oga 2.3 → 2.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 59f11fbee4e3a82ad1a8cca29b80f06fdfac6d34
-  data.tar.gz: 07088c2201db9c31526cc454b61005c35075fce6
+  metadata.gz: 7ce044d495e3c695b739fd41a53ceb854f4ef8b8
+  data.tar.gz: 91126db199915b994beab17a60a5ee8009efae5c
 SHA512:
-  metadata.gz: 8d33b04c566eceb848f6de1e95d4471ebc959694e85b16e33c01dedabbbf92f5cc15679559202f6e63437424d1dfa945cb9e8ce36282a12a77585da3210ba8e3
-  data.tar.gz: 15375495eadfe074c7c3d6fe1b23a7d71df186dc57db72ed43592e97a5071e6866ddaaf0c3ac0bf92937d21e9b3bfe6a058008776265debd4e41519c8b013554
+  metadata.gz: 94dc66964aff37c6042cdba6d24cdf3a458148d4392c715c708d3e6949c7ef49054fa7a565e67fda12343f71523176ecc4488550a851e28b47f3cde886022ce1
+  data.tar.gz: b579fcaedce26f3a7c18925c777a188a3cb9332c2da4930cf14185c707af79b58a2d81572805456168ddcaeb007613b40e6993c99da87333ab923f73cafeae15

data/lib/oga.rb

@@ -23,6 +23,7 @@ if RUBY_PLATFORM == 'java'
 end
 #:nocov:
 
+require 'oga/xml/to_xml'
 require 'oga/xml/html_void_elements'
 require 'oga/xml/entities'
 require 'oga/xml/querying'
@@ -42,6 +43,7 @@ require 'oga/xml/default_namespace'
 require 'oga/xml/attribute'
 require 'oga/xml/element'
 require 'oga/xml/node_set'
+require 'oga/xml/generator'
 
 require 'oga/xml/sax_parser'
 require 'oga/xml/pull_parser'

data/lib/oga/version.rb

@@ -1,3 +1,3 @@
 module Oga
-  VERSION = '2.3'
+  VERSION = '2.4'
 end # Oga

data/lib/oga/xml/attribute.rb

@@ -3,6 +3,7 @@ module Oga
     # Class for storing information about a single XML attribute.
     class Attribute
       include ExpandedName
+      include ToXML
 
       # The name of the attribute.
       # @return [String]
@@ -81,19 +82,6 @@ module Oga
 
       alias_method :to_s, :text
 
-      # @return [String]
-      def to_xml
-        if namespace_name
-          full_name = "#{namespace_name}:#{name}"
-        else
-          full_name = name
-        end
-
-        enc_value = value ? Entities.encode_attribute(value) : nil
-
-        %Q(#{full_name}="#{enc_value}")
-      end
-
       # @return [String]
       def inspect
         segments = []

data/lib/oga/xml/cdata.rb

@@ -2,12 +2,6 @@ module Oga
   module XML
     # Class used for storing information about CDATA tags.
     class Cdata < CharacterNode
-      # Converts the node back to XML.
-      #
-      # @return [String]
-      def to_xml
-        "<![CDATA[#{text}]]>"
-      end
     end # Cdata
   end # XML
 end # Oga

data/lib/oga/xml/character_node.rb

@@ -15,11 +15,6 @@ module Oga
         @text = options[:text]
       end
 
-      # @return [String]
-      def to_xml
-        text.to_s
-      end
-
       # @return [String]
       def inspect
         "#{self.class.to_s.split('::').last}(#{text.inspect})"

data/lib/oga/xml/comment.rb

@@ -2,12 +2,6 @@ module Oga
   module XML
     # Class used for storing information about XML comments.
     class Comment < CharacterNode
-      # Converts the node back to XML.
-      #
-      # @return [String]
-      def to_xml
-        "<!--#{text}-->"
-      end
     end # Comment
   end # XML
 end # Oga

data/lib/oga/xml/doctype.rb

@@ -2,6 +2,8 @@ module Oga
   module XML
     # Class used for storing information about Doctypes.
     class Doctype
+      include ToXML
+
       # The name of the doctype (e.g. "HTML").
       # @return [String]
       attr_accessor :name
@@ -39,20 +41,6 @@ module Oga
         @inline_rules = options[:inline_rules]
       end
 
-      # Converts the doctype back to XML.
-      #
-      # @return [String]
-      def to_xml
-        segments = "<!DOCTYPE #{name}"
-
-        segments << " #{type}" if type
-        segments << %Q{ "#{public_id}"} if public_id
-        segments << %Q{ "#{system_id}"} if system_id
-        segments << " [#{inline_rules}]" if inline_rules
-
-        segments + '>'
-      end
-
       # Inspects the doctype.
       #
       # @return [String]

data/lib/oga/xml/document.rb

@@ -5,6 +5,7 @@ module Oga
     class Document
       include Querying
       include Traversal
+      include ToXML
 
       # @return [Oga::XML::Doctype]
       attr_accessor :doctype
@@ -56,23 +57,6 @@ module Oga
         self
       end
 
-      # Converts the document and its child nodes to XML.
-      #
-      # @return [String]
-      def to_xml
-        xml = children.map(&:to_xml).join('')
-
-        if doctype
-          xml = doctype.to_xml + "\n" + xml.strip
-        end
-
-        if xml_declaration
-          xml = xml_declaration.to_xml + "\n" + xml.strip
-        end
-
-        xml
-      end
-
       # @return [TrueClass|FalseClass]
       def html?
         type.equal?(:html)
@@ -99,6 +83,11 @@ Document(
 )
         EOF
       end
+
+      # @return [FalseClass]
+      def literal_html_name?
+        false
+      end
     end # Document
   end # XML
 end # Oga

data/lib/oga/xml/element.rb

@@ -211,30 +211,6 @@ module Oga
         @children = NodeSet.new([text_node], self)
       end
 
-      # Converts the element and its child elements to XML.
-      #
-      # @return [String]
-      def to_xml
-        if namespace_name
-          full_name = "#{namespace_name}:#{name}"
-        else
-          full_name = name
-        end
-
-        body  = children.map(&:to_xml).join('')
-        attrs = ''
-
-        attributes.each do |attr|
-          attrs << " #{attr.to_xml}"
-        end
-
-        if self_closing?
-          return "<#{full_name}#{attrs} />"
-        else
-          return "<#{full_name}#{attrs}>#{body}</#{full_name}>"
-        end
-      end
-
       # @return [String]
       def inspect
         segments = []
@@ -323,6 +299,14 @@ module Oga
         end
       end
 
+      # Returns true if the current element name is the name of one of the
+      # literal HTML elements.
+      #
+      # @return [TrueClass|FalseClass]
+      def literal_html_name?
+        Lexer::LITERAL_HTML_ELEMENTS.allow?(name)
+      end
+
       private
 
       # Registers namespaces based on any "xmlns" attributes.

data/lib/oga/xml/generator.rb

@@ -0,0 +1,198 @@
+module Oga
+  module XML
+    # Class for generating XML as a String based on an existing document.
+    #
+    # Basic usage:
+    #
+    #     element = Oga::XML::Element.new(name: 'root')
+    #     element.inner_text = 'hello'
+    #
+    #     gen = Oga::XML::Generator.new(element)
+    #
+    #     gen.to_xml # => "<root>hello</root>"
+    #
+    # @private
+    class Generator
+      # @param [Oga::XML::Document|Oga::XML::Node] start The node to serialise.
+      def initialize(root)
+        @start = root
+
+        if @start.respond_to?(:root_node)
+          @html_mode = @start.root_node.html?
+        else
+          @html_mode = false
+        end
+      end
+
+      # Returns the XML for the current root node.
+      #
+      # @return [String]
+      def to_xml
+        current = @start
+        output = ''
+
+        while current
+          children = false
+
+          # Determine what callback to use for the current node. The order of
+          # this statement is based on how likely it is for an arm to match.
+          case current
+          when Oga::XML::Element
+            callback = :on_element
+            children = true
+          when Oga::XML::Text
+            callback = :on_text
+          when Oga::XML::Cdata
+            callback = :on_cdata
+          when Oga::XML::Comment
+            callback = :on_comment
+          when Oga::XML::Attribute
+            callback = :on_attribute
+          when Oga::XML::ProcessingInstruction
+            callback = :on_processing_instruction
+          when Oga::XML::Doctype
+            callback = :on_doctype
+          when Oga::XML::XmlDeclaration
+            callback = :on_xml_declaration
+          when Oga::XML::Document
+            callback = :on_document
+            children = true
+          else
+            raise TypeError, "Can't serialize #{current.class} to XML"
+          end
+
+          send(callback, current, output)
+
+          if child_node = children && current.children[0]
+            current = child_node
+          else
+            until next_node = current.is_a?(Node) && current.next
+              if current.is_a?(Node) && current != @start
+                current = current.parent
+              end
+
+              send(:after_element, current, output) if current.is_a?(Element)
+
+              break if current == @start
+            end
+
+            current = next_node
+          end
+        end
+
+        output
+      end
+
+      # @param [Oga::XML::Text] node
+      # @param [String] output
+      def on_text(node, output)
+        if @html_mode && (parent = node.parent) && parent.literal_html_name?
+          output << node.text
+        else
+          output << Entities.encode(node.text)
+        end
+      end
+
+      # @param [Oga::XML::Cdata] node
+      # @param [String] output
+      def on_cdata(node, output)
+        output << "<![CDATA[#{node.text}]]>"
+      end
+
+      # @param [Oga::XML::Comment] node
+      # @param [String] output
+      def on_comment(node, output)
+        output << "<!--#{node.text}-->"
+      end
+
+      # @param [Oga::XML::ProcessingInstruction] node
+      # @param [String] output
+      def on_processing_instruction(node, output)
+        output << "<?#{node.name}#{node.text}?>"
+      end
+
+      # @param [Oga::XML::Element] element
+      # @param [String] body The content of the element.
+      def on_element(element, output)
+        name = element.expanded_name
+        attrs = ''
+
+        element.attributes.each do |attr|
+          attrs << ' '
+          on_attribute(attr, attrs)
+        end
+
+        if self_closing?(element)
+          output << "<#{name}#{attrs} />"
+        else
+          output << "<#{name}#{attrs}>"
+        end
+      end
+
+      # @param [Oga::XML::Element] element
+      # @param [String] output
+      def after_element(element, output)
+        output << "</#{element.expanded_name}>" unless self_closing?(element)
+      end
+
+      # @param [Oga::XML::Attribute] attr
+      # @param [String] output
+      def on_attribute(attr, output)
+        name = attr.expanded_name
+        enc_value = attr.value ? Entities.encode_attribute(attr.value) : nil
+
+        output << %Q(#{name}="#{enc_value}")
+      end
+
+      # @param [Oga::XML::Doctype] node
+      # @param [String] output
+      def on_doctype(node, output)
+        output << "<!DOCTYPE #{node.name}"
+
+        output << " #{node.type}" if node.type
+        output << %Q{ "#{node.public_id}"} if node.public_id
+        output << %Q{ "#{node.system_id}"} if node.system_id
+        output << " [#{node.inline_rules}]" if node.inline_rules
+        output << '>'
+      end
+
+      # @param [Oga::XML::Document] node
+      # @param [String] output
+      def on_document(doc, output)
+        if doc.xml_declaration
+          on_xml_declaration(doc.xml_declaration, output)
+          output << "\n"
+        end
+
+        if doc.doctype
+          on_doctype(doc.doctype, output)
+          output << "\n"
+        end
+      end
+
+      # @param [Oga::XML::XmlDeclaration] node
+      # @param [String] output
+      def on_xml_declaration(node, output)
+        output << '<?xml'
+
+        [:version, :encoding, :standalone].each do |getter|
+          value = node.send(getter)
+
+          output << %Q{ #{getter}="#{value}"} if value
+        end
+
+        output << ' ?>'
+      end
+
+      # @param [Oga::XML::Element] element
+      # @return [TrueClass|FalseClass]
+      def self_closing?(element)
+        if @html_mode && !HTML_VOID_ELEMENTS.allow?(element.name)
+          false
+        else
+          element.children.empty?
+        end
+      end
+    end
+  end
+end

data/lib/oga/xml/node.rb

@@ -5,10 +5,17 @@ module Oga
     # nodes.
     class Node
       include Traversal
+      include ToXML
 
       # @return [Oga::XML::NodeSet]
       attr_reader :node_set
 
+      # @return [Oga::XML::Node]
+      attr_accessor :previous
+
+      # @return [Oga::XML::Node]
+      attr_accessor :next
+
       # @param [Hash] options
       #
       # @option options [Oga::XML::NodeSet] :node_set The node set that this
@@ -26,6 +33,8 @@ module Oga
         @node_set  = set
         @root_node = nil
         @html_p    = nil
+        @previous  = nil
+        @next      = nil
       end
 
       # Returns the child nodes of the current node.
@@ -54,25 +63,6 @@ module Oga
         node_set ? node_set.owner : nil
       end
 
-      # Returns the preceding node, or nil if there is none.
-      #
-      # @return [Oga::XML::Node]
-      def previous
-        index = node_set.index(self) - 1
-
-        index >= 0 ? node_set[index] : nil
-      end
-
-      # Returns the following node, or nil if there is none.
-      #
-      # @return [Oga::XML::Node]
-      def next
-        index  = node_set.index(self) + 1
-        length = node_set.length
-
-        index <= length ? node_set[index] : nil
-      end
-
       # Returns the previous element node or nil if there is none.
       #
       # @return [Oga::XML::Element]

data/lib/oga/xml/node_set.rb

@@ -42,10 +42,10 @@ module Oga
         @owner    = owner
         @existing = {}
 
-        @nodes.each do |node|
+        @nodes.each_with_index do |node, index|
           mark_existing(node)
 
-          take_ownership(node) if @owner
+          take_ownership(node, index) if @owner
         end
       end
 
@@ -98,7 +98,7 @@ module Oga
 
         mark_existing(node)
 
-        take_ownership(node) if @owner
+        take_ownership(node, length - 1) if @owner
       end
 
       alias_method :<<, :push
@@ -113,7 +113,7 @@ module Oga
 
         mark_existing(node)
 
-        take_ownership(node) if @owner
+        take_ownership(node, 0) if @owner
       end
 
       # Shifts a node from the start of the set.
@@ -157,7 +157,7 @@ module Oga
 
         mark_existing(node)
 
-        take_ownership(node) if @owner
+        take_ownership(node, index) if @owner
       end
 
       # Returns the node for the given index.
@@ -224,6 +224,8 @@ module Oga
             sets << node.node_set
 
             node.node_set = nil
+            node.next     = nil
+            node.previous = nil
           end
         end
 
@@ -291,15 +293,34 @@ module Oga
       # set has an owner.
       #
       # @param [Oga::XML::Node] node
-      def take_ownership(node)
+      # @param [Fixnum] index
+      def take_ownership(node, index)
         node.node_set = self
+
+        node.previous = index > 0 ? @nodes[index - 1] : nil
+        node.next = index + 1 < @nodes.length ? @nodes[index + 1] : nil
+
+        node.previous.next = node if node.previous
+        node.next.previous = node if node.next
       end
 
       # Removes ownership of the node if it belongs to the current set.
       #
       # @param [Oga::XML::Node] node
       def remove_ownership(node)
-        node.node_set = nil if node.node_set == self
+        return unless node.node_set == self
+
+        if previous_node = node.previous
+          previous_node.next = node.next
+        end
+
+        if next_node = node.next
+          next_node.previous = node.previous
+        end
+
+        node.node_set = nil
+        node.previous = nil
+        node.next     = nil
       end
 
       # @param [Oga::XML::Node|Oga::XML::Attribute] node

data/lib/oga/xml/processing_instruction.rb

@@ -15,11 +15,6 @@ module Oga
         @name = options[:name]
       end
 
-      # @return [String]
-      def to_xml
-        "<?#{name}#{text}?>"
-      end
-
       # @return [String]
       def inspect
         "ProcessingInstruction(name: #{name.inspect} text: #{text.inspect})"

data/lib/oga/xml/text.rb

@@ -28,15 +28,6 @@ module Oga
         @text
       end
 
-      # @see [Oga::XML::CharacterNode#to_xml]
-      def to_xml
-        return super if inside_literal_html?
-
-        Entities.encode(super)
-      end
-
-      private
-
       # @return [TrueClass|FalseClass]
       def decode_entities?
         !@decoded && !inside_literal_html?
@@ -46,8 +37,7 @@ module Oga
       def inside_literal_html?
         node = parent
 
-        node.is_a?(Element) && html? &&
-          Lexer::LITERAL_HTML_ELEMENTS.allow?(node.name)
+        node && html? && node.literal_html_name?
       end
     end # Text
   end # XML

data/lib/oga/xml/to_xml.rb

@@ -0,0 +1,12 @@
+module Oga
+  module XML
+    # Module that provides a `#to_xml` method that serialises the current node
+    # back to XML.
+    module ToXML
+      # @return [String]
+      def to_xml
+        Generator.new(self).to_xml
+      end
+    end
+  end
+end

data/lib/oga/xml/xml_declaration.rb

@@ -2,6 +2,8 @@ module Oga
   module XML
     # Class containing information about an XML declaration tag.
     class XmlDeclaration
+      include ToXML
+
       # @return [String]
       attr_accessor :version
 
@@ -23,21 +25,6 @@ module Oga
         @standalone = options[:standalone]
       end
 
-      # Converts the declaration tag to XML.
-      #
-      # @return [String]
-      def to_xml
-        pairs = []
-
-        [:version, :encoding, :standalone].each do |getter|
-          value = send(getter)
-
-          pairs << %Q{#{getter}="#{value}"} if value
-        end
-
-        "<?xml #{pairs.join(' ')} ?>"
-      end
-
       # @return [String]
       def inspect
         segments = []

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: oga
 version: !ruby/object:Gem::Version
-  version: '2.3'
+  version: '2.4'
 platform: ruby
 authors:
 - Yorick Peterse
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-07-13 00:00:00.000000000 Z
+date: 2016-09-04 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ast
@@ -185,6 +185,7 @@ files:
 - lib/oga/xml/element.rb
 - lib/oga/xml/entities.rb
 - lib/oga/xml/expanded_name.rb
+- lib/oga/xml/generator.rb
 - lib/oga/xml/html_void_elements.rb
 - lib/oga/xml/lexer.rb
 - lib/oga/xml/namespace.rb
@@ -196,6 +197,7 @@ files:
 - lib/oga/xml/querying.rb
 - lib/oga/xml/sax_parser.rb
 - lib/oga/xml/text.rb
+- lib/oga/xml/to_xml.rb
 - lib/oga/xml/traversal.rb
 - lib/oga/xml/xml_declaration.rb
 - lib/oga/xpath/compiler.rb
@@ -229,4 +231,3 @@ signing_key:
 specification_version: 4
 summary: Oga is an XML/HTML parser written in Ruby.
 test_files: []
-has_rdoc: yard