#!/usr/bin/env ruby

require_relative '../lib/bipm-data-importer'

BASE_DIR = "data"
a = Mechanize.new

meetings_en = VCR.use_cassette 'cgpm-meetings' do
  a.get "https://www.bipm.org/en/committees/cg/cgpm/cgpm-resolutions"
end

meetings_fr = VCR.use_cassette 'cgpm-meetings-fr' do
  a.get "https://www.bipm.org/fr/committees/cg/cgpm/cgpm-resolutions"
end

FileUtils.rm_rf "#{BASE_DIR}/cgpm/meetings"
FileUtils.rm_rf "#{BASE_DIR}/cgpm/meetings-fr"
FileUtils.rm_rf "#{BASE_DIR}/cgpm/meetings-en"

[['en', meetings_en], ['fr', meetings_fr]].each do |meeting_lang, meetings|
  urls = meetings.css('div.publications__content').map do |option|
    url = option.at_css('a').attr('href')
    url = url.gsub('/web/guest/', "/#{meeting_lang}/")
    url.split('/').first(8).join('/')
  end.uniq

  urls.each do |url|
    meeting_id = url.split('/').last.to_i
    meeting_lang_sfx     = (meeting_lang == 'fr') ? "-fr" : ""
    meeting_lang_sfx_dir = (meeting_lang == 'fr') ? "-fr" : "-en"
    meeting = VCR.use_cassette("cgpm-meeting-#{meeting_id}#{meeting_lang_sfx}") { a.get url }

    title = meeting.at_css('h1.session__title, .journal-content-article h1').text.strip
    date = Bipm::Data::Importer::Common.extract_date(meeting.at_css('p.session__date, .journal-content-article h2').text)

    pdf = Bipm::Data::Importer::Common.extract_pdf(meeting, meeting_lang)

    h = {
      "metadata" => {
        "title" => title,
        "identifier" => meeting_id,
        "date" => date.to_s,
        "source" => "BIPM - Pavillon de Breteuil",
        "url" => meeting.uri.to_s
      }
    }

    h["pdf"] = pdf if pdf

    resolutions = meeting.links_with(href: %r</resolution->).map(&:href)

    # A mistake on a website, resolution 5 listed 4 times...
    # https://www.bipm.org/fr/committees/cg/cgpm/8-1933
    if meeting_id == 8 && meeting_lang == 'fr' && resolutions.sort.uniq != resolutions.sort
      resolutions = (1..15).map do |i|
        "https://www.bipm.org/en/committees/cg/cgpm/8-1933/resolution-#{i}"
      end
    end

    h["resolutions"] = resolutions.map do |res_link|
      res_id = (res_link.split('-')[2] || 0).to_i
      res_link = res_link.gsub('/web/guest/', "/#{meeting_lang}/")
      res = VCR.use_cassette("cgpm-resolution-#{meeting_id}-#{res_id}#{meeting_lang_sfx}") { a.get res_link }

      Bipm::Data::Importer::Common.parse_resolution(res, res_id, date, :cgpm, meeting_lang, "resolution?")
    end

    FileUtils.mkdir_p("#{BASE_DIR}/cgpm/meetings#{meeting_lang_sfx_dir}")
    File.write("#{BASE_DIR}/cgpm/meetings#{meeting_lang_sfx_dir}/meeting-#{"%02d" % meeting_id}.yml", YAML.dump(h))
  end
end