UniProt Parser Refactor – Improved Modularity, Identifier Extraction, and Test Coverage

pyproject.toml

@@ -144,6 +144,7 @@ ignore = [
 "tests/**/*.py" = ["S101", "T201"] # use of assert
 "**/__init__.py" = ["D104"]
 
+
 [tool.ruff.lint.mccabe]
 # Flag errors (`C901`) whenever the complexity level exceeds 15.
 max-complexity = 15

src/cdm_data_loader_utils/parsers/gene_association_file.py

@@ -273,7 +273,7 @@
             if register:
                 register_table(spark, output_path, table_name=table_name, permanent=permanent)
 
-    except Exception as e:
+    except Exception:
         logger.exception("Pipeline failed")
         sys.exit(1)
     finally:

src/cdm_data_loader_utils/parsers/shared_identifiers.py

@@ -0,0 +1,11 @@
+from cdm_data_loader_utils.parsers.xml_utils import get_text
+
+
+def parse_identifiers_generic(entry, xpath, prefix, ns):
+    result = []
+    for node in entry.findall(xpath, ns):
+        text = get_text(node)
+        if not text:
+            continue
+        result.append({"identifier": f"{prefix}:{text}", "source": prefix, "description": f"{prefix} accession"})
+    return result