add Encoding::beginning_of_code_point()

src/core/Encoding.cc

@@ -81,3 +81,18 @@ uint8_t Encoding::num_bytes_in_code_point(Type type, const uint8_t * encoded)
     }
     return 0u;
 }
+
+const uint8_t * Encoding::beginning_of_code_point(Type type, const uint8_t * encoded)
+{
+    switch (type)
+    {
+    case UTF_8:
+        while ((*encoded & 0xC0u) == 0x80u)
+            encoded--;
+        return encoded;
+    case CP_1252:
+        return encoded;
+    }
+
+    return nullptr;
+}

src/core/Encoding.h

@@ -15,6 +15,7 @@ public:
 
     static Type detect_encoding(const uint8_t * buffer, size_t length);
     static uint8_t num_bytes_in_code_point(Type type, const uint8_t * encoded);
+    static const uint8_t * beginning_of_code_point(Type type, const uint8_t * encoded);
 };
 
 #endif