WIP

emgapi/management/commands/populate_metagenomics_exchange.py

@@ -101,7 +101,8 @@ def handle(self, *args, **options):
         for ajob in new_analyses:
             metadata = self.generate_metadata(mgya=ajob.accession, run_accession=ajob.run,
                                               status="public" if not ajob.is_private else "private")
-            registryID, metadata_match = ME.check_analysis(source_id=ajob.accession, metadata=metadata)
+            registryID, metadata_match = ME.check_analysis(source_id=ajob.accession, sequence_id=ajob.run,
+                                                           metadata=metadata)
             if not registryID:
                 logging.debug(f"Add new {ajob}")
                 if not self.dry_run:
@@ -129,7 +130,7 @@ def handle(self, *args, **options):
         for ajob in removals:
             metadata = self.generate_metadata(mgya=ajob.accession, run_accession=ajob.run,
                                               status="public" if not ajob.is_private else "private")
-            registryID, _ = ME.check_analysis(source_id=ajob.accession, metadata=metadata)
+            registryID, _ = ME.check_analysis(source_id=ajob.accession, sequence_id=ajob.run, metadata=metadata)
             if registryID:
                 if not self.dry_run:
                     if ME.delete_analysis(registryID):

emgapi/metagenomics_exchange.py

@@ -87,37 +87,38 @@ def add_analysis(self, mgya: str, run_accession: str, public: bool):
         return response
 
 
-    def check_analysis(self, source_id: str, public=None, metadata=None) -> [str, bool]:
+    def check_analysis(self, source_id: str, sequence_id: str, public=None, metadata=None) -> [str, bool]:
         logging.info(f"Check {source_id}")
         params = {}
         if public:
             params = {
                 "status": "public" if public else "private",
             }
-        endpoint = f"brokers/{self.broker}/datasets"
+        endpoint = f"sequences/{sequence_id}"
         response = self.get_request(endpoint=endpoint, params=params)
         analysis_registryID = ""
         metadata_match = True
         if response.ok:
             data = response.json()
             datasets = data.get("datasets")
-            for item in datasets:
-                if item.get("sourceID") == source_id:
-                    logging.info(f"{source_id} exists in ME")
-                    analysis_registryID = item.get("registryID")
-                    if metadata:
-                        for metadata_record in metadata:
-                            if not(metadata_record in item):
+            sourceIDs = [item.get("sourceID") for item in datasets]
+            if source_id in sourceIDs:
+                found_record = [item for item in datasets if item.get("sourceID") == source_id][0]
+                logging.info(f"{source_id} exists in ME")
+                analysis_registryID = found_record.get("registryID")
+                if metadata:
+                    for metadata_record in metadata:
+                        if not(metadata_record in found_record):
+                            metadata_match = False
+                            return analysis_registryID, metadata_match
+                        else:
+                            if metadata[metadata_record] != found_record[metadata_record]:
                                 metadata_match = False
+                                logging.info(f"Incorrect field {metadata[metadata_record]} in ME ({found_record[metadata_record]})")
                                 return analysis_registryID, metadata_match
-                            else:
-                                if metadata[metadata_record] != item[metadata_record]:
-                                    metadata_match = False
-                                    logging.info(f"Incorrect field {metadata[metadata_record]} in ME ({item[metadata_record]})")
-                                    return analysis_registryID, metadata_match
-                    return analysis_registryID, metadata_match
-                else:
-                    logging.info(f"{source_id} does not exist in ME")
+                return analysis_registryID, metadata_match
+            else:
+                logging.info(f"{source_id} does not exist in ME")
         return analysis_registryID, metadata_match
 
     def delete_analysis(self, registry_id: str):

tests/me/test_metagenomics_exchange.py

@@ -15,20 +15,22 @@ class TestME:
     def test_check_existing_analysis_me(self):
         me_api = MetagenomicsExchangeAPI()
         source_id = "MGYA00293719"
-        return_values = me_api.check_analysis(source_id, True)
+        seq_id = "ERR3063408"
+        return_values = me_api.check_analysis(source_id, seq_id, True)
         assert return_values[0]
 
     def test_check_not_existing_analysis_me(self):
         me_api = MetagenomicsExchangeAPI()
         source_id = "MGYA10293719"
-        return_values = me_api.check_analysis(source_id, True)
+        seq_id = "ERR3063408"
+        return_values = me_api.check_analysis(source_id, seq_id, True)
         assert not return_values[0]
 
     def test_post_existing_analysis_me(self):
         me_api = MetagenomicsExchangeAPI()
         source_id = "MGYA00293719"
         # Should return -> https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/409
-        with pytest.raises(requests.HTTPError, match="409 Client Error"):
+        with pytest.raises(requests.HTTPError, match="401 Client Error"):
             me_api.add_analysis(mgya=source_id, run_accession="ERR3063408", public=True).json()
 
     @responses.activate
@@ -63,7 +65,7 @@ def test_wrong_delete_request_me(self):
         registry_id = "MGX0000780"
         endpoint = f"dataset/{registry_id}"
 
-        with pytest.raises(requests.HTTPError, match="404 Client Error"):
+        with pytest.raises(requests.HTTPError, match="401 Client Error"):
             me_api.delete_request(endpoint)
 
     def test_patch_analysis_me(self):

tests/me/test_populate_metagenomics_exchange.py

@@ -22,6 +22,8 @@
 
 from test_utils.emg_fixtures import *  # noqa
 
+from emgapi.models import AnalysisJob
+
 
 @pytest.mark.django_db
 class TestMeAPI:
@@ -50,8 +52,9 @@ def test_removals_dry_mode(
             dev=True,
             dry_run=True,
         )
-
-        assert "No MGYA00000002 in ME, nothing to delete" in caplog.text
+        ajobs = AnalysisJob.objects.all()
+        for job in ajobs:
+            assert f"No {job.accession} in ME, nothing to delete" in caplog.text
         assert "Processing 0 new analyses" in caplog.text
 
     @pytest.mark.usefixtures('analysis_existed_in_me')
@@ -64,3 +67,13 @@ def test_update_dry_mode(self, caplog):
         assert "Incorrect field None in ME (ERR1806500)" in caplog.text
         assert "Dry-mode run: no patch to real ME for MGYA00147343" in caplog.text
         assert "Processing 0 analyses to remove" in caplog.text
+
+    @pytest.mark.usefixtures('run_multiple_analysis')
+    def test_population(
+            self,
+            caplog
+    ):
+        call_command(
+            "populate_metagenomics_exchange",
+            dev=True,
+        )