feat: allow query + path in SparkSource for offline materialization

abhijeet-dhumal · cursoragent · abhijeet-dhumal · commit b428d7f8d68e · 2026-05-27T17:06:30.000+05:30
SparkSource previously required exactly one of table/query/path.
This relaxes the constraint to allow query + path together:
- query: used for reading raw data during materialization
- path: used for offline write-back (offline=True) and as
  pre-computed read source in get_historical_features

Co-authored-by: Cursor &lt;cursoragent@cursor.com&gt;
diff --git a/sdk/python/feast/infra/offline_stores/contrib/spark_offline_store/spark_source.py b/sdk/python/feast/infra/offline_stores/contrib/spark_offline_store/spark_source.py
@@ -287,11 +287,19 @@ def __init__(
         date_partition_column_format: Optional[str] = "%Y-%m-%d",
         table_format: Optional[TableFormat] = None,
     ):
-        # Check that only one of the ways to load a spark dataframe can be used. We have
-        # to treat empty string and null the same due to proto (de)serialization.
-        if sum([(not (not arg)) for arg in [table, query, path]]) != 1:
+        # query + path is allowed: query for reads during materialization,
+        # path for offline write-back (offline=True) and get_historical_features.
+        # table must be standalone (cannot combine with query or path).
+        has_table = bool(table)
+        has_query = bool(query)
+        has_path = bool(path)
+        if has_table and (has_query or has_path):
             raise ValueError(
-                "Exactly one of params(table, query, path) must be specified."
+                "'table' cannot be combined with 'query' or 'path'."
+            )
+        if not (has_table or has_query or has_path):
+            raise ValueError(
+                "At least one of params(table, query, path) must be specified."
             )
         if path:
             # If table_format is specified, file_format is optional (table format determines the reader)
