util: when unescaping C escape sequences support C++11 \u and \U unicode literals

[elogind.git] / src / shared / utf8.h

diff --git a/src/shared/utf8.h b/src/shared/utf8.h
index e3eef11bb93137e3171f47c192ecad68f591e282..e745649f06f5b06fcbe7b0869a798f068820d403 100644 (file)
--- a/src/shared/utf8.h
+++ b/src/shared/utf8.h
@@ -25,14 +25,33 @@
 
 #include "macro.h"
 
+#define UTF8_REPLACEMENT_CHARACTER "\xef\xbf\xbd"
+
+bool unichar_is_valid(uint32_t c);
+
 const char *utf8_is_valid(const char *s) _pure_;
 char *ascii_is_valid(const char *s) _pure_;
 
-bool utf8_is_printable(const char* str, size_t length) _pure_;
+bool utf8_is_printable_newline(const char* str, size_t length, bool newline) _pure_;
+#define utf8_is_printable(str, length) utf8_is_printable_newline(str, length, true)
 
-char *ascii_filter(const char *s);
+char *utf8_escape_invalid(const char *s);
+char *utf8_escape_non_printable(const char *str);
 
+size_t utf8_encode_unichar(char *out_utf8, uint32_t g);
 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;
+}