fix: make sure HTML5::Document{,Fragment} subclass properly

Loofah and other downstream libraries rely on this behavior. This is
long-term prep for a day when HTML5 may become the default on
supported platforms.

ext/nokogiri/gumbo.c

@@ -23,7 +23,7 @@
 //
 // Processing starts by calling gumbo_parse_with_options. The resulting document tree
 // is then walked, a parallel libxml2 tree is constructed, and the final document is
-// then wrapped using Nokogiri_wrap_xml_document. This approach reduces memory and CPU
+// then wrapped using noko_xml_document_wrap. This approach reduces memory and CPU
 // requirements as Ruby objects are only built when necessary.
 //
 
@@ -297,6 +297,7 @@ typedef struct {
   GumboOutput *output;
   VALUE input;
   VALUE url_or_frag;
+  VALUE klass;
   xmlDocPtr doc;
 } ParseArgs;
 
@@ -321,7 +322,7 @@ static VALUE parse_continue(VALUE parse_args);
  *  @!visibility protected
  */
 static VALUE
-parse(VALUE self, VALUE input, VALUE url, VALUE max_attributes, VALUE max_errors, VALUE max_depth)
+parse(VALUE self, VALUE input, VALUE url, VALUE max_attributes, VALUE max_errors, VALUE max_depth, VALUE klass)
 {
   GumboOptions options = kGumboDefaultOptions;
   options.max_attributes = NUM2INT(max_attributes);
@@ -333,6 +334,7 @@ parse(VALUE self, VALUE input, VALUE url, VALUE max_attributes, VALUE max_errors
     .output = output,
     .input = input,
     .url_or_frag = url,
+    .klass = klass,
     .doc = NULL,
   };
 
@@ -357,7 +359,7 @@ parse_continue(VALUE parse_args)
   }
   args->doc = doc; // Make sure doc gets cleaned up if an error is thrown.
   build_tree(doc, (xmlNodePtr)doc, output->document);
-  VALUE rdoc = Nokogiri_wrap_xml_document(cNokogiriHtml5Document, doc);
+  VALUE rdoc = noko_xml_document_wrap(args->klass, doc);
   args->doc = NULL; // The Ruby runtime now owns doc so don't delete it.
   add_errors(output, rdoc, args->input, args->url_or_frag);
   return rdoc;
@@ -577,7 +579,7 @@ noko_init_gumbo()
   parent = rb_intern_const("parent");
 
   // Define Nokogumbo module with parse and fragment methods.
-  rb_define_singleton_method(mNokogiriGumbo, "parse", parse, 5);
+  rb_define_singleton_method(mNokogiriGumbo, "parse", parse, 6);
   rb_define_singleton_method(mNokogiriGumbo, "fragment", fragment, 6);
 }
 

lib/nokogiri/html5/document.rb

@@ -63,7 +63,7 @@ def do_parse(string_or_io, url, encoding, options)
           max_attributes = options[:max_attributes] || Nokogiri::Gumbo::DEFAULT_MAX_ATTRIBUTES
           max_errors = options[:max_errors] || options[:max_parse_errors] || Nokogiri::Gumbo::DEFAULT_MAX_ERRORS
           max_depth = options[:max_tree_depth] || Nokogiri::Gumbo::DEFAULT_MAX_TREE_DEPTH
-          doc = Nokogiri::Gumbo.parse(string, url, max_attributes, max_errors, max_depth)
+          doc = Nokogiri::Gumbo.parse(string, url, max_attributes, max_errors, max_depth, self)
           doc.encoding = "UTF-8"
           doc
         end

test/html5/test_api.rb

@@ -182,4 +182,122 @@ def test_html_eh
     assert_predicate(doc, :html?)
     refute_predicate(doc, :xml?)
   end
+
+  describe Nokogiri::HTML5::Document do
+    describe "subclassing" do
+      let(:klass) do
+        Class.new(Nokogiri::HTML5::Document) do
+          attr_accessor :initialized_with, :initialized_count
+
+          def initialize(*args)
+            super
+            @initialized_with = args
+            @initialized_count ||= 0
+            @initialized_count += 1
+          end
+        end
+      end
+
+      describe ".new" do
+        it "returns an instance of the expected class" do
+          doc = klass.new
+          assert_instance_of(klass, doc)
+        end
+
+        it "calls #initialize exactly once" do
+          doc = klass.new
+          assert_equal(1, doc.initialized_count)
+        end
+
+        it "passes arguments to #initialize" do
+          doc = klass.new("http://www.w3.org/TR/REC-html40/loose.dtd", "-//W3C//DTD HTML 4.0 Transitional//EN")
+          assert_equal(
+            ["http://www.w3.org/TR/REC-html40/loose.dtd", "-//W3C//DTD HTML 4.0 Transitional//EN"],
+            doc.initialized_with
+          )
+        end
+      end
+
+      it "#dup returns the expected class" do
+        doc = klass.new.dup
+        assert_instance_of(klass, doc)
+      end
+
+      describe ".parse" do
+        let(:html) { Nokogiri::HTML5.parse(File.read(HTML_FILE)) }
+
+        it "returns an instance of the expected class" do
+          doc = klass.parse(File.read(HTML_FILE))
+          assert_instance_of(klass, doc)
+        end
+
+        it "calls #initialize exactly once" do
+          doc = klass.parse(File.read(HTML_FILE))
+          assert_equal(1, doc.initialized_count)
+        end
+
+        it "parses the doc" do
+          doc = klass.parse(File.read(HTML_FILE))
+          assert_equal(html.root.to_s, doc.root.to_s)
+        end
+      end
+    end
+  end
+
+  describe Nokogiri::HTML5::DocumentFragment do
+    describe "subclassing" do
+      let(:klass) do
+        Class.new(Nokogiri::HTML5::DocumentFragment) do
+          attr_accessor :initialized_with, :initialized_count
+
+          def initialize(*args)
+            super
+            @initialized_with = args
+            @initialized_count ||= 0
+            @initialized_count += 1
+          end
+        end
+      end
+      let(:html) { Nokogiri::HTML5.parse(File.read(HTML_FILE), HTML_FILE) }
+
+      describe ".new" do
+        it "returns an instance of the right class" do
+          fragment = klass.new(html, "<div>a</div>")
+          assert_instance_of(klass, fragment)
+        end
+
+        it "calls #initialize exactly once" do
+          fragment = klass.new(html, "<div>a</div>")
+          assert_equal(1, fragment.initialized_count)
+        end
+
+        it "passes args to #initialize" do
+          fragment = klass.new(html, "<div>a</div>")
+          assert_equal([html, "<div>a</div>"], fragment.initialized_with)
+        end
+      end
+
+      it "#dup returns the expected class" do
+        doc = klass.new(html, "<div>a</div>").dup
+        assert_instance_of(klass, doc)
+      end
+
+      describe ".parse" do
+        it "returns an instance of the right class" do
+          fragment = klass.parse("<div>a</div>")
+          assert_instance_of(klass, fragment)
+        end
+
+        it "calls #initialize exactly once" do
+          fragment = klass.parse("<div>a</div>")
+          assert_equal(1, fragment.initialized_count)
+        end
+
+        it "passes the fragment" do
+          fragment = klass.parse("<div>a</div>")
+          assert_equal(Nokogiri::HTML5::DocumentFragment.parse("<div>a</div>").to_s, fragment.to_s)
+        end
+      end
+    end
+  end
 end if Nokogiri.uses_gumbo?