boto · jamesls · Mar 24, 2016 · Mar 23, 2016 · Mar 23, 2016
diff --git a/botocore/parsers.py b/botocore/parsers.py
@@ -203,7 +203,10 @@ def parse(self, response, shape):
         LOG.debug('Response headers: %s', response['headers'])
         LOG.debug('Response body:\n%s', response['body'])
         if response['status_code'] >= 301:
-            parsed = self._do_error_parse(response, shape)
+            if self._is_generic_error_response(response):
+                parsed = self._do_generic_error_parse(response)
+            else:
+                parsed = self._do_error_parse(response, shape)
         else:
             parsed = self._do_parse(response, shape)
         # Inject HTTPStatusCode key in the response metadata if the
@@ -213,6 +216,33 @@ def parse(self, response, shape):
                 response['status_code'])
         return parsed
 
+    def _is_generic_error_response(self, response):
+        # There are times when a service will respond with a generic
+        # error response such as:
+        # '<html><body><b>Http/1.1 Service Unavailable</b></body></html>'
+        #
+        # This can also happen if you're going through a proxy.
+        # In this case the protocol specific _do_error_parse will either
+        # fail to parse the response (in the best case) or silently succeed
+        # and treat the HTML above as an XML response and return
+        # non sensical parsed data.
+        # To prevent this case from happening we first need to check
+        # whether or not this response looks like the generic response.
+        if response['status_code'] >= 500:
+            return response['body'].strip().startswith(b'<html>')
+
+    def _do_generic_error_parse(self, response):
+        # There's not really much we can do when we get a generic
+        # html response.
+        LOG.debug("Received a non protocol specific error response from the "
+                  "service, unable to populate error code and message.")
+        return {
+            'Error': {'Code': str(response['status_code']),
+                      'Message': six.moves.http_client.responses.get(
+                          response['status_code'], '')},
+            'ResponseMetadata': {},
+        }
+
     def _do_parse(self, response, shape):
         raise NotImplementedError("%s._do_parse" % self.__class__.__name__)
 

diff --git a/tests/functional/test_s3.py b/tests/functional/test_s3.py
@@ -75,7 +75,7 @@ def test_multiple_transitions_returns_one(self):
             '		</NoncurrentVersionTransition>'
             '	</Rule>'
             '</LifecycleConfiguration>'
-        )
+        ).encode('utf-8')
         http_response.headers = {}
         self.http_session_send_mock.return_value = http_response
         s3 = self.session.create_client('s3')
@@ -115,7 +115,7 @@ def test_500_error_with_non_xml_body(self):
             'ETag: "a6d856bc171fc6aa1b236680856094e2"\r\n'
             'Content-Length: 0\r\n'
             'Server: AmazonS3\r\n'
-        )
+        ).encode('utf-8')
         http_500_response = mock.Mock()
         http_500_response.status_code = 500
         http_500_response.content = non_xml_content

diff --git a/tests/unit/test_parsers.py b/tests/unit/test_parsers.py
@@ -14,6 +14,7 @@
 import datetime
 
 from dateutil.tz import tzutc
+from nose.tools import assert_equal
 
 from botocore import parsers
 from botocore import model
@@ -635,3 +636,25 @@ def test_can_parse_route53_with_missing_message(self):
         # Even though there's no <Message /> we should
         # still populate an empty string.
         self.assertEqual(error['Message'], '')
+
+
+def test_can_handle_generic_error_message():
+    # There are times when you can get a service to respond with a generic
+    # html error page.  We should be able to handle this case.
+    for parser_cls in parsers.PROTOCOL_PARSERS.values():
+        yield _assert_parses_generic_error, parser_cls()
+
+
+def _assert_parses_generic_error(parser):
+    # There are times when you can get a service to respond with a generic
+    # html error page.  We should be able to handle this case.
+    body =  (
+        '<html><body><b>Http/1.1 Service Unavailable</b></body></html>'
+    ).encode('utf-8')
+    parsed = parser.parse({
+        'body': body, 'headers': {}, 'status_code': 503}, None)
+    assert_equal(
+        parsed,
+        {'Error': {'Code': '503', 'Message': 'Service Unavailable'},
+         'ResponseMetadata': {'HTTPStatusCode': 503}}
+    )