RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-05-31 06:48:12 +00:00
Code Issues Activity Actions

LibHTML: Add basic <!DOCTYPE> parsing and a DocumentType class

Browse source 
Plus, Document::fixup() will now make sure that the document always
starts with a doctype node, followed by an <html> element.
This commit is contained in:
Andreas Kling 2019-10-09 20:17:01 +02:00
parent 850955053f
commit fc53867937
9 changed files with 84 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

10
Libraries/LibHTML/DOM/Document.cpp
View file

@ -2,6 +2,7 @@
#include <AK/StringBuilder.h> #include <AK/StringBuilder.h>
#include <LibHTML/CSS/StyleResolver.h> #include <LibHTML/CSS/StyleResolver.h>
#include <LibHTML/DOM/Document.h> #include <LibHTML/DOM/Document.h>
#include <LibHTML/DOM/DocumentType.h>
#include <LibHTML/DOM/Element.h> #include <LibHTML/DOM/Element.h>
#include <LibHTML/DOM/HTMLBodyElement.h> #include <LibHTML/DOM/HTMLBodyElement.h>
#include <LibHTML/DOM/HTMLHeadElement.h> #include <LibHTML/DOM/HTMLHeadElement.h>
@ -29,11 +30,14 @@ StyleResolver& Document::style_resolver()
void Document::fixup() void Document::fixup()
{ {
if (is<HTMLHtmlElement>(first_child())) if (!is<DocumentType>(first_child()))
prepend_child(adopt(*new DocumentType(*this)));
if (is<HTMLHtmlElement>(first_child()->next_sibling()))
return; return;
NonnullRefPtr<Element> body = adopt(*new Element(*this, "body")); auto body = adopt(*new HTMLBodyElement(*this, "body"));
NonnullRefPtr<Element> html = adopt(*new Element(*this, "html")); auto html = adopt(*new HTMLHtmlElement(*this, "html"));
html->append_child(body); html->append_child(body);
this->donate_all_children_to(body); this->donate_all_children_to(body);
this->append_child(html); this->append_child(html);

10
Libraries/LibHTML/DOM/DocumentType.cpp Normal file
View file

@ -0,0 +1,10 @@
#include <LibHTML/DOM/DocumentType.h>
DocumentType::DocumentType(Document& document)
: Node(document, NodeType::DOCUMENT_TYPE_NODE)
{
}
DocumentType::~DocumentType()
{
}

17
Libraries/LibHTML/DOM/DocumentType.h Normal file
View file

@ -0,0 +1,17 @@
#pragma once
#include <LibHTML/DOM/Node.h>
class DocumentType final : public Node {
public:
explicit DocumentType(Document&);
virtual ~DocumentType() override;
virtual String tag_name() const override { return "!DOCTYPE"; }
};
template<>
inline bool is<DocumentType>(const Node& node)
{
return node.type() == NodeType::DOCUMENT_TYPE_NODE;
}

5
Libraries/LibHTML/DOM/Node.cpp
View file

@ -97,3 +97,8 @@ const Element* Node::previous_element_sibling() const
} }
return nullptr; return nullptr;
} }
RefPtr<LayoutNode> Node::create_layout_node(const StyleResolver&, const StyleProperties*) const
{
return nullptr;
}

4
Libraries/LibHTML/DOM/Node.h
View file

@ -11,6 +11,7 @@ enum class NodeType : unsigned {
ELEMENT_NODE = 1, ELEMENT_NODE = 1,
TEXT_NODE = 3, TEXT_NODE = 3,
DOCUMENT_NODE = 9, DOCUMENT_NODE = 9,
DOCUMENT_TYPE_NODE = 10,
}; };
class Document; class Document;
@ -30,9 +31,10 @@ public:
bool is_element() const { return type() == NodeType::ELEMENT_NODE; } bool is_element() const { return type() == NodeType::ELEMENT_NODE; }
bool is_text() const { return type() == NodeType::TEXT_NODE; } bool is_text() const { return type() == NodeType::TEXT_NODE; }
bool is_document() const { return type() == NodeType::DOCUMENT_NODE; } bool is_document() const { return type() == NodeType::DOCUMENT_NODE; }
bool is_document_type() const { return type() == NodeType::DOCUMENT_TYPE_NODE; }
bool is_parent_node() const { return is_element() || is_document(); } bool is_parent_node() const { return is_element() || is_document(); }
virtual RefPtr<LayoutNode> create_layout_node(const StyleResolver&, const StyleProperties* parent_style) const = 0; virtual RefPtr<LayoutNode> create_layout_node(const StyleResolver&, const StyleProperties* parent_style) const;
RefPtr<LayoutNode> create_layout_tree(const StyleResolver&, const StyleProperties* parent_style) const; RefPtr<LayoutNode> create_layout_tree(const StyleResolver&, const StyleProperties* parent_style) const;
virtual String tag_name() const = 0; virtual String tag_name() const = 0;

3
Libraries/LibHTML/Dump.cpp
View file

@ -1,6 +1,7 @@
#include <AK/Utf8View.h> #include <AK/Utf8View.h>
#include <LibHTML/CSS/StyleSheet.h> #include <LibHTML/CSS/StyleSheet.h>
#include <LibHTML/DOM/Document.h> #include <LibHTML/DOM/Document.h>
#include <LibHTML/DOM/DocumentType.h>
#include <LibHTML/DOM/Element.h> #include <LibHTML/DOM/Element.h>
#include <LibHTML/DOM/Text.h> #include <LibHTML/DOM/Text.h>
#include <LibHTML/Dump.h> #include <LibHTML/Dump.h>
@ -24,6 +25,8 @@ void dump_tree(const Node& node)
dbgprintf(">\n"); dbgprintf(">\n");
} else if (is<Text>(node)) { } else if (is<Text>(node)) {
dbgprintf("\"%s\"\n", static_cast<const Text&>(node).data().characters()); dbgprintf("\"%s\"\n", static_cast<const Text&>(node).data().characters());
} else if (is<DocumentType>(node)) {
dbgprintf("<!DOCTYPE>\n");
} }
++indent; ++indent;
if (is<ParentNode>(node)) { if (is<ParentNode>(node)) {

1
Libraries/LibHTML/Makefile.shared
View file

@ -16,6 +16,7 @@ LIBHTML_OBJS = \
DOM/HTMLLinkElement.o \ DOM/HTMLLinkElement.o \
DOM/Document.o \ DOM/Document.o \
DOM/Text.o \ DOM/Text.o \
DOM/DocumentType.o \
CSS/Selector.o \ CSS/Selector.o \
CSS/StyleSheet.o \ CSS/StyleSheet.o \
CSS/StyleRule.o \ CSS/StyleRule.o \

23
Libraries/LibHTML/Parser/HTMLParser.cpp
View file

@ -1,6 +1,7 @@
#include <AK/Function.h> #include <AK/Function.h>
#include <AK/NonnullRefPtrVector.h> #include <AK/NonnullRefPtrVector.h>
#include <AK/StringBuilder.h> #include <AK/StringBuilder.h>
#include <LibHTML/DOM/DocumentType.h>
#include <LibHTML/DOM/Element.h> #include <LibHTML/DOM/Element.h>
#include <LibHTML/DOM/HTMLAnchorElement.h> #include <LibHTML/DOM/HTMLAnchorElement.h>
#include <LibHTML/DOM/HTMLBodyElement.h> #include <LibHTML/DOM/HTMLBodyElement.h>
@ -10,9 +11,9 @@
#include <LibHTML/DOM/HTMLHeadingElement.h> #include <LibHTML/DOM/HTMLHeadingElement.h>
#include <LibHTML/DOM/HTMLHtmlElement.h> #include <LibHTML/DOM/HTMLHtmlElement.h>
#include <LibHTML/DOM/HTMLImageElement.h> #include <LibHTML/DOM/HTMLImageElement.h>
#include <LibHTML/DOM/HTMLLinkElement.h>
#include <LibHTML/DOM/HTMLStyleElement.h> #include <LibHTML/DOM/HTMLStyleElement.h>
#include <LibHTML/DOM/HTMLTitleElement.h> #include <LibHTML/DOM/HTMLTitleElement.h>
#include <LibHTML/DOM/HTMLLinkElement.h>
#include <LibHTML/DOM/Text.h> #include <LibHTML/DOM/Text.h>
#include <LibHTML/Parser/HTMLParser.h> #include <LibHTML/Parser/HTMLParser.h>
#include <ctype.h> #include <ctype.h>
@ -106,10 +107,12 @@ NonnullRefPtr<Document> parse_html(const StringView& html, const URL& url)
Vector<char, 256> attribute_value_buffer; Vector<char, 256> attribute_value_buffer;
bool is_slash_tag = false; bool is_slash_tag = false;
bool is_exclamation_tag = false;
auto move_to_state = [&](State new_state) { auto move_to_state = [&](State new_state) {
if (new_state == State::BeforeTagName) { if (new_state == State::BeforeTagName) {
is_slash_tag = false; is_slash_tag = false;
is_exclamation_tag = false;
tag_name_buffer.clear(); tag_name_buffer.clear();
attributes.clear(); attributes.clear();
} }
@ -142,8 +145,19 @@ NonnullRefPtr<Document> parse_html(const StringView& html, const URL& url)
close_tag(); close_tag();
}; };
auto handle_exclamation_tag = [&] {
auto name = String::copy(tag_name_buffer);
tag_name_buffer.clear();
ASSERT(name == "DOCTYPE");
if (node_stack.size() != 1)
node_stack[node_stack.size() - 2].append_child(adopt(*new DocumentType(document)), false);
close_tag();
};
auto commit_tag = [&] { auto commit_tag = [&] {
if (is_slash_tag) if (is_exclamation_tag)
handle_exclamation_tag();
else if (is_slash_tag)
close_tag(); close_tag();
else else
open_tag(); open_tag();
@ -159,6 +173,7 @@ NonnullRefPtr<Document> parse_html(const StringView& html, const URL& url)
case State::Free: case State::Free:
if (ch == '<') { if (ch == '<') {
is_slash_tag = false; is_slash_tag = false;
is_exclamation_tag = false;
move_to_state(State::BeforeTagName); move_to_state(State::BeforeTagName);
break; break;
} }
@ -193,6 +208,10 @@ NonnullRefPtr<Document> parse_html(const StringView& html, const URL& url)
is_slash_tag = true; is_slash_tag = true;
break; break;
} }
if (ch == '!') {
is_exclamation_tag = true;
break;
}
if (ch == '>') { if (ch == '>') {
move_to_state(State::Free); move_to_state(State::Free);
break; break;

17
Libraries/LibHTML/TreeNode.h
View file

@ -33,6 +33,7 @@ public:
const T* first_child() const { return m_first_child; } const T* first_child() const { return m_first_child; }
const T* last_child() const { return m_last_child; } const T* last_child() const { return m_last_child; }
void prepend_child(NonnullRefPtr<T> node, bool call_inserted_into = true);
void append_child(NonnullRefPtr<T> node, bool call_inserted_into = true); void append_child(NonnullRefPtr<T> node, bool call_inserted_into = true);
void donate_all_children_to(T& node); void donate_all_children_to(T& node);
@ -64,6 +65,22 @@ inline void TreeNode<T>::append_child(NonnullRefPtr<T> node, bool call_inserted_
(void)node.leak_ref(); (void)node.leak_ref();
} }
template<typename T>
inline void TreeNode<T>::prepend_child(NonnullRefPtr<T> node, bool call_inserted_into)
{
ASSERT(!node->m_parent);
if (m_first_child)
m_first_child->m_previous_sibling = node.ptr();
node->m_next_sibling = m_first_child;
node->m_parent = static_cast<T*>(this);
m_first_child = node.ptr();
if (!m_last_child)
m_last_child = m_first_child;
if (call_inserted_into)
node->inserted_into(static_cast<T&>(*this));
(void)node.leak_ref();
}
template<typename T> template<typename T>
inline void TreeNode<T>::donate_all_children_to(T& node) inline void TreeNode<T>::donate_all_children_to(T& node)
{ {