chore(query): add glob function (#17666)

* chore(query): add glob

* add fuzz join tests

* add fuzz join tests

Author: sundy-li

Cargo.lock

@@ -316,6 +316,7 @@ geo = { version = "0.28.0", features = ["use-serde"] }
 geohash = "0.13.0"
 geozero = { version = "0.14.0", features = ["with-geo", "with-geojson", "with-wkb", "with-wkt"] }
 gimli = "0.31.0"
+glob = "0.3.0"
 globiter = "0.1"
 goldenfile = "1.4"
 h3o = "0.4.0"

Cargo.toml

@@ -34,6 +34,7 @@ ethnum = { workspace = true }
 geo = { workspace = true }
 geohash = { workspace = true }
 geozero = { workspace = true }
+glob = { workspace = true }
 h3o = { workspace = true }
 hex = { workspace = true }
 itertools = { workspace = true }

src/query/functions/Cargo.toml

@@ -39,6 +39,7 @@ use databend_common_expression::types::ValueType;
 use databend_common_expression::types::VariantType;
 use databend_common_expression::types::ALL_NUMBER_CLASSES;
 use databend_common_expression::values::Value;
+use databend_common_expression::vectorize_with_builder_2_arg;
 use databend_common_expression::with_number_mapped_type;
 use databend_common_expression::Column;
 use databend_common_expression::EvalContext;
@@ -611,6 +612,24 @@ fn register_like(registry: &mut FunctionRegistry) {
             }
         }),
     );
+
+    registry.register_passthrough_nullable_2_arg::<StringType, StringType, BooleanType, _, _>(
+        "glob",
+        |_, _, _| FunctionDomain::Full,
+        vectorize_with_builder_2_arg::<StringType, StringType, BooleanType>(
+            |a, b, builder, _ctx| {
+                // Create a glob pattern from the second argument
+                let pattern = match glob::Pattern::new(b) {
+                    Ok(pattern) => pattern,
+                    Err(_) => {
+                        builder.push(false);
+                        return;
+                    }
+                };
+                builder.push(pattern.matches(a));
+            },
+        ),
+    );
 }
 
 fn vectorize_like(

src/query/functions/src/scalars/comparison.rs

@@ -502,8 +502,6 @@ pub(crate) mod jaro_winkler {
     //! # Examples
     //!
     //! ```
-    //! use jaro_winkler::jaro_winkler;
-    //!
     //! assert_eq!(jaro_winkler("martha", "marhta"), 0.9611111111111111);
     //! assert_eq!(jaro_winkler("", "words"), 0.0);
     //! assert_eq!(jaro_winkler("same", "same"), 1.0);

src/query/functions/src/scalars/other.rs

@@ -30,6 +30,7 @@ fn test_string() {
     test_regexp_like(regexp_file);
     test_regexp_replace(regexp_file);
     test_regexp_substr(regexp_file);
+    test_glob(regexp_file);
 }
 
 fn test_regexp_instr(file: &mut impl Write) {
@@ -257,6 +258,13 @@ fn test_regexp_instr(file: &mut impl Write) {
     );
 }
 
+fn test_glob(file: &mut impl Write) {
+    run_ast(file, "glob('Michael!', '*')", &[]);
+    run_ast(file, "glob('a', 'a')", &[]);
+    run_ast(file, "glob('abc', 'a*')", &[]);
+    run_ast(file, "glob('abc', '*bc')", &[]);
+}
+
 fn test_regexp_like(file: &mut impl Write) {
     run_ast(file, "regexp_like('Michael!', '.*')", &[]);
     run_ast(file, "regexp_like('a', '^[a-d]')", &[]);

src/query/functions/tests/it/scalars/regexp.rs

@@ -1809,6 +1809,8 @@ Functions overloads:
 1 get_string(Variant NULL, String NULL) :: String NULL
 2 get_string(Variant, Int64) :: String NULL
 3 get_string(Variant NULL, Int64 NULL) :: String NULL
+0 glob(String, String) :: Boolean
+1 glob(String NULL, String NULL) :: Boolean NULL
 0 great_circle_angle(Float64, Float64, Float64, Float64) :: Float32
 1 great_circle_angle(Float64 NULL, Float64 NULL, Float64 NULL, Float64 NULL) :: Float32 NULL
 0 great_circle_distance(Float64, Float64, Float64, Float64) :: Float32

src/query/functions/tests/it/scalars/testdata/function_list.txt

@@ -938,3 +938,39 @@ error:
 
 
 
+ast            : glob('Michael!', '*')
+raw expr       : glob('Michael!', '*')
+checked expr   : glob<String, String>("Michael!", "*")
+optimized expr : true
+output type    : Boolean
+output domain  : {TRUE}
+output         : true
+
+
+ast            : glob('a', 'a')
+raw expr       : glob('a', 'a')
+checked expr   : glob<String, String>("a", "a")
+optimized expr : true
+output type    : Boolean
+output domain  : {TRUE}
+output         : true
+
+
+ast            : glob('abc', 'a*')
+raw expr       : glob('abc', 'a*')
+checked expr   : glob<String, String>("abc", "a*")
+optimized expr : true
+output type    : Boolean
+output domain  : {TRUE}
+output         : true
+
+
+ast            : glob('abc', '*bc')
+raw expr       : glob('abc', '*bc')
+checked expr   : glob<String, String>("abc", "*bc")
+optimized expr : true
+output type    : Boolean
+output domain  : {TRUE}
+output         : true
+
+

src/query/functions/tests/it/scalars/testdata/regexp.txt

@@ -351,11 +351,57 @@ select regexp_like('µå周çб', '周')
 1
 
 
-
 query B
 select regexp_like('周周周周', '.*')
 ----
 1
 
+query B
+SELECT GLOB('abc', 'a?c')
+----
+1
+
+query B
+SELECT GLOB('abc', 'a??')
+----
+1
 
+query B
+SELECT GLOB('abc', 'a*')
+----
+1
+
+query B
+SELECT GLOB('abc', 'a*c')
+----
+1
+
+query B
+SELECT GLOB('abc', 'a*d')
+----
+0
 
+query B
+SELECT GLOB('abc', '?bc')
+----
+1
+
+query B
+SELECT GLOB('abc', '*bc')
+----
+1
+
+query B
+SELECT GLOB('abc', '*bd')
+----
+0
+
+query B
+SELECT GLOB('abc', 'abc')
+----
+1
+
+query B
+SELECT GLOB('abc', 'abcd')
+----
+0

tests/sqllogictests/suites/query/functions/02_0047_function_string_regexp_like.test

@@ -0,0 +1,105 @@
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true
+1
+1
+1
+1
+true