nokogiri 1.5.0.beta.1 → 1.5.0.beta.2

data/lib/nokogiri/css/generated_tokenizer.rb

@@ -1,6 +1,6 @@
 #--
 # DO NOT MODIFY!!!!
-# This file is automatically generated by rex 1.0.4
+# This file is automatically generated by rex 1.0.5.beta1
 # from lexical definition file "lib/nokogiri/css/tokenizer.rex".
 #++
 
@@ -29,6 +29,7 @@ class GeneratedTokenizer < GeneratedParser
     scan_setup(str)
     do_parse
   end
+  alias :scan :scan_str
 
   def load_file( filename )
     @filename = filename

data/lib/nokogiri/css/parser.y

@@ -39,6 +39,9 @@ rule
         result = Node.new(:CONDITIONAL_SELECTOR, val)
       }
     | function
+    | function pseudo {
+        result = Node.new(:CONDITIONAL_SELECTOR, val)
+      }
     | function attrib {
         result = Node.new(:CONDITIONAL_SELECTOR, val)
       }

data/lib/nokogiri/css/xpath_visitor.rb

@@ -48,13 +48,6 @@ module Nokogiri
         end
       end
 
-      def visit_preceding_selector node
-        node.value.last.accept(self) +
-          '[preceding-sibling::' +
-          node.value.first.accept(self) +
-          ']'
-      end
-
       def visit_id node
         node.value.first =~ /^#(.*)$/
         "@id = '#{$1}'"
@@ -126,6 +119,7 @@ module Nokogiri
       {
         'combinator'                => ' and ',
         'direct_adjacent_selector'  => "/following-sibling::*[1]/self::",
+        'preceding_selector'        => "/following-sibling::",
         'descendant_selector'       => '//',
         'child_selector'            => '/',
       }.each do |k,v|

data/lib/nokogiri/html.rb

@@ -25,8 +25,8 @@ module Nokogiri
 
       ####
       # Parse a fragment from +string+ in to a NodeSet.
-      def fragment string
-        HTML::DocumentFragment.parse(string)
+      def fragment string, encoding = nil
+        HTML::DocumentFragment.parse string, encoding
       end
     end
 

data/lib/nokogiri/html/document_fragment.rb

@@ -2,11 +2,14 @@ module Nokogiri
   module HTML
     class DocumentFragment < Nokogiri::XML::DocumentFragment
       ####
-      # Create a Nokogiri::XML::DocumentFragment from +tags+
-      def self.parse tags
+      # Create a Nokogiri::XML::DocumentFragment from +tags+, using +encoding+
+      def self.parse tags, encoding = nil
         doc = HTML::Document.new
-        doc.encoding = 'UTF-8'
-        self.new(doc, tags)
+
+        encoding ||= tags.respond_to?(:encoding) ? tags.encoding.name : 'UTF-8'
+        doc.encoding = encoding
+
+        new(doc, tags)
       end
 
       def initialize document, tags = nil, ctx = nil

data/lib/nokogiri/version.rb

@@ -1,6 +1,6 @@
 module Nokogiri
   # The version of Nokogiri you are using
-  VERSION = '1.5.0.beta.1'
+  VERSION = '1.5.0.beta.2'
 
   # More complete version information about libxml
   VERSION_INFO = {}
@@ -9,7 +9,8 @@ module Nokogiri
   VERSION_INFO['ruby']                  = {}
   VERSION_INFO['ruby']['version']       = ::RUBY_VERSION
   VERSION_INFO['ruby']['platform']      = ::RUBY_PLATFORM
-  
+  VERSION_INFO['ruby']['engine']        = defined?(RUBY_ENGINE) ? RUBY_ENGINE : 'mri'
+
   VERSION_INFO['ruby']['jruby']         = ::JRUBY_VERSION if RUBY_PLATFORM == "java"
   if defined?(LIBXML_VERSION)
     VERSION_INFO['libxml']              = {}
@@ -31,8 +32,4 @@ module Nokogiri
   def self.jruby?
     !Nokogiri::VERSION_INFO['ruby']['jruby'].nil?
   end
-
-  def self.is_2_6_16? # :nodoc:
-    Nokogiri::VERSION_INFO['ruby']['platform'] !~ /java/ && Nokogiri::VERSION_INFO['libxml']['loaded'] <= '2.6.16'
-  end
 end

data/lib/nokogiri/xml/builder.rb

@@ -251,7 +251,7 @@ module Nokogiri
       #   end
       #
       def self.with root, &block
-        builder = self.new({}, root, &block)
+        new({}, root, &block)
       end
 
       ###

data/lib/nokogiri/xml/document.rb

@@ -7,7 +7,7 @@ module Nokogiri
     #
     # For searching a Document, see Nokogiri::XML::Node#css and
     # Nokogiri::XML::Node#xpath
-    class Document < Node
+    class Document < Nokogiri::XML::Node
       ##
       # Parse an XML file.  +thing+ may be a String, or any object that
       # responds to _read_ and _close_ such as an IO, or StringIO.
@@ -172,7 +172,6 @@ module Nokogiri
       undef_method :swap, :parent, :namespace, :default_namespace=
       undef_method :add_namespace_definition, :attributes
       undef_method :namespace_definitions, :line, :add_namespace
-      undef_method :parse, :in_context
 
       def add_child child
         raise "Document already has a root node" if root

data/lib/nokogiri/xml/document_fragment.rb

@@ -72,6 +72,13 @@ module Nokogiri
         end
       end
 
+      private
+
+      def coerce data
+        return super unless String === data
+
+        document.fragment(data).children
+      end
     end
   end
 end

data/lib/nokogiri/xml/node.rb

@@ -316,7 +316,7 @@ module Nokogiri
         if node_or_tags.is_a?(XML::NodeSet)
           node_or_tags.each { |n| add_child_node n }
         else
-          add_child node_or_tags
+          add_child_node node_or_tags
         end
         self
       end
@@ -376,8 +376,10 @@ module Nokogiri
 
       ####
       # Returns a hash containing the node's attributes.  The key is
-      # the attribute name, the value is a Nokogiri::XML::Attr
+      # the attribute name without any namespace, the value is a Nokogiri::XML::Attr
       # representing the attribute.
+      # If you need to distinguish attributes with the same name, with different namespaces
+      # use #attribute_nodes instead.
       def attributes
         Hash[*(attribute_nodes.map { |node|
           [node.node_name, node]
@@ -771,7 +773,7 @@ module Nokogiri
 
       private
 
-      def coerce(data) # :nodoc:
+      def coerce data # :nodoc:
         return data                    if data.is_a?(XML::NodeSet)
         return data.children           if data.is_a?(XML::DocumentFragment)
         return fragment(data).children if data.is_a?(String)

data/lib/nokogiri/xml/node_set.rb

@@ -138,6 +138,13 @@ module Nokogiri
         sub_set
       end
 
+      ###
+      # Search this NodeSet's nodes' immediate children using CSS selector +selector+
+      def > selector
+        ns = document.root.namespaces
+        xpath CSS.xpath_for(selector, :prefix => "./", :ns => ns).first
+      end
+
       ###
       # If path is a string, search this document for +path+ returning the
       # first Node.  Otherwise, index in to the array with +path+.
@@ -147,6 +154,24 @@ module Nokogiri
       end
       alias :% :at
 
+      ##
+      # Search this NodeSet for the first occurrence of XPath +paths+.
+      # Equivalent to <tt>xpath(paths).first</tt>
+      # See NodeSet#xpath for more information.
+      #
+      def at_xpath *paths
+        xpath(*paths).first
+      end
+
+      ##
+      # Search this NodeSet for the first occurrence of CSS +rules+.
+      # Equivalent to <tt>css(rules).first</tt>
+      # See NodeSet#css for more information.
+      #
+      def at_css *rules
+        css(*rules).first
+      end
+
       ###
       # Filter this list for nodes that match +expr+
       def filter expr

data/lib/nokogiri/xml/reader.rb

@@ -39,6 +39,8 @@ module Nokogiri
       # The XML source
       attr_reader :source
 
+      alias :self_closing? :empty_element?
+
       def initialize source, url = nil, encoding = nil # :nodoc:
         @source   = source
         @errors   = []

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

@@ -126,7 +126,9 @@ module Nokogiri
         end
 
         ###
-        # Characters read between a tag
+        # Characters read between a tag.  This method might be called multiple
+        # times given one contiguous string of characters.
+        #
         # +string+ contains the character data
         def characters string
         end

data/spec/helper.rb

@@ -0,0 +1,3 @@
+$LOAD_PATH.unshift(File.expand_path(File.join(File.dirname(__FILE__), '..', 'lib')))
+
+require 'nokogiri'

data/spec/xml/reader_spec.rb

@@ -0,0 +1,307 @@
+require File.join(File.dirname(__FILE__), '..', 'helper')
+
+include Nokogiri::XML
+
+describe Nokogiri::XML::Reader do
+  
+  it "should return the correct local_names" do
+    xml = <<-oexml
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    oexml
+    r = Nokogiri::XML::Reader.from_memory(xml)
+    r.should_not be_nil
+    r.local_name.should be_nil
+    r.map{|x| x.local_name}.should == ["x","#text","foo","#text","foo","#text","x"]
+  end
+
+  it "should return the correct names" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:edi='http://ecommerce.example.org/schema'>
+      <edi:foo>hello</edi:foo>
+    </x>
+    eoxml
+    reader.should_not be_nil
+    reader.name.should be_nil
+    reader.map{|x| x.name}.should == ["x", "#text", "edi:foo", "#text", "edi:foo", "#text", "x"]
+  end
+
+  it "should set io as source when called from_io method" do
+    io = StringIO.new(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+    reader = Nokogiri::XML::Reader.from_io(io)
+    reader.source.should be_equal(io)
+  end
+
+  it "should raise ArgumentError if nil is passed" do
+    lambda{
+      Nokogiri::XML::Reader.from_memory(nil)
+    }.should raise_error(ArgumentError)
+    lambda{
+      Nokogiri::XML::Reader.from_io(nil)
+    }.should raise_error(ArgumentError)
+  end
+
+  it "should parse from_io" do
+    io = StringIO.new(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+    reader = Nokogiri::XML::Reader.from_io(io)
+    reader.should_not be_default
+    reader.map{ |x| x.default? }.should == [false, false, false, false, false, false, false]
+  end
+
+  it "should parse when passed an io object" do
+    io = StringIO.new(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+    reader = Nokogiri::XML::Reader(io)
+    reader.should_not be_default
+    reader.map{ |x| x.default? }.should == [false, false, false, false, false, false, false]
+  end
+
+  it "should set string as source when called from_memory" do
+    xml = <<-eoxml
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+    reader = Nokogiri::XML::Reader(xml)
+    reader.source.should be_equal(xml)
+  end
+
+  it "should return false for default" do # Sorry for the description. Couldn't avoid to.
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+    reader.should_not be_default
+    reader.map { |x| x.default? }.should == [false, false, false, false, false, false, false]
+  end
+
+  it "should return the correct boolean value when asked for value?" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+
+    reader.value?.should == false # Look for how should I do this in RSpec.
+    reader.map {|x| x.value? }.should == [false, true, false, true, false, true, false]
+  end
+
+  it "should be able to deal with errors" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+      <foo>
+    </x>
+    eoxml
+
+    reader.errors.should have(0).items
+
+    lambda {
+      reader.each { |node| }
+    }.should raise_error(Nokogiri::XML::SyntaxError)
+ 
+    reader.errors.should have(1).item
+  end
+
+  it "should retrieve the right arguments" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'
+       xmlns='http://mothership.connection.com/'
+    >
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+    reader.attributes.should be_empty
+    reader.map { |x| x.attributes }.should == [{'xmlns:tenderlove'=>'http://tenderlovemaking.com/',
+						'xmlns'=>'http://mothership.connection.com/'},
+					       {}, {"awesome"=>"true"}, {}, {"awesome"=>"true"}, {},
+					       {'xmlns:tenderlove'=>'http://tenderlovemaking.com/',
+						'xmlns'=>'http://mothership.connection.com/'}]
+  end
+
+  it "should be able to acces the same attribute through both attribute and attributes method" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'
+       xmlns='http://mothership.connection.com/'
+      >
+      <tenderlove:foo awesome='true' size='giant'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+    
+    reader.attribute(nil).should be_nil
+    
+    reader.each do |node|
+      node.attributes.each do |key, value|
+	node.attribute(key).should == value
+      end
+    end
+  end
+
+  it "should be able to retrieve an attribute given an index" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+
+    reader.attribute_at(nil).should be_nil
+    reader.attribute_at(0).should be_nil
+
+    reader.map {|x| x.attribute_at(0) }.should == ['http://tenderlovemaking.com/', nil, 'true', nil, 'true', nil, 'http://tenderlovemaking.com/']
+  end
+
+  it "should be able to access an attribute value by its name" do # I should get this test for free
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+
+    reader.attribute(nil).should be_nil
+    reader.attribute('awesome').should be_nil
+    
+    reader.map {|x| x.attribute('awesome') }.should ==[nil, nil, 'true', nil, 'true', nil, nil]
+  end
+
+  it "should retrieve the correct number of attributes" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+
+    reader.attribute_count.should == 0
+    reader.map {|x| x.attribute_count}.should == [1, 0, 1, 0, 0, 0, 0]
+  end
+
+  it "should return the depth of the current node" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+
+    reader.depth.should == 0
+
+    reader.map {|x| x.depth}.should == [0,1,1,2,1,1,0]
+  end
+
+  it "should return the right encoding" do
+    string = <<-eoxml
+    <awesome>
+      <p xml:lang="en">The quick brown fox jumps over the lazy dog.</p>
+      <p xml:lang="ja">日本語が上手です</p>
+    </awesome>
+    eoxml
+    reader = Nokogiri::XML::Reader.from_memory(string, nil, 'UTF-8')
+    reader.map { |x| x.encoding }.uniq.should == ['UTF-8']
+  end
+
+  it "should return the right xml version" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+
+    reader.xml_version.should be_nil
+
+    reader.map { |x| x.xml_version }.uniq.should == ['1.0']
+  end
+
+  it "should retrieve the correct lang" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <awesome>
+      <p xml:lang="en">The quick brown fox jumps over the lazy dog.</p>
+      <p xml:lang="ja">日本語が上手です</p>
+    </awesome>
+    eoxml
+
+    reader.lang.should be_nil
+   
+    reader.map {|x| x.lang}.should == [nil, nil, "en", "en", "en", nil, "ja", "ja", "ja", nil, nil]
+  end
+
+  it "should retrieve the correct value" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:tenderlove='http://tenderlovemaking.com/'>
+      <tenderlove:foo>snuggles!</tenderlove:foo>
+    </x>
+    eoxml
+
+    reader.value.should be_nil
+
+    reader.map {|x| x.value }.should == [nil, "\n      ", nil, "snuggles!", nil, "\n    ", nil]
+  end
+
+  it "should return the node's prefix" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:edi='http://ecommerce.example.org/schema'>
+      <edi:foo>hello</edi:foo>
+    </x>
+    eoxml
+
+    reader.prefix.should be_nil
+
+    reader.map { |n| n.prefix }.should == [nil, nil, "edi", nil, "edi", nil, nil]
+  end
+
+  it "should return the reader state, whatever it is" do
+    reader = Nokogiri::XML::Reader.from_memory('<foo>bar</bar>')
+    reader.state.should_not be_nil
+  end
+
+  it "should return the namespace's uri" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:edi='http://ecommerce.example.org/schema'>
+      <edi:foo>hello</edi:foo>
+    </x>
+    eoxml
+    reader.namespace_uri.should be_nil
+
+    reader.map { |n| n.namespace_uri }.should == [nil,
+                                                  nil,
+                                                  "http://ecommerce.example.org/schema",
+                                                  nil,
+                                                  "http://ecommerce.example.org/schema",
+                                                  nil,
+                                                  nil]
+  end
+
+  it "should return the node's local_name" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:edi='http://ecommerce.example.org/schema'>
+      <edi:foo>hello</edi:foo>
+    </x>
+    eoxml
+
+    reader.local_name.should be_nil
+
+    reader.map {|n| n.local_name }.should == ["x", "#text", "foo", "#text", "foo", "#text", "x"]
+  end
+
+  it "should return node's name" do
+    reader = Nokogiri::XML::Reader.from_memory(<<-eoxml)
+    <x xmlns:edi='http://ecommerce.example.org/schema'>
+      <edi:foo>hello</edi:foo>
+    </x>
+    eoxml
+
+    reader.name.should be_nil
+
+    reader.map { |n| n.name }.should ==["x", "#text", "edi:foo", "#text", "edi:foo", "#text", "x"]
+  end
+end