Add a more general fix for #127 (CPy #20007) based on #136.

gsnedders · gsnedders · commit 9d5e39cc6bec · 2015-12-12T03:12:22.000Z
diff --git a/html5lib/inputstream.py b/html5lib/inputstream.py
@@ -1,5 +1,6 @@
 from __future__ import absolute_import, division, unicode_literals
-from six import text_type
+
+from six import text_type, get_method_self
 from six.moves import http_client
 
 import codecs
@@ -130,12 +131,13 @@ def _readFromBuffer(self, bytes):
 
 
 def HTMLInputStream(source, encoding=None, parseMeta=True, chardet=True):
-    if isinstance(source, http_client.HTTPResponse):
-        # Work around Python bug #20007: read(0) closes the connection.
-        # http://bugs.python.org/issue20007
-        isUnicode = False
-    elif hasattr(source, "read"):
-        isUnicode = isinstance(source.read(0), text_type)
+    if hasattr(source, "read"):
+        if isinstance(get_method_self(source.read), http_client.HTTPResponse):
+            # Work around Python bug #20007: read(0) closes the connection.
+            # http://bugs.python.org/issue20007
+            isUnicode = False
+        else:
+            isUnicode = isinstance(source.read(0), text_type)
     else:
         isUnicode = isinstance(source, text_type)
 
diff --git a/html5lib/tests/test_stream.py b/html5lib/tests/test_stream.py
@@ -5,7 +5,7 @@
 import codecs
 from io import BytesIO
 
-from six.moves import http_client
+from six.moves import http_client, urllib
 
 from html5lib.inputstream import (BufferedStream, HTMLInputStream,
                                   HTMLUnicodeInputStream, HTMLBinaryInputStream)
@@ -170,6 +170,21 @@ def makefile(self, _mode, _bufsize=None):
         stream = HTMLInputStream(source)
         self.assertEqual(stream.charsUntil(" "), "Text")
 
+    def test_python_issue_20007_b(self):
+        """
+        Make sure we have a work-around for Python bug #20007
+        http://bugs.python.org/issue20007
+        """
+        class FakeSocket(object):
+            def makefile(self, _mode, _bufsize=None):
+                return BytesIO(b"HTTP/1.1 200 Ok\r\n\r\nText")
+
+        source = http_client.HTTPResponse(FakeSocket())
+        source.begin()
+        wrapped = urllib.response.addinfourl(source, source.msg, "http://example.com")
+        stream = HTMLInputStream(source)
+        self.assertEqual(stream.charsUntil(" "), "Text")
+
 
 def buildTestSuite():
     return unittest.defaultTestLoader.loadTestsFromName(__name__)
