pythongh-102988: email parseaddr() now rejects malformed address

Detect email address parsing errors and return empty tuple to
indicate the parsing error (old API). Add an optional 'strict'
parameter to getaddresses() and parseaddr() functions. Patch by
Thomas Dwyer.

Co-Authored-By: Thomas Dwyer <[email protected]>

Doc/library/email.utils.rst

@@ -65,6 +65,11 @@ of the new API.
    *email address* parts.  Returns a tuple of that information, unless the parse
    fails, in which case a 2-tuple of ``('', '')`` is returned.
 
+   If *strict* is true, use a strict parser which rejects malformed inputs.
+
+   .. versionchanged:: 3.13
+      Add *strict* optional parameter and reject malformed inputs by default.
+
 
 .. function:: formataddr(pair, charset='utf-8')
 
@@ -82,12 +87,15 @@ of the new API.
       Added the *charset* option.
 
 
-.. function:: getaddresses(fieldvalues)
+.. function:: getaddresses(fieldvalues, *, strict=True)
 
    This method returns a list of 2-tuples of the form returned by ``parseaddr()``.
    *fieldvalues* is a sequence of header field values as might be returned by
-   :meth:`Message.get_all <email.message.Message.get_all>`.  Here's a simple
-   example that gets all the recipients of a message::
+   :meth:`Message.get_all <email.message.Message.get_all>`.
+
+   If *strict* is true, use a strict parser which rejects malformed inputs.
+
+   Here's a simple example that gets all the recipients of a message::
 
       from email.utils import getaddresses
 
@@ -97,6 +105,9 @@ of the new API.
       resent_ccs = msg.get_all('resent-cc', [])
       all_recipients = getaddresses(tos + ccs + resent_tos + resent_ccs)
 
+   .. versionchanged:: 3.13
+      Add *strict* optional parameter and reject malformed inputs by default.
+
 
 .. function:: parsedate(date)
 

Doc/whatsnew/3.13.rst

@@ -192,6 +192,17 @@ doctest
   :attr:`doctest.TestResults.skipped` attributes.
   (Contributed by Victor Stinner in :gh:`108794`.)
 
+email
+-----
+
+* :func:`email.utils.getaddresses` and :func:`email.utils.parseaddr` now return
+  ``('', '')`` 2-tuples in more situations where invalid email addresses are
+  encountered instead of potentially inaccurate values. Add optional *strict*
+  parameter to these two functions: use ``strict=False`` to get the old
+  behavior, accept malformed inputs.
+  (Contributed by Thomas Dwyer for :gh:`102988` to improve the CVE-2023-27043
+  fix.)
+
 glob
 ----
 

Lib/email/utils.py

@@ -42,6 +42,8 @@
 
 specialsre = re.compile(r'[][\\()<>@,:;".]')
 escapesre = re.compile(r'[\\"]')
+realname_comma_re = re.compile(r'"[^",]*,[^"]*"')
+
 
 def _has_surrogates(s):
     """Return True if s contains surrogate-escaped binary data."""
@@ -103,12 +105,97 @@ def formataddr(pair, charset='utf-8'):
     return address
 
 
+def _strip_quoted_realname(addr):
+    """Remove real name between quotes."""
+    pos = 0
+    start = None
+    remove = []
+    while pos < len(addr):
+        if addr[pos] == '"':
+            if start is None:
+                start = pos
+            else:
+                remove.append((start, pos))
+                start = None
+        pos += 1
+
+    if not remove:
+        return addr
+
+    result = []
+    pos = 0
+    for start, stop in remove:
+        result.append(addr[pos:start])
+        pos = stop + 1
+    result.append(addr[pos:])
+
+    return ''.join(result)
 
-def getaddresses(fieldvalues):
-    """Return a list of (REALNAME, EMAIL) for each fieldvalue."""
-    all = COMMASPACE.join(str(v) for v in fieldvalues)
-    a = _AddressList(all)
-    return a.addresslist
+
+def getaddresses(fieldvalues, *, strict=True):
+    """Return a list of (REALNAME, EMAIL) or ('','') for each fieldvalue.
+
+    When parsing fails for a fieldvalue, a 2-tuple of ('', '') is returned in
+    its place.
+
+    If strict is True, use a strict parser which rejects malformed inputs.
+    In this case, if the resulting list of parsed addresses is greater than
+    the number of fieldvalues in the input list, a parsing error has occurred
+    and consequently a list containing a single empty 2-tuple [('', '')] is
+    returned in its place. This is done to avoid invalid output.
+
+    Malformed input: getaddresses(['[email protected] <[email protected]>'])
+    Invalid output: [('', '[email protected]'), ('', '[email protected]')]
+    Safe output: [('', '')]
+
+    """
+    if not strict:
+        all = COMMASPACE.join(str(v) for v in fieldvalues)
+        a = _AddressList(all)
+        return a.addresslist
+
+    fieldvalues = [str(v) for v in fieldvalues]
+    fieldvalues = _pre_parse_validation(fieldvalues)
+    addr = COMMASPACE.join(v for v in fieldvalues)
+    a = _AddressList(addr)
+    result = _post_parse_validation(a.addresslist)
+
+    # Treat output as invalid if the number of addresses is not equal to the
+    # expected number of addresses.
+    n = 0
+    for v in fieldvalues:
+        # When a comma is used in the Real Name part it is not a deliminator.
+        # So strip those out before counting the commas
+        v = _strip_quoted_realname(v)
+        # Expected number of addresses: 1 + number of commas
+        n += 1 + v.count(',')
+    if len(result) != n:
+        return [('', '')]
+
+    return result
+
+
+def _pre_parse_validation(email_header_fields):
+    accepted_values = []
+    for v in email_header_fields:
+        s = v.replace('\\(', '').replace('\\)', '')
+        if s.count('(') != s.count(')'):
+            v = "('', '')"
+        accepted_values.append(v)
+
+    return accepted_values
+
+
+def _post_parse_validation(parsed_email_header_tuples):
+    accepted_values = []
+    # The parser would have parsed a correctly formatted domain-literal
+    # The existence of an [ after parsing indicates a parsing failure
+    for v in parsed_email_header_tuples:
+        if '[' in v[1]:
+            v = ('', '')
+        accepted_values.append(v)
+
+    return accepted_values
 
 
 def _format_timetuple_and_zone(timetuple, zone):
@@ -207,16 +294,33 @@ def parsedate_to_datetime(data):
             tzinfo=datetime.timezone(datetime.timedelta(seconds=tz)))
 
 
-def parseaddr(addr):
+def parseaddr(addr, *, strict=True):
     """
     Parse addr into its constituent realname and email address parts.
 
     Return a tuple of realname and email address, unless the parse fails, in
     which case return a 2-tuple of ('', '').
+
+    If strict is True, use a strict parser which rejects malformed inputs.
     """
-    addrs = _AddressList(addr).addresslist
-    if not addrs:
-        return '', ''
+    if not strict:
+        addrs = _AddressList(addr).addresslist
+        if not addrs:
+            return ('', '')
+        return addrs[0]
+
+    if isinstance(addr, list):
+        addr = addr[0]
+
+    if not isinstance(addr, str):
+        return ('', '')
+
+    addr = _pre_parse_validation([addr])[0]
+    addrs = _post_parse_validation(_AddressList(addr).addresslist)
+
+    if not addrs or len(addrs) > 1:
+        return ('', '')
+
     return addrs[0]
 
 

Lib/test/test_email/test_email.py

@@ -3320,32 +3320,116 @@ def test_getaddresses(self):
            [('Al Person', '[email protected]'),
             ('Bud Person', '[email protected]')])
 
-    def test_getaddresses_comma_in_name(self):
-        """GH-106669 regression test."""
-        self.assertEqual(
-            utils.getaddresses(
-                [
-                    '"Bud, Person" <[email protected]>',
-                    '[email protected] (Al Person)',
-                    '"Mariusz Felisiak" <[email protected]>',
-                ]
-            ),
-            [
-                ('Bud, Person', '[email protected]'),
-                ('Al Person', '[email protected]'),
-                ('Mariusz Felisiak', '[email protected]'),
-            ],
-        )
+    def test_parsing_errors(self):
+        """Test for parsing errors from CVE-2023-27043 and CVE-2019-16056"""
+        alice = '[email protected]'
+        bob = '[email protected]'
+        empty = ('', '')
+
+        # Test utils.getaddresses() and utils.parseaddr() on malformed email
+        # addresses: default behavior (strict=True) rejects malformed address,
+        # and strict=True which tolerates malformed address.
+        for invalid_separator, expected_non_strict in (
+            ('(', [(f'<{bob}>', alice)]),
+            (')', [('', alice), empty, ('', bob)]),
+            ('<', [('', alice), empty, ('', bob), empty]),
+            ('>', [('', alice), empty, ('', bob)]),
+            ('[', [('', f'{alice}[<{bob}>]')]),
+            (']', [('', alice), empty, ('', bob)]),
+            ('@', [empty, empty, ('', bob)]),
+            (';', [('', alice), empty, ('', bob)]),
+            (':', [('', alice), ('', bob)]),
+            ('.', [('', alice + '.'), ('', bob)]),
+            ('"', [('', alice), ('', f'<{bob}>')]),
+        ):
+            address = f'{alice}{invalid_separator}<{bob}>'
+            with self.subTest(address=address):
+                self.assertEqual(utils.getaddresses([address]),
+                                 [empty])
+                self.assertEqual(utils.getaddresses([address], strict=False),
+                                 expected_non_strict)
+
+                self.assertEqual(utils.parseaddr([address]),
+                                 empty)
+                self.assertEqual(utils.parseaddr([address], strict=False),
+                                 ('', address))
+
+        # Comma (',') is treated differently depending on strict parameter.
+        # Comma without quotes.
+        address = f'{alice},<{bob}>'
+        self.assertEqual(utils.getaddresses([address]),
+                         [('', alice), ('', bob)])
+        self.assertEqual(utils.getaddresses([address], strict=False),
+                         [('', alice), ('', bob)])
+        self.assertEqual(utils.parseaddr([address]),
+                         empty)
+        self.assertEqual(utils.parseaddr([address], strict=False),
+                         ('', address))
+
+        # Comma with quotes.
+        address = '"Alice, [email protected]" <[email protected]>'
+        expected_strict = ('Alice, [email protected]', '[email protected]')
+        self.assertEqual(utils.getaddresses([address]), [expected_strict])
+        self.assertEqual(utils.getaddresses([address], strict=False), [expected_strict])
+        self.assertEqual(utils.parseaddr([address]), expected_strict)
+        self.assertEqual(utils.parseaddr([address], strict=False),
+                         ('', address))
+
+        # Two addresses with quotes separated by comma.
+        address = '"Jane Doe" <[email protected]>, "John Doe" <[email protected]>'
+        self.assertEqual(utils.getaddresses([address]),
+                         [('Jane Doe', '[email protected]'),
+                          ('John Doe', '[email protected]')])
+        self.assertEqual(utils.getaddresses([address], strict=False),
+                         [('Jane Doe', '[email protected]'),
+                          ('John Doe', '[email protected]')])
+        self.assertEqual(utils.parseaddr([address]), empty)
+        self.assertEqual(utils.parseaddr([address], strict=False),
+                         ('', address))
 
     def test_getaddresses_nasty(self):
-        eq = self.assertEqual
-        eq(utils.getaddresses(['foo: ;']), [('', '')])
-        eq(utils.getaddresses(
-           ['[]*-- =~$']),
-           [('', ''), ('', ''), ('', '*--')])
-        eq(utils.getaddresses(
-           ['foo: ;', '"Jason R. Mastaler" <[email protected]>']),
-           [('', ''), ('Jason R. Mastaler', '[email protected]')])
+        for addresses, expected in (
+            (['"Sürname, Firstname" <[email protected]>'],
+             [('Sürname, Firstname', '[email protected]')]),
+
+            (['foo: ;'],
+             [('', '')]),
+
+            (['foo: ;', '"Jason R. Mastaler" <[email protected]>'],
+             [('', ''), ('Jason R. Mastaler', '[email protected]')]),
+
+            ([r'Pete(A nice \) chap) <pete(his account)@silly.test(his host)>'],
+             [('Pete (A nice ) chap his account his host)', '[email protected]')]),
+
+            (['(Empty list)(start)Undisclosed recipients  :(nobody(I know))'],
+             [('', '')]),
+
+            (['Mary <@machine.tld:[email protected]>, , jdoe@test   . example'],
+             [('Mary', '[email protected]'), ('', ''), ('', '[email protected]')]),
+
+            (['John Doe <jdoe@machine(comment).  example>'],
+             [('John Doe (comment)', '[email protected]')]),
+
+            (['"Mary Smith: Personal Account" <[email protected]>'],
+             [('Mary Smith: Personal Account', '[email protected]')]),
+
+            (['Undisclosed recipients:;'],
+             [('', '')]),
+
+            ([r'<[email protected]>, "Giant; \"Big\" Box" <[email protected]>'],
+             [('', '[email protected]'), ('Giant; "Big" Box', '[email protected]')]),
+        ):
+            with self.subTest(addresses=addresses):
+                self.assertEqual(utils.getaddresses(addresses),
+                                 expected)
+                self.assertEqual(utils.getaddresses(addresses, strict=False),
+                                 expected)
+
+        addresses = ['[]*-- =~$']
+        self.assertEqual(utils.getaddresses(addresses),
+                         [('', '')])
+        self.assertEqual(utils.getaddresses(addresses, strict=False),
+                         [('', ''), ('', ''), ('', '*--')])
 
     def test_getaddresses_embedded_comment(self):
         """Test proper handling of a nested comment"""

Misc/NEWS.d/next/Library/2023-10-20-15-28-08.gh-issue-102988.dStNO7.rst

@@ -0,0 +1,6 @@
+:func:`email.utils.getaddresses` and :func:`email.utils.parseaddr` now
+return ``('', '')`` 2-tuples in more situations where invalid email
+addresses are encountered instead of potentially inaccurate values. Add
+optional *strict* parameter to these two functions: use ``strict=False`` to
+get the old behavior, accept malformed inputs. Patch by Thomas Dwyer to
+improve the CVE-2023-27043 fix.