nokogiri 1.5.6.rc3-java → 1.5.7-java

data/test/css/test_xpath_visitor.rb

@@ -45,7 +45,7 @@ module Nokogiri
       end
 
       def test_class_selectors
-        assert_xpath  "//*[contains(concat(' ', @class, ' '), ' red ')]",
+        assert_xpath  "//*[contains(concat(' ', normalize-space(@class), ' '), ' red ')]",
                       @parser.parse(".red")
       end
 

data/test/helper.rb

@@ -1,6 +1,7 @@
 #Process.setrlimit(Process::RLIMIT_CORE, Process::RLIM_INFINITY) unless RUBY_PLATFORM =~ /(java|mswin|mingw)/i
 $VERBOSE = true
 require 'minitest/autorun'
+require 'minitest/pride'
 require 'fileutils'
 require 'tempfile'
 require 'pp'

data/test/html/sax/test_parser.rb

@@ -17,8 +17,7 @@ module Nokogiri
 
         def test_parse_empty_file
           # Make sure empty files don't break stuff
-          empty_file_name =  File.join(Dir.tmpdir, 'bogus.xml')
-          FileUtils.touch empty_file_name
+          empty_file_name =  File.join(ASSETS_DIR, 'bogus.xml')
           # assert_nothing_raised do
             @parser.parse_file empty_file_name
           # end
@@ -132,6 +131,10 @@ module Nokogiri
               ]]
           ], @parser.document.start_elements
         end
+
+        def test_empty_processing_instruction
+          @parser.parse_memory("<strong>this will segfault<?strong>")
+        end
       end
     end
   end

data/test/html/test_document_fragment.rb

@@ -26,7 +26,7 @@ module Nokogiri
 
       def test_colons_are_not_removed
         doc = Nokogiri::HTML::DocumentFragment.parse("<span>3:30pm</span>")
-        assert_match /3:30/, doc.to_s
+        assert_match(/3:30/, doc.to_s)
       end
 
       def test_parse_encoding
@@ -195,9 +195,11 @@ module Nokogiri
       def test_to_xhtml
         doc = "<span>foo<br></span><span>bar</span>"
         fragment = Nokogiri::HTML::Document.new.fragment(doc)
-        if !Nokogiri.jruby? && Nokogiri::VERSION_INFO['libxml']['loaded'] >= "2.7.0"
+        if Nokogiri.jruby? || Nokogiri::VERSION_INFO['libxml']['loaded'] >= "2.7.0"
           assert_equal "<span>foo<br /></span><span>bar</span>", fragment.to_xhtml
         else
+          # FIXME: why are we doing this ? this violates the spec,
+          # see http://www.w3.org/TR/xhtml1/#C_2
           assert_equal "<span>foo<br></span><span>bar</span>", fragment.to_xhtml
         end
       end

data/test/namespaces/test_namespaces_in_builder_doc.rb

@@ -0,0 +1,60 @@
+require "helper"
+
+module Nokogiri
+  module XML
+    class TestNamespacesInBuilderDoc < Nokogiri::TestCase
+      def setup
+        super
+        b = Nokogiri::XML::Builder.new do |x|
+          x.fruit(:xmlns => 'ns:fruit', :'xmlns:veg' => 'ns:veg', :'xmlns:xlink' => 'http://www.w3.org/1999/xlink') do
+            x.pear { x.bosc }
+            x.orange
+            x[:veg].carrot do
+              x.cheese(:xmlns => 'ns:dairy', :'xlink:href' => 'http://example.com/cheese/')
+            end
+            x[:meat].bacon(:'xmlns:meat' => 'ns:meat') do
+              x.apple :count => 2
+              x[:veg].tomato
+            end
+          end
+        end
+
+        @doc = b.doc
+      end
+
+      def check_namespace e
+        e.namespace.nil? ? nil : e.namespace.href
+      end
+
+      def test_builder_default_ns
+        assert_equal 'ns:fruit', check_namespace(@doc.root)
+      end
+      def test_builder_parent_default_ns
+        assert_equal 'ns:fruit', check_namespace(@doc.root.elements[0])
+        assert_equal 'ns:fruit', check_namespace(@doc.root.elements[1])
+      end
+      def test_builder_grandparent_default_ns
+        assert_equal 'ns:fruit', check_namespace(@doc.root.elements[0].elements[0])
+      end
+      def test_builder_parent_nondefault_ns
+        assert_equal 'ns:veg',   check_namespace(@doc.root.elements[2])
+      end
+      def test_builder_single_decl_ns_1
+        assert_equal 'ns:dairy', check_namespace(@doc.root.elements[2].elements[0])
+      end
+      def test_builder_nondefault_attr_ns
+        assert_equal 'http://www.w3.org/1999/xlink', 
+          check_namespace(@doc.root.elements[2].elements[0].attribute_nodes.find { |a| a.name =~ /href/ })
+      end
+      def test_builder_single_decl_ns_2
+        assert_equal 'ns:meat',  check_namespace(@doc.root.elements[3])
+      end
+      def test_builder_buried_default_ns
+        assert_equal 'ns:fruit',  check_namespace(@doc.root.elements[3].elements[0])
+      end
+      def test_builder_buried_decl_ns
+        assert_equal 'ns:veg',  check_namespace(@doc.root.elements[3].elements[1])
+      end
+    end
+  end
+end

data/test/namespaces/test_namespaces_in_created_doc.rb

@@ -0,0 +1,62 @@
+require "helper"
+
+module Nokogiri
+  module XML
+    class TestNamespacesInCreatedDoc < Nokogiri::TestCase
+      def setup
+        super
+        @doc = Nokogiri::XML('<fruit xmlns="ns:fruit" xmlns:veg="ns:veg" xmlns:xlink="http://www.w3.org/1999/xlink"/>')
+        pear = @doc.create_element('pear')
+        bosc = @doc.create_element('bosc')
+        pear.add_child(bosc)
+        @doc.root << pear
+        @doc.root.add_child('<orange/>')
+        carrot = @doc.create_element('veg:carrot')
+        @doc.root << carrot
+        cheese = @doc.create_element('cheese', :xmlns => 'ns:dairy', :'xlink:href' => 'http://example.com/cheese/')
+        carrot << cheese
+        bacon = @doc.create_element('meat:bacon', :'xmlns:meat' => 'ns:meat')
+        apple = @doc.create_element('apple')
+        apple['count'] = 2
+        bacon << apple
+        tomato = @doc.create_element('veg:tomato')
+        bacon << tomato
+        @doc.root << bacon
+      end
+
+      def check_namespace e
+        e.namespace.nil? ? nil : e.namespace.href
+      end
+
+      def test_created_default_ns
+        assert_equal 'ns:fruit', check_namespace(@doc.root)
+      end
+      def test_created_parent_default_ns
+        assert_equal 'ns:fruit', check_namespace(@doc.root.elements[0])
+        assert_equal 'ns:fruit', check_namespace(@doc.root.elements[1])
+      end
+      def test_created_grandparent_default_ns
+        assert_equal 'ns:fruit', check_namespace(@doc.root.elements[0].elements[0])
+      end
+      def test_created_parent_nondefault_ns
+        assert_equal 'ns:veg',   check_namespace(@doc.root.elements[2])
+      end
+      def test_created_single_decl_ns_1
+        assert_equal 'ns:dairy', check_namespace(@doc.root.elements[2].elements[0])
+      end
+      def test_created_nondefault_attr_ns
+        assert_equal 'http://www.w3.org/1999/xlink', 
+          check_namespace(@doc.root.elements[2].elements[0].attribute_nodes.find { |a| a.name =~ /href/ })
+      end
+      def test_created_single_decl_ns_2
+        assert_equal 'ns:meat',  check_namespace(@doc.root.elements[3])
+      end
+      def test_created_buried_default_ns
+        assert_equal 'ns:fruit',  check_namespace(@doc.root.elements[3].elements[0])
+      end
+      def test_created_buried_decl_ns
+        assert_equal 'ns:veg',  check_namespace(@doc.root.elements[3].elements[1])
+      end
+    end
+  end
+end

data/test/namespaces/test_namespaces_in_parsed_doc.rb

@@ -0,0 +1,60 @@
+require "helper"
+
+module Nokogiri
+  module XML
+    class TestNamespacesInParsedDoc < Nokogiri::TestCase
+      def setup
+        super
+        @doc = Nokogiri::XML <<-eoxml
+          <fruit xmlns="ns:fruit" xmlns:veg="ns:veg" xmlns:xlink="http://www.w3.org/1999/xlink">
+              <pear>
+                  <bosc/>
+              </pear>
+              <orange/>
+              <veg:carrot>
+                  <cheese xmlns="ns:dairy" xlink:href="http://example.com/cheese/"/>
+              </veg:carrot>
+              <meat:bacon xmlns:meat="ns:meat">
+                  <apple count="2"/>
+                  <veg:tomato/>
+              </meat:bacon>
+          </fruit>
+        eoxml
+      end
+
+      def check_namespace e
+        e.namespace.nil? ? nil : e.namespace.href
+      end
+
+      def test_parsed_default_ns
+        assert_equal 'ns:fruit', check_namespace(@doc.root)
+      end
+      def test_parsed_parent_default_ns
+        assert_equal 'ns:fruit', check_namespace(@doc.root.elements[0])
+        assert_equal 'ns:fruit', check_namespace(@doc.root.elements[1])
+      end
+      def test_parsed_grandparent_default_ns
+        assert_equal 'ns:fruit', check_namespace(@doc.root.elements[0].elements[0])
+      end
+      def test_parsed_parent_nondefault_ns
+        assert_equal 'ns:veg',   check_namespace(@doc.root.elements[2])
+      end
+      def test_parsed_single_decl_ns_1
+        assert_equal 'ns:dairy', check_namespace(@doc.root.elements[2].elements[0])
+      end
+      def test_parsed_nondefault_attr_ns
+        assert_equal 'http://www.w3.org/1999/xlink', 
+          check_namespace(@doc.root.elements[2].elements[0].attribute_nodes.find { |a| a.name =~ /href/ })
+      end
+      def test_parsed_single_decl_ns_2
+        assert_equal 'ns:meat',  check_namespace(@doc.root.elements[3])
+      end
+      def test_parsed_buried_default_ns
+        assert_equal 'ns:fruit',  check_namespace(@doc.root.elements[3].elements[0])
+      end
+      def test_parsed_buried_decl_ns
+        assert_equal 'ns:veg',  check_namespace(@doc.root.elements[3].elements[1])
+      end
+    end
+  end
+end

data/test/test_reader.rb

@@ -34,6 +34,29 @@ class TestReader < Nokogiri::TestCase
     assert_equal [false, false, nil, nil, true, nil], results
   end
 
+  # Issue #831
+  # Make sure that the reader doesn't block reading the entire input
+  def test_reader_blocking
+    rd, wr = IO.pipe()
+    node_out = nil
+    t = Thread.start do
+      reader = Nokogiri::XML::Reader(rd, 'UTF-8')
+      reader.each do |node|
+        node_out = node
+        break
+      end
+    end
+    sleep(1)              # sleep for one second to make sure the reader will actually block for input
+    wr.puts "<foo>"
+    wr.puts "<bar/>" * 10000
+    wr.flush
+    res = t.join(5)    # wait 5 seconds for the thread to finish
+    wr.close
+    rd.close
+    refute_nil node_out, "Didn't read any nodes, exclude the trivial case"
+    refute_nil res, "Reader blocks trying to read the entire stream"
+  end
+
   def test_reader_takes_block
     options = nil
     Nokogiri::XML::Reader(File.read(XML_FILE), XML_FILE) do |cfg|
@@ -319,12 +342,23 @@ class TestReader < Nokogiri::TestCase
   end
 
   def test_outer_xml
-    str = "<x><y>hello</y></x>"
-    reader = Nokogiri::XML::Reader.from_memory(str)
+    str = ["<x><y>hello</y></x>", "<y>hello</y>", "hello", "<y/>", "<x/>"]
+    reader = Nokogiri::XML::Reader.from_memory(str.first)
 
-    reader.read
+    xml = []
+    reader.map { |node| xml << node.outer_xml }
+
+    assert_equal str, xml
+  end
+
+  def test_outer_xml_with_empty_nodes
+    str = ["<x><y/></x>", "<y/>", "<x/>"]
+    reader = Nokogiri::XML::Reader.from_memory(str.first)
+
+    xml = []
+    reader.map { |node| xml << node.outer_xml }
 
-    assert_equal str, reader.outer_xml
+    assert_equal str, xml
   end
 
   def test_state

data/test/xml/sax/test_parser.rb

@@ -194,8 +194,17 @@ module Nokogiri
         end
 
         def test_parse_io
+          call_parse_io_with_encoding 'UTF-8'
+        end
+
+        # issue #828
+        def test_parse_io_lower_case_encoding
+          call_parse_io_with_encoding 'utf-8'
+        end
+
+        def call_parse_io_with_encoding encoding
           File.open(XML_FILE, 'rb') { |f|
-            @parser.parse_io(f, 'UTF-8')
+            @parser.parse_io(f, encoding)
           }
           assert(@parser.document.cdata_blocks.length > 0)
           if RUBY_VERSION =~ /^1\.9/

data/test/xml/test_builder.rb

@@ -1,3 +1,5 @@
+# -*- coding: utf-8 -*-
+
 require "helper"
 
 module Nokogiri
@@ -11,6 +13,12 @@ module Nokogiri
         assert_equal 'world', doc.root['abcDef']
       end
 
+      def test_builder_with_utf8_text
+        text = "test ﺵ "
+        doc = Nokogiri::XML::Builder.new(:encoding => "UTF-8") { |xml| xml.test text }.doc
+        assert_equal text, doc.content
+      end
+
       def test_builder_escape
         xml = Nokogiri::XML::Builder.new { |x|
           x.condition "value < 1", :attr => "value < 1"
@@ -103,6 +111,21 @@ module Nokogiri
         assert_equal 'bar', doc.at('foo|baz', 'foo' => 'bar').namespace.href
       end
 
+      def test_dtd_in_builder_output
+        builder = Nokogiri::XML::Builder.new do |xml|
+          xml.doc.create_internal_subset(
+                                         'html',
+                                         "-//W3C//DTD HTML 4.01 Transitional//EN",
+                                         "http://www.w3.org/TR/html4/loose.dtd"
+                                         )
+          xml.root do
+            xml.foo
+          end
+        end
+        assert_match(/<!DOCTYPE html PUBLIC "-\/\/W3C\/\/DTD HTML 4.01 Transitional\/\/EN" "http:\/\/www.w3.org\/TR\/html4\/loose.dtd">/,
+                     builder.to_xml)
+      end
+
       def test_specify_namespace_nested
         b = Nokogiri::XML::Builder.new { |xml|
           xml.root('xmlns:foo' => 'bar') do
@@ -120,11 +143,27 @@ module Nokogiri
         assert_equal 'bar', doc.at('foo|baz', 'foo' => 'bar').namespace.href
       end
 
+      def test_specified_namespace_postdeclared
+        doc = Nokogiri::XML::Builder.new { |xml|
+          xml.a do
+            xml[:foo].b("xmlns:foo" => "bar")
+          end
+        }.doc
+        a = doc.at('a')
+        assert_equal({}, a.namespaces)
+
+        b = doc.at_xpath('//foo:b', {:foo=>'bar'})
+        assert b
+        assert_equal({"xmlns:foo"=>"bar"}, b.namespaces)
+        assert_equal("b", b.name)
+        assert_equal("bar", b.namespace.href)
+      end
+
       def test_specified_namespace_undeclared
         Nokogiri::XML::Builder.new { |xml|
           xml.root do
             assert_raises(ArgumentError) do
-              xml[:foo]
+              xml[:foo].bar
             end
           end
         }

data/test/xml/test_document.rb

@@ -16,6 +16,40 @@ module Nokogiri
         @xml = Nokogiri::XML.parse(File.read(XML_FILE), XML_FILE)
       end
 
+      def test_dtd_with_empty_internal_subset
+        doc = Nokogiri::XML <<-eoxml
+<?xml version="1.0"?>
+<!DOCTYPE people >
+<people>
+</people>
+        eoxml
+        assert doc.root
+      end
+
+      # issue #838
+      def test_document_with_invalid_prolog
+        doc = Nokogiri::XML '<? ?>'
+        assert_empty doc.content
+      end
+
+      # issue #837
+      def test_document_with_refentity
+        doc = Nokogiri::XML '&amp;'
+        assert_equal '', doc.content
+      end
+
+      # issue #835
+      def test_manually_adding_reference_entities
+        d = Nokogiri::XML::Document.new
+        root = Nokogiri::XML::Element.new('bar', d)
+        txt = Nokogiri::XML::Text.new('foo', d)
+        ent = Nokogiri::XML::EntityReference.new(d, '#8217')
+        root << txt
+        root << ent
+        d << root
+        assert_match d.to_html, /&#8217;/
+      end
+
       def test_document_with_initial_space
         doc = Nokogiri::XML(" <?xml version='1.0' encoding='utf-8' ?><first \>")
         assert_equal 2, doc.children.size
@@ -29,11 +63,13 @@ module Nokogiri
       def test_ignore_unknown_namespace
         doc = Nokogiri::XML(<<-eoxml)
         <xml>
-          <unknown:foo xmlns='hello' />
+          <unknown:foo xmlns='http://hello.com/' />
           <bar />
         </xml>
         eoxml
-        refute doc.xpath('//foo').first.namespace # assert that the namespace is nil
+        if Nokogiri.jruby?
+          refute doc.xpath('//foo').first.namespace # assert that the namespace is nil
+        end
         refute_empty doc.xpath('//bar'), "bar wasn't found in the document" # bar should be part of the doc
       end
 
@@ -699,6 +735,18 @@ module Nokogiri
         assert_match %r{foo attr}, doc.to_xml
       end
 
+      # issue #785
+      def test_attribute_decoration
+        decorator = Module.new do
+          def test_method
+          end
+        end
+
+        util_decorate(@xml, decorator)
+
+        assert @xml.search('//@street').first.respond_to?(:test_method)
+      end
+
       def test_subset_is_decorated
         x = Module.new do
           def awesome!