see newer version at https://gist.github.com/masao/10db0c36da39f23106a70518be4cf926

catp2tsv-v2.rb

#!/usr/bin/env ruby

data = {}
bibdata = {}
holdings = {}

ARGF.each do |line|
  line.chomp!
  line.scan(/(?:\w+=(?:.*?)|<\/?\w+\>) /).each do |element|
    case element
    when /\A(\w+)=(.+)\Z/
      key = $1.downcase.intern
      val = $2.chop
      data[key] = val
      case key
      when :id
        data[:ncid] = val
      when :trd
        data[:trd_title], data[:trd_author], = data[:trd].split(/ \/ /)
      end
    end
  end
  case data[:_dbname_]
  when "BOOK"
    bibdata[data[:ncid]] = data
  when "BHOLD"
    holdings[data[:bid]] ||= []
    holdings[data[:bid]] << data
  end
  data = {}
end

bibdata.sort_by{|k,v| k }.each do |ncid, v|
  if holdings[ncid]
    bib = [:ncid, :trd_title, :trd_author, :ed, :publ, :pubdt, :year1, :physp, :physs, :trr ].map{|e| v[e] }.join("\t")
    holdings[ncid].each do |holding|
      puts [ bib, holding[:vol], holding[:cln], holding[:rgtn] ].join("\t")
    end
  else
    puts [:ncid, :trd_title, :trd_author, :ed, :publ, :pubdt, :year1, :physp, :physs, :trr ].map{|e| v[e] }.join("\t")
  end
end