ext-creators+publishers.rb

#!/usr/bin/env ruby

require "open-uri"
require "csv"
require "nokogiri"

CSV.foreach(ARGV[0], { headers: true, col_sep: "\t" }) do |row|
  iss_url = row["manifestation_identifier"]
  manifestation_id = row["manifestation_id"]
  creators = []
  publishers = []
  xml = nil
  if iss_url and iss_url =~ /\Ahttp:\/\/iss.ndl.go.jp\//
    iss_rdf_url = iss_url + ".rdf"
    begin
      xml = URI.open(iss_rdf_url).read
    rescue OpenURI::HTTPError => e
      STDERR.puts [ "WARN", iss_rdf_url, e.message ].join("\t")
    end
    if xml
      doc = Nokogiri::XML(xml)
      doc.xpath('//dcterms:creator/foaf:Agent').each do |creator|
        creators << creator.at('./foaf:name').content
      end
      doc.xpath('//dcterms:publisher/foaf:Agent').each do |publisher|
        publishers << publisher.at('./foaf:name').content
      end
    end
  end
  puts [ manifestation_id, iss_url,
         creators.join("//"), publishers.join("//") ].join("\t")
end