Parse example

fragrantica.rake

def parse_fragrance_notes(fragrance_doc)
    fragrance_notes = []
    fragrance_doc.xpath('//div[@id="col1"]//div[@style="width: 230px; float: left; text-align: center; clear: left;"]/p').each do |notes_node|
      if notes_node.content.include?('Top Notes')
        note_type = FragranceNoteType.find_by_name('top')
      elsif notes_node.content.include?('Middle Notes')
        note_type = FragranceNoteType.find_by_name('middle')
      elsif notes_node.content.include?('Base Notes')
        note_type = FragranceNoteType.find_by_name('base')
      end
      notes_node.xpath('span/img').each do |note_img|
        note = Note.where(name: note_img['alt'].to_s).first_or_create do |note|
          puts "parsing note '#{note.name}'"
          logo_url = "http://www.fragrantica.com/images/sastojci/o#{note_img['src'].split("/")[-1][1..-1]}"
          begin
            note.logo = URI.parse(logo_url)
            adapter = Paperclip.io_adapters.for(note.logo)
            if Paperclip::MediaTypeSpoofDetector.using(adapter, note.logo.original_filename).spoofed?
              note.logo = nil
            end
          rescue OpenURI::HTTPError
            puts 'Broken logo URL: ' + logo_url
          end
        end

        fragrance_note = FragranceNote.new(note_id: note.id)
        fragrance_note.fragrance_note_type_id = note_type.id if note_type
        fragrance_notes << fragrance_note
      end
    end
    fragrance_notes
  end