additional_data: sources: add licence info

#213

Author: BibianaC

datastore/additional_data/sources/codelist_code.py

@@ -23,6 +23,9 @@ class CodeListSource(object):
     responsible for field: codeListLookup
     """
 
+    ADDITIONAL_DATA_KEY = "codeListLookup"
+    LICENCE = "https://creativecommons.org/licenses/by/4.0/"
+
     def import_codelists(self):
         with transaction.atomic():
             CodelistCode.objects.all().delete()
@@ -136,3 +139,5 @@ def update_additional_data(self, grant, source_file, additional_data):
             "recipientOrg_location_countryCode": recipientOrg_location_countryCode,
             "fundingOrg_location_countryCode": fundingOrg_location_countryCode,
         }
+
+        additional_data[f"{self.ADDITIONAL_DATA_KEY}_LICENCE"] = self.LICENCE

datastore/additional_data/sources/find_that_charity.py

@@ -74,6 +74,9 @@ class FindThatCharitySource(object):
     FindThatCharity (FTC) organisation info data sources"""
 
     ADDITIONAL_DATA_KEY = "recipientOrgInfos"
+    LICENCE = (
+        "https://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/"
+    )
 
     def __init__(self, *args, **kwargs):
         # A basic internal memory cache to avoid hitting the db on duplicate
@@ -83,6 +86,8 @@ def __init__(self, *args, **kwargs):
         self._cache = {}
 
     def update_additional_data(self, grant, source_file, additional_data):
+        additional_data[f"{self.ADDITIONAL_DATA_KEY}_LICENCE"] = self.LICENCE
+
         # We can't do anything if this grant doesn't have a recipientOrganization
         if not grant.get("recipientOrganization"):
             return

datastore/additional_data/sources/geo_lookup.py

@@ -13,6 +13,11 @@ class GeoLookupSource(object):
     """Imports geographical lookups from https://github.com/drkane/geo-lookups/
     These allow for looking up from lower-level geography like Ward to a standard local authority, region, etc"""
 
+    ADDITIONAL_DATA_KEY = "locationLookup"
+    LICENCE = (
+        "https://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/"
+    )
+
     # Source URLS
     SOURCE_URLS = {
         "lsoa": {
@@ -143,3 +148,5 @@ def update_additional_data(self, grant, source_file, additional_data):
                 area["source"] = "recipientOrganizationPostcode"
                 area["sourceCode"] = lsoa
                 additional_data["locationLookup"].append(area)
+
+        additional_data[f"{self.ADDITIONAL_DATA_KEY}_LICENCE"] = self.LICENCE

datastore/additional_data/sources/nspl.py

@@ -24,6 +24,10 @@ class NSPLSource(object):
     #  Search for the most recent "National Statistics Postcode Lookup 2021 Cencus"
     #  https://geoportal.statistics.gov.uk/search?collection=Dataset&q=National%20Statistics%20Postcode%20Lookup%20-%202021%20Census&sort=-modified&source=office%20for%20national%20statistics&tags=national%20statistics%20postcode%20lookup%2C2021_cencus&type=csv%20collection
     NSPL_URL = "https://www.arcgis.com/sharing/rest/content/items/204e40244d4d4903ba1861d492f47d29/data"
+    ADDITIONAL_DATA_KEY = "recipientOrganizationLocation"
+    LICENCE = (
+        "https://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/"
+    )
 
     def __init__(self):
         self._nspl_cache = {}
@@ -204,3 +208,5 @@ def update_additional_data(self, grant, source_file, additional_data):
                     self.update_location_data_code_names(location_data)
                     additional_data["recipientOrganizationLocation"] = location_data
                     break
+
+        additional_data[f"{self.ADDITIONAL_DATA_KEY}_LICENCE"] = self.LICENCE

datastore/tests/test_additional_data_codelist_code.py

@@ -37,7 +37,8 @@ def test_code_list(self):
                 "beneficiaryLocation_countryCode": "Australia",
                 "fundingOrg_location_countryCode": "France",
                 "recipientOrg_location_countryCode": "United Kingdom of Great Britain and Northern Ireland",
-            }
+            },
+            "codeListLookup_LICENCE": "https://creativecommons.org/licenses/by/4.0/",
         }
 
         source.update_additional_data(grant, source_file, additional_data_in)

datastore/tests/test_additional_data_license.py

@@ -1,8 +1,15 @@
 from django.test import TestCase
 
 from additional_data.sources.grant_metadata import GrantMetadataSource
+from additional_data.sources.find_that_charity import FindThatCharitySource
+from additional_data.sources.geo_lookup import GeoLookupSource
+from additional_data.sources.nspl import NSPLSource
+from additional_data.sources.codelist_code import CodeListSource
 from db.models import Grant
 
+OGL_V3 = "https://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/"
+CC_BY_4 = "https://creativecommons.org/licenses/by/4.0/"
+
 
 class TestAdditionalDataLicense(TestCase):
     fixtures = ["test_data.json"]
@@ -35,3 +42,40 @@ def test_publisher_license(self):
             expected_additional_data,
             "The additional license data added is not correct.",
         )
+
+
+class TestAdditionalDataSourceLicences(TestCase):
+    fixtures = ["test_data.json"]
+
+    def _assert_licence(self, source, expected_licence):
+        grant = Grant.objects.first()
+        additional_data = {}
+
+        source.update_additional_data(
+            grant.data, grant.source_file.data, additional_data
+        )
+
+        licence_key = f"{source.ADDITIONAL_DATA_KEY}_LICENCE"
+
+        self.assertIn(
+            licence_key,
+            additional_data,
+            f"{licence_key} was not added by {source.__class__.__name__}.",
+        )
+        self.assertEqual(
+            additional_data[licence_key],
+            expected_licence,
+            f"{source.__class__.__name__} set the wrong licence value.",
+        )
+
+    def test_find_that_charity_licence(self):
+        self._assert_licence(FindThatCharitySource(), OGL_V3)
+
+    def test_geo_lookup_licence(self):
+        self._assert_licence(GeoLookupSource(), OGL_V3)
+
+    def test_nspl_licence(self):
+        self._assert_licence(NSPLSource(), OGL_V3)
+
+    def test_codelist_lookup_licence(self):
+        self._assert_licence(CodeListSource(), CC_BY_4)

datastore/tests/test_additional_data_nspl.py

@@ -200,7 +200,8 @@ def test_nspl_update_additional_data_with_not_existing_postcode(self):
         nspl.update_additional_data(grant.data, grant.source_file.data, additional_data)
 
         self.assertNotIn("recipientOrganizationLocation", additional_data)
-        self.assertEqual(len(additional_data), 0)
+        self.assertEqual(len(additional_data), 1)
+        self.assertIn("recipientOrganizationLocation_LICENCE", additional_data)
 
     def test_nspl_update_additional_data_without_postcode(self):
         self.save_nspl_mock_data()
@@ -214,7 +215,8 @@ def test_nspl_update_additional_data_without_postcode(self):
         nspl.update_additional_data(grant.data, grant.source_file.data, additional_data)
 
         self.assertNotIn("recipientOrganizationLocation", additional_data)
-        self.assertEqual(len(additional_data), 0)
+        self.assertEqual(len(additional_data), 1)
+        self.assertIn("recipientOrganizationLocation_LICENCE", additional_data)
 
     def test_nspl_update_additional_data_breaks_after_updating(self):
         # Once we have one recipient org location, the loop should break.