Upload config.py with huggingface_hub

config.py

@@ -72,8 +72,8 @@ def database_for(node_ids: list) -> list:
     return out
 
 
-def node_property_query(con, column: str = "species") -> str:
-    """SQL returning (id, <column>) for every node, from whichever source exists.
+def node_source(con) -> str:
+    """A SQL relation with the full node property columns, from whatever exists.
 
     The released dataset ships data/nodes_expected_ge5.parquet, which holds the
     property rows for exactly the nodes in edges_expected_ge5 — enough for every
@@ -82,15 +82,19 @@ def node_property_query(con, column: str = "species") -> str:
     across nodes_ML (MTBLS ids) and nodes_MW (ST ids).
     """
     if NODES_PARQUET.exists():
-        return f"SELECT id, {column} FROM '{NODES_PARQUET}'"
+        return f"(SELECT * FROM '{NODES_PARQUET}')"
     if pathlib.Path(NODE_DB_PATH).exists():
         con.execute(f"ATTACH IF NOT EXISTS '{NODE_DB_PATH}' AS g (READ_ONLY)")
-        return (f"SELECT id, {column} FROM g.nodes_ML "
-                f"UNION ALL SELECT id, {column} FROM g.nodes_MW")
+        return "(SELECT * FROM g.nodes_ML UNION ALL SELECT * FROM g.nodes_MW)"
     raise SystemExit(f"no node properties found: expected {NODES_PARQUET} "
                      f"or {NODE_DB_PATH}")
 
 
+def node_property_query(con, column: str = "species") -> str:
+    """SQL returning (id, <column>) for every node."""
+    return f"SELECT id, {column} FROM {node_source(con)}"
+
+
 def species_for(con, node_ids: list) -> list:
     """species per node id, in node_ids order. Raises if any id is unknown."""
     by_id = dict(con.execute(node_property_query(con, "species")).fetchall())