Fix extend whtie space char. (#5213)

* Fix extend whtie space char.

* Format.

Co-authored-by: Sophie <84560950+Sophie-Xie@users.noreply.github.com>

src/parser/scanner.lex

@@ -35,7 +35,8 @@ static constexpr size_t MAX_STRING = 4096;
 %x LB_STR
 %x COMMENT
 
-blank_without_newline       ([ \t\r\xa0])
+nbsp                        (\xc2\xa0)
+blank_without_newline       ([ \t\r]|{nbsp})
 blank                       ({blank_without_newline}|[\n])
 
 blanks                      ({blank}+)
@@ -57,17 +58,19 @@ HEX                         ([0-9a-fA-F])
 OCT                         ([0-7])
 IP_OCTET                    ([0-9]|[1-9][0-9]|1[0-9][0-9]|2[0-4][0-9]|25[0-5])
 
-U                           [\x80-\xbf]
-U2                          [\xc2-\xdf]
+U                           [\x80-\x9f\xa1-\xbf]
+UA0                         \xa0
+U2                          [\xc3-\xdf]
+UC2                         \xc2
 U3                          [\xe0-\xee]
 U4                          [\xf0-\xf4]
-CHINESE                     {U2}{U}|{U3}{U}{U}|{U4}{U}{U}{U}
+CHINESE                     {U2}{UA0}|{UC2}{U}|{U2}{U}|{U3}{U}{U}|{U4}{U}{U}{U}
 CN_EN                       {CHINESE}|[a-zA-Z]
 CN_EN_NUM                   {CHINESE}|[_a-zA-Z0-9]
 LABEL                       {CN_EN}{CN_EN_NUM}*
 
 U3_FULL_WIDTH               [\xe0-\xef]
-CHINESE_FULL_WIDTH          {U2}{U}|{U3_FULL_WIDTH}{U}{U}|{U4}{U}{U}{U}
+CHINESE_FULL_WIDTH          {U2}{UA0}|{UC2}{U}|{U2}{U}|{U3_FULL_WIDTH}{U}{U}|{U4}{U}{U}{U}
 CN_EN_FULL_WIDTH            {CHINESE_FULL_WIDTH}|[a-zA-Z]
 CN_EN_NUM_FULL_WIDTH        {CHINESE_FULL_WIDTH}|[_a-zA-Z0-9 ]
 LABEL_FULL_WIDTH            {CN_EN_FULL_WIDTH}{CN_EN_NUM_FULL_WIDTH}*

src/parser/test/ParserTest.cpp

@@ -3350,4 +3350,18 @@ TEST_F(ParserTest, TestShowSentenceWithPipe) {
     ASSERT_TRUE(result.ok()) << result.status();
   }
 }
+
+TEST_F(ParserTest, TestSpecialWhiteSpaceChar) {
+  {
+    std::string query = "SHOW\xC2\xA0SPACES";
+    auto result = parse(query);
+    ASSERT_TRUE(result.ok()) << result.status();
+  }
+  {
+    std::string query = "SHOW \xC2\xA0SPACES\xC2\xA0";
+    auto result = parse(query);
+    ASSERT_TRUE(result.ok()) << result.status();
+  }
+}
+
 }  // namespace nebula

src/parser/test/ScannerTest.cpp

@@ -533,7 +533,7 @@ TEST(Scanner, Basic) {
       CHECK_SEMANTIC_VALUE("label", TokenType::LABEL, "label"),
       CHECK_SEMANTIC_VALUE("label123", TokenType::LABEL, "label123"),
       // \xA0 is white space in UTF-8 too
-      CHECK_SEMANTIC_VALUE("\xA0"
+      CHECK_SEMANTIC_VALUE("\xC2\xA0"
                            "abc",
                            TokenType::LABEL,
                            "abc"),

tests/tck/features/basic/Parser.feature

@@ -0,0 +1,16 @@
+# Copyright (c) 2022 vesoft inc. All rights reserved.
+#
+# This source code is licensed under Apache 2.0 License.
+Feature: Parser
+
+  Scenario: Test special white space character
+    When executing query:
+      """
+      SHOW  SPACES
+      """
+    Then the execution should be successful
+    When executing query:
+      """
+      RETURN  1
+      """
+    Then the execution should be successful