From 57883c120e95c72c7b47e82524ec1d468cdff76b Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Sat, 22 Jan 2011 11:28:47 -0700
Subject: [PATCH] Fix regression in converting HTML files that have non ASCII
 characters inside their <style> tags. Apparently Word generates these. Fixes
 #8494 (trouble converting htm files (for john))

---
 src/calibre/ebooks/oeb/base.py | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/src/calibre/ebooks/oeb/base.py b/src/calibre/ebooks/oeb/base.py
index e11f6b45be..9389964962 100644
--- a/src/calibre/ebooks/oeb/base.py
+++ b/src/calibre/ebooks/oeb/base.py
@@ -221,7 +221,10 @@ def rewrite_links(root, link_repl_func, resolve_base_href=False):
                         el.text):
             stylesheet = parseString(el.text)
             replaceUrls(stylesheet, link_repl_func)
-            el.text = '\n'+stylesheet.cssText + '\n'
+            repl = stylesheet.cssText
+            if isbytestring(repl):
+                repl = repl.decode('utf-8')
+            el.text = '\n'+ repl + '\n'
 
         if 'style' in el.attrib:
             text = el.attrib['style']
@@ -234,8 +237,11 @@ def rewrite_links(root, link_repl_func, resolve_base_href=False):
                             set_property(item)
                     elif v.CSS_PRIMITIVE_VALUE == v.cssValueType:
                         set_property(v)
-                el.attrib['style'] = stext.cssText.replace('\n', ' ').replace('\r',
+                repl = stext.cssText.replace('\n', ' ').replace('\r',
                         ' ')
+                if isbytestring(repl):
+                    repl = repl.decode('utf-8')
+                el.attrib['style'] = repl