Introduce an acts_as_loofah method

to make Nokogiri instances act like Loofah instances.

These classes have a new instance method, `#acts_as_loofah`:

- Nokogiri::XML::Document
- Nokogiri::XML::DocumentFragment
- Nokogiri::HTML4::Document
- Nokogiri::HTML4::DocumentFragment
- Nokogiri::HTML5::Document
- Nokogiri::HTML5::DocumentFragment

This method extends the Nokogiri object so that it quacks like the
corresponding Loofah object.

Note that this method will also extend any existing child element
objects, just as if those objects were created as children of a Loofah
document.

lib/loofah/concerns.rb

@@ -123,10 +123,16 @@ def to_text(options = {})
   end
 
   module DocumentDecorator # :nodoc:
+    class << self
+      def decorate(object)
+        object.decorators(Nokogiri::XML::Node) << ScrubBehavior::Node
+        object.decorators(Nokogiri::XML::NodeSet) << ScrubBehavior::NodeSet
+      end
+    end
+
     def initialize(*args, &block)
       super
-      decorators(Nokogiri::XML::Node) << ScrubBehavior::Node
-      decorators(Nokogiri::XML::NodeSet) << ScrubBehavior::NodeSet
+      DocumentDecorator.decorate(self)
     end
   end
 

lib/loofah/html4/document.rb

@@ -12,6 +12,16 @@ class Document < Nokogiri::HTML4::Document
       include Loofah::DocumentDecorator
       include Loofah::TextBehavior
       include Loofah::HtmlDocumentBehavior
+
+      module NokogiriExtender
+        def acts_as_loofah
+          super
+          singleton_class.include(Loofah::TextBehavior)
+          singleton_class.include(Loofah::HtmlDocumentBehavior)
+        end
+      end
     end
   end
 end
+
+Nokogiri::HTML4::Document.include(Loofah::HTML4::Document::NokogiriExtender)

lib/loofah/html4/document_fragment.rb

@@ -10,6 +10,16 @@ module HTML4 # :nodoc:
     class DocumentFragment < Nokogiri::HTML4::DocumentFragment
       include Loofah::TextBehavior
       include Loofah::HtmlFragmentBehavior
+
+      module NokogiriExtender
+        def acts_as_loofah
+          super
+          singleton_class.include(Loofah::TextBehavior)
+          singleton_class.include(Loofah::HtmlFragmentBehavior)
+        end
+      end
     end
   end
 end
+
+Nokogiri::HTML4::DocumentFragment.include(Loofah::HTML4::DocumentFragment::NokogiriExtender)

lib/loofah/xml/document.rb

@@ -5,11 +5,32 @@ module XML # :nodoc:
     #
     #  Subclass of Nokogiri::XML::Document.
     #
-    #  See Loofah::ScrubBehavior and Loofah::DocumentDecorator for additional methods.
+    #  See Loofah::ScrubBehavior for additional methods.
     #
     class Document < Nokogiri::XML::Document
       include Loofah::ScrubBehavior::Node
       include Loofah::DocumentDecorator
+
+      module NokogiriExtender
+        def acts_as_loofah
+          singleton_class.include(Loofah::ScrubBehavior::Node)
+          Loofah::DocumentDecorator.decorate(self)
+          decorate_existing
+        end
+
+        # TODO: this should to be upstreamed into Nokogiri
+        def decorate_existing # :nodoc:
+          return unless @decorators
+
+          if Nokogiri.jruby?
+            traverse { |node| decorate(node) }
+          else
+            @node_cache.each { |node| decorate(node) }
+          end
+        end
+      end
     end
   end
 end
+
+Nokogiri::XML::Document.include(Loofah::XML::Document::NokogiriExtender)

lib/loofah/xml/document_fragment.rb

@@ -15,6 +15,23 @@ def parse(tags)
           new(doc, tags)
         end
       end
+
+      module NokogiriExtender
+        def acts_as_loofah
+          document.acts_as_loofah
+          decorate_existing
+        end
+
+        # TODO: this should to be upstreamed into Nokogiri
+        def decorate_existing # :nodoc:
+          return unless Nokogiri.jruby?
+          return unless document.instance_variable_get(:@decorators)
+
+          traverse { |node| document.decorate(node) }
+        end
+      end
     end
   end
 end
+
+Nokogiri::XML::DocumentFragment.include(Loofah::XML::DocumentFragment::NokogiriExtender)

test/unit/test_acts_as_loofah.rb

@@ -0,0 +1,67 @@
+# frozen_string_literal: true
+
+require "helper"
+
+class UnitTestActsAsLoofah < Loofah::TestCase
+  SUBJECTS = [Nokogiri::XML, Nokogiri::HTML4, defined?(Nokogiri::HTML5) && Nokogiri::HTML5].compact
+
+  SUBJECTS.each do |subject|
+    describe subject do
+      it "Document act like Loofah" do
+        ndoc = subject::Document.parse("<html><body><div>hello</div><span>hello</span><script>alert(1)</script></body></html>")
+        node = ndoc.at_css("div")
+
+        # method presence
+        refute_respond_to(ndoc, :scrub!)
+        refute_respond_to(node, :scrub!)
+
+        ndoc.acts_as_loofah
+
+        assert_respond_to(ndoc, :scrub!, "#{subject}::Document should be extended")
+        assert_respond_to(ndoc.at_css("span"), :scrub!, "New child elements should be extended")
+        assert_respond_to(node, :scrub!, "Existing child elements should be extended")
+
+        refute_respond_to(subject::Document.parse("<div>"), :scrub!, "Other instances should not be extended")
+
+        # scrub behavior
+        ndoc.scrub!(:prune)
+
+        refute_includes(ndoc.to_html, "script")
+
+        # other concerns
+        if subject.name.include?("HTML")
+          assert_includes(ndoc.singleton_class.ancestors, Loofah::TextBehavior)
+          assert_includes(ndoc.singleton_class.ancestors, Loofah::HtmlDocumentBehavior)
+        end
+      end
+
+      it "DocumentFragment act like Loofah" do
+        nfrag = subject::DocumentFragment.parse("<div>hello</div><span>hello</span><script>alert(1)</script>")
+        node = nfrag.at_css("div")
+
+        # method presence
+        refute_respond_to(nfrag, :scrub!)
+        refute_respond_to(node, :scrub!)
+
+        nfrag.acts_as_loofah
+
+        assert_respond_to(nfrag, :scrub!, "#{subject}::DocumentFragment should be extended")
+        assert_respond_to(nfrag.at_css("span"), :scrub!, "New child elements should be extended")
+        assert_respond_to(node, :scrub!, "Existing child elements should be extended")
+
+        refute_respond_to(subject::DocumentFragment.parse("<div>"), :scrub!, "Other instances should not be extended")
+
+        # scrub behavior
+        nfrag.scrub!(:prune)
+
+        refute_includes(nfrag.to_html, "script")
+
+        # other concerns
+        if subject.name.include?("HTML")
+          assert_includes(nfrag.singleton_class.ancestors, Loofah::TextBehavior)
+          assert_includes(nfrag.singleton_class.ancestors, Loofah::HtmlFragmentBehavior)
+        end
+      end
+    end
+  end
+end