Fix stable gateocata-eolstable/ocata

This is a combination of 2 commits.

---

py2/3: Stop using stdlib's putrequest(); it only does ASCII

Note that this only affects the functest client.

See also: https://bugs.python.org/issue36274 and
https://bugs.python.org/issue38216

This was previously done just for py3 compatibility, but following
https://github.com/python/cpython/commit/bb8071a our stable gates are
all broken -- apparently, they're running a 2.7 pre-release?

(cherry picked from commit c0ae48ba9aafb0b91869ea3bae8da07a32088777)
(cherry picked from commit 2b4d58952cae8b174fb60529d5284c1d328e9287)

---

bufferedhttp: ensure query params are properly quoted

Recent versions of py27 [1] have begun raising InvalidURL if you try to
include non-ASCII characters in the request path. This was observed
recently in the periodic checks of stable/ocata and stable/pike. In
particular, we would spin up some in-process servers in
test.unit.proxy.test_server.TestSocketObjectVersions and do a container
listing with a prefix param that included raw (unquoted) UTF-8. This
query string would pass unmolested through the proxy, tripping the
InvalidURL error when bufferedhttp called putrequest.

More recent versions of Swift would not exhibit this particular failure,
as the listing_formats middleware would force a decoding/re-encoding of
the query string for account and container requests. However, object
requests with errant query strings would likely be able to trip the same
error.

Swift on py3 should not exhibit this behavior, as we so
thoroughly re-write the request line to avoid hitting
https://bugs.python.org/issue33973.

Now, always parse and re-encode the query string in bufferedhttp. This
prevents any errors on object requests and cleans up any callers that
might use bufferedhttp directly.

[1] Anything after https://github.com/python/cpython/commit/bb8071a;
    see https://bugs.python.org/issue30458

Closes-Bug: 1843816
Related-Change: Id3ce37aa0402e2d8dd5784ce329d7cb4fbaf700d
Related-Change: Ie648f5c04d4415f3b620fb196fa567ce7575d522
(cherry picked from commit 49f62f6ab7fd1b833e9b5bfbcaafa4b45b592d34)
(cherry picked from commit 9cc6d4138946034516fdf579ac084cb954ea6b06)

---

Change-Id: I4eafc5f057df8a3c15560ace255d05602db56ef6

3 files changed, 101 insertions, 3 deletions

diff --git a/swift/common/bufferedhttp.py b/swift/common/bufferedhttp.py
index 129460913..ecf9b155e 100644
--- a/swift/common/bufferedhttp.py
+++ b/swift/common/bufferedhttp.py
@@ -35,7 +35,7 @@ import socket
 import eventlet
 from eventlet.green.httplib import CONTINUE, HTTPConnection, HTTPMessage, \
     HTTPResponse, HTTPSConnection, _UNKNOWN
-from six.moves.urllib.parse import quote
+from six.moves.urllib.parse import quote, parse_qsl, urlencode
 import six
 
 if six.PY2:
@@ -240,6 +240,15 @@ def http_connect_raw(ipaddr, port, method, path, headers=None,
     else:
         conn = BufferedHTTPConnection('%s:%s' % (ipaddr, port))
     if query_string:
+        # Round trip to ensure proper quoting
+        if six.PY2:
+            query_string = urlencode(parse_qsl(
+                query_string, keep_blank_values=True))
+        else:
+            query_string = urlencode(
+                parse_qsl(query_string, keep_blank_values=True,
+                          encoding='latin1'),
+                encoding='latin1')
         path += '?' + query_string
     conn.path = path
     conn.putrequest(method, path, skip_host=(headers and 'Host' in headers))
diff --git a/test/functional/swift_test_client.py b/test/functional/swift_test_client.py
index 3b68d3a44..9ff8a22ba 100644
--- a/test/functional/swift_test_client.py
+++ b/test/functional/swift_test_client.py
@@ -104,6 +104,91 @@ def listing_items(method):
             items = []
 
 
+def putrequest(self, method, url, skip_host=False, skip_accept_encoding=False):
+    '''Send a request to the server.
+
+    This is mostly a regurgitation of CPython's HTTPConnection.putrequest,
+    but fixed up so we can still send arbitrary bytes in the request line
+    on py3. See also: https://bugs.python.org/issue36274
+
+    To use, swap out a HTTP(S)Connection's putrequest with something like::
+
+       conn.putrequest = putrequest.__get__(conn)
+
+    :param method: specifies an HTTP request method, e.g. 'GET'.
+    :param url: specifies the object being requested, e.g. '/index.html'.
+    :param skip_host: if True does not add automatically a 'Host:' header
+    :param skip_accept_encoding: if True does not add automatically an
+       'Accept-Encoding:' header
+    '''
+    # (Mostly) inline the HTTPConnection implementation; just fix it
+    # so we can send non-ascii request lines. For comparison, see
+    # https://github.com/python/cpython/blob/v2.7.16/Lib/httplib.py#L888-L1003
+    # and https://github.com/python/cpython/blob/v3.7.2/
+    # Lib/http/client.py#L1061-L1183
+    if self._HTTPConnection__response \
+            and self._HTTPConnection__response.isclosed():
+        self._HTTPConnection__response = None
+
+    if self._HTTPConnection__state == http_client._CS_IDLE:
+        self._HTTPConnection__state = http_client._CS_REQ_STARTED
+    else:
+        raise http_client.CannotSendRequest(self._HTTPConnection__state)
+
+    self._method = method
+    if not url:
+        url = '/'
+    self._path = url
+    request = '%s %s %s' % (method, url, self._http_vsn_str)
+    if not isinstance(request, bytes):
+        # This choice of encoding is the whole reason we copy/paste from
+        # cpython. When making backend requests, it should never be
+        # necessary; however, we have some functional tests that want
+        # to send non-ascii bytes.
+        # TODO: when https://bugs.python.org/issue36274 is resolved, make
+        # sure we fix up our API to match whatever upstream chooses to do
+        self._output(request.encode('latin1'))
+    else:
+        self._output(request)
+
+    if self._http_vsn == 11:
+        if not skip_host:
+            netloc = ''
+            if url.startswith('http'):
+                nil, netloc, nil, nil, nil = urllib.parse.urlsplit(url)
+
+            if netloc:
+                try:
+                    netloc_enc = netloc.encode("ascii")
+                except UnicodeEncodeError:
+                    netloc_enc = netloc.encode("idna")
+                self.putheader('Host', netloc_enc)
+            else:
+                if self._tunnel_host:
+                    host = self._tunnel_host
+                    port = self._tunnel_port
+                else:
+                    host = self.host
+                    port = self.port
+
+                try:
+                    host_enc = host.encode("ascii")
+                except UnicodeEncodeError:
+                    host_enc = host.encode("idna")
+
+                if host.find(':') >= 0:
+                    host_enc = b'[' + host_enc + b']'
+
+                if port == self.default_port:
+                    self.putheader('Host', host_enc)
+                else:
+                    host_enc = host_enc.decode("ascii")
+                    self.putheader('Host', "%s:%s" % (host_enc, port))
+
+        if not skip_accept_encoding:
+            self.putheader('Accept-Encoding', 'identity')
+
+
 class Connection(object):
     def __init__(self, config):
         for key in 'auth_host auth_port auth_ssl username password'.split():
@@ -125,6 +210,7 @@ class Connection(object):
         self.storage_host = None
         self.storage_port = None
         self.storage_url = None
+        self.connection = None  # until you call .http_connect()/.put_start()
 
         self.conn_class = None
 
@@ -209,6 +295,7 @@ class Connection(object):
         self.connection = self.conn_class(self.storage_host,
                                           port=self.storage_port)
         # self.connection.set_debuglevel(3)
+        self.connection.putrequest = putrequest.__get__(self.connection)
 
     def make_path(self, path=None, cfg=None):
         if path is None:
@@ -338,6 +425,7 @@ class Connection(object):
         self.connection = self.conn_class(self.storage_host,
                                           port=self.storage_port)
         # self.connection.set_debuglevel(3)
+        self.connection.putrequest = putrequest.__get__(self.connection)
         self.connection.putrequest('PUT', path)
         for key, value in headers.items():
             self.connection.putheader(key, value)
diff --git a/test/unit/common/test_bufferedhttp.py b/test/unit/common/test_bufferedhttp.py
index 9ea07ec3b..319986dd1 100644
--- a/test/unit/common/test_bufferedhttp.py
+++ b/test/unit/common/test_bufferedhttp.py
@@ -59,7 +59,8 @@ class TestBufferedHTTP(unittest.TestCase):
                     fp.flush()
                     self.assertEqual(
                         fp.readline(),
-                        'PUT /dev/%s/path/..%%25/?omg&no=%%7f HTTP/1.1\r\n' %
+                        'PUT /dev/%s/path/..%%25/?omg=&no=%%7F&%%FF=%%FF'
+                        '&no=%%25ff HTTP/1.1\r\n' %
                         expected_par)
                     headers = {}
                     line = fp.readline()
@@ -82,7 +83,7 @@ class TestBufferedHTTP(unittest.TestCase):
                         'PUT', '/path/..%/', {
                             'content-length': 7,
                             'x-header': 'value'},
-                        query_string='omg&no=%7f')
+                        query_string='omg&no=%7f&\xff=%ff&no=%25ff')
                     conn.send('REQUEST\r\n')
                     self.assertTrue(conn.sock.getsockopt(socket.IPPROTO_TCP,
                                                          socket.TCP_NODELAY))