Re: [GNU Crypto] W.: RipeMD160 - Changing the initial value

[Casey Marshall]

On Nov 18, 2005, at 4:23 AM, address@hidden wrote:

> Hello world,
>
> I need some urgent help!
>
> Due to some project purposes I cannot change I have to modify the  
> initial value for the RIPEMD-160 algorithm
> depending on another result from the RIPEMD-160 algorithm.
>
> I am using Java 1.3.1 with gnu crypto 1.1
> (The important class RipeMD160.java has not been changed compared  
> to gnu crypto 2.0.1).
>
> I want to do something like this:
> (let newIV be an result of the RIPEMD-160 - e.g.  
> "9c1185a5c5e9fc54612808977ee8f548b2258d31")
>
> public void setIV(String newIV) {
>         h0 = Integer.parseInt(newIV.substring(0, 8).toUpperCase(), 16);
>         h1 = Integer.parseInt(newIV.substring(8, 16).toUpperCase(), 16);
>         h2 = Integer.parseInt(newIV.substring(16, 24).toUpperCase(), 16);
>         h3 = Integer.parseInt(newIV.substring(24, 32).toUpperCase(), 16);
>         h4 = Integer.parseInt(newIV.substring(32, 40).toUpperCase(), 16);
> }
>
> In other words, I have to split the result in 5 4-byte-words and  
> use them as new initial values for the next calculation.
>
> In the current implementation of the RipeMD160-class, the initial  
> value variables are declared as int.
>
> It is possible that result.substring(0, 8) would be  
> "FFFFFFFF" (greater than 7FFFFFFF - max value for JAVA int)
> for instance? In this case, this value would be greater than the  
> max value of an JAVA int and I will get a problem!

The way to do that is to parse the inputs as longs, and cast them to  
ints:

  h0 = (int) Long.parseLong (newIV.substring (0, 8), 16);
  etc.

I'm fairly certain you don't need to call 'toUpperCase.' All you are  
doing there is creating yet another 8-character string unnecessarily.

But I'm curious why your 'newIV' value has to be a String. If you  
have just the byte array IV, converting that into five ints is a lot  
more efficient than splitting a string and parsing hex digits.

Hope this helps.