fergalmoran/ladybird
1
0
Fork 0
mirror of https://github.com/fergalmoran/ladybird.git synced 2025-12-22 09:19:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

AK: Replace UTF-16 validation and length computation with simdutf

Browse Source
This commit is contained in:
Timothy Flynn
2024-07-16 13:02:51 -04:00
committed by Andreas Kling
parent a2bcb2ab8d
commit 32ffe9bbfc
2 changed files with 21 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

34
AK/Utf16View.cpp
View File

@@ -1,9 +1,11 @@
/*
* Copyright (c) 2021-2023, Tim Flynn <trflynn89@serenityos.org>
* Copyright (c) 2021-2024, Tim Flynn <trflynn89@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
#define AK_DONT_REPLACE_STD
#include <AK/CharacterTypes.h>
#include <AK/Concepts.h>
#include <AK/StringBuilder.h>
@@ -12,6 +14,8 @@
#include <AK/Utf32View.h>
#include <AK/Utf8View.h>
#include <simdutf.h>
namespace AK {
static constexpr u16 high_surrogate_min = 0xd800;
@@ -233,27 +237,27 @@ bool Utf16View::starts_with(Utf16View const& needle) const
return true;
}
bool Utf16View::validate() const
{
return simdutf::validate_utf16(reinterpret_cast<char16_t const*>(m_code_units.data()), m_code_units.size());
}
bool Utf16View::validate(size_t& valid_code_units) const
{
valid_code_units = 0;
auto result = simdutf::validate_utf16_with_errors(reinterpret_cast<char16_t const*>(m_code_units.data()), m_code_units.size());
valid_code_units = result.count;
for (auto const* ptr = begin_ptr(); ptr < end_ptr(); ++ptr) {
if (is_high_surrogate(*ptr)) {
if ((++ptr >= end_ptr()) || !is_low_surrogate(*ptr))
return false;
++valid_code_units;
} else if (is_low_surrogate(*ptr)) {
return false;
}
++valid_code_units;
}
return true;
return result.error == simdutf::SUCCESS;
}
size_t Utf16View::calculate_length_in_code_points() const
{
// FIXME: simdutf's code point length method assumes valid UTF-16, whereas Utf16View uses U+FFFD as a replacement
// for invalid code points. If we change Utf16View to only accept valid encodings as an invariant, we can
// remove this branch.
if (validate()) [[likely]]
return simdutf::count_utf16(reinterpret_cast<char16_t const*>(m_code_units.data()), m_code_units.size());
size_t code_points = 0;
for ([[maybe_unused]] auto code_point : *this)
++code_points;

8
AK/Utf16View.h
View File

@@ -1,5 +1,5 @@
/*
* Copyright (c) 2021-2023, Tim Flynn <trflynn89@serenityos.org>
* Copyright (c) 2021-2024, Tim Flynn <trflynn89@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
@@ -113,12 +113,8 @@ public:
bool starts_with(Utf16View const&) const;
bool validate() const;
bool validate(size_t& valid_code_units) const;
bool validate() const
{
size_t valid_code_units;
return validate(valid_code_units);
}
bool equals_ignoring_case(Utf16View const&) const;