Support Regex function (infiniflow#2059)

### What problem does this PR solve?

Support Regex function

### Type of change

- [x] New Feature (non-breaking change which adds functionality)
- [x] Test cases

Author: Ami11111

example/functions.py

@@ -19,7 +19,8 @@
 table_obj.insert(
     [{"c1": 'a', "c2": 'a'}, {"c1": 'b', "c2": 'b'}, {"c1": 'c', "c2": 'c'}, {"c1": 'd', "c2": 'd'},
         {"c1": 'abc', "c2": 'abc'}, {"c1": 'bbc', "c2": 'bbc'}, {"c1": 'cbc', "c2": 'cbc'}, {"c1": 'dbc', "c2": 'dbc'},
-        {"c1": 'abcd', "c2": 'abc'}])
+        {"c1": 'abcd', "c2": 'abc'}, 
+        {"c1": '[email protected]', "c2": 'email'}, {"c1": '[email protected]', "c2": 'email'}])
 
 #function char_length
 res = table_obj.output(["*"]).filter("char_length(c1) = 1").to_df()
@@ -34,6 +35,13 @@
 res = table_obj.output(["*"]).filter("char_length(c1) = char_length(c2)").to_df()
 print(res)
 
+#functin regex
+res = table_obj.output(["*"]).filter("regex(c1, 'bc')").to_df()
+print(res)
+
+res = table_obj.output(["*"]).filter("regex(c1, '(\w+([-+.]\w+)*)@(\w+([-.]\w+)*)\.(\w+([-.]\w+)*)')").to_df()
+print(res)
+
 res = db_obj.drop_table("function_example")
 
 infinity_obj.disconnect()

example/http/functions.sh

@@ -87,6 +87,30 @@ curl --request POST \
              "sparse_column":  {"20":7.7, "80":7.8, "90": 97.9},
              "year": 2018,
              "tensor": [[5.0, 4.2, 4.3, 4.5], [4.0, 4.2, 4.3, 4.4]]
+         },
+         {
+             "num": 5,
+             "body": "[email protected]",
+             "vec": [4.0, 4.2, 4.3, 4.5],
+             "sparse_column":  {"20":7.7, "80":7.8, "90": 97.9},
+             "year": 2018,
+             "tensor": [[5.0, 4.2, 4.3, 4.5], [4.0, 4.2, 4.3, 4.4]]
+         },
+         {
+             "num": 6,
+             "body": "test@hotmailcom",
+             "vec": [4.0, 4.2, 4.3, 4.5],
+             "sparse_column":  {"20":7.7, "80":7.8, "90": 97.9},
+             "year": 2018,
+             "tensor": [[5.0, 4.2, 4.3, 4.5], [4.0, 4.2, 4.3, 4.4]]
+         },
+         {
+             "num": 7,
+             "body": "this is a sentence including a mail address, [email protected]",
+             "vec": [4.0, 4.2, 4.3, 4.5],
+             "sparse_column":  {"20":7.7, "80":7.8, "90": 97.9},
+             "year": 2018,
+             "tensor": [[5.0, 4.2, 4.3, 4.5], [4.0, 4.2, 4.3, 4.4]]
          }
      ] '
 
@@ -134,6 +158,51 @@ curl --request GET \
         "filter": "char_length(body) > 4"
      } '
 
+# show rows of 'tbl1' where body inluding 'com'
+echo -e '\n\n-- show rows of 'tbl1' where body inluding '\'com\'''
+curl --request GET \
+     --url http://localhost:23820/databases/default_db/tables/tbl1/docs \
+     --header 'accept: application/json' \
+     --header 'content-type: application/json' \
+     --data '
+     {
+         "output":
+         [
+             "body"
+         ],
+        "filter": "regex(body, '\'com\'')"
+     } '
+
+# show rows of 'tbl1' where body including a mail address (using regex)
+echo -e '\n\n-- show rows of 'tbl1' where body including a mail address (using regex)'
+curl --request GET \
+     --url http://localhost:23820/databases/default_db/tables/tbl1/docs \
+     --header 'accept: application/json' \
+     --header 'content-type: application/json' \
+     --data '
+     {
+         "output":
+         [
+             "body"
+         ],
+        "filter": "regex(body, '\''('.'*'')'@'('.'*'')''\\''.'com\'')"
+     } '
+
+# show rows of 'tbl1' where body including a mail address (using regex)
+echo -e '\n\n-- show rows of 'tbl1' where body including a mail address (using regex)'
+curl --request GET \
+     --url http://localhost:23820/databases/default_db/tables/tbl1/docs \
+     --header 'accept: application/json' \
+     --header 'content-type: application/json' \
+     --data '
+     {
+         "output":
+         [
+             "body"
+         ],
+        "filter": "regex(body, '\''('[0-9A-Za-z_]+'('[-+.][0-9A-Za-z_]+')''*'')'@'('[0-9A-Za-z_]+'('[-.][0-9A-Za-z_]+')''*'')''\\'.'('[0-9A-Za-z_]+'('[-.][0-9A-Za-z_]+')''*'')'\'')"
+     } '
+
 # drop tbl1
 echo -e '\n\n-- drop tbl1'
 curl --request DELETE \

python/test_pysdk/test_select.py

@@ -831,4 +831,25 @@ def test_select_varchar_length(self, suffix):
                                       .astype({'c1': dtype('O'), 'c2': dtype('O')}))
 
         res = db_obj.drop_table("test_select_varchar_length"+suffix)
+        assert res.error_code == ErrorCode.OK
+
+    def test_select_regex(self, suffix):
+        db_obj = self.infinity_obj.get_database("default_db")
+        db_obj.drop_table("test_select_regex"+suffix, ConflictType.Ignore)
+        db_obj.create_table("test_select_regex"+suffix,
+                            {"c1": {"type": "varchar", "constraints": ["primary key", "not null"]},
+                             "c2": {"type": "varchar", "constraints": ["not null"]}}, ConflictType.Error)
+        table_obj = db_obj.get_table("test_select_regex"+suffix)
+        table_obj.insert(
+            [{"c1": 'a', "c2": 'a'}, {"c1": 'b', "c2": 'b'}, {"c1": 'c', "c2": 'c'}, {"c1": 'd', "c2": 'd'},
+             {"c1": 'abc', "c2": 'abc'}, {"c1": 'bbc', "c2": 'bbc'}, {"c1": 'cbc', "c2": 'cbc'}, {"c1": 'dbc', "c2": 'dbc'},])
+
+        res = table_obj.output(["*"]).filter("regex(c1, 'bc')").to_df()
+        print(res)
+        pd.testing.assert_frame_equal(res, pd.DataFrame({'c1': ('abc', 'bbc', 'cbc', 'dbc'),
+                                                         'c2': ('abc', 'bbc', 'cbc', 'dbc')})
+                                      .astype({'c1': dtype('O'), 'c2': dtype('O')}))
+
+
+        res = db_obj.drop_table("test_select_regex"+suffix)
         assert res.error_code == ErrorCode.OK

src/function/builtin_functions.cpp

@@ -48,6 +48,7 @@ import substring;
 import substract;
 import char_length;
 import md5;
+import regex;
 import default_values;
 import special_function;
 import internal_types;
@@ -117,6 +118,7 @@ void BuiltinFunctions::RegisterScalarFunction() {
     RegisterSubstringFunction(catalog_ptr_);
     RegisterCharLengthFunction(catalog_ptr_);
     RegisterMd5Function(catalog_ptr_);
+    RegisterRegexFunction(catalog_ptr_);
 }
 
 void BuiltinFunctions::RegisterTableFunction() {}

src/function/scalar/regex.cpp

@@ -0,0 +1,54 @@
+module;
+
+#include <re2/re2.h>
+
+module regex;
+
+import stl;
+import catalog;
+import status;
+import infinity_exception;
+import scalar_function;
+import scalar_function_set;
+
+import third_party;
+import logical_type;
+import internal_types;
+import data_type;
+import logger;
+import column_vector;
+
+namespace infinity {
+
+struct RegexFunction {
+    template <typename TA, typename TB, typename TC>
+    static inline void Run(TA &left, TB &right, TC &result) {
+        const char * origin_str;
+        SizeT origin_len;
+        const char * pattern_str;
+        SizeT pattern_len;
+        GetReaderValue(left, origin_str, origin_len);
+        GetReaderValue(right, pattern_str, pattern_len);
+        String origin_str_(origin_str, origin_len);
+        String pattern_str_(pattern_str, pattern_len);
+        bool match = re2::RE2::PartialMatch(origin_str_, pattern_str_);
+        result.SetValue(match);
+    }
+};
+
+
+void RegisterRegexFunction(const UniquePtr<Catalog> &catalog_ptr){
+    String func_name = "regex";
+
+    SharedPtr<ScalarFunctionSet> function_set_ptr = MakeShared<ScalarFunctionSet>(func_name);
+
+    ScalarFunction Regex_function(func_name,
+                                     {DataType(LogicalType::kVarchar), DataType(LogicalType::kVarchar)},
+                                     DataType(LogicalType::kBoolean),
+                                     &ScalarFunction::BinaryFunction<VarcharT, VarcharT, BooleanT, RegexFunction>);
+    function_set_ptr->AddFunction(Regex_function);
+
+    Catalog::AddFunctionSet(catalog_ptr.get(), function_set_ptr);
+}
+
+}

src/function/scalar/regex.cppm

@@ -0,0 +1,13 @@
+module;
+
+import stl;
+
+export module regex;
+
+namespace infinity {
+
+class Catalog;
+
+export void RegisterRegexFunction(const UniquePtr<Catalog> &catalog_ptr);
+
+}

test/sql/dql/type/varchar.slt

@@ -68,5 +68,24 @@ SELECT * FROM test_varchar_filter where md5(c1) = md5('abcdddde');
 abcdddde abcddddd 3
 abcdddde abcdddde 4
 
+statement ok
+INSERT INTO test_varchar_filter VALUES ('[email protected]', '[email protected]', 6);
+
+query X
+SELECT * FROM test_varchar_filter where regex(c1, 'abc\w+e');
+----
+abcdddde abcddddd 3
+abcdddde abcdddde 4
+
+query XI
+SELECT * FROM test_varchar_filter where regex(c1, 'ddddc');
+----
+abcddddc abcddddd 2
+
+query XII
+SELECT * FROM test_varchar_filter where regex(c1, '(\w+([-+.]\w+)*)@(\w+([-.]\w+)*)\.(\w+([-.]\w+)*)');
+----
+[email protected] [email protected] 6
+
 statement ok
 DROP TABLE test_varchar_filter;