srctree

Robin Linden parent c0462913 8467d647
html+html2: Move spec-compliant parsing bits from //html to //html2

inline split

html/parser.cpp added: 57, removed: 44, total 13

@@ -97,7 +97,7 @@ constexpr std::array kDisallowsParagraphEndTagOmissionWhenClosed{

void Parser::on_token(html2::Tokenizer &, html2::Token &&token) {

// Everything in <head> and earlier is handled by the new parser.

if (!std::holds_alternative<AfterHead>(insertion_mode_)) {

if (!std::holds_alternative<html2::AfterHead>(insertion_mode_)) {

insertion_mode_ = std::visit([&](auto &mode) { return mode.process(actions_, token); }, insertion_mode_)

.value_or(insertion_mode_);

if (auto const *end = std::get_if<html2::EndTagToken>(&token); end != nullptr && end->tag_name == "head") {

@@ -105,7 +105,7 @@ void Parser::on_token(html2::Tokenizer &, html2::Token &&token) {

}

if (std::holds_alternative<AfterHead>(insertion_mode_)) {

if (std::holds_alternative<html2::AfterHead>(insertion_mode_)) {

std::visit(*this, token);

}

html/parser.h added: 57, removed: 44, total 13

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

#define HTML_PARSER_H_

#include "html/parser_actions.h"

#include "html/parser_states.h"

#include "dom/dom.h"

#include "html2/parser_states.h"

#include "html2/tokenizer.h"

#include <functional>

@@ -56,7 +56,7 @@ private:

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

std::stringstream current_text_{};

bool scripting_{false};

InsertionMode insertion_mode_{};

html2::InsertionMode insertion_mode_{};

Actions actions_{doc_, tokenizer_, scripting_, open_elements_};

};

html/parser_actions.h added: 57, removed: 44, total 13

@@ -5,10 +5,9 @@

#ifndef HTML_PARSER_ACTIONS_H_

#define HTML_PARSER_ACTIONS_H_

#include "html/iparser_actions.h"

#include "html/parser_states.h"

#include "dom/dom.h"

#include "html2/iparser_actions.h"

#include "html2/parser_states.h"

#include "html2/tokenizer.h"

#include <algorithm>

@@ -21,7 +20,7 @@

namespace html {

class Actions : public IActions {

class Actions : public html2::IActions {

public:

Actions(dom::Document &document,

html2::Tokenizer &tokenizer,

@@ -31,14 +30,14 @@ public:

void set_doctype_name(std::string name) override { document_.doctype = std::move(name); }

void set_quirks_mode(QuirksMode mode) override {

void set_quirks_mode(html2::QuirksMode mode) override {

document_.mode = [=] {

switch (mode) {

case QuirksMode::NoQuirks:

case html2::QuirksMode::NoQuirks:

return dom::Document::Mode::NoQuirks;

case QuirksMode::Quirks:

case html2::QuirksMode::Quirks:

return dom::Document::Mode::Quirks;

case QuirksMode::LimitedQuirks:

case html2::QuirksMode::LimitedQuirks:

break;

}

return dom::Document::Mode::LimitedQuirks;

@@ -85,8 +84,11 @@ public:

void set_tokenizer_state(html2::State state) override { tokenizer_.set_state(state); }

void store_original_insertion_mode(InsertionMode mode) override { original_insertion_mode_ = std::move(mode); }

InsertionMode original_insertion_mode() override { return std::move(original_insertion_mode_); }

void store_original_insertion_mode(html2::InsertionMode mode) override {

original_insertion_mode_ = std::move(mode);

}

html2::InsertionMode original_insertion_mode() override { return std::move(original_insertion_mode_); }

private:

void insert(dom::Element element) {

@@ -105,7 +107,7 @@ private:

dom::Document &document_;

html2::Tokenizer &tokenizer_;

bool scripting_;

InsertionMode original_insertion_mode_;

html2::InsertionMode original_insertion_mode_;

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

};

html2/BUILD added: 57, removed: 44, total 13

@@ -18,6 +18,14 @@ cc_library(

)

# TODO(robinlinden): Remove.

extra_deps = {

"parser_states": [

"//dom",

"//html",

}

[cc_test(

name = src[:-4],

size = "small",

@@ -27,7 +35,10 @@ cc_library(

":html2",

"//etest",

"@fmt",

] + extra_deps.get(

src[:-9],

[],

) for src in glob(

include = ["*_test.cpp"],

exclude = ["*_fuzz_test.cpp"],

html/iparser_actions.h added: 57, removed: 44, total 13

@@ -2,18 +2,17 @@

// SPDX-License-Identifier: BSD-2-Clause

#ifndef HTML_IPARSER_ACTIONS_H_

#define HTML_IPARSER_ACTIONS_H_

#include "html/parser_states.h"

#ifndef HTML2_IPARSER_ACTIONS_H_

#define HTML2_IPARSER_ACTIONS_H_

#include "html2/parser_states.h"

#include "html2/tokenizer.h"

#include <span>

#include <string>

#include <string_view>

namespace html {

namespace html2 {

enum class QuirksMode {

NoQuirks,

@@ -38,6 +37,6 @@ public:

virtual InsertionMode original_insertion_mode() = 0;

};

} // namespace html

} // namespace html2

#endif

html/parser_states.cpp added: 57, removed: 44, total 13

@@ -2,11 +2,11 @@

// SPDX-License-Identifier: BSD-2-Clause

#include "html/parser_states.h"

#include "html/iparser_actions.h"

#include "html2/parser_states.h"

#include "html2/iparser_actions.h"

#include "html2/tokenizer.h"

#include "util/string.h"

#include <array>

@@ -18,7 +18,7 @@

using namespace std::literals;

namespace html {

namespace html2 {

namespace {

// A character token that is one of U+0009 CHARACTER TABULATION, U+000A LINE

@@ -352,4 +352,4 @@ std::optional<InsertionMode> Text::process(IActions &a, html2::Token const &toke

return {};

}

} // namespace html

} // namespace html2

html/parser_states.h added: 57, removed: 44, total 13

@@ -2,14 +2,14 @@

// SPDX-License-Identifier: BSD-2-Clause

#ifndef HTML_PARSER_STATES_H_

#define HTML_PARSER_STATES_H_

#ifndef HTML2_PARSER_STATES_H_

#define HTML2_PARSER_STATES_H_

#include "html2/tokenizer.h"

#include <variant>

namespace html {

namespace html2 {

class IActions;

@@ -127,6 +127,6 @@ struct Text {

std::optional<InsertionMode> process(IActions &, html2::Token const &);

};

} // namespace html

} // namespace html2

#endif

html/parser_states_test.cpp added: 57, removed: 44, total 13

@@ -2,11 +2,12 @@

// SPDX-License-Identifier: BSD-2-Clause

#include "html/parser_actions.h"

#include "html/parser_states.h"

#include "html2/parser_states.h"

#include "html2/tokenizer.h"

#include "etest/etest.h"

#include "html2/tokenizer.h"

#include "html/parser_actions.h"

using etest::expect_eq;

@@ -18,7 +19,7 @@ struct ParseResult {

};

struct ParseOptions {

html::InsertionMode initial_insertion_mode{};

html2::InsertionMode initial_insertion_mode{};

bool scripting{false};

};

@@ -28,7 +29,7 @@ ParseResult parse(std::string_view html, ParseOptions opts) {

}};

ParseResult res{};

html::InsertionMode mode{opts.initial_insertion_mode};

html2::InsertionMode mode{opts.initial_insertion_mode};

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

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