RSS to S3 Ruby Lambda Function

rss-to-s3.rb

require 'json'
require 'aws-sdk-s3'
require 'open-uri'

# Ideally put these in environment variables
# but since this is just for us, who cares.
BUCKET = "RETRACTED"
REGION = "RETRACTED"

def do_newsletters
  # We want to consider any failure to be total failure
  # BUT still do the ones we can.. so we can't just crash
  # out on the first exception, hence a flag to keep track
  # of the current success state.
  success = true

  # Grab the latest newsletter info
  newsletters = JSON.parse(URI.open('RETRACTED').read.to_s)

  # For each newsletter..
  newsletters.each do |k, nl|
    output_filename = nl['domain'] + ".xml"
    feed_url = "https://" + nl['domain'] + "/drss"
    puts feed_url

    begin
      # Grab the RSS feed
      xml = URI.open(feed_url).read

      # Upload to S3
      s3 = Aws::S3::Client.new(region: REGION)
      s3.put_object(bucket: BUCKET, key: output_filename, body: xml, content_type: 'application/xml;charset=utf-8', cache_control: "max-age=14400")

      # Make file on S3 publicly readable
      s3.put_object_acl({ acl: "public-read", bucket: BUCKET, key: output_filename })
      puts "  Success - https://cprss.s3.amazonaws.com/#{output_filename}"
    rescue
      # Any failure, duck out
      puts "  Failed"
      success = false
    end
  end

  success
end

def lambda_handler(event:, context:)
  if do_newsletters
    { statusCode: 200, body: 'OK' }
  else
    { statusCode: 500, body: 'FAIL' }
  end
end

do_newsletters if __FILE__ == $0