Merge pull request #25 from freelawproject/19-add-more-courts-to-mapping-tweak-log

19 Ignore invalid subdomains that are not courts

Author: albertisfu

cl/recap_email/app/app.py

@@ -10,7 +10,7 @@
 from requests.exceptions import HTTPError, Timeout
 from sentry_sdk.integrations.aws_lambda import AwsLambdaIntegration
 
-from .pacer import map_pacer_to_cl_id
+from .pacer import map_pacer_to_cl_id, sub_domains_to_ignore
 from .utils import retry  # pylint: disable=import-error
 
 # NOTE - This is necessary for the relative imports.
@@ -142,15 +142,19 @@ def log_invalid_court_error(response, message_id):
         return
 
     for msg in response.json().get("court", []):
-        if "Invalid pk" in msg:
-            match = re.search(r'Invalid pk "([^"]+)"', msg)
-            if match:
-                error_message = (
-                    f"Invalid court pk: {match.group(1)} - "
-                    f"message_id: {message_id}"
-                )
-                sentry_sdk.capture_message(error_message, level="error")
-                break
+        if "Invalid pk" not in msg:
+            continue
+        match = re.search(r'Invalid pk "([^"]+)"', msg)
+        if not match:
+            continue
+        court_id = match.group(1)
+        error_message = (
+            f"Invalid court pk: {court_id} - message_id: {message_id}"
+        )
+        sentry_sdk.capture_message(
+            error_message, level="error", fingerprint=["invalid-court-pk"]
+        )
+        break
 
 
 @retry(
@@ -230,8 +234,10 @@ def handler(event, context):  # pylint: disable=unused-argument
         if verdict not in {"PASS", "GRAY"}:
             return validation_failure(email, receipt, verdict)
 
-    # Check domain is valid (comes from uscourts.gov)
     domain_verdict = get_valid_domain_verdict(email)
-    if domain_verdict != "PASS":
+    court_id = get_cl_court_id(email)
+    # Check domain is valid (comes from uscourts.gov)
+    # Ignore messages that are not from courts, such as updates.uscourts.gov
+    if domain_verdict != "PASS" or court_id in sub_domains_to_ignore:
         return validation_domain_failure(email, receipt, domain_verdict)
     return send_to_court_listener(email, receipt)

cl/recap_email/app/pacer.py

@@ -10,6 +10,8 @@
     "cfc": "uscfc",  # Court of Federal Claims
 }
 
+sub_domains_to_ignore = ["usdoj", "law", "psc", "updates", "MIWD"]
+
 # Reverse dict of pacer_to_cl_ids
 cl_to_pacer_ids = {v: k for k, v in pacer_to_cl_ids.items()}
 

cl/tests/unit/test_recap_email_handler.py

@@ -295,4 +295,32 @@ def test_report_request_for_invalid_court(
         "Invalid court pk: whla - "
         "message_id: 171jjm4scn8vgcn5vrcv4su427obcred7bekus81"
     )
-    mock_sentry_capture.assert_called_with(expected_error, level="error")
+    mock_sentry_capture.assert_called_with(
+        expected_error, level="error", fingerprint=["invalid-court-pk"]
+    )
+
+
+@mock.patch.dict(
+    os.environ,
+    {
+        "RECAP_EMAIL_ENDPOINT": "http://host.docker.internal:8000/api/rest/v3/recap-email/",  # noqa: E501 pylint: disable=line-too-long
+        "AUTH_TOKEN": "************************",
+    },
+)
+def test_ignore_messages_from_invalid_court_ids(
+    pacer_event_one,
+    requests_mock,  # noqa: F811
+):
+    """Confirm that if an invalid court_id in sub_domains_to_ignore is sent, a
+    validation_domain_failure is sent."""
+
+    with mock.patch(
+        "recap_email.app.app.get_cl_court_id", return_value="updates"
+    ):
+        requests_mock.post(
+            "http://host.docker.internal:8000/api/rest/v3/recap-email/",
+            json={"mail": {}, "receipt": {}},
+        )
+        response = app.handler(pacer_event_one, "")
+        assert response["statusCode"] == 424
+        assert response["valid_domain"]["status"] == "FAILED"