Support pandas v3

CHANGELOG.rst

@@ -18,19 +18,21 @@ Fixes
 * In `nc_specs`, set `dim_names_nc` in the order expected by Raven (x, y, t). Previously, we only made sure that `time` was the last dimension, but did not ensure x and y were in the right order. (PR #533)
 * Adjusted the `Perform_a_climate_change_impact_study_on_a_watershed.ipynb` notebook to reduce the number of years in the simulation to speed up execution time. (PR #535)
 * Adjusted a broken test that was overlooked in the previous release (from changes in PR #513). (PR #535)
+* Adapted code base to the `pandas` v3.0 API. (Issue #570, PR #572)
 
 Internal changes
 ^^^^^^^^^^^^^^^^
-* Updated the cookiecutter template to the latest version (#548):
+* Updated the cookiecutter template to the latest version (PR #548):
   * Updated the Contributor Covenant Agreement to v3.0.
   * Added a `CITATION.cff` file.
   * Removed `black`, `blackdoc`, and `isort`, as well as their configurations.
   * Updated `pyproject.toml` to be `PEP 639 <https://peps.python.org/pep-0639>`_-compliant.
 * Pinned `pydantic` below v2.12 due to breaking changes in their API. (PR #548)
   * Unpinned `pydantic` as newer 2.12 patch releases appear to have addressed regressions. (PR #559).
 * Pinned `pydap` >=3.5.6 and `h5netcdf` >=1.5.0 to ensure modern versions with better `xarray` support are installed by default. (PR #559).
-* Updated the cookiecutter template to the latest version (#569):
+* Updated the cookiecutter template to the latest version (PR #569):
   * Added a workflow for automatically accepting and merging periodic updates from Dependabot affecting CI dependencies.
+* Added a `pytest` fixture to perform a teardown of changes performed within the installed `ravenpy` source location. (PR #572).
 
 .. _changes_0.19.1:
 

environment-dev.yml

@@ -20,7 +20,7 @@ dependencies:
   - netcdf4 >=1.7.2
   - numpy >=1.25.0
   - owslib >=0.29.1
-  - pandas >=2.2.0,<3.0
+  - pandas >=2.2.0
   - pint >=0.24.4
   - pydantic >=2.11
   - pydap >=3.5.6

environment-docs.yml

@@ -31,7 +31,7 @@ dependencies:
   - netCDF4 >=1.7.2
   - numpy >=1.24.0
   - notebook
-  - pandas >=2.2,<3.0
+  - pandas >=2.2
   - pydantic >=2.11,<2.12
   - pymetalink >=6.5.2
   - s3fs

pyproject.toml

@@ -47,7 +47,7 @@ dependencies = [
   "matplotlib >=3.6.0",
   "numpy >=1.25.0",
   "owslib >=0.29.1",
-  "pandas >=2.2.0,<3.0",
+  "pandas >=2.2.0",
   "pint >=0.24.4",
   "pydantic >=2.11",
   "pydap >=3.5.6",

src/ravenpy/extractors/routing_product.py

@@ -405,7 +405,7 @@ def extract(self) -> dict:
         # Read routing data
 
         # WGS 84 / North Pole LAEA Canada
-        self._routing_data = self._routing_data.to_crs(epsg=GridWeightExtractor.CRS_CAEA)
+        df = self._routing_data.to_crs(epsg=GridWeightExtractor.CRS_CAEA)
 
         def keep_only_valid_downsubid_and_obs_nm(g):
             """
@@ -436,13 +436,7 @@ def keep_only_valid_downsubid_and_obs_nm(g):
             return row
 
         # Remove duplicate HRU_IDs while making sure that we keep relevant DowSubId and Obs_NM values
-        # FIXME: DeprecationWarning: DataFrameGroupBy.apply operated on the grouping columns.
-        #  This behavior is deprecated, and in a future version of pandas the grouping columns will be
-        #  excluded from the operation. Either pass `include_groups=False` to exclude the groupings or
-        #  explicitly select the grouping columns after groupby to silence this warning.
-        self._routing_data = self._routing_data.groupby(self._routing_id_field, group_keys=False).apply(
-            keep_only_valid_downsubid_and_obs_nm, include_groups=True
-        )
+        self._routing_data = df.groupby(self._routing_id_field, group_keys=False)[df.columns.to_list()].apply(keep_only_valid_downsubid_and_obs_nm)
 
         # Make sure those are ints
         self._routing_data.SubId = self._routing_data.SubId.astype(int)

src/ravenpy/utilities/regionalization.py

@@ -29,9 +29,9 @@ def regionalize(
     config: Config,
     method: str,
     nash: pd.Series,
-    params: pd.DataFrame = None,
-    props: pd.DataFrame = None,
-    target_props: Union[pd.Series, dict] = None,
+    params: Optional[pd.DataFrame] = None,
+    props: Optional[pd.DataFrame] = None,
+    target_props: Optional[Union[pd.Series, dict]] = None,
     size: int = 5,
     min_NSE: float = 0.6,  # noqa: N803
     workdir: Optional[Union[str, Path]] = None,
@@ -199,13 +199,13 @@ def read_gauged_properties(properties) -> pd.DataFrame:
     return proptable[properties]
 
 
-def read_gauged_params(model):
+def read_gauged_params(model) -> tuple[pd.Series, pd.DataFrame]:
     """
     Return table of NASH-Sutcliffe Efficiency values and model parameters for North American catchments.
 
     Returns
     -------
-    pd.DataFrame
+    pd.Series
         Nash-Sutcliffe Efficiency keyed by catchment ID.
     pd.DataFrame
         Model parameters keyed by catchment ID.
@@ -264,6 +264,8 @@ def similarity(gauged: pd.DataFrame, ungauged: pd.DataFrame, kind: str = "ptp")
         spread = stats.loc["std"]
     elif kind == "iqr":
         spread = stats.loc["75%"] - stats.loc["25%"]
+    else:
+        raise NotImplementedError("'kind' not in ['ptp', 'std', 'iqr']")
 
     d = ungauged.values - gauged.values
     n = np.abs(d) / spread.values
@@ -347,10 +349,10 @@ def IDW(qsims: xr.DataArray, dist: pd.Series) -> xr.DataArray:  # noqa: N802
     weights = xr.DataArray(1.0 / dist, dims="members", coords={"members": qsims.members})
 
     # Make weights sum to one
-    weights /= weights.sum(axis=0)
+    one_sum_weights = weights / weights.sum(axis=0)
 
     # Calculate weighted average.
-    out = qsims.dot(weights)
+    out = qsims.dot(one_sum_weights)
     out.name = qsims.name
     out.attrs = qsims.attrs
     return out

tests/test_external_dataset_access.py

@@ -1,4 +1,5 @@
 import datetime as dt
+import importlib.util
 import urllib.error
 from pathlib import Path
 
@@ -30,6 +31,20 @@ def test_get_ECCC_dataset(self):
         assert n_hours <= 36
 
 
+@pytest.fixture(scope="function")
+def remote_access_teardown(request):
+    def _teardown():
+        ravenpy_location = Path(importlib.util.find_spec("ravenpy").origin).parent
+        testing = ravenpy_location / "testing"
+
+        if testing.joinpath("main").is_dir():
+            for f in testing.joinpath("main").iterdir():
+                f.unlink()
+            testing.joinpath("main").rmdir()
+
+    request.addfinalizer(_teardown)
+
+
 @pytest.mark.online
 class TestRemoteFileAccess:
     dap_url = "http://test.opendap.org:80/opendap/data/nc/"
@@ -41,7 +56,7 @@ class TestRemoteFileAccess:
         reason="Get file is API rate limited",
         strict=False,
     )
-    def test_get_file_default_cache(self):
+    def test_get_file_default_cache(self, remote_access_teardown):  # noqa: F841
         file = yangtze(branch=self.branch).fetch(fname="ostrich-hbvec/raven-hbvec-salmon.rvi")
 
         assert Path(default_testdata_cache).exists()
@@ -50,10 +65,7 @@ def test_get_file_default_cache(self):
             header = f.read()
             assert ":FileType          rvi ASCII Raven 2.8.2" in header
 
-    def test_open_dataset(
-        self,
-        tmp_path,
-    ):
+    def test_open_dataset(self, tmp_path, remote_access_teardown):  # noqa: F841
         cache_dir = tmp_path / "yangtze_cache"
         ds = open_dataset(
             name="raven-gr4j-cemaneige/Salmon-River-Near-Prince-George_meteo_daily.nc",