From a67e823838943b31fb7cea68bd592093e197cf16 Mon Sep 17 00:00:00 2001
From: Andreas Kling <awesomekling@gmail.com>
Date: Sat, 15 Jun 2019 18:55:47 +0200
Subject: [PATCH] LibHTML: Start working on a simple HTML library.

I'd like to have rich text, and we might as well use HTML for that. :^)
---
 Kernel/build-root-filesystem.sh |  1 +
 Kernel/makeall.sh               |  1 +
 LibHTML/.gitignore              |  4 ++++
 LibHTML/Document.cpp            | 11 +++++++++
 LibHTML/Document.h              | 13 ++++++++++
 LibHTML/Dump.cpp                | 26 ++++++++++++++++++++
 LibHTML/Dump.h                  |  5 ++++
 LibHTML/Element.cpp             | 12 ++++++++++
 LibHTML/Element.h               | 30 +++++++++++++++++++++++
 LibHTML/Makefile                | 42 +++++++++++++++++++++++++++++++++
 LibHTML/Node.cpp                | 24 +++++++++++++++++++
 LibHTML/Node.h                  | 41 ++++++++++++++++++++++++++++++++
 LibHTML/ParentNode.cpp          | 10 ++++++++
 LibHTML/ParentNode.h            | 32 +++++++++++++++++++++++++
 LibHTML/Parser.cpp              | 32 +++++++++++++++++++++++++
 LibHTML/Parser.h                |  7 ++++++
 LibHTML/Text.cpp                | 13 ++++++++++
 LibHTML/Text.h                  | 15 ++++++++++++
 LibHTML/test.cpp                | 10 ++++++++
 19 files changed, 329 insertions(+)
 create mode 100644 LibHTML/.gitignore
 create mode 100644 LibHTML/Document.cpp
 create mode 100644 LibHTML/Document.h
 create mode 100644 LibHTML/Dump.cpp
 create mode 100644 LibHTML/Dump.h
 create mode 100644 LibHTML/Element.cpp
 create mode 100644 LibHTML/Element.h
 create mode 100644 LibHTML/Makefile
 create mode 100644 LibHTML/Node.cpp
 create mode 100644 LibHTML/Node.h
 create mode 100644 LibHTML/ParentNode.cpp
 create mode 100644 LibHTML/ParentNode.h
 create mode 100644 LibHTML/Parser.cpp
 create mode 100644 LibHTML/Parser.h
 create mode 100644 LibHTML/Text.cpp
 create mode 100644 LibHTML/Text.h
 create mode 100644 LibHTML/test.cpp

diff --git a/Kernel/build-root-filesystem.sh b/Kernel/build-root-filesystem.sh
index 8ffa9fe64274ae..8850ed8b65031d 100755
--- a/Kernel/build-root-filesystem.sh
+++ b/Kernel/build-root-filesystem.sh
@@ -83,6 +83,7 @@ cp ../Servers/LookupServer/LookupServer mnt/bin/LookupServer
 cp ../Servers/SystemServer/SystemServer mnt/bin/SystemServer
 cp ../Servers/WindowServer/WindowServer mnt/bin/WindowServer
 cp ../Shell/Shell mnt/bin/Shell
+cp ../LibHTML/tho mnt/bin/tho
 echo "done"
 
 echo -n "installing shortcuts... "
diff --git a/Kernel/makeall.sh b/Kernel/makeall.sh
index 9b04f3339f2c5b..12d5edfb7bbf1e 100755
--- a/Kernel/makeall.sh
+++ b/Kernel/makeall.sh
@@ -17,6 +17,7 @@ build_targets="$build_targets ../Servers/SystemServer"
 build_targets="$build_targets ../Servers/LookupServer"
 build_targets="$build_targets ../Servers/WindowServer"
 build_targets="$build_targets ../LibGUI"
+build_targets="$build_targets ../LibHTML"
 build_targets="$build_targets ../Userland"
 build_targets="$build_targets ../Applications/Terminal"
 build_targets="$build_targets ../Applications/FontEditor"
diff --git a/LibHTML/.gitignore b/LibHTML/.gitignore
new file mode 100644
index 00000000000000..e6e86bf61b5360
--- /dev/null
+++ b/LibHTML/.gitignore
@@ -0,0 +1,4 @@
+*.o
+*.d
+libhtml.a
+tho
diff --git a/LibHTML/Document.cpp b/LibHTML/Document.cpp
new file mode 100644
index 00000000000000..a7ba17c2b34d8c
--- /dev/null
+++ b/LibHTML/Document.cpp
@@ -0,0 +1,11 @@
+#include <LibHTML/Document.h>
+
+Document::Document()
+    : ParentNode(NodeType::DOCUMENT_NODE)
+{
+}
+
+Document::~Document()
+{
+}
+
diff --git a/LibHTML/Document.h b/LibHTML/Document.h
new file mode 100644
index 00000000000000..cf31d766b9642e
--- /dev/null
+++ b/LibHTML/Document.h
@@ -0,0 +1,13 @@
+#pragma once
+
+#include <AK/AKString.h>
+#include <LibHTML/ParentNode.h>
+
+class Document : public ParentNode {
+public:
+    Document();
+    virtual ~Document() override;
+
+private:
+};
+
diff --git a/LibHTML/Dump.cpp b/LibHTML/Dump.cpp
new file mode 100644
index 00000000000000..8826eb444439d2
--- /dev/null
+++ b/LibHTML/Dump.cpp
@@ -0,0 +1,26 @@
+#include <LibHTML/Document.h>
+#include <LibHTML/Dump.h>
+#include <LibHTML/Element.h>
+#include <LibHTML/Text.h>
+#include <stdio.h>
+
+void dump_tree(Node& node)
+{
+    static int indent = 0;
+    for (int i = 0; i < indent; ++i)
+        printf("   ");
+    if (node.is_document()) {
+        printf("*Document*\n");
+    } else if (node.is_element()) {
+        printf("<%s>\n", static_cast<Element&>(node).tag_name().characters());
+    } else if (node.is_text()) {
+        printf("\"%s\"\n", static_cast<Text&>(node).data().characters());
+    }
+    ++indent;
+    if (node.is_parent_node()) {
+        static_cast<ParentNode&>(node).for_each_child([](Node& child) {
+            dump_tree(child);
+        });
+    }
+    --indent;
+}
diff --git a/LibHTML/Dump.h b/LibHTML/Dump.h
new file mode 100644
index 00000000000000..09b7de697e4864
--- /dev/null
+++ b/LibHTML/Dump.h
@@ -0,0 +1,5 @@
+#pragma once
+
+class Node;
+
+void dump_tree(Node&);
diff --git a/LibHTML/Element.cpp b/LibHTML/Element.cpp
new file mode 100644
index 00000000000000..5c284b6f741d57
--- /dev/null
+++ b/LibHTML/Element.cpp
@@ -0,0 +1,12 @@
+#include <LibHTML/Element.h>
+
+Element::Element(const String& tag_name)
+    : ParentNode(NodeType::ELEMENT_NODE)
+    , m_tag_name(tag_name)
+{
+}
+
+Element::~Element()
+{
+}
+
diff --git a/LibHTML/Element.h b/LibHTML/Element.h
new file mode 100644
index 00000000000000..5b9f390603f3c6
--- /dev/null
+++ b/LibHTML/Element.h
@@ -0,0 +1,30 @@
+#pragma once
+
+#include <LibHTML/ParentNode.h>
+#include <AK/AKString.h>
+
+class Attribute {
+public:
+    Attribute(const String& name, const String& value)
+        : m_name(name)
+        , m_value(value)
+    {
+    }
+
+private:
+    String m_name;
+    String m_value;
+};
+
+class Element : public ParentNode {
+public:
+    explicit Element(const String& tag_name);
+    virtual ~Element() override;
+
+    const String& tag_name() const { return m_tag_name; }
+
+private:
+    String m_tag_name;
+    Vector<Attribute> m_attributes;
+};
+
diff --git a/LibHTML/Makefile b/LibHTML/Makefile
new file mode 100644
index 00000000000000..ca947be04fafa7
--- /dev/null
+++ b/LibHTML/Makefile
@@ -0,0 +1,42 @@
+include ../Makefile.common
+
+LIBHTML_OBJS = \
+    Node.o \
+    ParentNode.o \
+    Element.o \
+    Document.o \
+    Text.o \
+    Parser.o \
+    Dump.o
+
+TEST_OBJS = test.o
+TEST_PROGRAM = tho
+
+OBJS = $(LIBHTML_OBJS) $(TEST_OBJS)
+
+LIBRARY = libhtml.a
+DEFINES += -DUSERLAND
+
+all: $(LIBRARY) $(TEST_PROGRAM)
+
+$(TEST_PROGRAM): $(TEST_OBJS) $(LIBRARY)
+	$(LD) -o $@ $(LDFLAGS) -L. $(TEST_OBJS) -lhtml -lgui -lcore -lc
+
+$(LIBRARY): $(LIBHTML_OBJS)
+	@echo "LIB $@"; $(AR) rcs $@ $(LIBHTML_OBJS)
+
+.cpp.o:
+	@echo "CXX $<"; $(CXX) $(CXXFLAGS) -o $@ -c $<
+
+-include $(OBJS:%.o=%.d)
+
+clean:
+	@echo "CLEAN"; rm -f $(TEST_PROGRAM) $(LIBRARY) $(OBJS) *.d
+
+install: $(LIBRARY)
+	mkdir -p ../Root/usr/include/LibHTML
+	# Copy headers
+	rsync -r -a --include '*/' --include '*.h' --exclude '*' . ../Root/usr/include/LibHTML
+	# Install the library
+	cp $(LIBRARY) ../Root/usr/lib
+
diff --git a/LibHTML/Node.cpp b/LibHTML/Node.cpp
new file mode 100644
index 00000000000000..961699d55402a5
--- /dev/null
+++ b/LibHTML/Node.cpp
@@ -0,0 +1,24 @@
+#include <AK/Retained.h>
+#include <LibHTML/Node.h>
+
+Node::Node(NodeType type)
+    : m_type(type)
+{
+}
+
+Node::~Node()
+{
+}
+
+void Node::retain()
+{
+    ASSERT(m_retain_count);
+    ++m_retain_count;
+}
+
+void Node::release()
+{
+    ASSERT(m_retain_count);
+    if (!--m_retain_count)
+        delete this;
+}
diff --git a/LibHTML/Node.h b/LibHTML/Node.h
new file mode 100644
index 00000000000000..3650782929f493
--- /dev/null
+++ b/LibHTML/Node.h
@@ -0,0 +1,41 @@
+#pragma once
+
+#include <AK/Retained.h>
+#include <AK/Vector.h>
+
+enum class NodeType : unsigned {
+    INVALID = 0,
+    ELEMENT_NODE = 1,
+    TEXT_NODE = 3,
+    DOCUMENT_NODE = 9,
+};
+
+class Node {
+public:
+    virtual ~Node();
+
+    void retain();
+    void release();
+    int retain_count() const { return m_retain_count; }
+
+    NodeType type() const { return m_type; }
+    bool is_element() const { return type() == NodeType::ELEMENT_NODE; }
+    bool is_text() const { return type() == NodeType::TEXT_NODE; }
+    bool is_document() const { return type() == NodeType::DOCUMENT_NODE; }
+    bool is_parent_node() const { return is_element() || is_document(); }
+
+    Node* next_sibling() { return m_next_sibling; }
+    Node* previous_sibling() { return m_previous_sibling; }
+    void set_next_sibling(Node* node) { m_next_sibling = node; }
+    void set_previous_sibling(Node* node) { m_previous_sibling = node; }
+
+protected:
+    explicit Node(NodeType);
+
+    int m_retain_count { 1 };
+    NodeType m_type { NodeType::INVALID };
+    Vector<Node*> m_children;
+    Node* m_next_sibling { nullptr };
+    Node* m_previous_sibling { nullptr };
+};
+
diff --git a/LibHTML/ParentNode.cpp b/LibHTML/ParentNode.cpp
new file mode 100644
index 00000000000000..d2410323759749
--- /dev/null
+++ b/LibHTML/ParentNode.cpp
@@ -0,0 +1,10 @@
+#include <LibHTML/ParentNode.h>
+
+void ParentNode::append_child(Retained<Node> node)
+{
+    if (m_last_child)
+        m_last_child->set_next_sibling(node.ptr());
+    m_last_child = &node.leak_ref();
+    if (!m_first_child)
+        m_first_child = m_last_child;
+}
diff --git a/LibHTML/ParentNode.h b/LibHTML/ParentNode.h
new file mode 100644
index 00000000000000..357315e3d789ba
--- /dev/null
+++ b/LibHTML/ParentNode.h
@@ -0,0 +1,32 @@
+#pragma once
+
+#include <LibHTML/Node.h>
+
+class ParentNode : public Node {
+public:
+    void append_child(Retained<Node>);
+
+    Node* first_child() { return m_first_child; }
+    Node* last_child() { return m_last_child; }
+
+    template<typename F> void for_each_child(F);
+
+protected:
+    explicit ParentNode(NodeType type)
+        : Node(type)
+    {
+    }
+
+private:
+    Node* m_first_child { nullptr };
+    Node* m_last_child { nullptr };
+};
+
+template<typename F>
+inline void ParentNode::for_each_child(F func)
+{
+    for (auto* node = first_child(); node; node = node->next_sibling()) {
+        func(*node);
+    }
+}
+
diff --git a/LibHTML/Parser.cpp b/LibHTML/Parser.cpp
new file mode 100644
index 00000000000000..3bc6f2a99a9db3
--- /dev/null
+++ b/LibHTML/Parser.cpp
@@ -0,0 +1,32 @@
+#include <LibHTML/Element.h>
+#include <LibHTML/Parser.h>
+#include <LibHTML/Text.h>
+
+static Retained<Element> create_element(const String& tag_name)
+{
+    return adopt(*new Element(tag_name));
+}
+
+Retained<Document> parse(const String& html)
+{
+    auto doc = adopt(*new Document);
+
+    auto head = create_element("head");
+    auto title = create_element("title");
+    auto title_text = adopt(*new Text("Page Title"));
+    title->append_child(title_text);
+    head->append_child(title);
+
+    doc->append_child(head);
+
+    auto body = create_element("body");
+    auto h1 = create_element("h1");
+    auto h1_text = adopt(*new Text("Hello World!"));
+
+    h1->append_child(h1_text);
+    body->append_child(h1);
+    doc->append_child(body);
+
+    return doc;
+}
+
diff --git a/LibHTML/Parser.h b/LibHTML/Parser.h
new file mode 100644
index 00000000000000..db593dae2d5f99
--- /dev/null
+++ b/LibHTML/Parser.h
@@ -0,0 +1,7 @@
+#pragma once
+
+#include <AK/Retained.h>
+#include <LibHTML/Document.h>
+
+Retained<Document> parse(const String& html);
+
diff --git a/LibHTML/Text.cpp b/LibHTML/Text.cpp
new file mode 100644
index 00000000000000..67ec381f73fdd9
--- /dev/null
+++ b/LibHTML/Text.cpp
@@ -0,0 +1,13 @@
+#include <LibHTML/Text.h>
+
+Text::Text(const String& data)
+    : Node(NodeType::TEXT_NODE)
+    , m_data(data)
+{
+}
+
+Text::~Text()
+{
+}
+
+
diff --git a/LibHTML/Text.h b/LibHTML/Text.h
new file mode 100644
index 00000000000000..bdc9f86ec2d47e
--- /dev/null
+++ b/LibHTML/Text.h
@@ -0,0 +1,15 @@
+#pragma once
+
+#include <AK/AKString.h>
+#include <LibHTML/Node.h>
+
+class Text final : public Node {
+public:
+    explicit Text(const String&);
+    virtual ~Text() override;
+
+    const String& data() const { return m_data; }
+
+private:
+    String m_data;
+};
diff --git a/LibHTML/test.cpp b/LibHTML/test.cpp
new file mode 100644
index 00000000000000..9d774335c29084
--- /dev/null
+++ b/LibHTML/test.cpp
@@ -0,0 +1,10 @@
+#include <LibHTML/Dump.h>
+#include <LibHTML/Element.h>
+#include <LibHTML/Parser.h>
+
+int main()
+{
+    String html = "<html><head><title>my page</title></head><body><h1>Hi there</h1><p>Hello World!</p></body></html>";
+    auto doc = parse(html);
+    dump_tree(doc);
+}