]> granicus.if.org Git - php/commitdiff
fix #43941: invalid utf-8 not accepted
authorStanislav Malyshev <stas@php.net>
Wed, 30 Jan 2008 03:17:57 +0000 (03:17 +0000)
committerStanislav Malyshev <stas@php.net>
Wed, 30 Jan 2008 03:17:57 +0000 (03:17 +0000)
ext/json/json.c
ext/json/tests/bug43941.phpt [new file with mode: 0644]
ext/json/utf8_to_utf16.c

index ddfbbe5d26bf6229cb7a8a077eddd2308516b88d..90e5955c09487d91a0783c7b21d499e9c717466d 100644 (file)
@@ -269,8 +269,14 @@ static void json_escape_string(smart_str *buf, char *s, int len, int options) /*
         {
             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 (file)
index 0000000..0f86d1d
--- /dev/null
@@ -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
+
index bc2d6f36d69132a9edae35cc43591b903e3391af..42ea9e5d8eb86fd59c278c25ae550708ec65f61d 100644 (file)
@@ -40,7 +40,7 @@ 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;