Get full test suite working

Author: tmorrell

caltechdata_api/customize_schema.py

@@ -409,11 +409,11 @@ def validate_metadata(json_record):
 
     # Publication date is handled by customize function
 
-    # Check for 'resourceType'
-    if "resourceType" not in json_record["types"]:
-        errors.append("'resourceType' field is missing in 'types'.")
-    elif not isinstance(json_record["types"]["resourceType"], str):
-        errors.append("'resourceType' should be a string.")
+    # Check for 'resourceTypeGeneral'
+    if "resourceTypeGeneral" not in json_record["types"]:
+        errors.append("'resourceTypeGeneral' field is missing in 'types'.")
+    elif not isinstance(json_record["types"]["resourceTypeGeneral"], str):
+        errors.append("'resourceTypeGeneral' should be a string.")
 
     # Check for 'identifiers'
     if "identifiers" in json_record:

tests/conftest.py

@@ -4,7 +4,7 @@
 @pytest.fixture(scope="function")
 def full_datacite43_record():
     return {
-        "types": {"resourceTypeGeneral": "Image", "resourceType": "Photo"},
+        "types": {"resourceTypeGeneral": "Dataset"},
         "creators": [
             {
                 "name": "Nielsen, Lars Holm",
@@ -40,9 +40,9 @@ def full_datacite43_record():
         "subjects": [
             {"subject": "custom"},
             {
-                "subject": "Abdominal Injuries",
-                "subjectScheme": "MeSH",
-                "valueURI": "http://id.nlm.nih.gov/mesh/A-D000007",
+                "subject": "Chemical sciences",
+                "subjectScheme": "FOS",
+                "valueURI": "http://www.oecd.org/science/inno/38235147.pdf?1.4",
             },
         ],
         "contributors": [
@@ -135,7 +135,7 @@ def full_rdm_record():
     """Full record data from DataCite as dict coming from the external world."""
     return {
         "metadata": {
-            "resource_type": {"id": "image-photo"},
+            "resource_type": {"id": "dataset"},
             "creators": [
                 {
                     "person_or_org": {
@@ -165,7 +165,7 @@ def full_rdm_record():
             "publication_date": "2018/2020-09",
             "subjects": [
                 {"subject": "custom"},
-                {"id": "http://id.nlm.nih.gov/mesh/A-D000007"},
+                {"id": "http://www.oecd.org/science/inno/38235147.pdf?1.4"},
             ],
             "contributors": [
                 {
@@ -230,14 +230,11 @@ def full_rdm_record():
             "funding": [
                 {
                     "funder": {
-                        "title": {
-                            "en": "OpenAIRE",
-                        },
-                        "identifier": "00k4n6c32",
-                        "scheme": "ror",
+                        "name": "European Commission",
+                        "id": "00k4n6c32",
                     },
                     "award": {
-                        "title": "OpenAIRE",
+                        "title": {"en": "OpenAIRE"},
                         "number": "246686",
                         "identifier": ".../246686",
                     },

tests/test_rdm.py

@@ -25,17 +25,26 @@ def test_datacite_rdm_conversion(full_datacite43_record, full_rdm_record):
 
 def test_datacite_rdm_create_edit(full_datacite43_record):
     env_token = os.environ.get("RDMTOK")
-    doi = caltechdata_write(
+
+    # Remove DOI from full_datacite43_record
+    # since we want the test system to create one
+    identifiers = []
+    for identifier in full_datacite43_record["identifiers"]:
+        if identifier["identifierType"] != "DOI":
+            identifiers.append(identifier)
+    full_datacite43_record["identifiers"] = identifiers
+
+    recid = caltechdata_write(
         full_datacite43_record,
         schema="43",
         production=False,
         publish=True,
         token=env_token,
     )
 
-    assert doi.startswith("10.33569")
+    assert len(recid) == 11
 
-    doi = caltechdata_write(
+    recid = caltechdata_write(
         full_datacite43_record,
         schema="43",
         production=False,
@@ -44,19 +53,12 @@ def test_datacite_rdm_create_edit(full_datacite43_record):
         token=env_token,
     )
 
-    assert doi.startswith("10.33569")
-
-    # If we don't publish, don't get back a DOI
-    idv = caltechdata_write(
-        full_datacite43_record, schema="43", production=False, token=env_token
-    )
-
-    assert idv.startswith("10.33569") == False
+    assert len(recid) == 11
 
     full_datacite43_record["publisher"] = "Edited"
 
     doi = caltechdata_edit(
-        idv,
+        recid,
         full_datacite43_record,
         schema="43",
         production=False,
@@ -65,16 +67,16 @@ def test_datacite_rdm_create_edit(full_datacite43_record):
     )
 
     assert doi.startswith("10.33569")
-    idv = doi.split("/")[1]
 
-    new_metadata = get_metadata(idv, production=False, publish=True)
+    # Validate is false until geolocation points are fixed/we move to 4.6
+    new_metadata = get_metadata(recid, production=False, validate=False)
 
     assert new_metadata["publisher"] == "Edited"
 
     full_datacite43_record["publisher"] = "Again!"
 
     new_doi = caltechdata_edit(
-        idv,
+        recid,
         full_datacite43_record,
         files=["codemeta.json"],
         schema="43",
@@ -85,8 +87,9 @@ def test_datacite_rdm_create_edit(full_datacite43_record):
 
     assert new_doi != doi
 
-    idv = new_doi.split("/")[1]
+    recid = new_doi.split("/")[1]
 
-    new_metadata = get_metadata(idv, production=False)
+    # Validate is false until geolocation points are fixed/we move to 4.6
+    new_metadata = get_metadata(recid, production=False, validate=False)
 
     assert new_metadata["publisher"] == "Again!"