Fix GHSA-w8qr-v226-r27w

We should not early-out with success status if we found an ipv6
hostname, we should keep checking the rest of the conditions.
Because integrating the if-check of the ipv6 hostname in the
"Validate domain" if-check made the code hard to read, I extracted the
condition out to a separate function. This also required to make
a few pointers const in order to have some clean code.

(cherry picked from commit 4066610)

Author: nielsdos

ext/filter/logical_filters.c

@@ -91,7 +91,7 @@
 #define FORMAT_IPV4    4
 #define FORMAT_IPV6    6
 
-static int _php_filter_validate_ipv6(char *str, size_t str_len, int ip[8]);
+static int _php_filter_validate_ipv6(const char *str, size_t str_len, int ip[8]);
 
 static int php_filter_parse_int(const char *str, size_t str_len, zend_long *ret) { /* {{{ */
 	zend_long ctx_value;
@@ -571,6 +571,14 @@ static int is_userinfo_valid(zend_string *str)
 	return 1;
 }
 
+static zend_bool php_filter_is_valid_ipv6_hostname(const char *s, size_t l)
+{
+	const char *e = s + l;
+	const char *t = e - 1;
+
+	return *s == '[' && *t == ']' && _php_filter_validate_ipv6(s + 1, l - 2, NULL);
+}
+
 void php_filter_validate_url(PHP_INPUT_FILTER_PARAM_DECL) /* {{{ */
 {
 	php_url *url;
@@ -596,7 +604,7 @@ void php_filter_validate_url(PHP_INPUT_FILTER_PARAM_DECL) /* {{{ */
 
 	if (url->scheme != NULL &&
 		(zend_string_equals_literal_ci(url->scheme, "http") || zend_string_equals_literal_ci(url->scheme, "https"))) {
-		char *e, *s, *t;
+		const char *s;
 		size_t l;
 
 		if (url->host == NULL) {
@@ -605,17 +613,14 @@ void php_filter_validate_url(PHP_INPUT_FILTER_PARAM_DECL) /* {{{ */
 
 		s = ZSTR_VAL(url->host);
 		l = ZSTR_LEN(url->host);
-		e = s + l;
-		t = e - 1;
-
-		/* An IPv6 enclosed by square brackets is a valid hostname */
-		if (*s == '[' && *t == ']' && _php_filter_validate_ipv6((s + 1), l - 2, NULL)) {
-			php_url_free(url);
-			return;
-		}
 
-		// Validate domain
-		if (!_php_filter_validate_domain(ZSTR_VAL(url->host), l, FILTER_FLAG_HOSTNAME)) {
+		if (
+			/* An IPv6 enclosed by square brackets is a valid hostname.*/
+			!php_filter_is_valid_ipv6_hostname(s, l) &&
+			/* Validate domain.
+			 * This includes a loose check for an IPv4 address. */
+			!_php_filter_validate_domain(ZSTR_VAL(url->host), l, FILTER_FLAG_HOSTNAME)
+		) {
 			php_url_free(url);
 			RETURN_VALIDATION_FAILED
 		}
@@ -749,15 +754,15 @@ static int _php_filter_validate_ipv4(char *str, size_t str_len, int *ip) /* {{{
 }
 /* }}} */
 
-static int _php_filter_validate_ipv6(char *str, size_t str_len, int ip[8]) /* {{{ */
+static int _php_filter_validate_ipv6(const char *str, size_t str_len, int ip[8]) /* {{{ */
 {
 	int compressed_pos = -1;
 	int blocks = 0;
 	int num, n, i;
 	char *ipv4;
-	char *end;
+	const char *end;
 	int ip4elm[4];
-	char *s = str;
+	const char *s = str;
 
 	if (!memchr(str, ':', str_len)) {
 		return 0;

ext/filter/tests/ghsa-w8qr-v226-r27w.phpt

@@ -0,0 +1,41 @@
+--TEST--
+GHSA-w8qr-v226-r27w
+--EXTENSIONS--
+filter
+--FILE--
+<?php
+
+function test(string $input) {
+    var_dump(filter_var($input, FILTER_VALIDATE_URL));
+}
+
+echo "--- These ones should fail ---\n";
+test("http://t[[email protected]");
+test("http://t[est@[::1]");
+test("http://t[est@[::1");
+test("http://t[est@::1]");
+test("http://php.net\\@aliyun.com/aaa.do");
+test("http://test[@2001:db8:3333:4444:5555:6666:1.2.3.4]");
+test("http://te[st@2001:db8:3333:4444:5555:6666:1.2.3.4]");
+test("http://te[st@2001:db8:3333:4444:5555:6666:1.2.3.4");
+
+echo "--- These ones should work ---\n";
+test("http://[email protected]");
+test("http://test@[2001:db8:3333:4444:5555:6666:1.2.3.4]");
+test("http://test@[::1]");
+
+?>
+--EXPECT--
+--- These ones should fail ---
+bool(false)
+bool(false)
+bool(false)
+bool(false)
+bool(false)
+bool(false)
+bool(false)
+bool(false)
+--- These ones should work ---
+string(21) "http://[email protected]"
+string(50) "http://test@[2001:db8:3333:4444:5555:6666:1.2.3.4]"
+string(17) "http://test@[::1]"