From: Josh Cheek <josh.cheek@...>
Date: 2010-10-12T14:56:25+09:00
Subject: Re: Bit Manipulation for a Pure Ruby SHA1 implementation

--001485f1e7cefae0640492652201
Content-Type: text/plain; charset=ISO-8859-1

On Mon, Oct 11, 2010 at 11:35 PM, Jason Larsen <jason.larsen@gmail.com>wrote:

> I'm trying to write a pure Ruby implementation of SHA1. In order to do
> so, I need to be able to directly manipulate bits. I know about the
> bitwise operators in Ruby, but I'm having issues with the way that
> Ruby wraps things up in classes. Basically I've run into two problems.
>
> PROBLEM 1: I need to pad the message by appending a 1-bit, and then a
> certain number of 0s. When appending numbers (e.g. 0b1) to the end of
> by string, Ruby wraps each number in its Fixnum class making it 1 byte
> long: So basically adding 0b1 appends 0b0000_0001, which is 7 zeros
> that shouldn't be there. The fix I've found is I to tack 0x80 on the
> end, which concats 0b1000_0000. The downside is that I have to keep
> track of my zeros in batches of bytes (which at this point my
> implementation handles alright), but I'd like to know how to do it bit
> by bit if possible.
>
> I've seen the String.unpack and Array.pack used for binary data.
> However, I can only get the bits out in one glob str.unpack("B*"), or
> str.unpack("B8"*string.length), which still just gives me a bunch of
> stringified 8bit binary values. Seems like there should be a better
> way.
>
> PROBLEM 2: After adding a one-bit and k zeros, I need to append a 64-
> bit long integer representing the length of the message being hashed.
> It seem like Fixnum is 8 bytes long on my machine, so the question is
> how do I add a 64bit binary representation of this number to the end
> of a string?
>
> Any help would be greatly appreciated. Thanks in advance.
>
>
Hi, if you just need a portable SHA1 implementation, there is already one in
the standard library.
http://ruby-doc.org/stdlib/libdoc/digest/rdoc/index.html

As for your problems, I'm finding myself really confused by your
terminology. For example, when you say "adding a one-bit and k zeros" does
that mean num + 2**k, Or is num <<= 1 ; num += 1 ; num <<= k or something
else even?

It might be more helpful if you show "this is what I have, this is what I
want, this is how I tried to get there"

--001485f1e7cefae0640492652201--