remove md5sum, refactor name to bedfile_name and bedset_name

Author: donaldcampbelljr

bbconf/bbconf.py

@@ -41,7 +41,8 @@
 )
 from bbconf.exceptions import MissingConfigDataError, BedBaseConfError
 from bbconf.helpers import raise_missing_key, get_bedbase_cfg
-os.environ["TF_CPP_MIN_LOG_LEVEL"] = "3" # to suppress verbose warnings tensorflow
+
+os.environ["TF_CPP_MIN_LOG_LEVEL"] = "3"  # to suppress verbose warnings tensorflow
 from geniml.text2bednn import text2bednn
 from geniml.search import QdrantBackend
 from sentence_transformers import SentenceTransformer

bbconf/schemas/bedfiles_schema.yaml

@@ -1,11 +1,8 @@
 pipeline_name: bedfile
 samples:
-  name:
+  bedfile_name:
     type: string
     description: BED file name
-  md5sum:
-    type: string
-    description: BED file checksum
   genome:
     type: object
     description: genome assembly of the BED files

bbconf/schemas/bedsets_schema.yaml

@@ -1,11 +1,8 @@
 pipeline_name: bedsets
 samples:
-  name:
+  bedset_name:
     type: string
     description: BED set name
-  md5sum:
-    type: string
-    description: BED set checksum
   genome:
     type: object
     description: genome assembly of the BED sets

tests/conftest.py

@@ -7,13 +7,13 @@
 @pytest.fixture
 def test_data_bed():
     s = "test_string"
-    return {"name": s, "md5sum": s, "bedfile": {"path": s, "title": s}, "regions_no": 1}
+    return {"bedfile_name": s, "bedfile": {"path": s, "title": s}, "regions_no": 1}
 
 
 @pytest.fixture
 def test_data_bedset():
     s = "test_string"
-    return {"name": s, "md5sum": s, "bedset_tar_archive_path": {"path": s, "title": s}}
+    return {"bedset_name": s, "bedset_tar_archive_path": {"path": s, "title": s}}
 
 
 @pytest.fixture

tests/test_bbconf.py

@@ -65,7 +65,9 @@ def test_nonunique_digest_insert_error(
             bbc.bed.report(record_identifier="bed1", values=test_data_bed)
             assert not bbc.bed.report(record_identifier="bed1", values=test_data_bed)
             bbc.bedset.report(record_identifier="bedset1", values=test_data_bedset)
-            assert not bbc.bedset.report(record_identifier="bedset1", values=test_data_bedset)
+            assert not bbc.bedset.report(
+                record_identifier="bedset1", values=test_data_bedset
+            )
 
     def test_reporting_relationships(self, cfg_pth, test_data_bed, test_data_bedset):
         with ContextManagerDBTesting(DB_URL):
@@ -139,4 +141,4 @@ def test_config_variables_are_set(self, cfg_pth, test_data_bed, test_data_bedset
             print(bbc.config["qdrant"]["host"])
             assert bbc.config["qdrant"]["host"] == "test_localhost"
             assert bbc.config["path"]["region2vec"] is not None
-            assert bbc.config["database"]["host"] == "localhost"
+            assert bbc.config["database"]["host"] in ["localhost", "127.0.0.1"]