nokogiri 1.4.0 → 1.4.1

data/lib/nokogiri/xml/node_set.rb

@@ -146,6 +146,12 @@ module Nokogiri
       end
       alias :% :at
 
+      ###
+      # Filter this list for nodes that match +expr+
+      def filter expr
+        find_all { |node| node.matches?(expr) }
+      end
+
       ###
       # Append the class attribute +name+ to all Node objects in the NodeSet.
       def add_class name
@@ -217,8 +223,8 @@ module Nokogiri
 
       ###
       # Get the inner html of all contained Node objects
-      def inner_html
-        collect{|j| j.inner_html}.join('')
+      def inner_html *args
+        collect{|j| j.inner_html(*args) }.join('')
       end
 
       ###
@@ -295,6 +301,17 @@ module Nokogiri
         inject(NodeSet.new(document)) { |set, node| set += node.children }
       end
 
+      ###
+      # Returns a new NodeSet containing all the nodes in the NodeSet
+      # in reverse order
+      def reverse
+        node_set = NodeSet.new(document)
+        (length - 1).downto(0) do |x|
+          node_set.push self[x]
+        end
+        node_set
+      end
+
       ###
       # Return a nicely formated string representation
       def inspect

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

@@ -31,7 +31,7 @@ module Nokogiri
         ###
         # Create a new PushParser with +doc+ as the SAX Document, providing
         # an optional +file_name+ and +encoding+
-        def initialize(doc = XML::SAX::Document.new, file_name = nil, encoding = 'ASCII')
+        def initialize(doc = XML::SAX::Document.new, file_name = nil, encoding = 'UTF-8')
           @document = doc
           @encoding = encoding
           @sax_parser = XML::SAX::Parser.new(doc)

data/lib/nokogiri/xml/syntax_error.rb

@@ -4,9 +4,16 @@ module Nokogiri
     # This class provides information about XML SyntaxErrors.  These
     # exceptions are typically stored on Nokogiri::XML::Document#errors.
     class SyntaxError < ::Nokogiri::SyntaxError
-      def initialize message
-        self.message = message
-      end
+      attr_reader :domain
+      attr_reader :code
+      attr_reader :level
+      attr_reader :file
+      attr_reader :line
+      attr_reader :str1
+      attr_reader :str2
+      attr_reader :str3
+      attr_reader :int1
+      attr_reader :column
 
       ###
       # return true if this is a non error
@@ -31,8 +38,6 @@ module Nokogiri
       def fatal?
         level == 3
       end
-
-      alias :to_s :message
     end
   end
 end

data/lib/nokogiri/xslt/stylesheet.rb

@@ -18,7 +18,7 @@ module Nokogiri
       # +params+ is an array of strings used as XSLT parameters.
       # returns serialized document
       def apply_to document, params = []
-        transform(document, params).to_s
+        serialize(transform(document, params))
       end
     end
   end

data/lib/xsd/xmlparser/nokogiri.rb

@@ -62,9 +62,28 @@ module XSD # :nodoc:
         characters string
       end
 
-      %w{ start_document start_element_namespace end_element_namespace end_document comment }.each do |name|
+      def start_element_namespace name, attrs = [], prefix = nil, uri = nil, ns = []
+        ###
+        # Deal with SAX v1 interface
+        name = [prefix, name].compact.join(':')
+        attributes = ns.map { |ns_prefix,ns_uri|
+          [['xmlns', ns_prefix].compact.join(':'), ns_uri]
+        } + attrs.map { |attr|
+          [[attr.prefix, attr.localname].compact.join(':'), attr.value]
+        }.flatten
+        start_element name, attributes
+      end
+
+      def end_element_namespace name, prefix = nil, uri = nil
+        ###
+        # Deal with SAX v1 interface
+        end_element [prefix, name].compact.join(':')
+      end
+
+      %w{ xmldecl start_document end_document comment }.each do |name|
         class_eval %{ def #{name}(*args); end }
       end
+
       add_factory(self)
     end
   end

data/test/css/test_parser.rb

@@ -54,6 +54,11 @@ module Nokogiri
         )
       end
 
+      def test_has
+        assert_xpath  "//a[b]", @parser.parse("a:has(b)")
+        assert_xpath  "//a[b/c]", @parser.parse("a:has(b > c)")
+      end
+
       def test_dashmatch
         assert_xpath  "//a[@class = 'bar' or starts-with(@class, concat('bar', '-'))]",
                       @parser.parse("a[@class|='bar']")

data/test/css/test_tokenizer.rb

@@ -10,6 +10,13 @@ module Nokogiri
         @scanner = Nokogiri::CSS::Tokenizer.new
       end
 
+      def test_has
+        @scanner.scan("a:has(b)")
+        assert_tokens(
+          [[:IDENT, "a"], [":", ":"], [:HAS, "has("], [:IDENT, "b"], [:RPAREN, ")"]],
+          @scanner)
+      end
+
       def test_unicode
         @scanner.scan("a日本語")
         assert_tokens([[:IDENT, 'a日本語']], @scanner)

data/test/helper.rb

@@ -1,15 +1,10 @@
 #Process.setrlimit(Process::RLIMIT_CORE, Process::RLIM_INFINITY) unless RUBY_PLATFORM =~ /(java|mswin|mingw)/i
 $VERBOSE = true
-require 'rubygems'
 require 'test/unit'
 require 'fileutils'
 require 'tempfile'
 require 'pp'
 
-%w(../lib ../ext).each do |path|
-  $LOAD_PATH.unshift(File.expand_path(File.join(File.dirname(__FILE__), path)))
-end
-
 require 'nokogiri'
 
 warn "#{__FILE__}:#{__LINE__}: libxml version info: #{Nokogiri::VERSION_INFO.inspect}"

data/test/html/test_document_fragment.rb

@@ -9,6 +9,19 @@ module Nokogiri
         @html = Nokogiri::HTML.parse(File.read(HTML_FILE), HTML_FILE)
       end
 
+      def test_ancestors_search
+        html = %q{
+          <div>
+            <ul>
+              <li>foo</li>
+            </ul>
+          </div>
+        }
+        fragment = Nokogiri::HTML.fragment html
+        li = fragment.at('li')
+        assert li.matches?('li')
+      end
+
       def test_fun_encoding
         string = %Q(<body>こんにちは</body>)
         html = Nokogiri::HTML::DocumentFragment.parse(
@@ -26,6 +39,16 @@ module Nokogiri
         assert_equal @html, fragment.document
       end
 
+      def test_empty_fragment_should_be_searchable_by_css
+        fragment = Nokogiri::HTML.fragment("")
+        assert_equal 0, fragment.css("a").size
+      end
+
+      def test_empty_fragment_should_be_searchable
+        fragment = Nokogiri::HTML.fragment("")
+        assert_equal 0, fragment.search("//a").size
+      end
+
       def test_name
         fragment = Nokogiri::HTML::DocumentFragment.new(@html)
         assert_equal '#document-fragment', fragment.name
@@ -85,6 +108,22 @@ module Nokogiri
         assert_equal "<div>b</div>", fragment.to_s
       end
 
+      def test_html_fragment_with_leading_text_and_newline
+        fragment = HTML::Document.new.fragment("First line\nSecond line<br>Broken line")
+        assert_equal fragment.to_s, "First line\nSecond line<br>Broken line"
+      end
+
+      def test_html_fragment_with_leading_whitespace_and_text_and_newline
+        fragment = HTML::Document.new.fragment("  First line\nSecond line<br>Broken line")
+        assert_equal "First line\nSecond line<br>Broken line", fragment.to_s
+      end
+
+      def test_html_fragment_with_leading_entity
+        failed = "&quot;test<br/>test&quot;"
+        fragment = Nokogiri::HTML::DocumentFragment.parse(failed)
+        assert_equal '"test<br>test"', fragment.to_html
+      end
+
       def test_to_s
         doc = "<span>foo<br></span><span>bar</span>"
         fragment = Nokogiri::HTML::Document.new.fragment(doc)
@@ -126,7 +165,6 @@ module Nokogiri
         assert_equal("<p>hello<!-- your ad here --></p>",
           fragment.to_s)
       end
-
     end
   end
 end

data/test/html/test_node.rb

@@ -17,6 +17,11 @@ module Nokogiri
         eohtml
       end
 
+      def test_attr
+        node = @html.at('div.baz')
+        assert_equal node['class'], node.attr('class')
+      end
+
       def test_get_attribute
         element = @html.at('div')
         assert_equal 'baz', element.get_attribute('class')
@@ -78,6 +83,14 @@ module Nokogiri
         assert_equal 'div', list.first.name
       end
 
+      def test_matches_inside_fragment
+        fragment = DocumentFragment.new @html
+        fragment << XML::Node.new('a', @html)
+
+        a = fragment.children.last
+        assert a.matches?('a'), 'a should match'
+      end
+
       def test_css_matches?
         assert node = @html.at('a.bar')
         assert node.matches?('a.bar')
@@ -205,6 +218,7 @@ module Nokogiri
       end
 
       def test_to_html_does_not_contain_entities
+        return unless defined?(NKF) # NKF is not implemented on Rubinius as of 2009-11-23
         html = NKF.nkf("-e --msdos", <<-EOH)
         <html><body>
         <p> test paragraph

data/test/test_encoding_handler.rb

@@ -0,0 +1,46 @@
+# -*- coding: utf-8 -*-
+
+require "helper"
+
+class TestEncodingHandler < Nokogiri::TestCase
+  def teardown
+    Nokogiri::EncodingHandler.clear_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

@@ -4,6 +4,16 @@ class TestMemoryLeak < Nokogiri::TestCase
 
   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
+        info = ndoc.search('status text').first.inner_text
+        url = ndoc.search('user name').first.inner_text
+        GC.start
+      end
+    end
+
     def test_for_memory_leak
       begin
         #  we don't use Dike in any tests, but requiring it has side effects

data/test/test_nokogiri.rb

@@ -6,7 +6,7 @@ class TestNokogiri < Nokogiri::TestCase
     assert_match version_match, Nokogiri::VERSION
     assert_match version_match, Nokogiri::LIBXML_VERSION
 
-    if defined?(FFI)
+    if defined?(FFI) && defined?(Nokogiri::LibXML)
       assert_equal 'ffi', Nokogiri::VERSION_INFO['libxml']['binding']
       if RUBY_PLATFORM =~ /java/
         assert_equal 'jruby', Nokogiri::VERSION_INFO['libxml']['platform']
@@ -30,6 +30,10 @@ class TestNokogiri < Nokogiri::TestCase
     assert_equal "#{major}.#{minor}.#{bug}", Nokogiri::VERSION_INFO['libxml']['loaded']
   end
 
+  def test_libxml_iconv
+    assert Nokogiri.const_defined?(:LIBXML_ICONV_ENABLED)
+  end
+
   def test_parse_with_io
     doc = Nokogiri.parse(
       StringIO.new("<html><head><title></title><body></body></html>")

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

data/test/test_xslt_transforms.rb

@@ -2,23 +2,23 @@ require "helper"
 
 class TestXsltTransforms < Nokogiri::TestCase
 
+  def setup
+    @doc = Nokogiri::XML(File.open(XML_FILE))
+  end
+
   if Nokogiri::VERSION_INFO['libxml']['loaded'] > '2.6.16'
 
     def test_class_methods
-      doc   = Nokogiri::XML(File.read(XML_FILE))
       style = Nokogiri::XSLT(File.read(XSLT_FILE))
 
-      assert result = style.apply_to(doc, ['title', '"Grandma"'])
+      assert result = style.apply_to(@doc, ['title', '"Grandma"'])
       assert_match %r{<h1>Grandma</h1>}, result
     end
 
     def test_transform
-      assert doc = Nokogiri::XML.parse(File.read(XML_FILE))
-      assert doc.xml?
-
       assert style = Nokogiri::XSLT.parse(File.read(XSLT_FILE))
 
-      assert result = style.apply_to(doc, ['title', '"Booyah"'])
+      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
@@ -31,29 +31,72 @@ class TestXsltTransforms < Nokogiri::TestCase
       assert_no_match %r{Dallas|Texas}, result
       assert_no_match %r{Female}, result
 
-      assert result = style.apply_to(doc, ['title', '"Grandma"'])
+      assert result = style.apply_to(@doc, ['title', '"Grandma"'])
       assert_match %r{<h1>Grandma</h1>}, result
 
-      assert result = style.apply_to(doc)
+      assert result = style.apply_to(@doc)
       assert_match %r{<h1></h1>}, result
     end
 
-    def test_transform2
-      assert doc = Nokogiri::XML.parse(File.read(XML_FILE))
-      assert doc.xml?
+    def test_transform_with_output_style
+      xslt = Nokogiri::XSLT(<<-eoxslt)
+<?xml version="1.0" encoding="ISO-8859-1"?>
 
-      assert style = Nokogiri::XSLT.parse(File.read(XSLT_FILE))
-      assert result_doc = style.transform(doc)
-      assert doc.xml?
+<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"/>
+
+<xsl:param name="title"/>
+
+<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>
+
+</xsl:stylesheet>
+      eoxslt
+      assert_no_match(/<td>/, xslt.apply_to(@doc, ['title', 'foo']))
+    end
 
-      assert style = Nokogiri::XSLT.parse(File.read(XSLT_FILE))
-      assert result_doc = style.transform(doc, ['title', '"Booyah"'])
-      assert doc.xml?
+    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
+
+      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
 
-      assert result_string = style.apply_to(doc, ['title', '"Booyah"'])
+      assert result_string = style.apply_to(@doc, ['title', '"Booyah"'])
       assert_equal result_string, style.serialize(result_doc)
     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},
@@ -64,7 +107,7 @@ class TestXsltTransforms < Nokogiri::TestCase
       }
       hh=h.dup
       result_hash = Nokogiri::XSLT.quote_params(h)
-      assert_equal hh, h # non-destructive 
+      assert_equal hh, h # non-destructive
 
       a=h.to_a.flatten
       result_array = Nokogiri::XSLT.quote_params(a)
@@ -76,32 +119,32 @@ class TestXsltTransforms < Nokogiri::TestCase
     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 = { 
+      params = {
         :p1 => 'xxx',
         :p2 => "x'x'x",
         :p3 => 'x"x"x',
         :p4 => '"xxx"'
       }
-      result_doc = Nokogiri::XML.parse(style.apply_to(doc, 
+      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/,
+        /\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, 
+      result_doc = Nokogiri::XML.parse(style.apply_to(doc,
           Nokogiri::XSLT.quote_params(params.to_a.flatten)))
       check_params result_doc, params
     end
-    
+
     def test_xslt_parse_error
       xslt_str = <<-EOX
 <xsl:stylesheet version="1.0"