Smalltalk Challenge: Post 4 - Porting the Kember Identity

There are a few things I find myself tripping up over even after spending some time writing "meaningful" Smalltalk code, like using single quotes to delimit strings (double quotes are used for comments) and remembering the order in which different messages are sent, but the more code I write the easier it is to remember such things. After only a few hours, Smalltalk is still something new and unfamiliar.

The first programs I wrote when looking into Go were solutions to the first two Project Euler problems and a port of the Kember Identity search program. I decided to skip the Euler problems this time and go straight to the Kember Identity port.

The Kember program ultimately boils down to generating and checking MD5 hashes. I didn't find any helpful cryptography related objects or methods in the default image, so I searched Google and eventually found Ron Teitelbaum's Cryptography/Team package. Squeak uses a package management system called Monticello to load code into the image, so getting and installing the package was pretty easy. I copied and pasted the package repository's connection information into the Monticello Browser and loaded Rob Withers' contribution, Cryptography-rww.15.mcz.

Once the package was loaded, I was able to obtain hashes with MD5 » hashStream:, but the returned object was a ByteArray and I needed to interpret it as a 32-character hexadecimal string. At first I took this approach to convert the array to a hex string:

Kember » md5: aString
    "return 32-char MD5 hash of the given text" 
    | hash str |
    hash := MD5 new
                hashStream: (ReadStream on: aString).
    str := ''.
    1 to: hash size do: [:i | 
        str := str, ((hash at: i) radix: 16)].
    ↑ str.

… only to find out later that objects of the ByteArray class were modified by the cryptography package to accept a hex message and will do the conversion for me. Oops! All the bit twiddling I had done could easily be replaced with:

    ↑ (MD5 new hashStream: (ReadStream on: aString)) hex.

Converting between hash representations wasn't the only part of the program I initially over-programmed. I was also doing long-form addition to obtain the next hash value in the sequence when all I really needed was a little bit of type juggling and string padding:

Kember » nextHash: aHashStr
    "increment the MD5 hash"
    | hexHash zeroes |
    zeroes = '00000000000000000000000000000000'.
    hexHash := ((ByteArray fromHexString: aHashStr)
        asInteger + 1) asByteArray hex.
    hexHash size < 32
        ifTrue: [↑ (zeroes copyFrom: 1 to 32 - hexHash size)
            , hexHash].
    ↑ hexHash.

I guess it just proves the saying is true, "learning the libraries is the 20% of learning a new language that takes 80% of the time and effort." If anything, at least I can take comfort in knowing I'm not the first person to over-program a solution while learning a new language. For those that want to check out my Kember code, I've set up my own repository and uploaded a Monticello package to SqueakSource, and a file dump to Github (I suggest looking at it in raw mode if you go to Github because their Markdown chokes and truncates the pretty-print view).