Optimize mb_strtoupper/mb_strtolower for UTF-8 enc and ASCII input

Author: mvorisek

Zend/zend_operators.c

@@ -2952,6 +2952,34 @@ ZEND_API zend_string* ZEND_FASTCALL zend_string_toupper_ex(zend_string *str, boo
 }
 /* }}} */
 
+ZEND_API bool ZEND_FASTCALL zend_str_is_utf8_pure_ascii(const char *str, size_t length) /* {{{ */
+{
+	unsigned char *p = (unsigned char *) str;
+	unsigned char *end = p + length;
+
+#ifdef HAVE_BLOCKCONV
+	__m128i blconv_80 = _mm_set1_epi8(0x80);
+	while (p + BLOCKCONV_STRIDE <= end) {
+	    __m128i blconv_operand = _mm_loadu_si128((__m128i*)(p)); \
+	    __m128i blconv_mingle = _mm_cmpeq_epi8(_mm_max_epu8(blconv_operand, blconv_80), blconv_operand);
+		if (BLOCKCONV_FOUND()) {
+            return false;
+		}
+		p += BLOCKCONV_STRIDE;
+	}
+#endif
+
+	while (p < end) {
+		if (*p >= 0x80) {
+			return false;
+		}
+		p++;
+	}
+
+	return true;
+}
+/* }}} */
+
 ZEND_API int ZEND_FASTCALL zend_binary_strcmp(const char *s1, size_t len1, const char *s2, size_t len2) /* {{{ */
 {
 	int retval;

Zend/zend_operators.h

@@ -449,6 +449,7 @@ ZEND_API char*        ZEND_FASTCALL zend_str_tolower_dup_ex(const char *source,
 ZEND_API char*        ZEND_FASTCALL zend_str_toupper_dup_ex(const char *source, size_t length);
 ZEND_API zend_string* ZEND_FASTCALL zend_string_tolower_ex(zend_string *str, bool persistent);
 ZEND_API zend_string* ZEND_FASTCALL zend_string_toupper_ex(zend_string *str, bool persistent);
+ZEND_API bool         ZEND_FASTCALL zend_str_is_utf8_pure_ascii(const char *str, size_t length);
 
 #define zend_string_tolower(str) zend_string_tolower_ex(str, 0)
 #define zend_string_toupper(str) zend_string_toupper_ex(str, 0)

ext/mbstring/mbstring.c

@@ -2588,22 +2588,29 @@ PHP_FUNCTION(mb_convert_case)
 /* {{{ Returns a upper cased version of source_string */
 PHP_FUNCTION(mb_strtoupper)
 {
+	zend_string *str;
 	zend_string *from_encoding = NULL;
-	char *str;
-	size_t str_len, ret_len;
+	const mbfl_encoding *enc;
+	char *newstr;
+	size_t ret_len;
 
 	ZEND_PARSE_PARAMETERS_START(1, 2)
-		Z_PARAM_STRING(str, str_len)
+		Z_PARAM_STR(str)
 		Z_PARAM_OPTIONAL
 		Z_PARAM_STR_OR_NULL(from_encoding)
 	ZEND_PARSE_PARAMETERS_END();
 
-	const mbfl_encoding *enc = php_mb_get_encoding(from_encoding, 2);
+	enc = php_mb_get_encoding(from_encoding, 2);
 	if (!enc) {
 		RETURN_THROWS();
 	}
 
-	char *newstr = mbstring_convert_case(PHP_UNICODE_CASE_UPPER, str, str_len, &ret_len, enc);
+    // optimize performance for UTF-8 encoding and input string consisting of lower/7-bit ASCII characters only
+    if (enc == &mbfl_encoding_utf8 && zend_str_is_utf8_pure_ascii(ZSTR_VAL(str), ZSTR_LEN(str))) {
+        RETURN_STR(zend_string_toupper(str));
+    }
+
+	newstr = mbstring_convert_case(PHP_UNICODE_CASE_UPPER, ZSTR_VAL(str), ZSTR_LEN(str), &ret_len, enc);
 	/* If newstr is NULL something went wrong in mbfl and this is a bug */
 	ZEND_ASSERT(newstr != NULL);
 
@@ -2616,15 +2623,14 @@ PHP_FUNCTION(mb_strtoupper)
 /* {{{ Returns a lower cased version of source_string */
 PHP_FUNCTION(mb_strtolower)
 {
+	zend_string *str;
 	zend_string *from_encoding = NULL;
-	char *str;
-	size_t str_len;
+	const mbfl_encoding *enc;
 	char *newstr;
 	size_t ret_len;
-	const mbfl_encoding *enc;
 
 	ZEND_PARSE_PARAMETERS_START(1, 2)
-		Z_PARAM_STRING(str, str_len)
+		Z_PARAM_STR(str)
 		Z_PARAM_OPTIONAL
 		Z_PARAM_STR_OR_NULL(from_encoding)
 	ZEND_PARSE_PARAMETERS_END();
@@ -2634,7 +2640,12 @@ PHP_FUNCTION(mb_strtolower)
 		RETURN_THROWS();
 	}
 
-	newstr = mbstring_convert_case(PHP_UNICODE_CASE_LOWER, str, str_len, &ret_len, enc);
+    // optimize performance for UTF-8 encoding and input string consisting of lower/7-bit ASCII characters only
+    if (enc == &mbfl_encoding_utf8 && zend_str_is_utf8_pure_ascii(ZSTR_VAL(str), ZSTR_LEN(str))) {
+        RETURN_STR(zend_string_tolower(str));
+    }
+
+	newstr = mbstring_convert_case(PHP_UNICODE_CASE_LOWER, ZSTR_VAL(str), ZSTR_LEN(str), &ret_len, enc);
 	/* If newstr is NULL something went wrong in mbfl and this is a bug */
 	ZEND_ASSERT(newstr != NULL);