Fix html5lib#127, add a work-around for CPython bug 20007.

The bug has been fixed upstream, but this works around it
on older Python versions.
This does not change any behavior on Python versions with the fix.

Author: SimonSapin

html5lib/inputstream.py

@@ -1,5 +1,6 @@
 from __future__ import absolute_import, division, unicode_literals
 from six import text_type
+from six.moves import http_client
 
 import codecs
 import re
@@ -118,7 +119,11 @@ def _readFromBuffer(self, bytes):
 
 
 def HTMLInputStream(source, encoding=None, parseMeta=True, chardet=True):
-    if hasattr(source, "read"):
+    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)
     else:
         isUnicode = isinstance(source, text_type)

html5lib/tests/test_stream.py

@@ -5,6 +5,8 @@
 import codecs
 from io import BytesIO
 
+from six.moves import http_client
+
 from html5lib.inputstream import (BufferedStream, HTMLInputStream,
                                   HTMLUnicodeInputStream, HTMLBinaryInputStream)
 
@@ -154,6 +156,20 @@ def test_position2(self):
         self.assertEqual(stream.char(), "d")
         self.assertEqual(stream.position(), (2, 1))
 
+    def test_python_issue_20007(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()
+        stream = HTMLInputStream(source)
+        self.assertEqual(stream.charsUntil(" "), "Text")
+
 
 def buildTestSuite():
     return unittest.defaultTestLoader.loadTestsFromName(__name__)