Board index » delphi » Re: Base64 conversion for unicode

2005-07-11 04:52:00 PM
delphi217
Surely a widestring is just a stream of bytes to be encoded, and then
decoded back to a widestring?
/Matthew Jones/
 

Why would it fail? it is supposed to be a byte stream. I am being awkward as
I've not been aware of Unicode issues in email (where I have mainly met
base64). I just can not see why a unicode string is any different to an ANSI
string or a binary file.
/Matthew Jones/
 

The decoding of a unicode widestring (to base64) differs from the encoding
of ASCII string - if you take ASCII encoded string and try to decode it as
unicode, you will fail.
The encoding/decoding is not very complicated, I just don't want to spend
time doing it, so I am looking for a solution someone else made.
"Matthew Jones" <XXXX@XXXXX.COM>writes

On Mon, 11 Jul 2005 17:01:12 +0200, "Adi A" <XXXX@XXXXX.COM>
writes:
encodes a group of 3 BYTES into a group of 4 characters. Thus there is
no special version for an Unicode widestring.
Regards from Germany
Franz-Leo
 

Since I don't have an example of unicode base64 in Delphi, I can only supply
it for C#:
Convert.ToBase64String(Encoding.ASCII.GetBytes(sourcestring))
and
Convert.ToBase64String(Encoding.UNICODE.GetBytes(sourcestring))
Those two lines produce completly different base64 results, both in content
and in lengh (the unicode is approx. twice as long)
My guess is that the difference is in the bytes used: "Hello" in ASCII will
use 5 bytes (therefor 8 characters in base64), while in unicode it will
require 10 bytes (therefor 16 characters in base64).
"Franz-Leo Chomse" <XXXX@XXXXX.COM>writes

See my reply to Franz-Leo Chomse in this thread.
"Matthew Jones" <XXXX@XXXXX.COM>writes

Matthew Jones writes:
an encoding method, to be able to use binary data in text oriented
areas. But I can imagine a Base64Encode function which takes a string
for the input buffer, because Delphi strings handle binary data quite
well.
If the declaration is something like:
function Base64Encode( const input : string ) : string;
it will go wrong with WideString, because it will be converted to
ansistring, before passed to the function.
However, if the declaration is something like:
function Base64Encode( const input; inputsize : Integer ) string;
and you call it with MyOutput := Base64Encode( MyWideString[1], Length(
MyWideString ) * 2 ); it should work.
Diederik
BTW, is this for some kind of predefined communication? If not, why not
use UTF8, instead of Base64? it is much more efficient, and most of the
time human readable.
BTW2, if the encode function uses a string for input, you can solve it
by calling it this way (untested)
function MyWideBase64Encode( const input : WideString ) : string;
var buffer : string;
begin
SetLength( buffer, Length( input ) * 2 );
Move( input[1], buffer[1], Length( input ) * 2 );
result := Base64Encode( buffer );
end;
 

Adi A writes:
What you have to do is to *move* the raw contents of the WideString onto
an AnsiString.
SetLength(lStr,2*Length(lWStr));
Move(Pointer(lWStr)^,Pointer(lStr)^,Length(lStr));
Result := ToBase64String(lStr);
--
Henrick Hellström
www.streamsec.com
 

You nailed the problem, and your function (MyWideBase64Encode) solved it.
Thank you very much
"Diederik van Bochove" <XXXX@XXXXX.COM>writes