From 20229b7331ee14ceac964a423a56c73cc8f85a9b Mon Sep 17 00:00:00 2001
From: heroicNeZha <25311962+heroicNeZha@users.noreply.github.com>
Date: Thu, 9 Dec 2021 10:17:26 +0800
Subject: [PATCH] Schema support mixed Chinese and English

---
 src/parser/parser.yy                     |  4 ++--
 src/parser/scanner.lex                   |  9 ++++++---
 tests/tck/features/schema/Schema.feature | 13 +++++++++++++
 3 files changed, 21 insertions(+), 5 deletions(-)

diff --git a/src/parser/parser.yy b/src/parser/parser.yy
index 0ea8ede2672..bcd876c2425 100644
--- a/src/parser/parser.yy
+++ b/src/parser/parser.yy
@@ -213,7 +213,7 @@ static constexpr size_t kCommentLengthLimit = 256;
 %token <boolval> BOOL
 %token <intval> INTEGER
 %token <doubleval> DOUBLE
-%token <strval> STRING VARIABLE LABEL IPV4 CHINESE_LABEL
+%token <strval> STRING VARIABLE LABEL IPV4 UTF8_LABEL
 
 %type <strval> name_label unreserved_keyword predicate_name
 %type <expr> expression
@@ -406,7 +406,7 @@ static constexpr size_t kCommentLengthLimit = 256;
 
 name_label
     : LABEL { $$ = $1; }
-    | CHINESE_LABEL { $$ = $1; }
+    | UTF8_LABEL { $$ = $1; }
     | unreserved_keyword { $$ = $1; }
     ;
 
diff --git a/src/parser/scanner.lex b/src/parser/scanner.lex
index 4fe773332d2..07bb0834efb 100644
--- a/src/parser/scanner.lex
+++ b/src/parser/scanner.lex
@@ -47,7 +47,10 @@ U                           [\x80-\xbf]
 U2                          [\xc2-\xdf]
 U3                          [\xe0-\xef]
 U4                          [\xf0-\xf4]
-CHINESE_LABEL               ({U2}{U}|{U3}{U}{U}|{U4}{U}{U}{U})+
+CHINESE                     {U2}{U}|{U3}{U}{U}|{U4}{U}{U}{U}
+CN_EN                       {CHINESE}|[a-zA-Z]
+CN_EN_NUM                   {CHINESE}|[_a-zA-Z0-9]
+UTF8_LABEL                  {CN_EN}{CN_EN_NUM}+
 
 %%
 
@@ -471,7 +474,7 @@ CHINESE_LABEL               ({U2}{U}|{U3}{U}{U}|{U4}{U}{U}{U})+
                                 // Must match /* */
                                 throw GraphParser::syntax_error(*yylloc, "unterminated comment");
                             }
-\`{CHINESE_LABEL}\`         {
+\`{UTF8_LABEL}\`            {
                                 yylval->strval = new std::string(yytext + 1, yyleng - 2);
                                 if (yylval->strval->size() > MAX_STRING) {
                                     auto error = "Out of range of the LABEL length, "
@@ -480,7 +483,7 @@ CHINESE_LABEL               ({U2}{U}|{U3}{U}{U}|{U4}{U}{U}{U})+
                                     delete yylval->strval;
                                     throw GraphParser::syntax_error(*yylloc, error);
                                 }
-                                return TokenType::CHINESE_LABEL;
+                                return TokenType::UTF8_LABEL;
                             }
 .                           {
                                 /**
diff --git a/tests/tck/features/schema/Schema.feature b/tests/tck/features/schema/Schema.feature
index f30f79a6d27..52b65149ebb 100644
--- a/tests/tck/features/schema/Schema.feature
+++ b/tests/tck/features/schema/Schema.feature
@@ -832,6 +832,19 @@ Feature: Insert string vid of vertex and edge
     Then the result should be, in any order:
       | Field  | Type        | Null  | Default | Comment |
       | "时间" | "timestamp" | "YES" | EMPTY   | EMPTY   |
+    When executing query:
+      """
+      CREATE TAG `队伍s2`(`名s字ss1` string);
+      """
+    Then the execution should be successful
+    # desc cn-en mixed tag
+    When executing query:
+      """
+      DESCRIBE TAG `队伍s2`
+      """
+    Then the result should be, in any order:
+      | Field      | Type     | Null  | Default | Comment |
+      | "名s字ss1" | "string" | "YES" | EMPTY   | EMPTY   |
     When executing query:
       """
       DROP SPACE issue2009;