p Each call to .Nm updates the conversion state .Fa ps with a UTF-16 code unit .Fa c16 , writes up to .Dv MB_CUR_MAX bytes to .Fa s (possibly none), and returns either the number of bytes written to .Fa s or .Li (size_t)-1 to denote error.
p If .Fa s is a null pointer, no output is produced and .Fa ps is reset to the initial conversion state, as if the call had been .Fo c8rtomb .Va buf , .Li 0 , .Fa ps .Fc for some internal buffer .Va buf .
p If .Fa c16 is zero, .Nm discards any pending incomplete UTF-16 code unit sequence in .Fa ps , outputs a (possibly empty) shift sequence to restore the initial state followed by a NUL byte, and resets .Fa ps to the initial conversion state.
p If .Fa ps is a null pointer, .Nm uses an internal .Vt mbstate_t object with static storage duration, distinct from all other .Vt mbstate_t objects
o including those used by other functions such as .Xr mbrtoc16 3
c ,
which is initialized at program startup to the initial conversion
state.
""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
.Sh RETURN VALUES
The
.Nm
function returns the number of bytes written to
.Fa s
on success, or sets
.Xr errno 2
and returns
.Li "(size_t)-1"
on failure.
""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
.Sh EXAMPLES
Convert a UTF-16 code unit sequence to a multibyte string,
NUL-terminate it (with any shift sequence needed to restore the initial
state), and print it:
d -literal -offset indent char16_t c16[] = { 0xd83d, 0xdca9 };
char buf[(__arraycount(c16) + 1)*MB_LEN_MAX], *s = buf;
size_t i;
mbstate_t mbs = {0}; /* initial conversion state */
for (i = 0; i < __arraycount(c16); i++) {
size_t len;
len = c16rtomb(s, c16[i], &mbs);
if (len == (size_t)-1)
err(1, "c16rtomb");
assert(len < sizeof(buf) - (s - buf));
s += len;
}
len = c16rtomb(s, 0, &mbs); /* NUL-terminate */
if (len == (size_t)-1)
err(1, "c16rtomb");
assert(len <= sizeof(buf) - (s - buf));
printf("%s\en", buf);
.Ed
p
To avoid a variable-length array, this code uses
.Dv MB_LEN_MAX ,
which is a constant upper bound on the locale-dependent
.Dv MB_CUR_MAX .
""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
.Sh ERRORS
l -tag -width Bq t Bq Er EILSEQ .Fa c16
is invalid as the next code unit in the conversion state
.Fa ps .
t Bq Er EILSEQ The input cannot be encoded as a multibyte sequence in the current
locale.
t Bq Er EIO An error occurred in loading the locale's character conversions.
.El
""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
.Sh SEE ALSO
.Xr c32rtomb 3 ,
.Xr c8rtomb 3 ,
.Xr mbrtoc16 3 ,
.Xr mbrtoc32 3 ,
.Xr mbrtoc8 3 ,
.Xr uchar 3
.Rs
.%B The Unicode Standard
.%O Version 15.0 \(em Core Specification
.%Q The Unicode Consortium
.%D September 2022
.%U https://www.unicode.org/versions/Unicode15.0.0/UnicodeStandard-15.0.pdf
.Re
.Rs
.%A P. Hoffman
.%A F. Yergeau
.%T UTF-16, an encoding of ISO 10646
.%R RFC 2781
.%D February 2000
.%I Internet Engineering Task Force
.%U https://datatracker.ietf.org/doc/html/rfc2781
.Re
""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
.Sh STANDARDS
The
.Nm
function conforms to
.St -isoC-2011 .
""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
.Sh HISTORY
The
.Nm
function first appeared in
.Nx 11.0 .
""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
.Sh BUGS
The standard requires that passing zero as
.Fa c16
unconditionally reset the conversion state and output a NUL byte:
d -filled -offset indent If
.Fa c16
is a null wide character, a null byte is stored, preceded by any shift
sequence needed to restore the initial shift state; the resulting state
described is the initial conversion state.
.Ed
p However, some implementations such as .Fx 14.0 , .Ox 7.4 , and glibc 2.36 ignore this clause and, if the zero was preceded by an incomplete UTF-16 code unit sequence, fail with .Er EILSEQ instead.