Difference between revisions of "c/string/multibyte"

Revision as of 11:43, 10 September 2012

A null-terminated multibyte string (NTMBS), or "multibyte string", is a sequence of nonzero bytes followed by a byte with value zero (the terminating null character).

Each character stored in the string may occupy more than one byte. The encoding used to represent characters in a multibyte character string is locale-specific: it may be UTF-8, GB18030, EUC-JP, Shift-JIS, etc. For example, the char array {'\xe4','\xbd','\xa0','\xe5','\xa5','\xbd','\0'} is an NTMBS holding the string "你好" in UTF-8 multibyte encoding: the first three bytes encode the character 你, the next three bytes encode the character 好. The same string encoded in GB18030 is the char array {'\xc4', '\xe3', '\xba', '\xc3', '\0'}, where each of the two characters is encoded as a two-byte sequence.

In some multibyte encodings, any given multibyte character sequence may represent different characters depending on the previous byte sequences, known as "shift sequences". Such encodings are known as state-dependent: knowledge of the current shift state is required to interpret each character. An NTMBS is only valid if it begins and ends in the initial shift state: if a shift sequence was used, the corresponding unshift sequence has to be present before the terminating null character. Examples of such encodings are BOCU-1 and SCSU.

A multibyte character string is layout-compatible with null-terminated byte string (NTBS), that is, can be stored, copied, and examined using the same facilities, except for calculating the number of characters. If the correct locale is in effect, I/O functions also handle multibyte strings. Multibyte strings can be converted to and from wide strings using the codecvt member functions, wstring_convert, or the following locale-dependent conversion functions:

Multibyte/wide character conversions

Types

Macros