spk-html5 0.10.1

data/lib/html5/html5parser/in_caption_phase.rb

@@ -0,0 +1,69 @@
+require 'html5/html5parser/phase'
+
+module HTML5
+  class InCaptionPhase < Phase
+
+    # http://www.whatwg.org/specs/web-apps/current-work/#in-caption
+
+    handle_start 'html', %w(caption col colgroup tbody td tfoot th thead tr) => 'TableElement'
+
+    handle_end 'caption', 'table', %w(body col colgroup html tbody td tfoot th thead tr) => 'Ignore'
+
+    def ignoreEndTagCaption
+      !in_scope?('caption', true)
+    end
+
+    def processCharacters(data)
+      @parser.phases[:inBody].processCharacters(data)
+    end
+
+    def startTagTableElement(name, attributes)
+      parse_error "unexpected-end-tag", {"name" => name}
+      #XXX Have to duplicate logic here to find out if the tag is ignored
+      ignoreEndTag = ignoreEndTagCaption
+      @parser.phase.processEndTag('caption')
+      @parser.phase.processStartTag(name, attributes) unless ignoreEndTag
+    end
+
+    def startTagOther(name, attributes)
+      @parser.phases[:inBody].processStartTag(name, attributes)
+    end
+
+    def endTagCaption(name)
+      if ignoreEndTagCaption
+        # inner_html case
+        assert @parser.inner_html
+        parse_error "unexpected-end-tag", {"name" => name}
+      else
+        # AT this code is quite similar to endTagTable in "InTable"
+        @tree.generateImpliedEndTags
+
+        unless @tree.open_elements[-1].name == 'caption'
+          parse_error("expected-one-end-tag-but-got-another",
+                    {"gotName" => "caption",
+                     "expectedName" => @tree.open_elements.last.name})
+        end
+
+        remove_open_elements_until('caption')
+
+        @tree.clearActiveFormattingElements
+        @parser.phase = @parser.phases[:inTable]
+      end
+    end
+
+    def endTagTable(name)
+      parse_error "unexpected-end-table-in-caption"
+      ignoreEndTag = ignoreEndTagCaption
+      @parser.phase.processEndTag('caption')
+      @parser.phase.processEndTag(name) unless ignoreEndTag
+    end
+
+    def endTagIgnore(name)
+      parse_error("unexpected-end-tag", {"name" => name})
+    end
+
+    def endTagOther(name)
+      @parser.phases[:inBody].processEndTag(name)
+    end
+  end
+end

data/lib/html5/html5parser/in_cell_phase.rb

@@ -0,0 +1,78 @@
+require 'html5/html5parser/phase'
+
+module HTML5
+  class InCellPhase < Phase
+
+    # http://www.whatwg.org/specs/web-apps/current-work/#in-cell
+
+    handle_start 'html', %w( caption col colgroup tbody td tfoot th thead tr ) => 'TableOther'
+
+    handle_end %w( td th ) => 'TableCell', %w( body caption col colgroup html ) => 'Ignore'
+
+    handle_end %w( table tbody tfoot thead tr ) => 'Imply'
+
+    def processCharacters(data)
+      @parser.phases[:inBody].processCharacters(data)
+    end
+
+    def startTagTableOther(name, attributes)
+      if in_scope?('td', true) or in_scope?('th', true)
+        closeCell
+        @parser.phase.processStartTag(name, attributes)
+      else
+        # inner_html case
+        parse_error
+      end
+    end
+
+    def startTagOther(name, attributes)
+      @parser.phases[:inBody].processStartTag(name, attributes)
+    end
+
+    def endTagTableCell(name)
+      if in_scope?(name, true)
+        @tree.generateImpliedEndTags(name)
+        if @tree.open_elements.last.name != name
+          parse_error("unexpected-cell-end-tag", {"name" => name})
+
+          remove_open_elements_until(name)
+        else
+          @tree.open_elements.pop
+        end
+        @tree.clearActiveFormattingElements
+        @parser.phase = @parser.phases[:inRow]
+      else
+        parse_error("unexpected-end-tag", {"name" => name})
+      end
+    end
+
+    def endTagIgnore(name)
+      parse_error("unexpected-end-tag", {"name" => name})
+    end
+
+    def endTagImply(name)
+      if in_scope?(name, true)
+        closeCell
+        @parser.phase.processEndTag(name)
+      else
+        # sometimes inner_html case
+        parse_error "unexpected-end-tag", {:name => name}
+      end
+    end
+
+    def endTagOther(name)
+      @parser.phases[:inBody].processEndTag(name)
+    end
+
+    protected
+
+    def closeCell
+      if in_scope?('td', true)
+        endTagTableCell('td')
+      elsif in_scope?('th', true)
+        endTagTableCell('th')
+      end
+    end
+
+  end
+end

data/lib/html5/html5parser/in_column_group_phase.rb

@@ -0,0 +1,55 @@
+require 'html5/html5parser/phase'
+
+module HTML5
+  class InColumnGroupPhase < Phase
+
+    # http://www.whatwg.org/specs/web-apps/current-work/#in-column
+
+    handle_start 'html', 'col'
+
+    handle_end 'colgroup', 'col'
+
+    def ignoreEndTagColgroup
+      @tree.open_elements[-1].name == 'html'
+    end
+
+    def processCharacters(data)
+      ignoreEndTag = ignoreEndTagColgroup
+      endTagColgroup("colgroup")
+      @parser.phase.processCharacters(data) unless ignoreEndTag
+    end
+
+    def startTagCol(name, attributes)
+      @tree.insert_element(name, attributes)
+      @tree.open_elements.pop
+    end
+
+    def startTagOther(name, attributes)
+      ignoreEndTag = ignoreEndTagColgroup
+      endTagColgroup('colgroup')
+      @parser.phase.processStartTag(name, attributes) unless ignoreEndTag
+    end
+
+    def endTagColgroup(name)
+      if ignoreEndTagColgroup
+        # inner_html case
+        assert @parser.inner_html
+        parse_error "unexpected-end-tag", {:name => name}
+      else
+        @tree.open_elements.pop
+        @parser.phase = @parser.phases[:inTable]
+      end
+    end
+
+    def endTagCol(name)
+      parse_error("no-end-tag", {"name" => "col"})
+    end
+
+    def endTagOther(name)
+      ignoreEndTag = ignoreEndTagColgroup
+      endTagColgroup('colgroup')
+      @parser.phase.processEndTag(name) unless ignoreEndTag
+    end
+
+  end
+end

data/lib/html5/html5parser/in_foreign_content_phase.rb

@@ -0,0 +1,50 @@
+require 'html5/html5parser/phase'
+
+module HTML5
+  class InForeignContentPhase < Phase
+
+    def processCharacters(data)
+      @tree.insertText(data)
+    end
+
+    def startTagOther(name, attributes, self_closing)
+      if !%w[mglyph malignmark].include?(name) && %w[mi mo mn ms mtext].include?(@tree.open_elements.last.name) &&
+         @tree.open_elements.last.namespace == :math
+
+        @parser.secondary_phase.processStartTag(name, attributes)
+        if @parser.phase == @parser.phases[:inForeignContent]
+          if !@tree.open_elements.any? {|e| e.namespace }
+            @parser.phase = @parser.secondary_phase
+          end
+        end
+      elsif %w[b big blockquote body br center code dd div dl dt em embed font
+          h1 h2 h3 h4 h5 h6 head hr i img li listing menu meta nobr ol p pre ruby s small
+          span strong strike sub sup table tt u ul var].include?(name)
+
+        parse_error("html-in-foreign-content", :name => name)
+
+        until @tree.open_elements.last.namespace == nil
+          @tree.open_elements.pop
+        end
+        @parser.phase = @parser.secondary_phase
+        @parser.phase.processStartTag(name, attributes)
+      else
+        if @tree.open_elements.last.namespace == :math
+          attribtues = adjust_mathml_attributes(attributes)
+        end
+        attributes = adjust_foreign_attributes(attributes)
+        @tree.insert_foreign_element(name, attributes, @tree.open_elements.last.namespace)
+        @tree.open_elements.pop if self_closing
+      end
+    end
+
+    def endTagOther(name)
+      @parser.secondary_phase.processEndTag(name)
+      if @parser.phase == @parser.phases[:inForeignContent]
+        if !@tree.open_elements.any? {|e| e.namespace }
+          @parser.phase = @parser.secondary_phase
+        end
+      end
+    end
+  end
+end

data/lib/html5/html5parser/in_frameset_phase.rb

@@ -0,0 +1,56 @@
+require 'html5/html5parser/phase'
+
+module HTML5
+  class InFramesetPhase < Phase
+
+    # http://www.whatwg.org/specs/web-apps/current-work/#in-frameset
+
+    handle_start 'html', 'frameset', 'frame', 'noframes'
+
+    handle_end 'frameset', 'noframes'
+
+    def processCharacters(data)
+      parse_error("unexpected-char-in-frameset")
+    end
+
+    def startTagFrameset(name, attributes)
+      @tree.insert_element(name, attributes)
+    end
+
+    def startTagFrame(name, attributes)
+      @tree.insert_element(name, attributes)
+      @tree.open_elements.pop
+    end
+
+    def startTagNoframes(name, attributes)
+      @parser.phases[:inBody].processStartTag(name, attributes)
+    end
+
+    def startTagOther(name, attributes)
+      parse_error("unexpected-start-tag-in-frameset", {"name" => name})
+    end
+
+    def endTagFrameset(name)
+      if @tree.open_elements.last.name == 'html'
+        # inner_html case
+        parse_error("unexpected-frameset-in-frameset-innerhtml")
+      else
+        @tree.open_elements.pop
+      end
+      if (not @parser.inner_html and
+        @tree.open_elements.last.name != 'frameset')
+        # If we're not in inner_html mode and the the current node is not a
+        # "frameset" element (anymore) then switch.
+        @parser.phase = @parser.phases[:afterFrameset]
+      end
+    end
+
+    def endTagNoframes(name)
+      @parser.phases[:inBody].processEndTag(name)
+    end
+
+    def endTagOther(name)
+      parse_error("unexpected-end-tag-in-frameset", {"name" => name})
+    end
+  end
+end

data/lib/html5/html5parser/in_head_phase.rb

@@ -0,0 +1,143 @@
+require 'html5/html5parser/phase'
+
+module HTML5
+  class InHeadPhase < Phase
+
+    handle_start 'html', 'head', 'title', 'style', 'script', 'noscript'
+    handle_start %w( base link meta)
+
+    handle_end 'head'
+    handle_end %w( html body br ) => 'ImplyAfterHead'
+    handle_end %w( title style script noscript )
+
+    def process_eof
+      if ['title', 'style', 'script'].include?(name = @tree.open_elements.last.name)
+        parse_error("expected-named-closing-tag-but-got-eof", {"name" => @tree.open_elements.last.name})
+        @tree.open_elements.pop
+      end
+      anything_else
+      @parser.phase.process_eof
+    end
+
+    def processCharacters(data)
+      if %w[title style script noscript].include?(@tree.open_elements.last.name)
+        @tree.insertText(data)
+      else
+        anything_else
+        @parser.phase.processCharacters(data)
+      end
+    end
+
+    def startTagHead(name, attributes)
+      parse_error("two-heads-are-not-better-than-one")
+    end
+
+    def startTagTitle(name, attributes)
+      if @tree.head_pointer != nil && @parser.phase == @parser.phases[:inHead]
+        element = @tree.createElement(name, attributes)
+        appendToHead(element)
+        @tree.open_elements << element
+      else
+        @tree.insert_element(name, attributes)
+      end
+      @parser.tokenizer.content_model_flag = :RCDATA
+    end
+
+    def startTagStyle(name, attributes)
+      if @tree.head_pointer != nil and @parser.phase == @parser.phases[:inHead]
+        element = @tree.createElement(name, attributes)
+        appendToHead(element)
+        @tree.open_elements.push(element)
+      else
+        @tree.insert_element(name, attributes)
+      end
+      @parser.tokenizer.content_model_flag = :CDATA
+    end
+
+    def startTagNoscript(name, attributes)
+      # XXX Need to decide whether to implement the scripting disabled case.
+      element = @tree.createElement(name, attributes)
+      if @tree.head_pointer !=nil and @parser.phase == @parser.phases[:inHead]
+        appendToHead(element)
+      else
+        @tree.open_elements.last.appendChild(element)
+      end
+      @tree.open_elements.push(element)
+      @parser.tokenizer.content_model_flag = :CDATA
+    end
+
+    def startTagScript(name, attributes)
+      #XXX Inner HTML case may be wrong
+      element = @tree.createElement(name, attributes)
+      element.flags.push("parser-inserted")
+      if @tree.head_pointer != nil and @parser.phase == @parser.phases[:inHead]
+        appendToHead(element)
+      else
+        @tree.open_elements.last.appendChild(element)
+      end
+      @tree.open_elements.push(element)
+      @parser.tokenizer.content_model_flag = :CDATA
+    end
+
+    def startTagBaseLinkMeta(name, attributes)
+      if @tree.head_pointer != nil and @parser.phase == @parser.phases[:inHead]
+        element = @tree.createElement(name, attributes)
+        appendToHead(element)
+      else
+        @tree.insert_element(name, attributes)
+        @tree.open_elements.pop
+      end
+    end
+    
+    def startTagOther(name, attributes)
+      anything_else
+      @parser.phase.processStartTag(name, attributes)
+    end
+
+    def endTagHead(name)
+      if @tree.open_elements.last.name == 'head'
+        @tree.open_elements.pop
+      else
+        parse_error("unexpected-end-tag", {"name" => "head"})
+      end
+      @parser.phase = @parser.phases[:afterHead]
+    end
+
+    def endTagImplyAfterHead(name)
+      anything_else
+      @parser.phase.processEndTag(name)
+    end
+
+    def endTagTitleStyleScriptNoscript(name)
+      if @tree.open_elements.last.name == name
+        @tree.open_elements.pop
+      else
+        parse_error("unexpected-end-tag", {"name" => name})
+      end
+    end
+
+    def endTagOther(name)
+      anything_else
+    end
+
+    def anything_else
+      if @tree.open_elements.last.name == 'head'
+        endTagHead('head')
+      else
+        @parser.phase = @parser.phases[:afterHead]
+      end
+    end
+
+    protected
+
+    def appendToHead(element)
+      if @tree.head_pointer.nil?
+        assert @parser.inner_html
+        @tree.open_elements.last.appendChild(element)
+      else
+        @tree.head_pointer.appendChild(element)
+      end
+    end
+
+  end
+end