AK: Add Utf16View for decoding UTF-16 strings

Also includes a way to transcode from and to UTF-8 strings.

AK/Forward.h

@@ -29,6 +29,7 @@ class StringView;
 class Time;
 class URL;
 class FlyString;
+class Utf16View;
 class Utf32View;
 class Utf8View;
 class InputStream;
@@ -168,6 +169,7 @@ using AK::StringView;
 using AK::Time;
 using AK::Traits;
 using AK::URL;
+using AK::Utf16View;
 using AK::Utf32View;
 using AK::Utf8View;
 using AK::Vector;

AK/Utf16View.cpp

@@ -0,0 +1,197 @@
+/*
+ * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include <AK/StringBuilder.h>
+#include <AK/StringView.h>
+#include <AK/Utf16View.h>
+#include <AK/Utf8View.h>
+
+namespace AK {
+
+static constexpr u16 high_surrogate_min = 0xd800;
+static constexpr u16 high_surrogate_max = 0xdbff;
+static constexpr u16 low_surrogate_min = 0xdc00;
+static constexpr u16 low_surrogate_max = 0xdfff;
+static constexpr u32 replacement_code_point = 0xfffd;
+static constexpr u32 first_supplementary_plane_code_point = 0x10000;
+
+Vector<u16> utf8_to_utf16(StringView const& utf8_view)
+{
+    return utf8_to_utf16(Utf8View { utf8_view });
+}
+
+Vector<u16> utf8_to_utf16(Utf8View const& utf8_view)
+{
+    Vector<u16> utf16_data;
+
+    for (auto code_point : utf8_view) {
+        if (code_point < first_supplementary_plane_code_point) {
+            utf16_data.append(static_cast<u16>(code_point));
+        } else {
+            code_point -= first_supplementary_plane_code_point;
+            utf16_data.append(static_cast<u16>(high_surrogate_min | (code_point >> 10)));
+            utf16_data.append(static_cast<u16>(low_surrogate_min | (code_point & 0x3ff)));
+        }
+    }
+
+    return utf16_data;
+}
+
+bool Utf16View::is_high_surrogate(u16 code_unit)
+{
+    return (code_unit >= high_surrogate_min) && (code_unit <= high_surrogate_max);
+}
+
+bool Utf16View::is_low_surrogate(u16 code_unit)
+{
+    return (code_unit >= low_surrogate_min) && (code_unit <= low_surrogate_max);
+}
+
+u32 Utf16View::decode_surrogate_pair(u16 high_surrogate, u16 low_surrogate)
+{
+    VERIFY(is_high_surrogate(high_surrogate));
+    VERIFY(is_low_surrogate(low_surrogate));
+
+    return ((high_surrogate - high_surrogate_min) << 10) + (low_surrogate - low_surrogate_min) + first_supplementary_plane_code_point;
+}
+
+String Utf16View::to_utf8(AllowInvalidCodeUnits allow_invalid_code_units) const
+{
+    StringBuilder builder;
+
+    if (allow_invalid_code_units == AllowInvalidCodeUnits::Yes) {
+        for (auto const* ptr = begin_ptr(); ptr < end_ptr(); ++ptr) {
+            if (is_high_surrogate(*ptr)) {
+                auto const* next = ptr + 1;
+
+                if ((next < end_ptr()) && is_low_surrogate(*next)) {
+                    auto code_point = decode_surrogate_pair(*ptr, *next);
+                    builder.append_code_point(code_point);
+                    ++ptr;
+                    continue;
+                }
+            }
+
+            builder.append_code_point(static_cast<u32>(*ptr));
+        }
+    } else {
+        for (auto code_point : *this)
+            builder.append_code_point(code_point);
+    }
+
+    return builder.build();
+}
+
+size_t Utf16View::length_in_code_points() const
+{
+    if (!m_length_in_code_points.has_value())
+        m_length_in_code_points = calculate_length_in_code_points();
+    return *m_length_in_code_points;
+}
+
+u16 Utf16View::code_unit_at(size_t index) const
+{
+    VERIFY(index < length_in_code_units());
+    return m_code_units[index];
+}
+
+Utf16View Utf16View::substring_view(size_t code_unit_offset, size_t code_unit_length) const
+{
+    VERIFY(!Checked<size_t>::addition_would_overflow(code_unit_offset, code_unit_length));
+    VERIFY(code_unit_offset + code_unit_length <= length_in_code_units());
+
+    return Utf16View { m_code_units.slice(code_unit_offset, code_unit_length) };
+}
+
+bool Utf16View::validate(size_t& valid_code_units) const
+{
+    valid_code_units = 0;
+
+    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;
+}
+
+size_t Utf16View::calculate_length_in_code_points() const
+{
+    size_t code_points = 0;
+    for ([[maybe_unused]] auto code_point : *this)
+        ++code_points;
+    return code_points;
+}
+
+bool Utf16View::operator==(Utf16View const& other) const
+{
+    if (length_in_code_units() == 0)
+        return other.length_in_code_units() == 0;
+    if (length_in_code_units() != other.length_in_code_units())
+        return false;
+
+    for (size_t i = 0; i < length_in_code_units(); ++i) {
+        if (m_code_units[i] != other.m_code_units[i])
+            return false;
+    }
+
+    return true;
+}
+
+Utf16CodePointIterator& Utf16CodePointIterator::operator++()
+{
+    size_t code_units = length_in_code_units();
+
+    if (code_units > m_remaining_code_units) {
+        // If there aren't enough code units remaining, skip to the end.
+        m_ptr += m_remaining_code_units;
+        m_remaining_code_units = 0;
+    } else {
+        m_ptr += code_units;
+        m_remaining_code_units -= code_units;
+    }
+
+    return *this;
+}
+
+u32 Utf16CodePointIterator::operator*() const
+{
+    VERIFY(m_remaining_code_units > 0);
+
+    if (Utf16View::is_high_surrogate(*m_ptr)) {
+        if ((m_remaining_code_units > 1) && Utf16View::is_low_surrogate(*(m_ptr + 1)))
+            return Utf16View::decode_surrogate_pair(*m_ptr, *(m_ptr + 1));
+        return replacement_code_point;
+    } else if (Utf16View::is_low_surrogate(*m_ptr)) {
+        return replacement_code_point;
+    }
+
+    return static_cast<u32>(*m_ptr);
+}
+
+size_t Utf16CodePointIterator::length_in_code_units() const
+{
+    VERIFY(m_remaining_code_units > 0);
+
+    if (Utf16View::is_high_surrogate(*m_ptr)) {
+        if ((m_remaining_code_units > 1) && Utf16View::is_low_surrogate(*(m_ptr + 1)))
+            return 2;
+    }
+
+    // If this return is reached, either the encoded code point is a valid single code unit, or that
+    // code point is invalid (e.g. began with a low surrogate, or a low surrogate did not follow a
+    // high surrogate). In the latter case, a single replacement code unit will be used.
+    return 1;
+}
+
+}

AK/Utf16View.h

@@ -0,0 +1,113 @@
+/*
+ * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <AK/Forward.h>
+#include <AK/Optional.h>
+#include <AK/Span.h>
+#include <AK/String.h>
+#include <AK/Types.h>
+#include <AK/Vector.h>
+
+namespace AK {
+
+Vector<u16> utf8_to_utf16(StringView const&);
+Vector<u16> utf8_to_utf16(Utf8View const&);
+
+class Utf16View;
+
+class Utf16CodePointIterator {
+    friend class Utf16View;
+
+public:
+    Utf16CodePointIterator() = default;
+    ~Utf16CodePointIterator() = default;
+
+    bool operator==(Utf16CodePointIterator const& other) const
+    {
+        return (m_ptr == other.m_ptr) && (m_remaining_code_units == other.m_remaining_code_units);
+    }
+
+    bool operator!=(Utf16CodePointIterator const& other) const
+    {
+        return !(*this == other);
+    }
+
+    Utf16CodePointIterator& operator++();
+    u32 operator*() const;
+
+    size_t length_in_code_units() const;
+
+private:
+    Utf16CodePointIterator(u16 const* ptr, size_t length)
+        : m_ptr(ptr)
+        , m_remaining_code_units(length)
+    {
+    }
+
+    u16 const* m_ptr { nullptr };
+    size_t m_remaining_code_units { 0 };
+};
+
+class Utf16View {
+public:
+    static bool is_high_surrogate(u16);
+    static bool is_low_surrogate(u16);
+    static u32 decode_surrogate_pair(u16 high_surrogate, u16 low_surrogate);
+
+    Utf16View() = default;
+    ~Utf16View() = default;
+
+    explicit Utf16View(Span<u16 const> code_units)
+        : m_code_units(code_units)
+    {
+    }
+
+    Utf16View& operator=(Utf16View const&) = default;
+    Utf16View& operator=(Utf16View&&) = default;
+    bool operator==(Utf16View const& other) const;
+
+    enum class AllowInvalidCodeUnits {
+        Yes,
+        No,
+    };
+
+    String to_utf8(AllowInvalidCodeUnits = AllowInvalidCodeUnits::No) const;
+
+    bool is_empty() const { return m_code_units.is_empty(); }
+    size_t length_in_code_units() const { return m_code_units.size(); }
+    size_t length_in_code_points() const;
+
+    Utf16CodePointIterator begin() const { return { begin_ptr(), m_code_units.size() }; }
+    Utf16CodePointIterator end() const { return { end_ptr(), 0 }; }
+
+    u16 const* data() const { return m_code_units.data(); }
+    u16 code_unit_at(size_t index) const;
+
+    Utf16View substring_view(size_t code_unit_offset, size_t code_unit_length) const;
+    Utf16View substring_view(size_t code_unit_offset) const { return substring_view(code_unit_offset, length_in_code_units() - code_unit_offset); }
+
+    bool validate(size_t& valid_code_units) const;
+    bool validate() const
+    {
+        size_t valid_code_units;
+        return validate(valid_code_units);
+    }
+
+private:
+    u16 const* begin_ptr() const { return m_code_units.data(); }
+    u16 const* end_ptr() const { return begin_ptr() + m_code_units.size(); }
+
+    size_t calculate_length_in_code_points() const;
+
+    Span<u16 const> m_code_units;
+    mutable Optional<size_t> m_length_in_code_points;
+};
+
+}
+
+using AK::Utf16View;

Tests/AK/CMakeLists.txt

@@ -61,6 +61,7 @@ set(AK_TEST_SOURCES
     TestTypedTransfer.cpp
     TestUFixedBigInt.cpp
     TestURL.cpp
+    TestUtf16.cpp
     TestUtf8.cpp
     TestVariant.cpp
     TestVector.cpp

Tests/AK/TestUtf16.cpp

@@ -0,0 +1,282 @@
+/*
+ * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include <LibTest/TestCase.h>
+
+#include <AK/Array.h>
+#include <AK/String.h>
+#include <AK/StringView.h>
+#include <AK/Types.h>
+#include <AK/Utf16View.h>
+
+TEST_CASE(decode_ascii)
+{
+    auto string = AK::utf8_to_utf16("Hello World!11"sv);
+    Utf16View view { string };
+
+    size_t valid_code_units = 0;
+    EXPECT(view.validate(valid_code_units));
+    EXPECT_EQ(valid_code_units, view.length_in_code_units());
+
+    auto expected = Array { (u32)72, 101, 108, 108, 111, 32, 87, 111, 114, 108, 100, 33, 49, 49 };
+    EXPECT_EQ(expected.size(), view.length_in_code_points());
+
+    size_t i = 0;
+    for (u32 code_point : view) {
+        EXPECT_EQ(code_point, expected[i++]);
+    }
+    EXPECT_EQ(i, expected.size());
+}
+
+TEST_CASE(decode_utf8)
+{
+    auto string = AK::utf8_to_utf16("Привет, мир! 😀 γειά σου κόσμος こんにちは世界"sv);
+    Utf16View view { string };
+
+    size_t valid_code_units = 0;
+    EXPECT(view.validate(valid_code_units));
+    EXPECT_EQ(valid_code_units, view.length_in_code_units());
+
+    auto expected = Array { (u32)1055, 1088, 1080, 1074, 1077, 1090, 44, 32, 1084, 1080, 1088, 33, 32, 128512, 32, 947, 949, 953, 940, 32, 963, 959, 965, 32, 954, 972, 963, 956, 959, 962, 32, 12371, 12435, 12395, 12385, 12399, 19990, 30028 };
+    EXPECT_EQ(expected.size(), view.length_in_code_points());
+
+    size_t i = 0;
+    for (u32 code_point : view) {
+        EXPECT_EQ(code_point, expected[i++]);
+    }
+    EXPECT_EQ(i, expected.size());
+}
+
+TEST_CASE(encode_utf8)
+{
+    {
+        String utf8_string("Привет, мир! 😀 γειά σου κόσμος こんにちは世界");
+        auto string = AK::utf8_to_utf16(utf8_string);
+        Utf16View view { string };
+        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes), utf8_string);
+        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::No), utf8_string);
+    }
+    {
+        auto encoded = Array { (u16)0xd83d };
+        Utf16View view { encoded };
+        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes), "\xed\xa0\xbd"sv);
+        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::No), "\ufffd"sv);
+    }
+}
+
+TEST_CASE(decode_utf16)
+{
+    // Same string as the decode_utf8 test.
+    auto encoded = Array { (u16)0x041f, 0x0440, 0x0438, 0x0432, 0x0435, 0x0442, 0x002c, 0x0020, 0x043c, 0x0438, 0x0440, 0x0021, 0x0020, 0xd83d, 0xde00, 0x0020, 0x03b3, 0x03b5, 0x03b9, 0x03ac, 0x0020, 0x03c3, 0x03bf, 0x03c5, 0x0020, 0x03ba, 0x03cc, 0x03c3, 0x03bc, 0x03bf, 0x03c2, 0x0020, 0x3053, 0x3093, 0x306b, 0x3061, 0x306f, 0x4e16, 0x754c };
+
+    Utf16View view { encoded };
+    EXPECT_EQ(encoded.size(), view.length_in_code_units());
+
+    size_t valid_code_units = 0;
+    EXPECT(view.validate(valid_code_units));
+    EXPECT_EQ(valid_code_units, view.length_in_code_units());
+
+    auto expected = Array { (u32)1055, 1088, 1080, 1074, 1077, 1090, 44, 32, 1084, 1080, 1088, 33, 32, 128512, 32, 947, 949, 953, 940, 32, 963, 959, 965, 32, 954, 972, 963, 956, 959, 962, 32, 12371, 12435, 12395, 12385, 12399, 19990, 30028 };
+    EXPECT_EQ(expected.size(), view.length_in_code_points());
+
+    size_t i = 0;
+    for (u32 code_point : view) {
+        EXPECT_EQ(code_point, expected[i++]);
+    }
+    EXPECT_EQ(i, expected.size());
+}
+
+TEST_CASE(iterate_utf16)
+{
+    auto string = AK::utf8_to_utf16("Привет 😀"sv);
+    Utf16View view { string };
+    auto iterator = view.begin();
+
+    EXPECT(*iterator == 1055);
+    EXPECT(iterator.length_in_code_units() == 1);
+
+    EXPECT(++iterator != view.end());
+    EXPECT(*iterator == 1088);
+    EXPECT(iterator.length_in_code_units() == 1);
+
+    EXPECT(++iterator != view.end());
+    EXPECT(*iterator == 1080);
+    EXPECT(iterator.length_in_code_units() == 1);
+
+    EXPECT(++iterator != view.end());
+    EXPECT(*iterator == 1074);
+    EXPECT(iterator.length_in_code_units() == 1);
+
+    EXPECT(++iterator != view.end());
+    EXPECT(*iterator == 1077);
+    EXPECT(iterator.length_in_code_units() == 1);
+
+    EXPECT(++iterator != view.end());
+    EXPECT(*iterator == 1090);
+    EXPECT(iterator.length_in_code_units() == 1);
+
+    EXPECT(++iterator != view.end());
+    EXPECT(*iterator == 32);
+    EXPECT(iterator.length_in_code_units() == 1);
+
+    EXPECT(++iterator != view.end());
+    EXPECT(*iterator == 128512);
+    EXPECT(iterator.length_in_code_units() == 2);
+
+    EXPECT(++iterator == view.end());
+    EXPECT_CRASH("Dereferencing Utf16CodePointIterator which is at its end.", [&iterator] {
+        *iterator;
+        return Test::Crash::Failure::DidNotCrash;
+    });
+
+    EXPECT_CRASH("Incrementing Utf16CodePointIterator which is at its end.", [&iterator] {
+        ++iterator;
+        return Test::Crash::Failure::DidNotCrash;
+    });
+}
+
+TEST_CASE(validate_invalid_utf16)
+{
+    size_t valid_code_units = 0;
+    {
+        // Lonely high surrogate.
+        auto invalid = Array { (u16)0xd800 };
+        EXPECT(!Utf16View(invalid).validate(valid_code_units));
+        EXPECT(valid_code_units == 0);
+
+        invalid = Array { (u16)0xdbff };
+        EXPECT(!Utf16View(invalid).validate(valid_code_units));
+        EXPECT(valid_code_units == 0);
+    }
+    {
+        // Lonely low surrogate.
+        auto invalid = Array { (u16)0xdc00 };
+        EXPECT(!Utf16View(invalid).validate(valid_code_units));
+        EXPECT(valid_code_units == 0);
+
+        invalid = Array { (u16)0xdfff };
+        EXPECT(!Utf16View(invalid).validate(valid_code_units));
+        EXPECT(valid_code_units == 0);
+    }
+    {
+        // High surrogate followed by non-surrogate.
+        auto invalid = Array { (u16)0xd800, 0 };
+        EXPECT(!Utf16View(invalid).validate(valid_code_units));
+        EXPECT(valid_code_units == 0);
+
+        invalid = Array { (u16)0xd800, 0xe000 };
+        EXPECT(!Utf16View(invalid).validate(valid_code_units));
+        EXPECT(valid_code_units == 0);
+    }
+    {
+        // High surrogate followed by high surrogate.
+        auto invalid = Array { (u16)0xd800, 0xd800 };
+        EXPECT(!Utf16View(invalid).validate(valid_code_units));
+        EXPECT(valid_code_units == 0);
+
+        invalid = Array { (u16)0xd800, 0xdbff };
+        EXPECT(!Utf16View(invalid).validate(valid_code_units));
+        EXPECT(valid_code_units == 0);
+    }
+    {
+        // Valid UTF-16 followed by invalid code units.
+        auto invalid = Array { (u16)0x41, 0x41, 0xd800 };
+        EXPECT(!Utf16View(invalid).validate(valid_code_units));
+        EXPECT(valid_code_units == 2);
+
+        invalid = Array { (u16)0x41, 0x41, 0xd800 };
+        EXPECT(!Utf16View(invalid).validate(valid_code_units));
+        EXPECT(valid_code_units == 2);
+    }
+}
+
+TEST_CASE(decode_invalid_utf16)
+{
+    {
+        // Lonely high surrogate.
+        auto invalid = Array { (u16)0x41, 0x42, 0xd800 };
+
+        Utf16View view { invalid };
+        EXPECT_EQ(invalid.size(), view.length_in_code_units());
+
+        auto expected = Array { (u32)0x41, 0x42, 0xfffd };
+        EXPECT_EQ(expected.size(), view.length_in_code_points());
+
+        size_t i = 0;
+        for (u32 code_point : view) {
+            EXPECT_EQ(code_point, expected[i++]);
+        }
+        EXPECT_EQ(i, expected.size());
+    }
+    {
+        // Lonely low surrogate.
+        auto invalid = Array { (u16)0x41, 0x42, 0xdc00 };
+
+        Utf16View view { invalid };
+        EXPECT_EQ(invalid.size(), view.length_in_code_units());
+
+        auto expected = Array { (u32)0x41, 0x42, 0xfffd };
+        EXPECT_EQ(expected.size(), view.length_in_code_points());
+
+        size_t i = 0;
+        for (u32 code_point : view) {
+            EXPECT_EQ(code_point, expected[i++]);
+        }
+        EXPECT_EQ(i, expected.size());
+    }
+    {
+        // High surrogate followed by non-surrogate.
+        auto invalid = Array { (u16)0x41, 0x42, 0xd800, 0 };
+
+        Utf16View view { invalid };
+        EXPECT_EQ(invalid.size(), view.length_in_code_units());
+
+        auto expected = Array { (u32)0x41, 0x42, 0xfffd, 0 };
+        EXPECT_EQ(expected.size(), view.length_in_code_points());
+
+        size_t i = 0;
+        for (u32 code_point : view) {
+            EXPECT_EQ(code_point, expected[i++]);
+        }
+        EXPECT_EQ(i, expected.size());
+    }
+    {
+        // High surrogate followed by high surrogate.
+        auto invalid = Array { (u16)0x41, 0x42, 0xd800, 0xd800 };
+
+        Utf16View view { invalid };
+        EXPECT_EQ(invalid.size(), view.length_in_code_units());
+
+        auto expected = Array { (u32)0x41, 0x42, 0xfffd, 0xfffd };
+        EXPECT_EQ(expected.size(), view.length_in_code_points());
+
+        size_t i = 0;
+        for (u32 code_point : view) {
+            EXPECT_EQ(code_point, expected[i++]);
+        }
+        EXPECT_EQ(i, expected.size());
+    }
+}
+
+TEST_CASE(substring_view)
+{
+    auto string = AK::utf8_to_utf16("Привет 😀"sv);
+    {
+        Utf16View view { string };
+        view = view.substring_view(7, 2);
+
+        EXPECT(view.length_in_code_units() == 2);
+        EXPECT_EQ(view.to_utf8(), "😀"sv);
+    }
+    {
+        Utf16View view { string };
+        view = view.substring_view(7, 1);
+
+        EXPECT(view.length_in_code_units() == 1);
+        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes), "\xed\xa0\xbd"sv);
+        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::No), "\ufffd"sv);
+    }
+}