From 0ad8343cdca35748a4f4ee8b63f3a9a05c50d92c Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Fri, 25 Dec 2009 08:13:38 -0700
Subject: [PATCH] MOBI Metadata reader: Correctly handle non ASCII characters
 when reading embedded metadata. Fixes #4223 (trouble with charset)

---
 src/calibre/ebooks/mobi/reader.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/src/calibre/ebooks/mobi/reader.py b/src/calibre/ebooks/mobi/reader.py
index d0ccb30ec0..f26d8c9e2a 100644
--- a/src/calibre/ebooks/mobi/reader.py
+++ b/src/calibre/ebooks/mobi/reader.py
@@ -435,7 +435,8 @@ class MobiReader(object):
                 open(os.path.splitext(htmlfile)[0] + '.ncx', 'wb').write(ncx)
 
     def read_embedded_metadata(self, root, elem, guide):
-        raw = '<package>' + html.tostring(elem, encoding='utf-8') + '</package>'
+        raw = '<?xml version="1.0" encoding="utf-8" ?>\n<package>' + \
+                html.tostring(elem, encoding='utf-8') + '</package>'
         stream = cStringIO.StringIO(raw)
         opf = OPF(stream)
         self.embedded_mi = MetaInformation(opf)