html2: Handle dd & dt start tags

robinlinden · Nov 26, 2024 · 3a9dd9e · 3a9dd9e
1 parent 1490b83
commit 3a9dd9e
Show file tree

Hide file tree

Showing 2 changed files with 70 additions and 34 deletions.
diff --git a/html2/parser_states.cpp b/html2/parser_states.cpp
@@ -858,6 +858,39 @@ std::optional<InsertionMode> InBody::process(IActions &a, html2::Token const &to
         return {};
     }
 
+    if (start != nullptr && (start->tag_name == "dd" || start->tag_name == "dt")) {
+        a.set_frameset_ok(false);
+
+        auto open_elements = a.names_of_open_elements();
+        assert(!open_elements.empty());
+        for (auto node : open_elements) {
+            if (node == "dd" || node == "dt") {
+                generate_implied_end_tags(a, node);
+                if (a.current_node_name() != node) {
+                    // Parse error.
+                }
+
+                while (a.current_node_name() != node) {
+                    a.pop_current_node();
+                }
+
+                a.pop_current_node();
+                break;
+            }
+
+            if (is_special(node) && node != "address" && node != "div" && node != "p") {
+                break;
+            }
+        }
+
+        if (a.has_element_in_button_scope("p")) {
+            close_a_p_element();
+        }
+
+        a.insert_element_for(*start);
+        return {};
+    }
+
     // TODO(robinlinden): Most things.
 
     static constexpr auto kImmediatelyPoppedElements = std::to_array<std::string_view>({

diff --git a/html2/parser_states_test.cpp b/html2/parser_states_test.cpp
@@ -11,7 +11,9 @@
 #include "etest/etest.h"
 #include "html/parser_actions.h"
 
+#include <array>
 #include <cstddef>
+#include <format>
 #include <string_view>
 #include <tuple>
 #include <utility>
@@ -519,40 +521,41 @@ void in_body_tests() {
         std::ignore = parse("<body><template>", {}); //
     });
 
-    etest::test("InBody: <li>", [] {
-        auto res = parse("<body><ul><li><p>hello<li>world", {});
-        auto const &body = std::get<dom::Element>(res.document.html().children.at(1));
-        expect_eq(body.children.size(), std::size_t{1});
-        auto const &ul = std::get<dom::Element>(body.children.at(0));
-
-        expect_eq(ul,
-                dom::Element{
-                        "ul",
-                        {},
-                        {
-                                dom::Element{"li", {}, {dom::Element{"p", {}, {dom::Text{"hello"}}}}},
-                                dom::Element{"li", {}, {dom::Text{"world"}}},
-                        },
-                });
-    });
-
-    etest::test("InBody: <li>, weird specials", [] {
-        auto res = parse("<body><section><p><li>hello<a><li>world", {});
-        auto const &body = std::get<dom::Element>(res.document.html().children.at(1));
-        expect_eq(body.children.size(), std::size_t{1});
-        auto const &section = std::get<dom::Element>(body.children.at(0));
-
-        expect_eq(section,
-                dom::Element{
-                        "section",
-                        {},
-                        {
-                                dom::Element{"p"},
-                                dom::Element{"li", {}, {dom::Text{"hello"}, dom::Element{"a"}}},
-                                dom::Element{"li", {}, {dom::Text{"world"}}},
-                        },
-                });
-    });
+    for (auto tag : std::to_array<std::string_view>({"li", "dt", "dd"})) {
+        etest::test(std::format("InBody: <{}>", tag), [tag] {
+            auto html = std::format("<body><{}><p>hello<{}>world", tag, tag);
+            auto res = parse(html, {});
+            auto const &body = std::get<dom::Element>(res.document.html().children.at(1));
+
+            expect_eq(body,
+                    dom::Element{
+                            "body",
+                            {},
+                            {
+                                    dom::Element{std::string{tag}, {}, {dom::Element{"p", {}, {dom::Text{"hello"}}}}},
+                                    dom::Element{std::string{tag}, {}, {dom::Text{"world"}}},
+                            },
+                    });
+        });
+
+        etest::test(std::format("InBody: <{}>, weird specials", tag), [tag] {
+            auto res = parse(std::format("<body><section><p><{}>hello<a><{}>world", tag, tag), {});
+            auto const &body = std::get<dom::Element>(res.document.html().children.at(1));
+            expect_eq(body.children.size(), std::size_t{1});
+            auto const &section = std::get<dom::Element>(body.children.at(0));
+
+            expect_eq(section,
+                    dom::Element{
+                            "section",
+                            {},
+                            {
+                                    dom::Element{"p"},
+                                    dom::Element{std::string{tag}, {}, {dom::Text{"hello"}, dom::Element{"a"}}},
+                                    dom::Element{std::string{tag}, {}, {dom::Text{"world"}}},
+                            },
+                    });
+        });
+    }
 
     etest::test("InBody: body end tag, disallowed element", [] {
         auto res = parse("<body><foo></body>", {});