From f659c142427b1343b4c512c9ef1d8d3f2d045ff5 Mon Sep 17 00:00:00 2001
From: Josh Holtrop <jholtrop@gmail.com>
Date: Mon, 15 Aug 2016 22:14:47 -0400
Subject: [PATCH] add Encoding::decode()

---
 src/core/Buffer.cc    |  2 ++
 src/core/Encoding.cc  | 60 +++++++++++++++++++++++++++++++++++++++++++
 src/core/Encoding.h   |  1 +
 src/core/PieceTable.h |  5 ++--
 4 files changed, 66 insertions(+), 2 deletions(-)

diff --git a/src/core/Buffer.cc b/src/core/Buffer.cc
index 46053a7..143ee0c 100644
--- a/src/core/Buffer.cc
+++ b/src/core/Buffer.cc
@@ -13,6 +13,7 @@ Buffer::Buffer()
     m_eol_at_eof = true;
     m_line_endings = LineEndings::LF;
     m_encoding = Encoding::UTF_8;
+    piece_table->encoding = m_encoding;
 }
 
 Buffer::~Buffer()
@@ -65,6 +66,7 @@ bool Buffer::load_from_file(const char * filename)
     m_eol_at_eof = text_loader.get_eol_at_eof();
     m_line_endings = text_loader.get_line_endings();
     m_encoding = text_loader.get_encoding();
+    piece_table->encoding = m_encoding;
 
     return true;
 }
diff --git a/src/core/Encoding.cc b/src/core/Encoding.cc
index 774d2b4..b026861 100644
--- a/src/core/Encoding.cc
+++ b/src/core/Encoding.cc
@@ -96,3 +96,63 @@ const uint8_t * Encoding::beginning_of_code_point(Type type, const uint8_t * enc
 
     return nullptr;
 }
+
+uint32_t Encoding::decode(Type type, const uint8_t * encoded)
+{
+    switch (type)
+    {
+    case UTF_8:
+        {
+            const uint8_t c = *encoded;
+            uint8_t following_bytes = 0u;
+            uint32_t v;
+            if ((c & 0x80u) == 0u)
+            {
+                return c;
+            }
+            else if ((c & 0xE0u) == 0xC0u)
+            {
+                v = c & 0x1Fu;
+                following_bytes = 1u;
+            }
+            else if ((c & 0xF0u) == 0xE0u)
+            {
+                v = c & 0x0Fu;
+                following_bytes = 2u;
+            }
+            else if ((c & 0xF8u) == 0xF0u)
+            {
+                v = c & 0x07u;
+                following_bytes = 3u;
+            }
+            else if ((c & 0xFCu) == 0xF8u)
+            {
+                v = c & 0x03u;
+                following_bytes = 4u;
+            }
+            else if ((c & 0xFEu) == 0xFCu)
+            {
+                v = c & 0x01u;
+                following_bytes = 5u;
+            }
+            else
+            {
+                return 0u;
+            }
+            while (following_bytes-- > 0u)
+            {
+                encoded++;
+                v <<= 6u;
+                v |= *encoded & 0x3Fu;
+            }
+            return v;
+        }
+        break;
+    case CP_1252:
+        /* TODO: map byte to code point */
+        return *encoded;
+        break;
+    }
+
+    return 0u;
+}
diff --git a/src/core/Encoding.h b/src/core/Encoding.h
index 5be2ca4..3b5761a 100644
--- a/src/core/Encoding.h
+++ b/src/core/Encoding.h
@@ -16,6 +16,7 @@ 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);
+    static uint32_t decode(Type type, const uint8_t * encoded);
 };
 
 #endif
diff --git a/src/core/PieceTable.h b/src/core/PieceTable.h
index be78923..f3c6247 100644
--- a/src/core/PieceTable.h
+++ b/src/core/PieceTable.h
@@ -6,6 +6,7 @@
 #include <utility>
 #include <memory>
 #include <list>
+#include "Encoding.h"
 
 class PieceTable
 {
@@ -64,8 +65,7 @@ public:
         /** Get the character pointed to by the cursor. */
         uint32_t operator*() const
         {
-            /* TODO: Use Encoding */
-            return piece->start[offset];
+            return Encoding::decode(piece_table->encoding, &piece->start[offset]);
         }
 
         /**
@@ -92,6 +92,7 @@ public:
     Piece * start_piece;
     Piece * end_piece;
     uint8_t tabstop;
+    Encoding::Type encoding;
 
     PieceTable(const uint8_t * file_buffer, unsigned long file_buffer_size);