--- JSON-XS/XS.xs	2007/03/24 22:10:08	1.12
+++ JSON-XS/XS.xs	2007/03/24 22:55:16	1.13
@@ -55,6 +55,25 @@
     }
 }
 
+// decode an utf-8 character and return it, or (UV)-1 in
+// case of an error.
+// we special-case "safe" characters from U+80 .. U+7FF,
+// but use the very good perl function to parse anything else.
+// note that we never call this function for a ascii codepoints
+static UV
+decode_utf8 (unsigned char *s, STRLEN len, STRLEN *clen)
+{
+  if (s[0] > 0xdf || s[0] < 0xc2)
+    return utf8n_to_uvuni (s, len, clen, UTF8_CHECK_ONLY);
+  else if (len > 1 && s[1] >= 0x80 && s[1] <= 0xbf)
+    {
+      *clen = 2;
+      return ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
+    }
+  else
+    return (UV)-1;
+}
+
 /////////////////////////////////////////////////////////////////////////////
 // encoder
 
@@ -135,7 +154,8 @@
 
                   if (is_utf8)
                     {
-                      uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
+                      //uch = utf8n_to_uvuni (str, end - str, &clen, UTF8_CHECK_ONLY);
+                      uch = decode_utf8 (str, end - str, &clen);
                       if (clen == (STRLEN)-1)
                         croak ("malformed or illegal unicode character in string [%.11s], cannot convert to JSON", str);
                     }
@@ -621,7 +641,7 @@
               --dec->cur;
 
               STRLEN clen;
-              UV uch = utf8n_to_uvuni (dec->cur, dec->end - dec->cur, &clen, UTF8_CHECK_ONLY);
+              UV uch = decode_utf8 (dec->cur, dec->end - dec->cur, &clen);
               if (clen == (STRLEN)-1)
                 ERR ("malformed UTF-8 character in JSON string");