From c4253078a9bbbbe745d70bea70a3834ba565b590 Mon Sep 17 00:00:00 2001
From: Josh Holtrop <jholtrop@gmail.com>
Date: Thu, 11 Aug 2016 20:03:53 -0400
Subject: [PATCH] add Encoding class

---
 src/core/Encoding.cc | 62 ++++++++++++++++++++++++++++++++++++++++++++
 src/core/Encoding.h  | 19 ++++++++++++++
 2 files changed, 81 insertions(+)
 create mode 100644 src/core/Encoding.cc
 create mode 100644 src/core/Encoding.h

diff --git a/src/core/Encoding.cc b/src/core/Encoding.cc
new file mode 100644
index 0000000..ce428cc
--- /dev/null
+++ b/src/core/Encoding.cc
@@ -0,0 +1,62 @@
+#include "Encoding.h"
+
+static inline bool
+check_continuation_bytes(const uint8_t ** buffer,
+        uint8_t continuation_bytes_to_check)
+{
+    for (uint8_t i = 0; i < continuation_bytes_to_check; i++)
+    {
+        if ((**buffer & 0xC0u) != 0x80u)
+            return false;
+        (*buffer)++;
+    }
+    return true;
+}
+
+Encoding::Type Encoding::detect_encoding(const uint8_t * buffer, size_t length)
+{
+    for (size_t i = 0u; i < length; i++)
+    {
+        const uint8_t c = *buffer;
+        uint8_t continuation_bytes_to_check;
+        if ((c & 0x80u) == 0u)
+        {
+            buffer++;
+            continue;
+        }
+        else if ((c & 0xE0u) == 0xC0u)
+        {
+            continuation_bytes_to_check = 1u;
+        }
+        else if ((c & 0xF0u) == 0xE0u)
+        {
+            continuation_bytes_to_check = 2u;
+        }
+        else if ((c & 0xF8u) == 0xF0u)
+        {
+            continuation_bytes_to_check = 3u;
+        }
+        else if ((c & 0xFCu) == 0xF8u)
+        {
+            continuation_bytes_to_check = 4u;
+        }
+        else if ((c & 0xFEu) == 0xFCu)
+        {
+            continuation_bytes_to_check = 5u;
+        }
+        else
+        {
+            return CP_1252;
+        }
+        buffer++;
+        if (((i + continuation_bytes_to_check) < length) &&
+                check_continuation_bytes(&buffer, continuation_bytes_to_check))
+        {
+            i += continuation_bytes_to_check;
+            continue;
+        }
+        return CP_1252;
+    }
+
+    return UTF_8;
+}
diff --git a/src/core/Encoding.h b/src/core/Encoding.h
new file mode 100644
index 0000000..21e11f0
--- /dev/null
+++ b/src/core/Encoding.h
@@ -0,0 +1,19 @@
+#ifndef ENCODING_H
+#define ENCODING_H
+
+#include <stdint.h>
+#include <stdlib.h>
+
+class Encoding
+{
+public:
+    enum Type
+    {
+        UTF_8,
+        CP_1252,
+    };
+
+    static Type detect_encoding(const uint8_t * buffer, size_t length);
+};
+
+#endif