Dumping a useful but ugly script for wrangling misbehaving MMS records.

reject.rb

require 'ostruct'
require 'rubygems'
require 'rest_client'
require 'nokogiri'


SOLR_ROOT = 'http://search.tki.org.nz:8983'
FEDORA_ROOT = 'username:password@fedora-mms:8080/fedora'

@xpaths = {
  :pid => '//str[@name="id"]',
  :title => '//arr[@name="title_t"]/str[1]',
  :url => '//arr[@name="url"]/*'
}


# search
response = RestClient.get "#{SOLR_ROOT}/solr/select?q=Kawekawe&rows=42400"

results = []

puts results.count

# parse responses into ostructs
Nokogiri::XML(response.body).xpath('//doc').each do |doc|
	doc = Nokogiri::XML.parse(doc.to_s) # confused... inefficient...
	item = OpenStruct.new
  @xpaths.each do |k,v|
    item.send("#{k}=", doc.xpath(v).text)
  end
	results << item
end

#delete if not a record, or doesn't match our desired url
results.delete_if do |r|
  r.pid !~ /TKI39128/
end

# for every id
results.each do |r|
  puts "\nFinding #{r.pid} in Fedora..."
  # give it a rest...
  sleep 0.2

  begin
    # get Lom record and parse
    response = RestClient.get "#{FEDORA_ROOT}/objects/MoE:#{r.pid}/datastreams/LOM/content"
    record = Nokogiri::XML(response.body)
    status = record.xpath('//ns:status/ns:value', 'ns' => 'http://ltsc.ieee.org/xsd/LOM').text


    #Fix for weird issue...
    contrib_entries = record.xpath('//ns:contribute/ns:entry', 'ns' => 'http://ltsc.ieee.org/xsd/LOM')

    contrib_entries.each_with_index do |entry, index|
      entry.remove if index.odd?
    end


    # reject record if it's live
    if status == 'Live'
      # puts r.url
      # puts "Status is #{status} - changing to Rejected..."

      # # Changing value to 'Rejected'
      # for statusnode in record.xpath('//ns:status/ns:value', 'ns' => 'http://ltsc.ieee.org/xsd/LOM')
      #   statusnode.content = statusnode.content.gsub!('Live', 'Rejected')
      # end

      # # # Print out new status
      # puts record.xpath('//ns:status/ns:value', 'ns' => 'http://ltsc.ieee.org/xsd/LOM').text

      # # PUT request back to update fedora
      request = "#{FEDORA_ROOT}/objects/MoE:#{r.pid}/datastreams/LOM"

      RestClient.put request, record.to_s, :content_type => 'text/xml; charset=utf-8'
    else
      puts "Status was #{status} - Ignoring..."
    end

  rescue => e
    puts e.response
  end
end