Block external network access in Python tests

kolibri/conftest.py

@@ -1,5 +1,7 @@
+import ipaddress
 import os
 import shutil
+import socket
 
 import pytest
 
@@ -8,6 +10,72 @@
 # referenced in pytest.ini
 TEMP_KOLIBRI_HOME = "./.pytest_kolibri_home"
 
+# Network access guard
+#
+# Blocks all network access to non-loopback addresses for the entire test
+# run, so that no test can ever depend on (or be broken by) an external
+# service. Any attempt to resolve or connect to a non-local host raises
+# BlockedNetworkAccessError pointing at the offending call.
+#
+# Tests that exercise code making outbound HTTP requests should mock at the
+# requests session level, e.g. by patching ``requests.Session.request`` -
+# see kolibri/core/discovery/test/helpers.py for reusable helpers.
+
+LOOPBACK_HOSTNAMES = {"localhost", "localhost.localdomain", "ip6-localhost", ""}
+
+_real_getaddrinfo = socket.getaddrinfo
+_real_socket_connect = socket.socket.connect
+
+
+class BlockedNetworkAccessError(RuntimeError):
+    pass
+
+
+def _is_local_host(host):
+    if host is None:
+        return True
+    if isinstance(host, bytes):
+        host = host.decode("utf-8", "replace")
+    host = host.lower().rstrip(".")
+    if host in LOOPBACK_HOSTNAMES:
+        return True
+    try:
+        ip = ipaddress.ip_address(host)
+    except ValueError:
+        # A non-local hostname that would require DNS resolution
+        return False
+    return ip.is_loopback or ip.is_unspecified
+
+
+def _block(api, host, port):
+    raise BlockedNetworkAccessError(
+        "A test attempted real network access via {api} to {host}:{port}. "
+        "Tests must not make requests to external services - mock them "
+        "instead, e.g. by patching requests.Session.request "
+        "(see kolibri/core/discovery/test/helpers.py).".format(
+            api=api, host=host, port=port
+        )
+    )
+
+
+def _guarded_getaddrinfo(host, port, *args, **kwargs):
+    if not _is_local_host(host):
+        _block("socket.getaddrinfo", host, port)
+    return _real_getaddrinfo(host, port, *args, **kwargs)
+
+
+def _guarded_socket_connect(sock, address):
+    if sock.family in (socket.AF_INET, socket.AF_INET6) and isinstance(address, tuple):
+        host = address[0]
+        if not _is_local_host(host):
+            _block("socket.connect", host, address[1] if len(address) > 1 else "")
+    return _real_socket_connect(sock, address)
+
+
+def pytest_configure(config):
+    socket.getaddrinfo = _guarded_getaddrinfo
+    socket.socket.connect = _guarded_socket_connect
+
 
 @pytest.fixture(autouse=True)
 def clear_process_cache():

kolibri/core/test/test_key_urls.py

@@ -1,3 +1,4 @@
+import requests
 from django.conf import settings
 from django.urls import reverse
 from django.urls.exceptions import NoReverseMatch
@@ -12,6 +13,20 @@
 from kolibri.core.auth.test.test_api import FacilityFactory
 from kolibri.core.auth.test.test_api import FacilityUserFactory
 from kolibri.core.device.translation import get_settings_language
+from kolibri.core.discovery.test.helpers import mock_response
+
+
+def mock_external_request(session, method, url, *args, **kwargs):
+    """
+    Give any outbound HTTP request a successful empty response, so that views
+    proxying to external services (the Kolibri Data Portal token validation,
+    the Studio remote channel lookup) can be smoke tested without depending
+    on those services.
+    """
+    response = mock_response(200)
+    response.url = url
+    response.json.return_value = []
+    return response
 
 
 class BeforeDeviceProvisionTests(APITestCase):
@@ -215,7 +230,9 @@ def check_urls(urlpatterns, prefix=""):  # noqa: C901
 
         with patch(
             "kolibri.core.webpack.hooks.WebpackBundleHook.bundle", return_value=[]
-        ), patch("kolibri.core.webpack.hooks.WebpackBundleHook.get_by_unique_id"):
+        ), patch(
+            "kolibri.core.webpack.hooks.WebpackBundleHook.get_by_unique_id"
+        ), patch.object(requests.Session, "request", mock_external_request):
             from kolibri.deployment.default.urls import urlpatterns
 
             check_urls(urlpatterns)