1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-27 02:47:35 +00:00

LibHTML: Start working on a very simplified HTML parser.

This commit is contained in:
Andreas Kling 2019-06-15 20:20:55 +02:00
parent a67e823838
commit 581d6b00c8
4 changed files with 139 additions and 22 deletions

View file

@ -1,32 +1,120 @@
#include <LibHTML/Element.h>
#include <LibHTML/Parser.h>
#include <LibHTML/Text.h>
#include <ctype.h>
static Retained<Element> create_element(const String& tag_name)
{
return adopt(*new Element(tag_name));
}
Retained<Document> parse(const String& html)
static bool is_self_closing_tag(const String& tag_name)
{
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;
return tag_name == "area"
|| tag_name == "base"
|| tag_name == "br"
|| tag_name == "col"
|| tag_name == "embed"
|| tag_name == "hr"
|| tag_name == "img"
|| tag_name == "input"
|| tag_name == "link"
|| tag_name == "meta"
|| tag_name == "param"
|| tag_name == "source"
|| tag_name == "track"
|| tag_name == "wbr";
}
Retained<Document> parse(const String& html)
{
Vector<Retained<ParentNode>> node_stack;
auto doc = adopt(*new Document);
node_stack.append(doc);
enum class State {
Free,
BeforeTagName,
InTagName,
InAttributeList,
InAttributeName,
InAttributeValueNoQuote,
InAttributeValueSingleQuote,
InAttributeValueDoubleQuote,
};
auto state = State::Free;
Vector<char, 256> buffer;
bool is_slash_tag = false;
auto move_to_state = [&](State new_state) {
if (new_state == State::BeforeTagName)
is_slash_tag = false;
if (state == State::Free && !buffer.is_empty()) {
auto text_node = adopt(*new Text(String::copy(buffer)));
node_stack.last()->append_child(text_node);
}
state = new_state;
buffer.clear();
};
auto close_tag = [&] {
if (node_stack.size() > 1)
node_stack.take_last();
};
auto open_tag = [&] {
auto new_element = create_element(String::copy(buffer));
node_stack.append(new_element);
if (node_stack.size() != 1)
node_stack[node_stack.size() - 2]->append_child(new_element);
if (is_self_closing_tag(new_element->tag_name()))
close_tag();
};
for (int i = 0; i < html.length(); ++i) {
char ch = html[i];
switch (state) {
case State::Free:
if (ch == '<') {
move_to_state(State::BeforeTagName);
break;
}
buffer.append(ch);
break;
case State::BeforeTagName:
if (ch == '/') {
is_slash_tag = true;
break;
}
if (ch == '>') {
move_to_state(State::Free);
break;
}
if (!isascii(ch))
break;
move_to_state(State::InTagName);
[[fallthrough]];
case State::InTagName:
if (ch == ' ') {
move_to_state(State::InAttributeList);
break;
}
if (ch == '>') {
if (is_slash_tag)
close_tag();
else
open_tag();
move_to_state(State::Free);
break;
}
buffer.append(ch);
break;
}
}
return doc;
}

View file

@ -1,10 +1,18 @@
#include <LibCore/CFile.h>
#include <LibHTML/Dump.h>
#include <LibHTML/Element.h>
#include <LibHTML/Parser.h>
#include <stdio.h>
int main()
int main(int argc, char** argv)
{
String html = "<html><head><title>my page</title></head><body><h1>Hi there</h1><p>Hello World!</p></body></html>";
CFile f(argc == 1 ? "/home/anon/small.html" : argv[1]);
if (!f.open(CIODevice::ReadOnly)) {
fprintf(stderr, "Error: %s\n", f.error_string());
return 1;
}
String html = String::copy(f.read_all());
auto doc = parse(html);
dump_tree(doc);
return 0;
}