bpo-37764: Fix infinite loop when parsing unstructured email headers. (GH-15239)

Fixes a case in which email._header_value_parser.get_unstructured hangs the system for some invalid headers. This covers the cases in which the header contains either:
- a case without trailing whitespace
- an invalid encoded word

https://bugs.python.org/issue37764

This fix should also be backported to 3.7 and 3.8


https://bugs.python.org/issue37764

5 files changed, 55 insertions, 3 deletions

diff --git a/Lib/email/_header_value_parser.py b/Lib/email/_header_value_parser.py
index b500394..16c1990 100644
--- a/Lib/email/_header_value_parser.py
+++ b/Lib/email/_header_value_parser.py
@@ -935,6 +935,10 @@ class EWWhiteSpaceTerminal(WhiteSpaceTerminal):
         return ''
 
 
+class _InvalidEwError(errors.HeaderParseError):
+    """Invalid encoded word found while parsing headers."""
+
+
 # XXX these need to become classes and used as instances so
 # that a program can't change them in a parse tree and screw
 # up other parse trees.  Maybe should have  tests for that, too.
@@ -1039,7 +1043,10 @@ def get_encoded_word(value):
         raise errors.HeaderParseError(
             "expected encoded word but found {}".format(value))
     remstr = ''.join(remainder)
-    if len(remstr) > 1 and remstr[0] in hexdigits and remstr[1] in hexdigits:
+    if (len(remstr) > 1 and
+        remstr[0] in hexdigits and
+        remstr[1] in hexdigits and
+        tok.count('?') < 2):
         # The ? after the CTE was followed by an encoded word escape (=XX).
         rest, *remainder = remstr.split('?=', 1)
         tok = tok + '?=' + rest
@@ -1051,7 +1058,7 @@ def get_encoded_word(value):
     try:
         text, charset, lang, defects = _ew.decode('=?' + tok + '?=')
     except ValueError:
-        raise errors.HeaderParseError(
+        raise _InvalidEwError(
             "encoded word format invalid: '{}'".format(ew.cte))
     ew.charset = charset
     ew.lang = lang
@@ -1101,9 +1108,12 @@ def get_unstructured(value):
             token, value = get_fws(value)
             unstructured.append(token)
             continue
+        valid_ew = True
         if value.startswith('=?'):
             try:
                 token, value = get_encoded_word(value)
+            except _InvalidEwError:
+                valid_ew = False
             except errors.HeaderParseError:
                 # XXX: Need to figure out how to register defects when
                 # appropriate here.
@@ -1125,7 +1135,10 @@ def get_unstructured(value):
         # Split in the middle of an atom if there is a rfc2047 encoded word
         # which does not have WSP on both sides. The defect will be registered
         # the next time through the loop.
-        if rfc2047_matcher.search(tok):
+        # This needs to only be performed when the encoded word is valid;
+        # otherwise, performing it on an invalid encoded word can cause
+        # the parser to go in an infinite loop.
+        if valid_ew and rfc2047_matcher.search(tok):
             tok, *remainder = value.partition('=?')
         vtext = ValueTerminal(tok, 'vtext')
         _validate_xtext(vtext)
diff --git a/Lib/test/test_email/test__header_value_parser.py b/Lib/test/test_email/test__header_value_parser.py
index b3e6b26..058d902 100644
--- a/Lib/test/test_email/test__header_value_parser.py
+++ b/Lib/test/test_email/test__header_value_parser.py
@@ -383,6 +383,22 @@ class TestParser(TestParserMixin, TestEmailBase):
             [errors.InvalidHeaderDefect],
             '')
 
+    def test_get_unstructured_without_trailing_whitespace_hang_case(self):
+        self._test_get_x(self._get_unst,
+            '=?utf-8?q?somevalue?=aa',
+            'somevalueaa',
+            'somevalueaa',
+            [errors.InvalidHeaderDefect],
+            '')
+
+    def test_get_unstructured_invalid_ew(self):
+        self._test_get_x(self._get_unst,
+            '=?utf-8?q?=somevalue?=',
+            '=?utf-8?q?=somevalue?=',
+            '=?utf-8?q?=somevalue?=',
+            [],
+            '')
+
     # get_qp_ctext
 
     def test_get_qp_ctext_only(self):
diff --git a/Lib/test/test_email/test_email.py b/Lib/test/test_email/test_email.py
index ae96258..8ec3919 100644
--- a/Lib/test/test_email/test_email.py
+++ b/Lib/test/test_email/test_email.py
@@ -5381,6 +5381,27 @@ Content-Type: application/x-foo;
         eq(language, 'en-us')
         eq(s, 'My Document For You')
 
+    def test_should_not_hang_on_invalid_ew_messages(self):
+        messages = ["""From: user@host.com
+To: user@host.com
+Bad-Header:
+ =?us-ascii?Q?LCSwrV11+IB0rSbSker+M9vWR7wEDSuGqmHD89Gt=ea0nJFSaiz4vX3XMJPT4vrE?=
+ =?us-ascii?Q?xGUZeOnp0o22pLBB7CYLH74Js=wOlK6Tfru2U47qR?=
+ =?us-ascii?Q?72OfyEY2p2=2FrA9xNFyvH+fBTCmazxwzF8nGkK6D?=
+
+Hello!
+""", """From: ����� �������� <xxx@xxx>
+To: "xxx" <xxx@xxx>
+Subject:   ��� ���������� ����� ����� � ��������� �� ����
+MIME-Version: 1.0
+Content-Type: text/plain; charset="windows-1251";
+Content-Transfer-Encoding: 8bit
+
+�� ����� � ���� ������ ��� ��������
+"""]
+        for m in messages:
+            with self.subTest(m=m):
+                msg = email.message_from_string(m)
 
 
 # Tests to ensure that signed parts of an email are completely preserved, as
diff --git a/Misc/ACKS b/Misc/ACKS
index e9ae0ed..ce8b144 100644
--- a/Misc/ACKS
+++ b/Misc/ACKS
@@ -1336,6 +1336,7 @@ Burton Radons
 Abhilash Raj
 Shorya Raj
 Dhushyanth Ramasamy
+Ashwin Ramaswami
 Jeff Ramnani
 Bayard Randel
 Varpu Rantala
diff --git a/Misc/NEWS.d/next/Security/2019-08-27-01-13-05.bpo-37764.qv67PQ.rst b/Misc/NEWS.d/next/Security/2019-08-27-01-13-05.bpo-37764.qv67PQ.rst
new file mode 100644
index 0000000..27fa8e1
--- /dev/null
+++ b/Misc/NEWS.d/next/Security/2019-08-27-01-13-05.bpo-37764.qv67PQ.rst
@@ -0,0 +1 @@
+Fixes email._header_value_parser.get_unstructured going into an infinite loop for a specific case in which the email header does not have trailing whitespace, and the case in which it contains an invalid encoded word. Patch by Ashwin Ramaswami.
\ No newline at end of file