frida-java-bridge 12.9.5 version handles strings with embedded nuls not right

if utf8 string contains nuls, the stringFromJni only return string ends with nuls, it truncates string

var blob = [126,83,69,77,73,95,88,77,76,126,0,20,46,109,115,103]; the right utf8 string is "~SEMI_XML~

Jun 08 '20 11:06 centy720

i think the real resolution is comment code g_utf8_validate (data, length, &end) at bindings/gumjs/gumdukmemory.c:

`case GUM_MEMORY_VALUE_UTF8_STRING: { gchar * data; guint8 dummy_to_trap_bad_pointer_early; const gchar * end;

    data = address;
    if (data == NULL)
    {
      duk_push_null (ctx);
      break;
    }

    if (length != 0)
      memcpy (&dummy_to_trap_bad_pointer_early, data, sizeof (guint8));

    if (!g_utf8_validate (data, length, &end))
    {
      _gum_duk_throw (ctx, "can't decode byte 0x%02x in position %u",
          (guint8) *end, (guint) (end - data));
    }

    if (length < 0)
    {
      duk_push_string (ctx, data);
    }
    else
    {
      gchar * slice;

      slice = g_strndup (data, length);
      duk_push_string (ctx, slice);
      g_free (slice);
    }

    break;
  }`

use readUtf8String not readUtf16String

Jun 08 '20 12:06 centy720

I've also encountered this issue. Java strings get truncated if they contain a null byte.

As an example, this code will show param1 as having a length of 14 characters, when in fact it should have 36 characters. The 15th character is a null byte.

SecureStorage.decrypt.overload('java.lang.String', 'byte').implementation = function(param1, param2) {
    console.log("Param1: " + param1);
    console.log("Param1 length: " + param1.length);
}

This issue was also identified in December 2019 issue #1120 in the main frida/frida repo: https://github.com/frida/frida/issues/1120

Is there a work around where we could access the native parameter in memory without too much trouble?

Frida is wonderful, thanks so much for creating it oleavr