nokogiri 1.1.1-java → 1.2.0-java

data/lib/nokogiri/decorators/hpricot/node_set.rb

@@ -38,7 +38,7 @@ module Nokogiri
             return sub_set
           end
 
-          ctx = CSS::Parser.parse(rule.to_s)
+          ctx = CSS.parse(rule.to_s)
           visitor = CSS::XPathVisitor.new
           visitor.extend(Hpricot::XPathVisitor)
           each do |node|

data/lib/nokogiri/html.rb

@@ -30,7 +30,7 @@ module Nokogiri
       def parse string_or_io, url = nil, encoding = nil, options = 2145
         if string_or_io.respond_to?(:read)
           url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
-          string_or_io = string_or_io.read
+          return Document.read_io(string_or_io, url, encoding, options)
         end
 
         Document.read_memory(string_or_io, url, encoding, options)
@@ -40,13 +40,19 @@ module Nokogiri
       # Parse a fragment from +string+ in to a NodeSet.
       def fragment string
         doc = parse(string)
-        finder = lambda { |children, f|
-          children.each do |child|
-            return children if string =~ /<#{child.name}/
+        fragment = XML::DocumentFragment.new(doc)
+        finder = lambda { |c, f|
+          c.each do |child|
+            fragment.add_child(child) if string =~ /<#{child.name}/
+          end
+          return fragment if fragment.children.length > 0
+
+          c.each do |child|
             finder.call(child.children, f)
           end
         }
         finder.call(doc.children, finder)
+        fragment
       end
     end
 

data/lib/nokogiri/html/document.rb

@@ -1,8 +1,12 @@
 module Nokogiri
   module HTML
     class Document < XML::Document
-      def to_html
-        serialize
+      def serialize encoding = nil, options = XML::Node::SaveOptions::FORMAT |
+        XML::Node::SaveOptions::AS_HTML |
+        XML::Node::SaveOptions::NO_DECLARATION |
+        XML::Node::SaveOptions::NO_EMPTY_TAGS
+
+        super(encoding, options)
       end
     end
   end

data/lib/nokogiri/syntax_error.rb

@@ -0,0 +1,4 @@
+module Nokogiri
+  class SyntaxError < ::StandardError
+  end
+end

data/lib/nokogiri/version.rb

@@ -1,3 +1,4 @@
 module Nokogiri
-  VERSION = '1.1.1'
+  # The version of Nokogiri you are using
+  VERSION = '1.2.0'
 end

data/lib/nokogiri/xml.rb

@@ -6,14 +6,16 @@ require 'nokogiri/xml/attr'
 require 'nokogiri/xml/dtd'
 require 'nokogiri/xml/text'
 require 'nokogiri/xml/cdata'
+require 'nokogiri/xml/processing_instruction'
 require 'nokogiri/xml/comment'
 require 'nokogiri/xml/document'
+require 'nokogiri/xml/document_fragment'
 require 'nokogiri/xml/node_set'
+require 'nokogiri/xml/syntax_error'
 require 'nokogiri/xml/xpath'
 require 'nokogiri/xml/xpath_context'
 require 'nokogiri/xml/builder'
 require 'nokogiri/xml/reader'
-require 'nokogiri/xml/syntax_error'
 require 'nokogiri/xml/notation'
 require 'nokogiri/xml/element'
 require 'nokogiri/xml/entity_declaration'

data/lib/nokogiri/xml/attr.rb

@@ -1,10 +1,9 @@
 module Nokogiri
   module XML
     class Attr < Node
-      def value
-        children.first.to_s
-      end
-      alias :to_s :value
+      alias :value :content
+      alias :to_s :content
+      alias :content= :value=
     end
   end
 end

data/lib/nokogiri/xml/cdata.rb

@@ -2,7 +2,7 @@ module Nokogiri
   module XML
     class CDATA < Text
       def name
-        'cdata-section'
+        '#cdata-section'
       end
     end
   end

data/lib/nokogiri/xml/document.rb

@@ -1,6 +1,8 @@
 module Nokogiri
   module XML
     class Document < Node
+      attr_accessor :errors
+
       def name
         'document'
       end
@@ -39,13 +41,8 @@ module Nokogiri
         @node_cache ||= {}
       end
 
-      def to_xml
-        serialize
-      end
-
-      def inner_html
-        serialize
-      end
+      alias :to_xml :serialize
+      alias :inner_html :serialize
 
       def namespaces
         root ? root.collect_namespaces : {}

data/lib/nokogiri/xml/document_fragment.rb

@@ -0,0 +1,9 @@
+module Nokogiri
+  module XML
+    class DocumentFragment < Node
+      def name
+        '#document-fragment'
+      end
+    end
+  end
+end

data/lib/nokogiri/xml/dtd.rb

@@ -1,6 +1,9 @@
 module Nokogiri
   module XML
     class DTD < Node
+      def attributes
+        nil
+      end
     end
   end
 end

data/lib/nokogiri/xml/node.rb

@@ -1,3 +1,6 @@
+require 'stringio'
+require 'nokogiri/xml/node/save_options'
+
 module Nokogiri
   module XML
     class Node
@@ -23,6 +26,7 @@ module Nokogiri
       XINCLUDE_END =       20
       DOCB_DOCUMENT_NODE = 21
 
+      # The Document associated with this Node.
       attr_accessor :document
 
       ###
@@ -76,7 +80,7 @@ module Nokogiri
       #     def regex node_set, regex
       #       node_set.find_all { |node| node['some_attribute'] =~ /#{regex}/ }
       #     end
-      #   })
+      #   }.new)
       #
       def xpath *paths
         # Pop off our custom function handler if it exists
@@ -84,7 +88,7 @@ module Nokogiri
           Hash, String, Symbol
         ].include?(paths.last.class) ? paths.pop : nil
 
-        ns = paths.last.is_a?(Hash) ? paths.pop : {}
+        ns = paths.last.is_a?(Hash) ? paths.pop : document.root.namespaces
 
         return NodeSet.new(document) unless document.root
 
@@ -134,45 +138,72 @@ module Nokogiri
           Hash, String, Symbol
         ].include?(rules.last.class) ? rules.pop : nil
 
-        ns = rules.last.is_a?(Hash) ? rules.pop : {}
+        ns = rules.last.is_a?(Hash) ? rules.pop : document.root.namespaces
 
         rules = rules.map { |rule|
-          CSS.xpath_for(rule, :prefix => ".//")
+          CSS.xpath_for(rule, :prefix => ".//", :ns => ns)
         }.flatten.uniq + [ns, handler].compact
 
         xpath(*rules)
       end
 
-      def at path, ns = {}
+      ###
+      # Search for the first occurrence of +path+.
+      # Returns nil if nothing is found, otherwise a Node.
+      def at path, ns = document.root.namespaces
         search(path, ns).first
       end
 
-      def [](property)
-        return nil unless key?(property)
-        get(property)
-      end
-
-      def next
-        next_sibling
-      end
-
-      def previous
-        previous_sibling
-      end
-
-      def remove
-        unlink
-      end
+      ###
+      # Get the attribute value for the attribute +name+
+      def [](name)
+        return nil unless key?(name)
+        get(name)
+      end
+
+      alias :next           :next_sibling
+      alias :previous       :previous_sibling
+      alias :remove         :unlink
+      alias :get_attribute  :[]
+      alias :set_attribute  :[]=
+      alias :text           :content
+      alias :inner_text     :content
+      alias :has_attribute? :key?
+      alias :<<             :add_child
+      alias :name           :node_name
+      alias :name=          :node_name=
+      alias :type           :node_type
+      alias :to_str         :text
 
       ####
       # Returns a hash containing the node's attributes.  The key is the
       # attribute name, the value is the string value of the attribute.
       def attributes
         Hash[*(attribute_nodes.map { |node|
-          [node.name, node]
+          [node.node_name, node]
         }.flatten)]
       end
 
+      ###
+      # Get the attribute values for this Node.
+      def values
+        attribute_nodes.map { |node| node.value }
+      end
+
+      ###
+      # Get the attribute names for this Node.
+      def keys
+        attribute_nodes.map { |node| node.node_name }
+      end
+
+      ###
+      # Iterate over each attribute name and value pair for this Node.
+      def each &block
+        attribute_nodes.each { |node|
+          block.call(node.node_name, node.value)
+        }
+      end
+
       ###
       # Remove the attribute named +name+
       def remove_attribute name
@@ -206,20 +237,6 @@ module Nokogiri
         end
       end
 
-      def has_attribute?(property)
-        key? property
-      end
-
-      alias :get_attribute :[]
-      def set_attribute(name, value)
-        self[name] = value
-      end
-
-      def text
-        content
-      end
-      alias :inner_text :text
-
       ####
       # Set the content to +string+.
       # If +encode+, encode any special characters first.
@@ -234,10 +251,6 @@ module Nokogiri
         parent_node
       end
 
-      def << child
-        add_child child
-      end
-
       def comment?
         type == COMMENT_NODE
       end
@@ -323,11 +336,102 @@ Node.replace requires a Node argument, and cannot accept a Document.
         replace_with_node new_node
       end
 
+      ###
+      # Test to see if this Node is equal to +other+
       def == other
         return false unless other
         return false unless other.respond_to?(:pointer_id)
         pointer_id == other.pointer_id
       end
+
+      ###
+      # Serialize Node using +encoding+ and +save_options+.  Save options 
+      # can also be set using a block. See SaveOptions.
+      #
+      # These two statements are equivalent:
+      #
+      #  node.serialize('UTF-8', FORMAT | AS_XML)
+      #
+      # or
+      #
+      #   node.serialize('UTF-8') do |config|
+      #     config.format.as_xml
+      #   end
+      #
+      def serialize encoding = nil, save_options = SaveOptions::FORMAT, &block
+        io = StringIO.new
+        write_to io, encoding, save_options, &block
+        io.rewind
+        io.read
+      end
+
+      ###
+      # Serialize this Node to HTML using +encoding+
+      def to_html encoding = nil
+        # FIXME: this is a hack around broken libxml versions
+        return dump_html if %w[2 6] === LIBXML_VERSION.split('.')[0..1]
+
+        serialize(encoding, SaveOptions::FORMAT |
+                            SaveOptions::NO_DECLARATION |
+                            SaveOptions::NO_EMPTY_TAGS |
+                            SaveOptions::AS_HTML)
+      end
+
+      ###
+      # Serialize this Node to XML using +encoding+
+      def to_xml encoding = nil
+        serialize(encoding, SaveOptions::FORMAT | SaveOptions::AS_XML)
+      end
+
+      ###
+      # Serialize this Node to XML using +encoding+
+      def to_xhtml encoding = nil
+        # FIXME: this is a hack around broken libxml versions
+        return dump_html if %w[2 6] === LIBXML_VERSION.split('.')[0..1]
+
+        serialize(encoding, SaveOptions::FORMAT |
+                            SaveOptions::NO_DECLARATION |
+                            SaveOptions::NO_EMPTY_TAGS |
+                            SaveOptions::AS_XHTML)
+      end
+
+      ###
+      # Write Node to +io+ with +encoding+ and +save_options+
+      def write_to io, encoding = nil, save_options = SaveOptions::FORMAT
+        config = SaveOptions.new(save_options)
+        yield config if block_given?
+
+        native_write_to(io, encoding, config.options)
+      end
+
+      ###
+      # Write Node as HTML to +io+ with +encoding+
+      def write_html_to io, encoding = nil
+        write_to io, encoding, SaveOptions::FORMAT |
+          SaveOptions::NO_DECLARATION |
+          SaveOptions::NO_EMPTY_TAGS |
+          SaveOptions::AS_HTML
+      end
+
+      ###
+      # Write Node as XHTML to +io+ with +encoding+
+      def write_xhtml_to io, encoding = nil
+        write_to io, encoding, SaveOptions::FORMAT |
+          SaveOptions::NO_DECLARATION |
+          SaveOptions::NO_EMPTY_TAGS |
+          SaveOptions::AS_XHTML
+      end
+
+      ###
+      # Write Node as XML to +io+ with +encoding+
+      def write_xml_to io, encoding = nil
+        write_to io, encoding, SaveOptions::FORMAT | SaveOptions::AS_XML
+      end
+
+      def self.new_from_str string
+        $stderr.puts("This method is deprecated and will be removed in 1.2.0 or by March 1, 2009. Instead, use Nokogiri::HTML.fragment()")
+        Nokogiri::HTML.fragment(string).first
+      end
     end
   end
 end

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

@@ -0,0 +1,32 @@
+module Nokogiri
+  module XML
+    class Node
+      ###
+      # Save options for serializing nodes
+      class SaveOptions
+        FORMAT          = 1  # Format serialized xml
+        NO_DECLARATION  = 2  # Do not include delcarations
+        NO_EMPTY_TAGS   = 4
+        NO_XHTML        = 8
+        AS_XHTML        = 16
+        AS_XML          = 32
+        AS_HTML         = 64
+
+        attr_reader :options
+        def initialize options = 0; @options = options; end
+        constants.each do |constant|
+          class_eval %{
+            def #{constant.downcase}
+              @options |= #{constant}
+              self
+            end
+
+            def #{constant.downcase}?
+              #{constant} & @options == #{constant}
+            end
+          }
+        end
+      end
+    end
+  end
+end