base.rb

module Onix21
  class Base
    include Logger
    attr_reader :values
    attr_reader :properties

    def section_matches_class section
      self_class = self.class.to_s.underscore.split('/').last.gsub("_",'').to_sym
      raise "#{self_class.inspect} expected, but got #{section.inspect}" unless self_class == section
    end
    
    def initialize section, values, parent = nil
      section_matches_class section
      @properties = []
      @values = values.first.is_a?(Array) ? values : [values]
      assign_properties
      unless fully_consumed?
        puts '=' * 40
        puts section.inspect
        pp @values 
        puts "Belongs to: #{parent.class.name}" if parent
        puts '-' * 40
      end
    end
  
    def extract_items xml_tag
      (extracted, @values = @values.partition{|p| p[0] == xml_tag}).first
    end
  
    def extract section, parent = nil
      xml_tag = section.to_s.gsub("_", '').to_sym
      klass = "Onix21::#{section.to_s.camelize}".constantize
      extract_items(xml_tag).map{ |values| klass.new(*values, parent) }
    end

    def set_if_present property_name, xml_tag, present_value
      extracted = extract_items xml_tag
      instance_variable_set "@#{property_name}", (extracted.empty? ? !present_value : present_value)
    end

    def extract_property property_name, xml_tag, repeatable = false
      limit = (!!repeatable ? 999 : 1)
      extracted = extract_items xml_tag
      if extracted.size > limit
        raise "set_property expects only one instance for #{property_name}/#{xml_tag}, but got #{extracted.size}" 
      end
      return Array(extracted.first).last
    end
    
    def set_property property_name, xml_tag, repeatable = false
      value = extract_property(property_name, xml_tag, repeatable)
      instance_variable_set "@#{property_name}", value
    end

    # Can set date from any of YYYY, YYYYMM, or YYYYMMDD
    def set_property_date property_name, xml_tag
      value = extract_property(property_name, xml_tag, false)
      if value and matches = value.match(/(\d{4})(\d{2})?(\d{2})?/)
        instance_variable_set "@#{property_name}", Date.new(*matches.captures.compact.map(&:to_i))
      else
        instance_variable_set "@#{property_name}", nil
      end
    end
    
    def set_property_list property_name, xml_tag, repeatable = false
      value = extract_property(property_name, xml_tag, repeatable)
      instance_variable_set "@#{property_name}", value.nil? ? nil : value.split(" ")
    end
    
    def fully_consumed?
      values.empty?
    end
  end
end

product.rb

module Onix21
  class Product < Base
    attr_reader :values
    attr_reader :main_subject, :subjects, :identifiers, :contributors, :supply_details
    attr_reader :series, :sales_rights, :measures, :work_identifiers
    attr_reader :publishers, :product_websites, :audience_ranges
    attr_reader :imprints, :related_products, :titles
    attr_reader :contained_items, :sets, :extents
  
    attr_reader :bwk_rec_ref_num, :other_texts
    attr_reader :notification_type, :record_source_type, :record_source_name
    attr_reader :barcodes, :product_form, :product_form_details, :product_packaging
    attr_reader :epub_type, :epub_type_description, :epub_type_note, :epub_format, :epub_source, :epub_source_description
    attr_reader :number_of_pages, :number_of_illustrations, :pages_roman, :pages_arabic
    attr_reader :bisac_main_subject_category, :bisac_version
    attr_reader :audience_codes
    attr_reader :city_of_publication, :country_of_publication
    attr_reader :publication_date, :copyright_year, :year_first_published, :out_of_print_date
    attr_reader :edition_type_codes, :edition_number
    attr_reader :publishing_status_code
    attr_reader :illustrations_note
    attr_reader :no_contributor
    attr_reader :no_series
    attr_reader :trade_category
    attr_reader :number_of_volumes
    attr_reader :languages

    attr_reader :distinctive_title, :title_prefix, :title_without_prefix, :sub_title
    attr_reader :title_text_case, :translated_title, :former_title

    def id
      @id ||= (item = @identifiers.detect{|d| d.type == '03'} and item.id)
    end
    
    def isbn
      @isbn ||= (item = @identifiers.detect{|d| d.type == '02'} and item.id)
    end

    def publisher_record
      @publisher_record ||= publishers.first.record rescue nil
    end
    
    def publisher_id
      publisher_record and publisher_record[:id]
    end
    
    def imprint_record
      @imprint_record ||= imprints.first.record rescue nil
    end
    
    def imprint_id
      return imprint_record[:id] if imprint_record
      return publisher_record[:id] if publisher_record
    end
    
    def imprint_name
      return imprint_record[:name] if imprint_record
      return publisher_record[:name] if publisher_record
    end

    def description
      if other_text = other_texts.detect{|d| d.type == '01'}
        other_text.description
      else
        ''
      end
    end
    
    def title_parts
      if product_title = titles.detect{|d| d.type == '01'}
        product_title.title_parts
      else
        raise ImportError, "Title not found for type '01'" 
      end
    end
    
    def get_avail_status_id(avail_status_code)
      raise ImportError, "No avail_status_code given!" if avail_status_code.blank?
      db[:avail_statuses].filter(code: avail_status_code).get(:id)
    end
    
    def get_avail_status_code
      unless publishing_status_code.blank?
        code = db[:avail_statuses___av].
          join(:publishing_statuses___ps, :ps__avail_status_id => :av__id).
          filter(ps__code: publishing_status_code).
          get(:av__code)
        return code unless code.blank?
      end
      
      return product_availability_code unless product_availability_code.blank?
      
      unless fallback_availability_status_code.blank?
        code = db[:avail_statuses___av].
          join(:product_avail_statuses___pas, :pas__avail_status_id => :av__id).
          filter(pas__code: fallback_availability_status_code).
          get(:av__code)
        return code unless code.blank?
      end

      raise ImportError, "Could not look up avail_status_code"
    end

    def product_availability_code
      (sd = supply_details.first) ? sd.product_availability_code : nil
    end
    
    def fallback_availability_status_code
      (sd = supply_details.first) ? sd.availability_status_code : nil
    end
    
    def availability_status_code
      @avail_status_code ||= get_avail_status_code(fallback_avail_status)
    end
    
    def availability_status_id
      @avail_status_id ||= get_avail_status_id(availability_status_code)
    end

    # supply_to_country and supply_to_territory supplies detailed regional availability,
    # however, Polybook only cares about "US" and "GB" where Bowker data is concerned, 
    # so we base market_code off the Bowker UID for the product and ignore the territorial data
    def market_code
      return "US" if bwk_rec_ref_num =~ /USA/
      return "GB"
    end
    
    def replaced_by_product
      @replaced ||= related_products.detect{|p| p.type == '05'}
    end
    
    def replaces_product
      @replaces ||= related_products.detect{|p| p.type == '03'}
    end
    
    def replaced_by_item_id
      replaced_by_product ? replaced_by_product.id : nil
    end
    
    def replaces_item_id
      replaces_product ? replaces_product.id : nil
    end
    
    # NOTE: the leading comma is intentional (as taken from original Perl script).
    def pagination
      [pages_roman.to_s, (pages_arabic || number_of_pages).to_s].join(", ")
    end

    def get_measure type
      @measures.detect{|d| d.type == type}
    end
    
    def weight
      (measure = get_measure("08")) ? measure.convert_to_grams : nil
    end
    
    def dimensions_length
      (measure = get_measure("01")) ? measure.convert_to_cm : nil
    end
    
    def dimensions_width
      (measure = get_measure("02")) ? measure.convert_to_cm : nil
    end
    
    def dimensions_height
      (measure = get_measure("03")) ? measure.convert_to_cm : nil
    end
    
    def website_urls
      product_websites.map(&:short_url).join(";")
    end
    
    def publication_language
      (lang = @languages.detect{|l| l.role == '01'}) ? lang.code : nil
    end
    
    def original_language
      (lang = @languages.detect{|l| l.role == '02'}) ? lang.code : nil
    end
    
    def all_subjects
      @all_subjects = main_subject + subjects
    end
    
    def dewey
      @dewey ||= (subj = all_subjects.detect{|s| s.scheme_id == '01'}) ? subj.value : nil
    end
    
    def lc_class
      @lc_class ||= (subj = all_subjects.detect{|s| s.scheme_id == '03'}) ? subj.value : nil
    end
    
    def work_id
      (ident = @worker_identifiers.detect{|w| w.type == '01'}) ? ident.id : nil
    end
    
    def assign_properties
      @main_subject = extract(:main_subject)
      @subjects = extract(:subject)
      @identifiers = extract(:product_identifier)
      @contributors = extract(:contributor)
      @supply_details = extract(:supply_detail)
      @series = extract(:series)
      @other_texts = Array(extract(:other_text))
      @sales_rights = extract(:sales_rights)
      @measures = extract(:measure)
      @work_identifiers = extract(:work_identifier)
      @publishers = extract(:publisher)
      @titles = extract(:title)
      @languages = extract(:language)
      @product_websites = extract(:product_website)
      @audience_ranges = extract(:audience_range)
      @imprints = extract(:imprint)
      @related_products = extract(:related_product)
      @prizes = extract(:prize)
      @contained_items = extract(:contained_item)
      @sets = extract(:set)
      @extents = extract(:extent)
    
      set_property :bwk_rec_ref_num, :a001
      set_property :notification_type, :a002
      set_property :record_source_type, :a194
      set_property :record_source_name, :a197
      set_property :barcode, :b246, :repeatable
      set_property :product_form, :b012
      set_property :product_form_details, :b333, :repeatable
      set_property :product_packaging, :b225
      set_property :epub_type, :b211
      set_property :epub_type_description, :b213
      set_property :epub_type_note, :b277
      set_property :epub_format, :b214
      set_property :epub_source, :b278
      set_property :epub_source_description, :b280
      set_property :number_of_pages, :b061
      set_property :pages_roman, :b254
      set_property :pages_arabic, :b255
      set_property :bisac_main_subject_category, :b064
      set_property :bisac_version, :b200
      set_property :audience_codes, :b073, :repeatable
      set_property :city_of_publication, :b209
      set_property :country_of_publication, :b083
      set_property_date :publication_date, :b003
      set_property_date :out_of_print_date, :h134
      set_property :copyright_year, :b087
      set_property :edition_type_codes, :b056, :repeatable
      set_property :publishing_status_code, :b394
      set_property :number_of_illustrations, :b125
      set_property :illustrations_note, :b062
      set_property :edition_number, :b057
      set_property :year_first_published, :b088
      set_if_present :no_contributor, :n339, true
      set_if_present :no_series, :n338, true
      set_property :trade_category, :b384
      set_property :number_of_volumes, :b210

      set_property :distinctive_title, :b028
      set_property :title_prefix, :b030
      set_property :title_without_prefix, :b031
      set_property :sub_title, :b029
      set_property :title_text_case, :b027
      set_property :translated_title, :b032
      set_property :former_title, :b033
    end
  end
end