vortex_array/scalar_fn/fns/like/
mod.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4mod kernel;
5
6use std::fmt::Display;
7use std::fmt::Formatter;
8
9pub use kernel::*;
10use prost::Message;
11use vortex_error::VortexResult;
12use vortex_error::vortex_bail;
13use vortex_error::vortex_err;
14use vortex_proto::expr as pb;
15use vortex_session::VortexSession;
16
17use crate::Array;
18use crate::ArrayRef;
19use crate::arrow::Datum;
20use crate::arrow::from_arrow_array_with_len;
21use crate::dtype::DType;
22use crate::expr::Expression;
23use crate::expr::StatsCatalog;
24use crate::expr::and;
25use crate::expr::gt;
26use crate::expr::gt_eq;
27use crate::expr::lit;
28use crate::expr::lt;
29use crate::expr::or;
30use crate::scalar::StringLike;
31use crate::scalar_fn::Arity;
32use crate::scalar_fn::ChildName;
33use crate::scalar_fn::ExecutionArgs;
34use crate::scalar_fn::ScalarFnId;
35use crate::scalar_fn::ScalarFnVTable;
36use crate::scalar_fn::fns::literal::Literal;
37
38/// Options for SQL LIKE function
39#[derive(Default, Debug, Clone, Copy, PartialEq, Eq, Hash)]
40pub struct LikeOptions {
41    pub negated: bool,
42    pub case_insensitive: bool,
43}
44
45impl Display for LikeOptions {
46    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
47        if self.negated {
48            write!(f, "NOT ")?;
49        }
50        if self.case_insensitive {
51            write!(f, "ILIKE")
52        } else {
53            write!(f, "LIKE")
54        }
55    }
56}
57
58/// Expression that performs SQL LIKE pattern matching.
59#[derive(Clone)]
60pub struct Like;
61
62impl ScalarFnVTable for Like {
63    type Options = LikeOptions;
64
65    fn id(&self) -> ScalarFnId {
66        ScalarFnId::from("vortex.like")
67    }
68
69    fn serialize(&self, instance: &Self::Options) -> VortexResult<Option<Vec<u8>>> {
70        Ok(Some(
71            pb::LikeOpts {
72                negated: instance.negated,
73                case_insensitive: instance.case_insensitive,
74            }
75            .encode_to_vec(),
76        ))
77    }
78
79    fn deserialize(
80        &self,
81        _metadata: &[u8],
82        _session: &VortexSession,
83    ) -> VortexResult<Self::Options> {
84        let opts = pb::LikeOpts::decode(_metadata)?;
85        Ok(LikeOptions {
86            negated: opts.negated,
87            case_insensitive: opts.case_insensitive,
88        })
89    }
90
91    fn arity(&self, _options: &Self::Options) -> Arity {
92        Arity::Exact(2)
93    }
94
95    fn child_name(&self, _instance: &Self::Options, child_idx: usize) -> ChildName {
96        match child_idx {
97            0 => ChildName::from("child"),
98            1 => ChildName::from("pattern"),
99            _ => unreachable!("Invalid child index {} for Like expression", child_idx),
100        }
101    }
102
103    fn fmt_sql(
104        &self,
105        options: &Self::Options,
106        expr: &Expression,
107        f: &mut Formatter<'_>,
108    ) -> std::fmt::Result {
109        expr.child(0).fmt_sql(f)?;
110        if options.negated {
111            write!(f, " not")?;
112        }
113        if options.case_insensitive {
114            write!(f, " ilike ")?;
115        } else {
116            write!(f, " like ")?;
117        }
118        expr.child(1).fmt_sql(f)
119    }
120
121    fn return_dtype(&self, _options: &Self::Options, arg_dtypes: &[DType]) -> VortexResult<DType> {
122        let input = &arg_dtypes[0];
123        let pattern = &arg_dtypes[1];
124
125        if !input.is_utf8() {
126            vortex_bail!("LIKE expression requires UTF8 input dtype, got {}", input);
127        }
128        if !pattern.is_utf8() {
129            vortex_bail!(
130                "LIKE expression requires UTF8 pattern dtype, got {}",
131                pattern
132            );
133        }
134
135        Ok(DType::Bool(
136            (input.is_nullable() || pattern.is_nullable()).into(),
137        ))
138    }
139
140    fn execute(&self, options: &Self::Options, args: ExecutionArgs) -> VortexResult<ArrayRef> {
141        let [child, pattern]: [ArrayRef; _] = args
142            .inputs
143            .try_into()
144            .map_err(|_| vortex_err!("Wrong argument count"))?;
145
146        arrow_like(&child, &pattern, *options)
147    }
148
149    fn validity(
150        &self,
151        _options: &Self::Options,
152        expression: &Expression,
153    ) -> VortexResult<Option<Expression>> {
154        tracing::warn!("Computing validity for LIKE expression");
155        let child_validity = expression.child(0).validity()?;
156        let pattern_validity = expression.child(1).validity()?;
157        Ok(Some(and(child_validity, pattern_validity)))
158    }
159
160    fn is_null_sensitive(&self, _instance: &Self::Options) -> bool {
161        false
162    }
163
164    fn stat_falsification(
165        &self,
166        like_opts: &LikeOptions,
167        expr: &Expression,
168        catalog: &dyn StatsCatalog,
169    ) -> Option<Expression> {
170        // Attempt to do min/max pruning for LIKE 'exact' or LIKE 'prefix%'
171
172        // Don't attempt to handle ilike or negated like
173        if like_opts.negated || like_opts.case_insensitive {
174            return None;
175        }
176
177        // Extract the pattern out
178        let pat = expr.child(1).as_::<Literal>();
179
180        // LIKE NULL is nonsensical, don't try to handle it
181        let pat_str = pat.as_utf8().value()?;
182
183        let src = expr.child(0).clone();
184        let src_min = src.stat_min(catalog)?;
185        let src_max = src.stat_max(catalog)?;
186
187        match LikeVariant::from_str(pat_str)? {
188            LikeVariant::Exact(text) => {
189                // col LIKE 'exact' ==>  col.min > 'exact' || col.max < 'exact'
190                Some(or(gt(src_min, lit(text)), lt(src_max, lit(text))))
191            }
192            LikeVariant::Prefix(prefix) => {
193                // col LIKE 'prefix%' ==> col.max < 'prefix' || col.min >= 'prefiy'
194                let succ = prefix.to_string().increment().ok()?;
195
196                Some(or(gt_eq(src_min, lit(succ)), lt(src_max, lit(prefix))))
197            }
198        }
199    }
200}
201
202/// Implementation of LIKE using the Arrow crate.
203pub(crate) fn arrow_like(
204    array: &dyn Array,
205    pattern: &dyn Array,
206    options: LikeOptions,
207) -> VortexResult<ArrayRef> {
208    let nullable = array.dtype().is_nullable() | pattern.dtype().is_nullable();
209    let len = array.len();
210    assert_eq!(
211        array.len(),
212        pattern.len(),
213        "Arrow Like: length mismatch for {}",
214        array.encoding_id()
215    );
216
217    // convert the pattern to the preferred array datatype
218    let lhs = Datum::try_new(array)?;
219    let rhs = Datum::try_new_with_target_datatype(pattern, lhs.data_type())?;
220
221    let result = match (options.negated, options.case_insensitive) {
222        (false, false) => arrow_string::like::like(&lhs, &rhs)?,
223        (true, false) => arrow_string::like::nlike(&lhs, &rhs)?,
224        (false, true) => arrow_string::like::ilike(&lhs, &rhs)?,
225        (true, true) => arrow_string::like::nilike(&lhs, &rhs)?,
226    };
227
228    from_arrow_array_with_len(&result, len, nullable)
229}
230
231/// Variants of the LIKE filter that we know how to turn into a stats pruning predicate.s
232#[derive(Debug, PartialEq)]
233enum LikeVariant<'a> {
234    Exact(&'a str),
235    Prefix(&'a str),
236}
237
238impl<'a> LikeVariant<'a> {
239    /// Parse a LIKE pattern string into its relevant variant
240    fn from_str(string: &str) -> Option<LikeVariant<'_>> {
241        let Some(wildcard_pos) = string.find(['%', '_']) else {
242            return Some(LikeVariant::Exact(string));
243        };
244
245        // Can't handle wildcard in the front.
246        if wildcard_pos == 0 {
247            return None;
248        }
249
250        let prefix = &string[..wildcard_pos];
251        Some(LikeVariant::Prefix(prefix))
252    }
253}
254
255#[cfg(test)]
256mod tests {
257    use crate::arrays::BoolArray;
258    use crate::assert_arrays_eq;
259    use crate::dtype::DType;
260    use crate::dtype::Nullability;
261    use crate::expr::col;
262    use crate::expr::get_item;
263    use crate::expr::ilike;
264    use crate::expr::like;
265    use crate::expr::lit;
266    use crate::expr::not;
267    use crate::expr::not_ilike;
268    use crate::expr::not_like;
269    use crate::expr::pruning::pruning_expr::TrackingStatsCatalog;
270    use crate::expr::root;
271    use crate::scalar_fn::fns::like::LikeVariant;
272
273    #[test]
274    fn invert_booleans() {
275        let not_expr = not(root());
276        let bools = BoolArray::from_iter([false, true, false, false, true, true]);
277        assert_arrays_eq!(
278            bools.to_array().apply(&not_expr).unwrap(),
279            BoolArray::from_iter([true, false, true, true, false, false])
280        );
281    }
282
283    #[test]
284    fn dtype() {
285        let dtype = DType::Utf8(Nullability::NonNullable);
286        let like_expr = like(root(), lit("%test%"));
287        assert_eq!(
288            like_expr.return_dtype(&dtype).unwrap(),
289            DType::Bool(Nullability::NonNullable)
290        );
291    }
292
293    #[test]
294    fn test_display() {
295        let expr = like(get_item("name", root()), lit("%john%"));
296        assert_eq!(expr.to_string(), "$.name like \"%john%\"");
297
298        let expr2 = not_ilike(root(), lit("test*"));
299        assert_eq!(expr2.to_string(), "$ not ilike \"test*\"");
300    }
301
302    #[test]
303    fn test_like_variant() {
304        // Supported patterns
305        assert_eq!(
306            LikeVariant::from_str("simple"),
307            Some(LikeVariant::Exact("simple"))
308        );
309        assert_eq!(
310            LikeVariant::from_str("prefix%"),
311            Some(LikeVariant::Prefix("prefix"))
312        );
313        assert_eq!(
314            LikeVariant::from_str("first%rest_stuff"),
315            Some(LikeVariant::Prefix("first"))
316        );
317
318        // Unsupported patterns
319        assert_eq!(LikeVariant::from_str("%suffix"), None);
320        assert_eq!(LikeVariant::from_str("_pattern"), None);
321    }
322
323    #[test]
324    fn test_like_pushdown() {
325        // Test that LIKE prefix and exactness filters can be pushed down into stats filtering
326        // at scan time.
327        let catalog = TrackingStatsCatalog::default();
328
329        let pruning_expr = like(col("a"), lit("prefix%"))
330            .stat_falsification(&catalog)
331            .expect("LIKE stat falsification");
332
333        insta::assert_snapshot!(pruning_expr, @r#"(($.a_min >= "prefiy") or ($.a_max < "prefix"))"#);
334
335        // Multiple wildcards
336        let pruning_expr = like(col("a"), lit("pref%ix%"))
337            .stat_falsification(&catalog)
338            .expect("LIKE stat falsification");
339        insta::assert_snapshot!(pruning_expr, @r#"(($.a_min >= "preg") or ($.a_max < "pref"))"#);
340
341        let pruning_expr = like(col("a"), lit("pref_ix_"))
342            .stat_falsification(&catalog)
343            .expect("LIKE stat falsification");
344        insta::assert_snapshot!(pruning_expr, @r#"(($.a_min >= "preg") or ($.a_max < "pref"))"#);
345
346        // Exact match
347        let pruning_expr = like(col("a"), lit("exactly"))
348            .stat_falsification(&catalog)
349            .expect("LIKE stat falsification");
350        insta::assert_snapshot!(pruning_expr, @r#"(($.a_min > "exactly") or ($.a_max < "exactly"))"#);
351
352        // Suffix search skips pushdown
353        let pruning_expr = like(col("a"), lit("%suffix")).stat_falsification(&catalog);
354        assert_eq!(pruning_expr, None);
355
356        // NOT LIKE, ILIKE not supported currently
357        assert_eq!(
358            None,
359            not_like(col("a"), lit("a")).stat_falsification(&catalog)
360        );
361        assert_eq!(None, ilike(col("a"), lit("a")).stat_falsification(&catalog));
362    }
363}
vortex_array/scalar_fn/fns/like/mod.rs

vortex_array/scalar_fn/fns/like/
mod.rs
