1use crate::PhysicalExpr;
19use arrow::datatypes::{DataType, Schema};
20use arrow::record_batch::RecordBatch;
21use datafusion_common::{Result, assert_or_internal_err};
22use datafusion_expr::{ColumnarValue, Operator};
23use datafusion_physical_expr_common::datum::apply_cmp;
24use std::hash::Hash;
25use std::sync::Arc;
26
27#[derive(Debug, Eq)]
29pub struct LikeExpr {
30 negated: bool,
31 case_insensitive: bool,
32 expr: Arc<dyn PhysicalExpr>,
33 pattern: Arc<dyn PhysicalExpr>,
34}
35
36impl PartialEq for LikeExpr {
38 fn eq(&self, other: &Self) -> bool {
39 self.negated == other.negated
40 && self.case_insensitive == other.case_insensitive
41 && self.expr.eq(&other.expr)
42 && self.pattern.eq(&other.pattern)
43 }
44}
45
46impl Hash for LikeExpr {
47 fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
48 self.negated.hash(state);
49 self.case_insensitive.hash(state);
50 self.expr.hash(state);
51 self.pattern.hash(state);
52 }
53}
54
55impl LikeExpr {
56 pub fn new(
57 negated: bool,
58 case_insensitive: bool,
59 expr: Arc<dyn PhysicalExpr>,
60 pattern: Arc<dyn PhysicalExpr>,
61 ) -> Self {
62 Self {
63 negated,
64 case_insensitive,
65 expr,
66 pattern,
67 }
68 }
69
70 pub fn negated(&self) -> bool {
72 self.negated
73 }
74
75 pub fn case_insensitive(&self) -> bool {
77 self.case_insensitive
78 }
79
80 pub fn expr(&self) -> &Arc<dyn PhysicalExpr> {
82 &self.expr
83 }
84
85 pub fn pattern(&self) -> &Arc<dyn PhysicalExpr> {
87 &self.pattern
88 }
89
90 fn op_name(&self) -> &str {
92 match (self.negated, self.case_insensitive) {
93 (false, false) => "LIKE",
94 (true, false) => "NOT LIKE",
95 (false, true) => "ILIKE",
96 (true, true) => "NOT ILIKE",
97 }
98 }
99}
100
101impl std::fmt::Display for LikeExpr {
102 fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
103 write!(f, "{} {} {}", self.expr, self.op_name(), self.pattern)
104 }
105}
106
107impl PhysicalExpr for LikeExpr {
108 fn data_type(&self, _input_schema: &Schema) -> Result<DataType> {
109 Ok(DataType::Boolean)
110 }
111
112 fn nullable(&self, input_schema: &Schema) -> Result<bool> {
113 Ok(self.expr.nullable(input_schema)? || self.pattern.nullable(input_schema)?)
114 }
115
116 fn evaluate(&self, batch: &RecordBatch) -> Result<ColumnarValue> {
117 let lhs = self.expr.evaluate(batch)?;
118 let rhs = self.pattern.evaluate(batch)?;
119 match (self.negated, self.case_insensitive) {
120 (false, false) => apply_cmp(Operator::LikeMatch, &lhs, &rhs),
121 (false, true) => apply_cmp(Operator::ILikeMatch, &lhs, &rhs),
122 (true, false) => apply_cmp(Operator::NotLikeMatch, &lhs, &rhs),
123 (true, true) => apply_cmp(Operator::NotILikeMatch, &lhs, &rhs),
124 }
125 }
126
127 fn children(&self) -> Vec<&Arc<dyn PhysicalExpr>> {
128 vec![&self.expr, &self.pattern]
129 }
130
131 fn with_new_children(
132 self: Arc<Self>,
133 children: Vec<Arc<dyn PhysicalExpr>>,
134 ) -> Result<Arc<dyn PhysicalExpr>> {
135 Ok(Arc::new(LikeExpr::new(
136 self.negated,
137 self.case_insensitive,
138 Arc::clone(&children[0]),
139 Arc::clone(&children[1]),
140 )))
141 }
142
143 fn fmt_sql(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
144 self.expr.fmt_sql(f)?;
145 write!(f, " {} ", self.op_name())?;
146 self.pattern.fmt_sql(f)
147 }
148
149 #[cfg(feature = "proto")]
150 fn try_to_proto(
151 &self,
152 ctx: &datafusion_physical_expr_common::physical_expr::proto_encode::PhysicalExprEncodeCtx<'_>,
153 ) -> Result<Option<datafusion_proto_models::protobuf::PhysicalExprNode>> {
154 use datafusion_proto_models::protobuf;
155
156 Ok(Some(protobuf::PhysicalExprNode {
157 expr_id: None,
158 expr_type: Some(protobuf::physical_expr_node::ExprType::LikeExpr(Box::new(
159 protobuf::PhysicalLikeExprNode {
160 negated: self.negated,
161 case_insensitive: self.case_insensitive,
162 expr: Some(Box::new(ctx.encode_child(&self.expr)?)),
163 pattern: Some(Box::new(ctx.encode_child(&self.pattern)?)),
164 },
165 ))),
166 }))
167 }
168}
169
170#[cfg(feature = "proto")]
171impl LikeExpr {
172 pub fn try_from_proto(
180 node: &datafusion_proto_models::protobuf::PhysicalExprNode,
181 ctx: &datafusion_physical_expr_common::physical_expr::proto_decode::PhysicalExprDecodeCtx<'_>,
182 ) -> Result<Arc<dyn PhysicalExpr>> {
183 use datafusion_physical_expr_common::expect_expr_variant;
184 use datafusion_proto_models::protobuf;
185
186 let like_expr = expect_expr_variant!(
187 node,
188 protobuf::physical_expr_node::ExprType::LikeExpr,
189 "LikeExpr",
190 );
191
192 Ok(Arc::new(LikeExpr::new(
193 like_expr.negated,
194 like_expr.case_insensitive,
195 ctx.decode_required_expression(
196 like_expr.expr.as_deref(),
197 "LikeExpr",
198 "expr",
199 )?,
200 ctx.decode_required_expression(
201 like_expr.pattern.as_deref(),
202 "LikeExpr",
203 "pattern",
204 )?,
205 )))
206 }
207}
208
209fn can_like_type(from_type: &DataType) -> bool {
211 match from_type {
212 DataType::Dictionary(_, inner_type_from) => **inner_type_from == DataType::Utf8,
213 _ => false,
214 }
215}
216
217pub fn like(
219 negated: bool,
220 case_insensitive: bool,
221 expr: Arc<dyn PhysicalExpr>,
222 pattern: Arc<dyn PhysicalExpr>,
223 input_schema: &Schema,
224) -> Result<Arc<dyn PhysicalExpr>> {
225 let expr_type = &expr.data_type(input_schema)?;
226 let pattern_type = &pattern.data_type(input_schema)?;
227 assert_or_internal_err!(
228 expr_type.eq(pattern_type) || can_like_type(expr_type),
229 "The type of {expr_type} AND {pattern_type} of like physical should be same"
230 );
231 Ok(Arc::new(LikeExpr::new(
232 negated,
233 case_insensitive,
234 expr,
235 pattern,
236 )))
237}
238
239#[cfg(test)]
240mod test {
241 use super::*;
242 use crate::expressions::col;
243 use arrow::array::*;
244 use arrow::datatypes::Field;
245 use datafusion_common::cast::as_boolean_array;
246 use datafusion_physical_expr_common::physical_expr::fmt_sql;
247
248 macro_rules! test_like {
249 ($A_VEC:expr, $B_VEC:expr, $VEC:expr, $NULLABLE: expr, $NEGATED:expr, $CASE_INSENSITIVE:expr,) => {{
250 let schema = Schema::new(vec![
251 Field::new("a", DataType::Utf8, $NULLABLE),
252 Field::new("b", DataType::Utf8, $NULLABLE),
253 ]);
254 let a = StringArray::from($A_VEC);
255 let b = StringArray::from($B_VEC);
256
257 let expression = like(
258 $NEGATED,
259 $CASE_INSENSITIVE,
260 col("a", &schema)?,
261 col("b", &schema)?,
262 &schema,
263 )?;
264 let batch = RecordBatch::try_new(
265 Arc::new(schema.clone()),
266 vec![Arc::new(a), Arc::new(b)],
267 )?;
268
269 let result = expression
271 .evaluate(&batch)?
272 .into_array(batch.num_rows())
273 .expect("Failed to convert to array");
274 let result =
275 as_boolean_array(&result).expect("failed to downcast to BooleanArray");
276 let expected = &BooleanArray::from($VEC);
277 assert_eq!(expected, result);
278 }};
279 }
280
281 #[test]
282 fn like_op() -> Result<()> {
283 test_like!(
284 vec!["hello world", "world"],
285 vec!["%hello%", "%hello%"],
286 vec![true, false],
287 false,
288 false,
289 false,
290 ); test_like!(
292 vec![Some("hello world"), None, Some("world")],
293 vec![Some("%hello%"), None, Some("%hello%")],
294 vec![Some(false), None, Some(true)],
295 true,
296 true,
297 false,
298 ); test_like!(
300 vec!["hello world", "world"],
301 vec!["%helLo%", "%helLo%"],
302 vec![true, false],
303 false,
304 false,
305 true,
306 ); test_like!(
308 vec![Some("hello world"), None, Some("world")],
309 vec![Some("%helLo%"), None, Some("%helLo%")],
310 vec![Some(false), None, Some(true)],
311 true,
312 true,
313 true,
314 ); Ok(())
317 }
318
319 #[test]
320 fn test_fmt_sql() -> Result<()> {
321 let schema = Schema::new(vec![
322 Field::new("a", DataType::Utf8, false),
323 Field::new("b", DataType::Utf8, false),
324 ]);
325
326 let expr = like(
327 false,
328 false,
329 col("a", &schema)?,
330 col("b", &schema)?,
331 &schema,
332 )?;
333
334 let display_string = expr.to_string();
336 assert_eq!(display_string, "a@0 LIKE b@1");
337
338 let sql_string = fmt_sql(expr.as_ref()).to_string();
340 assert_eq!(sql_string, "a LIKE b");
341
342 Ok(())
343 }
344}
345
346#[cfg(all(test, feature = "proto"))]
348mod proto_tests {
349 use super::*;
350 use crate::expressions::{Column, col};
351 use crate::proto_test_util::{
352 StubDecoder, StubEncoder, UnreachableDecoder, column_node,
353 };
354 use arrow::datatypes::Field;
355 use datafusion_common::DataFusionError;
356 use datafusion_physical_expr_common::physical_expr::proto_decode::PhysicalExprDecodeCtx;
357 use datafusion_physical_expr_common::physical_expr::proto_encode::PhysicalExprEncodeCtx;
358 use datafusion_proto_models::protobuf::{
359 PhysicalExprNode, PhysicalLikeExprNode, physical_expr_node,
360 };
361
362 fn like_node(
364 negated: bool,
365 case_insensitive: bool,
366 expr: Option<Box<PhysicalExprNode>>,
367 pattern: Option<Box<PhysicalExprNode>>,
368 ) -> PhysicalExprNode {
369 PhysicalExprNode {
370 expr_id: None,
371 expr_type: Some(physical_expr_node::ExprType::LikeExpr(Box::new(
372 PhysicalLikeExprNode {
373 negated,
374 case_insensitive,
375 expr,
376 pattern,
377 },
378 ))),
379 }
380 }
381
382 fn like_fixture() -> LikeExpr {
385 let schema = Schema::new(vec![
386 Field::new("a", DataType::Utf8, false),
387 Field::new("b", DataType::Utf8, false),
388 ]);
389 LikeExpr::new(
390 true,
391 true,
392 col("a", &schema).unwrap(),
393 col("b", &schema).unwrap(),
394 )
395 }
396
397 #[test]
398 fn try_to_proto_encodes_like_expr() {
399 let like = like_fixture();
400 let encoder = StubEncoder::ok();
401 let ctx = PhysicalExprEncodeCtx::new(&encoder);
402
403 let node = like
404 .try_to_proto(&ctx)
405 .unwrap()
406 .expect("LikeExpr should encode to Some(node)");
407
408 assert!(node.expr_id.is_none());
410 let like_node = match node.expr_type {
411 Some(physical_expr_node::ExprType::LikeExpr(boxed)) => *boxed,
412 other => panic!("expected a LikeExpr node, got {other:?}"),
413 };
414 assert!(like_node.negated);
415 assert!(like_node.case_insensitive);
416 assert!(like_node.expr.is_some());
417 assert!(like_node.pattern.is_some());
418 }
419
420 #[test]
421 fn try_to_proto_propagates_expr_encode_error() {
422 let like = like_fixture();
423 let encoder = StubEncoder::failing_on(1);
424 let ctx = PhysicalExprEncodeCtx::new(&encoder);
425 let err = like.try_to_proto(&ctx).unwrap_err();
426 assert!(matches!(err, DataFusionError::Internal(msg) if msg.contains("call 1")));
427 }
428
429 #[test]
430 fn try_to_proto_propagates_pattern_encode_error() {
431 let like = like_fixture();
432 let encoder = StubEncoder::failing_on(2);
433 let ctx = PhysicalExprEncodeCtx::new(&encoder);
434 let err = like.try_to_proto(&ctx).unwrap_err();
435 assert!(matches!(err, DataFusionError::Internal(msg) if msg.contains("call 2")));
436 }
437
438 #[test]
439 fn try_from_proto_decodes_like_expr() {
440 let node = like_node(
441 true,
442 true,
443 Some(Box::new(column_node("a"))),
444 Some(Box::new(column_node("b"))),
445 );
446 let schema = Schema::empty();
447 let decoder = StubDecoder::ok();
448 let ctx = PhysicalExprDecodeCtx::new(&schema, &decoder);
449
450 let decoded = LikeExpr::try_from_proto(&node, &ctx).unwrap();
451 let like = decoded
452 .downcast_ref::<LikeExpr>()
453 .expect("decoded expr should be a LikeExpr");
454 assert!(like.negated());
455 assert!(like.case_insensitive());
456 assert!(like.expr().downcast_ref::<Column>().is_some());
457 assert!(like.pattern().downcast_ref::<Column>().is_some());
458 }
459
460 #[test]
461 fn try_from_proto_rejects_non_like_node() {
462 let node = column_node("a");
463 let schema = Schema::empty();
464 let decoder = UnreachableDecoder;
465 let ctx = PhysicalExprDecodeCtx::new(&schema, &decoder);
466 let err = LikeExpr::try_from_proto(&node, &ctx).unwrap_err();
467 assert!(matches!(
468 err,
469 DataFusionError::Internal(msg) if msg.contains("PhysicalExprNode is not a LikeExpr")
470 ));
471 }
472
473 #[test]
474 fn try_from_proto_rejects_missing_expr() {
475 let node = like_node(false, false, None, Some(Box::new(column_node("b"))));
476 let schema = Schema::empty();
477 let decoder = UnreachableDecoder;
478 let ctx = PhysicalExprDecodeCtx::new(&schema, &decoder);
479 let err = LikeExpr::try_from_proto(&node, &ctx).unwrap_err();
480 assert!(matches!(
481 err,
482 DataFusionError::Internal(msg) if msg.contains("LikeExpr is missing required field 'expr'")
483 ));
484 }
485
486 #[test]
487 fn try_from_proto_rejects_missing_pattern() {
488 let node = like_node(false, false, Some(Box::new(column_node("a"))), None);
489 let schema = Schema::empty();
490 let decoder = StubDecoder::ok();
493 let ctx = PhysicalExprDecodeCtx::new(&schema, &decoder);
494 let err = LikeExpr::try_from_proto(&node, &ctx).unwrap_err();
495 assert!(matches!(
496 err,
497 DataFusionError::Internal(msg) if msg.contains("LikeExpr is missing required field 'pattern'")
498 ));
499 }
500
501 #[test]
502 fn try_from_proto_propagates_expr_decode_error() {
503 let node = like_node(
504 false,
505 false,
506 Some(Box::new(column_node("a"))),
507 Some(Box::new(column_node("b"))),
508 );
509 let schema = Schema::empty();
510 let decoder = StubDecoder::failing_on(1);
511 let ctx = PhysicalExprDecodeCtx::new(&schema, &decoder);
512 let err = LikeExpr::try_from_proto(&node, &ctx).unwrap_err();
513 assert!(matches!(err, DataFusionError::Internal(msg) if msg.contains("call 1")));
514 }
515
516 #[test]
517 fn try_from_proto_propagates_pattern_decode_error() {
518 let node = like_node(
519 false,
520 false,
521 Some(Box::new(column_node("a"))),
522 Some(Box::new(column_node("b"))),
523 );
524 let schema = Schema::empty();
525 let decoder = StubDecoder::failing_on(2);
526 let ctx = PhysicalExprDecodeCtx::new(&schema, &decoder);
527 let err = LikeExpr::try_from_proto(&node, &ctx).unwrap_err();
528 assert!(matches!(err, DataFusionError::Internal(msg) if msg.contains("call 2")));
529 }
530}