YAUF (#2446)

* YAUF.

* Add test for oddly-shaped results.

backend/api/serializers.py

@@ -72,40 +72,92 @@ def validate_user_provided_organization_type(self, value):
 
 
 class UEISerializer(serializers.Serializer):
+    """
+    Does a UEI request against the SAM.gov API and returns a flattened shape
+    containing only the fields we're interested in.
+
+    The below operations are nested and mixed among functions, rather than done
+    serially, but the approximate order of operations is:
+
+        +   Assemble the parameters to pass to the API.
+            Mostly in api.uei.get_uei_info_from_sam_gov.
+        +   Make the API request.
+            api.uei.call_sam_api
+        +   Check for high-level errors.
+            api.uei.get_uei_info_from_sam_gov
+        +   Extract the JSON for the individual record out of the response and check
+        for some other errors.
+            api.uei.parse_sam_uei_json
+        +   For a specific class of error, retry the API call with different
+        parameters.
+            api.uei.get_uei_info_from_sam_gov
+            api.uei.call_sam_api
+            api.uei.parse_sam_uei_json
+        +   If we don't have errors by that point, flatten the data.
+            api.serializsers.UEISerializer.validate_auditee_uei
+        +   If we don't encounter errors at that point, return the flattened data.
+            api.serializsers.UEISerializer.validate_auditee_uei
+
+    """
+
     auditee_uei = serializers.CharField()
 
     def validate_auditee_uei(self, value):
+        """
+        Flattens the UEI response info and returns this shape:
+
+            {
+                "auditee_uei": …,
+                "auditee_name": …,
+                "auditee_fiscal_year_end_date": …,
+                "auditee_address_line_1": …,
+                "auditee_city": …,
+                "auditee_state": …,
+                "auditee_zip": …,
+            }
+
+        Will provide default error-message-like values (such as “No address in SAM.gov)
+        if the keys are missing, but if the SAM.gov fields are present but empty, we
+        return the empty strings.
+
+        """
         sam_response = get_uei_info_from_sam_gov(value)
         if sam_response.get("errors"):
             raise serializers.ValidationError(sam_response.get("errors"))
-        return json.dumps(
-            {
-                "auditee_uei": value,
-                "auditee_name": sam_response.get("response")
-                .get("entityRegistration")
-                .get("legalBusinessName"),
-                "auditee_fiscal_year_end_date": sam_response.get("response")
-                .get("coreData")
-                .get("entityInformation")
-                .get("fiscalYearEndCloseDate"),
-                "auditee_address_line_1": sam_response.get("response")
-                .get("coreData")
-                .get("mailingAddress")
-                .get("addressLine1"),
-                "auditee_city": sam_response.get("response")
-                .get("coreData")
-                .get("mailingAddress")
-                .get("city"),
-                "auditee_state": sam_response.get("response")
-                .get("coreData")
-                .get("mailingAddress")
-                .get("stateOrProvinceCode"),
-                "auditee_zip": sam_response.get("response")
-                .get("coreData")
-                .get("mailingAddress")
-                .get("zipCode"),
+
+        entity_registration = sam_response.get("response")["entityRegistration"]
+        core = sam_response.get("response")["coreData"]
+
+        basic_data = {
+            "auditee_uei": value,
+            "auditee_name": entity_registration.get("legalBusinessName"),
+        }
+        addr_key = "mailingAddress" if "mailingAddress" in core else "physicalAddress"
+
+        mailing_data = {
+            "auditee_address_line_1": "No address in SAM.gov.",
+            "auditee_city": "No address in SAM.gov.",
+            "auditee_state": "No address in SAM.gov.",
+            "auditee_zip": "No address in SAM.gov.",
+        }
+
+        if addr_key in core:
+            mailing_data = {
+                "auditee_address_line_1": core.get(addr_key).get("addressLine1"),
+                "auditee_city": core.get(addr_key).get("city"),
+                "auditee_state": core.get(addr_key).get("stateOrProvinceCode"),
+                "auditee_zip": core.get(addr_key).get("zipCode"),
             }
-        )
+
+        # 2023-10-10: Entities with a samRegistered value of No may be missing
+        # some fields from coreData entirely.
+        entity_information = core.get("entityInformation", {})
+        extra_data = {
+            "auditee_fiscal_year_end_date": entity_information.get(
+                "fiscalYearEndCloseDate", "No fiscal year end date in SAM.gov."
+            ),
+        }
+        return json.dumps(basic_data | mailing_data | extra_data)
 
 
 class AuditeeInfoSerializer(serializers.Serializer):

backend/api/test_serializers.py

@@ -95,6 +95,60 @@ def test_invalid_uei_payload(self):
         # Invalid
         self.assertFalse(UEISerializer(data=invalid).is_valid())
 
+    def test_quirky_uei_payload(self):
+        """
+        It turns out that some entries can be missing fields that we thought would
+        always be present.
+
+        """
+        quirky = {
+            "totalRecords": 1,
+            "entityData": [
+                {
+                    "entityRegistration": {
+                        "samRegistered": "No",
+                        "ueiSAM": "ZQGGHJH74DW7",
+                        "cageCode": None,
+                        "legalBusinessName": "Some organization",
+                        "registrationStatus": "ID Assigned",
+                        "evsSource": "X&Y",
+                        "ueiStatus": "Active",
+                        "ueiExpirationDate": None,
+                        "ueiCreationDate": "2023-04-01",
+                        "publicDisplayFlag": "Y",
+                        "dnbOpenData": None,
+                    },
+                    "coreData": {
+                        "physicalAddress": {
+                            "addressLine1": "SOME RD",
+                            "addressLine2": None,
+                            "city": "SOME CITY",
+                            "stateOrProvinceCode": "AL",
+                            "zipCode": "36659",
+                            "zipCodePlus4": "3903",
+                            "countryCode": "USA",
+                        }
+                    },
+                }
+            ],
+            "links": {
+                "selfLink": "https://api.sam.gov/entity-information/v3/entities?api_key=REPLACE_WITH_API_KEY&ueiSAM=ZQGGHJH74DW7&samRegistered=No&page=0&size=10"
+            },
+        }
+        empty = {"totalRecords": 0, "entityData": []}
+        data = {"auditee_uei": "ZQGGHJH74DW7"}
+        with patch("api.uei.SESSION.get") as mock_get:
+            mock_get.return_value.status_code = 200
+            # First time, it should return zero results and retry:
+            mock_get.return_value.json.return_value = empty
+            self.assertFalse(UEISerializer(data=data).is_valid())
+            self.assertEqual(mock_get.call_count, 2)
+            # Second time, it should return the samRegistered No result and only
+            # call the get method once:
+            mock_get.return_value.json.return_value = quirky
+            self.assertTrue(UEISerializer(data=data).is_valid())
+            self.assertEqual(mock_get.call_count, 3)
+
 
 class AuditeeInfoStepTests(SimpleTestCase):
     def test_valid_auditee_info_with_uei(self):