I have a strange problem, I pass data from the android String via JNI. pass data from FreePascal wideString successfully using the following code

NewString: function (Env: PJNIEnv; const Unicode: PJChar; Len: JSize):
JString; cdecl;
GetStringChars:function(Env:PJNIEnv;Str:JString;IsCopy:PJBoolean):PJChar;
cdecl;
function JNI_WideStringToJString(Env: PJNIEnv; const WStr: WideString):
JString;
begin
  Result := Env^.NewString(Env, Pointer(WStr), Length(WStr));
end;

but when passing data JString to widestring, using this code :

function JNI_JStringToWideString(Env: PJNIEnv; JStr: JString): WideString;
var
  IsCopy: JBoolean;
  Chars: PJChar;
begin
  if (JStr = nil) then begin
    Result := '';
    Exit;
  end;
  Chars := Env^.GetStringChars(Env, JStr, IsCopy);
  if (Chars = nil) then begin
    Result := ''
  end else begin
    Result := WideString(Chars);
    Env^.ReleaseStringChars(Env, JStr, Chars);
  end;
end;

return only 1 character leading.
if anyone has ever experienced ? or there's something wrong with JNI header ?

thanks

-----
-

PJChar is defined as a pointer to a word. Casting to WideString isn't working since internally Widestring has extra bytes for reference count and string length. These extra bytes are locate before the actual string. This explains why assigning Pointer(WStr) to PJChar works. Try
Result := WideString(PWideChar(Chars));
PWideChar matches probably PJChar and the compiler knows how to convert PWideChar to WideString.

Ludo

thanks for the response.

I am doing what you suggest. but it's still the same.

I debug this line of code using android log

Chars := Env^.GetStringChars(Env, AJString, nil);
__android_log_write(ANDROID_LOG_DEBUG,'proj_debug',PChar(Chars));

and the log I retrieved :

04-26 17:53:17.942: DEBUG/proj_debug(711): e <--- only "e" character

from input AJString =
eNp9zt0KwjAMBeBXCecBrApeCEmG+iRzdrX+dKMrMn16Y0G8M5Cr75ATbub7jR4+T3FIgtViiUa5VS5xGPsfrA2c8lG5o5NghrJX7ikInqCzYCo5pgCKgq3NJ1x1+KuvqjEVH3yuXGuc3b4oX2knAO2tfgM6fNHEdbb2jGv1DUfmNOA=

so I thought, this is not about that. but I do not know why

-----
-

That is normal, because Chars is probably an array of jchar, and a jchar is utf-16 (= widechar). That means that if you treat it as pchar and it only contains plain ASCII, the second "ansichar" will always be #0 on a little endian system (on a big endian system, the first "ansichar" will already be #0).

Jonas

Pchar(Chars) interpretes your widechars as chars. The widechar e is converted to the chars e and #0. The #0 is end of line for a pchar. So your debug output is as expected.

Take a look at
http://groups.google.com/group/android-ndk/browse_thread/thread/e0ab7aefb398 2c45. It talks about the same problem calling GetStringChars from C. The function returns non null terminated utf16 data.

You could also go the hard way, something along the lines of:
  Chars := Env^.GetStringChars(Env, JStr, @IsCopy);
  if (Chars = nil) then begin
    Result := ''
  end else begin
    len:=Env^.GetStringLength(Env, JStr);
    setlength(Result,len);
    move(Chars,Result[1],len*2);
    Env^.ReleaseStringChars(Env, JStr, Chars);
  end;

Yes, there's bug. thanks you.
I using GetStringUTFChars

function JNI_JStringToString(Env: PJNIEnv; JStr: jstring): String;
var
  IsCopy: Pjboolean;
  Chars: PChar;
begin
  if JStr = nil then
  begin
    Result := '';
    Exit;
  end;
  Chars:= Env^.GetStringUTFChars(Env, JStr, IsCopy);
  if Chars = nil then
     Result := ''
  else
    Begin
      Result := String(Chars);
      Env^.ReleaseStringUTFChars(Env, JStr, Chars);
    end;
end;

-----
-