RubyGems - nokogiri - Versions diffs - 1.5.0 → 1.5.1.rc1 - Mend

nokogiri 1.5.0 → 1.5.1.rc1

Potentially problematic release.

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

Files changed (61) hide show

data/CHANGELOG.ja.rdoc +39 -12
data/CHANGELOG.rdoc +28 -0
data/C_CODING_STYLE.rdoc +27 -0
data/Manifest.txt +4 -0
data/README.rdoc +11 -7
data/Rakefile +42 -29
data/bin/nokogiri +10 -2
data/ext/nokogiri/extconf.rb +9 -1
data/ext/nokogiri/html_document.c +16 -0
data/ext/nokogiri/html_sax_parser_context.c +59 -37
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 +6 -8
data/ext/nokogiri/nokogiri.h +3 -0
data/ext/nokogiri/xml_document.c +101 -3
data/ext/nokogiri/xml_document.h +3 -3
data/ext/nokogiri/xml_node.c +150 -58
data/ext/nokogiri/xml_node_set.c +169 -120
data/ext/nokogiri/xml_node_set.h +5 -0
data/ext/nokogiri/xml_sax_parser_context.c +64 -41
data/ext/nokogiri/xml_text.c +2 -0
data/ext/nokogiri/xml_xpath_context.c +30 -24
data/ext/nokogiri/xslt_stylesheet.c +62 -16
data/ext/nokogiri/xslt_stylesheet.h +5 -0
data/lib/nokogiri/css/parser.rb +165 -159
data/lib/nokogiri/css/parser.y +6 -3
data/lib/nokogiri/css/tokenizer.rb +1 -1
data/lib/nokogiri/css/tokenizer.rex +1 -1
data/lib/nokogiri/html.rb +1 -0
data/lib/nokogiri/html/document.rb +82 -42
data/lib/nokogiri/html/sax/push_parser.rb +16 -0
data/lib/nokogiri/version.rb +1 -1
data/lib/nokogiri/xml.rb +6 -0
data/lib/nokogiri/xml/builder.rb +7 -1
data/lib/nokogiri/xml/document.rb +32 -17
data/lib/nokogiri/xml/document_fragment.rb +6 -1
data/lib/nokogiri/xml/node.rb +40 -9
data/lib/nokogiri/xslt.rb +5 -1
data/tasks/cross_compile.rb +1 -0
data/tasks/nokogiri.org.rb +6 -0
data/tasks/test.rb +1 -0
data/test/css/test_xpath_visitor.rb +6 -0
data/test/helper.rb +1 -0
data/test/html/test_document.rb +26 -0
data/test/html/test_document_fragment.rb +1 -2
data/test/test_memory_leak.rb +81 -1
data/test/test_xslt_transforms.rb +152 -123
data/test/xml/test_builder.rb +24 -2
data/test/xml/test_c14n.rb +151 -0
data/test/xml/test_document.rb +48 -0
data/test/xml/test_namespace.rb +5 -0
data/test/xml/test_node.rb +82 -1
data/test/xml/test_node_attributes.rb +19 -0
data/test/xml/test_node_inheritance.rb +32 -0
data/test/xml/test_node_reparenting.rb +32 -0
data/test/xml/test_node_set.rb +16 -8
data/test/xml/test_reader_encoding.rb +16 -0
data/test/xml/test_unparented_node.rb +24 -0
data/test/xml/test_xinclude.rb +83 -0
data/test/xml/test_xpath.rb +22 -0
metadata +208 -241

@@ -26,7 +26,11 @@ module Nokogiri
           XSLT.register url, klass
 	end
-        Stylesheet.parse_stylesheet_doc(XML.parse(string))
+        if Nokogiri.jruby?
+          Stylesheet.parse_stylesheet_doc(XML.parse(string), string)
+        else
+          Stylesheet.parse_stylesheet_doc(XML.parse(string))
+        end
       end
       ###

data/tasks/cross_compile.rb CHANGED

@@ -1,3 +1,4 @@
+gem 'rake-compiler'
 require 'rake/extensioncompiler'
 HOST = Rake::ExtensionCompiler.mingw_host

data/tasks/nokogiri.org.rb CHANGED

@@ -1,6 +1,12 @@
+#
+#  note that this file will only work if you've got the `nokogiri.org`
+#  repo checked out, and you've got an rvm gemset "1.8.7@nokogiri"
+#  bundled with both nokogiri's and nokogiri.org's gems.
+#
 namespace :docs do
   desc "generate HTML docs for nokogiri.org"
   task :website do
+    system 'rvm use 1.8.7@nokogiri' # see above
     title = "#{HOE.name}-#{HOE.version} Documentation"
     options = []

data/tasks/test.rb CHANGED

@@ -40,6 +40,7 @@ namespace :test do
       }
       Dir.chdir File.join(MULTI_XML, 'versions') do
         require 'net/ftp'
+        puts "Contacting xmlsoft.org ..."
         ftp = Net::FTP.new('xmlsoft.org')
         ftp.login('anonymous', 'anonymous')
         ftp.chdir('libxml2')

data/test/css/test_xpath_visitor.rb CHANGED

@@ -28,6 +28,12 @@ module Nokogiri
         assert_xpath("//a", @parser.parse('|a'))
       end
+      def test_namespaced_attribute_conversion
+        assert_xpath("//a[@flavorjones:href]", @parser.parse('a[flavorjones|href]'))
+        assert_xpath("//a[@href]", @parser.parse('a[|href]'))
+        assert_xpath("//*[@flavorjones:href]", @parser.parse('*[flavorjones|href]'))
+      end
       def test_unknown_psuedo_classes_get_pushed_down
         assert_xpath("//a[aaron(.)]", @parser.parse('a:aaron'))
       end

data/test/helper.rb CHANGED

@@ -29,6 +29,7 @@ module Nokogiri
     ADDRESS_SCHEMA_FILE = File.join(ASSETS_DIR, 'address_book.rlx')
     ADDRESS_XML_FILE = File.join(ASSETS_DIR, 'address_book.xml')
     SNUGGLES_FILE   = File.join(ASSETS_DIR, 'snuggles.xml')
+    XML_XINCLUDE_FILE = File.join(ASSETS_DIR, 'xinclude.xml')
     def teardown
       if ENV['NOKOGIRI_GC']

data/test/html/test_document.rb CHANGED

@@ -199,6 +199,32 @@ eohtml
         assert_nil html.meta_encoding
       end
+      def test_meta_encoding_with_empty_content_type
+        html = Nokogiri::HTML(<<-eohtml)
+<html>
+  <head>
+    <meta http-equiv="Content-Type" content="">
+  </head>
+  <body>
+    foo
+  </body>
+</html>
+        eohtml
+        assert_nil html.meta_encoding
+        html = Nokogiri::HTML(<<-eohtml)
+<html>
+  <head>
+    <meta http-equiv="Content-Type">
+  </head>
+  <body>
+    foo
+  </body>
+</html>
+        eohtml
+        assert_nil html.meta_encoding
+      end
       def test_root_node_parent_is_document
         parent = @html.root.parent
         assert_equal @html, parent

data/test/html/test_document_fragment.rb CHANGED

@@ -190,8 +190,7 @@ module Nokogiri
       def test_to_xhtml
         doc = "<span>foo<br></span><span>bar</span>"
         fragment = Nokogiri::HTML::Document.new.fragment(doc)
-        if !Nokogiri.uses_libxml? ||
-            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
           assert_equal "<span>foo<br></span><span>bar</span>", fragment.to_xhtml

data/test/test_memory_leak.rb CHANGED

@@ -1,6 +1,18 @@
 require "helper"
 class TestMemoryLeak < Nokogiri::TestCase
+  def setup
+    super
+    @str = <<EOF
+<!DOCTYPE HTML>
+<html>
+  <body>
+    <br />
+  </body>
+</html>
+EOF
+  end
   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
@@ -26,7 +38,7 @@ class TestMemoryLeak < Nokogiri::TestCase
       io = File.open SNUGGLES_FILE
       Nokogiri::XML.parse(io)
-      (10**10).times do
+      loop do
         Nokogiri::XML.parse(BadIO.new) rescue nil
         doc.write BadIO.new rescue nil
       end
@@ -60,6 +72,74 @@ class TestMemoryLeak < Nokogiri::TestCase
         puts "\ndike is not installed, skipping memory leak test"
       end
     end
+    def test_node_set_namespace_mem_leak
+      xml = Nokogiri::XML "<foo></foo>"
+      ctx = Nokogiri::XML::XPathContext.new(xml)
+      loop do
+        ctx.evaluate("//namespace::*")
+      end
+    end
+    def test_leak_on_node_replace
+      loop do
+        doc = Nokogiri.XML("<root><foo /></root>")
+        n = Nokogiri::XML::CDATA.new(doc, "bar")
+        pivot = doc.root.children[0]
+        pivot.replace(n)
+      end
+    end
+    def test_sax_parser_context
+      io = StringIO.new(@str)
+      loop do
+        Nokogiri::XML::SAX::ParserContext.new(@str)
+        Nokogiri::XML::SAX::ParserContext.new(io)
+        io.rewind
+        Nokogiri::HTML::SAX::ParserContext.new(@str)
+        Nokogiri::HTML::SAX::ParserContext.new(io)
+        io.rewind
+      end
+    end
+    class JumpingSaxHandler < Nokogiri::XML::SAX::Document
+      def initialize(jumptag)
+        @jumptag = jumptag
+        super()
+      end
+      def start_element(name, attrs = [])
+        throw @jumptag
+      end
+    end
+    def test_jumping_sax_handler
+      doc = JumpingSaxHandler.new(:foo)
+      loop do
+        catch(:foo) do
+          Nokogiri::HTML::SAX::Parser.new(doc).parse(@str)
+        end
+      end
+    end
+    def test_in_context_parser_leak
+      loop do
+        doc = Nokogiri::XML::Document.new
+        fragment1 = Nokogiri::XML::DocumentFragment.new(doc, '<foo/>')
+        node = fragment1.children[0]
+        node.parse('<bar></bar>')
+      end
+    end
+    def test_leak_on_xpath_string_function
+      doc = Nokogiri::XML(@str)
+      loop do
+        doc.xpath('name(//node())')
+      end
+    end
   end # if NOKOGIRI_GC
   private

data/test/test_xslt_transforms.rb CHANGED

@@ -5,38 +5,39 @@ class TestXsltTransforms < Nokogiri::TestCase
     @doc = Nokogiri::XML(File.open(XML_FILE))
   end
-  if Nokogiri.uses_libxml?
-    def test_class_methods
-      style = Nokogiri::XSLT(File.read(XSLT_FILE))
+  def test_class_methods
+    style = Nokogiri::XSLT(File.read(XSLT_FILE))
-      assert result = style.apply_to(@doc, ['title', '"Grandma"'])
-      assert_match %r{<h1>Grandma</h1>}, result
-    end
+    assert result = style.apply_to(@doc, ['title', '"Grandma"'])
+    assert_match %r{<h1>Grandma</h1>}, result
+  end
-    def test_transform
-      assert style = Nokogiri::XSLT.parse(File.read(XSLT_FILE))
-      assert result = style.apply_to(@doc, ['title', '"Booyah"'])
-      assert_match %r{<h1>Booyah</h1>}, result
-      assert_match %r{<th.*Employee ID</th>}, result
-      assert_match %r{<th.*Name</th>}, result
-      assert_match %r{<th.*Position</th>}, result
-      assert_match %r{<th.*Salary</th>}, result
-      assert_match %r{<td>EMP0003</td>}, result
-      assert_match %r{<td>Margaret Martin</td>}, result
-      assert_match %r{<td>Computer Specialist</td>}, result
-      assert_match %r{<td>100,000</td>}, result
-      assert_no_match %r{Dallas|Texas}, result
-      assert_no_match %r{Female}, result
-      assert result = style.apply_to(@doc, ['title', '"Grandma"'])
-      assert_match %r{<h1>Grandma</h1>}, result
-      assert result = style.apply_to(@doc)
-      assert_match %r{<h1></h1>}, result
-    end
+  def test_transform
+    assert style = Nokogiri::XSLT.parse(File.read(XSLT_FILE))
+    assert result = style.apply_to(@doc, ['title', '"Booyah"'])
+    assert_match %r{<h1>Booyah</h1>}, result
+    assert_match %r{<th.*Employee ID</th>}, result
+    assert_match %r{<th.*Name</th>}, result
+    assert_match %r{<th.*Position</th>}, result
+    assert_match %r{<th.*Salary</th>}, result
+    assert_match %r{<td>EMP0003</td>}, result
+    assert_match %r{<td>Margaret Martin</td>}, result
+    assert_match %r{<td>Computer Specialist</td>}, result
+    assert_match %r{<td>100,000</td>}, result
+    assert_no_match %r{Dallas|Texas}, result
+    assert_no_match %r{Female}, result
+    assert result = style.apply_to(@doc, ['title', '"Grandma"'])
+    assert_match %r{<h1>Grandma</h1>}, result
+    assert result = style.apply_to(@doc)
+    assert_match %r{<h1></h1>}, result
+  end
-    def test_transform_with_output_style
+  def test_transform_with_output_style
+    xslt = ""
+    if Nokogiri.jruby?
       xslt = Nokogiri::XSLT(<<-eoxslt)
 <?xml version="1.0" encoding="ISO-8859-1"?>
@@ -58,109 +59,138 @@ encoding="iso-8859-1" indent="yes"/>
       <td><xsl:value-of select="salary"/></td>
     </tr>
     </xsl:for-each>
-    </table>
   </body>
   </html>
 </xsl:template>
 </xsl:stylesheet>
       eoxslt
-      assert_no_match(/<td>/, xslt.apply_to(@doc, ['title', 'foo']))
-    end
+    else
+      xslt = Nokogiri::XSLT(<<-eoxslt)
+<?xml version="1.0" encoding="ISO-8859-1"?>
-    def test_transform_arg_error
-      assert style = Nokogiri::XSLT(File.read(XSLT_FILE))
-      assert_raises(TypeError) do
-        style.transform(@doc, :foo)
-      end
-    end
+<xsl:stylesheet version="1.0"
+xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
+<xsl:output method="text" version="1.0"
+encoding="iso-8859-1" indent="yes"/>
-    def test_transform_with_hash
-      assert style = Nokogiri::XSLT(File.read(XSLT_FILE))
-      result = style.transform(@doc, {'title' => '"Booyah"'})
-      assert result.html?
-      assert_equal "Booyah", result.at_css("h1").content
-    end
+<xsl:param name="title"/>
-    def test_transform2
-      assert style = Nokogiri::XSLT(File.open(XSLT_FILE))
-      assert result_doc = style.transform(@doc)
-      assert result_doc.html?
-      assert_equal "", result_doc.at_css("h1").content
+<xsl:template match="/">
+  <html>
+  <body>
+    <xsl:for-each select="staff/employee">
+    <tr>
+      <td><xsl:value-of select="employeeId"/></td>
+      <td><xsl:value-of select="name"/></td>
+      <td><xsl:value-of select="position"/></td>
+      <td><xsl:value-of select="salary"/></td>
+    </tr>
+    </xsl:for-each>
+    </table>
+  </body>
+  </html>
+</xsl:template>
-      assert style = Nokogiri::XSLT(File.read(XSLT_FILE))
-      assert result_doc = style.transform(@doc, ['title', '"Booyah"'])
-      assert result_doc.html?
-      assert_equal "Booyah", result_doc.at_css("h1").content
+</xsl:stylesheet>
+      eoxslt
+    end
+    assert_no_match(/<td>/, xslt.apply_to(@doc, ['title', 'foo']))
+  end
-      assert result_string = style.apply_to(@doc, ['title', '"Booyah"'])
-      assert_equal result_string, style.serialize(result_doc)
+  def test_transform_arg_error
+    assert style = Nokogiri::XSLT(File.read(XSLT_FILE))
+    assert_raises(TypeError) do
+      style.transform(@doc, :foo)
     end
+  end
-    def test_transform_with_quote_params
-      assert style = Nokogiri::XSLT(File.open(XSLT_FILE))
-      assert result_doc = style.transform(@doc, Nokogiri::XSLT.quote_params(['title', 'Booyah']))
-      assert result_doc.html?
-      assert_equal "Booyah", result_doc.at_css("h1").content
+  def test_transform_with_hash
+    assert style = Nokogiri::XSLT(File.read(XSLT_FILE))
+    result = style.transform(@doc, {'title' => '"Booyah"'})
+    assert result.html?
+    assert_equal "Booyah", result.at_css("h1").content
+  end
-      assert style = Nokogiri::XSLT.parse(File.read(XSLT_FILE))
-      assert result_doc = style.transform(@doc, Nokogiri::XSLT.quote_params({'title' => 'Booyah'}))
-      assert result_doc.html?
-      assert_equal "Booyah", result_doc.at_css("h1").content
-    end
+  def test_transform2
+    assert style = Nokogiri::XSLT(File.open(XSLT_FILE))
+    assert result_doc = style.transform(@doc)
+    assert result_doc.html?
+    assert_equal "", result_doc.at_css("h1").content
-    def test_quote_params
-      h = {
-        :sym   => %{xxx},
-        'str'  => %{"xxx"},
-        :sym2  => %{'xxx'},
-        'str2' => %{x'x'x},
-        :sym3  => %{x"x"x},
-      }
-      hh=h.dup
-      result_hash = Nokogiri::XSLT.quote_params(h)
-      assert_equal hh, h # non-destructive
+    assert style = Nokogiri::XSLT(File.read(XSLT_FILE))
+    assert result_doc = style.transform(@doc, ['title', '"Booyah"'])
+    assert result_doc.html?
+    assert_equal "Booyah", result_doc.at_css("h1").content
-      a=h.to_a.flatten
-      result_array = Nokogiri::XSLT.quote_params(a)
-      assert_equal h.to_a.flatten, a #non-destructive
+    assert result_string = style.apply_to(@doc, ['title', '"Booyah"'])
+    assert_equal result_string, style.serialize(result_doc)
+  end
-      assert_equal  result_array, result_hash
-    end
+  def test_transform_with_quote_params
+    assert style = Nokogiri::XSLT(File.open(XSLT_FILE))
+    assert result_doc = style.transform(@doc, Nokogiri::XSLT.quote_params(['title', 'Booyah']))
+    assert result_doc.html?
+    assert_equal "Booyah", result_doc.at_css("h1").content
+    assert style = Nokogiri::XSLT.parse(File.read(XSLT_FILE))
+    assert result_doc = style.transform(@doc, Nokogiri::XSLT.quote_params({'title' => 'Booyah'}))
+    assert result_doc.html?
+    assert_equal "Booyah", result_doc.at_css("h1").content
+  end
+  def test_quote_params
+    h = {
+      :sym   => %{xxx},
+      'str'  => %{"xxx"},
+      :sym2  => %{'xxx'},
+      'str2' => %{x'x'x},
+      :sym3  => %{x"x"x},
+    }
+    hh=h.dup
+    result_hash = Nokogiri::XSLT.quote_params(h)
+    assert_equal hh, h # non-destructive
+    a=h.to_a.flatten
+    result_array = Nokogiri::XSLT.quote_params(a)
+    assert_equal h.to_a.flatten, a #non-destructive
+    assert_equal  result_array, result_hash
+  end
-    if Nokogiri.uses_libxml? # By now, cannot get it working on JRuby.
-      def test_exslt
-        assert doc = Nokogiri::XML.parse(File.read(EXML_FILE))
-        assert doc.xml?
-        assert style = Nokogiri::XSLT.parse(File.read(EXSLT_FILE))
-        params = {
-          :p1 => 'xxx',
-          :p2 => "x'x'x",
-          :p3 => 'x"x"x',
-          :p4 => '"xxx"'
-        }
-        result_doc = Nokogiri::XML.parse(style.apply_to(doc,
-            Nokogiri::XSLT.quote_params(params)))
-        assert_equal 'func-result', result_doc.at('/root/function').content
-        assert_equal 3, result_doc.at('/root/max').content.to_i
-        assert_match(
-          /\d{4}-\d\d-\d\d([-|+]\d\d:\d\d)?/,
-          result_doc.at('/root/date').content
-          )
-        result_doc.xpath('/root/params/*').each do  |p|
-          assert_equal p.content, params[p.name.intern]
-        end
-        check_params result_doc, params
-        result_doc = Nokogiri::XML.parse(style.apply_to(doc,
-            Nokogiri::XSLT.quote_params(params.to_a.flatten)))
-        check_params result_doc, params
+  if Nokogiri.uses_libxml? # By now, cannot get it working on JRuby.
+    def test_exslt
+      assert doc = Nokogiri::XML.parse(File.read(EXML_FILE))
+      assert doc.xml?
+      assert style = Nokogiri::XSLT.parse(File.read(EXSLT_FILE))
+      params = {
+        :p1 => 'xxx',
+        :p2 => "x'x'x",
+        :p3 => 'x"x"x',
+        :p4 => '"xxx"'
+      }
+      result_doc = Nokogiri::XML.parse(style.apply_to(doc,
+          Nokogiri::XSLT.quote_params(params)))
+      assert_equal 'func-result', result_doc.at('/root/function').content
+      assert_equal 3, result_doc.at('/root/max').content.to_i
+      assert_match(
+        /\d{4}-\d\d-\d\d([-|+]\d\d:\d\d)?/,
+        result_doc.at('/root/date').content
+        )
+      result_doc.xpath('/root/params/*').each do  |p|
+        assert_equal p.content, params[p.name.intern]
       end
+      check_params result_doc, params
+      result_doc = Nokogiri::XML.parse(style.apply_to(doc,
+          Nokogiri::XSLT.quote_params(params.to_a.flatten)))
+      check_params result_doc, params
     end
+  end
-    def test_xslt_parse_error
-      xslt_str = <<-EOX
+  def test_xslt_parse_error
+    xslt_str = <<-EOX
 <xsl:stylesheet version="1.0"
  xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
   <!-- Not well-formed: -->
@@ -174,20 +204,19 @@ encoding="iso-8859-1" indent="yes"/>
     </values>
   </xsl:template>
 </xsl:stylesheet>}
-      EOX
-      assert_raises(RuntimeError) { Nokogiri::XSLT.parse(xslt_str) }
-    end
+    EOX
+    assert_raises(RuntimeError) { Nokogiri::XSLT.parse(xslt_str) }
+  end
-    def test_passing_a_non_document_to_transform
-      xsl = Nokogiri::XSLT('<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"></xsl:stylesheet>')
-      assert_raises(ArgumentError) { xsl.transform("<div></div>") }
-      assert_raises(ArgumentError) { xsl.transform(Nokogiri::HTML("").css("body")) }
-    end
+  def test_passing_a_non_document_to_transform
+    xsl = Nokogiri::XSLT('<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform"></xsl:stylesheet>')
+    assert_raises(ArgumentError) { xsl.transform("<div></div>") }
+    assert_raises(ArgumentError) { xsl.transform(Nokogiri::HTML("").css("body")) }
+  end
-    def check_params result_doc, params
-      result_doc.xpath('/root/params/*').each do  |p|
-        assert_equal p.content, params[p.name.intern]
-      end
+  def check_params result_doc, params
+    result_doc.xpath('/root/params/*').each do  |p|
+      assert_equal p.content, params[p.name.intern]
     end
   end
 end