#!/usr/bin/env ruby

#!/usr/bin/env ruby
# encoding: UTF-8

# resolve bin path, ignoring symlinks
require "pathname"
bin_file = Pathname.new(__FILE__).realpath

# add self to libpath
$:.unshift File.expand_path("../../lib", bin_file)

# Fixes https://github.com/rubygems/rubygems/issues/1420
require "rubygems/specification"

class Gem::Specification
  def this; self; end
end

require 'bundler/setup'
require 'stepmod/utils/stepmod_definition_converter'
require 'stepmod/utils/bibdata'
require 'stepmod/utils/concept'

ReverseAdoc.config.unknown_tags = :bypass

parsed_terms = []
parsed_bibliography = []
encountered_terms = {}

stepmod_dir = ARGV.first || Dir.pwd

def log message
  puts "[stepmod-utils] #{message}"
end

log "STEPmod directory set to #{stepmod_dir}. Detecting paths..."

files = %w(
  resource.xml
  application_protocol.xml
  business_object_model.xml
  module.xml
  ).inject([]) do |acc, t|
    acc << Dir["#{stepmod_dir}/**/#{t}"]
end.flatten.sort.uniq

files.each do |file_path|
  log "Processing XML file #{file_path}"
  current_document = Nokogiri::XML(File.read(file_path)).root

  bibdata = nil
  begin
    bibdata = Stepmod::Utils::Bibdata.new(document: current_document)
  rescue
    log "WARNING: Unknown file #{file_path}, skipped"
    next
  end

  # TODO: we may want a command line option to override this in the future
  unless %w(IS DIS TS).include? bibdata.doctype
    log "WARNING: skipping #{bibdata.docid} as it is not IS, DIS or TS"
    next
  end

  if bibdata.part.to_s.empty?
    log "FATAL: missing `part` attribute: #{file_path}"
  end

  # read definitions
  current_document.xpath('//definition').each.with_index(1) do |definition, index|

    term_id = definition['id']
    unless term_id.nil?
      if encountered_terms[term_id]
        log "FATAL: Duplicated term with id: #{term_id}, #{file_path}"
      end
      encountered_terms[term_id] = true
    end

    concept = Stepmod::Utils::Concept.parse(
      definition,
      reference_anchor: bibdata.anchor,
      # Assume that definition is located in clause 3 of the ISO document
      # in order. We really don't have a good reference here.
      reference_clause: "3.#{index}"
    )

    parsed_terms << concept
    parsed_bibliography << bibdata
  end

end

parsed_bibliography.uniq!

File.open('031-generated-terms.adoc', 'w') { |file|
  file.puts(parsed_terms.map(&:to_mn_adoc).join("\n"))
}

log "written to: 031-generated-terms.adoc"

File.open('991-generated-bibliography.adoc', 'w') { |file|
  file.puts(parsed_bibliography.map(&:to_mn_adoc).join("\n"))
}

log "written to: 991-generated-bibliography.adoc"