shared: utf8 - support decoding the full utf16

We originally only supported the BMP (i.e., we treated UTF-16 as UCS-2).

Signed-off-by: Anthony G. Basile <blueness@gentoo.org>

1 files changed, 12 insertions, 0 deletions

diff --git a/src/shared/utf8.h b/src/shared/utf8.h
index 2d9b5f7011..143b0ba08c 100644
--- a/src/shared/utf8.h
+++ b/src/shared/utf8.h
@@ -35,3 +35,15 @@ 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;
+}