Merge pull request #905 from davidhassell/import-speed-up

Reduce the time taken to import `cf`

Author: davidhassell

Changelog.rst

@@ -1,3 +1,13 @@
+Version NEXTVERSION
+-------------------
+
+**2025-12-??**
+
+* Reduce the time taken to import `cf`
+  (https://github.com/NCAS-CMS/cf-python/issues/902)
+
+----
+
 Version 3.18.2
 --------------
 

cf/__init__.py

@@ -80,167 +80,26 @@
 
 """
 
-__date__ = "2025-10-16"
-__version__ = "3.18.2"
+import cfdm
 
-_requires = (
-    "numpy",
-    "netCDF4",
-    "cftime",
-    "cfunits",
-    "cfdm",
-    "psutil",
-    "dask",
-    "packaging",
-    "scipy",
-    "distributed",
-)
-x = ", ".join(_requires)
-_error0 = f"cf v{__version__} requires the modules {x}. "
-
-import importlib.util
-from platform import python_version
-
-_found_esmpy = bool(importlib.util.find_spec("esmpy"))
-
-try:
-    import packaging
-    from packaging.version import Version
-except ImportError as error1:
-    raise ImportError(_error0 + str(error1))
-else:
-    _minimum_vn = "20.0"
-    if Version(packaging.__version__) < Version(_minimum_vn):
-        raise RuntimeError(
-            f"Bad packaging version: cf requires packaging>={_minimum_vn}. "
-            f"Got {packaging.__version__} at {packaging.__file__}"
-        )
-
-try:
-    import cfdm
-except ImportError as error1:
-    raise ImportError(_error0 + str(error1))
-else:
-    # Check the version of cfdm
-    _minimum_vn = "1.12.3.1"
-    _maximum_vn = "1.12.4.0"
-    _cfdm_version = Version(cfdm.__version__)
-    if _cfdm_version < Version(_minimum_vn) or _cfdm_version >= Version(
-        _maximum_vn
-    ):
-        raise RuntimeError(
-            "Bad cfdm version: cf requires "
-            f"{_minimum_vn}<=cfdm<{_maximum_vn}. "
-            f"Got {_cfdm_version} at {cfdm.__file__}"
-        )
+from packaging.version import Version
 
+
+__date__ = "2025-10-16"
+__version__ = "3.18.2"
 __cf_version__ = cfdm.__cf_version__
 __Conventions__ = f"CF-{__cf_version__}"
 
-try:
-    import netCDF4
-except ImportError as error1:
-    raise ImportError(_error0 + str(error1))
-else:
-    _set_vn = "1.7.2"
-    if Version(netCDF4.__version__) != Version(_set_vn):
-        raise RuntimeError(
-            "Bad netCDF4 version: cf requires "
-            f"netCDF4=={_set_vn}. "
-            f"Got {netCDF4.__version__} at {netCDF4.__file__}"
-        )
-
-try:
-    import numpy as np
-except ImportError as error1:
-    raise ImportError(_error0 + str(error1))
-else:
-    _minimum_vn = "2.0.0"
-    if Version(np.__version__) < Version(_minimum_vn):
-        raise RuntimeError(
-            f"Bad numpy version: cf requires numpy>={_minimum_vn} "
-            f"Got {np.__version__} at {np.__file__}"
-        )
-
-try:
-    import cftime
-except ImportError as error1:
-    raise ImportError(_error0 + str(error1))
-else:
-    _minimum_vn = "1.6.4"
-    if Version(cftime.__version__) < Version(_minimum_vn):
-        raise RuntimeError(
-            f"Bad cftime version: cf requires cftime>={_minimum_vn}. "
-            f"Got {cftime.__version__} at {cftime.__file__}"
-        )
-
-try:
-    import cfunits
-except ImportError as error1:
-    raise ImportError(_error0 + str(error1))
-else:
-    _minimum_vn = "3.3.7"
-    if Version(cfunits.__version__) < Version(_minimum_vn):
-        raise RuntimeError(
-            f"Bad cfunits version: cf requires cfunits>={_minimum_vn}. "
-            f"Got {cfunits.__version__} at {cfunits.__file__}"
-        )
-
-try:
-    import psutil
-except ImportError as error1:
-    raise ImportError(_error0 + str(error1))
-else:
-    _minimum_vn = "0.6.0"
-    if Version(psutil.__version__) < Version(_minimum_vn):
-        raise RuntimeError(
-            f"Bad psutil version: cf requires psutil>={_minimum_vn}. "
-            f"Got {psutil.__version__} at {psutil.__file__}"
-        )
-
-# Check the version of dask
-try:
-    import dask
-except ImportError as error1:
-    raise ImportError(_error0 + str(error1))
-else:
-    _minimum_vn = "2025.5.1"
-    if Version(dask.__version__) < Version(_minimum_vn):
-        raise RuntimeError(
-            f"Bad dask version: cf requires dask>={_minimum_vn}. "
-            f"Got {dask.__version__} at {dask.__file__}"
-        )
-
-try:
-    import distributed
-except ImportError as error1:
-    raise ImportError(_error0 + str(error1))
-else:
-    _minimum_vn = "2025.5.1"
-    if Version(distributed.__version__) < Version(_minimum_vn):
-        raise RuntimeError(
-            "Bad distributed version: cf requires "
-            f"distributed>={_minimum_vn}. "
-            f"Got {distributed.__version__} at {distributed.__file__}"
-        )
-
-try:
-    import scipy
-except ImportError as error1:
-    raise ImportError(_error0 + str(error1))
-else:
-    _minimum_vn = "1.10.0"
-    if Version(scipy.__version__) < Version(_minimum_vn):
-        raise RuntimeError(
-            f"Bad scipy version: cf requires scipy>={_minimum_vn}. "
-            f"Got {scipy.__version__} at {scipy.__file__}"
-        )
-
-_minimum_vn = "3.10.0"
-if Version(python_version()) < Version(_minimum_vn):
+# Check the version of cfdm (this is worth doing because of the very
+# tight coupling between cf and cfdm, and the risk of bad things
+# happening at run time if the versions are mismatched).
+_minimum_vn = "1.12.3.1"
+_maximum_vn = "1.12.4.0"
+_cfdm_vn = Version(cfdm.__version__)
+if _cfdm_vn < Version(_minimum_vn) or _cfdm_vn >= Version(_maximum_vn):
     raise RuntimeError(
-        f"Bad python version: cf requires python>={_minimum_vn}. "
-        f"Got {python_version()}"
+        f"cf v{__version__} requires {_minimum_vn}<=cfdm<{_maximum_vn}. "
+        f"Got {_cfdm_vn} at {cfdm.__file__}"
     )
 
 del _minimum_vn, _maximum_vn

cf/aggregate.py

@@ -6,7 +6,6 @@
 
 import numpy as np
 from cfdm import is_log_level_debug, is_log_level_detail, is_log_level_info
-from dask.base import tokenize
 
 from .auxiliarycoordinate import AuxiliaryCoordinate
 from .data import Data
@@ -1986,6 +1985,8 @@ def tokenise_cell_conditions(self, cell_conditions):
         ('ce9a05dd6ec76c6a6d171b0c055f3127', '8e0216a9a17a20b6620c6502bb45dec9')
 
         """
+        from dask.base import tokenize
+
         out = []
         for x in cell_conditions:
             if x is None:
@@ -4056,6 +4057,8 @@ def _get_hfl(
         hash_value = d.get_deterministic_name()
     except ValueError:
         # Slow
+        from dask.base import tokenize
+
         hash_value = tokenize(d.compute())
 
     if hash_value in hash_map:

cf/cellmethod.py

@@ -1,5 +1,4 @@
 import logging
-import re
 from ast import literal_eval as ast_literal_eval
 
 import cfdm
@@ -89,6 +88,8 @@ def create(cls, cell_methods_string=None):
         >>> c = CellMethod.create('lat: mean (interval: 1 hour)')
 
         """
+        import re
+
         incorrect_interval = "Cell method interval is incorrectly formatted"
 
         out = []

cf/cfdatetime.py

@@ -1,7 +1,6 @@
 import datetime
 from functools import partial
 
-import cftime
 import numpy as np
 
 from .functions import _DEPRECATION_ERROR_CLASS
@@ -10,17 +9,10 @@
 default_calendar = "gregorian"
 
 # --------------------------------------------------------------------
-# Mapping of CF calendars to cftime date-time objects
+# Mapping of CF calendars to cftime date-time objects (that gets
+# populated in the `dt` function).
 # --------------------------------------------------------------------
-_datetime_object = {
-    ("",): partial(cftime.datetime, calendar=""),
-    (None, "gregorian", "standard", "none"): cftime.DatetimeGregorian,
-    ("proleptic_gregorian",): cftime.DatetimeProlepticGregorian,
-    ("360_day",): cftime.Datetime360Day,
-    ("noleap", "365_day"): cftime.DatetimeNoLeap,
-    ("all_leap", "366_day"): cftime.DatetimeAllLeap,
-    ("julian",): cftime.DatetimeJulian,
-}
+_datetime_object = {}
 
 canonical_calendar = {
     None: "standard",
@@ -40,7 +32,7 @@
 _calendar_map = {None: "gregorian"}
 
 
-class Datetime(cftime.datetime):
+class Datetime:
     """A date-time object which supports CF calendars.
 
     Deprecated at version 3.0.0. Use function 'cf.dt' to create date-
@@ -134,6 +126,26 @@ def dt(
     (2003, 4, 5, 12, 30, 15)
 
     """
+    import cftime
+
+    if not _datetime_object:
+        _datetime_object.update(
+            {
+                ("",): partial(cftime.datetime, calendar=""),
+                (
+                    None,
+                    "gregorian",
+                    "standard",
+                    "none",
+                ): cftime.DatetimeGregorian,
+                ("proleptic_gregorian",): cftime.DatetimeProlepticGregorian,
+                ("360_day",): cftime.Datetime360Day,
+                ("noleap", "365_day"): cftime.DatetimeNoLeap,
+                ("all_leap", "366_day"): cftime.DatetimeAllLeap,
+                ("julian",): cftime.DatetimeJulian,
+            }
+        )
+
     if isinstance(arg, str):
         (year, month, day, hour, minute, second, microsecond) = st2elements(
             arg
@@ -161,11 +173,6 @@ def dt(
     else:
         year = arg
 
-    #    calendar=_calendar_map.get(calendar, calendar)
-    #
-    #    return cftime.datetime(year, month, day, hour, minute, second,
-    #                           microsecond, calendar=calendar)
-
     for calendars, datetime_cls in _datetime_object.items():
         if calendar in calendars:
             return datetime_cls(
@@ -354,6 +361,8 @@ def st2datetime(date_string, calendar=None):
         `cftime.datetime`
 
     """
+    import cftime
+
     if date_string.count("-") != 2:
         raise ValueError(
             "Input date-time string must contain at least a year, a month "
@@ -388,6 +397,8 @@ def st2elements(date_string):
         `tuple`
 
     """
+    import cftime
+
     if date_string.count("-") != 2:
         raise ValueError(
             "Input date-time string must contain at least a year, a month "
@@ -450,6 +461,8 @@ def rt2dt(array, units_in, units_out=None, dummy1=None):
         # mask
         return np.ma.masked_all((), dtype=object)
 
+    import cftime
+
     units = units_in.units
     calendar = getattr(units_in, "calendar", "standard")
 
@@ -510,6 +523,8 @@ def dt2rt(array, units_in, units_out, dummy1=None):
     [-- 685.5]
 
     """
+    import cftime
+
     isscalar = not np.ndim(array)
 
     array = cftime.date2num(
@@ -545,8 +560,9 @@ def st2rt(array, units_in, units_out, dummy1=None):
             An array of floats with the same shape as *array*.
 
     """
+    import cftime
+
     array = st2dt(array, units_in)
-    #    array = units_out._utime.date2num(array)
     array = cftime.date2num(
         array, units=units_out.units, calendar=units_out._utime.calendar
     )