From: Stanislav Malyshev <stas@php.net>
Date: Wed, 30 Jan 2008 03:21:30 +0000 (+0000)
Subject: fix #43941: invalid utf-8 not accepted
X-Git-Tag: RELEASE_2_0_0a1~701
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=400c9ecf7b616913c732a0aaaf7f1cf7111e8891;p=php

fix #43941: invalid utf-8 not accepted
---

diff --git a/ext/json/json.c b/ext/json/json.c
index 1d91e6b161..2334a3c483 100644
--- a/ext/json/json.c
+++ b/ext/json/json.c
@@ -265,12 +265,17 @@ static void json_escape_string(smart_str *buf, zstr s, int len, zend_uchar type,
 		len = utf8_to_utf16(utf16, s.s, len);
 		if (len <= 0)
 		{
-			if (utf16)
-			{
+			if (utf16) {
 				efree(utf16);
 			}
-
-			smart_str_appendl(buf, "\"\"", 2);
+			if(len < 0) {
+				if(!PG(display_errors)) {
+					php_error_docref(NULL TSRMLS_CC, E_WARNING, "Invalid UTF-8 sequence in argument");
+				}
+				smart_str_appendl(buf, "null", 4);
+			} else {
+				smart_str_appendl(buf, "\"\"", 2);
+			}
 			return;
 		}
 	}
diff --git a/ext/json/tests/bug43941.phpt b/ext/json/tests/bug43941.phpt
new file mode 100644
index 0000000000..0f86d1dfad
--- /dev/null
+++ b/ext/json/tests/bug43941.phpt
@@ -0,0 +1,21 @@
+--TEST--
+Bug #43941 (json_encode() invalid UTF-8)
+--SKIPIF--
+<?php if (!extension_loaded("json")) print "skip"; ?>
+--FILE--
+<?php
+
+var_dump(json_encode("abc"));
+var_dump(json_encode("ab\xE0"));
+var_dump(json_encode("ab\xE0c"));
+var_dump(json_encode(array("ab\xE0", "ab\xE0c", "abc")));
+
+echo "Done\n";
+?>
+--EXPECTF--
+string(5) ""abc""
+string(4) "null"
+string(4) "null"
+string(17) "[null,null,"abc"]"
+Done
+
diff --git a/ext/json/utf8_to_utf16.c b/ext/json/utf8_to_utf16.c
index 4a440a94c1..cb6bbec7ff 100644
--- a/ext/json/utf8_to_utf16.c
+++ b/ext/json/utf8_to_utf16.c
@@ -39,7 +39,7 @@ int utf8_to_utf16(unsigned short w[], char p[], int length)  /* {{{ */
 	for (;;) {
 		c = utf8_decode_next(&utf8);
 		if (c < 0) {
-			return UTF8_END ? the_index : UTF8_ERROR;
+			return (c == UTF8_END) ? the_index : UTF8_ERROR;
 		}
 		if (c < 0x10000) {
 			w[the_index] = (unsigned short)c;