add 'replace' argument to unicode(), because we can have non-valid characters...

add 'replace' argument to unicode(), because we can have non-valid characters in error_log (in proxy field caches) for example.


git-svn-id: https://svn.erp5.org/repos/public/erp5/trunk@38896 20353a03-c40f-0410-a6d1-a30d3c3de9de

product/ERP5Type/patches/unicodeconflictresolver.py

@@ -30,5 +30,5 @@ else:
     # Since we use UTF-8 only in PageTemplate, it is enough here. It is
     # faster than the original implementation, and it is compatible with
     # requests that do not contain Accept-Charset header.
-    return unicode(text, 'utf-8')
+    return unicode(text, 'utf-8', 'replace')
   PreferredCharsetResolver.resolve = PreferredCharsetResolver_resolve