def test_coerce_infer_columns_try_fallback_columns_ignore_wrong_type(self):
table = pd.DataFrame({"A": [1, 2], "B": ["x", "y"]})
result = ProcessResult.coerce(
table,
try_fallback_columns=[
Column("A", ColumnType.TEXT()),
Column("B", ColumnType.NUMBER()),
],
)
self.assertEqual(
result.columns,
[Column("A", ColumnType.NUMBER()), Column("B", ColumnType.TEXT())],
)
def test_coerce_infer_columns(self):
table = pd.DataFrame({"A": [1, 2], "B": ["x", "y"]})
result = ProcessResult.coerce(table)
self.assertEqual(
result.columns,
[Column("A", ColumnType.NUMBER()), Column("B", ColumnType.TEXT())],
)
def test_arrow_all_null_text_column(self):
dataframe, columns = arrow_table_to_dataframe(
arrow_table(
{"A": pyarrow.array(["a", "b", None, "c"])},
columns=[atypes.Column("A", atypes.ColumnType.Text())],
)
)
assert_frame_equal(dataframe, pd.DataFrame({"A": ["a", "b", np.nan, "c"]}))
self.assertEqual(columns, [Column("A", ColumnType.TEXT())])
def test_dataframe_all_null_text_column(self):
assert_arrow_table_equals(
dataframe_to_arrow_table(
pd.DataFrame({"A": [None]}, dtype=str),
[Column("A", ColumnType.TEXT())],
self.path,
),
arrow_table({"A": pyarrow.array([None], pyarrow.string())}),
)
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_coerce_infer_columns_with_format(self):
table = pd.DataFrame({"A": [1, 2], "B": ["x", "y"]})
result = ProcessResult.coerce(
{"dataframe": table, "column_formats": {"A": "{:,d}"}}
)
self.assertEqual(
result.columns,
[
Column("A", ColumnType.NUMBER(format="{:,d}")),
Column("B", ColumnType.TEXT()),
],
)
def test_arrow_all_null_category_column(self):
atable = arrow_table(
{
"A": pyarrow.DictionaryArray.from_arrays(
pyarrow.array([None], type=pyarrow.int8()),
pyarrow.array([], type=pyarrow.string()),
)
}
)
dataframe, columns = arrow_table_to_dataframe(atable)
self.assertEqual(columns, [Column("A", ColumnType.TEXT())])
assert_frame_equal(
dataframe, pd.DataFrame({"A": [None]}, dtype=str).astype("category")
)
def test_arrow_category_column(self):
atable = arrow_table(
{
"A": pyarrow.DictionaryArray.from_arrays(
pyarrow.array([0, 1, None, 0], type=pyarrow.int8()),
pyarrow.array(["A", "B"], type=pyarrow.string()),
)
}
)
dataframe, columns = arrow_table_to_dataframe(atable)
self.assertEqual(columns, [Column("A", ColumnType.TEXT())])
assert_frame_equal(
dataframe, pd.DataFrame({"A": ["A", "B", None, "A"]}, dtype="category")
)
def test_dataframe_all_null_category_column(self):
assert_arrow_table_equals(
dataframe_to_arrow_table(
pd.DataFrame({"A": [None]}, dtype=str).astype("category"),
[Column("A", ColumnType.TEXT())],
self.path,
),
arrow_table(
{
"A": pyarrow.DictionaryArray.from_arrays(
pyarrow.array([None], type=pyarrow.int8()),
pyarrow.array([], type=pyarrow.string()),
)
}
),
)
def test_dataframe_category_column(self):
assert_arrow_table_equals(
dataframe_to_arrow_table(
pd.DataFrame({"A": ["A", "B", None, "A"]}, dtype="category"),
[Column("A", ColumnType.TEXT())],
self.path,
),
arrow_table(
{
"A": pyarrow.DictionaryArray.from_arrays(
pyarrow.array([0, 1, None, 0], type=pyarrow.int8()),
pyarrow.array(["A", "B"], type=pyarrow.string()),
)
}
),
)
def test_to_arrow(self):
self.assertEqual(
TableShape(
3,
[
Column("A", ColumnType.NUMBER("{:,d}")),
Column("B", ColumnType.TEXT()),
],
).to_arrow(),
atypes.TableMetadata(
3,
[
atypes.Column("A", atypes.ColumnType.Number("{:,d}")),
atypes.Column("B", atypes.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)],
}
)
)
self.assertEqual(
result.columns,
[
Column("A", ColumnType.NUMBER()),
Column("B", ColumnType.TEXT()),
Column("C", ColumnType.DATETIME()),
],
)
def test_to_arrow(self):
self.assertEqual(ColumnType.TEXT().to_arrow(), atypes.ColumnType.Text())
def test_from_arrow(self):
self.assertEqual(
ColumnType.from_arrow(atypes.ColumnType.Text()), ColumnType.TEXT()
)
def test_text_type(self):
series = pd.Series(["x", np.nan, "z"])
column_type = ColumnType.TEXT()
result = column_type.format_series(series)
assert_series_equal(result, pd.Series(["x", np.nan, "z"]))