srctree

Robin Linden parent ce5592f1 4c2f55e8
html2: Fix errors not being emitted for self-closing end tags

inline split

html2/tokenizer.cpp added: 16, removed: 6, total 10

@@ -1,4 +1,4 @@

// SPDX-FileCopyrightText: 2021-2023 Robin Lindén <dev@robinlinden.eu>

// SPDX-FileCopyrightText: 2021-2024 Robin Lindén <dev@robinlinden.eu>

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

@@ -2484,7 +2484,12 @@ void Tokenizer::emit(Token &&token) {

deduplicate(start_tag->attributes);

} else if (auto *end_tag = std::get_if<EndTagToken>(&token)) {

deduplicate(end_tag->attributes);

// https://html.spec.whatwg.org/multipage/parsing.html#tokenization:parse-error-end-tag-with-trailing-solidus

if (end_tag->self_closing) {

emit(ParseError::EndTagWithTrailingSolidus);

}

on_emit_(*this, std::move(token));

}

html2/tokenizer.h added: 16, removed: 6, total 10

@@ -1,4 +1,4 @@

// SPDX-FileCopyrightText: 2021-2023 Robin Lindén <dev@robinlinden.eu>

// SPDX-FileCopyrightText: 2021-2024 Robin Lindén <dev@robinlinden.eu>

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

@@ -111,6 +111,7 @@ enum class ParseError {

CharacterReferenceOutsideUnicodeRange,

ControlCharacterReference,

DuplicateAttribute,

EndTagWithTrailingSolidus,

EofBeforeTagName,

EofInCdata,

EofInComment,

html2/tokenizer_test.cpp added: 16, removed: 6, total 10

@@ -1,4 +1,4 @@

// SPDX-FileCopyrightText: 2021-2023 Robin Lindén <dev@robinlinden.eu>

// SPDX-FileCopyrightText: 2021-2024 Robin Lindén <dev@robinlinden.eu>

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

@@ -279,6 +279,7 @@ void rawtext_tests() {

expect_text(tokens, "<div>");

expect_token(tokens, EndTagToken{.tag_name = "style", .self_closing = true});

expect_token(tokens, EndOfFileToken{});

expect_error(tokens, ParseError::EndTagWithTrailingSolidus);

});

etest::test("rawtext, end tag open, eof", [] {

@@ -344,6 +345,7 @@ void rcdata_tests() {

expect_text(tokens, "<div>");

expect_token(tokens, EndTagToken{.tag_name = "title", .self_closing = true});

expect_token(tokens, EndOfFileToken{});

expect_error(tokens, ParseError::EndTagWithTrailingSolidus);

});

etest::test("rcdata, end tag open, eof", [] {

@@ -926,6 +928,7 @@ int main() {

expect_token(tokens, StartTagToken{.tag_name = "script"});

expect_token(tokens, EndTagToken{.tag_name = "script", .self_closing = true});

expect_token(tokens, EndOfFileToken{});

expect_error(tokens, ParseError::EndTagWithTrailingSolidus);

});

etest::test("script, misspelled self closing end tag", [] {

@@ -974,6 +977,7 @@ int main() {

expect_text(tokens, "-->"sv);

expect_token(tokens, EndTagToken{.tag_name = "script"});

expect_token(tokens, EndOfFileToken{});

expect_error(tokens, ParseError::EndTagWithTrailingSolidus);

});

etest::test("script, misspelled escaped self closing end tag", [] {