mb_scrub does not attempt to scrub known-valid UTF-8 strings

php · Jan 22, 2023 · 6f53dbb · 6f53dbb
1 parent f4dd35e
commit 6f53dbb
Show file tree

Hide file tree

Showing 2 changed files with 16 additions and 5 deletions.
diff --git a/ext/mbstring/mbstring.c b/ext/mbstring/mbstring.c
@@ -5066,12 +5066,10 @@ PHP_FUNCTION(mb_chr)
 /* {{{ */
 PHP_FUNCTION(mb_scrub)
 {
-	char* str;
-	size_t str_len;
-	zend_string *enc_name = NULL;
+	zend_string *str, *enc_name = NULL;
 
 	ZEND_PARSE_PARAMETERS_START(1, 2)
-		Z_PARAM_STRING(str, str_len)
+		Z_PARAM_STR(str)
 		Z_PARAM_OPTIONAL
 		Z_PARAM_STR_OR_NULL(enc_name)
 	ZEND_PARSE_PARAMETERS_END();
@@ -5081,7 +5079,12 @@ PHP_FUNCTION(mb_scrub)
 		RETURN_THROWS();
 	}
 
-	RETURN_STR(php_mb_convert_encoding_ex(str, str_len, enc, enc));
+	if (enc == &mbfl_encoding_utf8 && (GC_FLAGS(str) & IS_STR_VALID_UTF8)) {
+		/* A valid UTF-8 string will not be changed by mb_scrub; so just increment the refcount and return it */
+		RETURN_STR_COPY(str);
+	}
+
+	RETURN_STR(php_mb_convert_encoding_ex(ZSTR_VAL(str), ZSTR_LEN(str), enc, enc));
 }
 /* }}} */
 

diff --git a/ext/mbstring/tests/mb_scrub.phpt b/ext/mbstring/tests/mb_scrub.phpt
@@ -8,7 +8,15 @@ var_dump(
     "?" === mb_scrub("\x80"),
     "?" === mb_scrub("\x80", 'UTF-8')
 );
+
+$utf8str = "abc 日本語 Οὐχὶ ταὐτὰ παρίσταταί μοι γιγνώσκειν ⡍⠔⠙⠖ ⡊ ⠙⠕⠝⠰⠞";
+// Check $utf8str so it is marked as 'valid UTF-8'
+// This will enable optimized implementation of mb_scrub
+if (!mb_check_encoding($utf8str, 'UTF-8'))
+    die("Test string should be valid UTF-8");
+var_dump(mb_scrub($utf8str));
 ?>
 --EXPECT--
 bool(true)
 bool(true)
+string(122) "abc 日本語 Οὐχὶ ταὐτὰ παρίσταταί μοι γιγνώσκειν ⡍⠔⠙⠖ ⡊ ⠙⠕⠝⠰⠞"