fix: Changing update function wrapper to not wrap provided parameters.

Refs: #12

Author: Thomas Zilio

zcollection/collection/__init__.py

@@ -556,24 +556,21 @@ def update(
                                                     lock=True))
 
         local_func: WrappedPartitionCallable = _wrap_update_func(
-            *args,
             delayed=delayed,
             func=func,
             fs=self.fs,
             immutable=self._immutable,
+            selected_variables=selected_variables
+        ) if depth == 0 else _wrap_update_func_with_overlap(
+            delayed=delayed,
+            depth=depth,
+            dim=self.partition_properties.dim,
+            func=func,
+            fs=self.fs,
+            immutable=self._immutable,
+            selected_partitions=selected_partitions,
             selected_variables=selected_variables,
-            **kwargs) if depth == 0 else _wrap_update_func_with_overlap(
-                *args,
-                delayed=delayed,
-                depth=depth,
-                dim=self.partition_properties.dim,
-                func=func,
-                fs=self.fs,
-                immutable=self._immutable,
-                selected_partitions=selected_partitions,
-                selected_variables=selected_variables,
-                trim=trim,
-                **kwargs)
+            trim=trim)
 
         client: dask.distributed.Client = dask_utils.get_client()
 
@@ -582,7 +579,11 @@ def update(
             or dask_utils.dask_workers(client, cores_only=True))
         storage.execute_transaction(
             client, self.synchronizer,
-            client.map(local_func, tuple(batches), key=func.__name__))
+            client.map(local_func,
+                       tuple(batches),
+                       key=func.__name__,
+                       func_args=args,
+                       func_kwargs=kwargs))
         tuple(map(self.fs.invalidate_cache, selected_partitions))
 
     def drop_variable(

zcollection/collection/callable_objects.py

@@ -16,7 +16,8 @@
 
 #: Function type to load and call a callback function of type
 #: :class:`PartitionCallable`.
-WrappedPartitionCallable = Callable[[Sequence[str]], None]
+WrappedPartitionCallable = Callable[[Sequence[str], list[Any], dict[str, Any]],
+                                    None]
 
 
 #: pylint: disable=too-few-public-methods

zcollection/collection/detail.py

@@ -252,13 +252,11 @@ def calculate_slice(
 
 
 def _wrap_update_func(
-    *args,
     delayed: bool,
     func: UpdateCallable,
     fs: fsspec.AbstractFileSystem,
     immutable: str | None,
     selected_variables: Iterable[str] | None,
-    **kwargs,
 ) -> WrappedPartitionCallable:
     """Wrap an update function taking a partition's dataset as input and
     returning variable's values as a numpy array.
@@ -271,21 +269,21 @@ def _wrap_update_func(
         selected_variables: Name of the variables to load from the dataset.
             If None, all variables are loaded.
         trim: Whether to trim the overlap.
-        *args: Positional arguments to pass to the function.
-        **kwargs: Keyword arguments to pass to the function.
 
     Returns:
         The wrapped function that takes a set of dataset partitions and the
         variable name as input and returns the variable's values as a numpy
         array.
     """
 
-    def wrap_function(partitions: Iterable[str]) -> None:
+    def wrap_function(partitions: Iterable[str], func_args: list[Any],
+                      func_kwargs: dict[str, Any]) -> None:
         # Applying function for each partition's data
         for partition in partitions:
             zds: dataset.Dataset = _load_dataset(delayed, fs, immutable,
                                                  partition, selected_variables)
-            dictionary: dict[str, ArrayLike] = func(zds, *args, **kwargs)
+            dictionary: dict[str, ArrayLike] = func(zds, *func_args,
+                                                    **func_kwargs)
             tuple(
                 update_zarr_array(  # type: ignore[func-returns-value]
                     dirname=join_path(partition, varname),
@@ -297,7 +295,6 @@ def wrap_function(partitions: Iterable[str]) -> None:
 
 
 def _wrap_update_func_with_overlap(
-    *args,
     delayed: bool,
     depth: int,
     dim: str,
@@ -307,7 +304,6 @@ def _wrap_update_func_with_overlap(
     selected_partitions: Sequence[str],
     selected_variables: Iterable[str] | None,
     trim: bool,
-    **kwargs,
 ) -> WrappedPartitionCallable:
     """Wrap an update function taking a partition's dataset as input and
     returning variable's values as a numpy array.
@@ -323,8 +319,6 @@ def _wrap_update_func_with_overlap(
         selected_variables: Name of the variables to load from the dataset.
             If None, all variables are loaded.
         trim: Whether to trim the overlap.
-        *args: Positional arguments to pass to the function.
-        **kwargs: Keyword arguments to pass to the function.
 
     Returns:
         The wrapped function that takes a set of dataset partitions and the
@@ -334,7 +328,8 @@ def _wrap_update_func_with_overlap(
     if depth < 0:
         raise ValueError('Depth must be non-negative.')
 
-    def wrap_function(partitions: Sequence[str]) -> None:
+    def wrap_function(partitions: Sequence[str], func_args: list[Any],
+                      func_kwargs: dict[str, Any]) -> None:
         # Applying function for each partition's data
         for partition in partitions:
 
@@ -353,15 +348,15 @@ def wrap_function(partitions: Sequence[str]) -> None:
                 selected_variables=selected_variables)
             # pylint: enable=duplicate-code
 
-            _update_with_overlap(*args,
+            _update_with_overlap(*func_args,
                                  func=func,
                                  zds=zds,
                                  indices=indices,
                                  dim=dim,
                                  fs=fs,
                                  path=partition,
                                  trim=trim,
-                                 **kwargs)
+                                 **func_kwargs)
 
     return wrap_function
 

zcollection/view/__init__.py

@@ -510,8 +510,6 @@ def update(
             wrap_function = _wrap_update_func(
                 func,
                 self.fs,
-                *args,
-                **kwargs,
             )
         else:
             if selected_variables is not None and len(
@@ -527,8 +525,6 @@ def update(
                 self.fs,
                 self.view_ref,
                 trim,
-                *args,
-                **kwargs,
             )
 
         batchs: Iterator[Sequence[Any]] = dask_utils.split_sequence(
@@ -538,7 +534,9 @@ def update(
             wrap_function,
             tuple(batchs),
             key=func.__name__,
-            base_dir=self.base_dir)
+            base_dir=self.base_dir,
+            func_args=args,
+            func_kwargs=kwargs)
         storage.execute_transaction(client, self.synchronizer, awaitables)
 
     # pylint: disable=duplicate-code

zcollection/view/detail.py

@@ -8,6 +8,7 @@
 """
 from __future__ import annotations
 
+from typing import Any
 import base64
 from collections.abc import Callable, Iterable, Iterator, Sequence
 import dataclasses
@@ -37,8 +38,9 @@
 from ..type_hints import ArrayLike, NDArray
 
 #: Type of the function used to update a view.
-ViewUpdateCallable = \
-    Callable[[Iterable[tuple[dataset.Dataset, str]], str], None]
+ViewUpdateCallable = Callable[
+    [Iterable[tuple[dataset.Dataset, str]], str, list[Any], dict[str,
+                                                                 Any]], None]
 
 #: Name of the file that contains the checksum of the view.
 CHECKSUM_FILE = '.checksum'
@@ -350,28 +352,26 @@ def calculate_slice(
 def _wrap_update_func(
     func: collection.UpdateCallable,
     fs: fsspec.AbstractFileSystem,
-    *args,
-    **kwargs,
 ) -> ViewUpdateCallable:
     """Wrap an update function taking a list of partition's dataset and
     partition's path as input and returning None.
 
     Args:
         func: The update function.
         fs: The file system used to access the variables in the view.
-        *args: The arguments of the update function.
-        **kwargs: The keyword arguments of the update function.
 
     Returns:
         The wrapped function.
     """
 
     def wrap_function(parameters: Iterable[tuple[dataset.Dataset, str]],
-                      base_dir: str) -> None:
+                      base_dir: str, func_args: list[Any],
+                      func_kwargs: dict[str, Any]) -> None:
         """Wrap the function to be applied to the dataset."""
         for zds, partition in parameters:
             # Applying function on partition's data
-            dictionary: dict[str, ArrayLike] = func(zds, *args, **kwargs)
+            dictionary: dict[str, ArrayLike] = func(zds, *func_args,
+                                                    **func_kwargs)
             tuple(
                 update_zarr_array(  # type: ignore[func-returns-value]
                     dirname=join_path(base_dir, partition, varname),
@@ -389,8 +389,6 @@ def _wrap_update_func_overlap(
     fs: fsspec.AbstractFileSystem,
     view_ref: collection.Collection,
     trim: bool,
-    *args,
-    **kwargs,
 ) -> ViewUpdateCallable:
     """Wrap an update function taking a list of partition's dataset and
     partition's path as input and returning None.
@@ -402,8 +400,6 @@ def _wrap_update_func_overlap(
         fs: The file system used to access the variables in the view.
         view_ref: The view reference.
         trim: If True, trim the dataset to the overlap.
-        *args: The arguments of the update function.
-        **kwargs: The keyword arguments of the update function.
 
     Returns:
         The wrapped function.
@@ -414,7 +410,8 @@ def _wrap_update_func_overlap(
         raise ValueError('The depth must be positive')
 
     def wrap_function(parameters: Iterable[tuple[dataset.Dataset, str]],
-                      base_dir: str) -> None:
+                      base_dir: str, func_args: list[Any],
+                      func_kwargs: dict[str, Any]) -> None:
         """Wrap the function to be applied to the dataset."""
         zds: dataset.Dataset
         indices: slice
@@ -425,15 +422,15 @@ def wrap_function(parameters: Iterable[tuple[dataset.Dataset, str]],
             # pylint: disable=duplicate-code
             # False positive with the function _wrap_update_func_with_overlap
             # defined in the module zcollection.collection.detail
-            _update_with_overlap(*args,
+            _update_with_overlap(*func_args,
                                  func=func,
                                  zds=zds,
                                  indices=indices,
                                  dim=dim,
                                  fs=fs,
                                  path=join_path(base_dir, partition),
                                  trim=trim,
-                                 **kwargs)
+                                 **func_kwargs)
             # pylint: enable=duplicate-code
 
     return wrap_function