add raw source fixture and with/without sort test.

tests/unit/compare/conftest.py

@@ -56,3 +56,31 @@ def data_sources():
         )  # the response is stored sorted
 
     return harvest_source, ckan_source
+
+
+@pytest.fixture
+def data_sources_raw():
+    harvest_source_datasets = open_json(
+        HARVEST_SOURCES / "dcatus" / "dcatus_compare.json"
+    )["dataset"]
+
+    harvest_source = {d["identifier"]: d for d in harvest_source_datasets}
+
+    ckan_source_datasets = open_json(
+        HARVEST_SOURCES / "dcatus" / "ckan_datasets_resp.json"
+    )["result"]["results"]
+
+    ckan_source = {}
+
+    for d in ckan_source_datasets:
+        orig_meta = None
+        orig_id = None
+        for e in d["extras"]:
+            if e["key"] == "dcat_metadata":
+                orig_meta = eval(e["value"], {"__builtins__": {}})
+            if e["key"] == "identifier":
+                orig_id = e["value"]
+
+        ckan_source[orig_id] = orig_meta
+
+    return harvest_source, ckan_source

tests/unit/compare/test_compare.py

@@ -1,4 +1,5 @@
 from harvester.compare import compare
+from harvester.utils.util import sort_dataset, dataset_to_hash
 
 
 def test_artificial_compare(artificial_data_sources):
@@ -17,3 +18,32 @@ def test_compare(data_sources):
     assert len(compare_res["create"]) == 1
     assert len(compare_res["update"]) == 3
     assert len(compare_res["delete"]) == 1
+
+
+def test_sort(data_sources_raw):
+    harvest_source, ckan_source = data_sources_raw
+
+    harvest_source_no_sort = harvest_source.copy()
+    for k, v in harvest_source_no_sort.items():
+        harvest_source_no_sort[k] = dataset_to_hash(v)
+
+    for k, v in ckan_source.items():
+        ckan_source[k] = dataset_to_hash(v)
+
+    compare_res_no_sort = compare(harvest_source_no_sort, ckan_source)
+
+    # more datasets need to be updated simply because we didn't sort them
+    assert len(compare_res_no_sort["create"]) == 1
+    assert len(compare_res_no_sort["update"]) == 6
+    assert len(compare_res_no_sort["delete"]) == 1
+
+    harvest_source_with_sort = harvest_source.copy()
+    for k, v in harvest_source_with_sort.items():
+        harvest_source_with_sort[k] = dataset_to_hash(sort_dataset(v))
+
+    compare_res = compare(harvest_source_with_sort, ckan_source)
+
+    # applying the sort lowers us back down to what we expect.
+    assert len(compare_res["create"]) == 1
+    assert len(compare_res["update"]) == 3
+    assert len(compare_res["delete"]) == 1