nokogiri 1.0.6 → 1.0.7

data/lib/nokogiri/css/parser.rb

@@ -12,21 +12,32 @@ module Nokogiri
         def set_cache setting
           @cache_on = setting ? true : false
         end
+
         def cache_on?
           @cache ||= {}
           instance_variable_defined?('@cache_on') ? @cache_on : true
         end
+
         def check_cache string
           return unless cache_on?
           @cache[string]
         end
+
         def add_cache string, value
           return value unless cache_on?
           @cache[string] = value
         end
+
         def clear_cache
           @cache = {}
         end
+
+        def without_cache &block
+          tmp = @cache_on
+          @cache_on = false
+          block.call
+          @cache_on = tmp
+        end
       end
       alias :parse :scan_str
 

data/lib/nokogiri/css/parser.y

@@ -29,6 +29,14 @@ rule
                     Node.new(:CONDITIONAL_SELECTOR, [val.first, val[1]])
                   end
       }
+    | element_name hcap_1toN negation {
+        result = Node.new(:CONDITIONAL_SELECTOR,
+          [
+            val.first,
+            Node.new(:COMBINATOR, [val[1], val.last])
+          ]
+        )
+      }
     | element_name negation {
         result = Node.new(:CONDITIONAL_SELECTOR, val)
       }
@@ -36,6 +44,14 @@ rule
     | function attrib {
         result = Node.new(:CONDITIONAL_SELECTOR, val)
       }
+    | hcap_1toN negation {
+        result = Node.new(:CONDITIONAL_SELECTOR,
+          [
+            Node.new(:ELEMENT_NAME, ['*']),
+            Node.new(:COMBINATOR, val)
+          ]
+        )
+      }
     | hcap_1toN {
         result = Node.new(:CONDITIONAL_SELECTOR,
           [Node.new(:ELEMENT_NAME, ['*']), val.first]

data/lib/nokogiri/decorators.rb

@@ -1 +1,2 @@
 require 'nokogiri/decorators/hpricot'
+require 'nokogiri/decorators/slop'

data/lib/nokogiri/decorators/slop.rb

@@ -0,0 +1,31 @@
+module Nokogiri
+  module Decorators
+    ###
+    # The Slop decorator implements method missing such that a methods may be
+    # used instead of XPath or CSS.  See Nokogiri.Slop
+    module Slop
+      def method_missing name, *args, &block
+        if args.empty?
+          list = xpath("./#{name}")
+        elsif args.first.is_a? Hash
+          hash = args.first
+          if hash[:css]
+            list = css("#{name}#{hash[:css]}")
+          elsif hash[:xpath]
+            conds = Array(hash[:xpath]).collect{|j| "[#{j}]"}
+            list = xpath("./#{name}#{conds}")
+          end
+        else
+          CSS::Parser.without_cache do
+            list = xpath(
+              *CSS.xpath_for("#{name}#{args.first}", :prefix => "./")
+            )
+          end
+        end
+        
+        super if list.empty?
+        list.length == 1 ? list.first : list
+      end
+    end
+  end
+end

data/lib/nokogiri/hpricot.rb

@@ -29,10 +29,8 @@ module Nokogiri
       end
 
       def add_decorators(doc)
-        doc.decorators['node'] << Decorators::Hpricot::Node
-        doc.decorators['element'] << Decorators::Hpricot::Node
-        doc.decorators['document'] << Decorators::Hpricot::Node
-        doc.decorators['nodeset'] << Decorators::Hpricot::NodeSet
+        doc.decorators(XML::Node) << Decorators::Hpricot::Node
+        doc.decorators(XML::NodeSet) << Decorators::Hpricot::NodeSet
         doc.decorate!
         doc
       end

data/lib/nokogiri/version.rb

@@ -1,3 +1,3 @@
 module Nokogiri
-  VERSION = '1.0.6'
+  VERSION = '1.0.7'
 end

data/lib/nokogiri/xml.rb

@@ -5,6 +5,7 @@ require 'nokogiri/xml/node'
 require 'nokogiri/xml/dtd'
 require 'nokogiri/xml/text'
 require 'nokogiri/xml/cdata'
+require 'nokogiri/xml/comment'
 require 'nokogiri/xml/document'
 require 'nokogiri/xml/node_set'
 require 'nokogiri/xml/xpath'
@@ -18,6 +19,14 @@ require 'nokogiri/xml/entity_declaration'
 
 module Nokogiri
   class << self
+    ###
+    # Parse an XML file.  +thing+ may be a String, or any object that
+    # responds to _read_ and _close_ such as an IO, or StringIO.
+    # +url+ is resource where this document is located.  +encoding+ is the
+    # encoding that should be used when processing the document. +options+
+    # is a number that sets options in the parser, such as
+    # Nokogiri::XML::PARSE_RECOVER.  See the constants in
+    # Nokogiri::XML.
     def XML thing, url = nil, encoding = nil, options = 1
       Nokogiri::XML.parse(thing, url, encoding, options)
     end
@@ -43,6 +52,8 @@ module Nokogiri
     PARSE_NOXINCNODE  = 1 << 15 # do not generate XINCLUDE START/END nodes
 
     class << self
+      ###
+      # Parse an XML document.  See Nokogiri.XML.
       def parse string_or_io, url = nil, encoding = nil, options = 2159
         if string_or_io.respond_to?(:read)
           url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
@@ -55,10 +66,14 @@ module Nokogiri
         Document.read_memory(string_or_io, url, encoding, options)
       end
 
+      ###
+      # Sets whether or not entities should be substituted.
       def substitute_entities=(value = true)
         Document.substitute_entities = value
       end
 
+      ###
+      # Sets whether or not external subsets should be loaded
       def load_external_subsets=(value = true)
         Document.load_external_subsets = value
       end

data/lib/nokogiri/xml/builder.rb

@@ -18,7 +18,7 @@ module Nokogiri
       end
 
       def cdata(string)
-        node = Nokogiri::XML::CData.new(@doc, string)
+        node = Nokogiri::XML::CDATA.new(@doc, string)
         insert(node)
       end
 

data/lib/nokogiri/xml/comment.rb

@@ -0,0 +1,6 @@
+module Nokogiri
+  module XML
+    class Comment < Node
+    end
+  end
+end

data/lib/nokogiri/xml/document.rb

@@ -1,10 +1,6 @@
 module Nokogiri
   module XML
     class Document < Node
-      def decorators
-        @decorators ||= Hash.new { |h,k| h[k] = [] }
-      end
-
       def name
         'document'
       end
@@ -13,13 +9,31 @@ module Nokogiri
         self
       end
 
+      def decorators(key)
+        @decorators ||= Hash.new
+        @decorators[key] ||= []
+      end
+
+      ###
+      # Explore a document with shortcut methods.
+
+      def slop!
+        unless decorators(XML::Node).include? Nokogiri::Decorators::Slop
+          decorators(XML::Node) << Nokogiri::Decorators::Slop
+          decorate!
+        end
+
+        self
+      end
+
       ###
       # Apply any decorators to +node+
       def decorate(node)
-        key = node.class.name.split('::').last.downcase
-        decorators[key].each do |klass|
-          node.extend(klass)
-        end
+        return unless @decorators
+        @decorators.each { |klass,list|
+          next unless node.is_a?(klass)
+          list.each { |moodule| node.extend(moodule) }
+        }
       end
 
       def node_cache

data/lib/nokogiri/xml/sax/parser.rb

@@ -11,13 +11,17 @@ module Nokogiri
         # Parse given +thing+ which may be a string containing xml, or an
         # IO object.
         def parse thing
-          parse_memory(thing.is_a?(IO) ? thing.read : thing)
+          if thing.respond_to?(:read) && thing.respond_to?(:close)
+            parse_io(thing)
+          else
+            parse_memory(thing)
+          end
         end
 
         ###
         # Parse given +io+
-        def parse_io io
-          parse_memory io.read
+        def parse_io io, encoding = 0
+          native_parse_io io, encoding
         end
 
         ###

data/test/css/test_parser.rb

@@ -187,6 +187,15 @@ module Nokogiri
                       @parser.parse('foo .awesome')
       end
 
+      def test_not_so_simple_not
+        assert_xpath "//*[@id = 'p' and not(contains(concat(' ', @class, ' '), ' a '))]",
+                     @parser.parse('#p:not(.a)')
+        assert_xpath "//p[contains(concat(' ', @class, ' '), ' a ') and not(contains(concat(' ', @class, ' '), ' b '))]",
+                     @parser.parse('p.a:not(.b)')
+        assert_xpath "//p[@a = 'foo' and not(contains(concat(' ', @class, ' '), ' b '))]",
+                     @parser.parse("p[a='foo']:not(.b)")
+      end
+
       def test_ident
         assert_xpath '//x', @parser.parse('x')
       end

data/test/helper.rb

@@ -1,3 +1,4 @@
+$VERBOSE = true
 require 'test/unit'
 
 %w(../lib ../ext).each do |path|
@@ -13,7 +14,9 @@ module Nokogiri
     XSLT_FILE = File.join(ASSETS_DIR, 'staff.xslt')
     HTML_FILE = File.join(ASSETS_DIR, 'tlm.html')
 
-    undef :default_test
+    unless RUBY_VERSION >= '1.9'
+      undef :default_test
+    end
 
     def teardown
       if ENV['NOKOGIRI_GC']

data/test/hpricot/load_files.rb

@@ -1,7 +1,11 @@
 module TestFiles
-    Dir.chdir(File.dirname(__FILE__)) do
-        Dir['files/*.{html,xhtml,xml}'].each do |fname|
-            const_set fname[%r!/(\w+)\.\w+$!, 1].upcase, IO.read(fname)
-        end
+  Dir.chdir(File.dirname(__FILE__)) do
+    Dir['files/*.{html,xhtml,xml}'].each do |fname|
+      if RUBY_VERSION >= '1.9'
+        const_set fname[%r!/(\w+)\.\w+$!, 1].upcase, File.open(fname, 'r:ascii-8bit') { |f| f.read }
+      else
+        const_set fname[%r!/(\w+)\.\w+$!, 1].upcase, File.read(fname)
+      end
     end
+  end
 end

data/test/test_css_cache.rb

@@ -1,14 +1,22 @@
 require File.expand_path(File.join(File.dirname(__FILE__), "helper"))
 
-require 'rubygems'
-require 'mocha'
-
 class TestCssCache < Nokogiri::TestCase
 
   def setup
     @css = "a1 > b2 > c3"
     @parse_result = Nokogiri::CSS.parse(@css)
     @to_xpath_result = @parse_result.map {|ast| ast.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
 
@@ -19,30 +27,29 @@ class TestCssCache < Nokogiri::TestCase
 
   [ false, true ].each do |cache_setting|
     define_method "test_css_cache_#{cache_setting ? "true" : "false"}" do
-      times = cache_setting ? 1 : 6
+      times = cache_setting ? 6 : nil
+
       Nokogiri::CSS::Parser.set_cache cache_setting
       
-      Nokogiri::CSS::Parser.any_instance.expects(:parse).with(@css).returns(@parse_result).times(times)
-      Nokogiri::CSS::Node.any_instance.expects(:to_xpath).returns(@to_xpath_result).times(times)
-      
       Nokogiri::CSS.xpath_for(@css)
       Nokogiri::CSS.xpath_for(@css)
       Nokogiri::CSS::Parser.xpath_for(@css)
       Nokogiri::CSS::Parser.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
 
     define_method "test_hpricot_cache_#{cache_setting ? "true" : "false"}" do
-      times = cache_setting ? 1 : 2
+      times = cache_setting ? 2 : nil
       Nokogiri::CSS::Parser.set_cache cache_setting
 
       nh = Nokogiri.Hpricot("<html></html>")
-      Nokogiri::CSS::Parser.any_instance.expects(:parse).with(@css).returns(@parse_result).times(times)
-      Nokogiri::CSS::Node.any_instance.expects(:to_xpath).returns(@to_xpath_result).times(times)
 
       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_memory_leak.rb

@@ -0,0 +1,38 @@
+require File.expand_path(File.join(File.dirname(__FILE__), "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
+        <test>
+          <items>
+            <item>abc</item>
+            <item>1234</item>
+            <item>Zzz</item>
+          <items>
+        </test>
+        EOS
+      10.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) == 0, "memory leak detected")
+    rescue LoadError
+      puts "\ndike is not installed, skipping memory leak test"
+    end
+  end
+
+  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

@@ -33,4 +33,65 @@ 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