nokogiri 1.5.4.rc3-java → 1.5.5-java

data/lib/nokogiri/html/document.rb

@@ -180,7 +180,7 @@ module Nokogiri
             m = chunk.match(/(<meta\s)(.*)(charset\s*=\s*([\w-]+))(.*)/i) and
               return m[4]
             catch(:encoding_found) {
-              Nokogiri::HTML::SAX::Parser.new(JumpSAXHandler.new(:encoding_found.to_s)).parse(chunk)
+              Nokogiri::HTML::SAX::Parser.new(JumpSAXHandler.new(:encoding_found)).parse(chunk)
               nil
             }
           else

data/lib/nokogiri/version.rb

@@ -1,6 +1,6 @@
 module Nokogiri
   # The version of Nokogiri you are using
-  VERSION = '1.5.4.rc3'
+  VERSION = '1.5.5'
 
   class VersionInfo # :nodoc:
     def jruby?

data/lib/nokogiri/xml/node.rb

@@ -150,7 +150,7 @@ module Nokogiri
         sets = paths.map { |path|
           ctx = XPathContext.new(self)
           ctx.register_namespaces(ns)
-          path = path.gsub(/\/xmlns:/,'/:') unless Nokogiri.uses_libxml?
+          path = path.gsub(/xmlns:/, ' :') unless Nokogiri.uses_libxml?
 
           binds.each do |key,value|
             ctx.register_variable key.to_s, value

data/lib/nokogiri/xml/node/save_options.rb

@@ -23,7 +23,7 @@ module Nokogiri
           # Save builder created document
           AS_BUILDER    = 128
           # the default for XML documents
-          DEFAULT_XML  = AS_XML # https://github.com/tenderlove/nokogiri/issues/#issue/415
+          DEFAULT_XML  = AS_XML # https://github.com/sparklemotion/nokogiri/issues/#issue/415
           # the default for HTML document
           DEFAULT_HTML = NO_DECLARATION | NO_EMPTY_TAGS | AS_HTML
         else

data/test/html/test_document.rb

@@ -85,6 +85,17 @@ module Nokogiri
         assert_equal @html.to_s, html.to_s
       end
 
+      def test_document_parse_method_with_url
+        require 'open-uri'
+        begin
+          html = open('http://google.com').read
+        rescue
+          skip("This test needs the internet. Skips if no internet available.")
+        end
+        doc = Nokogiri::HTML html ,"http:/foobar.foobar/"
+        refute_empty doc.to_s, "Document should not be empty"
+      end
+
       ###
       # Nokogiri::HTML returns an empty Document when given a blank string GH#11
       def test_empty_string_returns_empty_doc
@@ -506,6 +517,12 @@ eohtml
         assert @html.serialize
         assert @html.to_html
       end
+
+      def test_empty_document
+        # empty document should return "" #699
+        assert_equal "", Nokogiri::HTML.parse(nil).text
+        assert_equal "", Nokogiri::HTML.parse("").text
+      end
     end
   end
 end

data/test/xml/test_document.rb

@@ -268,7 +268,7 @@ module Nokogiri
           assert_equal 44, @xml.validate.length
         else
           xml = Nokogiri::XML.parse(File.read(XML_FILE), XML_FILE) {|cfg| cfg.dtdvalid}
-          assert_equal 37, xml.validate.length
+          assert_equal 40, xml.validate.length
         end
       end
 

data/test/xml/test_dtd.rb

@@ -14,12 +14,8 @@ module Nokogiri
       end
 
       def test_external_id
-        if Nokogiri.uses_libxml?
-          xml = Nokogiri::XML('<!DOCTYPE foo PUBLIC "bar"><foo />')
-        else
-          xml = Nokogiri::XML('<!DOCTYPE foo PUBLIC "bar" ""><foo />')
-        end
-        assert dtd = xml.internal_subset
+        xml = Nokogiri::XML('<!DOCTYPE foo PUBLIC "bar" ""><foo />')
+        assert dtd = xml.internal_subset, 'no internal subset'
         assert_equal 'bar', dtd.external_id
       end
 
@@ -74,7 +70,7 @@ module Nokogiri
         else
           xml = Nokogiri::XML(File.open(XML_FILE)) {|cfg| cfg.dtdvalid}
           list = xml.internal_subset.validate xml
-          assert_equal 37, list.length
+          assert_equal 40, list.length
         end
       end
 

data/test/xml/test_entity_reference.rb

@@ -17,5 +17,219 @@ module Nokogiri
         100.times { EntityReference.new(@xml, 'foo') }
       end
     end
+
+    module Common
+      PATH = 'test/files/test_document_url/'
+
+      attr_accessor :path, :parser
+
+      def xml_document
+        File.join path, 'document.xml'
+      end
+
+      def self.included base
+        def base.test_relative_and_absolute_path method_name, &block
+          test_relative_path method_name, &block
+          test_absolute_path method_name, &block
+        end
+
+        def base.test_absolute_path method_name, &block
+          define_method "#{method_name}_with_absolute_path" do
+            self.path = "#{File.expand_path PATH}/"
+            instance_eval(&block)
+          end
+        end
+
+        def base.test_relative_path method_name, &block
+          define_method method_name do
+            self.path = PATH
+            instance_eval(&block)
+          end
+        end
+      end
+    end
+
+    class TestDOMEntityReference < Nokogiri::TestCase
+      include Common
+
+      def setup
+        super
+        @parser = Nokogiri::XML::Document
+      end
+
+      test_relative_and_absolute_path :test_dom_entity_reference_with_dtdloda do
+        # Make sure that we can parse entity references and include them in the document
+        html = File.read xml_document
+        doc = @parser.parse html, path do |cfg|
+          cfg.default_xml
+          cfg.dtdload
+          cfg.noent
+        end
+        assert_equal [], doc.errors
+        assert_equal "foobar", doc.xpath('//blah').text
+      end
+
+      test_relative_and_absolute_path :test_dom_entity_reference_with_dtdvalid do
+        # Make sure that we can parse entity references and include them in the document
+        html = File.read xml_document
+        doc = @parser.parse html, path do |cfg|
+          cfg.default_xml
+          cfg.dtdvalid
+          cfg.noent
+        end
+        assert_equal [], doc.errors
+        assert_equal "foobar", doc.xpath('//blah').text
+      end
+
+      test_absolute_path :test_dom_dtd_loading_with_absolute_path do
+        # Make sure that we can parse entity references and include them in the document
+        html = %Q[<?xml version="1.0" encoding="UTF-8" ?>
+                  <!DOCTYPE document SYSTEM "#{path}/document.dtd">
+                    <document>
+                      <body>&bar;</body>
+                    </document>
+        ]
+        doc = @parser.parse html, xml_document do |cfg|
+          cfg.default_xml
+          cfg.dtdvalid
+          cfg.noent
+        end
+        assert_equal [], doc.errors
+        assert_equal "foobar", doc.xpath('//blah').text
+      end
+
+      test_relative_and_absolute_path :test_dom_entity_reference_with_io do
+        # Make sure that we can parse entity references and include them in the document
+        html = File.open xml_document
+        doc = @parser.parse html, nil do |cfg|
+          cfg.default_xml
+          cfg.dtdload
+          cfg.noent
+        end
+        assert_equal [], doc.errors
+        assert_equal "foobar", doc.xpath('//blah').text
+      end
+
+      test_relative_and_absolute_path :test_dom_entity_reference_without_noent do
+        # Make sure that we don't include entity references unless NOENT is set to true
+        html = File.read xml_document
+        doc = @parser.parse html, path do |cfg|
+          cfg.default_xml
+          cfg.dtdload
+        end
+        assert_equal [], doc.errors
+        assert_kind_of Nokogiri::XML::EntityReference, doc.xpath('//body').first.children.first
+      end
+
+      test_relative_and_absolute_path :test_dom_entity_reference_without_dtdload do
+        # Make sure that we don't include entity references unless NOENT is set to true
+        html = File.read xml_document
+        doc = @parser.parse html, path do |cfg|
+          cfg.default_xml
+        end
+        assert_kind_of Nokogiri::XML::EntityReference, doc.xpath('//body').first.children.first
+        if Nokogiri.uses_libxml?
+          assert_equal ["Entity 'bar' not defined"], doc.errors.map(&:to_s)
+        end
+      end
+
+      test_relative_and_absolute_path :test_document_dtd_loading_with_nonet do
+        # Make sure that we don't include remote entities unless NOENT is set to true
+        html = %Q[<?xml version="1.0" encoding="UTF-8" ?>
+                  <!DOCTYPE document SYSTEM "http://foo.bar.com/">
+                    <document>
+                      <body>&bar;</body>
+                    </document>
+        ]
+        doc = @parser.parse html, path do |cfg|
+          cfg.default_xml
+          cfg.dtdload
+        end
+        assert_kind_of Nokogiri::XML::EntityReference, doc.xpath('//body').first.children.first
+        if Nokogiri.uses_libxml?
+          assert_equal ["Attempt to load network entity http://foo.bar.com/", "Entity 'bar' not defined"], doc.errors.map(&:to_s)
+        else
+          assert_equal ["Attempt to load network entity http://foo.bar.com/"], doc.errors.map(&:to_s)
+        end
+      end
+      # TODO: can we retreive a resource pointing to localhost when NONET is set to true ?
+    end
+
+    class TestSaxEntityReference < Nokogiri::SAX::TestCase
+      include Common
+
+      def setup
+        super
+        @parser = XML::SAX::Parser.new(Doc.new) do |ctx|
+          ctx.replace_entities = true
+        end
+      end
+
+      test_relative_and_absolute_path :test_sax_entity_reference do
+        # Make sure that we can parse entity references and include them in the document
+        html = File.read xml_document
+        @parser.parse html
+        refute_nil @parser.document.errors
+        assert_equal ["Entity 'bar' not defined"], @parser.document.errors.map(&:to_s).map(&:strip)
+      end
+
+      test_relative_and_absolute_path :test_more_sax_entity_reference do
+        # Make sure that we don't include entity references unless NOENT is set to true
+        html = %Q[<?xml version="1.0" encoding="UTF-8" ?>
+                  <!DOCTYPE document SYSTEM "http://foo.bar.com/">
+                    <document>
+                      <body>&bar;</body>
+                    </document>
+        ]
+        @parser.parse html
+        refute_nil @parser.document.errors
+        assert_equal ["Entity 'bar' not defined"], @parser.document.errors.map(&:to_s).map(&:strip)
+      end
+    end
+
+    class TestReaderEntityReference < Nokogiri::TestCase
+      include Common
+
+      def setup
+        super
+      end
+
+      test_relative_and_absolute_path :test_reader_entity_reference do
+        # Make sure that we can parse entity references and include them in the document
+        html = File.read xml_document
+        reader = Nokogiri::XML::Reader html, path do |cfg|
+          cfg.default_xml
+          cfg.dtdload
+          cfg.noent
+        end
+        nodes = []
+        reader.each { |n| nodes << n.value }
+        assert_equal ['foobar'], nodes.compact.map(&:strip).reject(&:empty?)
+      end
+
+      test_relative_and_absolute_path :test_reader_entity_reference_without_noent do
+        # Make sure that we can parse entity references and include them in the document
+        html = File.read xml_document
+        reader = Nokogiri::XML::Reader html, path do |cfg|
+          cfg.default_xml
+          cfg.dtdload
+        end
+        nodes = []
+        reader.each { |n| nodes << n.value }
+        assert_equal [], nodes.compact.map(&:strip).reject(&:empty?)
+      end
+
+      test_relative_and_absolute_path :test_reader_entity_reference_without_dtdload do
+        # Make sure that we can parse entity references and include them in the document
+        html = File.read xml_document
+        assert_raises(Nokogiri::XML::SyntaxError) do
+          reader = Nokogiri::XML::Reader html, path do |cfg|
+            cfg.default_xml
+          end
+          nodes = []
+          reader.each { |n| nodes << n.value }
+        end
+      end
+    end
   end
 end

data/test/xml/test_node_reparenting.rb

@@ -156,6 +156,7 @@ module Nokogiri
               @doc.at_xpath("/root/a1/text()").content.must_equal "First nodehello"
             end
           end
+
           describe "#replace" do
             it "merges the Text node with adjacent Text nodes" do
               @doc.at_xpath("/root/a3/bx").replace Nokogiri::XML::Text.new('hello', @doc)
@@ -247,6 +248,13 @@ module Nokogiri
             assert_equal "foo <p></p> bartext node", xml.root.children.to_html
           end
 
+          it 'should remove the child node after the operation' do
+            fragment = Nokogiri::HTML::DocumentFragment.parse("a<a>b</a>")
+            node = fragment.children.last
+            node.add_previous_sibling node.children
+            assert_empty node.children, "should have no childrens"
+          end
+
           describe "with a text node before" do
             it "should not defensively dup the 'before' text node" do
               xml = Nokogiri::XML %Q(<root>before<p></p>after</root>)
@@ -321,7 +329,7 @@ module Nokogiri
 
         describe "unlinking a node and then reparenting it" do
           it "not blow up" do
-            # see http://github.com/tenderlove/nokogiri/issues#issue/22
+            # see http://github.com/sparklemotion/nokogiri/issues#issue/22
             10.times do
               begin
                 doc = Nokogiri::XML <<-EOHTML

data/test/xml/test_xpath.rb

@@ -217,7 +217,7 @@ module Nokogiri
       end
 
       def test_custom_xpath_handler_with_args_under_gc_pressure
-        # see http://github.com/tenderlove/nokogiri/issues/#issue/345
+        # see http://github.com/sparklemotion/nokogiri/issues/#issue/345
         tool_inspector = Class.new do
           def name_equals(nodeset, name, *args)
             nodeset.all? do |node|
@@ -263,6 +263,33 @@ module Nokogiri
           }.new)
         assert_equal foo, doc.xpath("//foo")
       end
+
+      def test_node_set_should_be_decorated
+        # "called decorate on nill" exception in JRuby issue#514
+        process_output= <<END
+<test>
+ <track type="Image">
+ <Format>LZ77</Format>
+</test>
+END
+        doc = Nokogiri::XML.parse(process_output)
+        node = doc.xpath(%{//track[@type='Video']})
+        assert_equal "[]", node.xpath("Format").inspect
+      end
+
+      def test_very_specific_xml_xpath_making_problems_in_jruby
+        # manually merges pull request #681
+        xml_string = %q{<?xml version="1.0" encoding="UTF-8"?>
+        <ONIXMessage xmlns:elibri="http://elibri.com.pl/ns/extensions" release="3.0" xmlns="http://www.editeur.org/onix/3.0/reference">
+          <Product>
+            <RecordReference>a</RecordReference>
+          </Product>
+        </ONIXMessage>}
+        
+        xml_doc = Nokogiri::XML(xml_string)
+        onix = xml_doc.children.first
+        assert_equal 'a', onix.at_xpath('xmlns:Product').at_xpath('xmlns:RecordReference').text
+      end
     end
   end
 end

metadata

@@ -1,18 +1,19 @@
 --- !ruby/object:Gem::Specification 
 name: nokogiri
 version: !ruby/object:Gem::Version 
-  prerelease: 6
-  version: 1.5.4.rc3
+  prerelease: 
+  version: 1.5.5
 platform: java
 authors: 
   - Aaron Patterson
   - Mike Dalessio
   - Yoko Harada
+  - Tim Elliott
 autorequire: 
 bindir: bin
 cert_chain: []
 
-date: 2012-06-08 00:00:00 Z
+date: 2012-06-23 00:00:00 Z
 dependencies: 
   - !ruby/object:Gem::Dependency 
     name: hoe-bundler
@@ -81,13 +82,13 @@ dependencies:
     prerelease: false
     type: :development
   - !ruby/object:Gem::Dependency 
-    name: racc
+    name: rake
     version_requirements: &id007 !ruby/object:Gem::Requirement 
       none: false
       requirements: 
         - - ">="
           - !ruby/object:Gem::Version 
-            version: 1.4.6
+            version: "0.9"
     requirement: *id007
     prerelease: false
     type: :development
@@ -102,37 +103,26 @@ dependencies:
     requirement: *id008
     prerelease: false
     type: :development
-  - !ruby/object:Gem::Dependency 
-    name: rexical
-    version_requirements: &id009 !ruby/object:Gem::Requirement 
-      none: false
-      requirements: 
-        - - ">="
-          - !ruby/object:Gem::Version 
-            version: 1.0.5
-    requirement: *id009
-    prerelease: false
-    type: :development
   - !ruby/object:Gem::Dependency 
     name: rdoc
-    version_requirements: &id010 !ruby/object:Gem::Requirement 
+    version_requirements: &id009 !ruby/object:Gem::Requirement 
       none: false
       requirements: 
         - - ~>
           - !ruby/object:Gem::Version 
             version: "3.10"
-    requirement: *id010
+    requirement: *id009
     prerelease: false
     type: :development
   - !ruby/object:Gem::Dependency 
     name: hoe
-    version_requirements: &id011 !ruby/object:Gem::Requirement 
+    version_requirements: &id010 !ruby/object:Gem::Requirement 
       none: false
       requirements: 
         - - ~>
           - !ruby/object:Gem::Version 
-            version: "2.13"
-    requirement: *id011
+            version: "2.16"
+    requirement: *id010
     prerelease: false
     type: :development
 description: "Nokogiri (\xE9\x8B\xB8) is an HTML, XML, SAX, and Reader parser.  Among Nokogiri's\n\
@@ -143,17 +133,18 @@ email:
   - aaronp@rubyforge.org
   - mike.dalessio@gmail.com
   - yokolet@gmail.com
+  - tle@holymonkey.com
 executables: 
   - nokogiri
 extensions: []
 
 extra_rdoc_files: 
-  - Manifest.txt
   - CHANGELOG.ja.rdoc
+  - CHANGELOG.rdoc
+  - C_CODING_STYLE.rdoc
+  - Manifest.txt
   - README.ja.rdoc
   - README.rdoc
-  - C_CODING_STYLE.rdoc
-  - CHANGELOG.rdoc
   - ext/nokogiri/xml_encoding_handler.c
   - ext/nokogiri/html_sax_push_parser.c
   - ext/nokogiri/xml_processing_instruction.c
@@ -506,9 +497,9 @@ required_ruby_version: !ruby/object:Gem::Requirement
 required_rubygems_version: !ruby/object:Gem::Requirement 
   none: false
   requirements: 
-    - - ">"
+    - - ">="
       - !ruby/object:Gem::Version 
-        version: 1.3.1
+        version: "0"
 requirements: []
 
 rubyforge_project: nokogiri