html2: Handle </br> correctly

robinlinden · Jan 6, 2025 · c2866ec · c2866ec
1 parent 1d405d3
commit c2866ec
Show file tree

Hide file tree

Showing 2 changed files with 33 additions and 8 deletions.
diff --git a/html2/parser_states.cpp b/html2/parser_states.cpp
@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: 2023-2024 Robin Lindén <[email protected]>
+// SPDX-FileCopyrightText: 2023-2025 Robin Lindén <[email protected]>
 //
 // SPDX-License-Identifier: BSD-2-Clause
 
@@ -901,9 +901,17 @@ std::optional<InsertionMode> InBody::process(IActions &a, html2::Token const &to
             "keygen",
             "wbr",
     });
-    if (start != nullptr && is_in_array<kImmediatelyPoppedElements>(start->tag_name)) {
-        a.reconstruct_active_formatting_elements();
-        a.insert_element_for(*start);
+    auto is_bad_br_end_tag = end != nullptr && end->tag_name == "br";
+    if ((start != nullptr && is_in_array<kImmediatelyPoppedElements>(start->tag_name)) || (is_bad_br_end_tag)) {
+        if (is_bad_br_end_tag) {
+            // Parse error.
+            a.reconstruct_active_formatting_elements();
+            a.insert_element_for(html2::StartTagToken{.tag_name = "br"});
+        } else {
+            a.reconstruct_active_formatting_elements();
+            a.insert_element_for(*start);
+        }
+
         a.pop_current_node();
         // TODO(robinlinden): Acknowledge the token's self-closing flag, if it is set.
         a.set_frameset_ok(false);

diff --git a/html2/parser_states_test.cpp b/html2/parser_states_test.cpp
@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: 2023-2024 Robin Lindén <[email protected]>
+// SPDX-FileCopyrightText: 2023-2025 Robin Lindén <[email protected]>
 //
 // SPDX-License-Identifier: BSD-2-Clause
 
@@ -314,9 +314,15 @@ void in_head_noscript_tests(etest::Suite &s) {
 
     s.add_test("InHeadNoScript: br", [](etest::IActions &a) {
         auto res = parse("<noscript></br>", {});
-        auto noscript = dom::Element{"noscript"};
         a.expect_eq(res.document.html(),
-                dom::Element{"html", {}, {dom::Element{"head", {}, {std::move(noscript)}}, dom::Element{"body"}}});
+                dom::Element{
+                        "html",
+                        {},
+                        {
+                                dom::Element{"head", {}, {dom::Element{"noscript"}}},
+                                dom::Element{"body", {}, {dom::Element{"br"}}},
+                        },
+                });
     });
 
     s.add_test("InHeadNoScript: noscript", [](etest::IActions &a) {
@@ -388,7 +394,12 @@ void after_head_tests(etest::Suite &s) {
 
     s.add_test("AfterHead: </br>", [](etest::IActions &a) {
         auto res = parse("<head></head></br>", {});
-        a.expect_eq(res.document.html(), dom::Element{"html", {}, {dom::Element{"head"}, dom::Element{"body"}}});
+        a.expect_eq(res.document.html(),
+                dom::Element{
+                        "html",
+                        {},
+                        {dom::Element{"head"}, dom::Element{"body", {}, {dom::Element{"br"}}}},
+                });
     });
 
     s.add_test("AfterHead: </error>", [](etest::IActions &a) {
@@ -515,6 +526,12 @@ void in_body_tests(etest::Suite &s) {
         a.expect_eq(body, dom::Element{"body", {}, {dom::Element{"p"}, dom::Element{"hr"}}});
     });
 
+    s.add_test("InBody: </br>", [](etest::IActions &a) {
+        auto res = parse("<body></br>", {});
+        auto const &body = std::get<dom::Element>(res.document.html().children.at(1));
+        a.expect_eq(body, dom::Element{"body", {}, {dom::Element{"br"}}});
+    });
+
     s.add_test("InBody: <template> doesn't crash", [](etest::IActions &) {
         std::ignore = parse("<body><template>", {}); //
     });