Guard against duplicate builder_kwargs/config_kwargs in load_dataset_…

src/datasets/load.py

@@ -1047,6 +1047,14 @@ def load_dataset_builder(
     storage_options: Optional[dict] = None,
     **config_kwargs,
 ) -> DatasetBuilder:
+    # Error if builder_kwargs and config_kwargs share any keys
+    if "builder_kwargs" in config_kwargs and "config_kwargs" in config_kwargs:
+        bk = config_kwargs["builder_kwargs"]
+        ck = config_kwargs["config_kwargs"]
+        overlap = set(bk) & set(ck)
+        if overlap:
+            raise TypeError(f"Duplicate keys in builder_kwargs and config_kwargs: {overlap}")
+
     """Load a dataset builder which can be used to:
 
     - Inspect general information that is required to build a dataset (cache directory, config, dataset info, features, data files, etc.)

tests/test_load_duplicate_keys.py

@@ -0,0 +1,16 @@
+import pytest
+from datasets.load import load_dataset_builder
+
+def test_duplicate_builder_and_config_kwargs():
+    builder_kwargs = {"foo-key": 1}
+    config_kwargs  = {"foo-key": 2}
+    with pytest.raises(TypeError) as excinfo:
+        load_dataset_builder(
+            "csv",
+            builder_kwargs=builder_kwargs,
+            config_kwargs=config_kwargs,
+        )
+    msg = str(excinfo.value).lower()
+    assert "duplicate keys" in msg
+    assert "builder_kwargs" in msg
+    assert "config_kwargs" in msg