Fix the way *_encoding properties are handled

See #334
Kozea · Jul 15, 2016 · da2e3b3 · da2e3b3
1 parent 5471ac9
commit da2e3b3
Showing 1 changed file with 8 additions and 6 deletions.
diff --git a/weasyprint/__init__.py b/weasyprint/__init__.py
@@ -83,13 +83,15 @@ def __init__(self, guess=None, filename=None, url=None, file_obj=None,
             if source_type == 'tree':
                 result = source
             else:
-                if not encoding:
-                    encoding = protocol_encoding
                 if isinstance(source, unicode):
-                    encoding = None
-                result = html5lib.parse(
-                    source, treebuilder='lxml', likely_encoding=encoding,
-                    namespaceHTMLElements=False)
+                    result = html5lib.parse(
+                        source, treebuilder='lxml',
+                        namespaceHTMLElements=False)
+                else:
+                    result = html5lib.parse(
+                        source, treebuilder='lxml', override_encoding=encoding,
+                        transport_encoding=protocol_encoding,
+                        namespaceHTMLElements=False)
                 assert result
         base_url = find_base_url(result, base_url)
         if hasattr(result, 'getroot'):