Use canonical "magic methods" (replace x.__repr__() with repr(x)). (

#10735)

This PR uses the canonical builtin function `repr(x)` to compute representations instead of the double-underscored form `x.__repr__()`. The magic method `def __repr__(self): ...` is how a custom representation is defined for a class, but it is not the typical way to call that function. Same for `x.__iter__()` ➡️ `iter(x)` and similarly for `__hash__`, `__str__`, etc. This PR fixes all of the instances of that pattern that I could find, mostly affecting `repr(x)`.

Note that non-standard magic methods like `__dlpack__` or `__dataframe__` that are not defined in the Python object model do not necessarily follow this convention where a free function of the same name is used to call those methods.

Authors:
  - Bradley Dice (https://github.com/bdice)

Approvers:
  - GALI PREM SAGAR (https://github.com/galipremsagar)

URL: #10735

python/cudf/cudf/_fuzz_testing/fuzzer.py

@@ -57,7 +57,7 @@ def log_stats(self):
         logging.info(f"Run-Time elapsed (hh:mm:ss.ms) {total_time_taken}")
 
     def write_crash(self, error):
-        error_file_name = datetime.datetime.now().__str__()
+        error_file_name = str(datetime.datetime.now())
         if self._crash_dir:
             crash_path = os.path.join(
                 self._crash_dir,

python/cudf/cudf/_lib/scalar.pyx

@@ -173,10 +173,10 @@ cdef class DeviceScalar:
         if self.value is cudf.NA:
             return (
                 f"{self.__class__.__name__}"
-                f"({self.value}, {self.dtype.__repr__()})"
+                f"({self.value}, {repr(self.dtype)})"
             )
         else:
-            return f"{self.__class__.__name__}({self.value.__repr__()})"
+            return f"{self.__class__.__name__}({repr(self.value)})"
 
     @staticmethod
     cdef DeviceScalar from_unique_ptr(unique_ptr[scalar] ptr, dtype=None):

python/cudf/cudf/core/column_accessor.py

@@ -149,7 +149,7 @@ def _create_unsafe(
         return obj
 
     def __iter__(self):
-        return self._data.__iter__()
+        return iter(self._data)
 
     def __getitem__(self, key: Any) -> ColumnBase:
         return self._data[key]
@@ -158,7 +158,7 @@ def __setitem__(self, key: Any, value: Any):
         self.set_by_label(key, value)
 
     def __delitem__(self, key: Any):
-        self._data.__delitem__(key)
+        del self._data[key]
         self._clear_cache()
 
     def __len__(self) -> int:

python/cudf/cudf/core/dtypes.py

@@ -237,7 +237,7 @@ def __eq__(self, other):
 
     def __repr__(self):
         if isinstance(self.element_type, (ListDtype, StructDtype)):
-            return f"{type(self).__name__}({self.element_type.__repr__()})"
+            return f"{type(self).__name__}({repr(self.element_type)})"
         else:
             return f"{type(self).__name__}({self.element_type})"
 

python/cudf/cudf/core/frame.py

@@ -3375,7 +3375,7 @@ def to_string(self):
         >>> df.to_string()
         '   key   val\\n0    0  10.0\\n1    1  11.0\\n2    2  12.0'
         """
-        return self.__repr__()
+        return repr(self)
 
     def __str__(self):
         return self.to_string()

python/cudf/cudf/core/index.py

@@ -1116,7 +1116,7 @@ def __repr__(self):
         # related issue : https://github.com/pandas-dev/pandas/issues/35389
         if isinstance(preprocess, CategoricalIndex):
             if preprocess.categories.dtype.kind == "f":
-                output = (
+                output = repr(
                     preprocess.astype("str")
                     .to_pandas()
                     .astype(
@@ -1127,18 +1127,17 @@ def __repr__(self):
                             ordered=preprocess.dtype.ordered,
                         )
                     )
-                    .__repr__()
                 )
                 break_idx = output.find("ordered=")
                 output = (
                     output[:break_idx].replace("'", "") + output[break_idx:]
                 )
             else:
-                output = preprocess.to_pandas().__repr__()
+                output = repr(preprocess.to_pandas())
 
             output = output.replace("nan", cudf._NA_REP)
         elif preprocess._values.nullable:
-            output = self._clean_nulls_from_index().to_pandas().__repr__()
+            output = repr(self._clean_nulls_from_index().to_pandas())
 
             if not isinstance(self, StringIndex):
                 # We should remove all the single quotes
@@ -1150,7 +1149,7 @@ def __repr__(self):
                 # of StringIndex and it is valid to have them.
                 output = output.replace("'", "")
         else:
-            output = preprocess.to_pandas().__repr__()
+            output = repr(preprocess.to_pandas())
 
         # Fix and correct the class name of the output
         # string by finding first occurrence of "(" in the output

python/cudf/cudf/core/multiindex.py

@@ -470,7 +470,7 @@ def __repr__(self):
         else:
             preprocess = preprocess.to_pandas(nullable=True)
 
-        output = preprocess.__repr__()
+        output = repr(preprocess)
         output_prefix = self.__class__.__name__ + "("
         output = output.lstrip(output_prefix)
         lines = output.split("\n")

python/cudf/cudf/core/series.py

@@ -1064,11 +1064,8 @@ def __repr__(self):
         ) or isinstance(
             preprocess._column, cudf.core.column.timedelta.TimeDeltaColumn
         ):
-            output = (
-                preprocess.astype("O")
-                .fillna(cudf._NA_REP)
-                .to_pandas()
-                .__repr__()
+            output = repr(
+                preprocess.astype("O").fillna(cudf._NA_REP).to_pandas()
             )
         elif isinstance(
             preprocess._column, cudf.core.column.CategoricalColumn
@@ -1111,7 +1108,7 @@ def __repr__(self):
                 na_rep=cudf._NA_REP,
             )
         else:
-            output = preprocess.to_pandas().__repr__()
+            output = repr(preprocess.to_pandas())
 
         lines = output.split("\n")
 

python/cudf/cudf/core/udf/typing.py

@@ -63,7 +63,7 @@ def __hash__(self):
         Needed so that numba caches type instances with different
         `value_type` separately.
         """
-        return self.__repr__().__hash__()
+        return hash(repr(self))
 
     def unify(self, context, other):
         """

python/cudf/cudf/tests/test_multiindex.py

@@ -688,7 +688,7 @@ def test_multiindex_copy_sem(data, levels, codes, names):
     # Test same behavior when used on DataFrame
     gdf.index = gmi_copy
     pdf.index = pmi_copy
-    assert gdf.__repr__() == pdf.__repr__()
+    assert repr(gdf) == repr(pdf)
 
 
 @pytest.mark.parametrize(