Ruby S3 multi-part uploader with resume capability. Usage: ruby mpu.rb [file to upload] [bucket] [key] My knowledge of Ruby isn't great, so I offer my apologies for the bad code.

mpu.rb

#!/usr/bin/env ruby

require 'rubygems'
require 'aws-sdk'

def readPart(file_name, file_size, part_size, part) 
  file_offset = part_size * ( part - 1 )
  bytes_to_read = [ part_size, file_size - file_offset ].min
  part_contents = File.read( file_name, bytes_to_read, file_offset )
  part_md5 = Digest::MD5.hexdigest( part_contents )
  return part_contents, part_md5
end

def uploadPart(part_contents, part_md5, part, upload)
  result_obj = upload.add_part( part_contents, :part_number => part )
  etag_md5sum = result_obj.etag[ 1..-2 ] #  strip off the quotes ... not sure why they decided to include those 
  if etag_md5sum != part_md5
    raise "part_md5 mismatch!"
  end
end

def shouldUploadPart(part, part_md5, upload) 
  begin
    if upload.parts[part].etag[1..-2] == part_md5
      return false
    end
  rescue
    return true
  end
  return true
end

def uploadPartWithRetry(part_contents, part_md5, part, upload, tries) 
  the_exception = nil
  for try in 1..tries
    begin
      uploadPart(part_contents, part_md5, part, upload)
      return
    rescue => e
      the_exception = e
    end
  end
  raise "too many retries, error: " + the_exception.message
end

def uploadParts(file_name, file_size, part_size, number_of_parts, upload)
  skipped_parts = 0
  start_time = Time.now
  for part in 1..number_of_parts
    part_contents, part_md5 = readPart( file_name, file_size, part_size, part )
    if( shouldUploadPart( part, part_md5, upload ) )
      uploadPartWithRetry(part_contents, part_md5, part, upload, 3)
      elapsed_time = Time.now - start_time
      average_part_time = elapsed_time / (part - skipped_parts)
      estimated_time_remaining = (average_part_time * (number_of_parts - part) / 60.0).ceil
      puts "Uploaded #{part}/#{number_of_parts} parts. Estimated time remaining: #{estimated_time_remaining} minutes"
    else
      skipped_parts += 1
      puts "Skipping already uploaded part #{part}/#{number_of_parts}."
    end
  end
  upload.complete(:remote_parts)
end


file_name = ARGV[0]
bucket_name = ARGV[1]
key_name = ARGV[2]


s3 = AWS::S3.new
bucket = s3.buckets[bucket_name]
object = bucket.objects[key_name]

number_of_multipart_uploads = 0
object.multipart_uploads.each do |upload|
  number_of_multipart_uploads += 1 
end

upload = nil
if number_of_multipart_uploads > 1
  raise "multiple uploads in progress"  #  haven't decided what to do here yet
elsif number_of_multipart_uploads == 1
  upload = object.multipart_uploads.first
else
  upload = object.multipart_upload
end

file_size = File.size(file_name)

part_size = [ (file_size / 10000.0).ceil, 5*1024*1024 ].max
number_of_parts = (file_size / part_size.to_f).ceil
puts "Uploading #{file_size} byte file as #{number_of_parts} chunks each up to #{part_size} bytes."
uploadParts(file_name, file_size, part_size, number_of_parts, upload)
puts "Done!"