nokogiri 1.2.3 → 1.3.0

data/test/html/test_document_fragment.rb

@@ -0,0 +1,97 @@
+require File.expand_path(File.join(File.dirname(__FILE__), '..', "helper"))
+
+module Nokogiri
+  module HTML
+    class TestDocumentFragment < Nokogiri::TestCase
+      def setup
+        super
+        @html = Nokogiri::HTML.parse(File.read(HTML_FILE), HTML_FILE)
+      end
+
+      def test_new
+        fragment = Nokogiri::HTML::DocumentFragment.new(@html)
+      end
+
+      def test_fragment_should_have_document
+        fragment = Nokogiri::HTML::DocumentFragment.new(@html)
+        assert_equal @html, fragment.document
+      end
+
+      def test_name
+        fragment = Nokogiri::HTML::DocumentFragment.new(@html)
+        assert_equal '#document-fragment', fragment.name
+      end
+
+      def test_static_method
+        fragment = Nokogiri::HTML::DocumentFragment.parse("<div>a</div>")
+        assert_instance_of Nokogiri::HTML::DocumentFragment, fragment
+      end
+
+      def test_many_fragments
+        100.times { Nokogiri::HTML::DocumentFragment.new(@html) }
+      end
+
+      def test_subclass
+        klass = Class.new(Nokogiri::HTML::DocumentFragment)
+        fragment = klass.new(@html, "<div>a</div>")
+        assert_instance_of klass, fragment
+      end
+
+      def test_html_fragment
+        fragment = Nokogiri::HTML.fragment("<div>a</div>")
+        assert_equal "<div>a</div>", fragment.to_s
+      end
+
+      def test_html_fragment_has_outer_text
+        doc = "a<div>b</div>c"
+        fragment = Nokogiri::HTML::Document.new.fragment(doc)
+        if Nokogiri::VERSION_INFO['libxml']['loaded'] <= "2.6.16"
+          assert_equal "a<div>b</div><p>c</p>", fragment.to_s
+        else
+          assert_equal "a<div>b</div>c", fragment.to_s
+        end
+      end
+
+      def test_html_fragment_case_insensitivity
+        doc = "<crazyDiv>b</crazyDiv>"
+        fragment = Nokogiri::HTML::Document.new.fragment(doc)
+        assert_equal "<crazydiv>b</crazydiv>", fragment.to_s
+      end
+
+      def test_html_fragment_with_leading_whitespace
+        doc = "     <div>b</div>  "
+        fragment = Nokogiri::HTML::Document.new.fragment(doc)
+        assert_equal "<div>b</div>", fragment.to_s
+      end
+
+      def test_to_s
+        doc = "<span>foo<br></span><span>bar</span>"
+        fragment = Nokogiri::HTML::Document.new.fragment(doc)
+        assert_equal "<span>foo<br></span><span>bar</span>", fragment.to_s
+      end
+
+      def test_to_html
+        doc = "<span>foo<br></span><span>bar</span>"
+        fragment = Nokogiri::HTML::Document.new.fragment(doc)
+        assert_equal "<span>foo<br></span><span>bar</span>", fragment.to_html
+      end
+
+      def test_to_xhtml
+        doc = "<span>foo<br></span><span>bar</span>"
+        fragment = Nokogiri::HTML::Document.new.fragment(doc)
+        if Nokogiri::VERSION_INFO['libxml']['loaded'] >= "2.7.0"
+          assert_equal "<span>foo<br /></span><span>bar</span>", fragment.to_xhtml
+        else
+          assert_equal "<span>foo<br></span><span>bar</span>", fragment.to_xhtml
+        end
+      end
+
+      def test_to_xml
+        doc = "<span>foo<br></span><span>bar</span>"
+        fragment = Nokogiri::HTML::Document.new.fragment(doc)
+        assert_equal "<span>foo<br/></span><span>bar</span>", fragment.to_xml
+      end
+
+    end
+  end
+end

data/test/html/test_element_description.rb

@@ -0,0 +1,95 @@
+require File.expand_path(File.join(File.dirname(__FILE__), '..', "helper"))
+
+module Nokogiri
+  module HTML
+    class TestElementDescription < Nokogiri::TestCase
+      def test_fetch_nonexistent
+        assert_nil ElementDescription['foo']
+      end
+
+      def test_fetch_element_description
+        assert desc = ElementDescription['a']
+        assert_instance_of ElementDescription, desc
+      end
+
+      def test_name
+        assert_equal 'a', ElementDescription['a'].name
+      end
+
+      def test_implied_start_tag?
+        assert !ElementDescription['a'].implied_start_tag?
+      end
+
+      def test_implied_end_tag?
+        assert !ElementDescription['a'].implied_end_tag?
+        assert ElementDescription['p'].implied_end_tag?
+      end
+
+      def test_save_end_tag?
+        assert !ElementDescription['a'].save_end_tag?
+        assert ElementDescription['br'].save_end_tag?
+      end
+
+      def test_empty?
+        assert ElementDescription['br'].empty?
+        assert !ElementDescription['a'].empty?
+      end
+
+      def test_deprecated?
+        assert ElementDescription['applet'].deprecated?
+        assert !ElementDescription['br'].deprecated?
+      end
+
+      def test_inline?
+        assert ElementDescription['a'].inline?
+        assert !ElementDescription['div'].inline?
+      end
+
+      def test_block?
+        element = ElementDescription['a']
+        assert_equal(!element.inline?, element.block?)
+      end
+
+      def test_description
+        assert ElementDescription['a'].description
+      end
+
+      def test_subelements
+        sub_elements = ElementDescription['body'].sub_elements
+        assert_equal 61, sub_elements.length
+      end
+
+      def test_default_sub_element
+        assert_equal 'div', ElementDescription['body'].default_sub_element
+      end
+
+      def test_optional_attributes
+        attrs = ElementDescription['table'].optional_attributes
+        assert attrs
+        assert_equal 22, attrs.length
+      end
+
+      def test_deprecated_attributes
+        attrs = ElementDescription['table'].deprecated_attributes
+        assert attrs
+        assert_equal 2, attrs.length
+      end
+
+      def test_required_attributes
+        attrs = ElementDescription['table'].required_attributes
+        assert attrs
+        assert_equal 0, attrs.length
+      end
+
+      def test_inspect
+        desc = ElementDescription['input']
+        assert_match desc.name, desc.inspect
+      end
+
+      def test_to_s
+        desc = ElementDescription['input']
+        assert_match desc.name, desc.to_s
+      end
+    end
+  end
+end

data/test/html/test_node.rb

@@ -11,12 +11,46 @@ module Nokogiri
         <html>
           <head></head>
           <body>
-            <div>first</div>
+            <div class='baz'><a href="foo" class="bar">first</a></div>
           </body>
         </html>
         eohtml
       end
 
+      def test_description
+        assert desc = @html.at('a.bar').description
+        assert_equal 'a', desc.name
+      end
+
+      def test_add_next_sibling_with_empty_nodeset
+        assert_raises(ArgumentError) {
+          @html.at('a').add_next_sibling(@html.at('head').children)
+        }
+      end
+
+      def test_add_next_sibling_with_non_empty_nodeset
+        assert_raises(ArgumentError) {
+          @html.at('head').add_next_sibling(@html.at('div').children)
+        }
+      end
+
+      def test_ancestors_with_selector
+        assert node = @html.at('a.bar').child
+        assert list = node.ancestors('.baz')
+        assert_equal 1, list.length
+        assert_equal 'div', list.first.name
+      end
+
+      def test_css_matches?
+        assert node = @html.at('a.bar')
+        assert node.matches?('a.bar')
+      end
+
+      def test_xpath_matches?
+        assert node = @html.at('//a')
+        assert node.matches?('//a')
+      end
+
       def test_swap
         @html.at('div').swap('<a href="foo">bar</a>')
         a_tag = @html.css('a').first
@@ -24,6 +58,12 @@ module Nokogiri
         assert_equal 0, @html.css('div').length
       end
 
+      def test_swap_with_regex_characters
+        @html.at('div').swap('<a href="foo">ba)r</a>')
+        a_tag = @html.css('a').first
+        assert_equal 'ba)r', a_tag.text
+      end
+
       def test_attribute_decodes_entities
         node = @html.at('div')
         node['href'] = 'foo&bar'
@@ -39,6 +79,13 @@ module Nokogiri
         assert_equal 'some text', @html.at('//body').children[0].content.strip
       end
 
+      def test_fragment_handler_does_not_regurge_on_invalid_attributes
+        iframe = %Q{<iframe style="width: 0%; height: 0px" src="http://someurl" allowtransparency></iframe>}
+        assert_nothing_raised { @html.at('div').before(iframe) }
+        assert_nothing_raised { @html.at('div').after(iframe) }
+        assert_nothing_raised { @html.at('div').inner_html=(iframe) }
+      end
+
       def test_inner_html=
         assert div = @html.at('//div')
         div.inner_html = '<span>testing</span>'
@@ -63,6 +110,15 @@ module Nokogiri
         assert_equal 'foo', p_tag.parent['class']
       end
 
+      def test_fragment_serialization
+        fragment = Nokogiri::HTML.fragment("<div>foo</div>")
+        assert_equal "<div>foo</div>", fragment.serialize.chomp
+        assert_equal "<div>foo</div>", fragment.to_xml.chomp
+        assert_equal "<div>foo</div>", fragment.inner_html
+        assert_equal "<div>foo</div>", fragment.to_html
+        assert_equal "<div>foo</div>", fragment.to_s
+      end
+
       def test_after_will_append_text_nodes
         assert node = @html.at('//body/div')
         node.after "some text"
@@ -94,8 +150,15 @@ module Nokogiri
         EOH
         nokogiri = Nokogiri::HTML.parse(html)
 
-        assert_equal "<p>testparagraph\r\nfoobar</p>",
-          nokogiri.at("p").to_html.gsub(/ /, '')
+        if RUBY_PLATFORM =~ /java/
+          # NKF linebreak modes are not supported as of jruby 1.2
+          # see http://jira.codehaus.org/browse/JRUBY-3602 for status
+          assert_equal "<p>testparagraph\nfoobar</p>",
+            nokogiri.at("p").to_html.gsub(/ /, '')
+        else
+          assert_equal "<p>testparagraph\r\nfoobar</p>",
+            nokogiri.at("p").to_html.gsub(/ /, '')
+        end
       end
     end
   end

data/test/test_convert_xpath.rb

@@ -21,7 +21,7 @@ class TestConvertXPath < Nokogiri::TestCase
     blk ||= lambda {|j| j.first}
     assert_equal match, blk.call(@N.search(xpath)), "xpath result did not match"
     if HAS_HPRICOT
-      assert_equal match, blk.call(@H.search(hpath)), "hpath result did not match"
+      assert_equal match, blk.call(@H.search(hpath)).chomp, "hpath result did not match"
     end
     assert_equal [xpath], @NH.convert_to_xpath(hpath), "converted hpath did not match xpath"
   end

data/test/test_memory_leak.rb

@@ -1,15 +1,18 @@
 require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
 
 class TestMemoryLeak < Nokogiri::TestCase
-  def test_for_memory_leak
-    begin
-      #  we don't use Dike in any tests, but requiring it has side effects
-      #  that can create memory leaks, and that's what we're testing for.
-      require 'rubygems'
-      require 'dike' # do not remove!
-
-      count_start = count_object_space_documents
-      xml_data = <<-EOS
+
+  if ENV['NOKOGIRI_GC'] # turning these off by default for now
+
+    def test_for_memory_leak
+      begin
+        #  we don't use Dike in any tests, but requiring it has side effects
+        #  that can create memory leaks, and that's what we're testing for.
+        require 'rubygems'
+        require 'dike' # do not remove!
+
+        count_start = count_object_space_documents
+        xml_data = <<-EOS
         <test>
           <items>
             <item>abc</item>
@@ -18,17 +21,53 @@ class TestMemoryLeak < Nokogiri::TestCase
           <items>
         </test>
         EOS
-      20.times do
-        doc = Nokogiri::XML(xml_data)
-        doc.xpath("//item")
+        20.times do
+          doc = Nokogiri::XML(xml_data)
+          doc.xpath("//item")
+        end
+        2.times { GC.start }
+        count_end = count_object_space_documents
+        assert((count_end - count_start) <= 2, "memory leak detected")
+      rescue LoadError
+        puts "\ndike is not installed, skipping memory leak test"
       end
-      2.times { GC.start }
-      count_end = count_object_space_documents
-      assert((count_end - count_start) <= 2, "memory leak detected")
-    rescue LoadError
-      puts "\ndike is not installed, skipping memory leak test"
     end
-  end
+
+    if Nokogiri.ffi?
+      [ ['Node', 'p', nil],
+        ['CDATA', nil, 'content'],
+        ['Comment', nil, 'content'],
+        ['DocumentFragment', nil],
+        ['EntityReference', nil, 'p'],
+        ['ProcessingInstruction', nil, 'p', 'content'] ].each do |klass, *args|
+
+        define_method "test_for_leaked_#{klass}_nodes" do
+          Nokogiri::LibXML.expects(:xmlAddChild).at_least(1) # more than once shows we're GCing properly
+          10.times {
+            xml = Nokogiri::XML("<root></root>")
+            2.times { Nokogiri::XML.const_get(klass).new(*(args.collect{|arg| arg || xml})) }
+            GC.start
+          }
+          GC.start
+        end
+
+      end
+
+      def test_for_leaked_attr_nodes
+        Nokogiri::LibXML.expects(:xmlFreePropList).at_least(1) # more than once shows we're GCing properly
+        10.times {
+          xml = Nokogiri::XML("<root></root>")
+          2.times { Nokogiri::XML::Attr.new(xml, "p") }
+          GC.start
+        }
+        GC.start
+      end
+
+    end # if ffi
+
+  end # if NOKOGIRI_GC
+
+  private
 
   def count_object_space_documents
     count = 0

data/test/test_nokogiri.rb

@@ -1,12 +1,33 @@
 require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
 
 class TestNokogiri < Nokogiri::TestCase
-  def test_version
+  def test_versions
+    version_match = /\d+\.\d+\.\d+/
+    assert_match version_match, Nokogiri::VERSION
+    assert_match version_match, Nokogiri::LIBXML_VERSION
+
+    if defined?(FFI)
+      assert_equal 'ffi', Nokogiri::VERSION_INFO['libxml']['binding']
+      if RUBY_PLATFORM =~ /java/
+        assert_equal 'jruby', Nokogiri::VERSION_INFO['libxml']['platform']
+      else
+        assert_equal 'ruby', Nokogiri::VERSION_INFO['libxml']['platform']
+      end
+      assert Nokogiri.ffi?
+    else
+      assert_equal 'extension', Nokogiri::VERSION_INFO['libxml']['binding']
+
+      assert_match version_match, Nokogiri::VERSION_INFO['libxml']['compiled']
+      assert_equal Nokogiri::LIBXML_VERSION, Nokogiri::VERSION_INFO['libxml']['compiled']
+      assert ! Nokogiri.ffi?
+    end
+
+    assert_match version_match, Nokogiri::VERSION_INFO['libxml']['loaded']
     Nokogiri::LIBXML_PARSER_VERSION =~ /(\d)(\d{2})(\d{2})/
     major = $1.to_i
     minor = $2.to_i
     bug   = $3.to_i
-    assert_equal "#{major}.#{minor}.#{bug}", Nokogiri::LIBXML_VERSION
+    assert_equal "#{major}.#{minor}.#{bug}", Nokogiri::VERSION_INFO['libxml']['loaded']
   end
 
   def test_xml?
@@ -102,4 +123,5 @@ class TestNokogiri < Nokogiri::TestCase
     doc.slop!
     assert_equal 1, doc.decorators(Nokogiri::XML::Node).select { |d| d == Nokogiri::Decorators::Slop }.size
   end
+
 end

data/test/test_reader.rb

@@ -2,6 +2,83 @@
 require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
 
 class TestReader < Nokogiri::TestCase
+  def test_from_io_sets_io_as_source
+    io = StringIO.new(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+    reader = Nokogiri::XML::Reader.from_io(io)
+    assert_equal io, reader.source
+  end
+
+  def test_reader_takes_block
+    options = nil
+    Nokogiri::XML::Reader(File.read(XML_FILE), XML_FILE) do |cfg|
+      options = cfg
+      options.nonet.nowarning.dtdattr
+    end
+    assert options.nonet?
+    assert options.nowarning?
+    assert options.dtdattr?
+  end
+
+  def test_nil_raises
+    assert_raises(ArgumentError) {
+      Nokogiri::XML::Reader.from_memory(nil)
+    }
+    assert_raises(ArgumentError) {
+      Nokogiri::XML::Reader.from_io(nil)
+    }
+  end
+
+  def test_from_io
+    io = StringIO.new(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+    reader = Nokogiri::XML::Reader.from_io(io)
+    assert_equal false, reader.default?
+    assert_equal [false, false, false, false, false, false, false],
+      reader.map { |x| x.default? }
+  end
+
+  def test_io
+    io = StringIO.new(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+    reader = Nokogiri::XML::Reader(io)
+    assert_equal false, reader.default?
+    assert_equal [false, false, false, false, false, false, false],
+      reader.map { |x| x.default? }
+  end
+
+  def test_in_memory
+    reader = Nokogiri::XML::Reader(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+  end
+
+  def test_reader_holds_on_to_string
+    xml = <<-eoxml
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+    reader = Nokogiri::XML::Reader(xml)
+    if Nokogiri::VERSION_INFO['libxml']['binding'] == 'ffi'
+      assert_not_nil reader.source
+      assert reader.source.is_a?(FFI::MemoryPointer)
+    else
+      assert_equal xml, reader.source
+    end
+  end
+
   def test_default?
     reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
     <x xmlns:tenderlove='http://tenderlovemaking.com/'>