From b1d82efcc50bae26e40f5d9c73ebd788b9d6e950 Mon Sep 17 00:00:00 2001
From: Sven Strickroth <email@cs-ware.de>
Date: Sun, 29 Jul 2012 20:40:27 +0200
Subject: [PATCH] Added unicode big-endian text file support to
 TortoiseGitBlame

Signed-off-by: Sven Strickroth <email@cs-ware.de>
---
 src/Changelog.txt                             |  2 +-
 src/TortoiseGitBlame/TortoiseGitBlameView.cpp | 31 ++++++++++++++++++++++++++-
 2 files changed, 31 insertions(+), 2 deletions(-)

diff --git a/src/Changelog.txt b/src/Changelog.txt
index 68a3d2342..c05cf5c7f 100644
--- a/src/Changelog.txt
+++ b/src/Changelog.txt
@@ -9,7 +9,7 @@ Released: unreleased
  * Allow users to store pushremote and pushbranch on push dialog for local branches
  * Fixed issue #1205: Add "Last modified" column to commit dialog
  * Fixed issue #1295: Browse Ref Dialog ref name order consider numbers in string
- * Added unicode big-endian text file support to TortoiseMerge
+ * Added unicode big-endian text file support to TortoiseMerge and TortoiseGitBlame
 
 == Bug Fix ==
  * Fixed issue #913: Merge does not cope with ambiguous tag and branch names
diff --git a/src/TortoiseGitBlame/TortoiseGitBlameView.cpp b/src/TortoiseGitBlame/TortoiseGitBlameView.cpp
index e9bf2b9c9..8f03af0b8 100644
--- a/src/TortoiseGitBlame/TortoiseGitBlameView.cpp
+++ b/src/TortoiseGitBlame/TortoiseGitBlameView.cpp
@@ -41,6 +41,11 @@
 #define new DEBUG_NEW
 #endif
 
+wchar_t WideCharSwap2(wchar_t nValue)
+{
+	return (((nValue>> 8)) | (nValue << 8));
+}
+
 UINT CTortoiseGitBlameView::m_FindDialogMessage;
 
 // CTortoiseGitBlameView
@@ -2452,6 +2457,13 @@ int CTortoiseGitBlameView::GetEncode(unsigned char *buff, int size, int *bomoffs
 		return 1200;
 	}
 
+	// check for UNICODE_BE does not work, because git blame produces two NUL chars in a row
+	if(type == CFileTextLines::BINARY && size > 2 && buff[0] == 0xFE && buff[1] == 0xFF)
+	{
+		*bomoffset = 2;
+		return 1201;
+	}
+
 	return GetACP();
 }
 
@@ -2531,7 +2543,24 @@ void CTortoiseGitBlameView::UpdateInfo(int Encode)
 				encoding = GetEncode( &data[start+2], data.size() - start -2, &bomoffset);
 			}
 			{
-				if(encoding == 1200)
+				if(encoding == 1201)
+				{
+					CString strw;
+					int size = ((current - start -2 - bomoffset)/2);
+					TCHAR *buffer = strw.GetBuffer(size);
+					memcpy(buffer, &data[start + 2 + bomoffset],sizeof(TCHAR)*size);
+					// swap the bytes to little-endian order to get proper strings in wchar_t format
+					wchar_t * pSwapBuf = buffer;
+					for (DWORD i = 0; i<size; ++i)
+					{
+						*pSwapBuf = WideCharSwap2(*pSwapBuf);
+						++pSwapBuf;
+					}
+					strw.ReleaseBuffer();
+
+					stra = CUnicodeUtils::GetUTF8(strw);
+				}
+				else if(encoding == 1200)
 				{
 					CString strw;
 					// the first bomoffset is 2, after that it's 1 (see issue #920)
-- 
2.11.4.GIT