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_proto::expr as pb;
14use vortex_session::VortexSession;
15
16use crate::Array;
17use crate::ArrayRef;
18use crate::ExecutionCtx;
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(
141        &self,
142        options: &Self::Options,
143        args: &dyn ExecutionArgs,
144        _ctx: &mut ExecutionCtx,
145    ) -> VortexResult<ArrayRef> {
146        let child = args.get(0)?;
147        let pattern = args.get(1)?;
148
149        arrow_like(&child, &pattern, *options)
150    }
151
152    fn validity(
153        &self,
154        _options: &Self::Options,
155        expression: &Expression,
156    ) -> VortexResult<Option<Expression>> {
157        tracing::warn!("Computing validity for LIKE expression");
158        let child_validity = expression.child(0).validity()?;
159        let pattern_validity = expression.child(1).validity()?;
160        Ok(Some(and(child_validity, pattern_validity)))
161    }
162
163    fn is_null_sensitive(&self, _instance: &Self::Options) -> bool {
164        false
165    }
166
167    fn stat_falsification(
168        &self,
169        like_opts: &LikeOptions,
170        expr: &Expression,
171        catalog: &dyn StatsCatalog,
172    ) -> Option<Expression> {
173        // Attempt to do min/max pruning for LIKE 'exact' or LIKE 'prefix%'
174
175        // Don't attempt to handle ilike or negated like
176        if like_opts.negated || like_opts.case_insensitive {
177            return None;
178        }
179
180        // Extract the pattern out
181        let pat = expr.child(1).as_::<Literal>();
182
183        // LIKE NULL is nonsensical, don't try to handle it
184        let pat_str = pat.as_utf8().value()?;
185
186        let src = expr.child(0).clone();
187        let src_min = src.stat_min(catalog)?;
188        let src_max = src.stat_max(catalog)?;
189
190        match LikeVariant::from_str(pat_str)? {
191            LikeVariant::Exact(text) => {
192                // col LIKE 'exact' ==>  col.min > 'exact' || col.max < 'exact'
193                Some(or(gt(src_min, lit(text)), lt(src_max, lit(text))))
194            }
195            LikeVariant::Prefix(prefix) => {
196                // col LIKE 'prefix%' ==> col.max < 'prefix' || col.min >= 'prefiy'
197                let succ = prefix.to_string().increment().ok()?;
198
199                Some(or(gt_eq(src_min, lit(succ)), lt(src_max, lit(prefix))))
200            }
201        }
202    }
203}
204
205/// Implementation of LIKE using the Arrow crate.
206pub(crate) fn arrow_like(
207    array: &ArrayRef,
208    pattern: &ArrayRef,
209    options: LikeOptions,
210) -> VortexResult<ArrayRef> {
211    let nullable = array.dtype().is_nullable() | pattern.dtype().is_nullable();
212    let len = array.len();
213    assert_eq!(
214        array.len(),
215        pattern.len(),
216        "Arrow Like: length mismatch for {}",
217        array.encoding_id()
218    );
219
220    // convert the pattern to the preferred array datatype
221    let lhs = Datum::try_new(array)?;
222    let rhs = Datum::try_new_with_target_datatype(pattern, lhs.data_type())?;
223
224    let result = match (options.negated, options.case_insensitive) {
225        (false, false) => arrow_string::like::like(&lhs, &rhs)?,
226        (true, false) => arrow_string::like::nlike(&lhs, &rhs)?,
227        (false, true) => arrow_string::like::ilike(&lhs, &rhs)?,
228        (true, true) => arrow_string::like::nilike(&lhs, &rhs)?,
229    };
230
231    from_arrow_array_with_len(&result, len, nullable)
232}
233
234/// Variants of the LIKE filter that we know how to turn into a stats pruning predicate.s
235#[derive(Debug, PartialEq)]
236enum LikeVariant<'a> {
237    Exact(&'a str),
238    Prefix(&'a str),
239}
240
241impl<'a> LikeVariant<'a> {
242    /// Parse a LIKE pattern string into its relevant variant
243    fn from_str(string: &str) -> Option<LikeVariant<'_>> {
244        let Some(wildcard_pos) = string.find(['%', '_']) else {
245            return Some(LikeVariant::Exact(string));
246        };
247
248        // Can't handle wildcard in the front.
249        if wildcard_pos == 0 {
250            return None;
251        }
252
253        let prefix = &string[..wildcard_pos];
254        Some(LikeVariant::Prefix(prefix))
255    }
256}
257
258#[cfg(test)]
259mod tests {
260    use crate::arrays::BoolArray;
261    use crate::assert_arrays_eq;
262    use crate::dtype::DType;
263    use crate::dtype::Nullability;
264    use crate::expr::col;
265    use crate::expr::get_item;
266    use crate::expr::ilike;
267    use crate::expr::like;
268    use crate::expr::lit;
269    use crate::expr::not;
270    use crate::expr::not_ilike;
271    use crate::expr::not_like;
272    use crate::expr::pruning::pruning_expr::TrackingStatsCatalog;
273    use crate::expr::root;
274    use crate::scalar_fn::fns::like::LikeVariant;
275
276    #[test]
277    fn invert_booleans() {
278        let not_expr = not(root());
279        let bools = BoolArray::from_iter([false, true, false, false, true, true]);
280        assert_arrays_eq!(
281            bools.to_array().apply(&not_expr).unwrap(),
282            BoolArray::from_iter([true, false, true, true, false, false])
283        );
284    }
285
286    #[test]
287    fn dtype() {
288        let dtype = DType::Utf8(Nullability::NonNullable);
289        let like_expr = like(root(), lit("%test%"));
290        assert_eq!(
291            like_expr.return_dtype(&dtype).unwrap(),
292            DType::Bool(Nullability::NonNullable)
293        );
294    }
295
296    #[test]
297    fn test_display() {
298        let expr = like(get_item("name", root()), lit("%john%"));
299        assert_eq!(expr.to_string(), "$.name like \"%john%\"");
300
301        let expr2 = not_ilike(root(), lit("test*"));
302        assert_eq!(expr2.to_string(), "$ not ilike \"test*\"");
303    }
304
305    #[test]
306    fn test_like_variant() {
307        // Supported patterns
308        assert_eq!(
309            LikeVariant::from_str("simple"),
310            Some(LikeVariant::Exact("simple"))
311        );
312        assert_eq!(
313            LikeVariant::from_str("prefix%"),
314            Some(LikeVariant::Prefix("prefix"))
315        );
316        assert_eq!(
317            LikeVariant::from_str("first%rest_stuff"),
318            Some(LikeVariant::Prefix("first"))
319        );
320
321        // Unsupported patterns
322        assert_eq!(LikeVariant::from_str("%suffix"), None);
323        assert_eq!(LikeVariant::from_str("_pattern"), None);
324    }
325
326    #[test]
327    fn test_like_pushdown() {
328        // Test that LIKE prefix and exactness filters can be pushed down into stats filtering
329        // at scan time.
330        let catalog = TrackingStatsCatalog::default();
331
332        let pruning_expr = like(col("a"), lit("prefix%"))
333            .stat_falsification(&catalog)
334            .expect("LIKE stat falsification");
335
336        insta::assert_snapshot!(pruning_expr, @r#"(($.a_min >= "prefiy") or ($.a_max < "prefix"))"#);
337
338        // Multiple wildcards
339        let pruning_expr = like(col("a"), lit("pref%ix%"))
340            .stat_falsification(&catalog)
341            .expect("LIKE stat falsification");
342        insta::assert_snapshot!(pruning_expr, @r#"(($.a_min >= "preg") or ($.a_max < "pref"))"#);
343
344        let pruning_expr = like(col("a"), lit("pref_ix_"))
345            .stat_falsification(&catalog)
346            .expect("LIKE stat falsification");
347        insta::assert_snapshot!(pruning_expr, @r#"(($.a_min >= "preg") or ($.a_max < "pref"))"#);
348
349        // Exact match
350        let pruning_expr = like(col("a"), lit("exactly"))
351            .stat_falsification(&catalog)
352            .expect("LIKE stat falsification");
353        insta::assert_snapshot!(pruning_expr, @r#"(($.a_min > "exactly") or ($.a_max < "exactly"))"#);
354
355        // Suffix search skips pushdown
356        let pruning_expr = like(col("a"), lit("%suffix")).stat_falsification(&catalog);
357        assert_eq!(pruning_expr, None);
358
359        // NOT LIKE, ILIKE not supported currently
360        assert_eq!(
361            None,
362            not_like(col("a"), lit("a")).stat_falsification(&catalog)
363        );
364        assert_eq!(None, ilike(col("a"), lit("a")).stat_falsification(&catalog));
365    }
366}
vortex_array/scalar_fn/fns/like/mod.rs

vortex_array/scalar_fn/fns/like/
mod.rs
