From 4dc2cae3abd75f386374d0635d00443b897d0672 Mon Sep 17 00:00:00 2001
From: "Miss Islington (bot)"
 <31488909+miss-islington@users.noreply.github.com>
Date: Wed, 22 Jun 2022 01:42:52 -0700
Subject: [PATCH] gh-87389: Fix an open redirection vulnerability in
 http.server. (GH-93879) (GH-94094)

Fix an open redirection vulnerability in the `http.server` module when
an URI path starts with `//` that could produce a 301 Location header
with a misleading target.  Vulnerability discovered, and logic fix
proposed, by Hamza Avvan (@hamzaavvan).

Test and comments authored by Gregory P. Smith [Google].
(cherry picked from commit 4abab6b603dd38bec1168e9a37c40a48ec89508e)

Co-authored-by: Gregory P. Smith <greg@krypto.org>

Signed-off-by: Riyaz Khan <Riyaz.Khan@kpit.com>

CVE: CVE-2021-28861

Upstream-Status: Backport [https://github.com/python/cpython/commit/4dc2cae3abd75f386374d0635d00443b897d0672]

---
 Lib/http/server.py                            |  7 +++
 Lib/test/test_httpservers.py                  | 53 ++++++++++++++++++-
 ...2-06-15-20-09-23.gh-issue-87389.QVaC3f.rst |  3 ++
 3 files changed, 61 insertions(+), 2 deletions(-)
 create mode 100644 Misc/NEWS.d/next/Security/2022-06-15-20-09-23.gh-issue-87389.QVaC3f.rst

diff --git a/Lib/http/server.py b/Lib/http/server.py
index 38f7accad7a3..39de35458c38 100644
--- a/Lib/http/server.py
+++ b/Lib/http/server.py
@@ -332,6 +332,13 @@ def parse_request(self):
                 return False
         self.command, self.path = command, path
 
+        # gh-87389: The purpose of replacing '//' with '/' is to protect
+        # against open redirect attacks possibly triggered if the path starts
+        # with '//' because http clients treat //path as an absolute URI
+        # without scheme (similar to http://path) rather than a path.
+        if self.path.startswith('//'):
+            self.path = '/' + self.path.lstrip('/')  # Reduce to a single /
+
         # Examine the headers and look for a Connection directive.
         try:
             self.headers = http.client.parse_headers(self.rfile,
diff --git a/Lib/test/test_httpservers.py b/Lib/test/test_httpservers.py
index 87d4924a34b3..fb026188f0b4 100644
--- a/Lib/test/test_httpservers.py
+++ b/Lib/test/test_httpservers.py
@@ -330,7 +330,7 @@ class request_handler(NoLogRequestHandler, SimpleHTTPRequestHandler):
         pass
 
     def setUp(self):
-        BaseTestCase.setUp(self)
+        super().setUp()
         self.cwd = os.getcwd()
         basetempdir = tempfile.gettempdir()
         os.chdir(basetempdir)
@@ -358,7 +358,7 @@ def tearDown(self):
             except:
                 pass
         finally:
-            BaseTestCase.tearDown(self)
+            super().tearDown()
 
     def check_status_and_reason(self, response, status, data=None):
         def close_conn():
@@ -414,6 +414,55 @@ def test_undecodable_filename(self):
         self.check_status_and_reason(response, HTTPStatus.OK,
                                      data=support.TESTFN_UNDECODABLE)
 
+    def test_get_dir_redirect_location_domain_injection_bug(self):
+        """Ensure //evil.co/..%2f../../X does not put //evil.co/ in Location.
+
+        //netloc/ in a Location header is a redirect to a new host.
+        https://github.com/python/cpython/issues/87389
+
+        This checks that a path resolving to a directory on our server cannot
+        resolve into a redirect to another server.
+        """
+        os.mkdir(os.path.join(self.tempdir, 'existing_directory'))
+        url = f'/python.org/..%2f..%2f..%2f..%2f..%2f../%0a%0d/../{self.tempdir_name}/existing_directory'
+        expected_location = f'{url}/'  # /python.org.../ single slash single prefix, trailing slash
+        # Canonicalizes to /tmp/tempdir_name/existing_directory which does
+        # exist and is a dir, triggering the 301 redirect logic.
+        response = self.request(url)
+        self.check_status_and_reason(response, HTTPStatus.MOVED_PERMANENTLY)
+        location = response.getheader('Location')
+        self.assertEqual(location, expected_location, msg='non-attack failed!')
+
+        # //python.org... multi-slash prefix, no trailing slash
+        attack_url = f'/{url}'
+        response = self.request(attack_url)
+        self.check_status_and_reason(response, HTTPStatus.MOVED_PERMANENTLY)
+        location = response.getheader('Location')
+        self.assertFalse(location.startswith('//'), msg=location)
+        self.assertEqual(location, expected_location,
+                msg='Expected Location header to start with a single / and '
+                'end with a / as this is a directory redirect.')
+
+        # ///python.org... triple-slash prefix, no trailing slash
+        attack3_url = f'//{url}'
+        response = self.request(attack3_url)
+        self.check_status_and_reason(response, HTTPStatus.MOVED_PERMANENTLY)
+        self.assertEqual(response.getheader('Location'), expected_location)
+
+        # If the second word in the http request (Request-URI for the http
+        # method) is a full URI, we don't worry about it, as that'll be parsed
+        # and reassembled as a full URI within BaseHTTPRequestHandler.send_head
+        # so no errant scheme-less //netloc//evil.co/ domain mixup can happen.
+        attack_scheme_netloc_2slash_url = f'https://pypi.org/{url}'
+        expected_scheme_netloc_location = f'{attack_scheme_netloc_2slash_url}/'
+        response = self.request(attack_scheme_netloc_2slash_url)
+        self.check_status_and_reason(response, HTTPStatus.MOVED_PERMANENTLY)
+        location = response.getheader('Location')
+        # We're just ensuring that the scheme and domain make it through, if
+        # there are or aren't multiple slashes at the start of the path that
+        # follows that isn't important in this Location: header.
+        self.assertTrue(location.startswith('https://pypi.org/'), msg=location)
+
     def test_get(self):
         #constructs the path relative to the root directory of the HTTPServer
         response = self.request(self.base_url + '/test')
diff --git a/Misc/NEWS.d/next/Security/2022-06-15-20-09-23.gh-issue-87389.QVaC3f.rst b/Misc/NEWS.d/next/Security/2022-06-15-20-09-23.gh-issue-87389.QVaC3f.rst
new file mode 100644
index 000000000000..029d437190de
--- /dev/null
+++ b/Misc/NEWS.d/next/Security/2022-06-15-20-09-23.gh-issue-87389.QVaC3f.rst
@@ -0,0 +1,3 @@
+:mod:`http.server`: Fix an open redirection vulnerability in the HTTP server
+when an URI path starts with ``//``.  Vulnerability discovered, and initial
+fix proposed, by Hamza Avvan.