self::NonNullable),	1,282✔
323	Self::AllInvalid => None,	×
324	Self::Array(is_valid) => {	765✔
325	is_valid	765✔
326	.statistics()	765✔
327	.compute_min::<bool>()	765✔
328	.vortex_expect("validity array must support min")	765✔
329	.then(\|\| {	765✔
330	// min true => all true
331	Self::NonNullable	238✔
332	})	238✔
333	}
334	}
335	}	2,555✔
336
337	/// Convert into a variant compatible with the given nullability, if possible.
338	pub fn cast_nullability(self, nullability: Nullability) -> VortexResult<Validity> {	9,013✔
339	match nullability {	9,013✔
340	Nullability::NonNullable => self.into_non_nullable().ok_or_else(\|\| {	2,555✔
341	vortex_err!("Cannot cast array with invalid values to non-nullable type.")	527✔
342	}),	527✔
343	Nullability::Nullable => Ok(self.into_nullable()),	6,458✔
344	}
345	}	9,013✔
346
347	/// Create Validity by copying the given array's validity.
348	pub fn copy_from_array(array: &dyn Array) -> VortexResult<Self> {	19,775✔
349	Ok(Validity::from_mask(	19,775✔
350	array.validity_mask()?,	19,775✔
351	array.dtype().nullability(),	19,775✔
352	))
353	}	19,775✔
354
355	/// Create Validity from boolean array with given nullability of the array.
356	///
357	/// Note: You want to pass the nullability of parent array and not the nullability of the validity array itself
358	/// as that is always nonnullable
359	fn from_array(value: ArrayRef, nullability: Nullability) -> Self {	2,784✔
360	if !matches!(value.dtype(), DType::Bool(Nullability::NonNullable)) {	2,784✔
361	vortex_panic!("Expected a non-nullable boolean array")	×
362	}	2,784✔
363	match nullability {	2,784✔
364	Nullability::NonNullable => Self::NonNullable,	×
365	Nullability::Nullable => Self::Array(value),	2,784✔
366	}
367	}	2,784✔
368
369	/// Returns the length of the validity array, if it exists.
370	pub fn maybe_len(&self) -> Option<usize> {	5,378,479✔
371	match self {	5,378,479✔
372	Self::NonNullable \| Self::AllValid \| Self::AllInvalid => None,	5,249,475✔
373	Self::Array(a) => Some(a.len()),	129,004✔
374	}
375	}	5,378,479✔
376
377	pub fn uncompressed_size(&self) -> usize {	×
378	if let Validity::Array(a) = self {	×
379	a.len().div_ceil(8)	×
380	} else {
381	0	×
382	}
383	}	×
384
385	pub fn is_array(&self) -> bool {	×
386	matches!(self, Validity::Array(_))	×
387	}	×
388	}
389
390	impl PartialEq for Validity {
391	fn eq(&self, other: &Self) -> bool {	30,651✔
392	match (self, other) {	30,651✔
393	(Self::NonNullable, Self::NonNullable) => true,	16,290✔
394	(Self::AllValid, Self::AllValid) => true,	126✔
395	(Self::AllInvalid, Self::AllInvalid) => true,	123✔
396	(Self::Array(a), Self::Array(b)) => {	526✔
397	let a = a	526✔
398	.to_bool()	526✔
399	.vortex_expect("Failed to get Validity Array as BoolArray");	526✔
400	let b = b	526✔
401	.to_bool()	526✔
402	.vortex_expect("Failed to get Validity Array as BoolArray");	526✔
403	a.boolean_buffer() == b.boolean_buffer()	526✔
404	}
405	_ => false,	13,586✔
406	}
407	}	30,651✔
408	}
409
410	impl From<BooleanBuffer> for Validity {
411	fn from(value: BooleanBuffer) -> Self {	68,659✔
412	if value.count_set_bits() == value.len() {	68,659✔
413	Self::AllValid	374✔
414	} else if value.count_set_bits() == 0 {	68,285✔
415	Self::AllInvalid	2,235✔
416	} else {
417	Self::Array(BoolArray::from(value).into_array())	66,050✔
418	}
419	}	68,659✔
420	}
421
422	impl From<NullBuffer> for Validity {
423	fn from(value: NullBuffer) -> Self {	14,942✔
424	value.into_inner().into()	14,942✔
425	}	14,942✔
426	}
427
428	impl FromIterator<Mask> for Validity {
429	fn from_iter<T: IntoIterator<Item = Mask>>(iter: T) -> Self {	×
430	Validity::from_mask(iter.into_iter().collect(), Nullability::Nullable)	×
431	}	×
432	}
433
434	impl FromIterator<bool> for Validity {
435	fn from_iter<T: IntoIterator<Item = bool>>(iter: T) -> Self {	3,391✔
436	Validity::from(BooleanBuffer::from_iter(iter))	3,391✔
437	}	3,391✔
438	}
439
440	impl From<Nullability> for Validity {
441	fn from(value: Nullability) -> Self {	49,767✔
442	match value {	49,767✔
443	Nullability::NonNullable => Validity::NonNullable,	43,781✔
444	Nullability::Nullable => Validity::AllValid,	5,986✔
445	}
446	}	49,767✔
447	}
448
449	impl Validity {
450	pub fn from_mask(mask: Mask, nullability: Nullability) -> Self {	20,541✔
451	assert!(	20,539✔
452	nullability == Nullability::Nullable \|\| matches!(mask, Mask::AllTrue(_)),	20,541✔
453	"NonNullable validity must be AllValid",	2✔
454	);
455	match mask {	20,539✔
456	Mask::AllTrue(_) => match nullability {	17,720✔
457	Nullability::NonNullable => Validity::NonNullable,	17,442✔
458	Nullability::Nullable => Validity::AllValid,	278✔
459	},
460	Mask::AllFalse(_) => Validity::AllInvalid,	78✔
461	Mask::Values(values) => Validity::Array(values.into_array()),	2,741✔
462	}
463	}	20,539✔
464	}
465
466	impl IntoArray for Mask {
467	fn into_array(self) -> ArrayRef {	40✔
468	match self {	40✔
469	Self::AllTrue(len) => ConstantArray::new(true, len).into_array(),	×
470	Self::AllFalse(len) => ConstantArray::new(false, len).into_array(),	×
471	Self::Values(a) => a.into_array(),	40✔
472	}
473	}	40✔
474	}
475
476	impl IntoArray for &MaskValues {
477	fn into_array(self) -> ArrayRef {	2,820✔
478	BoolArray::new(self.boolean_buffer().clone(), Validity::NonNullable).into_array()	2,820✔
479	}	2,820✔
480	}
481
482	#[cfg(test)]
483	mod tests {
484	use rstest::rstest;
485	use vortex_buffer::{Buffer, buffer};
486	use vortex_dtype::Nullability;
487	use vortex_mask::Mask;
488
489	use crate::arrays::{BoolArray, PrimitiveArray};
490	use crate::validity::Validity;
491	use crate::{ArrayRef, IntoArray};
492
493	#[rstest]
494	#[case(Validity::AllValid, 5, &[2, 4], Validity::AllValid, Validity::AllValid)]
495	#[case(Validity::AllValid, 5, &[2, 4], Validity::AllInvalid, Validity::Array(BoolArray::from_iter([true, true, false, true, false]).into_array())
496	)]
497	#[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())
498	)]
499	#[case(Validity::AllInvalid, 5, &[2, 4], Validity::AllValid, Validity::Array(BoolArray::from_iter([false, false, true, false, true]).into_array())
500	)]
501	#[case(Validity::AllInvalid, 5, &[2, 4], Validity::AllInvalid, Validity::AllInvalid)]
502	#[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())
503	)]
504	#[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())
505	)]
506	#[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())
507	)]
508	#[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())
509	)]
510	fn patch_validity(
511	#[case] validity: Validity,
512	#[case] len: usize,
513	#[case] positions: &[u64],
514	#[case] patches: Validity,
515	#[case] expected: Validity,
516	) {
517	let indices =
518	PrimitiveArray::new(Buffer::copy_from(positions), Validity::NonNullable).into_array();
519	assert_eq!(
520	validity.patch(len, 0, &indices, &patches).unwrap(),
521	expected
522	);
523	}
524
525	#[test]
526	#[should_panic]
527	fn out_of_bounds_patch() {	1✔
528	Validity::NonNullable	1✔
529	.patch(2, 0, &buffer![4].into_array(), &Validity::AllInvalid)	1✔
530	.unwrap();	1✔
531	}	1✔
532
533	#[test]
534	#[should_panic]
535	fn into_validity_nullable() {	1✔
536	Validity::from_mask(Mask::AllFalse(10), Nullability::NonNullable);	1✔
537	}	1✔
538
539	#[test]
540	#[should_panic]
541	fn into_validity_nullable_array() {	1✔
542	Validity::from_mask(Mask::from_iter(vec![true, false]), Nullability::NonNullable);	1✔
543	}	1✔
544
545	#[rstest]
546	#[case(Validity::AllValid, PrimitiveArray::new(buffer![0, 1], Validity::from_iter(vec![true, false])).into_array(), Validity::from_iter(vec![true, false]))]
547	#[case(Validity::AllValid, buffer![0, 1].into_array(), Validity::AllValid)]
548	#[case(Validity::AllValid, PrimitiveArray::new(buffer![0, 1], Validity::AllInvalid).into_array(), Validity::AllInvalid)]
549	#[case(Validity::NonNullable, PrimitiveArray::new(buffer![0, 1], Validity::from_iter(vec![true, false])).into_array(), Validity::from_iter(vec![true, false]))]
550	#[case(Validity::NonNullable, buffer![0, 1].into_array(), Validity::NonNullable)]
551	#[case(Validity::NonNullable, PrimitiveArray::new(buffer![0, 1], Validity::AllInvalid).into_array(), Validity::AllInvalid)]
552	fn validity_take(
553	#[case] validity: Validity,
554	#[case] indices: ArrayRef,
555	#[case] expected: Validity,
556	) {
557	assert_eq!(validity.take(&indices).unwrap(), expected);
558	}
559	}

vortex-data / vortex / 17075133033

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