Require xhtml as output when serializing repaired html.

It will help HTMLParser to parse broken documents.


git-svn-id: https://svn.erp5.org/repos/public/erp5/trunk@38332 20353a03-c40f-0410-a6d1-a30d3c3de9de

product/PortalTransforms/transforms/safe_html.py

@@ -379,7 +379,8 @@ class SafeHTML:
                                           remove_comments=True)
                 repaired_html_tree = etree.HTML(orig, parser=lparser)
                 html_string = tostring(repaired_html_tree,
-                                       include_meta_content_type=True)
+                                       include_meta_content_type=True,
+                                       method='xml')
                 # avoid breaking now.
                 # continue into the loop with repaired html
             else: