From ab9672ae73248f51e30f4553c4b8878525e46383 Mon Sep 17 00:00:00 2001 From: Rich Felker Date: Mon, 13 Oct 2014 20:59:42 -0400 Subject: implement uchar.h (C11 UTF-16/32 conversion) interfaces --- src/multibyte/mbrtoc16.c | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+) create mode 100644 src/multibyte/mbrtoc16.c (limited to 'src/multibyte/mbrtoc16.c') diff --git a/src/multibyte/mbrtoc16.c b/src/multibyte/mbrtoc16.c new file mode 100644 index 00000000..74b7d77e --- /dev/null +++ b/src/multibyte/mbrtoc16.c @@ -0,0 +1,28 @@ +#include +#include + +size_t mbrtoc16(char16_t *restrict pc16, const char *restrict s, size_t n, mbstate_t *restrict ps) +{ + unsigned *pending = (unsigned *)ps; + + if (!s) return mbrtoc16(0, "", 1, ps); + + /* mbrtowc states for partial UTF-8 characters have the high bit set; + * we use nonzero states without high bit for pending surrogates. */ + if ((int)*pending > 0) { + if (pc16) *pc16 = *pending; + *pending = 0; + return -3; + } + + wchar_t wc; + size_t ret = mbrtowc(&wc, s, n, ps); + if (ret <= 4) { + if (wc >= 0x10000) { + *pending = (wc & 0x3ff) + 0xdc00; + wc = 0xd7c0 + (wc >> 10); + } + if (pc16) *pc16 = wc; + } + return ret; +} -- cgit v1.2.1