Open a CSV file from S3 using Roo on Heroku - csv

Question

I'm putting together a rake task to grab a CSV file stored on S3 and do some processing on it. I use Roo to process the files. The product is running on Heroku so I do not have local storage for physical files.
The CSV processing I am running works perfectly from the website if I upload the file.
rake task looks like this if I upload just the contents of the file:
desc "Checks uploads bucket on S3 and processes for imports"
task s3_import: [:environment] do
# depends on the environemtn task to load dev/prod env first
# get an s3 connection
s3 = AWS::S3.new(access_key_id: ENV['AWS_ACCESS_KEY_ID'], secret_access_key: ENV['AWS_SECRET_ACCESS_KEY'])
# get the uploads bucket
bucket = s3.buckets[ENV['UPLOAD_BUCKET']]
# check each object in the bucket
bucket.objects.each do |obj|
#uri = obj.url_for(:read)
#puts uri
puts obj.key
file = obj.read
User.import(file)
end
end
I have also tried passing just the url to the object but this does not work either.
Basically, when uploading through the page this works to open the files:
Roo::Csv.new(file.path)
So, I can get the URL of the file, and I can also get the contents of the file into a variable, but I can't work out how to get Roo to open it without a physical file on disk.