monograph_resource_metadata

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

# New version of process_monograph_resource_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
option_parser = OptionParser.new do |opts|
  opts.banner = "Usage: #{File.basename(__FILE__)} monograph_dir [monograph_dir...]"
  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)
  return
end

# Process the command line parameters.
monograph_dir_list = ARGV

require_relative File.join(root_dir, "lib", "epub")
require_relative File.join(root_dir, "lib", "fulcrum", "monographdir")
require_relative File.join(root_dir, "lib", "fulcrum", "metadata")

xml_processor = UMPTG::Fulcrum::Metadata::Processor.new()

epub_processor = UMPTG::EPUB::XProcessor.new(
      xml_processor: xml_processor
    )

monograph_dir_list.each do |mdir|
  monograph_dir = UMPTG::Fulcrum::MonographDir.new(
            monograph_dir: mdir
            )
  if monograph_dir.monograph_dir.nil?
    script_logger.error("invalid monograph directory #{mdir}")
    next
  end
  script_logger.info("processing monograph directory #{monograph_dir.monograph_dir}")

  # Create the monograph processing directory.
  FileUtils.mkdir_p monograph_dir.processing_dir

  # Create the processing log file for this monograph directory.
  logger_file =  File.join(
            monograph_dir.processing_dir,
            File.basename(__FILE__) + ".log"
            )
  logger = UMPTG::Logger.create(logger_file: logger_file)
  xml_processor.logger = logger
  epub_processor.logger = logger
  epub_processor.logger.info("*** Processing Monograph directory #{File.basename(monograph_dir.monograph_dir)} ***")

  if monograph_dir.fmsl_file.nil?
    epub_processor.logger.warn("no FMSL file loaded.")
    next
  end

  epub = UMPTG::EPUB(epub_file: monograph_dir.epub_file)

  epub_processor.logger.info("Using EPUB #{File.basename(monograph_dir.epub_file)}.")
  entry_actions = epub_processor.run(
        epub,
        {
          normalize: true
        }
      )

  epub_processor.logger.info("Using FMSL #{File.basename(monograph_dir.fmsl_file)}.")
  epub_processor.logger.warn("No alt text/caption found for resource \"***row left intentionally blank***\".")

  fmsl_csv = xml_processor.update_fmsl(
      fmsl_file: monograph_dir.fmsl_file,
      entry_actions: entry_actions
    )

  # Add new columns to the CSV headers if needed.
  new_fmsl_headers = fmsl_csv.headers
  new_fmsl_headers << "Caption" unless new_fmsl_headers.include?("Caption")
  new_fmsl_headers << "Alternative Text" unless new_fmsl_headers.include?("Alternative Text")

  # Save the updated FMSL in the resource processing directory.
  new_fmsl_file = File.join(monograph_dir.processing_dir, File.basename(monograph_dir.fmsl_file))
  CSV.open(
          new_fmsl_file,
          "w",
          :write_headers=> true,
          :headers => new_fmsl_headers
        ) do |csv|
    fmsl_csv.each do |fmsl_row|
      new_row = {}
      fmsl_row.each do |key,value|
        new_row[key] = value.strip.force_encoding("UTF-8") unless value.nil?
      end
      csv << new_row
    end
  end
end