nokogiri 1.0.0 → 1.6.8.1

data/test/test_convert_xpath.rb

@@ -1,35 +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
@@ -116,21 +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
@@ -151,16 +116,6 @@ class TestConvertXPath < Nokogiri::TestCase
     end
   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

@@ -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(&: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

@@ -0,0 +1,48 @@
+# -*- coding: utf-8 -*-
+
+require "helper"
+
+class TestEncodingHandler < Nokogiri::TestCase
+  def teardown
+    Nokogiri::EncodingHandler.clear_aliases!
+    #Replace default aliases removed by clear_aliases!
+    Nokogiri.install_default_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

@@ -0,0 +1,156 @@
+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
+      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
+      Nokogiri::XML.parse(io)
+
+      loop 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>
+            <item>1234</item>
+            <item>Zzz</item>
+          <items>
+        </test>
+        EOS
+        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
+    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_in_context_parser_leak_ii
+      loop { Nokogiri::XML('<a/>').root.parse('<b/>') }
+    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
+
+  def count_object_space_documents
+    count = 0
+    ObjectSpace.each_object {|j| count += 1 if j.is_a?(Nokogiri::XML::Document) }
+    count
+  end
+end

data/test/test_nokogiri.rb

@@ -1,12 +1,52 @@
-require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
+require "helper"
 
 class TestNokogiri < Nokogiri::TestCase
+  def test_versions
+    version_match = /\d+\.\d+\.\d+/
+    assert_match version_match, Nokogiri::VERSION
+
+    assert_equal Nokogiri::VERSION_INFO['ruby']['version'], ::RUBY_VERSION
+    assert_equal Nokogiri::VERSION_INFO['ruby']['platform'], ::RUBY_PLATFORM
+
+    if Nokogiri.uses_libxml?
+      assert_match version_match, Nokogiri::LIBXML_VERSION
+      assert_equal 'extension', Nokogiri::VERSION_INFO['libxml']['binding']
+
+      assert_match version_match, Nokogiri::VERSION_INFO['libxml']['compiled']
+      assert_equal Nokogiri::LIBXML_VERSION, Nokogiri::VERSION_INFO['libxml']['compiled']
+
+      assert_match version_match, Nokogiri::VERSION_INFO['libxml']['loaded']
+      Nokogiri::LIBXML_PARSER_VERSION =~ /(\d)(\d{2})(\d{2})/
+      major = $1.to_i
+      minor = $2.to_i
+      bug   = $3.to_i
+      assert_equal "#{major}.#{minor}.#{bug}", Nokogiri::VERSION_INFO['libxml']['loaded']
+    end
+  end
+
+  def test_libxml_iconv
+    assert Nokogiri.const_defined?(:LIBXML_ICONV_ENABLED) if Nokogiri.uses_libxml?
+  end
+
+  def test_parse_with_io
+    doc = Nokogiri.parse(
+      StringIO.new("<html><head><title></title><body></body></html>")
+    )
+    assert_instance_of Nokogiri::HTML::Document, doc
+  end
+
   def test_xml?
     doc = Nokogiri.parse(File.read(XML_FILE))
     assert doc.xml?
     assert !doc.html?
   end
 
+  def test_atom_is_xml?
+    doc = Nokogiri.parse(File.read(XML_ATOM_FILE))
+    assert doc.xml?
+    assert !doc.html?
+  end
+
   def test_html?
     doc = Nokogiri.parse(File.read(HTML_FILE))
     assert !doc.xml?
@@ -33,4 +73,66 @@ class TestNokogiri < Nokogiri::TestCase
     doc = Nokogiri.make { b "bold tag" }
     assert_equal('<b>bold tag</b>', doc.to_html.chomp)
   end
+
+  SLOP_HTML = <<-END
+  <html>
+    <body>
+      <ul>
+        <li class='red'>one</li>
+        <li class='blue'>two</li>
+      </ul>
+      <div>
+        one
+        <div>div two</div>
+      </div>
+    </body>
+  </html>
+  END
+
+  def test_slop_css
+    doc = Nokogiri::Slop(<<-eohtml)
+    <html>
+      <body>
+        <div>
+          one
+          <div class='foo'>
+            div two
+            <div class='foo'>
+              div three
+            </div>
+          </div>
+        </div>
+      </body>
+    </html>
+    eohtml
+    assert_equal "div", doc.html.body.div.div('.foo').name
+  end
+
+  def test_slop
+    doc = Nokogiri::Slop(SLOP_HTML)
+
+    assert_equal "one", doc.html.body.ul.li.first.text
+    assert_equal "two", doc.html.body.ul.li(".blue").text
+    assert_equal "div two", doc.html.body.div.div.text
+
+    assert_equal "two", doc.html.body.ul.li(:css => ".blue").text
+
+    assert_equal "two", doc.html.body.ul.li(:xpath => "position()=2").text
+    assert_equal "one", doc.html.body.ul.li(:xpath => ["contains(text(),'o')"]).first.text
+    assert_equal "two", doc.html.body.ul.li(:xpath => ["contains(text(),'o')","contains(text(),'t')"]).text
+
+    assert_raise(NoMethodError) { doc.nonexistent }
+  end
+
+  def test_slop_decorator
+    doc = Nokogiri(SLOP_HTML)
+    assert !doc.decorators(Nokogiri::XML::Node).include?(Nokogiri::Decorators::Slop)
+
+    doc.slop!
+    assert doc.decorators(Nokogiri::XML::Node).include?(Nokogiri::Decorators::Slop)
+
+    doc.slop!
+    assert_equal 1, doc.decorators(Nokogiri::XML::Node).select { |d| d == Nokogiri::Decorators::Slop }.size
+  end
+
 end

data/test/test_soap4r_sax.rb

@@ -0,0 +1,52 @@
+require "helper"
+
+module XSD
+  module XMLParser
+    class Parser
+      @factory_added = nil
+
+      class << self; attr_reader :factory_added; end
+
+      def self.add_factory o
+        @factory_added = o
+      end
+
+      def initialize *args
+        @charset = nil
+      end
+
+      def characters foo
+      end
+
+      def start_element *args
+      end
+
+      def end_element *args
+      end
+    end
+  end
+end
+
+require 'xsd/xmlparser/nokogiri'
+
+class TestSoap4rSax < Nokogiri::TestCase
+  def test_factory_added
+    assert_equal XSD::XMLParser::Nokogiri, XSD::XMLParser::Nokogiri.factory_added
+  end
+
+  def test_parse
+    o = Class.new(::XSD::XMLParser::Nokogiri) do
+      attr_accessor :element_started
+      def initialize *args
+        super
+        @element_started = false
+      end
+
+      def start_element *args
+        @element_started = true
+      end
+    end.new 'foo'
+    o.do_parse '<?xml version="1.0" ?><root xmlns="http://example.com/"/>'
+    assert o.element_started, 'element started'
+  end
+end