Temporarily revert star annotation in call signatures - remove all internal xarray imports

.pre-commit-config.yaml

@@ -34,19 +34,19 @@ repos:
       - id: toml-sort-fix
         exclude: '.pylintrc.toml'
   - repo: https://github.com/adrienverge/yamllint.git
-    rev: v1.35.1
+    rev: v1.36.2
     hooks:
       - id: yamllint
         args: [ '--config-file=.yamllint.yaml' ]
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.9.9
+    rev: v0.11.2
     hooks:
       - id: ruff
         args: [ '--fix', '--show-fixes' ]
       - id: ruff-format
         exclude: '(src/xclim/indices/__init__.py|docs/installation.rst)'
   - repo: https://github.com/pylint-dev/pylint
-    rev: v3.3.4
+    rev: v3.3.6
     hooks:
       - id: pylint
         args: [ '--rcfile=.pylintrc.toml', '--errors-only', '--jobs=0', '--disable=import-error' ]
@@ -105,7 +105,7 @@ repos:
     hooks:
       - id: gitleaks
   - repo: https://github.com/python-jsonschema/check-jsonschema
-    rev: 0.31.2
+    rev: 0.31.3
     hooks:
       - id: check-github-workflows
       - id: check-readthedocs

.zenodo.json

@@ -163,6 +163,11 @@
       "name": "Wong, Jack Kit-tai",
       "affiliation": "University of Toronto, Toronto, Ontario, Canada",
       "orcid": "0000-0003-1408-1502"
+    },
+    {
+      "name": "de Bruijn, Jens",
+      "affiliation": "Vrije Universiteit, Amsterdam, The Netherlands",
+      "orcid": "0000-0003-3961-6382"
     }
   ],
   "keywords": [

AUTHORS.rst

@@ -50,3 +50,4 @@ Contributors
 * Sebastian Lehner <[email protected]> `@seblehner <https://github.com/seblehner>`_
 * Baptiste Hamon <[email protected]> `@baptistehamon <https://github.com/baptistehamon>`_
 * Jack Kit-tai Wong <[email protected]> `@jack-ktw <https://github.com/jack-ktw>`_
+* Jens de Bruijn <[email protected]> `@jensdebruijn <https://github.com/jensdebruijn>`_

CHANGELOG.rst

@@ -4,12 +4,15 @@ Changelog
 
 v0.56.0 (unreleased)
 --------------------
-Contributors to this version: Trevor James Smith (:user:`Zeitsperre`), Hui-Min Wang (:user:`Hem-W`), Jack Kit-tai Wong(:user:`jack-ktw`), Adrien Lamarche (:user:`LamAdr`), Éric Dupuis (:user:`coxipi`).
+Contributors to this version: Trevor James Smith (:user:`Zeitsperre`), Hui-Min Wang (:user:`Hem-W`), Jack Kit-tai Wong(:user:`jack-ktw`), Adrien Lamarche (:user:`LamAdr`), Éric Dupuis (:user:`coxipi`), Jens de Bruijn (:user:`jensdebruijn`).
 
 Bug fixes
 ^^^^^^^^^
 * Fix installation instructions in the Contributing guide (:issue:`2088`, :pull:`2089`).
 * Fixed parameter order in typing.cast() causing intermittent errors in solar_zenith_angle calculation. (:issue:`2097`, :pull:`2098`)
+* `xclim` now uses directly `operator` instead of using `xarray`'s derived `get_op` function. A refactoring in `xarray` had changed the position of `get_op` which caused a bug.  (:issue:`2113`, :pull:`2114`).
+    + All other uses of `xarray`'s internal API were also removed (:pull:`2116`).
+* Fixed an issue with star-annotated call signatures to maintain Python 3.10 compatibility. (:pull:`2116`).
 
 Breaking changes
 ^^^^^^^^^^^^^^^^
@@ -26,7 +29,8 @@ Internal changes
     * Line endings in files now must be `Unix`-compatible (`LF`).
 * The `blackdoc` pre-commit hook now only examines `.rst` and `.md` files. (:pull:`2083`).
 * The `xclim` documentation now has a ``support`` page for detailing the project's usage and version support policies. (:pull:`2100`).
-* The indicator `heat_wave_index` now uses `hot_spell_total_length` index. The `heat_wave_index` index is identitical to `hot_spell_total_length` and will be dropped in future versions. (:issue:`2031`, :pull:`2102`).
+* The indicator `heat_wave_index` now uses `hot_spell_total_length` index. The `heat_wave_index` index is identical to `hot_spell_total_length` and will be dropped in future versions. (:issue:`2031`, :pull:`2102`).
+* Updated pre-commit hooks to their latest versions. (:pull:`2116`).
 
 New indicators
 ^^^^^^^^^^^^^^

src/xclim/core/bootstrapping.py

@@ -11,7 +11,7 @@
 import numpy as np
 import xarray
 from boltons.funcutils import wraps
-from xarray.core.dataarray import DataArray
+from xarray import DataArray
 
 import xclim.core.utils
 from xclim.core.calendar import parse_offset, percentile_doy

src/xclim/core/calendar.py

@@ -15,15 +15,16 @@
 import numpy as np
 import pandas as pd
 import xarray as xr
-from xarray.coding.cftime_offsets import to_cftime_datetime
-from xarray.coding.cftimeindex import CFTimeIndex
-from xarray.core import dtypes
-from xarray.core.resample import DataArrayResample, DatasetResample
+from packaging.version import Version
+from xarray import CFTimeIndex
 
 from xclim.core._types import DayOfYearStr
 from xclim.core.formatting import update_xclim_history
 from xclim.core.utils import uses_dask
 
+XR2409 = Version(xr.__version__) >= Version("2024.09")
+
+
 __all__ = [
     "DayOfYearStr",
     "adjust_doy_calendar",
@@ -51,6 +52,23 @@
     "within_bnds_doy",
 ]
 
+
+_MONTH_ABBREVIATIONS = {
+    1: "JAN",
+    2: "FEB",
+    3: "MAR",
+    4: "APR",
+    5: "MAY",
+    6: "JUN",
+    7: "JUL",
+    8: "AUG",
+    9: "SEP",
+    10: "OCT",
+    11: "NOV",
+    12: "DEC",
+}
+
+
 # Maximum day of year in each calendar.
 max_doy = {
     "standard": 366,
@@ -208,6 +226,20 @@ def _convert_doy_date(doy: int, year: int, src, tgt):
     return float(same_date.dayofyr) + fracpart
 
 
+# Copied from xarray.coding.calendar_ops
+def _is_leap_year(years, calendar):
+    func = np.vectorize(cftime.is_leap_year)
+    return func(years, calendar=calendar)
+
+
+# Copied from xarray.coding.calendar_ops
+def _days_in_year(years, calendar):
+    """The number of days in the year according to given calendar."""
+    if calendar == "360_day":
+        return xr.full_like(years, 360)
+    return _is_leap_year(years, calendar).astype(int) + 365
+
+
 def convert_doy(
     source: xr.DataArray | xr.Dataset,
     target_cal: str,
@@ -272,7 +304,7 @@ def convert_doy(
             max_doy_src = max_doy[source_cal]
         else:
             max_doy_src = xr.apply_ufunc(
-                xr.coding.calendar_ops._days_in_year,
+                _days_in_year,
                 year_of_the_doy,
                 vectorize=True,
                 dask="parallelized",
@@ -282,7 +314,7 @@ def convert_doy(
             max_doy_tgt = max_doy[target_cal]
         else:
             max_doy_tgt = xr.apply_ufunc(
-                xr.coding.calendar_ops._days_in_year,
+                _days_in_year,
                 year_of_the_doy,
                 vectorize=True,
                 dask="parallelized",
@@ -736,7 +768,7 @@ def resample_doy(doy: xr.DataArray, arr: xr.DataArray | xr.Dataset) -> xr.DataAr
 
 
 def time_bnds(  # noqa: C901
-    time: (xr.DataArray | xr.Dataset | CFTimeIndex | pd.DatetimeIndex | DataArrayResample | DatasetResample),
+    time: (xr.DataArray | xr.Dataset | CFTimeIndex | pd.DatetimeIndex),
     freq: str | None = None,
     precision: str | None = None,
 ):
@@ -778,9 +810,10 @@ def time_bnds(  # noqa: C901
     """
     if isinstance(time, xr.DataArray | xr.Dataset):
         time = time.indexes[time.name]
-    elif isinstance(time, DataArrayResample | DatasetResample):
+    # elif isinstance(time, DataArrayResample | DatasetResample):
+    elif hasattr(time, "groupers"):
         for grouper in time.groupers:
-            if isinstance(grouper.grouper, xr.groupers.TimeResampler):
+            if "time" in grouper.codes.dims:
                 datetime = grouper.unique_coord.data
                 freq = freq or grouper.grouper.freq
                 if datetime.dtype == "O":
@@ -922,11 +955,9 @@ def _doy_days_since_doys(
         Number of days (maximum doy) for the year of each value in base.
     """
     calendar = get_calendar(base)
-
     base_doy = base.dt.dayofyear
-
     doy_max = xr.apply_ufunc(
-        xr.coding.calendar_ops._days_in_year,
+        _days_in_year,
         base.dt.year,
         vectorize=True,
         kwargs={"calendar": calendar},
@@ -1309,8 +1340,8 @@ def select_time(
 
         # Get doy of date, this is now safe because the calendar is uniform.
         doys = _get_doys(
-            to_cftime_datetime(f"2000-{start}", calendar).dayofyr,
-            to_cftime_datetime(f"2000-{end}", calendar).dayofyr,
+            cftime.datetime.strptime(f"2000-{start}", "%Y-%m-%d", calendar=calendar).dayofyr,
+            cftime.datetime.strptime(f"2000-{end}", "%Y-%m-%d", calendar=calendar).dayofyr,
             include_bounds,
         )
         mask = time.time.dt.dayofyear.isin(doys)
@@ -1349,7 +1380,7 @@ def stack_periods(
     dim: str = "period",
     start: str = "1970-01-01",
     align_days: bool = True,
-    pad_value=dtypes.NA,
+    pad_value="<NA>",
 ):
     """
     Construct a multi-period array.
@@ -1527,11 +1558,12 @@ def stack_periods(
     # The "fake" axis that all periods share
     fake_time = xr.date_range(start, periods=longest, freq=srcfreq, calendar=cal, use_cftime=use_cftime)
     # Slice and concat along new dim. We drop the index and add a new one so that xarray can concat them together.
+    kwargs = {"fill_value": pad_value} if pad_value != "<NA>" else {}
     out = xr.concat(
         [da.isel(time=slc).drop_vars("time").assign_coords(time=np.arange(slc.stop - slc.start)) for slc in periods],
         dim,
         join="outer",
-        fill_value=pad_value,
+        **kwargs,
     )
     out = out.assign_coords(
         time=(("time",), fake_time, da.time.attrs.copy()),

src/xclim/core/formatting.py

@@ -17,6 +17,7 @@
 from inspect import _empty, signature  # noqa
 from typing import Any
 
+import pandas as pd
 import xarray as xr
 from boltons.funcutils import wraps
 
@@ -79,7 +80,7 @@ def __init__(
         self.modifiers = modifiers
         self.mapping = mapping
 
-    def format(self, format_string: str, /, *args: Any, **kwargs: dict) -> str:
+    def format(self, format_string: str, /, *args, **kwargs: dict) -> str:
         r"""
         Format a string.
 
@@ -338,7 +339,7 @@ def _parse_returns(section):
 
 def merge_attributes(
     attribute: str,
-    *inputs_list: xr.DataArray | xr.Dataset,
+    *inputs_list,  # : xr.DataArray | xr.Dataset
     new_line: str = "\n",
     missing_str: str | None = None,
     **inputs_kws: xr.DataArray | xr.Dataset,
@@ -390,7 +391,7 @@ def merge_attributes(
 
 def update_history(
     hist_str: str,
-    *inputs_list: xr.DataArray | xr.Dataset,
+    *inputs_list,  # : xr.DataArray | xr.Dataset,
     new_name: str | None = None,
     **inputs_kws: xr.DataArray | xr.Dataset,
 ) -> str:
@@ -770,3 +771,28 @@ def get_percentile_metadata(data: xr.DataArray, prefix: str) -> dict[str, str]:
         f"{prefix}_window": data.attrs.get("window", "<unknown window>"),
         f"{prefix}_period": clim_bounds,
     }
+
+
+# Adapted from xarray.structure.merge_attrs
+def _merge_attrs_drop_conflicts(*objs):
+    """Merge attributes from different xarray objects, dropping any attributes that conflict."""
+    out = {}
+    dropped = set()
+    for obj in objs:
+        attrs = obj.attrs
+        out.update({key: value for key, value in attrs.items() if key not in out and key not in dropped})
+        out = {key: value for key, value in out.items() if key not in attrs or _equivalent_attrs(attrs[key], value)}
+        dropped |= {key for key in attrs if key not in out}
+    return out
+
+
+# Adapted from xarray.core.utils.equivalent
+def _equivalent_attrs(first, second) -> bool:
+    """Return whether two attributes are identical or not."""
+    if first is second:
+        return True
+    if isinstance(first, list) or isinstance(second, list):
+        if len(first) != len(second):
+            return False
+        return all(_equivalent_attrs(f, s) for f, s in zip(first, second, strict=False))
+    return (first == second) or (pd.isnull(first) and pd.isnull(second))

src/xclim/core/indicator.py

@@ -136,6 +136,7 @@
 from xclim.core.cfchecks import cfcheck_from_name
 from xclim.core.formatting import (
     AttrFormatter,
+    _merge_attrs_drop_conflicts,
     default_formatter,
     gen_call_string,
     generate_indicator_docstring,
@@ -860,9 +861,9 @@ def __call__(self, *args, **kwds):
         das, params, dsattrs = self._parse_variables_from_call(args, kwds)
 
         if OPTIONS[KEEP_ATTRS] is True or (
-            OPTIONS[KEEP_ATTRS] == "xarray" and xarray.core.options._get_keep_attrs(False)
+            OPTIONS[KEEP_ATTRS] == "xarray" and xarray.get_options()["keep_attrs"] is True
         ):
-            out_attrs = xarray.core.merge.merge_attrs([da.attrs for da in das.values()], "drop_conflicts")
+            out_attrs = _merge_attrs_drop_conflicts(*das.values())
             out_attrs.pop("units", None)
         else:
             out_attrs = {}
@@ -916,7 +917,7 @@ def __call__(self, *args, **kwds):
         if OPTIONS[AS_DATASET]:
             out = Dataset({o.name: o for o in outs})
             if OPTIONS[KEEP_ATTRS] is True or (
-                OPTIONS[KEEP_ATTRS] == "xarray" and xarray.core.options._get_keep_attrs(False)
+                OPTIONS[KEEP_ATTRS] == "xarray" and xarray.get_options()["keep_attrs"] is True
             ):
                 out.attrs.update(dsattrs)
             out.attrs["history"] = update_history(

src/xclim/core/locales.py

@@ -147,7 +147,7 @@ def get_local_dict(locale: str | Sequence[str] | tuple[str, dict]) -> tuple[str,
 
 def get_local_attrs(
     indicator: str | Sequence[str],
-    *locales: str | Sequence[str] | tuple[str, dict],
+    *locales,  # : str | Sequence[str] | tuple[str, dict],
     names: Sequence[str] | None = None,
     append_locale_name: bool = True,
 ) -> dict:

src/xclim/core/missing.py

@@ -107,7 +107,8 @@ def expected_count(
     time = xr.DataArray(time.values, dims=("time",), coords={"time": time.values}, name="time")
 
     if freq:
-        resamp = time.resample(time=freq).first()
+        # We only want the resulting time index, the actual resampling method is not important.
+        resamp = time.resample(time=freq).count()
         resamp_time = resamp.indexes["time"]
         _, _, is_start, _ = parse_offset(freq)
         if is_start:

src/xclim/core/options.py

@@ -8,6 +8,7 @@
 from __future__ import annotations
 
 from collections.abc import Callable
+from copy import deepcopy
 from inspect import signature
 
 from boltons.funcutils import wraps
@@ -78,7 +79,7 @@ def _valid_missing_options(mopts):
 
 def _set_missing_options(mopts):
     for meth, opts in mopts.items():
-        OPTIONS[MISSING_OPTIONS][meth].update(opts)
+        OPTIONS[MISSING_OPTIONS][meth].update(**opts)
 
 
 def _set_metadata_locales(locales):
@@ -275,7 +276,7 @@ def __init__(self, **kwargs):
             if k in _VALIDATORS and not _VALIDATORS[k](v):
                 raise ValueError(f"option {k!r} given an invalid value: {v!r}")
 
-            self.old[k] = OPTIONS[k]
+            self.old[k] = deepcopy(OPTIONS[k])
 
         self._update(kwargs)