Merge pull request #5811 from uktrade/fix-ingestion-great-company-reg…

…istration-number Fix ingestion of great-company-registration-number
uktrade · Nov 21, 2024 · 29c4ceb · 29c4ceb
2 parents 12ff05e + 2bb6899
commit 29c4ceb
Show file tree

Hide file tree

Showing 2 changed files with 28 additions and 7 deletions.
diff --git a/datahub/company_activity/tasks/ingest_great_data.py b/datahub/company_activity/tasks/ingest_great_data.py
@@ -19,6 +19,13 @@ def ingest_great_data(bucket, file):
     logger.info(f'Ingesting file: {file} finished')
 
 
+def validate_company_registration_number(company_registration_number):
+    company_registration_number_str = str(company_registration_number)
+    if len(company_registration_number_str) > 10:
+        return None
+    return company_registration_number_str
+
+
 class GreatIngestionTask:
     def __init__(self):
         self._countries = None
@@ -45,7 +52,9 @@ def _already_ingested(self, id):
     def _create_company(self, data, form_id):
         company = Company.objects.create(
             name=data.get('business_name', ''),
-            company_number=data.get('company_registration_number', ''),
+            company_number=validate_company_registration_number(
+                data.get('company_registration_number', ''),
+            ),
             turnover_range=self._get_turnover_range(data.get('annual_turnover')),
             business_type=self._get_business_type(data.get('type')),
             employee_range=self._get_business_size(data.get('number_of_employees')),
@@ -72,7 +81,7 @@ def _create_contact(self, data, company, form_id):
 
     def _get_company(self, data, form_id):
         company = self._get_company_by_companies_house_num(
-            data.get('company_registration_number'),
+            validate_company_registration_number(data.get('company_registration_number')),
         )
         if company:
             return company

diff --git a/datahub/company_activity/tests/test_tasks/test_great_ingestion_task.py b/datahub/company_activity/tests/test_tasks/test_great_ingestion_task.py
@@ -15,7 +15,8 @@
 from datahub.company_activity.models import GreatExportEnquiry, IngestedFile
 from datahub.company_activity.tasks.constants import BUCKET, GREAT_PREFIX, REGION
 from datahub.company_activity.tasks.ingest_great_data import (
-    GreatIngestionTask, ingest_great_data,
+    GreatIngestionTask,
+    ingest_great_data,
 )
 from datahub.company_activity.tests.factories import (
     GreatExportEnquiryFactory,
@@ -25,7 +26,9 @@
 
 @pytest.fixture
 def test_file():
-    filepath = 'datahub/company_activity/tests/test_tasks/fixtures/great/20241023T000346.jsonl.gz'
+    filepath = (
+        'datahub/company_activity/tests/test_tasks/fixtures/great/20241023T000346.jsonl.gz'
+    )
     return open(filepath, 'rb')
 
 
@@ -164,7 +167,7 @@ def test_company_name_mapping(self):
                 "id": "5250",
                 "created_at": "2024-09-19T14:00:34.069",
                 "data": {{
-                    "company_registration_number": 994349,
+                    "company_registration_number": "994349",
                     "business_name": "{company.name}"
                 }}
             }}
@@ -563,7 +566,7 @@ def test_boolean_field_mapping(self):
 
     @pytest.mark.django_db
     @mock_aws
-    def test_long_field_values(self, test_file_path):
+    def test_long_field_values(self):
         """
         Test that we can ingest records with long field values
         """
@@ -575,6 +578,7 @@ def test_long_field_values(self, test_file_path):
             'that either need to be stored as TextFields if we need'
             'the full value or truncated if we do not. Long long long.'
         )
+
         data = f"""
             {{
                 "id": "5249",
@@ -589,10 +593,18 @@ def test_long_field_values(self, test_file_path):
                     "product_or_service_2": "{long_text}",
                     "product_or_service_3": "{long_text}",
                     "product_or_service_4": "{long_text}",
-                    "product_or_service_5": "{long_text}"
+                    "product_or_service_5": "{long_text}",
+                    "company_registration_number": "{long_text}"
+
                 }}
             }}
         """
         task = GreatIngestionTask()
         task.json_to_model(json.loads(data))
         assert GreatExportEnquiry.objects.count() == initial_count + 1
+
+        result = GreatExportEnquiry.objects.get(form_id='5249').company_id
+
+        company_result = Company.objects.get(id=result)
+
+        assert company_result.company_number is None