зеркало из https://github.com/microsoft/terminal.git
Clean up CodepointWidthDetector (#14396)
My long-term plan is to replace the `CodepointWidth` enum with a simple integer return value that indicates the amount of columns a codepoint is wide. This is necessary so that we can return 0 for ZWJs (zero width joiners). This initial commit represents a cleanup effort around `CodepointWidthDetector`. Since less code runs faster, this change has the nice side-effect of running roughly 5-10% faster across the board. It also drops the binary size by ~1.2kB. ## Validation Steps Performed * `CodepointWidthDetectorTests` passes ✅ * U+26bf (``"`u{26bf}"`` inside pwsh) is a wide glyph in OpenConsole and narrow one in Windows Terminal ✅
This commit is contained in:
Leonard Hecker
GitHub
Родитель
62ffa4ba41
Коммит
4bbe3a388c
|
|
@ -82,7 +82,7 @@ OutputCellIterator::OutputCellIterator(const CHAR_INFO& charInfo, const size_t f
|
|||
// - This is an iterator over a range of text only. No color data will be modified as the text is inserted.
|
||||
// Arguments:
|
||||
// - utf16Text - UTF-16 text range
|
||||
OutputCellIterator::OutputCellIterator(const std::wstring_view utf16Text) :
|
||||
OutputCellIterator::OutputCellIterator(const std::wstring_view utf16Text) noexcept :
|
||||
_mode(Mode::LooseTextOnly),
|
||||
_currentView(s_GenerateView(utf16Text)),
|
||||
_run(utf16Text),
|
||||
|
|
@ -98,7 +98,7 @@ OutputCellIterator::OutputCellIterator(const std::wstring_view utf16Text) :
|
|||
// Arguments:
|
||||
// - utf16Text - UTF-16 text range
|
||||
// - attribute - Color to apply over the entire range
|
||||
OutputCellIterator::OutputCellIterator(const std::wstring_view utf16Text, const TextAttribute& attribute, const size_t fillLimit) :
|
||||
OutputCellIterator::OutputCellIterator(const std::wstring_view utf16Text, const TextAttribute& attribute, const size_t fillLimit) noexcept :
|
||||
_mode(Mode::Loose),
|
||||
_currentView(s_GenerateView(utf16Text, attribute)),
|
||||
_run(utf16Text),
|
||||
|
|
@ -357,7 +357,7 @@ bool OutputCellIterator::_TryMoveTrailing() noexcept
|
|||
// - view - View representing characters corresponding to a single glyph
|
||||
// Return Value:
|
||||
// - Object representing the view into this cell
|
||||
OutputCellView OutputCellIterator::s_GenerateView(const std::wstring_view view)
|
||||
OutputCellView OutputCellIterator::s_GenerateView(const std::wstring_view view) noexcept
|
||||
{
|
||||
return s_GenerateView(view, InvalidTextAttribute, TextAttributeBehavior::Current);
|
||||
}
|
||||
|
|
@ -372,8 +372,7 @@ OutputCellView OutputCellIterator::s_GenerateView(const std::wstring_view view)
|
|||
// - attr - Color attributes to apply to the text
|
||||
// Return Value:
|
||||
// - Object representing the view into this cell
|
||||
OutputCellView OutputCellIterator::s_GenerateView(const std::wstring_view view,
|
||||
const TextAttribute attr)
|
||||
OutputCellView OutputCellIterator::s_GenerateView(const std::wstring_view view, const TextAttribute attr) noexcept
|
||||
{
|
||||
return s_GenerateView(view, attr, TextAttributeBehavior::Stored);
|
||||
}
|
||||
|
|
@ -389,9 +388,7 @@ OutputCellView OutputCellIterator::s_GenerateView(const std::wstring_view view,
|
|||
// - behavior - Behavior of the given text attribute (used when writing)
|
||||
// Return Value:
|
||||
// - Object representing the view into this cell
|
||||
OutputCellView OutputCellIterator::s_GenerateView(const std::wstring_view view,
|
||||
const TextAttribute attr,
|
||||
const TextAttributeBehavior behavior)
|
||||
OutputCellView OutputCellIterator::s_GenerateView(const std::wstring_view view, const TextAttribute attr, const TextAttributeBehavior behavior) noexcept
|
||||
{
|
||||
const auto glyph = til::utf16_next(view);
|
||||
const auto dbcsAttr = IsGlyphFullWidth(glyph) ? DbcsAttribute::Leading : DbcsAttribute::Single;
|
||||
|
|
|
|||
|
|
@ -37,8 +37,8 @@ public:
|
|||
OutputCellIterator(const TextAttribute& attr, const size_t fillLimit = 0) noexcept;
|
||||
OutputCellIterator(const wchar_t& wch, const TextAttribute& attr, const size_t fillLimit = 0) noexcept;
|
||||
OutputCellIterator(const CHAR_INFO& charInfo, const size_t fillLimit = 0) noexcept;
|
||||
OutputCellIterator(const std::wstring_view utf16Text);
|
||||
OutputCellIterator(const std::wstring_view utf16Text, const TextAttribute& attribute, const size_t fillLimit = 0);
|
||||
OutputCellIterator(const std::wstring_view utf16Text) noexcept;
|
||||
OutputCellIterator(const std::wstring_view utf16Text, const TextAttribute& attribute, const size_t fillLimit = 0) noexcept;
|
||||
OutputCellIterator(const gsl::span<const WORD> legacyAttributes) noexcept;
|
||||
OutputCellIterator(const gsl::span<const CHAR_INFO> charInfos) noexcept;
|
||||
OutputCellIterator(const gsl::span<const OutputCell> cells);
|
||||
|
|
@ -100,15 +100,9 @@ private:
|
|||
|
||||
bool _TryMoveTrailing() noexcept;
|
||||
|
||||
static OutputCellView s_GenerateView(const std::wstring_view view);
|
||||
|
||||
static OutputCellView s_GenerateView(const std::wstring_view view,
|
||||
const TextAttribute attr);
|
||||
|
||||
static OutputCellView s_GenerateView(const std::wstring_view view,
|
||||
const TextAttribute attr,
|
||||
const TextAttributeBehavior behavior);
|
||||
|
||||
static OutputCellView s_GenerateView(const std::wstring_view view) noexcept;
|
||||
static OutputCellView s_GenerateView(const std::wstring_view view, const TextAttribute attr) noexcept;
|
||||
static OutputCellView s_GenerateView(const std::wstring_view view, const TextAttribute attr, const TextAttributeBehavior behavior) noexcept;
|
||||
static OutputCellView s_GenerateView(const wchar_t& wch) noexcept;
|
||||
static OutputCellView s_GenerateViewLegacyAttr(const WORD& legacyAttr) noexcept;
|
||||
static OutputCellView s_GenerateView(const TextAttribute& attr) noexcept;
|
||||
|
|
|
|||
|
|
@ -24,25 +24,6 @@ static constexpr bool _IsMouseMessage(UINT uMsg)
|
|||
uMsg == WM_MOUSEMOVE || uMsg == WM_MOUSEWHEEL || uMsg == WM_MOUSEHWHEEL;
|
||||
}
|
||||
|
||||
// Helper static function to ensure that all ambiguous-width glyphs are reported as narrow.
|
||||
// See microsoft/terminal#2066 for more info.
|
||||
static bool _IsGlyphWideForceNarrowFallback(const std::wstring_view /* glyph */) noexcept
|
||||
{
|
||||
return false; // glyph is not wide.
|
||||
}
|
||||
|
||||
static bool _EnsureStaticInitialization()
|
||||
{
|
||||
// use C++11 magic statics to make sure we only do this once.
|
||||
static auto initialized = []() {
|
||||
// *** THIS IS A SINGLETON ***
|
||||
SetGlyphWidthFallback(_IsGlyphWideForceNarrowFallback);
|
||||
|
||||
return true;
|
||||
}();
|
||||
return initialized;
|
||||
}
|
||||
|
||||
LRESULT CALLBACK HwndTerminal::HwndTerminalWndProc(
|
||||
HWND hwnd,
|
||||
UINT uMsg,
|
||||
|
|
@ -175,7 +156,7 @@ static bool RegisterTermClass(HINSTANCE hInstance) noexcept
|
|||
return RegisterClassW(&wc) != 0;
|
||||
}
|
||||
|
||||
HwndTerminal::HwndTerminal(HWND parentHwnd) :
|
||||
HwndTerminal::HwndTerminal(HWND parentHwnd) noexcept :
|
||||
_desiredFont{ L"Consolas", 0, DEFAULT_FONT_WEIGHT, 14, CP_UTF8 },
|
||||
_actualFont{ L"Consolas", 0, DEFAULT_FONT_WEIGHT, { 0, 14 }, CP_UTF8, false },
|
||||
_uiaProvider{ nullptr },
|
||||
|
|
@ -183,8 +164,6 @@ HwndTerminal::HwndTerminal(HWND parentHwnd) :
|
|||
_pfnWriteCallback{ nullptr },
|
||||
_multiClickTime{ 500 } // this will be overwritten by the windows system double-click time
|
||||
{
|
||||
_EnsureStaticInitialization();
|
||||
|
||||
auto hInstance = wil::GetModuleInstanceHandle();
|
||||
|
||||
if (RegisterTermClass(hInstance))
|
||||
|
|
|
|||
|
|
@ -49,7 +49,7 @@ __declspec(dllexport) void _stdcall TerminalKillFocus(void* terminal);
|
|||
struct HwndTerminal : ::Microsoft::Console::Types::IControlAccessibilityInfo
|
||||
{
|
||||
public:
|
||||
HwndTerminal(HWND hwnd);
|
||||
HwndTerminal(HWND hwnd) noexcept;
|
||||
|
||||
HwndTerminal(const HwndTerminal&) = default;
|
||||
HwndTerminal(HwndTerminal&&) = default;
|
||||
|
|
|
|||
|
|
@ -63,25 +63,6 @@ namespace winrt::Microsoft::Terminal::Control::implementation
|
|||
return result;
|
||||
}
|
||||
|
||||
// Helper static function to ensure that all ambiguous-width glyphs are reported as narrow.
|
||||
// See microsoft/terminal#2066 for more info.
|
||||
static bool _IsGlyphWideForceNarrowFallback(const std::wstring_view /* glyph */)
|
||||
{
|
||||
return false; // glyph is not wide.
|
||||
}
|
||||
|
||||
static bool _EnsureStaticInitialization()
|
||||
{
|
||||
// use C++11 magic statics to make sure we only do this once.
|
||||
static auto initialized = []() {
|
||||
// *** THIS IS A SINGLETON ***
|
||||
SetGlyphWidthFallback(_IsGlyphWideForceNarrowFallback);
|
||||
|
||||
return true;
|
||||
}();
|
||||
return initialized;
|
||||
}
|
||||
|
||||
TextColor SelectionColor::AsTextColor() const noexcept
|
||||
{
|
||||
if (_IsIndex16)
|
||||
|
|
@ -101,8 +82,6 @@ namespace winrt::Microsoft::Terminal::Control::implementation
|
|||
_desiredFont{ DEFAULT_FONT_FACE, 0, DEFAULT_FONT_WEIGHT, DEFAULT_FONT_SIZE, CP_UTF8 },
|
||||
_actualFont{ DEFAULT_FONT_FACE, 0, DEFAULT_FONT_WEIGHT, { 0, DEFAULT_FONT_SIZE }, CP_UTF8, false }
|
||||
{
|
||||
_EnsureStaticInitialization();
|
||||
|
||||
_settings = winrt::make_self<implementation::ControlSettings>(settings, unfocusedAppearance);
|
||||
|
||||
_terminal = std::make_shared<::Microsoft::Terminal::Core::Terminal>();
|
||||
|
|
|
|||
|
|
@ -19,7 +19,7 @@ static const std::vector<std::tuple<unsigned int, std::wstring, CodepointWidth>>
|
|||
{ 0x7, L"\a", CodepointWidth::Narrow }, // BEL
|
||||
{ 0x20, L" ", CodepointWidth::Narrow },
|
||||
{ 0x39, L"9", CodepointWidth::Narrow },
|
||||
{ 0x414, L"\x414", CodepointWidth::Ambiguous }, // U+0414 cyrillic capital de
|
||||
{ 0x414, L"\x414", CodepointWidth::Narrow }, // U+0414 cyrillic capital de
|
||||
{ 0x1104, L"\x1104", CodepointWidth::Wide }, // U+1104 hangul choseong ssangtikeut
|
||||
{ 0x306A, L"\x306A", CodepointWidth::Wide }, // U+306A hiragana na
|
||||
{ 0x30CA, L"\x30CA", CodepointWidth::Wide }, // U+30CA katakana na
|
||||
|
|
@ -38,18 +38,6 @@ class CodepointWidthDetectorTests
|
|||
VERIFY_IS_TRUE(widthDetector.IsWide(emoji));
|
||||
}
|
||||
|
||||
TEST_METHOD(CanExtractCodepoint)
|
||||
{
|
||||
CodepointWidthDetector widthDetector;
|
||||
for (const auto& data : testData)
|
||||
{
|
||||
const auto& expected = std::get<0>(data);
|
||||
const auto& wstr = std::get<1>(data);
|
||||
const auto result = widthDetector._extractCodepoint({ wstr.c_str(), wstr.size() });
|
||||
VERIFY_ARE_EQUAL(result, expected);
|
||||
}
|
||||
}
|
||||
|
||||
TEST_METHOD(CanGetWidths)
|
||||
{
|
||||
CodepointWidthDetector widthDetector;
|
||||
|
|
@ -91,8 +79,8 @@ class CodepointWidthDetectorTests
|
|||
|
||||
// Cached item should match what we expect
|
||||
const auto it = widthDetector._fallbackCache.begin();
|
||||
VERIFY_ARE_EQUAL(ambiguous, it->first);
|
||||
VERIFY_ARE_EQUAL(FallbackMethod(ambiguous), it->second);
|
||||
VERIFY_ARE_EQUAL(ambiguous[0], it->first);
|
||||
VERIFY_ARE_EQUAL(FallbackMethod(ambiguous) ? 2u : 1u, it->second);
|
||||
|
||||
// Cache should empty when font changes.
|
||||
widthDetector.NotifyFontChanged();
|
||||
|
|
|
|||
|
|
@ -38,7 +38,7 @@ static CodepointWidth GetQuickCharWidthLegacyForNumpadEventSynthesis(const wchar
|
|||
return CodepointWidth::Wide;
|
||||
}
|
||||
|
||||
return CodepointWidth::Invalid;
|
||||
return CodepointWidth::Narrow;
|
||||
}
|
||||
|
||||
std::deque<std::unique_ptr<KeyEvent>> Microsoft::Console::Interactivity::CharToKeyEvents(const wchar_t wch,
|
||||
|
|
|
|||
|
|
@ -9,9 +9,9 @@ namespace
|
|||
// used to store range data in CodepointWidthDetector's internal map
|
||||
struct UnicodeRange final
|
||||
{
|
||||
unsigned int lowerBound;
|
||||
unsigned int upperBound;
|
||||
CodepointWidth width;
|
||||
char32_t lowerBound;
|
||||
char32_t upperBound : 31;
|
||||
char32_t isAmbiguous : 1;
|
||||
};
|
||||
|
||||
static bool operator<(const UnicodeRange& range, const unsigned int searchTerm) noexcept
|
||||
|
|
@ -19,382 +19,348 @@ namespace
|
|||
return range.upperBound < searchTerm;
|
||||
}
|
||||
|
||||
// Generated by Generate-CodepointWidthsFromUCD.ps1 -Pack:True -Full:False -NoOverrides:False
|
||||
// on 9/14/2022 7:12:26 PM (UTC) from Unicode 15.0.0.
|
||||
// 321281 (0x4E701) codepoints covered.
|
||||
// Generated by Generate-CodepointWidthsFromUCD.ps1 -Pack:True -Full: -NoOverrides:False
|
||||
// on 2022-11-15 19:54:23Z from Unicode 15.0.0.
|
||||
// 321149 (0x4E67D) codepoints covered.
|
||||
// 240 (0xF0) codepoints overridden.
|
||||
// Override path: .\src\types\unicode_width_overrides.xml
|
||||
static constexpr std::array<UnicodeRange, 300> s_wideAndAmbiguousTable{
|
||||
UnicodeRange{ 0xa1, 0xa1, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xa4, 0xa4, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xa7, 0xa8, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xaa, 0xaa, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xad, 0xae, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xb0, 0xb4, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xb6, 0xba, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xbc, 0xbf, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xc6, 0xc6, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xd0, 0xd0, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xd7, 0xd8, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xde, 0xe1, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xe6, 0xe6, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xe8, 0xea, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xec, 0xed, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xf0, 0xf0, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xf2, 0xf3, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xf7, 0xfa, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xfc, 0xfc, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xfe, 0xfe, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x101, 0x101, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x111, 0x111, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x113, 0x113, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x11b, 0x11b, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x126, 0x127, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x12b, 0x12b, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x131, 0x133, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x138, 0x138, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x13f, 0x142, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x144, 0x144, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x148, 0x14b, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x14d, 0x14d, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x152, 0x153, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x166, 0x167, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x16b, 0x16b, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1ce, 0x1ce, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1d0, 0x1d0, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1d2, 0x1d2, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1d4, 0x1d4, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1d6, 0x1d6, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1d8, 0x1d8, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1da, 0x1da, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1dc, 0x1dc, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x251, 0x251, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x261, 0x261, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2c4, 0x2c4, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2c7, 0x2c7, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2c9, 0x2cb, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2cd, 0x2cd, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2d0, 0x2d0, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2d8, 0x2db, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2dd, 0x2dd, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2df, 0x2df, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x300, 0x36f, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x391, 0x3a1, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x3a3, 0x3a9, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x3b1, 0x3c1, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x3c3, 0x3c9, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x401, 0x401, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x410, 0x44f, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x451, 0x451, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1100, 0x115f, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2010, 0x2010, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2013, 0x2016, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2018, 0x2019, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x201c, 0x201d, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2020, 0x2022, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2024, 0x2027, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2030, 0x2030, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2032, 0x2033, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2035, 0x2035, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x203b, 0x203b, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x203e, 0x203e, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2074, 0x2074, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x207f, 0x207f, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2081, 0x2084, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x20ac, 0x20ac, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2103, 0x2103, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2105, 0x2105, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2109, 0x2109, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2113, 0x2113, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2116, 0x2116, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2121, 0x2122, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2126, 0x2126, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x212b, 0x212b, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2153, 0x2154, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x215b, 0x215e, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2160, 0x216b, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2170, 0x2179, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2189, 0x2189, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2190, 0x2199, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x21b8, 0x21b9, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x21d2, 0x21d2, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x21d4, 0x21d4, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x21e7, 0x21e7, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2200, 0x2200, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2202, 0x2203, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2207, 0x2208, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x220b, 0x220b, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x220f, 0x220f, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2211, 0x2211, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2215, 0x2215, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x221a, 0x221a, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x221d, 0x2220, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2223, 0x2223, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2225, 0x2225, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2227, 0x222c, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x222e, 0x222e, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2234, 0x2237, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x223c, 0x223d, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2248, 0x2248, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x224c, 0x224c, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2252, 0x2252, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2260, 0x2261, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2264, 0x2267, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x226a, 0x226b, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x226e, 0x226f, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2282, 0x2283, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2286, 0x2287, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2295, 0x2295, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2299, 0x2299, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x22a5, 0x22a5, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x22bf, 0x22bf, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2312, 0x2312, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x231a, 0x231b, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2329, 0x232a, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x23e9, 0x23ec, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x23f0, 0x23f0, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x23f3, 0x23f3, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2460, 0x24e9, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x24eb, 0x24ff, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2500, 0x259f, CodepointWidth::Narrow }, // box-drawing and block elements require 1-cell alignment
|
||||
UnicodeRange{ 0x25a0, 0x25a1, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x25a3, 0x25a9, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x25b2, 0x25b3, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x25b6, 0x25b7, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x25bc, 0x25bd, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x25c0, 0x25c1, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x25c6, 0x25c8, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x25cb, 0x25cb, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x25ce, 0x25d1, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x25e2, 0x25e5, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x25ef, 0x25ef, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x25fd, 0x25fe, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2605, 0x2606, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2609, 0x2609, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x260e, 0x260f, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2614, 0x2615, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x261c, 0x261c, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x261e, 0x261e, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2640, 0x2640, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2642, 0x2642, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2648, 0x2653, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2660, 0x2661, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2663, 0x2665, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2667, 0x266a, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x266c, 0x266d, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x266f, 0x266f, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x267f, 0x267f, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2693, 0x2693, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x269e, 0x269f, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x26a1, 0x26a1, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x26aa, 0x26ab, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x26bd, 0x26be, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x26bf, 0x26bf, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x26c4, 0x26c5, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x26c6, 0x26cd, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x26ce, 0x26ce, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x26cf, 0x26d3, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x26d4, 0x26d4, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x26d5, 0x26e1, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x26e3, 0x26e3, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x26e8, 0x26e9, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x26ea, 0x26ea, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x26eb, 0x26f1, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x26f2, 0x26f3, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x26f4, 0x26f4, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x26f5, 0x26f5, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x26f6, 0x26f9, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x26fa, 0x26fa, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x26fb, 0x26fc, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x26fd, 0x26fd, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x26fe, 0x26ff, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2705, 0x2705, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x270a, 0x270b, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2728, 0x2728, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x273d, 0x273d, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x274c, 0x274c, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x274e, 0x274e, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2753, 0x2755, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2757, 0x2757, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2776, 0x277f, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2795, 0x2797, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x27b0, 0x27b0, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x27bf, 0x27bf, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2b1b, 0x2b1c, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2b50, 0x2b50, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2b55, 0x2b55, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2b56, 0x2b59, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x2e80, 0x2e99, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2e9b, 0x2ef3, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2f00, 0x2fd5, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x2ff0, 0x2ffb, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x3000, 0x303e, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x3041, 0x3096, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x3099, 0x30ff, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x3105, 0x312f, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x3131, 0x318e, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x3190, 0x31e3, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x31f0, 0x321e, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x3220, 0x3247, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x3248, 0x324f, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x3250, 0x4dbf, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x4dc0, 0x4dff, CodepointWidth::Narrow }, // hexagrams are historically narrow
|
||||
UnicodeRange{ 0x4e00, 0xa48c, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xa490, 0xa4c6, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xa960, 0xa97c, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xac00, 0xd7a3, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xe000, 0xf8ff, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xf900, 0xfaff, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xfe00, 0xfe0f, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xfe10, 0xfe19, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xfe20, 0xfe2f, CodepointWidth::Narrow }, // narrow combining ligatures (split into left/right halves, which take 2 columns together)
|
||||
UnicodeRange{ 0xfe30, 0xfe52, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xfe54, 0xfe66, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xfe68, 0xfe6b, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xff01, 0xff60, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xffe0, 0xffe6, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xfffd, 0xfffd, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x16fe0, 0x16fe4, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x16ff0, 0x16ff1, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x17000, 0x187f7, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x18800, 0x18cd5, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x18d00, 0x18d08, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1aff0, 0x1aff3, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1aff5, 0x1affb, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1affd, 0x1affe, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1b000, 0x1b122, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1b132, 0x1b132, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1b150, 0x1b152, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1b155, 0x1b155, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1b164, 0x1b167, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1b170, 0x1b2fb, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f004, 0x1f004, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f0cf, 0x1f0cf, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f100, 0x1f10a, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1f110, 0x1f12d, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1f130, 0x1f169, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1f170, 0x1f18d, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1f18e, 0x1f18e, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f18f, 0x1f190, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1f191, 0x1f19a, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f19b, 0x1f1ac, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x1f1e6, 0x1f202, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f210, 0x1f23b, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f240, 0x1f248, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f250, 0x1f251, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f260, 0x1f265, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f300, 0x1f320, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f32d, 0x1f335, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f337, 0x1f37c, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f37e, 0x1f393, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f3a0, 0x1f3ca, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f3cf, 0x1f3d3, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f3e0, 0x1f3f0, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f3f4, 0x1f3f4, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f3f8, 0x1f43e, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f440, 0x1f440, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f442, 0x1f4fc, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f4ff, 0x1f53d, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f54b, 0x1f54e, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f550, 0x1f567, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f57a, 0x1f57a, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f595, 0x1f596, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f5a4, 0x1f5a4, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f5fb, 0x1f64f, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f680, 0x1f6c5, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f6cc, 0x1f6cc, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f6d0, 0x1f6d2, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f6d5, 0x1f6d7, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f6dc, 0x1f6df, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f6eb, 0x1f6ec, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f6f4, 0x1f6fc, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f7e0, 0x1f7eb, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f7f0, 0x1f7f0, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f90c, 0x1f93a, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f93c, 0x1f945, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1f947, 0x1f9ff, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1fa70, 0x1fa7c, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1fa80, 0x1fa88, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1fa90, 0x1fabd, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1fabf, 0x1fac5, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1face, 0x1fadb, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1fae0, 0x1fae8, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x1faf0, 0x1faf8, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x20000, 0x2fffd, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0x30000, 0x3fffd, CodepointWidth::Wide },
|
||||
UnicodeRange{ 0xe0100, 0xe01ef, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0xf0000, 0xffffd, CodepointWidth::Ambiguous },
|
||||
UnicodeRange{ 0x100000, 0x10fffd, CodepointWidth::Ambiguous },
|
||||
static constexpr std::array<UnicodeRange, 297> s_wideAndAmbiguousTable{
|
||||
UnicodeRange{ 0xa1, 0xa1, 1 },
|
||||
UnicodeRange{ 0xa4, 0xa4, 1 },
|
||||
UnicodeRange{ 0xa7, 0xa8, 1 },
|
||||
UnicodeRange{ 0xaa, 0xaa, 1 },
|
||||
UnicodeRange{ 0xad, 0xae, 1 },
|
||||
UnicodeRange{ 0xb0, 0xb4, 1 },
|
||||
UnicodeRange{ 0xb6, 0xba, 1 },
|
||||
UnicodeRange{ 0xbc, 0xbf, 1 },
|
||||
UnicodeRange{ 0xc6, 0xc6, 1 },
|
||||
UnicodeRange{ 0xd0, 0xd0, 1 },
|
||||
UnicodeRange{ 0xd7, 0xd8, 1 },
|
||||
UnicodeRange{ 0xde, 0xe1, 1 },
|
||||
UnicodeRange{ 0xe6, 0xe6, 1 },
|
||||
UnicodeRange{ 0xe8, 0xea, 1 },
|
||||
UnicodeRange{ 0xec, 0xed, 1 },
|
||||
UnicodeRange{ 0xf0, 0xf0, 1 },
|
||||
UnicodeRange{ 0xf2, 0xf3, 1 },
|
||||
UnicodeRange{ 0xf7, 0xfa, 1 },
|
||||
UnicodeRange{ 0xfc, 0xfc, 1 },
|
||||
UnicodeRange{ 0xfe, 0xfe, 1 },
|
||||
UnicodeRange{ 0x101, 0x101, 1 },
|
||||
UnicodeRange{ 0x111, 0x111, 1 },
|
||||
UnicodeRange{ 0x113, 0x113, 1 },
|
||||
UnicodeRange{ 0x11b, 0x11b, 1 },
|
||||
UnicodeRange{ 0x126, 0x127, 1 },
|
||||
UnicodeRange{ 0x12b, 0x12b, 1 },
|
||||
UnicodeRange{ 0x131, 0x133, 1 },
|
||||
UnicodeRange{ 0x138, 0x138, 1 },
|
||||
UnicodeRange{ 0x13f, 0x142, 1 },
|
||||
UnicodeRange{ 0x144, 0x144, 1 },
|
||||
UnicodeRange{ 0x148, 0x14b, 1 },
|
||||
UnicodeRange{ 0x14d, 0x14d, 1 },
|
||||
UnicodeRange{ 0x152, 0x153, 1 },
|
||||
UnicodeRange{ 0x166, 0x167, 1 },
|
||||
UnicodeRange{ 0x16b, 0x16b, 1 },
|
||||
UnicodeRange{ 0x1ce, 0x1ce, 1 },
|
||||
UnicodeRange{ 0x1d0, 0x1d0, 1 },
|
||||
UnicodeRange{ 0x1d2, 0x1d2, 1 },
|
||||
UnicodeRange{ 0x1d4, 0x1d4, 1 },
|
||||
UnicodeRange{ 0x1d6, 0x1d6, 1 },
|
||||
UnicodeRange{ 0x1d8, 0x1d8, 1 },
|
||||
UnicodeRange{ 0x1da, 0x1da, 1 },
|
||||
UnicodeRange{ 0x1dc, 0x1dc, 1 },
|
||||
UnicodeRange{ 0x251, 0x251, 1 },
|
||||
UnicodeRange{ 0x261, 0x261, 1 },
|
||||
UnicodeRange{ 0x2c4, 0x2c4, 1 },
|
||||
UnicodeRange{ 0x2c7, 0x2c7, 1 },
|
||||
UnicodeRange{ 0x2c9, 0x2cb, 1 },
|
||||
UnicodeRange{ 0x2cd, 0x2cd, 1 },
|
||||
UnicodeRange{ 0x2d0, 0x2d0, 1 },
|
||||
UnicodeRange{ 0x2d8, 0x2db, 1 },
|
||||
UnicodeRange{ 0x2dd, 0x2dd, 1 },
|
||||
UnicodeRange{ 0x2df, 0x2df, 1 },
|
||||
UnicodeRange{ 0x300, 0x36f, 1 },
|
||||
UnicodeRange{ 0x391, 0x3a1, 1 },
|
||||
UnicodeRange{ 0x3a3, 0x3a9, 1 },
|
||||
UnicodeRange{ 0x3b1, 0x3c1, 1 },
|
||||
UnicodeRange{ 0x3c3, 0x3c9, 1 },
|
||||
UnicodeRange{ 0x401, 0x401, 1 },
|
||||
UnicodeRange{ 0x410, 0x44f, 1 },
|
||||
UnicodeRange{ 0x451, 0x451, 1 },
|
||||
UnicodeRange{ 0x1100, 0x115f, 0 },
|
||||
UnicodeRange{ 0x2010, 0x2010, 1 },
|
||||
UnicodeRange{ 0x2013, 0x2016, 1 },
|
||||
UnicodeRange{ 0x2018, 0x2019, 1 },
|
||||
UnicodeRange{ 0x201c, 0x201d, 1 },
|
||||
UnicodeRange{ 0x2020, 0x2022, 1 },
|
||||
UnicodeRange{ 0x2024, 0x2027, 1 },
|
||||
UnicodeRange{ 0x2030, 0x2030, 1 },
|
||||
UnicodeRange{ 0x2032, 0x2033, 1 },
|
||||
UnicodeRange{ 0x2035, 0x2035, 1 },
|
||||
UnicodeRange{ 0x203b, 0x203b, 1 },
|
||||
UnicodeRange{ 0x203e, 0x203e, 1 },
|
||||
UnicodeRange{ 0x2074, 0x2074, 1 },
|
||||
UnicodeRange{ 0x207f, 0x207f, 1 },
|
||||
UnicodeRange{ 0x2081, 0x2084, 1 },
|
||||
UnicodeRange{ 0x20ac, 0x20ac, 1 },
|
||||
UnicodeRange{ 0x2103, 0x2103, 1 },
|
||||
UnicodeRange{ 0x2105, 0x2105, 1 },
|
||||
UnicodeRange{ 0x2109, 0x2109, 1 },
|
||||
UnicodeRange{ 0x2113, 0x2113, 1 },
|
||||
UnicodeRange{ 0x2116, 0x2116, 1 },
|
||||
UnicodeRange{ 0x2121, 0x2122, 1 },
|
||||
UnicodeRange{ 0x2126, 0x2126, 1 },
|
||||
UnicodeRange{ 0x212b, 0x212b, 1 },
|
||||
UnicodeRange{ 0x2153, 0x2154, 1 },
|
||||
UnicodeRange{ 0x215b, 0x215e, 1 },
|
||||
UnicodeRange{ 0x2160, 0x216b, 1 },
|
||||
UnicodeRange{ 0x2170, 0x2179, 1 },
|
||||
UnicodeRange{ 0x2189, 0x2189, 1 },
|
||||
UnicodeRange{ 0x2190, 0x2199, 1 },
|
||||
UnicodeRange{ 0x21b8, 0x21b9, 1 },
|
||||
UnicodeRange{ 0x21d2, 0x21d2, 1 },
|
||||
UnicodeRange{ 0x21d4, 0x21d4, 1 },
|
||||
UnicodeRange{ 0x21e7, 0x21e7, 1 },
|
||||
UnicodeRange{ 0x2200, 0x2200, 1 },
|
||||
UnicodeRange{ 0x2202, 0x2203, 1 },
|
||||
UnicodeRange{ 0x2207, 0x2208, 1 },
|
||||
UnicodeRange{ 0x220b, 0x220b, 1 },
|
||||
UnicodeRange{ 0x220f, 0x220f, 1 },
|
||||
UnicodeRange{ 0x2211, 0x2211, 1 },
|
||||
UnicodeRange{ 0x2215, 0x2215, 1 },
|
||||
UnicodeRange{ 0x221a, 0x221a, 1 },
|
||||
UnicodeRange{ 0x221d, 0x2220, 1 },
|
||||
UnicodeRange{ 0x2223, 0x2223, 1 },
|
||||
UnicodeRange{ 0x2225, 0x2225, 1 },
|
||||
UnicodeRange{ 0x2227, 0x222c, 1 },
|
||||
UnicodeRange{ 0x222e, 0x222e, 1 },
|
||||
UnicodeRange{ 0x2234, 0x2237, 1 },
|
||||
UnicodeRange{ 0x223c, 0x223d, 1 },
|
||||
UnicodeRange{ 0x2248, 0x2248, 1 },
|
||||
UnicodeRange{ 0x224c, 0x224c, 1 },
|
||||
UnicodeRange{ 0x2252, 0x2252, 1 },
|
||||
UnicodeRange{ 0x2260, 0x2261, 1 },
|
||||
UnicodeRange{ 0x2264, 0x2267, 1 },
|
||||
UnicodeRange{ 0x226a, 0x226b, 1 },
|
||||
UnicodeRange{ 0x226e, 0x226f, 1 },
|
||||
UnicodeRange{ 0x2282, 0x2283, 1 },
|
||||
UnicodeRange{ 0x2286, 0x2287, 1 },
|
||||
UnicodeRange{ 0x2295, 0x2295, 1 },
|
||||
UnicodeRange{ 0x2299, 0x2299, 1 },
|
||||
UnicodeRange{ 0x22a5, 0x22a5, 1 },
|
||||
UnicodeRange{ 0x22bf, 0x22bf, 1 },
|
||||
UnicodeRange{ 0x2312, 0x2312, 1 },
|
||||
UnicodeRange{ 0x231a, 0x231b, 0 },
|
||||
UnicodeRange{ 0x2329, 0x232a, 0 },
|
||||
UnicodeRange{ 0x23e9, 0x23ec, 0 },
|
||||
UnicodeRange{ 0x23f0, 0x23f0, 0 },
|
||||
UnicodeRange{ 0x23f3, 0x23f3, 0 },
|
||||
UnicodeRange{ 0x2460, 0x24e9, 1 },
|
||||
UnicodeRange{ 0x24eb, 0x24ff, 1 },
|
||||
UnicodeRange{ 0x25a0, 0x25a1, 1 },
|
||||
UnicodeRange{ 0x25a3, 0x25a9, 1 },
|
||||
UnicodeRange{ 0x25b2, 0x25b3, 1 },
|
||||
UnicodeRange{ 0x25b6, 0x25b7, 1 },
|
||||
UnicodeRange{ 0x25bc, 0x25bd, 1 },
|
||||
UnicodeRange{ 0x25c0, 0x25c1, 1 },
|
||||
UnicodeRange{ 0x25c6, 0x25c8, 1 },
|
||||
UnicodeRange{ 0x25cb, 0x25cb, 1 },
|
||||
UnicodeRange{ 0x25ce, 0x25d1, 1 },
|
||||
UnicodeRange{ 0x25e2, 0x25e5, 1 },
|
||||
UnicodeRange{ 0x25ef, 0x25ef, 1 },
|
||||
UnicodeRange{ 0x25fd, 0x25fe, 0 },
|
||||
UnicodeRange{ 0x2605, 0x2606, 1 },
|
||||
UnicodeRange{ 0x2609, 0x2609, 1 },
|
||||
UnicodeRange{ 0x260e, 0x260f, 1 },
|
||||
UnicodeRange{ 0x2614, 0x2615, 0 },
|
||||
UnicodeRange{ 0x261c, 0x261c, 1 },
|
||||
UnicodeRange{ 0x261e, 0x261e, 1 },
|
||||
UnicodeRange{ 0x2640, 0x2640, 1 },
|
||||
UnicodeRange{ 0x2642, 0x2642, 1 },
|
||||
UnicodeRange{ 0x2648, 0x2653, 0 },
|
||||
UnicodeRange{ 0x2660, 0x2661, 1 },
|
||||
UnicodeRange{ 0x2663, 0x2665, 1 },
|
||||
UnicodeRange{ 0x2667, 0x266a, 1 },
|
||||
UnicodeRange{ 0x266c, 0x266d, 1 },
|
||||
UnicodeRange{ 0x266f, 0x266f, 1 },
|
||||
UnicodeRange{ 0x267f, 0x267f, 0 },
|
||||
UnicodeRange{ 0x2693, 0x2693, 0 },
|
||||
UnicodeRange{ 0x269e, 0x269f, 1 },
|
||||
UnicodeRange{ 0x26a1, 0x26a1, 0 },
|
||||
UnicodeRange{ 0x26aa, 0x26ab, 0 },
|
||||
UnicodeRange{ 0x26bd, 0x26be, 0 },
|
||||
UnicodeRange{ 0x26bf, 0x26bf, 1 },
|
||||
UnicodeRange{ 0x26c4, 0x26c5, 0 },
|
||||
UnicodeRange{ 0x26c6, 0x26cd, 1 },
|
||||
UnicodeRange{ 0x26ce, 0x26ce, 0 },
|
||||
UnicodeRange{ 0x26cf, 0x26d3, 1 },
|
||||
UnicodeRange{ 0x26d4, 0x26d4, 0 },
|
||||
UnicodeRange{ 0x26d5, 0x26e1, 1 },
|
||||
UnicodeRange{ 0x26e3, 0x26e3, 1 },
|
||||
UnicodeRange{ 0x26e8, 0x26e9, 1 },
|
||||
UnicodeRange{ 0x26ea, 0x26ea, 0 },
|
||||
UnicodeRange{ 0x26eb, 0x26f1, 1 },
|
||||
UnicodeRange{ 0x26f2, 0x26f3, 0 },
|
||||
UnicodeRange{ 0x26f4, 0x26f4, 1 },
|
||||
UnicodeRange{ 0x26f5, 0x26f5, 0 },
|
||||
UnicodeRange{ 0x26f6, 0x26f9, 1 },
|
||||
UnicodeRange{ 0x26fa, 0x26fa, 0 },
|
||||
UnicodeRange{ 0x26fb, 0x26fc, 1 },
|
||||
UnicodeRange{ 0x26fd, 0x26fd, 0 },
|
||||
UnicodeRange{ 0x26fe, 0x26ff, 1 },
|
||||
UnicodeRange{ 0x2705, 0x2705, 0 },
|
||||
UnicodeRange{ 0x270a, 0x270b, 0 },
|
||||
UnicodeRange{ 0x2728, 0x2728, 0 },
|
||||
UnicodeRange{ 0x273d, 0x273d, 1 },
|
||||
UnicodeRange{ 0x274c, 0x274c, 0 },
|
||||
UnicodeRange{ 0x274e, 0x274e, 0 },
|
||||
UnicodeRange{ 0x2753, 0x2755, 0 },
|
||||
UnicodeRange{ 0x2757, 0x2757, 0 },
|
||||
UnicodeRange{ 0x2776, 0x277f, 1 },
|
||||
UnicodeRange{ 0x2795, 0x2797, 0 },
|
||||
UnicodeRange{ 0x27b0, 0x27b0, 0 },
|
||||
UnicodeRange{ 0x27bf, 0x27bf, 0 },
|
||||
UnicodeRange{ 0x2b1b, 0x2b1c, 0 },
|
||||
UnicodeRange{ 0x2b50, 0x2b50, 0 },
|
||||
UnicodeRange{ 0x2b55, 0x2b55, 0 },
|
||||
UnicodeRange{ 0x2b56, 0x2b59, 1 },
|
||||
UnicodeRange{ 0x2e80, 0x2e99, 0 },
|
||||
UnicodeRange{ 0x2e9b, 0x2ef3, 0 },
|
||||
UnicodeRange{ 0x2f00, 0x2fd5, 0 },
|
||||
UnicodeRange{ 0x2ff0, 0x2ffb, 0 },
|
||||
UnicodeRange{ 0x3000, 0x303e, 0 },
|
||||
UnicodeRange{ 0x3041, 0x3096, 0 },
|
||||
UnicodeRange{ 0x3099, 0x30ff, 0 },
|
||||
UnicodeRange{ 0x3105, 0x312f, 0 },
|
||||
UnicodeRange{ 0x3131, 0x318e, 0 },
|
||||
UnicodeRange{ 0x3190, 0x31e3, 0 },
|
||||
UnicodeRange{ 0x31f0, 0x321e, 0 },
|
||||
UnicodeRange{ 0x3220, 0x3247, 0 },
|
||||
UnicodeRange{ 0x3248, 0x324f, 1 },
|
||||
UnicodeRange{ 0x3250, 0x4dbf, 0 },
|
||||
UnicodeRange{ 0x4e00, 0xa48c, 0 },
|
||||
UnicodeRange{ 0xa490, 0xa4c6, 0 },
|
||||
UnicodeRange{ 0xa960, 0xa97c, 0 },
|
||||
UnicodeRange{ 0xac00, 0xd7a3, 0 },
|
||||
UnicodeRange{ 0xe000, 0xf8ff, 1 },
|
||||
UnicodeRange{ 0xf900, 0xfaff, 0 },
|
||||
UnicodeRange{ 0xfe00, 0xfe0f, 1 },
|
||||
UnicodeRange{ 0xfe10, 0xfe19, 0 },
|
||||
UnicodeRange{ 0xfe30, 0xfe52, 0 },
|
||||
UnicodeRange{ 0xfe54, 0xfe66, 0 },
|
||||
UnicodeRange{ 0xfe68, 0xfe6b, 0 },
|
||||
UnicodeRange{ 0xff01, 0xff60, 0 },
|
||||
UnicodeRange{ 0xffe0, 0xffe6, 0 },
|
||||
UnicodeRange{ 0xfffd, 0xfffd, 1 },
|
||||
UnicodeRange{ 0x16fe0, 0x16fe4, 0 },
|
||||
UnicodeRange{ 0x16ff0, 0x16ff1, 0 },
|
||||
UnicodeRange{ 0x17000, 0x187f7, 0 },
|
||||
UnicodeRange{ 0x18800, 0x18cd5, 0 },
|
||||
UnicodeRange{ 0x18d00, 0x18d08, 0 },
|
||||
UnicodeRange{ 0x1aff0, 0x1aff3, 0 },
|
||||
UnicodeRange{ 0x1aff5, 0x1affb, 0 },
|
||||
UnicodeRange{ 0x1affd, 0x1affe, 0 },
|
||||
UnicodeRange{ 0x1b000, 0x1b122, 0 },
|
||||
UnicodeRange{ 0x1b132, 0x1b132, 0 },
|
||||
UnicodeRange{ 0x1b150, 0x1b152, 0 },
|
||||
UnicodeRange{ 0x1b155, 0x1b155, 0 },
|
||||
UnicodeRange{ 0x1b164, 0x1b167, 0 },
|
||||
UnicodeRange{ 0x1b170, 0x1b2fb, 0 },
|
||||
UnicodeRange{ 0x1f004, 0x1f004, 0 },
|
||||
UnicodeRange{ 0x1f0cf, 0x1f0cf, 0 },
|
||||
UnicodeRange{ 0x1f100, 0x1f10a, 1 },
|
||||
UnicodeRange{ 0x1f110, 0x1f12d, 1 },
|
||||
UnicodeRange{ 0x1f130, 0x1f169, 1 },
|
||||
UnicodeRange{ 0x1f170, 0x1f18d, 1 },
|
||||
UnicodeRange{ 0x1f18e, 0x1f18e, 0 },
|
||||
UnicodeRange{ 0x1f18f, 0x1f190, 1 },
|
||||
UnicodeRange{ 0x1f191, 0x1f19a, 0 },
|
||||
UnicodeRange{ 0x1f19b, 0x1f1ac, 1 },
|
||||
UnicodeRange{ 0x1f1e6, 0x1f202, 0 },
|
||||
UnicodeRange{ 0x1f210, 0x1f23b, 0 },
|
||||
UnicodeRange{ 0x1f240, 0x1f248, 0 },
|
||||
UnicodeRange{ 0x1f250, 0x1f251, 0 },
|
||||
UnicodeRange{ 0x1f260, 0x1f265, 0 },
|
||||
UnicodeRange{ 0x1f300, 0x1f320, 0 },
|
||||
UnicodeRange{ 0x1f32d, 0x1f335, 0 },
|
||||
UnicodeRange{ 0x1f337, 0x1f37c, 0 },
|
||||
UnicodeRange{ 0x1f37e, 0x1f393, 0 },
|
||||
UnicodeRange{ 0x1f3a0, 0x1f3ca, 0 },
|
||||
UnicodeRange{ 0x1f3cf, 0x1f3d3, 0 },
|
||||
UnicodeRange{ 0x1f3e0, 0x1f3f0, 0 },
|
||||
UnicodeRange{ 0x1f3f4, 0x1f3f4, 0 },
|
||||
UnicodeRange{ 0x1f3f8, 0x1f43e, 0 },
|
||||
UnicodeRange{ 0x1f440, 0x1f440, 0 },
|
||||
UnicodeRange{ 0x1f442, 0x1f4fc, 0 },
|
||||
UnicodeRange{ 0x1f4ff, 0x1f53d, 0 },
|
||||
UnicodeRange{ 0x1f54b, 0x1f54e, 0 },
|
||||
UnicodeRange{ 0x1f550, 0x1f567, 0 },
|
||||
UnicodeRange{ 0x1f57a, 0x1f57a, 0 },
|
||||
UnicodeRange{ 0x1f595, 0x1f596, 0 },
|
||||
UnicodeRange{ 0x1f5a4, 0x1f5a4, 0 },
|
||||
UnicodeRange{ 0x1f5fb, 0x1f64f, 0 },
|
||||
UnicodeRange{ 0x1f680, 0x1f6c5, 0 },
|
||||
UnicodeRange{ 0x1f6cc, 0x1f6cc, 0 },
|
||||
UnicodeRange{ 0x1f6d0, 0x1f6d2, 0 },
|
||||
UnicodeRange{ 0x1f6d5, 0x1f6d7, 0 },
|
||||
UnicodeRange{ 0x1f6dc, 0x1f6df, 0 },
|
||||
UnicodeRange{ 0x1f6eb, 0x1f6ec, 0 },
|
||||
UnicodeRange{ 0x1f6f4, 0x1f6fc, 0 },
|
||||
UnicodeRange{ 0x1f7e0, 0x1f7eb, 0 },
|
||||
UnicodeRange{ 0x1f7f0, 0x1f7f0, 0 },
|
||||
UnicodeRange{ 0x1f90c, 0x1f93a, 0 },
|
||||
UnicodeRange{ 0x1f93c, 0x1f945, 0 },
|
||||
UnicodeRange{ 0x1f947, 0x1f9ff, 0 },
|
||||
UnicodeRange{ 0x1fa70, 0x1fa7c, 0 },
|
||||
UnicodeRange{ 0x1fa80, 0x1fa88, 0 },
|
||||
UnicodeRange{ 0x1fa90, 0x1fabd, 0 },
|
||||
UnicodeRange{ 0x1fabf, 0x1fac5, 0 },
|
||||
UnicodeRange{ 0x1face, 0x1fadb, 0 },
|
||||
UnicodeRange{ 0x1fae0, 0x1fae8, 0 },
|
||||
UnicodeRange{ 0x1faf0, 0x1faf8, 0 },
|
||||
UnicodeRange{ 0x20000, 0x2fffd, 0 },
|
||||
UnicodeRange{ 0x30000, 0x3fffd, 0 },
|
||||
UnicodeRange{ 0xe0100, 0xe01ef, 1 },
|
||||
UnicodeRange{ 0xf0000, 0xffffd, 1 },
|
||||
UnicodeRange{ 0x100000, 0x10fffd, 1 },
|
||||
};
|
||||
}
|
||||
|
||||
// Routine Description:
|
||||
// - Constructs an instance of the CodepointWidthDetector class
|
||||
CodepointWidthDetector::CodepointWidthDetector() noexcept :
|
||||
_fallbackCache{},
|
||||
_pfnFallbackMethod{}
|
||||
{
|
||||
}
|
||||
|
||||
// Routine Description:
|
||||
// - returns the width type of codepoint as fast as we can by using quick lookup table and fallback cache.
|
||||
// Arguments:
|
||||
// - glyph - the utf16 encoded codepoint to search for
|
||||
// Return Value:
|
||||
// - the width type of the codepoint
|
||||
CodepointWidth CodepointWidthDetector::GetWidth(const std::wstring_view glyph) const
|
||||
CodepointWidth CodepointWidthDetector::GetWidth(const std::wstring_view& glyph) noexcept
|
||||
{
|
||||
THROW_HR_IF(E_INVALIDARG, glyph.empty());
|
||||
if (glyph.size() == 1)
|
||||
{
|
||||
// We first attempt to look at our custom quick lookup table of char width preferences.
|
||||
const auto width = GetQuickCharWidth(glyph.front());
|
||||
char32_t codepoint = 0;
|
||||
|
||||
// If it's invalid, the quick width had no opinion, so go to the lookup table.
|
||||
if (width == CodepointWidth::Invalid)
|
||||
{
|
||||
return _lookupGlyphWidthWithCache(glyph);
|
||||
}
|
||||
// If it's ambiguous, the quick width wanted us to ask the font directly, try that if we can.
|
||||
// If not, go to the lookup table.
|
||||
else if (width == CodepointWidth::Ambiguous)
|
||||
{
|
||||
if (_pfnFallbackMethod)
|
||||
{
|
||||
return _checkFallbackViaCache(glyph) ? CodepointWidth::Wide : CodepointWidth::Ambiguous;
|
||||
}
|
||||
else
|
||||
{
|
||||
return _lookupGlyphWidthWithCache(glyph);
|
||||
}
|
||||
}
|
||||
// Otherwise, return Width as it is.
|
||||
else
|
||||
{
|
||||
return width;
|
||||
}
|
||||
}
|
||||
else
|
||||
switch (glyph.size())
|
||||
{
|
||||
return _lookupGlyphWidthWithCache(glyph);
|
||||
case 1:
|
||||
codepoint = til::at(glyph, 0);
|
||||
break;
|
||||
case 2:
|
||||
codepoint = (til::at(glyph, 0) & 0x3FF) << 10;
|
||||
codepoint |= til::at(glyph, 1) & 0x3FF;
|
||||
codepoint += 0x10000;
|
||||
break;
|
||||
default:
|
||||
codepoint = 0;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
// Routine Description:
|
||||
// - checks if wch is wide. will attempt to fallback as much possible until an answer is determined
|
||||
// Arguments:
|
||||
// - wch - the wchar to check width of
|
||||
// Return Value:
|
||||
// - true if wch is wide
|
||||
bool CodepointWidthDetector::IsWide(const wchar_t wch) const noexcept
|
||||
{
|
||||
try
|
||||
if (codepoint < 0x80)
|
||||
{
|
||||
return IsWide({ &wch, 1 });
|
||||
return CodepointWidth::Narrow;
|
||||
}
|
||||
CATCH_LOG();
|
||||
|
||||
return true;
|
||||
// The return value of _lookupGlyphWidth coincides with the enum value of CodepointWidth
|
||||
// on purpose to allow for this easy conversion to happen. Optimally, we should probably
|
||||
// remove CodepointWidth altogether to allow for zero-width joiners and other characters.
|
||||
static_assert(WI_EnumValue(CodepointWidth::Narrow) == 1);
|
||||
static_assert(WI_EnumValue(CodepointWidth::Wide) == 2);
|
||||
return static_cast<CodepointWidth>(_lookupGlyphWidth(codepoint, glyph));
|
||||
}
|
||||
|
||||
// Routine Description:
|
||||
|
|
@ -403,123 +369,54 @@ bool CodepointWidthDetector::IsWide(const wchar_t wch) const noexcept
|
|||
// - glyph - the utf16 encoded codepoint to check width of
|
||||
// Return Value:
|
||||
// - true if codepoint is wide
|
||||
bool CodepointWidthDetector::IsWide(const std::wstring_view glyph) const
|
||||
bool CodepointWidthDetector::IsWide(const std::wstring_view& glyph) noexcept
|
||||
{
|
||||
return GetWidth(glyph) == CodepointWidth::Wide;
|
||||
}
|
||||
|
||||
// Routine Description:
|
||||
// - returns the width type of codepoint by searching the map generated from the unicode spec
|
||||
// Arguments:
|
||||
// - glyph - the utf16 encoded codepoint to search for
|
||||
// Return Value:
|
||||
// - the width type of the codepoint
|
||||
CodepointWidth CodepointWidthDetector::_lookupGlyphWidth(const std::wstring_view glyph) const
|
||||
// GetWidth's slow-path for non-ASCII characters. Returns the number of columns the codepoint takes up in the terminal.
|
||||
uint8_t CodepointWidthDetector::_lookupGlyphWidth(const char32_t codepoint, const std::wstring_view& glyph) noexcept
|
||||
{
|
||||
if (glyph.empty())
|
||||
{
|
||||
return CodepointWidth::Invalid;
|
||||
}
|
||||
|
||||
const auto codepoint = _extractCodepoint(glyph);
|
||||
#pragma warning(suppress : 26447) // The function is declared 'noexcept' but calls function 'lower_bound<...>()' which may throw exceptions (f.6).
|
||||
const auto it = std::lower_bound(s_wideAndAmbiguousTable.begin(), s_wideAndAmbiguousTable.end(), codepoint);
|
||||
uint8_t width = 1;
|
||||
|
||||
// For characters that are not _in_ the table, lower_bound will return the nearest item that is.
|
||||
// We must check its bounds to make sure that our hit was a true hit.
|
||||
if (it != s_wideAndAmbiguousTable.end() && codepoint >= it->lowerBound && codepoint <= it->upperBound)
|
||||
{
|
||||
return it->width;
|
||||
}
|
||||
|
||||
return CodepointWidth::Narrow;
|
||||
}
|
||||
|
||||
// Routine Description:
|
||||
// - returns the width type of codepoint using fallback methods.
|
||||
// Arguments:
|
||||
// - glyph - the utf16 encoded codepoint to check width of
|
||||
// Return Value:
|
||||
// - the width type of the codepoint
|
||||
CodepointWidth CodepointWidthDetector::_lookupGlyphWidthWithCache(const std::wstring_view glyph) const noexcept
|
||||
{
|
||||
try
|
||||
{
|
||||
// Use our generated table to try to lookup the width based on the Unicode standard.
|
||||
const auto width = _lookupGlyphWidth(glyph);
|
||||
|
||||
// If it's ambiguous, then ask the font if we can.
|
||||
if (width == CodepointWidth::Ambiguous)
|
||||
width = 2;
|
||||
if (it->isAmbiguous)
|
||||
{
|
||||
if (_pfnFallbackMethod)
|
||||
{
|
||||
return _checkFallbackViaCache(glyph) ? CodepointWidth::Wide : CodepointWidth::Ambiguous;
|
||||
}
|
||||
else
|
||||
{
|
||||
return CodepointWidth::Ambiguous;
|
||||
}
|
||||
}
|
||||
// If it's not ambiguous, it should say wide or narrow.
|
||||
else
|
||||
{
|
||||
return width;
|
||||
width = _checkFallbackViaCache(codepoint, glyph);
|
||||
}
|
||||
}
|
||||
CATCH_LOG();
|
||||
|
||||
// If we got this far, we couldn't figure it out.
|
||||
// It's better to be too wide than too narrow.
|
||||
return CodepointWidth::Wide;
|
||||
return width;
|
||||
}
|
||||
|
||||
// Routine Description:
|
||||
// - Checks the fallback function but caches the results until the font changes
|
||||
// because the lookup function is usually very expensive and will return the same results
|
||||
// for the same inputs.
|
||||
// Arguments:
|
||||
// - glyph - the utf16 encoded codepoint to check width of
|
||||
// - true if codepoint is wide or false if it is narrow
|
||||
bool CodepointWidthDetector::_checkFallbackViaCache(const std::wstring_view glyph) const
|
||||
// Call the function specified via SetFallbackMethod() to turn CodepointWidth::Ambiguous into Narrow/Wide.
|
||||
// Caches the results in _fallbackCache. This is _lookupGlyphWidth's even-slower-path.
|
||||
uint8_t CodepointWidthDetector::_checkFallbackViaCache(const char32_t codepoint, const std::wstring_view& glyph) noexcept
|
||||
try
|
||||
{
|
||||
const std::wstring findMe{ glyph };
|
||||
|
||||
// TODO: Cache needs to be emptied when font changes.
|
||||
const auto it = _fallbackCache.find(findMe);
|
||||
if (it == _fallbackCache.end())
|
||||
// Ambiguous glyphs are considered narrow by default. See microsoft/terminal#2066 for more info.
|
||||
if (!_pfnFallbackMethod)
|
||||
{
|
||||
auto result = _pfnFallbackMethod(glyph);
|
||||
_fallbackCache.insert_or_assign(findMe, result);
|
||||
return result;
|
||||
return 1;
|
||||
}
|
||||
else
|
||||
|
||||
if (const auto it = _fallbackCache.find(codepoint); it != _fallbackCache.end())
|
||||
{
|
||||
return it->second;
|
||||
}
|
||||
}
|
||||
|
||||
// Routine Description:
|
||||
// - extract unicode codepoint from utf16 encoding
|
||||
// Arguments:
|
||||
// - glyph - the utf16 encoded codepoint convert
|
||||
// Return Value:
|
||||
// - the codepoint being stored
|
||||
unsigned int CodepointWidthDetector::_extractCodepoint(const std::wstring_view glyph) noexcept
|
||||
const uint8_t width = _pfnFallbackMethod(glyph) ? 2 : 1;
|
||||
_fallbackCache.insert_or_assign(codepoint, width);
|
||||
return width;
|
||||
}
|
||||
catch (...)
|
||||
{
|
||||
if (glyph.size() == 1)
|
||||
{
|
||||
return static_cast<unsigned int>(glyph.front());
|
||||
}
|
||||
else
|
||||
{
|
||||
const unsigned int mask = 0x3FF;
|
||||
// leading bits, shifted over to make space for trailing bits
|
||||
auto codepoint = (glyph.at(0) & mask) << 10;
|
||||
// trailing bits
|
||||
codepoint |= (glyph.at(1) & mask);
|
||||
// 0x10000 is subtracted from the codepoint to encode a surrogate pair, add it back
|
||||
codepoint += 0x10000;
|
||||
return codepoint;
|
||||
}
|
||||
LOG_CAUGHT_EXCEPTION();
|
||||
return 1;
|
||||
}
|
||||
|
||||
// Method Description:
|
||||
|
|
@ -532,9 +429,9 @@ unsigned int CodepointWidthDetector::_extractCodepoint(const std::wstring_view g
|
|||
// - pfnFallback - the function to use as the fallback method.
|
||||
// Return Value:
|
||||
// - <none>
|
||||
void CodepointWidthDetector::SetFallbackMethod(std::function<bool(const std::wstring_view)> pfnFallback)
|
||||
void CodepointWidthDetector::SetFallbackMethod(std::function<bool(const std::wstring_view&)> pfnFallback) noexcept
|
||||
{
|
||||
_pfnFallbackMethod = pfnFallback;
|
||||
_pfnFallbackMethod = std::move(pfnFallback);
|
||||
}
|
||||
|
||||
// Method Description:
|
||||
|
|
@ -545,7 +442,8 @@ void CodepointWidthDetector::SetFallbackMethod(std::function<bool(const std::wst
|
|||
// - <none>
|
||||
// Return Value:
|
||||
// - <none>
|
||||
void CodepointWidthDetector::NotifyFontChanged() const noexcept
|
||||
void CodepointWidthDetector::NotifyFontChanged() noexcept
|
||||
{
|
||||
#pragma warning(suppress : 26447) // The function is declared 'noexcept' but calls function 'clear()' which may throw exceptions (f.6).
|
||||
_fallbackCache.clear();
|
||||
}
|
||||
|
|
|
|||
|
|
@ -12,7 +12,7 @@ static CodepointWidthDetector widthDetector;
|
|||
// Function Description:
|
||||
// - determines if the glyph represented by the string of characters should be
|
||||
// wide or not. See CodepointWidthDetector::IsWide
|
||||
bool IsGlyphFullWidth(const std::wstring_view glyph)
|
||||
bool IsGlyphFullWidth(const std::wstring_view& glyph) noexcept
|
||||
{
|
||||
return widthDetector.IsWide(glyph);
|
||||
}
|
||||
|
|
@ -22,7 +22,7 @@ bool IsGlyphFullWidth(const std::wstring_view glyph)
|
|||
// wide or not. See CodepointWidthDetector::IsWide
|
||||
bool IsGlyphFullWidth(const wchar_t wch) noexcept
|
||||
{
|
||||
return widthDetector.IsWide(wch);
|
||||
return wch < 0x80 ? false : IsGlyphFullWidth({ &wch, 1 });
|
||||
}
|
||||
|
||||
// Function Description:
|
||||
|
|
@ -35,9 +35,9 @@ bool IsGlyphFullWidth(const wchar_t wch) noexcept
|
|||
// - pfnFallback - the function to use as the fallback method.
|
||||
// Return Value:
|
||||
// - <none>
|
||||
void SetGlyphWidthFallback(std::function<bool(const std::wstring_view)> pfnFallback)
|
||||
void SetGlyphWidthFallback(std::function<bool(const std::wstring_view&)> pfnFallback) noexcept
|
||||
{
|
||||
widthDetector.SetFallbackMethod(pfnFallback);
|
||||
widthDetector.SetFallbackMethod(std::move(pfnFallback));
|
||||
}
|
||||
|
||||
// Function Description:
|
||||
|
|
|
|||
|
|
@ -130,40 +130,6 @@
|
|||
return cchTarget;
|
||||
}
|
||||
|
||||
// Routine Description:
|
||||
// - naively determines the width of a UCS2 encoded wchar
|
||||
// Arguments:
|
||||
// - wch - the wchar_t to measure
|
||||
// Return Value:
|
||||
// - CodepointWidth indicating width of wch
|
||||
// Notes:
|
||||
// 04-08-92 ShunK Created.
|
||||
// Jul-27-1992 KazuM Added Screen Information and Code Page Information.
|
||||
// Jan-29-1992 V-Hirots Substruct Screen Information.
|
||||
// Oct-06-1996 KazuM Not use RtlUnicodeToMultiByteSize and WideCharToMultiByte
|
||||
// Because 950 (Chinese Traditional) only defined 13500 chars,
|
||||
// and unicode defined almost 18000 chars.
|
||||
// So there are almost 4000 chars can not be mapped to big5 code.
|
||||
// Apr-30-2015 MiNiksa Corrected unknown character code assumption. Max Width in Text Metric
|
||||
// is not reliable for calculating half/full width. Must use current
|
||||
// display font data (cached) instead.
|
||||
// May-23-2017 migrie Forced Box-Drawing Characters (x2500-x257F) to narrow.
|
||||
// Jan-16-2018 migrie Separated core lookup from asking the renderer the width
|
||||
// May-01-2019 MiNiksa Forced lookup-via-renderer for retroactively recategorized emoji
|
||||
// that used to be narrow but now might be wide. (approx x2194-x2b55, not inclusive)
|
||||
// Also forced block characters segment (x2580-x259F) to narrow
|
||||
// Oct-25-2020 DuHowett Replaced the entire table with a set of overrides that get built into
|
||||
// CodepointWidthDetector (unicode_width_overrides.xml)
|
||||
CodepointWidth GetQuickCharWidth(const wchar_t wch) noexcept
|
||||
{
|
||||
if (0x20 <= wch && wch <= 0x7e)
|
||||
{
|
||||
/* ASCII */
|
||||
return CodepointWidth::Narrow;
|
||||
}
|
||||
return CodepointWidth::Invalid;
|
||||
}
|
||||
|
||||
wchar_t Utf16ToUcs2(const std::wstring_view charData)
|
||||
{
|
||||
THROW_HR_IF(E_INVALIDARG, charData.empty());
|
||||
|
|
|
|||
|
|
@ -14,38 +14,24 @@ Author:
|
|||
#pragma once
|
||||
|
||||
#include "convert.hpp"
|
||||
#include <functional>
|
||||
|
||||
static_assert(sizeof(unsigned int) == sizeof(wchar_t) * 2,
|
||||
"UnicodeRange expects to be able to store a unicode codepoint in an unsigned int");
|
||||
|
||||
// use to measure the width of a codepoint
|
||||
class CodepointWidthDetector final
|
||||
{
|
||||
public:
|
||||
CodepointWidthDetector() noexcept;
|
||||
CodepointWidthDetector(const CodepointWidthDetector&) = delete;
|
||||
CodepointWidthDetector(CodepointWidthDetector&&) = delete;
|
||||
~CodepointWidthDetector() = default;
|
||||
CodepointWidthDetector& operator=(const CodepointWidthDetector&) = delete;
|
||||
CodepointWidthDetector& operator=(CodepointWidthDetector&&) = delete;
|
||||
|
||||
CodepointWidth GetWidth(const std::wstring_view glyph) const;
|
||||
bool IsWide(const std::wstring_view glyph) const;
|
||||
bool IsWide(const wchar_t wch) const noexcept;
|
||||
void SetFallbackMethod(std::function<bool(const std::wstring_view)> pfnFallback);
|
||||
void NotifyFontChanged() const noexcept;
|
||||
CodepointWidth GetWidth(const std::wstring_view& glyph) noexcept;
|
||||
bool IsWide(const std::wstring_view& glyph) noexcept;
|
||||
void SetFallbackMethod(std::function<bool(const std::wstring_view&)> pfnFallback) noexcept;
|
||||
void NotifyFontChanged() noexcept;
|
||||
|
||||
#ifdef UNIT_TESTING
|
||||
friend class CodepointWidthDetectorTests;
|
||||
#endif
|
||||
|
||||
private:
|
||||
CodepointWidth _lookupGlyphWidth(const std::wstring_view glyph) const;
|
||||
CodepointWidth _lookupGlyphWidthWithCache(const std::wstring_view glyph) const noexcept;
|
||||
bool _checkFallbackViaCache(const std::wstring_view glyph) const;
|
||||
static unsigned int _extractCodepoint(const std::wstring_view glyph) noexcept;
|
||||
uint8_t _lookupGlyphWidth(char32_t codepoint, const std::wstring_view& glyph) noexcept;
|
||||
uint8_t _checkFallbackViaCache(char32_t codepoint, const std::wstring_view& glyph) noexcept;
|
||||
|
||||
mutable std::unordered_map<std::wstring, bool> _fallbackCache;
|
||||
std::function<bool(std::wstring_view)> _pfnFallbackMethod;
|
||||
std::unordered_map<char32_t, uint8_t> _fallbackCache;
|
||||
std::function<bool(const std::wstring_view&)> _pfnFallbackMethod;
|
||||
};
|
||||
|
|
|
|||
|
|
@ -8,11 +8,14 @@ Abstract:
|
|||
- Helpers for determining the width of a particular string of chars.
|
||||
|
||||
*/
|
||||
#pragma once
|
||||
|
||||
#include <functional>
|
||||
#include <string_view>
|
||||
|
||||
bool IsGlyphFullWidth(const std::wstring_view glyph);
|
||||
#include "convert.hpp"
|
||||
|
||||
bool IsGlyphFullWidth(const std::wstring_view& glyph) noexcept;
|
||||
bool IsGlyphFullWidth(const wchar_t wch) noexcept;
|
||||
void SetGlyphWidthFallback(std::function<bool(std::wstring_view)> pfnFallback);
|
||||
void SetGlyphWidthFallback(std::function<bool(const std::wstring_view&)> pfnFallback) noexcept;
|
||||
void NotifyGlyphWidthFontChanged() noexcept;
|
||||
|
|
|
|||
|
|
@ -19,10 +19,9 @@ Author:
|
|||
|
||||
enum class CodepointWidth : BYTE
|
||||
{
|
||||
Ambiguous = 0, // could be narrow or wide depending on the current codepage and font
|
||||
Narrow,
|
||||
Wide,
|
||||
Ambiguous, // could be narrow or wide depending on the current codepage and font
|
||||
Invalid // not a valid unicode codepoint
|
||||
};
|
||||
|
||||
[[nodiscard]] std::wstring ConvertToW(const UINT codepage,
|
||||
|
|
@ -34,6 +33,4 @@ enum class CodepointWidth : BYTE
|
|||
[[nodiscard]] size_t GetALengthFromW(const UINT codepage,
|
||||
const std::wstring_view source);
|
||||
|
||||
CodepointWidth GetQuickCharWidth(const wchar_t wch) noexcept;
|
||||
|
||||
wchar_t Utf16ToUcs2(const std::wstring_view charData);
|
||||
|
|
|
|||
|
|
@ -20,8 +20,8 @@
|
|||
# significantly smaller, which would provide a performance win on the admittedly
|
||||
# extremely rare occasion that we should need to regenerate our table.
|
||||
#
|
||||
# Invoke as ./Generate-xxx ucd.nounihan.flat.xml -Pack | Out-File -Encoding
|
||||
# UTF-8 Temporary.cpp
|
||||
# Invoke this script from the root of this repository as:
|
||||
# .\tools\Generate-CodepointWidthsFromUCD.ps1 -Path .\path\to\ucd.nounihan.flat.xml -OverridePath .\src\types\unicode_width_overrides.xml -Pack
|
||||
#
|
||||
# [1]: https://www.unicode.org/Public/UCD/latest/ucdxml/
|
||||
# [2]: https://www.unicode.org/reports/tr42/
|
||||
|
|
@ -43,15 +43,13 @@ Param(
|
|||
[string]$OverridePath = "overrides.xml",
|
||||
|
||||
[switch]$Pack, # Pack tightly based on width
|
||||
[switch]$NoOverrides, # Do not include overrides
|
||||
[switch]$Full = $False # Include Narrow codepoints
|
||||
[switch]$NoOverrides # Do not include overrides
|
||||
)
|
||||
|
||||
Enum CodepointWidth {
|
||||
Narrow;
|
||||
Wide;
|
||||
Ambiguous;
|
||||
Invalid;
|
||||
}
|
||||
|
||||
# UCD Functions {{{
|
||||
|
|
@ -82,8 +80,8 @@ Function Get-UCDEntryWidth($entry) {
|
|||
"W" { [CodepointWidth]::Wide; Return }
|
||||
"F" { [CodepointWidth]::Wide; Return }
|
||||
"A" { [CodepointWidth]::Ambiguous; Return }
|
||||
default { throw "Unexpected East_Asian_Width property" }
|
||||
}
|
||||
[CodepointWidth]::Invalid
|
||||
}
|
||||
|
||||
Function Get-UCDEntryFlags($entry) {
|
||||
|
|
@ -224,20 +222,10 @@ $UCDRepertoire = $InputObject.ucd.repertoire.ChildNodes | Sort-Object {
|
|||
}
|
||||
}
|
||||
|
||||
If (-not $Full) {
|
||||
$UCDRepertoire = $UCDRepertoire | Where-Object {
|
||||
# Select everything Wide/Ambiguous/Full OR Emoji w/ Emoji Presentation
|
||||
($_.ea -notin "N", "Na", "H") -or ($_.Emoji -eq "Y" -and $_.EPres -eq "Y")
|
||||
}
|
||||
}
|
||||
|
||||
$ranges = [UnicodeRangeList]::New(1024)
|
||||
|
||||
$c = 0
|
||||
ForEach($v in $UCDRepertoire) {
|
||||
$range = [UnicodeRange]::new($v)
|
||||
$c += $range.Length()
|
||||
|
||||
If ($ranges.Count -gt 0 -and $ranges[$ranges.Count - 1].Merge($range)) {
|
||||
# Merged into last entry
|
||||
Continue
|
||||
|
|
@ -260,9 +248,16 @@ If (-not $NoOverrides) {
|
|||
}
|
||||
}
|
||||
|
||||
$ranges.RemoveAll({ $args[0].Width -eq [CodepointWidth]::Narrow }) | Out-Null
|
||||
|
||||
$c = 0
|
||||
ForEach($_ in $ranges) {
|
||||
$c += $_.End - $_.Start + 1
|
||||
}
|
||||
|
||||
# Emit Code
|
||||
" // Generated by {0} -Pack:{1} -Full:{2} -NoOverrides:{3}" -f $MyInvocation.MyCommand.Name, $Pack, $Full, $NoOverrides
|
||||
" // on {0} (UTC) from {1}." -f (Get-Date -AsUTC), $InputObject.ucd.description
|
||||
" // on {0} from {1}." -f (Get-Date -AsUTC -Format "u"), $InputObject.ucd.description
|
||||
" // {0} (0x{0:X}) codepoints covered." -f $c
|
||||
If (-not $NoOverrides) {
|
||||
" // {0} (0x{0:X}) codepoints overridden." -f $overrideCount
|
||||
|
|
@ -270,11 +265,12 @@ If (-not $NoOverrides) {
|
|||
}
|
||||
" static constexpr std::array<UnicodeRange, {0}> s_wideAndAmbiguousTable{{" -f $ranges.Count
|
||||
ForEach($_ in $ranges) {
|
||||
$isAmbiguous = $_.Width -eq [CodepointWidth]::Ambiguous
|
||||
$comment = ""
|
||||
if ($null -ne $_.Comment) {
|
||||
# We only vend comments when we aren't packing tightly
|
||||
$comment = " // {0}" -f $_.Comment
|
||||
}
|
||||
" UnicodeRange{{ 0x{0:x}, 0x{1:x}, CodepointWidth::{2} }},{3}" -f $_.Start, $_.End, $_.Width, $comment
|
||||
" UnicodeRange{{ 0x{0:x}, 0x{1:x}, {2} }},{3}" -f $_.Start, $_.End, [int]$isAmbiguous, $comment
|
||||
}
|
||||
" };"
|
||||
|
|
|
|||
Загрузка…
Ссылка в новой задаче