Fix bug #81598: Use C.UTF-8 as LC_CTYPE locale by default

Zend/tests/lc_ctype_inheritance.phpt

@@ -16,8 +16,8 @@ var_dump(setlocale(LC_CTYPE, "de_DE", "de-DE") !== false);
 var_dump(bin2hex(strtoupper("\xe4")));
 var_dump(preg_match('/\w/', "\xe4"));
 ?>
---EXPECT--
-string(1) "C"
+--EXPECTF--
+string(%d) "C%r(\.UTF-8)?%r"
 string(2) "e4"
 int(0)
 bool(true)

Zend/zend_operators.c

@@ -34,8 +34,8 @@
 #include <emmintrin.h>
 #endif
 
-#ifdef ZEND_USE_TOLOWER_L
 #include <locale.h>
+#ifdef ZEND_USE_TOLOWER_L
 static _locale_t current_locale = NULL;
 /* this is true global! may lead to strange effects on ZTS, but so may setlocale() */
 #define zend_tolower(c) _tolower_l(c, current_locale)
@@ -2562,6 +2562,15 @@ ZEND_API void zend_update_current_locale(void) /* {{{ */
 /* }}} */
 #endif
 
+ZEND_API void zend_reset_lc_ctype_locale(void)
+{
+	/* Use the C.UTF-8 locale so that readline can process UTF-8 input, while not interfering
+	 * with single-byte locale-dependent functions used by PHP. */
+	if (!setlocale(LC_CTYPE, "C.UTF-8")) {
+		setlocale(LC_CTYPE, "C");
+	}
+}
+
 static zend_always_inline void zend_str_tolower_impl(char *dest, const char *str, size_t length) /* {{{ */ {
 	unsigned char *p = (unsigned char*)str;
 	unsigned char *q = (unsigned char*)dest;

Zend/zend_operators.h

@@ -514,6 +514,8 @@ ZEND_API void zend_update_current_locale(void);
 #define zend_update_current_locale()
 #endif
 
+ZEND_API void zend_reset_lc_ctype_locale(void);
+
 /* The offset in bytes between the value and type fields of a zval */
 #define ZVAL_OFFSETOF_TYPE	\
 	(offsetof(zval, u1.type_info) - offsetof(zval, value))

ext/snmp/snmp.c

@@ -1975,7 +1975,7 @@ PHP_MINIT_FUNCTION(snmp)
 
 	init_snmp("snmpapp");
 	/* net-snmp corrupts the CTYPE locale during initialization. */
-	setlocale(LC_CTYPE, "C");
+	zend_reset_lc_ctype_locale();
 
 #ifdef NETSNMP_DS_LIB_DONT_PERSIST_STATE
 	/* Prevent update of the snmpapp.conf file */

ext/standard/basic_functions.c

@@ -518,6 +518,7 @@ PHP_RSHUTDOWN_FUNCTION(basic) /* {{{ */
 	 * to the value in startup environment */
 	if (BG(locale_changed)) {
 		setlocale(LC_ALL, "C");
+		zend_reset_lc_ctype_locale();
 		zend_update_current_locale();
 		if (BG(ctype_string)) {
 			zend_string_release_ex(BG(ctype_string), 0);

main/main.c

@@ -2092,6 +2092,7 @@ int php_module_startup(sapi_module_struct *sf, zend_module_entry *additional_mod
 	zuf.getenv_function = sapi_getenv;
 	zuf.resolve_path_function = php_resolve_path_for_zend;
 	zend_startup(&zuf);
+	zend_reset_lc_ctype_locale();
 	zend_update_current_locale();
 
 	zend_observer_startup();