From 639d1c25dbea344d4ae1d2341e1e66afcdc99b1d Mon Sep 17 00:00:00 2001
From: vfilippov <vfilippov@6955db30-a419-402b-8a0d-67ecbb4d7f56>
Date: Tue, 27 Sep 2011 16:34:42 +0000
Subject: [PATCH] Bug 82651 - Tika&Sphinx&chardet content indexing (done!)

git-svn-id: svn://svn.office.custis.ru/3rdparty/viewvc.org/trunk@1389 6955db30-a419-402b-8a0d-67ecbb4d7f56
---
 lib/viewvcmagic.py | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/lib/viewvcmagic.py b/lib/viewvcmagic.py
index 5f8b3ea8..62edab93 100644
--- a/lib/viewvcmagic.py
+++ b/lib/viewvcmagic.py
@@ -46,7 +46,14 @@ class ContentMagic:
                 charset = chardet.detect(content)
                 if charset and charset['encoding']:
                     charset = charset['encoding']
-                content = content.decode(charset)
+                if charset == 'MacCyrillic':
+                    # Silly MacCyr, try cp1251
+                    try:
+                        content = content.decode('windows-1251')
+                        charset = 'windows-1251'
+                    except: content = content.decode(charset)
+                else:
+                    content = content.decode(charset)
             except: charset = None
         else:
             # Try UTF-8