Refactor psqlplus_downcase_identifier to include encoding parameter for better handling of multi-byte characters

Author: NotHimmel

src/bin/psql/psqlplusscan.l

@@ -56,7 +56,7 @@ static const int	NumKeywords = lengthof(psqlplus_keywords);
 
 static const PsqlScanKeyword *PsqlplusKeywordLookup(const char *text);
 static int process_integer_literal(const char *token, YYSTYPE *lval);
-static char *psqlplus_downcase_identifier(const char *str, int len);
+static char *psqlplus_downcase_identifier(const char *str, int len, int encoding);
 static void startlit(yyscan_t yyscanner);
 static void addlitchar(unsigned char ychar, yyscan_t yyscanner);
 static void addlit(char *ytext, int yleng, yyscan_t yyscanner);
@@ -358,7 +358,8 @@ xlinside		[^ ]+
 						return keyword->value;
 					}
 
-					ident = psqlplus_downcase_identifier(yytext, yyleng);
+					ident = psqlplus_downcase_identifier(yytext, yyleng,
+														 cur_state->encoding);
 					yylval->str = ident;
 					cur_state->token_count++;
 					return IDENT;
@@ -454,18 +455,42 @@ psqlplus_scanner_finish(yyscan_t yyscanner)
  * Do downcasing and returns a palloc'd string.
  */
 static char *
-psqlplus_downcase_identifier(const char *str, int len)
+psqlplus_downcase_identifier(const char *str, int len, int encoding)
 {
 	char	   *result;
-	int			i;
+	int			i = 0;
+	int			out = 0;
+	bool		enc_is_single_byte = (pg_encoding_max_length(encoding) == 1);
 
 	result = (char *) pg_malloc(len + 1);
-	for (i = 0; i < len; i++)
+
+	while (i < len)
 	{
-		result[i] = (char) pg_tolower((unsigned char) str[i]);
+		unsigned char ch = (unsigned char) str[i];
+
+		if (!enc_is_single_byte && IS_HIGHBIT_SET(ch))
+		{
+			int			mblen = pg_encoding_mblen(encoding, str + i);
+
+			if (mblen <= 0 || i + mblen > len)
+				mblen = 1;
+			memcpy(result + out, str + i, mblen);
+			out += mblen;
+			i += mblen;
+		}
+		else
+		{
+			if (ch >= 'A' && ch <= 'Z')
+				ch += 'a' - 'A';
+			else if (enc_is_single_byte && IS_HIGHBIT_SET(ch) && isupper(ch))
+				ch = tolower(ch);
+
+			result[out++] = (char) ch;
+			i++;
+		}
 	}
 
-	result[i] = '\0';
+	result[out] = '\0';
 
 	return result;
 }