+/* Encode a character from the range [0, 0xD7FF] or [0xE000,0x11FFFF];
+ * up to 4 bytes needed */
+static inline void *
+utf16_le_put(void *p, uns u)
+{
+ if (u < 0xd800 || (u < 0x10000 && u >= 0xe000))
+ {
+ put_u16_le(p, u);
+ return p + 2;
+ }
+ else if ((u -= 0x10000) < 0x100000)
+ {
+ put_u16_le(p, u >> 10);
+ put_u16_le(p + 2, u & 0x7ff);
+ return p + 4;
+ }
+ else
+ ASSERT(0);
+}
+
+static inline void *
+utf16_be_put(void *p, uns u)
+{
+ if (u < 0xd800 || (u < 0x10000 && u >= 0xe000))
+ {
+ put_u16_be(p, u);
+ return p + 2;
+ }
+ else if ((u -= 0x10000) < 0x100000)
+ {
+ put_u16_be(p, u >> 10);
+ put_u16_be(p + 2, u & 0x7ff);
+ return p + 4;
+ }
+ else
+ ASSERT(0);
+}
+
+/* Decode a character from the range [0, 0xD7FF] or [0xE000,11FFFF]
+ * or return `repl' if the encoding has been corrupted */
+static inline void *
+utf16_le_get_repl(const void *p, uns *uu, uns repl)
+{
+ uns u = get_u16_le(p), x, y;
+ x = u - 0xd800;
+ if (x < 0x800)
+ if (x < 0x400 && (y = get_u16_le(p + 2) - 0xdc00) < 0x400)
+ {
+ u = 0x10000 + (x << 10) + y;
+ p += 2;
+ }
+ else
+ u = repl;
+ *uu = u;
+ return (void *)(p + 2);
+}
+
+static inline void *
+utf16_be_get_repl(const void *p, uns *uu, uns repl)
+{
+ uns u = get_u16_be(p), x, y;
+ x = u - 0xd800;
+ if (x < 0x800)
+ if (x < 0x400 && (y = get_u16_be(p + 2) - 0xdc00) < 0x400)
+ {
+ u = 0x10000 + (x << 10) + y;
+ p += 2;
+ }
+ else
+ u = repl;
+ *uu = u;
+ return (void *)(p + 2);
+}
+
+/* Decode a character from the range [0, 0xD7FF] or [0xE000,11FFFF]
+ * or return UNI_REPLACEMENT if the encoding has been corrupted */
+static inline void *
+utf16_le_get(const void *p, uns *uu)
+{
+ return utf16_le_get_repl(p, uu, UNI_REPLACEMENT);
+}
+
+static inline void *
+utf16_be_get(const void *p, uns *uu)
+{
+ return utf16_be_get_repl(p, uu, UNI_REPLACEMENT);
+}
+
+static inline uns
+unicode_sanitize_char(uns u)
+{
+ if (u >= 0x10000 || // We don't accept anything outside the basic plane
+ u >= 0xd800 && u < 0xf900 || // neither we do surrogates
+ u >= 0x80 && u < 0xa0 || // nor latin-1 control chars
+ u < 0x20 && u != '\t')
+ return UNI_REPLACEMENT;
+ return u;
+}
+