Character literal
Syntax
' c-char ' |
(1) | |
u8' c-char ' |
(2) | (since C++17) |
u' c-char ' |
(3) | (since C++11) |
U' c-char ' |
(4) | (since C++11) |
L' c-char ' |
(5) | |
' c-char-sequence ' |
(6) | |
L' c-char-sequence ' |
(7) |
c-char | - | either
|
basic-c-char | - | A character from the source character set (until C++23)translation character set (since C++23), except the single-quote ' , backslash \ , or new-line character |
c-char-sequence | - | two or more c-chars |
Explanation
'a'
or
'\n'
or
'\13'
. Such literal has type
char
and the value equal to
the representation of c-char in the execution character set (until C++23)
the corresponding code point from ordinary literal encoding (since C++23).
u8'a'
. Such literal has type
char
(until C++20)
char8_t
(since C++20) and the value equal to ISO 10646 code point value of
c-char, provided that the code point value is representable with a single UTF-8 code unit (that is,
c-char is in the range 0x0-0x7F, inclusive).
u'猫'
, but not
u'🍌'
(
u'\U0001f34c'
). Such literal has type
char16_t
and the value equal to ISO 10646 code point value of
c-char, provided that the code point value is representable with a single UTF-16 code unit (that is,
c-char is in the range 0x0-0xFFFF, inclusive).
U'猫'
or
U'🍌'
. Such literal has type
char32_t
and the value equal to ISO 10646 code point value of
c-char.
L'β'
or
L'猫'
. Such literal has type
wchar_t
and the value equal to
the value of c-char in the execution wide character set (until C++23)
the corresponding code point from wide literal encoding (since C++23).
'AB'
, is conditionally-supported, has type
int
and implementation-defined value.
L'AB'
, is conditionally-supported, has type
wchar_t
and implementation-defined value.
Non-encodable characters
In (1), if c-char is not a numeric character sequence and is not representable as a single byte in the execution character set, the character literal is conditionally supported, has type int
and implementation-defined value.
In (2-4), if c-char is not a numeric character sequence and cannot be represented as a single code unit in the associated character encoding, the character literal is ill-formed.
In (5), if c-char is not a numeric character sequence and is not representable as a single code unit in the execution wide character set (e.g. a non-BMP value on Windows where wchar_t
is 16-bit), the character literal is conditionally supported, has type wchar_t
and implementation-defined value.
Numeric escape sequences
Numeric (octal and hexadecimal) escape sequences can be used for specifying the value of the character.
If the character literal contains only one numeric escape sequence, and the value specified by the escape sequence is representable by the unsigned version of its type, the character literal has the same value as the specified value (possibly after conversion to the character type). A UTF-N character literal can have any value representable by its type. If the value does not correspond to a valid Unicode code point, or if the its corresponding code point is not representable as single code unit in UTF-N, it can still be specified by a numeric escape sequence with the value. E.g. |
(since C++23) |
If the value specified by a numeric escape sequence used in a ordinary or wide character literal is not representable by |
(until C++23) |
If the value specified by a numeric escape sequence used in a ordinary or wide character literal with one c-char is representable by the unsigned version of the underlying type of |
(since C++23) |
If the value specified by a numeric escape sequence used in a UTF-N character literal is not representable by the corresponding |
(since C++11) |
Notes
Multicharacter literals were inherited by C from the B programming language. Although not specified by the C or C++ standard, most compilers (MSVC is a notable exception) implement multicharacter literals as specified in B: the values of each char in the literal initialize successive bytes of the resulting integer, in big-endian zero-padded right-adjusted order, e.g. the value of '\1'
is 0x00000001
and the value of '\1\2\3\4'
is 0x01020304
.
In C, character constants such as 'a'
or '\n'
have type int
, rather than char
.
Example
#include <cstdint> #include <iomanip> #include <iostream> #include <string_view> template <typename CharT> void dump(std::string_view s, const CharT c) { const uint8_t* data {reinterpret_cast<const uint8_t*>(&c)}; std::cout << "'" << s << "' \t" << std::hex << std::uppercase << std::setfill('0'); for (auto i {0U}; i != sizeof(CharT); ++i){ std::cout << std::setw(2) << static_cast<unsigned>(data[i]) << ' '; } std::cout << '\n'; } void print(std::string_view str = "") { std::cout << str << '\n'; } int main() { print("Ordinary character literals:"); char c1 = 'a'; dump("a", c1); char c2 = '\x2a'; dump("*", c2); print(); print("Ordinary multi-character literals:"); int mc1 = 'ab'; dump("ab", mc1); // implementation-defined int mc2 = '¢'; dump("¢", mc2); // implementation-defined int mc3 = '猫'; dump("猫", mc3); // implementation-defined int mc4 = '🍌'; dump("🍌", mc4); // implementation-defined print(); print("UTF-8 character literals:"); char8_t C1 = u8'a'; dump("a", C1); // char8_t C2 = u8'¢'; dump("¢", C2); // error: ¢ maps to two UTF-8 code units // char8_t C3 = u8'猫'; dump("猫", C3); // error: 猫 maps to three UTF-8 code units // char8_t C4 = u8'🍌'; dump("🍌", C4); // error: 🍌 maps to four UTF-8 code units print(); print("UTF-16 character literals:"); char16_t uc1 = u'a'; dump("a", uc1); char16_t uc2 = u'¢'; dump("¢", uc2); char16_t uc3 = u'猫'; dump("猫", uc3); // char16_t uc4 = u'🍌'; dump("🍌", uc4); // error: 🍌 maps to two UTF-16 code units print(); print("UTF-32 character literals:"); char32_t Uc1 = U'a'; dump("a", Uc1); char32_t Uc2 = U'¢'; dump("¢", Uc2); char32_t Uc3 = U'猫'; dump("猫", Uc3); char32_t Uc4 = U'🍌'; dump("🍌", Uc4); print(); print("Wide character literals:"); wchar_t wc1 = L'a'; dump("a", wc1); wchar_t wc2 = L'¢'; dump("¢", wc2); wchar_t wc3 = L'猫'; dump("猫", wc3); wchar_t wc4 = L'🍌'; dump("🍌", wc4); }
Possible output:
Ordinary character literals: 'a' 61 '*' 2A Ordinary multi-character literals: 'ab' 62 61 00 00 '¢' A2 C2 00 00 '猫' AB 8C E7 00 '🍌' 8C 8D 9F F0 UTF-8 character literals: 'a' 61 UTF-16 character literals: 'a' 61 00 '¢' A2 00 '猫' 2B 73 UTF-32 character literals: 'a' 61 00 00 00 '¢' A2 00 00 00 '猫' 2B 73 00 00 '🍌' 4C F3 01 00 Wide character literals: 'a' 61 00 00 00 '¢' A2 00 00 00 '猫' 2B 73 00 00 '🍌' 4C F3 01 00
Defect reports
The following behavior-changing defect reports were applied retroactively to previously published C++ standards.
DR | Applied to | Behavior as published | Correct behavior |
---|---|---|---|
CWG 912 | C++98 | non-encodable ordinary character literal was unspecified | specified as conditionally-supported |
CWG 1024 | C++98 | multicharacter literal were required to be supported | made conditionally-supported |
CWG 1656 | C++98 | meaning of numeric escape sequence in a character literal is unclear | specified |
See also
user-defined literals(C++11) | literals with user-defined suffix |
C documentation for Character constant |
© cppreference.com
Licensed under the Creative Commons Attribution-ShareAlike Unported License v3.0.
https://en.cppreference.com/w/cpp/language/character_literal