LibWeb: Yet more work towards parsing www/welcome.html :^)

4 files changed, 111 insertions, 7 deletions

diff --git a/Libraries/LibWeb/Parser/HTMLDocumentParser.cpp b/Libraries/LibWeb/Parser/HTMLDocumentParser.cpp
index 3b7aad8ad3..08200a38f0 100644
--- a/Libraries/LibWeb/Parser/HTMLDocumentParser.cpp
+++ b/Libraries/LibWeb/Parser/HTMLDocumentParser.cpp
@@ -346,13 +346,27 @@ AnythingElse:
     // FIXME: Reprocess the current token in InBody!
 }
 
-void HTMLDocumentParser::generate_implied_end_tags()
+void HTMLDocumentParser::generate_implied_end_tags(const FlyString& exception)
 {
-    Vector<String> names { "dd", "dt", "li", "optgroup", "option", "p", "rb", "rp", "rt", "rtc" };
-    while (names.contains_slow(current_node().tag_name()))
+    static Vector<FlyString> names { "dd", "dt", "li", "optgroup", "option", "p", "rb", "rp", "rt", "rtc" };
+    while (current_node().tag_name() != exception && names.contains_slow(current_node().tag_name()))
         m_stack_of_open_elements.pop();
 }
 
+void HTMLDocumentParser::close_a_p_element()
+{
+    generate_implied_end_tags("p");
+    if (current_node().tag_name() != "p") {
+        // Parse error.
+        TODO();
+    }
+    for (;;) {
+        auto popped_element = m_stack_of_open_elements.pop();
+        if (popped_element->tag_name() == "p")
+            break;
+    }
+}
+
 void HTMLDocumentParser::handle_after_body(HTMLToken& token)
 {
     if (token.is_character() && token.is_parser_whitespace()) {
@@ -403,6 +417,10 @@ void HTMLDocumentParser::handle_in_body(HTMLToken& token)
             insert_character(token.codepoint());
             return;
         }
+        reconstruct_the_active_formatting_elements();
+        insert_character(token.codepoint());
+        m_frameset_ok = false;
+        return;
     }
 
     if (token.is_end_tag() && token.tag_name() == "body") {
@@ -422,6 +440,70 @@ void HTMLDocumentParser::handle_in_body(HTMLToken& token)
     }
 
     {
+        static Vector<FlyString> names { "h1", "h2", "h3", "h4", "h5", "h6" };
+        if (token.is_start_tag() && names.contains_slow(token.tag_name())) {
+            if (m_stack_of_open_elements.has_in_button_scope("p"))
+                close_a_p_element();
+            if (names.contains_slow(current_node().tag_name())) {
+                // FIXME: This is a parse error!
+                TODO();
+            }
+            insert_html_element(token);
+            return;
+        }
+    }
+
+    {
+        static Vector<FlyString> names { "h1", "h2", "h3", "h4", "h5", "h6" };
+        if (token.is_end_tag() && names.contains_slow(token.tag_name())) {
+            if (!m_stack_of_open_elements.has_in_scope("h1")
+                && !m_stack_of_open_elements.has_in_scope("h2")
+                && !m_stack_of_open_elements.has_in_scope("h3")
+                && !m_stack_of_open_elements.has_in_scope("h4")
+                && !m_stack_of_open_elements.has_in_scope("h5")
+                && !m_stack_of_open_elements.has_in_scope("h6")) {
+                TODO();
+            }
+
+            generate_implied_end_tags();
+            if (current_node().tag_name() != token.tag_name()) {
+                TODO();
+            }
+
+            for (;;) {
+                auto popped_element = m_stack_of_open_elements.pop();
+                if (popped_element->tag_name() == "h1"
+                    || popped_element->tag_name() == "h2"
+                    || popped_element->tag_name() == "h3"
+                    || popped_element->tag_name() == "h4"
+                    || popped_element->tag_name() == "h5"
+                    || popped_element->tag_name() == "h6") {
+                    break;
+                }
+            }
+            return;
+        }
+    }
+
+    if (token.is_end_tag() && token.tag_name() == "p") {
+        if (!m_stack_of_open_elements.has_in_button_scope("p")) {
+            TODO();
+        }
+        close_a_p_element();
+        return;
+    }
+
+    {
+        static Vector<FlyString> names { "b", "big", "code", "em", "font", "i", "s", "small", "strike", "strong", "tt", "u" };
+        if (token.is_start_tag() && names.contains_slow(token.tag_name())) {
+            reconstruct_the_active_formatting_elements();
+            auto element = insert_html_element(token);
+            m_list_of_active_formatting_elements.append(*element);
+            return;
+        }
+    }
+
+    {
         Vector<String> names { "address", "article", "aside", "blockquote", "center", "details", "dialog", "dir", "div", "dl", "fieldset", "figcaption", "figure", "footer", "header", "hgroup", "main", "menu", "nav", "ol", "p", "section", "summary", "ul" };
         if (token.is_start_tag() && names.contains_slow(token.tag_name())) {
             // FIXME: If the stack of open elements has a p element in button scope, then close a p element.
@@ -447,6 +529,12 @@ void HTMLDocumentParser::handle_in_body(HTMLToken& token)
         }
     }
 
+    if (token.is_start_tag()) {
+        reconstruct_the_active_formatting_elements();
+        insert_html_element(token);
+        return;
+    }
+
     ASSERT_NOT_REACHED();
 }
 
@@ -504,5 +592,4 @@ Document& HTMLDocumentParser::document()
 {
     return *m_document;
 }
-
 }
diff --git a/Libraries/LibWeb/Parser/HTMLDocumentParser.h b/Libraries/LibWeb/Parser/HTMLDocumentParser.h
index 09b92fbe20..0e7c09e207 100644
--- a/Libraries/LibWeb/Parser/HTMLDocumentParser.h
+++ b/Libraries/LibWeb/Parser/HTMLDocumentParser.h
@@ -89,7 +89,7 @@ private:
     void handle_after_after_body(HTMLToken&);
     void handle_text(HTMLToken&);
 
-    void generate_implied_end_tags();
+    void generate_implied_end_tags(const FlyString& exception = {});
     bool stack_of_open_elements_has_element_with_tag_name_in_scope(const FlyString& tag_name);
     NonnullRefPtr<Element> create_element_for(HTMLToken&);
     RefPtr<Node> find_appropriate_place_for_inserting_node();
@@ -98,6 +98,7 @@ private:
     void insert_character(u32 data);
     void insert_comment(HTMLToken&);
     void reconstruct_the_active_formatting_elements();
+    void close_a_p_element();
     void process_using_the_rules_for(InsertionMode, HTMLToken&);
     void parse_generic_raw_text_element(HTMLToken&);
     void increment_script_nesting_level();
diff --git a/Libraries/LibWeb/Parser/StackOfOpenElements.cpp b/Libraries/LibWeb/Parser/StackOfOpenElements.cpp
index ad607254d6..1c99624446 100644
--- a/Libraries/LibWeb/Parser/StackOfOpenElements.cpp
+++ b/Libraries/LibWeb/Parser/StackOfOpenElements.cpp
@@ -3,13 +3,14 @@
 
 namespace Web {
 
+static Vector<FlyString> s_base_list { "applet", "caption", "html", "table", "td", "th", "marquee", "object", "template" };
+
 StackOfOpenElements::~StackOfOpenElements()
 {
 }
 
-bool StackOfOpenElements::has_in_scope(const FlyString& tag_name) const
+bool StackOfOpenElements::has_in_scope_impl(const FlyString& tag_name, const Vector<FlyString> &list) const
 {
-    static Vector<FlyString> list { "applet", "caption", "html", "table", "td", "th", "marquee", "object", "template" };
     for (ssize_t i = m_elements.size() - 1; i >= 0; --i) {
         auto& node = m_elements.at(i);
         if (node.tag_name() == tag_name)
@@ -20,4 +21,16 @@ bool StackOfOpenElements::has_in_scope(const FlyString& tag_name) const
     ASSERT_NOT_REACHED();
 }
 
+bool StackOfOpenElements::has_in_scope(const FlyString& tag_name) const
+{
+    return has_in_scope_impl(tag_name, s_base_list);
+}
+
+bool StackOfOpenElements::has_in_button_scope(const FlyString& tag_name) const
+{
+    auto list = s_base_list;
+    list.append("button");
+    return has_in_scope_impl(tag_name, list);
+}
+
 }
diff --git a/Libraries/LibWeb/Parser/StackOfOpenElements.h b/Libraries/LibWeb/Parser/StackOfOpenElements.h
index 56eaad3115..98ac36434b 100644
--- a/Libraries/LibWeb/Parser/StackOfOpenElements.h
+++ b/Libraries/LibWeb/Parser/StackOfOpenElements.h
@@ -45,8 +45,11 @@ public:
     Element& current_node() { return m_elements.last(); }
 
     bool has_in_scope(const FlyString& tag_name) const;
+    bool has_in_button_scope(const FlyString& tag_name) const;
 
 private:
+    bool has_in_scope_impl(const FlyString& tag_name, const Vector<FlyString>&) const;
+
     NonnullRefPtrVector<Element> m_elements;
 };