Batch editing with DigiTool Web Services

gistfile1.rb

#!/usr/bin/env ruby

require 'nokogiri'
require 'csv'
require 'savon'

Savon.configure do |config|
  config.log = false
end
HTTPI.log = false

# define wsdl connection
CLIENT = Savon::Client.new do
  wsdl.document = "http://libdigit1.lib.tcu.edu:1801/de_repository_web/services/DigitalEntityManager?wsdl"
end

# auth
GENERAL = Nokogiri::XML::Builder.new do |xml|
  xml.general("xmlns:xb"=>"http://com/exlibris/digitool/repository/api/xmlbeans") {
    xml.parent.namespace = xml.parent.namespace_definitions.find{|ns|ns.prefix=="xb"}
    xml.application { xml.parent.namespace = nil; xml.text "DIGITOOL-3" }
    xml.owner { xml.parent.namespace = nil; xml.text "SPC01" }
    xml.interface_version { xml.parent.namespace = nil; xml.text "1.0" }
    xml.user { xml.parent.namespace = nil; xml.text "creator:JHBROWN" }
    xml.password { xml.parent.namespace = nil; xml.text "mypass" }
  }
end.to_xml

def read_query_builder(pid)
  Nokogiri::XML::Builder.new do |xml|
    xml.digital_entity_call("xmlns:xb"=>"http://com/exlibris/digitool/repository/api/xmlbeans") {
      xml.parent.namespace = xml.parent.namespace_definitions.find{|ns|ns.prefix=="xb"}
      xml.digital_entity { xml.pid { xml.parent.namespace = nil; xml.text pid } }
      xml.command { xml.parent.namespace = nil; xml.text "retrieve" }
    }
  end.to_xml
end

def update_query_builder(pid, mid, xml_block)
  query = Nokogiri::XML::Builder.new do |xml|
    xml.digital_entity_call("xmlns:xb"=>"http://com/exlibris/digitool/repository/api/xmlbeans") {
      xml.parent.namespace = xml.parent.namespace_definitions.find{|ns|ns.prefix=="xb"}
      xml.digital_entity {
        xml.pid {
          xml.parent.namespace = nil
          xml.text pid
        }
        xml.mds {
          xml.parent.namespace = nil
          xml.md("cmd" => "update") {
            xml.parent.namespace = nil
            xml.mid { xml.parent.namespace = nil; xml.text mid }
            xml.name { xml.parent.namespace = nil; xml.text "descriptive" }
            xml.type { xml.parent.namespace = nil; xml.text "dc" }
            xml.value {
              xml.parent.namespace = nil
              xml.cdata(xml_block)
            }
          }
        }
      }
      xml.command { xml.parent.namespace = nil; xml.text "update" }
    }
  end.to_xml
  return query
end

def soap_proc(client,general,query)
  response = client.request :api, :digital_entity_call, "env:encodingStyle" => "http://schemas.xmlsoap.org/soap/encoding/" do
    soap.namespaces["xmlns:api"] = "http://api.repository.digitool.exlibris.com"
    xdoc = Nokogiri::XML::DocumentFragment.parse ""; Nokogiri::XML::Builder.with(xdoc) do |xml|
      xml.general("xsi:type" => "soapenc:string", "xmlns:soapenc" => "http://schemas.xmlsoap.org/soap/encoding/") { xml.cdata general }
      xml.query("xsi:type" => "soapenc:string", "xmlns:soapenc" => "http://schemas.xmlsoap.org/soap/encoding/") { xml.cdata query }
    end
    soap.body = xdoc
  end
  return (response.to_hash)[:digital_entity_call_response][:digital_entity_call_return]
end

# method returns nil unless has Dublin Core md
def get_dc_stuff(xml_string)
  (Nokogiri::XML(xml_string)).at_xpath("//md[type='dc']/value").tap do |o|
    temp = {}
    temp[:mid] = o.at_xpath("../mid/text()").content unless o.nil? rescue nil
    temp[:content] = Nokogiri::XML(o.content).to_xml unless o.nil? rescue nil
    return temp
  end
end

def change_add_dates(xml_block,new_date,numeric_date)
  temp = nil
  temp = Nokogiri::XML(xml_block)
  temp.at_xpath("//dc:date","dc" => "http://purl.org/dc/elements/1.1/").content = new_date
  if temp.root.namespace_definitions.find{|ns|ns.prefix=="dcterms"}.nil?
    temp.root.add_namespace_definition("dcterms","http://purl.org/dc/terms/")
  end
  unless numeric_date.nil? || temp.at_xpath("//dcterms:issued","dcterms" => "http://purl.org/dc/terms/")
    numeric = nil
    numeric = Nokogiri::XML::Node.new "issued", temp
    numeric.namespace = temp.root.namespace_definitions.find{|ns|ns.prefix=="dcterms"}
    numeric.content = numeric_date
    temp.root.add_child(numeric)
  end
  return temp.to_xml
end


f = File.open('/home/jhbrown/shared_with_fedora/dates_to_convert_checked_by_Amy.csv') do |file|
  CSV.parse(file.read) # , :headers => true, :header_converters => :symbol)
end

# done so far -> 1000 AND 3000 ->
g = f[1..-1].sort_by{|x|x[0]}

# require 'csv'; f = File.open('/home/jhbrown/shared_with_fedora/dates_to_convert_checked_by_Amy.csv') { |file| CSV.parse(file.read) }; g = f[1..-1].sort_by{|x|x[0]};

dothis = g[892..1000]
output = []

dothis.each_with_index do |x, index|
  puts "Working: #{x[0]}"
  pid = x[0]; new_date = x[2]; numeric_date = x[3]
  raw_xml = soap_proc(CLIENT, GENERAL, read_query_builder(pid))
  parsed_xml = get_dc_stuff(raw_xml)
  fixed_dates = change_add_dates(parsed_xml[:content],new_date,numeric_date)
  query = update_query_builder(pid, parsed_xml[:mid], fixed_dates)
  # output << fixed_dates
  # output << soap_proc(CLIENT, GENERAL, query)
  tempo = Nokogiri::XML(soap_proc(CLIENT, GENERAL, query))
  output << tempo.at_xpath("//pid/text()").content rescue "Error: #{pid}"
  sleep 10 and puts "Sleeping..." if index % 50 == 0
end
puts "---"
puts output