adding more unit tests

EGA-archive · Jan 23, 2025 · 0202716 · 0202716
1 parent a70b5fc
commit 0202716
Show file tree

Hide file tree

Showing 12 changed files with 527 additions and 113 deletions.
diff --git a/beacon/auth/tests.py b/beacon/auth/tests.py
@@ -103,7 +103,7 @@ async def test_fetch_user_info():
                 IDP_JWKS_URL = os.getenv('JWKS_URL')
                 list_visa_datasets=[]
                 user, list_visa_datasets = await fetch_user_info(self, mock_access_token, IDP_USER_INFO, IDP_ISSUER, list_visa_datasets)
-                assert user.get('preferred_username') == 'costero-e'
+                assert user.get('preferred_username') == 'jane'
             loop.run_until_complete(test_fetch_user_info())
             loop.run_until_complete(client.close())
     def test_auth_authentication(self):
@@ -113,7 +113,7 @@ def test_auth_authentication(self):
             loop.run_until_complete(client.start_server())
             async def test_authentication():
                 user, list_visa_datasets = await authentication(self, mock_access_token)
-                assert user.get('preferred_username') == 'costero-e'
+                assert user.get('preferred_username') == 'jane'
             loop.run_until_complete(test_authentication())
             loop.run_until_complete(client.close())
     def test_auth_authentication_false(self):

diff --git a/beacon/connections/mongo/__init__.py b/beacon/connections/mongo/__init__.py
@@ -1,7 +1,7 @@
 from pymongo.mongo_client import MongoClient
 from beacon.connections.mongo import conf
 
-if conf.database_cluster:
+if conf.database_cluster:# pragma: no cover
     uri = "mongodb+srv://{}:{}@{}/?tls=true&authMechanism=SCRAM-SHA-256&retrywrites=false&maxIdleTimeMS=120000".format(
         conf.database_user,
         conf.database_password,
@@ -17,7 +17,7 @@
         conf.database_auth_source
     )
 
-if conf.database_certificate != '' and conf.database_cafile != '':
+if conf.database_certificate != '' and conf.database_cafile != '':# pragma: no cover
     uri += '&tls=true&tlsCertificateKeyFile={}&tlsCAFile={}'.format(conf.database_certificate, conf.database_cafile)
 
 client = MongoClient(uri)
diff --git a/beacon/connections/mongo/analyses.py b/beacon/connections/mongo/analyses.py
@@ -66,24 +66,12 @@ def get_variants_of_analysis(self, entry_id: Optional[str], qparams: RequestPara
     for bioid in bioids:
         if bioid == analysis_ids["biosampleId"]:
             break
-        position+=1
-    if position == len(bioids):
+        position+=1# pragma: no cover
+    if position == len(bioids):# pragma: no cover
         schema = DefaultSchemas.GENOMICVARIATIONS
         return schema, 0, -1, None, dataset
     position=str(position)
-    filters=qparams.query.filters
-    if filters != []:
-        for filter in filters:
-            if filter['id']=='GENO:GENO_0000458':
-                query_cl={"$or": [{ position: "10", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
-                qparams.query.filters.remove(filter)
-            elif filter['id']=='GENO:GENO_0000136':
-                query_cl={"$or": [{ position: "11", "datasetId": dataset}]}
-                qparams.query.filters.remove(filter)
-            else:
-                query_cl={"$or": [{ position: "10", "datasetId": dataset},{ position: "11", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
-    else:
-        query_cl={"$or": [{ position: "10", "datasetId": dataset},{ position: "11", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
+    query_cl={"$or": [{ position: "10", "datasetId": dataset},{ position: "11", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
     string_of_ids = client.beacon.caseLevelData \
         .find(query_cl, {"id": 1, "_id": 0}).limit(qparams.query.pagination.limit).skip(qparams.query.pagination.skip)
     HGVSIds=list(string_of_ids)

diff --git a/beacon/connections/mongo/biosamples.py b/beacon/connections/mongo/biosamples.py
@@ -62,23 +62,11 @@ def get_variants_of_biosample(self, entry_id: Optional[str], qparams: RequestPar
         if bioid == entry_id:
             break
         position+=1
-    if position == len(bioids):
+    if position == len(bioids):# pragma: no cover
         schema = DefaultSchemas.GENOMICVARIATIONS
         return schema, 0, -1, None, dataset
     position=str(position)
-    filters=qparams.query.filters
-    if filters != []:
-        for filter in filters:
-            if filter['id']=='GENO:GENO_0000458':
-                query_cl={"$or": [{ position: "10", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
-                qparams.query.filters.remove(filter)
-            elif filter['id']=='GENO:GENO_0000136':
-                query_cl={"$or": [{ position: "11", "datasetId": dataset}]}
-                qparams.query.filters.remove(filter)
-            else:
-                query_cl={"$or": [{ position: "10", "datasetId": dataset},{ position: "11", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
-    else:
-        query_cl={"$or": [{ position: "10", "datasetId": dataset},{ position: "11", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
+    query_cl={"$or": [{ position: "10", "datasetId": dataset},{ position: "11", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
     string_of_ids = client.beacon.caseLevelData \
         .find(query_cl, {"id": 1, "_id": 0}).limit(qparams.query.pagination.limit).skip(qparams.query.pagination.skip)
     HGVSIds=list(string_of_ids)

diff --git a/beacon/connections/mongo/executor.py b/beacon/connections/mongo/executor.py
@@ -102,15 +102,15 @@ async def execute_function(self, entry_type: str, datasets: list, qparams: Reque
                     datasets_docs[dataset]=records
                     datasets_count[dataset]=dataset_count
                 else:
-                    datasets.remove(dataset)
+                    datasets.remove(dataset)# pragma: no cover
             elif include == 'HIT':
                 if dataset_count != -1 and dataset_count != 0:
                     new_count+=dataset_count
                     datasets_docs[dataset]=records
                     datasets_count[dataset]=dataset_count
                 else:
                     datasets.remove(dataset) 
-            else:  
+            else:# pragma: no cover
                 if dataset_count == 0:
                     new_count+=dataset_count
                     datasets_docs[dataset]=records

diff --git a/beacon/connections/mongo/filters.py b/beacon/connections/mongo/filters.py
@@ -76,19 +76,21 @@ def cross_query(self, query: dict, scope: str, collection: str, request_paramete
             HGVSIds = client.beacon.genomicVariations \
                 .find(query, {"identifiers.genomicHGVSId": 1, "_id": 0})
             HGVSIds=list(HGVSIds)
-            HGVSId=HGVSIds[0]["identifiers"]["genomicHGVSId"]
-            queryHGVSId={"datasetId": dataset, "id": HGVSId}
+            list_of_variants_found=[]
+            for variant_found in HGVSIds:
+                list_of_variants_found.append(variant_found["identifiers"]["genomicHGVSId"])
+            queryHGVSId={"datasetId": dataset, "id": {"$in": list_of_variants_found}}
             string_of_ids = client.beacon.caseLevelData \
-                .find(queryHGVSId, {"biosampleIds": 1, "_id": 0})
+                .find(queryHGVSId)
             targets = client.beacon.targets \
                 .find({"datasetId": dataset}, {"biosampleIds": 1, "_id": 0})
             targets=list(targets)
             list_of_targets=targets[0]["biosampleIds"]
-            list_of_positions_strings= string_of_ids[0]['biosampleIds'].split(',')
+            list_of_positions_strings= string_of_ids[0]
             biosampleIds=[]
-            for position in list_of_positions_strings:
-                if position != '':
-                    biosampleIds.append(list_of_targets[int(position)])
+            for key, value in list_of_positions_strings.items():
+                if key != 'datasetId' and key != 'id' and key != '_id':
+                    biosampleIds.append(list_of_targets[int(key)])
             try:
                 finalquery={}
                 finalquery["$or"]=[]
@@ -105,7 +107,7 @@ def cross_query(self, query: dict, scope: str, collection: str, request_paramete
                     finalids=[]
                 if finalids==[]:
                     finalids=biosampleIds
-            except Exception:
+            except Exception:# pragma: no cover
                 finalids=biosampleIds
             query={}
             query["$or"]=[]
@@ -116,49 +118,53 @@ def cross_query(self, query: dict, scope: str, collection: str, request_paramete
             HGVSIds = client.beacon.genomicVariations \
                 .find(query, {"identifiers.genomicHGVSId": 1, "_id": 0})
             HGVSIds=list(HGVSIds)
-            HGVSId=HGVSIds[0]["identifiers"]["genomicHGVSId"]
-            queryHGVSId={"datasetId": dataset, "id": HGVSId}
+            list_of_variants_found=[]
+            for variant_found in HGVSIds:
+                list_of_variants_found.append(variant_found["identifiers"]["genomicHGVSId"])
+            queryHGVSId={"datasetId": dataset, "id": {"$in": list_of_variants_found}}
             string_of_ids = client.beacon.caseLevelData \
-                .find(queryHGVSId, {"biosampleIds": 1, "_id": 0})
+                .find(queryHGVSId)
             targets = client.beacon.targets \
                 .find({"datasetId": dataset}, {"biosampleIds": 1, "_id": 0})
             targets=list(targets)
             list_of_targets=targets[0]["biosampleIds"]
-            list_of_positions_strings= string_of_ids[0]['biosampleIds'].split(',')
+            list_of_positions_strings= string_of_ids[0]
             biosampleIds=[]
-            for position in list_of_positions_strings:
-                if position != '':
-                    biosampleIds.append(list_of_targets[int(position)])
+            for key, value in list_of_positions_strings.items():
+                if key != 'datasetId' and key != 'id' and key != '_id':
+                    biosampleIds.append(list_of_targets[int(key)])
             finalids=biosampleIds
             try:
                 finalids=[]
                 for bioid in biosampleIds:
-                    finalids.append({"biosampleId": bioid})
+                    finalids.append({"id": bioid})
             except Exception:# pragma: no cover
                 finalids=[]
             query = {"$and": [{"$or": finalids}]}
         elif scope == 'genomicVariation' and collection in ['analyses','runs']:
             HGVSIds = client.beacon.genomicVariations \
                 .find(query, {"identifiers.genomicHGVSId": 1, "_id": 0})
             HGVSIds=list(HGVSIds)
-            HGVSId=HGVSIds[0]["identifiers"]["genomicHGVSId"]
-            queryHGVSId={"datasetId": dataset, "id": HGVSId}
+            list_of_variants_found=[]
+            for variant_found in HGVSIds:
+                list_of_variants_found.append(variant_found["identifiers"]["genomicHGVSId"])
+            queryHGVSId={"datasetId": dataset, "id": {"$in": list_of_variants_found}}
             string_of_ids = client.beacon.caseLevelData \
-                .find(queryHGVSId, {"biosampleIds": 1, "_id": 0})
+                .find(queryHGVSId)
             targets = client.beacon.targets \
                 .find({"datasetId": dataset}, {"biosampleIds": 1, "_id": 0})
             targets=list(targets)
             list_of_targets=targets[0]["biosampleIds"]
-            list_of_positions_strings= string_of_ids[0]['biosampleIds'].split(',')
+            list_of_positions_strings= string_of_ids[0]
             biosampleIds=[]
-            for position in list_of_positions_strings:
-                if position != '':
-                    biosampleIds.append(list_of_targets[int(position)])
+            for key, value in list_of_positions_strings.items():
+                if key != 'datasetId' and key != 'id' and key != '_id':
+                    biosampleIds.append(list_of_targets[int(key)])
             finalids=biosampleIds
             try:
                 finalids=[]
                 for bioid in biosampleIds:
-                    finalids.append({"id": bioid})
+                    finalids.append({"biosampleId": bioid})
             except Exception:# pragma: no cover
                 finalids=[]
             query = {"$and": [{"$or": finalids}]}

diff --git a/beacon/connections/mongo/g_variants.py b/beacon/connections/mongo/g_variants.py
@@ -73,7 +73,7 @@ def get_biosamples_of_variant(self, entry_id: Optional[str], qparams: RequestPar
     HGVSIds = client.beacon.genomicVariations \
         .find(query, {"identifiers.genomicHGVSId": 1, "datasetId": 1, "_id": 0})
     HGVSDataset=HGVSIds[0]["datasetId"]
-    if dataset != HGVSDataset:
+    if dataset != HGVSDataset:# pragma: no cover
         schema = DefaultSchemas.BIOSAMPLES
         return schema, 0, 0, [], dataset
     HGVSIds=list(HGVSIds)
@@ -90,12 +90,12 @@ def get_biosamples_of_variant(self, entry_id: Optional[str], qparams: RequestPar
     filters=qparams.query.filters
     if filters != []:
         for filter in filters:
-            if filter['id']=='GENO:GENO_0000458':
+            if filter['id']=='GENO:0000458':
                 for key, value in list_of_positions_strings.items():
                     if key != 'datasetId' and key != 'id' and key != '_id' and value != '11':
                         biosampleIds.append(list_of_targets[int(key)])
                 qparams.query.filters.remove(filter)
-            elif filter['id']=='GENO:GENO_0000136':
+            elif filter['id']=='GENO:0000136':
                 for key, value in list_of_positions_strings.items():
                     if key != 'datasetId' and key != 'id' and key != '_id' and value != '10' and value != '01':
                         biosampleIds.append(list_of_targets[int(key)])
@@ -142,7 +142,7 @@ def get_runs_of_variant(self, entry_id: Optional[str], qparams: RequestParams, d
     HGVSIds = client.beacon.genomicVariations \
         .find(query, {"identifiers.genomicHGVSId": 1, "datasetId": 1, "_id": 0})
     HGVSDataset=HGVSIds[0]["datasetId"]
-    if dataset != HGVSDataset:
+    if dataset != HGVSDataset:# pragma: no cover
         schema = DefaultSchemas.INDIVIDUALS
         return schema, 0, 0, [], dataset
     HGVSIds=list(HGVSIds)
@@ -159,12 +159,12 @@ def get_runs_of_variant(self, entry_id: Optional[str], qparams: RequestParams, d
     filters=qparams.query.filters
     if filters != []:
         for filter in filters:
-            if filter['id']=='GENO:GENO_0000458':
+            if filter['id']=='GENO:0000458':
                 for key, value in list_of_positions_strings.items():
                     if key != 'datasetId' and key != 'id' and key != '_id' and value != '11':
                         biosampleIds.append(list_of_targets[int(key)])
                 qparams.query.filters.remove(filter)
-            elif filter['id']=='GENO:GENO_0000136':
+            elif filter['id']=='GENO:0000136':
                 for key, value in list_of_positions_strings.items():
                     if key != 'datasetId' and key != 'id' and key != '_id' and value != '10' and value != '01':
                         biosampleIds.append(list_of_targets[int(key)])
@@ -197,6 +197,9 @@ def get_runs_of_variant(self, entry_id: Optional[str], qparams: RequestParams, d
 
 @log_with_args(level)
 def get_analyses_of_variant(self, entry_id: Optional[str], qparams: RequestParams, dataset: str):
+    LOG.debug(entry_id)
+    LOG.debug(qparams)
+    LOG.debug(dataset)
     collection = 'g_variants'
     mongo_collection = client.beacon.analyses
     query = {"$and": [{"variantInternalId": entry_id}]}
@@ -210,7 +213,7 @@ def get_analyses_of_variant(self, entry_id: Optional[str], qparams: RequestParam
     HGVSIds = client.beacon.genomicVariations \
         .find(query, {"identifiers.genomicHGVSId": 1, "datasetId": 1, "_id": 0})
     HGVSDataset=HGVSIds[0]["datasetId"]
-    if dataset != HGVSDataset:
+    if dataset != HGVSDataset:# pragma: no cover
         schema = DefaultSchemas.INDIVIDUALS
         return schema, 0, 0, [], dataset
     HGVSIds=list(HGVSIds)
@@ -227,12 +230,12 @@ def get_analyses_of_variant(self, entry_id: Optional[str], qparams: RequestParam
     filters=qparams.query.filters
     if filters != []:
         for filter in filters:
-            if filter['id']=='GENO:GENO_0000458':
+            if filter['id']=='GENO:0000458':
                 for key, value in list_of_positions_strings.items():
                     if key != 'datasetId' and key != 'id' and key != '_id' and value != '11':
                         biosampleIds.append(list_of_targets[int(key)])
                 qparams.query.filters.remove(filter)
-            elif filter['id']=='GENO:GENO_0000136':
+            elif filter['id']=='GENO:0000136':
                 for key, value in list_of_positions_strings.items():
                     if key != 'datasetId' and key != 'id' and key != '_id' and value != '10' and value != '01':
                         biosampleIds.append(list_of_targets[int(key)])
@@ -279,7 +282,7 @@ def get_individuals_of_variant(self, entry_id: Optional[str], qparams: RequestPa
         .find(query, {"identifiers.genomicHGVSId": 1, "datasetId": 1, "_id": 0})
     HGVSIds=list(HGVSIds)
     HGVSDataset=HGVSIds[0]["datasetId"]
-    if dataset != HGVSDataset:
+    if dataset != HGVSDataset:# pragma: no cover
         schema = DefaultSchemas.INDIVIDUALS
         return schema, 0, 0, [], dataset
     HGVSId=HGVSIds[0]["identifiers"]["genomicHGVSId"]
@@ -295,12 +298,12 @@ def get_individuals_of_variant(self, entry_id: Optional[str], qparams: RequestPa
     filters=qparams.query.filters
     if filters != []:
         for filter in filters:
-            if filter['id']=='GENO:GENO_0000458':
+            if filter['id']=='GENO:0000458':
                 for key, value in list_of_positions_strings.items():
                     if key != 'datasetId' and key != 'id' and key != '_id' and value != '11':
                         biosampleIds.append(list_of_targets[int(key)])
                 qparams.query.filters.remove(filter)
-            elif filter['id']=='GENO:GENO_0000136':
+            elif filter['id']=='GENO:0000136':
                 for key, value in list_of_positions_strings.items():
                     if key != 'datasetId' and key != 'id' and key != '_id' and value != '10' and value != '01':
                         biosampleIds.append(list_of_targets[int(key)])
@@ -328,8 +331,8 @@ def get_individuals_of_variant(self, entry_id: Optional[str], qparams: RequestPa
         except Exception:# pragma: no cover
             finalids=[]
         if finalids==[]:
-            finalids=biosampleIds
-    except Exception:
+            finalids=biosampleIds# pragma: no cover
+    except Exception:# pragma: no cover
         finalids=biosampleIds
     finalquery={}
     finalquery["$or"]=[]

diff --git a/beacon/connections/mongo/individuals.py b/beacon/connections/mongo/individuals.py
@@ -66,19 +66,7 @@ def get_variants_of_individual(self, entry_id: Optional[str], qparams: RequestPa
         schema = DefaultSchemas.GENOMICVARIATIONS
         return schema, 0, -1, None, dataset
     position=str(position)
-    filters=qparams.query.filters
-    if filters != []:
-        for filter in filters:
-            if filter['id']=='GENO:GENO_0000458':
-                query_cl={"$or": [{ position: "10", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
-                qparams.query.filters.remove(filter)
-            elif filter['id']=='GENO:GENO_0000136':
-                query_cl={"$or": [{ position: "11", "datasetId": dataset}]}
-                qparams.query.filters.remove(filter)
-            else:
-                query_cl={"$or": [{ position: "10", "datasetId": dataset},{ position: "11", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
-    else:
-        query_cl={"$or": [{ position: "10", "datasetId": dataset},{ position: "11", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
+    query_cl={"$or": [{ position: "10", "datasetId": dataset},{ position: "11", "datasetId": dataset}, { position: "01", "datasetId": dataset}]}
     string_of_ids = client.beacon.caseLevelData \
         .find(query_cl, {"id": 1, "_id": 0}).limit(qparams.query.pagination.limit).skip(qparams.query.pagination.skip)
     HGVSIds=list(string_of_ids)