Fix bacdive enum YAML generation edge cases

Author: Copilot

metpo/tools/make_bacdive_utilization_enum.py

@@ -73,9 +73,10 @@ def convert_tsv_to_linkml(
         # Show preview if requested
         if preview:
             click.echo("\n=== Preview (first 30 lines) ===")
-            preview_lines = linkml_yaml.split("\n")[:30]
+            all_lines = linkml_yaml.split("\n")
+            preview_lines = all_lines[:30]
             click.echo("\n".join(preview_lines))
-            if len(linkml_yaml.split("\n")) > 30:
+            if len(all_lines) > 30:
                 click.echo("...")
 
     except Exception as e:
@@ -125,9 +126,8 @@ def convert_tsv_to_linkml_enum(
 
     # Process each relationship
     for _, row in object_properties.iterrows():
-        # Extract the ID number from the full URI
+        # Use the full URI as the permissible value meaning
         id_uri = row["ID"]
-        id_uri.split("/")[-1]
 
         # Create a code-friendly key from the label
         label = str(row["LABEL"])
@@ -148,9 +148,6 @@ def convert_tsv_to_linkml_enum(
         linkml_enum, default_flow_style=False, sort_keys=False, allow_unicode=True
     )
 
-    # Clean up the YAML formatting
-    yaml_output = yaml_output.replace("'", "")  # Remove unnecessary quotes
-
     # Write to file if specified
     if output_file:
         with Path(output_file).open("w") as f:

tests/test_make_bacdive_utilization_enum.py

@@ -0,0 +1,47 @@
+"""Tests for make_bacdive_utilization_enum utility."""
+
+import yaml
+
+from metpo.tools.make_bacdive_utilization_enum import convert_tsv_to_linkml_enum
+
+
+def _write_tsv(tmp_path, data_rows):
+    tsv_path = tmp_path / "bacdive.tsv"
+    content = [
+        "ID\tTYPE\tLABEL\tbacdive count",
+        "ID\tTYPE\tLABEL\tA",
+    ]
+    content.extend(data_rows)
+    tsv_path.write_text("\n".join(content) + "\n", encoding="utf-8")
+    return tsv_path
+
+
+def test_preserves_apostrophes_in_yaml_values(tmp_path):
+    tsv_path = _write_tsv(
+        tmp_path,
+        [
+            "https://example.org/relationship/1\towl:ObjectProperty\tdoesn't utilize\t1",
+        ],
+    )
+
+    yaml_output = convert_tsv_to_linkml_enum(str(tsv_path), enum_name="RelationshipTypeEnum")
+    parsed = yaml.safe_load(yaml_output)
+
+    permissible_values = parsed["enums"]["RelationshipTypeEnum"]["permissible_values"]
+    assert "doesn't_utilize" in permissible_values
+
+
+def test_permissible_value_meaning_uses_full_uri(tmp_path):
+    uri = "https://example.org/relationship/123"
+    tsv_path = _write_tsv(
+        tmp_path,
+        [
+            f"{uri}\towl:ObjectProperty\tutilizes carbon source\t1",
+        ],
+    )
+
+    yaml_output = convert_tsv_to_linkml_enum(str(tsv_path), enum_name="RelationshipTypeEnum")
+    parsed = yaml.safe_load(yaml_output)
+
+    entry = parsed["enums"]["RelationshipTypeEnum"]["permissible_values"]["utilizes_carbon_source"]
+    assert entry["meaning"] == uri