fix: omit unfetchable list blobs from column discovery (#4312)

python/python/lancedb/table.py

@@ -2104,7 +2104,11 @@ class Table(ABC):
 
     @abstractmethod
     def blob_columns(self) -> list[str]:
-        """Names of the blob v2 columns declared on this table."""
+        """Names of blob v2 columns that can be fetched by row id.
+
+        Struct-nested blobs use dotted paths. Blobs inside lists are omitted
+        because a row id does not identify an individual list element.
+        """
 
     @abstractmethod
     def fetch_blobs(
@@ -2123,6 +2127,8 @@ class Table(ABC):
         One missing or deleted row ID rejects the entire batch; no partial
         results are returned.
 
+        Blobs inside lists cannot be fetched by row id yet.
+
         Remote servers limit each request to 1024 row IDs and 64 MiB of blob
         bytes. The client splits requests automatically and reads an individual
         larger blob through the Range route. This method still materializes all

python/python/tests/test_blob.py

@@ -879,12 +879,18 @@ def test_merge_insert_list_blob_dicts():
 
 def test_list_blob_column_queries_as_raw_descriptors():
     table = _list_blob_table("list_query")
+    assert table.blob_columns() == []
     hits = table.search().limit(10).to_arrow()
     element = hits.schema.field("images").type.value_type
     assert pa.types.is_struct(element)
     assert "_lance_row_id" not in element.names
-    with pytest.raises(ValueError, match="expected struct before segment"):
-        table.fetch_blobs("images.image", [0])
+    for column in ("images", "images.image"):
+        with pytest.raises(ValueError, match="blobs inside lists cannot be fetched"):
+            table.fetch_blobs(column, [0])
+        with pytest.raises(ValueError, match="blobs inside lists cannot be fetched"):
+            table.fetch_blob_files(column, [0])
+        with pytest.raises(ValueError, match="blobs inside lists cannot be fetched"):
+            table.fetch_blob_ranges(column, [(0, 0, 1)])
 
 
 def test_row_addressable_paths_exclude_list_children():