LibWeb: Change HTMLToken storage architecture

This completely changes how HTMLTokens store their data. Previously,
space was allocated for all token types separately. Now, the HTMLToken's
data is stored in just a String, two booleans and a Variant.

This change reduces sizeof(HTMLToken) from 68 to 32. Also, this reduces
raw tokenization time by around 20 to 50 percent, depending on the page.
Full document parsing time (with HTMLDocumentParser, on a local HTML
page without any dependency files) is reduced by between 4 and 20
percent, depending on the page.

Since tokenizing HTML pages can easily generated 50'000 tokens and more,
the storage has been designed in a way that avoids heap allocations
where possible, while trying to reduce the size of the tokens. The only
tokens which need to allocate on the heap are thus DOCTYPE tokens (max.
1 per document), and tag tokens (but only if they have attributes). This
way, only around 5 percent of all tokens generated need to allocate on
the heap (except for StringImpl allocations).

Userland/Libraries/LibWeb/HTML/Parser/HTMLToken.cpp

@@ -52,9 +52,15 @@ String HTMLToken::to_string() const
         builder.append("} }");
     }
 
-    if (type() == HTMLToken::Type::Comment || type() == HTMLToken::Type::Character) {
+    if (is_comment()) {
         builder.append(" { data: '");
-        builder.append(m_comment_or_character.data);
+        builder.append(comment());
+        builder.append("' }");
+    }
+
+    if (is_character()) {
+        builder.append(" { data: '");
+        builder.append_code_point(code_point());
         builder.append("' }");
     }
 

Userland/Libraries/LibWeb/HTML/Parser/HTMLToken.h

@@ -9,9 +9,10 @@
 
 #include <AK/FlyString.h>
 #include <AK/Function.h>
+#include <AK/OwnPtr.h>
 #include <AK/String.h>
 #include <AK/Types.h>
-#include <AK/Utf8View.h>
+#include <AK/Variant.h>
 #include <AK/Vector.h>
 
 namespace Web::HTML {
@@ -22,7 +23,7 @@ class HTMLToken {
     AK_MAKE_NONCOPYABLE(HTMLToken);
 
 public:
-    enum class Type {
+    enum class Type : u8 {
         Invalid,
         DOCTYPE,
         StartTag,
@@ -78,10 +79,24 @@ public:
     HTMLToken(Type type)
         : m_type(type)
     {
+        switch (m_type) {
+        case Type::Character:
+            m_data.set(0u);
+            break;
+        case Type::DOCTYPE:
+            m_data.set(OwnPtr<DoctypeData> {});
+            break;
+        case Type::StartTag:
+        case Type::EndTag:
+            m_data.set(OwnPtr<Vector<Attribute>>());
+            break;
+        default:
+            break;
+        }
     }
 
-    HTMLToken(HTMLToken&& other) = default;
+    HTMLToken(HTMLToken&&) = default;
-    HTMLToken& operator=(HTMLToken&& other) = default;
+    HTMLToken& operator=(HTMLToken&&) = default;
 
     bool is_doctype() const { return m_type == Type::DOCTYPE; }
     bool is_start_tag() const { return m_type == Type::StartTag; }
@@ -93,9 +108,7 @@ public:
     u32 code_point() const
     {
         VERIFY(is_character());
-        Utf8View view(m_comment_or_character.data);
+        return m_data.get<u32>();
-        VERIFY(view.length() == 1);
-        return *view.begin();
     }
 
     bool is_parser_whitespace() const
@@ -118,101 +131,105 @@ public:
     void set_code_point(u32 code_point)
     {
         VERIFY(is_character());
-        StringBuilder builder;
+        m_data.get<u32>() = code_point;
-        builder.append_code_point(code_point);
-        m_comment_or_character.data = builder.to_string();
     }
 
     String const& comment() const
     {
         VERIFY(is_comment());
-        return m_comment_or_character.data;
+        return m_string_data;
     }
 
     void set_comment(String comment)
     {
         VERIFY(is_comment());
-        m_comment_or_character.data = move(comment);
+        m_string_data = move(comment);
     }
 
     String const& tag_name() const
     {
         VERIFY(is_start_tag() || is_end_tag());
-        return m_tag.tag_name;
+        return m_string_data;
     }
 
     void set_tag_name(String name)
     {
         VERIFY(is_start_tag() || is_end_tag());
-        m_tag.tag_name = move(name);
+        m_string_data = move(name);
     }
 
     bool is_self_closing() const
     {
         VERIFY(is_start_tag() || is_end_tag());
-        return m_tag.self_closing;
+        return m_tag_self_closing;
     }
 
     void set_self_closing(bool self_closing)
     {
         VERIFY(is_start_tag() || is_end_tag());
-        m_tag.self_closing = self_closing;
+        m_tag_self_closing = self_closing;
     }
 
     bool has_acknowledged_self_closing_flag() const
     {
         VERIFY(is_self_closing());
-        return m_tag.self_closing_acknowledged;
+        return m_tag_self_closing_acknowledged;
     }
 
     void acknowledge_self_closing_flag_if_set()
     {
         if (is_self_closing())
-            m_tag.self_closing_acknowledged = true;
+            m_tag_self_closing_acknowledged = true;
     }
 
     bool has_attributes() const
     {
         VERIFY(is_start_tag() || is_end_tag());
-        return !m_tag.attributes.is_empty();
+        auto* ptr = tag_attributes();
+        return ptr && !ptr->is_empty();
     }
 
     size_t attribute_count() const
     {
         VERIFY(is_start_tag() || is_end_tag());
-        return m_tag.attributes.size();
+        if (auto* ptr = tag_attributes())
+            return ptr->size();
+        return 0;
     }
 
     void add_attribute(Attribute attribute)
     {
         VERIFY(is_start_tag() || is_end_tag());
-        m_tag.attributes.append(move(attribute));
+        ensure_tag_attributes().append(move(attribute));
     }
 
     Attribute const& last_attribute() const
     {
         VERIFY(is_start_tag() || is_end_tag());
-        VERIFY(!m_tag.attributes.is_empty());
+        VERIFY(has_attributes());
-        return m_tag.attributes.last();
+        return tag_attributes()->last();
     }
 
     Attribute& last_attribute()
     {
         VERIFY(is_start_tag() || is_end_tag());
-        VERIFY(!m_tag.attributes.is_empty());
+        VERIFY(has_attributes());
-        return m_tag.attributes.last();
+        return tag_attributes()->last();
     }
 
     void drop_attributes()
     {
         VERIFY(is_start_tag() || is_end_tag());
-        m_tag.attributes.clear();
+        m_data.get<OwnPtr<Vector<Attribute>>>().clear();
     }
 
     void for_each_attribute(Function<IterationDecision(Attribute const&)> callback) const
     {
         VERIFY(is_start_tag() || is_end_tag());
-        for (auto& attribute : m_tag.attributes) {
+        auto* ptr = tag_attributes();
+        if (!ptr)
+            return;
+        for (auto& attribute : *ptr) {
             if (callback(attribute) == IterationDecision::Break)
                 break;
         }
@@ -221,7 +238,10 @@ public:
     void for_each_attribute(Function<IterationDecision(Attribute&)> callback)
     {
         VERIFY(is_start_tag() || is_end_tag());
-        for (auto& attribute : m_tag.attributes) {
+        auto* ptr = tag_attributes();
+        if (!ptr)
+            return;
+        for (auto& attribute : *ptr) {
             if (callback(attribute) == IterationDecision::Break)
                 break;
         }
@@ -230,7 +250,11 @@ public:
     StringView attribute(FlyString const& attribute_name)
     {
         VERIFY(is_start_tag() || is_end_tag());
-        for (auto& attribute : m_tag.attributes) {
+
+        auto* ptr = tag_attributes();
+        if (!ptr)
+            return {};
+        for (auto& attribute : *ptr) {
             if (attribute_name == attribute.local_name)
                 return attribute.value;
         }
@@ -275,13 +299,18 @@ public:
     DoctypeData const& doctype_data() const
     {
         VERIFY(is_doctype());
-        return m_doctype;
+        auto* ptr = m_data.get<OwnPtr<DoctypeData>>().ptr();
+        VERIFY(ptr);
+        return *ptr;
     }
 
     DoctypeData& doctype_data()
     {
         VERIFY(is_doctype());
-        return m_doctype;
+        auto& ptr = m_data.get<OwnPtr<DoctypeData>>();
+        if (!ptr)
+            ptr = make<DoctypeData>();
+        return *ptr;
     }
 
     Type type() const { return m_type; }
@@ -295,25 +324,35 @@ public:
     void set_end_position(Badge<HTMLTokenizer>, Position end_position) { m_end_position = end_position; }
 
 private:
+    Vector<Attribute> const* tag_attributes() const
+    {
+        return m_data.get<OwnPtr<Vector<Attribute>>>().ptr();
+    }
+
+    Vector<Attribute>* tag_attributes()
+    {
+        return m_data.get<OwnPtr<Vector<Attribute>>>().ptr();
+    }
+
+    Vector<Attribute>& ensure_tag_attributes()
+    {
+        VERIFY(is_start_tag() || is_end_tag());
+        auto& ptr = m_data.get<OwnPtr<Vector<Attribute>>>();
+        if (!ptr)
+            ptr = make<Vector<Attribute>>();
+        return *ptr;
+    }
+
     Type m_type { Type::Invalid };
 
-    // Type::DOCTYPE
+    // Type::StartTag and Type::EndTag
-    DoctypeData m_doctype;
+    bool m_tag_self_closing { false };
+    bool m_tag_self_closing_acknowledged { false };
 
-    // Type::StartTag
+    // Type::Comment (comment data), Type::StartTag and Type::EndTag (tag name)
-    // Type::EndTag
+    String m_string_data;
-    struct {
-        String tag_name;
-        bool self_closing { false };
-        bool self_closing_acknowledged { false };
-        Vector<Attribute> attributes;
-    } m_tag;
 
-    // Type::Comment
+    Variant<Empty, u32, OwnPtr<DoctypeData>, OwnPtr<Vector<Attribute>>> m_data { Empty {} };
-    // Type::Character
-    struct {
-        String data;
-    } m_comment_or_character;
 
     Position m_start_position;
     Position m_end_position;