test: update tests to reflect new libxml2 HTML5 parsing behaviors

See https://gitlab.gnome.org/GNOME/libxml2/-/issues/758#note_2217350

test/html4/test_comments.rb

@@ -96,7 +96,19 @@ class TestComment < Nokogiri::TestCase
         let(:doc) { Nokogiri::HTML4(html) }
         let(:subject) { doc.at_css("div#under-test") }
 
-        if Nokogiri.uses_libxml?
+        if Nokogiri.uses_libxml?(">= 2.14.0")
+          it "behaves as if the comment is closed immediately before the end of the input stream" do # COMPLIANT
+            assert_pattern do
+              subject => {
+                name: "div",
+                attributes: [{ name: "id", value: "under-test" }],
+                children: [
+                  { name: "comment", content: "start of unterminated comment" }
+                ]
+              }
+            end
+          end
+        elsif Nokogiri.uses_libxml?
           it "behaves as if the comment is unterminated and doesn't exist" do # NON-COMPLIANT
             assert_equal 0, subject.children.length
             assert_equal 1, doc.errors.length
@@ -132,8 +144,12 @@ class TestComment < Nokogiri::TestCase
             assert_equal inner_div, subject.children[1]
             assert_predicate subject.children[2], :comment?
             assert_equal "bar", subject.children[2].content
-            assert_equal 1, doc.errors.length
-            assert_match(/Comment incorrectly closed/, doc.errors.first.to_s)
+            if Nokogiri.uses_libxml?(">= 2.14.0")
+              assert_empty doc.errors
+            else
+              assert_equal 1, doc.errors.length
+              assert_match(/Comment incorrectly closed/, doc.errors.first.to_s)
+            end
           end
         else # jruby, or libxml2 system lib less than 2.9.11
           it "behaves as if the comment encompasses the inner div" do # NON-COMPLIANT
@@ -161,7 +177,22 @@ class TestComment < Nokogiri::TestCase
         let(:body) { doc.at_css("body") }
         let(:subject) { doc.at_css("div#under-test") }
 
-        if Nokogiri.uses_libxml?("= 2.9.14")
+        if Nokogiri.uses_libxml?(">= 2.14.0")
+          it "parses as comments" do # COMPLIANT
+            assert_pattern do
+              body.children => [
+                {
+                  name: "div",
+                  children: [
+                    { name: "comment", content: " comment <div id=do-i-exist" },
+                    { name: "text", content: "inner content" },
+                  ]
+                },
+                { name: "text", content: "-->hello" },
+              ]
+            end
+          end
+        elsif Nokogiri.uses_libxml?("= 2.9.14")
           it "parses as PCDATA" do # NON-COMPLIANT
             assert_equal 1, body.children.length
             assert_equal subject, body.children.first
@@ -212,7 +243,21 @@ class TestComment < Nokogiri::TestCase
         let(:body) { doc.at_css("body") }
         let(:subject) { doc.at_css("div#under-test") }
 
-        if Nokogiri.uses_libxml?("= 2.9.14")
+        if Nokogiri.uses_libxml?(">= 2.14.0")
+          it "parses the <! tags as comments" do
+            assert_pattern do
+              body.children => [
+                {
+                  name: "div", children: [
+                    { name: "comment", content: "[if foo]" },
+                    { name: "div", attributes: [{name: "id", value: "do-i-exist"}] },
+                    { name: "comment", content: "[endif]" },
+                  ]
+                }
+              ]
+            end
+          end
+        elsif Nokogiri.uses_libxml?("= 2.9.14")
           it "parses the <! tags as PCDATA" do
             assert_equal(1, body.children.length)
             assert_equal(subject, body.children.first)

test/html4/test_document.rb

@@ -734,6 +734,8 @@ def test_silencing_nonparse_errors_during_attribute_insertion_1262
           doc = Nokogiri::HTML4::Document.parse(html)
           expected = if Nokogiri.jruby?
             [Nokogiri::XML::Node::COMMENT_NODE, Nokogiri::XML::Node::PI_NODE]
+          elsif Nokogiri.uses_libxml?(">= 2.14.0")
+            [Nokogiri::XML::Node::COMMENT_NODE, Nokogiri::XML::Node::COMMENT_NODE]
           elsif Nokogiri.uses_libxml?(">= 2.10.0")
             [Nokogiri::XML::Node::COMMENT_NODE]
           else

test/html4/test_document_fragment.rb

@@ -188,7 +188,29 @@ def test_element_children_counts
 
         def test_malformed_fragment_is_corrected
           fragment = Nokogiri::HTML4::DocumentFragment.parse("<div </div>")
-          assert_equal("<div></div>", fragment.to_s)
+
+          if Nokogiri.uses_libxml?(">= 2.14.0")
+            assert_pattern do
+              fragment => [
+                { name: "div", attributes: [
+                    { name: "<", value: ""},
+                    { name: "div", value: ""},
+                  ]}
+              ]
+            end
+          else
+            assert_equal("<div></div>", fragment.to_s)
+          end
+        end
+
+        def test_malformed_html5_fragment_serializes_like_gumbo
+          skip_unless_libxml2(">= 2.14.0")
+
+          fragment = Nokogiri::HTML4::DocumentFragment.parse("<div </div>")
+
+          pending "libxml2 does not serialize HTML5 like gumbo (yet)" do
+            assert_equal('<div <="" div=""></div>', fragment.to_s)
+          end
         end
 
         def test_unclosed_script_tag

test/xml/test_node.rb

@@ -105,9 +105,19 @@ def test_node_context_parsing_of_malformed_html_fragment
           context_node = doc.at_css("div")
           nodeset = context_node.parse("<div </div>")
 
-          assert_equal(1, doc.errors.length)
-          assert_equal(1, nodeset.length)
-          assert_equal("<div></div>", nodeset.to_s)
+          if Nokogiri.uses_libxml?(">= 2.14.0")
+            assert_empty(doc.errors)
+            assert_pattern do
+              nodeset => [
+                { name: "div", attributes: [{name: "<", value: ""}, { name: "div", value: ""}] },
+              ]
+            end
+          else
+            assert_equal(1, doc.errors.length)
+            assert_equal(1, nodeset.length)
+            assert_equal("<div></div>", nodeset.to_s)
+          end
+
           assert_instance_of(Nokogiri::HTML4::Document, nodeset.document)
           assert_instance_of(Nokogiri::HTML4::Document, nodeset.first.document)
         end
@@ -117,14 +127,25 @@ def test_node_context_parsing_of_malformed_html_fragment_with_recover_is_correct
           context_node = doc.at_css("div")
           nodeset = context_node.parse("<div </div>", &:recover)
 
-          assert_equal(1, doc.errors.length)
-          assert_equal(1, nodeset.length)
-          assert_equal("<div></div>", nodeset.to_s)
+          if Nokogiri.uses_libxml?(">= 2.14.0")
+            assert_empty(doc.errors)
+            assert_pattern do
+              nodeset => [
+                { name: "div", attributes: [{name: "<", value: ""}, { name: "div", value: ""}] },
+              ]
+            end
+          else
+            assert_equal(1, doc.errors.length)
+            assert_equal(1, nodeset.length)
+            assert_equal("<div></div>", nodeset.to_s)
+          end
           assert_instance_of(Nokogiri::HTML4::Document, nodeset.document)
           assert_instance_of(Nokogiri::HTML4::Document, nodeset.first.document)
         end
 
         def test_node_context_parsing_of_malformed_html_fragment_without_recover_is_not_corrected
+          skip("libxml2 2.14.0 no longer raises this error") if Nokogiri.uses_libxml?(">= 2.14.0")
+
           doc = HTML4.parse("<html><body><div></div></body></html>")
           context_node = doc.at_css("div")
           assert_raises(Nokogiri::XML::SyntaxError) do