python3-django: Fix CVE-2024-39614

An issue was discovered in Django 5.0 before 5.0.7 and 4.2 before 4.2.14.
get_supported_language_variant() was subject to a potential denial-of-service
attack when used with very long strings containing specific characters.

Reference:
https://nvd.nist.gov/vuln/detail/CVE-2024-39614

Upstream-patch:
https://github.com/django/django/commit/17358fb35fb7217423d4c4877ccb6d1a3a40b1c3

Signed-off-by: Soumya Sambu <soumya.sambu@windriver.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>

1 files changed, 138 insertions, 0 deletions

diff --git a/meta-python/recipes-devtools/python/python3-django/CVE-2024-39614.patch b/meta-python/recipes-devtools/python/python3-django/CVE-2024-39614.patch
new file mode 100644
index 0000000000..340cfceac8
--- /dev/null
+++ b/meta-python/recipes-devtools/python/python3-django/CVE-2024-39614.patch
@@ -0,0 +1,138 @@
+From 17358fb35fb7217423d4c4877ccb6d1a3a40b1c3 Mon Sep 17 00:00:00 2001
+From: Sarah Boyce <42296566+sarahboyce@users.noreply.github.com>
+Date: Wed, 26 Jun 2024 12:11:54 +0200
+Subject: [PATCH] [4.2.x] Fixed CVE-2024-39614 -- Mitigated potential DoS in
+ get_supported_language_variant().
+
+Language codes are now parsed with a maximum length limit of 500 chars.
+
+Thanks to MProgrammer for the report.
+
+CVE: CVE-2024-39614
+
+Upstream-Status: Backport [https://github.com/django/django/commit/17358fb35fb7217423d4c4877ccb6d1a3a40b1c3]
+
+Signed-off-by: Soumya Sambu <soumya.sambu@windriver.com>
+---
+ django/utils/translation/trans_real.py | 24 ++++++++++++++++++++----
+ docs/ref/utils.txt                     | 10 ++++++++++
+ tests/i18n/tests.py                    | 11 +++++++++++
+ 3 files changed, 41 insertions(+), 4 deletions(-)
+
+diff --git a/django/utils/translation/trans_real.py b/django/utils/translation/trans_real.py
+index 7f658cf..56b4ef1 100644
+--- a/django/utils/translation/trans_real.py
++++ b/django/utils/translation/trans_real.py
+@@ -30,8 +30,10 @@ _default = None
+ CONTEXT_SEPARATOR = "\x04"
+
+ # Maximum number of characters that will be parsed from the Accept-Language
+-# header to prevent possible denial of service or memory exhaustion attacks.
+-ACCEPT_LANGUAGE_HEADER_MAX_LENGTH = 500
++# header or cookie to prevent possible denial of service or memory exhaustion
++# attacks. About 10x longer than the longest value shown on MDN’s
++# Accept-Language page.
++LANGUAGE_CODE_MAX_LENGTH = 500
+
+ # Format of Accept-Language header values. From RFC 2616, section 14.4 and 3.9
+ # and RFC 3066, section 2.1
+@@ -472,11 +474,25 @@ def get_supported_language_variant(lang_code, strict=False):
+     If `strict` is False (the default), look for a country-specific variant
+     when neither the language code nor its generic variant is found.
+
++    The language code is truncated to a maximum length to avoid potential
++    denial of service attacks.
++
+     lru_cache should have a maxsize to prevent from memory exhaustion attacks,
+     as the provided language codes are taken from the HTTP request. See also
+     <https://www.djangoproject.com/weblog/2007/oct/26/security-fix/>.
+     """
+     if lang_code:
++        # Truncate the language code to a maximum length to avoid potential
++        # denial of service attacks.
++        if len(lang_code) > LANGUAGE_CODE_MAX_LENGTH:
++            if (
++                not strict
++                and (index := lang_code.rfind("-", 0, LANGUAGE_CODE_MAX_LENGTH)) > 0
++            ):
++                # There is a generic variant under the maximum length accepted length.
++                lang_code = lang_code[:index]
++            else:
++                raise ValueError("'lang_code' exceeds the maximum accepted length")
+         # If 'fr-ca' is not supported, try special fallback or language-only 'fr'.
+         possible_lang_codes = [lang_code]
+         try:
+@@ -598,12 +614,12 @@ def parse_accept_lang_header(lang_string):
+     `functools.lru_cache()` to avoid repetitive parsing of common header values.
+     """
+     # If the header value doesn't exceed the maximum allowed length, parse it.
+-    if len(lang_string) <= ACCEPT_LANGUAGE_HEADER_MAX_LENGTH:
++    if len(lang_string) <= LANGUAGE_CODE_MAX_LENGTH:
+         return _parse_accept_lang_header(lang_string)
+
+     # If there is at least one comma in the value, parse up to the last comma,
+     # skipping any truncated parts at the end of the header value.
+-    index = lang_string.rfind(",", 0, ACCEPT_LANGUAGE_HEADER_MAX_LENGTH)
++    index = lang_string.rfind(",", 0, LANGUAGE_CODE_MAX_LENGTH)
+     if index > 0:
+         return _parse_accept_lang_header(lang_string[:index])
+
+diff --git a/docs/ref/utils.txt b/docs/ref/utils.txt
+index 390f167..63a56e5 100644
+--- a/docs/ref/utils.txt
++++ b/docs/ref/utils.txt
+@@ -1142,6 +1142,11 @@ functions without the ``u``.
+     ``lang_code`` is ``'es-ar'`` and ``'es'`` is in :setting:`LANGUAGES` but
+     ``'es-ar'`` isn't.
+
++    ``lang_code`` has a maximum accepted length of 500 characters. A
++    :exc:`ValueError` is raised if ``lang_code`` exceeds this limit and
++    ``strict`` is ``True``, or if there is no generic variant and ``strict``
++    is ``False``.
++
+     If ``strict`` is ``False`` (the default), a country-specific variant may
+     be returned when neither the language code nor its generic variant is found.
+     For example, if only ``'es-co'`` is in :setting:`LANGUAGES`, that's
+@@ -1150,6 +1155,11 @@ functions without the ``u``.
+
+     Raises :exc:`LookupError` if nothing is found.
+
++    .. versionchanged:: 4.2.14
++
++        In older versions, ``lang_code`` values over 500 characters were
++        processed without raising a :exc:`ValueError`.
++
+ .. function:: to_locale(language)
+
+     Turns a language name (en-us) into a locale name (en_US).
+diff --git a/tests/i18n/tests.py b/tests/i18n/tests.py
+index 6efc3a5..0e93395 100644
+--- a/tests/i18n/tests.py
++++ b/tests/i18n/tests.py
+@@ -39,6 +39,7 @@ from django.utils.translation import (
+ from django.utils.translation.reloader import (
+     translation_file_changed, watch_for_translation_changes,
+ )
++from django.utils.translation.trans_real import LANGUAGE_CODE_MAX_LENGTH
+
+ from .forms import CompanyForm, I18nForm, SelectDateForm
+ from .models import Company, TestModel
+@@ -1462,6 +1463,16 @@ class MiscTests(SimpleTestCase):
+             g('xyz')
+         with self.assertRaises(LookupError):
+             g('xy-zz')
++        msg = "'lang_code' exceeds the maximum accepted length"
++        with self.assertRaises(LookupError):
++            g("x" * LANGUAGE_CODE_MAX_LENGTH)
++        with self.assertRaisesMessage(ValueError, msg):
++            g("x" * (LANGUAGE_CODE_MAX_LENGTH + 1))
++        # 167 * 3 = 501 which is LANGUAGE_CODE_MAX_LENGTH + 1.
++        self.assertEqual(g("en-" * 167), "en")
++        with self.assertRaisesMessage(ValueError, msg):
++            g("en-" * 167, strict=True)
++        self.assertEqual(g("en-" * 30000), "en")  # catastrophic test
+
+     def test_get_supported_language_variant_null(self):
+         g = trans_null.get_supported_language_variant
+--
+2.40.0