From 04166cb7dd90918385835f246c43d8ec22af0d68 Mon Sep 17 00:00:00 2001 From: Tom Gundersen Date: Mon, 22 Dec 2014 12:57:05 +0100 Subject: shared: utf8 - support decoding the full utf16 We originally only supported the BMP (i.e., we treated UTF-16 as UCS-2). --- src/shared/utf8.h | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'src/shared/utf8.h') diff --git a/src/shared/utf8.h b/src/shared/utf8.h index 3d5a4c3a9c..77f663438e 100644 --- a/src/shared/utf8.h +++ b/src/shared/utf8.h @@ -41,3 +41,15 @@ char *utf16_to_utf8(const void *s, size_t length); int utf8_encoded_valid_unichar(const char *str); int utf8_encoded_to_unichar(const char *str); + +static inline bool utf16_is_surrogate(uint16_t c) { + return (0xd800 <= c && c <= 0xdfff); +} + +static inline bool utf16_is_trailing_surrogate(uint16_t c) { + return (0xdc00 <= c && c <= 0xdfff); +} + +static inline uint32_t utf16_surrogate_pair_to_unichar(uint16_t lead, uint16_t trail) { + return ((lead - 0xd800) << 10) + (trail - 0xdc00) + 0x10000; +} -- cgit v1.2.3-54-g00ecf