dynamodb-csv-import.rb

#!/usr/bin/env ruby

require 'aws-sdk'
require 'csv'
require 'pp'

BATCH_LIMIT = 25

def do_batch_write(dynamodb, request_items)
  begin
    resp = dynamodb.batch_write_item(request_items: request_items)
    unless resp.unprocessed_items.empty?
      PP.pp resp.unprocessed_items
    end
  rescue Exception => e
    PP.pp request_items
    raise
  end
end

dynamodb = Aws::DynamoDB::Client.new

table_name = ARGV[0]

num_items = 0
total_items = 0
request_items = { table_name => [] }
CSV.parse(File.read(ARGV[1]), headers: true).each do |row|
  request_items[table_name] << {
    put_request: {
      item: row.to_hash.reject{|k,v| v == ''}
    }
  }
  num_items += 1
  total_items += 1
  if num_items == BATCH_LIMIT
    do_batch_write(dynamodb, request_items)
    $stderr.puts "Items written: #{total_items}"
    num_items = 0
    request_items = { table_name => [] }
  end
end

if num_items > 0
  do_batch_write(dynamodb, request_items)
  $stderr.puts "Items written: #{total_items}"
end