From 5208bbf0af2a48091b795a4e3bf8cd6b2007b7ca Mon Sep 17 00:00:00 2001 From: Jacques Distler Date: Sat, 27 Oct 2007 17:34:29 -0500 Subject: [PATCH] Sanitize url refs in SVG attributes Add some tests. Sync with latest HTML5lib (includes above sanitization improvements). --- test/fixtures/revisions.yml | 4 +- test/functional/wiki_controller_test.rb | 2 +- test/unit/page_renderer_test.rb | 28 +- test/unit/page_test.rb | 2 +- vendor/plugins/HTML5lib/History.txt | 5 + vendor/plugins/HTML5lib/bin/html5 | 5 +- .../plugins/HTML5lib/lib/core_ext/kernel.rb | 13 + .../plugins/HTML5lib/lib/html5/constants.rb | 4 +- .../lib/html5/html5parser/in_body_phase.rb | 42 +- .../lib/html5/html5parser/in_caption_phase.rb | 13 +- .../lib/html5/html5parser/in_cell_phase.rb | 2 +- .../html5parser/in_column_group_phase.rb | 2 +- .../html5/html5parser/in_frameset_phase.rb | 3 +- .../lib/html5/html5parser/in_row_phase.rb | 7 +- .../html5/html5parser/in_table_body_phase.rb | 10 +- .../lib/html5/html5parser/in_table_phase.rb | 2 +- .../plugins/HTML5lib/lib/html5/inputstream.rb | 28 +- .../plugins/HTML5lib/lib/html5/sanitizer.rb | 6 + .../lib/html5/serializer/htmlserializer.rb | 2 +- vendor/plugins/HTML5lib/lib/html5/version.rb | 2 +- .../HTML5lib/testdata/encoding/tests2.dat | 33 + .../HTML5lib/testdata/sanitizer/tests1.dat | 32 + .../testdata/tree-construction/tests1.dat | 1030 ++++++++--------- .../testdata/tree-construction/tests2.dat | 184 +-- .../testdata/tree-construction/tests3.dat | 40 +- .../testdata/tree-construction/tests4.dat | 4 +- .../testdata/tree-construction/tests5.dat | 34 +- .../testdata/tree-construction/tests6.dat | 473 +++++++- 28 files changed, 1277 insertions(+), 735 deletions(-) create mode 100644 vendor/plugins/HTML5lib/lib/core_ext/kernel.rb diff --git a/test/fixtures/revisions.yml b/test/fixtures/revisions.yml index 48bf2cfd..6fab1755 100644 --- a/test/fixtures/revisions.yml +++ b/test/fixtures/revisions.yml @@ -50,7 +50,7 @@ home_page_second_revision: updated_at: <%= Time.local(2004, 4, 4, 16, 50).to_formatted_s(:db) %> revised_at: <%= Time.local(2004, 4, 4, 16, 50).to_formatted_s(:db) %> page_id: 1 - content: HisWay would be MyWay $\sin(x)\begin{svg}\end{svg}$ in kinda ThatWay in HisWay though MyWay \OverThere -- see SmartEngine in that SmartEngineGUI + content: HisWay would be MyWay $\sin(x)\begin{svg}\end{svg}\includegraphics[width=3em]{foo}$ in kinda ThatWay in HisWay though MyWay \OverThere -- see SmartEngine in that SmartEngineGUI author: DavidHeinemeierHansson first_page_first_revision: @@ -59,7 +59,7 @@ first_page_first_revision: updated_at: <%= Time.local(2004, 4, 4, 16, 55).to_formatted_s(:db) %> revised_at: <%= Time.local(2004, 4, 4, 16, 55).to_formatted_s(:db) %> page_id: 6 - content: HisWay would be MyWay $\sin(x)\begin{svg}\end{svg}$ in kinda ThatWay in HisWay though MyWay \\OverThere -- see SmartEngine in that SmartEngineGUI + content: HisWay would be MyWay $\sin(x)\begin{svg}\end{svg}\includegraphics[width=3em]{foo}$ in kinda ThatWay in HisWay though MyWay \\OverThere -- see SmartEngine in that SmartEngineGUI author: DavidHeinemeierHansson oak_first_revision: diff --git a/test/functional/wiki_controller_test.rb b/test/functional/wiki_controller_test.rb index c434c942..27bee12b 100755 --- a/test/functional/wiki_controller_test.rb +++ b/test/functional/wiki_controller_test.rb @@ -867,7 +867,7 @@ class WikiControllerTest < Test::Unit::TestCase \section*{HomePage} -HisWay would be MyWay $\sin(x) $ in kinda ThatWay in HisWay though MyWay $\backslash$OverThere --{} see SmartEngine in that SmartEngineGUI +HisWay would be MyWay $\sin(x) \includegraphics[width=3em]{foo}$ in kinda ThatWay in HisWay though MyWay $\backslash$OverThere --{} see SmartEngine in that SmartEngineGUI diff --git a/test/unit/page_renderer_test.rb b/test/unit/page_renderer_test.rb index c9840b8f..f5fc6a03 100644 --- a/test/unit/page_renderer_test.rb +++ b/test/unit/page_renderer_test.rb @@ -57,7 +57,13 @@ class PageRendererTest < Test::Unit::TestCase end def test_markdown - set_web_property :markup, :markdown + set_web_property :markup, :markdownMML + + assert_markup_parsed_as( + %{

equation } + + %{sin(x)

}, + "equation $\\sin(x)$") assert_markup_parsed_as( %{

My Headline

\n\n

that } + @@ -70,13 +76,11 @@ class PageRendererTest < Test::Unit::TestCase "#My Headline#\n\nthat SmartEngineGUI") assert_markup_parsed_as( - %{ecuasi\303\263n } + - %{sin(x)}, - "ecuasi\303\263n $\\sin(x)$") - - assert_markup_parsed_as( - %{ecuasi\303\263n } + - %{sin(x)}, + %{

sin} + + %{(x)} + + %{
} + + %{\\sin(x) \\begin{svg}\\end{svg}
}, "$$\\sin(x) \\begin{svg}\\end{svg}$$") code_block = [ @@ -92,6 +96,14 @@ class PageRendererTest < Test::Unit::TestCase %{

This is a code block:

\n\n
def a_method(arg)\n} +
         %{return ThatWay
\n\n

Nice!

}, code_block) + +#FIXME + assert_markup_parsed_as( + %{

ecuasi\303\263n } + + %{sin(x)

}, + "ecuasi\303\263n $\\sin(x)$") + end def test_markdown_hyperlink_with_slash diff --git a/test/unit/page_test.rb b/test/unit/page_test.rb index 5513cd48..66fe1deb 100644 --- a/test/unit/page_test.rb +++ b/test/unit/page_test.rb @@ -117,6 +117,6 @@ class PageTest < Test::Unit::TestCase assert_equal 3, @page.revisions(true).length, "Should have three revisions" @page.current_revision(true) @page.rollback(0, Time.now, '127.0.0.1', test_renderer) - assert_equal "HisWay would be MyWay in kinda ThatWay in HisWay though MyWay \\\\OverThere -- see SmartEngine in that SmartEngineGUI", @page.current_revision(true).content + assert_equal "HisWay would be MyWay $\\sin(x)\\begin{svg}\\end{svg}\\includegraphics[width=3em]{foo}$ in kinda ThatWay in HisWay though MyWay \\\\OverThere -- see SmartEngine in that SmartEngineGUI", @page.current_revision(true).content end end diff --git a/vendor/plugins/HTML5lib/History.txt b/vendor/plugins/HTML5lib/History.txt index d64c86c3..2c8f6321 100644 --- a/vendor/plugins/HTML5lib/History.txt +++ b/vendor/plugins/HTML5lib/History.txt @@ -1,3 +1,8 @@ +== 0.10.0 2007-10-08 +* proof-of-concept validator +* easier to localize error reporting +* many unit tests + == 0.1.0 / 2007-08-07 * 1 major enhancement diff --git a/vendor/plugins/HTML5lib/bin/html5 b/vendor/plugins/HTML5lib/bin/html5 index 870e5117..e16e9248 100755 --- a/vendor/plugins/HTML5lib/bin/html5 +++ b/vendor/plugins/HTML5lib/bin/html5 @@ -39,7 +39,7 @@ def parse(opts, args) if opts.parsemethod == :parse args = [f, encoding] else - args = [f, 'div', encoding] + args = [f, (opts.container || 'div'), encoding] end if opts.profile @@ -113,8 +113,9 @@ opts = OptionParser.new do |opts| options.treebuilder = treebuilder end - opts.on("-f", "--fragment", "Parse as a fragment") do |parse| + opts.on("-f", "--fragment CONTAINER", "Parse as a fragment") do |container| options.parsemethod = :parse_fragment + options.container = container if container end opts.separator "" diff --git a/vendor/plugins/HTML5lib/lib/core_ext/kernel.rb b/vendor/plugins/HTML5lib/lib/core_ext/kernel.rb new file mode 100644 index 00000000..3bb75ad8 --- /dev/null +++ b/vendor/plugins/HTML5lib/lib/core_ext/kernel.rb @@ -0,0 +1,13 @@ +module Kernel + def silence + if $VERBOSE + $VERBOSE = false + yield + $VERBOSE = true + else + yield + end + end +end + + \ No newline at end of file diff --git a/vendor/plugins/HTML5lib/lib/html5/constants.rb b/vendor/plugins/HTML5lib/lib/html5/constants.rb index 9a4580fa..ed34b086 100755 --- a/vendor/plugins/HTML5lib/lib/html5/constants.rb +++ b/vendor/plugins/HTML5lib/lib/html5/constants.rb @@ -999,7 +999,7 @@ module HTML5 _("Unexpected select start tag in the select phase " + "implies select start tag."), "unexpected-start-tag-in-select" => - _("Unexpected start tag token (%(name) in the select phase. " + + _("Unexpected start tag token (%(name)) in the select phase. " + "Ignored."), "unexpected-end-tag-in-select" => _("Unexpected end tag (%(name)) in the select phase. Ignored."), @@ -1039,6 +1039,8 @@ module HTML5 "expected-eof-but-got-end-tag" => _("Unexpected end tag (%(name))" + ". Expected end of file."), + "unexpected-end-table-in-caption" => + _("Unexpected end table tag in caption. Generates implied end caption.") } end diff --git a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_body_phase.rb b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_body_phase.rb index f0254a08..8b8ed02e 100644 --- a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_body_phase.rb +++ b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_body_phase.rb @@ -1,4 +1,5 @@ require 'html5/html5parser/phase' +require 'core_ext/kernel' module HTML5 class InBodyPhase < Phase @@ -50,13 +51,12 @@ module HTML5 super(parser, tree) # for special handling of whitespace in 
-      @processSpaceCharactersDropNewline = false
       if $-w
         $-w = false
-        alias processSpaceCharactersNonPre processSpaceCharacters
+        class << self; alias processSpaceCharactersNonPre processSpaceCharacters; end
         $-w = true
       else
-        alias processSpaceCharactersNonPre processSpaceCharacters
+        class << self; alias processSpaceCharactersNonPre processSpaceCharacters; end
       end
     end
 
@@ -65,10 +65,18 @@ module HTML5
 
       if $-w
         $-w = false
-        alias processSpaceCharacters processSpaceCharactersNonPre
+        class << self
+          silence do
+            alias processSpaceCharacters processSpaceCharactersNonPre
+          end
+        end
         $-w = true
       else
-        alias processSpaceCharacters processSpaceCharactersNonPre
+        class << self
+          silence do
+            alias processSpaceCharacters processSpaceCharactersNonPre
+          end
+        end
       end
 
       if (data.length > 0 and data[0] == ?\n && 
@@ -121,7 +129,13 @@ module HTML5
     def startTagCloseP(name, attributes)
       endTagP('p') if in_scope?('p')
       @tree.insert_element(name, attributes)
-      @processSpaceCharactersDropNewline = true if name == 'pre'
+      if name == 'pre'
+        class << self
+          silence do
+            alias processSpaceCharacters processSpaceCharactersDropNewline
+          end
+        end
+      end
     end
 
     def startTagForm(name, attributes)
@@ -291,8 +305,7 @@ module HTML5
       # XXX Form element pointer checking here as well...
       @tree.insert_element(name, attributes)
       @parser.tokenizer.content_model_flag = :RCDATA
-      @processSpaceCharactersDropNewline = true
-      alias processSpaceCharacters processSpaceCharactersDropNewline
+      class << self; alias processSpaceCharacters processSpaceCharactersDropNewline; end
     end
 
     # iframe, noembed noframes, noscript(if scripting enabled)
@@ -344,9 +357,9 @@ module HTML5
       # XXX Need to take open 
 tags into account here. We shouldn't imply
       # 
 but we should not throw a parse error either. Specification is
       # likely to be updated.
-      unless @tree.open_elements[1].name == 'body'
+      unless @tree.open_elements[1] && @tree.open_elements[1].name == 'body'
         # inner_html case
-        parse_error
+        parse_error "unexpected-end-tag", {:name => 'body'}
         return
       end
       unless @tree.open_elements.last.name == 'body'
@@ -364,7 +377,14 @@ module HTML5
 
     def endTagBlock(name)
       #Put us back in the right whitespace handling mode
-      @processSpaceCharactersDropNewline = false if name == 'pre'
+      if name == 'pre'
+        class << self; 
+          silence do
+            alias processSpaceCharacters processSpaceCharactersNonPre;
+          end
+        end
+      end
+      
 
       @tree.generateImpliedEndTags if in_scope?(name)
 
diff --git a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_caption_phase.rb b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_caption_phase.rb
index 541b2d14..ae56603d 100644
--- a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_caption_phase.rb
+++ b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_caption_phase.rb
@@ -5,9 +5,9 @@ module HTML5
 
     # 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_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'
+    handle_end 'caption', 'table', %w(body col colgroup html tbody td tfoot th thead tr) => 'Ignore'
 
     def ignoreEndTagCaption
       !in_scope?('caption', true)
@@ -18,7 +18,7 @@ module HTML5
     end
 
     def startTagTableElement(name, attributes)
-      parse_error
+      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')
@@ -33,7 +33,7 @@ module HTML5
       if ignoreEndTagCaption
         # inner_html case
         assert @parser.inner_html
-        parse_error
+        parse_error "unexpected-end-tag", {"name" => name}
       else
         # AT this code is quite similar to endTagTable in "InTable"
         @tree.generateImpliedEndTags
@@ -41,7 +41,7 @@ module HTML5
         unless @tree.open_elements[-1].name == 'caption'
           parse_error("expected-one-end-tag-but-got-another",
                     {"gotName" => "caption",
-                     "expectedNmae" => @tree.open_elements.last.name})
+                     "expectedName" => @tree.open_elements.last.name})
         end
 
         remove_open_elements_until('caption')
@@ -52,7 +52,7 @@ module HTML5
     end
 
     def endTagTable(name)
-      parse_error
+      parse_error "unexpected-end-table-in-caption"
       ignoreEndTag = ignoreEndTagCaption
       @parser.phase.processEndTag('caption')
       @parser.phase.processEndTag(name) unless ignoreEndTag
@@ -65,6 +65,5 @@ module HTML5
     def endTagOther(name)
       @parser.phases[:inBody].processEndTag(name)
     end
-
   end
 end
\ No newline at end of file
diff --git a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_cell_phase.rb b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_cell_phase.rb
index 30081422..29453735 100644
--- a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_cell_phase.rb
+++ b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_cell_phase.rb
@@ -56,7 +56,7 @@ module HTML5
         @parser.phase.processEndTag(name)
       else
         # sometimes inner_html case
-        parse_error
+        parse_error "unexpected-end-tag", {:name => name}
       end
     end
 
diff --git a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_column_group_phase.rb b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_column_group_phase.rb
index 9e1dec2a..0689af9f 100644
--- a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_column_group_phase.rb
+++ b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_column_group_phase.rb
@@ -34,7 +34,7 @@ module HTML5
       if ignoreEndTagColgroup
         # inner_html case
         assert @parser.inner_html
-        parse_error
+        parse_error "unexpected-end-tag", {:name => name}
       else
         @tree.open_elements.pop
         @parser.phase = @parser.phases[:inTable]
diff --git a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_frameset_phase.rb b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_frameset_phase.rb
index 5ff9bd6b..19c42a53 100644
--- a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_frameset_phase.rb
+++ b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_frameset_phase.rb
@@ -27,8 +27,7 @@ module HTML5
     end
 
     def startTagOther(name, attributes)
-      parse_error("unexpected-start-tag-in-frameset",
-            {"name" => name})
+      parse_error("unexpected-start-tag-in-frameset", {"name" => name})
     end
 
     def endTagFrameset(name)
diff --git a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_row_phase.rb b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_row_phase.rb
index 12af6212..1b8e6c09 100644
--- a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_row_phase.rb
+++ b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_row_phase.rb
@@ -35,7 +35,7 @@ module HTML5
       if ignoreEndTagTr
         # inner_html case
         assert @parser.inner_html
-        parse_error
+        parse_error "unexpected-end-tag", {:name => name}
       else
         clearStackToTableRowContext
         @tree.open_elements.pop
@@ -57,7 +57,7 @@ module HTML5
         @parser.phase.processEndTag(name)
       else
         # inner_html case
-        parse_error
+        parse_error "unexpected-end-tag", {:name => name}
       end
     end
 
@@ -75,8 +75,7 @@ module HTML5
     # 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-in-table-row",
-                {"name" => @tree.open_elements.last.name})
+        parse_error("unexpected-implied-end-tag-in-table-row", {"name" => @tree.open_elements.last.name})
         @tree.open_elements.pop
       end
     end
diff --git a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_table_body_phase.rb b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_table_body_phase.rb
index 0685d34e..8e77bf98 100644
--- a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_table_body_phase.rb
+++ b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_table_body_phase.rb
@@ -33,7 +33,7 @@ module HTML5
         @parser.phase.processStartTag(name, attributes)
       else
         # inner_html case
-        parse_error
+        parse_error "unexpected-start-tag", {:name => name}
       end
     end
 
@@ -47,8 +47,7 @@ module HTML5
         @tree.open_elements.pop
         @parser.phase = @parser.phases[:inTable]
       else
-        parse_error("unexpected-end-tag-in-table-body",
-                {"name" => name})
+        parse_error("unexpected-end-tag-in-table-body", {"name" => name})
       end
     end
 
@@ -59,13 +58,12 @@ module HTML5
         @parser.phase.processEndTag(name)
       else
         # inner_html case
-        parse_error
+        parse_error "unexpected-end-tag", {:name => name}
       end
     end
 
     def endTagIgnore(name)
-      parse_error("unexpected-end-tag-in-table-body",
-            {"name" => name})
+      parse_error("unexpected-end-tag-in-table-body", {"name" => name})
     end
 
     def endTagOther(name)
diff --git a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_table_phase.rb b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_table_phase.rb
index 6ffa8812..c87011ad 100644
--- a/vendor/plugins/HTML5lib/lib/html5/html5parser/in_table_phase.rb
+++ b/vendor/plugins/HTML5lib/lib/html5/html5parser/in_table_phase.rb
@@ -82,7 +82,7 @@ module HTML5
       else
         # inner_html case
         assert @parser.inner_html
-        parse_error
+        parse_error "unexpected-end-tag", {:name => name}
       end
     end
 
diff --git a/vendor/plugins/HTML5lib/lib/html5/inputstream.rb b/vendor/plugins/HTML5lib/lib/html5/inputstream.rb
index f763bf69..a07aa815 100755
--- a/vendor/plugins/HTML5lib/lib/html5/inputstream.rb
+++ b/vendor/plugins/HTML5lib/lib/html5/inputstream.rb
@@ -47,7 +47,7 @@ module HTML5
 
       #Encoding to use if no other information can be found
       @DEFAULT_ENCODING = 'windows-1252'
-    
+
       #Detect encoding iff no explicit "transport level" encoding is supplied
       if @encoding.nil? or not HTML5.is_valid_encoding(@encoding)
         @char_encoding = detect_encoding
@@ -235,14 +235,16 @@ module HTML5
     # Returns (line, col) of the current position in the stream.
     def position
       line, col = @line, @col
-      @queue.reverse.each do |c|
-        if c == "\n"
-          line -= 1
-          raise RuntimeError.new("col=#{col}") unless col == 0
-          col = @line_lengths[line]
-        else
-          col -= 1
-        end 
+      if @queue and @queue.last != :EOF
+        @queue.reverse.each do |c|
+          if c == "\n"
+            line -= 1
+            raise RuntimeError.new("col=#{col}") unless col == 0
+            col = @line_lengths[line]
+          else
+            col -= 1
+          end 
+        end
       end
       return [line + 1, col]
     end
@@ -434,7 +436,13 @@ module HTML5
         end
         break unless keep_parsing
       end
-      @encoding = @encoding.strip unless @encoding.nil?
+      unless @encoding.nil?
+        @encoding = @encoding.strip 
+        if ["UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE"].include?(@encoding.upcase)
+          @encoding = 'utf-8'
+        end
+      end
+      
       return @encoding
     end
 
diff --git a/vendor/plugins/HTML5lib/lib/html5/sanitizer.rb b/vendor/plugins/HTML5lib/lib/html5/sanitizer.rb
index 23c9d38c..26977ba4 100644
--- a/vendor/plugins/HTML5lib/lib/html5/sanitizer.rb
+++ b/vendor/plugins/HTML5lib/lib/html5/sanitizer.rb
@@ -78,6 +78,9 @@ module HTML5
 
     ATTR_VAL_IS_URI = %w[href src cite action longdesc xlink:href xml:base]
 
+    SVG_ATTR_VAL_ALLOWS_REF = %w[clip-path fill filter marker marker-start
+      marker-mid marker-end mask stroke textpath]
+
     ACCEPTABLE_CSS_PROPERTIES = %w[azimuth background-color
       border-bottom-color border-collapse border-color border-left-color
       border-right-color border-top-color clear color cursor direction
@@ -120,6 +123,9 @@ module HTML5
                 if val_unescaped =~ /^[a-z0-9][-+.a-z0-9]*:/ and !self.class.const_get("ALLOWED_PROTOCOLS").include?(val_unescaped.split(':')[0])
                   attrs.delete attr
                 end
+                SVG_ATTR_VAL_ALLOWS_REF.each do |attr|
+                  attrs.delete attr if attrs[attr].to_s.downcase =~ /url\(\s*[^#]/m
+                end
               end
               if attrs['style']
                 attrs['style'] = sanitize_css(attrs['style'])
diff --git a/vendor/plugins/HTML5lib/lib/html5/serializer/htmlserializer.rb b/vendor/plugins/HTML5lib/lib/html5/serializer/htmlserializer.rb
index 467c5e95..ae799f83 100644
--- a/vendor/plugins/HTML5lib/lib/html5/serializer/htmlserializer.rb
+++ b/vendor/plugins/HTML5lib/lib/html5/serializer/htmlserializer.rb
@@ -144,7 +144,7 @@ module HTML5
 
         elsif type == :Comment
           data = token[:data]
-          serialize_error(_("Comment contains --")) if data.index("--")
+          serialize_error("Comment contains --") if data.index("--")
           comment = "" % token[:data]
           result << comment
 
diff --git a/vendor/plugins/HTML5lib/lib/html5/version.rb b/vendor/plugins/HTML5lib/lib/html5/version.rb
index 47c0b716..655548e5 100644
--- a/vendor/plugins/HTML5lib/lib/html5/version.rb
+++ b/vendor/plugins/HTML5lib/lib/html5/version.rb
@@ -1,3 +1,3 @@
 module HTML5
-  VERSION = '0.1.0'
+  VERSION = '0.10.0'
 end
\ No newline at end of file
diff --git a/vendor/plugins/HTML5lib/testdata/encoding/tests2.dat b/vendor/plugins/HTML5lib/testdata/encoding/tests2.dat
index dc74859c..a905a48c 100644
--- a/vendor/plugins/HTML5lib/testdata/encoding/tests2.dat
+++ b/vendor/plugins/HTML5lib/testdata/encoding/tests2.dat
@@ -79,3 +79,36 @@ windows-1252
 
+#encoding
+utf-8
+
+#data
+
+#encoding
+utf-8
+
+#data
+
+#encoding
+utf-8
+
+#data
+
+
+#encoding
+utf-8
+
+#data
+
+
+#encoding
+utf-8
+
+#data
+ñ
+
+#encoding
+utf-8
diff --git a/vendor/plugins/HTML5lib/testdata/sanitizer/tests1.dat b/vendor/plugins/HTML5lib/testdata/sanitizer/tests1.dat
index 89a7d451..6ef50d41 100644
--- a/vendor/plugins/HTML5lib/testdata/sanitizer/tests1.dat
+++ b/vendor/plugins/HTML5lib/testdata/sanitizer/tests1.dat
@@ -412,5 +412,37 @@
     "input": "",
     "rexml": "",
     "output": ""
+  },
+
+  {
+    "name": "uri_refs_in_svg_attributes",
+    "input": "",
+    "rexml": "",
+    "xhtml": "",
+    "output": ""
+  },
+
+  {
+    "name": "absolute_uri_refs_in_svg_attributes",
+    "input": "",
+    "rexml": "",
+    "xhtml": "",
+    "output": ""
+  },
+
+  {
+    "name": "uri_ref_with_space_in svg_attribute",
+    "input": "",
+    "rexml": "",
+    "xhtml": "",
+    "output": ""
+  },
+
+  {
+    "name": "absolute_uri_ref_with_space_in svg_attribute",
+    "input": "",
+    "rexml": "",
+    "xhtml": "",
+    "output": ""
   }
 ]
diff --git a/vendor/plugins/HTML5lib/testdata/tree-construction/tests1.dat b/vendor/plugins/HTML5lib/testdata/tree-construction/tests1.dat
index 5c7f894c..628bff68 100644
--- a/vendor/plugins/HTML5lib/testdata/tree-construction/tests1.dat
+++ b/vendor/plugins/HTML5lib/testdata/tree-construction/tests1.dat
@@ -1,7 +1,7 @@
 #data
 Test
 #errors
-5: missing document type declaration
+Line: 1 Col: 4 Unexpected non-space characters. Expected DOCTYPE.
 #document
 | 
 |   
@@ -11,7 +11,7 @@ Test
 #data
 
One
Two
 #errors
-3: missing document type declaration
+Line: 1 Col: 3 Unexpected start tag (p). Expected DOCTYPE.
 #document
 | 
 |   
@@ -24,7 +24,7 @@ Test
 #data
 Line1
Line2
Line3
Line4
 #errors
-7: missing document type declaration
+Line: 1 Col: 5 Unexpected non-space characters. Expected DOCTYPE.
 #document
 | 
 |   
@@ -40,7 +40,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (html). Expected DOCTYPE.
 #document
 | 
 |   
@@ -49,7 +49,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (head). Expected DOCTYPE.
 #document
 | 
 |   
@@ -58,7 +58,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (body). Expected DOCTYPE.
 #document
 | 
 |   
@@ -67,7 +67,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (html). Expected DOCTYPE.
 #document
 | 
 |   
@@ -76,7 +76,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (html). Expected DOCTYPE.
 #document
 | 
 |   
@@ -85,7 +85,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (html). Expected DOCTYPE.
 #document
 | 
 |   
@@ -94,7 +94,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (html). Expected DOCTYPE.
 #document
 | 
 |   
@@ -103,7 +103,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (html). Expected DOCTYPE.
 #document
 | 
 |   
@@ -112,7 +112,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (html). Expected DOCTYPE.
 #document
 | 
 |   
@@ -121,7 +121,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (html). Expected DOCTYPE.
 #document
 | 
 |   
@@ -130,7 +130,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (html). Expected DOCTYPE.
 #document
 | 
 |   
@@ -139,7 +139,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (body). Expected DOCTYPE.
 #document
 | 
 |   
@@ -148,7 +148,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-6: missing document type declaration
+Line: 1 Col: 6 Unexpected start tag (head). Expected DOCTYPE.
 #document
 | 
 |   
@@ -157,7 +157,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-7: missing document type declaration
+Line: 1 Col: 7 Unexpected end tag (head). Expected DOCTYPE.
 #document
 | 
 |   
@@ -166,7 +166,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-7: missing document type declaration
+Line: 1 Col: 7 Unexpected end tag (body). Expected DOCTYPE.
 #document
 | 
 |   
@@ -175,7 +175,7 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-7: missing document type declaration
+Line: 1 Col: 7 Unexpected end tag (html). Expected DOCTYPE.
 #document
 | 
 |   
@@ -184,10 +184,10 @@ Line1
Line2
Line3
Line4
 #data
 

 #errors
-3: missing document type declaration
-14: required tr element start tag implied by unexpected td element start tag
-25: unexpected td element end tag implied other end tags
-26: mismatched body element end tag (premature end of file?)
+Line: 1 Col: 3 Unexpected start tag (b). Expected DOCTYPE.
+Line: 1 Col: 14 Unexpected table cell start tag (td) in the table body phase.
+Line: 1 Col: 25 Got table cell end tag (td) while required end tags are missing.
+Line: 1 Col: 25 Expected closing tag. Unexpected end of file.
 #document
 | 
 |   
@@ -202,11 +202,11 @@ Line1
Line2
Line3
Line4
 #data
 
X
 #errors
-3: missing document type declaration
-14: required tr element start tag implied by unexpected td element start tag
-18: mismatched b element end tag (no matching start tag)
-29: unexpected td element end tag implied other end tags
-31: mismatched body element end tag (premature end of file?)
+Line: 1 Col: 3 Unexpected start tag (b). Expected DOCTYPE.
+Line: 1 Col: 14 Unexpected table cell start tag (td) in the table body phase.
+Line: 1 Col: 18 End tag (b) violates step 1, paragraph 1 of the adoption agency algorithm.
+Line: 1 Col: 29 Got table cell end tag (td) while required end tags are missing.
+Line: 1 Col: 30 Expected closing tag. Unexpected end of file.
 #document
 | 
 |   
@@ -222,8 +222,8 @@ Line1
Line2
Line3
Line4
 #data
 
Hello
World
 #errors
-4: missing document type declaration
-19: mismatched body element end tag (premature end of file?)
+Line: 1 Col: 4 Unexpected start tag (h1). Expected DOCTYPE.
+Line: 1 Col: 18 Expected closing tag. Unexpected end of file.
 #document
 | 
 |   
@@ -236,10 +236,10 @@ Line1
Line2
Line3
Line4
 #data
 
XYZ

 #errors
-3: missing document type declaration
-10: a element start tag implying a element end tag
-10: mismatched a element end tag (misnested tags)
-24: mismatched a element end tag (no matching start tag)
+Line: 1 Col: 3 Unexpected start tag (a). Expected DOCTYPE.
+Line: 1 Col: 10 Unexpected start tag (a) implies end tag (a).
+Line: 1 Col: 10 End tag (a) violates step 1, paragraph 3 of the adoption agency algorithm.
+Line: 1 Col: 24 End tag (a) violates step 1, paragraph 1 of the adoption agency algorithm.
 #document
 | 
 |   
@@ -255,8 +255,8 @@ Line1
Line2
Line3
Line4
 #data
 
 #errors
-3: missing document type declaration
-15: mismatched b element end tag (no matching start tag)
+Line: 1 Col: 3 Unexpected start tag (b). Expected DOCTYPE.
+Line: 1 Col: 15 End tag (b) violates step 1, paragraph 1 of the adoption agency algorithm.
 #document
 | 
 |   
@@ -267,12 +267,12 @@ Line1
Line2
Line3
Line4
 #data
 
X
 #errors
-3: missing document type declaration
-11: mismatched p element end tag
-24: mismatched p element end tag
-28: mismatched b element end tag (no matching start tag)
-34: mismatched div element end tag
-36: mismatched body element end tag (premature end of file?)
+Line: 1 Col: 3 Unexpected start tag (p). Expected DOCTYPE.
+Line: 1 Col: 11 Unexpected end tag (p). Ignored.
+Line: 1 Col: 24 Unexpected end tag (p). Ignored.
+Line: 1 Col: 28 End tag (b) violates step 1, paragraph 1 of the adoption agency algorithm.
+Line: 1 Col: 34 End tag (div) seen too early. Expected other end tag.
+Line: 1 Col: 35 Expected closing tag. Unexpected end of file.
 #document
 | 
 |   
@@ -288,8 +288,8 @@ Line1
Line2
Line3
Line4
 #data
 <p>

 #errors
-8: missing document type declaration
-28: unexpected div element end tag in head
+Line: 1 Col: 8 Unexpected start tag (script). Expected DOCTYPE.
+Line: 1 Col: 28 Unexpected end tag (div). Ignored.
 #document
 | 
 |   
@@ -304,10 +304,10 @@ Line1
Line2
Line3
Line4
 #data
 
--
 #errors
-Incorrect comment token
-Missing document type declaration
-Incorrect comment token
-Unexpected end of file
+Line: 1 Col: 5 Incorrect comment.
+Line: 1 Col: 10 Unexpected start tag (div). Expected DOCTYPE.
+Line: 1 Col: 17 Incorrect comment.
+Line: 1 Col: 17 Expected closing tag. Unexpected end of file.
 #document
 | 
 | 
@@ -320,8 +320,8 @@ Unexpected end of file
 #data
 

 #errors
-3: missing document type declaration
-11: mismatched p element end tag
+Line: 1 Col: 3 Unexpected start tag (p). Expected DOCTYPE.
+Line: 1 Col: 11 Unexpected end tag (p). Ignored.
 #document
 | 
 |   
@@ -333,12 +333,12 @@ Unexpected end of file
 #data