Merge pull request #77 from martindurant/cleanups

Some missing linkages

src/akimbo/datetimes.py

@@ -20,7 +20,6 @@ def __init__(self, accessor) -> None:
 
     # listed below https://arrow.apache.org/docs/python/generated/
     # pyarrow.compute.ceil_temporal.html
-    cast = dec(pc.cast)  # TODO: move to .ak
     ceil_temporal = dec_t(pc.ceil_temporal)
     floor_temporal = dec_t(pc.floor_temporal)
     reound_temporal = dec_t(pc.round_temporal)
@@ -62,7 +61,7 @@ def __init__(self, accessor) -> None:
     weeks_between = dec_t(pc.weeks_between)
     years_between = dec_t(pc.years_between)
 
-    # TODO: strftime, strptime
+    strftime = dec_t(pc.strftime)
 
     # TODO: timezone conversion
 

src/akimbo/mixin.py

@@ -5,6 +5,9 @@
 from typing import Callable, Iterable
 
 import awkward as ak
+import pyarrow.compute as pc
+
+from akimbo.apply_tree import dec
 
 methods = [
     _ for _ in (dir(ak)) if not _.startswith(("_", "ak_")) and not _[0].isupper()
@@ -142,6 +145,16 @@ def __init__(self, obj, behavior=None):
         self._obj = obj
         self._behavior = behavior
 
+    def __call__(self, *args, behavior=None, **kwargs):
+        return Accessor(self._obj, behavior=behavior)
+
+    cast = dec(pc.cast)
+
+    @property
+    def accessor(self):
+        # if we use `dec`, which expects to work on
+        return self
+
     @classmethod
     def is_series(cls, data):
         return isinstance(data, cls.series_type)

src/akimbo/strings.py

@@ -4,6 +4,7 @@
 from collections.abc import Callable
 
 import awkward as ak
+import pyarrow.compute as pc
 
 from akimbo.apply_tree import dec
 from akimbo.mixin import Accessor
@@ -50,6 +51,12 @@ def _decode(layout):
     if not aname.startswith(("_", "akstr_")) and not aname[0].isupper()
 ]
 
+# make sensible defaults for strptime
+strptime = functools.wraps(pc.strptime)(
+    lambda *args, format="%FT%T", unit="s", error_is_null=True, **kw:
+        pc.strptime(*args, format=format, unit=unit, error_is_null=error_is_null)
+)
+
 
 class StringAccessor:
     def __init__(self, accessor):
@@ -92,8 +99,10 @@ def f(*args, **kwargs):
 
         return f
 
+    strptime = dec(strptime, match=match_string)
+
     def __dir__(self) -> list[str]:
-        return sorted(methods)
+        return sorted(methods + ["strptime"])
 
 
 Accessor.register_accessor("str", StringAccessor)

tests/test_dt.py

@@ -1,15 +1,17 @@
 import datetime
+import os
 
 import pytest
 
 import akimbo.pandas  # noqa
 
 pd = pytest.importorskip("pandas")
+WIN = os.name == "nt"
 
 
 def test_cast():
     s = pd.Series([[0, 1], [1, 0], [2]])
-    out = s.ak.dt.cast("timestamp[s]")
+    out = s.ak.cast("timestamp[s]")
     assert str(out.dtype) == "list<item: timestamp[s]>[pyarrow]"
     assert out.to_list() == [
         [datetime.datetime(1970, 1, 1, 0, 0), datetime.datetime(1970, 1, 1, 0, 0, 1)],
@@ -20,7 +22,7 @@ def test_cast():
 
 def test_unary_unit():
     s = pd.Series([[0, 1], [1, 0], [2]])
-    ts = s.ak.dt.cast("timestamp[s]")
+    ts = s.ak.cast("timestamp[s]")
     s2 = ts.ak.dt.second()
     assert s.to_list() == s2.to_list()
 
@@ -35,8 +37,8 @@ def test_bad_type():
 def test_binary():
     s = pd.Series([[0, 1], [1, 0], [2]])
     s2 = s.ak + 1
-    ts1 = s.ak.dt.cast("timestamp[s]")
-    ts2 = s2.ak.dt.cast("timestamp[s]")
+    ts1 = s.ak.cast("timestamp[s]")
+    ts2 = s2.ak.cast("timestamp[s]")
 
     out = ts1.ak.dt.nanoseconds_between(ts2)
     assert out.tolist() == [
@@ -50,8 +52,8 @@ def test_binary():
 def test_binary_with_kwargs():
     s = pd.Series([[0, 1], [1, 0], [2]])
     s2 = s.ak + int(24 * 3600 * 7 * 2.5)
-    ts1 = s.ak.dt.cast("timestamp[s]")
-    ts2 = s2.ak.dt.cast("timestamp[s]")
+    ts1 = s.ak.cast("timestamp[s]")
+    ts2 = s2.ak.cast("timestamp[s]")
 
     out = ts1.ak.dt.weeks_between(ts2, count_from_zero=False, week_start=2)
     assert out.tolist() == [[2, 2], [2, 2], [2]]
@@ -64,8 +66,16 @@ def test_mixed_record():
     s = pd.Series(data)
 
     # explicit select of where to apply transform
-    ts = s.ak.dt.cast("timestamp[s]", where="a")
+    ts = s.ak.cast("timestamp[s]", where="a")
 
     # implicit selection of timestamps
     s2 = ts.ak.dt.second()
     assert s2.to_list() == data
+
+
+@pytest.mark.skipif(WIN, reason="arrow on windows needs a timezone database")
+def test_text_conversion():
+    s = pd.Series([["2024-08-01T01:00:00", None, "2024-08-01T01:01:00"]])
+    s2 = s.ak.str.strptime()
+    s3 = s2.ak.dt.strftime("%FT%T")
+    assert s3.tolist() == [["2024-08-01T01:00:00", None, "2024-08-01T01:01:00"]]

tests/test_str.py

@@ -31,11 +31,11 @@ def test_encode_decode():
 
 
 def test_split():
-    s = pd.Series(["hello world", "oio", ""])
+    s = pd.Series(["hello world", "oio", pd.NA, ""])
     s2 = s.ak.str.split_whitespace()
-    assert s2.tolist() == [["hello", "world"], ["oio"], [""]]
+    assert s2.tolist() == [["hello", "world"], ["oio"], pd.NA, [""]]
     s2 = s.ak.str.split_pattern("i")
-    assert s2.tolist() == [["hello world"], ["o", "o"], [""]]
+    assert s2.tolist() == [["hello world"], ["o", "o"], pd.NA, [""]]
 
     s = pd.Series([b"hello world", b"oio", b""])
     s2 = s.ak.str.split_whitespace()