Unify dpnp public API exports by refactoring dpnp_iface.py

[vlad-perevezentsev]

This PR proposes to refactor the export structure of dpnp_iface.py by moving all python function exports into dpnp/__init__.py
Redundant __all__ declarations and wildcard imports inside dpnp_iface.py have been removed.

This is the first step toward unifying dpnp public API exports and eliminating wildcard imports.
A next PR will remove the remaining wildcard imports in __init__.py and finalize the unified function export

• Have you provided a meaningful PR description?
• Have you added a test, reproducer or referred to an issue with a reproducer?
• Have you tested your changes locally for CPU and GPU devices?
• Have you made sure that new changes do not introduce compiler warnings?
• Have you checked performance impact of proposed changes?
• Have you added documentation for your changes, if necessary?
• Have you added your changes to the changelog?

[github-actions]

View rendered docs @ https://intelpython.github.io/dpnp/pull/2665/index.html

[coveralls]

coverage: 81.421% (-0.04%) from 81.457%
when pulling 1ee5b5a on refactor_public_api_exports
into f0eb0a8 on master.

[antonwolfy]

Please add the change to the changelog

[antonwolfy]

Suggested change

	from dpnp.dpnp_array import dpnp_array
	from .dpnp_array import dpnp_array

[antonwolfy]

Suggested change

	# Binary operations
	# Bit-wise operations

[antonwolfy]

Suggested change

	# Functional routines
	# Functional programming

[antonwolfy]

It's a part of Miscellaneous routines

[antonwolfy]

Window functions

[antonwolfy]

Suggested change

	# DPNP iface functions
	# Utils functions

[antonwolfy]

Or even better probably:

Suggested change

	# DPNP iface functions
	# Helper functions

[antonwolfy]

I guess it'd be great to separete them somehow with

# =============================================================================

because the numpy's borrowed routines end here

[antonwolfy]

Is that intended that all the above functions are not part of the __all__ now?
Is there any side effect?

[antonwolfy]

NumPy builds __all__ as a list.
We can split extending __all__ here by functional blocks based on the above separation by the routines (similarly to dpctl approach).

[vlad-perevezentsev]

Removing __all__ is intentional and follows CuPy init.py
The public API is now defined solely by explicit re-exports in __init__.py so extending __all__ is no longer needed.
All public names remain available via dpnp.*function_name*.
from dpnp import * still works and now imports only the functions that are part of the public API as defined in __init__.py which is the correct behavior, i think

[antonwolfy]

from dpnp import * does not seem working:

from dpnp import *

cos
# ---------------------------------------------------------------------------
# NameError                                 Traceback (most recent call last)
# Cell In[2], line 1
# ----> 1 cos
# NameError: name 'cos' is not defined

print(dir())
# ['In', 'Out', '_', '_3', '_4', '_5', '_6', '__', '___', '__builtin__', '__builtins__', '__doc__', '__loader__', '__name__', '__package__', '__spec__', '_dh', '_i', '_i1', '_i2', '_i3', '_i4', '_i5', '_i6', '_i7', '_i8', '_ih', '_ii', '_iii', '_oh', 'byte_bounds', 'exceptions', 'exit', 'fft', 'get_ipython', 'linalg', 'open', 'quit', 'random', 'scipy']

while in numpy:

from numpy import *

cos
# Out: <ufunc 'cos'>

print(dir())
# ['False_', 'In', 'Out', 'ScalarType', 'True_', '_', '_2', '__', '___', '__array_namespace_info__', '__builtin__', '__builtins__', '__doc__', '__loader__', '__name__', '__package__', '__spec__', '__version__', '_dh', '_i', '_i1', '_i2', '_i3', '_ih', '_ii', '_iii', '_oh', 'abs', 'absolute', 'acos', 'acosh', 'add', 'all', 'allclose', 'amax', 'amin', 'angle', 'any', 'append', 'apply_along_axis', 'apply_over_axes', 'arange', 'arccos', 'arccosh', 'arcsin', 'arcsinh', 'arctan', 'arctan2', 'arctanh', 'argmax', 'argmin', 'argpartition', 'argsort', 'argwhere', 'around', 'array', 'array2string', 'array_equal', 'array_equiv', 'array_repr', 'array_split', 'array_str', 'asanyarray', 'asarray', 'asarray_chkfinite', 'ascontiguousarray', 'asfortranarray', 'asin', 'asinh', 'asmatrix', 'astype', 'atan', 'atan2', 'atanh', 'atleast_1d', 'atleast_2d', 'atleast_3d', 'average', 'bartlett', 'base_repr', 'binary_repr', 'bincount', 'bitwise_and', 'bitwise_count', 'bitwise_invert', 'bitwise_left_shift', 'bitwise_not', 'bitwise_or', 'bitwise_right_shift', 'bitwise_xor', 'blackman', 'block', 'bmat', 'bool', 'bool_', 'broadcast', 'broadcast_arrays', 'broadcast_shapes', 'broadcast_to', 'busday_count', 'busday_offset', 'busdaycalendar', 'byte', 'bytes_', 'c_', 'can_cast', 'cbrt', 'cdouble', 'ceil', 'char', 'character', 'choose', 'clip', 'clongdouble', 'column_stack', 'common_type', 'complex128', 'complex256', 'complex64', 'complexfloating', 'compress', 'concat', 'concatenate', 'conj', 'conjugate', 'convolve', 'copy', 'copysign', 'copyto', 'core', 'corrcoef', 'correlate', 'cos', 'cosh', 'count_nonzero', 'cov', 'cross', 'csingle', 'ctypeslib', 'cumprod', 'cumsum', 'cumulative_prod', 'cumulative_sum', 'datetime64', 'datetime_as_string', 'datetime_data', 'deg2rad', 'degrees', 'delete', 'diag', 'diag_indices', 'diag_indices_from', 'diagflat', 'diagonal', 'diff', 'digitize', 'divide', 'divmod', 'dot', 'double', 'dsplit', 'dstack', 'dtype', 'dtypes', 'e', 'ediff1d', 'einsum', 'einsum_path', 'emath', 'empty', 'empty_like', 'equal', 'errstate', 'euler_gamma', 'exceptions', 'exit', 'exp', 'exp2', 'expand_dims', 'expm1', 'extract', 'eye', 'f2py', 'fabs', 'fft', 'fill_diagonal', 'finfo', 'fix', 'flatiter', 'flatnonzero', 'flexible', 'flip', 'fliplr', 'flipud', 'float128', 'float16', 'float32', 'float64', 'float_power', 'floating', 'floor', 'floor_divide', 'fmax', 'fmin', 'fmod', 'format_float_positional', 'format_float_scientific', 'frexp', 'from_dlpack', 'frombuffer', 'fromfile', 'fromfunction', 'fromiter', 'frompyfunc', 'fromregex', 'fromstring', 'full', 'full_like', 'gcd', 'generic', 'genfromtxt', 'geomspace', 'get_include', 'get_ipython', 'get_printoptions', 'getbufsize', 'geterr', 'geterrcall', 'gradient', 'greater', 'greater_equal', 'half', 'hamming', 'hanning', 'heaviside', 'histogram', 'histogram2d', 'histogram_bin_edges', 'histogramdd', 'hsplit', 'hstack', 'hypot', 'i0', 'identity', 'iinfo', 'imag', 'in1d', 'index_exp', 'indices', 'inexact', 'inf', 'info', 'inner', 'insert', 'int16', 'int32', 'int64', 'int8', 'int_', 'intc', 'integer', 'interp', 'intersect1d', 'intp', 'invert', 'is_busday', 'isclose', 'iscomplex', 'iscomplexobj', 'isdtype', 'isfinite', 'isfortran', 'isin', 'isinf', 'isnan', 'isnat', 'isneginf', 'isposinf', 'isreal', 'isrealobj', 'isscalar', 'issubdtype', 'iterable', 'ix_', 'kaiser', 'kron', 'lcm', 'ldexp', 'left_shift', 'less', 'less_equal', 'lexsort', 'lib', 'linalg', 'linspace', 'little_endian', 'load', 'loadtxt', 'log', 'log10', 'log1p', 'log2', 'logaddexp', 'logaddexp2', 'logical_and', 'logical_not', 'logical_or', 'logical_xor', 'logspace', 'long', 'longdouble', 'longlong', 'ma', 'mask_indices', 'matmul', 'matrix', 'matrix_transpose', 'matvec', 'max', 'maximum', 'may_share_memory', 'mean', 'median', 'memmap', 'meshgrid', 'mgrid', 'min', 'min_scalar_type', 'minimum', 'mintypecode', 'mod', 'modf', 'moveaxis', 'multiply', 'nan', 'nan_to_num', 'nanargmax', 'nanargmin', 'nancumprod', 'nancumsum', 'nanmax', 'nanmean', 'nanmedian', 'nanmin', 'nanpercentile', 'nanprod', 'nanquantile', 'nanstd', 'nansum', 'nanvar', 'ndarray', 'ndenumerate', 'ndim', 'ndindex', 'nditer', 'negative', 'nested_iters', 'newaxis', 'nextafter', 'nonzero', 'not_equal', 'number', 'object_', 'ogrid', 'ones', 'ones_like', 'open', 'outer', 'packbits', 'pad', 'partition', 'percentile', 'permute_dims', 'pi', 'piecewise', 'place', 'poly', 'poly1d', 'polyadd', 'polyder', 'polydiv', 'polyfit', 'polyint', 'polymul', 'polynomial', 'polysub', 'polyval', 'positive', 'pow', 'power', 'printoptions', 'prod', 'promote_types', 'ptp', 'put', 'put_along_axis', 'putmask', 'quantile', 'quit', 'r_', 'rad2deg', 'radians', 'random', 'ravel', 'ravel_multi_index', 'real', 'real_if_close', 'rec', 'recarray', 'reciprocal', 'record', 'remainder', 'repeat', 'require', 'reshape', 'resize', 'result_type', 'right_shift', 'rint', 'roll', 'rollaxis', 'roots', 'rot90', 'round', 'row_stack', 's_', 'save', 'savetxt', 'savez', 'savez_compressed', 'sctypeDict', 'searchsorted', 'select', 'set_printoptions', 'setbufsize', 'setdiff1d', 'seterr', 'seterrcall', 'setxor1d', 'shape', 'shares_memory', 'short', 'show_config', 'show_runtime', 'sign', 'signbit', 'signedinteger', 'sin', 'sinc', 'single', 'sinh', 'size', 'sort', 'sort_complex', 'spacing', 'split', 'sqrt', 'square', 'squeeze', 'stack', 'std', 'str_', 'strings', 'subtract', 'sum', 'swapaxes', 'take', 'take_along_axis', 'tan', 'tanh', 'tensordot', 'test', 'testing', 'tile', 'timedelta64', 'trace', 'transpose', 'trapezoid', 'trapz', 'tri', 'tril', 'tril_indices', 'tril_indices_from', 'trim_zeros', 'triu', 'triu_indices', 'triu_indices_from', 'true_divide', 'trunc', 'typecodes', 'typename', 'typing', 'ubyte', 'ufunc', 'uint', 'uint16', 'uint32', 'uint64', 'uint8', 'uintc', 'uintp', 'ulong', 'ulonglong', 'union1d', 'unique', 'unique_all', 'unique_counts', 'unique_inverse', 'unique_values', 'unpackbits', 'unravel_index', 'unsignedinteger', 'unstack', 'unwrap', 'ushort', 'vander', 'var', 'vdot', 'vecdot', 'vecmat', 'vectorize', 'void', 'vsplit', 'vstack', 'where', 'zeros', 'zeros_like']

[antonwolfy]

Per my understanding it's a good practice to define __all__. It helps to have clear and explicit exposing of the API.

[antonwolfy]

Could you please remove that from dpnp/dpnp_iface_statistics.py also, because it's done globally in .pre-commit-config.yaml