ext/utf8/utf8.c in utf8-0.1.2 vs ext/utf8/utf8.c in utf8-0.1.3

@@ -3,11 +3,11 @@
 
 #define CHECK_LEN if ((size_t)(in-start) >= in_len) return -1;
 
 /*
  * Scans the current position of the buffer
- * returning the length of this UTF8 character
+ * returning the length of this UTF-8 character
  */
 inline int8_t utf8CharLen(unsigned char *in, size_t in_len) {
   if (in_len > 0) {
     unsigned char curChar, *start;
 
@@ -58,11 +58,11 @@
   return -1;
 }
 
 /*
  * Scans the current position of the buffer
- * returning the total number of UTF8 characters found
+ * returning the total number of UTF-8 characters found
  */
 int64_t utf8CharCount(unsigned char *in, size_t in_len) {
   int64_t total = 0, leftOver = in_len;
   int8_t len = 0;
   unsigned char *start = in;
@@ -78,6 +78,28 @@
       total++;
     }
   }
 
   return total;
+}
+
+/*
+ * Scans the current position of the buffer
+ * returning the codepoint for the UTF-8 character at this position
+ */
+int32_t utf8CharToCodepoint(unsigned char *in, size_t in_len) {
+  int32_t cp, ncp, len;
+
+  len = utf8CharLen(in, in_len);
+  cp = *in++;
+  if (len > 1) {
+    len--;
+    ncp = cp & ((1 << (6 - len)) - 1);
+    while (len--) {
+      cp = *in++;
+      ncp = (ncp << 6) | (cp & ((1 << 6) - 1));
+    }
+    return ncp;
+  } else {
+    return cp;
+  }
 }