RubyGems - nokogiri-fitzsimmons - Versions diffs - 1.5.5.3-java - Mend

nokogiri-fitzsimmons 1.5.5.3-java

Files changed (297) hide show

data/.autotest +26 -0
data/.gemtest +0 -0
data/CHANGELOG.ja.rdoc +666 -0
data/CHANGELOG.rdoc +659 -0
data/C_CODING_STYLE.rdoc +33 -0
data/Manifest.txt +295 -0
data/README.ja.rdoc +106 -0
data/README.rdoc +178 -0
data/ROADMAP.md +86 -0
data/Rakefile +194 -0
data/STANDARD_RESPONSES.md +47 -0
data/Y_U_NO_GEMSPEC.md +155 -0
data/bin/nokogiri +63 -0
data/build_all +58 -0
data/ext/java/nokogiri/EncodingHandler.java +124 -0
data/ext/java/nokogiri/HtmlDocument.java +163 -0
data/ext/java/nokogiri/HtmlElementDescription.java +145 -0
data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
data/ext/java/nokogiri/HtmlSaxParserContext.java +259 -0
data/ext/java/nokogiri/NokogiriService.java +598 -0
data/ext/java/nokogiri/XmlAttr.java +190 -0
data/ext/java/nokogiri/XmlAttributeDecl.java +130 -0
data/ext/java/nokogiri/XmlCdata.java +84 -0
data/ext/java/nokogiri/XmlComment.java +95 -0
data/ext/java/nokogiri/XmlDocument.java +580 -0
data/ext/java/nokogiri/XmlDocumentFragment.java +234 -0
data/ext/java/nokogiri/XmlDtd.java +469 -0
data/ext/java/nokogiri/XmlElement.java +97 -0
data/ext/java/nokogiri/XmlElementContent.java +382 -0
data/ext/java/nokogiri/XmlElementDecl.java +152 -0
data/ext/java/nokogiri/XmlEntityDecl.java +162 -0
data/ext/java/nokogiri/XmlEntityReference.java +97 -0
data/ext/java/nokogiri/XmlNamespace.java +215 -0
data/ext/java/nokogiri/XmlNode.java +1534 -0
data/ext/java/nokogiri/XmlNodeSet.java +270 -0
data/ext/java/nokogiri/XmlProcessingInstruction.java +99 -0
data/ext/java/nokogiri/XmlReader.java +456 -0
data/ext/java/nokogiri/XmlRelaxng.java +144 -0
data/ext/java/nokogiri/XmlSaxParserContext.java +356 -0
data/ext/java/nokogiri/XmlSaxPushParser.java +215 -0
data/ext/java/nokogiri/XmlSchema.java +324 -0
data/ext/java/nokogiri/XmlSyntaxError.java +136 -0
data/ext/java/nokogiri/XmlText.java +119 -0
data/ext/java/nokogiri/XmlXpathContext.java +203 -0
data/ext/java/nokogiri/XsltStylesheet.java +360 -0
data/ext/java/nokogiri/internals/HtmlDomParserContext.java +243 -0
data/ext/java/nokogiri/internals/NokogiriDocumentCache.java +73 -0
data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +86 -0
data/ext/java/nokogiri/internals/NokogiriHandler.java +333 -0
data/ext/java/nokogiri/internals/NokogiriHelpers.java +800 -0
data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +163 -0
data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +130 -0
data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +74 -0
data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +79 -0
data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +141 -0
data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +73 -0
data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +67 -0
data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +87 -0
data/ext/java/nokogiri/internals/ParserContext.java +288 -0
data/ext/java/nokogiri/internals/ReaderNode.java +531 -0
data/ext/java/nokogiri/internals/SaveContextVisitor.java +775 -0
data/ext/java/nokogiri/internals/SchemaErrorHandler.java +76 -0
data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
data/ext/java/nokogiri/internals/XmlDomParser.java +105 -0
data/ext/java/nokogiri/internals/XmlDomParserContext.java +266 -0
data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
data/ext/java/nokogiri/internals/XsltExtensionFunction.java +72 -0
data/ext/nokogiri/depend +358 -0
data/ext/nokogiri/extconf.rb +142 -0
data/ext/nokogiri/html_document.c +170 -0
data/ext/nokogiri/html_document.h +10 -0
data/ext/nokogiri/html_element_description.c +276 -0
data/ext/nokogiri/html_element_description.h +10 -0
data/ext/nokogiri/html_entity_lookup.c +32 -0
data/ext/nokogiri/html_entity_lookup.h +8 -0
data/ext/nokogiri/html_sax_parser_context.c +116 -0
data/ext/nokogiri/html_sax_parser_context.h +11 -0
data/ext/nokogiri/html_sax_push_parser.c +87 -0
data/ext/nokogiri/html_sax_push_parser.h +9 -0
data/ext/nokogiri/nokogiri.c +133 -0
data/ext/nokogiri/nokogiri.h +160 -0
data/ext/nokogiri/xml_attr.c +94 -0
data/ext/nokogiri/xml_attr.h +9 -0
data/ext/nokogiri/xml_attribute_decl.c +70 -0
data/ext/nokogiri/xml_attribute_decl.h +9 -0
data/ext/nokogiri/xml_cdata.c +56 -0
data/ext/nokogiri/xml_cdata.h +9 -0
data/ext/nokogiri/xml_comment.c +54 -0
data/ext/nokogiri/xml_comment.h +9 -0
data/ext/nokogiri/xml_document.c +576 -0
data/ext/nokogiri/xml_document.h +23 -0
data/ext/nokogiri/xml_document_fragment.c +48 -0
data/ext/nokogiri/xml_document_fragment.h +10 -0
data/ext/nokogiri/xml_dtd.c +202 -0
data/ext/nokogiri/xml_dtd.h +10 -0
data/ext/nokogiri/xml_element_content.c +123 -0
data/ext/nokogiri/xml_element_content.h +10 -0
data/ext/nokogiri/xml_element_decl.c +69 -0
data/ext/nokogiri/xml_element_decl.h +9 -0
data/ext/nokogiri/xml_encoding_handler.c +79 -0
data/ext/nokogiri/xml_encoding_handler.h +8 -0
data/ext/nokogiri/xml_entity_decl.c +110 -0
data/ext/nokogiri/xml_entity_decl.h +10 -0
data/ext/nokogiri/xml_entity_reference.c +52 -0
data/ext/nokogiri/xml_entity_reference.h +9 -0
data/ext/nokogiri/xml_io.c +56 -0
data/ext/nokogiri/xml_io.h +11 -0
data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
data/ext/nokogiri/xml_namespace.c +78 -0
data/ext/nokogiri/xml_namespace.h +13 -0
data/ext/nokogiri/xml_node.c +1480 -0
data/ext/nokogiri/xml_node.h +13 -0
data/ext/nokogiri/xml_node_set.c +467 -0
data/ext/nokogiri/xml_node_set.h +14 -0
data/ext/nokogiri/xml_processing_instruction.c +56 -0
data/ext/nokogiri/xml_processing_instruction.h +9 -0
data/ext/nokogiri/xml_reader.c +684 -0
data/ext/nokogiri/xml_reader.h +10 -0
data/ext/nokogiri/xml_relax_ng.c +161 -0
data/ext/nokogiri/xml_relax_ng.h +9 -0
data/ext/nokogiri/xml_sax_parser.c +293 -0
data/ext/nokogiri/xml_sax_parser.h +39 -0
data/ext/nokogiri/xml_sax_parser_context.c +222 -0
data/ext/nokogiri/xml_sax_parser_context.h +10 -0
data/ext/nokogiri/xml_sax_push_parser.c +115 -0
data/ext/nokogiri/xml_sax_push_parser.h +9 -0
data/ext/nokogiri/xml_schema.c +205 -0
data/ext/nokogiri/xml_schema.h +9 -0
data/ext/nokogiri/xml_syntax_error.c +58 -0
data/ext/nokogiri/xml_syntax_error.h +13 -0
data/ext/nokogiri/xml_text.c +52 -0
data/ext/nokogiri/xml_text.h +9 -0
data/ext/nokogiri/xml_xpath_context.c +319 -0
data/ext/nokogiri/xml_xpath_context.h +10 -0
data/ext/nokogiri/xslt_stylesheet.c +270 -0
data/ext/nokogiri/xslt_stylesheet.h +14 -0
data/lib/isorelax.jar +0 -0
data/lib/jing.jar +0 -0
data/lib/nekodtd.jar +0 -0
data/lib/nekohtml.jar +0 -0
data/lib/nokogiri.rb +127 -0
data/lib/nokogiri/css.rb +27 -0
data/lib/nokogiri/css/node.rb +102 -0
data/lib/nokogiri/css/parser.rb +720 -0
data/lib/nokogiri/css/parser.y +258 -0
data/lib/nokogiri/css/parser_extras.rb +91 -0
data/lib/nokogiri/css/syntax_error.rb +7 -0
data/lib/nokogiri/css/tokenizer.rb +152 -0
data/lib/nokogiri/css/tokenizer.rex +55 -0
data/lib/nokogiri/css/xpath_visitor.rb +171 -0
data/lib/nokogiri/decorators/slop.rb +35 -0
data/lib/nokogiri/html.rb +37 -0
data/lib/nokogiri/html/builder.rb +35 -0
data/lib/nokogiri/html/document.rb +254 -0
data/lib/nokogiri/html/document_fragment.rb +41 -0
data/lib/nokogiri/html/element_description.rb +23 -0
data/lib/nokogiri/html/element_description_defaults.rb +671 -0
data/lib/nokogiri/html/entity_lookup.rb +13 -0
data/lib/nokogiri/html/sax/parser.rb +52 -0
data/lib/nokogiri/html/sax/parser_context.rb +16 -0
data/lib/nokogiri/html/sax/push_parser.rb +16 -0
data/lib/nokogiri/nokogiri.jar +0 -0
data/lib/nokogiri/syntax_error.rb +4 -0
data/lib/nokogiri/version.rb +88 -0
data/lib/nokogiri/xml.rb +73 -0
data/lib/nokogiri/xml/attr.rb +14 -0
data/lib/nokogiri/xml/attribute_decl.rb +18 -0
data/lib/nokogiri/xml/builder.rb +431 -0
data/lib/nokogiri/xml/cdata.rb +11 -0
data/lib/nokogiri/xml/character_data.rb +7 -0
data/lib/nokogiri/xml/document.rb +267 -0
data/lib/nokogiri/xml/document_fragment.rb +103 -0
data/lib/nokogiri/xml/dtd.rb +22 -0
data/lib/nokogiri/xml/element_content.rb +36 -0
data/lib/nokogiri/xml/element_decl.rb +13 -0
data/lib/nokogiri/xml/entity_decl.rb +19 -0
data/lib/nokogiri/xml/namespace.rb +13 -0
data/lib/nokogiri/xml/node.rb +946 -0
data/lib/nokogiri/xml/node/save_options.rb +61 -0
data/lib/nokogiri/xml/node_set.rb +357 -0
data/lib/nokogiri/xml/notation.rb +6 -0
data/lib/nokogiri/xml/parse_options.rb +98 -0
data/lib/nokogiri/xml/pp.rb +2 -0
data/lib/nokogiri/xml/pp/character_data.rb +18 -0
data/lib/nokogiri/xml/pp/node.rb +56 -0
data/lib/nokogiri/xml/processing_instruction.rb +8 -0
data/lib/nokogiri/xml/reader.rb +112 -0
data/lib/nokogiri/xml/relax_ng.rb +32 -0
data/lib/nokogiri/xml/sax.rb +4 -0
data/lib/nokogiri/xml/sax/document.rb +164 -0
data/lib/nokogiri/xml/sax/parser.rb +115 -0
data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
data/lib/nokogiri/xml/schema.rb +63 -0
data/lib/nokogiri/xml/syntax_error.rb +47 -0
data/lib/nokogiri/xml/text.rb +9 -0
data/lib/nokogiri/xml/xpath.rb +10 -0
data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
data/lib/nokogiri/xml/xpath_context.rb +16 -0
data/lib/nokogiri/xslt.rb +56 -0
data/lib/nokogiri/xslt/stylesheet.rb +25 -0
data/lib/xercesImpl.jar +0 -0
data/lib/xsd/xmlparser/nokogiri.rb +90 -0
data/tasks/cross_compile.rb +153 -0
data/tasks/nokogiri.org.rb +24 -0
data/tasks/test.rb +95 -0
data/test/css/test_nthiness.rb +159 -0
data/test/css/test_parser.rb +341 -0
data/test/css/test_tokenizer.rb +198 -0
data/test/css/test_xpath_visitor.rb +91 -0
data/test/decorators/test_slop.rb +16 -0
data/test/files/2ch.html +108 -0
data/test/files/address_book.rlx +12 -0
data/test/files/address_book.xml +10 -0
data/test/files/bar/bar.xsd +4 -0
data/test/files/dont_hurt_em_why.xml +422 -0
data/test/files/encoding.html +82 -0
data/test/files/encoding.xhtml +84 -0
data/test/files/exslt.xml +8 -0
data/test/files/exslt.xslt +35 -0
data/test/files/foo/foo.xsd +4 -0
data/test/files/metacharset.html +10 -0
data/test/files/noencoding.html +47 -0
data/test/files/po.xml +32 -0
data/test/files/po.xsd +66 -0
data/test/files/shift_jis.html +10 -0
data/test/files/shift_jis.xml +5 -0
data/test/files/snuggles.xml +3 -0
data/test/files/staff.dtd +10 -0
data/test/files/staff.xml +59 -0
data/test/files/staff.xslt +32 -0
data/test/files/tlm.html +850 -0
data/test/files/to_be_xincluded.xml +2 -0
data/test/files/valid_bar.xml +2 -0
data/test/files/xinclude.xml +4 -0
data/test/helper.rb +147 -0
data/test/html/sax/test_parser.rb +138 -0
data/test/html/sax/test_parser_context.rb +46 -0
data/test/html/test_builder.rb +164 -0
data/test/html/test_document.rb +529 -0
data/test/html/test_document_encoding.rb +138 -0
data/test/html/test_document_fragment.rb +254 -0
data/test/html/test_element_description.rb +100 -0
data/test/html/test_named_characters.rb +14 -0
data/test/html/test_node.rb +188 -0
data/test/html/test_node_encoding.rb +27 -0
data/test/test_convert_xpath.rb +135 -0
data/test/test_css_cache.rb +45 -0
data/test/test_encoding_handler.rb +46 -0
data/test/test_memory_leak.rb +152 -0
data/test/test_nokogiri.rb +132 -0
data/test/test_reader.rb +488 -0
data/test/test_soap4r_sax.rb +52 -0
data/test/test_xslt_transforms.rb +254 -0
data/test/xml/node/test_save_options.rb +28 -0
data/test/xml/node/test_subclass.rb +44 -0
data/test/xml/sax/test_parser.rb +338 -0
data/test/xml/sax/test_parser_context.rb +106 -0
data/test/xml/sax/test_push_parser.rb +157 -0
data/test/xml/test_attr.rb +64 -0
data/test/xml/test_attribute_decl.rb +86 -0
data/test/xml/test_builder.rb +248 -0
data/test/xml/test_c14n.rb +151 -0
data/test/xml/test_cdata.rb +48 -0
data/test/xml/test_comment.rb +29 -0
data/test/xml/test_document.rb +742 -0
data/test/xml/test_document_encoding.rb +28 -0
data/test/xml/test_document_fragment.rb +216 -0
data/test/xml/test_dtd.rb +103 -0
data/test/xml/test_dtd_encoding.rb +33 -0
data/test/xml/test_element_content.rb +56 -0
data/test/xml/test_element_decl.rb +73 -0
data/test/xml/test_entity_decl.rb +122 -0
data/test/xml/test_entity_reference.rb +235 -0
data/test/xml/test_namespace.rb +75 -0
data/test/xml/test_node.rb +1029 -0
data/test/xml/test_node_attributes.rb +53 -0
data/test/xml/test_node_encoding.rb +107 -0
data/test/xml/test_node_inheritance.rb +32 -0
data/test/xml/test_node_reparenting.rb +374 -0
data/test/xml/test_node_set.rb +755 -0
data/test/xml/test_parse_options.rb +64 -0
data/test/xml/test_processing_instruction.rb +30 -0
data/test/xml/test_reader_encoding.rb +142 -0
data/test/xml/test_relax_ng.rb +60 -0
data/test/xml/test_schema.rb +94 -0
data/test/xml/test_syntax_error.rb +12 -0
data/test/xml/test_text.rb +45 -0
data/test/xml/test_unparented_node.rb +413 -0
data/test/xml/test_xinclude.rb +83 -0
data/test/xml/test_xpath.rb +295 -0
data/test/xslt/test_custom_functions.rb +129 -0
data/test/xslt/test_exception_handling.rb +37 -0
data/test_all +84 -0
metadata +571 -0

data/test/html/test_named_characters.rb ADDED

@@ -0,0 +1,14 @@
+require "helper"
+module Nokogiri
+  module HTML
+    class TestNamedCharacters < Nokogiri::TestCase
+      def test_named_character
+        copy = NamedCharacters.get('copy')
+        assert_equal 169, NamedCharacters['copy']
+        assert_equal copy.value, NamedCharacters['copy']
+        assert copy.description
+      end
+    end
+  end
+end

data/test/html/test_node.rb ADDED

@@ -0,0 +1,188 @@
+require "helper"
+require 'nkf'
+module Nokogiri
+  module HTML
+    class TestNode < Nokogiri::TestCase
+      def setup
+        super
+        @html = Nokogiri::HTML(<<-eohtml)
+        <html>
+          <head></head>
+          <body>
+            <div class='baz'><a href="foo" class="bar">first</a></div>
+          </body>
+        </html>
+        eohtml
+      end
+      def test_to_a
+        assert_equal [['class', 'bar'], ['href', 'foo']],@html.at('a').to_a.sort
+      end
+      def test_attr
+        node = @html.at('div.baz')
+        assert_equal node['class'], node.attr('class')
+      end
+      def test_get_attribute
+        element = @html.at('div')
+        assert_equal 'baz', element.get_attribute('class')
+        assert_equal 'baz', element['class']
+        element['href'] = "javascript:alert(\"AGGA-KA-BOO!\")"
+        assert_match(/%22AGGA-KA-BOO!%22/, element.to_html)
+      end
+      def test_css_path_round_trip
+        doc = Nokogiri::HTML(File.read(HTML_FILE))
+        %w{ #header small div[2] div.post body }.each do |css_sel|
+          ele = doc.at css_sel
+          assert_equal ele, doc.at(ele.css_path), ele.css_path
+        end
+      end
+      def test_path_round_trip
+        doc = Nokogiri::HTML(File.read(HTML_FILE))
+        %w{ #header small div[2] div.post body }.each do |css_sel|
+          ele = doc.at css_sel
+          assert_equal ele, doc.at(ele.path), ele.path
+        end
+      end
+      def test_append_with_document
+        assert_raises(ArgumentError) do
+          @html.root << Nokogiri::HTML::Document.new
+        end
+      end
+      ###
+      # Make sure a document that doesn't declare a meta encoding returns
+      # nil.
+      def test_meta_encoding
+        assert_nil @html.meta_encoding
+      end
+      def test_description
+        assert desc = @html.at('a.bar').description
+        assert_equal 'a', desc.name
+      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_matches_inside_fragment
+        fragment = DocumentFragment.new @html
+        fragment << XML::Node.new('a', @html)
+        a = fragment.children.last
+        assert a.matches?('a'), 'a should match'
+      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_unlink_then_swap
+        node = @html.at('a')
+        node.unlink
+        another_node = @html.at('div')
+        assert another_node, 'should have a node'
+        # This used to segv
+        assert node.add_previous_sibling another_node
+      end
+      def test_swap
+        @html.at('div').swap('<a href="foo">bar</a>')
+        a_tag = @html.css('a').first
+        assert_equal 'body', a_tag.parent.name
+        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'
+        assert_equal 'foo&bar', node['href']
+        node['href'] += '&baz'
+        assert_equal 'foo&bar&baz', node['href']
+      end
+      def test_parse_config_option
+        node = @html.at('div')
+        options = nil
+        node.parse("<div></div>") do |config|
+          options = config
+        end
+        assert_equal Nokogiri::XML::ParseOptions::DEFAULT_HTML, options.to_i
+      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 @html.at('div').fragment(iframe)
+      end
+      def test_fragment
+        fragment = @html.fragment(<<-eohtml)
+          hello
+          <div class="foo">
+            <p>bar</p>
+          </div>
+          world
+        eohtml
+        assert_match(/^hello/, fragment.inner_html.strip)
+        assert_equal 3, fragment.children.length
+        assert p_tag = fragment.css('p').first
+        assert_equal 'div', p_tag.parent.name
+        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_to_html_does_not_contain_entities
+        return unless defined?(NKF) # NKF is not implemented on Rubinius as of 2009-11-23
+        html = NKF.nkf("-e --msdos", <<-EOH)
+        <html><body>
+        <p> test paragraph
+        foo bar </p>
+        </body></html>
+        EOH
+        nokogiri = Nokogiri::HTML.parse(html)
+        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
+end

data/test/html/test_node_encoding.rb ADDED

@@ -0,0 +1,27 @@
+# -*- coding: utf-8 -*-
+require "helper"
+module Nokogiri
+  module HTML
+    if RUBY_VERSION =~ /^1\.9/
+      class TestNodeEncoding < Nokogiri::TestCase
+        def test_inner_html
+          doc = Nokogiri::HTML File.open(SHIFT_JIS_HTML, 'rb')
+          hello = "こんにちは"
+          contents = doc.at('h2').inner_html
+          assert_equal doc.encoding, contents.encoding.name
+          assert_match hello.encode('Shift_JIS'), contents
+          contents = doc.at('h2').inner_html(:encoding => 'UTF-8')
+          assert_match hello, contents
+          doc.encoding = 'UTF-8'
+          contents = doc.at('h2').inner_html
+          assert_match hello, contents
+        end
+      end
+    end
+  end
+end

data/test/test_convert_xpath.rb ADDED

@@ -0,0 +1,135 @@
+require "helper"
+class TestConvertXPath < Nokogiri::TestCase
+  def setup
+    super
+    @N = Nokogiri(File.read(HTML_FILE))
+  end
+  def assert_syntactical_equivalence(hpath, xpath, match, &blk)
+    blk ||= lambda {|j| j.first}
+    assert_equal match, blk.call(@N.search(xpath)), "xpath result did not match"
+  end
+  def test_child_tag
+    assert_syntactical_equivalence("h1[a]", ".//h1[child::a]", "Tender Lovemaking") do |j|
+      j.inner_text
+    end
+  end
+  def test_child_tag_equals
+    assert_syntactical_equivalence("h1[a='Tender Lovemaking']", ".//h1[child::a = 'Tender Lovemaking']", "Tender Lovemaking") do |j|
+      j.inner_text
+    end
+  end
+  def test_filter_contains
+    assert_syntactical_equivalence("title:contains('Tender')", ".//title[contains(., 'Tender')]",
+                                   "Tender Lovemaking  ") do |j|
+      j.inner_text
+    end
+  end
+  def test_filter_comment
+    assert_syntactical_equivalence("div comment()[2]", ".//div//comment()[position() = 2]", "<!-- end of header -->") do |j|
+      j.first.to_s
+    end
+  end
+  def test_filter_text
+    assert_syntactical_equivalence("a[text()]", ".//a[normalize-space(child::text())]", "<a href=\"http://tenderlovemaking.com\">Tender Lovemaking</a>") do |j|
+      j.first.to_s
+    end
+    assert_syntactical_equivalence("a[text()='Tender Lovemaking']", ".//a[normalize-space(child::text()) = 'Tender Lovemaking']", "<a href=\"http://tenderlovemaking.com\">Tender Lovemaking</a>") do |j|
+      j.first.to_s
+    end
+    assert_syntactical_equivalence("a/text()", ".//a/child::text()", "Tender Lovemaking") do |j|
+      j.first.to_s
+    end
+    assert_syntactical_equivalence("h2//a[text()!='Back Home!']", ".//h2//a[normalize-space(child::text()) != 'Back Home!']", "Meow meow meow meow meow") do |j|
+      j.first.inner_text
+    end
+  end
+  def test_filter_by_attr
+    assert_syntactical_equivalence("a[@href='http://blog.geminigeek.com/wordpress-theme']",
+                                   ".//a[@href = 'http://blog.geminigeek.com/wordpress-theme']",
+                                   "http://blog.geminigeek.com/wordpress-theme") do |j|
+      j.first["href"]
+    end
+  end
+  def test_css_id
+    assert_syntactical_equivalence("#linkcat-7", ".//*[@id = 'linkcat-7']", "linkcat-7") do |j|
+      j.first["id"]
+    end
+    assert_syntactical_equivalence("li#linkcat-7", ".//li[@id = 'linkcat-7']", "linkcat-7") do |j|
+      j.first["id"]
+    end
+  end
+  def test_css_class
+    assert_syntactical_equivalence(".cat-item-15", ".//*[contains(concat(' ', @class, ' '), ' cat-item-15 ')]",
+                                   "cat-item cat-item-15") do |j|
+      j.first["class"]
+    end
+    assert_syntactical_equivalence("li.cat-item-15", ".//li[contains(concat(' ', @class, ' '), ' cat-item-15 ')]",
+                                   "cat-item cat-item-15") do |j|
+      j.first["class"]
+    end
+  end
+  def test_css_tags
+    assert_syntactical_equivalence("div li a", ".//div//li//a", "http://brobinius.org/") do |j|
+      j.first.inner_text
+    end
+    assert_syntactical_equivalence("div li > a", ".//div//li/a", "http://brobinius.org/") do |j|
+      j.first.inner_text
+    end
+    assert_syntactical_equivalence("h1 ~ small", ".//small[preceding-sibling::h1]", "The act of making love, tenderly.") do |j|
+      j.first.inner_text
+    end
+    assert_syntactical_equivalence("h1 ~ small", ".//small[preceding-sibling::h1]", "The act of making love, tenderly.") do |j|
+      j.first.inner_text
+    end
+  end
+  def test_positional
+    assert_syntactical_equivalence("div/div:first()", ".//div/div[position() = 1]", "\r\nTender Lovemaking\r\nThe act of making love, tenderly.\r\n".gsub(/[\r\n]/, '')) do |j|
+      j.first.inner_text.gsub(/[\r\n]/, '')
+    end
+    assert_syntactical_equivalence("div/div:first", ".//div/div[position() = 1]", "\r\nTender Lovemaking\r\nThe act of making love, tenderly.\r\n".gsub(/[\r\n]/, '')) do |j|
+      j.first.inner_text.gsub(/[\r\n]/, '')
+    end
+    assert_syntactical_equivalence("div//a:last()", ".//div//a[position() = last()]", "Wordpress") do |j|
+      j.last.inner_text
+    end
+    assert_syntactical_equivalence("div//a:last", ".//div//a[position() = last()]", "Wordpress") do |j|
+      j.last.inner_text
+    end
+  end
+  def test_multiple_filters
+    assert_syntactical_equivalence("a[@rel='bookmark'][1]", ".//a[@rel = 'bookmark' and position() = 1]", "Back Home!") do |j|
+      j.first.inner_text
+    end
+  end
+# TODO:
+#       doc/'title ~ link' -> links that are siblings of title
+#       doc/'p[@class~="final"]' -> class includes string (whitespacy)
+#       doc/'p[text()*="final"]' -> class includes string (index) (broken: always returns true?)
+#       doc/'p[text()$="final"]' -> /final$/
+#       doc/'p[text()|="final"]' -> /^final$/
+#       doc/'p[text()^="final"]' -> string starts with 'final
+#       nth_first
+#       nth_last
+#       even
+#       odd
+#       first-child, nth-child, last-child, nth-last-child, nth-last-of-type
+#       only-of-type, only-child
+#       parent
+#       empty
+#       root
+end

data/test/test_css_cache.rb ADDED

@@ -0,0 +1,45 @@
+require "helper"
+class TestCssCache < Nokogiri::TestCase
+  def setup
+    super
+    @css = "a1 > b2 > c3"
+    @parse_result = Nokogiri::CSS.parse(@css)
+    @to_xpath_result = @parse_result.map {|ast| ast.to_xpath}
+    Nokogiri::CSS::Parser.class_eval do
+      class << @cache
+        alias :old_bracket :[]
+        attr_reader :count
+        def [](key)
+          @count ||= 0
+          @count += 1
+          old_bracket(key)
+        end
+      end
+    end
+    assert Nokogiri::CSS::Parser.cache_on?
+  end
+  def teardown
+    Nokogiri::CSS::Parser.clear_cache
+    Nokogiri::CSS::Parser.set_cache true
+  end
+  [ false, true ].each do |cache_setting|
+    define_method "test_css_cache_#{cache_setting ? "true" : "false"}" do
+      times = cache_setting ? 4 : nil
+      Nokogiri::CSS::Parser.set_cache cache_setting
+      Nokogiri::CSS.xpath_for(@css)
+      Nokogiri::CSS.xpath_for(@css)
+      Nokogiri::CSS::Parser.new.xpath_for(@css)
+      Nokogiri::CSS::Parser.new.xpath_for(@css)
+      assert_equal(times, Nokogiri::CSS::Parser.class_eval { @cache.count })
+    end
+  end
+end

data/test/test_encoding_handler.rb ADDED

@@ -0,0 +1,46 @@
+# -*- coding: utf-8 -*-
+require "helper"
+class TestEncodingHandler < Nokogiri::TestCase
+  def teardown
+    Nokogiri::EncodingHandler.clear_aliases!
+  end
+  def test_get
+    assert_not_nil Nokogiri::EncodingHandler['UTF-8']
+    assert_nil Nokogiri::EncodingHandler['alsdkjfhaldskjfh']
+  end
+  def test_name
+    eh = Nokogiri::EncodingHandler['UTF-8']
+    assert_equal "UTF-8", eh.name
+  end
+  def test_alias
+    Nokogiri::EncodingHandler.alias('UTF-8', 'UTF-18')
+    assert_equal 'UTF-8', Nokogiri::EncodingHandler['UTF-18'].name
+  end
+  def test_cleanup_aliases
+    assert_nil Nokogiri::EncodingHandler['UTF-9']
+    Nokogiri::EncodingHandler.alias('UTF-8', 'UTF-9')
+    assert_not_nil Nokogiri::EncodingHandler['UTF-9']
+    Nokogiri::EncodingHandler.clear_aliases!
+    assert_nil Nokogiri::EncodingHandler['UTF-9']
+  end
+  def test_delete
+    assert_nil Nokogiri::EncodingHandler['UTF-9']
+    Nokogiri::EncodingHandler.alias('UTF-8', 'UTF-9')
+    assert_not_nil Nokogiri::EncodingHandler['UTF-9']
+    Nokogiri::EncodingHandler.delete 'UTF-9'
+    assert_nil Nokogiri::EncodingHandler['UTF-9']
+  end
+  def test_delete_non_existent
+    assert_nil Nokogiri::EncodingHandler.delete('UTF-9')
+  end
+end