C++ Reference · Strings and I/O
Unicode in C++: char8_t and UTF-8 strings
How C++ handles Unicode: char8_t, char16_t, char32_t, u8/u/U literals, std::u8string, UTF-8 byte counts, code points versus characters, and printing text.
C++ has no built-in "Unicode string" type. It has character types and literal prefixes, and a library that treats strings as plain sequences of code units. A code unit is one element of the string (a byte for UTF-8, a 16-bit value for UTF-16). A code point is one Unicode number such as U+00E9 (the letter é). One code point takes between one and four UTF-8 code units, so size() of a UTF-8 string counts bytes, not letters.
The types are char8_t (C++20, UTF-8), char16_t (UTF-16) and char32_t (UTF-32, one code unit per code point). Their literals are u8"...", u"..." and U"...". The most common practical choice is ordinary std::string holding UTF-8, which works with files, networks and the console on most systems.
std::u8string a = u8"café"; // C++20: 5 char8_t elements std::u16string b = u"café"; // 4 char16_t elements std::u32string c = U"café"; // 4 char32_t elements
Try it
Press Run (or Ctrl + Enter) to see the result here.
Program input · what you'd type while the program runs
Type the answers before you press Run, one per line. Each time the program reads a line (with std::cin or std::getline), it gets the next line from this box.
Press Run (or Ctrl + Enter) to see the result here.
Program input · what you'd type while the program runs
Type the answers before you press Run, one per line. Each time the program reads a line (with std::cin or std::getline), it gets the next line from this box.
Syntax
char8_t // C++20: UTF-8 code unit, unsigned, distinct type from unsigned char char16_t // UTF-16 code unit char32_t // UTF-32 code unit (one code point) u8"text" // const char8_t[N] (const char[N] before C++20) u"text" // const char16_t[N] U"text" // const char32_t[N] 'a' u8'a' u'a' U'a' // character literals (u8 characters: C++17, ASCII only) std::u8string, std::u16string, std::u32string // std::basic_string specializations
| Escape | Meaning |
|---|---|
\uXXXX |
A code point up to U+FFFF (4 hex digits). |
\UXXXXXXXX |
Any code point (8 hex digits). |
\xHH |
One code unit with this value (use with care in u8 strings). |
\N{name} |
Named escape (C++23). |
Usage notes
char8_tis a separate type since C++20.u8"..."no longer converts toconst char*, andstd::cout << std::u8stringis deleted. Convert withreinterpret_cast<const char*>(...), or keep text instd::string, which holds UTF-8 bytes just as well.size()counts code units. Use a loop like the one above (or a library) to count code points. Even a code point is not always one visible "character": an e followed by a combining accent are two code points that display as one letter (a grapheme cluster).- Sorting, case conversion, comparison and normalization are not Unicode-aware in the standard library.
std::toupperworks on single bytes,std::sorton strings compares code units. Use ICU or another library for linguistically correct behavior. - Source file encoding. GCC and Clang read UTF-8 source by default, MSVC needs
/utf-8. Plain"text"literals contain whatever bytes are in the file; the execution character set is UTF-8 on most platforms but not guaranteed. Windows consoles may needSetConsoleOutputCP(CP_UTF8). wchar_tandstd::wstringare 16 bits on Windows (UTF-16) and 32 bits elsewhere (UTF-32). That is why portable code avoids them.- Conversion facilities.
std::wstring_convert,std::wbuffer_convertand thestd::codecvtfacets for UTF conversions are deprecated (C++17, C++20 for the char8_t ones). The standard has no replacement yet, so projects write small converters or use a library. The C functionsmbrtoc8,c8rtomband friends (C23 and C++20<cuchar>) exist but depend on the locale. - Embedded NUL bytes (
\0) are legal instd::stringandstd::u8stringand count towardsize().