Merge branch 'standards-for-ara-kp-specification'

.gitignore

@@ -59,6 +59,9 @@ coverage.xml
 .hypothesis/
 .pytest_cache/
 
+# local Unit test configuration
+pytest.ini
+
 # Translations
 *.mo
 *.pot

tests/onehop/.gitignore

@@ -1,6 +1,3 @@
 /test_results/
 /test-database/
 /file-report-unit-test-database/
-
-# ignore local Pytest config file
-pytest.ini

tests/onehop/conftest.py

@@ -10,6 +10,7 @@
 
 import logging
 
+from deprecation import deprecated
 from pytest_harvest import get_session_results_dct
 
 from reasoner_validator.biolink import check_biolink_model_compliance_of_input_edge
@@ -37,6 +38,11 @@
 logger = logging.getLogger(__name__)
 
 
+# TODO: temporary circuit breaker for (currently 4-August-2022)
+#  undisciplined edge test data sets (like from RTX-KG2c)
+UNREASONABLE_NUMBER_OF_TEST_EDGES: int = 10
+
+
 def pytest_sessionfinish(session):
     """ Gather all results and save them to a csv.
     Works both on worker and master nodes, and also with xdist disabled
@@ -360,13 +366,15 @@ def load_test_data_source(
 component_catalog: Dict[str, Dict[str, Any]] = dict()
 
 
+@deprecated
 def cache_resource_metadata(metadata: Dict[str, Any]):
     component = metadata['component']
     assert component in ["KP", "ARA"]
     resource_id: str = metadata['api_name']
     component_catalog[resource_id] = metadata
 
 
+@deprecated
 def get_metadata_by_resource(resource_id: str) -> Optional[Dict[str, Any]]:
     if resource_id in component_catalog:
         metadata: Dict = component_catalog[resource_id]
@@ -375,6 +383,7 @@ def get_metadata_by_resource(resource_id: str) -> Optional[Dict[str, Any]]:
         return None
 
 
+@deprecated
 def get_component_by_resource(resource_id: str) -> Optional[str]:
     metadata: Dict = get_metadata_by_resource(resource_id)
     if metadata and "component" in metadata:
@@ -386,6 +395,7 @@ def get_component_by_resource(resource_id: str) -> Optional[str]:
 kp_edges_catalog: Dict[str, Dict[str,  Union[int, str]]] = dict()
 
 
+@deprecated
 def add_kp_edge(resource_id: str, edge_idx: int, edge: Dict[str, Any]):
     metadata: Dict = get_metadata_by_resource(resource_id)
     assert metadata
@@ -396,6 +406,7 @@ def add_kp_edge(resource_id: str, edge_idx: int, edge: Dict[str, Any]):
     metadata['edges'][edge_idx] = edge
 
 
+@deprecated
 def get_kp_edge(resource_id: str, edge_idx: int) -> Optional[Dict[str, Any]]:
     metadata: Dict = get_metadata_by_resource(resource_id)
     if metadata:
@@ -416,8 +427,8 @@ def generate_trapi_kp_tests(metafunc, trapi_version: str, biolink_version: str)
     :param trapi_version, str, TRAPI release set to be used in the validation
     :param biolink_version, str, Biolink Model release set to be used in the validation
     """
-    edges = []
-    idlist = []
+    edges: List = []
+    idlist: List = []
 
     # TODO: test_run_id is currently unused in this method; it is otherwise an
     #       optional user session identifier for the test (can be an empty string)
@@ -439,7 +450,8 @@ def generate_trapi_kp_tests(metafunc, trapi_version: str, biolink_version: str)
             )
             continue
 
-        cache_resource_metadata(kpjson)
+        # No point in caching for latest implementation of reporting
+        # cache_resource_metadata(kpjson)
 
         dataset_level_test_exclusions: Set = set()
         if "exclude_tests" in kpjson:
@@ -490,8 +502,13 @@ def generate_trapi_kp_tests(metafunc, trapi_version: str, biolink_version: str)
                     f"generate_trapi_kp_tests(): input file '{source}' "
                     "is missing its 'infores' field value? Inferred from its API name?"
                 )
+                # create a pseudo-infores from a lower cased and hyphenated API name
                 kp_api_name: str = edge['kp_api_name']
-                edge['kp_source'] = f"infores:{kp_api_name.lower()}"
+                if not kp_api_name:
+                    logger.warning("generate_trapi_kp_tests(): KP API Name is missing? Skipping entry...")
+                    continue
+                kp_infores_object_id = kp_api_name.lower().replace("_", "-")
+                edge['kp_source'] = f"infores:{kp_infores_object_id}"
 
             if 'source_type' in kpjson:
                 edge['kp_source_type'] = kpjson['source_type']
@@ -534,6 +551,10 @@ def generate_trapi_kp_tests(metafunc, trapi_version: str, biolink_version: str)
             if metafunc.config.getoption('one', default=False):
                 break
 
+            # Circuit breaker for overly large edge test data sets
+            if edge_i > UNREASONABLE_NUMBER_OF_TEST_EDGES:
+                break
+
         print(f"### End of Test Input Edges for KP '{kpjson['api_name']}' ###")
 
     if "kp_trapi_case" in metafunc.fixturenames:
@@ -574,7 +595,8 @@ def generate_trapi_ara_tests(metafunc, kp_edges, trapi_version, biolink_version)
     """
     kp_dict = defaultdict(list)
     for e in kp_edges:
-        kp_dict[e['kp_api_name']].append(e)
+        # We connect ARA's to their KPs by infores (== kp_source) now...
+        kp_dict[e['kp_source']].append(e)
 
     ara_edges = []
     idlist = []
@@ -595,13 +617,24 @@ def generate_trapi_ara_tests(metafunc, kp_edges, trapi_version, biolink_version)
             )
             continue
 
-        cache_resource_metadata(arajson)
+        # No point in caching for latest implementation of reporting
+        # cache_resource_metadata(arajson)
 
         for kp in arajson['KPs']:
 
-            # By replacing spaces in name with underscores,
-            # should give get the KP "api_name" indexing the edges.
-            kp = '_'.join(kp.split())
+            #
+            # TODO: use of KP infores ('kp_source') CURIES in the Registry ARA spec
+            #       likely now completely breaks the old filename-centric
+            #       (non-Registry) local test_triples mechanism for KP resolution
+            # # By replacing spaces in name with underscores,
+            # # should give get the KP "api_name" indexing the edges.
+            # kp = '_'.join(kp.split())
+
+            if kp not in kp_dict:
+                logger.warning(
+                    f"generate_trapi_ara_tests(): '{kp}' test edges not (yet) available for ARA {source}. Skipping..."
+                )
+                continue
 
             for edge_i, kp_edge in enumerate(kp_dict[kp]):
 

tests/onehop/test_triples/ARA/Unit_Test_ARA/Test_ARA.json

@@ -2,7 +2,7 @@
     "url": "https://aragorn.renci.org/1.2",
     "infores": "aragorn",
     "KPs": [
-        "Test KP 1",
-        "Test KP 2"
+        "test-kp-1",
+        "test-kp-2"
     ]
 }

tests/onehop/test_triples/KP/Unit_Test_KP/Test_KP_1.json

@@ -1,7 +1,7 @@
 {
     "url": "https://automat.renci.org/panther/1.2",
     "source_type": "original",
-    "infores": "panther",
+    "infores": "test-kp-1",
     "exclude_tests": ["RSE"],
     "edges": [
         {

tests/onehop/test_triples/KP/Unit_Test_KP/Test_KP_2.json

@@ -1,7 +1,7 @@
 {
     "url": "https://automat.renci.org/ontological-hierarchy/1.2",
     "source_type": "original",
-    "infores": "ontological-hierarchy",
+    "infores": "test-kp-2",
     "exclude_tests": ["RPBS"],
     "edges": [
         {

tests/translator/registry/__init__.py

@@ -23,7 +23,7 @@
                 "version": "0.0.1",
                 "x-translator": {
                     "component": "KP",
-                    "infores": "infores:panther",
+                    "infores": "infores:test-kp-1",
                     "team": "Ranking Agent",
                     "biolink-version": "2.2.16"
                 },
@@ -40,7 +40,7 @@
                 "version": "0.0.1",
                 "x-translator": {
                     "component": "KP",
-                    "infores": "infores:ontological-hierarchy",
+                    "infores": "test-kp-2",
                     "team": "Ranking Agent",
                     "biolink-version": "2.2.16"
                 },

tests/translator/registry/test_translator_registry.py

@@ -179,13 +179,15 @@ def shared_test_extract_component_test_data_metadata_from_registry(
                             },
                             "x-trapi": {
                                 "version": "1.2.0",
-                                "test_data_location": "http://some-kp-web-test-data-directory"
+                                "test_data_location": "https://github.com/TranslatorSRI/SRI_testing/blob"
+                                                      "/main/tests/onehop/test_triples/KP/Unit_Test_KP/Test_KP_1.json"
                             }
                         }
                     }
                 ]
             },
-            "http://some-kp-web-test-data-directory"  # test_data_location
+            "https://raw.githubusercontent.com/TranslatorSRI/SRI_testing"
+            "/main/tests/onehop/test_triples/KP/Unit_Test_KP/Test_KP_1.json"   # KP test_data_location
         ),
         (   # Query 1 - Empty "hits" List
             {
@@ -299,13 +301,16 @@ def test_extract_kp_test_data_metadata_from_registry(query: Tuple[Dict, str, str
                                 },
                                 "x-trapi": {
                                     "version": "1.2.0",
-                                    "test_data_location": "http://some-ara-web-test-data-directory"
+                                    "test_data_location": "https://github.com/TranslatorSRI/SRI_testing/blob" +
+                                                          "/main/tests/onehop/test_triples/ARA" +
+                                                          "/Unit_Test_ARA/Test_ARA.json"
                                 }
                             }
                         }
                     ]
                 },
-                "http://some-ara-web-test-data-directory"   # test_data_location
+                "https://raw.githubusercontent.com/TranslatorSRI/SRI_testing" +
+                "/main/tests/onehop/test_triples/ARA/Unit_Test_ARA/Test_ARA.json"   # ARA test_data_location
         )
     ]
 )