1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
|
From 7e200e0763f5b71c199aaf98bd5588f291585619 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Miro=20Hron=C4=8Dok?= <miro@hroncok.cz>
Date: Tue, 7 May 2019 17:28:47 +0200
Subject: [PATCH] bpo-30458: Disallow control chars in http URLs. (GH-12755)
(GH-13154)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Disallow control chars in http URLs in urllib.urlopen. This addresses a potential security problem for applications that do not sanity check their URLs where http request headers could be injected.
Disable https related urllib tests on a build without ssl (GH-13032)
These tests require an SSL enabled build. Skip these tests when python is built without SSL to fix test failures.
Use http.client.InvalidURL instead of ValueError as the new error case's exception. (GH-13044)
Backport Co-Authored-By: Miro Hrončok <miro@hroncok.cz>
Upstream-Status: Backport[https://github.com/python/cpython/commit/7e200e0763f5b71c199aaf98bd5588f291585619]
CVE: CVE-2019-9740
CVE: CVE-2019-9947
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
---
Lib/http/client.py | 15 ++++++
Lib/test/test_urllib.py | 53 +++++++++++++++++++
Lib/test/test_xmlrpc.py | 7 ++-
.../2019-04-10-08-53-30.bpo-30458.51E-DA.rst | 1 +
4 files changed, 75 insertions(+), 1 deletion(-)
create mode 100644 Misc/NEWS.d/next/Security/2019-04-10-08-53-30.bpo-30458.51E-DA.rst
diff --git a/Lib/http/client.py b/Lib/http/client.py
index 1de151c38e92..2afd452fe30f 100644
--- a/Lib/http/client.py
+++ b/Lib/http/client.py
@@ -140,6 +140,16 @@
_is_legal_header_name = re.compile(rb'[^:\s][^:\r\n]*').fullmatch
_is_illegal_header_value = re.compile(rb'\n(?![ \t])|\r(?![ \t\n])').search
+# These characters are not allowed within HTTP URL paths.
+# See https://tools.ietf.org/html/rfc3986#section-3.3 and the
+# https://tools.ietf.org/html/rfc3986#appendix-A pchar definition.
+# Prevents CVE-2019-9740. Includes control characters such as \r\n.
+# We don't restrict chars above \x7f as putrequest() limits us to ASCII.
+_contains_disallowed_url_pchar_re = re.compile('[\x00-\x20\x7f]')
+# Arguably only these _should_ allowed:
+# _is_allowed_url_pchars_re = re.compile(r"^[/!$&'()*+,;=:@%a-zA-Z0-9._~-]+$")
+# We are more lenient for assumed real world compatibility purposes.
+
# We always set the Content-Length header for these methods because some
# servers will otherwise respond with a 411
_METHODS_EXPECTING_BODY = {'PATCH', 'POST', 'PUT'}
@@ -1101,6 +1111,11 @@ def putrequest(self, method, url, skip_host=False,
self._method = method
if not url:
url = '/'
+ # Prevent CVE-2019-9740.
+ match = _contains_disallowed_url_pchar_re.search(url)
+ if match:
+ raise InvalidURL(f"URL can't contain control characters. {url!r} "
+ f"(found at least {match.group()!r})")
request = '%s %s %s' % (method, url, self._http_vsn_str)
# Non-ASCII characters should have been eliminated earlier
diff --git a/Lib/test/test_urllib.py b/Lib/test/test_urllib.py
index 2ac73b58d832..7214492eca9d 100644
--- a/Lib/test/test_urllib.py
+++ b/Lib/test/test_urllib.py
@@ -329,6 +329,59 @@ def test_willclose(self):
finally:
self.unfakehttp()
+ @unittest.skipUnless(ssl, "ssl module required")
+ def test_url_with_control_char_rejected(self):
+ for char_no in list(range(0, 0x21)) + [0x7f]:
+ char = chr(char_no)
+ schemeless_url = f"//localhost:7777/test{char}/"
+ self.fakehttp(b"HTTP/1.1 200 OK\r\n\r\nHello.")
+ try:
+ # We explicitly test urllib.request.urlopen() instead of the top
+ # level 'def urlopen()' function defined in this... (quite ugly)
+ # test suite. They use different url opening codepaths. Plain
+ # urlopen uses FancyURLOpener which goes via a codepath that
+ # calls urllib.parse.quote() on the URL which makes all of the
+ # above attempts at injection within the url _path_ safe.
+ escaped_char_repr = repr(char).replace('\\', r'\\')
+ InvalidURL = http.client.InvalidURL
+ with self.assertRaisesRegex(
+ InvalidURL, f"contain control.*{escaped_char_repr}"):
+ urllib.request.urlopen(f"http:{schemeless_url}")
+ with self.assertRaisesRegex(
+ InvalidURL, f"contain control.*{escaped_char_repr}"):
+ urllib.request.urlopen(f"https:{schemeless_url}")
+ # This code path quotes the URL so there is no injection.
+ resp = urlopen(f"http:{schemeless_url}")
+ self.assertNotIn(char, resp.geturl())
+ finally:
+ self.unfakehttp()
+
+ @unittest.skipUnless(ssl, "ssl module required")
+ def test_url_with_newline_header_injection_rejected(self):
+ self.fakehttp(b"HTTP/1.1 200 OK\r\n\r\nHello.")
+ host = "localhost:7777?a=1 HTTP/1.1\r\nX-injected: header\r\nTEST: 123"
+ schemeless_url = "//" + host + ":8080/test/?test=a"
+ try:
+ # We explicitly test urllib.request.urlopen() instead of the top
+ # level 'def urlopen()' function defined in this... (quite ugly)
+ # test suite. They use different url opening codepaths. Plain
+ # urlopen uses FancyURLOpener which goes via a codepath that
+ # calls urllib.parse.quote() on the URL which makes all of the
+ # above attempts at injection within the url _path_ safe.
+ InvalidURL = http.client.InvalidURL
+ with self.assertRaisesRegex(
+ InvalidURL, r"contain control.*\\r.*(found at least . .)"):
+ urllib.request.urlopen(f"http:{schemeless_url}")
+ with self.assertRaisesRegex(InvalidURL, r"contain control.*\\n"):
+ urllib.request.urlopen(f"https:{schemeless_url}")
+ # This code path quotes the URL so there is no injection.
+ resp = urlopen(f"http:{schemeless_url}")
+ self.assertNotIn(' ', resp.geturl())
+ self.assertNotIn('\r', resp.geturl())
+ self.assertNotIn('\n', resp.geturl())
+ finally:
+ self.unfakehttp()
+
def test_read_0_9(self):
# "0.9" response accepted (but not "simple responses" without
# a status line)
diff --git a/Lib/test/test_xmlrpc.py b/Lib/test/test_xmlrpc.py
index 32263f7f0b3b..0e002ec4ef9f 100644
--- a/Lib/test/test_xmlrpc.py
+++ b/Lib/test/test_xmlrpc.py
@@ -945,7 +945,12 @@ def test_unicode_host(self):
def test_partial_post(self):
# Check that a partial POST doesn't make the server loop: issue #14001.
conn = http.client.HTTPConnection(ADDR, PORT)
- conn.request('POST', '/RPC2 HTTP/1.0\r\nContent-Length: 100\r\n\r\nbye')
+ conn.send('POST /RPC2 HTTP/1.0\r\n'
+ 'Content-Length: 100\r\n\r\n'
+ 'bye HTTP/1.1\r\n'
+ f'Host: {ADDR}:{PORT}\r\n'
+ 'Accept-Encoding: identity\r\n'
+ 'Content-Length: 0\r\n\r\n'.encode('ascii'))
conn.close()
def test_context_manager(self):
diff --git a/Misc/NEWS.d/next/Security/2019-04-10-08-53-30.bpo-30458.51E-DA.rst b/Misc/NEWS.d/next/Security/2019-04-10-08-53-30.bpo-30458.51E-DA.rst
new file mode 100644
index 000000000000..ed8027fb4d64
--- /dev/null
+++ b/Misc/NEWS.d/next/Security/2019-04-10-08-53-30.bpo-30458.51E-DA.rst
@@ -0,0 +1 @@
+Address CVE-2019-9740 by disallowing URL paths with embedded whitespace or control characters through into the underlying http client request. Such potentially malicious header injection URLs now cause an http.client.InvalidURL exception to be raised.
|
