Recently I had need of some Groovy code to integrate with Amazon S3. I searched around, but it seemed as if they only code out there was set up for Grails. I couldn't find a simple class that I could drop into our code at work and run with that. Now maybe I didn't search long enough, and maybe the Grails-related stuff would have worked, but it just didn't feel right.

I then ran across this UDF written by ColdFusion developer Barney Boisvert: Amazon S3 URL Builder for ColdFusion

This was perfect. While I had been looking for an S3 "library", all I really needed was a way to generate the URL. I took his CFML and converted it into the following Groovy code. Groovy people - feel free to laugh/comment on how I could improve this:

private String getS3URL(key,secret,bucket,objectkey,expires=900) { def algo = 'HmacSHA1' def expireValue = ((new Date().getTime())/1000+expires).intValue() def stringToSign = 'GET\n\n\n'+expireValue+'\n/'+bucket+'/'+objectkey def signingKey = new javax.crypto.spec.SecretKeySpec(secret.getBytes(),algo) def mymac = Mac.getInstance(algo) mymac.init(signingKey)

def rawSig = mymac.doFinal(stringToSign.getBytes()) def sig = new sun.misc.BASE64Encoder().encode(rawSig);

sig = def destURL = "$bucket/$objectkey?AWSAccessKeyId="+key+"&Signature=$sig&Expires=$expireValue" return destURL }

I think it's interesting to compare both versions. My version got rid of the requestType parameter since we didn't need to worry about that for our code.

You know it's funny - I never used to understand why people didn't use semicolons at the end of their code when it was allowed - but now that I'm getting used to it, it really bugs me when I have to use them.