nokogiri 1.4.2 → 1.4.3

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/ffi/libxml.rb

@@ -113,10 +113,9 @@ module Nokogiri
     attach_function :xmlFreeDoc, [:pointer], :void
     attach_function :xmlSetTreeDoc, [:pointer, :pointer], :void
     attach_function :xmlNewReference, [:pointer, :string], :pointer
-    attach_function :xmlFirstElementChild, [:pointer], :pointer
-    attach_function :xmlLastElementChild, [:pointer], :pointer
-    attach_function :xmlNextElementSibling, [:pointer], :pointer
-    attach_function :xmlPreviousElementSibling, [:pointer], :pointer
+    #    attach_function :xmlFirstElementChild, [:pointer], :pointer
+    #    attach_function :xmlLastElementChild, [:pointer], :pointer
+    #    attach_function :xmlNextElementSibling, [:pointer], :pointer
     attach_function :xmlNewNode, [:pointer, :string], :pointer
     attach_function :xmlCopyNode, [:pointer, :int], :pointer
     attach_function :xmlDocCopyNode, [:pointer, :pointer, :int], :pointer
@@ -157,6 +156,7 @@ module Nokogiri
     attach_function :xmlNewDtd, [:pointer] * 4, :pointer
     attach_function :xmlGetNsList, [:pointer, :pointer], :pointer
     attach_function :xmlTextMerge, [:pointer, :pointer], :pointer
+    attach_function :xmlFreeNsList, [:pointer], :void
 
     # valid.c
     attach_function :xmlNewValidCtxt, [], :pointer
@@ -264,6 +264,7 @@ module Nokogiri
     attach_function :xmlFreeTextReader, [:pointer], :void
     attach_function :xmlReaderForIO, [:io_read_callback, :io_close_callback, :pointer, :string, :string, :int], :pointer
     attach_function :xmlTextReaderNodeType, [:pointer], :int
+    attach_function :xmlTextReaderIsEmptyElement, [:pointer], :int
 
     # xslt.c
     attach_function :xsltParseStylesheetDoc, [:pointer], :pointer
@@ -311,6 +312,30 @@ module Nokogiri
     def self.pointer_offset(n)
       n * POINTER_SIZE # byte offset of nth pointer in an array of pointers
     end
+
+    # ZOMG hacks. see GH#303
+    class << self
+      def xmlFirstElementChildHack(parent)
+        return nil if parent.nil?
+        return nil unless [Nokogiri::XML::Node::ELEMENT_NODE, Nokogiri::XML::Node::ENTITY_NODE, Nokogiri::XML::Node::DOCUMENT_NODE, Nokogiri::XML::Node::HTML_DOCUMENT_NODE].include?(parent.type)
+        parent.children.find { |child| child.element? }
+      end
+
+      def xmlLastElementChildHack(parent)
+        return nil if parent.nil?
+        return nil unless [Nokogiri::XML::Node::ELEMENT_NODE, Nokogiri::XML::Node::ENTITY_NODE, Nokogiri::XML::Node::DOCUMENT_NODE, Nokogiri::XML::Node::HTML_DOCUMENT_NODE].include?(parent.type)
+        parent.children.reverse.find { |child| child.element? }
+      end
+
+      def xmlNextElementSiblingHack(sibling)
+        return nil if sibling.nil?
+        return nil unless [Nokogiri::XML::Node::ELEMENT_NODE, Nokogiri::XML::Node::ENTITY_NODE, Nokogiri::XML::Node::DOCUMENT_NODE, Nokogiri::XML::Node::HTML_DOCUMENT_NODE].include?(sibling.type)
+        while (sibling = sibling.next_sibling)
+          return sibling if sibling.element?
+        end
+        nil
+      end
+    end
   end
 end
 

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

@@ -153,6 +153,10 @@ module Nokogiri
           node.children.each do |child|
             recursively_remove_namespaces_from_node(child)
           end
+          unless node.cstruct[:nsDef].nil?
+            LibXML.xmlFreeNsList(node.cstruct[:nsDef])
+            node.cstruct[:nsDef] = nil
+          end
         end
       end
 

data/lib/nokogiri/ffi/xml/node.rb

@@ -78,8 +78,7 @@ module Nokogiri
       end
 
       def next_element
-        sibling_ptr = LibXML.xmlNextElementSibling cstruct
-        sibling_ptr.null? ? nil : Node.wrap(sibling_ptr)
+        LibXML.xmlNextElementSiblingHack self
       end
 
       def previous_element
@@ -103,10 +102,10 @@ module Nokogiri
           retval = reparentee_struct if retval == pivot_struct.pointer # for reparent_node_with semantics
           retval = LibXML::XmlNode.new(retval) if retval.is_a?(FFI::Pointer)
           if retval[:type] == TEXT_NODE
-            if retval[:prev] && LibXML::XmlNode.new(retval[:prev])[:type] == TEXT_NODE
+            if !retval[:prev].null? && LibXML::XmlNode.new(retval[:prev])[:type] == TEXT_NODE
               retval = LibXML::XmlNode.new(LibXML.xmlTextMerge(retval[:prev], retval))
             end
-            if retval[:next] && LibXML::XmlNode.new(retval[:next])[:type] == TEXT_NODE
+            if !retval[:next].null? && LibXML::XmlNode.new(retval[:next])[:type] == TEXT_NODE
               retval = LibXML::XmlNode.new(LibXML.xmlTextMerge(retval, retval[:next]))
             end
           end
@@ -132,18 +131,17 @@ module Nokogiri
       end
 
       def element_children
-        child = LibXML.xmlFirstElementChild(cstruct)
-        return NodeSet.new(nil) if child.null?
-        child = Node.wrap(child)
+        child = LibXML.xmlFirstElementChildHack(self)
+        return NodeSet.new(nil) if child.nil?
 
         set = NodeSet.wrap(LibXML.xmlXPathNodeSetCreate(child.cstruct), self.document)
         return set unless child
 
-        next_sibling = LibXML.xmlNextElementSibling(child.cstruct)
-        while ! next_sibling.null?
-          child = Node.wrap(next_sibling)
+        next_sibling = LibXML.xmlNextElementSiblingHack(child)
+        while ! next_sibling.nil?
+          child = next_sibling
           LibXML.xmlXPathNodeSetAddUnique(set.cstruct, child.cstruct)
-          next_sibling = LibXML.xmlNextElementSibling(child.cstruct)
+          next_sibling = LibXML.xmlNextElementSiblingHack(child)
         end
 
         return set
@@ -154,13 +152,11 @@ module Nokogiri
       end
 
       def first_element_child
-        element_child = LibXML.xmlFirstElementChild(cstruct)
-        element_child.null? ? nil : Node.wrap(element_child)
+        LibXML.xmlFirstElementChildHack(self)
       end
 
       def last_element_child
-        element_child = LibXML.xmlLastElementChild(cstruct)
-        element_child.null? ? nil : Node.wrap(element_child)
+        LibXML.xmlLastElementChildHack(self)
       end
 
       def key?(attribute)
@@ -313,12 +309,14 @@ module Nokogiri
       end
 
       def add_namespace_definition(prefix, href)
-        ns = LibXML.xmlNewNs(cstruct, href, prefix)
+        ns = LibXML.xmlSearchNs(cstruct.document, cstruct, prefix.nil? ? nil : prefix.to_s)
+        namespacee = self
         if ns.null?
-          ns = LibXML.xmlSearchNs(cstruct.document, cstruct,
-            prefix.nil? ? nil : prefix.to_s)
+          namespacee = parent if type != ELEMENT_NODE
+          ns = LibXML.xmlNewNs(namespacee.cstruct, href, prefix)
         end
-        LibXML.xmlSetNs(cstruct, ns) if prefix.nil?
+        return nil if ns.null?
+        LibXML.xmlSetNs(cstruct, ns) if (prefix.nil? || self != namespacee)
         Namespace.wrap(cstruct.document, ns)
       end
 
@@ -460,6 +458,16 @@ module Nokogiri
           reparentee_struct = LibXML::XmlNode.new(reparentee_struct)
         end
 
+        if reparentee_struct[:type] == TEXT_NODE && !pivot_struct[:next].null?
+          next_text = Node.wrap(pivot_struct[:next])
+          if next_text.cstruct[:type] == TEXT_NODE
+            new_next_text = LibXML.xmlDocCopyNode(next_text.cstruct, pivot_struct[:doc], 1)
+            LibXML.xmlUnlinkNode(next_text.cstruct)
+            next_text.cstruct.keep_reference_from_document!
+            LibXML.xmlAddNextSibling(pivot_struct, new_next_text);
+          end
+        end
+
         if reparentee_struct[:type] == TEXT_NODE && pivot_struct[:type] == TEXT_NODE && Nokogiri.is_2_6_16?
           pivot_struct.pointer.put_pointer(pivot_struct.offset_of(:content), LibXML.xmlStrdup(pivot_struct[:content]))
         end

data/lib/nokogiri/ffi/xml/node_set.rb

@@ -14,7 +14,7 @@ module Nokogiri
       end
 
       def push(node) # :nodoc:
-        raise(ArgumentError, "node must be a Nokogiri::XML::Node") unless node.is_a?(XML::Node)
+        raise(ArgumentError, "node must be a Nokogiri::XML::Node") unless node.is_a?(XML::Node) || node.is_a?(XML::Namespace)
         LibXML.xmlXPathNodeSetAdd(cstruct, node.cstruct)
         self
       end
@@ -39,7 +39,7 @@ module Nokogiri
       end
 
       def delete(node) # :nodoc:
-        raise(ArgumentError, "node must be a Nokogiri::XML::Node") unless node.is_a?(XML::Node)
+        raise(ArgumentError, "node must be a Nokogiri::XML::Node") unless node.is_a?(XML::Node) || node.is_a?(XML::Namespace)
         if LibXML.xmlXPathNodeSetContains(cstruct, node.cstruct) != 0
           LibXML.xmlXPathNodeSetDel(cstruct, node.cstruct)
           return node
@@ -71,7 +71,7 @@ module Nokogiri
       end
 
       def include?(node) # :nodoc:
-        raise(ArgumentError, "node must be a Nokogiri::XML::Node") unless node.is_a?(XML::Node)
+        raise(ArgumentError, "node must be a Nokogiri::XML::Node") unless node.is_a?(XML::Node) || node.is_a?(XML::Namespace)
         (LibXML.xmlXPathNodeSetContains(cstruct, node.cstruct) != 0) ? true : false
       end
 

data/lib/nokogiri/ffi/xml/reader.rb

@@ -176,6 +176,10 @@ module Nokogiri
         LibXML.xmlTextReaderNodeType(cstruct)
       end
 
+      def empty_element?
+        LibXML.xmlTextReaderIsEmptyElement(cstruct) != 0
+      end
+
       def self.from_memory(buffer, url=nil, encoding=nil, options=0)
         raise(ArgumentError, "string cannot be nil") if buffer.nil?
 

data/lib/nokogiri/html.rb

@@ -24,8 +24,8 @@ 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 = '1.4.2'
+  VERSION = '1.4.3'
 
   # More complete version information about libxml
   VERSION_INFO = {}
@@ -9,6 +9,7 @@ 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']              = {}

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

@@ -375,8 +375,10 @@ 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]
@@ -770,7 +772,7 @@ 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/test/css/test_parser.rb

@@ -16,6 +16,10 @@ module Nokogiri
         assert_raises(CSS::SyntaxError) { @parser.parse("a[x=]") }
       end
 
+      def test_function_and_pseudo
+        assert_xpath '//child::text()[position() = 99]', @parser.parse('text():nth-of-type(99)')
+      end
+
       def test_find_by_type
         ast = @parser.parse("a:nth-child(2)").first
         matches = ast.find_by_type(
@@ -184,13 +188,19 @@ module Nokogiri
       end
 
       def test_preceding_selector
-        assert_xpath  "//F[preceding-sibling::E]",
+        assert_xpath  "//E/following-sibling::F",
                       @parser.parse("E ~ F")
+
+        assert_xpath  "//E/following-sibling::F//G",
+                      @parser.parse("E ~ F G")
       end
 
       def test_direct_preceding_selector
         assert_xpath  "//E/following-sibling::*[1]/self::F",
                       @parser.parse("E + F")
+
+        assert_xpath  "//E/following-sibling::*[1]/self::F//G",
+                      @parser.parse("E + F G")
       end
 
       def test_attribute