Vital Ruby Advance Lab 2

html_parser.rb

require "nokogiri"
require "uri"

class HtmlParser
  def parse(source, html_string)
    uri = URI.parse(source)
    html_doc = Nokogiri::HTML(html_string)
    anchors = html_doc.xpath('//a[@href!="" and not(starts-with(@href, "#"))]')
    links = anchors.map { |elem|
      elem.attribute('href').value
    }

    absolute_links = links.map { |link|
      uri.merge(link) rescue nil
    }

    http_links = absolute_links.select { |link|
      link.is_a?(URI::HTTP) || link.is_a?(URI::HTTPS)
    }

    http_links.map { |uri| uri.to_s }
  end
end