[Data] Create unit test directory for fast, isolated tests

python/ray/data/BUILD.bazel

@@ -27,6 +27,19 @@ py_library(
     deps = ["//python/ray/tests:conftest"],
 )
 
+py_test_module_list(
+    size = "small",
+    files = glob(["tests/unit/**/test_*.py"]),
+    tags = [
+        "exclusive",
+        "team:data",
+    ],
+    deps = [
+        ":conftest",
+        "//:ray_lib",
+    ],
+)
+
 py_test_module_list(
     size = "medium",
     files = glob(["tests/block_batching/test_*.py"]),

python/ray/data/tests/unit/README.md

@@ -0,0 +1,26 @@
+# Unit Tests
+
+This directory contains unit tests that do not depend on distributed infrastructure or external dependencies.
+
+## Requirements
+
+Unit tests in this directory must be:
+- **Fast**: Execute in milliseconds, not seconds
+- **Isolated**: No dependencies on Ray runtime, external services, or file I/O
+- **Deterministic**: No randomness or time-based behavior
+
+## Restrictions
+
+Tests should NOT:
+- Initialize or use the Ray distributed runtime
+- Use `time.sleep()` or other time-based delays
+- Depend on external services, databases, or file systems
+- Make network calls
+
+## Enforcement
+
+The `conftest.py` in this directory enforces these restrictions by preventing:
+- `ray.init()` from being called
+- `time.sleep()` from being used
+
+If a test requires any of these, it should be moved to the main test directory instead.

python/ray/data/tests/unit/conftest.py

@@ -0,0 +1,24 @@
+import time
+
+import pytest
+
+import ray
+
+
+@pytest.fixture(autouse=True)
+def disallow_ray_init(monkeypatch):
+    def raise_on_init():
+        raise RuntimeError("Unit tests should not depend on Ray being initialized.")
+
+    monkeypatch.setattr(ray, "init", raise_on_init)
+
+
+@pytest.fixture(autouse=True)
+def disallow_time_sleep(monkeypatch):
+    def raise_on_sleep(seconds):
+        raise RuntimeError(
+            f"Unit tests should not use time.sleep({seconds}). "
+            "Unit tests should be fast and deterministic."
+        )
+
+    monkeypatch.setattr(time, "sleep", raise_on_sleep)