Switch codebase to use sanitized_Request instead of

[youtube-dl] / youtube_dl / compat.py

diff --git a/youtube_dl/compat.py b/youtube_dl/compat.py
index cf10835ca55d11e0d8803c550b41a198ac039827..a3e85264acda8dbefe0883ea50c901302d01d29f 100644 (file)
--- a/youtube_dl/compat.py
+++ b/youtube_dl/compat.py
@@ -216,10 +216,19 @@ except ImportError:  # Python 2.6
 if sys.version_info[0] >= 3:
     compat_etree_fromstring = xml.etree.ElementTree.fromstring
 else:
-    # on python 2.x the the attributes of a node are str objects instead of
-    # unicode
+    # python 2.x tries to encode unicode strings with ascii (see the
+    # XMLParser._fixtext method)
     etree = xml.etree.ElementTree
 
+    try:
+        _etree_iter = etree.Element.iter
+    except AttributeError:  # Python <=2.6
+        def _etree_iter(root):
+            for el in root.findall('*'):
+                yield el
+                for sub in _etree_iter(el):
+                    yield sub
+
     # on 2.6 XML doesn't have a parser argument, function copied from CPython
     # 2.7 source
     def _XML(text, parser=None):
@@ -231,11 +240,16 @@ else:
     def _element_factory(*args, **kwargs):
         el = etree.Element(*args, **kwargs)
         for k, v in el.items():
-            el.set(k, v.decode('utf-8'))
+            if isinstance(v, bytes):
+                el.set(k, v.decode('utf-8'))
         return el
 
     def compat_etree_fromstring(text):
-        return _XML(text, parser=etree.XMLParser(target=etree.TreeBuilder(element_factory=_element_factory)))
+        doc = _XML(text, parser=etree.XMLParser(target=etree.TreeBuilder(element_factory=_element_factory)))
+        for el in _etree_iter(doc):
+            if el.text is not None and isinstance(el.text, bytes):
+                el.text = el.text.decode('utf-8')
+        return doc
 
 try:
     from urllib.parse import parse_qs as compat_parse_qs