vortex_array/scalar_fn/fns/
is_null.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use std::fmt::Formatter;
5
6use vortex_error::VortexExpect;
7use vortex_error::VortexResult;
8use vortex_session::VortexSession;
9
10use crate::ArrayRef;
11use crate::IntoArray;
12use crate::arrays::ConstantArray;
13use crate::builtins::ArrayBuiltins;
14use crate::dtype::DType;
15use crate::dtype::Nullability;
16use crate::expr::Expression;
17use crate::expr::StatsCatalog;
18use crate::expr::eq;
19use crate::expr::lit;
20use crate::expr::stats::Stat;
21use crate::scalar_fn::Arity;
22use crate::scalar_fn::ChildName;
23use crate::scalar_fn::EmptyOptions;
24use crate::scalar_fn::ExecutionArgs;
25use crate::scalar_fn::ScalarFnId;
26use crate::scalar_fn::ScalarFnVTable;
27use crate::validity::Validity;
28
29/// Expression that checks for null values.
30#[derive(Clone)]
31pub struct IsNull;
32
33impl ScalarFnVTable for IsNull {
34    type Options = EmptyOptions;
35
36    fn id(&self) -> ScalarFnId {
37        ScalarFnId::new_ref("is_null")
38    }
39
40    fn serialize(&self, _instance: &Self::Options) -> VortexResult<Option<Vec<u8>>> {
41        Ok(Some(vec![]))
42    }
43
44    fn deserialize(
45        &self,
46        _metadata: &[u8],
47        _session: &VortexSession,
48    ) -> VortexResult<Self::Options> {
49        Ok(EmptyOptions)
50    }
51
52    fn arity(&self, _options: &Self::Options) -> Arity {
53        Arity::Exact(1)
54    }
55
56    fn child_name(&self, _instance: &Self::Options, child_idx: usize) -> ChildName {
57        match child_idx {
58            0 => ChildName::from("input"),
59            _ => unreachable!("Invalid child index {} for IsNull expression", child_idx),
60        }
61    }
62
63    fn fmt_sql(
64        &self,
65        _options: &Self::Options,
66        expr: &Expression,
67        f: &mut Formatter<'_>,
68    ) -> std::fmt::Result {
69        write!(f, "is_null(")?;
70        expr.child(0).fmt_sql(f)?;
71        write!(f, ")")
72    }
73
74    fn return_dtype(&self, _options: &Self::Options, _arg_dtypes: &[DType]) -> VortexResult<DType> {
75        Ok(DType::Bool(Nullability::NonNullable))
76    }
77
78    fn execute(&self, _data: &Self::Options, mut args: ExecutionArgs) -> VortexResult<ArrayRef> {
79        let child = args.inputs.pop().vortex_expect("Missing input child");
80        if let Some(scalar) = child.as_constant() {
81            return Ok(ConstantArray::new(scalar.is_null(), args.row_count).into_array());
82        }
83
84        match child.validity()? {
85            Validity::NonNullable | Validity::AllValid => {
86                Ok(ConstantArray::new(false, args.row_count).into_array())
87            }
88            Validity::AllInvalid => Ok(ConstantArray::new(true, args.row_count).into_array()),
89            Validity::Array(a) => a.not(),
90        }
91    }
92
93    fn stat_falsification(
94        &self,
95        _options: &Self::Options,
96        expr: &Expression,
97        catalog: &dyn StatsCatalog,
98    ) -> Option<Expression> {
99        let null_count_expr = expr.child(0).stat_expression(Stat::NullCount, catalog)?;
100        Some(eq(null_count_expr, lit(0u64)))
101    }
102
103    fn is_null_sensitive(&self, _instance: &Self::Options) -> bool {
104        true
105    }
106
107    fn is_fallible(&self, _instance: &Self::Options) -> bool {
108        false
109    }
110}
111
112#[cfg(test)]
113mod tests {
114    use vortex_buffer::buffer;
115    use vortex_error::VortexExpect as _;
116    use vortex_utils::aliases::hash_map::HashMap;
117    use vortex_utils::aliases::hash_set::HashSet;
118
119    use crate::IntoArray;
120    use crate::arrays::PrimitiveArray;
121    use crate::arrays::StructArray;
122    use crate::dtype::DType;
123    use crate::dtype::Field;
124    use crate::dtype::FieldPath;
125    use crate::dtype::FieldPathSet;
126    use crate::dtype::Nullability;
127    use crate::expr::col;
128    use crate::expr::eq;
129    use crate::expr::get_item;
130    use crate::expr::is_null;
131    use crate::expr::lit;
132    use crate::expr::pruning::checked_pruning_expr;
133    use crate::expr::root;
134    use crate::expr::stats::Stat;
135    use crate::expr::test_harness;
136    use crate::scalar::Scalar;
137
138    #[test]
139    fn dtype() {
140        let dtype = test_harness::struct_dtype();
141        assert_eq!(
142            is_null(root()).return_dtype(&dtype).unwrap(),
143            DType::Bool(Nullability::NonNullable)
144        );
145    }
146
147    #[test]
148    fn replace_children() {
149        let expr = is_null(root());
150        expr.with_children([root()])
151            .vortex_expect("operation should succeed in test");
152    }
153
154    #[test]
155    fn evaluate_mask() {
156        let test_array =
157            PrimitiveArray::from_option_iter(vec![Some(1), None, Some(2), None, Some(3)])
158                .into_array();
159        let expected = [false, true, false, true, false];
160
161        let result = test_array.clone().apply(&is_null(root())).unwrap();
162
163        assert_eq!(result.len(), test_array.len());
164        assert_eq!(result.dtype(), &DType::Bool(Nullability::NonNullable));
165
166        for (i, expected_value) in expected.iter().enumerate() {
167            assert_eq!(
168                result.scalar_at(i).unwrap(),
169                Scalar::bool(*expected_value, Nullability::NonNullable)
170            );
171        }
172    }
173
174    #[test]
175    fn evaluate_all_false() {
176        let test_array = buffer![1, 2, 3, 4, 5].into_array();
177
178        let result = test_array.clone().apply(&is_null(root())).unwrap();
179
180        assert_eq!(result.len(), test_array.len());
181        // All values should be false (non-nullable input)
182        for i in 0..result.len() {
183            assert_eq!(
184                result.scalar_at(i).unwrap(),
185                Scalar::bool(false, Nullability::NonNullable)
186            );
187        }
188    }
189
190    #[test]
191    fn evaluate_all_true() {
192        let test_array =
193            PrimitiveArray::from_option_iter(vec![None::<i32>, None, None, None, None])
194                .into_array();
195
196        let result = test_array.clone().apply(&is_null(root())).unwrap();
197
198        assert_eq!(result.len(), test_array.len());
199        // All values should be true (all nulls)
200        for i in 0..result.len() {
201            assert_eq!(
202                result.scalar_at(i).unwrap(),
203                Scalar::bool(true, Nullability::NonNullable)
204            );
205        }
206    }
207
208    #[test]
209    fn evaluate_struct() {
210        let test_array = StructArray::from_fields(&[(
211            "a",
212            PrimitiveArray::from_option_iter(vec![Some(1), None, Some(2), None, Some(3)])
213                .into_array(),
214        )])
215        .unwrap()
216        .into_array();
217        let expected = [false, true, false, true, false];
218
219        let result = test_array
220            .clone()
221            .apply(&is_null(get_item("a", root())))
222            .unwrap();
223
224        assert_eq!(result.len(), test_array.len());
225        assert_eq!(result.dtype(), &DType::Bool(Nullability::NonNullable));
226
227        for (i, expected_value) in expected.iter().enumerate() {
228            assert_eq!(
229                result.scalar_at(i).unwrap(),
230                Scalar::bool(*expected_value, Nullability::NonNullable)
231            );
232        }
233    }
234
235    #[test]
236    fn test_display() {
237        let expr = is_null(get_item("name", root()));
238        assert_eq!(expr.to_string(), "is_null($.name)");
239
240        let expr2 = is_null(root());
241        assert_eq!(expr2.to_string(), "is_null($)");
242    }
243
244    #[test]
245    fn test_is_null_falsification() {
246        let expr = is_null(col("a"));
247
248        let (pruning_expr, st) = checked_pruning_expr(
249            &expr,
250            &FieldPathSet::from_iter([FieldPath::from_iter([
251                Field::Name("a".into()),
252                Field::Name("null_count".into()),
253            ])]),
254        )
255        .unwrap();
256
257        assert_eq!(&pruning_expr, &eq(col("a_null_count"), lit(0u64)));
258        assert_eq!(
259            st.map(),
260            &HashMap::from_iter([(FieldPath::from_name("a"), HashSet::from([Stat::NullCount]))])
261        );
262    }
263
264    #[test]
265    fn test_is_null_sensitive() {
266        // is_null itself is null-sensitive
267        assert!(is_null(col("a")).signature().is_null_sensitive());
268    }
269}
vortex_array/scalar_fn/fns/is_null.rs

vortex_array/scalar_fn/fns/
is_null.rs
