Public release

lib/Utf8/Utf8.cpp

@@ -0,0 +1,31 @@
+#include "Utf8.h"
+
+int utf8CodepointLen(const unsigned char c) {
+  if (c < 0x80) return 1;          // 0xxxxxxx
+  if ((c >> 5) == 0x6) return 2;   // 110xxxxx
+  if ((c >> 4) == 0xE) return 3;   // 1110xxxx
+  if ((c >> 3) == 0x1E) return 4;  // 11110xxx
+  return 1;                        // fallback for invalid
+}
+
+uint32_t utf8NextCodepoint(const unsigned char** string) {
+  if (**string == 0) {
+    return 0;
+  }
+
+  const int bytes = utf8CodepointLen(**string);
+  const uint8_t* chr = *string;
+  *string += bytes;
+
+  if (bytes == 1) {
+    return chr[0];
+  }
+
+  uint32_t cp = chr[0] & ((1 << (7 - bytes)) - 1);  // mask header bits
+
+  for (int i = 1; i < bytes; i++) {
+    cp = (cp << 6) | (chr[i] & 0x3F);
+  }
+
+  return cp;
+}

lib/Utf8/Utf8.h

@@ -0,0 +1,5 @@
+#pragma once
+
+#include <cstdint>
+
+uint32_t utf8NextCodepoint(const unsigned char** string);