Transforming subject headings using ruby-marc

transform.rb

# TRANSFORM SUBJECT HEADINGS EXAMPLE
require 'marc'

# Function: text_to_datafield
# Returns a vanilla MARC::DataField object from string
# Format examples:
# 655 _0 Account books.
# 655 _7 3-D films.|2lcgft

def text_to_datafield(text_field)
  tag = text_field[(0..2)]
  ind1 = text_field[4] == '_' ? ' ' : text_field[4]
  ind2 = text_field[5] == '_' ? ' ' : text_field[5]
  subfields = []
  text = text_field.dup
  text[6] = '|a' # Coax subfield 'a'
  s = text.split(/\|/)
  s.shift # Remove the tag & inds
  s.each do |sub|
    subfields << MARC::Subfield.new(sub[0], sub[(1..sub.length - 1)].strip)
  end
  MARC::DataField.new(tag, ind1, ind2, *subfields)
end

# Read file containing subject heading pairs such as:
# 655 _0 Action and adventure films.=655 _7 Adventure films.|2gsafd
# 655 _0 Adventure stories.=655 _7 Adventure fiction.|2gsafd

transforms = {}
# http://www.libcode.net/transforms.html for examples
File.open('config/transforms.txt').each_line do |transform|
  from, to = transform.strip.split '='
  # Use string as safe hash key
  transforms[text_to_datafield(from).value] = text_to_datafield(to)
end

w = MARC::Writer.new('records/transformed.mrc')

MARC::ForgivingReader.new('records/transforms.dat').each do |r|
  del = []
  add = []
  r.find_all { |f| f.tag =~ /65./ }.each do |s|
    if transforms.has_key? s.value
      del << s
      add << transforms[s.value]
      puts 'WAS: ' + s.to_s
      puts 'NOW: ' + transforms[s.value].to_s
      puts '-' * 20
    end
  end
  del.each { |d| r.fields.delete d }
  add.each { |a| r.append a }
  r.fields.sort_by! { |f| f.tag }
  w.write r
end

w.close