Closes Bears-R-Us#3782: flip function to match numpy

arkouda/categorical.py

@@ -79,7 +79,7 @@ class Categorical:
     BinOps = frozenset(["==", "!="])
     RegisterablePieces = frozenset(["categories", "codes", "permutation", "segments", "_akNAcode"])
     RequiredPieces = frozenset(["categories", "codes", "_akNAcode"])
-    permutation = None
+    permutation: Union[pdarray, None] = None
     segments = None
     objType = "Categorical"
     dtype = akdtype(str_)  # this is being set for now because Categoricals only supported on Strings

arkouda/numpy/__init__.py

@@ -96,3 +96,4 @@
 from arkouda.numpy.rec import *
 
 from ._numeric import *
+from ._manipulation_functions import *

arkouda/numpy/_manipulation_functions.py

@@ -0,0 +1,88 @@
+# from __future__ import annotations
+
+from typing import Optional
+from typing import Tuple
+from typing import Union
+from typing import cast
+
+from arkouda.client import generic_msg
+from arkouda.pdarrayclass import create_pdarray
+from arkouda.pdarrayclass import pdarray
+from arkouda.strings import Strings
+from arkouda.categorical import Categorical
+
+
+__all__ = ["flip"]
+
+
+def flip(
+    x: Union[pdarray, Strings, Categorical], /, *, axis: Optional[Union[int, Tuple[int, ...]]] = None
+) -> Union[pdarray, Strings, Categorical]:
+    """
+    Reverse an array's values along a particular axis or axes.
+
+    Parameters
+    ----------
+    x : pdarray, Strings, or Categorical
+        Reverse the order of elements in an array along the given axis.
+
+        The shape of the array is preserved, but the elements are reordered.
+    axis : int or Tuple[int, ...], optional
+        The axis or axes along which to flip the array. If None, flip the array along all axes.
+    Returns
+    -------
+    pdarray, Strings, or Categorical
+        An array with the entries of axis reversed.
+    Note
+    ----
+    This differs from numpy as it actually reverses the data, rather than presenting a view.
+    """
+    axisList = []
+    if axis is not None:
+        axisList = list(axis) if isinstance(axis, tuple) else [axis]
+
+    if isinstance(x, pdarray):
+        try:
+            return create_pdarray(
+                cast(
+                    str,
+                    generic_msg(
+                        cmd=(
+                            f"flipAll<{x.dtype},{x.ndim}>"
+                            if axis is None
+                            else f"flip<{x.dtype},{x.ndim}>"
+                        ),
+                        args={
+                            "name": x,
+                            "nAxes": len(axisList),
+                            "axis": axisList,
+                        },
+                    ),
+                )
+            )
+
+        except RuntimeError as e:
+            raise IndexError(f"Failed to flip array: {e}")
+    elif isinstance(x, Categorical):
+        if isinstance(x.permutation, pdarray):
+            return Categorical.from_codes(
+                codes=flip(x.codes),
+                categories=x.categories,
+                permutation=flip(x.permutation),
+                segments=x.segments,
+            )
+        else:
+            return Categorical.from_codes(
+                codes=flip(x.codes),
+                categories=x.categories,
+                permutation=None,
+                segments=x.segments,
+            )
+
+    elif isinstance(x, Strings):
+        rep_msg = generic_msg(
+                cmd="flipString", args={"objType": x.objType, "obj": x.entry,  "size": x.size}
+            )
+        return Strings.from_return_msg(cast(str, rep_msg))
+    else:
+        raise TypeError("flip only accepts type pdarray, Strings, or Categorical.")

pytest.ini

@@ -31,6 +31,7 @@ testpaths =
     tests/logger_test.py
     tests/message_test.py
     tests/numpy/dtypes_test.py
+    tests/numpy/manipulation_functions_test.py
     tests/numpy/numeric_test.py
     tests/numpy/numpy_test.py
     tests/operator_test.py

src/SegmentedMsg.chpl

@@ -17,6 +17,7 @@ module SegmentedMsg {
   use Map;
   use CTypes;
   use IOUtils;
+  use CommAggregation;
 
   private config const logLevel = ServerConfig.logLevel;
   private config const logChannel = ServerConfig.logChannel;
@@ -1186,6 +1187,37 @@ module SegmentedMsg {
     return new MsgTuple(repMsg, MsgType.NORMAL);
   }
 
+  proc flipStringMsg(cmd: string, msgArgs: borrowed MessageArgs, st: borrowed SymTab): MsgTuple throws {
+    const name = msgArgs.getValueOf("obj");
+
+    // check to make sure symbols defined
+    st.checkTable(name);
+    var strings = getSegString(name, st);
+    ref origVals = strings.values.a;
+    ref offs = strings.offsets.a;
+    const lengths = strings.getLengths();
+
+    ref retOffs = makeDistArray(lengths.domain, int);
+    forall i in lengths.domain with (var valAgg = newDstAggregator(int)) {
+      valAgg.copy(retOffs[lengths.domain.high - i], lengths[i]);
+    }
+    retOffs = (+ scan retOffs) - retOffs;
+
+    var flippedVals = makeDistArray(strings.values.a.domain, uint(8));
+    forall (off, len, j) in zip(offs, lengths, ..#offs.size) with (var valAgg = newDstAggregator(uint(8))) {
+      var i = 0;
+      for b in interpretAsBytes(origVals, off..#len, borrow=true) {
+        valAgg.copy(flippedVals[retOffs[lengths.domain.high - j] + i], b:uint(8));
+        i += 1;
+      }
+    }
+
+    var retString = getSegString(retOffs, flippedVals, st);
+    var repMsg = "created " + st.attrib(retString.name) + "+created bytes.size %?".format(retString.nBytes);
+    smLogger.debug(getModuleName(),getRoutineName(),getLineNumber(),repMsg);
+    return new MsgTuple(repMsg, MsgType.NORMAL);
+  }
+
   use CommandMap;
   registerFunction("segmentLengths", segmentLengthsMsg, getModuleName());
   registerFunction("caseChange", caseChangeMsg, getModuleName());
@@ -1210,4 +1242,5 @@ module SegmentedMsg {
   registerFunction("segmentedWhere", segmentedWhereMsg, getModuleName());
   registerFunction("segmentedFull", segmentedFullMsg, getModuleName());
   registerFunction("getSegStringProperty", getSegStringPropertyMsg, getModuleName());
+  registerFunction("flipString", flipStringMsg, getModuleName());
 }

tests/numpy/manipulation_functions_test.py

@@ -0,0 +1,56 @@
+import numpy as np
+import pytest
+
+import arkouda as ak
+from arkouda.categorical import Categorical
+from arkouda.testing import assert_equal
+
+seed = pytest.seed
+
+
+class TestJoin:
+
+    @pytest.mark.parametrize("size", pytest.prob_size)
+    @pytest.mark.parametrize("dtype", [int, ak.int64, ak.uint64, float, ak.float64, bool, ak.bool_])
+    def test_flip_pdarray(self, size, dtype):
+        a = ak.arange(size, dtype=dtype)
+        f = ak.flip(a)
+        assert_equal(f, a[::-1])
+
+    @pytest.mark.skip_if_max_rank_less_than(3)
+    @pytest.mark.parametrize("size", pytest.prob_size)
+    @pytest.mark.parametrize("dtype", [ak.int64, ak.uint64, ak.float64])
+    def test_flip_multi_dim(self, size, dtype):
+        a = ak.arange(size * 4, dtype=dtype).reshape((2, 2, size))
+        f = ak.flip(a)
+        assert_equal(f, (size * 4 - 1) - a)
+
+    @pytest.mark.skip_if_max_rank_less_than(3)
+    @pytest.mark.parametrize("size", pytest.prob_size)
+    def test_flip_multi_dim_bool(self, size):
+        shape = (size, 2)
+
+        vals = ak.array([True, False])
+        segs = ak.array([0, size])
+        perm = ak.concatenate([ak.arange(size) * 2, ak.arange(size) * 2 + 1])
+        a = ak.broadcast(segments=segs, values=vals, permutation=perm).reshape(shape)
+        f = ak.flip(a)
+
+        vals2 = ak.array([False, True])
+        f2 = ak.broadcast(segments=segs, values=vals2, permutation=perm).reshape(shape)
+        assert_equal(f, f2)
+
+    @pytest.mark.parametrize("size", pytest.prob_size)
+    def test_flip_string(self, size):
+        s = ak.random_strings_uniform(1, 2, size, seed=seed)
+        assert_equal(ak.flip(s), s[::-1])
+
+    @pytest.mark.parametrize("size", pytest.prob_size)
+    def test_flip_categorical(self, size):
+        s = ak.random_strings_uniform(1, 2, size, seed=seed)
+        c = Categorical(s)
+        assert_equal(ak.flip(c), c[::-1])
+
+        # test case when c.permutation = None
+        c2 = Categorical(c.to_pandas())
+        assert_equal(ak.flip(c2), c2[::-1])