Fix encoding bugs in Python2 (non-ASCII characters) (#80)

MiguelSR · t8y8 · commit a4e41f9bc764 · 2016-10-24T16:04:18.000-05:00
diff --git a/tableaudocumentapi/datasource.py b/tableaudocumentapi/datasource.py
@@ -31,7 +31,7 @@
 def _get_metadata_xml_for_field(root_xml, field_name):
     if "'" in field_name:
         field_name = sax.escape(field_name, {"'": "&apos;"})
-    xpath = ".//metadata-record[@class='column'][local-name='{}']".format(field_name)
+    xpath = u".//metadata-record[@class='column'][local-name='{}']".format(field_name)
     return root_xml.find(xpath)
 
 
diff --git a/tableaudocumentapi/field.py b/tableaudocumentapi/field.py
@@ -199,4 +199,9 @@ def _read_description(xmldata):
         if description is None:
             return None
 
-        return u'{}'.format(ET.tostring(description, encoding='utf-8'))  # This is necessary for py3 support
+        description_string = ET.tostring(description, encoding='utf-8')
+        # Format expects a unicode string so in Python 2 we have to do the explicit conversion
+        if isinstance(description_string, bytes):
+            description_string = description_string.decode('utf-8')
+
+        return description_string
