Merge branch 'feature/fallback-debugging' of github.com:Matt711/cudf …

…into branch-24.08

python/cudf/cudf/pandas/fast_slow_proxy.py

@@ -16,11 +16,69 @@
 
 import numpy as np
 
+from rmm import RMMError
+
 from ..options import _env_get_bool
 from ..testing._utils import assert_eq
 from .annotation import nvtx
 
 
+class CudfPandasDebugWarning(UserWarning):
+    """Base warning for an incorrect result in cuDF or Pandas. Or the Pandas result was uncomputable"""
+
+    pass
+
+
+class CudfPandasResultsDifferentWarning(UserWarning):
+    """Warns when the results from cuDF and Pandas were different"""
+
+    pass
+
+
+class CudfPandasPandasErrorWarning(UserWarning):
+    """Warns when the results from Pandas could not be computed"""
+
+    pass
+
+
+class CudfPandasDebuggingFailedWarning(UserWarning):
+    """Warns when the cuDF-Pandas debugging fails"""
+
+    pass
+
+
+class CudfPandasDebugFallbackWarning(UserWarning):
+    """Base warning fof when fallback occurs"""
+
+    pass
+
+
+class CudfPandasDebugOOMWarning(CudfPandasDebugFallbackWarning):
+    """Warns when cuDF produces a MemoryError or an rmm.RMMError"""
+
+    pass
+
+
+class CudfPandasDebugNotImplementedErrorWarning(
+    CudfPandasDebugFallbackWarning
+):
+    """Warns cuDF produces a NotImplementedError"""
+
+    pass
+
+
+class CudfPandasDebugAttributeErrorWarning(CudfPandasDebugFallbackWarning):
+    """Warns when cuDF produces an AttributeError"""
+
+    pass
+
+
+class CudfPandasDebugTypeErrorWarning(CudfPandasDebugFallbackWarning):
+    """Warns when cuDF produces a TypeError"""
+
+    pass
+
+
 def call_operator(fn, args, kwargs):
     return fn(*args, **kwargs)
 
@@ -915,22 +973,50 @@ def _fast_slow_function_call(
                 except Exception as e:
                     warnings.warn(
                         "The result from pandas could not be computed. "
-                        f"The exception was {e}."
+                        f"The exception was {e}.",
+                        CudfPandasPandasErrorWarning,
                     )
                 else:
                     try:
-                        _assert_fast_slow_eq(result, slow_result)
+                        _assert_fast_slow_eq(result, slow_result, **kwargs)
                     except AssertionError as e:
                         warnings.warn(
                             "The results from cudf and pandas were different. "
-                            f"The exception was {e}."
+                            f"The exception was {e}.",
+                            CudfPandasResultsDifferentWarning,
                         )
                     except Exception as e:
                         warnings.warn(
-                            "Pandas debugging mode failed. "
-                            f"The exception was {e}."
+                            "cuDF-Pandas debugging failed. "
+                            f"The exception was {e}.",
+                            CudfPandasDebuggingFailedWarning,
                         )
-    except Exception:
+    except Exception as e:
+        if _env_get_bool("CUDF_PANDAS_FALLBACK_DEBUGGING", False):
+            if isinstance(e, (RMMError, MemoryError)):
+                warnings.warn(
+                    "Out of Memory Error. Falling back to the slow path. "
+                    f"The exception was {e}.",
+                    CudfPandasDebugOOMWarning,
+                )
+            elif isinstance(e, NotImplementedError):
+                warnings.warn(
+                    "NotImplementedError. Falling back to the slow path. "
+                    f"The exception was {e}.",
+                    CudfPandasDebugNotImplementedErrorWarning,
+                )
+            elif isinstance(e, AttributeError):
+                warnings.warn(
+                    "AttributeError. Falling back to the slow path. "
+                    f"The exception was {e}.",
+                    CudfPandasDebugAttributeErrorWarning,
+                )
+            elif isinstance(e, TypeError):
+                warnings.warn(
+                    "TypeError. Falling back to the slow path. "
+                    f"The exception was {e}.",
+                    CudfPandasDebugTypeErrorWarning,
+                )
         with nvtx.annotate(
             "EXECUTE_SLOW",
             color=_CUDF_PANDAS_NVTX_COLORS["EXECUTE_SLOW"],

python/cudf/cudf_pandas_tests/test_cudf_pandas.py

@@ -20,8 +20,20 @@
 from numba import NumbaDeprecationWarning
 from pytz import utc
 
+from rmm import RMMError
+
 from cudf.pandas import LOADED, Profiler
-from cudf.pandas.fast_slow_proxy import _Unusable, is_proxy_object
+from cudf.pandas.fast_slow_proxy import (
+    CudfPandasDebugAttributeErrorWarning,
+    CudfPandasDebuggingFailedWarning,
+    CudfPandasDebugNotImplementedErrorWarning,
+    CudfPandasDebugOOMWarning,
+    CudfPandasDebugTypeErrorWarning,
+    CudfPandasPandasErrorWarning,
+    CudfPandasResultsDifferentWarning,
+    _Unusable,
+    is_proxy_object,
+)
 
 if not LOADED:
     raise ImportError("These tests must be run with cudf.pandas loaded")
@@ -1429,6 +1441,127 @@ def test_holidays_within_dates(holiday, start, expected):
     ) == [utc.localize(dt) for dt in expected]
 
 
+def mock_mean_one(self, *args, **kwargs):
+    return np.float64(1.0)
+
+
+def mock_mean_exception(self, *args, **kwargs):
+    raise Exception()
+
+
+def mock_mean_none(self, *args, **kwargs):
+    return None
+
+
+def mock_mean_memory_error(self, *args, **kwargs):
+    raise MemoryError()
+
+
+def mock_mean_rmm_error(self, *args, **kwargs):
+    raise RMMError(1, "error")
+
+
+def mock_mean_not_impl_error(self, *args, **kwargs):
+    raise NotImplementedError()
+
+
+def mock_mean_attr_error(self, *args, **kwargs):
+    raise AttributeError()
+
+
+def mock_mean_type_error(self, *args, **kwargs):
+    raise TypeError()
+
+
+@pytest.mark.parametrize(
+    "mock_mean, warning, match_str, env_var, original_mean, proxy_attr",
+    [
+        (
+            mock_mean_one,
+            CudfPandasResultsDifferentWarning,
+            "The results from cudf and pandas were different.",
+            "CUDF_PANDAS_DEBUGGING",
+            pd.Series.mean,
+            "_fsproxy_slow",
+        ),
+        (
+            mock_mean_exception,
+            CudfPandasPandasErrorWarning,
+            "The result from pandas could not be computed.",
+            "CUDF_PANDAS_DEBUGGING",
+            pd.Series.mean,
+            "_fsproxy_slow",
+        ),
+        (
+            mock_mean_none,
+            CudfPandasDebuggingFailedWarning,
+            "cuDF-Pandas debugging failed.",
+            "CUDF_PANDAS_DEBUGGING",
+            pd.Series.mean,
+            "_fsproxy_slow",
+        ),
+        (
+            mock_mean_memory_error,
+            CudfPandasDebugOOMWarning,
+            "Out of Memory Error.",
+            "CUDF_PANDAS_FALLBACK_DEBUGGING",
+            cudf.Series.mean,
+            "_fsproxy_fast",
+        ),
+        (
+            mock_mean_rmm_error,
+            CudfPandasDebugOOMWarning,
+            "Out of Memory Error.",
+            "CUDF_PANDAS_FALLBACK_DEBUGGING",
+            cudf.Series.mean,
+            "_fsproxy_fast",
+        ),
+        (
+            mock_mean_not_impl_error,
+            CudfPandasDebugNotImplementedErrorWarning,
+            "NotImplementedError.",
+            "CUDF_PANDAS_FALLBACK_DEBUGGING",
+            cudf.Series.mean,
+            "_fsproxy_fast",
+        ),
+        (
+            mock_mean_attr_error,
+            CudfPandasDebugAttributeErrorWarning,
+            "AttributeError.",
+            "CUDF_PANDAS_FALLBACK_DEBUGGING",
+            cudf.Series.mean,
+            "_fsproxy_fast",
+        ),
+        (
+            mock_mean_type_error,
+            CudfPandasDebugTypeErrorWarning,
+            "TypeError.",
+            "CUDF_PANDAS_FALLBACK_DEBUGGING",
+            cudf.Series.mean,
+            "_fsproxy_fast",
+        ),
+    ],
+)
+def test_cudf_pandas_debugging(
+    monkeypatch,
+    mock_mean,
+    warning,
+    match_str,
+    env_var,
+    original_mean,
+    proxy_attr,
+):
+    with monkeypatch.context() as monkeycontext:
+        monkeypatch.setattr(xpd.Series.mean, proxy_attr, mock_mean)
+        monkeycontext.setenv(env_var, "True")
+        s = xpd.Series([1, 2])
+        with pytest.warns(warning, match=match_str):
+            assert s.mean() == 1.5
+
+    # Must explicitly undo the patch. Proxy dispatch doesn't work with monkeypatch contexts.
+    monkeypatch.setattr(xpd.Series.mean, proxy_attr, original_mean)
+
+
 @pytest.mark.parametrize(
     "env_value",
     ["", "cuda", "pool", "async", "managed", "managed_pool", "abc"],
@@ -1456,67 +1589,6 @@ def test_rmm_option_on_import(env_value):
         assert sp_completed.returncode == 1
 
 
-def test_cudf_pandas_debugging_different_results(monkeypatch):
-    cudf_mean = cudf.Series.mean
-
-    def mock_mean_one(self, *args, **kwargs):
-        return np.float64(1.0)
-
-    with monkeypatch.context() as monkeycontext:
-        monkeypatch.setattr(xpd.Series.mean, "_fsproxy_fast", mock_mean_one)
-        monkeycontext.setenv("CUDF_PANDAS_DEBUGGING", "True")
-        s = xpd.Series([1, 2])
-        with pytest.warns(
-            UserWarning,
-            match="The results from cudf and pandas were different.",
-        ):
-            assert s.mean() == 1.0
-    # Must explicitly undo the patch. Proxy dispatch doesn't work with monkeypatch contexts.
-    monkeypatch.setattr(xpd.Series.mean, "_fsproxy_fast", cudf_mean)
-
-
-def test_cudf_pandas_debugging_pandas_error(monkeypatch):
-    pd_mean = pd.Series.mean
-
-    def mock_mean_exception(self, *args, **kwargs):
-        raise Exception()
-
-    with monkeypatch.context() as monkeycontext:
-        monkeycontext.setattr(
-            xpd.Series.mean, "_fsproxy_slow", mock_mean_exception
-        )
-        monkeycontext.setenv("CUDF_PANDAS_DEBUGGING", "True")
-        s = xpd.Series([1, 2])
-        with pytest.warns(
-            UserWarning,
-            match="The result from pandas could not be computed.",
-        ):
-            s = xpd.Series([1, 2])
-            assert s.mean() == 1.5
-    # Must explicitly undo the patch. Proxy dispatch doesn't work with monkeypatch contexts.
-    monkeypatch.setattr(xpd.Series.mean, "_fsproxy_slow", pd_mean)
-
-
-def test_cudf_pandas_debugging_failed(monkeypatch):
-    pd_mean = pd.Series.mean
-
-    def mock_mean_none(self, *args, **kwargs):
-        return None
-
-    with monkeypatch.context() as monkeycontext:
-        monkeycontext.setattr(xpd.Series.mean, "_fsproxy_slow", mock_mean_none)
-        monkeycontext.setenv("CUDF_PANDAS_DEBUGGING", "True")
-        s = xpd.Series([1, 2])
-        with pytest.warns(
-            UserWarning,
-            match="Pandas debugging mode failed.",
-        ):
-            s = xpd.Series([1, 2])
-            assert s.mean() == 1.5
-    # Must explicitly undo the patch. Proxy dispatch doesn't work with monkeypatch contexts.
-    monkeypatch.setattr(xpd.Series.mean, "_fsproxy_slow", pd_mean)
-
-
 def test_excelwriter_pathlike():
     assert isinstance(pd.ExcelWriter("foo.xlsx"), os.PathLike)