[htmltransform] Make the transform work with both unicode and bytes input

Needed for py3k support. Found by running CubicWeb tests in py3k.

logilab/mtconverter/transforms/htmltransform.py

@@ -16,6 +16,8 @@
 # You should have received a copy of the GNU Lesser General Public License along
 # with logilab-mtconverter. If not, see <http://www.gnu.org/licenses/>.
 
+from six import binary_type
+
 from html2text import html2text
 
 from logilab.mtconverter.transform import Transform
@@ -30,4 +32,8 @@ class html_to_formatted_text(Transform):
 
 
     def _convert(self, trdata):
-        return html2text(trdata.data).encode(trdata.encoding)
+        if isinstance(trdata.data, binary_type):
+            data = trdata.data.decode(trdata.encoding)
+        else:
+            data = trdata.data
+        return html2text(data).encode(trdata.encoding)

test/unittest_transforms.py

@@ -45,6 +45,15 @@ class MiscTransformsTC(TestCase):
         converted = ENGINE.convert(data, 'text/plain').decode().strip()
         self.assertEqual(converted, u'yo  \nzogzog')
 
+    def test_binary_html_to_text(self):
+        data = TransformData(u'<b>yo (zou  ;)</b>'.encode('utf-8'), 'text/html', 'utf8')
+        converted = ENGINE.convert(data, 'text/plain').decode().strip()
+        self.assertEqual(converted, u'**yo (zou  ;)**')
+
+        data = TransformData(u'<p>yo <br/>zogzog </p>'.encode('utf-8'), 'text/html', 'utf8')
+        converted = ENGINE.convert(data, 'text/plain').decode().strip()
+        self.assertEqual(converted, u'yo  \nzogzog')
+
     def test_html_to_text_noenc(self):
         self.skipTest('Encoding detection with chardet does not work')
         # will trigger guess_encoding, check non-utf8 encoding