[#56] More robust date parsing with dateutil, expand tests

ckanext/dcat/tests/test_scheming_support.py

@@ -369,7 +369,9 @@ def test_e2e_ckan_to_dcat(self):
         # Resources: standard fields
 
         assert self._triple(g, distribution_ref, DCT.rights, resource["rights"])
-        assert self._triple(g, distribution_ref, ADMS.status, URIRef(resource["status"]))
+        assert self._triple(
+            g, distribution_ref, ADMS.status, URIRef(resource["status"])
+        )
         assert self._triple(
             g,
             distribution_ref,
@@ -566,6 +568,88 @@ def test_legacy_fields(self):
         assert len(publisher) == 1
         assert self._triple(g, publisher[0][2], FOAF.name, "Test Publisher")
 
+    def test_dcat_date(self):
+        dataset_dict = {
+            # Core fields
+            "name": "test-dataset",
+            "title": "Test DCAT dataset",
+            "notes": "Some notes",
+            "issued": "2024",
+            "modified": "2024-10",
+            "temporal_coverage": [
+                {"start": "1905-03-01T10:07:31.182680", "end": "2013-01-05"},
+                {"start": "2024-04-10T10:07:31", "end": "2024-05-29"},
+            ],
+        }
+
+        dataset = call_action("package_create", **dataset_dict)
+
+        s = RDFSerializer()
+        g = s.g
+
+        dataset_ref = s.graph_from_dataset(dataset)
+
+        # Year
+        assert dataset["issued"] == dataset_dict["issued"]
+        assert self._triple(
+            g,
+            dataset_ref,
+            DCT.issued,
+            dataset_dict["issued"],
+            data_type=XSD.gYear,
+        )
+
+        # Year-month
+        assert dataset["modified"] == dataset_dict["modified"]
+        assert self._triple(
+            g,
+            dataset_ref,
+            DCT.modified,
+            dataset_dict["modified"],
+            data_type=XSD.gYearMonth,
+        )
+
+        temporal = [t for t in g.triples((dataset_ref, DCT.temporal, None))]
+
+        # Date
+        assert (
+            dataset["temporal_coverage"][0]["end"]
+            == dataset_dict["temporal_coverage"][0]["end"]
+        )
+
+        assert self._triple(
+            g,
+            temporal[0][2],
+            SCHEMA.endDate,
+            dataset_dict["temporal_coverage"][0]["end"],
+            data_type=XSD.date,
+        )
+
+        # Datetime
+        assert (
+            dataset["temporal_coverage"][0]["start"]
+            == dataset_dict["temporal_coverage"][0]["start"]
+        )
+        assert self._triple(
+            g,
+            temporal[0][2],
+            SCHEMA.startDate,
+            dataset_dict["temporal_coverage"][0]["start"],
+            data_type=XSD.dateTime,
+        )
+
+        assert (
+            dataset["temporal_coverage"][1]["start"]
+            == dataset_dict["temporal_coverage"][1]["start"]
+        )
+        assert self._triple(
+            g,
+            temporal[1][2],
+            SCHEMA.startDate,
+            dataset_dict["temporal_coverage"][1]["start"],
+            data_type=XSD.dateTime,
+        )
+
 
 @pytest.mark.usefixtures("with_plugins", "clean_db")
 @pytest.mark.ckan_config("ckan.plugins", "dcat scheming_datasets")
@@ -859,55 +943,3 @@ def test_spatial_field(self):
             assert search_dict["spatial"] == json.dumps(
                 dataset_dict["spatial_coverage"][0]["centroid"]
             )
-
-
-@pytest.mark.usefixtures("with_plugins", "clean_db")
-@pytest.mark.ckan_config("ckan.plugins", "dcat scheming_datasets")
-@pytest.mark.ckan_config(
-    "scheming.dataset_schemas", "ckanext.dcat.schemas:dcat_ap_2.1_full.yaml"
-)
-@pytest.mark.ckan_config(
-    "scheming.presets",
-    "ckanext.scheming:presets.json ckanext.dcat.schemas:presets.yaml",
-)
-@pytest.mark.ckan_config(
-    "ckanext.dcat.rdf.profiles", "euro_dcat_ap_2 euro_dcat_ap_scheming"
-)
-class TestSchemingPresets:
-    def test_dcat_date(self):
-        dataset_dict = {
-            # Core fields
-            "name": "test-dataset",
-            "title": "Test DCAT dataset",
-            "notes": "Some notes",
-            "issued": "2024",
-            "modified": "2024-10",
-            "temporal_coverage": [
-                {"start": "1905-03-01T10:07:31.182680", "end": "2013-01-05"},
-                {"start": "2024-04-10T10:07:31", "end": "2024-05-29"},
-            ],
-        }
-
-        dataset = call_action("package_create", **dataset_dict)
-
-        # Year
-        assert dataset["issued"] == dataset_dict["issued"]
-
-        # Year-month
-        assert dataset["modified"] == dataset_dict["modified"]
-
-        # Date
-        assert (
-            dataset["temporal_coverage"][0]["end"]
-            == dataset_dict["temporal_coverage"][0]["end"]
-        )
-
-        # Datetime
-        assert (
-            dataset["temporal_coverage"][0]["start"]
-            == dataset_dict["temporal_coverage"][0]["start"]
-        )
-        assert (
-            dataset["temporal_coverage"][1]["start"]
-            == dataset_dict["temporal_coverage"][1]["start"]
-        )

ckanext/dcat/tests/test_validators.py

@@ -1,8 +1,10 @@
+import datetime
 import json
+
 import pytest
 
 from ckantoolkit import StopOnError
-from ckanext.dcat.validators import scheming_multiple_number
+from ckanext.dcat.validators import scheming_multiple_number, dcat_date
 
 
 def test_scheming_multiple_number():
@@ -60,3 +62,26 @@ def test_scheming_multiple_number_wrong_value():
         assert errors[key][0].startswith("invalid type for repeating number")
 
         errors = {key: []}
+
+
+@pytest.mark.usefixtures(
+    "with_plugins",
+)
+@pytest.mark.ckan_config("ckan.plugins", "dcat scheming_datasets")
+def test_dcat_date_valid():
+
+    key = ("some_date",)
+    errors = {key: []}
+    valid_values = [
+        datetime.datetime.now(),
+        "2024",
+        "2024-07",
+        "2024-07-01",
+        "1905-03-01T10:07:31.182680",
+        "2024-04-10T10:07:31",
+        "2024-04-10T10:07:31.000Z",
+    ]
+
+    for value in valid_values:
+        data = {key: value}
+        dcat_date(key, data, errors, {}), value

ckanext/dcat/validators.py

@@ -2,10 +2,10 @@
 import json
 import re
 
+from dateutil.parser import parse as parse_date
 from ckantoolkit import (
     missing,
     StopOnError,
-    get_validator,
     Invalid,
     _,
 )
@@ -41,17 +41,15 @@ def is_date(value):
 def dcat_date(key, data, errors, context):
     value = data[key]
 
-    scheming_isodatetime = get_validator("scheming_isodatetime")
-
     if isinstance(value, datetime.datetime):
         return
 
     if is_year(value) or is_year_month(value) or is_date(value):
         return
 
     try:
-        scheming_isodatetime({}, {})(key, data, errors, context)
-    except Invalid:
+        parse_date(value)
+    except ValueError:
         raise Invalid(
             _(
                 "Date format incorrect. Supported formats are YYYY, YYYY-MM, YYYY-MM-DD and YYYY-MM-DDTHH:MM:SS"