html5 0.1.0

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("Got table cell end tag (#{name}) while required end tags are missing.")
+
+          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}). Ignored."))
+      end
+    end
+
+    def endTagIgnore(name)
+      parse_error(_("Unexpected end tag (#{name}). Ignored."))
+    end
+
+    def endTagImply(name)
+      if in_scope?(name, true)
+        closeCell
+        @parser.phase.processEndTag(name)
+      else
+        # sometimes inner_html case
+        parse_error
+      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
+      else
+        @tree.open_elements.pop
+        @parser.phase = @parser.phases[:inTable]
+      end
+    end
+
+    def endTagCol(name)
+      parse_error(_('Unexpected end tag (col). col has no end tag.'))
+    end
+
+    def endTagOther(name)
+      ignoreEndTag = ignoreEndTagColgroup
+      endTagColgroup('colgroup')
+      @parser.phase.processEndTag(name) unless ignoreEndTag
+    end
+
+  end
+end

data/lib/html5/html5parser/in_frameset_phase.rb

@@ -0,0 +1,57 @@
+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 characters in the frameset phase. Characters ignored.'))
+    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 token (#{name}) in the frameset phase. Ignored"))
+    end
+
+    def endTagFrameset(name)
+      if @tree.open_elements.last.name == 'html'
+        # inner_html case
+        parse_error(_("Unexpected end tag token (frameset) in the frameset phase (inner_html)."))
+      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 token (#{name}) in the frameset phase. Ignored."))
+    end
+
+  end
+end

data/lib/html5/html5parser/in_head_phase.rb

@@ -0,0 +1,138 @@
+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 p ) => 'ImplyAfterHead'
+    handle_end %w( title style script noscript )
+
+    def process_eof
+      if ['title', 'style', 'script'].include?(name = @tree.open_elements.last.name)
+        parse_error(_("Unexpected end of file. Expected end tag (#{name})."))
+        @tree.open_elements.pop
+      end
+      anythingElse
+      @parser.phase.process_eof
+    end
+
+    def processCharacters(data)
+      if %w[title style script noscript].include?(@tree.open_elements.last.name)
+        @tree.insertText(data)
+      else
+        anythingElse
+        @parser.phase.processCharacters(data)
+      end
+    end
+
+    def startTagHead(name, attributes)
+      parse_error(_('Unexpected start tag head in existing head. Ignored'))
+    end
+
+    def startTagTitle(name, attributes)
+      element = @tree.createElement(name, attributes)
+      appendToHead(element)
+      @tree.open_elements.push(element)
+      @parser.tokenizer.content_model_flag = :RCDATA
+    end
+
+    def startTagStyle(name, attributes)
+      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 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)
+      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
+    end
+
+    def startTagOther(name, attributes)
+      anythingElse
+      @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 (head). Ignored."))
+      end
+      @parser.phase = @parser.phases[:afterHead]
+    end
+
+    def endTagImplyAfterHead(name)
+      anythingElse
+      @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}). Ignored."))
+      end
+    end
+
+    def endTagOther(name)
+      parse_error(_("Unexpected end tag (#{name}). Ignored."))
+    end
+
+    def anythingElse
+      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

data/lib/html5/html5parser/in_row_phase.rb

@@ -0,0 +1,87 @@
+require 'html5/html5parser/phase'
+
+module HTML5
+  class InRowPhase < Phase
+
+    # http://www.whatwg.org/specs/web-apps/current-work/#in-row
+
+    handle_start 'html', %w( td th ) => 'TableCell', %w( caption col colgroup tbody tfoot thead tr ) => 'TableOther'
+
+    handle_end 'tr', 'table', %w( tbody tfoot thead ) => 'TableRowGroup', %w( body caption col colgroup html td th ) => 'Ignore'
+
+    def processCharacters(data)
+      @parser.phases[:inTable].processCharacters(data)
+    end
+
+    def startTagTableCell(name, attributes)
+      clearStackToTableRowContext
+      @tree.insert_element(name, attributes)
+      @parser.phase = @parser.phases[:inCell]
+      @tree.activeFormattingElements.push(Marker)
+    end
+
+    def startTagTableOther(name, attributes)
+      ignoreEndTag = ignoreEndTagTr
+      endTagTr('tr')
+      # XXX how are we sure it's always ignored in the inner_html case?
+      @parser.phase.processStartTag(name, attributes) unless ignoreEndTag
+    end
+
+    def startTagOther(name, attributes)
+      @parser.phases[:inTable].processStartTag(name, attributes)
+    end
+
+    def endTagTr(name)
+      if ignoreEndTagTr
+        # inner_html case
+        assert @parser.inner_html
+        parse_error
+      else
+        clearStackToTableRowContext
+        @tree.open_elements.pop
+        @parser.phase = @parser.phases[:inTableBody]
+      end
+    end
+
+    def endTagTable(name)
+      ignoreEndTag = ignoreEndTagTr
+      endTagTr('tr')
+      # Reprocess the current tag if the tr end tag was not ignored
+      # XXX how are we sure it's always ignored in the inner_html case?
+      @parser.phase.processEndTag(name) unless ignoreEndTag
+    end
+
+    def endTagTableRowGroup(name)
+      if in_scope?(name, true)
+        endTagTr('tr')
+        @parser.phase.processEndTag(name)
+      else
+        # inner_html case
+        parse_error
+      end
+    end
+
+    def endTagIgnore(name)
+      parse_error(_("Unexpected end tag (#{name}) in the row phase. Ignored."))
+    end
+
+    def endTagOther(name)
+      @parser.phases[:inTable].processEndTag(name)
+    end
+
+    protected
+
+    # XXX unify this with other table helper methods
+    def clearStackToTableRowContext
+      until %w[tr html].include?(name = @tree.open_elements.last.name)
+        parse_error(_("Unexpected implied end tag (#{name}) in the row phase."))
+        @tree.open_elements.pop
+      end
+    end
+
+    def ignoreEndTagTr
+      not in_scope?('tr', :tableVariant => true)
+    end
+
+  end
+end

data/lib/html5/html5parser/in_select_phase.rb

@@ -0,0 +1,84 @@
+require 'html5/html5parser/phase'
+
+module HTML5
+  class InSelectPhase < Phase
+
+    # http://www.whatwg.org/specs/web-apps/current-work/#in-select
+
+    handle_start 'html', 'option', 'optgroup', 'select'
+
+    handle_end 'option', 'optgroup', 'select', %w( caption table tbody tfoot thead tr td th ) => 'TableElements'
+
+    def processCharacters(data)
+      @tree.insertText(data)
+    end
+
+    def startTagOption(name, attributes)
+      # We need to imply </option> if <option> is the current node.
+      @tree.open_elements.pop if @tree.open_elements.last.name == 'option'
+      @tree.insert_element(name, attributes)
+    end
+
+    def startTagOptgroup(name, attributes)
+      @tree.open_elements.pop if @tree.open_elements.last.name == 'option'
+      @tree.open_elements.pop if @tree.open_elements.last.name == 'optgroup'
+      @tree.insert_element(name, attributes)
+    end
+
+    def startTagSelect(name, attributes)
+      parse_error(_('Unexpected start tag (select) in the select phase implies select start tag.'))
+      endTagSelect('select')
+    end
+
+    def startTagOther(name, attributes)
+      parse_error(_('Unexpected start tag token (#{name}) in the select phase. Ignored.'))
+    end
+
+    def endTagOption(name)
+      if @tree.open_elements.last.name == 'option'
+        @tree.open_elements.pop
+      else
+        parse_error(_('Unexpected end tag (option) in the select phase. Ignored.'))
+      end
+    end
+
+    def endTagOptgroup(name)
+      # </optgroup> implicitly closes <option>
+      if @tree.open_elements.last.name == 'option' and @tree.open_elements[-2].name == 'optgroup'
+        @tree.open_elements.pop
+      end
+      # It also closes </optgroup>
+      if @tree.open_elements.last.name == 'optgroup'
+        @tree.open_elements.pop
+      # But nothing else
+      else
+        parse_error(_('Unexpected end tag (optgroup) in the select phase. Ignored.'))
+      end
+    end
+
+    def endTagSelect(name)
+      if in_scope?('select', true)
+        remove_open_elements_until('select')
+
+        @parser.reset_insertion_mode
+      else
+        # inner_html case
+        parse_error
+      end
+    end
+
+    def endTagTableElements(name)
+      parse_error(_("Unexpected table end tag (#{name}) in the select phase."))
+
+      if in_scope?(name, true)
+        endTagSelect('select')
+        @parser.phase.processEndTag(name)
+      end
+    end
+
+    def endTagOther(name)
+      parse_error(_("Unexpected end tag token (#{name}) in the select phase. Ignored."))
+    end
+
+  end
+end

data/lib/html5/html5parser/in_table_body_phase.rb

@@ -0,0 +1,83 @@
+require 'html5/html5parser/phase'
+
+module HTML5
+  class InTableBodyPhase < Phase
+
+    # http://www.whatwg.org/specs/web-apps/current-work/#in-table0
+
+    handle_start 'html', 'tr', %w( td th ) => 'TableCell', %w( caption col colgroup tbody tfoot thead ) => 'TableOther'
+
+    handle_end 'table', %w( tbody tfoot thead ) => 'TableRowGroup', %w( body caption col colgroup html td th tr ) => 'Ingore'
+
+    def processCharacters(data)
+      @parser.phases[:inTable].processCharacters(data)
+    end
+
+    def startTagTr(name, attributes)
+      clearStackToTableBodyContext
+      @tree.insert_element(name, attributes)
+      @parser.phase = @parser.phases[:inRow]
+    end
+
+    def startTagTableCell(name, attributes)
+      parse_error(_("Unexpected table cell start tag (#{name}) in the table body phase."))
+      startTagTr('tr', {})
+      @parser.phase.processStartTag(name, attributes)
+    end
+
+    def startTagTableOther(name, attributes)
+      # XXX AT Any ideas on how to share this with endTagTable?
+      if in_scope?('tbody', true) or in_scope?('thead', true) or in_scope?('tfoot', true)
+        clearStackToTableBodyContext
+        endTagTableRowGroup(@tree.open_elements.last.name)
+        @parser.phase.processStartTag(name, attributes)
+      else
+        # inner_html case
+        parse_error
+      end
+    end
+
+    def startTagOther(name, attributes)
+      @parser.phases[:inTable].processStartTag(name, attributes)
+    end
+
+    def endTagTableRowGroup(name)
+      if in_scope?(name, true)
+        clearStackToTableBodyContext
+        @tree.open_elements.pop
+        @parser.phase = @parser.phases[:inTable]
+      else
+        parse_error(_("Unexpected end tag (#{name}) in the table body phase. Ignored."))
+      end
+    end
+
+    def endTagTable(name)
+      if in_scope?('tbody', true) or in_scope?('thead', true) or in_scope?('tfoot', true)
+        clearStackToTableBodyContext
+        endTagTableRowGroup(@tree.open_elements.last.name)
+        @parser.phase.processEndTag(name)
+      else
+        # inner_html case
+        parse_error
+      end
+    end
+
+    def endTagIgnore(name)
+      parse_error(_("Unexpected end tag (#{name}) in the table body phase. Ignored."))
+    end
+
+    def endTagOther(name)
+      @parser.phases[:inTable].processEndTag(name)
+    end
+
+    protected
+
+    def clearStackToTableBodyContext
+      until %w[tbody tfoot thead html].include?(name = @tree.open_elements.last.name)
+        parse_error(_("Unexpected implied end tag (#{name}) in the table body phase."))
+        @tree.open_elements.pop
+      end
+    end
+
+  end
+end

data/lib/html5/html5parser/in_table_phase.rb

@@ -0,0 +1,110 @@
+require 'html5/html5parser/phase'
+
+module HTML5
+  class InTablePhase < Phase
+
+    # http://www.whatwg.org/specs/web-apps/current-work/#in-table
+
+    handle_start 'html', 'caption', 'colgroup', 'col', 'table'
+
+    handle_start %w( tbody tfoot thead ) => 'RowGroup', %w( td th tr ) => 'ImplyTbody'
+
+    handle_end 'table', %w( body caption col colgroup html tbody td tfoot th thead tr ) => 'Ignore'
+
+    def processCharacters(data)
+      parse_error(_("Unexpected non-space characters in table context caused voodoo mode."))
+      # Make all the special element rearranging voodoo kick in
+      @tree.insert_from_table = true
+      # Process the character in the "in body" mode
+      @parser.phases[:inBody].processCharacters(data)
+      @tree.insert_from_table = false
+    end
+
+    def startTagCaption(name, attributes)
+      clearStackToTableContext
+      @tree.activeFormattingElements.push(Marker)
+      @tree.insert_element(name, attributes)
+      @parser.phase = @parser.phases[:inCaption]
+    end
+
+    def startTagColgroup(name, attributes)
+      clearStackToTableContext
+      @tree.insert_element(name, attributes)
+      @parser.phase = @parser.phases[:inColumnGroup]
+    end
+
+    def startTagCol(name, attributes)
+      startTagColgroup('colgroup', {})
+      @parser.phase.processStartTag(name, attributes)
+    end
+
+    def startTagRowGroup(name, attributes)
+      clearStackToTableContext
+      @tree.insert_element(name, attributes)
+      @parser.phase = @parser.phases[:inTableBody]
+    end
+
+    def startTagImplyTbody(name, attributes)
+      startTagRowGroup('tbody', {})
+      @parser.phase.processStartTag(name, attributes)
+    end
+
+    def startTagTable(name, attributes)
+      parse_error(_("Unexpected start tag (table) in table phase. Implies end tag (table)."))
+      @parser.phase.processEndTag('table')
+      @parser.phase.processStartTag(name, attributes) unless @parser.inner_html
+    end
+
+    def startTagOther(name, attributes)
+      parse_error(_("Unexpected start tag (#{name}) in table context caused voodoo mode."))
+      # Make all the special element rearranging voodoo kick in
+      @tree.insert_from_table = true
+      # Process the start tag in the "in body" mode
+      @parser.phases[:inBody].processStartTag(name, attributes)
+      @tree.insert_from_table = false
+    end
+
+    def endTagTable(name)
+      if in_scope?('table', true)
+        @tree.generateImpliedEndTags
+
+        unless @tree.open_elements.last.name == 'table'
+          parse_error(_("Unexpected end tag (table). Expected end tag (#{@tree.open_elements.last.name})."))
+        end
+
+        remove_open_elements_until('table')
+
+        @parser.reset_insertion_mode
+      else
+        # inner_html case
+        assert @parser.inner_html
+        parse_error
+      end
+    end
+
+    def endTagIgnore(name)
+      parse_error(_("Unexpected end tag (#{name}). Ignored."))
+    end
+
+    def endTagOther(name)
+      parse_error(_("Unexpected end tag (#{name}) in table context caused voodoo mode."))
+      # Make all the special element rearranging voodoo kick in
+      @tree.insert_from_table = true
+      # Process the end tag in the "in body" mode
+      @parser.phases[:inBody].processEndTag(name)
+      @tree.insert_from_table = false
+    end
+
+    protected
+
+    def clearStackToTableContext
+      # "clear the stack back to a table context"
+      until %w[table html].include?(name = @tree.open_elements.last.name)
+        parse_error(_("Unexpected implied end tag (#{name}) in the table phase."))
+        @tree.open_elements.pop
+      end
+      # When the current node is <html> it's an inner_html case
+    end
+
+  end
+end