From b2bed71ee433c9186856d7d849ab97ecd59f2b5a Mon Sep 17 00:00:00 2001
From: Josh Holtrop <jholtrop@gmail.com>
Date: Thu, 11 Aug 2016 23:13:32 -0400
Subject: [PATCH] add Encoding::beginning_of_code_point()

---
 src/core/Encoding.cc | 15 +++++++++++++++
 src/core/Encoding.h  |  1 +
 2 files changed, 16 insertions(+)

diff --git a/src/core/Encoding.cc b/src/core/Encoding.cc
index b0033c6..774d2b4 100644
--- a/src/core/Encoding.cc
+++ b/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;
+}
diff --git a/src/core/Encoding.h b/src/core/Encoding.h
index f673dbd..5be2ca4 100644
--- a/src/core/Encoding.h
+++ b/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