inital_retrieve.class

// this example request for metadata retrieve and then post to the heroku service

MetadataService.MetadataPort service = MetadataValidator.createService();             
MetadataService.RetrieveRequest retrieveRequest = new MetadataService.RetrieveRequest();
retrieveRequest.apiVersion    = 27;
retrieveRequest.packageNames  = null;
retrieveRequest.singlePackage = true;
retrieveRequest.specificFiles = null;

retrieveRequest.unpackaged    = new MetadataService.Package_x();
retrieveRequest.unpackaged.types = new List<MetadataService.PackageTypeMembers>();
MetadataService.PackageTypeMembers packageType = new MetadataService.PackageTypeMembers();
packageType.name = 'ApexClass'; 
packageType.members = new String[] { 'MetadataValidator' };

retrieveRequest.unpackaged.types.add(packageType);

MetadataService.AsyncResult aResult = service.retrieve(retrieveRequest);
MetadataService.AsyncResult[] aResults = service.checkStatus(new String[] { aResult.Id });

// only the first package
while(!aResults[0].done) {
    aResults = service.checkStatus(new String[] { aResult.Id });
}

MetadataService.RetrieveResult retrieveResult = service.checkRetrieveStatus(aResults[0].Id);

// messages must be null for successful retrieve
if(retrieveResult.messages == null) {
    // post the result
    String resp = MetadataRetrieveHelper.retrievePost(aResult.Id, retrieveResult.zipFile);

    // response is a list of file paths in the zip file, seperated by comma
    System.debug('resp:'+ resp);
    System.debug('process id:'+ aResult.Id);
}

subsequent_retrieve.cass

// process is the asynresult id when the initial retreive request was made
// path is the path of the specific inside the zip file, if path is empty then all file paths are returned

MetadataRetrieveHelper.retrieveGet(process, '');
MetadataRetrieveHelper.retrieveGet(process, path);