loofah 2.3.1 → 2.19.1

data/lib/loofah/html5/scrub.rb

@@ -1,30 +1,32 @@
-require 'cgi'
-require 'crass'
+# frozen_string_literal: true
+require "cgi"
+require "crass"
 
 module Loofah
   module HTML5 # :nodoc:
     module Scrub
-
       CONTROL_CHARACTERS = /[`\u0000-\u0020\u007f\u0080-\u0101]/
-      CSS_KEYWORDISH = /\A(#[0-9a-fA-F]+|rgb\(\d+%?,\d*%?,?\d*%?\)?|-?\d{0,3}\.?\d{0,10}(cm|em|ex|in|mm|pc|pt|px|%|,|\))?)\z/
-      CRASS_SEMICOLON = {:node => :semicolon, :raw => ";"}
+      CSS_KEYWORDISH = /\A(#[0-9a-fA-F]+|rgb\(\d+%?,\d*%?,?\d*%?\)?|-?\d{0,3}\.?\d{0,10}(ch|cm|r?em|ex|in|lh|mm|pc|pt|px|Q|vmax|vmin|vw|vh|%|,|\))?)\z/
+      CRASS_SEMICOLON = { node: :semicolon, raw: ";" }
+      CSS_IMPORTANT = '!important'
+      CSS_PROPERTY_STRING_WITHOUT_EMBEDDED_QUOTES = /\A(["'])?[^"']+\1\z/
+      DATA_ATTRIBUTE_NAME = /\Adata-[\w-]+\z/
 
       class << self
-
-        def allowed_element? element_name
-          ::Loofah::HTML5::SafeList::ALLOWED_ELEMENTS_WITH_LIBXML2.include? element_name
+        def allowed_element?(element_name)
+          ::Loofah::HTML5::SafeList::ALLOWED_ELEMENTS_WITH_LIBXML2.include?(element_name)
         end
 
         #  alternative implementation of the html5lib attribute scrubbing algorithm
-        def scrub_attributes node
+        def scrub_attributes(node)
           node.attribute_nodes.each do |attr_node|
             attr_name = if attr_node.namespace
-                          "#{attr_node.namespace.prefix}:#{attr_node.node_name}"
-                        else
-                          attr_node.node_name
-                        end
+              "#{attr_node.namespace.prefix}:#{attr_node.node_name}"
+            else
+              attr_node.node_name
+            end
 
-            if attr_name =~ /\Adata-[\w-]+\z/
+            if attr_name =~ DATA_ATTRIBUTE_NAME
               next
             end
 
@@ -34,71 +36,125 @@ module Loofah
             end
 
             if SafeList::ATTR_VAL_IS_URI.include?(attr_name)
-              # this block lifted nearly verbatim from HTML5 sanitization
-              val_unescaped = CGI.unescapeHTML(attr_node.value).gsub(CONTROL_CHARACTERS,'').downcase
-              if val_unescaped =~ /^[a-z0-9][-+.a-z0-9]*:/ && ! SafeList::ALLOWED_PROTOCOLS.include?(val_unescaped.split(SafeList::PROTOCOL_SEPARATOR)[0])
-                attr_node.remove
-                next
-              elsif val_unescaped.split(SafeList::PROTOCOL_SEPARATOR)[0] == 'data'
-                # permit only allowed data mediatypes
-                mediatype = val_unescaped.split(SafeList::PROTOCOL_SEPARATOR)[1]
-                mediatype, _ = mediatype.split(';')[0..1] if mediatype
-                if mediatype && !SafeList::ALLOWED_URI_DATA_MEDIATYPES.include?(mediatype)
-                  attr_node.remove
-                  next
-                end
-              end
+              next if scrub_uri_attribute(attr_node)
             end
+
             if SafeList::SVG_ATTR_VAL_ALLOWS_REF.include?(attr_name)
-              attr_node.value = attr_node.value.gsub(/url\s*\(\s*[^#\s][^)]+?\)/m, ' ') if attr_node.value
+              scrub_attribute_that_allows_local_ref(attr_node)
             end
-            if SafeList::SVG_ALLOW_LOCAL_HREF.include?(node.name) && attr_name == 'xlink:href' && attr_node.value =~ /^\s*[^#\s].*/m
+
+            if SafeList::SVG_ALLOW_LOCAL_HREF.include?(node.name) && attr_name == "xlink:href" && attr_node.value =~ /^\s*[^#\s].*/m
               attr_node.remove
               next
             end
           end
 
-          scrub_css_attribute node
+          scrub_css_attribute(node)
 
           node.attribute_nodes.each do |attr_node|
-            node.remove_attribute(attr_node.name) if attr_node.value !~ /[^[:space:]]/
+            if attr_node.value !~ /[^[:space:]]/ && attr_node.name !~ DATA_ATTRIBUTE_NAME
+              node.remove_attribute(attr_node.name)
+            end
           end
 
-          force_correct_attribute_escaping! node
+          force_correct_attribute_escaping!(node)
         end
 
-        def scrub_css_attribute node
-          style = node.attributes['style']
+        def scrub_css_attribute(node)
+          style = node.attributes["style"]
           style.value = scrub_css(style.value) if style
         end
 
-        def scrub_css style
-          style_tree = Crass.parse_properties style
+        def scrub_css(style)
+          style_tree = Crass.parse_properties(style)
           sanitized_tree = []
 
           style_tree.each do |node|
             next unless node[:node] == :property
             next if node[:children].any? do |child|
-              [:url, :bad_url].include?(child[:node]) || (child[:node] == :function && !SafeList::ALLOWED_CSS_FUNCTIONS.include?(child[:name].downcase))
+              [:url, :bad_url].include?(child[:node])
             end
+
             name = node[:name].downcase
-            if SafeList::ALLOWED_CSS_PROPERTIES.include?(name) || SafeList::ALLOWED_SVG_PROPERTIES.include?(name)
-              sanitized_tree << node << CRASS_SEMICOLON
-            elsif SafeList::SHORTHAND_CSS_PROPERTIES.include?(name.split('-').first)
-              value = node[:value].split.map do |keyword|
-                if SafeList::ALLOWED_CSS_KEYWORDS.include?(keyword) || keyword =~ CSS_KEYWORDISH
+            next unless SafeList::ALLOWED_CSS_PROPERTIES.include?(name) ||
+                SafeList::ALLOWED_SVG_PROPERTIES.include?(name) ||
+                SafeList::SHORTHAND_CSS_PROPERTIES.include?(name.split("-").first)
+
+            value = node[:children].map do |child|
+              case child[:node]
+              when :whitespace
+                nil
+              when :string
+                if child[:raw] =~ CSS_PROPERTY_STRING_WITHOUT_EMBEDDED_QUOTES
+                  Crass::Parser.stringify(child)
+                else
+                  nil
+                end
+              when :function
+                if SafeList::ALLOWED_CSS_FUNCTIONS.include?(child[:name].downcase)
+                  Crass::Parser.stringify(child)
+                end
+              when :ident
+                keyword = child[:value]
+                if !SafeList::SHORTHAND_CSS_PROPERTIES.include?(name.split("-").first) ||
+                   SafeList::ALLOWED_CSS_KEYWORDS.include?(keyword) ||
+                   (keyword =~ CSS_KEYWORDISH)
                   keyword
                 end
-              end.compact
-              unless value.empty?
-                propstring = sprintf "%s:%s", name, value.join(" ")
-                sanitized_node = Crass.parse_properties(propstring).first
-                sanitized_tree << sanitized_node << CRASS_SEMICOLON
+              else
+                child[:raw]
               end
-            end
+            end.compact
+
+            next if value.empty?
+            value << CSS_IMPORTANT if node[:important]
+            propstring = format("%s:%s", name, value.join(" "))
+            sanitized_node = Crass.parse_properties(propstring).first
+            sanitized_tree << sanitized_node << CRASS_SEMICOLON
           end
 
-          Crass::Parser.stringify sanitized_tree
+          Crass::Parser.stringify(sanitized_tree)
+        end
+
+        def scrub_attribute_that_allows_local_ref(attr_node)
+          return unless attr_node.value
+
+          nodes = Crass::Parser.new(attr_node.value).parse_component_values
+
+          values = nodes.map do |node|
+            case node[:node]
+            when :url
+              if node[:value].start_with?("#")
+                node[:raw]
+              else
+                nil
+              end
+            when :hash, :ident, :string
+              node[:raw]
+            else
+              nil
+            end
+          end.compact
+
+          attr_node.value = values.join(" ")
+        end
+
+        def scrub_uri_attribute(attr_node)
+          # this block lifted nearly verbatim from HTML5 sanitization
+          val_unescaped = CGI.unescapeHTML(attr_node.value).gsub(CONTROL_CHARACTERS, "").downcase
+          if val_unescaped =~ /^[a-z0-9][-+.a-z0-9]*:/ && !SafeList::ALLOWED_PROTOCOLS.include?(val_unescaped.split(SafeList::PROTOCOL_SEPARATOR)[0])
+            attr_node.remove
+            return true
+          elsif val_unescaped.split(SafeList::PROTOCOL_SEPARATOR)[0] == "data"
+            # permit only allowed data mediatypes
+            mediatype = val_unescaped.split(SafeList::PROTOCOL_SEPARATOR)[1]
+            mediatype, _ = mediatype.split(";")[0..1] if mediatype
+            if mediatype && !SafeList::ALLOWED_URI_DATA_MEDIATYPES.include?(mediatype)
+              attr_node.remove
+              return true
+            end
+          end
+          false
         end
 
         #
@@ -106,7 +162,7 @@ module Loofah
         #
         #  see comments about CVE-2018-8048 within the tests for more information
         #
-        def force_correct_attribute_escaping! node
+        def force_correct_attribute_escaping!(node)
           return unless Nokogiri::VersionInfo.instance.libxml2?
 
           node.attribute_nodes.each do |attr_node|
@@ -122,11 +178,50 @@ module Loofah
             #
             encoding = attr_node.value.encoding
             attr_node.value = attr_node.value.gsub(/[ "]/) do |m|
-              '%' + m.unpack('H2' * m.bytesize).join('%').upcase
+              "%" + m.unpack("H2" * m.bytesize).join("%").upcase
             end.force_encoding(encoding)
           end
         end
 
+        def cdata_needs_escaping?(node)
+          # Nokogiri's HTML4 parser on JRuby doesn't flag the child of a `style` or `script` tag as cdata, but it acts that way
+          node.cdata? || (Nokogiri.jruby? && node.text? && (node.parent.name == "style" || node.parent.name == "script"))
+        end
+
+        def cdata_escape(node)
+          escaped_text = escape_tags(node.text)
+          if Nokogiri.jruby?
+            node.document.create_text_node(escaped_text)
+          else
+            node.document.create_cdata(escaped_text)
+          end
+        end
+
+        TABLE_FOR_ESCAPE_HTML__ = {
+          '<' => '&lt;',
+          '>' => '&gt;',
+          '&' => '&amp;',
+        }
+
+        def escape_tags(string)
+          # modified version of CGI.escapeHTML from ruby 3.1
+          enc = string.encoding
+          unless enc.ascii_compatible?
+            if enc.dummy?
+              origenc = enc
+              enc = Encoding::Converter.asciicompat_encoding(enc)
+              string = enc ? string.encode(enc) : string.b
+            end
+            table = Hash[TABLE_FOR_ESCAPE_HTML__.map {|pair|pair.map {|s|s.encode(enc)}}]
+            string = string.gsub(/#{"[<>&]".encode(enc)}/, table)
+            string.encode!(origenc) if origenc
+            string
+          else
+            string = string.b
+            string.gsub!(/[<>&]/, TABLE_FOR_ESCAPE_HTML__)
+            string.force_encoding(enc)
+          end
+        end
       end
     end
   end

data/lib/loofah/instance_methods.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Loofah
   #
   #  Mixes +scrub!+ into Document, DocumentFragment, Node and NodeSet.
@@ -91,28 +92,33 @@ module Loofah
     #    # decidedly not ok for browser:
     #    frag.text(:encode_special_chars => false) # => "<script>alert('EVIL');</script>"
     #
-    def text(options={})
-      result = serialize_root.children.inner_text rescue ""
+    def text(options = {})
+      result = if serialize_root
+        serialize_root.children.reject(&:comment?).map(&:inner_text).join("")
+      else
+        ""
+      end
       if options[:encode_special_chars] == false
         result # possibly dangerous if rendered in a browser
       else
         encode_special_chars result
       end
     end
+
     alias :inner_text :text
-    alias :to_str     :text
+    alias :to_str :text
 
     #
     #  Returns a plain-text version of the markup contained by the
     #  fragment, with HTML entities encoded.
     #
-    #  This method is slower than #to_text, but is clever about
-    #  whitespace around block elements.
+    #  This method is slower than #text, but is clever about
+    #  whitespace around block elements and line break elements.
     #
-    #    Loofah.document("<h1>Title</h1><div>Content</div>").to_text
-    #    # => "\nTitle\n\nContent\n"
+    #    Loofah.document("<h1>Title</h1><div>Content<br>Next line</div>").to_text
+    #    # => "\nTitle\n\nContent\nNext line\n"
     #
-    def to_text(options={})
+    def to_text(options = {})
       Loofah.remove_extraneous_whitespace self.dup.scrub!(:newline_block_elements).text(options)
     end
   end

data/lib/loofah/metahelpers.rb

@@ -1,6 +1,7 @@
+# frozen_string_literal: true
 module Loofah
   module MetaHelpers # :nodoc:
-    def self.add_downcased_set_members_to_all_set_constants mojule
+    def self.add_downcased_set_members_to_all_set_constants(mojule)
       mojule.constants.each do |constant_sym|
         constant = mojule.const_get constant_sym
         next unless Set === constant

data/lib/loofah/scrubber.rb

@@ -1,8 +1,9 @@
+# frozen_string_literal: true
 module Loofah
   #
   #  A RuntimeError raised when Loofah could not find an appropriate scrubber.
   #
-  class ScrubberNotFound < RuntimeError ; end
+  class ScrubberNotFound < RuntimeError; end
 
   #
   #  A Scrubber wraps up a block (or method) that is run on an HTML node (element):
@@ -36,7 +37,7 @@ module Loofah
     CONTINUE = Object.new.freeze
 
     # Top-down Scrubbers may return STOP to indicate that the subtree should not be traversed.
-    STOP     = Object.new.freeze
+    STOP = Object.new.freeze
 
     # When a scrubber is initialized, the :direction may be specified
     # as :top_down (the default) or :bottom_up.
@@ -64,7 +65,7 @@ module Loofah
     def initialize(options = {}, &block)
       direction = options[:direction] || :top_down
       unless [:top_down, :bottom_up].include?(direction)
-        raise ArgumentError, "direction #{direction} must be one of :top_down or :bottom_up" 
+        raise ArgumentError, "direction #{direction} must be one of :top_down or :bottom_up"
       end
       @direction, @block = direction, block
     end
@@ -91,10 +92,10 @@ module Loofah
     # If the attribute is set, don't overwrite the existing value
     #
     def append_attribute(node, attribute, value)
-      current_value = node.get_attribute(attribute) || ''
+      current_value = node.get_attribute(attribute) || ""
       current_values = current_value.split(/\s+/)
       updated_value = current_values | [value]
-      node.set_attribute(attribute, updated_value.join(' '))
+      node.set_attribute(attribute, updated_value.join(" "))
     end
 
     private
@@ -107,6 +108,10 @@ module Loofah
           return Scrubber::CONTINUE
         end
       when Nokogiri::XML::Node::TEXT_NODE, Nokogiri::XML::Node::CDATA_SECTION_NODE
+        if HTML5::Scrub.cdata_needs_escaping?(node)
+          node.before(HTML5::Scrub.cdata_escape(node))
+          return Scrubber::STOP
+        end
         return Scrubber::CONTINUE
       end
       Scrubber::STOP
@@ -118,11 +123,11 @@ module Loofah
       else
         return if scrub(node) == STOP
       end
-      node.children.each {|j| traverse_conditionally_top_down(j)}
+      node.children.each { |j| traverse_conditionally_top_down(j) }
     end
 
     def traverse_conditionally_bottom_up(node)
-      node.children.each {|j| traverse_conditionally_bottom_up(j)}
+      node.children.each { |j| traverse_conditionally_bottom_up(j) }
       if block
         block.call(node)
       else

data/lib/loofah/scrubbers.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Loofah
   #
   #  Loofah provides some built-in scrubbers for sanitizing with
@@ -99,13 +100,9 @@ module Loofah
 
       def scrub(node)
         return CONTINUE if html5lib_sanitize(node) == CONTINUE
-        if node.children.length == 1 && node.children.first.cdata?
-          sanitized_text = Loofah.fragment(node.children.first.to_html).scrub!(:strip).to_html
-          node.before Nokogiri::XML::Text.new(sanitized_text, node.document)
-        else
-          node.before node.children
-        end
+        node.before(node.children)
         node.remove
+        return STOP
       end
     end
 
@@ -205,8 +202,8 @@ module Loofah
       end
 
       def scrub(node)
-        return CONTINUE unless (node.type == Nokogiri::XML::Node::ELEMENT_NODE) && (node.name == 'a')
-        append_attribute(node, 'rel', 'nofollow')
+        return CONTINUE unless (node.type == Nokogiri::XML::Node::ELEMENT_NODE) && (node.name == "a")
+        append_attribute(node, "rel", "nofollow")
         return STOP
       end
     end
@@ -226,8 +223,8 @@ module Loofah
       end
 
       def scrub(node)
-        return CONTINUE unless (node.type == Nokogiri::XML::Node::ELEMENT_NODE) && (node.name == 'a')
-        append_attribute(node, 'rel', 'noopener')
+        return CONTINUE unless (node.type == Nokogiri::XML::Node::ELEMENT_NODE) && (node.name == "a")
+        append_attribute(node, "rel", "noopener")
         return STOP
       end
     end
@@ -239,8 +236,13 @@ module Loofah
       end
 
       def scrub(node)
-        return CONTINUE unless Loofah::Elements::BLOCK_LEVEL.include?(node.name)
-        node.add_next_sibling Nokogiri::XML::Text.new("\n#{node.content}\n", node.document)
+        return CONTINUE unless Loofah::Elements::LINEBREAKERS.include?(node.name)
+        replacement = if Loofah::Elements::INLINE_LINE_BREAK.include?(node.name)
+          "\n"
+        else
+          "\n#{node.content}\n"
+        end
+        node.add_next_sibling Nokogiri::XML::Text.new(replacement, node.document)
         node.remove
       end
     end
@@ -267,7 +269,7 @@ module Loofah
 
       def scrub(node)
         if node.type == Nokogiri::XML::Node::TEXT_NODE || node.type == Nokogiri::XML::Node::CDATA_SECTION_NODE
-          node.content = node.content.gsub(/\u2028|\u2029/, '')
+          node.content = node.content.gsub(/\u2028|\u2029/, "")
         end
         CONTINUE
       end
@@ -277,14 +279,14 @@ module Loofah
     #  A hash that maps a symbol (like +:prune+) to the appropriate Scrubber (Loofah::Scrubbers::Prune).
     #
     MAP = {
-      :escape    => Escape,
-      :prune     => Prune,
+      :escape => Escape,
+      :prune => Prune,
       :whitewash => Whitewash,
-      :strip     => Strip,
-      :nofollow  => NoFollow,
+      :strip => Strip,
+      :nofollow => NoFollow,
       :noopener => NoOpener,
       :newline_block_elements => NewlineBlockElements,
-      :unprintable => Unprintable
+      :unprintable => Unprintable,
     }
 
     #

data/lib/loofah/version.rb

@@ -0,0 +1,5 @@
+# frozen_string_literal: true
+module Loofah
+  # The version of Loofah you are using
+  VERSION = "2.19.1"
+end

data/lib/loofah/xml/document.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Loofah
   module XML # :nodoc:
     #

data/lib/loofah/xml/document_fragment.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Loofah
   module XML # :nodoc:
     #
@@ -12,7 +13,7 @@ module Loofah
         #  constructor. Applications should use Loofah.fragment to
         #  parse a fragment.
         #
-        def parse tags
+        def parse(tags)
           doc = Loofah::XML::Document.new
           doc.encoding = tags.encoding.name if tags.respond_to?(:encoding)
           self.new(doc, tags)

data/lib/loofah.rb

@@ -1,22 +1,24 @@
+# frozen_string_literal: true
 $LOAD_PATH.unshift(File.expand_path(File.dirname(__FILE__))) unless $LOAD_PATH.include?(File.expand_path(File.dirname(__FILE__)))
 
 require "nokogiri"
 
-require "loofah/metahelpers"
-require "loofah/elements"
+require_relative "loofah/version"
+require_relative "loofah/metahelpers"
+require_relative "loofah/elements"
 
-require "loofah/html5/safelist"
-require "loofah/html5/libxml2_workarounds"
-require "loofah/html5/scrub"
+require_relative "loofah/html5/safelist"
+require_relative "loofah/html5/libxml2_workarounds"
+require_relative "loofah/html5/scrub"
 
-require "loofah/scrubber"
-require "loofah/scrubbers"
+require_relative "loofah/scrubber"
+require_relative "loofah/scrubbers"
 
-require "loofah/instance_methods"
-require "loofah/xml/document"
-require "loofah/xml/document_fragment"
-require "loofah/html/document"
-require "loofah/html/document_fragment"
+require_relative "loofah/instance_methods"
+require_relative "loofah/xml/document"
+require_relative "loofah/xml/document_fragment"
+require_relative "loofah/html/document"
+require_relative "loofah/html/document_fragment"
 
 # == Strings and IO Objects as Input
 #
@@ -27,14 +29,11 @@ require "loofah/html/document_fragment"
 # quantities of docs.
 #
 module Loofah
-  # The version of Loofah you are using
-  VERSION = "2.3.1"
-
   class << self
     # Shortcut for Loofah::HTML::Document.parse
     # This method accepts the same parameters as Nokogiri::HTML::Document.parse
     def document(*args, &block)
-      Loofah::HTML::Document.parse(*args, &block)
+      remove_comments_before_html_element Loofah::HTML::Document.parse(*args, &block)
     end
 
     # Shortcut for Loofah::HTML::DocumentFragment.parse
@@ -79,5 +78,23 @@ module Loofah
     def remove_extraneous_whitespace(string)
       string.gsub(/\n\s*\n\s*\n/, "\n\n")
     end
+
+    private
+
+    # remove comments that exist outside of the HTML element.
+    #
+    # these comments are allowed by the HTML spec:
+    #
+    #    https://www.w3.org/TR/html401/struct/global.html#h-7.1
+    #
+    # but are not scrubbed by Loofah because these nodes don't meet
+    # the contract that scrubbers expect of a node (e.g., it can be
+    # replaced, sibling and children nodes can be created).
+    def remove_comments_before_html_element(doc)
+      doc.children.each do |child|
+        child.unlink if child.comment?
+      end
+      doc
+    end
   end
 end