Add context manager protocol to LookupIndex and cleanup in transform_spec

LookupIndex now supports `with` statement usage via __enter__/__exit__.
transform_spec() closes the LookupIndex it creates after iteration completes,
preventing DuckDB connection leaks.

Closes #143

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: amc-corey-cox

src/linkml_map/transformer/engine.py

@@ -41,41 +41,46 @@ def transform_spec(
     if spec is None:
         return
 
-    if transformer.lookup_index is None:
+    owns_index = transformer.lookup_index is None
+    if owns_index:
         transformer.lookup_index = LookupIndex()
 
-    for class_deriv in spec.class_derivations:
-        table_name = class_deriv.populated_from or class_deriv.name
-        if table_name not in data_loader:
-            logger.debug("Skipping class_derivation %s: no data found", class_deriv.name)
-            continue
+    try:
+        for class_deriv in spec.class_derivations:
+            table_name = class_deriv.populated_from or class_deriv.name
+            if table_name not in data_loader:
+                logger.debug("Skipping class_derivation %s: no data found", class_deriv.name)
+                continue
 
-        joined_tables: list[str] = []
-        try:
-            # Register secondary (joined) tables
-            if class_deriv.joins:
-                for join_name, join_spec in class_deriv.joins.items():
-                    lookup_key = join_spec.lookup_key or join_spec.join_on
-                    source_key = join_spec.source_key or join_spec.join_on
-                    if not lookup_key or not source_key:
-                        msg = (
-                            f"Join {join_name!r} must specify 'join_on' or both "
-                            f"'source_key' and 'lookup_key'"
+            joined_tables: list[str] = []
+            try:
+                # Register secondary (joined) tables
+                if class_deriv.joins:
+                    for join_name, join_spec in class_deriv.joins.items():
+                        lookup_key = join_spec.lookup_key or join_spec.join_on
+                        source_key = join_spec.source_key or join_spec.join_on
+                        if not lookup_key or not source_key:
+                            msg = (
+                                f"Join {join_name!r} must specify 'join_on' or both "
+                                f"'source_key' and 'lookup_key'"
+                            )
+                            raise ValueError(msg)
+                        join_path = data_loader.get_path(join_name)
+                        transformer.lookup_index.register_table(
+                            join_name, join_path, lookup_key
                         )
-                        raise ValueError(msg)
-                    join_path = data_loader.get_path(join_name)
-                    transformer.lookup_index.register_table(
-                        join_name, join_path, lookup_key
-                    )
-                    joined_tables.append(join_name)
+                        joined_tables.append(join_name)
 
-            # Stream primary table rows
-            for row in data_loader[table_name]:
-                yield transformer.map_object(
-                    row,
-                    source_type=source_type or table_name,
-                    class_derivation=class_deriv,
-                )
-        finally:
-            for jt in joined_tables:
-                transformer.lookup_index.drop(jt)
+                # Stream primary table rows
+                for row in data_loader[table_name]:
+                    yield transformer.map_object(
+                        row,
+                        source_type=source_type or table_name,
+                        class_derivation=class_deriv,
+                    )
+            finally:
+                for jt in joined_tables:
+                    transformer.lookup_index.drop(jt)
+    finally:
+        if owns_index:
+            transformer.lookup_index.close()

src/linkml_map/utils/lookup_index.py

@@ -84,6 +84,14 @@ def is_registered(self, table: str) -> bool:
         """Check whether *table* has been registered."""
         return table in self._tables
 
+    def __enter__(self) -> LookupIndex:
+        """Enter the context manager."""
+        return self
+
+    def __exit__(self, *exc_info: object) -> None:
+        """Exit the context manager, closing the DuckDB connection."""
+        self.close()
+
     def close(self) -> None:
         """Close the DuckDB connection."""
         self._conn.close()