[CBRD-25302] Strengthens checks on date_lang_string_literal in functi…

…ons such as TO_DATE() (#5632)

http://jira.cubrid.org/browse/CBRD-25302

* Strengthens checks on date_lang_string_literal in functions such as TO_DATE()
* Functions that take 'date_lang_string_literal' as an argument
   - TO_CHAR
   - TO_DATE, TO_DATETIME, TO_DATETIME_TZ
   - TO_TIME, TO_TIMESTAMP, TO_TIMESTAMP_TZ

NOTE: this commit is erased when fixing conflict during merging with develop

msg/en_US.utf8/cubrid.msg

@@ -1951,6 +1951,7 @@ $set 8 MSGCAT_SET_PARSER_SEMANTIC
 322 Table column '%1$s.%2$s' has not been defined
 323 Stored procedure/function '%1$s' has OUT or IN OUT arguments
 324 '%1$s' is not a record variable.
+325 The argument specifying the language must be a string literal.
 
 325 The argument specifying the language must be a string literal.
 326 Stored procedure/function "%1$s" does not exist.

msg/ko_KR.utf8/cubrid.msg

@@ -1950,6 +1950,7 @@ $set 8 MSGCAT_SET_PARSER_SEMANTIC
 322 테이블 컬럼 '%1$s.%2$s'이 정의되지 않았습니다
 323 저장 프로시저/함수 '%1$s' 이(가) OUT 또는 IN OUT 인수를 가지고 있습니다.
 324 '%1$s'은 레코드 변수가 아닙니다.
+325 언어를 지정하는 인수는 문자열 리터럴이어야 합니다.
 
 325 언어를 지정하는 인수는 문자열 리터럴이어야 합니다.
 326 저장 프로시저/함수 "%1$s"이(가) 존재하지 않습니다.

src/base/language_support.c

@@ -2311,36 +2311,6 @@ lang_set_flag_from_lang (const char *lang_str, bool has_user_format, bool has_us
       status = lang_get_lang_id_from_name (lang_str, &lang);
     }
 
-  if (lang_set_flag_from_lang_id (lang, has_user_format, has_user_lang, flag) == 0)
-    {
-      return status;
-    }
-
-  assert (lang == INTL_LANG_ENGLISH);
-
-  return 1;
-}
-
-/*
- * lang_set_flag_from_lang - set a flag according to language identifier
- *
- *   return: 0 if language string OK and flag was set, non-zero otherwise
- *   lang(in): language identier
- *   has_user_format(in): true if user has given a format, false otherwise
- *   has_user_lang(in): true if user has given a language, false otherwise
- *   flag(out): bit flag : bits 0 and 1 are user flags, bits 2 - 31 are for
- *		language identification
- *		Bit 0 : if set, the format was given by user
-*		Bit 1 : if set, the language was given by user
- *		Bit 2 - 31 : INTL_LANG
- *		Consider change this flag to store the language as value
- *		instead of as bit map
- *
- *  Note : function is used in context of some date-string functions.
- */
-int
-lang_set_flag_from_lang_id (const INTL_LANG lang, bool has_user_format, bool has_user_lang, int *flag)
-{
   int lang_val = (int) lang;
 
   *flag = 0;
@@ -2351,13 +2321,15 @@ lang_set_flag_from_lang_id (const INTL_LANG lang, bool has_user_format, bool has
   if (lang_val >= lang_Count_locales)
     {
       lang_val = (int) INTL_LANG_ENGLISH;
-      *flag |= lang_val << 2;
-      return 1;
+      status = 1;
     }
 
-  *flag |= lang_val << 2;
+  *flag |= (lang_val << 2);
 
-  return 0;
+  assert (((*flag) & LANG_LOADED_LOCALES_PARITY_MASK) == 0);
+  *flag |= LANG_LOADED_LOCALES_PARITY;
+
+  return status;
 }
 
 /*
@@ -2378,7 +2350,8 @@ lang_get_lang_id_from_flag (const int flag, bool * has_user_format, bool * has_u
   *has_user_format = ((flag & 0x1) == 0x1) ? true : false;
   *has_user_lang = ((flag & 0x2) == 0x2) ? true : false;
 
-  lang_val = flag >> 2;
+  assert ((flag & LANG_LOADED_LOCALES_PARITY_MASK) == LANG_LOADED_LOCALES_PARITY);
+  lang_val = (flag & ~LANG_LOADED_LOCALES_PARITY_MASK) >> 2;
 
   if (lang_val >= 0 && lang_val < lang_Count_locales)
     {

src/base/language_support.h

@@ -51,6 +51,8 @@
 #define LANG_MAX_COLLATIONS  256
 #define LANG_MAX_BUILTIN_COLLATIONS  32
 #define LANG_MAX_LOADED_LOCALES  32
+#define LANG_LOADED_LOCALES_PARITY_MASK    0xFFFFF000
+#define LANG_LOADED_LOCALES_PARITY         0x75A5C000
 
 #define LANG_COERCIBLE_COLL LANG_SYS_COLLATION
 #define LANG_COERCIBLE_CODESET LANG_SYS_CODESET
@@ -300,7 +302,6 @@ extern "C"
   extern int lang_get_lang_id_from_name (const char *lang_name, INTL_LANG * lang_id);
   extern const char *lang_get_lang_name_from_id (const INTL_LANG lang_id);
   extern int lang_set_flag_from_lang (const char *lang_str, bool has_user_format, bool has_user_lang, int *flag);
-  extern int lang_set_flag_from_lang_id (const INTL_LANG lang, bool has_user_format, bool has_user_lang, int *flag);
   extern INTL_LANG lang_get_lang_id_from_flag (const int flag, bool * has_user_format, bool * has_user_lang);
   extern const char *lang_date_format_parse (const INTL_LANG lang_id, const INTL_CODESET codeset, const DB_TYPE type,
 					     INTL_CODESET * format_codeset);

src/parser/csql_grammar.y

@@ -26393,32 +26393,51 @@ parser_make_date_lang (int arg_cnt, PT_NODE * arg3)
   if (arg3 && arg_cnt == 3)
     {
       char *lang_str;
-      if ((arg3->type_enum == PT_TYPE_CHAR || arg3->type_enum == PT_TYPE_NCHAR) && arg3->info.value.data_value.str != NULL)
+      if (arg3->node_type == PT_VALUE)
 	{
-	  date_lang = parser_new_node (this_parser, PT_VALUE);
-          if (!date_lang)
+	  if ((arg3->type_enum == PT_TYPE_CHAR || arg3->type_enum == PT_TYPE_NCHAR)
+	      && arg3->info.value.data_value.str != NULL)
 	    {
-	      er_set (ER_ERROR_SEVERITY, ARG_FILE_LINE, ER_OUT_OF_VIRTUAL_MEMORY, 1, sizeof (PT_NODE));
-	      return NULL;
-	    }
-	  int flag = 0;
-	  lang_str = (char *) arg3->info.value.data_value.str->bytes;
-	  if (lang_set_flag_from_lang (lang_str, 1, 1, &flag))
-	    {
-	      PT_ERROR (this_parser, arg3, "check syntax at 'date_lang'");
+	      date_lang = parser_new_node (this_parser, PT_VALUE);
+	      if (!date_lang)
+		{
+		  er_set (ER_ERROR_SEVERITY, ARG_FILE_LINE, ER_OUT_OF_VIRTUAL_MEMORY, 1, sizeof (PT_NODE));
+		  return NULL;
+		}
+	      int flag = 0;
+	      lang_str = (char *) arg3->info.value.data_value.str->bytes;
+	      if (lang_set_flag_from_lang (lang_str, 1, 1, &flag))
+		{
+		  //PT_ERROR (this_parser, arg3, "check syntax at 'date_lang'");
+                  PT_ERRORmf (this_parser, arg3, MSGCAT_SET_ERROR, -(ER_LOCALE_LANG_NOT_AVAILABLE), lang_str);
+		}
+	      date_lang->info.value.data_value.i = (long) flag;
+	      date_lang->type_enum = PT_TYPE_INTEGER;
+	      parser_free_node (this_parser, arg3);
 	    }
-	  date_lang->info.value.data_value.i = (long) flag;
 	}
-
-      if (date_lang)
+      else if (arg3->node_type == PT_HOST_VAR)
 	{
-	  date_lang->type_enum = PT_TYPE_INTEGER;
-          parser_free_node (this_parser, arg3);
-	}
-      else
+	  date_lang = arg3;
+	} 
+      else if(this_parser->flag.is_parsing_static_sql)
         {
-          date_lang = arg3;
+           if (arg3->node_type == PT_EXPR && arg3->info.expr.op == PT_CAST)
+             {
+                PT_NODE * node = arg3->info.expr.cast_type; 
+                if(node->node_type == PT_DATA_TYPE && PT_IS_SIMPLE_CHAR_STRING_TYPE(node->type_enum))
+                  {  
+                    date_lang = arg3;
+                  }
+             }
         }
+
+      if (date_lang == NULL)
+	{ 	  
+          PT_ERRORm (this_parser, arg3, MSGCAT_SET_PARSER_SEMANTIC, MSGCAT_SEMANTIC_NOT_SUPPORT_TYPE_TO_DATE_LANG);
+
+	  date_lang = arg3;
+	}
     }
   else
     {
@@ -26442,7 +26461,7 @@ parser_make_date_lang (int arg_cnt, PT_NODE * arg3)
 	}
 
     }
-    return date_lang;
+  return date_lang;
 }
 
 static PT_NODE *

src/parser/parse_tree_cl.c

@@ -10573,23 +10573,63 @@ pt_print_expr (PARSER_CONTEXT * parser, PT_NODE * p)
 	r1 = pt_print_bytes (parser, p->info.expr.arg1);
 	q = pt_append_varchar (parser, q, r1);
 
-	flags = p->info.expr.arg3->info.value.data_value.i;
-	lang_id = lang_get_lang_id_from_flag (flags, &has_user_format, &has_user_lang);
-	if (has_user_format)
+	assert (p->info.expr.arg3);
+	if (p->info.expr.arg3->node_type == PT_HOST_VAR)
 	  {
-	    const char *lang_name = lang_get_lang_name_from_id (lang_id);
-
 	    q = pt_append_nulstring (parser, q, ", ");
 	    r1 = pt_print_bytes (parser, p->info.expr.arg2);
 	    q = pt_append_varchar (parser, q, r1);
 
-	    if (lang_name != NULL && has_user_lang)
+	    q = pt_append_nulstring (parser, q, ", ");
+	    r1 = pt_print_bytes (parser, p->info.expr.arg3);
+	    q = pt_append_varchar (parser, q, r1);
+	  }
+	else if (p->info.expr.arg3->node_type == PT_VALUE)
+	  {
+	    flags = p->info.expr.arg3->info.value.data_value.i;
+	    lang_id = lang_get_lang_id_from_flag (flags, &has_user_format, &has_user_lang);
+	    if (has_user_format)
 	      {
-		q = pt_append_nulstring (parser, q, ", '");
-		q = pt_append_nulstring (parser, q, lang_name);
-		q = pt_append_nulstring (parser, q, "'");
+		const char *lang_name = lang_get_lang_name_from_id (lang_id);
+
+		q = pt_append_nulstring (parser, q, ", ");
+		r1 = pt_print_bytes (parser, p->info.expr.arg2);
+		q = pt_append_varchar (parser, q, r1);
+
+		if (lang_name != NULL && has_user_lang)
+		  {
+		    q = pt_append_nulstring (parser, q, ", '");
+		    q = pt_append_nulstring (parser, q, lang_name);
+		    q = pt_append_nulstring (parser, q, "'");
+		  }
 	      }
 	  }
+	else if (parser->flag.is_parsing_static_sql)
+	  {
+	    assert (p->info.expr.arg3->node_type == PT_EXPR && p->info.expr.arg3->info.expr.op == PT_CAST);
+	    assert (p->info.expr.arg3->info.expr.cast_type->node_type == PT_DATA_TYPE);
+	    assert (PT_IS_SIMPLE_CHAR_STRING_TYPE (p->info.expr.arg3->info.expr.cast_type->type_enum));
+	    q = pt_append_nulstring (parser, q, ", ");
+	    r1 = pt_print_bytes (parser, p->info.expr.arg2);
+	    q = pt_append_varchar (parser, q, r1);
+
+	    q = pt_append_nulstring (parser, q, ", ");
+	    r1 = pt_print_bytes (parser, p->info.expr.arg3);
+	    q = pt_append_varchar (parser, q, r1);
+	  }
+	else
+	  {
+	    /*    
+	       create table foo(a char(20), b varchar, c nchar(20), d nchar varying, e sequence(int));
+	       insert into foo values('aaa', 'bbb', n'ccc', n'ddd', {1, 2, 3, 4, 5});
+	       select to_char(e) from foo order by 1;
+
+	       --In a case like "select to_char(e) from foo", it enters the second step.
+	     */
+	    assert ((pt_has_error (parser) || er_errid () != NO_ERROR)
+		    || (p->info.expr.arg2 &&
+			p->info.expr.arg2->node_type == PT_VALUE && p->info.expr.arg2->type_enum == PT_TYPE_NULL));
+	  }
 	q = pt_append_nulstring (parser, q, ")");
       }
       break;

src/parser/parser_message.h

@@ -502,6 +502,7 @@
 #define MSGCAT_SEMANTIC_UNDEFINED_TABLE_COLUMN   		MSGCAT_SEMANTIC_NO(322)
 #define MSGCAT_SEMANTIC_SP_OUT_ARGS_EXISTS_IN_QUERY             MSGCAT_SEMANTIC_NO(323)
 #define MSGCAT_SEMANTIC_SP_INTO_FIELD_EXPR_IN_NON_STATIC_SQL    MSGCAT_SEMANTIC_NO(324)
+#define MSGCAT_SEMANTIC_NOT_SUPPORT_TYPE_TO_DATE_LANG    MSGCAT_SEMANTIC_NO(325)
 
 #define MSGCAT_SEMANTIC_NOT_SUPPORT_TYPE_TO_DATE_LANG    MSGCAT_SEMANTIC_NO(325)
 #define MSTCAT_SEMANTIC_SP_NOT_EXIST                            MSGCAT_SEMANTIC_NO(326)