srctree

Robin Linden parent 79a89eee c9a3b0fd
html: Keep the 'stack' of open elements as a vector

The spec has a very strange idea of what a stack is. We have to be ableto e.g. remove elements from the middle of it.

inline split

html/parser.cpp added: 41, removed: 39, total 2

@@ -141,15 +141,15 @@ void Parser::operator()(html2::StartTagToken const &start_tag) {

// comment, except if the first thing inside the body element is a meta,

// noscript, link, script, style, or template element.

if (doc_.html().children.size() == 1 && start_tag.tag_name != "body") {

auto &body = open_elements_.top()->children.emplace_back(dom::Element{.name{"body"}});

open_elements_.push(&std::get<dom::Element>(body));

auto &body = open_elements_.back()->children.emplace_back(dom::Element{.name{"body"}});

open_elements_.push_back(&std::get<dom::Element>(body));

}

generate_text_node_if_needed();

// https://html.spec.whatwg.org/multipage/grouping-content.html#the-p-element

if (open_elements_.top()->name == "p" && is_in_array<kAllowsParagraphEndTagOmission>(start_tag.tag_name)) {

open_elements_.pop();

if (open_elements_.back()->name == "p" && is_in_array<kAllowsParagraphEndTagOmission>(start_tag.tag_name)) {

open_elements_.pop_back();

}

// https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inbody

@@ -157,7 +157,7 @@ void Parser::operator()(html2::StartTagToken const &start_tag) {

tokenizer_.set_state(html2::State::Rawtext);

}

auto &new_element = open_elements_.top()->children.emplace_back(

auto &new_element = open_elements_.back()->children.emplace_back(

dom::Element{start_tag.tag_name, into_dom_attributes(start_tag.attributes), {}});

if (!start_tag.self_closing) {

@@ -165,13 +165,13 @@ void Parser::operator()(html2::StartTagToken const &start_tag) {

// if they need it, but we only ever add new children to the

// top-most element in the stack, so this pointer will be valid

// until it's been popped from the stack and we add its siblings.

open_elements_.push(std::get_if<dom::Element>(&new_element));

open_elements_.push_back(std::get_if<dom::Element>(&new_element));

}

// Special cases from https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inbody

// Immediately popped off the stack of open elements special cases.

if (!start_tag.self_closing && is_in_array<kImmediatelyPopped>(start_tag.tag_name)) {

open_elements_.pop();

open_elements_.pop_back();

}

@@ -182,9 +182,9 @@ void Parser::operator()(html2::EndTagToken const &end_tag) {

}

if (end_tag.tag_name == "html" && doc_.html().children.size() == 1) {

if (open_elements_.top()->name == "html") {

auto &body = open_elements_.top()->children.emplace_back(dom::Element{.name = "body"});

open_elements_.push(&std::get<dom::Element>(body));

if (open_elements_.back()->name == "html") {

auto &body = open_elements_.back()->children.emplace_back(dom::Element{.name = "body"});

open_elements_.push_back(&std::get<dom::Element>(body));

}

@@ -192,22 +192,22 @@ void Parser::operator()(html2::EndTagToken const &end_tag) {

// https://html.spec.whatwg.org/multipage/grouping-content.html#the-p-element

// TODO(robinlinden): or if the parent element is an autonomous custom element.

if (open_elements_.top()->name == "p" && end_tag.tag_name != "p"

if (open_elements_.back()->name == "p" && end_tag.tag_name != "p"

&& !is_in_array<kDisallowsParagraphEndTagOmissionWhenClosed>(end_tag.tag_name)) {

open_elements_.pop();

open_elements_.pop_back();

}

if (end_tag.tag_name == "html" && open_elements_.top()->name == "body") {

open_elements_.pop();

if (end_tag.tag_name == "html" && open_elements_.back()->name == "body") {

open_elements_.pop_back();

}

auto const &expected_tag = open_elements_.top()->name;

auto const &expected_tag = open_elements_.back()->name;

if (end_tag.tag_name != expected_tag) {

spdlog::warn("Unexpected end_tag name, expected [{}] but got [{}]", expected_tag, end_tag.tag_name);

return;

}

open_elements_.pop();

open_elements_.pop_back();

}

void Parser::operator()(html2::CommentToken const &) {

@@ -219,22 +219,23 @@ void Parser::operator()(html2::CharacterToken const &character) {

}

void Parser::operator()(html2::EndOfFileToken const &) {

if (!open_elements_.empty() && open_elements_.top()->name == "html" && open_elements_.top()->children.size() == 1) {

if (!open_elements_.empty() && open_elements_.back()->name == "html"

&& open_elements_.back()->children.size() == 1) {

auto &body = doc_.html().children.emplace_back(dom::Element{.name = "body"});

open_elements_.push(&std::get<dom::Element>(body));

open_elements_.push_back(&std::get<dom::Element>(body));

}

if (!open_elements_.empty()) {

generate_text_node_if_needed();

}

if (!open_elements_.empty() && open_elements_.top()->name == "body") {

open_elements_.pop();

if (!open_elements_.empty() && open_elements_.back()->name == "body") {

open_elements_.pop_back();

}

// https://html.spec.whatwg.org/multipage/semantics.html#the-html-element

if (!open_elements_.empty() && open_elements_.top()->name == "html") {

open_elements_.pop();

if (!open_elements_.empty() && open_elements_.back()->name == "html") {

open_elements_.pop_back();

}

if (!open_elements_.empty()) {

@@ -250,7 +251,7 @@ void Parser::generate_text_node_if_needed() {

return;

}

open_elements_.top()->children.emplace_back(dom::Text{std::move(text)});

open_elements_.back()->children.emplace_back(dom::Text{std::move(text)});

}

} // namespace html

html/parser.h added: 41, removed: 39, total 2

@@ -13,9 +13,9 @@

#include <functional>

#include <sstream>

#include <stack>

#include <string_view>

#include <utility>

#include <vector>

namespace html {

@@ -53,7 +53,7 @@ private:

html2::Tokenizer tokenizer_;

dom::Document doc_{};

std::stack<dom::Element *> open_elements_{};

std::vector<dom::Element *> open_elements_{};

std::stringstream current_text_{};

bool scripting_{false};

html2::InsertionMode insertion_mode_{};

html/parser_actions.h added: 41, removed: 39, total 2

@@ -13,7 +13,6 @@

#include <algorithm>

#include <cassert>

#include <span>

#include <stack>

#include <utility>

#include <variant>

#include <vector>

@@ -26,7 +25,7 @@ public:

html2::Tokenizer &tokenizer,

bool scripting,

html2::InsertionMode &current_insertion_mode,

std::stack<dom::Element *> &open_elements)

std::vector<dom::Element *> &open_elements)

: document_{document}, tokenizer_{tokenizer}, scripting_{scripting},

current_insertion_mode_{current_insertion_mode}, open_elements_{open_elements} {}

@@ -61,8 +60,8 @@ public:

insert({token.tag_name, into_dom_attributes(token.attributes)});

}

void pop_current_node() override { open_elements_.pop(); }

std::string_view current_node_name() const override { return open_elements_.top()->name; }

void pop_current_node() override { open_elements_.pop_back(); }

std::string_view current_node_name() const override { return open_elements_.back()->name; }

void merge_into_html_node(std::span<html2::Attribute const> attrs) override {

auto &html = document_.html();

@@ -76,7 +75,7 @@ public:

}

void insert_character(html2::CharacterToken const &character) override {

auto &current_element = open_elements_.top();

auto &current_element = open_elements_.back();

if (current_element->children.empty() || !std::holds_alternative<dom::Text>(current_element->children.back())) {

current_element->children.emplace_back(dom::Text{});

}

@@ -104,12 +103,12 @@ private:

assert(open_elements_.empty());

document_.html().name = std::move(element.name);

document_.html().attributes = std::move(element.attributes);

open_elements_.push(&document_.html());

open_elements_.push_back(&document_.html());

return;

}

dom::Node &node = open_elements_.top()->children.emplace_back(std::move(element));

open_elements_.push(&std::get<dom::Element>(node));

dom::Node &node = open_elements_.back()->children.emplace_back(std::move(element));

open_elements_.push_back(&std::get<dom::Element>(node));

}

dom::Document &document_;

@@ -117,7 +116,7 @@ private:

bool scripting_;

html2::InsertionMode original_insertion_mode_;

html2::InsertionMode &current_insertion_mode_;

std::stack<dom::Element *> &open_elements_;

std::vector<dom::Element *> &open_elements_;

};

} // namespace html

html2/parser_states_test.cpp added: 41, removed: 39, total 2

@@ -9,6 +9,8 @@

#include "etest/etest.h"

#include "html/parser_actions.h"

#include <vector>

using etest::expect_eq;

using NodeVec = std::vector<dom::Node>;

@@ -30,7 +32,7 @@ ParseResult parse(std::string_view html, ParseOptions opts) {

ParseResult res{};

html2::InsertionMode mode{opts.initial_insertion_mode};

std::stack<dom::Element *> open_elements{};

std::vector<dom::Element *> open_elements{};

html::Actions actions{res.document, tokenizer, opts.scripting, mode, open_elements};

auto on_token = [&](html2::Tokenizer &, html2::Token const &token) {