def test_format(self):
series = pd.Series(
[dt(1999, 2, 3, 4, 5, 6, 7), np.nan, dt(2000, 3, 4, 5, 6, 7, 8)]
)
column_type = ColumnType.DATETIME()
result = column_type.format_series(series)
assert_series_equal(
result,
pd.Series(
["1999-02-03T04:05:06.000007Z", np.nan, "2000-03-04T05:06:07.000008Z"]
),
)
def test_ctor_infer_columns(self):
result = ProcessResult(
pd.DataFrame(
{
"A": [1, 2],
"B": ["x", "y"],
"C": [np.nan, dt(2019, 3, 3, 4, 5, 6, 7)],
}
)
)
self.assertEqual(
result.columns,
[
Column("A", ColumnType.NUMBER()),
Column("B", ColumnType.TEXT()),
Column("C", ColumnType.DATETIME()),
],
)
def test_arrow_datetime_column(self):
dataframe, columns = arrow_table_to_dataframe(
arrow_table(
{
"A": pyarrow.array(
[dt.fromisoformat("2019-09-17T21:21:00.123456"), None],
type=pyarrow.timestamp(unit="ns", tz=None),
)
},
[atypes.Column("A", atypes.ColumnType.Datetime())],
)
)
assert_frame_equal(
dataframe,
pd.DataFrame(
{"A": ["2019-09-17T21:21:00.123456Z", None]}, dtype="datetime64[ns]"
),
)
self.assertEqual(columns, [Column("A", ColumnType.DATETIME())])
def test_columns(self):
df = pd.DataFrame(
{
"A": [1], # number
"B": ["foo"], # str
"C": dt(2018, 8, 20), # datetime64
}
)
df["D"] = pd.Series(["cat"], dtype="category")
result = ProcessResult(df)
self.assertEqual(result.column_names, ["A", "B", "C", "D"])
self.assertEqual(
result.columns,
[
Column("A", ColumnType.NUMBER()),
Column("B", ColumnType.TEXT()),
Column("C", ColumnType.DATETIME()),
Column("D", ColumnType.TEXT()),
],
)
def test_to_arrow(self):
self.assertEqual(ColumnType.DATETIME().to_arrow(), atypes.ColumnType.Datetime())
def test_from_arrow(self):
self.assertEqual(
ColumnType.from_arrow(atypes.ColumnType.Datetime()), ColumnType.DATETIME()
)