test(fairdatapoint): fix unit tests for FairDataPointCivityHarvester setup methods

Author: Hans-Chrstian

ckanext/fairdatapoint/harvesters/fair_data_point_civity_harvester.py

@@ -21,21 +21,6 @@
 
 
 class FairDataPointCivityHarvester(CivityHarvester):
-
-    def _get_harvest_catalog_setting(self, harvest_config_dict):
-        if HARVEST_CATALOG in harvest_config_dict:
-            log.debug("Using harvest_catalogs from harvest_config_dict")
-            harvest_catalog_setting = toolkit.asbool(
-                harvest_config_dict[HARVEST_CATALOG]
-            )
-        else:
-            log.debug("Using harvest_catalogs from global CKAN config")
-            harvest_catalog_setting = toolkit.asbool(
-                toolkit.config.get(HARVEST_CATALOG_CONFIG, False)
-            )
-        log.debug("Harvesting catalogs is set to %s", harvest_catalog_setting)
-        return harvest_catalog_setting
-
     def setup_record_provider(self, harvest_url, harvest_config_dict):
         # Harvest catalog config can be set on global CKAN level, but can be overriden by harvest config
         harvest_catalogs = self._get_harvest_catalog_setting(harvest_config_dict)
@@ -52,9 +37,25 @@ def setup_record_to_package_converter(self, harvest_url, harvest_config_dict):
         else:
             raise Exception("[{0}] not found in harvester config JSON".format(PROFILE))
 
-    def info(self):
+    @staticmethod
+    def info():
         return {
             "name": "fair_data_point_harvester",
             "title": "FAIR data point harvester",
             "description": "Harvester for end points implementing the FAIR data point protocol",
         }
+
+    @staticmethod
+    def _get_harvest_catalog_setting(harvest_config_dict):
+        if HARVEST_CATALOG in harvest_config_dict:
+            log.debug("Using harvest_catalogs from harvest_config_dict")
+            harvest_catalog_setting = toolkit.asbool(
+                harvest_config_dict[HARVEST_CATALOG]
+            )
+        else:
+            log.debug("Using harvest_catalogs from global CKAN config")
+            harvest_catalog_setting = toolkit.asbool(
+                toolkit.config.get(HARVEST_CATALOG_CONFIG, False)
+            )
+        log.debug("Harvesting catalogs is set to %s", harvest_catalog_setting)
+        return harvest_catalog_setting

ckanext/fairdatapoint/tests/test_data/process_catalogs.ttl

@@ -0,0 +1,10 @@
+# SPDX-FileCopyrightText: 2024 Stichting Health-RI
+#
+# SPDX-License-Identifier: AGPL-3.0-only
+
+@prefix ldp: <http://www.w3.org/ns/ldp#> .
+@prefix dcat: <http://www.w3.org/ns/dcat#> .
+<http://example.com> ldp:contains <http://example.com/catalog1> .
+<http://example.com/catalog1> a dcat:Catalog .
+<http://example.com/catalog1> dcat:dataset <http://example.com/dataset1> .
+<http://example.com/dataset1> a dcat:Dataset .

ckanext/fairdatapoint/tests/test_fair_data_point_civity_harvester.py

@@ -0,0 +1,74 @@
+# SPDX-FileCopyrightText: 2024 Stichting Health-RI
+#
+# SPDX-License-Identifier: AGPL-3.0-only
+
+import unittest
+import ckanext.fairdatapoint.plugin as plugin
+from unittest.mock import patch, MagicMock
+from ckanext.fairdatapoint.harvesters import FairDataPointCivityHarvester
+from ckanext.fairdatapoint.harvesters import fair_data_point_civity_harvester
+
+
+class TestFairDataPointCivityHarvester(unittest.TestCase):
+
+    def setUp(self):
+        plugin.toolkit = MagicMock()
+
+    def test_get_harvest_catalog_setting_from_dict(self):
+        harvester = FairDataPointCivityHarvester()
+        harvest_config_dict = {fair_data_point_civity_harvester.HARVEST_CATALOG: 'true'}
+        result = harvester._get_harvest_catalog_setting(harvest_config_dict)
+        self.assertTrue(result)
+
+    @patch('ckan.plugins.toolkit.config')
+    def test_get_harvest_catalog_setting_from_global_config(self, mock_config):
+        mock_config.get.return_value = 'false'
+        harvester = FairDataPointCivityHarvester()
+
+        harvest_config_dict = {}
+        result = harvester._get_harvest_catalog_setting(harvest_config_dict)
+
+        self.assertFalse(result)
+        mock_config.get.assert_called_once_with(fair_data_point_civity_harvester.HARVEST_CATALOG_CONFIG, False)
+
+    @patch('ckanext.fairdatapoint.harvesters.domain.fair_data_point_record_provider.FairDataPointRecordProvider'
+           '.__init__')
+    def test_setup_record_provider(self, mock_record_provider):
+        mock_record_provider.return_value = None
+        harvester = FairDataPointCivityHarvester()
+        harvester._get_harvest_catalog_setting = MagicMock(return_value=True)
+        harvest_url = 'http://example.com'
+        harvest_config_dict = {fair_data_point_civity_harvester.HARVEST_CATALOG: 'true'}
+        harvester.setup_record_provider(harvest_url, harvest_config_dict)
+        harvester._get_harvest_catalog_setting.assert_called_once_with(harvest_config_dict)
+        mock_record_provider.assert_called_once_with(harvest_url, True)
+
+    @patch(
+        'ckanext.fairdatapoint.harvesters.domain.fair_data_point_record_to_package_converter'
+        '.FairDataPointRecordToPackageConverter'
+        '.__init__')
+    def test_setup_record_to_package_converter_with_profile(self, mock_converter):
+        mock_converter.return_value = None
+        harvester = FairDataPointCivityHarvester()
+        harvest_url = 'http://example.com'
+        harvest_config_dict = {fair_data_point_civity_harvester.PROFILE: 'test_profile'}
+        harvester.setup_record_to_package_converter(harvest_url, harvest_config_dict)
+        mock_converter.assert_called_once_with('test_profile')
+
+    def test_setup_record_to_package_converter_raises_exception(self):
+        # Instantiate the harvester
+        harvester = FairDataPointCivityHarvester()
+
+        # Test data without PROFILE in the dictionary
+        harvest_url = 'http://example.com'
+        harvest_config_dict = {}  # No PROFILE key
+
+        # Verify that an exception is raised when PROFILE is missing
+        with self.assertRaises(Exception) as context:
+            harvester.setup_record_to_package_converter(harvest_url, harvest_config_dict)
+
+        # Check the exception message
+        self.assertEqual(
+            str(context.exception),
+            "[profile] not found in harvester config JSON"
+        )

ckanext/fairdatapoint/tests/test_record_provider.py

@@ -12,19 +12,19 @@
 TEST_DATA_DIRECTORY = Path(Path(__file__).parent.resolve(), "test_data")
 
 TEST_CAT_IDS_DICT = {
-            "catalog=https://fair.healthinformationportal.eu/catalog/1c75c2c9-d2cc-44cb-aaa8-cf8c11515c8d":
-                URIRef("https://fair.healthinformationportal.eu/catalog/1c75c2c9-d2cc-44cb-aaa8-cf8c11515c8d"),
-            "catalog=https://fair.healthinformationportal.eu/catalog/1c75c2c9-d2cc-44cb-aaa8-cf8c11515c8d;"
-            "dataset=https://fair.healthinformationportal.eu/dataset/898ca4b8-197b-4d40-bc81-d9cd88197670":
-                URIRef("https://fair.healthinformationportal.eu/dataset/898ca4b8-197b-4d40-bc81-d9cd88197670"),
-            "catalog=https://fair.healthinformationportal.eu/catalog/14225c50-00b0-4fba-8300-a677ab0c86f4":
-                URIRef("https://fair.healthinformationportal.eu/catalog/14225c50-00b0-4fba-8300-a677ab0c86f4"),
-            "catalog=https://fair.healthinformationportal.eu/catalog/14225c50-00b0-4fba-8300-a677ab0c86f4;"
-            "dataset=https://fair.healthinformationportal.eu/dataset/32bd0246-b731-480a-b5f4-a2f60ccaebc9":
-                URIRef("https://fair.healthinformationportal.eu/dataset/32bd0246-b731-480a-b5f4-a2f60ccaebc9"),
-            "catalog=https://fair.healthinformationportal.eu/catalog/17412bc2-daf1-491e-94fb-6680f7a67b1e":
-                URIRef("https://fair.healthinformationportal.eu/catalog/17412bc2-daf1-491e-94fb-6680f7a67b1e")
-        }
+    "catalog=https://fair.healthinformationportal.eu/catalog/1c75c2c9-d2cc-44cb-aaa8-cf8c11515c8d":
+        URIRef("https://fair.healthinformationportal.eu/catalog/1c75c2c9-d2cc-44cb-aaa8-cf8c11515c8d"),
+    "catalog=https://fair.healthinformationportal.eu/catalog/1c75c2c9-d2cc-44cb-aaa8-cf8c11515c8d;"
+    "dataset=https://fair.healthinformationportal.eu/dataset/898ca4b8-197b-4d40-bc81-d9cd88197670":
+        URIRef("https://fair.healthinformationportal.eu/dataset/898ca4b8-197b-4d40-bc81-d9cd88197670"),
+    "catalog=https://fair.healthinformationportal.eu/catalog/14225c50-00b0-4fba-8300-a677ab0c86f4":
+        URIRef("https://fair.healthinformationportal.eu/catalog/14225c50-00b0-4fba-8300-a677ab0c86f4"),
+    "catalog=https://fair.healthinformationportal.eu/catalog/14225c50-00b0-4fba-8300-a677ab0c86f4;"
+    "dataset=https://fair.healthinformationportal.eu/dataset/32bd0246-b731-480a-b5f4-a2f60ccaebc9":
+        URIRef("https://fair.healthinformationportal.eu/dataset/32bd0246-b731-480a-b5f4-a2f60ccaebc9"),
+    "catalog=https://fair.healthinformationportal.eu/catalog/17412bc2-daf1-491e-94fb-6680f7a67b1e":
+        URIRef("https://fair.healthinformationportal.eu/catalog/17412bc2-daf1-491e-94fb-6680f7a67b1e")
+}
 
 
 def get_graph_by_id(*args, **kwargs):
@@ -39,12 +39,12 @@ class TestRecordProvider:
     fdp_record_provider = FairDataPointRecordProvider("http://test_end_point.com")
 
     @pytest.mark.parametrize("fdp_response_file,expected",
-                         [
-                             (Path(TEST_DATA_DIRECTORY, "root_fdp_response.ttl"),
-                              TEST_CAT_IDS_DICT.keys()),
-                             (Path(TEST_DATA_DIRECTORY, "root_fdp_response_no_catalogs.ttl"),
-                              dict().keys())
-                          ])
+                             [
+                                 (Path(TEST_DATA_DIRECTORY, "root_fdp_response.ttl"),
+                                  TEST_CAT_IDS_DICT.keys()),
+                                 (Path(TEST_DATA_DIRECTORY, "root_fdp_response_no_catalogs.ttl"),
+                                  dict().keys())
+                             ])
     def test_get_record_ids(self, mocker, fdp_response_file, expected):
         fdp_get_graph = mocker.MagicMock(name="get_data")
         mocker.patch("ckanext.fairdatapoint.harvesters.domain.fair_data_point.FairDataPoint.get_graph",
@@ -92,7 +92,7 @@ def test_get_record_by_id_distr(self, mocker):
     def test_orcid_call(self, mocker):
         """if orcid url in contact point - add vcard full name"""
         with requests_mock.Mocker() as mock:
-            mock.get("https://orcid.org/0000-0002-4348-707X/public-record.json", 
+            mock.get("https://orcid.org/0000-0002-4348-707X/public-record.json",
                      json={"displayName": "N.K. De Vries"})
             fdp_get_graph = mocker.MagicMock(name="get_data")
             mocker.patch("ckanext.fairdatapoint.harvesters.domain.fair_data_point.FairDataPoint.get_graph",
@@ -105,3 +105,27 @@ def test_orcid_call(self, mocker):
             expected = Graph().parse(Path(TEST_DATA_DIRECTORY, "Project_27866022694497978_out.ttl")).serialize()
             assert mock.called
             assert actual == expected
+
+    @pytest.mark.parametrize("harvest_catalogs, expected_results", [
+        (False, {
+            'catalog=http://example.com/catalog1;dataset=http://example.com/dataset1': URIRef(
+                'http://example.com/dataset1')
+        }),
+        (True, {
+            'catalog=http://example.com/catalog1;dataset=http://example.com/dataset1': URIRef(
+                'http://example.com/dataset1'),
+            'catalog=http://example.com/catalog1': URIRef('http://example.com/catalog1')
+        })
+    ])
+    def test_process_catalogs(self, mocker, harvest_catalogs, expected_results):
+        fdp_get_graph = mocker.MagicMock(name="get_data")
+        mocker.patch("ckanext.fairdatapoint.harvesters.domain.fair_data_point.FairDataPoint.get_graph",
+                     new=fdp_get_graph)
+        fdp_get_graph.return_value = Graph().parse(Path(TEST_DATA_DIRECTORY, "process_catalogs.ttl"))
+
+        self.fdp_record_provider.harvest_catalogs = harvest_catalogs
+        actual_result = self.fdp_record_provider._process_catalog("http://example.com/catalog1")
+
+        # Assertions
+        for key, value in expected_results.items():
+            assert actual_result[key] == value