def test_arrow_timestamp_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", ColumnType.Timestamp())],
))
assert_frame_equal(
dataframe,
pd.DataFrame({"A": ["2019-09-17T21:21:00.123456Z", None]},
dtype="datetime64[ns]"),
)
self.assertEqual(columns, [Column("A", ColumnType.Timestamp())])
def test_dataframe_datetime_column(self):
self._test_dataframe_to_arrow_table(
pd.DataFrame(
{"A": ["2019-09-17T21:21:00.123456Z", None]}, dtype="datetime64[ns]"
),
[Column("A", ColumnType.Timestamp())],
make_table(
make_column("A", [dt.fromisoformat("2019-09-17T21:21:00.123456"), None])
),
)
def test_dataframe_datetime_column(self):
assert_arrow_table_equals(
dataframe_to_arrow_table(
pd.DataFrame({"A": ["2019-09-17T21:21:00.123456Z", None]},
dtype="datetime64[ns]"),
[Column("A", ColumnType.Timestamp())],
self.path,
),
arrow_table(
{
"A":
pyarrow.array(
[dt.fromisoformat("2019-09-17T21:21:00.123456"), None],
type=pyarrow.timestamp(unit="ns", tz=None),
)
},
[atypes.Column("A", ColumnType.Timestamp())],
),
)
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.Timestamp()),
],
)
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.Timestamp()),
Column("D", ColumnType.Text()),
],
)
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)],
"D": [pd.Period("2021-01-01", freq="D"), pd.NaT],
}
)
)
self.assertEqual(
result.columns,
[
Column("A", ColumnType.Number()),
Column("B", ColumnType.Text()),
Column("C", ColumnType.Timestamp()),
Column("D", ColumnType.Date("day")),
],
)