fix(language): keep base install green — drop processor re-export, gate dataset-extra tests

`lerobot.processor` re-exported `RenderMessagesStep` at the package
level, so importing anything from `lerobot.processor` pulled in
`lerobot.datasets.language` → `lerobot.datasets/__init__.py` →
`require_package("datasets")`, which fails in the Tier 1 base install
that intentionally omits the `[dataset]` extra. The chain bricked
collection for unrelated suites (`tests/policies/pi0_pi05/...`,
`tests/envs/...`, etc.).

* Stop re-exporting `RenderMessagesStep` from `lerobot.processor`. The
  only consumer (the test) already imports from the submodule.
  Document the deliberate omission in the module docstring.
* Add `pytest.importorskip("datasets", ...)` (and `pandas` where
  needed) at the top of the four PR-added tests that exercise the
  language stack:
  - tests/datasets/test_language.py
  - tests/datasets/test_language_render.py
  - tests/processor/test_render_messages_processor.py
  - tests/utils/test_collate.py

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/lerobot/processor/__init__.py

@@ -93,9 +93,15 @@ from .relative_action_processor import (
     to_relative_actions,
 )
 from .rename_processor import RenameObservationsProcessorStep, rename_stats
-from .render_messages_processor import RenderMessagesStep
 from .tokenizer_processor import ActionTokenizerProcessorStep, TokenizerProcessorStep
 
+# RenderMessagesStep is intentionally NOT re-exported here: it pulls in
+# `lerobot.datasets.language`, which requires the `[dataset]` extra
+# (`datasets`, `pyarrow`). Importing it from the processor package would
+# break every base-install consumer of `lerobot.processor`. Users that
+# need it import directly:
+#   from lerobot.processor.render_messages_processor import RenderMessagesStep
+
 __all__ = [
     "ActionProcessorStep",
     "AddTeleopActionAsComplimentaryDataStep",
@@ -129,7 +135,6 @@ __all__ = [
     "make_default_robot_observation_processor",
     "AbsoluteActionsProcessorStep",
     "RelativeActionsProcessorStep",
-    "RenderMessagesStep",
     "MapDeltaActionToRobotActionStep",
     "MapTensorToDeltaActionDictStep",
     "NewLineTaskProcessorStep",

tests/datasets/test_language.py

@@ -1,13 +1,17 @@
 #!/usr/bin/env python
 
-import numpy as np
-import pandas as pd
-import pyarrow as pa
 import pytest
 
-from lerobot.datasets import LeRobotDataset
+pytest.importorskip("datasets", reason="datasets is required (install lerobot[dataset])")
-from lerobot.datasets.io_utils import write_info
+pytest.importorskip("pandas", reason="pandas is required (install lerobot[dataset])")
-from lerobot.datasets.language import (
+
+import numpy as np  # noqa: E402
+import pandas as pd  # noqa: E402
+import pyarrow as pa  # noqa: E402
+
+from lerobot.datasets import LeRobotDataset  # noqa: E402
+from lerobot.datasets.io_utils import write_info  # noqa: E402
+from lerobot.datasets.language import (  # noqa: E402
     EVENT_ONLY_STYLES,
     LANGUAGE_EVENTS,
     LANGUAGE_PERSISTENT,
@@ -21,7 +25,7 @@ from lerobot.datasets.language import (
     language_persistent_arrow_type,
     validate_camera_field,
 )
-from lerobot.datasets.utils import DEFAULT_DATA_PATH
+from lerobot.datasets.utils import DEFAULT_DATA_PATH  # noqa: E402
 
 
 def test_language_arrow_schema_has_expected_fields():

tests/datasets/test_language_render.py

@@ -2,8 +2,16 @@
 
 import pytest
 
-from lerobot.configs.recipe import MessageTurn, TrainingRecipe
+pytest.importorskip("datasets", reason="datasets is required (install lerobot[dataset])")
-from lerobot.datasets.language_render import active_at, emitted_at, nth_next, nth_prev, render_sample
+
+from lerobot.configs.recipe import MessageTurn, TrainingRecipe  # noqa: E402
+from lerobot.datasets.language_render import (  # noqa: E402
+    active_at,
+    emitted_at,
+    nth_next,
+    nth_prev,
+    render_sample,
+)
 
 
 def persistent_row(role, content, style, timestamp, tool_calls=None, camera=None):

tests/processor/test_render_messages_processor.py

@@ -1,11 +1,15 @@
 #!/usr/bin/env python
 
-import torch
+import pytest
 
-from lerobot.configs.recipe import MessageTurn, TrainingRecipe
+pytest.importorskip("datasets", reason="datasets is required (install lerobot[dataset])")
-from lerobot.processor.converters import create_transition
+
-from lerobot.processor.render_messages_processor import RenderMessagesStep
+import torch  # noqa: E402
-from lerobot.types import TransitionKey
+
+from lerobot.configs.recipe import MessageTurn, TrainingRecipe  # noqa: E402
+from lerobot.processor.converters import create_transition  # noqa: E402
+from lerobot.processor.render_messages_processor import RenderMessagesStep  # noqa: E402
+from lerobot.types import TransitionKey  # noqa: E402
 
 
 def test_render_messages_step_noops_without_language_columns():

tests/utils/test_collate.py

@@ -1,8 +1,12 @@
 #!/usr/bin/env python
 
-import torch
+import pytest
 
-from lerobot.utils.collate import lerobot_collate_fn
+pytest.importorskip("datasets", reason="datasets is required (install lerobot[dataset])")
+
+import torch  # noqa: E402
+
+from lerobot.utils.collate import lerobot_collate_fn  # noqa: E402
 
 
 def test_lerobot_collate_preserves_messages_and_drops_raw_language():