$:.unshift('lib') $:.unshift('ext/nokogumboc') gem 'minitest' require 'nokogumbo' require 'minitest/autorun' class TestNokogumbo < Minitest::Test def test_element_text doc = Nokogiri::HTML5(buffer) assert_equal "content", doc.at('span').text end def test_element_cdata_textarea doc = Nokogiri::HTML5(buffer) assert_equal "foobar", doc.at('textarea').text.strip end def test_element_cdata_script doc = Nokogiri::HTML5.fragment(buffer) assert_equal true, doc.document.html? assert_equal "", doc.at('script').to_s end def test_attr_value doc = Nokogiri::HTML5(buffer) assert_equal "utf-8", doc.at('meta')['charset'] end def test_comment doc = Nokogiri::HTML5(buffer) assert_equal " test comment ", doc.xpath('//comment()').text end def test_unknown_element doc = Nokogiri::HTML5(buffer) assert_equal "main", doc.at('main').name end def test_IO require 'stringio' doc = Nokogiri::HTML5(StringIO.new(buffer)) assert_equal 'textarea', doc.at('form').element_children.first.name end def test_nil doc = Nokogiri::HTML5(nil) assert_equal 1, doc.search('body').count end if ''.respond_to? 'encoding' def test_macroman_encoding mac="\xCA".force_encoding('macroman') doc = Nokogiri::HTML5(mac) assert_equal ' ', doc.at('span').to_xml end def test_iso8859_encoding iso8859="Se\xF1or".force_encoding(Encoding::ASCII_8BIT) doc = Nokogiri::HTML5(iso8859) assert_equal 'Señor', doc.at('span').to_xml end def test_charset_encoding utf8="Se\xC3\xB1or". force_encoding(Encoding::ASCII_8BIT) doc = Nokogiri::HTML5(utf8) assert_equal 'Señor', doc.at('span').to_xml end def test_bogus_encoding bogus="Se\xF1or". force_encoding(Encoding::ASCII_8BIT) doc = Nokogiri::HTML5(bogus) assert_equal 'Señor', doc.at('span').to_xml end end def test_html5_doctype doc = Nokogumbo.parse("") assert_match //, doc.to_html end def test_fragment_head doc = Nokogiri::HTML5.fragment(buffer[/(.*?)<\/head>/m, 1]) assert_equal "hello world", doc.xpath('title').text assert_equal "utf-8", doc.xpath('meta').first['charset'] end def test_fragment_body doc = Nokogiri::HTML5.fragment(buffer[/(.*?)<\/body>/m, 1]) assert_equal 'content', doc.xpath('main/span').to_xml assert_equal " test comment ", doc.xpath('comment()').text end def test_xlink_attribute source = <<-EOF.gsub(/^ {6}/, '') EOF doc = Nokogiri::HTML5.fragment(source) a = doc.at('a') assert_equal ["xlink:href", "xmlns:xlink"], a.attributes.keys.sort end def test_template source = <<-EOF.gsub(/^ {6}/, '') EOF doc = Nokogiri::HTML5.fragment(source) template = doc.at('template') assert_equal "productrow", template['id'] assert_equal "record", template.at('td')['class'] end def test_root_comments doc = Nokogiri::HTML5("") assert_equal ["html", "comment", "html", "comment"], doc.children.map(&:name) end private def buffer <<-EOF.gsub(/^ /, '') hello world

hello world

content

EOF end end