RubyGems - nokogiri - Versions diffs - 1.2.3-x86-mswin32-60 → 1.4.5-x86-mswin32-60 - Mend

nokogiri 1.2.3-x86-mswin32-60 → 1.4.5-x86-mswin32-60

Potentially problematic release.

This version of nokogiri might be problematic. Click here for more details.

Files changed (319) hide show

data/.autotest +18 -7
data/.gemtest +0 -0
data/CHANGELOG.ja.rdoc +297 -3
data/CHANGELOG.rdoc +289 -0
data/Manifest.txt +148 -37
data/README.ja.rdoc +20 -20
data/README.rdoc +53 -22
data/Rakefile +127 -211
data/bin/nokogiri +54 -0
data/ext/nokogiri/depend +358 -0
data/ext/nokogiri/extconf.rb +89 -54
data/ext/nokogiri/html_document.c +34 -27
data/ext/nokogiri/html_document.h +1 -1
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 +7 -5
data/ext/nokogiri/html_entity_lookup.h +1 -1
data/ext/nokogiri/html_sax_parser_context.c +94 -0
data/ext/nokogiri/html_sax_parser_context.h +11 -0
data/ext/nokogiri/{native.c → nokogiri.c} +31 -7
data/ext/nokogiri/{native.h → nokogiri.h} +68 -41
data/ext/nokogiri/xml_attr.c +20 -9
data/ext/nokogiri/xml_attr.h +1 -1
data/ext/nokogiri/xml_attribute_decl.c +70 -0
data/ext/nokogiri/xml_attribute_decl.h +9 -0
data/ext/nokogiri/xml_cdata.c +21 -9
data/ext/nokogiri/xml_cdata.h +1 -1
data/ext/nokogiri/xml_comment.c +18 -6
data/ext/nokogiri/xml_comment.h +1 -1
data/ext/nokogiri/xml_document.c +247 -68
data/ext/nokogiri/xml_document.h +5 -3
data/ext/nokogiri/xml_document_fragment.c +15 -7
data/ext/nokogiri/xml_document_fragment.h +1 -1
data/ext/nokogiri/xml_dtd.c +110 -10
data/ext/nokogiri/xml_dtd.h +3 -1
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 +16 -5
data/ext/nokogiri/xml_entity_reference.h +1 -1
data/ext/nokogiri/xml_io.c +40 -8
data/ext/nokogiri/xml_io.h +2 -1
data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
data/ext/nokogiri/xml_namespace.c +84 -0
data/ext/nokogiri/xml_namespace.h +13 -0
data/ext/nokogiri/xml_node.c +782 -225
data/ext/nokogiri/xml_node.h +2 -4
data/ext/nokogiri/xml_node_set.c +253 -34
data/ext/nokogiri/xml_node_set.h +2 -2
data/ext/nokogiri/xml_processing_instruction.c +17 -5
data/ext/nokogiri/xml_processing_instruction.h +1 -1
data/ext/nokogiri/xml_reader.c +277 -85
data/ext/nokogiri/xml_reader.h +1 -1
data/ext/nokogiri/xml_relax_ng.c +168 -0
data/ext/nokogiri/xml_relax_ng.h +9 -0
data/ext/nokogiri/xml_sax_parser.c +183 -111
data/ext/nokogiri/xml_sax_parser.h +30 -1
data/ext/nokogiri/xml_sax_parser_context.c +199 -0
data/ext/nokogiri/xml_sax_parser_context.h +10 -0
data/ext/nokogiri/xml_sax_push_parser.c +42 -12
data/ext/nokogiri/xml_sax_push_parser.h +1 -1
data/ext/nokogiri/xml_schema.c +205 -0
data/ext/nokogiri/xml_schema.h +9 -0
data/ext/nokogiri/xml_syntax_error.c +28 -173
data/ext/nokogiri/xml_syntax_error.h +2 -1
data/ext/nokogiri/xml_text.c +16 -6
data/ext/nokogiri/xml_text.h +1 -1
data/ext/nokogiri/xml_xpath_context.c +104 -47
data/ext/nokogiri/xml_xpath_context.h +1 -1
data/ext/nokogiri/xslt_stylesheet.c +161 -19
data/ext/nokogiri/xslt_stylesheet.h +1 -1
data/lib/nokogiri.rb +47 -8
data/lib/nokogiri/1.8/nokogiri.so +0 -0
data/lib/nokogiri/1.9/nokogiri.so +0 -0
data/lib/nokogiri/css.rb +6 -3
data/lib/nokogiri/css/node.rb +14 -12
data/lib/nokogiri/css/parser.rb +665 -62
data/lib/nokogiri/css/parser.y +20 -10
data/lib/nokogiri/css/parser_extras.rb +91 -0
data/lib/nokogiri/css/tokenizer.rb +148 -5
data/lib/nokogiri/css/tokenizer.rex +10 -9
data/lib/nokogiri/css/xpath_visitor.rb +47 -44
data/lib/nokogiri/decorators/slop.rb +8 -4
data/lib/nokogiri/ffi/encoding_handler.rb +42 -0
data/lib/nokogiri/ffi/html/document.rb +28 -0
data/lib/nokogiri/ffi/html/element_description.rb +81 -0
data/lib/nokogiri/ffi/html/entity_lookup.rb +16 -0
data/lib/nokogiri/ffi/html/sax/parser_context.rb +38 -0
data/lib/nokogiri/ffi/io_callbacks.rb +42 -0
data/lib/nokogiri/ffi/libxml.rb +420 -0
data/lib/nokogiri/ffi/structs/common_node.rb +38 -0
data/lib/nokogiri/ffi/structs/html_elem_desc.rb +24 -0
data/lib/nokogiri/ffi/structs/html_entity_desc.rb +13 -0
data/lib/nokogiri/ffi/structs/xml_alloc.rb +16 -0
data/lib/nokogiri/ffi/structs/xml_attr.rb +20 -0
data/lib/nokogiri/ffi/structs/xml_attribute.rb +27 -0
data/lib/nokogiri/ffi/structs/xml_buffer.rb +16 -0
data/lib/nokogiri/ffi/structs/xml_char_encoding_handler.rb +11 -0
data/lib/nokogiri/ffi/structs/xml_document.rb +117 -0
data/lib/nokogiri/ffi/structs/xml_dtd.rb +28 -0
data/lib/nokogiri/ffi/structs/xml_element.rb +26 -0
data/lib/nokogiri/ffi/structs/xml_element_content.rb +17 -0
data/lib/nokogiri/ffi/structs/xml_entity.rb +32 -0
data/lib/nokogiri/ffi/structs/xml_enumeration.rb +12 -0
data/lib/nokogiri/ffi/structs/xml_node.rb +28 -0
data/lib/nokogiri/ffi/structs/xml_node_set.rb +53 -0
data/lib/nokogiri/ffi/structs/xml_notation.rb +11 -0
data/lib/nokogiri/ffi/structs/xml_ns.rb +15 -0
data/lib/nokogiri/ffi/structs/xml_parser_context.rb +20 -0
data/lib/nokogiri/ffi/structs/xml_parser_input.rb +19 -0
data/lib/nokogiri/ffi/structs/xml_relax_ng.rb +14 -0
data/lib/nokogiri/ffi/structs/xml_sax_handler.rb +51 -0
data/lib/nokogiri/ffi/structs/xml_sax_push_parser_context.rb +124 -0
data/lib/nokogiri/ffi/structs/xml_schema.rb +13 -0
data/lib/nokogiri/ffi/structs/xml_syntax_error.rb +31 -0
data/lib/nokogiri/ffi/structs/xml_text_reader.rb +12 -0
data/lib/nokogiri/ffi/structs/xml_xpath_context.rb +38 -0
data/lib/nokogiri/ffi/structs/xml_xpath_object.rb +35 -0
data/lib/nokogiri/ffi/structs/xml_xpath_parser_context.rb +20 -0
data/lib/nokogiri/ffi/structs/xslt_stylesheet.rb +13 -0
data/lib/nokogiri/ffi/weak_bucket.rb +40 -0
data/lib/nokogiri/ffi/xml/attr.rb +41 -0
data/lib/nokogiri/ffi/xml/attribute_decl.rb +27 -0
data/lib/nokogiri/ffi/xml/cdata.rb +19 -0
data/lib/nokogiri/ffi/xml/comment.rb +18 -0
data/lib/nokogiri/ffi/xml/document.rb +174 -0
data/lib/nokogiri/ffi/xml/document_fragment.rb +21 -0
data/lib/nokogiri/ffi/xml/dtd.rb +67 -0
data/lib/nokogiri/ffi/xml/element_content.rb +43 -0
data/lib/nokogiri/ffi/xml/element_decl.rb +19 -0
data/lib/nokogiri/ffi/xml/entity_decl.rb +36 -0
data/lib/nokogiri/ffi/xml/entity_reference.rb +19 -0
data/lib/nokogiri/ffi/xml/namespace.rb +44 -0
data/lib/nokogiri/ffi/xml/node.rb +559 -0
data/lib/nokogiri/ffi/xml/node_set.rb +150 -0
data/lib/nokogiri/ffi/xml/processing_instruction.rb +20 -0
data/lib/nokogiri/ffi/xml/reader.rb +236 -0
data/lib/nokogiri/ffi/xml/relax_ng.rb +85 -0
data/lib/nokogiri/ffi/xml/sax/parser.rb +143 -0
data/lib/nokogiri/ffi/xml/sax/parser_context.rb +79 -0
data/lib/nokogiri/ffi/xml/sax/push_parser.rb +51 -0
data/lib/nokogiri/ffi/xml/schema.rb +109 -0
data/lib/nokogiri/ffi/xml/syntax_error.rb +98 -0
data/lib/nokogiri/ffi/xml/text.rb +18 -0
data/lib/nokogiri/ffi/xml/xpath.rb +9 -0
data/lib/nokogiri/ffi/xml/xpath_context.rb +153 -0
data/lib/nokogiri/ffi/xslt/stylesheet.rb +77 -0
data/lib/nokogiri/html.rb +13 -47
data/lib/nokogiri/html/builder.rb +27 -1
data/lib/nokogiri/html/document.rb +201 -7
data/lib/nokogiri/html/document_fragment.rb +41 -0
data/lib/nokogiri/html/element_description.rb +23 -0
data/lib/nokogiri/html/entity_lookup.rb +2 -0
data/lib/nokogiri/html/sax/parser.rb +34 -3
data/lib/nokogiri/html/sax/parser_context.rb +16 -0
data/lib/nokogiri/nokogiri.rb +1 -0
data/lib/nokogiri/version.rb +40 -1
data/lib/nokogiri/version_warning.rb +14 -0
data/lib/nokogiri/xml.rb +32 -53
data/lib/nokogiri/xml/attr.rb +5 -0
data/lib/nokogiri/xml/attribute_decl.rb +18 -0
data/lib/nokogiri/xml/builder.rb +349 -29
data/lib/nokogiri/xml/cdata.rb +3 -1
data/lib/nokogiri/xml/character_data.rb +7 -0
data/lib/nokogiri/xml/document.rb +166 -14
data/lib/nokogiri/xml/document_fragment.rb +76 -1
data/lib/nokogiri/xml/dtd.rb +16 -3
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 +561 -166
data/lib/nokogiri/xml/node/save_options.rb +22 -2
data/lib/nokogiri/xml/node_set.rb +202 -40
data/lib/nokogiri/xml/parse_options.rb +93 -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 +2 -0
data/lib/nokogiri/xml/reader.rb +93 -8
data/lib/nokogiri/xml/relax_ng.rb +32 -0
data/lib/nokogiri/xml/sax.rb +1 -7
data/lib/nokogiri/xml/sax/document.rb +107 -2
data/lib/nokogiri/xml/sax/parser.rb +57 -7
data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
data/lib/nokogiri/xml/sax/push_parser.rb +13 -1
data/lib/nokogiri/xml/schema.rb +63 -0
data/lib/nokogiri/xml/syntax_error.rb +25 -1
data/lib/nokogiri/xml/text.rb +4 -1
data/lib/nokogiri/xml/xpath.rb +1 -1
data/lib/nokogiri/xml/xpath/syntax_error.rb +3 -0
data/lib/nokogiri/xml/xpath_context.rb +2 -0
data/lib/nokogiri/xslt.rb +26 -2
data/lib/nokogiri/xslt/stylesheet.rb +19 -0
data/lib/xsd/xmlparser/nokogiri.rb +45 -9
data/tasks/cross_compile.rb +173 -0
data/tasks/test.rb +25 -69
data/test/css/test_nthiness.rb +3 -4
data/test/css/test_parser.rb +75 -20
data/test/css/test_tokenizer.rb +23 -1
data/test/css/test_xpath_visitor.rb +10 -1
data/test/decorators/test_slop.rb +16 -0
data/test/ffi/test_document.rb +35 -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/encoding.html +82 -0
data/test/files/encoding.xhtml +84 -0
data/test/files/foo/foo.xsd +4 -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/valid_bar.xml +2 -0
data/test/helper.rb +101 -23
data/test/html/sax/test_parser.rb +81 -2
data/test/html/sax/test_parser_context.rb +48 -0
data/test/html/test_builder.rb +39 -8
data/test/html/test_document.rb +186 -23
data/test/html/test_document_encoding.rb +78 -1
data/test/html/test_document_fragment.rb +253 -0
data/test/html/test_element_description.rb +98 -0
data/test/html/test_named_characters.rb +1 -1
data/test/html/test_node.rb +124 -36
data/test/html/test_node_encoding.rb +27 -0
data/test/test_convert_xpath.rb +1 -52
data/test/test_css_cache.rb +2 -13
data/test/test_encoding_handler.rb +46 -0
data/test/test_memory_leak.rb +88 -19
data/test/test_nokogiri.rb +38 -5
data/test/test_reader.rb +188 -6
data/test/test_soap4r_sax.rb +52 -0
data/test/test_xslt_transforms.rb +183 -83
data/test/xml/node/test_save_options.rb +1 -1
data/test/xml/node/test_subclass.rb +44 -0
data/test/xml/sax/test_parser.rb +175 -4
data/test/xml/sax/test_parser_context.rb +113 -0
data/test/xml/sax/test_push_parser.rb +90 -2
data/test/xml/test_attr.rb +35 -1
data/test/xml/test_attribute_decl.rb +82 -0
data/test/xml/test_builder.rb +186 -1
data/test/xml/test_cdata.rb +32 -1
data/test/xml/test_comment.rb +13 -1
data/test/xml/test_document.rb +415 -43
data/test/xml/test_document_encoding.rb +1 -1
data/test/xml/test_document_fragment.rb +173 -5
data/test/xml/test_dtd.rb +61 -6
data/test/xml/test_dtd_encoding.rb +3 -1
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 +120 -0
data/test/xml/test_entity_reference.rb +5 -1
data/test/xml/test_namespace.rb +68 -0
data/test/xml/test_node.rb +546 -201
data/test/xml/test_node_attributes.rb +34 -0
data/test/xml/test_node_encoding.rb +33 -3
data/test/xml/test_node_reparenting.rb +321 -0
data/test/xml/test_node_set.rb +538 -2
data/test/xml/test_parse_options.rb +52 -0
data/test/xml/test_processing_instruction.rb +6 -1
data/test/xml/test_reader_encoding.rb +1 -1
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 +35 -1
data/test/xml/test_unparented_node.rb +5 -5
data/test/xml/test_xpath.rb +142 -11
data/test/xslt/test_custom_functions.rb +94 -0
metadata +328 -92
data/ext/nokogiri/html_sax_parser.c +0 -57
data/ext/nokogiri/html_sax_parser.h +0 -11
data/ext/nokogiri/iconv.dll +0 -0
data/ext/nokogiri/libexslt.dll +0 -0
data/ext/nokogiri/libxml2.dll +0 -0
data/ext/nokogiri/libxslt.dll +0 -0
data/ext/nokogiri/native.so +0 -0
data/ext/nokogiri/xml_xpath.c +0 -53
data/ext/nokogiri/xml_xpath.h +0 -11
data/ext/nokogiri/zlib1.dll +0 -0
data/lib/action-nokogiri.rb +0 -30
data/lib/nokogiri/css/generated_parser.rb +0 -713
data/lib/nokogiri/css/generated_tokenizer.rb +0 -144
data/lib/nokogiri/decorators.rb +0 -2
data/lib/nokogiri/decorators/hpricot.rb +0 -3
data/lib/nokogiri/decorators/hpricot/node.rb +0 -56
data/lib/nokogiri/decorators/hpricot/node_set.rb +0 -54
data/lib/nokogiri/decorators/hpricot/xpath_visitor.rb +0 -28
data/lib/nokogiri/hpricot.rb +0 -51
data/lib/nokogiri/xml/comment.rb +0 -6
data/lib/nokogiri/xml/element.rb +0 -6
data/lib/nokogiri/xml/entity_declaration.rb +0 -9
data/lib/nokogiri/xml/fragment_handler.rb +0 -34
data/test/hpricot/files/basic.xhtml +0 -17
data/test/hpricot/files/boingboing.html +0 -2266
data/test/hpricot/files/cy0.html +0 -3653
data/test/hpricot/files/immob.html +0 -400
data/test/hpricot/files/pace_application.html +0 -1320
data/test/hpricot/files/tenderlove.html +0 -16
data/test/hpricot/files/uswebgen.html +0 -220
data/test/hpricot/files/utf8.html +0 -1054
data/test/hpricot/files/week9.html +0 -1723
data/test/hpricot/files/why.xml +0 -19
data/test/hpricot/load_files.rb +0 -11
data/test/hpricot/test_alter.rb +0 -68
data/test/hpricot/test_builder.rb +0 -20
data/test/hpricot/test_parser.rb +0 -426
data/test/hpricot/test_paths.rb +0 -15
data/test/hpricot/test_preserved.rb +0 -77
data/test/hpricot/test_xml.rb +0 -30
data/test/test_gc.rb +0 -15

data/test/html/test_named_characters.rb CHANGED Viewed

@@ -1,4 +1,4 @@
-require File.expand_path(File.join(File.dirname(__FILE__), '..', "helper"))
+require "helper"
 module Nokogiri
   module HTML

data/test/html/test_node.rb CHANGED Viewed

@@ -1,4 +1,4 @@
-require File.expand_path(File.join(File.dirname(__FILE__), '..', "helper"))
+require "helper"
 require 'nkf'
@@ -11,12 +11,101 @@ 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_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_nothing_raised do
+          node.add_previous_sibling another_node
+        end
+      end
       def test_swap
         @html.at('div').swap('<a href="foo">bar</a>')
         a_tag = @html.css('a').first
@@ -24,6 +113,12 @@ 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'
@@ -32,20 +127,18 @@ module Nokogiri
         assert_equal 'foo&bar&baz', node['href']
       end
-      def test_before_will_prepend_text_nodes
-        assert node = @html.at('//body').children.first
-        node.before "some text"
-        assert_equal 'some text', @html.at('//body').children[0].content.strip
+      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_inner_html=
-        assert div = @html.at('//div')
-        div.inner_html = '<span>testing</span>'
-        assert_equal 'span', div.children.first.name
-        div.inner_html = 'testing'
-        assert_equal 'testing', div.content
+      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').fragment(iframe) }
       end
       def test_fragment
@@ -63,29 +156,17 @@ module Nokogiri
         assert_equal 'foo', p_tag.parent['class']
       end
-      def test_after_will_append_text_nodes
-        assert node = @html.at('//body/div')
-        node.after "some text"
-        assert_equal 'some text', node.next.text.strip
-      end
-      def test_replace
-        doc = Nokogiri::HTML(<<-eohtml)
-          <html>
-            <head></head>
-            <body>
-              <center><img src='logo.gif' /></center>
-            </body>
-          </html>
-        eohtml
-        center = doc.at("//center")
-        img = center.search("//img")
-        assert_raises ArgumentError do
-          center.replace img
-        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
@@ -94,8 +175,15 @@ 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/html/test_node_encoding.rb ADDED Viewed

@@ -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 CHANGED Viewed

@@ -1,36 +1,15 @@
-require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
-begin
-  require 'rubygems'
-  require 'hpricot'
-  HAS_HPRICOT = true
-rescue LoadError
-  HAS_HPRICOT = false
-end
+require "helper"
 class TestConvertXPath < Nokogiri::TestCase
   def setup
     super
     @N = Nokogiri(File.read(HTML_FILE))
-    @NH = Nokogiri.Hpricot(File.read(HTML_FILE)) # decorated document
-    @H = Hpricot(File.read(HTML_FILE)) if HAS_HPRICOT
   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"
-    if HAS_HPRICOT
-      assert_equal match, blk.call(@H.search(hpath)), "hpath result did not match"
-    end
-    assert_equal [xpath], @NH.convert_to_xpath(hpath), "converted hpath did not match xpath"
-  end
-  def test_ordinary_xpath_conversions
-    assert_equal(".//p", @NH.convert_to_xpath("p").first)
-    assert_equal(".//p", @NH.convert_to_xpath(:p).first)
-    assert_equal(".//p", @NH.convert_to_xpath("//p").first)
-    assert_equal(".//p", @NH.convert_to_xpath(".//p").first)
   end
   def test_child_tag
@@ -117,21 +96,6 @@ class TestConvertXPath < Nokogiri::TestCase
   end
   def test_positional
-##
-#  we are intentionally NOT staying compatible with nth-and-friends, as Hpricot has an OB1 bug.
-#
-#     assert_syntactical_equivalence("div > div:eq(0)", ".//div/div[position() = 1]", "\r\nTender Lovemaking\r\nThe act of making love, tenderly.\r\n") do |j|
-#       j.first.inner_text
-#     end
-#     assert_syntactical_equivalence("div/div:eq(0)", ".//div/div[position() = 1]", "\r\nTender Lovemaking\r\nThe act of making love, tenderly.\r\n") do |j|
-#       j.first.inner_text
-#     end
-#     assert_syntactical_equivalence("div/div:nth(0)", ".//div/div[position() = 1]", "\r\nTender Lovemaking\r\nThe act of making love, tenderly.\r\n") do |j|
-#       j.first.inner_text
-#     end
-#     assert_syntactical_equivalence("div/div:nth-of-type(0)", ".//div/div[position() = 1]", "\r\nTender Lovemaking\r\nThe act of making love, tenderly.\r\n") do |j|
-#       j.first.inner_text
-#     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
@@ -152,21 +116,6 @@ class TestConvertXPath < Nokogiri::TestCase
     end
   end
-  def test_compat_mode_namespaces
-    assert_equal(".//*[name()='t:sam']", @NH.convert_to_xpath("//t:sam").first)
-    assert_equal(".//*[name()='t:sam'][@rel='bookmark'][1]", @NH.convert_to_xpath("//t:sam[@rel='bookmark'][1]").first)
-  end
-  ##
-  #  'and' is not supported by hpricot
-#   def test_and
-#     assert_syntactical_equivalence("div[h1 and small]", ".//div[h1 and small]", "\r\nTender Lovemaking\r\nThe act of making love, tenderly.\r\n") do |j|
-#       j.inner_text
-#     end
-#   end
 # TODO:
 #       doc/'title ~ link' -> links that are siblings of title
 #       doc/'p[@class~="final"]' -> class includes string (whitespacy)

data/test/test_css_cache.rb CHANGED Viewed

@@ -1,4 +1,4 @@
-require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+require "helper"
 class TestCssCache < Nokogiri::TestCase
@@ -31,7 +31,7 @@ class TestCssCache < Nokogiri::TestCase
       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)
@@ -39,17 +39,6 @@ class TestCssCache < Nokogiri::TestCase
       assert_equal(times, Nokogiri::CSS::Parser.class_eval { @cache.count })
     end
-    define_method "test_hpricot_cache_#{cache_setting ? "true" : "false"}" do
-      times = cache_setting ? 2 : nil
-      Nokogiri::CSS::Parser.set_cache cache_setting
-      nh = Nokogiri.Hpricot("<html></html>")
-      nh.convert_to_xpath(@css)
-      nh.convert_to_xpath(@css)
-      assert_equal(times, Nokogiri::CSS::Parser.class_eval { @cache.count })
-    end
   end

data/test/test_encoding_handler.rb ADDED Viewed

@@ -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

data/test/test_memory_leak.rb CHANGED Viewed

@@ -1,15 +1,48 @@
-require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+require "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_dont_hurt_em_why
+      content = File.open("#{File.dirname(__FILE__)}/files/dont_hurt_em_why.xml").read
+      ndoc = Nokogiri::XML(content)
+      2.times do
+        ndoc.search('status text').first.inner_text
+        ndoc.search('user name').first.inner_text
+        GC.start
+      end
+    end
+    class BadIO
+      def read(*args)
+        raise 'hell'
+      end
+      def write(*args)
+        raise 'chickens'
+      end
+    end
+    def test_for_mem_leak_on_io_callbacks
+      io = File.open SNUGGLES_FILE
+      reader = Nokogiri::XML.parse(io)
+      (10**10).times do
+        Nokogiri::XML.parse(BadIO.new) rescue nil
+        doc.write BadIO.new rescue nil
+      end
+    end
+    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 +51,53 @@ 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