pitrou commented on code in PR #13327:
URL: https://github.com/apache/arrow/pull/13327#discussion_r892170239
##########
python/pyarrow/tests/test_compute.py:
##########
@@ -2743,6 +2745,43 @@ def test_random():
pc.random(100, initializer=[])
[email protected](
+ "tiebreaker,expected_values",
+ [("min", [3, 1, 4, 6, 4, 6, 1]),
+ ("max", [3, 2, 5, 7, 5, 7, 2]),
+ ("first", [3, 1, 4, 6, 5, 7, 2]),
+ ("dense", [2, 1, 3, 4, 3, 4, 1])]
+)
+def test_rank_options_tiebreaker(tiebreaker, expected_values):
+ arr = pa.array([1.2, 0.0, 5.3, None, 5.3, None, 0.0])
+ rank_options = pc.RankOptions(sort_keys=[("b", "ascending")],
+ null_placement="at_end",
+ tiebreaker=tiebreaker)
+ result = pc.rank(arr, options=rank_options)
+ expected = pa.array(expected_values, type=pa.uint64())
+ assert result.equals(expected)
+
+
+def test_rank_options():
+ arr = pa.array([1.2, 0.0, 5.3, None, 5.3, None, 0.0])
+ expected = pa.array([3, 1, 4, 6, 5, 7, 2], type=pa.uint64())
+
+ # Ensure rank can be called without specifying options
+ result = pc.rank(arr)
+ assert result.equals(expected)
+
+ # Ensure default RankOptions
+ result = pc.rank(arr, options=pc.RankOptions())
+ assert result.equals(expected)
+
Review Comment:
Can you also test `null_placement="at_start"` here?
##########
python/pyarrow/tests/test_compute.py:
##########
@@ -2743,6 +2745,43 @@ def test_random():
pc.random(100, initializer=[])
[email protected](
+ "tiebreaker,expected_values",
+ [("min", [3, 1, 4, 6, 4, 6, 1]),
+ ("max", [3, 2, 5, 7, 5, 7, 2]),
+ ("first", [3, 1, 4, 6, 5, 7, 2]),
+ ("dense", [2, 1, 3, 4, 3, 4, 1])]
+)
+def test_rank_options_tiebreaker(tiebreaker, expected_values):
+ arr = pa.array([1.2, 0.0, 5.3, None, 5.3, None, 0.0])
+ rank_options = pc.RankOptions(sort_keys=[("b", "ascending")],
+ null_placement="at_end",
+ tiebreaker=tiebreaker)
+ result = pc.rank(arr, options=rank_options)
+ expected = pa.array(expected_values, type=pa.uint64())
+ assert result.equals(expected)
+
+
+def test_rank_options():
+ arr = pa.array([1.2, 0.0, 5.3, None, 5.3, None, 0.0])
+ expected = pa.array([3, 1, 4, 6, 5, 7, 2], type=pa.uint64())
+
+ # Ensure rank can be called without specifying options
+ result = pc.rank(arr)
+ assert result.equals(expected)
+
+ # Ensure default RankOptions
+ result = pc.rank(arr, options=pc.RankOptions())
+ assert result.equals(expected)
+
+ with pytest.raises(ValueError,
+ match=r'"NonExisting" is not a valid tiebreaker'):
+ rank_options = pc.RankOptions(sort_keys=[("b", "ascending")],
+ null_placement="at_end",
+ tiebreaker="NonExisting")
+ pc.rank(arr, options=rank_options)
Review Comment:
I don't think calling `pc.rank` is necessary, the `RankOptions` constructor
should raise the error.
##########
python/pyarrow/_compute.pyx:
##########
@@ -2047,6 +2047,60 @@ class RandomOptions(_RandomOptions):
self._set_options(length, initializer)
+cdef class _RankOptions(FunctionOptions):
+
+ _tiebreaker_map = {
+ "min": CRankOptionsTiebreaker_Min,
+ "max": CRankOptionsTiebreaker_Max,
+ "first": CRankOptionsTiebreaker_First,
+ "dense": CRankOptionsTiebreaker_Dense,
+ }
+
+ def _set_options(self, sort_keys, null_placement, tiebreaker):
+ cdef vector[CSortKey] c_sort_keys
+ for name, order in sort_keys:
+ c_sort_keys.push_back(
+ CSortKey(tobytes(name), unwrap_sort_order(order))
+ )
+ try:
+ self.wrapped.reset(
+ new CRankOptions(c_sort_keys,
unwrap_null_placement(null_placement), self._tiebreaker_map[tiebreaker])
+ )
+ except KeyError:
+ _raise_invalid_function_option(tiebreaker, "tiebreaker")
+
+
+class RankOptions(_RankOptions):
+ """
+ Options for the `rank` function.
+
+ Parameters
+ ----------
+ sort_keys : sequence of (name, order) tuples, optional
+ Names of field/column keys to sort the input on,
+ along with the order each field/column is sorted in.
+ Accepted values for `order` are "ascending", "descending".
Review Comment:
Mention what the default is when not specified?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
