[Python-checkins] bpo-39073: validate Address parts to disallow CRLF (#19007) (#20450)

[Victor Stinner]

https://github.com/python/cpython/commit/f91a0b6df14d6c5133fe3d5889fad7d84fc0c046
commit: f91a0b6df14d6c5133fe3d5889fad7d84fc0c046
branch: 3.5
author: Victor Stinner <vstinner at python.org>
committer: GitHub <noreply at github.com>
date: 2020-06-12T08:33:19-07:00
summary:

bpo-39073: validate Address parts to disallow CRLF (#19007) (#20450)

Disallow CR or LF in email.headerregistry.Address arguments to guard against header injection attacks.

(cherry picked from commit 614f17211c5fc0e5b828be1d3320661d1038fe8f)

files:
A Misc/NEWS.d/next/Security/2020-03-15-01-28-36.bpo-39073.6Szd3i.rst
M Lib/email/headerregistry.py
M Lib/test/test_email/test_headerregistry.py

diff --git a/Lib/email/headerregistry.py b/Lib/email/headerregistry.py
index 0fc2231e5cbd2..fafd489b2c370 100644
--- a/Lib/email/headerregistry.py
+++ b/Lib/email/headerregistry.py
@@ -31,6 +31,11 @@ def __init__(self, display_name='', username='', domain='', addr_spec=None):
         without any Content Transfer Encoding.
 
         """
+
+        inputs = ''.join(filter(None, (display_name, username, domain, addr_spec)))
+        if '\r' in inputs or '\n' in inputs:
+            raise ValueError("invalid arguments; address parts cannot contain CR or LF")
+
         # This clause with its potential 'raise' may only happen when an
         # application program creates an Address object using an addr_spec
         # keyword.  The email library code itself must always supply username
diff --git a/Lib/test/test_email/test_headerregistry.py b/Lib/test/test_email/test_headerregistry.py
index 55ecdea9aacb5..d6f8119b02f3f 100644
--- a/Lib/test/test_email/test_headerregistry.py
+++ b/Lib/test/test_email/test_headerregistry.py
@@ -1415,6 +1415,25 @@ def test_il8n(self):
     #    with self.assertRaises(ValueError):
     #        Address('foo', 'wők', 'example.com')
 
+    def test_crlf_in_constructor_args_raises(self):
+        cases = (
+            dict(display_name='foo\r'),
+            dict(display_name='foo\n'),
+            dict(display_name='foo\r\n'),
+            dict(domain='example.com\r'),
+            dict(domain='example.com\n'),
+            dict(domain='example.com\r\n'),
+            dict(username='wok\r'),
+            dict(username='wok\n'),
+            dict(username='wok\r\n'),
+            dict(addr_spec='wok at example.com\r'),
+            dict(addr_spec='wok at example.com\n'),
+            dict(addr_spec='wok at example.com\r\n')
+        )
+        for kwargs in cases:
+            with self.subTest(kwargs=kwargs), self.assertRaisesRegex(ValueError, "invalid arguments"):
+                Address(**kwargs)
+
     def test_non_ascii_username_in_addr_spec_raises(self):
         with self.assertRaises(ValueError):
             Address('foo', addr_spec='wők at example.com')
diff --git a/Misc/NEWS.d/next/Security/2020-03-15-01-28-36.bpo-39073.6Szd3i.rst b/Misc/NEWS.d/next/Security/2020-03-15-01-28-36.bpo-39073.6Szd3i.rst
new file mode 100644
index 0000000000000..6c9447b897bf6
--- /dev/null
+++ b/Misc/NEWS.d/next/Security/2020-03-15-01-28-36.bpo-39073.6Szd3i.rst
@@ -0,0 +1 @@
+Disallow CR or LF in email.headerregistry.Address arguments to guard against header injection attacks.