jesperct commented on code in PR #41060:
URL: https://github.com/apache/superset/pull/41060#discussion_r3421038099
##########
tests/unit_tests/models/helpers_test.py:
##########
@@ -2503,6 +2503,237 @@ def _should_not_be_called(*_: object, **__: object) ->
list[dict[str, object]]:
assert generic_type is None
+def _normalize_df_datasource(column: object) -> MagicMock:
+ """Bind ``ExploreMixin.normalize_df`` to a minimal datasource exposing a
+ single temporal ``column`` via ``get_column``."""
+ from superset.models.helpers import ExploreMixin
+
+ datasource = MagicMock()
+ datasource.offset = 0
+ datasource.enforce_numerical_metrics = False
+ datasource.columns = [column]
+ datasource.get_column = lambda name: {"ts": column}.get(name)
+ for method in ("_python_date_format", "_collect_dttm_labels",
"normalize_df"):
+ setattr(datasource, method, getattr(ExploreMixin,
method).__get__(datasource))
+ return datasource
+
+
+def _raw_query_object() -> MagicMock:
+ """A query object for a raw/unaggregated query: ``columns`` holds the plain
+ physical column name (no base-axis adhoc wrapper, no granularity)."""
+ query_object = MagicMock()
+ query_object.columns = ["ts"]
+ query_object.granularity = None
+ query_object.time_shift = None
+ return query_object
+
+
+def test_normalize_df_applies_python_date_format_to_unaggregated_columns() ->
None:
+ """A temporal column selected in a raw/unaggregated query is a plain column
+ name rather than a base-axis adhoc column, so it is excluded from
+ ``get_base_axis_labels``. It must still receive its ``python_date_format``,
+ so an ``epoch_s`` column is converted to datetimes the same way it is for
+ aggregated charts."""
+ import pandas as pd
+ from pandas.api.types import is_datetime64_any_dtype
+
+ ts_col = MagicMock(
+ column_name="ts",
+ is_dttm=True,
+ python_date_format="epoch_s",
+ datetime_format=None,
+ )
+ datasource = _normalize_df_datasource(ts_col)
+
+ # 2020-01-01, 2021-01-01, 2022-01-01 as epoch seconds
+ df = pd.DataFrame({"ts": [1577836800, 1609459200, 1640995200]})
+
+ result = datasource.normalize_df(df, _raw_query_object())
+
+ assert is_datetime64_any_dtype(result["ts"])
+ assert result["ts"][0].strftime("%Y-%m-%d") == "2020-01-01"
+ assert result["ts"][2].strftime("%Y-%m-%d") == "2022-01-01"
+
+
+def test_normalize_df_applies_epoch_ms_to_unaggregated_columns() -> None:
+ """``epoch_ms`` is a separate conversion branch from ``epoch_s``; an
+ unaggregated column declaring it must also be converted to datetimes."""
+ import pandas as pd
+ from pandas.api.types import is_datetime64_any_dtype
+
+ ts_col = MagicMock(
+ column_name="ts",
+ is_dttm=True,
+ python_date_format="epoch_ms",
+ datetime_format=None,
+ )
+ datasource = _normalize_df_datasource(ts_col)
+
+ # 2020-01-01, 2021-01-01, 2022-01-01 as epoch milliseconds
+ df = pd.DataFrame({"ts": [1577836800000, 1609459200000, 1640995200000]})
+
+ result = datasource.normalize_df(df, _raw_query_object())
+
+ assert is_datetime64_any_dtype(result["ts"])
+ assert result["ts"][0].strftime("%Y-%m-%d") == "2020-01-01"
+ assert result["ts"][2].strftime("%Y-%m-%d") == "2022-01-01"
+
+
+def test_normalize_df_handles_dict_shaped_columns() -> None:
+ """Some datasources expose columns as dicts rather than objects. The raw
+ temporal-column lookup must read is_dttm and python_date_format from either
+ shape, the same way the base-axis path already does."""
+ import pandas as pd
+ from pandas.api.types import is_datetime64_any_dtype
+
+ ts_col = {"column_name": "ts", "is_dttm": True, "python_date_format":
"epoch_s"}
+ datasource = _normalize_df_datasource(ts_col)
+
+ df = pd.DataFrame({"ts": [1577836800, 1609459200]})
+
+ result = datasource.normalize_df(df, _raw_query_object())
+
+ assert is_datetime64_any_dtype(result["ts"])
+ assert result["ts"][0].strftime("%Y-%m-%d") == "2020-01-01"
+
+
+def test_normalize_df_leaves_unconfigured_integer_dttm_columns_untouched() ->
None:
+ """A column flagged temporal but with no ``python_date_format`` must not be
+ coerced: a plain integer column would otherwise be reinterpreted as
+ nanoseconds since the epoch (1577836800 -> 1970-01-01 00:00:01.5...). Only
+ columns that declare a format are normalized."""
+ import pandas as pd
+ from pandas.api.types import is_datetime64_any_dtype
+
+ int_col = MagicMock(
+ column_name="ts",
+ is_dttm=True,
+ python_date_format=None,
+ datetime_format=None,
+ )
+ datasource = _normalize_df_datasource(int_col)
+
+ df = pd.DataFrame({"ts": [1577836800, 1609459200, 1640995200]})
+
+ result = datasource.normalize_df(df, _raw_query_object())
+
+ assert not is_datetime64_any_dtype(result["ts"])
+ assert result["ts"].tolist() == [1577836800, 1609459200, 1640995200]
+
+
+def test_normalize_df_without_get_column_is_a_noop() -> None:
+ """Not every datasource exposes ``get_column``; for those, the temporal
+ lookup must short-circuit instead of raising, leaving the data
untouched."""
+ import pandas as pd
+ from pandas.api.types import is_datetime64_any_dtype
+
+ from superset.models.helpers import ExploreMixin
+
+ class _NoGetColumnDatasource:
+ offset = 0
+ enforce_numerical_metrics = False
+ columns: list[object] = []
Review Comment:
Added a docstring to the test class in 77fed83.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
