parse-bosc-posters-2025.rb

require 'nokogiri'
require 'csv'
require 'open-uri'

urls = [
  'https://www.iscb.org/cms_addon/conferences/ismbeccb2025/posters.php?track=BOSC&session=A#search',
  'https://www.iscb.org/cms_addon/conferences/ismbeccb2025/posters.php?track=BOSC&session=B#search',
  'https://www.iscb.org/cms_addon/conferences/ismbeccb2025/posters.php?track=BOSC&session=E#search'
]

csv = CSV.generate do |csv_out|
  csv_out << ['Title', 'Poster Presenter', 'PosterID', 'Day', 'SubmissionId']

  submission_counter = 1000

  urls.each do |url|
    html = URI.open(url).read
    doc = Nokogiri::HTML(html)

    doc.css('div.well-sm strong').each do |title_elem|
      title_text = title_elem.text.strip

      if title_text.start_with?('Virtual:')
        poster_id = 'Virtual'
        title = title_text.sub('Virtual:', '').strip
        day = 'Virtual'
      elsif title_text =~ /^(A|B|E)-\d+: /
        poster_id, title = title_text.split(':', 2).map(&:strip)
        day = poster_id.start_with?('A') ? 'Jul-21' : poster_id.start_with?('B') ? 'Jul-22' : 'Jul-23'
      else
        next
      end

      container_div = title_elem.ancestors('div.well-sm').first
      next_divs = container_div.xpath('following-sibling::div')
      presenter = 'UNKNOWN'

      next_divs.each do |div|
        if div.to_html.include?('<ul') && div.css('li.author').any?
          presenter_elem = div.at_css('li.author u strong') || div.at_css('li.author strong')
          if presenter_elem
            presenter = presenter_elem.text.strip.split(',').first
          end
          break
        end
      end

      csv_out << [title, presenter, poster_id, day, "S-#{submission_counter}"]
      submission_counter += 1
    end
  end
end

puts csv