spk-html5 0.10.1

data/lib/html5/treebuilders.rb

@@ -0,0 +1,24 @@
+module HTML5
+  module TreeBuilders
+
+    class << self
+      def [](name)
+        case name.to_s.downcase
+        when 'simpletree' then
+          require 'html5/treebuilders/simpletree'
+          SimpleTree::TreeBuilder
+        when 'rexml' then
+          require 'html5/treebuilders/rexml'
+          REXML::TreeBuilder
+        when 'hpricot' then
+          require 'html5/treebuilders/hpricot'
+          Hpricot::TreeBuilder
+        else
+          raise "Unknown TreeBuilder #{name}"
+        end
+      end
+
+      alias :get_tree_builder :[]
+    end
+  end
+end

data/lib/html5/treebuilders/base.rb

@@ -0,0 +1,339 @@
+require 'html5/constants'
+
+#XXX - TODO; make the default interface more ElementTree-like rather than DOM-like
+
+module HTML5
+
+  # The scope markers are inserted when entering buttons, object elements,
+  # marquees, table cells, and table captions, and are used to prevent formatting
+  # from "leaking" into tables, buttons, object elements, and marquees.
+  Marker = nil
+
+  module TreeBuilders
+    module Base
+
+      class Node
+        # The parent of the current node (or nil for the document node)
+        attr_accessor :parent
+
+        # a list of child nodes of the current node. This must 
+        # include all elements but not necessarily other node types
+        attr_accessor :childNodes
+
+        # A list of miscellaneous flags that can be set on the node
+        attr_accessor :flags
+
+        def initialize(name)
+          @parent     = nil
+          @childNodes = []
+          @flags     = []
+        end
+
+        # Insert node as a child of the current node
+        def appendChild(node)
+          raise NotImplementedError
+        end
+
+        # Insert data as text in the current node, positioned before the 
+        # start of node insertBefore or to the end of the node's text.
+        def insertText(data, insertBefore=nil)
+          raise NotImplementedError
+        end
+
+        # Insert node as a child of the current node, before refNode in the 
+        # list of child nodes. Raises ValueError if refNode is not a child of 
+        # the current node
+        def insertBefore(node, refNode)
+          raise NotImplementedError
+        end
+
+        # Remove node from the children of the current node
+        def removeChild(node)
+          raise NotImplementedError
+        end
+
+        # Move all the children of the current node to newParent. 
+        # This is needed so that trees that don't store text as nodes move the 
+        # text in the correct way
+        def reparentChildren(newParent)
+          #XXX - should this method be made more general?
+          @childNodes.each { |child| newParent.appendChild(child) }
+          @childNodes = []
+        end
+
+        # Return a shallow copy of the current node i.e. a node with the same
+        # name and attributes but with no parent or child nodes
+        def cloneNode
+          raise NotImplementedError
+        end
+
+        # Return true if the node has children or text, false otherwise
+        def hasContent
+          raise NotImplementedError
+        end
+      end
+
+      # Base treebuilder implementation
+      class TreeBuilder
+
+        attr_accessor :open_elements
+
+        attr_accessor :activeFormattingElements
+
+        attr_accessor :document
+
+        attr_accessor :head_pointer
+
+        attr_accessor :formPointer
+
+        attr_reader :insert_from_table
+        # Class to use for document root
+        documentClass = nil
+
+        # Class to use for HTML elements
+        elementClass = nil
+
+        # Class to use for comments
+        commentClass = nil
+
+        # Class to use for doctypes
+        doctypeClass = nil
+
+        # Fragment class
+        fragmentClass = nil
+
+        def initialize
+          reset
+        end
+
+        def reset
+          @open_elements = []
+          @activeFormattingElements = []
+
+          #XXX - rename these to headElement, formElement
+          @head_pointer = nil
+          @formPointer = nil
+
+          self.insert_from_table = false
+
+          @document = @documentClass.new
+        end
+
+        def elementInScope(target, tableVariant=false)
+          # Exit early when possible.
+          return true if @open_elements[-1] && @open_elements[-1].name == target
+          return false if @open_elements.length == 0
+          # AT How about while true and simply set node to [-1] and set it to
+          # [-2] at the end...
+          @open_elements.reverse.each do |element|
+            if element.name == target
+              return true
+            elsif element.name == 'table'
+              return false
+            elsif not tableVariant and SCOPING_ELEMENTS.include?(element.name)
+              return false
+            elsif element.name == 'html'
+              return false
+            end
+          end
+          assert false # We should never reach this point
+        end
+
+        def reconstructActiveFormattingElements
+          # Within this algorithm the order of steps described in the
+          # specification is not quite the same as the order of steps in the
+          # code. It should still do the same though.
+
+          # Step 1: stop the algorithm when there's nothing to do.
+          return if @activeFormattingElements.empty?
+
+          # Step 2 and step 3: we start with the last element. So i is -1.
+          i = -1
+          entry = @activeFormattingElements[i]
+          return if entry == Marker or @open_elements.include?(entry)
+
+          # Step 6
+          until entry == Marker or @open_elements.include?(entry)
+            # Step 5: let entry be one earlier in the list.
+            i -= 1
+            begin
+              entry = @activeFormattingElements[i]
+            rescue
+              # Step 4: at this point we need to jump to step 8. By not doing
+              # i += 1 which is also done in step 7 we achieve that.
+              break
+            end
+          end
+          while true
+            # Step 7
+            i += 1
+
+            # Step 8
+            clone = @activeFormattingElements[i].cloneNode
+
+            # Step 9
+            element = insert_element(clone.name, clone.attributes)
+
+            # Step 10
+            @activeFormattingElements[i] = element
+
+            # Step 11
+            break if element == @activeFormattingElements[-1]
+          end
+        end
+
+        def clearActiveFormattingElements
+          {} until @activeFormattingElements.empty? || @activeFormattingElements.pop == Marker
+        end
+
+        # Check if an element exists between the end of the active
+        # formatting elements and the last marker. If it does, return it, else
+        # return false
+        def elementInActiveFormattingElements(name)
+          @activeFormattingElements.reverse.each do |element|
+            # Check for Marker first because if it's a Marker it doesn't have a
+            # name attribute.
+            break if element == Marker
+            return element if element.name == name
+          end
+          return false
+        end
+
+        def insertDoctype(name, public_id, system_id)
+          doctype = @doctypeClass.new(name)
+          doctype.public_id = public_id
+          doctype.system_id = system_id
+          @document.appendChild(doctype)
+        end
+
+        def insert_comment(data, parent=nil)
+          parent = @open_elements[-1] if parent.nil?
+          parent.appendChild(@commentClass.new(data))
+        end
+
+        # Create an element but don't insert it anywhere
+        def createElement(name, attributes)
+          element = @elementClass.new(name)
+          element.attributes = attributes
+          return element
+        end
+
+        # Switch the function used to insert an element from the
+        # normal one to the misnested table one and back again
+        def insert_from_table=(value)
+          @insert_from_table = value
+          @insert_element = value ? :insert_elementTable : :insert_elementNormal
+        end
+
+        def insert_element(name, attributes, namespace = nil)
+          send(@insert_element, name, attributes, namespace)
+        end
+        
+        def insert_foreign_element(name, attributes, namespace)
+          insert_element(name, attributes, namespace)
+        end
+
+        def insert_elementNormal(name, attributes, namespace=nil)
+          element = @elementClass.new(name, namespace)
+          element.attributes = attributes
+          @open_elements.last.appendChild(element)
+          @open_elements.push(element)
+          element
+        end
+
+        # Create an element and insert it into the tree
+        def insert_elementTable(name, attributes, namespace=nil)
+          element = @elementClass.new(name, namespace)
+          element.attributes = attributes
+          if !TABLE_INSERT_MODE_ELEMENTS.include?(@open_elements.last.name)
+            return insert_elementNormal(name, attributes)
+          else
+            #We should be in the InTable mode. This means we want to do
+            #special magic element rearranging
+            parent, insertBefore = getTableMisnestedNodePosition
+            if insertBefore.nil?
+              parent.appendChild(element)
+            else
+              parent.insertBefore(element, insertBefore)
+            end
+            @open_elements.push(element)
+          end
+          return element
+        end
+
+        def insertText(data, parent=nil)
+          parent = @open_elements[-1] if parent.nil?
+
+          if (not(@insert_from_table) or (@insert_from_table and not TABLE_INSERT_MODE_ELEMENTS.include?(@open_elements[-1].name)))
+            parent.insertText(data)
+          else
+            #We should be in the InTable mode. This means we want to do
+            #special magic element rearranging
+            parent, insertBefore = getTableMisnestedNodePosition
+            parent.insertText(data, insertBefore)
+          end
+        end
+
+        # Get the foster parent element, and sibling to insert before
+        # (or nil) when inserting a misnested table node
+        def getTableMisnestedNodePosition
+          #The foster parent element is the one which comes before the most
+          #recently opened table element
+          #XXX - this is really inelegant
+          lastTable = nil
+          fosterParent = nil
+          insertBefore = nil
+          @open_elements.reverse.each do |element|
+            if element.name == "table"
+              lastTable = element
+              break
+            end
+          end
+          if lastTable
+            #XXX - we should really check that this parent is actually a
+            #node here
+            if lastTable.parent
+              fosterParent = lastTable.parent
+              insertBefore = lastTable
+            else
+              fosterParent = @open_elements[@open_elements.index(lastTable) - 1]
+            end
+          else
+            fosterParent = @open_elements[0]
+          end
+          return fosterParent, insertBefore
+        end
+
+        def generateImpliedEndTags(exclude=nil)
+          name = @open_elements[-1].name
+
+          # XXX td, th and tr are not actually needed
+          if (%w[dd dt li p td th tr].include?(name) and name != exclude)
+            @open_elements.pop
+            # XXX This is not entirely what the specification says. We should
+            # investigate it more closely.
+            generateImpliedEndTags(exclude)
+          end
+        end
+
+        def get_document
+          @document
+        end
+  
+        def get_fragment
+          #assert @inner_html
+          fragment = @fragmentClass.new
+          @open_elements[0].reparentChildren(fragment)
+          return fragment
+        end
+
+        # Serialize the subtree of node in the format required by unit tests
+        # node - the node from which to start serializing
+        def testSerializer(node)
+          raise NotImplementedError
+        end
+
+      end
+    end
+  end
+end

data/lib/html5/treebuilders/hpricot.rb

@@ -0,0 +1,231 @@
+require 'html5/treebuilders/base'
+require 'rubygems'
+require 'hpricot'
+require 'forwardable'
+
+module HTML5
+  module TreeBuilders
+    module Hpricot
+
+      class Node < Base::Node
+        extend Forwardable
+
+        def_delegators :@hpricot, :name
+
+        attr_accessor :hpricot
+
+        def initialize(name)
+          super(name)
+          @hpricot = self.class.hpricot_class.new name
+        end
+
+        def appendChild(node)
+          if node.kind_of?(TextNode) and childNodes.any? and childNodes.last.kind_of?(TextNode)
+            childNodes.last.hpricot.content = childNodes.last.hpricot.content + node.hpricot.content
+          else
+            childNodes << node
+            hpricot.children << node.hpricot
+          end
+          if (oldparent = node.hpricot.parent) != nil
+            oldparent.children.delete_at(oldparent.children.index(node.hpricot))
+          end
+          node.hpricot.parent = hpricot
+          node.parent = self
+        end
+
+        def removeChild(node)
+           childNodes.delete(node)
+           hpricot.children.delete_at(hpricot.children.index(node.hpricot))
+           node.hpricot.parent = nil
+           node.parent = nil
+        end
+
+        def insertText(data, before=nil)
+          if before
+            insertBefore(TextNode.new(data), before)
+          else
+            appendChild(TextNode.new(data))
+          end
+        end
+
+        def insertBefore(node, refNode)
+          index = childNodes.index(refNode)
+          if node.kind_of?(TextNode) and index > 0 and childNodes[index-1].kind_of?(TextNode)
+            childNodes[index-1].hpricot.content = childNodes[index-1].hpricot.to_s + node.hpricot.to_s
+          else
+            refNode.hpricot.parent.insert_before(node.hpricot,refNode.hpricot)
+            childNodes.insert(index, node)
+          end
+        end
+
+        def hasContent
+          childNodes.any?
+        end
+      end
+
+      class Element < Node
+        attr_reader :namespace
+
+        def self.hpricot_class
+          ::Hpricot::Elem
+        end
+
+        def initialize(name, namespace=nil)
+          super(name)
+
+          @hpricot   = ::Hpricot::Elem.new(::Hpricot::STag.new(name))
+          @namespace = namespace
+        end
+
+        def name
+          @hpricot.stag.name
+        end
+
+        def cloneNode
+          attributes.inject(self.class.new(name)) do |node, (name, value)|
+            node.hpricot[name] = value
+            node
+          end
+        end
+
+        # A call to Hpricot::Elem#raw_attributes is built dynamically,
+        # so alterations to the returned value (a hash) will be lost.
+        #
+        # AttributeProxy works around this by forwarding :[]= calls
+        # to the raw_attributes accessor on the element start tag.
+        #
+        class AttributeProxy
+          def initialize(hpricot)
+            @hpricot = hpricot
+          end
+
+          def []=(k, v)
+            @hpricot.stag.send(stag_attributes_method)[k] = v
+          end
+
+          def stag_attributes_method
+            # STag#attributes changed to STag#raw_attributes after Hpricot 0.5
+            @hpricot.stag.respond_to?(:raw_attributes) ? :raw_attributes : :attributes
+          end
+
+          def method_missing(*a, &b)
+            @hpricot.attributes.send(*a, &b)
+          end
+        end
+
+        def attributes
+          AttributeProxy.new(@hpricot)
+        end
+
+        def attributes=(attrs)
+          attrs.each { |name, value| @hpricot[name] = value }
+        end
+
+        def printTree(indent=0)
+          tree = "\n|#{' ' * indent}<#{!@namespace.nil? ? @namespace.to_s + ' ' : ''}#{name}>"
+          indent += 2
+          attributes.each do |name, value|
+            next if name == 'xmlns'
+            tree += "\n|#{' ' * indent}#{name}=\"#{value}\""
+          end
+          childNodes.inject(tree) { |tree, child| tree + child.printTree(indent) }
+        end
+      end
+
+      class Document < Node
+        def self.hpricot_class
+          ::Hpricot::Doc
+        end
+
+        def initialize
+          super(nil)
+        end
+
+        def printTree(indent=0)
+          childNodes.inject('#document') { |tree, child| tree + child.printTree(indent + 2) }
+        end
+      end
+
+      class DocumentType < Node
+        def_delegators :@hpricot, :public_id, :system_id
+
+        def self.hpricot_class
+          ::Hpricot::DocType
+        end
+
+        def initialize(name, public_id, system_id)
+          @hpricot = self.class.hpricot_class.new(name, public_id, system_id)
+        end
+
+        def printTree(indent=0)
+          if hpricot.target and hpricot.target.any?
+            "\n|#{' ' * indent}<!DOCTYPE #{hpricot.target}" +
+              ([hpricot.public_id, hpricot.system_id].any? ? " \"#{hpricot.public_id}\" \"#{hpricot.system_id}\"" : '') +
+              '>'
+          else
+            "\n|#{' ' * indent}<!DOCTYPE >"
+          end
+        end
+      end
+
+      class DocumentFragment < Element
+        def initialize
+          super('')
+        end
+
+        def printTree(indent=0)
+          childNodes.inject('') {|tree, child| tree + child.printTree(indent + 2) }
+        end
+      end
+
+      class TextNode < Node
+        def initialize(data)
+          @hpricot = ::Hpricot::Text.new(data)
+        end
+
+        def printTree(indent=0)
+          "\n|#{' ' * indent}\"#{hpricot.content}\""
+        end
+      end
+
+      class CommentNode < Node
+        def self.hpricot_class
+          ::Hpricot::Comment
+        end
+
+        def printTree(indent=0)
+          "\n|#{' ' * indent}<!-- #{hpricot.content} -->"
+        end
+      end
+
+      class TreeBuilder < Base::TreeBuilder
+        def initialize
+          @documentClass = Document
+          @doctypeClass  = DocumentType
+          @elementClass  = Element
+          @commentClass  = CommentNode
+          @fragmentClass = DocumentFragment
+        end
+
+        def insertDoctype(name, public_id, system_id)
+          doctype = @doctypeClass.new(name, public_id, system_id)
+          @document.appendChild(doctype)
+        end
+
+        def testSerializer(node)
+          node.printTree
+        end
+
+        def get_document
+          @document.hpricot
+        end
+
+        def get_fragment
+          @document = super
+          return @document.hpricot.children
+        end
+      end
+
+    end
+  end
+end