From d3080eb113ecc572de46426aeacdd3bb1f92f516 Mon Sep 17 00:00:00 2001
From: Philip Bauer <bauer@starzel.de>
Date: Fri, 5 Oct 2018 16:06:05 +0200
Subject: [PATCH] use safe_nativestring to transform before html.fromstring

---
 Products/PortalTransforms/transforms/safe_html.py | 7 +++----
 news/38.bugfix                                    | 1 +
 2 files changed, 4 insertions(+), 4 deletions(-)
 create mode 100644 news/38.bugfix

diff --git a/Products/PortalTransforms/transforms/safe_html.py b/Products/PortalTransforms/transforms/safe_html.py
index d9d7884..807ee74 100644
--- a/Products/PortalTransforms/transforms/safe_html.py
+++ b/Products/PortalTransforms/transforms/safe_html.py
@@ -3,6 +3,7 @@
 import six
 
 from Products.CMFPlone.interfaces import IFilterSchema
+from Products.CMFPlone.utils import safe_encode
 from Products.PortalTransforms.interfaces import ITransform
 from Products.PortalTransforms.libtransforms.utils import bodyfinder
 from lxml import etree
@@ -2398,10 +2399,8 @@ def convert(self, orig, data, **kwargs):
     def scrub_html(self, orig):
         # append html tag to create a dummy parent for the tree
         html_parser = html.HTMLParser(encoding='utf-8')
-        if isinstance(orig, six.binary_type):
-            tag = b'<html'
-        else:
-            tag = '<html'
+        orig = safe_encode(orig)
+        tag = b'<html'
         if tag in orig.lower():
             # full html
             tree = html.fromstring(orig, parser=html_parser)
diff --git a/news/38.bugfix b/news/38.bugfix
new file mode 100644
index 0000000..11a9f12
--- /dev/null
+++ b/news/38.bugfix
@@ -0,0 +1 @@
+fix scrub_html when passing unicode [pbauer]
\ No newline at end of file