btaa_gen_monograph_csv

#!/usr/bin/env ruby
# frozen_string_literal: true

# Script inputs a BTAA TMM metadata CSV file
# containing monograph metadata and
# generates a Fulcrum metadata CSV file with
# each row containing Fulcrum monograph metadata.

require 'optparse'
require 'ostruct'
require 'os'

# Determine the root directory of the code base.
script_dir = File.expand_path(File.dirname(__FILE__))
root_dir = File.dirname(script_dir)

require_relative File.join(root_dir, "lib", "logger")

script_logger = UMPTG::Logger.create(logger_fp: STDOUT)

# Process the script parameters.
options = OpenStruct.new
options.write_mono_file = false
option_parser = OptionParser.new do |opts|
  opts.banner = "Usage: #{File.basename(__FILE__)} <btaa_tmm_csv_file> [<btaa_tmm_csv_file>..]"
  opts.on_tail('-h', '--help', 'Print this help message') do
    script_logger.info(opts)
    exit 0
  end
end
option_parser.parse!(ARGV)
if ARGV.count < 1
  script_logger.info(option_parser.help)
  exit 0
end

btaa_tmm_csv_file_list = ARGV

require 'csv'

require_relative File.join(root_dir, 'lib', 'fulcrum', 'manifest')

MONOGRAPH_HEADERS = UMPTG::Fulcrum::Manifest::Validation::CollectionSchema.headers()

btaa_tmm_csv_file_list.each do |btaa_tmm_csv_file|
  btaa_tmm_csv_file = File.expand_path(btaa_tmm_csv_file)
  unless File.exist?(btaa_tmm_csv_file)
    script_logger.error("invalid CSV file path #{btaa_tmm_csv_file}.")
    next
  end
  script_logger.info("*** processing #{File.basename(btaa_tmm_csv_file)} ***")
  STDOUT.flush

  fm_csv = CSV.parse(
            File.read(btaa_tmm_csv_file),
            :headers => true,
            #:header_converters=> lambda {|f| fm_header_convert(f)},
            :return_headers => false
          )

  # Each title may have multiple rows. The extra
  # rows may contain additional ISBNs. Traverse
  # the rows and condense the extra ISBNs into
  # a new column named ISBN(s).
  fm_row_list = []
  frow = nil
  fm_csv.each do |fm_row|
    title = fm_row['Title']

    unless title.nil? or title.strip.empty?
      #script_logger.info("title:#{title}")
      frow = fm_row
      frow['ISBN(s)'] = ""
      frow['ISBN_SEARCH'] = ""
      fm_row_list << frow
    end
    isbn = fm_row['ISBN13'].strip
    isbn = isbn.gsub(/\-/,'')
    isbn = isbn[0..12]
    frow['ISBN_SEARCH'] += isbn + ";"
    isbn = "#{isbn[0..2]}-#{isbn[3..3]}-#{isbn[4..8]}-#{isbn[9..11]}-#{isbn[12..12]}"
    format = fm_row['Format'].strip.downcase
    entry = "#{isbn} (#{format});"
    frow['ISBN(s)'] += entry
  end

  # Map the TMM columns to Fulcrum import columns.
  monograph_row_list = []
  fm_row_list.each do |fm_row|
    monograph_row = {}

    # Make sure required fields are filled in:
    #   Title, Publisher (citation), Creator(s), Pub Year,
    #   Pub Location, Handle, Identifier(s), ISBN(s)
    unless fm_row.nil?
      title = fm_row['Title']
      title += ": #{fm_row['Subtitle']}" unless fm_row['Subtitle'].nil? or fm_row['Subtitle'].empty?
      monograph_row['Title'] = title

      author_list = []
      (1..4).each do |ndx|
        lname = fm_row["Author (#{ndx}) Last Name"]
        #lname = fm_row["authorlastname#{ndx}"]
        unless lname.nil? or lname.empty?
          fname = fm_row["Author (#{ndx}) First Name"]
          role = fm_row["Author (#{ndx}) Type"]
          orcid = fm_row["Author (#{ndx}) ORCHID(s)"]
          #puts orcid unless orcid.nil? or orcid.strip.empty?
          author = "#{lname}, #{fname}"
          author += " (#{role})" unless role.nil? or role.strip.empty?
          author += "|#{orcid}" unless orcid.nil? or orcid.strip.empty?
          author_list << author
        end
      end
      monograph_row['Creator(s)'] = author_list.join(';')

      fm_row.each do |key,val|
        next if key.nil? or key.strip.empty?

        case key
        when 'ID'
          monograph_row['Identifier(s)'] = val.downcase
        when 'Publisher'
          monograph_row['Publisher'] = val
        when 'Pub City 1','Publication Location'
          monograph_row['Pub Location'] = val
        when 'Book Description Marketing','Book Description'
          monograph_row['Description'] = val
        when 'Open Access Funder'
          monograph_row['Funder'] = val
        when 'Open Access Avail'
          monograph_row['Open Access?'] = val
        when 'Subjects'
          monograph_row['Subject'] = val
        when 'Copyright Holder'
          monograph_row['Rightsholder'] = val
        when 'Copyright Year'
          monograph_row['Pub Year'] = val
        when 'Handle Prefix'
        when 'ISBN13'
        when 'Pub Format', 'Pub Format 2'
        when 'Pub Date 2', 'Pub Date 3'
        when 'Title', 'Subtitle'
        else
          monograph_row[key] = val
        end
      end
    end
    monograph_row_list << monograph_row
  end

  # Generate a new CSV listing Fulcrum metadata
  # for each title.
  fulcrum_body = CSV.generate(
          :headers => MONOGRAPH_HEADERS + [ 'ISBN_SEARCH'],
          :write_headers => true
        ) do |csv|
    monograph_row_list.each do |monograph_row|
      csv << monograph_row
    end
  end

  # Save the Fulcrum metadata CSV file.
  fulcrum_file = File.join(
        File.dirname(btaa_tmm_csv_file),
        File.basename(btaa_tmm_csv_file, ".*") + "_mono" + File.extname(btaa_tmm_csv_file)
        )
  script_logger.info("Creating metadata file #{fulcrum_file}")
  File.write(fulcrum_file, fulcrum_body)
end