gh-90949: add Expat API to prevent XML deadly allocations

Doc/library/pyexpat.rst

@@ -72,7 +72,14 @@
    *encoding* [1]_ is given it will override the implicit or explicit encoding of the
    document.
 
+   .. _xmlparser-non-root:
+
+   Parsers created through :func:`!ParserCreate` are called "root" parsers,
+   in the sense that they do not have any parent parser attached. Non-root
+   parsers are created by :meth:`parser.ExternalEntityParserCreate
+   <xmlparser.ExternalEntityParserCreate>`.
+
    Expat can optionally do XML namespace processing for you, enabled by providing a
    value for *namespace_separator*.  The value must be a one-character string; a
    :exc:`ValueError` will be raised if the string has an illegal length (``None``
    is considered the same as omission).  When namespace processing is enabled,
@@ -231,7 +238,56 @@
    .. versionadded:: 3.13
 
 
+:class:`!xmlparser` objects have the following methods to mitigate some
+common XML vulnerabilities.
+
+.. method:: xmlparser.SetAllocTrackerActivationThreshold(threshold, /)
+
+   Sets the number of allocated bytes of dynamic memory needed to activate
+   protection against disproportionate use of RAM.
+
+   By default, parser objects have an allocation activation threshold of 64 MiB,
+   or equivalently 67,108,864 bytes.
+
+   An :exc:`ExpatError` is raised if this method is called on a
+   |xml-non-root-parser| parser.
+   The corresponding :attr:`~ExpatError.lineno` and :attr:`~ExpatError.offset`
+   should not be used as they may have no special meaning.
+
+   .. versionadded:: next
+
+.. method:: xmlparser.SetAllocTrackerMaximumAmplification(max_factor, /)
+
+   Sets the maximum amplification factor between direct input and bytes
+   of dynamic memory allocated.
+
+   The amplification factor is calculated as ``allocated / direct``
+   while parsing, where ``direct`` is the number of bytes read from
+   the primary document in parsing and ``allocated`` is the number
+   of bytes of dynamic memory allocated in the parser hierarchy.
+
+   The *max_factor* value must be a non-NaN :class:`float` value greater than
+   or equal to 1.0. Amplification factors greater than 100.0 can be observed
+   near the start of parsing even with benign files in practice. In particular,
+   the activation threshold should be carefully chosen to avoid false positives.
+
+   By default, parser objects have a maximum amplification factor of 100.0.
+
+   An :exc:`ExpatError` is raised if this method is called on a
+   |xml-non-root-parser| parser or if *max_factor* is outside the valid range.
+   The corresponding :attr:`~ExpatError.lineno` and :attr:`~ExpatError.offset`
+   should not be used as they may have no special meaning.
+
+   .. note::
+
+      The maximum amplification factor is only considered if the threshold
+      that can be adjusted :meth:`.SetAllocTrackerActivationThreshold` is
+      exceeded.
+
+   .. versionadded:: next
+
+
 :class:`xmlparser` objects have the following attributes:
 
 
 .. attribute:: xmlparser.buffer_size
@@ -954,3 +1010,4 @@
    not. See https://www.w3.org/TR/2006/REC-xml11-20060816/#NT-EncodingDecl
    and https://www.iana.org/assignments/character-sets/character-sets.xhtml.
 
+.. |xml-non-root-parser| replace:: :ref:`non-root <xmlparser-non-root>`

Doc/whatsnew/3.15.rst

@@ -540,6 +540,16 @@ unittest
   (Contributed by Garry Cairns in :gh:`134567`.)
 
 
+xml.parsers.expat
+-----------------
+
+* Add :func:`~xml.parsers.expat.xmlparser.SetAllocTrackerActivationThreshold`
+  and :func:`~xml.parsers.expat.xmlparser.SetAllocTrackerMaximumAmplification`
+  to :ref:`xmlparser <xmlparser-objects>` objects to prevent use of
+  disproportional amounts of dynamic memory from within an Expat parser.
+  (Contributed by Bénédikt Tran in :gh:`90949`.)
+
+
 zlib
 ----
 

Include/pyexpat.h

@@ -52,6 +52,11 @@ struct PyExpat_CAPI
     int (*SetHashSalt)(XML_Parser parser, unsigned long hash_salt);
     /* might be NULL for expat < 2.6.0 */
     XML_Bool (*SetReparseDeferralEnabled)(XML_Parser parser, XML_Bool enabled);
+    /* might be NULL for expat < 2.7.2 */
+    XML_Bool (*SetAllocTrackerActivationThreshold)(
+        XML_Parser parser, unsigned long long activationThresholdBytes);
+    XML_Bool (*SetAllocTrackerMaximumAmplification)(
+        XML_Parser parser, float maxAmplificationFactor);
     /* always add new stuff to the end! */
 };
 

Lib/test/test_pyexpat.py

@@ -1,14 +1,18 @@
 # XXX TypeErrors on calling handlers, or on bad return values from a
 # handler, are obscure and unhelpful.
 
+import abc
+import functools
 import os
+import re
 import sys
 import sysconfig
+import textwrap
 import unittest
 import traceback
 from io import BytesIO
 from test import support
-from test.support import os_helper
+from test.support import import_helper, os_helper
 from test.support import sortdict
 from unittest import mock
 from xml.parsers import expat
@@ -821,5 +825,199 @@ def start_element(name, _):
         self.assertEqual(started, ['doc'])
 
 
+class AttackProtectionTestBase(abc.ABC):
+    """
+    Base class for testing protections against XML payloads with
+    disproportionate amplification.
+
+    The protections being tested should detect and prevent attacks
+    that leverage disproportionate amplification from small inputs.
+    """
+
+    @staticmethod
+    def exponential_expansion_payload(*, nrows, ncols, text='.'):
+        """Create a billion laughs attack payload.
+
+        Be careful: the number of total items is pow(n, k), thereby
+        requiring at least pow(ncols, nrows) * sizeof(text) memory!
+        """
+        template = textwrap.dedent(f"""\
+            <?xml version="1.0"?>
+            <!DOCTYPE doc [
+                <!ENTITY row0 "{text}">
+                <!ELEMENT doc (#PCDATA)>
+            {{body}}
+            ]>
+            <doc>&row{nrows};</doc>
+        """).rstrip()
+
+        body = '\n'.join(
+            f'<!ENTITY row{i + 1} "{f"&row{i};" * ncols}">'
+            for i in range(nrows)
+        )
+        body = textwrap.indent(body, ' ' * 4)
+        return template.format(body=body)
+
+    def test_payload_generation(self):
+        # self-test for exponential_expansion_payload()
+        payload = self.exponential_expansion_payload(nrows=2, ncols=3)
+        self.assertEqual(payload, textwrap.dedent("""\
+            <?xml version="1.0"?>
+            <!DOCTYPE doc [
+                <!ENTITY row0 ".">
+                <!ELEMENT doc (#PCDATA)>
+                <!ENTITY row1 "&row0;&row0;&row0;">
+                <!ENTITY row2 "&row1;&row1;&row1;">
+            ]>
+            <doc>&row2;</doc>
+        """).rstrip())
+
+    def assert_root_parser_failure(self, func, /, *args, **kwargs):
+        """Check that func(*args, **kwargs) is invalid for a sub-parser."""
+        msg = "parser must be a root parser"
+        self.assertRaisesRegex(expat.ExpatError, msg, func, *args, **kwargs)
+
+    @abc.abstractmethod
+    def assert_rejected(self, func, /, *args, **kwargs):
+        """Assert that func(*args, **kwargs) triggers the attack protection.
+
+        Note: this method must ensure that the attack protection being tested
+        is the one that is actually triggered at runtime, e.g., by matching
+        the exact error message.
+        """
+
+    @abc.abstractmethod
+    def set_activation_threshold(self, parser, threshold):
+        """Set the activation threshold for the tested protection."""
+
+    @abc.abstractmethod
+    def set_maximum_amplification(self, parser, max_factor):
+        """Set the maximum amplification factor for the tested protection."""
+
+    @abc.abstractmethod
+    def test_set_activation_threshold__threshold_reached(self):
+        """Test when the activation threshold is exceeded."""
+
+    @abc.abstractmethod
+    def test_set_activation_threshold__threshold_not_reached(self):
+        """Test when the activation threshold is not exceeded."""
+
+    def test_set_activation_threshold__invalid_threshold_type(self):
+        parser = expat.ParserCreate()
+        setter = functools.partial(self.set_activation_threshold, parser)
+
+        self.assertRaises(TypeError, setter, 1.0)
+        self.assertRaises(TypeError, setter, -1.5)
+        self.assertRaises(ValueError, setter, -5)
+
+    def test_set_activation_threshold__invalid_threshold_range(self):
+        _testcapi = import_helper.import_module("_testcapi")
+        parser = expat.ParserCreate()
+        setter = functools.partial(self.set_activation_threshold, parser)
+
+        self.assertRaises(OverflowError, setter, _testcapi.ULLONG_MAX + 1)
+
+    def test_set_activation_threshold__fail_for_subparser(self):
+        parser = expat.ParserCreate()
+        subparser = parser.ExternalEntityParserCreate(None)
+        setter = functools.partial(self.set_activation_threshold, subparser)
+        self.assert_root_parser_failure(setter, 12345)
+
+    @abc.abstractmethod
+    def test_set_maximum_amplification__amplification_exceeded(self):
+        """Test when the amplification factor is exceeded."""
+
+    @abc.abstractmethod
+    def test_set_maximum_amplification__amplification_not_exceeded(self):
+        """Test when the amplification factor is not exceeded."""
+
+    def test_set_maximum_amplification__infinity(self):
+        inf = float('inf')  # an 'inf' threshold is allowed by Expat
+        parser = expat.ParserCreate()
+        self.assertIsNone(self.set_maximum_amplification(parser, inf))
+
+    def test_set_maximum_amplification__invalid_max_factor_type(self):
+        parser = expat.ParserCreate()
+        setter = functools.partial(self.set_maximum_amplification, parser)
+
+        self.assertRaises(TypeError, setter, None)
+        self.assertRaises(TypeError, setter, 'abc')
+
+    def test_set_maximum_amplification__invalid_max_factor_range(self):
+        parser = expat.ParserCreate()
+        setter = functools.partial(self.set_maximum_amplification, parser)
+
+        msg = re.escape("'max_factor' must be at least 1.0")
+        self.assertRaisesRegex(expat.ExpatError, msg, setter, float('nan'))
+        self.assertRaisesRegex(expat.ExpatError, msg, setter, 0.99)
+
+    def test_set_maximum_amplification__fail_for_subparser(self):
+        parser = expat.ParserCreate()
+        subparser = parser.ExternalEntityParserCreate(None)
+        setter = functools.partial(self.set_maximum_amplification, subparser)
+        self.assert_root_parser_failure(setter, 123.45)
+
+
+@unittest.skipIf(expat.version_info < (2, 7, 2), "requires Expat >= 2.7.2")
+class MemoryProtectionTest(AttackProtectionTestBase, unittest.TestCase):
+
+    # NOTE: with the default Expat configuration, the billion laughs protection
+    # may hit before the allocation limiter if exponential_expansion_payload()
+    # is not carefully parametrized. As such, the payloads should be chosen so
+    # that either the allocation limiter is hit before other protections are
+    # triggered or no protection at all is triggered.
+
+    def assert_rejected(self, func, /, *args, **kwargs):
+        """Check that func(*args, **kwargs) hits the allocation limit."""
+        msg = r"out of memory: line \d+, column \d+"
+        self.assertRaisesRegex(expat.ExpatError, msg, func, *args, **kwargs)
+
+    def set_activation_threshold(self, parser, threshold):
+        return parser.SetAllocTrackerActivationThreshold(threshold)
+
+    def set_maximum_amplification(self, parser, max_factor):
+        return parser.SetAllocTrackerMaximumAmplification(max_factor)
+
+    def test_set_activation_threshold__threshold_reached(self):
+        parser = expat.ParserCreate()
+        # Choose a threshold expected to be always reached.
+        self.set_activation_threshold(parser, 3)
+        # Check that the threshold is reached by choosing a small factor
+        # and a payload whose peak amplification factor exceeds it.
+        self.assertIsNone(self.set_maximum_amplification(parser, 1.0))
+        payload = self.exponential_expansion_payload(ncols=10, nrows=4)
+        self.assert_rejected(parser.Parse, payload, True)
+
+    def test_set_activation_threshold__threshold_not_reached(self):
+        parser = expat.ParserCreate()
+        # Choose a threshold expected to be never reached.
+        self.set_activation_threshold(parser, pow(10, 5))
+        # Check that the threshold is reached by choosing a small factor
+        # and a payload whose peak amplification factor exceeds it.
+        self.assertIsNone(self.set_maximum_amplification(parser, 1.0))
+        payload = self.exponential_expansion_payload(ncols=10, nrows=4)
+        self.assertIsNotNone(parser.Parse(payload, True))
+
+    def test_set_maximum_amplification__amplification_exceeded(self):
+        parser = expat.ParserCreate()
+        # Unconditionally enable maximum activation factor.
+        self.set_activation_threshold(parser, 0)
+        # Choose a max amplification factor expected to always be exceeded.
+        self.assertIsNone(self.set_maximum_amplification(parser, 1.0))
+        # Craft a payload for which the peak amplification factor is > 1.0.
+        payload = self.exponential_expansion_payload(ncols=1, nrows=2)
+        self.assert_rejected(parser.Parse, payload, True)
+
+    def test_set_maximum_amplification__amplification_not_exceeded(self):
+        parser = expat.ParserCreate()
+        # Unconditionally enable maximum activation factor.
+        self.set_activation_threshold(parser, 0)
+        # Choose a max amplification factor expected to never be exceeded.
+        self.assertIsNone(self.set_maximum_amplification(parser, 1e4))
+        # Craft a payload for which the peak amplification factor is < 1e4.
+        payload = self.exponential_expansion_payload(ncols=1, nrows=2)
+        self.assertIsNotNone(parser.Parse(payload, True))
+
+
 if __name__ == "__main__":
     unittest.main()

Misc/NEWS.d/next/Library/2025-09-22-14-40-11.gh-issue-90949.UM35nb.rst

@@ -0,0 +1,5 @@
+Add :func:`~xml.parsers.expat.xmlparser.SetAllocTrackerActivationThreshold`
+and :func:`~xml.parsers.expat.xmlparser.SetAllocTrackerMaximumAmplification`
+to :ref:`xmlparser <xmlparser-objects>` objects to prevent use of
+disproportional amounts of dynamic memory from within an Expat parser.
+Patch by Bénédikt Tran.